{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9999099301959018,
  "eval_steps": 500,
  "global_step": 11102,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00018013960819635217,
      "grad_norm": 1.0146561861038208,
      "learning_rate": 1e-05,
      "loss": 3.4013,
      "step": 1
    },
    {
      "epoch": 0.00036027921639270434,
      "grad_norm": 1.078662395477295,
      "learning_rate": 2e-05,
      "loss": 3.6395,
      "step": 2
    },
    {
      "epoch": 0.0005404188245890565,
      "grad_norm": 1.1184042692184448,
      "learning_rate": 3e-05,
      "loss": 3.638,
      "step": 3
    },
    {
      "epoch": 0.0007205584327854087,
      "grad_norm": 1.2305233478546143,
      "learning_rate": 4e-05,
      "loss": 3.6723,
      "step": 4
    },
    {
      "epoch": 0.0009006980409817609,
      "grad_norm": 1.2428853511810303,
      "learning_rate": 5e-05,
      "loss": 4.0646,
      "step": 5
    },
    {
      "epoch": 0.001080837649178113,
      "grad_norm": 1.4851632118225098,
      "learning_rate": 6e-05,
      "loss": 4.1793,
      "step": 6
    },
    {
      "epoch": 0.0012609772573744652,
      "grad_norm": 1.4113008975982666,
      "learning_rate": 7e-05,
      "loss": 4.257,
      "step": 7
    },
    {
      "epoch": 0.0014411168655708174,
      "grad_norm": 1.4709892272949219,
      "learning_rate": 8e-05,
      "loss": 4.5698,
      "step": 8
    },
    {
      "epoch": 0.0016212564737671695,
      "grad_norm": 1.5990041494369507,
      "learning_rate": 9e-05,
      "loss": 4.3774,
      "step": 9
    },
    {
      "epoch": 0.0018013960819635217,
      "grad_norm": 1.0094465017318726,
      "learning_rate": 0.0001,
      "loss": 3.1328,
      "step": 10
    },
    {
      "epoch": 0.001981535690159874,
      "grad_norm": 0.9330719709396362,
      "learning_rate": 9.999999799451219e-05,
      "loss": 3.2082,
      "step": 11
    },
    {
      "epoch": 0.002161675298356226,
      "grad_norm": 0.9791777729988098,
      "learning_rate": 9.999999197804887e-05,
      "loss": 2.9366,
      "step": 12
    },
    {
      "epoch": 0.0023418149065525785,
      "grad_norm": 0.9647117853164673,
      "learning_rate": 9.999998195061056e-05,
      "loss": 2.8834,
      "step": 13
    },
    {
      "epoch": 0.0025219545147489304,
      "grad_norm": 1.0648342370986938,
      "learning_rate": 9.999996791219805e-05,
      "loss": 3.1131,
      "step": 14
    },
    {
      "epoch": 0.002702094122945283,
      "grad_norm": 0.9934209585189819,
      "learning_rate": 9.999994986281247e-05,
      "loss": 2.5361,
      "step": 15
    },
    {
      "epoch": 0.0028822337311416348,
      "grad_norm": 1.257798671722412,
      "learning_rate": 9.999992780245527e-05,
      "loss": 2.8522,
      "step": 16
    },
    {
      "epoch": 0.003062373339337987,
      "grad_norm": 1.149506688117981,
      "learning_rate": 9.99999017311282e-05,
      "loss": 2.9118,
      "step": 17
    },
    {
      "epoch": 0.003242512947534339,
      "grad_norm": 1.1489620208740234,
      "learning_rate": 9.99998716488334e-05,
      "loss": 2.8234,
      "step": 18
    },
    {
      "epoch": 0.0034226525557306915,
      "grad_norm": 1.0399941205978394,
      "learning_rate": 9.999983755557323e-05,
      "loss": 2.8068,
      "step": 19
    },
    {
      "epoch": 0.0036027921639270434,
      "grad_norm": 0.8986930847167969,
      "learning_rate": 9.999979945135043e-05,
      "loss": 2.8921,
      "step": 20
    },
    {
      "epoch": 0.003782931772123396,
      "grad_norm": 0.9371277689933777,
      "learning_rate": 9.999975733616809e-05,
      "loss": 2.4691,
      "step": 21
    },
    {
      "epoch": 0.003963071380319748,
      "grad_norm": 0.9938897490501404,
      "learning_rate": 9.999971121002958e-05,
      "loss": 2.8946,
      "step": 22
    },
    {
      "epoch": 0.0041432109885161,
      "grad_norm": 0.8986033201217651,
      "learning_rate": 9.999966107293857e-05,
      "loss": 2.5737,
      "step": 23
    },
    {
      "epoch": 0.004323350596712452,
      "grad_norm": 0.8785312175750732,
      "learning_rate": 9.999960692489911e-05,
      "loss": 2.6496,
      "step": 24
    },
    {
      "epoch": 0.004503490204908804,
      "grad_norm": 0.863671600818634,
      "learning_rate": 9.999954876591552e-05,
      "loss": 2.3118,
      "step": 25
    },
    {
      "epoch": 0.004683629813105157,
      "grad_norm": 0.9351087212562561,
      "learning_rate": 9.999948659599252e-05,
      "loss": 2.7481,
      "step": 26
    },
    {
      "epoch": 0.004863769421301509,
      "grad_norm": 0.9674515724182129,
      "learning_rate": 9.999942041513504e-05,
      "loss": 2.7264,
      "step": 27
    },
    {
      "epoch": 0.005043909029497861,
      "grad_norm": 1.075107455253601,
      "learning_rate": 9.99993502233484e-05,
      "loss": 2.7795,
      "step": 28
    },
    {
      "epoch": 0.005224048637694213,
      "grad_norm": 1.000078797340393,
      "learning_rate": 9.999927602063825e-05,
      "loss": 2.7193,
      "step": 29
    },
    {
      "epoch": 0.005404188245890566,
      "grad_norm": 0.9467408061027527,
      "learning_rate": 9.999919780701052e-05,
      "loss": 2.4211,
      "step": 30
    },
    {
      "epoch": 0.0055843278540869176,
      "grad_norm": 1.0002549886703491,
      "learning_rate": 9.999911558247151e-05,
      "loss": 2.4868,
      "step": 31
    },
    {
      "epoch": 0.0057644674622832695,
      "grad_norm": 0.9504591226577759,
      "learning_rate": 9.999902934702778e-05,
      "loss": 2.6444,
      "step": 32
    },
    {
      "epoch": 0.0059446070704796215,
      "grad_norm": 0.9449986815452576,
      "learning_rate": 9.99989391006863e-05,
      "loss": 2.5808,
      "step": 33
    },
    {
      "epoch": 0.006124746678675974,
      "grad_norm": 0.8339290618896484,
      "learning_rate": 9.999884484345426e-05,
      "loss": 2.4875,
      "step": 34
    },
    {
      "epoch": 0.006304886286872326,
      "grad_norm": 0.829587459564209,
      "learning_rate": 9.999874657533923e-05,
      "loss": 2.5427,
      "step": 35
    },
    {
      "epoch": 0.006485025895068678,
      "grad_norm": 0.7929829359054565,
      "learning_rate": 9.999864429634913e-05,
      "loss": 2.1261,
      "step": 36
    },
    {
      "epoch": 0.00666516550326503,
      "grad_norm": 0.8820804357528687,
      "learning_rate": 9.999853800649211e-05,
      "loss": 3.0714,
      "step": 37
    },
    {
      "epoch": 0.006845305111461383,
      "grad_norm": 0.8300914764404297,
      "learning_rate": 9.999842770577675e-05,
      "loss": 2.5948,
      "step": 38
    },
    {
      "epoch": 0.007025444719657735,
      "grad_norm": 0.8988324403762817,
      "learning_rate": 9.999831339421187e-05,
      "loss": 2.6299,
      "step": 39
    },
    {
      "epoch": 0.007205584327854087,
      "grad_norm": 0.8685336709022522,
      "learning_rate": 9.999819507180662e-05,
      "loss": 2.8515,
      "step": 40
    },
    {
      "epoch": 0.007385723936050439,
      "grad_norm": 0.7915685176849365,
      "learning_rate": 9.999807273857054e-05,
      "loss": 2.3942,
      "step": 41
    },
    {
      "epoch": 0.007565863544246792,
      "grad_norm": 0.8272666335105896,
      "learning_rate": 9.99979463945134e-05,
      "loss": 2.669,
      "step": 42
    },
    {
      "epoch": 0.007746003152443144,
      "grad_norm": 0.845380425453186,
      "learning_rate": 9.999781603964535e-05,
      "loss": 2.6562,
      "step": 43
    },
    {
      "epoch": 0.007926142760639496,
      "grad_norm": 0.8954911231994629,
      "learning_rate": 9.999768167397687e-05,
      "loss": 2.9604,
      "step": 44
    },
    {
      "epoch": 0.008106282368835848,
      "grad_norm": 0.7897361516952515,
      "learning_rate": 9.999754329751871e-05,
      "loss": 2.7354,
      "step": 45
    },
    {
      "epoch": 0.0082864219770322,
      "grad_norm": 0.8153886795043945,
      "learning_rate": 9.999740091028198e-05,
      "loss": 2.5302,
      "step": 46
    },
    {
      "epoch": 0.008466561585228552,
      "grad_norm": 0.8083345293998718,
      "learning_rate": 9.99972545122781e-05,
      "loss": 2.6665,
      "step": 47
    },
    {
      "epoch": 0.008646701193424904,
      "grad_norm": 0.7844825387001038,
      "learning_rate": 9.999710410351881e-05,
      "loss": 2.4497,
      "step": 48
    },
    {
      "epoch": 0.008826840801621256,
      "grad_norm": 0.7885056734085083,
      "learning_rate": 9.999694968401618e-05,
      "loss": 2.6281,
      "step": 49
    },
    {
      "epoch": 0.009006980409817608,
      "grad_norm": 0.9038015007972717,
      "learning_rate": 9.999679125378262e-05,
      "loss": 2.4357,
      "step": 50
    },
    {
      "epoch": 0.00918712001801396,
      "grad_norm": 1.2260198593139648,
      "learning_rate": 9.999662881283081e-05,
      "loss": 2.7279,
      "step": 51
    },
    {
      "epoch": 0.009367259626210314,
      "grad_norm": 1.0754985809326172,
      "learning_rate": 9.999646236117377e-05,
      "loss": 2.9244,
      "step": 52
    },
    {
      "epoch": 0.009547399234406666,
      "grad_norm": 1.1802886724472046,
      "learning_rate": 9.999629189882489e-05,
      "loss": 3.1561,
      "step": 53
    },
    {
      "epoch": 0.009727538842603018,
      "grad_norm": 1.0842028856277466,
      "learning_rate": 9.999611742579783e-05,
      "loss": 2.9517,
      "step": 54
    },
    {
      "epoch": 0.00990767845079937,
      "grad_norm": 1.1068226099014282,
      "learning_rate": 9.999593894210657e-05,
      "loss": 3.2007,
      "step": 55
    },
    {
      "epoch": 0.010087818058995722,
      "grad_norm": 0.9489306211471558,
      "learning_rate": 9.999575644776544e-05,
      "loss": 2.9234,
      "step": 56
    },
    {
      "epoch": 0.010267957667192074,
      "grad_norm": 1.0008602142333984,
      "learning_rate": 9.99955699427891e-05,
      "loss": 2.8981,
      "step": 57
    },
    {
      "epoch": 0.010448097275388426,
      "grad_norm": 1.174797773361206,
      "learning_rate": 9.999537942719247e-05,
      "loss": 3.5682,
      "step": 58
    },
    {
      "epoch": 0.010628236883584778,
      "grad_norm": 1.3938766717910767,
      "learning_rate": 9.999518490099086e-05,
      "loss": 3.8545,
      "step": 59
    },
    {
      "epoch": 0.010808376491781131,
      "grad_norm": 1.3305302858352661,
      "learning_rate": 9.999498636419988e-05,
      "loss": 3.0109,
      "step": 60
    },
    {
      "epoch": 0.010988516099977483,
      "grad_norm": 0.8620555996894836,
      "learning_rate": 9.999478381683543e-05,
      "loss": 2.2525,
      "step": 61
    },
    {
      "epoch": 0.011168655708173835,
      "grad_norm": 0.7407914400100708,
      "learning_rate": 9.999457725891377e-05,
      "loss": 2.3131,
      "step": 62
    },
    {
      "epoch": 0.011348795316370187,
      "grad_norm": 0.9008898138999939,
      "learning_rate": 9.999436669045147e-05,
      "loss": 2.252,
      "step": 63
    },
    {
      "epoch": 0.011528934924566539,
      "grad_norm": 0.8526875972747803,
      "learning_rate": 9.999415211146545e-05,
      "loss": 2.3424,
      "step": 64
    },
    {
      "epoch": 0.011709074532762891,
      "grad_norm": 0.8313794732093811,
      "learning_rate": 9.999393352197289e-05,
      "loss": 2.3411,
      "step": 65
    },
    {
      "epoch": 0.011889214140959243,
      "grad_norm": 0.8745887875556946,
      "learning_rate": 9.999371092199132e-05,
      "loss": 2.354,
      "step": 66
    },
    {
      "epoch": 0.012069353749155595,
      "grad_norm": 0.7536364197731018,
      "learning_rate": 9.999348431153863e-05,
      "loss": 2.358,
      "step": 67
    },
    {
      "epoch": 0.012249493357351949,
      "grad_norm": 0.7970806360244751,
      "learning_rate": 9.999325369063297e-05,
      "loss": 2.1937,
      "step": 68
    },
    {
      "epoch": 0.0124296329655483,
      "grad_norm": 0.8265585899353027,
      "learning_rate": 9.999301905929286e-05,
      "loss": 2.4513,
      "step": 69
    },
    {
      "epoch": 0.012609772573744652,
      "grad_norm": 0.6983076930046082,
      "learning_rate": 9.99927804175371e-05,
      "loss": 2.2879,
      "step": 70
    },
    {
      "epoch": 0.012789912181941004,
      "grad_norm": 0.7495995759963989,
      "learning_rate": 9.999253776538484e-05,
      "loss": 2.5069,
      "step": 71
    },
    {
      "epoch": 0.012970051790137356,
      "grad_norm": 0.776552677154541,
      "learning_rate": 9.999229110285557e-05,
      "loss": 2.5313,
      "step": 72
    },
    {
      "epoch": 0.013150191398333708,
      "grad_norm": 0.7193177938461304,
      "learning_rate": 9.999204042996904e-05,
      "loss": 2.268,
      "step": 73
    },
    {
      "epoch": 0.01333033100653006,
      "grad_norm": 0.7583246231079102,
      "learning_rate": 9.999178574674539e-05,
      "loss": 2.2903,
      "step": 74
    },
    {
      "epoch": 0.013510470614726412,
      "grad_norm": 0.8359841108322144,
      "learning_rate": 9.999152705320502e-05,
      "loss": 2.1628,
      "step": 75
    },
    {
      "epoch": 0.013690610222922766,
      "grad_norm": 0.8961155414581299,
      "learning_rate": 9.999126434936871e-05,
      "loss": 2.4106,
      "step": 76
    },
    {
      "epoch": 0.013870749831119118,
      "grad_norm": 0.814487099647522,
      "learning_rate": 9.999099763525755e-05,
      "loss": 2.5886,
      "step": 77
    },
    {
      "epoch": 0.01405088943931547,
      "grad_norm": 0.8167582750320435,
      "learning_rate": 9.999072691089287e-05,
      "loss": 2.3747,
      "step": 78
    },
    {
      "epoch": 0.014231029047511822,
      "grad_norm": 0.7982463836669922,
      "learning_rate": 9.999045217629644e-05,
      "loss": 2.4233,
      "step": 79
    },
    {
      "epoch": 0.014411168655708174,
      "grad_norm": 0.7659124732017517,
      "learning_rate": 9.999017343149029e-05,
      "loss": 2.2224,
      "step": 80
    },
    {
      "epoch": 0.014591308263904526,
      "grad_norm": 0.8021962642669678,
      "learning_rate": 9.998989067649678e-05,
      "loss": 2.418,
      "step": 81
    },
    {
      "epoch": 0.014771447872100878,
      "grad_norm": 0.8060923218727112,
      "learning_rate": 9.998960391133857e-05,
      "loss": 2.3914,
      "step": 82
    },
    {
      "epoch": 0.01495158748029723,
      "grad_norm": 0.8500896692276001,
      "learning_rate": 9.998931313603871e-05,
      "loss": 2.8956,
      "step": 83
    },
    {
      "epoch": 0.015131727088493583,
      "grad_norm": 0.7829626798629761,
      "learning_rate": 9.998901835062048e-05,
      "loss": 2.3679,
      "step": 84
    },
    {
      "epoch": 0.015311866696689935,
      "grad_norm": 0.9021100401878357,
      "learning_rate": 9.998871955510756e-05,
      "loss": 2.712,
      "step": 85
    },
    {
      "epoch": 0.015492006304886287,
      "grad_norm": 0.8967096209526062,
      "learning_rate": 9.99884167495239e-05,
      "loss": 2.6505,
      "step": 86
    },
    {
      "epoch": 0.01567214591308264,
      "grad_norm": 0.7971676588058472,
      "learning_rate": 9.99881099338938e-05,
      "loss": 2.4672,
      "step": 87
    },
    {
      "epoch": 0.015852285521278993,
      "grad_norm": 0.9014323353767395,
      "learning_rate": 9.998779910824185e-05,
      "loss": 2.4235,
      "step": 88
    },
    {
      "epoch": 0.016032425129475343,
      "grad_norm": 0.8840922117233276,
      "learning_rate": 9.998748427259302e-05,
      "loss": 2.4347,
      "step": 89
    },
    {
      "epoch": 0.016212564737671697,
      "grad_norm": 0.7735182046890259,
      "learning_rate": 9.998716542697255e-05,
      "loss": 2.4093,
      "step": 90
    },
    {
      "epoch": 0.016392704345868047,
      "grad_norm": 0.8273934125900269,
      "learning_rate": 9.998684257140603e-05,
      "loss": 2.7378,
      "step": 91
    },
    {
      "epoch": 0.0165728439540644,
      "grad_norm": 0.8422622680664062,
      "learning_rate": 9.998651570591932e-05,
      "loss": 2.367,
      "step": 92
    },
    {
      "epoch": 0.01675298356226075,
      "grad_norm": 0.780792236328125,
      "learning_rate": 9.998618483053868e-05,
      "loss": 2.2097,
      "step": 93
    },
    {
      "epoch": 0.016933123170457105,
      "grad_norm": 0.8539138436317444,
      "learning_rate": 9.998584994529063e-05,
      "loss": 2.5079,
      "step": 94
    },
    {
      "epoch": 0.017113262778653455,
      "grad_norm": 0.8521108627319336,
      "learning_rate": 9.998551105020206e-05,
      "loss": 2.1245,
      "step": 95
    },
    {
      "epoch": 0.01729340238684981,
      "grad_norm": 0.9224622845649719,
      "learning_rate": 9.998516814530015e-05,
      "loss": 2.2187,
      "step": 96
    },
    {
      "epoch": 0.017473541995046162,
      "grad_norm": 0.9476681351661682,
      "learning_rate": 9.998482123061237e-05,
      "loss": 2.5253,
      "step": 97
    },
    {
      "epoch": 0.017653681603242512,
      "grad_norm": 1.0359331369400024,
      "learning_rate": 9.99844703061666e-05,
      "loss": 2.7389,
      "step": 98
    },
    {
      "epoch": 0.017833821211438866,
      "grad_norm": 0.9494062662124634,
      "learning_rate": 9.998411537199096e-05,
      "loss": 2.0257,
      "step": 99
    },
    {
      "epoch": 0.018013960819635216,
      "grad_norm": 0.9466662406921387,
      "learning_rate": 9.998375642811393e-05,
      "loss": 2.555,
      "step": 100
    },
    {
      "epoch": 0.01819410042783157,
      "grad_norm": 1.4059162139892578,
      "learning_rate": 9.99833934745643e-05,
      "loss": 3.1052,
      "step": 101
    },
    {
      "epoch": 0.01837424003602792,
      "grad_norm": 1.4918389320373535,
      "learning_rate": 9.99830265113712e-05,
      "loss": 3.09,
      "step": 102
    },
    {
      "epoch": 0.018554379644224274,
      "grad_norm": 1.5020196437835693,
      "learning_rate": 9.998265553856406e-05,
      "loss": 3.1729,
      "step": 103
    },
    {
      "epoch": 0.018734519252420628,
      "grad_norm": 1.1906907558441162,
      "learning_rate": 9.998228055617263e-05,
      "loss": 2.8921,
      "step": 104
    },
    {
      "epoch": 0.018914658860616978,
      "grad_norm": 1.053817868232727,
      "learning_rate": 9.9981901564227e-05,
      "loss": 2.9665,
      "step": 105
    },
    {
      "epoch": 0.01909479846881333,
      "grad_norm": 0.9323416352272034,
      "learning_rate": 9.99815185627576e-05,
      "loss": 2.8324,
      "step": 106
    },
    {
      "epoch": 0.019274938077009682,
      "grad_norm": 1.0798654556274414,
      "learning_rate": 9.99811315517951e-05,
      "loss": 3.2274,
      "step": 107
    },
    {
      "epoch": 0.019455077685206035,
      "grad_norm": 1.1817671060562134,
      "learning_rate": 9.998074053137056e-05,
      "loss": 3.2837,
      "step": 108
    },
    {
      "epoch": 0.019635217293402386,
      "grad_norm": 1.2873262166976929,
      "learning_rate": 9.998034550151538e-05,
      "loss": 3.337,
      "step": 109
    },
    {
      "epoch": 0.01981535690159874,
      "grad_norm": 1.780294418334961,
      "learning_rate": 9.997994646226122e-05,
      "loss": 3.371,
      "step": 110
    },
    {
      "epoch": 0.01999549650979509,
      "grad_norm": 0.9122216105461121,
      "learning_rate": 9.997954341364011e-05,
      "loss": 2.2638,
      "step": 111
    },
    {
      "epoch": 0.020175636117991443,
      "grad_norm": 0.8674176335334778,
      "learning_rate": 9.997913635568435e-05,
      "loss": 2.2437,
      "step": 112
    },
    {
      "epoch": 0.020355775726187797,
      "grad_norm": 0.8664303421974182,
      "learning_rate": 9.997872528842663e-05,
      "loss": 2.2444,
      "step": 113
    },
    {
      "epoch": 0.020535915334384147,
      "grad_norm": 0.890316903591156,
      "learning_rate": 9.997831021189992e-05,
      "loss": 2.3717,
      "step": 114
    },
    {
      "epoch": 0.0207160549425805,
      "grad_norm": 0.776483952999115,
      "learning_rate": 9.997789112613749e-05,
      "loss": 2.1567,
      "step": 115
    },
    {
      "epoch": 0.02089619455077685,
      "grad_norm": 0.9575796127319336,
      "learning_rate": 9.997746803117298e-05,
      "loss": 2.2566,
      "step": 116
    },
    {
      "epoch": 0.021076334158973205,
      "grad_norm": 0.8641085624694824,
      "learning_rate": 9.997704092704034e-05,
      "loss": 2.1461,
      "step": 117
    },
    {
      "epoch": 0.021256473767169555,
      "grad_norm": 0.9085921049118042,
      "learning_rate": 9.997660981377381e-05,
      "loss": 2.3909,
      "step": 118
    },
    {
      "epoch": 0.02143661337536591,
      "grad_norm": 0.8731765747070312,
      "learning_rate": 9.997617469140799e-05,
      "loss": 2.0962,
      "step": 119
    },
    {
      "epoch": 0.021616752983562262,
      "grad_norm": 0.7569247484207153,
      "learning_rate": 9.997573555997777e-05,
      "loss": 1.9514,
      "step": 120
    },
    {
      "epoch": 0.021796892591758613,
      "grad_norm": 0.8122623562812805,
      "learning_rate": 9.99752924195184e-05,
      "loss": 2.2195,
      "step": 121
    },
    {
      "epoch": 0.021977032199954966,
      "grad_norm": 0.7952985763549805,
      "learning_rate": 9.99748452700654e-05,
      "loss": 2.2251,
      "step": 122
    },
    {
      "epoch": 0.022157171808151317,
      "grad_norm": 0.889175295829773,
      "learning_rate": 9.997439411165466e-05,
      "loss": 2.0404,
      "step": 123
    },
    {
      "epoch": 0.02233731141634767,
      "grad_norm": 0.9200299978256226,
      "learning_rate": 9.997393894432237e-05,
      "loss": 2.3201,
      "step": 124
    },
    {
      "epoch": 0.02251745102454402,
      "grad_norm": 0.8736909627914429,
      "learning_rate": 9.997347976810503e-05,
      "loss": 1.9624,
      "step": 125
    },
    {
      "epoch": 0.022697590632740374,
      "grad_norm": 0.8899530172348022,
      "learning_rate": 9.997301658303948e-05,
      "loss": 2.2782,
      "step": 126
    },
    {
      "epoch": 0.022877730240936724,
      "grad_norm": 0.9963721036911011,
      "learning_rate": 9.997254938916289e-05,
      "loss": 2.2359,
      "step": 127
    },
    {
      "epoch": 0.023057869849133078,
      "grad_norm": 0.8973841071128845,
      "learning_rate": 9.997207818651274e-05,
      "loss": 2.2099,
      "step": 128
    },
    {
      "epoch": 0.02323800945732943,
      "grad_norm": 0.9693061113357544,
      "learning_rate": 9.99716029751268e-05,
      "loss": 2.0216,
      "step": 129
    },
    {
      "epoch": 0.023418149065525782,
      "grad_norm": 0.8820716142654419,
      "learning_rate": 9.997112375504322e-05,
      "loss": 2.2054,
      "step": 130
    },
    {
      "epoch": 0.023598288673722136,
      "grad_norm": 1.0860403776168823,
      "learning_rate": 9.997064052630044e-05,
      "loss": 2.4929,
      "step": 131
    },
    {
      "epoch": 0.023778428281918486,
      "grad_norm": 0.9099593758583069,
      "learning_rate": 9.99701532889372e-05,
      "loss": 2.0855,
      "step": 132
    },
    {
      "epoch": 0.02395856789011484,
      "grad_norm": 1.0118968486785889,
      "learning_rate": 9.99696620429926e-05,
      "loss": 2.5576,
      "step": 133
    },
    {
      "epoch": 0.02413870749831119,
      "grad_norm": 0.9647431373596191,
      "learning_rate": 9.996916678850605e-05,
      "loss": 2.2156,
      "step": 134
    },
    {
      "epoch": 0.024318847106507543,
      "grad_norm": 0.9351106882095337,
      "learning_rate": 9.99686675255173e-05,
      "loss": 2.5907,
      "step": 135
    },
    {
      "epoch": 0.024498986714703897,
      "grad_norm": 0.9223719835281372,
      "learning_rate": 9.996816425406636e-05,
      "loss": 2.4192,
      "step": 136
    },
    {
      "epoch": 0.024679126322900247,
      "grad_norm": 0.9520527720451355,
      "learning_rate": 9.996765697419363e-05,
      "loss": 2.3176,
      "step": 137
    },
    {
      "epoch": 0.0248592659310966,
      "grad_norm": 0.8900474309921265,
      "learning_rate": 9.996714568593981e-05,
      "loss": 2.3632,
      "step": 138
    },
    {
      "epoch": 0.02503940553929295,
      "grad_norm": 1.0412280559539795,
      "learning_rate": 9.996663038934588e-05,
      "loss": 2.6637,
      "step": 139
    },
    {
      "epoch": 0.025219545147489305,
      "grad_norm": 0.9756955504417419,
      "learning_rate": 9.99661110844532e-05,
      "loss": 2.4556,
      "step": 140
    },
    {
      "epoch": 0.025399684755685655,
      "grad_norm": 0.9872828125953674,
      "learning_rate": 9.996558777130344e-05,
      "loss": 2.5443,
      "step": 141
    },
    {
      "epoch": 0.02557982436388201,
      "grad_norm": 0.8600561022758484,
      "learning_rate": 9.996506044993855e-05,
      "loss": 2.1805,
      "step": 142
    },
    {
      "epoch": 0.02575996397207836,
      "grad_norm": 0.8384175896644592,
      "learning_rate": 9.996452912040086e-05,
      "loss": 2.0727,
      "step": 143
    },
    {
      "epoch": 0.025940103580274713,
      "grad_norm": 1.015053629875183,
      "learning_rate": 9.996399378273298e-05,
      "loss": 2.5225,
      "step": 144
    },
    {
      "epoch": 0.026120243188471066,
      "grad_norm": 1.0503270626068115,
      "learning_rate": 9.996345443697785e-05,
      "loss": 2.3433,
      "step": 145
    },
    {
      "epoch": 0.026300382796667417,
      "grad_norm": 0.904880940914154,
      "learning_rate": 9.996291108317874e-05,
      "loss": 2.1061,
      "step": 146
    },
    {
      "epoch": 0.02648052240486377,
      "grad_norm": 0.9697627425193787,
      "learning_rate": 9.996236372137924e-05,
      "loss": 2.4878,
      "step": 147
    },
    {
      "epoch": 0.02666066201306012,
      "grad_norm": 0.8747907280921936,
      "learning_rate": 9.996181235162327e-05,
      "loss": 1.8756,
      "step": 148
    },
    {
      "epoch": 0.026840801621256474,
      "grad_norm": 0.9519774913787842,
      "learning_rate": 9.996125697395503e-05,
      "loss": 2.2033,
      "step": 149
    },
    {
      "epoch": 0.027020941229452825,
      "grad_norm": 0.9820895195007324,
      "learning_rate": 9.99606975884191e-05,
      "loss": 2.2792,
      "step": 150
    },
    {
      "epoch": 0.027201080837649178,
      "grad_norm": 1.1771305799484253,
      "learning_rate": 9.996013419506034e-05,
      "loss": 2.8351,
      "step": 151
    },
    {
      "epoch": 0.027381220445845532,
      "grad_norm": 1.1806966066360474,
      "learning_rate": 9.995956679392395e-05,
      "loss": 2.8213,
      "step": 152
    },
    {
      "epoch": 0.027561360054041882,
      "grad_norm": 1.113235354423523,
      "learning_rate": 9.995899538505546e-05,
      "loss": 2.5992,
      "step": 153
    },
    {
      "epoch": 0.027741499662238236,
      "grad_norm": 1.0962517261505127,
      "learning_rate": 9.995841996850068e-05,
      "loss": 3.0989,
      "step": 154
    },
    {
      "epoch": 0.027921639270434586,
      "grad_norm": 1.1463273763656616,
      "learning_rate": 9.99578405443058e-05,
      "loss": 2.9798,
      "step": 155
    },
    {
      "epoch": 0.02810177887863094,
      "grad_norm": 1.1385432481765747,
      "learning_rate": 9.995725711251726e-05,
      "loss": 2.9343,
      "step": 156
    },
    {
      "epoch": 0.02828191848682729,
      "grad_norm": 1.2719049453735352,
      "learning_rate": 9.995666967318191e-05,
      "loss": 2.9438,
      "step": 157
    },
    {
      "epoch": 0.028462058095023644,
      "grad_norm": 1.303350806236267,
      "learning_rate": 9.995607822634684e-05,
      "loss": 3.0166,
      "step": 158
    },
    {
      "epoch": 0.028642197703219994,
      "grad_norm": 1.7089077234268188,
      "learning_rate": 9.99554827720595e-05,
      "loss": 3.5575,
      "step": 159
    },
    {
      "epoch": 0.028822337311416348,
      "grad_norm": 0.9399608969688416,
      "learning_rate": 9.995488331036769e-05,
      "loss": 2.5122,
      "step": 160
    },
    {
      "epoch": 0.0290024769196127,
      "grad_norm": 0.9248188734054565,
      "learning_rate": 9.995427984131945e-05,
      "loss": 2.0853,
      "step": 161
    },
    {
      "epoch": 0.02918261652780905,
      "grad_norm": 1.037304401397705,
      "learning_rate": 9.995367236496322e-05,
      "loss": 2.085,
      "step": 162
    },
    {
      "epoch": 0.029362756136005405,
      "grad_norm": 1.0212739706039429,
      "learning_rate": 9.995306088134773e-05,
      "loss": 2.2672,
      "step": 163
    },
    {
      "epoch": 0.029542895744201755,
      "grad_norm": 0.9299241900444031,
      "learning_rate": 9.995244539052202e-05,
      "loss": 2.2869,
      "step": 164
    },
    {
      "epoch": 0.02972303535239811,
      "grad_norm": 1.0138578414916992,
      "learning_rate": 9.995182589253548e-05,
      "loss": 2.0205,
      "step": 165
    },
    {
      "epoch": 0.02990317496059446,
      "grad_norm": 0.8964521884918213,
      "learning_rate": 9.995120238743779e-05,
      "loss": 1.9681,
      "step": 166
    },
    {
      "epoch": 0.030083314568790813,
      "grad_norm": 0.9747411012649536,
      "learning_rate": 9.995057487527898e-05,
      "loss": 2.3657,
      "step": 167
    },
    {
      "epoch": 0.030263454176987167,
      "grad_norm": 0.9115554690361023,
      "learning_rate": 9.994994335610939e-05,
      "loss": 2.0457,
      "step": 168
    },
    {
      "epoch": 0.030443593785183517,
      "grad_norm": 0.920690655708313,
      "learning_rate": 9.994930782997966e-05,
      "loss": 2.2838,
      "step": 169
    },
    {
      "epoch": 0.03062373339337987,
      "grad_norm": 0.935954213142395,
      "learning_rate": 9.994866829694079e-05,
      "loss": 2.2176,
      "step": 170
    },
    {
      "epoch": 0.03080387300157622,
      "grad_norm": 0.8992125391960144,
      "learning_rate": 9.994802475704408e-05,
      "loss": 2.0229,
      "step": 171
    },
    {
      "epoch": 0.030984012609772574,
      "grad_norm": 0.9858156442642212,
      "learning_rate": 9.994737721034114e-05,
      "loss": 2.0928,
      "step": 172
    },
    {
      "epoch": 0.031164152217968925,
      "grad_norm": 0.8882651925086975,
      "learning_rate": 9.994672565688394e-05,
      "loss": 2.1093,
      "step": 173
    },
    {
      "epoch": 0.03134429182616528,
      "grad_norm": 1.0130804777145386,
      "learning_rate": 9.994607009672474e-05,
      "loss": 2.412,
      "step": 174
    },
    {
      "epoch": 0.03152443143436163,
      "grad_norm": 1.0048799514770508,
      "learning_rate": 9.994541052991611e-05,
      "loss": 2.0407,
      "step": 175
    },
    {
      "epoch": 0.031704571042557986,
      "grad_norm": 0.9575945138931274,
      "learning_rate": 9.9944746956511e-05,
      "loss": 2.1829,
      "step": 176
    },
    {
      "epoch": 0.031884710650754336,
      "grad_norm": 0.9418357610702515,
      "learning_rate": 9.994407937656259e-05,
      "loss": 2.2905,
      "step": 177
    },
    {
      "epoch": 0.032064850258950686,
      "grad_norm": 0.9341018795967102,
      "learning_rate": 9.994340779012448e-05,
      "loss": 2.1959,
      "step": 178
    },
    {
      "epoch": 0.032244989867147036,
      "grad_norm": 0.9014979004859924,
      "learning_rate": 9.994273219725052e-05,
      "loss": 1.8595,
      "step": 179
    },
    {
      "epoch": 0.032425129475343394,
      "grad_norm": 0.9666259288787842,
      "learning_rate": 9.99420525979949e-05,
      "loss": 2.1748,
      "step": 180
    },
    {
      "epoch": 0.032605269083539744,
      "grad_norm": 0.9988476037979126,
      "learning_rate": 9.994136899241214e-05,
      "loss": 2.3861,
      "step": 181
    },
    {
      "epoch": 0.032785408691736094,
      "grad_norm": 0.9823298454284668,
      "learning_rate": 9.99406813805571e-05,
      "loss": 2.338,
      "step": 182
    },
    {
      "epoch": 0.032965548299932444,
      "grad_norm": 1.0125535726547241,
      "learning_rate": 9.993998976248491e-05,
      "loss": 2.151,
      "step": 183
    },
    {
      "epoch": 0.0331456879081288,
      "grad_norm": 0.9595600366592407,
      "learning_rate": 9.993929413825109e-05,
      "loss": 2.0875,
      "step": 184
    },
    {
      "epoch": 0.03332582751632515,
      "grad_norm": 0.9910955429077148,
      "learning_rate": 9.99385945079114e-05,
      "loss": 2.2156,
      "step": 185
    },
    {
      "epoch": 0.0335059671245215,
      "grad_norm": 0.9795382618904114,
      "learning_rate": 9.9937890871522e-05,
      "loss": 2.3917,
      "step": 186
    },
    {
      "epoch": 0.03368610673271786,
      "grad_norm": 1.0755410194396973,
      "learning_rate": 9.99371832291393e-05,
      "loss": 2.3763,
      "step": 187
    },
    {
      "epoch": 0.03386624634091421,
      "grad_norm": 0.968889594078064,
      "learning_rate": 9.993647158082009e-05,
      "loss": 2.2037,
      "step": 188
    },
    {
      "epoch": 0.03404638594911056,
      "grad_norm": 1.0878701210021973,
      "learning_rate": 9.993575592662145e-05,
      "loss": 2.425,
      "step": 189
    },
    {
      "epoch": 0.03422652555730691,
      "grad_norm": 1.120072364807129,
      "learning_rate": 9.993503626660079e-05,
      "loss": 2.3524,
      "step": 190
    },
    {
      "epoch": 0.03440666516550327,
      "grad_norm": 1.0771347284317017,
      "learning_rate": 9.993431260081585e-05,
      "loss": 2.4268,
      "step": 191
    },
    {
      "epoch": 0.03458680477369962,
      "grad_norm": 0.9888401627540588,
      "learning_rate": 9.993358492932467e-05,
      "loss": 1.8791,
      "step": 192
    },
    {
      "epoch": 0.03476694438189597,
      "grad_norm": 1.0292469263076782,
      "learning_rate": 9.993285325218564e-05,
      "loss": 2.4085,
      "step": 193
    },
    {
      "epoch": 0.034947083990092324,
      "grad_norm": 0.9969762563705444,
      "learning_rate": 9.993211756945744e-05,
      "loss": 2.3485,
      "step": 194
    },
    {
      "epoch": 0.035127223598288675,
      "grad_norm": 1.1246824264526367,
      "learning_rate": 9.993137788119909e-05,
      "loss": 2.4197,
      "step": 195
    },
    {
      "epoch": 0.035307363206485025,
      "grad_norm": 1.1985913515090942,
      "learning_rate": 9.993063418746991e-05,
      "loss": 2.3737,
      "step": 196
    },
    {
      "epoch": 0.035487502814681375,
      "grad_norm": 1.054831862449646,
      "learning_rate": 9.99298864883296e-05,
      "loss": 2.4413,
      "step": 197
    },
    {
      "epoch": 0.03566764242287773,
      "grad_norm": 1.0391515493392944,
      "learning_rate": 9.99291347838381e-05,
      "loss": 2.2235,
      "step": 198
    },
    {
      "epoch": 0.03584778203107408,
      "grad_norm": 1.234522819519043,
      "learning_rate": 9.992837907405574e-05,
      "loss": 2.5434,
      "step": 199
    },
    {
      "epoch": 0.03602792163927043,
      "grad_norm": 1.055237054824829,
      "learning_rate": 9.992761935904312e-05,
      "loss": 2.2512,
      "step": 200
    },
    {
      "epoch": 0.03620806124746679,
      "grad_norm": 1.484744906425476,
      "learning_rate": 9.99268556388612e-05,
      "loss": 3.1837,
      "step": 201
    },
    {
      "epoch": 0.03638820085566314,
      "grad_norm": 1.3243892192840576,
      "learning_rate": 9.992608791357123e-05,
      "loss": 2.682,
      "step": 202
    },
    {
      "epoch": 0.03656834046385949,
      "grad_norm": 1.2395107746124268,
      "learning_rate": 9.992531618323482e-05,
      "loss": 2.5428,
      "step": 203
    },
    {
      "epoch": 0.03674848007205584,
      "grad_norm": 1.220672607421875,
      "learning_rate": 9.992454044791384e-05,
      "loss": 2.7916,
      "step": 204
    },
    {
      "epoch": 0.0369286196802522,
      "grad_norm": 1.2384480237960815,
      "learning_rate": 9.992376070767056e-05,
      "loss": 2.9702,
      "step": 205
    },
    {
      "epoch": 0.03710875928844855,
      "grad_norm": 1.176816463470459,
      "learning_rate": 9.992297696256752e-05,
      "loss": 2.7499,
      "step": 206
    },
    {
      "epoch": 0.0372888988966449,
      "grad_norm": 1.1946957111358643,
      "learning_rate": 9.992218921266756e-05,
      "loss": 2.638,
      "step": 207
    },
    {
      "epoch": 0.037469038504841255,
      "grad_norm": 1.4310799837112427,
      "learning_rate": 9.992139745803393e-05,
      "loss": 2.5702,
      "step": 208
    },
    {
      "epoch": 0.037649178113037605,
      "grad_norm": 1.6133946180343628,
      "learning_rate": 9.992060169873008e-05,
      "loss": 3.4747,
      "step": 209
    },
    {
      "epoch": 0.037829317721233956,
      "grad_norm": 1.2044761180877686,
      "learning_rate": 9.991980193481989e-05,
      "loss": 2.7726,
      "step": 210
    },
    {
      "epoch": 0.038009457329430306,
      "grad_norm": 1.0784586668014526,
      "learning_rate": 9.991899816636751e-05,
      "loss": 2.0399,
      "step": 211
    },
    {
      "epoch": 0.03818959693762666,
      "grad_norm": 1.1064428091049194,
      "learning_rate": 9.991819039343741e-05,
      "loss": 2.3068,
      "step": 212
    },
    {
      "epoch": 0.03836973654582301,
      "grad_norm": 1.1013522148132324,
      "learning_rate": 9.991737861609438e-05,
      "loss": 2.0892,
      "step": 213
    },
    {
      "epoch": 0.038549876154019364,
      "grad_norm": 1.0337615013122559,
      "learning_rate": 9.991656283440356e-05,
      "loss": 2.2709,
      "step": 214
    },
    {
      "epoch": 0.038730015762215714,
      "grad_norm": 1.0415656566619873,
      "learning_rate": 9.991574304843038e-05,
      "loss": 2.0989,
      "step": 215
    },
    {
      "epoch": 0.03891015537041207,
      "grad_norm": 0.9581682085990906,
      "learning_rate": 9.991491925824062e-05,
      "loss": 2.2555,
      "step": 216
    },
    {
      "epoch": 0.03909029497860842,
      "grad_norm": 1.040091872215271,
      "learning_rate": 9.991409146390034e-05,
      "loss": 1.9537,
      "step": 217
    },
    {
      "epoch": 0.03927043458680477,
      "grad_norm": 0.9496813416481018,
      "learning_rate": 9.991325966547596e-05,
      "loss": 2.1856,
      "step": 218
    },
    {
      "epoch": 0.03945057419500113,
      "grad_norm": 0.9699931740760803,
      "learning_rate": 9.991242386303419e-05,
      "loss": 2.1698,
      "step": 219
    },
    {
      "epoch": 0.03963071380319748,
      "grad_norm": 1.062461018562317,
      "learning_rate": 9.991158405664209e-05,
      "loss": 2.158,
      "step": 220
    },
    {
      "epoch": 0.03981085341139383,
      "grad_norm": 0.9707486629486084,
      "learning_rate": 9.991074024636704e-05,
      "loss": 2.0502,
      "step": 221
    },
    {
      "epoch": 0.03999099301959018,
      "grad_norm": 1.0586658716201782,
      "learning_rate": 9.99098924322767e-05,
      "loss": 2.2018,
      "step": 222
    },
    {
      "epoch": 0.040171132627786536,
      "grad_norm": 1.1378344297409058,
      "learning_rate": 9.990904061443911e-05,
      "loss": 2.4697,
      "step": 223
    },
    {
      "epoch": 0.04035127223598289,
      "grad_norm": 1.0444320440292358,
      "learning_rate": 9.990818479292259e-05,
      "loss": 1.8717,
      "step": 224
    },
    {
      "epoch": 0.04053141184417924,
      "grad_norm": 1.0241833925247192,
      "learning_rate": 9.99073249677958e-05,
      "loss": 2.3334,
      "step": 225
    },
    {
      "epoch": 0.040711551452375594,
      "grad_norm": 1.0246652364730835,
      "learning_rate": 9.99064611391277e-05,
      "loss": 2.0989,
      "step": 226
    },
    {
      "epoch": 0.040891691060571944,
      "grad_norm": 1.0386786460876465,
      "learning_rate": 9.99055933069876e-05,
      "loss": 2.1809,
      "step": 227
    },
    {
      "epoch": 0.041071830668768294,
      "grad_norm": 1.0378458499908447,
      "learning_rate": 9.990472147144511e-05,
      "loss": 2.292,
      "step": 228
    },
    {
      "epoch": 0.041251970276964645,
      "grad_norm": 1.0621331930160522,
      "learning_rate": 9.990384563257018e-05,
      "loss": 2.1973,
      "step": 229
    },
    {
      "epoch": 0.041432109885161,
      "grad_norm": 1.0714662075042725,
      "learning_rate": 9.990296579043305e-05,
      "loss": 2.1582,
      "step": 230
    },
    {
      "epoch": 0.04161224949335735,
      "grad_norm": 1.0897587537765503,
      "learning_rate": 9.990208194510433e-05,
      "loss": 2.2179,
      "step": 231
    },
    {
      "epoch": 0.0417923891015537,
      "grad_norm": 1.0470248460769653,
      "learning_rate": 9.990119409665489e-05,
      "loss": 2.3186,
      "step": 232
    },
    {
      "epoch": 0.04197252870975006,
      "grad_norm": 1.0715101957321167,
      "learning_rate": 9.990030224515596e-05,
      "loss": 2.2274,
      "step": 233
    },
    {
      "epoch": 0.04215266831794641,
      "grad_norm": 1.1052443981170654,
      "learning_rate": 9.989940639067911e-05,
      "loss": 2.3154,
      "step": 234
    },
    {
      "epoch": 0.04233280792614276,
      "grad_norm": 1.107999563217163,
      "learning_rate": 9.989850653329617e-05,
      "loss": 2.3555,
      "step": 235
    },
    {
      "epoch": 0.04251294753433911,
      "grad_norm": 1.1312811374664307,
      "learning_rate": 9.989760267307934e-05,
      "loss": 2.5028,
      "step": 236
    },
    {
      "epoch": 0.04269308714253547,
      "grad_norm": 1.0960156917572021,
      "learning_rate": 9.989669481010113e-05,
      "loss": 2.4164,
      "step": 237
    },
    {
      "epoch": 0.04287322675073182,
      "grad_norm": 1.0567772388458252,
      "learning_rate": 9.989578294443436e-05,
      "loss": 2.0846,
      "step": 238
    },
    {
      "epoch": 0.04305336635892817,
      "grad_norm": 1.0982437133789062,
      "learning_rate": 9.989486707615219e-05,
      "loss": 2.322,
      "step": 239
    },
    {
      "epoch": 0.043233505967124525,
      "grad_norm": 1.1053307056427002,
      "learning_rate": 9.98939472053281e-05,
      "loss": 2.4231,
      "step": 240
    },
    {
      "epoch": 0.043413645575320875,
      "grad_norm": 1.0721138715744019,
      "learning_rate": 9.989302333203585e-05,
      "loss": 2.3685,
      "step": 241
    },
    {
      "epoch": 0.043593785183517225,
      "grad_norm": 1.0529550313949585,
      "learning_rate": 9.989209545634957e-05,
      "loss": 1.8775,
      "step": 242
    },
    {
      "epoch": 0.043773924791713575,
      "grad_norm": 1.1814721822738647,
      "learning_rate": 9.98911635783437e-05,
      "loss": 2.475,
      "step": 243
    },
    {
      "epoch": 0.04395406439990993,
      "grad_norm": 1.2148950099945068,
      "learning_rate": 9.989022769809299e-05,
      "loss": 2.5961,
      "step": 244
    },
    {
      "epoch": 0.04413420400810628,
      "grad_norm": 1.1676557064056396,
      "learning_rate": 9.98892878156725e-05,
      "loss": 2.0822,
      "step": 245
    },
    {
      "epoch": 0.04431434361630263,
      "grad_norm": 1.2208837270736694,
      "learning_rate": 9.988834393115767e-05,
      "loss": 2.3525,
      "step": 246
    },
    {
      "epoch": 0.04449448322449899,
      "grad_norm": 1.1107949018478394,
      "learning_rate": 9.988739604462415e-05,
      "loss": 1.9607,
      "step": 247
    },
    {
      "epoch": 0.04467462283269534,
      "grad_norm": 1.096728801727295,
      "learning_rate": 9.988644415614806e-05,
      "loss": 2.2089,
      "step": 248
    },
    {
      "epoch": 0.04485476244089169,
      "grad_norm": 1.2097132205963135,
      "learning_rate": 9.98854882658057e-05,
      "loss": 2.1281,
      "step": 249
    },
    {
      "epoch": 0.04503490204908804,
      "grad_norm": 1.3903559446334839,
      "learning_rate": 9.988452837367377e-05,
      "loss": 2.2447,
      "step": 250
    },
    {
      "epoch": 0.0452150416572844,
      "grad_norm": 1.250702142715454,
      "learning_rate": 9.988356447982927e-05,
      "loss": 2.3893,
      "step": 251
    },
    {
      "epoch": 0.04539518126548075,
      "grad_norm": 1.2847386598587036,
      "learning_rate": 9.988259658434952e-05,
      "loss": 2.7229,
      "step": 252
    },
    {
      "epoch": 0.0455753208736771,
      "grad_norm": 1.2485649585723877,
      "learning_rate": 9.988162468731219e-05,
      "loss": 2.5685,
      "step": 253
    },
    {
      "epoch": 0.04575546048187345,
      "grad_norm": 1.2486999034881592,
      "learning_rate": 9.98806487887952e-05,
      "loss": 2.6679,
      "step": 254
    },
    {
      "epoch": 0.045935600090069806,
      "grad_norm": 1.3331091403961182,
      "learning_rate": 9.987966888887687e-05,
      "loss": 2.7473,
      "step": 255
    },
    {
      "epoch": 0.046115739698266156,
      "grad_norm": 1.2322219610214233,
      "learning_rate": 9.987868498763581e-05,
      "loss": 2.6721,
      "step": 256
    },
    {
      "epoch": 0.046295879306462506,
      "grad_norm": 1.2584573030471802,
      "learning_rate": 9.987769708515091e-05,
      "loss": 2.5292,
      "step": 257
    },
    {
      "epoch": 0.04647601891465886,
      "grad_norm": 1.2774226665496826,
      "learning_rate": 9.987670518150147e-05,
      "loss": 2.6983,
      "step": 258
    },
    {
      "epoch": 0.046656158522855214,
      "grad_norm": 1.4696308374404907,
      "learning_rate": 9.987570927676701e-05,
      "loss": 2.7173,
      "step": 259
    },
    {
      "epoch": 0.046836298131051564,
      "grad_norm": 1.8001126050949097,
      "learning_rate": 9.987470937102747e-05,
      "loss": 3.094,
      "step": 260
    },
    {
      "epoch": 0.047016437739247914,
      "grad_norm": 1.1973381042480469,
      "learning_rate": 9.9873705464363e-05,
      "loss": 2.3736,
      "step": 261
    },
    {
      "epoch": 0.04719657734744427,
      "grad_norm": 1.4078965187072754,
      "learning_rate": 9.98726975568542e-05,
      "loss": 2.316,
      "step": 262
    },
    {
      "epoch": 0.04737671695564062,
      "grad_norm": 1.417547583580017,
      "learning_rate": 9.98716856485819e-05,
      "loss": 2.7383,
      "step": 263
    },
    {
      "epoch": 0.04755685656383697,
      "grad_norm": 1.435121774673462,
      "learning_rate": 9.987066973962726e-05,
      "loss": 2.0731,
      "step": 264
    },
    {
      "epoch": 0.04773699617203333,
      "grad_norm": 1.2506507635116577,
      "learning_rate": 9.98696498300718e-05,
      "loss": 2.1467,
      "step": 265
    },
    {
      "epoch": 0.04791713578022968,
      "grad_norm": 1.2805836200714111,
      "learning_rate": 9.986862591999728e-05,
      "loss": 2.2571,
      "step": 266
    },
    {
      "epoch": 0.04809727538842603,
      "grad_norm": 1.1107738018035889,
      "learning_rate": 9.986759800948591e-05,
      "loss": 1.9794,
      "step": 267
    },
    {
      "epoch": 0.04827741499662238,
      "grad_norm": 1.0665452480316162,
      "learning_rate": 9.986656609862011e-05,
      "loss": 1.9837,
      "step": 268
    },
    {
      "epoch": 0.04845755460481874,
      "grad_norm": 0.9659749269485474,
      "learning_rate": 9.986553018748266e-05,
      "loss": 2.0036,
      "step": 269
    },
    {
      "epoch": 0.04863769421301509,
      "grad_norm": 0.935899555683136,
      "learning_rate": 9.986449027615667e-05,
      "loss": 1.9917,
      "step": 270
    },
    {
      "epoch": 0.04881783382121144,
      "grad_norm": 1.054198145866394,
      "learning_rate": 9.986344636472558e-05,
      "loss": 1.9455,
      "step": 271
    },
    {
      "epoch": 0.048997973429407794,
      "grad_norm": 1.116992473602295,
      "learning_rate": 9.986239845327308e-05,
      "loss": 2.13,
      "step": 272
    },
    {
      "epoch": 0.049178113037604144,
      "grad_norm": 1.214431643486023,
      "learning_rate": 9.986134654188328e-05,
      "loss": 2.0619,
      "step": 273
    },
    {
      "epoch": 0.049358252645800495,
      "grad_norm": 1.1717039346694946,
      "learning_rate": 9.986029063064053e-05,
      "loss": 1.9885,
      "step": 274
    },
    {
      "epoch": 0.049538392253996845,
      "grad_norm": 1.2670177221298218,
      "learning_rate": 9.985923071962957e-05,
      "loss": 2.1736,
      "step": 275
    },
    {
      "epoch": 0.0497185318621932,
      "grad_norm": 1.127020239830017,
      "learning_rate": 9.98581668089354e-05,
      "loss": 2.1314,
      "step": 276
    },
    {
      "epoch": 0.04989867147038955,
      "grad_norm": 1.0725072622299194,
      "learning_rate": 9.985709889864337e-05,
      "loss": 2.0381,
      "step": 277
    },
    {
      "epoch": 0.0500788110785859,
      "grad_norm": 1.148308277130127,
      "learning_rate": 9.985602698883916e-05,
      "loss": 2.1644,
      "step": 278
    },
    {
      "epoch": 0.05025895068678226,
      "grad_norm": 1.0978087186813354,
      "learning_rate": 9.985495107960876e-05,
      "loss": 2.0789,
      "step": 279
    },
    {
      "epoch": 0.05043909029497861,
      "grad_norm": 1.1935482025146484,
      "learning_rate": 9.985387117103843e-05,
      "loss": 2.2864,
      "step": 280
    },
    {
      "epoch": 0.05061922990317496,
      "grad_norm": 1.1536046266555786,
      "learning_rate": 9.985278726321487e-05,
      "loss": 2.2444,
      "step": 281
    },
    {
      "epoch": 0.05079936951137131,
      "grad_norm": 1.1859569549560547,
      "learning_rate": 9.985169935622498e-05,
      "loss": 2.1967,
      "step": 282
    },
    {
      "epoch": 0.05097950911956767,
      "grad_norm": 1.2108591794967651,
      "learning_rate": 9.985060745015607e-05,
      "loss": 2.309,
      "step": 283
    },
    {
      "epoch": 0.05115964872776402,
      "grad_norm": 1.1978607177734375,
      "learning_rate": 9.98495115450957e-05,
      "loss": 2.4844,
      "step": 284
    },
    {
      "epoch": 0.05133978833596037,
      "grad_norm": 1.105621337890625,
      "learning_rate": 9.98484116411318e-05,
      "loss": 2.3288,
      "step": 285
    },
    {
      "epoch": 0.05151992794415672,
      "grad_norm": 1.1028847694396973,
      "learning_rate": 9.984730773835258e-05,
      "loss": 2.1256,
      "step": 286
    },
    {
      "epoch": 0.051700067552353075,
      "grad_norm": 1.182299017906189,
      "learning_rate": 9.984619983684662e-05,
      "loss": 2.4203,
      "step": 287
    },
    {
      "epoch": 0.051880207160549426,
      "grad_norm": 1.120054006576538,
      "learning_rate": 9.984508793670279e-05,
      "loss": 2.3129,
      "step": 288
    },
    {
      "epoch": 0.052060346768745776,
      "grad_norm": 1.0880308151245117,
      "learning_rate": 9.98439720380103e-05,
      "loss": 2.2993,
      "step": 289
    },
    {
      "epoch": 0.05224048637694213,
      "grad_norm": 1.1723957061767578,
      "learning_rate": 9.984285214085865e-05,
      "loss": 2.4264,
      "step": 290
    },
    {
      "epoch": 0.05242062598513848,
      "grad_norm": 1.0767836570739746,
      "learning_rate": 9.984172824533766e-05,
      "loss": 1.9417,
      "step": 291
    },
    {
      "epoch": 0.05260076559333483,
      "grad_norm": 1.090864658355713,
      "learning_rate": 9.98406003515375e-05,
      "loss": 2.2895,
      "step": 292
    },
    {
      "epoch": 0.052780905201531184,
      "grad_norm": 1.0966577529907227,
      "learning_rate": 9.983946845954868e-05,
      "loss": 2.2194,
      "step": 293
    },
    {
      "epoch": 0.05296104480972754,
      "grad_norm": 1.108288049697876,
      "learning_rate": 9.983833256946198e-05,
      "loss": 2.0902,
      "step": 294
    },
    {
      "epoch": 0.05314118441792389,
      "grad_norm": 1.1045280694961548,
      "learning_rate": 9.98371926813685e-05,
      "loss": 2.382,
      "step": 295
    },
    {
      "epoch": 0.05332132402612024,
      "grad_norm": 1.0808862447738647,
      "learning_rate": 9.98360487953597e-05,
      "loss": 2.1628,
      "step": 296
    },
    {
      "epoch": 0.0535014636343166,
      "grad_norm": 1.1434890031814575,
      "learning_rate": 9.983490091152734e-05,
      "loss": 2.1914,
      "step": 297
    },
    {
      "epoch": 0.05368160324251295,
      "grad_norm": 1.0393102169036865,
      "learning_rate": 9.98337490299635e-05,
      "loss": 2.069,
      "step": 298
    },
    {
      "epoch": 0.0538617428507093,
      "grad_norm": 1.1679807901382446,
      "learning_rate": 9.98325931507606e-05,
      "loss": 2.1102,
      "step": 299
    },
    {
      "epoch": 0.05404188245890565,
      "grad_norm": 1.3268619775772095,
      "learning_rate": 9.983143327401134e-05,
      "loss": 2.2969,
      "step": 300
    },
    {
      "epoch": 0.054222022067102006,
      "grad_norm": 1.3637564182281494,
      "learning_rate": 9.983026939980877e-05,
      "loss": 2.6549,
      "step": 301
    },
    {
      "epoch": 0.054402161675298356,
      "grad_norm": 1.292595386505127,
      "learning_rate": 9.982910152824627e-05,
      "loss": 2.7285,
      "step": 302
    },
    {
      "epoch": 0.05458230128349471,
      "grad_norm": 1.1744555234909058,
      "learning_rate": 9.982792965941751e-05,
      "loss": 2.5136,
      "step": 303
    },
    {
      "epoch": 0.054762440891691064,
      "grad_norm": 1.2590298652648926,
      "learning_rate": 9.982675379341651e-05,
      "loss": 2.6739,
      "step": 304
    },
    {
      "epoch": 0.054942580499887414,
      "grad_norm": 1.2690422534942627,
      "learning_rate": 9.982557393033758e-05,
      "loss": 2.4736,
      "step": 305
    },
    {
      "epoch": 0.055122720108083764,
      "grad_norm": 1.2975382804870605,
      "learning_rate": 9.982439007027538e-05,
      "loss": 2.6993,
      "step": 306
    },
    {
      "epoch": 0.055302859716280114,
      "grad_norm": 1.5365309715270996,
      "learning_rate": 9.982320221332488e-05,
      "loss": 3.1287,
      "step": 307
    },
    {
      "epoch": 0.05548299932447647,
      "grad_norm": 1.339965581893921,
      "learning_rate": 9.982201035958138e-05,
      "loss": 2.6284,
      "step": 308
    },
    {
      "epoch": 0.05566313893267282,
      "grad_norm": 1.4245731830596924,
      "learning_rate": 9.982081450914046e-05,
      "loss": 2.4764,
      "step": 309
    },
    {
      "epoch": 0.05584327854086917,
      "grad_norm": 1.81796133518219,
      "learning_rate": 9.981961466209808e-05,
      "loss": 3.0644,
      "step": 310
    },
    {
      "epoch": 0.05602341814906553,
      "grad_norm": 1.4154448509216309,
      "learning_rate": 9.981841081855046e-05,
      "loss": 2.5412,
      "step": 311
    },
    {
      "epoch": 0.05620355775726188,
      "grad_norm": 1.485663890838623,
      "learning_rate": 9.98172029785942e-05,
      "loss": 2.1247,
      "step": 312
    },
    {
      "epoch": 0.05638369736545823,
      "grad_norm": 1.418679118156433,
      "learning_rate": 9.981599114232618e-05,
      "loss": 2.0103,
      "step": 313
    },
    {
      "epoch": 0.05656383697365458,
      "grad_norm": 1.4113951921463013,
      "learning_rate": 9.981477530984362e-05,
      "loss": 2.2682,
      "step": 314
    },
    {
      "epoch": 0.05674397658185094,
      "grad_norm": 1.078120231628418,
      "learning_rate": 9.981355548124405e-05,
      "loss": 1.8057,
      "step": 315
    },
    {
      "epoch": 0.05692411619004729,
      "grad_norm": 1.241740345954895,
      "learning_rate": 9.981233165662531e-05,
      "loss": 2.1231,
      "step": 316
    },
    {
      "epoch": 0.05710425579824364,
      "grad_norm": 1.234967827796936,
      "learning_rate": 9.98111038360856e-05,
      "loss": 1.99,
      "step": 317
    },
    {
      "epoch": 0.05728439540643999,
      "grad_norm": 1.0713143348693848,
      "learning_rate": 9.980987201972338e-05,
      "loss": 1.9586,
      "step": 318
    },
    {
      "epoch": 0.057464535014636345,
      "grad_norm": 1.058562159538269,
      "learning_rate": 9.980863620763752e-05,
      "loss": 1.966,
      "step": 319
    },
    {
      "epoch": 0.057644674622832695,
      "grad_norm": 1.1675026416778564,
      "learning_rate": 9.980739639992711e-05,
      "loss": 2.3292,
      "step": 320
    },
    {
      "epoch": 0.057824814231029045,
      "grad_norm": 1.0847362279891968,
      "learning_rate": 9.980615259669162e-05,
      "loss": 2.0149,
      "step": 321
    },
    {
      "epoch": 0.0580049538392254,
      "grad_norm": 1.1505693197250366,
      "learning_rate": 9.980490479803082e-05,
      "loss": 2.0683,
      "step": 322
    },
    {
      "epoch": 0.05818509344742175,
      "grad_norm": 1.2553009986877441,
      "learning_rate": 9.980365300404483e-05,
      "loss": 1.9848,
      "step": 323
    },
    {
      "epoch": 0.0583652330556181,
      "grad_norm": 1.2744611501693726,
      "learning_rate": 9.980239721483404e-05,
      "loss": 2.1334,
      "step": 324
    },
    {
      "epoch": 0.05854537266381445,
      "grad_norm": 1.248488187789917,
      "learning_rate": 9.980113743049922e-05,
      "loss": 2.1967,
      "step": 325
    },
    {
      "epoch": 0.05872551227201081,
      "grad_norm": 1.2903262376785278,
      "learning_rate": 9.97998736511414e-05,
      "loss": 2.1005,
      "step": 326
    },
    {
      "epoch": 0.05890565188020716,
      "grad_norm": 1.1605138778686523,
      "learning_rate": 9.979860587686198e-05,
      "loss": 2.2164,
      "step": 327
    },
    {
      "epoch": 0.05908579148840351,
      "grad_norm": 1.053455114364624,
      "learning_rate": 9.979733410776266e-05,
      "loss": 1.8878,
      "step": 328
    },
    {
      "epoch": 0.05926593109659987,
      "grad_norm": 1.1907683610916138,
      "learning_rate": 9.979605834394544e-05,
      "loss": 2.0901,
      "step": 329
    },
    {
      "epoch": 0.05944607070479622,
      "grad_norm": 1.1980634927749634,
      "learning_rate": 9.979477858551267e-05,
      "loss": 2.1962,
      "step": 330
    },
    {
      "epoch": 0.05962621031299257,
      "grad_norm": 1.1212539672851562,
      "learning_rate": 9.979349483256704e-05,
      "loss": 1.8947,
      "step": 331
    },
    {
      "epoch": 0.05980634992118892,
      "grad_norm": 1.1518847942352295,
      "learning_rate": 9.979220708521148e-05,
      "loss": 2.1186,
      "step": 332
    },
    {
      "epoch": 0.059986489529385276,
      "grad_norm": 1.1192442178726196,
      "learning_rate": 9.979091534354933e-05,
      "loss": 1.8505,
      "step": 333
    },
    {
      "epoch": 0.060166629137581626,
      "grad_norm": 1.2988682985305786,
      "learning_rate": 9.978961960768421e-05,
      "loss": 2.2051,
      "step": 334
    },
    {
      "epoch": 0.060346768745777976,
      "grad_norm": 1.1861850023269653,
      "learning_rate": 9.978831987772006e-05,
      "loss": 2.1588,
      "step": 335
    },
    {
      "epoch": 0.06052690835397433,
      "grad_norm": 1.2539981603622437,
      "learning_rate": 9.978701615376113e-05,
      "loss": 2.17,
      "step": 336
    },
    {
      "epoch": 0.060707047962170683,
      "grad_norm": 1.2473936080932617,
      "learning_rate": 9.978570843591202e-05,
      "loss": 2.4724,
      "step": 337
    },
    {
      "epoch": 0.060887187570367034,
      "grad_norm": 1.2099696397781372,
      "learning_rate": 9.978439672427762e-05,
      "loss": 1.9898,
      "step": 338
    },
    {
      "epoch": 0.061067327178563384,
      "grad_norm": 1.1351348161697388,
      "learning_rate": 9.978308101896318e-05,
      "loss": 2.1373,
      "step": 339
    },
    {
      "epoch": 0.06124746678675974,
      "grad_norm": 1.206483006477356,
      "learning_rate": 9.978176132007422e-05,
      "loss": 2.0748,
      "step": 340
    },
    {
      "epoch": 0.06142760639495609,
      "grad_norm": 1.1359015703201294,
      "learning_rate": 9.97804376277166e-05,
      "loss": 1.9103,
      "step": 341
    },
    {
      "epoch": 0.06160774600315244,
      "grad_norm": 1.3475091457366943,
      "learning_rate": 9.977910994199655e-05,
      "loss": 2.2672,
      "step": 342
    },
    {
      "epoch": 0.0617878856113488,
      "grad_norm": 1.226010799407959,
      "learning_rate": 9.977777826302053e-05,
      "loss": 2.0521,
      "step": 343
    },
    {
      "epoch": 0.06196802521954515,
      "grad_norm": 1.1900585889816284,
      "learning_rate": 9.977644259089539e-05,
      "loss": 2.1845,
      "step": 344
    },
    {
      "epoch": 0.0621481648277415,
      "grad_norm": 1.2761820554733276,
      "learning_rate": 9.977510292572826e-05,
      "loss": 2.2941,
      "step": 345
    },
    {
      "epoch": 0.06232830443593785,
      "grad_norm": 1.282155990600586,
      "learning_rate": 9.977375926762662e-05,
      "loss": 2.4086,
      "step": 346
    },
    {
      "epoch": 0.0625084440441342,
      "grad_norm": 1.2463797330856323,
      "learning_rate": 9.977241161669825e-05,
      "loss": 2.3348,
      "step": 347
    },
    {
      "epoch": 0.06268858365233056,
      "grad_norm": 1.20982825756073,
      "learning_rate": 9.97710599730513e-05,
      "loss": 1.9883,
      "step": 348
    },
    {
      "epoch": 0.06286872326052691,
      "grad_norm": 1.2739825248718262,
      "learning_rate": 9.976970433679413e-05,
      "loss": 2.2029,
      "step": 349
    },
    {
      "epoch": 0.06304886286872326,
      "grad_norm": 1.2550883293151855,
      "learning_rate": 9.976834470803552e-05,
      "loss": 2.2877,
      "step": 350
    },
    {
      "epoch": 0.06322900247691961,
      "grad_norm": 1.5917472839355469,
      "learning_rate": 9.976698108688454e-05,
      "loss": 3.0127,
      "step": 351
    },
    {
      "epoch": 0.06340914208511597,
      "grad_norm": 1.3110361099243164,
      "learning_rate": 9.976561347345059e-05,
      "loss": 2.3605,
      "step": 352
    },
    {
      "epoch": 0.06358928169331232,
      "grad_norm": 1.384914755821228,
      "learning_rate": 9.976424186784336e-05,
      "loss": 2.8826,
      "step": 353
    },
    {
      "epoch": 0.06376942130150867,
      "grad_norm": 1.401537299156189,
      "learning_rate": 9.976286627017289e-05,
      "loss": 2.9413,
      "step": 354
    },
    {
      "epoch": 0.06394956090970502,
      "grad_norm": 1.298903465270996,
      "learning_rate": 9.976148668054951e-05,
      "loss": 2.6467,
      "step": 355
    },
    {
      "epoch": 0.06412970051790137,
      "grad_norm": 1.4152841567993164,
      "learning_rate": 9.976010309908392e-05,
      "loss": 2.3555,
      "step": 356
    },
    {
      "epoch": 0.06430984012609772,
      "grad_norm": 1.5927162170410156,
      "learning_rate": 9.975871552588709e-05,
      "loss": 3.0816,
      "step": 357
    },
    {
      "epoch": 0.06448997973429407,
      "grad_norm": 1.6605229377746582,
      "learning_rate": 9.975732396107035e-05,
      "loss": 3.0563,
      "step": 358
    },
    {
      "epoch": 0.06467011934249044,
      "grad_norm": 1.9879968166351318,
      "learning_rate": 9.97559284047453e-05,
      "loss": 2.9739,
      "step": 359
    },
    {
      "epoch": 0.06485025895068679,
      "grad_norm": 1.1902703046798706,
      "learning_rate": 9.975452885702392e-05,
      "loss": 2.1506,
      "step": 360
    },
    {
      "epoch": 0.06503039855888314,
      "grad_norm": 1.2667981386184692,
      "learning_rate": 9.975312531801849e-05,
      "loss": 1.9611,
      "step": 361
    },
    {
      "epoch": 0.06521053816707949,
      "grad_norm": 1.1608258485794067,
      "learning_rate": 9.975171778784154e-05,
      "loss": 1.9572,
      "step": 362
    },
    {
      "epoch": 0.06539067777527584,
      "grad_norm": 1.2240402698516846,
      "learning_rate": 9.975030626660604e-05,
      "loss": 2.0274,
      "step": 363
    },
    {
      "epoch": 0.06557081738347219,
      "grad_norm": 1.2387436628341675,
      "learning_rate": 9.974889075442521e-05,
      "loss": 2.0846,
      "step": 364
    },
    {
      "epoch": 0.06575095699166854,
      "grad_norm": 1.1588557958602905,
      "learning_rate": 9.97474712514126e-05,
      "loss": 2.1116,
      "step": 365
    },
    {
      "epoch": 0.06593109659986489,
      "grad_norm": 1.1860038042068481,
      "learning_rate": 9.974604775768206e-05,
      "loss": 2.069,
      "step": 366
    },
    {
      "epoch": 0.06611123620806125,
      "grad_norm": 1.1692122220993042,
      "learning_rate": 9.97446202733478e-05,
      "loss": 2.1168,
      "step": 367
    },
    {
      "epoch": 0.0662913758162576,
      "grad_norm": 1.1010984182357788,
      "learning_rate": 9.974318879852435e-05,
      "loss": 1.8769,
      "step": 368
    },
    {
      "epoch": 0.06647151542445395,
      "grad_norm": 1.1738924980163574,
      "learning_rate": 9.97417533333265e-05,
      "loss": 2.0203,
      "step": 369
    },
    {
      "epoch": 0.0666516550326503,
      "grad_norm": 1.1776379346847534,
      "learning_rate": 9.974031387786945e-05,
      "loss": 2.0117,
      "step": 370
    },
    {
      "epoch": 0.06683179464084665,
      "grad_norm": 1.176415205001831,
      "learning_rate": 9.973887043226863e-05,
      "loss": 1.9938,
      "step": 371
    },
    {
      "epoch": 0.067011934249043,
      "grad_norm": 1.03434157371521,
      "learning_rate": 9.973742299663986e-05,
      "loss": 1.9566,
      "step": 372
    },
    {
      "epoch": 0.06719207385723935,
      "grad_norm": 1.062901258468628,
      "learning_rate": 9.973597157109924e-05,
      "loss": 1.6733,
      "step": 373
    },
    {
      "epoch": 0.06737221346543572,
      "grad_norm": 1.09894859790802,
      "learning_rate": 9.973451615576321e-05,
      "loss": 1.6909,
      "step": 374
    },
    {
      "epoch": 0.06755235307363207,
      "grad_norm": 1.1642646789550781,
      "learning_rate": 9.973305675074851e-05,
      "loss": 2.1543,
      "step": 375
    },
    {
      "epoch": 0.06773249268182842,
      "grad_norm": 1.2013028860092163,
      "learning_rate": 9.973159335617223e-05,
      "loss": 2.0974,
      "step": 376
    },
    {
      "epoch": 0.06791263229002477,
      "grad_norm": 1.1819990873336792,
      "learning_rate": 9.973012597215175e-05,
      "loss": 2.1242,
      "step": 377
    },
    {
      "epoch": 0.06809277189822112,
      "grad_norm": 1.2293394804000854,
      "learning_rate": 9.972865459880479e-05,
      "loss": 2.0918,
      "step": 378
    },
    {
      "epoch": 0.06827291150641747,
      "grad_norm": 1.1342253684997559,
      "learning_rate": 9.972717923624938e-05,
      "loss": 2.0022,
      "step": 379
    },
    {
      "epoch": 0.06845305111461382,
      "grad_norm": 1.1995207071304321,
      "learning_rate": 9.972569988460387e-05,
      "loss": 2.3192,
      "step": 380
    },
    {
      "epoch": 0.06863319072281018,
      "grad_norm": 1.2198859453201294,
      "learning_rate": 9.972421654398693e-05,
      "loss": 2.1539,
      "step": 381
    },
    {
      "epoch": 0.06881333033100653,
      "grad_norm": 1.200929880142212,
      "learning_rate": 9.972272921451758e-05,
      "loss": 2.1581,
      "step": 382
    },
    {
      "epoch": 0.06899346993920288,
      "grad_norm": 1.1735786199569702,
      "learning_rate": 9.972123789631509e-05,
      "loss": 1.9714,
      "step": 383
    },
    {
      "epoch": 0.06917360954739923,
      "grad_norm": 1.1791009902954102,
      "learning_rate": 9.971974258949913e-05,
      "loss": 2.3169,
      "step": 384
    },
    {
      "epoch": 0.06935374915559558,
      "grad_norm": 1.164649248123169,
      "learning_rate": 9.971824329418961e-05,
      "loss": 2.0622,
      "step": 385
    },
    {
      "epoch": 0.06953388876379193,
      "grad_norm": 1.1468734741210938,
      "learning_rate": 9.971674001050686e-05,
      "loss": 1.9853,
      "step": 386
    },
    {
      "epoch": 0.06971402837198828,
      "grad_norm": 1.3305964469909668,
      "learning_rate": 9.971523273857145e-05,
      "loss": 2.1466,
      "step": 387
    },
    {
      "epoch": 0.06989416798018465,
      "grad_norm": 1.1789056062698364,
      "learning_rate": 9.971372147850426e-05,
      "loss": 2.2511,
      "step": 388
    },
    {
      "epoch": 0.070074307588381,
      "grad_norm": 1.2829426527023315,
      "learning_rate": 9.971220623042655e-05,
      "loss": 2.2993,
      "step": 389
    },
    {
      "epoch": 0.07025444719657735,
      "grad_norm": 1.1618152856826782,
      "learning_rate": 9.971068699445987e-05,
      "loss": 1.9169,
      "step": 390
    },
    {
      "epoch": 0.0704345868047737,
      "grad_norm": 1.3183369636535645,
      "learning_rate": 9.97091637707261e-05,
      "loss": 2.2636,
      "step": 391
    },
    {
      "epoch": 0.07061472641297005,
      "grad_norm": 1.3134520053863525,
      "learning_rate": 9.970763655934741e-05,
      "loss": 2.1004,
      "step": 392
    },
    {
      "epoch": 0.0707948660211664,
      "grad_norm": 1.2930353879928589,
      "learning_rate": 9.970610536044634e-05,
      "loss": 2.2754,
      "step": 393
    },
    {
      "epoch": 0.07097500562936275,
      "grad_norm": 1.212755799293518,
      "learning_rate": 9.970457017414572e-05,
      "loss": 2.0809,
      "step": 394
    },
    {
      "epoch": 0.07115514523755911,
      "grad_norm": 1.222426414489746,
      "learning_rate": 9.970303100056867e-05,
      "loss": 2.0221,
      "step": 395
    },
    {
      "epoch": 0.07133528484575546,
      "grad_norm": 1.2207961082458496,
      "learning_rate": 9.970148783983869e-05,
      "loss": 1.9381,
      "step": 396
    },
    {
      "epoch": 0.07151542445395181,
      "grad_norm": 1.205553412437439,
      "learning_rate": 9.969994069207958e-05,
      "loss": 2.045,
      "step": 397
    },
    {
      "epoch": 0.07169556406214816,
      "grad_norm": 1.2328301668167114,
      "learning_rate": 9.969838955741541e-05,
      "loss": 1.9693,
      "step": 398
    },
    {
      "epoch": 0.07187570367034452,
      "grad_norm": 1.2753703594207764,
      "learning_rate": 9.969683443597064e-05,
      "loss": 1.8625,
      "step": 399
    },
    {
      "epoch": 0.07205584327854087,
      "grad_norm": 1.1866049766540527,
      "learning_rate": 9.969527532787003e-05,
      "loss": 2.0762,
      "step": 400
    },
    {
      "epoch": 0.07223598288673722,
      "grad_norm": 1.3906890153884888,
      "learning_rate": 9.969371223323865e-05,
      "loss": 2.7344,
      "step": 401
    },
    {
      "epoch": 0.07241612249493358,
      "grad_norm": 1.3052465915679932,
      "learning_rate": 9.969214515220186e-05,
      "loss": 2.5495,
      "step": 402
    },
    {
      "epoch": 0.07259626210312993,
      "grad_norm": 1.3904844522476196,
      "learning_rate": 9.969057408488539e-05,
      "loss": 2.8292,
      "step": 403
    },
    {
      "epoch": 0.07277640171132628,
      "grad_norm": 1.2776987552642822,
      "learning_rate": 9.968899903141526e-05,
      "loss": 2.4234,
      "step": 404
    },
    {
      "epoch": 0.07295654131952263,
      "grad_norm": 1.3141162395477295,
      "learning_rate": 9.968741999191787e-05,
      "loss": 2.4546,
      "step": 405
    },
    {
      "epoch": 0.07313668092771898,
      "grad_norm": 1.2972636222839355,
      "learning_rate": 9.96858369665198e-05,
      "loss": 2.4995,
      "step": 406
    },
    {
      "epoch": 0.07331682053591533,
      "grad_norm": 1.5233914852142334,
      "learning_rate": 9.968424995534813e-05,
      "loss": 2.5775,
      "step": 407
    },
    {
      "epoch": 0.07349696014411168,
      "grad_norm": 1.6031200885772705,
      "learning_rate": 9.96826589585301e-05,
      "loss": 2.622,
      "step": 408
    },
    {
      "epoch": 0.07367709975230805,
      "grad_norm": 2.019437789916992,
      "learning_rate": 9.968106397619339e-05,
      "loss": 3.0999,
      "step": 409
    },
    {
      "epoch": 0.0738572393605044,
      "grad_norm": 1.4470237493515015,
      "learning_rate": 9.967946500846592e-05,
      "loss": 2.7362,
      "step": 410
    },
    {
      "epoch": 0.07403737896870075,
      "grad_norm": 1.2874963283538818,
      "learning_rate": 9.967786205547595e-05,
      "loss": 2.2517,
      "step": 411
    },
    {
      "epoch": 0.0742175185768971,
      "grad_norm": 1.3089977502822876,
      "learning_rate": 9.96762551173521e-05,
      "loss": 2.1858,
      "step": 412
    },
    {
      "epoch": 0.07439765818509345,
      "grad_norm": 1.2105944156646729,
      "learning_rate": 9.967464419422327e-05,
      "loss": 1.9773,
      "step": 413
    },
    {
      "epoch": 0.0745777977932898,
      "grad_norm": 1.3345972299575806,
      "learning_rate": 9.967302928621867e-05,
      "loss": 2.2248,
      "step": 414
    },
    {
      "epoch": 0.07475793740148615,
      "grad_norm": 1.153934359550476,
      "learning_rate": 9.967141039346784e-05,
      "loss": 2.1041,
      "step": 415
    },
    {
      "epoch": 0.07493807700968251,
      "grad_norm": 1.1744452714920044,
      "learning_rate": 9.966978751610068e-05,
      "loss": 2.0261,
      "step": 416
    },
    {
      "epoch": 0.07511821661787886,
      "grad_norm": 1.134575366973877,
      "learning_rate": 9.966816065424736e-05,
      "loss": 1.9197,
      "step": 417
    },
    {
      "epoch": 0.07529835622607521,
      "grad_norm": 1.2032464742660522,
      "learning_rate": 9.966652980803836e-05,
      "loss": 1.8453,
      "step": 418
    },
    {
      "epoch": 0.07547849583427156,
      "grad_norm": 1.1819320917129517,
      "learning_rate": 9.966489497760456e-05,
      "loss": 1.9449,
      "step": 419
    },
    {
      "epoch": 0.07565863544246791,
      "grad_norm": 1.150172472000122,
      "learning_rate": 9.966325616307706e-05,
      "loss": 2.0224,
      "step": 420
    },
    {
      "epoch": 0.07583877505066426,
      "grad_norm": 1.143376111984253,
      "learning_rate": 9.966161336458734e-05,
      "loss": 1.9967,
      "step": 421
    },
    {
      "epoch": 0.07601891465886061,
      "grad_norm": 1.153085470199585,
      "learning_rate": 9.965996658226719e-05,
      "loss": 1.816,
      "step": 422
    },
    {
      "epoch": 0.07619905426705698,
      "grad_norm": 1.2253894805908203,
      "learning_rate": 9.965831581624871e-05,
      "loss": 1.9349,
      "step": 423
    },
    {
      "epoch": 0.07637919387525333,
      "grad_norm": 1.3817367553710938,
      "learning_rate": 9.965666106666432e-05,
      "loss": 1.996,
      "step": 424
    },
    {
      "epoch": 0.07655933348344968,
      "grad_norm": 1.1546779870986938,
      "learning_rate": 9.965500233364678e-05,
      "loss": 2.0808,
      "step": 425
    },
    {
      "epoch": 0.07673947309164603,
      "grad_norm": 1.1899226903915405,
      "learning_rate": 9.965333961732912e-05,
      "loss": 1.8567,
      "step": 426
    },
    {
      "epoch": 0.07691961269984238,
      "grad_norm": 1.3877899646759033,
      "learning_rate": 9.965167291784474e-05,
      "loss": 1.7732,
      "step": 427
    },
    {
      "epoch": 0.07709975230803873,
      "grad_norm": 1.4410600662231445,
      "learning_rate": 9.965000223532736e-05,
      "loss": 2.4459,
      "step": 428
    },
    {
      "epoch": 0.07727989191623508,
      "grad_norm": 1.2743172645568848,
      "learning_rate": 9.964832756991097e-05,
      "loss": 2.0879,
      "step": 429
    },
    {
      "epoch": 0.07746003152443143,
      "grad_norm": 1.1992241144180298,
      "learning_rate": 9.964664892172992e-05,
      "loss": 2.0258,
      "step": 430
    },
    {
      "epoch": 0.07764017113262779,
      "grad_norm": 1.101402759552002,
      "learning_rate": 9.96449662909189e-05,
      "loss": 2.0255,
      "step": 431
    },
    {
      "epoch": 0.07782031074082414,
      "grad_norm": 1.2350220680236816,
      "learning_rate": 9.964327967761285e-05,
      "loss": 2.0689,
      "step": 432
    },
    {
      "epoch": 0.07800045034902049,
      "grad_norm": 1.2699592113494873,
      "learning_rate": 9.964158908194708e-05,
      "loss": 2.0516,
      "step": 433
    },
    {
      "epoch": 0.07818058995721684,
      "grad_norm": 1.3503687381744385,
      "learning_rate": 9.963989450405722e-05,
      "loss": 2.1425,
      "step": 434
    },
    {
      "epoch": 0.07836072956541319,
      "grad_norm": 1.3755441904067993,
      "learning_rate": 9.963819594407921e-05,
      "loss": 2.328,
      "step": 435
    },
    {
      "epoch": 0.07854086917360954,
      "grad_norm": 1.378748893737793,
      "learning_rate": 9.963649340214929e-05,
      "loss": 2.4802,
      "step": 436
    },
    {
      "epoch": 0.07872100878180589,
      "grad_norm": 1.2597651481628418,
      "learning_rate": 9.963478687840404e-05,
      "loss": 2.3289,
      "step": 437
    },
    {
      "epoch": 0.07890114839000226,
      "grad_norm": 1.2250365018844604,
      "learning_rate": 9.963307637298037e-05,
      "loss": 2.1271,
      "step": 438
    },
    {
      "epoch": 0.07908128799819861,
      "grad_norm": 1.2458264827728271,
      "learning_rate": 9.96313618860155e-05,
      "loss": 1.952,
      "step": 439
    },
    {
      "epoch": 0.07926142760639496,
      "grad_norm": 1.347388744354248,
      "learning_rate": 9.962964341764694e-05,
      "loss": 2.0552,
      "step": 440
    },
    {
      "epoch": 0.07944156721459131,
      "grad_norm": 1.1496285200119019,
      "learning_rate": 9.962792096801255e-05,
      "loss": 1.7487,
      "step": 441
    },
    {
      "epoch": 0.07962170682278766,
      "grad_norm": 1.318347692489624,
      "learning_rate": 9.962619453725052e-05,
      "loss": 2.0939,
      "step": 442
    },
    {
      "epoch": 0.07980184643098401,
      "grad_norm": 1.3763155937194824,
      "learning_rate": 9.962446412549935e-05,
      "loss": 2.0468,
      "step": 443
    },
    {
      "epoch": 0.07998198603918036,
      "grad_norm": 1.2991896867752075,
      "learning_rate": 9.962272973289782e-05,
      "loss": 2.0533,
      "step": 444
    },
    {
      "epoch": 0.08016212564737672,
      "grad_norm": 1.1254770755767822,
      "learning_rate": 9.962099135958509e-05,
      "loss": 1.5903,
      "step": 445
    },
    {
      "epoch": 0.08034226525557307,
      "grad_norm": 1.3234833478927612,
      "learning_rate": 9.96192490057006e-05,
      "loss": 2.2038,
      "step": 446
    },
    {
      "epoch": 0.08052240486376942,
      "grad_norm": 1.536211609840393,
      "learning_rate": 9.961750267138414e-05,
      "loss": 2.2408,
      "step": 447
    },
    {
      "epoch": 0.08070254447196577,
      "grad_norm": 1.2849133014678955,
      "learning_rate": 9.961575235677576e-05,
      "loss": 1.6949,
      "step": 448
    },
    {
      "epoch": 0.08088268408016212,
      "grad_norm": 1.3605488538742065,
      "learning_rate": 9.96139980620159e-05,
      "loss": 2.1481,
      "step": 449
    },
    {
      "epoch": 0.08106282368835847,
      "grad_norm": 1.335553526878357,
      "learning_rate": 9.961223978724528e-05,
      "loss": 2.2895,
      "step": 450
    },
    {
      "epoch": 0.08124296329655482,
      "grad_norm": 1.488923192024231,
      "learning_rate": 9.961047753260496e-05,
      "loss": 2.6168,
      "step": 451
    },
    {
      "epoch": 0.08142310290475119,
      "grad_norm": 1.4191235303878784,
      "learning_rate": 9.960871129823631e-05,
      "loss": 2.5639,
      "step": 452
    },
    {
      "epoch": 0.08160324251294754,
      "grad_norm": 1.596569538116455,
      "learning_rate": 9.960694108428097e-05,
      "loss": 2.497,
      "step": 453
    },
    {
      "epoch": 0.08178338212114389,
      "grad_norm": 1.3627617359161377,
      "learning_rate": 9.960516689088099e-05,
      "loss": 2.6283,
      "step": 454
    },
    {
      "epoch": 0.08196352172934024,
      "grad_norm": 1.267029881477356,
      "learning_rate": 9.960338871817869e-05,
      "loss": 2.5063,
      "step": 455
    },
    {
      "epoch": 0.08214366133753659,
      "grad_norm": 1.4305446147918701,
      "learning_rate": 9.96016065663167e-05,
      "loss": 2.5778,
      "step": 456
    },
    {
      "epoch": 0.08232380094573294,
      "grad_norm": 1.4641321897506714,
      "learning_rate": 9.959982043543799e-05,
      "loss": 2.3723,
      "step": 457
    },
    {
      "epoch": 0.08250394055392929,
      "grad_norm": 1.4689418077468872,
      "learning_rate": 9.959803032568587e-05,
      "loss": 2.2516,
      "step": 458
    },
    {
      "epoch": 0.08268408016212565,
      "grad_norm": 1.5636807680130005,
      "learning_rate": 9.959623623720388e-05,
      "loss": 2.3476,
      "step": 459
    },
    {
      "epoch": 0.082864219770322,
      "grad_norm": 1.840383768081665,
      "learning_rate": 9.9594438170136e-05,
      "loss": 2.8303,
      "step": 460
    },
    {
      "epoch": 0.08304435937851835,
      "grad_norm": 1.9453020095825195,
      "learning_rate": 9.959263612462644e-05,
      "loss": 2.7647,
      "step": 461
    },
    {
      "epoch": 0.0832244989867147,
      "grad_norm": 1.5214613676071167,
      "learning_rate": 9.959083010081976e-05,
      "loss": 2.1241,
      "step": 462
    },
    {
      "epoch": 0.08340463859491105,
      "grad_norm": 1.2775464057922363,
      "learning_rate": 9.958902009886085e-05,
      "loss": 1.8959,
      "step": 463
    },
    {
      "epoch": 0.0835847782031074,
      "grad_norm": 1.4920369386672974,
      "learning_rate": 9.95872061188949e-05,
      "loss": 2.1968,
      "step": 464
    },
    {
      "epoch": 0.08376491781130375,
      "grad_norm": 1.3108103275299072,
      "learning_rate": 9.958538816106742e-05,
      "loss": 2.1947,
      "step": 465
    },
    {
      "epoch": 0.08394505741950012,
      "grad_norm": 1.2641619443893433,
      "learning_rate": 9.958356622552427e-05,
      "loss": 2.1109,
      "step": 466
    },
    {
      "epoch": 0.08412519702769647,
      "grad_norm": 1.1450035572052002,
      "learning_rate": 9.958174031241157e-05,
      "loss": 1.6667,
      "step": 467
    },
    {
      "epoch": 0.08430533663589282,
      "grad_norm": 1.2238646745681763,
      "learning_rate": 9.957991042187582e-05,
      "loss": 2.1029,
      "step": 468
    },
    {
      "epoch": 0.08448547624408917,
      "grad_norm": 1.1442573070526123,
      "learning_rate": 9.957807655406382e-05,
      "loss": 1.9957,
      "step": 469
    },
    {
      "epoch": 0.08466561585228552,
      "grad_norm": 1.2300710678100586,
      "learning_rate": 9.957623870912266e-05,
      "loss": 1.908,
      "step": 470
    },
    {
      "epoch": 0.08484575546048187,
      "grad_norm": 1.249601125717163,
      "learning_rate": 9.957439688719976e-05,
      "loss": 2.1251,
      "step": 471
    },
    {
      "epoch": 0.08502589506867822,
      "grad_norm": 1.2331570386886597,
      "learning_rate": 9.95725510884429e-05,
      "loss": 2.3607,
      "step": 472
    },
    {
      "epoch": 0.08520603467687458,
      "grad_norm": 1.259178638458252,
      "learning_rate": 9.957070131300014e-05,
      "loss": 2.2836,
      "step": 473
    },
    {
      "epoch": 0.08538617428507093,
      "grad_norm": 1.2483408451080322,
      "learning_rate": 9.956884756101987e-05,
      "loss": 2.0932,
      "step": 474
    },
    {
      "epoch": 0.08556631389326728,
      "grad_norm": 1.187480092048645,
      "learning_rate": 9.956698983265081e-05,
      "loss": 1.9143,
      "step": 475
    },
    {
      "epoch": 0.08574645350146363,
      "grad_norm": 1.2999680042266846,
      "learning_rate": 9.956512812804194e-05,
      "loss": 2.1693,
      "step": 476
    },
    {
      "epoch": 0.08592659310965998,
      "grad_norm": 1.2362725734710693,
      "learning_rate": 9.956326244734265e-05,
      "loss": 2.0334,
      "step": 477
    },
    {
      "epoch": 0.08610673271785634,
      "grad_norm": 1.1153935194015503,
      "learning_rate": 9.956139279070257e-05,
      "loss": 1.8978,
      "step": 478
    },
    {
      "epoch": 0.08628687232605269,
      "grad_norm": 1.1800506114959717,
      "learning_rate": 9.955951915827171e-05,
      "loss": 1.941,
      "step": 479
    },
    {
      "epoch": 0.08646701193424905,
      "grad_norm": 1.2130142450332642,
      "learning_rate": 9.955764155020037e-05,
      "loss": 1.9348,
      "step": 480
    },
    {
      "epoch": 0.0866471515424454,
      "grad_norm": 1.2380461692810059,
      "learning_rate": 9.955575996663916e-05,
      "loss": 2.0916,
      "step": 481
    },
    {
      "epoch": 0.08682729115064175,
      "grad_norm": 1.1962411403656006,
      "learning_rate": 9.9553874407739e-05,
      "loss": 2.089,
      "step": 482
    },
    {
      "epoch": 0.0870074307588381,
      "grad_norm": 1.161903738975525,
      "learning_rate": 9.95519848736512e-05,
      "loss": 1.883,
      "step": 483
    },
    {
      "epoch": 0.08718757036703445,
      "grad_norm": 1.2464240789413452,
      "learning_rate": 9.95500913645273e-05,
      "loss": 2.0199,
      "step": 484
    },
    {
      "epoch": 0.0873677099752308,
      "grad_norm": 1.1132943630218506,
      "learning_rate": 9.954819388051921e-05,
      "loss": 1.6719,
      "step": 485
    },
    {
      "epoch": 0.08754784958342715,
      "grad_norm": 1.287757396697998,
      "learning_rate": 9.954629242177913e-05,
      "loss": 1.8318,
      "step": 486
    },
    {
      "epoch": 0.08772798919162351,
      "grad_norm": 1.2616487741470337,
      "learning_rate": 9.954438698845962e-05,
      "loss": 2.0988,
      "step": 487
    },
    {
      "epoch": 0.08790812879981987,
      "grad_norm": 1.2310196161270142,
      "learning_rate": 9.95424775807135e-05,
      "loss": 2.1358,
      "step": 488
    },
    {
      "epoch": 0.08808826840801622,
      "grad_norm": 1.3051484823226929,
      "learning_rate": 9.954056419869398e-05,
      "loss": 2.051,
      "step": 489
    },
    {
      "epoch": 0.08826840801621257,
      "grad_norm": 1.2223469018936157,
      "learning_rate": 9.953864684255452e-05,
      "loss": 2.0419,
      "step": 490
    },
    {
      "epoch": 0.08844854762440892,
      "grad_norm": 1.3044649362564087,
      "learning_rate": 9.953672551244895e-05,
      "loss": 2.2188,
      "step": 491
    },
    {
      "epoch": 0.08862868723260527,
      "grad_norm": 1.2248224020004272,
      "learning_rate": 9.953480020853138e-05,
      "loss": 1.961,
      "step": 492
    },
    {
      "epoch": 0.08880882684080162,
      "grad_norm": 1.324702262878418,
      "learning_rate": 9.953287093095626e-05,
      "loss": 2.4466,
      "step": 493
    },
    {
      "epoch": 0.08898896644899798,
      "grad_norm": 1.3229643106460571,
      "learning_rate": 9.953093767987835e-05,
      "loss": 1.9756,
      "step": 494
    },
    {
      "epoch": 0.08916910605719433,
      "grad_norm": 1.4614118337631226,
      "learning_rate": 9.952900045545278e-05,
      "loss": 2.2114,
      "step": 495
    },
    {
      "epoch": 0.08934924566539068,
      "grad_norm": 1.5767461061477661,
      "learning_rate": 9.952705925783488e-05,
      "loss": 2.1117,
      "step": 496
    },
    {
      "epoch": 0.08952938527358703,
      "grad_norm": 1.3316841125488281,
      "learning_rate": 9.952511408718044e-05,
      "loss": 1.978,
      "step": 497
    },
    {
      "epoch": 0.08970952488178338,
      "grad_norm": 1.2290759086608887,
      "learning_rate": 9.952316494364546e-05,
      "loss": 2.0191,
      "step": 498
    },
    {
      "epoch": 0.08988966448997973,
      "grad_norm": 1.276228904724121,
      "learning_rate": 9.952121182738629e-05,
      "loss": 1.6621,
      "step": 499
    },
    {
      "epoch": 0.09006980409817608,
      "grad_norm": 1.3325002193450928,
      "learning_rate": 9.951925473855965e-05,
      "loss": 2.2832,
      "step": 500
    },
    {
      "epoch": 0.09024994370637243,
      "grad_norm": 1.5747864246368408,
      "learning_rate": 9.951729367732252e-05,
      "loss": 2.6645,
      "step": 501
    },
    {
      "epoch": 0.0904300833145688,
      "grad_norm": 1.3309484720230103,
      "learning_rate": 9.95153286438322e-05,
      "loss": 2.5603,
      "step": 502
    },
    {
      "epoch": 0.09061022292276515,
      "grad_norm": 1.4444763660430908,
      "learning_rate": 9.951335963824633e-05,
      "loss": 2.565,
      "step": 503
    },
    {
      "epoch": 0.0907903625309615,
      "grad_norm": 1.4311542510986328,
      "learning_rate": 9.951138666072288e-05,
      "loss": 2.4572,
      "step": 504
    },
    {
      "epoch": 0.09097050213915785,
      "grad_norm": 1.3777457475662231,
      "learning_rate": 9.95094097114201e-05,
      "loss": 2.713,
      "step": 505
    },
    {
      "epoch": 0.0911506417473542,
      "grad_norm": 1.3411471843719482,
      "learning_rate": 9.95074287904966e-05,
      "loss": 2.2585,
      "step": 506
    },
    {
      "epoch": 0.09133078135555055,
      "grad_norm": 1.5575672388076782,
      "learning_rate": 9.950544389811125e-05,
      "loss": 2.9659,
      "step": 507
    },
    {
      "epoch": 0.0915109209637469,
      "grad_norm": 1.4797890186309814,
      "learning_rate": 9.950345503442332e-05,
      "loss": 2.6608,
      "step": 508
    },
    {
      "epoch": 0.09169106057194326,
      "grad_norm": 1.6868726015090942,
      "learning_rate": 9.950146219959233e-05,
      "loss": 2.6066,
      "step": 509
    },
    {
      "epoch": 0.09187120018013961,
      "grad_norm": 1.8538469076156616,
      "learning_rate": 9.949946539377817e-05,
      "loss": 2.7629,
      "step": 510
    },
    {
      "epoch": 0.09205133978833596,
      "grad_norm": 1.512001872062683,
      "learning_rate": 9.9497464617141e-05,
      "loss": 2.3833,
      "step": 511
    },
    {
      "epoch": 0.09223147939653231,
      "grad_norm": 1.3509187698364258,
      "learning_rate": 9.949545986984133e-05,
      "loss": 2.2007,
      "step": 512
    },
    {
      "epoch": 0.09241161900472866,
      "grad_norm": 1.2238082885742188,
      "learning_rate": 9.949345115203998e-05,
      "loss": 2.1222,
      "step": 513
    },
    {
      "epoch": 0.09259175861292501,
      "grad_norm": 1.3921561241149902,
      "learning_rate": 9.949143846389808e-05,
      "loss": 2.0566,
      "step": 514
    },
    {
      "epoch": 0.09277189822112136,
      "grad_norm": 1.2336784601211548,
      "learning_rate": 9.948942180557708e-05,
      "loss": 2.0548,
      "step": 515
    },
    {
      "epoch": 0.09295203782931773,
      "grad_norm": 1.3000797033309937,
      "learning_rate": 9.948740117723879e-05,
      "loss": 2.065,
      "step": 516
    },
    {
      "epoch": 0.09313217743751408,
      "grad_norm": 1.1365313529968262,
      "learning_rate": 9.948537657904526e-05,
      "loss": 1.8607,
      "step": 517
    },
    {
      "epoch": 0.09331231704571043,
      "grad_norm": 1.190427541732788,
      "learning_rate": 9.948334801115895e-05,
      "loss": 1.8012,
      "step": 518
    },
    {
      "epoch": 0.09349245665390678,
      "grad_norm": 1.1424087285995483,
      "learning_rate": 9.948131547374255e-05,
      "loss": 1.9257,
      "step": 519
    },
    {
      "epoch": 0.09367259626210313,
      "grad_norm": 1.2099260091781616,
      "learning_rate": 9.947927896695913e-05,
      "loss": 2.039,
      "step": 520
    },
    {
      "epoch": 0.09385273587029948,
      "grad_norm": 1.2156577110290527,
      "learning_rate": 9.947723849097203e-05,
      "loss": 2.0548,
      "step": 521
    },
    {
      "epoch": 0.09403287547849583,
      "grad_norm": 1.2186594009399414,
      "learning_rate": 9.947519404594498e-05,
      "loss": 2.0695,
      "step": 522
    },
    {
      "epoch": 0.09421301508669219,
      "grad_norm": 1.2354419231414795,
      "learning_rate": 9.947314563204198e-05,
      "loss": 2.1092,
      "step": 523
    },
    {
      "epoch": 0.09439315469488854,
      "grad_norm": 1.2534072399139404,
      "learning_rate": 9.947109324942729e-05,
      "loss": 1.9764,
      "step": 524
    },
    {
      "epoch": 0.09457329430308489,
      "grad_norm": 1.1973639726638794,
      "learning_rate": 9.946903689826563e-05,
      "loss": 1.754,
      "step": 525
    },
    {
      "epoch": 0.09475343391128124,
      "grad_norm": 1.3052338361740112,
      "learning_rate": 9.946697657872191e-05,
      "loss": 2.1353,
      "step": 526
    },
    {
      "epoch": 0.0949335735194776,
      "grad_norm": 1.2879843711853027,
      "learning_rate": 9.946491229096144e-05,
      "loss": 1.889,
      "step": 527
    },
    {
      "epoch": 0.09511371312767394,
      "grad_norm": 1.2169817686080933,
      "learning_rate": 9.946284403514979e-05,
      "loss": 2.0355,
      "step": 528
    },
    {
      "epoch": 0.0952938527358703,
      "grad_norm": 1.2379153966903687,
      "learning_rate": 9.946077181145287e-05,
      "loss": 2.125,
      "step": 529
    },
    {
      "epoch": 0.09547399234406666,
      "grad_norm": 1.3447768688201904,
      "learning_rate": 9.945869562003694e-05,
      "loss": 2.2768,
      "step": 530
    },
    {
      "epoch": 0.09565413195226301,
      "grad_norm": 1.1286956071853638,
      "learning_rate": 9.945661546106853e-05,
      "loss": 1.8762,
      "step": 531
    },
    {
      "epoch": 0.09583427156045936,
      "grad_norm": 1.4641311168670654,
      "learning_rate": 9.945453133471453e-05,
      "loss": 2.4861,
      "step": 532
    },
    {
      "epoch": 0.09601441116865571,
      "grad_norm": 1.2305301427841187,
      "learning_rate": 9.945244324114211e-05,
      "loss": 1.9035,
      "step": 533
    },
    {
      "epoch": 0.09619455077685206,
      "grad_norm": 1.40278160572052,
      "learning_rate": 9.945035118051878e-05,
      "loss": 2.3066,
      "step": 534
    },
    {
      "epoch": 0.09637469038504841,
      "grad_norm": 1.3533518314361572,
      "learning_rate": 9.944825515301235e-05,
      "loss": 2.1885,
      "step": 535
    },
    {
      "epoch": 0.09655482999324476,
      "grad_norm": 1.3937537670135498,
      "learning_rate": 9.9446155158791e-05,
      "loss": 2.252,
      "step": 536
    },
    {
      "epoch": 0.09673496960144112,
      "grad_norm": 1.3002718687057495,
      "learning_rate": 9.944405119802316e-05,
      "loss": 1.9532,
      "step": 537
    },
    {
      "epoch": 0.09691510920963747,
      "grad_norm": 1.3723276853561401,
      "learning_rate": 9.944194327087761e-05,
      "loss": 2.2575,
      "step": 538
    },
    {
      "epoch": 0.09709524881783382,
      "grad_norm": 1.3745765686035156,
      "learning_rate": 9.943983137752346e-05,
      "loss": 2.1827,
      "step": 539
    },
    {
      "epoch": 0.09727538842603017,
      "grad_norm": 1.3019641637802124,
      "learning_rate": 9.943771551813011e-05,
      "loss": 2.205,
      "step": 540
    },
    {
      "epoch": 0.09745552803422652,
      "grad_norm": 1.3320200443267822,
      "learning_rate": 9.94355956928673e-05,
      "loss": 2.1291,
      "step": 541
    },
    {
      "epoch": 0.09763566764242287,
      "grad_norm": 1.3012250661849976,
      "learning_rate": 9.943347190190511e-05,
      "loss": 2.005,
      "step": 542
    },
    {
      "epoch": 0.09781580725061922,
      "grad_norm": 1.3021081686019897,
      "learning_rate": 9.943134414541386e-05,
      "loss": 2.0511,
      "step": 543
    },
    {
      "epoch": 0.09799594685881559,
      "grad_norm": 1.3133217096328735,
      "learning_rate": 9.942921242356425e-05,
      "loss": 2.1142,
      "step": 544
    },
    {
      "epoch": 0.09817608646701194,
      "grad_norm": 1.343146562576294,
      "learning_rate": 9.942707673652731e-05,
      "loss": 1.8386,
      "step": 545
    },
    {
      "epoch": 0.09835622607520829,
      "grad_norm": 1.2855020761489868,
      "learning_rate": 9.942493708447435e-05,
      "loss": 1.7687,
      "step": 546
    },
    {
      "epoch": 0.09853636568340464,
      "grad_norm": 1.188632845878601,
      "learning_rate": 9.9422793467577e-05,
      "loss": 1.8733,
      "step": 547
    },
    {
      "epoch": 0.09871650529160099,
      "grad_norm": 1.3299243450164795,
      "learning_rate": 9.942064588600724e-05,
      "loss": 2.0229,
      "step": 548
    },
    {
      "epoch": 0.09889664489979734,
      "grad_norm": 1.2409086227416992,
      "learning_rate": 9.941849433993734e-05,
      "loss": 1.8813,
      "step": 549
    },
    {
      "epoch": 0.09907678450799369,
      "grad_norm": 1.4136215448379517,
      "learning_rate": 9.941633882953989e-05,
      "loss": 2.5996,
      "step": 550
    },
    {
      "epoch": 0.09925692411619005,
      "grad_norm": 1.5979270935058594,
      "learning_rate": 9.941417935498782e-05,
      "loss": 2.3478,
      "step": 551
    },
    {
      "epoch": 0.0994370637243864,
      "grad_norm": 1.4183084964752197,
      "learning_rate": 9.941201591645434e-05,
      "loss": 2.5067,
      "step": 552
    },
    {
      "epoch": 0.09961720333258275,
      "grad_norm": 1.3027455806732178,
      "learning_rate": 9.940984851411301e-05,
      "loss": 2.3411,
      "step": 553
    },
    {
      "epoch": 0.0997973429407791,
      "grad_norm": 1.8705165386199951,
      "learning_rate": 9.94076771481377e-05,
      "loss": 2.3631,
      "step": 554
    },
    {
      "epoch": 0.09997748254897545,
      "grad_norm": 1.4732381105422974,
      "learning_rate": 9.94055018187026e-05,
      "loss": 2.5289,
      "step": 555
    },
    {
      "epoch": 0.1001576221571718,
      "grad_norm": 1.4549360275268555,
      "learning_rate": 9.94033225259822e-05,
      "loss": 2.1177,
      "step": 556
    },
    {
      "epoch": 0.10033776176536816,
      "grad_norm": 1.69441556930542,
      "learning_rate": 9.940113927015134e-05,
      "loss": 2.4761,
      "step": 557
    },
    {
      "epoch": 0.10051790137356452,
      "grad_norm": 1.782146692276001,
      "learning_rate": 9.939895205138514e-05,
      "loss": 2.1732,
      "step": 558
    },
    {
      "epoch": 0.10069804098176087,
      "grad_norm": 1.762495756149292,
      "learning_rate": 9.939676086985907e-05,
      "loss": 2.7288,
      "step": 559
    },
    {
      "epoch": 0.10087818058995722,
      "grad_norm": 2.0288562774658203,
      "learning_rate": 9.939456572574892e-05,
      "loss": 2.1508,
      "step": 560
    },
    {
      "epoch": 0.10105832019815357,
      "grad_norm": 1.2310386896133423,
      "learning_rate": 9.939236661923074e-05,
      "loss": 2.1423,
      "step": 561
    },
    {
      "epoch": 0.10123845980634992,
      "grad_norm": 1.186572551727295,
      "learning_rate": 9.939016355048099e-05,
      "loss": 1.9507,
      "step": 562
    },
    {
      "epoch": 0.10141859941454627,
      "grad_norm": 1.2622029781341553,
      "learning_rate": 9.938795651967637e-05,
      "loss": 1.7998,
      "step": 563
    },
    {
      "epoch": 0.10159873902274262,
      "grad_norm": 1.3796643018722534,
      "learning_rate": 9.938574552699394e-05,
      "loss": 1.9648,
      "step": 564
    },
    {
      "epoch": 0.10177887863093897,
      "grad_norm": 1.236175537109375,
      "learning_rate": 9.938353057261105e-05,
      "loss": 1.9793,
      "step": 565
    },
    {
      "epoch": 0.10195901823913534,
      "grad_norm": 1.7577669620513916,
      "learning_rate": 9.938131165670538e-05,
      "loss": 1.9435,
      "step": 566
    },
    {
      "epoch": 0.10213915784733169,
      "grad_norm": 1.2309681177139282,
      "learning_rate": 9.937908877945496e-05,
      "loss": 2.1315,
      "step": 567
    },
    {
      "epoch": 0.10231929745552804,
      "grad_norm": 1.2463933229446411,
      "learning_rate": 9.93768619410381e-05,
      "loss": 2.2162,
      "step": 568
    },
    {
      "epoch": 0.10249943706372439,
      "grad_norm": 1.2483500242233276,
      "learning_rate": 9.93746311416334e-05,
      "loss": 1.7632,
      "step": 569
    },
    {
      "epoch": 0.10267957667192074,
      "grad_norm": 1.1438194513320923,
      "learning_rate": 9.937239638141984e-05,
      "loss": 1.8644,
      "step": 570
    },
    {
      "epoch": 0.10285971628011709,
      "grad_norm": 1.1547142267227173,
      "learning_rate": 9.937015766057671e-05,
      "loss": 1.985,
      "step": 571
    },
    {
      "epoch": 0.10303985588831344,
      "grad_norm": 1.1651068925857544,
      "learning_rate": 9.936791497928357e-05,
      "loss": 1.932,
      "step": 572
    },
    {
      "epoch": 0.1032199954965098,
      "grad_norm": 1.2023049592971802,
      "learning_rate": 9.936566833772034e-05,
      "loss": 2.0256,
      "step": 573
    },
    {
      "epoch": 0.10340013510470615,
      "grad_norm": 1.1891133785247803,
      "learning_rate": 9.936341773606723e-05,
      "loss": 2.0444,
      "step": 574
    },
    {
      "epoch": 0.1035802747129025,
      "grad_norm": 1.382871150970459,
      "learning_rate": 9.936116317450482e-05,
      "loss": 2.0966,
      "step": 575
    },
    {
      "epoch": 0.10376041432109885,
      "grad_norm": 1.181156039237976,
      "learning_rate": 9.935890465321392e-05,
      "loss": 1.7597,
      "step": 576
    },
    {
      "epoch": 0.1039405539292952,
      "grad_norm": 1.1654242277145386,
      "learning_rate": 9.935664217237573e-05,
      "loss": 1.6947,
      "step": 577
    },
    {
      "epoch": 0.10412069353749155,
      "grad_norm": 1.2307331562042236,
      "learning_rate": 9.935437573217175e-05,
      "loss": 1.9475,
      "step": 578
    },
    {
      "epoch": 0.1043008331456879,
      "grad_norm": 1.3564519882202148,
      "learning_rate": 9.935210533278379e-05,
      "loss": 2.3476,
      "step": 579
    },
    {
      "epoch": 0.10448097275388427,
      "grad_norm": 1.1362614631652832,
      "learning_rate": 9.934983097439396e-05,
      "loss": 1.9735,
      "step": 580
    },
    {
      "epoch": 0.10466111236208062,
      "grad_norm": 1.245951533317566,
      "learning_rate": 9.934755265718476e-05,
      "loss": 2.2328,
      "step": 581
    },
    {
      "epoch": 0.10484125197027697,
      "grad_norm": 1.2103431224822998,
      "learning_rate": 9.934527038133889e-05,
      "loss": 2.2116,
      "step": 582
    },
    {
      "epoch": 0.10502139157847332,
      "grad_norm": 1.2476712465286255,
      "learning_rate": 9.934298414703948e-05,
      "loss": 2.0962,
      "step": 583
    },
    {
      "epoch": 0.10520153118666967,
      "grad_norm": 1.24763822555542,
      "learning_rate": 9.934069395446993e-05,
      "loss": 1.9395,
      "step": 584
    },
    {
      "epoch": 0.10538167079486602,
      "grad_norm": 1.3773177862167358,
      "learning_rate": 9.933839980381393e-05,
      "loss": 2.2344,
      "step": 585
    },
    {
      "epoch": 0.10556181040306237,
      "grad_norm": 1.348358392715454,
      "learning_rate": 9.933610169525554e-05,
      "loss": 2.0153,
      "step": 586
    },
    {
      "epoch": 0.10574195001125873,
      "grad_norm": 1.327081561088562,
      "learning_rate": 9.933379962897909e-05,
      "loss": 2.0961,
      "step": 587
    },
    {
      "epoch": 0.10592208961945508,
      "grad_norm": 1.2437586784362793,
      "learning_rate": 9.933149360516927e-05,
      "loss": 1.9906,
      "step": 588
    },
    {
      "epoch": 0.10610222922765143,
      "grad_norm": 1.3158669471740723,
      "learning_rate": 9.932918362401104e-05,
      "loss": 2.0608,
      "step": 589
    },
    {
      "epoch": 0.10628236883584778,
      "grad_norm": 1.361436128616333,
      "learning_rate": 9.932686968568976e-05,
      "loss": 2.0768,
      "step": 590
    },
    {
      "epoch": 0.10646250844404413,
      "grad_norm": 1.3792731761932373,
      "learning_rate": 9.932455179039101e-05,
      "loss": 1.9328,
      "step": 591
    },
    {
      "epoch": 0.10664264805224048,
      "grad_norm": 1.528085708618164,
      "learning_rate": 9.932222993830074e-05,
      "loss": 2.3766,
      "step": 592
    },
    {
      "epoch": 0.10682278766043683,
      "grad_norm": 1.2961821556091309,
      "learning_rate": 9.931990412960518e-05,
      "loss": 1.9681,
      "step": 593
    },
    {
      "epoch": 0.1070029272686332,
      "grad_norm": 1.3054449558258057,
      "learning_rate": 9.931757436449097e-05,
      "loss": 1.9601,
      "step": 594
    },
    {
      "epoch": 0.10718306687682955,
      "grad_norm": 1.3940820693969727,
      "learning_rate": 9.931524064314496e-05,
      "loss": 2.2602,
      "step": 595
    },
    {
      "epoch": 0.1073632064850259,
      "grad_norm": 1.380031943321228,
      "learning_rate": 9.931290296575436e-05,
      "loss": 2.4307,
      "step": 596
    },
    {
      "epoch": 0.10754334609322225,
      "grad_norm": 1.3479726314544678,
      "learning_rate": 9.93105613325067e-05,
      "loss": 2.1546,
      "step": 597
    },
    {
      "epoch": 0.1077234857014186,
      "grad_norm": 1.179189682006836,
      "learning_rate": 9.930821574358984e-05,
      "loss": 1.6836,
      "step": 598
    },
    {
      "epoch": 0.10790362530961495,
      "grad_norm": 1.339950680732727,
      "learning_rate": 9.930586619919192e-05,
      "loss": 2.2317,
      "step": 599
    },
    {
      "epoch": 0.1080837649178113,
      "grad_norm": 1.2645717859268188,
      "learning_rate": 9.930351269950143e-05,
      "loss": 1.8206,
      "step": 600
    },
    {
      "epoch": 0.10826390452600766,
      "grad_norm": 1.6206364631652832,
      "learning_rate": 9.930115524470718e-05,
      "loss": 2.6552,
      "step": 601
    },
    {
      "epoch": 0.10844404413420401,
      "grad_norm": 1.2960243225097656,
      "learning_rate": 9.929879383499825e-05,
      "loss": 2.2248,
      "step": 602
    },
    {
      "epoch": 0.10862418374240036,
      "grad_norm": 1.2325005531311035,
      "learning_rate": 9.92964284705641e-05,
      "loss": 2.3235,
      "step": 603
    },
    {
      "epoch": 0.10880432335059671,
      "grad_norm": 1.3556349277496338,
      "learning_rate": 9.929405915159446e-05,
      "loss": 2.3627,
      "step": 604
    },
    {
      "epoch": 0.10898446295879306,
      "grad_norm": 1.3524086475372314,
      "learning_rate": 9.929168587827942e-05,
      "loss": 2.5927,
      "step": 605
    },
    {
      "epoch": 0.10916460256698941,
      "grad_norm": 1.334237813949585,
      "learning_rate": 9.928930865080934e-05,
      "loss": 2.1805,
      "step": 606
    },
    {
      "epoch": 0.10934474217518576,
      "grad_norm": 1.5597326755523682,
      "learning_rate": 9.928692746937494e-05,
      "loss": 2.5088,
      "step": 607
    },
    {
      "epoch": 0.10952488178338213,
      "grad_norm": 1.8527363538742065,
      "learning_rate": 9.92845423341672e-05,
      "loss": 2.6455,
      "step": 608
    },
    {
      "epoch": 0.10970502139157848,
      "grad_norm": 1.8527334928512573,
      "learning_rate": 9.92821532453775e-05,
      "loss": 2.71,
      "step": 609
    },
    {
      "epoch": 0.10988516099977483,
      "grad_norm": 1.2585922479629517,
      "learning_rate": 9.927976020319747e-05,
      "loss": 2.1503,
      "step": 610
    },
    {
      "epoch": 0.11006530060797118,
      "grad_norm": 1.2161167860031128,
      "learning_rate": 9.927736320781908e-05,
      "loss": 1.7714,
      "step": 611
    },
    {
      "epoch": 0.11024544021616753,
      "grad_norm": 1.2532986402511597,
      "learning_rate": 9.92749622594346e-05,
      "loss": 2.0686,
      "step": 612
    },
    {
      "epoch": 0.11042557982436388,
      "grad_norm": 1.2780698537826538,
      "learning_rate": 9.927255735823667e-05,
      "loss": 2.0918,
      "step": 613
    },
    {
      "epoch": 0.11060571943256023,
      "grad_norm": 1.2372409105300903,
      "learning_rate": 9.927014850441818e-05,
      "loss": 1.9126,
      "step": 614
    },
    {
      "epoch": 0.11078585904075659,
      "grad_norm": 1.3066140413284302,
      "learning_rate": 9.926773569817237e-05,
      "loss": 1.9936,
      "step": 615
    },
    {
      "epoch": 0.11096599864895294,
      "grad_norm": 1.2493231296539307,
      "learning_rate": 9.926531893969279e-05,
      "loss": 1.9355,
      "step": 616
    },
    {
      "epoch": 0.1111461382571493,
      "grad_norm": 1.3704819679260254,
      "learning_rate": 9.926289822917335e-05,
      "loss": 2.0541,
      "step": 617
    },
    {
      "epoch": 0.11132627786534564,
      "grad_norm": 1.2036651372909546,
      "learning_rate": 9.926047356680819e-05,
      "loss": 2.0361,
      "step": 618
    },
    {
      "epoch": 0.111506417473542,
      "grad_norm": 1.1734963655471802,
      "learning_rate": 9.925804495279185e-05,
      "loss": 2.0089,
      "step": 619
    },
    {
      "epoch": 0.11168655708173834,
      "grad_norm": 1.3296834230422974,
      "learning_rate": 9.925561238731912e-05,
      "loss": 2.1309,
      "step": 620
    },
    {
      "epoch": 0.1118666966899347,
      "grad_norm": 1.1076639890670776,
      "learning_rate": 9.925317587058516e-05,
      "loss": 1.974,
      "step": 621
    },
    {
      "epoch": 0.11204683629813106,
      "grad_norm": 1.2039434909820557,
      "learning_rate": 9.925073540278542e-05,
      "loss": 1.9346,
      "step": 622
    },
    {
      "epoch": 0.11222697590632741,
      "grad_norm": 1.2502390146255493,
      "learning_rate": 9.924829098411568e-05,
      "loss": 1.81,
      "step": 623
    },
    {
      "epoch": 0.11240711551452376,
      "grad_norm": 1.2662842273712158,
      "learning_rate": 9.924584261477203e-05,
      "loss": 1.9783,
      "step": 624
    },
    {
      "epoch": 0.11258725512272011,
      "grad_norm": 1.18838632106781,
      "learning_rate": 9.924339029495086e-05,
      "loss": 1.7056,
      "step": 625
    },
    {
      "epoch": 0.11276739473091646,
      "grad_norm": 1.202418565750122,
      "learning_rate": 9.924093402484891e-05,
      "loss": 1.9405,
      "step": 626
    },
    {
      "epoch": 0.11294753433911281,
      "grad_norm": 1.2927911281585693,
      "learning_rate": 9.923847380466323e-05,
      "loss": 1.7644,
      "step": 627
    },
    {
      "epoch": 0.11312767394730916,
      "grad_norm": 1.2021912336349487,
      "learning_rate": 9.923600963459115e-05,
      "loss": 1.7018,
      "step": 628
    },
    {
      "epoch": 0.11330781355550551,
      "grad_norm": 1.2121084928512573,
      "learning_rate": 9.923354151483035e-05,
      "loss": 2.0384,
      "step": 629
    },
    {
      "epoch": 0.11348795316370187,
      "grad_norm": 1.2217655181884766,
      "learning_rate": 9.923106944557884e-05,
      "loss": 1.9361,
      "step": 630
    },
    {
      "epoch": 0.11366809277189822,
      "grad_norm": 1.297149658203125,
      "learning_rate": 9.922859342703492e-05,
      "loss": 2.1438,
      "step": 631
    },
    {
      "epoch": 0.11384823238009457,
      "grad_norm": 1.2163119316101074,
      "learning_rate": 9.922611345939723e-05,
      "loss": 1.8925,
      "step": 632
    },
    {
      "epoch": 0.11402837198829092,
      "grad_norm": 1.2779686450958252,
      "learning_rate": 9.922362954286468e-05,
      "loss": 1.9334,
      "step": 633
    },
    {
      "epoch": 0.11420851159648727,
      "grad_norm": 1.3244131803512573,
      "learning_rate": 9.922114167763655e-05,
      "loss": 2.0773,
      "step": 634
    },
    {
      "epoch": 0.11438865120468363,
      "grad_norm": 1.2796010971069336,
      "learning_rate": 9.92186498639124e-05,
      "loss": 1.8922,
      "step": 635
    },
    {
      "epoch": 0.11456879081287998,
      "grad_norm": 1.3653994798660278,
      "learning_rate": 9.921615410189213e-05,
      "loss": 1.9424,
      "step": 636
    },
    {
      "epoch": 0.11474893042107634,
      "grad_norm": 1.3693561553955078,
      "learning_rate": 9.921365439177596e-05,
      "loss": 2.1637,
      "step": 637
    },
    {
      "epoch": 0.11492907002927269,
      "grad_norm": 1.5115869045257568,
      "learning_rate": 9.921115073376439e-05,
      "loss": 1.9723,
      "step": 638
    },
    {
      "epoch": 0.11510920963746904,
      "grad_norm": 1.4102026224136353,
      "learning_rate": 9.920864312805828e-05,
      "loss": 2.0927,
      "step": 639
    },
    {
      "epoch": 0.11528934924566539,
      "grad_norm": 1.5333706140518188,
      "learning_rate": 9.920613157485881e-05,
      "loss": 2.269,
      "step": 640
    },
    {
      "epoch": 0.11546948885386174,
      "grad_norm": 1.3293869495391846,
      "learning_rate": 9.920361607436742e-05,
      "loss": 1.9739,
      "step": 641
    },
    {
      "epoch": 0.11564962846205809,
      "grad_norm": 1.318129062652588,
      "learning_rate": 9.920109662678592e-05,
      "loss": 1.8895,
      "step": 642
    },
    {
      "epoch": 0.11582976807025444,
      "grad_norm": 1.3061244487762451,
      "learning_rate": 9.91985732323164e-05,
      "loss": 2.176,
      "step": 643
    },
    {
      "epoch": 0.1160099076784508,
      "grad_norm": 1.2987688779830933,
      "learning_rate": 9.919604589116132e-05,
      "loss": 2.0634,
      "step": 644
    },
    {
      "epoch": 0.11619004728664716,
      "grad_norm": 1.2750600576400757,
      "learning_rate": 9.919351460352338e-05,
      "loss": 1.9271,
      "step": 645
    },
    {
      "epoch": 0.1163701868948435,
      "grad_norm": 1.3042123317718506,
      "learning_rate": 9.919097936960567e-05,
      "loss": 1.8597,
      "step": 646
    },
    {
      "epoch": 0.11655032650303986,
      "grad_norm": 1.2460150718688965,
      "learning_rate": 9.918844018961155e-05,
      "loss": 1.8877,
      "step": 647
    },
    {
      "epoch": 0.1167304661112362,
      "grad_norm": 1.3470635414123535,
      "learning_rate": 9.918589706374472e-05,
      "loss": 2.1862,
      "step": 648
    },
    {
      "epoch": 0.11691060571943256,
      "grad_norm": 1.4191900491714478,
      "learning_rate": 9.918334999220917e-05,
      "loss": 1.7834,
      "step": 649
    },
    {
      "epoch": 0.1170907453276289,
      "grad_norm": 1.1623694896697998,
      "learning_rate": 9.918079897520927e-05,
      "loss": 1.8351,
      "step": 650
    },
    {
      "epoch": 0.11727088493582527,
      "grad_norm": 1.44710373878479,
      "learning_rate": 9.917824401294962e-05,
      "loss": 2.5077,
      "step": 651
    },
    {
      "epoch": 0.11745102454402162,
      "grad_norm": 1.5147384405136108,
      "learning_rate": 9.917568510563517e-05,
      "loss": 2.2504,
      "step": 652
    },
    {
      "epoch": 0.11763116415221797,
      "grad_norm": 1.3511815071105957,
      "learning_rate": 9.917312225347124e-05,
      "loss": 2.2882,
      "step": 653
    },
    {
      "epoch": 0.11781130376041432,
      "grad_norm": 1.301245093345642,
      "learning_rate": 9.917055545666339e-05,
      "loss": 2.2157,
      "step": 654
    },
    {
      "epoch": 0.11799144336861067,
      "grad_norm": 1.642171025276184,
      "learning_rate": 9.91679847154175e-05,
      "loss": 2.3195,
      "step": 655
    },
    {
      "epoch": 0.11817158297680702,
      "grad_norm": 1.4102668762207031,
      "learning_rate": 9.916541002993987e-05,
      "loss": 2.32,
      "step": 656
    },
    {
      "epoch": 0.11835172258500337,
      "grad_norm": 1.4176762104034424,
      "learning_rate": 9.916283140043696e-05,
      "loss": 2.4767,
      "step": 657
    },
    {
      "epoch": 0.11853186219319974,
      "grad_norm": 1.7280687093734741,
      "learning_rate": 9.916024882711567e-05,
      "loss": 2.3133,
      "step": 658
    },
    {
      "epoch": 0.11871200180139609,
      "grad_norm": 1.9185752868652344,
      "learning_rate": 9.915766231018318e-05,
      "loss": 2.5846,
      "step": 659
    },
    {
      "epoch": 0.11889214140959244,
      "grad_norm": 1.6732910871505737,
      "learning_rate": 9.915507184984695e-05,
      "loss": 2.8271,
      "step": 660
    },
    {
      "epoch": 0.11907228101778879,
      "grad_norm": 1.233290195465088,
      "learning_rate": 9.91524774463148e-05,
      "loss": 1.9402,
      "step": 661
    },
    {
      "epoch": 0.11925242062598514,
      "grad_norm": 1.248557209968567,
      "learning_rate": 9.914987909979485e-05,
      "loss": 1.8964,
      "step": 662
    },
    {
      "epoch": 0.11943256023418149,
      "grad_norm": 1.296522617340088,
      "learning_rate": 9.914727681049554e-05,
      "loss": 2.2381,
      "step": 663
    },
    {
      "epoch": 0.11961269984237784,
      "grad_norm": 1.274907112121582,
      "learning_rate": 9.914467057862563e-05,
      "loss": 2.1219,
      "step": 664
    },
    {
      "epoch": 0.1197928394505742,
      "grad_norm": 1.119384765625,
      "learning_rate": 9.914206040439418e-05,
      "loss": 1.8136,
      "step": 665
    },
    {
      "epoch": 0.11997297905877055,
      "grad_norm": 1.323623538017273,
      "learning_rate": 9.913944628801058e-05,
      "loss": 1.9493,
      "step": 666
    },
    {
      "epoch": 0.1201531186669669,
      "grad_norm": 1.13485848903656,
      "learning_rate": 9.913682822968454e-05,
      "loss": 1.9245,
      "step": 667
    },
    {
      "epoch": 0.12033325827516325,
      "grad_norm": 1.34122633934021,
      "learning_rate": 9.913420622962606e-05,
      "loss": 2.3114,
      "step": 668
    },
    {
      "epoch": 0.1205133978833596,
      "grad_norm": 1.2279052734375,
      "learning_rate": 9.913158028804548e-05,
      "loss": 1.9898,
      "step": 669
    },
    {
      "epoch": 0.12069353749155595,
      "grad_norm": 1.147538185119629,
      "learning_rate": 9.912895040515349e-05,
      "loss": 1.8738,
      "step": 670
    },
    {
      "epoch": 0.1208736770997523,
      "grad_norm": 1.3366378545761108,
      "learning_rate": 9.912631658116101e-05,
      "loss": 1.8985,
      "step": 671
    },
    {
      "epoch": 0.12105381670794867,
      "grad_norm": 1.2011362314224243,
      "learning_rate": 9.912367881627933e-05,
      "loss": 2.0348,
      "step": 672
    },
    {
      "epoch": 0.12123395631614502,
      "grad_norm": 1.2105849981307983,
      "learning_rate": 9.912103711072008e-05,
      "loss": 2.0519,
      "step": 673
    },
    {
      "epoch": 0.12141409592434137,
      "grad_norm": 1.245194435119629,
      "learning_rate": 9.911839146469515e-05,
      "loss": 1.9079,
      "step": 674
    },
    {
      "epoch": 0.12159423553253772,
      "grad_norm": 1.2677587270736694,
      "learning_rate": 9.91157418784168e-05,
      "loss": 2.1401,
      "step": 675
    },
    {
      "epoch": 0.12177437514073407,
      "grad_norm": 1.2442809343338013,
      "learning_rate": 9.911308835209755e-05,
      "loss": 1.7792,
      "step": 676
    },
    {
      "epoch": 0.12195451474893042,
      "grad_norm": 1.4197555780410767,
      "learning_rate": 9.911043088595027e-05,
      "loss": 1.9049,
      "step": 677
    },
    {
      "epoch": 0.12213465435712677,
      "grad_norm": 1.268538475036621,
      "learning_rate": 9.910776948018814e-05,
      "loss": 1.8835,
      "step": 678
    },
    {
      "epoch": 0.12231479396532313,
      "grad_norm": 1.2134947776794434,
      "learning_rate": 9.910510413502467e-05,
      "loss": 1.8627,
      "step": 679
    },
    {
      "epoch": 0.12249493357351948,
      "grad_norm": 1.3159133195877075,
      "learning_rate": 9.910243485067367e-05,
      "loss": 2.1532,
      "step": 680
    },
    {
      "epoch": 0.12267507318171583,
      "grad_norm": 1.3118300437927246,
      "learning_rate": 9.909976162734926e-05,
      "loss": 2.0355,
      "step": 681
    },
    {
      "epoch": 0.12285521278991218,
      "grad_norm": 1.230215311050415,
      "learning_rate": 9.90970844652659e-05,
      "loss": 1.8318,
      "step": 682
    },
    {
      "epoch": 0.12303535239810853,
      "grad_norm": 1.428345799446106,
      "learning_rate": 9.90944033646383e-05,
      "loss": 2.1659,
      "step": 683
    },
    {
      "epoch": 0.12321549200630488,
      "grad_norm": 1.2812304496765137,
      "learning_rate": 9.90917183256816e-05,
      "loss": 1.9663,
      "step": 684
    },
    {
      "epoch": 0.12339563161450123,
      "grad_norm": 1.4184229373931885,
      "learning_rate": 9.908902934861117e-05,
      "loss": 1.7655,
      "step": 685
    },
    {
      "epoch": 0.1235757712226976,
      "grad_norm": 1.2999763488769531,
      "learning_rate": 9.908633643364271e-05,
      "loss": 1.885,
      "step": 686
    },
    {
      "epoch": 0.12375591083089395,
      "grad_norm": 1.1869958639144897,
      "learning_rate": 9.908363958099225e-05,
      "loss": 1.866,
      "step": 687
    },
    {
      "epoch": 0.1239360504390903,
      "grad_norm": 1.1575794219970703,
      "learning_rate": 9.908093879087614e-05,
      "loss": 1.6819,
      "step": 688
    },
    {
      "epoch": 0.12411619004728665,
      "grad_norm": 1.4000781774520874,
      "learning_rate": 9.907823406351102e-05,
      "loss": 2.1885,
      "step": 689
    },
    {
      "epoch": 0.124296329655483,
      "grad_norm": 1.44357430934906,
      "learning_rate": 9.907552539911387e-05,
      "loss": 2.0515,
      "step": 690
    },
    {
      "epoch": 0.12447646926367935,
      "grad_norm": 1.3052406311035156,
      "learning_rate": 9.907281279790197e-05,
      "loss": 1.822,
      "step": 691
    },
    {
      "epoch": 0.1246566088718757,
      "grad_norm": 1.3275206089019775,
      "learning_rate": 9.907009626009294e-05,
      "loss": 2.0095,
      "step": 692
    },
    {
      "epoch": 0.12483674848007205,
      "grad_norm": 1.3867100477218628,
      "learning_rate": 9.906737578590468e-05,
      "loss": 2.0162,
      "step": 693
    },
    {
      "epoch": 0.1250168880882684,
      "grad_norm": 1.2267770767211914,
      "learning_rate": 9.906465137555543e-05,
      "loss": 1.8557,
      "step": 694
    },
    {
      "epoch": 0.12519702769646476,
      "grad_norm": 1.3638699054718018,
      "learning_rate": 9.906192302926376e-05,
      "loss": 2.0727,
      "step": 695
    },
    {
      "epoch": 0.1253771673046611,
      "grad_norm": 1.5085831880569458,
      "learning_rate": 9.905919074724852e-05,
      "loss": 1.7744,
      "step": 696
    },
    {
      "epoch": 0.12555730691285746,
      "grad_norm": 1.4602805376052856,
      "learning_rate": 9.905645452972889e-05,
      "loss": 2.3713,
      "step": 697
    },
    {
      "epoch": 0.12573744652105381,
      "grad_norm": 1.391291856765747,
      "learning_rate": 9.905371437692437e-05,
      "loss": 2.0513,
      "step": 698
    },
    {
      "epoch": 0.12591758612925016,
      "grad_norm": 1.3481264114379883,
      "learning_rate": 9.905097028905477e-05,
      "loss": 1.8174,
      "step": 699
    },
    {
      "epoch": 0.12609772573744651,
      "grad_norm": 1.2525372505187988,
      "learning_rate": 9.904822226634024e-05,
      "loss": 2.0188,
      "step": 700
    },
    {
      "epoch": 0.12627786534564286,
      "grad_norm": 1.3850762844085693,
      "learning_rate": 9.90454703090012e-05,
      "loss": 2.1517,
      "step": 701
    },
    {
      "epoch": 0.12645800495383921,
      "grad_norm": 1.3955851793289185,
      "learning_rate": 9.904271441725843e-05,
      "loss": 2.3126,
      "step": 702
    },
    {
      "epoch": 0.12663814456203557,
      "grad_norm": 1.3821715116500854,
      "learning_rate": 9.9039954591333e-05,
      "loss": 2.3915,
      "step": 703
    },
    {
      "epoch": 0.12681828417023194,
      "grad_norm": 1.4405173063278198,
      "learning_rate": 9.903719083144627e-05,
      "loss": 2.0821,
      "step": 704
    },
    {
      "epoch": 0.1269984237784283,
      "grad_norm": 1.4351264238357544,
      "learning_rate": 9.903442313782e-05,
      "loss": 2.5997,
      "step": 705
    },
    {
      "epoch": 0.12717856338662464,
      "grad_norm": 1.5886861085891724,
      "learning_rate": 9.90316515106762e-05,
      "loss": 2.6303,
      "step": 706
    },
    {
      "epoch": 0.127358702994821,
      "grad_norm": 1.8986220359802246,
      "learning_rate": 9.902887595023718e-05,
      "loss": 2.2789,
      "step": 707
    },
    {
      "epoch": 0.12753884260301734,
      "grad_norm": 1.804407000541687,
      "learning_rate": 9.902609645672562e-05,
      "loss": 2.6275,
      "step": 708
    },
    {
      "epoch": 0.1277189822112137,
      "grad_norm": 1.7212653160095215,
      "learning_rate": 9.902331303036448e-05,
      "loss": 2.4336,
      "step": 709
    },
    {
      "epoch": 0.12789912181941004,
      "grad_norm": 2.3265233039855957,
      "learning_rate": 9.902052567137706e-05,
      "loss": 2.7222,
      "step": 710
    },
    {
      "epoch": 0.1280792614276064,
      "grad_norm": 1.2806710004806519,
      "learning_rate": 9.901773437998692e-05,
      "loss": 1.8323,
      "step": 711
    },
    {
      "epoch": 0.12825940103580274,
      "grad_norm": 1.2179723978042603,
      "learning_rate": 9.901493915641802e-05,
      "loss": 2.0353,
      "step": 712
    },
    {
      "epoch": 0.1284395406439991,
      "grad_norm": 1.2533454895019531,
      "learning_rate": 9.901214000089456e-05,
      "loss": 1.9294,
      "step": 713
    },
    {
      "epoch": 0.12861968025219545,
      "grad_norm": 1.3815363645553589,
      "learning_rate": 9.900933691364112e-05,
      "loss": 2.107,
      "step": 714
    },
    {
      "epoch": 0.1287998198603918,
      "grad_norm": 1.126511573791504,
      "learning_rate": 9.900652989488255e-05,
      "loss": 1.7513,
      "step": 715
    },
    {
      "epoch": 0.12897995946858815,
      "grad_norm": 1.1627007722854614,
      "learning_rate": 9.900371894484402e-05,
      "loss": 1.6158,
      "step": 716
    },
    {
      "epoch": 0.1291600990767845,
      "grad_norm": 1.2627477645874023,
      "learning_rate": 9.9000904063751e-05,
      "loss": 1.9606,
      "step": 717
    },
    {
      "epoch": 0.12934023868498087,
      "grad_norm": 1.299331545829773,
      "learning_rate": 9.899808525182935e-05,
      "loss": 1.8576,
      "step": 718
    },
    {
      "epoch": 0.12952037829317722,
      "grad_norm": 1.1861523389816284,
      "learning_rate": 9.899526250930516e-05,
      "loss": 1.8284,
      "step": 719
    },
    {
      "epoch": 0.12970051790137357,
      "grad_norm": 1.1307034492492676,
      "learning_rate": 9.899243583640488e-05,
      "loss": 1.786,
      "step": 720
    },
    {
      "epoch": 0.12988065750956992,
      "grad_norm": 1.1576093435287476,
      "learning_rate": 9.898960523335524e-05,
      "loss": 1.7738,
      "step": 721
    },
    {
      "epoch": 0.13006079711776627,
      "grad_norm": 1.2430235147476196,
      "learning_rate": 9.898677070038334e-05,
      "loss": 1.8535,
      "step": 722
    },
    {
      "epoch": 0.13024093672596262,
      "grad_norm": 1.2052149772644043,
      "learning_rate": 9.898393223771656e-05,
      "loss": 1.624,
      "step": 723
    },
    {
      "epoch": 0.13042107633415898,
      "grad_norm": 1.35765540599823,
      "learning_rate": 9.898108984558261e-05,
      "loss": 1.9485,
      "step": 724
    },
    {
      "epoch": 0.13060121594235533,
      "grad_norm": 1.1707983016967773,
      "learning_rate": 9.897824352420946e-05,
      "loss": 1.8295,
      "step": 725
    },
    {
      "epoch": 0.13078135555055168,
      "grad_norm": 1.2256286144256592,
      "learning_rate": 9.89753932738255e-05,
      "loss": 1.8786,
      "step": 726
    },
    {
      "epoch": 0.13096149515874803,
      "grad_norm": 1.2813303470611572,
      "learning_rate": 9.897253909465934e-05,
      "loss": 2.0545,
      "step": 727
    },
    {
      "epoch": 0.13114163476694438,
      "grad_norm": 1.2567967176437378,
      "learning_rate": 9.896968098693993e-05,
      "loss": 2.0479,
      "step": 728
    },
    {
      "epoch": 0.13132177437514073,
      "grad_norm": 1.2023850679397583,
      "learning_rate": 9.896681895089657e-05,
      "loss": 1.6083,
      "step": 729
    },
    {
      "epoch": 0.13150191398333708,
      "grad_norm": 1.270164966583252,
      "learning_rate": 9.896395298675885e-05,
      "loss": 2.1973,
      "step": 730
    },
    {
      "epoch": 0.13168205359153343,
      "grad_norm": 1.2565652132034302,
      "learning_rate": 9.896108309475666e-05,
      "loss": 1.7755,
      "step": 731
    },
    {
      "epoch": 0.13186219319972978,
      "grad_norm": 1.235974907875061,
      "learning_rate": 9.895820927512025e-05,
      "loss": 1.8644,
      "step": 732
    },
    {
      "epoch": 0.13204233280792615,
      "grad_norm": 1.3333117961883545,
      "learning_rate": 9.895533152808013e-05,
      "loss": 2.1768,
      "step": 733
    },
    {
      "epoch": 0.1322224724161225,
      "grad_norm": 1.1892369985580444,
      "learning_rate": 9.895244985386715e-05,
      "loss": 1.8334,
      "step": 734
    },
    {
      "epoch": 0.13240261202431886,
      "grad_norm": 1.1675918102264404,
      "learning_rate": 9.894956425271251e-05,
      "loss": 1.8035,
      "step": 735
    },
    {
      "epoch": 0.1325827516325152,
      "grad_norm": 1.21050226688385,
      "learning_rate": 9.894667472484765e-05,
      "loss": 1.8119,
      "step": 736
    },
    {
      "epoch": 0.13276289124071156,
      "grad_norm": 1.3085192441940308,
      "learning_rate": 9.894378127050439e-05,
      "loss": 2.0245,
      "step": 737
    },
    {
      "epoch": 0.1329430308489079,
      "grad_norm": 1.241247296333313,
      "learning_rate": 9.894088388991483e-05,
      "loss": 1.7271,
      "step": 738
    },
    {
      "epoch": 0.13312317045710426,
      "grad_norm": 1.2851693630218506,
      "learning_rate": 9.893798258331142e-05,
      "loss": 2.0219,
      "step": 739
    },
    {
      "epoch": 0.1333033100653006,
      "grad_norm": 1.2805635929107666,
      "learning_rate": 9.893507735092687e-05,
      "loss": 1.9879,
      "step": 740
    },
    {
      "epoch": 0.13348344967349696,
      "grad_norm": 1.2987934350967407,
      "learning_rate": 9.893216819299425e-05,
      "loss": 2.0181,
      "step": 741
    },
    {
      "epoch": 0.1336635892816933,
      "grad_norm": 1.425493597984314,
      "learning_rate": 9.892925510974693e-05,
      "loss": 1.9566,
      "step": 742
    },
    {
      "epoch": 0.13384372888988966,
      "grad_norm": 1.3339426517486572,
      "learning_rate": 9.89263381014186e-05,
      "loss": 1.8115,
      "step": 743
    },
    {
      "epoch": 0.134023868498086,
      "grad_norm": 1.229731559753418,
      "learning_rate": 9.892341716824327e-05,
      "loss": 1.7261,
      "step": 744
    },
    {
      "epoch": 0.13420400810628236,
      "grad_norm": 1.5051815509796143,
      "learning_rate": 9.892049231045524e-05,
      "loss": 2.0755,
      "step": 745
    },
    {
      "epoch": 0.1343841477144787,
      "grad_norm": 1.2894476652145386,
      "learning_rate": 9.891756352828914e-05,
      "loss": 1.954,
      "step": 746
    },
    {
      "epoch": 0.13456428732267509,
      "grad_norm": 1.5340901613235474,
      "learning_rate": 9.891463082197992e-05,
      "loss": 1.9708,
      "step": 747
    },
    {
      "epoch": 0.13474442693087144,
      "grad_norm": 1.2935574054718018,
      "learning_rate": 9.891169419176283e-05,
      "loss": 1.7517,
      "step": 748
    },
    {
      "epoch": 0.1349245665390678,
      "grad_norm": 1.2900742292404175,
      "learning_rate": 9.890875363787347e-05,
      "loss": 1.7476,
      "step": 749
    },
    {
      "epoch": 0.13510470614726414,
      "grad_norm": 1.5172815322875977,
      "learning_rate": 9.890580916054772e-05,
      "loss": 1.905,
      "step": 750
    },
    {
      "epoch": 0.1352848457554605,
      "grad_norm": 1.405549168586731,
      "learning_rate": 9.890286076002178e-05,
      "loss": 2.2225,
      "step": 751
    },
    {
      "epoch": 0.13546498536365684,
      "grad_norm": 1.4486700296401978,
      "learning_rate": 9.889990843653216e-05,
      "loss": 2.3053,
      "step": 752
    },
    {
      "epoch": 0.1356451249718532,
      "grad_norm": 1.4098986387252808,
      "learning_rate": 9.88969521903157e-05,
      "loss": 2.1699,
      "step": 753
    },
    {
      "epoch": 0.13582526458004954,
      "grad_norm": 1.4319884777069092,
      "learning_rate": 9.889399202160956e-05,
      "loss": 2.6816,
      "step": 754
    },
    {
      "epoch": 0.1360054041882459,
      "grad_norm": 1.3512017726898193,
      "learning_rate": 9.889102793065119e-05,
      "loss": 2.3135,
      "step": 755
    },
    {
      "epoch": 0.13618554379644224,
      "grad_norm": 1.548256278038025,
      "learning_rate": 9.888805991767838e-05,
      "loss": 2.1343,
      "step": 756
    },
    {
      "epoch": 0.1363656834046386,
      "grad_norm": 1.5729053020477295,
      "learning_rate": 9.888508798292921e-05,
      "loss": 2.3763,
      "step": 757
    },
    {
      "epoch": 0.13654582301283494,
      "grad_norm": 1.7666906118392944,
      "learning_rate": 9.888211212664211e-05,
      "loss": 2.0297,
      "step": 758
    },
    {
      "epoch": 0.1367259626210313,
      "grad_norm": 2.469851493835449,
      "learning_rate": 9.887913234905577e-05,
      "loss": 2.4429,
      "step": 759
    },
    {
      "epoch": 0.13690610222922764,
      "grad_norm": 1.450822353363037,
      "learning_rate": 9.887614865040924e-05,
      "loss": 2.1066,
      "step": 760
    },
    {
      "epoch": 0.13708624183742402,
      "grad_norm": 1.3207193613052368,
      "learning_rate": 9.887316103094188e-05,
      "loss": 1.7213,
      "step": 761
    },
    {
      "epoch": 0.13726638144562037,
      "grad_norm": 1.3421610593795776,
      "learning_rate": 9.887016949089333e-05,
      "loss": 1.8533,
      "step": 762
    },
    {
      "epoch": 0.13744652105381672,
      "grad_norm": 1.3048672676086426,
      "learning_rate": 9.88671740305036e-05,
      "loss": 1.7998,
      "step": 763
    },
    {
      "epoch": 0.13762666066201307,
      "grad_norm": 1.3105459213256836,
      "learning_rate": 9.886417465001299e-05,
      "loss": 1.6609,
      "step": 764
    },
    {
      "epoch": 0.13780680027020942,
      "grad_norm": 1.2985025644302368,
      "learning_rate": 9.886117134966207e-05,
      "loss": 1.628,
      "step": 765
    },
    {
      "epoch": 0.13798693987840577,
      "grad_norm": 1.3812878131866455,
      "learning_rate": 9.885816412969178e-05,
      "loss": 1.8243,
      "step": 766
    },
    {
      "epoch": 0.13816707948660212,
      "grad_norm": 1.3445358276367188,
      "learning_rate": 9.885515299034336e-05,
      "loss": 1.9206,
      "step": 767
    },
    {
      "epoch": 0.13834721909479847,
      "grad_norm": 1.2129161357879639,
      "learning_rate": 9.885213793185837e-05,
      "loss": 1.8118,
      "step": 768
    },
    {
      "epoch": 0.13852735870299482,
      "grad_norm": 1.299033761024475,
      "learning_rate": 9.884911895447867e-05,
      "loss": 1.8389,
      "step": 769
    },
    {
      "epoch": 0.13870749831119117,
      "grad_norm": 1.2037889957427979,
      "learning_rate": 9.884609605844644e-05,
      "loss": 1.6472,
      "step": 770
    },
    {
      "epoch": 0.13888763791938752,
      "grad_norm": 1.268749713897705,
      "learning_rate": 9.884306924400418e-05,
      "loss": 1.923,
      "step": 771
    },
    {
      "epoch": 0.13906777752758387,
      "grad_norm": 1.2923638820648193,
      "learning_rate": 9.88400385113947e-05,
      "loss": 1.9507,
      "step": 772
    },
    {
      "epoch": 0.13924791713578022,
      "grad_norm": 1.2170230150222778,
      "learning_rate": 9.883700386086111e-05,
      "loss": 2.1415,
      "step": 773
    },
    {
      "epoch": 0.13942805674397657,
      "grad_norm": 1.2914255857467651,
      "learning_rate": 9.883396529264686e-05,
      "loss": 1.8496,
      "step": 774
    },
    {
      "epoch": 0.13960819635217295,
      "grad_norm": 1.294809103012085,
      "learning_rate": 9.883092280699571e-05,
      "loss": 2.0246,
      "step": 775
    },
    {
      "epoch": 0.1397883359603693,
      "grad_norm": 1.315855622291565,
      "learning_rate": 9.882787640415169e-05,
      "loss": 1.7098,
      "step": 776
    },
    {
      "epoch": 0.13996847556856565,
      "grad_norm": 1.3406991958618164,
      "learning_rate": 9.882482608435923e-05,
      "loss": 1.9858,
      "step": 777
    },
    {
      "epoch": 0.140148615176762,
      "grad_norm": 1.2104747295379639,
      "learning_rate": 9.882177184786302e-05,
      "loss": 1.9567,
      "step": 778
    },
    {
      "epoch": 0.14032875478495835,
      "grad_norm": 1.2148669958114624,
      "learning_rate": 9.881871369490802e-05,
      "loss": 2.0177,
      "step": 779
    },
    {
      "epoch": 0.1405088943931547,
      "grad_norm": 1.4571545124053955,
      "learning_rate": 9.881565162573961e-05,
      "loss": 1.8855,
      "step": 780
    },
    {
      "epoch": 0.14068903400135105,
      "grad_norm": 1.2827662229537964,
      "learning_rate": 9.881258564060338e-05,
      "loss": 1.7891,
      "step": 781
    },
    {
      "epoch": 0.1408691736095474,
      "grad_norm": 1.3293724060058594,
      "learning_rate": 9.880951573974533e-05,
      "loss": 2.0578,
      "step": 782
    },
    {
      "epoch": 0.14104931321774375,
      "grad_norm": 1.3720998764038086,
      "learning_rate": 9.880644192341169e-05,
      "loss": 2.0085,
      "step": 783
    },
    {
      "epoch": 0.1412294528259401,
      "grad_norm": 1.2281889915466309,
      "learning_rate": 9.880336419184905e-05,
      "loss": 1.8419,
      "step": 784
    },
    {
      "epoch": 0.14140959243413645,
      "grad_norm": 1.2522205114364624,
      "learning_rate": 9.88002825453043e-05,
      "loss": 2.0256,
      "step": 785
    },
    {
      "epoch": 0.1415897320423328,
      "grad_norm": 1.2879754304885864,
      "learning_rate": 9.879719698402467e-05,
      "loss": 1.8484,
      "step": 786
    },
    {
      "epoch": 0.14176987165052915,
      "grad_norm": 1.3683329820632935,
      "learning_rate": 9.879410750825766e-05,
      "loss": 2.2531,
      "step": 787
    },
    {
      "epoch": 0.1419500112587255,
      "grad_norm": 1.1800748109817505,
      "learning_rate": 9.879101411825111e-05,
      "loss": 1.6993,
      "step": 788
    },
    {
      "epoch": 0.14213015086692188,
      "grad_norm": 1.282247543334961,
      "learning_rate": 9.878791681425317e-05,
      "loss": 1.7951,
      "step": 789
    },
    {
      "epoch": 0.14231029047511823,
      "grad_norm": 1.316503882408142,
      "learning_rate": 9.87848155965123e-05,
      "loss": 1.7466,
      "step": 790
    },
    {
      "epoch": 0.14249043008331458,
      "grad_norm": 1.4377599954605103,
      "learning_rate": 9.87817104652773e-05,
      "loss": 2.1601,
      "step": 791
    },
    {
      "epoch": 0.14267056969151093,
      "grad_norm": 1.289066195487976,
      "learning_rate": 9.877860142079725e-05,
      "loss": 2.1348,
      "step": 792
    },
    {
      "epoch": 0.14285070929970728,
      "grad_norm": 1.2391852140426636,
      "learning_rate": 9.877548846332154e-05,
      "loss": 1.8587,
      "step": 793
    },
    {
      "epoch": 0.14303084890790363,
      "grad_norm": 1.2917535305023193,
      "learning_rate": 9.877237159309991e-05,
      "loss": 1.9722,
      "step": 794
    },
    {
      "epoch": 0.14321098851609998,
      "grad_norm": 1.6067614555358887,
      "learning_rate": 9.876925081038238e-05,
      "loss": 2.2133,
      "step": 795
    },
    {
      "epoch": 0.14339112812429633,
      "grad_norm": 1.5059887170791626,
      "learning_rate": 9.87661261154193e-05,
      "loss": 1.9933,
      "step": 796
    },
    {
      "epoch": 0.14357126773249268,
      "grad_norm": 1.3656240701675415,
      "learning_rate": 9.876299750846136e-05,
      "loss": 1.7474,
      "step": 797
    },
    {
      "epoch": 0.14375140734068903,
      "grad_norm": 1.4455832242965698,
      "learning_rate": 9.87598649897595e-05,
      "loss": 1.7519,
      "step": 798
    },
    {
      "epoch": 0.14393154694888538,
      "grad_norm": 1.288589596748352,
      "learning_rate": 9.875672855956501e-05,
      "loss": 1.7428,
      "step": 799
    },
    {
      "epoch": 0.14411168655708173,
      "grad_norm": 1.4480016231536865,
      "learning_rate": 9.875358821812953e-05,
      "loss": 1.7862,
      "step": 800
    },
    {
      "epoch": 0.14429182616527808,
      "grad_norm": 1.372703194618225,
      "learning_rate": 9.875044396570493e-05,
      "loss": 2.5034,
      "step": 801
    },
    {
      "epoch": 0.14447196577347443,
      "grad_norm": 1.4766262769699097,
      "learning_rate": 9.874729580254345e-05,
      "loss": 2.696,
      "step": 802
    },
    {
      "epoch": 0.14465210538167078,
      "grad_norm": 1.2882176637649536,
      "learning_rate": 9.874414372889767e-05,
      "loss": 2.3181,
      "step": 803
    },
    {
      "epoch": 0.14483224498986716,
      "grad_norm": 1.3298519849777222,
      "learning_rate": 9.87409877450204e-05,
      "loss": 2.3199,
      "step": 804
    },
    {
      "epoch": 0.1450123845980635,
      "grad_norm": 1.3550612926483154,
      "learning_rate": 9.873782785116484e-05,
      "loss": 2.4582,
      "step": 805
    },
    {
      "epoch": 0.14519252420625986,
      "grad_norm": 1.4804792404174805,
      "learning_rate": 9.873466404758446e-05,
      "loss": 2.2796,
      "step": 806
    },
    {
      "epoch": 0.1453726638144562,
      "grad_norm": 1.3525737524032593,
      "learning_rate": 9.873149633453308e-05,
      "loss": 2.0421,
      "step": 807
    },
    {
      "epoch": 0.14555280342265256,
      "grad_norm": 1.615544319152832,
      "learning_rate": 9.872832471226482e-05,
      "loss": 2.4053,
      "step": 808
    },
    {
      "epoch": 0.1457329430308489,
      "grad_norm": 1.6876574754714966,
      "learning_rate": 9.872514918103406e-05,
      "loss": 2.3352,
      "step": 809
    },
    {
      "epoch": 0.14591308263904526,
      "grad_norm": 2.0150530338287354,
      "learning_rate": 9.872196974109558e-05,
      "loss": 2.4234,
      "step": 810
    },
    {
      "epoch": 0.1460932222472416,
      "grad_norm": 1.2573726177215576,
      "learning_rate": 9.871878639270441e-05,
      "loss": 1.7109,
      "step": 811
    },
    {
      "epoch": 0.14627336185543796,
      "grad_norm": 1.3346951007843018,
      "learning_rate": 9.871559913611592e-05,
      "loss": 1.9011,
      "step": 812
    },
    {
      "epoch": 0.1464535014636343,
      "grad_norm": 1.3603603839874268,
      "learning_rate": 9.871240797158581e-05,
      "loss": 2.0636,
      "step": 813
    },
    {
      "epoch": 0.14663364107183066,
      "grad_norm": 1.3373414278030396,
      "learning_rate": 9.870921289937006e-05,
      "loss": 1.9809,
      "step": 814
    },
    {
      "epoch": 0.146813780680027,
      "grad_norm": 1.360762357711792,
      "learning_rate": 9.870601391972496e-05,
      "loss": 2.0151,
      "step": 815
    },
    {
      "epoch": 0.14699392028822336,
      "grad_norm": 1.4189239740371704,
      "learning_rate": 9.870281103290717e-05,
      "loss": 1.9082,
      "step": 816
    },
    {
      "epoch": 0.1471740598964197,
      "grad_norm": 1.1659568548202515,
      "learning_rate": 9.869960423917357e-05,
      "loss": 1.7785,
      "step": 817
    },
    {
      "epoch": 0.1473541995046161,
      "grad_norm": 1.305368423461914,
      "learning_rate": 9.869639353878146e-05,
      "loss": 1.9978,
      "step": 818
    },
    {
      "epoch": 0.14753433911281244,
      "grad_norm": 1.2636337280273438,
      "learning_rate": 9.869317893198838e-05,
      "loss": 1.7232,
      "step": 819
    },
    {
      "epoch": 0.1477144787210088,
      "grad_norm": 1.2035678625106812,
      "learning_rate": 9.86899604190522e-05,
      "loss": 1.7567,
      "step": 820
    },
    {
      "epoch": 0.14789461832920514,
      "grad_norm": 1.2082552909851074,
      "learning_rate": 9.868673800023111e-05,
      "loss": 2.0055,
      "step": 821
    },
    {
      "epoch": 0.1480747579374015,
      "grad_norm": 1.3273311853408813,
      "learning_rate": 9.868351167578363e-05,
      "loss": 1.9704,
      "step": 822
    },
    {
      "epoch": 0.14825489754559784,
      "grad_norm": 1.3554540872573853,
      "learning_rate": 9.868028144596852e-05,
      "loss": 2.0078,
      "step": 823
    },
    {
      "epoch": 0.1484350371537942,
      "grad_norm": 1.3239822387695312,
      "learning_rate": 9.867704731104496e-05,
      "loss": 1.8729,
      "step": 824
    },
    {
      "epoch": 0.14861517676199054,
      "grad_norm": 1.246198296546936,
      "learning_rate": 9.867380927127239e-05,
      "loss": 1.8706,
      "step": 825
    },
    {
      "epoch": 0.1487953163701869,
      "grad_norm": 1.2795337438583374,
      "learning_rate": 9.867056732691054e-05,
      "loss": 2.0535,
      "step": 826
    },
    {
      "epoch": 0.14897545597838324,
      "grad_norm": 1.2475967407226562,
      "learning_rate": 9.866732147821947e-05,
      "loss": 1.7275,
      "step": 827
    },
    {
      "epoch": 0.1491555955865796,
      "grad_norm": 1.3331667184829712,
      "learning_rate": 9.86640717254596e-05,
      "loss": 1.8053,
      "step": 828
    },
    {
      "epoch": 0.14933573519477594,
      "grad_norm": 1.2305011749267578,
      "learning_rate": 9.866081806889158e-05,
      "loss": 1.7696,
      "step": 829
    },
    {
      "epoch": 0.1495158748029723,
      "grad_norm": 1.368841528892517,
      "learning_rate": 9.865756050877645e-05,
      "loss": 2.1196,
      "step": 830
    },
    {
      "epoch": 0.14969601441116864,
      "grad_norm": 1.2340091466903687,
      "learning_rate": 9.86542990453755e-05,
      "loss": 1.7208,
      "step": 831
    },
    {
      "epoch": 0.14987615401936502,
      "grad_norm": 1.3266440629959106,
      "learning_rate": 9.865103367895039e-05,
      "loss": 2.0684,
      "step": 832
    },
    {
      "epoch": 0.15005629362756137,
      "grad_norm": 1.3821064233779907,
      "learning_rate": 9.864776440976306e-05,
      "loss": 2.1309,
      "step": 833
    },
    {
      "epoch": 0.15023643323575772,
      "grad_norm": 1.1727502346038818,
      "learning_rate": 9.864449123807576e-05,
      "loss": 1.6972,
      "step": 834
    },
    {
      "epoch": 0.15041657284395407,
      "grad_norm": 1.2775639295578003,
      "learning_rate": 9.864121416415106e-05,
      "loss": 1.8604,
      "step": 835
    },
    {
      "epoch": 0.15059671245215042,
      "grad_norm": 1.3789913654327393,
      "learning_rate": 9.863793318825186e-05,
      "loss": 1.9378,
      "step": 836
    },
    {
      "epoch": 0.15077685206034677,
      "grad_norm": 1.3935092687606812,
      "learning_rate": 9.863464831064135e-05,
      "loss": 1.8925,
      "step": 837
    },
    {
      "epoch": 0.15095699166854312,
      "grad_norm": 1.3984404802322388,
      "learning_rate": 9.863135953158305e-05,
      "loss": 1.9013,
      "step": 838
    },
    {
      "epoch": 0.15113713127673947,
      "grad_norm": 1.3366034030914307,
      "learning_rate": 9.862806685134077e-05,
      "loss": 2.0315,
      "step": 839
    },
    {
      "epoch": 0.15131727088493582,
      "grad_norm": 1.4392826557159424,
      "learning_rate": 9.862477027017864e-05,
      "loss": 2.0174,
      "step": 840
    },
    {
      "epoch": 0.15149741049313217,
      "grad_norm": 1.4083927869796753,
      "learning_rate": 9.862146978836113e-05,
      "loss": 2.0956,
      "step": 841
    },
    {
      "epoch": 0.15167755010132852,
      "grad_norm": 1.296589970588684,
      "learning_rate": 9.861816540615299e-05,
      "loss": 1.9268,
      "step": 842
    },
    {
      "epoch": 0.15185768970952487,
      "grad_norm": 1.2528111934661865,
      "learning_rate": 9.861485712381931e-05,
      "loss": 1.923,
      "step": 843
    },
    {
      "epoch": 0.15203782931772122,
      "grad_norm": 1.3625739812850952,
      "learning_rate": 9.861154494162548e-05,
      "loss": 2.1309,
      "step": 844
    },
    {
      "epoch": 0.15221796892591757,
      "grad_norm": 1.4396804571151733,
      "learning_rate": 9.860822885983718e-05,
      "loss": 1.8609,
      "step": 845
    },
    {
      "epoch": 0.15239810853411395,
      "grad_norm": 1.4618682861328125,
      "learning_rate": 9.860490887872045e-05,
      "loss": 2.0644,
      "step": 846
    },
    {
      "epoch": 0.1525782481423103,
      "grad_norm": 1.2852728366851807,
      "learning_rate": 9.860158499854159e-05,
      "loss": 1.8212,
      "step": 847
    },
    {
      "epoch": 0.15275838775050665,
      "grad_norm": 1.3814423084259033,
      "learning_rate": 9.859825721956725e-05,
      "loss": 1.9965,
      "step": 848
    },
    {
      "epoch": 0.152938527358703,
      "grad_norm": 1.4789983034133911,
      "learning_rate": 9.85949255420644e-05,
      "loss": 2.0927,
      "step": 849
    },
    {
      "epoch": 0.15311866696689935,
      "grad_norm": 1.369570255279541,
      "learning_rate": 9.859158996630029e-05,
      "loss": 1.9551,
      "step": 850
    },
    {
      "epoch": 0.1532988065750957,
      "grad_norm": 1.4670217037200928,
      "learning_rate": 9.85882504925425e-05,
      "loss": 2.3491,
      "step": 851
    },
    {
      "epoch": 0.15347894618329205,
      "grad_norm": 1.451915979385376,
      "learning_rate": 9.858490712105892e-05,
      "loss": 2.3814,
      "step": 852
    },
    {
      "epoch": 0.1536590857914884,
      "grad_norm": 1.3465337753295898,
      "learning_rate": 9.858155985211776e-05,
      "loss": 2.3684,
      "step": 853
    },
    {
      "epoch": 0.15383922539968475,
      "grad_norm": 1.309025526046753,
      "learning_rate": 9.857820868598753e-05,
      "loss": 2.166,
      "step": 854
    },
    {
      "epoch": 0.1540193650078811,
      "grad_norm": 1.430438756942749,
      "learning_rate": 9.857485362293707e-05,
      "loss": 2.3894,
      "step": 855
    },
    {
      "epoch": 0.15419950461607745,
      "grad_norm": 1.4471404552459717,
      "learning_rate": 9.857149466323549e-05,
      "loss": 2.2494,
      "step": 856
    },
    {
      "epoch": 0.1543796442242738,
      "grad_norm": 1.853070855140686,
      "learning_rate": 9.856813180715229e-05,
      "loss": 2.4151,
      "step": 857
    },
    {
      "epoch": 0.15455978383247015,
      "grad_norm": 1.7586486339569092,
      "learning_rate": 9.856476505495721e-05,
      "loss": 2.3734,
      "step": 858
    },
    {
      "epoch": 0.1547399234406665,
      "grad_norm": 5.4407501220703125,
      "learning_rate": 9.856139440692031e-05,
      "loss": 2.6805,
      "step": 859
    },
    {
      "epoch": 0.15492006304886285,
      "grad_norm": 1.71763014793396,
      "learning_rate": 9.855801986331203e-05,
      "loss": 2.3634,
      "step": 860
    },
    {
      "epoch": 0.15510020265705923,
      "grad_norm": 1.3074826002120972,
      "learning_rate": 9.855464142440303e-05,
      "loss": 2.038,
      "step": 861
    },
    {
      "epoch": 0.15528034226525558,
      "grad_norm": 1.149901032447815,
      "learning_rate": 9.855125909046436e-05,
      "loss": 1.601,
      "step": 862
    },
    {
      "epoch": 0.15546048187345193,
      "grad_norm": 1.2345138788223267,
      "learning_rate": 9.854787286176732e-05,
      "loss": 1.832,
      "step": 863
    },
    {
      "epoch": 0.15564062148164828,
      "grad_norm": 1.199136734008789,
      "learning_rate": 9.854448273858359e-05,
      "loss": 1.7501,
      "step": 864
    },
    {
      "epoch": 0.15582076108984463,
      "grad_norm": 1.298190951347351,
      "learning_rate": 9.854108872118508e-05,
      "loss": 1.8705,
      "step": 865
    },
    {
      "epoch": 0.15600090069804098,
      "grad_norm": 1.2769882678985596,
      "learning_rate": 9.853769080984407e-05,
      "loss": 1.9791,
      "step": 866
    },
    {
      "epoch": 0.15618104030623733,
      "grad_norm": 1.2338829040527344,
      "learning_rate": 9.853428900483316e-05,
      "loss": 1.9113,
      "step": 867
    },
    {
      "epoch": 0.15636117991443368,
      "grad_norm": 1.3071916103363037,
      "learning_rate": 9.853088330642522e-05,
      "loss": 1.719,
      "step": 868
    },
    {
      "epoch": 0.15654131952263003,
      "grad_norm": 1.2592660188674927,
      "learning_rate": 9.852747371489346e-05,
      "loss": 1.7682,
      "step": 869
    },
    {
      "epoch": 0.15672145913082638,
      "grad_norm": 1.2901333570480347,
      "learning_rate": 9.85240602305114e-05,
      "loss": 2.0187,
      "step": 870
    },
    {
      "epoch": 0.15690159873902274,
      "grad_norm": 1.2896544933319092,
      "learning_rate": 9.852064285355285e-05,
      "loss": 1.8367,
      "step": 871
    },
    {
      "epoch": 0.15708173834721909,
      "grad_norm": 1.3103440999984741,
      "learning_rate": 9.851722158429198e-05,
      "loss": 1.879,
      "step": 872
    },
    {
      "epoch": 0.15726187795541544,
      "grad_norm": 1.1968687772750854,
      "learning_rate": 9.851379642300323e-05,
      "loss": 1.9501,
      "step": 873
    },
    {
      "epoch": 0.15744201756361179,
      "grad_norm": 1.1210064888000488,
      "learning_rate": 9.851036736996133e-05,
      "loss": 1.5667,
      "step": 874
    },
    {
      "epoch": 0.15762215717180816,
      "grad_norm": 1.279301404953003,
      "learning_rate": 9.850693442544141e-05,
      "loss": 2.0346,
      "step": 875
    },
    {
      "epoch": 0.15780229678000451,
      "grad_norm": 1.2971493005752563,
      "learning_rate": 9.850349758971883e-05,
      "loss": 1.7209,
      "step": 876
    },
    {
      "epoch": 0.15798243638820086,
      "grad_norm": 1.4926193952560425,
      "learning_rate": 9.850005686306932e-05,
      "loss": 2.3681,
      "step": 877
    },
    {
      "epoch": 0.15816257599639721,
      "grad_norm": 1.486655831336975,
      "learning_rate": 9.849661224576885e-05,
      "loss": 2.2133,
      "step": 878
    },
    {
      "epoch": 0.15834271560459356,
      "grad_norm": 1.5211142301559448,
      "learning_rate": 9.849316373809378e-05,
      "loss": 2.1244,
      "step": 879
    },
    {
      "epoch": 0.15852285521278991,
      "grad_norm": 1.3286242485046387,
      "learning_rate": 9.848971134032072e-05,
      "loss": 2.1348,
      "step": 880
    },
    {
      "epoch": 0.15870299482098627,
      "grad_norm": 1.3326750993728638,
      "learning_rate": 9.848625505272665e-05,
      "loss": 2.2662,
      "step": 881
    },
    {
      "epoch": 0.15888313442918262,
      "grad_norm": 1.3165205717086792,
      "learning_rate": 9.848279487558881e-05,
      "loss": 1.9833,
      "step": 882
    },
    {
      "epoch": 0.15906327403737897,
      "grad_norm": 1.3390347957611084,
      "learning_rate": 9.847933080918479e-05,
      "loss": 1.8351,
      "step": 883
    },
    {
      "epoch": 0.15924341364557532,
      "grad_norm": 1.303579568862915,
      "learning_rate": 9.847586285379244e-05,
      "loss": 1.9421,
      "step": 884
    },
    {
      "epoch": 0.15942355325377167,
      "grad_norm": 1.2818201780319214,
      "learning_rate": 9.847239100969001e-05,
      "loss": 1.9231,
      "step": 885
    },
    {
      "epoch": 0.15960369286196802,
      "grad_norm": 1.3924510478973389,
      "learning_rate": 9.846891527715597e-05,
      "loss": 2.1783,
      "step": 886
    },
    {
      "epoch": 0.15978383247016437,
      "grad_norm": 1.353502869606018,
      "learning_rate": 9.846543565646915e-05,
      "loss": 2.0035,
      "step": 887
    },
    {
      "epoch": 0.15996397207836072,
      "grad_norm": 1.3508769273757935,
      "learning_rate": 9.84619521479087e-05,
      "loss": 2.0585,
      "step": 888
    },
    {
      "epoch": 0.1601441116865571,
      "grad_norm": 1.3455395698547363,
      "learning_rate": 9.845846475175406e-05,
      "loss": 1.915,
      "step": 889
    },
    {
      "epoch": 0.16032425129475344,
      "grad_norm": 1.3258179426193237,
      "learning_rate": 9.845497346828496e-05,
      "loss": 1.9067,
      "step": 890
    },
    {
      "epoch": 0.1605043909029498,
      "grad_norm": 1.2612024545669556,
      "learning_rate": 9.84514782977815e-05,
      "loss": 1.8582,
      "step": 891
    },
    {
      "epoch": 0.16068453051114615,
      "grad_norm": 1.455338478088379,
      "learning_rate": 9.844797924052406e-05,
      "loss": 2.2347,
      "step": 892
    },
    {
      "epoch": 0.1608646701193425,
      "grad_norm": 1.5008792877197266,
      "learning_rate": 9.844447629679331e-05,
      "loss": 2.1904,
      "step": 893
    },
    {
      "epoch": 0.16104480972753885,
      "grad_norm": 1.4137682914733887,
      "learning_rate": 9.844096946687028e-05,
      "loss": 1.9771,
      "step": 894
    },
    {
      "epoch": 0.1612249493357352,
      "grad_norm": 1.2500028610229492,
      "learning_rate": 9.843745875103627e-05,
      "loss": 2.134,
      "step": 895
    },
    {
      "epoch": 0.16140508894393155,
      "grad_norm": 1.375946283340454,
      "learning_rate": 9.843394414957289e-05,
      "loss": 1.8825,
      "step": 896
    },
    {
      "epoch": 0.1615852285521279,
      "grad_norm": 1.4914226531982422,
      "learning_rate": 9.843042566276214e-05,
      "loss": 2.0495,
      "step": 897
    },
    {
      "epoch": 0.16176536816032425,
      "grad_norm": 1.3528846502304077,
      "learning_rate": 9.842690329088622e-05,
      "loss": 1.857,
      "step": 898
    },
    {
      "epoch": 0.1619455077685206,
      "grad_norm": 1.4280825853347778,
      "learning_rate": 9.84233770342277e-05,
      "loss": 1.7073,
      "step": 899
    },
    {
      "epoch": 0.16212564737671695,
      "grad_norm": 1.6007810831069946,
      "learning_rate": 9.841984689306945e-05,
      "loss": 1.743,
      "step": 900
    },
    {
      "epoch": 0.1623057869849133,
      "grad_norm": 1.26369309425354,
      "learning_rate": 9.841631286769468e-05,
      "loss": 2.4067,
      "step": 901
    },
    {
      "epoch": 0.16248592659310965,
      "grad_norm": 1.3648098707199097,
      "learning_rate": 9.841277495838688e-05,
      "loss": 2.3882,
      "step": 902
    },
    {
      "epoch": 0.16266606620130603,
      "grad_norm": 1.3654829263687134,
      "learning_rate": 9.840923316542983e-05,
      "loss": 2.2245,
      "step": 903
    },
    {
      "epoch": 0.16284620580950238,
      "grad_norm": 1.4620901346206665,
      "learning_rate": 9.84056874891077e-05,
      "loss": 2.1419,
      "step": 904
    },
    {
      "epoch": 0.16302634541769873,
      "grad_norm": 1.4209767580032349,
      "learning_rate": 9.840213792970489e-05,
      "loss": 2.4339,
      "step": 905
    },
    {
      "epoch": 0.16320648502589508,
      "grad_norm": 1.3212268352508545,
      "learning_rate": 9.839858448750614e-05,
      "loss": 2.2154,
      "step": 906
    },
    {
      "epoch": 0.16338662463409143,
      "grad_norm": 1.3348406553268433,
      "learning_rate": 9.839502716279653e-05,
      "loss": 2.0729,
      "step": 907
    },
    {
      "epoch": 0.16356676424228778,
      "grad_norm": 1.3997691869735718,
      "learning_rate": 9.83914659558614e-05,
      "loss": 2.1104,
      "step": 908
    },
    {
      "epoch": 0.16374690385048413,
      "grad_norm": 1.744264006614685,
      "learning_rate": 9.838790086698646e-05,
      "loss": 2.2759,
      "step": 909
    },
    {
      "epoch": 0.16392704345868048,
      "grad_norm": 2.3655107021331787,
      "learning_rate": 9.838433189645768e-05,
      "loss": 2.2623,
      "step": 910
    },
    {
      "epoch": 0.16410718306687683,
      "grad_norm": 1.341048240661621,
      "learning_rate": 9.838075904456134e-05,
      "loss": 2.057,
      "step": 911
    },
    {
      "epoch": 0.16428732267507318,
      "grad_norm": 1.2010446786880493,
      "learning_rate": 9.83771823115841e-05,
      "loss": 1.8994,
      "step": 912
    },
    {
      "epoch": 0.16446746228326953,
      "grad_norm": 1.3203390836715698,
      "learning_rate": 9.837360169781285e-05,
      "loss": 1.8322,
      "step": 913
    },
    {
      "epoch": 0.16464760189146588,
      "grad_norm": 1.4009171724319458,
      "learning_rate": 9.837001720353483e-05,
      "loss": 2.0131,
      "step": 914
    },
    {
      "epoch": 0.16482774149966223,
      "grad_norm": 1.2876688241958618,
      "learning_rate": 9.836642882903759e-05,
      "loss": 1.807,
      "step": 915
    },
    {
      "epoch": 0.16500788110785858,
      "grad_norm": 1.2223199605941772,
      "learning_rate": 9.836283657460899e-05,
      "loss": 1.7567,
      "step": 916
    },
    {
      "epoch": 0.16518802071605496,
      "grad_norm": 1.208533525466919,
      "learning_rate": 9.835924044053719e-05,
      "loss": 1.8141,
      "step": 917
    },
    {
      "epoch": 0.1653681603242513,
      "grad_norm": 1.410413384437561,
      "learning_rate": 9.835564042711068e-05,
      "loss": 2.05,
      "step": 918
    },
    {
      "epoch": 0.16554829993244766,
      "grad_norm": 1.1603425741195679,
      "learning_rate": 9.835203653461824e-05,
      "loss": 1.6845,
      "step": 919
    },
    {
      "epoch": 0.165728439540644,
      "grad_norm": 1.264150857925415,
      "learning_rate": 9.8348428763349e-05,
      "loss": 2.1106,
      "step": 920
    },
    {
      "epoch": 0.16590857914884036,
      "grad_norm": 1.314612865447998,
      "learning_rate": 9.834481711359234e-05,
      "loss": 1.8999,
      "step": 921
    },
    {
      "epoch": 0.1660887187570367,
      "grad_norm": 1.2794193029403687,
      "learning_rate": 9.8341201585638e-05,
      "loss": 1.7304,
      "step": 922
    },
    {
      "epoch": 0.16626885836523306,
      "grad_norm": 1.303610920906067,
      "learning_rate": 9.833758217977603e-05,
      "loss": 1.8447,
      "step": 923
    },
    {
      "epoch": 0.1664489979734294,
      "grad_norm": 1.4079903364181519,
      "learning_rate": 9.833395889629675e-05,
      "loss": 1.9543,
      "step": 924
    },
    {
      "epoch": 0.16662913758162576,
      "grad_norm": 1.2935470342636108,
      "learning_rate": 9.833033173549083e-05,
      "loss": 1.7859,
      "step": 925
    },
    {
      "epoch": 0.1668092771898221,
      "grad_norm": 1.25351083278656,
      "learning_rate": 9.832670069764924e-05,
      "loss": 2.0095,
      "step": 926
    },
    {
      "epoch": 0.16698941679801846,
      "grad_norm": 1.3944593667984009,
      "learning_rate": 9.832306578306326e-05,
      "loss": 2.0233,
      "step": 927
    },
    {
      "epoch": 0.1671695564062148,
      "grad_norm": 1.2526030540466309,
      "learning_rate": 9.831942699202448e-05,
      "loss": 1.677,
      "step": 928
    },
    {
      "epoch": 0.16734969601441116,
      "grad_norm": 1.5346907377243042,
      "learning_rate": 9.83157843248248e-05,
      "loss": 1.9569,
      "step": 929
    },
    {
      "epoch": 0.1675298356226075,
      "grad_norm": 1.342502236366272,
      "learning_rate": 9.831213778175644e-05,
      "loss": 2.0435,
      "step": 930
    },
    {
      "epoch": 0.16770997523080386,
      "grad_norm": 1.2255526781082153,
      "learning_rate": 9.830848736311193e-05,
      "loss": 1.7563,
      "step": 931
    },
    {
      "epoch": 0.16789011483900024,
      "grad_norm": 1.268193006515503,
      "learning_rate": 9.830483306918407e-05,
      "loss": 1.54,
      "step": 932
    },
    {
      "epoch": 0.1680702544471966,
      "grad_norm": 1.3726140260696411,
      "learning_rate": 9.830117490026604e-05,
      "loss": 1.7818,
      "step": 933
    },
    {
      "epoch": 0.16825039405539294,
      "grad_norm": 1.3046773672103882,
      "learning_rate": 9.829751285665128e-05,
      "loss": 2.1845,
      "step": 934
    },
    {
      "epoch": 0.1684305336635893,
      "grad_norm": 1.2928993701934814,
      "learning_rate": 9.829384693863357e-05,
      "loss": 1.8105,
      "step": 935
    },
    {
      "epoch": 0.16861067327178564,
      "grad_norm": 1.4520810842514038,
      "learning_rate": 9.829017714650699e-05,
      "loss": 2.1978,
      "step": 936
    },
    {
      "epoch": 0.168790812879982,
      "grad_norm": 1.5034133195877075,
      "learning_rate": 9.828650348056591e-05,
      "loss": 2.1587,
      "step": 937
    },
    {
      "epoch": 0.16897095248817834,
      "grad_norm": 1.3858444690704346,
      "learning_rate": 9.828282594110505e-05,
      "loss": 1.9391,
      "step": 938
    },
    {
      "epoch": 0.1691510920963747,
      "grad_norm": 1.4061371088027954,
      "learning_rate": 9.82791445284194e-05,
      "loss": 1.8593,
      "step": 939
    },
    {
      "epoch": 0.16933123170457104,
      "grad_norm": 1.2276051044464111,
      "learning_rate": 9.82754592428043e-05,
      "loss": 1.6239,
      "step": 940
    },
    {
      "epoch": 0.1695113713127674,
      "grad_norm": 1.2832549810409546,
      "learning_rate": 9.827177008455536e-05,
      "loss": 1.6908,
      "step": 941
    },
    {
      "epoch": 0.16969151092096374,
      "grad_norm": 1.3262428045272827,
      "learning_rate": 9.826807705396854e-05,
      "loss": 1.8058,
      "step": 942
    },
    {
      "epoch": 0.1698716505291601,
      "grad_norm": 1.3945311307907104,
      "learning_rate": 9.82643801513401e-05,
      "loss": 2.1552,
      "step": 943
    },
    {
      "epoch": 0.17005179013735644,
      "grad_norm": 1.315379023551941,
      "learning_rate": 9.826067937696659e-05,
      "loss": 1.9155,
      "step": 944
    },
    {
      "epoch": 0.1702319297455528,
      "grad_norm": 1.521011233329773,
      "learning_rate": 9.825697473114488e-05,
      "loss": 2.152,
      "step": 945
    },
    {
      "epoch": 0.17041206935374917,
      "grad_norm": 1.483184814453125,
      "learning_rate": 9.825326621417216e-05,
      "loss": 2.0287,
      "step": 946
    },
    {
      "epoch": 0.17059220896194552,
      "grad_norm": 1.3443012237548828,
      "learning_rate": 9.824955382634594e-05,
      "loss": 1.6047,
      "step": 947
    },
    {
      "epoch": 0.17077234857014187,
      "grad_norm": 1.398842453956604,
      "learning_rate": 9.824583756796402e-05,
      "loss": 1.9133,
      "step": 948
    },
    {
      "epoch": 0.17095248817833822,
      "grad_norm": 1.4118281602859497,
      "learning_rate": 9.824211743932449e-05,
      "loss": 1.9895,
      "step": 949
    },
    {
      "epoch": 0.17113262778653457,
      "grad_norm": 1.3446736335754395,
      "learning_rate": 9.82383934407258e-05,
      "loss": 1.7331,
      "step": 950
    },
    {
      "epoch": 0.17131276739473092,
      "grad_norm": 1.295842170715332,
      "learning_rate": 9.823466557246668e-05,
      "loss": 2.3869,
      "step": 951
    },
    {
      "epoch": 0.17149290700292727,
      "grad_norm": 1.4812614917755127,
      "learning_rate": 9.82309338348462e-05,
      "loss": 2.3976,
      "step": 952
    },
    {
      "epoch": 0.17167304661112362,
      "grad_norm": 1.270684838294983,
      "learning_rate": 9.822719822816368e-05,
      "loss": 2.3213,
      "step": 953
    },
    {
      "epoch": 0.17185318621931997,
      "grad_norm": 1.2606884241104126,
      "learning_rate": 9.822345875271883e-05,
      "loss": 2.0548,
      "step": 954
    },
    {
      "epoch": 0.17203332582751632,
      "grad_norm": 1.3441598415374756,
      "learning_rate": 9.821971540881161e-05,
      "loss": 2.329,
      "step": 955
    },
    {
      "epoch": 0.17221346543571267,
      "grad_norm": 1.5862243175506592,
      "learning_rate": 9.821596819674228e-05,
      "loss": 2.4331,
      "step": 956
    },
    {
      "epoch": 0.17239360504390902,
      "grad_norm": 1.4265897274017334,
      "learning_rate": 9.821221711681148e-05,
      "loss": 2.3035,
      "step": 957
    },
    {
      "epoch": 0.17257374465210537,
      "grad_norm": 1.9258174896240234,
      "learning_rate": 9.820846216932011e-05,
      "loss": 2.3316,
      "step": 958
    },
    {
      "epoch": 0.17275388426030172,
      "grad_norm": 1.8160074949264526,
      "learning_rate": 9.82047033545694e-05,
      "loss": 2.6953,
      "step": 959
    },
    {
      "epoch": 0.1729340238684981,
      "grad_norm": 2.8916215896606445,
      "learning_rate": 9.820094067286083e-05,
      "loss": 2.4196,
      "step": 960
    },
    {
      "epoch": 0.17311416347669445,
      "grad_norm": 1.324589490890503,
      "learning_rate": 9.819717412449631e-05,
      "loss": 1.8192,
      "step": 961
    },
    {
      "epoch": 0.1732943030848908,
      "grad_norm": 1.2401609420776367,
      "learning_rate": 9.819340370977794e-05,
      "loss": 1.7033,
      "step": 962
    },
    {
      "epoch": 0.17347444269308715,
      "grad_norm": 1.2908941507339478,
      "learning_rate": 9.81896294290082e-05,
      "loss": 1.9744,
      "step": 963
    },
    {
      "epoch": 0.1736545823012835,
      "grad_norm": 1.2396693229675293,
      "learning_rate": 9.818585128248988e-05,
      "loss": 1.8415,
      "step": 964
    },
    {
      "epoch": 0.17383472190947985,
      "grad_norm": 1.175499439239502,
      "learning_rate": 9.818206927052602e-05,
      "loss": 1.6938,
      "step": 965
    },
    {
      "epoch": 0.1740148615176762,
      "grad_norm": 1.2146129608154297,
      "learning_rate": 9.817828339342003e-05,
      "loss": 1.8406,
      "step": 966
    },
    {
      "epoch": 0.17419500112587255,
      "grad_norm": 1.224776029586792,
      "learning_rate": 9.817449365147561e-05,
      "loss": 1.7464,
      "step": 967
    },
    {
      "epoch": 0.1743751407340689,
      "grad_norm": 1.324190616607666,
      "learning_rate": 9.817070004499677e-05,
      "loss": 1.9447,
      "step": 968
    },
    {
      "epoch": 0.17455528034226525,
      "grad_norm": 1.2665823698043823,
      "learning_rate": 9.816690257428787e-05,
      "loss": 1.9,
      "step": 969
    },
    {
      "epoch": 0.1747354199504616,
      "grad_norm": 1.2253963947296143,
      "learning_rate": 9.816310123965348e-05,
      "loss": 1.7283,
      "step": 970
    },
    {
      "epoch": 0.17491555955865795,
      "grad_norm": 1.2518510818481445,
      "learning_rate": 9.815929604139857e-05,
      "loss": 1.9728,
      "step": 971
    },
    {
      "epoch": 0.1750956991668543,
      "grad_norm": 1.2972339391708374,
      "learning_rate": 9.81554869798284e-05,
      "loss": 1.8105,
      "step": 972
    },
    {
      "epoch": 0.17527583877505065,
      "grad_norm": 1.2224445343017578,
      "learning_rate": 9.81516740552485e-05,
      "loss": 1.7032,
      "step": 973
    },
    {
      "epoch": 0.17545597838324703,
      "grad_norm": 1.270284652709961,
      "learning_rate": 9.814785726796478e-05,
      "loss": 1.9332,
      "step": 974
    },
    {
      "epoch": 0.17563611799144338,
      "grad_norm": 1.159845232963562,
      "learning_rate": 9.81440366182834e-05,
      "loss": 1.7398,
      "step": 975
    },
    {
      "epoch": 0.17581625759963973,
      "grad_norm": 1.171785831451416,
      "learning_rate": 9.814021210651084e-05,
      "loss": 1.9278,
      "step": 976
    },
    {
      "epoch": 0.17599639720783608,
      "grad_norm": 1.3585643768310547,
      "learning_rate": 9.813638373295392e-05,
      "loss": 2.0651,
      "step": 977
    },
    {
      "epoch": 0.17617653681603243,
      "grad_norm": 1.1806076765060425,
      "learning_rate": 9.813255149791975e-05,
      "loss": 1.6853,
      "step": 978
    },
    {
      "epoch": 0.17635667642422878,
      "grad_norm": 1.3248120546340942,
      "learning_rate": 9.812871540171574e-05,
      "loss": 1.8361,
      "step": 979
    },
    {
      "epoch": 0.17653681603242513,
      "grad_norm": 1.4118931293487549,
      "learning_rate": 9.81248754446496e-05,
      "loss": 2.1368,
      "step": 980
    },
    {
      "epoch": 0.17671695564062148,
      "grad_norm": 1.1947461366653442,
      "learning_rate": 9.812103162702942e-05,
      "loss": 1.5551,
      "step": 981
    },
    {
      "epoch": 0.17689709524881783,
      "grad_norm": 1.2409764528274536,
      "learning_rate": 9.811718394916351e-05,
      "loss": 1.7858,
      "step": 982
    },
    {
      "epoch": 0.17707723485701418,
      "grad_norm": 1.4271025657653809,
      "learning_rate": 9.811333241136056e-05,
      "loss": 2.1236,
      "step": 983
    },
    {
      "epoch": 0.17725737446521053,
      "grad_norm": 1.3088583946228027,
      "learning_rate": 9.81094770139295e-05,
      "loss": 1.7706,
      "step": 984
    },
    {
      "epoch": 0.17743751407340688,
      "grad_norm": 1.5288505554199219,
      "learning_rate": 9.810561775717964e-05,
      "loss": 2.3629,
      "step": 985
    },
    {
      "epoch": 0.17761765368160323,
      "grad_norm": 1.3391587734222412,
      "learning_rate": 9.810175464142053e-05,
      "loss": 2.1055,
      "step": 986
    },
    {
      "epoch": 0.17779779328979958,
      "grad_norm": 1.3647550344467163,
      "learning_rate": 9.809788766696213e-05,
      "loss": 1.8956,
      "step": 987
    },
    {
      "epoch": 0.17797793289799596,
      "grad_norm": 1.4144048690795898,
      "learning_rate": 9.809401683411457e-05,
      "loss": 1.9614,
      "step": 988
    },
    {
      "epoch": 0.1781580725061923,
      "grad_norm": 1.405476689338684,
      "learning_rate": 9.809014214318843e-05,
      "loss": 1.829,
      "step": 989
    },
    {
      "epoch": 0.17833821211438866,
      "grad_norm": 1.3766098022460938,
      "learning_rate": 9.808626359449451e-05,
      "loss": 1.8353,
      "step": 990
    },
    {
      "epoch": 0.178518351722585,
      "grad_norm": 1.3525934219360352,
      "learning_rate": 9.808238118834394e-05,
      "loss": 1.72,
      "step": 991
    },
    {
      "epoch": 0.17869849133078136,
      "grad_norm": 1.4221618175506592,
      "learning_rate": 9.807849492504817e-05,
      "loss": 2.0379,
      "step": 992
    },
    {
      "epoch": 0.1788786309389777,
      "grad_norm": 1.3941259384155273,
      "learning_rate": 9.807460480491895e-05,
      "loss": 2.0559,
      "step": 993
    },
    {
      "epoch": 0.17905877054717406,
      "grad_norm": 1.195777177810669,
      "learning_rate": 9.807071082826836e-05,
      "loss": 1.6323,
      "step": 994
    },
    {
      "epoch": 0.1792389101553704,
      "grad_norm": 1.3851172924041748,
      "learning_rate": 9.806681299540877e-05,
      "loss": 1.7501,
      "step": 995
    },
    {
      "epoch": 0.17941904976356676,
      "grad_norm": 1.3659976720809937,
      "learning_rate": 9.806291130665284e-05,
      "loss": 1.8423,
      "step": 996
    },
    {
      "epoch": 0.1795991893717631,
      "grad_norm": 1.4213128089904785,
      "learning_rate": 9.805900576231358e-05,
      "loss": 1.9137,
      "step": 997
    },
    {
      "epoch": 0.17977932897995946,
      "grad_norm": 1.3705589771270752,
      "learning_rate": 9.805509636270427e-05,
      "loss": 1.9406,
      "step": 998
    },
    {
      "epoch": 0.1799594685881558,
      "grad_norm": 1.6332011222839355,
      "learning_rate": 9.805118310813856e-05,
      "loss": 2.113,
      "step": 999
    },
    {
      "epoch": 0.18013960819635216,
      "grad_norm": 1.3458186388015747,
      "learning_rate": 9.804726599893033e-05,
      "loss": 1.694,
      "step": 1000
    },
    {
      "epoch": 0.1803197478045485,
      "grad_norm": 1.216593623161316,
      "learning_rate": 9.804334503539383e-05,
      "loss": 2.0628,
      "step": 1001
    },
    {
      "epoch": 0.18049988741274486,
      "grad_norm": 1.2084629535675049,
      "learning_rate": 9.803942021784361e-05,
      "loss": 2.1192,
      "step": 1002
    },
    {
      "epoch": 0.18068002702094124,
      "grad_norm": 1.3153703212738037,
      "learning_rate": 9.803549154659448e-05,
      "loss": 2.2023,
      "step": 1003
    },
    {
      "epoch": 0.1808601666291376,
      "grad_norm": 1.354103446006775,
      "learning_rate": 9.803155902196161e-05,
      "loss": 2.3206,
      "step": 1004
    },
    {
      "epoch": 0.18104030623733394,
      "grad_norm": 1.3447707891464233,
      "learning_rate": 9.802762264426048e-05,
      "loss": 2.3337,
      "step": 1005
    },
    {
      "epoch": 0.1812204458455303,
      "grad_norm": 1.8964850902557373,
      "learning_rate": 9.802368241380685e-05,
      "loss": 2.29,
      "step": 1006
    },
    {
      "epoch": 0.18140058545372664,
      "grad_norm": 1.4371596574783325,
      "learning_rate": 9.801973833091682e-05,
      "loss": 2.2448,
      "step": 1007
    },
    {
      "epoch": 0.181580725061923,
      "grad_norm": 1.5542829036712646,
      "learning_rate": 9.801579039590675e-05,
      "loss": 2.1635,
      "step": 1008
    },
    {
      "epoch": 0.18176086467011934,
      "grad_norm": 1.7047879695892334,
      "learning_rate": 9.801183860909338e-05,
      "loss": 2.541,
      "step": 1009
    },
    {
      "epoch": 0.1819410042783157,
      "grad_norm": 1.702187180519104,
      "learning_rate": 9.80078829707937e-05,
      "loss": 2.3559,
      "step": 1010
    },
    {
      "epoch": 0.18212114388651204,
      "grad_norm": 1.2553833723068237,
      "learning_rate": 9.800392348132504e-05,
      "loss": 1.978,
      "step": 1011
    },
    {
      "epoch": 0.1823012834947084,
      "grad_norm": 1.3444942235946655,
      "learning_rate": 9.7999960141005e-05,
      "loss": 2.0664,
      "step": 1012
    },
    {
      "epoch": 0.18248142310290474,
      "grad_norm": 1.2549513578414917,
      "learning_rate": 9.799599295015154e-05,
      "loss": 2.0013,
      "step": 1013
    },
    {
      "epoch": 0.1826615627111011,
      "grad_norm": 1.3910187482833862,
      "learning_rate": 9.799202190908293e-05,
      "loss": 1.9403,
      "step": 1014
    },
    {
      "epoch": 0.18284170231929744,
      "grad_norm": 1.1417717933654785,
      "learning_rate": 9.798804701811765e-05,
      "loss": 1.5851,
      "step": 1015
    },
    {
      "epoch": 0.1830218419274938,
      "grad_norm": 1.232528567314148,
      "learning_rate": 9.798406827757465e-05,
      "loss": 1.9015,
      "step": 1016
    },
    {
      "epoch": 0.18320198153569017,
      "grad_norm": 1.3617945909500122,
      "learning_rate": 9.798008568777306e-05,
      "loss": 1.8823,
      "step": 1017
    },
    {
      "epoch": 0.18338212114388652,
      "grad_norm": 1.2282837629318237,
      "learning_rate": 9.797609924903237e-05,
      "loss": 1.8085,
      "step": 1018
    },
    {
      "epoch": 0.18356226075208287,
      "grad_norm": 1.1524983644485474,
      "learning_rate": 9.797210896167234e-05,
      "loss": 1.8068,
      "step": 1019
    },
    {
      "epoch": 0.18374240036027922,
      "grad_norm": 1.2424622774124146,
      "learning_rate": 9.796811482601314e-05,
      "loss": 1.9911,
      "step": 1020
    },
    {
      "epoch": 0.18392253996847557,
      "grad_norm": 1.1579972505569458,
      "learning_rate": 9.79641168423751e-05,
      "loss": 1.7695,
      "step": 1021
    },
    {
      "epoch": 0.18410267957667192,
      "grad_norm": 1.2499160766601562,
      "learning_rate": 9.796011501107898e-05,
      "loss": 1.675,
      "step": 1022
    },
    {
      "epoch": 0.18428281918486827,
      "grad_norm": 1.3865952491760254,
      "learning_rate": 9.795610933244579e-05,
      "loss": 1.841,
      "step": 1023
    },
    {
      "epoch": 0.18446295879306462,
      "grad_norm": 1.2480659484863281,
      "learning_rate": 9.795209980679687e-05,
      "loss": 1.9389,
      "step": 1024
    },
    {
      "epoch": 0.18464309840126097,
      "grad_norm": 1.3107167482376099,
      "learning_rate": 9.794808643445385e-05,
      "loss": 1.9855,
      "step": 1025
    },
    {
      "epoch": 0.18482323800945732,
      "grad_norm": 1.2713347673416138,
      "learning_rate": 9.794406921573871e-05,
      "loss": 1.7889,
      "step": 1026
    },
    {
      "epoch": 0.18500337761765367,
      "grad_norm": 1.401381254196167,
      "learning_rate": 9.794004815097367e-05,
      "loss": 2.0242,
      "step": 1027
    },
    {
      "epoch": 0.18518351722585003,
      "grad_norm": 1.3585182428359985,
      "learning_rate": 9.793602324048134e-05,
      "loss": 2.1322,
      "step": 1028
    },
    {
      "epoch": 0.18536365683404638,
      "grad_norm": 1.3229870796203613,
      "learning_rate": 9.793199448458456e-05,
      "loss": 1.9269,
      "step": 1029
    },
    {
      "epoch": 0.18554379644224273,
      "grad_norm": 1.3098610639572144,
      "learning_rate": 9.792796188360654e-05,
      "loss": 2.0335,
      "step": 1030
    },
    {
      "epoch": 0.1857239360504391,
      "grad_norm": 1.2868019342422485,
      "learning_rate": 9.792392543787075e-05,
      "loss": 1.9837,
      "step": 1031
    },
    {
      "epoch": 0.18590407565863545,
      "grad_norm": 1.2790473699569702,
      "learning_rate": 9.791988514770101e-05,
      "loss": 1.8837,
      "step": 1032
    },
    {
      "epoch": 0.1860842152668318,
      "grad_norm": 1.3025767803192139,
      "learning_rate": 9.791584101342142e-05,
      "loss": 1.7508,
      "step": 1033
    },
    {
      "epoch": 0.18626435487502815,
      "grad_norm": 1.2877564430236816,
      "learning_rate": 9.79117930353564e-05,
      "loss": 1.6398,
      "step": 1034
    },
    {
      "epoch": 0.1864444944832245,
      "grad_norm": 1.1584537029266357,
      "learning_rate": 9.790774121383068e-05,
      "loss": 1.6753,
      "step": 1035
    },
    {
      "epoch": 0.18662463409142085,
      "grad_norm": 1.5099139213562012,
      "learning_rate": 9.790368554916931e-05,
      "loss": 2.2535,
      "step": 1036
    },
    {
      "epoch": 0.1868047736996172,
      "grad_norm": 1.408600926399231,
      "learning_rate": 9.789962604169761e-05,
      "loss": 1.8433,
      "step": 1037
    },
    {
      "epoch": 0.18698491330781356,
      "grad_norm": 1.3460115194320679,
      "learning_rate": 9.789556269174123e-05,
      "loss": 1.8245,
      "step": 1038
    },
    {
      "epoch": 0.1871650529160099,
      "grad_norm": 1.4837392568588257,
      "learning_rate": 9.789149549962617e-05,
      "loss": 1.9556,
      "step": 1039
    },
    {
      "epoch": 0.18734519252420626,
      "grad_norm": 1.480675220489502,
      "learning_rate": 9.788742446567865e-05,
      "loss": 1.8344,
      "step": 1040
    },
    {
      "epoch": 0.1875253321324026,
      "grad_norm": 1.3863763809204102,
      "learning_rate": 9.788334959022524e-05,
      "loss": 1.6853,
      "step": 1041
    },
    {
      "epoch": 0.18770547174059896,
      "grad_norm": 1.4679621458053589,
      "learning_rate": 9.787927087359287e-05,
      "loss": 1.9679,
      "step": 1042
    },
    {
      "epoch": 0.1878856113487953,
      "grad_norm": 1.2725722789764404,
      "learning_rate": 9.787518831610873e-05,
      "loss": 1.8375,
      "step": 1043
    },
    {
      "epoch": 0.18806575095699166,
      "grad_norm": 1.3111138343811035,
      "learning_rate": 9.787110191810027e-05,
      "loss": 1.6635,
      "step": 1044
    },
    {
      "epoch": 0.18824589056518803,
      "grad_norm": 1.3817356824874878,
      "learning_rate": 9.786701167989536e-05,
      "loss": 2.0924,
      "step": 1045
    },
    {
      "epoch": 0.18842603017338438,
      "grad_norm": 1.47190260887146,
      "learning_rate": 9.786291760182207e-05,
      "loss": 1.7857,
      "step": 1046
    },
    {
      "epoch": 0.18860616978158073,
      "grad_norm": 1.297044277191162,
      "learning_rate": 9.785881968420885e-05,
      "loss": 1.8661,
      "step": 1047
    },
    {
      "epoch": 0.18878630938977708,
      "grad_norm": 1.4383862018585205,
      "learning_rate": 9.785471792738443e-05,
      "loss": 1.8995,
      "step": 1048
    },
    {
      "epoch": 0.18896644899797344,
      "grad_norm": 1.407568335533142,
      "learning_rate": 9.785061233167783e-05,
      "loss": 1.557,
      "step": 1049
    },
    {
      "epoch": 0.18914658860616979,
      "grad_norm": 1.3079746961593628,
      "learning_rate": 9.784650289741844e-05,
      "loss": 1.7306,
      "step": 1050
    },
    {
      "epoch": 0.18932672821436614,
      "grad_norm": 1.3035038709640503,
      "learning_rate": 9.784238962493587e-05,
      "loss": 2.2453,
      "step": 1051
    },
    {
      "epoch": 0.18950686782256249,
      "grad_norm": 1.310741662979126,
      "learning_rate": 9.783827251456012e-05,
      "loss": 2.6441,
      "step": 1052
    },
    {
      "epoch": 0.18968700743075884,
      "grad_norm": 1.2539050579071045,
      "learning_rate": 9.783415156662145e-05,
      "loss": 2.1461,
      "step": 1053
    },
    {
      "epoch": 0.1898671470389552,
      "grad_norm": 1.2706090211868286,
      "learning_rate": 9.783002678145043e-05,
      "loss": 1.9084,
      "step": 1054
    },
    {
      "epoch": 0.19004728664715154,
      "grad_norm": 1.1846846342086792,
      "learning_rate": 9.782589815937797e-05,
      "loss": 1.988,
      "step": 1055
    },
    {
      "epoch": 0.1902274262553479,
      "grad_norm": 1.547398567199707,
      "learning_rate": 9.782176570073525e-05,
      "loss": 2.3427,
      "step": 1056
    },
    {
      "epoch": 0.19040756586354424,
      "grad_norm": 1.5276302099227905,
      "learning_rate": 9.781762940585379e-05,
      "loss": 2.3789,
      "step": 1057
    },
    {
      "epoch": 0.1905877054717406,
      "grad_norm": 1.664921522140503,
      "learning_rate": 9.781348927506538e-05,
      "loss": 2.2988,
      "step": 1058
    },
    {
      "epoch": 0.19076784507993694,
      "grad_norm": 1.8396921157836914,
      "learning_rate": 9.780934530870214e-05,
      "loss": 2.0596,
      "step": 1059
    },
    {
      "epoch": 0.19094798468813332,
      "grad_norm": 1.892615795135498,
      "learning_rate": 9.780519750709651e-05,
      "loss": 2.5788,
      "step": 1060
    },
    {
      "epoch": 0.19112812429632967,
      "grad_norm": 1.3459168672561646,
      "learning_rate": 9.780104587058123e-05,
      "loss": 1.8824,
      "step": 1061
    },
    {
      "epoch": 0.19130826390452602,
      "grad_norm": 1.4003496170043945,
      "learning_rate": 9.779689039948934e-05,
      "loss": 2.008,
      "step": 1062
    },
    {
      "epoch": 0.19148840351272237,
      "grad_norm": 1.4113889932632446,
      "learning_rate": 9.779273109415417e-05,
      "loss": 2.186,
      "step": 1063
    },
    {
      "epoch": 0.19166854312091872,
      "grad_norm": 1.4023385047912598,
      "learning_rate": 9.778856795490939e-05,
      "loss": 1.9166,
      "step": 1064
    },
    {
      "epoch": 0.19184868272911507,
      "grad_norm": 1.434439778327942,
      "learning_rate": 9.778440098208897e-05,
      "loss": 2.1106,
      "step": 1065
    },
    {
      "epoch": 0.19202882233731142,
      "grad_norm": 1.31666100025177,
      "learning_rate": 9.778023017602718e-05,
      "loss": 1.8378,
      "step": 1066
    },
    {
      "epoch": 0.19220896194550777,
      "grad_norm": 1.1839903593063354,
      "learning_rate": 9.77760555370586e-05,
      "loss": 1.6132,
      "step": 1067
    },
    {
      "epoch": 0.19238910155370412,
      "grad_norm": 1.455063819885254,
      "learning_rate": 9.777187706551811e-05,
      "loss": 2.036,
      "step": 1068
    },
    {
      "epoch": 0.19256924116190047,
      "grad_norm": 1.2490276098251343,
      "learning_rate": 9.776769476174091e-05,
      "loss": 1.8584,
      "step": 1069
    },
    {
      "epoch": 0.19274938077009682,
      "grad_norm": 1.2353384494781494,
      "learning_rate": 9.776350862606251e-05,
      "loss": 1.73,
      "step": 1070
    },
    {
      "epoch": 0.19292952037829317,
      "grad_norm": 1.334887146949768,
      "learning_rate": 9.775931865881872e-05,
      "loss": 1.7689,
      "step": 1071
    },
    {
      "epoch": 0.19310965998648952,
      "grad_norm": 1.2303911447525024,
      "learning_rate": 9.775512486034563e-05,
      "loss": 1.6101,
      "step": 1072
    },
    {
      "epoch": 0.19328979959468587,
      "grad_norm": 1.2655179500579834,
      "learning_rate": 9.775092723097969e-05,
      "loss": 1.8952,
      "step": 1073
    },
    {
      "epoch": 0.19346993920288225,
      "grad_norm": 1.2273050546646118,
      "learning_rate": 9.774672577105763e-05,
      "loss": 1.8603,
      "step": 1074
    },
    {
      "epoch": 0.1936500788110786,
      "grad_norm": 1.1224472522735596,
      "learning_rate": 9.774252048091648e-05,
      "loss": 1.63,
      "step": 1075
    },
    {
      "epoch": 0.19383021841927495,
      "grad_norm": 1.3293113708496094,
      "learning_rate": 9.77383113608936e-05,
      "loss": 2.0011,
      "step": 1076
    },
    {
      "epoch": 0.1940103580274713,
      "grad_norm": 1.2797737121582031,
      "learning_rate": 9.773409841132664e-05,
      "loss": 1.9221,
      "step": 1077
    },
    {
      "epoch": 0.19419049763566765,
      "grad_norm": 1.392388105392456,
      "learning_rate": 9.772988163255354e-05,
      "loss": 1.9118,
      "step": 1078
    },
    {
      "epoch": 0.194370637243864,
      "grad_norm": 1.3774945735931396,
      "learning_rate": 9.772566102491259e-05,
      "loss": 1.7821,
      "step": 1079
    },
    {
      "epoch": 0.19455077685206035,
      "grad_norm": 1.4170559644699097,
      "learning_rate": 9.772143658874235e-05,
      "loss": 2.0068,
      "step": 1080
    },
    {
      "epoch": 0.1947309164602567,
      "grad_norm": 1.3845994472503662,
      "learning_rate": 9.771720832438172e-05,
      "loss": 2.0581,
      "step": 1081
    },
    {
      "epoch": 0.19491105606845305,
      "grad_norm": 1.238254427909851,
      "learning_rate": 9.771297623216988e-05,
      "loss": 1.6188,
      "step": 1082
    },
    {
      "epoch": 0.1950911956766494,
      "grad_norm": 1.5110121965408325,
      "learning_rate": 9.770874031244634e-05,
      "loss": 2.0153,
      "step": 1083
    },
    {
      "epoch": 0.19527133528484575,
      "grad_norm": 1.2862738370895386,
      "learning_rate": 9.770450056555087e-05,
      "loss": 1.6687,
      "step": 1084
    },
    {
      "epoch": 0.1954514748930421,
      "grad_norm": 1.4597505331039429,
      "learning_rate": 9.770025699182362e-05,
      "loss": 1.7867,
      "step": 1085
    },
    {
      "epoch": 0.19563161450123845,
      "grad_norm": 1.3708995580673218,
      "learning_rate": 9.769600959160497e-05,
      "loss": 1.9031,
      "step": 1086
    },
    {
      "epoch": 0.1958117541094348,
      "grad_norm": 1.2739589214324951,
      "learning_rate": 9.769175836523568e-05,
      "loss": 1.8369,
      "step": 1087
    },
    {
      "epoch": 0.19599189371763118,
      "grad_norm": 1.3774598836898804,
      "learning_rate": 9.768750331305673e-05,
      "loss": 1.9478,
      "step": 1088
    },
    {
      "epoch": 0.19617203332582753,
      "grad_norm": 1.405414342880249,
      "learning_rate": 9.768324443540953e-05,
      "loss": 1.9954,
      "step": 1089
    },
    {
      "epoch": 0.19635217293402388,
      "grad_norm": 1.3449652194976807,
      "learning_rate": 9.767898173263568e-05,
      "loss": 1.8365,
      "step": 1090
    },
    {
      "epoch": 0.19653231254222023,
      "grad_norm": 1.4840807914733887,
      "learning_rate": 9.767471520507711e-05,
      "loss": 2.1271,
      "step": 1091
    },
    {
      "epoch": 0.19671245215041658,
      "grad_norm": 1.4266517162322998,
      "learning_rate": 9.767044485307615e-05,
      "loss": 1.8355,
      "step": 1092
    },
    {
      "epoch": 0.19689259175861293,
      "grad_norm": 1.3680309057235718,
      "learning_rate": 9.76661706769753e-05,
      "loss": 1.5311,
      "step": 1093
    },
    {
      "epoch": 0.19707273136680928,
      "grad_norm": 1.4910508394241333,
      "learning_rate": 9.766189267711745e-05,
      "loss": 2.1129,
      "step": 1094
    },
    {
      "epoch": 0.19725287097500563,
      "grad_norm": 1.3345222473144531,
      "learning_rate": 9.76576108538458e-05,
      "loss": 1.9193,
      "step": 1095
    },
    {
      "epoch": 0.19743301058320198,
      "grad_norm": 1.3052685260772705,
      "learning_rate": 9.765332520750381e-05,
      "loss": 1.7407,
      "step": 1096
    },
    {
      "epoch": 0.19761315019139833,
      "grad_norm": 1.4958308935165405,
      "learning_rate": 9.76490357384353e-05,
      "loss": 1.9617,
      "step": 1097
    },
    {
      "epoch": 0.19779328979959468,
      "grad_norm": 1.4788193702697754,
      "learning_rate": 9.764474244698432e-05,
      "loss": 1.8313,
      "step": 1098
    },
    {
      "epoch": 0.19797342940779103,
      "grad_norm": 1.4373680353164673,
      "learning_rate": 9.764044533349532e-05,
      "loss": 1.9792,
      "step": 1099
    },
    {
      "epoch": 0.19815356901598738,
      "grad_norm": 1.3391166925430298,
      "learning_rate": 9.763614439831301e-05,
      "loss": 1.794,
      "step": 1100
    },
    {
      "epoch": 0.19833370862418373,
      "grad_norm": 1.3338404893875122,
      "learning_rate": 9.76318396417824e-05,
      "loss": 2.3145,
      "step": 1101
    },
    {
      "epoch": 0.1985138482323801,
      "grad_norm": 1.3932822942733765,
      "learning_rate": 9.762753106424882e-05,
      "loss": 2.3292,
      "step": 1102
    },
    {
      "epoch": 0.19869398784057646,
      "grad_norm": 1.2400325536727905,
      "learning_rate": 9.762321866605789e-05,
      "loss": 2.212,
      "step": 1103
    },
    {
      "epoch": 0.1988741274487728,
      "grad_norm": 1.3097000122070312,
      "learning_rate": 9.761890244755554e-05,
      "loss": 2.3767,
      "step": 1104
    },
    {
      "epoch": 0.19905426705696916,
      "grad_norm": 1.9655861854553223,
      "learning_rate": 9.761458240908804e-05,
      "loss": 2.4407,
      "step": 1105
    },
    {
      "epoch": 0.1992344066651655,
      "grad_norm": 1.4316792488098145,
      "learning_rate": 9.761025855100193e-05,
      "loss": 1.8224,
      "step": 1106
    },
    {
      "epoch": 0.19941454627336186,
      "grad_norm": 1.503657579421997,
      "learning_rate": 9.760593087364408e-05,
      "loss": 2.0246,
      "step": 1107
    },
    {
      "epoch": 0.1995946858815582,
      "grad_norm": 1.9716205596923828,
      "learning_rate": 9.760159937736163e-05,
      "loss": 2.5355,
      "step": 1108
    },
    {
      "epoch": 0.19977482548975456,
      "grad_norm": 1.720569133758545,
      "learning_rate": 9.759726406250207e-05,
      "loss": 2.3835,
      "step": 1109
    },
    {
      "epoch": 0.1999549650979509,
      "grad_norm": 1.3845270872116089,
      "learning_rate": 9.759292492941317e-05,
      "loss": 1.8727,
      "step": 1110
    },
    {
      "epoch": 0.20013510470614726,
      "grad_norm": 1.1699467897415161,
      "learning_rate": 9.758858197844302e-05,
      "loss": 1.791,
      "step": 1111
    },
    {
      "epoch": 0.2003152443143436,
      "grad_norm": 1.3896262645721436,
      "learning_rate": 9.758423520993999e-05,
      "loss": 1.9561,
      "step": 1112
    },
    {
      "epoch": 0.20049538392253996,
      "grad_norm": 1.3396421670913696,
      "learning_rate": 9.75798846242528e-05,
      "loss": 1.7734,
      "step": 1113
    },
    {
      "epoch": 0.2006755235307363,
      "grad_norm": 1.2620508670806885,
      "learning_rate": 9.757553022173043e-05,
      "loss": 1.9857,
      "step": 1114
    },
    {
      "epoch": 0.20085566313893266,
      "grad_norm": 1.3318549394607544,
      "learning_rate": 9.757117200272221e-05,
      "loss": 1.9463,
      "step": 1115
    },
    {
      "epoch": 0.20103580274712904,
      "grad_norm": 2.266291856765747,
      "learning_rate": 9.756680996757774e-05,
      "loss": 2.0614,
      "step": 1116
    },
    {
      "epoch": 0.2012159423553254,
      "grad_norm": 1.1756788492202759,
      "learning_rate": 9.756244411664694e-05,
      "loss": 1.5352,
      "step": 1117
    },
    {
      "epoch": 0.20139608196352174,
      "grad_norm": 1.4018486738204956,
      "learning_rate": 9.755807445028005e-05,
      "loss": 2.0813,
      "step": 1118
    },
    {
      "epoch": 0.2015762215717181,
      "grad_norm": 1.1505703926086426,
      "learning_rate": 9.755370096882757e-05,
      "loss": 1.614,
      "step": 1119
    },
    {
      "epoch": 0.20175636117991444,
      "grad_norm": 1.2846778631210327,
      "learning_rate": 9.754932367264038e-05,
      "loss": 1.9285,
      "step": 1120
    },
    {
      "epoch": 0.2019365007881108,
      "grad_norm": 1.3055049180984497,
      "learning_rate": 9.754494256206961e-05,
      "loss": 1.9963,
      "step": 1121
    },
    {
      "epoch": 0.20211664039630714,
      "grad_norm": 1.2917466163635254,
      "learning_rate": 9.754055763746668e-05,
      "loss": 1.7182,
      "step": 1122
    },
    {
      "epoch": 0.2022967800045035,
      "grad_norm": 1.3191360235214233,
      "learning_rate": 9.75361688991834e-05,
      "loss": 2.0444,
      "step": 1123
    },
    {
      "epoch": 0.20247691961269984,
      "grad_norm": 1.312503695487976,
      "learning_rate": 9.753177634757179e-05,
      "loss": 1.8112,
      "step": 1124
    },
    {
      "epoch": 0.2026570592208962,
      "grad_norm": 1.2230874300003052,
      "learning_rate": 9.752737998298425e-05,
      "loss": 1.8059,
      "step": 1125
    },
    {
      "epoch": 0.20283719882909254,
      "grad_norm": 1.3253501653671265,
      "learning_rate": 9.75229798057734e-05,
      "loss": 1.8656,
      "step": 1126
    },
    {
      "epoch": 0.2030173384372889,
      "grad_norm": 1.2441682815551758,
      "learning_rate": 9.75185758162923e-05,
      "loss": 1.7846,
      "step": 1127
    },
    {
      "epoch": 0.20319747804548524,
      "grad_norm": 1.2234008312225342,
      "learning_rate": 9.751416801489417e-05,
      "loss": 1.9963,
      "step": 1128
    },
    {
      "epoch": 0.2033776176536816,
      "grad_norm": 1.2510011196136475,
      "learning_rate": 9.750975640193263e-05,
      "loss": 1.971,
      "step": 1129
    },
    {
      "epoch": 0.20355775726187794,
      "grad_norm": 1.238528847694397,
      "learning_rate": 9.750534097776158e-05,
      "loss": 1.7935,
      "step": 1130
    },
    {
      "epoch": 0.20373789687007432,
      "grad_norm": 1.2961379289627075,
      "learning_rate": 9.750092174273521e-05,
      "loss": 2.0679,
      "step": 1131
    },
    {
      "epoch": 0.20391803647827067,
      "grad_norm": 1.3668330907821655,
      "learning_rate": 9.749649869720803e-05,
      "loss": 1.8844,
      "step": 1132
    },
    {
      "epoch": 0.20409817608646702,
      "grad_norm": 1.3781975507736206,
      "learning_rate": 9.749207184153486e-05,
      "loss": 2.0807,
      "step": 1133
    },
    {
      "epoch": 0.20427831569466337,
      "grad_norm": 1.3022228479385376,
      "learning_rate": 9.748764117607084e-05,
      "loss": 2.0386,
      "step": 1134
    },
    {
      "epoch": 0.20445845530285972,
      "grad_norm": 1.3531427383422852,
      "learning_rate": 9.748320670117134e-05,
      "loss": 1.9187,
      "step": 1135
    },
    {
      "epoch": 0.20463859491105607,
      "grad_norm": 1.3155783414840698,
      "learning_rate": 9.747876841719214e-05,
      "loss": 1.9167,
      "step": 1136
    },
    {
      "epoch": 0.20481873451925242,
      "grad_norm": 1.29123055934906,
      "learning_rate": 9.747432632448928e-05,
      "loss": 1.8631,
      "step": 1137
    },
    {
      "epoch": 0.20499887412744877,
      "grad_norm": 1.2773045301437378,
      "learning_rate": 9.746988042341906e-05,
      "loss": 1.9638,
      "step": 1138
    },
    {
      "epoch": 0.20517901373564512,
      "grad_norm": 1.3281618356704712,
      "learning_rate": 9.746543071433818e-05,
      "loss": 1.9104,
      "step": 1139
    },
    {
      "epoch": 0.20535915334384147,
      "grad_norm": 1.2960795164108276,
      "learning_rate": 9.746097719760355e-05,
      "loss": 1.6743,
      "step": 1140
    },
    {
      "epoch": 0.20553929295203782,
      "grad_norm": 1.339294195175171,
      "learning_rate": 9.745651987357245e-05,
      "loss": 1.7139,
      "step": 1141
    },
    {
      "epoch": 0.20571943256023417,
      "grad_norm": 1.2710645198822021,
      "learning_rate": 9.745205874260246e-05,
      "loss": 1.8765,
      "step": 1142
    },
    {
      "epoch": 0.20589957216843052,
      "grad_norm": 1.3489642143249512,
      "learning_rate": 9.744759380505141e-05,
      "loss": 1.8905,
      "step": 1143
    },
    {
      "epoch": 0.20607971177662687,
      "grad_norm": 1.3232749700546265,
      "learning_rate": 9.744312506127749e-05,
      "loss": 1.913,
      "step": 1144
    },
    {
      "epoch": 0.20625985138482325,
      "grad_norm": 1.3831884860992432,
      "learning_rate": 9.74386525116392e-05,
      "loss": 1.8077,
      "step": 1145
    },
    {
      "epoch": 0.2064399909930196,
      "grad_norm": 1.3234407901763916,
      "learning_rate": 9.74341761564953e-05,
      "loss": 1.7466,
      "step": 1146
    },
    {
      "epoch": 0.20662013060121595,
      "grad_norm": 1.478615403175354,
      "learning_rate": 9.74296959962049e-05,
      "loss": 1.8893,
      "step": 1147
    },
    {
      "epoch": 0.2068002702094123,
      "grad_norm": 1.304890751838684,
      "learning_rate": 9.74252120311274e-05,
      "loss": 1.7499,
      "step": 1148
    },
    {
      "epoch": 0.20698040981760865,
      "grad_norm": 1.286595106124878,
      "learning_rate": 9.742072426162249e-05,
      "loss": 1.7273,
      "step": 1149
    },
    {
      "epoch": 0.207160549425805,
      "grad_norm": 1.437442421913147,
      "learning_rate": 9.741623268805017e-05,
      "loss": 1.9995,
      "step": 1150
    },
    {
      "epoch": 0.20734068903400135,
      "grad_norm": 1.3560956716537476,
      "learning_rate": 9.741173731077075e-05,
      "loss": 2.2523,
      "step": 1151
    },
    {
      "epoch": 0.2075208286421977,
      "grad_norm": 1.4598274230957031,
      "learning_rate": 9.740723813014487e-05,
      "loss": 2.3113,
      "step": 1152
    },
    {
      "epoch": 0.20770096825039405,
      "grad_norm": 1.3802415132522583,
      "learning_rate": 9.740273514653344e-05,
      "loss": 2.3313,
      "step": 1153
    },
    {
      "epoch": 0.2078811078585904,
      "grad_norm": 1.4327166080474854,
      "learning_rate": 9.739822836029768e-05,
      "loss": 2.3032,
      "step": 1154
    },
    {
      "epoch": 0.20806124746678675,
      "grad_norm": 1.2819279432296753,
      "learning_rate": 9.739371777179912e-05,
      "loss": 2.3655,
      "step": 1155
    },
    {
      "epoch": 0.2082413870749831,
      "grad_norm": 1.4347268342971802,
      "learning_rate": 9.738920338139961e-05,
      "loss": 2.2786,
      "step": 1156
    },
    {
      "epoch": 0.20842152668317945,
      "grad_norm": 1.5550678968429565,
      "learning_rate": 9.738468518946128e-05,
      "loss": 2.435,
      "step": 1157
    },
    {
      "epoch": 0.2086016662913758,
      "grad_norm": 1.6036711931228638,
      "learning_rate": 9.73801631963466e-05,
      "loss": 2.441,
      "step": 1158
    },
    {
      "epoch": 0.20878180589957218,
      "grad_norm": 2.26991605758667,
      "learning_rate": 9.73756374024183e-05,
      "loss": 2.1703,
      "step": 1159
    },
    {
      "epoch": 0.20896194550776853,
      "grad_norm": 1.752753734588623,
      "learning_rate": 9.737110780803945e-05,
      "loss": 2.2458,
      "step": 1160
    },
    {
      "epoch": 0.20914208511596488,
      "grad_norm": 1.2708410024642944,
      "learning_rate": 9.736657441357339e-05,
      "loss": 1.7728,
      "step": 1161
    },
    {
      "epoch": 0.20932222472416123,
      "grad_norm": 1.3556185960769653,
      "learning_rate": 9.736203721938382e-05,
      "loss": 1.9593,
      "step": 1162
    },
    {
      "epoch": 0.20950236433235758,
      "grad_norm": 1.2849525213241577,
      "learning_rate": 9.735749622583468e-05,
      "loss": 2.0225,
      "step": 1163
    },
    {
      "epoch": 0.20968250394055393,
      "grad_norm": 1.2545552253723145,
      "learning_rate": 9.735295143329027e-05,
      "loss": 1.6616,
      "step": 1164
    },
    {
      "epoch": 0.20986264354875028,
      "grad_norm": 1.3480777740478516,
      "learning_rate": 9.734840284211515e-05,
      "loss": 1.7657,
      "step": 1165
    },
    {
      "epoch": 0.21004278315694663,
      "grad_norm": 1.4599077701568604,
      "learning_rate": 9.734385045267422e-05,
      "loss": 1.8052,
      "step": 1166
    },
    {
      "epoch": 0.21022292276514298,
      "grad_norm": 1.4238650798797607,
      "learning_rate": 9.733929426533268e-05,
      "loss": 1.9854,
      "step": 1167
    },
    {
      "epoch": 0.21040306237333933,
      "grad_norm": 1.3731627464294434,
      "learning_rate": 9.733473428045599e-05,
      "loss": 1.872,
      "step": 1168
    },
    {
      "epoch": 0.21058320198153568,
      "grad_norm": 1.2766082286834717,
      "learning_rate": 9.733017049840999e-05,
      "loss": 1.8957,
      "step": 1169
    },
    {
      "epoch": 0.21076334158973203,
      "grad_norm": 1.3454746007919312,
      "learning_rate": 9.732560291956076e-05,
      "loss": 2.0359,
      "step": 1170
    },
    {
      "epoch": 0.21094348119792838,
      "grad_norm": 1.3586333990097046,
      "learning_rate": 9.73210315442747e-05,
      "loss": 2.0876,
      "step": 1171
    },
    {
      "epoch": 0.21112362080612473,
      "grad_norm": 1.3223870992660522,
      "learning_rate": 9.731645637291857e-05,
      "loss": 1.8035,
      "step": 1172
    },
    {
      "epoch": 0.2113037604143211,
      "grad_norm": 1.6285736560821533,
      "learning_rate": 9.731187740585934e-05,
      "loss": 1.8659,
      "step": 1173
    },
    {
      "epoch": 0.21148390002251746,
      "grad_norm": 1.1811076402664185,
      "learning_rate": 9.730729464346435e-05,
      "loss": 1.7004,
      "step": 1174
    },
    {
      "epoch": 0.2116640396307138,
      "grad_norm": 1.2941184043884277,
      "learning_rate": 9.730270808610123e-05,
      "loss": 2.1722,
      "step": 1175
    },
    {
      "epoch": 0.21184417923891016,
      "grad_norm": 1.2476855516433716,
      "learning_rate": 9.72981177341379e-05,
      "loss": 1.9381,
      "step": 1176
    },
    {
      "epoch": 0.2120243188471065,
      "grad_norm": 1.3430800437927246,
      "learning_rate": 9.729352358794259e-05,
      "loss": 1.826,
      "step": 1177
    },
    {
      "epoch": 0.21220445845530286,
      "grad_norm": 1.2400137186050415,
      "learning_rate": 9.728892564788387e-05,
      "loss": 1.5523,
      "step": 1178
    },
    {
      "epoch": 0.2123845980634992,
      "grad_norm": 1.167266607284546,
      "learning_rate": 9.728432391433058e-05,
      "loss": 1.7809,
      "step": 1179
    },
    {
      "epoch": 0.21256473767169556,
      "grad_norm": 1.388493537902832,
      "learning_rate": 9.727971838765184e-05,
      "loss": 1.8449,
      "step": 1180
    },
    {
      "epoch": 0.21274487727989191,
      "grad_norm": 1.307047963142395,
      "learning_rate": 9.727510906821712e-05,
      "loss": 1.5885,
      "step": 1181
    },
    {
      "epoch": 0.21292501688808826,
      "grad_norm": 1.3642369508743286,
      "learning_rate": 9.727049595639619e-05,
      "loss": 1.8891,
      "step": 1182
    },
    {
      "epoch": 0.21310515649628461,
      "grad_norm": 1.2851828336715698,
      "learning_rate": 9.72658790525591e-05,
      "loss": 1.9321,
      "step": 1183
    },
    {
      "epoch": 0.21328529610448096,
      "grad_norm": 1.4932687282562256,
      "learning_rate": 9.72612583570762e-05,
      "loss": 2.1038,
      "step": 1184
    },
    {
      "epoch": 0.21346543571267731,
      "grad_norm": 1.402328372001648,
      "learning_rate": 9.725663387031818e-05,
      "loss": 1.8707,
      "step": 1185
    },
    {
      "epoch": 0.21364557532087367,
      "grad_norm": 1.4498121738433838,
      "learning_rate": 9.725200559265601e-05,
      "loss": 2.0383,
      "step": 1186
    },
    {
      "epoch": 0.21382571492907002,
      "grad_norm": 1.4633519649505615,
      "learning_rate": 9.724737352446097e-05,
      "loss": 1.9944,
      "step": 1187
    },
    {
      "epoch": 0.2140058545372664,
      "grad_norm": 1.463475227355957,
      "learning_rate": 9.724273766610463e-05,
      "loss": 2.1831,
      "step": 1188
    },
    {
      "epoch": 0.21418599414546274,
      "grad_norm": 1.3918044567108154,
      "learning_rate": 9.72380980179589e-05,
      "loss": 2.0299,
      "step": 1189
    },
    {
      "epoch": 0.2143661337536591,
      "grad_norm": 1.3234190940856934,
      "learning_rate": 9.723345458039594e-05,
      "loss": 1.8762,
      "step": 1190
    },
    {
      "epoch": 0.21454627336185544,
      "grad_norm": 1.3141950368881226,
      "learning_rate": 9.722880735378827e-05,
      "loss": 1.8001,
      "step": 1191
    },
    {
      "epoch": 0.2147264129700518,
      "grad_norm": 1.2665444612503052,
      "learning_rate": 9.722415633850867e-05,
      "loss": 1.84,
      "step": 1192
    },
    {
      "epoch": 0.21490655257824814,
      "grad_norm": 1.313765287399292,
      "learning_rate": 9.721950153493026e-05,
      "loss": 1.6348,
      "step": 1193
    },
    {
      "epoch": 0.2150866921864445,
      "grad_norm": 1.4050348997116089,
      "learning_rate": 9.721484294342644e-05,
      "loss": 1.8332,
      "step": 1194
    },
    {
      "epoch": 0.21526683179464084,
      "grad_norm": 1.3703303337097168,
      "learning_rate": 9.721018056437091e-05,
      "loss": 1.9306,
      "step": 1195
    },
    {
      "epoch": 0.2154469714028372,
      "grad_norm": 1.4321315288543701,
      "learning_rate": 9.72055143981377e-05,
      "loss": 1.9754,
      "step": 1196
    },
    {
      "epoch": 0.21562711101103355,
      "grad_norm": 1.4123356342315674,
      "learning_rate": 9.720084444510109e-05,
      "loss": 1.8759,
      "step": 1197
    },
    {
      "epoch": 0.2158072506192299,
      "grad_norm": 1.4252748489379883,
      "learning_rate": 9.719617070563575e-05,
      "loss": 1.8527,
      "step": 1198
    },
    {
      "epoch": 0.21598739022742625,
      "grad_norm": 1.3909029960632324,
      "learning_rate": 9.719149318011659e-05,
      "loss": 1.6909,
      "step": 1199
    },
    {
      "epoch": 0.2161675298356226,
      "grad_norm": 1.535224437713623,
      "learning_rate": 9.718681186891883e-05,
      "loss": 2.043,
      "step": 1200
    },
    {
      "epoch": 0.21634766944381895,
      "grad_norm": 1.3515214920043945,
      "learning_rate": 9.718212677241801e-05,
      "loss": 2.3651,
      "step": 1201
    },
    {
      "epoch": 0.21652780905201532,
      "grad_norm": 1.2250727415084839,
      "learning_rate": 9.717743789098994e-05,
      "loss": 2.0236,
      "step": 1202
    },
    {
      "epoch": 0.21670794866021167,
      "grad_norm": 1.2858576774597168,
      "learning_rate": 9.717274522501078e-05,
      "loss": 2.1827,
      "step": 1203
    },
    {
      "epoch": 0.21688808826840802,
      "grad_norm": 1.3357107639312744,
      "learning_rate": 9.7168048774857e-05,
      "loss": 2.0166,
      "step": 1204
    },
    {
      "epoch": 0.21706822787660437,
      "grad_norm": 1.3120057582855225,
      "learning_rate": 9.716334854090532e-05,
      "loss": 2.1984,
      "step": 1205
    },
    {
      "epoch": 0.21724836748480073,
      "grad_norm": 2.0808050632476807,
      "learning_rate": 9.715864452353278e-05,
      "loss": 2.2997,
      "step": 1206
    },
    {
      "epoch": 0.21742850709299708,
      "grad_norm": 1.8010993003845215,
      "learning_rate": 9.715393672311674e-05,
      "loss": 2.0266,
      "step": 1207
    },
    {
      "epoch": 0.21760864670119343,
      "grad_norm": 1.7874611616134644,
      "learning_rate": 9.714922514003489e-05,
      "loss": 2.7595,
      "step": 1208
    },
    {
      "epoch": 0.21778878630938978,
      "grad_norm": 1.9151959419250488,
      "learning_rate": 9.714450977466514e-05,
      "loss": 2.819,
      "step": 1209
    },
    {
      "epoch": 0.21796892591758613,
      "grad_norm": 2.263363838195801,
      "learning_rate": 9.713979062738579e-05,
      "loss": 2.839,
      "step": 1210
    },
    {
      "epoch": 0.21814906552578248,
      "grad_norm": 1.2915294170379639,
      "learning_rate": 9.71350676985754e-05,
      "loss": 1.9654,
      "step": 1211
    },
    {
      "epoch": 0.21832920513397883,
      "grad_norm": 1.2647360563278198,
      "learning_rate": 9.713034098861283e-05,
      "loss": 2.0295,
      "step": 1212
    },
    {
      "epoch": 0.21850934474217518,
      "grad_norm": 1.2217167615890503,
      "learning_rate": 9.712561049787727e-05,
      "loss": 1.8334,
      "step": 1213
    },
    {
      "epoch": 0.21868948435037153,
      "grad_norm": 1.156005620956421,
      "learning_rate": 9.712087622674819e-05,
      "loss": 1.6766,
      "step": 1214
    },
    {
      "epoch": 0.21886962395856788,
      "grad_norm": 1.2325598001480103,
      "learning_rate": 9.711613817560537e-05,
      "loss": 1.6672,
      "step": 1215
    },
    {
      "epoch": 0.21904976356676426,
      "grad_norm": 1.2970725297927856,
      "learning_rate": 9.711139634482889e-05,
      "loss": 1.6545,
      "step": 1216
    },
    {
      "epoch": 0.2192299031749606,
      "grad_norm": 1.331709384918213,
      "learning_rate": 9.710665073479914e-05,
      "loss": 1.7171,
      "step": 1217
    },
    {
      "epoch": 0.21941004278315696,
      "grad_norm": 1.2705309391021729,
      "learning_rate": 9.710190134589681e-05,
      "loss": 1.8049,
      "step": 1218
    },
    {
      "epoch": 0.2195901823913533,
      "grad_norm": 1.3399676084518433,
      "learning_rate": 9.709714817850291e-05,
      "loss": 1.8443,
      "step": 1219
    },
    {
      "epoch": 0.21977032199954966,
      "grad_norm": 1.283126950263977,
      "learning_rate": 9.709239123299871e-05,
      "loss": 1.6588,
      "step": 1220
    },
    {
      "epoch": 0.219950461607746,
      "grad_norm": 1.600463628768921,
      "learning_rate": 9.708763050976583e-05,
      "loss": 2.2678,
      "step": 1221
    },
    {
      "epoch": 0.22013060121594236,
      "grad_norm": 1.2552045583724976,
      "learning_rate": 9.708286600918617e-05,
      "loss": 1.7007,
      "step": 1222
    },
    {
      "epoch": 0.2203107408241387,
      "grad_norm": 1.3178097009658813,
      "learning_rate": 9.707809773164192e-05,
      "loss": 1.8726,
      "step": 1223
    },
    {
      "epoch": 0.22049088043233506,
      "grad_norm": 1.349375605583191,
      "learning_rate": 9.707332567751562e-05,
      "loss": 1.9585,
      "step": 1224
    },
    {
      "epoch": 0.2206710200405314,
      "grad_norm": 1.3276695013046265,
      "learning_rate": 9.706854984719005e-05,
      "loss": 2.1645,
      "step": 1225
    },
    {
      "epoch": 0.22085115964872776,
      "grad_norm": 1.1302896738052368,
      "learning_rate": 9.706377024104834e-05,
      "loss": 1.628,
      "step": 1226
    },
    {
      "epoch": 0.2210312992569241,
      "grad_norm": 1.2439773082733154,
      "learning_rate": 9.70589868594739e-05,
      "loss": 1.8528,
      "step": 1227
    },
    {
      "epoch": 0.22121143886512046,
      "grad_norm": 1.3040409088134766,
      "learning_rate": 9.705419970285046e-05,
      "loss": 1.7976,
      "step": 1228
    },
    {
      "epoch": 0.2213915784733168,
      "grad_norm": 1.3220973014831543,
      "learning_rate": 9.704940877156204e-05,
      "loss": 2.0121,
      "step": 1229
    },
    {
      "epoch": 0.22157171808151319,
      "grad_norm": 1.348577618598938,
      "learning_rate": 9.704461406599296e-05,
      "loss": 1.9989,
      "step": 1230
    },
    {
      "epoch": 0.22175185768970954,
      "grad_norm": 1.2426236867904663,
      "learning_rate": 9.703981558652786e-05,
      "loss": 1.8465,
      "step": 1231
    },
    {
      "epoch": 0.2219319972979059,
      "grad_norm": 1.3763753175735474,
      "learning_rate": 9.703501333355168e-05,
      "loss": 2.0208,
      "step": 1232
    },
    {
      "epoch": 0.22211213690610224,
      "grad_norm": 1.3459206819534302,
      "learning_rate": 9.70302073074496e-05,
      "loss": 1.8847,
      "step": 1233
    },
    {
      "epoch": 0.2222922765142986,
      "grad_norm": 1.2877014875411987,
      "learning_rate": 9.702539750860723e-05,
      "loss": 1.747,
      "step": 1234
    },
    {
      "epoch": 0.22247241612249494,
      "grad_norm": 1.3653275966644287,
      "learning_rate": 9.702058393741038e-05,
      "loss": 2.045,
      "step": 1235
    },
    {
      "epoch": 0.2226525557306913,
      "grad_norm": 1.3644788265228271,
      "learning_rate": 9.701576659424517e-05,
      "loss": 2.1115,
      "step": 1236
    },
    {
      "epoch": 0.22283269533888764,
      "grad_norm": 1.3194355964660645,
      "learning_rate": 9.701094547949807e-05,
      "loss": 1.9783,
      "step": 1237
    },
    {
      "epoch": 0.223012834947084,
      "grad_norm": 1.2422782182693481,
      "learning_rate": 9.700612059355582e-05,
      "loss": 1.8706,
      "step": 1238
    },
    {
      "epoch": 0.22319297455528034,
      "grad_norm": 1.231799602508545,
      "learning_rate": 9.700129193680549e-05,
      "loss": 1.629,
      "step": 1239
    },
    {
      "epoch": 0.2233731141634767,
      "grad_norm": 1.4027040004730225,
      "learning_rate": 9.699645950963441e-05,
      "loss": 1.8705,
      "step": 1240
    },
    {
      "epoch": 0.22355325377167304,
      "grad_norm": 1.355323076248169,
      "learning_rate": 9.699162331243021e-05,
      "loss": 1.8316,
      "step": 1241
    },
    {
      "epoch": 0.2237333933798694,
      "grad_norm": 1.4202133417129517,
      "learning_rate": 9.698678334558092e-05,
      "loss": 2.1323,
      "step": 1242
    },
    {
      "epoch": 0.22391353298806574,
      "grad_norm": 1.334398865699768,
      "learning_rate": 9.698193960947472e-05,
      "loss": 1.6633,
      "step": 1243
    },
    {
      "epoch": 0.22409367259626212,
      "grad_norm": 1.416087031364441,
      "learning_rate": 9.697709210450022e-05,
      "loss": 1.9239,
      "step": 1244
    },
    {
      "epoch": 0.22427381220445847,
      "grad_norm": 1.3756368160247803,
      "learning_rate": 9.697224083104627e-05,
      "loss": 1.8295,
      "step": 1245
    },
    {
      "epoch": 0.22445395181265482,
      "grad_norm": 1.3327268362045288,
      "learning_rate": 9.696738578950206e-05,
      "loss": 1.8389,
      "step": 1246
    },
    {
      "epoch": 0.22463409142085117,
      "grad_norm": 1.5033972263336182,
      "learning_rate": 9.696252698025702e-05,
      "loss": 1.5961,
      "step": 1247
    },
    {
      "epoch": 0.22481423102904752,
      "grad_norm": 1.4197916984558105,
      "learning_rate": 9.695766440370095e-05,
      "loss": 1.8635,
      "step": 1248
    },
    {
      "epoch": 0.22499437063724387,
      "grad_norm": 1.3862701654434204,
      "learning_rate": 9.69527980602239e-05,
      "loss": 1.6812,
      "step": 1249
    },
    {
      "epoch": 0.22517451024544022,
      "grad_norm": 1.514758586883545,
      "learning_rate": 9.694792795021628e-05,
      "loss": 2.1818,
      "step": 1250
    },
    {
      "epoch": 0.22535464985363657,
      "grad_norm": 1.2962361574172974,
      "learning_rate": 9.694305407406873e-05,
      "loss": 2.2989,
      "step": 1251
    },
    {
      "epoch": 0.22553478946183292,
      "grad_norm": 1.4761441946029663,
      "learning_rate": 9.693817643217224e-05,
      "loss": 2.233,
      "step": 1252
    },
    {
      "epoch": 0.22571492907002927,
      "grad_norm": 1.2174078226089478,
      "learning_rate": 9.693329502491813e-05,
      "loss": 2.134,
      "step": 1253
    },
    {
      "epoch": 0.22589506867822562,
      "grad_norm": 1.3961092233657837,
      "learning_rate": 9.692840985269793e-05,
      "loss": 2.3063,
      "step": 1254
    },
    {
      "epoch": 0.22607520828642197,
      "grad_norm": 1.3094801902770996,
      "learning_rate": 9.692352091590356e-05,
      "loss": 2.1216,
      "step": 1255
    },
    {
      "epoch": 0.22625534789461832,
      "grad_norm": 1.478244423866272,
      "learning_rate": 9.691862821492719e-05,
      "loss": 2.1102,
      "step": 1256
    },
    {
      "epoch": 0.22643548750281467,
      "grad_norm": 1.435144305229187,
      "learning_rate": 9.691373175016132e-05,
      "loss": 2.1021,
      "step": 1257
    },
    {
      "epoch": 0.22661562711101102,
      "grad_norm": 1.5235644578933716,
      "learning_rate": 9.690883152199873e-05,
      "loss": 2.337,
      "step": 1258
    },
    {
      "epoch": 0.2267957667192074,
      "grad_norm": 1.8211218118667603,
      "learning_rate": 9.690392753083254e-05,
      "loss": 2.2778,
      "step": 1259
    },
    {
      "epoch": 0.22697590632740375,
      "grad_norm": 1.7948099374771118,
      "learning_rate": 9.689901977705613e-05,
      "loss": 2.2129,
      "step": 1260
    },
    {
      "epoch": 0.2271560459356001,
      "grad_norm": 1.3149867057800293,
      "learning_rate": 9.689410826106317e-05,
      "loss": 1.8688,
      "step": 1261
    },
    {
      "epoch": 0.22733618554379645,
      "grad_norm": 1.3334577083587646,
      "learning_rate": 9.688919298324772e-05,
      "loss": 1.8792,
      "step": 1262
    },
    {
      "epoch": 0.2275163251519928,
      "grad_norm": 1.3534876108169556,
      "learning_rate": 9.688427394400404e-05,
      "loss": 1.7669,
      "step": 1263
    },
    {
      "epoch": 0.22769646476018915,
      "grad_norm": 1.2005510330200195,
      "learning_rate": 9.687935114372673e-05,
      "loss": 1.7123,
      "step": 1264
    },
    {
      "epoch": 0.2278766043683855,
      "grad_norm": 1.2224311828613281,
      "learning_rate": 9.68744245828107e-05,
      "loss": 1.5392,
      "step": 1265
    },
    {
      "epoch": 0.22805674397658185,
      "grad_norm": 1.3176820278167725,
      "learning_rate": 9.686949426165119e-05,
      "loss": 1.8028,
      "step": 1266
    },
    {
      "epoch": 0.2282368835847782,
      "grad_norm": 1.3573458194732666,
      "learning_rate": 9.686456018064367e-05,
      "loss": 1.7664,
      "step": 1267
    },
    {
      "epoch": 0.22841702319297455,
      "grad_norm": 1.3003135919570923,
      "learning_rate": 9.685962234018395e-05,
      "loss": 1.9275,
      "step": 1268
    },
    {
      "epoch": 0.2285971628011709,
      "grad_norm": 1.3143268823623657,
      "learning_rate": 9.685468074066815e-05,
      "loss": 1.9996,
      "step": 1269
    },
    {
      "epoch": 0.22877730240936725,
      "grad_norm": 1.291113257408142,
      "learning_rate": 9.684973538249267e-05,
      "loss": 1.7173,
      "step": 1270
    },
    {
      "epoch": 0.2289574420175636,
      "grad_norm": 1.305135726928711,
      "learning_rate": 9.684478626605427e-05,
      "loss": 2.1342,
      "step": 1271
    },
    {
      "epoch": 0.22913758162575995,
      "grad_norm": 1.3201669454574585,
      "learning_rate": 9.683983339174991e-05,
      "loss": 2.1349,
      "step": 1272
    },
    {
      "epoch": 0.22931772123395633,
      "grad_norm": 1.1403651237487793,
      "learning_rate": 9.683487675997693e-05,
      "loss": 1.5152,
      "step": 1273
    },
    {
      "epoch": 0.22949786084215268,
      "grad_norm": 1.1839333772659302,
      "learning_rate": 9.682991637113296e-05,
      "loss": 1.7999,
      "step": 1274
    },
    {
      "epoch": 0.22967800045034903,
      "grad_norm": 1.2619744539260864,
      "learning_rate": 9.68249522256159e-05,
      "loss": 1.8163,
      "step": 1275
    },
    {
      "epoch": 0.22985814005854538,
      "grad_norm": 1.321630835533142,
      "learning_rate": 9.681998432382398e-05,
      "loss": 1.9633,
      "step": 1276
    },
    {
      "epoch": 0.23003827966674173,
      "grad_norm": 1.3500498533248901,
      "learning_rate": 9.681501266615574e-05,
      "loss": 1.8286,
      "step": 1277
    },
    {
      "epoch": 0.23021841927493808,
      "grad_norm": 1.1742342710494995,
      "learning_rate": 9.681003725300997e-05,
      "loss": 1.732,
      "step": 1278
    },
    {
      "epoch": 0.23039855888313443,
      "grad_norm": 1.2814358472824097,
      "learning_rate": 9.680505808478582e-05,
      "loss": 1.8687,
      "step": 1279
    },
    {
      "epoch": 0.23057869849133078,
      "grad_norm": 1.2051386833190918,
      "learning_rate": 9.68000751618827e-05,
      "loss": 1.7608,
      "step": 1280
    },
    {
      "epoch": 0.23075883809952713,
      "grad_norm": 1.4166581630706787,
      "learning_rate": 9.679508848470033e-05,
      "loss": 1.7312,
      "step": 1281
    },
    {
      "epoch": 0.23093897770772348,
      "grad_norm": 1.256353497505188,
      "learning_rate": 9.679009805363879e-05,
      "loss": 1.8639,
      "step": 1282
    },
    {
      "epoch": 0.23111911731591983,
      "grad_norm": 1.284873366355896,
      "learning_rate": 9.678510386909836e-05,
      "loss": 1.8801,
      "step": 1283
    },
    {
      "epoch": 0.23129925692411618,
      "grad_norm": 1.3541619777679443,
      "learning_rate": 9.678010593147968e-05,
      "loss": 1.5903,
      "step": 1284
    },
    {
      "epoch": 0.23147939653231253,
      "grad_norm": 1.3045016527175903,
      "learning_rate": 9.67751042411837e-05,
      "loss": 1.9373,
      "step": 1285
    },
    {
      "epoch": 0.23165953614050888,
      "grad_norm": 1.3718358278274536,
      "learning_rate": 9.677009879861164e-05,
      "loss": 1.7672,
      "step": 1286
    },
    {
      "epoch": 0.23183967574870526,
      "grad_norm": 1.2938722372055054,
      "learning_rate": 9.676508960416502e-05,
      "loss": 1.9419,
      "step": 1287
    },
    {
      "epoch": 0.2320198153569016,
      "grad_norm": 1.3558599948883057,
      "learning_rate": 9.676007665824568e-05,
      "loss": 2.0485,
      "step": 1288
    },
    {
      "epoch": 0.23219995496509796,
      "grad_norm": 1.4920417070388794,
      "learning_rate": 9.67550599612558e-05,
      "loss": 2.2428,
      "step": 1289
    },
    {
      "epoch": 0.2323800945732943,
      "grad_norm": 1.3960587978363037,
      "learning_rate": 9.675003951359778e-05,
      "loss": 2.025,
      "step": 1290
    },
    {
      "epoch": 0.23256023418149066,
      "grad_norm": 1.442877173423767,
      "learning_rate": 9.674501531567434e-05,
      "loss": 2.0415,
      "step": 1291
    },
    {
      "epoch": 0.232740373789687,
      "grad_norm": 1.3591831922531128,
      "learning_rate": 9.673998736788855e-05,
      "loss": 2.0451,
      "step": 1292
    },
    {
      "epoch": 0.23292051339788336,
      "grad_norm": 1.408906102180481,
      "learning_rate": 9.673495567064375e-05,
      "loss": 1.8846,
      "step": 1293
    },
    {
      "epoch": 0.2331006530060797,
      "grad_norm": 1.3663188219070435,
      "learning_rate": 9.672992022434355e-05,
      "loss": 1.8281,
      "step": 1294
    },
    {
      "epoch": 0.23328079261427606,
      "grad_norm": 1.3005244731903076,
      "learning_rate": 9.672488102939193e-05,
      "loss": 1.7782,
      "step": 1295
    },
    {
      "epoch": 0.2334609322224724,
      "grad_norm": 1.3708949089050293,
      "learning_rate": 9.671983808619309e-05,
      "loss": 1.7935,
      "step": 1296
    },
    {
      "epoch": 0.23364107183066876,
      "grad_norm": 1.3792146444320679,
      "learning_rate": 9.671479139515162e-05,
      "loss": 1.9796,
      "step": 1297
    },
    {
      "epoch": 0.2338212114388651,
      "grad_norm": 1.336287260055542,
      "learning_rate": 9.670974095667232e-05,
      "loss": 1.8319,
      "step": 1298
    },
    {
      "epoch": 0.23400135104706146,
      "grad_norm": 1.4283175468444824,
      "learning_rate": 9.670468677116035e-05,
      "loss": 1.6735,
      "step": 1299
    },
    {
      "epoch": 0.2341814906552578,
      "grad_norm": 1.498521327972412,
      "learning_rate": 9.669962883902118e-05,
      "loss": 2.2208,
      "step": 1300
    },
    {
      "epoch": 0.2343616302634542,
      "grad_norm": 1.418279767036438,
      "learning_rate": 9.669456716066052e-05,
      "loss": 2.3285,
      "step": 1301
    },
    {
      "epoch": 0.23454176987165054,
      "grad_norm": 1.287266731262207,
      "learning_rate": 9.668950173648443e-05,
      "loss": 2.114,
      "step": 1302
    },
    {
      "epoch": 0.2347219094798469,
      "grad_norm": 1.5345252752304077,
      "learning_rate": 9.668443256689924e-05,
      "loss": 2.0366,
      "step": 1303
    },
    {
      "epoch": 0.23490204908804324,
      "grad_norm": 1.3359532356262207,
      "learning_rate": 9.667935965231161e-05,
      "loss": 2.1441,
      "step": 1304
    },
    {
      "epoch": 0.2350821886962396,
      "grad_norm": 1.2325955629348755,
      "learning_rate": 9.667428299312849e-05,
      "loss": 2.0052,
      "step": 1305
    },
    {
      "epoch": 0.23526232830443594,
      "grad_norm": 1.2899880409240723,
      "learning_rate": 9.666920258975712e-05,
      "loss": 1.7652,
      "step": 1306
    },
    {
      "epoch": 0.2354424679126323,
      "grad_norm": 1.4716989994049072,
      "learning_rate": 9.666411844260506e-05,
      "loss": 2.1174,
      "step": 1307
    },
    {
      "epoch": 0.23562260752082864,
      "grad_norm": 1.713647484779358,
      "learning_rate": 9.665903055208014e-05,
      "loss": 2.3615,
      "step": 1308
    },
    {
      "epoch": 0.235802747129025,
      "grad_norm": 1.6967394351959229,
      "learning_rate": 9.665393891859051e-05,
      "loss": 1.9691,
      "step": 1309
    },
    {
      "epoch": 0.23598288673722134,
      "grad_norm": 2.2882301807403564,
      "learning_rate": 9.664884354254463e-05,
      "loss": 2.504,
      "step": 1310
    },
    {
      "epoch": 0.2361630263454177,
      "grad_norm": 1.6432644128799438,
      "learning_rate": 9.664374442435124e-05,
      "loss": 2.1923,
      "step": 1311
    },
    {
      "epoch": 0.23634316595361404,
      "grad_norm": 1.2447680234909058,
      "learning_rate": 9.66386415644194e-05,
      "loss": 1.7736,
      "step": 1312
    },
    {
      "epoch": 0.2365233055618104,
      "grad_norm": 1.3229706287384033,
      "learning_rate": 9.663353496315844e-05,
      "loss": 1.9034,
      "step": 1313
    },
    {
      "epoch": 0.23670344517000674,
      "grad_norm": 1.295579195022583,
      "learning_rate": 9.662842462097802e-05,
      "loss": 1.9027,
      "step": 1314
    },
    {
      "epoch": 0.23688358477820312,
      "grad_norm": 1.2668462991714478,
      "learning_rate": 9.66233105382881e-05,
      "loss": 1.8819,
      "step": 1315
    },
    {
      "epoch": 0.23706372438639947,
      "grad_norm": 1.2516711950302124,
      "learning_rate": 9.661819271549892e-05,
      "loss": 1.7095,
      "step": 1316
    },
    {
      "epoch": 0.23724386399459582,
      "grad_norm": 1.3345041275024414,
      "learning_rate": 9.661307115302101e-05,
      "loss": 2.007,
      "step": 1317
    },
    {
      "epoch": 0.23742400360279217,
      "grad_norm": 1.4656058549880981,
      "learning_rate": 9.660794585126525e-05,
      "loss": 1.7817,
      "step": 1318
    },
    {
      "epoch": 0.23760414321098852,
      "grad_norm": 1.3824762105941772,
      "learning_rate": 9.660281681064276e-05,
      "loss": 2.1394,
      "step": 1319
    },
    {
      "epoch": 0.23778428281918487,
      "grad_norm": 1.351065993309021,
      "learning_rate": 9.659768403156502e-05,
      "loss": 1.7363,
      "step": 1320
    },
    {
      "epoch": 0.23796442242738122,
      "grad_norm": 1.3922151327133179,
      "learning_rate": 9.659254751444376e-05,
      "loss": 1.9992,
      "step": 1321
    },
    {
      "epoch": 0.23814456203557757,
      "grad_norm": 1.269749402999878,
      "learning_rate": 9.658740725969105e-05,
      "loss": 1.9123,
      "step": 1322
    },
    {
      "epoch": 0.23832470164377392,
      "grad_norm": 1.2521382570266724,
      "learning_rate": 9.65822632677192e-05,
      "loss": 1.833,
      "step": 1323
    },
    {
      "epoch": 0.23850484125197027,
      "grad_norm": 1.2525455951690674,
      "learning_rate": 9.657711553894088e-05,
      "loss": 1.871,
      "step": 1324
    },
    {
      "epoch": 0.23868498086016662,
      "grad_norm": 1.3701800107955933,
      "learning_rate": 9.657196407376904e-05,
      "loss": 2.071,
      "step": 1325
    },
    {
      "epoch": 0.23886512046836297,
      "grad_norm": 1.2621268033981323,
      "learning_rate": 9.656680887261693e-05,
      "loss": 1.9383,
      "step": 1326
    },
    {
      "epoch": 0.23904526007655932,
      "grad_norm": 1.344577670097351,
      "learning_rate": 9.65616499358981e-05,
      "loss": 1.7889,
      "step": 1327
    },
    {
      "epoch": 0.23922539968475567,
      "grad_norm": 1.2507952451705933,
      "learning_rate": 9.655648726402637e-05,
      "loss": 1.9257,
      "step": 1328
    },
    {
      "epoch": 0.23940553929295202,
      "grad_norm": 1.2649699449539185,
      "learning_rate": 9.655132085741593e-05,
      "loss": 1.8827,
      "step": 1329
    },
    {
      "epoch": 0.2395856789011484,
      "grad_norm": 1.362486481666565,
      "learning_rate": 9.65461507164812e-05,
      "loss": 1.5458,
      "step": 1330
    },
    {
      "epoch": 0.23976581850934475,
      "grad_norm": 1.2885550260543823,
      "learning_rate": 9.654097684163692e-05,
      "loss": 1.8555,
      "step": 1331
    },
    {
      "epoch": 0.2399459581175411,
      "grad_norm": 1.217755675315857,
      "learning_rate": 9.653579923329816e-05,
      "loss": 1.718,
      "step": 1332
    },
    {
      "epoch": 0.24012609772573745,
      "grad_norm": 1.3939847946166992,
      "learning_rate": 9.653061789188025e-05,
      "loss": 1.8666,
      "step": 1333
    },
    {
      "epoch": 0.2403062373339338,
      "grad_norm": 1.2871462106704712,
      "learning_rate": 9.652543281779884e-05,
      "loss": 1.7333,
      "step": 1334
    },
    {
      "epoch": 0.24048637694213015,
      "grad_norm": 1.3200628757476807,
      "learning_rate": 9.652024401146986e-05,
      "loss": 1.7965,
      "step": 1335
    },
    {
      "epoch": 0.2406665165503265,
      "grad_norm": 1.3936206102371216,
      "learning_rate": 9.651505147330958e-05,
      "loss": 1.6797,
      "step": 1336
    },
    {
      "epoch": 0.24084665615852285,
      "grad_norm": 1.3318467140197754,
      "learning_rate": 9.650985520373452e-05,
      "loss": 1.7061,
      "step": 1337
    },
    {
      "epoch": 0.2410267957667192,
      "grad_norm": 1.389526128768921,
      "learning_rate": 9.650465520316152e-05,
      "loss": 1.9818,
      "step": 1338
    },
    {
      "epoch": 0.24120693537491555,
      "grad_norm": 1.3966667652130127,
      "learning_rate": 9.649945147200774e-05,
      "loss": 1.714,
      "step": 1339
    },
    {
      "epoch": 0.2413870749831119,
      "grad_norm": 1.4819436073303223,
      "learning_rate": 9.649424401069062e-05,
      "loss": 1.9911,
      "step": 1340
    },
    {
      "epoch": 0.24156721459130825,
      "grad_norm": 1.4604023694992065,
      "learning_rate": 9.648903281962789e-05,
      "loss": 1.6126,
      "step": 1341
    },
    {
      "epoch": 0.2417473541995046,
      "grad_norm": 1.5306700468063354,
      "learning_rate": 9.648381789923757e-05,
      "loss": 1.8541,
      "step": 1342
    },
    {
      "epoch": 0.24192749380770096,
      "grad_norm": 1.387073278427124,
      "learning_rate": 9.647859924993804e-05,
      "loss": 1.662,
      "step": 1343
    },
    {
      "epoch": 0.24210763341589733,
      "grad_norm": 1.1959772109985352,
      "learning_rate": 9.647337687214792e-05,
      "loss": 1.4285,
      "step": 1344
    },
    {
      "epoch": 0.24228777302409368,
      "grad_norm": 1.3606319427490234,
      "learning_rate": 9.646815076628612e-05,
      "loss": 1.7306,
      "step": 1345
    },
    {
      "epoch": 0.24246791263229003,
      "grad_norm": 1.376552939414978,
      "learning_rate": 9.646292093277192e-05,
      "loss": 1.6686,
      "step": 1346
    },
    {
      "epoch": 0.24264805224048638,
      "grad_norm": 1.4229816198349,
      "learning_rate": 9.645768737202483e-05,
      "loss": 1.8728,
      "step": 1347
    },
    {
      "epoch": 0.24282819184868273,
      "grad_norm": 1.5853666067123413,
      "learning_rate": 9.645245008446468e-05,
      "loss": 2.1374,
      "step": 1348
    },
    {
      "epoch": 0.24300833145687908,
      "grad_norm": 1.2157371044158936,
      "learning_rate": 9.644720907051163e-05,
      "loss": 1.2887,
      "step": 1349
    },
    {
      "epoch": 0.24318847106507543,
      "grad_norm": 1.4207323789596558,
      "learning_rate": 9.644196433058608e-05,
      "loss": 1.7471,
      "step": 1350
    },
    {
      "epoch": 0.24336861067327178,
      "grad_norm": 1.4169080257415771,
      "learning_rate": 9.643671586510876e-05,
      "loss": 2.2114,
      "step": 1351
    },
    {
      "epoch": 0.24354875028146813,
      "grad_norm": 1.8168373107910156,
      "learning_rate": 9.643146367450073e-05,
      "loss": 2.3965,
      "step": 1352
    },
    {
      "epoch": 0.24372888988966449,
      "grad_norm": 1.3238023519515991,
      "learning_rate": 9.64262077591833e-05,
      "loss": 2.2433,
      "step": 1353
    },
    {
      "epoch": 0.24390902949786084,
      "grad_norm": 1.3712788820266724,
      "learning_rate": 9.642094811957809e-05,
      "loss": 2.2986,
      "step": 1354
    },
    {
      "epoch": 0.24408916910605719,
      "grad_norm": 1.4252575635910034,
      "learning_rate": 9.641568475610703e-05,
      "loss": 2.3612,
      "step": 1355
    },
    {
      "epoch": 0.24426930871425354,
      "grad_norm": 1.229488492012024,
      "learning_rate": 9.641041766919234e-05,
      "loss": 1.9304,
      "step": 1356
    },
    {
      "epoch": 0.24444944832244989,
      "grad_norm": 1.4757345914840698,
      "learning_rate": 9.640514685925656e-05,
      "loss": 2.0317,
      "step": 1357
    },
    {
      "epoch": 0.24462958793064626,
      "grad_norm": 1.4099195003509521,
      "learning_rate": 9.639987232672252e-05,
      "loss": 2.2138,
      "step": 1358
    },
    {
      "epoch": 0.24480972753884261,
      "grad_norm": 1.7490659952163696,
      "learning_rate": 9.63945940720133e-05,
      "loss": 2.1144,
      "step": 1359
    },
    {
      "epoch": 0.24498986714703896,
      "grad_norm": 2.476339101791382,
      "learning_rate": 9.638931209555234e-05,
      "loss": 2.7539,
      "step": 1360
    },
    {
      "epoch": 0.24517000675523531,
      "grad_norm": 1.3424603939056396,
      "learning_rate": 9.638402639776337e-05,
      "loss": 1.8887,
      "step": 1361
    },
    {
      "epoch": 0.24535014636343166,
      "grad_norm": 1.328654170036316,
      "learning_rate": 9.637873697907039e-05,
      "loss": 1.6789,
      "step": 1362
    },
    {
      "epoch": 0.24553028597162802,
      "grad_norm": 1.3418844938278198,
      "learning_rate": 9.637344383989773e-05,
      "loss": 2.1178,
      "step": 1363
    },
    {
      "epoch": 0.24571042557982437,
      "grad_norm": 1.2037616968154907,
      "learning_rate": 9.636814698066998e-05,
      "loss": 1.7504,
      "step": 1364
    },
    {
      "epoch": 0.24589056518802072,
      "grad_norm": 1.2933865785598755,
      "learning_rate": 9.636284640181205e-05,
      "loss": 1.7595,
      "step": 1365
    },
    {
      "epoch": 0.24607070479621707,
      "grad_norm": 2.4373340606689453,
      "learning_rate": 9.635754210374919e-05,
      "loss": 1.9394,
      "step": 1366
    },
    {
      "epoch": 0.24625084440441342,
      "grad_norm": 1.310178279876709,
      "learning_rate": 9.635223408690688e-05,
      "loss": 1.8679,
      "step": 1367
    },
    {
      "epoch": 0.24643098401260977,
      "grad_norm": 1.2312427759170532,
      "learning_rate": 9.634692235171092e-05,
      "loss": 1.704,
      "step": 1368
    },
    {
      "epoch": 0.24661112362080612,
      "grad_norm": 1.3104429244995117,
      "learning_rate": 9.634160689858743e-05,
      "loss": 1.6453,
      "step": 1369
    },
    {
      "epoch": 0.24679126322900247,
      "grad_norm": 1.2865937948226929,
      "learning_rate": 9.63362877279628e-05,
      "loss": 1.8056,
      "step": 1370
    },
    {
      "epoch": 0.24697140283719882,
      "grad_norm": 1.2945168018341064,
      "learning_rate": 9.633096484026375e-05,
      "loss": 1.9038,
      "step": 1371
    },
    {
      "epoch": 0.2471515424453952,
      "grad_norm": 1.4774664640426636,
      "learning_rate": 9.632563823591727e-05,
      "loss": 1.9085,
      "step": 1372
    },
    {
      "epoch": 0.24733168205359155,
      "grad_norm": 1.2102530002593994,
      "learning_rate": 9.632030791535063e-05,
      "loss": 1.7314,
      "step": 1373
    },
    {
      "epoch": 0.2475118216617879,
      "grad_norm": 1.210584282875061,
      "learning_rate": 9.631497387899146e-05,
      "loss": 1.6257,
      "step": 1374
    },
    {
      "epoch": 0.24769196126998425,
      "grad_norm": 1.19646418094635,
      "learning_rate": 9.630963612726766e-05,
      "loss": 1.7282,
      "step": 1375
    },
    {
      "epoch": 0.2478721008781806,
      "grad_norm": 1.3051313161849976,
      "learning_rate": 9.63042946606074e-05,
      "loss": 1.9294,
      "step": 1376
    },
    {
      "epoch": 0.24805224048637695,
      "grad_norm": 1.1502336263656616,
      "learning_rate": 9.629894947943916e-05,
      "loss": 1.8746,
      "step": 1377
    },
    {
      "epoch": 0.2482323800945733,
      "grad_norm": 1.210191011428833,
      "learning_rate": 9.629360058419176e-05,
      "loss": 1.7942,
      "step": 1378
    },
    {
      "epoch": 0.24841251970276965,
      "grad_norm": 1.3472601175308228,
      "learning_rate": 9.628824797529428e-05,
      "loss": 1.9549,
      "step": 1379
    },
    {
      "epoch": 0.248592659310966,
      "grad_norm": 1.316145420074463,
      "learning_rate": 9.628289165317608e-05,
      "loss": 1.6018,
      "step": 1380
    },
    {
      "epoch": 0.24877279891916235,
      "grad_norm": 1.2660845518112183,
      "learning_rate": 9.627753161826685e-05,
      "loss": 1.738,
      "step": 1381
    },
    {
      "epoch": 0.2489529385273587,
      "grad_norm": 1.3599636554718018,
      "learning_rate": 9.627216787099658e-05,
      "loss": 1.9489,
      "step": 1382
    },
    {
      "epoch": 0.24913307813555505,
      "grad_norm": 1.2624542713165283,
      "learning_rate": 9.626680041179555e-05,
      "loss": 1.7027,
      "step": 1383
    },
    {
      "epoch": 0.2493132177437514,
      "grad_norm": 1.419079303741455,
      "learning_rate": 9.626142924109432e-05,
      "loss": 1.6361,
      "step": 1384
    },
    {
      "epoch": 0.24949335735194775,
      "grad_norm": 1.394989013671875,
      "learning_rate": 9.625605435932377e-05,
      "loss": 1.8748,
      "step": 1385
    },
    {
      "epoch": 0.2496734969601441,
      "grad_norm": 1.3571964502334595,
      "learning_rate": 9.625067576691505e-05,
      "loss": 1.6741,
      "step": 1386
    },
    {
      "epoch": 0.24985363656834048,
      "grad_norm": 1.4658979177474976,
      "learning_rate": 9.624529346429967e-05,
      "loss": 1.8919,
      "step": 1387
    },
    {
      "epoch": 0.2500337761765368,
      "grad_norm": 1.3335614204406738,
      "learning_rate": 9.623990745190938e-05,
      "loss": 1.8162,
      "step": 1388
    },
    {
      "epoch": 0.2502139157847332,
      "grad_norm": 1.2806657552719116,
      "learning_rate": 9.623451773017622e-05,
      "loss": 1.7118,
      "step": 1389
    },
    {
      "epoch": 0.2503940553929295,
      "grad_norm": 1.41310453414917,
      "learning_rate": 9.622912429953257e-05,
      "loss": 1.9767,
      "step": 1390
    },
    {
      "epoch": 0.2505741950011259,
      "grad_norm": 1.4300659894943237,
      "learning_rate": 9.62237271604111e-05,
      "loss": 1.9149,
      "step": 1391
    },
    {
      "epoch": 0.2507543346093222,
      "grad_norm": 1.3532480001449585,
      "learning_rate": 9.621832631324474e-05,
      "loss": 1.7847,
      "step": 1392
    },
    {
      "epoch": 0.2509344742175186,
      "grad_norm": 1.4936203956604004,
      "learning_rate": 9.621292175846675e-05,
      "loss": 2.0031,
      "step": 1393
    },
    {
      "epoch": 0.2511146138257149,
      "grad_norm": 1.3195195198059082,
      "learning_rate": 9.620751349651069e-05,
      "loss": 1.6065,
      "step": 1394
    },
    {
      "epoch": 0.2512947534339113,
      "grad_norm": 1.4474780559539795,
      "learning_rate": 9.620210152781042e-05,
      "loss": 1.9245,
      "step": 1395
    },
    {
      "epoch": 0.25147489304210763,
      "grad_norm": 1.5137771368026733,
      "learning_rate": 9.619668585280005e-05,
      "loss": 2.1028,
      "step": 1396
    },
    {
      "epoch": 0.251655032650304,
      "grad_norm": 1.3129740953445435,
      "learning_rate": 9.619126647191405e-05,
      "loss": 1.7898,
      "step": 1397
    },
    {
      "epoch": 0.25183517225850033,
      "grad_norm": 1.4312471151351929,
      "learning_rate": 9.618584338558716e-05,
      "loss": 2.0074,
      "step": 1398
    },
    {
      "epoch": 0.2520153118666967,
      "grad_norm": 1.2129664421081543,
      "learning_rate": 9.618041659425439e-05,
      "loss": 1.6612,
      "step": 1399
    },
    {
      "epoch": 0.25219545147489303,
      "grad_norm": 1.4841365814208984,
      "learning_rate": 9.617498609835112e-05,
      "loss": 1.9451,
      "step": 1400
    },
    {
      "epoch": 0.2523755910830894,
      "grad_norm": 1.3020541667938232,
      "learning_rate": 9.616955189831293e-05,
      "loss": 2.1399,
      "step": 1401
    },
    {
      "epoch": 0.25255573069128573,
      "grad_norm": 1.1937135457992554,
      "learning_rate": 9.616411399457579e-05,
      "loss": 2.135,
      "step": 1402
    },
    {
      "epoch": 0.2527358702994821,
      "grad_norm": 1.26156485080719,
      "learning_rate": 9.615867238757593e-05,
      "loss": 2.1286,
      "step": 1403
    },
    {
      "epoch": 0.25291600990767843,
      "grad_norm": 1.1347700357437134,
      "learning_rate": 9.615322707774982e-05,
      "loss": 2.0889,
      "step": 1404
    },
    {
      "epoch": 0.2530961495158748,
      "grad_norm": 1.3031673431396484,
      "learning_rate": 9.614777806553432e-05,
      "loss": 1.6973,
      "step": 1405
    },
    {
      "epoch": 0.25327628912407113,
      "grad_norm": 1.3179755210876465,
      "learning_rate": 9.614232535136656e-05,
      "loss": 1.8961,
      "step": 1406
    },
    {
      "epoch": 0.2534564287322675,
      "grad_norm": 1.5497852563858032,
      "learning_rate": 9.613686893568392e-05,
      "loss": 2.3143,
      "step": 1407
    },
    {
      "epoch": 0.2536365683404639,
      "grad_norm": 1.6290380954742432,
      "learning_rate": 9.613140881892413e-05,
      "loss": 2.3828,
      "step": 1408
    },
    {
      "epoch": 0.25381670794866024,
      "grad_norm": 1.9030406475067139,
      "learning_rate": 9.61259450015252e-05,
      "loss": 2.4538,
      "step": 1409
    },
    {
      "epoch": 0.2539968475568566,
      "grad_norm": 2.0345888137817383,
      "learning_rate": 9.612047748392543e-05,
      "loss": 2.5833,
      "step": 1410
    },
    {
      "epoch": 0.25417698716505294,
      "grad_norm": 1.375148892402649,
      "learning_rate": 9.61150062665634e-05,
      "loss": 1.782,
      "step": 1411
    },
    {
      "epoch": 0.2543571267732493,
      "grad_norm": 1.2778311967849731,
      "learning_rate": 9.610953134987806e-05,
      "loss": 1.7237,
      "step": 1412
    },
    {
      "epoch": 0.25453726638144564,
      "grad_norm": 1.3053090572357178,
      "learning_rate": 9.610405273430855e-05,
      "loss": 1.9499,
      "step": 1413
    },
    {
      "epoch": 0.254717405989642,
      "grad_norm": 1.39927339553833,
      "learning_rate": 9.60985704202944e-05,
      "loss": 1.8895,
      "step": 1414
    },
    {
      "epoch": 0.25489754559783834,
      "grad_norm": 1.3640800714492798,
      "learning_rate": 9.609308440827538e-05,
      "loss": 1.936,
      "step": 1415
    },
    {
      "epoch": 0.2550776852060347,
      "grad_norm": 1.265154480934143,
      "learning_rate": 9.608759469869158e-05,
      "loss": 1.7464,
      "step": 1416
    },
    {
      "epoch": 0.25525782481423104,
      "grad_norm": 1.2529346942901611,
      "learning_rate": 9.608210129198338e-05,
      "loss": 1.6161,
      "step": 1417
    },
    {
      "epoch": 0.2554379644224274,
      "grad_norm": 1.263816237449646,
      "learning_rate": 9.607660418859146e-05,
      "loss": 1.851,
      "step": 1418
    },
    {
      "epoch": 0.25561810403062374,
      "grad_norm": 1.346397876739502,
      "learning_rate": 9.60711033889568e-05,
      "loss": 1.6665,
      "step": 1419
    },
    {
      "epoch": 0.2557982436388201,
      "grad_norm": 1.2287291288375854,
      "learning_rate": 9.606559889352064e-05,
      "loss": 1.7784,
      "step": 1420
    },
    {
      "epoch": 0.25597838324701644,
      "grad_norm": 1.246948480606079,
      "learning_rate": 9.606009070272461e-05,
      "loss": 1.6907,
      "step": 1421
    },
    {
      "epoch": 0.2561585228552128,
      "grad_norm": 1.3155512809753418,
      "learning_rate": 9.605457881701051e-05,
      "loss": 1.6913,
      "step": 1422
    },
    {
      "epoch": 0.25633866246340914,
      "grad_norm": 1.2273707389831543,
      "learning_rate": 9.604906323682057e-05,
      "loss": 1.7287,
      "step": 1423
    },
    {
      "epoch": 0.2565188020716055,
      "grad_norm": 1.3157439231872559,
      "learning_rate": 9.604354396259718e-05,
      "loss": 1.7937,
      "step": 1424
    },
    {
      "epoch": 0.25669894167980184,
      "grad_norm": 1.4950356483459473,
      "learning_rate": 9.603802099478312e-05,
      "loss": 2.0248,
      "step": 1425
    },
    {
      "epoch": 0.2568790812879982,
      "grad_norm": 1.3335496187210083,
      "learning_rate": 9.603249433382144e-05,
      "loss": 1.831,
      "step": 1426
    },
    {
      "epoch": 0.25705922089619454,
      "grad_norm": 1.334462285041809,
      "learning_rate": 9.60269639801555e-05,
      "loss": 1.8213,
      "step": 1427
    },
    {
      "epoch": 0.2572393605043909,
      "grad_norm": 1.2646679878234863,
      "learning_rate": 9.602142993422892e-05,
      "loss": 1.7785,
      "step": 1428
    },
    {
      "epoch": 0.25741950011258724,
      "grad_norm": 1.2021260261535645,
      "learning_rate": 9.601589219648563e-05,
      "loss": 1.6934,
      "step": 1429
    },
    {
      "epoch": 0.2575996397207836,
      "grad_norm": 1.4244736433029175,
      "learning_rate": 9.60103507673699e-05,
      "loss": 1.7952,
      "step": 1430
    },
    {
      "epoch": 0.25777977932897994,
      "grad_norm": 1.3532769680023193,
      "learning_rate": 9.600480564732623e-05,
      "loss": 2.0208,
      "step": 1431
    },
    {
      "epoch": 0.2579599189371763,
      "grad_norm": 1.259838581085205,
      "learning_rate": 9.599925683679948e-05,
      "loss": 1.7265,
      "step": 1432
    },
    {
      "epoch": 0.25814005854537264,
      "grad_norm": 1.320878028869629,
      "learning_rate": 9.599370433623474e-05,
      "loss": 1.7461,
      "step": 1433
    },
    {
      "epoch": 0.258320198153569,
      "grad_norm": 1.3004517555236816,
      "learning_rate": 9.598814814607744e-05,
      "loss": 1.7427,
      "step": 1434
    },
    {
      "epoch": 0.25850033776176534,
      "grad_norm": 1.3136227130889893,
      "learning_rate": 9.598258826677328e-05,
      "loss": 1.8261,
      "step": 1435
    },
    {
      "epoch": 0.25868047736996175,
      "grad_norm": 1.2894428968429565,
      "learning_rate": 9.597702469876832e-05,
      "loss": 1.9278,
      "step": 1436
    },
    {
      "epoch": 0.2588606169781581,
      "grad_norm": 1.3688976764678955,
      "learning_rate": 9.597145744250882e-05,
      "loss": 1.6501,
      "step": 1437
    },
    {
      "epoch": 0.25904075658635445,
      "grad_norm": 1.3156013488769531,
      "learning_rate": 9.596588649844138e-05,
      "loss": 1.7856,
      "step": 1438
    },
    {
      "epoch": 0.2592208961945508,
      "grad_norm": 1.3687058687210083,
      "learning_rate": 9.596031186701294e-05,
      "loss": 1.7681,
      "step": 1439
    },
    {
      "epoch": 0.25940103580274715,
      "grad_norm": 1.341892957687378,
      "learning_rate": 9.595473354867066e-05,
      "loss": 2.0313,
      "step": 1440
    },
    {
      "epoch": 0.2595811754109435,
      "grad_norm": 1.2842971086502075,
      "learning_rate": 9.594915154386201e-05,
      "loss": 1.8891,
      "step": 1441
    },
    {
      "epoch": 0.25976131501913985,
      "grad_norm": 1.5312058925628662,
      "learning_rate": 9.594356585303484e-05,
      "loss": 1.9411,
      "step": 1442
    },
    {
      "epoch": 0.2599414546273362,
      "grad_norm": 1.495235800743103,
      "learning_rate": 9.593797647663718e-05,
      "loss": 1.8542,
      "step": 1443
    },
    {
      "epoch": 0.26012159423553255,
      "grad_norm": 1.4159555435180664,
      "learning_rate": 9.593238341511743e-05,
      "loss": 1.5517,
      "step": 1444
    },
    {
      "epoch": 0.2603017338437289,
      "grad_norm": 1.3420803546905518,
      "learning_rate": 9.592678666892425e-05,
      "loss": 2.0079,
      "step": 1445
    },
    {
      "epoch": 0.26048187345192525,
      "grad_norm": 1.3957349061965942,
      "learning_rate": 9.592118623850661e-05,
      "loss": 1.8886,
      "step": 1446
    },
    {
      "epoch": 0.2606620130601216,
      "grad_norm": 1.4691698551177979,
      "learning_rate": 9.591558212431378e-05,
      "loss": 2.0943,
      "step": 1447
    },
    {
      "epoch": 0.26084215266831795,
      "grad_norm": 1.4037094116210938,
      "learning_rate": 9.590997432679532e-05,
      "loss": 1.6345,
      "step": 1448
    },
    {
      "epoch": 0.2610222922765143,
      "grad_norm": 1.4972895383834839,
      "learning_rate": 9.590436284640107e-05,
      "loss": 2.0783,
      "step": 1449
    },
    {
      "epoch": 0.26120243188471065,
      "grad_norm": 1.3901547193527222,
      "learning_rate": 9.589874768358121e-05,
      "loss": 1.767,
      "step": 1450
    },
    {
      "epoch": 0.261382571492907,
      "grad_norm": 1.3561162948608398,
      "learning_rate": 9.589312883878615e-05,
      "loss": 2.0377,
      "step": 1451
    },
    {
      "epoch": 0.26156271110110335,
      "grad_norm": 1.2409238815307617,
      "learning_rate": 9.588750631246667e-05,
      "loss": 2.2851,
      "step": 1452
    },
    {
      "epoch": 0.2617428507092997,
      "grad_norm": 1.6290526390075684,
      "learning_rate": 9.588188010507378e-05,
      "loss": 2.5258,
      "step": 1453
    },
    {
      "epoch": 0.26192299031749605,
      "grad_norm": 1.3076307773590088,
      "learning_rate": 9.58762502170588e-05,
      "loss": 2.4793,
      "step": 1454
    },
    {
      "epoch": 0.2621031299256924,
      "grad_norm": 1.3256698846817017,
      "learning_rate": 9.587061664887338e-05,
      "loss": 2.3607,
      "step": 1455
    },
    {
      "epoch": 0.26228326953388875,
      "grad_norm": 1.5011816024780273,
      "learning_rate": 9.586497940096944e-05,
      "loss": 2.2197,
      "step": 1456
    },
    {
      "epoch": 0.2624634091420851,
      "grad_norm": 1.44423246383667,
      "learning_rate": 9.585933847379919e-05,
      "loss": 2.1764,
      "step": 1457
    },
    {
      "epoch": 0.26264354875028145,
      "grad_norm": 1.5568348169326782,
      "learning_rate": 9.585369386781514e-05,
      "loss": 2.3642,
      "step": 1458
    },
    {
      "epoch": 0.2628236883584778,
      "grad_norm": 1.6485800743103027,
      "learning_rate": 9.58480455834701e-05,
      "loss": 2.1984,
      "step": 1459
    },
    {
      "epoch": 0.26300382796667415,
      "grad_norm": 2.164614200592041,
      "learning_rate": 9.584239362121719e-05,
      "loss": 2.6011,
      "step": 1460
    },
    {
      "epoch": 0.2631839675748705,
      "grad_norm": 1.3199414014816284,
      "learning_rate": 9.583673798150979e-05,
      "loss": 2.0123,
      "step": 1461
    },
    {
      "epoch": 0.26336410718306685,
      "grad_norm": 1.286777138710022,
      "learning_rate": 9.583107866480159e-05,
      "loss": 1.901,
      "step": 1462
    },
    {
      "epoch": 0.2635442467912632,
      "grad_norm": 1.2382673025131226,
      "learning_rate": 9.582541567154657e-05,
      "loss": 1.653,
      "step": 1463
    },
    {
      "epoch": 0.26372438639945955,
      "grad_norm": 1.1522717475891113,
      "learning_rate": 9.581974900219905e-05,
      "loss": 1.4788,
      "step": 1464
    },
    {
      "epoch": 0.26390452600765596,
      "grad_norm": 1.2540417909622192,
      "learning_rate": 9.581407865721357e-05,
      "loss": 1.7519,
      "step": 1465
    },
    {
      "epoch": 0.2640846656158523,
      "grad_norm": 1.323789119720459,
      "learning_rate": 9.580840463704501e-05,
      "loss": 1.9791,
      "step": 1466
    },
    {
      "epoch": 0.26426480522404866,
      "grad_norm": 1.247562050819397,
      "learning_rate": 9.580272694214854e-05,
      "loss": 1.6467,
      "step": 1467
    },
    {
      "epoch": 0.264444944832245,
      "grad_norm": 1.389323115348816,
      "learning_rate": 9.579704557297964e-05,
      "loss": 1.9478,
      "step": 1468
    },
    {
      "epoch": 0.26462508444044136,
      "grad_norm": 1.429758906364441,
      "learning_rate": 9.579136052999405e-05,
      "loss": 2.1681,
      "step": 1469
    },
    {
      "epoch": 0.2648052240486377,
      "grad_norm": 1.4671255350112915,
      "learning_rate": 9.578567181364781e-05,
      "loss": 1.9981,
      "step": 1470
    },
    {
      "epoch": 0.26498536365683406,
      "grad_norm": 1.3760862350463867,
      "learning_rate": 9.577997942439729e-05,
      "loss": 1.976,
      "step": 1471
    },
    {
      "epoch": 0.2651655032650304,
      "grad_norm": 1.1407766342163086,
      "learning_rate": 9.577428336269912e-05,
      "loss": 1.5791,
      "step": 1472
    },
    {
      "epoch": 0.26534564287322676,
      "grad_norm": 1.1693896055221558,
      "learning_rate": 9.576858362901023e-05,
      "loss": 1.7841,
      "step": 1473
    },
    {
      "epoch": 0.2655257824814231,
      "grad_norm": 1.250060796737671,
      "learning_rate": 9.576288022378784e-05,
      "loss": 1.7979,
      "step": 1474
    },
    {
      "epoch": 0.26570592208961946,
      "grad_norm": 1.1622016429901123,
      "learning_rate": 9.575717314748951e-05,
      "loss": 1.8498,
      "step": 1475
    },
    {
      "epoch": 0.2658860616978158,
      "grad_norm": 1.2893506288528442,
      "learning_rate": 9.575146240057302e-05,
      "loss": 1.8537,
      "step": 1476
    },
    {
      "epoch": 0.26606620130601216,
      "grad_norm": 1.386186957359314,
      "learning_rate": 9.574574798349651e-05,
      "loss": 1.8894,
      "step": 1477
    },
    {
      "epoch": 0.2662463409142085,
      "grad_norm": 1.2083815336227417,
      "learning_rate": 9.57400298967184e-05,
      "loss": 1.7881,
      "step": 1478
    },
    {
      "epoch": 0.26642648052240486,
      "grad_norm": 1.4141194820404053,
      "learning_rate": 9.573430814069734e-05,
      "loss": 1.7631,
      "step": 1479
    },
    {
      "epoch": 0.2666066201306012,
      "grad_norm": 1.270307183265686,
      "learning_rate": 9.572858271589237e-05,
      "loss": 2.0846,
      "step": 1480
    },
    {
      "epoch": 0.26678675973879756,
      "grad_norm": 1.4181755781173706,
      "learning_rate": 9.572285362276277e-05,
      "loss": 1.975,
      "step": 1481
    },
    {
      "epoch": 0.2669668993469939,
      "grad_norm": 1.3155264854431152,
      "learning_rate": 9.571712086176812e-05,
      "loss": 1.9158,
      "step": 1482
    },
    {
      "epoch": 0.26714703895519026,
      "grad_norm": 1.1629223823547363,
      "learning_rate": 9.57113844333683e-05,
      "loss": 1.6506,
      "step": 1483
    },
    {
      "epoch": 0.2673271785633866,
      "grad_norm": 1.3701138496398926,
      "learning_rate": 9.570564433802348e-05,
      "loss": 1.6617,
      "step": 1484
    },
    {
      "epoch": 0.26750731817158296,
      "grad_norm": 1.2416373491287231,
      "learning_rate": 9.569990057619414e-05,
      "loss": 1.962,
      "step": 1485
    },
    {
      "epoch": 0.2676874577797793,
      "grad_norm": 1.343294620513916,
      "learning_rate": 9.569415314834104e-05,
      "loss": 1.7463,
      "step": 1486
    },
    {
      "epoch": 0.26786759738797566,
      "grad_norm": 1.3468525409698486,
      "learning_rate": 9.568840205492522e-05,
      "loss": 1.8854,
      "step": 1487
    },
    {
      "epoch": 0.268047736996172,
      "grad_norm": 1.349214792251587,
      "learning_rate": 9.568264729640805e-05,
      "loss": 1.7979,
      "step": 1488
    },
    {
      "epoch": 0.26822787660436836,
      "grad_norm": 1.4037396907806396,
      "learning_rate": 9.567688887325116e-05,
      "loss": 1.9524,
      "step": 1489
    },
    {
      "epoch": 0.2684080162125647,
      "grad_norm": 1.317078948020935,
      "learning_rate": 9.567112678591651e-05,
      "loss": 1.7042,
      "step": 1490
    },
    {
      "epoch": 0.26858815582076107,
      "grad_norm": 1.3170051574707031,
      "learning_rate": 9.56653610348663e-05,
      "loss": 1.4587,
      "step": 1491
    },
    {
      "epoch": 0.2687682954289574,
      "grad_norm": 1.2764688730239868,
      "learning_rate": 9.565959162056307e-05,
      "loss": 1.6758,
      "step": 1492
    },
    {
      "epoch": 0.2689484350371538,
      "grad_norm": 1.2370458841323853,
      "learning_rate": 9.565381854346965e-05,
      "loss": 1.456,
      "step": 1493
    },
    {
      "epoch": 0.26912857464535017,
      "grad_norm": 1.325675368309021,
      "learning_rate": 9.564804180404914e-05,
      "loss": 1.5876,
      "step": 1494
    },
    {
      "epoch": 0.2693087142535465,
      "grad_norm": 1.4240556955337524,
      "learning_rate": 9.564226140276493e-05,
      "loss": 1.6717,
      "step": 1495
    },
    {
      "epoch": 0.26948885386174287,
      "grad_norm": 1.353955864906311,
      "learning_rate": 9.563647734008077e-05,
      "loss": 1.725,
      "step": 1496
    },
    {
      "epoch": 0.2696689934699392,
      "grad_norm": 1.3728160858154297,
      "learning_rate": 9.56306896164606e-05,
      "loss": 1.7114,
      "step": 1497
    },
    {
      "epoch": 0.2698491330781356,
      "grad_norm": 1.401940941810608,
      "learning_rate": 9.562489823236877e-05,
      "loss": 1.8394,
      "step": 1498
    },
    {
      "epoch": 0.2700292726863319,
      "grad_norm": 1.4119722843170166,
      "learning_rate": 9.561910318826981e-05,
      "loss": 1.5947,
      "step": 1499
    },
    {
      "epoch": 0.2702094122945283,
      "grad_norm": 1.5551724433898926,
      "learning_rate": 9.56133044846286e-05,
      "loss": 2.0708,
      "step": 1500
    },
    {
      "epoch": 0.2703895519027246,
      "grad_norm": 1.380489706993103,
      "learning_rate": 9.560750212191033e-05,
      "loss": 2.3114,
      "step": 1501
    },
    {
      "epoch": 0.270569691510921,
      "grad_norm": 1.2792739868164062,
      "learning_rate": 9.560169610058045e-05,
      "loss": 2.0318,
      "step": 1502
    },
    {
      "epoch": 0.2707498311191173,
      "grad_norm": 2.988199472427368,
      "learning_rate": 9.559588642110473e-05,
      "loss": 2.5278,
      "step": 1503
    },
    {
      "epoch": 0.2709299707273137,
      "grad_norm": 1.3033597469329834,
      "learning_rate": 9.559007308394921e-05,
      "loss": 2.0703,
      "step": 1504
    },
    {
      "epoch": 0.27111011033551,
      "grad_norm": 1.2994680404663086,
      "learning_rate": 9.558425608958023e-05,
      "loss": 2.2384,
      "step": 1505
    },
    {
      "epoch": 0.2712902499437064,
      "grad_norm": 1.3633158206939697,
      "learning_rate": 9.557843543846443e-05,
      "loss": 2.1388,
      "step": 1506
    },
    {
      "epoch": 0.2714703895519027,
      "grad_norm": 1.3668885231018066,
      "learning_rate": 9.557261113106875e-05,
      "loss": 1.9701,
      "step": 1507
    },
    {
      "epoch": 0.2716505291600991,
      "grad_norm": 1.5243639945983887,
      "learning_rate": 9.556678316786038e-05,
      "loss": 2.4586,
      "step": 1508
    },
    {
      "epoch": 0.2718306687682954,
      "grad_norm": 1.534784197807312,
      "learning_rate": 9.556095154930688e-05,
      "loss": 2.1406,
      "step": 1509
    },
    {
      "epoch": 0.2720108083764918,
      "grad_norm": 1.9293186664581299,
      "learning_rate": 9.555511627587603e-05,
      "loss": 2.6017,
      "step": 1510
    },
    {
      "epoch": 0.2721909479846881,
      "grad_norm": 1.9219658374786377,
      "learning_rate": 9.554927734803594e-05,
      "loss": 2.334,
      "step": 1511
    },
    {
      "epoch": 0.2723710875928845,
      "grad_norm": 1.1688010692596436,
      "learning_rate": 9.5543434766255e-05,
      "loss": 1.7778,
      "step": 1512
    },
    {
      "epoch": 0.2725512272010808,
      "grad_norm": 1.214694619178772,
      "learning_rate": 9.553758853100192e-05,
      "loss": 1.7873,
      "step": 1513
    },
    {
      "epoch": 0.2727313668092772,
      "grad_norm": 1.2433154582977295,
      "learning_rate": 9.553173864274567e-05,
      "loss": 1.8045,
      "step": 1514
    },
    {
      "epoch": 0.2729115064174735,
      "grad_norm": 1.2065643072128296,
      "learning_rate": 9.552588510195552e-05,
      "loss": 1.7442,
      "step": 1515
    },
    {
      "epoch": 0.2730916460256699,
      "grad_norm": 1.1479398012161255,
      "learning_rate": 9.552002790910102e-05,
      "loss": 1.7399,
      "step": 1516
    },
    {
      "epoch": 0.2732717856338662,
      "grad_norm": 1.2715672254562378,
      "learning_rate": 9.551416706465208e-05,
      "loss": 1.6718,
      "step": 1517
    },
    {
      "epoch": 0.2734519252420626,
      "grad_norm": 1.1501915454864502,
      "learning_rate": 9.550830256907884e-05,
      "loss": 1.6586,
      "step": 1518
    },
    {
      "epoch": 0.2736320648502589,
      "grad_norm": 1.1713447570800781,
      "learning_rate": 9.550243442285169e-05,
      "loss": 1.6865,
      "step": 1519
    },
    {
      "epoch": 0.2738122044584553,
      "grad_norm": 1.2487322092056274,
      "learning_rate": 9.549656262644144e-05,
      "loss": 1.8284,
      "step": 1520
    },
    {
      "epoch": 0.2739923440666516,
      "grad_norm": 1.3894429206848145,
      "learning_rate": 9.54906871803191e-05,
      "loss": 2.1056,
      "step": 1521
    },
    {
      "epoch": 0.27417248367484803,
      "grad_norm": 1.2422899007797241,
      "learning_rate": 9.548480808495599e-05,
      "loss": 1.7347,
      "step": 1522
    },
    {
      "epoch": 0.2743526232830444,
      "grad_norm": 1.3186671733856201,
      "learning_rate": 9.547892534082372e-05,
      "loss": 1.9204,
      "step": 1523
    },
    {
      "epoch": 0.27453276289124073,
      "grad_norm": 1.209747076034546,
      "learning_rate": 9.547303894839422e-05,
      "loss": 1.572,
      "step": 1524
    },
    {
      "epoch": 0.2747129024994371,
      "grad_norm": 1.217474341392517,
      "learning_rate": 9.546714890813968e-05,
      "loss": 1.7755,
      "step": 1525
    },
    {
      "epoch": 0.27489304210763343,
      "grad_norm": 1.2285915613174438,
      "learning_rate": 9.546125522053261e-05,
      "loss": 1.7449,
      "step": 1526
    },
    {
      "epoch": 0.2750731817158298,
      "grad_norm": 1.1651476621627808,
      "learning_rate": 9.545535788604578e-05,
      "loss": 1.4557,
      "step": 1527
    },
    {
      "epoch": 0.27525332132402613,
      "grad_norm": 1.236716866493225,
      "learning_rate": 9.544945690515229e-05,
      "loss": 1.9389,
      "step": 1528
    },
    {
      "epoch": 0.2754334609322225,
      "grad_norm": 1.240716814994812,
      "learning_rate": 9.54435522783255e-05,
      "loss": 1.8538,
      "step": 1529
    },
    {
      "epoch": 0.27561360054041883,
      "grad_norm": 1.2599238157272339,
      "learning_rate": 9.543764400603908e-05,
      "loss": 1.8716,
      "step": 1530
    },
    {
      "epoch": 0.2757937401486152,
      "grad_norm": 1.4449176788330078,
      "learning_rate": 9.543173208876698e-05,
      "loss": 1.9514,
      "step": 1531
    },
    {
      "epoch": 0.27597387975681154,
      "grad_norm": 1.2154200077056885,
      "learning_rate": 9.542581652698348e-05,
      "loss": 1.5626,
      "step": 1532
    },
    {
      "epoch": 0.2761540193650079,
      "grad_norm": 1.3090604543685913,
      "learning_rate": 9.54198973211631e-05,
      "loss": 1.7569,
      "step": 1533
    },
    {
      "epoch": 0.27633415897320424,
      "grad_norm": 1.328647255897522,
      "learning_rate": 9.541397447178067e-05,
      "loss": 1.6839,
      "step": 1534
    },
    {
      "epoch": 0.2765142985814006,
      "grad_norm": 1.2050557136535645,
      "learning_rate": 9.540804797931134e-05,
      "loss": 1.7401,
      "step": 1535
    },
    {
      "epoch": 0.27669443818959694,
      "grad_norm": 1.3914463520050049,
      "learning_rate": 9.54021178442305e-05,
      "loss": 1.8738,
      "step": 1536
    },
    {
      "epoch": 0.2768745777977933,
      "grad_norm": 1.3764175176620483,
      "learning_rate": 9.539618406701389e-05,
      "loss": 1.654,
      "step": 1537
    },
    {
      "epoch": 0.27705471740598964,
      "grad_norm": 1.413156270980835,
      "learning_rate": 9.539024664813752e-05,
      "loss": 1.9631,
      "step": 1538
    },
    {
      "epoch": 0.277234857014186,
      "grad_norm": 1.3338452577590942,
      "learning_rate": 9.538430558807766e-05,
      "loss": 1.7974,
      "step": 1539
    },
    {
      "epoch": 0.27741499662238234,
      "grad_norm": 1.2033109664916992,
      "learning_rate": 9.537836088731091e-05,
      "loss": 1.5907,
      "step": 1540
    },
    {
      "epoch": 0.2775951362305787,
      "grad_norm": 1.3737608194351196,
      "learning_rate": 9.537241254631416e-05,
      "loss": 1.8458,
      "step": 1541
    },
    {
      "epoch": 0.27777527583877504,
      "grad_norm": 1.5115687847137451,
      "learning_rate": 9.536646056556458e-05,
      "loss": 1.9156,
      "step": 1542
    },
    {
      "epoch": 0.2779554154469714,
      "grad_norm": 1.53287935256958,
      "learning_rate": 9.536050494553961e-05,
      "loss": 1.9097,
      "step": 1543
    },
    {
      "epoch": 0.27813555505516774,
      "grad_norm": 1.5483609437942505,
      "learning_rate": 9.535454568671704e-05,
      "loss": 2.058,
      "step": 1544
    },
    {
      "epoch": 0.2783156946633641,
      "grad_norm": 1.3104422092437744,
      "learning_rate": 9.534858278957491e-05,
      "loss": 1.6419,
      "step": 1545
    },
    {
      "epoch": 0.27849583427156044,
      "grad_norm": 1.3753620386123657,
      "learning_rate": 9.534261625459156e-05,
      "loss": 1.581,
      "step": 1546
    },
    {
      "epoch": 0.2786759738797568,
      "grad_norm": 1.4918774366378784,
      "learning_rate": 9.53366460822456e-05,
      "loss": 1.8023,
      "step": 1547
    },
    {
      "epoch": 0.27885611348795314,
      "grad_norm": 1.5932422876358032,
      "learning_rate": 9.5330672273016e-05,
      "loss": 2.1818,
      "step": 1548
    },
    {
      "epoch": 0.2790362530961495,
      "grad_norm": 1.3778334856033325,
      "learning_rate": 9.532469482738193e-05,
      "loss": 1.9875,
      "step": 1549
    },
    {
      "epoch": 0.2792163927043459,
      "grad_norm": 1.2964246273040771,
      "learning_rate": 9.531871374582293e-05,
      "loss": 1.5889,
      "step": 1550
    },
    {
      "epoch": 0.27939653231254225,
      "grad_norm": 1.4452760219573975,
      "learning_rate": 9.531272902881878e-05,
      "loss": 2.2888,
      "step": 1551
    },
    {
      "epoch": 0.2795766719207386,
      "grad_norm": 1.2659543752670288,
      "learning_rate": 9.530674067684958e-05,
      "loss": 1.8517,
      "step": 1552
    },
    {
      "epoch": 0.27975681152893495,
      "grad_norm": 1.2677006721496582,
      "learning_rate": 9.530074869039572e-05,
      "loss": 1.9753,
      "step": 1553
    },
    {
      "epoch": 0.2799369511371313,
      "grad_norm": 1.2493791580200195,
      "learning_rate": 9.529475306993784e-05,
      "loss": 2.1794,
      "step": 1554
    },
    {
      "epoch": 0.28011709074532765,
      "grad_norm": 1.2145565748214722,
      "learning_rate": 9.528875381595696e-05,
      "loss": 1.9815,
      "step": 1555
    },
    {
      "epoch": 0.280297230353524,
      "grad_norm": 1.2580119371414185,
      "learning_rate": 9.528275092893428e-05,
      "loss": 1.8955,
      "step": 1556
    },
    {
      "epoch": 0.28047736996172035,
      "grad_norm": 1.4347100257873535,
      "learning_rate": 9.527674440935141e-05,
      "loss": 2.1757,
      "step": 1557
    },
    {
      "epoch": 0.2806575095699167,
      "grad_norm": 1.522442102432251,
      "learning_rate": 9.527073425769013e-05,
      "loss": 2.2382,
      "step": 1558
    },
    {
      "epoch": 0.28083764917811305,
      "grad_norm": 1.6864714622497559,
      "learning_rate": 9.52647204744326e-05,
      "loss": 2.3198,
      "step": 1559
    },
    {
      "epoch": 0.2810177887863094,
      "grad_norm": 1.8145338296890259,
      "learning_rate": 9.525870306006126e-05,
      "loss": 2.525,
      "step": 1560
    },
    {
      "epoch": 0.28119792839450575,
      "grad_norm": 2.089353561401367,
      "learning_rate": 9.52526820150588e-05,
      "loss": 2.6572,
      "step": 1561
    },
    {
      "epoch": 0.2813780680027021,
      "grad_norm": 1.4616459608078003,
      "learning_rate": 9.524665733990822e-05,
      "loss": 1.9956,
      "step": 1562
    },
    {
      "epoch": 0.28155820761089845,
      "grad_norm": 1.352818489074707,
      "learning_rate": 9.524062903509283e-05,
      "loss": 1.8813,
      "step": 1563
    },
    {
      "epoch": 0.2817383472190948,
      "grad_norm": 1.2800897359848022,
      "learning_rate": 9.523459710109622e-05,
      "loss": 1.9067,
      "step": 1564
    },
    {
      "epoch": 0.28191848682729115,
      "grad_norm": 1.2382099628448486,
      "learning_rate": 9.522856153840226e-05,
      "loss": 1.6283,
      "step": 1565
    },
    {
      "epoch": 0.2820986264354875,
      "grad_norm": 1.1803381443023682,
      "learning_rate": 9.522252234749512e-05,
      "loss": 1.6379,
      "step": 1566
    },
    {
      "epoch": 0.28227876604368385,
      "grad_norm": 1.1737301349639893,
      "learning_rate": 9.521647952885927e-05,
      "loss": 1.7575,
      "step": 1567
    },
    {
      "epoch": 0.2824589056518802,
      "grad_norm": 1.4332079887390137,
      "learning_rate": 9.521043308297945e-05,
      "loss": 1.7779,
      "step": 1568
    },
    {
      "epoch": 0.28263904526007655,
      "grad_norm": 1.3486615419387817,
      "learning_rate": 9.520438301034072e-05,
      "loss": 1.974,
      "step": 1569
    },
    {
      "epoch": 0.2828191848682729,
      "grad_norm": 1.2072014808654785,
      "learning_rate": 9.51983293114284e-05,
      "loss": 1.5324,
      "step": 1570
    },
    {
      "epoch": 0.28299932447646925,
      "grad_norm": 1.3316431045532227,
      "learning_rate": 9.519227198672811e-05,
      "loss": 1.8893,
      "step": 1571
    },
    {
      "epoch": 0.2831794640846656,
      "grad_norm": 1.3074681758880615,
      "learning_rate": 9.518621103672578e-05,
      "loss": 1.8394,
      "step": 1572
    },
    {
      "epoch": 0.28335960369286195,
      "grad_norm": 1.3992279767990112,
      "learning_rate": 9.51801464619076e-05,
      "loss": 1.9081,
      "step": 1573
    },
    {
      "epoch": 0.2835397433010583,
      "grad_norm": 1.3060131072998047,
      "learning_rate": 9.51740782627601e-05,
      "loss": 1.627,
      "step": 1574
    },
    {
      "epoch": 0.28371988290925465,
      "grad_norm": 1.4052181243896484,
      "learning_rate": 9.516800643977003e-05,
      "loss": 1.952,
      "step": 1575
    },
    {
      "epoch": 0.283900022517451,
      "grad_norm": 1.1823112964630127,
      "learning_rate": 9.516193099342448e-05,
      "loss": 1.612,
      "step": 1576
    },
    {
      "epoch": 0.28408016212564735,
      "grad_norm": 1.3419382572174072,
      "learning_rate": 9.515585192421084e-05,
      "loss": 2.0059,
      "step": 1577
    },
    {
      "epoch": 0.28426030173384376,
      "grad_norm": 1.2367485761642456,
      "learning_rate": 9.514976923261674e-05,
      "loss": 1.7249,
      "step": 1578
    },
    {
      "epoch": 0.2844404413420401,
      "grad_norm": 1.3635226488113403,
      "learning_rate": 9.514368291913013e-05,
      "loss": 1.8586,
      "step": 1579
    },
    {
      "epoch": 0.28462058095023646,
      "grad_norm": 1.2817710638046265,
      "learning_rate": 9.513759298423929e-05,
      "loss": 2.049,
      "step": 1580
    },
    {
      "epoch": 0.2848007205584328,
      "grad_norm": 1.2486647367477417,
      "learning_rate": 9.513149942843271e-05,
      "loss": 1.5934,
      "step": 1581
    },
    {
      "epoch": 0.28498086016662916,
      "grad_norm": 1.3252346515655518,
      "learning_rate": 9.512540225219925e-05,
      "loss": 2.1179,
      "step": 1582
    },
    {
      "epoch": 0.2851609997748255,
      "grad_norm": 1.2106174230575562,
      "learning_rate": 9.511930145602798e-05,
      "loss": 1.9192,
      "step": 1583
    },
    {
      "epoch": 0.28534113938302186,
      "grad_norm": 1.3011929988861084,
      "learning_rate": 9.511319704040833e-05,
      "loss": 1.8247,
      "step": 1584
    },
    {
      "epoch": 0.2855212789912182,
      "grad_norm": 1.2832356691360474,
      "learning_rate": 9.510708900582999e-05,
      "loss": 1.8141,
      "step": 1585
    },
    {
      "epoch": 0.28570141859941456,
      "grad_norm": 1.4250215291976929,
      "learning_rate": 9.510097735278294e-05,
      "loss": 1.8305,
      "step": 1586
    },
    {
      "epoch": 0.2858815582076109,
      "grad_norm": 1.4618161916732788,
      "learning_rate": 9.509486208175746e-05,
      "loss": 2.0115,
      "step": 1587
    },
    {
      "epoch": 0.28606169781580726,
      "grad_norm": 1.4392646551132202,
      "learning_rate": 9.50887431932441e-05,
      "loss": 1.8492,
      "step": 1588
    },
    {
      "epoch": 0.2862418374240036,
      "grad_norm": 1.4258909225463867,
      "learning_rate": 9.508262068773372e-05,
      "loss": 1.7784,
      "step": 1589
    },
    {
      "epoch": 0.28642197703219996,
      "grad_norm": 1.2192386388778687,
      "learning_rate": 9.507649456571748e-05,
      "loss": 1.6844,
      "step": 1590
    },
    {
      "epoch": 0.2866021166403963,
      "grad_norm": 1.3005166053771973,
      "learning_rate": 9.507036482768679e-05,
      "loss": 1.6884,
      "step": 1591
    },
    {
      "epoch": 0.28678225624859266,
      "grad_norm": 1.3351885080337524,
      "learning_rate": 9.50642314741334e-05,
      "loss": 1.6317,
      "step": 1592
    },
    {
      "epoch": 0.286962395856789,
      "grad_norm": 1.4735064506530762,
      "learning_rate": 9.505809450554928e-05,
      "loss": 1.7504,
      "step": 1593
    },
    {
      "epoch": 0.28714253546498536,
      "grad_norm": 1.5064492225646973,
      "learning_rate": 9.50519539224268e-05,
      "loss": 1.8461,
      "step": 1594
    },
    {
      "epoch": 0.2873226750731817,
      "grad_norm": 1.577451229095459,
      "learning_rate": 9.50458097252585e-05,
      "loss": 1.8405,
      "step": 1595
    },
    {
      "epoch": 0.28750281468137806,
      "grad_norm": 1.4107273817062378,
      "learning_rate": 9.503966191453731e-05,
      "loss": 1.543,
      "step": 1596
    },
    {
      "epoch": 0.2876829542895744,
      "grad_norm": 1.431510329246521,
      "learning_rate": 9.503351049075635e-05,
      "loss": 1.6636,
      "step": 1597
    },
    {
      "epoch": 0.28786309389777076,
      "grad_norm": 1.4518580436706543,
      "learning_rate": 9.502735545440912e-05,
      "loss": 1.7394,
      "step": 1598
    },
    {
      "epoch": 0.2880432335059671,
      "grad_norm": 1.3683980703353882,
      "learning_rate": 9.502119680598939e-05,
      "loss": 1.6494,
      "step": 1599
    },
    {
      "epoch": 0.28822337311416346,
      "grad_norm": 1.3465979099273682,
      "learning_rate": 9.501503454599116e-05,
      "loss": 1.7304,
      "step": 1600
    },
    {
      "epoch": 0.2884035127223598,
      "grad_norm": 2.3761870861053467,
      "learning_rate": 9.500886867490879e-05,
      "loss": 2.4285,
      "step": 1601
    },
    {
      "epoch": 0.28858365233055616,
      "grad_norm": 1.8500492572784424,
      "learning_rate": 9.50026991932369e-05,
      "loss": 2.3104,
      "step": 1602
    },
    {
      "epoch": 0.2887637919387525,
      "grad_norm": 1.3993734121322632,
      "learning_rate": 9.49965261014704e-05,
      "loss": 2.1481,
      "step": 1603
    },
    {
      "epoch": 0.28894393154694886,
      "grad_norm": 1.2521560192108154,
      "learning_rate": 9.49903494001045e-05,
      "loss": 1.9191,
      "step": 1604
    },
    {
      "epoch": 0.2891240711551452,
      "grad_norm": 1.3501262664794922,
      "learning_rate": 9.498416908963465e-05,
      "loss": 2.0639,
      "step": 1605
    },
    {
      "epoch": 0.28930421076334156,
      "grad_norm": 1.3352537155151367,
      "learning_rate": 9.49779851705567e-05,
      "loss": 2.2062,
      "step": 1606
    },
    {
      "epoch": 0.28948435037153797,
      "grad_norm": 1.6244553327560425,
      "learning_rate": 9.497179764336669e-05,
      "loss": 2.2306,
      "step": 1607
    },
    {
      "epoch": 0.2896644899797343,
      "grad_norm": 1.4581143856048584,
      "learning_rate": 9.496560650856097e-05,
      "loss": 2.0957,
      "step": 1608
    },
    {
      "epoch": 0.28984462958793067,
      "grad_norm": 1.7605836391448975,
      "learning_rate": 9.495941176663619e-05,
      "loss": 2.2289,
      "step": 1609
    },
    {
      "epoch": 0.290024769196127,
      "grad_norm": 2.2980453968048096,
      "learning_rate": 9.495321341808931e-05,
      "loss": 2.5397,
      "step": 1610
    },
    {
      "epoch": 0.29020490880432337,
      "grad_norm": 1.374963641166687,
      "learning_rate": 9.494701146341753e-05,
      "loss": 1.6998,
      "step": 1611
    },
    {
      "epoch": 0.2903850484125197,
      "grad_norm": 1.308967113494873,
      "learning_rate": 9.49408059031184e-05,
      "loss": 1.8211,
      "step": 1612
    },
    {
      "epoch": 0.29056518802071607,
      "grad_norm": 1.3360024690628052,
      "learning_rate": 9.49345967376897e-05,
      "loss": 2.0419,
      "step": 1613
    },
    {
      "epoch": 0.2907453276289124,
      "grad_norm": 1.4949694871902466,
      "learning_rate": 9.492838396762954e-05,
      "loss": 1.8544,
      "step": 1614
    },
    {
      "epoch": 0.29092546723710877,
      "grad_norm": 1.2362403869628906,
      "learning_rate": 9.49221675934363e-05,
      "loss": 1.7997,
      "step": 1615
    },
    {
      "epoch": 0.2911056068453051,
      "grad_norm": 1.343570351600647,
      "learning_rate": 9.491594761560866e-05,
      "loss": 1.8882,
      "step": 1616
    },
    {
      "epoch": 0.29128574645350147,
      "grad_norm": 1.3654160499572754,
      "learning_rate": 9.490972403464557e-05,
      "loss": 1.8538,
      "step": 1617
    },
    {
      "epoch": 0.2914658860616978,
      "grad_norm": 1.1694066524505615,
      "learning_rate": 9.490349685104632e-05,
      "loss": 1.7157,
      "step": 1618
    },
    {
      "epoch": 0.29164602566989417,
      "grad_norm": 1.2941679954528809,
      "learning_rate": 9.489726606531039e-05,
      "loss": 1.6227,
      "step": 1619
    },
    {
      "epoch": 0.2918261652780905,
      "grad_norm": 1.3347948789596558,
      "learning_rate": 9.489103167793767e-05,
      "loss": 1.6148,
      "step": 1620
    },
    {
      "epoch": 0.29200630488628687,
      "grad_norm": 1.274156928062439,
      "learning_rate": 9.488479368942825e-05,
      "loss": 1.6794,
      "step": 1621
    },
    {
      "epoch": 0.2921864444944832,
      "grad_norm": 1.1951353549957275,
      "learning_rate": 9.487855210028252e-05,
      "loss": 1.7496,
      "step": 1622
    },
    {
      "epoch": 0.29236658410267957,
      "grad_norm": 1.2419942617416382,
      "learning_rate": 9.487230691100122e-05,
      "loss": 1.7391,
      "step": 1623
    },
    {
      "epoch": 0.2925467237108759,
      "grad_norm": 1.2361663579940796,
      "learning_rate": 9.48660581220853e-05,
      "loss": 1.8085,
      "step": 1624
    },
    {
      "epoch": 0.2927268633190723,
      "grad_norm": 1.2734806537628174,
      "learning_rate": 9.485980573403607e-05,
      "loss": 1.7154,
      "step": 1625
    },
    {
      "epoch": 0.2929070029272686,
      "grad_norm": 1.2362664937973022,
      "learning_rate": 9.485354974735507e-05,
      "loss": 1.8179,
      "step": 1626
    },
    {
      "epoch": 0.293087142535465,
      "grad_norm": 1.2693657875061035,
      "learning_rate": 9.484729016254416e-05,
      "loss": 1.751,
      "step": 1627
    },
    {
      "epoch": 0.2932672821436613,
      "grad_norm": 1.3136745691299438,
      "learning_rate": 9.484102698010548e-05,
      "loss": 1.7883,
      "step": 1628
    },
    {
      "epoch": 0.2934474217518577,
      "grad_norm": 1.2337639331817627,
      "learning_rate": 9.483476020054145e-05,
      "loss": 1.6342,
      "step": 1629
    },
    {
      "epoch": 0.293627561360054,
      "grad_norm": 1.240913987159729,
      "learning_rate": 9.482848982435479e-05,
      "loss": 1.8566,
      "step": 1630
    },
    {
      "epoch": 0.2938077009682504,
      "grad_norm": 1.2837787866592407,
      "learning_rate": 9.482221585204851e-05,
      "loss": 1.8245,
      "step": 1631
    },
    {
      "epoch": 0.2939878405764467,
      "grad_norm": 1.2207119464874268,
      "learning_rate": 9.481593828412591e-05,
      "loss": 1.5952,
      "step": 1632
    },
    {
      "epoch": 0.2941679801846431,
      "grad_norm": 1.2471157312393188,
      "learning_rate": 9.480965712109058e-05,
      "loss": 1.6466,
      "step": 1633
    },
    {
      "epoch": 0.2943481197928394,
      "grad_norm": 1.2496901750564575,
      "learning_rate": 9.480337236344639e-05,
      "loss": 1.6434,
      "step": 1634
    },
    {
      "epoch": 0.29452825940103583,
      "grad_norm": 1.2381601333618164,
      "learning_rate": 9.479708401169747e-05,
      "loss": 1.6075,
      "step": 1635
    },
    {
      "epoch": 0.2947083990092322,
      "grad_norm": 1.3611130714416504,
      "learning_rate": 9.479079206634829e-05,
      "loss": 1.9765,
      "step": 1636
    },
    {
      "epoch": 0.29488853861742853,
      "grad_norm": 1.3782380819320679,
      "learning_rate": 9.47844965279036e-05,
      "loss": 1.7475,
      "step": 1637
    },
    {
      "epoch": 0.2950686782256249,
      "grad_norm": 1.4967594146728516,
      "learning_rate": 9.47781973968684e-05,
      "loss": 1.8748,
      "step": 1638
    },
    {
      "epoch": 0.29524881783382123,
      "grad_norm": 1.3652241230010986,
      "learning_rate": 9.477189467374803e-05,
      "loss": 1.8639,
      "step": 1639
    },
    {
      "epoch": 0.2954289574420176,
      "grad_norm": 1.3141024112701416,
      "learning_rate": 9.476558835904806e-05,
      "loss": 2.0563,
      "step": 1640
    },
    {
      "epoch": 0.29560909705021393,
      "grad_norm": 1.3570927381515503,
      "learning_rate": 9.475927845327441e-05,
      "loss": 1.8514,
      "step": 1641
    },
    {
      "epoch": 0.2957892366584103,
      "grad_norm": 1.3338172435760498,
      "learning_rate": 9.475296495693325e-05,
      "loss": 1.7215,
      "step": 1642
    },
    {
      "epoch": 0.29596937626660663,
      "grad_norm": 1.4011904001235962,
      "learning_rate": 9.474664787053102e-05,
      "loss": 1.8308,
      "step": 1643
    },
    {
      "epoch": 0.296149515874803,
      "grad_norm": 1.2640546560287476,
      "learning_rate": 9.47403271945745e-05,
      "loss": 1.4881,
      "step": 1644
    },
    {
      "epoch": 0.29632965548299933,
      "grad_norm": 1.3274166584014893,
      "learning_rate": 9.473400292957072e-05,
      "loss": 1.8128,
      "step": 1645
    },
    {
      "epoch": 0.2965097950911957,
      "grad_norm": 1.2972867488861084,
      "learning_rate": 9.472767507602702e-05,
      "loss": 1.6003,
      "step": 1646
    },
    {
      "epoch": 0.29668993469939203,
      "grad_norm": 1.3326082229614258,
      "learning_rate": 9.472134363445102e-05,
      "loss": 1.7415,
      "step": 1647
    },
    {
      "epoch": 0.2968700743075884,
      "grad_norm": 1.4534722566604614,
      "learning_rate": 9.471500860535061e-05,
      "loss": 1.9167,
      "step": 1648
    },
    {
      "epoch": 0.29705021391578473,
      "grad_norm": 1.3006587028503418,
      "learning_rate": 9.470866998923398e-05,
      "loss": 1.6788,
      "step": 1649
    },
    {
      "epoch": 0.2972303535239811,
      "grad_norm": 1.4854328632354736,
      "learning_rate": 9.470232778660964e-05,
      "loss": 1.8662,
      "step": 1650
    },
    {
      "epoch": 0.29741049313217743,
      "grad_norm": 1.7819095849990845,
      "learning_rate": 9.469598199798632e-05,
      "loss": 2.3337,
      "step": 1651
    },
    {
      "epoch": 0.2975906327403738,
      "grad_norm": 1.4240829944610596,
      "learning_rate": 9.468963262387309e-05,
      "loss": 2.2743,
      "step": 1652
    },
    {
      "epoch": 0.29777077234857013,
      "grad_norm": 1.9591439962387085,
      "learning_rate": 9.468327966477932e-05,
      "loss": 2.3144,
      "step": 1653
    },
    {
      "epoch": 0.2979509119567665,
      "grad_norm": 1.3398984670639038,
      "learning_rate": 9.467692312121461e-05,
      "loss": 2.1823,
      "step": 1654
    },
    {
      "epoch": 0.29813105156496283,
      "grad_norm": 1.3944469690322876,
      "learning_rate": 9.467056299368889e-05,
      "loss": 2.1413,
      "step": 1655
    },
    {
      "epoch": 0.2983111911731592,
      "grad_norm": 1.399717926979065,
      "learning_rate": 9.466419928271236e-05,
      "loss": 2.0515,
      "step": 1656
    },
    {
      "epoch": 0.29849133078135553,
      "grad_norm": 1.5255688428878784,
      "learning_rate": 9.465783198879552e-05,
      "loss": 2.4554,
      "step": 1657
    },
    {
      "epoch": 0.2986714703895519,
      "grad_norm": 1.4890297651290894,
      "learning_rate": 9.465146111244915e-05,
      "loss": 2.1461,
      "step": 1658
    },
    {
      "epoch": 0.29885160999774824,
      "grad_norm": 1.5976670980453491,
      "learning_rate": 9.464508665418433e-05,
      "loss": 2.0651,
      "step": 1659
    },
    {
      "epoch": 0.2990317496059446,
      "grad_norm": 1.8347368240356445,
      "learning_rate": 9.463870861451239e-05,
      "loss": 2.177,
      "step": 1660
    },
    {
      "epoch": 0.29921188921414094,
      "grad_norm": 1.11990487575531,
      "learning_rate": 9.4632326993945e-05,
      "loss": 1.6319,
      "step": 1661
    },
    {
      "epoch": 0.2993920288223373,
      "grad_norm": 1.2626057863235474,
      "learning_rate": 9.462594179299406e-05,
      "loss": 1.7326,
      "step": 1662
    },
    {
      "epoch": 0.29957216843053364,
      "grad_norm": 1.4078112840652466,
      "learning_rate": 9.461955301217182e-05,
      "loss": 1.948,
      "step": 1663
    },
    {
      "epoch": 0.29975230803873004,
      "grad_norm": 1.2782460451126099,
      "learning_rate": 9.461316065199077e-05,
      "loss": 1.7655,
      "step": 1664
    },
    {
      "epoch": 0.2999324476469264,
      "grad_norm": 1.3490705490112305,
      "learning_rate": 9.46067647129637e-05,
      "loss": 1.7881,
      "step": 1665
    },
    {
      "epoch": 0.30011258725512274,
      "grad_norm": 1.3608444929122925,
      "learning_rate": 9.460036519560369e-05,
      "loss": 1.8756,
      "step": 1666
    },
    {
      "epoch": 0.3002927268633191,
      "grad_norm": 1.2374787330627441,
      "learning_rate": 9.45939621004241e-05,
      "loss": 1.5984,
      "step": 1667
    },
    {
      "epoch": 0.30047286647151544,
      "grad_norm": 1.29380202293396,
      "learning_rate": 9.458755542793862e-05,
      "loss": 1.7261,
      "step": 1668
    },
    {
      "epoch": 0.3006530060797118,
      "grad_norm": 1.2572122812271118,
      "learning_rate": 9.458114517866115e-05,
      "loss": 1.698,
      "step": 1669
    },
    {
      "epoch": 0.30083314568790814,
      "grad_norm": 1.1971521377563477,
      "learning_rate": 9.45747313531059e-05,
      "loss": 1.5534,
      "step": 1670
    },
    {
      "epoch": 0.3010132852961045,
      "grad_norm": 1.2243391275405884,
      "learning_rate": 9.456831395178745e-05,
      "loss": 1.8837,
      "step": 1671
    },
    {
      "epoch": 0.30119342490430084,
      "grad_norm": 1.3460075855255127,
      "learning_rate": 9.456189297522053e-05,
      "loss": 1.7708,
      "step": 1672
    },
    {
      "epoch": 0.3013735645124972,
      "grad_norm": 1.1608455181121826,
      "learning_rate": 9.455546842392029e-05,
      "loss": 1.5252,
      "step": 1673
    },
    {
      "epoch": 0.30155370412069354,
      "grad_norm": 1.2588447332382202,
      "learning_rate": 9.454904029840206e-05,
      "loss": 1.7134,
      "step": 1674
    },
    {
      "epoch": 0.3017338437288899,
      "grad_norm": 1.3225351572036743,
      "learning_rate": 9.454260859918153e-05,
      "loss": 1.8432,
      "step": 1675
    },
    {
      "epoch": 0.30191398333708624,
      "grad_norm": 1.3037313222885132,
      "learning_rate": 9.453617332677461e-05,
      "loss": 1.9354,
      "step": 1676
    },
    {
      "epoch": 0.3020941229452826,
      "grad_norm": 1.3181267976760864,
      "learning_rate": 9.452973448169758e-05,
      "loss": 2.0047,
      "step": 1677
    },
    {
      "epoch": 0.30227426255347895,
      "grad_norm": 1.3081953525543213,
      "learning_rate": 9.452329206446693e-05,
      "loss": 1.9375,
      "step": 1678
    },
    {
      "epoch": 0.3024544021616753,
      "grad_norm": 1.3248902559280396,
      "learning_rate": 9.451684607559947e-05,
      "loss": 1.7659,
      "step": 1679
    },
    {
      "epoch": 0.30263454176987165,
      "grad_norm": 1.4065247774124146,
      "learning_rate": 9.451039651561231e-05,
      "loss": 2.1117,
      "step": 1680
    },
    {
      "epoch": 0.302814681378068,
      "grad_norm": 1.2488824129104614,
      "learning_rate": 9.450394338502284e-05,
      "loss": 1.9707,
      "step": 1681
    },
    {
      "epoch": 0.30299482098626435,
      "grad_norm": 1.3439654111862183,
      "learning_rate": 9.449748668434868e-05,
      "loss": 1.976,
      "step": 1682
    },
    {
      "epoch": 0.3031749605944607,
      "grad_norm": 1.2611677646636963,
      "learning_rate": 9.449102641410783e-05,
      "loss": 1.8581,
      "step": 1683
    },
    {
      "epoch": 0.30335510020265705,
      "grad_norm": 1.3079493045806885,
      "learning_rate": 9.44845625748185e-05,
      "loss": 1.8,
      "step": 1684
    },
    {
      "epoch": 0.3035352398108534,
      "grad_norm": 1.2610013484954834,
      "learning_rate": 9.447809516699924e-05,
      "loss": 1.9368,
      "step": 1685
    },
    {
      "epoch": 0.30371537941904975,
      "grad_norm": 1.4763059616088867,
      "learning_rate": 9.447162419116884e-05,
      "loss": 2.096,
      "step": 1686
    },
    {
      "epoch": 0.3038955190272461,
      "grad_norm": 1.380873680114746,
      "learning_rate": 9.446514964784641e-05,
      "loss": 1.8453,
      "step": 1687
    },
    {
      "epoch": 0.30407565863544245,
      "grad_norm": 1.2984628677368164,
      "learning_rate": 9.445867153755132e-05,
      "loss": 1.7129,
      "step": 1688
    },
    {
      "epoch": 0.3042557982436388,
      "grad_norm": 1.2476941347122192,
      "learning_rate": 9.445218986080329e-05,
      "loss": 1.7545,
      "step": 1689
    },
    {
      "epoch": 0.30443593785183515,
      "grad_norm": 1.246397614479065,
      "learning_rate": 9.444570461812221e-05,
      "loss": 1.5413,
      "step": 1690
    },
    {
      "epoch": 0.3046160774600315,
      "grad_norm": 1.2983617782592773,
      "learning_rate": 9.443921581002837e-05,
      "loss": 1.8814,
      "step": 1691
    },
    {
      "epoch": 0.3047962170682279,
      "grad_norm": 1.3920047283172607,
      "learning_rate": 9.443272343704227e-05,
      "loss": 1.768,
      "step": 1692
    },
    {
      "epoch": 0.30497635667642425,
      "grad_norm": 1.2787226438522339,
      "learning_rate": 9.442622749968475e-05,
      "loss": 1.6886,
      "step": 1693
    },
    {
      "epoch": 0.3051564962846206,
      "grad_norm": 1.3687013387680054,
      "learning_rate": 9.44197279984769e-05,
      "loss": 1.8526,
      "step": 1694
    },
    {
      "epoch": 0.30533663589281695,
      "grad_norm": 1.2241677045822144,
      "learning_rate": 9.441322493394009e-05,
      "loss": 1.5742,
      "step": 1695
    },
    {
      "epoch": 0.3055167755010133,
      "grad_norm": 1.394643783569336,
      "learning_rate": 9.440671830659602e-05,
      "loss": 1.8381,
      "step": 1696
    },
    {
      "epoch": 0.30569691510920965,
      "grad_norm": 1.514260172843933,
      "learning_rate": 9.440020811696664e-05,
      "loss": 1.9431,
      "step": 1697
    },
    {
      "epoch": 0.305877054717406,
      "grad_norm": 1.326114296913147,
      "learning_rate": 9.43936943655742e-05,
      "loss": 1.7265,
      "step": 1698
    },
    {
      "epoch": 0.30605719432560236,
      "grad_norm": 1.2670493125915527,
      "learning_rate": 9.43871770529412e-05,
      "loss": 1.4803,
      "step": 1699
    },
    {
      "epoch": 0.3062373339337987,
      "grad_norm": 1.274014949798584,
      "learning_rate": 9.438065617959048e-05,
      "loss": 1.5406,
      "step": 1700
    },
    {
      "epoch": 0.30641747354199506,
      "grad_norm": 1.5541276931762695,
      "learning_rate": 9.437413174604515e-05,
      "loss": 2.3882,
      "step": 1701
    },
    {
      "epoch": 0.3065976131501914,
      "grad_norm": 1.4880536794662476,
      "learning_rate": 9.436760375282859e-05,
      "loss": 2.1364,
      "step": 1702
    },
    {
      "epoch": 0.30677775275838776,
      "grad_norm": 1.2172539234161377,
      "learning_rate": 9.436107220046445e-05,
      "loss": 1.9677,
      "step": 1703
    },
    {
      "epoch": 0.3069578923665841,
      "grad_norm": 1.3161697387695312,
      "learning_rate": 9.435453708947672e-05,
      "loss": 2.2429,
      "step": 1704
    },
    {
      "epoch": 0.30713803197478046,
      "grad_norm": 1.3544458150863647,
      "learning_rate": 9.434799842038961e-05,
      "loss": 2.1697,
      "step": 1705
    },
    {
      "epoch": 0.3073181715829768,
      "grad_norm": 1.2765963077545166,
      "learning_rate": 9.43414561937277e-05,
      "loss": 2.1297,
      "step": 1706
    },
    {
      "epoch": 0.30749831119117316,
      "grad_norm": 1.4520622491836548,
      "learning_rate": 9.433491041001575e-05,
      "loss": 2.2196,
      "step": 1707
    },
    {
      "epoch": 0.3076784507993695,
      "grad_norm": 1.6171400547027588,
      "learning_rate": 9.432836106977888e-05,
      "loss": 2.1238,
      "step": 1708
    },
    {
      "epoch": 0.30785859040756586,
      "grad_norm": 1.802198886871338,
      "learning_rate": 9.432180817354249e-05,
      "loss": 2.0303,
      "step": 1709
    },
    {
      "epoch": 0.3080387300157622,
      "grad_norm": 1.7998108863830566,
      "learning_rate": 9.431525172183222e-05,
      "loss": 1.9403,
      "step": 1710
    },
    {
      "epoch": 0.30821886962395856,
      "grad_norm": 1.5913798809051514,
      "learning_rate": 9.430869171517405e-05,
      "loss": 2.0044,
      "step": 1711
    },
    {
      "epoch": 0.3083990092321549,
      "grad_norm": 1.3948698043823242,
      "learning_rate": 9.430212815409421e-05,
      "loss": 1.7397,
      "step": 1712
    },
    {
      "epoch": 0.30857914884035126,
      "grad_norm": 1.3813555240631104,
      "learning_rate": 9.429556103911923e-05,
      "loss": 1.7361,
      "step": 1713
    },
    {
      "epoch": 0.3087592884485476,
      "grad_norm": 1.2981288433074951,
      "learning_rate": 9.428899037077592e-05,
      "loss": 1.6294,
      "step": 1714
    },
    {
      "epoch": 0.30893942805674396,
      "grad_norm": 1.3228983879089355,
      "learning_rate": 9.428241614959137e-05,
      "loss": 1.8087,
      "step": 1715
    },
    {
      "epoch": 0.3091195676649403,
      "grad_norm": 1.376579761505127,
      "learning_rate": 9.427583837609297e-05,
      "loss": 1.8035,
      "step": 1716
    },
    {
      "epoch": 0.30929970727313666,
      "grad_norm": 1.0601602792739868,
      "learning_rate": 9.426925705080837e-05,
      "loss": 1.4258,
      "step": 1717
    },
    {
      "epoch": 0.309479846881333,
      "grad_norm": 1.2452287673950195,
      "learning_rate": 9.426267217426556e-05,
      "loss": 1.6219,
      "step": 1718
    },
    {
      "epoch": 0.30965998648952936,
      "grad_norm": 1.1422358751296997,
      "learning_rate": 9.425608374699273e-05,
      "loss": 1.6694,
      "step": 1719
    },
    {
      "epoch": 0.3098401260977257,
      "grad_norm": 1.3704848289489746,
      "learning_rate": 9.424949176951843e-05,
      "loss": 1.9035,
      "step": 1720
    },
    {
      "epoch": 0.3100202657059221,
      "grad_norm": 1.2918815612792969,
      "learning_rate": 9.424289624237144e-05,
      "loss": 1.8343,
      "step": 1721
    },
    {
      "epoch": 0.31020040531411847,
      "grad_norm": 1.3253055810928345,
      "learning_rate": 9.423629716608087e-05,
      "loss": 1.853,
      "step": 1722
    },
    {
      "epoch": 0.3103805449223148,
      "grad_norm": 1.381380319595337,
      "learning_rate": 9.422969454117607e-05,
      "loss": 2.0804,
      "step": 1723
    },
    {
      "epoch": 0.31056068453051117,
      "grad_norm": 1.270503282546997,
      "learning_rate": 9.422308836818675e-05,
      "loss": 1.6293,
      "step": 1724
    },
    {
      "epoch": 0.3107408241387075,
      "grad_norm": 1.3035050630569458,
      "learning_rate": 9.42164786476428e-05,
      "loss": 1.8483,
      "step": 1725
    },
    {
      "epoch": 0.31092096374690387,
      "grad_norm": 1.42676842212677,
      "learning_rate": 9.420986538007446e-05,
      "loss": 1.8387,
      "step": 1726
    },
    {
      "epoch": 0.3111011033551002,
      "grad_norm": 1.2072376012802124,
      "learning_rate": 9.420324856601227e-05,
      "loss": 1.9353,
      "step": 1727
    },
    {
      "epoch": 0.31128124296329657,
      "grad_norm": 1.205321192741394,
      "learning_rate": 9.4196628205987e-05,
      "loss": 1.7275,
      "step": 1728
    },
    {
      "epoch": 0.3114613825714929,
      "grad_norm": 1.257953405380249,
      "learning_rate": 9.419000430052976e-05,
      "loss": 1.7968,
      "step": 1729
    },
    {
      "epoch": 0.31164152217968927,
      "grad_norm": 1.274452805519104,
      "learning_rate": 9.41833768501719e-05,
      "loss": 1.8045,
      "step": 1730
    },
    {
      "epoch": 0.3118216617878856,
      "grad_norm": 1.2059699296951294,
      "learning_rate": 9.417674585544505e-05,
      "loss": 1.8465,
      "step": 1731
    },
    {
      "epoch": 0.31200180139608197,
      "grad_norm": 1.16840660572052,
      "learning_rate": 9.417011131688117e-05,
      "loss": 1.4997,
      "step": 1732
    },
    {
      "epoch": 0.3121819410042783,
      "grad_norm": 1.2271968126296997,
      "learning_rate": 9.416347323501248e-05,
      "loss": 1.7907,
      "step": 1733
    },
    {
      "epoch": 0.31236208061247467,
      "grad_norm": 1.349408745765686,
      "learning_rate": 9.415683161037148e-05,
      "loss": 1.7231,
      "step": 1734
    },
    {
      "epoch": 0.312542220220671,
      "grad_norm": 1.4745874404907227,
      "learning_rate": 9.415018644349095e-05,
      "loss": 2.1468,
      "step": 1735
    },
    {
      "epoch": 0.31272235982886737,
      "grad_norm": 1.3756881952285767,
      "learning_rate": 9.414353773490398e-05,
      "loss": 1.7441,
      "step": 1736
    },
    {
      "epoch": 0.3129024994370637,
      "grad_norm": 1.3152180910110474,
      "learning_rate": 9.41368854851439e-05,
      "loss": 1.6037,
      "step": 1737
    },
    {
      "epoch": 0.31308263904526007,
      "grad_norm": 1.4125536680221558,
      "learning_rate": 9.413022969474437e-05,
      "loss": 1.8707,
      "step": 1738
    },
    {
      "epoch": 0.3132627786534564,
      "grad_norm": 1.378162145614624,
      "learning_rate": 9.412357036423931e-05,
      "loss": 1.8994,
      "step": 1739
    },
    {
      "epoch": 0.31344291826165277,
      "grad_norm": 1.3343230485916138,
      "learning_rate": 9.411690749416294e-05,
      "loss": 1.8816,
      "step": 1740
    },
    {
      "epoch": 0.3136230578698491,
      "grad_norm": 1.3739832639694214,
      "learning_rate": 9.411024108504972e-05,
      "loss": 1.7611,
      "step": 1741
    },
    {
      "epoch": 0.31380319747804547,
      "grad_norm": 1.2668706178665161,
      "learning_rate": 9.410357113743444e-05,
      "loss": 1.7515,
      "step": 1742
    },
    {
      "epoch": 0.3139833370862418,
      "grad_norm": 1.3751226663589478,
      "learning_rate": 9.409689765185218e-05,
      "loss": 1.7097,
      "step": 1743
    },
    {
      "epoch": 0.31416347669443817,
      "grad_norm": 1.5417399406433105,
      "learning_rate": 9.409022062883828e-05,
      "loss": 1.9759,
      "step": 1744
    },
    {
      "epoch": 0.3143436163026345,
      "grad_norm": 1.4709678888320923,
      "learning_rate": 9.408354006892833e-05,
      "loss": 1.9505,
      "step": 1745
    },
    {
      "epoch": 0.31452375591083087,
      "grad_norm": 1.3311420679092407,
      "learning_rate": 9.407685597265828e-05,
      "loss": 1.6615,
      "step": 1746
    },
    {
      "epoch": 0.3147038955190272,
      "grad_norm": 1.3597052097320557,
      "learning_rate": 9.407016834056432e-05,
      "loss": 1.6673,
      "step": 1747
    },
    {
      "epoch": 0.31488403512722357,
      "grad_norm": 1.3709936141967773,
      "learning_rate": 9.406347717318291e-05,
      "loss": 1.5944,
      "step": 1748
    },
    {
      "epoch": 0.31506417473542,
      "grad_norm": 1.5791631937026978,
      "learning_rate": 9.405678247105083e-05,
      "loss": 1.9664,
      "step": 1749
    },
    {
      "epoch": 0.3152443143436163,
      "grad_norm": 1.2927601337432861,
      "learning_rate": 9.405008423470513e-05,
      "loss": 1.6538,
      "step": 1750
    },
    {
      "epoch": 0.3154244539518127,
      "grad_norm": 1.4359164237976074,
      "learning_rate": 9.404338246468311e-05,
      "loss": 2.0641,
      "step": 1751
    },
    {
      "epoch": 0.31560459356000903,
      "grad_norm": 1.5348880290985107,
      "learning_rate": 9.403667716152242e-05,
      "loss": 2.4473,
      "step": 1752
    },
    {
      "epoch": 0.3157847331682054,
      "grad_norm": 1.9629950523376465,
      "learning_rate": 9.402996832576091e-05,
      "loss": 2.0529,
      "step": 1753
    },
    {
      "epoch": 0.31596487277640173,
      "grad_norm": 1.2311313152313232,
      "learning_rate": 9.402325595793682e-05,
      "loss": 1.8946,
      "step": 1754
    },
    {
      "epoch": 0.3161450123845981,
      "grad_norm": 1.2952783107757568,
      "learning_rate": 9.401654005858855e-05,
      "loss": 1.957,
      "step": 1755
    },
    {
      "epoch": 0.31632515199279443,
      "grad_norm": 1.3070710897445679,
      "learning_rate": 9.40098206282549e-05,
      "loss": 2.223,
      "step": 1756
    },
    {
      "epoch": 0.3165052916009908,
      "grad_norm": 1.441473126411438,
      "learning_rate": 9.400309766747486e-05,
      "loss": 2.2275,
      "step": 1757
    },
    {
      "epoch": 0.31668543120918713,
      "grad_norm": 1.5374830961227417,
      "learning_rate": 9.399637117678776e-05,
      "loss": 2.2977,
      "step": 1758
    },
    {
      "epoch": 0.3168655708173835,
      "grad_norm": 1.617869257926941,
      "learning_rate": 9.398964115673321e-05,
      "loss": 2.3844,
      "step": 1759
    },
    {
      "epoch": 0.31704571042557983,
      "grad_norm": 1.7544715404510498,
      "learning_rate": 9.398290760785105e-05,
      "loss": 2.1444,
      "step": 1760
    },
    {
      "epoch": 0.3172258500337762,
      "grad_norm": 1.6874605417251587,
      "learning_rate": 9.397617053068149e-05,
      "loss": 2.0249,
      "step": 1761
    },
    {
      "epoch": 0.31740598964197253,
      "grad_norm": 1.2498189210891724,
      "learning_rate": 9.396942992576493e-05,
      "loss": 1.7133,
      "step": 1762
    },
    {
      "epoch": 0.3175861292501689,
      "grad_norm": 1.309888243675232,
      "learning_rate": 9.396268579364212e-05,
      "loss": 1.7252,
      "step": 1763
    },
    {
      "epoch": 0.31776626885836523,
      "grad_norm": 1.2649097442626953,
      "learning_rate": 9.395593813485408e-05,
      "loss": 1.5637,
      "step": 1764
    },
    {
      "epoch": 0.3179464084665616,
      "grad_norm": 1.283597707748413,
      "learning_rate": 9.394918694994208e-05,
      "loss": 1.5893,
      "step": 1765
    },
    {
      "epoch": 0.31812654807475793,
      "grad_norm": 1.2372602224349976,
      "learning_rate": 9.394243223944771e-05,
      "loss": 1.7177,
      "step": 1766
    },
    {
      "epoch": 0.3183066876829543,
      "grad_norm": 1.2657066583633423,
      "learning_rate": 9.393567400391285e-05,
      "loss": 1.6995,
      "step": 1767
    },
    {
      "epoch": 0.31848682729115063,
      "grad_norm": 1.3661051988601685,
      "learning_rate": 9.392891224387961e-05,
      "loss": 1.678,
      "step": 1768
    },
    {
      "epoch": 0.318666966899347,
      "grad_norm": 1.1339210271835327,
      "learning_rate": 9.392214695989041e-05,
      "loss": 1.6526,
      "step": 1769
    },
    {
      "epoch": 0.31884710650754333,
      "grad_norm": 1.2592577934265137,
      "learning_rate": 9.391537815248799e-05,
      "loss": 1.618,
      "step": 1770
    },
    {
      "epoch": 0.3190272461157397,
      "grad_norm": 1.4403939247131348,
      "learning_rate": 9.390860582221532e-05,
      "loss": 1.7451,
      "step": 1771
    },
    {
      "epoch": 0.31920738572393603,
      "grad_norm": 1.1700098514556885,
      "learning_rate": 9.390182996961569e-05,
      "loss": 1.5627,
      "step": 1772
    },
    {
      "epoch": 0.3193875253321324,
      "grad_norm": 1.1356571912765503,
      "learning_rate": 9.389505059523263e-05,
      "loss": 1.5484,
      "step": 1773
    },
    {
      "epoch": 0.31956766494032873,
      "grad_norm": 1.2463253736495972,
      "learning_rate": 9.388826769961e-05,
      "loss": 1.6142,
      "step": 1774
    },
    {
      "epoch": 0.3197478045485251,
      "grad_norm": 1.32034170627594,
      "learning_rate": 9.388148128329191e-05,
      "loss": 1.6295,
      "step": 1775
    },
    {
      "epoch": 0.31992794415672143,
      "grad_norm": 1.277449131011963,
      "learning_rate": 9.387469134682277e-05,
      "loss": 1.7325,
      "step": 1776
    },
    {
      "epoch": 0.32010808376491784,
      "grad_norm": 1.25298273563385,
      "learning_rate": 9.386789789074726e-05,
      "loss": 1.6039,
      "step": 1777
    },
    {
      "epoch": 0.3202882233731142,
      "grad_norm": 1.286400318145752,
      "learning_rate": 9.386110091561033e-05,
      "loss": 1.8384,
      "step": 1778
    },
    {
      "epoch": 0.32046836298131054,
      "grad_norm": 1.3443750143051147,
      "learning_rate": 9.385430042195726e-05,
      "loss": 1.8617,
      "step": 1779
    },
    {
      "epoch": 0.3206485025895069,
      "grad_norm": 1.327980637550354,
      "learning_rate": 9.384749641033359e-05,
      "loss": 2.0242,
      "step": 1780
    },
    {
      "epoch": 0.32082864219770324,
      "grad_norm": 1.197792887687683,
      "learning_rate": 9.384068888128509e-05,
      "loss": 1.6825,
      "step": 1781
    },
    {
      "epoch": 0.3210087818058996,
      "grad_norm": 1.3268731832504272,
      "learning_rate": 9.383387783535789e-05,
      "loss": 1.8954,
      "step": 1782
    },
    {
      "epoch": 0.32118892141409594,
      "grad_norm": 1.2550289630889893,
      "learning_rate": 9.382706327309836e-05,
      "loss": 1.8276,
      "step": 1783
    },
    {
      "epoch": 0.3213690610222923,
      "grad_norm": 1.3409785032272339,
      "learning_rate": 9.382024519505316e-05,
      "loss": 1.8141,
      "step": 1784
    },
    {
      "epoch": 0.32154920063048864,
      "grad_norm": 1.304237961769104,
      "learning_rate": 9.381342360176924e-05,
      "loss": 1.5877,
      "step": 1785
    },
    {
      "epoch": 0.321729340238685,
      "grad_norm": 1.364743709564209,
      "learning_rate": 9.380659849379382e-05,
      "loss": 1.7072,
      "step": 1786
    },
    {
      "epoch": 0.32190947984688134,
      "grad_norm": 1.255950927734375,
      "learning_rate": 9.37997698716744e-05,
      "loss": 1.7769,
      "step": 1787
    },
    {
      "epoch": 0.3220896194550777,
      "grad_norm": 1.3073413372039795,
      "learning_rate": 9.379293773595879e-05,
      "loss": 1.7811,
      "step": 1788
    },
    {
      "epoch": 0.32226975906327404,
      "grad_norm": 1.4029983282089233,
      "learning_rate": 9.378610208719502e-05,
      "loss": 1.8239,
      "step": 1789
    },
    {
      "epoch": 0.3224498986714704,
      "grad_norm": 1.5238877534866333,
      "learning_rate": 9.377926292593149e-05,
      "loss": 1.854,
      "step": 1790
    },
    {
      "epoch": 0.32263003827966674,
      "grad_norm": 1.1692688465118408,
      "learning_rate": 9.377242025271679e-05,
      "loss": 1.4625,
      "step": 1791
    },
    {
      "epoch": 0.3228101778878631,
      "grad_norm": 1.5369685888290405,
      "learning_rate": 9.376557406809988e-05,
      "loss": 2.0254,
      "step": 1792
    },
    {
      "epoch": 0.32299031749605944,
      "grad_norm": 1.4477719068527222,
      "learning_rate": 9.375872437262993e-05,
      "loss": 2.0603,
      "step": 1793
    },
    {
      "epoch": 0.3231704571042558,
      "grad_norm": 1.3892316818237305,
      "learning_rate": 9.375187116685642e-05,
      "loss": 1.8796,
      "step": 1794
    },
    {
      "epoch": 0.32335059671245214,
      "grad_norm": 1.3259357213974,
      "learning_rate": 9.374501445132912e-05,
      "loss": 1.5953,
      "step": 1795
    },
    {
      "epoch": 0.3235307363206485,
      "grad_norm": 1.4816466569900513,
      "learning_rate": 9.373815422659806e-05,
      "loss": 2.0084,
      "step": 1796
    },
    {
      "epoch": 0.32371087592884484,
      "grad_norm": 1.2615822553634644,
      "learning_rate": 9.373129049321359e-05,
      "loss": 1.4401,
      "step": 1797
    },
    {
      "epoch": 0.3238910155370412,
      "grad_norm": 1.6288646459579468,
      "learning_rate": 9.372442325172628e-05,
      "loss": 1.8927,
      "step": 1798
    },
    {
      "epoch": 0.32407115514523754,
      "grad_norm": 1.5668270587921143,
      "learning_rate": 9.371755250268704e-05,
      "loss": 1.6474,
      "step": 1799
    },
    {
      "epoch": 0.3242512947534339,
      "grad_norm": 1.380558729171753,
      "learning_rate": 9.371067824664703e-05,
      "loss": 1.4893,
      "step": 1800
    },
    {
      "epoch": 0.32443143436163024,
      "grad_norm": 1.5281416177749634,
      "learning_rate": 9.370380048415771e-05,
      "loss": 2.2178,
      "step": 1801
    },
    {
      "epoch": 0.3246115739698266,
      "grad_norm": 1.3303380012512207,
      "learning_rate": 9.36969192157708e-05,
      "loss": 2.0225,
      "step": 1802
    },
    {
      "epoch": 0.32479171357802294,
      "grad_norm": 1.4043996334075928,
      "learning_rate": 9.369003444203831e-05,
      "loss": 2.2893,
      "step": 1803
    },
    {
      "epoch": 0.3249718531862193,
      "grad_norm": 1.7341221570968628,
      "learning_rate": 9.368314616351253e-05,
      "loss": 2.1961,
      "step": 1804
    },
    {
      "epoch": 0.32515199279441565,
      "grad_norm": 1.37443208694458,
      "learning_rate": 9.367625438074606e-05,
      "loss": 1.9917,
      "step": 1805
    },
    {
      "epoch": 0.32533213240261205,
      "grad_norm": 1.9039502143859863,
      "learning_rate": 9.366935909429174e-05,
      "loss": 2.2042,
      "step": 1806
    },
    {
      "epoch": 0.3255122720108084,
      "grad_norm": 1.275850534439087,
      "learning_rate": 9.366246030470268e-05,
      "loss": 1.9785,
      "step": 1807
    },
    {
      "epoch": 0.32569241161900475,
      "grad_norm": 1.369605302810669,
      "learning_rate": 9.365555801253234e-05,
      "loss": 1.8925,
      "step": 1808
    },
    {
      "epoch": 0.3258725512272011,
      "grad_norm": 1.7239036560058594,
      "learning_rate": 9.364865221833439e-05,
      "loss": 2.5742,
      "step": 1809
    },
    {
      "epoch": 0.32605269083539745,
      "grad_norm": 2.1848268508911133,
      "learning_rate": 9.364174292266284e-05,
      "loss": 2.4709,
      "step": 1810
    },
    {
      "epoch": 0.3262328304435938,
      "grad_norm": 1.3390096426010132,
      "learning_rate": 9.36348301260719e-05,
      "loss": 2.0305,
      "step": 1811
    },
    {
      "epoch": 0.32641297005179015,
      "grad_norm": 1.4194066524505615,
      "learning_rate": 9.362791382911615e-05,
      "loss": 1.9321,
      "step": 1812
    },
    {
      "epoch": 0.3265931096599865,
      "grad_norm": 1.1762449741363525,
      "learning_rate": 9.362099403235041e-05,
      "loss": 1.7828,
      "step": 1813
    },
    {
      "epoch": 0.32677324926818285,
      "grad_norm": 1.2199952602386475,
      "learning_rate": 9.361407073632977e-05,
      "loss": 1.4752,
      "step": 1814
    },
    {
      "epoch": 0.3269533888763792,
      "grad_norm": 1.3213188648223877,
      "learning_rate": 9.360714394160964e-05,
      "loss": 1.8439,
      "step": 1815
    },
    {
      "epoch": 0.32713352848457555,
      "grad_norm": 1.196629524230957,
      "learning_rate": 9.360021364874563e-05,
      "loss": 1.4599,
      "step": 1816
    },
    {
      "epoch": 0.3273136680927719,
      "grad_norm": 1.333306074142456,
      "learning_rate": 9.359327985829374e-05,
      "loss": 1.82,
      "step": 1817
    },
    {
      "epoch": 0.32749380770096825,
      "grad_norm": 1.2444415092468262,
      "learning_rate": 9.358634257081016e-05,
      "loss": 1.8003,
      "step": 1818
    },
    {
      "epoch": 0.3276739473091646,
      "grad_norm": 1.317900538444519,
      "learning_rate": 9.357940178685142e-05,
      "loss": 1.5071,
      "step": 1819
    },
    {
      "epoch": 0.32785408691736095,
      "grad_norm": 1.3687241077423096,
      "learning_rate": 9.35724575069743e-05,
      "loss": 1.8539,
      "step": 1820
    },
    {
      "epoch": 0.3280342265255573,
      "grad_norm": 1.2784526348114014,
      "learning_rate": 9.356550973173585e-05,
      "loss": 1.7341,
      "step": 1821
    },
    {
      "epoch": 0.32821436613375365,
      "grad_norm": 1.607740044593811,
      "learning_rate": 9.355855846169344e-05,
      "loss": 1.7791,
      "step": 1822
    },
    {
      "epoch": 0.32839450574195,
      "grad_norm": 1.2505072355270386,
      "learning_rate": 9.355160369740467e-05,
      "loss": 1.5663,
      "step": 1823
    },
    {
      "epoch": 0.32857464535014635,
      "grad_norm": 1.281872272491455,
      "learning_rate": 9.354464543942747e-05,
      "loss": 1.7267,
      "step": 1824
    },
    {
      "epoch": 0.3287547849583427,
      "grad_norm": 1.2124817371368408,
      "learning_rate": 9.353768368832004e-05,
      "loss": 1.4913,
      "step": 1825
    },
    {
      "epoch": 0.32893492456653906,
      "grad_norm": 1.2875710725784302,
      "learning_rate": 9.353071844464081e-05,
      "loss": 1.7825,
      "step": 1826
    },
    {
      "epoch": 0.3291150641747354,
      "grad_norm": 1.3107181787490845,
      "learning_rate": 9.352374970894856e-05,
      "loss": 1.62,
      "step": 1827
    },
    {
      "epoch": 0.32929520378293176,
      "grad_norm": 1.2903443574905396,
      "learning_rate": 9.35167774818023e-05,
      "loss": 1.8159,
      "step": 1828
    },
    {
      "epoch": 0.3294753433911281,
      "grad_norm": 1.3100007772445679,
      "learning_rate": 9.350980176376137e-05,
      "loss": 1.5669,
      "step": 1829
    },
    {
      "epoch": 0.32965548299932446,
      "grad_norm": 1.2323390245437622,
      "learning_rate": 9.350282255538531e-05,
      "loss": 1.7937,
      "step": 1830
    },
    {
      "epoch": 0.3298356226075208,
      "grad_norm": 1.3646736145019531,
      "learning_rate": 9.349583985723402e-05,
      "loss": 2.0431,
      "step": 1831
    },
    {
      "epoch": 0.33001576221571716,
      "grad_norm": 1.346835970878601,
      "learning_rate": 9.348885366986765e-05,
      "loss": 1.8195,
      "step": 1832
    },
    {
      "epoch": 0.3301959018239135,
      "grad_norm": 1.4719043970108032,
      "learning_rate": 9.348186399384662e-05,
      "loss": 1.7148,
      "step": 1833
    },
    {
      "epoch": 0.3303760414321099,
      "grad_norm": 1.3289631605148315,
      "learning_rate": 9.347487082973163e-05,
      "loss": 1.8976,
      "step": 1834
    },
    {
      "epoch": 0.33055618104030626,
      "grad_norm": 1.4127905368804932,
      "learning_rate": 9.346787417808369e-05,
      "loss": 1.9273,
      "step": 1835
    },
    {
      "epoch": 0.3307363206485026,
      "grad_norm": 1.272700548171997,
      "learning_rate": 9.346087403946404e-05,
      "loss": 1.8676,
      "step": 1836
    },
    {
      "epoch": 0.33091646025669896,
      "grad_norm": 1.1171737909317017,
      "learning_rate": 9.345387041443426e-05,
      "loss": 1.5226,
      "step": 1837
    },
    {
      "epoch": 0.3310965998648953,
      "grad_norm": 1.3365083932876587,
      "learning_rate": 9.344686330355614e-05,
      "loss": 1.8689,
      "step": 1838
    },
    {
      "epoch": 0.33127673947309166,
      "grad_norm": 1.411765217781067,
      "learning_rate": 9.343985270739182e-05,
      "loss": 1.8712,
      "step": 1839
    },
    {
      "epoch": 0.331456879081288,
      "grad_norm": 1.3648669719696045,
      "learning_rate": 9.34328386265037e-05,
      "loss": 1.9321,
      "step": 1840
    },
    {
      "epoch": 0.33163701868948436,
      "grad_norm": 1.2787798643112183,
      "learning_rate": 9.342582106145439e-05,
      "loss": 1.5218,
      "step": 1841
    },
    {
      "epoch": 0.3318171582976807,
      "grad_norm": 1.4488646984100342,
      "learning_rate": 9.341880001280687e-05,
      "loss": 1.9806,
      "step": 1842
    },
    {
      "epoch": 0.33199729790587706,
      "grad_norm": 1.433493971824646,
      "learning_rate": 9.341177548112436e-05,
      "loss": 1.9,
      "step": 1843
    },
    {
      "epoch": 0.3321774375140734,
      "grad_norm": 1.4048950672149658,
      "learning_rate": 9.340474746697036e-05,
      "loss": 1.6446,
      "step": 1844
    },
    {
      "epoch": 0.33235757712226977,
      "grad_norm": 1.245510220527649,
      "learning_rate": 9.339771597090869e-05,
      "loss": 1.6488,
      "step": 1845
    },
    {
      "epoch": 0.3325377167304661,
      "grad_norm": 1.2737582921981812,
      "learning_rate": 9.339068099350337e-05,
      "loss": 1.4383,
      "step": 1846
    },
    {
      "epoch": 0.33271785633866247,
      "grad_norm": 1.3664889335632324,
      "learning_rate": 9.338364253531874e-05,
      "loss": 1.8731,
      "step": 1847
    },
    {
      "epoch": 0.3328979959468588,
      "grad_norm": 1.336047887802124,
      "learning_rate": 9.337660059691946e-05,
      "loss": 1.63,
      "step": 1848
    },
    {
      "epoch": 0.33307813555505517,
      "grad_norm": 1.3905073404312134,
      "learning_rate": 9.336955517887039e-05,
      "loss": 1.7776,
      "step": 1849
    },
    {
      "epoch": 0.3332582751632515,
      "grad_norm": 1.3520543575286865,
      "learning_rate": 9.336250628173673e-05,
      "loss": 1.5627,
      "step": 1850
    },
    {
      "epoch": 0.33343841477144787,
      "grad_norm": 1.3104815483093262,
      "learning_rate": 9.335545390608393e-05,
      "loss": 2.0544,
      "step": 1851
    },
    {
      "epoch": 0.3336185543796442,
      "grad_norm": 1.2332983016967773,
      "learning_rate": 9.334839805247775e-05,
      "loss": 2.0492,
      "step": 1852
    },
    {
      "epoch": 0.33379869398784057,
      "grad_norm": 1.460875153541565,
      "learning_rate": 9.334133872148418e-05,
      "loss": 2.1039,
      "step": 1853
    },
    {
      "epoch": 0.3339788335960369,
      "grad_norm": 1.4161427021026611,
      "learning_rate": 9.333427591366954e-05,
      "loss": 2.0569,
      "step": 1854
    },
    {
      "epoch": 0.33415897320423327,
      "grad_norm": 1.276246428489685,
      "learning_rate": 9.33272096296004e-05,
      "loss": 2.0297,
      "step": 1855
    },
    {
      "epoch": 0.3343391128124296,
      "grad_norm": 1.3764469623565674,
      "learning_rate": 9.33201398698436e-05,
      "loss": 2.4107,
      "step": 1856
    },
    {
      "epoch": 0.33451925242062597,
      "grad_norm": 1.3868846893310547,
      "learning_rate": 9.331306663496627e-05,
      "loss": 1.7107,
      "step": 1857
    },
    {
      "epoch": 0.3346993920288223,
      "grad_norm": 2.7034528255462646,
      "learning_rate": 9.330598992553585e-05,
      "loss": 2.3245,
      "step": 1858
    },
    {
      "epoch": 0.33487953163701867,
      "grad_norm": 1.936898112297058,
      "learning_rate": 9.329890974211999e-05,
      "loss": 2.3913,
      "step": 1859
    },
    {
      "epoch": 0.335059671245215,
      "grad_norm": 1.9338799715042114,
      "learning_rate": 9.329182608528671e-05,
      "loss": 2.6362,
      "step": 1860
    },
    {
      "epoch": 0.33523981085341137,
      "grad_norm": 1.5080702304840088,
      "learning_rate": 9.32847389556042e-05,
      "loss": 2.0042,
      "step": 1861
    },
    {
      "epoch": 0.3354199504616077,
      "grad_norm": 1.262425422668457,
      "learning_rate": 9.327764835364102e-05,
      "loss": 1.7909,
      "step": 1862
    },
    {
      "epoch": 0.3356000900698041,
      "grad_norm": 1.2286747694015503,
      "learning_rate": 9.327055427996598e-05,
      "loss": 1.5188,
      "step": 1863
    },
    {
      "epoch": 0.3357802296780005,
      "grad_norm": 1.3243101835250854,
      "learning_rate": 9.326345673514813e-05,
      "loss": 1.6541,
      "step": 1864
    },
    {
      "epoch": 0.3359603692861968,
      "grad_norm": 1.3640878200531006,
      "learning_rate": 9.325635571975688e-05,
      "loss": 1.9318,
      "step": 1865
    },
    {
      "epoch": 0.3361405088943932,
      "grad_norm": 1.3129050731658936,
      "learning_rate": 9.324925123436183e-05,
      "loss": 1.8943,
      "step": 1866
    },
    {
      "epoch": 0.3363206485025895,
      "grad_norm": 1.4156001806259155,
      "learning_rate": 9.324214327953292e-05,
      "loss": 1.8235,
      "step": 1867
    },
    {
      "epoch": 0.3365007881107859,
      "grad_norm": 1.2496826648712158,
      "learning_rate": 9.323503185584032e-05,
      "loss": 1.6156,
      "step": 1868
    },
    {
      "epoch": 0.3366809277189822,
      "grad_norm": 1.4432305097579956,
      "learning_rate": 9.322791696385454e-05,
      "loss": 1.9378,
      "step": 1869
    },
    {
      "epoch": 0.3368610673271786,
      "grad_norm": 1.3068046569824219,
      "learning_rate": 9.322079860414632e-05,
      "loss": 1.8379,
      "step": 1870
    },
    {
      "epoch": 0.3370412069353749,
      "grad_norm": 1.2956987619400024,
      "learning_rate": 9.321367677728667e-05,
      "loss": 1.862,
      "step": 1871
    },
    {
      "epoch": 0.3372213465435713,
      "grad_norm": 1.2337764501571655,
      "learning_rate": 9.320655148384694e-05,
      "loss": 1.6232,
      "step": 1872
    },
    {
      "epoch": 0.3374014861517676,
      "grad_norm": 1.186627745628357,
      "learning_rate": 9.319942272439867e-05,
      "loss": 1.6559,
      "step": 1873
    },
    {
      "epoch": 0.337581625759964,
      "grad_norm": 1.2476738691329956,
      "learning_rate": 9.319229049951378e-05,
      "loss": 1.8412,
      "step": 1874
    },
    {
      "epoch": 0.3377617653681603,
      "grad_norm": 1.2603083848953247,
      "learning_rate": 9.318515480976437e-05,
      "loss": 1.6934,
      "step": 1875
    },
    {
      "epoch": 0.3379419049763567,
      "grad_norm": 1.255574345588684,
      "learning_rate": 9.317801565572287e-05,
      "loss": 1.615,
      "step": 1876
    },
    {
      "epoch": 0.338122044584553,
      "grad_norm": 1.4196699857711792,
      "learning_rate": 9.3170873037962e-05,
      "loss": 1.872,
      "step": 1877
    },
    {
      "epoch": 0.3383021841927494,
      "grad_norm": 1.3192800283432007,
      "learning_rate": 9.316372695705471e-05,
      "loss": 1.8323,
      "step": 1878
    },
    {
      "epoch": 0.33848232380094573,
      "grad_norm": 1.3001998662948608,
      "learning_rate": 9.315657741357427e-05,
      "loss": 1.6802,
      "step": 1879
    },
    {
      "epoch": 0.3386624634091421,
      "grad_norm": 1.3281223773956299,
      "learning_rate": 9.314942440809421e-05,
      "loss": 1.702,
      "step": 1880
    },
    {
      "epoch": 0.33884260301733843,
      "grad_norm": 1.3211722373962402,
      "learning_rate": 9.314226794118834e-05,
      "loss": 1.7073,
      "step": 1881
    },
    {
      "epoch": 0.3390227426255348,
      "grad_norm": 1.2838819026947021,
      "learning_rate": 9.313510801343075e-05,
      "loss": 1.7984,
      "step": 1882
    },
    {
      "epoch": 0.33920288223373113,
      "grad_norm": 1.1914340257644653,
      "learning_rate": 9.312794462539581e-05,
      "loss": 1.7057,
      "step": 1883
    },
    {
      "epoch": 0.3393830218419275,
      "grad_norm": 1.2363718748092651,
      "learning_rate": 9.312077777765816e-05,
      "loss": 1.531,
      "step": 1884
    },
    {
      "epoch": 0.33956316145012383,
      "grad_norm": 1.5167639255523682,
      "learning_rate": 9.311360747079272e-05,
      "loss": 1.8029,
      "step": 1885
    },
    {
      "epoch": 0.3397433010583202,
      "grad_norm": 1.523821473121643,
      "learning_rate": 9.310643370537469e-05,
      "loss": 1.8245,
      "step": 1886
    },
    {
      "epoch": 0.33992344066651653,
      "grad_norm": 1.2223461866378784,
      "learning_rate": 9.309925648197952e-05,
      "loss": 1.5311,
      "step": 1887
    },
    {
      "epoch": 0.3401035802747129,
      "grad_norm": 1.393572211265564,
      "learning_rate": 9.309207580118302e-05,
      "loss": 1.9235,
      "step": 1888
    },
    {
      "epoch": 0.34028371988290923,
      "grad_norm": 1.4057350158691406,
      "learning_rate": 9.308489166356117e-05,
      "loss": 1.8884,
      "step": 1889
    },
    {
      "epoch": 0.3404638594911056,
      "grad_norm": 1.3276938199996948,
      "learning_rate": 9.30777040696903e-05,
      "loss": 1.6955,
      "step": 1890
    },
    {
      "epoch": 0.340643999099302,
      "grad_norm": 1.316297173500061,
      "learning_rate": 9.307051302014701e-05,
      "loss": 1.7297,
      "step": 1891
    },
    {
      "epoch": 0.34082413870749834,
      "grad_norm": 1.4817277193069458,
      "learning_rate": 9.306331851550813e-05,
      "loss": 1.9366,
      "step": 1892
    },
    {
      "epoch": 0.3410042783156947,
      "grad_norm": 1.260643720626831,
      "learning_rate": 9.30561205563508e-05,
      "loss": 1.7532,
      "step": 1893
    },
    {
      "epoch": 0.34118441792389104,
      "grad_norm": 1.4127812385559082,
      "learning_rate": 9.304891914325247e-05,
      "loss": 1.69,
      "step": 1894
    },
    {
      "epoch": 0.3413645575320874,
      "grad_norm": 1.4184190034866333,
      "learning_rate": 9.304171427679081e-05,
      "loss": 1.8321,
      "step": 1895
    },
    {
      "epoch": 0.34154469714028374,
      "grad_norm": 1.4236093759536743,
      "learning_rate": 9.30345059575438e-05,
      "loss": 1.825,
      "step": 1896
    },
    {
      "epoch": 0.3417248367484801,
      "grad_norm": 1.3687212467193604,
      "learning_rate": 9.302729418608969e-05,
      "loss": 1.6191,
      "step": 1897
    },
    {
      "epoch": 0.34190497635667644,
      "grad_norm": 1.453348994255066,
      "learning_rate": 9.302007896300698e-05,
      "loss": 1.6804,
      "step": 1898
    },
    {
      "epoch": 0.3420851159648728,
      "grad_norm": 1.274766445159912,
      "learning_rate": 9.301286028887449e-05,
      "loss": 1.4069,
      "step": 1899
    },
    {
      "epoch": 0.34226525557306914,
      "grad_norm": 1.2411935329437256,
      "learning_rate": 9.30056381642713e-05,
      "loss": 1.8631,
      "step": 1900
    },
    {
      "epoch": 0.3424453951812655,
      "grad_norm": 1.3289551734924316,
      "learning_rate": 9.299841258977678e-05,
      "loss": 2.3693,
      "step": 1901
    },
    {
      "epoch": 0.34262553478946184,
      "grad_norm": 1.3542392253875732,
      "learning_rate": 9.299118356597053e-05,
      "loss": 2.1098,
      "step": 1902
    },
    {
      "epoch": 0.3428056743976582,
      "grad_norm": 1.5168741941452026,
      "learning_rate": 9.298395109343247e-05,
      "loss": 2.1701,
      "step": 1903
    },
    {
      "epoch": 0.34298581400585454,
      "grad_norm": 1.4525636434555054,
      "learning_rate": 9.29767151727428e-05,
      "loss": 2.2253,
      "step": 1904
    },
    {
      "epoch": 0.3431659536140509,
      "grad_norm": 1.313673496246338,
      "learning_rate": 9.296947580448199e-05,
      "loss": 2.1184,
      "step": 1905
    },
    {
      "epoch": 0.34334609322224724,
      "grad_norm": 1.2006787061691284,
      "learning_rate": 9.296223298923073e-05,
      "loss": 1.9584,
      "step": 1906
    },
    {
      "epoch": 0.3435262328304436,
      "grad_norm": 1.3107112646102905,
      "learning_rate": 9.295498672757009e-05,
      "loss": 2.2213,
      "step": 1907
    },
    {
      "epoch": 0.34370637243863994,
      "grad_norm": 2.0840630531311035,
      "learning_rate": 9.294773702008131e-05,
      "loss": 2.2139,
      "step": 1908
    },
    {
      "epoch": 0.3438865120468363,
      "grad_norm": 1.8646247386932373,
      "learning_rate": 9.2940483867346e-05,
      "loss": 1.9821,
      "step": 1909
    },
    {
      "epoch": 0.34406665165503264,
      "grad_norm": 1.9690399169921875,
      "learning_rate": 9.2933227269946e-05,
      "loss": 2.6646,
      "step": 1910
    },
    {
      "epoch": 0.344246791263229,
      "grad_norm": 1.2502071857452393,
      "learning_rate": 9.29259672284634e-05,
      "loss": 1.6817,
      "step": 1911
    },
    {
      "epoch": 0.34442693087142534,
      "grad_norm": 1.3751085996627808,
      "learning_rate": 9.291870374348063e-05,
      "loss": 1.8343,
      "step": 1912
    },
    {
      "epoch": 0.3446070704796217,
      "grad_norm": 1.3243052959442139,
      "learning_rate": 9.291143681558035e-05,
      "loss": 1.788,
      "step": 1913
    },
    {
      "epoch": 0.34478721008781804,
      "grad_norm": 1.3268070220947266,
      "learning_rate": 9.29041664453455e-05,
      "loss": 1.4985,
      "step": 1914
    },
    {
      "epoch": 0.3449673496960144,
      "grad_norm": 1.3265726566314697,
      "learning_rate": 9.289689263335933e-05,
      "loss": 1.8567,
      "step": 1915
    },
    {
      "epoch": 0.34514748930421074,
      "grad_norm": 1.3508665561676025,
      "learning_rate": 9.288961538020533e-05,
      "loss": 1.6855,
      "step": 1916
    },
    {
      "epoch": 0.3453276289124071,
      "grad_norm": 1.24409818649292,
      "learning_rate": 9.288233468646726e-05,
      "loss": 1.6284,
      "step": 1917
    },
    {
      "epoch": 0.34550776852060344,
      "grad_norm": 1.3220150470733643,
      "learning_rate": 9.28750505527292e-05,
      "loss": 1.7913,
      "step": 1918
    },
    {
      "epoch": 0.3456879081287998,
      "grad_norm": 1.2090907096862793,
      "learning_rate": 9.286776297957546e-05,
      "loss": 1.5898,
      "step": 1919
    },
    {
      "epoch": 0.3458680477369962,
      "grad_norm": 1.1936092376708984,
      "learning_rate": 9.286047196759066e-05,
      "loss": 1.6078,
      "step": 1920
    },
    {
      "epoch": 0.34604818734519255,
      "grad_norm": 1.2637219429016113,
      "learning_rate": 9.285317751735967e-05,
      "loss": 1.7801,
      "step": 1921
    },
    {
      "epoch": 0.3462283269533889,
      "grad_norm": 1.2839415073394775,
      "learning_rate": 9.284587962946765e-05,
      "loss": 1.9403,
      "step": 1922
    },
    {
      "epoch": 0.34640846656158525,
      "grad_norm": 1.4117120504379272,
      "learning_rate": 9.283857830450004e-05,
      "loss": 2.1124,
      "step": 1923
    },
    {
      "epoch": 0.3465886061697816,
      "grad_norm": 1.3012604713439941,
      "learning_rate": 9.283127354304255e-05,
      "loss": 1.6945,
      "step": 1924
    },
    {
      "epoch": 0.34676874577797795,
      "grad_norm": 1.2102127075195312,
      "learning_rate": 9.282396534568116e-05,
      "loss": 1.7112,
      "step": 1925
    },
    {
      "epoch": 0.3469488853861743,
      "grad_norm": 1.2092266082763672,
      "learning_rate": 9.281665371300212e-05,
      "loss": 1.8522,
      "step": 1926
    },
    {
      "epoch": 0.34712902499437065,
      "grad_norm": 1.4207414388656616,
      "learning_rate": 9.280933864559198e-05,
      "loss": 1.9325,
      "step": 1927
    },
    {
      "epoch": 0.347309164602567,
      "grad_norm": 1.3066776990890503,
      "learning_rate": 9.280202014403754e-05,
      "loss": 1.8754,
      "step": 1928
    },
    {
      "epoch": 0.34748930421076335,
      "grad_norm": 1.2406049966812134,
      "learning_rate": 9.27946982089259e-05,
      "loss": 1.8758,
      "step": 1929
    },
    {
      "epoch": 0.3476694438189597,
      "grad_norm": 1.3070012331008911,
      "learning_rate": 9.27873728408444e-05,
      "loss": 1.9121,
      "step": 1930
    },
    {
      "epoch": 0.34784958342715605,
      "grad_norm": 1.199819803237915,
      "learning_rate": 9.278004404038069e-05,
      "loss": 1.8932,
      "step": 1931
    },
    {
      "epoch": 0.3480297230353524,
      "grad_norm": 1.3043848276138306,
      "learning_rate": 9.277271180812271e-05,
      "loss": 1.668,
      "step": 1932
    },
    {
      "epoch": 0.34820986264354875,
      "grad_norm": 1.2169181108474731,
      "learning_rate": 9.276537614465862e-05,
      "loss": 1.8083,
      "step": 1933
    },
    {
      "epoch": 0.3483900022517451,
      "grad_norm": 1.2651370763778687,
      "learning_rate": 9.275803705057686e-05,
      "loss": 1.565,
      "step": 1934
    },
    {
      "epoch": 0.34857014185994145,
      "grad_norm": 1.3813928365707397,
      "learning_rate": 9.275069452646621e-05,
      "loss": 1.6281,
      "step": 1935
    },
    {
      "epoch": 0.3487502814681378,
      "grad_norm": 1.2649669647216797,
      "learning_rate": 9.274334857291567e-05,
      "loss": 1.5485,
      "step": 1936
    },
    {
      "epoch": 0.34893042107633415,
      "grad_norm": 1.406266212463379,
      "learning_rate": 9.273599919051452e-05,
      "loss": 1.7461,
      "step": 1937
    },
    {
      "epoch": 0.3491105606845305,
      "grad_norm": 1.3586225509643555,
      "learning_rate": 9.272864637985234e-05,
      "loss": 2.2013,
      "step": 1938
    },
    {
      "epoch": 0.34929070029272685,
      "grad_norm": 1.3642197847366333,
      "learning_rate": 9.272129014151896e-05,
      "loss": 1.9002,
      "step": 1939
    },
    {
      "epoch": 0.3494708399009232,
      "grad_norm": 1.472520112991333,
      "learning_rate": 9.271393047610448e-05,
      "loss": 1.9047,
      "step": 1940
    },
    {
      "epoch": 0.34965097950911955,
      "grad_norm": 1.470763921737671,
      "learning_rate": 9.27065673841993e-05,
      "loss": 1.9646,
      "step": 1941
    },
    {
      "epoch": 0.3498311191173159,
      "grad_norm": 1.3430933952331543,
      "learning_rate": 9.26992008663941e-05,
      "loss": 1.6854,
      "step": 1942
    },
    {
      "epoch": 0.35001125872551225,
      "grad_norm": 1.2034293413162231,
      "learning_rate": 9.26918309232798e-05,
      "loss": 1.5191,
      "step": 1943
    },
    {
      "epoch": 0.3501913983337086,
      "grad_norm": 1.3794699907302856,
      "learning_rate": 9.268445755544763e-05,
      "loss": 1.7684,
      "step": 1944
    },
    {
      "epoch": 0.35037153794190495,
      "grad_norm": 1.440076470375061,
      "learning_rate": 9.267708076348904e-05,
      "loss": 1.6696,
      "step": 1945
    },
    {
      "epoch": 0.3505516775501013,
      "grad_norm": 1.313086748123169,
      "learning_rate": 9.266970054799584e-05,
      "loss": 1.4966,
      "step": 1946
    },
    {
      "epoch": 0.35073181715829765,
      "grad_norm": 1.2346845865249634,
      "learning_rate": 9.266231690956004e-05,
      "loss": 1.6228,
      "step": 1947
    },
    {
      "epoch": 0.35091195676649406,
      "grad_norm": 1.6065713167190552,
      "learning_rate": 9.265492984877396e-05,
      "loss": 1.9179,
      "step": 1948
    },
    {
      "epoch": 0.3510920963746904,
      "grad_norm": 1.4420204162597656,
      "learning_rate": 9.264753936623017e-05,
      "loss": 1.6355,
      "step": 1949
    },
    {
      "epoch": 0.35127223598288676,
      "grad_norm": 1.321545958518982,
      "learning_rate": 9.264014546252155e-05,
      "loss": 1.6243,
      "step": 1950
    },
    {
      "epoch": 0.3514523755910831,
      "grad_norm": 1.5253962278366089,
      "learning_rate": 9.263274813824125e-05,
      "loss": 2.0982,
      "step": 1951
    },
    {
      "epoch": 0.35163251519927946,
      "grad_norm": 2.1106340885162354,
      "learning_rate": 9.262534739398264e-05,
      "loss": 2.1145,
      "step": 1952
    },
    {
      "epoch": 0.3518126548074758,
      "grad_norm": 1.6434231996536255,
      "learning_rate": 9.261794323033943e-05,
      "loss": 2.0523,
      "step": 1953
    },
    {
      "epoch": 0.35199279441567216,
      "grad_norm": 1.1787556409835815,
      "learning_rate": 9.261053564790558e-05,
      "loss": 2.0348,
      "step": 1954
    },
    {
      "epoch": 0.3521729340238685,
      "grad_norm": 1.3213731050491333,
      "learning_rate": 9.26031246472753e-05,
      "loss": 1.8265,
      "step": 1955
    },
    {
      "epoch": 0.35235307363206486,
      "grad_norm": 1.4476991891860962,
      "learning_rate": 9.259571022904312e-05,
      "loss": 2.2012,
      "step": 1956
    },
    {
      "epoch": 0.3525332132402612,
      "grad_norm": 1.560680627822876,
      "learning_rate": 9.25882923938038e-05,
      "loss": 2.1016,
      "step": 1957
    },
    {
      "epoch": 0.35271335284845756,
      "grad_norm": 1.5593386888504028,
      "learning_rate": 9.258087114215243e-05,
      "loss": 2.3905,
      "step": 1958
    },
    {
      "epoch": 0.3528934924566539,
      "grad_norm": 2.4309866428375244,
      "learning_rate": 9.257344647468431e-05,
      "loss": 2.441,
      "step": 1959
    },
    {
      "epoch": 0.35307363206485026,
      "grad_norm": 1.4332079887390137,
      "learning_rate": 9.256601839199506e-05,
      "loss": 2.1198,
      "step": 1960
    },
    {
      "epoch": 0.3532537716730466,
      "grad_norm": 1.2350066900253296,
      "learning_rate": 9.255858689468054e-05,
      "loss": 1.687,
      "step": 1961
    },
    {
      "epoch": 0.35343391128124296,
      "grad_norm": 1.3483506441116333,
      "learning_rate": 9.255115198333689e-05,
      "loss": 1.9088,
      "step": 1962
    },
    {
      "epoch": 0.3536140508894393,
      "grad_norm": 1.2272214889526367,
      "learning_rate": 9.254371365856058e-05,
      "loss": 1.7149,
      "step": 1963
    },
    {
      "epoch": 0.35379419049763566,
      "grad_norm": 1.189282774925232,
      "learning_rate": 9.253627192094827e-05,
      "loss": 1.6502,
      "step": 1964
    },
    {
      "epoch": 0.353974330105832,
      "grad_norm": 1.2303739786148071,
      "learning_rate": 9.252882677109693e-05,
      "loss": 1.7893,
      "step": 1965
    },
    {
      "epoch": 0.35415446971402836,
      "grad_norm": 1.2383229732513428,
      "learning_rate": 9.252137820960384e-05,
      "loss": 1.7523,
      "step": 1966
    },
    {
      "epoch": 0.3543346093222247,
      "grad_norm": 1.2676115036010742,
      "learning_rate": 9.251392623706651e-05,
      "loss": 1.675,
      "step": 1967
    },
    {
      "epoch": 0.35451474893042106,
      "grad_norm": 1.1664024591445923,
      "learning_rate": 9.25064708540827e-05,
      "loss": 1.5168,
      "step": 1968
    },
    {
      "epoch": 0.3546948885386174,
      "grad_norm": 1.1627534627914429,
      "learning_rate": 9.249901206125053e-05,
      "loss": 1.4811,
      "step": 1969
    },
    {
      "epoch": 0.35487502814681376,
      "grad_norm": 1.2342137098312378,
      "learning_rate": 9.249154985916831e-05,
      "loss": 1.5882,
      "step": 1970
    },
    {
      "epoch": 0.3550551677550101,
      "grad_norm": 1.2362819910049438,
      "learning_rate": 9.248408424843464e-05,
      "loss": 1.7096,
      "step": 1971
    },
    {
      "epoch": 0.35523530736320647,
      "grad_norm": 1.2362438440322876,
      "learning_rate": 9.247661522964844e-05,
      "loss": 1.6364,
      "step": 1972
    },
    {
      "epoch": 0.3554154469714028,
      "grad_norm": 1.391287922859192,
      "learning_rate": 9.246914280340884e-05,
      "loss": 1.9177,
      "step": 1973
    },
    {
      "epoch": 0.35559558657959917,
      "grad_norm": 1.316846489906311,
      "learning_rate": 9.246166697031529e-05,
      "loss": 1.6523,
      "step": 1974
    },
    {
      "epoch": 0.3557757261877955,
      "grad_norm": 1.258087158203125,
      "learning_rate": 9.245418773096751e-05,
      "loss": 1.6798,
      "step": 1975
    },
    {
      "epoch": 0.3559558657959919,
      "grad_norm": 1.675063133239746,
      "learning_rate": 9.244670508596547e-05,
      "loss": 2.1753,
      "step": 1976
    },
    {
      "epoch": 0.35613600540418827,
      "grad_norm": 1.3191180229187012,
      "learning_rate": 9.243921903590942e-05,
      "loss": 1.6388,
      "step": 1977
    },
    {
      "epoch": 0.3563161450123846,
      "grad_norm": 1.2297879457473755,
      "learning_rate": 9.243172958139989e-05,
      "loss": 1.6644,
      "step": 1978
    },
    {
      "epoch": 0.35649628462058097,
      "grad_norm": 1.2635867595672607,
      "learning_rate": 9.242423672303768e-05,
      "loss": 1.7228,
      "step": 1979
    },
    {
      "epoch": 0.3566764242287773,
      "grad_norm": 1.3461564779281616,
      "learning_rate": 9.241674046142386e-05,
      "loss": 1.9923,
      "step": 1980
    },
    {
      "epoch": 0.3568565638369737,
      "grad_norm": 1.2456141710281372,
      "learning_rate": 9.240924079715977e-05,
      "loss": 1.7294,
      "step": 1981
    },
    {
      "epoch": 0.35703670344517,
      "grad_norm": 1.3183419704437256,
      "learning_rate": 9.240173773084706e-05,
      "loss": 1.747,
      "step": 1982
    },
    {
      "epoch": 0.3572168430533664,
      "grad_norm": 1.2270807027816772,
      "learning_rate": 9.239423126308759e-05,
      "loss": 1.6177,
      "step": 1983
    },
    {
      "epoch": 0.3573969826615627,
      "grad_norm": 1.2759424448013306,
      "learning_rate": 9.238672139448354e-05,
      "loss": 1.7872,
      "step": 1984
    },
    {
      "epoch": 0.3575771222697591,
      "grad_norm": 1.3207776546478271,
      "learning_rate": 9.237920812563736e-05,
      "loss": 1.7999,
      "step": 1985
    },
    {
      "epoch": 0.3577572618779554,
      "grad_norm": 1.4106075763702393,
      "learning_rate": 9.237169145715175e-05,
      "loss": 1.7686,
      "step": 1986
    },
    {
      "epoch": 0.3579374014861518,
      "grad_norm": 1.3376574516296387,
      "learning_rate": 9.236417138962966e-05,
      "loss": 1.6492,
      "step": 1987
    },
    {
      "epoch": 0.3581175410943481,
      "grad_norm": 1.3242768049240112,
      "learning_rate": 9.23566479236744e-05,
      "loss": 1.6214,
      "step": 1988
    },
    {
      "epoch": 0.3582976807025445,
      "grad_norm": 1.199057936668396,
      "learning_rate": 9.234912105988947e-05,
      "loss": 1.7488,
      "step": 1989
    },
    {
      "epoch": 0.3584778203107408,
      "grad_norm": 1.4324913024902344,
      "learning_rate": 9.234159079887867e-05,
      "loss": 1.5512,
      "step": 1990
    },
    {
      "epoch": 0.3586579599189372,
      "grad_norm": 1.341431975364685,
      "learning_rate": 9.233405714124608e-05,
      "loss": 1.699,
      "step": 1991
    },
    {
      "epoch": 0.3588380995271335,
      "grad_norm": 1.3021507263183594,
      "learning_rate": 9.232652008759605e-05,
      "loss": 1.8533,
      "step": 1992
    },
    {
      "epoch": 0.3590182391353299,
      "grad_norm": 1.3398795127868652,
      "learning_rate": 9.231897963853318e-05,
      "loss": 1.7793,
      "step": 1993
    },
    {
      "epoch": 0.3591983787435262,
      "grad_norm": 1.301111102104187,
      "learning_rate": 9.231143579466239e-05,
      "loss": 1.6081,
      "step": 1994
    },
    {
      "epoch": 0.3593785183517226,
      "grad_norm": 1.285243034362793,
      "learning_rate": 9.230388855658884e-05,
      "loss": 1.6836,
      "step": 1995
    },
    {
      "epoch": 0.3595586579599189,
      "grad_norm": 1.3567155599594116,
      "learning_rate": 9.229633792491795e-05,
      "loss": 1.6616,
      "step": 1996
    },
    {
      "epoch": 0.3597387975681153,
      "grad_norm": 1.3073797225952148,
      "learning_rate": 9.228878390025543e-05,
      "loss": 1.7619,
      "step": 1997
    },
    {
      "epoch": 0.3599189371763116,
      "grad_norm": 1.5632058382034302,
      "learning_rate": 9.228122648320725e-05,
      "loss": 2.1415,
      "step": 1998
    },
    {
      "epoch": 0.360099076784508,
      "grad_norm": 1.3863821029663086,
      "learning_rate": 9.227366567437969e-05,
      "loss": 1.6922,
      "step": 1999
    },
    {
      "epoch": 0.3602792163927043,
      "grad_norm": 1.2061693668365479,
      "learning_rate": 9.226610147437926e-05,
      "loss": 1.3197,
      "step": 2000
    },
    {
      "epoch": 0.3604593560009007,
      "grad_norm": 1.421900749206543,
      "learning_rate": 9.225853388381275e-05,
      "loss": 2.2009,
      "step": 2001
    },
    {
      "epoch": 0.360639495609097,
      "grad_norm": 1.434679627418518,
      "learning_rate": 9.225096290328725e-05,
      "loss": 2.0606,
      "step": 2002
    },
    {
      "epoch": 0.3608196352172934,
      "grad_norm": 1.2855474948883057,
      "learning_rate": 9.224338853341008e-05,
      "loss": 2.006,
      "step": 2003
    },
    {
      "epoch": 0.3609997748254897,
      "grad_norm": 1.268730640411377,
      "learning_rate": 9.223581077478885e-05,
      "loss": 1.8865,
      "step": 2004
    },
    {
      "epoch": 0.36117991443368613,
      "grad_norm": 1.4431568384170532,
      "learning_rate": 9.222822962803146e-05,
      "loss": 2.3008,
      "step": 2005
    },
    {
      "epoch": 0.3613600540418825,
      "grad_norm": 1.2382725477218628,
      "learning_rate": 9.222064509374605e-05,
      "loss": 1.6778,
      "step": 2006
    },
    {
      "epoch": 0.36154019365007883,
      "grad_norm": 1.622361183166504,
      "learning_rate": 9.221305717254105e-05,
      "loss": 2.029,
      "step": 2007
    },
    {
      "epoch": 0.3617203332582752,
      "grad_norm": 1.5106502771377563,
      "learning_rate": 9.220546586502519e-05,
      "loss": 2.3058,
      "step": 2008
    },
    {
      "epoch": 0.36190047286647153,
      "grad_norm": 1.5847405195236206,
      "learning_rate": 9.21978711718074e-05,
      "loss": 2.3521,
      "step": 2009
    },
    {
      "epoch": 0.3620806124746679,
      "grad_norm": 1.8725775480270386,
      "learning_rate": 9.219027309349692e-05,
      "loss": 2.6225,
      "step": 2010
    },
    {
      "epoch": 0.36226075208286423,
      "grad_norm": 1.914020299911499,
      "learning_rate": 9.218267163070332e-05,
      "loss": 2.2776,
      "step": 2011
    },
    {
      "epoch": 0.3624408916910606,
      "grad_norm": 1.284096598625183,
      "learning_rate": 9.217506678403633e-05,
      "loss": 1.7909,
      "step": 2012
    },
    {
      "epoch": 0.36262103129925694,
      "grad_norm": 1.2840335369110107,
      "learning_rate": 9.216745855410602e-05,
      "loss": 1.6491,
      "step": 2013
    },
    {
      "epoch": 0.3628011709074533,
      "grad_norm": 1.3496419191360474,
      "learning_rate": 9.215984694152273e-05,
      "loss": 1.9515,
      "step": 2014
    },
    {
      "epoch": 0.36298131051564964,
      "grad_norm": 1.2450848817825317,
      "learning_rate": 9.215223194689705e-05,
      "loss": 1.6875,
      "step": 2015
    },
    {
      "epoch": 0.363161450123846,
      "grad_norm": 1.1996676921844482,
      "learning_rate": 9.214461357083985e-05,
      "loss": 1.6659,
      "step": 2016
    },
    {
      "epoch": 0.36334158973204234,
      "grad_norm": 1.3575389385223389,
      "learning_rate": 9.21369918139623e-05,
      "loss": 1.622,
      "step": 2017
    },
    {
      "epoch": 0.3635217293402387,
      "grad_norm": 1.3158403635025024,
      "learning_rate": 9.212936667687578e-05,
      "loss": 1.7217,
      "step": 2018
    },
    {
      "epoch": 0.36370186894843504,
      "grad_norm": 1.3368136882781982,
      "learning_rate": 9.212173816019197e-05,
      "loss": 1.7732,
      "step": 2019
    },
    {
      "epoch": 0.3638820085566314,
      "grad_norm": 1.40514075756073,
      "learning_rate": 9.211410626452286e-05,
      "loss": 1.9252,
      "step": 2020
    },
    {
      "epoch": 0.36406214816482774,
      "grad_norm": 1.1953845024108887,
      "learning_rate": 9.210647099048064e-05,
      "loss": 1.6109,
      "step": 2021
    },
    {
      "epoch": 0.3642422877730241,
      "grad_norm": 1.245178461074829,
      "learning_rate": 9.209883233867784e-05,
      "loss": 1.693,
      "step": 2022
    },
    {
      "epoch": 0.36442242738122044,
      "grad_norm": 1.197798252105713,
      "learning_rate": 9.209119030972722e-05,
      "loss": 1.703,
      "step": 2023
    },
    {
      "epoch": 0.3646025669894168,
      "grad_norm": 1.398685336112976,
      "learning_rate": 9.20835449042418e-05,
      "loss": 1.9915,
      "step": 2024
    },
    {
      "epoch": 0.36478270659761314,
      "grad_norm": 1.2733819484710693,
      "learning_rate": 9.207589612283492e-05,
      "loss": 1.7829,
      "step": 2025
    },
    {
      "epoch": 0.3649628462058095,
      "grad_norm": 1.304255485534668,
      "learning_rate": 9.206824396612013e-05,
      "loss": 1.8843,
      "step": 2026
    },
    {
      "epoch": 0.36514298581400584,
      "grad_norm": 1.1876626014709473,
      "learning_rate": 9.206058843471133e-05,
      "loss": 1.7658,
      "step": 2027
    },
    {
      "epoch": 0.3653231254222022,
      "grad_norm": 1.302866816520691,
      "learning_rate": 9.205292952922257e-05,
      "loss": 1.9191,
      "step": 2028
    },
    {
      "epoch": 0.36550326503039854,
      "grad_norm": 1.3426549434661865,
      "learning_rate": 9.204526725026832e-05,
      "loss": 1.7502,
      "step": 2029
    },
    {
      "epoch": 0.3656834046385949,
      "grad_norm": 1.3348923921585083,
      "learning_rate": 9.20376015984632e-05,
      "loss": 1.835,
      "step": 2030
    },
    {
      "epoch": 0.36586354424679124,
      "grad_norm": 1.2460559606552124,
      "learning_rate": 9.202993257442217e-05,
      "loss": 1.6431,
      "step": 2031
    },
    {
      "epoch": 0.3660436838549876,
      "grad_norm": 1.2204155921936035,
      "learning_rate": 9.20222601787604e-05,
      "loss": 1.7716,
      "step": 2032
    },
    {
      "epoch": 0.366223823463184,
      "grad_norm": 1.249332070350647,
      "learning_rate": 9.20145844120934e-05,
      "loss": 1.6243,
      "step": 2033
    },
    {
      "epoch": 0.36640396307138035,
      "grad_norm": 1.2118375301361084,
      "learning_rate": 9.200690527503689e-05,
      "loss": 1.7388,
      "step": 2034
    },
    {
      "epoch": 0.3665841026795767,
      "grad_norm": 1.2627040147781372,
      "learning_rate": 9.199922276820691e-05,
      "loss": 1.5562,
      "step": 2035
    },
    {
      "epoch": 0.36676424228777305,
      "grad_norm": 1.2564849853515625,
      "learning_rate": 9.199153689221974e-05,
      "loss": 1.6902,
      "step": 2036
    },
    {
      "epoch": 0.3669443818959694,
      "grad_norm": 1.3405301570892334,
      "learning_rate": 9.198384764769192e-05,
      "loss": 1.7691,
      "step": 2037
    },
    {
      "epoch": 0.36712452150416575,
      "grad_norm": 1.261576533317566,
      "learning_rate": 9.19761550352403e-05,
      "loss": 1.595,
      "step": 2038
    },
    {
      "epoch": 0.3673046611123621,
      "grad_norm": 1.299034595489502,
      "learning_rate": 9.196845905548197e-05,
      "loss": 1.6305,
      "step": 2039
    },
    {
      "epoch": 0.36748480072055845,
      "grad_norm": 1.407024621963501,
      "learning_rate": 9.19607597090343e-05,
      "loss": 1.7265,
      "step": 2040
    },
    {
      "epoch": 0.3676649403287548,
      "grad_norm": 1.4784094095230103,
      "learning_rate": 9.195305699651491e-05,
      "loss": 1.9685,
      "step": 2041
    },
    {
      "epoch": 0.36784507993695115,
      "grad_norm": 1.4655539989471436,
      "learning_rate": 9.194535091854173e-05,
      "loss": 1.8251,
      "step": 2042
    },
    {
      "epoch": 0.3680252195451475,
      "grad_norm": 1.6981803178787231,
      "learning_rate": 9.193764147573294e-05,
      "loss": 2.1593,
      "step": 2043
    },
    {
      "epoch": 0.36820535915334385,
      "grad_norm": 1.373761534690857,
      "learning_rate": 9.192992866870695e-05,
      "loss": 1.7459,
      "step": 2044
    },
    {
      "epoch": 0.3683854987615402,
      "grad_norm": 1.385209083557129,
      "learning_rate": 9.192221249808254e-05,
      "loss": 1.5596,
      "step": 2045
    },
    {
      "epoch": 0.36856563836973655,
      "grad_norm": 1.4608234167099,
      "learning_rate": 9.191449296447865e-05,
      "loss": 1.6597,
      "step": 2046
    },
    {
      "epoch": 0.3687457779779329,
      "grad_norm": 1.4122939109802246,
      "learning_rate": 9.190677006851454e-05,
      "loss": 1.5083,
      "step": 2047
    },
    {
      "epoch": 0.36892591758612925,
      "grad_norm": 1.3078893423080444,
      "learning_rate": 9.189904381080977e-05,
      "loss": 1.5587,
      "step": 2048
    },
    {
      "epoch": 0.3691060571943256,
      "grad_norm": 1.304000973701477,
      "learning_rate": 9.18913141919841e-05,
      "loss": 1.3948,
      "step": 2049
    },
    {
      "epoch": 0.36928619680252195,
      "grad_norm": 1.4559122323989868,
      "learning_rate": 9.188358121265759e-05,
      "loss": 1.8596,
      "step": 2050
    },
    {
      "epoch": 0.3694663364107183,
      "grad_norm": 1.5265893936157227,
      "learning_rate": 9.187584487345062e-05,
      "loss": 2.3415,
      "step": 2051
    },
    {
      "epoch": 0.36964647601891465,
      "grad_norm": 1.9684908390045166,
      "learning_rate": 9.186810517498376e-05,
      "loss": 2.2039,
      "step": 2052
    },
    {
      "epoch": 0.369826615627111,
      "grad_norm": 2.2677884101867676,
      "learning_rate": 9.186036211787792e-05,
      "loss": 2.1073,
      "step": 2053
    },
    {
      "epoch": 0.37000675523530735,
      "grad_norm": 1.3882420063018799,
      "learning_rate": 9.18526157027542e-05,
      "loss": 2.0612,
      "step": 2054
    },
    {
      "epoch": 0.3701868948435037,
      "grad_norm": 1.5191304683685303,
      "learning_rate": 9.184486593023402e-05,
      "loss": 2.1988,
      "step": 2055
    },
    {
      "epoch": 0.37036703445170005,
      "grad_norm": 1.4260764122009277,
      "learning_rate": 9.18371128009391e-05,
      "loss": 2.2295,
      "step": 2056
    },
    {
      "epoch": 0.3705471740598964,
      "grad_norm": 1.4145503044128418,
      "learning_rate": 9.182935631549137e-05,
      "loss": 2.2642,
      "step": 2057
    },
    {
      "epoch": 0.37072731366809275,
      "grad_norm": 1.5801762342453003,
      "learning_rate": 9.182159647451305e-05,
      "loss": 2.1425,
      "step": 2058
    },
    {
      "epoch": 0.3709074532762891,
      "grad_norm": 1.551311731338501,
      "learning_rate": 9.181383327862662e-05,
      "loss": 2.1386,
      "step": 2059
    },
    {
      "epoch": 0.37108759288448545,
      "grad_norm": 1.894332766532898,
      "learning_rate": 9.180606672845486e-05,
      "loss": 2.6294,
      "step": 2060
    },
    {
      "epoch": 0.3712677324926818,
      "grad_norm": 1.365515112876892,
      "learning_rate": 9.179829682462078e-05,
      "loss": 2.0323,
      "step": 2061
    },
    {
      "epoch": 0.3714478721008782,
      "grad_norm": 1.3314865827560425,
      "learning_rate": 9.179052356774768e-05,
      "loss": 2.0089,
      "step": 2062
    },
    {
      "epoch": 0.37162801170907456,
      "grad_norm": 1.2352780103683472,
      "learning_rate": 9.178274695845917e-05,
      "loss": 1.7733,
      "step": 2063
    },
    {
      "epoch": 0.3718081513172709,
      "grad_norm": 1.2196687459945679,
      "learning_rate": 9.177496699737901e-05,
      "loss": 1.7665,
      "step": 2064
    },
    {
      "epoch": 0.37198829092546726,
      "grad_norm": 1.2358906269073486,
      "learning_rate": 9.176718368513137e-05,
      "loss": 1.7103,
      "step": 2065
    },
    {
      "epoch": 0.3721684305336636,
      "grad_norm": 1.3376888036727905,
      "learning_rate": 9.17593970223406e-05,
      "loss": 1.9167,
      "step": 2066
    },
    {
      "epoch": 0.37234857014185996,
      "grad_norm": 1.2779468297958374,
      "learning_rate": 9.175160700963134e-05,
      "loss": 1.8454,
      "step": 2067
    },
    {
      "epoch": 0.3725287097500563,
      "grad_norm": 1.2241846323013306,
      "learning_rate": 9.17438136476285e-05,
      "loss": 1.649,
      "step": 2068
    },
    {
      "epoch": 0.37270884935825266,
      "grad_norm": 1.262999176979065,
      "learning_rate": 9.173601693695725e-05,
      "loss": 1.6635,
      "step": 2069
    },
    {
      "epoch": 0.372888988966449,
      "grad_norm": 1.2626407146453857,
      "learning_rate": 9.172821687824307e-05,
      "loss": 1.7404,
      "step": 2070
    },
    {
      "epoch": 0.37306912857464536,
      "grad_norm": 1.2384082078933716,
      "learning_rate": 9.172041347211165e-05,
      "loss": 1.7461,
      "step": 2071
    },
    {
      "epoch": 0.3732492681828417,
      "grad_norm": 1.2913539409637451,
      "learning_rate": 9.171260671918897e-05,
      "loss": 1.5681,
      "step": 2072
    },
    {
      "epoch": 0.37342940779103806,
      "grad_norm": 1.1793841123580933,
      "learning_rate": 9.170479662010129e-05,
      "loss": 1.6358,
      "step": 2073
    },
    {
      "epoch": 0.3736095473992344,
      "grad_norm": 1.239508032798767,
      "learning_rate": 9.169698317547517e-05,
      "loss": 1.8285,
      "step": 2074
    },
    {
      "epoch": 0.37378968700743076,
      "grad_norm": 1.3518551588058472,
      "learning_rate": 9.168916638593736e-05,
      "loss": 1.8222,
      "step": 2075
    },
    {
      "epoch": 0.3739698266156271,
      "grad_norm": 1.2116771936416626,
      "learning_rate": 9.168134625211492e-05,
      "loss": 1.4637,
      "step": 2076
    },
    {
      "epoch": 0.37414996622382346,
      "grad_norm": 1.2727361917495728,
      "learning_rate": 9.167352277463519e-05,
      "loss": 1.6008,
      "step": 2077
    },
    {
      "epoch": 0.3743301058320198,
      "grad_norm": 1.3380126953125,
      "learning_rate": 9.166569595412575e-05,
      "loss": 1.8369,
      "step": 2078
    },
    {
      "epoch": 0.37451024544021616,
      "grad_norm": 1.2954001426696777,
      "learning_rate": 9.16578657912145e-05,
      "loss": 1.6829,
      "step": 2079
    },
    {
      "epoch": 0.3746903850484125,
      "grad_norm": 1.3621466159820557,
      "learning_rate": 9.165003228652952e-05,
      "loss": 1.9403,
      "step": 2080
    },
    {
      "epoch": 0.37487052465660886,
      "grad_norm": 1.5151996612548828,
      "learning_rate": 9.164219544069926e-05,
      "loss": 1.935,
      "step": 2081
    },
    {
      "epoch": 0.3750506642648052,
      "grad_norm": 1.4557955265045166,
      "learning_rate": 9.163435525435234e-05,
      "loss": 1.8357,
      "step": 2082
    },
    {
      "epoch": 0.37523080387300156,
      "grad_norm": 1.3402683734893799,
      "learning_rate": 9.162651172811773e-05,
      "loss": 1.7679,
      "step": 2083
    },
    {
      "epoch": 0.3754109434811979,
      "grad_norm": 1.3840535879135132,
      "learning_rate": 9.161866486262463e-05,
      "loss": 1.6648,
      "step": 2084
    },
    {
      "epoch": 0.37559108308939426,
      "grad_norm": 1.25978422164917,
      "learning_rate": 9.161081465850252e-05,
      "loss": 1.8722,
      "step": 2085
    },
    {
      "epoch": 0.3757712226975906,
      "grad_norm": 1.357959270477295,
      "learning_rate": 9.160296111638111e-05,
      "loss": 1.871,
      "step": 2086
    },
    {
      "epoch": 0.37595136230578696,
      "grad_norm": 1.4812606573104858,
      "learning_rate": 9.159510423689042e-05,
      "loss": 1.9384,
      "step": 2087
    },
    {
      "epoch": 0.3761315019139833,
      "grad_norm": 1.4350230693817139,
      "learning_rate": 9.158724402066074e-05,
      "loss": 1.9554,
      "step": 2088
    },
    {
      "epoch": 0.37631164152217966,
      "grad_norm": 1.4070546627044678,
      "learning_rate": 9.157938046832259e-05,
      "loss": 1.685,
      "step": 2089
    },
    {
      "epoch": 0.37649178113037607,
      "grad_norm": 1.4612318277359009,
      "learning_rate": 9.15715135805068e-05,
      "loss": 1.7132,
      "step": 2090
    },
    {
      "epoch": 0.3766719207385724,
      "grad_norm": 1.3602055311203003,
      "learning_rate": 9.156364335784443e-05,
      "loss": 1.6161,
      "step": 2091
    },
    {
      "epoch": 0.37685206034676877,
      "grad_norm": 1.3037409782409668,
      "learning_rate": 9.155576980096685e-05,
      "loss": 1.782,
      "step": 2092
    },
    {
      "epoch": 0.3770321999549651,
      "grad_norm": 1.3262662887573242,
      "learning_rate": 9.154789291050564e-05,
      "loss": 1.8636,
      "step": 2093
    },
    {
      "epoch": 0.37721233956316147,
      "grad_norm": 1.504162311553955,
      "learning_rate": 9.154001268709272e-05,
      "loss": 1.8956,
      "step": 2094
    },
    {
      "epoch": 0.3773924791713578,
      "grad_norm": 1.4728641510009766,
      "learning_rate": 9.153212913136018e-05,
      "loss": 1.9784,
      "step": 2095
    },
    {
      "epoch": 0.37757261877955417,
      "grad_norm": 1.2264498472213745,
      "learning_rate": 9.15242422439405e-05,
      "loss": 1.3746,
      "step": 2096
    },
    {
      "epoch": 0.3777527583877505,
      "grad_norm": 1.3264819383621216,
      "learning_rate": 9.151635202546634e-05,
      "loss": 1.7244,
      "step": 2097
    },
    {
      "epoch": 0.37793289799594687,
      "grad_norm": 1.2909444570541382,
      "learning_rate": 9.150845847657061e-05,
      "loss": 1.4397,
      "step": 2098
    },
    {
      "epoch": 0.3781130376041432,
      "grad_norm": 1.2654004096984863,
      "learning_rate": 9.150056159788658e-05,
      "loss": 1.5312,
      "step": 2099
    },
    {
      "epoch": 0.37829317721233957,
      "grad_norm": 1.4643536806106567,
      "learning_rate": 9.149266139004771e-05,
      "loss": 1.4035,
      "step": 2100
    },
    {
      "epoch": 0.3784733168205359,
      "grad_norm": 1.5059345960617065,
      "learning_rate": 9.148475785368776e-05,
      "loss": 2.4415,
      "step": 2101
    },
    {
      "epoch": 0.37865345642873227,
      "grad_norm": 1.2910499572753906,
      "learning_rate": 9.147685098944072e-05,
      "loss": 1.8699,
      "step": 2102
    },
    {
      "epoch": 0.3788335960369286,
      "grad_norm": 1.306274175643921,
      "learning_rate": 9.146894079794092e-05,
      "loss": 2.0729,
      "step": 2103
    },
    {
      "epoch": 0.37901373564512497,
      "grad_norm": 1.284865379333496,
      "learning_rate": 9.146102727982287e-05,
      "loss": 2.1571,
      "step": 2104
    },
    {
      "epoch": 0.3791938752533213,
      "grad_norm": 1.2388386726379395,
      "learning_rate": 9.145311043572142e-05,
      "loss": 1.8669,
      "step": 2105
    },
    {
      "epoch": 0.37937401486151767,
      "grad_norm": 1.7818042039871216,
      "learning_rate": 9.144519026627164e-05,
      "loss": 2.1748,
      "step": 2106
    },
    {
      "epoch": 0.379554154469714,
      "grad_norm": 1.6017647981643677,
      "learning_rate": 9.143726677210888e-05,
      "loss": 2.1431,
      "step": 2107
    },
    {
      "epoch": 0.3797342940779104,
      "grad_norm": 1.6454085111618042,
      "learning_rate": 9.142933995386876e-05,
      "loss": 2.3816,
      "step": 2108
    },
    {
      "epoch": 0.3799144336861067,
      "grad_norm": 1.982255458831787,
      "learning_rate": 9.142140981218717e-05,
      "loss": 2.4602,
      "step": 2109
    },
    {
      "epoch": 0.3800945732943031,
      "grad_norm": 1.2942668199539185,
      "learning_rate": 9.141347634770027e-05,
      "loss": 1.6573,
      "step": 2110
    },
    {
      "epoch": 0.3802747129024994,
      "grad_norm": 1.2739670276641846,
      "learning_rate": 9.140553956104447e-05,
      "loss": 1.7545,
      "step": 2111
    },
    {
      "epoch": 0.3804548525106958,
      "grad_norm": 1.1825495958328247,
      "learning_rate": 9.139759945285645e-05,
      "loss": 1.6799,
      "step": 2112
    },
    {
      "epoch": 0.3806349921188921,
      "grad_norm": 1.2051327228546143,
      "learning_rate": 9.138965602377315e-05,
      "loss": 1.4395,
      "step": 2113
    },
    {
      "epoch": 0.3808151317270885,
      "grad_norm": 1.3557251691818237,
      "learning_rate": 9.138170927443181e-05,
      "loss": 1.7576,
      "step": 2114
    },
    {
      "epoch": 0.3809952713352848,
      "grad_norm": 1.453099250793457,
      "learning_rate": 9.137375920546992e-05,
      "loss": 2.0789,
      "step": 2115
    },
    {
      "epoch": 0.3811754109434812,
      "grad_norm": 1.2727937698364258,
      "learning_rate": 9.136580581752522e-05,
      "loss": 1.9015,
      "step": 2116
    },
    {
      "epoch": 0.3813555505516775,
      "grad_norm": 1.2283263206481934,
      "learning_rate": 9.135784911123571e-05,
      "loss": 1.6213,
      "step": 2117
    },
    {
      "epoch": 0.3815356901598739,
      "grad_norm": 1.3137099742889404,
      "learning_rate": 9.13498890872397e-05,
      "loss": 1.7475,
      "step": 2118
    },
    {
      "epoch": 0.3817158297680703,
      "grad_norm": 1.197319746017456,
      "learning_rate": 9.134192574617573e-05,
      "loss": 1.4516,
      "step": 2119
    },
    {
      "epoch": 0.38189596937626663,
      "grad_norm": 1.2807143926620483,
      "learning_rate": 9.13339590886826e-05,
      "loss": 1.5196,
      "step": 2120
    },
    {
      "epoch": 0.382076108984463,
      "grad_norm": 1.1984219551086426,
      "learning_rate": 9.132598911539942e-05,
      "loss": 1.6206,
      "step": 2121
    },
    {
      "epoch": 0.38225624859265933,
      "grad_norm": 1.1745525598526,
      "learning_rate": 9.131801582696553e-05,
      "loss": 1.4356,
      "step": 2122
    },
    {
      "epoch": 0.3824363882008557,
      "grad_norm": 1.289275050163269,
      "learning_rate": 9.131003922402052e-05,
      "loss": 1.8689,
      "step": 2123
    },
    {
      "epoch": 0.38261652780905203,
      "grad_norm": 1.4010155200958252,
      "learning_rate": 9.130205930720429e-05,
      "loss": 1.8454,
      "step": 2124
    },
    {
      "epoch": 0.3827966674172484,
      "grad_norm": 1.3265488147735596,
      "learning_rate": 9.129407607715698e-05,
      "loss": 1.656,
      "step": 2125
    },
    {
      "epoch": 0.38297680702544473,
      "grad_norm": 1.2692373991012573,
      "learning_rate": 9.128608953451899e-05,
      "loss": 1.618,
      "step": 2126
    },
    {
      "epoch": 0.3831569466336411,
      "grad_norm": 1.3315633535385132,
      "learning_rate": 9.127809967993102e-05,
      "loss": 1.6082,
      "step": 2127
    },
    {
      "epoch": 0.38333708624183743,
      "grad_norm": 1.2551548480987549,
      "learning_rate": 9.1270106514034e-05,
      "loss": 1.5285,
      "step": 2128
    },
    {
      "epoch": 0.3835172258500338,
      "grad_norm": 1.3189990520477295,
      "learning_rate": 9.126211003746913e-05,
      "loss": 1.6632,
      "step": 2129
    },
    {
      "epoch": 0.38369736545823013,
      "grad_norm": 1.2506258487701416,
      "learning_rate": 9.125411025087789e-05,
      "loss": 1.4936,
      "step": 2130
    },
    {
      "epoch": 0.3838775050664265,
      "grad_norm": 1.243418574333191,
      "learning_rate": 9.124610715490203e-05,
      "loss": 1.8767,
      "step": 2131
    },
    {
      "epoch": 0.38405764467462283,
      "grad_norm": 1.2705342769622803,
      "learning_rate": 9.123810075018354e-05,
      "loss": 1.6632,
      "step": 2132
    },
    {
      "epoch": 0.3842377842828192,
      "grad_norm": 1.2149455547332764,
      "learning_rate": 9.123009103736468e-05,
      "loss": 1.7818,
      "step": 2133
    },
    {
      "epoch": 0.38441792389101553,
      "grad_norm": 1.465367317199707,
      "learning_rate": 9.122207801708802e-05,
      "loss": 1.6318,
      "step": 2134
    },
    {
      "epoch": 0.3845980634992119,
      "grad_norm": 1.3344510793685913,
      "learning_rate": 9.121406168999633e-05,
      "loss": 1.7328,
      "step": 2135
    },
    {
      "epoch": 0.38477820310740823,
      "grad_norm": 1.2665818929672241,
      "learning_rate": 9.120604205673268e-05,
      "loss": 1.6465,
      "step": 2136
    },
    {
      "epoch": 0.3849583427156046,
      "grad_norm": 1.3166686296463013,
      "learning_rate": 9.119801911794043e-05,
      "loss": 1.4931,
      "step": 2137
    },
    {
      "epoch": 0.38513848232380093,
      "grad_norm": 1.3134535551071167,
      "learning_rate": 9.118999287426314e-05,
      "loss": 1.7463,
      "step": 2138
    },
    {
      "epoch": 0.3853186219319973,
      "grad_norm": 1.2184779644012451,
      "learning_rate": 9.118196332634467e-05,
      "loss": 1.4493,
      "step": 2139
    },
    {
      "epoch": 0.38549876154019364,
      "grad_norm": 1.2748123407363892,
      "learning_rate": 9.117393047482919e-05,
      "loss": 1.7025,
      "step": 2140
    },
    {
      "epoch": 0.38567890114839,
      "grad_norm": 1.312063217163086,
      "learning_rate": 9.116589432036104e-05,
      "loss": 1.6307,
      "step": 2141
    },
    {
      "epoch": 0.38585904075658634,
      "grad_norm": 1.442418098449707,
      "learning_rate": 9.11578548635849e-05,
      "loss": 1.8966,
      "step": 2142
    },
    {
      "epoch": 0.3860391803647827,
      "grad_norm": 1.4317374229431152,
      "learning_rate": 9.11498121051457e-05,
      "loss": 1.7603,
      "step": 2143
    },
    {
      "epoch": 0.38621931997297904,
      "grad_norm": 1.4041088819503784,
      "learning_rate": 9.114176604568861e-05,
      "loss": 1.8934,
      "step": 2144
    },
    {
      "epoch": 0.3863994595811754,
      "grad_norm": 1.3317939043045044,
      "learning_rate": 9.113371668585908e-05,
      "loss": 1.6703,
      "step": 2145
    },
    {
      "epoch": 0.38657959918937174,
      "grad_norm": 1.3929203748703003,
      "learning_rate": 9.112566402630285e-05,
      "loss": 1.7624,
      "step": 2146
    },
    {
      "epoch": 0.38675973879756814,
      "grad_norm": 1.4137219190597534,
      "learning_rate": 9.111760806766587e-05,
      "loss": 1.7475,
      "step": 2147
    },
    {
      "epoch": 0.3869398784057645,
      "grad_norm": 1.4739933013916016,
      "learning_rate": 9.110954881059439e-05,
      "loss": 1.7192,
      "step": 2148
    },
    {
      "epoch": 0.38712001801396084,
      "grad_norm": 1.3204176425933838,
      "learning_rate": 9.110148625573496e-05,
      "loss": 1.6008,
      "step": 2149
    },
    {
      "epoch": 0.3873001576221572,
      "grad_norm": 1.271134614944458,
      "learning_rate": 9.10934204037343e-05,
      "loss": 1.5157,
      "step": 2150
    },
    {
      "epoch": 0.38748029723035354,
      "grad_norm": 1.3271753787994385,
      "learning_rate": 9.108535125523947e-05,
      "loss": 2.2178,
      "step": 2151
    },
    {
      "epoch": 0.3876604368385499,
      "grad_norm": 3.78749680519104,
      "learning_rate": 9.107727881089779e-05,
      "loss": 2.047,
      "step": 2152
    },
    {
      "epoch": 0.38784057644674624,
      "grad_norm": 1.2255138158798218,
      "learning_rate": 9.106920307135679e-05,
      "loss": 2.1348,
      "step": 2153
    },
    {
      "epoch": 0.3880207160549426,
      "grad_norm": 1.3508459329605103,
      "learning_rate": 9.106112403726435e-05,
      "loss": 2.129,
      "step": 2154
    },
    {
      "epoch": 0.38820085566313894,
      "grad_norm": 1.3681482076644897,
      "learning_rate": 9.105304170926851e-05,
      "loss": 2.1737,
      "step": 2155
    },
    {
      "epoch": 0.3883809952713353,
      "grad_norm": 1.2568097114562988,
      "learning_rate": 9.104495608801768e-05,
      "loss": 2.089,
      "step": 2156
    },
    {
      "epoch": 0.38856113487953164,
      "grad_norm": 1.3417366743087769,
      "learning_rate": 9.103686717416046e-05,
      "loss": 2.0041,
      "step": 2157
    },
    {
      "epoch": 0.388741274487728,
      "grad_norm": 1.3027949333190918,
      "learning_rate": 9.102877496834575e-05,
      "loss": 1.9694,
      "step": 2158
    },
    {
      "epoch": 0.38892141409592434,
      "grad_norm": 1.4856226444244385,
      "learning_rate": 9.102067947122269e-05,
      "loss": 2.1128,
      "step": 2159
    },
    {
      "epoch": 0.3891015537041207,
      "grad_norm": 1.7984919548034668,
      "learning_rate": 9.101258068344069e-05,
      "loss": 2.2167,
      "step": 2160
    },
    {
      "epoch": 0.38928169331231705,
      "grad_norm": 3.168741464614868,
      "learning_rate": 9.100447860564945e-05,
      "loss": 2.1743,
      "step": 2161
    },
    {
      "epoch": 0.3894618329205134,
      "grad_norm": 1.22146737575531,
      "learning_rate": 9.099637323849893e-05,
      "loss": 1.6083,
      "step": 2162
    },
    {
      "epoch": 0.38964197252870975,
      "grad_norm": 1.1351044178009033,
      "learning_rate": 9.098826458263929e-05,
      "loss": 1.5818,
      "step": 2163
    },
    {
      "epoch": 0.3898221121369061,
      "grad_norm": 1.2672655582427979,
      "learning_rate": 9.098015263872104e-05,
      "loss": 1.7528,
      "step": 2164
    },
    {
      "epoch": 0.39000225174510245,
      "grad_norm": 1.3483572006225586,
      "learning_rate": 9.097203740739491e-05,
      "loss": 1.7113,
      "step": 2165
    },
    {
      "epoch": 0.3901823913532988,
      "grad_norm": 1.204759955406189,
      "learning_rate": 9.09639188893119e-05,
      "loss": 1.5575,
      "step": 2166
    },
    {
      "epoch": 0.39036253096149515,
      "grad_norm": 1.3749037981033325,
      "learning_rate": 9.095579708512327e-05,
      "loss": 1.6415,
      "step": 2167
    },
    {
      "epoch": 0.3905426705696915,
      "grad_norm": 1.3593950271606445,
      "learning_rate": 9.094767199548054e-05,
      "loss": 1.7345,
      "step": 2168
    },
    {
      "epoch": 0.39072281017788785,
      "grad_norm": 1.252854824066162,
      "learning_rate": 9.093954362103551e-05,
      "loss": 1.6888,
      "step": 2169
    },
    {
      "epoch": 0.3909029497860842,
      "grad_norm": 1.2474712133407593,
      "learning_rate": 9.093141196244022e-05,
      "loss": 1.4746,
      "step": 2170
    },
    {
      "epoch": 0.39108308939428055,
      "grad_norm": 1.3356012105941772,
      "learning_rate": 9.092327702034702e-05,
      "loss": 1.7455,
      "step": 2171
    },
    {
      "epoch": 0.3912632290024769,
      "grad_norm": 1.4254230260849,
      "learning_rate": 9.091513879540845e-05,
      "loss": 1.6512,
      "step": 2172
    },
    {
      "epoch": 0.39144336861067325,
      "grad_norm": 1.3403728008270264,
      "learning_rate": 9.090699728827738e-05,
      "loss": 1.7819,
      "step": 2173
    },
    {
      "epoch": 0.3916235082188696,
      "grad_norm": 1.3594458103179932,
      "learning_rate": 9.089885249960693e-05,
      "loss": 1.8249,
      "step": 2174
    },
    {
      "epoch": 0.39180364782706595,
      "grad_norm": 1.326656460762024,
      "learning_rate": 9.089070443005044e-05,
      "loss": 1.9626,
      "step": 2175
    },
    {
      "epoch": 0.39198378743526235,
      "grad_norm": 1.3090012073516846,
      "learning_rate": 9.088255308026156e-05,
      "loss": 1.7413,
      "step": 2176
    },
    {
      "epoch": 0.3921639270434587,
      "grad_norm": 1.1989049911499023,
      "learning_rate": 9.087439845089418e-05,
      "loss": 1.5912,
      "step": 2177
    },
    {
      "epoch": 0.39234406665165505,
      "grad_norm": 1.2044060230255127,
      "learning_rate": 9.086624054260247e-05,
      "loss": 1.6615,
      "step": 2178
    },
    {
      "epoch": 0.3925242062598514,
      "grad_norm": 1.2591339349746704,
      "learning_rate": 9.085807935604085e-05,
      "loss": 1.5324,
      "step": 2179
    },
    {
      "epoch": 0.39270434586804776,
      "grad_norm": 1.2562506198883057,
      "learning_rate": 9.0849914891864e-05,
      "loss": 1.6469,
      "step": 2180
    },
    {
      "epoch": 0.3928844854762441,
      "grad_norm": 1.2304260730743408,
      "learning_rate": 9.084174715072689e-05,
      "loss": 1.7989,
      "step": 2181
    },
    {
      "epoch": 0.39306462508444046,
      "grad_norm": 1.3507219552993774,
      "learning_rate": 9.08335761332847e-05,
      "loss": 1.7477,
      "step": 2182
    },
    {
      "epoch": 0.3932447646926368,
      "grad_norm": 1.3541796207427979,
      "learning_rate": 9.082540184019293e-05,
      "loss": 1.6694,
      "step": 2183
    },
    {
      "epoch": 0.39342490430083316,
      "grad_norm": 1.3877840042114258,
      "learning_rate": 9.081722427210731e-05,
      "loss": 1.787,
      "step": 2184
    },
    {
      "epoch": 0.3936050439090295,
      "grad_norm": 1.205797791481018,
      "learning_rate": 9.080904342968384e-05,
      "loss": 1.6819,
      "step": 2185
    },
    {
      "epoch": 0.39378518351722586,
      "grad_norm": 1.3570153713226318,
      "learning_rate": 9.080085931357877e-05,
      "loss": 1.7425,
      "step": 2186
    },
    {
      "epoch": 0.3939653231254222,
      "grad_norm": 1.3105854988098145,
      "learning_rate": 9.079267192444865e-05,
      "loss": 1.8101,
      "step": 2187
    },
    {
      "epoch": 0.39414546273361856,
      "grad_norm": 1.3009496927261353,
      "learning_rate": 9.078448126295026e-05,
      "loss": 1.5877,
      "step": 2188
    },
    {
      "epoch": 0.3943256023418149,
      "grad_norm": 1.3271325826644897,
      "learning_rate": 9.077628732974065e-05,
      "loss": 1.9116,
      "step": 2189
    },
    {
      "epoch": 0.39450574195001126,
      "grad_norm": 1.250184178352356,
      "learning_rate": 9.076809012547712e-05,
      "loss": 1.6684,
      "step": 2190
    },
    {
      "epoch": 0.3946858815582076,
      "grad_norm": 1.4337053298950195,
      "learning_rate": 9.075988965081728e-05,
      "loss": 2.0014,
      "step": 2191
    },
    {
      "epoch": 0.39486602116640396,
      "grad_norm": 1.456337332725525,
      "learning_rate": 9.075168590641891e-05,
      "loss": 1.8061,
      "step": 2192
    },
    {
      "epoch": 0.3950461607746003,
      "grad_norm": 1.2794229984283447,
      "learning_rate": 9.074347889294016e-05,
      "loss": 1.6618,
      "step": 2193
    },
    {
      "epoch": 0.39522630038279666,
      "grad_norm": 1.2905025482177734,
      "learning_rate": 9.073526861103941e-05,
      "loss": 1.6935,
      "step": 2194
    },
    {
      "epoch": 0.395406439990993,
      "grad_norm": 1.2809932231903076,
      "learning_rate": 9.072705506137522e-05,
      "loss": 1.6264,
      "step": 2195
    },
    {
      "epoch": 0.39558657959918936,
      "grad_norm": 1.3347852230072021,
      "learning_rate": 9.071883824460652e-05,
      "loss": 1.6993,
      "step": 2196
    },
    {
      "epoch": 0.3957667192073857,
      "grad_norm": 1.4605298042297363,
      "learning_rate": 9.071061816139245e-05,
      "loss": 1.7853,
      "step": 2197
    },
    {
      "epoch": 0.39594685881558206,
      "grad_norm": 1.5141929388046265,
      "learning_rate": 9.070239481239243e-05,
      "loss": 1.9662,
      "step": 2198
    },
    {
      "epoch": 0.3961269984237784,
      "grad_norm": 1.4083324670791626,
      "learning_rate": 9.06941681982661e-05,
      "loss": 1.5646,
      "step": 2199
    },
    {
      "epoch": 0.39630713803197476,
      "grad_norm": 1.2718425989151,
      "learning_rate": 9.068593831967344e-05,
      "loss": 1.4153,
      "step": 2200
    },
    {
      "epoch": 0.3964872776401711,
      "grad_norm": 1.4546648263931274,
      "learning_rate": 9.06777051772746e-05,
      "loss": 2.1496,
      "step": 2201
    },
    {
      "epoch": 0.39666741724836746,
      "grad_norm": 1.33214271068573,
      "learning_rate": 9.066946877173009e-05,
      "loss": 2.3049,
      "step": 2202
    },
    {
      "epoch": 0.3968475568565638,
      "grad_norm": 1.179665207862854,
      "learning_rate": 9.06612291037006e-05,
      "loss": 1.8641,
      "step": 2203
    },
    {
      "epoch": 0.3970276964647602,
      "grad_norm": 1.496773362159729,
      "learning_rate": 9.065298617384712e-05,
      "loss": 2.2468,
      "step": 2204
    },
    {
      "epoch": 0.39720783607295657,
      "grad_norm": 1.2146224975585938,
      "learning_rate": 9.064473998283088e-05,
      "loss": 1.9496,
      "step": 2205
    },
    {
      "epoch": 0.3973879756811529,
      "grad_norm": 1.3933171033859253,
      "learning_rate": 9.06364905313134e-05,
      "loss": 1.9137,
      "step": 2206
    },
    {
      "epoch": 0.39756811528934927,
      "grad_norm": 1.5050772428512573,
      "learning_rate": 9.062823781995644e-05,
      "loss": 2.2607,
      "step": 2207
    },
    {
      "epoch": 0.3977482548975456,
      "grad_norm": 2.351269006729126,
      "learning_rate": 9.061998184942203e-05,
      "loss": 2.3468,
      "step": 2208
    },
    {
      "epoch": 0.39792839450574197,
      "grad_norm": 1.691042423248291,
      "learning_rate": 9.061172262037247e-05,
      "loss": 2.0413,
      "step": 2209
    },
    {
      "epoch": 0.3981085341139383,
      "grad_norm": 1.6214369535446167,
      "learning_rate": 9.060346013347029e-05,
      "loss": 1.9874,
      "step": 2210
    },
    {
      "epoch": 0.39828867372213467,
      "grad_norm": 1.6569087505340576,
      "learning_rate": 9.059519438937833e-05,
      "loss": 2.0923,
      "step": 2211
    },
    {
      "epoch": 0.398468813330331,
      "grad_norm": 1.254485011100769,
      "learning_rate": 9.058692538875964e-05,
      "loss": 1.564,
      "step": 2212
    },
    {
      "epoch": 0.39864895293852737,
      "grad_norm": 1.2168151140213013,
      "learning_rate": 9.057865313227756e-05,
      "loss": 1.6159,
      "step": 2213
    },
    {
      "epoch": 0.3988290925467237,
      "grad_norm": 1.2943649291992188,
      "learning_rate": 9.05703776205957e-05,
      "loss": 1.8458,
      "step": 2214
    },
    {
      "epoch": 0.39900923215492007,
      "grad_norm": 1.3149343729019165,
      "learning_rate": 9.05620988543779e-05,
      "loss": 1.6536,
      "step": 2215
    },
    {
      "epoch": 0.3991893717631164,
      "grad_norm": 1.247033953666687,
      "learning_rate": 9.055381683428829e-05,
      "loss": 1.6111,
      "step": 2216
    },
    {
      "epoch": 0.39936951137131277,
      "grad_norm": 1.3952230215072632,
      "learning_rate": 9.054553156099126e-05,
      "loss": 1.7637,
      "step": 2217
    },
    {
      "epoch": 0.3995496509795091,
      "grad_norm": 1.394584059715271,
      "learning_rate": 9.053724303515142e-05,
      "loss": 1.9503,
      "step": 2218
    },
    {
      "epoch": 0.39972979058770547,
      "grad_norm": 1.1841211318969727,
      "learning_rate": 9.05289512574337e-05,
      "loss": 1.6442,
      "step": 2219
    },
    {
      "epoch": 0.3999099301959018,
      "grad_norm": 1.3564656972885132,
      "learning_rate": 9.052065622850324e-05,
      "loss": 1.6907,
      "step": 2220
    },
    {
      "epoch": 0.40009006980409817,
      "grad_norm": 1.2043156623840332,
      "learning_rate": 9.051235794902549e-05,
      "loss": 1.4155,
      "step": 2221
    },
    {
      "epoch": 0.4002702094122945,
      "grad_norm": 1.3666579723358154,
      "learning_rate": 9.050405641966611e-05,
      "loss": 1.757,
      "step": 2222
    },
    {
      "epoch": 0.40045034902049087,
      "grad_norm": 1.2404276132583618,
      "learning_rate": 9.049575164109104e-05,
      "loss": 1.7984,
      "step": 2223
    },
    {
      "epoch": 0.4006304886286872,
      "grad_norm": 1.263825535774231,
      "learning_rate": 9.04874436139665e-05,
      "loss": 1.8548,
      "step": 2224
    },
    {
      "epoch": 0.40081062823688357,
      "grad_norm": 1.2468438148498535,
      "learning_rate": 9.047913233895898e-05,
      "loss": 1.7792,
      "step": 2225
    },
    {
      "epoch": 0.4009907678450799,
      "grad_norm": 1.3407175540924072,
      "learning_rate": 9.047081781673517e-05,
      "loss": 1.9603,
      "step": 2226
    },
    {
      "epoch": 0.40117090745327627,
      "grad_norm": 1.2416698932647705,
      "learning_rate": 9.046250004796205e-05,
      "loss": 1.5941,
      "step": 2227
    },
    {
      "epoch": 0.4013510470614726,
      "grad_norm": 1.1872090101242065,
      "learning_rate": 9.04541790333069e-05,
      "loss": 1.4569,
      "step": 2228
    },
    {
      "epoch": 0.40153118666966897,
      "grad_norm": 1.309704303741455,
      "learning_rate": 9.044585477343722e-05,
      "loss": 1.8411,
      "step": 2229
    },
    {
      "epoch": 0.4017113262778653,
      "grad_norm": 1.2670727968215942,
      "learning_rate": 9.043752726902075e-05,
      "loss": 1.6368,
      "step": 2230
    },
    {
      "epoch": 0.40189146588606167,
      "grad_norm": 1.3420205116271973,
      "learning_rate": 9.042919652072554e-05,
      "loss": 1.8833,
      "step": 2231
    },
    {
      "epoch": 0.4020716054942581,
      "grad_norm": 1.4256051778793335,
      "learning_rate": 9.042086252921989e-05,
      "loss": 1.791,
      "step": 2232
    },
    {
      "epoch": 0.4022517451024544,
      "grad_norm": 1.324985384941101,
      "learning_rate": 9.041252529517232e-05,
      "loss": 1.8372,
      "step": 2233
    },
    {
      "epoch": 0.4024318847106508,
      "grad_norm": 1.19207763671875,
      "learning_rate": 9.040418481925168e-05,
      "loss": 1.464,
      "step": 2234
    },
    {
      "epoch": 0.40261202431884713,
      "grad_norm": 1.420929193496704,
      "learning_rate": 9.0395841102127e-05,
      "loss": 1.8441,
      "step": 2235
    },
    {
      "epoch": 0.4027921639270435,
      "grad_norm": 1.3216580152511597,
      "learning_rate": 9.038749414446763e-05,
      "loss": 1.7486,
      "step": 2236
    },
    {
      "epoch": 0.40297230353523983,
      "grad_norm": 1.4530539512634277,
      "learning_rate": 9.037914394694313e-05,
      "loss": 1.7013,
      "step": 2237
    },
    {
      "epoch": 0.4031524431434362,
      "grad_norm": 1.3988749980926514,
      "learning_rate": 9.037079051022339e-05,
      "loss": 1.6117,
      "step": 2238
    },
    {
      "epoch": 0.40333258275163253,
      "grad_norm": 1.1622987985610962,
      "learning_rate": 9.03624338349785e-05,
      "loss": 1.5733,
      "step": 2239
    },
    {
      "epoch": 0.4035127223598289,
      "grad_norm": 1.2403991222381592,
      "learning_rate": 9.035407392187883e-05,
      "loss": 1.7356,
      "step": 2240
    },
    {
      "epoch": 0.40369286196802523,
      "grad_norm": 1.3406028747558594,
      "learning_rate": 9.0345710771595e-05,
      "loss": 1.6973,
      "step": 2241
    },
    {
      "epoch": 0.4038730015762216,
      "grad_norm": 1.459892988204956,
      "learning_rate": 9.033734438479789e-05,
      "loss": 1.7797,
      "step": 2242
    },
    {
      "epoch": 0.40405314118441793,
      "grad_norm": 1.3988192081451416,
      "learning_rate": 9.032897476215866e-05,
      "loss": 1.9643,
      "step": 2243
    },
    {
      "epoch": 0.4042332807926143,
      "grad_norm": 1.5672204494476318,
      "learning_rate": 9.032060190434874e-05,
      "loss": 1.7598,
      "step": 2244
    },
    {
      "epoch": 0.40441342040081063,
      "grad_norm": 1.5195800065994263,
      "learning_rate": 9.031222581203977e-05,
      "loss": 2.0137,
      "step": 2245
    },
    {
      "epoch": 0.404593560009007,
      "grad_norm": 1.3430835008621216,
      "learning_rate": 9.030384648590369e-05,
      "loss": 1.6662,
      "step": 2246
    },
    {
      "epoch": 0.40477369961720333,
      "grad_norm": 1.395531415939331,
      "learning_rate": 9.029546392661265e-05,
      "loss": 1.619,
      "step": 2247
    },
    {
      "epoch": 0.4049538392253997,
      "grad_norm": 1.475770115852356,
      "learning_rate": 9.028707813483913e-05,
      "loss": 1.7312,
      "step": 2248
    },
    {
      "epoch": 0.40513397883359603,
      "grad_norm": 1.3561451435089111,
      "learning_rate": 9.027868911125583e-05,
      "loss": 1.6794,
      "step": 2249
    },
    {
      "epoch": 0.4053141184417924,
      "grad_norm": 1.49905526638031,
      "learning_rate": 9.02702968565357e-05,
      "loss": 1.5964,
      "step": 2250
    },
    {
      "epoch": 0.40549425804998873,
      "grad_norm": 1.259384036064148,
      "learning_rate": 9.0261901371352e-05,
      "loss": 1.9782,
      "step": 2251
    },
    {
      "epoch": 0.4056743976581851,
      "grad_norm": 1.2289361953735352,
      "learning_rate": 9.025350265637815e-05,
      "loss": 2.0043,
      "step": 2252
    },
    {
      "epoch": 0.40585453726638143,
      "grad_norm": 1.1787949800491333,
      "learning_rate": 9.024510071228793e-05,
      "loss": 1.962,
      "step": 2253
    },
    {
      "epoch": 0.4060346768745778,
      "grad_norm": 1.4196516275405884,
      "learning_rate": 9.023669553975534e-05,
      "loss": 2.4036,
      "step": 2254
    },
    {
      "epoch": 0.40621481648277413,
      "grad_norm": 1.3713552951812744,
      "learning_rate": 9.022828713945463e-05,
      "loss": 2.1738,
      "step": 2255
    },
    {
      "epoch": 0.4063949560909705,
      "grad_norm": 1.5663354396820068,
      "learning_rate": 9.021987551206032e-05,
      "loss": 2.2402,
      "step": 2256
    },
    {
      "epoch": 0.40657509569916683,
      "grad_norm": 1.4110584259033203,
      "learning_rate": 9.021146065824719e-05,
      "loss": 2.1178,
      "step": 2257
    },
    {
      "epoch": 0.4067552353073632,
      "grad_norm": 1.68704092502594,
      "learning_rate": 9.020304257869028e-05,
      "loss": 2.3081,
      "step": 2258
    },
    {
      "epoch": 0.40693537491555953,
      "grad_norm": 1.7063883543014526,
      "learning_rate": 9.019462127406487e-05,
      "loss": 2.3804,
      "step": 2259
    },
    {
      "epoch": 0.4071155145237559,
      "grad_norm": 1.3695518970489502,
      "learning_rate": 9.018619674504652e-05,
      "loss": 1.8772,
      "step": 2260
    },
    {
      "epoch": 0.4072956541319523,
      "grad_norm": 1.1992802619934082,
      "learning_rate": 9.017776899231103e-05,
      "loss": 1.5364,
      "step": 2261
    },
    {
      "epoch": 0.40747579374014864,
      "grad_norm": 1.2105752229690552,
      "learning_rate": 9.016933801653449e-05,
      "loss": 1.5297,
      "step": 2262
    },
    {
      "epoch": 0.407655933348345,
      "grad_norm": 1.304094672203064,
      "learning_rate": 9.016090381839322e-05,
      "loss": 1.5659,
      "step": 2263
    },
    {
      "epoch": 0.40783607295654134,
      "grad_norm": 1.4258983135223389,
      "learning_rate": 9.015246639856381e-05,
      "loss": 2.0664,
      "step": 2264
    },
    {
      "epoch": 0.4080162125647377,
      "grad_norm": 1.2939250469207764,
      "learning_rate": 9.01440257577231e-05,
      "loss": 1.6481,
      "step": 2265
    },
    {
      "epoch": 0.40819635217293404,
      "grad_norm": 1.3380907773971558,
      "learning_rate": 9.013558189654819e-05,
      "loss": 1.7788,
      "step": 2266
    },
    {
      "epoch": 0.4083764917811304,
      "grad_norm": 1.2485278844833374,
      "learning_rate": 9.012713481571645e-05,
      "loss": 1.5974,
      "step": 2267
    },
    {
      "epoch": 0.40855663138932674,
      "grad_norm": 1.2398656606674194,
      "learning_rate": 9.01186845159055e-05,
      "loss": 1.6576,
      "step": 2268
    },
    {
      "epoch": 0.4087367709975231,
      "grad_norm": 1.2962778806686401,
      "learning_rate": 9.011023099779323e-05,
      "loss": 1.6924,
      "step": 2269
    },
    {
      "epoch": 0.40891691060571944,
      "grad_norm": 1.2225233316421509,
      "learning_rate": 9.010177426205777e-05,
      "loss": 1.5986,
      "step": 2270
    },
    {
      "epoch": 0.4090970502139158,
      "grad_norm": 1.1218363046646118,
      "learning_rate": 9.009331430937751e-05,
      "loss": 1.4091,
      "step": 2271
    },
    {
      "epoch": 0.40927718982211214,
      "grad_norm": 1.1273895502090454,
      "learning_rate": 9.008485114043108e-05,
      "loss": 1.475,
      "step": 2272
    },
    {
      "epoch": 0.4094573294303085,
      "grad_norm": 1.3381619453430176,
      "learning_rate": 9.007638475589745e-05,
      "loss": 1.6634,
      "step": 2273
    },
    {
      "epoch": 0.40963746903850484,
      "grad_norm": 1.352177619934082,
      "learning_rate": 9.006791515645573e-05,
      "loss": 1.7489,
      "step": 2274
    },
    {
      "epoch": 0.4098176086467012,
      "grad_norm": 1.398384690284729,
      "learning_rate": 9.005944234278538e-05,
      "loss": 1.7822,
      "step": 2275
    },
    {
      "epoch": 0.40999774825489754,
      "grad_norm": 1.241804599761963,
      "learning_rate": 9.005096631556609e-05,
      "loss": 1.4318,
      "step": 2276
    },
    {
      "epoch": 0.4101778878630939,
      "grad_norm": 1.1617920398712158,
      "learning_rate": 9.004248707547777e-05,
      "loss": 1.4568,
      "step": 2277
    },
    {
      "epoch": 0.41035802747129024,
      "grad_norm": 1.2524787187576294,
      "learning_rate": 9.003400462320067e-05,
      "loss": 1.358,
      "step": 2278
    },
    {
      "epoch": 0.4105381670794866,
      "grad_norm": 1.269258737564087,
      "learning_rate": 9.002551895941519e-05,
      "loss": 1.8369,
      "step": 2279
    },
    {
      "epoch": 0.41071830668768294,
      "grad_norm": 1.2870339155197144,
      "learning_rate": 9.001703008480208e-05,
      "loss": 1.7499,
      "step": 2280
    },
    {
      "epoch": 0.4108984462958793,
      "grad_norm": 1.3752247095108032,
      "learning_rate": 9.000853800004233e-05,
      "loss": 1.6058,
      "step": 2281
    },
    {
      "epoch": 0.41107858590407564,
      "grad_norm": 1.4035537242889404,
      "learning_rate": 9.000004270581712e-05,
      "loss": 1.8562,
      "step": 2282
    },
    {
      "epoch": 0.411258725512272,
      "grad_norm": 1.258719563484192,
      "learning_rate": 8.999154420280798e-05,
      "loss": 1.6019,
      "step": 2283
    },
    {
      "epoch": 0.41143886512046834,
      "grad_norm": 1.2502412796020508,
      "learning_rate": 8.998304249169664e-05,
      "loss": 1.4756,
      "step": 2284
    },
    {
      "epoch": 0.4116190047286647,
      "grad_norm": 1.4311153888702393,
      "learning_rate": 8.997453757316512e-05,
      "loss": 1.6999,
      "step": 2285
    },
    {
      "epoch": 0.41179914433686104,
      "grad_norm": 1.5594477653503418,
      "learning_rate": 8.996602944789565e-05,
      "loss": 2.0204,
      "step": 2286
    },
    {
      "epoch": 0.4119792839450574,
      "grad_norm": 1.323693871498108,
      "learning_rate": 8.995751811657077e-05,
      "loss": 1.7253,
      "step": 2287
    },
    {
      "epoch": 0.41215942355325375,
      "grad_norm": 1.3957284688949585,
      "learning_rate": 8.994900357987326e-05,
      "loss": 1.6376,
      "step": 2288
    },
    {
      "epoch": 0.41233956316145015,
      "grad_norm": 1.3958390951156616,
      "learning_rate": 8.994048583848614e-05,
      "loss": 1.6135,
      "step": 2289
    },
    {
      "epoch": 0.4125197027696465,
      "grad_norm": 1.339706540107727,
      "learning_rate": 8.993196489309267e-05,
      "loss": 1.4103,
      "step": 2290
    },
    {
      "epoch": 0.41269984237784285,
      "grad_norm": 1.302673578262329,
      "learning_rate": 8.992344074437645e-05,
      "loss": 1.6156,
      "step": 2291
    },
    {
      "epoch": 0.4128799819860392,
      "grad_norm": 1.3750134706497192,
      "learning_rate": 8.991491339302126e-05,
      "loss": 1.8009,
      "step": 2292
    },
    {
      "epoch": 0.41306012159423555,
      "grad_norm": 1.3975346088409424,
      "learning_rate": 8.990638283971116e-05,
      "loss": 1.7812,
      "step": 2293
    },
    {
      "epoch": 0.4132402612024319,
      "grad_norm": 1.4291768074035645,
      "learning_rate": 8.989784908513046e-05,
      "loss": 1.701,
      "step": 2294
    },
    {
      "epoch": 0.41342040081062825,
      "grad_norm": 1.4330488443374634,
      "learning_rate": 8.988931212996374e-05,
      "loss": 1.9773,
      "step": 2295
    },
    {
      "epoch": 0.4136005404188246,
      "grad_norm": 1.3990591764450073,
      "learning_rate": 8.988077197489583e-05,
      "loss": 1.6276,
      "step": 2296
    },
    {
      "epoch": 0.41378068002702095,
      "grad_norm": 1.2557814121246338,
      "learning_rate": 8.987222862061182e-05,
      "loss": 1.3885,
      "step": 2297
    },
    {
      "epoch": 0.4139608196352173,
      "grad_norm": 1.2447737455368042,
      "learning_rate": 8.986368206779706e-05,
      "loss": 1.4814,
      "step": 2298
    },
    {
      "epoch": 0.41414095924341365,
      "grad_norm": 1.497323989868164,
      "learning_rate": 8.985513231713713e-05,
      "loss": 1.9769,
      "step": 2299
    },
    {
      "epoch": 0.41432109885161,
      "grad_norm": 1.3988481760025024,
      "learning_rate": 8.984657936931791e-05,
      "loss": 2.0068,
      "step": 2300
    },
    {
      "epoch": 0.41450123845980635,
      "grad_norm": 1.3377095460891724,
      "learning_rate": 8.983802322502548e-05,
      "loss": 2.2395,
      "step": 2301
    },
    {
      "epoch": 0.4146813780680027,
      "grad_norm": 1.3252151012420654,
      "learning_rate": 8.982946388494625e-05,
      "loss": 2.2211,
      "step": 2302
    },
    {
      "epoch": 0.41486151767619905,
      "grad_norm": 1.776610016822815,
      "learning_rate": 8.982090134976683e-05,
      "loss": 2.2165,
      "step": 2303
    },
    {
      "epoch": 0.4150416572843954,
      "grad_norm": 1.233462929725647,
      "learning_rate": 8.981233562017408e-05,
      "loss": 1.9882,
      "step": 2304
    },
    {
      "epoch": 0.41522179689259175,
      "grad_norm": 1.1811257600784302,
      "learning_rate": 8.980376669685517e-05,
      "loss": 1.8871,
      "step": 2305
    },
    {
      "epoch": 0.4154019365007881,
      "grad_norm": 2.043971538543701,
      "learning_rate": 8.97951945804975e-05,
      "loss": 2.0592,
      "step": 2306
    },
    {
      "epoch": 0.41558207610898446,
      "grad_norm": 1.7642455101013184,
      "learning_rate": 8.978661927178867e-05,
      "loss": 1.9924,
      "step": 2307
    },
    {
      "epoch": 0.4157622157171808,
      "grad_norm": 1.3262183666229248,
      "learning_rate": 8.977804077141665e-05,
      "loss": 2.1365,
      "step": 2308
    },
    {
      "epoch": 0.41594235532537716,
      "grad_norm": 1.703677773475647,
      "learning_rate": 8.976945908006956e-05,
      "loss": 1.981,
      "step": 2309
    },
    {
      "epoch": 0.4161224949335735,
      "grad_norm": 1.7459417581558228,
      "learning_rate": 8.976087419843584e-05,
      "loss": 2.5611,
      "step": 2310
    },
    {
      "epoch": 0.41630263454176986,
      "grad_norm": 1.5708870887756348,
      "learning_rate": 8.975228612720416e-05,
      "loss": 1.8458,
      "step": 2311
    },
    {
      "epoch": 0.4164827741499662,
      "grad_norm": 1.2064553499221802,
      "learning_rate": 8.974369486706345e-05,
      "loss": 1.7064,
      "step": 2312
    },
    {
      "epoch": 0.41666291375816256,
      "grad_norm": 1.3049978017807007,
      "learning_rate": 8.973510041870287e-05,
      "loss": 1.6698,
      "step": 2313
    },
    {
      "epoch": 0.4168430533663589,
      "grad_norm": 1.2189924716949463,
      "learning_rate": 8.972650278281191e-05,
      "loss": 1.7325,
      "step": 2314
    },
    {
      "epoch": 0.41702319297455526,
      "grad_norm": 1.329236626625061,
      "learning_rate": 8.971790196008025e-05,
      "loss": 1.5717,
      "step": 2315
    },
    {
      "epoch": 0.4172033325827516,
      "grad_norm": 1.280090093612671,
      "learning_rate": 8.970929795119784e-05,
      "loss": 1.868,
      "step": 2316
    },
    {
      "epoch": 0.41738347219094796,
      "grad_norm": 1.1924283504486084,
      "learning_rate": 8.97006907568549e-05,
      "loss": 1.5763,
      "step": 2317
    },
    {
      "epoch": 0.41756361179914436,
      "grad_norm": 1.3362699747085571,
      "learning_rate": 8.969208037774186e-05,
      "loss": 1.7364,
      "step": 2318
    },
    {
      "epoch": 0.4177437514073407,
      "grad_norm": 1.326858401298523,
      "learning_rate": 8.968346681454949e-05,
      "loss": 1.6928,
      "step": 2319
    },
    {
      "epoch": 0.41792389101553706,
      "grad_norm": 1.1357793807983398,
      "learning_rate": 8.967485006796872e-05,
      "loss": 1.6156,
      "step": 2320
    },
    {
      "epoch": 0.4181040306237334,
      "grad_norm": 1.3107483386993408,
      "learning_rate": 8.966623013869082e-05,
      "loss": 1.7734,
      "step": 2321
    },
    {
      "epoch": 0.41828417023192976,
      "grad_norm": 1.5365502834320068,
      "learning_rate": 8.965760702740725e-05,
      "loss": 1.71,
      "step": 2322
    },
    {
      "epoch": 0.4184643098401261,
      "grad_norm": 1.3306937217712402,
      "learning_rate": 8.964898073480976e-05,
      "loss": 1.4933,
      "step": 2323
    },
    {
      "epoch": 0.41864444944832246,
      "grad_norm": 1.3033932447433472,
      "learning_rate": 8.964035126159035e-05,
      "loss": 1.7984,
      "step": 2324
    },
    {
      "epoch": 0.4188245890565188,
      "grad_norm": 1.354055643081665,
      "learning_rate": 8.963171860844128e-05,
      "loss": 1.8457,
      "step": 2325
    },
    {
      "epoch": 0.41900472866471516,
      "grad_norm": 1.2634997367858887,
      "learning_rate": 8.962308277605502e-05,
      "loss": 1.8136,
      "step": 2326
    },
    {
      "epoch": 0.4191848682729115,
      "grad_norm": 1.3568263053894043,
      "learning_rate": 8.961444376512438e-05,
      "loss": 1.6783,
      "step": 2327
    },
    {
      "epoch": 0.41936500788110787,
      "grad_norm": 1.4491347074508667,
      "learning_rate": 8.960580157634235e-05,
      "loss": 1.9684,
      "step": 2328
    },
    {
      "epoch": 0.4195451474893042,
      "grad_norm": 1.267566442489624,
      "learning_rate": 8.959715621040221e-05,
      "loss": 1.5467,
      "step": 2329
    },
    {
      "epoch": 0.41972528709750057,
      "grad_norm": 1.183724284172058,
      "learning_rate": 8.95885076679975e-05,
      "loss": 1.7789,
      "step": 2330
    },
    {
      "epoch": 0.4199054267056969,
      "grad_norm": 1.364312767982483,
      "learning_rate": 8.957985594982197e-05,
      "loss": 1.7984,
      "step": 2331
    },
    {
      "epoch": 0.42008556631389327,
      "grad_norm": 1.3236796855926514,
      "learning_rate": 8.957120105656968e-05,
      "loss": 1.8618,
      "step": 2332
    },
    {
      "epoch": 0.4202657059220896,
      "grad_norm": 1.254014492034912,
      "learning_rate": 8.956254298893491e-05,
      "loss": 1.7109,
      "step": 2333
    },
    {
      "epoch": 0.42044584553028597,
      "grad_norm": 1.1174708604812622,
      "learning_rate": 8.955388174761222e-05,
      "loss": 1.3256,
      "step": 2334
    },
    {
      "epoch": 0.4206259851384823,
      "grad_norm": 1.2135961055755615,
      "learning_rate": 8.95452173332964e-05,
      "loss": 1.6955,
      "step": 2335
    },
    {
      "epoch": 0.42080612474667867,
      "grad_norm": 1.3413887023925781,
      "learning_rate": 8.953654974668251e-05,
      "loss": 1.7351,
      "step": 2336
    },
    {
      "epoch": 0.420986264354875,
      "grad_norm": 1.1912155151367188,
      "learning_rate": 8.952787898846585e-05,
      "loss": 1.495,
      "step": 2337
    },
    {
      "epoch": 0.42116640396307137,
      "grad_norm": 1.3948345184326172,
      "learning_rate": 8.951920505934199e-05,
      "loss": 1.6379,
      "step": 2338
    },
    {
      "epoch": 0.4213465435712677,
      "grad_norm": 1.446315884590149,
      "learning_rate": 8.951052796000676e-05,
      "loss": 1.8504,
      "step": 2339
    },
    {
      "epoch": 0.42152668317946407,
      "grad_norm": 1.2895922660827637,
      "learning_rate": 8.950184769115621e-05,
      "loss": 1.6966,
      "step": 2340
    },
    {
      "epoch": 0.4217068227876604,
      "grad_norm": 1.405571460723877,
      "learning_rate": 8.94931642534867e-05,
      "loss": 1.5646,
      "step": 2341
    },
    {
      "epoch": 0.42188696239585677,
      "grad_norm": 1.4101899862289429,
      "learning_rate": 8.948447764769477e-05,
      "loss": 1.679,
      "step": 2342
    },
    {
      "epoch": 0.4220671020040531,
      "grad_norm": 1.411247968673706,
      "learning_rate": 8.947578787447728e-05,
      "loss": 1.8322,
      "step": 2343
    },
    {
      "epoch": 0.42224724161224947,
      "grad_norm": 1.4264168739318848,
      "learning_rate": 8.946709493453132e-05,
      "loss": 1.7271,
      "step": 2344
    },
    {
      "epoch": 0.4224273812204458,
      "grad_norm": 1.330457329750061,
      "learning_rate": 8.945839882855422e-05,
      "loss": 1.5499,
      "step": 2345
    },
    {
      "epoch": 0.4226075208286422,
      "grad_norm": 1.3925718069076538,
      "learning_rate": 8.944969955724358e-05,
      "loss": 1.6385,
      "step": 2346
    },
    {
      "epoch": 0.4227876604368386,
      "grad_norm": 1.4525631666183472,
      "learning_rate": 8.944099712129727e-05,
      "loss": 1.7932,
      "step": 2347
    },
    {
      "epoch": 0.4229678000450349,
      "grad_norm": 1.4268920421600342,
      "learning_rate": 8.943229152141337e-05,
      "loss": 1.6502,
      "step": 2348
    },
    {
      "epoch": 0.4231479396532313,
      "grad_norm": 1.4260139465332031,
      "learning_rate": 8.942358275829026e-05,
      "loss": 1.4499,
      "step": 2349
    },
    {
      "epoch": 0.4233280792614276,
      "grad_norm": 1.4666041135787964,
      "learning_rate": 8.941487083262654e-05,
      "loss": 1.8307,
      "step": 2350
    },
    {
      "epoch": 0.423508218869624,
      "grad_norm": 1.4299721717834473,
      "learning_rate": 8.940615574512108e-05,
      "loss": 2.2927,
      "step": 2351
    },
    {
      "epoch": 0.4236883584778203,
      "grad_norm": 1.2465479373931885,
      "learning_rate": 8.939743749647301e-05,
      "loss": 1.8718,
      "step": 2352
    },
    {
      "epoch": 0.4238684980860167,
      "grad_norm": 1.1579290628433228,
      "learning_rate": 8.938871608738168e-05,
      "loss": 1.7701,
      "step": 2353
    },
    {
      "epoch": 0.424048637694213,
      "grad_norm": 1.1931153535842896,
      "learning_rate": 8.937999151854674e-05,
      "loss": 2.0909,
      "step": 2354
    },
    {
      "epoch": 0.4242287773024094,
      "grad_norm": 1.3782215118408203,
      "learning_rate": 8.937126379066804e-05,
      "loss": 2.0862,
      "step": 2355
    },
    {
      "epoch": 0.4244089169106057,
      "grad_norm": 1.3637079000473022,
      "learning_rate": 8.936253290444577e-05,
      "loss": 2.0026,
      "step": 2356
    },
    {
      "epoch": 0.4245890565188021,
      "grad_norm": 2.3987016677856445,
      "learning_rate": 8.935379886058026e-05,
      "loss": 2.5389,
      "step": 2357
    },
    {
      "epoch": 0.4247691961269984,
      "grad_norm": 1.3973356485366821,
      "learning_rate": 8.934506165977218e-05,
      "loss": 2.1329,
      "step": 2358
    },
    {
      "epoch": 0.4249493357351948,
      "grad_norm": 1.603434443473816,
      "learning_rate": 8.933632130272242e-05,
      "loss": 2.1958,
      "step": 2359
    },
    {
      "epoch": 0.42512947534339113,
      "grad_norm": 1.675324559211731,
      "learning_rate": 8.932757779013214e-05,
      "loss": 2.3248,
      "step": 2360
    },
    {
      "epoch": 0.4253096149515875,
      "grad_norm": 1.3479998111724854,
      "learning_rate": 8.93188311227027e-05,
      "loss": 1.4798,
      "step": 2361
    },
    {
      "epoch": 0.42548975455978383,
      "grad_norm": 1.265716791152954,
      "learning_rate": 8.931008130113578e-05,
      "loss": 1.8728,
      "step": 2362
    },
    {
      "epoch": 0.4256698941679802,
      "grad_norm": 1.2824552059173584,
      "learning_rate": 8.93013283261333e-05,
      "loss": 1.8325,
      "step": 2363
    },
    {
      "epoch": 0.42585003377617653,
      "grad_norm": 1.2346662282943726,
      "learning_rate": 8.92925721983974e-05,
      "loss": 1.511,
      "step": 2364
    },
    {
      "epoch": 0.4260301733843729,
      "grad_norm": 1.2194435596466064,
      "learning_rate": 8.928381291863049e-05,
      "loss": 1.6183,
      "step": 2365
    },
    {
      "epoch": 0.42621031299256923,
      "grad_norm": 1.2850176095962524,
      "learning_rate": 8.927505048753526e-05,
      "loss": 1.8856,
      "step": 2366
    },
    {
      "epoch": 0.4263904526007656,
      "grad_norm": 1.2203431129455566,
      "learning_rate": 8.92662849058146e-05,
      "loss": 1.7079,
      "step": 2367
    },
    {
      "epoch": 0.42657059220896193,
      "grad_norm": 1.2614035606384277,
      "learning_rate": 8.925751617417169e-05,
      "loss": 1.5353,
      "step": 2368
    },
    {
      "epoch": 0.4267507318171583,
      "grad_norm": 1.1229199171066284,
      "learning_rate": 8.924874429330994e-05,
      "loss": 1.2559,
      "step": 2369
    },
    {
      "epoch": 0.42693087142535463,
      "grad_norm": 1.325992226600647,
      "learning_rate": 8.923996926393305e-05,
      "loss": 1.9604,
      "step": 2370
    },
    {
      "epoch": 0.427111011033551,
      "grad_norm": 1.2546918392181396,
      "learning_rate": 8.923119108674494e-05,
      "loss": 1.5735,
      "step": 2371
    },
    {
      "epoch": 0.42729115064174733,
      "grad_norm": 1.375550627708435,
      "learning_rate": 8.92224097624498e-05,
      "loss": 1.7835,
      "step": 2372
    },
    {
      "epoch": 0.4274712902499437,
      "grad_norm": 1.4299111366271973,
      "learning_rate": 8.921362529175204e-05,
      "loss": 1.9749,
      "step": 2373
    },
    {
      "epoch": 0.42765142985814003,
      "grad_norm": 1.3563504219055176,
      "learning_rate": 8.920483767535636e-05,
      "loss": 1.5832,
      "step": 2374
    },
    {
      "epoch": 0.42783156946633644,
      "grad_norm": 1.3628417253494263,
      "learning_rate": 8.91960469139677e-05,
      "loss": 1.7148,
      "step": 2375
    },
    {
      "epoch": 0.4280117090745328,
      "grad_norm": 1.388820767402649,
      "learning_rate": 8.918725300829126e-05,
      "loss": 2.0205,
      "step": 2376
    },
    {
      "epoch": 0.42819184868272914,
      "grad_norm": 1.3120652437210083,
      "learning_rate": 8.917845595903246e-05,
      "loss": 1.6295,
      "step": 2377
    },
    {
      "epoch": 0.4283719882909255,
      "grad_norm": 1.3631128072738647,
      "learning_rate": 8.9169655766897e-05,
      "loss": 1.7328,
      "step": 2378
    },
    {
      "epoch": 0.42855212789912184,
      "grad_norm": 1.178661584854126,
      "learning_rate": 8.916085243259084e-05,
      "loss": 1.4739,
      "step": 2379
    },
    {
      "epoch": 0.4287322675073182,
      "grad_norm": 1.2599750757217407,
      "learning_rate": 8.915204595682018e-05,
      "loss": 1.5898,
      "step": 2380
    },
    {
      "epoch": 0.42891240711551454,
      "grad_norm": 1.2569741010665894,
      "learning_rate": 8.914323634029145e-05,
      "loss": 1.568,
      "step": 2381
    },
    {
      "epoch": 0.4290925467237109,
      "grad_norm": 1.245320439338684,
      "learning_rate": 8.913442358371138e-05,
      "loss": 1.6056,
      "step": 2382
    },
    {
      "epoch": 0.42927268633190724,
      "grad_norm": 1.3373866081237793,
      "learning_rate": 8.91256076877869e-05,
      "loss": 1.7986,
      "step": 2383
    },
    {
      "epoch": 0.4294528259401036,
      "grad_norm": 1.4051498174667358,
      "learning_rate": 8.911678865322524e-05,
      "loss": 1.8037,
      "step": 2384
    },
    {
      "epoch": 0.42963296554829994,
      "grad_norm": 1.3707317113876343,
      "learning_rate": 8.910796648073384e-05,
      "loss": 1.7514,
      "step": 2385
    },
    {
      "epoch": 0.4298131051564963,
      "grad_norm": 1.3537299633026123,
      "learning_rate": 8.909914117102041e-05,
      "loss": 1.6818,
      "step": 2386
    },
    {
      "epoch": 0.42999324476469264,
      "grad_norm": 1.2826321125030518,
      "learning_rate": 8.909031272479294e-05,
      "loss": 1.6717,
      "step": 2387
    },
    {
      "epoch": 0.430173384372889,
      "grad_norm": 1.5074782371520996,
      "learning_rate": 8.908148114275961e-05,
      "loss": 2.0749,
      "step": 2388
    },
    {
      "epoch": 0.43035352398108534,
      "grad_norm": 1.5770882368087769,
      "learning_rate": 8.907264642562889e-05,
      "loss": 1.8662,
      "step": 2389
    },
    {
      "epoch": 0.4305336635892817,
      "grad_norm": 1.3784534931182861,
      "learning_rate": 8.906380857410952e-05,
      "loss": 1.7692,
      "step": 2390
    },
    {
      "epoch": 0.43071380319747804,
      "grad_norm": 1.253674864768982,
      "learning_rate": 8.905496758891045e-05,
      "loss": 1.7131,
      "step": 2391
    },
    {
      "epoch": 0.4308939428056744,
      "grad_norm": 1.551912546157837,
      "learning_rate": 8.90461234707409e-05,
      "loss": 1.7812,
      "step": 2392
    },
    {
      "epoch": 0.43107408241387074,
      "grad_norm": 1.2576667070388794,
      "learning_rate": 8.903727622031035e-05,
      "loss": 1.5691,
      "step": 2393
    },
    {
      "epoch": 0.4312542220220671,
      "grad_norm": 1.4048371315002441,
      "learning_rate": 8.902842583832851e-05,
      "loss": 1.7556,
      "step": 2394
    },
    {
      "epoch": 0.43143436163026344,
      "grad_norm": 1.3117891550064087,
      "learning_rate": 8.901957232550537e-05,
      "loss": 1.752,
      "step": 2395
    },
    {
      "epoch": 0.4316145012384598,
      "grad_norm": 1.3268016576766968,
      "learning_rate": 8.901071568255112e-05,
      "loss": 1.708,
      "step": 2396
    },
    {
      "epoch": 0.43179464084665614,
      "grad_norm": 1.4855804443359375,
      "learning_rate": 8.90018559101763e-05,
      "loss": 1.6548,
      "step": 2397
    },
    {
      "epoch": 0.4319747804548525,
      "grad_norm": 1.2910501956939697,
      "learning_rate": 8.899299300909156e-05,
      "loss": 1.5835,
      "step": 2398
    },
    {
      "epoch": 0.43215492006304884,
      "grad_norm": 1.2075519561767578,
      "learning_rate": 8.898412698000792e-05,
      "loss": 1.5984,
      "step": 2399
    },
    {
      "epoch": 0.4323350596712452,
      "grad_norm": 1.361940860748291,
      "learning_rate": 8.897525782363661e-05,
      "loss": 1.6462,
      "step": 2400
    },
    {
      "epoch": 0.43251519927944154,
      "grad_norm": 1.417803406715393,
      "learning_rate": 8.89663855406891e-05,
      "loss": 2.45,
      "step": 2401
    },
    {
      "epoch": 0.4326953388876379,
      "grad_norm": 1.3750723600387573,
      "learning_rate": 8.895751013187713e-05,
      "loss": 2.2356,
      "step": 2402
    },
    {
      "epoch": 0.4328754784958343,
      "grad_norm": 1.3236726522445679,
      "learning_rate": 8.894863159791265e-05,
      "loss": 2.0514,
      "step": 2403
    },
    {
      "epoch": 0.43305561810403065,
      "grad_norm": 1.2322497367858887,
      "learning_rate": 8.893974993950794e-05,
      "loss": 1.9954,
      "step": 2404
    },
    {
      "epoch": 0.433235757712227,
      "grad_norm": 1.177254319190979,
      "learning_rate": 8.893086515737543e-05,
      "loss": 2.1942,
      "step": 2405
    },
    {
      "epoch": 0.43341589732042335,
      "grad_norm": 1.2370191812515259,
      "learning_rate": 8.89219772522279e-05,
      "loss": 1.884,
      "step": 2406
    },
    {
      "epoch": 0.4335960369286197,
      "grad_norm": 1.7734014987945557,
      "learning_rate": 8.89130862247783e-05,
      "loss": 2.4715,
      "step": 2407
    },
    {
      "epoch": 0.43377617653681605,
      "grad_norm": 1.4193809032440186,
      "learning_rate": 8.890419207573988e-05,
      "loss": 2.1092,
      "step": 2408
    },
    {
      "epoch": 0.4339563161450124,
      "grad_norm": 1.8293203115463257,
      "learning_rate": 8.889529480582612e-05,
      "loss": 2.215,
      "step": 2409
    },
    {
      "epoch": 0.43413645575320875,
      "grad_norm": 1.8702088594436646,
      "learning_rate": 8.888639441575077e-05,
      "loss": 2.294,
      "step": 2410
    },
    {
      "epoch": 0.4343165953614051,
      "grad_norm": 1.2626774311065674,
      "learning_rate": 8.887749090622779e-05,
      "loss": 1.6463,
      "step": 2411
    },
    {
      "epoch": 0.43449673496960145,
      "grad_norm": 1.2881629467010498,
      "learning_rate": 8.886858427797143e-05,
      "loss": 1.665,
      "step": 2412
    },
    {
      "epoch": 0.4346768745777978,
      "grad_norm": 1.1848063468933105,
      "learning_rate": 8.885967453169618e-05,
      "loss": 1.5145,
      "step": 2413
    },
    {
      "epoch": 0.43485701418599415,
      "grad_norm": 1.3114615678787231,
      "learning_rate": 8.885076166811674e-05,
      "loss": 1.7421,
      "step": 2414
    },
    {
      "epoch": 0.4350371537941905,
      "grad_norm": 1.295041799545288,
      "learning_rate": 8.884184568794814e-05,
      "loss": 1.7812,
      "step": 2415
    },
    {
      "epoch": 0.43521729340238685,
      "grad_norm": 1.3562287092208862,
      "learning_rate": 8.88329265919056e-05,
      "loss": 1.8488,
      "step": 2416
    },
    {
      "epoch": 0.4353974330105832,
      "grad_norm": 1.2571269273757935,
      "learning_rate": 8.88240043807046e-05,
      "loss": 1.6278,
      "step": 2417
    },
    {
      "epoch": 0.43557757261877955,
      "grad_norm": 1.1439836025238037,
      "learning_rate": 8.881507905506087e-05,
      "loss": 1.4527,
      "step": 2418
    },
    {
      "epoch": 0.4357577122269759,
      "grad_norm": 1.3284558057785034,
      "learning_rate": 8.880615061569041e-05,
      "loss": 1.8035,
      "step": 2419
    },
    {
      "epoch": 0.43593785183517225,
      "grad_norm": 1.3843868970870972,
      "learning_rate": 8.879721906330947e-05,
      "loss": 1.6351,
      "step": 2420
    },
    {
      "epoch": 0.4361179914433686,
      "grad_norm": 1.2090736627578735,
      "learning_rate": 8.878828439863449e-05,
      "loss": 1.5604,
      "step": 2421
    },
    {
      "epoch": 0.43629813105156495,
      "grad_norm": 1.3676234483718872,
      "learning_rate": 8.877934662238223e-05,
      "loss": 1.5865,
      "step": 2422
    },
    {
      "epoch": 0.4364782706597613,
      "grad_norm": 1.2687270641326904,
      "learning_rate": 8.877040573526969e-05,
      "loss": 1.5578,
      "step": 2423
    },
    {
      "epoch": 0.43665841026795765,
      "grad_norm": 1.3585126399993896,
      "learning_rate": 8.876146173801407e-05,
      "loss": 1.6335,
      "step": 2424
    },
    {
      "epoch": 0.436838549876154,
      "grad_norm": 1.355819821357727,
      "learning_rate": 8.875251463133288e-05,
      "loss": 1.6906,
      "step": 2425
    },
    {
      "epoch": 0.43701868948435035,
      "grad_norm": 1.2871942520141602,
      "learning_rate": 8.874356441594382e-05,
      "loss": 1.7136,
      "step": 2426
    },
    {
      "epoch": 0.4371988290925467,
      "grad_norm": 1.287707805633545,
      "learning_rate": 8.873461109256492e-05,
      "loss": 1.7524,
      "step": 2427
    },
    {
      "epoch": 0.43737896870074305,
      "grad_norm": 1.2620363235473633,
      "learning_rate": 8.872565466191438e-05,
      "loss": 1.7398,
      "step": 2428
    },
    {
      "epoch": 0.4375591083089394,
      "grad_norm": 1.3467801809310913,
      "learning_rate": 8.871669512471068e-05,
      "loss": 1.6144,
      "step": 2429
    },
    {
      "epoch": 0.43773924791713575,
      "grad_norm": 1.2558860778808594,
      "learning_rate": 8.870773248167254e-05,
      "loss": 1.5698,
      "step": 2430
    },
    {
      "epoch": 0.43791938752533216,
      "grad_norm": 1.2626712322235107,
      "learning_rate": 8.869876673351898e-05,
      "loss": 1.6514,
      "step": 2431
    },
    {
      "epoch": 0.4380995271335285,
      "grad_norm": 1.370428204536438,
      "learning_rate": 8.86897978809692e-05,
      "loss": 1.831,
      "step": 2432
    },
    {
      "epoch": 0.43827966674172486,
      "grad_norm": 1.2651333808898926,
      "learning_rate": 8.868082592474267e-05,
      "loss": 1.6784,
      "step": 2433
    },
    {
      "epoch": 0.4384598063499212,
      "grad_norm": 1.1817269325256348,
      "learning_rate": 8.867185086555913e-05,
      "loss": 1.489,
      "step": 2434
    },
    {
      "epoch": 0.43863994595811756,
      "grad_norm": 1.3302593231201172,
      "learning_rate": 8.866287270413854e-05,
      "loss": 1.713,
      "step": 2435
    },
    {
      "epoch": 0.4388200855663139,
      "grad_norm": 1.2751165628433228,
      "learning_rate": 8.865389144120114e-05,
      "loss": 1.4732,
      "step": 2436
    },
    {
      "epoch": 0.43900022517451026,
      "grad_norm": 1.3313978910446167,
      "learning_rate": 8.864490707746739e-05,
      "loss": 1.8413,
      "step": 2437
    },
    {
      "epoch": 0.4391803647827066,
      "grad_norm": 1.3491672277450562,
      "learning_rate": 8.863591961365803e-05,
      "loss": 1.5525,
      "step": 2438
    },
    {
      "epoch": 0.43936050439090296,
      "grad_norm": 1.1956068277359009,
      "learning_rate": 8.862692905049401e-05,
      "loss": 1.2878,
      "step": 2439
    },
    {
      "epoch": 0.4395406439990993,
      "grad_norm": 1.2738324403762817,
      "learning_rate": 8.861793538869656e-05,
      "loss": 1.5003,
      "step": 2440
    },
    {
      "epoch": 0.43972078360729566,
      "grad_norm": 1.3016011714935303,
      "learning_rate": 8.860893862898713e-05,
      "loss": 1.5044,
      "step": 2441
    },
    {
      "epoch": 0.439900923215492,
      "grad_norm": 1.393778920173645,
      "learning_rate": 8.859993877208747e-05,
      "loss": 1.6511,
      "step": 2442
    },
    {
      "epoch": 0.44008106282368836,
      "grad_norm": 1.3200653791427612,
      "learning_rate": 8.859093581871952e-05,
      "loss": 1.4509,
      "step": 2443
    },
    {
      "epoch": 0.4402612024318847,
      "grad_norm": 1.5782687664031982,
      "learning_rate": 8.858192976960549e-05,
      "loss": 1.5456,
      "step": 2444
    },
    {
      "epoch": 0.44044134204008106,
      "grad_norm": 1.5603652000427246,
      "learning_rate": 8.857292062546783e-05,
      "loss": 1.8388,
      "step": 2445
    },
    {
      "epoch": 0.4406214816482774,
      "grad_norm": 1.380617618560791,
      "learning_rate": 8.856390838702928e-05,
      "loss": 1.692,
      "step": 2446
    },
    {
      "epoch": 0.44080162125647376,
      "grad_norm": 1.5715570449829102,
      "learning_rate": 8.855489305501279e-05,
      "loss": 1.5373,
      "step": 2447
    },
    {
      "epoch": 0.4409817608646701,
      "grad_norm": 1.4420361518859863,
      "learning_rate": 8.854587463014155e-05,
      "loss": 1.7806,
      "step": 2448
    },
    {
      "epoch": 0.44116190047286646,
      "grad_norm": 1.3811311721801758,
      "learning_rate": 8.853685311313903e-05,
      "loss": 1.5826,
      "step": 2449
    },
    {
      "epoch": 0.4413420400810628,
      "grad_norm": 1.3375993967056274,
      "learning_rate": 8.852782850472892e-05,
      "loss": 1.8287,
      "step": 2450
    },
    {
      "epoch": 0.44152217968925916,
      "grad_norm": 1.2993003129959106,
      "learning_rate": 8.851880080563516e-05,
      "loss": 2.1843,
      "step": 2451
    },
    {
      "epoch": 0.4417023192974555,
      "grad_norm": 1.493749976158142,
      "learning_rate": 8.850977001658198e-05,
      "loss": 1.9254,
      "step": 2452
    },
    {
      "epoch": 0.44188245890565186,
      "grad_norm": 1.3017258644104004,
      "learning_rate": 8.850073613829379e-05,
      "loss": 1.9634,
      "step": 2453
    },
    {
      "epoch": 0.4420625985138482,
      "grad_norm": 1.6557326316833496,
      "learning_rate": 8.849169917149531e-05,
      "loss": 2.0286,
      "step": 2454
    },
    {
      "epoch": 0.44224273812204457,
      "grad_norm": 1.2409926652908325,
      "learning_rate": 8.848265911691147e-05,
      "loss": 1.9505,
      "step": 2455
    },
    {
      "epoch": 0.4424228777302409,
      "grad_norm": 1.5028578042984009,
      "learning_rate": 8.847361597526746e-05,
      "loss": 1.8375,
      "step": 2456
    },
    {
      "epoch": 0.44260301733843727,
      "grad_norm": 1.4043923616409302,
      "learning_rate": 8.846456974728873e-05,
      "loss": 1.982,
      "step": 2457
    },
    {
      "epoch": 0.4427831569466336,
      "grad_norm": 1.4624378681182861,
      "learning_rate": 8.845552043370093e-05,
      "loss": 2.3653,
      "step": 2458
    },
    {
      "epoch": 0.44296329655482997,
      "grad_norm": 1.6124484539031982,
      "learning_rate": 8.844646803523002e-05,
      "loss": 2.1662,
      "step": 2459
    },
    {
      "epoch": 0.44314343616302637,
      "grad_norm": 1.7372701168060303,
      "learning_rate": 8.843741255260217e-05,
      "loss": 2.0882,
      "step": 2460
    },
    {
      "epoch": 0.4433235757712227,
      "grad_norm": 1.8878483772277832,
      "learning_rate": 8.842835398654381e-05,
      "loss": 1.8989,
      "step": 2461
    },
    {
      "epoch": 0.4435037153794191,
      "grad_norm": 1.1851613521575928,
      "learning_rate": 8.841929233778161e-05,
      "loss": 1.6491,
      "step": 2462
    },
    {
      "epoch": 0.4436838549876154,
      "grad_norm": 1.2387031316757202,
      "learning_rate": 8.841022760704248e-05,
      "loss": 1.9347,
      "step": 2463
    },
    {
      "epoch": 0.4438639945958118,
      "grad_norm": 1.2441588640213013,
      "learning_rate": 8.840115979505361e-05,
      "loss": 1.855,
      "step": 2464
    },
    {
      "epoch": 0.4440441342040081,
      "grad_norm": 1.280192494392395,
      "learning_rate": 8.839208890254241e-05,
      "loss": 1.6512,
      "step": 2465
    },
    {
      "epoch": 0.4442242738122045,
      "grad_norm": 1.2847806215286255,
      "learning_rate": 8.838301493023654e-05,
      "loss": 1.6675,
      "step": 2466
    },
    {
      "epoch": 0.4444044134204008,
      "grad_norm": 1.3067476749420166,
      "learning_rate": 8.83739378788639e-05,
      "loss": 1.661,
      "step": 2467
    },
    {
      "epoch": 0.4445845530285972,
      "grad_norm": 1.2300266027450562,
      "learning_rate": 8.836485774915267e-05,
      "loss": 1.6697,
      "step": 2468
    },
    {
      "epoch": 0.4447646926367935,
      "grad_norm": 1.2237114906311035,
      "learning_rate": 8.835577454183122e-05,
      "loss": 1.6357,
      "step": 2469
    },
    {
      "epoch": 0.4449448322449899,
      "grad_norm": 1.2065696716308594,
      "learning_rate": 8.834668825762821e-05,
      "loss": 1.694,
      "step": 2470
    },
    {
      "epoch": 0.4451249718531862,
      "grad_norm": 1.2156944274902344,
      "learning_rate": 8.833759889727257e-05,
      "loss": 1.6118,
      "step": 2471
    },
    {
      "epoch": 0.4453051114613826,
      "grad_norm": 1.3187237977981567,
      "learning_rate": 8.83285064614934e-05,
      "loss": 1.6126,
      "step": 2472
    },
    {
      "epoch": 0.4454852510695789,
      "grad_norm": 1.2941577434539795,
      "learning_rate": 8.831941095102013e-05,
      "loss": 1.6511,
      "step": 2473
    },
    {
      "epoch": 0.4456653906777753,
      "grad_norm": 1.389487624168396,
      "learning_rate": 8.831031236658236e-05,
      "loss": 1.9561,
      "step": 2474
    },
    {
      "epoch": 0.4458455302859716,
      "grad_norm": 1.3510010242462158,
      "learning_rate": 8.830121070891e-05,
      "loss": 1.8267,
      "step": 2475
    },
    {
      "epoch": 0.446025669894168,
      "grad_norm": 1.2735846042633057,
      "learning_rate": 8.829210597873318e-05,
      "loss": 1.777,
      "step": 2476
    },
    {
      "epoch": 0.4462058095023643,
      "grad_norm": 1.5049409866333008,
      "learning_rate": 8.828299817678226e-05,
      "loss": 2.0296,
      "step": 2477
    },
    {
      "epoch": 0.4463859491105607,
      "grad_norm": 1.2562586069107056,
      "learning_rate": 8.827388730378787e-05,
      "loss": 1.649,
      "step": 2478
    },
    {
      "epoch": 0.446566088718757,
      "grad_norm": 1.282714605331421,
      "learning_rate": 8.82647733604809e-05,
      "loss": 1.8501,
      "step": 2479
    },
    {
      "epoch": 0.4467462283269534,
      "grad_norm": 1.325710415840149,
      "learning_rate": 8.825565634759243e-05,
      "loss": 1.8766,
      "step": 2480
    },
    {
      "epoch": 0.4469263679351497,
      "grad_norm": 1.2347376346588135,
      "learning_rate": 8.824653626585386e-05,
      "loss": 1.5214,
      "step": 2481
    },
    {
      "epoch": 0.4471065075433461,
      "grad_norm": 1.355336308479309,
      "learning_rate": 8.823741311599678e-05,
      "loss": 1.8545,
      "step": 2482
    },
    {
      "epoch": 0.4472866471515424,
      "grad_norm": 1.1751785278320312,
      "learning_rate": 8.822828689875301e-05,
      "loss": 1.3873,
      "step": 2483
    },
    {
      "epoch": 0.4474667867597388,
      "grad_norm": 1.2634822130203247,
      "learning_rate": 8.821915761485472e-05,
      "loss": 1.6765,
      "step": 2484
    },
    {
      "epoch": 0.4476469263679351,
      "grad_norm": 1.3524913787841797,
      "learning_rate": 8.821002526503422e-05,
      "loss": 1.7428,
      "step": 2485
    },
    {
      "epoch": 0.4478270659761315,
      "grad_norm": 1.4607428312301636,
      "learning_rate": 8.82008898500241e-05,
      "loss": 1.5735,
      "step": 2486
    },
    {
      "epoch": 0.44800720558432783,
      "grad_norm": 1.215192198753357,
      "learning_rate": 8.81917513705572e-05,
      "loss": 1.6297,
      "step": 2487
    },
    {
      "epoch": 0.44818734519252423,
      "grad_norm": 1.2971477508544922,
      "learning_rate": 8.818260982736661e-05,
      "loss": 1.6451,
      "step": 2488
    },
    {
      "epoch": 0.4483674848007206,
      "grad_norm": 1.378607988357544,
      "learning_rate": 8.817346522118566e-05,
      "loss": 1.878,
      "step": 2489
    },
    {
      "epoch": 0.44854762440891693,
      "grad_norm": 1.387888789176941,
      "learning_rate": 8.816431755274791e-05,
      "loss": 1.6493,
      "step": 2490
    },
    {
      "epoch": 0.4487277640171133,
      "grad_norm": 1.3238481283187866,
      "learning_rate": 8.81551668227872e-05,
      "loss": 1.6445,
      "step": 2491
    },
    {
      "epoch": 0.44890790362530963,
      "grad_norm": 1.3662227392196655,
      "learning_rate": 8.81460130320376e-05,
      "loss": 1.5854,
      "step": 2492
    },
    {
      "epoch": 0.449088043233506,
      "grad_norm": 1.3880327939987183,
      "learning_rate": 8.813685618123341e-05,
      "loss": 1.7137,
      "step": 2493
    },
    {
      "epoch": 0.44926818284170233,
      "grad_norm": 1.4108017683029175,
      "learning_rate": 8.81276962711092e-05,
      "loss": 1.8754,
      "step": 2494
    },
    {
      "epoch": 0.4494483224498987,
      "grad_norm": 1.4106652736663818,
      "learning_rate": 8.811853330239976e-05,
      "loss": 1.476,
      "step": 2495
    },
    {
      "epoch": 0.44962846205809504,
      "grad_norm": 1.3727869987487793,
      "learning_rate": 8.810936727584015e-05,
      "loss": 1.6171,
      "step": 2496
    },
    {
      "epoch": 0.4498086016662914,
      "grad_norm": 1.503373384475708,
      "learning_rate": 8.810019819216566e-05,
      "loss": 1.7728,
      "step": 2497
    },
    {
      "epoch": 0.44998874127448774,
      "grad_norm": 1.345131754875183,
      "learning_rate": 8.809102605211181e-05,
      "loss": 1.6648,
      "step": 2498
    },
    {
      "epoch": 0.4501688808826841,
      "grad_norm": 1.2526774406433105,
      "learning_rate": 8.808185085641443e-05,
      "loss": 1.5028,
      "step": 2499
    },
    {
      "epoch": 0.45034902049088044,
      "grad_norm": 1.2435864210128784,
      "learning_rate": 8.807267260580952e-05,
      "loss": 1.4284,
      "step": 2500
    },
    {
      "epoch": 0.4505291600990768,
      "grad_norm": 1.3523681163787842,
      "learning_rate": 8.806349130103333e-05,
      "loss": 2.2496,
      "step": 2501
    },
    {
      "epoch": 0.45070929970727314,
      "grad_norm": 1.1527364253997803,
      "learning_rate": 8.805430694282244e-05,
      "loss": 1.9082,
      "step": 2502
    },
    {
      "epoch": 0.4508894393154695,
      "grad_norm": 1.2345834970474243,
      "learning_rate": 8.804511953191357e-05,
      "loss": 2.1592,
      "step": 2503
    },
    {
      "epoch": 0.45106957892366584,
      "grad_norm": 1.2807146310806274,
      "learning_rate": 8.803592906904374e-05,
      "loss": 2.3721,
      "step": 2504
    },
    {
      "epoch": 0.4512497185318622,
      "grad_norm": 1.3325930833816528,
      "learning_rate": 8.80267355549502e-05,
      "loss": 2.0378,
      "step": 2505
    },
    {
      "epoch": 0.45142985814005854,
      "grad_norm": 1.3501343727111816,
      "learning_rate": 8.801753899037048e-05,
      "loss": 2.243,
      "step": 2506
    },
    {
      "epoch": 0.4516099977482549,
      "grad_norm": 1.3663647174835205,
      "learning_rate": 8.800833937604227e-05,
      "loss": 1.8671,
      "step": 2507
    },
    {
      "epoch": 0.45179013735645124,
      "grad_norm": 1.3206526041030884,
      "learning_rate": 8.799913671270361e-05,
      "loss": 1.9298,
      "step": 2508
    },
    {
      "epoch": 0.4519702769646476,
      "grad_norm": 1.6166900396347046,
      "learning_rate": 8.798993100109271e-05,
      "loss": 2.3065,
      "step": 2509
    },
    {
      "epoch": 0.45215041657284394,
      "grad_norm": 1.8101261854171753,
      "learning_rate": 8.798072224194805e-05,
      "loss": 2.2197,
      "step": 2510
    },
    {
      "epoch": 0.4523305561810403,
      "grad_norm": 1.2505074739456177,
      "learning_rate": 8.797151043600834e-05,
      "loss": 1.493,
      "step": 2511
    },
    {
      "epoch": 0.45251069578923664,
      "grad_norm": 1.354472279548645,
      "learning_rate": 8.796229558401257e-05,
      "loss": 1.7866,
      "step": 2512
    },
    {
      "epoch": 0.452690835397433,
      "grad_norm": 1.2613309621810913,
      "learning_rate": 8.795307768669992e-05,
      "loss": 1.8656,
      "step": 2513
    },
    {
      "epoch": 0.45287097500562934,
      "grad_norm": 1.1571128368377686,
      "learning_rate": 8.794385674480988e-05,
      "loss": 1.5235,
      "step": 2514
    },
    {
      "epoch": 0.4530511146138257,
      "grad_norm": 1.310507893562317,
      "learning_rate": 8.793463275908214e-05,
      "loss": 1.6617,
      "step": 2515
    },
    {
      "epoch": 0.45323125422202204,
      "grad_norm": 1.1342276334762573,
      "learning_rate": 8.792540573025663e-05,
      "loss": 1.4005,
      "step": 2516
    },
    {
      "epoch": 0.45341139383021845,
      "grad_norm": 1.0309433937072754,
      "learning_rate": 8.791617565907353e-05,
      "loss": 1.3134,
      "step": 2517
    },
    {
      "epoch": 0.4535915334384148,
      "grad_norm": 1.2996028661727905,
      "learning_rate": 8.79069425462733e-05,
      "loss": 1.7006,
      "step": 2518
    },
    {
      "epoch": 0.45377167304661115,
      "grad_norm": 1.231593370437622,
      "learning_rate": 8.78977063925966e-05,
      "loss": 1.6629,
      "step": 2519
    },
    {
      "epoch": 0.4539518126548075,
      "grad_norm": 1.2369225025177002,
      "learning_rate": 8.788846719878436e-05,
      "loss": 1.6645,
      "step": 2520
    },
    {
      "epoch": 0.45413195226300385,
      "grad_norm": 1.2013051509857178,
      "learning_rate": 8.787922496557773e-05,
      "loss": 1.7033,
      "step": 2521
    },
    {
      "epoch": 0.4543120918712002,
      "grad_norm": 1.249882698059082,
      "learning_rate": 8.786997969371813e-05,
      "loss": 1.8146,
      "step": 2522
    },
    {
      "epoch": 0.45449223147939655,
      "grad_norm": 1.301825761795044,
      "learning_rate": 8.786073138394721e-05,
      "loss": 1.6801,
      "step": 2523
    },
    {
      "epoch": 0.4546723710875929,
      "grad_norm": 1.1768606901168823,
      "learning_rate": 8.785148003700685e-05,
      "loss": 1.5777,
      "step": 2524
    },
    {
      "epoch": 0.45485251069578925,
      "grad_norm": 1.1592276096343994,
      "learning_rate": 8.784222565363918e-05,
      "loss": 1.4041,
      "step": 2525
    },
    {
      "epoch": 0.4550326503039856,
      "grad_norm": 1.1781882047653198,
      "learning_rate": 8.783296823458662e-05,
      "loss": 1.5239,
      "step": 2526
    },
    {
      "epoch": 0.45521278991218195,
      "grad_norm": 1.2495073080062866,
      "learning_rate": 8.782370778059177e-05,
      "loss": 1.6619,
      "step": 2527
    },
    {
      "epoch": 0.4553929295203783,
      "grad_norm": 1.3477360010147095,
      "learning_rate": 8.781444429239752e-05,
      "loss": 2.0455,
      "step": 2528
    },
    {
      "epoch": 0.45557306912857465,
      "grad_norm": 1.3403608798980713,
      "learning_rate": 8.780517777074695e-05,
      "loss": 1.6951,
      "step": 2529
    },
    {
      "epoch": 0.455753208736771,
      "grad_norm": 1.1876440048217773,
      "learning_rate": 8.779590821638344e-05,
      "loss": 1.3576,
      "step": 2530
    },
    {
      "epoch": 0.45593334834496735,
      "grad_norm": 1.2472617626190186,
      "learning_rate": 8.778663563005059e-05,
      "loss": 1.6684,
      "step": 2531
    },
    {
      "epoch": 0.4561134879531637,
      "grad_norm": 1.2174952030181885,
      "learning_rate": 8.777736001249224e-05,
      "loss": 1.5275,
      "step": 2532
    },
    {
      "epoch": 0.45629362756136005,
      "grad_norm": 1.2705187797546387,
      "learning_rate": 8.776808136445245e-05,
      "loss": 1.5641,
      "step": 2533
    },
    {
      "epoch": 0.4564737671695564,
      "grad_norm": 1.1995683908462524,
      "learning_rate": 8.775879968667557e-05,
      "loss": 1.5224,
      "step": 2534
    },
    {
      "epoch": 0.45665390677775275,
      "grad_norm": 1.3762696981430054,
      "learning_rate": 8.774951497990617e-05,
      "loss": 1.8633,
      "step": 2535
    },
    {
      "epoch": 0.4568340463859491,
      "grad_norm": 1.2934454679489136,
      "learning_rate": 8.774022724488908e-05,
      "loss": 1.7836,
      "step": 2536
    },
    {
      "epoch": 0.45701418599414545,
      "grad_norm": 1.3924291133880615,
      "learning_rate": 8.773093648236934e-05,
      "loss": 1.7315,
      "step": 2537
    },
    {
      "epoch": 0.4571943256023418,
      "grad_norm": 1.357987880706787,
      "learning_rate": 8.772164269309225e-05,
      "loss": 1.7143,
      "step": 2538
    },
    {
      "epoch": 0.45737446521053815,
      "grad_norm": 1.3550989627838135,
      "learning_rate": 8.771234587780337e-05,
      "loss": 1.5903,
      "step": 2539
    },
    {
      "epoch": 0.4575546048187345,
      "grad_norm": 1.3092279434204102,
      "learning_rate": 8.770304603724846e-05,
      "loss": 1.7098,
      "step": 2540
    },
    {
      "epoch": 0.45773474442693085,
      "grad_norm": 1.2539315223693848,
      "learning_rate": 8.769374317217356e-05,
      "loss": 1.6217,
      "step": 2541
    },
    {
      "epoch": 0.4579148840351272,
      "grad_norm": 1.490655541419983,
      "learning_rate": 8.768443728332494e-05,
      "loss": 1.7979,
      "step": 2542
    },
    {
      "epoch": 0.45809502364332355,
      "grad_norm": 1.288400411605835,
      "learning_rate": 8.767512837144911e-05,
      "loss": 1.7497,
      "step": 2543
    },
    {
      "epoch": 0.4582751632515199,
      "grad_norm": 1.5470545291900635,
      "learning_rate": 8.766581643729286e-05,
      "loss": 1.8006,
      "step": 2544
    },
    {
      "epoch": 0.4584553028597163,
      "grad_norm": 1.3264930248260498,
      "learning_rate": 8.765650148160314e-05,
      "loss": 1.6516,
      "step": 2545
    },
    {
      "epoch": 0.45863544246791266,
      "grad_norm": 1.4090070724487305,
      "learning_rate": 8.764718350512721e-05,
      "loss": 1.6554,
      "step": 2546
    },
    {
      "epoch": 0.458815582076109,
      "grad_norm": 1.4938548803329468,
      "learning_rate": 8.763786250861256e-05,
      "loss": 1.7768,
      "step": 2547
    },
    {
      "epoch": 0.45899572168430536,
      "grad_norm": 1.2691638469696045,
      "learning_rate": 8.762853849280693e-05,
      "loss": 1.4626,
      "step": 2548
    },
    {
      "epoch": 0.4591758612925017,
      "grad_norm": 1.5443065166473389,
      "learning_rate": 8.761921145845826e-05,
      "loss": 1.6673,
      "step": 2549
    },
    {
      "epoch": 0.45935600090069806,
      "grad_norm": 1.3302720785140991,
      "learning_rate": 8.760988140631476e-05,
      "loss": 1.506,
      "step": 2550
    },
    {
      "epoch": 0.4595361405088944,
      "grad_norm": 1.1516814231872559,
      "learning_rate": 8.76005483371249e-05,
      "loss": 2.1592,
      "step": 2551
    },
    {
      "epoch": 0.45971628011709076,
      "grad_norm": 1.2670003175735474,
      "learning_rate": 8.759121225163738e-05,
      "loss": 1.9908,
      "step": 2552
    },
    {
      "epoch": 0.4598964197252871,
      "grad_norm": 1.3853363990783691,
      "learning_rate": 8.758187315060111e-05,
      "loss": 2.1705,
      "step": 2553
    },
    {
      "epoch": 0.46007655933348346,
      "grad_norm": 1.2478830814361572,
      "learning_rate": 8.757253103476528e-05,
      "loss": 1.9577,
      "step": 2554
    },
    {
      "epoch": 0.4602566989416798,
      "grad_norm": 1.3179986476898193,
      "learning_rate": 8.75631859048793e-05,
      "loss": 2.2253,
      "step": 2555
    },
    {
      "epoch": 0.46043683854987616,
      "grad_norm": 1.29220449924469,
      "learning_rate": 8.755383776169286e-05,
      "loss": 1.7994,
      "step": 2556
    },
    {
      "epoch": 0.4606169781580725,
      "grad_norm": 1.5291500091552734,
      "learning_rate": 8.754448660595585e-05,
      "loss": 1.9425,
      "step": 2557
    },
    {
      "epoch": 0.46079711776626886,
      "grad_norm": 1.4553630352020264,
      "learning_rate": 8.75351324384184e-05,
      "loss": 2.0551,
      "step": 2558
    },
    {
      "epoch": 0.4609772573744652,
      "grad_norm": 1.6553764343261719,
      "learning_rate": 8.752577525983093e-05,
      "loss": 2.1199,
      "step": 2559
    },
    {
      "epoch": 0.46115739698266156,
      "grad_norm": 1.1331473588943481,
      "learning_rate": 8.751641507094401e-05,
      "loss": 1.5915,
      "step": 2560
    },
    {
      "epoch": 0.4613375365908579,
      "grad_norm": 1.4233940839767456,
      "learning_rate": 8.750705187250858e-05,
      "loss": 2.0584,
      "step": 2561
    },
    {
      "epoch": 0.46151767619905426,
      "grad_norm": 1.2926568984985352,
      "learning_rate": 8.74976856652757e-05,
      "loss": 1.5486,
      "step": 2562
    },
    {
      "epoch": 0.4616978158072506,
      "grad_norm": 1.3534621000289917,
      "learning_rate": 8.748831644999674e-05,
      "loss": 1.6501,
      "step": 2563
    },
    {
      "epoch": 0.46187795541544696,
      "grad_norm": 1.1807377338409424,
      "learning_rate": 8.74789442274233e-05,
      "loss": 1.526,
      "step": 2564
    },
    {
      "epoch": 0.4620580950236433,
      "grad_norm": 1.4298193454742432,
      "learning_rate": 8.746956899830721e-05,
      "loss": 1.4228,
      "step": 2565
    },
    {
      "epoch": 0.46223823463183966,
      "grad_norm": 1.1970449686050415,
      "learning_rate": 8.746019076340055e-05,
      "loss": 1.6464,
      "step": 2566
    },
    {
      "epoch": 0.462418374240036,
      "grad_norm": 1.2095990180969238,
      "learning_rate": 8.745080952345562e-05,
      "loss": 1.4759,
      "step": 2567
    },
    {
      "epoch": 0.46259851384823236,
      "grad_norm": 1.303605079650879,
      "learning_rate": 8.744142527922501e-05,
      "loss": 1.8078,
      "step": 2568
    },
    {
      "epoch": 0.4627786534564287,
      "grad_norm": 1.3396903276443481,
      "learning_rate": 8.743203803146148e-05,
      "loss": 1.5931,
      "step": 2569
    },
    {
      "epoch": 0.46295879306462506,
      "grad_norm": 1.4176502227783203,
      "learning_rate": 8.742264778091812e-05,
      "loss": 1.8256,
      "step": 2570
    },
    {
      "epoch": 0.4631389326728214,
      "grad_norm": 1.3381545543670654,
      "learning_rate": 8.741325452834814e-05,
      "loss": 1.7019,
      "step": 2571
    },
    {
      "epoch": 0.46331907228101776,
      "grad_norm": 1.2189345359802246,
      "learning_rate": 8.740385827450514e-05,
      "loss": 1.4363,
      "step": 2572
    },
    {
      "epoch": 0.4634992118892141,
      "grad_norm": 1.2670601606369019,
      "learning_rate": 8.739445902014284e-05,
      "loss": 1.6716,
      "step": 2573
    },
    {
      "epoch": 0.4636793514974105,
      "grad_norm": 1.3506355285644531,
      "learning_rate": 8.738505676601525e-05,
      "loss": 1.6444,
      "step": 2574
    },
    {
      "epoch": 0.46385949110560687,
      "grad_norm": 1.1819381713867188,
      "learning_rate": 8.737565151287662e-05,
      "loss": 1.346,
      "step": 2575
    },
    {
      "epoch": 0.4640396307138032,
      "grad_norm": 1.245508074760437,
      "learning_rate": 8.736624326148142e-05,
      "loss": 1.6375,
      "step": 2576
    },
    {
      "epoch": 0.46421977032199957,
      "grad_norm": 1.2968839406967163,
      "learning_rate": 8.73568320125844e-05,
      "loss": 1.6468,
      "step": 2577
    },
    {
      "epoch": 0.4643999099301959,
      "grad_norm": 1.289516806602478,
      "learning_rate": 8.734741776694048e-05,
      "loss": 1.7434,
      "step": 2578
    },
    {
      "epoch": 0.46458004953839227,
      "grad_norm": 1.2348601818084717,
      "learning_rate": 8.733800052530495e-05,
      "loss": 1.3735,
      "step": 2579
    },
    {
      "epoch": 0.4647601891465886,
      "grad_norm": 1.2062407732009888,
      "learning_rate": 8.732858028843316e-05,
      "loss": 1.6988,
      "step": 2580
    },
    {
      "epoch": 0.46494032875478497,
      "grad_norm": 1.302638053894043,
      "learning_rate": 8.731915705708086e-05,
      "loss": 1.7326,
      "step": 2581
    },
    {
      "epoch": 0.4651204683629813,
      "grad_norm": 1.3212897777557373,
      "learning_rate": 8.730973083200395e-05,
      "loss": 1.8311,
      "step": 2582
    },
    {
      "epoch": 0.46530060797117767,
      "grad_norm": 1.2713528871536255,
      "learning_rate": 8.730030161395862e-05,
      "loss": 1.6469,
      "step": 2583
    },
    {
      "epoch": 0.465480747579374,
      "grad_norm": 1.217058539390564,
      "learning_rate": 8.729086940370125e-05,
      "loss": 1.6528,
      "step": 2584
    },
    {
      "epoch": 0.46566088718757037,
      "grad_norm": 1.425415277481079,
      "learning_rate": 8.728143420198851e-05,
      "loss": 1.8322,
      "step": 2585
    },
    {
      "epoch": 0.4658410267957667,
      "grad_norm": 1.35520601272583,
      "learning_rate": 8.727199600957728e-05,
      "loss": 1.7416,
      "step": 2586
    },
    {
      "epoch": 0.46602116640396307,
      "grad_norm": 1.2844891548156738,
      "learning_rate": 8.726255482722467e-05,
      "loss": 1.6008,
      "step": 2587
    },
    {
      "epoch": 0.4662013060121594,
      "grad_norm": 1.2533422708511353,
      "learning_rate": 8.725311065568806e-05,
      "loss": 1.5579,
      "step": 2588
    },
    {
      "epoch": 0.4663814456203558,
      "grad_norm": 1.4437865018844604,
      "learning_rate": 8.724366349572508e-05,
      "loss": 1.6766,
      "step": 2589
    },
    {
      "epoch": 0.4665615852285521,
      "grad_norm": 1.2465699911117554,
      "learning_rate": 8.723421334809354e-05,
      "loss": 1.4116,
      "step": 2590
    },
    {
      "epoch": 0.4667417248367485,
      "grad_norm": 1.3845593929290771,
      "learning_rate": 8.722476021355155e-05,
      "loss": 1.9713,
      "step": 2591
    },
    {
      "epoch": 0.4669218644449448,
      "grad_norm": 1.3377214670181274,
      "learning_rate": 8.72153040928574e-05,
      "loss": 1.5961,
      "step": 2592
    },
    {
      "epoch": 0.4671020040531412,
      "grad_norm": 1.4602549076080322,
      "learning_rate": 8.720584498676971e-05,
      "loss": 2.0016,
      "step": 2593
    },
    {
      "epoch": 0.4672821436613375,
      "grad_norm": 1.2804954051971436,
      "learning_rate": 8.719638289604726e-05,
      "loss": 1.5078,
      "step": 2594
    },
    {
      "epoch": 0.4674622832695339,
      "grad_norm": 1.462478518486023,
      "learning_rate": 8.718691782144908e-05,
      "loss": 1.6792,
      "step": 2595
    },
    {
      "epoch": 0.4676424228777302,
      "grad_norm": 1.4092472791671753,
      "learning_rate": 8.717744976373447e-05,
      "loss": 1.6286,
      "step": 2596
    },
    {
      "epoch": 0.4678225624859266,
      "grad_norm": 1.367961049079895,
      "learning_rate": 8.716797872366293e-05,
      "loss": 1.3877,
      "step": 2597
    },
    {
      "epoch": 0.4680027020941229,
      "grad_norm": 1.4596394300460815,
      "learning_rate": 8.715850470199428e-05,
      "loss": 1.8017,
      "step": 2598
    },
    {
      "epoch": 0.4681828417023193,
      "grad_norm": 1.36800217628479,
      "learning_rate": 8.714902769948846e-05,
      "loss": 1.4744,
      "step": 2599
    },
    {
      "epoch": 0.4683629813105156,
      "grad_norm": 1.3347901105880737,
      "learning_rate": 8.713954771690572e-05,
      "loss": 1.4369,
      "step": 2600
    },
    {
      "epoch": 0.468543120918712,
      "grad_norm": 2.5180611610412598,
      "learning_rate": 8.713006475500657e-05,
      "loss": 2.0333,
      "step": 2601
    },
    {
      "epoch": 0.4687232605269084,
      "grad_norm": 1.3184151649475098,
      "learning_rate": 8.71205788145517e-05,
      "loss": 2.2133,
      "step": 2602
    },
    {
      "epoch": 0.46890340013510473,
      "grad_norm": 1.5204957723617554,
      "learning_rate": 8.711108989630207e-05,
      "loss": 2.0981,
      "step": 2603
    },
    {
      "epoch": 0.4690835397433011,
      "grad_norm": 1.3316802978515625,
      "learning_rate": 8.710159800101891e-05,
      "loss": 2.0305,
      "step": 2604
    },
    {
      "epoch": 0.46926367935149743,
      "grad_norm": 1.493296504020691,
      "learning_rate": 8.709210312946363e-05,
      "loss": 2.19,
      "step": 2605
    },
    {
      "epoch": 0.4694438189596938,
      "grad_norm": 1.3044030666351318,
      "learning_rate": 8.708260528239788e-05,
      "loss": 1.9925,
      "step": 2606
    },
    {
      "epoch": 0.46962395856789013,
      "grad_norm": 1.5854933261871338,
      "learning_rate": 8.707310446058361e-05,
      "loss": 2.0451,
      "step": 2607
    },
    {
      "epoch": 0.4698040981760865,
      "grad_norm": 1.5118924379348755,
      "learning_rate": 8.706360066478296e-05,
      "loss": 2.2421,
      "step": 2608
    },
    {
      "epoch": 0.46998423778428283,
      "grad_norm": 1.749561071395874,
      "learning_rate": 8.705409389575831e-05,
      "loss": 2.4843,
      "step": 2609
    },
    {
      "epoch": 0.4701643773924792,
      "grad_norm": 1.6904290914535522,
      "learning_rate": 8.70445841542723e-05,
      "loss": 2.1574,
      "step": 2610
    },
    {
      "epoch": 0.47034451700067553,
      "grad_norm": 1.1131564378738403,
      "learning_rate": 8.70350714410878e-05,
      "loss": 1.6179,
      "step": 2611
    },
    {
      "epoch": 0.4705246566088719,
      "grad_norm": 1.2533875703811646,
      "learning_rate": 8.702555575696789e-05,
      "loss": 1.786,
      "step": 2612
    },
    {
      "epoch": 0.47070479621706823,
      "grad_norm": 1.1966012716293335,
      "learning_rate": 8.701603710267593e-05,
      "loss": 1.7227,
      "step": 2613
    },
    {
      "epoch": 0.4708849358252646,
      "grad_norm": 1.196203589439392,
      "learning_rate": 8.700651547897553e-05,
      "loss": 1.8332,
      "step": 2614
    },
    {
      "epoch": 0.47106507543346093,
      "grad_norm": 1.352143406867981,
      "learning_rate": 8.699699088663047e-05,
      "loss": 1.8101,
      "step": 2615
    },
    {
      "epoch": 0.4712452150416573,
      "grad_norm": 1.2338167428970337,
      "learning_rate": 8.698746332640482e-05,
      "loss": 1.5321,
      "step": 2616
    },
    {
      "epoch": 0.47142535464985363,
      "grad_norm": 1.1382373571395874,
      "learning_rate": 8.697793279906288e-05,
      "loss": 1.5785,
      "step": 2617
    },
    {
      "epoch": 0.47160549425805,
      "grad_norm": 1.2017070055007935,
      "learning_rate": 8.696839930536919e-05,
      "loss": 1.4744,
      "step": 2618
    },
    {
      "epoch": 0.47178563386624633,
      "grad_norm": 1.2269024848937988,
      "learning_rate": 8.69588628460885e-05,
      "loss": 1.4611,
      "step": 2619
    },
    {
      "epoch": 0.4719657734744427,
      "grad_norm": 1.2656941413879395,
      "learning_rate": 8.694932342198585e-05,
      "loss": 1.5712,
      "step": 2620
    },
    {
      "epoch": 0.47214591308263903,
      "grad_norm": 1.6248953342437744,
      "learning_rate": 8.693978103382646e-05,
      "loss": 1.6579,
      "step": 2621
    },
    {
      "epoch": 0.4723260526908354,
      "grad_norm": 1.3158235549926758,
      "learning_rate": 8.693023568237582e-05,
      "loss": 1.7402,
      "step": 2622
    },
    {
      "epoch": 0.47250619229903174,
      "grad_norm": 1.175398826599121,
      "learning_rate": 8.692068736839968e-05,
      "loss": 1.4395,
      "step": 2623
    },
    {
      "epoch": 0.4726863319072281,
      "grad_norm": 1.2811686992645264,
      "learning_rate": 8.691113609266398e-05,
      "loss": 1.6473,
      "step": 2624
    },
    {
      "epoch": 0.47286647151542444,
      "grad_norm": 1.2921103239059448,
      "learning_rate": 8.690158185593491e-05,
      "loss": 1.8152,
      "step": 2625
    },
    {
      "epoch": 0.4730466111236208,
      "grad_norm": 1.3153115510940552,
      "learning_rate": 8.689202465897893e-05,
      "loss": 1.728,
      "step": 2626
    },
    {
      "epoch": 0.47322675073181714,
      "grad_norm": 1.215086579322815,
      "learning_rate": 8.688246450256267e-05,
      "loss": 1.4166,
      "step": 2627
    },
    {
      "epoch": 0.4734068903400135,
      "grad_norm": 1.4470634460449219,
      "learning_rate": 8.687290138745313e-05,
      "loss": 1.6833,
      "step": 2628
    },
    {
      "epoch": 0.47358702994820984,
      "grad_norm": 1.2941168546676636,
      "learning_rate": 8.686333531441734e-05,
      "loss": 1.6226,
      "step": 2629
    },
    {
      "epoch": 0.47376716955640624,
      "grad_norm": 1.4168493747711182,
      "learning_rate": 8.685376628422279e-05,
      "loss": 1.8653,
      "step": 2630
    },
    {
      "epoch": 0.4739473091646026,
      "grad_norm": 1.3069216012954712,
      "learning_rate": 8.684419429763703e-05,
      "loss": 1.6843,
      "step": 2631
    },
    {
      "epoch": 0.47412744877279894,
      "grad_norm": 1.3069379329681396,
      "learning_rate": 8.683461935542798e-05,
      "loss": 1.4671,
      "step": 2632
    },
    {
      "epoch": 0.4743075883809953,
      "grad_norm": 1.3075499534606934,
      "learning_rate": 8.682504145836369e-05,
      "loss": 1.7295,
      "step": 2633
    },
    {
      "epoch": 0.47448772798919164,
      "grad_norm": 1.2161720991134644,
      "learning_rate": 8.68154606072125e-05,
      "loss": 1.4187,
      "step": 2634
    },
    {
      "epoch": 0.474667867597388,
      "grad_norm": 1.4876856803894043,
      "learning_rate": 8.680587680274302e-05,
      "loss": 2.0348,
      "step": 2635
    },
    {
      "epoch": 0.47484800720558434,
      "grad_norm": 1.3415518999099731,
      "learning_rate": 8.679629004572401e-05,
      "loss": 1.6867,
      "step": 2636
    },
    {
      "epoch": 0.4750281468137807,
      "grad_norm": 1.288272500038147,
      "learning_rate": 8.678670033692454e-05,
      "loss": 1.4819,
      "step": 2637
    },
    {
      "epoch": 0.47520828642197704,
      "grad_norm": 1.5106487274169922,
      "learning_rate": 8.677710767711389e-05,
      "loss": 2.0366,
      "step": 2638
    },
    {
      "epoch": 0.4753884260301734,
      "grad_norm": 1.5960053205490112,
      "learning_rate": 8.676751206706157e-05,
      "loss": 1.6012,
      "step": 2639
    },
    {
      "epoch": 0.47556856563836974,
      "grad_norm": 1.4855772256851196,
      "learning_rate": 8.675791350753735e-05,
      "loss": 1.988,
      "step": 2640
    },
    {
      "epoch": 0.4757487052465661,
      "grad_norm": 1.5577479600906372,
      "learning_rate": 8.674831199931121e-05,
      "loss": 1.7417,
      "step": 2641
    },
    {
      "epoch": 0.47592884485476245,
      "grad_norm": 1.4745912551879883,
      "learning_rate": 8.673870754315336e-05,
      "loss": 1.6599,
      "step": 2642
    },
    {
      "epoch": 0.4761089844629588,
      "grad_norm": 1.434037685394287,
      "learning_rate": 8.672910013983431e-05,
      "loss": 1.7926,
      "step": 2643
    },
    {
      "epoch": 0.47628912407115515,
      "grad_norm": 1.387356162071228,
      "learning_rate": 8.671948979012474e-05,
      "loss": 1.703,
      "step": 2644
    },
    {
      "epoch": 0.4764692636793515,
      "grad_norm": 1.4359241724014282,
      "learning_rate": 8.670987649479557e-05,
      "loss": 1.6394,
      "step": 2645
    },
    {
      "epoch": 0.47664940328754785,
      "grad_norm": 1.3842597007751465,
      "learning_rate": 8.6700260254618e-05,
      "loss": 1.3689,
      "step": 2646
    },
    {
      "epoch": 0.4768295428957442,
      "grad_norm": 1.463671088218689,
      "learning_rate": 8.66906410703634e-05,
      "loss": 1.5767,
      "step": 2647
    },
    {
      "epoch": 0.47700968250394055,
      "grad_norm": 1.48291015625,
      "learning_rate": 8.668101894280347e-05,
      "loss": 1.6378,
      "step": 2648
    },
    {
      "epoch": 0.4771898221121369,
      "grad_norm": 1.493817925453186,
      "learning_rate": 8.667139387271007e-05,
      "loss": 1.5672,
      "step": 2649
    },
    {
      "epoch": 0.47736996172033325,
      "grad_norm": 1.350455641746521,
      "learning_rate": 8.66617658608553e-05,
      "loss": 1.3829,
      "step": 2650
    },
    {
      "epoch": 0.4775501013285296,
      "grad_norm": 1.316884994506836,
      "learning_rate": 8.665213490801154e-05,
      "loss": 2.2399,
      "step": 2651
    },
    {
      "epoch": 0.47773024093672595,
      "grad_norm": 1.3907532691955566,
      "learning_rate": 8.664250101495137e-05,
      "loss": 2.1095,
      "step": 2652
    },
    {
      "epoch": 0.4779103805449223,
      "grad_norm": 2.0396082401275635,
      "learning_rate": 8.663286418244763e-05,
      "loss": 2.266,
      "step": 2653
    },
    {
      "epoch": 0.47809052015311865,
      "grad_norm": 1.4065428972244263,
      "learning_rate": 8.662322441127334e-05,
      "loss": 2.2639,
      "step": 2654
    },
    {
      "epoch": 0.478270659761315,
      "grad_norm": 1.420646071434021,
      "learning_rate": 8.661358170220183e-05,
      "loss": 2.092,
      "step": 2655
    },
    {
      "epoch": 0.47845079936951135,
      "grad_norm": 1.272267460823059,
      "learning_rate": 8.660393605600666e-05,
      "loss": 1.9187,
      "step": 2656
    },
    {
      "epoch": 0.4786309389777077,
      "grad_norm": 1.359609842300415,
      "learning_rate": 8.659428747346154e-05,
      "loss": 1.9691,
      "step": 2657
    },
    {
      "epoch": 0.47881107858590405,
      "grad_norm": 1.4266998767852783,
      "learning_rate": 8.658463595534051e-05,
      "loss": 1.9072,
      "step": 2658
    },
    {
      "epoch": 0.47899121819410045,
      "grad_norm": 1.6890134811401367,
      "learning_rate": 8.657498150241781e-05,
      "loss": 2.1801,
      "step": 2659
    },
    {
      "epoch": 0.4791713578022968,
      "grad_norm": 1.6149895191192627,
      "learning_rate": 8.65653241154679e-05,
      "loss": 2.0892,
      "step": 2660
    },
    {
      "epoch": 0.47935149741049315,
      "grad_norm": 1.251991629600525,
      "learning_rate": 8.655566379526551e-05,
      "loss": 1.8521,
      "step": 2661
    },
    {
      "epoch": 0.4795316370186895,
      "grad_norm": 1.3477108478546143,
      "learning_rate": 8.654600054258557e-05,
      "loss": 1.8777,
      "step": 2662
    },
    {
      "epoch": 0.47971177662688586,
      "grad_norm": 1.3829909563064575,
      "learning_rate": 8.653633435820326e-05,
      "loss": 1.7965,
      "step": 2663
    },
    {
      "epoch": 0.4798919162350822,
      "grad_norm": 1.1786986589431763,
      "learning_rate": 8.6526665242894e-05,
      "loss": 1.5584,
      "step": 2664
    },
    {
      "epoch": 0.48007205584327856,
      "grad_norm": 1.1727650165557861,
      "learning_rate": 8.651699319743347e-05,
      "loss": 1.4325,
      "step": 2665
    },
    {
      "epoch": 0.4802521954514749,
      "grad_norm": 1.2339653968811035,
      "learning_rate": 8.650731822259753e-05,
      "loss": 1.6374,
      "step": 2666
    },
    {
      "epoch": 0.48043233505967126,
      "grad_norm": 1.2392654418945312,
      "learning_rate": 8.649764031916227e-05,
      "loss": 1.6147,
      "step": 2667
    },
    {
      "epoch": 0.4806124746678676,
      "grad_norm": 1.35507333278656,
      "learning_rate": 8.648795948790411e-05,
      "loss": 1.6223,
      "step": 2668
    },
    {
      "epoch": 0.48079261427606396,
      "grad_norm": 1.3900197744369507,
      "learning_rate": 8.647827572959962e-05,
      "loss": 1.8806,
      "step": 2669
    },
    {
      "epoch": 0.4809727538842603,
      "grad_norm": 1.4021410942077637,
      "learning_rate": 8.646858904502559e-05,
      "loss": 1.9184,
      "step": 2670
    },
    {
      "epoch": 0.48115289349245666,
      "grad_norm": 1.2401812076568604,
      "learning_rate": 8.645889943495913e-05,
      "loss": 1.6602,
      "step": 2671
    },
    {
      "epoch": 0.481333033100653,
      "grad_norm": 1.418326497077942,
      "learning_rate": 8.644920690017752e-05,
      "loss": 1.5734,
      "step": 2672
    },
    {
      "epoch": 0.48151317270884936,
      "grad_norm": 1.2083072662353516,
      "learning_rate": 8.643951144145828e-05,
      "loss": 1.5713,
      "step": 2673
    },
    {
      "epoch": 0.4816933123170457,
      "grad_norm": 1.2850606441497803,
      "learning_rate": 8.642981305957919e-05,
      "loss": 1.7279,
      "step": 2674
    },
    {
      "epoch": 0.48187345192524206,
      "grad_norm": 1.326866626739502,
      "learning_rate": 8.642011175531822e-05,
      "loss": 1.7855,
      "step": 2675
    },
    {
      "epoch": 0.4820535915334384,
      "grad_norm": 1.4067503213882446,
      "learning_rate": 8.641040752945363e-05,
      "loss": 1.7365,
      "step": 2676
    },
    {
      "epoch": 0.48223373114163476,
      "grad_norm": 1.415247917175293,
      "learning_rate": 8.640070038276388e-05,
      "loss": 1.5787,
      "step": 2677
    },
    {
      "epoch": 0.4824138707498311,
      "grad_norm": 1.2646092176437378,
      "learning_rate": 8.639099031602769e-05,
      "loss": 1.4971,
      "step": 2678
    },
    {
      "epoch": 0.48259401035802746,
      "grad_norm": 1.294352412223816,
      "learning_rate": 8.638127733002397e-05,
      "loss": 1.5892,
      "step": 2679
    },
    {
      "epoch": 0.4827741499662238,
      "grad_norm": 1.2876085042953491,
      "learning_rate": 8.637156142553191e-05,
      "loss": 1.5463,
      "step": 2680
    },
    {
      "epoch": 0.48295428957442016,
      "grad_norm": 1.224370002746582,
      "learning_rate": 8.636184260333091e-05,
      "loss": 1.651,
      "step": 2681
    },
    {
      "epoch": 0.4831344291826165,
      "grad_norm": 1.3199914693832397,
      "learning_rate": 8.635212086420061e-05,
      "loss": 1.6555,
      "step": 2682
    },
    {
      "epoch": 0.48331456879081286,
      "grad_norm": 1.2267916202545166,
      "learning_rate": 8.634239620892088e-05,
      "loss": 1.5521,
      "step": 2683
    },
    {
      "epoch": 0.4834947083990092,
      "grad_norm": 1.3754695653915405,
      "learning_rate": 8.633266863827181e-05,
      "loss": 1.6656,
      "step": 2684
    },
    {
      "epoch": 0.48367484800720556,
      "grad_norm": 1.2926530838012695,
      "learning_rate": 8.632293815303378e-05,
      "loss": 1.566,
      "step": 2685
    },
    {
      "epoch": 0.4838549876154019,
      "grad_norm": 1.33713960647583,
      "learning_rate": 8.631320475398734e-05,
      "loss": 1.727,
      "step": 2686
    },
    {
      "epoch": 0.4840351272235983,
      "grad_norm": 1.3771313428878784,
      "learning_rate": 8.630346844191329e-05,
      "loss": 1.509,
      "step": 2687
    },
    {
      "epoch": 0.48421526683179467,
      "grad_norm": 1.4455842971801758,
      "learning_rate": 8.629372921759269e-05,
      "loss": 1.7759,
      "step": 2688
    },
    {
      "epoch": 0.484395406439991,
      "grad_norm": 1.2864117622375488,
      "learning_rate": 8.628398708180681e-05,
      "loss": 1.6044,
      "step": 2689
    },
    {
      "epoch": 0.48457554604818737,
      "grad_norm": 1.3856821060180664,
      "learning_rate": 8.627424203533716e-05,
      "loss": 1.9426,
      "step": 2690
    },
    {
      "epoch": 0.4847556856563837,
      "grad_norm": 1.5106086730957031,
      "learning_rate": 8.626449407896548e-05,
      "loss": 1.7021,
      "step": 2691
    },
    {
      "epoch": 0.48493582526458007,
      "grad_norm": 1.3311948776245117,
      "learning_rate": 8.625474321347376e-05,
      "loss": 1.8208,
      "step": 2692
    },
    {
      "epoch": 0.4851159648727764,
      "grad_norm": 1.3760781288146973,
      "learning_rate": 8.624498943964419e-05,
      "loss": 1.6965,
      "step": 2693
    },
    {
      "epoch": 0.48529610448097277,
      "grad_norm": 1.385847806930542,
      "learning_rate": 8.623523275825922e-05,
      "loss": 1.5341,
      "step": 2694
    },
    {
      "epoch": 0.4854762440891691,
      "grad_norm": 1.301340937614441,
      "learning_rate": 8.622547317010154e-05,
      "loss": 1.5803,
      "step": 2695
    },
    {
      "epoch": 0.48565638369736547,
      "grad_norm": 1.405910611152649,
      "learning_rate": 8.621571067595404e-05,
      "loss": 1.7003,
      "step": 2696
    },
    {
      "epoch": 0.4858365233055618,
      "grad_norm": 1.2262637615203857,
      "learning_rate": 8.620594527659987e-05,
      "loss": 1.3236,
      "step": 2697
    },
    {
      "epoch": 0.48601666291375817,
      "grad_norm": 1.442098617553711,
      "learning_rate": 8.61961769728224e-05,
      "loss": 1.5305,
      "step": 2698
    },
    {
      "epoch": 0.4861968025219545,
      "grad_norm": 1.2316488027572632,
      "learning_rate": 8.618640576540525e-05,
      "loss": 1.3083,
      "step": 2699
    },
    {
      "epoch": 0.48637694213015087,
      "grad_norm": 1.3544647693634033,
      "learning_rate": 8.617663165513223e-05,
      "loss": 1.6582,
      "step": 2700
    },
    {
      "epoch": 0.4865570817383472,
      "grad_norm": 1.1886539459228516,
      "learning_rate": 8.616685464278748e-05,
      "loss": 2.0696,
      "step": 2701
    },
    {
      "epoch": 0.48673722134654357,
      "grad_norm": 1.2951563596725464,
      "learning_rate": 8.615707472915525e-05,
      "loss": 2.2145,
      "step": 2702
    },
    {
      "epoch": 0.4869173609547399,
      "grad_norm": 1.3158202171325684,
      "learning_rate": 8.614729191502008e-05,
      "loss": 2.1534,
      "step": 2703
    },
    {
      "epoch": 0.48709750056293627,
      "grad_norm": 1.250584602355957,
      "learning_rate": 8.613750620116679e-05,
      "loss": 1.9095,
      "step": 2704
    },
    {
      "epoch": 0.4872776401711326,
      "grad_norm": 1.3280251026153564,
      "learning_rate": 8.612771758838033e-05,
      "loss": 2.0795,
      "step": 2705
    },
    {
      "epoch": 0.48745777977932897,
      "grad_norm": 1.3332096338272095,
      "learning_rate": 8.611792607744597e-05,
      "loss": 1.896,
      "step": 2706
    },
    {
      "epoch": 0.4876379193875253,
      "grad_norm": 1.3256202936172485,
      "learning_rate": 8.610813166914917e-05,
      "loss": 1.8772,
      "step": 2707
    },
    {
      "epoch": 0.48781805899572167,
      "grad_norm": 1.649614930152893,
      "learning_rate": 8.609833436427563e-05,
      "loss": 2.2148,
      "step": 2708
    },
    {
      "epoch": 0.487998198603918,
      "grad_norm": 1.8630517721176147,
      "learning_rate": 8.608853416361131e-05,
      "loss": 2.4517,
      "step": 2709
    },
    {
      "epoch": 0.48817833821211437,
      "grad_norm": 1.3139984607696533,
      "learning_rate": 8.607873106794235e-05,
      "loss": 1.7884,
      "step": 2710
    },
    {
      "epoch": 0.4883584778203107,
      "grad_norm": 1.335869550704956,
      "learning_rate": 8.606892507805515e-05,
      "loss": 1.7496,
      "step": 2711
    },
    {
      "epoch": 0.48853861742850707,
      "grad_norm": 1.2606089115142822,
      "learning_rate": 8.605911619473635e-05,
      "loss": 1.597,
      "step": 2712
    },
    {
      "epoch": 0.4887187570367034,
      "grad_norm": 1.2731069326400757,
      "learning_rate": 8.604930441877281e-05,
      "loss": 1.4671,
      "step": 2713
    },
    {
      "epoch": 0.48889889664489977,
      "grad_norm": 1.2373595237731934,
      "learning_rate": 8.603948975095165e-05,
      "loss": 1.5107,
      "step": 2714
    },
    {
      "epoch": 0.4890790362530961,
      "grad_norm": 1.2658288478851318,
      "learning_rate": 8.602967219206015e-05,
      "loss": 1.7882,
      "step": 2715
    },
    {
      "epoch": 0.48925917586129253,
      "grad_norm": 1.2807252407073975,
      "learning_rate": 8.60198517428859e-05,
      "loss": 1.5173,
      "step": 2716
    },
    {
      "epoch": 0.4894393154694889,
      "grad_norm": 1.3331763744354248,
      "learning_rate": 8.601002840421671e-05,
      "loss": 1.6888,
      "step": 2717
    },
    {
      "epoch": 0.48961945507768523,
      "grad_norm": 1.3163950443267822,
      "learning_rate": 8.600020217684055e-05,
      "loss": 1.7147,
      "step": 2718
    },
    {
      "epoch": 0.4897995946858816,
      "grad_norm": 1.1635547876358032,
      "learning_rate": 8.599037306154572e-05,
      "loss": 1.5312,
      "step": 2719
    },
    {
      "epoch": 0.48997973429407793,
      "grad_norm": 1.3463550806045532,
      "learning_rate": 8.59805410591207e-05,
      "loss": 1.7128,
      "step": 2720
    },
    {
      "epoch": 0.4901598739022743,
      "grad_norm": 1.1776913404464722,
      "learning_rate": 8.597070617035419e-05,
      "loss": 1.4958,
      "step": 2721
    },
    {
      "epoch": 0.49034001351047063,
      "grad_norm": 1.2490227222442627,
      "learning_rate": 8.596086839603517e-05,
      "loss": 1.5679,
      "step": 2722
    },
    {
      "epoch": 0.490520153118667,
      "grad_norm": 1.2558070421218872,
      "learning_rate": 8.595102773695278e-05,
      "loss": 1.5767,
      "step": 2723
    },
    {
      "epoch": 0.49070029272686333,
      "grad_norm": 1.3319237232208252,
      "learning_rate": 8.594118419389647e-05,
      "loss": 1.9236,
      "step": 2724
    },
    {
      "epoch": 0.4908804323350597,
      "grad_norm": 1.2993637323379517,
      "learning_rate": 8.593133776765587e-05,
      "loss": 1.8092,
      "step": 2725
    },
    {
      "epoch": 0.49106057194325603,
      "grad_norm": 1.3291493654251099,
      "learning_rate": 8.592148845902086e-05,
      "loss": 1.677,
      "step": 2726
    },
    {
      "epoch": 0.4912407115514524,
      "grad_norm": 1.222584843635559,
      "learning_rate": 8.591163626878153e-05,
      "loss": 1.6296,
      "step": 2727
    },
    {
      "epoch": 0.49142085115964873,
      "grad_norm": 1.2929731607437134,
      "learning_rate": 8.590178119772824e-05,
      "loss": 1.5407,
      "step": 2728
    },
    {
      "epoch": 0.4916009907678451,
      "grad_norm": 1.325379729270935,
      "learning_rate": 8.589192324665154e-05,
      "loss": 1.7361,
      "step": 2729
    },
    {
      "epoch": 0.49178113037604143,
      "grad_norm": 1.2696601152420044,
      "learning_rate": 8.588206241634225e-05,
      "loss": 1.5384,
      "step": 2730
    },
    {
      "epoch": 0.4919612699842378,
      "grad_norm": 1.283871054649353,
      "learning_rate": 8.587219870759138e-05,
      "loss": 1.8387,
      "step": 2731
    },
    {
      "epoch": 0.49214140959243413,
      "grad_norm": 1.4020068645477295,
      "learning_rate": 8.586233212119021e-05,
      "loss": 1.8382,
      "step": 2732
    },
    {
      "epoch": 0.4923215492006305,
      "grad_norm": 1.1911858320236206,
      "learning_rate": 8.585246265793023e-05,
      "loss": 1.4332,
      "step": 2733
    },
    {
      "epoch": 0.49250168880882683,
      "grad_norm": 1.401396632194519,
      "learning_rate": 8.584259031860315e-05,
      "loss": 1.7361,
      "step": 2734
    },
    {
      "epoch": 0.4926818284170232,
      "grad_norm": 1.230306625366211,
      "learning_rate": 8.583271510400093e-05,
      "loss": 1.5654,
      "step": 2735
    },
    {
      "epoch": 0.49286196802521953,
      "grad_norm": 1.318452000617981,
      "learning_rate": 8.582283701491576e-05,
      "loss": 1.8634,
      "step": 2736
    },
    {
      "epoch": 0.4930421076334159,
      "grad_norm": 1.3202909231185913,
      "learning_rate": 8.581295605214005e-05,
      "loss": 1.5214,
      "step": 2737
    },
    {
      "epoch": 0.49322224724161223,
      "grad_norm": 1.329823613166809,
      "learning_rate": 8.580307221646647e-05,
      "loss": 1.5694,
      "step": 2738
    },
    {
      "epoch": 0.4934023868498086,
      "grad_norm": 1.308545708656311,
      "learning_rate": 8.579318550868786e-05,
      "loss": 1.5803,
      "step": 2739
    },
    {
      "epoch": 0.49358252645800493,
      "grad_norm": 1.420293927192688,
      "learning_rate": 8.578329592959734e-05,
      "loss": 1.9692,
      "step": 2740
    },
    {
      "epoch": 0.4937626660662013,
      "grad_norm": 1.4217355251312256,
      "learning_rate": 8.577340347998825e-05,
      "loss": 1.4416,
      "step": 2741
    },
    {
      "epoch": 0.49394280567439763,
      "grad_norm": 1.490926742553711,
      "learning_rate": 8.576350816065417e-05,
      "loss": 1.7305,
      "step": 2742
    },
    {
      "epoch": 0.494122945282594,
      "grad_norm": 1.429819941520691,
      "learning_rate": 8.575360997238887e-05,
      "loss": 1.5678,
      "step": 2743
    },
    {
      "epoch": 0.4943030848907904,
      "grad_norm": 1.4071948528289795,
      "learning_rate": 8.57437089159864e-05,
      "loss": 1.5747,
      "step": 2744
    },
    {
      "epoch": 0.49448322449898674,
      "grad_norm": 1.3541555404663086,
      "learning_rate": 8.573380499224101e-05,
      "loss": 1.5497,
      "step": 2745
    },
    {
      "epoch": 0.4946633641071831,
      "grad_norm": 1.4640523195266724,
      "learning_rate": 8.572389820194719e-05,
      "loss": 1.8185,
      "step": 2746
    },
    {
      "epoch": 0.49484350371537944,
      "grad_norm": 1.3823490142822266,
      "learning_rate": 8.571398854589966e-05,
      "loss": 1.6214,
      "step": 2747
    },
    {
      "epoch": 0.4950236433235758,
      "grad_norm": 1.3801867961883545,
      "learning_rate": 8.570407602489335e-05,
      "loss": 1.4755,
      "step": 2748
    },
    {
      "epoch": 0.49520378293177214,
      "grad_norm": 1.57958984375,
      "learning_rate": 8.569416063972347e-05,
      "loss": 1.7264,
      "step": 2749
    },
    {
      "epoch": 0.4953839225399685,
      "grad_norm": 1.2653738260269165,
      "learning_rate": 8.56842423911854e-05,
      "loss": 1.3777,
      "step": 2750
    },
    {
      "epoch": 0.49556406214816484,
      "grad_norm": 1.2235174179077148,
      "learning_rate": 8.56743212800748e-05,
      "loss": 2.1362,
      "step": 2751
    },
    {
      "epoch": 0.4957442017563612,
      "grad_norm": 1.1713038682937622,
      "learning_rate": 8.566439730718749e-05,
      "loss": 1.9403,
      "step": 2752
    },
    {
      "epoch": 0.49592434136455754,
      "grad_norm": 1.2549755573272705,
      "learning_rate": 8.565447047331962e-05,
      "loss": 2.2198,
      "step": 2753
    },
    {
      "epoch": 0.4961044809727539,
      "grad_norm": 1.2807815074920654,
      "learning_rate": 8.564454077926751e-05,
      "loss": 2.0376,
      "step": 2754
    },
    {
      "epoch": 0.49628462058095024,
      "grad_norm": 1.4784092903137207,
      "learning_rate": 8.56346082258277e-05,
      "loss": 2.4307,
      "step": 2755
    },
    {
      "epoch": 0.4964647601891466,
      "grad_norm": 1.3252829313278198,
      "learning_rate": 8.562467281379695e-05,
      "loss": 2.0048,
      "step": 2756
    },
    {
      "epoch": 0.49664489979734294,
      "grad_norm": 1.576277494430542,
      "learning_rate": 8.56147345439723e-05,
      "loss": 2.0626,
      "step": 2757
    },
    {
      "epoch": 0.4968250394055393,
      "grad_norm": 1.568812608718872,
      "learning_rate": 8.560479341715102e-05,
      "loss": 1.934,
      "step": 2758
    },
    {
      "epoch": 0.49700517901373564,
      "grad_norm": 1.515198826789856,
      "learning_rate": 8.559484943413053e-05,
      "loss": 1.9284,
      "step": 2759
    },
    {
      "epoch": 0.497185318621932,
      "grad_norm": 1.9978443384170532,
      "learning_rate": 8.558490259570857e-05,
      "loss": 2.2414,
      "step": 2760
    },
    {
      "epoch": 0.49736545823012834,
      "grad_norm": 1.7489478588104248,
      "learning_rate": 8.557495290268305e-05,
      "loss": 2.1514,
      "step": 2761
    },
    {
      "epoch": 0.4975455978383247,
      "grad_norm": 1.2738151550292969,
      "learning_rate": 8.556500035585213e-05,
      "loss": 1.8158,
      "step": 2762
    },
    {
      "epoch": 0.49772573744652104,
      "grad_norm": 1.302860140800476,
      "learning_rate": 8.55550449560142e-05,
      "loss": 1.7584,
      "step": 2763
    },
    {
      "epoch": 0.4979058770547174,
      "grad_norm": 1.2008062601089478,
      "learning_rate": 8.554508670396791e-05,
      "loss": 1.6243,
      "step": 2764
    },
    {
      "epoch": 0.49808601666291374,
      "grad_norm": 1.253883957862854,
      "learning_rate": 8.553512560051206e-05,
      "loss": 1.4351,
      "step": 2765
    },
    {
      "epoch": 0.4982661562711101,
      "grad_norm": 1.179694652557373,
      "learning_rate": 8.552516164644575e-05,
      "loss": 1.4304,
      "step": 2766
    },
    {
      "epoch": 0.49844629587930644,
      "grad_norm": 1.204978585243225,
      "learning_rate": 8.551519484256826e-05,
      "loss": 1.4948,
      "step": 2767
    },
    {
      "epoch": 0.4986264354875028,
      "grad_norm": 1.367863655090332,
      "learning_rate": 8.550522518967915e-05,
      "loss": 1.715,
      "step": 2768
    },
    {
      "epoch": 0.49880657509569915,
      "grad_norm": 1.2354141473770142,
      "learning_rate": 8.549525268857816e-05,
      "loss": 1.4642,
      "step": 2769
    },
    {
      "epoch": 0.4989867147038955,
      "grad_norm": 1.155656099319458,
      "learning_rate": 8.54852773400653e-05,
      "loss": 1.3461,
      "step": 2770
    },
    {
      "epoch": 0.49916685431209185,
      "grad_norm": 1.3603318929672241,
      "learning_rate": 8.547529914494078e-05,
      "loss": 1.9171,
      "step": 2771
    },
    {
      "epoch": 0.4993469939202882,
      "grad_norm": 1.308967113494873,
      "learning_rate": 8.546531810400503e-05,
      "loss": 1.6232,
      "step": 2772
    },
    {
      "epoch": 0.4995271335284846,
      "grad_norm": 1.3593995571136475,
      "learning_rate": 8.545533421805875e-05,
      "loss": 1.8749,
      "step": 2773
    },
    {
      "epoch": 0.49970727313668095,
      "grad_norm": 1.3304009437561035,
      "learning_rate": 8.544534748790281e-05,
      "loss": 1.8702,
      "step": 2774
    },
    {
      "epoch": 0.4998874127448773,
      "grad_norm": 1.2001197338104248,
      "learning_rate": 8.543535791433838e-05,
      "loss": 1.4296,
      "step": 2775
    },
    {
      "epoch": 0.5000675523530737,
      "grad_norm": 1.204925537109375,
      "learning_rate": 8.542536549816678e-05,
      "loss": 1.5712,
      "step": 2776
    },
    {
      "epoch": 0.50024769196127,
      "grad_norm": 1.2322618961334229,
      "learning_rate": 8.541537024018962e-05,
      "loss": 1.4776,
      "step": 2777
    },
    {
      "epoch": 0.5004278315694664,
      "grad_norm": 1.3218423128128052,
      "learning_rate": 8.540537214120872e-05,
      "loss": 1.6754,
      "step": 2778
    },
    {
      "epoch": 0.5006079711776626,
      "grad_norm": 1.4041993618011475,
      "learning_rate": 8.53953712020261e-05,
      "loss": 1.8543,
      "step": 2779
    },
    {
      "epoch": 0.500788110785859,
      "grad_norm": 1.2976808547973633,
      "learning_rate": 8.538536742344406e-05,
      "loss": 1.6144,
      "step": 2780
    },
    {
      "epoch": 0.5009682503940553,
      "grad_norm": 1.2562553882598877,
      "learning_rate": 8.537536080626505e-05,
      "loss": 1.5163,
      "step": 2781
    },
    {
      "epoch": 0.5011483900022518,
      "grad_norm": 1.319531798362732,
      "learning_rate": 8.536535135129184e-05,
      "loss": 1.6357,
      "step": 2782
    },
    {
      "epoch": 0.501328529610448,
      "grad_norm": 1.371364712715149,
      "learning_rate": 8.535533905932738e-05,
      "loss": 1.5351,
      "step": 2783
    },
    {
      "epoch": 0.5015086692186445,
      "grad_norm": 1.1721785068511963,
      "learning_rate": 8.534532393117484e-05,
      "loss": 1.555,
      "step": 2784
    },
    {
      "epoch": 0.5016888088268407,
      "grad_norm": 1.2747962474822998,
      "learning_rate": 8.533530596763763e-05,
      "loss": 1.5807,
      "step": 2785
    },
    {
      "epoch": 0.5018689484350372,
      "grad_norm": 1.2756599187850952,
      "learning_rate": 8.532528516951939e-05,
      "loss": 1.547,
      "step": 2786
    },
    {
      "epoch": 0.5020490880432336,
      "grad_norm": 1.349286675453186,
      "learning_rate": 8.531526153762396e-05,
      "loss": 1.4839,
      "step": 2787
    },
    {
      "epoch": 0.5022292276514299,
      "grad_norm": 1.324385643005371,
      "learning_rate": 8.530523507275546e-05,
      "loss": 1.4951,
      "step": 2788
    },
    {
      "epoch": 0.5024093672596263,
      "grad_norm": 1.485912799835205,
      "learning_rate": 8.52952057757182e-05,
      "loss": 1.6979,
      "step": 2789
    },
    {
      "epoch": 0.5025895068678226,
      "grad_norm": 1.3031331300735474,
      "learning_rate": 8.528517364731673e-05,
      "loss": 1.4363,
      "step": 2790
    },
    {
      "epoch": 0.502769646476019,
      "grad_norm": 1.3996716737747192,
      "learning_rate": 8.527513868835582e-05,
      "loss": 1.7178,
      "step": 2791
    },
    {
      "epoch": 0.5029497860842153,
      "grad_norm": 1.4848968982696533,
      "learning_rate": 8.526510089964045e-05,
      "loss": 1.662,
      "step": 2792
    },
    {
      "epoch": 0.5031299256924117,
      "grad_norm": 1.3520406484603882,
      "learning_rate": 8.525506028197589e-05,
      "loss": 1.7115,
      "step": 2793
    },
    {
      "epoch": 0.503310065300608,
      "grad_norm": 1.38235604763031,
      "learning_rate": 8.524501683616755e-05,
      "loss": 1.6243,
      "step": 2794
    },
    {
      "epoch": 0.5034902049088044,
      "grad_norm": 1.3430402278900146,
      "learning_rate": 8.523497056302112e-05,
      "loss": 1.4513,
      "step": 2795
    },
    {
      "epoch": 0.5036703445170007,
      "grad_norm": 1.3040351867675781,
      "learning_rate": 8.522492146334251e-05,
      "loss": 1.5241,
      "step": 2796
    },
    {
      "epoch": 0.5038504841251971,
      "grad_norm": 1.4564930200576782,
      "learning_rate": 8.521486953793787e-05,
      "loss": 1.8308,
      "step": 2797
    },
    {
      "epoch": 0.5040306237333934,
      "grad_norm": 1.4538522958755493,
      "learning_rate": 8.520481478761355e-05,
      "loss": 1.8094,
      "step": 2798
    },
    {
      "epoch": 0.5042107633415898,
      "grad_norm": 1.3144654035568237,
      "learning_rate": 8.519475721317614e-05,
      "loss": 1.4744,
      "step": 2799
    },
    {
      "epoch": 0.5043909029497861,
      "grad_norm": 1.4392014741897583,
      "learning_rate": 8.518469681543244e-05,
      "loss": 1.5725,
      "step": 2800
    },
    {
      "epoch": 0.5045710425579825,
      "grad_norm": 1.2421153783798218,
      "learning_rate": 8.517463359518949e-05,
      "loss": 2.1194,
      "step": 2801
    },
    {
      "epoch": 0.5047511821661788,
      "grad_norm": 1.2600113153457642,
      "learning_rate": 8.516456755325457e-05,
      "loss": 1.9421,
      "step": 2802
    },
    {
      "epoch": 0.5049313217743752,
      "grad_norm": 1.2511544227600098,
      "learning_rate": 8.515449869043518e-05,
      "loss": 2.0808,
      "step": 2803
    },
    {
      "epoch": 0.5051114613825715,
      "grad_norm": 1.3157927989959717,
      "learning_rate": 8.514442700753902e-05,
      "loss": 1.9794,
      "step": 2804
    },
    {
      "epoch": 0.5052916009907679,
      "grad_norm": 1.231649398803711,
      "learning_rate": 8.513435250537405e-05,
      "loss": 1.9295,
      "step": 2805
    },
    {
      "epoch": 0.5054717405989642,
      "grad_norm": 1.3225748538970947,
      "learning_rate": 8.512427518474844e-05,
      "loss": 1.8283,
      "step": 2806
    },
    {
      "epoch": 0.5056518802071606,
      "grad_norm": 1.4478939771652222,
      "learning_rate": 8.511419504647056e-05,
      "loss": 2.4938,
      "step": 2807
    },
    {
      "epoch": 0.5058320198153569,
      "grad_norm": 1.4256596565246582,
      "learning_rate": 8.510411209134908e-05,
      "loss": 1.9922,
      "step": 2808
    },
    {
      "epoch": 0.5060121594235533,
      "grad_norm": 1.8600107431411743,
      "learning_rate": 8.509402632019282e-05,
      "loss": 2.0058,
      "step": 2809
    },
    {
      "epoch": 0.5061922990317496,
      "grad_norm": 2.1335132122039795,
      "learning_rate": 8.508393773381085e-05,
      "loss": 2.6432,
      "step": 2810
    },
    {
      "epoch": 0.506372438639946,
      "grad_norm": 1.2174696922302246,
      "learning_rate": 8.507384633301249e-05,
      "loss": 1.4564,
      "step": 2811
    },
    {
      "epoch": 0.5065525782481423,
      "grad_norm": 1.5074141025543213,
      "learning_rate": 8.506375211860728e-05,
      "loss": 1.6933,
      "step": 2812
    },
    {
      "epoch": 0.5067327178563387,
      "grad_norm": 1.2732199430465698,
      "learning_rate": 8.505365509140494e-05,
      "loss": 1.6016,
      "step": 2813
    },
    {
      "epoch": 0.506912857464535,
      "grad_norm": 1.3593729734420776,
      "learning_rate": 8.504355525221545e-05,
      "loss": 1.6226,
      "step": 2814
    },
    {
      "epoch": 0.5070929970727314,
      "grad_norm": 1.4237853288650513,
      "learning_rate": 8.503345260184905e-05,
      "loss": 1.5724,
      "step": 2815
    },
    {
      "epoch": 0.5072731366809278,
      "grad_norm": 1.2254680395126343,
      "learning_rate": 8.502334714111612e-05,
      "loss": 1.3983,
      "step": 2816
    },
    {
      "epoch": 0.5074532762891241,
      "grad_norm": 1.4216575622558594,
      "learning_rate": 8.501323887082737e-05,
      "loss": 1.5055,
      "step": 2817
    },
    {
      "epoch": 0.5076334158973205,
      "grad_norm": 1.3354389667510986,
      "learning_rate": 8.500312779179365e-05,
      "loss": 1.7029,
      "step": 2818
    },
    {
      "epoch": 0.5078135555055168,
      "grad_norm": 1.3727309703826904,
      "learning_rate": 8.499301390482605e-05,
      "loss": 1.675,
      "step": 2819
    },
    {
      "epoch": 0.5079936951137132,
      "grad_norm": 1.2982864379882812,
      "learning_rate": 8.498289721073593e-05,
      "loss": 1.6344,
      "step": 2820
    },
    {
      "epoch": 0.5081738347219095,
      "grad_norm": 1.417824625968933,
      "learning_rate": 8.497277771033482e-05,
      "loss": 1.6598,
      "step": 2821
    },
    {
      "epoch": 0.5083539743301059,
      "grad_norm": 1.2452620267868042,
      "learning_rate": 8.496265540443453e-05,
      "loss": 1.4911,
      "step": 2822
    },
    {
      "epoch": 0.5085341139383022,
      "grad_norm": 1.3113662004470825,
      "learning_rate": 8.495253029384705e-05,
      "loss": 1.8271,
      "step": 2823
    },
    {
      "epoch": 0.5087142535464986,
      "grad_norm": 1.2392932176589966,
      "learning_rate": 8.494240237938461e-05,
      "loss": 1.561,
      "step": 2824
    },
    {
      "epoch": 0.5088943931546949,
      "grad_norm": 1.3330018520355225,
      "learning_rate": 8.493227166185968e-05,
      "loss": 1.5807,
      "step": 2825
    },
    {
      "epoch": 0.5090745327628913,
      "grad_norm": 1.315157175064087,
      "learning_rate": 8.492213814208494e-05,
      "loss": 1.5571,
      "step": 2826
    },
    {
      "epoch": 0.5092546723710876,
      "grad_norm": 1.2412223815917969,
      "learning_rate": 8.491200182087326e-05,
      "loss": 1.5673,
      "step": 2827
    },
    {
      "epoch": 0.509434811979284,
      "grad_norm": 1.2737836837768555,
      "learning_rate": 8.490186269903782e-05,
      "loss": 1.6543,
      "step": 2828
    },
    {
      "epoch": 0.5096149515874803,
      "grad_norm": 1.2240816354751587,
      "learning_rate": 8.489172077739195e-05,
      "loss": 1.6392,
      "step": 2829
    },
    {
      "epoch": 0.5097950911956767,
      "grad_norm": 1.350205659866333,
      "learning_rate": 8.488157605674925e-05,
      "loss": 1.7706,
      "step": 2830
    },
    {
      "epoch": 0.509975230803873,
      "grad_norm": 1.1772468090057373,
      "learning_rate": 8.487142853792348e-05,
      "loss": 1.4376,
      "step": 2831
    },
    {
      "epoch": 0.5101553704120694,
      "grad_norm": 1.2457062005996704,
      "learning_rate": 8.486127822172871e-05,
      "loss": 1.5642,
      "step": 2832
    },
    {
      "epoch": 0.5103355100202657,
      "grad_norm": 1.4030654430389404,
      "learning_rate": 8.485112510897918e-05,
      "loss": 1.7351,
      "step": 2833
    },
    {
      "epoch": 0.5105156496284621,
      "grad_norm": 1.4837732315063477,
      "learning_rate": 8.484096920048939e-05,
      "loss": 1.7033,
      "step": 2834
    },
    {
      "epoch": 0.5106957892366584,
      "grad_norm": 1.4558892250061035,
      "learning_rate": 8.4830810497074e-05,
      "loss": 1.9693,
      "step": 2835
    },
    {
      "epoch": 0.5108759288448548,
      "grad_norm": 1.3075482845306396,
      "learning_rate": 8.482064899954796e-05,
      "loss": 1.7349,
      "step": 2836
    },
    {
      "epoch": 0.5110560684530511,
      "grad_norm": 1.214443325996399,
      "learning_rate": 8.481048470872641e-05,
      "loss": 1.5078,
      "step": 2837
    },
    {
      "epoch": 0.5112362080612475,
      "grad_norm": 1.3043891191482544,
      "learning_rate": 8.480031762542474e-05,
      "loss": 1.7372,
      "step": 2838
    },
    {
      "epoch": 0.5114163476694438,
      "grad_norm": 1.3345760107040405,
      "learning_rate": 8.479014775045855e-05,
      "loss": 1.8883,
      "step": 2839
    },
    {
      "epoch": 0.5115964872776402,
      "grad_norm": 1.4989759922027588,
      "learning_rate": 8.477997508464364e-05,
      "loss": 2.0237,
      "step": 2840
    },
    {
      "epoch": 0.5117766268858365,
      "grad_norm": 1.4189120531082153,
      "learning_rate": 8.476979962879608e-05,
      "loss": 1.9285,
      "step": 2841
    },
    {
      "epoch": 0.5119567664940329,
      "grad_norm": 1.379510521888733,
      "learning_rate": 8.475962138373213e-05,
      "loss": 1.5603,
      "step": 2842
    },
    {
      "epoch": 0.5121369061022292,
      "grad_norm": 1.4271522760391235,
      "learning_rate": 8.474944035026826e-05,
      "loss": 1.8155,
      "step": 2843
    },
    {
      "epoch": 0.5123170457104256,
      "grad_norm": 1.439161777496338,
      "learning_rate": 8.473925652922124e-05,
      "loss": 1.6182,
      "step": 2844
    },
    {
      "epoch": 0.512497185318622,
      "grad_norm": 1.286539077758789,
      "learning_rate": 8.472906992140797e-05,
      "loss": 1.6494,
      "step": 2845
    },
    {
      "epoch": 0.5126773249268183,
      "grad_norm": 1.517162799835205,
      "learning_rate": 8.471888052764563e-05,
      "loss": 1.478,
      "step": 2846
    },
    {
      "epoch": 0.5128574645350147,
      "grad_norm": 1.3631229400634766,
      "learning_rate": 8.470868834875159e-05,
      "loss": 1.3688,
      "step": 2847
    },
    {
      "epoch": 0.513037604143211,
      "grad_norm": 1.3169572353363037,
      "learning_rate": 8.46984933855435e-05,
      "loss": 1.4934,
      "step": 2848
    },
    {
      "epoch": 0.5132177437514074,
      "grad_norm": 1.3168138265609741,
      "learning_rate": 8.468829563883916e-05,
      "loss": 1.5146,
      "step": 2849
    },
    {
      "epoch": 0.5133978833596037,
      "grad_norm": 1.397937297821045,
      "learning_rate": 8.467809510945664e-05,
      "loss": 1.5788,
      "step": 2850
    },
    {
      "epoch": 0.5135780229678001,
      "grad_norm": 1.2191779613494873,
      "learning_rate": 8.466789179821422e-05,
      "loss": 2.0042,
      "step": 2851
    },
    {
      "epoch": 0.5137581625759964,
      "grad_norm": 1.2275811433792114,
      "learning_rate": 8.46576857059304e-05,
      "loss": 1.9563,
      "step": 2852
    },
    {
      "epoch": 0.5139383021841928,
      "grad_norm": 1.231316328048706,
      "learning_rate": 8.464747683342392e-05,
      "loss": 2.3349,
      "step": 2853
    },
    {
      "epoch": 0.5141184417923891,
      "grad_norm": 1.2736597061157227,
      "learning_rate": 8.463726518151372e-05,
      "loss": 2.2835,
      "step": 2854
    },
    {
      "epoch": 0.5142985814005855,
      "grad_norm": 1.3382141590118408,
      "learning_rate": 8.4627050751019e-05,
      "loss": 1.8476,
      "step": 2855
    },
    {
      "epoch": 0.5144787210087818,
      "grad_norm": 1.2106196880340576,
      "learning_rate": 8.461683354275911e-05,
      "loss": 2.0188,
      "step": 2856
    },
    {
      "epoch": 0.5146588606169782,
      "grad_norm": 1.3403714895248413,
      "learning_rate": 8.46066135575537e-05,
      "loss": 1.8018,
      "step": 2857
    },
    {
      "epoch": 0.5148390002251745,
      "grad_norm": 1.5464999675750732,
      "learning_rate": 8.459639079622261e-05,
      "loss": 2.2572,
      "step": 2858
    },
    {
      "epoch": 0.5150191398333709,
      "grad_norm": 1.7590980529785156,
      "learning_rate": 8.45861652595859e-05,
      "loss": 2.4232,
      "step": 2859
    },
    {
      "epoch": 0.5151992794415672,
      "grad_norm": 1.77105712890625,
      "learning_rate": 8.457593694846386e-05,
      "loss": 2.22,
      "step": 2860
    },
    {
      "epoch": 0.5153794190497636,
      "grad_norm": 1.3987563848495483,
      "learning_rate": 8.4565705863677e-05,
      "loss": 1.8002,
      "step": 2861
    },
    {
      "epoch": 0.5155595586579599,
      "grad_norm": 1.375073790550232,
      "learning_rate": 8.455547200604606e-05,
      "loss": 1.8871,
      "step": 2862
    },
    {
      "epoch": 0.5157396982661563,
      "grad_norm": 1.266856074333191,
      "learning_rate": 8.454523537639199e-05,
      "loss": 1.6751,
      "step": 2863
    },
    {
      "epoch": 0.5159198378743526,
      "grad_norm": 1.2251063585281372,
      "learning_rate": 8.453499597553595e-05,
      "loss": 1.7471,
      "step": 2864
    },
    {
      "epoch": 0.516099977482549,
      "grad_norm": 1.2130104303359985,
      "learning_rate": 8.452475380429935e-05,
      "loss": 1.519,
      "step": 2865
    },
    {
      "epoch": 0.5162801170907453,
      "grad_norm": 1.157072901725769,
      "learning_rate": 8.451450886350382e-05,
      "loss": 1.3935,
      "step": 2866
    },
    {
      "epoch": 0.5164602566989417,
      "grad_norm": 1.2801405191421509,
      "learning_rate": 8.450426115397121e-05,
      "loss": 1.786,
      "step": 2867
    },
    {
      "epoch": 0.516640396307138,
      "grad_norm": 1.3811697959899902,
      "learning_rate": 8.449401067652358e-05,
      "loss": 1.7417,
      "step": 2868
    },
    {
      "epoch": 0.5168205359153344,
      "grad_norm": 1.3319814205169678,
      "learning_rate": 8.44837574319832e-05,
      "loss": 1.8402,
      "step": 2869
    },
    {
      "epoch": 0.5170006755235307,
      "grad_norm": 1.2061325311660767,
      "learning_rate": 8.447350142117261e-05,
      "loss": 1.61,
      "step": 2870
    },
    {
      "epoch": 0.5171808151317271,
      "grad_norm": 1.3511368036270142,
      "learning_rate": 8.446324264491452e-05,
      "loss": 1.7815,
      "step": 2871
    },
    {
      "epoch": 0.5173609547399235,
      "grad_norm": 1.3119224309921265,
      "learning_rate": 8.445298110403188e-05,
      "loss": 1.6436,
      "step": 2872
    },
    {
      "epoch": 0.5175410943481198,
      "grad_norm": 1.19091796875,
      "learning_rate": 8.44427167993479e-05,
      "loss": 1.4899,
      "step": 2873
    },
    {
      "epoch": 0.5177212339563162,
      "grad_norm": 1.3423573970794678,
      "learning_rate": 8.443244973168594e-05,
      "loss": 1.6115,
      "step": 2874
    },
    {
      "epoch": 0.5179013735645125,
      "grad_norm": 1.301041603088379,
      "learning_rate": 8.442217990186962e-05,
      "loss": 1.4322,
      "step": 2875
    },
    {
      "epoch": 0.5180815131727089,
      "grad_norm": 1.451704740524292,
      "learning_rate": 8.441190731072281e-05,
      "loss": 1.7627,
      "step": 2876
    },
    {
      "epoch": 0.5182616527809052,
      "grad_norm": 1.2976250648498535,
      "learning_rate": 8.440163195906958e-05,
      "loss": 1.7752,
      "step": 2877
    },
    {
      "epoch": 0.5184417923891016,
      "grad_norm": 1.2227054834365845,
      "learning_rate": 8.439135384773415e-05,
      "loss": 1.5108,
      "step": 2878
    },
    {
      "epoch": 0.5186219319972979,
      "grad_norm": 1.3561028242111206,
      "learning_rate": 8.43810729775411e-05,
      "loss": 1.7148,
      "step": 2879
    },
    {
      "epoch": 0.5188020716054943,
      "grad_norm": 1.4228743314743042,
      "learning_rate": 8.43707893493151e-05,
      "loss": 1.9676,
      "step": 2880
    },
    {
      "epoch": 0.5189822112136906,
      "grad_norm": 1.234188199043274,
      "learning_rate": 8.436050296388114e-05,
      "loss": 1.4886,
      "step": 2881
    },
    {
      "epoch": 0.519162350821887,
      "grad_norm": 1.3529471158981323,
      "learning_rate": 8.435021382206433e-05,
      "loss": 1.6212,
      "step": 2882
    },
    {
      "epoch": 0.5193424904300833,
      "grad_norm": 1.2622463703155518,
      "learning_rate": 8.433992192469013e-05,
      "loss": 1.585,
      "step": 2883
    },
    {
      "epoch": 0.5195226300382797,
      "grad_norm": 1.3103488683700562,
      "learning_rate": 8.432962727258413e-05,
      "loss": 1.7917,
      "step": 2884
    },
    {
      "epoch": 0.519702769646476,
      "grad_norm": 1.2690023183822632,
      "learning_rate": 8.431932986657214e-05,
      "loss": 1.6042,
      "step": 2885
    },
    {
      "epoch": 0.5198829092546724,
      "grad_norm": 1.359317421913147,
      "learning_rate": 8.430902970748022e-05,
      "loss": 1.7418,
      "step": 2886
    },
    {
      "epoch": 0.5200630488628687,
      "grad_norm": 1.2745059728622437,
      "learning_rate": 8.429872679613464e-05,
      "loss": 1.7374,
      "step": 2887
    },
    {
      "epoch": 0.5202431884710651,
      "grad_norm": 1.293550968170166,
      "learning_rate": 8.428842113336191e-05,
      "loss": 1.6381,
      "step": 2888
    },
    {
      "epoch": 0.5204233280792614,
      "grad_norm": 1.4161497354507446,
      "learning_rate": 8.427811271998873e-05,
      "loss": 1.7489,
      "step": 2889
    },
    {
      "epoch": 0.5206034676874578,
      "grad_norm": 1.393950343132019,
      "learning_rate": 8.426780155684206e-05,
      "loss": 1.8912,
      "step": 2890
    },
    {
      "epoch": 0.5207836072956541,
      "grad_norm": 1.380697250366211,
      "learning_rate": 8.425748764474903e-05,
      "loss": 1.8741,
      "step": 2891
    },
    {
      "epoch": 0.5209637469038505,
      "grad_norm": 1.2994813919067383,
      "learning_rate": 8.424717098453703e-05,
      "loss": 1.4546,
      "step": 2892
    },
    {
      "epoch": 0.5211438865120468,
      "grad_norm": 1.4606106281280518,
      "learning_rate": 8.423685157703366e-05,
      "loss": 1.828,
      "step": 2893
    },
    {
      "epoch": 0.5213240261202432,
      "grad_norm": 1.217456340789795,
      "learning_rate": 8.422652942306672e-05,
      "loss": 1.695,
      "step": 2894
    },
    {
      "epoch": 0.5215041657284395,
      "grad_norm": 1.4949731826782227,
      "learning_rate": 8.421620452346428e-05,
      "loss": 1.9131,
      "step": 2895
    },
    {
      "epoch": 0.5216843053366359,
      "grad_norm": 1.394025444984436,
      "learning_rate": 8.420587687905457e-05,
      "loss": 1.7087,
      "step": 2896
    },
    {
      "epoch": 0.5218644449448322,
      "grad_norm": 1.5372185707092285,
      "learning_rate": 8.419554649066606e-05,
      "loss": 1.4535,
      "step": 2897
    },
    {
      "epoch": 0.5220445845530286,
      "grad_norm": 1.3466992378234863,
      "learning_rate": 8.418521335912749e-05,
      "loss": 1.4944,
      "step": 2898
    },
    {
      "epoch": 0.5222247241612249,
      "grad_norm": 1.2983089685440063,
      "learning_rate": 8.417487748526774e-05,
      "loss": 1.4455,
      "step": 2899
    },
    {
      "epoch": 0.5224048637694213,
      "grad_norm": 1.4654840230941772,
      "learning_rate": 8.416453886991598e-05,
      "loss": 1.6367,
      "step": 2900
    },
    {
      "epoch": 0.5225850033776177,
      "grad_norm": 1.2391570806503296,
      "learning_rate": 8.415419751390155e-05,
      "loss": 2.1621,
      "step": 2901
    },
    {
      "epoch": 0.522765142985814,
      "grad_norm": 1.38381826877594,
      "learning_rate": 8.414385341805402e-05,
      "loss": 2.0932,
      "step": 2902
    },
    {
      "epoch": 0.5229452825940104,
      "grad_norm": 1.2105937004089355,
      "learning_rate": 8.41335065832032e-05,
      "loss": 2.1338,
      "step": 2903
    },
    {
      "epoch": 0.5231254222022067,
      "grad_norm": 1.2993297576904297,
      "learning_rate": 8.412315701017912e-05,
      "loss": 1.859,
      "step": 2904
    },
    {
      "epoch": 0.5233055618104031,
      "grad_norm": 1.3384552001953125,
      "learning_rate": 8.4112804699812e-05,
      "loss": 2.1587,
      "step": 2905
    },
    {
      "epoch": 0.5234857014185994,
      "grad_norm": 1.508028507232666,
      "learning_rate": 8.410244965293231e-05,
      "loss": 2.4367,
      "step": 2906
    },
    {
      "epoch": 0.5236658410267958,
      "grad_norm": 1.4038044214248657,
      "learning_rate": 8.409209187037071e-05,
      "loss": 1.888,
      "step": 2907
    },
    {
      "epoch": 0.5238459806349921,
      "grad_norm": 1.602329969406128,
      "learning_rate": 8.408173135295809e-05,
      "loss": 2.1134,
      "step": 2908
    },
    {
      "epoch": 0.5240261202431885,
      "grad_norm": 1.6898146867752075,
      "learning_rate": 8.407136810152561e-05,
      "loss": 2.1636,
      "step": 2909
    },
    {
      "epoch": 0.5242062598513848,
      "grad_norm": 1.287115454673767,
      "learning_rate": 8.406100211690456e-05,
      "loss": 1.7853,
      "step": 2910
    },
    {
      "epoch": 0.5243863994595812,
      "grad_norm": 1.2718130350112915,
      "learning_rate": 8.405063339992651e-05,
      "loss": 1.6375,
      "step": 2911
    },
    {
      "epoch": 0.5245665390677775,
      "grad_norm": 1.2713121175765991,
      "learning_rate": 8.404026195142326e-05,
      "loss": 1.6516,
      "step": 2912
    },
    {
      "epoch": 0.5247466786759739,
      "grad_norm": 1.1569982767105103,
      "learning_rate": 8.402988777222674e-05,
      "loss": 1.4313,
      "step": 2913
    },
    {
      "epoch": 0.5249268182841702,
      "grad_norm": 1.328687310218811,
      "learning_rate": 8.401951086316921e-05,
      "loss": 1.7319,
      "step": 2914
    },
    {
      "epoch": 0.5251069578923666,
      "grad_norm": 1.3176500797271729,
      "learning_rate": 8.400913122508309e-05,
      "loss": 1.8114,
      "step": 2915
    },
    {
      "epoch": 0.5252870975005629,
      "grad_norm": 1.2282111644744873,
      "learning_rate": 8.399874885880104e-05,
      "loss": 1.6335,
      "step": 2916
    },
    {
      "epoch": 0.5254672371087593,
      "grad_norm": 1.2876404523849487,
      "learning_rate": 8.39883637651559e-05,
      "loss": 1.4814,
      "step": 2917
    },
    {
      "epoch": 0.5256473767169556,
      "grad_norm": 1.2772941589355469,
      "learning_rate": 8.39779759449808e-05,
      "loss": 1.5264,
      "step": 2918
    },
    {
      "epoch": 0.525827516325152,
      "grad_norm": 1.4183640480041504,
      "learning_rate": 8.3967585399109e-05,
      "loss": 1.6784,
      "step": 2919
    },
    {
      "epoch": 0.5260076559333483,
      "grad_norm": 1.2522879838943481,
      "learning_rate": 8.395719212837404e-05,
      "loss": 1.7072,
      "step": 2920
    },
    {
      "epoch": 0.5261877955415447,
      "grad_norm": 1.3981165885925293,
      "learning_rate": 8.394679613360968e-05,
      "loss": 1.764,
      "step": 2921
    },
    {
      "epoch": 0.526367935149741,
      "grad_norm": 1.313857078552246,
      "learning_rate": 8.393639741564984e-05,
      "loss": 1.7087,
      "step": 2922
    },
    {
      "epoch": 0.5265480747579374,
      "grad_norm": 1.3553773164749146,
      "learning_rate": 8.392599597532877e-05,
      "loss": 1.6911,
      "step": 2923
    },
    {
      "epoch": 0.5267282143661337,
      "grad_norm": 1.2365988492965698,
      "learning_rate": 8.391559181348082e-05,
      "loss": 1.5675,
      "step": 2924
    },
    {
      "epoch": 0.5269083539743301,
      "grad_norm": 1.2551406621932983,
      "learning_rate": 8.39051849309406e-05,
      "loss": 1.3187,
      "step": 2925
    },
    {
      "epoch": 0.5270884935825264,
      "grad_norm": 1.305245041847229,
      "learning_rate": 8.389477532854297e-05,
      "loss": 1.5808,
      "step": 2926
    },
    {
      "epoch": 0.5272686331907228,
      "grad_norm": 1.2748825550079346,
      "learning_rate": 8.388436300712295e-05,
      "loss": 1.46,
      "step": 2927
    },
    {
      "epoch": 0.5274487727989191,
      "grad_norm": 1.2135010957717896,
      "learning_rate": 8.387394796751585e-05,
      "loss": 1.4507,
      "step": 2928
    },
    {
      "epoch": 0.5276289124071155,
      "grad_norm": 1.290830373764038,
      "learning_rate": 8.386353021055715e-05,
      "loss": 1.5783,
      "step": 2929
    },
    {
      "epoch": 0.5278090520153119,
      "grad_norm": 1.295822024345398,
      "learning_rate": 8.385310973708254e-05,
      "loss": 1.522,
      "step": 2930
    },
    {
      "epoch": 0.5279891916235082,
      "grad_norm": 1.4165290594100952,
      "learning_rate": 8.384268654792797e-05,
      "loss": 1.7107,
      "step": 2931
    },
    {
      "epoch": 0.5281693312317046,
      "grad_norm": 1.2215039730072021,
      "learning_rate": 8.383226064392956e-05,
      "loss": 1.4735,
      "step": 2932
    },
    {
      "epoch": 0.5283494708399009,
      "grad_norm": 1.3790441751480103,
      "learning_rate": 8.382183202592367e-05,
      "loss": 1.8254,
      "step": 2933
    },
    {
      "epoch": 0.5285296104480973,
      "grad_norm": 1.2985281944274902,
      "learning_rate": 8.381140069474691e-05,
      "loss": 1.5506,
      "step": 2934
    },
    {
      "epoch": 0.5287097500562936,
      "grad_norm": 1.2924715280532837,
      "learning_rate": 8.380096665123604e-05,
      "loss": 1.5431,
      "step": 2935
    },
    {
      "epoch": 0.52888988966449,
      "grad_norm": 1.3524894714355469,
      "learning_rate": 8.379052989622809e-05,
      "loss": 1.7336,
      "step": 2936
    },
    {
      "epoch": 0.5290700292726863,
      "grad_norm": 1.2735430002212524,
      "learning_rate": 8.37800904305603e-05,
      "loss": 1.5671,
      "step": 2937
    },
    {
      "epoch": 0.5292501688808827,
      "grad_norm": 1.356387734413147,
      "learning_rate": 8.37696482550701e-05,
      "loss": 1.5586,
      "step": 2938
    },
    {
      "epoch": 0.529430308489079,
      "grad_norm": 1.3299981355667114,
      "learning_rate": 8.375920337059518e-05,
      "loss": 1.4931,
      "step": 2939
    },
    {
      "epoch": 0.5296104480972754,
      "grad_norm": 1.4164447784423828,
      "learning_rate": 8.374875577797339e-05,
      "loss": 1.7236,
      "step": 2940
    },
    {
      "epoch": 0.5297905877054717,
      "grad_norm": 1.3486334085464478,
      "learning_rate": 8.373830547804285e-05,
      "loss": 1.7085,
      "step": 2941
    },
    {
      "epoch": 0.5299707273136681,
      "grad_norm": 1.2500412464141846,
      "learning_rate": 8.37278524716419e-05,
      "loss": 1.3937,
      "step": 2942
    },
    {
      "epoch": 0.5301508669218644,
      "grad_norm": 1.350296139717102,
      "learning_rate": 8.371739675960904e-05,
      "loss": 1.759,
      "step": 2943
    },
    {
      "epoch": 0.5303310065300608,
      "grad_norm": 1.2812758684158325,
      "learning_rate": 8.370693834278302e-05,
      "loss": 1.693,
      "step": 2944
    },
    {
      "epoch": 0.5305111461382571,
      "grad_norm": 1.4055895805358887,
      "learning_rate": 8.369647722200285e-05,
      "loss": 1.5049,
      "step": 2945
    },
    {
      "epoch": 0.5306912857464535,
      "grad_norm": 1.4454180002212524,
      "learning_rate": 8.368601339810767e-05,
      "loss": 1.6285,
      "step": 2946
    },
    {
      "epoch": 0.5308714253546498,
      "grad_norm": 1.3234490156173706,
      "learning_rate": 8.367554687193691e-05,
      "loss": 1.5159,
      "step": 2947
    },
    {
      "epoch": 0.5310515649628462,
      "grad_norm": 1.4226984977722168,
      "learning_rate": 8.366507764433018e-05,
      "loss": 1.7194,
      "step": 2948
    },
    {
      "epoch": 0.5312317045710425,
      "grad_norm": 1.2859829664230347,
      "learning_rate": 8.365460571612733e-05,
      "loss": 1.6281,
      "step": 2949
    },
    {
      "epoch": 0.5314118441792389,
      "grad_norm": 1.2395504713058472,
      "learning_rate": 8.36441310881684e-05,
      "loss": 1.5818,
      "step": 2950
    },
    {
      "epoch": 0.5315919837874352,
      "grad_norm": 1.1794137954711914,
      "learning_rate": 8.363365376129365e-05,
      "loss": 2.0086,
      "step": 2951
    },
    {
      "epoch": 0.5317721233956316,
      "grad_norm": 1.2856978178024292,
      "learning_rate": 8.362317373634359e-05,
      "loss": 2.0545,
      "step": 2952
    },
    {
      "epoch": 0.5319522630038279,
      "grad_norm": 1.2867718935012817,
      "learning_rate": 8.36126910141589e-05,
      "loss": 2.0849,
      "step": 2953
    },
    {
      "epoch": 0.5321324026120243,
      "grad_norm": 1.1966649293899536,
      "learning_rate": 8.360220559558051e-05,
      "loss": 2.1832,
      "step": 2954
    },
    {
      "epoch": 0.5323125422202206,
      "grad_norm": 1.377406358718872,
      "learning_rate": 8.359171748144956e-05,
      "loss": 2.394,
      "step": 2955
    },
    {
      "epoch": 0.532492681828417,
      "grad_norm": 2.2069456577301025,
      "learning_rate": 8.35812266726074e-05,
      "loss": 2.1278,
      "step": 2956
    },
    {
      "epoch": 0.5326728214366134,
      "grad_norm": 1.2921961545944214,
      "learning_rate": 8.357073316989559e-05,
      "loss": 2.0947,
      "step": 2957
    },
    {
      "epoch": 0.5328529610448097,
      "grad_norm": 1.3869844675064087,
      "learning_rate": 8.356023697415591e-05,
      "loss": 1.8076,
      "step": 2958
    },
    {
      "epoch": 0.5330331006530061,
      "grad_norm": 1.7190313339233398,
      "learning_rate": 8.354973808623037e-05,
      "loss": 2.0916,
      "step": 2959
    },
    {
      "epoch": 0.5332132402612024,
      "grad_norm": 1.565635323524475,
      "learning_rate": 8.353923650696118e-05,
      "loss": 1.7942,
      "step": 2960
    },
    {
      "epoch": 0.5333933798693988,
      "grad_norm": 1.1866713762283325,
      "learning_rate": 8.352873223719078e-05,
      "loss": 1.5644,
      "step": 2961
    },
    {
      "epoch": 0.5335735194775951,
      "grad_norm": 1.2483091354370117,
      "learning_rate": 8.351822527776181e-05,
      "loss": 1.6697,
      "step": 2962
    },
    {
      "epoch": 0.5337536590857915,
      "grad_norm": 1.2404404878616333,
      "learning_rate": 8.350771562951712e-05,
      "loss": 1.654,
      "step": 2963
    },
    {
      "epoch": 0.5339337986939878,
      "grad_norm": 1.2751595973968506,
      "learning_rate": 8.349720329329981e-05,
      "loss": 1.5721,
      "step": 2964
    },
    {
      "epoch": 0.5341139383021842,
      "grad_norm": 1.1954346895217896,
      "learning_rate": 8.348668826995319e-05,
      "loss": 1.5188,
      "step": 2965
    },
    {
      "epoch": 0.5342940779103805,
      "grad_norm": 1.3471052646636963,
      "learning_rate": 8.347617056032072e-05,
      "loss": 1.6529,
      "step": 2966
    },
    {
      "epoch": 0.5344742175185769,
      "grad_norm": 1.2993495464324951,
      "learning_rate": 8.346565016524617e-05,
      "loss": 1.6368,
      "step": 2967
    },
    {
      "epoch": 0.5346543571267732,
      "grad_norm": 1.2527159452438354,
      "learning_rate": 8.345512708557345e-05,
      "loss": 1.6186,
      "step": 2968
    },
    {
      "epoch": 0.5348344967349696,
      "grad_norm": 1.088618516921997,
      "learning_rate": 8.344460132214675e-05,
      "loss": 1.2741,
      "step": 2969
    },
    {
      "epoch": 0.5350146363431659,
      "grad_norm": 1.2552549839019775,
      "learning_rate": 8.34340728758104e-05,
      "loss": 1.6054,
      "step": 2970
    },
    {
      "epoch": 0.5351947759513623,
      "grad_norm": 1.1767199039459229,
      "learning_rate": 8.342354174740902e-05,
      "loss": 1.3718,
      "step": 2971
    },
    {
      "epoch": 0.5353749155595586,
      "grad_norm": 1.22327721118927,
      "learning_rate": 8.34130079377874e-05,
      "loss": 1.5924,
      "step": 2972
    },
    {
      "epoch": 0.535555055167755,
      "grad_norm": 1.234145998954773,
      "learning_rate": 8.340247144779055e-05,
      "loss": 1.5154,
      "step": 2973
    },
    {
      "epoch": 0.5357351947759513,
      "grad_norm": 1.227959394454956,
      "learning_rate": 8.339193227826373e-05,
      "loss": 1.5425,
      "step": 2974
    },
    {
      "epoch": 0.5359153343841477,
      "grad_norm": 1.2257956266403198,
      "learning_rate": 8.338139043005235e-05,
      "loss": 1.5687,
      "step": 2975
    },
    {
      "epoch": 0.536095473992344,
      "grad_norm": 1.2666376829147339,
      "learning_rate": 8.33708459040021e-05,
      "loss": 1.7125,
      "step": 2976
    },
    {
      "epoch": 0.5362756136005404,
      "grad_norm": 1.416784644126892,
      "learning_rate": 8.336029870095885e-05,
      "loss": 1.6145,
      "step": 2977
    },
    {
      "epoch": 0.5364557532087367,
      "grad_norm": 1.275352954864502,
      "learning_rate": 8.334974882176868e-05,
      "loss": 1.6509,
      "step": 2978
    },
    {
      "epoch": 0.5366358928169331,
      "grad_norm": 1.3713093996047974,
      "learning_rate": 8.333919626727788e-05,
      "loss": 1.7314,
      "step": 2979
    },
    {
      "epoch": 0.5368160324251294,
      "grad_norm": 1.369529366493225,
      "learning_rate": 8.332864103833302e-05,
      "loss": 1.6948,
      "step": 2980
    },
    {
      "epoch": 0.5369961720333258,
      "grad_norm": 1.2536239624023438,
      "learning_rate": 8.331808313578082e-05,
      "loss": 1.6574,
      "step": 2981
    },
    {
      "epoch": 0.5371763116415221,
      "grad_norm": 1.2864121198654175,
      "learning_rate": 8.33075225604682e-05,
      "loss": 1.6983,
      "step": 2982
    },
    {
      "epoch": 0.5373564512497185,
      "grad_norm": 1.2520129680633545,
      "learning_rate": 8.329695931324235e-05,
      "loss": 1.6456,
      "step": 2983
    },
    {
      "epoch": 0.5375365908579148,
      "grad_norm": 1.2824212312698364,
      "learning_rate": 8.328639339495065e-05,
      "loss": 1.6142,
      "step": 2984
    },
    {
      "epoch": 0.5377167304661112,
      "grad_norm": 1.303770661354065,
      "learning_rate": 8.327582480644067e-05,
      "loss": 1.5105,
      "step": 2985
    },
    {
      "epoch": 0.5378968700743076,
      "grad_norm": 1.3146182298660278,
      "learning_rate": 8.326525354856025e-05,
      "loss": 1.6884,
      "step": 2986
    },
    {
      "epoch": 0.5380770096825039,
      "grad_norm": 1.4487546682357788,
      "learning_rate": 8.325467962215738e-05,
      "loss": 1.5976,
      "step": 2987
    },
    {
      "epoch": 0.5382571492907003,
      "grad_norm": 1.4768322706222534,
      "learning_rate": 8.32441030280803e-05,
      "loss": 1.5234,
      "step": 2988
    },
    {
      "epoch": 0.5384372888988966,
      "grad_norm": 1.3447977304458618,
      "learning_rate": 8.323352376717748e-05,
      "loss": 1.7109,
      "step": 2989
    },
    {
      "epoch": 0.538617428507093,
      "grad_norm": 1.355639100074768,
      "learning_rate": 8.322294184029758e-05,
      "loss": 1.5224,
      "step": 2990
    },
    {
      "epoch": 0.5387975681152893,
      "grad_norm": 1.4150404930114746,
      "learning_rate": 8.321235724828945e-05,
      "loss": 1.7687,
      "step": 2991
    },
    {
      "epoch": 0.5389777077234857,
      "grad_norm": 1.295420527458191,
      "learning_rate": 8.32017699920022e-05,
      "loss": 1.5991,
      "step": 2992
    },
    {
      "epoch": 0.539157847331682,
      "grad_norm": 1.2863636016845703,
      "learning_rate": 8.319118007228515e-05,
      "loss": 1.5894,
      "step": 2993
    },
    {
      "epoch": 0.5393379869398784,
      "grad_norm": 1.288286566734314,
      "learning_rate": 8.318058748998778e-05,
      "loss": 1.5178,
      "step": 2994
    },
    {
      "epoch": 0.5395181265480747,
      "grad_norm": 1.325046181678772,
      "learning_rate": 8.316999224595986e-05,
      "loss": 1.7127,
      "step": 2995
    },
    {
      "epoch": 0.5396982661562711,
      "grad_norm": 1.339909553527832,
      "learning_rate": 8.315939434105132e-05,
      "loss": 1.6117,
      "step": 2996
    },
    {
      "epoch": 0.5398784057644674,
      "grad_norm": 1.467614769935608,
      "learning_rate": 8.314879377611232e-05,
      "loss": 1.8017,
      "step": 2997
    },
    {
      "epoch": 0.5400585453726638,
      "grad_norm": 1.4138083457946777,
      "learning_rate": 8.313819055199321e-05,
      "loss": 1.5596,
      "step": 2998
    },
    {
      "epoch": 0.5402386849808601,
      "grad_norm": 1.4147588014602661,
      "learning_rate": 8.31275846695446e-05,
      "loss": 1.5653,
      "step": 2999
    },
    {
      "epoch": 0.5404188245890565,
      "grad_norm": 1.4119365215301514,
      "learning_rate": 8.311697612961729e-05,
      "loss": 1.5692,
      "step": 3000
    },
    {
      "epoch": 0.5405989641972528,
      "grad_norm": 1.2690507173538208,
      "learning_rate": 8.310636493306227e-05,
      "loss": 1.7542,
      "step": 3001
    },
    {
      "epoch": 0.5407791038054492,
      "grad_norm": 1.212904930114746,
      "learning_rate": 8.30957510807308e-05,
      "loss": 1.948,
      "step": 3002
    },
    {
      "epoch": 0.5409592434136455,
      "grad_norm": 1.2991600036621094,
      "learning_rate": 8.30851345734743e-05,
      "loss": 1.9466,
      "step": 3003
    },
    {
      "epoch": 0.541139383021842,
      "grad_norm": 1.3232090473175049,
      "learning_rate": 8.307451541214441e-05,
      "loss": 1.7802,
      "step": 3004
    },
    {
      "epoch": 0.5413195226300382,
      "grad_norm": 1.348582148551941,
      "learning_rate": 8.306389359759302e-05,
      "loss": 1.8812,
      "step": 3005
    },
    {
      "epoch": 0.5414996622382346,
      "grad_norm": 1.344104528427124,
      "learning_rate": 8.305326913067215e-05,
      "loss": 1.9632,
      "step": 3006
    },
    {
      "epoch": 0.5416798018464309,
      "grad_norm": 1.4267274141311646,
      "learning_rate": 8.304264201223417e-05,
      "loss": 2.3278,
      "step": 3007
    },
    {
      "epoch": 0.5418599414546273,
      "grad_norm": 1.5221933126449585,
      "learning_rate": 8.303201224313153e-05,
      "loss": 2.5895,
      "step": 3008
    },
    {
      "epoch": 0.5420400810628236,
      "grad_norm": 1.5019080638885498,
      "learning_rate": 8.302137982421696e-05,
      "loss": 1.8229,
      "step": 3009
    },
    {
      "epoch": 0.54222022067102,
      "grad_norm": 1.8297483921051025,
      "learning_rate": 8.301074475634337e-05,
      "loss": 2.2933,
      "step": 3010
    },
    {
      "epoch": 0.5424003602792163,
      "grad_norm": 1.3127143383026123,
      "learning_rate": 8.300010704036393e-05,
      "loss": 1.71,
      "step": 3011
    },
    {
      "epoch": 0.5425804998874127,
      "grad_norm": 1.1981914043426514,
      "learning_rate": 8.298946667713197e-05,
      "loss": 1.631,
      "step": 3012
    },
    {
      "epoch": 0.542760639495609,
      "grad_norm": 1.332061529159546,
      "learning_rate": 8.297882366750107e-05,
      "loss": 1.9382,
      "step": 3013
    },
    {
      "epoch": 0.5429407791038054,
      "grad_norm": 1.3686964511871338,
      "learning_rate": 8.2968178012325e-05,
      "loss": 2.0074,
      "step": 3014
    },
    {
      "epoch": 0.5431209187120019,
      "grad_norm": 1.3442156314849854,
      "learning_rate": 8.295752971245773e-05,
      "loss": 1.57,
      "step": 3015
    },
    {
      "epoch": 0.5433010583201981,
      "grad_norm": 1.2571334838867188,
      "learning_rate": 8.294687876875349e-05,
      "loss": 1.7784,
      "step": 3016
    },
    {
      "epoch": 0.5434811979283946,
      "grad_norm": 1.2542588710784912,
      "learning_rate": 8.293622518206668e-05,
      "loss": 1.625,
      "step": 3017
    },
    {
      "epoch": 0.5436613375365908,
      "grad_norm": 1.3364351987838745,
      "learning_rate": 8.292556895325194e-05,
      "loss": 1.6184,
      "step": 3018
    },
    {
      "epoch": 0.5438414771447873,
      "grad_norm": 1.201956033706665,
      "learning_rate": 8.291491008316409e-05,
      "loss": 1.5305,
      "step": 3019
    },
    {
      "epoch": 0.5440216167529836,
      "grad_norm": 1.166782259941101,
      "learning_rate": 8.290424857265818e-05,
      "loss": 1.5747,
      "step": 3020
    },
    {
      "epoch": 0.54420175636118,
      "grad_norm": 1.1946702003479004,
      "learning_rate": 8.289358442258947e-05,
      "loss": 1.3544,
      "step": 3021
    },
    {
      "epoch": 0.5443818959693763,
      "grad_norm": 1.2366557121276855,
      "learning_rate": 8.288291763381346e-05,
      "loss": 1.6255,
      "step": 3022
    },
    {
      "epoch": 0.5445620355775727,
      "grad_norm": 1.291358232498169,
      "learning_rate": 8.28722482071858e-05,
      "loss": 1.5542,
      "step": 3023
    },
    {
      "epoch": 0.544742175185769,
      "grad_norm": 1.306465744972229,
      "learning_rate": 8.28615761435624e-05,
      "loss": 1.6861,
      "step": 3024
    },
    {
      "epoch": 0.5449223147939654,
      "grad_norm": 1.3511592149734497,
      "learning_rate": 8.285090144379937e-05,
      "loss": 1.6909,
      "step": 3025
    },
    {
      "epoch": 0.5451024544021617,
      "grad_norm": 1.2512991428375244,
      "learning_rate": 8.284022410875305e-05,
      "loss": 1.6008,
      "step": 3026
    },
    {
      "epoch": 0.5452825940103581,
      "grad_norm": 1.2254688739776611,
      "learning_rate": 8.282954413927992e-05,
      "loss": 1.5745,
      "step": 3027
    },
    {
      "epoch": 0.5454627336185544,
      "grad_norm": 1.3068994283676147,
      "learning_rate": 8.281886153623675e-05,
      "loss": 1.704,
      "step": 3028
    },
    {
      "epoch": 0.5456428732267508,
      "grad_norm": 1.375957727432251,
      "learning_rate": 8.28081763004805e-05,
      "loss": 1.8602,
      "step": 3029
    },
    {
      "epoch": 0.545823012834947,
      "grad_norm": 1.5064692497253418,
      "learning_rate": 8.279748843286833e-05,
      "loss": 1.7836,
      "step": 3030
    },
    {
      "epoch": 0.5460031524431435,
      "grad_norm": 1.173319935798645,
      "learning_rate": 8.278679793425761e-05,
      "loss": 1.4822,
      "step": 3031
    },
    {
      "epoch": 0.5461832920513398,
      "grad_norm": 1.4948939085006714,
      "learning_rate": 8.277610480550593e-05,
      "loss": 1.7825,
      "step": 3032
    },
    {
      "epoch": 0.5463634316595362,
      "grad_norm": 1.5304152965545654,
      "learning_rate": 8.27654090474711e-05,
      "loss": 2.0278,
      "step": 3033
    },
    {
      "epoch": 0.5465435712677325,
      "grad_norm": 1.337569236755371,
      "learning_rate": 8.275471066101109e-05,
      "loss": 1.7703,
      "step": 3034
    },
    {
      "epoch": 0.5467237108759289,
      "grad_norm": 1.3904756307601929,
      "learning_rate": 8.274400964698417e-05,
      "loss": 1.5607,
      "step": 3035
    },
    {
      "epoch": 0.5469038504841252,
      "grad_norm": 1.4075552225112915,
      "learning_rate": 8.273330600624872e-05,
      "loss": 1.8431,
      "step": 3036
    },
    {
      "epoch": 0.5470839900923216,
      "grad_norm": 1.4117883443832397,
      "learning_rate": 8.272259973966343e-05,
      "loss": 1.7731,
      "step": 3037
    },
    {
      "epoch": 0.5472641297005179,
      "grad_norm": 1.3136579990386963,
      "learning_rate": 8.27118908480871e-05,
      "loss": 1.4595,
      "step": 3038
    },
    {
      "epoch": 0.5474442693087143,
      "grad_norm": 1.3299778699874878,
      "learning_rate": 8.270117933237884e-05,
      "loss": 1.7849,
      "step": 3039
    },
    {
      "epoch": 0.5476244089169106,
      "grad_norm": 1.3349429368972778,
      "learning_rate": 8.269046519339792e-05,
      "loss": 1.8017,
      "step": 3040
    },
    {
      "epoch": 0.547804548525107,
      "grad_norm": 1.4871941804885864,
      "learning_rate": 8.267974843200376e-05,
      "loss": 1.8816,
      "step": 3041
    },
    {
      "epoch": 0.5479846881333033,
      "grad_norm": 1.3881686925888062,
      "learning_rate": 8.266902904905613e-05,
      "loss": 1.6137,
      "step": 3042
    },
    {
      "epoch": 0.5481648277414997,
      "grad_norm": 1.5226101875305176,
      "learning_rate": 8.26583070454149e-05,
      "loss": 1.7101,
      "step": 3043
    },
    {
      "epoch": 0.5483449673496961,
      "grad_norm": 1.45222008228302,
      "learning_rate": 8.264758242194017e-05,
      "loss": 1.9807,
      "step": 3044
    },
    {
      "epoch": 0.5485251069578924,
      "grad_norm": 1.4130432605743408,
      "learning_rate": 8.263685517949228e-05,
      "loss": 1.6995,
      "step": 3045
    },
    {
      "epoch": 0.5487052465660888,
      "grad_norm": 1.3273085355758667,
      "learning_rate": 8.262612531893179e-05,
      "loss": 1.5851,
      "step": 3046
    },
    {
      "epoch": 0.5488853861742851,
      "grad_norm": 1.3767088651657104,
      "learning_rate": 8.261539284111941e-05,
      "loss": 1.4744,
      "step": 3047
    },
    {
      "epoch": 0.5490655257824815,
      "grad_norm": 1.264435052871704,
      "learning_rate": 8.26046577469161e-05,
      "loss": 1.5227,
      "step": 3048
    },
    {
      "epoch": 0.5492456653906778,
      "grad_norm": 1.2981853485107422,
      "learning_rate": 8.259392003718301e-05,
      "loss": 1.4457,
      "step": 3049
    },
    {
      "epoch": 0.5494258049988742,
      "grad_norm": 1.4427740573883057,
      "learning_rate": 8.258317971278155e-05,
      "loss": 1.6294,
      "step": 3050
    },
    {
      "epoch": 0.5496059446070705,
      "grad_norm": 1.6608119010925293,
      "learning_rate": 8.257243677457328e-05,
      "loss": 2.1272,
      "step": 3051
    },
    {
      "epoch": 0.5497860842152669,
      "grad_norm": 1.3499118089675903,
      "learning_rate": 8.256169122341998e-05,
      "loss": 2.2511,
      "step": 3052
    },
    {
      "epoch": 0.5499662238234632,
      "grad_norm": 1.202763319015503,
      "learning_rate": 8.25509430601837e-05,
      "loss": 1.9001,
      "step": 3053
    },
    {
      "epoch": 0.5501463634316596,
      "grad_norm": 1.3304518461227417,
      "learning_rate": 8.25401922857266e-05,
      "loss": 2.0743,
      "step": 3054
    },
    {
      "epoch": 0.5503265030398559,
      "grad_norm": 1.2752487659454346,
      "learning_rate": 8.252943890091113e-05,
      "loss": 1.9138,
      "step": 3055
    },
    {
      "epoch": 0.5505066426480523,
      "grad_norm": 1.3434410095214844,
      "learning_rate": 8.251868290659993e-05,
      "loss": 1.9507,
      "step": 3056
    },
    {
      "epoch": 0.5506867822562486,
      "grad_norm": 1.4837720394134521,
      "learning_rate": 8.250792430365582e-05,
      "loss": 2.0902,
      "step": 3057
    },
    {
      "epoch": 0.550866921864445,
      "grad_norm": 1.5025969743728638,
      "learning_rate": 8.249716309294184e-05,
      "loss": 1.8903,
      "step": 3058
    },
    {
      "epoch": 0.5510470614726413,
      "grad_norm": 1.7507224082946777,
      "learning_rate": 8.248639927532127e-05,
      "loss": 2.0086,
      "step": 3059
    },
    {
      "epoch": 0.5512272010808377,
      "grad_norm": 1.455481767654419,
      "learning_rate": 8.247563285165759e-05,
      "loss": 1.6329,
      "step": 3060
    },
    {
      "epoch": 0.551407340689034,
      "grad_norm": 1.2602249383926392,
      "learning_rate": 8.246486382281444e-05,
      "loss": 1.6726,
      "step": 3061
    },
    {
      "epoch": 0.5515874802972304,
      "grad_norm": 1.2342243194580078,
      "learning_rate": 8.245409218965575e-05,
      "loss": 1.6463,
      "step": 3062
    },
    {
      "epoch": 0.5517676199054267,
      "grad_norm": 1.3046754598617554,
      "learning_rate": 8.244331795304559e-05,
      "loss": 1.6362,
      "step": 3063
    },
    {
      "epoch": 0.5519477595136231,
      "grad_norm": 1.353340744972229,
      "learning_rate": 8.243254111384825e-05,
      "loss": 1.6933,
      "step": 3064
    },
    {
      "epoch": 0.5521278991218194,
      "grad_norm": 1.2738736867904663,
      "learning_rate": 8.242176167292826e-05,
      "loss": 1.516,
      "step": 3065
    },
    {
      "epoch": 0.5523080387300158,
      "grad_norm": 1.4168224334716797,
      "learning_rate": 8.241097963115035e-05,
      "loss": 1.5416,
      "step": 3066
    },
    {
      "epoch": 0.5524881783382121,
      "grad_norm": 1.4942160844802856,
      "learning_rate": 8.240019498937944e-05,
      "loss": 1.9458,
      "step": 3067
    },
    {
      "epoch": 0.5526683179464085,
      "grad_norm": 1.2541728019714355,
      "learning_rate": 8.238940774848066e-05,
      "loss": 1.4761,
      "step": 3068
    },
    {
      "epoch": 0.5528484575546048,
      "grad_norm": 1.240328073501587,
      "learning_rate": 8.237861790931938e-05,
      "loss": 1.7644,
      "step": 3069
    },
    {
      "epoch": 0.5530285971628012,
      "grad_norm": 1.3574550151824951,
      "learning_rate": 8.236782547276114e-05,
      "loss": 1.5677,
      "step": 3070
    },
    {
      "epoch": 0.5532087367709976,
      "grad_norm": 1.2949200868606567,
      "learning_rate": 8.23570304396717e-05,
      "loss": 2.0283,
      "step": 3071
    },
    {
      "epoch": 0.5533888763791939,
      "grad_norm": 1.3146377801895142,
      "learning_rate": 8.234623281091704e-05,
      "loss": 1.6428,
      "step": 3072
    },
    {
      "epoch": 0.5535690159873903,
      "grad_norm": 1.2935655117034912,
      "learning_rate": 8.233543258736334e-05,
      "loss": 1.7052,
      "step": 3073
    },
    {
      "epoch": 0.5537491555955866,
      "grad_norm": 1.128687858581543,
      "learning_rate": 8.232462976987699e-05,
      "loss": 1.4706,
      "step": 3074
    },
    {
      "epoch": 0.553929295203783,
      "grad_norm": 1.203316330909729,
      "learning_rate": 8.231382435932459e-05,
      "loss": 1.4951,
      "step": 3075
    },
    {
      "epoch": 0.5541094348119793,
      "grad_norm": 1.3685472011566162,
      "learning_rate": 8.230301635657293e-05,
      "loss": 1.8669,
      "step": 3076
    },
    {
      "epoch": 0.5542895744201757,
      "grad_norm": 1.301174283027649,
      "learning_rate": 8.229220576248904e-05,
      "loss": 1.6304,
      "step": 3077
    },
    {
      "epoch": 0.554469714028372,
      "grad_norm": 1.19906485080719,
      "learning_rate": 8.228139257794012e-05,
      "loss": 1.4816,
      "step": 3078
    },
    {
      "epoch": 0.5546498536365684,
      "grad_norm": 1.3333052396774292,
      "learning_rate": 8.227057680379362e-05,
      "loss": 1.5358,
      "step": 3079
    },
    {
      "epoch": 0.5548299932447647,
      "grad_norm": 1.2843226194381714,
      "learning_rate": 8.225975844091717e-05,
      "loss": 1.5784,
      "step": 3080
    },
    {
      "epoch": 0.5550101328529611,
      "grad_norm": 1.4443106651306152,
      "learning_rate": 8.224893749017862e-05,
      "loss": 1.6562,
      "step": 3081
    },
    {
      "epoch": 0.5551902724611574,
      "grad_norm": 1.2481725215911865,
      "learning_rate": 8.223811395244598e-05,
      "loss": 1.479,
      "step": 3082
    },
    {
      "epoch": 0.5553704120693538,
      "grad_norm": 1.3599581718444824,
      "learning_rate": 8.222728782858757e-05,
      "loss": 1.7141,
      "step": 3083
    },
    {
      "epoch": 0.5555505516775501,
      "grad_norm": 1.281242847442627,
      "learning_rate": 8.221645911947182e-05,
      "loss": 1.4411,
      "step": 3084
    },
    {
      "epoch": 0.5557306912857465,
      "grad_norm": 1.414431095123291,
      "learning_rate": 8.220562782596742e-05,
      "loss": 1.7247,
      "step": 3085
    },
    {
      "epoch": 0.5559108308939428,
      "grad_norm": 1.418641209602356,
      "learning_rate": 8.219479394894323e-05,
      "loss": 1.8093,
      "step": 3086
    },
    {
      "epoch": 0.5560909705021392,
      "grad_norm": 1.356667160987854,
      "learning_rate": 8.218395748926834e-05,
      "loss": 1.6301,
      "step": 3087
    },
    {
      "epoch": 0.5562711101103355,
      "grad_norm": 1.4201664924621582,
      "learning_rate": 8.217311844781208e-05,
      "loss": 1.7891,
      "step": 3088
    },
    {
      "epoch": 0.5564512497185319,
      "grad_norm": 1.3570770025253296,
      "learning_rate": 8.21622768254439e-05,
      "loss": 1.6598,
      "step": 3089
    },
    {
      "epoch": 0.5566313893267282,
      "grad_norm": 1.3093458414077759,
      "learning_rate": 8.215143262303358e-05,
      "loss": 1.5541,
      "step": 3090
    },
    {
      "epoch": 0.5568115289349246,
      "grad_norm": 1.5533108711242676,
      "learning_rate": 8.214058584145096e-05,
      "loss": 2.0327,
      "step": 3091
    },
    {
      "epoch": 0.5569916685431209,
      "grad_norm": 1.458650827407837,
      "learning_rate": 8.212973648156622e-05,
      "loss": 1.9261,
      "step": 3092
    },
    {
      "epoch": 0.5571718081513173,
      "grad_norm": 1.324681043624878,
      "learning_rate": 8.211888454424966e-05,
      "loss": 1.762,
      "step": 3093
    },
    {
      "epoch": 0.5573519477595136,
      "grad_norm": 1.4270514249801636,
      "learning_rate": 8.210803003037182e-05,
      "loss": 1.6771,
      "step": 3094
    },
    {
      "epoch": 0.55753208736771,
      "grad_norm": 1.2860157489776611,
      "learning_rate": 8.209717294080345e-05,
      "loss": 1.503,
      "step": 3095
    },
    {
      "epoch": 0.5577122269759063,
      "grad_norm": 1.3806312084197998,
      "learning_rate": 8.208631327641551e-05,
      "loss": 1.5762,
      "step": 3096
    },
    {
      "epoch": 0.5578923665841027,
      "grad_norm": 1.3448361158370972,
      "learning_rate": 8.207545103807916e-05,
      "loss": 1.7357,
      "step": 3097
    },
    {
      "epoch": 0.558072506192299,
      "grad_norm": 1.4819183349609375,
      "learning_rate": 8.206458622666574e-05,
      "loss": 1.8043,
      "step": 3098
    },
    {
      "epoch": 0.5582526458004954,
      "grad_norm": 1.247347116470337,
      "learning_rate": 8.205371884304682e-05,
      "loss": 1.4419,
      "step": 3099
    },
    {
      "epoch": 0.5584327854086918,
      "grad_norm": 1.2945557832717896,
      "learning_rate": 8.20428488880942e-05,
      "loss": 1.3948,
      "step": 3100
    },
    {
      "epoch": 0.5586129250168881,
      "grad_norm": 1.2828346490859985,
      "learning_rate": 8.203197636267985e-05,
      "loss": 2.2183,
      "step": 3101
    },
    {
      "epoch": 0.5587930646250845,
      "grad_norm": 1.1839442253112793,
      "learning_rate": 8.202110126767595e-05,
      "loss": 2.0172,
      "step": 3102
    },
    {
      "epoch": 0.5589732042332808,
      "grad_norm": 1.301910638809204,
      "learning_rate": 8.20102236039549e-05,
      "loss": 2.1956,
      "step": 3103
    },
    {
      "epoch": 0.5591533438414772,
      "grad_norm": 1.1977522373199463,
      "learning_rate": 8.199934337238933e-05,
      "loss": 1.8359,
      "step": 3104
    },
    {
      "epoch": 0.5593334834496735,
      "grad_norm": 1.2712671756744385,
      "learning_rate": 8.1988460573852e-05,
      "loss": 2.0776,
      "step": 3105
    },
    {
      "epoch": 0.5595136230578699,
      "grad_norm": 1.2882437705993652,
      "learning_rate": 8.197757520921594e-05,
      "loss": 2.1507,
      "step": 3106
    },
    {
      "epoch": 0.5596937626660662,
      "grad_norm": 1.2889759540557861,
      "learning_rate": 8.196668727935438e-05,
      "loss": 1.9329,
      "step": 3107
    },
    {
      "epoch": 0.5598739022742626,
      "grad_norm": 1.398176908493042,
      "learning_rate": 8.195579678514074e-05,
      "loss": 1.919,
      "step": 3108
    },
    {
      "epoch": 0.5600540418824589,
      "grad_norm": 1.539720058441162,
      "learning_rate": 8.194490372744863e-05,
      "loss": 2.1278,
      "step": 3109
    },
    {
      "epoch": 0.5602341814906553,
      "grad_norm": 1.7101788520812988,
      "learning_rate": 8.193400810715192e-05,
      "loss": 2.1057,
      "step": 3110
    },
    {
      "epoch": 0.5604143210988516,
      "grad_norm": 1.318769097328186,
      "learning_rate": 8.192310992512463e-05,
      "loss": 1.738,
      "step": 3111
    },
    {
      "epoch": 0.560594460707048,
      "grad_norm": 1.2736765146255493,
      "learning_rate": 8.191220918224101e-05,
      "loss": 1.8106,
      "step": 3112
    },
    {
      "epoch": 0.5607746003152443,
      "grad_norm": 1.1861085891723633,
      "learning_rate": 8.190130587937551e-05,
      "loss": 1.3742,
      "step": 3113
    },
    {
      "epoch": 0.5609547399234407,
      "grad_norm": 1.3282548189163208,
      "learning_rate": 8.18904000174028e-05,
      "loss": 1.9661,
      "step": 3114
    },
    {
      "epoch": 0.561134879531637,
      "grad_norm": 1.2139323949813843,
      "learning_rate": 8.187949159719772e-05,
      "loss": 1.6599,
      "step": 3115
    },
    {
      "epoch": 0.5613150191398334,
      "grad_norm": 1.1640039682388306,
      "learning_rate": 8.186858061963537e-05,
      "loss": 1.4607,
      "step": 3116
    },
    {
      "epoch": 0.5614951587480297,
      "grad_norm": 1.270864725112915,
      "learning_rate": 8.185766708559099e-05,
      "loss": 1.4509,
      "step": 3117
    },
    {
      "epoch": 0.5616752983562261,
      "grad_norm": 1.2682238817214966,
      "learning_rate": 8.184675099594008e-05,
      "loss": 1.4128,
      "step": 3118
    },
    {
      "epoch": 0.5618554379644224,
      "grad_norm": 1.3159164190292358,
      "learning_rate": 8.183583235155832e-05,
      "loss": 1.6746,
      "step": 3119
    },
    {
      "epoch": 0.5620355775726188,
      "grad_norm": 1.3651436567306519,
      "learning_rate": 8.18249111533216e-05,
      "loss": 1.5824,
      "step": 3120
    },
    {
      "epoch": 0.5622157171808151,
      "grad_norm": 1.2646636962890625,
      "learning_rate": 8.1813987402106e-05,
      "loss": 1.7706,
      "step": 3121
    },
    {
      "epoch": 0.5623958567890115,
      "grad_norm": 1.222787857055664,
      "learning_rate": 8.180306109878782e-05,
      "loss": 1.4056,
      "step": 3122
    },
    {
      "epoch": 0.5625759963972078,
      "grad_norm": 1.2767640352249146,
      "learning_rate": 8.179213224424358e-05,
      "loss": 1.6311,
      "step": 3123
    },
    {
      "epoch": 0.5627561360054042,
      "grad_norm": 1.295236349105835,
      "learning_rate": 8.178120083934997e-05,
      "loss": 1.6952,
      "step": 3124
    },
    {
      "epoch": 0.5629362756136005,
      "grad_norm": 1.268667221069336,
      "learning_rate": 8.17702668849839e-05,
      "loss": 1.6174,
      "step": 3125
    },
    {
      "epoch": 0.5631164152217969,
      "grad_norm": 1.116782784461975,
      "learning_rate": 8.175933038202252e-05,
      "loss": 1.2679,
      "step": 3126
    },
    {
      "epoch": 0.5632965548299932,
      "grad_norm": 1.4525080919265747,
      "learning_rate": 8.174839133134311e-05,
      "loss": 2.047,
      "step": 3127
    },
    {
      "epoch": 0.5634766944381896,
      "grad_norm": 1.267739176750183,
      "learning_rate": 8.173744973382323e-05,
      "loss": 1.7404,
      "step": 3128
    },
    {
      "epoch": 0.563656834046386,
      "grad_norm": 1.4082084894180298,
      "learning_rate": 8.172650559034056e-05,
      "loss": 1.7778,
      "step": 3129
    },
    {
      "epoch": 0.5638369736545823,
      "grad_norm": 1.1513056755065918,
      "learning_rate": 8.171555890177309e-05,
      "loss": 1.4083,
      "step": 3130
    },
    {
      "epoch": 0.5640171132627787,
      "grad_norm": 1.3695405721664429,
      "learning_rate": 8.170460966899891e-05,
      "loss": 1.8081,
      "step": 3131
    },
    {
      "epoch": 0.564197252870975,
      "grad_norm": 1.3276681900024414,
      "learning_rate": 8.16936578928964e-05,
      "loss": 1.7415,
      "step": 3132
    },
    {
      "epoch": 0.5643773924791714,
      "grad_norm": 1.310463547706604,
      "learning_rate": 8.16827035743441e-05,
      "loss": 1.7445,
      "step": 3133
    },
    {
      "epoch": 0.5645575320873677,
      "grad_norm": 1.3860929012298584,
      "learning_rate": 8.167174671422074e-05,
      "loss": 1.7978,
      "step": 3134
    },
    {
      "epoch": 0.5647376716955641,
      "grad_norm": 1.1992087364196777,
      "learning_rate": 8.16607873134053e-05,
      "loss": 1.5431,
      "step": 3135
    },
    {
      "epoch": 0.5649178113037604,
      "grad_norm": 1.1616803407669067,
      "learning_rate": 8.16498253727769e-05,
      "loss": 1.4783,
      "step": 3136
    },
    {
      "epoch": 0.5650979509119568,
      "grad_norm": 1.2000281810760498,
      "learning_rate": 8.163886089321493e-05,
      "loss": 1.4426,
      "step": 3137
    },
    {
      "epoch": 0.5652780905201531,
      "grad_norm": 1.3088774681091309,
      "learning_rate": 8.162789387559894e-05,
      "loss": 1.5736,
      "step": 3138
    },
    {
      "epoch": 0.5654582301283495,
      "grad_norm": 1.4206700325012207,
      "learning_rate": 8.161692432080873e-05,
      "loss": 1.5989,
      "step": 3139
    },
    {
      "epoch": 0.5656383697365458,
      "grad_norm": 1.4686074256896973,
      "learning_rate": 8.160595222972426e-05,
      "loss": 1.7029,
      "step": 3140
    },
    {
      "epoch": 0.5658185093447422,
      "grad_norm": 1.3710918426513672,
      "learning_rate": 8.159497760322567e-05,
      "loss": 1.4908,
      "step": 3141
    },
    {
      "epoch": 0.5659986489529385,
      "grad_norm": 1.4399117231369019,
      "learning_rate": 8.158400044219338e-05,
      "loss": 1.9285,
      "step": 3142
    },
    {
      "epoch": 0.5661787885611349,
      "grad_norm": 1.6106703281402588,
      "learning_rate": 8.157302074750797e-05,
      "loss": 1.9503,
      "step": 3143
    },
    {
      "epoch": 0.5663589281693312,
      "grad_norm": 1.3452823162078857,
      "learning_rate": 8.15620385200502e-05,
      "loss": 1.5154,
      "step": 3144
    },
    {
      "epoch": 0.5665390677775276,
      "grad_norm": 1.34783136844635,
      "learning_rate": 8.155105376070107e-05,
      "loss": 1.598,
      "step": 3145
    },
    {
      "epoch": 0.5667192073857239,
      "grad_norm": 1.367321252822876,
      "learning_rate": 8.154006647034179e-05,
      "loss": 1.5694,
      "step": 3146
    },
    {
      "epoch": 0.5668993469939203,
      "grad_norm": 1.5204110145568848,
      "learning_rate": 8.152907664985373e-05,
      "loss": 1.8233,
      "step": 3147
    },
    {
      "epoch": 0.5670794866021166,
      "grad_norm": 1.4078739881515503,
      "learning_rate": 8.151808430011848e-05,
      "loss": 1.5152,
      "step": 3148
    },
    {
      "epoch": 0.567259626210313,
      "grad_norm": 1.4590213298797607,
      "learning_rate": 8.150708942201788e-05,
      "loss": 1.5588,
      "step": 3149
    },
    {
      "epoch": 0.5674397658185093,
      "grad_norm": 1.358292818069458,
      "learning_rate": 8.149609201643391e-05,
      "loss": 1.3822,
      "step": 3150
    },
    {
      "epoch": 0.5676199054267057,
      "grad_norm": 1.3277286291122437,
      "learning_rate": 8.148509208424878e-05,
      "loss": 2.0926,
      "step": 3151
    },
    {
      "epoch": 0.567800045034902,
      "grad_norm": 1.2633247375488281,
      "learning_rate": 8.147408962634488e-05,
      "loss": 2.0861,
      "step": 3152
    },
    {
      "epoch": 0.5679801846430984,
      "grad_norm": 1.2951682806015015,
      "learning_rate": 8.146308464360489e-05,
      "loss": 2.0527,
      "step": 3153
    },
    {
      "epoch": 0.5681603242512947,
      "grad_norm": 1.1878705024719238,
      "learning_rate": 8.145207713691153e-05,
      "loss": 1.8376,
      "step": 3154
    },
    {
      "epoch": 0.5683404638594911,
      "grad_norm": 1.253411889076233,
      "learning_rate": 8.144106710714788e-05,
      "loss": 1.8045,
      "step": 3155
    },
    {
      "epoch": 0.5685206034676875,
      "grad_norm": 1.3438165187835693,
      "learning_rate": 8.143005455519715e-05,
      "loss": 2.0449,
      "step": 3156
    },
    {
      "epoch": 0.5687007430758838,
      "grad_norm": 1.3496830463409424,
      "learning_rate": 8.141903948194271e-05,
      "loss": 2.1844,
      "step": 3157
    },
    {
      "epoch": 0.5688808826840802,
      "grad_norm": 1.3739079236984253,
      "learning_rate": 8.140802188826827e-05,
      "loss": 1.8802,
      "step": 3158
    },
    {
      "epoch": 0.5690610222922765,
      "grad_norm": 1.5568959712982178,
      "learning_rate": 8.13970017750576e-05,
      "loss": 1.8717,
      "step": 3159
    },
    {
      "epoch": 0.5692411619004729,
      "grad_norm": 1.7015992403030396,
      "learning_rate": 8.138597914319474e-05,
      "loss": 1.9087,
      "step": 3160
    },
    {
      "epoch": 0.5694213015086692,
      "grad_norm": 1.2822550535202026,
      "learning_rate": 8.137495399356391e-05,
      "loss": 1.6845,
      "step": 3161
    },
    {
      "epoch": 0.5696014411168656,
      "grad_norm": 1.450877070426941,
      "learning_rate": 8.136392632704957e-05,
      "loss": 1.8716,
      "step": 3162
    },
    {
      "epoch": 0.5697815807250619,
      "grad_norm": 1.3381742238998413,
      "learning_rate": 8.135289614453633e-05,
      "loss": 1.679,
      "step": 3163
    },
    {
      "epoch": 0.5699617203332583,
      "grad_norm": 1.353865623474121,
      "learning_rate": 8.134186344690903e-05,
      "loss": 1.798,
      "step": 3164
    },
    {
      "epoch": 0.5701418599414546,
      "grad_norm": 1.2209184169769287,
      "learning_rate": 8.133082823505271e-05,
      "loss": 1.61,
      "step": 3165
    },
    {
      "epoch": 0.570321999549651,
      "grad_norm": 1.2871285676956177,
      "learning_rate": 8.131979050985262e-05,
      "loss": 1.4535,
      "step": 3166
    },
    {
      "epoch": 0.5705021391578473,
      "grad_norm": 1.202979326248169,
      "learning_rate": 8.130875027219417e-05,
      "loss": 1.5035,
      "step": 3167
    },
    {
      "epoch": 0.5706822787660437,
      "grad_norm": 1.2236607074737549,
      "learning_rate": 8.129770752296305e-05,
      "loss": 1.5575,
      "step": 3168
    },
    {
      "epoch": 0.57086241837424,
      "grad_norm": 1.18218994140625,
      "learning_rate": 8.128666226304507e-05,
      "loss": 1.5653,
      "step": 3169
    },
    {
      "epoch": 0.5710425579824364,
      "grad_norm": 1.3156682252883911,
      "learning_rate": 8.127561449332627e-05,
      "loss": 1.4856,
      "step": 3170
    },
    {
      "epoch": 0.5712226975906327,
      "grad_norm": 1.2198225259780884,
      "learning_rate": 8.126456421469291e-05,
      "loss": 1.5367,
      "step": 3171
    },
    {
      "epoch": 0.5714028371988291,
      "grad_norm": 1.2979079484939575,
      "learning_rate": 8.125351142803144e-05,
      "loss": 1.7432,
      "step": 3172
    },
    {
      "epoch": 0.5715829768070254,
      "grad_norm": 1.4458940029144287,
      "learning_rate": 8.124245613422851e-05,
      "loss": 1.6537,
      "step": 3173
    },
    {
      "epoch": 0.5717631164152218,
      "grad_norm": 1.4343434572219849,
      "learning_rate": 8.123139833417095e-05,
      "loss": 1.7432,
      "step": 3174
    },
    {
      "epoch": 0.5719432560234181,
      "grad_norm": 1.2733614444732666,
      "learning_rate": 8.122033802874585e-05,
      "loss": 1.8152,
      "step": 3175
    },
    {
      "epoch": 0.5721233956316145,
      "grad_norm": 1.32044517993927,
      "learning_rate": 8.120927521884044e-05,
      "loss": 1.8011,
      "step": 3176
    },
    {
      "epoch": 0.5723035352398108,
      "grad_norm": 1.2091959714889526,
      "learning_rate": 8.119820990534218e-05,
      "loss": 1.7056,
      "step": 3177
    },
    {
      "epoch": 0.5724836748480072,
      "grad_norm": 1.3584576845169067,
      "learning_rate": 8.11871420891387e-05,
      "loss": 1.8433,
      "step": 3178
    },
    {
      "epoch": 0.5726638144562035,
      "grad_norm": 1.2639418840408325,
      "learning_rate": 8.117607177111787e-05,
      "loss": 1.6686,
      "step": 3179
    },
    {
      "epoch": 0.5728439540643999,
      "grad_norm": 1.385926604270935,
      "learning_rate": 8.116499895216774e-05,
      "loss": 1.883,
      "step": 3180
    },
    {
      "epoch": 0.5730240936725962,
      "grad_norm": 1.3171542882919312,
      "learning_rate": 8.11539236331766e-05,
      "loss": 1.851,
      "step": 3181
    },
    {
      "epoch": 0.5732042332807926,
      "grad_norm": 1.26935875415802,
      "learning_rate": 8.114284581503287e-05,
      "loss": 1.5329,
      "step": 3182
    },
    {
      "epoch": 0.5733843728889889,
      "grad_norm": 1.4185380935668945,
      "learning_rate": 8.113176549862522e-05,
      "loss": 1.6581,
      "step": 3183
    },
    {
      "epoch": 0.5735645124971853,
      "grad_norm": 1.344731092453003,
      "learning_rate": 8.112068268484249e-05,
      "loss": 1.5609,
      "step": 3184
    },
    {
      "epoch": 0.5737446521053817,
      "grad_norm": 1.2733148336410522,
      "learning_rate": 8.110959737457377e-05,
      "loss": 1.6929,
      "step": 3185
    },
    {
      "epoch": 0.573924791713578,
      "grad_norm": 1.4895830154418945,
      "learning_rate": 8.10985095687083e-05,
      "loss": 1.8285,
      "step": 3186
    },
    {
      "epoch": 0.5741049313217744,
      "grad_norm": 1.3092437982559204,
      "learning_rate": 8.108741926813553e-05,
      "loss": 1.5719,
      "step": 3187
    },
    {
      "epoch": 0.5742850709299707,
      "grad_norm": 1.3088788986206055,
      "learning_rate": 8.107632647374514e-05,
      "loss": 1.6501,
      "step": 3188
    },
    {
      "epoch": 0.5744652105381671,
      "grad_norm": 1.3658629655838013,
      "learning_rate": 8.106523118642697e-05,
      "loss": 1.6845,
      "step": 3189
    },
    {
      "epoch": 0.5746453501463634,
      "grad_norm": 1.3632906675338745,
      "learning_rate": 8.105413340707109e-05,
      "loss": 1.8674,
      "step": 3190
    },
    {
      "epoch": 0.5748254897545598,
      "grad_norm": 1.3251127004623413,
      "learning_rate": 8.104303313656775e-05,
      "loss": 1.6598,
      "step": 3191
    },
    {
      "epoch": 0.5750056293627561,
      "grad_norm": 1.3589049577713013,
      "learning_rate": 8.10319303758074e-05,
      "loss": 1.6211,
      "step": 3192
    },
    {
      "epoch": 0.5751857689709525,
      "grad_norm": 1.352040410041809,
      "learning_rate": 8.102082512568075e-05,
      "loss": 1.4777,
      "step": 3193
    },
    {
      "epoch": 0.5753659085791488,
      "grad_norm": 1.4846113920211792,
      "learning_rate": 8.100971738707857e-05,
      "loss": 1.9339,
      "step": 3194
    },
    {
      "epoch": 0.5755460481873452,
      "grad_norm": 1.3877880573272705,
      "learning_rate": 8.0998607160892e-05,
      "loss": 1.6246,
      "step": 3195
    },
    {
      "epoch": 0.5757261877955415,
      "grad_norm": 1.3888198137283325,
      "learning_rate": 8.098749444801224e-05,
      "loss": 1.7753,
      "step": 3196
    },
    {
      "epoch": 0.5759063274037379,
      "grad_norm": 1.5230215787887573,
      "learning_rate": 8.09763792493308e-05,
      "loss": 1.6742,
      "step": 3197
    },
    {
      "epoch": 0.5760864670119342,
      "grad_norm": 1.4500510692596436,
      "learning_rate": 8.096526156573926e-05,
      "loss": 1.4329,
      "step": 3198
    },
    {
      "epoch": 0.5762666066201306,
      "grad_norm": 1.4574159383773804,
      "learning_rate": 8.095414139812956e-05,
      "loss": 1.6202,
      "step": 3199
    },
    {
      "epoch": 0.5764467462283269,
      "grad_norm": 1.290509581565857,
      "learning_rate": 8.09430187473937e-05,
      "loss": 1.2,
      "step": 3200
    },
    {
      "epoch": 0.5766268858365233,
      "grad_norm": 1.2402974367141724,
      "learning_rate": 8.093189361442394e-05,
      "loss": 1.7644,
      "step": 3201
    },
    {
      "epoch": 0.5768070254447196,
      "grad_norm": 1.2125972509384155,
      "learning_rate": 8.092076600011276e-05,
      "loss": 2.1467,
      "step": 3202
    },
    {
      "epoch": 0.576987165052916,
      "grad_norm": 1.2628872394561768,
      "learning_rate": 8.090963590535277e-05,
      "loss": 2.2245,
      "step": 3203
    },
    {
      "epoch": 0.5771673046611123,
      "grad_norm": 1.1404106616973877,
      "learning_rate": 8.089850333103684e-05,
      "loss": 1.856,
      "step": 3204
    },
    {
      "epoch": 0.5773474442693087,
      "grad_norm": 1.3384531736373901,
      "learning_rate": 8.088736827805805e-05,
      "loss": 1.935,
      "step": 3205
    },
    {
      "epoch": 0.577527583877505,
      "grad_norm": 1.2109678983688354,
      "learning_rate": 8.08762307473096e-05,
      "loss": 1.5656,
      "step": 3206
    },
    {
      "epoch": 0.5777077234857014,
      "grad_norm": 1.3816996812820435,
      "learning_rate": 8.086509073968496e-05,
      "loss": 1.7886,
      "step": 3207
    },
    {
      "epoch": 0.5778878630938977,
      "grad_norm": 1.5806138515472412,
      "learning_rate": 8.085394825607778e-05,
      "loss": 2.3206,
      "step": 3208
    },
    {
      "epoch": 0.5780680027020941,
      "grad_norm": 1.739669919013977,
      "learning_rate": 8.084280329738191e-05,
      "loss": 2.2176,
      "step": 3209
    },
    {
      "epoch": 0.5782481423102904,
      "grad_norm": 1.7253975868225098,
      "learning_rate": 8.083165586449138e-05,
      "loss": 2.2416,
      "step": 3210
    },
    {
      "epoch": 0.5784282819184868,
      "grad_norm": 1.3106157779693604,
      "learning_rate": 8.082050595830042e-05,
      "loss": 1.811,
      "step": 3211
    },
    {
      "epoch": 0.5786084215266831,
      "grad_norm": 1.2804299592971802,
      "learning_rate": 8.080935357970348e-05,
      "loss": 1.6173,
      "step": 3212
    },
    {
      "epoch": 0.5787885611348795,
      "grad_norm": 1.19856595993042,
      "learning_rate": 8.079819872959522e-05,
      "loss": 1.66,
      "step": 3213
    },
    {
      "epoch": 0.5789687007430759,
      "grad_norm": 1.1803557872772217,
      "learning_rate": 8.078704140887046e-05,
      "loss": 1.4163,
      "step": 3214
    },
    {
      "epoch": 0.5791488403512722,
      "grad_norm": 1.1865688562393188,
      "learning_rate": 8.077588161842422e-05,
      "loss": 1.475,
      "step": 3215
    },
    {
      "epoch": 0.5793289799594686,
      "grad_norm": 1.3032927513122559,
      "learning_rate": 8.076471935915176e-05,
      "loss": 1.6544,
      "step": 3216
    },
    {
      "epoch": 0.5795091195676649,
      "grad_norm": 1.2634093761444092,
      "learning_rate": 8.075355463194847e-05,
      "loss": 1.7275,
      "step": 3217
    },
    {
      "epoch": 0.5796892591758613,
      "grad_norm": 1.341326355934143,
      "learning_rate": 8.074238743771003e-05,
      "loss": 1.7533,
      "step": 3218
    },
    {
      "epoch": 0.5798693987840576,
      "grad_norm": 1.3142240047454834,
      "learning_rate": 8.073121777733226e-05,
      "loss": 1.8086,
      "step": 3219
    },
    {
      "epoch": 0.580049538392254,
      "grad_norm": 1.1341456174850464,
      "learning_rate": 8.072004565171114e-05,
      "loss": 1.4552,
      "step": 3220
    },
    {
      "epoch": 0.5802296780004503,
      "grad_norm": 1.1689975261688232,
      "learning_rate": 8.070887106174293e-05,
      "loss": 1.3858,
      "step": 3221
    },
    {
      "epoch": 0.5804098176086467,
      "grad_norm": 1.2383785247802734,
      "learning_rate": 8.069769400832404e-05,
      "loss": 1.5716,
      "step": 3222
    },
    {
      "epoch": 0.580589957216843,
      "grad_norm": 1.3554126024246216,
      "learning_rate": 8.06865144923511e-05,
      "loss": 1.7538,
      "step": 3223
    },
    {
      "epoch": 0.5807700968250394,
      "grad_norm": 1.5515005588531494,
      "learning_rate": 8.06753325147209e-05,
      "loss": 1.8926,
      "step": 3224
    },
    {
      "epoch": 0.5809502364332357,
      "grad_norm": 1.3006205558776855,
      "learning_rate": 8.066414807633047e-05,
      "loss": 1.5122,
      "step": 3225
    },
    {
      "epoch": 0.5811303760414321,
      "grad_norm": 1.249218463897705,
      "learning_rate": 8.065296117807703e-05,
      "loss": 1.7046,
      "step": 3226
    },
    {
      "epoch": 0.5813105156496284,
      "grad_norm": 1.3515232801437378,
      "learning_rate": 8.064177182085795e-05,
      "loss": 1.6931,
      "step": 3227
    },
    {
      "epoch": 0.5814906552578248,
      "grad_norm": 1.1837708950042725,
      "learning_rate": 8.06305800055709e-05,
      "loss": 1.4618,
      "step": 3228
    },
    {
      "epoch": 0.5816707948660211,
      "grad_norm": 1.3043564558029175,
      "learning_rate": 8.061938573311359e-05,
      "loss": 1.5691,
      "step": 3229
    },
    {
      "epoch": 0.5818509344742175,
      "grad_norm": 1.4088311195373535,
      "learning_rate": 8.060818900438411e-05,
      "loss": 1.4785,
      "step": 3230
    },
    {
      "epoch": 0.5820310740824138,
      "grad_norm": 1.2701168060302734,
      "learning_rate": 8.059698982028059e-05,
      "loss": 1.4091,
      "step": 3231
    },
    {
      "epoch": 0.5822112136906102,
      "grad_norm": 1.3696273565292358,
      "learning_rate": 8.058578818170147e-05,
      "loss": 1.7556,
      "step": 3232
    },
    {
      "epoch": 0.5823913532988065,
      "grad_norm": 1.3635327816009521,
      "learning_rate": 8.057458408954532e-05,
      "loss": 1.8731,
      "step": 3233
    },
    {
      "epoch": 0.5825714929070029,
      "grad_norm": 1.3212649822235107,
      "learning_rate": 8.05633775447109e-05,
      "loss": 1.726,
      "step": 3234
    },
    {
      "epoch": 0.5827516325151992,
      "grad_norm": 1.2060922384262085,
      "learning_rate": 8.055216854809726e-05,
      "loss": 1.4458,
      "step": 3235
    },
    {
      "epoch": 0.5829317721233956,
      "grad_norm": 1.4185734987258911,
      "learning_rate": 8.054095710060351e-05,
      "loss": 1.4539,
      "step": 3236
    },
    {
      "epoch": 0.5831119117315919,
      "grad_norm": 1.3189666271209717,
      "learning_rate": 8.052974320312907e-05,
      "loss": 1.8016,
      "step": 3237
    },
    {
      "epoch": 0.5832920513397883,
      "grad_norm": 1.191605806350708,
      "learning_rate": 8.05185268565735e-05,
      "loss": 1.6538,
      "step": 3238
    },
    {
      "epoch": 0.5834721909479846,
      "grad_norm": 1.3001261949539185,
      "learning_rate": 8.050730806183657e-05,
      "loss": 1.5849,
      "step": 3239
    },
    {
      "epoch": 0.583652330556181,
      "grad_norm": 1.5563225746154785,
      "learning_rate": 8.049608681981824e-05,
      "loss": 1.877,
      "step": 3240
    },
    {
      "epoch": 0.5838324701643773,
      "grad_norm": 1.3988347053527832,
      "learning_rate": 8.048486313141868e-05,
      "loss": 1.7948,
      "step": 3241
    },
    {
      "epoch": 0.5840126097725737,
      "grad_norm": 1.4530854225158691,
      "learning_rate": 8.047363699753827e-05,
      "loss": 1.8618,
      "step": 3242
    },
    {
      "epoch": 0.5841927493807701,
      "grad_norm": 1.329079270362854,
      "learning_rate": 8.046240841907752e-05,
      "loss": 1.717,
      "step": 3243
    },
    {
      "epoch": 0.5843728889889664,
      "grad_norm": 1.4676613807678223,
      "learning_rate": 8.045117739693722e-05,
      "loss": 1.62,
      "step": 3244
    },
    {
      "epoch": 0.5845530285971628,
      "grad_norm": 1.367106556892395,
      "learning_rate": 8.043994393201828e-05,
      "loss": 1.5984,
      "step": 3245
    },
    {
      "epoch": 0.5847331682053591,
      "grad_norm": 1.3436119556427002,
      "learning_rate": 8.042870802522189e-05,
      "loss": 1.5044,
      "step": 3246
    },
    {
      "epoch": 0.5849133078135555,
      "grad_norm": 1.3227373361587524,
      "learning_rate": 8.041746967744934e-05,
      "loss": 1.6565,
      "step": 3247
    },
    {
      "epoch": 0.5850934474217518,
      "grad_norm": 1.3055144548416138,
      "learning_rate": 8.040622888960221e-05,
      "loss": 1.5221,
      "step": 3248
    },
    {
      "epoch": 0.5852735870299483,
      "grad_norm": 1.421113133430481,
      "learning_rate": 8.03949856625822e-05,
      "loss": 1.4749,
      "step": 3249
    },
    {
      "epoch": 0.5854537266381445,
      "grad_norm": 1.351487159729004,
      "learning_rate": 8.038373999729123e-05,
      "loss": 1.4543,
      "step": 3250
    },
    {
      "epoch": 0.585633866246341,
      "grad_norm": 1.2402830123901367,
      "learning_rate": 8.037249189463145e-05,
      "loss": 1.9116,
      "step": 3251
    },
    {
      "epoch": 0.5858140058545372,
      "grad_norm": 1.2029560804367065,
      "learning_rate": 8.036124135550516e-05,
      "loss": 1.8752,
      "step": 3252
    },
    {
      "epoch": 0.5859941454627337,
      "grad_norm": 1.6283750534057617,
      "learning_rate": 8.03499883808149e-05,
      "loss": 2.3594,
      "step": 3253
    },
    {
      "epoch": 0.58617428507093,
      "grad_norm": 1.185128092765808,
      "learning_rate": 8.033873297146333e-05,
      "loss": 1.9623,
      "step": 3254
    },
    {
      "epoch": 0.5863544246791264,
      "grad_norm": 1.2744601964950562,
      "learning_rate": 8.032747512835337e-05,
      "loss": 2.067,
      "step": 3255
    },
    {
      "epoch": 0.5865345642873226,
      "grad_norm": 1.3466190099716187,
      "learning_rate": 8.031621485238815e-05,
      "loss": 2.1384,
      "step": 3256
    },
    {
      "epoch": 0.586714703895519,
      "grad_norm": 1.4176702499389648,
      "learning_rate": 8.030495214447093e-05,
      "loss": 1.8959,
      "step": 3257
    },
    {
      "epoch": 0.5868948435037153,
      "grad_norm": 1.4075653553009033,
      "learning_rate": 8.029368700550522e-05,
      "loss": 2.0826,
      "step": 3258
    },
    {
      "epoch": 0.5870749831119118,
      "grad_norm": 1.6770472526550293,
      "learning_rate": 8.028241943639468e-05,
      "loss": 2.1306,
      "step": 3259
    },
    {
      "epoch": 0.587255122720108,
      "grad_norm": 1.672384262084961,
      "learning_rate": 8.027114943804322e-05,
      "loss": 2.3935,
      "step": 3260
    },
    {
      "epoch": 0.5874352623283045,
      "grad_norm": 1.516764760017395,
      "learning_rate": 8.025987701135487e-05,
      "loss": 1.9128,
      "step": 3261
    },
    {
      "epoch": 0.5876154019365007,
      "grad_norm": 1.1755174398422241,
      "learning_rate": 8.024860215723396e-05,
      "loss": 1.556,
      "step": 3262
    },
    {
      "epoch": 0.5877955415446972,
      "grad_norm": 1.195533275604248,
      "learning_rate": 8.023732487658489e-05,
      "loss": 1.7773,
      "step": 3263
    },
    {
      "epoch": 0.5879756811528934,
      "grad_norm": 1.2383601665496826,
      "learning_rate": 8.022604517031236e-05,
      "loss": 1.6818,
      "step": 3264
    },
    {
      "epoch": 0.5881558207610899,
      "grad_norm": 1.1856685876846313,
      "learning_rate": 8.021476303932121e-05,
      "loss": 1.4957,
      "step": 3265
    },
    {
      "epoch": 0.5883359603692861,
      "grad_norm": 1.2532113790512085,
      "learning_rate": 8.020347848451647e-05,
      "loss": 1.5602,
      "step": 3266
    },
    {
      "epoch": 0.5885160999774826,
      "grad_norm": 1.2764549255371094,
      "learning_rate": 8.019219150680341e-05,
      "loss": 1.6345,
      "step": 3267
    },
    {
      "epoch": 0.5886962395856788,
      "grad_norm": 1.1813106536865234,
      "learning_rate": 8.018090210708746e-05,
      "loss": 1.4156,
      "step": 3268
    },
    {
      "epoch": 0.5888763791938753,
      "grad_norm": 1.1520346403121948,
      "learning_rate": 8.016961028627423e-05,
      "loss": 1.4903,
      "step": 3269
    },
    {
      "epoch": 0.5890565188020717,
      "grad_norm": 1.2903872728347778,
      "learning_rate": 8.015831604526957e-05,
      "loss": 1.553,
      "step": 3270
    },
    {
      "epoch": 0.589236658410268,
      "grad_norm": 1.2544065713882446,
      "learning_rate": 8.014701938497947e-05,
      "loss": 1.6497,
      "step": 3271
    },
    {
      "epoch": 0.5894167980184644,
      "grad_norm": 1.2569499015808105,
      "learning_rate": 8.013572030631017e-05,
      "loss": 1.5701,
      "step": 3272
    },
    {
      "epoch": 0.5895969376266607,
      "grad_norm": 1.1802705526351929,
      "learning_rate": 8.012441881016806e-05,
      "loss": 1.4772,
      "step": 3273
    },
    {
      "epoch": 0.5897770772348571,
      "grad_norm": 1.3752899169921875,
      "learning_rate": 8.011311489745976e-05,
      "loss": 1.5558,
      "step": 3274
    },
    {
      "epoch": 0.5899572168430534,
      "grad_norm": 1.2956390380859375,
      "learning_rate": 8.010180856909204e-05,
      "loss": 1.4747,
      "step": 3275
    },
    {
      "epoch": 0.5901373564512498,
      "grad_norm": 1.2420748472213745,
      "learning_rate": 8.00904998259719e-05,
      "loss": 1.664,
      "step": 3276
    },
    {
      "epoch": 0.5903174960594461,
      "grad_norm": 1.2409071922302246,
      "learning_rate": 8.007918866900653e-05,
      "loss": 1.4659,
      "step": 3277
    },
    {
      "epoch": 0.5904976356676425,
      "grad_norm": 1.4191685914993286,
      "learning_rate": 8.006787509910328e-05,
      "loss": 1.7158,
      "step": 3278
    },
    {
      "epoch": 0.5906777752758388,
      "grad_norm": 1.2597142457962036,
      "learning_rate": 8.005655911716974e-05,
      "loss": 1.5931,
      "step": 3279
    },
    {
      "epoch": 0.5908579148840352,
      "grad_norm": 1.317113995552063,
      "learning_rate": 8.004524072411367e-05,
      "loss": 1.4726,
      "step": 3280
    },
    {
      "epoch": 0.5910380544922315,
      "grad_norm": 1.3195549249649048,
      "learning_rate": 8.003391992084302e-05,
      "loss": 1.7517,
      "step": 3281
    },
    {
      "epoch": 0.5912181941004279,
      "grad_norm": 1.504854440689087,
      "learning_rate": 8.002259670826596e-05,
      "loss": 1.65,
      "step": 3282
    },
    {
      "epoch": 0.5913983337086242,
      "grad_norm": 1.3624268770217896,
      "learning_rate": 8.001127108729079e-05,
      "loss": 1.7472,
      "step": 3283
    },
    {
      "epoch": 0.5915784733168206,
      "grad_norm": 1.3879119157791138,
      "learning_rate": 7.999994305882608e-05,
      "loss": 1.7861,
      "step": 3284
    },
    {
      "epoch": 0.5917586129250169,
      "grad_norm": 1.6236234903335571,
      "learning_rate": 7.998861262378054e-05,
      "loss": 1.9822,
      "step": 3285
    },
    {
      "epoch": 0.5919387525332133,
      "grad_norm": 1.463417887687683,
      "learning_rate": 7.997727978306313e-05,
      "loss": 1.8017,
      "step": 3286
    },
    {
      "epoch": 0.5921188921414096,
      "grad_norm": 1.429961919784546,
      "learning_rate": 7.996594453758292e-05,
      "loss": 1.8323,
      "step": 3287
    },
    {
      "epoch": 0.592299031749606,
      "grad_norm": 1.2735933065414429,
      "learning_rate": 7.995460688824924e-05,
      "loss": 1.6421,
      "step": 3288
    },
    {
      "epoch": 0.5924791713578023,
      "grad_norm": 1.447253942489624,
      "learning_rate": 7.994326683597158e-05,
      "loss": 1.5854,
      "step": 3289
    },
    {
      "epoch": 0.5926593109659987,
      "grad_norm": 1.3713699579238892,
      "learning_rate": 7.993192438165964e-05,
      "loss": 1.5027,
      "step": 3290
    },
    {
      "epoch": 0.592839450574195,
      "grad_norm": 1.3277535438537598,
      "learning_rate": 7.992057952622332e-05,
      "loss": 1.6642,
      "step": 3291
    },
    {
      "epoch": 0.5930195901823914,
      "grad_norm": 1.32603919506073,
      "learning_rate": 7.990923227057266e-05,
      "loss": 1.6168,
      "step": 3292
    },
    {
      "epoch": 0.5931997297905877,
      "grad_norm": 1.4450396299362183,
      "learning_rate": 7.989788261561797e-05,
      "loss": 1.5499,
      "step": 3293
    },
    {
      "epoch": 0.5933798693987841,
      "grad_norm": 1.3398592472076416,
      "learning_rate": 7.98865305622697e-05,
      "loss": 1.6352,
      "step": 3294
    },
    {
      "epoch": 0.5935600090069804,
      "grad_norm": 1.3279551267623901,
      "learning_rate": 7.98751761114385e-05,
      "loss": 1.4219,
      "step": 3295
    },
    {
      "epoch": 0.5937401486151768,
      "grad_norm": 1.3421669006347656,
      "learning_rate": 7.986381926403524e-05,
      "loss": 1.7277,
      "step": 3296
    },
    {
      "epoch": 0.5939202882233731,
      "grad_norm": 1.4003654718399048,
      "learning_rate": 7.985246002097092e-05,
      "loss": 1.6725,
      "step": 3297
    },
    {
      "epoch": 0.5941004278315695,
      "grad_norm": 1.4394348859786987,
      "learning_rate": 7.984109838315682e-05,
      "loss": 1.6146,
      "step": 3298
    },
    {
      "epoch": 0.5942805674397659,
      "grad_norm": 1.2530721426010132,
      "learning_rate": 7.982973435150433e-05,
      "loss": 1.4136,
      "step": 3299
    },
    {
      "epoch": 0.5944607070479622,
      "grad_norm": 1.3988275527954102,
      "learning_rate": 7.981836792692508e-05,
      "loss": 1.3519,
      "step": 3300
    },
    {
      "epoch": 0.5946408466561586,
      "grad_norm": 1.0750325918197632,
      "learning_rate": 7.980699911033087e-05,
      "loss": 2.0638,
      "step": 3301
    },
    {
      "epoch": 0.5948209862643549,
      "grad_norm": 1.0915641784667969,
      "learning_rate": 7.979562790263373e-05,
      "loss": 1.9258,
      "step": 3302
    },
    {
      "epoch": 0.5950011258725513,
      "grad_norm": 1.232743501663208,
      "learning_rate": 7.978425430474581e-05,
      "loss": 2.2529,
      "step": 3303
    },
    {
      "epoch": 0.5951812654807476,
      "grad_norm": 1.2405080795288086,
      "learning_rate": 7.977287831757954e-05,
      "loss": 2.0744,
      "step": 3304
    },
    {
      "epoch": 0.595361405088944,
      "grad_norm": 1.2133256196975708,
      "learning_rate": 7.976149994204747e-05,
      "loss": 1.8397,
      "step": 3305
    },
    {
      "epoch": 0.5955415446971403,
      "grad_norm": 1.2922027111053467,
      "learning_rate": 7.975011917906236e-05,
      "loss": 2.122,
      "step": 3306
    },
    {
      "epoch": 0.5957216843053367,
      "grad_norm": 1.2730096578598022,
      "learning_rate": 7.973873602953719e-05,
      "loss": 1.9076,
      "step": 3307
    },
    {
      "epoch": 0.595901823913533,
      "grad_norm": 1.3810216188430786,
      "learning_rate": 7.972735049438509e-05,
      "loss": 1.6815,
      "step": 3308
    },
    {
      "epoch": 0.5960819635217294,
      "grad_norm": 1.4814426898956299,
      "learning_rate": 7.97159625745194e-05,
      "loss": 2.0047,
      "step": 3309
    },
    {
      "epoch": 0.5962621031299257,
      "grad_norm": 1.6553380489349365,
      "learning_rate": 7.97045722708537e-05,
      "loss": 2.2769,
      "step": 3310
    },
    {
      "epoch": 0.5964422427381221,
      "grad_norm": 1.4031059741973877,
      "learning_rate": 7.969317958430166e-05,
      "loss": 2.0157,
      "step": 3311
    },
    {
      "epoch": 0.5966223823463184,
      "grad_norm": 1.3077967166900635,
      "learning_rate": 7.968178451577724e-05,
      "loss": 1.7834,
      "step": 3312
    },
    {
      "epoch": 0.5968025219545148,
      "grad_norm": 1.2406649589538574,
      "learning_rate": 7.96703870661945e-05,
      "loss": 1.558,
      "step": 3313
    },
    {
      "epoch": 0.5969826615627111,
      "grad_norm": 1.2428460121154785,
      "learning_rate": 7.965898723646776e-05,
      "loss": 1.5469,
      "step": 3314
    },
    {
      "epoch": 0.5971628011709075,
      "grad_norm": 1.2896435260772705,
      "learning_rate": 7.964758502751153e-05,
      "loss": 1.4554,
      "step": 3315
    },
    {
      "epoch": 0.5973429407791038,
      "grad_norm": 1.2954893112182617,
      "learning_rate": 7.963618044024046e-05,
      "loss": 1.5566,
      "step": 3316
    },
    {
      "epoch": 0.5975230803873002,
      "grad_norm": 1.3373665809631348,
      "learning_rate": 7.962477347556941e-05,
      "loss": 1.7136,
      "step": 3317
    },
    {
      "epoch": 0.5977032199954965,
      "grad_norm": 1.358614206314087,
      "learning_rate": 7.96133641344135e-05,
      "loss": 1.5006,
      "step": 3318
    },
    {
      "epoch": 0.5978833596036929,
      "grad_norm": 1.3425312042236328,
      "learning_rate": 7.960195241768791e-05,
      "loss": 1.6973,
      "step": 3319
    },
    {
      "epoch": 0.5980634992118892,
      "grad_norm": 1.2611958980560303,
      "learning_rate": 7.959053832630813e-05,
      "loss": 1.5634,
      "step": 3320
    },
    {
      "epoch": 0.5982436388200856,
      "grad_norm": 1.2140748500823975,
      "learning_rate": 7.957912186118979e-05,
      "loss": 1.2989,
      "step": 3321
    },
    {
      "epoch": 0.5984237784282819,
      "grad_norm": 1.2744649648666382,
      "learning_rate": 7.956770302324867e-05,
      "loss": 1.5584,
      "step": 3322
    },
    {
      "epoch": 0.5986039180364783,
      "grad_norm": 1.2696504592895508,
      "learning_rate": 7.955628181340084e-05,
      "loss": 1.6822,
      "step": 3323
    },
    {
      "epoch": 0.5987840576446746,
      "grad_norm": 1.260656714439392,
      "learning_rate": 7.954485823256247e-05,
      "loss": 1.5254,
      "step": 3324
    },
    {
      "epoch": 0.598964197252871,
      "grad_norm": 1.2681546211242676,
      "learning_rate": 7.953343228164998e-05,
      "loss": 1.6567,
      "step": 3325
    },
    {
      "epoch": 0.5991443368610673,
      "grad_norm": 1.3946624994277954,
      "learning_rate": 7.952200396157991e-05,
      "loss": 1.803,
      "step": 3326
    },
    {
      "epoch": 0.5993244764692637,
      "grad_norm": 1.3119937181472778,
      "learning_rate": 7.951057327326908e-05,
      "loss": 1.772,
      "step": 3327
    },
    {
      "epoch": 0.5995046160774601,
      "grad_norm": 1.2493733167648315,
      "learning_rate": 7.949914021763444e-05,
      "loss": 1.5276,
      "step": 3328
    },
    {
      "epoch": 0.5996847556856564,
      "grad_norm": 1.3633266687393188,
      "learning_rate": 7.948770479559314e-05,
      "loss": 1.6384,
      "step": 3329
    },
    {
      "epoch": 0.5998648952938528,
      "grad_norm": 1.3602796792984009,
      "learning_rate": 7.94762670080625e-05,
      "loss": 1.6157,
      "step": 3330
    },
    {
      "epoch": 0.6000450349020491,
      "grad_norm": 1.277429461479187,
      "learning_rate": 7.946482685596011e-05,
      "loss": 1.4277,
      "step": 3331
    },
    {
      "epoch": 0.6002251745102455,
      "grad_norm": 1.2905783653259277,
      "learning_rate": 7.945338434020364e-05,
      "loss": 1.5763,
      "step": 3332
    },
    {
      "epoch": 0.6004053141184418,
      "grad_norm": 1.4707800149917603,
      "learning_rate": 7.944193946171103e-05,
      "loss": 1.8454,
      "step": 3333
    },
    {
      "epoch": 0.6005854537266382,
      "grad_norm": 1.331000804901123,
      "learning_rate": 7.943049222140037e-05,
      "loss": 1.6466,
      "step": 3334
    },
    {
      "epoch": 0.6007655933348345,
      "grad_norm": 1.3378723859786987,
      "learning_rate": 7.941904262018998e-05,
      "loss": 1.6085,
      "step": 3335
    },
    {
      "epoch": 0.6009457329430309,
      "grad_norm": 1.4617573022842407,
      "learning_rate": 7.94075906589983e-05,
      "loss": 1.7044,
      "step": 3336
    },
    {
      "epoch": 0.6011258725512272,
      "grad_norm": 1.3141578435897827,
      "learning_rate": 7.939613633874403e-05,
      "loss": 1.598,
      "step": 3337
    },
    {
      "epoch": 0.6013060121594236,
      "grad_norm": 1.4842551946640015,
      "learning_rate": 7.938467966034601e-05,
      "loss": 1.8941,
      "step": 3338
    },
    {
      "epoch": 0.6014861517676199,
      "grad_norm": 1.3281524181365967,
      "learning_rate": 7.937322062472333e-05,
      "loss": 1.567,
      "step": 3339
    },
    {
      "epoch": 0.6016662913758163,
      "grad_norm": 1.2148057222366333,
      "learning_rate": 7.936175923279519e-05,
      "loss": 1.4619,
      "step": 3340
    },
    {
      "epoch": 0.6018464309840126,
      "grad_norm": 1.467674732208252,
      "learning_rate": 7.935029548548099e-05,
      "loss": 1.7453,
      "step": 3341
    },
    {
      "epoch": 0.602026570592209,
      "grad_norm": 1.2933822870254517,
      "learning_rate": 7.933882938370042e-05,
      "loss": 1.4134,
      "step": 3342
    },
    {
      "epoch": 0.6022067102004053,
      "grad_norm": 1.387030839920044,
      "learning_rate": 7.932736092837323e-05,
      "loss": 1.4154,
      "step": 3343
    },
    {
      "epoch": 0.6023868498086017,
      "grad_norm": 1.3232898712158203,
      "learning_rate": 7.931589012041944e-05,
      "loss": 1.5633,
      "step": 3344
    },
    {
      "epoch": 0.602566989416798,
      "grad_norm": 1.2709810733795166,
      "learning_rate": 7.930441696075921e-05,
      "loss": 1.4484,
      "step": 3345
    },
    {
      "epoch": 0.6027471290249944,
      "grad_norm": 1.5326019525527954,
      "learning_rate": 7.929294145031294e-05,
      "loss": 1.8422,
      "step": 3346
    },
    {
      "epoch": 0.6029272686331907,
      "grad_norm": 1.559929370880127,
      "learning_rate": 7.928146359000117e-05,
      "loss": 1.6572,
      "step": 3347
    },
    {
      "epoch": 0.6031074082413871,
      "grad_norm": 1.3563951253890991,
      "learning_rate": 7.926998338074465e-05,
      "loss": 1.3689,
      "step": 3348
    },
    {
      "epoch": 0.6032875478495834,
      "grad_norm": 1.3260586261749268,
      "learning_rate": 7.925850082346433e-05,
      "loss": 1.4621,
      "step": 3349
    },
    {
      "epoch": 0.6034676874577798,
      "grad_norm": 1.1549409627914429,
      "learning_rate": 7.924701591908132e-05,
      "loss": 1.1978,
      "step": 3350
    },
    {
      "epoch": 0.6036478270659761,
      "grad_norm": 1.2072349786758423,
      "learning_rate": 7.923552866851693e-05,
      "loss": 1.7823,
      "step": 3351
    },
    {
      "epoch": 0.6038279666741725,
      "grad_norm": 1.190442442893982,
      "learning_rate": 7.922403907269267e-05,
      "loss": 1.893,
      "step": 3352
    },
    {
      "epoch": 0.6040081062823688,
      "grad_norm": 1.2276901006698608,
      "learning_rate": 7.921254713253024e-05,
      "loss": 1.9304,
      "step": 3353
    },
    {
      "epoch": 0.6041882458905652,
      "grad_norm": 1.2261402606964111,
      "learning_rate": 7.920105284895152e-05,
      "loss": 2.0447,
      "step": 3354
    },
    {
      "epoch": 0.6043683854987616,
      "grad_norm": 1.3613996505737305,
      "learning_rate": 7.918955622287854e-05,
      "loss": 1.9871,
      "step": 3355
    },
    {
      "epoch": 0.6045485251069579,
      "grad_norm": 1.402151346206665,
      "learning_rate": 7.91780572552336e-05,
      "loss": 2.2685,
      "step": 3356
    },
    {
      "epoch": 0.6047286647151543,
      "grad_norm": 1.4284284114837646,
      "learning_rate": 7.916655594693909e-05,
      "loss": 1.8568,
      "step": 3357
    },
    {
      "epoch": 0.6049088043233506,
      "grad_norm": 1.518607258796692,
      "learning_rate": 7.915505229891769e-05,
      "loss": 2.0351,
      "step": 3358
    },
    {
      "epoch": 0.605088943931547,
      "grad_norm": 1.6458548307418823,
      "learning_rate": 7.914354631209219e-05,
      "loss": 1.9328,
      "step": 3359
    },
    {
      "epoch": 0.6052690835397433,
      "grad_norm": 1.7217801809310913,
      "learning_rate": 7.91320379873856e-05,
      "loss": 2.081,
      "step": 3360
    },
    {
      "epoch": 0.6054492231479397,
      "grad_norm": 1.547681450843811,
      "learning_rate": 7.912052732572111e-05,
      "loss": 1.6799,
      "step": 3361
    },
    {
      "epoch": 0.605629362756136,
      "grad_norm": 1.3306622505187988,
      "learning_rate": 7.910901432802212e-05,
      "loss": 1.63,
      "step": 3362
    },
    {
      "epoch": 0.6058095023643324,
      "grad_norm": 1.4279226064682007,
      "learning_rate": 7.909749899521215e-05,
      "loss": 1.6464,
      "step": 3363
    },
    {
      "epoch": 0.6059896419725287,
      "grad_norm": 1.4367555379867554,
      "learning_rate": 7.9085981328215e-05,
      "loss": 1.7141,
      "step": 3364
    },
    {
      "epoch": 0.6061697815807251,
      "grad_norm": 1.1761301755905151,
      "learning_rate": 7.907446132795459e-05,
      "loss": 1.4359,
      "step": 3365
    },
    {
      "epoch": 0.6063499211889214,
      "grad_norm": 1.3685128688812256,
      "learning_rate": 7.906293899535505e-05,
      "loss": 1.5781,
      "step": 3366
    },
    {
      "epoch": 0.6065300607971178,
      "grad_norm": 1.2375048398971558,
      "learning_rate": 7.905141433134071e-05,
      "loss": 1.4272,
      "step": 3367
    },
    {
      "epoch": 0.6067102004053141,
      "grad_norm": 1.3058501482009888,
      "learning_rate": 7.903988733683606e-05,
      "loss": 1.582,
      "step": 3368
    },
    {
      "epoch": 0.6068903400135105,
      "grad_norm": 1.3604326248168945,
      "learning_rate": 7.902835801276578e-05,
      "loss": 1.8541,
      "step": 3369
    },
    {
      "epoch": 0.6070704796217068,
      "grad_norm": 1.1521631479263306,
      "learning_rate": 7.901682636005478e-05,
      "loss": 1.485,
      "step": 3370
    },
    {
      "epoch": 0.6072506192299032,
      "grad_norm": 1.3001232147216797,
      "learning_rate": 7.90052923796281e-05,
      "loss": 1.6517,
      "step": 3371
    },
    {
      "epoch": 0.6074307588380995,
      "grad_norm": 1.4358909130096436,
      "learning_rate": 7.899375607241097e-05,
      "loss": 1.7774,
      "step": 3372
    },
    {
      "epoch": 0.6076108984462959,
      "grad_norm": 1.2707207202911377,
      "learning_rate": 7.898221743932888e-05,
      "loss": 1.6063,
      "step": 3373
    },
    {
      "epoch": 0.6077910380544922,
      "grad_norm": 1.3460183143615723,
      "learning_rate": 7.89706764813074e-05,
      "loss": 1.6385,
      "step": 3374
    },
    {
      "epoch": 0.6079711776626886,
      "grad_norm": 1.3746737241744995,
      "learning_rate": 7.895913319927239e-05,
      "loss": 1.9249,
      "step": 3375
    },
    {
      "epoch": 0.6081513172708849,
      "grad_norm": 1.3329097032546997,
      "learning_rate": 7.89475875941498e-05,
      "loss": 1.6834,
      "step": 3376
    },
    {
      "epoch": 0.6083314568790813,
      "grad_norm": 1.2083162069320679,
      "learning_rate": 7.893603966686584e-05,
      "loss": 1.4734,
      "step": 3377
    },
    {
      "epoch": 0.6085115964872776,
      "grad_norm": 1.1004382371902466,
      "learning_rate": 7.892448941834687e-05,
      "loss": 1.4186,
      "step": 3378
    },
    {
      "epoch": 0.608691736095474,
      "grad_norm": 1.3928534984588623,
      "learning_rate": 7.891293684951945e-05,
      "loss": 1.7284,
      "step": 3379
    },
    {
      "epoch": 0.6088718757036703,
      "grad_norm": 1.2312918901443481,
      "learning_rate": 7.890138196131031e-05,
      "loss": 1.6094,
      "step": 3380
    },
    {
      "epoch": 0.6090520153118667,
      "grad_norm": 1.2159086465835571,
      "learning_rate": 7.88898247546464e-05,
      "loss": 1.5268,
      "step": 3381
    },
    {
      "epoch": 0.609232154920063,
      "grad_norm": 1.2257739305496216,
      "learning_rate": 7.887826523045481e-05,
      "loss": 1.665,
      "step": 3382
    },
    {
      "epoch": 0.6094122945282594,
      "grad_norm": 1.3303282260894775,
      "learning_rate": 7.886670338966284e-05,
      "loss": 1.5839,
      "step": 3383
    },
    {
      "epoch": 0.6095924341364558,
      "grad_norm": 1.3492027521133423,
      "learning_rate": 7.8855139233198e-05,
      "loss": 1.6221,
      "step": 3384
    },
    {
      "epoch": 0.6097725737446521,
      "grad_norm": 1.1684311628341675,
      "learning_rate": 7.884357276198795e-05,
      "loss": 1.2783,
      "step": 3385
    },
    {
      "epoch": 0.6099527133528485,
      "grad_norm": 1.3742403984069824,
      "learning_rate": 7.883200397696053e-05,
      "loss": 1.4889,
      "step": 3386
    },
    {
      "epoch": 0.6101328529610448,
      "grad_norm": 1.3382236957550049,
      "learning_rate": 7.882043287904379e-05,
      "loss": 1.6217,
      "step": 3387
    },
    {
      "epoch": 0.6103129925692412,
      "grad_norm": 1.3607343435287476,
      "learning_rate": 7.880885946916598e-05,
      "loss": 1.4183,
      "step": 3388
    },
    {
      "epoch": 0.6104931321774375,
      "grad_norm": 1.3939534425735474,
      "learning_rate": 7.879728374825551e-05,
      "loss": 1.5689,
      "step": 3389
    },
    {
      "epoch": 0.6106732717856339,
      "grad_norm": 1.3255269527435303,
      "learning_rate": 7.878570571724092e-05,
      "loss": 1.7329,
      "step": 3390
    },
    {
      "epoch": 0.6108534113938302,
      "grad_norm": 1.2737993001937866,
      "learning_rate": 7.877412537705107e-05,
      "loss": 1.7746,
      "step": 3391
    },
    {
      "epoch": 0.6110335510020266,
      "grad_norm": 1.463544487953186,
      "learning_rate": 7.876254272861487e-05,
      "loss": 1.7846,
      "step": 3392
    },
    {
      "epoch": 0.6112136906102229,
      "grad_norm": 1.3744786977767944,
      "learning_rate": 7.875095777286153e-05,
      "loss": 1.5589,
      "step": 3393
    },
    {
      "epoch": 0.6113938302184193,
      "grad_norm": 1.3302916288375854,
      "learning_rate": 7.873937051072035e-05,
      "loss": 1.7601,
      "step": 3394
    },
    {
      "epoch": 0.6115739698266156,
      "grad_norm": 1.4363634586334229,
      "learning_rate": 7.872778094312086e-05,
      "loss": 1.6493,
      "step": 3395
    },
    {
      "epoch": 0.611754109434812,
      "grad_norm": 1.3024040460586548,
      "learning_rate": 7.871618907099279e-05,
      "loss": 1.6299,
      "step": 3396
    },
    {
      "epoch": 0.6119342490430083,
      "grad_norm": 1.340654730796814,
      "learning_rate": 7.870459489526602e-05,
      "loss": 1.477,
      "step": 3397
    },
    {
      "epoch": 0.6121143886512047,
      "grad_norm": 1.3368935585021973,
      "learning_rate": 7.869299841687061e-05,
      "loss": 1.2291,
      "step": 3398
    },
    {
      "epoch": 0.612294528259401,
      "grad_norm": 1.4286428689956665,
      "learning_rate": 7.868139963673686e-05,
      "loss": 1.5098,
      "step": 3399
    },
    {
      "epoch": 0.6124746678675974,
      "grad_norm": 1.2529758214950562,
      "learning_rate": 7.86697985557952e-05,
      "loss": 1.5391,
      "step": 3400
    },
    {
      "epoch": 0.6126548074757937,
      "grad_norm": 1.1250114440917969,
      "learning_rate": 7.865819517497626e-05,
      "loss": 2.1686,
      "step": 3401
    },
    {
      "epoch": 0.6128349470839901,
      "grad_norm": 1.1796998977661133,
      "learning_rate": 7.864658949521087e-05,
      "loss": 1.8919,
      "step": 3402
    },
    {
      "epoch": 0.6130150866921864,
      "grad_norm": 1.1717329025268555,
      "learning_rate": 7.863498151743e-05,
      "loss": 1.9769,
      "step": 3403
    },
    {
      "epoch": 0.6131952263003828,
      "grad_norm": 1.2340620756149292,
      "learning_rate": 7.862337124256487e-05,
      "loss": 1.8723,
      "step": 3404
    },
    {
      "epoch": 0.6133753659085791,
      "grad_norm": 1.4481773376464844,
      "learning_rate": 7.861175867154686e-05,
      "loss": 2.3697,
      "step": 3405
    },
    {
      "epoch": 0.6135555055167755,
      "grad_norm": 1.1795804500579834,
      "learning_rate": 7.860014380530749e-05,
      "loss": 1.8402,
      "step": 3406
    },
    {
      "epoch": 0.6137356451249718,
      "grad_norm": 1.2762295007705688,
      "learning_rate": 7.858852664477851e-05,
      "loss": 1.7181,
      "step": 3407
    },
    {
      "epoch": 0.6139157847331682,
      "grad_norm": 1.367079257965088,
      "learning_rate": 7.857690719089185e-05,
      "loss": 2.0054,
      "step": 3408
    },
    {
      "epoch": 0.6140959243413645,
      "grad_norm": 1.3670467138290405,
      "learning_rate": 7.856528544457964e-05,
      "loss": 1.8649,
      "step": 3409
    },
    {
      "epoch": 0.6142760639495609,
      "grad_norm": 1.7702361345291138,
      "learning_rate": 7.855366140677412e-05,
      "loss": 2.2306,
      "step": 3410
    },
    {
      "epoch": 0.6144562035577572,
      "grad_norm": 1.5775545835494995,
      "learning_rate": 7.854203507840778e-05,
      "loss": 2.0586,
      "step": 3411
    },
    {
      "epoch": 0.6146363431659536,
      "grad_norm": 1.212159514427185,
      "learning_rate": 7.853040646041332e-05,
      "loss": 1.6627,
      "step": 3412
    },
    {
      "epoch": 0.61481648277415,
      "grad_norm": 1.323357343673706,
      "learning_rate": 7.851877555372353e-05,
      "loss": 1.7685,
      "step": 3413
    },
    {
      "epoch": 0.6149966223823463,
      "grad_norm": 1.1844624280929565,
      "learning_rate": 7.850714235927145e-05,
      "loss": 1.5377,
      "step": 3414
    },
    {
      "epoch": 0.6151767619905427,
      "grad_norm": 1.4404406547546387,
      "learning_rate": 7.84955068779903e-05,
      "loss": 1.5404,
      "step": 3415
    },
    {
      "epoch": 0.615356901598739,
      "grad_norm": 1.3945233821868896,
      "learning_rate": 7.84838691108135e-05,
      "loss": 2.0254,
      "step": 3416
    },
    {
      "epoch": 0.6155370412069354,
      "grad_norm": 1.2886253595352173,
      "learning_rate": 7.847222905867455e-05,
      "loss": 1.583,
      "step": 3417
    },
    {
      "epoch": 0.6157171808151317,
      "grad_norm": 1.2599767446517944,
      "learning_rate": 7.846058672250727e-05,
      "loss": 1.5903,
      "step": 3418
    },
    {
      "epoch": 0.6158973204233281,
      "grad_norm": 1.1815823316574097,
      "learning_rate": 7.844894210324559e-05,
      "loss": 1.3637,
      "step": 3419
    },
    {
      "epoch": 0.6160774600315244,
      "grad_norm": 1.2556649446487427,
      "learning_rate": 7.843729520182362e-05,
      "loss": 1.6547,
      "step": 3420
    },
    {
      "epoch": 0.6162575996397208,
      "grad_norm": 1.3473434448242188,
      "learning_rate": 7.842564601917569e-05,
      "loss": 1.4851,
      "step": 3421
    },
    {
      "epoch": 0.6164377392479171,
      "grad_norm": 1.2386751174926758,
      "learning_rate": 7.841399455623629e-05,
      "loss": 1.5664,
      "step": 3422
    },
    {
      "epoch": 0.6166178788561135,
      "grad_norm": 1.5451143980026245,
      "learning_rate": 7.840234081394007e-05,
      "loss": 1.7077,
      "step": 3423
    },
    {
      "epoch": 0.6167980184643098,
      "grad_norm": 1.1659245491027832,
      "learning_rate": 7.83906847932219e-05,
      "loss": 1.5656,
      "step": 3424
    },
    {
      "epoch": 0.6169781580725062,
      "grad_norm": 1.2365696430206299,
      "learning_rate": 7.837902649501684e-05,
      "loss": 1.6032,
      "step": 3425
    },
    {
      "epoch": 0.6171582976807025,
      "grad_norm": 1.2708405256271362,
      "learning_rate": 7.836736592026009e-05,
      "loss": 1.543,
      "step": 3426
    },
    {
      "epoch": 0.6173384372888989,
      "grad_norm": 1.3413358926773071,
      "learning_rate": 7.835570306988706e-05,
      "loss": 1.6978,
      "step": 3427
    },
    {
      "epoch": 0.6175185768970952,
      "grad_norm": 1.3258224725723267,
      "learning_rate": 7.834403794483336e-05,
      "loss": 1.5797,
      "step": 3428
    },
    {
      "epoch": 0.6176987165052916,
      "grad_norm": 1.3221250772476196,
      "learning_rate": 7.833237054603473e-05,
      "loss": 1.5992,
      "step": 3429
    },
    {
      "epoch": 0.6178788561134879,
      "grad_norm": 1.235971450805664,
      "learning_rate": 7.832070087442714e-05,
      "loss": 1.3491,
      "step": 3430
    },
    {
      "epoch": 0.6180589957216843,
      "grad_norm": 1.4612895250320435,
      "learning_rate": 7.83090289309467e-05,
      "loss": 1.6087,
      "step": 3431
    },
    {
      "epoch": 0.6182391353298806,
      "grad_norm": 1.3738644123077393,
      "learning_rate": 7.829735471652978e-05,
      "loss": 1.5013,
      "step": 3432
    },
    {
      "epoch": 0.618419274938077,
      "grad_norm": 1.282684087753296,
      "learning_rate": 7.828567823211282e-05,
      "loss": 1.6167,
      "step": 3433
    },
    {
      "epoch": 0.6185994145462733,
      "grad_norm": 1.2626640796661377,
      "learning_rate": 7.827399947863254e-05,
      "loss": 1.4793,
      "step": 3434
    },
    {
      "epoch": 0.6187795541544697,
      "grad_norm": 1.330369234085083,
      "learning_rate": 7.826231845702578e-05,
      "loss": 1.5556,
      "step": 3435
    },
    {
      "epoch": 0.618959693762666,
      "grad_norm": 1.2297042608261108,
      "learning_rate": 7.825063516822961e-05,
      "loss": 1.4004,
      "step": 3436
    },
    {
      "epoch": 0.6191398333708624,
      "grad_norm": 1.3232755661010742,
      "learning_rate": 7.823894961318125e-05,
      "loss": 1.5041,
      "step": 3437
    },
    {
      "epoch": 0.6193199729790587,
      "grad_norm": 1.2605990171432495,
      "learning_rate": 7.822726179281812e-05,
      "loss": 1.3582,
      "step": 3438
    },
    {
      "epoch": 0.6195001125872551,
      "grad_norm": 1.4318681955337524,
      "learning_rate": 7.821557170807778e-05,
      "loss": 1.8279,
      "step": 3439
    },
    {
      "epoch": 0.6196802521954514,
      "grad_norm": 1.4431785345077515,
      "learning_rate": 7.820387935989802e-05,
      "loss": 1.9802,
      "step": 3440
    },
    {
      "epoch": 0.6198603918036478,
      "grad_norm": 1.3473280668258667,
      "learning_rate": 7.81921847492168e-05,
      "loss": 1.4405,
      "step": 3441
    },
    {
      "epoch": 0.6200405314118442,
      "grad_norm": 1.4384108781814575,
      "learning_rate": 7.818048787697225e-05,
      "loss": 1.8097,
      "step": 3442
    },
    {
      "epoch": 0.6202206710200405,
      "grad_norm": 1.236392617225647,
      "learning_rate": 7.816878874410269e-05,
      "loss": 1.5365,
      "step": 3443
    },
    {
      "epoch": 0.6204008106282369,
      "grad_norm": 1.3704609870910645,
      "learning_rate": 7.815708735154663e-05,
      "loss": 1.5112,
      "step": 3444
    },
    {
      "epoch": 0.6205809502364332,
      "grad_norm": 1.439429759979248,
      "learning_rate": 7.814538370024271e-05,
      "loss": 1.9872,
      "step": 3445
    },
    {
      "epoch": 0.6207610898446296,
      "grad_norm": 1.3481985330581665,
      "learning_rate": 7.813367779112984e-05,
      "loss": 1.5367,
      "step": 3446
    },
    {
      "epoch": 0.6209412294528259,
      "grad_norm": 1.5427281856536865,
      "learning_rate": 7.812196962514704e-05,
      "loss": 1.6443,
      "step": 3447
    },
    {
      "epoch": 0.6211213690610223,
      "grad_norm": 1.3741356134414673,
      "learning_rate": 7.811025920323354e-05,
      "loss": 1.6081,
      "step": 3448
    },
    {
      "epoch": 0.6213015086692186,
      "grad_norm": 1.262168049812317,
      "learning_rate": 7.809854652632872e-05,
      "loss": 1.3483,
      "step": 3449
    },
    {
      "epoch": 0.621481648277415,
      "grad_norm": 1.1154948472976685,
      "learning_rate": 7.808683159537219e-05,
      "loss": 1.1129,
      "step": 3450
    },
    {
      "epoch": 0.6216617878856113,
      "grad_norm": 1.092190146446228,
      "learning_rate": 7.807511441130371e-05,
      "loss": 1.8382,
      "step": 3451
    },
    {
      "epoch": 0.6218419274938077,
      "grad_norm": 1.0987615585327148,
      "learning_rate": 7.806339497506324e-05,
      "loss": 1.7497,
      "step": 3452
    },
    {
      "epoch": 0.622022067102004,
      "grad_norm": 1.2533376216888428,
      "learning_rate": 7.805167328759087e-05,
      "loss": 2.0765,
      "step": 3453
    },
    {
      "epoch": 0.6222022067102004,
      "grad_norm": 2.7416746616363525,
      "learning_rate": 7.803994934982696e-05,
      "loss": 2.1189,
      "step": 3454
    },
    {
      "epoch": 0.6223823463183967,
      "grad_norm": 1.2867873907089233,
      "learning_rate": 7.802822316271195e-05,
      "loss": 2.0103,
      "step": 3455
    },
    {
      "epoch": 0.6225624859265931,
      "grad_norm": 1.3870503902435303,
      "learning_rate": 7.801649472718652e-05,
      "loss": 1.8653,
      "step": 3456
    },
    {
      "epoch": 0.6227426255347894,
      "grad_norm": 1.4988107681274414,
      "learning_rate": 7.800476404419156e-05,
      "loss": 2.1285,
      "step": 3457
    },
    {
      "epoch": 0.6229227651429858,
      "grad_norm": 1.4061899185180664,
      "learning_rate": 7.799303111466806e-05,
      "loss": 1.7172,
      "step": 3458
    },
    {
      "epoch": 0.6231029047511821,
      "grad_norm": 1.9066200256347656,
      "learning_rate": 7.798129593955723e-05,
      "loss": 2.2153,
      "step": 3459
    },
    {
      "epoch": 0.6232830443593785,
      "grad_norm": 1.7115209102630615,
      "learning_rate": 7.796955851980048e-05,
      "loss": 2.2916,
      "step": 3460
    },
    {
      "epoch": 0.6234631839675748,
      "grad_norm": 1.2237215042114258,
      "learning_rate": 7.795781885633936e-05,
      "loss": 1.7763,
      "step": 3461
    },
    {
      "epoch": 0.6236433235757712,
      "grad_norm": 1.2384086847305298,
      "learning_rate": 7.794607695011562e-05,
      "loss": 1.4757,
      "step": 3462
    },
    {
      "epoch": 0.6238234631839675,
      "grad_norm": 1.2109719514846802,
      "learning_rate": 7.793433280207122e-05,
      "loss": 1.5239,
      "step": 3463
    },
    {
      "epoch": 0.6240036027921639,
      "grad_norm": 1.2383770942687988,
      "learning_rate": 7.792258641314824e-05,
      "loss": 1.5054,
      "step": 3464
    },
    {
      "epoch": 0.6241837424003602,
      "grad_norm": 1.2524224519729614,
      "learning_rate": 7.791083778428898e-05,
      "loss": 1.5824,
      "step": 3465
    },
    {
      "epoch": 0.6243638820085566,
      "grad_norm": 1.2961297035217285,
      "learning_rate": 7.78990869164359e-05,
      "loss": 1.5858,
      "step": 3466
    },
    {
      "epoch": 0.6245440216167529,
      "grad_norm": 1.2047690153121948,
      "learning_rate": 7.788733381053166e-05,
      "loss": 1.6125,
      "step": 3467
    },
    {
      "epoch": 0.6247241612249493,
      "grad_norm": 1.3629313707351685,
      "learning_rate": 7.78755784675191e-05,
      "loss": 1.5742,
      "step": 3468
    },
    {
      "epoch": 0.6249043008331457,
      "grad_norm": 1.32099187374115,
      "learning_rate": 7.786382088834119e-05,
      "loss": 1.6325,
      "step": 3469
    },
    {
      "epoch": 0.625084440441342,
      "grad_norm": 1.3721431493759155,
      "learning_rate": 7.785206107394116e-05,
      "loss": 1.6982,
      "step": 3470
    },
    {
      "epoch": 0.6252645800495384,
      "grad_norm": 1.186661720275879,
      "learning_rate": 7.784029902526233e-05,
      "loss": 1.3455,
      "step": 3471
    },
    {
      "epoch": 0.6254447196577347,
      "grad_norm": 1.2110214233398438,
      "learning_rate": 7.78285347432483e-05,
      "loss": 1.385,
      "step": 3472
    },
    {
      "epoch": 0.6256248592659311,
      "grad_norm": 1.2676270008087158,
      "learning_rate": 7.781676822884277e-05,
      "loss": 1.507,
      "step": 3473
    },
    {
      "epoch": 0.6258049988741274,
      "grad_norm": 1.1804970502853394,
      "learning_rate": 7.780499948298962e-05,
      "loss": 1.47,
      "step": 3474
    },
    {
      "epoch": 0.6259851384823238,
      "grad_norm": 1.3457781076431274,
      "learning_rate": 7.779322850663296e-05,
      "loss": 1.8081,
      "step": 3475
    },
    {
      "epoch": 0.6261652780905201,
      "grad_norm": 1.367093801498413,
      "learning_rate": 7.778145530071706e-05,
      "loss": 1.6013,
      "step": 3476
    },
    {
      "epoch": 0.6263454176987165,
      "grad_norm": 1.2647483348846436,
      "learning_rate": 7.776967986618633e-05,
      "loss": 1.7504,
      "step": 3477
    },
    {
      "epoch": 0.6265255573069128,
      "grad_norm": 1.2595716714859009,
      "learning_rate": 7.775790220398542e-05,
      "loss": 1.6164,
      "step": 3478
    },
    {
      "epoch": 0.6267056969151092,
      "grad_norm": 1.2980504035949707,
      "learning_rate": 7.77461223150591e-05,
      "loss": 1.6195,
      "step": 3479
    },
    {
      "epoch": 0.6268858365233055,
      "grad_norm": 1.312412977218628,
      "learning_rate": 7.773434020035238e-05,
      "loss": 1.7071,
      "step": 3480
    },
    {
      "epoch": 0.627065976131502,
      "grad_norm": 1.3567476272583008,
      "learning_rate": 7.77225558608104e-05,
      "loss": 1.3982,
      "step": 3481
    },
    {
      "epoch": 0.6272461157396982,
      "grad_norm": 1.2952780723571777,
      "learning_rate": 7.771076929737849e-05,
      "loss": 1.536,
      "step": 3482
    },
    {
      "epoch": 0.6274262553478946,
      "grad_norm": 1.4046076536178589,
      "learning_rate": 7.769898051100214e-05,
      "loss": 1.832,
      "step": 3483
    },
    {
      "epoch": 0.6276063949560909,
      "grad_norm": 1.4581197500228882,
      "learning_rate": 7.76871895026271e-05,
      "loss": 1.8457,
      "step": 3484
    },
    {
      "epoch": 0.6277865345642873,
      "grad_norm": 1.439780354499817,
      "learning_rate": 7.767539627319917e-05,
      "loss": 1.7358,
      "step": 3485
    },
    {
      "epoch": 0.6279666741724836,
      "grad_norm": 1.2732691764831543,
      "learning_rate": 7.766360082366447e-05,
      "loss": 1.6066,
      "step": 3486
    },
    {
      "epoch": 0.62814681378068,
      "grad_norm": 1.406550407409668,
      "learning_rate": 7.765180315496916e-05,
      "loss": 1.8004,
      "step": 3487
    },
    {
      "epoch": 0.6283269533888763,
      "grad_norm": 1.4022067785263062,
      "learning_rate": 7.764000326805967e-05,
      "loss": 1.6685,
      "step": 3488
    },
    {
      "epoch": 0.6285070929970727,
      "grad_norm": 1.453587532043457,
      "learning_rate": 7.76282011638826e-05,
      "loss": 1.6558,
      "step": 3489
    },
    {
      "epoch": 0.628687232605269,
      "grad_norm": 1.4871257543563843,
      "learning_rate": 7.761639684338466e-05,
      "loss": 1.5302,
      "step": 3490
    },
    {
      "epoch": 0.6288673722134654,
      "grad_norm": 1.3871734142303467,
      "learning_rate": 7.760459030751284e-05,
      "loss": 1.6276,
      "step": 3491
    },
    {
      "epoch": 0.6290475118216617,
      "grad_norm": 1.40336275100708,
      "learning_rate": 7.759278155721422e-05,
      "loss": 1.727,
      "step": 3492
    },
    {
      "epoch": 0.6292276514298581,
      "grad_norm": 1.3767496347427368,
      "learning_rate": 7.758097059343611e-05,
      "loss": 1.7256,
      "step": 3493
    },
    {
      "epoch": 0.6294077910380544,
      "grad_norm": 1.3373011350631714,
      "learning_rate": 7.756915741712597e-05,
      "loss": 1.5523,
      "step": 3494
    },
    {
      "epoch": 0.6295879306462508,
      "grad_norm": 1.527956247329712,
      "learning_rate": 7.755734202923145e-05,
      "loss": 1.8352,
      "step": 3495
    },
    {
      "epoch": 0.6297680702544471,
      "grad_norm": 1.320296287536621,
      "learning_rate": 7.754552443070038e-05,
      "loss": 1.3598,
      "step": 3496
    },
    {
      "epoch": 0.6299482098626435,
      "grad_norm": 1.4245721101760864,
      "learning_rate": 7.753370462248076e-05,
      "loss": 1.6712,
      "step": 3497
    },
    {
      "epoch": 0.63012834947084,
      "grad_norm": 1.3571979999542236,
      "learning_rate": 7.752188260552075e-05,
      "loss": 1.4392,
      "step": 3498
    },
    {
      "epoch": 0.6303084890790362,
      "grad_norm": 1.582578182220459,
      "learning_rate": 7.751005838076872e-05,
      "loss": 1.7333,
      "step": 3499
    },
    {
      "epoch": 0.6304886286872327,
      "grad_norm": 1.4914798736572266,
      "learning_rate": 7.749823194917323e-05,
      "loss": 1.4464,
      "step": 3500
    },
    {
      "epoch": 0.630668768295429,
      "grad_norm": 1.2188926935195923,
      "learning_rate": 7.748640331168296e-05,
      "loss": 1.9885,
      "step": 3501
    },
    {
      "epoch": 0.6308489079036254,
      "grad_norm": 1.196022629737854,
      "learning_rate": 7.74745724692468e-05,
      "loss": 1.9895,
      "step": 3502
    },
    {
      "epoch": 0.6310290475118217,
      "grad_norm": 1.2456132173538208,
      "learning_rate": 7.746273942281383e-05,
      "loss": 1.9596,
      "step": 3503
    },
    {
      "epoch": 0.6312091871200181,
      "grad_norm": 1.303802728652954,
      "learning_rate": 7.745090417333325e-05,
      "loss": 2.1779,
      "step": 3504
    },
    {
      "epoch": 0.6313893267282144,
      "grad_norm": 1.2232509851455688,
      "learning_rate": 7.743906672175453e-05,
      "loss": 1.765,
      "step": 3505
    },
    {
      "epoch": 0.6315694663364108,
      "grad_norm": 1.3437637090682983,
      "learning_rate": 7.742722706902724e-05,
      "loss": 2.0879,
      "step": 3506
    },
    {
      "epoch": 0.631749605944607,
      "grad_norm": 1.3863576650619507,
      "learning_rate": 7.741538521610115e-05,
      "loss": 1.6599,
      "step": 3507
    },
    {
      "epoch": 0.6319297455528035,
      "grad_norm": 1.4730451107025146,
      "learning_rate": 7.740354116392622e-05,
      "loss": 1.9449,
      "step": 3508
    },
    {
      "epoch": 0.6321098851609998,
      "grad_norm": 1.678127408027649,
      "learning_rate": 7.739169491345256e-05,
      "loss": 2.3293,
      "step": 3509
    },
    {
      "epoch": 0.6322900247691962,
      "grad_norm": 1.487926721572876,
      "learning_rate": 7.737984646563048e-05,
      "loss": 1.6924,
      "step": 3510
    },
    {
      "epoch": 0.6324701643773925,
      "grad_norm": 1.4182573556900024,
      "learning_rate": 7.736799582141046e-05,
      "loss": 2.035,
      "step": 3511
    },
    {
      "epoch": 0.6326503039855889,
      "grad_norm": 1.2465749979019165,
      "learning_rate": 7.735614298174313e-05,
      "loss": 1.8594,
      "step": 3512
    },
    {
      "epoch": 0.6328304435937852,
      "grad_norm": 1.2813730239868164,
      "learning_rate": 7.734428794757932e-05,
      "loss": 1.3947,
      "step": 3513
    },
    {
      "epoch": 0.6330105832019816,
      "grad_norm": 1.1685377359390259,
      "learning_rate": 7.733243071987009e-05,
      "loss": 1.3803,
      "step": 3514
    },
    {
      "epoch": 0.6331907228101779,
      "grad_norm": 1.2509543895721436,
      "learning_rate": 7.732057129956656e-05,
      "loss": 1.616,
      "step": 3515
    },
    {
      "epoch": 0.6333708624183743,
      "grad_norm": 1.3205718994140625,
      "learning_rate": 7.730870968762011e-05,
      "loss": 1.7883,
      "step": 3516
    },
    {
      "epoch": 0.6335510020265706,
      "grad_norm": 1.3118411302566528,
      "learning_rate": 7.729684588498226e-05,
      "loss": 1.6799,
      "step": 3517
    },
    {
      "epoch": 0.633731141634767,
      "grad_norm": 1.2695672512054443,
      "learning_rate": 7.728497989260473e-05,
      "loss": 1.4795,
      "step": 3518
    },
    {
      "epoch": 0.6339112812429633,
      "grad_norm": 1.3269842863082886,
      "learning_rate": 7.727311171143943e-05,
      "loss": 1.6241,
      "step": 3519
    },
    {
      "epoch": 0.6340914208511597,
      "grad_norm": 1.3300689458847046,
      "learning_rate": 7.726124134243836e-05,
      "loss": 1.4836,
      "step": 3520
    },
    {
      "epoch": 0.634271560459356,
      "grad_norm": 1.1789056062698364,
      "learning_rate": 7.72493687865538e-05,
      "loss": 1.6328,
      "step": 3521
    },
    {
      "epoch": 0.6344517000675524,
      "grad_norm": 1.3083393573760986,
      "learning_rate": 7.723749404473816e-05,
      "loss": 1.5419,
      "step": 3522
    },
    {
      "epoch": 0.6346318396757487,
      "grad_norm": 1.500576376914978,
      "learning_rate": 7.7225617117944e-05,
      "loss": 1.8592,
      "step": 3523
    },
    {
      "epoch": 0.6348119792839451,
      "grad_norm": 1.169145941734314,
      "learning_rate": 7.72137380071241e-05,
      "loss": 1.5003,
      "step": 3524
    },
    {
      "epoch": 0.6349921188921414,
      "grad_norm": 1.3244119882583618,
      "learning_rate": 7.720185671323138e-05,
      "loss": 1.8247,
      "step": 3525
    },
    {
      "epoch": 0.6351722585003378,
      "grad_norm": 1.2466422319412231,
      "learning_rate": 7.718997323721899e-05,
      "loss": 1.5434,
      "step": 3526
    },
    {
      "epoch": 0.6353523981085342,
      "grad_norm": 1.283230185508728,
      "learning_rate": 7.717808758004018e-05,
      "loss": 1.4864,
      "step": 3527
    },
    {
      "epoch": 0.6355325377167305,
      "grad_norm": 1.433368444442749,
      "learning_rate": 7.716619974264843e-05,
      "loss": 1.6029,
      "step": 3528
    },
    {
      "epoch": 0.6357126773249269,
      "grad_norm": 1.4116008281707764,
      "learning_rate": 7.715430972599736e-05,
      "loss": 1.8188,
      "step": 3529
    },
    {
      "epoch": 0.6358928169331232,
      "grad_norm": 1.2602133750915527,
      "learning_rate": 7.714241753104079e-05,
      "loss": 1.6417,
      "step": 3530
    },
    {
      "epoch": 0.6360729565413196,
      "grad_norm": 1.436623215675354,
      "learning_rate": 7.713052315873269e-05,
      "loss": 1.5603,
      "step": 3531
    },
    {
      "epoch": 0.6362530961495159,
      "grad_norm": 1.2853972911834717,
      "learning_rate": 7.711862661002727e-05,
      "loss": 1.7582,
      "step": 3532
    },
    {
      "epoch": 0.6364332357577123,
      "grad_norm": 1.3522992134094238,
      "learning_rate": 7.71067278858788e-05,
      "loss": 1.6433,
      "step": 3533
    },
    {
      "epoch": 0.6366133753659086,
      "grad_norm": 1.2607392072677612,
      "learning_rate": 7.709482698724185e-05,
      "loss": 1.459,
      "step": 3534
    },
    {
      "epoch": 0.636793514974105,
      "grad_norm": 1.2452760934829712,
      "learning_rate": 7.708292391507105e-05,
      "loss": 1.4935,
      "step": 3535
    },
    {
      "epoch": 0.6369736545823013,
      "grad_norm": 1.3124128580093384,
      "learning_rate": 7.707101867032129e-05,
      "loss": 1.5672,
      "step": 3536
    },
    {
      "epoch": 0.6371537941904977,
      "grad_norm": 1.2509071826934814,
      "learning_rate": 7.705911125394762e-05,
      "loss": 1.5853,
      "step": 3537
    },
    {
      "epoch": 0.637333933798694,
      "grad_norm": 1.3260079622268677,
      "learning_rate": 7.70472016669052e-05,
      "loss": 1.6166,
      "step": 3538
    },
    {
      "epoch": 0.6375140734068904,
      "grad_norm": 1.3136029243469238,
      "learning_rate": 7.703528991014946e-05,
      "loss": 1.507,
      "step": 3539
    },
    {
      "epoch": 0.6376942130150867,
      "grad_norm": 1.2431551218032837,
      "learning_rate": 7.702337598463592e-05,
      "loss": 1.5781,
      "step": 3540
    },
    {
      "epoch": 0.6378743526232831,
      "grad_norm": 1.3854396343231201,
      "learning_rate": 7.701145989132032e-05,
      "loss": 1.5438,
      "step": 3541
    },
    {
      "epoch": 0.6380544922314794,
      "grad_norm": 1.5077077150344849,
      "learning_rate": 7.699954163115857e-05,
      "loss": 1.8064,
      "step": 3542
    },
    {
      "epoch": 0.6382346318396758,
      "grad_norm": 1.3464741706848145,
      "learning_rate": 7.698762120510674e-05,
      "loss": 1.7212,
      "step": 3543
    },
    {
      "epoch": 0.6384147714478721,
      "grad_norm": 1.5715100765228271,
      "learning_rate": 7.697569861412106e-05,
      "loss": 1.6226,
      "step": 3544
    },
    {
      "epoch": 0.6385949110560685,
      "grad_norm": 1.283296823501587,
      "learning_rate": 7.6963773859158e-05,
      "loss": 1.5795,
      "step": 3545
    },
    {
      "epoch": 0.6387750506642648,
      "grad_norm": 1.3279004096984863,
      "learning_rate": 7.695184694117414e-05,
      "loss": 1.4647,
      "step": 3546
    },
    {
      "epoch": 0.6389551902724612,
      "grad_norm": 1.182403564453125,
      "learning_rate": 7.693991786112625e-05,
      "loss": 1.2688,
      "step": 3547
    },
    {
      "epoch": 0.6391353298806575,
      "grad_norm": 1.2587999105453491,
      "learning_rate": 7.692798661997123e-05,
      "loss": 1.4922,
      "step": 3548
    },
    {
      "epoch": 0.6393154694888539,
      "grad_norm": 1.4625664949417114,
      "learning_rate": 7.691605321866628e-05,
      "loss": 1.6578,
      "step": 3549
    },
    {
      "epoch": 0.6394956090970502,
      "grad_norm": 1.4390089511871338,
      "learning_rate": 7.690411765816864e-05,
      "loss": 1.4838,
      "step": 3550
    },
    {
      "epoch": 0.6396757487052466,
      "grad_norm": 1.2063490152359009,
      "learning_rate": 7.689217993943578e-05,
      "loss": 1.8046,
      "step": 3551
    },
    {
      "epoch": 0.6398558883134429,
      "grad_norm": 1.2386894226074219,
      "learning_rate": 7.688024006342535e-05,
      "loss": 2.1889,
      "step": 3552
    },
    {
      "epoch": 0.6400360279216393,
      "grad_norm": 1.2151871919631958,
      "learning_rate": 7.686829803109516e-05,
      "loss": 2.2384,
      "step": 3553
    },
    {
      "epoch": 0.6402161675298357,
      "grad_norm": 1.3600953817367554,
      "learning_rate": 7.685635384340317e-05,
      "loss": 2.2246,
      "step": 3554
    },
    {
      "epoch": 0.640396307138032,
      "grad_norm": 1.4402754306793213,
      "learning_rate": 7.684440750130758e-05,
      "loss": 1.8696,
      "step": 3555
    },
    {
      "epoch": 0.6405764467462284,
      "grad_norm": 1.2156671285629272,
      "learning_rate": 7.683245900576668e-05,
      "loss": 1.8921,
      "step": 3556
    },
    {
      "epoch": 0.6407565863544247,
      "grad_norm": 1.6027262210845947,
      "learning_rate": 7.6820508357739e-05,
      "loss": 1.993,
      "step": 3557
    },
    {
      "epoch": 0.6409367259626211,
      "grad_norm": 1.6761640310287476,
      "learning_rate": 7.680855555818318e-05,
      "loss": 2.0825,
      "step": 3558
    },
    {
      "epoch": 0.6411168655708174,
      "grad_norm": 1.5012297630310059,
      "learning_rate": 7.679660060805811e-05,
      "loss": 1.7486,
      "step": 3559
    },
    {
      "epoch": 0.6412970051790138,
      "grad_norm": 1.969394564628601,
      "learning_rate": 7.678464350832279e-05,
      "loss": 2.1835,
      "step": 3560
    },
    {
      "epoch": 0.6414771447872101,
      "grad_norm": 1.3221994638442993,
      "learning_rate": 7.677268425993642e-05,
      "loss": 1.8194,
      "step": 3561
    },
    {
      "epoch": 0.6416572843954065,
      "grad_norm": 1.1828910112380981,
      "learning_rate": 7.676072286385835e-05,
      "loss": 1.4909,
      "step": 3562
    },
    {
      "epoch": 0.6418374240036028,
      "grad_norm": 1.3463126420974731,
      "learning_rate": 7.674875932104813e-05,
      "loss": 1.52,
      "step": 3563
    },
    {
      "epoch": 0.6420175636117992,
      "grad_norm": 1.3547868728637695,
      "learning_rate": 7.673679363246546e-05,
      "loss": 1.591,
      "step": 3564
    },
    {
      "epoch": 0.6421977032199955,
      "grad_norm": 1.189975380897522,
      "learning_rate": 7.672482579907023e-05,
      "loss": 1.4636,
      "step": 3565
    },
    {
      "epoch": 0.6423778428281919,
      "grad_norm": 1.3249268531799316,
      "learning_rate": 7.671285582182249e-05,
      "loss": 1.4856,
      "step": 3566
    },
    {
      "epoch": 0.6425579824363882,
      "grad_norm": 1.3356399536132812,
      "learning_rate": 7.670088370168246e-05,
      "loss": 1.5946,
      "step": 3567
    },
    {
      "epoch": 0.6427381220445846,
      "grad_norm": 1.4490073919296265,
      "learning_rate": 7.668890943961058e-05,
      "loss": 1.6814,
      "step": 3568
    },
    {
      "epoch": 0.6429182616527809,
      "grad_norm": 1.2150338888168335,
      "learning_rate": 7.667693303656734e-05,
      "loss": 1.5836,
      "step": 3569
    },
    {
      "epoch": 0.6430984012609773,
      "grad_norm": 1.2231847047805786,
      "learning_rate": 7.666495449351355e-05,
      "loss": 1.3912,
      "step": 3570
    },
    {
      "epoch": 0.6432785408691736,
      "grad_norm": 1.2491759061813354,
      "learning_rate": 7.66529738114101e-05,
      "loss": 1.618,
      "step": 3571
    },
    {
      "epoch": 0.64345868047737,
      "grad_norm": 1.2275420427322388,
      "learning_rate": 7.664099099121807e-05,
      "loss": 1.6126,
      "step": 3572
    },
    {
      "epoch": 0.6436388200855663,
      "grad_norm": 1.346206784248352,
      "learning_rate": 7.662900603389871e-05,
      "loss": 1.6733,
      "step": 3573
    },
    {
      "epoch": 0.6438189596937627,
      "grad_norm": 1.1287707090377808,
      "learning_rate": 7.661701894041346e-05,
      "loss": 1.4035,
      "step": 3574
    },
    {
      "epoch": 0.643999099301959,
      "grad_norm": 1.2175657749176025,
      "learning_rate": 7.660502971172391e-05,
      "loss": 1.3145,
      "step": 3575
    },
    {
      "epoch": 0.6441792389101554,
      "grad_norm": 1.3306307792663574,
      "learning_rate": 7.659303834879185e-05,
      "loss": 1.6304,
      "step": 3576
    },
    {
      "epoch": 0.6443593785183517,
      "grad_norm": 1.4126652479171753,
      "learning_rate": 7.658104485257922e-05,
      "loss": 1.8582,
      "step": 3577
    },
    {
      "epoch": 0.6445395181265481,
      "grad_norm": 1.4104547500610352,
      "learning_rate": 7.656904922404809e-05,
      "loss": 1.634,
      "step": 3578
    },
    {
      "epoch": 0.6447196577347444,
      "grad_norm": 1.3894020318984985,
      "learning_rate": 7.655705146416079e-05,
      "loss": 1.776,
      "step": 3579
    },
    {
      "epoch": 0.6448997973429408,
      "grad_norm": 1.3726348876953125,
      "learning_rate": 7.654505157387974e-05,
      "loss": 1.8422,
      "step": 3580
    },
    {
      "epoch": 0.6450799369511371,
      "grad_norm": 1.3133918046951294,
      "learning_rate": 7.653304955416758e-05,
      "loss": 1.5783,
      "step": 3581
    },
    {
      "epoch": 0.6452600765593335,
      "grad_norm": 1.3153074979782104,
      "learning_rate": 7.652104540598712e-05,
      "loss": 1.5851,
      "step": 3582
    },
    {
      "epoch": 0.6454402161675299,
      "grad_norm": 1.233579397201538,
      "learning_rate": 7.650903913030132e-05,
      "loss": 1.516,
      "step": 3583
    },
    {
      "epoch": 0.6456203557757262,
      "grad_norm": 1.2603578567504883,
      "learning_rate": 7.649703072807332e-05,
      "loss": 1.6104,
      "step": 3584
    },
    {
      "epoch": 0.6458004953839226,
      "grad_norm": 1.2925444841384888,
      "learning_rate": 7.64850202002664e-05,
      "loss": 1.3514,
      "step": 3585
    },
    {
      "epoch": 0.6459806349921189,
      "grad_norm": 1.2838647365570068,
      "learning_rate": 7.647300754784407e-05,
      "loss": 1.3989,
      "step": 3586
    },
    {
      "epoch": 0.6461607746003153,
      "grad_norm": 1.3049839735031128,
      "learning_rate": 7.646099277176997e-05,
      "loss": 1.5073,
      "step": 3587
    },
    {
      "epoch": 0.6463409142085116,
      "grad_norm": 1.2988070249557495,
      "learning_rate": 7.644897587300793e-05,
      "loss": 1.4433,
      "step": 3588
    },
    {
      "epoch": 0.646521053816708,
      "grad_norm": 1.3294895887374878,
      "learning_rate": 7.643695685252191e-05,
      "loss": 1.5193,
      "step": 3589
    },
    {
      "epoch": 0.6467011934249043,
      "grad_norm": 1.4931018352508545,
      "learning_rate": 7.64249357112761e-05,
      "loss": 1.5881,
      "step": 3590
    },
    {
      "epoch": 0.6468813330331007,
      "grad_norm": 1.4193544387817383,
      "learning_rate": 7.641291245023481e-05,
      "loss": 1.7283,
      "step": 3591
    },
    {
      "epoch": 0.647061472641297,
      "grad_norm": 1.3652536869049072,
      "learning_rate": 7.640088707036253e-05,
      "loss": 1.4064,
      "step": 3592
    },
    {
      "epoch": 0.6472416122494934,
      "grad_norm": 1.3515150547027588,
      "learning_rate": 7.638885957262398e-05,
      "loss": 1.5234,
      "step": 3593
    },
    {
      "epoch": 0.6474217518576897,
      "grad_norm": 1.448966145515442,
      "learning_rate": 7.637682995798396e-05,
      "loss": 1.5217,
      "step": 3594
    },
    {
      "epoch": 0.6476018914658861,
      "grad_norm": 1.4578299522399902,
      "learning_rate": 7.636479822740749e-05,
      "loss": 1.6968,
      "step": 3595
    },
    {
      "epoch": 0.6477820310740824,
      "grad_norm": 1.3756144046783447,
      "learning_rate": 7.635276438185975e-05,
      "loss": 1.4141,
      "step": 3596
    },
    {
      "epoch": 0.6479621706822788,
      "grad_norm": 1.4369256496429443,
      "learning_rate": 7.634072842230607e-05,
      "loss": 1.6725,
      "step": 3597
    },
    {
      "epoch": 0.6481423102904751,
      "grad_norm": 1.1872689723968506,
      "learning_rate": 7.6328690349712e-05,
      "loss": 1.2286,
      "step": 3598
    },
    {
      "epoch": 0.6483224498986715,
      "grad_norm": 1.4163745641708374,
      "learning_rate": 7.631665016504321e-05,
      "loss": 1.5682,
      "step": 3599
    },
    {
      "epoch": 0.6485025895068678,
      "grad_norm": 1.5728036165237427,
      "learning_rate": 7.630460786926556e-05,
      "loss": 1.7093,
      "step": 3600
    },
    {
      "epoch": 0.6486827291150642,
      "grad_norm": 1.23725163936615,
      "learning_rate": 7.629256346334508e-05,
      "loss": 2.2935,
      "step": 3601
    },
    {
      "epoch": 0.6488628687232605,
      "grad_norm": 1.1671984195709229,
      "learning_rate": 7.628051694824798e-05,
      "loss": 2.1548,
      "step": 3602
    },
    {
      "epoch": 0.6490430083314569,
      "grad_norm": 1.3207608461380005,
      "learning_rate": 7.62684683249406e-05,
      "loss": 2.1959,
      "step": 3603
    },
    {
      "epoch": 0.6492231479396532,
      "grad_norm": 1.455020785331726,
      "learning_rate": 7.625641759438948e-05,
      "loss": 1.8492,
      "step": 3604
    },
    {
      "epoch": 0.6494032875478496,
      "grad_norm": 1.3297969102859497,
      "learning_rate": 7.624436475756134e-05,
      "loss": 2.0218,
      "step": 3605
    },
    {
      "epoch": 0.6495834271560459,
      "grad_norm": 1.240293264389038,
      "learning_rate": 7.623230981542304e-05,
      "loss": 1.6975,
      "step": 3606
    },
    {
      "epoch": 0.6497635667642423,
      "grad_norm": 1.381180763244629,
      "learning_rate": 7.622025276894163e-05,
      "loss": 2.02,
      "step": 3607
    },
    {
      "epoch": 0.6499437063724386,
      "grad_norm": 1.4141510725021362,
      "learning_rate": 7.620819361908429e-05,
      "loss": 1.9788,
      "step": 3608
    },
    {
      "epoch": 0.650123845980635,
      "grad_norm": 1.3985809087753296,
      "learning_rate": 7.619613236681843e-05,
      "loss": 2.0135,
      "step": 3609
    },
    {
      "epoch": 0.6503039855888313,
      "grad_norm": 1.6117537021636963,
      "learning_rate": 7.618406901311162e-05,
      "loss": 2.1882,
      "step": 3610
    },
    {
      "epoch": 0.6504841251970277,
      "grad_norm": 1.2975481748580933,
      "learning_rate": 7.617200355893153e-05,
      "loss": 1.7086,
      "step": 3611
    },
    {
      "epoch": 0.6506642648052241,
      "grad_norm": 1.243620753288269,
      "learning_rate": 7.615993600524606e-05,
      "loss": 1.6497,
      "step": 3612
    },
    {
      "epoch": 0.6508444044134204,
      "grad_norm": 1.3089145421981812,
      "learning_rate": 7.614786635302326e-05,
      "loss": 1.6266,
      "step": 3613
    },
    {
      "epoch": 0.6510245440216168,
      "grad_norm": 1.2470273971557617,
      "learning_rate": 7.613579460323137e-05,
      "loss": 1.4206,
      "step": 3614
    },
    {
      "epoch": 0.6512046836298131,
      "grad_norm": 1.2418755292892456,
      "learning_rate": 7.612372075683875e-05,
      "loss": 1.6923,
      "step": 3615
    },
    {
      "epoch": 0.6513848232380095,
      "grad_norm": 1.3111896514892578,
      "learning_rate": 7.6111644814814e-05,
      "loss": 1.5096,
      "step": 3616
    },
    {
      "epoch": 0.6515649628462058,
      "grad_norm": 1.229321837425232,
      "learning_rate": 7.609956677812579e-05,
      "loss": 1.6228,
      "step": 3617
    },
    {
      "epoch": 0.6517451024544022,
      "grad_norm": 1.2228953838348389,
      "learning_rate": 7.608748664774308e-05,
      "loss": 1.5049,
      "step": 3618
    },
    {
      "epoch": 0.6519252420625985,
      "grad_norm": 1.3195760250091553,
      "learning_rate": 7.607540442463487e-05,
      "loss": 1.7859,
      "step": 3619
    },
    {
      "epoch": 0.6521053816707949,
      "grad_norm": 1.26304292678833,
      "learning_rate": 7.606332010977044e-05,
      "loss": 1.3972,
      "step": 3620
    },
    {
      "epoch": 0.6522855212789912,
      "grad_norm": 1.2393215894699097,
      "learning_rate": 7.605123370411915e-05,
      "loss": 1.5209,
      "step": 3621
    },
    {
      "epoch": 0.6524656608871876,
      "grad_norm": 1.2372890710830688,
      "learning_rate": 7.603914520865059e-05,
      "loss": 1.5373,
      "step": 3622
    },
    {
      "epoch": 0.6526458004953839,
      "grad_norm": 1.1615149974822998,
      "learning_rate": 7.602705462433448e-05,
      "loss": 1.2234,
      "step": 3623
    },
    {
      "epoch": 0.6528259401035803,
      "grad_norm": 1.280493140220642,
      "learning_rate": 7.601496195214073e-05,
      "loss": 1.5755,
      "step": 3624
    },
    {
      "epoch": 0.6530060797117766,
      "grad_norm": 1.3271478414535522,
      "learning_rate": 7.600286719303939e-05,
      "loss": 1.551,
      "step": 3625
    },
    {
      "epoch": 0.653186219319973,
      "grad_norm": 1.2108567953109741,
      "learning_rate": 7.599077034800073e-05,
      "loss": 1.5421,
      "step": 3626
    },
    {
      "epoch": 0.6533663589281693,
      "grad_norm": 1.4844692945480347,
      "learning_rate": 7.597867141799512e-05,
      "loss": 1.996,
      "step": 3627
    },
    {
      "epoch": 0.6535464985363657,
      "grad_norm": 1.3217929601669312,
      "learning_rate": 7.596657040399315e-05,
      "loss": 1.7011,
      "step": 3628
    },
    {
      "epoch": 0.653726638144562,
      "grad_norm": 1.2655220031738281,
      "learning_rate": 7.595446730696554e-05,
      "loss": 1.4463,
      "step": 3629
    },
    {
      "epoch": 0.6539067777527584,
      "grad_norm": 1.3344367742538452,
      "learning_rate": 7.594236212788321e-05,
      "loss": 1.823,
      "step": 3630
    },
    {
      "epoch": 0.6540869173609547,
      "grad_norm": 1.3263784646987915,
      "learning_rate": 7.593025486771722e-05,
      "loss": 1.5615,
      "step": 3631
    },
    {
      "epoch": 0.6542670569691511,
      "grad_norm": 1.420602798461914,
      "learning_rate": 7.591814552743883e-05,
      "loss": 1.9178,
      "step": 3632
    },
    {
      "epoch": 0.6544471965773474,
      "grad_norm": 1.235182285308838,
      "learning_rate": 7.590603410801942e-05,
      "loss": 1.5472,
      "step": 3633
    },
    {
      "epoch": 0.6546273361855438,
      "grad_norm": 1.3650022745132446,
      "learning_rate": 7.589392061043057e-05,
      "loss": 1.5285,
      "step": 3634
    },
    {
      "epoch": 0.6548074757937401,
      "grad_norm": 1.3641724586486816,
      "learning_rate": 7.588180503564401e-05,
      "loss": 1.7199,
      "step": 3635
    },
    {
      "epoch": 0.6549876154019365,
      "grad_norm": 1.3221230506896973,
      "learning_rate": 7.586968738463168e-05,
      "loss": 1.6271,
      "step": 3636
    },
    {
      "epoch": 0.6551677550101328,
      "grad_norm": 1.2659703493118286,
      "learning_rate": 7.585756765836563e-05,
      "loss": 1.5409,
      "step": 3637
    },
    {
      "epoch": 0.6553478946183292,
      "grad_norm": 1.246110200881958,
      "learning_rate": 7.584544585781808e-05,
      "loss": 1.6047,
      "step": 3638
    },
    {
      "epoch": 0.6555280342265255,
      "grad_norm": 1.3502000570297241,
      "learning_rate": 7.583332198396144e-05,
      "loss": 1.6643,
      "step": 3639
    },
    {
      "epoch": 0.6557081738347219,
      "grad_norm": 1.3185523748397827,
      "learning_rate": 7.582119603776833e-05,
      "loss": 1.5902,
      "step": 3640
    },
    {
      "epoch": 0.6558883134429183,
      "grad_norm": 1.426889181137085,
      "learning_rate": 7.580906802021142e-05,
      "loss": 1.6414,
      "step": 3641
    },
    {
      "epoch": 0.6560684530511146,
      "grad_norm": 1.2787363529205322,
      "learning_rate": 7.579693793226367e-05,
      "loss": 1.5154,
      "step": 3642
    },
    {
      "epoch": 0.656248592659311,
      "grad_norm": 1.4704679250717163,
      "learning_rate": 7.57848057748981e-05,
      "loss": 1.7166,
      "step": 3643
    },
    {
      "epoch": 0.6564287322675073,
      "grad_norm": 1.2037187814712524,
      "learning_rate": 7.577267154908798e-05,
      "loss": 1.3593,
      "step": 3644
    },
    {
      "epoch": 0.6566088718757037,
      "grad_norm": 1.4547653198242188,
      "learning_rate": 7.576053525580671e-05,
      "loss": 1.6249,
      "step": 3645
    },
    {
      "epoch": 0.6567890114839,
      "grad_norm": 1.2622307538986206,
      "learning_rate": 7.574839689602784e-05,
      "loss": 1.4856,
      "step": 3646
    },
    {
      "epoch": 0.6569691510920964,
      "grad_norm": 1.4857465028762817,
      "learning_rate": 7.57362564707251e-05,
      "loss": 1.7151,
      "step": 3647
    },
    {
      "epoch": 0.6571492907002927,
      "grad_norm": 1.4397097826004028,
      "learning_rate": 7.572411398087244e-05,
      "loss": 1.6776,
      "step": 3648
    },
    {
      "epoch": 0.6573294303084891,
      "grad_norm": 1.4924947023391724,
      "learning_rate": 7.571196942744386e-05,
      "loss": 1.4068,
      "step": 3649
    },
    {
      "epoch": 0.6575095699166854,
      "grad_norm": 1.354737639427185,
      "learning_rate": 7.569982281141362e-05,
      "loss": 1.6787,
      "step": 3650
    },
    {
      "epoch": 0.6576897095248818,
      "grad_norm": 1.1923655271530151,
      "learning_rate": 7.56876741337561e-05,
      "loss": 1.8675,
      "step": 3651
    },
    {
      "epoch": 0.6578698491330781,
      "grad_norm": 2.02211332321167,
      "learning_rate": 7.567552339544589e-05,
      "loss": 2.2728,
      "step": 3652
    },
    {
      "epoch": 0.6580499887412745,
      "grad_norm": 1.2227410078048706,
      "learning_rate": 7.56633705974577e-05,
      "loss": 1.8893,
      "step": 3653
    },
    {
      "epoch": 0.6582301283494708,
      "grad_norm": 1.4408692121505737,
      "learning_rate": 7.565121574076641e-05,
      "loss": 1.9994,
      "step": 3654
    },
    {
      "epoch": 0.6584102679576672,
      "grad_norm": 1.3394439220428467,
      "learning_rate": 7.563905882634709e-05,
      "loss": 1.9256,
      "step": 3655
    },
    {
      "epoch": 0.6585904075658635,
      "grad_norm": 1.3568943738937378,
      "learning_rate": 7.562689985517496e-05,
      "loss": 1.9424,
      "step": 3656
    },
    {
      "epoch": 0.6587705471740599,
      "grad_norm": 1.3870124816894531,
      "learning_rate": 7.56147388282254e-05,
      "loss": 1.8813,
      "step": 3657
    },
    {
      "epoch": 0.6589506867822562,
      "grad_norm": 1.4717187881469727,
      "learning_rate": 7.560257574647399e-05,
      "loss": 2.007,
      "step": 3658
    },
    {
      "epoch": 0.6591308263904526,
      "grad_norm": 1.610626220703125,
      "learning_rate": 7.55904106108964e-05,
      "loss": 2.2868,
      "step": 3659
    },
    {
      "epoch": 0.6593109659986489,
      "grad_norm": 1.8786755800247192,
      "learning_rate": 7.557824342246856e-05,
      "loss": 2.6466,
      "step": 3660
    },
    {
      "epoch": 0.6594911056068453,
      "grad_norm": 1.259275197982788,
      "learning_rate": 7.556607418216646e-05,
      "loss": 1.6087,
      "step": 3661
    },
    {
      "epoch": 0.6596712452150416,
      "grad_norm": 1.2575100660324097,
      "learning_rate": 7.555390289096635e-05,
      "loss": 1.5655,
      "step": 3662
    },
    {
      "epoch": 0.659851384823238,
      "grad_norm": 1.3455612659454346,
      "learning_rate": 7.554172954984462e-05,
      "loss": 1.7662,
      "step": 3663
    },
    {
      "epoch": 0.6600315244314343,
      "grad_norm": 1.0780529975891113,
      "learning_rate": 7.552955415977776e-05,
      "loss": 1.3136,
      "step": 3664
    },
    {
      "epoch": 0.6602116640396307,
      "grad_norm": 1.188529372215271,
      "learning_rate": 7.551737672174252e-05,
      "loss": 1.3991,
      "step": 3665
    },
    {
      "epoch": 0.660391803647827,
      "grad_norm": 1.2434971332550049,
      "learning_rate": 7.550519723671575e-05,
      "loss": 1.5803,
      "step": 3666
    },
    {
      "epoch": 0.6605719432560234,
      "grad_norm": 1.098416805267334,
      "learning_rate": 7.549301570567446e-05,
      "loss": 1.2345,
      "step": 3667
    },
    {
      "epoch": 0.6607520828642198,
      "grad_norm": 1.3781863451004028,
      "learning_rate": 7.548083212959588e-05,
      "loss": 1.5225,
      "step": 3668
    },
    {
      "epoch": 0.6609322224724161,
      "grad_norm": 1.3204600811004639,
      "learning_rate": 7.546864650945735e-05,
      "loss": 1.6693,
      "step": 3669
    },
    {
      "epoch": 0.6611123620806125,
      "grad_norm": 1.3403867483139038,
      "learning_rate": 7.545645884623643e-05,
      "loss": 1.5242,
      "step": 3670
    },
    {
      "epoch": 0.6612925016888088,
      "grad_norm": 1.3336831331253052,
      "learning_rate": 7.544426914091073e-05,
      "loss": 1.6023,
      "step": 3671
    },
    {
      "epoch": 0.6614726412970052,
      "grad_norm": 1.2926417589187622,
      "learning_rate": 7.543207739445818e-05,
      "loss": 1.5581,
      "step": 3672
    },
    {
      "epoch": 0.6616527809052015,
      "grad_norm": 1.2402985095977783,
      "learning_rate": 7.541988360785678e-05,
      "loss": 1.5536,
      "step": 3673
    },
    {
      "epoch": 0.6618329205133979,
      "grad_norm": 1.277171015739441,
      "learning_rate": 7.54076877820847e-05,
      "loss": 1.6265,
      "step": 3674
    },
    {
      "epoch": 0.6620130601215942,
      "grad_norm": 1.300055742263794,
      "learning_rate": 7.539548991812027e-05,
      "loss": 1.6525,
      "step": 3675
    },
    {
      "epoch": 0.6621931997297906,
      "grad_norm": 1.2308604717254639,
      "learning_rate": 7.5383290016942e-05,
      "loss": 1.6456,
      "step": 3676
    },
    {
      "epoch": 0.6623733393379869,
      "grad_norm": 1.2776856422424316,
      "learning_rate": 7.537108807952858e-05,
      "loss": 1.5924,
      "step": 3677
    },
    {
      "epoch": 0.6625534789461833,
      "grad_norm": 1.5007275342941284,
      "learning_rate": 7.535888410685883e-05,
      "loss": 1.8818,
      "step": 3678
    },
    {
      "epoch": 0.6627336185543796,
      "grad_norm": 1.3374208211898804,
      "learning_rate": 7.534667809991175e-05,
      "loss": 1.6561,
      "step": 3679
    },
    {
      "epoch": 0.662913758162576,
      "grad_norm": 1.2487257719039917,
      "learning_rate": 7.533447005966649e-05,
      "loss": 1.5636,
      "step": 3680
    },
    {
      "epoch": 0.6630938977707723,
      "grad_norm": 1.3374063968658447,
      "learning_rate": 7.532225998710241e-05,
      "loss": 1.6567,
      "step": 3681
    },
    {
      "epoch": 0.6632740373789687,
      "grad_norm": 1.2387478351593018,
      "learning_rate": 7.531004788319895e-05,
      "loss": 1.2708,
      "step": 3682
    },
    {
      "epoch": 0.663454176987165,
      "grad_norm": 1.387755274772644,
      "learning_rate": 7.529783374893576e-05,
      "loss": 1.6619,
      "step": 3683
    },
    {
      "epoch": 0.6636343165953614,
      "grad_norm": 1.277577519416809,
      "learning_rate": 7.528561758529269e-05,
      "loss": 1.7869,
      "step": 3684
    },
    {
      "epoch": 0.6638144562035577,
      "grad_norm": 1.2916918992996216,
      "learning_rate": 7.527339939324968e-05,
      "loss": 1.2151,
      "step": 3685
    },
    {
      "epoch": 0.6639945958117541,
      "grad_norm": 1.2440828084945679,
      "learning_rate": 7.52611791737869e-05,
      "loss": 1.4988,
      "step": 3686
    },
    {
      "epoch": 0.6641747354199504,
      "grad_norm": 1.3986138105392456,
      "learning_rate": 7.524895692788461e-05,
      "loss": 1.7273,
      "step": 3687
    },
    {
      "epoch": 0.6643548750281468,
      "grad_norm": 1.285841464996338,
      "learning_rate": 7.52367326565233e-05,
      "loss": 1.4586,
      "step": 3688
    },
    {
      "epoch": 0.6645350146363431,
      "grad_norm": 1.357838749885559,
      "learning_rate": 7.522450636068359e-05,
      "loss": 1.7751,
      "step": 3689
    },
    {
      "epoch": 0.6647151542445395,
      "grad_norm": 1.3919072151184082,
      "learning_rate": 7.521227804134626e-05,
      "loss": 1.3994,
      "step": 3690
    },
    {
      "epoch": 0.6648952938527358,
      "grad_norm": 1.3824599981307983,
      "learning_rate": 7.520004769949229e-05,
      "loss": 1.6173,
      "step": 3691
    },
    {
      "epoch": 0.6650754334609322,
      "grad_norm": 1.237833857536316,
      "learning_rate": 7.518781533610273e-05,
      "loss": 1.3284,
      "step": 3692
    },
    {
      "epoch": 0.6652555730691285,
      "grad_norm": 1.2499383687973022,
      "learning_rate": 7.517558095215894e-05,
      "loss": 1.4224,
      "step": 3693
    },
    {
      "epoch": 0.6654357126773249,
      "grad_norm": 1.377846598625183,
      "learning_rate": 7.516334454864226e-05,
      "loss": 1.7126,
      "step": 3694
    },
    {
      "epoch": 0.6656158522855212,
      "grad_norm": 1.4783602952957153,
      "learning_rate": 7.515110612653436e-05,
      "loss": 1.5385,
      "step": 3695
    },
    {
      "epoch": 0.6657959918937176,
      "grad_norm": 1.3596348762512207,
      "learning_rate": 7.513886568681699e-05,
      "loss": 1.4503,
      "step": 3696
    },
    {
      "epoch": 0.665976131501914,
      "grad_norm": 1.4587466716766357,
      "learning_rate": 7.512662323047204e-05,
      "loss": 1.4007,
      "step": 3697
    },
    {
      "epoch": 0.6661562711101103,
      "grad_norm": 1.4428508281707764,
      "learning_rate": 7.511437875848162e-05,
      "loss": 1.564,
      "step": 3698
    },
    {
      "epoch": 0.6663364107183067,
      "grad_norm": 1.2985502481460571,
      "learning_rate": 7.510213227182796e-05,
      "loss": 1.5423,
      "step": 3699
    },
    {
      "epoch": 0.666516550326503,
      "grad_norm": 1.4289222955703735,
      "learning_rate": 7.508988377149349e-05,
      "loss": 1.3352,
      "step": 3700
    },
    {
      "epoch": 0.6666966899346994,
      "grad_norm": 1.477471947669983,
      "learning_rate": 7.507763325846075e-05,
      "loss": 1.9711,
      "step": 3701
    },
    {
      "epoch": 0.6668768295428957,
      "grad_norm": 1.3349413871765137,
      "learning_rate": 7.50653807337125e-05,
      "loss": 2.1618,
      "step": 3702
    },
    {
      "epoch": 0.6670569691510921,
      "grad_norm": 1.2045387029647827,
      "learning_rate": 7.505312619823162e-05,
      "loss": 2.0315,
      "step": 3703
    },
    {
      "epoch": 0.6672371087592884,
      "grad_norm": 1.3169463872909546,
      "learning_rate": 7.504086965300113e-05,
      "loss": 1.9713,
      "step": 3704
    },
    {
      "epoch": 0.6674172483674848,
      "grad_norm": 1.3186359405517578,
      "learning_rate": 7.50286110990043e-05,
      "loss": 2.0726,
      "step": 3705
    },
    {
      "epoch": 0.6675973879756811,
      "grad_norm": 1.4061086177825928,
      "learning_rate": 7.501635053722447e-05,
      "loss": 1.8744,
      "step": 3706
    },
    {
      "epoch": 0.6677775275838775,
      "grad_norm": 1.3869818449020386,
      "learning_rate": 7.50040879686452e-05,
      "loss": 2.1019,
      "step": 3707
    },
    {
      "epoch": 0.6679576671920738,
      "grad_norm": 1.4747493267059326,
      "learning_rate": 7.499182339425014e-05,
      "loss": 2.0295,
      "step": 3708
    },
    {
      "epoch": 0.6681378068002702,
      "grad_norm": 1.6722146272659302,
      "learning_rate": 7.49795568150232e-05,
      "loss": 1.8864,
      "step": 3709
    },
    {
      "epoch": 0.6683179464084665,
      "grad_norm": 1.7746806144714355,
      "learning_rate": 7.496728823194839e-05,
      "loss": 2.143,
      "step": 3710
    },
    {
      "epoch": 0.6684980860166629,
      "grad_norm": 1.3278132677078247,
      "learning_rate": 7.495501764600985e-05,
      "loss": 1.9036,
      "step": 3711
    },
    {
      "epoch": 0.6686782256248592,
      "grad_norm": 1.2284557819366455,
      "learning_rate": 7.494274505819197e-05,
      "loss": 1.5252,
      "step": 3712
    },
    {
      "epoch": 0.6688583652330556,
      "grad_norm": 1.3370000123977661,
      "learning_rate": 7.493047046947923e-05,
      "loss": 1.6233,
      "step": 3713
    },
    {
      "epoch": 0.6690385048412519,
      "grad_norm": 1.2801949977874756,
      "learning_rate": 7.491819388085629e-05,
      "loss": 1.5844,
      "step": 3714
    },
    {
      "epoch": 0.6692186444494483,
      "grad_norm": 1.3016551733016968,
      "learning_rate": 7.490591529330797e-05,
      "loss": 1.4833,
      "step": 3715
    },
    {
      "epoch": 0.6693987840576446,
      "grad_norm": 1.253571629524231,
      "learning_rate": 7.489363470781926e-05,
      "loss": 1.5693,
      "step": 3716
    },
    {
      "epoch": 0.669578923665841,
      "grad_norm": 1.256401777267456,
      "learning_rate": 7.48813521253753e-05,
      "loss": 1.5879,
      "step": 3717
    },
    {
      "epoch": 0.6697590632740373,
      "grad_norm": 1.175365924835205,
      "learning_rate": 7.48690675469614e-05,
      "loss": 1.4255,
      "step": 3718
    },
    {
      "epoch": 0.6699392028822337,
      "grad_norm": 1.2581830024719238,
      "learning_rate": 7.4856780973563e-05,
      "loss": 1.446,
      "step": 3719
    },
    {
      "epoch": 0.67011934249043,
      "grad_norm": 1.3508281707763672,
      "learning_rate": 7.484449240616573e-05,
      "loss": 1.6363,
      "step": 3720
    },
    {
      "epoch": 0.6702994820986264,
      "grad_norm": 1.3509913682937622,
      "learning_rate": 7.48322018457554e-05,
      "loss": 1.6833,
      "step": 3721
    },
    {
      "epoch": 0.6704796217068227,
      "grad_norm": 1.2553876638412476,
      "learning_rate": 7.481990929331793e-05,
      "loss": 1.3355,
      "step": 3722
    },
    {
      "epoch": 0.6706597613150191,
      "grad_norm": 1.4013798236846924,
      "learning_rate": 7.480761474983943e-05,
      "loss": 1.6521,
      "step": 3723
    },
    {
      "epoch": 0.6708399009232154,
      "grad_norm": 1.413506031036377,
      "learning_rate": 7.479531821630615e-05,
      "loss": 1.7168,
      "step": 3724
    },
    {
      "epoch": 0.6710200405314118,
      "grad_norm": 1.5890861749649048,
      "learning_rate": 7.478301969370452e-05,
      "loss": 1.8377,
      "step": 3725
    },
    {
      "epoch": 0.6712001801396082,
      "grad_norm": 1.3385136127471924,
      "learning_rate": 7.477071918302112e-05,
      "loss": 1.6891,
      "step": 3726
    },
    {
      "epoch": 0.6713803197478045,
      "grad_norm": 1.281814455986023,
      "learning_rate": 7.475841668524268e-05,
      "loss": 1.5864,
      "step": 3727
    },
    {
      "epoch": 0.671560459356001,
      "grad_norm": 1.1952110528945923,
      "learning_rate": 7.474611220135614e-05,
      "loss": 1.2592,
      "step": 3728
    },
    {
      "epoch": 0.6717405989641972,
      "grad_norm": 1.3743549585342407,
      "learning_rate": 7.473380573234852e-05,
      "loss": 1.6243,
      "step": 3729
    },
    {
      "epoch": 0.6719207385723936,
      "grad_norm": 1.3610057830810547,
      "learning_rate": 7.472149727920705e-05,
      "loss": 1.6303,
      "step": 3730
    },
    {
      "epoch": 0.67210087818059,
      "grad_norm": 1.4856343269348145,
      "learning_rate": 7.47091868429191e-05,
      "loss": 1.7969,
      "step": 3731
    },
    {
      "epoch": 0.6722810177887864,
      "grad_norm": 1.369631052017212,
      "learning_rate": 7.469687442447223e-05,
      "loss": 1.7481,
      "step": 3732
    },
    {
      "epoch": 0.6724611573969826,
      "grad_norm": 1.2857675552368164,
      "learning_rate": 7.468456002485413e-05,
      "loss": 1.5861,
      "step": 3733
    },
    {
      "epoch": 0.672641297005179,
      "grad_norm": 1.4559111595153809,
      "learning_rate": 7.467224364505264e-05,
      "loss": 1.6241,
      "step": 3734
    },
    {
      "epoch": 0.6728214366133753,
      "grad_norm": 1.4399436712265015,
      "learning_rate": 7.465992528605579e-05,
      "loss": 1.6622,
      "step": 3735
    },
    {
      "epoch": 0.6730015762215718,
      "grad_norm": 1.5655626058578491,
      "learning_rate": 7.464760494885172e-05,
      "loss": 2.0022,
      "step": 3736
    },
    {
      "epoch": 0.673181715829768,
      "grad_norm": 1.2919420003890991,
      "learning_rate": 7.46352826344288e-05,
      "loss": 1.481,
      "step": 3737
    },
    {
      "epoch": 0.6733618554379645,
      "grad_norm": 1.2697975635528564,
      "learning_rate": 7.46229583437755e-05,
      "loss": 1.5485,
      "step": 3738
    },
    {
      "epoch": 0.6735419950461607,
      "grad_norm": 1.586836814880371,
      "learning_rate": 7.46106320778805e-05,
      "loss": 1.8011,
      "step": 3739
    },
    {
      "epoch": 0.6737221346543572,
      "grad_norm": 1.3497823476791382,
      "learning_rate": 7.459830383773258e-05,
      "loss": 1.5316,
      "step": 3740
    },
    {
      "epoch": 0.6739022742625534,
      "grad_norm": 1.3906464576721191,
      "learning_rate": 7.458597362432068e-05,
      "loss": 1.6649,
      "step": 3741
    },
    {
      "epoch": 0.6740824138707499,
      "grad_norm": 1.3394267559051514,
      "learning_rate": 7.457364143863397e-05,
      "loss": 1.4355,
      "step": 3742
    },
    {
      "epoch": 0.6742625534789461,
      "grad_norm": 1.230089545249939,
      "learning_rate": 7.456130728166171e-05,
      "loss": 1.462,
      "step": 3743
    },
    {
      "epoch": 0.6744426930871426,
      "grad_norm": 1.3259518146514893,
      "learning_rate": 7.454897115439333e-05,
      "loss": 1.5647,
      "step": 3744
    },
    {
      "epoch": 0.6746228326953388,
      "grad_norm": 1.3547018766403198,
      "learning_rate": 7.453663305781847e-05,
      "loss": 1.5953,
      "step": 3745
    },
    {
      "epoch": 0.6748029723035353,
      "grad_norm": 1.4750478267669678,
      "learning_rate": 7.452429299292684e-05,
      "loss": 1.5652,
      "step": 3746
    },
    {
      "epoch": 0.6749831119117315,
      "grad_norm": 1.5076578855514526,
      "learning_rate": 7.451195096070838e-05,
      "loss": 1.8026,
      "step": 3747
    },
    {
      "epoch": 0.675163251519928,
      "grad_norm": 1.308268666267395,
      "learning_rate": 7.449960696215312e-05,
      "loss": 1.5457,
      "step": 3748
    },
    {
      "epoch": 0.6753433911281242,
      "grad_norm": 1.465030312538147,
      "learning_rate": 7.448726099825137e-05,
      "loss": 1.8723,
      "step": 3749
    },
    {
      "epoch": 0.6755235307363207,
      "grad_norm": 1.153728723526001,
      "learning_rate": 7.447491306999344e-05,
      "loss": 1.2589,
      "step": 3750
    },
    {
      "epoch": 0.675703670344517,
      "grad_norm": 1.2619167566299438,
      "learning_rate": 7.446256317836991e-05,
      "loss": 2.0856,
      "step": 3751
    },
    {
      "epoch": 0.6758838099527134,
      "grad_norm": 1.1553776264190674,
      "learning_rate": 7.445021132437147e-05,
      "loss": 1.6955,
      "step": 3752
    },
    {
      "epoch": 0.6760639495609098,
      "grad_norm": 1.1766729354858398,
      "learning_rate": 7.4437857508989e-05,
      "loss": 2.0043,
      "step": 3753
    },
    {
      "epoch": 0.676244089169106,
      "grad_norm": 1.2143126726150513,
      "learning_rate": 7.442550173321351e-05,
      "loss": 1.8816,
      "step": 3754
    },
    {
      "epoch": 0.6764242287773025,
      "grad_norm": 1.2775821685791016,
      "learning_rate": 7.441314399803616e-05,
      "loss": 2.0566,
      "step": 3755
    },
    {
      "epoch": 0.6766043683854988,
      "grad_norm": 1.3005644083023071,
      "learning_rate": 7.44007843044483e-05,
      "loss": 1.6926,
      "step": 3756
    },
    {
      "epoch": 0.6767845079936952,
      "grad_norm": 1.4249347448349,
      "learning_rate": 7.438842265344139e-05,
      "loss": 2.0715,
      "step": 3757
    },
    {
      "epoch": 0.6769646476018915,
      "grad_norm": 1.5612343549728394,
      "learning_rate": 7.437605904600709e-05,
      "loss": 1.9976,
      "step": 3758
    },
    {
      "epoch": 0.6771447872100879,
      "grad_norm": 1.5529882907867432,
      "learning_rate": 7.436369348313721e-05,
      "loss": 2.0521,
      "step": 3759
    },
    {
      "epoch": 0.6773249268182842,
      "grad_norm": 1.8146275281906128,
      "learning_rate": 7.435132596582373e-05,
      "loss": 2.164,
      "step": 3760
    },
    {
      "epoch": 0.6775050664264806,
      "grad_norm": 1.321893334388733,
      "learning_rate": 7.433895649505873e-05,
      "loss": 1.5386,
      "step": 3761
    },
    {
      "epoch": 0.6776852060346769,
      "grad_norm": 1.2590678930282593,
      "learning_rate": 7.432658507183449e-05,
      "loss": 1.6997,
      "step": 3762
    },
    {
      "epoch": 0.6778653456428733,
      "grad_norm": 1.2367582321166992,
      "learning_rate": 7.431421169714344e-05,
      "loss": 1.5306,
      "step": 3763
    },
    {
      "epoch": 0.6780454852510696,
      "grad_norm": 1.3217625617980957,
      "learning_rate": 7.430183637197818e-05,
      "loss": 1.7033,
      "step": 3764
    },
    {
      "epoch": 0.678225624859266,
      "grad_norm": 1.1774741411209106,
      "learning_rate": 7.428945909733143e-05,
      "loss": 1.456,
      "step": 3765
    },
    {
      "epoch": 0.6784057644674623,
      "grad_norm": 1.3103166818618774,
      "learning_rate": 7.427707987419612e-05,
      "loss": 1.5354,
      "step": 3766
    },
    {
      "epoch": 0.6785859040756587,
      "grad_norm": 1.3158172369003296,
      "learning_rate": 7.426469870356528e-05,
      "loss": 1.6131,
      "step": 3767
    },
    {
      "epoch": 0.678766043683855,
      "grad_norm": 1.333551287651062,
      "learning_rate": 7.425231558643213e-05,
      "loss": 1.7648,
      "step": 3768
    },
    {
      "epoch": 0.6789461832920514,
      "grad_norm": 1.4817004203796387,
      "learning_rate": 7.423993052379003e-05,
      "loss": 1.4387,
      "step": 3769
    },
    {
      "epoch": 0.6791263229002477,
      "grad_norm": 1.3711568117141724,
      "learning_rate": 7.422754351663252e-05,
      "loss": 1.5995,
      "step": 3770
    },
    {
      "epoch": 0.6793064625084441,
      "grad_norm": 1.212541103363037,
      "learning_rate": 7.421515456595325e-05,
      "loss": 1.3797,
      "step": 3771
    },
    {
      "epoch": 0.6794866021166404,
      "grad_norm": 1.3111616373062134,
      "learning_rate": 7.42027636727461e-05,
      "loss": 1.7147,
      "step": 3772
    },
    {
      "epoch": 0.6796667417248368,
      "grad_norm": 1.2291555404663086,
      "learning_rate": 7.419037083800502e-05,
      "loss": 1.3334,
      "step": 3773
    },
    {
      "epoch": 0.6798468813330331,
      "grad_norm": 1.2658121585845947,
      "learning_rate": 7.417797606272418e-05,
      "loss": 1.4482,
      "step": 3774
    },
    {
      "epoch": 0.6800270209412295,
      "grad_norm": 1.3814055919647217,
      "learning_rate": 7.416557934789788e-05,
      "loss": 1.5043,
      "step": 3775
    },
    {
      "epoch": 0.6802071605494258,
      "grad_norm": 1.2001444101333618,
      "learning_rate": 7.415318069452058e-05,
      "loss": 1.4524,
      "step": 3776
    },
    {
      "epoch": 0.6803873001576222,
      "grad_norm": 1.3388934135437012,
      "learning_rate": 7.414078010358688e-05,
      "loss": 1.6208,
      "step": 3777
    },
    {
      "epoch": 0.6805674397658185,
      "grad_norm": 1.206924557685852,
      "learning_rate": 7.412837757609157e-05,
      "loss": 1.3059,
      "step": 3778
    },
    {
      "epoch": 0.6807475793740149,
      "grad_norm": 1.3403254747390747,
      "learning_rate": 7.411597311302955e-05,
      "loss": 1.4764,
      "step": 3779
    },
    {
      "epoch": 0.6809277189822112,
      "grad_norm": 1.3341001272201538,
      "learning_rate": 7.410356671539592e-05,
      "loss": 1.5923,
      "step": 3780
    },
    {
      "epoch": 0.6811078585904076,
      "grad_norm": 1.3420345783233643,
      "learning_rate": 7.40911583841859e-05,
      "loss": 1.5379,
      "step": 3781
    },
    {
      "epoch": 0.681287998198604,
      "grad_norm": 1.4320402145385742,
      "learning_rate": 7.407874812039489e-05,
      "loss": 1.7614,
      "step": 3782
    },
    {
      "epoch": 0.6814681378068003,
      "grad_norm": 1.3970222473144531,
      "learning_rate": 7.406633592501845e-05,
      "loss": 1.7229,
      "step": 3783
    },
    {
      "epoch": 0.6816482774149967,
      "grad_norm": 1.4104418754577637,
      "learning_rate": 7.405392179905225e-05,
      "loss": 1.7245,
      "step": 3784
    },
    {
      "epoch": 0.681828417023193,
      "grad_norm": 1.3820829391479492,
      "learning_rate": 7.404150574349216e-05,
      "loss": 1.7106,
      "step": 3785
    },
    {
      "epoch": 0.6820085566313894,
      "grad_norm": 1.3436079025268555,
      "learning_rate": 7.402908775933419e-05,
      "loss": 1.7077,
      "step": 3786
    },
    {
      "epoch": 0.6821886962395857,
      "grad_norm": 1.3855971097946167,
      "learning_rate": 7.40166678475745e-05,
      "loss": 1.7022,
      "step": 3787
    },
    {
      "epoch": 0.6823688358477821,
      "grad_norm": 1.3236973285675049,
      "learning_rate": 7.400424600920943e-05,
      "loss": 1.7128,
      "step": 3788
    },
    {
      "epoch": 0.6825489754559784,
      "grad_norm": 1.3929835557937622,
      "learning_rate": 7.399182224523543e-05,
      "loss": 1.8028,
      "step": 3789
    },
    {
      "epoch": 0.6827291150641748,
      "grad_norm": 1.3194581270217896,
      "learning_rate": 7.397939655664912e-05,
      "loss": 1.4563,
      "step": 3790
    },
    {
      "epoch": 0.6829092546723711,
      "grad_norm": 1.4082924127578735,
      "learning_rate": 7.396696894444732e-05,
      "loss": 1.4215,
      "step": 3791
    },
    {
      "epoch": 0.6830893942805675,
      "grad_norm": 1.3762539625167847,
      "learning_rate": 7.395453940962693e-05,
      "loss": 1.68,
      "step": 3792
    },
    {
      "epoch": 0.6832695338887638,
      "grad_norm": 1.3264034986495972,
      "learning_rate": 7.394210795318506e-05,
      "loss": 1.3545,
      "step": 3793
    },
    {
      "epoch": 0.6834496734969602,
      "grad_norm": 1.429064154624939,
      "learning_rate": 7.392967457611897e-05,
      "loss": 1.5334,
      "step": 3794
    },
    {
      "epoch": 0.6836298131051565,
      "grad_norm": 1.2954745292663574,
      "learning_rate": 7.391723927942602e-05,
      "loss": 1.4719,
      "step": 3795
    },
    {
      "epoch": 0.6838099527133529,
      "grad_norm": 1.4566138982772827,
      "learning_rate": 7.390480206410379e-05,
      "loss": 1.5146,
      "step": 3796
    },
    {
      "epoch": 0.6839900923215492,
      "grad_norm": 1.558975338935852,
      "learning_rate": 7.389236293114998e-05,
      "loss": 1.4991,
      "step": 3797
    },
    {
      "epoch": 0.6841702319297456,
      "grad_norm": 1.516238808631897,
      "learning_rate": 7.387992188156245e-05,
      "loss": 1.3562,
      "step": 3798
    },
    {
      "epoch": 0.6843503715379419,
      "grad_norm": 1.395850419998169,
      "learning_rate": 7.386747891633923e-05,
      "loss": 1.4575,
      "step": 3799
    },
    {
      "epoch": 0.6845305111461383,
      "grad_norm": 1.4928802251815796,
      "learning_rate": 7.385503403647847e-05,
      "loss": 1.7443,
      "step": 3800
    },
    {
      "epoch": 0.6847106507543346,
      "grad_norm": 1.6430546045303345,
      "learning_rate": 7.38425872429785e-05,
      "loss": 1.6772,
      "step": 3801
    },
    {
      "epoch": 0.684890790362531,
      "grad_norm": 1.5847761631011963,
      "learning_rate": 7.383013853683781e-05,
      "loss": 2.0852,
      "step": 3802
    },
    {
      "epoch": 0.6850709299707273,
      "grad_norm": 1.4791513681411743,
      "learning_rate": 7.381768791905499e-05,
      "loss": 2.1373,
      "step": 3803
    },
    {
      "epoch": 0.6852510695789237,
      "grad_norm": 1.3154712915420532,
      "learning_rate": 7.380523539062885e-05,
      "loss": 2.1147,
      "step": 3804
    },
    {
      "epoch": 0.68543120918712,
      "grad_norm": 1.2487972974777222,
      "learning_rate": 7.379278095255833e-05,
      "loss": 1.6921,
      "step": 3805
    },
    {
      "epoch": 0.6856113487953164,
      "grad_norm": 1.2755018472671509,
      "learning_rate": 7.37803246058425e-05,
      "loss": 1.6568,
      "step": 3806
    },
    {
      "epoch": 0.6857914884035127,
      "grad_norm": 1.4299025535583496,
      "learning_rate": 7.376786635148063e-05,
      "loss": 2.0746,
      "step": 3807
    },
    {
      "epoch": 0.6859716280117091,
      "grad_norm": 1.5345515012741089,
      "learning_rate": 7.375540619047207e-05,
      "loss": 2.4002,
      "step": 3808
    },
    {
      "epoch": 0.6861517676199054,
      "grad_norm": 1.5227692127227783,
      "learning_rate": 7.374294412381643e-05,
      "loss": 1.9818,
      "step": 3809
    },
    {
      "epoch": 0.6863319072281018,
      "grad_norm": 1.543299674987793,
      "learning_rate": 7.373048015251337e-05,
      "loss": 2.36,
      "step": 3810
    },
    {
      "epoch": 0.6865120468362982,
      "grad_norm": 2.098301649093628,
      "learning_rate": 7.371801427756274e-05,
      "loss": 1.8631,
      "step": 3811
    },
    {
      "epoch": 0.6866921864444945,
      "grad_norm": 1.3070650100708008,
      "learning_rate": 7.370554649996454e-05,
      "loss": 1.6812,
      "step": 3812
    },
    {
      "epoch": 0.6868723260526909,
      "grad_norm": 1.2857000827789307,
      "learning_rate": 7.369307682071898e-05,
      "loss": 1.7798,
      "step": 3813
    },
    {
      "epoch": 0.6870524656608872,
      "grad_norm": 1.3998967409133911,
      "learning_rate": 7.368060524082631e-05,
      "loss": 1.7286,
      "step": 3814
    },
    {
      "epoch": 0.6872326052690836,
      "grad_norm": 1.2834653854370117,
      "learning_rate": 7.366813176128705e-05,
      "loss": 1.6483,
      "step": 3815
    },
    {
      "epoch": 0.6874127448772799,
      "grad_norm": 1.3898488283157349,
      "learning_rate": 7.365565638310177e-05,
      "loss": 1.4193,
      "step": 3816
    },
    {
      "epoch": 0.6875928844854763,
      "grad_norm": 1.345453143119812,
      "learning_rate": 7.364317910727128e-05,
      "loss": 1.6951,
      "step": 3817
    },
    {
      "epoch": 0.6877730240936726,
      "grad_norm": 1.1374995708465576,
      "learning_rate": 7.363069993479644e-05,
      "loss": 1.3432,
      "step": 3818
    },
    {
      "epoch": 0.687953163701869,
      "grad_norm": 1.245538592338562,
      "learning_rate": 7.361821886667839e-05,
      "loss": 1.4805,
      "step": 3819
    },
    {
      "epoch": 0.6881333033100653,
      "grad_norm": 1.3383915424346924,
      "learning_rate": 7.360573590391831e-05,
      "loss": 1.6369,
      "step": 3820
    },
    {
      "epoch": 0.6883134429182617,
      "grad_norm": 1.427017092704773,
      "learning_rate": 7.359325104751762e-05,
      "loss": 1.5755,
      "step": 3821
    },
    {
      "epoch": 0.688493582526458,
      "grad_norm": 1.3233128786087036,
      "learning_rate": 7.358076429847779e-05,
      "loss": 1.4975,
      "step": 3822
    },
    {
      "epoch": 0.6886737221346544,
      "grad_norm": 1.3432520627975464,
      "learning_rate": 7.356827565780055e-05,
      "loss": 1.6049,
      "step": 3823
    },
    {
      "epoch": 0.6888538617428507,
      "grad_norm": 1.3402349948883057,
      "learning_rate": 7.355578512648772e-05,
      "loss": 1.7118,
      "step": 3824
    },
    {
      "epoch": 0.6890340013510471,
      "grad_norm": 1.343283772468567,
      "learning_rate": 7.354329270554127e-05,
      "loss": 1.5753,
      "step": 3825
    },
    {
      "epoch": 0.6892141409592434,
      "grad_norm": 1.2998533248901367,
      "learning_rate": 7.353079839596336e-05,
      "loss": 1.7031,
      "step": 3826
    },
    {
      "epoch": 0.6893942805674398,
      "grad_norm": 1.4211757183074951,
      "learning_rate": 7.351830219875625e-05,
      "loss": 1.7855,
      "step": 3827
    },
    {
      "epoch": 0.6895744201756361,
      "grad_norm": 1.2932559251785278,
      "learning_rate": 7.35058041149224e-05,
      "loss": 1.5938,
      "step": 3828
    },
    {
      "epoch": 0.6897545597838325,
      "grad_norm": 1.3493373394012451,
      "learning_rate": 7.34933041454644e-05,
      "loss": 1.7244,
      "step": 3829
    },
    {
      "epoch": 0.6899346993920288,
      "grad_norm": 1.2789973020553589,
      "learning_rate": 7.348080229138496e-05,
      "loss": 1.5126,
      "step": 3830
    },
    {
      "epoch": 0.6901148390002252,
      "grad_norm": 1.3149967193603516,
      "learning_rate": 7.346829855368702e-05,
      "loss": 1.628,
      "step": 3831
    },
    {
      "epoch": 0.6902949786084215,
      "grad_norm": 1.2477061748504639,
      "learning_rate": 7.34557929333736e-05,
      "loss": 1.6282,
      "step": 3832
    },
    {
      "epoch": 0.6904751182166179,
      "grad_norm": 1.2891724109649658,
      "learning_rate": 7.34432854314479e-05,
      "loss": 1.6348,
      "step": 3833
    },
    {
      "epoch": 0.6906552578248142,
      "grad_norm": 1.3005331754684448,
      "learning_rate": 7.343077604891323e-05,
      "loss": 1.5863,
      "step": 3834
    },
    {
      "epoch": 0.6908353974330106,
      "grad_norm": 1.3337929248809814,
      "learning_rate": 7.341826478677316e-05,
      "loss": 1.4212,
      "step": 3835
    },
    {
      "epoch": 0.6910155370412069,
      "grad_norm": 1.405542016029358,
      "learning_rate": 7.340575164603127e-05,
      "loss": 1.7336,
      "step": 3836
    },
    {
      "epoch": 0.6911956766494033,
      "grad_norm": 1.415505051612854,
      "learning_rate": 7.339323662769139e-05,
      "loss": 1.6209,
      "step": 3837
    },
    {
      "epoch": 0.6913758162575996,
      "grad_norm": 1.2408785820007324,
      "learning_rate": 7.338071973275746e-05,
      "loss": 1.349,
      "step": 3838
    },
    {
      "epoch": 0.691555955865796,
      "grad_norm": 1.3061391115188599,
      "learning_rate": 7.33682009622336e-05,
      "loss": 1.4186,
      "step": 3839
    },
    {
      "epoch": 0.6917360954739924,
      "grad_norm": 1.4277821779251099,
      "learning_rate": 7.335568031712401e-05,
      "loss": 1.5794,
      "step": 3840
    },
    {
      "epoch": 0.6919162350821887,
      "grad_norm": 1.5519523620605469,
      "learning_rate": 7.334315779843313e-05,
      "loss": 1.8449,
      "step": 3841
    },
    {
      "epoch": 0.6920963746903851,
      "grad_norm": 1.314356803894043,
      "learning_rate": 7.333063340716551e-05,
      "loss": 1.6588,
      "step": 3842
    },
    {
      "epoch": 0.6922765142985814,
      "grad_norm": 1.575269103050232,
      "learning_rate": 7.331810714432585e-05,
      "loss": 1.6854,
      "step": 3843
    },
    {
      "epoch": 0.6924566539067778,
      "grad_norm": 1.36978018283844,
      "learning_rate": 7.330557901091898e-05,
      "loss": 1.6053,
      "step": 3844
    },
    {
      "epoch": 0.6926367935149741,
      "grad_norm": 1.4974101781845093,
      "learning_rate": 7.329304900794991e-05,
      "loss": 1.6429,
      "step": 3845
    },
    {
      "epoch": 0.6928169331231705,
      "grad_norm": 1.35093092918396,
      "learning_rate": 7.32805171364238e-05,
      "loss": 1.6877,
      "step": 3846
    },
    {
      "epoch": 0.6929970727313668,
      "grad_norm": 1.3871434926986694,
      "learning_rate": 7.326798339734594e-05,
      "loss": 1.6827,
      "step": 3847
    },
    {
      "epoch": 0.6931772123395632,
      "grad_norm": 1.3733747005462646,
      "learning_rate": 7.32554477917218e-05,
      "loss": 1.4125,
      "step": 3848
    },
    {
      "epoch": 0.6933573519477595,
      "grad_norm": 1.4671399593353271,
      "learning_rate": 7.324291032055696e-05,
      "loss": 1.483,
      "step": 3849
    },
    {
      "epoch": 0.6935374915559559,
      "grad_norm": 1.3165206909179688,
      "learning_rate": 7.323037098485716e-05,
      "loss": 1.5879,
      "step": 3850
    },
    {
      "epoch": 0.6937176311641522,
      "grad_norm": 1.335375428199768,
      "learning_rate": 7.321782978562833e-05,
      "loss": 1.92,
      "step": 3851
    },
    {
      "epoch": 0.6938977707723486,
      "grad_norm": 1.2365530729293823,
      "learning_rate": 7.320528672387648e-05,
      "loss": 2.0427,
      "step": 3852
    },
    {
      "epoch": 0.6940779103805449,
      "grad_norm": 1.2713297605514526,
      "learning_rate": 7.319274180060786e-05,
      "loss": 1.9989,
      "step": 3853
    },
    {
      "epoch": 0.6942580499887413,
      "grad_norm": 1.3556358814239502,
      "learning_rate": 7.318019501682879e-05,
      "loss": 2.07,
      "step": 3854
    },
    {
      "epoch": 0.6944381895969376,
      "grad_norm": 1.236850380897522,
      "learning_rate": 7.316764637354575e-05,
      "loss": 1.8912,
      "step": 3855
    },
    {
      "epoch": 0.694618329205134,
      "grad_norm": 1.2358862161636353,
      "learning_rate": 7.315509587176541e-05,
      "loss": 1.9052,
      "step": 3856
    },
    {
      "epoch": 0.6947984688133303,
      "grad_norm": 1.446398377418518,
      "learning_rate": 7.314254351249454e-05,
      "loss": 2.0899,
      "step": 3857
    },
    {
      "epoch": 0.6949786084215267,
      "grad_norm": 1.5219694375991821,
      "learning_rate": 7.312998929674013e-05,
      "loss": 2.0933,
      "step": 3858
    },
    {
      "epoch": 0.695158748029723,
      "grad_norm": 1.619506597518921,
      "learning_rate": 7.311743322550923e-05,
      "loss": 2.1694,
      "step": 3859
    },
    {
      "epoch": 0.6953388876379194,
      "grad_norm": 1.593291997909546,
      "learning_rate": 7.310487529980911e-05,
      "loss": 1.5654,
      "step": 3860
    },
    {
      "epoch": 0.6955190272461157,
      "grad_norm": 1.2240742444992065,
      "learning_rate": 7.309231552064714e-05,
      "loss": 1.5291,
      "step": 3861
    },
    {
      "epoch": 0.6956991668543121,
      "grad_norm": 1.2461670637130737,
      "learning_rate": 7.307975388903086e-05,
      "loss": 1.505,
      "step": 3862
    },
    {
      "epoch": 0.6958793064625084,
      "grad_norm": 1.1853049993515015,
      "learning_rate": 7.306719040596797e-05,
      "loss": 1.5592,
      "step": 3863
    },
    {
      "epoch": 0.6960594460707048,
      "grad_norm": 1.2626633644104004,
      "learning_rate": 7.30546250724663e-05,
      "loss": 1.3658,
      "step": 3864
    },
    {
      "epoch": 0.6962395856789011,
      "grad_norm": 1.3056353330612183,
      "learning_rate": 7.304205788953385e-05,
      "loss": 1.5105,
      "step": 3865
    },
    {
      "epoch": 0.6964197252870975,
      "grad_norm": 1.3683557510375977,
      "learning_rate": 7.30294888581787e-05,
      "loss": 1.6711,
      "step": 3866
    },
    {
      "epoch": 0.6965998648952939,
      "grad_norm": 1.394254207611084,
      "learning_rate": 7.301691797940921e-05,
      "loss": 1.6437,
      "step": 3867
    },
    {
      "epoch": 0.6967800045034902,
      "grad_norm": 1.3778191804885864,
      "learning_rate": 7.300434525423375e-05,
      "loss": 1.6292,
      "step": 3868
    },
    {
      "epoch": 0.6969601441116866,
      "grad_norm": 1.2305551767349243,
      "learning_rate": 7.299177068366094e-05,
      "loss": 1.4671,
      "step": 3869
    },
    {
      "epoch": 0.6971402837198829,
      "grad_norm": 1.4378727674484253,
      "learning_rate": 7.297919426869947e-05,
      "loss": 1.7736,
      "step": 3870
    },
    {
      "epoch": 0.6973204233280793,
      "grad_norm": 1.4748146533966064,
      "learning_rate": 7.296661601035822e-05,
      "loss": 1.7749,
      "step": 3871
    },
    {
      "epoch": 0.6975005629362756,
      "grad_norm": 1.2209627628326416,
      "learning_rate": 7.295403590964623e-05,
      "loss": 1.6068,
      "step": 3872
    },
    {
      "epoch": 0.697680702544472,
      "grad_norm": 1.2132511138916016,
      "learning_rate": 7.294145396757266e-05,
      "loss": 1.5123,
      "step": 3873
    },
    {
      "epoch": 0.6978608421526683,
      "grad_norm": 1.3248944282531738,
      "learning_rate": 7.292887018514683e-05,
      "loss": 1.5667,
      "step": 3874
    },
    {
      "epoch": 0.6980409817608647,
      "grad_norm": 1.392854928970337,
      "learning_rate": 7.29162845633782e-05,
      "loss": 1.7377,
      "step": 3875
    },
    {
      "epoch": 0.698221121369061,
      "grad_norm": 1.2801711559295654,
      "learning_rate": 7.290369710327638e-05,
      "loss": 1.4187,
      "step": 3876
    },
    {
      "epoch": 0.6984012609772574,
      "grad_norm": 1.2746928930282593,
      "learning_rate": 7.289110780585115e-05,
      "loss": 1.3883,
      "step": 3877
    },
    {
      "epoch": 0.6985814005854537,
      "grad_norm": 1.3822486400604248,
      "learning_rate": 7.287851667211238e-05,
      "loss": 1.6047,
      "step": 3878
    },
    {
      "epoch": 0.6987615401936501,
      "grad_norm": 1.3795428276062012,
      "learning_rate": 7.286592370307014e-05,
      "loss": 1.5998,
      "step": 3879
    },
    {
      "epoch": 0.6989416798018464,
      "grad_norm": 1.468690276145935,
      "learning_rate": 7.285332889973467e-05,
      "loss": 1.7966,
      "step": 3880
    },
    {
      "epoch": 0.6991218194100428,
      "grad_norm": 1.3186126947402954,
      "learning_rate": 7.284073226311627e-05,
      "loss": 1.5719,
      "step": 3881
    },
    {
      "epoch": 0.6993019590182391,
      "grad_norm": 1.2671804428100586,
      "learning_rate": 7.282813379422546e-05,
      "loss": 1.5765,
      "step": 3882
    },
    {
      "epoch": 0.6994820986264355,
      "grad_norm": 1.311571478843689,
      "learning_rate": 7.281553349407287e-05,
      "loss": 1.7042,
      "step": 3883
    },
    {
      "epoch": 0.6996622382346318,
      "grad_norm": 1.355029821395874,
      "learning_rate": 7.280293136366929e-05,
      "loss": 1.6722,
      "step": 3884
    },
    {
      "epoch": 0.6998423778428282,
      "grad_norm": 1.5546504259109497,
      "learning_rate": 7.279032740402567e-05,
      "loss": 1.6475,
      "step": 3885
    },
    {
      "epoch": 0.7000225174510245,
      "grad_norm": 1.3103325366973877,
      "learning_rate": 7.277772161615309e-05,
      "loss": 1.4711,
      "step": 3886
    },
    {
      "epoch": 0.7002026570592209,
      "grad_norm": 1.3622653484344482,
      "learning_rate": 7.276511400106277e-05,
      "loss": 1.4725,
      "step": 3887
    },
    {
      "epoch": 0.7003827966674172,
      "grad_norm": 1.3578202724456787,
      "learning_rate": 7.275250455976609e-05,
      "loss": 1.6427,
      "step": 3888
    },
    {
      "epoch": 0.7005629362756136,
      "grad_norm": 1.2511574029922485,
      "learning_rate": 7.273989329327459e-05,
      "loss": 1.7029,
      "step": 3889
    },
    {
      "epoch": 0.7007430758838099,
      "grad_norm": 1.3640930652618408,
      "learning_rate": 7.272728020259992e-05,
      "loss": 1.7626,
      "step": 3890
    },
    {
      "epoch": 0.7009232154920063,
      "grad_norm": 1.3834847211837769,
      "learning_rate": 7.27146652887539e-05,
      "loss": 1.6992,
      "step": 3891
    },
    {
      "epoch": 0.7011033551002026,
      "grad_norm": 1.414367437362671,
      "learning_rate": 7.270204855274849e-05,
      "loss": 1.7793,
      "step": 3892
    },
    {
      "epoch": 0.701283494708399,
      "grad_norm": 1.5253984928131104,
      "learning_rate": 7.26894299955958e-05,
      "loss": 1.8531,
      "step": 3893
    },
    {
      "epoch": 0.7014636343165953,
      "grad_norm": 1.3273879289627075,
      "learning_rate": 7.26768096183081e-05,
      "loss": 1.3552,
      "step": 3894
    },
    {
      "epoch": 0.7016437739247917,
      "grad_norm": 1.562617301940918,
      "learning_rate": 7.266418742189778e-05,
      "loss": 1.6576,
      "step": 3895
    },
    {
      "epoch": 0.7018239135329881,
      "grad_norm": 1.5604127645492554,
      "learning_rate": 7.265156340737737e-05,
      "loss": 1.5371,
      "step": 3896
    },
    {
      "epoch": 0.7020040531411844,
      "grad_norm": 1.3432213068008423,
      "learning_rate": 7.263893757575959e-05,
      "loss": 1.4072,
      "step": 3897
    },
    {
      "epoch": 0.7021841927493808,
      "grad_norm": 1.4777671098709106,
      "learning_rate": 7.262630992805725e-05,
      "loss": 1.6485,
      "step": 3898
    },
    {
      "epoch": 0.7023643323575771,
      "grad_norm": 1.305596113204956,
      "learning_rate": 7.261368046528337e-05,
      "loss": 1.3085,
      "step": 3899
    },
    {
      "epoch": 0.7025444719657735,
      "grad_norm": 1.323996901512146,
      "learning_rate": 7.260104918845105e-05,
      "loss": 1.5566,
      "step": 3900
    },
    {
      "epoch": 0.7027246115739698,
      "grad_norm": 1.1090295314788818,
      "learning_rate": 7.258841609857357e-05,
      "loss": 1.5064,
      "step": 3901
    },
    {
      "epoch": 0.7029047511821662,
      "grad_norm": 1.2259312868118286,
      "learning_rate": 7.257578119666434e-05,
      "loss": 1.9847,
      "step": 3902
    },
    {
      "epoch": 0.7030848907903625,
      "grad_norm": 1.179762601852417,
      "learning_rate": 7.256314448373696e-05,
      "loss": 1.7255,
      "step": 3903
    },
    {
      "epoch": 0.7032650303985589,
      "grad_norm": 1.2060153484344482,
      "learning_rate": 7.255050596080509e-05,
      "loss": 1.8005,
      "step": 3904
    },
    {
      "epoch": 0.7034451700067552,
      "grad_norm": 1.2332481145858765,
      "learning_rate": 7.253786562888263e-05,
      "loss": 1.7718,
      "step": 3905
    },
    {
      "epoch": 0.7036253096149516,
      "grad_norm": 1.3497898578643799,
      "learning_rate": 7.252522348898359e-05,
      "loss": 1.8627,
      "step": 3906
    },
    {
      "epoch": 0.7038054492231479,
      "grad_norm": 1.3237775564193726,
      "learning_rate": 7.251257954212207e-05,
      "loss": 1.9482,
      "step": 3907
    },
    {
      "epoch": 0.7039855888313443,
      "grad_norm": 1.4205254316329956,
      "learning_rate": 7.249993378931239e-05,
      "loss": 1.8779,
      "step": 3908
    },
    {
      "epoch": 0.7041657284395406,
      "grad_norm": 1.6139785051345825,
      "learning_rate": 7.248728623156899e-05,
      "loss": 2.2808,
      "step": 3909
    },
    {
      "epoch": 0.704345868047737,
      "grad_norm": 1.4392380714416504,
      "learning_rate": 7.247463686990643e-05,
      "loss": 1.7677,
      "step": 3910
    },
    {
      "epoch": 0.7045260076559333,
      "grad_norm": 1.341322898864746,
      "learning_rate": 7.246198570533944e-05,
      "loss": 1.6774,
      "step": 3911
    },
    {
      "epoch": 0.7047061472641297,
      "grad_norm": 1.24701726436615,
      "learning_rate": 7.24493327388829e-05,
      "loss": 1.5218,
      "step": 3912
    },
    {
      "epoch": 0.704886286872326,
      "grad_norm": 1.2835712432861328,
      "learning_rate": 7.243667797155185e-05,
      "loss": 1.6536,
      "step": 3913
    },
    {
      "epoch": 0.7050664264805224,
      "grad_norm": 1.197148323059082,
      "learning_rate": 7.24240214043614e-05,
      "loss": 1.541,
      "step": 3914
    },
    {
      "epoch": 0.7052465660887187,
      "grad_norm": 1.166495442390442,
      "learning_rate": 7.241136303832687e-05,
      "loss": 1.5309,
      "step": 3915
    },
    {
      "epoch": 0.7054267056969151,
      "grad_norm": 1.3184785842895508,
      "learning_rate": 7.239870287446372e-05,
      "loss": 1.4287,
      "step": 3916
    },
    {
      "epoch": 0.7056068453051114,
      "grad_norm": 1.2774484157562256,
      "learning_rate": 7.238604091378753e-05,
      "loss": 1.4549,
      "step": 3917
    },
    {
      "epoch": 0.7057869849133078,
      "grad_norm": 1.3129510879516602,
      "learning_rate": 7.237337715731405e-05,
      "loss": 1.726,
      "step": 3918
    },
    {
      "epoch": 0.7059671245215041,
      "grad_norm": 1.2660243511199951,
      "learning_rate": 7.236071160605914e-05,
      "loss": 1.4032,
      "step": 3919
    },
    {
      "epoch": 0.7061472641297005,
      "grad_norm": 1.1676615476608276,
      "learning_rate": 7.234804426103885e-05,
      "loss": 1.4168,
      "step": 3920
    },
    {
      "epoch": 0.7063274037378968,
      "grad_norm": 1.321829915046692,
      "learning_rate": 7.233537512326934e-05,
      "loss": 1.509,
      "step": 3921
    },
    {
      "epoch": 0.7065075433460932,
      "grad_norm": 1.3350731134414673,
      "learning_rate": 7.23227041937669e-05,
      "loss": 1.4685,
      "step": 3922
    },
    {
      "epoch": 0.7066876829542895,
      "grad_norm": 1.320751667022705,
      "learning_rate": 7.2310031473548e-05,
      "loss": 1.7096,
      "step": 3923
    },
    {
      "epoch": 0.7068678225624859,
      "grad_norm": 1.2472236156463623,
      "learning_rate": 7.229735696362927e-05,
      "loss": 1.3546,
      "step": 3924
    },
    {
      "epoch": 0.7070479621706823,
      "grad_norm": 1.34446382522583,
      "learning_rate": 7.228468066502741e-05,
      "loss": 1.6708,
      "step": 3925
    },
    {
      "epoch": 0.7072281017788786,
      "grad_norm": 1.2939053773880005,
      "learning_rate": 7.227200257875931e-05,
      "loss": 1.515,
      "step": 3926
    },
    {
      "epoch": 0.707408241387075,
      "grad_norm": 1.395036220550537,
      "learning_rate": 7.225932270584202e-05,
      "loss": 1.6256,
      "step": 3927
    },
    {
      "epoch": 0.7075883809952713,
      "grad_norm": 1.2223687171936035,
      "learning_rate": 7.224664104729271e-05,
      "loss": 1.5721,
      "step": 3928
    },
    {
      "epoch": 0.7077685206034677,
      "grad_norm": 1.5163313150405884,
      "learning_rate": 7.223395760412868e-05,
      "loss": 1.7254,
      "step": 3929
    },
    {
      "epoch": 0.707948660211664,
      "grad_norm": 1.196772813796997,
      "learning_rate": 7.22212723773674e-05,
      "loss": 1.2991,
      "step": 3930
    },
    {
      "epoch": 0.7081287998198604,
      "grad_norm": 1.1617646217346191,
      "learning_rate": 7.22085853680265e-05,
      "loss": 1.506,
      "step": 3931
    },
    {
      "epoch": 0.7083089394280567,
      "grad_norm": 1.4280757904052734,
      "learning_rate": 7.219589657712367e-05,
      "loss": 1.6633,
      "step": 3932
    },
    {
      "epoch": 0.7084890790362531,
      "grad_norm": 1.447654366493225,
      "learning_rate": 7.218320600567683e-05,
      "loss": 1.7713,
      "step": 3933
    },
    {
      "epoch": 0.7086692186444494,
      "grad_norm": 1.418230652809143,
      "learning_rate": 7.217051365470403e-05,
      "loss": 1.6651,
      "step": 3934
    },
    {
      "epoch": 0.7088493582526458,
      "grad_norm": 1.446295142173767,
      "learning_rate": 7.21578195252234e-05,
      "loss": 1.7017,
      "step": 3935
    },
    {
      "epoch": 0.7090294978608421,
      "grad_norm": 1.2032145261764526,
      "learning_rate": 7.214512361825328e-05,
      "loss": 1.6577,
      "step": 3936
    },
    {
      "epoch": 0.7092096374690385,
      "grad_norm": 1.3188484907150269,
      "learning_rate": 7.213242593481214e-05,
      "loss": 1.5251,
      "step": 3937
    },
    {
      "epoch": 0.7093897770772348,
      "grad_norm": 1.3633226156234741,
      "learning_rate": 7.211972647591858e-05,
      "loss": 1.5452,
      "step": 3938
    },
    {
      "epoch": 0.7095699166854312,
      "grad_norm": 1.3551186323165894,
      "learning_rate": 7.210702524259132e-05,
      "loss": 1.7061,
      "step": 3939
    },
    {
      "epoch": 0.7097500562936275,
      "grad_norm": 1.3911857604980469,
      "learning_rate": 7.209432223584928e-05,
      "loss": 1.6126,
      "step": 3940
    },
    {
      "epoch": 0.7099301959018239,
      "grad_norm": 1.241748332977295,
      "learning_rate": 7.208161745671146e-05,
      "loss": 1.6858,
      "step": 3941
    },
    {
      "epoch": 0.7101103355100202,
      "grad_norm": 1.389390230178833,
      "learning_rate": 7.206891090619706e-05,
      "loss": 1.6792,
      "step": 3942
    },
    {
      "epoch": 0.7102904751182166,
      "grad_norm": 1.5643136501312256,
      "learning_rate": 7.205620258532534e-05,
      "loss": 1.4664,
      "step": 3943
    },
    {
      "epoch": 0.7104706147264129,
      "grad_norm": 1.4613926410675049,
      "learning_rate": 7.204349249511582e-05,
      "loss": 1.8153,
      "step": 3944
    },
    {
      "epoch": 0.7106507543346093,
      "grad_norm": 1.3153282403945923,
      "learning_rate": 7.203078063658807e-05,
      "loss": 1.5791,
      "step": 3945
    },
    {
      "epoch": 0.7108308939428056,
      "grad_norm": 1.5081466436386108,
      "learning_rate": 7.201806701076181e-05,
      "loss": 1.7991,
      "step": 3946
    },
    {
      "epoch": 0.711011033551002,
      "grad_norm": 1.3599598407745361,
      "learning_rate": 7.200535161865696e-05,
      "loss": 1.8239,
      "step": 3947
    },
    {
      "epoch": 0.7111911731591983,
      "grad_norm": 1.339067816734314,
      "learning_rate": 7.19926344612935e-05,
      "loss": 1.3831,
      "step": 3948
    },
    {
      "epoch": 0.7113713127673947,
      "grad_norm": 1.2522943019866943,
      "learning_rate": 7.197991553969164e-05,
      "loss": 1.3998,
      "step": 3949
    },
    {
      "epoch": 0.711551452375591,
      "grad_norm": 1.3630602359771729,
      "learning_rate": 7.196719485487164e-05,
      "loss": 1.5852,
      "step": 3950
    },
    {
      "epoch": 0.7117315919837874,
      "grad_norm": 1.2519137859344482,
      "learning_rate": 7.195447240785399e-05,
      "loss": 2.053,
      "step": 3951
    },
    {
      "epoch": 0.7119117315919838,
      "grad_norm": 1.4044971466064453,
      "learning_rate": 7.194174819965925e-05,
      "loss": 2.0164,
      "step": 3952
    },
    {
      "epoch": 0.7120918712001801,
      "grad_norm": 1.2754586935043335,
      "learning_rate": 7.192902223130814e-05,
      "loss": 2.0396,
      "step": 3953
    },
    {
      "epoch": 0.7122720108083765,
      "grad_norm": 1.1970125436782837,
      "learning_rate": 7.191629450382155e-05,
      "loss": 1.632,
      "step": 3954
    },
    {
      "epoch": 0.7124521504165728,
      "grad_norm": 1.2506103515625,
      "learning_rate": 7.19035650182205e-05,
      "loss": 1.9135,
      "step": 3955
    },
    {
      "epoch": 0.7126322900247692,
      "grad_norm": 1.161482572555542,
      "learning_rate": 7.189083377552615e-05,
      "loss": 1.4625,
      "step": 3956
    },
    {
      "epoch": 0.7128124296329655,
      "grad_norm": 1.3697588443756104,
      "learning_rate": 7.187810077675975e-05,
      "loss": 1.9631,
      "step": 3957
    },
    {
      "epoch": 0.7129925692411619,
      "grad_norm": 1.3391778469085693,
      "learning_rate": 7.186536602294278e-05,
      "loss": 1.8479,
      "step": 3958
    },
    {
      "epoch": 0.7131727088493582,
      "grad_norm": 1.4021704196929932,
      "learning_rate": 7.185262951509678e-05,
      "loss": 1.9375,
      "step": 3959
    },
    {
      "epoch": 0.7133528484575546,
      "grad_norm": 1.9403471946716309,
      "learning_rate": 7.183989125424349e-05,
      "loss": 2.463,
      "step": 3960
    },
    {
      "epoch": 0.7135329880657509,
      "grad_norm": 1.3209342956542969,
      "learning_rate": 7.182715124140478e-05,
      "loss": 1.7369,
      "step": 3961
    },
    {
      "epoch": 0.7137131276739473,
      "grad_norm": 1.35624098777771,
      "learning_rate": 7.181440947760261e-05,
      "loss": 1.7724,
      "step": 3962
    },
    {
      "epoch": 0.7138932672821436,
      "grad_norm": 1.3411002159118652,
      "learning_rate": 7.180166596385914e-05,
      "loss": 1.627,
      "step": 3963
    },
    {
      "epoch": 0.71407340689034,
      "grad_norm": 1.2416220903396606,
      "learning_rate": 7.178892070119665e-05,
      "loss": 1.6305,
      "step": 3964
    },
    {
      "epoch": 0.7142535464985363,
      "grad_norm": 1.2578668594360352,
      "learning_rate": 7.177617369063755e-05,
      "loss": 1.6771,
      "step": 3965
    },
    {
      "epoch": 0.7144336861067327,
      "grad_norm": 1.3047893047332764,
      "learning_rate": 7.17634249332044e-05,
      "loss": 1.5447,
      "step": 3966
    },
    {
      "epoch": 0.714613825714929,
      "grad_norm": 1.4433828592300415,
      "learning_rate": 7.175067442991992e-05,
      "loss": 1.5483,
      "step": 3967
    },
    {
      "epoch": 0.7147939653231254,
      "grad_norm": 1.3985286951065063,
      "learning_rate": 7.173792218180692e-05,
      "loss": 1.6664,
      "step": 3968
    },
    {
      "epoch": 0.7149741049313217,
      "grad_norm": 1.3115431070327759,
      "learning_rate": 7.172516818988838e-05,
      "loss": 1.6203,
      "step": 3969
    },
    {
      "epoch": 0.7151542445395181,
      "grad_norm": 1.1956276893615723,
      "learning_rate": 7.171241245518744e-05,
      "loss": 1.5123,
      "step": 3970
    },
    {
      "epoch": 0.7153343841477144,
      "grad_norm": 1.3231576681137085,
      "learning_rate": 7.169965497872735e-05,
      "loss": 1.4703,
      "step": 3971
    },
    {
      "epoch": 0.7155145237559108,
      "grad_norm": 1.3194525241851807,
      "learning_rate": 7.168689576153151e-05,
      "loss": 1.4696,
      "step": 3972
    },
    {
      "epoch": 0.7156946633641071,
      "grad_norm": 1.4018319845199585,
      "learning_rate": 7.167413480462344e-05,
      "loss": 1.6021,
      "step": 3973
    },
    {
      "epoch": 0.7158748029723035,
      "grad_norm": 1.2673927545547485,
      "learning_rate": 7.166137210902685e-05,
      "loss": 1.5744,
      "step": 3974
    },
    {
      "epoch": 0.7160549425804998,
      "grad_norm": 1.2739746570587158,
      "learning_rate": 7.164860767576553e-05,
      "loss": 1.4467,
      "step": 3975
    },
    {
      "epoch": 0.7162350821886962,
      "grad_norm": 1.2330594062805176,
      "learning_rate": 7.163584150586345e-05,
      "loss": 1.2784,
      "step": 3976
    },
    {
      "epoch": 0.7164152217968925,
      "grad_norm": 1.389986276626587,
      "learning_rate": 7.16230736003447e-05,
      "loss": 1.6499,
      "step": 3977
    },
    {
      "epoch": 0.716595361405089,
      "grad_norm": 1.3281902074813843,
      "learning_rate": 7.161030396023353e-05,
      "loss": 1.746,
      "step": 3978
    },
    {
      "epoch": 0.7167755010132852,
      "grad_norm": 1.2783664464950562,
      "learning_rate": 7.15975325865543e-05,
      "loss": 1.7524,
      "step": 3979
    },
    {
      "epoch": 0.7169556406214816,
      "grad_norm": 1.2845207452774048,
      "learning_rate": 7.158475948033151e-05,
      "loss": 1.5091,
      "step": 3980
    },
    {
      "epoch": 0.717135780229678,
      "grad_norm": 1.2399158477783203,
      "learning_rate": 7.157198464258984e-05,
      "loss": 1.463,
      "step": 3981
    },
    {
      "epoch": 0.7173159198378743,
      "grad_norm": 1.195797085762024,
      "learning_rate": 7.155920807435406e-05,
      "loss": 1.3059,
      "step": 3982
    },
    {
      "epoch": 0.7174960594460708,
      "grad_norm": 1.3450088500976562,
      "learning_rate": 7.154642977664912e-05,
      "loss": 1.6668,
      "step": 3983
    },
    {
      "epoch": 0.717676199054267,
      "grad_norm": 1.4196999073028564,
      "learning_rate": 7.153364975050009e-05,
      "loss": 1.6847,
      "step": 3984
    },
    {
      "epoch": 0.7178563386624635,
      "grad_norm": 1.3191006183624268,
      "learning_rate": 7.152086799693214e-05,
      "loss": 1.5292,
      "step": 3985
    },
    {
      "epoch": 0.7180364782706598,
      "grad_norm": 1.2243523597717285,
      "learning_rate": 7.150808451697066e-05,
      "loss": 1.6042,
      "step": 3986
    },
    {
      "epoch": 0.7182166178788562,
      "grad_norm": 1.3338755369186401,
      "learning_rate": 7.14952993116411e-05,
      "loss": 1.5491,
      "step": 3987
    },
    {
      "epoch": 0.7183967574870525,
      "grad_norm": 1.3634601831436157,
      "learning_rate": 7.14825123819691e-05,
      "loss": 1.5375,
      "step": 3988
    },
    {
      "epoch": 0.7185768970952489,
      "grad_norm": 1.427426815032959,
      "learning_rate": 7.146972372898045e-05,
      "loss": 1.7399,
      "step": 3989
    },
    {
      "epoch": 0.7187570367034452,
      "grad_norm": 1.548447608947754,
      "learning_rate": 7.145693335370098e-05,
      "loss": 1.8068,
      "step": 3990
    },
    {
      "epoch": 0.7189371763116416,
      "grad_norm": 1.4484425783157349,
      "learning_rate": 7.14441412571568e-05,
      "loss": 1.6907,
      "step": 3991
    },
    {
      "epoch": 0.7191173159198379,
      "grad_norm": 1.2436573505401611,
      "learning_rate": 7.143134744037403e-05,
      "loss": 1.357,
      "step": 3992
    },
    {
      "epoch": 0.7192974555280343,
      "grad_norm": 1.5230742692947388,
      "learning_rate": 7.141855190437901e-05,
      "loss": 1.5732,
      "step": 3993
    },
    {
      "epoch": 0.7194775951362306,
      "grad_norm": 1.416967511177063,
      "learning_rate": 7.140575465019819e-05,
      "loss": 1.6801,
      "step": 3994
    },
    {
      "epoch": 0.719657734744427,
      "grad_norm": 1.3700101375579834,
      "learning_rate": 7.139295567885818e-05,
      "loss": 1.475,
      "step": 3995
    },
    {
      "epoch": 0.7198378743526233,
      "grad_norm": 1.3727749586105347,
      "learning_rate": 7.138015499138566e-05,
      "loss": 1.5376,
      "step": 3996
    },
    {
      "epoch": 0.7200180139608197,
      "grad_norm": 1.3151254653930664,
      "learning_rate": 7.136735258880754e-05,
      "loss": 1.6662,
      "step": 3997
    },
    {
      "epoch": 0.720198153569016,
      "grad_norm": 1.365763783454895,
      "learning_rate": 7.135454847215079e-05,
      "loss": 1.4079,
      "step": 3998
    },
    {
      "epoch": 0.7203782931772124,
      "grad_norm": 1.4193084239959717,
      "learning_rate": 7.134174264244256e-05,
      "loss": 1.5276,
      "step": 3999
    },
    {
      "epoch": 0.7205584327854087,
      "grad_norm": 1.2450289726257324,
      "learning_rate": 7.132893510071014e-05,
      "loss": 1.3972,
      "step": 4000
    },
    {
      "epoch": 0.7207385723936051,
      "grad_norm": 1.3603992462158203,
      "learning_rate": 7.131612584798093e-05,
      "loss": 2.3097,
      "step": 4001
    },
    {
      "epoch": 0.7209187120018014,
      "grad_norm": 1.162360668182373,
      "learning_rate": 7.130331488528249e-05,
      "loss": 1.7911,
      "step": 4002
    },
    {
      "epoch": 0.7210988516099978,
      "grad_norm": 1.2884531021118164,
      "learning_rate": 7.12905022136425e-05,
      "loss": 2.0206,
      "step": 4003
    },
    {
      "epoch": 0.721278991218194,
      "grad_norm": 1.523433804512024,
      "learning_rate": 7.127768783408879e-05,
      "loss": 1.9963,
      "step": 4004
    },
    {
      "epoch": 0.7214591308263905,
      "grad_norm": 1.2343974113464355,
      "learning_rate": 7.126487174764936e-05,
      "loss": 1.7754,
      "step": 4005
    },
    {
      "epoch": 0.7216392704345868,
      "grad_norm": 1.3891587257385254,
      "learning_rate": 7.125205395535223e-05,
      "loss": 1.6804,
      "step": 4006
    },
    {
      "epoch": 0.7218194100427832,
      "grad_norm": 1.4183688163757324,
      "learning_rate": 7.12392344582257e-05,
      "loss": 1.948,
      "step": 4007
    },
    {
      "epoch": 0.7219995496509795,
      "grad_norm": 1.459185242652893,
      "learning_rate": 7.122641325729813e-05,
      "loss": 1.9604,
      "step": 4008
    },
    {
      "epoch": 0.7221796892591759,
      "grad_norm": 1.7456086874008179,
      "learning_rate": 7.121359035359802e-05,
      "loss": 2.0318,
      "step": 4009
    },
    {
      "epoch": 0.7223598288673723,
      "grad_norm": 1.3245593309402466,
      "learning_rate": 7.120076574815404e-05,
      "loss": 1.6815,
      "step": 4010
    },
    {
      "epoch": 0.7225399684755686,
      "grad_norm": 1.2667951583862305,
      "learning_rate": 7.118793944199496e-05,
      "loss": 1.31,
      "step": 4011
    },
    {
      "epoch": 0.722720108083765,
      "grad_norm": 1.2869495153427124,
      "learning_rate": 7.117511143614968e-05,
      "loss": 1.6299,
      "step": 4012
    },
    {
      "epoch": 0.7229002476919613,
      "grad_norm": 1.3147913217544556,
      "learning_rate": 7.116228173164727e-05,
      "loss": 1.5726,
      "step": 4013
    },
    {
      "epoch": 0.7230803873001577,
      "grad_norm": 1.3100680112838745,
      "learning_rate": 7.114945032951696e-05,
      "loss": 1.5938,
      "step": 4014
    },
    {
      "epoch": 0.723260526908354,
      "grad_norm": 1.3373206853866577,
      "learning_rate": 7.113661723078801e-05,
      "loss": 1.4637,
      "step": 4015
    },
    {
      "epoch": 0.7234406665165504,
      "grad_norm": 1.221670389175415,
      "learning_rate": 7.112378243648995e-05,
      "loss": 1.6497,
      "step": 4016
    },
    {
      "epoch": 0.7236208061247467,
      "grad_norm": 1.302420735359192,
      "learning_rate": 7.111094594765234e-05,
      "loss": 1.6215,
      "step": 4017
    },
    {
      "epoch": 0.7238009457329431,
      "grad_norm": 1.2747881412506104,
      "learning_rate": 7.109810776530492e-05,
      "loss": 1.4865,
      "step": 4018
    },
    {
      "epoch": 0.7239810853411394,
      "grad_norm": 1.280052661895752,
      "learning_rate": 7.108526789047758e-05,
      "loss": 1.4805,
      "step": 4019
    },
    {
      "epoch": 0.7241612249493358,
      "grad_norm": 1.3133503198623657,
      "learning_rate": 7.107242632420032e-05,
      "loss": 1.372,
      "step": 4020
    },
    {
      "epoch": 0.7243413645575321,
      "grad_norm": 1.3217718601226807,
      "learning_rate": 7.10595830675033e-05,
      "loss": 1.5532,
      "step": 4021
    },
    {
      "epoch": 0.7245215041657285,
      "grad_norm": 1.302534580230713,
      "learning_rate": 7.104673812141675e-05,
      "loss": 1.5096,
      "step": 4022
    },
    {
      "epoch": 0.7247016437739248,
      "grad_norm": 1.3490378856658936,
      "learning_rate": 7.103389148697114e-05,
      "loss": 1.5999,
      "step": 4023
    },
    {
      "epoch": 0.7248817833821212,
      "grad_norm": 1.358303427696228,
      "learning_rate": 7.1021043165197e-05,
      "loss": 1.5972,
      "step": 4024
    },
    {
      "epoch": 0.7250619229903175,
      "grad_norm": 1.3473252058029175,
      "learning_rate": 7.100819315712501e-05,
      "loss": 1.644,
      "step": 4025
    },
    {
      "epoch": 0.7252420625985139,
      "grad_norm": 1.2381086349487305,
      "learning_rate": 7.0995341463786e-05,
      "loss": 1.4305,
      "step": 4026
    },
    {
      "epoch": 0.7254222022067102,
      "grad_norm": 1.2565757036209106,
      "learning_rate": 7.098248808621093e-05,
      "loss": 1.4568,
      "step": 4027
    },
    {
      "epoch": 0.7256023418149066,
      "grad_norm": 1.3137116432189941,
      "learning_rate": 7.096963302543088e-05,
      "loss": 1.5448,
      "step": 4028
    },
    {
      "epoch": 0.7257824814231029,
      "grad_norm": 1.2895878553390503,
      "learning_rate": 7.095677628247708e-05,
      "loss": 1.4827,
      "step": 4029
    },
    {
      "epoch": 0.7259626210312993,
      "grad_norm": 1.4240233898162842,
      "learning_rate": 7.094391785838091e-05,
      "loss": 1.7361,
      "step": 4030
    },
    {
      "epoch": 0.7261427606394956,
      "grad_norm": 1.5960263013839722,
      "learning_rate": 7.093105775417384e-05,
      "loss": 1.9167,
      "step": 4031
    },
    {
      "epoch": 0.726322900247692,
      "grad_norm": 1.33319091796875,
      "learning_rate": 7.091819597088751e-05,
      "loss": 1.6361,
      "step": 4032
    },
    {
      "epoch": 0.7265030398558883,
      "grad_norm": 1.4717464447021484,
      "learning_rate": 7.090533250955369e-05,
      "loss": 1.6254,
      "step": 4033
    },
    {
      "epoch": 0.7266831794640847,
      "grad_norm": 1.4692511558532715,
      "learning_rate": 7.089246737120428e-05,
      "loss": 1.6136,
      "step": 4034
    },
    {
      "epoch": 0.726863319072281,
      "grad_norm": 1.498299241065979,
      "learning_rate": 7.08796005568713e-05,
      "loss": 1.9383,
      "step": 4035
    },
    {
      "epoch": 0.7270434586804774,
      "grad_norm": 1.4138383865356445,
      "learning_rate": 7.086673206758694e-05,
      "loss": 1.6045,
      "step": 4036
    },
    {
      "epoch": 0.7272235982886737,
      "grad_norm": 1.4371824264526367,
      "learning_rate": 7.085386190438352e-05,
      "loss": 1.5348,
      "step": 4037
    },
    {
      "epoch": 0.7274037378968701,
      "grad_norm": 1.4831973314285278,
      "learning_rate": 7.084099006829344e-05,
      "loss": 1.7428,
      "step": 4038
    },
    {
      "epoch": 0.7275838775050665,
      "grad_norm": 1.272243618965149,
      "learning_rate": 7.082811656034928e-05,
      "loss": 1.4864,
      "step": 4039
    },
    {
      "epoch": 0.7277640171132628,
      "grad_norm": 1.374729871749878,
      "learning_rate": 7.081524138158377e-05,
      "loss": 1.5607,
      "step": 4040
    },
    {
      "epoch": 0.7279441567214592,
      "grad_norm": 1.5588423013687134,
      "learning_rate": 7.08023645330297e-05,
      "loss": 1.9319,
      "step": 4041
    },
    {
      "epoch": 0.7281242963296555,
      "grad_norm": 1.4162218570709229,
      "learning_rate": 7.078948601572012e-05,
      "loss": 1.6257,
      "step": 4042
    },
    {
      "epoch": 0.7283044359378519,
      "grad_norm": 1.40138840675354,
      "learning_rate": 7.077660583068808e-05,
      "loss": 1.6812,
      "step": 4043
    },
    {
      "epoch": 0.7284845755460482,
      "grad_norm": 1.5075501203536987,
      "learning_rate": 7.076372397896685e-05,
      "loss": 1.8057,
      "step": 4044
    },
    {
      "epoch": 0.7286647151542446,
      "grad_norm": 1.4859552383422852,
      "learning_rate": 7.07508404615898e-05,
      "loss": 1.5626,
      "step": 4045
    },
    {
      "epoch": 0.7288448547624409,
      "grad_norm": 1.4579800367355347,
      "learning_rate": 7.073795527959041e-05,
      "loss": 1.469,
      "step": 4046
    },
    {
      "epoch": 0.7290249943706373,
      "grad_norm": 1.393281102180481,
      "learning_rate": 7.072506843400237e-05,
      "loss": 1.4869,
      "step": 4047
    },
    {
      "epoch": 0.7292051339788336,
      "grad_norm": 1.4526339769363403,
      "learning_rate": 7.071217992585943e-05,
      "loss": 1.5421,
      "step": 4048
    },
    {
      "epoch": 0.72938527358703,
      "grad_norm": 1.3382951021194458,
      "learning_rate": 7.069928975619551e-05,
      "loss": 1.6026,
      "step": 4049
    },
    {
      "epoch": 0.7295654131952263,
      "grad_norm": 1.3069384098052979,
      "learning_rate": 7.068639792604463e-05,
      "loss": 1.2638,
      "step": 4050
    },
    {
      "epoch": 0.7297455528034227,
      "grad_norm": 1.2935819625854492,
      "learning_rate": 7.0673504436441e-05,
      "loss": 2.17,
      "step": 4051
    },
    {
      "epoch": 0.729925692411619,
      "grad_norm": 1.1897939443588257,
      "learning_rate": 7.066060928841892e-05,
      "loss": 1.9364,
      "step": 4052
    },
    {
      "epoch": 0.7301058320198154,
      "grad_norm": 1.3020228147506714,
      "learning_rate": 7.064771248301282e-05,
      "loss": 2.1757,
      "step": 4053
    },
    {
      "epoch": 0.7302859716280117,
      "grad_norm": 1.299071192741394,
      "learning_rate": 7.063481402125728e-05,
      "loss": 1.9235,
      "step": 4054
    },
    {
      "epoch": 0.7304661112362081,
      "grad_norm": 1.297268033027649,
      "learning_rate": 7.0621913904187e-05,
      "loss": 2.168,
      "step": 4055
    },
    {
      "epoch": 0.7306462508444044,
      "grad_norm": 1.1980663537979126,
      "learning_rate": 7.060901213283684e-05,
      "loss": 1.8966,
      "step": 4056
    },
    {
      "epoch": 0.7308263904526008,
      "grad_norm": 1.4256212711334229,
      "learning_rate": 7.059610870824176e-05,
      "loss": 2.0502,
      "step": 4057
    },
    {
      "epoch": 0.7310065300607971,
      "grad_norm": 1.5272873640060425,
      "learning_rate": 7.058320363143688e-05,
      "loss": 1.9595,
      "step": 4058
    },
    {
      "epoch": 0.7311866696689935,
      "grad_norm": 1.5679125785827637,
      "learning_rate": 7.057029690345742e-05,
      "loss": 2.1143,
      "step": 4059
    },
    {
      "epoch": 0.7313668092771898,
      "grad_norm": 1.5130186080932617,
      "learning_rate": 7.055738852533877e-05,
      "loss": 1.9007,
      "step": 4060
    },
    {
      "epoch": 0.7315469488853862,
      "grad_norm": 1.260037899017334,
      "learning_rate": 7.054447849811642e-05,
      "loss": 1.6843,
      "step": 4061
    },
    {
      "epoch": 0.7317270884935825,
      "grad_norm": 1.2341854572296143,
      "learning_rate": 7.053156682282603e-05,
      "loss": 1.4969,
      "step": 4062
    },
    {
      "epoch": 0.7319072281017789,
      "grad_norm": 1.2701035737991333,
      "learning_rate": 7.051865350050332e-05,
      "loss": 1.5142,
      "step": 4063
    },
    {
      "epoch": 0.7320873677099752,
      "grad_norm": 1.486148476600647,
      "learning_rate": 7.050573853218425e-05,
      "loss": 1.9954,
      "step": 4064
    },
    {
      "epoch": 0.7322675073181716,
      "grad_norm": 1.300268292427063,
      "learning_rate": 7.04928219189048e-05,
      "loss": 1.7385,
      "step": 4065
    },
    {
      "epoch": 0.732447646926368,
      "grad_norm": 1.2946431636810303,
      "learning_rate": 7.047990366170116e-05,
      "loss": 1.6872,
      "step": 4066
    },
    {
      "epoch": 0.7326277865345643,
      "grad_norm": 1.3352512121200562,
      "learning_rate": 7.046698376160964e-05,
      "loss": 1.7155,
      "step": 4067
    },
    {
      "epoch": 0.7328079261427607,
      "grad_norm": 1.4107164144515991,
      "learning_rate": 7.045406221966663e-05,
      "loss": 1.6769,
      "step": 4068
    },
    {
      "epoch": 0.732988065750957,
      "grad_norm": 1.2012137174606323,
      "learning_rate": 7.044113903690873e-05,
      "loss": 1.5363,
      "step": 4069
    },
    {
      "epoch": 0.7331682053591534,
      "grad_norm": 1.133810043334961,
      "learning_rate": 7.042821421437262e-05,
      "loss": 1.541,
      "step": 4070
    },
    {
      "epoch": 0.7333483449673497,
      "grad_norm": 1.2481111288070679,
      "learning_rate": 7.04152877530951e-05,
      "loss": 1.5736,
      "step": 4071
    },
    {
      "epoch": 0.7335284845755461,
      "grad_norm": 1.3325014114379883,
      "learning_rate": 7.040235965411314e-05,
      "loss": 1.6684,
      "step": 4072
    },
    {
      "epoch": 0.7337086241837424,
      "grad_norm": 1.0985866785049438,
      "learning_rate": 7.038942991846382e-05,
      "loss": 1.3557,
      "step": 4073
    },
    {
      "epoch": 0.7338887637919388,
      "grad_norm": 1.3543462753295898,
      "learning_rate": 7.037649854718438e-05,
      "loss": 1.6038,
      "step": 4074
    },
    {
      "epoch": 0.7340689034001351,
      "grad_norm": 1.2914390563964844,
      "learning_rate": 7.036356554131214e-05,
      "loss": 1.3374,
      "step": 4075
    },
    {
      "epoch": 0.7342490430083315,
      "grad_norm": 1.1897563934326172,
      "learning_rate": 7.03506309018846e-05,
      "loss": 1.4623,
      "step": 4076
    },
    {
      "epoch": 0.7344291826165278,
      "grad_norm": 1.2917678356170654,
      "learning_rate": 7.033769462993936e-05,
      "loss": 1.4686,
      "step": 4077
    },
    {
      "epoch": 0.7346093222247242,
      "grad_norm": 1.387580394744873,
      "learning_rate": 7.032475672651415e-05,
      "loss": 1.7882,
      "step": 4078
    },
    {
      "epoch": 0.7347894618329205,
      "grad_norm": 1.1911344528198242,
      "learning_rate": 7.031181719264687e-05,
      "loss": 1.3947,
      "step": 4079
    },
    {
      "epoch": 0.7349696014411169,
      "grad_norm": 1.3339983224868774,
      "learning_rate": 7.02988760293755e-05,
      "loss": 1.5524,
      "step": 4080
    },
    {
      "epoch": 0.7351497410493132,
      "grad_norm": 1.2976665496826172,
      "learning_rate": 7.02859332377382e-05,
      "loss": 1.7326,
      "step": 4081
    },
    {
      "epoch": 0.7353298806575096,
      "grad_norm": 1.389215111732483,
      "learning_rate": 7.02729888187732e-05,
      "loss": 1.6981,
      "step": 4082
    },
    {
      "epoch": 0.7355100202657059,
      "grad_norm": 1.3509610891342163,
      "learning_rate": 7.026004277351889e-05,
      "loss": 1.6327,
      "step": 4083
    },
    {
      "epoch": 0.7356901598739023,
      "grad_norm": 1.4561971426010132,
      "learning_rate": 7.024709510301385e-05,
      "loss": 1.6417,
      "step": 4084
    },
    {
      "epoch": 0.7358702994820986,
      "grad_norm": 1.2138378620147705,
      "learning_rate": 7.023414580829668e-05,
      "loss": 1.21,
      "step": 4085
    },
    {
      "epoch": 0.736050439090295,
      "grad_norm": 1.3451802730560303,
      "learning_rate": 7.022119489040621e-05,
      "loss": 1.5169,
      "step": 4086
    },
    {
      "epoch": 0.7362305786984913,
      "grad_norm": 1.398783564567566,
      "learning_rate": 7.020824235038133e-05,
      "loss": 1.5407,
      "step": 4087
    },
    {
      "epoch": 0.7364107183066877,
      "grad_norm": 1.3204797506332397,
      "learning_rate": 7.019528818926108e-05,
      "loss": 1.5791,
      "step": 4088
    },
    {
      "epoch": 0.736590857914884,
      "grad_norm": 1.3674566745758057,
      "learning_rate": 7.018233240808467e-05,
      "loss": 1.5936,
      "step": 4089
    },
    {
      "epoch": 0.7367709975230804,
      "grad_norm": 1.2220652103424072,
      "learning_rate": 7.016937500789137e-05,
      "loss": 1.4255,
      "step": 4090
    },
    {
      "epoch": 0.7369511371312767,
      "grad_norm": 1.4028562307357788,
      "learning_rate": 7.015641598972064e-05,
      "loss": 1.4876,
      "step": 4091
    },
    {
      "epoch": 0.7371312767394731,
      "grad_norm": 1.2851715087890625,
      "learning_rate": 7.014345535461202e-05,
      "loss": 1.5646,
      "step": 4092
    },
    {
      "epoch": 0.7373114163476694,
      "grad_norm": 1.415369987487793,
      "learning_rate": 7.013049310360524e-05,
      "loss": 1.39,
      "step": 4093
    },
    {
      "epoch": 0.7374915559558658,
      "grad_norm": 1.3153975009918213,
      "learning_rate": 7.011752923774009e-05,
      "loss": 1.4428,
      "step": 4094
    },
    {
      "epoch": 0.7376716955640622,
      "grad_norm": 1.3487350940704346,
      "learning_rate": 7.010456375805656e-05,
      "loss": 1.457,
      "step": 4095
    },
    {
      "epoch": 0.7378518351722585,
      "grad_norm": 1.3977594375610352,
      "learning_rate": 7.009159666559472e-05,
      "loss": 1.5324,
      "step": 4096
    },
    {
      "epoch": 0.7380319747804549,
      "grad_norm": 1.3922799825668335,
      "learning_rate": 7.007862796139477e-05,
      "loss": 1.3293,
      "step": 4097
    },
    {
      "epoch": 0.7382121143886512,
      "grad_norm": 1.6159543991088867,
      "learning_rate": 7.006565764649708e-05,
      "loss": 1.4438,
      "step": 4098
    },
    {
      "epoch": 0.7383922539968476,
      "grad_norm": 1.3941423892974854,
      "learning_rate": 7.005268572194208e-05,
      "loss": 1.6014,
      "step": 4099
    },
    {
      "epoch": 0.7385723936050439,
      "grad_norm": 1.4492136240005493,
      "learning_rate": 7.003971218877043e-05,
      "loss": 1.5172,
      "step": 4100
    },
    {
      "epoch": 0.7387525332132403,
      "grad_norm": 1.1980843544006348,
      "learning_rate": 7.00267370480228e-05,
      "loss": 1.8891,
      "step": 4101
    },
    {
      "epoch": 0.7389326728214366,
      "grad_norm": 1.1812303066253662,
      "learning_rate": 7.00137603007401e-05,
      "loss": 1.8001,
      "step": 4102
    },
    {
      "epoch": 0.739112812429633,
      "grad_norm": 1.2779420614242554,
      "learning_rate": 7.00007819479633e-05,
      "loss": 2.0786,
      "step": 4103
    },
    {
      "epoch": 0.7392929520378293,
      "grad_norm": 1.2915014028549194,
      "learning_rate": 6.998780199073349e-05,
      "loss": 1.8119,
      "step": 4104
    },
    {
      "epoch": 0.7394730916460257,
      "grad_norm": 1.2572461366653442,
      "learning_rate": 6.997482043009196e-05,
      "loss": 1.9241,
      "step": 4105
    },
    {
      "epoch": 0.739653231254222,
      "grad_norm": 1.255854606628418,
      "learning_rate": 6.996183726708005e-05,
      "loss": 1.7558,
      "step": 4106
    },
    {
      "epoch": 0.7398333708624184,
      "grad_norm": 1.4942491054534912,
      "learning_rate": 6.994885250273929e-05,
      "loss": 1.8149,
      "step": 4107
    },
    {
      "epoch": 0.7400135104706147,
      "grad_norm": 1.391896367073059,
      "learning_rate": 6.99358661381113e-05,
      "loss": 2.0134,
      "step": 4108
    },
    {
      "epoch": 0.7401936500788111,
      "grad_norm": 1.651044487953186,
      "learning_rate": 6.992287817423784e-05,
      "loss": 2.124,
      "step": 4109
    },
    {
      "epoch": 0.7403737896870074,
      "grad_norm": 1.4316761493682861,
      "learning_rate": 6.99098886121608e-05,
      "loss": 1.5978,
      "step": 4110
    },
    {
      "epoch": 0.7405539292952038,
      "grad_norm": 1.2705094814300537,
      "learning_rate": 6.989689745292217e-05,
      "loss": 1.6258,
      "step": 4111
    },
    {
      "epoch": 0.7407340689034001,
      "grad_norm": 1.2856587171554565,
      "learning_rate": 6.988390469756415e-05,
      "loss": 1.6745,
      "step": 4112
    },
    {
      "epoch": 0.7409142085115965,
      "grad_norm": 1.1784669160842896,
      "learning_rate": 6.987091034712894e-05,
      "loss": 1.291,
      "step": 4113
    },
    {
      "epoch": 0.7410943481197928,
      "grad_norm": 1.2560909986495972,
      "learning_rate": 6.985791440265902e-05,
      "loss": 1.6296,
      "step": 4114
    },
    {
      "epoch": 0.7412744877279892,
      "grad_norm": 1.3841191530227661,
      "learning_rate": 6.984491686519685e-05,
      "loss": 1.694,
      "step": 4115
    },
    {
      "epoch": 0.7414546273361855,
      "grad_norm": 1.1775438785552979,
      "learning_rate": 6.983191773578513e-05,
      "loss": 1.5431,
      "step": 4116
    },
    {
      "epoch": 0.7416347669443819,
      "grad_norm": 1.1501076221466064,
      "learning_rate": 6.981891701546663e-05,
      "loss": 1.4605,
      "step": 4117
    },
    {
      "epoch": 0.7418149065525782,
      "grad_norm": 1.2753643989562988,
      "learning_rate": 6.980591470528425e-05,
      "loss": 1.5419,
      "step": 4118
    },
    {
      "epoch": 0.7419950461607746,
      "grad_norm": 1.362391710281372,
      "learning_rate": 6.979291080628105e-05,
      "loss": 1.7005,
      "step": 4119
    },
    {
      "epoch": 0.7421751857689709,
      "grad_norm": 1.2860534191131592,
      "learning_rate": 6.977990531950018e-05,
      "loss": 1.731,
      "step": 4120
    },
    {
      "epoch": 0.7423553253771673,
      "grad_norm": 1.203721046447754,
      "learning_rate": 6.976689824598494e-05,
      "loss": 1.5181,
      "step": 4121
    },
    {
      "epoch": 0.7425354649853636,
      "grad_norm": 1.4677066802978516,
      "learning_rate": 6.975388958677875e-05,
      "loss": 1.6474,
      "step": 4122
    },
    {
      "epoch": 0.74271560459356,
      "grad_norm": 1.301697850227356,
      "learning_rate": 6.974087934292517e-05,
      "loss": 1.5654,
      "step": 4123
    },
    {
      "epoch": 0.7428957442017564,
      "grad_norm": 1.2527819871902466,
      "learning_rate": 6.972786751546787e-05,
      "loss": 1.5692,
      "step": 4124
    },
    {
      "epoch": 0.7430758838099527,
      "grad_norm": 1.1882433891296387,
      "learning_rate": 6.971485410545062e-05,
      "loss": 1.41,
      "step": 4125
    },
    {
      "epoch": 0.7432560234181491,
      "grad_norm": 1.2155542373657227,
      "learning_rate": 6.970183911391739e-05,
      "loss": 1.238,
      "step": 4126
    },
    {
      "epoch": 0.7434361630263454,
      "grad_norm": 1.3560665845870972,
      "learning_rate": 6.968882254191223e-05,
      "loss": 1.6155,
      "step": 4127
    },
    {
      "epoch": 0.7436163026345418,
      "grad_norm": 1.3439843654632568,
      "learning_rate": 6.96758043904793e-05,
      "loss": 1.6426,
      "step": 4128
    },
    {
      "epoch": 0.7437964422427381,
      "grad_norm": 1.5166290998458862,
      "learning_rate": 6.966278466066294e-05,
      "loss": 1.6941,
      "step": 4129
    },
    {
      "epoch": 0.7439765818509345,
      "grad_norm": 1.2749342918395996,
      "learning_rate": 6.964976335350757e-05,
      "loss": 1.3667,
      "step": 4130
    },
    {
      "epoch": 0.7441567214591308,
      "grad_norm": 1.3304468393325806,
      "learning_rate": 6.963674047005774e-05,
      "loss": 1.5901,
      "step": 4131
    },
    {
      "epoch": 0.7443368610673272,
      "grad_norm": 1.2431001663208008,
      "learning_rate": 6.962371601135816e-05,
      "loss": 1.5014,
      "step": 4132
    },
    {
      "epoch": 0.7445170006755235,
      "grad_norm": 1.2925816774368286,
      "learning_rate": 6.961068997845366e-05,
      "loss": 1.596,
      "step": 4133
    },
    {
      "epoch": 0.7446971402837199,
      "grad_norm": 1.2649071216583252,
      "learning_rate": 6.959766237238913e-05,
      "loss": 1.5276,
      "step": 4134
    },
    {
      "epoch": 0.7448772798919162,
      "grad_norm": 1.4556392431259155,
      "learning_rate": 6.95846331942097e-05,
      "loss": 1.7455,
      "step": 4135
    },
    {
      "epoch": 0.7450574195001126,
      "grad_norm": 1.383287787437439,
      "learning_rate": 6.957160244496051e-05,
      "loss": 1.6727,
      "step": 4136
    },
    {
      "epoch": 0.7452375591083089,
      "grad_norm": 1.2267539501190186,
      "learning_rate": 6.955857012568692e-05,
      "loss": 1.3615,
      "step": 4137
    },
    {
      "epoch": 0.7454176987165053,
      "grad_norm": 1.2379580736160278,
      "learning_rate": 6.954553623743437e-05,
      "loss": 1.3082,
      "step": 4138
    },
    {
      "epoch": 0.7455978383247016,
      "grad_norm": 1.3733998537063599,
      "learning_rate": 6.95325007812484e-05,
      "loss": 1.5761,
      "step": 4139
    },
    {
      "epoch": 0.745777977932898,
      "grad_norm": 1.5266066789627075,
      "learning_rate": 6.951946375817474e-05,
      "loss": 1.8277,
      "step": 4140
    },
    {
      "epoch": 0.7459581175410943,
      "grad_norm": 1.3549997806549072,
      "learning_rate": 6.950642516925921e-05,
      "loss": 1.4367,
      "step": 4141
    },
    {
      "epoch": 0.7461382571492907,
      "grad_norm": 1.323388934135437,
      "learning_rate": 6.949338501554776e-05,
      "loss": 1.4384,
      "step": 4142
    },
    {
      "epoch": 0.746318396757487,
      "grad_norm": 1.2775802612304688,
      "learning_rate": 6.948034329808644e-05,
      "loss": 1.497,
      "step": 4143
    },
    {
      "epoch": 0.7464985363656834,
      "grad_norm": 1.405749797821045,
      "learning_rate": 6.946730001792148e-05,
      "loss": 1.5867,
      "step": 4144
    },
    {
      "epoch": 0.7466786759738797,
      "grad_norm": 1.3147780895233154,
      "learning_rate": 6.945425517609921e-05,
      "loss": 1.4402,
      "step": 4145
    },
    {
      "epoch": 0.7468588155820761,
      "grad_norm": 1.2919516563415527,
      "learning_rate": 6.944120877366604e-05,
      "loss": 1.5231,
      "step": 4146
    },
    {
      "epoch": 0.7470389551902724,
      "grad_norm": 1.4937280416488647,
      "learning_rate": 6.942816081166859e-05,
      "loss": 1.6328,
      "step": 4147
    },
    {
      "epoch": 0.7472190947984688,
      "grad_norm": 1.3586223125457764,
      "learning_rate": 6.941511129115352e-05,
      "loss": 1.457,
      "step": 4148
    },
    {
      "epoch": 0.7473992344066651,
      "grad_norm": 1.3802465200424194,
      "learning_rate": 6.94020602131677e-05,
      "loss": 1.5487,
      "step": 4149
    },
    {
      "epoch": 0.7475793740148615,
      "grad_norm": 1.2864220142364502,
      "learning_rate": 6.938900757875804e-05,
      "loss": 1.3871,
      "step": 4150
    },
    {
      "epoch": 0.7477595136230579,
      "grad_norm": 1.2333890199661255,
      "learning_rate": 6.937595338897167e-05,
      "loss": 1.9039,
      "step": 4151
    },
    {
      "epoch": 0.7479396532312542,
      "grad_norm": 1.222285270690918,
      "learning_rate": 6.936289764485574e-05,
      "loss": 1.9035,
      "step": 4152
    },
    {
      "epoch": 0.7481197928394506,
      "grad_norm": 1.3959789276123047,
      "learning_rate": 6.934984034745757e-05,
      "loss": 2.0104,
      "step": 4153
    },
    {
      "epoch": 0.7482999324476469,
      "grad_norm": 1.1939700841903687,
      "learning_rate": 6.933678149782465e-05,
      "loss": 1.8061,
      "step": 4154
    },
    {
      "epoch": 0.7484800720558433,
      "grad_norm": 1.3192189931869507,
      "learning_rate": 6.932372109700454e-05,
      "loss": 2.06,
      "step": 4155
    },
    {
      "epoch": 0.7486602116640396,
      "grad_norm": 1.5580344200134277,
      "learning_rate": 6.931065914604493e-05,
      "loss": 2.0652,
      "step": 4156
    },
    {
      "epoch": 0.748840351272236,
      "grad_norm": 1.3030911684036255,
      "learning_rate": 6.929759564599365e-05,
      "loss": 1.7373,
      "step": 4157
    },
    {
      "epoch": 0.7490204908804323,
      "grad_norm": 1.580299973487854,
      "learning_rate": 6.928453059789866e-05,
      "loss": 2.1873,
      "step": 4158
    },
    {
      "epoch": 0.7492006304886287,
      "grad_norm": 1.5764505863189697,
      "learning_rate": 6.927146400280801e-05,
      "loss": 1.7591,
      "step": 4159
    },
    {
      "epoch": 0.749380770096825,
      "grad_norm": 2.004701852798462,
      "learning_rate": 6.92583958617699e-05,
      "loss": 2.1502,
      "step": 4160
    },
    {
      "epoch": 0.7495609097050214,
      "grad_norm": 1.2407548427581787,
      "learning_rate": 6.924532617583266e-05,
      "loss": 1.5723,
      "step": 4161
    },
    {
      "epoch": 0.7497410493132177,
      "grad_norm": 1.4992525577545166,
      "learning_rate": 6.923225494604473e-05,
      "loss": 1.8719,
      "step": 4162
    },
    {
      "epoch": 0.7499211889214141,
      "grad_norm": 1.2989745140075684,
      "learning_rate": 6.921918217345467e-05,
      "loss": 1.7208,
      "step": 4163
    },
    {
      "epoch": 0.7501013285296104,
      "grad_norm": 1.203352928161621,
      "learning_rate": 6.920610785911117e-05,
      "loss": 1.4796,
      "step": 4164
    },
    {
      "epoch": 0.7502814681378068,
      "grad_norm": 1.3173727989196777,
      "learning_rate": 6.919303200406307e-05,
      "loss": 1.5884,
      "step": 4165
    },
    {
      "epoch": 0.7504616077460031,
      "grad_norm": 1.3721061944961548,
      "learning_rate": 6.917995460935928e-05,
      "loss": 1.8345,
      "step": 4166
    },
    {
      "epoch": 0.7506417473541995,
      "grad_norm": 1.2218767404556274,
      "learning_rate": 6.916687567604887e-05,
      "loss": 1.4724,
      "step": 4167
    },
    {
      "epoch": 0.7508218869623958,
      "grad_norm": 1.3902645111083984,
      "learning_rate": 6.915379520518103e-05,
      "loss": 1.7459,
      "step": 4168
    },
    {
      "epoch": 0.7510020265705922,
      "grad_norm": 1.3177253007888794,
      "learning_rate": 6.914071319780507e-05,
      "loss": 1.5862,
      "step": 4169
    },
    {
      "epoch": 0.7511821661787885,
      "grad_norm": 1.1674782037734985,
      "learning_rate": 6.912762965497043e-05,
      "loss": 1.5232,
      "step": 4170
    },
    {
      "epoch": 0.7513623057869849,
      "grad_norm": 1.3822455406188965,
      "learning_rate": 6.911454457772664e-05,
      "loss": 1.5746,
      "step": 4171
    },
    {
      "epoch": 0.7515424453951812,
      "grad_norm": 1.274601936340332,
      "learning_rate": 6.910145796712341e-05,
      "loss": 1.5619,
      "step": 4172
    },
    {
      "epoch": 0.7517225850033776,
      "grad_norm": 1.2890963554382324,
      "learning_rate": 6.908836982421052e-05,
      "loss": 1.5745,
      "step": 4173
    },
    {
      "epoch": 0.7519027246115739,
      "grad_norm": 1.3160080909729004,
      "learning_rate": 6.90752801500379e-05,
      "loss": 1.5005,
      "step": 4174
    },
    {
      "epoch": 0.7520828642197703,
      "grad_norm": 1.3277052640914917,
      "learning_rate": 6.90621889456556e-05,
      "loss": 1.4722,
      "step": 4175
    },
    {
      "epoch": 0.7522630038279666,
      "grad_norm": 1.3834278583526611,
      "learning_rate": 6.904909621211378e-05,
      "loss": 1.6741,
      "step": 4176
    },
    {
      "epoch": 0.752443143436163,
      "grad_norm": 1.2452369928359985,
      "learning_rate": 6.903600195046275e-05,
      "loss": 1.3404,
      "step": 4177
    },
    {
      "epoch": 0.7526232830443593,
      "grad_norm": 1.198986291885376,
      "learning_rate": 6.902290616175291e-05,
      "loss": 1.3517,
      "step": 4178
    },
    {
      "epoch": 0.7528034226525557,
      "grad_norm": 1.3502898216247559,
      "learning_rate": 6.90098088470348e-05,
      "loss": 1.5992,
      "step": 4179
    },
    {
      "epoch": 0.7529835622607521,
      "grad_norm": 1.3791543245315552,
      "learning_rate": 6.899671000735909e-05,
      "loss": 1.7495,
      "step": 4180
    },
    {
      "epoch": 0.7531637018689484,
      "grad_norm": 1.263975739479065,
      "learning_rate": 6.898360964377654e-05,
      "loss": 1.5498,
      "step": 4181
    },
    {
      "epoch": 0.7533438414771448,
      "grad_norm": 1.4190168380737305,
      "learning_rate": 6.89705077573381e-05,
      "loss": 1.8724,
      "step": 4182
    },
    {
      "epoch": 0.7535239810853411,
      "grad_norm": 1.3742786645889282,
      "learning_rate": 6.895740434909476e-05,
      "loss": 1.5981,
      "step": 4183
    },
    {
      "epoch": 0.7537041206935375,
      "grad_norm": 1.5329620838165283,
      "learning_rate": 6.894429942009767e-05,
      "loss": 1.745,
      "step": 4184
    },
    {
      "epoch": 0.7538842603017338,
      "grad_norm": 1.347888708114624,
      "learning_rate": 6.893119297139809e-05,
      "loss": 1.6484,
      "step": 4185
    },
    {
      "epoch": 0.7540643999099302,
      "grad_norm": 1.4926917552947998,
      "learning_rate": 6.891808500404747e-05,
      "loss": 1.7861,
      "step": 4186
    },
    {
      "epoch": 0.7542445395181265,
      "grad_norm": 1.3732911348342896,
      "learning_rate": 6.890497551909725e-05,
      "loss": 1.703,
      "step": 4187
    },
    {
      "epoch": 0.7544246791263229,
      "grad_norm": 1.3571734428405762,
      "learning_rate": 6.889186451759911e-05,
      "loss": 1.6648,
      "step": 4188
    },
    {
      "epoch": 0.7546048187345192,
      "grad_norm": 1.3788052797317505,
      "learning_rate": 6.887875200060483e-05,
      "loss": 1.3385,
      "step": 4189
    },
    {
      "epoch": 0.7547849583427156,
      "grad_norm": 1.2850556373596191,
      "learning_rate": 6.886563796916622e-05,
      "loss": 1.5291,
      "step": 4190
    },
    {
      "epoch": 0.7549650979509119,
      "grad_norm": 1.4293962717056274,
      "learning_rate": 6.885252242433534e-05,
      "loss": 1.7287,
      "step": 4191
    },
    {
      "epoch": 0.7551452375591083,
      "grad_norm": 1.4468601942062378,
      "learning_rate": 6.883940536716428e-05,
      "loss": 1.3396,
      "step": 4192
    },
    {
      "epoch": 0.7553253771673046,
      "grad_norm": 1.3453272581100464,
      "learning_rate": 6.882628679870532e-05,
      "loss": 1.549,
      "step": 4193
    },
    {
      "epoch": 0.755505516775501,
      "grad_norm": 1.3950121402740479,
      "learning_rate": 6.88131667200108e-05,
      "loss": 1.4155,
      "step": 4194
    },
    {
      "epoch": 0.7556856563836973,
      "grad_norm": 1.3736757040023804,
      "learning_rate": 6.88000451321332e-05,
      "loss": 1.6363,
      "step": 4195
    },
    {
      "epoch": 0.7558657959918937,
      "grad_norm": 1.5825588703155518,
      "learning_rate": 6.878692203612513e-05,
      "loss": 1.6647,
      "step": 4196
    },
    {
      "epoch": 0.75604593560009,
      "grad_norm": 1.3325579166412354,
      "learning_rate": 6.877379743303933e-05,
      "loss": 1.4773,
      "step": 4197
    },
    {
      "epoch": 0.7562260752082864,
      "grad_norm": 1.3661730289459229,
      "learning_rate": 6.876067132392866e-05,
      "loss": 1.4619,
      "step": 4198
    },
    {
      "epoch": 0.7564062148164827,
      "grad_norm": 1.414772391319275,
      "learning_rate": 6.874754370984606e-05,
      "loss": 1.5437,
      "step": 4199
    },
    {
      "epoch": 0.7565863544246791,
      "grad_norm": 1.3531521558761597,
      "learning_rate": 6.873441459184464e-05,
      "loss": 1.3423,
      "step": 4200
    },
    {
      "epoch": 0.7567664940328754,
      "grad_norm": 1.2674922943115234,
      "learning_rate": 6.87212839709776e-05,
      "loss": 2.0647,
      "step": 4201
    },
    {
      "epoch": 0.7569466336410718,
      "grad_norm": 1.2441178560256958,
      "learning_rate": 6.870815184829828e-05,
      "loss": 1.7841,
      "step": 4202
    },
    {
      "epoch": 0.7571267732492681,
      "grad_norm": 1.375637412071228,
      "learning_rate": 6.869501822486014e-05,
      "loss": 1.6896,
      "step": 4203
    },
    {
      "epoch": 0.7573069128574645,
      "grad_norm": 1.2367630004882812,
      "learning_rate": 6.868188310171673e-05,
      "loss": 2.0089,
      "step": 4204
    },
    {
      "epoch": 0.7574870524656608,
      "grad_norm": 1.1467173099517822,
      "learning_rate": 6.866874647992177e-05,
      "loss": 1.491,
      "step": 4205
    },
    {
      "epoch": 0.7576671920738572,
      "grad_norm": 1.4076244831085205,
      "learning_rate": 6.865560836052905e-05,
      "loss": 2.4381,
      "step": 4206
    },
    {
      "epoch": 0.7578473316820535,
      "grad_norm": 1.4978952407836914,
      "learning_rate": 6.864246874459252e-05,
      "loss": 2.082,
      "step": 4207
    },
    {
      "epoch": 0.7580274712902499,
      "grad_norm": 1.4958641529083252,
      "learning_rate": 6.862932763316623e-05,
      "loss": 2.0228,
      "step": 4208
    },
    {
      "epoch": 0.7582076108984463,
      "grad_norm": 1.7537269592285156,
      "learning_rate": 6.861618502730435e-05,
      "loss": 2.2037,
      "step": 4209
    },
    {
      "epoch": 0.7583877505066426,
      "grad_norm": 1.6714410781860352,
      "learning_rate": 6.860304092806119e-05,
      "loss": 1.8447,
      "step": 4210
    },
    {
      "epoch": 0.758567890114839,
      "grad_norm": 1.3466695547103882,
      "learning_rate": 6.858989533649112e-05,
      "loss": 1.6541,
      "step": 4211
    },
    {
      "epoch": 0.7587480297230353,
      "grad_norm": 1.2620398998260498,
      "learning_rate": 6.857674825364871e-05,
      "loss": 1.6686,
      "step": 4212
    },
    {
      "epoch": 0.7589281693312318,
      "grad_norm": 1.2540416717529297,
      "learning_rate": 6.856359968058861e-05,
      "loss": 1.6794,
      "step": 4213
    },
    {
      "epoch": 0.759108308939428,
      "grad_norm": 1.3351025581359863,
      "learning_rate": 6.855044961836557e-05,
      "loss": 1.543,
      "step": 4214
    },
    {
      "epoch": 0.7592884485476245,
      "grad_norm": 1.3714812994003296,
      "learning_rate": 6.85372980680345e-05,
      "loss": 1.5582,
      "step": 4215
    },
    {
      "epoch": 0.7594685881558207,
      "grad_norm": 1.3614323139190674,
      "learning_rate": 6.852414503065044e-05,
      "loss": 1.5234,
      "step": 4216
    },
    {
      "epoch": 0.7596487277640172,
      "grad_norm": 1.2694581747055054,
      "learning_rate": 6.851099050726847e-05,
      "loss": 1.4151,
      "step": 4217
    },
    {
      "epoch": 0.7598288673722134,
      "grad_norm": 1.311307430267334,
      "learning_rate": 6.849783449894384e-05,
      "loss": 1.5952,
      "step": 4218
    },
    {
      "epoch": 0.7600090069804099,
      "grad_norm": 1.3133302927017212,
      "learning_rate": 6.848467700673196e-05,
      "loss": 1.5335,
      "step": 4219
    },
    {
      "epoch": 0.7601891465886061,
      "grad_norm": 1.120577335357666,
      "learning_rate": 6.847151803168828e-05,
      "loss": 1.2536,
      "step": 4220
    },
    {
      "epoch": 0.7603692861968026,
      "grad_norm": 1.4306323528289795,
      "learning_rate": 6.845835757486841e-05,
      "loss": 1.7583,
      "step": 4221
    },
    {
      "epoch": 0.7605494258049988,
      "grad_norm": 1.3486709594726562,
      "learning_rate": 6.844519563732811e-05,
      "loss": 1.7007,
      "step": 4222
    },
    {
      "epoch": 0.7607295654131953,
      "grad_norm": 1.2146520614624023,
      "learning_rate": 6.84320322201232e-05,
      "loss": 1.7896,
      "step": 4223
    },
    {
      "epoch": 0.7609097050213915,
      "grad_norm": 1.3226094245910645,
      "learning_rate": 6.841886732430963e-05,
      "loss": 1.6607,
      "step": 4224
    },
    {
      "epoch": 0.761089844629588,
      "grad_norm": 1.2693378925323486,
      "learning_rate": 6.84057009509435e-05,
      "loss": 1.5153,
      "step": 4225
    },
    {
      "epoch": 0.7612699842377842,
      "grad_norm": 1.232830286026001,
      "learning_rate": 6.8392533101081e-05,
      "loss": 1.6261,
      "step": 4226
    },
    {
      "epoch": 0.7614501238459807,
      "grad_norm": 1.4661186933517456,
      "learning_rate": 6.837936377577847e-05,
      "loss": 1.8412,
      "step": 4227
    },
    {
      "epoch": 0.761630263454177,
      "grad_norm": 1.3075085878372192,
      "learning_rate": 6.83661929760923e-05,
      "loss": 1.5111,
      "step": 4228
    },
    {
      "epoch": 0.7618104030623734,
      "grad_norm": 1.337407112121582,
      "learning_rate": 6.835302070307909e-05,
      "loss": 1.7119,
      "step": 4229
    },
    {
      "epoch": 0.7619905426705696,
      "grad_norm": 1.382531762123108,
      "learning_rate": 6.833984695779551e-05,
      "loss": 1.6738,
      "step": 4230
    },
    {
      "epoch": 0.762170682278766,
      "grad_norm": 1.2840282917022705,
      "learning_rate": 6.832667174129834e-05,
      "loss": 1.6193,
      "step": 4231
    },
    {
      "epoch": 0.7623508218869623,
      "grad_norm": 1.369227647781372,
      "learning_rate": 6.831349505464448e-05,
      "loss": 1.626,
      "step": 4232
    },
    {
      "epoch": 0.7625309614951588,
      "grad_norm": 1.3168357610702515,
      "learning_rate": 6.830031689889097e-05,
      "loss": 1.6952,
      "step": 4233
    },
    {
      "epoch": 0.762711101103355,
      "grad_norm": 1.2343453168869019,
      "learning_rate": 6.828713727509494e-05,
      "loss": 1.504,
      "step": 4234
    },
    {
      "epoch": 0.7628912407115515,
      "grad_norm": 1.294103980064392,
      "learning_rate": 6.827395618431367e-05,
      "loss": 1.5967,
      "step": 4235
    },
    {
      "epoch": 0.7630713803197477,
      "grad_norm": 1.285144329071045,
      "learning_rate": 6.826077362760454e-05,
      "loss": 1.5652,
      "step": 4236
    },
    {
      "epoch": 0.7632515199279442,
      "grad_norm": 1.2821578979492188,
      "learning_rate": 6.824758960602504e-05,
      "loss": 1.3802,
      "step": 4237
    },
    {
      "epoch": 0.7634316595361406,
      "grad_norm": 1.5116546154022217,
      "learning_rate": 6.823440412063282e-05,
      "loss": 1.8018,
      "step": 4238
    },
    {
      "epoch": 0.7636117991443369,
      "grad_norm": 1.4021544456481934,
      "learning_rate": 6.822121717248554e-05,
      "loss": 1.6717,
      "step": 4239
    },
    {
      "epoch": 0.7637919387525333,
      "grad_norm": 1.3740514516830444,
      "learning_rate": 6.820802876264112e-05,
      "loss": 1.4027,
      "step": 4240
    },
    {
      "epoch": 0.7639720783607296,
      "grad_norm": 1.4727224111557007,
      "learning_rate": 6.819483889215749e-05,
      "loss": 1.468,
      "step": 4241
    },
    {
      "epoch": 0.764152217968926,
      "grad_norm": 1.3883235454559326,
      "learning_rate": 6.818164756209274e-05,
      "loss": 1.3633,
      "step": 4242
    },
    {
      "epoch": 0.7643323575771223,
      "grad_norm": 1.2938035726547241,
      "learning_rate": 6.81684547735051e-05,
      "loss": 1.4315,
      "step": 4243
    },
    {
      "epoch": 0.7645124971853187,
      "grad_norm": 1.5009466409683228,
      "learning_rate": 6.815526052745284e-05,
      "loss": 1.5758,
      "step": 4244
    },
    {
      "epoch": 0.764692636793515,
      "grad_norm": 1.509669303894043,
      "learning_rate": 6.814206482499445e-05,
      "loss": 1.6465,
      "step": 4245
    },
    {
      "epoch": 0.7648727764017114,
      "grad_norm": 1.4357390403747559,
      "learning_rate": 6.812886766718845e-05,
      "loss": 1.4487,
      "step": 4246
    },
    {
      "epoch": 0.7650529160099077,
      "grad_norm": 1.476682186126709,
      "learning_rate": 6.811566905509351e-05,
      "loss": 1.332,
      "step": 4247
    },
    {
      "epoch": 0.7652330556181041,
      "grad_norm": 1.4407345056533813,
      "learning_rate": 6.810246898976841e-05,
      "loss": 1.463,
      "step": 4248
    },
    {
      "epoch": 0.7654131952263004,
      "grad_norm": 1.3901293277740479,
      "learning_rate": 6.808926747227209e-05,
      "loss": 1.3918,
      "step": 4249
    },
    {
      "epoch": 0.7655933348344968,
      "grad_norm": 1.2813270092010498,
      "learning_rate": 6.807606450366352e-05,
      "loss": 1.2823,
      "step": 4250
    },
    {
      "epoch": 0.7657734744426931,
      "grad_norm": 1.3141233921051025,
      "learning_rate": 6.806286008500189e-05,
      "loss": 2.0169,
      "step": 4251
    },
    {
      "epoch": 0.7659536140508895,
      "grad_norm": 1.40625,
      "learning_rate": 6.80496542173464e-05,
      "loss": 2.0799,
      "step": 4252
    },
    {
      "epoch": 0.7661337536590858,
      "grad_norm": 1.2948371171951294,
      "learning_rate": 6.803644690175644e-05,
      "loss": 2.0991,
      "step": 4253
    },
    {
      "epoch": 0.7663138932672822,
      "grad_norm": 1.2399791479110718,
      "learning_rate": 6.802323813929151e-05,
      "loss": 1.9906,
      "step": 4254
    },
    {
      "epoch": 0.7664940328754785,
      "grad_norm": 1.303386926651001,
      "learning_rate": 6.801002793101119e-05,
      "loss": 2.0277,
      "step": 4255
    },
    {
      "epoch": 0.7666741724836749,
      "grad_norm": 1.3103928565979004,
      "learning_rate": 6.799681627797519e-05,
      "loss": 1.938,
      "step": 4256
    },
    {
      "epoch": 0.7668543120918712,
      "grad_norm": 1.369431734085083,
      "learning_rate": 6.798360318124336e-05,
      "loss": 1.9729,
      "step": 4257
    },
    {
      "epoch": 0.7670344517000676,
      "grad_norm": 1.257573127746582,
      "learning_rate": 6.797038864187564e-05,
      "loss": 1.7535,
      "step": 4258
    },
    {
      "epoch": 0.7672145913082639,
      "grad_norm": 1.5795159339904785,
      "learning_rate": 6.795717266093211e-05,
      "loss": 1.887,
      "step": 4259
    },
    {
      "epoch": 0.7673947309164603,
      "grad_norm": 1.7241616249084473,
      "learning_rate": 6.794395523947293e-05,
      "loss": 2.0207,
      "step": 4260
    },
    {
      "epoch": 0.7675748705246566,
      "grad_norm": 1.4644179344177246,
      "learning_rate": 6.79307363785584e-05,
      "loss": 1.8781,
      "step": 4261
    },
    {
      "epoch": 0.767755010132853,
      "grad_norm": 1.1703157424926758,
      "learning_rate": 6.791751607924892e-05,
      "loss": 1.5761,
      "step": 4262
    },
    {
      "epoch": 0.7679351497410493,
      "grad_norm": 1.4546563625335693,
      "learning_rate": 6.790429434260504e-05,
      "loss": 1.8715,
      "step": 4263
    },
    {
      "epoch": 0.7681152893492457,
      "grad_norm": 1.3338547945022583,
      "learning_rate": 6.789107116968738e-05,
      "loss": 1.624,
      "step": 4264
    },
    {
      "epoch": 0.7682954289574421,
      "grad_norm": 1.3146785497665405,
      "learning_rate": 6.787784656155671e-05,
      "loss": 1.5431,
      "step": 4265
    },
    {
      "epoch": 0.7684755685656384,
      "grad_norm": 1.4626420736312866,
      "learning_rate": 6.78646205192739e-05,
      "loss": 1.5782,
      "step": 4266
    },
    {
      "epoch": 0.7686557081738348,
      "grad_norm": 1.3771897554397583,
      "learning_rate": 6.785139304389993e-05,
      "loss": 1.4225,
      "step": 4267
    },
    {
      "epoch": 0.7688358477820311,
      "grad_norm": 1.2438147068023682,
      "learning_rate": 6.783816413649591e-05,
      "loss": 1.3842,
      "step": 4268
    },
    {
      "epoch": 0.7690159873902275,
      "grad_norm": 1.3158282041549683,
      "learning_rate": 6.782493379812302e-05,
      "loss": 1.5977,
      "step": 4269
    },
    {
      "epoch": 0.7691961269984238,
      "grad_norm": 1.3996762037277222,
      "learning_rate": 6.781170202984267e-05,
      "loss": 1.5792,
      "step": 4270
    },
    {
      "epoch": 0.7693762666066202,
      "grad_norm": 1.3123843669891357,
      "learning_rate": 6.779846883271624e-05,
      "loss": 1.4833,
      "step": 4271
    },
    {
      "epoch": 0.7695564062148165,
      "grad_norm": 1.3152434825897217,
      "learning_rate": 6.77852342078053e-05,
      "loss": 1.5138,
      "step": 4272
    },
    {
      "epoch": 0.7697365458230129,
      "grad_norm": 1.3313504457473755,
      "learning_rate": 6.777199815617154e-05,
      "loss": 1.5747,
      "step": 4273
    },
    {
      "epoch": 0.7699166854312092,
      "grad_norm": 1.3116765022277832,
      "learning_rate": 6.775876067887674e-05,
      "loss": 1.4184,
      "step": 4274
    },
    {
      "epoch": 0.7700968250394056,
      "grad_norm": 1.412165880203247,
      "learning_rate": 6.774552177698282e-05,
      "loss": 1.6018,
      "step": 4275
    },
    {
      "epoch": 0.7702769646476019,
      "grad_norm": 1.22885000705719,
      "learning_rate": 6.77322814515518e-05,
      "loss": 1.4306,
      "step": 4276
    },
    {
      "epoch": 0.7704571042557983,
      "grad_norm": 1.2525964975357056,
      "learning_rate": 6.771903970364577e-05,
      "loss": 1.5575,
      "step": 4277
    },
    {
      "epoch": 0.7706372438639946,
      "grad_norm": 1.3322268724441528,
      "learning_rate": 6.770579653432701e-05,
      "loss": 1.4084,
      "step": 4278
    },
    {
      "epoch": 0.770817383472191,
      "grad_norm": 1.3104209899902344,
      "learning_rate": 6.76925519446579e-05,
      "loss": 1.6766,
      "step": 4279
    },
    {
      "epoch": 0.7709975230803873,
      "grad_norm": 1.3410226106643677,
      "learning_rate": 6.767930593570086e-05,
      "loss": 1.3541,
      "step": 4280
    },
    {
      "epoch": 0.7711776626885837,
      "grad_norm": 1.3326008319854736,
      "learning_rate": 6.766605850851853e-05,
      "loss": 1.6574,
      "step": 4281
    },
    {
      "epoch": 0.77135780229678,
      "grad_norm": 1.4369840621948242,
      "learning_rate": 6.765280966417359e-05,
      "loss": 2.0069,
      "step": 4282
    },
    {
      "epoch": 0.7715379419049764,
      "grad_norm": 1.2933682203292847,
      "learning_rate": 6.763955940372883e-05,
      "loss": 1.5017,
      "step": 4283
    },
    {
      "epoch": 0.7717180815131727,
      "grad_norm": 1.5855871438980103,
      "learning_rate": 6.762630772824724e-05,
      "loss": 1.9075,
      "step": 4284
    },
    {
      "epoch": 0.7718982211213691,
      "grad_norm": 1.2598789930343628,
      "learning_rate": 6.761305463879181e-05,
      "loss": 1.395,
      "step": 4285
    },
    {
      "epoch": 0.7720783607295654,
      "grad_norm": 1.3627471923828125,
      "learning_rate": 6.759980013642573e-05,
      "loss": 1.5435,
      "step": 4286
    },
    {
      "epoch": 0.7722585003377618,
      "grad_norm": 1.2259958982467651,
      "learning_rate": 6.758654422221224e-05,
      "loss": 1.3809,
      "step": 4287
    },
    {
      "epoch": 0.7724386399459581,
      "grad_norm": 1.3748903274536133,
      "learning_rate": 6.757328689721476e-05,
      "loss": 1.3938,
      "step": 4288
    },
    {
      "epoch": 0.7726187795541545,
      "grad_norm": 1.361642599105835,
      "learning_rate": 6.756002816249676e-05,
      "loss": 1.5747,
      "step": 4289
    },
    {
      "epoch": 0.7727989191623508,
      "grad_norm": 1.4628759622573853,
      "learning_rate": 6.754676801912183e-05,
      "loss": 1.7951,
      "step": 4290
    },
    {
      "epoch": 0.7729790587705472,
      "grad_norm": 1.324305772781372,
      "learning_rate": 6.753350646815373e-05,
      "loss": 1.5718,
      "step": 4291
    },
    {
      "epoch": 0.7731591983787435,
      "grad_norm": 1.3084319829940796,
      "learning_rate": 6.752024351065629e-05,
      "loss": 1.5159,
      "step": 4292
    },
    {
      "epoch": 0.7733393379869399,
      "grad_norm": 1.4280668497085571,
      "learning_rate": 6.750697914769345e-05,
      "loss": 1.433,
      "step": 4293
    },
    {
      "epoch": 0.7735194775951363,
      "grad_norm": 1.4539705514907837,
      "learning_rate": 6.749371338032925e-05,
      "loss": 1.7837,
      "step": 4294
    },
    {
      "epoch": 0.7736996172033326,
      "grad_norm": 1.4157899618148804,
      "learning_rate": 6.74804462096279e-05,
      "loss": 1.4403,
      "step": 4295
    },
    {
      "epoch": 0.773879756811529,
      "grad_norm": 1.3355616331100464,
      "learning_rate": 6.746717763665366e-05,
      "loss": 1.4053,
      "step": 4296
    },
    {
      "epoch": 0.7740598964197253,
      "grad_norm": 1.419180989265442,
      "learning_rate": 6.745390766247096e-05,
      "loss": 1.4461,
      "step": 4297
    },
    {
      "epoch": 0.7742400360279217,
      "grad_norm": 1.342730164527893,
      "learning_rate": 6.744063628814428e-05,
      "loss": 1.55,
      "step": 4298
    },
    {
      "epoch": 0.774420175636118,
      "grad_norm": 1.462121605873108,
      "learning_rate": 6.742736351473823e-05,
      "loss": 1.4211,
      "step": 4299
    },
    {
      "epoch": 0.7746003152443144,
      "grad_norm": 1.4189155101776123,
      "learning_rate": 6.741408934331759e-05,
      "loss": 1.3856,
      "step": 4300
    },
    {
      "epoch": 0.7747804548525107,
      "grad_norm": 1.4132473468780518,
      "learning_rate": 6.740081377494715e-05,
      "loss": 2.1445,
      "step": 4301
    },
    {
      "epoch": 0.7749605944607071,
      "grad_norm": 1.2943170070648193,
      "learning_rate": 6.738753681069194e-05,
      "loss": 2.0852,
      "step": 4302
    },
    {
      "epoch": 0.7751407340689034,
      "grad_norm": 1.2062082290649414,
      "learning_rate": 6.737425845161701e-05,
      "loss": 1.9833,
      "step": 4303
    },
    {
      "epoch": 0.7753208736770998,
      "grad_norm": 1.3156678676605225,
      "learning_rate": 6.73609786987875e-05,
      "loss": 1.851,
      "step": 4304
    },
    {
      "epoch": 0.7755010132852961,
      "grad_norm": 1.290466070175171,
      "learning_rate": 6.734769755326875e-05,
      "loss": 1.7542,
      "step": 4305
    },
    {
      "epoch": 0.7756811528934925,
      "grad_norm": 1.2796297073364258,
      "learning_rate": 6.733441501612612e-05,
      "loss": 1.8778,
      "step": 4306
    },
    {
      "epoch": 0.7758612925016888,
      "grad_norm": 1.3409113883972168,
      "learning_rate": 6.73211310884252e-05,
      "loss": 1.6384,
      "step": 4307
    },
    {
      "epoch": 0.7760414321098852,
      "grad_norm": 1.4278721809387207,
      "learning_rate": 6.730784577123157e-05,
      "loss": 1.974,
      "step": 4308
    },
    {
      "epoch": 0.7762215717180815,
      "grad_norm": 1.5057724714279175,
      "learning_rate": 6.729455906561098e-05,
      "loss": 2.1269,
      "step": 4309
    },
    {
      "epoch": 0.7764017113262779,
      "grad_norm": 1.9017735719680786,
      "learning_rate": 6.728127097262929e-05,
      "loss": 2.1665,
      "step": 4310
    },
    {
      "epoch": 0.7765818509344742,
      "grad_norm": 1.3164198398590088,
      "learning_rate": 6.726798149335244e-05,
      "loss": 1.5831,
      "step": 4311
    },
    {
      "epoch": 0.7767619905426706,
      "grad_norm": 1.2864744663238525,
      "learning_rate": 6.725469062884655e-05,
      "loss": 1.6175,
      "step": 4312
    },
    {
      "epoch": 0.7769421301508669,
      "grad_norm": 1.3113449811935425,
      "learning_rate": 6.724139838017778e-05,
      "loss": 1.5805,
      "step": 4313
    },
    {
      "epoch": 0.7771222697590633,
      "grad_norm": 1.3403681516647339,
      "learning_rate": 6.722810474841242e-05,
      "loss": 1.6504,
      "step": 4314
    },
    {
      "epoch": 0.7773024093672596,
      "grad_norm": 1.253453254699707,
      "learning_rate": 6.721480973461688e-05,
      "loss": 1.5868,
      "step": 4315
    },
    {
      "epoch": 0.777482548975456,
      "grad_norm": 1.21878182888031,
      "learning_rate": 6.72015133398577e-05,
      "loss": 1.4491,
      "step": 4316
    },
    {
      "epoch": 0.7776626885836523,
      "grad_norm": 1.2276949882507324,
      "learning_rate": 6.718821556520151e-05,
      "loss": 1.4078,
      "step": 4317
    },
    {
      "epoch": 0.7778428281918487,
      "grad_norm": 1.2296260595321655,
      "learning_rate": 6.7174916411715e-05,
      "loss": 1.5385,
      "step": 4318
    },
    {
      "epoch": 0.778022967800045,
      "grad_norm": 1.1654185056686401,
      "learning_rate": 6.71616158804651e-05,
      "loss": 1.3349,
      "step": 4319
    },
    {
      "epoch": 0.7782031074082414,
      "grad_norm": 1.3187497854232788,
      "learning_rate": 6.714831397251873e-05,
      "loss": 1.4194,
      "step": 4320
    },
    {
      "epoch": 0.7783832470164377,
      "grad_norm": 1.3843474388122559,
      "learning_rate": 6.713501068894296e-05,
      "loss": 1.6823,
      "step": 4321
    },
    {
      "epoch": 0.7785633866246341,
      "grad_norm": 1.1842639446258545,
      "learning_rate": 6.712170603080497e-05,
      "loss": 1.4052,
      "step": 4322
    },
    {
      "epoch": 0.7787435262328305,
      "grad_norm": 1.280438780784607,
      "learning_rate": 6.710839999917207e-05,
      "loss": 1.4829,
      "step": 4323
    },
    {
      "epoch": 0.7789236658410268,
      "grad_norm": 1.2648844718933105,
      "learning_rate": 6.709509259511165e-05,
      "loss": 1.653,
      "step": 4324
    },
    {
      "epoch": 0.7791038054492232,
      "grad_norm": 1.4239405393600464,
      "learning_rate": 6.708178381969123e-05,
      "loss": 1.5412,
      "step": 4325
    },
    {
      "epoch": 0.7792839450574195,
      "grad_norm": 1.4355626106262207,
      "learning_rate": 6.706847367397843e-05,
      "loss": 1.5768,
      "step": 4326
    },
    {
      "epoch": 0.7794640846656159,
      "grad_norm": 1.4034491777420044,
      "learning_rate": 6.7055162159041e-05,
      "loss": 1.5767,
      "step": 4327
    },
    {
      "epoch": 0.7796442242738122,
      "grad_norm": 1.4894843101501465,
      "learning_rate": 6.704184927594675e-05,
      "loss": 1.6383,
      "step": 4328
    },
    {
      "epoch": 0.7798243638820086,
      "grad_norm": 1.4030094146728516,
      "learning_rate": 6.702853502576367e-05,
      "loss": 1.7633,
      "step": 4329
    },
    {
      "epoch": 0.7800045034902049,
      "grad_norm": 1.428145408630371,
      "learning_rate": 6.70152194095598e-05,
      "loss": 1.3597,
      "step": 4330
    },
    {
      "epoch": 0.7801846430984013,
      "grad_norm": 1.4601353406906128,
      "learning_rate": 6.70019024284033e-05,
      "loss": 1.8679,
      "step": 4331
    },
    {
      "epoch": 0.7803647827065976,
      "grad_norm": 1.2155267000198364,
      "learning_rate": 6.698858408336248e-05,
      "loss": 1.3697,
      "step": 4332
    },
    {
      "epoch": 0.780544922314794,
      "grad_norm": 1.322020173072815,
      "learning_rate": 6.697526437550571e-05,
      "loss": 1.641,
      "step": 4333
    },
    {
      "epoch": 0.7807250619229903,
      "grad_norm": 1.2564399242401123,
      "learning_rate": 6.696194330590151e-05,
      "loss": 1.4124,
      "step": 4334
    },
    {
      "epoch": 0.7809052015311867,
      "grad_norm": 1.3006970882415771,
      "learning_rate": 6.694862087561849e-05,
      "loss": 1.4655,
      "step": 4335
    },
    {
      "epoch": 0.781085341139383,
      "grad_norm": 1.2777646780014038,
      "learning_rate": 6.693529708572535e-05,
      "loss": 1.4395,
      "step": 4336
    },
    {
      "epoch": 0.7812654807475794,
      "grad_norm": 1.3497413396835327,
      "learning_rate": 6.692197193729092e-05,
      "loss": 1.5953,
      "step": 4337
    },
    {
      "epoch": 0.7814456203557757,
      "grad_norm": 1.4025439023971558,
      "learning_rate": 6.690864543138415e-05,
      "loss": 1.4995,
      "step": 4338
    },
    {
      "epoch": 0.7816257599639721,
      "grad_norm": 1.3810391426086426,
      "learning_rate": 6.689531756907408e-05,
      "loss": 1.7339,
      "step": 4339
    },
    {
      "epoch": 0.7818058995721684,
      "grad_norm": 1.2977656126022339,
      "learning_rate": 6.688198835142986e-05,
      "loss": 1.4848,
      "step": 4340
    },
    {
      "epoch": 0.7819860391803648,
      "grad_norm": 1.365507960319519,
      "learning_rate": 6.686865777952076e-05,
      "loss": 1.6372,
      "step": 4341
    },
    {
      "epoch": 0.7821661787885611,
      "grad_norm": 1.2283918857574463,
      "learning_rate": 6.685532585441615e-05,
      "loss": 1.4993,
      "step": 4342
    },
    {
      "epoch": 0.7823463183967575,
      "grad_norm": 1.3399330377578735,
      "learning_rate": 6.68419925771855e-05,
      "loss": 1.45,
      "step": 4343
    },
    {
      "epoch": 0.7825264580049538,
      "grad_norm": 1.4427807331085205,
      "learning_rate": 6.682865794889843e-05,
      "loss": 1.5582,
      "step": 4344
    },
    {
      "epoch": 0.7827065976131502,
      "grad_norm": 1.5503398180007935,
      "learning_rate": 6.68153219706246e-05,
      "loss": 1.6297,
      "step": 4345
    },
    {
      "epoch": 0.7828867372213465,
      "grad_norm": 1.4203894138336182,
      "learning_rate": 6.680198464343383e-05,
      "loss": 1.624,
      "step": 4346
    },
    {
      "epoch": 0.7830668768295429,
      "grad_norm": 1.4910606145858765,
      "learning_rate": 6.678864596839605e-05,
      "loss": 1.4961,
      "step": 4347
    },
    {
      "epoch": 0.7832470164377392,
      "grad_norm": 1.3974127769470215,
      "learning_rate": 6.677530594658126e-05,
      "loss": 1.4824,
      "step": 4348
    },
    {
      "epoch": 0.7834271560459356,
      "grad_norm": 1.389898419380188,
      "learning_rate": 6.67619645790596e-05,
      "loss": 1.3888,
      "step": 4349
    },
    {
      "epoch": 0.7836072956541319,
      "grad_norm": 1.4136215448379517,
      "learning_rate": 6.67486218669013e-05,
      "loss": 1.5896,
      "step": 4350
    },
    {
      "epoch": 0.7837874352623283,
      "grad_norm": 1.2181212902069092,
      "learning_rate": 6.673527781117672e-05,
      "loss": 2.1319,
      "step": 4351
    },
    {
      "epoch": 0.7839675748705247,
      "grad_norm": 1.2846685647964478,
      "learning_rate": 6.672193241295631e-05,
      "loss": 2.0701,
      "step": 4352
    },
    {
      "epoch": 0.784147714478721,
      "grad_norm": 1.307144045829773,
      "learning_rate": 6.67085856733106e-05,
      "loss": 2.1257,
      "step": 4353
    },
    {
      "epoch": 0.7843278540869174,
      "grad_norm": 1.1837550401687622,
      "learning_rate": 6.669523759331031e-05,
      "loss": 1.8665,
      "step": 4354
    },
    {
      "epoch": 0.7845079936951137,
      "grad_norm": 1.3024024963378906,
      "learning_rate": 6.668188817402617e-05,
      "loss": 2.0207,
      "step": 4355
    },
    {
      "epoch": 0.7846881333033101,
      "grad_norm": 1.3469128608703613,
      "learning_rate": 6.66685374165291e-05,
      "loss": 1.7316,
      "step": 4356
    },
    {
      "epoch": 0.7848682729115064,
      "grad_norm": 1.4680286645889282,
      "learning_rate": 6.66551853218901e-05,
      "loss": 1.9258,
      "step": 4357
    },
    {
      "epoch": 0.7850484125197028,
      "grad_norm": 1.6727474927902222,
      "learning_rate": 6.664183189118022e-05,
      "loss": 2.0131,
      "step": 4358
    },
    {
      "epoch": 0.7852285521278991,
      "grad_norm": 1.9772467613220215,
      "learning_rate": 6.66284771254707e-05,
      "loss": 2.3818,
      "step": 4359
    },
    {
      "epoch": 0.7854086917360955,
      "grad_norm": 1.3002800941467285,
      "learning_rate": 6.661512102583283e-05,
      "loss": 1.6019,
      "step": 4360
    },
    {
      "epoch": 0.7855888313442918,
      "grad_norm": 1.2263084650039673,
      "learning_rate": 6.660176359333806e-05,
      "loss": 1.701,
      "step": 4361
    },
    {
      "epoch": 0.7857689709524882,
      "grad_norm": 1.3445242643356323,
      "learning_rate": 6.658840482905791e-05,
      "loss": 1.4601,
      "step": 4362
    },
    {
      "epoch": 0.7859491105606845,
      "grad_norm": 1.2192333936691284,
      "learning_rate": 6.657504473406398e-05,
      "loss": 1.5035,
      "step": 4363
    },
    {
      "epoch": 0.7861292501688809,
      "grad_norm": 1.333082914352417,
      "learning_rate": 6.656168330942804e-05,
      "loss": 1.6493,
      "step": 4364
    },
    {
      "epoch": 0.7863093897770772,
      "grad_norm": 1.2969721555709839,
      "learning_rate": 6.654832055622195e-05,
      "loss": 1.4837,
      "step": 4365
    },
    {
      "epoch": 0.7864895293852736,
      "grad_norm": 1.3417179584503174,
      "learning_rate": 6.653495647551763e-05,
      "loss": 1.4244,
      "step": 4366
    },
    {
      "epoch": 0.7866696689934699,
      "grad_norm": 1.2683868408203125,
      "learning_rate": 6.652159106838714e-05,
      "loss": 1.5175,
      "step": 4367
    },
    {
      "epoch": 0.7868498086016663,
      "grad_norm": 1.3362749814987183,
      "learning_rate": 6.65082243359027e-05,
      "loss": 1.6481,
      "step": 4368
    },
    {
      "epoch": 0.7870299482098626,
      "grad_norm": 1.3211076259613037,
      "learning_rate": 6.649485627913652e-05,
      "loss": 1.4736,
      "step": 4369
    },
    {
      "epoch": 0.787210087818059,
      "grad_norm": 1.3011056184768677,
      "learning_rate": 6.648148689916101e-05,
      "loss": 1.4945,
      "step": 4370
    },
    {
      "epoch": 0.7873902274262553,
      "grad_norm": 1.336059808731079,
      "learning_rate": 6.646811619704863e-05,
      "loss": 1.4135,
      "step": 4371
    },
    {
      "epoch": 0.7875703670344517,
      "grad_norm": 1.3346209526062012,
      "learning_rate": 6.645474417387202e-05,
      "loss": 1.4704,
      "step": 4372
    },
    {
      "epoch": 0.787750506642648,
      "grad_norm": 1.3304309844970703,
      "learning_rate": 6.644137083070383e-05,
      "loss": 1.6287,
      "step": 4373
    },
    {
      "epoch": 0.7879306462508444,
      "grad_norm": 1.3033826351165771,
      "learning_rate": 6.64279961686169e-05,
      "loss": 1.6459,
      "step": 4374
    },
    {
      "epoch": 0.7881107858590407,
      "grad_norm": 1.3721983432769775,
      "learning_rate": 6.641462018868409e-05,
      "loss": 1.6047,
      "step": 4375
    },
    {
      "epoch": 0.7882909254672371,
      "grad_norm": 1.2553279399871826,
      "learning_rate": 6.640124289197845e-05,
      "loss": 1.3705,
      "step": 4376
    },
    {
      "epoch": 0.7884710650754334,
      "grad_norm": 1.3671807050704956,
      "learning_rate": 6.638786427957311e-05,
      "loss": 1.4814,
      "step": 4377
    },
    {
      "epoch": 0.7886512046836298,
      "grad_norm": 1.3198288679122925,
      "learning_rate": 6.637448435254128e-05,
      "loss": 1.5643,
      "step": 4378
    },
    {
      "epoch": 0.7888313442918262,
      "grad_norm": 1.3639233112335205,
      "learning_rate": 6.636110311195628e-05,
      "loss": 1.5882,
      "step": 4379
    },
    {
      "epoch": 0.7890114839000225,
      "grad_norm": 1.2690165042877197,
      "learning_rate": 6.634772055889156e-05,
      "loss": 1.5313,
      "step": 4380
    },
    {
      "epoch": 0.7891916235082189,
      "grad_norm": 1.244288682937622,
      "learning_rate": 6.633433669442065e-05,
      "loss": 1.4042,
      "step": 4381
    },
    {
      "epoch": 0.7893717631164152,
      "grad_norm": 1.3455307483673096,
      "learning_rate": 6.632095151961722e-05,
      "loss": 1.7174,
      "step": 4382
    },
    {
      "epoch": 0.7895519027246116,
      "grad_norm": 1.3777587413787842,
      "learning_rate": 6.6307565035555e-05,
      "loss": 1.6765,
      "step": 4383
    },
    {
      "epoch": 0.7897320423328079,
      "grad_norm": 1.2678066492080688,
      "learning_rate": 6.629417724330788e-05,
      "loss": 1.6305,
      "step": 4384
    },
    {
      "epoch": 0.7899121819410043,
      "grad_norm": 1.4724981784820557,
      "learning_rate": 6.628078814394976e-05,
      "loss": 1.6793,
      "step": 4385
    },
    {
      "epoch": 0.7900923215492006,
      "grad_norm": 1.2241454124450684,
      "learning_rate": 6.626739773855477e-05,
      "loss": 1.5104,
      "step": 4386
    },
    {
      "epoch": 0.790272461157397,
      "grad_norm": 1.3004575967788696,
      "learning_rate": 6.625400602819705e-05,
      "loss": 1.5002,
      "step": 4387
    },
    {
      "epoch": 0.7904526007655933,
      "grad_norm": 1.4709972143173218,
      "learning_rate": 6.624061301395088e-05,
      "loss": 1.935,
      "step": 4388
    },
    {
      "epoch": 0.7906327403737897,
      "grad_norm": 1.4223140478134155,
      "learning_rate": 6.622721869689065e-05,
      "loss": 1.5612,
      "step": 4389
    },
    {
      "epoch": 0.790812879981986,
      "grad_norm": 1.456362247467041,
      "learning_rate": 6.621382307809084e-05,
      "loss": 1.7898,
      "step": 4390
    },
    {
      "epoch": 0.7909930195901824,
      "grad_norm": 1.4476219415664673,
      "learning_rate": 6.620042615862604e-05,
      "loss": 1.7348,
      "step": 4391
    },
    {
      "epoch": 0.7911731591983787,
      "grad_norm": 1.3089605569839478,
      "learning_rate": 6.618702793957094e-05,
      "loss": 1.4495,
      "step": 4392
    },
    {
      "epoch": 0.7913532988065751,
      "grad_norm": 1.4579360485076904,
      "learning_rate": 6.617362842200034e-05,
      "loss": 1.7474,
      "step": 4393
    },
    {
      "epoch": 0.7915334384147714,
      "grad_norm": 1.3035341501235962,
      "learning_rate": 6.616022760698917e-05,
      "loss": 1.4203,
      "step": 4394
    },
    {
      "epoch": 0.7917135780229678,
      "grad_norm": 1.3709262609481812,
      "learning_rate": 6.614682549561239e-05,
      "loss": 1.5029,
      "step": 4395
    },
    {
      "epoch": 0.7918937176311641,
      "grad_norm": 1.169980525970459,
      "learning_rate": 6.613342208894515e-05,
      "loss": 1.197,
      "step": 4396
    },
    {
      "epoch": 0.7920738572393605,
      "grad_norm": 1.4016973972320557,
      "learning_rate": 6.612001738806261e-05,
      "loss": 1.5222,
      "step": 4397
    },
    {
      "epoch": 0.7922539968475568,
      "grad_norm": 1.3692858219146729,
      "learning_rate": 6.610661139404018e-05,
      "loss": 1.5008,
      "step": 4398
    },
    {
      "epoch": 0.7924341364557532,
      "grad_norm": 1.5063860416412354,
      "learning_rate": 6.609320410795319e-05,
      "loss": 1.5498,
      "step": 4399
    },
    {
      "epoch": 0.7926142760639495,
      "grad_norm": 1.3322750329971313,
      "learning_rate": 6.607979553087722e-05,
      "loss": 1.1885,
      "step": 4400
    },
    {
      "epoch": 0.7927944156721459,
      "grad_norm": 1.312860369682312,
      "learning_rate": 6.606638566388787e-05,
      "loss": 1.9776,
      "step": 4401
    },
    {
      "epoch": 0.7929745552803422,
      "grad_norm": 1.1298643350601196,
      "learning_rate": 6.605297450806091e-05,
      "loss": 1.5361,
      "step": 4402
    },
    {
      "epoch": 0.7931546948885386,
      "grad_norm": 1.2165077924728394,
      "learning_rate": 6.603956206447215e-05,
      "loss": 1.8651,
      "step": 4403
    },
    {
      "epoch": 0.7933348344967349,
      "grad_norm": 1.3561855554580688,
      "learning_rate": 6.602614833419753e-05,
      "loss": 2.1797,
      "step": 4404
    },
    {
      "epoch": 0.7935149741049313,
      "grad_norm": 1.2631462812423706,
      "learning_rate": 6.601273331831309e-05,
      "loss": 1.7613,
      "step": 4405
    },
    {
      "epoch": 0.7936951137131276,
      "grad_norm": 1.247037410736084,
      "learning_rate": 6.599931701789498e-05,
      "loss": 1.8524,
      "step": 4406
    },
    {
      "epoch": 0.793875253321324,
      "grad_norm": 1.4577915668487549,
      "learning_rate": 6.598589943401948e-05,
      "loss": 2.0792,
      "step": 4407
    },
    {
      "epoch": 0.7940553929295204,
      "grad_norm": 1.3760621547698975,
      "learning_rate": 6.597248056776288e-05,
      "loss": 1.9623,
      "step": 4408
    },
    {
      "epoch": 0.7942355325377167,
      "grad_norm": 1.5641895532608032,
      "learning_rate": 6.595906042020169e-05,
      "loss": 2.1334,
      "step": 4409
    },
    {
      "epoch": 0.7944156721459131,
      "grad_norm": 1.5525637865066528,
      "learning_rate": 6.594563899241244e-05,
      "loss": 2.1652,
      "step": 4410
    },
    {
      "epoch": 0.7945958117541094,
      "grad_norm": 1.8955398797988892,
      "learning_rate": 6.59322162854718e-05,
      "loss": 2.1922,
      "step": 4411
    },
    {
      "epoch": 0.7947759513623058,
      "grad_norm": 1.248736023902893,
      "learning_rate": 6.591879230045653e-05,
      "loss": 1.513,
      "step": 4412
    },
    {
      "epoch": 0.7949560909705021,
      "grad_norm": 1.3473514318466187,
      "learning_rate": 6.590536703844348e-05,
      "loss": 1.6981,
      "step": 4413
    },
    {
      "epoch": 0.7951362305786985,
      "grad_norm": 1.3260043859481812,
      "learning_rate": 6.589194050050964e-05,
      "loss": 1.5863,
      "step": 4414
    },
    {
      "epoch": 0.7953163701868948,
      "grad_norm": 1.4116684198379517,
      "learning_rate": 6.587851268773208e-05,
      "loss": 1.5412,
      "step": 4415
    },
    {
      "epoch": 0.7954965097950912,
      "grad_norm": 1.2888160943984985,
      "learning_rate": 6.586508360118798e-05,
      "loss": 1.5163,
      "step": 4416
    },
    {
      "epoch": 0.7956766494032875,
      "grad_norm": 1.4023644924163818,
      "learning_rate": 6.585165324195458e-05,
      "loss": 1.7653,
      "step": 4417
    },
    {
      "epoch": 0.7958567890114839,
      "grad_norm": 1.179885983467102,
      "learning_rate": 6.583822161110928e-05,
      "loss": 1.2653,
      "step": 4418
    },
    {
      "epoch": 0.7960369286196802,
      "grad_norm": 1.1852091550827026,
      "learning_rate": 6.582478870972956e-05,
      "loss": 1.4764,
      "step": 4419
    },
    {
      "epoch": 0.7962170682278766,
      "grad_norm": 1.2814651727676392,
      "learning_rate": 6.5811354538893e-05,
      "loss": 1.571,
      "step": 4420
    },
    {
      "epoch": 0.7963972078360729,
      "grad_norm": 1.3784024715423584,
      "learning_rate": 6.579791909967727e-05,
      "loss": 1.5953,
      "step": 4421
    },
    {
      "epoch": 0.7965773474442693,
      "grad_norm": 1.428084135055542,
      "learning_rate": 6.578448239316017e-05,
      "loss": 1.8014,
      "step": 4422
    },
    {
      "epoch": 0.7967574870524656,
      "grad_norm": 1.3210147619247437,
      "learning_rate": 6.577104442041959e-05,
      "loss": 1.6047,
      "step": 4423
    },
    {
      "epoch": 0.796937626660662,
      "grad_norm": 1.315963625907898,
      "learning_rate": 6.575760518253351e-05,
      "loss": 1.5545,
      "step": 4424
    },
    {
      "epoch": 0.7971177662688583,
      "grad_norm": 1.185929536819458,
      "learning_rate": 6.574416468058e-05,
      "loss": 1.4976,
      "step": 4425
    },
    {
      "epoch": 0.7972979058770547,
      "grad_norm": 1.3967103958129883,
      "learning_rate": 6.573072291563729e-05,
      "loss": 1.6291,
      "step": 4426
    },
    {
      "epoch": 0.797478045485251,
      "grad_norm": 1.2888225317001343,
      "learning_rate": 6.571727988878364e-05,
      "loss": 1.4863,
      "step": 4427
    },
    {
      "epoch": 0.7976581850934474,
      "grad_norm": 1.2711877822875977,
      "learning_rate": 6.570383560109745e-05,
      "loss": 1.5938,
      "step": 4428
    },
    {
      "epoch": 0.7978383247016437,
      "grad_norm": 1.3336187601089478,
      "learning_rate": 6.569039005365721e-05,
      "loss": 1.5705,
      "step": 4429
    },
    {
      "epoch": 0.7980184643098401,
      "grad_norm": 1.2123491764068604,
      "learning_rate": 6.567694324754153e-05,
      "loss": 1.5141,
      "step": 4430
    },
    {
      "epoch": 0.7981986039180364,
      "grad_norm": 1.2405171394348145,
      "learning_rate": 6.56634951838291e-05,
      "loss": 1.4064,
      "step": 4431
    },
    {
      "epoch": 0.7983787435262328,
      "grad_norm": 1.3566830158233643,
      "learning_rate": 6.565004586359872e-05,
      "loss": 1.7046,
      "step": 4432
    },
    {
      "epoch": 0.7985588831344291,
      "grad_norm": 1.2678134441375732,
      "learning_rate": 6.563659528792928e-05,
      "loss": 1.5103,
      "step": 4433
    },
    {
      "epoch": 0.7987390227426255,
      "grad_norm": 1.3577377796173096,
      "learning_rate": 6.562314345789977e-05,
      "loss": 1.6721,
      "step": 4434
    },
    {
      "epoch": 0.7989191623508218,
      "grad_norm": 1.4468982219696045,
      "learning_rate": 6.560969037458933e-05,
      "loss": 1.4349,
      "step": 4435
    },
    {
      "epoch": 0.7990993019590182,
      "grad_norm": 1.3974382877349854,
      "learning_rate": 6.559623603907709e-05,
      "loss": 1.6136,
      "step": 4436
    },
    {
      "epoch": 0.7992794415672146,
      "grad_norm": 1.2704823017120361,
      "learning_rate": 6.558278045244243e-05,
      "loss": 1.4007,
      "step": 4437
    },
    {
      "epoch": 0.7994595811754109,
      "grad_norm": 1.4970290660858154,
      "learning_rate": 6.55693236157647e-05,
      "loss": 1.5104,
      "step": 4438
    },
    {
      "epoch": 0.7996397207836073,
      "grad_norm": 1.4043169021606445,
      "learning_rate": 6.555586553012341e-05,
      "loss": 1.522,
      "step": 4439
    },
    {
      "epoch": 0.7998198603918036,
      "grad_norm": 1.3219150304794312,
      "learning_rate": 6.554240619659817e-05,
      "loss": 1.3923,
      "step": 4440
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.389901876449585,
      "learning_rate": 6.552894561626865e-05,
      "loss": 1.5099,
      "step": 4441
    },
    {
      "epoch": 0.8001801396081963,
      "grad_norm": 1.392430067062378,
      "learning_rate": 6.55154837902147e-05,
      "loss": 1.5431,
      "step": 4442
    },
    {
      "epoch": 0.8003602792163927,
      "grad_norm": 1.4318674802780151,
      "learning_rate": 6.550202071951622e-05,
      "loss": 1.853,
      "step": 4443
    },
    {
      "epoch": 0.800540418824589,
      "grad_norm": 1.4515800476074219,
      "learning_rate": 6.548855640525315e-05,
      "loss": 1.6175,
      "step": 4444
    },
    {
      "epoch": 0.8007205584327854,
      "grad_norm": 1.4249720573425293,
      "learning_rate": 6.547509084850564e-05,
      "loss": 1.7187,
      "step": 4445
    },
    {
      "epoch": 0.8009006980409817,
      "grad_norm": 1.5754344463348389,
      "learning_rate": 6.546162405035388e-05,
      "loss": 1.4609,
      "step": 4446
    },
    {
      "epoch": 0.8010808376491781,
      "grad_norm": 1.4884302616119385,
      "learning_rate": 6.544815601187818e-05,
      "loss": 1.536,
      "step": 4447
    },
    {
      "epoch": 0.8012609772573744,
      "grad_norm": 1.4362431764602661,
      "learning_rate": 6.543468673415892e-05,
      "loss": 1.5394,
      "step": 4448
    },
    {
      "epoch": 0.8014411168655708,
      "grad_norm": 1.4859565496444702,
      "learning_rate": 6.542121621827661e-05,
      "loss": 1.5504,
      "step": 4449
    },
    {
      "epoch": 0.8016212564737671,
      "grad_norm": 1.3708226680755615,
      "learning_rate": 6.540774446531183e-05,
      "loss": 1.3096,
      "step": 4450
    },
    {
      "epoch": 0.8018013960819635,
      "grad_norm": 1.3131113052368164,
      "learning_rate": 6.539427147634533e-05,
      "loss": 2.0142,
      "step": 4451
    },
    {
      "epoch": 0.8019815356901598,
      "grad_norm": 1.1867454051971436,
      "learning_rate": 6.538079725245785e-05,
      "loss": 1.9549,
      "step": 4452
    },
    {
      "epoch": 0.8021616752983562,
      "grad_norm": 1.275564193725586,
      "learning_rate": 6.53673217947303e-05,
      "loss": 2.028,
      "step": 4453
    },
    {
      "epoch": 0.8023418149065525,
      "grad_norm": 1.316580057144165,
      "learning_rate": 6.535384510424369e-05,
      "loss": 2.0151,
      "step": 4454
    },
    {
      "epoch": 0.802521954514749,
      "grad_norm": 1.3737013339996338,
      "learning_rate": 6.53403671820791e-05,
      "loss": 2.1012,
      "step": 4455
    },
    {
      "epoch": 0.8027020941229452,
      "grad_norm": 1.249172568321228,
      "learning_rate": 6.532688802931773e-05,
      "loss": 2.169,
      "step": 4456
    },
    {
      "epoch": 0.8028822337311416,
      "grad_norm": 1.3895149230957031,
      "learning_rate": 6.531340764704085e-05,
      "loss": 1.9085,
      "step": 4457
    },
    {
      "epoch": 0.8030623733393379,
      "grad_norm": 1.4911143779754639,
      "learning_rate": 6.529992603632989e-05,
      "loss": 2.0766,
      "step": 4458
    },
    {
      "epoch": 0.8032425129475343,
      "grad_norm": 1.6138501167297363,
      "learning_rate": 6.52864431982663e-05,
      "loss": 2.2572,
      "step": 4459
    },
    {
      "epoch": 0.8034226525557306,
      "grad_norm": 1.610435962677002,
      "learning_rate": 6.527295913393171e-05,
      "loss": 2.3306,
      "step": 4460
    },
    {
      "epoch": 0.803602792163927,
      "grad_norm": 1.9049159288406372,
      "learning_rate": 6.525947384440776e-05,
      "loss": 2.2983,
      "step": 4461
    },
    {
      "epoch": 0.8037829317721233,
      "grad_norm": 1.2327874898910522,
      "learning_rate": 6.524598733077625e-05,
      "loss": 1.618,
      "step": 4462
    },
    {
      "epoch": 0.8039630713803197,
      "grad_norm": 1.1832700967788696,
      "learning_rate": 6.523249959411906e-05,
      "loss": 1.4639,
      "step": 4463
    },
    {
      "epoch": 0.8041432109885162,
      "grad_norm": 1.218001365661621,
      "learning_rate": 6.521901063551817e-05,
      "loss": 1.4186,
      "step": 4464
    },
    {
      "epoch": 0.8043233505967124,
      "grad_norm": 1.575579285621643,
      "learning_rate": 6.520552045605566e-05,
      "loss": 1.6536,
      "step": 4465
    },
    {
      "epoch": 0.8045034902049089,
      "grad_norm": 1.3112882375717163,
      "learning_rate": 6.519202905681371e-05,
      "loss": 1.4742,
      "step": 4466
    },
    {
      "epoch": 0.8046836298131052,
      "grad_norm": 1.222264051437378,
      "learning_rate": 6.51785364388746e-05,
      "loss": 1.5827,
      "step": 4467
    },
    {
      "epoch": 0.8048637694213016,
      "grad_norm": 1.1691155433654785,
      "learning_rate": 6.516504260332067e-05,
      "loss": 1.5971,
      "step": 4468
    },
    {
      "epoch": 0.8050439090294979,
      "grad_norm": 1.3767268657684326,
      "learning_rate": 6.515154755123442e-05,
      "loss": 1.4404,
      "step": 4469
    },
    {
      "epoch": 0.8052240486376943,
      "grad_norm": 1.291096568107605,
      "learning_rate": 6.51380512836984e-05,
      "loss": 1.3579,
      "step": 4470
    },
    {
      "epoch": 0.8054041882458906,
      "grad_norm": 1.3260118961334229,
      "learning_rate": 6.512455380179528e-05,
      "loss": 1.5205,
      "step": 4471
    },
    {
      "epoch": 0.805584327854087,
      "grad_norm": 1.1601088047027588,
      "learning_rate": 6.511105510660782e-05,
      "loss": 1.4457,
      "step": 4472
    },
    {
      "epoch": 0.8057644674622833,
      "grad_norm": 1.3309887647628784,
      "learning_rate": 6.509755519921889e-05,
      "loss": 1.7781,
      "step": 4473
    },
    {
      "epoch": 0.8059446070704797,
      "grad_norm": 1.3190155029296875,
      "learning_rate": 6.508405408071141e-05,
      "loss": 1.5145,
      "step": 4474
    },
    {
      "epoch": 0.806124746678676,
      "grad_norm": 1.4214673042297363,
      "learning_rate": 6.50705517521685e-05,
      "loss": 1.5623,
      "step": 4475
    },
    {
      "epoch": 0.8063048862868724,
      "grad_norm": 1.2425867319107056,
      "learning_rate": 6.505704821467325e-05,
      "loss": 1.3369,
      "step": 4476
    },
    {
      "epoch": 0.8064850258950687,
      "grad_norm": 1.3489048480987549,
      "learning_rate": 6.504354346930892e-05,
      "loss": 1.6381,
      "step": 4477
    },
    {
      "epoch": 0.8066651655032651,
      "grad_norm": 1.3433719873428345,
      "learning_rate": 6.503003751715885e-05,
      "loss": 1.5532,
      "step": 4478
    },
    {
      "epoch": 0.8068453051114614,
      "grad_norm": 1.1665107011795044,
      "learning_rate": 6.501653035930651e-05,
      "loss": 1.2765,
      "step": 4479
    },
    {
      "epoch": 0.8070254447196578,
      "grad_norm": 1.2134331464767456,
      "learning_rate": 6.500302199683541e-05,
      "loss": 1.3399,
      "step": 4480
    },
    {
      "epoch": 0.807205584327854,
      "grad_norm": 1.3646905422210693,
      "learning_rate": 6.49895124308292e-05,
      "loss": 1.6784,
      "step": 4481
    },
    {
      "epoch": 0.8073857239360505,
      "grad_norm": 1.249129295349121,
      "learning_rate": 6.497600166237161e-05,
      "loss": 1.3701,
      "step": 4482
    },
    {
      "epoch": 0.8075658635442468,
      "grad_norm": 1.4498471021652222,
      "learning_rate": 6.496248969254643e-05,
      "loss": 1.7075,
      "step": 4483
    },
    {
      "epoch": 0.8077460031524432,
      "grad_norm": 1.4726353883743286,
      "learning_rate": 6.494897652243765e-05,
      "loss": 1.8316,
      "step": 4484
    },
    {
      "epoch": 0.8079261427606395,
      "grad_norm": 1.5187658071517944,
      "learning_rate": 6.493546215312922e-05,
      "loss": 1.6956,
      "step": 4485
    },
    {
      "epoch": 0.8081062823688359,
      "grad_norm": 1.5023746490478516,
      "learning_rate": 6.492194658570532e-05,
      "loss": 1.7556,
      "step": 4486
    },
    {
      "epoch": 0.8082864219770322,
      "grad_norm": 1.2612223625183105,
      "learning_rate": 6.490842982125013e-05,
      "loss": 1.3557,
      "step": 4487
    },
    {
      "epoch": 0.8084665615852286,
      "grad_norm": 1.339063048362732,
      "learning_rate": 6.489491186084796e-05,
      "loss": 1.4407,
      "step": 4488
    },
    {
      "epoch": 0.8086467011934249,
      "grad_norm": 1.4253846406936646,
      "learning_rate": 6.488139270558321e-05,
      "loss": 1.5354,
      "step": 4489
    },
    {
      "epoch": 0.8088268408016213,
      "grad_norm": 1.4491996765136719,
      "learning_rate": 6.486787235654039e-05,
      "loss": 1.6168,
      "step": 4490
    },
    {
      "epoch": 0.8090069804098176,
      "grad_norm": 1.3849133253097534,
      "learning_rate": 6.485435081480409e-05,
      "loss": 1.3404,
      "step": 4491
    },
    {
      "epoch": 0.809187120018014,
      "grad_norm": 1.5375938415527344,
      "learning_rate": 6.484082808145899e-05,
      "loss": 1.396,
      "step": 4492
    },
    {
      "epoch": 0.8093672596262104,
      "grad_norm": 1.4248143434524536,
      "learning_rate": 6.482730415758991e-05,
      "loss": 1.7538,
      "step": 4493
    },
    {
      "epoch": 0.8095473992344067,
      "grad_norm": 1.438437581062317,
      "learning_rate": 6.481377904428171e-05,
      "loss": 1.7606,
      "step": 4494
    },
    {
      "epoch": 0.8097275388426031,
      "grad_norm": 1.3876696825027466,
      "learning_rate": 6.480025274261936e-05,
      "loss": 1.5314,
      "step": 4495
    },
    {
      "epoch": 0.8099076784507994,
      "grad_norm": 1.3699356317520142,
      "learning_rate": 6.478672525368796e-05,
      "loss": 1.4884,
      "step": 4496
    },
    {
      "epoch": 0.8100878180589958,
      "grad_norm": 1.3716552257537842,
      "learning_rate": 6.477319657857263e-05,
      "loss": 1.681,
      "step": 4497
    },
    {
      "epoch": 0.8102679576671921,
      "grad_norm": 1.4728550910949707,
      "learning_rate": 6.47596667183587e-05,
      "loss": 1.6941,
      "step": 4498
    },
    {
      "epoch": 0.8104480972753885,
      "grad_norm": 1.3962680101394653,
      "learning_rate": 6.474613567413148e-05,
      "loss": 1.4986,
      "step": 4499
    },
    {
      "epoch": 0.8106282368835848,
      "grad_norm": 1.5372809171676636,
      "learning_rate": 6.473260344697645e-05,
      "loss": 1.7452,
      "step": 4500
    },
    {
      "epoch": 0.8108083764917812,
      "grad_norm": 1.1778926849365234,
      "learning_rate": 6.471907003797913e-05,
      "loss": 1.9501,
      "step": 4501
    },
    {
      "epoch": 0.8109885160999775,
      "grad_norm": 1.2058022022247314,
      "learning_rate": 6.470553544822519e-05,
      "loss": 1.6338,
      "step": 4502
    },
    {
      "epoch": 0.8111686557081739,
      "grad_norm": 1.358543038368225,
      "learning_rate": 6.469199967880035e-05,
      "loss": 2.0036,
      "step": 4503
    },
    {
      "epoch": 0.8113487953163702,
      "grad_norm": 1.2895220518112183,
      "learning_rate": 6.467846273079047e-05,
      "loss": 1.8362,
      "step": 4504
    },
    {
      "epoch": 0.8115289349245666,
      "grad_norm": 1.280074954032898,
      "learning_rate": 6.466492460528145e-05,
      "loss": 1.9504,
      "step": 4505
    },
    {
      "epoch": 0.8117090745327629,
      "grad_norm": 1.3310102224349976,
      "learning_rate": 6.465138530335932e-05,
      "loss": 1.8365,
      "step": 4506
    },
    {
      "epoch": 0.8118892141409593,
      "grad_norm": 1.3557909727096558,
      "learning_rate": 6.463784482611019e-05,
      "loss": 2.0871,
      "step": 4507
    },
    {
      "epoch": 0.8120693537491556,
      "grad_norm": 1.3947160243988037,
      "learning_rate": 6.462430317462029e-05,
      "loss": 1.6909,
      "step": 4508
    },
    {
      "epoch": 0.812249493357352,
      "grad_norm": 1.5817337036132812,
      "learning_rate": 6.461076034997591e-05,
      "loss": 1.828,
      "step": 4509
    },
    {
      "epoch": 0.8124296329655483,
      "grad_norm": 1.9269813299179077,
      "learning_rate": 6.459721635326346e-05,
      "loss": 2.0909,
      "step": 4510
    },
    {
      "epoch": 0.8126097725737447,
      "grad_norm": 1.4996401071548462,
      "learning_rate": 6.45836711855694e-05,
      "loss": 1.8779,
      "step": 4511
    },
    {
      "epoch": 0.812789912181941,
      "grad_norm": 1.3707562685012817,
      "learning_rate": 6.457012484798037e-05,
      "loss": 1.6666,
      "step": 4512
    },
    {
      "epoch": 0.8129700517901374,
      "grad_norm": 1.1545947790145874,
      "learning_rate": 6.4556577341583e-05,
      "loss": 1.401,
      "step": 4513
    },
    {
      "epoch": 0.8131501913983337,
      "grad_norm": 1.2503443956375122,
      "learning_rate": 6.454302866746411e-05,
      "loss": 1.5507,
      "step": 4514
    },
    {
      "epoch": 0.8133303310065301,
      "grad_norm": 1.2434924840927124,
      "learning_rate": 6.452947882671052e-05,
      "loss": 1.4838,
      "step": 4515
    },
    {
      "epoch": 0.8135104706147264,
      "grad_norm": 1.281829595565796,
      "learning_rate": 6.451592782040922e-05,
      "loss": 1.5137,
      "step": 4516
    },
    {
      "epoch": 0.8136906102229228,
      "grad_norm": 1.3889199495315552,
      "learning_rate": 6.450237564964727e-05,
      "loss": 1.4431,
      "step": 4517
    },
    {
      "epoch": 0.8138707498311191,
      "grad_norm": 1.4919050931930542,
      "learning_rate": 6.448882231551178e-05,
      "loss": 1.7569,
      "step": 4518
    },
    {
      "epoch": 0.8140508894393155,
      "grad_norm": 1.3197606801986694,
      "learning_rate": 6.447526781909006e-05,
      "loss": 1.4634,
      "step": 4519
    },
    {
      "epoch": 0.8142310290475118,
      "grad_norm": 1.3520443439483643,
      "learning_rate": 6.44617121614694e-05,
      "loss": 1.626,
      "step": 4520
    },
    {
      "epoch": 0.8144111686557082,
      "grad_norm": 1.4526621103286743,
      "learning_rate": 6.444815534373721e-05,
      "loss": 1.4709,
      "step": 4521
    },
    {
      "epoch": 0.8145913082639046,
      "grad_norm": 1.4153876304626465,
      "learning_rate": 6.443459736698105e-05,
      "loss": 1.4838,
      "step": 4522
    },
    {
      "epoch": 0.8147714478721009,
      "grad_norm": 1.320319414138794,
      "learning_rate": 6.442103823228852e-05,
      "loss": 1.4079,
      "step": 4523
    },
    {
      "epoch": 0.8149515874802973,
      "grad_norm": 1.3222471475601196,
      "learning_rate": 6.440747794074733e-05,
      "loss": 1.6301,
      "step": 4524
    },
    {
      "epoch": 0.8151317270884936,
      "grad_norm": 1.212831974029541,
      "learning_rate": 6.439391649344528e-05,
      "loss": 1.2774,
      "step": 4525
    },
    {
      "epoch": 0.81531186669669,
      "grad_norm": 1.3025630712509155,
      "learning_rate": 6.438035389147026e-05,
      "loss": 1.3928,
      "step": 4526
    },
    {
      "epoch": 0.8154920063048863,
      "grad_norm": 1.2681788206100464,
      "learning_rate": 6.436679013591023e-05,
      "loss": 1.3501,
      "step": 4527
    },
    {
      "epoch": 0.8156721459130827,
      "grad_norm": 1.411774754524231,
      "learning_rate": 6.435322522785332e-05,
      "loss": 1.6519,
      "step": 4528
    },
    {
      "epoch": 0.815852285521279,
      "grad_norm": 1.3411880731582642,
      "learning_rate": 6.433965916838766e-05,
      "loss": 1.6624,
      "step": 4529
    },
    {
      "epoch": 0.8160324251294754,
      "grad_norm": 1.4375041723251343,
      "learning_rate": 6.432609195860155e-05,
      "loss": 1.8248,
      "step": 4530
    },
    {
      "epoch": 0.8162125647376717,
      "grad_norm": 1.380291223526001,
      "learning_rate": 6.43125235995833e-05,
      "loss": 1.6202,
      "step": 4531
    },
    {
      "epoch": 0.8163927043458681,
      "grad_norm": 1.2709763050079346,
      "learning_rate": 6.429895409242139e-05,
      "loss": 1.5164,
      "step": 4532
    },
    {
      "epoch": 0.8165728439540644,
      "grad_norm": 1.4841139316558838,
      "learning_rate": 6.428538343820434e-05,
      "loss": 1.7763,
      "step": 4533
    },
    {
      "epoch": 0.8167529835622608,
      "grad_norm": 1.257549524307251,
      "learning_rate": 6.427181163802077e-05,
      "loss": 1.4864,
      "step": 4534
    },
    {
      "epoch": 0.8169331231704571,
      "grad_norm": 1.3358641862869263,
      "learning_rate": 6.425823869295945e-05,
      "loss": 1.7005,
      "step": 4535
    },
    {
      "epoch": 0.8171132627786535,
      "grad_norm": 1.4645570516586304,
      "learning_rate": 6.424466460410916e-05,
      "loss": 1.7505,
      "step": 4536
    },
    {
      "epoch": 0.8172934023868498,
      "grad_norm": 1.2917605638504028,
      "learning_rate": 6.423108937255881e-05,
      "loss": 1.5402,
      "step": 4537
    },
    {
      "epoch": 0.8174735419950462,
      "grad_norm": 1.4276437759399414,
      "learning_rate": 6.42175129993974e-05,
      "loss": 1.7942,
      "step": 4538
    },
    {
      "epoch": 0.8176536816032425,
      "grad_norm": 1.325857400894165,
      "learning_rate": 6.420393548571402e-05,
      "loss": 1.5461,
      "step": 4539
    },
    {
      "epoch": 0.8178338212114389,
      "grad_norm": 1.3716787099838257,
      "learning_rate": 6.419035683259786e-05,
      "loss": 1.4073,
      "step": 4540
    },
    {
      "epoch": 0.8180139608196352,
      "grad_norm": 1.3536577224731445,
      "learning_rate": 6.41767770411382e-05,
      "loss": 1.564,
      "step": 4541
    },
    {
      "epoch": 0.8181941004278316,
      "grad_norm": 1.3137818574905396,
      "learning_rate": 6.416319611242437e-05,
      "loss": 1.4814,
      "step": 4542
    },
    {
      "epoch": 0.8183742400360279,
      "grad_norm": 1.3973814249038696,
      "learning_rate": 6.414961404754584e-05,
      "loss": 1.6267,
      "step": 4543
    },
    {
      "epoch": 0.8185543796442243,
      "grad_norm": 1.3066155910491943,
      "learning_rate": 6.413603084759219e-05,
      "loss": 1.4406,
      "step": 4544
    },
    {
      "epoch": 0.8187345192524206,
      "grad_norm": 1.4038729667663574,
      "learning_rate": 6.412244651365302e-05,
      "loss": 1.6981,
      "step": 4545
    },
    {
      "epoch": 0.818914658860617,
      "grad_norm": 1.2938554286956787,
      "learning_rate": 6.410886104681806e-05,
      "loss": 1.3584,
      "step": 4546
    },
    {
      "epoch": 0.8190947984688133,
      "grad_norm": 1.4238848686218262,
      "learning_rate": 6.409527444817716e-05,
      "loss": 1.3944,
      "step": 4547
    },
    {
      "epoch": 0.8192749380770097,
      "grad_norm": 1.4574675559997559,
      "learning_rate": 6.40816867188202e-05,
      "loss": 1.6258,
      "step": 4548
    },
    {
      "epoch": 0.819455077685206,
      "grad_norm": 1.344557523727417,
      "learning_rate": 6.406809785983717e-05,
      "loss": 1.3529,
      "step": 4549
    },
    {
      "epoch": 0.8196352172934024,
      "grad_norm": 1.442880630493164,
      "learning_rate": 6.40545078723182e-05,
      "loss": 1.5632,
      "step": 4550
    },
    {
      "epoch": 0.8198153569015988,
      "grad_norm": 1.537803292274475,
      "learning_rate": 6.404091675735344e-05,
      "loss": 2.1995,
      "step": 4551
    },
    {
      "epoch": 0.8199954965097951,
      "grad_norm": 1.1940757036209106,
      "learning_rate": 6.402732451603322e-05,
      "loss": 1.9971,
      "step": 4552
    },
    {
      "epoch": 0.8201756361179915,
      "grad_norm": 1.6424672603607178,
      "learning_rate": 6.401373114944781e-05,
      "loss": 1.6392,
      "step": 4553
    },
    {
      "epoch": 0.8203557757261878,
      "grad_norm": 1.2726942300796509,
      "learning_rate": 6.400013665868774e-05,
      "loss": 1.7334,
      "step": 4554
    },
    {
      "epoch": 0.8205359153343842,
      "grad_norm": 1.3184664249420166,
      "learning_rate": 6.398654104484351e-05,
      "loss": 1.7732,
      "step": 4555
    },
    {
      "epoch": 0.8207160549425805,
      "grad_norm": 1.3863942623138428,
      "learning_rate": 6.397294430900578e-05,
      "loss": 1.7735,
      "step": 4556
    },
    {
      "epoch": 0.8208961945507769,
      "grad_norm": 1.3770524263381958,
      "learning_rate": 6.395934645226526e-05,
      "loss": 1.7526,
      "step": 4557
    },
    {
      "epoch": 0.8210763341589732,
      "grad_norm": 1.5593171119689941,
      "learning_rate": 6.394574747571277e-05,
      "loss": 2.1031,
      "step": 4558
    },
    {
      "epoch": 0.8212564737671696,
      "grad_norm": 1.5171849727630615,
      "learning_rate": 6.393214738043922e-05,
      "loss": 1.875,
      "step": 4559
    },
    {
      "epoch": 0.8214366133753659,
      "grad_norm": 1.7083256244659424,
      "learning_rate": 6.391854616753556e-05,
      "loss": 2.2154,
      "step": 4560
    },
    {
      "epoch": 0.8216167529835623,
      "grad_norm": 1.2111754417419434,
      "learning_rate": 6.390494383809293e-05,
      "loss": 1.3375,
      "step": 4561
    },
    {
      "epoch": 0.8217968925917586,
      "grad_norm": 1.3022698163986206,
      "learning_rate": 6.389134039320248e-05,
      "loss": 1.5758,
      "step": 4562
    },
    {
      "epoch": 0.821977032199955,
      "grad_norm": 1.5092780590057373,
      "learning_rate": 6.387773583395548e-05,
      "loss": 1.7321,
      "step": 4563
    },
    {
      "epoch": 0.8221571718081513,
      "grad_norm": 1.3205093145370483,
      "learning_rate": 6.386413016144324e-05,
      "loss": 1.4703,
      "step": 4564
    },
    {
      "epoch": 0.8223373114163477,
      "grad_norm": 1.437686562538147,
      "learning_rate": 6.385052337675725e-05,
      "loss": 1.6962,
      "step": 4565
    },
    {
      "epoch": 0.822517451024544,
      "grad_norm": 1.3500137329101562,
      "learning_rate": 6.383691548098902e-05,
      "loss": 1.7265,
      "step": 4566
    },
    {
      "epoch": 0.8226975906327404,
      "grad_norm": 1.4039119482040405,
      "learning_rate": 6.382330647523017e-05,
      "loss": 1.6105,
      "step": 4567
    },
    {
      "epoch": 0.8228777302409367,
      "grad_norm": 1.4360295534133911,
      "learning_rate": 6.380969636057241e-05,
      "loss": 1.6602,
      "step": 4568
    },
    {
      "epoch": 0.8230578698491331,
      "grad_norm": 1.3655681610107422,
      "learning_rate": 6.379608513810753e-05,
      "loss": 1.5239,
      "step": 4569
    },
    {
      "epoch": 0.8232380094573294,
      "grad_norm": 1.4226715564727783,
      "learning_rate": 6.378247280892742e-05,
      "loss": 1.6426,
      "step": 4570
    },
    {
      "epoch": 0.8234181490655258,
      "grad_norm": 1.2794270515441895,
      "learning_rate": 6.376885937412404e-05,
      "loss": 1.5989,
      "step": 4571
    },
    {
      "epoch": 0.8235982886737221,
      "grad_norm": 1.2326371669769287,
      "learning_rate": 6.37552448347895e-05,
      "loss": 1.2904,
      "step": 4572
    },
    {
      "epoch": 0.8237784282819185,
      "grad_norm": 1.397631287574768,
      "learning_rate": 6.374162919201591e-05,
      "loss": 1.6154,
      "step": 4573
    },
    {
      "epoch": 0.8239585678901148,
      "grad_norm": 1.369861364364624,
      "learning_rate": 6.372801244689551e-05,
      "loss": 1.6403,
      "step": 4574
    },
    {
      "epoch": 0.8241387074983112,
      "grad_norm": 1.3602043390274048,
      "learning_rate": 6.371439460052064e-05,
      "loss": 1.4651,
      "step": 4575
    },
    {
      "epoch": 0.8243188471065075,
      "grad_norm": 1.4389811754226685,
      "learning_rate": 6.370077565398371e-05,
      "loss": 1.4494,
      "step": 4576
    },
    {
      "epoch": 0.8244989867147039,
      "grad_norm": 1.3502315282821655,
      "learning_rate": 6.368715560837723e-05,
      "loss": 1.5568,
      "step": 4577
    },
    {
      "epoch": 0.8246791263229003,
      "grad_norm": 1.3633898496627808,
      "learning_rate": 6.36735344647938e-05,
      "loss": 1.7038,
      "step": 4578
    },
    {
      "epoch": 0.8248592659310966,
      "grad_norm": 1.2685598134994507,
      "learning_rate": 6.36599122243261e-05,
      "loss": 1.3985,
      "step": 4579
    },
    {
      "epoch": 0.825039405539293,
      "grad_norm": 1.3337403535842896,
      "learning_rate": 6.364628888806689e-05,
      "loss": 1.4886,
      "step": 4580
    },
    {
      "epoch": 0.8252195451474893,
      "grad_norm": 1.527099609375,
      "learning_rate": 6.363266445710903e-05,
      "loss": 1.5541,
      "step": 4581
    },
    {
      "epoch": 0.8253996847556857,
      "grad_norm": 1.592868685722351,
      "learning_rate": 6.361903893254548e-05,
      "loss": 1.3807,
      "step": 4582
    },
    {
      "epoch": 0.825579824363882,
      "grad_norm": 1.319865345954895,
      "learning_rate": 6.360541231546924e-05,
      "loss": 1.4804,
      "step": 4583
    },
    {
      "epoch": 0.8257599639720784,
      "grad_norm": 1.3298275470733643,
      "learning_rate": 6.359178460697346e-05,
      "loss": 1.4019,
      "step": 4584
    },
    {
      "epoch": 0.8259401035802747,
      "grad_norm": 1.3917150497436523,
      "learning_rate": 6.357815580815135e-05,
      "loss": 1.5151,
      "step": 4585
    },
    {
      "epoch": 0.8261202431884711,
      "grad_norm": 1.2952529191970825,
      "learning_rate": 6.356452592009619e-05,
      "loss": 1.5054,
      "step": 4586
    },
    {
      "epoch": 0.8263003827966674,
      "grad_norm": 1.3699872493743896,
      "learning_rate": 6.355089494390136e-05,
      "loss": 1.5985,
      "step": 4587
    },
    {
      "epoch": 0.8264805224048638,
      "grad_norm": 1.3411661386489868,
      "learning_rate": 6.353726288066033e-05,
      "loss": 1.3733,
      "step": 4588
    },
    {
      "epoch": 0.8266606620130601,
      "grad_norm": 1.4760723114013672,
      "learning_rate": 6.352362973146669e-05,
      "loss": 1.6383,
      "step": 4589
    },
    {
      "epoch": 0.8268408016212565,
      "grad_norm": 1.5501197576522827,
      "learning_rate": 6.350999549741404e-05,
      "loss": 1.7465,
      "step": 4590
    },
    {
      "epoch": 0.8270209412294528,
      "grad_norm": 1.340526819229126,
      "learning_rate": 6.349636017959615e-05,
      "loss": 1.5862,
      "step": 4591
    },
    {
      "epoch": 0.8272010808376492,
      "grad_norm": 1.433256983757019,
      "learning_rate": 6.34827237791068e-05,
      "loss": 1.631,
      "step": 4592
    },
    {
      "epoch": 0.8273812204458455,
      "grad_norm": 1.3091191053390503,
      "learning_rate": 6.346908629703992e-05,
      "loss": 1.4199,
      "step": 4593
    },
    {
      "epoch": 0.8275613600540419,
      "grad_norm": 1.3807443380355835,
      "learning_rate": 6.34554477344895e-05,
      "loss": 1.5952,
      "step": 4594
    },
    {
      "epoch": 0.8277414996622382,
      "grad_norm": 1.2758647203445435,
      "learning_rate": 6.344180809254963e-05,
      "loss": 1.373,
      "step": 4595
    },
    {
      "epoch": 0.8279216392704346,
      "grad_norm": 1.3240890502929688,
      "learning_rate": 6.342816737231446e-05,
      "loss": 1.4888,
      "step": 4596
    },
    {
      "epoch": 0.8281017788786309,
      "grad_norm": 1.3974522352218628,
      "learning_rate": 6.341452557487821e-05,
      "loss": 1.3543,
      "step": 4597
    },
    {
      "epoch": 0.8282819184868273,
      "grad_norm": 1.3768842220306396,
      "learning_rate": 6.340088270133528e-05,
      "loss": 1.5455,
      "step": 4598
    },
    {
      "epoch": 0.8284620580950236,
      "grad_norm": 1.2990660667419434,
      "learning_rate": 6.338723875278006e-05,
      "loss": 1.3839,
      "step": 4599
    },
    {
      "epoch": 0.82864219770322,
      "grad_norm": 1.3482491970062256,
      "learning_rate": 6.337359373030708e-05,
      "loss": 1.5414,
      "step": 4600
    },
    {
      "epoch": 0.8288223373114163,
      "grad_norm": 1.3061233758926392,
      "learning_rate": 6.335994763501094e-05,
      "loss": 2.0464,
      "step": 4601
    },
    {
      "epoch": 0.8290024769196127,
      "grad_norm": 1.2174551486968994,
      "learning_rate": 6.334630046798627e-05,
      "loss": 1.835,
      "step": 4602
    },
    {
      "epoch": 0.829182616527809,
      "grad_norm": 1.326425552368164,
      "learning_rate": 6.333265223032789e-05,
      "loss": 1.984,
      "step": 4603
    },
    {
      "epoch": 0.8293627561360054,
      "grad_norm": 1.3246726989746094,
      "learning_rate": 6.331900292313063e-05,
      "loss": 1.9577,
      "step": 4604
    },
    {
      "epoch": 0.8295428957442017,
      "grad_norm": 1.3895658254623413,
      "learning_rate": 6.330535254748946e-05,
      "loss": 2.085,
      "step": 4605
    },
    {
      "epoch": 0.8297230353523981,
      "grad_norm": 1.2714837789535522,
      "learning_rate": 6.32917011044994e-05,
      "loss": 1.5644,
      "step": 4606
    },
    {
      "epoch": 0.8299031749605945,
      "grad_norm": 1.3417433500289917,
      "learning_rate": 6.327804859525553e-05,
      "loss": 2.0656,
      "step": 4607
    },
    {
      "epoch": 0.8300833145687908,
      "grad_norm": 1.440290927886963,
      "learning_rate": 6.326439502085309e-05,
      "loss": 2.0938,
      "step": 4608
    },
    {
      "epoch": 0.8302634541769872,
      "grad_norm": 1.4658920764923096,
      "learning_rate": 6.325074038238732e-05,
      "loss": 2.0801,
      "step": 4609
    },
    {
      "epoch": 0.8304435937851835,
      "grad_norm": 1.7624857425689697,
      "learning_rate": 6.323708468095362e-05,
      "loss": 2.0864,
      "step": 4610
    },
    {
      "epoch": 0.8306237333933799,
      "grad_norm": 1.8614895343780518,
      "learning_rate": 6.322342791764744e-05,
      "loss": 2.0584,
      "step": 4611
    },
    {
      "epoch": 0.8308038730015762,
      "grad_norm": 1.2200448513031006,
      "learning_rate": 6.320977009356431e-05,
      "loss": 1.6163,
      "step": 4612
    },
    {
      "epoch": 0.8309840126097726,
      "grad_norm": 1.2457724809646606,
      "learning_rate": 6.319611120979984e-05,
      "loss": 1.4773,
      "step": 4613
    },
    {
      "epoch": 0.8311641522179689,
      "grad_norm": 1.3264724016189575,
      "learning_rate": 6.318245126744978e-05,
      "loss": 1.5159,
      "step": 4614
    },
    {
      "epoch": 0.8313442918261653,
      "grad_norm": 1.3391426801681519,
      "learning_rate": 6.316879026760989e-05,
      "loss": 1.6065,
      "step": 4615
    },
    {
      "epoch": 0.8315244314343616,
      "grad_norm": 1.300076961517334,
      "learning_rate": 6.315512821137606e-05,
      "loss": 1.4951,
      "step": 4616
    },
    {
      "epoch": 0.831704571042558,
      "grad_norm": 1.2662957906723022,
      "learning_rate": 6.314146509984427e-05,
      "loss": 1.5941,
      "step": 4617
    },
    {
      "epoch": 0.8318847106507543,
      "grad_norm": 1.310592770576477,
      "learning_rate": 6.312780093411053e-05,
      "loss": 1.6058,
      "step": 4618
    },
    {
      "epoch": 0.8320648502589507,
      "grad_norm": 1.329297423362732,
      "learning_rate": 6.311413571527099e-05,
      "loss": 1.414,
      "step": 4619
    },
    {
      "epoch": 0.832244989867147,
      "grad_norm": 1.2725532054901123,
      "learning_rate": 6.310046944442187e-05,
      "loss": 1.5976,
      "step": 4620
    },
    {
      "epoch": 0.8324251294753434,
      "grad_norm": 1.4283620119094849,
      "learning_rate": 6.30868021226595e-05,
      "loss": 1.492,
      "step": 4621
    },
    {
      "epoch": 0.8326052690835397,
      "grad_norm": 1.4487082958221436,
      "learning_rate": 6.307313375108021e-05,
      "loss": 1.8385,
      "step": 4622
    },
    {
      "epoch": 0.8327854086917361,
      "grad_norm": 1.2685261964797974,
      "learning_rate": 6.305946433078051e-05,
      "loss": 1.3864,
      "step": 4623
    },
    {
      "epoch": 0.8329655482999324,
      "grad_norm": 1.3122773170471191,
      "learning_rate": 6.304579386285694e-05,
      "loss": 1.4753,
      "step": 4624
    },
    {
      "epoch": 0.8331456879081288,
      "grad_norm": 1.407189965248108,
      "learning_rate": 6.303212234840615e-05,
      "loss": 1.7095,
      "step": 4625
    },
    {
      "epoch": 0.8333258275163251,
      "grad_norm": 1.3137348890304565,
      "learning_rate": 6.301844978852484e-05,
      "loss": 1.3032,
      "step": 4626
    },
    {
      "epoch": 0.8335059671245215,
      "grad_norm": 1.4981633424758911,
      "learning_rate": 6.300477618430983e-05,
      "loss": 1.9303,
      "step": 4627
    },
    {
      "epoch": 0.8336861067327178,
      "grad_norm": 1.3229765892028809,
      "learning_rate": 6.299110153685802e-05,
      "loss": 1.6261,
      "step": 4628
    },
    {
      "epoch": 0.8338662463409142,
      "grad_norm": 1.231129765510559,
      "learning_rate": 6.297742584726635e-05,
      "loss": 1.4318,
      "step": 4629
    },
    {
      "epoch": 0.8340463859491105,
      "grad_norm": 1.4217721223831177,
      "learning_rate": 6.296374911663191e-05,
      "loss": 1.5453,
      "step": 4630
    },
    {
      "epoch": 0.8342265255573069,
      "grad_norm": 1.2488220930099487,
      "learning_rate": 6.295007134605184e-05,
      "loss": 1.5537,
      "step": 4631
    },
    {
      "epoch": 0.8344066651655032,
      "grad_norm": 1.3453364372253418,
      "learning_rate": 6.293639253662333e-05,
      "loss": 1.5872,
      "step": 4632
    },
    {
      "epoch": 0.8345868047736996,
      "grad_norm": 1.4354774951934814,
      "learning_rate": 6.292271268944375e-05,
      "loss": 1.3761,
      "step": 4633
    },
    {
      "epoch": 0.8347669443818959,
      "grad_norm": 1.30548894405365,
      "learning_rate": 6.290903180561044e-05,
      "loss": 1.3112,
      "step": 4634
    },
    {
      "epoch": 0.8349470839900923,
      "grad_norm": 1.2789276838302612,
      "learning_rate": 6.289534988622087e-05,
      "loss": 1.3596,
      "step": 4635
    },
    {
      "epoch": 0.8351272235982887,
      "grad_norm": 1.4940729141235352,
      "learning_rate": 6.28816669323726e-05,
      "loss": 1.89,
      "step": 4636
    },
    {
      "epoch": 0.835307363206485,
      "grad_norm": 1.353440284729004,
      "learning_rate": 6.286798294516331e-05,
      "loss": 1.2853,
      "step": 4637
    },
    {
      "epoch": 0.8354875028146814,
      "grad_norm": 1.6403306722640991,
      "learning_rate": 6.28542979256907e-05,
      "loss": 1.7644,
      "step": 4638
    },
    {
      "epoch": 0.8356676424228777,
      "grad_norm": 1.4719386100769043,
      "learning_rate": 6.284061187505257e-05,
      "loss": 1.5504,
      "step": 4639
    },
    {
      "epoch": 0.8358477820310741,
      "grad_norm": 1.3217520713806152,
      "learning_rate": 6.28269247943468e-05,
      "loss": 1.4542,
      "step": 4640
    },
    {
      "epoch": 0.8360279216392704,
      "grad_norm": 1.378509521484375,
      "learning_rate": 6.281323668467136e-05,
      "loss": 1.5927,
      "step": 4641
    },
    {
      "epoch": 0.8362080612474668,
      "grad_norm": 1.2521916627883911,
      "learning_rate": 6.279954754712433e-05,
      "loss": 1.384,
      "step": 4642
    },
    {
      "epoch": 0.8363882008556631,
      "grad_norm": 1.5763472318649292,
      "learning_rate": 6.278585738280384e-05,
      "loss": 1.5358,
      "step": 4643
    },
    {
      "epoch": 0.8365683404638595,
      "grad_norm": 1.3223928213119507,
      "learning_rate": 6.277216619280808e-05,
      "loss": 1.4529,
      "step": 4644
    },
    {
      "epoch": 0.8367484800720558,
      "grad_norm": 1.3129265308380127,
      "learning_rate": 6.275847397823537e-05,
      "loss": 1.3774,
      "step": 4645
    },
    {
      "epoch": 0.8369286196802522,
      "grad_norm": 1.4142290353775024,
      "learning_rate": 6.27447807401841e-05,
      "loss": 1.5104,
      "step": 4646
    },
    {
      "epoch": 0.8371087592884485,
      "grad_norm": 1.4789080619812012,
      "learning_rate": 6.273108647975273e-05,
      "loss": 1.3418,
      "step": 4647
    },
    {
      "epoch": 0.8372888988966449,
      "grad_norm": 1.406337857246399,
      "learning_rate": 6.271739119803979e-05,
      "loss": 1.3727,
      "step": 4648
    },
    {
      "epoch": 0.8374690385048412,
      "grad_norm": 1.3417208194732666,
      "learning_rate": 6.270369489614394e-05,
      "loss": 1.3004,
      "step": 4649
    },
    {
      "epoch": 0.8376491781130376,
      "grad_norm": 1.2350142002105713,
      "learning_rate": 6.268999757516385e-05,
      "loss": 1.1123,
      "step": 4650
    },
    {
      "epoch": 0.8378293177212339,
      "grad_norm": 1.3327823877334595,
      "learning_rate": 6.267629923619835e-05,
      "loss": 2.035,
      "step": 4651
    },
    {
      "epoch": 0.8380094573294303,
      "grad_norm": 1.2755335569381714,
      "learning_rate": 6.26625998803463e-05,
      "loss": 1.9764,
      "step": 4652
    },
    {
      "epoch": 0.8381895969376266,
      "grad_norm": 1.1044981479644775,
      "learning_rate": 6.264889950870664e-05,
      "loss": 1.6571,
      "step": 4653
    },
    {
      "epoch": 0.838369736545823,
      "grad_norm": 1.3979949951171875,
      "learning_rate": 6.263519812237845e-05,
      "loss": 2.1235,
      "step": 4654
    },
    {
      "epoch": 0.8385498761540193,
      "grad_norm": 1.342162847518921,
      "learning_rate": 6.262149572246081e-05,
      "loss": 1.9552,
      "step": 4655
    },
    {
      "epoch": 0.8387300157622157,
      "grad_norm": 1.4222033023834229,
      "learning_rate": 6.260779231005293e-05,
      "loss": 2.1543,
      "step": 4656
    },
    {
      "epoch": 0.838910155370412,
      "grad_norm": 1.440486192703247,
      "learning_rate": 6.259408788625408e-05,
      "loss": 1.9623,
      "step": 4657
    },
    {
      "epoch": 0.8390902949786084,
      "grad_norm": 1.5896884202957153,
      "learning_rate": 6.258038245216365e-05,
      "loss": 2.1164,
      "step": 4658
    },
    {
      "epoch": 0.8392704345868047,
      "grad_norm": 1.6232459545135498,
      "learning_rate": 6.256667600888106e-05,
      "loss": 2.089,
      "step": 4659
    },
    {
      "epoch": 0.8394505741950011,
      "grad_norm": 1.4081693887710571,
      "learning_rate": 6.255296855750584e-05,
      "loss": 1.8755,
      "step": 4660
    },
    {
      "epoch": 0.8396307138031974,
      "grad_norm": 1.3497090339660645,
      "learning_rate": 6.253926009913761e-05,
      "loss": 1.7496,
      "step": 4661
    },
    {
      "epoch": 0.8398108534113938,
      "grad_norm": 1.423877239227295,
      "learning_rate": 6.252555063487602e-05,
      "loss": 1.8525,
      "step": 4662
    },
    {
      "epoch": 0.8399909930195902,
      "grad_norm": 1.276597261428833,
      "learning_rate": 6.251184016582088e-05,
      "loss": 1.819,
      "step": 4663
    },
    {
      "epoch": 0.8401711326277865,
      "grad_norm": 1.2613213062286377,
      "learning_rate": 6.249812869307198e-05,
      "loss": 1.569,
      "step": 4664
    },
    {
      "epoch": 0.8403512722359829,
      "grad_norm": 1.2881382703781128,
      "learning_rate": 6.248441621772933e-05,
      "loss": 1.491,
      "step": 4665
    },
    {
      "epoch": 0.8405314118441792,
      "grad_norm": 1.4041613340377808,
      "learning_rate": 6.247070274089288e-05,
      "loss": 1.5604,
      "step": 4666
    },
    {
      "epoch": 0.8407115514523756,
      "grad_norm": 1.2167034149169922,
      "learning_rate": 6.245698826366272e-05,
      "loss": 1.4768,
      "step": 4667
    },
    {
      "epoch": 0.8408916910605719,
      "grad_norm": 1.2529959678649902,
      "learning_rate": 6.244327278713904e-05,
      "loss": 1.2245,
      "step": 4668
    },
    {
      "epoch": 0.8410718306687683,
      "grad_norm": 1.3273762464523315,
      "learning_rate": 6.242955631242205e-05,
      "loss": 1.6981,
      "step": 4669
    },
    {
      "epoch": 0.8412519702769646,
      "grad_norm": 1.301316738128662,
      "learning_rate": 6.241583884061214e-05,
      "loss": 1.445,
      "step": 4670
    },
    {
      "epoch": 0.841432109885161,
      "grad_norm": 1.2993309497833252,
      "learning_rate": 6.240212037280966e-05,
      "loss": 1.5646,
      "step": 4671
    },
    {
      "epoch": 0.8416122494933573,
      "grad_norm": 1.2444121837615967,
      "learning_rate": 6.238840091011515e-05,
      "loss": 1.5712,
      "step": 4672
    },
    {
      "epoch": 0.8417923891015537,
      "grad_norm": 1.3461843729019165,
      "learning_rate": 6.237468045362915e-05,
      "loss": 1.5655,
      "step": 4673
    },
    {
      "epoch": 0.84197252870975,
      "grad_norm": 1.4032658338546753,
      "learning_rate": 6.236095900445229e-05,
      "loss": 1.5982,
      "step": 4674
    },
    {
      "epoch": 0.8421526683179464,
      "grad_norm": 1.3040753602981567,
      "learning_rate": 6.234723656368533e-05,
      "loss": 1.5227,
      "step": 4675
    },
    {
      "epoch": 0.8423328079261427,
      "grad_norm": 1.2950160503387451,
      "learning_rate": 6.233351313242907e-05,
      "loss": 1.532,
      "step": 4676
    },
    {
      "epoch": 0.8425129475343391,
      "grad_norm": 1.5120325088500977,
      "learning_rate": 6.23197887117844e-05,
      "loss": 1.7143,
      "step": 4677
    },
    {
      "epoch": 0.8426930871425354,
      "grad_norm": 1.3828502893447876,
      "learning_rate": 6.230606330285227e-05,
      "loss": 1.4519,
      "step": 4678
    },
    {
      "epoch": 0.8428732267507318,
      "grad_norm": 1.1508655548095703,
      "learning_rate": 6.229233690673375e-05,
      "loss": 1.369,
      "step": 4679
    },
    {
      "epoch": 0.8430533663589281,
      "grad_norm": 1.3323533535003662,
      "learning_rate": 6.227860952452993e-05,
      "loss": 1.5739,
      "step": 4680
    },
    {
      "epoch": 0.8432335059671245,
      "grad_norm": 1.2645061016082764,
      "learning_rate": 6.226488115734204e-05,
      "loss": 1.2532,
      "step": 4681
    },
    {
      "epoch": 0.8434136455753208,
      "grad_norm": 1.4220880270004272,
      "learning_rate": 6.225115180627137e-05,
      "loss": 1.5388,
      "step": 4682
    },
    {
      "epoch": 0.8435937851835172,
      "grad_norm": 1.3079980611801147,
      "learning_rate": 6.223742147241926e-05,
      "loss": 1.5281,
      "step": 4683
    },
    {
      "epoch": 0.8437739247917135,
      "grad_norm": 1.3803733587265015,
      "learning_rate": 6.222369015688716e-05,
      "loss": 1.5258,
      "step": 4684
    },
    {
      "epoch": 0.8439540643999099,
      "grad_norm": 1.4696791172027588,
      "learning_rate": 6.220995786077658e-05,
      "loss": 1.6131,
      "step": 4685
    },
    {
      "epoch": 0.8441342040081062,
      "grad_norm": 1.318091630935669,
      "learning_rate": 6.219622458518915e-05,
      "loss": 1.3615,
      "step": 4686
    },
    {
      "epoch": 0.8443143436163026,
      "grad_norm": 1.4679065942764282,
      "learning_rate": 6.21824903312265e-05,
      "loss": 1.6684,
      "step": 4687
    },
    {
      "epoch": 0.8444944832244989,
      "grad_norm": 1.5002509355545044,
      "learning_rate": 6.216875509999043e-05,
      "loss": 1.6468,
      "step": 4688
    },
    {
      "epoch": 0.8446746228326953,
      "grad_norm": 1.4470481872558594,
      "learning_rate": 6.215501889258275e-05,
      "loss": 1.5381,
      "step": 4689
    },
    {
      "epoch": 0.8448547624408916,
      "grad_norm": 1.3229215145111084,
      "learning_rate": 6.214128171010536e-05,
      "loss": 1.4441,
      "step": 4690
    },
    {
      "epoch": 0.845034902049088,
      "grad_norm": 1.3469367027282715,
      "learning_rate": 6.212754355366029e-05,
      "loss": 1.4857,
      "step": 4691
    },
    {
      "epoch": 0.8452150416572844,
      "grad_norm": 1.3363783359527588,
      "learning_rate": 6.211380442434957e-05,
      "loss": 1.4232,
      "step": 4692
    },
    {
      "epoch": 0.8453951812654807,
      "grad_norm": 1.422350287437439,
      "learning_rate": 6.210006432327535e-05,
      "loss": 1.501,
      "step": 4693
    },
    {
      "epoch": 0.8455753208736771,
      "grad_norm": 1.4966259002685547,
      "learning_rate": 6.208632325153988e-05,
      "loss": 1.5585,
      "step": 4694
    },
    {
      "epoch": 0.8457554604818734,
      "grad_norm": 1.3642261028289795,
      "learning_rate": 6.207258121024543e-05,
      "loss": 1.4925,
      "step": 4695
    },
    {
      "epoch": 0.8459356000900699,
      "grad_norm": 1.379480242729187,
      "learning_rate": 6.205883820049441e-05,
      "loss": 1.413,
      "step": 4696
    },
    {
      "epoch": 0.8461157396982661,
      "grad_norm": 1.2150439023971558,
      "learning_rate": 6.204509422338924e-05,
      "loss": 1.3691,
      "step": 4697
    },
    {
      "epoch": 0.8462958793064626,
      "grad_norm": 1.382617712020874,
      "learning_rate": 6.203134928003251e-05,
      "loss": 1.3898,
      "step": 4698
    },
    {
      "epoch": 0.8464760189146588,
      "grad_norm": 1.2476613521575928,
      "learning_rate": 6.201760337152677e-05,
      "loss": 1.4084,
      "step": 4699
    },
    {
      "epoch": 0.8466561585228553,
      "grad_norm": 1.4892241954803467,
      "learning_rate": 6.200385649897477e-05,
      "loss": 1.8027,
      "step": 4700
    },
    {
      "epoch": 0.8468362981310515,
      "grad_norm": 1.2168999910354614,
      "learning_rate": 6.199010866347922e-05,
      "loss": 1.9943,
      "step": 4701
    },
    {
      "epoch": 0.847016437739248,
      "grad_norm": 1.3281121253967285,
      "learning_rate": 6.197635986614302e-05,
      "loss": 1.9106,
      "step": 4702
    },
    {
      "epoch": 0.8471965773474442,
      "grad_norm": 1.2331305742263794,
      "learning_rate": 6.196261010806905e-05,
      "loss": 1.8294,
      "step": 4703
    },
    {
      "epoch": 0.8473767169556407,
      "grad_norm": 1.5338815450668335,
      "learning_rate": 6.194885939036033e-05,
      "loss": 1.6992,
      "step": 4704
    },
    {
      "epoch": 0.847556856563837,
      "grad_norm": 1.3370227813720703,
      "learning_rate": 6.193510771411993e-05,
      "loss": 1.9032,
      "step": 4705
    },
    {
      "epoch": 0.8477369961720334,
      "grad_norm": 1.3217793703079224,
      "learning_rate": 6.1921355080451e-05,
      "loss": 2.1385,
      "step": 4706
    },
    {
      "epoch": 0.8479171357802296,
      "grad_norm": 1.5505423545837402,
      "learning_rate": 6.190760149045677e-05,
      "loss": 2.1535,
      "step": 4707
    },
    {
      "epoch": 0.848097275388426,
      "grad_norm": 1.4694255590438843,
      "learning_rate": 6.189384694524056e-05,
      "loss": 1.8964,
      "step": 4708
    },
    {
      "epoch": 0.8482774149966223,
      "grad_norm": 1.6660085916519165,
      "learning_rate": 6.188009144590574e-05,
      "loss": 2.0454,
      "step": 4709
    },
    {
      "epoch": 0.8484575546048188,
      "grad_norm": 1.3536443710327148,
      "learning_rate": 6.186633499355576e-05,
      "loss": 1.5905,
      "step": 4710
    },
    {
      "epoch": 0.848637694213015,
      "grad_norm": 1.2272223234176636,
      "learning_rate": 6.185257758929415e-05,
      "loss": 1.5948,
      "step": 4711
    },
    {
      "epoch": 0.8488178338212115,
      "grad_norm": 1.3076916933059692,
      "learning_rate": 6.183881923422457e-05,
      "loss": 1.675,
      "step": 4712
    },
    {
      "epoch": 0.8489979734294077,
      "grad_norm": 1.2455263137817383,
      "learning_rate": 6.182505992945064e-05,
      "loss": 1.564,
      "step": 4713
    },
    {
      "epoch": 0.8491781130376042,
      "grad_norm": 1.254154086112976,
      "learning_rate": 6.181129967607621e-05,
      "loss": 1.3722,
      "step": 4714
    },
    {
      "epoch": 0.8493582526458004,
      "grad_norm": 1.2478543519973755,
      "learning_rate": 6.179753847520506e-05,
      "loss": 1.5416,
      "step": 4715
    },
    {
      "epoch": 0.8495383922539969,
      "grad_norm": 1.4882732629776,
      "learning_rate": 6.17837763279411e-05,
      "loss": 1.6731,
      "step": 4716
    },
    {
      "epoch": 0.8497185318621931,
      "grad_norm": 1.309693694114685,
      "learning_rate": 6.177001323538836e-05,
      "loss": 1.4801,
      "step": 4717
    },
    {
      "epoch": 0.8498986714703896,
      "grad_norm": 1.261641263961792,
      "learning_rate": 6.175624919865087e-05,
      "loss": 1.6143,
      "step": 4718
    },
    {
      "epoch": 0.8500788110785858,
      "grad_norm": 1.4516394138336182,
      "learning_rate": 6.174248421883282e-05,
      "loss": 1.5207,
      "step": 4719
    },
    {
      "epoch": 0.8502589506867823,
      "grad_norm": 1.4540899991989136,
      "learning_rate": 6.172871829703839e-05,
      "loss": 1.6793,
      "step": 4720
    },
    {
      "epoch": 0.8504390902949787,
      "grad_norm": 1.2702233791351318,
      "learning_rate": 6.17149514343719e-05,
      "loss": 1.3969,
      "step": 4721
    },
    {
      "epoch": 0.850619229903175,
      "grad_norm": 1.4019607305526733,
      "learning_rate": 6.170118363193772e-05,
      "loss": 1.8609,
      "step": 4722
    },
    {
      "epoch": 0.8507993695113714,
      "grad_norm": 1.2587826251983643,
      "learning_rate": 6.168741489084026e-05,
      "loss": 1.3129,
      "step": 4723
    },
    {
      "epoch": 0.8509795091195677,
      "grad_norm": 1.452796459197998,
      "learning_rate": 6.16736452121841e-05,
      "loss": 1.5949,
      "step": 4724
    },
    {
      "epoch": 0.8511596487277641,
      "grad_norm": 1.39859139919281,
      "learning_rate": 6.16598745970738e-05,
      "loss": 1.6331,
      "step": 4725
    },
    {
      "epoch": 0.8513397883359604,
      "grad_norm": 1.2477107048034668,
      "learning_rate": 6.164610304661403e-05,
      "loss": 1.4915,
      "step": 4726
    },
    {
      "epoch": 0.8515199279441568,
      "grad_norm": 1.4244507551193237,
      "learning_rate": 6.163233056190955e-05,
      "loss": 1.7,
      "step": 4727
    },
    {
      "epoch": 0.8517000675523531,
      "grad_norm": 1.3881311416625977,
      "learning_rate": 6.161855714406517e-05,
      "loss": 1.4208,
      "step": 4728
    },
    {
      "epoch": 0.8518802071605495,
      "grad_norm": 1.4379562139511108,
      "learning_rate": 6.16047827941858e-05,
      "loss": 1.6672,
      "step": 4729
    },
    {
      "epoch": 0.8520603467687458,
      "grad_norm": 1.3581207990646362,
      "learning_rate": 6.159100751337642e-05,
      "loss": 1.8151,
      "step": 4730
    },
    {
      "epoch": 0.8522404863769422,
      "grad_norm": 1.5168596506118774,
      "learning_rate": 6.157723130274205e-05,
      "loss": 1.8394,
      "step": 4731
    },
    {
      "epoch": 0.8524206259851385,
      "grad_norm": 1.4738023281097412,
      "learning_rate": 6.156345416338783e-05,
      "loss": 1.7189,
      "step": 4732
    },
    {
      "epoch": 0.8526007655933349,
      "grad_norm": 1.3453974723815918,
      "learning_rate": 6.154967609641895e-05,
      "loss": 1.3966,
      "step": 4733
    },
    {
      "epoch": 0.8527809052015312,
      "grad_norm": 1.2744262218475342,
      "learning_rate": 6.153589710294067e-05,
      "loss": 1.3853,
      "step": 4734
    },
    {
      "epoch": 0.8529610448097276,
      "grad_norm": 1.3842501640319824,
      "learning_rate": 6.152211718405835e-05,
      "loss": 1.769,
      "step": 4735
    },
    {
      "epoch": 0.8531411844179239,
      "grad_norm": 1.38172447681427,
      "learning_rate": 6.15083363408774e-05,
      "loss": 1.3845,
      "step": 4736
    },
    {
      "epoch": 0.8533213240261203,
      "grad_norm": 1.382222056388855,
      "learning_rate": 6.149455457450332e-05,
      "loss": 1.7328,
      "step": 4737
    },
    {
      "epoch": 0.8535014636343166,
      "grad_norm": 1.5157599449157715,
      "learning_rate": 6.148077188604167e-05,
      "loss": 1.6104,
      "step": 4738
    },
    {
      "epoch": 0.853681603242513,
      "grad_norm": 1.3676745891571045,
      "learning_rate": 6.146698827659809e-05,
      "loss": 1.4409,
      "step": 4739
    },
    {
      "epoch": 0.8538617428507093,
      "grad_norm": 1.319949984550476,
      "learning_rate": 6.145320374727828e-05,
      "loss": 1.4274,
      "step": 4740
    },
    {
      "epoch": 0.8540418824589057,
      "grad_norm": 1.4198298454284668,
      "learning_rate": 6.143941829918807e-05,
      "loss": 1.444,
      "step": 4741
    },
    {
      "epoch": 0.854222022067102,
      "grad_norm": 1.4113118648529053,
      "learning_rate": 6.142563193343328e-05,
      "loss": 1.5287,
      "step": 4742
    },
    {
      "epoch": 0.8544021616752984,
      "grad_norm": 1.3260498046875,
      "learning_rate": 6.141184465111985e-05,
      "loss": 1.5862,
      "step": 4743
    },
    {
      "epoch": 0.8545823012834947,
      "grad_norm": 1.374909520149231,
      "learning_rate": 6.139805645335383e-05,
      "loss": 1.4731,
      "step": 4744
    },
    {
      "epoch": 0.8547624408916911,
      "grad_norm": 1.4753024578094482,
      "learning_rate": 6.138426734124125e-05,
      "loss": 1.7286,
      "step": 4745
    },
    {
      "epoch": 0.8549425804998874,
      "grad_norm": 1.388883352279663,
      "learning_rate": 6.13704773158883e-05,
      "loss": 1.401,
      "step": 4746
    },
    {
      "epoch": 0.8551227201080838,
      "grad_norm": 1.352304458618164,
      "learning_rate": 6.13566863784012e-05,
      "loss": 1.131,
      "step": 4747
    },
    {
      "epoch": 0.8553028597162801,
      "grad_norm": 1.4158308506011963,
      "learning_rate": 6.134289452988624e-05,
      "loss": 1.4943,
      "step": 4748
    },
    {
      "epoch": 0.8554829993244765,
      "grad_norm": 1.1839171648025513,
      "learning_rate": 6.132910177144981e-05,
      "loss": 1.0694,
      "step": 4749
    },
    {
      "epoch": 0.8556631389326729,
      "grad_norm": 1.160314679145813,
      "learning_rate": 6.131530810419836e-05,
      "loss": 1.1007,
      "step": 4750
    },
    {
      "epoch": 0.8558432785408692,
      "grad_norm": 1.2519688606262207,
      "learning_rate": 6.13015135292384e-05,
      "loss": 2.079,
      "step": 4751
    },
    {
      "epoch": 0.8560234181490656,
      "grad_norm": 1.377999186515808,
      "learning_rate": 6.128771804767653e-05,
      "loss": 2.2511,
      "step": 4752
    },
    {
      "epoch": 0.8562035577572619,
      "grad_norm": 1.21393883228302,
      "learning_rate": 6.127392166061941e-05,
      "loss": 1.8693,
      "step": 4753
    },
    {
      "epoch": 0.8563836973654583,
      "grad_norm": 1.2158241271972656,
      "learning_rate": 6.12601243691738e-05,
      "loss": 1.9023,
      "step": 4754
    },
    {
      "epoch": 0.8565638369736546,
      "grad_norm": 1.2721275091171265,
      "learning_rate": 6.124632617444649e-05,
      "loss": 1.6744,
      "step": 4755
    },
    {
      "epoch": 0.856743976581851,
      "grad_norm": 1.254005789756775,
      "learning_rate": 6.123252707754438e-05,
      "loss": 1.9115,
      "step": 4756
    },
    {
      "epoch": 0.8569241161900473,
      "grad_norm": 1.3052823543548584,
      "learning_rate": 6.121872707957441e-05,
      "loss": 1.626,
      "step": 4757
    },
    {
      "epoch": 0.8571042557982437,
      "grad_norm": 1.4470106363296509,
      "learning_rate": 6.120492618164362e-05,
      "loss": 1.7882,
      "step": 4758
    },
    {
      "epoch": 0.85728439540644,
      "grad_norm": 1.6067529916763306,
      "learning_rate": 6.11911243848591e-05,
      "loss": 1.9188,
      "step": 4759
    },
    {
      "epoch": 0.8574645350146364,
      "grad_norm": 1.6848806142807007,
      "learning_rate": 6.117732169032805e-05,
      "loss": 2.2247,
      "step": 4760
    },
    {
      "epoch": 0.8576446746228327,
      "grad_norm": 1.375502109527588,
      "learning_rate": 6.116351809915769e-05,
      "loss": 1.6211,
      "step": 4761
    },
    {
      "epoch": 0.8578248142310291,
      "grad_norm": 1.215942144393921,
      "learning_rate": 6.114971361245534e-05,
      "loss": 1.428,
      "step": 4762
    },
    {
      "epoch": 0.8580049538392254,
      "grad_norm": 1.2597742080688477,
      "learning_rate": 6.11359082313284e-05,
      "loss": 1.5426,
      "step": 4763
    },
    {
      "epoch": 0.8581850934474218,
      "grad_norm": 1.2725377082824707,
      "learning_rate": 6.112210195688433e-05,
      "loss": 1.585,
      "step": 4764
    },
    {
      "epoch": 0.8583652330556181,
      "grad_norm": 1.2748991250991821,
      "learning_rate": 6.110829479023066e-05,
      "loss": 1.6192,
      "step": 4765
    },
    {
      "epoch": 0.8585453726638145,
      "grad_norm": 1.4489142894744873,
      "learning_rate": 6.109448673247498e-05,
      "loss": 1.6671,
      "step": 4766
    },
    {
      "epoch": 0.8587255122720108,
      "grad_norm": 1.25693941116333,
      "learning_rate": 6.108067778472498e-05,
      "loss": 1.6059,
      "step": 4767
    },
    {
      "epoch": 0.8589056518802072,
      "grad_norm": 1.4101102352142334,
      "learning_rate": 6.106686794808842e-05,
      "loss": 1.6158,
      "step": 4768
    },
    {
      "epoch": 0.8590857914884035,
      "grad_norm": 1.3865492343902588,
      "learning_rate": 6.105305722367309e-05,
      "loss": 1.5778,
      "step": 4769
    },
    {
      "epoch": 0.8592659310965999,
      "grad_norm": 1.3160815238952637,
      "learning_rate": 6.10392456125869e-05,
      "loss": 1.5773,
      "step": 4770
    },
    {
      "epoch": 0.8594460707047962,
      "grad_norm": 1.3327451944351196,
      "learning_rate": 6.10254331159378e-05,
      "loss": 1.4592,
      "step": 4771
    },
    {
      "epoch": 0.8596262103129926,
      "grad_norm": 1.2754762172698975,
      "learning_rate": 6.101161973483383e-05,
      "loss": 1.4346,
      "step": 4772
    },
    {
      "epoch": 0.8598063499211889,
      "grad_norm": 1.375786304473877,
      "learning_rate": 6.0997805470383085e-05,
      "loss": 1.5247,
      "step": 4773
    },
    {
      "epoch": 0.8599864895293853,
      "grad_norm": 1.1936523914337158,
      "learning_rate": 6.098399032369375e-05,
      "loss": 1.3325,
      "step": 4774
    },
    {
      "epoch": 0.8601666291375816,
      "grad_norm": 1.3287545442581177,
      "learning_rate": 6.097017429587406e-05,
      "loss": 1.5158,
      "step": 4775
    },
    {
      "epoch": 0.860346768745778,
      "grad_norm": 1.32796311378479,
      "learning_rate": 6.095635738803231e-05,
      "loss": 1.4306,
      "step": 4776
    },
    {
      "epoch": 0.8605269083539744,
      "grad_norm": 1.3871411085128784,
      "learning_rate": 6.0942539601276924e-05,
      "loss": 1.5126,
      "step": 4777
    },
    {
      "epoch": 0.8607070479621707,
      "grad_norm": 1.3941295146942139,
      "learning_rate": 6.0928720936716335e-05,
      "loss": 1.5253,
      "step": 4778
    },
    {
      "epoch": 0.8608871875703671,
      "grad_norm": 1.3492759466171265,
      "learning_rate": 6.09149013954591e-05,
      "loss": 1.3709,
      "step": 4779
    },
    {
      "epoch": 0.8610673271785634,
      "grad_norm": 1.380707025527954,
      "learning_rate": 6.090108097861377e-05,
      "loss": 1.6046,
      "step": 4780
    },
    {
      "epoch": 0.8612474667867598,
      "grad_norm": 1.2643134593963623,
      "learning_rate": 6.0887259687289035e-05,
      "loss": 1.3613,
      "step": 4781
    },
    {
      "epoch": 0.8614276063949561,
      "grad_norm": 1.3875644207000732,
      "learning_rate": 6.0873437522593626e-05,
      "loss": 1.6409,
      "step": 4782
    },
    {
      "epoch": 0.8616077460031525,
      "grad_norm": 1.3749431371688843,
      "learning_rate": 6.085961448563635e-05,
      "loss": 1.5268,
      "step": 4783
    },
    {
      "epoch": 0.8617878856113488,
      "grad_norm": 1.2980433702468872,
      "learning_rate": 6.084579057752611e-05,
      "loss": 1.4625,
      "step": 4784
    },
    {
      "epoch": 0.8619680252195452,
      "grad_norm": 1.4089449644088745,
      "learning_rate": 6.083196579937183e-05,
      "loss": 1.559,
      "step": 4785
    },
    {
      "epoch": 0.8621481648277415,
      "grad_norm": 1.3613237142562866,
      "learning_rate": 6.081814015228252e-05,
      "loss": 1.5107,
      "step": 4786
    },
    {
      "epoch": 0.8623283044359379,
      "grad_norm": 1.4567334651947021,
      "learning_rate": 6.080431363736728e-05,
      "loss": 1.5551,
      "step": 4787
    },
    {
      "epoch": 0.8625084440441342,
      "grad_norm": 1.2888206243515015,
      "learning_rate": 6.079048625573525e-05,
      "loss": 1.4785,
      "step": 4788
    },
    {
      "epoch": 0.8626885836523306,
      "grad_norm": 1.4184306859970093,
      "learning_rate": 6.077665800849568e-05,
      "loss": 1.6856,
      "step": 4789
    },
    {
      "epoch": 0.8628687232605269,
      "grad_norm": 1.4704737663269043,
      "learning_rate": 6.0762828896757864e-05,
      "loss": 1.6755,
      "step": 4790
    },
    {
      "epoch": 0.8630488628687233,
      "grad_norm": 1.4484741687774658,
      "learning_rate": 6.074899892163115e-05,
      "loss": 1.6205,
      "step": 4791
    },
    {
      "epoch": 0.8632290024769196,
      "grad_norm": 1.4339154958724976,
      "learning_rate": 6.073516808422496e-05,
      "loss": 1.5008,
      "step": 4792
    },
    {
      "epoch": 0.863409142085116,
      "grad_norm": 1.2951631546020508,
      "learning_rate": 6.072133638564883e-05,
      "loss": 1.5817,
      "step": 4793
    },
    {
      "epoch": 0.8635892816933123,
      "grad_norm": 1.449933409690857,
      "learning_rate": 6.070750382701232e-05,
      "loss": 1.5311,
      "step": 4794
    },
    {
      "epoch": 0.8637694213015087,
      "grad_norm": 1.4684382677078247,
      "learning_rate": 6.0693670409425074e-05,
      "loss": 1.4948,
      "step": 4795
    },
    {
      "epoch": 0.863949560909705,
      "grad_norm": 1.30257248878479,
      "learning_rate": 6.0679836133996783e-05,
      "loss": 1.3349,
      "step": 4796
    },
    {
      "epoch": 0.8641297005179014,
      "grad_norm": 1.504602313041687,
      "learning_rate": 6.0666001001837236e-05,
      "loss": 1.4596,
      "step": 4797
    },
    {
      "epoch": 0.8643098401260977,
      "grad_norm": 1.3995978832244873,
      "learning_rate": 6.0652165014056294e-05,
      "loss": 1.3697,
      "step": 4798
    },
    {
      "epoch": 0.8644899797342941,
      "grad_norm": 1.389730453491211,
      "learning_rate": 6.063832817176386e-05,
      "loss": 1.2978,
      "step": 4799
    },
    {
      "epoch": 0.8646701193424904,
      "grad_norm": 1.2073769569396973,
      "learning_rate": 6.0624490476069916e-05,
      "loss": 1.2034,
      "step": 4800
    },
    {
      "epoch": 0.8648502589506868,
      "grad_norm": 1.3232203722000122,
      "learning_rate": 6.061065192808455e-05,
      "loss": 2.1753,
      "step": 4801
    },
    {
      "epoch": 0.8650303985588831,
      "grad_norm": 1.2441351413726807,
      "learning_rate": 6.0596812528917827e-05,
      "loss": 1.921,
      "step": 4802
    },
    {
      "epoch": 0.8652105381670795,
      "grad_norm": 1.377205729484558,
      "learning_rate": 6.058297227967996e-05,
      "loss": 1.8491,
      "step": 4803
    },
    {
      "epoch": 0.8653906777752758,
      "grad_norm": 1.2158855199813843,
      "learning_rate": 6.056913118148122e-05,
      "loss": 1.6871,
      "step": 4804
    },
    {
      "epoch": 0.8655708173834722,
      "grad_norm": 1.1970714330673218,
      "learning_rate": 6.055528923543192e-05,
      "loss": 1.8951,
      "step": 4805
    },
    {
      "epoch": 0.8657509569916686,
      "grad_norm": 1.2609047889709473,
      "learning_rate": 6.0541446442642466e-05,
      "loss": 1.8977,
      "step": 4806
    },
    {
      "epoch": 0.8659310965998649,
      "grad_norm": 1.351499080657959,
      "learning_rate": 6.052760280422332e-05,
      "loss": 1.9532,
      "step": 4807
    },
    {
      "epoch": 0.8661112362080613,
      "grad_norm": 1.4403001070022583,
      "learning_rate": 6.0513758321285e-05,
      "loss": 2.0017,
      "step": 4808
    },
    {
      "epoch": 0.8662913758162576,
      "grad_norm": 1.7267508506774902,
      "learning_rate": 6.049991299493809e-05,
      "loss": 2.5492,
      "step": 4809
    },
    {
      "epoch": 0.866471515424454,
      "grad_norm": 1.7751617431640625,
      "learning_rate": 6.0486066826293295e-05,
      "loss": 2.268,
      "step": 4810
    },
    {
      "epoch": 0.8666516550326503,
      "grad_norm": 1.392783522605896,
      "learning_rate": 6.047221981646132e-05,
      "loss": 1.8072,
      "step": 4811
    },
    {
      "epoch": 0.8668317946408467,
      "grad_norm": 1.4296787977218628,
      "learning_rate": 6.045837196655299e-05,
      "loss": 1.6098,
      "step": 4812
    },
    {
      "epoch": 0.867011934249043,
      "grad_norm": 1.2642333507537842,
      "learning_rate": 6.044452327767912e-05,
      "loss": 1.3019,
      "step": 4813
    },
    {
      "epoch": 0.8671920738572394,
      "grad_norm": 1.1790757179260254,
      "learning_rate": 6.0430673750950703e-05,
      "loss": 1.4551,
      "step": 4814
    },
    {
      "epoch": 0.8673722134654357,
      "grad_norm": 1.4662318229675293,
      "learning_rate": 6.0416823387478715e-05,
      "loss": 1.6536,
      "step": 4815
    },
    {
      "epoch": 0.8675523530736321,
      "grad_norm": 1.3558217287063599,
      "learning_rate": 6.040297218837422e-05,
      "loss": 1.6962,
      "step": 4816
    },
    {
      "epoch": 0.8677324926818284,
      "grad_norm": 1.2178094387054443,
      "learning_rate": 6.038912015474837e-05,
      "loss": 1.3872,
      "step": 4817
    },
    {
      "epoch": 0.8679126322900248,
      "grad_norm": 1.3234713077545166,
      "learning_rate": 6.0375267287712366e-05,
      "loss": 1.7961,
      "step": 4818
    },
    {
      "epoch": 0.8680927718982211,
      "grad_norm": 1.1687045097351074,
      "learning_rate": 6.036141358837747e-05,
      "loss": 1.4426,
      "step": 4819
    },
    {
      "epoch": 0.8682729115064175,
      "grad_norm": 1.3265058994293213,
      "learning_rate": 6.0347559057855006e-05,
      "loss": 1.6344,
      "step": 4820
    },
    {
      "epoch": 0.8684530511146138,
      "grad_norm": 1.4721804857254028,
      "learning_rate": 6.033370369725642e-05,
      "loss": 1.5868,
      "step": 4821
    },
    {
      "epoch": 0.8686331907228102,
      "grad_norm": 1.2043962478637695,
      "learning_rate": 6.031984750769315e-05,
      "loss": 1.4309,
      "step": 4822
    },
    {
      "epoch": 0.8688133303310065,
      "grad_norm": 1.3187940120697021,
      "learning_rate": 6.0305990490276745e-05,
      "loss": 1.6828,
      "step": 4823
    },
    {
      "epoch": 0.8689934699392029,
      "grad_norm": 1.3083454370498657,
      "learning_rate": 6.029213264611879e-05,
      "loss": 1.6196,
      "step": 4824
    },
    {
      "epoch": 0.8691736095473992,
      "grad_norm": 1.3861478567123413,
      "learning_rate": 6.027827397633097e-05,
      "loss": 1.7051,
      "step": 4825
    },
    {
      "epoch": 0.8693537491555956,
      "grad_norm": 1.3731266260147095,
      "learning_rate": 6.026441448202502e-05,
      "loss": 1.4161,
      "step": 4826
    },
    {
      "epoch": 0.8695338887637919,
      "grad_norm": 1.438948154449463,
      "learning_rate": 6.025055416431273e-05,
      "loss": 1.6066,
      "step": 4827
    },
    {
      "epoch": 0.8697140283719883,
      "grad_norm": 1.2723557949066162,
      "learning_rate": 6.0236693024306e-05,
      "loss": 1.4009,
      "step": 4828
    },
    {
      "epoch": 0.8698941679801846,
      "grad_norm": 1.3104726076126099,
      "learning_rate": 6.022283106311673e-05,
      "loss": 1.4465,
      "step": 4829
    },
    {
      "epoch": 0.870074307588381,
      "grad_norm": 1.391049861907959,
      "learning_rate": 6.0208968281856927e-05,
      "loss": 1.7052,
      "step": 4830
    },
    {
      "epoch": 0.8702544471965773,
      "grad_norm": 1.4700539112091064,
      "learning_rate": 6.019510468163866e-05,
      "loss": 1.6007,
      "step": 4831
    },
    {
      "epoch": 0.8704345868047737,
      "grad_norm": 1.3410675525665283,
      "learning_rate": 6.0181240263574056e-05,
      "loss": 1.6324,
      "step": 4832
    },
    {
      "epoch": 0.87061472641297,
      "grad_norm": 1.3417143821716309,
      "learning_rate": 6.016737502877533e-05,
      "loss": 1.482,
      "step": 4833
    },
    {
      "epoch": 0.8707948660211664,
      "grad_norm": 1.4190144538879395,
      "learning_rate": 6.0153508978354734e-05,
      "loss": 1.6507,
      "step": 4834
    },
    {
      "epoch": 0.8709750056293628,
      "grad_norm": 1.297910213470459,
      "learning_rate": 6.013964211342459e-05,
      "loss": 1.5162,
      "step": 4835
    },
    {
      "epoch": 0.8711551452375591,
      "grad_norm": 1.40993332862854,
      "learning_rate": 6.012577443509728e-05,
      "loss": 1.4281,
      "step": 4836
    },
    {
      "epoch": 0.8713352848457555,
      "grad_norm": 1.4083045721054077,
      "learning_rate": 6.0111905944485304e-05,
      "loss": 1.6359,
      "step": 4837
    },
    {
      "epoch": 0.8715154244539518,
      "grad_norm": 1.3539639711380005,
      "learning_rate": 6.009803664270114e-05,
      "loss": 1.3878,
      "step": 4838
    },
    {
      "epoch": 0.8716955640621482,
      "grad_norm": 1.5067014694213867,
      "learning_rate": 6.00841665308574e-05,
      "loss": 1.6564,
      "step": 4839
    },
    {
      "epoch": 0.8718757036703445,
      "grad_norm": 1.4218924045562744,
      "learning_rate": 6.007029561006674e-05,
      "loss": 1.4749,
      "step": 4840
    },
    {
      "epoch": 0.8720558432785409,
      "grad_norm": 1.3239388465881348,
      "learning_rate": 6.005642388144185e-05,
      "loss": 1.4957,
      "step": 4841
    },
    {
      "epoch": 0.8722359828867372,
      "grad_norm": 1.4342950582504272,
      "learning_rate": 6.0042551346095554e-05,
      "loss": 1.6117,
      "step": 4842
    },
    {
      "epoch": 0.8724161224949336,
      "grad_norm": 1.4106281995773315,
      "learning_rate": 6.002867800514066e-05,
      "loss": 1.4038,
      "step": 4843
    },
    {
      "epoch": 0.8725962621031299,
      "grad_norm": 1.6074198484420776,
      "learning_rate": 6.001480385969013e-05,
      "loss": 1.9554,
      "step": 4844
    },
    {
      "epoch": 0.8727764017113263,
      "grad_norm": 1.4746546745300293,
      "learning_rate": 6.00009289108569e-05,
      "loss": 1.5582,
      "step": 4845
    },
    {
      "epoch": 0.8729565413195226,
      "grad_norm": 1.5003559589385986,
      "learning_rate": 5.998705315975402e-05,
      "loss": 1.5545,
      "step": 4846
    },
    {
      "epoch": 0.873136680927719,
      "grad_norm": 1.4416735172271729,
      "learning_rate": 5.9973176607494596e-05,
      "loss": 1.6708,
      "step": 4847
    },
    {
      "epoch": 0.8733168205359153,
      "grad_norm": 1.5088763236999512,
      "learning_rate": 5.99592992551918e-05,
      "loss": 1.6303,
      "step": 4848
    },
    {
      "epoch": 0.8734969601441117,
      "grad_norm": 1.268067717552185,
      "learning_rate": 5.994542110395889e-05,
      "loss": 1.4367,
      "step": 4849
    },
    {
      "epoch": 0.873677099752308,
      "grad_norm": 1.4419240951538086,
      "learning_rate": 5.993154215490913e-05,
      "loss": 1.6497,
      "step": 4850
    },
    {
      "epoch": 0.8738572393605044,
      "grad_norm": 1.3228414058685303,
      "learning_rate": 5.99176624091559e-05,
      "loss": 1.8495,
      "step": 4851
    },
    {
      "epoch": 0.8740373789687007,
      "grad_norm": 1.24224853515625,
      "learning_rate": 5.990378186781261e-05,
      "loss": 1.6382,
      "step": 4852
    },
    {
      "epoch": 0.8742175185768971,
      "grad_norm": 1.3494120836257935,
      "learning_rate": 5.988990053199277e-05,
      "loss": 1.9537,
      "step": 4853
    },
    {
      "epoch": 0.8743976581850934,
      "grad_norm": 1.288163185119629,
      "learning_rate": 5.987601840280993e-05,
      "loss": 1.9257,
      "step": 4854
    },
    {
      "epoch": 0.8745777977932898,
      "grad_norm": 1.2223747968673706,
      "learning_rate": 5.98621354813777e-05,
      "loss": 1.7363,
      "step": 4855
    },
    {
      "epoch": 0.8747579374014861,
      "grad_norm": 1.2455908060073853,
      "learning_rate": 5.984825176880977e-05,
      "loss": 1.8408,
      "step": 4856
    },
    {
      "epoch": 0.8749380770096825,
      "grad_norm": 1.3384571075439453,
      "learning_rate": 5.983436726621987e-05,
      "loss": 1.8955,
      "step": 4857
    },
    {
      "epoch": 0.8751182166178788,
      "grad_norm": 1.524652361869812,
      "learning_rate": 5.982048197472183e-05,
      "loss": 1.89,
      "step": 4858
    },
    {
      "epoch": 0.8752983562260752,
      "grad_norm": 1.3820980787277222,
      "learning_rate": 5.980659589542951e-05,
      "loss": 1.6155,
      "step": 4859
    },
    {
      "epoch": 0.8754784958342715,
      "grad_norm": 1.9295732975006104,
      "learning_rate": 5.979270902945684e-05,
      "loss": 2.4923,
      "step": 4860
    },
    {
      "epoch": 0.8756586354424679,
      "grad_norm": 1.2807426452636719,
      "learning_rate": 5.977882137791781e-05,
      "loss": 1.6508,
      "step": 4861
    },
    {
      "epoch": 0.8758387750506643,
      "grad_norm": 1.2610318660736084,
      "learning_rate": 5.97649329419265e-05,
      "loss": 1.6568,
      "step": 4862
    },
    {
      "epoch": 0.8760189146588606,
      "grad_norm": 1.335521936416626,
      "learning_rate": 5.975104372259703e-05,
      "loss": 1.5188,
      "step": 4863
    },
    {
      "epoch": 0.876199054267057,
      "grad_norm": 1.3706283569335938,
      "learning_rate": 5.973715372104357e-05,
      "loss": 1.4579,
      "step": 4864
    },
    {
      "epoch": 0.8763791938752533,
      "grad_norm": 1.3073537349700928,
      "learning_rate": 5.9723262938380396e-05,
      "loss": 1.4596,
      "step": 4865
    },
    {
      "epoch": 0.8765593334834497,
      "grad_norm": 1.4208894968032837,
      "learning_rate": 5.9709371375721815e-05,
      "loss": 1.6413,
      "step": 4866
    },
    {
      "epoch": 0.876739473091646,
      "grad_norm": 1.373203992843628,
      "learning_rate": 5.9695479034182164e-05,
      "loss": 1.7215,
      "step": 4867
    },
    {
      "epoch": 0.8769196126998424,
      "grad_norm": 1.3259773254394531,
      "learning_rate": 5.968158591487592e-05,
      "loss": 1.4677,
      "step": 4868
    },
    {
      "epoch": 0.8770997523080387,
      "grad_norm": 1.464622974395752,
      "learning_rate": 5.966769201891756e-05,
      "loss": 1.5597,
      "step": 4869
    },
    {
      "epoch": 0.8772798919162351,
      "grad_norm": 1.361344337463379,
      "learning_rate": 5.965379734742167e-05,
      "loss": 1.7767,
      "step": 4870
    },
    {
      "epoch": 0.8774600315244314,
      "grad_norm": 1.3051594495773315,
      "learning_rate": 5.963990190150286e-05,
      "loss": 1.5666,
      "step": 4871
    },
    {
      "epoch": 0.8776401711326278,
      "grad_norm": 1.4447046518325806,
      "learning_rate": 5.962600568227581e-05,
      "loss": 1.683,
      "step": 4872
    },
    {
      "epoch": 0.8778203107408241,
      "grad_norm": 1.3636035919189453,
      "learning_rate": 5.961210869085527e-05,
      "loss": 1.5195,
      "step": 4873
    },
    {
      "epoch": 0.8780004503490205,
      "grad_norm": 1.2971924543380737,
      "learning_rate": 5.959821092835605e-05,
      "loss": 1.5972,
      "step": 4874
    },
    {
      "epoch": 0.8781805899572168,
      "grad_norm": 1.2787104845046997,
      "learning_rate": 5.9584312395893026e-05,
      "loss": 1.3256,
      "step": 4875
    },
    {
      "epoch": 0.8783607295654132,
      "grad_norm": 1.255251407623291,
      "learning_rate": 5.957041309458113e-05,
      "loss": 1.3018,
      "step": 4876
    },
    {
      "epoch": 0.8785408691736095,
      "grad_norm": 1.3704200983047485,
      "learning_rate": 5.9556513025535374e-05,
      "loss": 1.5456,
      "step": 4877
    },
    {
      "epoch": 0.8787210087818059,
      "grad_norm": 1.4923502206802368,
      "learning_rate": 5.954261218987077e-05,
      "loss": 1.671,
      "step": 4878
    },
    {
      "epoch": 0.8789011483900022,
      "grad_norm": 1.2371340990066528,
      "learning_rate": 5.9528710588702486e-05,
      "loss": 1.541,
      "step": 4879
    },
    {
      "epoch": 0.8790812879981986,
      "grad_norm": 1.314180850982666,
      "learning_rate": 5.951480822314567e-05,
      "loss": 1.4818,
      "step": 4880
    },
    {
      "epoch": 0.8792614276063949,
      "grad_norm": 1.5361371040344238,
      "learning_rate": 5.9500905094315565e-05,
      "loss": 1.5173,
      "step": 4881
    },
    {
      "epoch": 0.8794415672145913,
      "grad_norm": 1.1707242727279663,
      "learning_rate": 5.94870012033275e-05,
      "loss": 1.4226,
      "step": 4882
    },
    {
      "epoch": 0.8796217068227876,
      "grad_norm": 1.41873037815094,
      "learning_rate": 5.94730965512968e-05,
      "loss": 1.5139,
      "step": 4883
    },
    {
      "epoch": 0.879801846430984,
      "grad_norm": 1.5557864904403687,
      "learning_rate": 5.945919113933893e-05,
      "loss": 1.7769,
      "step": 4884
    },
    {
      "epoch": 0.8799819860391803,
      "grad_norm": 1.3495111465454102,
      "learning_rate": 5.9445284968569335e-05,
      "loss": 1.5695,
      "step": 4885
    },
    {
      "epoch": 0.8801621256473767,
      "grad_norm": 1.4541488885879517,
      "learning_rate": 5.943137804010359e-05,
      "loss": 1.5186,
      "step": 4886
    },
    {
      "epoch": 0.880342265255573,
      "grad_norm": 1.466526746749878,
      "learning_rate": 5.9417470355057295e-05,
      "loss": 1.6613,
      "step": 4887
    },
    {
      "epoch": 0.8805224048637694,
      "grad_norm": 1.3073025941848755,
      "learning_rate": 5.940356191454611e-05,
      "loss": 1.6232,
      "step": 4888
    },
    {
      "epoch": 0.8807025444719657,
      "grad_norm": 1.3118188381195068,
      "learning_rate": 5.938965271968577e-05,
      "loss": 1.4054,
      "step": 4889
    },
    {
      "epoch": 0.8808826840801621,
      "grad_norm": 1.4506570100784302,
      "learning_rate": 5.9375742771592056e-05,
      "loss": 1.6646,
      "step": 4890
    },
    {
      "epoch": 0.8810628236883585,
      "grad_norm": 1.4313631057739258,
      "learning_rate": 5.936183207138084e-05,
      "loss": 1.585,
      "step": 4891
    },
    {
      "epoch": 0.8812429632965548,
      "grad_norm": 1.5904314517974854,
      "learning_rate": 5.9347920620167996e-05,
      "loss": 1.7123,
      "step": 4892
    },
    {
      "epoch": 0.8814231029047512,
      "grad_norm": 1.330359935760498,
      "learning_rate": 5.933400841906953e-05,
      "loss": 1.3484,
      "step": 4893
    },
    {
      "epoch": 0.8816032425129475,
      "grad_norm": 1.6149178743362427,
      "learning_rate": 5.9320095469201454e-05,
      "loss": 1.571,
      "step": 4894
    },
    {
      "epoch": 0.8817833821211439,
      "grad_norm": 1.3532160520553589,
      "learning_rate": 5.930618177167984e-05,
      "loss": 1.3791,
      "step": 4895
    },
    {
      "epoch": 0.8819635217293402,
      "grad_norm": 1.385862946510315,
      "learning_rate": 5.9292267327620876e-05,
      "loss": 1.3973,
      "step": 4896
    },
    {
      "epoch": 0.8821436613375366,
      "grad_norm": 1.3089302778244019,
      "learning_rate": 5.927835213814073e-05,
      "loss": 1.3804,
      "step": 4897
    },
    {
      "epoch": 0.8823238009457329,
      "grad_norm": 1.2697385549545288,
      "learning_rate": 5.9264436204355724e-05,
      "loss": 1.2888,
      "step": 4898
    },
    {
      "epoch": 0.8825039405539293,
      "grad_norm": 1.3289599418640137,
      "learning_rate": 5.9250519527382144e-05,
      "loss": 1.4597,
      "step": 4899
    },
    {
      "epoch": 0.8826840801621256,
      "grad_norm": 1.3576087951660156,
      "learning_rate": 5.9236602108336395e-05,
      "loss": 1.2186,
      "step": 4900
    },
    {
      "epoch": 0.882864219770322,
      "grad_norm": 1.56609308719635,
      "learning_rate": 5.922268394833492e-05,
      "loss": 2.0012,
      "step": 4901
    },
    {
      "epoch": 0.8830443593785183,
      "grad_norm": 1.3856627941131592,
      "learning_rate": 5.920876504849423e-05,
      "loss": 2.1709,
      "step": 4902
    },
    {
      "epoch": 0.8832244989867147,
      "grad_norm": 1.3094737529754639,
      "learning_rate": 5.91948454099309e-05,
      "loss": 2.0393,
      "step": 4903
    },
    {
      "epoch": 0.883404638594911,
      "grad_norm": 1.206174612045288,
      "learning_rate": 5.918092503376156e-05,
      "loss": 1.8737,
      "step": 4904
    },
    {
      "epoch": 0.8835847782031074,
      "grad_norm": 1.4572250843048096,
      "learning_rate": 5.916700392110287e-05,
      "loss": 2.1271,
      "step": 4905
    },
    {
      "epoch": 0.8837649178113037,
      "grad_norm": 1.272459626197815,
      "learning_rate": 5.9153082073071586e-05,
      "loss": 1.796,
      "step": 4906
    },
    {
      "epoch": 0.8839450574195001,
      "grad_norm": 1.350561499595642,
      "learning_rate": 5.913915949078452e-05,
      "loss": 1.6258,
      "step": 4907
    },
    {
      "epoch": 0.8841251970276964,
      "grad_norm": 1.4500226974487305,
      "learning_rate": 5.912523617535855e-05,
      "loss": 1.8637,
      "step": 4908
    },
    {
      "epoch": 0.8843053366358928,
      "grad_norm": 1.6210451126098633,
      "learning_rate": 5.911131212791056e-05,
      "loss": 2.1012,
      "step": 4909
    },
    {
      "epoch": 0.8844854762440891,
      "grad_norm": 2.0391478538513184,
      "learning_rate": 5.9097387349557553e-05,
      "loss": 2.7092,
      "step": 4910
    },
    {
      "epoch": 0.8846656158522855,
      "grad_norm": 1.4408419132232666,
      "learning_rate": 5.9083461841416554e-05,
      "loss": 1.7117,
      "step": 4911
    },
    {
      "epoch": 0.8848457554604818,
      "grad_norm": 1.1679744720458984,
      "learning_rate": 5.906953560460469e-05,
      "loss": 1.4988,
      "step": 4912
    },
    {
      "epoch": 0.8850258950686782,
      "grad_norm": 1.324597954750061,
      "learning_rate": 5.905560864023908e-05,
      "loss": 1.7074,
      "step": 4913
    },
    {
      "epoch": 0.8852060346768745,
      "grad_norm": 1.2151472568511963,
      "learning_rate": 5.904168094943696e-05,
      "loss": 1.353,
      "step": 4914
    },
    {
      "epoch": 0.8853861742850709,
      "grad_norm": 1.1258124113082886,
      "learning_rate": 5.902775253331562e-05,
      "loss": 1.3196,
      "step": 4915
    },
    {
      "epoch": 0.8855663138932672,
      "grad_norm": 1.2189654111862183,
      "learning_rate": 5.901382339299234e-05,
      "loss": 1.6053,
      "step": 4916
    },
    {
      "epoch": 0.8857464535014636,
      "grad_norm": 1.2708308696746826,
      "learning_rate": 5.8999893529584546e-05,
      "loss": 1.5039,
      "step": 4917
    },
    {
      "epoch": 0.8859265931096599,
      "grad_norm": 1.2425771951675415,
      "learning_rate": 5.898596294420968e-05,
      "loss": 1.3228,
      "step": 4918
    },
    {
      "epoch": 0.8861067327178563,
      "grad_norm": 1.4810388088226318,
      "learning_rate": 5.897203163798524e-05,
      "loss": 1.7695,
      "step": 4919
    },
    {
      "epoch": 0.8862868723260527,
      "grad_norm": 1.410508394241333,
      "learning_rate": 5.8958099612028804e-05,
      "loss": 1.4703,
      "step": 4920
    },
    {
      "epoch": 0.886467011934249,
      "grad_norm": 1.3662093877792358,
      "learning_rate": 5.894416686745797e-05,
      "loss": 1.5843,
      "step": 4921
    },
    {
      "epoch": 0.8866471515424454,
      "grad_norm": 1.4293785095214844,
      "learning_rate": 5.893023340539043e-05,
      "loss": 1.4908,
      "step": 4922
    },
    {
      "epoch": 0.8868272911506417,
      "grad_norm": 1.2521655559539795,
      "learning_rate": 5.8916299226943905e-05,
      "loss": 1.4374,
      "step": 4923
    },
    {
      "epoch": 0.8870074307588381,
      "grad_norm": 1.3169161081314087,
      "learning_rate": 5.8902364333236214e-05,
      "loss": 1.5272,
      "step": 4924
    },
    {
      "epoch": 0.8871875703670344,
      "grad_norm": 1.3253748416900635,
      "learning_rate": 5.888842872538519e-05,
      "loss": 1.4615,
      "step": 4925
    },
    {
      "epoch": 0.8873677099752308,
      "grad_norm": 1.3979521989822388,
      "learning_rate": 5.887449240450874e-05,
      "loss": 1.4807,
      "step": 4926
    },
    {
      "epoch": 0.8875478495834271,
      "grad_norm": 1.3415324687957764,
      "learning_rate": 5.886055537172482e-05,
      "loss": 1.4248,
      "step": 4927
    },
    {
      "epoch": 0.8877279891916235,
      "grad_norm": 1.2935024499893188,
      "learning_rate": 5.884661762815148e-05,
      "loss": 1.5159,
      "step": 4928
    },
    {
      "epoch": 0.8879081287998198,
      "grad_norm": 1.2247368097305298,
      "learning_rate": 5.883267917490678e-05,
      "loss": 1.3982,
      "step": 4929
    },
    {
      "epoch": 0.8880882684080162,
      "grad_norm": 1.396319031715393,
      "learning_rate": 5.8818740013108854e-05,
      "loss": 1.5317,
      "step": 4930
    },
    {
      "epoch": 0.8882684080162125,
      "grad_norm": 1.4318525791168213,
      "learning_rate": 5.8804800143875915e-05,
      "loss": 1.6454,
      "step": 4931
    },
    {
      "epoch": 0.888448547624409,
      "grad_norm": 1.4536333084106445,
      "learning_rate": 5.8790859568326176e-05,
      "loss": 1.9068,
      "step": 4932
    },
    {
      "epoch": 0.8886286872326052,
      "grad_norm": 1.271181344985962,
      "learning_rate": 5.877691828757798e-05,
      "loss": 1.397,
      "step": 4933
    },
    {
      "epoch": 0.8888088268408016,
      "grad_norm": 1.2889289855957031,
      "learning_rate": 5.876297630274965e-05,
      "loss": 1.3531,
      "step": 4934
    },
    {
      "epoch": 0.8889889664489979,
      "grad_norm": 1.4697061777114868,
      "learning_rate": 5.874903361495965e-05,
      "loss": 1.5059,
      "step": 4935
    },
    {
      "epoch": 0.8891691060571943,
      "grad_norm": 1.4518953561782837,
      "learning_rate": 5.8735090225326436e-05,
      "loss": 1.5801,
      "step": 4936
    },
    {
      "epoch": 0.8893492456653906,
      "grad_norm": 1.4577720165252686,
      "learning_rate": 5.872114613496854e-05,
      "loss": 1.5614,
      "step": 4937
    },
    {
      "epoch": 0.889529385273587,
      "grad_norm": 1.35596764087677,
      "learning_rate": 5.870720134500454e-05,
      "loss": 1.3468,
      "step": 4938
    },
    {
      "epoch": 0.8897095248817833,
      "grad_norm": 1.431528091430664,
      "learning_rate": 5.869325585655309e-05,
      "loss": 1.4518,
      "step": 4939
    },
    {
      "epoch": 0.8898896644899797,
      "grad_norm": 1.4274637699127197,
      "learning_rate": 5.8679309670732896e-05,
      "loss": 1.5935,
      "step": 4940
    },
    {
      "epoch": 0.890069804098176,
      "grad_norm": 1.4080716371536255,
      "learning_rate": 5.8665362788662706e-05,
      "loss": 1.7318,
      "step": 4941
    },
    {
      "epoch": 0.8902499437063724,
      "grad_norm": 1.3980385065078735,
      "learning_rate": 5.8651415211461335e-05,
      "loss": 1.4891,
      "step": 4942
    },
    {
      "epoch": 0.8904300833145687,
      "grad_norm": 1.5835100412368774,
      "learning_rate": 5.8637466940247645e-05,
      "loss": 1.7582,
      "step": 4943
    },
    {
      "epoch": 0.8906102229227651,
      "grad_norm": 1.3957884311676025,
      "learning_rate": 5.8623517976140554e-05,
      "loss": 1.5943,
      "step": 4944
    },
    {
      "epoch": 0.8907903625309614,
      "grad_norm": 1.3227876424789429,
      "learning_rate": 5.8609568320259066e-05,
      "loss": 1.3689,
      "step": 4945
    },
    {
      "epoch": 0.8909705021391578,
      "grad_norm": 1.4396952390670776,
      "learning_rate": 5.859561797372219e-05,
      "loss": 1.3578,
      "step": 4946
    },
    {
      "epoch": 0.8911506417473541,
      "grad_norm": 1.3999418020248413,
      "learning_rate": 5.858166693764905e-05,
      "loss": 1.4781,
      "step": 4947
    },
    {
      "epoch": 0.8913307813555505,
      "grad_norm": 1.877793312072754,
      "learning_rate": 5.856771521315875e-05,
      "loss": 1.6333,
      "step": 4948
    },
    {
      "epoch": 0.891510920963747,
      "grad_norm": 1.3292099237442017,
      "learning_rate": 5.85537628013705e-05,
      "loss": 1.5068,
      "step": 4949
    },
    {
      "epoch": 0.8916910605719433,
      "grad_norm": 1.2942478656768799,
      "learning_rate": 5.853980970340358e-05,
      "loss": 1.3592,
      "step": 4950
    },
    {
      "epoch": 0.8918712001801397,
      "grad_norm": 1.351731777191162,
      "learning_rate": 5.8525855920377285e-05,
      "loss": 2.1554,
      "step": 4951
    },
    {
      "epoch": 0.892051339788336,
      "grad_norm": 1.1347591876983643,
      "learning_rate": 5.851190145341098e-05,
      "loss": 1.6675,
      "step": 4952
    },
    {
      "epoch": 0.8922314793965324,
      "grad_norm": 1.4482944011688232,
      "learning_rate": 5.849794630362408e-05,
      "loss": 2.5191,
      "step": 4953
    },
    {
      "epoch": 0.8924116190047287,
      "grad_norm": 1.2418495416641235,
      "learning_rate": 5.8483990472136074e-05,
      "loss": 2.0076,
      "step": 4954
    },
    {
      "epoch": 0.8925917586129251,
      "grad_norm": 1.3946818113327026,
      "learning_rate": 5.847003396006647e-05,
      "loss": 1.9167,
      "step": 4955
    },
    {
      "epoch": 0.8927718982211214,
      "grad_norm": 1.3768799304962158,
      "learning_rate": 5.8456076768534886e-05,
      "loss": 1.881,
      "step": 4956
    },
    {
      "epoch": 0.8929520378293178,
      "grad_norm": 1.40651273727417,
      "learning_rate": 5.844211889866094e-05,
      "loss": 1.9443,
      "step": 4957
    },
    {
      "epoch": 0.893132177437514,
      "grad_norm": 1.4150489568710327,
      "learning_rate": 5.842816035156432e-05,
      "loss": 1.9269,
      "step": 4958
    },
    {
      "epoch": 0.8933123170457105,
      "grad_norm": 1.6062489748001099,
      "learning_rate": 5.841420112836479e-05,
      "loss": 2.1852,
      "step": 4959
    },
    {
      "epoch": 0.8934924566539068,
      "grad_norm": 1.503406286239624,
      "learning_rate": 5.840024123018213e-05,
      "loss": 1.8622,
      "step": 4960
    },
    {
      "epoch": 0.8936725962621032,
      "grad_norm": 1.3765223026275635,
      "learning_rate": 5.838628065813622e-05,
      "loss": 1.8461,
      "step": 4961
    },
    {
      "epoch": 0.8938527358702995,
      "grad_norm": 1.267340898513794,
      "learning_rate": 5.8372319413346944e-05,
      "loss": 1.5935,
      "step": 4962
    },
    {
      "epoch": 0.8940328754784959,
      "grad_norm": 1.3180469274520874,
      "learning_rate": 5.8358357496934304e-05,
      "loss": 1.8769,
      "step": 4963
    },
    {
      "epoch": 0.8942130150866922,
      "grad_norm": 1.2231675386428833,
      "learning_rate": 5.83443949100183e-05,
      "loss": 1.6831,
      "step": 4964
    },
    {
      "epoch": 0.8943931546948886,
      "grad_norm": 1.4154112339019775,
      "learning_rate": 5.8330431653718986e-05,
      "loss": 1.7242,
      "step": 4965
    },
    {
      "epoch": 0.8945732943030849,
      "grad_norm": 1.4427021741867065,
      "learning_rate": 5.831646772915651e-05,
      "loss": 1.6218,
      "step": 4966
    },
    {
      "epoch": 0.8947534339112813,
      "grad_norm": 1.2480283975601196,
      "learning_rate": 5.830250313745104e-05,
      "loss": 1.5665,
      "step": 4967
    },
    {
      "epoch": 0.8949335735194776,
      "grad_norm": 1.3867474794387817,
      "learning_rate": 5.828853787972281e-05,
      "loss": 1.7147,
      "step": 4968
    },
    {
      "epoch": 0.895113713127674,
      "grad_norm": 1.3857214450836182,
      "learning_rate": 5.827457195709212e-05,
      "loss": 1.6213,
      "step": 4969
    },
    {
      "epoch": 0.8952938527358703,
      "grad_norm": 1.2099359035491943,
      "learning_rate": 5.8260605370679286e-05,
      "loss": 1.5305,
      "step": 4970
    },
    {
      "epoch": 0.8954739923440667,
      "grad_norm": 1.4401071071624756,
      "learning_rate": 5.8246638121604715e-05,
      "loss": 1.822,
      "step": 4971
    },
    {
      "epoch": 0.895654131952263,
      "grad_norm": 1.2437294721603394,
      "learning_rate": 5.8232670210988847e-05,
      "loss": 1.2666,
      "step": 4972
    },
    {
      "epoch": 0.8958342715604594,
      "grad_norm": 1.378402590751648,
      "learning_rate": 5.82187016399522e-05,
      "loss": 1.6252,
      "step": 4973
    },
    {
      "epoch": 0.8960144111686557,
      "grad_norm": 1.3699381351470947,
      "learning_rate": 5.8204732409615304e-05,
      "loss": 1.4978,
      "step": 4974
    },
    {
      "epoch": 0.8961945507768521,
      "grad_norm": 1.4028576612472534,
      "learning_rate": 5.819076252109876e-05,
      "loss": 1.6407,
      "step": 4975
    },
    {
      "epoch": 0.8963746903850485,
      "grad_norm": 1.397886872291565,
      "learning_rate": 5.8176791975523235e-05,
      "loss": 1.5976,
      "step": 4976
    },
    {
      "epoch": 0.8965548299932448,
      "grad_norm": 1.4487146139144897,
      "learning_rate": 5.816282077400945e-05,
      "loss": 1.492,
      "step": 4977
    },
    {
      "epoch": 0.8967349696014412,
      "grad_norm": 1.2671177387237549,
      "learning_rate": 5.814884891767816e-05,
      "loss": 1.4081,
      "step": 4978
    },
    {
      "epoch": 0.8969151092096375,
      "grad_norm": 1.3465325832366943,
      "learning_rate": 5.813487640765018e-05,
      "loss": 1.6068,
      "step": 4979
    },
    {
      "epoch": 0.8970952488178339,
      "grad_norm": 1.4604675769805908,
      "learning_rate": 5.812090324504638e-05,
      "loss": 1.7003,
      "step": 4980
    },
    {
      "epoch": 0.8972753884260302,
      "grad_norm": 1.330579161643982,
      "learning_rate": 5.8106929430987653e-05,
      "loss": 1.445,
      "step": 4981
    },
    {
      "epoch": 0.8974555280342266,
      "grad_norm": 1.292406678199768,
      "learning_rate": 5.8092954966595024e-05,
      "loss": 1.4893,
      "step": 4982
    },
    {
      "epoch": 0.8976356676424229,
      "grad_norm": 1.3988697528839111,
      "learning_rate": 5.8078979852989466e-05,
      "loss": 1.684,
      "step": 4983
    },
    {
      "epoch": 0.8978158072506193,
      "grad_norm": 1.3910832405090332,
      "learning_rate": 5.80650040912921e-05,
      "loss": 1.7591,
      "step": 4984
    },
    {
      "epoch": 0.8979959468588156,
      "grad_norm": 1.4246978759765625,
      "learning_rate": 5.8051027682624026e-05,
      "loss": 1.7317,
      "step": 4985
    },
    {
      "epoch": 0.898176086467012,
      "grad_norm": 1.3134583234786987,
      "learning_rate": 5.803705062810644e-05,
      "loss": 1.3337,
      "step": 4986
    },
    {
      "epoch": 0.8983562260752083,
      "grad_norm": 1.3355578184127808,
      "learning_rate": 5.802307292886056e-05,
      "loss": 1.4894,
      "step": 4987
    },
    {
      "epoch": 0.8985363656834047,
      "grad_norm": 1.3627285957336426,
      "learning_rate": 5.800909458600767e-05,
      "loss": 1.46,
      "step": 4988
    },
    {
      "epoch": 0.898716505291601,
      "grad_norm": 1.3580554723739624,
      "learning_rate": 5.799511560066912e-05,
      "loss": 1.679,
      "step": 4989
    },
    {
      "epoch": 0.8988966448997974,
      "grad_norm": 1.4573010206222534,
      "learning_rate": 5.798113597396629e-05,
      "loss": 1.54,
      "step": 4990
    },
    {
      "epoch": 0.8990767845079937,
      "grad_norm": 1.4165380001068115,
      "learning_rate": 5.7967155707020623e-05,
      "loss": 1.5047,
      "step": 4991
    },
    {
      "epoch": 0.8992569241161901,
      "grad_norm": 1.4199260473251343,
      "learning_rate": 5.7953174800953604e-05,
      "loss": 1.5042,
      "step": 4992
    },
    {
      "epoch": 0.8994370637243864,
      "grad_norm": 1.4115427732467651,
      "learning_rate": 5.793919325688678e-05,
      "loss": 1.3498,
      "step": 4993
    },
    {
      "epoch": 0.8996172033325828,
      "grad_norm": 1.4568222761154175,
      "learning_rate": 5.7925211075941744e-05,
      "loss": 1.3783,
      "step": 4994
    },
    {
      "epoch": 0.8997973429407791,
      "grad_norm": 1.3275940418243408,
      "learning_rate": 5.791122825924011e-05,
      "loss": 1.5193,
      "step": 4995
    },
    {
      "epoch": 0.8999774825489755,
      "grad_norm": 1.3888188600540161,
      "learning_rate": 5.7897244807903636e-05,
      "loss": 1.5633,
      "step": 4996
    },
    {
      "epoch": 0.9001576221571718,
      "grad_norm": 1.4054131507873535,
      "learning_rate": 5.7883260723054e-05,
      "loss": 1.4336,
      "step": 4997
    },
    {
      "epoch": 0.9003377617653682,
      "grad_norm": 1.4035102128982544,
      "learning_rate": 5.7869276005813044e-05,
      "loss": 1.2238,
      "step": 4998
    },
    {
      "epoch": 0.9005179013735645,
      "grad_norm": 1.352142572402954,
      "learning_rate": 5.785529065730258e-05,
      "loss": 1.4641,
      "step": 4999
    },
    {
      "epoch": 0.9006980409817609,
      "grad_norm": 1.409900188446045,
      "learning_rate": 5.784130467864455e-05,
      "loss": 1.5026,
      "step": 5000
    },
    {
      "epoch": 0.9008781805899572,
      "grad_norm": 1.3493738174438477,
      "learning_rate": 5.782731807096088e-05,
      "loss": 2.1084,
      "step": 5001
    },
    {
      "epoch": 0.9010583201981536,
      "grad_norm": 1.326295256614685,
      "learning_rate": 5.7813330835373545e-05,
      "loss": 2.0093,
      "step": 5002
    },
    {
      "epoch": 0.9012384598063499,
      "grad_norm": 1.1491564512252808,
      "learning_rate": 5.779934297300463e-05,
      "loss": 1.8514,
      "step": 5003
    },
    {
      "epoch": 0.9014185994145463,
      "grad_norm": 1.203399658203125,
      "learning_rate": 5.778535448497622e-05,
      "loss": 1.7148,
      "step": 5004
    },
    {
      "epoch": 0.9015987390227427,
      "grad_norm": 1.3038862943649292,
      "learning_rate": 5.7771365372410454e-05,
      "loss": 2.0116,
      "step": 5005
    },
    {
      "epoch": 0.901778878630939,
      "grad_norm": 1.2065067291259766,
      "learning_rate": 5.7757375636429564e-05,
      "loss": 1.9362,
      "step": 5006
    },
    {
      "epoch": 0.9019590182391354,
      "grad_norm": 1.418507695198059,
      "learning_rate": 5.774338527815577e-05,
      "loss": 1.9639,
      "step": 5007
    },
    {
      "epoch": 0.9021391578473317,
      "grad_norm": 1.4580485820770264,
      "learning_rate": 5.772939429871138e-05,
      "loss": 1.8525,
      "step": 5008
    },
    {
      "epoch": 0.9023192974555281,
      "grad_norm": 1.5523029565811157,
      "learning_rate": 5.771540269921874e-05,
      "loss": 2.0376,
      "step": 5009
    },
    {
      "epoch": 0.9024994370637244,
      "grad_norm": 1.748618721961975,
      "learning_rate": 5.770141048080026e-05,
      "loss": 2.2565,
      "step": 5010
    },
    {
      "epoch": 0.9026795766719208,
      "grad_norm": 1.416452407836914,
      "learning_rate": 5.768741764457836e-05,
      "loss": 1.7559,
      "step": 5011
    },
    {
      "epoch": 0.9028597162801171,
      "grad_norm": 1.4235934019088745,
      "learning_rate": 5.7673424191675595e-05,
      "loss": 1.6576,
      "step": 5012
    },
    {
      "epoch": 0.9030398558883135,
      "grad_norm": 1.2068952322006226,
      "learning_rate": 5.7659430123214455e-05,
      "loss": 1.3391,
      "step": 5013
    },
    {
      "epoch": 0.9032199954965098,
      "grad_norm": 1.3004122972488403,
      "learning_rate": 5.764543544031757e-05,
      "loss": 1.3112,
      "step": 5014
    },
    {
      "epoch": 0.9034001351047062,
      "grad_norm": 1.3666245937347412,
      "learning_rate": 5.763144014410759e-05,
      "loss": 1.7867,
      "step": 5015
    },
    {
      "epoch": 0.9035802747129025,
      "grad_norm": 1.366073727607727,
      "learning_rate": 5.761744423570716e-05,
      "loss": 1.5905,
      "step": 5016
    },
    {
      "epoch": 0.9037604143210989,
      "grad_norm": 1.282131314277649,
      "learning_rate": 5.76034477162391e-05,
      "loss": 1.4685,
      "step": 5017
    },
    {
      "epoch": 0.9039405539292952,
      "grad_norm": 1.3294016122817993,
      "learning_rate": 5.7589450586826165e-05,
      "loss": 1.65,
      "step": 5018
    },
    {
      "epoch": 0.9041206935374916,
      "grad_norm": 1.3729841709136963,
      "learning_rate": 5.7575452848591196e-05,
      "loss": 1.6355,
      "step": 5019
    },
    {
      "epoch": 0.9043008331456879,
      "grad_norm": 1.4479531049728394,
      "learning_rate": 5.7561454502657075e-05,
      "loss": 1.7243,
      "step": 5020
    },
    {
      "epoch": 0.9044809727538843,
      "grad_norm": 1.2787152528762817,
      "learning_rate": 5.754745555014677e-05,
      "loss": 1.5639,
      "step": 5021
    },
    {
      "epoch": 0.9046611123620806,
      "grad_norm": 1.2977980375289917,
      "learning_rate": 5.753345599218326e-05,
      "loss": 1.534,
      "step": 5022
    },
    {
      "epoch": 0.904841251970277,
      "grad_norm": 1.342494249343872,
      "learning_rate": 5.751945582988958e-05,
      "loss": 1.4455,
      "step": 5023
    },
    {
      "epoch": 0.9050213915784733,
      "grad_norm": 1.2599451541900635,
      "learning_rate": 5.75054550643888e-05,
      "loss": 1.2797,
      "step": 5024
    },
    {
      "epoch": 0.9052015311866697,
      "grad_norm": 1.482111930847168,
      "learning_rate": 5.749145369680407e-05,
      "loss": 1.6278,
      "step": 5025
    },
    {
      "epoch": 0.905381670794866,
      "grad_norm": 1.3941726684570312,
      "learning_rate": 5.7477451728258605e-05,
      "loss": 1.552,
      "step": 5026
    },
    {
      "epoch": 0.9055618104030624,
      "grad_norm": 1.3854631185531616,
      "learning_rate": 5.746344915987557e-05,
      "loss": 1.4391,
      "step": 5027
    },
    {
      "epoch": 0.9057419500112587,
      "grad_norm": 1.315888524055481,
      "learning_rate": 5.7449445992778304e-05,
      "loss": 1.4151,
      "step": 5028
    },
    {
      "epoch": 0.9059220896194551,
      "grad_norm": 1.4299246072769165,
      "learning_rate": 5.74354422280901e-05,
      "loss": 1.6382,
      "step": 5029
    },
    {
      "epoch": 0.9061022292276514,
      "grad_norm": 1.3035376071929932,
      "learning_rate": 5.742143786693434e-05,
      "loss": 1.4697,
      "step": 5030
    },
    {
      "epoch": 0.9062823688358478,
      "grad_norm": 1.3902029991149902,
      "learning_rate": 5.7407432910434444e-05,
      "loss": 1.6825,
      "step": 5031
    },
    {
      "epoch": 0.9064625084440441,
      "grad_norm": 1.3576289415359497,
      "learning_rate": 5.739342735971389e-05,
      "loss": 1.4541,
      "step": 5032
    },
    {
      "epoch": 0.9066426480522405,
      "grad_norm": 1.5524572134017944,
      "learning_rate": 5.737942121589619e-05,
      "loss": 1.9006,
      "step": 5033
    },
    {
      "epoch": 0.9068227876604369,
      "grad_norm": 1.3840633630752563,
      "learning_rate": 5.736541448010493e-05,
      "loss": 1.5157,
      "step": 5034
    },
    {
      "epoch": 0.9070029272686332,
      "grad_norm": 1.4073817729949951,
      "learning_rate": 5.73514071534637e-05,
      "loss": 1.7386,
      "step": 5035
    },
    {
      "epoch": 0.9071830668768296,
      "grad_norm": 1.3186254501342773,
      "learning_rate": 5.733739923709618e-05,
      "loss": 1.3219,
      "step": 5036
    },
    {
      "epoch": 0.9073632064850259,
      "grad_norm": 1.3760497570037842,
      "learning_rate": 5.732339073212605e-05,
      "loss": 1.527,
      "step": 5037
    },
    {
      "epoch": 0.9075433460932223,
      "grad_norm": 1.493990182876587,
      "learning_rate": 5.7309381639677085e-05,
      "loss": 1.7089,
      "step": 5038
    },
    {
      "epoch": 0.9077234857014186,
      "grad_norm": 1.4230737686157227,
      "learning_rate": 5.729537196087309e-05,
      "loss": 1.6311,
      "step": 5039
    },
    {
      "epoch": 0.907903625309615,
      "grad_norm": 1.4095138311386108,
      "learning_rate": 5.728136169683791e-05,
      "loss": 1.4789,
      "step": 5040
    },
    {
      "epoch": 0.9080837649178113,
      "grad_norm": 1.494604229927063,
      "learning_rate": 5.726735084869543e-05,
      "loss": 1.4799,
      "step": 5041
    },
    {
      "epoch": 0.9082639045260077,
      "grad_norm": 1.3156670331954956,
      "learning_rate": 5.725333941756962e-05,
      "loss": 1.2875,
      "step": 5042
    },
    {
      "epoch": 0.908444044134204,
      "grad_norm": 1.2558436393737793,
      "learning_rate": 5.7239327404584454e-05,
      "loss": 1.2869,
      "step": 5043
    },
    {
      "epoch": 0.9086241837424004,
      "grad_norm": 1.484230399131775,
      "learning_rate": 5.722531481086395e-05,
      "loss": 1.7496,
      "step": 5044
    },
    {
      "epoch": 0.9088043233505967,
      "grad_norm": 1.3950295448303223,
      "learning_rate": 5.7211301637532245e-05,
      "loss": 1.5871,
      "step": 5045
    },
    {
      "epoch": 0.9089844629587931,
      "grad_norm": 1.4383394718170166,
      "learning_rate": 5.7197287885713405e-05,
      "loss": 1.2528,
      "step": 5046
    },
    {
      "epoch": 0.9091646025669894,
      "grad_norm": 1.4085705280303955,
      "learning_rate": 5.718327355653166e-05,
      "loss": 1.6171,
      "step": 5047
    },
    {
      "epoch": 0.9093447421751858,
      "grad_norm": 1.2919187545776367,
      "learning_rate": 5.716925865111118e-05,
      "loss": 1.3425,
      "step": 5048
    },
    {
      "epoch": 0.9095248817833821,
      "grad_norm": 1.2899022102355957,
      "learning_rate": 5.7155243170576276e-05,
      "loss": 1.205,
      "step": 5049
    },
    {
      "epoch": 0.9097050213915785,
      "grad_norm": 1.4842193126678467,
      "learning_rate": 5.714122711605127e-05,
      "loss": 1.6139,
      "step": 5050
    },
    {
      "epoch": 0.9098851609997748,
      "grad_norm": 1.2801039218902588,
      "learning_rate": 5.712721048866049e-05,
      "loss": 1.7893,
      "step": 5051
    },
    {
      "epoch": 0.9100653006079712,
      "grad_norm": 1.4865806102752686,
      "learning_rate": 5.711319328952836e-05,
      "loss": 2.3504,
      "step": 5052
    },
    {
      "epoch": 0.9102454402161675,
      "grad_norm": 1.2550746202468872,
      "learning_rate": 5.7099175519779304e-05,
      "loss": 1.9261,
      "step": 5053
    },
    {
      "epoch": 0.9104255798243639,
      "grad_norm": 1.2748281955718994,
      "learning_rate": 5.7085157180537865e-05,
      "loss": 1.7573,
      "step": 5054
    },
    {
      "epoch": 0.9106057194325602,
      "grad_norm": 1.2243000268936157,
      "learning_rate": 5.707113827292857e-05,
      "loss": 1.778,
      "step": 5055
    },
    {
      "epoch": 0.9107858590407566,
      "grad_norm": 1.3699009418487549,
      "learning_rate": 5.7057118798076004e-05,
      "loss": 2.1405,
      "step": 5056
    },
    {
      "epoch": 0.9109659986489529,
      "grad_norm": 1.46440851688385,
      "learning_rate": 5.7043098757104805e-05,
      "loss": 1.959,
      "step": 5057
    },
    {
      "epoch": 0.9111461382571493,
      "grad_norm": 1.5501600503921509,
      "learning_rate": 5.702907815113966e-05,
      "loss": 2.0403,
      "step": 5058
    },
    {
      "epoch": 0.9113262778653456,
      "grad_norm": 1.734147310256958,
      "learning_rate": 5.701505698130528e-05,
      "loss": 2.2871,
      "step": 5059
    },
    {
      "epoch": 0.911506417473542,
      "grad_norm": 1.8098869323730469,
      "learning_rate": 5.700103524872644e-05,
      "loss": 2.0413,
      "step": 5060
    },
    {
      "epoch": 0.9116865570817384,
      "grad_norm": 1.3115081787109375,
      "learning_rate": 5.698701295452798e-05,
      "loss": 1.6129,
      "step": 5061
    },
    {
      "epoch": 0.9118666966899347,
      "grad_norm": 1.2694413661956787,
      "learning_rate": 5.6972990099834736e-05,
      "loss": 1.6414,
      "step": 5062
    },
    {
      "epoch": 0.9120468362981311,
      "grad_norm": 1.2579491138458252,
      "learning_rate": 5.695896668577163e-05,
      "loss": 1.5723,
      "step": 5063
    },
    {
      "epoch": 0.9122269759063274,
      "grad_norm": 1.39626145362854,
      "learning_rate": 5.694494271346359e-05,
      "loss": 1.483,
      "step": 5064
    },
    {
      "epoch": 0.9124071155145238,
      "grad_norm": 1.137127161026001,
      "learning_rate": 5.6930918184035645e-05,
      "loss": 1.2465,
      "step": 5065
    },
    {
      "epoch": 0.9125872551227201,
      "grad_norm": 1.3818987607955933,
      "learning_rate": 5.6916893098612814e-05,
      "loss": 1.7448,
      "step": 5066
    },
    {
      "epoch": 0.9127673947309165,
      "grad_norm": 1.291425347328186,
      "learning_rate": 5.690286745832019e-05,
      "loss": 1.6185,
      "step": 5067
    },
    {
      "epoch": 0.9129475343391128,
      "grad_norm": 1.3024672269821167,
      "learning_rate": 5.688884126428289e-05,
      "loss": 1.58,
      "step": 5068
    },
    {
      "epoch": 0.9131276739473092,
      "grad_norm": 1.1630114316940308,
      "learning_rate": 5.6874814517626094e-05,
      "loss": 1.3182,
      "step": 5069
    },
    {
      "epoch": 0.9133078135555055,
      "grad_norm": 1.2996869087219238,
      "learning_rate": 5.686078721947504e-05,
      "loss": 1.3772,
      "step": 5070
    },
    {
      "epoch": 0.9134879531637019,
      "grad_norm": 1.221467137336731,
      "learning_rate": 5.684675937095498e-05,
      "loss": 1.2569,
      "step": 5071
    },
    {
      "epoch": 0.9136680927718982,
      "grad_norm": 1.3229528665542603,
      "learning_rate": 5.683273097319122e-05,
      "loss": 1.624,
      "step": 5072
    },
    {
      "epoch": 0.9138482323800946,
      "grad_norm": 1.3927230834960938,
      "learning_rate": 5.68187020273091e-05,
      "loss": 1.6236,
      "step": 5073
    },
    {
      "epoch": 0.9140283719882909,
      "grad_norm": 1.4482344388961792,
      "learning_rate": 5.680467253443402e-05,
      "loss": 1.5324,
      "step": 5074
    },
    {
      "epoch": 0.9142085115964873,
      "grad_norm": 1.3677200078964233,
      "learning_rate": 5.679064249569144e-05,
      "loss": 1.5572,
      "step": 5075
    },
    {
      "epoch": 0.9143886512046836,
      "grad_norm": 1.3752690553665161,
      "learning_rate": 5.67766119122068e-05,
      "loss": 1.5922,
      "step": 5076
    },
    {
      "epoch": 0.91456879081288,
      "grad_norm": 1.2365682125091553,
      "learning_rate": 5.676258078510569e-05,
      "loss": 1.5804,
      "step": 5077
    },
    {
      "epoch": 0.9147489304210763,
      "grad_norm": 1.3693547248840332,
      "learning_rate": 5.674854911551362e-05,
      "loss": 1.4754,
      "step": 5078
    },
    {
      "epoch": 0.9149290700292727,
      "grad_norm": 1.393280029296875,
      "learning_rate": 5.673451690455622e-05,
      "loss": 1.4322,
      "step": 5079
    },
    {
      "epoch": 0.915109209637469,
      "grad_norm": 1.3403124809265137,
      "learning_rate": 5.6720484153359174e-05,
      "loss": 1.4891,
      "step": 5080
    },
    {
      "epoch": 0.9152893492456654,
      "grad_norm": 1.2650532722473145,
      "learning_rate": 5.6706450863048145e-05,
      "loss": 1.5332,
      "step": 5081
    },
    {
      "epoch": 0.9154694888538617,
      "grad_norm": 1.4569473266601562,
      "learning_rate": 5.66924170347489e-05,
      "loss": 1.7839,
      "step": 5082
    },
    {
      "epoch": 0.9156496284620581,
      "grad_norm": 1.3612486124038696,
      "learning_rate": 5.667838266958723e-05,
      "loss": 1.5323,
      "step": 5083
    },
    {
      "epoch": 0.9158297680702544,
      "grad_norm": 1.4201631546020508,
      "learning_rate": 5.666434776868895e-05,
      "loss": 1.5092,
      "step": 5084
    },
    {
      "epoch": 0.9160099076784508,
      "grad_norm": 1.459028720855713,
      "learning_rate": 5.6650312333179923e-05,
      "loss": 1.5687,
      "step": 5085
    },
    {
      "epoch": 0.9161900472866471,
      "grad_norm": 1.4282561540603638,
      "learning_rate": 5.6636276364186105e-05,
      "loss": 1.3516,
      "step": 5086
    },
    {
      "epoch": 0.9163701868948435,
      "grad_norm": 1.4773622751235962,
      "learning_rate": 5.662223986283341e-05,
      "loss": 1.6518,
      "step": 5087
    },
    {
      "epoch": 0.9165503265030398,
      "grad_norm": 1.4208288192749023,
      "learning_rate": 5.660820283024788e-05,
      "loss": 1.7851,
      "step": 5088
    },
    {
      "epoch": 0.9167304661112362,
      "grad_norm": 1.382665753364563,
      "learning_rate": 5.659416526755552e-05,
      "loss": 1.4614,
      "step": 5089
    },
    {
      "epoch": 0.9169106057194326,
      "grad_norm": 1.3633079528808594,
      "learning_rate": 5.658012717588244e-05,
      "loss": 1.4254,
      "step": 5090
    },
    {
      "epoch": 0.9170907453276289,
      "grad_norm": 1.4343020915985107,
      "learning_rate": 5.6566088556354754e-05,
      "loss": 1.5308,
      "step": 5091
    },
    {
      "epoch": 0.9172708849358253,
      "grad_norm": 1.2589569091796875,
      "learning_rate": 5.655204941009864e-05,
      "loss": 1.3062,
      "step": 5092
    },
    {
      "epoch": 0.9174510245440216,
      "grad_norm": 1.4107121229171753,
      "learning_rate": 5.653800973824034e-05,
      "loss": 1.5481,
      "step": 5093
    },
    {
      "epoch": 0.917631164152218,
      "grad_norm": 1.4062985181808472,
      "learning_rate": 5.652396954190608e-05,
      "loss": 1.573,
      "step": 5094
    },
    {
      "epoch": 0.9178113037604143,
      "grad_norm": 1.4348030090332031,
      "learning_rate": 5.650992882222212e-05,
      "loss": 1.5402,
      "step": 5095
    },
    {
      "epoch": 0.9179914433686107,
      "grad_norm": 1.2970614433288574,
      "learning_rate": 5.6495887580314876e-05,
      "loss": 1.5557,
      "step": 5096
    },
    {
      "epoch": 0.918171582976807,
      "grad_norm": 1.3836398124694824,
      "learning_rate": 5.6481845817310685e-05,
      "loss": 1.269,
      "step": 5097
    },
    {
      "epoch": 0.9183517225850034,
      "grad_norm": 1.4323362112045288,
      "learning_rate": 5.6467803534335986e-05,
      "loss": 1.6282,
      "step": 5098
    },
    {
      "epoch": 0.9185318621931997,
      "grad_norm": 1.3195542097091675,
      "learning_rate": 5.645376073251724e-05,
      "loss": 1.2351,
      "step": 5099
    },
    {
      "epoch": 0.9187120018013961,
      "grad_norm": 1.3150973320007324,
      "learning_rate": 5.643971741298094e-05,
      "loss": 1.5404,
      "step": 5100
    },
    {
      "epoch": 0.9188921414095924,
      "grad_norm": 1.3054282665252686,
      "learning_rate": 5.642567357685365e-05,
      "loss": 2.3153,
      "step": 5101
    },
    {
      "epoch": 0.9190722810177888,
      "grad_norm": 1.2936956882476807,
      "learning_rate": 5.6411629225261954e-05,
      "loss": 1.9611,
      "step": 5102
    },
    {
      "epoch": 0.9192524206259851,
      "grad_norm": 1.2296570539474487,
      "learning_rate": 5.639758435933249e-05,
      "loss": 2.037,
      "step": 5103
    },
    {
      "epoch": 0.9194325602341815,
      "grad_norm": 1.2555570602416992,
      "learning_rate": 5.638353898019192e-05,
      "loss": 1.9361,
      "step": 5104
    },
    {
      "epoch": 0.9196126998423778,
      "grad_norm": 1.2353285551071167,
      "learning_rate": 5.6369493088966964e-05,
      "loss": 1.8856,
      "step": 5105
    },
    {
      "epoch": 0.9197928394505742,
      "grad_norm": 1.2829554080963135,
      "learning_rate": 5.6355446686784375e-05,
      "loss": 2.0834,
      "step": 5106
    },
    {
      "epoch": 0.9199729790587705,
      "grad_norm": 1.3782570362091064,
      "learning_rate": 5.634139977477094e-05,
      "loss": 2.1177,
      "step": 5107
    },
    {
      "epoch": 0.9201531186669669,
      "grad_norm": 1.4753156900405884,
      "learning_rate": 5.632735235405352e-05,
      "loss": 1.8709,
      "step": 5108
    },
    {
      "epoch": 0.9203332582751632,
      "grad_norm": 1.5350147485733032,
      "learning_rate": 5.631330442575896e-05,
      "loss": 1.9023,
      "step": 5109
    },
    {
      "epoch": 0.9205133978833596,
      "grad_norm": 1.48201322555542,
      "learning_rate": 5.629925599101422e-05,
      "loss": 1.7216,
      "step": 5110
    },
    {
      "epoch": 0.9206935374915559,
      "grad_norm": 1.543350338935852,
      "learning_rate": 5.62852070509462e-05,
      "loss": 2.1403,
      "step": 5111
    },
    {
      "epoch": 0.9208736770997523,
      "grad_norm": 1.331264853477478,
      "learning_rate": 5.6271157606681954e-05,
      "loss": 1.7228,
      "step": 5112
    },
    {
      "epoch": 0.9210538167079486,
      "grad_norm": 1.2444875240325928,
      "learning_rate": 5.6257107659348484e-05,
      "loss": 1.6073,
      "step": 5113
    },
    {
      "epoch": 0.921233956316145,
      "grad_norm": 1.1544668674468994,
      "learning_rate": 5.624305721007289e-05,
      "loss": 1.384,
      "step": 5114
    },
    {
      "epoch": 0.9214140959243413,
      "grad_norm": 1.244156837463379,
      "learning_rate": 5.62290062599823e-05,
      "loss": 1.5327,
      "step": 5115
    },
    {
      "epoch": 0.9215942355325377,
      "grad_norm": 1.4060598611831665,
      "learning_rate": 5.621495481020386e-05,
      "loss": 1.4705,
      "step": 5116
    },
    {
      "epoch": 0.921774375140734,
      "grad_norm": 1.629783034324646,
      "learning_rate": 5.6200902861864756e-05,
      "loss": 1.5371,
      "step": 5117
    },
    {
      "epoch": 0.9219545147489304,
      "grad_norm": 1.2417926788330078,
      "learning_rate": 5.618685041609225e-05,
      "loss": 1.3576,
      "step": 5118
    },
    {
      "epoch": 0.9221346543571268,
      "grad_norm": 1.4018644094467163,
      "learning_rate": 5.617279747401363e-05,
      "loss": 1.5638,
      "step": 5119
    },
    {
      "epoch": 0.9223147939653231,
      "grad_norm": 1.2647839784622192,
      "learning_rate": 5.61587440367562e-05,
      "loss": 1.4678,
      "step": 5120
    },
    {
      "epoch": 0.9224949335735195,
      "grad_norm": 1.2668280601501465,
      "learning_rate": 5.6144690105447305e-05,
      "loss": 1.3495,
      "step": 5121
    },
    {
      "epoch": 0.9226750731817158,
      "grad_norm": 1.2567169666290283,
      "learning_rate": 5.6130635681214386e-05,
      "loss": 1.325,
      "step": 5122
    },
    {
      "epoch": 0.9228552127899122,
      "grad_norm": 1.3366979360580444,
      "learning_rate": 5.6116580765184834e-05,
      "loss": 1.4925,
      "step": 5123
    },
    {
      "epoch": 0.9230353523981085,
      "grad_norm": 1.288009524345398,
      "learning_rate": 5.610252535848617e-05,
      "loss": 1.4696,
      "step": 5124
    },
    {
      "epoch": 0.9232154920063049,
      "grad_norm": 1.2684028148651123,
      "learning_rate": 5.608846946224589e-05,
      "loss": 1.4457,
      "step": 5125
    },
    {
      "epoch": 0.9233956316145012,
      "grad_norm": 1.3418757915496826,
      "learning_rate": 5.607441307759156e-05,
      "loss": 1.6583,
      "step": 5126
    },
    {
      "epoch": 0.9235757712226976,
      "grad_norm": 1.343349575996399,
      "learning_rate": 5.6060356205650766e-05,
      "loss": 1.4726,
      "step": 5127
    },
    {
      "epoch": 0.9237559108308939,
      "grad_norm": 1.4082485437393188,
      "learning_rate": 5.604629884755115e-05,
      "loss": 1.4287,
      "step": 5128
    },
    {
      "epoch": 0.9239360504390903,
      "grad_norm": 1.2966614961624146,
      "learning_rate": 5.603224100442038e-05,
      "loss": 1.3629,
      "step": 5129
    },
    {
      "epoch": 0.9241161900472866,
      "grad_norm": 1.2937183380126953,
      "learning_rate": 5.6018182677386176e-05,
      "loss": 1.2554,
      "step": 5130
    },
    {
      "epoch": 0.924296329655483,
      "grad_norm": 1.3303674459457397,
      "learning_rate": 5.60041238675763e-05,
      "loss": 1.6817,
      "step": 5131
    },
    {
      "epoch": 0.9244764692636793,
      "grad_norm": 1.4202079772949219,
      "learning_rate": 5.5990064576118525e-05,
      "loss": 1.5546,
      "step": 5132
    },
    {
      "epoch": 0.9246566088718757,
      "grad_norm": 1.3486336469650269,
      "learning_rate": 5.5976004804140695e-05,
      "loss": 1.4283,
      "step": 5133
    },
    {
      "epoch": 0.924836748480072,
      "grad_norm": 1.3125056028366089,
      "learning_rate": 5.596194455277065e-05,
      "loss": 1.4568,
      "step": 5134
    },
    {
      "epoch": 0.9250168880882684,
      "grad_norm": 1.5667637586593628,
      "learning_rate": 5.594788382313634e-05,
      "loss": 1.8937,
      "step": 5135
    },
    {
      "epoch": 0.9251970276964647,
      "grad_norm": 1.442713975906372,
      "learning_rate": 5.593382261636567e-05,
      "loss": 1.7078,
      "step": 5136
    },
    {
      "epoch": 0.9253771673046611,
      "grad_norm": 1.2807390689849854,
      "learning_rate": 5.591976093358665e-05,
      "loss": 1.4063,
      "step": 5137
    },
    {
      "epoch": 0.9255573069128574,
      "grad_norm": 1.4069725275039673,
      "learning_rate": 5.590569877592729e-05,
      "loss": 1.4569,
      "step": 5138
    },
    {
      "epoch": 0.9257374465210538,
      "grad_norm": 1.344852328300476,
      "learning_rate": 5.589163614451565e-05,
      "loss": 1.6277,
      "step": 5139
    },
    {
      "epoch": 0.9259175861292501,
      "grad_norm": 1.4091660976409912,
      "learning_rate": 5.587757304047983e-05,
      "loss": 1.5884,
      "step": 5140
    },
    {
      "epoch": 0.9260977257374465,
      "grad_norm": 1.3231937885284424,
      "learning_rate": 5.586350946494795e-05,
      "loss": 1.3841,
      "step": 5141
    },
    {
      "epoch": 0.9262778653456428,
      "grad_norm": 1.48690664768219,
      "learning_rate": 5.5849445419048216e-05,
      "loss": 1.5442,
      "step": 5142
    },
    {
      "epoch": 0.9264580049538392,
      "grad_norm": 1.3540806770324707,
      "learning_rate": 5.583538090390882e-05,
      "loss": 1.6341,
      "step": 5143
    },
    {
      "epoch": 0.9266381445620355,
      "grad_norm": 1.44573974609375,
      "learning_rate": 5.582131592065799e-05,
      "loss": 1.6569,
      "step": 5144
    },
    {
      "epoch": 0.9268182841702319,
      "grad_norm": 1.4059885740280151,
      "learning_rate": 5.580725047042405e-05,
      "loss": 1.5414,
      "step": 5145
    },
    {
      "epoch": 0.9269984237784282,
      "grad_norm": 1.3429491519927979,
      "learning_rate": 5.579318455433529e-05,
      "loss": 1.364,
      "step": 5146
    },
    {
      "epoch": 0.9271785633866246,
      "grad_norm": 1.2300504446029663,
      "learning_rate": 5.57791181735201e-05,
      "loss": 1.339,
      "step": 5147
    },
    {
      "epoch": 0.927358702994821,
      "grad_norm": 1.3269760608673096,
      "learning_rate": 5.5765051329106864e-05,
      "loss": 1.3428,
      "step": 5148
    },
    {
      "epoch": 0.9275388426030173,
      "grad_norm": 1.476166009902954,
      "learning_rate": 5.5750984022224015e-05,
      "loss": 1.328,
      "step": 5149
    },
    {
      "epoch": 0.9277189822112137,
      "grad_norm": 1.4089046716690063,
      "learning_rate": 5.5736916254000036e-05,
      "loss": 1.2934,
      "step": 5150
    },
    {
      "epoch": 0.92789912181941,
      "grad_norm": 1.237349271774292,
      "learning_rate": 5.572284802556341e-05,
      "loss": 1.688,
      "step": 5151
    },
    {
      "epoch": 0.9280792614276064,
      "grad_norm": 1.4099372625350952,
      "learning_rate": 5.5708779338042725e-05,
      "loss": 2.1427,
      "step": 5152
    },
    {
      "epoch": 0.9282594010358027,
      "grad_norm": 1.3852771520614624,
      "learning_rate": 5.569471019256654e-05,
      "loss": 1.8432,
      "step": 5153
    },
    {
      "epoch": 0.9284395406439991,
      "grad_norm": 1.2227874994277954,
      "learning_rate": 5.568064059026347e-05,
      "loss": 1.7545,
      "step": 5154
    },
    {
      "epoch": 0.9286196802521954,
      "grad_norm": 1.2323182821273804,
      "learning_rate": 5.5666570532262166e-05,
      "loss": 1.8028,
      "step": 5155
    },
    {
      "epoch": 0.9287998198603918,
      "grad_norm": 1.3899558782577515,
      "learning_rate": 5.565250001969136e-05,
      "loss": 1.8758,
      "step": 5156
    },
    {
      "epoch": 0.9289799594685881,
      "grad_norm": 1.551007628440857,
      "learning_rate": 5.563842905367974e-05,
      "loss": 2.0121,
      "step": 5157
    },
    {
      "epoch": 0.9291600990767845,
      "grad_norm": 1.4386353492736816,
      "learning_rate": 5.56243576353561e-05,
      "loss": 1.8246,
      "step": 5158
    },
    {
      "epoch": 0.9293402386849808,
      "grad_norm": 1.6415252685546875,
      "learning_rate": 5.5610285765849214e-05,
      "loss": 2.0491,
      "step": 5159
    },
    {
      "epoch": 0.9295203782931772,
      "grad_norm": 1.8444405794143677,
      "learning_rate": 5.5596213446287935e-05,
      "loss": 2.2638,
      "step": 5160
    },
    {
      "epoch": 0.9297005179013735,
      "grad_norm": 1.513251781463623,
      "learning_rate": 5.558214067780115e-05,
      "loss": 1.8008,
      "step": 5161
    },
    {
      "epoch": 0.9298806575095699,
      "grad_norm": 1.319232702255249,
      "learning_rate": 5.556806746151775e-05,
      "loss": 1.5633,
      "step": 5162
    },
    {
      "epoch": 0.9300607971177662,
      "grad_norm": 1.4036771059036255,
      "learning_rate": 5.5553993798566696e-05,
      "loss": 1.6758,
      "step": 5163
    },
    {
      "epoch": 0.9302409367259626,
      "grad_norm": 1.3583061695098877,
      "learning_rate": 5.553991969007698e-05,
      "loss": 1.5427,
      "step": 5164
    },
    {
      "epoch": 0.9304210763341589,
      "grad_norm": 1.3591159582138062,
      "learning_rate": 5.552584513717758e-05,
      "loss": 1.7411,
      "step": 5165
    },
    {
      "epoch": 0.9306012159423553,
      "grad_norm": 1.3450148105621338,
      "learning_rate": 5.551177014099759e-05,
      "loss": 1.6463,
      "step": 5166
    },
    {
      "epoch": 0.9307813555505516,
      "grad_norm": 1.3095580339431763,
      "learning_rate": 5.549769470266607e-05,
      "loss": 1.5818,
      "step": 5167
    },
    {
      "epoch": 0.930961495158748,
      "grad_norm": 1.3924058675765991,
      "learning_rate": 5.548361882331217e-05,
      "loss": 1.4851,
      "step": 5168
    },
    {
      "epoch": 0.9311416347669443,
      "grad_norm": 1.2917842864990234,
      "learning_rate": 5.546954250406504e-05,
      "loss": 1.3244,
      "step": 5169
    },
    {
      "epoch": 0.9313217743751407,
      "grad_norm": 1.3035470247268677,
      "learning_rate": 5.545546574605387e-05,
      "loss": 1.5441,
      "step": 5170
    },
    {
      "epoch": 0.931501913983337,
      "grad_norm": 1.3958535194396973,
      "learning_rate": 5.544138855040789e-05,
      "loss": 1.613,
      "step": 5171
    },
    {
      "epoch": 0.9316820535915334,
      "grad_norm": 1.4021573066711426,
      "learning_rate": 5.542731091825638e-05,
      "loss": 1.5517,
      "step": 5172
    },
    {
      "epoch": 0.9318621931997297,
      "grad_norm": 1.3312063217163086,
      "learning_rate": 5.5413232850728634e-05,
      "loss": 1.3708,
      "step": 5173
    },
    {
      "epoch": 0.9320423328079261,
      "grad_norm": 1.330467939376831,
      "learning_rate": 5.5399154348953977e-05,
      "loss": 1.4821,
      "step": 5174
    },
    {
      "epoch": 0.9322224724161225,
      "grad_norm": 1.4916101694107056,
      "learning_rate": 5.5385075414061795e-05,
      "loss": 1.7005,
      "step": 5175
    },
    {
      "epoch": 0.9324026120243188,
      "grad_norm": 1.3027833700180054,
      "learning_rate": 5.537099604718148e-05,
      "loss": 1.4846,
      "step": 5176
    },
    {
      "epoch": 0.9325827516325152,
      "grad_norm": 1.2656091451644897,
      "learning_rate": 5.5356916249442484e-05,
      "loss": 1.4978,
      "step": 5177
    },
    {
      "epoch": 0.9327628912407115,
      "grad_norm": 1.2995755672454834,
      "learning_rate": 5.534283602197425e-05,
      "loss": 1.4641,
      "step": 5178
    },
    {
      "epoch": 0.932943030848908,
      "grad_norm": 1.477933645248413,
      "learning_rate": 5.532875536590635e-05,
      "loss": 1.8889,
      "step": 5179
    },
    {
      "epoch": 0.9331231704571042,
      "grad_norm": 1.5306252241134644,
      "learning_rate": 5.5314674282368275e-05,
      "loss": 1.4539,
      "step": 5180
    },
    {
      "epoch": 0.9333033100653007,
      "grad_norm": 1.2270597219467163,
      "learning_rate": 5.5300592772489625e-05,
      "loss": 1.5087,
      "step": 5181
    },
    {
      "epoch": 0.933483449673497,
      "grad_norm": 1.3239166736602783,
      "learning_rate": 5.5286510837399995e-05,
      "loss": 1.5424,
      "step": 5182
    },
    {
      "epoch": 0.9336635892816934,
      "grad_norm": 1.3206679821014404,
      "learning_rate": 5.5272428478229035e-05,
      "loss": 1.6499,
      "step": 5183
    },
    {
      "epoch": 0.9338437288898896,
      "grad_norm": 1.2962802648544312,
      "learning_rate": 5.525834569610644e-05,
      "loss": 1.4514,
      "step": 5184
    },
    {
      "epoch": 0.934023868498086,
      "grad_norm": 1.245829701423645,
      "learning_rate": 5.524426249216191e-05,
      "loss": 1.3677,
      "step": 5185
    },
    {
      "epoch": 0.9342040081062823,
      "grad_norm": 1.3724788427352905,
      "learning_rate": 5.523017886752521e-05,
      "loss": 1.5466,
      "step": 5186
    },
    {
      "epoch": 0.9343841477144788,
      "grad_norm": 1.3739076852798462,
      "learning_rate": 5.5216094823326106e-05,
      "loss": 1.7048,
      "step": 5187
    },
    {
      "epoch": 0.934564287322675,
      "grad_norm": 1.4113438129425049,
      "learning_rate": 5.520201036069439e-05,
      "loss": 1.589,
      "step": 5188
    },
    {
      "epoch": 0.9347444269308715,
      "grad_norm": 1.4559268951416016,
      "learning_rate": 5.518792548075996e-05,
      "loss": 1.5094,
      "step": 5189
    },
    {
      "epoch": 0.9349245665390677,
      "grad_norm": 1.3730664253234863,
      "learning_rate": 5.517384018465267e-05,
      "loss": 1.5086,
      "step": 5190
    },
    {
      "epoch": 0.9351047061472642,
      "grad_norm": 1.371534824371338,
      "learning_rate": 5.515975447350244e-05,
      "loss": 1.4017,
      "step": 5191
    },
    {
      "epoch": 0.9352848457554604,
      "grad_norm": 1.32597815990448,
      "learning_rate": 5.5145668348439216e-05,
      "loss": 1.406,
      "step": 5192
    },
    {
      "epoch": 0.9354649853636569,
      "grad_norm": 1.3546141386032104,
      "learning_rate": 5.5131581810592994e-05,
      "loss": 1.2636,
      "step": 5193
    },
    {
      "epoch": 0.9356451249718531,
      "grad_norm": 1.5192584991455078,
      "learning_rate": 5.5117494861093774e-05,
      "loss": 1.7882,
      "step": 5194
    },
    {
      "epoch": 0.9358252645800496,
      "grad_norm": 1.4189032316207886,
      "learning_rate": 5.51034075010716e-05,
      "loss": 1.6248,
      "step": 5195
    },
    {
      "epoch": 0.9360054041882458,
      "grad_norm": 1.5195542573928833,
      "learning_rate": 5.508931973165658e-05,
      "loss": 1.432,
      "step": 5196
    },
    {
      "epoch": 0.9361855437964423,
      "grad_norm": 1.570252776145935,
      "learning_rate": 5.50752315539788e-05,
      "loss": 1.6429,
      "step": 5197
    },
    {
      "epoch": 0.9363656834046385,
      "grad_norm": 1.3162167072296143,
      "learning_rate": 5.506114296916841e-05,
      "loss": 1.1696,
      "step": 5198
    },
    {
      "epoch": 0.936545823012835,
      "grad_norm": 1.3737947940826416,
      "learning_rate": 5.50470539783556e-05,
      "loss": 1.2468,
      "step": 5199
    },
    {
      "epoch": 0.9367259626210312,
      "grad_norm": 1.3073331117630005,
      "learning_rate": 5.5032964582670584e-05,
      "loss": 1.4755,
      "step": 5200
    },
    {
      "epoch": 0.9369061022292277,
      "grad_norm": 1.3313778638839722,
      "learning_rate": 5.50188747832436e-05,
      "loss": 2.0013,
      "step": 5201
    },
    {
      "epoch": 0.937086241837424,
      "grad_norm": 1.3424785137176514,
      "learning_rate": 5.5004784581204927e-05,
      "loss": 1.7619,
      "step": 5202
    },
    {
      "epoch": 0.9372663814456204,
      "grad_norm": 1.2780787944793701,
      "learning_rate": 5.499069397768487e-05,
      "loss": 1.9991,
      "step": 5203
    },
    {
      "epoch": 0.9374465210538168,
      "grad_norm": 1.291505217552185,
      "learning_rate": 5.497660297381376e-05,
      "loss": 1.9561,
      "step": 5204
    },
    {
      "epoch": 0.9376266606620131,
      "grad_norm": 1.309662938117981,
      "learning_rate": 5.4962511570721995e-05,
      "loss": 1.9757,
      "step": 5205
    },
    {
      "epoch": 0.9378068002702095,
      "grad_norm": 1.3198344707489014,
      "learning_rate": 5.494841976953996e-05,
      "loss": 2.0563,
      "step": 5206
    },
    {
      "epoch": 0.9379869398784058,
      "grad_norm": 1.5110433101654053,
      "learning_rate": 5.493432757139813e-05,
      "loss": 1.5722,
      "step": 5207
    },
    {
      "epoch": 0.9381670794866022,
      "grad_norm": 1.486543893814087,
      "learning_rate": 5.492023497742693e-05,
      "loss": 1.8125,
      "step": 5208
    },
    {
      "epoch": 0.9383472190947985,
      "grad_norm": 1.8968520164489746,
      "learning_rate": 5.4906141988756865e-05,
      "loss": 2.1808,
      "step": 5209
    },
    {
      "epoch": 0.9385273587029949,
      "grad_norm": 1.479502558708191,
      "learning_rate": 5.4892048606518484e-05,
      "loss": 1.9078,
      "step": 5210
    },
    {
      "epoch": 0.9387074983111912,
      "grad_norm": 1.4141347408294678,
      "learning_rate": 5.4877954831842346e-05,
      "loss": 1.7568,
      "step": 5211
    },
    {
      "epoch": 0.9388876379193876,
      "grad_norm": 1.2972908020019531,
      "learning_rate": 5.486386066585904e-05,
      "loss": 1.4539,
      "step": 5212
    },
    {
      "epoch": 0.9390677775275839,
      "grad_norm": 1.2835732698440552,
      "learning_rate": 5.484976610969923e-05,
      "loss": 1.3333,
      "step": 5213
    },
    {
      "epoch": 0.9392479171357803,
      "grad_norm": 1.2514272928237915,
      "learning_rate": 5.483567116449353e-05,
      "loss": 1.3357,
      "step": 5214
    },
    {
      "epoch": 0.9394280567439766,
      "grad_norm": 1.2403721809387207,
      "learning_rate": 5.482157583137264e-05,
      "loss": 1.4374,
      "step": 5215
    },
    {
      "epoch": 0.939608196352173,
      "grad_norm": 1.257214903831482,
      "learning_rate": 5.480748011146728e-05,
      "loss": 1.4023,
      "step": 5216
    },
    {
      "epoch": 0.9397883359603693,
      "grad_norm": 1.4019343852996826,
      "learning_rate": 5.479338400590821e-05,
      "loss": 1.441,
      "step": 5217
    },
    {
      "epoch": 0.9399684755685657,
      "grad_norm": 1.4403142929077148,
      "learning_rate": 5.4779287515826216e-05,
      "loss": 1.7232,
      "step": 5218
    },
    {
      "epoch": 0.940148615176762,
      "grad_norm": 1.3482422828674316,
      "learning_rate": 5.4765190642352106e-05,
      "loss": 1.6092,
      "step": 5219
    },
    {
      "epoch": 0.9403287547849584,
      "grad_norm": 1.3199315071105957,
      "learning_rate": 5.4751093386616716e-05,
      "loss": 1.6753,
      "step": 5220
    },
    {
      "epoch": 0.9405088943931547,
      "grad_norm": 1.8255246877670288,
      "learning_rate": 5.473699574975093e-05,
      "loss": 1.6225,
      "step": 5221
    },
    {
      "epoch": 0.9406890340013511,
      "grad_norm": 1.4141913652420044,
      "learning_rate": 5.472289773288566e-05,
      "loss": 1.7127,
      "step": 5222
    },
    {
      "epoch": 0.9408691736095474,
      "grad_norm": 1.280982255935669,
      "learning_rate": 5.4708799337151826e-05,
      "loss": 1.393,
      "step": 5223
    },
    {
      "epoch": 0.9410493132177438,
      "grad_norm": 1.3213261365890503,
      "learning_rate": 5.469470056368041e-05,
      "loss": 1.5745,
      "step": 5224
    },
    {
      "epoch": 0.9412294528259401,
      "grad_norm": 1.2625030279159546,
      "learning_rate": 5.4680601413602396e-05,
      "loss": 1.3813,
      "step": 5225
    },
    {
      "epoch": 0.9414095924341365,
      "grad_norm": 1.4939217567443848,
      "learning_rate": 5.4666501888048815e-05,
      "loss": 1.7354,
      "step": 5226
    },
    {
      "epoch": 0.9415897320423328,
      "grad_norm": 1.3313169479370117,
      "learning_rate": 5.4652401988150725e-05,
      "loss": 1.3806,
      "step": 5227
    },
    {
      "epoch": 0.9417698716505292,
      "grad_norm": 1.310395359992981,
      "learning_rate": 5.463830171503922e-05,
      "loss": 1.6968,
      "step": 5228
    },
    {
      "epoch": 0.9419500112587255,
      "grad_norm": 1.4383376836776733,
      "learning_rate": 5.4624201069845416e-05,
      "loss": 1.379,
      "step": 5229
    },
    {
      "epoch": 0.9421301508669219,
      "grad_norm": 1.3266233205795288,
      "learning_rate": 5.461010005370044e-05,
      "loss": 1.453,
      "step": 5230
    },
    {
      "epoch": 0.9423102904751182,
      "grad_norm": 1.4859930276870728,
      "learning_rate": 5.459599866773549e-05,
      "loss": 1.6295,
      "step": 5231
    },
    {
      "epoch": 0.9424904300833146,
      "grad_norm": 1.4545331001281738,
      "learning_rate": 5.458189691308177e-05,
      "loss": 1.7269,
      "step": 5232
    },
    {
      "epoch": 0.942670569691511,
      "grad_norm": 1.4907279014587402,
      "learning_rate": 5.4567794790870495e-05,
      "loss": 1.7467,
      "step": 5233
    },
    {
      "epoch": 0.9428507092997073,
      "grad_norm": 1.418258547782898,
      "learning_rate": 5.455369230223296e-05,
      "loss": 1.7117,
      "step": 5234
    },
    {
      "epoch": 0.9430308489079037,
      "grad_norm": 1.373248815536499,
      "learning_rate": 5.453958944830045e-05,
      "loss": 1.683,
      "step": 5235
    },
    {
      "epoch": 0.9432109885161,
      "grad_norm": 1.3248751163482666,
      "learning_rate": 5.452548623020428e-05,
      "loss": 1.4341,
      "step": 5236
    },
    {
      "epoch": 0.9433911281242964,
      "grad_norm": 1.4791078567504883,
      "learning_rate": 5.45113826490758e-05,
      "loss": 1.6996,
      "step": 5237
    },
    {
      "epoch": 0.9435712677324927,
      "grad_norm": 1.4098297357559204,
      "learning_rate": 5.449727870604642e-05,
      "loss": 1.5332,
      "step": 5238
    },
    {
      "epoch": 0.9437514073406891,
      "grad_norm": 1.3441678285598755,
      "learning_rate": 5.4483174402247526e-05,
      "loss": 1.5943,
      "step": 5239
    },
    {
      "epoch": 0.9439315469488854,
      "grad_norm": 1.3465396165847778,
      "learning_rate": 5.446906973881056e-05,
      "loss": 1.4731,
      "step": 5240
    },
    {
      "epoch": 0.9441116865570818,
      "grad_norm": 1.401462435722351,
      "learning_rate": 5.4454964716866997e-05,
      "loss": 1.4924,
      "step": 5241
    },
    {
      "epoch": 0.9442918261652781,
      "grad_norm": 1.3031507730484009,
      "learning_rate": 5.4440859337548346e-05,
      "loss": 1.3929,
      "step": 5242
    },
    {
      "epoch": 0.9444719657734745,
      "grad_norm": 1.3358888626098633,
      "learning_rate": 5.44267536019861e-05,
      "loss": 1.5051,
      "step": 5243
    },
    {
      "epoch": 0.9446521053816708,
      "grad_norm": 1.4784542322158813,
      "learning_rate": 5.4412647511311855e-05,
      "loss": 1.5675,
      "step": 5244
    },
    {
      "epoch": 0.9448322449898672,
      "grad_norm": 1.5374906063079834,
      "learning_rate": 5.439854106665717e-05,
      "loss": 1.6853,
      "step": 5245
    },
    {
      "epoch": 0.9450123845980635,
      "grad_norm": 1.4937000274658203,
      "learning_rate": 5.4384434269153664e-05,
      "loss": 1.4855,
      "step": 5246
    },
    {
      "epoch": 0.9451925242062599,
      "grad_norm": 1.3594037294387817,
      "learning_rate": 5.437032711993297e-05,
      "loss": 1.4149,
      "step": 5247
    },
    {
      "epoch": 0.9453726638144562,
      "grad_norm": 1.2588589191436768,
      "learning_rate": 5.435621962012676e-05,
      "loss": 1.1674,
      "step": 5248
    },
    {
      "epoch": 0.9455528034226526,
      "grad_norm": 1.3163026571273804,
      "learning_rate": 5.434211177086674e-05,
      "loss": 1.3618,
      "step": 5249
    },
    {
      "epoch": 0.9457329430308489,
      "grad_norm": 1.2332024574279785,
      "learning_rate": 5.4328003573284645e-05,
      "loss": 1.1261,
      "step": 5250
    },
    {
      "epoch": 0.9459130826390453,
      "grad_norm": 1.2786064147949219,
      "learning_rate": 5.4313895028512194e-05,
      "loss": 1.9925,
      "step": 5251
    },
    {
      "epoch": 0.9460932222472416,
      "grad_norm": 1.2586089372634888,
      "learning_rate": 5.4299786137681194e-05,
      "loss": 1.9319,
      "step": 5252
    },
    {
      "epoch": 0.946273361855438,
      "grad_norm": 1.431467890739441,
      "learning_rate": 5.4285676901923435e-05,
      "loss": 1.9859,
      "step": 5253
    },
    {
      "epoch": 0.9464535014636343,
      "grad_norm": 1.2929086685180664,
      "learning_rate": 5.4271567322370765e-05,
      "loss": 2.0411,
      "step": 5254
    },
    {
      "epoch": 0.9466336410718307,
      "grad_norm": 1.330033540725708,
      "learning_rate": 5.425745740015504e-05,
      "loss": 1.8186,
      "step": 5255
    },
    {
      "epoch": 0.946813780680027,
      "grad_norm": 1.259856939315796,
      "learning_rate": 5.4243347136408176e-05,
      "loss": 1.7864,
      "step": 5256
    },
    {
      "epoch": 0.9469939202882234,
      "grad_norm": 1.3493809700012207,
      "learning_rate": 5.4229236532262064e-05,
      "loss": 1.7646,
      "step": 5257
    },
    {
      "epoch": 0.9471740598964197,
      "grad_norm": 1.4474732875823975,
      "learning_rate": 5.4215125588848635e-05,
      "loss": 2.0963,
      "step": 5258
    },
    {
      "epoch": 0.9473541995046161,
      "grad_norm": 1.9803376197814941,
      "learning_rate": 5.4201014307299904e-05,
      "loss": 2.3726,
      "step": 5259
    },
    {
      "epoch": 0.9475343391128125,
      "grad_norm": 1.6465718746185303,
      "learning_rate": 5.418690268874784e-05,
      "loss": 2.2566,
      "step": 5260
    },
    {
      "epoch": 0.9477144787210088,
      "grad_norm": 1.874721884727478,
      "learning_rate": 5.41727907343245e-05,
      "loss": 2.2286,
      "step": 5261
    },
    {
      "epoch": 0.9478946183292052,
      "grad_norm": 1.2656484842300415,
      "learning_rate": 5.4158678445161906e-05,
      "loss": 1.615,
      "step": 5262
    },
    {
      "epoch": 0.9480747579374015,
      "grad_norm": 1.3518365621566772,
      "learning_rate": 5.414456582239216e-05,
      "loss": 1.5953,
      "step": 5263
    },
    {
      "epoch": 0.9482548975455979,
      "grad_norm": 1.2650874853134155,
      "learning_rate": 5.4130452867147366e-05,
      "loss": 1.4918,
      "step": 5264
    },
    {
      "epoch": 0.9484350371537942,
      "grad_norm": 1.3905199766159058,
      "learning_rate": 5.411633958055963e-05,
      "loss": 1.7087,
      "step": 5265
    },
    {
      "epoch": 0.9486151767619906,
      "grad_norm": 1.3632307052612305,
      "learning_rate": 5.4102225963761155e-05,
      "loss": 1.7486,
      "step": 5266
    },
    {
      "epoch": 0.9487953163701869,
      "grad_norm": 1.2745888233184814,
      "learning_rate": 5.408811201788412e-05,
      "loss": 1.5132,
      "step": 5267
    },
    {
      "epoch": 0.9489754559783833,
      "grad_norm": 1.410667896270752,
      "learning_rate": 5.407399774406072e-05,
      "loss": 1.7726,
      "step": 5268
    },
    {
      "epoch": 0.9491555955865796,
      "grad_norm": 1.3012992143630981,
      "learning_rate": 5.4059883143423204e-05,
      "loss": 1.467,
      "step": 5269
    },
    {
      "epoch": 0.949335735194776,
      "grad_norm": 1.3403841257095337,
      "learning_rate": 5.4045768217103844e-05,
      "loss": 1.5374,
      "step": 5270
    },
    {
      "epoch": 0.9495158748029723,
      "grad_norm": 1.2738533020019531,
      "learning_rate": 5.4031652966234933e-05,
      "loss": 1.5176,
      "step": 5271
    },
    {
      "epoch": 0.9496960144111687,
      "grad_norm": 1.3155604600906372,
      "learning_rate": 5.401753739194878e-05,
      "loss": 1.607,
      "step": 5272
    },
    {
      "epoch": 0.949876154019365,
      "grad_norm": 1.4210665225982666,
      "learning_rate": 5.4003421495377735e-05,
      "loss": 1.6619,
      "step": 5273
    },
    {
      "epoch": 0.9500562936275614,
      "grad_norm": 1.4584317207336426,
      "learning_rate": 5.3989305277654156e-05,
      "loss": 1.5684,
      "step": 5274
    },
    {
      "epoch": 0.9502364332357577,
      "grad_norm": 1.4001803398132324,
      "learning_rate": 5.3975188739910464e-05,
      "loss": 1.4645,
      "step": 5275
    },
    {
      "epoch": 0.9504165728439541,
      "grad_norm": 1.3127332925796509,
      "learning_rate": 5.396107188327906e-05,
      "loss": 1.463,
      "step": 5276
    },
    {
      "epoch": 0.9505967124521504,
      "grad_norm": 1.3617123365402222,
      "learning_rate": 5.394695470889241e-05,
      "loss": 1.3824,
      "step": 5277
    },
    {
      "epoch": 0.9507768520603468,
      "grad_norm": 1.3839503526687622,
      "learning_rate": 5.3932837217882984e-05,
      "loss": 1.5643,
      "step": 5278
    },
    {
      "epoch": 0.9509569916685431,
      "grad_norm": 1.2879011631011963,
      "learning_rate": 5.391871941138324e-05,
      "loss": 1.3389,
      "step": 5279
    },
    {
      "epoch": 0.9511371312767395,
      "grad_norm": 1.3183971643447876,
      "learning_rate": 5.390460129052576e-05,
      "loss": 1.3732,
      "step": 5280
    },
    {
      "epoch": 0.9513172708849358,
      "grad_norm": 1.370556116104126,
      "learning_rate": 5.3890482856443035e-05,
      "loss": 1.4709,
      "step": 5281
    },
    {
      "epoch": 0.9514974104931322,
      "grad_norm": 1.3111011981964111,
      "learning_rate": 5.38763641102677e-05,
      "loss": 1.4616,
      "step": 5282
    },
    {
      "epoch": 0.9516775501013285,
      "grad_norm": 1.2944639921188354,
      "learning_rate": 5.386224505313232e-05,
      "loss": 1.577,
      "step": 5283
    },
    {
      "epoch": 0.9518576897095249,
      "grad_norm": 1.4301786422729492,
      "learning_rate": 5.3848125686169524e-05,
      "loss": 1.5857,
      "step": 5284
    },
    {
      "epoch": 0.9520378293177212,
      "grad_norm": 1.413000464439392,
      "learning_rate": 5.383400601051194e-05,
      "loss": 1.6484,
      "step": 5285
    },
    {
      "epoch": 0.9522179689259176,
      "grad_norm": 1.4588044881820679,
      "learning_rate": 5.381988602729229e-05,
      "loss": 1.64,
      "step": 5286
    },
    {
      "epoch": 0.9523981085341139,
      "grad_norm": 1.338789463043213,
      "learning_rate": 5.380576573764322e-05,
      "loss": 1.3751,
      "step": 5287
    },
    {
      "epoch": 0.9525782481423103,
      "grad_norm": 1.3800326585769653,
      "learning_rate": 5.37916451426975e-05,
      "loss": 1.501,
      "step": 5288
    },
    {
      "epoch": 0.9527583877505067,
      "grad_norm": 1.4623042345046997,
      "learning_rate": 5.377752424358784e-05,
      "loss": 1.7401,
      "step": 5289
    },
    {
      "epoch": 0.952938527358703,
      "grad_norm": 1.474949836730957,
      "learning_rate": 5.376340304144701e-05,
      "loss": 1.6748,
      "step": 5290
    },
    {
      "epoch": 0.9531186669668994,
      "grad_norm": 1.5639631748199463,
      "learning_rate": 5.3749281537407845e-05,
      "loss": 1.535,
      "step": 5291
    },
    {
      "epoch": 0.9532988065750957,
      "grad_norm": 1.2978333234786987,
      "learning_rate": 5.373515973260313e-05,
      "loss": 1.616,
      "step": 5292
    },
    {
      "epoch": 0.9534789461832921,
      "grad_norm": 1.3505828380584717,
      "learning_rate": 5.3721037628165716e-05,
      "loss": 1.3829,
      "step": 5293
    },
    {
      "epoch": 0.9536590857914884,
      "grad_norm": 1.3966565132141113,
      "learning_rate": 5.3706915225228493e-05,
      "loss": 1.3703,
      "step": 5294
    },
    {
      "epoch": 0.9538392253996848,
      "grad_norm": 1.3113038539886475,
      "learning_rate": 5.3692792524924315e-05,
      "loss": 1.3557,
      "step": 5295
    },
    {
      "epoch": 0.9540193650078811,
      "grad_norm": 1.5705491304397583,
      "learning_rate": 5.367866952838614e-05,
      "loss": 1.6088,
      "step": 5296
    },
    {
      "epoch": 0.9541995046160775,
      "grad_norm": 1.3828707933425903,
      "learning_rate": 5.3664546236746874e-05,
      "loss": 1.5108,
      "step": 5297
    },
    {
      "epoch": 0.9543796442242738,
      "grad_norm": 1.3983142375946045,
      "learning_rate": 5.36504226511395e-05,
      "loss": 1.2467,
      "step": 5298
    },
    {
      "epoch": 0.9545597838324702,
      "grad_norm": 1.3830968141555786,
      "learning_rate": 5.363629877269699e-05,
      "loss": 1.4298,
      "step": 5299
    },
    {
      "epoch": 0.9547399234406665,
      "grad_norm": 1.2593024969100952,
      "learning_rate": 5.3622174602552376e-05,
      "loss": 1.2716,
      "step": 5300
    },
    {
      "epoch": 0.9549200630488629,
      "grad_norm": 1.445265531539917,
      "learning_rate": 5.360805014183867e-05,
      "loss": 2.1269,
      "step": 5301
    },
    {
      "epoch": 0.9551002026570592,
      "grad_norm": 1.355506181716919,
      "learning_rate": 5.3593925391688937e-05,
      "loss": 1.935,
      "step": 5302
    },
    {
      "epoch": 0.9552803422652556,
      "grad_norm": 1.340486764907837,
      "learning_rate": 5.357980035323628e-05,
      "loss": 1.9725,
      "step": 5303
    },
    {
      "epoch": 0.9554604818734519,
      "grad_norm": 1.3715964555740356,
      "learning_rate": 5.356567502761376e-05,
      "loss": 2.3223,
      "step": 5304
    },
    {
      "epoch": 0.9556406214816483,
      "grad_norm": 1.3033766746520996,
      "learning_rate": 5.355154941595454e-05,
      "loss": 2.0322,
      "step": 5305
    },
    {
      "epoch": 0.9558207610898446,
      "grad_norm": 1.383777141571045,
      "learning_rate": 5.3537423519391736e-05,
      "loss": 1.8917,
      "step": 5306
    },
    {
      "epoch": 0.956000900698041,
      "grad_norm": 1.3475384712219238,
      "learning_rate": 5.352329733905856e-05,
      "loss": 1.6404,
      "step": 5307
    },
    {
      "epoch": 0.9561810403062373,
      "grad_norm": 1.4337648153305054,
      "learning_rate": 5.350917087608818e-05,
      "loss": 1.9408,
      "step": 5308
    },
    {
      "epoch": 0.9563611799144337,
      "grad_norm": 1.7661281824111938,
      "learning_rate": 5.349504413161381e-05,
      "loss": 2.6733,
      "step": 5309
    },
    {
      "epoch": 0.95654131952263,
      "grad_norm": 1.6380430459976196,
      "learning_rate": 5.348091710676874e-05,
      "loss": 1.8912,
      "step": 5310
    },
    {
      "epoch": 0.9567214591308264,
      "grad_norm": 1.5622223615646362,
      "learning_rate": 5.3466789802686155e-05,
      "loss": 1.8553,
      "step": 5311
    },
    {
      "epoch": 0.9569015987390227,
      "grad_norm": 1.3471565246582031,
      "learning_rate": 5.345266222049941e-05,
      "loss": 1.4785,
      "step": 5312
    },
    {
      "epoch": 0.9570817383472191,
      "grad_norm": 1.3371397256851196,
      "learning_rate": 5.3438534361341755e-05,
      "loss": 1.7472,
      "step": 5313
    },
    {
      "epoch": 0.9572618779554154,
      "grad_norm": 1.2839196920394897,
      "learning_rate": 5.342440622634657e-05,
      "loss": 1.421,
      "step": 5314
    },
    {
      "epoch": 0.9574420175636118,
      "grad_norm": 1.3307623863220215,
      "learning_rate": 5.341027781664717e-05,
      "loss": 1.5156,
      "step": 5315
    },
    {
      "epoch": 0.9576221571718081,
      "grad_norm": 1.386922001838684,
      "learning_rate": 5.339614913337696e-05,
      "loss": 1.5564,
      "step": 5316
    },
    {
      "epoch": 0.9578022967800045,
      "grad_norm": 1.2679773569107056,
      "learning_rate": 5.3382020177669314e-05,
      "loss": 1.3923,
      "step": 5317
    },
    {
      "epoch": 0.9579824363882009,
      "grad_norm": 1.313116431236267,
      "learning_rate": 5.336789095065765e-05,
      "loss": 1.4542,
      "step": 5318
    },
    {
      "epoch": 0.9581625759963972,
      "grad_norm": 1.2313501834869385,
      "learning_rate": 5.335376145347542e-05,
      "loss": 1.5989,
      "step": 5319
    },
    {
      "epoch": 0.9583427156045936,
      "grad_norm": 1.5348842144012451,
      "learning_rate": 5.3339631687256084e-05,
      "loss": 1.7846,
      "step": 5320
    },
    {
      "epoch": 0.9585228552127899,
      "grad_norm": 1.3411052227020264,
      "learning_rate": 5.3325501653133125e-05,
      "loss": 1.4423,
      "step": 5321
    },
    {
      "epoch": 0.9587029948209863,
      "grad_norm": 1.2594385147094727,
      "learning_rate": 5.331137135224003e-05,
      "loss": 1.4513,
      "step": 5322
    },
    {
      "epoch": 0.9588831344291826,
      "grad_norm": 1.2734466791152954,
      "learning_rate": 5.3297240785710335e-05,
      "loss": 1.4999,
      "step": 5323
    },
    {
      "epoch": 0.959063274037379,
      "grad_norm": 1.2433655261993408,
      "learning_rate": 5.3283109954677604e-05,
      "loss": 1.297,
      "step": 5324
    },
    {
      "epoch": 0.9592434136455753,
      "grad_norm": 1.2244817018508911,
      "learning_rate": 5.326897886027537e-05,
      "loss": 1.4525,
      "step": 5325
    },
    {
      "epoch": 0.9594235532537717,
      "grad_norm": 1.2821660041809082,
      "learning_rate": 5.3254847503637285e-05,
      "loss": 1.4262,
      "step": 5326
    },
    {
      "epoch": 0.959603692861968,
      "grad_norm": 1.274381399154663,
      "learning_rate": 5.324071588589689e-05,
      "loss": 1.435,
      "step": 5327
    },
    {
      "epoch": 0.9597838324701644,
      "grad_norm": 1.259706974029541,
      "learning_rate": 5.322658400818784e-05,
      "loss": 1.2774,
      "step": 5328
    },
    {
      "epoch": 0.9599639720783607,
      "grad_norm": 1.3823274374008179,
      "learning_rate": 5.3212451871643806e-05,
      "loss": 1.5523,
      "step": 5329
    },
    {
      "epoch": 0.9601441116865571,
      "grad_norm": 1.4042102098464966,
      "learning_rate": 5.319831947739844e-05,
      "loss": 1.4561,
      "step": 5330
    },
    {
      "epoch": 0.9603242512947534,
      "grad_norm": 1.4428564310073853,
      "learning_rate": 5.318418682658545e-05,
      "loss": 1.4929,
      "step": 5331
    },
    {
      "epoch": 0.9605043909029498,
      "grad_norm": 1.3917436599731445,
      "learning_rate": 5.317005392033854e-05,
      "loss": 1.3568,
      "step": 5332
    },
    {
      "epoch": 0.9606845305111461,
      "grad_norm": 1.3827439546585083,
      "learning_rate": 5.3155920759791446e-05,
      "loss": 1.5766,
      "step": 5333
    },
    {
      "epoch": 0.9608646701193425,
      "grad_norm": 1.5625269412994385,
      "learning_rate": 5.314178734607792e-05,
      "loss": 1.5136,
      "step": 5334
    },
    {
      "epoch": 0.9610448097275388,
      "grad_norm": 1.2335017919540405,
      "learning_rate": 5.3127653680331755e-05,
      "loss": 1.3161,
      "step": 5335
    },
    {
      "epoch": 0.9612249493357352,
      "grad_norm": 1.426091194152832,
      "learning_rate": 5.311351976368674e-05,
      "loss": 1.6226,
      "step": 5336
    },
    {
      "epoch": 0.9614050889439315,
      "grad_norm": 1.402161955833435,
      "learning_rate": 5.309938559727667e-05,
      "loss": 1.5161,
      "step": 5337
    },
    {
      "epoch": 0.9615852285521279,
      "grad_norm": 1.4057179689407349,
      "learning_rate": 5.3085251182235416e-05,
      "loss": 1.7032,
      "step": 5338
    },
    {
      "epoch": 0.9617653681603242,
      "grad_norm": 1.460518479347229,
      "learning_rate": 5.30711165196968e-05,
      "loss": 1.5467,
      "step": 5339
    },
    {
      "epoch": 0.9619455077685206,
      "grad_norm": 1.2929974794387817,
      "learning_rate": 5.305698161079471e-05,
      "loss": 1.3473,
      "step": 5340
    },
    {
      "epoch": 0.9621256473767169,
      "grad_norm": 1.2966831922531128,
      "learning_rate": 5.304284645666304e-05,
      "loss": 1.419,
      "step": 5341
    },
    {
      "epoch": 0.9623057869849133,
      "grad_norm": 1.3768166303634644,
      "learning_rate": 5.302871105843573e-05,
      "loss": 1.5191,
      "step": 5342
    },
    {
      "epoch": 0.9624859265931096,
      "grad_norm": 1.446094036102295,
      "learning_rate": 5.3014575417246706e-05,
      "loss": 1.4516,
      "step": 5343
    },
    {
      "epoch": 0.962666066201306,
      "grad_norm": 1.294031023979187,
      "learning_rate": 5.3000439534229875e-05,
      "loss": 1.336,
      "step": 5344
    },
    {
      "epoch": 0.9628462058095023,
      "grad_norm": 1.4943344593048096,
      "learning_rate": 5.2986303410519254e-05,
      "loss": 1.6072,
      "step": 5345
    },
    {
      "epoch": 0.9630263454176987,
      "grad_norm": 1.4249428510665894,
      "learning_rate": 5.297216704724883e-05,
      "loss": 1.4949,
      "step": 5346
    },
    {
      "epoch": 0.9632064850258951,
      "grad_norm": 1.536133050918579,
      "learning_rate": 5.295803044555262e-05,
      "loss": 1.4801,
      "step": 5347
    },
    {
      "epoch": 0.9633866246340914,
      "grad_norm": 1.5198663473129272,
      "learning_rate": 5.294389360656465e-05,
      "loss": 1.641,
      "step": 5348
    },
    {
      "epoch": 0.9635667642422878,
      "grad_norm": 1.4848164319992065,
      "learning_rate": 5.292975653141896e-05,
      "loss": 1.5108,
      "step": 5349
    },
    {
      "epoch": 0.9637469038504841,
      "grad_norm": 1.4578449726104736,
      "learning_rate": 5.2915619221249645e-05,
      "loss": 1.6459,
      "step": 5350
    },
    {
      "epoch": 0.9639270434586805,
      "grad_norm": 1.1815444231033325,
      "learning_rate": 5.2901481677190755e-05,
      "loss": 1.9762,
      "step": 5351
    },
    {
      "epoch": 0.9641071830668768,
      "grad_norm": 1.2879889011383057,
      "learning_rate": 5.288734390037643e-05,
      "loss": 2.2266,
      "step": 5352
    },
    {
      "epoch": 0.9642873226750732,
      "grad_norm": 1.2407768964767456,
      "learning_rate": 5.287320589194078e-05,
      "loss": 2.1087,
      "step": 5353
    },
    {
      "epoch": 0.9644674622832695,
      "grad_norm": 1.252458930015564,
      "learning_rate": 5.285906765301796e-05,
      "loss": 1.9604,
      "step": 5354
    },
    {
      "epoch": 0.9646476018914659,
      "grad_norm": 1.2389116287231445,
      "learning_rate": 5.2844929184742106e-05,
      "loss": 1.6724,
      "step": 5355
    },
    {
      "epoch": 0.9648277414996622,
      "grad_norm": 1.3210970163345337,
      "learning_rate": 5.283079048824743e-05,
      "loss": 1.6816,
      "step": 5356
    },
    {
      "epoch": 0.9650078811078586,
      "grad_norm": 1.4765480756759644,
      "learning_rate": 5.281665156466813e-05,
      "loss": 1.9629,
      "step": 5357
    },
    {
      "epoch": 0.9651880207160549,
      "grad_norm": 1.600051760673523,
      "learning_rate": 5.28025124151384e-05,
      "loss": 2.0372,
      "step": 5358
    },
    {
      "epoch": 0.9653681603242513,
      "grad_norm": 1.7439708709716797,
      "learning_rate": 5.278837304079251e-05,
      "loss": 2.0927,
      "step": 5359
    },
    {
      "epoch": 0.9655482999324476,
      "grad_norm": 1.5792269706726074,
      "learning_rate": 5.277423344276465e-05,
      "loss": 1.873,
      "step": 5360
    },
    {
      "epoch": 0.965728439540644,
      "grad_norm": 1.2552424669265747,
      "learning_rate": 5.276009362218917e-05,
      "loss": 1.4666,
      "step": 5361
    },
    {
      "epoch": 0.9659085791488403,
      "grad_norm": 1.3538386821746826,
      "learning_rate": 5.2745953580200314e-05,
      "loss": 1.6019,
      "step": 5362
    },
    {
      "epoch": 0.9660887187570367,
      "grad_norm": 1.4427562952041626,
      "learning_rate": 5.27318133179324e-05,
      "loss": 1.6921,
      "step": 5363
    },
    {
      "epoch": 0.966268858365233,
      "grad_norm": 1.3343034982681274,
      "learning_rate": 5.271767283651975e-05,
      "loss": 1.4431,
      "step": 5364
    },
    {
      "epoch": 0.9664489979734294,
      "grad_norm": 1.2728177309036255,
      "learning_rate": 5.270353213709671e-05,
      "loss": 1.462,
      "step": 5365
    },
    {
      "epoch": 0.9666291375816257,
      "grad_norm": 1.2118480205535889,
      "learning_rate": 5.268939122079764e-05,
      "loss": 1.5252,
      "step": 5366
    },
    {
      "epoch": 0.9668092771898221,
      "grad_norm": 1.1633142232894897,
      "learning_rate": 5.267525008875692e-05,
      "loss": 1.4332,
      "step": 5367
    },
    {
      "epoch": 0.9669894167980184,
      "grad_norm": 1.4071487188339233,
      "learning_rate": 5.2661108742108935e-05,
      "loss": 1.672,
      "step": 5368
    },
    {
      "epoch": 0.9671695564062148,
      "grad_norm": 1.3663479089736938,
      "learning_rate": 5.264696718198811e-05,
      "loss": 1.7068,
      "step": 5369
    },
    {
      "epoch": 0.9673496960144111,
      "grad_norm": 1.3703925609588623,
      "learning_rate": 5.263282540952886e-05,
      "loss": 1.7422,
      "step": 5370
    },
    {
      "epoch": 0.9675298356226075,
      "grad_norm": 1.47991144657135,
      "learning_rate": 5.261868342586565e-05,
      "loss": 1.7098,
      "step": 5371
    },
    {
      "epoch": 0.9677099752308038,
      "grad_norm": 1.2913236618041992,
      "learning_rate": 5.260454123213291e-05,
      "loss": 1.3145,
      "step": 5372
    },
    {
      "epoch": 0.9678901148390002,
      "grad_norm": 1.4022060632705688,
      "learning_rate": 5.259039882946516e-05,
      "loss": 1.6299,
      "step": 5373
    },
    {
      "epoch": 0.9680702544471966,
      "grad_norm": 1.493638515472412,
      "learning_rate": 5.257625621899686e-05,
      "loss": 1.4015,
      "step": 5374
    },
    {
      "epoch": 0.9682503940553929,
      "grad_norm": 1.3100414276123047,
      "learning_rate": 5.2562113401862556e-05,
      "loss": 1.4389,
      "step": 5375
    },
    {
      "epoch": 0.9684305336635893,
      "grad_norm": 1.3837242126464844,
      "learning_rate": 5.254797037919675e-05,
      "loss": 1.4667,
      "step": 5376
    },
    {
      "epoch": 0.9686106732717856,
      "grad_norm": 1.3499776124954224,
      "learning_rate": 5.253382715213402e-05,
      "loss": 1.4416,
      "step": 5377
    },
    {
      "epoch": 0.968790812879982,
      "grad_norm": 1.4175798892974854,
      "learning_rate": 5.2519683721808895e-05,
      "loss": 1.4218,
      "step": 5378
    },
    {
      "epoch": 0.9689709524881783,
      "grad_norm": 1.3439507484436035,
      "learning_rate": 5.250554008935596e-05,
      "loss": 1.5618,
      "step": 5379
    },
    {
      "epoch": 0.9691510920963747,
      "grad_norm": 1.3955005407333374,
      "learning_rate": 5.249139625590984e-05,
      "loss": 1.4376,
      "step": 5380
    },
    {
      "epoch": 0.969331231704571,
      "grad_norm": 1.3289790153503418,
      "learning_rate": 5.247725222260512e-05,
      "loss": 1.3678,
      "step": 5381
    },
    {
      "epoch": 0.9695113713127674,
      "grad_norm": 1.2933189868927002,
      "learning_rate": 5.2463107990576444e-05,
      "loss": 1.433,
      "step": 5382
    },
    {
      "epoch": 0.9696915109209637,
      "grad_norm": 1.374428629875183,
      "learning_rate": 5.244896356095842e-05,
      "loss": 1.5303,
      "step": 5383
    },
    {
      "epoch": 0.9698716505291601,
      "grad_norm": 1.5264179706573486,
      "learning_rate": 5.2434818934885745e-05,
      "loss": 1.731,
      "step": 5384
    },
    {
      "epoch": 0.9700517901373564,
      "grad_norm": 1.3887882232666016,
      "learning_rate": 5.2420674113493084e-05,
      "loss": 1.3395,
      "step": 5385
    },
    {
      "epoch": 0.9702319297455528,
      "grad_norm": 1.511896014213562,
      "learning_rate": 5.2406529097915125e-05,
      "loss": 1.4429,
      "step": 5386
    },
    {
      "epoch": 0.9704120693537491,
      "grad_norm": 1.4407421350479126,
      "learning_rate": 5.2392383889286566e-05,
      "loss": 1.4556,
      "step": 5387
    },
    {
      "epoch": 0.9705922089619455,
      "grad_norm": 1.4015072584152222,
      "learning_rate": 5.237823848874214e-05,
      "loss": 1.4712,
      "step": 5388
    },
    {
      "epoch": 0.9707723485701418,
      "grad_norm": 1.5028202533721924,
      "learning_rate": 5.236409289741658e-05,
      "loss": 1.772,
      "step": 5389
    },
    {
      "epoch": 0.9709524881783382,
      "grad_norm": 1.4634078741073608,
      "learning_rate": 5.234994711644463e-05,
      "loss": 1.5136,
      "step": 5390
    },
    {
      "epoch": 0.9711326277865345,
      "grad_norm": 1.4175139665603638,
      "learning_rate": 5.2335801146961095e-05,
      "loss": 1.4582,
      "step": 5391
    },
    {
      "epoch": 0.9713127673947309,
      "grad_norm": 1.3558861017227173,
      "learning_rate": 5.2321654990100713e-05,
      "loss": 1.5888,
      "step": 5392
    },
    {
      "epoch": 0.9714929070029272,
      "grad_norm": 1.2589038610458374,
      "learning_rate": 5.230750864699829e-05,
      "loss": 1.2624,
      "step": 5393
    },
    {
      "epoch": 0.9716730466111236,
      "grad_norm": 1.311826229095459,
      "learning_rate": 5.229336211878866e-05,
      "loss": 1.2215,
      "step": 5394
    },
    {
      "epoch": 0.9718531862193199,
      "grad_norm": 1.3919540643692017,
      "learning_rate": 5.2279215406606627e-05,
      "loss": 1.5494,
      "step": 5395
    },
    {
      "epoch": 0.9720333258275163,
      "grad_norm": 1.5327457189559937,
      "learning_rate": 5.226506851158706e-05,
      "loss": 1.6599,
      "step": 5396
    },
    {
      "epoch": 0.9722134654357126,
      "grad_norm": 1.4243777990341187,
      "learning_rate": 5.2250921434864786e-05,
      "loss": 1.4678,
      "step": 5397
    },
    {
      "epoch": 0.972393605043909,
      "grad_norm": 1.501854419708252,
      "learning_rate": 5.22367741775747e-05,
      "loss": 1.3979,
      "step": 5398
    },
    {
      "epoch": 0.9725737446521053,
      "grad_norm": 1.27566397190094,
      "learning_rate": 5.222262674085166e-05,
      "loss": 0.9885,
      "step": 5399
    },
    {
      "epoch": 0.9727538842603017,
      "grad_norm": 1.4059970378875732,
      "learning_rate": 5.220847912583061e-05,
      "loss": 1.5126,
      "step": 5400
    },
    {
      "epoch": 0.972934023868498,
      "grad_norm": 1.3445314168930054,
      "learning_rate": 5.219433133364643e-05,
      "loss": 1.8674,
      "step": 5401
    },
    {
      "epoch": 0.9731141634766944,
      "grad_norm": 1.2173614501953125,
      "learning_rate": 5.218018336543407e-05,
      "loss": 1.8643,
      "step": 5402
    },
    {
      "epoch": 0.9732943030848908,
      "grad_norm": 1.5966296195983887,
      "learning_rate": 5.2166035222328446e-05,
      "loss": 1.9428,
      "step": 5403
    },
    {
      "epoch": 0.9734744426930871,
      "grad_norm": 1.4709405899047852,
      "learning_rate": 5.215188690546453e-05,
      "loss": 2.2181,
      "step": 5404
    },
    {
      "epoch": 0.9736545823012835,
      "grad_norm": 1.2857550382614136,
      "learning_rate": 5.21377384159773e-05,
      "loss": 1.6885,
      "step": 5405
    },
    {
      "epoch": 0.9738347219094798,
      "grad_norm": 1.477520227432251,
      "learning_rate": 5.2123589755001733e-05,
      "loss": 1.8622,
      "step": 5406
    },
    {
      "epoch": 0.9740148615176762,
      "grad_norm": 1.4609161615371704,
      "learning_rate": 5.2109440923672836e-05,
      "loss": 1.7777,
      "step": 5407
    },
    {
      "epoch": 0.9741950011258725,
      "grad_norm": 1.6814249753952026,
      "learning_rate": 5.2095291923125635e-05,
      "loss": 2.0174,
      "step": 5408
    },
    {
      "epoch": 0.974375140734069,
      "grad_norm": 1.603654146194458,
      "learning_rate": 5.2081142754495106e-05,
      "loss": 1.9189,
      "step": 5409
    },
    {
      "epoch": 0.9745552803422652,
      "grad_norm": 1.7143933773040771,
      "learning_rate": 5.2066993418916335e-05,
      "loss": 1.9721,
      "step": 5410
    },
    {
      "epoch": 0.9747354199504616,
      "grad_norm": 1.3116461038589478,
      "learning_rate": 5.205284391752434e-05,
      "loss": 1.3644,
      "step": 5411
    },
    {
      "epoch": 0.9749155595586579,
      "grad_norm": 1.3820116519927979,
      "learning_rate": 5.203869425145422e-05,
      "loss": 1.3874,
      "step": 5412
    },
    {
      "epoch": 0.9750956991668543,
      "grad_norm": 1.427774429321289,
      "learning_rate": 5.202454442184104e-05,
      "loss": 1.641,
      "step": 5413
    },
    {
      "epoch": 0.9752758387750506,
      "grad_norm": 1.3567496538162231,
      "learning_rate": 5.20103944298199e-05,
      "loss": 1.6077,
      "step": 5414
    },
    {
      "epoch": 0.975455978383247,
      "grad_norm": 1.3524290323257446,
      "learning_rate": 5.199624427652588e-05,
      "loss": 1.5241,
      "step": 5415
    },
    {
      "epoch": 0.9756361179914433,
      "grad_norm": 1.2697441577911377,
      "learning_rate": 5.1982093963094126e-05,
      "loss": 1.3101,
      "step": 5416
    },
    {
      "epoch": 0.9758162575996397,
      "grad_norm": 1.3722608089447021,
      "learning_rate": 5.196794349065975e-05,
      "loss": 1.6009,
      "step": 5417
    },
    {
      "epoch": 0.975996397207836,
      "grad_norm": 1.2859597206115723,
      "learning_rate": 5.195379286035793e-05,
      "loss": 1.4874,
      "step": 5418
    },
    {
      "epoch": 0.9761765368160324,
      "grad_norm": 1.218653917312622,
      "learning_rate": 5.193964207332378e-05,
      "loss": 1.4239,
      "step": 5419
    },
    {
      "epoch": 0.9763566764242287,
      "grad_norm": 1.4374889135360718,
      "learning_rate": 5.1925491130692486e-05,
      "loss": 1.5437,
      "step": 5420
    },
    {
      "epoch": 0.9765368160324251,
      "grad_norm": 1.3119416236877441,
      "learning_rate": 5.1911340033599244e-05,
      "loss": 1.4213,
      "step": 5421
    },
    {
      "epoch": 0.9767169556406214,
      "grad_norm": 1.3285481929779053,
      "learning_rate": 5.189718878317923e-05,
      "loss": 1.4009,
      "step": 5422
    },
    {
      "epoch": 0.9768970952488178,
      "grad_norm": 1.3340466022491455,
      "learning_rate": 5.188303738056765e-05,
      "loss": 1.3934,
      "step": 5423
    },
    {
      "epoch": 0.9770772348570141,
      "grad_norm": 1.4003435373306274,
      "learning_rate": 5.186888582689974e-05,
      "loss": 1.359,
      "step": 5424
    },
    {
      "epoch": 0.9772573744652105,
      "grad_norm": 1.3494806289672852,
      "learning_rate": 5.18547341233107e-05,
      "loss": 1.4999,
      "step": 5425
    },
    {
      "epoch": 0.9774375140734068,
      "grad_norm": 1.427824854850769,
      "learning_rate": 5.184058227093581e-05,
      "loss": 1.6114,
      "step": 5426
    },
    {
      "epoch": 0.9776176536816032,
      "grad_norm": 1.2543511390686035,
      "learning_rate": 5.1826430270910286e-05,
      "loss": 1.2611,
      "step": 5427
    },
    {
      "epoch": 0.9777977932897995,
      "grad_norm": 1.4440135955810547,
      "learning_rate": 5.181227812436944e-05,
      "loss": 1.7121,
      "step": 5428
    },
    {
      "epoch": 0.977977932897996,
      "grad_norm": 1.5105116367340088,
      "learning_rate": 5.179812583244851e-05,
      "loss": 1.7424,
      "step": 5429
    },
    {
      "epoch": 0.9781580725061922,
      "grad_norm": 1.45107901096344,
      "learning_rate": 5.1783973396282816e-05,
      "loss": 1.6719,
      "step": 5430
    },
    {
      "epoch": 0.9783382121143887,
      "grad_norm": 1.2329802513122559,
      "learning_rate": 5.176982081700763e-05,
      "loss": 1.4013,
      "step": 5431
    },
    {
      "epoch": 0.9785183517225851,
      "grad_norm": 1.3970863819122314,
      "learning_rate": 5.175566809575828e-05,
      "loss": 1.5069,
      "step": 5432
    },
    {
      "epoch": 0.9786984913307814,
      "grad_norm": 1.3058767318725586,
      "learning_rate": 5.17415152336701e-05,
      "loss": 1.5483,
      "step": 5433
    },
    {
      "epoch": 0.9788786309389778,
      "grad_norm": 1.3232342004776,
      "learning_rate": 5.172736223187842e-05,
      "loss": 1.4056,
      "step": 5434
    },
    {
      "epoch": 0.979058770547174,
      "grad_norm": 1.5994749069213867,
      "learning_rate": 5.171320909151858e-05,
      "loss": 1.8552,
      "step": 5435
    },
    {
      "epoch": 0.9792389101553705,
      "grad_norm": 1.5129034519195557,
      "learning_rate": 5.1699055813725936e-05,
      "loss": 1.5214,
      "step": 5436
    },
    {
      "epoch": 0.9794190497635668,
      "grad_norm": 1.393964171409607,
      "learning_rate": 5.168490239963586e-05,
      "loss": 1.4593,
      "step": 5437
    },
    {
      "epoch": 0.9795991893717632,
      "grad_norm": 1.2762224674224854,
      "learning_rate": 5.167074885038373e-05,
      "loss": 1.2972,
      "step": 5438
    },
    {
      "epoch": 0.9797793289799595,
      "grad_norm": 1.3246617317199707,
      "learning_rate": 5.1656595167104946e-05,
      "loss": 1.3516,
      "step": 5439
    },
    {
      "epoch": 0.9799594685881559,
      "grad_norm": 1.3270987272262573,
      "learning_rate": 5.164244135093492e-05,
      "loss": 1.5593,
      "step": 5440
    },
    {
      "epoch": 0.9801396081963522,
      "grad_norm": 1.4755040407180786,
      "learning_rate": 5.1628287403009026e-05,
      "loss": 1.6389,
      "step": 5441
    },
    {
      "epoch": 0.9803197478045486,
      "grad_norm": 1.5281306505203247,
      "learning_rate": 5.161413332446272e-05,
      "loss": 1.6293,
      "step": 5442
    },
    {
      "epoch": 0.9804998874127449,
      "grad_norm": 1.4726852178573608,
      "learning_rate": 5.159997911643143e-05,
      "loss": 1.4763,
      "step": 5443
    },
    {
      "epoch": 0.9806800270209413,
      "grad_norm": 1.5389952659606934,
      "learning_rate": 5.158582478005059e-05,
      "loss": 1.746,
      "step": 5444
    },
    {
      "epoch": 0.9808601666291376,
      "grad_norm": 1.358837604522705,
      "learning_rate": 5.1571670316455654e-05,
      "loss": 1.412,
      "step": 5445
    },
    {
      "epoch": 0.981040306237334,
      "grad_norm": 1.2964686155319214,
      "learning_rate": 5.155751572678209e-05,
      "loss": 1.3202,
      "step": 5446
    },
    {
      "epoch": 0.9812204458455303,
      "grad_norm": 1.2630119323730469,
      "learning_rate": 5.1543361012165384e-05,
      "loss": 1.3314,
      "step": 5447
    },
    {
      "epoch": 0.9814005854537267,
      "grad_norm": 1.2962652444839478,
      "learning_rate": 5.1529206173741e-05,
      "loss": 1.3861,
      "step": 5448
    },
    {
      "epoch": 0.981580725061923,
      "grad_norm": 1.461641788482666,
      "learning_rate": 5.151505121264444e-05,
      "loss": 1.4171,
      "step": 5449
    },
    {
      "epoch": 0.9817608646701194,
      "grad_norm": 1.2739298343658447,
      "learning_rate": 5.1500896130011225e-05,
      "loss": 1.2318,
      "step": 5450
    },
    {
      "epoch": 0.9819410042783157,
      "grad_norm": 1.3509132862091064,
      "learning_rate": 5.148674092697684e-05,
      "loss": 2.063,
      "step": 5451
    },
    {
      "epoch": 0.9821211438865121,
      "grad_norm": 1.2478272914886475,
      "learning_rate": 5.147258560467683e-05,
      "loss": 1.914,
      "step": 5452
    },
    {
      "epoch": 0.9823012834947084,
      "grad_norm": 1.295697569847107,
      "learning_rate": 5.1458430164246706e-05,
      "loss": 1.984,
      "step": 5453
    },
    {
      "epoch": 0.9824814231029048,
      "grad_norm": 1.3407965898513794,
      "learning_rate": 5.144427460682203e-05,
      "loss": 1.8529,
      "step": 5454
    },
    {
      "epoch": 0.982661562711101,
      "grad_norm": 1.3061943054199219,
      "learning_rate": 5.143011893353835e-05,
      "loss": 1.9232,
      "step": 5455
    },
    {
      "epoch": 0.9828417023192975,
      "grad_norm": 1.3617926836013794,
      "learning_rate": 5.141596314553124e-05,
      "loss": 2.0016,
      "step": 5456
    },
    {
      "epoch": 0.9830218419274938,
      "grad_norm": 1.422461748123169,
      "learning_rate": 5.140180724393624e-05,
      "loss": 1.8131,
      "step": 5457
    },
    {
      "epoch": 0.9832019815356902,
      "grad_norm": 2.5082414150238037,
      "learning_rate": 5.1387651229888945e-05,
      "loss": 2.0925,
      "step": 5458
    },
    {
      "epoch": 0.9833821211438866,
      "grad_norm": 2.0095503330230713,
      "learning_rate": 5.137349510452495e-05,
      "loss": 2.3096,
      "step": 5459
    },
    {
      "epoch": 0.9835622607520829,
      "grad_norm": 1.7564771175384521,
      "learning_rate": 5.1359338868979834e-05,
      "loss": 2.1726,
      "step": 5460
    },
    {
      "epoch": 0.9837424003602793,
      "grad_norm": 1.3236967325210571,
      "learning_rate": 5.1345182524389236e-05,
      "loss": 1.6855,
      "step": 5461
    },
    {
      "epoch": 0.9839225399684756,
      "grad_norm": 1.2989704608917236,
      "learning_rate": 5.133102607188874e-05,
      "loss": 1.5133,
      "step": 5462
    },
    {
      "epoch": 0.984102679576672,
      "grad_norm": 1.2168680429458618,
      "learning_rate": 5.1316869512613984e-05,
      "loss": 1.4162,
      "step": 5463
    },
    {
      "epoch": 0.9842828191848683,
      "grad_norm": 1.3532770872116089,
      "learning_rate": 5.1302712847700604e-05,
      "loss": 1.2964,
      "step": 5464
    },
    {
      "epoch": 0.9844629587930647,
      "grad_norm": 1.3319768905639648,
      "learning_rate": 5.1288556078284223e-05,
      "loss": 1.6427,
      "step": 5465
    },
    {
      "epoch": 0.984643098401261,
      "grad_norm": 1.2991929054260254,
      "learning_rate": 5.12743992055005e-05,
      "loss": 1.479,
      "step": 5466
    },
    {
      "epoch": 0.9848232380094574,
      "grad_norm": 1.2791972160339355,
      "learning_rate": 5.12602422304851e-05,
      "loss": 1.6004,
      "step": 5467
    },
    {
      "epoch": 0.9850033776176537,
      "grad_norm": 1.3292728662490845,
      "learning_rate": 5.124608515437368e-05,
      "loss": 1.4978,
      "step": 5468
    },
    {
      "epoch": 0.9851835172258501,
      "grad_norm": 1.4439105987548828,
      "learning_rate": 5.1231927978301906e-05,
      "loss": 1.4954,
      "step": 5469
    },
    {
      "epoch": 0.9853636568340464,
      "grad_norm": 1.3360912799835205,
      "learning_rate": 5.121777070340549e-05,
      "loss": 1.5736,
      "step": 5470
    },
    {
      "epoch": 0.9855437964422428,
      "grad_norm": 1.3444323539733887,
      "learning_rate": 5.120361333082009e-05,
      "loss": 1.5729,
      "step": 5471
    },
    {
      "epoch": 0.9857239360504391,
      "grad_norm": 1.2622430324554443,
      "learning_rate": 5.1189455861681425e-05,
      "loss": 1.4671,
      "step": 5472
    },
    {
      "epoch": 0.9859040756586355,
      "grad_norm": 1.5953948497772217,
      "learning_rate": 5.117529829712517e-05,
      "loss": 1.6837,
      "step": 5473
    },
    {
      "epoch": 0.9860842152668318,
      "grad_norm": 1.3227770328521729,
      "learning_rate": 5.116114063828706e-05,
      "loss": 1.4916,
      "step": 5474
    },
    {
      "epoch": 0.9862643548750282,
      "grad_norm": 1.3851453065872192,
      "learning_rate": 5.114698288630283e-05,
      "loss": 1.3644,
      "step": 5475
    },
    {
      "epoch": 0.9864444944832245,
      "grad_norm": 1.356567621231079,
      "learning_rate": 5.1132825042308186e-05,
      "loss": 1.4013,
      "step": 5476
    },
    {
      "epoch": 0.9866246340914209,
      "grad_norm": 1.4187607765197754,
      "learning_rate": 5.1118667107438864e-05,
      "loss": 1.5896,
      "step": 5477
    },
    {
      "epoch": 0.9868047736996172,
      "grad_norm": 1.2975939512252808,
      "learning_rate": 5.110450908283062e-05,
      "loss": 1.3973,
      "step": 5478
    },
    {
      "epoch": 0.9869849133078136,
      "grad_norm": 1.4173749685287476,
      "learning_rate": 5.109035096961919e-05,
      "loss": 1.7043,
      "step": 5479
    },
    {
      "epoch": 0.9871650529160099,
      "grad_norm": 1.3138997554779053,
      "learning_rate": 5.1076192768940333e-05,
      "loss": 1.4212,
      "step": 5480
    },
    {
      "epoch": 0.9873451925242063,
      "grad_norm": 1.4722062349319458,
      "learning_rate": 5.106203448192981e-05,
      "loss": 1.5128,
      "step": 5481
    },
    {
      "epoch": 0.9875253321324026,
      "grad_norm": 1.4500669240951538,
      "learning_rate": 5.104787610972341e-05,
      "loss": 1.6007,
      "step": 5482
    },
    {
      "epoch": 0.987705471740599,
      "grad_norm": 1.5035229921340942,
      "learning_rate": 5.10337176534569e-05,
      "loss": 1.6317,
      "step": 5483
    },
    {
      "epoch": 0.9878856113487953,
      "grad_norm": 1.3849323987960815,
      "learning_rate": 5.101955911426607e-05,
      "loss": 1.6066,
      "step": 5484
    },
    {
      "epoch": 0.9880657509569917,
      "grad_norm": 1.3403972387313843,
      "learning_rate": 5.100540049328669e-05,
      "loss": 1.5829,
      "step": 5485
    },
    {
      "epoch": 0.988245890565188,
      "grad_norm": 1.5764458179473877,
      "learning_rate": 5.0991241791654576e-05,
      "loss": 1.5613,
      "step": 5486
    },
    {
      "epoch": 0.9884260301733844,
      "grad_norm": 1.3477911949157715,
      "learning_rate": 5.097708301050552e-05,
      "loss": 1.5545,
      "step": 5487
    },
    {
      "epoch": 0.9886061697815808,
      "grad_norm": 1.506456971168518,
      "learning_rate": 5.096292415097534e-05,
      "loss": 1.5169,
      "step": 5488
    },
    {
      "epoch": 0.9887863093897771,
      "grad_norm": 1.4868850708007812,
      "learning_rate": 5.094876521419989e-05,
      "loss": 1.573,
      "step": 5489
    },
    {
      "epoch": 0.9889664489979735,
      "grad_norm": 1.4440927505493164,
      "learning_rate": 5.093460620131492e-05,
      "loss": 1.5389,
      "step": 5490
    },
    {
      "epoch": 0.9891465886061698,
      "grad_norm": 1.4233566522598267,
      "learning_rate": 5.0920447113456306e-05,
      "loss": 1.7414,
      "step": 5491
    },
    {
      "epoch": 0.9893267282143662,
      "grad_norm": 1.4849672317504883,
      "learning_rate": 5.0906287951759866e-05,
      "loss": 1.5977,
      "step": 5492
    },
    {
      "epoch": 0.9895068678225625,
      "grad_norm": 1.417472004890442,
      "learning_rate": 5.0892128717361456e-05,
      "loss": 1.5074,
      "step": 5493
    },
    {
      "epoch": 0.9896870074307589,
      "grad_norm": 1.471031665802002,
      "learning_rate": 5.087796941139691e-05,
      "loss": 1.7083,
      "step": 5494
    },
    {
      "epoch": 0.9898671470389552,
      "grad_norm": 1.314753532409668,
      "learning_rate": 5.086381003500209e-05,
      "loss": 1.4298,
      "step": 5495
    },
    {
      "epoch": 0.9900472866471516,
      "grad_norm": 1.4409737586975098,
      "learning_rate": 5.084965058931286e-05,
      "loss": 1.5632,
      "step": 5496
    },
    {
      "epoch": 0.9902274262553479,
      "grad_norm": 1.2563427686691284,
      "learning_rate": 5.0835491075465045e-05,
      "loss": 1.3845,
      "step": 5497
    },
    {
      "epoch": 0.9904075658635443,
      "grad_norm": 1.5046136379241943,
      "learning_rate": 5.082133149459457e-05,
      "loss": 1.4388,
      "step": 5498
    },
    {
      "epoch": 0.9905877054717406,
      "grad_norm": 1.3909345865249634,
      "learning_rate": 5.0807171847837274e-05,
      "loss": 1.279,
      "step": 5499
    },
    {
      "epoch": 0.990767845079937,
      "grad_norm": 1.4372528791427612,
      "learning_rate": 5.0793012136329044e-05,
      "loss": 1.4589,
      "step": 5500
    },
    {
      "epoch": 0.9909479846881333,
      "grad_norm": 1.443535566329956,
      "learning_rate": 5.077885236120577e-05,
      "loss": 2.205,
      "step": 5501
    },
    {
      "epoch": 0.9911281242963297,
      "grad_norm": 1.1812182664871216,
      "learning_rate": 5.076469252360333e-05,
      "loss": 1.7309,
      "step": 5502
    },
    {
      "epoch": 0.991308263904526,
      "grad_norm": 1.285457730293274,
      "learning_rate": 5.075053262465763e-05,
      "loss": 1.6994,
      "step": 5503
    },
    {
      "epoch": 0.9914884035127224,
      "grad_norm": 1.4190510511398315,
      "learning_rate": 5.073637266550456e-05,
      "loss": 1.9109,
      "step": 5504
    },
    {
      "epoch": 0.9916685431209187,
      "grad_norm": 1.3112819194793701,
      "learning_rate": 5.072221264728007e-05,
      "loss": 1.9456,
      "step": 5505
    },
    {
      "epoch": 0.9918486827291151,
      "grad_norm": 1.3437947034835815,
      "learning_rate": 5.070805257111999e-05,
      "loss": 1.6904,
      "step": 5506
    },
    {
      "epoch": 0.9920288223373114,
      "grad_norm": 1.30347740650177,
      "learning_rate": 5.069389243816031e-05,
      "loss": 1.9173,
      "step": 5507
    },
    {
      "epoch": 0.9922089619455078,
      "grad_norm": 1.4260587692260742,
      "learning_rate": 5.0679732249536905e-05,
      "loss": 1.9504,
      "step": 5508
    },
    {
      "epoch": 0.9923891015537041,
      "grad_norm": 1.5754131078720093,
      "learning_rate": 5.0665572006385695e-05,
      "loss": 1.8174,
      "step": 5509
    },
    {
      "epoch": 0.9925692411619005,
      "grad_norm": 1.795831561088562,
      "learning_rate": 5.065141170984263e-05,
      "loss": 2.2449,
      "step": 5510
    },
    {
      "epoch": 0.9927493807700968,
      "grad_norm": 1.3534268140792847,
      "learning_rate": 5.063725136104365e-05,
      "loss": 1.6678,
      "step": 5511
    },
    {
      "epoch": 0.9929295203782932,
      "grad_norm": 1.368416666984558,
      "learning_rate": 5.062309096112467e-05,
      "loss": 1.785,
      "step": 5512
    },
    {
      "epoch": 0.9931096599864895,
      "grad_norm": 1.3788455724716187,
      "learning_rate": 5.060893051122163e-05,
      "loss": 1.5657,
      "step": 5513
    },
    {
      "epoch": 0.9932897995946859,
      "grad_norm": 1.2808939218521118,
      "learning_rate": 5.059477001247048e-05,
      "loss": 1.4714,
      "step": 5514
    },
    {
      "epoch": 0.9934699392028822,
      "grad_norm": 1.336501121520996,
      "learning_rate": 5.0580609466007176e-05,
      "loss": 1.6352,
      "step": 5515
    },
    {
      "epoch": 0.9936500788110786,
      "grad_norm": 1.1835541725158691,
      "learning_rate": 5.056644887296766e-05,
      "loss": 1.5222,
      "step": 5516
    },
    {
      "epoch": 0.993830218419275,
      "grad_norm": 1.2137409448623657,
      "learning_rate": 5.05522882344879e-05,
      "loss": 1.3276,
      "step": 5517
    },
    {
      "epoch": 0.9940103580274713,
      "grad_norm": 1.2530930042266846,
      "learning_rate": 5.053812755170383e-05,
      "loss": 1.4916,
      "step": 5518
    },
    {
      "epoch": 0.9941904976356677,
      "grad_norm": 1.301798701286316,
      "learning_rate": 5.052396682575145e-05,
      "loss": 1.2784,
      "step": 5519
    },
    {
      "epoch": 0.994370637243864,
      "grad_norm": 1.2902483940124512,
      "learning_rate": 5.050980605776668e-05,
      "loss": 1.3717,
      "step": 5520
    },
    {
      "epoch": 0.9945507768520604,
      "grad_norm": 1.370252251625061,
      "learning_rate": 5.049564524888555e-05,
      "loss": 1.6479,
      "step": 5521
    },
    {
      "epoch": 0.9947309164602567,
      "grad_norm": 1.4862064123153687,
      "learning_rate": 5.048148440024398e-05,
      "loss": 1.8263,
      "step": 5522
    },
    {
      "epoch": 0.9949110560684531,
      "grad_norm": 1.2226202487945557,
      "learning_rate": 5.046732351297796e-05,
      "loss": 1.3234,
      "step": 5523
    },
    {
      "epoch": 0.9950911956766494,
      "grad_norm": 1.2787630558013916,
      "learning_rate": 5.045316258822349e-05,
      "loss": 1.4156,
      "step": 5524
    },
    {
      "epoch": 0.9952713352848458,
      "grad_norm": 1.4211828708648682,
      "learning_rate": 5.043900162711652e-05,
      "loss": 1.869,
      "step": 5525
    },
    {
      "epoch": 0.9954514748930421,
      "grad_norm": 1.5089571475982666,
      "learning_rate": 5.042484063079307e-05,
      "loss": 1.7412,
      "step": 5526
    },
    {
      "epoch": 0.9956316145012385,
      "grad_norm": 1.439766526222229,
      "learning_rate": 5.041067960038911e-05,
      "loss": 1.5093,
      "step": 5527
    },
    {
      "epoch": 0.9958117541094348,
      "grad_norm": 1.345784068107605,
      "learning_rate": 5.039651853704063e-05,
      "loss": 1.4175,
      "step": 5528
    },
    {
      "epoch": 0.9959918937176312,
      "grad_norm": 1.3591045141220093,
      "learning_rate": 5.038235744188362e-05,
      "loss": 1.3927,
      "step": 5529
    },
    {
      "epoch": 0.9961720333258275,
      "grad_norm": 1.3794161081314087,
      "learning_rate": 5.036819631605409e-05,
      "loss": 1.4927,
      "step": 5530
    },
    {
      "epoch": 0.9963521729340239,
      "grad_norm": 1.3785758018493652,
      "learning_rate": 5.035403516068803e-05,
      "loss": 1.4733,
      "step": 5531
    },
    {
      "epoch": 0.9965323125422202,
      "grad_norm": 1.4453762769699097,
      "learning_rate": 5.033987397692144e-05,
      "loss": 1.5312,
      "step": 5532
    },
    {
      "epoch": 0.9967124521504166,
      "grad_norm": 1.569035291671753,
      "learning_rate": 5.032571276589032e-05,
      "loss": 1.4063,
      "step": 5533
    },
    {
      "epoch": 0.9968925917586129,
      "grad_norm": 1.4120374917984009,
      "learning_rate": 5.0311551528730685e-05,
      "loss": 1.3971,
      "step": 5534
    },
    {
      "epoch": 0.9970727313668093,
      "grad_norm": 1.3823720216751099,
      "learning_rate": 5.029739026657854e-05,
      "loss": 1.4651,
      "step": 5535
    },
    {
      "epoch": 0.9972528709750056,
      "grad_norm": 1.3192176818847656,
      "learning_rate": 5.0283228980569883e-05,
      "loss": 1.3163,
      "step": 5536
    },
    {
      "epoch": 0.997433010583202,
      "grad_norm": 1.4467960596084595,
      "learning_rate": 5.026906767184075e-05,
      "loss": 1.5889,
      "step": 5537
    },
    {
      "epoch": 0.9976131501913983,
      "grad_norm": 1.411865234375,
      "learning_rate": 5.025490634152714e-05,
      "loss": 1.5999,
      "step": 5538
    },
    {
      "epoch": 0.9977932897995947,
      "grad_norm": 1.5603841543197632,
      "learning_rate": 5.0240744990765055e-05,
      "loss": 1.535,
      "step": 5539
    },
    {
      "epoch": 0.997973429407791,
      "grad_norm": 1.41175377368927,
      "learning_rate": 5.022658362069053e-05,
      "loss": 1.4259,
      "step": 5540
    },
    {
      "epoch": 0.9981535690159874,
      "grad_norm": 1.5095746517181396,
      "learning_rate": 5.021242223243956e-05,
      "loss": 1.464,
      "step": 5541
    },
    {
      "epoch": 0.9983337086241837,
      "grad_norm": 1.488194465637207,
      "learning_rate": 5.0198260827148206e-05,
      "loss": 1.6052,
      "step": 5542
    },
    {
      "epoch": 0.9985138482323801,
      "grad_norm": 1.5387918949127197,
      "learning_rate": 5.018409940595247e-05,
      "loss": 1.6405,
      "step": 5543
    },
    {
      "epoch": 0.9986939878405764,
      "grad_norm": 1.4464972019195557,
      "learning_rate": 5.016993796998835e-05,
      "loss": 1.3603,
      "step": 5544
    },
    {
      "epoch": 0.9988741274487728,
      "grad_norm": 1.234643816947937,
      "learning_rate": 5.0155776520391896e-05,
      "loss": 1.2346,
      "step": 5545
    },
    {
      "epoch": 0.9990542670569692,
      "grad_norm": 1.556142807006836,
      "learning_rate": 5.014161505829912e-05,
      "loss": 1.7073,
      "step": 5546
    },
    {
      "epoch": 0.9992344066651655,
      "grad_norm": 1.4931926727294922,
      "learning_rate": 5.012745358484605e-05,
      "loss": 1.5738,
      "step": 5547
    },
    {
      "epoch": 0.9994145462733619,
      "grad_norm": 1.6659613847732544,
      "learning_rate": 5.011329210116873e-05,
      "loss": 1.3528,
      "step": 5548
    },
    {
      "epoch": 0.9995946858815582,
      "grad_norm": 1.372674584388733,
      "learning_rate": 5.0099130608403164e-05,
      "loss": 1.4698,
      "step": 5549
    },
    {
      "epoch": 0.9997748254897546,
      "grad_norm": 1.2082171440124512,
      "learning_rate": 5.00849691076854e-05,
      "loss": 1.1453,
      "step": 5550
    },
    {
      "epoch": 0.9999549650979509,
      "grad_norm": 1.3851683139801025,
      "learning_rate": 5.007080760015144e-05,
      "loss": 1.8766,
      "step": 5551
    },
    {
      "epoch": 0.9999549650979509,
      "eval_loss": 1.5813547372817993,
      "eval_runtime": 185.5209,
      "eval_samples_per_second": 50.399,
      "eval_steps_per_second": 12.602,
      "step": 5551
    },
    {
      "epoch": 1.0001351047061473,
      "grad_norm": 3.594057559967041,
      "learning_rate": 5.0056646086937344e-05,
      "loss": 2.2345,
      "step": 5552
    },
    {
      "epoch": 1.0003152443143437,
      "grad_norm": 1.1515469551086426,
      "learning_rate": 5.0042484569179114e-05,
      "loss": 1.7658,
      "step": 5553
    },
    {
      "epoch": 1.00049538392254,
      "grad_norm": 1.23834228515625,
      "learning_rate": 5.00283230480128e-05,
      "loss": 1.7991,
      "step": 5554
    },
    {
      "epoch": 1.0006755235307363,
      "grad_norm": 1.2544416189193726,
      "learning_rate": 5.001416152457441e-05,
      "loss": 1.9802,
      "step": 5555
    },
    {
      "epoch": 1.0008556631389327,
      "grad_norm": 1.2491496801376343,
      "learning_rate": 5e-05,
      "loss": 1.6689,
      "step": 5556
    },
    {
      "epoch": 1.001035802747129,
      "grad_norm": 1.207411766052246,
      "learning_rate": 4.99858384754256e-05,
      "loss": 1.6938,
      "step": 5557
    },
    {
      "epoch": 1.0012159423553253,
      "grad_norm": 1.329949975013733,
      "learning_rate": 4.997167695198722e-05,
      "loss": 1.7802,
      "step": 5558
    },
    {
      "epoch": 1.0013960819635217,
      "grad_norm": 1.3830840587615967,
      "learning_rate": 4.99575154308209e-05,
      "loss": 1.665,
      "step": 5559
    },
    {
      "epoch": 1.001576221571718,
      "grad_norm": 1.5742459297180176,
      "learning_rate": 4.9943353913062674e-05,
      "loss": 2.0974,
      "step": 5560
    },
    {
      "epoch": 1.0017563611799145,
      "grad_norm": 1.754102110862732,
      "learning_rate": 4.992919239984857e-05,
      "loss": 2.1248,
      "step": 5561
    },
    {
      "epoch": 1.0019365007881107,
      "grad_norm": 1.3664779663085938,
      "learning_rate": 4.9915030892314625e-05,
      "loss": 1.7125,
      "step": 5562
    },
    {
      "epoch": 1.002116640396307,
      "grad_norm": 1.286925196647644,
      "learning_rate": 4.990086939159684e-05,
      "loss": 1.5467,
      "step": 5563
    },
    {
      "epoch": 1.0022967800045035,
      "grad_norm": 1.2146855592727661,
      "learning_rate": 4.988670789883128e-05,
      "loss": 1.2794,
      "step": 5564
    },
    {
      "epoch": 1.0024769196127,
      "grad_norm": 1.346280574798584,
      "learning_rate": 4.9872546415153954e-05,
      "loss": 1.3649,
      "step": 5565
    },
    {
      "epoch": 1.002657059220896,
      "grad_norm": 1.497022271156311,
      "learning_rate": 4.98583849417009e-05,
      "loss": 1.6313,
      "step": 5566
    },
    {
      "epoch": 1.0028371988290925,
      "grad_norm": 1.3814294338226318,
      "learning_rate": 4.984422347960813e-05,
      "loss": 1.575,
      "step": 5567
    },
    {
      "epoch": 1.003017338437289,
      "grad_norm": 1.2965688705444336,
      "learning_rate": 4.9830062030011657e-05,
      "loss": 1.3843,
      "step": 5568
    },
    {
      "epoch": 1.0031974780454853,
      "grad_norm": 1.2462544441223145,
      "learning_rate": 4.981590059404754e-05,
      "loss": 1.2929,
      "step": 5569
    },
    {
      "epoch": 1.0033776176536815,
      "grad_norm": 1.2687498331069946,
      "learning_rate": 4.9801739172851806e-05,
      "loss": 1.528,
      "step": 5570
    },
    {
      "epoch": 1.003557757261878,
      "grad_norm": 1.3159748315811157,
      "learning_rate": 4.978757776756045e-05,
      "loss": 1.4044,
      "step": 5571
    },
    {
      "epoch": 1.0037378968700743,
      "grad_norm": 1.2301899194717407,
      "learning_rate": 4.9773416379309496e-05,
      "loss": 1.3239,
      "step": 5572
    },
    {
      "epoch": 1.0039180364782707,
      "grad_norm": 1.318537950515747,
      "learning_rate": 4.975925500923495e-05,
      "loss": 1.6227,
      "step": 5573
    },
    {
      "epoch": 1.0040981760864671,
      "grad_norm": 1.2856943607330322,
      "learning_rate": 4.9745093658472876e-05,
      "loss": 1.6121,
      "step": 5574
    },
    {
      "epoch": 1.0042783156946633,
      "grad_norm": 1.3267793655395508,
      "learning_rate": 4.9730932328159255e-05,
      "loss": 1.2396,
      "step": 5575
    },
    {
      "epoch": 1.0044584553028597,
      "grad_norm": 1.5443915128707886,
      "learning_rate": 4.971677101943013e-05,
      "loss": 1.244,
      "step": 5576
    },
    {
      "epoch": 1.0046385949110561,
      "grad_norm": 1.492397427558899,
      "learning_rate": 4.970260973342146e-05,
      "loss": 1.6181,
      "step": 5577
    },
    {
      "epoch": 1.0048187345192525,
      "grad_norm": 1.349518895149231,
      "learning_rate": 4.968844847126932e-05,
      "loss": 1.6127,
      "step": 5578
    },
    {
      "epoch": 1.0049988741274487,
      "grad_norm": 1.4305379390716553,
      "learning_rate": 4.967428723410969e-05,
      "loss": 1.5743,
      "step": 5579
    },
    {
      "epoch": 1.005179013735645,
      "grad_norm": 1.5587505102157593,
      "learning_rate": 4.966012602307857e-05,
      "loss": 1.6561,
      "step": 5580
    },
    {
      "epoch": 1.0053591533438415,
      "grad_norm": 1.4036973714828491,
      "learning_rate": 4.9645964839311985e-05,
      "loss": 1.3369,
      "step": 5581
    },
    {
      "epoch": 1.005539292952038,
      "grad_norm": 1.286837100982666,
      "learning_rate": 4.963180368394592e-05,
      "loss": 1.1369,
      "step": 5582
    },
    {
      "epoch": 1.005719432560234,
      "grad_norm": 1.6943107843399048,
      "learning_rate": 4.9617642558116384e-05,
      "loss": 1.7455,
      "step": 5583
    },
    {
      "epoch": 1.0058995721684305,
      "grad_norm": 1.4491300582885742,
      "learning_rate": 4.9603481462959384e-05,
      "loss": 1.6461,
      "step": 5584
    },
    {
      "epoch": 1.006079711776627,
      "grad_norm": 1.4178802967071533,
      "learning_rate": 4.9589320399610904e-05,
      "loss": 1.3058,
      "step": 5585
    },
    {
      "epoch": 1.0062598513848233,
      "grad_norm": 1.4115400314331055,
      "learning_rate": 4.9575159369206944e-05,
      "loss": 1.4939,
      "step": 5586
    },
    {
      "epoch": 1.0064399909930195,
      "grad_norm": 1.3227558135986328,
      "learning_rate": 4.9560998372883476e-05,
      "loss": 1.2713,
      "step": 5587
    },
    {
      "epoch": 1.006620130601216,
      "grad_norm": 1.4719223976135254,
      "learning_rate": 4.954683741177652e-05,
      "loss": 1.7106,
      "step": 5588
    },
    {
      "epoch": 1.0068002702094123,
      "grad_norm": 1.3001450300216675,
      "learning_rate": 4.9532676487022046e-05,
      "loss": 1.4296,
      "step": 5589
    },
    {
      "epoch": 1.0069804098176087,
      "grad_norm": 1.227793574333191,
      "learning_rate": 4.9518515599756035e-05,
      "loss": 1.1813,
      "step": 5590
    },
    {
      "epoch": 1.007160549425805,
      "grad_norm": 1.5430632829666138,
      "learning_rate": 4.950435475111446e-05,
      "loss": 1.5679,
      "step": 5591
    },
    {
      "epoch": 1.0073406890340013,
      "grad_norm": 1.5058188438415527,
      "learning_rate": 4.9490193942233315e-05,
      "loss": 1.6437,
      "step": 5592
    },
    {
      "epoch": 1.0075208286421977,
      "grad_norm": 1.3336890935897827,
      "learning_rate": 4.9476033174248564e-05,
      "loss": 1.187,
      "step": 5593
    },
    {
      "epoch": 1.0077009682503941,
      "grad_norm": 1.5512518882751465,
      "learning_rate": 4.946187244829618e-05,
      "loss": 1.5464,
      "step": 5594
    },
    {
      "epoch": 1.0078811078585903,
      "grad_norm": 1.434260368347168,
      "learning_rate": 4.944771176551212e-05,
      "loss": 1.4756,
      "step": 5595
    },
    {
      "epoch": 1.0080612474667867,
      "grad_norm": 1.3953254222869873,
      "learning_rate": 4.943355112703234e-05,
      "loss": 1.4091,
      "step": 5596
    },
    {
      "epoch": 1.0082413870749831,
      "grad_norm": 1.5238646268844604,
      "learning_rate": 4.9419390533992835e-05,
      "loss": 1.5481,
      "step": 5597
    },
    {
      "epoch": 1.0084215266831795,
      "grad_norm": 1.3699073791503906,
      "learning_rate": 4.940522998752953e-05,
      "loss": 1.2815,
      "step": 5598
    },
    {
      "epoch": 1.0086016662913757,
      "grad_norm": 1.3145962953567505,
      "learning_rate": 4.939106948877839e-05,
      "loss": 1.2487,
      "step": 5599
    },
    {
      "epoch": 1.0087818058995721,
      "grad_norm": 1.2763704061508179,
      "learning_rate": 4.937690903887535e-05,
      "loss": 1.0733,
      "step": 5600
    },
    {
      "epoch": 1.0089619455077685,
      "grad_norm": 1.2779762744903564,
      "learning_rate": 4.936274863895635e-05,
      "loss": 1.3373,
      "step": 5601
    },
    {
      "epoch": 1.009142085115965,
      "grad_norm": 1.236710548400879,
      "learning_rate": 4.9348588290157375e-05,
      "loss": 1.8283,
      "step": 5602
    },
    {
      "epoch": 1.0093222247241613,
      "grad_norm": 1.3207132816314697,
      "learning_rate": 4.933442799361432e-05,
      "loss": 1.7135,
      "step": 5603
    },
    {
      "epoch": 1.0095023643323575,
      "grad_norm": 1.2202510833740234,
      "learning_rate": 4.932026775046312e-05,
      "loss": 1.5694,
      "step": 5604
    },
    {
      "epoch": 1.009682503940554,
      "grad_norm": 1.2876667976379395,
      "learning_rate": 4.9306107561839696e-05,
      "loss": 1.7128,
      "step": 5605
    },
    {
      "epoch": 1.0098626435487503,
      "grad_norm": 1.3892180919647217,
      "learning_rate": 4.9291947428880005e-05,
      "loss": 1.9629,
      "step": 5606
    },
    {
      "epoch": 1.0100427831569467,
      "grad_norm": 1.3952772617340088,
      "learning_rate": 4.9277787352719944e-05,
      "loss": 1.8067,
      "step": 5607
    },
    {
      "epoch": 1.010222922765143,
      "grad_norm": 1.5145434141159058,
      "learning_rate": 4.9263627334495446e-05,
      "loss": 1.8333,
      "step": 5608
    },
    {
      "epoch": 1.0104030623733393,
      "grad_norm": 1.5198681354522705,
      "learning_rate": 4.924946737534239e-05,
      "loss": 1.6511,
      "step": 5609
    },
    {
      "epoch": 1.0105832019815357,
      "grad_norm": 1.6591224670410156,
      "learning_rate": 4.923530747639668e-05,
      "loss": 2.1028,
      "step": 5610
    },
    {
      "epoch": 1.0107633415897321,
      "grad_norm": 1.8929842710494995,
      "learning_rate": 4.922114763879424e-05,
      "loss": 1.9754,
      "step": 5611
    },
    {
      "epoch": 1.0109434811979283,
      "grad_norm": 1.4713486433029175,
      "learning_rate": 4.920698786367097e-05,
      "loss": 1.6352,
      "step": 5612
    },
    {
      "epoch": 1.0111236208061247,
      "grad_norm": 1.2287005186080933,
      "learning_rate": 4.919282815216274e-05,
      "loss": 1.3001,
      "step": 5613
    },
    {
      "epoch": 1.0113037604143211,
      "grad_norm": 1.1148327589035034,
      "learning_rate": 4.9178668505405454e-05,
      "loss": 1.4141,
      "step": 5614
    },
    {
      "epoch": 1.0114839000225175,
      "grad_norm": 1.3809200525283813,
      "learning_rate": 4.916450892453495e-05,
      "loss": 1.4582,
      "step": 5615
    },
    {
      "epoch": 1.0116640396307137,
      "grad_norm": 1.37238609790802,
      "learning_rate": 4.915034941068716e-05,
      "loss": 1.6725,
      "step": 5616
    },
    {
      "epoch": 1.0118441792389101,
      "grad_norm": 1.2395992279052734,
      "learning_rate": 4.913618996499792e-05,
      "loss": 1.4422,
      "step": 5617
    },
    {
      "epoch": 1.0120243188471065,
      "grad_norm": 1.3363028764724731,
      "learning_rate": 4.91220305886031e-05,
      "loss": 1.3323,
      "step": 5618
    },
    {
      "epoch": 1.012204458455303,
      "grad_norm": 1.382542610168457,
      "learning_rate": 4.910787128263855e-05,
      "loss": 1.6322,
      "step": 5619
    },
    {
      "epoch": 1.0123845980634991,
      "grad_norm": 1.3107426166534424,
      "learning_rate": 4.909371204824014e-05,
      "loss": 1.6006,
      "step": 5620
    },
    {
      "epoch": 1.0125647376716955,
      "grad_norm": 1.3757141828536987,
      "learning_rate": 4.9079552886543706e-05,
      "loss": 1.4538,
      "step": 5621
    },
    {
      "epoch": 1.012744877279892,
      "grad_norm": 1.3708990812301636,
      "learning_rate": 4.906539379868509e-05,
      "loss": 1.5539,
      "step": 5622
    },
    {
      "epoch": 1.0129250168880883,
      "grad_norm": 1.3575074672698975,
      "learning_rate": 4.905123478580014e-05,
      "loss": 1.427,
      "step": 5623
    },
    {
      "epoch": 1.0131051564962845,
      "grad_norm": 1.1885992288589478,
      "learning_rate": 4.903707584902465e-05,
      "loss": 1.1832,
      "step": 5624
    },
    {
      "epoch": 1.013285296104481,
      "grad_norm": 1.426460862159729,
      "learning_rate": 4.902291698949448e-05,
      "loss": 1.6044,
      "step": 5625
    },
    {
      "epoch": 1.0134654357126773,
      "grad_norm": 1.3127444982528687,
      "learning_rate": 4.900875820834544e-05,
      "loss": 1.2975,
      "step": 5626
    },
    {
      "epoch": 1.0136455753208737,
      "grad_norm": 1.4574682712554932,
      "learning_rate": 4.899459950671333e-05,
      "loss": 1.3662,
      "step": 5627
    },
    {
      "epoch": 1.01382571492907,
      "grad_norm": 1.389467477798462,
      "learning_rate": 4.898044088573395e-05,
      "loss": 1.3365,
      "step": 5628
    },
    {
      "epoch": 1.0140058545372663,
      "grad_norm": 1.3078991174697876,
      "learning_rate": 4.896628234654311e-05,
      "loss": 1.403,
      "step": 5629
    },
    {
      "epoch": 1.0141859941454627,
      "grad_norm": 1.4988057613372803,
      "learning_rate": 4.89521238902766e-05,
      "loss": 1.6475,
      "step": 5630
    },
    {
      "epoch": 1.0143661337536591,
      "grad_norm": 1.3802160024642944,
      "learning_rate": 4.8937965518070195e-05,
      "loss": 1.3365,
      "step": 5631
    },
    {
      "epoch": 1.0145462733618555,
      "grad_norm": 1.4270375967025757,
      "learning_rate": 4.8923807231059685e-05,
      "loss": 1.4756,
      "step": 5632
    },
    {
      "epoch": 1.0147264129700517,
      "grad_norm": 1.396551251411438,
      "learning_rate": 4.890964903038081e-05,
      "loss": 1.4774,
      "step": 5633
    },
    {
      "epoch": 1.0149065525782481,
      "grad_norm": 1.4541343450546265,
      "learning_rate": 4.8895490917169395e-05,
      "loss": 1.5586,
      "step": 5634
    },
    {
      "epoch": 1.0150866921864445,
      "grad_norm": 1.3801171779632568,
      "learning_rate": 4.888133289256115e-05,
      "loss": 1.3555,
      "step": 5635
    },
    {
      "epoch": 1.015266831794641,
      "grad_norm": 1.4155704975128174,
      "learning_rate": 4.886717495769183e-05,
      "loss": 1.5705,
      "step": 5636
    },
    {
      "epoch": 1.0154469714028371,
      "grad_norm": 1.3111226558685303,
      "learning_rate": 4.8853017113697186e-05,
      "loss": 1.2058,
      "step": 5637
    },
    {
      "epoch": 1.0156271110110335,
      "grad_norm": 1.4586259126663208,
      "learning_rate": 4.883885936171293e-05,
      "loss": 1.5129,
      "step": 5638
    },
    {
      "epoch": 1.01580725061923,
      "grad_norm": 1.357161521911621,
      "learning_rate": 4.8824701702874834e-05,
      "loss": 1.3479,
      "step": 5639
    },
    {
      "epoch": 1.0159873902274263,
      "grad_norm": 1.5355088710784912,
      "learning_rate": 4.881054413831859e-05,
      "loss": 1.7294,
      "step": 5640
    },
    {
      "epoch": 1.0161675298356225,
      "grad_norm": 1.4661026000976562,
      "learning_rate": 4.879638666917993e-05,
      "loss": 1.5689,
      "step": 5641
    },
    {
      "epoch": 1.016347669443819,
      "grad_norm": 1.311046838760376,
      "learning_rate": 4.878222929659453e-05,
      "loss": 1.3014,
      "step": 5642
    },
    {
      "epoch": 1.0165278090520153,
      "grad_norm": 1.6099871397018433,
      "learning_rate": 4.8768072021698085e-05,
      "loss": 1.6142,
      "step": 5643
    },
    {
      "epoch": 1.0167079486602117,
      "grad_norm": 1.4108574390411377,
      "learning_rate": 4.875391484562632e-05,
      "loss": 1.3236,
      "step": 5644
    },
    {
      "epoch": 1.016888088268408,
      "grad_norm": 1.4383701086044312,
      "learning_rate": 4.873975776951491e-05,
      "loss": 1.2987,
      "step": 5645
    },
    {
      "epoch": 1.0170682278766043,
      "grad_norm": 1.5219625234603882,
      "learning_rate": 4.872560079449951e-05,
      "loss": 1.6096,
      "step": 5646
    },
    {
      "epoch": 1.0172483674848007,
      "grad_norm": 1.450549602508545,
      "learning_rate": 4.871144392171578e-05,
      "loss": 1.3545,
      "step": 5647
    },
    {
      "epoch": 1.0174285070929971,
      "grad_norm": 1.5298703908920288,
      "learning_rate": 4.869728715229941e-05,
      "loss": 1.5569,
      "step": 5648
    },
    {
      "epoch": 1.0176086467011933,
      "grad_norm": 1.4178144931793213,
      "learning_rate": 4.868313048738602e-05,
      "loss": 1.1251,
      "step": 5649
    },
    {
      "epoch": 1.0177887863093897,
      "grad_norm": 1.3194656372070312,
      "learning_rate": 4.866897392811126e-05,
      "loss": 1.2814,
      "step": 5650
    },
    {
      "epoch": 1.0179689259175861,
      "grad_norm": 1.3082959651947021,
      "learning_rate": 4.8654817475610776e-05,
      "loss": 1.0645,
      "step": 5651
    },
    {
      "epoch": 1.0181490655257825,
      "grad_norm": 1.2600536346435547,
      "learning_rate": 4.864066113102016e-05,
      "loss": 1.658,
      "step": 5652
    },
    {
      "epoch": 1.0183292051339787,
      "grad_norm": 1.3712444305419922,
      "learning_rate": 4.862650489547506e-05,
      "loss": 1.9719,
      "step": 5653
    },
    {
      "epoch": 1.0185093447421751,
      "grad_norm": 1.3431187868118286,
      "learning_rate": 4.8612348770111066e-05,
      "loss": 1.6319,
      "step": 5654
    },
    {
      "epoch": 1.0186894843503715,
      "grad_norm": 1.2910689115524292,
      "learning_rate": 4.8598192756063774e-05,
      "loss": 1.9287,
      "step": 5655
    },
    {
      "epoch": 1.018869623958568,
      "grad_norm": 1.3449338674545288,
      "learning_rate": 4.8584036854468785e-05,
      "loss": 1.7589,
      "step": 5656
    },
    {
      "epoch": 1.0190497635667644,
      "grad_norm": 1.3084237575531006,
      "learning_rate": 4.856988106646165e-05,
      "loss": 1.7129,
      "step": 5657
    },
    {
      "epoch": 1.0192299031749605,
      "grad_norm": 1.4933860301971436,
      "learning_rate": 4.855572539317798e-05,
      "loss": 1.7005,
      "step": 5658
    },
    {
      "epoch": 1.019410042783157,
      "grad_norm": 1.3637412786483765,
      "learning_rate": 4.8541569835753306e-05,
      "loss": 1.5829,
      "step": 5659
    },
    {
      "epoch": 1.0195901823913534,
      "grad_norm": 1.538561463356018,
      "learning_rate": 4.8527414395323196e-05,
      "loss": 1.7994,
      "step": 5660
    },
    {
      "epoch": 1.0197703219995498,
      "grad_norm": 1.685618281364441,
      "learning_rate": 4.851325907302317e-05,
      "loss": 1.7449,
      "step": 5661
    },
    {
      "epoch": 1.019950461607746,
      "grad_norm": 1.3408700227737427,
      "learning_rate": 4.849910386998879e-05,
      "loss": 1.5771,
      "step": 5662
    },
    {
      "epoch": 1.0201306012159423,
      "grad_norm": 1.4169869422912598,
      "learning_rate": 4.8484948787355565e-05,
      "loss": 1.4825,
      "step": 5663
    },
    {
      "epoch": 1.0203107408241388,
      "grad_norm": 1.2307233810424805,
      "learning_rate": 4.8470793826259015e-05,
      "loss": 1.3632,
      "step": 5664
    },
    {
      "epoch": 1.0204908804323352,
      "grad_norm": 1.3273478746414185,
      "learning_rate": 4.8456638987834635e-05,
      "loss": 1.3481,
      "step": 5665
    },
    {
      "epoch": 1.0206710200405313,
      "grad_norm": 1.2714406251907349,
      "learning_rate": 4.84424842732179e-05,
      "loss": 1.6143,
      "step": 5666
    },
    {
      "epoch": 1.0208511596487277,
      "grad_norm": 1.3952585458755493,
      "learning_rate": 4.842832968354436e-05,
      "loss": 1.5994,
      "step": 5667
    },
    {
      "epoch": 1.0210312992569242,
      "grad_norm": 1.3672749996185303,
      "learning_rate": 4.841417521994943e-05,
      "loss": 1.5182,
      "step": 5668
    },
    {
      "epoch": 1.0212114388651206,
      "grad_norm": 1.3039138317108154,
      "learning_rate": 4.8400020883568594e-05,
      "loss": 1.4702,
      "step": 5669
    },
    {
      "epoch": 1.0213915784733167,
      "grad_norm": 1.3019334077835083,
      "learning_rate": 4.838586667553728e-05,
      "loss": 1.3222,
      "step": 5670
    },
    {
      "epoch": 1.0215717180815131,
      "grad_norm": 1.3701164722442627,
      "learning_rate": 4.837171259699097e-05,
      "loss": 1.5042,
      "step": 5671
    },
    {
      "epoch": 1.0217518576897096,
      "grad_norm": 1.2783962488174438,
      "learning_rate": 4.8357558649065096e-05,
      "loss": 1.3902,
      "step": 5672
    },
    {
      "epoch": 1.021931997297906,
      "grad_norm": 1.3674273490905762,
      "learning_rate": 4.8343404832895066e-05,
      "loss": 1.3956,
      "step": 5673
    },
    {
      "epoch": 1.0221121369061021,
      "grad_norm": 1.4398524761199951,
      "learning_rate": 4.832925114961629e-05,
      "loss": 1.6261,
      "step": 5674
    },
    {
      "epoch": 1.0222922765142985,
      "grad_norm": 1.490523338317871,
      "learning_rate": 4.8315097600364145e-05,
      "loss": 1.618,
      "step": 5675
    },
    {
      "epoch": 1.022472416122495,
      "grad_norm": 1.2867107391357422,
      "learning_rate": 4.8300944186274076e-05,
      "loss": 1.2315,
      "step": 5676
    },
    {
      "epoch": 1.0226525557306914,
      "grad_norm": 1.539480209350586,
      "learning_rate": 4.828679090848143e-05,
      "loss": 1.5947,
      "step": 5677
    },
    {
      "epoch": 1.0228326953388875,
      "grad_norm": 1.5215333700180054,
      "learning_rate": 4.827263776812159e-05,
      "loss": 1.4624,
      "step": 5678
    },
    {
      "epoch": 1.023012834947084,
      "grad_norm": 1.510625958442688,
      "learning_rate": 4.825848476632992e-05,
      "loss": 1.4863,
      "step": 5679
    },
    {
      "epoch": 1.0231929745552804,
      "grad_norm": 1.3307843208312988,
      "learning_rate": 4.824433190424172e-05,
      "loss": 1.2927,
      "step": 5680
    },
    {
      "epoch": 1.0233731141634768,
      "grad_norm": 1.5096434354782104,
      "learning_rate": 4.8230179182992375e-05,
      "loss": 1.4389,
      "step": 5681
    },
    {
      "epoch": 1.023553253771673,
      "grad_norm": 1.4708826541900635,
      "learning_rate": 4.82160266037172e-05,
      "loss": 1.4199,
      "step": 5682
    },
    {
      "epoch": 1.0237333933798693,
      "grad_norm": 1.5468809604644775,
      "learning_rate": 4.82018741675515e-05,
      "loss": 1.6074,
      "step": 5683
    },
    {
      "epoch": 1.0239135329880658,
      "grad_norm": 1.3394520282745361,
      "learning_rate": 4.8187721875630565e-05,
      "loss": 1.2816,
      "step": 5684
    },
    {
      "epoch": 1.0240936725962622,
      "grad_norm": 1.495758295059204,
      "learning_rate": 4.817356972908971e-05,
      "loss": 1.6142,
      "step": 5685
    },
    {
      "epoch": 1.0242738122044583,
      "grad_norm": 1.5061441659927368,
      "learning_rate": 4.815941772906421e-05,
      "loss": 1.5169,
      "step": 5686
    },
    {
      "epoch": 1.0244539518126548,
      "grad_norm": 1.2849524021148682,
      "learning_rate": 4.814526587668931e-05,
      "loss": 1.3454,
      "step": 5687
    },
    {
      "epoch": 1.0246340914208512,
      "grad_norm": 1.5339057445526123,
      "learning_rate": 4.8131114173100286e-05,
      "loss": 1.3045,
      "step": 5688
    },
    {
      "epoch": 1.0248142310290476,
      "grad_norm": 1.389390230178833,
      "learning_rate": 4.811696261943236e-05,
      "loss": 1.3284,
      "step": 5689
    },
    {
      "epoch": 1.024994370637244,
      "grad_norm": 1.5021467208862305,
      "learning_rate": 4.810281121682079e-05,
      "loss": 1.5879,
      "step": 5690
    },
    {
      "epoch": 1.0251745102454402,
      "grad_norm": 1.5365718603134155,
      "learning_rate": 4.8088659966400774e-05,
      "loss": 1.641,
      "step": 5691
    },
    {
      "epoch": 1.0253546498536366,
      "grad_norm": 1.3402938842773438,
      "learning_rate": 4.8074508869307526e-05,
      "loss": 1.1189,
      "step": 5692
    },
    {
      "epoch": 1.025534789461833,
      "grad_norm": 1.5072957277297974,
      "learning_rate": 4.8060357926676244e-05,
      "loss": 1.7261,
      "step": 5693
    },
    {
      "epoch": 1.0257149290700294,
      "grad_norm": 1.3815417289733887,
      "learning_rate": 4.804620713964208e-05,
      "loss": 1.2545,
      "step": 5694
    },
    {
      "epoch": 1.0258950686782256,
      "grad_norm": 1.4508488178253174,
      "learning_rate": 4.8032056509340254e-05,
      "loss": 1.4157,
      "step": 5695
    },
    {
      "epoch": 1.026075208286422,
      "grad_norm": 1.3612934350967407,
      "learning_rate": 4.801790603690589e-05,
      "loss": 1.2882,
      "step": 5696
    },
    {
      "epoch": 1.0262553478946184,
      "grad_norm": 1.4334169626235962,
      "learning_rate": 4.800375572347414e-05,
      "loss": 1.2776,
      "step": 5697
    },
    {
      "epoch": 1.0264354875028148,
      "grad_norm": 1.533494234085083,
      "learning_rate": 4.7989605570180106e-05,
      "loss": 1.6476,
      "step": 5698
    },
    {
      "epoch": 1.026615627111011,
      "grad_norm": 1.666290283203125,
      "learning_rate": 4.797545557815897e-05,
      "loss": 1.6083,
      "step": 5699
    },
    {
      "epoch": 1.0267957667192074,
      "grad_norm": 1.4549219608306885,
      "learning_rate": 4.7961305748545795e-05,
      "loss": 1.2693,
      "step": 5700
    },
    {
      "epoch": 1.0269759063274038,
      "grad_norm": 1.545725703239441,
      "learning_rate": 4.794715608247567e-05,
      "loss": 1.5388,
      "step": 5701
    },
    {
      "epoch": 1.0271560459356002,
      "grad_norm": 1.2349741458892822,
      "learning_rate": 4.79330065810837e-05,
      "loss": 1.843,
      "step": 5702
    },
    {
      "epoch": 1.0273361855437964,
      "grad_norm": 1.222366452217102,
      "learning_rate": 4.791885724550489e-05,
      "loss": 1.7908,
      "step": 5703
    },
    {
      "epoch": 1.0275163251519928,
      "grad_norm": 1.40467369556427,
      "learning_rate": 4.7904708076874383e-05,
      "loss": 1.9344,
      "step": 5704
    },
    {
      "epoch": 1.0276964647601892,
      "grad_norm": 1.4209883213043213,
      "learning_rate": 4.7890559076327175e-05,
      "loss": 1.8369,
      "step": 5705
    },
    {
      "epoch": 1.0278766043683856,
      "grad_norm": 1.4214222431182861,
      "learning_rate": 4.787641024499828e-05,
      "loss": 1.9501,
      "step": 5706
    },
    {
      "epoch": 1.0280567439765818,
      "grad_norm": 1.3862560987472534,
      "learning_rate": 4.7862261584022716e-05,
      "loss": 1.9316,
      "step": 5707
    },
    {
      "epoch": 1.0282368835847782,
      "grad_norm": 1.3517098426818848,
      "learning_rate": 4.7848113094535466e-05,
      "loss": 1.5606,
      "step": 5708
    },
    {
      "epoch": 1.0284170231929746,
      "grad_norm": 1.4003384113311768,
      "learning_rate": 4.7833964777671565e-05,
      "loss": 1.8409,
      "step": 5709
    },
    {
      "epoch": 1.028597162801171,
      "grad_norm": 1.5580077171325684,
      "learning_rate": 4.781981663456595e-05,
      "loss": 1.897,
      "step": 5710
    },
    {
      "epoch": 1.0287773024093672,
      "grad_norm": 1.5970946550369263,
      "learning_rate": 4.7805668666353584e-05,
      "loss": 1.9684,
      "step": 5711
    },
    {
      "epoch": 1.0289574420175636,
      "grad_norm": 1.526471495628357,
      "learning_rate": 4.7791520874169395e-05,
      "loss": 1.6483,
      "step": 5712
    },
    {
      "epoch": 1.02913758162576,
      "grad_norm": 1.3739126920700073,
      "learning_rate": 4.7777373259148335e-05,
      "loss": 1.7123,
      "step": 5713
    },
    {
      "epoch": 1.0293177212339564,
      "grad_norm": 1.33206045627594,
      "learning_rate": 4.776322582242531e-05,
      "loss": 1.261,
      "step": 5714
    },
    {
      "epoch": 1.0294978608421528,
      "grad_norm": 1.2750029563903809,
      "learning_rate": 4.7749078565135225e-05,
      "loss": 1.4291,
      "step": 5715
    },
    {
      "epoch": 1.029678000450349,
      "grad_norm": 1.3429150581359863,
      "learning_rate": 4.773493148841296e-05,
      "loss": 1.4563,
      "step": 5716
    },
    {
      "epoch": 1.0298581400585454,
      "grad_norm": 1.4499880075454712,
      "learning_rate": 4.772078459339338e-05,
      "loss": 1.5551,
      "step": 5717
    },
    {
      "epoch": 1.0300382796667418,
      "grad_norm": 1.3189891576766968,
      "learning_rate": 4.7706637881211346e-05,
      "loss": 1.2808,
      "step": 5718
    },
    {
      "epoch": 1.0302184192749382,
      "grad_norm": 1.411763072013855,
      "learning_rate": 4.769249135300172e-05,
      "loss": 1.527,
      "step": 5719
    },
    {
      "epoch": 1.0303985588831344,
      "grad_norm": 1.4098269939422607,
      "learning_rate": 4.7678345009899305e-05,
      "loss": 1.5637,
      "step": 5720
    },
    {
      "epoch": 1.0305786984913308,
      "grad_norm": 1.2264735698699951,
      "learning_rate": 4.7664198853038924e-05,
      "loss": 1.3567,
      "step": 5721
    },
    {
      "epoch": 1.0307588380995272,
      "grad_norm": 1.4814525842666626,
      "learning_rate": 4.765005288355536e-05,
      "loss": 1.806,
      "step": 5722
    },
    {
      "epoch": 1.0309389777077236,
      "grad_norm": 1.3225632905960083,
      "learning_rate": 4.763590710258343e-05,
      "loss": 1.3965,
      "step": 5723
    },
    {
      "epoch": 1.0311191173159198,
      "grad_norm": 1.2582075595855713,
      "learning_rate": 4.762176151125787e-05,
      "loss": 1.5195,
      "step": 5724
    },
    {
      "epoch": 1.0312992569241162,
      "grad_norm": 1.3020389080047607,
      "learning_rate": 4.7607616110713446e-05,
      "loss": 1.3946,
      "step": 5725
    },
    {
      "epoch": 1.0314793965323126,
      "grad_norm": 1.3780348300933838,
      "learning_rate": 4.759347090208489e-05,
      "loss": 1.5789,
      "step": 5726
    },
    {
      "epoch": 1.031659536140509,
      "grad_norm": 1.4056990146636963,
      "learning_rate": 4.757932588650693e-05,
      "loss": 1.3888,
      "step": 5727
    },
    {
      "epoch": 1.0318396757487052,
      "grad_norm": 1.4529263973236084,
      "learning_rate": 4.7565181065114267e-05,
      "loss": 1.5747,
      "step": 5728
    },
    {
      "epoch": 1.0320198153569016,
      "grad_norm": 1.5221878290176392,
      "learning_rate": 4.755103643904159e-05,
      "loss": 1.5775,
      "step": 5729
    },
    {
      "epoch": 1.032199954965098,
      "grad_norm": 1.277483582496643,
      "learning_rate": 4.753689200942358e-05,
      "loss": 1.2928,
      "step": 5730
    },
    {
      "epoch": 1.0323800945732944,
      "grad_norm": 1.3328691720962524,
      "learning_rate": 4.752274777739488e-05,
      "loss": 1.2877,
      "step": 5731
    },
    {
      "epoch": 1.0325602341814906,
      "grad_norm": 1.3660414218902588,
      "learning_rate": 4.750860374409017e-05,
      "loss": 1.558,
      "step": 5732
    },
    {
      "epoch": 1.032740373789687,
      "grad_norm": 1.5818179845809937,
      "learning_rate": 4.749445991064404e-05,
      "loss": 1.6719,
      "step": 5733
    },
    {
      "epoch": 1.0329205133978834,
      "grad_norm": 1.4644523859024048,
      "learning_rate": 4.748031627819112e-05,
      "loss": 1.5023,
      "step": 5734
    },
    {
      "epoch": 1.0331006530060798,
      "grad_norm": 1.5040640830993652,
      "learning_rate": 4.746617284786601e-05,
      "loss": 1.7585,
      "step": 5735
    },
    {
      "epoch": 1.033280792614276,
      "grad_norm": 1.3353251218795776,
      "learning_rate": 4.7452029620803246e-05,
      "loss": 1.3647,
      "step": 5736
    },
    {
      "epoch": 1.0334609322224724,
      "grad_norm": 1.4378539323806763,
      "learning_rate": 4.743788659813745e-05,
      "loss": 1.5239,
      "step": 5737
    },
    {
      "epoch": 1.0336410718306688,
      "grad_norm": 1.3207393884658813,
      "learning_rate": 4.7423743781003153e-05,
      "loss": 1.3488,
      "step": 5738
    },
    {
      "epoch": 1.0338212114388652,
      "grad_norm": 1.3925564289093018,
      "learning_rate": 4.740960117053487e-05,
      "loss": 1.4494,
      "step": 5739
    },
    {
      "epoch": 1.0340013510470614,
      "grad_norm": 1.521926999092102,
      "learning_rate": 4.739545876786709e-05,
      "loss": 1.5359,
      "step": 5740
    },
    {
      "epoch": 1.0341814906552578,
      "grad_norm": 1.3042010068893433,
      "learning_rate": 4.7381316574134365e-05,
      "loss": 1.2302,
      "step": 5741
    },
    {
      "epoch": 1.0343616302634542,
      "grad_norm": 1.4988808631896973,
      "learning_rate": 4.7367174590471145e-05,
      "loss": 1.3834,
      "step": 5742
    },
    {
      "epoch": 1.0345417698716506,
      "grad_norm": 1.567260503768921,
      "learning_rate": 4.73530328180119e-05,
      "loss": 1.5681,
      "step": 5743
    },
    {
      "epoch": 1.0347219094798468,
      "grad_norm": 1.4177359342575073,
      "learning_rate": 4.7338891257891084e-05,
      "loss": 1.398,
      "step": 5744
    },
    {
      "epoch": 1.0349020490880432,
      "grad_norm": 1.4603091478347778,
      "learning_rate": 4.7324749911243085e-05,
      "loss": 1.4455,
      "step": 5745
    },
    {
      "epoch": 1.0350821886962396,
      "grad_norm": 1.4425009489059448,
      "learning_rate": 4.731060877920236e-05,
      "loss": 1.2133,
      "step": 5746
    },
    {
      "epoch": 1.035262328304436,
      "grad_norm": 1.3918821811676025,
      "learning_rate": 4.7296467862903295e-05,
      "loss": 1.3172,
      "step": 5747
    },
    {
      "epoch": 1.0354424679126324,
      "grad_norm": 1.505253791809082,
      "learning_rate": 4.728232716348025e-05,
      "loss": 1.5475,
      "step": 5748
    },
    {
      "epoch": 1.0356226075208286,
      "grad_norm": 1.3155560493469238,
      "learning_rate": 4.726818668206761e-05,
      "loss": 1.1824,
      "step": 5749
    },
    {
      "epoch": 1.035802747129025,
      "grad_norm": 1.3571481704711914,
      "learning_rate": 4.725404641979969e-05,
      "loss": 1.3126,
      "step": 5750
    },
    {
      "epoch": 1.0359828867372214,
      "grad_norm": 1.384545922279358,
      "learning_rate": 4.723990637781084e-05,
      "loss": 1.3437,
      "step": 5751
    },
    {
      "epoch": 1.0361630263454178,
      "grad_norm": 1.3022403717041016,
      "learning_rate": 4.7225766557235354e-05,
      "loss": 1.649,
      "step": 5752
    },
    {
      "epoch": 1.036343165953614,
      "grad_norm": 1.3195267915725708,
      "learning_rate": 4.7211626959207524e-05,
      "loss": 1.8459,
      "step": 5753
    },
    {
      "epoch": 1.0365233055618104,
      "grad_norm": 1.3528718948364258,
      "learning_rate": 4.719748758486161e-05,
      "loss": 1.9176,
      "step": 5754
    },
    {
      "epoch": 1.0367034451700068,
      "grad_norm": 1.3536591529846191,
      "learning_rate": 4.718334843533189e-05,
      "loss": 1.9696,
      "step": 5755
    },
    {
      "epoch": 1.0368835847782032,
      "grad_norm": 1.287377953529358,
      "learning_rate": 4.7169209511752574e-05,
      "loss": 1.5935,
      "step": 5756
    },
    {
      "epoch": 1.0370637243863994,
      "grad_norm": 1.296950340270996,
      "learning_rate": 4.7155070815257905e-05,
      "loss": 1.7355,
      "step": 5757
    },
    {
      "epoch": 1.0372438639945958,
      "grad_norm": 1.4993720054626465,
      "learning_rate": 4.714093234698207e-05,
      "loss": 1.9432,
      "step": 5758
    },
    {
      "epoch": 1.0374240036027922,
      "grad_norm": 1.4126299619674683,
      "learning_rate": 4.712679410805923e-05,
      "loss": 1.7028,
      "step": 5759
    },
    {
      "epoch": 1.0376041432109886,
      "grad_norm": 1.6363420486450195,
      "learning_rate": 4.7112656099623584e-05,
      "loss": 1.9907,
      "step": 5760
    },
    {
      "epoch": 1.0377842828191848,
      "grad_norm": 1.7183916568756104,
      "learning_rate": 4.709851832280926e-05,
      "loss": 2.1713,
      "step": 5761
    },
    {
      "epoch": 1.0379644224273812,
      "grad_norm": 1.4752726554870605,
      "learning_rate": 4.708438077875038e-05,
      "loss": 1.5608,
      "step": 5762
    },
    {
      "epoch": 1.0381445620355776,
      "grad_norm": 1.3664577007293701,
      "learning_rate": 4.707024346858105e-05,
      "loss": 1.5523,
      "step": 5763
    },
    {
      "epoch": 1.038324701643774,
      "grad_norm": 1.333034873008728,
      "learning_rate": 4.705610639343536e-05,
      "loss": 1.5955,
      "step": 5764
    },
    {
      "epoch": 1.0385048412519702,
      "grad_norm": 1.3754521608352661,
      "learning_rate": 4.704196955444739e-05,
      "loss": 1.6484,
      "step": 5765
    },
    {
      "epoch": 1.0386849808601666,
      "grad_norm": 1.2431342601776123,
      "learning_rate": 4.7027832952751175e-05,
      "loss": 1.3816,
      "step": 5766
    },
    {
      "epoch": 1.038865120468363,
      "grad_norm": 1.32204008102417,
      "learning_rate": 4.701369658948076e-05,
      "loss": 1.4545,
      "step": 5767
    },
    {
      "epoch": 1.0390452600765594,
      "grad_norm": 1.2806951999664307,
      "learning_rate": 4.699956046577012e-05,
      "loss": 1.2847,
      "step": 5768
    },
    {
      "epoch": 1.0392253996847556,
      "grad_norm": 1.33052659034729,
      "learning_rate": 4.698542458275331e-05,
      "loss": 1.4433,
      "step": 5769
    },
    {
      "epoch": 1.039405539292952,
      "grad_norm": 1.435530662536621,
      "learning_rate": 4.697128894156428e-05,
      "loss": 1.6555,
      "step": 5770
    },
    {
      "epoch": 1.0395856789011484,
      "grad_norm": 1.381945252418518,
      "learning_rate": 4.6957153543336965e-05,
      "loss": 1.5089,
      "step": 5771
    },
    {
      "epoch": 1.0397658185093448,
      "grad_norm": 1.315426230430603,
      "learning_rate": 4.694301838920531e-05,
      "loss": 1.3656,
      "step": 5772
    },
    {
      "epoch": 1.0399459581175412,
      "grad_norm": 1.4077342748641968,
      "learning_rate": 4.6928883480303206e-05,
      "loss": 1.4299,
      "step": 5773
    },
    {
      "epoch": 1.0401260977257374,
      "grad_norm": 1.274634599685669,
      "learning_rate": 4.6914748817764596e-05,
      "loss": 1.2912,
      "step": 5774
    },
    {
      "epoch": 1.0403062373339338,
      "grad_norm": 1.373408555984497,
      "learning_rate": 4.6900614402723334e-05,
      "loss": 1.44,
      "step": 5775
    },
    {
      "epoch": 1.0404863769421302,
      "grad_norm": 1.3952643871307373,
      "learning_rate": 4.688648023631327e-05,
      "loss": 1.4443,
      "step": 5776
    },
    {
      "epoch": 1.0406665165503266,
      "grad_norm": 1.4027163982391357,
      "learning_rate": 4.687234631966826e-05,
      "loss": 1.5012,
      "step": 5777
    },
    {
      "epoch": 1.0408466561585228,
      "grad_norm": 1.400055170059204,
      "learning_rate": 4.685821265392208e-05,
      "loss": 1.489,
      "step": 5778
    },
    {
      "epoch": 1.0410267957667192,
      "grad_norm": 1.3491933345794678,
      "learning_rate": 4.684407924020856e-05,
      "loss": 1.3453,
      "step": 5779
    },
    {
      "epoch": 1.0412069353749156,
      "grad_norm": 1.4953827857971191,
      "learning_rate": 4.682994607966147e-05,
      "loss": 1.6989,
      "step": 5780
    },
    {
      "epoch": 1.041387074983112,
      "grad_norm": 1.3640010356903076,
      "learning_rate": 4.681581317341457e-05,
      "loss": 1.4085,
      "step": 5781
    },
    {
      "epoch": 1.0415672145913082,
      "grad_norm": 1.3998982906341553,
      "learning_rate": 4.680168052260156e-05,
      "loss": 1.4891,
      "step": 5782
    },
    {
      "epoch": 1.0417473541995046,
      "grad_norm": 1.3895014524459839,
      "learning_rate": 4.6787548128356205e-05,
      "loss": 1.4404,
      "step": 5783
    },
    {
      "epoch": 1.041927493807701,
      "grad_norm": 1.475217342376709,
      "learning_rate": 4.677341599181217e-05,
      "loss": 1.6394,
      "step": 5784
    },
    {
      "epoch": 1.0421076334158974,
      "grad_norm": 1.3195574283599854,
      "learning_rate": 4.675928411410313e-05,
      "loss": 1.2674,
      "step": 5785
    },
    {
      "epoch": 1.0422877730240936,
      "grad_norm": 1.4098390340805054,
      "learning_rate": 4.674515249636274e-05,
      "loss": 1.3154,
      "step": 5786
    },
    {
      "epoch": 1.04246791263229,
      "grad_norm": 1.3156152963638306,
      "learning_rate": 4.6731021139724625e-05,
      "loss": 1.3817,
      "step": 5787
    },
    {
      "epoch": 1.0426480522404864,
      "grad_norm": 1.2982183694839478,
      "learning_rate": 4.671689004532241e-05,
      "loss": 1.3423,
      "step": 5788
    },
    {
      "epoch": 1.0428281918486828,
      "grad_norm": 1.4725338220596313,
      "learning_rate": 4.670275921428967e-05,
      "loss": 1.3816,
      "step": 5789
    },
    {
      "epoch": 1.043008331456879,
      "grad_norm": 1.4499729871749878,
      "learning_rate": 4.668862864775999e-05,
      "loss": 1.4162,
      "step": 5790
    },
    {
      "epoch": 1.0431884710650754,
      "grad_norm": 1.3309400081634521,
      "learning_rate": 4.667449834686689e-05,
      "loss": 1.3185,
      "step": 5791
    },
    {
      "epoch": 1.0433686106732718,
      "grad_norm": 1.342346429824829,
      "learning_rate": 4.666036831274392e-05,
      "loss": 1.2237,
      "step": 5792
    },
    {
      "epoch": 1.0435487502814682,
      "grad_norm": 1.3679298162460327,
      "learning_rate": 4.664623854652459e-05,
      "loss": 1.3494,
      "step": 5793
    },
    {
      "epoch": 1.0437288898896644,
      "grad_norm": 1.3507254123687744,
      "learning_rate": 4.663210904934237e-05,
      "loss": 1.3174,
      "step": 5794
    },
    {
      "epoch": 1.0439090294978608,
      "grad_norm": 1.377913475036621,
      "learning_rate": 4.661797982233071e-05,
      "loss": 1.2674,
      "step": 5795
    },
    {
      "epoch": 1.0440891691060572,
      "grad_norm": 1.5450230836868286,
      "learning_rate": 4.6603850866623037e-05,
      "loss": 1.5028,
      "step": 5796
    },
    {
      "epoch": 1.0442693087142536,
      "grad_norm": 1.4675216674804688,
      "learning_rate": 4.658972218335284e-05,
      "loss": 1.4726,
      "step": 5797
    },
    {
      "epoch": 1.0444494483224498,
      "grad_norm": 1.5356272459030151,
      "learning_rate": 4.657559377365345e-05,
      "loss": 1.2549,
      "step": 5798
    },
    {
      "epoch": 1.0446295879306462,
      "grad_norm": 1.4964866638183594,
      "learning_rate": 4.656146563865826e-05,
      "loss": 1.4362,
      "step": 5799
    },
    {
      "epoch": 1.0448097275388426,
      "grad_norm": 1.5085233449935913,
      "learning_rate": 4.654733777950062e-05,
      "loss": 1.4774,
      "step": 5800
    },
    {
      "epoch": 1.044989867147039,
      "grad_norm": 1.521103858947754,
      "learning_rate": 4.653321019731384e-05,
      "loss": 1.4785,
      "step": 5801
    },
    {
      "epoch": 1.0451700067552354,
      "grad_norm": 1.3913700580596924,
      "learning_rate": 4.6519082893231273e-05,
      "loss": 1.8465,
      "step": 5802
    },
    {
      "epoch": 1.0453501463634316,
      "grad_norm": 1.3035120964050293,
      "learning_rate": 4.6504955868386194e-05,
      "loss": 1.6488,
      "step": 5803
    },
    {
      "epoch": 1.045530285971628,
      "grad_norm": 1.3929150104522705,
      "learning_rate": 4.649082912391184e-05,
      "loss": 1.6733,
      "step": 5804
    },
    {
      "epoch": 1.0457104255798244,
      "grad_norm": 1.2299821376800537,
      "learning_rate": 4.647670266094144e-05,
      "loss": 1.5208,
      "step": 5805
    },
    {
      "epoch": 1.0458905651880208,
      "grad_norm": 1.302491545677185,
      "learning_rate": 4.646257648060826e-05,
      "loss": 1.8307,
      "step": 5806
    },
    {
      "epoch": 1.046070704796217,
      "grad_norm": 1.3841044902801514,
      "learning_rate": 4.644845058404547e-05,
      "loss": 1.6925,
      "step": 5807
    },
    {
      "epoch": 1.0462508444044134,
      "grad_norm": 1.4533723592758179,
      "learning_rate": 4.643432497238625e-05,
      "loss": 2.1432,
      "step": 5808
    },
    {
      "epoch": 1.0464309840126098,
      "grad_norm": 1.8999817371368408,
      "learning_rate": 4.6420199646763754e-05,
      "loss": 1.8443,
      "step": 5809
    },
    {
      "epoch": 1.0466111236208062,
      "grad_norm": 1.6552634239196777,
      "learning_rate": 4.6406074608311055e-05,
      "loss": 1.6317,
      "step": 5810
    },
    {
      "epoch": 1.0467912632290024,
      "grad_norm": 1.7162585258483887,
      "learning_rate": 4.639194985816133e-05,
      "loss": 2.0495,
      "step": 5811
    },
    {
      "epoch": 1.0469714028371988,
      "grad_norm": 1.5610268115997314,
      "learning_rate": 4.6377825397447635e-05,
      "loss": 1.8608,
      "step": 5812
    },
    {
      "epoch": 1.0471515424453952,
      "grad_norm": 1.3679322004318237,
      "learning_rate": 4.636370122730302e-05,
      "loss": 1.5618,
      "step": 5813
    },
    {
      "epoch": 1.0473316820535916,
      "grad_norm": 1.4862409830093384,
      "learning_rate": 4.634957734886052e-05,
      "loss": 1.6521,
      "step": 5814
    },
    {
      "epoch": 1.0475118216617878,
      "grad_norm": 1.3192274570465088,
      "learning_rate": 4.633545376325313e-05,
      "loss": 1.4314,
      "step": 5815
    },
    {
      "epoch": 1.0476919612699842,
      "grad_norm": 1.335257887840271,
      "learning_rate": 4.632133047161387e-05,
      "loss": 1.3611,
      "step": 5816
    },
    {
      "epoch": 1.0478721008781806,
      "grad_norm": 1.3837766647338867,
      "learning_rate": 4.630720747507569e-05,
      "loss": 1.4771,
      "step": 5817
    },
    {
      "epoch": 1.048052240486377,
      "grad_norm": 1.3603489398956299,
      "learning_rate": 4.6293084774771525e-05,
      "loss": 1.3864,
      "step": 5818
    },
    {
      "epoch": 1.0482323800945732,
      "grad_norm": 1.3667148351669312,
      "learning_rate": 4.627896237183428e-05,
      "loss": 1.4564,
      "step": 5819
    },
    {
      "epoch": 1.0484125197027696,
      "grad_norm": 1.3413883447647095,
      "learning_rate": 4.626484026739688e-05,
      "loss": 1.4014,
      "step": 5820
    },
    {
      "epoch": 1.048592659310966,
      "grad_norm": 1.3456906080245972,
      "learning_rate": 4.625071846259217e-05,
      "loss": 1.4981,
      "step": 5821
    },
    {
      "epoch": 1.0487727989191624,
      "grad_norm": 1.468139886856079,
      "learning_rate": 4.6236596958553e-05,
      "loss": 1.5266,
      "step": 5822
    },
    {
      "epoch": 1.0489529385273586,
      "grad_norm": 1.2669085264205933,
      "learning_rate": 4.6222475756412184e-05,
      "loss": 1.4241,
      "step": 5823
    },
    {
      "epoch": 1.049133078135555,
      "grad_norm": 1.2660475969314575,
      "learning_rate": 4.6208354857302514e-05,
      "loss": 1.3017,
      "step": 5824
    },
    {
      "epoch": 1.0493132177437514,
      "grad_norm": 1.2904064655303955,
      "learning_rate": 4.6194234262356785e-05,
      "loss": 1.4019,
      "step": 5825
    },
    {
      "epoch": 1.0494933573519478,
      "grad_norm": 1.3185572624206543,
      "learning_rate": 4.618011397270773e-05,
      "loss": 1.5035,
      "step": 5826
    },
    {
      "epoch": 1.049673496960144,
      "grad_norm": 1.4489637613296509,
      "learning_rate": 4.6165993989488065e-05,
      "loss": 1.522,
      "step": 5827
    },
    {
      "epoch": 1.0498536365683404,
      "grad_norm": 1.3474037647247314,
      "learning_rate": 4.61518743138305e-05,
      "loss": 1.3995,
      "step": 5828
    },
    {
      "epoch": 1.0500337761765368,
      "grad_norm": 1.378036618232727,
      "learning_rate": 4.613775494686769e-05,
      "loss": 1.5432,
      "step": 5829
    },
    {
      "epoch": 1.0502139157847332,
      "grad_norm": 1.4725327491760254,
      "learning_rate": 4.612363588973231e-05,
      "loss": 1.1525,
      "step": 5830
    },
    {
      "epoch": 1.0503940553929296,
      "grad_norm": 1.4765253067016602,
      "learning_rate": 4.610951714355697e-05,
      "loss": 1.4696,
      "step": 5831
    },
    {
      "epoch": 1.0505741950011258,
      "grad_norm": 1.4648715257644653,
      "learning_rate": 4.609539870947427e-05,
      "loss": 1.4586,
      "step": 5832
    },
    {
      "epoch": 1.0507543346093222,
      "grad_norm": 1.3569397926330566,
      "learning_rate": 4.608128058861676e-05,
      "loss": 1.3504,
      "step": 5833
    },
    {
      "epoch": 1.0509344742175186,
      "grad_norm": 1.3940021991729736,
      "learning_rate": 4.6067162782117034e-05,
      "loss": 1.35,
      "step": 5834
    },
    {
      "epoch": 1.051114613825715,
      "grad_norm": 1.5747038125991821,
      "learning_rate": 4.605304529110761e-05,
      "loss": 1.6744,
      "step": 5835
    },
    {
      "epoch": 1.0512947534339112,
      "grad_norm": 1.4956122636795044,
      "learning_rate": 4.603892811672095e-05,
      "loss": 1.5604,
      "step": 5836
    },
    {
      "epoch": 1.0514748930421076,
      "grad_norm": 1.3946802616119385,
      "learning_rate": 4.6024811260089554e-05,
      "loss": 1.4269,
      "step": 5837
    },
    {
      "epoch": 1.051655032650304,
      "grad_norm": 1.3431092500686646,
      "learning_rate": 4.601069472234584e-05,
      "loss": 1.3158,
      "step": 5838
    },
    {
      "epoch": 1.0518351722585004,
      "grad_norm": 1.4149354696273804,
      "learning_rate": 4.5996578504622276e-05,
      "loss": 1.3701,
      "step": 5839
    },
    {
      "epoch": 1.0520153118666966,
      "grad_norm": 1.423290729522705,
      "learning_rate": 4.598246260805123e-05,
      "loss": 1.0842,
      "step": 5840
    },
    {
      "epoch": 1.052195451474893,
      "grad_norm": 1.358107089996338,
      "learning_rate": 4.596834703376509e-05,
      "loss": 1.3242,
      "step": 5841
    },
    {
      "epoch": 1.0523755910830894,
      "grad_norm": 1.4556576013565063,
      "learning_rate": 4.595423178289618e-05,
      "loss": 1.3552,
      "step": 5842
    },
    {
      "epoch": 1.0525557306912858,
      "grad_norm": 1.4829086065292358,
      "learning_rate": 4.59401168565768e-05,
      "loss": 1.3519,
      "step": 5843
    },
    {
      "epoch": 1.052735870299482,
      "grad_norm": 1.6029837131500244,
      "learning_rate": 4.5926002255939285e-05,
      "loss": 1.722,
      "step": 5844
    },
    {
      "epoch": 1.0529160099076784,
      "grad_norm": 1.5289759635925293,
      "learning_rate": 4.591188798211589e-05,
      "loss": 1.4243,
      "step": 5845
    },
    {
      "epoch": 1.0530961495158748,
      "grad_norm": 1.2917219400405884,
      "learning_rate": 4.589777403623885e-05,
      "loss": 1.1035,
      "step": 5846
    },
    {
      "epoch": 1.0532762891240712,
      "grad_norm": 1.4382274150848389,
      "learning_rate": 4.588366041944037e-05,
      "loss": 1.3642,
      "step": 5847
    },
    {
      "epoch": 1.0534564287322674,
      "grad_norm": 1.3584870100021362,
      "learning_rate": 4.586954713285265e-05,
      "loss": 1.2684,
      "step": 5848
    },
    {
      "epoch": 1.0536365683404638,
      "grad_norm": 1.4906058311462402,
      "learning_rate": 4.5855434177607855e-05,
      "loss": 1.2704,
      "step": 5849
    },
    {
      "epoch": 1.0538167079486602,
      "grad_norm": 1.67153000831604,
      "learning_rate": 4.5841321554838105e-05,
      "loss": 1.5062,
      "step": 5850
    },
    {
      "epoch": 1.0539968475568566,
      "grad_norm": 1.4223164319992065,
      "learning_rate": 4.582720926567552e-05,
      "loss": 1.2765,
      "step": 5851
    },
    {
      "epoch": 1.0541769871650528,
      "grad_norm": 1.279051423072815,
      "learning_rate": 4.5813097311252165e-05,
      "loss": 1.7388,
      "step": 5852
    },
    {
      "epoch": 1.0543571267732492,
      "grad_norm": 1.3545536994934082,
      "learning_rate": 4.579898569270011e-05,
      "loss": 1.9797,
      "step": 5853
    },
    {
      "epoch": 1.0545372663814456,
      "grad_norm": 1.3983289003372192,
      "learning_rate": 4.5784874411151376e-05,
      "loss": 1.9432,
      "step": 5854
    },
    {
      "epoch": 1.054717405989642,
      "grad_norm": 1.4242984056472778,
      "learning_rate": 4.577076346773797e-05,
      "loss": 1.9216,
      "step": 5855
    },
    {
      "epoch": 1.0548975455978384,
      "grad_norm": 1.3120312690734863,
      "learning_rate": 4.575665286359185e-05,
      "loss": 1.6461,
      "step": 5856
    },
    {
      "epoch": 1.0550776852060346,
      "grad_norm": 1.3882075548171997,
      "learning_rate": 4.574254259984496e-05,
      "loss": 1.8425,
      "step": 5857
    },
    {
      "epoch": 1.055257824814231,
      "grad_norm": 1.3838682174682617,
      "learning_rate": 4.5728432677629246e-05,
      "loss": 1.7617,
      "step": 5858
    },
    {
      "epoch": 1.0554379644224274,
      "grad_norm": 1.5086780786514282,
      "learning_rate": 4.5714323098076576e-05,
      "loss": 1.7995,
      "step": 5859
    },
    {
      "epoch": 1.0556181040306238,
      "grad_norm": 1.463242769241333,
      "learning_rate": 4.570021386231883e-05,
      "loss": 1.9989,
      "step": 5860
    },
    {
      "epoch": 1.05579824363882,
      "grad_norm": 1.5298558473587036,
      "learning_rate": 4.568610497148782e-05,
      "loss": 1.7859,
      "step": 5861
    },
    {
      "epoch": 1.0559783832470164,
      "grad_norm": 1.228330373764038,
      "learning_rate": 4.5671996426715366e-05,
      "loss": 1.4762,
      "step": 5862
    },
    {
      "epoch": 1.0561585228552128,
      "grad_norm": 1.23577880859375,
      "learning_rate": 4.5657888229133264e-05,
      "loss": 1.5135,
      "step": 5863
    },
    {
      "epoch": 1.0563386624634092,
      "grad_norm": 1.3579933643341064,
      "learning_rate": 4.564378037987325e-05,
      "loss": 1.5408,
      "step": 5864
    },
    {
      "epoch": 1.0565188020716054,
      "grad_norm": 1.2992329597473145,
      "learning_rate": 4.562967288006705e-05,
      "loss": 1.5238,
      "step": 5865
    },
    {
      "epoch": 1.0566989416798018,
      "grad_norm": 1.2924782037734985,
      "learning_rate": 4.561556573084634e-05,
      "loss": 1.3862,
      "step": 5866
    },
    {
      "epoch": 1.0568790812879982,
      "grad_norm": 1.387759804725647,
      "learning_rate": 4.560145893334283e-05,
      "loss": 1.4879,
      "step": 5867
    },
    {
      "epoch": 1.0570592208961946,
      "grad_norm": 1.2933061122894287,
      "learning_rate": 4.558735248868816e-05,
      "loss": 1.5111,
      "step": 5868
    },
    {
      "epoch": 1.0572393605043908,
      "grad_norm": 1.3263051509857178,
      "learning_rate": 4.5573246398013916e-05,
      "loss": 1.508,
      "step": 5869
    },
    {
      "epoch": 1.0574195001125872,
      "grad_norm": 1.3353958129882812,
      "learning_rate": 4.555914066245168e-05,
      "loss": 1.4281,
      "step": 5870
    },
    {
      "epoch": 1.0575996397207836,
      "grad_norm": 1.2505006790161133,
      "learning_rate": 4.5545035283133e-05,
      "loss": 1.4233,
      "step": 5871
    },
    {
      "epoch": 1.05777977932898,
      "grad_norm": 1.2882511615753174,
      "learning_rate": 4.553093026118944e-05,
      "loss": 1.3904,
      "step": 5872
    },
    {
      "epoch": 1.0579599189371762,
      "grad_norm": 1.498822569847107,
      "learning_rate": 4.5516825597752485e-05,
      "loss": 1.6502,
      "step": 5873
    },
    {
      "epoch": 1.0581400585453726,
      "grad_norm": 1.428572416305542,
      "learning_rate": 4.55027212939536e-05,
      "loss": 1.5253,
      "step": 5874
    },
    {
      "epoch": 1.058320198153569,
      "grad_norm": 1.2035534381866455,
      "learning_rate": 4.548861735092419e-05,
      "loss": 1.2131,
      "step": 5875
    },
    {
      "epoch": 1.0585003377617654,
      "grad_norm": 1.3449424505233765,
      "learning_rate": 4.547451376979572e-05,
      "loss": 1.4028,
      "step": 5876
    },
    {
      "epoch": 1.0586804773699616,
      "grad_norm": 1.3589001893997192,
      "learning_rate": 4.546041055169956e-05,
      "loss": 1.6379,
      "step": 5877
    },
    {
      "epoch": 1.058860616978158,
      "grad_norm": 1.363996148109436,
      "learning_rate": 4.5446307697767044e-05,
      "loss": 1.3631,
      "step": 5878
    },
    {
      "epoch": 1.0590407565863544,
      "grad_norm": 1.513213872909546,
      "learning_rate": 4.543220520912951e-05,
      "loss": 1.5737,
      "step": 5879
    },
    {
      "epoch": 1.0592208961945508,
      "grad_norm": 1.6133108139038086,
      "learning_rate": 4.541810308691824e-05,
      "loss": 1.752,
      "step": 5880
    },
    {
      "epoch": 1.059401035802747,
      "grad_norm": 1.4693034887313843,
      "learning_rate": 4.540400133226452e-05,
      "loss": 1.4857,
      "step": 5881
    },
    {
      "epoch": 1.0595811754109434,
      "grad_norm": 1.2282518148422241,
      "learning_rate": 4.538989994629957e-05,
      "loss": 1.1928,
      "step": 5882
    },
    {
      "epoch": 1.0597613150191398,
      "grad_norm": 1.4172661304473877,
      "learning_rate": 4.53757989301546e-05,
      "loss": 1.4922,
      "step": 5883
    },
    {
      "epoch": 1.0599414546273362,
      "grad_norm": 1.4527961015701294,
      "learning_rate": 4.5361698284960794e-05,
      "loss": 1.5314,
      "step": 5884
    },
    {
      "epoch": 1.0601215942355324,
      "grad_norm": 1.4484930038452148,
      "learning_rate": 4.534759801184928e-05,
      "loss": 1.4391,
      "step": 5885
    },
    {
      "epoch": 1.0603017338437288,
      "grad_norm": 1.4354983568191528,
      "learning_rate": 4.5333498111951196e-05,
      "loss": 1.386,
      "step": 5886
    },
    {
      "epoch": 1.0604818734519252,
      "grad_norm": 1.336592674255371,
      "learning_rate": 4.5319398586397616e-05,
      "loss": 1.2486,
      "step": 5887
    },
    {
      "epoch": 1.0606620130601216,
      "grad_norm": 1.3638192415237427,
      "learning_rate": 4.530529943631961e-05,
      "loss": 1.3237,
      "step": 5888
    },
    {
      "epoch": 1.060842152668318,
      "grad_norm": 1.3784631490707397,
      "learning_rate": 4.529120066284818e-05,
      "loss": 1.2917,
      "step": 5889
    },
    {
      "epoch": 1.0610222922765142,
      "grad_norm": 1.6139980554580688,
      "learning_rate": 4.527710226711435e-05,
      "loss": 1.7189,
      "step": 5890
    },
    {
      "epoch": 1.0612024318847106,
      "grad_norm": 1.4127531051635742,
      "learning_rate": 4.526300425024908e-05,
      "loss": 1.6182,
      "step": 5891
    },
    {
      "epoch": 1.061382571492907,
      "grad_norm": 1.4597697257995605,
      "learning_rate": 4.52489066133833e-05,
      "loss": 1.2911,
      "step": 5892
    },
    {
      "epoch": 1.0615627111011035,
      "grad_norm": 1.6121646165847778,
      "learning_rate": 4.523480935764792e-05,
      "loss": 1.7029,
      "step": 5893
    },
    {
      "epoch": 1.0617428507092996,
      "grad_norm": 1.3753849267959595,
      "learning_rate": 4.5220712484173796e-05,
      "loss": 1.2042,
      "step": 5894
    },
    {
      "epoch": 1.061922990317496,
      "grad_norm": 1.5212903022766113,
      "learning_rate": 4.52066159940918e-05,
      "loss": 1.6569,
      "step": 5895
    },
    {
      "epoch": 1.0621031299256924,
      "grad_norm": 1.5773612260818481,
      "learning_rate": 4.5192519888532734e-05,
      "loss": 1.4344,
      "step": 5896
    },
    {
      "epoch": 1.0622832695338889,
      "grad_norm": 1.4069286584854126,
      "learning_rate": 4.517842416862739e-05,
      "loss": 1.3726,
      "step": 5897
    },
    {
      "epoch": 1.062463409142085,
      "grad_norm": 1.7123864889144897,
      "learning_rate": 4.5164328835506476e-05,
      "loss": 1.5374,
      "step": 5898
    },
    {
      "epoch": 1.0626435487502814,
      "grad_norm": 1.2970200777053833,
      "learning_rate": 4.5150233890300776e-05,
      "loss": 1.2195,
      "step": 5899
    },
    {
      "epoch": 1.0628236883584778,
      "grad_norm": 1.3803675174713135,
      "learning_rate": 4.513613933414096e-05,
      "loss": 1.2504,
      "step": 5900
    },
    {
      "epoch": 1.0630038279666743,
      "grad_norm": 1.3868794441223145,
      "learning_rate": 4.5122045168157666e-05,
      "loss": 1.2987,
      "step": 5901
    },
    {
      "epoch": 1.0631839675748704,
      "grad_norm": 1.298801302909851,
      "learning_rate": 4.5107951393481534e-05,
      "loss": 2.0606,
      "step": 5902
    },
    {
      "epoch": 1.0633641071830668,
      "grad_norm": 1.3050814867019653,
      "learning_rate": 4.509385801124314e-05,
      "loss": 1.9173,
      "step": 5903
    },
    {
      "epoch": 1.0635442467912632,
      "grad_norm": 1.687298059463501,
      "learning_rate": 4.507976502257308e-05,
      "loss": 1.8127,
      "step": 5904
    },
    {
      "epoch": 1.0637243863994597,
      "grad_norm": 1.3398560285568237,
      "learning_rate": 4.506567242860188e-05,
      "loss": 1.7186,
      "step": 5905
    },
    {
      "epoch": 1.0639045260076558,
      "grad_norm": 1.3587313890457153,
      "learning_rate": 4.5051580230460046e-05,
      "loss": 1.922,
      "step": 5906
    },
    {
      "epoch": 1.0640846656158522,
      "grad_norm": 1.3958826065063477,
      "learning_rate": 4.503748842927802e-05,
      "loss": 1.7471,
      "step": 5907
    },
    {
      "epoch": 1.0642648052240486,
      "grad_norm": 1.5311014652252197,
      "learning_rate": 4.502339702618624e-05,
      "loss": 1.6232,
      "step": 5908
    },
    {
      "epoch": 1.064444944832245,
      "grad_norm": 1.4137617349624634,
      "learning_rate": 4.5009306022315136e-05,
      "loss": 1.7105,
      "step": 5909
    },
    {
      "epoch": 1.0646250844404412,
      "grad_norm": 1.5777653455734253,
      "learning_rate": 4.4995215418795085e-05,
      "loss": 1.7451,
      "step": 5910
    },
    {
      "epoch": 1.0648052240486376,
      "grad_norm": 1.6742364168167114,
      "learning_rate": 4.498112521675641e-05,
      "loss": 1.81,
      "step": 5911
    },
    {
      "epoch": 1.064985363656834,
      "grad_norm": 1.5370639562606812,
      "learning_rate": 4.496703541732942e-05,
      "loss": 1.5559,
      "step": 5912
    },
    {
      "epoch": 1.0651655032650305,
      "grad_norm": 1.3023706674575806,
      "learning_rate": 4.4952946021644396e-05,
      "loss": 1.3688,
      "step": 5913
    },
    {
      "epoch": 1.0653456428732269,
      "grad_norm": 1.5068551301956177,
      "learning_rate": 4.493885703083159e-05,
      "loss": 1.5758,
      "step": 5914
    },
    {
      "epoch": 1.065525782481423,
      "grad_norm": 1.2598767280578613,
      "learning_rate": 4.492476844602122e-05,
      "loss": 1.2191,
      "step": 5915
    },
    {
      "epoch": 1.0657059220896195,
      "grad_norm": 1.4634027481079102,
      "learning_rate": 4.4910680268343435e-05,
      "loss": 1.7271,
      "step": 5916
    },
    {
      "epoch": 1.0658860616978159,
      "grad_norm": 1.409731149673462,
      "learning_rate": 4.48965924989284e-05,
      "loss": 1.5372,
      "step": 5917
    },
    {
      "epoch": 1.0660662013060123,
      "grad_norm": 1.2124196290969849,
      "learning_rate": 4.488250513890623e-05,
      "loss": 1.3647,
      "step": 5918
    },
    {
      "epoch": 1.0662463409142084,
      "grad_norm": 1.2886885404586792,
      "learning_rate": 4.486841818940701e-05,
      "loss": 1.3609,
      "step": 5919
    },
    {
      "epoch": 1.0664264805224049,
      "grad_norm": 1.2771979570388794,
      "learning_rate": 4.485433165156079e-05,
      "loss": 1.2345,
      "step": 5920
    },
    {
      "epoch": 1.0666066201306013,
      "grad_norm": 1.3354061841964722,
      "learning_rate": 4.484024552649757e-05,
      "loss": 1.5148,
      "step": 5921
    },
    {
      "epoch": 1.0667867597387977,
      "grad_norm": 1.3562688827514648,
      "learning_rate": 4.482615981534734e-05,
      "loss": 1.2987,
      "step": 5922
    },
    {
      "epoch": 1.0669668993469938,
      "grad_norm": 1.5325772762298584,
      "learning_rate": 4.4812074519240046e-05,
      "loss": 1.5195,
      "step": 5923
    },
    {
      "epoch": 1.0671470389551903,
      "grad_norm": 1.3717572689056396,
      "learning_rate": 4.4797989639305615e-05,
      "loss": 1.3244,
      "step": 5924
    },
    {
      "epoch": 1.0673271785633867,
      "grad_norm": 1.3823553323745728,
      "learning_rate": 4.4783905176673926e-05,
      "loss": 1.3844,
      "step": 5925
    },
    {
      "epoch": 1.067507318171583,
      "grad_norm": 1.352996587753296,
      "learning_rate": 4.4769821132474805e-05,
      "loss": 1.5602,
      "step": 5926
    },
    {
      "epoch": 1.0676874577797792,
      "grad_norm": 1.401050329208374,
      "learning_rate": 4.47557375078381e-05,
      "loss": 1.4517,
      "step": 5927
    },
    {
      "epoch": 1.0678675973879757,
      "grad_norm": 1.4393761157989502,
      "learning_rate": 4.474165430389357e-05,
      "loss": 1.3229,
      "step": 5928
    },
    {
      "epoch": 1.068047736996172,
      "grad_norm": 1.4201732873916626,
      "learning_rate": 4.472757152177098e-05,
      "loss": 1.3306,
      "step": 5929
    },
    {
      "epoch": 1.0682278766043685,
      "grad_norm": 1.4842742681503296,
      "learning_rate": 4.471348916260003e-05,
      "loss": 1.484,
      "step": 5930
    },
    {
      "epoch": 1.0684080162125646,
      "grad_norm": 1.4290024042129517,
      "learning_rate": 4.469940722751038e-05,
      "loss": 1.6884,
      "step": 5931
    },
    {
      "epoch": 1.068588155820761,
      "grad_norm": 1.4492005109786987,
      "learning_rate": 4.4685325717631736e-05,
      "loss": 1.547,
      "step": 5932
    },
    {
      "epoch": 1.0687682954289575,
      "grad_norm": 1.4960498809814453,
      "learning_rate": 4.467124463409367e-05,
      "loss": 1.6342,
      "step": 5933
    },
    {
      "epoch": 1.0689484350371539,
      "grad_norm": 1.4255046844482422,
      "learning_rate": 4.4657163978025754e-05,
      "loss": 1.4318,
      "step": 5934
    },
    {
      "epoch": 1.06912857464535,
      "grad_norm": 1.4704073667526245,
      "learning_rate": 4.464308375055755e-05,
      "loss": 1.4211,
      "step": 5935
    },
    {
      "epoch": 1.0693087142535465,
      "grad_norm": 1.43056321144104,
      "learning_rate": 4.4629003952818523e-05,
      "loss": 1.3927,
      "step": 5936
    },
    {
      "epoch": 1.0694888538617429,
      "grad_norm": 1.4655250310897827,
      "learning_rate": 4.461492458593821e-05,
      "loss": 1.5348,
      "step": 5937
    },
    {
      "epoch": 1.0696689934699393,
      "grad_norm": 1.4300577640533447,
      "learning_rate": 4.460084565104603e-05,
      "loss": 1.5673,
      "step": 5938
    },
    {
      "epoch": 1.0698491330781357,
      "grad_norm": 1.3906139135360718,
      "learning_rate": 4.458676714927139e-05,
      "loss": 1.2609,
      "step": 5939
    },
    {
      "epoch": 1.0700292726863319,
      "grad_norm": 1.521336555480957,
      "learning_rate": 4.457268908174362e-05,
      "loss": 1.6969,
      "step": 5940
    },
    {
      "epoch": 1.0702094122945283,
      "grad_norm": 1.5022079944610596,
      "learning_rate": 4.4558611449592106e-05,
      "loss": 1.2934,
      "step": 5941
    },
    {
      "epoch": 1.0703895519027247,
      "grad_norm": 1.4440133571624756,
      "learning_rate": 4.4544534253946136e-05,
      "loss": 1.4334,
      "step": 5942
    },
    {
      "epoch": 1.0705696915109209,
      "grad_norm": 1.4926468133926392,
      "learning_rate": 4.453045749593498e-05,
      "loss": 1.5811,
      "step": 5943
    },
    {
      "epoch": 1.0707498311191173,
      "grad_norm": 1.5202566385269165,
      "learning_rate": 4.4516381176687835e-05,
      "loss": 1.4628,
      "step": 5944
    },
    {
      "epoch": 1.0709299707273137,
      "grad_norm": 1.6362404823303223,
      "learning_rate": 4.450230529733393e-05,
      "loss": 1.7735,
      "step": 5945
    },
    {
      "epoch": 1.07111011033551,
      "grad_norm": 1.5577993392944336,
      "learning_rate": 4.448822985900242e-05,
      "loss": 1.6431,
      "step": 5946
    },
    {
      "epoch": 1.0712902499437065,
      "grad_norm": 1.553818702697754,
      "learning_rate": 4.447415486282242e-05,
      "loss": 1.46,
      "step": 5947
    },
    {
      "epoch": 1.0714703895519027,
      "grad_norm": 1.4666478633880615,
      "learning_rate": 4.446008030992304e-05,
      "loss": 1.4502,
      "step": 5948
    },
    {
      "epoch": 1.071650529160099,
      "grad_norm": 1.5311304330825806,
      "learning_rate": 4.444600620143331e-05,
      "loss": 1.2108,
      "step": 5949
    },
    {
      "epoch": 1.0718306687682955,
      "grad_norm": 1.4128271341323853,
      "learning_rate": 4.4431932538482255e-05,
      "loss": 1.2548,
      "step": 5950
    },
    {
      "epoch": 1.0720108083764919,
      "grad_norm": 1.2450443506240845,
      "learning_rate": 4.441785932219886e-05,
      "loss": 0.9277,
      "step": 5951
    },
    {
      "epoch": 1.072190947984688,
      "grad_norm": 1.3749061822891235,
      "learning_rate": 4.440378655371208e-05,
      "loss": 2.1365,
      "step": 5952
    },
    {
      "epoch": 1.0723710875928845,
      "grad_norm": 1.2680636644363403,
      "learning_rate": 4.438971423415081e-05,
      "loss": 1.8767,
      "step": 5953
    },
    {
      "epoch": 1.0725512272010809,
      "grad_norm": 1.4089096784591675,
      "learning_rate": 4.437564236464392e-05,
      "loss": 1.8431,
      "step": 5954
    },
    {
      "epoch": 1.0727313668092773,
      "grad_norm": 1.3442976474761963,
      "learning_rate": 4.4361570946320276e-05,
      "loss": 2.0213,
      "step": 5955
    },
    {
      "epoch": 1.0729115064174735,
      "grad_norm": 1.2464367151260376,
      "learning_rate": 4.434749998030866e-05,
      "loss": 1.5912,
      "step": 5956
    },
    {
      "epoch": 1.0730916460256699,
      "grad_norm": 1.347922921180725,
      "learning_rate": 4.433342946773784e-05,
      "loss": 1.7482,
      "step": 5957
    },
    {
      "epoch": 1.0732717856338663,
      "grad_norm": 1.4396653175354004,
      "learning_rate": 4.431935940973656e-05,
      "loss": 1.7131,
      "step": 5958
    },
    {
      "epoch": 1.0734519252420627,
      "grad_norm": 1.484238862991333,
      "learning_rate": 4.430528980743347e-05,
      "loss": 1.9303,
      "step": 5959
    },
    {
      "epoch": 1.0736320648502589,
      "grad_norm": 1.5621731281280518,
      "learning_rate": 4.429122066195729e-05,
      "loss": 1.8973,
      "step": 5960
    },
    {
      "epoch": 1.0738122044584553,
      "grad_norm": 1.7975538969039917,
      "learning_rate": 4.42771519744366e-05,
      "loss": 1.9678,
      "step": 5961
    },
    {
      "epoch": 1.0739923440666517,
      "grad_norm": 1.3895522356033325,
      "learning_rate": 4.426308374599999e-05,
      "loss": 1.7068,
      "step": 5962
    },
    {
      "epoch": 1.074172483674848,
      "grad_norm": 1.3360257148742676,
      "learning_rate": 4.4249015977776e-05,
      "loss": 1.4722,
      "step": 5963
    },
    {
      "epoch": 1.0743526232830443,
      "grad_norm": 1.3033549785614014,
      "learning_rate": 4.4234948670893134e-05,
      "loss": 1.2835,
      "step": 5964
    },
    {
      "epoch": 1.0745327628912407,
      "grad_norm": 1.2391910552978516,
      "learning_rate": 4.4220881826479906e-05,
      "loss": 1.133,
      "step": 5965
    },
    {
      "epoch": 1.074712902499437,
      "grad_norm": 1.3447905778884888,
      "learning_rate": 4.420681544566472e-05,
      "loss": 1.4391,
      "step": 5966
    },
    {
      "epoch": 1.0748930421076335,
      "grad_norm": 1.3464741706848145,
      "learning_rate": 4.419274952957597e-05,
      "loss": 1.4672,
      "step": 5967
    },
    {
      "epoch": 1.0750731817158297,
      "grad_norm": 1.3003981113433838,
      "learning_rate": 4.417868407934201e-05,
      "loss": 1.3032,
      "step": 5968
    },
    {
      "epoch": 1.075253321324026,
      "grad_norm": 1.3012442588806152,
      "learning_rate": 4.416461909609119e-05,
      "loss": 1.3675,
      "step": 5969
    },
    {
      "epoch": 1.0754334609322225,
      "grad_norm": 1.449258804321289,
      "learning_rate": 4.415055458095179e-05,
      "loss": 1.6051,
      "step": 5970
    },
    {
      "epoch": 1.0756136005404189,
      "grad_norm": 1.3937456607818604,
      "learning_rate": 4.4136490535052064e-05,
      "loss": 1.5427,
      "step": 5971
    },
    {
      "epoch": 1.0757937401486153,
      "grad_norm": 1.3208332061767578,
      "learning_rate": 4.412242695952019e-05,
      "loss": 1.4454,
      "step": 5972
    },
    {
      "epoch": 1.0759738797568115,
      "grad_norm": 1.5141353607177734,
      "learning_rate": 4.4108363855484356e-05,
      "loss": 1.6584,
      "step": 5973
    },
    {
      "epoch": 1.0761540193650079,
      "grad_norm": 1.4284627437591553,
      "learning_rate": 4.409430122407271e-05,
      "loss": 1.5334,
      "step": 5974
    },
    {
      "epoch": 1.0763341589732043,
      "grad_norm": 1.6092393398284912,
      "learning_rate": 4.408023906641335e-05,
      "loss": 1.6227,
      "step": 5975
    },
    {
      "epoch": 1.0765142985814007,
      "grad_norm": 1.332499623298645,
      "learning_rate": 4.406617738363433e-05,
      "loss": 1.4038,
      "step": 5976
    },
    {
      "epoch": 1.0766944381895969,
      "grad_norm": 1.417228102684021,
      "learning_rate": 4.4052116176863686e-05,
      "loss": 1.4458,
      "step": 5977
    },
    {
      "epoch": 1.0768745777977933,
      "grad_norm": 1.5306553840637207,
      "learning_rate": 4.403805544722935e-05,
      "loss": 1.5597,
      "step": 5978
    },
    {
      "epoch": 1.0770547174059897,
      "grad_norm": 1.4487805366516113,
      "learning_rate": 4.402399519585932e-05,
      "loss": 1.6045,
      "step": 5979
    },
    {
      "epoch": 1.077234857014186,
      "grad_norm": 1.3833152055740356,
      "learning_rate": 4.400993542388148e-05,
      "loss": 1.4061,
      "step": 5980
    },
    {
      "epoch": 1.0774149966223823,
      "grad_norm": 1.2953568696975708,
      "learning_rate": 4.399587613242371e-05,
      "loss": 1.3023,
      "step": 5981
    },
    {
      "epoch": 1.0775951362305787,
      "grad_norm": 1.4946660995483398,
      "learning_rate": 4.398181732261382e-05,
      "loss": 1.4403,
      "step": 5982
    },
    {
      "epoch": 1.077775275838775,
      "grad_norm": 1.4172781705856323,
      "learning_rate": 4.3967758995579624e-05,
      "loss": 1.4389,
      "step": 5983
    },
    {
      "epoch": 1.0779554154469715,
      "grad_norm": 1.2224206924438477,
      "learning_rate": 4.395370115244886e-05,
      "loss": 1.2448,
      "step": 5984
    },
    {
      "epoch": 1.0781355550551677,
      "grad_norm": 1.4250576496124268,
      "learning_rate": 4.393964379434925e-05,
      "loss": 1.4833,
      "step": 5985
    },
    {
      "epoch": 1.078315694663364,
      "grad_norm": 1.4825717210769653,
      "learning_rate": 4.392558692240846e-05,
      "loss": 1.5569,
      "step": 5986
    },
    {
      "epoch": 1.0784958342715605,
      "grad_norm": 1.4613025188446045,
      "learning_rate": 4.3911530537754117e-05,
      "loss": 1.5337,
      "step": 5987
    },
    {
      "epoch": 1.0786759738797569,
      "grad_norm": 1.5007274150848389,
      "learning_rate": 4.389747464151384e-05,
      "loss": 1.498,
      "step": 5988
    },
    {
      "epoch": 1.078856113487953,
      "grad_norm": 1.3157531023025513,
      "learning_rate": 4.388341923481518e-05,
      "loss": 1.2244,
      "step": 5989
    },
    {
      "epoch": 1.0790362530961495,
      "grad_norm": 1.4924142360687256,
      "learning_rate": 4.386936431878564e-05,
      "loss": 1.5304,
      "step": 5990
    },
    {
      "epoch": 1.0792163927043459,
      "grad_norm": 1.5481082201004028,
      "learning_rate": 4.3855309894552707e-05,
      "loss": 1.5226,
      "step": 5991
    },
    {
      "epoch": 1.0793965323125423,
      "grad_norm": 1.411333441734314,
      "learning_rate": 4.384125596324382e-05,
      "loss": 1.477,
      "step": 5992
    },
    {
      "epoch": 1.0795766719207385,
      "grad_norm": 1.5101584196090698,
      "learning_rate": 4.382720252598639e-05,
      "loss": 1.5143,
      "step": 5993
    },
    {
      "epoch": 1.0797568115289349,
      "grad_norm": 1.5340688228607178,
      "learning_rate": 4.381314958390776e-05,
      "loss": 1.6914,
      "step": 5994
    },
    {
      "epoch": 1.0799369511371313,
      "grad_norm": 1.4889111518859863,
      "learning_rate": 4.379909713813526e-05,
      "loss": 1.2961,
      "step": 5995
    },
    {
      "epoch": 1.0801170907453277,
      "grad_norm": 1.476090669631958,
      "learning_rate": 4.378504518979615e-05,
      "loss": 1.5113,
      "step": 5996
    },
    {
      "epoch": 1.080297230353524,
      "grad_norm": 1.4502959251403809,
      "learning_rate": 4.377099374001771e-05,
      "loss": 1.3697,
      "step": 5997
    },
    {
      "epoch": 1.0804773699617203,
      "grad_norm": 1.3926317691802979,
      "learning_rate": 4.375694278992712e-05,
      "loss": 1.3198,
      "step": 5998
    },
    {
      "epoch": 1.0806575095699167,
      "grad_norm": 1.4504799842834473,
      "learning_rate": 4.3742892340651534e-05,
      "loss": 1.4005,
      "step": 5999
    },
    {
      "epoch": 1.080837649178113,
      "grad_norm": 1.4456108808517456,
      "learning_rate": 4.372884239331807e-05,
      "loss": 1.3226,
      "step": 6000
    },
    {
      "epoch": 1.0810177887863093,
      "grad_norm": 1.344207525253296,
      "learning_rate": 4.37147929490538e-05,
      "loss": 1.2448,
      "step": 6001
    },
    {
      "epoch": 1.0811979283945057,
      "grad_norm": 1.4364771842956543,
      "learning_rate": 4.370074400898579e-05,
      "loss": 1.7137,
      "step": 6002
    },
    {
      "epoch": 1.081378068002702,
      "grad_norm": 1.269330382347107,
      "learning_rate": 4.368669557424105e-05,
      "loss": 1.8557,
      "step": 6003
    },
    {
      "epoch": 1.0815582076108985,
      "grad_norm": 1.5894314050674438,
      "learning_rate": 4.36726476459465e-05,
      "loss": 1.9153,
      "step": 6004
    },
    {
      "epoch": 1.081738347219095,
      "grad_norm": 1.3860946893692017,
      "learning_rate": 4.365860022522905e-05,
      "loss": 1.6729,
      "step": 6005
    },
    {
      "epoch": 1.081918486827291,
      "grad_norm": 1.4932881593704224,
      "learning_rate": 4.364455331321563e-05,
      "loss": 2.2214,
      "step": 6006
    },
    {
      "epoch": 1.0820986264354875,
      "grad_norm": 1.2864151000976562,
      "learning_rate": 4.363050691103304e-05,
      "loss": 1.5079,
      "step": 6007
    },
    {
      "epoch": 1.082278766043684,
      "grad_norm": 1.3679653406143188,
      "learning_rate": 4.3616461019808094e-05,
      "loss": 1.7773,
      "step": 6008
    },
    {
      "epoch": 1.0824589056518803,
      "grad_norm": 1.3484203815460205,
      "learning_rate": 4.360241564066753e-05,
      "loss": 1.6213,
      "step": 6009
    },
    {
      "epoch": 1.0826390452600765,
      "grad_norm": 1.747590184211731,
      "learning_rate": 4.358837077473805e-05,
      "loss": 1.9143,
      "step": 6010
    },
    {
      "epoch": 1.0828191848682729,
      "grad_norm": 1.7672019004821777,
      "learning_rate": 4.357432642314636e-05,
      "loss": 2.0902,
      "step": 6011
    },
    {
      "epoch": 1.0829993244764693,
      "grad_norm": 1.4902225732803345,
      "learning_rate": 4.3560282587019075e-05,
      "loss": 1.6244,
      "step": 6012
    },
    {
      "epoch": 1.0831794640846657,
      "grad_norm": 1.4241995811462402,
      "learning_rate": 4.354623926748278e-05,
      "loss": 1.5463,
      "step": 6013
    },
    {
      "epoch": 1.0833596036928619,
      "grad_norm": 1.3042106628417969,
      "learning_rate": 4.3532196465664025e-05,
      "loss": 1.402,
      "step": 6014
    },
    {
      "epoch": 1.0835397433010583,
      "grad_norm": 1.3135544061660767,
      "learning_rate": 4.351815418268932e-05,
      "loss": 1.3167,
      "step": 6015
    },
    {
      "epoch": 1.0837198829092547,
      "grad_norm": 1.4225006103515625,
      "learning_rate": 4.350411241968513e-05,
      "loss": 1.3696,
      "step": 6016
    },
    {
      "epoch": 1.083900022517451,
      "grad_norm": 1.2467031478881836,
      "learning_rate": 4.3490071177777883e-05,
      "loss": 1.4926,
      "step": 6017
    },
    {
      "epoch": 1.0840801621256473,
      "grad_norm": 1.2875277996063232,
      "learning_rate": 4.3476030458093955e-05,
      "loss": 1.3543,
      "step": 6018
    },
    {
      "epoch": 1.0842603017338437,
      "grad_norm": 1.3541805744171143,
      "learning_rate": 4.3461990261759673e-05,
      "loss": 1.3853,
      "step": 6019
    },
    {
      "epoch": 1.08444044134204,
      "grad_norm": 1.2741343975067139,
      "learning_rate": 4.3447950589901356e-05,
      "loss": 1.4867,
      "step": 6020
    },
    {
      "epoch": 1.0846205809502365,
      "grad_norm": 1.5485223531723022,
      "learning_rate": 4.343391144364525e-05,
      "loss": 1.6307,
      "step": 6021
    },
    {
      "epoch": 1.0848007205584327,
      "grad_norm": 1.3859827518463135,
      "learning_rate": 4.341987282411758e-05,
      "loss": 1.4764,
      "step": 6022
    },
    {
      "epoch": 1.084980860166629,
      "grad_norm": 1.3100894689559937,
      "learning_rate": 4.34058347324445e-05,
      "loss": 1.3163,
      "step": 6023
    },
    {
      "epoch": 1.0851609997748255,
      "grad_norm": 1.332659363746643,
      "learning_rate": 4.339179716975214e-05,
      "loss": 1.4089,
      "step": 6024
    },
    {
      "epoch": 1.085341139383022,
      "grad_norm": 1.4047330617904663,
      "learning_rate": 4.33777601371666e-05,
      "loss": 1.2615,
      "step": 6025
    },
    {
      "epoch": 1.085521278991218,
      "grad_norm": 1.4291481971740723,
      "learning_rate": 4.336372363581391e-05,
      "loss": 1.4737,
      "step": 6026
    },
    {
      "epoch": 1.0857014185994145,
      "grad_norm": 1.3553016185760498,
      "learning_rate": 4.334968766682008e-05,
      "loss": 1.5754,
      "step": 6027
    },
    {
      "epoch": 1.085881558207611,
      "grad_norm": 1.5707887411117554,
      "learning_rate": 4.333565223131107e-05,
      "loss": 1.7987,
      "step": 6028
    },
    {
      "epoch": 1.0860616978158073,
      "grad_norm": 1.5148882865905762,
      "learning_rate": 4.332161733041277e-05,
      "loss": 1.6844,
      "step": 6029
    },
    {
      "epoch": 1.0862418374240037,
      "grad_norm": 1.5114638805389404,
      "learning_rate": 4.330758296525111e-05,
      "loss": 1.5731,
      "step": 6030
    },
    {
      "epoch": 1.0864219770322,
      "grad_norm": 1.44391667842865,
      "learning_rate": 4.329354913695187e-05,
      "loss": 1.4934,
      "step": 6031
    },
    {
      "epoch": 1.0866021166403963,
      "grad_norm": 1.3163456916809082,
      "learning_rate": 4.327951584664085e-05,
      "loss": 1.3704,
      "step": 6032
    },
    {
      "epoch": 1.0867822562485927,
      "grad_norm": 1.4682332277297974,
      "learning_rate": 4.3265483095443775e-05,
      "loss": 1.4003,
      "step": 6033
    },
    {
      "epoch": 1.086962395856789,
      "grad_norm": 1.3556357622146606,
      "learning_rate": 4.325145088448639e-05,
      "loss": 1.2178,
      "step": 6034
    },
    {
      "epoch": 1.0871425354649853,
      "grad_norm": 1.542004108428955,
      "learning_rate": 4.3237419214894325e-05,
      "loss": 1.507,
      "step": 6035
    },
    {
      "epoch": 1.0873226750731817,
      "grad_norm": 1.406859278678894,
      "learning_rate": 4.322338808779321e-05,
      "loss": 1.4783,
      "step": 6036
    },
    {
      "epoch": 1.087502814681378,
      "grad_norm": 1.5269654989242554,
      "learning_rate": 4.320935750430859e-05,
      "loss": 1.5656,
      "step": 6037
    },
    {
      "epoch": 1.0876829542895745,
      "grad_norm": 1.5209910869598389,
      "learning_rate": 4.3195327465565984e-05,
      "loss": 1.4596,
      "step": 6038
    },
    {
      "epoch": 1.0878630938977707,
      "grad_norm": 1.37599515914917,
      "learning_rate": 4.3181297972690906e-05,
      "loss": 1.2902,
      "step": 6039
    },
    {
      "epoch": 1.088043233505967,
      "grad_norm": 1.4066253900527954,
      "learning_rate": 4.3167269026808794e-05,
      "loss": 1.496,
      "step": 6040
    },
    {
      "epoch": 1.0882233731141635,
      "grad_norm": 1.4396476745605469,
      "learning_rate": 4.315324062904503e-05,
      "loss": 1.3581,
      "step": 6041
    },
    {
      "epoch": 1.08840351272236,
      "grad_norm": 1.4745769500732422,
      "learning_rate": 4.313921278052498e-05,
      "loss": 1.4068,
      "step": 6042
    },
    {
      "epoch": 1.088583652330556,
      "grad_norm": 1.3182926177978516,
      "learning_rate": 4.3125185482373904e-05,
      "loss": 1.2551,
      "step": 6043
    },
    {
      "epoch": 1.0887637919387525,
      "grad_norm": 1.494860053062439,
      "learning_rate": 4.311115873571712e-05,
      "loss": 1.5242,
      "step": 6044
    },
    {
      "epoch": 1.088943931546949,
      "grad_norm": 1.4621728658676147,
      "learning_rate": 4.309713254167983e-05,
      "loss": 1.3596,
      "step": 6045
    },
    {
      "epoch": 1.0891240711551453,
      "grad_norm": 1.654954195022583,
      "learning_rate": 4.3083106901387205e-05,
      "loss": 1.6727,
      "step": 6046
    },
    {
      "epoch": 1.0893042107633415,
      "grad_norm": 1.3115841150283813,
      "learning_rate": 4.306908181596436e-05,
      "loss": 1.1973,
      "step": 6047
    },
    {
      "epoch": 1.089484350371538,
      "grad_norm": 1.5648349523544312,
      "learning_rate": 4.305505728653641e-05,
      "loss": 1.6711,
      "step": 6048
    },
    {
      "epoch": 1.0896644899797343,
      "grad_norm": 1.4789600372314453,
      "learning_rate": 4.304103331422839e-05,
      "loss": 1.4837,
      "step": 6049
    },
    {
      "epoch": 1.0898446295879307,
      "grad_norm": 1.466086983680725,
      "learning_rate": 4.302700990016527e-05,
      "loss": 1.399,
      "step": 6050
    },
    {
      "epoch": 1.090024769196127,
      "grad_norm": 1.5863369703292847,
      "learning_rate": 4.301298704547203e-05,
      "loss": 1.4064,
      "step": 6051
    },
    {
      "epoch": 1.0902049088043233,
      "grad_norm": 1.335128664970398,
      "learning_rate": 4.2998964751273556e-05,
      "loss": 1.9362,
      "step": 6052
    },
    {
      "epoch": 1.0903850484125197,
      "grad_norm": 1.3851066827774048,
      "learning_rate": 4.298494301869473e-05,
      "loss": 1.9888,
      "step": 6053
    },
    {
      "epoch": 1.0905651880207161,
      "grad_norm": 1.3395419120788574,
      "learning_rate": 4.297092184886036e-05,
      "loss": 1.993,
      "step": 6054
    },
    {
      "epoch": 1.0907453276289125,
      "grad_norm": 1.314893126487732,
      "learning_rate": 4.295690124289521e-05,
      "loss": 1.8678,
      "step": 6055
    },
    {
      "epoch": 1.0909254672371087,
      "grad_norm": 1.322532296180725,
      "learning_rate": 4.2942881201924015e-05,
      "loss": 2.1428,
      "step": 6056
    },
    {
      "epoch": 1.091105606845305,
      "grad_norm": 1.3147462606430054,
      "learning_rate": 4.292886172707144e-05,
      "loss": 1.4492,
      "step": 6057
    },
    {
      "epoch": 1.0912857464535015,
      "grad_norm": 1.4396051168441772,
      "learning_rate": 4.291484281946214e-05,
      "loss": 1.7382,
      "step": 6058
    },
    {
      "epoch": 1.0914658860616977,
      "grad_norm": 1.4112471342086792,
      "learning_rate": 4.290082448022071e-05,
      "loss": 1.7363,
      "step": 6059
    },
    {
      "epoch": 1.091646025669894,
      "grad_norm": 1.578538179397583,
      "learning_rate": 4.2886806710471674e-05,
      "loss": 1.8698,
      "step": 6060
    },
    {
      "epoch": 1.0918261652780905,
      "grad_norm": 1.9945176839828491,
      "learning_rate": 4.2872789511339517e-05,
      "loss": 2.2201,
      "step": 6061
    },
    {
      "epoch": 1.092006304886287,
      "grad_norm": 1.579054355621338,
      "learning_rate": 4.2858772883948744e-05,
      "loss": 1.6256,
      "step": 6062
    },
    {
      "epoch": 1.0921864444944833,
      "grad_norm": 1.2972469329833984,
      "learning_rate": 4.284475682942373e-05,
      "loss": 1.3623,
      "step": 6063
    },
    {
      "epoch": 1.0923665841026795,
      "grad_norm": 1.48723566532135,
      "learning_rate": 4.2830741348888834e-05,
      "loss": 1.7508,
      "step": 6064
    },
    {
      "epoch": 1.092546723710876,
      "grad_norm": 1.3057808876037598,
      "learning_rate": 4.2816726443468366e-05,
      "loss": 1.3124,
      "step": 6065
    },
    {
      "epoch": 1.0927268633190723,
      "grad_norm": 1.4572094678878784,
      "learning_rate": 4.2802712114286593e-05,
      "loss": 1.4669,
      "step": 6066
    },
    {
      "epoch": 1.0929070029272687,
      "grad_norm": 1.3565524816513062,
      "learning_rate": 4.278869836246777e-05,
      "loss": 1.4857,
      "step": 6067
    },
    {
      "epoch": 1.093087142535465,
      "grad_norm": 1.3721365928649902,
      "learning_rate": 4.2774685189136055e-05,
      "loss": 1.2895,
      "step": 6068
    },
    {
      "epoch": 1.0932672821436613,
      "grad_norm": 1.3331338167190552,
      "learning_rate": 4.276067259541557e-05,
      "loss": 1.2973,
      "step": 6069
    },
    {
      "epoch": 1.0934474217518577,
      "grad_norm": 1.4717903137207031,
      "learning_rate": 4.2746660582430404e-05,
      "loss": 1.522,
      "step": 6070
    },
    {
      "epoch": 1.0936275613600541,
      "grad_norm": 1.4906092882156372,
      "learning_rate": 4.273264915130457e-05,
      "loss": 1.6026,
      "step": 6071
    },
    {
      "epoch": 1.0938077009682503,
      "grad_norm": 1.4196577072143555,
      "learning_rate": 4.27186383031621e-05,
      "loss": 1.4763,
      "step": 6072
    },
    {
      "epoch": 1.0939878405764467,
      "grad_norm": 1.445973515510559,
      "learning_rate": 4.270462803912692e-05,
      "loss": 1.5939,
      "step": 6073
    },
    {
      "epoch": 1.0941679801846431,
      "grad_norm": 1.4378174543380737,
      "learning_rate": 4.269061836032293e-05,
      "loss": 1.5654,
      "step": 6074
    },
    {
      "epoch": 1.0943481197928395,
      "grad_norm": 1.4773672819137573,
      "learning_rate": 4.2676609267873955e-05,
      "loss": 1.7185,
      "step": 6075
    },
    {
      "epoch": 1.0945282594010357,
      "grad_norm": 1.503448724746704,
      "learning_rate": 4.266260076290384e-05,
      "loss": 1.491,
      "step": 6076
    },
    {
      "epoch": 1.0947083990092321,
      "grad_norm": 1.4960205554962158,
      "learning_rate": 4.26485928465363e-05,
      "loss": 1.6752,
      "step": 6077
    },
    {
      "epoch": 1.0948885386174285,
      "grad_norm": 1.517636775970459,
      "learning_rate": 4.263458551989508e-05,
      "loss": 1.393,
      "step": 6078
    },
    {
      "epoch": 1.095068678225625,
      "grad_norm": 1.5256741046905518,
      "learning_rate": 4.2620578784103817e-05,
      "loss": 1.3955,
      "step": 6079
    },
    {
      "epoch": 1.095248817833821,
      "grad_norm": 1.5632925033569336,
      "learning_rate": 4.260657264028611e-05,
      "loss": 1.781,
      "step": 6080
    },
    {
      "epoch": 1.0954289574420175,
      "grad_norm": 1.4825087785720825,
      "learning_rate": 4.259256708956557e-05,
      "loss": 1.3952,
      "step": 6081
    },
    {
      "epoch": 1.095609097050214,
      "grad_norm": 1.4218758344650269,
      "learning_rate": 4.257856213306568e-05,
      "loss": 1.3979,
      "step": 6082
    },
    {
      "epoch": 1.0957892366584103,
      "grad_norm": 1.3864850997924805,
      "learning_rate": 4.256455777190992e-05,
      "loss": 1.4768,
      "step": 6083
    },
    {
      "epoch": 1.0959693762666065,
      "grad_norm": 1.3884128332138062,
      "learning_rate": 4.2550554007221714e-05,
      "loss": 1.3947,
      "step": 6084
    },
    {
      "epoch": 1.096149515874803,
      "grad_norm": 1.3406847715377808,
      "learning_rate": 4.2536550840124437e-05,
      "loss": 1.3267,
      "step": 6085
    },
    {
      "epoch": 1.0963296554829993,
      "grad_norm": 1.5844337940216064,
      "learning_rate": 4.252254827174141e-05,
      "loss": 1.6502,
      "step": 6086
    },
    {
      "epoch": 1.0965097950911957,
      "grad_norm": 1.4591584205627441,
      "learning_rate": 4.250854630319593e-05,
      "loss": 1.3718,
      "step": 6087
    },
    {
      "epoch": 1.0966899346993921,
      "grad_norm": 1.604280710220337,
      "learning_rate": 4.249454493561121e-05,
      "loss": 1.5846,
      "step": 6088
    },
    {
      "epoch": 1.0968700743075883,
      "grad_norm": 1.5479811429977417,
      "learning_rate": 4.2480544170110434e-05,
      "loss": 1.5252,
      "step": 6089
    },
    {
      "epoch": 1.0970502139157847,
      "grad_norm": 1.4364222288131714,
      "learning_rate": 4.2466544007816757e-05,
      "loss": 1.5394,
      "step": 6090
    },
    {
      "epoch": 1.0972303535239811,
      "grad_norm": 1.6261520385742188,
      "learning_rate": 4.245254444985324e-05,
      "loss": 1.4463,
      "step": 6091
    },
    {
      "epoch": 1.0974104931321775,
      "grad_norm": 1.571993112564087,
      "learning_rate": 4.2438545497342944e-05,
      "loss": 1.5242,
      "step": 6092
    },
    {
      "epoch": 1.0975906327403737,
      "grad_norm": 1.4871073961257935,
      "learning_rate": 4.2424547151408835e-05,
      "loss": 1.1836,
      "step": 6093
    },
    {
      "epoch": 1.0977707723485701,
      "grad_norm": 1.4581156969070435,
      "learning_rate": 4.241054941317384e-05,
      "loss": 1.4042,
      "step": 6094
    },
    {
      "epoch": 1.0979509119567665,
      "grad_norm": 1.4372199773788452,
      "learning_rate": 4.239655228376091e-05,
      "loss": 1.4195,
      "step": 6095
    },
    {
      "epoch": 1.098131051564963,
      "grad_norm": 1.6576018333435059,
      "learning_rate": 4.2382555764292844e-05,
      "loss": 1.4222,
      "step": 6096
    },
    {
      "epoch": 1.0983111911731591,
      "grad_norm": 1.435503363609314,
      "learning_rate": 4.2368559855892445e-05,
      "loss": 1.3131,
      "step": 6097
    },
    {
      "epoch": 1.0984913307813555,
      "grad_norm": 1.5506922006607056,
      "learning_rate": 4.235456455968245e-05,
      "loss": 1.5346,
      "step": 6098
    },
    {
      "epoch": 1.098671470389552,
      "grad_norm": 1.3754539489746094,
      "learning_rate": 4.234056987678554e-05,
      "loss": 1.212,
      "step": 6099
    },
    {
      "epoch": 1.0988516099977483,
      "grad_norm": 1.4971472024917603,
      "learning_rate": 4.232657580832442e-05,
      "loss": 1.0587,
      "step": 6100
    },
    {
      "epoch": 1.0990317496059445,
      "grad_norm": 1.2814102172851562,
      "learning_rate": 4.231258235542165e-05,
      "loss": 1.0463,
      "step": 6101
    },
    {
      "epoch": 1.099211889214141,
      "grad_norm": 1.3388750553131104,
      "learning_rate": 4.2298589519199774e-05,
      "loss": 1.9516,
      "step": 6102
    },
    {
      "epoch": 1.0993920288223373,
      "grad_norm": 1.3929924964904785,
      "learning_rate": 4.228459730078127e-05,
      "loss": 2.0303,
      "step": 6103
    },
    {
      "epoch": 1.0995721684305337,
      "grad_norm": 1.508176565170288,
      "learning_rate": 4.227060570128863e-05,
      "loss": 1.859,
      "step": 6104
    },
    {
      "epoch": 1.09975230803873,
      "grad_norm": 1.3213920593261719,
      "learning_rate": 4.225661472184424e-05,
      "loss": 1.6766,
      "step": 6105
    },
    {
      "epoch": 1.0999324476469263,
      "grad_norm": 1.2895421981811523,
      "learning_rate": 4.224262436357045e-05,
      "loss": 1.6967,
      "step": 6106
    },
    {
      "epoch": 1.1001125872551227,
      "grad_norm": 1.4962865114212036,
      "learning_rate": 4.222863462758956e-05,
      "loss": 1.9401,
      "step": 6107
    },
    {
      "epoch": 1.1002927268633191,
      "grad_norm": 1.7972993850708008,
      "learning_rate": 4.2214645515023785e-05,
      "loss": 1.8173,
      "step": 6108
    },
    {
      "epoch": 1.1004728664715153,
      "grad_norm": 1.3864963054656982,
      "learning_rate": 4.2200657026995373e-05,
      "loss": 1.7349,
      "step": 6109
    },
    {
      "epoch": 1.1006530060797117,
      "grad_norm": 1.4920361042022705,
      "learning_rate": 4.218666916462646e-05,
      "loss": 1.8326,
      "step": 6110
    },
    {
      "epoch": 1.1008331456879081,
      "grad_norm": 1.8557910919189453,
      "learning_rate": 4.217268192903914e-05,
      "loss": 2.0118,
      "step": 6111
    },
    {
      "epoch": 1.1010132852961045,
      "grad_norm": 1.4169776439666748,
      "learning_rate": 4.215869532135544e-05,
      "loss": 1.6487,
      "step": 6112
    },
    {
      "epoch": 1.101193424904301,
      "grad_norm": 1.309471607208252,
      "learning_rate": 4.214470934269741e-05,
      "loss": 1.4913,
      "step": 6113
    },
    {
      "epoch": 1.1013735645124971,
      "grad_norm": 1.285409688949585,
      "learning_rate": 4.213072399418697e-05,
      "loss": 1.4006,
      "step": 6114
    },
    {
      "epoch": 1.1015537041206935,
      "grad_norm": 1.4673672914505005,
      "learning_rate": 4.2116739276946004e-05,
      "loss": 1.4354,
      "step": 6115
    },
    {
      "epoch": 1.10173384372889,
      "grad_norm": 1.3363569974899292,
      "learning_rate": 4.210275519209639e-05,
      "loss": 1.3883,
      "step": 6116
    },
    {
      "epoch": 1.1019139833370863,
      "grad_norm": 1.3354419469833374,
      "learning_rate": 4.208877174075989e-05,
      "loss": 1.3839,
      "step": 6117
    },
    {
      "epoch": 1.1020941229452825,
      "grad_norm": 1.4057977199554443,
      "learning_rate": 4.2074788924058274e-05,
      "loss": 1.3925,
      "step": 6118
    },
    {
      "epoch": 1.102274262553479,
      "grad_norm": 1.295888900756836,
      "learning_rate": 4.206080674311323e-05,
      "loss": 1.3169,
      "step": 6119
    },
    {
      "epoch": 1.1024544021616753,
      "grad_norm": 1.305772304534912,
      "learning_rate": 4.204682519904641e-05,
      "loss": 1.3151,
      "step": 6120
    },
    {
      "epoch": 1.1026345417698717,
      "grad_norm": 1.3515914678573608,
      "learning_rate": 4.203284429297939e-05,
      "loss": 1.3692,
      "step": 6121
    },
    {
      "epoch": 1.102814681378068,
      "grad_norm": 1.3483145236968994,
      "learning_rate": 4.2018864026033713e-05,
      "loss": 1.2599,
      "step": 6122
    },
    {
      "epoch": 1.1029948209862643,
      "grad_norm": 1.3273777961730957,
      "learning_rate": 4.200488439933089e-05,
      "loss": 1.315,
      "step": 6123
    },
    {
      "epoch": 1.1031749605944607,
      "grad_norm": 1.4054166078567505,
      "learning_rate": 4.1990905413992345e-05,
      "loss": 1.613,
      "step": 6124
    },
    {
      "epoch": 1.1033551002026571,
      "grad_norm": 1.3974183797836304,
      "learning_rate": 4.197692707113946e-05,
      "loss": 1.3795,
      "step": 6125
    },
    {
      "epoch": 1.1035352398108533,
      "grad_norm": 1.3704001903533936,
      "learning_rate": 4.196294937189356e-05,
      "loss": 1.4559,
      "step": 6126
    },
    {
      "epoch": 1.1037153794190497,
      "grad_norm": 1.3539764881134033,
      "learning_rate": 4.1948972317375986e-05,
      "loss": 1.4313,
      "step": 6127
    },
    {
      "epoch": 1.1038955190272461,
      "grad_norm": 1.3365424871444702,
      "learning_rate": 4.1934995908707914e-05,
      "loss": 1.2696,
      "step": 6128
    },
    {
      "epoch": 1.1040756586354425,
      "grad_norm": 1.4005861282348633,
      "learning_rate": 4.1921020147010546e-05,
      "loss": 1.3777,
      "step": 6129
    },
    {
      "epoch": 1.1042557982436387,
      "grad_norm": 1.512091040611267,
      "learning_rate": 4.1907045033405e-05,
      "loss": 1.6633,
      "step": 6130
    },
    {
      "epoch": 1.1044359378518351,
      "grad_norm": 1.5257515907287598,
      "learning_rate": 4.189307056901234e-05,
      "loss": 1.3437,
      "step": 6131
    },
    {
      "epoch": 1.1046160774600315,
      "grad_norm": 1.5249392986297607,
      "learning_rate": 4.1879096754953635e-05,
      "loss": 1.4375,
      "step": 6132
    },
    {
      "epoch": 1.104796217068228,
      "grad_norm": 1.3857553005218506,
      "learning_rate": 4.186512359234984e-05,
      "loss": 1.409,
      "step": 6133
    },
    {
      "epoch": 1.1049763566764241,
      "grad_norm": 1.7051955461502075,
      "learning_rate": 4.185115108232186e-05,
      "loss": 1.5983,
      "step": 6134
    },
    {
      "epoch": 1.1051564962846205,
      "grad_norm": 1.4331978559494019,
      "learning_rate": 4.183717922599057e-05,
      "loss": 1.4122,
      "step": 6135
    },
    {
      "epoch": 1.105336635892817,
      "grad_norm": 1.413873314857483,
      "learning_rate": 4.1823208024476763e-05,
      "loss": 1.5692,
      "step": 6136
    },
    {
      "epoch": 1.1055167755010133,
      "grad_norm": 1.3154181241989136,
      "learning_rate": 4.180923747890125e-05,
      "loss": 1.3426,
      "step": 6137
    },
    {
      "epoch": 1.1056969151092098,
      "grad_norm": 1.4994357824325562,
      "learning_rate": 4.1795267590384715e-05,
      "loss": 1.3224,
      "step": 6138
    },
    {
      "epoch": 1.105877054717406,
      "grad_norm": 1.5189157724380493,
      "learning_rate": 4.178129836004783e-05,
      "loss": 1.3851,
      "step": 6139
    },
    {
      "epoch": 1.1060571943256023,
      "grad_norm": 1.490999698638916,
      "learning_rate": 4.176732978901115e-05,
      "loss": 1.508,
      "step": 6140
    },
    {
      "epoch": 1.1062373339337987,
      "grad_norm": 1.5014995336532593,
      "learning_rate": 4.175336187839529e-05,
      "loss": 1.4084,
      "step": 6141
    },
    {
      "epoch": 1.106417473541995,
      "grad_norm": 1.5019289255142212,
      "learning_rate": 4.1739394629320725e-05,
      "loss": 1.4676,
      "step": 6142
    },
    {
      "epoch": 1.1065976131501913,
      "grad_norm": 1.5706757307052612,
      "learning_rate": 4.17254280429079e-05,
      "loss": 1.5517,
      "step": 6143
    },
    {
      "epoch": 1.1067777527583877,
      "grad_norm": 1.3584853410720825,
      "learning_rate": 4.17114621202772e-05,
      "loss": 1.2703,
      "step": 6144
    },
    {
      "epoch": 1.1069578923665842,
      "grad_norm": 1.4021124839782715,
      "learning_rate": 4.1697496862548965e-05,
      "loss": 1.4356,
      "step": 6145
    },
    {
      "epoch": 1.1071380319747806,
      "grad_norm": 1.4259263277053833,
      "learning_rate": 4.1683532270843504e-05,
      "loss": 1.3867,
      "step": 6146
    },
    {
      "epoch": 1.1073181715829767,
      "grad_norm": 1.5630364418029785,
      "learning_rate": 4.1669568346281025e-05,
      "loss": 1.4107,
      "step": 6147
    },
    {
      "epoch": 1.1074983111911731,
      "grad_norm": 1.4057809114456177,
      "learning_rate": 4.165560508998172e-05,
      "loss": 1.3747,
      "step": 6148
    },
    {
      "epoch": 1.1076784507993696,
      "grad_norm": 1.383571743965149,
      "learning_rate": 4.164164250306571e-05,
      "loss": 1.2607,
      "step": 6149
    },
    {
      "epoch": 1.107858590407566,
      "grad_norm": 1.5991369485855103,
      "learning_rate": 4.162768058665305e-05,
      "loss": 1.3144,
      "step": 6150
    },
    {
      "epoch": 1.1080387300157621,
      "grad_norm": 1.554791808128357,
      "learning_rate": 4.1613719341863795e-05,
      "loss": 1.2946,
      "step": 6151
    },
    {
      "epoch": 1.1082188696239585,
      "grad_norm": 1.2826147079467773,
      "learning_rate": 4.1599758769817884e-05,
      "loss": 1.629,
      "step": 6152
    },
    {
      "epoch": 1.108399009232155,
      "grad_norm": 1.3295612335205078,
      "learning_rate": 4.1585798871635235e-05,
      "loss": 1.7586,
      "step": 6153
    },
    {
      "epoch": 1.1085791488403514,
      "grad_norm": 1.305613398551941,
      "learning_rate": 4.157183964843569e-05,
      "loss": 1.6992,
      "step": 6154
    },
    {
      "epoch": 1.1087592884485475,
      "grad_norm": 1.301514983177185,
      "learning_rate": 4.155788110133908e-05,
      "loss": 1.5269,
      "step": 6155
    },
    {
      "epoch": 1.108939428056744,
      "grad_norm": 1.3898534774780273,
      "learning_rate": 4.154392323146513e-05,
      "loss": 1.9453,
      "step": 6156
    },
    {
      "epoch": 1.1091195676649404,
      "grad_norm": 1.3027302026748657,
      "learning_rate": 4.152996603993354e-05,
      "loss": 1.5309,
      "step": 6157
    },
    {
      "epoch": 1.1092997072731368,
      "grad_norm": 1.3024576902389526,
      "learning_rate": 4.151600952786395e-05,
      "loss": 1.681,
      "step": 6158
    },
    {
      "epoch": 1.109479846881333,
      "grad_norm": 1.5968878269195557,
      "learning_rate": 4.150205369637593e-05,
      "loss": 2.0805,
      "step": 6159
    },
    {
      "epoch": 1.1096599864895293,
      "grad_norm": 1.414453387260437,
      "learning_rate": 4.148809854658904e-05,
      "loss": 1.7257,
      "step": 6160
    },
    {
      "epoch": 1.1098401260977258,
      "grad_norm": 1.6638097763061523,
      "learning_rate": 4.1474144079622734e-05,
      "loss": 1.6771,
      "step": 6161
    },
    {
      "epoch": 1.1100202657059222,
      "grad_norm": 1.6642818450927734,
      "learning_rate": 4.146019029659643e-05,
      "loss": 1.7272,
      "step": 6162
    },
    {
      "epoch": 1.1102004053141183,
      "grad_norm": 1.2560365200042725,
      "learning_rate": 4.1446237198629515e-05,
      "loss": 1.4937,
      "step": 6163
    },
    {
      "epoch": 1.1103805449223147,
      "grad_norm": 1.5085198879241943,
      "learning_rate": 4.143228478684126e-05,
      "loss": 1.7527,
      "step": 6164
    },
    {
      "epoch": 1.1105606845305112,
      "grad_norm": 1.3567441701889038,
      "learning_rate": 4.141833306235096e-05,
      "loss": 1.4795,
      "step": 6165
    },
    {
      "epoch": 1.1107408241387076,
      "grad_norm": 1.3132731914520264,
      "learning_rate": 4.140438202627782e-05,
      "loss": 1.3473,
      "step": 6166
    },
    {
      "epoch": 1.1109209637469037,
      "grad_norm": 1.3185608386993408,
      "learning_rate": 4.139043167974095e-05,
      "loss": 1.3879,
      "step": 6167
    },
    {
      "epoch": 1.1111011033551002,
      "grad_norm": 1.4361873865127563,
      "learning_rate": 4.1376482023859444e-05,
      "loss": 1.3423,
      "step": 6168
    },
    {
      "epoch": 1.1112812429632966,
      "grad_norm": 1.4034762382507324,
      "learning_rate": 4.1362533059752367e-05,
      "loss": 1.2948,
      "step": 6169
    },
    {
      "epoch": 1.111461382571493,
      "grad_norm": 1.4269036054611206,
      "learning_rate": 4.134858478853868e-05,
      "loss": 1.4175,
      "step": 6170
    },
    {
      "epoch": 1.1116415221796894,
      "grad_norm": 1.3170973062515259,
      "learning_rate": 4.133463721133731e-05,
      "loss": 1.3649,
      "step": 6171
    },
    {
      "epoch": 1.1118216617878856,
      "grad_norm": 1.3620795011520386,
      "learning_rate": 4.132069032926713e-05,
      "loss": 1.3632,
      "step": 6172
    },
    {
      "epoch": 1.112001801396082,
      "grad_norm": 1.4967515468597412,
      "learning_rate": 4.1306744143446916e-05,
      "loss": 1.4926,
      "step": 6173
    },
    {
      "epoch": 1.1121819410042784,
      "grad_norm": 1.3666870594024658,
      "learning_rate": 4.129279865499547e-05,
      "loss": 1.4527,
      "step": 6174
    },
    {
      "epoch": 1.1123620806124748,
      "grad_norm": 1.4432941675186157,
      "learning_rate": 4.1278853865031476e-05,
      "loss": 1.4963,
      "step": 6175
    },
    {
      "epoch": 1.112542220220671,
      "grad_norm": 1.399491310119629,
      "learning_rate": 4.1264909774673575e-05,
      "loss": 1.3659,
      "step": 6176
    },
    {
      "epoch": 1.1127223598288674,
      "grad_norm": 1.400530219078064,
      "learning_rate": 4.125096638504036e-05,
      "loss": 1.4273,
      "step": 6177
    },
    {
      "epoch": 1.1129024994370638,
      "grad_norm": 1.2825788259506226,
      "learning_rate": 4.123702369725035e-05,
      "loss": 1.3412,
      "step": 6178
    },
    {
      "epoch": 1.1130826390452602,
      "grad_norm": 1.600413203239441,
      "learning_rate": 4.1223081712422034e-05,
      "loss": 1.4961,
      "step": 6179
    },
    {
      "epoch": 1.1132627786534564,
      "grad_norm": 1.4961121082305908,
      "learning_rate": 4.1209140431673836e-05,
      "loss": 1.5415,
      "step": 6180
    },
    {
      "epoch": 1.1134429182616528,
      "grad_norm": 1.5522667169570923,
      "learning_rate": 4.119519985612411e-05,
      "loss": 1.4536,
      "step": 6181
    },
    {
      "epoch": 1.1136230578698492,
      "grad_norm": 1.3497360944747925,
      "learning_rate": 4.118125998689115e-05,
      "loss": 1.2582,
      "step": 6182
    },
    {
      "epoch": 1.1138031974780456,
      "grad_norm": 1.4879268407821655,
      "learning_rate": 4.116732082509323e-05,
      "loss": 1.3136,
      "step": 6183
    },
    {
      "epoch": 1.1139833370862418,
      "grad_norm": 1.3646485805511475,
      "learning_rate": 4.115338237184853e-05,
      "loss": 1.4057,
      "step": 6184
    },
    {
      "epoch": 1.1141634766944382,
      "grad_norm": 1.564008116722107,
      "learning_rate": 4.1139444628275184e-05,
      "loss": 1.6474,
      "step": 6185
    },
    {
      "epoch": 1.1143436163026346,
      "grad_norm": 1.4447704553604126,
      "learning_rate": 4.112550759549128e-05,
      "loss": 1.4193,
      "step": 6186
    },
    {
      "epoch": 1.114523755910831,
      "grad_norm": 1.3231905698776245,
      "learning_rate": 4.1111571274614826e-05,
      "loss": 1.3128,
      "step": 6187
    },
    {
      "epoch": 1.1147038955190272,
      "grad_norm": 1.4894272089004517,
      "learning_rate": 4.1097635666763804e-05,
      "loss": 1.4481,
      "step": 6188
    },
    {
      "epoch": 1.1148840351272236,
      "grad_norm": 1.4499069452285767,
      "learning_rate": 4.1083700773056106e-05,
      "loss": 1.5206,
      "step": 6189
    },
    {
      "epoch": 1.11506417473542,
      "grad_norm": 1.5077327489852905,
      "learning_rate": 4.106976659460959e-05,
      "loss": 1.2893,
      "step": 6190
    },
    {
      "epoch": 1.1152443143436164,
      "grad_norm": 1.443794846534729,
      "learning_rate": 4.105583313254205e-05,
      "loss": 1.3777,
      "step": 6191
    },
    {
      "epoch": 1.1154244539518126,
      "grad_norm": 1.6584303379058838,
      "learning_rate": 4.104190038797121e-05,
      "loss": 1.5709,
      "step": 6192
    },
    {
      "epoch": 1.115604593560009,
      "grad_norm": 1.5060536861419678,
      "learning_rate": 4.1027968362014767e-05,
      "loss": 1.452,
      "step": 6193
    },
    {
      "epoch": 1.1157847331682054,
      "grad_norm": 1.540624737739563,
      "learning_rate": 4.101403705579033e-05,
      "loss": 1.3916,
      "step": 6194
    },
    {
      "epoch": 1.1159648727764018,
      "grad_norm": 1.7504982948303223,
      "learning_rate": 4.1000106470415465e-05,
      "loss": 1.7559,
      "step": 6195
    },
    {
      "epoch": 1.1161450123845982,
      "grad_norm": 1.3713117837905884,
      "learning_rate": 4.098617660700766e-05,
      "loss": 1.3707,
      "step": 6196
    },
    {
      "epoch": 1.1163251519927944,
      "grad_norm": 1.4045027494430542,
      "learning_rate": 4.097224746668439e-05,
      "loss": 1.3817,
      "step": 6197
    },
    {
      "epoch": 1.1165052916009908,
      "grad_norm": 1.4067449569702148,
      "learning_rate": 4.095831905056305e-05,
      "loss": 1.2176,
      "step": 6198
    },
    {
      "epoch": 1.1166854312091872,
      "grad_norm": 1.4754767417907715,
      "learning_rate": 4.0944391359760936e-05,
      "loss": 1.4074,
      "step": 6199
    },
    {
      "epoch": 1.1168655708173834,
      "grad_norm": 1.6160401105880737,
      "learning_rate": 4.093046439539534e-05,
      "loss": 1.4791,
      "step": 6200
    },
    {
      "epoch": 1.1170457104255798,
      "grad_norm": 1.523664951324463,
      "learning_rate": 4.0916538158583444e-05,
      "loss": 1.471,
      "step": 6201
    },
    {
      "epoch": 1.1172258500337762,
      "grad_norm": 1.2320739030838013,
      "learning_rate": 4.090261265044246e-05,
      "loss": 1.7411,
      "step": 6202
    },
    {
      "epoch": 1.1174059896419726,
      "grad_norm": 1.2761849164962769,
      "learning_rate": 4.0888687872089454e-05,
      "loss": 1.7275,
      "step": 6203
    },
    {
      "epoch": 1.117586129250169,
      "grad_norm": 1.428338885307312,
      "learning_rate": 4.0874763824641485e-05,
      "loss": 1.7193,
      "step": 6204
    },
    {
      "epoch": 1.1177662688583652,
      "grad_norm": 1.2932162284851074,
      "learning_rate": 4.0860840509215496e-05,
      "loss": 1.8825,
      "step": 6205
    },
    {
      "epoch": 1.1179464084665616,
      "grad_norm": 1.3267757892608643,
      "learning_rate": 4.084691792692842e-05,
      "loss": 1.7919,
      "step": 6206
    },
    {
      "epoch": 1.118126548074758,
      "grad_norm": 1.3492305278778076,
      "learning_rate": 4.0832996078897144e-05,
      "loss": 1.7378,
      "step": 6207
    },
    {
      "epoch": 1.1183066876829544,
      "grad_norm": 1.4578297138214111,
      "learning_rate": 4.0819074966238454e-05,
      "loss": 1.6261,
      "step": 6208
    },
    {
      "epoch": 1.1184868272911506,
      "grad_norm": 1.4927153587341309,
      "learning_rate": 4.0805154590069105e-05,
      "loss": 1.7397,
      "step": 6209
    },
    {
      "epoch": 1.118666966899347,
      "grad_norm": 1.5851614475250244,
      "learning_rate": 4.079123495150577e-05,
      "loss": 1.7325,
      "step": 6210
    },
    {
      "epoch": 1.1188471065075434,
      "grad_norm": 1.8917169570922852,
      "learning_rate": 4.0777316051665083e-05,
      "loss": 2.1976,
      "step": 6211
    },
    {
      "epoch": 1.1190272461157398,
      "grad_norm": 1.5172230005264282,
      "learning_rate": 4.076339789166361e-05,
      "loss": 1.6581,
      "step": 6212
    },
    {
      "epoch": 1.119207385723936,
      "grad_norm": 1.4962286949157715,
      "learning_rate": 4.074948047261787e-05,
      "loss": 1.7189,
      "step": 6213
    },
    {
      "epoch": 1.1193875253321324,
      "grad_norm": 1.5342788696289062,
      "learning_rate": 4.0735563795644294e-05,
      "loss": 1.4977,
      "step": 6214
    },
    {
      "epoch": 1.1195676649403288,
      "grad_norm": 1.3722361326217651,
      "learning_rate": 4.072164786185927e-05,
      "loss": 1.4322,
      "step": 6215
    },
    {
      "epoch": 1.1197478045485252,
      "grad_norm": 1.3871073722839355,
      "learning_rate": 4.070773267237914e-05,
      "loss": 1.2903,
      "step": 6216
    },
    {
      "epoch": 1.1199279441567214,
      "grad_norm": 1.4097100496292114,
      "learning_rate": 4.0693818228320166e-05,
      "loss": 1.5195,
      "step": 6217
    },
    {
      "epoch": 1.1201080837649178,
      "grad_norm": 1.2969202995300293,
      "learning_rate": 4.067990453079857e-05,
      "loss": 1.3181,
      "step": 6218
    },
    {
      "epoch": 1.1202882233731142,
      "grad_norm": 1.3555386066436768,
      "learning_rate": 4.066599158093049e-05,
      "loss": 1.4059,
      "step": 6219
    },
    {
      "epoch": 1.1204683629813106,
      "grad_norm": 1.2622871398925781,
      "learning_rate": 4.065207937983201e-05,
      "loss": 1.3708,
      "step": 6220
    },
    {
      "epoch": 1.1206485025895068,
      "grad_norm": 1.2996799945831299,
      "learning_rate": 4.063816792861918e-05,
      "loss": 1.3131,
      "step": 6221
    },
    {
      "epoch": 1.1208286421977032,
      "grad_norm": 1.4033098220825195,
      "learning_rate": 4.0624257228407956e-05,
      "loss": 1.3551,
      "step": 6222
    },
    {
      "epoch": 1.1210087818058996,
      "grad_norm": 1.4054595232009888,
      "learning_rate": 4.0610347280314246e-05,
      "loss": 1.6246,
      "step": 6223
    },
    {
      "epoch": 1.121188921414096,
      "grad_norm": 1.4164077043533325,
      "learning_rate": 4.0596438085453895e-05,
      "loss": 1.494,
      "step": 6224
    },
    {
      "epoch": 1.1213690610222922,
      "grad_norm": 1.368730068206787,
      "learning_rate": 4.0582529644942717e-05,
      "loss": 1.3676,
      "step": 6225
    },
    {
      "epoch": 1.1215492006304886,
      "grad_norm": 1.3133366107940674,
      "learning_rate": 4.056862195989642e-05,
      "loss": 1.5143,
      "step": 6226
    },
    {
      "epoch": 1.121729340238685,
      "grad_norm": 1.3156968355178833,
      "learning_rate": 4.0554715031430676e-05,
      "loss": 1.406,
      "step": 6227
    },
    {
      "epoch": 1.1219094798468814,
      "grad_norm": 1.4045811891555786,
      "learning_rate": 4.0540808860661096e-05,
      "loss": 1.3386,
      "step": 6228
    },
    {
      "epoch": 1.1220896194550778,
      "grad_norm": 1.4746567010879517,
      "learning_rate": 4.052690344870319e-05,
      "loss": 1.4049,
      "step": 6229
    },
    {
      "epoch": 1.122269759063274,
      "grad_norm": 1.46615731716156,
      "learning_rate": 4.051299879667251e-05,
      "loss": 1.6106,
      "step": 6230
    },
    {
      "epoch": 1.1224498986714704,
      "grad_norm": 1.4413039684295654,
      "learning_rate": 4.049909490568444e-05,
      "loss": 1.5601,
      "step": 6231
    },
    {
      "epoch": 1.1226300382796668,
      "grad_norm": 1.422850489616394,
      "learning_rate": 4.048519177685435e-05,
      "loss": 1.4184,
      "step": 6232
    },
    {
      "epoch": 1.1228101778878632,
      "grad_norm": 1.4803673028945923,
      "learning_rate": 4.047128941129752e-05,
      "loss": 1.3622,
      "step": 6233
    },
    {
      "epoch": 1.1229903174960594,
      "grad_norm": 1.3702976703643799,
      "learning_rate": 4.045738781012923e-05,
      "loss": 1.2834,
      "step": 6234
    },
    {
      "epoch": 1.1231704571042558,
      "grad_norm": 1.5145506858825684,
      "learning_rate": 4.044348697446464e-05,
      "loss": 1.3921,
      "step": 6235
    },
    {
      "epoch": 1.1233505967124522,
      "grad_norm": 1.624612808227539,
      "learning_rate": 4.042958690541887e-05,
      "loss": 1.3908,
      "step": 6236
    },
    {
      "epoch": 1.1235307363206486,
      "grad_norm": 1.5227570533752441,
      "learning_rate": 4.0415687604106986e-05,
      "loss": 1.5937,
      "step": 6237
    },
    {
      "epoch": 1.1237108759288448,
      "grad_norm": 1.5994679927825928,
      "learning_rate": 4.040178907164395e-05,
      "loss": 1.4782,
      "step": 6238
    },
    {
      "epoch": 1.1238910155370412,
      "grad_norm": 1.4512572288513184,
      "learning_rate": 4.038789130914474e-05,
      "loss": 1.4218,
      "step": 6239
    },
    {
      "epoch": 1.1240711551452376,
      "grad_norm": 1.345016360282898,
      "learning_rate": 4.03739943177242e-05,
      "loss": 1.2091,
      "step": 6240
    },
    {
      "epoch": 1.124251294753434,
      "grad_norm": 1.4000985622406006,
      "learning_rate": 4.036009809849715e-05,
      "loss": 1.2851,
      "step": 6241
    },
    {
      "epoch": 1.1244314343616302,
      "grad_norm": 1.5489205121994019,
      "learning_rate": 4.0346202652578335e-05,
      "loss": 1.4182,
      "step": 6242
    },
    {
      "epoch": 1.1246115739698266,
      "grad_norm": 1.365464448928833,
      "learning_rate": 4.0332307981082436e-05,
      "loss": 1.2173,
      "step": 6243
    },
    {
      "epoch": 1.124791713578023,
      "grad_norm": 1.4468257427215576,
      "learning_rate": 4.031841408512409e-05,
      "loss": 1.2509,
      "step": 6244
    },
    {
      "epoch": 1.1249718531862194,
      "grad_norm": 1.3929481506347656,
      "learning_rate": 4.030452096581785e-05,
      "loss": 1.3307,
      "step": 6245
    },
    {
      "epoch": 1.1251519927944156,
      "grad_norm": 1.3547942638397217,
      "learning_rate": 4.029062862427821e-05,
      "loss": 1.2824,
      "step": 6246
    },
    {
      "epoch": 1.125332132402612,
      "grad_norm": 1.6060343980789185,
      "learning_rate": 4.027673706161961e-05,
      "loss": 1.5025,
      "step": 6247
    },
    {
      "epoch": 1.1255122720108084,
      "grad_norm": 1.5124266147613525,
      "learning_rate": 4.0262846278956426e-05,
      "loss": 1.4452,
      "step": 6248
    },
    {
      "epoch": 1.1256924116190048,
      "grad_norm": 1.726059913635254,
      "learning_rate": 4.0248956277402985e-05,
      "loss": 1.6267,
      "step": 6249
    },
    {
      "epoch": 1.125872551227201,
      "grad_norm": 1.3140368461608887,
      "learning_rate": 4.0235067058073514e-05,
      "loss": 0.9682,
      "step": 6250
    },
    {
      "epoch": 1.1260526908353974,
      "grad_norm": 1.5666542053222656,
      "learning_rate": 4.0221178622082206e-05,
      "loss": 1.3454,
      "step": 6251
    },
    {
      "epoch": 1.1262328304435938,
      "grad_norm": 1.1789259910583496,
      "learning_rate": 4.0207290970543175e-05,
      "loss": 1.6018,
      "step": 6252
    },
    {
      "epoch": 1.1264129700517902,
      "grad_norm": 1.466103196144104,
      "learning_rate": 4.019340410457051e-05,
      "loss": 1.8465,
      "step": 6253
    },
    {
      "epoch": 1.1265931096599866,
      "grad_norm": 1.2758469581604004,
      "learning_rate": 4.017951802527819e-05,
      "loss": 1.5849,
      "step": 6254
    },
    {
      "epoch": 1.1267732492681828,
      "grad_norm": 1.4252521991729736,
      "learning_rate": 4.0165632733780144e-05,
      "loss": 1.9982,
      "step": 6255
    },
    {
      "epoch": 1.1269533888763792,
      "grad_norm": 1.3692508935928345,
      "learning_rate": 4.015174823119025e-05,
      "loss": 1.6104,
      "step": 6256
    },
    {
      "epoch": 1.1271335284845756,
      "grad_norm": 1.4817110300064087,
      "learning_rate": 4.013786451862231e-05,
      "loss": 1.9699,
      "step": 6257
    },
    {
      "epoch": 1.1273136680927718,
      "grad_norm": 1.4180225133895874,
      "learning_rate": 4.012398159719009e-05,
      "loss": 1.8913,
      "step": 6258
    },
    {
      "epoch": 1.1274938077009682,
      "grad_norm": 1.4201819896697998,
      "learning_rate": 4.011009946800724e-05,
      "loss": 1.6075,
      "step": 6259
    },
    {
      "epoch": 1.1276739473091646,
      "grad_norm": 1.7947778701782227,
      "learning_rate": 4.0096218132187406e-05,
      "loss": 2.1542,
      "step": 6260
    },
    {
      "epoch": 1.127854086917361,
      "grad_norm": 1.7639341354370117,
      "learning_rate": 4.008233759084411e-05,
      "loss": 1.7316,
      "step": 6261
    },
    {
      "epoch": 1.1280342265255574,
      "grad_norm": 1.5179598331451416,
      "learning_rate": 4.006845784509087e-05,
      "loss": 1.5267,
      "step": 6262
    },
    {
      "epoch": 1.1282143661337536,
      "grad_norm": 1.4841554164886475,
      "learning_rate": 4.005457889604113e-05,
      "loss": 1.5748,
      "step": 6263
    },
    {
      "epoch": 1.12839450574195,
      "grad_norm": 1.2656651735305786,
      "learning_rate": 4.0040700744808204e-05,
      "loss": 1.3786,
      "step": 6264
    },
    {
      "epoch": 1.1285746453501464,
      "grad_norm": 1.4172027111053467,
      "learning_rate": 4.002682339250542e-05,
      "loss": 1.4976,
      "step": 6265
    },
    {
      "epoch": 1.1287547849583428,
      "grad_norm": 1.3436676263809204,
      "learning_rate": 4.001294684024599e-05,
      "loss": 1.3844,
      "step": 6266
    },
    {
      "epoch": 1.128934924566539,
      "grad_norm": 1.3546191453933716,
      "learning_rate": 3.999907108914311e-05,
      "loss": 1.3296,
      "step": 6267
    },
    {
      "epoch": 1.1291150641747354,
      "grad_norm": 1.3178660869598389,
      "learning_rate": 3.9985196140309876e-05,
      "loss": 1.3973,
      "step": 6268
    },
    {
      "epoch": 1.1292952037829318,
      "grad_norm": 1.2529330253601074,
      "learning_rate": 3.9971321994859345e-05,
      "loss": 1.3467,
      "step": 6269
    },
    {
      "epoch": 1.1294753433911282,
      "grad_norm": 1.3508087396621704,
      "learning_rate": 3.995744865390447e-05,
      "loss": 1.4668,
      "step": 6270
    },
    {
      "epoch": 1.1296554829993244,
      "grad_norm": 1.421682357788086,
      "learning_rate": 3.994357611855815e-05,
      "loss": 1.579,
      "step": 6271
    },
    {
      "epoch": 1.1298356226075208,
      "grad_norm": 1.3042895793914795,
      "learning_rate": 3.992970438993327e-05,
      "loss": 1.2686,
      "step": 6272
    },
    {
      "epoch": 1.1300157622157172,
      "grad_norm": 1.4298288822174072,
      "learning_rate": 3.9915833469142607e-05,
      "loss": 1.4867,
      "step": 6273
    },
    {
      "epoch": 1.1301959018239136,
      "grad_norm": 1.3123995065689087,
      "learning_rate": 3.990196335729887e-05,
      "loss": 1.3188,
      "step": 6274
    },
    {
      "epoch": 1.1303760414321098,
      "grad_norm": 1.3402804136276245,
      "learning_rate": 3.98880940555147e-05,
      "loss": 1.2947,
      "step": 6275
    },
    {
      "epoch": 1.1305561810403062,
      "grad_norm": 1.485793113708496,
      "learning_rate": 3.987422556490271e-05,
      "loss": 1.6032,
      "step": 6276
    },
    {
      "epoch": 1.1307363206485026,
      "grad_norm": 1.3809276819229126,
      "learning_rate": 3.986035788657542e-05,
      "loss": 1.2968,
      "step": 6277
    },
    {
      "epoch": 1.130916460256699,
      "grad_norm": 1.3818897008895874,
      "learning_rate": 3.984649102164528e-05,
      "loss": 1.3279,
      "step": 6278
    },
    {
      "epoch": 1.1310965998648954,
      "grad_norm": 1.5038496255874634,
      "learning_rate": 3.9832624971224675e-05,
      "loss": 1.4183,
      "step": 6279
    },
    {
      "epoch": 1.1312767394730916,
      "grad_norm": 1.48307204246521,
      "learning_rate": 3.9818759736425935e-05,
      "loss": 1.5878,
      "step": 6280
    },
    {
      "epoch": 1.131456879081288,
      "grad_norm": 1.4590240716934204,
      "learning_rate": 3.980489531836135e-05,
      "loss": 1.5134,
      "step": 6281
    },
    {
      "epoch": 1.1316370186894844,
      "grad_norm": 1.5738403797149658,
      "learning_rate": 3.979103171814309e-05,
      "loss": 1.778,
      "step": 6282
    },
    {
      "epoch": 1.1318171582976806,
      "grad_norm": 1.4653578996658325,
      "learning_rate": 3.977716893688329e-05,
      "loss": 1.5657,
      "step": 6283
    },
    {
      "epoch": 1.131997297905877,
      "grad_norm": 1.470521092414856,
      "learning_rate": 3.976330697569403e-05,
      "loss": 1.4279,
      "step": 6284
    },
    {
      "epoch": 1.1321774375140734,
      "grad_norm": 1.4195396900177002,
      "learning_rate": 3.974944583568727e-05,
      "loss": 1.3877,
      "step": 6285
    },
    {
      "epoch": 1.1323575771222698,
      "grad_norm": 1.4443098306655884,
      "learning_rate": 3.973558551797499e-05,
      "loss": 1.2852,
      "step": 6286
    },
    {
      "epoch": 1.1325377167304662,
      "grad_norm": 1.5698120594024658,
      "learning_rate": 3.972172602366905e-05,
      "loss": 1.6662,
      "step": 6287
    },
    {
      "epoch": 1.1327178563386624,
      "grad_norm": 1.5509881973266602,
      "learning_rate": 3.9707867353881234e-05,
      "loss": 1.5111,
      "step": 6288
    },
    {
      "epoch": 1.1328979959468588,
      "grad_norm": 1.4579596519470215,
      "learning_rate": 3.969400950972327e-05,
      "loss": 1.4525,
      "step": 6289
    },
    {
      "epoch": 1.1330781355550552,
      "grad_norm": 1.4359208345413208,
      "learning_rate": 3.968015249230686e-05,
      "loss": 1.4731,
      "step": 6290
    },
    {
      "epoch": 1.1332582751632516,
      "grad_norm": 1.5758246183395386,
      "learning_rate": 3.966629630274359e-05,
      "loss": 1.6345,
      "step": 6291
    },
    {
      "epoch": 1.1334384147714478,
      "grad_norm": 1.3557735681533813,
      "learning_rate": 3.9652440942145e-05,
      "loss": 1.2489,
      "step": 6292
    },
    {
      "epoch": 1.1336185543796442,
      "grad_norm": 1.5547388792037964,
      "learning_rate": 3.963858641162255e-05,
      "loss": 1.3588,
      "step": 6293
    },
    {
      "epoch": 1.1337986939878406,
      "grad_norm": 1.4470330476760864,
      "learning_rate": 3.962473271228764e-05,
      "loss": 1.3628,
      "step": 6294
    },
    {
      "epoch": 1.133978833596037,
      "grad_norm": 1.5260344743728638,
      "learning_rate": 3.9610879845251643e-05,
      "loss": 1.4396,
      "step": 6295
    },
    {
      "epoch": 1.1341589732042332,
      "grad_norm": 1.450937271118164,
      "learning_rate": 3.95970278116258e-05,
      "loss": 1.3569,
      "step": 6296
    },
    {
      "epoch": 1.1343391128124296,
      "grad_norm": 1.525468111038208,
      "learning_rate": 3.958317661252131e-05,
      "loss": 1.3179,
      "step": 6297
    },
    {
      "epoch": 1.134519252420626,
      "grad_norm": 1.706613302230835,
      "learning_rate": 3.956932624904932e-05,
      "loss": 1.3127,
      "step": 6298
    },
    {
      "epoch": 1.1346993920288224,
      "grad_norm": 1.4617421627044678,
      "learning_rate": 3.955547672232088e-05,
      "loss": 1.1749,
      "step": 6299
    },
    {
      "epoch": 1.1348795316370186,
      "grad_norm": 1.493533730506897,
      "learning_rate": 3.954162803344703e-05,
      "loss": 1.2902,
      "step": 6300
    },
    {
      "epoch": 1.135059671245215,
      "grad_norm": 1.4127315282821655,
      "learning_rate": 3.95277801835387e-05,
      "loss": 1.253,
      "step": 6301
    },
    {
      "epoch": 1.1352398108534114,
      "grad_norm": 1.4898011684417725,
      "learning_rate": 3.951393317370672e-05,
      "loss": 2.1482,
      "step": 6302
    },
    {
      "epoch": 1.1354199504616078,
      "grad_norm": 1.21418035030365,
      "learning_rate": 3.9500087005061905e-05,
      "loss": 1.7092,
      "step": 6303
    },
    {
      "epoch": 1.135600090069804,
      "grad_norm": 1.4689902067184448,
      "learning_rate": 3.9486241678715006e-05,
      "loss": 1.9027,
      "step": 6304
    },
    {
      "epoch": 1.1357802296780004,
      "grad_norm": 1.3548260927200317,
      "learning_rate": 3.947239719577669e-05,
      "loss": 1.9257,
      "step": 6305
    },
    {
      "epoch": 1.1359603692861968,
      "grad_norm": 1.430167317390442,
      "learning_rate": 3.945855355735754e-05,
      "loss": 2.0,
      "step": 6306
    },
    {
      "epoch": 1.1361405088943932,
      "grad_norm": 1.4146255254745483,
      "learning_rate": 3.944471076456808e-05,
      "loss": 1.6218,
      "step": 6307
    },
    {
      "epoch": 1.1363206485025894,
      "grad_norm": 1.4098167419433594,
      "learning_rate": 3.9430868818518784e-05,
      "loss": 1.7763,
      "step": 6308
    },
    {
      "epoch": 1.1365007881107858,
      "grad_norm": 1.6797902584075928,
      "learning_rate": 3.941702772032004e-05,
      "loss": 1.9914,
      "step": 6309
    },
    {
      "epoch": 1.1366809277189822,
      "grad_norm": 1.7100614309310913,
      "learning_rate": 3.940318747108219e-05,
      "loss": 2.1759,
      "step": 6310
    },
    {
      "epoch": 1.1368610673271786,
      "grad_norm": 1.419112205505371,
      "learning_rate": 3.938934807191548e-05,
      "loss": 1.7008,
      "step": 6311
    },
    {
      "epoch": 1.137041206935375,
      "grad_norm": 1.3717199563980103,
      "learning_rate": 3.937550952393009e-05,
      "loss": 1.2475,
      "step": 6312
    },
    {
      "epoch": 1.1372213465435712,
      "grad_norm": 1.3663500547409058,
      "learning_rate": 3.936167182823615e-05,
      "loss": 1.4451,
      "step": 6313
    },
    {
      "epoch": 1.1374014861517676,
      "grad_norm": 1.3883980512619019,
      "learning_rate": 3.934783498594372e-05,
      "loss": 1.4397,
      "step": 6314
    },
    {
      "epoch": 1.137581625759964,
      "grad_norm": 1.3909910917282104,
      "learning_rate": 3.9333998998162776e-05,
      "loss": 1.4214,
      "step": 6315
    },
    {
      "epoch": 1.1377617653681602,
      "grad_norm": 1.3119492530822754,
      "learning_rate": 3.9320163866003235e-05,
      "loss": 1.4058,
      "step": 6316
    },
    {
      "epoch": 1.1379419049763566,
      "grad_norm": 1.498169183731079,
      "learning_rate": 3.9306329590574945e-05,
      "loss": 1.5825,
      "step": 6317
    },
    {
      "epoch": 1.138122044584553,
      "grad_norm": 1.2926905155181885,
      "learning_rate": 3.9292496172987695e-05,
      "loss": 1.3581,
      "step": 6318
    },
    {
      "epoch": 1.1383021841927494,
      "grad_norm": 1.4041759967803955,
      "learning_rate": 3.927866361435118e-05,
      "loss": 1.3549,
      "step": 6319
    },
    {
      "epoch": 1.1384823238009458,
      "grad_norm": 1.4187194108963013,
      "learning_rate": 3.9264831915775047e-05,
      "loss": 1.8183,
      "step": 6320
    },
    {
      "epoch": 1.138662463409142,
      "grad_norm": 1.3764961957931519,
      "learning_rate": 3.9251001078368874e-05,
      "loss": 1.3475,
      "step": 6321
    },
    {
      "epoch": 1.1388426030173384,
      "grad_norm": 1.3768562078475952,
      "learning_rate": 3.923717110324215e-05,
      "loss": 1.4672,
      "step": 6322
    },
    {
      "epoch": 1.1390227426255348,
      "grad_norm": 1.3775190114974976,
      "learning_rate": 3.922334199150432e-05,
      "loss": 1.3738,
      "step": 6323
    },
    {
      "epoch": 1.1392028822337312,
      "grad_norm": 1.4008256196975708,
      "learning_rate": 3.920951374426476e-05,
      "loss": 1.6002,
      "step": 6324
    },
    {
      "epoch": 1.1393830218419274,
      "grad_norm": 1.4886822700500488,
      "learning_rate": 3.9195686362632745e-05,
      "loss": 1.4508,
      "step": 6325
    },
    {
      "epoch": 1.1395631614501238,
      "grad_norm": 1.3538321256637573,
      "learning_rate": 3.9181859847717505e-05,
      "loss": 1.4591,
      "step": 6326
    },
    {
      "epoch": 1.1397433010583202,
      "grad_norm": 1.4685170650482178,
      "learning_rate": 3.916803420062818e-05,
      "loss": 1.6289,
      "step": 6327
    },
    {
      "epoch": 1.1399234406665166,
      "grad_norm": 1.4800105094909668,
      "learning_rate": 3.91542094224739e-05,
      "loss": 1.5703,
      "step": 6328
    },
    {
      "epoch": 1.1401035802747128,
      "grad_norm": 1.6208146810531616,
      "learning_rate": 3.914038551436366e-05,
      "loss": 1.4553,
      "step": 6329
    },
    {
      "epoch": 1.1402837198829092,
      "grad_norm": 1.3834105730056763,
      "learning_rate": 3.912656247740639e-05,
      "loss": 1.2279,
      "step": 6330
    },
    {
      "epoch": 1.1404638594911056,
      "grad_norm": 1.4585415124893188,
      "learning_rate": 3.911274031271097e-05,
      "loss": 1.3839,
      "step": 6331
    },
    {
      "epoch": 1.140643999099302,
      "grad_norm": 1.326542854309082,
      "learning_rate": 3.909891902138624e-05,
      "loss": 1.2692,
      "step": 6332
    },
    {
      "epoch": 1.1408241387074982,
      "grad_norm": 1.4512786865234375,
      "learning_rate": 3.908509860454092e-05,
      "loss": 1.3632,
      "step": 6333
    },
    {
      "epoch": 1.1410042783156946,
      "grad_norm": 1.4155592918395996,
      "learning_rate": 3.907127906328367e-05,
      "loss": 1.4285,
      "step": 6334
    },
    {
      "epoch": 1.141184417923891,
      "grad_norm": 1.5636345148086548,
      "learning_rate": 3.9057460398723095e-05,
      "loss": 1.5357,
      "step": 6335
    },
    {
      "epoch": 1.1413645575320874,
      "grad_norm": 1.5808826684951782,
      "learning_rate": 3.904364261196769e-05,
      "loss": 1.6927,
      "step": 6336
    },
    {
      "epoch": 1.1415446971402838,
      "grad_norm": 1.5937719345092773,
      "learning_rate": 3.902982570412596e-05,
      "loss": 1.4371,
      "step": 6337
    },
    {
      "epoch": 1.14172483674848,
      "grad_norm": 1.3962398767471313,
      "learning_rate": 3.9016009676306266e-05,
      "loss": 1.441,
      "step": 6338
    },
    {
      "epoch": 1.1419049763566764,
      "grad_norm": 1.26748526096344,
      "learning_rate": 3.900219452961692e-05,
      "loss": 1.2156,
      "step": 6339
    },
    {
      "epoch": 1.1420851159648728,
      "grad_norm": 1.5198264122009277,
      "learning_rate": 3.898838026516617e-05,
      "loss": 1.4083,
      "step": 6340
    },
    {
      "epoch": 1.142265255573069,
      "grad_norm": 1.433039665222168,
      "learning_rate": 3.8974566884062204e-05,
      "loss": 1.4679,
      "step": 6341
    },
    {
      "epoch": 1.1424453951812654,
      "grad_norm": 1.4591983556747437,
      "learning_rate": 3.8960754387413106e-05,
      "loss": 1.5379,
      "step": 6342
    },
    {
      "epoch": 1.1426255347894618,
      "grad_norm": 1.4905363321304321,
      "learning_rate": 3.894694277632692e-05,
      "loss": 1.4098,
      "step": 6343
    },
    {
      "epoch": 1.1428056743976582,
      "grad_norm": 1.4621517658233643,
      "learning_rate": 3.893313205191159e-05,
      "loss": 1.5027,
      "step": 6344
    },
    {
      "epoch": 1.1429858140058546,
      "grad_norm": 1.5439268350601196,
      "learning_rate": 3.8919322215275015e-05,
      "loss": 1.4637,
      "step": 6345
    },
    {
      "epoch": 1.1431659536140508,
      "grad_norm": 1.6386984586715698,
      "learning_rate": 3.890551326752503e-05,
      "loss": 1.6819,
      "step": 6346
    },
    {
      "epoch": 1.1433460932222472,
      "grad_norm": 1.6549303531646729,
      "learning_rate": 3.8891705209769355e-05,
      "loss": 1.5999,
      "step": 6347
    },
    {
      "epoch": 1.1435262328304436,
      "grad_norm": 1.6477750539779663,
      "learning_rate": 3.887789804311568e-05,
      "loss": 1.605,
      "step": 6348
    },
    {
      "epoch": 1.14370637243864,
      "grad_norm": 1.4324318170547485,
      "learning_rate": 3.886409176867161e-05,
      "loss": 1.2021,
      "step": 6349
    },
    {
      "epoch": 1.1438865120468362,
      "grad_norm": 1.399327039718628,
      "learning_rate": 3.885028638754466e-05,
      "loss": 1.3076,
      "step": 6350
    },
    {
      "epoch": 1.1440666516550326,
      "grad_norm": 1.422559142112732,
      "learning_rate": 3.883648190084232e-05,
      "loss": 1.4297,
      "step": 6351
    },
    {
      "epoch": 1.144246791263229,
      "grad_norm": 1.458188533782959,
      "learning_rate": 3.882267830967197e-05,
      "loss": 2.1258,
      "step": 6352
    },
    {
      "epoch": 1.1444269308714254,
      "grad_norm": 1.3275213241577148,
      "learning_rate": 3.880887561514091e-05,
      "loss": 1.7881,
      "step": 6353
    },
    {
      "epoch": 1.1446070704796216,
      "grad_norm": 1.3926016092300415,
      "learning_rate": 3.879507381835639e-05,
      "loss": 2.0682,
      "step": 6354
    },
    {
      "epoch": 1.144787210087818,
      "grad_norm": 1.3775486946105957,
      "learning_rate": 3.8781272920425605e-05,
      "loss": 1.8687,
      "step": 6355
    },
    {
      "epoch": 1.1449673496960144,
      "grad_norm": 1.4278829097747803,
      "learning_rate": 3.876747292245564e-05,
      "loss": 1.8642,
      "step": 6356
    },
    {
      "epoch": 1.1451474893042108,
      "grad_norm": 1.298592209815979,
      "learning_rate": 3.875367382555352e-05,
      "loss": 1.7746,
      "step": 6357
    },
    {
      "epoch": 1.145327628912407,
      "grad_norm": 1.5147939920425415,
      "learning_rate": 3.873987563082622e-05,
      "loss": 1.8696,
      "step": 6358
    },
    {
      "epoch": 1.1455077685206034,
      "grad_norm": 1.5675779581069946,
      "learning_rate": 3.872607833938058e-05,
      "loss": 1.76,
      "step": 6359
    },
    {
      "epoch": 1.1456879081287998,
      "grad_norm": 1.6738656759262085,
      "learning_rate": 3.871228195232348e-05,
      "loss": 2.1328,
      "step": 6360
    },
    {
      "epoch": 1.1458680477369962,
      "grad_norm": 1.5009450912475586,
      "learning_rate": 3.869848647076161e-05,
      "loss": 1.6758,
      "step": 6361
    },
    {
      "epoch": 1.1460481873451926,
      "grad_norm": 1.436381220817566,
      "learning_rate": 3.8684691895801656e-05,
      "loss": 1.5397,
      "step": 6362
    },
    {
      "epoch": 1.1462283269533888,
      "grad_norm": 1.3255953788757324,
      "learning_rate": 3.867089822855021e-05,
      "loss": 1.542,
      "step": 6363
    },
    {
      "epoch": 1.1464084665615852,
      "grad_norm": 1.4282640218734741,
      "learning_rate": 3.8657105470113755e-05,
      "loss": 1.6246,
      "step": 6364
    },
    {
      "epoch": 1.1465886061697816,
      "grad_norm": 1.4649337530136108,
      "learning_rate": 3.8643313621598807e-05,
      "loss": 1.5199,
      "step": 6365
    },
    {
      "epoch": 1.1467687457779778,
      "grad_norm": 1.2402459383010864,
      "learning_rate": 3.862952268411172e-05,
      "loss": 1.1631,
      "step": 6366
    },
    {
      "epoch": 1.1469488853861742,
      "grad_norm": 1.5456969738006592,
      "learning_rate": 3.8615732658758764e-05,
      "loss": 1.6976,
      "step": 6367
    },
    {
      "epoch": 1.1471290249943706,
      "grad_norm": 1.3947255611419678,
      "learning_rate": 3.8601943546646175e-05,
      "loss": 1.4114,
      "step": 6368
    },
    {
      "epoch": 1.147309164602567,
      "grad_norm": 1.5307798385620117,
      "learning_rate": 3.8588155348880146e-05,
      "loss": 1.4185,
      "step": 6369
    },
    {
      "epoch": 1.1474893042107634,
      "grad_norm": 1.4056857824325562,
      "learning_rate": 3.857436806656673e-05,
      "loss": 1.5179,
      "step": 6370
    },
    {
      "epoch": 1.1476694438189596,
      "grad_norm": 1.3568029403686523,
      "learning_rate": 3.8560581700811946e-05,
      "loss": 1.3396,
      "step": 6371
    },
    {
      "epoch": 1.147849583427156,
      "grad_norm": 1.7604305744171143,
      "learning_rate": 3.854679625272174e-05,
      "loss": 1.6465,
      "step": 6372
    },
    {
      "epoch": 1.1480297230353524,
      "grad_norm": 1.3852447271347046,
      "learning_rate": 3.8533011723401924e-05,
      "loss": 1.454,
      "step": 6373
    },
    {
      "epoch": 1.1482098626435486,
      "grad_norm": 1.4612922668457031,
      "learning_rate": 3.851922811395834e-05,
      "loss": 1.481,
      "step": 6374
    },
    {
      "epoch": 1.148390002251745,
      "grad_norm": 1.456812858581543,
      "learning_rate": 3.850544542549669e-05,
      "loss": 1.3663,
      "step": 6375
    },
    {
      "epoch": 1.1485701418599414,
      "grad_norm": 1.247881531715393,
      "learning_rate": 3.849166365912261e-05,
      "loss": 1.1812,
      "step": 6376
    },
    {
      "epoch": 1.1487502814681378,
      "grad_norm": 1.5341315269470215,
      "learning_rate": 3.8477882815941665e-05,
      "loss": 1.5675,
      "step": 6377
    },
    {
      "epoch": 1.1489304210763343,
      "grad_norm": 1.4741030931472778,
      "learning_rate": 3.846410289705933e-05,
      "loss": 1.5101,
      "step": 6378
    },
    {
      "epoch": 1.1491105606845304,
      "grad_norm": 1.4194914102554321,
      "learning_rate": 3.8450323903581066e-05,
      "loss": 1.466,
      "step": 6379
    },
    {
      "epoch": 1.1492907002927268,
      "grad_norm": 1.3781687021255493,
      "learning_rate": 3.843654583661218e-05,
      "loss": 1.4657,
      "step": 6380
    },
    {
      "epoch": 1.1494708399009232,
      "grad_norm": 1.627516746520996,
      "learning_rate": 3.842276869725796e-05,
      "loss": 1.6975,
      "step": 6381
    },
    {
      "epoch": 1.1496509795091197,
      "grad_norm": 1.3546347618103027,
      "learning_rate": 3.840899248662358e-05,
      "loss": 1.3582,
      "step": 6382
    },
    {
      "epoch": 1.1498311191173158,
      "grad_norm": 1.63339102268219,
      "learning_rate": 3.83952172058142e-05,
      "loss": 1.6237,
      "step": 6383
    },
    {
      "epoch": 1.1500112587255122,
      "grad_norm": 1.3516658544540405,
      "learning_rate": 3.8381442855934836e-05,
      "loss": 1.2936,
      "step": 6384
    },
    {
      "epoch": 1.1501913983337086,
      "grad_norm": 1.4490880966186523,
      "learning_rate": 3.836766943809047e-05,
      "loss": 1.405,
      "step": 6385
    },
    {
      "epoch": 1.150371537941905,
      "grad_norm": 1.4267319440841675,
      "learning_rate": 3.835389695338599e-05,
      "loss": 1.3181,
      "step": 6386
    },
    {
      "epoch": 1.1505516775501012,
      "grad_norm": 1.4693578481674194,
      "learning_rate": 3.834012540292622e-05,
      "loss": 1.4233,
      "step": 6387
    },
    {
      "epoch": 1.1507318171582976,
      "grad_norm": 1.3248298168182373,
      "learning_rate": 3.832635478781592e-05,
      "loss": 1.2684,
      "step": 6388
    },
    {
      "epoch": 1.150911956766494,
      "grad_norm": 1.3616430759429932,
      "learning_rate": 3.831258510915974e-05,
      "loss": 1.4677,
      "step": 6389
    },
    {
      "epoch": 1.1510920963746905,
      "grad_norm": 1.4244612455368042,
      "learning_rate": 3.829881636806231e-05,
      "loss": 1.4384,
      "step": 6390
    },
    {
      "epoch": 1.1512722359828866,
      "grad_norm": 1.437246561050415,
      "learning_rate": 3.828504856562812e-05,
      "loss": 1.5431,
      "step": 6391
    },
    {
      "epoch": 1.151452375591083,
      "grad_norm": 1.3519972562789917,
      "learning_rate": 3.82712817029616e-05,
      "loss": 1.3013,
      "step": 6392
    },
    {
      "epoch": 1.1516325151992794,
      "grad_norm": 1.6793372631072998,
      "learning_rate": 3.8257515781167196e-05,
      "loss": 1.3241,
      "step": 6393
    },
    {
      "epoch": 1.1518126548074759,
      "grad_norm": 1.5597165822982788,
      "learning_rate": 3.824375080134914e-05,
      "loss": 1.6509,
      "step": 6394
    },
    {
      "epoch": 1.1519927944156723,
      "grad_norm": 1.557692289352417,
      "learning_rate": 3.822998676461166e-05,
      "loss": 1.3878,
      "step": 6395
    },
    {
      "epoch": 1.1521729340238684,
      "grad_norm": 1.5494650602340698,
      "learning_rate": 3.8216223672058896e-05,
      "loss": 1.3849,
      "step": 6396
    },
    {
      "epoch": 1.1523530736320649,
      "grad_norm": 1.5278879404067993,
      "learning_rate": 3.8202461524794955e-05,
      "loss": 1.0978,
      "step": 6397
    },
    {
      "epoch": 1.1525332132402613,
      "grad_norm": 1.3613522052764893,
      "learning_rate": 3.81887003239238e-05,
      "loss": 1.1918,
      "step": 6398
    },
    {
      "epoch": 1.1527133528484574,
      "grad_norm": 1.652655005455017,
      "learning_rate": 3.8174940070549356e-05,
      "loss": 1.5971,
      "step": 6399
    },
    {
      "epoch": 1.1528934924566538,
      "grad_norm": 1.5285753011703491,
      "learning_rate": 3.816118076577546e-05,
      "loss": 1.1683,
      "step": 6400
    },
    {
      "epoch": 1.1530736320648503,
      "grad_norm": 1.4251883029937744,
      "learning_rate": 3.8147422410705846e-05,
      "loss": 1.0841,
      "step": 6401
    },
    {
      "epoch": 1.1532537716730467,
      "grad_norm": 1.289496660232544,
      "learning_rate": 3.8133665006444255e-05,
      "loss": 1.6405,
      "step": 6402
    },
    {
      "epoch": 1.153433911281243,
      "grad_norm": 1.2119741439819336,
      "learning_rate": 3.811990855409428e-05,
      "loss": 1.6039,
      "step": 6403
    },
    {
      "epoch": 1.1536140508894392,
      "grad_norm": 1.4235894680023193,
      "learning_rate": 3.8106153054759456e-05,
      "loss": 1.9263,
      "step": 6404
    },
    {
      "epoch": 1.1537941904976357,
      "grad_norm": 1.3450112342834473,
      "learning_rate": 3.8092398509543257e-05,
      "loss": 1.8604,
      "step": 6405
    },
    {
      "epoch": 1.153974330105832,
      "grad_norm": 1.3842041492462158,
      "learning_rate": 3.807864491954901e-05,
      "loss": 1.8604,
      "step": 6406
    },
    {
      "epoch": 1.1541544697140285,
      "grad_norm": 1.4505120515823364,
      "learning_rate": 3.806489228588008e-05,
      "loss": 1.6408,
      "step": 6407
    },
    {
      "epoch": 1.1543346093222246,
      "grad_norm": 1.4167990684509277,
      "learning_rate": 3.805114060963968e-05,
      "loss": 1.6922,
      "step": 6408
    },
    {
      "epoch": 1.154514748930421,
      "grad_norm": 1.6952906847000122,
      "learning_rate": 3.803738989193096e-05,
      "loss": 1.6504,
      "step": 6409
    },
    {
      "epoch": 1.1546948885386175,
      "grad_norm": 1.7385038137435913,
      "learning_rate": 3.8023640133856985e-05,
      "loss": 1.8393,
      "step": 6410
    },
    {
      "epoch": 1.1548750281468139,
      "grad_norm": 1.6009682416915894,
      "learning_rate": 3.8009891336520774e-05,
      "loss": 1.4323,
      "step": 6411
    },
    {
      "epoch": 1.15505516775501,
      "grad_norm": 1.4975910186767578,
      "learning_rate": 3.7996143501025245e-05,
      "loss": 1.5291,
      "step": 6412
    },
    {
      "epoch": 1.1552353073632065,
      "grad_norm": 1.3827592134475708,
      "learning_rate": 3.798239662847323e-05,
      "loss": 1.2069,
      "step": 6413
    },
    {
      "epoch": 1.1554154469714029,
      "grad_norm": 1.3769804239273071,
      "learning_rate": 3.796865071996751e-05,
      "loss": 1.2456,
      "step": 6414
    },
    {
      "epoch": 1.1555955865795993,
      "grad_norm": 1.2939969301223755,
      "learning_rate": 3.795490577661076e-05,
      "loss": 1.3427,
      "step": 6415
    },
    {
      "epoch": 1.1557757261877954,
      "grad_norm": 1.6286332607269287,
      "learning_rate": 3.794116179950561e-05,
      "loss": 1.4717,
      "step": 6416
    },
    {
      "epoch": 1.1559558657959919,
      "grad_norm": 1.4103859663009644,
      "learning_rate": 3.792741878975458e-05,
      "loss": 1.3882,
      "step": 6417
    },
    {
      "epoch": 1.1561360054041883,
      "grad_norm": 1.2872146368026733,
      "learning_rate": 3.791367674846015e-05,
      "loss": 1.3637,
      "step": 6418
    },
    {
      "epoch": 1.1563161450123847,
      "grad_norm": 1.3618946075439453,
      "learning_rate": 3.789993567672467e-05,
      "loss": 1.4437,
      "step": 6419
    },
    {
      "epoch": 1.156496284620581,
      "grad_norm": 1.3749947547912598,
      "learning_rate": 3.788619557565045e-05,
      "loss": 1.3932,
      "step": 6420
    },
    {
      "epoch": 1.1566764242287773,
      "grad_norm": 1.3605971336364746,
      "learning_rate": 3.7872456446339736e-05,
      "loss": 1.3635,
      "step": 6421
    },
    {
      "epoch": 1.1568565638369737,
      "grad_norm": 1.3279310464859009,
      "learning_rate": 3.785871828989465e-05,
      "loss": 1.356,
      "step": 6422
    },
    {
      "epoch": 1.15703670344517,
      "grad_norm": 1.3740218877792358,
      "learning_rate": 3.784498110741727e-05,
      "loss": 1.4392,
      "step": 6423
    },
    {
      "epoch": 1.1572168430533663,
      "grad_norm": 1.3229327201843262,
      "learning_rate": 3.7831244900009564e-05,
      "loss": 1.2514,
      "step": 6424
    },
    {
      "epoch": 1.1573969826615627,
      "grad_norm": 1.4733837842941284,
      "learning_rate": 3.78175096687735e-05,
      "loss": 1.4161,
      "step": 6425
    },
    {
      "epoch": 1.157577122269759,
      "grad_norm": 1.3571406602859497,
      "learning_rate": 3.780377541481087e-05,
      "loss": 1.1683,
      "step": 6426
    },
    {
      "epoch": 1.1577572618779555,
      "grad_norm": 1.405378818511963,
      "learning_rate": 3.779004213922343e-05,
      "loss": 1.4102,
      "step": 6427
    },
    {
      "epoch": 1.1579374014861519,
      "grad_norm": 1.3658502101898193,
      "learning_rate": 3.777630984311287e-05,
      "loss": 1.3832,
      "step": 6428
    },
    {
      "epoch": 1.158117541094348,
      "grad_norm": 1.374346375465393,
      "learning_rate": 3.776257852758074e-05,
      "loss": 1.3915,
      "step": 6429
    },
    {
      "epoch": 1.1582976807025445,
      "grad_norm": 1.5733004808425903,
      "learning_rate": 3.7748848193728644e-05,
      "loss": 1.6897,
      "step": 6430
    },
    {
      "epoch": 1.1584778203107409,
      "grad_norm": 1.362378716468811,
      "learning_rate": 3.7735118842657974e-05,
      "loss": 1.4484,
      "step": 6431
    },
    {
      "epoch": 1.158657959918937,
      "grad_norm": 1.5261523723602295,
      "learning_rate": 3.7721390475470094e-05,
      "loss": 1.548,
      "step": 6432
    },
    {
      "epoch": 1.1588380995271335,
      "grad_norm": 1.442342758178711,
      "learning_rate": 3.770766309326628e-05,
      "loss": 1.427,
      "step": 6433
    },
    {
      "epoch": 1.1590182391353299,
      "grad_norm": 1.503861665725708,
      "learning_rate": 3.7693936697147737e-05,
      "loss": 1.587,
      "step": 6434
    },
    {
      "epoch": 1.1591983787435263,
      "grad_norm": 1.4095183610916138,
      "learning_rate": 3.7680211288215614e-05,
      "loss": 1.3114,
      "step": 6435
    },
    {
      "epoch": 1.1593785183517227,
      "grad_norm": 1.5870410203933716,
      "learning_rate": 3.766648686757094e-05,
      "loss": 1.4534,
      "step": 6436
    },
    {
      "epoch": 1.1595586579599189,
      "grad_norm": 1.478755235671997,
      "learning_rate": 3.7652763436314686e-05,
      "loss": 1.4959,
      "step": 6437
    },
    {
      "epoch": 1.1597387975681153,
      "grad_norm": 1.4625827074050903,
      "learning_rate": 3.7639040995547706e-05,
      "loss": 1.3592,
      "step": 6438
    },
    {
      "epoch": 1.1599189371763117,
      "grad_norm": 1.5260828733444214,
      "learning_rate": 3.762531954637086e-05,
      "loss": 1.5229,
      "step": 6439
    },
    {
      "epoch": 1.160099076784508,
      "grad_norm": 1.5431325435638428,
      "learning_rate": 3.761159908988486e-05,
      "loss": 1.4521,
      "step": 6440
    },
    {
      "epoch": 1.1602792163927043,
      "grad_norm": 1.4463703632354736,
      "learning_rate": 3.7597879627190334e-05,
      "loss": 1.3324,
      "step": 6441
    },
    {
      "epoch": 1.1604593560009007,
      "grad_norm": 1.5385984182357788,
      "learning_rate": 3.7584161159387875e-05,
      "loss": 1.7039,
      "step": 6442
    },
    {
      "epoch": 1.160639495609097,
      "grad_norm": 1.502137303352356,
      "learning_rate": 3.7570443687577944e-05,
      "loss": 1.4978,
      "step": 6443
    },
    {
      "epoch": 1.1608196352172935,
      "grad_norm": 1.4563536643981934,
      "learning_rate": 3.755672721286098e-05,
      "loss": 1.5419,
      "step": 6444
    },
    {
      "epoch": 1.1609997748254897,
      "grad_norm": 1.407694697380066,
      "learning_rate": 3.7543011736337294e-05,
      "loss": 1.3218,
      "step": 6445
    },
    {
      "epoch": 1.161179914433686,
      "grad_norm": 1.5893535614013672,
      "learning_rate": 3.752929725910714e-05,
      "loss": 1.7216,
      "step": 6446
    },
    {
      "epoch": 1.1613600540418825,
      "grad_norm": 1.3737194538116455,
      "learning_rate": 3.751558378227067e-05,
      "loss": 1.2127,
      "step": 6447
    },
    {
      "epoch": 1.1615401936500789,
      "grad_norm": 1.387589693069458,
      "learning_rate": 3.750187130692801e-05,
      "loss": 1.3567,
      "step": 6448
    },
    {
      "epoch": 1.161720333258275,
      "grad_norm": 1.58565354347229,
      "learning_rate": 3.748815983417914e-05,
      "loss": 1.3706,
      "step": 6449
    },
    {
      "epoch": 1.1619004728664715,
      "grad_norm": 1.4688928127288818,
      "learning_rate": 3.747444936512399e-05,
      "loss": 1.2583,
      "step": 6450
    },
    {
      "epoch": 1.1620806124746679,
      "grad_norm": 1.393837332725525,
      "learning_rate": 3.746073990086242e-05,
      "loss": 1.2619,
      "step": 6451
    },
    {
      "epoch": 1.1622607520828643,
      "grad_norm": 1.3083038330078125,
      "learning_rate": 3.744703144249416e-05,
      "loss": 1.79,
      "step": 6452
    },
    {
      "epoch": 1.1624408916910607,
      "grad_norm": 1.3120161294937134,
      "learning_rate": 3.743332399111895e-05,
      "loss": 1.8202,
      "step": 6453
    },
    {
      "epoch": 1.1626210312992569,
      "grad_norm": 1.4391287565231323,
      "learning_rate": 3.741961754783636e-05,
      "loss": 2.0526,
      "step": 6454
    },
    {
      "epoch": 1.1628011709074533,
      "grad_norm": 1.375231146812439,
      "learning_rate": 3.7405912113745933e-05,
      "loss": 1.9177,
      "step": 6455
    },
    {
      "epoch": 1.1629813105156497,
      "grad_norm": 1.3305304050445557,
      "learning_rate": 3.739220768994709e-05,
      "loss": 1.6191,
      "step": 6456
    },
    {
      "epoch": 1.1631614501238459,
      "grad_norm": 1.34432852268219,
      "learning_rate": 3.737850427753921e-05,
      "loss": 1.7354,
      "step": 6457
    },
    {
      "epoch": 1.1633415897320423,
      "grad_norm": 1.469430923461914,
      "learning_rate": 3.7364801877621565e-05,
      "loss": 1.7996,
      "step": 6458
    },
    {
      "epoch": 1.1635217293402387,
      "grad_norm": 1.4734642505645752,
      "learning_rate": 3.735110049129336e-05,
      "loss": 1.8147,
      "step": 6459
    },
    {
      "epoch": 1.163701868948435,
      "grad_norm": 1.6314880847930908,
      "learning_rate": 3.733740011965372e-05,
      "loss": 1.5816,
      "step": 6460
    },
    {
      "epoch": 1.1638820085566315,
      "grad_norm": 1.8882904052734375,
      "learning_rate": 3.732370076380165e-05,
      "loss": 2.0899,
      "step": 6461
    },
    {
      "epoch": 1.1640621481648277,
      "grad_norm": 1.7348037958145142,
      "learning_rate": 3.731000242483615e-05,
      "loss": 1.6417,
      "step": 6462
    },
    {
      "epoch": 1.164242287773024,
      "grad_norm": 1.3300294876098633,
      "learning_rate": 3.7296305103856076e-05,
      "loss": 1.3842,
      "step": 6463
    },
    {
      "epoch": 1.1644224273812205,
      "grad_norm": 1.3609557151794434,
      "learning_rate": 3.7282608801960227e-05,
      "loss": 1.6424,
      "step": 6464
    },
    {
      "epoch": 1.1646025669894169,
      "grad_norm": 1.4475817680358887,
      "learning_rate": 3.72689135202473e-05,
      "loss": 1.6811,
      "step": 6465
    },
    {
      "epoch": 1.164782706597613,
      "grad_norm": 1.361619472503662,
      "learning_rate": 3.72552192598159e-05,
      "loss": 1.2482,
      "step": 6466
    },
    {
      "epoch": 1.1649628462058095,
      "grad_norm": 1.3050634860992432,
      "learning_rate": 3.724152602176463e-05,
      "loss": 1.2937,
      "step": 6467
    },
    {
      "epoch": 1.1651429858140059,
      "grad_norm": 1.3399953842163086,
      "learning_rate": 3.7227833807191926e-05,
      "loss": 1.1783,
      "step": 6468
    },
    {
      "epoch": 1.1653231254222023,
      "grad_norm": 1.4773786067962646,
      "learning_rate": 3.7214142617196176e-05,
      "loss": 1.2402,
      "step": 6469
    },
    {
      "epoch": 1.1655032650303985,
      "grad_norm": 1.4315987825393677,
      "learning_rate": 3.720045245287569e-05,
      "loss": 1.4231,
      "step": 6470
    },
    {
      "epoch": 1.1656834046385949,
      "grad_norm": 1.4008889198303223,
      "learning_rate": 3.7186763315328635e-05,
      "loss": 1.3352,
      "step": 6471
    },
    {
      "epoch": 1.1658635442467913,
      "grad_norm": 1.4410016536712646,
      "learning_rate": 3.7173075205653206e-05,
      "loss": 1.5302,
      "step": 6472
    },
    {
      "epoch": 1.1660436838549877,
      "grad_norm": 1.399985909461975,
      "learning_rate": 3.7159388124947445e-05,
      "loss": 1.4895,
      "step": 6473
    },
    {
      "epoch": 1.1662238234631839,
      "grad_norm": 1.3843098878860474,
      "learning_rate": 3.714570207430931e-05,
      "loss": 1.4967,
      "step": 6474
    },
    {
      "epoch": 1.1664039630713803,
      "grad_norm": 1.3149001598358154,
      "learning_rate": 3.713201705483668e-05,
      "loss": 1.304,
      "step": 6475
    },
    {
      "epoch": 1.1665841026795767,
      "grad_norm": 1.4783203601837158,
      "learning_rate": 3.7118333067627385e-05,
      "loss": 1.5237,
      "step": 6476
    },
    {
      "epoch": 1.166764242287773,
      "grad_norm": 1.3466395139694214,
      "learning_rate": 3.7104650113779145e-05,
      "loss": 1.46,
      "step": 6477
    },
    {
      "epoch": 1.1669443818959695,
      "grad_norm": 1.4924614429473877,
      "learning_rate": 3.709096819438958e-05,
      "loss": 1.3535,
      "step": 6478
    },
    {
      "epoch": 1.1671245215041657,
      "grad_norm": 1.5412014722824097,
      "learning_rate": 3.707728731055627e-05,
      "loss": 1.5989,
      "step": 6479
    },
    {
      "epoch": 1.167304661112362,
      "grad_norm": 1.3754088878631592,
      "learning_rate": 3.706360746337666e-05,
      "loss": 1.3466,
      "step": 6480
    },
    {
      "epoch": 1.1674848007205585,
      "grad_norm": 1.4865368604660034,
      "learning_rate": 3.704992865394817e-05,
      "loss": 1.3704,
      "step": 6481
    },
    {
      "epoch": 1.1676649403287547,
      "grad_norm": 1.5877918004989624,
      "learning_rate": 3.703625088336809e-05,
      "loss": 1.5244,
      "step": 6482
    },
    {
      "epoch": 1.167845079936951,
      "grad_norm": 1.6087439060211182,
      "learning_rate": 3.702257415273366e-05,
      "loss": 1.8559,
      "step": 6483
    },
    {
      "epoch": 1.1680252195451475,
      "grad_norm": 1.4541953802108765,
      "learning_rate": 3.700889846314201e-05,
      "loss": 1.363,
      "step": 6484
    },
    {
      "epoch": 1.168205359153344,
      "grad_norm": 1.4501092433929443,
      "learning_rate": 3.6995223815690184e-05,
      "loss": 1.5096,
      "step": 6485
    },
    {
      "epoch": 1.1683854987615403,
      "grad_norm": 1.5299330949783325,
      "learning_rate": 3.698155021147518e-05,
      "loss": 1.6382,
      "step": 6486
    },
    {
      "epoch": 1.1685656383697365,
      "grad_norm": 1.4811174869537354,
      "learning_rate": 3.696787765159388e-05,
      "loss": 1.4464,
      "step": 6487
    },
    {
      "epoch": 1.1687457779779329,
      "grad_norm": 1.4445598125457764,
      "learning_rate": 3.695420613714308e-05,
      "loss": 1.3327,
      "step": 6488
    },
    {
      "epoch": 1.1689259175861293,
      "grad_norm": 1.3719395399093628,
      "learning_rate": 3.694053566921949e-05,
      "loss": 1.3368,
      "step": 6489
    },
    {
      "epoch": 1.1691060571943257,
      "grad_norm": 1.5457484722137451,
      "learning_rate": 3.69268662489198e-05,
      "loss": 1.5565,
      "step": 6490
    },
    {
      "epoch": 1.1692861968025219,
      "grad_norm": 1.4242305755615234,
      "learning_rate": 3.691319787734052e-05,
      "loss": 1.2629,
      "step": 6491
    },
    {
      "epoch": 1.1694663364107183,
      "grad_norm": 1.4743695259094238,
      "learning_rate": 3.689953055557813e-05,
      "loss": 1.4033,
      "step": 6492
    },
    {
      "epoch": 1.1696464760189147,
      "grad_norm": 1.5519685745239258,
      "learning_rate": 3.688586428472902e-05,
      "loss": 1.4642,
      "step": 6493
    },
    {
      "epoch": 1.169826615627111,
      "grad_norm": 1.5706673860549927,
      "learning_rate": 3.687219906588947e-05,
      "loss": 1.6092,
      "step": 6494
    },
    {
      "epoch": 1.1700067552353073,
      "grad_norm": 1.5480272769927979,
      "learning_rate": 3.6858534900155736e-05,
      "loss": 1.3821,
      "step": 6495
    },
    {
      "epoch": 1.1701868948435037,
      "grad_norm": 1.4981380701065063,
      "learning_rate": 3.6844871788623945e-05,
      "loss": 1.5754,
      "step": 6496
    },
    {
      "epoch": 1.1703670344517,
      "grad_norm": 1.4690871238708496,
      "learning_rate": 3.6831209732390126e-05,
      "loss": 1.3932,
      "step": 6497
    },
    {
      "epoch": 1.1705471740598965,
      "grad_norm": 1.5345977544784546,
      "learning_rate": 3.681754873255023e-05,
      "loss": 1.4397,
      "step": 6498
    },
    {
      "epoch": 1.1707273136680927,
      "grad_norm": 1.3727781772613525,
      "learning_rate": 3.680388879020015e-05,
      "loss": 1.215,
      "step": 6499
    },
    {
      "epoch": 1.170907453276289,
      "grad_norm": 1.418428659439087,
      "learning_rate": 3.6790229906435705e-05,
      "loss": 1.4243,
      "step": 6500
    },
    {
      "epoch": 1.1710875928844855,
      "grad_norm": 1.4023103713989258,
      "learning_rate": 3.677657208235257e-05,
      "loss": 1.2126,
      "step": 6501
    },
    {
      "epoch": 1.171267732492682,
      "grad_norm": 1.33457612991333,
      "learning_rate": 3.67629153190464e-05,
      "loss": 1.8205,
      "step": 6502
    },
    {
      "epoch": 1.171447872100878,
      "grad_norm": 1.3318908214569092,
      "learning_rate": 3.674925961761268e-05,
      "loss": 1.7812,
      "step": 6503
    },
    {
      "epoch": 1.1716280117090745,
      "grad_norm": 1.2840074300765991,
      "learning_rate": 3.673560497914692e-05,
      "loss": 1.6893,
      "step": 6504
    },
    {
      "epoch": 1.171808151317271,
      "grad_norm": 1.3347468376159668,
      "learning_rate": 3.672195140474447e-05,
      "loss": 1.6688,
      "step": 6505
    },
    {
      "epoch": 1.1719882909254673,
      "grad_norm": 1.3650994300842285,
      "learning_rate": 3.6708298895500616e-05,
      "loss": 1.9315,
      "step": 6506
    },
    {
      "epoch": 1.1721684305336635,
      "grad_norm": 1.4201208353042603,
      "learning_rate": 3.669464745251054e-05,
      "loss": 1.9089,
      "step": 6507
    },
    {
      "epoch": 1.17234857014186,
      "grad_norm": 1.414838433265686,
      "learning_rate": 3.668099707686936e-05,
      "loss": 1.8584,
      "step": 6508
    },
    {
      "epoch": 1.1725287097500563,
      "grad_norm": 1.4731675386428833,
      "learning_rate": 3.666734776967212e-05,
      "loss": 1.9729,
      "step": 6509
    },
    {
      "epoch": 1.1727088493582527,
      "grad_norm": 1.6170134544372559,
      "learning_rate": 3.6653699532013745e-05,
      "loss": 1.7726,
      "step": 6510
    },
    {
      "epoch": 1.172888988966449,
      "grad_norm": 1.7776848077774048,
      "learning_rate": 3.6640052364989096e-05,
      "loss": 1.8517,
      "step": 6511
    },
    {
      "epoch": 1.1730691285746453,
      "grad_norm": 1.449196696281433,
      "learning_rate": 3.6626406269692934e-05,
      "loss": 1.7625,
      "step": 6512
    },
    {
      "epoch": 1.1732492681828417,
      "grad_norm": 1.324354887008667,
      "learning_rate": 3.661276124721994e-05,
      "loss": 1.3314,
      "step": 6513
    },
    {
      "epoch": 1.173429407791038,
      "grad_norm": 1.3754370212554932,
      "learning_rate": 3.6599117298664725e-05,
      "loss": 1.4978,
      "step": 6514
    },
    {
      "epoch": 1.1736095473992343,
      "grad_norm": 1.4485403299331665,
      "learning_rate": 3.658547442512179e-05,
      "loss": 1.4981,
      "step": 6515
    },
    {
      "epoch": 1.1737896870074307,
      "grad_norm": 1.3009345531463623,
      "learning_rate": 3.657183262768557e-05,
      "loss": 1.3513,
      "step": 6516
    },
    {
      "epoch": 1.173969826615627,
      "grad_norm": 1.4406200647354126,
      "learning_rate": 3.6558191907450385e-05,
      "loss": 1.3956,
      "step": 6517
    },
    {
      "epoch": 1.1741499662238235,
      "grad_norm": 1.3275244235992432,
      "learning_rate": 3.6544552265510505e-05,
      "loss": 1.3103,
      "step": 6518
    },
    {
      "epoch": 1.17433010583202,
      "grad_norm": 1.371840476989746,
      "learning_rate": 3.653091370296009e-05,
      "loss": 1.547,
      "step": 6519
    },
    {
      "epoch": 1.174510245440216,
      "grad_norm": 1.2835233211517334,
      "learning_rate": 3.6517276220893214e-05,
      "loss": 1.4501,
      "step": 6520
    },
    {
      "epoch": 1.1746903850484125,
      "grad_norm": 1.268537998199463,
      "learning_rate": 3.650363982040388e-05,
      "loss": 1.3726,
      "step": 6521
    },
    {
      "epoch": 1.174870524656609,
      "grad_norm": 1.2959752082824707,
      "learning_rate": 3.6490004502585965e-05,
      "loss": 1.3328,
      "step": 6522
    },
    {
      "epoch": 1.1750506642648053,
      "grad_norm": 1.4207144975662231,
      "learning_rate": 3.647637026853333e-05,
      "loss": 1.3504,
      "step": 6523
    },
    {
      "epoch": 1.1752308038730015,
      "grad_norm": 1.4284688234329224,
      "learning_rate": 3.646273711933967e-05,
      "loss": 1.4336,
      "step": 6524
    },
    {
      "epoch": 1.175410943481198,
      "grad_norm": 1.2378349304199219,
      "learning_rate": 3.644910505609866e-05,
      "loss": 1.122,
      "step": 6525
    },
    {
      "epoch": 1.1755910830893943,
      "grad_norm": 1.335356593132019,
      "learning_rate": 3.643547407990384e-05,
      "loss": 1.3188,
      "step": 6526
    },
    {
      "epoch": 1.1757712226975907,
      "grad_norm": 1.39003324508667,
      "learning_rate": 3.642184419184865e-05,
      "loss": 1.5439,
      "step": 6527
    },
    {
      "epoch": 1.175951362305787,
      "grad_norm": 1.451756477355957,
      "learning_rate": 3.640821539302654e-05,
      "loss": 1.4459,
      "step": 6528
    },
    {
      "epoch": 1.1761315019139833,
      "grad_norm": 1.557690978050232,
      "learning_rate": 3.639458768453077e-05,
      "loss": 1.5089,
      "step": 6529
    },
    {
      "epoch": 1.1763116415221797,
      "grad_norm": 1.441338062286377,
      "learning_rate": 3.638096106745454e-05,
      "loss": 1.299,
      "step": 6530
    },
    {
      "epoch": 1.1764917811303761,
      "grad_norm": 1.384159803390503,
      "learning_rate": 3.636733554289097e-05,
      "loss": 1.3679,
      "step": 6531
    },
    {
      "epoch": 1.1766719207385723,
      "grad_norm": 1.4989361763000488,
      "learning_rate": 3.635371111193311e-05,
      "loss": 1.522,
      "step": 6532
    },
    {
      "epoch": 1.1768520603467687,
      "grad_norm": 1.581574559211731,
      "learning_rate": 3.6340087775673903e-05,
      "loss": 1.474,
      "step": 6533
    },
    {
      "epoch": 1.177032199954965,
      "grad_norm": 1.382921814918518,
      "learning_rate": 3.63264655352062e-05,
      "loss": 1.5329,
      "step": 6534
    },
    {
      "epoch": 1.1772123395631615,
      "grad_norm": 1.5133802890777588,
      "learning_rate": 3.631284439162278e-05,
      "loss": 1.4656,
      "step": 6535
    },
    {
      "epoch": 1.177392479171358,
      "grad_norm": 1.3116607666015625,
      "learning_rate": 3.629922434601629e-05,
      "loss": 1.2995,
      "step": 6536
    },
    {
      "epoch": 1.177572618779554,
      "grad_norm": 1.3999848365783691,
      "learning_rate": 3.628560539947937e-05,
      "loss": 1.2268,
      "step": 6537
    },
    {
      "epoch": 1.1777527583877505,
      "grad_norm": 1.499196171760559,
      "learning_rate": 3.6271987553104504e-05,
      "loss": 1.3645,
      "step": 6538
    },
    {
      "epoch": 1.177932897995947,
      "grad_norm": 1.4488658905029297,
      "learning_rate": 3.625837080798411e-05,
      "loss": 1.5458,
      "step": 6539
    },
    {
      "epoch": 1.178113037604143,
      "grad_norm": 1.327940583229065,
      "learning_rate": 3.624475516521051e-05,
      "loss": 1.2288,
      "step": 6540
    },
    {
      "epoch": 1.1782931772123395,
      "grad_norm": 1.5767016410827637,
      "learning_rate": 3.623114062587595e-05,
      "loss": 1.4652,
      "step": 6541
    },
    {
      "epoch": 1.178473316820536,
      "grad_norm": 1.592172622680664,
      "learning_rate": 3.6217527191072585e-05,
      "loss": 1.7449,
      "step": 6542
    },
    {
      "epoch": 1.1786534564287323,
      "grad_norm": 1.4359644651412964,
      "learning_rate": 3.620391486189248e-05,
      "loss": 1.4,
      "step": 6543
    },
    {
      "epoch": 1.1788335960369287,
      "grad_norm": 1.545168161392212,
      "learning_rate": 3.6190303639427604e-05,
      "loss": 1.4624,
      "step": 6544
    },
    {
      "epoch": 1.179013735645125,
      "grad_norm": 1.4305164813995361,
      "learning_rate": 3.617669352476983e-05,
      "loss": 1.4609,
      "step": 6545
    },
    {
      "epoch": 1.1791938752533213,
      "grad_norm": 1.4097996950149536,
      "learning_rate": 3.6163084519010984e-05,
      "loss": 1.2905,
      "step": 6546
    },
    {
      "epoch": 1.1793740148615177,
      "grad_norm": 1.4933857917785645,
      "learning_rate": 3.6149476623242754e-05,
      "loss": 1.2906,
      "step": 6547
    },
    {
      "epoch": 1.1795541544697141,
      "grad_norm": 1.5967057943344116,
      "learning_rate": 3.6135869838556766e-05,
      "loss": 1.4725,
      "step": 6548
    },
    {
      "epoch": 1.1797342940779103,
      "grad_norm": 1.6981664896011353,
      "learning_rate": 3.612226416604455e-05,
      "loss": 1.7249,
      "step": 6549
    },
    {
      "epoch": 1.1799144336861067,
      "grad_norm": 1.554762363433838,
      "learning_rate": 3.610865960679752e-05,
      "loss": 1.2583,
      "step": 6550
    },
    {
      "epoch": 1.1800945732943031,
      "grad_norm": 1.509871244430542,
      "learning_rate": 3.6095056161907074e-05,
      "loss": 1.1946,
      "step": 6551
    },
    {
      "epoch": 1.1802747129024995,
      "grad_norm": 1.244280219078064,
      "learning_rate": 3.608145383246444e-05,
      "loss": 1.7441,
      "step": 6552
    },
    {
      "epoch": 1.1804548525106957,
      "grad_norm": 1.366241216659546,
      "learning_rate": 3.606785261956081e-05,
      "loss": 1.9998,
      "step": 6553
    },
    {
      "epoch": 1.1806349921188921,
      "grad_norm": 1.330600619316101,
      "learning_rate": 3.605425252428725e-05,
      "loss": 1.9065,
      "step": 6554
    },
    {
      "epoch": 1.1808151317270885,
      "grad_norm": 1.2577922344207764,
      "learning_rate": 3.604065354773475e-05,
      "loss": 1.893,
      "step": 6555
    },
    {
      "epoch": 1.180995271335285,
      "grad_norm": 1.3520675897598267,
      "learning_rate": 3.6027055690994235e-05,
      "loss": 2.0387,
      "step": 6556
    },
    {
      "epoch": 1.181175410943481,
      "grad_norm": 1.5209952592849731,
      "learning_rate": 3.601345895515651e-05,
      "loss": 1.8802,
      "step": 6557
    },
    {
      "epoch": 1.1813555505516775,
      "grad_norm": 1.3552844524383545,
      "learning_rate": 3.599986334131229e-05,
      "loss": 1.4301,
      "step": 6558
    },
    {
      "epoch": 1.181535690159874,
      "grad_norm": 1.4343005418777466,
      "learning_rate": 3.598626885055219e-05,
      "loss": 1.8323,
      "step": 6559
    },
    {
      "epoch": 1.1817158297680703,
      "grad_norm": 1.4352056980133057,
      "learning_rate": 3.59726754839668e-05,
      "loss": 1.8075,
      "step": 6560
    },
    {
      "epoch": 1.1818959693762667,
      "grad_norm": 1.5641226768493652,
      "learning_rate": 3.595908324264656e-05,
      "loss": 1.7918,
      "step": 6561
    },
    {
      "epoch": 1.182076108984463,
      "grad_norm": 1.930543065071106,
      "learning_rate": 3.5945492127681816e-05,
      "loss": 2.145,
      "step": 6562
    },
    {
      "epoch": 1.1822562485926593,
      "grad_norm": 1.615303635597229,
      "learning_rate": 3.593190214016285e-05,
      "loss": 1.8811,
      "step": 6563
    },
    {
      "epoch": 1.1824363882008557,
      "grad_norm": 1.411547303199768,
      "learning_rate": 3.591831328117982e-05,
      "loss": 1.4036,
      "step": 6564
    },
    {
      "epoch": 1.182616527809052,
      "grad_norm": 1.3096319437026978,
      "learning_rate": 3.590472555182285e-05,
      "loss": 1.3405,
      "step": 6565
    },
    {
      "epoch": 1.1827966674172483,
      "grad_norm": 1.263176679611206,
      "learning_rate": 3.589113895318194e-05,
      "loss": 1.252,
      "step": 6566
    },
    {
      "epoch": 1.1829768070254447,
      "grad_norm": 1.35726797580719,
      "learning_rate": 3.5877553486347e-05,
      "loss": 1.5132,
      "step": 6567
    },
    {
      "epoch": 1.1831569466336411,
      "grad_norm": 1.3140900135040283,
      "learning_rate": 3.586396915240781e-05,
      "loss": 1.3644,
      "step": 6568
    },
    {
      "epoch": 1.1833370862418375,
      "grad_norm": 1.448294758796692,
      "learning_rate": 3.585038595245415e-05,
      "loss": 1.4053,
      "step": 6569
    },
    {
      "epoch": 1.1835172258500337,
      "grad_norm": 1.3514827489852905,
      "learning_rate": 3.5836803887575634e-05,
      "loss": 1.4455,
      "step": 6570
    },
    {
      "epoch": 1.1836973654582301,
      "grad_norm": 1.4190112352371216,
      "learning_rate": 3.5823222958861814e-05,
      "loss": 1.4431,
      "step": 6571
    },
    {
      "epoch": 1.1838775050664265,
      "grad_norm": 1.328613519668579,
      "learning_rate": 3.5809643167402145e-05,
      "loss": 1.4305,
      "step": 6572
    },
    {
      "epoch": 1.1840576446746227,
      "grad_norm": 1.4426491260528564,
      "learning_rate": 3.579606451428598e-05,
      "loss": 1.5225,
      "step": 6573
    },
    {
      "epoch": 1.1842377842828191,
      "grad_norm": 1.3465397357940674,
      "learning_rate": 3.57824870006026e-05,
      "loss": 1.3794,
      "step": 6574
    },
    {
      "epoch": 1.1844179238910155,
      "grad_norm": 1.4426008462905884,
      "learning_rate": 3.57689106274412e-05,
      "loss": 1.5571,
      "step": 6575
    },
    {
      "epoch": 1.184598063499212,
      "grad_norm": 1.393587350845337,
      "learning_rate": 3.5755335395890855e-05,
      "loss": 1.3809,
      "step": 6576
    },
    {
      "epoch": 1.1847782031074083,
      "grad_norm": 1.320299744606018,
      "learning_rate": 3.5741761307040564e-05,
      "loss": 1.3174,
      "step": 6577
    },
    {
      "epoch": 1.1849583427156045,
      "grad_norm": 1.3964899778366089,
      "learning_rate": 3.572818836197923e-05,
      "loss": 1.4343,
      "step": 6578
    },
    {
      "epoch": 1.185138482323801,
      "grad_norm": 1.4703927040100098,
      "learning_rate": 3.5714616561795675e-05,
      "loss": 1.4531,
      "step": 6579
    },
    {
      "epoch": 1.1853186219319973,
      "grad_norm": 1.4955027103424072,
      "learning_rate": 3.570104590757863e-05,
      "loss": 1.3522,
      "step": 6580
    },
    {
      "epoch": 1.1854987615401937,
      "grad_norm": 1.3167777061462402,
      "learning_rate": 3.5687476400416714e-05,
      "loss": 1.3425,
      "step": 6581
    },
    {
      "epoch": 1.18567890114839,
      "grad_norm": 1.4521174430847168,
      "learning_rate": 3.5673908041398464e-05,
      "loss": 1.2777,
      "step": 6582
    },
    {
      "epoch": 1.1858590407565863,
      "grad_norm": 1.509108543395996,
      "learning_rate": 3.566034083161234e-05,
      "loss": 1.5353,
      "step": 6583
    },
    {
      "epoch": 1.1860391803647827,
      "grad_norm": 1.5208263397216797,
      "learning_rate": 3.5646774772146684e-05,
      "loss": 1.6614,
      "step": 6584
    },
    {
      "epoch": 1.1862193199729791,
      "grad_norm": 1.4845576286315918,
      "learning_rate": 3.5633209864089766e-05,
      "loss": 1.3558,
      "step": 6585
    },
    {
      "epoch": 1.1863994595811753,
      "grad_norm": 1.5247132778167725,
      "learning_rate": 3.561964610852976e-05,
      "loss": 1.3456,
      "step": 6586
    },
    {
      "epoch": 1.1865795991893717,
      "grad_norm": 1.523598551750183,
      "learning_rate": 3.560608350655473e-05,
      "loss": 1.6709,
      "step": 6587
    },
    {
      "epoch": 1.1867597387975681,
      "grad_norm": 1.4654392004013062,
      "learning_rate": 3.5592522059252674e-05,
      "loss": 1.519,
      "step": 6588
    },
    {
      "epoch": 1.1869398784057645,
      "grad_norm": 1.4512569904327393,
      "learning_rate": 3.557896176771149e-05,
      "loss": 1.2494,
      "step": 6589
    },
    {
      "epoch": 1.1871200180139607,
      "grad_norm": 1.5786213874816895,
      "learning_rate": 3.556540263301896e-05,
      "loss": 1.5712,
      "step": 6590
    },
    {
      "epoch": 1.1873001576221571,
      "grad_norm": 1.3032244443893433,
      "learning_rate": 3.55518446562628e-05,
      "loss": 1.2322,
      "step": 6591
    },
    {
      "epoch": 1.1874802972303535,
      "grad_norm": 1.6659175157546997,
      "learning_rate": 3.553828783853061e-05,
      "loss": 1.582,
      "step": 6592
    },
    {
      "epoch": 1.18766043683855,
      "grad_norm": 1.2859901189804077,
      "learning_rate": 3.552473218090995e-05,
      "loss": 1.1052,
      "step": 6593
    },
    {
      "epoch": 1.1878405764467463,
      "grad_norm": 1.5602586269378662,
      "learning_rate": 3.5511177684488216e-05,
      "loss": 1.5606,
      "step": 6594
    },
    {
      "epoch": 1.1880207160549425,
      "grad_norm": 1.5847281217575073,
      "learning_rate": 3.549762435035275e-05,
      "loss": 1.3735,
      "step": 6595
    },
    {
      "epoch": 1.188200855663139,
      "grad_norm": 1.463122844696045,
      "learning_rate": 3.5484072179590776e-05,
      "loss": 1.4242,
      "step": 6596
    },
    {
      "epoch": 1.1883809952713353,
      "grad_norm": 1.4870685338974,
      "learning_rate": 3.5470521173289486e-05,
      "loss": 1.4618,
      "step": 6597
    },
    {
      "epoch": 1.1885611348795315,
      "grad_norm": 1.5126922130584717,
      "learning_rate": 3.54569713325359e-05,
      "loss": 1.4509,
      "step": 6598
    },
    {
      "epoch": 1.188741274487728,
      "grad_norm": 1.701738953590393,
      "learning_rate": 3.544342265841701e-05,
      "loss": 1.4731,
      "step": 6599
    },
    {
      "epoch": 1.1889214140959243,
      "grad_norm": 1.4555048942565918,
      "learning_rate": 3.5429875152019655e-05,
      "loss": 1.275,
      "step": 6600
    },
    {
      "epoch": 1.1891015537041207,
      "grad_norm": 1.3295918703079224,
      "learning_rate": 3.5416328814430586e-05,
      "loss": 1.0307,
      "step": 6601
    },
    {
      "epoch": 1.1892816933123171,
      "grad_norm": 1.3476378917694092,
      "learning_rate": 3.540278364673655e-05,
      "loss": 1.7332,
      "step": 6602
    },
    {
      "epoch": 1.1894618329205133,
      "grad_norm": 1.3140027523040771,
      "learning_rate": 3.538923965002409e-05,
      "loss": 1.9636,
      "step": 6603
    },
    {
      "epoch": 1.1896419725287097,
      "grad_norm": 1.2953075170516968,
      "learning_rate": 3.537569682537971e-05,
      "loss": 1.8666,
      "step": 6604
    },
    {
      "epoch": 1.1898221121369061,
      "grad_norm": 1.4390342235565186,
      "learning_rate": 3.536215517388981e-05,
      "loss": 1.8716,
      "step": 6605
    },
    {
      "epoch": 1.1900022517451025,
      "grad_norm": 1.352335810661316,
      "learning_rate": 3.534861469664068e-05,
      "loss": 1.8955,
      "step": 6606
    },
    {
      "epoch": 1.1901823913532987,
      "grad_norm": 1.4060144424438477,
      "learning_rate": 3.533507539471856e-05,
      "loss": 1.5629,
      "step": 6607
    },
    {
      "epoch": 1.1903625309614951,
      "grad_norm": 1.4898676872253418,
      "learning_rate": 3.532153726920954e-05,
      "loss": 1.9746,
      "step": 6608
    },
    {
      "epoch": 1.1905426705696915,
      "grad_norm": 1.6041971445083618,
      "learning_rate": 3.530800032119965e-05,
      "loss": 1.9493,
      "step": 6609
    },
    {
      "epoch": 1.190722810177888,
      "grad_norm": 1.5663259029388428,
      "learning_rate": 3.529446455177481e-05,
      "loss": 1.8175,
      "step": 6610
    },
    {
      "epoch": 1.1909029497860841,
      "grad_norm": 1.8397724628448486,
      "learning_rate": 3.5280929962020884e-05,
      "loss": 2.1634,
      "step": 6611
    },
    {
      "epoch": 1.1910830893942805,
      "grad_norm": 1.4266645908355713,
      "learning_rate": 3.5267396553023565e-05,
      "loss": 1.5405,
      "step": 6612
    },
    {
      "epoch": 1.191263229002477,
      "grad_norm": 1.4743822813034058,
      "learning_rate": 3.5253864325868536e-05,
      "loss": 1.7425,
      "step": 6613
    },
    {
      "epoch": 1.1914433686106733,
      "grad_norm": 1.421815037727356,
      "learning_rate": 3.524033328164132e-05,
      "loss": 1.5163,
      "step": 6614
    },
    {
      "epoch": 1.1916235082188695,
      "grad_norm": 1.3058090209960938,
      "learning_rate": 3.5226803421427366e-05,
      "loss": 1.3694,
      "step": 6615
    },
    {
      "epoch": 1.191803647827066,
      "grad_norm": 1.467496395111084,
      "learning_rate": 3.521327474631206e-05,
      "loss": 1.6727,
      "step": 6616
    },
    {
      "epoch": 1.1919837874352623,
      "grad_norm": 1.3814575672149658,
      "learning_rate": 3.5199747257380654e-05,
      "loss": 1.3561,
      "step": 6617
    },
    {
      "epoch": 1.1921639270434587,
      "grad_norm": 1.4303704500198364,
      "learning_rate": 3.5186220955718306e-05,
      "loss": 1.5032,
      "step": 6618
    },
    {
      "epoch": 1.1923440666516552,
      "grad_norm": 1.372300148010254,
      "learning_rate": 3.517269584241011e-05,
      "loss": 1.606,
      "step": 6619
    },
    {
      "epoch": 1.1925242062598513,
      "grad_norm": 1.3018789291381836,
      "learning_rate": 3.5159171918541015e-05,
      "loss": 1.3698,
      "step": 6620
    },
    {
      "epoch": 1.1927043458680477,
      "grad_norm": 1.3748687505722046,
      "learning_rate": 3.5145649185195926e-05,
      "loss": 1.5707,
      "step": 6621
    },
    {
      "epoch": 1.1928844854762441,
      "grad_norm": 1.305811882019043,
      "learning_rate": 3.5132127643459626e-05,
      "loss": 1.404,
      "step": 6622
    },
    {
      "epoch": 1.1930646250844403,
      "grad_norm": 1.4802151918411255,
      "learning_rate": 3.5118607294416806e-05,
      "loss": 1.4995,
      "step": 6623
    },
    {
      "epoch": 1.1932447646926367,
      "grad_norm": 1.418360710144043,
      "learning_rate": 3.510508813915204e-05,
      "loss": 1.4648,
      "step": 6624
    },
    {
      "epoch": 1.1934249043008331,
      "grad_norm": 1.2332953214645386,
      "learning_rate": 3.5091570178749863e-05,
      "loss": 1.1852,
      "step": 6625
    },
    {
      "epoch": 1.1936050439090296,
      "grad_norm": 1.437515377998352,
      "learning_rate": 3.507805341429469e-05,
      "loss": 1.5245,
      "step": 6626
    },
    {
      "epoch": 1.193785183517226,
      "grad_norm": 1.522303819656372,
      "learning_rate": 3.506453784687078e-05,
      "loss": 1.3664,
      "step": 6627
    },
    {
      "epoch": 1.1939653231254221,
      "grad_norm": 1.4197916984558105,
      "learning_rate": 3.505102347756237e-05,
      "loss": 1.4164,
      "step": 6628
    },
    {
      "epoch": 1.1941454627336185,
      "grad_norm": 1.382589340209961,
      "learning_rate": 3.503751030745356e-05,
      "loss": 1.3275,
      "step": 6629
    },
    {
      "epoch": 1.194325602341815,
      "grad_norm": 1.3801850080490112,
      "learning_rate": 3.502399833762841e-05,
      "loss": 1.2015,
      "step": 6630
    },
    {
      "epoch": 1.1945057419500111,
      "grad_norm": 1.4676933288574219,
      "learning_rate": 3.5010487569170806e-05,
      "loss": 1.7089,
      "step": 6631
    },
    {
      "epoch": 1.1946858815582075,
      "grad_norm": 1.4051316976547241,
      "learning_rate": 3.499697800316461e-05,
      "loss": 1.3133,
      "step": 6632
    },
    {
      "epoch": 1.194866021166404,
      "grad_norm": 1.3830493688583374,
      "learning_rate": 3.498346964069351e-05,
      "loss": 1.1997,
      "step": 6633
    },
    {
      "epoch": 1.1950461607746004,
      "grad_norm": 1.3955798149108887,
      "learning_rate": 3.4969962482841145e-05,
      "loss": 1.3241,
      "step": 6634
    },
    {
      "epoch": 1.1952263003827968,
      "grad_norm": 1.3250913619995117,
      "learning_rate": 3.4956456530691094e-05,
      "loss": 1.3055,
      "step": 6635
    },
    {
      "epoch": 1.195406439990993,
      "grad_norm": 1.4945846796035767,
      "learning_rate": 3.494295178532677e-05,
      "loss": 1.4199,
      "step": 6636
    },
    {
      "epoch": 1.1955865795991893,
      "grad_norm": 1.5242571830749512,
      "learning_rate": 3.492944824783152e-05,
      "loss": 1.4864,
      "step": 6637
    },
    {
      "epoch": 1.1957667192073858,
      "grad_norm": 1.3969535827636719,
      "learning_rate": 3.4915945919288576e-05,
      "loss": 1.2787,
      "step": 6638
    },
    {
      "epoch": 1.1959468588155822,
      "grad_norm": 1.3807302713394165,
      "learning_rate": 3.490244480078112e-05,
      "loss": 1.379,
      "step": 6639
    },
    {
      "epoch": 1.1961269984237783,
      "grad_norm": 1.5274091958999634,
      "learning_rate": 3.488894489339218e-05,
      "loss": 1.621,
      "step": 6640
    },
    {
      "epoch": 1.1963071380319747,
      "grad_norm": 1.5574896335601807,
      "learning_rate": 3.487544619820473e-05,
      "loss": 1.4443,
      "step": 6641
    },
    {
      "epoch": 1.1964872776401712,
      "grad_norm": 1.5202059745788574,
      "learning_rate": 3.4861948716301615e-05,
      "loss": 1.4528,
      "step": 6642
    },
    {
      "epoch": 1.1966674172483676,
      "grad_norm": 1.5719102621078491,
      "learning_rate": 3.484845244876559e-05,
      "loss": 1.2942,
      "step": 6643
    },
    {
      "epoch": 1.1968475568565637,
      "grad_norm": 1.289324402809143,
      "learning_rate": 3.4834957396679336e-05,
      "loss": 1.2822,
      "step": 6644
    },
    {
      "epoch": 1.1970276964647601,
      "grad_norm": 1.6651968955993652,
      "learning_rate": 3.482146356112542e-05,
      "loss": 1.7346,
      "step": 6645
    },
    {
      "epoch": 1.1972078360729566,
      "grad_norm": 1.5542699098587036,
      "learning_rate": 3.48079709431863e-05,
      "loss": 1.5366,
      "step": 6646
    },
    {
      "epoch": 1.197387975681153,
      "grad_norm": 1.4728922843933105,
      "learning_rate": 3.4794479543944355e-05,
      "loss": 1.3818,
      "step": 6647
    },
    {
      "epoch": 1.1975681152893491,
      "grad_norm": 1.511291265487671,
      "learning_rate": 3.4780989364481836e-05,
      "loss": 1.2833,
      "step": 6648
    },
    {
      "epoch": 1.1977482548975456,
      "grad_norm": 1.5342804193496704,
      "learning_rate": 3.476750040588095e-05,
      "loss": 1.485,
      "step": 6649
    },
    {
      "epoch": 1.197928394505742,
      "grad_norm": 1.5278414487838745,
      "learning_rate": 3.475401266922377e-05,
      "loss": 1.3269,
      "step": 6650
    },
    {
      "epoch": 1.1981085341139384,
      "grad_norm": 1.4570335149765015,
      "learning_rate": 3.474052615559226e-05,
      "loss": 1.2613,
      "step": 6651
    },
    {
      "epoch": 1.1982886737221348,
      "grad_norm": 1.3100883960723877,
      "learning_rate": 3.47270408660683e-05,
      "loss": 1.5977,
      "step": 6652
    },
    {
      "epoch": 1.198468813330331,
      "grad_norm": 1.361167311668396,
      "learning_rate": 3.471355680173369e-05,
      "loss": 1.8322,
      "step": 6653
    },
    {
      "epoch": 1.1986489529385274,
      "grad_norm": 1.3418521881103516,
      "learning_rate": 3.470007396367012e-05,
      "loss": 1.8426,
      "step": 6654
    },
    {
      "epoch": 1.1988290925467238,
      "grad_norm": 1.4249584674835205,
      "learning_rate": 3.468659235295916e-05,
      "loss": 2.033,
      "step": 6655
    },
    {
      "epoch": 1.19900923215492,
      "grad_norm": 1.7422164678573608,
      "learning_rate": 3.467311197068229e-05,
      "loss": 1.9195,
      "step": 6656
    },
    {
      "epoch": 1.1991893717631164,
      "grad_norm": 1.4534685611724854,
      "learning_rate": 3.46596328179209e-05,
      "loss": 1.872,
      "step": 6657
    },
    {
      "epoch": 1.1993695113713128,
      "grad_norm": 1.381417155265808,
      "learning_rate": 3.464615489575632e-05,
      "loss": 1.7978,
      "step": 6658
    },
    {
      "epoch": 1.1995496509795092,
      "grad_norm": 1.6449260711669922,
      "learning_rate": 3.463267820526972e-05,
      "loss": 1.695,
      "step": 6659
    },
    {
      "epoch": 1.1997297905877056,
      "grad_norm": 1.7877861261367798,
      "learning_rate": 3.461920274754218e-05,
      "loss": 1.8696,
      "step": 6660
    },
    {
      "epoch": 1.1999099301959018,
      "grad_norm": 1.7133687734603882,
      "learning_rate": 3.4605728523654704e-05,
      "loss": 1.9319,
      "step": 6661
    },
    {
      "epoch": 1.2000900698040982,
      "grad_norm": 1.4877243041992188,
      "learning_rate": 3.459225553468815e-05,
      "loss": 1.5891,
      "step": 6662
    },
    {
      "epoch": 1.2002702094122946,
      "grad_norm": 1.4385182857513428,
      "learning_rate": 3.45787837817234e-05,
      "loss": 1.5529,
      "step": 6663
    },
    {
      "epoch": 1.200450349020491,
      "grad_norm": 1.4236807823181152,
      "learning_rate": 3.45653132658411e-05,
      "loss": 1.5168,
      "step": 6664
    },
    {
      "epoch": 1.2006304886286872,
      "grad_norm": 1.3317898511886597,
      "learning_rate": 3.4551843988121844e-05,
      "loss": 1.4586,
      "step": 6665
    },
    {
      "epoch": 1.2008106282368836,
      "grad_norm": 1.3173550367355347,
      "learning_rate": 3.4538375949646116e-05,
      "loss": 1.4256,
      "step": 6666
    },
    {
      "epoch": 1.20099076784508,
      "grad_norm": 1.3368550539016724,
      "learning_rate": 3.452490915149436e-05,
      "loss": 1.4086,
      "step": 6667
    },
    {
      "epoch": 1.2011709074532764,
      "grad_norm": 1.414510726928711,
      "learning_rate": 3.4511443594746854e-05,
      "loss": 1.3895,
      "step": 6668
    },
    {
      "epoch": 1.2013510470614726,
      "grad_norm": 1.5006206035614014,
      "learning_rate": 3.44979792804838e-05,
      "loss": 1.3912,
      "step": 6669
    },
    {
      "epoch": 1.201531186669669,
      "grad_norm": 1.3819299936294556,
      "learning_rate": 3.448451620978531e-05,
      "loss": 1.5017,
      "step": 6670
    },
    {
      "epoch": 1.2017113262778654,
      "grad_norm": 1.3423086404800415,
      "learning_rate": 3.447105438373134e-05,
      "loss": 1.4041,
      "step": 6671
    },
    {
      "epoch": 1.2018914658860618,
      "grad_norm": 1.47804856300354,
      "learning_rate": 3.445759380340185e-05,
      "loss": 1.5716,
      "step": 6672
    },
    {
      "epoch": 1.202071605494258,
      "grad_norm": 1.4944829940795898,
      "learning_rate": 3.44441344698766e-05,
      "loss": 1.5392,
      "step": 6673
    },
    {
      "epoch": 1.2022517451024544,
      "grad_norm": 1.5020502805709839,
      "learning_rate": 3.443067638423532e-05,
      "loss": 1.3804,
      "step": 6674
    },
    {
      "epoch": 1.2024318847106508,
      "grad_norm": 1.3079289197921753,
      "learning_rate": 3.441721954755757e-05,
      "loss": 1.3554,
      "step": 6675
    },
    {
      "epoch": 1.2026120243188472,
      "grad_norm": 1.3910051584243774,
      "learning_rate": 3.44037639609229e-05,
      "loss": 1.2899,
      "step": 6676
    },
    {
      "epoch": 1.2027921639270436,
      "grad_norm": 1.4919273853302002,
      "learning_rate": 3.4390309625410686e-05,
      "loss": 1.5703,
      "step": 6677
    },
    {
      "epoch": 1.2029723035352398,
      "grad_norm": 1.4217326641082764,
      "learning_rate": 3.437685654210023e-05,
      "loss": 1.3826,
      "step": 6678
    },
    {
      "epoch": 1.2031524431434362,
      "grad_norm": 1.3119813203811646,
      "learning_rate": 3.436340471207074e-05,
      "loss": 1.2576,
      "step": 6679
    },
    {
      "epoch": 1.2033325827516326,
      "grad_norm": 1.313380241394043,
      "learning_rate": 3.434995413640129e-05,
      "loss": 1.2351,
      "step": 6680
    },
    {
      "epoch": 1.2035127223598288,
      "grad_norm": 1.389377474784851,
      "learning_rate": 3.4336504816170905e-05,
      "loss": 1.4679,
      "step": 6681
    },
    {
      "epoch": 1.2036928619680252,
      "grad_norm": 1.4892104864120483,
      "learning_rate": 3.4323056752458476e-05,
      "loss": 1.4898,
      "step": 6682
    },
    {
      "epoch": 1.2038730015762216,
      "grad_norm": 1.5369975566864014,
      "learning_rate": 3.43096099463428e-05,
      "loss": 1.3933,
      "step": 6683
    },
    {
      "epoch": 1.204053141184418,
      "grad_norm": 1.5410183668136597,
      "learning_rate": 3.429616439890258e-05,
      "loss": 1.3769,
      "step": 6684
    },
    {
      "epoch": 1.2042332807926144,
      "grad_norm": 1.416324257850647,
      "learning_rate": 3.428272011121637e-05,
      "loss": 1.1954,
      "step": 6685
    },
    {
      "epoch": 1.2044134204008106,
      "grad_norm": 1.549851417541504,
      "learning_rate": 3.4269277084362724e-05,
      "loss": 1.527,
      "step": 6686
    },
    {
      "epoch": 1.204593560009007,
      "grad_norm": 1.5699172019958496,
      "learning_rate": 3.425583531942e-05,
      "loss": 1.4905,
      "step": 6687
    },
    {
      "epoch": 1.2047736996172034,
      "grad_norm": 1.7205044031143188,
      "learning_rate": 3.42423948174665e-05,
      "loss": 1.8317,
      "step": 6688
    },
    {
      "epoch": 1.2049538392253998,
      "grad_norm": 1.5982245206832886,
      "learning_rate": 3.42289555795804e-05,
      "loss": 1.6165,
      "step": 6689
    },
    {
      "epoch": 1.205133978833596,
      "grad_norm": 1.4603633880615234,
      "learning_rate": 3.421551760683982e-05,
      "loss": 1.2945,
      "step": 6690
    },
    {
      "epoch": 1.2053141184417924,
      "grad_norm": 1.384820818901062,
      "learning_rate": 3.420208090032274e-05,
      "loss": 1.2724,
      "step": 6691
    },
    {
      "epoch": 1.2054942580499888,
      "grad_norm": 1.553351879119873,
      "learning_rate": 3.418864546110701e-05,
      "loss": 1.4566,
      "step": 6692
    },
    {
      "epoch": 1.2056743976581852,
      "grad_norm": 1.5097345113754272,
      "learning_rate": 3.417521129027046e-05,
      "loss": 1.4959,
      "step": 6693
    },
    {
      "epoch": 1.2058545372663814,
      "grad_norm": 1.501902461051941,
      "learning_rate": 3.416177838889072e-05,
      "loss": 1.4165,
      "step": 6694
    },
    {
      "epoch": 1.2060346768745778,
      "grad_norm": 1.6061943769454956,
      "learning_rate": 3.4148346758045426e-05,
      "loss": 1.6726,
      "step": 6695
    },
    {
      "epoch": 1.2062148164827742,
      "grad_norm": 1.3930976390838623,
      "learning_rate": 3.4134916398812034e-05,
      "loss": 1.1994,
      "step": 6696
    },
    {
      "epoch": 1.2063949560909706,
      "grad_norm": 1.3988239765167236,
      "learning_rate": 3.412148731226793e-05,
      "loss": 1.2137,
      "step": 6697
    },
    {
      "epoch": 1.2065750956991668,
      "grad_norm": 1.3684662580490112,
      "learning_rate": 3.4108059499490376e-05,
      "loss": 1.2246,
      "step": 6698
    },
    {
      "epoch": 1.2067552353073632,
      "grad_norm": 1.584137201309204,
      "learning_rate": 3.409463296155652e-05,
      "loss": 1.5362,
      "step": 6699
    },
    {
      "epoch": 1.2069353749155596,
      "grad_norm": 1.3989081382751465,
      "learning_rate": 3.4081207699543485e-05,
      "loss": 1.178,
      "step": 6700
    },
    {
      "epoch": 1.207115514523756,
      "grad_norm": 1.605372667312622,
      "learning_rate": 3.4067783714528215e-05,
      "loss": 1.4927,
      "step": 6701
    },
    {
      "epoch": 1.2072956541319522,
      "grad_norm": 1.2538365125656128,
      "learning_rate": 3.4054361007587576e-05,
      "loss": 1.4983,
      "step": 6702
    },
    {
      "epoch": 1.2074757937401486,
      "grad_norm": 1.2905033826828003,
      "learning_rate": 3.4040939579798316e-05,
      "loss": 1.9806,
      "step": 6703
    },
    {
      "epoch": 1.207655933348345,
      "grad_norm": 1.3866654634475708,
      "learning_rate": 3.402751943223712e-05,
      "loss": 1.9978,
      "step": 6704
    },
    {
      "epoch": 1.2078360729565414,
      "grad_norm": 1.3942734003067017,
      "learning_rate": 3.4014100565980534e-05,
      "loss": 1.6547,
      "step": 6705
    },
    {
      "epoch": 1.2080162125647376,
      "grad_norm": 1.3620405197143555,
      "learning_rate": 3.4000682982105014e-05,
      "loss": 1.7364,
      "step": 6706
    },
    {
      "epoch": 1.208196352172934,
      "grad_norm": 1.4084973335266113,
      "learning_rate": 3.398726668168692e-05,
      "loss": 1.6698,
      "step": 6707
    },
    {
      "epoch": 1.2083764917811304,
      "grad_norm": 1.6366336345672607,
      "learning_rate": 3.397385166580248e-05,
      "loss": 2.0062,
      "step": 6708
    },
    {
      "epoch": 1.2085566313893268,
      "grad_norm": 1.6566872596740723,
      "learning_rate": 3.3960437935527855e-05,
      "loss": 2.061,
      "step": 6709
    },
    {
      "epoch": 1.2087367709975232,
      "grad_norm": 1.6653355360031128,
      "learning_rate": 3.394702549193909e-05,
      "loss": 1.7207,
      "step": 6710
    },
    {
      "epoch": 1.2089169106057194,
      "grad_norm": 1.474739909172058,
      "learning_rate": 3.393361433611213e-05,
      "loss": 1.5235,
      "step": 6711
    },
    {
      "epoch": 1.2090970502139158,
      "grad_norm": 1.298464298248291,
      "learning_rate": 3.392020446912279e-05,
      "loss": 1.388,
      "step": 6712
    },
    {
      "epoch": 1.2092771898221122,
      "grad_norm": 1.2222894430160522,
      "learning_rate": 3.390679589204682e-05,
      "loss": 1.2057,
      "step": 6713
    },
    {
      "epoch": 1.2094573294303084,
      "grad_norm": 1.3018053770065308,
      "learning_rate": 3.389338860595984e-05,
      "loss": 1.23,
      "step": 6714
    },
    {
      "epoch": 1.2096374690385048,
      "grad_norm": 1.3688886165618896,
      "learning_rate": 3.3879982611937386e-05,
      "loss": 1.3463,
      "step": 6715
    },
    {
      "epoch": 1.2098176086467012,
      "grad_norm": 1.47565495967865,
      "learning_rate": 3.386657791105488e-05,
      "loss": 1.593,
      "step": 6716
    },
    {
      "epoch": 1.2099977482548976,
      "grad_norm": 1.3760709762573242,
      "learning_rate": 3.385317450438762e-05,
      "loss": 1.3085,
      "step": 6717
    },
    {
      "epoch": 1.210177887863094,
      "grad_norm": 1.4314104318618774,
      "learning_rate": 3.383977239301085e-05,
      "loss": 1.5979,
      "step": 6718
    },
    {
      "epoch": 1.2103580274712902,
      "grad_norm": 1.3776984214782715,
      "learning_rate": 3.382637157799966e-05,
      "loss": 1.5581,
      "step": 6719
    },
    {
      "epoch": 1.2105381670794866,
      "grad_norm": 1.4690535068511963,
      "learning_rate": 3.381297206042907e-05,
      "loss": 1.5545,
      "step": 6720
    },
    {
      "epoch": 1.210718306687683,
      "grad_norm": 1.3706868886947632,
      "learning_rate": 3.3799573841373975e-05,
      "loss": 1.3295,
      "step": 6721
    },
    {
      "epoch": 1.2108984462958794,
      "grad_norm": 1.4949100017547607,
      "learning_rate": 3.3786176921909155e-05,
      "loss": 1.4161,
      "step": 6722
    },
    {
      "epoch": 1.2110785859040756,
      "grad_norm": 1.3225620985031128,
      "learning_rate": 3.3772781303109355e-05,
      "loss": 1.3062,
      "step": 6723
    },
    {
      "epoch": 1.211258725512272,
      "grad_norm": 1.4657484292984009,
      "learning_rate": 3.3759386986049135e-05,
      "loss": 1.4986,
      "step": 6724
    },
    {
      "epoch": 1.2114388651204684,
      "grad_norm": 1.4000439643859863,
      "learning_rate": 3.3745993971802966e-05,
      "loss": 1.2593,
      "step": 6725
    },
    {
      "epoch": 1.2116190047286648,
      "grad_norm": 1.436929702758789,
      "learning_rate": 3.373260226144525e-05,
      "loss": 1.3059,
      "step": 6726
    },
    {
      "epoch": 1.211799144336861,
      "grad_norm": 1.4112604856491089,
      "learning_rate": 3.3719211856050234e-05,
      "loss": 1.3099,
      "step": 6727
    },
    {
      "epoch": 1.2119792839450574,
      "grad_norm": 1.4148492813110352,
      "learning_rate": 3.370582275669214e-05,
      "loss": 1.4562,
      "step": 6728
    },
    {
      "epoch": 1.2121594235532538,
      "grad_norm": 1.5359387397766113,
      "learning_rate": 3.369243496444501e-05,
      "loss": 1.5223,
      "step": 6729
    },
    {
      "epoch": 1.2123395631614502,
      "grad_norm": 1.3944507837295532,
      "learning_rate": 3.36790484803828e-05,
      "loss": 1.4496,
      "step": 6730
    },
    {
      "epoch": 1.2125197027696464,
      "grad_norm": 1.5073171854019165,
      "learning_rate": 3.366566330557935e-05,
      "loss": 1.3671,
      "step": 6731
    },
    {
      "epoch": 1.2126998423778428,
      "grad_norm": 1.4034618139266968,
      "learning_rate": 3.3652279441108446e-05,
      "loss": 1.2737,
      "step": 6732
    },
    {
      "epoch": 1.2128799819860392,
      "grad_norm": 1.4668939113616943,
      "learning_rate": 3.363889688804373e-05,
      "loss": 1.4895,
      "step": 6733
    },
    {
      "epoch": 1.2130601215942356,
      "grad_norm": 1.2924970388412476,
      "learning_rate": 3.362551564745874e-05,
      "loss": 1.2958,
      "step": 6734
    },
    {
      "epoch": 1.213240261202432,
      "grad_norm": 1.6009396314620972,
      "learning_rate": 3.3612135720426904e-05,
      "loss": 1.6714,
      "step": 6735
    },
    {
      "epoch": 1.2134204008106282,
      "grad_norm": 1.5855759382247925,
      "learning_rate": 3.3598757108021546e-05,
      "loss": 1.3441,
      "step": 6736
    },
    {
      "epoch": 1.2136005404188246,
      "grad_norm": 1.610321044921875,
      "learning_rate": 3.3585379811315914e-05,
      "loss": 1.4576,
      "step": 6737
    },
    {
      "epoch": 1.213780680027021,
      "grad_norm": 1.5988126993179321,
      "learning_rate": 3.357200383138312e-05,
      "loss": 1.4523,
      "step": 6738
    },
    {
      "epoch": 1.2139608196352172,
      "grad_norm": 1.5248401165008545,
      "learning_rate": 3.355862916929618e-05,
      "loss": 1.3886,
      "step": 6739
    },
    {
      "epoch": 1.2141409592434136,
      "grad_norm": 1.531491994857788,
      "learning_rate": 3.3545255826128e-05,
      "loss": 1.4907,
      "step": 6740
    },
    {
      "epoch": 1.21432109885161,
      "grad_norm": 1.582066297531128,
      "learning_rate": 3.3531883802951364e-05,
      "loss": 1.3933,
      "step": 6741
    },
    {
      "epoch": 1.2145012384598064,
      "grad_norm": 1.5020043849945068,
      "learning_rate": 3.3518513100839e-05,
      "loss": 1.4564,
      "step": 6742
    },
    {
      "epoch": 1.2146813780680028,
      "grad_norm": 1.484968900680542,
      "learning_rate": 3.3505143720863497e-05,
      "loss": 1.3698,
      "step": 6743
    },
    {
      "epoch": 1.214861517676199,
      "grad_norm": 1.5655699968338013,
      "learning_rate": 3.3491775664097326e-05,
      "loss": 1.4121,
      "step": 6744
    },
    {
      "epoch": 1.2150416572843954,
      "grad_norm": 1.5694249868392944,
      "learning_rate": 3.3478408931612856e-05,
      "loss": 1.4314,
      "step": 6745
    },
    {
      "epoch": 1.2152217968925918,
      "grad_norm": 1.4166194200515747,
      "learning_rate": 3.3465043524482384e-05,
      "loss": 1.219,
      "step": 6746
    },
    {
      "epoch": 1.2154019365007882,
      "grad_norm": 1.4143685102462769,
      "learning_rate": 3.345167944377807e-05,
      "loss": 1.3164,
      "step": 6747
    },
    {
      "epoch": 1.2155820761089844,
      "grad_norm": 1.499548316001892,
      "learning_rate": 3.343831669057197e-05,
      "loss": 1.476,
      "step": 6748
    },
    {
      "epoch": 1.2157622157171808,
      "grad_norm": 1.3988614082336426,
      "learning_rate": 3.342495526593604e-05,
      "loss": 1.1152,
      "step": 6749
    },
    {
      "epoch": 1.2159423553253772,
      "grad_norm": 1.4815285205841064,
      "learning_rate": 3.3411595170942115e-05,
      "loss": 1.2164,
      "step": 6750
    },
    {
      "epoch": 1.2161224949335736,
      "grad_norm": 1.4810343980789185,
      "learning_rate": 3.3398236406661955e-05,
      "loss": 1.0439,
      "step": 6751
    },
    {
      "epoch": 1.2163026345417698,
      "grad_norm": 1.4816360473632812,
      "learning_rate": 3.338487897416718e-05,
      "loss": 2.1415,
      "step": 6752
    },
    {
      "epoch": 1.2164827741499662,
      "grad_norm": 1.4066493511199951,
      "learning_rate": 3.337152287452932e-05,
      "loss": 1.8022,
      "step": 6753
    },
    {
      "epoch": 1.2166629137581626,
      "grad_norm": 1.4352459907531738,
      "learning_rate": 3.33581681088198e-05,
      "loss": 1.7989,
      "step": 6754
    },
    {
      "epoch": 1.216843053366359,
      "grad_norm": 1.3035228252410889,
      "learning_rate": 3.334481467810992e-05,
      "loss": 1.6342,
      "step": 6755
    },
    {
      "epoch": 1.2170231929745552,
      "grad_norm": 1.297839641571045,
      "learning_rate": 3.3331462583470894e-05,
      "loss": 1.8681,
      "step": 6756
    },
    {
      "epoch": 1.2172033325827516,
      "grad_norm": 1.612751841545105,
      "learning_rate": 3.331811182597383e-05,
      "loss": 2.1145,
      "step": 6757
    },
    {
      "epoch": 1.217383472190948,
      "grad_norm": 1.5627317428588867,
      "learning_rate": 3.330476240668971e-05,
      "loss": 1.9384,
      "step": 6758
    },
    {
      "epoch": 1.2175636117991444,
      "grad_norm": 1.390958309173584,
      "learning_rate": 3.3291414326689396e-05,
      "loss": 1.5293,
      "step": 6759
    },
    {
      "epoch": 1.2177437514073408,
      "grad_norm": 1.6251673698425293,
      "learning_rate": 3.3278067587043705e-05,
      "loss": 1.904,
      "step": 6760
    },
    {
      "epoch": 1.217923891015537,
      "grad_norm": 1.6113286018371582,
      "learning_rate": 3.3264722188823295e-05,
      "loss": 1.8217,
      "step": 6761
    },
    {
      "epoch": 1.2181040306237334,
      "grad_norm": 1.5598195791244507,
      "learning_rate": 3.3251378133098723e-05,
      "loss": 1.7962,
      "step": 6762
    },
    {
      "epoch": 1.2182841702319298,
      "grad_norm": 1.4992597103118896,
      "learning_rate": 3.323803542094043e-05,
      "loss": 1.6648,
      "step": 6763
    },
    {
      "epoch": 1.218464309840126,
      "grad_norm": 1.30910325050354,
      "learning_rate": 3.322469405341875e-05,
      "loss": 1.3617,
      "step": 6764
    },
    {
      "epoch": 1.2186444494483224,
      "grad_norm": 1.3764314651489258,
      "learning_rate": 3.321135403160396e-05,
      "loss": 1.4496,
      "step": 6765
    },
    {
      "epoch": 1.2188245890565188,
      "grad_norm": 1.4324058294296265,
      "learning_rate": 3.319801535656617e-05,
      "loss": 1.3841,
      "step": 6766
    },
    {
      "epoch": 1.2190047286647152,
      "grad_norm": 1.372729778289795,
      "learning_rate": 3.318467802937541e-05,
      "loss": 1.3912,
      "step": 6767
    },
    {
      "epoch": 1.2191848682729116,
      "grad_norm": 1.3461893796920776,
      "learning_rate": 3.317134205110159e-05,
      "loss": 1.3233,
      "step": 6768
    },
    {
      "epoch": 1.2193650078811078,
      "grad_norm": 1.3404725790023804,
      "learning_rate": 3.3158007422814496e-05,
      "loss": 1.1978,
      "step": 6769
    },
    {
      "epoch": 1.2195451474893042,
      "grad_norm": 1.4907500743865967,
      "learning_rate": 3.314467414558385e-05,
      "loss": 1.5287,
      "step": 6770
    },
    {
      "epoch": 1.2197252870975006,
      "grad_norm": 1.4512618780136108,
      "learning_rate": 3.313134222047924e-05,
      "loss": 1.2156,
      "step": 6771
    },
    {
      "epoch": 1.2199054267056968,
      "grad_norm": 1.4059784412384033,
      "learning_rate": 3.311801164857015e-05,
      "loss": 1.4038,
      "step": 6772
    },
    {
      "epoch": 1.2200855663138932,
      "grad_norm": 1.3922499418258667,
      "learning_rate": 3.3104682430925924e-05,
      "loss": 1.4809,
      "step": 6773
    },
    {
      "epoch": 1.2202657059220896,
      "grad_norm": 1.4750616550445557,
      "learning_rate": 3.3091354568615854e-05,
      "loss": 1.5713,
      "step": 6774
    },
    {
      "epoch": 1.220445845530286,
      "grad_norm": 1.6020843982696533,
      "learning_rate": 3.307802806270909e-05,
      "loss": 1.8264,
      "step": 6775
    },
    {
      "epoch": 1.2206259851384824,
      "grad_norm": 1.324859380722046,
      "learning_rate": 3.3064702914274665e-05,
      "loss": 1.2919,
      "step": 6776
    },
    {
      "epoch": 1.2208061247466786,
      "grad_norm": 1.4866806268692017,
      "learning_rate": 3.3051379124381526e-05,
      "loss": 1.4767,
      "step": 6777
    },
    {
      "epoch": 1.220986264354875,
      "grad_norm": 1.5042155981063843,
      "learning_rate": 3.303805669409848e-05,
      "loss": 1.4528,
      "step": 6778
    },
    {
      "epoch": 1.2211664039630714,
      "grad_norm": 1.5368664264678955,
      "learning_rate": 3.302473562449429e-05,
      "loss": 1.6127,
      "step": 6779
    },
    {
      "epoch": 1.2213465435712678,
      "grad_norm": 1.3732553720474243,
      "learning_rate": 3.3011415916637536e-05,
      "loss": 1.2218,
      "step": 6780
    },
    {
      "epoch": 1.221526683179464,
      "grad_norm": 1.4287132024765015,
      "learning_rate": 3.2998097571596715e-05,
      "loss": 1.3873,
      "step": 6781
    },
    {
      "epoch": 1.2217068227876604,
      "grad_norm": 1.8109266757965088,
      "learning_rate": 3.2984780590440215e-05,
      "loss": 1.4497,
      "step": 6782
    },
    {
      "epoch": 1.2218869623958568,
      "grad_norm": 1.3814113140106201,
      "learning_rate": 3.2971464974236346e-05,
      "loss": 1.366,
      "step": 6783
    },
    {
      "epoch": 1.2220671020040532,
      "grad_norm": 1.4273579120635986,
      "learning_rate": 3.295815072405326e-05,
      "loss": 1.3397,
      "step": 6784
    },
    {
      "epoch": 1.2222472416122494,
      "grad_norm": 1.5226681232452393,
      "learning_rate": 3.294483784095902e-05,
      "loss": 1.5882,
      "step": 6785
    },
    {
      "epoch": 1.2224273812204458,
      "grad_norm": 1.3168551921844482,
      "learning_rate": 3.293152632602158e-05,
      "loss": 1.2676,
      "step": 6786
    },
    {
      "epoch": 1.2226075208286422,
      "grad_norm": 1.4255211353302002,
      "learning_rate": 3.291821618030877e-05,
      "loss": 1.3754,
      "step": 6787
    },
    {
      "epoch": 1.2227876604368386,
      "grad_norm": 1.4688334465026855,
      "learning_rate": 3.2904907404888366e-05,
      "loss": 1.365,
      "step": 6788
    },
    {
      "epoch": 1.2229678000450348,
      "grad_norm": 1.517831802368164,
      "learning_rate": 3.2891600000827946e-05,
      "loss": 1.4551,
      "step": 6789
    },
    {
      "epoch": 1.2231479396532312,
      "grad_norm": 1.399332880973816,
      "learning_rate": 3.287829396919505e-05,
      "loss": 1.1788,
      "step": 6790
    },
    {
      "epoch": 1.2233280792614276,
      "grad_norm": 1.6225569248199463,
      "learning_rate": 3.286498931105706e-05,
      "loss": 1.3243,
      "step": 6791
    },
    {
      "epoch": 1.223508218869624,
      "grad_norm": 1.4913194179534912,
      "learning_rate": 3.285168602748128e-05,
      "loss": 1.3081,
      "step": 6792
    },
    {
      "epoch": 1.2236883584778204,
      "grad_norm": 1.4559649229049683,
      "learning_rate": 3.28383841195349e-05,
      "loss": 1.3107,
      "step": 6793
    },
    {
      "epoch": 1.2238684980860166,
      "grad_norm": 1.4115616083145142,
      "learning_rate": 3.2825083588285e-05,
      "loss": 1.0701,
      "step": 6794
    },
    {
      "epoch": 1.224048637694213,
      "grad_norm": 1.4900233745574951,
      "learning_rate": 3.281178443479852e-05,
      "loss": 1.3649,
      "step": 6795
    },
    {
      "epoch": 1.2242287773024094,
      "grad_norm": 1.3989254236221313,
      "learning_rate": 3.27984866601423e-05,
      "loss": 1.1522,
      "step": 6796
    },
    {
      "epoch": 1.2244089169106056,
      "grad_norm": 1.4771958589553833,
      "learning_rate": 3.278519026538312e-05,
      "loss": 1.2725,
      "step": 6797
    },
    {
      "epoch": 1.224589056518802,
      "grad_norm": 1.3885525465011597,
      "learning_rate": 3.277189525158759e-05,
      "loss": 1.3949,
      "step": 6798
    },
    {
      "epoch": 1.2247691961269984,
      "grad_norm": 1.324245810508728,
      "learning_rate": 3.275860161982224e-05,
      "loss": 1.1143,
      "step": 6799
    },
    {
      "epoch": 1.2249493357351948,
      "grad_norm": 1.5012863874435425,
      "learning_rate": 3.274530937115348e-05,
      "loss": 1.1181,
      "step": 6800
    },
    {
      "epoch": 1.2251294753433912,
      "grad_norm": 1.40353262424469,
      "learning_rate": 3.2732018506647553e-05,
      "loss": 1.1923,
      "step": 6801
    },
    {
      "epoch": 1.2253096149515874,
      "grad_norm": 1.4508329629898071,
      "learning_rate": 3.271872902737072e-05,
      "loss": 1.968,
      "step": 6802
    },
    {
      "epoch": 1.2254897545597838,
      "grad_norm": 1.3287415504455566,
      "learning_rate": 3.270544093438903e-05,
      "loss": 1.8131,
      "step": 6803
    },
    {
      "epoch": 1.2256698941679802,
      "grad_norm": 1.3261590003967285,
      "learning_rate": 3.269215422876844e-05,
      "loss": 1.793,
      "step": 6804
    },
    {
      "epoch": 1.2258500337761766,
      "grad_norm": 1.4887539148330688,
      "learning_rate": 3.2678868911574814e-05,
      "loss": 1.9968,
      "step": 6805
    },
    {
      "epoch": 1.2260301733843728,
      "grad_norm": 1.391884684562683,
      "learning_rate": 3.266558498387386e-05,
      "loss": 1.6587,
      "step": 6806
    },
    {
      "epoch": 1.2262103129925692,
      "grad_norm": 1.4738408327102661,
      "learning_rate": 3.2652302446731266e-05,
      "loss": 1.7724,
      "step": 6807
    },
    {
      "epoch": 1.2263904526007656,
      "grad_norm": 1.4325567483901978,
      "learning_rate": 3.2639021301212513e-05,
      "loss": 1.873,
      "step": 6808
    },
    {
      "epoch": 1.226570592208962,
      "grad_norm": 1.4966789484024048,
      "learning_rate": 3.262574154838301e-05,
      "loss": 1.6836,
      "step": 6809
    },
    {
      "epoch": 1.2267507318171582,
      "grad_norm": 1.8694946765899658,
      "learning_rate": 3.2612463189308054e-05,
      "loss": 1.8775,
      "step": 6810
    },
    {
      "epoch": 1.2269308714253546,
      "grad_norm": 1.651310920715332,
      "learning_rate": 3.259918622505284e-05,
      "loss": 1.6738,
      "step": 6811
    },
    {
      "epoch": 1.227111011033551,
      "grad_norm": 1.4324439764022827,
      "learning_rate": 3.258591065668243e-05,
      "loss": 1.5847,
      "step": 6812
    },
    {
      "epoch": 1.2272911506417474,
      "grad_norm": 1.497910499572754,
      "learning_rate": 3.257263648526178e-05,
      "loss": 1.6083,
      "step": 6813
    },
    {
      "epoch": 1.2274712902499436,
      "grad_norm": 1.367472767829895,
      "learning_rate": 3.255936371185575e-05,
      "loss": 1.5481,
      "step": 6814
    },
    {
      "epoch": 1.22765142985814,
      "grad_norm": 1.2517704963684082,
      "learning_rate": 3.254609233752906e-05,
      "loss": 1.2703,
      "step": 6815
    },
    {
      "epoch": 1.2278315694663364,
      "grad_norm": 1.38472580909729,
      "learning_rate": 3.253282236334634e-05,
      "loss": 1.2486,
      "step": 6816
    },
    {
      "epoch": 1.2280117090745328,
      "grad_norm": 1.2442814111709595,
      "learning_rate": 3.251955379037211e-05,
      "loss": 1.3852,
      "step": 6817
    },
    {
      "epoch": 1.2281918486827292,
      "grad_norm": 1.413346290588379,
      "learning_rate": 3.250628661967076e-05,
      "loss": 1.4071,
      "step": 6818
    },
    {
      "epoch": 1.2283719882909254,
      "grad_norm": 1.603061556816101,
      "learning_rate": 3.249302085230658e-05,
      "loss": 1.5594,
      "step": 6819
    },
    {
      "epoch": 1.2285521278991218,
      "grad_norm": 1.2397353649139404,
      "learning_rate": 3.2479756489343714e-05,
      "loss": 1.1788,
      "step": 6820
    },
    {
      "epoch": 1.2287322675073182,
      "grad_norm": 1.375002145767212,
      "learning_rate": 3.2466493531846274e-05,
      "loss": 1.2393,
      "step": 6821
    },
    {
      "epoch": 1.2289124071155144,
      "grad_norm": 1.4281266927719116,
      "learning_rate": 3.245323198087818e-05,
      "loss": 1.5349,
      "step": 6822
    },
    {
      "epoch": 1.2290925467237108,
      "grad_norm": 1.3915725946426392,
      "learning_rate": 3.2439971837503266e-05,
      "loss": 1.4838,
      "step": 6823
    },
    {
      "epoch": 1.2292726863319072,
      "grad_norm": 1.503351092338562,
      "learning_rate": 3.242671310278525e-05,
      "loss": 1.6898,
      "step": 6824
    },
    {
      "epoch": 1.2294528259401036,
      "grad_norm": 1.3596590757369995,
      "learning_rate": 3.241345577778775e-05,
      "loss": 1.587,
      "step": 6825
    },
    {
      "epoch": 1.2296329655483,
      "grad_norm": 1.49898099899292,
      "learning_rate": 3.240019986357429e-05,
      "loss": 1.4074,
      "step": 6826
    },
    {
      "epoch": 1.2298131051564962,
      "grad_norm": 1.4580934047698975,
      "learning_rate": 3.23869453612082e-05,
      "loss": 1.396,
      "step": 6827
    },
    {
      "epoch": 1.2299932447646926,
      "grad_norm": 1.3868801593780518,
      "learning_rate": 3.237369227175278e-05,
      "loss": 1.1905,
      "step": 6828
    },
    {
      "epoch": 1.230173384372889,
      "grad_norm": 1.2144285440444946,
      "learning_rate": 3.236044059627116e-05,
      "loss": 1.1591,
      "step": 6829
    },
    {
      "epoch": 1.2303535239810852,
      "grad_norm": 1.482460618019104,
      "learning_rate": 3.234719033582642e-05,
      "loss": 1.5664,
      "step": 6830
    },
    {
      "epoch": 1.2305336635892816,
      "grad_norm": 1.5039933919906616,
      "learning_rate": 3.233394149148148e-05,
      "loss": 1.6836,
      "step": 6831
    },
    {
      "epoch": 1.230713803197478,
      "grad_norm": 1.5417627096176147,
      "learning_rate": 3.2320694064299146e-05,
      "loss": 1.4081,
      "step": 6832
    },
    {
      "epoch": 1.2308939428056744,
      "grad_norm": 1.4270129203796387,
      "learning_rate": 3.230744805534213e-05,
      "loss": 1.3792,
      "step": 6833
    },
    {
      "epoch": 1.2310740824138708,
      "grad_norm": 1.448017954826355,
      "learning_rate": 3.2294203465672986e-05,
      "loss": 1.3987,
      "step": 6834
    },
    {
      "epoch": 1.231254222022067,
      "grad_norm": 1.5489007234573364,
      "learning_rate": 3.228096029635423e-05,
      "loss": 1.4949,
      "step": 6835
    },
    {
      "epoch": 1.2314343616302634,
      "grad_norm": 1.4830647706985474,
      "learning_rate": 3.226771854844822e-05,
      "loss": 1.479,
      "step": 6836
    },
    {
      "epoch": 1.2316145012384598,
      "grad_norm": 1.587178349494934,
      "learning_rate": 3.225447822301718e-05,
      "loss": 1.6389,
      "step": 6837
    },
    {
      "epoch": 1.2317946408466562,
      "grad_norm": 1.553702473640442,
      "learning_rate": 3.2241239321123254e-05,
      "loss": 1.5346,
      "step": 6838
    },
    {
      "epoch": 1.2319747804548524,
      "grad_norm": 1.398423194885254,
      "learning_rate": 3.222800184382846e-05,
      "loss": 1.1183,
      "step": 6839
    },
    {
      "epoch": 1.2321549200630488,
      "grad_norm": 1.6206096410751343,
      "learning_rate": 3.2214765792194706e-05,
      "loss": 1.6472,
      "step": 6840
    },
    {
      "epoch": 1.2323350596712452,
      "grad_norm": 1.6678884029388428,
      "learning_rate": 3.2201531167283774e-05,
      "loss": 1.6587,
      "step": 6841
    },
    {
      "epoch": 1.2325151992794416,
      "grad_norm": 1.458854079246521,
      "learning_rate": 3.218829797015735e-05,
      "loss": 1.3503,
      "step": 6842
    },
    {
      "epoch": 1.2326953388876378,
      "grad_norm": 1.5148917436599731,
      "learning_rate": 3.217506620187697e-05,
      "loss": 1.3895,
      "step": 6843
    },
    {
      "epoch": 1.2328754784958342,
      "grad_norm": 1.3003181219100952,
      "learning_rate": 3.216183586350411e-05,
      "loss": 1.1938,
      "step": 6844
    },
    {
      "epoch": 1.2330556181040306,
      "grad_norm": 1.507581114768982,
      "learning_rate": 3.214860695610008e-05,
      "loss": 1.364,
      "step": 6845
    },
    {
      "epoch": 1.233235757712227,
      "grad_norm": 1.55634343624115,
      "learning_rate": 3.2135379480726116e-05,
      "loss": 1.5315,
      "step": 6846
    },
    {
      "epoch": 1.2334158973204232,
      "grad_norm": 1.4182003736495972,
      "learning_rate": 3.212215343844331e-05,
      "loss": 1.2499,
      "step": 6847
    },
    {
      "epoch": 1.2335960369286196,
      "grad_norm": 1.4784570932388306,
      "learning_rate": 3.210892883031263e-05,
      "loss": 1.3,
      "step": 6848
    },
    {
      "epoch": 1.233776176536816,
      "grad_norm": 1.5924310684204102,
      "learning_rate": 3.209570565739498e-05,
      "loss": 1.2403,
      "step": 6849
    },
    {
      "epoch": 1.2339563161450124,
      "grad_norm": 1.3215382099151611,
      "learning_rate": 3.2082483920751095e-05,
      "loss": 1.0135,
      "step": 6850
    },
    {
      "epoch": 1.2341364557532088,
      "grad_norm": 1.370424747467041,
      "learning_rate": 3.2069263621441626e-05,
      "loss": 1.1475,
      "step": 6851
    },
    {
      "epoch": 1.234316595361405,
      "grad_norm": 1.4670120477676392,
      "learning_rate": 3.205604476052707e-05,
      "loss": 2.2251,
      "step": 6852
    },
    {
      "epoch": 1.2344967349696014,
      "grad_norm": 1.2843053340911865,
      "learning_rate": 3.2042827339067903e-05,
      "loss": 1.671,
      "step": 6853
    },
    {
      "epoch": 1.2346768745777978,
      "grad_norm": 1.268125295639038,
      "learning_rate": 3.202961135812437e-05,
      "loss": 1.6883,
      "step": 6854
    },
    {
      "epoch": 1.234857014185994,
      "grad_norm": 1.2623155117034912,
      "learning_rate": 3.201639681875665e-05,
      "loss": 1.6978,
      "step": 6855
    },
    {
      "epoch": 1.2350371537941904,
      "grad_norm": 1.4933207035064697,
      "learning_rate": 3.200318372202483e-05,
      "loss": 1.7517,
      "step": 6856
    },
    {
      "epoch": 1.2352172934023868,
      "grad_norm": 1.4884768724441528,
      "learning_rate": 3.1989972068988815e-05,
      "loss": 1.686,
      "step": 6857
    },
    {
      "epoch": 1.2353974330105832,
      "grad_norm": 1.4897816181182861,
      "learning_rate": 3.197676186070849e-05,
      "loss": 1.8471,
      "step": 6858
    },
    {
      "epoch": 1.2355775726187797,
      "grad_norm": 1.6314853429794312,
      "learning_rate": 3.196355309824357e-05,
      "loss": 2.145,
      "step": 6859
    },
    {
      "epoch": 1.2357577122269758,
      "grad_norm": 1.7411549091339111,
      "learning_rate": 3.195034578265362e-05,
      "loss": 2.0476,
      "step": 6860
    },
    {
      "epoch": 1.2359378518351722,
      "grad_norm": 1.785402774810791,
      "learning_rate": 3.1937139914998135e-05,
      "loss": 2.2141,
      "step": 6861
    },
    {
      "epoch": 1.2361179914433686,
      "grad_norm": 1.5895336866378784,
      "learning_rate": 3.192393549633647e-05,
      "loss": 1.7239,
      "step": 6862
    },
    {
      "epoch": 1.236298131051565,
      "grad_norm": 1.421332836151123,
      "learning_rate": 3.191073252772791e-05,
      "loss": 1.6081,
      "step": 6863
    },
    {
      "epoch": 1.2364782706597612,
      "grad_norm": 1.5207263231277466,
      "learning_rate": 3.1897531010231584e-05,
      "loss": 1.465,
      "step": 6864
    },
    {
      "epoch": 1.2366584102679576,
      "grad_norm": 1.3388986587524414,
      "learning_rate": 3.188433094490652e-05,
      "loss": 1.3126,
      "step": 6865
    },
    {
      "epoch": 1.236838549876154,
      "grad_norm": 1.3267252445220947,
      "learning_rate": 3.187113233281156e-05,
      "loss": 1.246,
      "step": 6866
    },
    {
      "epoch": 1.2370186894843505,
      "grad_norm": 1.4351184368133545,
      "learning_rate": 3.185793517500555e-05,
      "loss": 1.3043,
      "step": 6867
    },
    {
      "epoch": 1.2371988290925466,
      "grad_norm": 1.4483933448791504,
      "learning_rate": 3.1844739472547156e-05,
      "loss": 1.3921,
      "step": 6868
    },
    {
      "epoch": 1.237378968700743,
      "grad_norm": 1.3321502208709717,
      "learning_rate": 3.183154522649492e-05,
      "loss": 1.1965,
      "step": 6869
    },
    {
      "epoch": 1.2375591083089394,
      "grad_norm": 1.4448031187057495,
      "learning_rate": 3.181835243790727e-05,
      "loss": 1.5186,
      "step": 6870
    },
    {
      "epoch": 1.2377392479171359,
      "grad_norm": 1.4353901147842407,
      "learning_rate": 3.180516110784252e-05,
      "loss": 1.4443,
      "step": 6871
    },
    {
      "epoch": 1.237919387525332,
      "grad_norm": 1.4049485921859741,
      "learning_rate": 3.179197123735889e-05,
      "loss": 1.4224,
      "step": 6872
    },
    {
      "epoch": 1.2380995271335284,
      "grad_norm": 1.414174199104309,
      "learning_rate": 3.177878282751446e-05,
      "loss": 1.2071,
      "step": 6873
    },
    {
      "epoch": 1.2382796667417248,
      "grad_norm": 1.3704897165298462,
      "learning_rate": 3.17655958793672e-05,
      "loss": 1.4004,
      "step": 6874
    },
    {
      "epoch": 1.2384598063499213,
      "grad_norm": 1.7496007680892944,
      "learning_rate": 3.1752410393974965e-05,
      "loss": 1.6258,
      "step": 6875
    },
    {
      "epoch": 1.2386399459581177,
      "grad_norm": 1.510542392730713,
      "learning_rate": 3.173922637239546e-05,
      "loss": 1.3983,
      "step": 6876
    },
    {
      "epoch": 1.2388200855663138,
      "grad_norm": 1.4122511148452759,
      "learning_rate": 3.172604381568634e-05,
      "loss": 1.3087,
      "step": 6877
    },
    {
      "epoch": 1.2390002251745103,
      "grad_norm": 1.4079084396362305,
      "learning_rate": 3.1712862724905076e-05,
      "loss": 1.3244,
      "step": 6878
    },
    {
      "epoch": 1.2391803647827067,
      "grad_norm": 1.5156313180923462,
      "learning_rate": 3.169968310110906e-05,
      "loss": 1.5998,
      "step": 6879
    },
    {
      "epoch": 1.2393605043909028,
      "grad_norm": 1.4952698945999146,
      "learning_rate": 3.1686504945355536e-05,
      "loss": 1.6639,
      "step": 6880
    },
    {
      "epoch": 1.2395406439990992,
      "grad_norm": 1.3995941877365112,
      "learning_rate": 3.167332825870168e-05,
      "loss": 1.6085,
      "step": 6881
    },
    {
      "epoch": 1.2397207836072957,
      "grad_norm": 1.5575815439224243,
      "learning_rate": 3.16601530422045e-05,
      "loss": 1.5242,
      "step": 6882
    },
    {
      "epoch": 1.239900923215492,
      "grad_norm": 1.4111615419387817,
      "learning_rate": 3.164697929692092e-05,
      "loss": 1.3462,
      "step": 6883
    },
    {
      "epoch": 1.2400810628236885,
      "grad_norm": 1.5483009815216064,
      "learning_rate": 3.1633807023907716e-05,
      "loss": 1.4555,
      "step": 6884
    },
    {
      "epoch": 1.2402612024318846,
      "grad_norm": 1.4701731204986572,
      "learning_rate": 3.162063622422155e-05,
      "loss": 1.3976,
      "step": 6885
    },
    {
      "epoch": 1.240441342040081,
      "grad_norm": 1.435465693473816,
      "learning_rate": 3.160746689891901e-05,
      "loss": 1.4032,
      "step": 6886
    },
    {
      "epoch": 1.2406214816482775,
      "grad_norm": 1.362147569656372,
      "learning_rate": 3.159429904905652e-05,
      "loss": 1.4076,
      "step": 6887
    },
    {
      "epoch": 1.2408016212564739,
      "grad_norm": 1.590501070022583,
      "learning_rate": 3.158113267569039e-05,
      "loss": 1.5094,
      "step": 6888
    },
    {
      "epoch": 1.24098176086467,
      "grad_norm": 1.5492151975631714,
      "learning_rate": 3.1567967779876804e-05,
      "loss": 1.7515,
      "step": 6889
    },
    {
      "epoch": 1.2411619004728665,
      "grad_norm": 1.4863865375518799,
      "learning_rate": 3.1554804362671884e-05,
      "loss": 1.3005,
      "step": 6890
    },
    {
      "epoch": 1.2413420400810629,
      "grad_norm": 1.5774645805358887,
      "learning_rate": 3.154164242513159e-05,
      "loss": 1.3072,
      "step": 6891
    },
    {
      "epoch": 1.2415221796892593,
      "grad_norm": 1.7997633218765259,
      "learning_rate": 3.152848196831174e-05,
      "loss": 1.6606,
      "step": 6892
    },
    {
      "epoch": 1.2417023192974554,
      "grad_norm": 1.5264493227005005,
      "learning_rate": 3.151532299326806e-05,
      "loss": 1.4024,
      "step": 6893
    },
    {
      "epoch": 1.2418824589056519,
      "grad_norm": 1.5488672256469727,
      "learning_rate": 3.1502165501056156e-05,
      "loss": 1.2862,
      "step": 6894
    },
    {
      "epoch": 1.2420625985138483,
      "grad_norm": 1.4287611246109009,
      "learning_rate": 3.148900949273155e-05,
      "loss": 1.4463,
      "step": 6895
    },
    {
      "epoch": 1.2422427381220447,
      "grad_norm": 1.6001183986663818,
      "learning_rate": 3.147585496934957e-05,
      "loss": 1.4893,
      "step": 6896
    },
    {
      "epoch": 1.2424228777302408,
      "grad_norm": 1.5124192237854004,
      "learning_rate": 3.14627019319655e-05,
      "loss": 1.3634,
      "step": 6897
    },
    {
      "epoch": 1.2426030173384373,
      "grad_norm": 1.4344770908355713,
      "learning_rate": 3.144955038163444e-05,
      "loss": 1.1965,
      "step": 6898
    },
    {
      "epoch": 1.2427831569466337,
      "grad_norm": 1.4477797746658325,
      "learning_rate": 3.14364003194114e-05,
      "loss": 1.2443,
      "step": 6899
    },
    {
      "epoch": 1.24296329655483,
      "grad_norm": 1.4012515544891357,
      "learning_rate": 3.14232517463513e-05,
      "loss": 1.1767,
      "step": 6900
    },
    {
      "epoch": 1.2431434361630262,
      "grad_norm": 1.3314661979675293,
      "learning_rate": 3.141010466350889e-05,
      "loss": 1.2804,
      "step": 6901
    },
    {
      "epoch": 1.2433235757712227,
      "grad_norm": 1.3033140897750854,
      "learning_rate": 3.139695907193884e-05,
      "loss": 1.801,
      "step": 6902
    },
    {
      "epoch": 1.243503715379419,
      "grad_norm": 1.3037872314453125,
      "learning_rate": 3.1383814972695646e-05,
      "loss": 1.8049,
      "step": 6903
    },
    {
      "epoch": 1.2436838549876155,
      "grad_norm": 1.3784995079040527,
      "learning_rate": 3.137067236683378e-05,
      "loss": 1.9693,
      "step": 6904
    },
    {
      "epoch": 1.2438639945958117,
      "grad_norm": 1.3589712381362915,
      "learning_rate": 3.135753125540748e-05,
      "loss": 1.8749,
      "step": 6905
    },
    {
      "epoch": 1.244044134204008,
      "grad_norm": 1.5621092319488525,
      "learning_rate": 3.134439163947095e-05,
      "loss": 1.6441,
      "step": 6906
    },
    {
      "epoch": 1.2442242738122045,
      "grad_norm": 1.3298524618148804,
      "learning_rate": 3.1331253520078244e-05,
      "loss": 1.5181,
      "step": 6907
    },
    {
      "epoch": 1.2444044134204009,
      "grad_norm": 1.46505868434906,
      "learning_rate": 3.131811689828327e-05,
      "loss": 1.762,
      "step": 6908
    },
    {
      "epoch": 1.2445845530285973,
      "grad_norm": 1.7069956064224243,
      "learning_rate": 3.1304981775139875e-05,
      "loss": 2.0042,
      "step": 6909
    },
    {
      "epoch": 1.2447646926367935,
      "grad_norm": 1.6842715740203857,
      "learning_rate": 3.129184815170173e-05,
      "loss": 1.8457,
      "step": 6910
    },
    {
      "epoch": 1.2449448322449899,
      "grad_norm": 1.6530653238296509,
      "learning_rate": 3.127871602902241e-05,
      "loss": 1.8401,
      "step": 6911
    },
    {
      "epoch": 1.2451249718531863,
      "grad_norm": 1.4249504804611206,
      "learning_rate": 3.126558540815538e-05,
      "loss": 1.3493,
      "step": 6912
    },
    {
      "epoch": 1.2453051114613825,
      "grad_norm": 1.3376333713531494,
      "learning_rate": 3.1252456290153954e-05,
      "loss": 1.5235,
      "step": 6913
    },
    {
      "epoch": 1.2454852510695789,
      "grad_norm": 1.407130479812622,
      "learning_rate": 3.1239328676071355e-05,
      "loss": 1.4224,
      "step": 6914
    },
    {
      "epoch": 1.2456653906777753,
      "grad_norm": 1.4102033376693726,
      "learning_rate": 3.122620256696067e-05,
      "loss": 1.6468,
      "step": 6915
    },
    {
      "epoch": 1.2458455302859717,
      "grad_norm": 1.3153547048568726,
      "learning_rate": 3.121307796387488e-05,
      "loss": 1.251,
      "step": 6916
    },
    {
      "epoch": 1.246025669894168,
      "grad_norm": 1.3923897743225098,
      "learning_rate": 3.11999548678668e-05,
      "loss": 1.5507,
      "step": 6917
    },
    {
      "epoch": 1.2462058095023643,
      "grad_norm": 1.4627549648284912,
      "learning_rate": 3.1186833279989215e-05,
      "loss": 1.7288,
      "step": 6918
    },
    {
      "epoch": 1.2463859491105607,
      "grad_norm": 1.350342035293579,
      "learning_rate": 3.117371320129469e-05,
      "loss": 1.4016,
      "step": 6919
    },
    {
      "epoch": 1.246566088718757,
      "grad_norm": 1.4918733835220337,
      "learning_rate": 3.116059463283573e-05,
      "loss": 1.5074,
      "step": 6920
    },
    {
      "epoch": 1.2467462283269535,
      "grad_norm": 1.5650421380996704,
      "learning_rate": 3.1147477575664675e-05,
      "loss": 1.7041,
      "step": 6921
    },
    {
      "epoch": 1.2469263679351497,
      "grad_norm": 1.3583215475082397,
      "learning_rate": 3.1134362030833775e-05,
      "loss": 1.2494,
      "step": 6922
    },
    {
      "epoch": 1.247106507543346,
      "grad_norm": 1.5212762355804443,
      "learning_rate": 3.1121247999395185e-05,
      "loss": 1.5584,
      "step": 6923
    },
    {
      "epoch": 1.2472866471515425,
      "grad_norm": 1.6421043872833252,
      "learning_rate": 3.110813548240089e-05,
      "loss": 1.5528,
      "step": 6924
    },
    {
      "epoch": 1.2474667867597389,
      "grad_norm": 1.327778935432434,
      "learning_rate": 3.1095024480902765e-05,
      "loss": 1.1003,
      "step": 6925
    },
    {
      "epoch": 1.247646926367935,
      "grad_norm": 1.5680298805236816,
      "learning_rate": 3.1081914995952564e-05,
      "loss": 1.588,
      "step": 6926
    },
    {
      "epoch": 1.2478270659761315,
      "grad_norm": 1.5872795581817627,
      "learning_rate": 3.1068807028601906e-05,
      "loss": 1.7875,
      "step": 6927
    },
    {
      "epoch": 1.2480072055843279,
      "grad_norm": 1.3631765842437744,
      "learning_rate": 3.1055700579902345e-05,
      "loss": 1.3209,
      "step": 6928
    },
    {
      "epoch": 1.2481873451925243,
      "grad_norm": 1.4500107765197754,
      "learning_rate": 3.104259565090526e-05,
      "loss": 1.3819,
      "step": 6929
    },
    {
      "epoch": 1.2483674848007205,
      "grad_norm": 1.4599426984786987,
      "learning_rate": 3.102949224266192e-05,
      "loss": 1.4269,
      "step": 6930
    },
    {
      "epoch": 1.2485476244089169,
      "grad_norm": 1.5013508796691895,
      "learning_rate": 3.101639035622345e-05,
      "loss": 1.3436,
      "step": 6931
    },
    {
      "epoch": 1.2487277640171133,
      "grad_norm": 1.3377251625061035,
      "learning_rate": 3.100328999264092e-05,
      "loss": 1.3642,
      "step": 6932
    },
    {
      "epoch": 1.2489079036253097,
      "grad_norm": 1.4358447790145874,
      "learning_rate": 3.099019115296521e-05,
      "loss": 1.5565,
      "step": 6933
    },
    {
      "epoch": 1.249088043233506,
      "grad_norm": 1.505300521850586,
      "learning_rate": 3.09770938382471e-05,
      "loss": 1.6027,
      "step": 6934
    },
    {
      "epoch": 1.2492681828417023,
      "grad_norm": 1.44719660282135,
      "learning_rate": 3.096399804953727e-05,
      "loss": 1.4905,
      "step": 6935
    },
    {
      "epoch": 1.2494483224498987,
      "grad_norm": 1.3589566946029663,
      "learning_rate": 3.095090378788622e-05,
      "loss": 1.3473,
      "step": 6936
    },
    {
      "epoch": 1.249628462058095,
      "grad_norm": 1.4965473413467407,
      "learning_rate": 3.093781105434441e-05,
      "loss": 1.517,
      "step": 6937
    },
    {
      "epoch": 1.2498086016662913,
      "grad_norm": 1.4648877382278442,
      "learning_rate": 3.092471984996211e-05,
      "loss": 1.4618,
      "step": 6938
    },
    {
      "epoch": 1.2499887412744877,
      "grad_norm": 1.5023448467254639,
      "learning_rate": 3.091163017578949e-05,
      "loss": 1.3562,
      "step": 6939
    },
    {
      "epoch": 1.250168880882684,
      "grad_norm": 1.553214192390442,
      "learning_rate": 3.0898542032876605e-05,
      "loss": 1.4443,
      "step": 6940
    },
    {
      "epoch": 1.2503490204908805,
      "grad_norm": 1.4273412227630615,
      "learning_rate": 3.088545542227336e-05,
      "loss": 1.4012,
      "step": 6941
    },
    {
      "epoch": 1.2505291600990769,
      "grad_norm": 1.4459377527236938,
      "learning_rate": 3.087237034502958e-05,
      "loss": 1.4371,
      "step": 6942
    },
    {
      "epoch": 1.250709299707273,
      "grad_norm": 1.450222134590149,
      "learning_rate": 3.085928680219494e-05,
      "loss": 1.5046,
      "step": 6943
    },
    {
      "epoch": 1.2508894393154695,
      "grad_norm": 1.4296579360961914,
      "learning_rate": 3.084620479481898e-05,
      "loss": 1.3686,
      "step": 6944
    },
    {
      "epoch": 1.2510695789236659,
      "grad_norm": 1.3699074983596802,
      "learning_rate": 3.083312432395113e-05,
      "loss": 1.2766,
      "step": 6945
    },
    {
      "epoch": 1.251249718531862,
      "grad_norm": 1.496502161026001,
      "learning_rate": 3.082004539064074e-05,
      "loss": 1.4537,
      "step": 6946
    },
    {
      "epoch": 1.2514298581400585,
      "grad_norm": 1.3376675844192505,
      "learning_rate": 3.080696799593695e-05,
      "loss": 1.1308,
      "step": 6947
    },
    {
      "epoch": 1.2516099977482549,
      "grad_norm": 1.4920117855072021,
      "learning_rate": 3.079389214088885e-05,
      "loss": 1.2374,
      "step": 6948
    },
    {
      "epoch": 1.2517901373564513,
      "grad_norm": 1.442320704460144,
      "learning_rate": 3.0780817826545356e-05,
      "loss": 1.1047,
      "step": 6949
    },
    {
      "epoch": 1.2519702769646477,
      "grad_norm": 1.4925241470336914,
      "learning_rate": 3.076774505395529e-05,
      "loss": 1.1603,
      "step": 6950
    },
    {
      "epoch": 1.2521504165728439,
      "grad_norm": 1.7362186908721924,
      "learning_rate": 3.075467382416736e-05,
      "loss": 1.4566,
      "step": 6951
    },
    {
      "epoch": 1.2523305561810403,
      "grad_norm": 1.2892107963562012,
      "learning_rate": 3.074160413823012e-05,
      "loss": 1.8149,
      "step": 6952
    },
    {
      "epoch": 1.2525106957892367,
      "grad_norm": 1.2806421518325806,
      "learning_rate": 3.072853599719201e-05,
      "loss": 1.6805,
      "step": 6953
    },
    {
      "epoch": 1.252690835397433,
      "grad_norm": 1.3249517679214478,
      "learning_rate": 3.071546940210136e-05,
      "loss": 1.8209,
      "step": 6954
    },
    {
      "epoch": 1.2528709750056293,
      "grad_norm": 1.3965421915054321,
      "learning_rate": 3.0702404354006345e-05,
      "loss": 1.6613,
      "step": 6955
    },
    {
      "epoch": 1.2530511146138257,
      "grad_norm": 1.4450734853744507,
      "learning_rate": 3.068934085395508e-05,
      "loss": 1.8908,
      "step": 6956
    },
    {
      "epoch": 1.253231254222022,
      "grad_norm": 1.4606119394302368,
      "learning_rate": 3.0676278902995475e-05,
      "loss": 1.7144,
      "step": 6957
    },
    {
      "epoch": 1.2534113938302185,
      "grad_norm": 1.5273696184158325,
      "learning_rate": 3.066321850217536e-05,
      "loss": 1.7206,
      "step": 6958
    },
    {
      "epoch": 1.253591533438415,
      "grad_norm": 1.531860589981079,
      "learning_rate": 3.065015965254243e-05,
      "loss": 1.8984,
      "step": 6959
    },
    {
      "epoch": 1.253771673046611,
      "grad_norm": 1.8183238506317139,
      "learning_rate": 3.0637102355144276e-05,
      "loss": 2.0119,
      "step": 6960
    },
    {
      "epoch": 1.2539518126548075,
      "grad_norm": 1.486114501953125,
      "learning_rate": 3.062404661102835e-05,
      "loss": 1.5818,
      "step": 6961
    },
    {
      "epoch": 1.254131952263004,
      "grad_norm": 1.3486278057098389,
      "learning_rate": 3.061099242124197e-05,
      "loss": 1.3069,
      "step": 6962
    },
    {
      "epoch": 1.2543120918712,
      "grad_norm": 1.2821143865585327,
      "learning_rate": 3.0597939786832324e-05,
      "loss": 1.325,
      "step": 6963
    },
    {
      "epoch": 1.2544922314793965,
      "grad_norm": 1.303511142730713,
      "learning_rate": 3.058488870884648e-05,
      "loss": 1.4196,
      "step": 6964
    },
    {
      "epoch": 1.2546723710875929,
      "grad_norm": 1.3509190082550049,
      "learning_rate": 3.057183918833143e-05,
      "loss": 1.2932,
      "step": 6965
    },
    {
      "epoch": 1.2548525106957893,
      "grad_norm": 1.4364148378372192,
      "learning_rate": 3.055879122633397e-05,
      "loss": 1.4473,
      "step": 6966
    },
    {
      "epoch": 1.2550326503039857,
      "grad_norm": 1.4226536750793457,
      "learning_rate": 3.054574482390081e-05,
      "loss": 1.4111,
      "step": 6967
    },
    {
      "epoch": 1.2552127899121819,
      "grad_norm": 1.3846882581710815,
      "learning_rate": 3.053269998207854e-05,
      "loss": 1.4484,
      "step": 6968
    },
    {
      "epoch": 1.2553929295203783,
      "grad_norm": 1.4553942680358887,
      "learning_rate": 3.0519656701913555e-05,
      "loss": 1.5139,
      "step": 6969
    },
    {
      "epoch": 1.2555730691285747,
      "grad_norm": 1.4015153646469116,
      "learning_rate": 3.0506614984452254e-05,
      "loss": 1.5118,
      "step": 6970
    },
    {
      "epoch": 1.2557532087367709,
      "grad_norm": 1.5202937126159668,
      "learning_rate": 3.04935748307408e-05,
      "loss": 1.493,
      "step": 6971
    },
    {
      "epoch": 1.2559333483449673,
      "grad_norm": 1.4150903224945068,
      "learning_rate": 3.0480536241825263e-05,
      "loss": 1.248,
      "step": 6972
    },
    {
      "epoch": 1.2561134879531637,
      "grad_norm": 1.2660809755325317,
      "learning_rate": 3.0467499218751605e-05,
      "loss": 1.338,
      "step": 6973
    },
    {
      "epoch": 1.25629362756136,
      "grad_norm": 1.5115392208099365,
      "learning_rate": 3.0454463762565643e-05,
      "loss": 1.592,
      "step": 6974
    },
    {
      "epoch": 1.2564737671695565,
      "grad_norm": 1.511918544769287,
      "learning_rate": 3.044142987431309e-05,
      "loss": 1.6061,
      "step": 6975
    },
    {
      "epoch": 1.2566539067777527,
      "grad_norm": 1.3115757703781128,
      "learning_rate": 3.042839755503949e-05,
      "loss": 1.2508,
      "step": 6976
    },
    {
      "epoch": 1.256834046385949,
      "grad_norm": 1.4423494338989258,
      "learning_rate": 3.0415366805790317e-05,
      "loss": 1.3977,
      "step": 6977
    },
    {
      "epoch": 1.2570141859941455,
      "grad_norm": 1.3164818286895752,
      "learning_rate": 3.0402337627610865e-05,
      "loss": 1.2694,
      "step": 6978
    },
    {
      "epoch": 1.2571943256023417,
      "grad_norm": 1.4696060419082642,
      "learning_rate": 3.0389310021546358e-05,
      "loss": 1.3781,
      "step": 6979
    },
    {
      "epoch": 1.257374465210538,
      "grad_norm": 1.4752376079559326,
      "learning_rate": 3.037628398864184e-05,
      "loss": 1.468,
      "step": 6980
    },
    {
      "epoch": 1.2575546048187345,
      "grad_norm": 1.4918098449707031,
      "learning_rate": 3.0363259529942273e-05,
      "loss": 1.42,
      "step": 6981
    },
    {
      "epoch": 1.257734744426931,
      "grad_norm": 1.4050174951553345,
      "learning_rate": 3.0350236646492457e-05,
      "loss": 1.4017,
      "step": 6982
    },
    {
      "epoch": 1.2579148840351273,
      "grad_norm": 1.383993148803711,
      "learning_rate": 3.0337215339337078e-05,
      "loss": 1.53,
      "step": 6983
    },
    {
      "epoch": 1.2580950236433237,
      "grad_norm": 1.379915475845337,
      "learning_rate": 3.032419560952071e-05,
      "loss": 1.1459,
      "step": 6984
    },
    {
      "epoch": 1.25827516325152,
      "grad_norm": 1.3971757888793945,
      "learning_rate": 3.031117745808779e-05,
      "loss": 1.3178,
      "step": 6985
    },
    {
      "epoch": 1.2584553028597163,
      "grad_norm": 1.4547984600067139,
      "learning_rate": 3.0298160886082628e-05,
      "loss": 1.3794,
      "step": 6986
    },
    {
      "epoch": 1.2586354424679127,
      "grad_norm": 1.431549072265625,
      "learning_rate": 3.0285145894549377e-05,
      "loss": 1.3986,
      "step": 6987
    },
    {
      "epoch": 1.2588155820761089,
      "grad_norm": 1.49636971950531,
      "learning_rate": 3.0272132484532145e-05,
      "loss": 1.5619,
      "step": 6988
    },
    {
      "epoch": 1.2589957216843053,
      "grad_norm": 1.364831805229187,
      "learning_rate": 3.0259120657074846e-05,
      "loss": 1.1707,
      "step": 6989
    },
    {
      "epoch": 1.2591758612925017,
      "grad_norm": 1.6209914684295654,
      "learning_rate": 3.0246110413221258e-05,
      "loss": 1.6512,
      "step": 6990
    },
    {
      "epoch": 1.259356000900698,
      "grad_norm": 1.5467591285705566,
      "learning_rate": 3.023310175401507e-05,
      "loss": 1.5577,
      "step": 6991
    },
    {
      "epoch": 1.2595361405088945,
      "grad_norm": 1.3463491201400757,
      "learning_rate": 3.022009468049982e-05,
      "loss": 1.242,
      "step": 6992
    },
    {
      "epoch": 1.2597162801170907,
      "grad_norm": 1.4590662717819214,
      "learning_rate": 3.0207089193718953e-05,
      "loss": 1.468,
      "step": 6993
    },
    {
      "epoch": 1.259896419725287,
      "grad_norm": 1.480472445487976,
      "learning_rate": 3.019408529471575e-05,
      "loss": 1.4001,
      "step": 6994
    },
    {
      "epoch": 1.2600765593334835,
      "grad_norm": 1.5388211011886597,
      "learning_rate": 3.0181082984533393e-05,
      "loss": 1.2213,
      "step": 6995
    },
    {
      "epoch": 1.2602566989416797,
      "grad_norm": 1.40041983127594,
      "learning_rate": 3.0168082264214893e-05,
      "loss": 1.3281,
      "step": 6996
    },
    {
      "epoch": 1.260436838549876,
      "grad_norm": 1.4714363813400269,
      "learning_rate": 3.015508313480315e-05,
      "loss": 1.4441,
      "step": 6997
    },
    {
      "epoch": 1.2606169781580725,
      "grad_norm": 1.4824481010437012,
      "learning_rate": 3.0142085597340997e-05,
      "loss": 1.22,
      "step": 6998
    },
    {
      "epoch": 1.260797117766269,
      "grad_norm": 1.4296058416366577,
      "learning_rate": 3.0129089652871064e-05,
      "loss": 1.3198,
      "step": 6999
    },
    {
      "epoch": 1.2609772573744653,
      "grad_norm": 1.6829460859298706,
      "learning_rate": 3.0116095302435876e-05,
      "loss": 1.3227,
      "step": 7000
    },
    {
      "epoch": 1.2611573969826615,
      "grad_norm": 1.3450504541397095,
      "learning_rate": 3.0103102547077834e-05,
      "loss": 1.1022,
      "step": 7001
    },
    {
      "epoch": 1.261337536590858,
      "grad_norm": 1.2480241060256958,
      "learning_rate": 3.0090111387839214e-05,
      "loss": 1.7888,
      "step": 7002
    },
    {
      "epoch": 1.2615176761990543,
      "grad_norm": 1.3191943168640137,
      "learning_rate": 3.0077121825762172e-05,
      "loss": 1.989,
      "step": 7003
    },
    {
      "epoch": 1.2616978158072505,
      "grad_norm": 1.4725444316864014,
      "learning_rate": 3.0064133861888712e-05,
      "loss": 2.083,
      "step": 7004
    },
    {
      "epoch": 1.261877955415447,
      "grad_norm": 1.3447837829589844,
      "learning_rate": 3.0051147497260717e-05,
      "loss": 1.8028,
      "step": 7005
    },
    {
      "epoch": 1.2620580950236433,
      "grad_norm": 1.4679718017578125,
      "learning_rate": 3.0038162732919946e-05,
      "loss": 1.8906,
      "step": 7006
    },
    {
      "epoch": 1.2622382346318397,
      "grad_norm": 1.4556591510772705,
      "learning_rate": 3.0025179569908047e-05,
      "loss": 1.6007,
      "step": 7007
    },
    {
      "epoch": 1.2624183742400361,
      "grad_norm": 1.4762115478515625,
      "learning_rate": 3.001219800926651e-05,
      "loss": 1.7174,
      "step": 7008
    },
    {
      "epoch": 1.2625985138482325,
      "grad_norm": 1.5539400577545166,
      "learning_rate": 2.9999218052036722e-05,
      "loss": 1.8245,
      "step": 7009
    },
    {
      "epoch": 1.2627786534564287,
      "grad_norm": 1.7731585502624512,
      "learning_rate": 2.9986239699259906e-05,
      "loss": 1.9254,
      "step": 7010
    },
    {
      "epoch": 1.262958793064625,
      "grad_norm": 1.9975221157073975,
      "learning_rate": 2.9973262951977198e-05,
      "loss": 2.1392,
      "step": 7011
    },
    {
      "epoch": 1.2631389326728215,
      "grad_norm": 1.57728111743927,
      "learning_rate": 2.9960287811229586e-05,
      "loss": 1.6129,
      "step": 7012
    },
    {
      "epoch": 1.2633190722810177,
      "grad_norm": 1.324569582939148,
      "learning_rate": 2.9947314278057927e-05,
      "loss": 1.3379,
      "step": 7013
    },
    {
      "epoch": 1.263499211889214,
      "grad_norm": 1.3974930047988892,
      "learning_rate": 2.9934342353502943e-05,
      "loss": 1.5554,
      "step": 7014
    },
    {
      "epoch": 1.2636793514974105,
      "grad_norm": 1.2897553443908691,
      "learning_rate": 2.9921372038605232e-05,
      "loss": 1.2555,
      "step": 7015
    },
    {
      "epoch": 1.263859491105607,
      "grad_norm": 1.3799039125442505,
      "learning_rate": 2.9908403334405295e-05,
      "loss": 1.3825,
      "step": 7016
    },
    {
      "epoch": 1.2640396307138033,
      "grad_norm": 1.441825032234192,
      "learning_rate": 2.989543624194345e-05,
      "loss": 1.5057,
      "step": 7017
    },
    {
      "epoch": 1.2642197703219995,
      "grad_norm": 1.4043766260147095,
      "learning_rate": 2.9882470762259917e-05,
      "loss": 1.3594,
      "step": 7018
    },
    {
      "epoch": 1.264399909930196,
      "grad_norm": 1.2256760597229004,
      "learning_rate": 2.9869506896394784e-05,
      "loss": 1.1936,
      "step": 7019
    },
    {
      "epoch": 1.2645800495383923,
      "grad_norm": 1.401892900466919,
      "learning_rate": 2.9856544645387974e-05,
      "loss": 1.4807,
      "step": 7020
    },
    {
      "epoch": 1.2647601891465885,
      "grad_norm": 1.5168848037719727,
      "learning_rate": 2.9843584010279378e-05,
      "loss": 1.5941,
      "step": 7021
    },
    {
      "epoch": 1.264940328754785,
      "grad_norm": 1.2363992929458618,
      "learning_rate": 2.9830624992108645e-05,
      "loss": 1.0569,
      "step": 7022
    },
    {
      "epoch": 1.2651204683629813,
      "grad_norm": 1.335463523864746,
      "learning_rate": 2.9817667591915345e-05,
      "loss": 1.414,
      "step": 7023
    },
    {
      "epoch": 1.2653006079711777,
      "grad_norm": 1.3160192966461182,
      "learning_rate": 2.9804711810738906e-05,
      "loss": 1.2895,
      "step": 7024
    },
    {
      "epoch": 1.2654807475793741,
      "grad_norm": 1.521207332611084,
      "learning_rate": 2.979175764961867e-05,
      "loss": 1.3923,
      "step": 7025
    },
    {
      "epoch": 1.2656608871875703,
      "grad_norm": 1.4431427717208862,
      "learning_rate": 2.977880510959379e-05,
      "loss": 1.3236,
      "step": 7026
    },
    {
      "epoch": 1.2658410267957667,
      "grad_norm": 1.5625979900360107,
      "learning_rate": 2.9765854191703326e-05,
      "loss": 1.6278,
      "step": 7027
    },
    {
      "epoch": 1.2660211664039631,
      "grad_norm": 1.393539309501648,
      "learning_rate": 2.975290489698617e-05,
      "loss": 1.4421,
      "step": 7028
    },
    {
      "epoch": 1.2662013060121593,
      "grad_norm": 1.6322964429855347,
      "learning_rate": 2.97399572264811e-05,
      "loss": 1.4769,
      "step": 7029
    },
    {
      "epoch": 1.2663814456203557,
      "grad_norm": 1.3377214670181274,
      "learning_rate": 2.9727011181226818e-05,
      "loss": 1.3165,
      "step": 7030
    },
    {
      "epoch": 1.2665615852285521,
      "grad_norm": 1.4006478786468506,
      "learning_rate": 2.9714066762261823e-05,
      "loss": 1.299,
      "step": 7031
    },
    {
      "epoch": 1.2667417248367485,
      "grad_norm": 1.383147120475769,
      "learning_rate": 2.9701123970624513e-05,
      "loss": 1.3153,
      "step": 7032
    },
    {
      "epoch": 1.266921864444945,
      "grad_norm": 1.3741413354873657,
      "learning_rate": 2.9688182807353156e-05,
      "loss": 1.3696,
      "step": 7033
    },
    {
      "epoch": 1.267102004053141,
      "grad_norm": 1.4689764976501465,
      "learning_rate": 2.9675243273485852e-05,
      "loss": 1.3693,
      "step": 7034
    },
    {
      "epoch": 1.2672821436613375,
      "grad_norm": 1.4250136613845825,
      "learning_rate": 2.966230537006065e-05,
      "loss": 1.5007,
      "step": 7035
    },
    {
      "epoch": 1.267462283269534,
      "grad_norm": 1.639159917831421,
      "learning_rate": 2.964936909811541e-05,
      "loss": 1.5369,
      "step": 7036
    },
    {
      "epoch": 1.2676424228777303,
      "grad_norm": 1.4493772983551025,
      "learning_rate": 2.9636434458687866e-05,
      "loss": 1.4696,
      "step": 7037
    },
    {
      "epoch": 1.2678225624859265,
      "grad_norm": 1.5093430280685425,
      "learning_rate": 2.9623501452815626e-05,
      "loss": 1.5496,
      "step": 7038
    },
    {
      "epoch": 1.268002702094123,
      "grad_norm": 1.5053790807724,
      "learning_rate": 2.9610570081536183e-05,
      "loss": 1.3887,
      "step": 7039
    },
    {
      "epoch": 1.2681828417023193,
      "grad_norm": 1.7349344491958618,
      "learning_rate": 2.959764034588687e-05,
      "loss": 1.1887,
      "step": 7040
    },
    {
      "epoch": 1.2683629813105157,
      "grad_norm": 1.6846961975097656,
      "learning_rate": 2.9584712246904918e-05,
      "loss": 1.5954,
      "step": 7041
    },
    {
      "epoch": 1.2685431209187121,
      "grad_norm": 1.5604982376098633,
      "learning_rate": 2.9571785785627405e-05,
      "loss": 1.6034,
      "step": 7042
    },
    {
      "epoch": 1.2687232605269083,
      "grad_norm": 1.5936533212661743,
      "learning_rate": 2.9558860963091274e-05,
      "loss": 1.4396,
      "step": 7043
    },
    {
      "epoch": 1.2689034001351047,
      "grad_norm": 1.3839609622955322,
      "learning_rate": 2.9545937780333367e-05,
      "loss": 1.3386,
      "step": 7044
    },
    {
      "epoch": 1.2690835397433011,
      "grad_norm": 1.5781564712524414,
      "learning_rate": 2.953301623839037e-05,
      "loss": 1.4553,
      "step": 7045
    },
    {
      "epoch": 1.2692636793514973,
      "grad_norm": 1.4864450693130493,
      "learning_rate": 2.952009633829884e-05,
      "loss": 1.4066,
      "step": 7046
    },
    {
      "epoch": 1.2694438189596937,
      "grad_norm": 1.6834286451339722,
      "learning_rate": 2.950717808109521e-05,
      "loss": 1.5763,
      "step": 7047
    },
    {
      "epoch": 1.2696239585678901,
      "grad_norm": 1.5803136825561523,
      "learning_rate": 2.9494261467815765e-05,
      "loss": 1.5549,
      "step": 7048
    },
    {
      "epoch": 1.2698040981760865,
      "grad_norm": 1.354555606842041,
      "learning_rate": 2.9481346499496682e-05,
      "loss": 1.0512,
      "step": 7049
    },
    {
      "epoch": 1.269984237784283,
      "grad_norm": 1.3255127668380737,
      "learning_rate": 2.9468433177173994e-05,
      "loss": 1.1053,
      "step": 7050
    },
    {
      "epoch": 1.2701643773924791,
      "grad_norm": 1.2747818231582642,
      "learning_rate": 2.9455521501883586e-05,
      "loss": 1.2028,
      "step": 7051
    },
    {
      "epoch": 1.2703445170006755,
      "grad_norm": 1.291144609451294,
      "learning_rate": 2.9442611474661226e-05,
      "loss": 1.7835,
      "step": 7052
    },
    {
      "epoch": 1.270524656608872,
      "grad_norm": 1.305700659751892,
      "learning_rate": 2.942970309654258e-05,
      "loss": 1.7217,
      "step": 7053
    },
    {
      "epoch": 1.270704796217068,
      "grad_norm": 1.4463049173355103,
      "learning_rate": 2.9416796368563136e-05,
      "loss": 2.2301,
      "step": 7054
    },
    {
      "epoch": 1.2708849358252645,
      "grad_norm": 1.3374251127243042,
      "learning_rate": 2.9403891291758256e-05,
      "loss": 1.8841,
      "step": 7055
    },
    {
      "epoch": 1.271065075433461,
      "grad_norm": 1.3790265321731567,
      "learning_rate": 2.9390987867163178e-05,
      "loss": 1.8791,
      "step": 7056
    },
    {
      "epoch": 1.2712452150416573,
      "grad_norm": 1.3896771669387817,
      "learning_rate": 2.9378086095813007e-05,
      "loss": 1.7174,
      "step": 7057
    },
    {
      "epoch": 1.2714253546498537,
      "grad_norm": 1.5280636548995972,
      "learning_rate": 2.936518597874274e-05,
      "loss": 1.606,
      "step": 7058
    },
    {
      "epoch": 1.27160549425805,
      "grad_norm": 1.5006309747695923,
      "learning_rate": 2.9352287516987198e-05,
      "loss": 1.7131,
      "step": 7059
    },
    {
      "epoch": 1.2717856338662463,
      "grad_norm": 1.5878580808639526,
      "learning_rate": 2.9339390711581105e-05,
      "loss": 1.9328,
      "step": 7060
    },
    {
      "epoch": 1.2719657734744427,
      "grad_norm": 1.8862226009368896,
      "learning_rate": 2.932649556355902e-05,
      "loss": 1.7538,
      "step": 7061
    },
    {
      "epoch": 1.272145913082639,
      "grad_norm": 1.4710617065429688,
      "learning_rate": 2.9313602073955364e-05,
      "loss": 1.5457,
      "step": 7062
    },
    {
      "epoch": 1.2723260526908353,
      "grad_norm": 1.3599189519882202,
      "learning_rate": 2.9300710243804498e-05,
      "loss": 1.5432,
      "step": 7063
    },
    {
      "epoch": 1.2725061922990317,
      "grad_norm": 1.2542777061462402,
      "learning_rate": 2.9287820074140576e-05,
      "loss": 1.3678,
      "step": 7064
    },
    {
      "epoch": 1.2726863319072281,
      "grad_norm": 1.3461740016937256,
      "learning_rate": 2.9274931565997642e-05,
      "loss": 1.4137,
      "step": 7065
    },
    {
      "epoch": 1.2728664715154245,
      "grad_norm": 1.3523504734039307,
      "learning_rate": 2.9262044720409588e-05,
      "loss": 1.3191,
      "step": 7066
    },
    {
      "epoch": 1.273046611123621,
      "grad_norm": 1.3251413106918335,
      "learning_rate": 2.9249159538410214e-05,
      "loss": 1.3222,
      "step": 7067
    },
    {
      "epoch": 1.2732267507318171,
      "grad_norm": 1.2500574588775635,
      "learning_rate": 2.923627602103316e-05,
      "loss": 1.2112,
      "step": 7068
    },
    {
      "epoch": 1.2734068903400135,
      "grad_norm": 1.4667209386825562,
      "learning_rate": 2.9223394169311924e-05,
      "loss": 1.4481,
      "step": 7069
    },
    {
      "epoch": 1.27358702994821,
      "grad_norm": 1.368598222732544,
      "learning_rate": 2.9210513984279884e-05,
      "loss": 1.3773,
      "step": 7070
    },
    {
      "epoch": 1.2737671695564061,
      "grad_norm": 1.486831545829773,
      "learning_rate": 2.9197635466970286e-05,
      "loss": 1.4937,
      "step": 7071
    },
    {
      "epoch": 1.2739473091646025,
      "grad_norm": 1.4855842590332031,
      "learning_rate": 2.9184758618416248e-05,
      "loss": 1.6067,
      "step": 7072
    },
    {
      "epoch": 1.274127448772799,
      "grad_norm": 1.3872230052947998,
      "learning_rate": 2.9171883439650728e-05,
      "loss": 1.4845,
      "step": 7073
    },
    {
      "epoch": 1.2743075883809953,
      "grad_norm": 1.47398042678833,
      "learning_rate": 2.9159009931706583e-05,
      "loss": 1.3943,
      "step": 7074
    },
    {
      "epoch": 1.2744877279891917,
      "grad_norm": 1.4635913372039795,
      "learning_rate": 2.9146138095616504e-05,
      "loss": 1.3138,
      "step": 7075
    },
    {
      "epoch": 1.274667867597388,
      "grad_norm": 1.5388904809951782,
      "learning_rate": 2.9133267932413046e-05,
      "loss": 1.5103,
      "step": 7076
    },
    {
      "epoch": 1.2748480072055843,
      "grad_norm": 1.4420770406723022,
      "learning_rate": 2.9120399443128692e-05,
      "loss": 1.3431,
      "step": 7077
    },
    {
      "epoch": 1.2750281468137807,
      "grad_norm": 1.5094555616378784,
      "learning_rate": 2.910753262879573e-05,
      "loss": 1.4352,
      "step": 7078
    },
    {
      "epoch": 1.275208286421977,
      "grad_norm": 1.5543296337127686,
      "learning_rate": 2.9094667490446338e-05,
      "loss": 1.6147,
      "step": 7079
    },
    {
      "epoch": 1.2753884260301733,
      "grad_norm": 1.4683845043182373,
      "learning_rate": 2.9081804029112503e-05,
      "loss": 1.3719,
      "step": 7080
    },
    {
      "epoch": 1.2755685656383697,
      "grad_norm": 1.3315753936767578,
      "learning_rate": 2.9068942245826177e-05,
      "loss": 1.4237,
      "step": 7081
    },
    {
      "epoch": 1.2757487052465661,
      "grad_norm": 1.4258633852005005,
      "learning_rate": 2.9056082141619106e-05,
      "loss": 1.3707,
      "step": 7082
    },
    {
      "epoch": 1.2759288448547625,
      "grad_norm": 1.4926718473434448,
      "learning_rate": 2.9043223717522927e-05,
      "loss": 1.526,
      "step": 7083
    },
    {
      "epoch": 1.2761089844629587,
      "grad_norm": 1.361735224723816,
      "learning_rate": 2.9030366974569136e-05,
      "loss": 1.2437,
      "step": 7084
    },
    {
      "epoch": 1.2762891240711551,
      "grad_norm": 1.4047073125839233,
      "learning_rate": 2.901751191378907e-05,
      "loss": 1.4558,
      "step": 7085
    },
    {
      "epoch": 1.2764692636793515,
      "grad_norm": 1.6143884658813477,
      "learning_rate": 2.9004658536213996e-05,
      "loss": 1.4707,
      "step": 7086
    },
    {
      "epoch": 1.2766494032875477,
      "grad_norm": 1.6838817596435547,
      "learning_rate": 2.899180684287499e-05,
      "loss": 1.6467,
      "step": 7087
    },
    {
      "epoch": 1.2768295428957441,
      "grad_norm": 1.6498833894729614,
      "learning_rate": 2.8978956834803006e-05,
      "loss": 1.4667,
      "step": 7088
    },
    {
      "epoch": 1.2770096825039405,
      "grad_norm": 1.5781311988830566,
      "learning_rate": 2.8966108513028868e-05,
      "loss": 1.263,
      "step": 7089
    },
    {
      "epoch": 1.277189822112137,
      "grad_norm": 1.5843472480773926,
      "learning_rate": 2.895326187858326e-05,
      "loss": 1.3936,
      "step": 7090
    },
    {
      "epoch": 1.2773699617203333,
      "grad_norm": 1.3745026588439941,
      "learning_rate": 2.8940416932496727e-05,
      "loss": 1.2256,
      "step": 7091
    },
    {
      "epoch": 1.2775501013285295,
      "grad_norm": 1.5958194732666016,
      "learning_rate": 2.8927573675799697e-05,
      "loss": 1.316,
      "step": 7092
    },
    {
      "epoch": 1.277730240936726,
      "grad_norm": 1.5327242612838745,
      "learning_rate": 2.8914732109522434e-05,
      "loss": 1.5621,
      "step": 7093
    },
    {
      "epoch": 1.2779103805449223,
      "grad_norm": 1.5197958946228027,
      "learning_rate": 2.8901892234695083e-05,
      "loss": 1.2194,
      "step": 7094
    },
    {
      "epoch": 1.2780905201531187,
      "grad_norm": 1.63090980052948,
      "learning_rate": 2.8889054052347668e-05,
      "loss": 1.5198,
      "step": 7095
    },
    {
      "epoch": 1.278270659761315,
      "grad_norm": 1.5595170259475708,
      "learning_rate": 2.887621756351006e-05,
      "loss": 1.5265,
      "step": 7096
    },
    {
      "epoch": 1.2784507993695113,
      "grad_norm": 1.6263370513916016,
      "learning_rate": 2.8863382769211988e-05,
      "loss": 1.4035,
      "step": 7097
    },
    {
      "epoch": 1.2786309389777077,
      "grad_norm": 1.5030648708343506,
      "learning_rate": 2.8850549670483063e-05,
      "loss": 1.1608,
      "step": 7098
    },
    {
      "epoch": 1.2788110785859041,
      "grad_norm": 1.4987504482269287,
      "learning_rate": 2.8837718268352732e-05,
      "loss": 1.307,
      "step": 7099
    },
    {
      "epoch": 1.2789912181941006,
      "grad_norm": 1.453192114830017,
      "learning_rate": 2.8824888563850337e-05,
      "loss": 1.4041,
      "step": 7100
    },
    {
      "epoch": 1.2791713578022967,
      "grad_norm": 1.5771466493606567,
      "learning_rate": 2.881206055800507e-05,
      "loss": 1.3558,
      "step": 7101
    },
    {
      "epoch": 1.2793514974104931,
      "grad_norm": 1.2904399633407593,
      "learning_rate": 2.879923425184598e-05,
      "loss": 1.8712,
      "step": 7102
    },
    {
      "epoch": 1.2795316370186895,
      "grad_norm": 1.2609823942184448,
      "learning_rate": 2.8786409646401997e-05,
      "loss": 1.7608,
      "step": 7103
    },
    {
      "epoch": 1.2797117766268857,
      "grad_norm": 1.3678944110870361,
      "learning_rate": 2.8773586742701873e-05,
      "loss": 1.8384,
      "step": 7104
    },
    {
      "epoch": 1.2798919162350821,
      "grad_norm": 1.3234736919403076,
      "learning_rate": 2.8760765541774303e-05,
      "loss": 1.8296,
      "step": 7105
    },
    {
      "epoch": 1.2800720558432785,
      "grad_norm": 1.474744200706482,
      "learning_rate": 2.8747946044647776e-05,
      "loss": 1.7938,
      "step": 7106
    },
    {
      "epoch": 1.280252195451475,
      "grad_norm": 1.4485154151916504,
      "learning_rate": 2.8735128252350673e-05,
      "loss": 1.6593,
      "step": 7107
    },
    {
      "epoch": 1.2804323350596714,
      "grad_norm": 1.4289631843566895,
      "learning_rate": 2.8722312165911204e-05,
      "loss": 1.7681,
      "step": 7108
    },
    {
      "epoch": 1.2806124746678675,
      "grad_norm": 1.5767881870269775,
      "learning_rate": 2.87094977863575e-05,
      "loss": 1.671,
      "step": 7109
    },
    {
      "epoch": 1.280792614276064,
      "grad_norm": 1.5129060745239258,
      "learning_rate": 2.8696685114717516e-05,
      "loss": 1.4741,
      "step": 7110
    },
    {
      "epoch": 1.2809727538842604,
      "grad_norm": 1.9507790803909302,
      "learning_rate": 2.8683874152019097e-05,
      "loss": 1.9831,
      "step": 7111
    },
    {
      "epoch": 1.2811528934924565,
      "grad_norm": 1.731133222579956,
      "learning_rate": 2.867106489928989e-05,
      "loss": 1.8499,
      "step": 7112
    },
    {
      "epoch": 1.281333033100653,
      "grad_norm": 1.4464892148971558,
      "learning_rate": 2.865825735755745e-05,
      "loss": 1.4283,
      "step": 7113
    },
    {
      "epoch": 1.2815131727088493,
      "grad_norm": 1.3960230350494385,
      "learning_rate": 2.8645451527849226e-05,
      "loss": 1.2875,
      "step": 7114
    },
    {
      "epoch": 1.2816933123170458,
      "grad_norm": 1.3529244661331177,
      "learning_rate": 2.8632647411192483e-05,
      "loss": 1.3898,
      "step": 7115
    },
    {
      "epoch": 1.2818734519252422,
      "grad_norm": 1.2842944860458374,
      "learning_rate": 2.8619845008614355e-05,
      "loss": 1.3809,
      "step": 7116
    },
    {
      "epoch": 1.2820535915334383,
      "grad_norm": 1.2405726909637451,
      "learning_rate": 2.8607044321141825e-05,
      "loss": 1.2132,
      "step": 7117
    },
    {
      "epoch": 1.2822337311416347,
      "grad_norm": 1.3347744941711426,
      "learning_rate": 2.8594245349801803e-05,
      "loss": 1.1413,
      "step": 7118
    },
    {
      "epoch": 1.2824138707498312,
      "grad_norm": 1.3706387281417847,
      "learning_rate": 2.858144809562099e-05,
      "loss": 1.452,
      "step": 7119
    },
    {
      "epoch": 1.2825940103580273,
      "grad_norm": 1.2984391450881958,
      "learning_rate": 2.8568652559625976e-05,
      "loss": 1.1992,
      "step": 7120
    },
    {
      "epoch": 1.2827741499662237,
      "grad_norm": 1.4117110967636108,
      "learning_rate": 2.8555858742843223e-05,
      "loss": 1.3602,
      "step": 7121
    },
    {
      "epoch": 1.2829542895744201,
      "grad_norm": 1.4401477575302124,
      "learning_rate": 2.8543066646299025e-05,
      "loss": 1.3341,
      "step": 7122
    },
    {
      "epoch": 1.2831344291826166,
      "grad_norm": 1.5523301362991333,
      "learning_rate": 2.8530276271019573e-05,
      "loss": 1.4858,
      "step": 7123
    },
    {
      "epoch": 1.283314568790813,
      "grad_norm": 1.37791907787323,
      "learning_rate": 2.851748761803091e-05,
      "loss": 1.2366,
      "step": 7124
    },
    {
      "epoch": 1.2834947083990094,
      "grad_norm": 1.5780267715454102,
      "learning_rate": 2.850470068835892e-05,
      "loss": 1.6152,
      "step": 7125
    },
    {
      "epoch": 1.2836748480072055,
      "grad_norm": 1.437329649925232,
      "learning_rate": 2.849191548302937e-05,
      "loss": 1.3599,
      "step": 7126
    },
    {
      "epoch": 1.283854987615402,
      "grad_norm": 1.451629877090454,
      "learning_rate": 2.8479132003067866e-05,
      "loss": 1.3958,
      "step": 7127
    },
    {
      "epoch": 1.2840351272235984,
      "grad_norm": 1.5319790840148926,
      "learning_rate": 2.846635024949993e-05,
      "loss": 1.3025,
      "step": 7128
    },
    {
      "epoch": 1.2842152668317945,
      "grad_norm": 1.39559805393219,
      "learning_rate": 2.8453570223350885e-05,
      "loss": 1.3347,
      "step": 7129
    },
    {
      "epoch": 1.284395406439991,
      "grad_norm": 1.4030119180679321,
      "learning_rate": 2.8440791925645942e-05,
      "loss": 1.2593,
      "step": 7130
    },
    {
      "epoch": 1.2845755460481874,
      "grad_norm": 1.4799177646636963,
      "learning_rate": 2.842801535741017e-05,
      "loss": 1.3758,
      "step": 7131
    },
    {
      "epoch": 1.2847556856563838,
      "grad_norm": 1.396166205406189,
      "learning_rate": 2.84152405196685e-05,
      "loss": 1.2077,
      "step": 7132
    },
    {
      "epoch": 1.2849358252645802,
      "grad_norm": 1.44206702709198,
      "learning_rate": 2.840246741344572e-05,
      "loss": 1.4144,
      "step": 7133
    },
    {
      "epoch": 1.2851159648727764,
      "grad_norm": 1.60263192653656,
      "learning_rate": 2.838969603976649e-05,
      "loss": 1.611,
      "step": 7134
    },
    {
      "epoch": 1.2852961044809728,
      "grad_norm": 1.6154299974441528,
      "learning_rate": 2.8376926399655312e-05,
      "loss": 1.6654,
      "step": 7135
    },
    {
      "epoch": 1.2854762440891692,
      "grad_norm": 1.5466952323913574,
      "learning_rate": 2.8364158494136552e-05,
      "loss": 1.5981,
      "step": 7136
    },
    {
      "epoch": 1.2856563836973653,
      "grad_norm": 1.4793764352798462,
      "learning_rate": 2.835139232423447e-05,
      "loss": 1.2664,
      "step": 7137
    },
    {
      "epoch": 1.2858365233055618,
      "grad_norm": 1.489059329032898,
      "learning_rate": 2.8338627890973158e-05,
      "loss": 1.2028,
      "step": 7138
    },
    {
      "epoch": 1.2860166629137582,
      "grad_norm": 1.7054269313812256,
      "learning_rate": 2.8325865195376566e-05,
      "loss": 1.65,
      "step": 7139
    },
    {
      "epoch": 1.2861968025219546,
      "grad_norm": 1.7295113801956177,
      "learning_rate": 2.8313104238468512e-05,
      "loss": 1.5,
      "step": 7140
    },
    {
      "epoch": 1.286376942130151,
      "grad_norm": 1.5239750146865845,
      "learning_rate": 2.8300345021272643e-05,
      "loss": 1.306,
      "step": 7141
    },
    {
      "epoch": 1.2865570817383472,
      "grad_norm": 1.3846197128295898,
      "learning_rate": 2.8287587544812555e-05,
      "loss": 1.1421,
      "step": 7142
    },
    {
      "epoch": 1.2867372213465436,
      "grad_norm": 1.5397863388061523,
      "learning_rate": 2.8274831810111625e-05,
      "loss": 1.5609,
      "step": 7143
    },
    {
      "epoch": 1.28691736095474,
      "grad_norm": 1.4834880828857422,
      "learning_rate": 2.826207781819311e-05,
      "loss": 1.213,
      "step": 7144
    },
    {
      "epoch": 1.2870975005629361,
      "grad_norm": 1.5617412328720093,
      "learning_rate": 2.824932557008009e-05,
      "loss": 1.4283,
      "step": 7145
    },
    {
      "epoch": 1.2872776401711326,
      "grad_norm": 1.2447694540023804,
      "learning_rate": 2.82365750667956e-05,
      "loss": 0.9551,
      "step": 7146
    },
    {
      "epoch": 1.287457779779329,
      "grad_norm": 1.4038580656051636,
      "learning_rate": 2.822382630936246e-05,
      "loss": 1.2449,
      "step": 7147
    },
    {
      "epoch": 1.2876379193875254,
      "grad_norm": 1.379562258720398,
      "learning_rate": 2.8211079298803365e-05,
      "loss": 1.2643,
      "step": 7148
    },
    {
      "epoch": 1.2878180589957218,
      "grad_norm": 1.6397968530654907,
      "learning_rate": 2.8198334036140874e-05,
      "loss": 1.2982,
      "step": 7149
    },
    {
      "epoch": 1.287998198603918,
      "grad_norm": 1.5758934020996094,
      "learning_rate": 2.8185590522397388e-05,
      "loss": 1.536,
      "step": 7150
    },
    {
      "epoch": 1.2881783382121144,
      "grad_norm": 1.4082257747650146,
      "learning_rate": 2.8172848758595227e-05,
      "loss": 1.1042,
      "step": 7151
    },
    {
      "epoch": 1.2883584778203108,
      "grad_norm": 1.3642839193344116,
      "learning_rate": 2.8160108745756508e-05,
      "loss": 1.7206,
      "step": 7152
    },
    {
      "epoch": 1.2885386174285072,
      "grad_norm": 1.303656816482544,
      "learning_rate": 2.8147370484903224e-05,
      "loss": 1.5467,
      "step": 7153
    },
    {
      "epoch": 1.2887187570367034,
      "grad_norm": 1.3603367805480957,
      "learning_rate": 2.8134633977057235e-05,
      "loss": 1.8569,
      "step": 7154
    },
    {
      "epoch": 1.2888988966448998,
      "grad_norm": 1.3758049011230469,
      "learning_rate": 2.8121899223240256e-05,
      "loss": 1.7634,
      "step": 7155
    },
    {
      "epoch": 1.2890790362530962,
      "grad_norm": 1.4273656606674194,
      "learning_rate": 2.810916622447387e-05,
      "loss": 1.7321,
      "step": 7156
    },
    {
      "epoch": 1.2892591758612926,
      "grad_norm": 1.3923115730285645,
      "learning_rate": 2.8096434981779508e-05,
      "loss": 1.7377,
      "step": 7157
    },
    {
      "epoch": 1.289439315469489,
      "grad_norm": 1.323893427848816,
      "learning_rate": 2.8083705496178457e-05,
      "loss": 1.4578,
      "step": 7158
    },
    {
      "epoch": 1.2896194550776852,
      "grad_norm": 1.5742992162704468,
      "learning_rate": 2.8070977768691863e-05,
      "loss": 1.6492,
      "step": 7159
    },
    {
      "epoch": 1.2897995946858816,
      "grad_norm": 1.5723830461502075,
      "learning_rate": 2.8058251800340762e-05,
      "loss": 1.6464,
      "step": 7160
    },
    {
      "epoch": 1.289979734294078,
      "grad_norm": 1.666792631149292,
      "learning_rate": 2.804552759214602e-05,
      "loss": 1.87,
      "step": 7161
    },
    {
      "epoch": 1.2901598739022742,
      "grad_norm": 1.5731443166732788,
      "learning_rate": 2.803280514512836e-05,
      "loss": 1.8203,
      "step": 7162
    },
    {
      "epoch": 1.2903400135104706,
      "grad_norm": 1.3674746751785278,
      "learning_rate": 2.8020084460308372e-05,
      "loss": 1.4863,
      "step": 7163
    },
    {
      "epoch": 1.290520153118667,
      "grad_norm": 1.3326148986816406,
      "learning_rate": 2.8007365538706503e-05,
      "loss": 1.3019,
      "step": 7164
    },
    {
      "epoch": 1.2907002927268634,
      "grad_norm": 1.3222126960754395,
      "learning_rate": 2.7994648381343058e-05,
      "loss": 1.2766,
      "step": 7165
    },
    {
      "epoch": 1.2908804323350598,
      "grad_norm": 1.3349494934082031,
      "learning_rate": 2.7981932989238203e-05,
      "loss": 1.2268,
      "step": 7166
    },
    {
      "epoch": 1.291060571943256,
      "grad_norm": 1.4164543151855469,
      "learning_rate": 2.796921936341196e-05,
      "loss": 1.4091,
      "step": 7167
    },
    {
      "epoch": 1.2912407115514524,
      "grad_norm": 1.3065745830535889,
      "learning_rate": 2.7956507504884198e-05,
      "loss": 1.5263,
      "step": 7168
    },
    {
      "epoch": 1.2914208511596488,
      "grad_norm": 1.3957542181015015,
      "learning_rate": 2.7943797414674656e-05,
      "loss": 1.3652,
      "step": 7169
    },
    {
      "epoch": 1.291600990767845,
      "grad_norm": 1.4634047746658325,
      "learning_rate": 2.7931089093802964e-05,
      "loss": 1.5178,
      "step": 7170
    },
    {
      "epoch": 1.2917811303760414,
      "grad_norm": 1.3141335248947144,
      "learning_rate": 2.7918382543288553e-05,
      "loss": 1.2781,
      "step": 7171
    },
    {
      "epoch": 1.2919612699842378,
      "grad_norm": 1.4111889600753784,
      "learning_rate": 2.790567776415074e-05,
      "loss": 1.3847,
      "step": 7172
    },
    {
      "epoch": 1.2921414095924342,
      "grad_norm": 1.3061233758926392,
      "learning_rate": 2.7892974757408673e-05,
      "loss": 1.2763,
      "step": 7173
    },
    {
      "epoch": 1.2923215492006306,
      "grad_norm": 1.4354208707809448,
      "learning_rate": 2.7880273524081423e-05,
      "loss": 1.3865,
      "step": 7174
    },
    {
      "epoch": 1.2925016888088268,
      "grad_norm": 1.4313865900039673,
      "learning_rate": 2.786757406518785e-05,
      "loss": 1.2978,
      "step": 7175
    },
    {
      "epoch": 1.2926818284170232,
      "grad_norm": 1.3059842586517334,
      "learning_rate": 2.7854876381746735e-05,
      "loss": 1.1552,
      "step": 7176
    },
    {
      "epoch": 1.2928619680252196,
      "grad_norm": 1.370823860168457,
      "learning_rate": 2.784218047477663e-05,
      "loss": 1.2846,
      "step": 7177
    },
    {
      "epoch": 1.2930421076334158,
      "grad_norm": 1.4305886030197144,
      "learning_rate": 2.782948634529599e-05,
      "loss": 1.4535,
      "step": 7178
    },
    {
      "epoch": 1.2932222472416122,
      "grad_norm": 1.459396481513977,
      "learning_rate": 2.7816793994323176e-05,
      "loss": 1.2932,
      "step": 7179
    },
    {
      "epoch": 1.2934023868498086,
      "grad_norm": 1.5988876819610596,
      "learning_rate": 2.780410342287635e-05,
      "loss": 1.5864,
      "step": 7180
    },
    {
      "epoch": 1.293582526458005,
      "grad_norm": 1.4623887538909912,
      "learning_rate": 2.7791414631973533e-05,
      "loss": 1.3431,
      "step": 7181
    },
    {
      "epoch": 1.2937626660662014,
      "grad_norm": 1.5250016450881958,
      "learning_rate": 2.7778727622632618e-05,
      "loss": 1.3522,
      "step": 7182
    },
    {
      "epoch": 1.2939428056743978,
      "grad_norm": 1.4584288597106934,
      "learning_rate": 2.7766042395871324e-05,
      "loss": 1.5749,
      "step": 7183
    },
    {
      "epoch": 1.294122945282594,
      "grad_norm": 1.3903683423995972,
      "learning_rate": 2.7753358952707297e-05,
      "loss": 1.1608,
      "step": 7184
    },
    {
      "epoch": 1.2943030848907904,
      "grad_norm": 1.421125888824463,
      "learning_rate": 2.7740677294157985e-05,
      "loss": 1.3196,
      "step": 7185
    },
    {
      "epoch": 1.2944832244989868,
      "grad_norm": 1.492103934288025,
      "learning_rate": 2.7727997421240697e-05,
      "loss": 1.4954,
      "step": 7186
    },
    {
      "epoch": 1.294663364107183,
      "grad_norm": 1.5505269765853882,
      "learning_rate": 2.771531933497261e-05,
      "loss": 1.3766,
      "step": 7187
    },
    {
      "epoch": 1.2948435037153794,
      "grad_norm": 1.4632881879806519,
      "learning_rate": 2.7702643036370747e-05,
      "loss": 1.2459,
      "step": 7188
    },
    {
      "epoch": 1.2950236433235758,
      "grad_norm": 1.5282247066497803,
      "learning_rate": 2.7689968526452003e-05,
      "loss": 1.5709,
      "step": 7189
    },
    {
      "epoch": 1.2952037829317722,
      "grad_norm": 1.4550132751464844,
      "learning_rate": 2.7677295806233118e-05,
      "loss": 1.4013,
      "step": 7190
    },
    {
      "epoch": 1.2953839225399686,
      "grad_norm": 1.7055833339691162,
      "learning_rate": 2.7664624876730683e-05,
      "loss": 1.4501,
      "step": 7191
    },
    {
      "epoch": 1.2955640621481648,
      "grad_norm": 1.4734947681427002,
      "learning_rate": 2.7651955738961144e-05,
      "loss": 1.3351,
      "step": 7192
    },
    {
      "epoch": 1.2957442017563612,
      "grad_norm": 1.4047929048538208,
      "learning_rate": 2.7639288393940853e-05,
      "loss": 1.2174,
      "step": 7193
    },
    {
      "epoch": 1.2959243413645576,
      "grad_norm": 1.8051222562789917,
      "learning_rate": 2.7626622842685955e-05,
      "loss": 1.6539,
      "step": 7194
    },
    {
      "epoch": 1.2961044809727538,
      "grad_norm": 1.5454577207565308,
      "learning_rate": 2.7613959086212475e-05,
      "loss": 1.4145,
      "step": 7195
    },
    {
      "epoch": 1.2962846205809502,
      "grad_norm": 1.4186687469482422,
      "learning_rate": 2.7601297125536295e-05,
      "loss": 1.2348,
      "step": 7196
    },
    {
      "epoch": 1.2964647601891466,
      "grad_norm": 1.3061327934265137,
      "learning_rate": 2.7588636961673148e-05,
      "loss": 1.0776,
      "step": 7197
    },
    {
      "epoch": 1.296644899797343,
      "grad_norm": 1.9398189783096313,
      "learning_rate": 2.7575978595638625e-05,
      "loss": 1.5619,
      "step": 7198
    },
    {
      "epoch": 1.2968250394055394,
      "grad_norm": 1.518505573272705,
      "learning_rate": 2.756332202844818e-05,
      "loss": 1.4203,
      "step": 7199
    },
    {
      "epoch": 1.2970051790137356,
      "grad_norm": 1.5813530683517456,
      "learning_rate": 2.7550667261117113e-05,
      "loss": 1.3887,
      "step": 7200
    },
    {
      "epoch": 1.297185318621932,
      "grad_norm": 1.55479097366333,
      "learning_rate": 2.7538014294660565e-05,
      "loss": 1.4567,
      "step": 7201
    },
    {
      "epoch": 1.2973654582301284,
      "grad_norm": 1.4022865295410156,
      "learning_rate": 2.7525363130093586e-05,
      "loss": 1.8743,
      "step": 7202
    },
    {
      "epoch": 1.2975455978383246,
      "grad_norm": 1.2930684089660645,
      "learning_rate": 2.751271376843103e-05,
      "loss": 1.8295,
      "step": 7203
    },
    {
      "epoch": 1.297725737446521,
      "grad_norm": 1.4828613996505737,
      "learning_rate": 2.7500066210687626e-05,
      "loss": 2.0995,
      "step": 7204
    },
    {
      "epoch": 1.2979058770547174,
      "grad_norm": 1.4192568063735962,
      "learning_rate": 2.7487420457877942e-05,
      "loss": 1.8038,
      "step": 7205
    },
    {
      "epoch": 1.2980860166629138,
      "grad_norm": 1.4027570486068726,
      "learning_rate": 2.7474776511016408e-05,
      "loss": 1.9283,
      "step": 7206
    },
    {
      "epoch": 1.2982661562711102,
      "grad_norm": 1.4486907720565796,
      "learning_rate": 2.7462134371117354e-05,
      "loss": 1.8553,
      "step": 7207
    },
    {
      "epoch": 1.2984462958793066,
      "grad_norm": 1.5106427669525146,
      "learning_rate": 2.74494940391949e-05,
      "loss": 1.567,
      "step": 7208
    },
    {
      "epoch": 1.2986264354875028,
      "grad_norm": 1.640448808670044,
      "learning_rate": 2.7436855516263073e-05,
      "loss": 2.043,
      "step": 7209
    },
    {
      "epoch": 1.2988065750956992,
      "grad_norm": 1.6078855991363525,
      "learning_rate": 2.742421880333569e-05,
      "loss": 1.7148,
      "step": 7210
    },
    {
      "epoch": 1.2989867147038956,
      "grad_norm": 1.810054063796997,
      "learning_rate": 2.7411583901426452e-05,
      "loss": 2.0094,
      "step": 7211
    },
    {
      "epoch": 1.2991668543120918,
      "grad_norm": 1.6464828252792358,
      "learning_rate": 2.7398950811548975e-05,
      "loss": 1.7425,
      "step": 7212
    },
    {
      "epoch": 1.2993469939202882,
      "grad_norm": 1.3600871562957764,
      "learning_rate": 2.7386319534716647e-05,
      "loss": 1.441,
      "step": 7213
    },
    {
      "epoch": 1.2995271335284846,
      "grad_norm": 1.462981104850769,
      "learning_rate": 2.7373690071942758e-05,
      "loss": 1.4793,
      "step": 7214
    },
    {
      "epoch": 1.299707273136681,
      "grad_norm": 1.445478916168213,
      "learning_rate": 2.7361062424240413e-05,
      "loss": 1.5306,
      "step": 7215
    },
    {
      "epoch": 1.2998874127448774,
      "grad_norm": 1.46418297290802,
      "learning_rate": 2.734843659262263e-05,
      "loss": 1.5081,
      "step": 7216
    },
    {
      "epoch": 1.3000675523530736,
      "grad_norm": 1.4876571893692017,
      "learning_rate": 2.7335812578102226e-05,
      "loss": 1.5218,
      "step": 7217
    },
    {
      "epoch": 1.30024769196127,
      "grad_norm": 1.3805893659591675,
      "learning_rate": 2.7323190381691903e-05,
      "loss": 1.4711,
      "step": 7218
    },
    {
      "epoch": 1.3004278315694664,
      "grad_norm": 1.5682705640792847,
      "learning_rate": 2.7310570004404202e-05,
      "loss": 1.5264,
      "step": 7219
    },
    {
      "epoch": 1.3006079711776626,
      "grad_norm": 1.3763188123703003,
      "learning_rate": 2.7297951447251524e-05,
      "loss": 1.3485,
      "step": 7220
    },
    {
      "epoch": 1.300788110785859,
      "grad_norm": 1.4283044338226318,
      "learning_rate": 2.728533471124612e-05,
      "loss": 1.4363,
      "step": 7221
    },
    {
      "epoch": 1.3009682503940554,
      "grad_norm": 1.2529871463775635,
      "learning_rate": 2.7272719797400102e-05,
      "loss": 1.128,
      "step": 7222
    },
    {
      "epoch": 1.3011483900022518,
      "grad_norm": 1.43123459815979,
      "learning_rate": 2.726010670672543e-05,
      "loss": 1.4492,
      "step": 7223
    },
    {
      "epoch": 1.3013285296104482,
      "grad_norm": 1.3271316289901733,
      "learning_rate": 2.7247495440233905e-05,
      "loss": 1.3584,
      "step": 7224
    },
    {
      "epoch": 1.3015086692186444,
      "grad_norm": 1.3395841121673584,
      "learning_rate": 2.723488599893724e-05,
      "loss": 1.3353,
      "step": 7225
    },
    {
      "epoch": 1.3016888088268408,
      "grad_norm": 1.3886768817901611,
      "learning_rate": 2.722227838384692e-05,
      "loss": 1.2368,
      "step": 7226
    },
    {
      "epoch": 1.3018689484350372,
      "grad_norm": 1.5242255926132202,
      "learning_rate": 2.7209672595974338e-05,
      "loss": 1.6965,
      "step": 7227
    },
    {
      "epoch": 1.3020490880432334,
      "grad_norm": 1.426235556602478,
      "learning_rate": 2.7197068636330723e-05,
      "loss": 1.409,
      "step": 7228
    },
    {
      "epoch": 1.3022292276514298,
      "grad_norm": 1.4048899412155151,
      "learning_rate": 2.7184466505927152e-05,
      "loss": 1.3561,
      "step": 7229
    },
    {
      "epoch": 1.3024093672596262,
      "grad_norm": 1.4246550798416138,
      "learning_rate": 2.7171866205774566e-05,
      "loss": 1.4794,
      "step": 7230
    },
    {
      "epoch": 1.3025895068678226,
      "grad_norm": 1.4218900203704834,
      "learning_rate": 2.7159267736883752e-05,
      "loss": 1.451,
      "step": 7231
    },
    {
      "epoch": 1.302769646476019,
      "grad_norm": 1.3649016618728638,
      "learning_rate": 2.7146671100265353e-05,
      "loss": 1.3299,
      "step": 7232
    },
    {
      "epoch": 1.3029497860842152,
      "grad_norm": 1.578447937965393,
      "learning_rate": 2.7134076296929866e-05,
      "loss": 1.6683,
      "step": 7233
    },
    {
      "epoch": 1.3031299256924116,
      "grad_norm": 1.5134035348892212,
      "learning_rate": 2.712148332788763e-05,
      "loss": 1.5424,
      "step": 7234
    },
    {
      "epoch": 1.303310065300608,
      "grad_norm": 1.4058001041412354,
      "learning_rate": 2.710889219414887e-05,
      "loss": 1.2976,
      "step": 7235
    },
    {
      "epoch": 1.3034902049088044,
      "grad_norm": 1.4663184881210327,
      "learning_rate": 2.709630289672363e-05,
      "loss": 1.3552,
      "step": 7236
    },
    {
      "epoch": 1.3036703445170006,
      "grad_norm": 1.6714893579483032,
      "learning_rate": 2.7083715436621815e-05,
      "loss": 1.7139,
      "step": 7237
    },
    {
      "epoch": 1.303850484125197,
      "grad_norm": 1.4534332752227783,
      "learning_rate": 2.7071129814853168e-05,
      "loss": 1.5504,
      "step": 7238
    },
    {
      "epoch": 1.3040306237333934,
      "grad_norm": 1.4148374795913696,
      "learning_rate": 2.7058546032427333e-05,
      "loss": 1.2533,
      "step": 7239
    },
    {
      "epoch": 1.3042107633415898,
      "grad_norm": 1.5228055715560913,
      "learning_rate": 2.704596409035377e-05,
      "loss": 1.2581,
      "step": 7240
    },
    {
      "epoch": 1.3043909029497862,
      "grad_norm": 1.5890846252441406,
      "learning_rate": 2.7033383989641804e-05,
      "loss": 1.3985,
      "step": 7241
    },
    {
      "epoch": 1.3045710425579824,
      "grad_norm": 1.4719725847244263,
      "learning_rate": 2.7020805731300556e-05,
      "loss": 1.451,
      "step": 7242
    },
    {
      "epoch": 1.3047511821661788,
      "grad_norm": 1.3679713010787964,
      "learning_rate": 2.7008229316339073e-05,
      "loss": 1.3359,
      "step": 7243
    },
    {
      "epoch": 1.3049313217743752,
      "grad_norm": 1.6730425357818604,
      "learning_rate": 2.6995654745766252e-05,
      "loss": 1.6712,
      "step": 7244
    },
    {
      "epoch": 1.3051114613825714,
      "grad_norm": 1.4474024772644043,
      "learning_rate": 2.6983082020590805e-05,
      "loss": 1.3266,
      "step": 7245
    },
    {
      "epoch": 1.3052916009907678,
      "grad_norm": 1.617191195487976,
      "learning_rate": 2.6970511141821307e-05,
      "loss": 1.498,
      "step": 7246
    },
    {
      "epoch": 1.3054717405989642,
      "grad_norm": 1.7057318687438965,
      "learning_rate": 2.6957942110466184e-05,
      "loss": 1.5397,
      "step": 7247
    },
    {
      "epoch": 1.3056518802071606,
      "grad_norm": 1.5168033838272095,
      "learning_rate": 2.69453749275337e-05,
      "loss": 1.4322,
      "step": 7248
    },
    {
      "epoch": 1.305832019815357,
      "grad_norm": 1.8632164001464844,
      "learning_rate": 2.6932809594032038e-05,
      "loss": 1.536,
      "step": 7249
    },
    {
      "epoch": 1.3060121594235532,
      "grad_norm": 1.5360517501831055,
      "learning_rate": 2.692024611096915e-05,
      "loss": 1.2614,
      "step": 7250
    },
    {
      "epoch": 1.3061922990317496,
      "grad_norm": 1.3333938121795654,
      "learning_rate": 2.6907684479352875e-05,
      "loss": 1.0758,
      "step": 7251
    },
    {
      "epoch": 1.306372438639946,
      "grad_norm": 1.1890655755996704,
      "learning_rate": 2.68951247001909e-05,
      "loss": 1.5101,
      "step": 7252
    },
    {
      "epoch": 1.3065525782481422,
      "grad_norm": 1.3975564241409302,
      "learning_rate": 2.6882566774490775e-05,
      "loss": 1.9661,
      "step": 7253
    },
    {
      "epoch": 1.3067327178563386,
      "grad_norm": 1.3566436767578125,
      "learning_rate": 2.6870010703259878e-05,
      "loss": 1.8625,
      "step": 7254
    },
    {
      "epoch": 1.306912857464535,
      "grad_norm": 1.277154803276062,
      "learning_rate": 2.685745648750546e-05,
      "loss": 1.653,
      "step": 7255
    },
    {
      "epoch": 1.3070929970727314,
      "grad_norm": 1.278106927871704,
      "learning_rate": 2.684490412823461e-05,
      "loss": 1.6864,
      "step": 7256
    },
    {
      "epoch": 1.3072731366809278,
      "grad_norm": 1.4362905025482178,
      "learning_rate": 2.6832353626454254e-05,
      "loss": 1.6785,
      "step": 7257
    },
    {
      "epoch": 1.307453276289124,
      "grad_norm": 1.507697343826294,
      "learning_rate": 2.681980498317122e-05,
      "loss": 1.7905,
      "step": 7258
    },
    {
      "epoch": 1.3076334158973204,
      "grad_norm": 1.7259743213653564,
      "learning_rate": 2.6807258199392144e-05,
      "loss": 1.9113,
      "step": 7259
    },
    {
      "epoch": 1.3078135555055168,
      "grad_norm": 1.672095775604248,
      "learning_rate": 2.6794713276123517e-05,
      "loss": 1.9838,
      "step": 7260
    },
    {
      "epoch": 1.307993695113713,
      "grad_norm": 1.7213810682296753,
      "learning_rate": 2.6782170214371686e-05,
      "loss": 1.8816,
      "step": 7261
    },
    {
      "epoch": 1.3081738347219094,
      "grad_norm": 1.4583821296691895,
      "learning_rate": 2.6769629015142857e-05,
      "loss": 1.5813,
      "step": 7262
    },
    {
      "epoch": 1.3083539743301058,
      "grad_norm": 1.152076244354248,
      "learning_rate": 2.675708967944307e-05,
      "loss": 1.2458,
      "step": 7263
    },
    {
      "epoch": 1.3085341139383022,
      "grad_norm": 1.4271306991577148,
      "learning_rate": 2.674455220827823e-05,
      "loss": 1.4204,
      "step": 7264
    },
    {
      "epoch": 1.3087142535464986,
      "grad_norm": 1.304468035697937,
      "learning_rate": 2.6732016602654074e-05,
      "loss": 1.3272,
      "step": 7265
    },
    {
      "epoch": 1.308894393154695,
      "grad_norm": 1.491913914680481,
      "learning_rate": 2.6719482863576207e-05,
      "loss": 1.5809,
      "step": 7266
    },
    {
      "epoch": 1.3090745327628912,
      "grad_norm": 1.234576940536499,
      "learning_rate": 2.6706950992050094e-05,
      "loss": 1.241,
      "step": 7267
    },
    {
      "epoch": 1.3092546723710876,
      "grad_norm": 1.2868402004241943,
      "learning_rate": 2.6694420989081038e-05,
      "loss": 1.2072,
      "step": 7268
    },
    {
      "epoch": 1.309434811979284,
      "grad_norm": 1.428783893585205,
      "learning_rate": 2.6681892855674168e-05,
      "loss": 1.6418,
      "step": 7269
    },
    {
      "epoch": 1.3096149515874802,
      "grad_norm": 1.375267505645752,
      "learning_rate": 2.66693665928345e-05,
      "loss": 1.3525,
      "step": 7270
    },
    {
      "epoch": 1.3097950911956766,
      "grad_norm": 1.5284193754196167,
      "learning_rate": 2.6656842201566855e-05,
      "loss": 1.4183,
      "step": 7271
    },
    {
      "epoch": 1.309975230803873,
      "grad_norm": 1.4594486951828003,
      "learning_rate": 2.6644319682875984e-05,
      "loss": 1.2222,
      "step": 7272
    },
    {
      "epoch": 1.3101553704120694,
      "grad_norm": 1.4691972732543945,
      "learning_rate": 2.6631799037766436e-05,
      "loss": 1.2174,
      "step": 7273
    },
    {
      "epoch": 1.3103355100202658,
      "grad_norm": 1.3081021308898926,
      "learning_rate": 2.6619280267242557e-05,
      "loss": 1.3187,
      "step": 7274
    },
    {
      "epoch": 1.310515649628462,
      "grad_norm": 1.6386879682540894,
      "learning_rate": 2.660676337230863e-05,
      "loss": 1.4259,
      "step": 7275
    },
    {
      "epoch": 1.3106957892366584,
      "grad_norm": 1.4025543928146362,
      "learning_rate": 2.6594248353968738e-05,
      "loss": 1.4932,
      "step": 7276
    },
    {
      "epoch": 1.3108759288448548,
      "grad_norm": 1.5073736906051636,
      "learning_rate": 2.6581735213226855e-05,
      "loss": 1.5067,
      "step": 7277
    },
    {
      "epoch": 1.311056068453051,
      "grad_norm": 1.4584907293319702,
      "learning_rate": 2.656922395108677e-05,
      "loss": 1.325,
      "step": 7278
    },
    {
      "epoch": 1.3112362080612474,
      "grad_norm": 1.4322322607040405,
      "learning_rate": 2.655671456855212e-05,
      "loss": 1.2326,
      "step": 7279
    },
    {
      "epoch": 1.3114163476694438,
      "grad_norm": 1.452122449874878,
      "learning_rate": 2.6544207066626397e-05,
      "loss": 1.4408,
      "step": 7280
    },
    {
      "epoch": 1.3115964872776402,
      "grad_norm": 1.5400147438049316,
      "learning_rate": 2.6531701446312974e-05,
      "loss": 1.429,
      "step": 7281
    },
    {
      "epoch": 1.3117766268858366,
      "grad_norm": 1.3828176259994507,
      "learning_rate": 2.6519197708615028e-05,
      "loss": 1.3844,
      "step": 7282
    },
    {
      "epoch": 1.3119567664940328,
      "grad_norm": 1.5221953392028809,
      "learning_rate": 2.6506695854535614e-05,
      "loss": 1.5843,
      "step": 7283
    },
    {
      "epoch": 1.3121369061022292,
      "grad_norm": 1.341744065284729,
      "learning_rate": 2.649419588507761e-05,
      "loss": 1.2313,
      "step": 7284
    },
    {
      "epoch": 1.3123170457104256,
      "grad_norm": 1.5417107343673706,
      "learning_rate": 2.648169780124376e-05,
      "loss": 1.4715,
      "step": 7285
    },
    {
      "epoch": 1.3124971853186218,
      "grad_norm": 1.6844868659973145,
      "learning_rate": 2.6469201604036657e-05,
      "loss": 1.8396,
      "step": 7286
    },
    {
      "epoch": 1.3126773249268182,
      "grad_norm": 1.4083354473114014,
      "learning_rate": 2.6456707294458742e-05,
      "loss": 1.2694,
      "step": 7287
    },
    {
      "epoch": 1.3128574645350146,
      "grad_norm": 1.430038332939148,
      "learning_rate": 2.6444214873512303e-05,
      "loss": 1.2239,
      "step": 7288
    },
    {
      "epoch": 1.313037604143211,
      "grad_norm": 1.4533299207687378,
      "learning_rate": 2.643172434219947e-05,
      "loss": 1.3972,
      "step": 7289
    },
    {
      "epoch": 1.3132177437514074,
      "grad_norm": 1.5285234451293945,
      "learning_rate": 2.641923570152221e-05,
      "loss": 1.4355,
      "step": 7290
    },
    {
      "epoch": 1.3133978833596036,
      "grad_norm": 1.6282265186309814,
      "learning_rate": 2.6406748952482397e-05,
      "loss": 1.404,
      "step": 7291
    },
    {
      "epoch": 1.3135780229678,
      "grad_norm": 1.5235157012939453,
      "learning_rate": 2.639426409608169e-05,
      "loss": 1.5045,
      "step": 7292
    },
    {
      "epoch": 1.3137581625759964,
      "grad_norm": 1.5599039793014526,
      "learning_rate": 2.638178113332162e-05,
      "loss": 1.4187,
      "step": 7293
    },
    {
      "epoch": 1.3139383021841928,
      "grad_norm": 1.4678492546081543,
      "learning_rate": 2.636930006520356e-05,
      "loss": 1.3195,
      "step": 7294
    },
    {
      "epoch": 1.314118441792389,
      "grad_norm": 1.5117708444595337,
      "learning_rate": 2.635682089272875e-05,
      "loss": 1.5204,
      "step": 7295
    },
    {
      "epoch": 1.3142985814005854,
      "grad_norm": 1.659947395324707,
      "learning_rate": 2.634434361689824e-05,
      "loss": 1.632,
      "step": 7296
    },
    {
      "epoch": 1.3144787210087818,
      "grad_norm": 1.5277550220489502,
      "learning_rate": 2.633186823871297e-05,
      "loss": 1.6065,
      "step": 7297
    },
    {
      "epoch": 1.3146588606169782,
      "grad_norm": 1.4691060781478882,
      "learning_rate": 2.63193947591737e-05,
      "loss": 1.3656,
      "step": 7298
    },
    {
      "epoch": 1.3148390002251746,
      "grad_norm": 1.492888331413269,
      "learning_rate": 2.630692317928103e-05,
      "loss": 1.3314,
      "step": 7299
    },
    {
      "epoch": 1.3150191398333708,
      "grad_norm": 1.4734760522842407,
      "learning_rate": 2.6294453500035453e-05,
      "loss": 1.3116,
      "step": 7300
    },
    {
      "epoch": 1.3151992794415672,
      "grad_norm": 1.5717759132385254,
      "learning_rate": 2.6281985722437273e-05,
      "loss": 1.4146,
      "step": 7301
    },
    {
      "epoch": 1.3153794190497636,
      "grad_norm": 1.3600084781646729,
      "learning_rate": 2.6269519847486657e-05,
      "loss": 1.8444,
      "step": 7302
    },
    {
      "epoch": 1.3155595586579598,
      "grad_norm": 1.3618195056915283,
      "learning_rate": 2.625705587618359e-05,
      "loss": 1.771,
      "step": 7303
    },
    {
      "epoch": 1.3157396982661562,
      "grad_norm": 1.3526585102081299,
      "learning_rate": 2.6244593809527918e-05,
      "loss": 1.7317,
      "step": 7304
    },
    {
      "epoch": 1.3159198378743526,
      "grad_norm": 1.41840398311615,
      "learning_rate": 2.6232133648519374e-05,
      "loss": 2.0681,
      "step": 7305
    },
    {
      "epoch": 1.316099977482549,
      "grad_norm": 1.3685473203659058,
      "learning_rate": 2.6219675394157516e-05,
      "loss": 1.7488,
      "step": 7306
    },
    {
      "epoch": 1.3162801170907454,
      "grad_norm": 1.4109721183776855,
      "learning_rate": 2.62072190474417e-05,
      "loss": 1.889,
      "step": 7307
    },
    {
      "epoch": 1.3164602566989416,
      "grad_norm": 1.4641557931900024,
      "learning_rate": 2.619476460937116e-05,
      "loss": 1.8409,
      "step": 7308
    },
    {
      "epoch": 1.316640396307138,
      "grad_norm": 1.449506163597107,
      "learning_rate": 2.6182312080945028e-05,
      "loss": 1.7929,
      "step": 7309
    },
    {
      "epoch": 1.3168205359153344,
      "grad_norm": 1.5877101421356201,
      "learning_rate": 2.6169861463162215e-05,
      "loss": 1.8019,
      "step": 7310
    },
    {
      "epoch": 1.3170006755235306,
      "grad_norm": 1.931904673576355,
      "learning_rate": 2.615741275702151e-05,
      "loss": 2.1412,
      "step": 7311
    },
    {
      "epoch": 1.317180815131727,
      "grad_norm": 1.7633779048919678,
      "learning_rate": 2.614496596352154e-05,
      "loss": 1.7031,
      "step": 7312
    },
    {
      "epoch": 1.3173609547399234,
      "grad_norm": 1.4810068607330322,
      "learning_rate": 2.6132521083660765e-05,
      "loss": 1.6009,
      "step": 7313
    },
    {
      "epoch": 1.3175410943481198,
      "grad_norm": 1.4078516960144043,
      "learning_rate": 2.612007811843754e-05,
      "loss": 1.4455,
      "step": 7314
    },
    {
      "epoch": 1.3177212339563162,
      "grad_norm": 1.3709933757781982,
      "learning_rate": 2.610763706885002e-05,
      "loss": 1.5757,
      "step": 7315
    },
    {
      "epoch": 1.3179013735645124,
      "grad_norm": 1.4359428882598877,
      "learning_rate": 2.6095197935896216e-05,
      "loss": 1.7435,
      "step": 7316
    },
    {
      "epoch": 1.3180815131727088,
      "grad_norm": 1.259642243385315,
      "learning_rate": 2.6082760720573985e-05,
      "loss": 1.2472,
      "step": 7317
    },
    {
      "epoch": 1.3182616527809052,
      "grad_norm": 1.457924723625183,
      "learning_rate": 2.607032542388105e-05,
      "loss": 1.6078,
      "step": 7318
    },
    {
      "epoch": 1.3184417923891014,
      "grad_norm": 1.337712287902832,
      "learning_rate": 2.605789204681494e-05,
      "loss": 1.372,
      "step": 7319
    },
    {
      "epoch": 1.3186219319972978,
      "grad_norm": 1.3357133865356445,
      "learning_rate": 2.6045460590373078e-05,
      "loss": 1.1382,
      "step": 7320
    },
    {
      "epoch": 1.3188020716054942,
      "grad_norm": 1.3546286821365356,
      "learning_rate": 2.603303105555269e-05,
      "loss": 1.5586,
      "step": 7321
    },
    {
      "epoch": 1.3189822112136906,
      "grad_norm": 1.471184492111206,
      "learning_rate": 2.6020603443350866e-05,
      "loss": 1.584,
      "step": 7322
    },
    {
      "epoch": 1.319162350821887,
      "grad_norm": 1.3240504264831543,
      "learning_rate": 2.6008177754764574e-05,
      "loss": 1.2226,
      "step": 7323
    },
    {
      "epoch": 1.3193424904300834,
      "grad_norm": 1.342200517654419,
      "learning_rate": 2.5995753990790573e-05,
      "loss": 1.1993,
      "step": 7324
    },
    {
      "epoch": 1.3195226300382796,
      "grad_norm": 1.262457251548767,
      "learning_rate": 2.59833321524255e-05,
      "loss": 1.2465,
      "step": 7325
    },
    {
      "epoch": 1.319702769646476,
      "grad_norm": 1.4489136934280396,
      "learning_rate": 2.5970912240665813e-05,
      "loss": 1.5042,
      "step": 7326
    },
    {
      "epoch": 1.3198829092546724,
      "grad_norm": 1.3953566551208496,
      "learning_rate": 2.5958494256507847e-05,
      "loss": 1.2865,
      "step": 7327
    },
    {
      "epoch": 1.3200630488628686,
      "grad_norm": 1.4870502948760986,
      "learning_rate": 2.5946078200947764e-05,
      "loss": 1.5319,
      "step": 7328
    },
    {
      "epoch": 1.320243188471065,
      "grad_norm": 1.491736650466919,
      "learning_rate": 2.5933664074981573e-05,
      "loss": 1.6044,
      "step": 7329
    },
    {
      "epoch": 1.3204233280792614,
      "grad_norm": 1.6358563899993896,
      "learning_rate": 2.5921251879605123e-05,
      "loss": 1.4241,
      "step": 7330
    },
    {
      "epoch": 1.3206034676874578,
      "grad_norm": 1.419915795326233,
      "learning_rate": 2.5908841615814105e-05,
      "loss": 1.4875,
      "step": 7331
    },
    {
      "epoch": 1.3207836072956542,
      "grad_norm": 1.488810420036316,
      "learning_rate": 2.5896433284604092e-05,
      "loss": 1.5792,
      "step": 7332
    },
    {
      "epoch": 1.3209637469038504,
      "grad_norm": 1.5382776260375977,
      "learning_rate": 2.588402688697046e-05,
      "loss": 1.4145,
      "step": 7333
    },
    {
      "epoch": 1.3211438865120468,
      "grad_norm": 1.418113350868225,
      "learning_rate": 2.5871622423908448e-05,
      "loss": 1.3075,
      "step": 7334
    },
    {
      "epoch": 1.3213240261202432,
      "grad_norm": 1.4907478094100952,
      "learning_rate": 2.5859219896413135e-05,
      "loss": 1.4728,
      "step": 7335
    },
    {
      "epoch": 1.3215041657284394,
      "grad_norm": 1.4709175825119019,
      "learning_rate": 2.584681930547942e-05,
      "loss": 1.4941,
      "step": 7336
    },
    {
      "epoch": 1.3216843053366358,
      "grad_norm": 1.6379138231277466,
      "learning_rate": 2.5834420652102114e-05,
      "loss": 1.5501,
      "step": 7337
    },
    {
      "epoch": 1.3218644449448322,
      "grad_norm": 1.6066511869430542,
      "learning_rate": 2.582202393727583e-05,
      "loss": 1.5001,
      "step": 7338
    },
    {
      "epoch": 1.3220445845530286,
      "grad_norm": 1.4030712842941284,
      "learning_rate": 2.5809629161994997e-05,
      "loss": 1.1984,
      "step": 7339
    },
    {
      "epoch": 1.322224724161225,
      "grad_norm": 1.468855857849121,
      "learning_rate": 2.5797236327253926e-05,
      "loss": 1.6492,
      "step": 7340
    },
    {
      "epoch": 1.3224048637694212,
      "grad_norm": 1.5526490211486816,
      "learning_rate": 2.578484543404675e-05,
      "loss": 1.438,
      "step": 7341
    },
    {
      "epoch": 1.3225850033776176,
      "grad_norm": 1.6583173274993896,
      "learning_rate": 2.5772456483367497e-05,
      "loss": 1.617,
      "step": 7342
    },
    {
      "epoch": 1.322765142985814,
      "grad_norm": 1.4880729913711548,
      "learning_rate": 2.576006947620998e-05,
      "loss": 1.4121,
      "step": 7343
    },
    {
      "epoch": 1.3229452825940102,
      "grad_norm": 1.4405739307403564,
      "learning_rate": 2.574768441356789e-05,
      "loss": 1.2916,
      "step": 7344
    },
    {
      "epoch": 1.3231254222022066,
      "grad_norm": 1.4879804849624634,
      "learning_rate": 2.573530129643472e-05,
      "loss": 1.4511,
      "step": 7345
    },
    {
      "epoch": 1.323305561810403,
      "grad_norm": 1.2999303340911865,
      "learning_rate": 2.5722920125803883e-05,
      "loss": 1.1854,
      "step": 7346
    },
    {
      "epoch": 1.3234857014185994,
      "grad_norm": 1.438851237297058,
      "learning_rate": 2.5710540902668568e-05,
      "loss": 1.2726,
      "step": 7347
    },
    {
      "epoch": 1.3236658410267959,
      "grad_norm": 1.591162919998169,
      "learning_rate": 2.5698163628021828e-05,
      "loss": 1.3435,
      "step": 7348
    },
    {
      "epoch": 1.323845980634992,
      "grad_norm": 1.3374578952789307,
      "learning_rate": 2.5685788302856567e-05,
      "loss": 1.0728,
      "step": 7349
    },
    {
      "epoch": 1.3240261202431884,
      "grad_norm": 1.6100682020187378,
      "learning_rate": 2.5673414928165523e-05,
      "loss": 1.339,
      "step": 7350
    },
    {
      "epoch": 1.3242062598513848,
      "grad_norm": 1.516313910484314,
      "learning_rate": 2.5661043504941285e-05,
      "loss": 1.2676,
      "step": 7351
    },
    {
      "epoch": 1.3243863994595813,
      "grad_norm": 1.4258757829666138,
      "learning_rate": 2.5648674034176285e-05,
      "loss": 1.9373,
      "step": 7352
    },
    {
      "epoch": 1.3245665390677774,
      "grad_norm": 1.2540708780288696,
      "learning_rate": 2.563630651686279e-05,
      "loss": 1.5331,
      "step": 7353
    },
    {
      "epoch": 1.3247466786759738,
      "grad_norm": 1.3426744937896729,
      "learning_rate": 2.562394095399292e-05,
      "loss": 2.0205,
      "step": 7354
    },
    {
      "epoch": 1.3249268182841702,
      "grad_norm": 1.2861336469650269,
      "learning_rate": 2.5611577346558612e-05,
      "loss": 1.7842,
      "step": 7355
    },
    {
      "epoch": 1.3251069578923667,
      "grad_norm": 1.3309264183044434,
      "learning_rate": 2.5599215695551714e-05,
      "loss": 1.5667,
      "step": 7356
    },
    {
      "epoch": 1.325287097500563,
      "grad_norm": 1.2930220365524292,
      "learning_rate": 2.5586856001963843e-05,
      "loss": 1.5863,
      "step": 7357
    },
    {
      "epoch": 1.3254672371087592,
      "grad_norm": 1.698184847831726,
      "learning_rate": 2.5574498266786496e-05,
      "loss": 1.7746,
      "step": 7358
    },
    {
      "epoch": 1.3256473767169556,
      "grad_norm": 1.5733425617218018,
      "learning_rate": 2.5562142491011e-05,
      "loss": 1.7823,
      "step": 7359
    },
    {
      "epoch": 1.325827516325152,
      "grad_norm": 1.4867327213287354,
      "learning_rate": 2.5549788675628527e-05,
      "loss": 1.7456,
      "step": 7360
    },
    {
      "epoch": 1.3260076559333482,
      "grad_norm": 1.7311596870422363,
      "learning_rate": 2.55374368216301e-05,
      "loss": 1.9907,
      "step": 7361
    },
    {
      "epoch": 1.3261877955415446,
      "grad_norm": 1.9057351350784302,
      "learning_rate": 2.552508693000658e-05,
      "loss": 2.1653,
      "step": 7362
    },
    {
      "epoch": 1.326367935149741,
      "grad_norm": 1.521274447441101,
      "learning_rate": 2.5512739001748663e-05,
      "loss": 1.6859,
      "step": 7363
    },
    {
      "epoch": 1.3265480747579375,
      "grad_norm": 1.3585267066955566,
      "learning_rate": 2.550039303784687e-05,
      "loss": 1.3846,
      "step": 7364
    },
    {
      "epoch": 1.3267282143661339,
      "grad_norm": 1.3551506996154785,
      "learning_rate": 2.548804903929164e-05,
      "loss": 1.4298,
      "step": 7365
    },
    {
      "epoch": 1.32690835397433,
      "grad_norm": 1.466292381286621,
      "learning_rate": 2.5475707007073176e-05,
      "loss": 1.4875,
      "step": 7366
    },
    {
      "epoch": 1.3270884935825265,
      "grad_norm": 1.3474199771881104,
      "learning_rate": 2.546336694218155e-05,
      "loss": 1.17,
      "step": 7367
    },
    {
      "epoch": 1.3272686331907229,
      "grad_norm": 1.3042031526565552,
      "learning_rate": 2.5451028845606673e-05,
      "loss": 1.3004,
      "step": 7368
    },
    {
      "epoch": 1.327448772798919,
      "grad_norm": 1.3461599349975586,
      "learning_rate": 2.5438692718338287e-05,
      "loss": 1.4078,
      "step": 7369
    },
    {
      "epoch": 1.3276289124071154,
      "grad_norm": 1.455161690711975,
      "learning_rate": 2.5426358561366027e-05,
      "loss": 1.5689,
      "step": 7370
    },
    {
      "epoch": 1.3278090520153119,
      "grad_norm": 1.3472305536270142,
      "learning_rate": 2.5414026375679335e-05,
      "loss": 1.3143,
      "step": 7371
    },
    {
      "epoch": 1.3279891916235083,
      "grad_norm": 1.335525393486023,
      "learning_rate": 2.5401696162267454e-05,
      "loss": 1.2696,
      "step": 7372
    },
    {
      "epoch": 1.3281693312317047,
      "grad_norm": 1.3115664720535278,
      "learning_rate": 2.5389367922119502e-05,
      "loss": 1.2453,
      "step": 7373
    },
    {
      "epoch": 1.3283494708399008,
      "grad_norm": 1.3081218004226685,
      "learning_rate": 2.5377041656224488e-05,
      "loss": 1.1292,
      "step": 7374
    },
    {
      "epoch": 1.3285296104480973,
      "grad_norm": 1.314295768737793,
      "learning_rate": 2.53647173655712e-05,
      "loss": 1.311,
      "step": 7375
    },
    {
      "epoch": 1.3287097500562937,
      "grad_norm": 1.3531287908554077,
      "learning_rate": 2.5352395051148292e-05,
      "loss": 1.2743,
      "step": 7376
    },
    {
      "epoch": 1.3288898896644898,
      "grad_norm": 1.4227545261383057,
      "learning_rate": 2.534007471394424e-05,
      "loss": 1.3979,
      "step": 7377
    },
    {
      "epoch": 1.3290700292726862,
      "grad_norm": 1.3450112342834473,
      "learning_rate": 2.5327756354947362e-05,
      "loss": 1.4086,
      "step": 7378
    },
    {
      "epoch": 1.3292501688808827,
      "grad_norm": 1.2894139289855957,
      "learning_rate": 2.5315439975145872e-05,
      "loss": 1.2673,
      "step": 7379
    },
    {
      "epoch": 1.329430308489079,
      "grad_norm": 1.369187593460083,
      "learning_rate": 2.5303125575527764e-05,
      "loss": 1.317,
      "step": 7380
    },
    {
      "epoch": 1.3296104480972755,
      "grad_norm": 1.5643956661224365,
      "learning_rate": 2.5290813157080895e-05,
      "loss": 1.5074,
      "step": 7381
    },
    {
      "epoch": 1.3297905877054719,
      "grad_norm": 1.4602714776992798,
      "learning_rate": 2.5278502720792963e-05,
      "loss": 1.2168,
      "step": 7382
    },
    {
      "epoch": 1.329970727313668,
      "grad_norm": 1.5449684858322144,
      "learning_rate": 2.52661942676515e-05,
      "loss": 1.5887,
      "step": 7383
    },
    {
      "epoch": 1.3301508669218645,
      "grad_norm": 1.3460958003997803,
      "learning_rate": 2.5253887798643876e-05,
      "loss": 1.2643,
      "step": 7384
    },
    {
      "epoch": 1.3303310065300609,
      "grad_norm": 1.3849482536315918,
      "learning_rate": 2.5241583314757327e-05,
      "loss": 1.3234,
      "step": 7385
    },
    {
      "epoch": 1.330511146138257,
      "grad_norm": 1.4551743268966675,
      "learning_rate": 2.5229280816978905e-05,
      "loss": 1.2632,
      "step": 7386
    },
    {
      "epoch": 1.3306912857464535,
      "grad_norm": 1.5274627208709717,
      "learning_rate": 2.521698030629549e-05,
      "loss": 1.5798,
      "step": 7387
    },
    {
      "epoch": 1.3308714253546499,
      "grad_norm": 1.4003686904907227,
      "learning_rate": 2.520468178369386e-05,
      "loss": 1.2641,
      "step": 7388
    },
    {
      "epoch": 1.3310515649628463,
      "grad_norm": 1.6020455360412598,
      "learning_rate": 2.5192385250160588e-05,
      "loss": 1.6004,
      "step": 7389
    },
    {
      "epoch": 1.3312317045710427,
      "grad_norm": 1.4557509422302246,
      "learning_rate": 2.518009070668208e-05,
      "loss": 1.3596,
      "step": 7390
    },
    {
      "epoch": 1.3314118441792389,
      "grad_norm": 1.5017496347427368,
      "learning_rate": 2.516779815424461e-05,
      "loss": 1.402,
      "step": 7391
    },
    {
      "epoch": 1.3315919837874353,
      "grad_norm": 1.5260274410247803,
      "learning_rate": 2.515550759383427e-05,
      "loss": 1.502,
      "step": 7392
    },
    {
      "epoch": 1.3317721233956317,
      "grad_norm": 1.3746528625488281,
      "learning_rate": 2.5143219026437022e-05,
      "loss": 1.3034,
      "step": 7393
    },
    {
      "epoch": 1.3319522630038279,
      "grad_norm": 1.4807591438293457,
      "learning_rate": 2.5130932453038626e-05,
      "loss": 1.3978,
      "step": 7394
    },
    {
      "epoch": 1.3321324026120243,
      "grad_norm": 1.4649735689163208,
      "learning_rate": 2.5118647874624723e-05,
      "loss": 1.3293,
      "step": 7395
    },
    {
      "epoch": 1.3323125422202207,
      "grad_norm": 1.4470797777175903,
      "learning_rate": 2.510636529218076e-05,
      "loss": 1.1737,
      "step": 7396
    },
    {
      "epoch": 1.332492681828417,
      "grad_norm": 1.6314496994018555,
      "learning_rate": 2.5094084706692033e-05,
      "loss": 1.3086,
      "step": 7397
    },
    {
      "epoch": 1.3326728214366135,
      "grad_norm": 1.5507962703704834,
      "learning_rate": 2.508180611914372e-05,
      "loss": 1.3525,
      "step": 7398
    },
    {
      "epoch": 1.3328529610448097,
      "grad_norm": 1.630357265472412,
      "learning_rate": 2.506952953052078e-05,
      "loss": 1.364,
      "step": 7399
    },
    {
      "epoch": 1.333033100653006,
      "grad_norm": 1.5774949789047241,
      "learning_rate": 2.505725494180804e-05,
      "loss": 1.264,
      "step": 7400
    },
    {
      "epoch": 1.3332132402612025,
      "grad_norm": 1.452436089515686,
      "learning_rate": 2.504498235399014e-05,
      "loss": 1.1149,
      "step": 7401
    },
    {
      "epoch": 1.3333933798693987,
      "grad_norm": 1.3649039268493652,
      "learning_rate": 2.503271176805162e-05,
      "loss": 1.966,
      "step": 7402
    },
    {
      "epoch": 1.333573519477595,
      "grad_norm": 1.2950999736785889,
      "learning_rate": 2.5020443184976816e-05,
      "loss": 1.8575,
      "step": 7403
    },
    {
      "epoch": 1.3337536590857915,
      "grad_norm": 1.31584632396698,
      "learning_rate": 2.500817660574988e-05,
      "loss": 1.9418,
      "step": 7404
    },
    {
      "epoch": 1.3339337986939879,
      "grad_norm": 1.3670109510421753,
      "learning_rate": 2.499591203135484e-05,
      "loss": 1.7432,
      "step": 7405
    },
    {
      "epoch": 1.3341139383021843,
      "grad_norm": 1.4113842248916626,
      "learning_rate": 2.498364946277554e-05,
      "loss": 1.8612,
      "step": 7406
    },
    {
      "epoch": 1.3342940779103807,
      "grad_norm": 1.447308897972107,
      "learning_rate": 2.4971388900995712e-05,
      "loss": 1.4803,
      "step": 7407
    },
    {
      "epoch": 1.3344742175185769,
      "grad_norm": 1.4973483085632324,
      "learning_rate": 2.4959130346998877e-05,
      "loss": 1.5137,
      "step": 7408
    },
    {
      "epoch": 1.3346543571267733,
      "grad_norm": 1.720306634902954,
      "learning_rate": 2.4946873801768407e-05,
      "loss": 1.8015,
      "step": 7409
    },
    {
      "epoch": 1.3348344967349697,
      "grad_norm": 1.6993073225021362,
      "learning_rate": 2.4934619266287516e-05,
      "loss": 1.6798,
      "step": 7410
    },
    {
      "epoch": 1.3350146363431659,
      "grad_norm": 1.5025004148483276,
      "learning_rate": 2.4922366741539243e-05,
      "loss": 1.6906,
      "step": 7411
    },
    {
      "epoch": 1.3351947759513623,
      "grad_norm": 1.539351224899292,
      "learning_rate": 2.4910116228506515e-05,
      "loss": 1.744,
      "step": 7412
    },
    {
      "epoch": 1.3353749155595587,
      "grad_norm": 1.4526416063308716,
      "learning_rate": 2.4897867728172043e-05,
      "loss": 1.6141,
      "step": 7413
    },
    {
      "epoch": 1.335555055167755,
      "grad_norm": 1.4780164957046509,
      "learning_rate": 2.4885621241518392e-05,
      "loss": 1.5224,
      "step": 7414
    },
    {
      "epoch": 1.3357351947759515,
      "grad_norm": 1.6061023473739624,
      "learning_rate": 2.487337676952797e-05,
      "loss": 1.7414,
      "step": 7415
    },
    {
      "epoch": 1.3359153343841477,
      "grad_norm": 1.3931218385696411,
      "learning_rate": 2.4861134313183027e-05,
      "loss": 1.4208,
      "step": 7416
    },
    {
      "epoch": 1.336095473992344,
      "grad_norm": 1.5883382558822632,
      "learning_rate": 2.4848893873465647e-05,
      "loss": 1.6416,
      "step": 7417
    },
    {
      "epoch": 1.3362756136005405,
      "grad_norm": 1.4585665464401245,
      "learning_rate": 2.4836655451357748e-05,
      "loss": 1.4885,
      "step": 7418
    },
    {
      "epoch": 1.3364557532087367,
      "grad_norm": 1.4005669355392456,
      "learning_rate": 2.482441904784109e-05,
      "loss": 1.4526,
      "step": 7419
    },
    {
      "epoch": 1.336635892816933,
      "grad_norm": 1.3478909730911255,
      "learning_rate": 2.481218466389726e-05,
      "loss": 1.2833,
      "step": 7420
    },
    {
      "epoch": 1.3368160324251295,
      "grad_norm": 1.5120099782943726,
      "learning_rate": 2.479995230050772e-05,
      "loss": 1.3637,
      "step": 7421
    },
    {
      "epoch": 1.3369961720333259,
      "grad_norm": 1.517782211303711,
      "learning_rate": 2.4787721958653737e-05,
      "loss": 1.5427,
      "step": 7422
    },
    {
      "epoch": 1.3371763116415223,
      "grad_norm": 1.373405933380127,
      "learning_rate": 2.4775493639316417e-05,
      "loss": 1.3379,
      "step": 7423
    },
    {
      "epoch": 1.3373564512497185,
      "grad_norm": 1.3311572074890137,
      "learning_rate": 2.476326734347671e-05,
      "loss": 1.3233,
      "step": 7424
    },
    {
      "epoch": 1.3375365908579149,
      "grad_norm": 1.3962302207946777,
      "learning_rate": 2.4751043072115404e-05,
      "loss": 1.5041,
      "step": 7425
    },
    {
      "epoch": 1.3377167304661113,
      "grad_norm": 1.4598067998886108,
      "learning_rate": 2.4738820826213127e-05,
      "loss": 1.5001,
      "step": 7426
    },
    {
      "epoch": 1.3378968700743075,
      "grad_norm": 1.3542522192001343,
      "learning_rate": 2.472660060675034e-05,
      "loss": 1.1943,
      "step": 7427
    },
    {
      "epoch": 1.3380770096825039,
      "grad_norm": 1.6554187536239624,
      "learning_rate": 2.4714382414707337e-05,
      "loss": 1.6644,
      "step": 7428
    },
    {
      "epoch": 1.3382571492907003,
      "grad_norm": 1.2895430326461792,
      "learning_rate": 2.470216625106424e-05,
      "loss": 1.2796,
      "step": 7429
    },
    {
      "epoch": 1.3384372888988967,
      "grad_norm": 1.3304355144500732,
      "learning_rate": 2.4689952116801067e-05,
      "loss": 1.3059,
      "step": 7430
    },
    {
      "epoch": 1.338617428507093,
      "grad_norm": 1.6134988069534302,
      "learning_rate": 2.467774001289761e-05,
      "loss": 1.5491,
      "step": 7431
    },
    {
      "epoch": 1.3387975681152893,
      "grad_norm": 1.3640635013580322,
      "learning_rate": 2.466552994033351e-05,
      "loss": 1.2825,
      "step": 7432
    },
    {
      "epoch": 1.3389777077234857,
      "grad_norm": 1.4744008779525757,
      "learning_rate": 2.465332190008826e-05,
      "loss": 1.6094,
      "step": 7433
    },
    {
      "epoch": 1.339157847331682,
      "grad_norm": 1.4444788694381714,
      "learning_rate": 2.4641115893141163e-05,
      "loss": 1.4454,
      "step": 7434
    },
    {
      "epoch": 1.3393379869398785,
      "grad_norm": 1.457944393157959,
      "learning_rate": 2.4628911920471416e-05,
      "loss": 1.4873,
      "step": 7435
    },
    {
      "epoch": 1.3395181265480747,
      "grad_norm": 1.4250948429107666,
      "learning_rate": 2.4616709983058018e-05,
      "loss": 1.3701,
      "step": 7436
    },
    {
      "epoch": 1.339698266156271,
      "grad_norm": 1.445516586303711,
      "learning_rate": 2.4604510081879756e-05,
      "loss": 1.4646,
      "step": 7437
    },
    {
      "epoch": 1.3398784057644675,
      "grad_norm": 1.5809369087219238,
      "learning_rate": 2.459231221791533e-05,
      "loss": 1.546,
      "step": 7438
    },
    {
      "epoch": 1.340058545372664,
      "grad_norm": 1.6723449230194092,
      "learning_rate": 2.4580116392143226e-05,
      "loss": 1.3507,
      "step": 7439
    },
    {
      "epoch": 1.3402386849808603,
      "grad_norm": 1.6411149501800537,
      "learning_rate": 2.4567922605541817e-05,
      "loss": 1.5675,
      "step": 7440
    },
    {
      "epoch": 1.3404188245890565,
      "grad_norm": 1.4077147245407104,
      "learning_rate": 2.455573085908927e-05,
      "loss": 1.294,
      "step": 7441
    },
    {
      "epoch": 1.3405989641972529,
      "grad_norm": 1.5889328718185425,
      "learning_rate": 2.4543541153763604e-05,
      "loss": 1.2656,
      "step": 7442
    },
    {
      "epoch": 1.3407791038054493,
      "grad_norm": 1.4541394710540771,
      "learning_rate": 2.4531353490542647e-05,
      "loss": 1.4248,
      "step": 7443
    },
    {
      "epoch": 1.3409592434136455,
      "grad_norm": 1.4782243967056274,
      "learning_rate": 2.4519167870404125e-05,
      "loss": 1.3322,
      "step": 7444
    },
    {
      "epoch": 1.3411393830218419,
      "grad_norm": 1.4873660802841187,
      "learning_rate": 2.4506984294325542e-05,
      "loss": 1.506,
      "step": 7445
    },
    {
      "epoch": 1.3413195226300383,
      "grad_norm": 1.4146454334259033,
      "learning_rate": 2.449480276328427e-05,
      "loss": 1.1864,
      "step": 7446
    },
    {
      "epoch": 1.3414996622382347,
      "grad_norm": 1.3631362915039062,
      "learning_rate": 2.4482623278257487e-05,
      "loss": 1.2718,
      "step": 7447
    },
    {
      "epoch": 1.341679801846431,
      "grad_norm": 1.581967830657959,
      "learning_rate": 2.4470445840222245e-05,
      "loss": 1.4732,
      "step": 7448
    },
    {
      "epoch": 1.3418599414546273,
      "grad_norm": 1.649193525314331,
      "learning_rate": 2.4458270450155396e-05,
      "loss": 1.4134,
      "step": 7449
    },
    {
      "epoch": 1.3420400810628237,
      "grad_norm": 1.3684557676315308,
      "learning_rate": 2.4446097109033657e-05,
      "loss": 0.9573,
      "step": 7450
    },
    {
      "epoch": 1.34222022067102,
      "grad_norm": 1.4057847261428833,
      "learning_rate": 2.443392581783355e-05,
      "loss": 1.2568,
      "step": 7451
    },
    {
      "epoch": 1.3424003602792163,
      "grad_norm": 1.3378297090530396,
      "learning_rate": 2.442175657753145e-05,
      "loss": 1.8224,
      "step": 7452
    },
    {
      "epoch": 1.3425804998874127,
      "grad_norm": 1.2546147108078003,
      "learning_rate": 2.44095893891036e-05,
      "loss": 1.5809,
      "step": 7453
    },
    {
      "epoch": 1.342760639495609,
      "grad_norm": 1.3928111791610718,
      "learning_rate": 2.4397424253526018e-05,
      "loss": 1.761,
      "step": 7454
    },
    {
      "epoch": 1.3429407791038055,
      "grad_norm": 1.4378395080566406,
      "learning_rate": 2.4385261171774597e-05,
      "loss": 2.167,
      "step": 7455
    },
    {
      "epoch": 1.343120918712002,
      "grad_norm": 1.4332181215286255,
      "learning_rate": 2.4373100144825052e-05,
      "loss": 1.8079,
      "step": 7456
    },
    {
      "epoch": 1.343301058320198,
      "grad_norm": 1.5470361709594727,
      "learning_rate": 2.4360941173652923e-05,
      "loss": 1.9301,
      "step": 7457
    },
    {
      "epoch": 1.3434811979283945,
      "grad_norm": 1.8123210668563843,
      "learning_rate": 2.4348784259233603e-05,
      "loss": 2.1844,
      "step": 7458
    },
    {
      "epoch": 1.343661337536591,
      "grad_norm": 1.8186489343643188,
      "learning_rate": 2.433662940254232e-05,
      "loss": 1.9889,
      "step": 7459
    },
    {
      "epoch": 1.343841477144787,
      "grad_norm": 1.817042589187622,
      "learning_rate": 2.432447660455413e-05,
      "loss": 1.8337,
      "step": 7460
    },
    {
      "epoch": 1.3440216167529835,
      "grad_norm": 1.5014989376068115,
      "learning_rate": 2.4312325866243918e-05,
      "loss": 1.617,
      "step": 7461
    },
    {
      "epoch": 1.34420175636118,
      "grad_norm": 1.4371072053909302,
      "learning_rate": 2.430017718858639e-05,
      "loss": 1.6313,
      "step": 7462
    },
    {
      "epoch": 1.3443818959693763,
      "grad_norm": 1.3879990577697754,
      "learning_rate": 2.4288030572556153e-05,
      "loss": 1.5081,
      "step": 7463
    },
    {
      "epoch": 1.3445620355775727,
      "grad_norm": 1.3962204456329346,
      "learning_rate": 2.4275886019127573e-05,
      "loss": 1.4773,
      "step": 7464
    },
    {
      "epoch": 1.344742175185769,
      "grad_norm": 1.3746598958969116,
      "learning_rate": 2.4263743529274897e-05,
      "loss": 1.4569,
      "step": 7465
    },
    {
      "epoch": 1.3449223147939653,
      "grad_norm": 1.3780293464660645,
      "learning_rate": 2.4251603103972154e-05,
      "loss": 1.3165,
      "step": 7466
    },
    {
      "epoch": 1.3451024544021617,
      "grad_norm": 1.3271169662475586,
      "learning_rate": 2.423946474419329e-05,
      "loss": 1.2733,
      "step": 7467
    },
    {
      "epoch": 1.345282594010358,
      "grad_norm": 1.3466829061508179,
      "learning_rate": 2.4227328450912034e-05,
      "loss": 1.3988,
      "step": 7468
    },
    {
      "epoch": 1.3454627336185543,
      "grad_norm": 1.4426360130310059,
      "learning_rate": 2.4215194225101918e-05,
      "loss": 1.3421,
      "step": 7469
    },
    {
      "epoch": 1.3456428732267507,
      "grad_norm": 1.5536413192749023,
      "learning_rate": 2.420306206773636e-05,
      "loss": 1.7887,
      "step": 7470
    },
    {
      "epoch": 1.345823012834947,
      "grad_norm": 1.3275916576385498,
      "learning_rate": 2.4190931979788577e-05,
      "loss": 1.4786,
      "step": 7471
    },
    {
      "epoch": 1.3460031524431435,
      "grad_norm": 1.328144907951355,
      "learning_rate": 2.4178803962231682e-05,
      "loss": 1.2672,
      "step": 7472
    },
    {
      "epoch": 1.34618329205134,
      "grad_norm": 1.2790991067886353,
      "learning_rate": 2.416667801603856e-05,
      "loss": 1.2191,
      "step": 7473
    },
    {
      "epoch": 1.346363431659536,
      "grad_norm": 1.2651121616363525,
      "learning_rate": 2.415455414218194e-05,
      "loss": 1.3081,
      "step": 7474
    },
    {
      "epoch": 1.3465435712677325,
      "grad_norm": 1.2784348726272583,
      "learning_rate": 2.4142432341634397e-05,
      "loss": 1.3523,
      "step": 7475
    },
    {
      "epoch": 1.346723710875929,
      "grad_norm": 1.40240478515625,
      "learning_rate": 2.4130312615368323e-05,
      "loss": 1.337,
      "step": 7476
    },
    {
      "epoch": 1.346903850484125,
      "grad_norm": 1.5089629888534546,
      "learning_rate": 2.4118194964355985e-05,
      "loss": 1.3523,
      "step": 7477
    },
    {
      "epoch": 1.3470839900923215,
      "grad_norm": 1.3649535179138184,
      "learning_rate": 2.4106079389569435e-05,
      "loss": 1.2622,
      "step": 7478
    },
    {
      "epoch": 1.347264129700518,
      "grad_norm": 1.5747923851013184,
      "learning_rate": 2.4093965891980592e-05,
      "loss": 1.5236,
      "step": 7479
    },
    {
      "epoch": 1.3474442693087143,
      "grad_norm": 1.4119004011154175,
      "learning_rate": 2.4081854472561184e-05,
      "loss": 1.2261,
      "step": 7480
    },
    {
      "epoch": 1.3476244089169107,
      "grad_norm": 1.328283667564392,
      "learning_rate": 2.406974513228279e-05,
      "loss": 1.3119,
      "step": 7481
    },
    {
      "epoch": 1.347804548525107,
      "grad_norm": 1.467720627784729,
      "learning_rate": 2.4057637872116808e-05,
      "loss": 1.5919,
      "step": 7482
    },
    {
      "epoch": 1.3479846881333033,
      "grad_norm": 1.3560082912445068,
      "learning_rate": 2.404553269303448e-05,
      "loss": 1.3715,
      "step": 7483
    },
    {
      "epoch": 1.3481648277414997,
      "grad_norm": 1.4915086030960083,
      "learning_rate": 2.4033429596006875e-05,
      "loss": 1.4866,
      "step": 7484
    },
    {
      "epoch": 1.3483449673496959,
      "grad_norm": 1.5317884683609009,
      "learning_rate": 2.4021328582004883e-05,
      "loss": 1.4028,
      "step": 7485
    },
    {
      "epoch": 1.3485251069578923,
      "grad_norm": 1.2985628843307495,
      "learning_rate": 2.4009229651999277e-05,
      "loss": 1.1345,
      "step": 7486
    },
    {
      "epoch": 1.3487052465660887,
      "grad_norm": 1.5905373096466064,
      "learning_rate": 2.399713280696061e-05,
      "loss": 1.4265,
      "step": 7487
    },
    {
      "epoch": 1.348885386174285,
      "grad_norm": 1.4643833637237549,
      "learning_rate": 2.398503804785928e-05,
      "loss": 1.2912,
      "step": 7488
    },
    {
      "epoch": 1.3490655257824815,
      "grad_norm": 1.410137414932251,
      "learning_rate": 2.3972945375665528e-05,
      "loss": 1.3174,
      "step": 7489
    },
    {
      "epoch": 1.3492456653906777,
      "grad_norm": 1.5130493640899658,
      "learning_rate": 2.3960854791349425e-05,
      "loss": 1.61,
      "step": 7490
    },
    {
      "epoch": 1.349425804998874,
      "grad_norm": 1.4430396556854248,
      "learning_rate": 2.3948766295880864e-05,
      "loss": 1.1788,
      "step": 7491
    },
    {
      "epoch": 1.3496059446070705,
      "grad_norm": 1.6037979125976562,
      "learning_rate": 2.393667989022958e-05,
      "loss": 1.5669,
      "step": 7492
    },
    {
      "epoch": 1.349786084215267,
      "grad_norm": 1.68007230758667,
      "learning_rate": 2.3924595575365138e-05,
      "loss": 1.6145,
      "step": 7493
    },
    {
      "epoch": 1.349966223823463,
      "grad_norm": 1.621848464012146,
      "learning_rate": 2.3912513352256927e-05,
      "loss": 1.5698,
      "step": 7494
    },
    {
      "epoch": 1.3501463634316595,
      "grad_norm": 1.3762249946594238,
      "learning_rate": 2.39004332218742e-05,
      "loss": 1.286,
      "step": 7495
    },
    {
      "epoch": 1.350326503039856,
      "grad_norm": 1.394285798072815,
      "learning_rate": 2.3888355185186014e-05,
      "loss": 1.3227,
      "step": 7496
    },
    {
      "epoch": 1.3505066426480523,
      "grad_norm": 1.5705859661102295,
      "learning_rate": 2.3876279243161255e-05,
      "loss": 1.4613,
      "step": 7497
    },
    {
      "epoch": 1.3506867822562487,
      "grad_norm": 1.4476512670516968,
      "learning_rate": 2.3864205396768645e-05,
      "loss": 1.4022,
      "step": 7498
    },
    {
      "epoch": 1.350866921864445,
      "grad_norm": 1.5966507196426392,
      "learning_rate": 2.3852133646976733e-05,
      "loss": 1.3322,
      "step": 7499
    },
    {
      "epoch": 1.3510470614726413,
      "grad_norm": 1.504564642906189,
      "learning_rate": 2.384006399475396e-05,
      "loss": 1.1595,
      "step": 7500
    },
    {
      "epoch": 1.3512272010808377,
      "grad_norm": 1.6095887422561646,
      "learning_rate": 2.3827996441068496e-05,
      "loss": 1.5881,
      "step": 7501
    },
    {
      "epoch": 1.351407340689034,
      "grad_norm": 1.4006251096725464,
      "learning_rate": 2.3815930986888406e-05,
      "loss": 1.9172,
      "step": 7502
    },
    {
      "epoch": 1.3515874802972303,
      "grad_norm": 1.2943110466003418,
      "learning_rate": 2.3803867633181574e-05,
      "loss": 1.7611,
      "step": 7503
    },
    {
      "epoch": 1.3517676199054267,
      "grad_norm": 1.3380517959594727,
      "learning_rate": 2.379180638091571e-05,
      "loss": 1.6191,
      "step": 7504
    },
    {
      "epoch": 1.3519477595136231,
      "grad_norm": 1.3716108798980713,
      "learning_rate": 2.377974723105839e-05,
      "loss": 1.9239,
      "step": 7505
    },
    {
      "epoch": 1.3521278991218195,
      "grad_norm": 1.3542637825012207,
      "learning_rate": 2.3767690184576975e-05,
      "loss": 1.8896,
      "step": 7506
    },
    {
      "epoch": 1.3523080387300157,
      "grad_norm": 1.525539755821228,
      "learning_rate": 2.3755635242438674e-05,
      "loss": 1.807,
      "step": 7507
    },
    {
      "epoch": 1.352488178338212,
      "grad_norm": 1.6884543895721436,
      "learning_rate": 2.3743582405610514e-05,
      "loss": 1.8735,
      "step": 7508
    },
    {
      "epoch": 1.3526683179464085,
      "grad_norm": 1.6111360788345337,
      "learning_rate": 2.3731531675059404e-05,
      "loss": 2.069,
      "step": 7509
    },
    {
      "epoch": 1.3528484575546047,
      "grad_norm": 1.5447648763656616,
      "learning_rate": 2.371948305175202e-05,
      "loss": 1.5468,
      "step": 7510
    },
    {
      "epoch": 1.353028597162801,
      "grad_norm": 1.9726524353027344,
      "learning_rate": 2.3707436536654913e-05,
      "loss": 2.3478,
      "step": 7511
    },
    {
      "epoch": 1.3532087367709975,
      "grad_norm": 1.633581519126892,
      "learning_rate": 2.3695392130734445e-05,
      "loss": 1.6174,
      "step": 7512
    },
    {
      "epoch": 1.353388876379194,
      "grad_norm": 1.3164770603179932,
      "learning_rate": 2.3683349834956797e-05,
      "loss": 1.3278,
      "step": 7513
    },
    {
      "epoch": 1.3535690159873903,
      "grad_norm": 1.4568259716033936,
      "learning_rate": 2.3671309650288017e-05,
      "loss": 1.5152,
      "step": 7514
    },
    {
      "epoch": 1.3537491555955865,
      "grad_norm": 1.4861228466033936,
      "learning_rate": 2.3659271577693942e-05,
      "loss": 1.4971,
      "step": 7515
    },
    {
      "epoch": 1.353929295203783,
      "grad_norm": 1.3664424419403076,
      "learning_rate": 2.364723561814028e-05,
      "loss": 1.5865,
      "step": 7516
    },
    {
      "epoch": 1.3541094348119793,
      "grad_norm": 1.6298110485076904,
      "learning_rate": 2.3635201772592537e-05,
      "loss": 1.7763,
      "step": 7517
    },
    {
      "epoch": 1.3542895744201755,
      "grad_norm": 1.426570177078247,
      "learning_rate": 2.3623170042016045e-05,
      "loss": 1.3826,
      "step": 7518
    },
    {
      "epoch": 1.354469714028372,
      "grad_norm": 1.3917127847671509,
      "learning_rate": 2.361114042737602e-05,
      "loss": 1.4356,
      "step": 7519
    },
    {
      "epoch": 1.3546498536365683,
      "grad_norm": 1.3347135782241821,
      "learning_rate": 2.359911292963747e-05,
      "loss": 1.2705,
      "step": 7520
    },
    {
      "epoch": 1.3548299932447647,
      "grad_norm": 1.599500298500061,
      "learning_rate": 2.3587087549765207e-05,
      "loss": 1.6841,
      "step": 7521
    },
    {
      "epoch": 1.3550101328529611,
      "grad_norm": 1.3436247110366821,
      "learning_rate": 2.3575064288723925e-05,
      "loss": 1.4051,
      "step": 7522
    },
    {
      "epoch": 1.3551902724611575,
      "grad_norm": 1.3791656494140625,
      "learning_rate": 2.3563043147478108e-05,
      "loss": 1.5266,
      "step": 7523
    },
    {
      "epoch": 1.3553704120693537,
      "grad_norm": 1.4408009052276611,
      "learning_rate": 2.35510241269921e-05,
      "loss": 1.4167,
      "step": 7524
    },
    {
      "epoch": 1.3555505516775501,
      "grad_norm": 1.5563013553619385,
      "learning_rate": 2.3539007228230048e-05,
      "loss": 1.5547,
      "step": 7525
    },
    {
      "epoch": 1.3557306912857465,
      "grad_norm": 1.438392996788025,
      "learning_rate": 2.352699245215595e-05,
      "loss": 1.3598,
      "step": 7526
    },
    {
      "epoch": 1.3559108308939427,
      "grad_norm": 1.457525372505188,
      "learning_rate": 2.35149797997336e-05,
      "loss": 1.4185,
      "step": 7527
    },
    {
      "epoch": 1.3560909705021391,
      "grad_norm": 1.6293107271194458,
      "learning_rate": 2.3502969271926694e-05,
      "loss": 1.2576,
      "step": 7528
    },
    {
      "epoch": 1.3562711101103355,
      "grad_norm": 1.534010410308838,
      "learning_rate": 2.349096086969868e-05,
      "loss": 1.325,
      "step": 7529
    },
    {
      "epoch": 1.356451249718532,
      "grad_norm": 1.5167186260223389,
      "learning_rate": 2.347895459401288e-05,
      "loss": 1.5054,
      "step": 7530
    },
    {
      "epoch": 1.3566313893267283,
      "grad_norm": 1.6355319023132324,
      "learning_rate": 2.3466950445832425e-05,
      "loss": 1.4301,
      "step": 7531
    },
    {
      "epoch": 1.3568115289349245,
      "grad_norm": 1.4158082008361816,
      "learning_rate": 2.345494842612026e-05,
      "loss": 1.3088,
      "step": 7532
    },
    {
      "epoch": 1.356991668543121,
      "grad_norm": 1.5112338066101074,
      "learning_rate": 2.3442948535839236e-05,
      "loss": 1.5923,
      "step": 7533
    },
    {
      "epoch": 1.3571718081513173,
      "grad_norm": 1.4129796028137207,
      "learning_rate": 2.343095077595193e-05,
      "loss": 1.3166,
      "step": 7534
    },
    {
      "epoch": 1.3573519477595135,
      "grad_norm": 1.4962005615234375,
      "learning_rate": 2.3418955147420813e-05,
      "loss": 1.4034,
      "step": 7535
    },
    {
      "epoch": 1.35753208736771,
      "grad_norm": 1.4700889587402344,
      "learning_rate": 2.3406961651208147e-05,
      "loss": 1.5653,
      "step": 7536
    },
    {
      "epoch": 1.3577122269759063,
      "grad_norm": 1.6155070066452026,
      "learning_rate": 2.3394970288276086e-05,
      "loss": 1.4549,
      "step": 7537
    },
    {
      "epoch": 1.3578923665841027,
      "grad_norm": 1.526595115661621,
      "learning_rate": 2.3382981059586546e-05,
      "loss": 1.497,
      "step": 7538
    },
    {
      "epoch": 1.3580725061922991,
      "grad_norm": 1.4552953243255615,
      "learning_rate": 2.33709939661013e-05,
      "loss": 1.4246,
      "step": 7539
    },
    {
      "epoch": 1.3582526458004953,
      "grad_norm": 1.4766429662704468,
      "learning_rate": 2.3359009008781945e-05,
      "loss": 1.6351,
      "step": 7540
    },
    {
      "epoch": 1.3584327854086917,
      "grad_norm": 1.5850279331207275,
      "learning_rate": 2.33470261885899e-05,
      "loss": 1.4329,
      "step": 7541
    },
    {
      "epoch": 1.3586129250168881,
      "grad_norm": 1.4105831384658813,
      "learning_rate": 2.3335045506486447e-05,
      "loss": 1.3036,
      "step": 7542
    },
    {
      "epoch": 1.3587930646250843,
      "grad_norm": 1.6465802192687988,
      "learning_rate": 2.3323066963432656e-05,
      "loss": 1.5756,
      "step": 7543
    },
    {
      "epoch": 1.3589732042332807,
      "grad_norm": 1.5706514120101929,
      "learning_rate": 2.3311090560389436e-05,
      "loss": 1.527,
      "step": 7544
    },
    {
      "epoch": 1.3591533438414771,
      "grad_norm": 1.4764642715454102,
      "learning_rate": 2.3299116298317553e-05,
      "loss": 1.3045,
      "step": 7545
    },
    {
      "epoch": 1.3593334834496735,
      "grad_norm": 1.35222327709198,
      "learning_rate": 2.3287144178177517e-05,
      "loss": 1.1929,
      "step": 7546
    },
    {
      "epoch": 1.35951362305787,
      "grad_norm": 1.383483648300171,
      "learning_rate": 2.3275174200929784e-05,
      "loss": 1.1262,
      "step": 7547
    },
    {
      "epoch": 1.3596937626660661,
      "grad_norm": 1.3822935819625854,
      "learning_rate": 2.3263206367534556e-05,
      "loss": 1.0622,
      "step": 7548
    },
    {
      "epoch": 1.3598739022742625,
      "grad_norm": 1.5210989713668823,
      "learning_rate": 2.325124067895189e-05,
      "loss": 1.3703,
      "step": 7549
    },
    {
      "epoch": 1.360054041882459,
      "grad_norm": 1.4833420515060425,
      "learning_rate": 2.3239277136141657e-05,
      "loss": 1.4321,
      "step": 7550
    },
    {
      "epoch": 1.3602341814906553,
      "grad_norm": 1.4262073040008545,
      "learning_rate": 2.3227315740063587e-05,
      "loss": 1.2162,
      "step": 7551
    },
    {
      "epoch": 1.3604143210988515,
      "grad_norm": 1.285927653312683,
      "learning_rate": 2.3215356491677216e-05,
      "loss": 1.6129,
      "step": 7552
    },
    {
      "epoch": 1.360594460707048,
      "grad_norm": 1.3243283033370972,
      "learning_rate": 2.3203399391941894e-05,
      "loss": 1.9194,
      "step": 7553
    },
    {
      "epoch": 1.3607746003152443,
      "grad_norm": 1.4116501808166504,
      "learning_rate": 2.3191444441816824e-05,
      "loss": 1.9162,
      "step": 7554
    },
    {
      "epoch": 1.3609547399234407,
      "grad_norm": 1.4555381536483765,
      "learning_rate": 2.317949164226102e-05,
      "loss": 1.7655,
      "step": 7555
    },
    {
      "epoch": 1.3611348795316371,
      "grad_norm": 1.432364821434021,
      "learning_rate": 2.3167540994233332e-05,
      "loss": 1.8505,
      "step": 7556
    },
    {
      "epoch": 1.3613150191398333,
      "grad_norm": 1.5342450141906738,
      "learning_rate": 2.315559249869244e-05,
      "loss": 1.8613,
      "step": 7557
    },
    {
      "epoch": 1.3614951587480297,
      "grad_norm": 1.6430089473724365,
      "learning_rate": 2.3143646156596845e-05,
      "loss": 2.2923,
      "step": 7558
    },
    {
      "epoch": 1.3616752983562261,
      "grad_norm": 1.593636393547058,
      "learning_rate": 2.3131701968904844e-05,
      "loss": 1.9379,
      "step": 7559
    },
    {
      "epoch": 1.3618554379644223,
      "grad_norm": 1.5477970838546753,
      "learning_rate": 2.3119759936574653e-05,
      "loss": 1.8448,
      "step": 7560
    },
    {
      "epoch": 1.3620355775726187,
      "grad_norm": 1.6547980308532715,
      "learning_rate": 2.3107820060564222e-05,
      "loss": 1.6139,
      "step": 7561
    },
    {
      "epoch": 1.3622157171808151,
      "grad_norm": 1.6107239723205566,
      "learning_rate": 2.3095882341831372e-05,
      "loss": 1.437,
      "step": 7562
    },
    {
      "epoch": 1.3623958567890115,
      "grad_norm": 1.5189756155014038,
      "learning_rate": 2.308394678133373e-05,
      "loss": 1.6508,
      "step": 7563
    },
    {
      "epoch": 1.362575996397208,
      "grad_norm": 1.6922158002853394,
      "learning_rate": 2.307201338002875e-05,
      "loss": 1.8281,
      "step": 7564
    },
    {
      "epoch": 1.3627561360054041,
      "grad_norm": 1.3784329891204834,
      "learning_rate": 2.306008213887378e-05,
      "loss": 1.4116,
      "step": 7565
    },
    {
      "epoch": 1.3629362756136005,
      "grad_norm": 1.5929391384124756,
      "learning_rate": 2.3048153058825878e-05,
      "loss": 1.7402,
      "step": 7566
    },
    {
      "epoch": 1.363116415221797,
      "grad_norm": 1.3289282321929932,
      "learning_rate": 2.303622614084201e-05,
      "loss": 1.3152,
      "step": 7567
    },
    {
      "epoch": 1.3632965548299931,
      "grad_norm": 1.4627569913864136,
      "learning_rate": 2.302430138587895e-05,
      "loss": 1.5799,
      "step": 7568
    },
    {
      "epoch": 1.3634766944381895,
      "grad_norm": 1.4813125133514404,
      "learning_rate": 2.3012378794893275e-05,
      "loss": 1.3834,
      "step": 7569
    },
    {
      "epoch": 1.363656834046386,
      "grad_norm": 1.4358408451080322,
      "learning_rate": 2.3000458368841442e-05,
      "loss": 1.3643,
      "step": 7570
    },
    {
      "epoch": 1.3638369736545823,
      "grad_norm": 1.3621000051498413,
      "learning_rate": 2.2988540108679692e-05,
      "loss": 1.4555,
      "step": 7571
    },
    {
      "epoch": 1.3640171132627787,
      "grad_norm": 1.3113750219345093,
      "learning_rate": 2.2976624015364095e-05,
      "loss": 1.1542,
      "step": 7572
    },
    {
      "epoch": 1.364197252870975,
      "grad_norm": 1.494115948677063,
      "learning_rate": 2.296471008985054e-05,
      "loss": 1.5979,
      "step": 7573
    },
    {
      "epoch": 1.3643773924791713,
      "grad_norm": 1.272529125213623,
      "learning_rate": 2.2952798333094788e-05,
      "loss": 1.2037,
      "step": 7574
    },
    {
      "epoch": 1.3645575320873677,
      "grad_norm": 1.3943688869476318,
      "learning_rate": 2.2940888746052386e-05,
      "loss": 1.2098,
      "step": 7575
    },
    {
      "epoch": 1.364737671695564,
      "grad_norm": 1.3633955717086792,
      "learning_rate": 2.29289813296787e-05,
      "loss": 1.4044,
      "step": 7576
    },
    {
      "epoch": 1.3649178113037603,
      "grad_norm": 1.6230963468551636,
      "learning_rate": 2.291707608492895e-05,
      "loss": 1.4544,
      "step": 7577
    },
    {
      "epoch": 1.3650979509119567,
      "grad_norm": 1.3644753694534302,
      "learning_rate": 2.2905173012758168e-05,
      "loss": 1.1211,
      "step": 7578
    },
    {
      "epoch": 1.3652780905201531,
      "grad_norm": 1.446577548980713,
      "learning_rate": 2.2893272114121206e-05,
      "loss": 1.4278,
      "step": 7579
    },
    {
      "epoch": 1.3654582301283495,
      "grad_norm": 1.5131093263626099,
      "learning_rate": 2.2881373389972753e-05,
      "loss": 1.3811,
      "step": 7580
    },
    {
      "epoch": 1.365638369736546,
      "grad_norm": 1.3555972576141357,
      "learning_rate": 2.286947684126732e-05,
      "loss": 1.2185,
      "step": 7581
    },
    {
      "epoch": 1.3658185093447421,
      "grad_norm": 1.4976664781570435,
      "learning_rate": 2.285758246895924e-05,
      "loss": 1.4818,
      "step": 7582
    },
    {
      "epoch": 1.3659986489529385,
      "grad_norm": 1.4231165647506714,
      "learning_rate": 2.2845690274002653e-05,
      "loss": 1.3611,
      "step": 7583
    },
    {
      "epoch": 1.366178788561135,
      "grad_norm": 1.4566525220870972,
      "learning_rate": 2.2833800257351586e-05,
      "loss": 1.3703,
      "step": 7584
    },
    {
      "epoch": 1.3663589281693311,
      "grad_norm": 1.4260296821594238,
      "learning_rate": 2.2821912419959825e-05,
      "loss": 1.3481,
      "step": 7585
    },
    {
      "epoch": 1.3665390677775275,
      "grad_norm": 1.4992929697036743,
      "learning_rate": 2.2810026762781017e-05,
      "loss": 1.3258,
      "step": 7586
    },
    {
      "epoch": 1.366719207385724,
      "grad_norm": 1.5576725006103516,
      "learning_rate": 2.2798143286768614e-05,
      "loss": 1.5788,
      "step": 7587
    },
    {
      "epoch": 1.3668993469939204,
      "grad_norm": 1.5120137929916382,
      "learning_rate": 2.2786261992875913e-05,
      "loss": 1.4573,
      "step": 7588
    },
    {
      "epoch": 1.3670794866021168,
      "grad_norm": 1.7288219928741455,
      "learning_rate": 2.2774382882056012e-05,
      "loss": 1.5498,
      "step": 7589
    },
    {
      "epoch": 1.367259626210313,
      "grad_norm": 1.5634130239486694,
      "learning_rate": 2.276250595526186e-05,
      "loss": 1.5107,
      "step": 7590
    },
    {
      "epoch": 1.3674397658185093,
      "grad_norm": 1.546922206878662,
      "learning_rate": 2.275063121344621e-05,
      "loss": 1.3574,
      "step": 7591
    },
    {
      "epoch": 1.3676199054267058,
      "grad_norm": 1.651639699935913,
      "learning_rate": 2.2738758657561636e-05,
      "loss": 1.5064,
      "step": 7592
    },
    {
      "epoch": 1.367800045034902,
      "grad_norm": 1.446218490600586,
      "learning_rate": 2.2726888288560582e-05,
      "loss": 1.5245,
      "step": 7593
    },
    {
      "epoch": 1.3679801846430983,
      "grad_norm": 1.435522437095642,
      "learning_rate": 2.271502010739527e-05,
      "loss": 1.325,
      "step": 7594
    },
    {
      "epoch": 1.3681603242512947,
      "grad_norm": 1.5314617156982422,
      "learning_rate": 2.270315411501775e-05,
      "loss": 1.4824,
      "step": 7595
    },
    {
      "epoch": 1.3683404638594912,
      "grad_norm": 1.4876232147216797,
      "learning_rate": 2.269129031237991e-05,
      "loss": 1.3691,
      "step": 7596
    },
    {
      "epoch": 1.3685206034676876,
      "grad_norm": 1.5717575550079346,
      "learning_rate": 2.2679428700433443e-05,
      "loss": 1.3863,
      "step": 7597
    },
    {
      "epoch": 1.3687007430758837,
      "grad_norm": 1.6156270503997803,
      "learning_rate": 2.266756928012993e-05,
      "loss": 1.2923,
      "step": 7598
    },
    {
      "epoch": 1.3688808826840801,
      "grad_norm": 1.5814160108566284,
      "learning_rate": 2.2655712052420687e-05,
      "loss": 1.5252,
      "step": 7599
    },
    {
      "epoch": 1.3690610222922766,
      "grad_norm": 1.4465832710266113,
      "learning_rate": 2.2643857018256898e-05,
      "loss": 1.2232,
      "step": 7600
    },
    {
      "epoch": 1.3692411619004727,
      "grad_norm": 1.3768585920333862,
      "learning_rate": 2.2632004178589555e-05,
      "loss": 1.0957,
      "step": 7601
    },
    {
      "epoch": 1.3694213015086691,
      "grad_norm": 1.2455940246582031,
      "learning_rate": 2.2620153534369528e-05,
      "loss": 1.6452,
      "step": 7602
    },
    {
      "epoch": 1.3696014411168655,
      "grad_norm": 1.3670068979263306,
      "learning_rate": 2.260830508654745e-05,
      "loss": 1.8486,
      "step": 7603
    },
    {
      "epoch": 1.369781580725062,
      "grad_norm": 1.342037320137024,
      "learning_rate": 2.2596458836073792e-05,
      "loss": 1.6132,
      "step": 7604
    },
    {
      "epoch": 1.3699617203332584,
      "grad_norm": 1.4381548166275024,
      "learning_rate": 2.2584614783898854e-05,
      "loss": 1.8604,
      "step": 7605
    },
    {
      "epoch": 1.3701418599414548,
      "grad_norm": 1.346655011177063,
      "learning_rate": 2.257277293097276e-05,
      "loss": 1.788,
      "step": 7606
    },
    {
      "epoch": 1.370321999549651,
      "grad_norm": 1.3799290657043457,
      "learning_rate": 2.2560933278245468e-05,
      "loss": 1.4876,
      "step": 7607
    },
    {
      "epoch": 1.3705021391578474,
      "grad_norm": 1.3988481760025024,
      "learning_rate": 2.254909582666675e-05,
      "loss": 1.7434,
      "step": 7608
    },
    {
      "epoch": 1.3706822787660438,
      "grad_norm": 1.4916144609451294,
      "learning_rate": 2.2537260577186186e-05,
      "loss": 1.7952,
      "step": 7609
    },
    {
      "epoch": 1.37086241837424,
      "grad_norm": 1.638384222984314,
      "learning_rate": 2.252542753075323e-05,
      "loss": 1.9852,
      "step": 7610
    },
    {
      "epoch": 1.3710425579824363,
      "grad_norm": 1.7106362581253052,
      "learning_rate": 2.2513596688317052e-05,
      "loss": 1.7957,
      "step": 7611
    },
    {
      "epoch": 1.3712226975906328,
      "grad_norm": 1.400754690170288,
      "learning_rate": 2.2501768050826782e-05,
      "loss": 1.4118,
      "step": 7612
    },
    {
      "epoch": 1.3714028371988292,
      "grad_norm": 1.4144703149795532,
      "learning_rate": 2.248994161923128e-05,
      "loss": 1.4364,
      "step": 7613
    },
    {
      "epoch": 1.3715829768070256,
      "grad_norm": 1.5763905048370361,
      "learning_rate": 2.2478117394479263e-05,
      "loss": 1.7449,
      "step": 7614
    },
    {
      "epoch": 1.3717631164152218,
      "grad_norm": 1.326772928237915,
      "learning_rate": 2.246629537751925e-05,
      "loss": 1.341,
      "step": 7615
    },
    {
      "epoch": 1.3719432560234182,
      "grad_norm": 1.3690229654312134,
      "learning_rate": 2.2454475569299627e-05,
      "loss": 1.4092,
      "step": 7616
    },
    {
      "epoch": 1.3721233956316146,
      "grad_norm": 1.367480993270874,
      "learning_rate": 2.2442657970768554e-05,
      "loss": 1.3212,
      "step": 7617
    },
    {
      "epoch": 1.3723035352398107,
      "grad_norm": 1.364949345588684,
      "learning_rate": 2.2430842582874034e-05,
      "loss": 1.2856,
      "step": 7618
    },
    {
      "epoch": 1.3724836748480072,
      "grad_norm": 1.5036259889602661,
      "learning_rate": 2.2419029406563895e-05,
      "loss": 1.7055,
      "step": 7619
    },
    {
      "epoch": 1.3726638144562036,
      "grad_norm": 1.3882821798324585,
      "learning_rate": 2.240721844278579e-05,
      "loss": 1.3846,
      "step": 7620
    },
    {
      "epoch": 1.3728439540644,
      "grad_norm": 1.9397857189178467,
      "learning_rate": 2.2395409692487175e-05,
      "loss": 1.4153,
      "step": 7621
    },
    {
      "epoch": 1.3730240936725964,
      "grad_norm": 1.4227429628372192,
      "learning_rate": 2.238360315661535e-05,
      "loss": 1.4628,
      "step": 7622
    },
    {
      "epoch": 1.3732042332807926,
      "grad_norm": 1.4572449922561646,
      "learning_rate": 2.237179883611743e-05,
      "loss": 1.4179,
      "step": 7623
    },
    {
      "epoch": 1.373384372888989,
      "grad_norm": 1.4244329929351807,
      "learning_rate": 2.235999673194035e-05,
      "loss": 1.1817,
      "step": 7624
    },
    {
      "epoch": 1.3735645124971854,
      "grad_norm": 1.6989593505859375,
      "learning_rate": 2.2348196845030846e-05,
      "loss": 1.4607,
      "step": 7625
    },
    {
      "epoch": 1.3737446521053815,
      "grad_norm": 1.462368130683899,
      "learning_rate": 2.233639917633554e-05,
      "loss": 1.5085,
      "step": 7626
    },
    {
      "epoch": 1.373924791713578,
      "grad_norm": 1.441068410873413,
      "learning_rate": 2.2324603726800823e-05,
      "loss": 1.2522,
      "step": 7627
    },
    {
      "epoch": 1.3741049313217744,
      "grad_norm": 1.4737187623977661,
      "learning_rate": 2.2312810497372917e-05,
      "loss": 1.3806,
      "step": 7628
    },
    {
      "epoch": 1.3742850709299708,
      "grad_norm": 1.33798086643219,
      "learning_rate": 2.2301019488997844e-05,
      "loss": 1.2479,
      "step": 7629
    },
    {
      "epoch": 1.3744652105381672,
      "grad_norm": 1.5182621479034424,
      "learning_rate": 2.2289230702621533e-05,
      "loss": 1.5805,
      "step": 7630
    },
    {
      "epoch": 1.3746453501463634,
      "grad_norm": 1.3349831104278564,
      "learning_rate": 2.227744413918962e-05,
      "loss": 1.244,
      "step": 7631
    },
    {
      "epoch": 1.3748254897545598,
      "grad_norm": 1.5687263011932373,
      "learning_rate": 2.226565979964763e-05,
      "loss": 1.5459,
      "step": 7632
    },
    {
      "epoch": 1.3750056293627562,
      "grad_norm": 1.5157266855239868,
      "learning_rate": 2.2253877684940912e-05,
      "loss": 1.5024,
      "step": 7633
    },
    {
      "epoch": 1.3751857689709526,
      "grad_norm": 1.406571626663208,
      "learning_rate": 2.2242097796014588e-05,
      "loss": 1.3794,
      "step": 7634
    },
    {
      "epoch": 1.3753659085791488,
      "grad_norm": 1.4581559896469116,
      "learning_rate": 2.2230320133813676e-05,
      "loss": 1.2482,
      "step": 7635
    },
    {
      "epoch": 1.3755460481873452,
      "grad_norm": 1.5845005512237549,
      "learning_rate": 2.2218544699282955e-05,
      "loss": 1.5547,
      "step": 7636
    },
    {
      "epoch": 1.3757261877955416,
      "grad_norm": 1.4608479738235474,
      "learning_rate": 2.2206771493367052e-05,
      "loss": 1.3299,
      "step": 7637
    },
    {
      "epoch": 1.375906327403738,
      "grad_norm": 1.4873682260513306,
      "learning_rate": 2.2195000517010395e-05,
      "loss": 1.3317,
      "step": 7638
    },
    {
      "epoch": 1.3760864670119344,
      "grad_norm": 1.5904486179351807,
      "learning_rate": 2.2183231771157243e-05,
      "loss": 1.5558,
      "step": 7639
    },
    {
      "epoch": 1.3762666066201306,
      "grad_norm": 1.5699273347854614,
      "learning_rate": 2.21714652567517e-05,
      "loss": 1.4763,
      "step": 7640
    },
    {
      "epoch": 1.376446746228327,
      "grad_norm": 1.5878998041152954,
      "learning_rate": 2.2159700974737662e-05,
      "loss": 1.5285,
      "step": 7641
    },
    {
      "epoch": 1.3766268858365234,
      "grad_norm": 1.5779407024383545,
      "learning_rate": 2.214793892605887e-05,
      "loss": 1.5131,
      "step": 7642
    },
    {
      "epoch": 1.3768070254447196,
      "grad_norm": 1.655317783355713,
      "learning_rate": 2.2136179111658816e-05,
      "loss": 1.3801,
      "step": 7643
    },
    {
      "epoch": 1.376987165052916,
      "grad_norm": 1.5089733600616455,
      "learning_rate": 2.212442153248092e-05,
      "loss": 1.5128,
      "step": 7644
    },
    {
      "epoch": 1.3771673046611124,
      "grad_norm": 1.4732587337493896,
      "learning_rate": 2.211266618946835e-05,
      "loss": 1.236,
      "step": 7645
    },
    {
      "epoch": 1.3773474442693088,
      "grad_norm": 1.4909825325012207,
      "learning_rate": 2.2100913083564113e-05,
      "loss": 1.3546,
      "step": 7646
    },
    {
      "epoch": 1.3775275838775052,
      "grad_norm": 1.6412643194198608,
      "learning_rate": 2.208916221571104e-05,
      "loss": 1.4825,
      "step": 7647
    },
    {
      "epoch": 1.3777077234857014,
      "grad_norm": 1.5115211009979248,
      "learning_rate": 2.2077413586851763e-05,
      "loss": 1.2195,
      "step": 7648
    },
    {
      "epoch": 1.3778878630938978,
      "grad_norm": 1.4692808389663696,
      "learning_rate": 2.2065667197928785e-05,
      "loss": 1.2303,
      "step": 7649
    },
    {
      "epoch": 1.3780680027020942,
      "grad_norm": 1.536497712135315,
      "learning_rate": 2.2053923049884378e-05,
      "loss": 1.2947,
      "step": 7650
    },
    {
      "epoch": 1.3782481423102904,
      "grad_norm": 1.413273811340332,
      "learning_rate": 2.2042181143660652e-05,
      "loss": 1.2069,
      "step": 7651
    },
    {
      "epoch": 1.3784282819184868,
      "grad_norm": 1.3353500366210938,
      "learning_rate": 2.2030441480199536e-05,
      "loss": 1.7804,
      "step": 7652
    },
    {
      "epoch": 1.3786084215266832,
      "grad_norm": 1.3828232288360596,
      "learning_rate": 2.201870406044278e-05,
      "loss": 1.7261,
      "step": 7653
    },
    {
      "epoch": 1.3787885611348796,
      "grad_norm": 1.3390487432479858,
      "learning_rate": 2.200696888533196e-05,
      "loss": 1.7177,
      "step": 7654
    },
    {
      "epoch": 1.378968700743076,
      "grad_norm": 1.2968525886535645,
      "learning_rate": 2.1995235955808452e-05,
      "loss": 1.4678,
      "step": 7655
    },
    {
      "epoch": 1.3791488403512722,
      "grad_norm": 1.4880366325378418,
      "learning_rate": 2.198350527281348e-05,
      "loss": 1.8316,
      "step": 7656
    },
    {
      "epoch": 1.3793289799594686,
      "grad_norm": 1.4430006742477417,
      "learning_rate": 2.1971776837288056e-05,
      "loss": 1.7934,
      "step": 7657
    },
    {
      "epoch": 1.379509119567665,
      "grad_norm": 1.4531675577163696,
      "learning_rate": 2.1960050650173054e-05,
      "loss": 1.8011,
      "step": 7658
    },
    {
      "epoch": 1.3796892591758612,
      "grad_norm": 1.6168450117111206,
      "learning_rate": 2.1948326712409128e-05,
      "loss": 1.7135,
      "step": 7659
    },
    {
      "epoch": 1.3798693987840576,
      "grad_norm": 1.6906057596206665,
      "learning_rate": 2.1936605024936775e-05,
      "loss": 1.882,
      "step": 7660
    },
    {
      "epoch": 1.380049538392254,
      "grad_norm": 1.670449137687683,
      "learning_rate": 2.1924885588696297e-05,
      "loss": 1.7769,
      "step": 7661
    },
    {
      "epoch": 1.3802296780004504,
      "grad_norm": 1.4827964305877686,
      "learning_rate": 2.191316840462782e-05,
      "loss": 1.4404,
      "step": 7662
    },
    {
      "epoch": 1.3804098176086468,
      "grad_norm": 1.3820644617080688,
      "learning_rate": 2.1901453473671295e-05,
      "loss": 1.4777,
      "step": 7663
    },
    {
      "epoch": 1.3805899572168432,
      "grad_norm": 1.4032403230667114,
      "learning_rate": 2.188974079676649e-05,
      "loss": 1.4918,
      "step": 7664
    },
    {
      "epoch": 1.3807700968250394,
      "grad_norm": 1.4772493839263916,
      "learning_rate": 2.1878030374852982e-05,
      "loss": 1.3491,
      "step": 7665
    },
    {
      "epoch": 1.3809502364332358,
      "grad_norm": 1.542166829109192,
      "learning_rate": 2.186632220887016e-05,
      "loss": 1.576,
      "step": 7666
    },
    {
      "epoch": 1.3811303760414322,
      "grad_norm": 1.3230671882629395,
      "learning_rate": 2.1854616299757285e-05,
      "loss": 1.2233,
      "step": 7667
    },
    {
      "epoch": 1.3813105156496284,
      "grad_norm": 1.523544192314148,
      "learning_rate": 2.1842912648453384e-05,
      "loss": 1.5538,
      "step": 7668
    },
    {
      "epoch": 1.3814906552578248,
      "grad_norm": 1.3466370105743408,
      "learning_rate": 2.1831211255897316e-05,
      "loss": 1.1174,
      "step": 7669
    },
    {
      "epoch": 1.3816707948660212,
      "grad_norm": 1.5028636455535889,
      "learning_rate": 2.1819512123027757e-05,
      "loss": 1.6191,
      "step": 7670
    },
    {
      "epoch": 1.3818509344742176,
      "grad_norm": 1.4414459466934204,
      "learning_rate": 2.1807815250783192e-05,
      "loss": 1.3771,
      "step": 7671
    },
    {
      "epoch": 1.382031074082414,
      "grad_norm": 1.4875317811965942,
      "learning_rate": 2.1796120640101976e-05,
      "loss": 1.5545,
      "step": 7672
    },
    {
      "epoch": 1.3822112136906102,
      "grad_norm": 1.4802964925765991,
      "learning_rate": 2.1784428291922222e-05,
      "loss": 1.6311,
      "step": 7673
    },
    {
      "epoch": 1.3823913532988066,
      "grad_norm": 1.5269854068756104,
      "learning_rate": 2.177273820718189e-05,
      "loss": 1.4111,
      "step": 7674
    },
    {
      "epoch": 1.382571492907003,
      "grad_norm": 1.4962408542633057,
      "learning_rate": 2.1761050386818765e-05,
      "loss": 1.4957,
      "step": 7675
    },
    {
      "epoch": 1.3827516325151992,
      "grad_norm": 1.3393136262893677,
      "learning_rate": 2.174936483177039e-05,
      "loss": 1.2472,
      "step": 7676
    },
    {
      "epoch": 1.3829317721233956,
      "grad_norm": 1.3940626382827759,
      "learning_rate": 2.1737681542974226e-05,
      "loss": 1.5632,
      "step": 7677
    },
    {
      "epoch": 1.383111911731592,
      "grad_norm": 1.353214144706726,
      "learning_rate": 2.172600052136748e-05,
      "loss": 1.301,
      "step": 7678
    },
    {
      "epoch": 1.3832920513397884,
      "grad_norm": 1.4809770584106445,
      "learning_rate": 2.1714321767887202e-05,
      "loss": 1.4772,
      "step": 7679
    },
    {
      "epoch": 1.3834721909479848,
      "grad_norm": 1.5792030096054077,
      "learning_rate": 2.1702645283470236e-05,
      "loss": 1.4207,
      "step": 7680
    },
    {
      "epoch": 1.383652330556181,
      "grad_norm": 1.4190375804901123,
      "learning_rate": 2.16909710690533e-05,
      "loss": 1.2509,
      "step": 7681
    },
    {
      "epoch": 1.3838324701643774,
      "grad_norm": 1.5044050216674805,
      "learning_rate": 2.1679299125572877e-05,
      "loss": 1.4268,
      "step": 7682
    },
    {
      "epoch": 1.3840126097725738,
      "grad_norm": 1.4418790340423584,
      "learning_rate": 2.1667629453965284e-05,
      "loss": 1.3362,
      "step": 7683
    },
    {
      "epoch": 1.38419274938077,
      "grad_norm": 1.4872791767120361,
      "learning_rate": 2.165596205516665e-05,
      "loss": 1.5002,
      "step": 7684
    },
    {
      "epoch": 1.3843728889889664,
      "grad_norm": 1.5594156980514526,
      "learning_rate": 2.164429693011294e-05,
      "loss": 1.5615,
      "step": 7685
    },
    {
      "epoch": 1.3845530285971628,
      "grad_norm": 1.5328456163406372,
      "learning_rate": 2.1632634079739922e-05,
      "loss": 1.5734,
      "step": 7686
    },
    {
      "epoch": 1.3847331682053592,
      "grad_norm": 1.5469337701797485,
      "learning_rate": 2.1620973504983172e-05,
      "loss": 1.5205,
      "step": 7687
    },
    {
      "epoch": 1.3849133078135556,
      "grad_norm": 1.5033612251281738,
      "learning_rate": 2.1609315206778108e-05,
      "loss": 1.1651,
      "step": 7688
    },
    {
      "epoch": 1.3850934474217518,
      "grad_norm": 1.5737769603729248,
      "learning_rate": 2.1597659186059955e-05,
      "loss": 1.5629,
      "step": 7689
    },
    {
      "epoch": 1.3852735870299482,
      "grad_norm": 1.4151990413665771,
      "learning_rate": 2.1586005443763722e-05,
      "loss": 1.2574,
      "step": 7690
    },
    {
      "epoch": 1.3854537266381446,
      "grad_norm": 1.4447829723358154,
      "learning_rate": 2.1574353980824315e-05,
      "loss": 1.4316,
      "step": 7691
    },
    {
      "epoch": 1.385633866246341,
      "grad_norm": 1.4259389638900757,
      "learning_rate": 2.1562704798176385e-05,
      "loss": 1.3603,
      "step": 7692
    },
    {
      "epoch": 1.3858140058545372,
      "grad_norm": 1.6830917596817017,
      "learning_rate": 2.1551057896754422e-05,
      "loss": 1.582,
      "step": 7693
    },
    {
      "epoch": 1.3859941454627336,
      "grad_norm": 1.476889729499817,
      "learning_rate": 2.153941327749272e-05,
      "loss": 1.2161,
      "step": 7694
    },
    {
      "epoch": 1.38617428507093,
      "grad_norm": 1.7358243465423584,
      "learning_rate": 2.1527770941325463e-05,
      "loss": 1.6931,
      "step": 7695
    },
    {
      "epoch": 1.3863544246791264,
      "grad_norm": 1.5765501260757446,
      "learning_rate": 2.1516130889186532e-05,
      "loss": 1.3892,
      "step": 7696
    },
    {
      "epoch": 1.3865345642873228,
      "grad_norm": 1.5216618776321411,
      "learning_rate": 2.1504493122009707e-05,
      "loss": 1.457,
      "step": 7697
    },
    {
      "epoch": 1.386714703895519,
      "grad_norm": 1.5311753749847412,
      "learning_rate": 2.1492857640728564e-05,
      "loss": 1.4901,
      "step": 7698
    },
    {
      "epoch": 1.3868948435037154,
      "grad_norm": 1.5865960121154785,
      "learning_rate": 2.1481224446276475e-05,
      "loss": 1.3783,
      "step": 7699
    },
    {
      "epoch": 1.3870749831119118,
      "grad_norm": 1.4798338413238525,
      "learning_rate": 2.1469593539586695e-05,
      "loss": 1.1062,
      "step": 7700
    },
    {
      "epoch": 1.387255122720108,
      "grad_norm": 1.4690207242965698,
      "learning_rate": 2.145796492159222e-05,
      "loss": 1.3676,
      "step": 7701
    },
    {
      "epoch": 1.3874352623283044,
      "grad_norm": 1.4369398355484009,
      "learning_rate": 2.1446338593225897e-05,
      "loss": 1.964,
      "step": 7702
    },
    {
      "epoch": 1.3876154019365008,
      "grad_norm": 1.4196062088012695,
      "learning_rate": 2.1434714555420383e-05,
      "loss": 2.0065,
      "step": 7703
    },
    {
      "epoch": 1.3877955415446972,
      "grad_norm": 1.4292179346084595,
      "learning_rate": 2.1423092809108138e-05,
      "loss": 1.974,
      "step": 7704
    },
    {
      "epoch": 1.3879756811528936,
      "grad_norm": 1.4649410247802734,
      "learning_rate": 2.1411473355221484e-05,
      "loss": 1.8785,
      "step": 7705
    },
    {
      "epoch": 1.3881558207610898,
      "grad_norm": 1.6712722778320312,
      "learning_rate": 2.1399856194692513e-05,
      "loss": 1.9159,
      "step": 7706
    },
    {
      "epoch": 1.3883359603692862,
      "grad_norm": 1.5062775611877441,
      "learning_rate": 2.1388241328453168e-05,
      "loss": 1.8301,
      "step": 7707
    },
    {
      "epoch": 1.3885160999774826,
      "grad_norm": 1.4207886457443237,
      "learning_rate": 2.137662875743513e-05,
      "loss": 1.6508,
      "step": 7708
    },
    {
      "epoch": 1.3886962395856788,
      "grad_norm": 1.4575504064559937,
      "learning_rate": 2.136501848257001e-05,
      "loss": 1.6889,
      "step": 7709
    },
    {
      "epoch": 1.3888763791938752,
      "grad_norm": 1.6852158308029175,
      "learning_rate": 2.1353410504789155e-05,
      "loss": 1.996,
      "step": 7710
    },
    {
      "epoch": 1.3890565188020716,
      "grad_norm": 1.7862882614135742,
      "learning_rate": 2.134180482502376e-05,
      "loss": 2.0916,
      "step": 7711
    },
    {
      "epoch": 1.389236658410268,
      "grad_norm": 1.588097333908081,
      "learning_rate": 2.1330201444204824e-05,
      "loss": 1.7535,
      "step": 7712
    },
    {
      "epoch": 1.3894167980184644,
      "grad_norm": 1.5554462671279907,
      "learning_rate": 2.131860036326314e-05,
      "loss": 1.5682,
      "step": 7713
    },
    {
      "epoch": 1.3895969376266606,
      "grad_norm": 1.4419959783554077,
      "learning_rate": 2.1307001583129387e-05,
      "loss": 1.3517,
      "step": 7714
    },
    {
      "epoch": 1.389777077234857,
      "grad_norm": 1.3815621137619019,
      "learning_rate": 2.1295405104733994e-05,
      "loss": 1.525,
      "step": 7715
    },
    {
      "epoch": 1.3899572168430534,
      "grad_norm": 1.4243804216384888,
      "learning_rate": 2.128381092900722e-05,
      "loss": 1.4571,
      "step": 7716
    },
    {
      "epoch": 1.3901373564512496,
      "grad_norm": 1.5259944200515747,
      "learning_rate": 2.1272219056879145e-05,
      "loss": 1.6857,
      "step": 7717
    },
    {
      "epoch": 1.390317496059446,
      "grad_norm": 1.5324000120162964,
      "learning_rate": 2.126062948927966e-05,
      "loss": 1.5487,
      "step": 7718
    },
    {
      "epoch": 1.3904976356676424,
      "grad_norm": 1.4288406372070312,
      "learning_rate": 2.1249042227138482e-05,
      "loss": 1.3615,
      "step": 7719
    },
    {
      "epoch": 1.3906777752758388,
      "grad_norm": 1.6133449077606201,
      "learning_rate": 2.1237457271385135e-05,
      "loss": 1.6416,
      "step": 7720
    },
    {
      "epoch": 1.3908579148840352,
      "grad_norm": 1.3956414461135864,
      "learning_rate": 2.1225874622948953e-05,
      "loss": 1.3475,
      "step": 7721
    },
    {
      "epoch": 1.3910380544922316,
      "grad_norm": 1.3993265628814697,
      "learning_rate": 2.1214294282759078e-05,
      "loss": 1.4546,
      "step": 7722
    },
    {
      "epoch": 1.3912181941004278,
      "grad_norm": 1.4617594480514526,
      "learning_rate": 2.1202716251744514e-05,
      "loss": 1.3715,
      "step": 7723
    },
    {
      "epoch": 1.3913983337086242,
      "grad_norm": 1.438252568244934,
      "learning_rate": 2.1191140530834024e-05,
      "loss": 1.4534,
      "step": 7724
    },
    {
      "epoch": 1.3915784733168206,
      "grad_norm": 1.406053900718689,
      "learning_rate": 2.1179567120956206e-05,
      "loss": 1.2718,
      "step": 7725
    },
    {
      "epoch": 1.3917586129250168,
      "grad_norm": 1.5558401346206665,
      "learning_rate": 2.1167996023039476e-05,
      "loss": 1.414,
      "step": 7726
    },
    {
      "epoch": 1.3919387525332132,
      "grad_norm": 1.377637267112732,
      "learning_rate": 2.1156427238012067e-05,
      "loss": 1.3197,
      "step": 7727
    },
    {
      "epoch": 1.3921188921414096,
      "grad_norm": 1.4500179290771484,
      "learning_rate": 2.114486076680201e-05,
      "loss": 1.3714,
      "step": 7728
    },
    {
      "epoch": 1.392299031749606,
      "grad_norm": 1.4364067316055298,
      "learning_rate": 2.1133296610337168e-05,
      "loss": 1.5057,
      "step": 7729
    },
    {
      "epoch": 1.3924791713578024,
      "grad_norm": 1.4090951681137085,
      "learning_rate": 2.1121734769545216e-05,
      "loss": 1.3518,
      "step": 7730
    },
    {
      "epoch": 1.3926593109659986,
      "grad_norm": 1.2920494079589844,
      "learning_rate": 2.1110175245353632e-05,
      "loss": 1.1892,
      "step": 7731
    },
    {
      "epoch": 1.392839450574195,
      "grad_norm": 1.4387471675872803,
      "learning_rate": 2.1098618038689694e-05,
      "loss": 1.3064,
      "step": 7732
    },
    {
      "epoch": 1.3930195901823914,
      "grad_norm": 1.4395453929901123,
      "learning_rate": 2.1087063150480562e-05,
      "loss": 1.4981,
      "step": 7733
    },
    {
      "epoch": 1.3931997297905876,
      "grad_norm": 1.7355272769927979,
      "learning_rate": 2.1075510581653146e-05,
      "loss": 1.6893,
      "step": 7734
    },
    {
      "epoch": 1.393379869398784,
      "grad_norm": 1.5270005464553833,
      "learning_rate": 2.1063960333134176e-05,
      "loss": 1.2634,
      "step": 7735
    },
    {
      "epoch": 1.3935600090069804,
      "grad_norm": 1.7191475629806519,
      "learning_rate": 2.1052412405850197e-05,
      "loss": 1.508,
      "step": 7736
    },
    {
      "epoch": 1.3937401486151768,
      "grad_norm": 1.381528377532959,
      "learning_rate": 2.1040866800727615e-05,
      "loss": 1.1787,
      "step": 7737
    },
    {
      "epoch": 1.3939202882233732,
      "grad_norm": 1.5310136079788208,
      "learning_rate": 2.1029323518692594e-05,
      "loss": 1.3795,
      "step": 7738
    },
    {
      "epoch": 1.3941004278315694,
      "grad_norm": 1.6061034202575684,
      "learning_rate": 2.1017782560671123e-05,
      "loss": 1.4098,
      "step": 7739
    },
    {
      "epoch": 1.3942805674397658,
      "grad_norm": 1.6720610857009888,
      "learning_rate": 2.1006243927589043e-05,
      "loss": 1.4564,
      "step": 7740
    },
    {
      "epoch": 1.3944607070479622,
      "grad_norm": 1.5601205825805664,
      "learning_rate": 2.099470762037191e-05,
      "loss": 1.3781,
      "step": 7741
    },
    {
      "epoch": 1.3946408466561584,
      "grad_norm": 1.4646220207214355,
      "learning_rate": 2.0983173639945232e-05,
      "loss": 1.3786,
      "step": 7742
    },
    {
      "epoch": 1.3948209862643548,
      "grad_norm": 1.5586062669754028,
      "learning_rate": 2.0971641987234227e-05,
      "loss": 1.4233,
      "step": 7743
    },
    {
      "epoch": 1.3950011258725512,
      "grad_norm": 1.6915987730026245,
      "learning_rate": 2.0960112663163957e-05,
      "loss": 1.4978,
      "step": 7744
    },
    {
      "epoch": 1.3951812654807476,
      "grad_norm": 1.725235939025879,
      "learning_rate": 2.0948585668659308e-05,
      "loss": 1.7309,
      "step": 7745
    },
    {
      "epoch": 1.395361405088944,
      "grad_norm": 1.4718632698059082,
      "learning_rate": 2.093706100464495e-05,
      "loss": 1.3467,
      "step": 7746
    },
    {
      "epoch": 1.3955415446971402,
      "grad_norm": 1.59532630443573,
      "learning_rate": 2.0925538672045412e-05,
      "loss": 1.3294,
      "step": 7747
    },
    {
      "epoch": 1.3957216843053366,
      "grad_norm": 1.705466628074646,
      "learning_rate": 2.0914018671785008e-05,
      "loss": 1.3364,
      "step": 7748
    },
    {
      "epoch": 1.395901823913533,
      "grad_norm": 1.4785093069076538,
      "learning_rate": 2.0902501004787862e-05,
      "loss": 1.5218,
      "step": 7749
    },
    {
      "epoch": 1.3960819635217294,
      "grad_norm": 1.6198146343231201,
      "learning_rate": 2.0890985671977904e-05,
      "loss": 1.3571,
      "step": 7750
    },
    {
      "epoch": 1.3962621031299256,
      "grad_norm": 1.4123060703277588,
      "learning_rate": 2.08794726742789e-05,
      "loss": 1.1225,
      "step": 7751
    },
    {
      "epoch": 1.396442242738122,
      "grad_norm": 1.4546451568603516,
      "learning_rate": 2.0867962012614418e-05,
      "loss": 1.6534,
      "step": 7752
    },
    {
      "epoch": 1.3966223823463184,
      "grad_norm": 1.234961748123169,
      "learning_rate": 2.0856453687907827e-05,
      "loss": 1.5735,
      "step": 7753
    },
    {
      "epoch": 1.3968025219545148,
      "grad_norm": 1.319766640663147,
      "learning_rate": 2.084494770108233e-05,
      "loss": 1.8374,
      "step": 7754
    },
    {
      "epoch": 1.3969826615627112,
      "grad_norm": 1.3299893140792847,
      "learning_rate": 2.083344405306091e-05,
      "loss": 1.8847,
      "step": 7755
    },
    {
      "epoch": 1.3971628011709074,
      "grad_norm": 1.3398196697235107,
      "learning_rate": 2.0821942744766414e-05,
      "loss": 1.683,
      "step": 7756
    },
    {
      "epoch": 1.3973429407791038,
      "grad_norm": 1.4054995775222778,
      "learning_rate": 2.0810443777121464e-05,
      "loss": 2.0796,
      "step": 7757
    },
    {
      "epoch": 1.3975230803873002,
      "grad_norm": 1.5112829208374023,
      "learning_rate": 2.07989471510485e-05,
      "loss": 1.8288,
      "step": 7758
    },
    {
      "epoch": 1.3977032199954964,
      "grad_norm": 1.4432308673858643,
      "learning_rate": 2.0787452867469764e-05,
      "loss": 1.8768,
      "step": 7759
    },
    {
      "epoch": 1.3978833596036928,
      "grad_norm": 1.4694452285766602,
      "learning_rate": 2.0775960927307337e-05,
      "loss": 2.0351,
      "step": 7760
    },
    {
      "epoch": 1.3980634992118892,
      "grad_norm": 1.7296438217163086,
      "learning_rate": 2.0764471331483083e-05,
      "loss": 2.1114,
      "step": 7761
    },
    {
      "epoch": 1.3982436388200856,
      "grad_norm": 1.6175496578216553,
      "learning_rate": 2.0752984080918703e-05,
      "loss": 1.753,
      "step": 7762
    },
    {
      "epoch": 1.398423778428282,
      "grad_norm": 1.4557983875274658,
      "learning_rate": 2.0741499176535695e-05,
      "loss": 1.6097,
      "step": 7763
    },
    {
      "epoch": 1.3986039180364782,
      "grad_norm": 1.279748558998108,
      "learning_rate": 2.073001661925535e-05,
      "loss": 1.408,
      "step": 7764
    },
    {
      "epoch": 1.3987840576446746,
      "grad_norm": 1.370500087738037,
      "learning_rate": 2.0718536409998834e-05,
      "loss": 1.6582,
      "step": 7765
    },
    {
      "epoch": 1.398964197252871,
      "grad_norm": 1.5277808904647827,
      "learning_rate": 2.0707058549687068e-05,
      "loss": 1.5586,
      "step": 7766
    },
    {
      "epoch": 1.3991443368610672,
      "grad_norm": 1.3419978618621826,
      "learning_rate": 2.0695583039240796e-05,
      "loss": 1.2688,
      "step": 7767
    },
    {
      "epoch": 1.3993244764692636,
      "grad_norm": 1.29177987575531,
      "learning_rate": 2.0684109879580577e-05,
      "loss": 1.3396,
      "step": 7768
    },
    {
      "epoch": 1.39950461607746,
      "grad_norm": 1.367904782295227,
      "learning_rate": 2.0672639071626765e-05,
      "loss": 1.41,
      "step": 7769
    },
    {
      "epoch": 1.3996847556856564,
      "grad_norm": 1.4549157619476318,
      "learning_rate": 2.0661170616299586e-05,
      "loss": 1.4626,
      "step": 7770
    },
    {
      "epoch": 1.3998648952938528,
      "grad_norm": 1.330974817276001,
      "learning_rate": 2.0649704514519004e-05,
      "loss": 1.3388,
      "step": 7771
    },
    {
      "epoch": 1.400045034902049,
      "grad_norm": 1.4854329824447632,
      "learning_rate": 2.0638240767204847e-05,
      "loss": 1.504,
      "step": 7772
    },
    {
      "epoch": 1.4002251745102454,
      "grad_norm": 1.4392707347869873,
      "learning_rate": 2.06267793752767e-05,
      "loss": 1.4376,
      "step": 7773
    },
    {
      "epoch": 1.4004053141184418,
      "grad_norm": 1.3751403093338013,
      "learning_rate": 2.0615320339653988e-05,
      "loss": 1.3028,
      "step": 7774
    },
    {
      "epoch": 1.400585453726638,
      "grad_norm": 1.3719784021377563,
      "learning_rate": 2.060386366125598e-05,
      "loss": 1.2964,
      "step": 7775
    },
    {
      "epoch": 1.4007655933348344,
      "grad_norm": 1.4615179300308228,
      "learning_rate": 2.0592409341001706e-05,
      "loss": 1.4476,
      "step": 7776
    },
    {
      "epoch": 1.4009457329430308,
      "grad_norm": 1.4169727563858032,
      "learning_rate": 2.058095737981004e-05,
      "loss": 1.1946,
      "step": 7777
    },
    {
      "epoch": 1.4011258725512272,
      "grad_norm": 1.5135514736175537,
      "learning_rate": 2.0569507778599617e-05,
      "loss": 1.3585,
      "step": 7778
    },
    {
      "epoch": 1.4013060121594236,
      "grad_norm": 1.5087056159973145,
      "learning_rate": 2.055806053828897e-05,
      "loss": 1.3662,
      "step": 7779
    },
    {
      "epoch": 1.40148615176762,
      "grad_norm": 1.3473104238510132,
      "learning_rate": 2.0546615659796364e-05,
      "loss": 1.3774,
      "step": 7780
    },
    {
      "epoch": 1.4016662913758162,
      "grad_norm": 1.464226484298706,
      "learning_rate": 2.0535173144039903e-05,
      "loss": 1.522,
      "step": 7781
    },
    {
      "epoch": 1.4018464309840126,
      "grad_norm": 1.3973422050476074,
      "learning_rate": 2.0523732991937498e-05,
      "loss": 1.2291,
      "step": 7782
    },
    {
      "epoch": 1.402026570592209,
      "grad_norm": 1.606042504310608,
      "learning_rate": 2.0512295204406878e-05,
      "loss": 1.6585,
      "step": 7783
    },
    {
      "epoch": 1.4022067102004052,
      "grad_norm": 1.6926218271255493,
      "learning_rate": 2.0500859782365573e-05,
      "loss": 1.7902,
      "step": 7784
    },
    {
      "epoch": 1.4023868498086016,
      "grad_norm": 1.49717116355896,
      "learning_rate": 2.048942672673093e-05,
      "loss": 1.5828,
      "step": 7785
    },
    {
      "epoch": 1.402566989416798,
      "grad_norm": 1.4541187286376953,
      "learning_rate": 2.04779960384201e-05,
      "loss": 1.326,
      "step": 7786
    },
    {
      "epoch": 1.4027471290249944,
      "grad_norm": 1.4698036909103394,
      "learning_rate": 2.0466567718350026e-05,
      "loss": 1.168,
      "step": 7787
    },
    {
      "epoch": 1.4029272686331908,
      "grad_norm": 1.5216681957244873,
      "learning_rate": 2.0455141767437524e-05,
      "loss": 1.5124,
      "step": 7788
    },
    {
      "epoch": 1.403107408241387,
      "grad_norm": 1.5490877628326416,
      "learning_rate": 2.0443718186599164e-05,
      "loss": 1.4346,
      "step": 7789
    },
    {
      "epoch": 1.4032875478495834,
      "grad_norm": 1.5008727312088013,
      "learning_rate": 2.043229697675133e-05,
      "loss": 1.2558,
      "step": 7790
    },
    {
      "epoch": 1.4034676874577798,
      "grad_norm": 1.8694827556610107,
      "learning_rate": 2.0420878138810227e-05,
      "loss": 1.7307,
      "step": 7791
    },
    {
      "epoch": 1.403647827065976,
      "grad_norm": 1.5814460515975952,
      "learning_rate": 2.040946167369188e-05,
      "loss": 1.4703,
      "step": 7792
    },
    {
      "epoch": 1.4038279666741724,
      "grad_norm": 1.6565017700195312,
      "learning_rate": 2.03980475823121e-05,
      "loss": 1.3891,
      "step": 7793
    },
    {
      "epoch": 1.4040081062823688,
      "grad_norm": 1.4405680894851685,
      "learning_rate": 2.0386635865586528e-05,
      "loss": 1.3573,
      "step": 7794
    },
    {
      "epoch": 1.4041882458905652,
      "grad_norm": 1.5605217218399048,
      "learning_rate": 2.0375226524430595e-05,
      "loss": 1.3605,
      "step": 7795
    },
    {
      "epoch": 1.4043683854987616,
      "grad_norm": 1.5561670064926147,
      "learning_rate": 2.0363819559759567e-05,
      "loss": 1.3627,
      "step": 7796
    },
    {
      "epoch": 1.4045485251069578,
      "grad_norm": 1.5371930599212646,
      "learning_rate": 2.035241497248848e-05,
      "loss": 1.3571,
      "step": 7797
    },
    {
      "epoch": 1.4047286647151542,
      "grad_norm": 1.4775609970092773,
      "learning_rate": 2.0341012763532243e-05,
      "loss": 1.4113,
      "step": 7798
    },
    {
      "epoch": 1.4049088043233506,
      "grad_norm": 1.5602045059204102,
      "learning_rate": 2.0329612933805515e-05,
      "loss": 1.2775,
      "step": 7799
    },
    {
      "epoch": 1.4050889439315468,
      "grad_norm": 1.3970675468444824,
      "learning_rate": 2.031821548422278e-05,
      "loss": 1.015,
      "step": 7800
    },
    {
      "epoch": 1.4052690835397432,
      "grad_norm": 1.5551159381866455,
      "learning_rate": 2.030682041569833e-05,
      "loss": 1.502,
      "step": 7801
    },
    {
      "epoch": 1.4054492231479396,
      "grad_norm": 1.2985392808914185,
      "learning_rate": 2.0295427729146293e-05,
      "loss": 1.5912,
      "step": 7802
    },
    {
      "epoch": 1.405629362756136,
      "grad_norm": 1.272081732749939,
      "learning_rate": 2.0284037425480583e-05,
      "loss": 1.6425,
      "step": 7803
    },
    {
      "epoch": 1.4058095023643324,
      "grad_norm": 1.3881675004959106,
      "learning_rate": 2.0272649505614915e-05,
      "loss": 2.0241,
      "step": 7804
    },
    {
      "epoch": 1.4059896419725288,
      "grad_norm": 1.5090227127075195,
      "learning_rate": 2.026126397046284e-05,
      "loss": 1.7893,
      "step": 7805
    },
    {
      "epoch": 1.406169781580725,
      "grad_norm": 1.481903314590454,
      "learning_rate": 2.024988082093765e-05,
      "loss": 1.6751,
      "step": 7806
    },
    {
      "epoch": 1.4063499211889214,
      "grad_norm": 1.3776445388793945,
      "learning_rate": 2.0238500057952547e-05,
      "loss": 1.8501,
      "step": 7807
    },
    {
      "epoch": 1.4065300607971178,
      "grad_norm": 1.4470198154449463,
      "learning_rate": 2.022712168242047e-05,
      "loss": 1.6271,
      "step": 7808
    },
    {
      "epoch": 1.406710200405314,
      "grad_norm": 1.6375354528427124,
      "learning_rate": 2.0215745695254194e-05,
      "loss": 1.9413,
      "step": 7809
    },
    {
      "epoch": 1.4068903400135104,
      "grad_norm": 1.6327897310256958,
      "learning_rate": 2.0204372097366287e-05,
      "loss": 1.878,
      "step": 7810
    },
    {
      "epoch": 1.4070704796217068,
      "grad_norm": 1.7835969924926758,
      "learning_rate": 2.019300088966912e-05,
      "loss": 1.8846,
      "step": 7811
    },
    {
      "epoch": 1.4072506192299032,
      "grad_norm": 1.527450442314148,
      "learning_rate": 2.0181632073074926e-05,
      "loss": 1.6527,
      "step": 7812
    },
    {
      "epoch": 1.4074307588380996,
      "grad_norm": 1.3308790922164917,
      "learning_rate": 2.017026564849568e-05,
      "loss": 1.4574,
      "step": 7813
    },
    {
      "epoch": 1.4076108984462958,
      "grad_norm": 1.3878746032714844,
      "learning_rate": 2.0158901616843196e-05,
      "loss": 1.5306,
      "step": 7814
    },
    {
      "epoch": 1.4077910380544922,
      "grad_norm": 1.296222448348999,
      "learning_rate": 2.0147539979029083e-05,
      "loss": 1.2255,
      "step": 7815
    },
    {
      "epoch": 1.4079711776626886,
      "grad_norm": 1.4993832111358643,
      "learning_rate": 2.0136180735964784e-05,
      "loss": 1.3738,
      "step": 7816
    },
    {
      "epoch": 1.4081513172708848,
      "grad_norm": 1.3295066356658936,
      "learning_rate": 2.0124823888561513e-05,
      "loss": 1.3389,
      "step": 7817
    },
    {
      "epoch": 1.4083314568790812,
      "grad_norm": 1.3483281135559082,
      "learning_rate": 2.0113469437730315e-05,
      "loss": 1.266,
      "step": 7818
    },
    {
      "epoch": 1.4085115964872776,
      "grad_norm": 1.450639009475708,
      "learning_rate": 2.0102117384382042e-05,
      "loss": 1.3115,
      "step": 7819
    },
    {
      "epoch": 1.408691736095474,
      "grad_norm": 1.4307678937911987,
      "learning_rate": 2.0090767729427336e-05,
      "loss": 1.2492,
      "step": 7820
    },
    {
      "epoch": 1.4088718757036705,
      "grad_norm": 1.321068286895752,
      "learning_rate": 2.007942047377669e-05,
      "loss": 1.1637,
      "step": 7821
    },
    {
      "epoch": 1.4090520153118666,
      "grad_norm": 1.4131584167480469,
      "learning_rate": 2.006807561834036e-05,
      "loss": 1.4678,
      "step": 7822
    },
    {
      "epoch": 1.409232154920063,
      "grad_norm": 1.5661684274673462,
      "learning_rate": 2.0056733164028422e-05,
      "loss": 1.4491,
      "step": 7823
    },
    {
      "epoch": 1.4094122945282594,
      "grad_norm": 1.3830509185791016,
      "learning_rate": 2.004539311175077e-05,
      "loss": 1.4107,
      "step": 7824
    },
    {
      "epoch": 1.4095924341364556,
      "grad_norm": 1.5468964576721191,
      "learning_rate": 2.003405546241709e-05,
      "loss": 1.6252,
      "step": 7825
    },
    {
      "epoch": 1.409772573744652,
      "grad_norm": 1.552641749382019,
      "learning_rate": 2.0022720216936886e-05,
      "loss": 1.6328,
      "step": 7826
    },
    {
      "epoch": 1.4099527133528484,
      "grad_norm": 1.480945110321045,
      "learning_rate": 2.0011387376219466e-05,
      "loss": 1.3122,
      "step": 7827
    },
    {
      "epoch": 1.4101328529610448,
      "grad_norm": 1.5036386251449585,
      "learning_rate": 2.000005694117394e-05,
      "loss": 1.6018,
      "step": 7828
    },
    {
      "epoch": 1.4103129925692413,
      "grad_norm": 1.361175537109375,
      "learning_rate": 1.9988728912709216e-05,
      "loss": 1.2247,
      "step": 7829
    },
    {
      "epoch": 1.4104931321774374,
      "grad_norm": 1.6424611806869507,
      "learning_rate": 1.997740329173406e-05,
      "loss": 1.6818,
      "step": 7830
    },
    {
      "epoch": 1.4106732717856338,
      "grad_norm": 1.5645921230316162,
      "learning_rate": 1.996608007915699e-05,
      "loss": 1.5336,
      "step": 7831
    },
    {
      "epoch": 1.4108534113938302,
      "grad_norm": 1.4245039224624634,
      "learning_rate": 1.995475927588634e-05,
      "loss": 1.3254,
      "step": 7832
    },
    {
      "epoch": 1.4110335510020267,
      "grad_norm": 1.4106066226959229,
      "learning_rate": 1.9943440882830273e-05,
      "loss": 1.4345,
      "step": 7833
    },
    {
      "epoch": 1.4112136906102228,
      "grad_norm": 1.5403735637664795,
      "learning_rate": 1.9932124900896722e-05,
      "loss": 1.5772,
      "step": 7834
    },
    {
      "epoch": 1.4113938302184192,
      "grad_norm": 1.509137749671936,
      "learning_rate": 1.9920811330993477e-05,
      "loss": 1.4109,
      "step": 7835
    },
    {
      "epoch": 1.4115739698266156,
      "grad_norm": 1.4068652391433716,
      "learning_rate": 1.99095001740281e-05,
      "loss": 1.2621,
      "step": 7836
    },
    {
      "epoch": 1.411754109434812,
      "grad_norm": 1.4332618713378906,
      "learning_rate": 1.9898191430907982e-05,
      "loss": 1.3474,
      "step": 7837
    },
    {
      "epoch": 1.4119342490430085,
      "grad_norm": 1.524052619934082,
      "learning_rate": 1.9886885102540264e-05,
      "loss": 1.329,
      "step": 7838
    },
    {
      "epoch": 1.4121143886512046,
      "grad_norm": 1.625367522239685,
      "learning_rate": 1.9875581189831943e-05,
      "loss": 1.3351,
      "step": 7839
    },
    {
      "epoch": 1.412294528259401,
      "grad_norm": 1.3909921646118164,
      "learning_rate": 1.986427969368984e-05,
      "loss": 1.2213,
      "step": 7840
    },
    {
      "epoch": 1.4124746678675975,
      "grad_norm": 1.5296363830566406,
      "learning_rate": 1.985298061502054e-05,
      "loss": 1.3934,
      "step": 7841
    },
    {
      "epoch": 1.4126548074757936,
      "grad_norm": 1.659833312034607,
      "learning_rate": 1.984168395473045e-05,
      "loss": 1.6676,
      "step": 7842
    },
    {
      "epoch": 1.41283494708399,
      "grad_norm": 1.5220714807510376,
      "learning_rate": 1.9830389713725773e-05,
      "loss": 1.3332,
      "step": 7843
    },
    {
      "epoch": 1.4130150866921865,
      "grad_norm": 1.4969030618667603,
      "learning_rate": 1.9819097892912548e-05,
      "loss": 1.3891,
      "step": 7844
    },
    {
      "epoch": 1.4131952263003829,
      "grad_norm": 1.4387528896331787,
      "learning_rate": 1.9807808493196594e-05,
      "loss": 1.3337,
      "step": 7845
    },
    {
      "epoch": 1.4133753659085793,
      "grad_norm": 1.5578863620758057,
      "learning_rate": 1.9796521515483533e-05,
      "loss": 1.4219,
      "step": 7846
    },
    {
      "epoch": 1.4135555055167754,
      "grad_norm": 1.5699931383132935,
      "learning_rate": 1.978523696067881e-05,
      "loss": 1.4139,
      "step": 7847
    },
    {
      "epoch": 1.4137356451249719,
      "grad_norm": 1.4896538257598877,
      "learning_rate": 1.977395482968766e-05,
      "loss": 1.3393,
      "step": 7848
    },
    {
      "epoch": 1.4139157847331683,
      "grad_norm": 1.4160312414169312,
      "learning_rate": 1.9762675123415124e-05,
      "loss": 1.0709,
      "step": 7849
    },
    {
      "epoch": 1.4140959243413644,
      "grad_norm": 1.5764027833938599,
      "learning_rate": 1.9751397842766067e-05,
      "loss": 1.2805,
      "step": 7850
    },
    {
      "epoch": 1.4142760639495608,
      "grad_norm": 1.6334166526794434,
      "learning_rate": 1.9740122988645138e-05,
      "loss": 1.2337,
      "step": 7851
    },
    {
      "epoch": 1.4144562035577573,
      "grad_norm": 1.3740922212600708,
      "learning_rate": 1.972885056195681e-05,
      "loss": 1.8473,
      "step": 7852
    },
    {
      "epoch": 1.4146363431659537,
      "grad_norm": 1.3389655351638794,
      "learning_rate": 1.971758056360532e-05,
      "loss": 1.9542,
      "step": 7853
    },
    {
      "epoch": 1.41481648277415,
      "grad_norm": 1.3268479108810425,
      "learning_rate": 1.970631299449479e-05,
      "loss": 1.7906,
      "step": 7854
    },
    {
      "epoch": 1.4149966223823462,
      "grad_norm": 1.448621392250061,
      "learning_rate": 1.9695047855529076e-05,
      "loss": 1.7253,
      "step": 7855
    },
    {
      "epoch": 1.4151767619905427,
      "grad_norm": 1.3105289936065674,
      "learning_rate": 1.968378514761186e-05,
      "loss": 1.6009,
      "step": 7856
    },
    {
      "epoch": 1.415356901598739,
      "grad_norm": 1.4124598503112793,
      "learning_rate": 1.967252487164663e-05,
      "loss": 1.6538,
      "step": 7857
    },
    {
      "epoch": 1.4155370412069352,
      "grad_norm": 1.8643112182617188,
      "learning_rate": 1.966126702853669e-05,
      "loss": 1.8717,
      "step": 7858
    },
    {
      "epoch": 1.4157171808151316,
      "grad_norm": 1.5061299800872803,
      "learning_rate": 1.9650011619185126e-05,
      "loss": 1.7982,
      "step": 7859
    },
    {
      "epoch": 1.415897320423328,
      "grad_norm": 1.7730798721313477,
      "learning_rate": 1.9638758644494848e-05,
      "loss": 1.9192,
      "step": 7860
    },
    {
      "epoch": 1.4160774600315245,
      "grad_norm": 1.5577809810638428,
      "learning_rate": 1.9627508105368564e-05,
      "loss": 1.5209,
      "step": 7861
    },
    {
      "epoch": 1.4162575996397209,
      "grad_norm": 1.4238172769546509,
      "learning_rate": 1.961626000270877e-05,
      "loss": 1.5473,
      "step": 7862
    },
    {
      "epoch": 1.4164377392479173,
      "grad_norm": 1.4410889148712158,
      "learning_rate": 1.9605014337417815e-05,
      "loss": 1.512,
      "step": 7863
    },
    {
      "epoch": 1.4166178788561135,
      "grad_norm": 1.244889736175537,
      "learning_rate": 1.9593771110397808e-05,
      "loss": 1.3998,
      "step": 7864
    },
    {
      "epoch": 1.4167980184643099,
      "grad_norm": 1.4741644859313965,
      "learning_rate": 1.9582530322550668e-05,
      "loss": 1.8002,
      "step": 7865
    },
    {
      "epoch": 1.4169781580725063,
      "grad_norm": 1.4130538702011108,
      "learning_rate": 1.9571291974778132e-05,
      "loss": 1.3553,
      "step": 7866
    },
    {
      "epoch": 1.4171582976807025,
      "grad_norm": 1.3103638887405396,
      "learning_rate": 1.9560056067981715e-05,
      "loss": 1.3527,
      "step": 7867
    },
    {
      "epoch": 1.4173384372888989,
      "grad_norm": 1.4565075635910034,
      "learning_rate": 1.9548822603062788e-05,
      "loss": 1.4666,
      "step": 7868
    },
    {
      "epoch": 1.4175185768970953,
      "grad_norm": 1.4128637313842773,
      "learning_rate": 1.95375915809225e-05,
      "loss": 1.3565,
      "step": 7869
    },
    {
      "epoch": 1.4176987165052917,
      "grad_norm": 1.4785748720169067,
      "learning_rate": 1.9526363002461758e-05,
      "loss": 1.3128,
      "step": 7870
    },
    {
      "epoch": 1.417878856113488,
      "grad_norm": 1.4779640436172485,
      "learning_rate": 1.9515136868581317e-05,
      "loss": 1.4045,
      "step": 7871
    },
    {
      "epoch": 1.4180589957216843,
      "grad_norm": 1.352829098701477,
      "learning_rate": 1.9503913180181767e-05,
      "loss": 1.2457,
      "step": 7872
    },
    {
      "epoch": 1.4182391353298807,
      "grad_norm": 1.3769781589508057,
      "learning_rate": 1.9492691938163443e-05,
      "loss": 1.2114,
      "step": 7873
    },
    {
      "epoch": 1.418419274938077,
      "grad_norm": 1.3709018230438232,
      "learning_rate": 1.9481473143426516e-05,
      "loss": 1.3453,
      "step": 7874
    },
    {
      "epoch": 1.4185994145462733,
      "grad_norm": 1.4437788724899292,
      "learning_rate": 1.9470256796870947e-05,
      "loss": 1.3438,
      "step": 7875
    },
    {
      "epoch": 1.4187795541544697,
      "grad_norm": 1.4368840456008911,
      "learning_rate": 1.9459042899396486e-05,
      "loss": 1.6159,
      "step": 7876
    },
    {
      "epoch": 1.418959693762666,
      "grad_norm": 1.4402023553848267,
      "learning_rate": 1.944783145190275e-05,
      "loss": 1.2944,
      "step": 7877
    },
    {
      "epoch": 1.4191398333708625,
      "grad_norm": 1.479618787765503,
      "learning_rate": 1.9436622455289093e-05,
      "loss": 1.2564,
      "step": 7878
    },
    {
      "epoch": 1.4193199729790589,
      "grad_norm": 1.4730867147445679,
      "learning_rate": 1.942541591045469e-05,
      "loss": 1.388,
      "step": 7879
    },
    {
      "epoch": 1.419500112587255,
      "grad_norm": 1.4355342388153076,
      "learning_rate": 1.941421181829854e-05,
      "loss": 1.4469,
      "step": 7880
    },
    {
      "epoch": 1.4196802521954515,
      "grad_norm": 1.4885388612747192,
      "learning_rate": 1.9403010179719417e-05,
      "loss": 1.2546,
      "step": 7881
    },
    {
      "epoch": 1.4198603918036479,
      "grad_norm": 1.3629711866378784,
      "learning_rate": 1.939181099561591e-05,
      "loss": 1.2152,
      "step": 7882
    },
    {
      "epoch": 1.420040531411844,
      "grad_norm": 1.5354400873184204,
      "learning_rate": 1.938061426688642e-05,
      "loss": 1.6531,
      "step": 7883
    },
    {
      "epoch": 1.4202206710200405,
      "grad_norm": 1.5731403827667236,
      "learning_rate": 1.9369419994429138e-05,
      "loss": 1.4817,
      "step": 7884
    },
    {
      "epoch": 1.4204008106282369,
      "grad_norm": 1.4022859334945679,
      "learning_rate": 1.9358228179142048e-05,
      "loss": 1.3337,
      "step": 7885
    },
    {
      "epoch": 1.4205809502364333,
      "grad_norm": 1.6270155906677246,
      "learning_rate": 1.9347038821922984e-05,
      "loss": 1.6178,
      "step": 7886
    },
    {
      "epoch": 1.4207610898446297,
      "grad_norm": 1.361730933189392,
      "learning_rate": 1.9335851923669536e-05,
      "loss": 1.2329,
      "step": 7887
    },
    {
      "epoch": 1.4209412294528259,
      "grad_norm": 1.5242886543273926,
      "learning_rate": 1.932466748527911e-05,
      "loss": 1.452,
      "step": 7888
    },
    {
      "epoch": 1.4211213690610223,
      "grad_norm": 1.4447822570800781,
      "learning_rate": 1.9313485507648922e-05,
      "loss": 1.5588,
      "step": 7889
    },
    {
      "epoch": 1.4213015086692187,
      "grad_norm": 1.4923052787780762,
      "learning_rate": 1.9302305991675973e-05,
      "loss": 1.5019,
      "step": 7890
    },
    {
      "epoch": 1.421481648277415,
      "grad_norm": 1.5655585527420044,
      "learning_rate": 1.9291128938257087e-05,
      "loss": 1.378,
      "step": 7891
    },
    {
      "epoch": 1.4216617878856113,
      "grad_norm": 1.5809687376022339,
      "learning_rate": 1.927995434828888e-05,
      "loss": 1.5549,
      "step": 7892
    },
    {
      "epoch": 1.4218419274938077,
      "grad_norm": 1.546939492225647,
      "learning_rate": 1.926878222266777e-05,
      "loss": 1.4146,
      "step": 7893
    },
    {
      "epoch": 1.422022067102004,
      "grad_norm": 1.3812103271484375,
      "learning_rate": 1.9257612562289967e-05,
      "loss": 1.231,
      "step": 7894
    },
    {
      "epoch": 1.4222022067102005,
      "grad_norm": 1.5994952917099,
      "learning_rate": 1.924644536805153e-05,
      "loss": 1.4235,
      "step": 7895
    },
    {
      "epoch": 1.4223823463183969,
      "grad_norm": 1.6997147798538208,
      "learning_rate": 1.9235280640848262e-05,
      "loss": 1.5143,
      "step": 7896
    },
    {
      "epoch": 1.422562485926593,
      "grad_norm": 1.5993354320526123,
      "learning_rate": 1.9224118381575797e-05,
      "loss": 1.1977,
      "step": 7897
    },
    {
      "epoch": 1.4227426255347895,
      "grad_norm": 1.4947186708450317,
      "learning_rate": 1.9212958591129565e-05,
      "loss": 1.3158,
      "step": 7898
    },
    {
      "epoch": 1.4229227651429859,
      "grad_norm": 1.6370208263397217,
      "learning_rate": 1.9201801270404778e-05,
      "loss": 1.398,
      "step": 7899
    },
    {
      "epoch": 1.423102904751182,
      "grad_norm": 1.4944663047790527,
      "learning_rate": 1.9190646420296517e-05,
      "loss": 1.216,
      "step": 7900
    },
    {
      "epoch": 1.4232830443593785,
      "grad_norm": 1.5353169441223145,
      "learning_rate": 1.9179494041699586e-05,
      "loss": 1.4229,
      "step": 7901
    },
    {
      "epoch": 1.4234631839675749,
      "grad_norm": 1.3661359548568726,
      "learning_rate": 1.9168344135508654e-05,
      "loss": 1.7543,
      "step": 7902
    },
    {
      "epoch": 1.4236433235757713,
      "grad_norm": 1.4645752906799316,
      "learning_rate": 1.915719670261812e-05,
      "loss": 1.8437,
      "step": 7903
    },
    {
      "epoch": 1.4238234631839677,
      "grad_norm": 1.3700424432754517,
      "learning_rate": 1.9146051743922223e-05,
      "loss": 1.8131,
      "step": 7904
    },
    {
      "epoch": 1.4240036027921639,
      "grad_norm": 1.4430216550827026,
      "learning_rate": 1.9134909260315044e-05,
      "loss": 1.6698,
      "step": 7905
    },
    {
      "epoch": 1.4241837424003603,
      "grad_norm": 1.344657063484192,
      "learning_rate": 1.912376925269041e-05,
      "loss": 1.5202,
      "step": 7906
    },
    {
      "epoch": 1.4243638820085567,
      "grad_norm": 1.4975886344909668,
      "learning_rate": 1.9112631721941972e-05,
      "loss": 1.9506,
      "step": 7907
    },
    {
      "epoch": 1.4245440216167529,
      "grad_norm": 1.504418969154358,
      "learning_rate": 1.9101496668963148e-05,
      "loss": 1.8791,
      "step": 7908
    },
    {
      "epoch": 1.4247241612249493,
      "grad_norm": 1.7046763896942139,
      "learning_rate": 1.9090364094647235e-05,
      "loss": 2.0661,
      "step": 7909
    },
    {
      "epoch": 1.4249043008331457,
      "grad_norm": 1.9011411666870117,
      "learning_rate": 1.907923399988725e-05,
      "loss": 2.2164,
      "step": 7910
    },
    {
      "epoch": 1.425084440441342,
      "grad_norm": 1.4955675601959229,
      "learning_rate": 1.9068106385576063e-05,
      "loss": 1.4977,
      "step": 7911
    },
    {
      "epoch": 1.4252645800495385,
      "grad_norm": 1.605182409286499,
      "learning_rate": 1.905698125260631e-05,
      "loss": 1.6462,
      "step": 7912
    },
    {
      "epoch": 1.4254447196577347,
      "grad_norm": 1.4227882623672485,
      "learning_rate": 1.9045858601870454e-05,
      "loss": 1.6901,
      "step": 7913
    },
    {
      "epoch": 1.425624859265931,
      "grad_norm": 1.2416832447052002,
      "learning_rate": 1.903473843426074e-05,
      "loss": 1.2759,
      "step": 7914
    },
    {
      "epoch": 1.4258049988741275,
      "grad_norm": 1.4357246160507202,
      "learning_rate": 1.9023620750669224e-05,
      "loss": 1.3224,
      "step": 7915
    },
    {
      "epoch": 1.4259851384823237,
      "grad_norm": 1.2815290689468384,
      "learning_rate": 1.9012505551987765e-05,
      "loss": 1.3066,
      "step": 7916
    },
    {
      "epoch": 1.42616527809052,
      "grad_norm": 1.4816241264343262,
      "learning_rate": 1.9001392839108018e-05,
      "loss": 1.4395,
      "step": 7917
    },
    {
      "epoch": 1.4263454176987165,
      "grad_norm": 1.3726794719696045,
      "learning_rate": 1.8990282612921424e-05,
      "loss": 1.3944,
      "step": 7918
    },
    {
      "epoch": 1.4265255573069129,
      "grad_norm": 1.4431177377700806,
      "learning_rate": 1.8979174874319265e-05,
      "loss": 1.4747,
      "step": 7919
    },
    {
      "epoch": 1.4267056969151093,
      "grad_norm": 1.4041320085525513,
      "learning_rate": 1.8968069624192593e-05,
      "loss": 1.3927,
      "step": 7920
    },
    {
      "epoch": 1.4268858365233057,
      "grad_norm": 1.5095077753067017,
      "learning_rate": 1.8956966863432256e-05,
      "loss": 1.5318,
      "step": 7921
    },
    {
      "epoch": 1.4270659761315019,
      "grad_norm": 1.327257513999939,
      "learning_rate": 1.8945866592928922e-05,
      "loss": 1.3714,
      "step": 7922
    },
    {
      "epoch": 1.4272461157396983,
      "grad_norm": 1.2723485231399536,
      "learning_rate": 1.8934768813573045e-05,
      "loss": 1.2151,
      "step": 7923
    },
    {
      "epoch": 1.4274262553478947,
      "grad_norm": 1.5590929985046387,
      "learning_rate": 1.8923673526254874e-05,
      "loss": 1.5145,
      "step": 7924
    },
    {
      "epoch": 1.4276063949560909,
      "grad_norm": 1.3956323862075806,
      "learning_rate": 1.8912580731864487e-05,
      "loss": 1.3634,
      "step": 7925
    },
    {
      "epoch": 1.4277865345642873,
      "grad_norm": 1.4058585166931152,
      "learning_rate": 1.8901490431291726e-05,
      "loss": 1.4765,
      "step": 7926
    },
    {
      "epoch": 1.4279666741724837,
      "grad_norm": 1.505709171295166,
      "learning_rate": 1.8890402625426235e-05,
      "loss": 1.4271,
      "step": 7927
    },
    {
      "epoch": 1.42814681378068,
      "grad_norm": 1.4091767072677612,
      "learning_rate": 1.8879317315157514e-05,
      "loss": 1.1701,
      "step": 7928
    },
    {
      "epoch": 1.4283269533888765,
      "grad_norm": 1.4743874073028564,
      "learning_rate": 1.88682345013748e-05,
      "loss": 1.3883,
      "step": 7929
    },
    {
      "epoch": 1.4285070929970727,
      "grad_norm": 1.495491862297058,
      "learning_rate": 1.8857154184967148e-05,
      "loss": 1.5767,
      "step": 7930
    },
    {
      "epoch": 1.428687232605269,
      "grad_norm": 1.622897744178772,
      "learning_rate": 1.8846076366823417e-05,
      "loss": 1.5282,
      "step": 7931
    },
    {
      "epoch": 1.4288673722134655,
      "grad_norm": 1.4698102474212646,
      "learning_rate": 1.883500104783225e-05,
      "loss": 1.2853,
      "step": 7932
    },
    {
      "epoch": 1.4290475118216617,
      "grad_norm": 1.4751147031784058,
      "learning_rate": 1.8823928228882137e-05,
      "loss": 1.3005,
      "step": 7933
    },
    {
      "epoch": 1.429227651429858,
      "grad_norm": 1.612786054611206,
      "learning_rate": 1.881285791086133e-05,
      "loss": 1.4869,
      "step": 7934
    },
    {
      "epoch": 1.4294077910380545,
      "grad_norm": 1.468007206916809,
      "learning_rate": 1.8801790094657856e-05,
      "loss": 1.3834,
      "step": 7935
    },
    {
      "epoch": 1.429587930646251,
      "grad_norm": 1.5922496318817139,
      "learning_rate": 1.879072478115957e-05,
      "loss": 1.5296,
      "step": 7936
    },
    {
      "epoch": 1.4297680702544473,
      "grad_norm": 1.5834506750106812,
      "learning_rate": 1.8779661971254153e-05,
      "loss": 1.4453,
      "step": 7937
    },
    {
      "epoch": 1.4299482098626435,
      "grad_norm": 1.535792350769043,
      "learning_rate": 1.876860166582905e-05,
      "loss": 1.1589,
      "step": 7938
    },
    {
      "epoch": 1.4301283494708399,
      "grad_norm": 1.5236412286758423,
      "learning_rate": 1.8757543865771504e-05,
      "loss": 1.2876,
      "step": 7939
    },
    {
      "epoch": 1.4303084890790363,
      "grad_norm": 1.5519678592681885,
      "learning_rate": 1.8746488571968574e-05,
      "loss": 1.4189,
      "step": 7940
    },
    {
      "epoch": 1.4304886286872325,
      "grad_norm": 1.5685690641403198,
      "learning_rate": 1.8735435785307094e-05,
      "loss": 1.4846,
      "step": 7941
    },
    {
      "epoch": 1.4306687682954289,
      "grad_norm": 1.5382126569747925,
      "learning_rate": 1.8724385506673735e-05,
      "loss": 1.5148,
      "step": 7942
    },
    {
      "epoch": 1.4308489079036253,
      "grad_norm": 1.5386821031570435,
      "learning_rate": 1.8713337736954945e-05,
      "loss": 1.3786,
      "step": 7943
    },
    {
      "epoch": 1.4310290475118217,
      "grad_norm": 1.5386455059051514,
      "learning_rate": 1.870229247703696e-05,
      "loss": 1.2948,
      "step": 7944
    },
    {
      "epoch": 1.431209187120018,
      "grad_norm": 1.5205636024475098,
      "learning_rate": 1.8691249727805832e-05,
      "loss": 1.3204,
      "step": 7945
    },
    {
      "epoch": 1.4313893267282143,
      "grad_norm": 1.6280258893966675,
      "learning_rate": 1.8680209490147395e-05,
      "loss": 1.3852,
      "step": 7946
    },
    {
      "epoch": 1.4315694663364107,
      "grad_norm": 1.5553555488586426,
      "learning_rate": 1.8669171764947297e-05,
      "loss": 1.4096,
      "step": 7947
    },
    {
      "epoch": 1.431749605944607,
      "grad_norm": 1.4723114967346191,
      "learning_rate": 1.865813655309098e-05,
      "loss": 1.2935,
      "step": 7948
    },
    {
      "epoch": 1.4319297455528035,
      "grad_norm": 1.5613638162612915,
      "learning_rate": 1.8647103855463688e-05,
      "loss": 1.3377,
      "step": 7949
    },
    {
      "epoch": 1.4321098851609997,
      "grad_norm": 1.4356449842453003,
      "learning_rate": 1.8636073672950433e-05,
      "loss": 1.38,
      "step": 7950
    },
    {
      "epoch": 1.432290024769196,
      "grad_norm": 1.3043460845947266,
      "learning_rate": 1.8625046006436086e-05,
      "loss": 1.018,
      "step": 7951
    },
    {
      "epoch": 1.4324701643773925,
      "grad_norm": 1.4227840900421143,
      "learning_rate": 1.861402085680527e-05,
      "loss": 1.9708,
      "step": 7952
    },
    {
      "epoch": 1.432650303985589,
      "grad_norm": 1.3337846994400024,
      "learning_rate": 1.860299822494241e-05,
      "loss": 1.9448,
      "step": 7953
    },
    {
      "epoch": 1.4328304435937853,
      "grad_norm": 1.4255822896957397,
      "learning_rate": 1.859197811173174e-05,
      "loss": 1.7514,
      "step": 7954
    },
    {
      "epoch": 1.4330105832019815,
      "grad_norm": 1.3270381689071655,
      "learning_rate": 1.8580960518057288e-05,
      "loss": 1.8815,
      "step": 7955
    },
    {
      "epoch": 1.433190722810178,
      "grad_norm": 1.4608460664749146,
      "learning_rate": 1.8569945444802882e-05,
      "loss": 1.8592,
      "step": 7956
    },
    {
      "epoch": 1.4333708624183743,
      "grad_norm": 1.541144609451294,
      "learning_rate": 1.8558932892852145e-05,
      "loss": 1.7285,
      "step": 7957
    },
    {
      "epoch": 1.4335510020265705,
      "grad_norm": 1.4751203060150146,
      "learning_rate": 1.854792286308849e-05,
      "loss": 1.6398,
      "step": 7958
    },
    {
      "epoch": 1.433731141634767,
      "grad_norm": 1.4892191886901855,
      "learning_rate": 1.8536915356395147e-05,
      "loss": 1.7831,
      "step": 7959
    },
    {
      "epoch": 1.4339112812429633,
      "grad_norm": 1.4279849529266357,
      "learning_rate": 1.852591037365511e-05,
      "loss": 1.5828,
      "step": 7960
    },
    {
      "epoch": 1.4340914208511597,
      "grad_norm": 1.8383522033691406,
      "learning_rate": 1.851490791575123e-05,
      "loss": 2.109,
      "step": 7961
    },
    {
      "epoch": 1.434271560459356,
      "grad_norm": 1.5862197875976562,
      "learning_rate": 1.8503907983566103e-05,
      "loss": 1.5968,
      "step": 7962
    },
    {
      "epoch": 1.4344517000675523,
      "grad_norm": 1.5333952903747559,
      "learning_rate": 1.8492910577982138e-05,
      "loss": 1.7282,
      "step": 7963
    },
    {
      "epoch": 1.4346318396757487,
      "grad_norm": 1.3018234968185425,
      "learning_rate": 1.8481915699881513e-05,
      "loss": 1.32,
      "step": 7964
    },
    {
      "epoch": 1.434811979283945,
      "grad_norm": 1.5543546676635742,
      "learning_rate": 1.8470923350146286e-05,
      "loss": 1.798,
      "step": 7965
    },
    {
      "epoch": 1.4349921188921413,
      "grad_norm": 1.4211477041244507,
      "learning_rate": 1.845993352965823e-05,
      "loss": 1.5113,
      "step": 7966
    },
    {
      "epoch": 1.4351722585003377,
      "grad_norm": 1.4054722785949707,
      "learning_rate": 1.8448946239298952e-05,
      "loss": 1.6904,
      "step": 7967
    },
    {
      "epoch": 1.435352398108534,
      "grad_norm": 1.287160038948059,
      "learning_rate": 1.8437961479949832e-05,
      "loss": 1.2892,
      "step": 7968
    },
    {
      "epoch": 1.4355325377167305,
      "grad_norm": 1.432945966720581,
      "learning_rate": 1.8426979252492045e-05,
      "loss": 1.3424,
      "step": 7969
    },
    {
      "epoch": 1.435712677324927,
      "grad_norm": 1.5032153129577637,
      "learning_rate": 1.8415999557806624e-05,
      "loss": 1.5446,
      "step": 7970
    },
    {
      "epoch": 1.435892816933123,
      "grad_norm": 1.3255146741867065,
      "learning_rate": 1.8405022396774337e-05,
      "loss": 1.2592,
      "step": 7971
    },
    {
      "epoch": 1.4360729565413195,
      "grad_norm": 1.4102063179016113,
      "learning_rate": 1.839404777027576e-05,
      "loss": 1.367,
      "step": 7972
    },
    {
      "epoch": 1.436253096149516,
      "grad_norm": 1.4789421558380127,
      "learning_rate": 1.8383075679191273e-05,
      "loss": 1.4415,
      "step": 7973
    },
    {
      "epoch": 1.436433235757712,
      "grad_norm": 1.4706363677978516,
      "learning_rate": 1.8372106124401045e-05,
      "loss": 1.4336,
      "step": 7974
    },
    {
      "epoch": 1.4366133753659085,
      "grad_norm": 1.3293424844741821,
      "learning_rate": 1.836113910678507e-05,
      "loss": 1.327,
      "step": 7975
    },
    {
      "epoch": 1.436793514974105,
      "grad_norm": 1.2669920921325684,
      "learning_rate": 1.8350174627223105e-05,
      "loss": 1.1628,
      "step": 7976
    },
    {
      "epoch": 1.4369736545823013,
      "grad_norm": 1.499607801437378,
      "learning_rate": 1.833921268659472e-05,
      "loss": 1.5666,
      "step": 7977
    },
    {
      "epoch": 1.4371537941904977,
      "grad_norm": 1.4095861911773682,
      "learning_rate": 1.8328253285779267e-05,
      "loss": 1.3281,
      "step": 7978
    },
    {
      "epoch": 1.4373339337986941,
      "grad_norm": 1.3976359367370605,
      "learning_rate": 1.831729642565591e-05,
      "loss": 1.377,
      "step": 7979
    },
    {
      "epoch": 1.4375140734068903,
      "grad_norm": 1.5039989948272705,
      "learning_rate": 1.830634210710361e-05,
      "loss": 1.5736,
      "step": 7980
    },
    {
      "epoch": 1.4376942130150867,
      "grad_norm": 1.4818403720855713,
      "learning_rate": 1.8295390331001096e-05,
      "loss": 1.487,
      "step": 7981
    },
    {
      "epoch": 1.4378743526232831,
      "grad_norm": 1.472092866897583,
      "learning_rate": 1.8284441098226934e-05,
      "loss": 1.5115,
      "step": 7982
    },
    {
      "epoch": 1.4380544922314793,
      "grad_norm": 1.629014253616333,
      "learning_rate": 1.8273494409659435e-05,
      "loss": 1.583,
      "step": 7983
    },
    {
      "epoch": 1.4382346318396757,
      "grad_norm": 1.5817917585372925,
      "learning_rate": 1.8262550266176787e-05,
      "loss": 1.3209,
      "step": 7984
    },
    {
      "epoch": 1.438414771447872,
      "grad_norm": 1.271173357963562,
      "learning_rate": 1.8251608668656894e-05,
      "loss": 1.1632,
      "step": 7985
    },
    {
      "epoch": 1.4385949110560685,
      "grad_norm": 1.4208099842071533,
      "learning_rate": 1.824066961797749e-05,
      "loss": 1.5138,
      "step": 7986
    },
    {
      "epoch": 1.438775050664265,
      "grad_norm": 1.5053945779800415,
      "learning_rate": 1.82297331150161e-05,
      "loss": 1.4114,
      "step": 7987
    },
    {
      "epoch": 1.438955190272461,
      "grad_norm": 1.5340497493743896,
      "learning_rate": 1.8218799160650047e-05,
      "loss": 1.4193,
      "step": 7988
    },
    {
      "epoch": 1.4391353298806575,
      "grad_norm": 1.4334325790405273,
      "learning_rate": 1.8207867755756442e-05,
      "loss": 1.4446,
      "step": 7989
    },
    {
      "epoch": 1.439315469488854,
      "grad_norm": 1.4936779737472534,
      "learning_rate": 1.8196938901212196e-05,
      "loss": 1.3615,
      "step": 7990
    },
    {
      "epoch": 1.43949560909705,
      "grad_norm": 1.3821886777877808,
      "learning_rate": 1.8186012597894026e-05,
      "loss": 1.1397,
      "step": 7991
    },
    {
      "epoch": 1.4396757487052465,
      "grad_norm": 1.6085041761398315,
      "learning_rate": 1.8175088846678412e-05,
      "loss": 1.3564,
      "step": 7992
    },
    {
      "epoch": 1.439855888313443,
      "grad_norm": 1.6198670864105225,
      "learning_rate": 1.8164167648441686e-05,
      "loss": 1.536,
      "step": 7993
    },
    {
      "epoch": 1.4400360279216393,
      "grad_norm": 1.6322762966156006,
      "learning_rate": 1.8153249004059925e-05,
      "loss": 1.454,
      "step": 7994
    },
    {
      "epoch": 1.4402161675298357,
      "grad_norm": 1.3694039583206177,
      "learning_rate": 1.8142332914409016e-05,
      "loss": 1.2246,
      "step": 7995
    },
    {
      "epoch": 1.440396307138032,
      "grad_norm": 1.5592377185821533,
      "learning_rate": 1.8131419380364644e-05,
      "loss": 1.352,
      "step": 7996
    },
    {
      "epoch": 1.4405764467462283,
      "grad_norm": 1.4743783473968506,
      "learning_rate": 1.812050840280227e-05,
      "loss": 1.3983,
      "step": 7997
    },
    {
      "epoch": 1.4407565863544247,
      "grad_norm": 1.5487103462219238,
      "learning_rate": 1.8109599982597203e-05,
      "loss": 1.2405,
      "step": 7998
    },
    {
      "epoch": 1.440936725962621,
      "grad_norm": 1.5841177701950073,
      "learning_rate": 1.8098694120624505e-05,
      "loss": 1.2875,
      "step": 7999
    },
    {
      "epoch": 1.4411168655708173,
      "grad_norm": 1.50828218460083,
      "learning_rate": 1.808779081775901e-05,
      "loss": 1.3283,
      "step": 8000
    },
    {
      "epoch": 1.4412970051790137,
      "grad_norm": 1.606026291847229,
      "learning_rate": 1.8076890074875374e-05,
      "loss": 1.3342,
      "step": 8001
    },
    {
      "epoch": 1.4414771447872101,
      "grad_norm": 1.2386553287506104,
      "learning_rate": 1.806599189284809e-05,
      "loss": 1.5013,
      "step": 8002
    },
    {
      "epoch": 1.4416572843954065,
      "grad_norm": 1.337441086769104,
      "learning_rate": 1.8055096272551375e-05,
      "loss": 1.7192,
      "step": 8003
    },
    {
      "epoch": 1.441837424003603,
      "grad_norm": 1.2560787200927734,
      "learning_rate": 1.804420321485928e-05,
      "loss": 1.5599,
      "step": 8004
    },
    {
      "epoch": 1.4420175636117991,
      "grad_norm": 1.3062278032302856,
      "learning_rate": 1.8033312720645635e-05,
      "loss": 1.6616,
      "step": 8005
    },
    {
      "epoch": 1.4421977032199955,
      "grad_norm": 1.3483303785324097,
      "learning_rate": 1.802242479078406e-05,
      "loss": 1.6165,
      "step": 8006
    },
    {
      "epoch": 1.442377842828192,
      "grad_norm": 1.520615577697754,
      "learning_rate": 1.801153942614801e-05,
      "loss": 1.7802,
      "step": 8007
    },
    {
      "epoch": 1.442557982436388,
      "grad_norm": 1.624891996383667,
      "learning_rate": 1.8000656627610683e-05,
      "loss": 1.9756,
      "step": 8008
    },
    {
      "epoch": 1.4427381220445845,
      "grad_norm": 1.5343197584152222,
      "learning_rate": 1.79897763960451e-05,
      "loss": 1.7663,
      "step": 8009
    },
    {
      "epoch": 1.442918261652781,
      "grad_norm": 1.9937756061553955,
      "learning_rate": 1.7978898732324055e-05,
      "loss": 1.8343,
      "step": 8010
    },
    {
      "epoch": 1.4430984012609773,
      "grad_norm": 1.3978335857391357,
      "learning_rate": 1.7968023637320164e-05,
      "loss": 1.3927,
      "step": 8011
    },
    {
      "epoch": 1.4432785408691737,
      "grad_norm": 1.4265486001968384,
      "learning_rate": 1.7957151111905813e-05,
      "loss": 1.36,
      "step": 8012
    },
    {
      "epoch": 1.44345868047737,
      "grad_norm": 1.3184529542922974,
      "learning_rate": 1.794628115695319e-05,
      "loss": 1.4875,
      "step": 8013
    },
    {
      "epoch": 1.4436388200855663,
      "grad_norm": 1.3715662956237793,
      "learning_rate": 1.793541377333428e-05,
      "loss": 1.3782,
      "step": 8014
    },
    {
      "epoch": 1.4438189596937627,
      "grad_norm": 1.4320085048675537,
      "learning_rate": 1.792454896192084e-05,
      "loss": 1.3684,
      "step": 8015
    },
    {
      "epoch": 1.443999099301959,
      "grad_norm": 1.4699143171310425,
      "learning_rate": 1.7913686723584478e-05,
      "loss": 1.4742,
      "step": 8016
    },
    {
      "epoch": 1.4441792389101553,
      "grad_norm": 1.3905134201049805,
      "learning_rate": 1.7902827059196542e-05,
      "loss": 1.572,
      "step": 8017
    },
    {
      "epoch": 1.4443593785183517,
      "grad_norm": 1.3693205118179321,
      "learning_rate": 1.7891969969628182e-05,
      "loss": 1.2682,
      "step": 8018
    },
    {
      "epoch": 1.4445395181265481,
      "grad_norm": 1.5497798919677734,
      "learning_rate": 1.788111545575035e-05,
      "loss": 1.5367,
      "step": 8019
    },
    {
      "epoch": 1.4447196577347445,
      "grad_norm": 1.3003859519958496,
      "learning_rate": 1.7870263518433793e-05,
      "loss": 1.1805,
      "step": 8020
    },
    {
      "epoch": 1.4448997973429407,
      "grad_norm": 1.261290431022644,
      "learning_rate": 1.785941415854905e-05,
      "loss": 1.2311,
      "step": 8021
    },
    {
      "epoch": 1.4450799369511371,
      "grad_norm": 1.5115928649902344,
      "learning_rate": 1.7848567376966447e-05,
      "loss": 1.6328,
      "step": 8022
    },
    {
      "epoch": 1.4452600765593335,
      "grad_norm": 1.4348713159561157,
      "learning_rate": 1.7837723174556102e-05,
      "loss": 1.3894,
      "step": 8023
    },
    {
      "epoch": 1.4454402161675297,
      "grad_norm": 1.4605377912521362,
      "learning_rate": 1.7826881552187946e-05,
      "loss": 1.4607,
      "step": 8024
    },
    {
      "epoch": 1.4456203557757261,
      "grad_norm": 1.4186657667160034,
      "learning_rate": 1.7816042510731663e-05,
      "loss": 1.2278,
      "step": 8025
    },
    {
      "epoch": 1.4458004953839225,
      "grad_norm": 1.4589608907699585,
      "learning_rate": 1.7805206051056788e-05,
      "loss": 1.2834,
      "step": 8026
    },
    {
      "epoch": 1.445980634992119,
      "grad_norm": 1.4106963872909546,
      "learning_rate": 1.7794372174032603e-05,
      "loss": 1.3708,
      "step": 8027
    },
    {
      "epoch": 1.4461607746003153,
      "grad_norm": 1.5547178983688354,
      "learning_rate": 1.7783540880528194e-05,
      "loss": 1.4757,
      "step": 8028
    },
    {
      "epoch": 1.4463409142085115,
      "grad_norm": 1.4539644718170166,
      "learning_rate": 1.7772712171412424e-05,
      "loss": 1.2575,
      "step": 8029
    },
    {
      "epoch": 1.446521053816708,
      "grad_norm": 1.3594244718551636,
      "learning_rate": 1.776188604755401e-05,
      "loss": 1.3355,
      "step": 8030
    },
    {
      "epoch": 1.4467011934249043,
      "grad_norm": 1.402726173400879,
      "learning_rate": 1.7751062509821392e-05,
      "loss": 1.4013,
      "step": 8031
    },
    {
      "epoch": 1.4468813330331007,
      "grad_norm": 1.6147027015686035,
      "learning_rate": 1.774024155908285e-05,
      "loss": 1.4934,
      "step": 8032
    },
    {
      "epoch": 1.447061472641297,
      "grad_norm": 1.512836217880249,
      "learning_rate": 1.7729423196206396e-05,
      "loss": 1.4415,
      "step": 8033
    },
    {
      "epoch": 1.4472416122494933,
      "grad_norm": 1.3208867311477661,
      "learning_rate": 1.771860742205988e-05,
      "loss": 1.2731,
      "step": 8034
    },
    {
      "epoch": 1.4474217518576897,
      "grad_norm": 1.4216922521591187,
      "learning_rate": 1.7707794237510973e-05,
      "loss": 1.3592,
      "step": 8035
    },
    {
      "epoch": 1.4476018914658861,
      "grad_norm": 1.3850549459457397,
      "learning_rate": 1.7696983643427078e-05,
      "loss": 1.3191,
      "step": 8036
    },
    {
      "epoch": 1.4477820310740825,
      "grad_norm": 1.3908941745758057,
      "learning_rate": 1.768617564067542e-05,
      "loss": 1.3333,
      "step": 8037
    },
    {
      "epoch": 1.4479621706822787,
      "grad_norm": 1.4673216342926025,
      "learning_rate": 1.7675370230123017e-05,
      "loss": 1.3204,
      "step": 8038
    },
    {
      "epoch": 1.4481423102904751,
      "grad_norm": 1.5468907356262207,
      "learning_rate": 1.766456741263665e-05,
      "loss": 1.3236,
      "step": 8039
    },
    {
      "epoch": 1.4483224498986715,
      "grad_norm": 1.6482319831848145,
      "learning_rate": 1.7653767189082955e-05,
      "loss": 1.5268,
      "step": 8040
    },
    {
      "epoch": 1.4485025895068677,
      "grad_norm": 1.6609482765197754,
      "learning_rate": 1.76429695603283e-05,
      "loss": 1.5336,
      "step": 8041
    },
    {
      "epoch": 1.4486827291150641,
      "grad_norm": 1.6685922145843506,
      "learning_rate": 1.7632174527238866e-05,
      "loss": 1.4974,
      "step": 8042
    },
    {
      "epoch": 1.4488628687232605,
      "grad_norm": 1.558226227760315,
      "learning_rate": 1.7621382090680628e-05,
      "loss": 1.4893,
      "step": 8043
    },
    {
      "epoch": 1.449043008331457,
      "grad_norm": 1.8582466840744019,
      "learning_rate": 1.761059225151934e-05,
      "loss": 1.6828,
      "step": 8044
    },
    {
      "epoch": 1.4492231479396533,
      "grad_norm": 1.4842251539230347,
      "learning_rate": 1.7599805010620574e-05,
      "loss": 1.4216,
      "step": 8045
    },
    {
      "epoch": 1.4494032875478495,
      "grad_norm": 1.576616644859314,
      "learning_rate": 1.7589020368849664e-05,
      "loss": 1.3209,
      "step": 8046
    },
    {
      "epoch": 1.449583427156046,
      "grad_norm": 1.6401101350784302,
      "learning_rate": 1.7578238327071748e-05,
      "loss": 1.3026,
      "step": 8047
    },
    {
      "epoch": 1.4497635667642423,
      "grad_norm": 1.5385409593582153,
      "learning_rate": 1.7567458886151754e-05,
      "loss": 1.3966,
      "step": 8048
    },
    {
      "epoch": 1.4499437063724385,
      "grad_norm": 1.4719061851501465,
      "learning_rate": 1.755668204695442e-05,
      "loss": 1.1243,
      "step": 8049
    },
    {
      "epoch": 1.450123845980635,
      "grad_norm": 1.4997024536132812,
      "learning_rate": 1.7545907810344254e-05,
      "loss": 1.3566,
      "step": 8050
    },
    {
      "epoch": 1.4503039855888313,
      "grad_norm": 1.4546380043029785,
      "learning_rate": 1.7535136177185556e-05,
      "loss": 1.3899,
      "step": 8051
    },
    {
      "epoch": 1.4504841251970277,
      "grad_norm": 1.4713420867919922,
      "learning_rate": 1.752436714834242e-05,
      "loss": 2.0031,
      "step": 8052
    },
    {
      "epoch": 1.4506642648052241,
      "grad_norm": 1.2983648777008057,
      "learning_rate": 1.751360072467873e-05,
      "loss": 1.857,
      "step": 8053
    },
    {
      "epoch": 1.4508444044134203,
      "grad_norm": 1.4595049619674683,
      "learning_rate": 1.7502836907058174e-05,
      "loss": 1.9324,
      "step": 8054
    },
    {
      "epoch": 1.4510245440216167,
      "grad_norm": 1.3958483934402466,
      "learning_rate": 1.7492075696344207e-05,
      "loss": 1.7987,
      "step": 8055
    },
    {
      "epoch": 1.4512046836298131,
      "grad_norm": 1.489006757736206,
      "learning_rate": 1.748131709340009e-05,
      "loss": 1.9766,
      "step": 8056
    },
    {
      "epoch": 1.4513848232380093,
      "grad_norm": 1.471548318862915,
      "learning_rate": 1.7470561099088867e-05,
      "loss": 1.6195,
      "step": 8057
    },
    {
      "epoch": 1.4515649628462057,
      "grad_norm": 1.4130973815917969,
      "learning_rate": 1.7459807714273402e-05,
      "loss": 1.677,
      "step": 8058
    },
    {
      "epoch": 1.4517451024544021,
      "grad_norm": 1.848110318183899,
      "learning_rate": 1.744905693981631e-05,
      "loss": 1.92,
      "step": 8059
    },
    {
      "epoch": 1.4519252420625985,
      "grad_norm": 1.632275104522705,
      "learning_rate": 1.743830877658002e-05,
      "loss": 1.9258,
      "step": 8060
    },
    {
      "epoch": 1.452105381670795,
      "grad_norm": 1.9830628633499146,
      "learning_rate": 1.7427563225426734e-05,
      "loss": 1.7233,
      "step": 8061
    },
    {
      "epoch": 1.4522855212789914,
      "grad_norm": 1.3974878787994385,
      "learning_rate": 1.741682028721845e-05,
      "loss": 1.3587,
      "step": 8062
    },
    {
      "epoch": 1.4524656608871875,
      "grad_norm": 1.3284026384353638,
      "learning_rate": 1.7406079962816986e-05,
      "loss": 1.3569,
      "step": 8063
    },
    {
      "epoch": 1.452645800495384,
      "grad_norm": 1.5110300779342651,
      "learning_rate": 1.7395342253083925e-05,
      "loss": 1.3713,
      "step": 8064
    },
    {
      "epoch": 1.4528259401035803,
      "grad_norm": 1.3864331245422363,
      "learning_rate": 1.738460715888061e-05,
      "loss": 1.5585,
      "step": 8065
    },
    {
      "epoch": 1.4530060797117765,
      "grad_norm": 1.436155080795288,
      "learning_rate": 1.737387468106823e-05,
      "loss": 1.359,
      "step": 8066
    },
    {
      "epoch": 1.453186219319973,
      "grad_norm": 1.6059166193008423,
      "learning_rate": 1.736314482050771e-05,
      "loss": 1.7009,
      "step": 8067
    },
    {
      "epoch": 1.4533663589281693,
      "grad_norm": 1.3695902824401855,
      "learning_rate": 1.735241757805983e-05,
      "loss": 1.3248,
      "step": 8068
    },
    {
      "epoch": 1.4535464985363657,
      "grad_norm": 1.3821109533309937,
      "learning_rate": 1.7341692954585113e-05,
      "loss": 1.2743,
      "step": 8069
    },
    {
      "epoch": 1.4537266381445622,
      "grad_norm": 1.43107008934021,
      "learning_rate": 1.733097095094388e-05,
      "loss": 1.4647,
      "step": 8070
    },
    {
      "epoch": 1.4539067777527583,
      "grad_norm": 1.3265044689178467,
      "learning_rate": 1.7320251567996232e-05,
      "loss": 1.3245,
      "step": 8071
    },
    {
      "epoch": 1.4540869173609547,
      "grad_norm": 1.2950725555419922,
      "learning_rate": 1.7309534806602097e-05,
      "loss": 1.3646,
      "step": 8072
    },
    {
      "epoch": 1.4542670569691512,
      "grad_norm": 1.5401910543441772,
      "learning_rate": 1.7298820667621156e-05,
      "loss": 1.7558,
      "step": 8073
    },
    {
      "epoch": 1.4544471965773473,
      "grad_norm": 1.3882255554199219,
      "learning_rate": 1.7288109151912895e-05,
      "loss": 1.3777,
      "step": 8074
    },
    {
      "epoch": 1.4546273361855437,
      "grad_norm": 1.4878950119018555,
      "learning_rate": 1.7277400260336584e-05,
      "loss": 1.3309,
      "step": 8075
    },
    {
      "epoch": 1.4548074757937401,
      "grad_norm": 1.3252559900283813,
      "learning_rate": 1.726669399375128e-05,
      "loss": 1.1664,
      "step": 8076
    },
    {
      "epoch": 1.4549876154019366,
      "grad_norm": 1.6111396551132202,
      "learning_rate": 1.725599035301585e-05,
      "loss": 1.5897,
      "step": 8077
    },
    {
      "epoch": 1.455167755010133,
      "grad_norm": 1.436450719833374,
      "learning_rate": 1.7245289338988917e-05,
      "loss": 1.341,
      "step": 8078
    },
    {
      "epoch": 1.4553478946183291,
      "grad_norm": 1.3740235567092896,
      "learning_rate": 1.7234590952528924e-05,
      "loss": 1.2053,
      "step": 8079
    },
    {
      "epoch": 1.4555280342265255,
      "grad_norm": 1.5159800052642822,
      "learning_rate": 1.722389519449408e-05,
      "loss": 1.5903,
      "step": 8080
    },
    {
      "epoch": 1.455708173834722,
      "grad_norm": 1.4350194931030273,
      "learning_rate": 1.721320206574239e-05,
      "loss": 1.2955,
      "step": 8081
    },
    {
      "epoch": 1.4558883134429181,
      "grad_norm": 1.4336975812911987,
      "learning_rate": 1.7202511567131672e-05,
      "loss": 1.3889,
      "step": 8082
    },
    {
      "epoch": 1.4560684530511145,
      "grad_norm": 1.5354095697402954,
      "learning_rate": 1.7191823699519506e-05,
      "loss": 1.4385,
      "step": 8083
    },
    {
      "epoch": 1.456248592659311,
      "grad_norm": 1.5346239805221558,
      "learning_rate": 1.718113846376326e-05,
      "loss": 1.5583,
      "step": 8084
    },
    {
      "epoch": 1.4564287322675074,
      "grad_norm": 1.5364880561828613,
      "learning_rate": 1.7170455860720096e-05,
      "loss": 1.5527,
      "step": 8085
    },
    {
      "epoch": 1.4566088718757038,
      "grad_norm": 1.5669749975204468,
      "learning_rate": 1.7159775891246976e-05,
      "loss": 1.4579,
      "step": 8086
    },
    {
      "epoch": 1.4567890114839,
      "grad_norm": 1.5393072366714478,
      "learning_rate": 1.7149098556200637e-05,
      "loss": 1.3381,
      "step": 8087
    },
    {
      "epoch": 1.4569691510920963,
      "grad_norm": 1.3815019130706787,
      "learning_rate": 1.7138423856437606e-05,
      "loss": 1.3856,
      "step": 8088
    },
    {
      "epoch": 1.4571492907002928,
      "grad_norm": 1.5448921918869019,
      "learning_rate": 1.7127751792814217e-05,
      "loss": 1.6371,
      "step": 8089
    },
    {
      "epoch": 1.4573294303084892,
      "grad_norm": 1.5260528326034546,
      "learning_rate": 1.7117082366186543e-05,
      "loss": 1.4921,
      "step": 8090
    },
    {
      "epoch": 1.4575095699166853,
      "grad_norm": 1.4143253564834595,
      "learning_rate": 1.710641557741053e-05,
      "loss": 1.2475,
      "step": 8091
    },
    {
      "epoch": 1.4576897095248817,
      "grad_norm": 1.579866886138916,
      "learning_rate": 1.7095751427341826e-05,
      "loss": 1.4775,
      "step": 8092
    },
    {
      "epoch": 1.4578698491330782,
      "grad_norm": 1.6168063879013062,
      "learning_rate": 1.7085089916835923e-05,
      "loss": 1.4495,
      "step": 8093
    },
    {
      "epoch": 1.4580499887412746,
      "grad_norm": 1.4802608489990234,
      "learning_rate": 1.7074431046748075e-05,
      "loss": 1.4819,
      "step": 8094
    },
    {
      "epoch": 1.458230128349471,
      "grad_norm": 1.4431006908416748,
      "learning_rate": 1.7063774817933313e-05,
      "loss": 1.4312,
      "step": 8095
    },
    {
      "epoch": 1.4584102679576672,
      "grad_norm": 1.596966028213501,
      "learning_rate": 1.705312123124653e-05,
      "loss": 1.4235,
      "step": 8096
    },
    {
      "epoch": 1.4585904075658636,
      "grad_norm": 1.4923477172851562,
      "learning_rate": 1.7042470287542288e-05,
      "loss": 1.2696,
      "step": 8097
    },
    {
      "epoch": 1.45877054717406,
      "grad_norm": 1.5950254201889038,
      "learning_rate": 1.703182198767503e-05,
      "loss": 1.3189,
      "step": 8098
    },
    {
      "epoch": 1.4589506867822561,
      "grad_norm": 1.7292513847351074,
      "learning_rate": 1.7021176332498938e-05,
      "loss": 1.7945,
      "step": 8099
    },
    {
      "epoch": 1.4591308263904526,
      "grad_norm": 1.3566813468933105,
      "learning_rate": 1.7010533322868034e-05,
      "loss": 1.2219,
      "step": 8100
    },
    {
      "epoch": 1.459310965998649,
      "grad_norm": 1.4696025848388672,
      "learning_rate": 1.6999892959636082e-05,
      "loss": 1.3622,
      "step": 8101
    },
    {
      "epoch": 1.4594911056068454,
      "grad_norm": 1.4046486616134644,
      "learning_rate": 1.6989255243656637e-05,
      "loss": 1.7419,
      "step": 8102
    },
    {
      "epoch": 1.4596712452150418,
      "grad_norm": 1.257022738456726,
      "learning_rate": 1.6978620175783062e-05,
      "loss": 1.725,
      "step": 8103
    },
    {
      "epoch": 1.459851384823238,
      "grad_norm": 1.4738199710845947,
      "learning_rate": 1.696798775686847e-05,
      "loss": 2.1063,
      "step": 8104
    },
    {
      "epoch": 1.4600315244314344,
      "grad_norm": 1.4084230661392212,
      "learning_rate": 1.6957357987765832e-05,
      "loss": 1.6651,
      "step": 8105
    },
    {
      "epoch": 1.4602116640396308,
      "grad_norm": 1.3796098232269287,
      "learning_rate": 1.6946730869327844e-05,
      "loss": 1.7655,
      "step": 8106
    },
    {
      "epoch": 1.460391803647827,
      "grad_norm": 1.490235686302185,
      "learning_rate": 1.6936106402406998e-05,
      "loss": 1.7923,
      "step": 8107
    },
    {
      "epoch": 1.4605719432560234,
      "grad_norm": 1.4591264724731445,
      "learning_rate": 1.69254845878556e-05,
      "loss": 1.6397,
      "step": 8108
    },
    {
      "epoch": 1.4607520828642198,
      "grad_norm": 1.6688789129257202,
      "learning_rate": 1.691486542652571e-05,
      "loss": 1.7579,
      "step": 8109
    },
    {
      "epoch": 1.4609322224724162,
      "grad_norm": 1.699428677558899,
      "learning_rate": 1.6904248919269212e-05,
      "loss": 1.9655,
      "step": 8110
    },
    {
      "epoch": 1.4611123620806126,
      "grad_norm": 1.4834977388381958,
      "learning_rate": 1.6893635066937737e-05,
      "loss": 1.7917,
      "step": 8111
    },
    {
      "epoch": 1.4612925016888088,
      "grad_norm": 1.4167309999465942,
      "learning_rate": 1.688302387038273e-05,
      "loss": 1.6533,
      "step": 8112
    },
    {
      "epoch": 1.4614726412970052,
      "grad_norm": 1.3667678833007812,
      "learning_rate": 1.6872415330455406e-05,
      "loss": 1.3328,
      "step": 8113
    },
    {
      "epoch": 1.4616527809052016,
      "grad_norm": 1.3453348875045776,
      "learning_rate": 1.6861809448006798e-05,
      "loss": 1.4076,
      "step": 8114
    },
    {
      "epoch": 1.4618329205133977,
      "grad_norm": 1.3253042697906494,
      "learning_rate": 1.68512062238877e-05,
      "loss": 1.3529,
      "step": 8115
    },
    {
      "epoch": 1.4620130601215942,
      "grad_norm": 1.3924373388290405,
      "learning_rate": 1.6840605658948695e-05,
      "loss": 1.4758,
      "step": 8116
    },
    {
      "epoch": 1.4621931997297906,
      "grad_norm": 1.377272367477417,
      "learning_rate": 1.683000775404015e-05,
      "loss": 1.4557,
      "step": 8117
    },
    {
      "epoch": 1.462373339337987,
      "grad_norm": 1.3477051258087158,
      "learning_rate": 1.6819412510012222e-05,
      "loss": 1.303,
      "step": 8118
    },
    {
      "epoch": 1.4625534789461834,
      "grad_norm": 1.280418038368225,
      "learning_rate": 1.680881992771487e-05,
      "loss": 1.3418,
      "step": 8119
    },
    {
      "epoch": 1.4627336185543798,
      "grad_norm": 1.3980178833007812,
      "learning_rate": 1.679823000799781e-05,
      "loss": 1.323,
      "step": 8120
    },
    {
      "epoch": 1.462913758162576,
      "grad_norm": 1.5017340183258057,
      "learning_rate": 1.678764275171057e-05,
      "loss": 1.4848,
      "step": 8121
    },
    {
      "epoch": 1.4630938977707724,
      "grad_norm": 1.4132311344146729,
      "learning_rate": 1.677705815970243e-05,
      "loss": 1.4837,
      "step": 8122
    },
    {
      "epoch": 1.4632740373789688,
      "grad_norm": 1.404493808746338,
      "learning_rate": 1.676647623282252e-05,
      "loss": 1.4461,
      "step": 8123
    },
    {
      "epoch": 1.463454176987165,
      "grad_norm": 1.4357677698135376,
      "learning_rate": 1.6755896971919704e-05,
      "loss": 1.3661,
      "step": 8124
    },
    {
      "epoch": 1.4636343165953614,
      "grad_norm": 1.599934697151184,
      "learning_rate": 1.674532037784264e-05,
      "loss": 1.6575,
      "step": 8125
    },
    {
      "epoch": 1.4638144562035578,
      "grad_norm": 1.3421075344085693,
      "learning_rate": 1.673474645143977e-05,
      "loss": 1.2776,
      "step": 8126
    },
    {
      "epoch": 1.4639945958117542,
      "grad_norm": 1.392279028892517,
      "learning_rate": 1.6724175193559327e-05,
      "loss": 1.3085,
      "step": 8127
    },
    {
      "epoch": 1.4641747354199506,
      "grad_norm": 1.3773834705352783,
      "learning_rate": 1.6713606605049355e-05,
      "loss": 1.2416,
      "step": 8128
    },
    {
      "epoch": 1.4643548750281468,
      "grad_norm": 1.3469650745391846,
      "learning_rate": 1.6703040686757666e-05,
      "loss": 1.2943,
      "step": 8129
    },
    {
      "epoch": 1.4645350146363432,
      "grad_norm": 1.4431020021438599,
      "learning_rate": 1.6692477439531817e-05,
      "loss": 1.2578,
      "step": 8130
    },
    {
      "epoch": 1.4647151542445396,
      "grad_norm": 1.6385438442230225,
      "learning_rate": 1.6681916864219206e-05,
      "loss": 1.6553,
      "step": 8131
    },
    {
      "epoch": 1.4648952938527358,
      "grad_norm": 1.5582624673843384,
      "learning_rate": 1.6671358961666977e-05,
      "loss": 1.3421,
      "step": 8132
    },
    {
      "epoch": 1.4650754334609322,
      "grad_norm": 1.5187559127807617,
      "learning_rate": 1.6660803732722114e-05,
      "loss": 1.484,
      "step": 8133
    },
    {
      "epoch": 1.4652555730691286,
      "grad_norm": 1.604551911354065,
      "learning_rate": 1.665025117823134e-05,
      "loss": 1.4912,
      "step": 8134
    },
    {
      "epoch": 1.465435712677325,
      "grad_norm": 1.59600830078125,
      "learning_rate": 1.6639701299041172e-05,
      "loss": 1.3918,
      "step": 8135
    },
    {
      "epoch": 1.4656158522855214,
      "grad_norm": 1.6133476495742798,
      "learning_rate": 1.6629154095997894e-05,
      "loss": 1.5449,
      "step": 8136
    },
    {
      "epoch": 1.4657959918937176,
      "grad_norm": 1.5654728412628174,
      "learning_rate": 1.6618609569947645e-05,
      "loss": 1.4359,
      "step": 8137
    },
    {
      "epoch": 1.465976131501914,
      "grad_norm": 1.5909291505813599,
      "learning_rate": 1.6608067721736277e-05,
      "loss": 1.4428,
      "step": 8138
    },
    {
      "epoch": 1.4661562711101104,
      "grad_norm": 1.531416654586792,
      "learning_rate": 1.6597528552209447e-05,
      "loss": 1.3344,
      "step": 8139
    },
    {
      "epoch": 1.4663364107183066,
      "grad_norm": 1.518561601638794,
      "learning_rate": 1.658699206221261e-05,
      "loss": 1.4462,
      "step": 8140
    },
    {
      "epoch": 1.466516550326503,
      "grad_norm": 1.4122933149337769,
      "learning_rate": 1.6576458252590986e-05,
      "loss": 1.0855,
      "step": 8141
    },
    {
      "epoch": 1.4666966899346994,
      "grad_norm": 1.4264907836914062,
      "learning_rate": 1.6565927124189613e-05,
      "loss": 1.346,
      "step": 8142
    },
    {
      "epoch": 1.4668768295428958,
      "grad_norm": 1.615363597869873,
      "learning_rate": 1.6555398677853273e-05,
      "loss": 1.5272,
      "step": 8143
    },
    {
      "epoch": 1.4670569691510922,
      "grad_norm": 1.4662690162658691,
      "learning_rate": 1.6544872914426563e-05,
      "loss": 1.4121,
      "step": 8144
    },
    {
      "epoch": 1.4672371087592884,
      "grad_norm": 1.3685420751571655,
      "learning_rate": 1.6534349834753855e-05,
      "loss": 1.1788,
      "step": 8145
    },
    {
      "epoch": 1.4674172483674848,
      "grad_norm": 1.6929986476898193,
      "learning_rate": 1.6523829439679284e-05,
      "loss": 1.3447,
      "step": 8146
    },
    {
      "epoch": 1.4675973879756812,
      "grad_norm": 1.7316865921020508,
      "learning_rate": 1.6513311730046828e-05,
      "loss": 1.5403,
      "step": 8147
    },
    {
      "epoch": 1.4677775275838776,
      "grad_norm": 1.5148394107818604,
      "learning_rate": 1.650279670670019e-05,
      "loss": 1.1846,
      "step": 8148
    },
    {
      "epoch": 1.4679576671920738,
      "grad_norm": 1.6269383430480957,
      "learning_rate": 1.6492284370482884e-05,
      "loss": 1.4486,
      "step": 8149
    },
    {
      "epoch": 1.4681378068002702,
      "grad_norm": 1.5044218301773071,
      "learning_rate": 1.648177472223821e-05,
      "loss": 1.1348,
      "step": 8150
    },
    {
      "epoch": 1.4683179464084666,
      "grad_norm": 1.5331660509109497,
      "learning_rate": 1.6471267762809238e-05,
      "loss": 1.1829,
      "step": 8151
    },
    {
      "epoch": 1.468498086016663,
      "grad_norm": 1.3183190822601318,
      "learning_rate": 1.646076349303884e-05,
      "loss": 1.6851,
      "step": 8152
    },
    {
      "epoch": 1.4686782256248594,
      "grad_norm": 1.2717913389205933,
      "learning_rate": 1.645026191376965e-05,
      "loss": 1.71,
      "step": 8153
    },
    {
      "epoch": 1.4688583652330556,
      "grad_norm": 1.273877501487732,
      "learning_rate": 1.6439763025844114e-05,
      "loss": 1.807,
      "step": 8154
    },
    {
      "epoch": 1.469038504841252,
      "grad_norm": 1.4363099336624146,
      "learning_rate": 1.6429266830104424e-05,
      "loss": 1.7865,
      "step": 8155
    },
    {
      "epoch": 1.4692186444494484,
      "grad_norm": 1.4654121398925781,
      "learning_rate": 1.6418773327392612e-05,
      "loss": 1.9925,
      "step": 8156
    },
    {
      "epoch": 1.4693987840576446,
      "grad_norm": 1.468063473701477,
      "learning_rate": 1.6408282518550444e-05,
      "loss": 1.7062,
      "step": 8157
    },
    {
      "epoch": 1.469578923665841,
      "grad_norm": 1.6224817037582397,
      "learning_rate": 1.6397794404419497e-05,
      "loss": 1.9611,
      "step": 8158
    },
    {
      "epoch": 1.4697590632740374,
      "grad_norm": 1.5204607248306274,
      "learning_rate": 1.638730898584111e-05,
      "loss": 1.7028,
      "step": 8159
    },
    {
      "epoch": 1.4699392028822338,
      "grad_norm": 1.6457748413085938,
      "learning_rate": 1.6376826263656413e-05,
      "loss": 1.6574,
      "step": 8160
    },
    {
      "epoch": 1.4701193424904302,
      "grad_norm": 1.8649173974990845,
      "learning_rate": 1.6366346238706365e-05,
      "loss": 1.9153,
      "step": 8161
    },
    {
      "epoch": 1.4702994820986264,
      "grad_norm": 1.559179663658142,
      "learning_rate": 1.6355868911831625e-05,
      "loss": 1.4524,
      "step": 8162
    },
    {
      "epoch": 1.4704796217068228,
      "grad_norm": 1.5441402196884155,
      "learning_rate": 1.634539428387269e-05,
      "loss": 1.663,
      "step": 8163
    },
    {
      "epoch": 1.4706597613150192,
      "grad_norm": 1.3922497034072876,
      "learning_rate": 1.6334922355669818e-05,
      "loss": 1.4285,
      "step": 8164
    },
    {
      "epoch": 1.4708399009232154,
      "grad_norm": 1.4064722061157227,
      "learning_rate": 1.6324453128063095e-05,
      "loss": 1.4873,
      "step": 8165
    },
    {
      "epoch": 1.4710200405314118,
      "grad_norm": 1.421367883682251,
      "learning_rate": 1.631398660189234e-05,
      "loss": 1.5008,
      "step": 8166
    },
    {
      "epoch": 1.4712001801396082,
      "grad_norm": 1.622439980506897,
      "learning_rate": 1.630352277799717e-05,
      "loss": 1.6112,
      "step": 8167
    },
    {
      "epoch": 1.4713803197478046,
      "grad_norm": 1.567834496498108,
      "learning_rate": 1.629306165721699e-05,
      "loss": 1.4846,
      "step": 8168
    },
    {
      "epoch": 1.471560459356001,
      "grad_norm": 1.425642728805542,
      "learning_rate": 1.628260324039097e-05,
      "loss": 1.2851,
      "step": 8169
    },
    {
      "epoch": 1.4717405989641972,
      "grad_norm": 1.3928115367889404,
      "learning_rate": 1.627214752835811e-05,
      "loss": 1.3105,
      "step": 8170
    },
    {
      "epoch": 1.4719207385723936,
      "grad_norm": 1.5022599697113037,
      "learning_rate": 1.626169452195715e-05,
      "loss": 1.4828,
      "step": 8171
    },
    {
      "epoch": 1.47210087818059,
      "grad_norm": 1.4822044372558594,
      "learning_rate": 1.625124422202662e-05,
      "loss": 1.4494,
      "step": 8172
    },
    {
      "epoch": 1.4722810177887862,
      "grad_norm": 1.4153070449829102,
      "learning_rate": 1.624079662940484e-05,
      "loss": 1.3555,
      "step": 8173
    },
    {
      "epoch": 1.4724611573969826,
      "grad_norm": 1.4269487857818604,
      "learning_rate": 1.623035174492991e-05,
      "loss": 1.2848,
      "step": 8174
    },
    {
      "epoch": 1.472641297005179,
      "grad_norm": 1.3872058391571045,
      "learning_rate": 1.6219909569439717e-05,
      "loss": 1.3322,
      "step": 8175
    },
    {
      "epoch": 1.4728214366133754,
      "grad_norm": 1.395554542541504,
      "learning_rate": 1.6209470103771923e-05,
      "loss": 1.1155,
      "step": 8176
    },
    {
      "epoch": 1.4730015762215718,
      "grad_norm": 1.4002515077590942,
      "learning_rate": 1.6199033348763975e-05,
      "loss": 1.271,
      "step": 8177
    },
    {
      "epoch": 1.4731817158297682,
      "grad_norm": 1.5238114595413208,
      "learning_rate": 1.61885993052531e-05,
      "loss": 1.5608,
      "step": 8178
    },
    {
      "epoch": 1.4733618554379644,
      "grad_norm": 1.462597370147705,
      "learning_rate": 1.617816797407633e-05,
      "loss": 1.3474,
      "step": 8179
    },
    {
      "epoch": 1.4735419950461608,
      "grad_norm": 1.5667364597320557,
      "learning_rate": 1.616773935607045e-05,
      "loss": 1.4376,
      "step": 8180
    },
    {
      "epoch": 1.4737221346543572,
      "grad_norm": 1.6095837354660034,
      "learning_rate": 1.6157313452072042e-05,
      "loss": 1.5954,
      "step": 8181
    },
    {
      "epoch": 1.4739022742625534,
      "grad_norm": 1.4854878187179565,
      "learning_rate": 1.614689026291747e-05,
      "loss": 1.3969,
      "step": 8182
    },
    {
      "epoch": 1.4740824138707498,
      "grad_norm": 1.3541326522827148,
      "learning_rate": 1.6136469789442864e-05,
      "loss": 1.3586,
      "step": 8183
    },
    {
      "epoch": 1.4742625534789462,
      "grad_norm": 1.4171067476272583,
      "learning_rate": 1.6126052032484157e-05,
      "loss": 1.3603,
      "step": 8184
    },
    {
      "epoch": 1.4744426930871426,
      "grad_norm": 1.4077274799346924,
      "learning_rate": 1.611563699287706e-05,
      "loss": 1.2536,
      "step": 8185
    },
    {
      "epoch": 1.474622832695339,
      "grad_norm": 1.596418023109436,
      "learning_rate": 1.610522467145706e-05,
      "loss": 1.3672,
      "step": 8186
    },
    {
      "epoch": 1.4748029723035352,
      "grad_norm": 1.4944181442260742,
      "learning_rate": 1.6094815069059428e-05,
      "loss": 1.3574,
      "step": 8187
    },
    {
      "epoch": 1.4749831119117316,
      "grad_norm": 1.5643279552459717,
      "learning_rate": 1.6084408186519196e-05,
      "loss": 1.5955,
      "step": 8188
    },
    {
      "epoch": 1.475163251519928,
      "grad_norm": 1.5402911901474,
      "learning_rate": 1.6074004024671234e-05,
      "loss": 1.5891,
      "step": 8189
    },
    {
      "epoch": 1.4753433911281242,
      "grad_norm": 1.412166714668274,
      "learning_rate": 1.606360258435015e-05,
      "loss": 1.4377,
      "step": 8190
    },
    {
      "epoch": 1.4755235307363206,
      "grad_norm": 1.6369582414627075,
      "learning_rate": 1.6053203866390336e-05,
      "loss": 1.4425,
      "step": 8191
    },
    {
      "epoch": 1.475703670344517,
      "grad_norm": 1.497342824935913,
      "learning_rate": 1.6042807871625953e-05,
      "loss": 1.2867,
      "step": 8192
    },
    {
      "epoch": 1.4758838099527134,
      "grad_norm": 1.5337201356887817,
      "learning_rate": 1.6032414600891005e-05,
      "loss": 1.408,
      "step": 8193
    },
    {
      "epoch": 1.4760639495609098,
      "grad_norm": 1.6361474990844727,
      "learning_rate": 1.6022024055019224e-05,
      "loss": 1.523,
      "step": 8194
    },
    {
      "epoch": 1.476244089169106,
      "grad_norm": 1.427764654159546,
      "learning_rate": 1.6011636234844106e-05,
      "loss": 1.243,
      "step": 8195
    },
    {
      "epoch": 1.4764242287773024,
      "grad_norm": 1.5448062419891357,
      "learning_rate": 1.600125114119898e-05,
      "loss": 1.3187,
      "step": 8196
    },
    {
      "epoch": 1.4766043683854988,
      "grad_norm": 1.6024812459945679,
      "learning_rate": 1.5990868774916906e-05,
      "loss": 1.5284,
      "step": 8197
    },
    {
      "epoch": 1.476784507993695,
      "grad_norm": 1.610843300819397,
      "learning_rate": 1.5980489136830794e-05,
      "loss": 1.3388,
      "step": 8198
    },
    {
      "epoch": 1.4769646476018914,
      "grad_norm": 1.6666816473007202,
      "learning_rate": 1.597011222777327e-05,
      "loss": 1.4057,
      "step": 8199
    },
    {
      "epoch": 1.4771447872100878,
      "grad_norm": 1.4728418588638306,
      "learning_rate": 1.5959738048576767e-05,
      "loss": 1.1331,
      "step": 8200
    },
    {
      "epoch": 1.4773249268182842,
      "grad_norm": 1.389033317565918,
      "learning_rate": 1.5949366600073502e-05,
      "loss": 1.0723,
      "step": 8201
    },
    {
      "epoch": 1.4775050664264806,
      "grad_norm": 1.417446255683899,
      "learning_rate": 1.5938997883095442e-05,
      "loss": 2.1171,
      "step": 8202
    },
    {
      "epoch": 1.477685206034677,
      "grad_norm": 1.3437901735305786,
      "learning_rate": 1.5928631898474393e-05,
      "loss": 1.7456,
      "step": 8203
    },
    {
      "epoch": 1.4778653456428732,
      "grad_norm": 1.3137058019638062,
      "learning_rate": 1.5918268647041908e-05,
      "loss": 1.7551,
      "step": 8204
    },
    {
      "epoch": 1.4780454852510696,
      "grad_norm": 1.312812089920044,
      "learning_rate": 1.5907908129629305e-05,
      "loss": 1.5694,
      "step": 8205
    },
    {
      "epoch": 1.478225624859266,
      "grad_norm": 1.5139602422714233,
      "learning_rate": 1.5897550347067706e-05,
      "loss": 1.7434,
      "step": 8206
    },
    {
      "epoch": 1.4784057644674622,
      "grad_norm": 1.526865005493164,
      "learning_rate": 1.588719530018801e-05,
      "loss": 1.6474,
      "step": 8207
    },
    {
      "epoch": 1.4785859040756586,
      "grad_norm": 1.5279940366744995,
      "learning_rate": 1.5876842989820895e-05,
      "loss": 1.6849,
      "step": 8208
    },
    {
      "epoch": 1.478766043683855,
      "grad_norm": 1.6920865774154663,
      "learning_rate": 1.5866493416796808e-05,
      "loss": 2.2204,
      "step": 8209
    },
    {
      "epoch": 1.4789461832920514,
      "grad_norm": 1.884311318397522,
      "learning_rate": 1.5856146581945995e-05,
      "loss": 1.8946,
      "step": 8210
    },
    {
      "epoch": 1.4791263229002478,
      "grad_norm": 1.6508115530014038,
      "learning_rate": 1.584580248609846e-05,
      "loss": 1.7426,
      "step": 8211
    },
    {
      "epoch": 1.479306462508444,
      "grad_norm": 1.573927879333496,
      "learning_rate": 1.583546113008402e-05,
      "loss": 1.6406,
      "step": 8212
    },
    {
      "epoch": 1.4794866021166404,
      "grad_norm": 1.3276898860931396,
      "learning_rate": 1.582512251473226e-05,
      "loss": 1.3468,
      "step": 8213
    },
    {
      "epoch": 1.4796667417248368,
      "grad_norm": 1.588862657546997,
      "learning_rate": 1.581478664087252e-05,
      "loss": 1.5124,
      "step": 8214
    },
    {
      "epoch": 1.479846881333033,
      "grad_norm": 1.2988438606262207,
      "learning_rate": 1.5804453509333944e-05,
      "loss": 1.3068,
      "step": 8215
    },
    {
      "epoch": 1.4800270209412294,
      "grad_norm": 1.3855034112930298,
      "learning_rate": 1.5794123120945447e-05,
      "loss": 1.4494,
      "step": 8216
    },
    {
      "epoch": 1.4802071605494258,
      "grad_norm": 1.3087507486343384,
      "learning_rate": 1.578379547653574e-05,
      "loss": 1.312,
      "step": 8217
    },
    {
      "epoch": 1.4803873001576222,
      "grad_norm": 1.395614504814148,
      "learning_rate": 1.5773470576933292e-05,
      "loss": 1.3759,
      "step": 8218
    },
    {
      "epoch": 1.4805674397658186,
      "grad_norm": 1.3604282140731812,
      "learning_rate": 1.5763148422966363e-05,
      "loss": 1.1742,
      "step": 8219
    },
    {
      "epoch": 1.4807475793740148,
      "grad_norm": 1.4736851453781128,
      "learning_rate": 1.575282901546297e-05,
      "loss": 1.5942,
      "step": 8220
    },
    {
      "epoch": 1.4809277189822112,
      "grad_norm": 1.3666789531707764,
      "learning_rate": 1.574251235525097e-05,
      "loss": 1.2418,
      "step": 8221
    },
    {
      "epoch": 1.4811078585904076,
      "grad_norm": 1.466553807258606,
      "learning_rate": 1.5732198443157947e-05,
      "loss": 1.4776,
      "step": 8222
    },
    {
      "epoch": 1.4812879981986038,
      "grad_norm": 1.5183740854263306,
      "learning_rate": 1.572188728001127e-05,
      "loss": 1.4796,
      "step": 8223
    },
    {
      "epoch": 1.4814681378068002,
      "grad_norm": 1.3407610654830933,
      "learning_rate": 1.5711578866638104e-05,
      "loss": 1.25,
      "step": 8224
    },
    {
      "epoch": 1.4816482774149966,
      "grad_norm": 1.4137070178985596,
      "learning_rate": 1.5701273203865358e-05,
      "loss": 1.3589,
      "step": 8225
    },
    {
      "epoch": 1.481828417023193,
      "grad_norm": 1.4186859130859375,
      "learning_rate": 1.5690970292519807e-05,
      "loss": 1.4058,
      "step": 8226
    },
    {
      "epoch": 1.4820085566313894,
      "grad_norm": 1.342305064201355,
      "learning_rate": 1.5680670133427888e-05,
      "loss": 1.2409,
      "step": 8227
    },
    {
      "epoch": 1.4821886962395856,
      "grad_norm": 1.4955202341079712,
      "learning_rate": 1.5670372727415895e-05,
      "loss": 1.525,
      "step": 8228
    },
    {
      "epoch": 1.482368835847782,
      "grad_norm": 1.556486964225769,
      "learning_rate": 1.5660078075309865e-05,
      "loss": 1.5135,
      "step": 8229
    },
    {
      "epoch": 1.4825489754559784,
      "grad_norm": 1.4578251838684082,
      "learning_rate": 1.564978617793566e-05,
      "loss": 1.3671,
      "step": 8230
    },
    {
      "epoch": 1.4827291150641748,
      "grad_norm": 1.5932451486587524,
      "learning_rate": 1.5639497036118877e-05,
      "loss": 1.4459,
      "step": 8231
    },
    {
      "epoch": 1.482909254672371,
      "grad_norm": 1.3448258638381958,
      "learning_rate": 1.5629210650684906e-05,
      "loss": 1.2717,
      "step": 8232
    },
    {
      "epoch": 1.4830893942805674,
      "grad_norm": 1.6008577346801758,
      "learning_rate": 1.5618927022458917e-05,
      "loss": 1.6105,
      "step": 8233
    },
    {
      "epoch": 1.4832695338887638,
      "grad_norm": 1.4718765020370483,
      "learning_rate": 1.560864615226584e-05,
      "loss": 1.3592,
      "step": 8234
    },
    {
      "epoch": 1.4834496734969602,
      "grad_norm": 1.5544114112854004,
      "learning_rate": 1.5598368040930426e-05,
      "loss": 1.5255,
      "step": 8235
    },
    {
      "epoch": 1.4836298131051566,
      "grad_norm": 1.5295480489730835,
      "learning_rate": 1.558809268927718e-05,
      "loss": 1.2372,
      "step": 8236
    },
    {
      "epoch": 1.4838099527133528,
      "grad_norm": 1.587200403213501,
      "learning_rate": 1.557782009813037e-05,
      "loss": 1.3634,
      "step": 8237
    },
    {
      "epoch": 1.4839900923215492,
      "grad_norm": 1.5351425409317017,
      "learning_rate": 1.5567550268314084e-05,
      "loss": 1.4644,
      "step": 8238
    },
    {
      "epoch": 1.4841702319297456,
      "grad_norm": 1.5293402671813965,
      "learning_rate": 1.5557283200652118e-05,
      "loss": 1.3356,
      "step": 8239
    },
    {
      "epoch": 1.4843503715379418,
      "grad_norm": 1.566402792930603,
      "learning_rate": 1.5547018895968126e-05,
      "loss": 1.4628,
      "step": 8240
    },
    {
      "epoch": 1.4845305111461382,
      "grad_norm": 1.6386185884475708,
      "learning_rate": 1.5536757355085497e-05,
      "loss": 1.4203,
      "step": 8241
    },
    {
      "epoch": 1.4847106507543346,
      "grad_norm": 1.5085557699203491,
      "learning_rate": 1.5526498578827407e-05,
      "loss": 1.5069,
      "step": 8242
    },
    {
      "epoch": 1.484890790362531,
      "grad_norm": 1.5629349946975708,
      "learning_rate": 1.55162425680168e-05,
      "loss": 1.4052,
      "step": 8243
    },
    {
      "epoch": 1.4850709299707274,
      "grad_norm": 1.669832468032837,
      "learning_rate": 1.5505989323476428e-05,
      "loss": 1.4681,
      "step": 8244
    },
    {
      "epoch": 1.4852510695789236,
      "grad_norm": 1.5910685062408447,
      "learning_rate": 1.549573884602879e-05,
      "loss": 1.3384,
      "step": 8245
    },
    {
      "epoch": 1.48543120918712,
      "grad_norm": 1.6272332668304443,
      "learning_rate": 1.5485491136496176e-05,
      "loss": 1.4858,
      "step": 8246
    },
    {
      "epoch": 1.4856113487953164,
      "grad_norm": 1.501037836074829,
      "learning_rate": 1.5475246195700653e-05,
      "loss": 1.2718,
      "step": 8247
    },
    {
      "epoch": 1.4857914884035126,
      "grad_norm": 1.5413844585418701,
      "learning_rate": 1.5465004024464063e-05,
      "loss": 1.2205,
      "step": 8248
    },
    {
      "epoch": 1.485971628011709,
      "grad_norm": 1.4610179662704468,
      "learning_rate": 1.5454764623608032e-05,
      "loss": 1.2234,
      "step": 8249
    },
    {
      "epoch": 1.4861517676199054,
      "grad_norm": 1.7109169960021973,
      "learning_rate": 1.544452799395395e-05,
      "loss": 1.4434,
      "step": 8250
    },
    {
      "epoch": 1.4863319072281018,
      "grad_norm": 1.5412770509719849,
      "learning_rate": 1.5434294136323006e-05,
      "loss": 1.537,
      "step": 8251
    },
    {
      "epoch": 1.4865120468362982,
      "grad_norm": 1.4056484699249268,
      "learning_rate": 1.5424063051536147e-05,
      "loss": 1.9708,
      "step": 8252
    },
    {
      "epoch": 1.4866921864444944,
      "grad_norm": 1.4181978702545166,
      "learning_rate": 1.5413834740414096e-05,
      "loss": 1.8486,
      "step": 8253
    },
    {
      "epoch": 1.4868723260526908,
      "grad_norm": 1.3953505754470825,
      "learning_rate": 1.540360920377739e-05,
      "loss": 1.8603,
      "step": 8254
    },
    {
      "epoch": 1.4870524656608872,
      "grad_norm": 1.4219179153442383,
      "learning_rate": 1.53933864424463e-05,
      "loss": 1.663,
      "step": 8255
    },
    {
      "epoch": 1.4872326052690834,
      "grad_norm": 1.4097083806991577,
      "learning_rate": 1.53831664572409e-05,
      "loss": 1.7531,
      "step": 8256
    },
    {
      "epoch": 1.4874127448772798,
      "grad_norm": 1.4414830207824707,
      "learning_rate": 1.5372949248981005e-05,
      "loss": 1.7384,
      "step": 8257
    },
    {
      "epoch": 1.4875928844854762,
      "grad_norm": 1.6061363220214844,
      "learning_rate": 1.5362734818486284e-05,
      "loss": 1.9557,
      "step": 8258
    },
    {
      "epoch": 1.4877730240936726,
      "grad_norm": 1.4928147792816162,
      "learning_rate": 1.5352523166576095e-05,
      "loss": 1.6007,
      "step": 8259
    },
    {
      "epoch": 1.487953163701869,
      "grad_norm": 1.912382960319519,
      "learning_rate": 1.5342314294069614e-05,
      "loss": 2.0942,
      "step": 8260
    },
    {
      "epoch": 1.4881333033100654,
      "grad_norm": 1.760590672492981,
      "learning_rate": 1.5332108201785807e-05,
      "loss": 1.8964,
      "step": 8261
    },
    {
      "epoch": 1.4883134429182616,
      "grad_norm": 1.6435855627059937,
      "learning_rate": 1.5321904890543377e-05,
      "loss": 1.8126,
      "step": 8262
    },
    {
      "epoch": 1.488493582526458,
      "grad_norm": 1.5011520385742188,
      "learning_rate": 1.5311704361160854e-05,
      "loss": 1.4877,
      "step": 8263
    },
    {
      "epoch": 1.4886737221346544,
      "grad_norm": 1.3183709383010864,
      "learning_rate": 1.5301506614456516e-05,
      "loss": 1.1734,
      "step": 8264
    },
    {
      "epoch": 1.4888538617428506,
      "grad_norm": 1.4418071508407593,
      "learning_rate": 1.5291311651248417e-05,
      "loss": 1.4261,
      "step": 8265
    },
    {
      "epoch": 1.489034001351047,
      "grad_norm": 1.5169801712036133,
      "learning_rate": 1.5281119472354395e-05,
      "loss": 1.533,
      "step": 8266
    },
    {
      "epoch": 1.4892141409592434,
      "grad_norm": 1.3048747777938843,
      "learning_rate": 1.527093007859204e-05,
      "loss": 1.3612,
      "step": 8267
    },
    {
      "epoch": 1.4893942805674398,
      "grad_norm": 1.378918170928955,
      "learning_rate": 1.5260743470778766e-05,
      "loss": 1.3104,
      "step": 8268
    },
    {
      "epoch": 1.4895744201756362,
      "grad_norm": 1.417015552520752,
      "learning_rate": 1.5250559649731739e-05,
      "loss": 1.4097,
      "step": 8269
    },
    {
      "epoch": 1.4897545597838324,
      "grad_norm": 1.5416224002838135,
      "learning_rate": 1.5240378616267886e-05,
      "loss": 1.4881,
      "step": 8270
    },
    {
      "epoch": 1.4899346993920288,
      "grad_norm": 1.4898866415023804,
      "learning_rate": 1.5230200371203934e-05,
      "loss": 1.5489,
      "step": 8271
    },
    {
      "epoch": 1.4901148390002252,
      "grad_norm": 1.4133309125900269,
      "learning_rate": 1.5220024915356369e-05,
      "loss": 1.2889,
      "step": 8272
    },
    {
      "epoch": 1.4902949786084214,
      "grad_norm": 1.4658639430999756,
      "learning_rate": 1.5209852249541468e-05,
      "loss": 1.5949,
      "step": 8273
    },
    {
      "epoch": 1.4904751182166178,
      "grad_norm": 1.3493504524230957,
      "learning_rate": 1.519968237457527e-05,
      "loss": 1.2713,
      "step": 8274
    },
    {
      "epoch": 1.4906552578248142,
      "grad_norm": 1.4170231819152832,
      "learning_rate": 1.5189515291273604e-05,
      "loss": 1.3645,
      "step": 8275
    },
    {
      "epoch": 1.4908353974330106,
      "grad_norm": 1.4936116933822632,
      "learning_rate": 1.5179351000452046e-05,
      "loss": 1.4379,
      "step": 8276
    },
    {
      "epoch": 1.491015537041207,
      "grad_norm": 1.4865164756774902,
      "learning_rate": 1.5169189502926008e-05,
      "loss": 1.5397,
      "step": 8277
    },
    {
      "epoch": 1.4911956766494032,
      "grad_norm": 1.4955211877822876,
      "learning_rate": 1.5159030799510627e-05,
      "loss": 1.3052,
      "step": 8278
    },
    {
      "epoch": 1.4913758162575996,
      "grad_norm": 1.407193899154663,
      "learning_rate": 1.514887489102082e-05,
      "loss": 1.3724,
      "step": 8279
    },
    {
      "epoch": 1.491555955865796,
      "grad_norm": 1.3906277418136597,
      "learning_rate": 1.5138721778271292e-05,
      "loss": 1.2316,
      "step": 8280
    },
    {
      "epoch": 1.4917360954739922,
      "grad_norm": 1.5691041946411133,
      "learning_rate": 1.5128571462076524e-05,
      "loss": 1.4848,
      "step": 8281
    },
    {
      "epoch": 1.4919162350821886,
      "grad_norm": 1.4598407745361328,
      "learning_rate": 1.5118423943250771e-05,
      "loss": 1.3737,
      "step": 8282
    },
    {
      "epoch": 1.492096374690385,
      "grad_norm": 1.4482777118682861,
      "learning_rate": 1.5108279222608058e-05,
      "loss": 1.2949,
      "step": 8283
    },
    {
      "epoch": 1.4922765142985814,
      "grad_norm": 1.4670007228851318,
      "learning_rate": 1.509813730096219e-05,
      "loss": 1.4747,
      "step": 8284
    },
    {
      "epoch": 1.4924566539067778,
      "grad_norm": 1.4364105463027954,
      "learning_rate": 1.5087998179126733e-05,
      "loss": 1.2779,
      "step": 8285
    },
    {
      "epoch": 1.492636793514974,
      "grad_norm": 1.527872085571289,
      "learning_rate": 1.5077861857915071e-05,
      "loss": 1.4822,
      "step": 8286
    },
    {
      "epoch": 1.4928169331231704,
      "grad_norm": 1.6420036554336548,
      "learning_rate": 1.5067728338140324e-05,
      "loss": 1.4718,
      "step": 8287
    },
    {
      "epoch": 1.4929970727313668,
      "grad_norm": 1.4591742753982544,
      "learning_rate": 1.5057597620615394e-05,
      "loss": 1.1637,
      "step": 8288
    },
    {
      "epoch": 1.4931772123395632,
      "grad_norm": 1.4519966840744019,
      "learning_rate": 1.5047469706152962e-05,
      "loss": 1.3354,
      "step": 8289
    },
    {
      "epoch": 1.4933573519477594,
      "grad_norm": 1.5276570320129395,
      "learning_rate": 1.5037344595565467e-05,
      "loss": 1.2669,
      "step": 8290
    },
    {
      "epoch": 1.4935374915559558,
      "grad_norm": 1.4279955625534058,
      "learning_rate": 1.5027222289665194e-05,
      "loss": 1.1563,
      "step": 8291
    },
    {
      "epoch": 1.4937176311641522,
      "grad_norm": 1.5719612836837769,
      "learning_rate": 1.5017102789264099e-05,
      "loss": 1.5391,
      "step": 8292
    },
    {
      "epoch": 1.4938977707723486,
      "grad_norm": 1.5116100311279297,
      "learning_rate": 1.5006986095173974e-05,
      "loss": 1.4209,
      "step": 8293
    },
    {
      "epoch": 1.494077910380545,
      "grad_norm": 1.625718593597412,
      "learning_rate": 1.499687220820638e-05,
      "loss": 1.6656,
      "step": 8294
    },
    {
      "epoch": 1.4942580499887412,
      "grad_norm": 1.6613056659698486,
      "learning_rate": 1.4986761129172638e-05,
      "loss": 1.4554,
      "step": 8295
    },
    {
      "epoch": 1.4944381895969376,
      "grad_norm": 1.6278865337371826,
      "learning_rate": 1.4976652858883877e-05,
      "loss": 1.4606,
      "step": 8296
    },
    {
      "epoch": 1.494618329205134,
      "grad_norm": 1.567407250404358,
      "learning_rate": 1.4966547398150966e-05,
      "loss": 1.3358,
      "step": 8297
    },
    {
      "epoch": 1.4947984688133302,
      "grad_norm": 1.6071711778640747,
      "learning_rate": 1.4956444747784554e-05,
      "loss": 1.5608,
      "step": 8298
    },
    {
      "epoch": 1.4949786084215266,
      "grad_norm": 1.6230558156967163,
      "learning_rate": 1.4946344908595061e-05,
      "loss": 1.5254,
      "step": 8299
    },
    {
      "epoch": 1.495158748029723,
      "grad_norm": 1.419877529144287,
      "learning_rate": 1.4936247881392723e-05,
      "loss": 1.1523,
      "step": 8300
    },
    {
      "epoch": 1.4953388876379194,
      "grad_norm": 1.37412691116333,
      "learning_rate": 1.4926153666987503e-05,
      "loss": 1.1959,
      "step": 8301
    },
    {
      "epoch": 1.4955190272461159,
      "grad_norm": 1.4982872009277344,
      "learning_rate": 1.4916062266189145e-05,
      "loss": 2.0673,
      "step": 8302
    },
    {
      "epoch": 1.495699166854312,
      "grad_norm": 1.4154436588287354,
      "learning_rate": 1.4905973679807205e-05,
      "loss": 1.9184,
      "step": 8303
    },
    {
      "epoch": 1.4958793064625084,
      "grad_norm": 1.2990942001342773,
      "learning_rate": 1.489588790865093e-05,
      "loss": 1.8955,
      "step": 8304
    },
    {
      "epoch": 1.4960594460707048,
      "grad_norm": 1.3932422399520874,
      "learning_rate": 1.4885804953529443e-05,
      "loss": 1.8519,
      "step": 8305
    },
    {
      "epoch": 1.496239585678901,
      "grad_norm": 1.3909327983856201,
      "learning_rate": 1.487572481525158e-05,
      "loss": 1.8113,
      "step": 8306
    },
    {
      "epoch": 1.4964197252870974,
      "grad_norm": 1.4670466184616089,
      "learning_rate": 1.4865647494625961e-05,
      "loss": 1.8791,
      "step": 8307
    },
    {
      "epoch": 1.4965998648952938,
      "grad_norm": 1.5743836164474487,
      "learning_rate": 1.4855572992460992e-05,
      "loss": 1.7762,
      "step": 8308
    },
    {
      "epoch": 1.4967800045034902,
      "grad_norm": 1.6498053073883057,
      "learning_rate": 1.484550130956482e-05,
      "loss": 2.1521,
      "step": 8309
    },
    {
      "epoch": 1.4969601441116867,
      "grad_norm": 1.7136973142623901,
      "learning_rate": 1.4835432446745428e-05,
      "loss": 1.7715,
      "step": 8310
    },
    {
      "epoch": 1.4971402837198828,
      "grad_norm": 2.12508225440979,
      "learning_rate": 1.4825366404810514e-05,
      "loss": 2.0838,
      "step": 8311
    },
    {
      "epoch": 1.4973204233280792,
      "grad_norm": 1.7046000957489014,
      "learning_rate": 1.4815303184567575e-05,
      "loss": 1.6362,
      "step": 8312
    },
    {
      "epoch": 1.4975005629362756,
      "grad_norm": 1.3992172479629517,
      "learning_rate": 1.480524278682388e-05,
      "loss": 1.4775,
      "step": 8313
    },
    {
      "epoch": 1.4976807025444718,
      "grad_norm": 1.513905644416809,
      "learning_rate": 1.4795185212386465e-05,
      "loss": 1.6337,
      "step": 8314
    },
    {
      "epoch": 1.4978608421526682,
      "grad_norm": 1.4843629598617554,
      "learning_rate": 1.4785130462062142e-05,
      "loss": 1.4203,
      "step": 8315
    },
    {
      "epoch": 1.4980409817608646,
      "grad_norm": 1.4706205129623413,
      "learning_rate": 1.47750785366575e-05,
      "loss": 1.1799,
      "step": 8316
    },
    {
      "epoch": 1.498221121369061,
      "grad_norm": 1.5224649906158447,
      "learning_rate": 1.47650294369789e-05,
      "loss": 1.3566,
      "step": 8317
    },
    {
      "epoch": 1.4984012609772575,
      "grad_norm": 1.3990535736083984,
      "learning_rate": 1.4754983163832464e-05,
      "loss": 1.4349,
      "step": 8318
    },
    {
      "epoch": 1.4985814005854539,
      "grad_norm": 1.396898865699768,
      "learning_rate": 1.4744939718024126e-05,
      "loss": 1.2324,
      "step": 8319
    },
    {
      "epoch": 1.49876154019365,
      "grad_norm": 1.420369267463684,
      "learning_rate": 1.4734899100359551e-05,
      "loss": 1.4519,
      "step": 8320
    },
    {
      "epoch": 1.4989416798018464,
      "grad_norm": 1.3734114170074463,
      "learning_rate": 1.4724861311644189e-05,
      "loss": 1.4056,
      "step": 8321
    },
    {
      "epoch": 1.4991218194100429,
      "grad_norm": 1.3760509490966797,
      "learning_rate": 1.4714826352683275e-05,
      "loss": 1.3771,
      "step": 8322
    },
    {
      "epoch": 1.499301959018239,
      "grad_norm": 1.2371797561645508,
      "learning_rate": 1.470479422428181e-05,
      "loss": 1.2243,
      "step": 8323
    },
    {
      "epoch": 1.4994820986264354,
      "grad_norm": 1.5186550617218018,
      "learning_rate": 1.4694764927244553e-05,
      "loss": 1.6891,
      "step": 8324
    },
    {
      "epoch": 1.4996622382346319,
      "grad_norm": 1.4722535610198975,
      "learning_rate": 1.4684738462376057e-05,
      "loss": 1.3139,
      "step": 8325
    },
    {
      "epoch": 1.4998423778428283,
      "grad_norm": 1.4951199293136597,
      "learning_rate": 1.467471483048064e-05,
      "loss": 1.4969,
      "step": 8326
    },
    {
      "epoch": 1.5000225174510247,
      "grad_norm": 1.3578238487243652,
      "learning_rate": 1.466469403236238e-05,
      "loss": 1.3798,
      "step": 8327
    },
    {
      "epoch": 1.5002026570592208,
      "grad_norm": 1.335727334022522,
      "learning_rate": 1.4654676068825169e-05,
      "loss": 1.2296,
      "step": 8328
    },
    {
      "epoch": 1.5003827966674173,
      "grad_norm": 1.554796576499939,
      "learning_rate": 1.4644660940672627e-05,
      "loss": 1.2578,
      "step": 8329
    },
    {
      "epoch": 1.5005629362756134,
      "grad_norm": 1.4066554307937622,
      "learning_rate": 1.4634648648708165e-05,
      "loss": 1.3122,
      "step": 8330
    },
    {
      "epoch": 1.5007430758838098,
      "grad_norm": 1.4520604610443115,
      "learning_rate": 1.4624639193734962e-05,
      "loss": 1.5269,
      "step": 8331
    },
    {
      "epoch": 1.5009232154920062,
      "grad_norm": 1.507185697555542,
      "learning_rate": 1.4614632576555954e-05,
      "loss": 1.3771,
      "step": 8332
    },
    {
      "epoch": 1.5011033551002027,
      "grad_norm": 1.4310791492462158,
      "learning_rate": 1.46046287979739e-05,
      "loss": 1.2709,
      "step": 8333
    },
    {
      "epoch": 1.501283494708399,
      "grad_norm": 1.4253461360931396,
      "learning_rate": 1.4594627858791287e-05,
      "loss": 1.3773,
      "step": 8334
    },
    {
      "epoch": 1.5014636343165955,
      "grad_norm": 1.449432134628296,
      "learning_rate": 1.458462975981038e-05,
      "loss": 1.4128,
      "step": 8335
    },
    {
      "epoch": 1.5016437739247919,
      "grad_norm": 1.4495176076889038,
      "learning_rate": 1.4574634501833224e-05,
      "loss": 1.3953,
      "step": 8336
    },
    {
      "epoch": 1.501823913532988,
      "grad_norm": 1.549580693244934,
      "learning_rate": 1.4564642085661634e-05,
      "loss": 1.5367,
      "step": 8337
    },
    {
      "epoch": 1.5020040531411845,
      "grad_norm": 1.4680416584014893,
      "learning_rate": 1.4554652512097194e-05,
      "loss": 1.3608,
      "step": 8338
    },
    {
      "epoch": 1.5021841927493806,
      "grad_norm": 1.7308632135391235,
      "learning_rate": 1.4544665781941263e-05,
      "loss": 1.5143,
      "step": 8339
    },
    {
      "epoch": 1.502364332357577,
      "grad_norm": 1.5684367418289185,
      "learning_rate": 1.4534681895994978e-05,
      "loss": 1.2402,
      "step": 8340
    },
    {
      "epoch": 1.5025444719657735,
      "grad_norm": 1.4874870777130127,
      "learning_rate": 1.452470085505922e-05,
      "loss": 1.3738,
      "step": 8341
    },
    {
      "epoch": 1.5027246115739699,
      "grad_norm": 1.5809177160263062,
      "learning_rate": 1.4514722659934693e-05,
      "loss": 1.3483,
      "step": 8342
    },
    {
      "epoch": 1.5029047511821663,
      "grad_norm": 1.3894163370132446,
      "learning_rate": 1.4504747311421835e-05,
      "loss": 1.3323,
      "step": 8343
    },
    {
      "epoch": 1.5030848907903627,
      "grad_norm": 1.480005145072937,
      "learning_rate": 1.4494774810320855e-05,
      "loss": 1.3526,
      "step": 8344
    },
    {
      "epoch": 1.5032650303985589,
      "grad_norm": 1.4905613660812378,
      "learning_rate": 1.4484805157431747e-05,
      "loss": 1.2468,
      "step": 8345
    },
    {
      "epoch": 1.5034451700067553,
      "grad_norm": 1.3716827630996704,
      "learning_rate": 1.4474838353554266e-05,
      "loss": 1.3311,
      "step": 8346
    },
    {
      "epoch": 1.5036253096149514,
      "grad_norm": 1.5514984130859375,
      "learning_rate": 1.4464874399487955e-05,
      "loss": 1.3753,
      "step": 8347
    },
    {
      "epoch": 1.5038054492231478,
      "grad_norm": 1.5787135362625122,
      "learning_rate": 1.4454913296032108e-05,
      "loss": 1.4103,
      "step": 8348
    },
    {
      "epoch": 1.5039855888313443,
      "grad_norm": 1.5407532453536987,
      "learning_rate": 1.4444955043985803e-05,
      "loss": 1.3342,
      "step": 8349
    },
    {
      "epoch": 1.5041657284395407,
      "grad_norm": 1.5277503728866577,
      "learning_rate": 1.4434999644147867e-05,
      "loss": 1.1601,
      "step": 8350
    },
    {
      "epoch": 1.504345868047737,
      "grad_norm": 1.4100755453109741,
      "learning_rate": 1.4425047097316957e-05,
      "loss": 1.1101,
      "step": 8351
    },
    {
      "epoch": 1.5045260076559335,
      "grad_norm": 1.469811201095581,
      "learning_rate": 1.4415097404291438e-05,
      "loss": 1.9185,
      "step": 8352
    },
    {
      "epoch": 1.5047061472641297,
      "grad_norm": 1.3146687746047974,
      "learning_rate": 1.4405150565869473e-05,
      "loss": 1.6853,
      "step": 8353
    },
    {
      "epoch": 1.504886286872326,
      "grad_norm": 1.4314987659454346,
      "learning_rate": 1.4395206582848997e-05,
      "loss": 2.0995,
      "step": 8354
    },
    {
      "epoch": 1.5050664264805222,
      "grad_norm": 1.3999438285827637,
      "learning_rate": 1.4385265456027685e-05,
      "loss": 1.7929,
      "step": 8355
    },
    {
      "epoch": 1.5052465660887187,
      "grad_norm": 1.340320348739624,
      "learning_rate": 1.4375327186203063e-05,
      "loss": 1.7079,
      "step": 8356
    },
    {
      "epoch": 1.505426705696915,
      "grad_norm": 1.5397825241088867,
      "learning_rate": 1.4365391774172327e-05,
      "loss": 1.5989,
      "step": 8357
    },
    {
      "epoch": 1.5056068453051115,
      "grad_norm": 1.6149042844772339,
      "learning_rate": 1.435545922073251e-05,
      "loss": 1.9178,
      "step": 8358
    },
    {
      "epoch": 1.5057869849133079,
      "grad_norm": 1.5382217168807983,
      "learning_rate": 1.4345529526680385e-05,
      "loss": 1.7802,
      "step": 8359
    },
    {
      "epoch": 1.5059671245215043,
      "grad_norm": 1.7813255786895752,
      "learning_rate": 1.4335602692812506e-05,
      "loss": 2.0355,
      "step": 8360
    },
    {
      "epoch": 1.5061472641297007,
      "grad_norm": 1.8131059408187866,
      "learning_rate": 1.4325678719925217e-05,
      "loss": 1.9474,
      "step": 8361
    },
    {
      "epoch": 1.5063274037378969,
      "grad_norm": 1.4389322996139526,
      "learning_rate": 1.4315757608814613e-05,
      "loss": 1.4792,
      "step": 8362
    },
    {
      "epoch": 1.5065075433460933,
      "grad_norm": 1.405253291130066,
      "learning_rate": 1.4305839360276541e-05,
      "loss": 1.6627,
      "step": 8363
    },
    {
      "epoch": 1.5066876829542895,
      "grad_norm": 1.3129521608352661,
      "learning_rate": 1.429592397510664e-05,
      "loss": 1.4173,
      "step": 8364
    },
    {
      "epoch": 1.5068678225624859,
      "grad_norm": 1.51478910446167,
      "learning_rate": 1.4286011454100345e-05,
      "loss": 1.7013,
      "step": 8365
    },
    {
      "epoch": 1.5070479621706823,
      "grad_norm": 1.4286258220672607,
      "learning_rate": 1.427610179805281e-05,
      "loss": 1.5803,
      "step": 8366
    },
    {
      "epoch": 1.5072281017788787,
      "grad_norm": 1.5034891366958618,
      "learning_rate": 1.4266195007758992e-05,
      "loss": 1.4936,
      "step": 8367
    },
    {
      "epoch": 1.507408241387075,
      "grad_norm": 1.4833680391311646,
      "learning_rate": 1.4256291084013623e-05,
      "loss": 1.498,
      "step": 8368
    },
    {
      "epoch": 1.5075883809952715,
      "grad_norm": 1.495717167854309,
      "learning_rate": 1.424639002761114e-05,
      "loss": 1.5051,
      "step": 8369
    },
    {
      "epoch": 1.5077685206034677,
      "grad_norm": 1.3419215679168701,
      "learning_rate": 1.4236491839345844e-05,
      "loss": 1.3533,
      "step": 8370
    },
    {
      "epoch": 1.507948660211664,
      "grad_norm": 1.4875454902648926,
      "learning_rate": 1.4226596520011759e-05,
      "loss": 1.6129,
      "step": 8371
    },
    {
      "epoch": 1.5081287998198603,
      "grad_norm": 1.600865125656128,
      "learning_rate": 1.4216704070402675e-05,
      "loss": 1.6304,
      "step": 8372
    },
    {
      "epoch": 1.5083089394280567,
      "grad_norm": 1.4899908304214478,
      "learning_rate": 1.420681449131216e-05,
      "loss": 1.2069,
      "step": 8373
    },
    {
      "epoch": 1.508489079036253,
      "grad_norm": 1.4615834951400757,
      "learning_rate": 1.4196927783533537e-05,
      "loss": 1.4899,
      "step": 8374
    },
    {
      "epoch": 1.5086692186444495,
      "grad_norm": 1.5631928443908691,
      "learning_rate": 1.4187043947859945e-05,
      "loss": 1.5518,
      "step": 8375
    },
    {
      "epoch": 1.5088493582526459,
      "grad_norm": 1.4729647636413574,
      "learning_rate": 1.4177162985084242e-05,
      "loss": 1.4203,
      "step": 8376
    },
    {
      "epoch": 1.5090294978608423,
      "grad_norm": 1.4309210777282715,
      "learning_rate": 1.4167284895999079e-05,
      "loss": 1.4526,
      "step": 8377
    },
    {
      "epoch": 1.5092096374690385,
      "grad_norm": 1.4715665578842163,
      "learning_rate": 1.4157409681396866e-05,
      "loss": 1.3414,
      "step": 8378
    },
    {
      "epoch": 1.5093897770772349,
      "grad_norm": 1.4777817726135254,
      "learning_rate": 1.414753734206979e-05,
      "loss": 1.5413,
      "step": 8379
    },
    {
      "epoch": 1.509569916685431,
      "grad_norm": 1.5771071910858154,
      "learning_rate": 1.4137667878809802e-05,
      "loss": 1.3945,
      "step": 8380
    },
    {
      "epoch": 1.5097500562936275,
      "grad_norm": 1.687414526939392,
      "learning_rate": 1.4127801292408633e-05,
      "loss": 1.5659,
      "step": 8381
    },
    {
      "epoch": 1.5099301959018239,
      "grad_norm": 1.4822789430618286,
      "learning_rate": 1.411793758365777e-05,
      "loss": 1.3918,
      "step": 8382
    },
    {
      "epoch": 1.5101103355100203,
      "grad_norm": 1.4594035148620605,
      "learning_rate": 1.4108076753348465e-05,
      "loss": 1.4652,
      "step": 8383
    },
    {
      "epoch": 1.5102904751182167,
      "grad_norm": 1.484213948249817,
      "learning_rate": 1.4098218802271773e-05,
      "loss": 1.4436,
      "step": 8384
    },
    {
      "epoch": 1.510470614726413,
      "grad_norm": 1.4155118465423584,
      "learning_rate": 1.4088363731218479e-05,
      "loss": 1.2711,
      "step": 8385
    },
    {
      "epoch": 1.5106507543346093,
      "grad_norm": 1.5056843757629395,
      "learning_rate": 1.4078511540979156e-05,
      "loss": 1.4301,
      "step": 8386
    },
    {
      "epoch": 1.5108308939428057,
      "grad_norm": 1.61029851436615,
      "learning_rate": 1.4068662232344143e-05,
      "loss": 1.5778,
      "step": 8387
    },
    {
      "epoch": 1.511011033551002,
      "grad_norm": 1.5662527084350586,
      "learning_rate": 1.4058815806103542e-05,
      "loss": 1.4208,
      "step": 8388
    },
    {
      "epoch": 1.5111911731591983,
      "grad_norm": 1.535174012184143,
      "learning_rate": 1.4048972263047228e-05,
      "loss": 1.6055,
      "step": 8389
    },
    {
      "epoch": 1.5113713127673947,
      "grad_norm": 1.526949405670166,
      "learning_rate": 1.403913160396485e-05,
      "loss": 1.2925,
      "step": 8390
    },
    {
      "epoch": 1.511551452375591,
      "grad_norm": 1.6564172506332397,
      "learning_rate": 1.4029293829645823e-05,
      "loss": 1.5368,
      "step": 8391
    },
    {
      "epoch": 1.5117315919837875,
      "grad_norm": 1.6053402423858643,
      "learning_rate": 1.4019458940879304e-05,
      "loss": 1.477,
      "step": 8392
    },
    {
      "epoch": 1.5119117315919839,
      "grad_norm": 1.4639637470245361,
      "learning_rate": 1.4009626938454279e-05,
      "loss": 1.2375,
      "step": 8393
    },
    {
      "epoch": 1.5120918712001803,
      "grad_norm": 1.4958629608154297,
      "learning_rate": 1.3999797823159455e-05,
      "loss": 1.4628,
      "step": 8394
    },
    {
      "epoch": 1.5122720108083765,
      "grad_norm": 1.4583030939102173,
      "learning_rate": 1.3989971595783313e-05,
      "loss": 1.2396,
      "step": 8395
    },
    {
      "epoch": 1.5124521504165729,
      "grad_norm": 1.4834370613098145,
      "learning_rate": 1.3980148257114107e-05,
      "loss": 1.2876,
      "step": 8396
    },
    {
      "epoch": 1.512632290024769,
      "grad_norm": 1.7045390605926514,
      "learning_rate": 1.3970327807939849e-05,
      "loss": 1.5498,
      "step": 8397
    },
    {
      "epoch": 1.5128124296329655,
      "grad_norm": 1.590372920036316,
      "learning_rate": 1.3960510249048358e-05,
      "loss": 1.3776,
      "step": 8398
    },
    {
      "epoch": 1.5129925692411619,
      "grad_norm": 1.6044691801071167,
      "learning_rate": 1.3950695581227186e-05,
      "loss": 1.2668,
      "step": 8399
    },
    {
      "epoch": 1.5131727088493583,
      "grad_norm": 1.6583843231201172,
      "learning_rate": 1.394088380526365e-05,
      "loss": 1.6151,
      "step": 8400
    },
    {
      "epoch": 1.5133528484575547,
      "grad_norm": 1.5915814638137817,
      "learning_rate": 1.393107492194487e-05,
      "loss": 1.6831,
      "step": 8401
    },
    {
      "epoch": 1.513532988065751,
      "grad_norm": 1.2983510494232178,
      "learning_rate": 1.3921268932057663e-05,
      "loss": 1.7787,
      "step": 8402
    },
    {
      "epoch": 1.5137131276739473,
      "grad_norm": 1.352957844734192,
      "learning_rate": 1.3911465836388699e-05,
      "loss": 1.7723,
      "step": 8403
    },
    {
      "epoch": 1.5138932672821437,
      "grad_norm": 1.4373608827590942,
      "learning_rate": 1.390166563572437e-05,
      "loss": 1.7958,
      "step": 8404
    },
    {
      "epoch": 1.5140734068903399,
      "grad_norm": 1.3368573188781738,
      "learning_rate": 1.3891868330850838e-05,
      "loss": 1.8473,
      "step": 8405
    },
    {
      "epoch": 1.5142535464985363,
      "grad_norm": 1.392109751701355,
      "learning_rate": 1.3882073922554029e-05,
      "loss": 1.8177,
      "step": 8406
    },
    {
      "epoch": 1.5144336861067327,
      "grad_norm": 1.4237622022628784,
      "learning_rate": 1.3872282411619669e-05,
      "loss": 1.5074,
      "step": 8407
    },
    {
      "epoch": 1.514613825714929,
      "grad_norm": 1.5282012224197388,
      "learning_rate": 1.386249379883322e-05,
      "loss": 2.0453,
      "step": 8408
    },
    {
      "epoch": 1.5147939653231255,
      "grad_norm": 1.5520250797271729,
      "learning_rate": 1.3852708084979921e-05,
      "loss": 1.5891,
      "step": 8409
    },
    {
      "epoch": 1.514974104931322,
      "grad_norm": 1.579027771949768,
      "learning_rate": 1.3842925270844764e-05,
      "loss": 1.7107,
      "step": 8410
    },
    {
      "epoch": 1.515154244539518,
      "grad_norm": 1.7695492506027222,
      "learning_rate": 1.3833145357212535e-05,
      "loss": 1.8077,
      "step": 8411
    },
    {
      "epoch": 1.5153343841477145,
      "grad_norm": 1.5642077922821045,
      "learning_rate": 1.3823368344867765e-05,
      "loss": 1.4737,
      "step": 8412
    },
    {
      "epoch": 1.5155145237559107,
      "grad_norm": 1.3876491785049438,
      "learning_rate": 1.3813594234594773e-05,
      "loss": 1.3701,
      "step": 8413
    },
    {
      "epoch": 1.515694663364107,
      "grad_norm": 1.4011150598526,
      "learning_rate": 1.380382302717762e-05,
      "loss": 1.4581,
      "step": 8414
    },
    {
      "epoch": 1.5158748029723035,
      "grad_norm": 1.4258620738983154,
      "learning_rate": 1.3794054723400152e-05,
      "loss": 1.4507,
      "step": 8415
    },
    {
      "epoch": 1.5160549425804999,
      "grad_norm": 1.419507622718811,
      "learning_rate": 1.3784289324045969e-05,
      "loss": 1.2879,
      "step": 8416
    },
    {
      "epoch": 1.5162350821886963,
      "grad_norm": 1.3497275114059448,
      "learning_rate": 1.3774526829898465e-05,
      "loss": 1.1851,
      "step": 8417
    },
    {
      "epoch": 1.5164152217968927,
      "grad_norm": 1.3190562725067139,
      "learning_rate": 1.3764767241740778e-05,
      "loss": 1.3401,
      "step": 8418
    },
    {
      "epoch": 1.516595361405089,
      "grad_norm": 1.534015417098999,
      "learning_rate": 1.3755010560355814e-05,
      "loss": 1.5609,
      "step": 8419
    },
    {
      "epoch": 1.5167755010132853,
      "grad_norm": 1.3188953399658203,
      "learning_rate": 1.3745256786526235e-05,
      "loss": 1.256,
      "step": 8420
    },
    {
      "epoch": 1.5169556406214817,
      "grad_norm": 1.4393165111541748,
      "learning_rate": 1.3735505921034526e-05,
      "loss": 1.2934,
      "step": 8421
    },
    {
      "epoch": 1.5171357802296779,
      "grad_norm": 1.4774068593978882,
      "learning_rate": 1.372575796466285e-05,
      "loss": 1.5429,
      "step": 8422
    },
    {
      "epoch": 1.5173159198378743,
      "grad_norm": 1.5270869731903076,
      "learning_rate": 1.3716012918193205e-05,
      "loss": 1.6515,
      "step": 8423
    },
    {
      "epoch": 1.5174960594460707,
      "grad_norm": 1.541034460067749,
      "learning_rate": 1.3706270782407326e-05,
      "loss": 1.6425,
      "step": 8424
    },
    {
      "epoch": 1.517676199054267,
      "grad_norm": 1.4326303005218506,
      "learning_rate": 1.3696531558086717e-05,
      "loss": 1.3795,
      "step": 8425
    },
    {
      "epoch": 1.5178563386624635,
      "grad_norm": 1.4946346282958984,
      "learning_rate": 1.3686795246012673e-05,
      "loss": 1.4344,
      "step": 8426
    },
    {
      "epoch": 1.51803647827066,
      "grad_norm": 1.567612648010254,
      "learning_rate": 1.3677061846966233e-05,
      "loss": 1.6448,
      "step": 8427
    },
    {
      "epoch": 1.518216617878856,
      "grad_norm": 1.353171944618225,
      "learning_rate": 1.3667331361728197e-05,
      "loss": 1.1792,
      "step": 8428
    },
    {
      "epoch": 1.5183967574870525,
      "grad_norm": 1.4202461242675781,
      "learning_rate": 1.3657603791079143e-05,
      "loss": 1.2336,
      "step": 8429
    },
    {
      "epoch": 1.5185768970952487,
      "grad_norm": 1.4545245170593262,
      "learning_rate": 1.3647879135799396e-05,
      "loss": 1.4073,
      "step": 8430
    },
    {
      "epoch": 1.518757036703445,
      "grad_norm": 1.5074716806411743,
      "learning_rate": 1.3638157396669088e-05,
      "loss": 1.5228,
      "step": 8431
    },
    {
      "epoch": 1.5189371763116415,
      "grad_norm": 1.3752111196517944,
      "learning_rate": 1.362843857446809e-05,
      "loss": 1.4161,
      "step": 8432
    },
    {
      "epoch": 1.519117315919838,
      "grad_norm": 1.363057255744934,
      "learning_rate": 1.3618722669976047e-05,
      "loss": 1.286,
      "step": 8433
    },
    {
      "epoch": 1.5192974555280343,
      "grad_norm": 1.432416319847107,
      "learning_rate": 1.3609009683972318e-05,
      "loss": 1.1755,
      "step": 8434
    },
    {
      "epoch": 1.5194775951362307,
      "grad_norm": 1.5536136627197266,
      "learning_rate": 1.3599299617236122e-05,
      "loss": 1.5197,
      "step": 8435
    },
    {
      "epoch": 1.519657734744427,
      "grad_norm": 1.5242769718170166,
      "learning_rate": 1.3589592470546381e-05,
      "loss": 1.4136,
      "step": 8436
    },
    {
      "epoch": 1.5198378743526233,
      "grad_norm": 1.5510059595108032,
      "learning_rate": 1.3579888244681798e-05,
      "loss": 1.3505,
      "step": 8437
    },
    {
      "epoch": 1.5200180139608195,
      "grad_norm": 1.4648244380950928,
      "learning_rate": 1.3570186940420836e-05,
      "loss": 1.2745,
      "step": 8438
    },
    {
      "epoch": 1.5201981535690159,
      "grad_norm": 1.4273651838302612,
      "learning_rate": 1.3560488558541724e-05,
      "loss": 1.2076,
      "step": 8439
    },
    {
      "epoch": 1.5203782931772123,
      "grad_norm": 1.5484436750411987,
      "learning_rate": 1.3550793099822484e-05,
      "loss": 1.452,
      "step": 8440
    },
    {
      "epoch": 1.5205584327854087,
      "grad_norm": 1.5063382387161255,
      "learning_rate": 1.3541100565040871e-05,
      "loss": 1.246,
      "step": 8441
    },
    {
      "epoch": 1.520738572393605,
      "grad_norm": 1.4981944561004639,
      "learning_rate": 1.3531410954974406e-05,
      "loss": 1.2674,
      "step": 8442
    },
    {
      "epoch": 1.5209187120018015,
      "grad_norm": 1.6496658325195312,
      "learning_rate": 1.3521724270400393e-05,
      "loss": 1.5407,
      "step": 8443
    },
    {
      "epoch": 1.5210988516099977,
      "grad_norm": 1.4470162391662598,
      "learning_rate": 1.3512040512095891e-05,
      "loss": 1.412,
      "step": 8444
    },
    {
      "epoch": 1.521278991218194,
      "grad_norm": 1.5615785121917725,
      "learning_rate": 1.3502359680837728e-05,
      "loss": 1.4588,
      "step": 8445
    },
    {
      "epoch": 1.5214591308263905,
      "grad_norm": 1.448725938796997,
      "learning_rate": 1.3492681777402494e-05,
      "loss": 1.1237,
      "step": 8446
    },
    {
      "epoch": 1.5216392704345867,
      "grad_norm": 1.5661933422088623,
      "learning_rate": 1.3483006802566544e-05,
      "loss": 1.2942,
      "step": 8447
    },
    {
      "epoch": 1.521819410042783,
      "grad_norm": 1.6213743686676025,
      "learning_rate": 1.3473334757105988e-05,
      "loss": 1.3139,
      "step": 8448
    },
    {
      "epoch": 1.5219995496509795,
      "grad_norm": 1.6604522466659546,
      "learning_rate": 1.3463665641796741e-05,
      "loss": 1.2844,
      "step": 8449
    },
    {
      "epoch": 1.522179689259176,
      "grad_norm": 1.5329192876815796,
      "learning_rate": 1.345399945741444e-05,
      "loss": 1.2786,
      "step": 8450
    },
    {
      "epoch": 1.5223598288673723,
      "grad_norm": 1.6042360067367554,
      "learning_rate": 1.3444336204734504e-05,
      "loss": 1.3243,
      "step": 8451
    },
    {
      "epoch": 1.5225399684755687,
      "grad_norm": 1.5694210529327393,
      "learning_rate": 1.343467588453211e-05,
      "loss": 2.1818,
      "step": 8452
    },
    {
      "epoch": 1.522720108083765,
      "grad_norm": 1.4153255224227905,
      "learning_rate": 1.3425018497582204e-05,
      "loss": 1.6971,
      "step": 8453
    },
    {
      "epoch": 1.5229002476919613,
      "grad_norm": 1.4332348108291626,
      "learning_rate": 1.34153640446595e-05,
      "loss": 1.9152,
      "step": 8454
    },
    {
      "epoch": 1.5230803873001575,
      "grad_norm": 1.4661483764648438,
      "learning_rate": 1.3405712526538478e-05,
      "loss": 1.9448,
      "step": 8455
    },
    {
      "epoch": 1.523260526908354,
      "grad_norm": 1.3632382154464722,
      "learning_rate": 1.3396063943993365e-05,
      "loss": 1.4259,
      "step": 8456
    },
    {
      "epoch": 1.5234406665165503,
      "grad_norm": 1.4935468435287476,
      "learning_rate": 1.3386418297798165e-05,
      "loss": 1.5784,
      "step": 8457
    },
    {
      "epoch": 1.5236208061247467,
      "grad_norm": 1.4813027381896973,
      "learning_rate": 1.3376775588726664e-05,
      "loss": 2.0049,
      "step": 8458
    },
    {
      "epoch": 1.5238009457329431,
      "grad_norm": 1.437522530555725,
      "learning_rate": 1.336713581755239e-05,
      "loss": 1.6885,
      "step": 8459
    },
    {
      "epoch": 1.5239810853411395,
      "grad_norm": 1.6935616731643677,
      "learning_rate": 1.3357498985048638e-05,
      "loss": 2.0892,
      "step": 8460
    },
    {
      "epoch": 1.5241612249493357,
      "grad_norm": 1.7641370296478271,
      "learning_rate": 1.3347865091988465e-05,
      "loss": 1.8358,
      "step": 8461
    },
    {
      "epoch": 1.524341364557532,
      "grad_norm": 1.5443445444107056,
      "learning_rate": 1.333823413914469e-05,
      "loss": 1.4988,
      "step": 8462
    },
    {
      "epoch": 1.5245215041657283,
      "grad_norm": 1.286849856376648,
      "learning_rate": 1.3328606127289934e-05,
      "loss": 1.2725,
      "step": 8463
    },
    {
      "epoch": 1.5247016437739247,
      "grad_norm": 1.5175888538360596,
      "learning_rate": 1.3318981057196529e-05,
      "loss": 1.5116,
      "step": 8464
    },
    {
      "epoch": 1.524881783382121,
      "grad_norm": 1.5164495706558228,
      "learning_rate": 1.3309358929636612e-05,
      "loss": 1.4129,
      "step": 8465
    },
    {
      "epoch": 1.5250619229903175,
      "grad_norm": 1.4837632179260254,
      "learning_rate": 1.3299739745382033e-05,
      "loss": 1.2452,
      "step": 8466
    },
    {
      "epoch": 1.525242062598514,
      "grad_norm": 1.3592618703842163,
      "learning_rate": 1.3290123505204444e-05,
      "loss": 1.3376,
      "step": 8467
    },
    {
      "epoch": 1.5254222022067103,
      "grad_norm": 1.4977359771728516,
      "learning_rate": 1.328051020987528e-05,
      "loss": 1.5322,
      "step": 8468
    },
    {
      "epoch": 1.5256023418149065,
      "grad_norm": 1.494917631149292,
      "learning_rate": 1.3270899860165698e-05,
      "loss": 1.6695,
      "step": 8469
    },
    {
      "epoch": 1.525782481423103,
      "grad_norm": 1.4956611394882202,
      "learning_rate": 1.3261292456846647e-05,
      "loss": 1.4165,
      "step": 8470
    },
    {
      "epoch": 1.525962621031299,
      "grad_norm": 1.5039005279541016,
      "learning_rate": 1.32516880006888e-05,
      "loss": 1.5244,
      "step": 8471
    },
    {
      "epoch": 1.5261427606394955,
      "grad_norm": 1.2097077369689941,
      "learning_rate": 1.3242086492462658e-05,
      "loss": 1.0662,
      "step": 8472
    },
    {
      "epoch": 1.526322900247692,
      "grad_norm": 1.3735398054122925,
      "learning_rate": 1.3232487932938431e-05,
      "loss": 1.3092,
      "step": 8473
    },
    {
      "epoch": 1.5265030398558883,
      "grad_norm": 1.441636562347412,
      "learning_rate": 1.3222892322886115e-05,
      "loss": 1.5211,
      "step": 8474
    },
    {
      "epoch": 1.5266831794640847,
      "grad_norm": 1.4389063119888306,
      "learning_rate": 1.3213299663075468e-05,
      "loss": 1.3159,
      "step": 8475
    },
    {
      "epoch": 1.5268633190722811,
      "grad_norm": 1.383254051208496,
      "learning_rate": 1.3203709954276e-05,
      "loss": 1.2766,
      "step": 8476
    },
    {
      "epoch": 1.5270434586804775,
      "grad_norm": 1.38627290725708,
      "learning_rate": 1.3194123197256996e-05,
      "loss": 1.3993,
      "step": 8477
    },
    {
      "epoch": 1.5272235982886737,
      "grad_norm": 1.471824049949646,
      "learning_rate": 1.3184539392787505e-05,
      "loss": 1.4703,
      "step": 8478
    },
    {
      "epoch": 1.5274037378968701,
      "grad_norm": 1.5234025716781616,
      "learning_rate": 1.3174958541636328e-05,
      "loss": 1.4725,
      "step": 8479
    },
    {
      "epoch": 1.5275838775050663,
      "grad_norm": 1.4237037897109985,
      "learning_rate": 1.316538064457204e-05,
      "loss": 1.4489,
      "step": 8480
    },
    {
      "epoch": 1.5277640171132627,
      "grad_norm": 1.7419742345809937,
      "learning_rate": 1.3155805702362961e-05,
      "loss": 1.6456,
      "step": 8481
    },
    {
      "epoch": 1.5279441567214591,
      "grad_norm": 1.4707708358764648,
      "learning_rate": 1.3146233715777218e-05,
      "loss": 1.3318,
      "step": 8482
    },
    {
      "epoch": 1.5281242963296555,
      "grad_norm": 1.4871906042099,
      "learning_rate": 1.3136664685582656e-05,
      "loss": 1.375,
      "step": 8483
    },
    {
      "epoch": 1.528304435937852,
      "grad_norm": 1.5011922121047974,
      "learning_rate": 1.3127098612546895e-05,
      "loss": 1.5781,
      "step": 8484
    },
    {
      "epoch": 1.5284845755460483,
      "grad_norm": 1.263919472694397,
      "learning_rate": 1.3117535497437327e-05,
      "loss": 1.1398,
      "step": 8485
    },
    {
      "epoch": 1.5286647151542445,
      "grad_norm": 1.6035559177398682,
      "learning_rate": 1.3107975341021094e-05,
      "loss": 1.4106,
      "step": 8486
    },
    {
      "epoch": 1.528844854762441,
      "grad_norm": 1.5318305492401123,
      "learning_rate": 1.3098418144065111e-05,
      "loss": 1.456,
      "step": 8487
    },
    {
      "epoch": 1.529024994370637,
      "grad_norm": 1.5810279846191406,
      "learning_rate": 1.3088863907336047e-05,
      "loss": 1.6726,
      "step": 8488
    },
    {
      "epoch": 1.5292051339788335,
      "grad_norm": 1.5620367527008057,
      "learning_rate": 1.3079312631600344e-05,
      "loss": 1.4859,
      "step": 8489
    },
    {
      "epoch": 1.52938527358703,
      "grad_norm": 1.5283035039901733,
      "learning_rate": 1.3069764317624183e-05,
      "loss": 1.3507,
      "step": 8490
    },
    {
      "epoch": 1.5295654131952263,
      "grad_norm": 1.5750842094421387,
      "learning_rate": 1.3060218966173554e-05,
      "loss": 1.4837,
      "step": 8491
    },
    {
      "epoch": 1.5297455528034227,
      "grad_norm": 1.3221359252929688,
      "learning_rate": 1.305067657801417e-05,
      "loss": 1.1077,
      "step": 8492
    },
    {
      "epoch": 1.5299256924116191,
      "grad_norm": 1.6749347448349,
      "learning_rate": 1.3041137153911515e-05,
      "loss": 1.4105,
      "step": 8493
    },
    {
      "epoch": 1.5301058320198153,
      "grad_norm": 1.623597264289856,
      "learning_rate": 1.303160069463083e-05,
      "loss": 1.3625,
      "step": 8494
    },
    {
      "epoch": 1.5302859716280117,
      "grad_norm": 1.5514707565307617,
      "learning_rate": 1.3022067200937116e-05,
      "loss": 1.5256,
      "step": 8495
    },
    {
      "epoch": 1.530466111236208,
      "grad_norm": 1.5476737022399902,
      "learning_rate": 1.3012536673595177e-05,
      "loss": 1.3902,
      "step": 8496
    },
    {
      "epoch": 1.5306462508444043,
      "grad_norm": 1.446799635887146,
      "learning_rate": 1.300300911336953e-05,
      "loss": 1.3214,
      "step": 8497
    },
    {
      "epoch": 1.5308263904526007,
      "grad_norm": 1.6114540100097656,
      "learning_rate": 1.299348452102449e-05,
      "loss": 1.3176,
      "step": 8498
    },
    {
      "epoch": 1.5310065300607971,
      "grad_norm": 1.3178235292434692,
      "learning_rate": 1.2983962897324059e-05,
      "loss": 1.1009,
      "step": 8499
    },
    {
      "epoch": 1.5311866696689935,
      "grad_norm": 1.5109543800354004,
      "learning_rate": 1.2974444243032114e-05,
      "loss": 1.2304,
      "step": 8500
    },
    {
      "epoch": 1.53136680927719,
      "grad_norm": 1.3390576839447021,
      "learning_rate": 1.296492855891222e-05,
      "loss": 1.1027,
      "step": 8501
    },
    {
      "epoch": 1.5315469488853863,
      "grad_norm": 1.2594733238220215,
      "learning_rate": 1.2955415845727714e-05,
      "loss": 1.5455,
      "step": 8502
    },
    {
      "epoch": 1.5317270884935825,
      "grad_norm": 1.3651126623153687,
      "learning_rate": 1.2945906104241706e-05,
      "loss": 1.8172,
      "step": 8503
    },
    {
      "epoch": 1.531907228101779,
      "grad_norm": 1.351478099822998,
      "learning_rate": 1.2936399335217042e-05,
      "loss": 1.7814,
      "step": 8504
    },
    {
      "epoch": 1.5320873677099751,
      "grad_norm": 1.3493454456329346,
      "learning_rate": 1.2926895539416394e-05,
      "loss": 1.4755,
      "step": 8505
    },
    {
      "epoch": 1.5322675073181715,
      "grad_norm": 1.450313687324524,
      "learning_rate": 1.2917394717602121e-05,
      "loss": 1.7402,
      "step": 8506
    },
    {
      "epoch": 1.532447646926368,
      "grad_norm": 1.495362639427185,
      "learning_rate": 1.2907896870536385e-05,
      "loss": 1.9777,
      "step": 8507
    },
    {
      "epoch": 1.5326277865345643,
      "grad_norm": 1.3716552257537842,
      "learning_rate": 1.2898401998981096e-05,
      "loss": 1.6616,
      "step": 8508
    },
    {
      "epoch": 1.5328079261427607,
      "grad_norm": 1.6226818561553955,
      "learning_rate": 1.2888910103697926e-05,
      "loss": 1.8247,
      "step": 8509
    },
    {
      "epoch": 1.5329880657509571,
      "grad_norm": 1.6547714471817017,
      "learning_rate": 1.2879421185448309e-05,
      "loss": 1.5338,
      "step": 8510
    },
    {
      "epoch": 1.5331682053591533,
      "grad_norm": 1.6544560194015503,
      "learning_rate": 1.2869935244993447e-05,
      "loss": 1.8283,
      "step": 8511
    },
    {
      "epoch": 1.5333483449673497,
      "grad_norm": 2.0752575397491455,
      "learning_rate": 1.2860452283094293e-05,
      "loss": 1.9163,
      "step": 8512
    },
    {
      "epoch": 1.533528484575546,
      "grad_norm": 1.375159740447998,
      "learning_rate": 1.285097230051155e-05,
      "loss": 1.4159,
      "step": 8513
    },
    {
      "epoch": 1.5337086241837423,
      "grad_norm": 1.5268782377243042,
      "learning_rate": 1.2841495298005734e-05,
      "loss": 1.5116,
      "step": 8514
    },
    {
      "epoch": 1.5338887637919387,
      "grad_norm": 1.589659571647644,
      "learning_rate": 1.2832021276337065e-05,
      "loss": 1.5728,
      "step": 8515
    },
    {
      "epoch": 1.5340689034001351,
      "grad_norm": 1.4498947858810425,
      "learning_rate": 1.282255023626554e-05,
      "loss": 1.4522,
      "step": 8516
    },
    {
      "epoch": 1.5342490430083315,
      "grad_norm": 1.3239405155181885,
      "learning_rate": 1.2813082178550929e-05,
      "loss": 1.2979,
      "step": 8517
    },
    {
      "epoch": 1.534429182616528,
      "grad_norm": 1.4620486497879028,
      "learning_rate": 1.2803617103952752e-05,
      "loss": 1.3065,
      "step": 8518
    },
    {
      "epoch": 1.5346093222247241,
      "grad_norm": 1.4922089576721191,
      "learning_rate": 1.2794155013230296e-05,
      "loss": 1.5853,
      "step": 8519
    },
    {
      "epoch": 1.5347894618329205,
      "grad_norm": 1.3442966938018799,
      "learning_rate": 1.2784695907142602e-05,
      "loss": 1.3448,
      "step": 8520
    },
    {
      "epoch": 1.5349696014411167,
      "grad_norm": 1.5153357982635498,
      "learning_rate": 1.2775239786448473e-05,
      "loss": 1.6984,
      "step": 8521
    },
    {
      "epoch": 1.5351497410493131,
      "grad_norm": 1.4586541652679443,
      "learning_rate": 1.2765786651906481e-05,
      "loss": 1.1627,
      "step": 8522
    },
    {
      "epoch": 1.5353298806575095,
      "grad_norm": 1.424187421798706,
      "learning_rate": 1.275633650427493e-05,
      "loss": 1.3935,
      "step": 8523
    },
    {
      "epoch": 1.535510020265706,
      "grad_norm": 1.5735300779342651,
      "learning_rate": 1.274688934431194e-05,
      "loss": 1.3414,
      "step": 8524
    },
    {
      "epoch": 1.5356901598739023,
      "grad_norm": 1.4543471336364746,
      "learning_rate": 1.2737445172775342e-05,
      "loss": 1.3258,
      "step": 8525
    },
    {
      "epoch": 1.5358702994820987,
      "grad_norm": 1.348741888999939,
      "learning_rate": 1.2728003990422737e-05,
      "loss": 1.2971,
      "step": 8526
    },
    {
      "epoch": 1.536050439090295,
      "grad_norm": 1.3355053663253784,
      "learning_rate": 1.271856579801149e-05,
      "loss": 1.2999,
      "step": 8527
    },
    {
      "epoch": 1.5362305786984913,
      "grad_norm": 1.512103796005249,
      "learning_rate": 1.2709130596298746e-05,
      "loss": 1.6508,
      "step": 8528
    },
    {
      "epoch": 1.5364107183066875,
      "grad_norm": 1.4918079376220703,
      "learning_rate": 1.269969838604138e-05,
      "loss": 1.4288,
      "step": 8529
    },
    {
      "epoch": 1.536590857914884,
      "grad_norm": 1.3475688695907593,
      "learning_rate": 1.269026916799606e-05,
      "loss": 1.2065,
      "step": 8530
    },
    {
      "epoch": 1.5367709975230803,
      "grad_norm": 1.5625334978103638,
      "learning_rate": 1.2680842942919158e-05,
      "loss": 1.2757,
      "step": 8531
    },
    {
      "epoch": 1.5369511371312767,
      "grad_norm": 1.5092414617538452,
      "learning_rate": 1.267141971156684e-05,
      "loss": 1.3177,
      "step": 8532
    },
    {
      "epoch": 1.5371312767394731,
      "grad_norm": 1.4448717832565308,
      "learning_rate": 1.2661999474695069e-05,
      "loss": 1.3904,
      "step": 8533
    },
    {
      "epoch": 1.5373114163476695,
      "grad_norm": 1.5039159059524536,
      "learning_rate": 1.2652582233059512e-05,
      "loss": 1.4093,
      "step": 8534
    },
    {
      "epoch": 1.537491555955866,
      "grad_norm": 1.530164361000061,
      "learning_rate": 1.2643167987415622e-05,
      "loss": 1.5026,
      "step": 8535
    },
    {
      "epoch": 1.5376716955640621,
      "grad_norm": 1.509382963180542,
      "learning_rate": 1.2633756738518593e-05,
      "loss": 1.3776,
      "step": 8536
    },
    {
      "epoch": 1.5378518351722585,
      "grad_norm": 1.6182764768600464,
      "learning_rate": 1.2624348487123388e-05,
      "loss": 1.4133,
      "step": 8537
    },
    {
      "epoch": 1.5380319747804547,
      "grad_norm": 1.5430703163146973,
      "learning_rate": 1.2614943233984755e-05,
      "loss": 1.5458,
      "step": 8538
    },
    {
      "epoch": 1.5382121143886511,
      "grad_norm": 1.3606374263763428,
      "learning_rate": 1.2605540979857167e-05,
      "loss": 1.2496,
      "step": 8539
    },
    {
      "epoch": 1.5383922539968475,
      "grad_norm": 1.6063823699951172,
      "learning_rate": 1.2596141725494865e-05,
      "loss": 1.4912,
      "step": 8540
    },
    {
      "epoch": 1.538572393605044,
      "grad_norm": 1.3954979181289673,
      "learning_rate": 1.2586745471651856e-05,
      "loss": 1.2741,
      "step": 8541
    },
    {
      "epoch": 1.5387525332132403,
      "grad_norm": 1.4321208000183105,
      "learning_rate": 1.2577352219081905e-05,
      "loss": 1.3587,
      "step": 8542
    },
    {
      "epoch": 1.5389326728214368,
      "grad_norm": 1.566691517829895,
      "learning_rate": 1.2567961968538528e-05,
      "loss": 1.4978,
      "step": 8543
    },
    {
      "epoch": 1.539112812429633,
      "grad_norm": 1.4372515678405762,
      "learning_rate": 1.2558574720775013e-05,
      "loss": 1.2638,
      "step": 8544
    },
    {
      "epoch": 1.5392929520378293,
      "grad_norm": 1.4506967067718506,
      "learning_rate": 1.2549190476544393e-05,
      "loss": 1.2999,
      "step": 8545
    },
    {
      "epoch": 1.5394730916460255,
      "grad_norm": 1.5573097467422485,
      "learning_rate": 1.2539809236599453e-05,
      "loss": 1.4117,
      "step": 8546
    },
    {
      "epoch": 1.539653231254222,
      "grad_norm": 1.500807285308838,
      "learning_rate": 1.2530431001692795e-05,
      "loss": 1.2566,
      "step": 8547
    },
    {
      "epoch": 1.5398333708624183,
      "grad_norm": 1.485992670059204,
      "learning_rate": 1.2521055772576701e-05,
      "loss": 1.188,
      "step": 8548
    },
    {
      "epoch": 1.5400135104706147,
      "grad_norm": 1.6053001880645752,
      "learning_rate": 1.2511683550003261e-05,
      "loss": 1.2021,
      "step": 8549
    },
    {
      "epoch": 1.5401936500788111,
      "grad_norm": 1.5648465156555176,
      "learning_rate": 1.2502314334724313e-05,
      "loss": 1.3417,
      "step": 8550
    },
    {
      "epoch": 1.5403737896870076,
      "grad_norm": 1.3934253454208374,
      "learning_rate": 1.2492948127491438e-05,
      "loss": 1.0801,
      "step": 8551
    },
    {
      "epoch": 1.5405539292952037,
      "grad_norm": 1.1661428213119507,
      "learning_rate": 1.2483584929055997e-05,
      "loss": 1.5584,
      "step": 8552
    },
    {
      "epoch": 1.5407340689034001,
      "grad_norm": 1.3349653482437134,
      "learning_rate": 1.24742247401691e-05,
      "loss": 1.7135,
      "step": 8553
    },
    {
      "epoch": 1.5409142085115963,
      "grad_norm": 1.3491472005844116,
      "learning_rate": 1.2464867561581612e-05,
      "loss": 1.6945,
      "step": 8554
    },
    {
      "epoch": 1.5410943481197927,
      "grad_norm": 1.5111083984375,
      "learning_rate": 1.2455513394044156e-05,
      "loss": 1.9851,
      "step": 8555
    },
    {
      "epoch": 1.5412744877279891,
      "grad_norm": 1.4901374578475952,
      "learning_rate": 1.244616223830714e-05,
      "loss": 1.8563,
      "step": 8556
    },
    {
      "epoch": 1.5414546273361855,
      "grad_norm": 1.4646332263946533,
      "learning_rate": 1.2436814095120696e-05,
      "loss": 2.0575,
      "step": 8557
    },
    {
      "epoch": 1.541634766944382,
      "grad_norm": 1.5264774560928345,
      "learning_rate": 1.2427468965234735e-05,
      "loss": 1.6241,
      "step": 8558
    },
    {
      "epoch": 1.5418149065525784,
      "grad_norm": 1.4877851009368896,
      "learning_rate": 1.2418126849398909e-05,
      "loss": 1.7371,
      "step": 8559
    },
    {
      "epoch": 1.5419950461607748,
      "grad_norm": 1.741468906402588,
      "learning_rate": 1.2408787748362626e-05,
      "loss": 1.8813,
      "step": 8560
    },
    {
      "epoch": 1.542175185768971,
      "grad_norm": 1.8243359327316284,
      "learning_rate": 1.2399451662875094e-05,
      "loss": 1.8387,
      "step": 8561
    },
    {
      "epoch": 1.5423553253771674,
      "grad_norm": 1.5647733211517334,
      "learning_rate": 1.2390118593685235e-05,
      "loss": 1.3895,
      "step": 8562
    },
    {
      "epoch": 1.5425354649853635,
      "grad_norm": 1.4561927318572998,
      "learning_rate": 1.238078854154176e-05,
      "loss": 1.4689,
      "step": 8563
    },
    {
      "epoch": 1.54271560459356,
      "grad_norm": 1.5138221979141235,
      "learning_rate": 1.2371461507193078e-05,
      "loss": 1.4717,
      "step": 8564
    },
    {
      "epoch": 1.5428957442017563,
      "grad_norm": 1.284593939781189,
      "learning_rate": 1.2362137491387432e-05,
      "loss": 1.2359,
      "step": 8565
    },
    {
      "epoch": 1.5430758838099528,
      "grad_norm": 1.575137972831726,
      "learning_rate": 1.2352816494872793e-05,
      "loss": 1.5824,
      "step": 8566
    },
    {
      "epoch": 1.5432560234181492,
      "grad_norm": 1.377861499786377,
      "learning_rate": 1.2343498518396873e-05,
      "loss": 1.2198,
      "step": 8567
    },
    {
      "epoch": 1.5434361630263456,
      "grad_norm": 1.2628295421600342,
      "learning_rate": 1.2334183562707158e-05,
      "loss": 1.2969,
      "step": 8568
    },
    {
      "epoch": 1.5436163026345417,
      "grad_norm": 1.365851879119873,
      "learning_rate": 1.2324871628550877e-05,
      "loss": 1.3515,
      "step": 8569
    },
    {
      "epoch": 1.5437964422427382,
      "grad_norm": 1.325805902481079,
      "learning_rate": 1.2315562716675062e-05,
      "loss": 1.1759,
      "step": 8570
    },
    {
      "epoch": 1.5439765818509343,
      "grad_norm": 1.4242055416107178,
      "learning_rate": 1.2306256827826445e-05,
      "loss": 1.3285,
      "step": 8571
    },
    {
      "epoch": 1.5441567214591307,
      "grad_norm": 1.4825645685195923,
      "learning_rate": 1.229695396275155e-05,
      "loss": 1.4229,
      "step": 8572
    },
    {
      "epoch": 1.5443368610673271,
      "grad_norm": 1.4225372076034546,
      "learning_rate": 1.228765412219664e-05,
      "loss": 1.3283,
      "step": 8573
    },
    {
      "epoch": 1.5445170006755236,
      "grad_norm": 1.410843014717102,
      "learning_rate": 1.227835730690775e-05,
      "loss": 1.2663,
      "step": 8574
    },
    {
      "epoch": 1.54469714028372,
      "grad_norm": 1.3840010166168213,
      "learning_rate": 1.2269063517630663e-05,
      "loss": 1.3268,
      "step": 8575
    },
    {
      "epoch": 1.5448772798919164,
      "grad_norm": 1.4351036548614502,
      "learning_rate": 1.225977275511092e-05,
      "loss": 1.3796,
      "step": 8576
    },
    {
      "epoch": 1.5450574195001125,
      "grad_norm": 1.5183186531066895,
      "learning_rate": 1.2250485020093827e-05,
      "loss": 1.6787,
      "step": 8577
    },
    {
      "epoch": 1.545237559108309,
      "grad_norm": 1.4989542961120605,
      "learning_rate": 1.2241200313324425e-05,
      "loss": 1.4513,
      "step": 8578
    },
    {
      "epoch": 1.5454176987165051,
      "grad_norm": 1.4515091180801392,
      "learning_rate": 1.2231918635547556e-05,
      "loss": 1.437,
      "step": 8579
    },
    {
      "epoch": 1.5455978383247015,
      "grad_norm": 1.346286654472351,
      "learning_rate": 1.2222639987507778e-05,
      "loss": 1.1592,
      "step": 8580
    },
    {
      "epoch": 1.545777977932898,
      "grad_norm": 1.5514501333236694,
      "learning_rate": 1.2213364369949414e-05,
      "loss": 1.3718,
      "step": 8581
    },
    {
      "epoch": 1.5459581175410944,
      "grad_norm": 1.404780626296997,
      "learning_rate": 1.2204091783616562e-05,
      "loss": 1.3441,
      "step": 8582
    },
    {
      "epoch": 1.5461382571492908,
      "grad_norm": 1.6831082105636597,
      "learning_rate": 1.2194822229253051e-05,
      "loss": 1.5636,
      "step": 8583
    },
    {
      "epoch": 1.5463183967574872,
      "grad_norm": 1.6924959421157837,
      "learning_rate": 1.2185555707602492e-05,
      "loss": 1.5399,
      "step": 8584
    },
    {
      "epoch": 1.5464985363656834,
      "grad_norm": 1.4593991041183472,
      "learning_rate": 1.217629221940823e-05,
      "loss": 1.1899,
      "step": 8585
    },
    {
      "epoch": 1.5466786759738798,
      "grad_norm": 1.3241924047470093,
      "learning_rate": 1.216703176541339e-05,
      "loss": 1.0874,
      "step": 8586
    },
    {
      "epoch": 1.5468588155820762,
      "grad_norm": 1.5010648965835571,
      "learning_rate": 1.2157774346360828e-05,
      "loss": 1.3995,
      "step": 8587
    },
    {
      "epoch": 1.5470389551902723,
      "grad_norm": 1.7728914022445679,
      "learning_rate": 1.2148519962993166e-05,
      "loss": 1.6562,
      "step": 8588
    },
    {
      "epoch": 1.5472190947984688,
      "grad_norm": 1.4874451160430908,
      "learning_rate": 1.2139268616052807e-05,
      "loss": 1.3433,
      "step": 8589
    },
    {
      "epoch": 1.5473992344066652,
      "grad_norm": 1.477943778038025,
      "learning_rate": 1.2130020306281881e-05,
      "loss": 1.3239,
      "step": 8590
    },
    {
      "epoch": 1.5475793740148616,
      "grad_norm": 1.4550631046295166,
      "learning_rate": 1.2120775034422276e-05,
      "loss": 1.3155,
      "step": 8591
    },
    {
      "epoch": 1.547759513623058,
      "grad_norm": 1.3794419765472412,
      "learning_rate": 1.2111532801215636e-05,
      "loss": 1.1849,
      "step": 8592
    },
    {
      "epoch": 1.5479396532312544,
      "grad_norm": 1.4824434518814087,
      "learning_rate": 1.2102293607403391e-05,
      "loss": 1.4421,
      "step": 8593
    },
    {
      "epoch": 1.5481197928394506,
      "grad_norm": 1.513271689414978,
      "learning_rate": 1.2093057453726697e-05,
      "loss": 1.3528,
      "step": 8594
    },
    {
      "epoch": 1.548299932447647,
      "grad_norm": 1.6666795015335083,
      "learning_rate": 1.2083824340926486e-05,
      "loss": 1.5194,
      "step": 8595
    },
    {
      "epoch": 1.5484800720558431,
      "grad_norm": 1.5944799184799194,
      "learning_rate": 1.20745942697434e-05,
      "loss": 1.3517,
      "step": 8596
    },
    {
      "epoch": 1.5486602116640396,
      "grad_norm": 1.3408061265945435,
      "learning_rate": 1.2065367240917874e-05,
      "loss": 1.193,
      "step": 8597
    },
    {
      "epoch": 1.548840351272236,
      "grad_norm": 1.4287751913070679,
      "learning_rate": 1.2056143255190128e-05,
      "loss": 1.138,
      "step": 8598
    },
    {
      "epoch": 1.5490204908804324,
      "grad_norm": 1.4982671737670898,
      "learning_rate": 1.2046922313300086e-05,
      "loss": 1.1075,
      "step": 8599
    },
    {
      "epoch": 1.5492006304886288,
      "grad_norm": 1.4751111268997192,
      "learning_rate": 1.2037704415987454e-05,
      "loss": 1.2775,
      "step": 8600
    },
    {
      "epoch": 1.5493807700968252,
      "grad_norm": 1.3570735454559326,
      "learning_rate": 1.2028489563991679e-05,
      "loss": 1.2004,
      "step": 8601
    },
    {
      "epoch": 1.5495609097050214,
      "grad_norm": 1.3632498979568481,
      "learning_rate": 1.201927775805196e-05,
      "loss": 1.7741,
      "step": 8602
    },
    {
      "epoch": 1.5497410493132178,
      "grad_norm": 1.2828961610794067,
      "learning_rate": 1.2010068998907298e-05,
      "loss": 1.7022,
      "step": 8603
    },
    {
      "epoch": 1.549921188921414,
      "grad_norm": 1.2892394065856934,
      "learning_rate": 1.2000863287296394e-05,
      "loss": 1.6525,
      "step": 8604
    },
    {
      "epoch": 1.5501013285296104,
      "grad_norm": 1.4127252101898193,
      "learning_rate": 1.1991660623957728e-05,
      "loss": 1.8569,
      "step": 8605
    },
    {
      "epoch": 1.5502814681378068,
      "grad_norm": 1.4736652374267578,
      "learning_rate": 1.1982461009629536e-05,
      "loss": 1.7014,
      "step": 8606
    },
    {
      "epoch": 1.5504616077460032,
      "grad_norm": 1.449845314025879,
      "learning_rate": 1.1973264445049804e-05,
      "loss": 1.6665,
      "step": 8607
    },
    {
      "epoch": 1.5506417473541996,
      "grad_norm": 1.4491453170776367,
      "learning_rate": 1.1964070930956273e-05,
      "loss": 1.5762,
      "step": 8608
    },
    {
      "epoch": 1.550821886962396,
      "grad_norm": 1.5829616785049438,
      "learning_rate": 1.195488046808645e-05,
      "loss": 1.7513,
      "step": 8609
    },
    {
      "epoch": 1.5510020265705922,
      "grad_norm": 1.8279757499694824,
      "learning_rate": 1.194569305717758e-05,
      "loss": 1.8266,
      "step": 8610
    },
    {
      "epoch": 1.5511821661787886,
      "grad_norm": 1.6994513273239136,
      "learning_rate": 1.1936508698966664e-05,
      "loss": 1.7551,
      "step": 8611
    },
    {
      "epoch": 1.5513623057869848,
      "grad_norm": 1.5221192836761475,
      "learning_rate": 1.1927327394190496e-05,
      "loss": 1.5015,
      "step": 8612
    },
    {
      "epoch": 1.5515424453951812,
      "grad_norm": 1.4585553407669067,
      "learning_rate": 1.1918149143585582e-05,
      "loss": 1.3764,
      "step": 8613
    },
    {
      "epoch": 1.5517225850033776,
      "grad_norm": 1.5565168857574463,
      "learning_rate": 1.1908973947888191e-05,
      "loss": 1.4802,
      "step": 8614
    },
    {
      "epoch": 1.551902724611574,
      "grad_norm": 1.4045219421386719,
      "learning_rate": 1.189980180783436e-05,
      "loss": 1.2302,
      "step": 8615
    },
    {
      "epoch": 1.5520828642197704,
      "grad_norm": 1.4543061256408691,
      "learning_rate": 1.1890632724159866e-05,
      "loss": 1.4389,
      "step": 8616
    },
    {
      "epoch": 1.5522630038279668,
      "grad_norm": 1.552564024925232,
      "learning_rate": 1.1881466697600252e-05,
      "loss": 1.5942,
      "step": 8617
    },
    {
      "epoch": 1.5524431434361632,
      "grad_norm": 1.3444478511810303,
      "learning_rate": 1.1872303728890816e-05,
      "loss": 1.4607,
      "step": 8618
    },
    {
      "epoch": 1.5526232830443594,
      "grad_norm": 1.3649972677230835,
      "learning_rate": 1.1863143818766603e-05,
      "loss": 1.3506,
      "step": 8619
    },
    {
      "epoch": 1.5528034226525558,
      "grad_norm": 1.3263585567474365,
      "learning_rate": 1.18539869679624e-05,
      "loss": 1.212,
      "step": 8620
    },
    {
      "epoch": 1.552983562260752,
      "grad_norm": 1.3795740604400635,
      "learning_rate": 1.18448331772128e-05,
      "loss": 1.4327,
      "step": 8621
    },
    {
      "epoch": 1.5531637018689484,
      "grad_norm": 1.6514915227890015,
      "learning_rate": 1.1835682447252095e-05,
      "loss": 1.5076,
      "step": 8622
    },
    {
      "epoch": 1.5533438414771448,
      "grad_norm": 1.4656131267547607,
      "learning_rate": 1.1826534778814358e-05,
      "loss": 1.4235,
      "step": 8623
    },
    {
      "epoch": 1.5535239810853412,
      "grad_norm": 1.4518013000488281,
      "learning_rate": 1.1817390172633403e-05,
      "loss": 1.2619,
      "step": 8624
    },
    {
      "epoch": 1.5537041206935376,
      "grad_norm": 1.571103811264038,
      "learning_rate": 1.1808248629442803e-05,
      "loss": 1.3302,
      "step": 8625
    },
    {
      "epoch": 1.553884260301734,
      "grad_norm": 1.4875376224517822,
      "learning_rate": 1.17991101499759e-05,
      "loss": 1.4112,
      "step": 8626
    },
    {
      "epoch": 1.5540643999099302,
      "grad_norm": 1.4393737316131592,
      "learning_rate": 1.1789974734965797e-05,
      "loss": 1.2955,
      "step": 8627
    },
    {
      "epoch": 1.5542445395181266,
      "grad_norm": 1.4137779474258423,
      "learning_rate": 1.178084238514529e-05,
      "loss": 1.384,
      "step": 8628
    },
    {
      "epoch": 1.5544246791263228,
      "grad_norm": 1.520317792892456,
      "learning_rate": 1.1771713101246995e-05,
      "loss": 1.1605,
      "step": 8629
    },
    {
      "epoch": 1.5546048187345192,
      "grad_norm": 1.4095158576965332,
      "learning_rate": 1.1762586884003235e-05,
      "loss": 1.3425,
      "step": 8630
    },
    {
      "epoch": 1.5547849583427156,
      "grad_norm": 1.4087046384811401,
      "learning_rate": 1.1753463734146148e-05,
      "loss": 1.2212,
      "step": 8631
    },
    {
      "epoch": 1.554965097950912,
      "grad_norm": 1.5679315328598022,
      "learning_rate": 1.174434365240757e-05,
      "loss": 1.4374,
      "step": 8632
    },
    {
      "epoch": 1.5551452375591084,
      "grad_norm": 1.5481170415878296,
      "learning_rate": 1.173522663951911e-05,
      "loss": 1.5784,
      "step": 8633
    },
    {
      "epoch": 1.5553253771673048,
      "grad_norm": 1.4032551050186157,
      "learning_rate": 1.1726112696212121e-05,
      "loss": 1.394,
      "step": 8634
    },
    {
      "epoch": 1.555505516775501,
      "grad_norm": 1.5116862058639526,
      "learning_rate": 1.1717001823217738e-05,
      "loss": 1.3542,
      "step": 8635
    },
    {
      "epoch": 1.5556856563836974,
      "grad_norm": 1.334087610244751,
      "learning_rate": 1.1707894021266824e-05,
      "loss": 1.1887,
      "step": 8636
    },
    {
      "epoch": 1.5558657959918936,
      "grad_norm": 1.5564337968826294,
      "learning_rate": 1.1698789291090001e-05,
      "loss": 1.3328,
      "step": 8637
    },
    {
      "epoch": 1.55604593560009,
      "grad_norm": 1.647221326828003,
      "learning_rate": 1.168968763341764e-05,
      "loss": 1.6553,
      "step": 8638
    },
    {
      "epoch": 1.5562260752082864,
      "grad_norm": 1.248002290725708,
      "learning_rate": 1.1680589048979885e-05,
      "loss": 1.0252,
      "step": 8639
    },
    {
      "epoch": 1.5564062148164828,
      "grad_norm": 1.4508148431777954,
      "learning_rate": 1.1671493538506606e-05,
      "loss": 1.395,
      "step": 8640
    },
    {
      "epoch": 1.5565863544246792,
      "grad_norm": 1.4866935014724731,
      "learning_rate": 1.166240110272745e-05,
      "loss": 1.3563,
      "step": 8641
    },
    {
      "epoch": 1.5567664940328756,
      "grad_norm": 1.5526775121688843,
      "learning_rate": 1.1653311742371797e-05,
      "loss": 1.4164,
      "step": 8642
    },
    {
      "epoch": 1.5569466336410718,
      "grad_norm": 1.5380525588989258,
      "learning_rate": 1.1644225458168806e-05,
      "loss": 1.4179,
      "step": 8643
    },
    {
      "epoch": 1.5571267732492682,
      "grad_norm": 1.6545181274414062,
      "learning_rate": 1.1635142250847347e-05,
      "loss": 1.3571,
      "step": 8644
    },
    {
      "epoch": 1.5573069128574646,
      "grad_norm": 1.5748405456542969,
      "learning_rate": 1.1626062121136106e-05,
      "loss": 1.2285,
      "step": 8645
    },
    {
      "epoch": 1.5574870524656608,
      "grad_norm": 1.5831753015518188,
      "learning_rate": 1.161698506976347e-05,
      "loss": 1.2879,
      "step": 8646
    },
    {
      "epoch": 1.5576671920738572,
      "grad_norm": 1.5573097467422485,
      "learning_rate": 1.1607911097457596e-05,
      "loss": 1.2037,
      "step": 8647
    },
    {
      "epoch": 1.5578473316820536,
      "grad_norm": 1.6150676012039185,
      "learning_rate": 1.1598840204946394e-05,
      "loss": 1.4767,
      "step": 8648
    },
    {
      "epoch": 1.55802747129025,
      "grad_norm": 1.5323644876480103,
      "learning_rate": 1.1589772392957526e-05,
      "loss": 1.2217,
      "step": 8649
    },
    {
      "epoch": 1.5582076108984464,
      "grad_norm": 1.5842088460922241,
      "learning_rate": 1.158070766221841e-05,
      "loss": 1.2831,
      "step": 8650
    },
    {
      "epoch": 1.5583877505066428,
      "grad_norm": 1.4812527894973755,
      "learning_rate": 1.1571646013456206e-05,
      "loss": 1.088,
      "step": 8651
    },
    {
      "epoch": 1.558567890114839,
      "grad_norm": 1.3270530700683594,
      "learning_rate": 1.1562587447397844e-05,
      "loss": 1.7395,
      "step": 8652
    },
    {
      "epoch": 1.5587480297230354,
      "grad_norm": 1.3473864793777466,
      "learning_rate": 1.1553531964769982e-05,
      "loss": 1.6648,
      "step": 8653
    },
    {
      "epoch": 1.5589281693312316,
      "grad_norm": 1.386386513710022,
      "learning_rate": 1.1544479566299076e-05,
      "loss": 1.7889,
      "step": 8654
    },
    {
      "epoch": 1.559108308939428,
      "grad_norm": 1.4458447694778442,
      "learning_rate": 1.1535430252711282e-05,
      "loss": 1.7524,
      "step": 8655
    },
    {
      "epoch": 1.5592884485476244,
      "grad_norm": 1.5281251668930054,
      "learning_rate": 1.1526384024732539e-05,
      "loss": 1.477,
      "step": 8656
    },
    {
      "epoch": 1.5594685881558208,
      "grad_norm": 1.5493708848953247,
      "learning_rate": 1.1517340883088534e-05,
      "loss": 1.8962,
      "step": 8657
    },
    {
      "epoch": 1.5596487277640172,
      "grad_norm": 1.7481637001037598,
      "learning_rate": 1.150830082850468e-05,
      "loss": 2.1099,
      "step": 8658
    },
    {
      "epoch": 1.5598288673722136,
      "grad_norm": 1.6272221803665161,
      "learning_rate": 1.1499263861706205e-05,
      "loss": 1.7286,
      "step": 8659
    },
    {
      "epoch": 1.5600090069804098,
      "grad_norm": 1.947574257850647,
      "learning_rate": 1.1490229983418038e-05,
      "loss": 2.13,
      "step": 8660
    },
    {
      "epoch": 1.5601891465886062,
      "grad_norm": 1.7890394926071167,
      "learning_rate": 1.1481199194364856e-05,
      "loss": 1.952,
      "step": 8661
    },
    {
      "epoch": 1.5603692861968024,
      "grad_norm": 1.5355799198150635,
      "learning_rate": 1.1472171495271095e-05,
      "loss": 1.437,
      "step": 8662
    },
    {
      "epoch": 1.5605494258049988,
      "grad_norm": 1.4334068298339844,
      "learning_rate": 1.1463146886860981e-05,
      "loss": 1.6783,
      "step": 8663
    },
    {
      "epoch": 1.5607295654131952,
      "grad_norm": 1.4627766609191895,
      "learning_rate": 1.1454125369858459e-05,
      "loss": 1.5114,
      "step": 8664
    },
    {
      "epoch": 1.5609097050213916,
      "grad_norm": 1.4727448225021362,
      "learning_rate": 1.1445106944987222e-05,
      "loss": 1.4182,
      "step": 8665
    },
    {
      "epoch": 1.561089844629588,
      "grad_norm": 1.345745325088501,
      "learning_rate": 1.1436091612970728e-05,
      "loss": 1.207,
      "step": 8666
    },
    {
      "epoch": 1.5612699842377844,
      "grad_norm": 1.46124267578125,
      "learning_rate": 1.1427079374532162e-05,
      "loss": 1.5473,
      "step": 8667
    },
    {
      "epoch": 1.5614501238459806,
      "grad_norm": 1.4334505796432495,
      "learning_rate": 1.1418070230394523e-05,
      "loss": 1.417,
      "step": 8668
    },
    {
      "epoch": 1.561630263454177,
      "grad_norm": 1.4410444498062134,
      "learning_rate": 1.1409064181280493e-05,
      "loss": 1.2146,
      "step": 8669
    },
    {
      "epoch": 1.5618104030623732,
      "grad_norm": 1.5506155490875244,
      "learning_rate": 1.1400061227912534e-05,
      "loss": 1.5874,
      "step": 8670
    },
    {
      "epoch": 1.5619905426705696,
      "grad_norm": 1.5400606393814087,
      "learning_rate": 1.1391061371012867e-05,
      "loss": 1.5057,
      "step": 8671
    },
    {
      "epoch": 1.562170682278766,
      "grad_norm": 1.4226422309875488,
      "learning_rate": 1.138206461130345e-05,
      "loss": 1.3539,
      "step": 8672
    },
    {
      "epoch": 1.5623508218869624,
      "grad_norm": 1.448081374168396,
      "learning_rate": 1.1373070949505998e-05,
      "loss": 1.2107,
      "step": 8673
    },
    {
      "epoch": 1.5625309614951588,
      "grad_norm": 1.4487507343292236,
      "learning_rate": 1.136408038634198e-05,
      "loss": 1.3585,
      "step": 8674
    },
    {
      "epoch": 1.5627111011033552,
      "grad_norm": 1.63332200050354,
      "learning_rate": 1.1355092922532618e-05,
      "loss": 1.6179,
      "step": 8675
    },
    {
      "epoch": 1.5628912407115516,
      "grad_norm": 1.4600070714950562,
      "learning_rate": 1.1346108558798862e-05,
      "loss": 1.3085,
      "step": 8676
    },
    {
      "epoch": 1.5630713803197478,
      "grad_norm": 1.3906023502349854,
      "learning_rate": 1.1337127295861465e-05,
      "loss": 1.3178,
      "step": 8677
    },
    {
      "epoch": 1.5632515199279442,
      "grad_norm": 1.3391629457473755,
      "learning_rate": 1.132814913444088e-05,
      "loss": 1.2214,
      "step": 8678
    },
    {
      "epoch": 1.5634316595361404,
      "grad_norm": 1.5805213451385498,
      "learning_rate": 1.1319174075257338e-05,
      "loss": 1.5774,
      "step": 8679
    },
    {
      "epoch": 1.5636117991443368,
      "grad_norm": 1.6098612546920776,
      "learning_rate": 1.1310202119030815e-05,
      "loss": 1.5041,
      "step": 8680
    },
    {
      "epoch": 1.5637919387525332,
      "grad_norm": 1.6560418605804443,
      "learning_rate": 1.1301233266481025e-05,
      "loss": 1.5958,
      "step": 8681
    },
    {
      "epoch": 1.5639720783607296,
      "grad_norm": 1.4953657388687134,
      "learning_rate": 1.129226751832746e-05,
      "loss": 1.3866,
      "step": 8682
    },
    {
      "epoch": 1.564152217968926,
      "grad_norm": 1.575158715248108,
      "learning_rate": 1.1283304875289336e-05,
      "loss": 1.3636,
      "step": 8683
    },
    {
      "epoch": 1.5643323575771224,
      "grad_norm": 1.5307196378707886,
      "learning_rate": 1.127434533808564e-05,
      "loss": 1.3651,
      "step": 8684
    },
    {
      "epoch": 1.5645124971853186,
      "grad_norm": 1.5176597833633423,
      "learning_rate": 1.1265388907435081e-05,
      "loss": 1.389,
      "step": 8685
    },
    {
      "epoch": 1.564692636793515,
      "grad_norm": 1.6231279373168945,
      "learning_rate": 1.1256435584056173e-05,
      "loss": 1.3781,
      "step": 8686
    },
    {
      "epoch": 1.5648727764017112,
      "grad_norm": 1.5850392580032349,
      "learning_rate": 1.1247485368667132e-05,
      "loss": 1.583,
      "step": 8687
    },
    {
      "epoch": 1.5650529160099076,
      "grad_norm": 1.4881261587142944,
      "learning_rate": 1.1238538261985937e-05,
      "loss": 1.3389,
      "step": 8688
    },
    {
      "epoch": 1.565233055618104,
      "grad_norm": 1.3968855142593384,
      "learning_rate": 1.1229594264730326e-05,
      "loss": 1.1427,
      "step": 8689
    },
    {
      "epoch": 1.5654131952263004,
      "grad_norm": 1.5665854215621948,
      "learning_rate": 1.1220653377617763e-05,
      "loss": 1.4724,
      "step": 8690
    },
    {
      "epoch": 1.5655933348344968,
      "grad_norm": 1.4646254777908325,
      "learning_rate": 1.1211715601365507e-05,
      "loss": 1.3174,
      "step": 8691
    },
    {
      "epoch": 1.5657734744426932,
      "grad_norm": 1.413590908050537,
      "learning_rate": 1.1202780936690549e-05,
      "loss": 1.3014,
      "step": 8692
    },
    {
      "epoch": 1.5659536140508894,
      "grad_norm": 1.6356289386749268,
      "learning_rate": 1.1193849384309595e-05,
      "loss": 1.5751,
      "step": 8693
    },
    {
      "epoch": 1.5661337536590858,
      "grad_norm": 1.5420008897781372,
      "learning_rate": 1.1184920944939143e-05,
      "loss": 1.3866,
      "step": 8694
    },
    {
      "epoch": 1.566313893267282,
      "grad_norm": 1.489952802658081,
      "learning_rate": 1.117599561929541e-05,
      "loss": 1.2339,
      "step": 8695
    },
    {
      "epoch": 1.5664940328754784,
      "grad_norm": 1.566988468170166,
      "learning_rate": 1.1167073408094409e-05,
      "loss": 1.5127,
      "step": 8696
    },
    {
      "epoch": 1.5666741724836748,
      "grad_norm": 1.7159370183944702,
      "learning_rate": 1.1158154312051866e-05,
      "loss": 1.5014,
      "step": 8697
    },
    {
      "epoch": 1.5668543120918712,
      "grad_norm": 1.6454249620437622,
      "learning_rate": 1.1149238331883266e-05,
      "loss": 1.557,
      "step": 8698
    },
    {
      "epoch": 1.5670344517000676,
      "grad_norm": 1.5364835262298584,
      "learning_rate": 1.1140325468303831e-05,
      "loss": 1.4042,
      "step": 8699
    },
    {
      "epoch": 1.567214591308264,
      "grad_norm": 1.4747416973114014,
      "learning_rate": 1.1131415722028571e-05,
      "loss": 1.1796,
      "step": 8700
    },
    {
      "epoch": 1.5673947309164604,
      "grad_norm": 1.4206022024154663,
      "learning_rate": 1.1122509093772209e-05,
      "loss": 1.24,
      "step": 8701
    },
    {
      "epoch": 1.5675748705246566,
      "grad_norm": 1.3547396659851074,
      "learning_rate": 1.1113605584249237e-05,
      "loss": 1.9937,
      "step": 8702
    },
    {
      "epoch": 1.567755010132853,
      "grad_norm": 1.353610873222351,
      "learning_rate": 1.1104705194173876e-05,
      "loss": 1.7046,
      "step": 8703
    },
    {
      "epoch": 1.5679351497410492,
      "grad_norm": 1.425837755203247,
      "learning_rate": 1.1095807924260121e-05,
      "loss": 1.9371,
      "step": 8704
    },
    {
      "epoch": 1.5681152893492456,
      "grad_norm": 1.339847445487976,
      "learning_rate": 1.1086913775221709e-05,
      "loss": 1.6754,
      "step": 8705
    },
    {
      "epoch": 1.568295428957442,
      "grad_norm": 1.453599452972412,
      "learning_rate": 1.1078022747772116e-05,
      "loss": 1.979,
      "step": 8706
    },
    {
      "epoch": 1.5684755685656384,
      "grad_norm": 1.5275856256484985,
      "learning_rate": 1.1069134842624585e-05,
      "loss": 1.8355,
      "step": 8707
    },
    {
      "epoch": 1.5686557081738348,
      "grad_norm": 1.4347549676895142,
      "learning_rate": 1.106025006049209e-05,
      "loss": 1.5313,
      "step": 8708
    },
    {
      "epoch": 1.5688358477820312,
      "grad_norm": 1.5533716678619385,
      "learning_rate": 1.1051368402087353e-05,
      "loss": 1.6628,
      "step": 8709
    },
    {
      "epoch": 1.5690159873902274,
      "grad_norm": 1.687168836593628,
      "learning_rate": 1.1042489868122885e-05,
      "loss": 1.7631,
      "step": 8710
    },
    {
      "epoch": 1.5691961269984238,
      "grad_norm": 2.1341562271118164,
      "learning_rate": 1.1033614459310909e-05,
      "loss": 2.064,
      "step": 8711
    },
    {
      "epoch": 1.56937626660662,
      "grad_norm": 1.457038164138794,
      "learning_rate": 1.1024742176363401e-05,
      "loss": 1.4305,
      "step": 8712
    },
    {
      "epoch": 1.5695564062148164,
      "grad_norm": 1.4856160879135132,
      "learning_rate": 1.1015873019992085e-05,
      "loss": 1.6931,
      "step": 8713
    },
    {
      "epoch": 1.5697365458230128,
      "grad_norm": 1.5376152992248535,
      "learning_rate": 1.1007006990908452e-05,
      "loss": 1.5286,
      "step": 8714
    },
    {
      "epoch": 1.5699166854312092,
      "grad_norm": 1.4504339694976807,
      "learning_rate": 1.0998144089823725e-05,
      "loss": 1.4434,
      "step": 8715
    },
    {
      "epoch": 1.5700968250394056,
      "grad_norm": 1.3297656774520874,
      "learning_rate": 1.0989284317448883e-05,
      "loss": 1.3496,
      "step": 8716
    },
    {
      "epoch": 1.570276964647602,
      "grad_norm": 1.4830807447433472,
      "learning_rate": 1.0980427674494648e-05,
      "loss": 1.5667,
      "step": 8717
    },
    {
      "epoch": 1.5704571042557982,
      "grad_norm": 1.1919097900390625,
      "learning_rate": 1.0971574161671489e-05,
      "loss": 0.9992,
      "step": 8718
    },
    {
      "epoch": 1.5706372438639946,
      "grad_norm": 1.2866847515106201,
      "learning_rate": 1.0962723779689654e-05,
      "loss": 1.3087,
      "step": 8719
    },
    {
      "epoch": 1.5708173834721908,
      "grad_norm": 1.39894700050354,
      "learning_rate": 1.0953876529259099e-05,
      "loss": 1.4516,
      "step": 8720
    },
    {
      "epoch": 1.5709975230803872,
      "grad_norm": 1.2938555479049683,
      "learning_rate": 1.0945032411089557e-05,
      "loss": 1.295,
      "step": 8721
    },
    {
      "epoch": 1.5711776626885836,
      "grad_norm": 1.3598833084106445,
      "learning_rate": 1.0936191425890486e-05,
      "loss": 1.3623,
      "step": 8722
    },
    {
      "epoch": 1.57135780229678,
      "grad_norm": 1.4263765811920166,
      "learning_rate": 1.0927353574371101e-05,
      "loss": 1.4898,
      "step": 8723
    },
    {
      "epoch": 1.5715379419049764,
      "grad_norm": 1.3458033800125122,
      "learning_rate": 1.0918518857240395e-05,
      "loss": 1.4364,
      "step": 8724
    },
    {
      "epoch": 1.5717180815131728,
      "grad_norm": 1.3973339796066284,
      "learning_rate": 1.090968727520708e-05,
      "loss": 1.3795,
      "step": 8725
    },
    {
      "epoch": 1.571898221121369,
      "grad_norm": 1.4609380960464478,
      "learning_rate": 1.0900858828979598e-05,
      "loss": 1.2824,
      "step": 8726
    },
    {
      "epoch": 1.5720783607295654,
      "grad_norm": 1.5024406909942627,
      "learning_rate": 1.0892033519266164e-05,
      "loss": 1.4705,
      "step": 8727
    },
    {
      "epoch": 1.5722585003377616,
      "grad_norm": 1.7645328044891357,
      "learning_rate": 1.0883211346774768e-05,
      "loss": 1.5015,
      "step": 8728
    },
    {
      "epoch": 1.572438639945958,
      "grad_norm": 1.6559275388717651,
      "learning_rate": 1.0874392312213106e-05,
      "loss": 1.5571,
      "step": 8729
    },
    {
      "epoch": 1.5726187795541544,
      "grad_norm": 1.4846141338348389,
      "learning_rate": 1.086557641628863e-05,
      "loss": 1.3511,
      "step": 8730
    },
    {
      "epoch": 1.5727989191623508,
      "grad_norm": 1.466202974319458,
      "learning_rate": 1.0856763659708558e-05,
      "loss": 1.468,
      "step": 8731
    },
    {
      "epoch": 1.5729790587705472,
      "grad_norm": 1.507413625717163,
      "learning_rate": 1.084795404317982e-05,
      "loss": 1.378,
      "step": 8732
    },
    {
      "epoch": 1.5731591983787436,
      "grad_norm": 1.5746922492980957,
      "learning_rate": 1.0839147567409158e-05,
      "loss": 1.5838,
      "step": 8733
    },
    {
      "epoch": 1.57333933798694,
      "grad_norm": 1.5736324787139893,
      "learning_rate": 1.0830344233103002e-05,
      "loss": 1.5805,
      "step": 8734
    },
    {
      "epoch": 1.5735194775951362,
      "grad_norm": 1.5431368350982666,
      "learning_rate": 1.0821544040967546e-05,
      "loss": 1.4128,
      "step": 8735
    },
    {
      "epoch": 1.5736996172033326,
      "grad_norm": 1.585395336151123,
      "learning_rate": 1.081274699170875e-05,
      "loss": 1.3835,
      "step": 8736
    },
    {
      "epoch": 1.5738797568115288,
      "grad_norm": 1.4899364709854126,
      "learning_rate": 1.0803953086032303e-05,
      "loss": 1.4272,
      "step": 8737
    },
    {
      "epoch": 1.5740598964197252,
      "grad_norm": 1.4143012762069702,
      "learning_rate": 1.0795162324643648e-05,
      "loss": 1.485,
      "step": 8738
    },
    {
      "epoch": 1.5742400360279216,
      "grad_norm": 1.39919912815094,
      "learning_rate": 1.078637470824797e-05,
      "loss": 1.3085,
      "step": 8739
    },
    {
      "epoch": 1.574420175636118,
      "grad_norm": 1.431362509727478,
      "learning_rate": 1.0777590237550217e-05,
      "loss": 1.3317,
      "step": 8740
    },
    {
      "epoch": 1.5746003152443144,
      "grad_norm": 1.5391736030578613,
      "learning_rate": 1.0768808913255057e-05,
      "loss": 1.2625,
      "step": 8741
    },
    {
      "epoch": 1.5747804548525108,
      "grad_norm": 1.604718804359436,
      "learning_rate": 1.0760030736066951e-05,
      "loss": 1.5051,
      "step": 8742
    },
    {
      "epoch": 1.574960594460707,
      "grad_norm": 1.5461753606796265,
      "learning_rate": 1.0751255706690066e-05,
      "loss": 1.2147,
      "step": 8743
    },
    {
      "epoch": 1.5751407340689034,
      "grad_norm": 1.4858249425888062,
      "learning_rate": 1.074248382582833e-05,
      "loss": 1.3766,
      "step": 8744
    },
    {
      "epoch": 1.5753208736770996,
      "grad_norm": 1.5699418783187866,
      "learning_rate": 1.073371509418542e-05,
      "loss": 1.477,
      "step": 8745
    },
    {
      "epoch": 1.575501013285296,
      "grad_norm": 1.7523880004882812,
      "learning_rate": 1.072494951246476e-05,
      "loss": 1.4182,
      "step": 8746
    },
    {
      "epoch": 1.5756811528934924,
      "grad_norm": 1.4491461515426636,
      "learning_rate": 1.0716187081369516e-05,
      "loss": 1.3406,
      "step": 8747
    },
    {
      "epoch": 1.5758612925016888,
      "grad_norm": 1.5331447124481201,
      "learning_rate": 1.0707427801602615e-05,
      "loss": 1.2322,
      "step": 8748
    },
    {
      "epoch": 1.5760414321098852,
      "grad_norm": 1.5976214408874512,
      "learning_rate": 1.0698671673866718e-05,
      "loss": 1.2848,
      "step": 8749
    },
    {
      "epoch": 1.5762215717180816,
      "grad_norm": 1.499138593673706,
      "learning_rate": 1.0689918698864232e-05,
      "loss": 1.2546,
      "step": 8750
    },
    {
      "epoch": 1.5764017113262778,
      "grad_norm": 1.5307953357696533,
      "learning_rate": 1.0681168877297303e-05,
      "loss": 1.3887,
      "step": 8751
    },
    {
      "epoch": 1.5765818509344742,
      "grad_norm": 1.4458649158477783,
      "learning_rate": 1.0672422209867878e-05,
      "loss": 2.1768,
      "step": 8752
    },
    {
      "epoch": 1.5767619905426704,
      "grad_norm": 1.2190377712249756,
      "learning_rate": 1.0663678697277586e-05,
      "loss": 1.6207,
      "step": 8753
    },
    {
      "epoch": 1.5769421301508668,
      "grad_norm": 1.38332998752594,
      "learning_rate": 1.0654938340227821e-05,
      "loss": 1.7679,
      "step": 8754
    },
    {
      "epoch": 1.5771222697590632,
      "grad_norm": 1.447637677192688,
      "learning_rate": 1.0646201139419732e-05,
      "loss": 1.6161,
      "step": 8755
    },
    {
      "epoch": 1.5773024093672596,
      "grad_norm": 1.3600705862045288,
      "learning_rate": 1.0637467095554231e-05,
      "loss": 1.6255,
      "step": 8756
    },
    {
      "epoch": 1.577482548975456,
      "grad_norm": 1.5435174703598022,
      "learning_rate": 1.0628736209331962e-05,
      "loss": 1.7634,
      "step": 8757
    },
    {
      "epoch": 1.5776626885836524,
      "grad_norm": 1.5020148754119873,
      "learning_rate": 1.062000848145328e-05,
      "loss": 1.9926,
      "step": 8758
    },
    {
      "epoch": 1.5778428281918488,
      "grad_norm": 1.4623502492904663,
      "learning_rate": 1.0611283912618342e-05,
      "loss": 1.6484,
      "step": 8759
    },
    {
      "epoch": 1.578022967800045,
      "grad_norm": 1.5832890272140503,
      "learning_rate": 1.0602562503527003e-05,
      "loss": 1.6794,
      "step": 8760
    },
    {
      "epoch": 1.5782031074082414,
      "grad_norm": 1.8005592823028564,
      "learning_rate": 1.0593844254878926e-05,
      "loss": 2.0905,
      "step": 8761
    },
    {
      "epoch": 1.5783832470164376,
      "grad_norm": 1.6440430879592896,
      "learning_rate": 1.058512916737347e-05,
      "loss": 1.7672,
      "step": 8762
    },
    {
      "epoch": 1.578563386624634,
      "grad_norm": 1.4857702255249023,
      "learning_rate": 1.0576417241709751e-05,
      "loss": 1.6237,
      "step": 8763
    },
    {
      "epoch": 1.5787435262328304,
      "grad_norm": 1.4623029232025146,
      "learning_rate": 1.0567708478586642e-05,
      "loss": 1.7061,
      "step": 8764
    },
    {
      "epoch": 1.5789236658410268,
      "grad_norm": 1.4218295812606812,
      "learning_rate": 1.0559002878702734e-05,
      "loss": 1.3973,
      "step": 8765
    },
    {
      "epoch": 1.5791038054492232,
      "grad_norm": 1.3829760551452637,
      "learning_rate": 1.0550300442756422e-05,
      "loss": 1.2975,
      "step": 8766
    },
    {
      "epoch": 1.5792839450574196,
      "grad_norm": 1.3502843379974365,
      "learning_rate": 1.0541601171445791e-05,
      "loss": 1.3248,
      "step": 8767
    },
    {
      "epoch": 1.5794640846656158,
      "grad_norm": 1.4077351093292236,
      "learning_rate": 1.0532905065468695e-05,
      "loss": 1.3455,
      "step": 8768
    },
    {
      "epoch": 1.5796442242738122,
      "grad_norm": 1.3918960094451904,
      "learning_rate": 1.0524212125522732e-05,
      "loss": 1.203,
      "step": 8769
    },
    {
      "epoch": 1.5798243638820084,
      "grad_norm": 1.37799870967865,
      "learning_rate": 1.0515522352305246e-05,
      "loss": 1.5085,
      "step": 8770
    },
    {
      "epoch": 1.5800045034902048,
      "grad_norm": 1.4105336666107178,
      "learning_rate": 1.0506835746513321e-05,
      "loss": 1.1745,
      "step": 8771
    },
    {
      "epoch": 1.5801846430984012,
      "grad_norm": 1.4574953317642212,
      "learning_rate": 1.0498152308843796e-05,
      "loss": 1.4043,
      "step": 8772
    },
    {
      "epoch": 1.5803647827065976,
      "grad_norm": 1.4460452795028687,
      "learning_rate": 1.0489472039993254e-05,
      "loss": 1.6298,
      "step": 8773
    },
    {
      "epoch": 1.580544922314794,
      "grad_norm": 1.3603572845458984,
      "learning_rate": 1.0480794940658006e-05,
      "loss": 1.1927,
      "step": 8774
    },
    {
      "epoch": 1.5807250619229904,
      "grad_norm": 1.5552875995635986,
      "learning_rate": 1.0472121011534153e-05,
      "loss": 1.4476,
      "step": 8775
    },
    {
      "epoch": 1.5809052015311866,
      "grad_norm": 1.4770456552505493,
      "learning_rate": 1.0463450253317497e-05,
      "loss": 1.2685,
      "step": 8776
    },
    {
      "epoch": 1.581085341139383,
      "grad_norm": 1.4912656545639038,
      "learning_rate": 1.0454782666703605e-05,
      "loss": 1.3914,
      "step": 8777
    },
    {
      "epoch": 1.5812654807475792,
      "grad_norm": 1.3550387620925903,
      "learning_rate": 1.0446118252387787e-05,
      "loss": 1.3277,
      "step": 8778
    },
    {
      "epoch": 1.5814456203557756,
      "grad_norm": 1.4726510047912598,
      "learning_rate": 1.0437457011065093e-05,
      "loss": 1.3966,
      "step": 8779
    },
    {
      "epoch": 1.581625759963972,
      "grad_norm": 1.511576533317566,
      "learning_rate": 1.0428798943430328e-05,
      "loss": 1.4058,
      "step": 8780
    },
    {
      "epoch": 1.5818058995721684,
      "grad_norm": 1.4912424087524414,
      "learning_rate": 1.0420144050178043e-05,
      "loss": 1.4979,
      "step": 8781
    },
    {
      "epoch": 1.5819860391803648,
      "grad_norm": 1.489510416984558,
      "learning_rate": 1.041149233200252e-05,
      "loss": 1.5692,
      "step": 8782
    },
    {
      "epoch": 1.5821661787885613,
      "grad_norm": 1.7289700508117676,
      "learning_rate": 1.0402843789597783e-05,
      "loss": 1.5418,
      "step": 8783
    },
    {
      "epoch": 1.5823463183967574,
      "grad_norm": 1.5123237371444702,
      "learning_rate": 1.0394198423657648e-05,
      "loss": 1.3976,
      "step": 8784
    },
    {
      "epoch": 1.5825264580049538,
      "grad_norm": 1.5076872110366821,
      "learning_rate": 1.0385556234875622e-05,
      "loss": 1.3539,
      "step": 8785
    },
    {
      "epoch": 1.5827065976131502,
      "grad_norm": 1.6992595195770264,
      "learning_rate": 1.0376917223944981e-05,
      "loss": 1.6975,
      "step": 8786
    },
    {
      "epoch": 1.5828867372213464,
      "grad_norm": 1.6513583660125732,
      "learning_rate": 1.0368281391558737e-05,
      "loss": 1.6185,
      "step": 8787
    },
    {
      "epoch": 1.5830668768295428,
      "grad_norm": 1.449255108833313,
      "learning_rate": 1.0359648738409645e-05,
      "loss": 1.4475,
      "step": 8788
    },
    {
      "epoch": 1.5832470164377392,
      "grad_norm": 1.5623551607131958,
      "learning_rate": 1.0351019265190237e-05,
      "loss": 1.4849,
      "step": 8789
    },
    {
      "epoch": 1.5834271560459356,
      "grad_norm": 1.5850846767425537,
      "learning_rate": 1.0342392972592762e-05,
      "loss": 1.3448,
      "step": 8790
    },
    {
      "epoch": 1.583607295654132,
      "grad_norm": 1.5581655502319336,
      "learning_rate": 1.03337698613092e-05,
      "loss": 1.3803,
      "step": 8791
    },
    {
      "epoch": 1.5837874352623285,
      "grad_norm": 1.5412063598632812,
      "learning_rate": 1.032514993203128e-05,
      "loss": 1.3988,
      "step": 8792
    },
    {
      "epoch": 1.5839675748705246,
      "grad_norm": 1.5423779487609863,
      "learning_rate": 1.0316533185450522e-05,
      "loss": 1.4385,
      "step": 8793
    },
    {
      "epoch": 1.584147714478721,
      "grad_norm": 1.5907313823699951,
      "learning_rate": 1.0307919622258139e-05,
      "loss": 1.3948,
      "step": 8794
    },
    {
      "epoch": 1.5843278540869172,
      "grad_norm": 1.5788952112197876,
      "learning_rate": 1.0299309243145117e-05,
      "loss": 1.3184,
      "step": 8795
    },
    {
      "epoch": 1.5845079936951136,
      "grad_norm": 1.5469717979431152,
      "learning_rate": 1.0290702048802164e-05,
      "loss": 1.2234,
      "step": 8796
    },
    {
      "epoch": 1.58468813330331,
      "grad_norm": 1.5656107664108276,
      "learning_rate": 1.028209803991974e-05,
      "loss": 1.2106,
      "step": 8797
    },
    {
      "epoch": 1.5848682729115064,
      "grad_norm": 1.4900894165039062,
      "learning_rate": 1.0273497217188076e-05,
      "loss": 1.1863,
      "step": 8798
    },
    {
      "epoch": 1.5850484125197029,
      "grad_norm": 1.8199594020843506,
      "learning_rate": 1.0264899581297122e-05,
      "loss": 1.5803,
      "step": 8799
    },
    {
      "epoch": 1.5852285521278993,
      "grad_norm": 1.5054301023483276,
      "learning_rate": 1.0256305132936567e-05,
      "loss": 1.2635,
      "step": 8800
    },
    {
      "epoch": 1.5854086917360954,
      "grad_norm": 1.4701454639434814,
      "learning_rate": 1.024771387279585e-05,
      "loss": 0.9962,
      "step": 8801
    },
    {
      "epoch": 1.5855888313442918,
      "grad_norm": 1.417427897453308,
      "learning_rate": 1.023912580156417e-05,
      "loss": 1.5873,
      "step": 8802
    },
    {
      "epoch": 1.585768970952488,
      "grad_norm": 1.3216639757156372,
      "learning_rate": 1.0230540919930448e-05,
      "loss": 1.5739,
      "step": 8803
    },
    {
      "epoch": 1.5859491105606844,
      "grad_norm": 1.4012728929519653,
      "learning_rate": 1.0221959228583366e-05,
      "loss": 1.6963,
      "step": 8804
    },
    {
      "epoch": 1.5861292501688808,
      "grad_norm": 1.4364832639694214,
      "learning_rate": 1.0213380728211336e-05,
      "loss": 1.8545,
      "step": 8805
    },
    {
      "epoch": 1.5863093897770773,
      "grad_norm": 1.3880946636199951,
      "learning_rate": 1.0204805419502511e-05,
      "loss": 1.7965,
      "step": 8806
    },
    {
      "epoch": 1.5864895293852737,
      "grad_norm": 1.4016976356506348,
      "learning_rate": 1.0196233303144826e-05,
      "loss": 1.5051,
      "step": 8807
    },
    {
      "epoch": 1.58666966899347,
      "grad_norm": 1.522895336151123,
      "learning_rate": 1.0187664379825917e-05,
      "loss": 1.9196,
      "step": 8808
    },
    {
      "epoch": 1.5868498086016662,
      "grad_norm": 1.7216529846191406,
      "learning_rate": 1.0179098650233182e-05,
      "loss": 2.0875,
      "step": 8809
    },
    {
      "epoch": 1.5870299482098627,
      "grad_norm": 1.7687259912490845,
      "learning_rate": 1.0170536115053753e-05,
      "loss": 1.7505,
      "step": 8810
    },
    {
      "epoch": 1.5872100878180588,
      "grad_norm": 1.7006326913833618,
      "learning_rate": 1.0161976774974519e-05,
      "loss": 1.8563,
      "step": 8811
    },
    {
      "epoch": 1.5873902274262552,
      "grad_norm": 1.5786316394805908,
      "learning_rate": 1.015342063068211e-05,
      "loss": 1.668,
      "step": 8812
    },
    {
      "epoch": 1.5875703670344516,
      "grad_norm": 1.4506906270980835,
      "learning_rate": 1.014486768286288e-05,
      "loss": 1.6047,
      "step": 8813
    },
    {
      "epoch": 1.587750506642648,
      "grad_norm": 1.423282265663147,
      "learning_rate": 1.0136317932202955e-05,
      "loss": 1.4585,
      "step": 8814
    },
    {
      "epoch": 1.5879306462508445,
      "grad_norm": 1.341646671295166,
      "learning_rate": 1.012777137938819e-05,
      "loss": 1.2959,
      "step": 8815
    },
    {
      "epoch": 1.5881107858590409,
      "grad_norm": 1.3415254354476929,
      "learning_rate": 1.0119228025104166e-05,
      "loss": 1.3727,
      "step": 8816
    },
    {
      "epoch": 1.5882909254672373,
      "grad_norm": 1.4170063734054565,
      "learning_rate": 1.0110687870036262e-05,
      "loss": 1.0576,
      "step": 8817
    },
    {
      "epoch": 1.5884710650754335,
      "grad_norm": 1.4202964305877686,
      "learning_rate": 1.0102150914869545e-05,
      "loss": 1.4034,
      "step": 8818
    },
    {
      "epoch": 1.5886512046836299,
      "grad_norm": 1.484820008277893,
      "learning_rate": 1.0093617160288853e-05,
      "loss": 1.5881,
      "step": 8819
    },
    {
      "epoch": 1.588831344291826,
      "grad_norm": 1.463269829750061,
      "learning_rate": 1.0085086606978739e-05,
      "loss": 1.3581,
      "step": 8820
    },
    {
      "epoch": 1.5890114839000224,
      "grad_norm": 1.4107099771499634,
      "learning_rate": 1.0076559255623546e-05,
      "loss": 1.323,
      "step": 8821
    },
    {
      "epoch": 1.5891916235082189,
      "grad_norm": 1.4247010946273804,
      "learning_rate": 1.0068035106907342e-05,
      "loss": 1.3528,
      "step": 8822
    },
    {
      "epoch": 1.5893717631164153,
      "grad_norm": 1.4731537103652954,
      "learning_rate": 1.0059514161513895e-05,
      "loss": 1.4214,
      "step": 8823
    },
    {
      "epoch": 1.5895519027246117,
      "grad_norm": 1.4545427560806274,
      "learning_rate": 1.0050996420126768e-05,
      "loss": 1.3684,
      "step": 8824
    },
    {
      "epoch": 1.589732042332808,
      "grad_norm": 1.497930884361267,
      "learning_rate": 1.0042481883429234e-05,
      "loss": 1.3873,
      "step": 8825
    },
    {
      "epoch": 1.5899121819410043,
      "grad_norm": 1.3994158506393433,
      "learning_rate": 1.0033970552104354e-05,
      "loss": 1.1974,
      "step": 8826
    },
    {
      "epoch": 1.5900923215492007,
      "grad_norm": 1.3913419246673584,
      "learning_rate": 1.0025462426834897e-05,
      "loss": 1.2426,
      "step": 8827
    },
    {
      "epoch": 1.5902724611573968,
      "grad_norm": 1.438128113746643,
      "learning_rate": 1.0016957508303365e-05,
      "loss": 1.4723,
      "step": 8828
    },
    {
      "epoch": 1.5904526007655932,
      "grad_norm": 1.4592293500900269,
      "learning_rate": 1.0008455797192029e-05,
      "loss": 1.3464,
      "step": 8829
    },
    {
      "epoch": 1.5906327403737897,
      "grad_norm": 1.5633761882781982,
      "learning_rate": 9.999957294182877e-06,
      "loss": 1.4671,
      "step": 8830
    },
    {
      "epoch": 1.590812879981986,
      "grad_norm": 1.4533586502075195,
      "learning_rate": 9.991461999957686e-06,
      "loss": 1.4118,
      "step": 8831
    },
    {
      "epoch": 1.5909930195901825,
      "grad_norm": 1.4530785083770752,
      "learning_rate": 9.982969915197915e-06,
      "loss": 1.3829,
      "step": 8832
    },
    {
      "epoch": 1.5911731591983789,
      "grad_norm": 1.5011948347091675,
      "learning_rate": 9.974481040584815e-06,
      "loss": 1.4278,
      "step": 8833
    },
    {
      "epoch": 1.591353298806575,
      "grad_norm": 1.4997286796569824,
      "learning_rate": 9.965995376799348e-06,
      "loss": 1.5595,
      "step": 8834
    },
    {
      "epoch": 1.5915334384147715,
      "grad_norm": 1.6865862607955933,
      "learning_rate": 9.957512924522227e-06,
      "loss": 1.7658,
      "step": 8835
    },
    {
      "epoch": 1.5917135780229676,
      "grad_norm": 1.48899245262146,
      "learning_rate": 9.949033684433922e-06,
      "loss": 1.5466,
      "step": 8836
    },
    {
      "epoch": 1.591893717631164,
      "grad_norm": 1.644566535949707,
      "learning_rate": 9.940557657214622e-06,
      "loss": 1.2391,
      "step": 8837
    },
    {
      "epoch": 1.5920738572393605,
      "grad_norm": 1.5699700117111206,
      "learning_rate": 9.932084843544277e-06,
      "loss": 1.3142,
      "step": 8838
    },
    {
      "epoch": 1.5922539968475569,
      "grad_norm": 1.5089904069900513,
      "learning_rate": 9.923615244102558e-06,
      "loss": 1.3419,
      "step": 8839
    },
    {
      "epoch": 1.5924341364557533,
      "grad_norm": 1.5472806692123413,
      "learning_rate": 9.915148859568912e-06,
      "loss": 1.41,
      "step": 8840
    },
    {
      "epoch": 1.5926142760639497,
      "grad_norm": 1.543796181678772,
      "learning_rate": 9.906685690622503e-06,
      "loss": 1.5795,
      "step": 8841
    },
    {
      "epoch": 1.5927944156721459,
      "grad_norm": 1.555656909942627,
      "learning_rate": 9.89822573794224e-06,
      "loss": 1.3925,
      "step": 8842
    },
    {
      "epoch": 1.5929745552803423,
      "grad_norm": 1.4736770391464233,
      "learning_rate": 9.889769002206773e-06,
      "loss": 1.2486,
      "step": 8843
    },
    {
      "epoch": 1.5931546948885387,
      "grad_norm": 1.7369734048843384,
      "learning_rate": 9.8813154840945e-06,
      "loss": 1.4568,
      "step": 8844
    },
    {
      "epoch": 1.5933348344967349,
      "grad_norm": 1.5936263799667358,
      "learning_rate": 9.872865184283559e-06,
      "loss": 1.4271,
      "step": 8845
    },
    {
      "epoch": 1.5935149741049313,
      "grad_norm": 1.5042322874069214,
      "learning_rate": 9.864418103451828e-06,
      "loss": 1.2519,
      "step": 8846
    },
    {
      "epoch": 1.5936951137131277,
      "grad_norm": 1.5108041763305664,
      "learning_rate": 9.855974242276923e-06,
      "loss": 1.4058,
      "step": 8847
    },
    {
      "epoch": 1.593875253321324,
      "grad_norm": 1.5147268772125244,
      "learning_rate": 9.847533601436204e-06,
      "loss": 1.2817,
      "step": 8848
    },
    {
      "epoch": 1.5940553929295205,
      "grad_norm": 1.6327966451644897,
      "learning_rate": 9.839096181606789e-06,
      "loss": 1.3359,
      "step": 8849
    },
    {
      "epoch": 1.5942355325377169,
      "grad_norm": 1.4893242120742798,
      "learning_rate": 9.830661983465522e-06,
      "loss": 1.2338,
      "step": 8850
    },
    {
      "epoch": 1.594415672145913,
      "grad_norm": 1.377326250076294,
      "learning_rate": 9.822231007688981e-06,
      "loss": 1.1019,
      "step": 8851
    },
    {
      "epoch": 1.5945958117541095,
      "grad_norm": 1.2911468744277954,
      "learning_rate": 9.813803254953503e-06,
      "loss": 1.391,
      "step": 8852
    },
    {
      "epoch": 1.5947759513623057,
      "grad_norm": 1.2680950164794922,
      "learning_rate": 9.805378725935133e-06,
      "loss": 1.4327,
      "step": 8853
    },
    {
      "epoch": 1.594956090970502,
      "grad_norm": 1.289388656616211,
      "learning_rate": 9.796957421309738e-06,
      "loss": 1.7376,
      "step": 8854
    },
    {
      "epoch": 1.5951362305786985,
      "grad_norm": 1.5124446153640747,
      "learning_rate": 9.788539341752816e-06,
      "loss": 2.1833,
      "step": 8855
    },
    {
      "epoch": 1.5953163701868949,
      "grad_norm": 1.6897313594818115,
      "learning_rate": 9.780124487939691e-06,
      "loss": 2.0108,
      "step": 8856
    },
    {
      "epoch": 1.5954965097950913,
      "grad_norm": 1.4944745302200317,
      "learning_rate": 9.771712860545379e-06,
      "loss": 1.4695,
      "step": 8857
    },
    {
      "epoch": 1.5956766494032877,
      "grad_norm": 1.509650468826294,
      "learning_rate": 9.76330446024466e-06,
      "loss": 1.5227,
      "step": 8858
    },
    {
      "epoch": 1.5958567890114839,
      "grad_norm": 1.6214643716812134,
      "learning_rate": 9.754899287712072e-06,
      "loss": 1.9105,
      "step": 8859
    },
    {
      "epoch": 1.5960369286196803,
      "grad_norm": 1.7892626523971558,
      "learning_rate": 9.746497343621857e-06,
      "loss": 1.9426,
      "step": 8860
    },
    {
      "epoch": 1.5962170682278765,
      "grad_norm": 1.7339506149291992,
      "learning_rate": 9.738098628648023e-06,
      "loss": 2.1087,
      "step": 8861
    },
    {
      "epoch": 1.5963972078360729,
      "grad_norm": 1.6403619050979614,
      "learning_rate": 9.729703143464292e-06,
      "loss": 1.6197,
      "step": 8862
    },
    {
      "epoch": 1.5965773474442693,
      "grad_norm": 1.554610013961792,
      "learning_rate": 9.721310888744167e-06,
      "loss": 1.6336,
      "step": 8863
    },
    {
      "epoch": 1.5967574870524657,
      "grad_norm": 1.4719735383987427,
      "learning_rate": 9.712921865160867e-06,
      "loss": 1.5205,
      "step": 8864
    },
    {
      "epoch": 1.596937626660662,
      "grad_norm": 1.3846261501312256,
      "learning_rate": 9.704536073387355e-06,
      "loss": 1.3065,
      "step": 8865
    },
    {
      "epoch": 1.5971177662688585,
      "grad_norm": 1.4613518714904785,
      "learning_rate": 9.696153514096329e-06,
      "loss": 1.4671,
      "step": 8866
    },
    {
      "epoch": 1.5972979058770547,
      "grad_norm": 1.4590027332305908,
      "learning_rate": 9.687774187960236e-06,
      "loss": 1.1502,
      "step": 8867
    },
    {
      "epoch": 1.597478045485251,
      "grad_norm": 1.4454160928726196,
      "learning_rate": 9.679398095651265e-06,
      "loss": 1.3929,
      "step": 8868
    },
    {
      "epoch": 1.5976581850934473,
      "grad_norm": 1.4497108459472656,
      "learning_rate": 9.671025237841335e-06,
      "loss": 1.4831,
      "step": 8869
    },
    {
      "epoch": 1.5978383247016437,
      "grad_norm": 1.4551922082901,
      "learning_rate": 9.66265561520212e-06,
      "loss": 1.4239,
      "step": 8870
    },
    {
      "epoch": 1.59801846430984,
      "grad_norm": 1.5123260021209717,
      "learning_rate": 9.654289228405028e-06,
      "loss": 1.5833,
      "step": 8871
    },
    {
      "epoch": 1.5981986039180365,
      "grad_norm": 1.4101616144180298,
      "learning_rate": 9.645926078121181e-06,
      "loss": 1.296,
      "step": 8872
    },
    {
      "epoch": 1.5983787435262329,
      "grad_norm": 1.456728219985962,
      "learning_rate": 9.637566165021506e-06,
      "loss": 1.5427,
      "step": 8873
    },
    {
      "epoch": 1.5985588831344293,
      "grad_norm": 1.4670246839523315,
      "learning_rate": 9.629209489776614e-06,
      "loss": 1.4958,
      "step": 8874
    },
    {
      "epoch": 1.5987390227426257,
      "grad_norm": 1.5197744369506836,
      "learning_rate": 9.620856053056871e-06,
      "loss": 1.5192,
      "step": 8875
    },
    {
      "epoch": 1.5989191623508219,
      "grad_norm": 1.3986202478408813,
      "learning_rate": 9.61250585553239e-06,
      "loss": 1.4055,
      "step": 8876
    },
    {
      "epoch": 1.5990993019590183,
      "grad_norm": 1.4559756517410278,
      "learning_rate": 9.604158897873012e-06,
      "loss": 1.3765,
      "step": 8877
    },
    {
      "epoch": 1.5992794415672145,
      "grad_norm": 1.3098924160003662,
      "learning_rate": 9.595815180748335e-06,
      "loss": 1.2494,
      "step": 8878
    },
    {
      "epoch": 1.5994595811754109,
      "grad_norm": 1.5794737339019775,
      "learning_rate": 9.58747470482768e-06,
      "loss": 1.522,
      "step": 8879
    },
    {
      "epoch": 1.5996397207836073,
      "grad_norm": 1.500978946685791,
      "learning_rate": 9.579137470780125e-06,
      "loss": 1.554,
      "step": 8880
    },
    {
      "epoch": 1.5998198603918037,
      "grad_norm": 1.487963080406189,
      "learning_rate": 9.57080347927446e-06,
      "loss": 1.3358,
      "step": 8881
    },
    {
      "epoch": 1.6,
      "grad_norm": 1.5781112909317017,
      "learning_rate": 9.562472730979261e-06,
      "loss": 1.4216,
      "step": 8882
    },
    {
      "epoch": 1.6001801396081965,
      "grad_norm": 1.4555845260620117,
      "learning_rate": 9.5541452265628e-06,
      "loss": 1.4076,
      "step": 8883
    },
    {
      "epoch": 1.6003602792163927,
      "grad_norm": 1.5419166088104248,
      "learning_rate": 9.545820966693115e-06,
      "loss": 1.4619,
      "step": 8884
    },
    {
      "epoch": 1.600540418824589,
      "grad_norm": 1.4731258153915405,
      "learning_rate": 9.537499952037948e-06,
      "loss": 1.3367,
      "step": 8885
    },
    {
      "epoch": 1.6007205584327853,
      "grad_norm": 1.6260346174240112,
      "learning_rate": 9.52918218326484e-06,
      "loss": 1.5014,
      "step": 8886
    },
    {
      "epoch": 1.6009006980409817,
      "grad_norm": 1.6821283102035522,
      "learning_rate": 9.520867661041038e-06,
      "loss": 1.5689,
      "step": 8887
    },
    {
      "epoch": 1.601080837649178,
      "grad_norm": 1.597687005996704,
      "learning_rate": 9.512556386033505e-06,
      "loss": 1.5017,
      "step": 8888
    },
    {
      "epoch": 1.6012609772573745,
      "grad_norm": 1.5506452322006226,
      "learning_rate": 9.504248358908974e-06,
      "loss": 1.4065,
      "step": 8889
    },
    {
      "epoch": 1.601441116865571,
      "grad_norm": 1.5384083986282349,
      "learning_rate": 9.495943580333905e-06,
      "loss": 1.4699,
      "step": 8890
    },
    {
      "epoch": 1.6016212564737673,
      "grad_norm": 1.5358400344848633,
      "learning_rate": 9.487642050974526e-06,
      "loss": 1.4864,
      "step": 8891
    },
    {
      "epoch": 1.6018013960819635,
      "grad_norm": 1.7625243663787842,
      "learning_rate": 9.479343771496768e-06,
      "loss": 1.6111,
      "step": 8892
    },
    {
      "epoch": 1.6019815356901599,
      "grad_norm": 1.5033992528915405,
      "learning_rate": 9.471048742566313e-06,
      "loss": 1.2479,
      "step": 8893
    },
    {
      "epoch": 1.602161675298356,
      "grad_norm": 1.6612857580184937,
      "learning_rate": 9.462756964848591e-06,
      "loss": 1.3696,
      "step": 8894
    },
    {
      "epoch": 1.6023418149065525,
      "grad_norm": 1.4729400873184204,
      "learning_rate": 9.454468439008745e-06,
      "loss": 1.3871,
      "step": 8895
    },
    {
      "epoch": 1.6025219545147489,
      "grad_norm": 1.7338024377822876,
      "learning_rate": 9.446183165711708e-06,
      "loss": 1.5381,
      "step": 8896
    },
    {
      "epoch": 1.6027020941229453,
      "grad_norm": 1.5807453393936157,
      "learning_rate": 9.437901145622102e-06,
      "loss": 1.461,
      "step": 8897
    },
    {
      "epoch": 1.6028822337311417,
      "grad_norm": 1.515012502670288,
      "learning_rate": 9.429622379404308e-06,
      "loss": 1.1404,
      "step": 8898
    },
    {
      "epoch": 1.603062373339338,
      "grad_norm": 1.58295738697052,
      "learning_rate": 9.421346867722447e-06,
      "loss": 1.2525,
      "step": 8899
    },
    {
      "epoch": 1.6032425129475345,
      "grad_norm": 1.4897652864456177,
      "learning_rate": 9.413074611240374e-06,
      "loss": 1.2147,
      "step": 8900
    },
    {
      "epoch": 1.6034226525557307,
      "grad_norm": 1.432344675064087,
      "learning_rate": 9.404805610621687e-06,
      "loss": 1.2557,
      "step": 8901
    },
    {
      "epoch": 1.603602792163927,
      "grad_norm": 1.4904383420944214,
      "learning_rate": 9.396539866529718e-06,
      "loss": 2.1437,
      "step": 8902
    },
    {
      "epoch": 1.6037829317721233,
      "grad_norm": 1.2578002214431763,
      "learning_rate": 9.388277379627546e-06,
      "loss": 1.4787,
      "step": 8903
    },
    {
      "epoch": 1.6039630713803197,
      "grad_norm": 1.3708674907684326,
      "learning_rate": 9.380018150577975e-06,
      "loss": 1.8865,
      "step": 8904
    },
    {
      "epoch": 1.604143210988516,
      "grad_norm": 1.3499674797058105,
      "learning_rate": 9.37176218004357e-06,
      "loss": 1.6424,
      "step": 8905
    },
    {
      "epoch": 1.6043233505967125,
      "grad_norm": 1.3097155094146729,
      "learning_rate": 9.36350946868661e-06,
      "loss": 1.6283,
      "step": 8906
    },
    {
      "epoch": 1.604503490204909,
      "grad_norm": 1.6581426858901978,
      "learning_rate": 9.355260017169131e-06,
      "loss": 2.1028,
      "step": 8907
    },
    {
      "epoch": 1.6046836298131053,
      "grad_norm": 1.4766243696212769,
      "learning_rate": 9.347013826152895e-06,
      "loss": 1.7667,
      "step": 8908
    },
    {
      "epoch": 1.6048637694213015,
      "grad_norm": 1.5135400295257568,
      "learning_rate": 9.33877089629941e-06,
      "loss": 1.6915,
      "step": 8909
    },
    {
      "epoch": 1.605043909029498,
      "grad_norm": 1.5758635997772217,
      "learning_rate": 9.330531228269917e-06,
      "loss": 2.1053,
      "step": 8910
    },
    {
      "epoch": 1.605224048637694,
      "grad_norm": 2.0041654109954834,
      "learning_rate": 9.322294822725403e-06,
      "loss": 2.1725,
      "step": 8911
    },
    {
      "epoch": 1.6054041882458905,
      "grad_norm": 1.4589606523513794,
      "learning_rate": 9.31406168032658e-06,
      "loss": 1.6749,
      "step": 8912
    },
    {
      "epoch": 1.605584327854087,
      "grad_norm": 1.4244757890701294,
      "learning_rate": 9.3058318017339e-06,
      "loss": 1.6318,
      "step": 8913
    },
    {
      "epoch": 1.6057644674622833,
      "grad_norm": 1.3773863315582275,
      "learning_rate": 9.297605187607584e-06,
      "loss": 1.239,
      "step": 8914
    },
    {
      "epoch": 1.6059446070704797,
      "grad_norm": 1.3393505811691284,
      "learning_rate": 9.289381838607552e-06,
      "loss": 1.2184,
      "step": 8915
    },
    {
      "epoch": 1.606124746678676,
      "grad_norm": 1.317397117614746,
      "learning_rate": 9.281161755393486e-06,
      "loss": 1.1994,
      "step": 8916
    },
    {
      "epoch": 1.6063048862868723,
      "grad_norm": 1.5283825397491455,
      "learning_rate": 9.27294493862479e-06,
      "loss": 1.2776,
      "step": 8917
    },
    {
      "epoch": 1.6064850258950687,
      "grad_norm": 1.5297576189041138,
      "learning_rate": 9.264731388960595e-06,
      "loss": 1.478,
      "step": 8918
    },
    {
      "epoch": 1.6066651655032649,
      "grad_norm": 1.592772364616394,
      "learning_rate": 9.256521107059834e-06,
      "loss": 1.5737,
      "step": 8919
    },
    {
      "epoch": 1.6068453051114613,
      "grad_norm": 1.403741478919983,
      "learning_rate": 9.248314093581095e-06,
      "loss": 1.2837,
      "step": 8920
    },
    {
      "epoch": 1.6070254447196577,
      "grad_norm": 1.459051251411438,
      "learning_rate": 9.240110349182745e-06,
      "loss": 1.4449,
      "step": 8921
    },
    {
      "epoch": 1.607205584327854,
      "grad_norm": 1.3451504707336426,
      "learning_rate": 9.231909874522893e-06,
      "loss": 1.2766,
      "step": 8922
    },
    {
      "epoch": 1.6073857239360505,
      "grad_norm": 1.545535922050476,
      "learning_rate": 9.22371267025936e-06,
      "loss": 1.5564,
      "step": 8923
    },
    {
      "epoch": 1.607565863544247,
      "grad_norm": 1.5190519094467163,
      "learning_rate": 9.21551873704975e-06,
      "loss": 1.4685,
      "step": 8924
    },
    {
      "epoch": 1.607746003152443,
      "grad_norm": 1.5024069547653198,
      "learning_rate": 9.207328075551359e-06,
      "loss": 1.4392,
      "step": 8925
    },
    {
      "epoch": 1.6079261427606395,
      "grad_norm": 1.4219722747802734,
      "learning_rate": 9.199140686421237e-06,
      "loss": 1.4907,
      "step": 8926
    },
    {
      "epoch": 1.6081062823688357,
      "grad_norm": 1.3777090311050415,
      "learning_rate": 9.190956570316167e-06,
      "loss": 1.4123,
      "step": 8927
    },
    {
      "epoch": 1.608286421977032,
      "grad_norm": 1.3739651441574097,
      "learning_rate": 9.182775727892696e-06,
      "loss": 1.257,
      "step": 8928
    },
    {
      "epoch": 1.6084665615852285,
      "grad_norm": 1.3471442461013794,
      "learning_rate": 9.174598159807073e-06,
      "loss": 1.3525,
      "step": 8929
    },
    {
      "epoch": 1.608646701193425,
      "grad_norm": 1.5186573266983032,
      "learning_rate": 9.166423866715302e-06,
      "loss": 1.4414,
      "step": 8930
    },
    {
      "epoch": 1.6088268408016213,
      "grad_norm": 1.4564911127090454,
      "learning_rate": 9.15825284927312e-06,
      "loss": 1.3723,
      "step": 8931
    },
    {
      "epoch": 1.6090069804098177,
      "grad_norm": 1.4761847257614136,
      "learning_rate": 9.150085108136003e-06,
      "loss": 1.4438,
      "step": 8932
    },
    {
      "epoch": 1.6091871200180141,
      "grad_norm": 1.5565615892410278,
      "learning_rate": 9.141920643959156e-06,
      "loss": 1.5325,
      "step": 8933
    },
    {
      "epoch": 1.6093672596262103,
      "grad_norm": 1.4822925329208374,
      "learning_rate": 9.133759457397534e-06,
      "loss": 1.492,
      "step": 8934
    },
    {
      "epoch": 1.6095473992344067,
      "grad_norm": 1.4930733442306519,
      "learning_rate": 9.125601549105827e-06,
      "loss": 1.3302,
      "step": 8935
    },
    {
      "epoch": 1.609727538842603,
      "grad_norm": 1.7272508144378662,
      "learning_rate": 9.117446919738453e-06,
      "loss": 1.5707,
      "step": 8936
    },
    {
      "epoch": 1.6099076784507993,
      "grad_norm": 1.5761549472808838,
      "learning_rate": 9.10929556994956e-06,
      "loss": 1.3571,
      "step": 8937
    },
    {
      "epoch": 1.6100878180589957,
      "grad_norm": 1.6181126832962036,
      "learning_rate": 9.101147500393076e-06,
      "loss": 1.423,
      "step": 8938
    },
    {
      "epoch": 1.610267957667192,
      "grad_norm": 1.5674889087677002,
      "learning_rate": 9.093002711722615e-06,
      "loss": 1.4229,
      "step": 8939
    },
    {
      "epoch": 1.6104480972753885,
      "grad_norm": 1.5214717388153076,
      "learning_rate": 9.084861204591549e-06,
      "loss": 1.2723,
      "step": 8940
    },
    {
      "epoch": 1.610628236883585,
      "grad_norm": 1.5210275650024414,
      "learning_rate": 9.07672297965299e-06,
      "loss": 1.3522,
      "step": 8941
    },
    {
      "epoch": 1.610808376491781,
      "grad_norm": 1.4901388883590698,
      "learning_rate": 9.068588037559782e-06,
      "loss": 1.1276,
      "step": 8942
    },
    {
      "epoch": 1.6109885160999775,
      "grad_norm": 1.5613468885421753,
      "learning_rate": 9.060456378964504e-06,
      "loss": 1.2203,
      "step": 8943
    },
    {
      "epoch": 1.6111686557081737,
      "grad_norm": 1.7145702838897705,
      "learning_rate": 9.052328004519473e-06,
      "loss": 1.4383,
      "step": 8944
    },
    {
      "epoch": 1.61134879531637,
      "grad_norm": 1.5754945278167725,
      "learning_rate": 9.044202914876748e-06,
      "loss": 1.1826,
      "step": 8945
    },
    {
      "epoch": 1.6115289349245665,
      "grad_norm": 1.55842924118042,
      "learning_rate": 9.0360811106881e-06,
      "loss": 1.2462,
      "step": 8946
    },
    {
      "epoch": 1.611709074532763,
      "grad_norm": 1.5444365739822388,
      "learning_rate": 9.027962592605089e-06,
      "loss": 1.4904,
      "step": 8947
    },
    {
      "epoch": 1.6118892141409593,
      "grad_norm": 1.647307276725769,
      "learning_rate": 9.01984736127896e-06,
      "loss": 1.4093,
      "step": 8948
    },
    {
      "epoch": 1.6120693537491557,
      "grad_norm": 1.537060260772705,
      "learning_rate": 9.011735417360712e-06,
      "loss": 1.2332,
      "step": 8949
    },
    {
      "epoch": 1.612249493357352,
      "grad_norm": 1.8365521430969238,
      "learning_rate": 9.003626761501089e-06,
      "loss": 1.6036,
      "step": 8950
    },
    {
      "epoch": 1.6124296329655483,
      "grad_norm": 1.367088794708252,
      "learning_rate": 8.995521394350542e-06,
      "loss": 1.0866,
      "step": 8951
    },
    {
      "epoch": 1.6126097725737445,
      "grad_norm": 1.4737699031829834,
      "learning_rate": 8.987419316559326e-06,
      "loss": 1.9999,
      "step": 8952
    },
    {
      "epoch": 1.612789912181941,
      "grad_norm": 1.35165274143219,
      "learning_rate": 8.979320528777341e-06,
      "loss": 1.7983,
      "step": 8953
    },
    {
      "epoch": 1.6129700517901373,
      "grad_norm": 1.4746110439300537,
      "learning_rate": 8.971225031654278e-06,
      "loss": 1.7234,
      "step": 8954
    },
    {
      "epoch": 1.6131501913983337,
      "grad_norm": 1.5031001567840576,
      "learning_rate": 8.963132825839554e-06,
      "loss": 2.0904,
      "step": 8955
    },
    {
      "epoch": 1.6133303310065301,
      "grad_norm": 1.5136126279830933,
      "learning_rate": 8.95504391198233e-06,
      "loss": 1.8051,
      "step": 8956
    },
    {
      "epoch": 1.6135104706147265,
      "grad_norm": 1.6458314657211304,
      "learning_rate": 8.9469582907315e-06,
      "loss": 1.9362,
      "step": 8957
    },
    {
      "epoch": 1.613690610222923,
      "grad_norm": 1.4249788522720337,
      "learning_rate": 8.938875962735676e-06,
      "loss": 1.6812,
      "step": 8958
    },
    {
      "epoch": 1.6138707498311191,
      "grad_norm": 1.5732581615447998,
      "learning_rate": 8.93079692864322e-06,
      "loss": 1.8222,
      "step": 8959
    },
    {
      "epoch": 1.6140508894393155,
      "grad_norm": 1.5990383625030518,
      "learning_rate": 8.922721189102217e-06,
      "loss": 1.5322,
      "step": 8960
    },
    {
      "epoch": 1.6142310290475117,
      "grad_norm": 1.8646399974822998,
      "learning_rate": 8.914648744760528e-06,
      "loss": 2.2475,
      "step": 8961
    },
    {
      "epoch": 1.614411168655708,
      "grad_norm": 1.4296687841415405,
      "learning_rate": 8.906579596265707e-06,
      "loss": 1.5818,
      "step": 8962
    },
    {
      "epoch": 1.6145913082639045,
      "grad_norm": 1.4705822467803955,
      "learning_rate": 8.898513744265052e-06,
      "loss": 1.5749,
      "step": 8963
    },
    {
      "epoch": 1.614771447872101,
      "grad_norm": 1.3082259893417358,
      "learning_rate": 8.890451189405618e-06,
      "loss": 1.2823,
      "step": 8964
    },
    {
      "epoch": 1.6149515874802973,
      "grad_norm": 1.5380620956420898,
      "learning_rate": 8.882391932334139e-06,
      "loss": 1.4484,
      "step": 8965
    },
    {
      "epoch": 1.6151317270884937,
      "grad_norm": 1.3446846008300781,
      "learning_rate": 8.874335973697168e-06,
      "loss": 1.0838,
      "step": 8966
    },
    {
      "epoch": 1.61531186669669,
      "grad_norm": 1.515052080154419,
      "learning_rate": 8.866283314140927e-06,
      "loss": 1.5796,
      "step": 8967
    },
    {
      "epoch": 1.6154920063048863,
      "grad_norm": 1.3485757112503052,
      "learning_rate": 8.858233954311406e-06,
      "loss": 1.252,
      "step": 8968
    },
    {
      "epoch": 1.6156721459130825,
      "grad_norm": 1.3139350414276123,
      "learning_rate": 8.850187894854306e-06,
      "loss": 1.1304,
      "step": 8969
    },
    {
      "epoch": 1.615852285521279,
      "grad_norm": 1.3755433559417725,
      "learning_rate": 8.842145136415103e-06,
      "loss": 1.2409,
      "step": 8970
    },
    {
      "epoch": 1.6160324251294753,
      "grad_norm": 1.446356177330017,
      "learning_rate": 8.834105679638972e-06,
      "loss": 1.4494,
      "step": 8971
    },
    {
      "epoch": 1.6162125647376717,
      "grad_norm": 1.4663615226745605,
      "learning_rate": 8.826069525170827e-06,
      "loss": 1.4345,
      "step": 8972
    },
    {
      "epoch": 1.6163927043458681,
      "grad_norm": 1.426906943321228,
      "learning_rate": 8.818036673655333e-06,
      "loss": 1.3279,
      "step": 8973
    },
    {
      "epoch": 1.6165728439540645,
      "grad_norm": 1.5902485847473145,
      "learning_rate": 8.810007125736874e-06,
      "loss": 1.2541,
      "step": 8974
    },
    {
      "epoch": 1.6167529835622607,
      "grad_norm": 1.474242925643921,
      "learning_rate": 8.801980882059586e-06,
      "loss": 1.3267,
      "step": 8975
    },
    {
      "epoch": 1.6169331231704571,
      "grad_norm": 1.34149169921875,
      "learning_rate": 8.793957943267323e-06,
      "loss": 1.1325,
      "step": 8976
    },
    {
      "epoch": 1.6171132627786533,
      "grad_norm": 1.5356731414794922,
      "learning_rate": 8.785938310003683e-06,
      "loss": 1.5951,
      "step": 8977
    },
    {
      "epoch": 1.6172934023868497,
      "grad_norm": 1.5027077198028564,
      "learning_rate": 8.777921982911996e-06,
      "loss": 1.3397,
      "step": 8978
    },
    {
      "epoch": 1.6174735419950461,
      "grad_norm": 1.3907259702682495,
      "learning_rate": 8.769908962635315e-06,
      "loss": 1.1936,
      "step": 8979
    },
    {
      "epoch": 1.6176536816032425,
      "grad_norm": 1.6222527027130127,
      "learning_rate": 8.76189924981647e-06,
      "loss": 1.6759,
      "step": 8980
    },
    {
      "epoch": 1.617833821211439,
      "grad_norm": 1.7455780506134033,
      "learning_rate": 8.753892845097978e-06,
      "loss": 1.6083,
      "step": 8981
    },
    {
      "epoch": 1.6180139608196353,
      "grad_norm": 1.456418752670288,
      "learning_rate": 8.745889749122115e-06,
      "loss": 1.4658,
      "step": 8982
    },
    {
      "epoch": 1.6181941004278315,
      "grad_norm": 1.4400017261505127,
      "learning_rate": 8.737889962530865e-06,
      "loss": 1.4093,
      "step": 8983
    },
    {
      "epoch": 1.618374240036028,
      "grad_norm": 1.4574759006500244,
      "learning_rate": 8.729893485966012e-06,
      "loss": 1.4719,
      "step": 8984
    },
    {
      "epoch": 1.6185543796442243,
      "grad_norm": 1.6979870796203613,
      "learning_rate": 8.721900320068992e-06,
      "loss": 1.3265,
      "step": 8985
    },
    {
      "epoch": 1.6187345192524205,
      "grad_norm": 1.6142722368240356,
      "learning_rate": 8.713910465481017e-06,
      "loss": 1.5635,
      "step": 8986
    },
    {
      "epoch": 1.618914658860617,
      "grad_norm": 1.4823176860809326,
      "learning_rate": 8.70592392284304e-06,
      "loss": 1.3805,
      "step": 8987
    },
    {
      "epoch": 1.6190947984688133,
      "grad_norm": 1.5214725732803345,
      "learning_rate": 8.69794069279572e-06,
      "loss": 1.4628,
      "step": 8988
    },
    {
      "epoch": 1.6192749380770097,
      "grad_norm": 1.5264830589294434,
      "learning_rate": 8.689960775979489e-06,
      "loss": 1.4645,
      "step": 8989
    },
    {
      "epoch": 1.6194550776852061,
      "grad_norm": 1.5529534816741943,
      "learning_rate": 8.68198417303448e-06,
      "loss": 1.506,
      "step": 8990
    },
    {
      "epoch": 1.6196352172934025,
      "grad_norm": 1.557588815689087,
      "learning_rate": 8.67401088460058e-06,
      "loss": 1.4007,
      "step": 8991
    },
    {
      "epoch": 1.6198153569015987,
      "grad_norm": 1.444764494895935,
      "learning_rate": 8.666040911317397e-06,
      "loss": 1.3443,
      "step": 8992
    },
    {
      "epoch": 1.6199954965097951,
      "grad_norm": 1.6465357542037964,
      "learning_rate": 8.658074253824266e-06,
      "loss": 1.421,
      "step": 8993
    },
    {
      "epoch": 1.6201756361179913,
      "grad_norm": 1.5755488872528076,
      "learning_rate": 8.650110912760296e-06,
      "loss": 1.478,
      "step": 8994
    },
    {
      "epoch": 1.6203557757261877,
      "grad_norm": 1.746640682220459,
      "learning_rate": 8.642150888764289e-06,
      "loss": 1.7575,
      "step": 8995
    },
    {
      "epoch": 1.6205359153343841,
      "grad_norm": 1.4601953029632568,
      "learning_rate": 8.634194182474786e-06,
      "loss": 1.3062,
      "step": 8996
    },
    {
      "epoch": 1.6207160549425805,
      "grad_norm": 1.610750675201416,
      "learning_rate": 8.626240794530083e-06,
      "loss": 1.6286,
      "step": 8997
    },
    {
      "epoch": 1.620896194550777,
      "grad_norm": 1.4666494131088257,
      "learning_rate": 8.61829072556819e-06,
      "loss": 1.3697,
      "step": 8998
    },
    {
      "epoch": 1.6210763341589733,
      "grad_norm": 1.5435643196105957,
      "learning_rate": 8.610343976226858e-06,
      "loss": 1.4625,
      "step": 8999
    },
    {
      "epoch": 1.6212564737671695,
      "grad_norm": 1.5185898542404175,
      "learning_rate": 8.602400547143574e-06,
      "loss": 1.2593,
      "step": 9000
    },
    {
      "epoch": 1.621436613375366,
      "grad_norm": 1.493996024131775,
      "learning_rate": 8.594460438955548e-06,
      "loss": 1.1408,
      "step": 9001
    },
    {
      "epoch": 1.6216167529835621,
      "grad_norm": 1.2784415483474731,
      "learning_rate": 8.586523652299733e-06,
      "loss": 1.6844,
      "step": 9002
    },
    {
      "epoch": 1.6217968925917585,
      "grad_norm": 1.3722058534622192,
      "learning_rate": 8.57859018781283e-06,
      "loss": 1.712,
      "step": 9003
    },
    {
      "epoch": 1.621977032199955,
      "grad_norm": 1.3358838558197021,
      "learning_rate": 8.57066004613124e-06,
      "loss": 1.8055,
      "step": 9004
    },
    {
      "epoch": 1.6221571718081513,
      "grad_norm": 1.402470588684082,
      "learning_rate": 8.562733227891128e-06,
      "loss": 1.8546,
      "step": 9005
    },
    {
      "epoch": 1.6223373114163477,
      "grad_norm": 1.435928225517273,
      "learning_rate": 8.55480973372837e-06,
      "loss": 1.861,
      "step": 9006
    },
    {
      "epoch": 1.6225174510245441,
      "grad_norm": 1.5125575065612793,
      "learning_rate": 8.546889564278587e-06,
      "loss": 1.9096,
      "step": 9007
    },
    {
      "epoch": 1.6226975906327403,
      "grad_norm": 1.5165061950683594,
      "learning_rate": 8.538972720177135e-06,
      "loss": 1.6049,
      "step": 9008
    },
    {
      "epoch": 1.6228777302409367,
      "grad_norm": 1.7191143035888672,
      "learning_rate": 8.531059202059094e-06,
      "loss": 2.1135,
      "step": 9009
    },
    {
      "epoch": 1.623057869849133,
      "grad_norm": 1.563788652420044,
      "learning_rate": 8.52314901055929e-06,
      "loss": 1.7805,
      "step": 9010
    },
    {
      "epoch": 1.6232380094573293,
      "grad_norm": 1.6980916261672974,
      "learning_rate": 8.515242146312252e-06,
      "loss": 1.954,
      "step": 9011
    },
    {
      "epoch": 1.6234181490655257,
      "grad_norm": 1.6332604885101318,
      "learning_rate": 8.507338609952298e-06,
      "loss": 1.717,
      "step": 9012
    },
    {
      "epoch": 1.6235982886737221,
      "grad_norm": 1.4301986694335938,
      "learning_rate": 8.49943840211343e-06,
      "loss": 1.4415,
      "step": 9013
    },
    {
      "epoch": 1.6237784282819185,
      "grad_norm": 1.519019365310669,
      "learning_rate": 8.491541523429397e-06,
      "loss": 1.5076,
      "step": 9014
    },
    {
      "epoch": 1.623958567890115,
      "grad_norm": 1.4183235168457031,
      "learning_rate": 8.483647974533688e-06,
      "loss": 1.4486,
      "step": 9015
    },
    {
      "epoch": 1.6241387074983114,
      "grad_norm": 1.4618350267410278,
      "learning_rate": 8.4757577560595e-06,
      "loss": 1.4031,
      "step": 9016
    },
    {
      "epoch": 1.6243188471065075,
      "grad_norm": 1.3977642059326172,
      "learning_rate": 8.467870868639826e-06,
      "loss": 1.248,
      "step": 9017
    },
    {
      "epoch": 1.624498986714704,
      "grad_norm": 1.4015337228775024,
      "learning_rate": 8.459987312907308e-06,
      "loss": 1.3719,
      "step": 9018
    },
    {
      "epoch": 1.6246791263229001,
      "grad_norm": 1.4100414514541626,
      "learning_rate": 8.452107089494376e-06,
      "loss": 1.4763,
      "step": 9019
    },
    {
      "epoch": 1.6248592659310965,
      "grad_norm": 1.5313351154327393,
      "learning_rate": 8.444230199033166e-06,
      "loss": 1.6206,
      "step": 9020
    },
    {
      "epoch": 1.625039405539293,
      "grad_norm": 1.4793330430984497,
      "learning_rate": 8.436356642155573e-06,
      "loss": 1.3862,
      "step": 9021
    },
    {
      "epoch": 1.6252195451474893,
      "grad_norm": 1.4847387075424194,
      "learning_rate": 8.428486419493208e-06,
      "loss": 1.475,
      "step": 9022
    },
    {
      "epoch": 1.6253996847556857,
      "grad_norm": 1.28648042678833,
      "learning_rate": 8.42061953167742e-06,
      "loss": 1.2547,
      "step": 9023
    },
    {
      "epoch": 1.6255798243638822,
      "grad_norm": 1.4302302598953247,
      "learning_rate": 8.412755979339275e-06,
      "loss": 1.3175,
      "step": 9024
    },
    {
      "epoch": 1.6257599639720783,
      "grad_norm": 1.51126229763031,
      "learning_rate": 8.404895763109578e-06,
      "loss": 1.2884,
      "step": 9025
    },
    {
      "epoch": 1.6259401035802747,
      "grad_norm": 1.4226932525634766,
      "learning_rate": 8.397038883618896e-06,
      "loss": 1.2495,
      "step": 9026
    },
    {
      "epoch": 1.626120243188471,
      "grad_norm": 1.3277056217193604,
      "learning_rate": 8.38918534149749e-06,
      "loss": 1.2746,
      "step": 9027
    },
    {
      "epoch": 1.6263003827966673,
      "grad_norm": 1.2354819774627686,
      "learning_rate": 8.381335137375367e-06,
      "loss": 1.116,
      "step": 9028
    },
    {
      "epoch": 1.6264805224048637,
      "grad_norm": 1.5247491598129272,
      "learning_rate": 8.37348827188228e-06,
      "loss": 1.4898,
      "step": 9029
    },
    {
      "epoch": 1.6266606620130601,
      "grad_norm": 1.4630602598190308,
      "learning_rate": 8.365644745647666e-06,
      "loss": 1.5886,
      "step": 9030
    },
    {
      "epoch": 1.6268408016212565,
      "grad_norm": 1.4480348825454712,
      "learning_rate": 8.357804559300758e-06,
      "loss": 1.447,
      "step": 9031
    },
    {
      "epoch": 1.627020941229453,
      "grad_norm": 1.3790920972824097,
      "learning_rate": 8.349967713470491e-06,
      "loss": 1.2481,
      "step": 9032
    },
    {
      "epoch": 1.6272010808376491,
      "grad_norm": 1.5948432683944702,
      "learning_rate": 8.342134208785523e-06,
      "loss": 1.5668,
      "step": 9033
    },
    {
      "epoch": 1.6273812204458455,
      "grad_norm": 1.5799834728240967,
      "learning_rate": 8.334304045874247e-06,
      "loss": 1.407,
      "step": 9034
    },
    {
      "epoch": 1.6275613600540417,
      "grad_norm": 1.5855259895324707,
      "learning_rate": 8.326477225364821e-06,
      "loss": 1.5919,
      "step": 9035
    },
    {
      "epoch": 1.6277414996622381,
      "grad_norm": 1.464565634727478,
      "learning_rate": 8.318653747885092e-06,
      "loss": 1.2988,
      "step": 9036
    },
    {
      "epoch": 1.6279216392704345,
      "grad_norm": 1.5801305770874023,
      "learning_rate": 8.310833614062651e-06,
      "loss": 1.4633,
      "step": 9037
    },
    {
      "epoch": 1.628101778878631,
      "grad_norm": 1.6051896810531616,
      "learning_rate": 8.30301682452484e-06,
      "loss": 1.4006,
      "step": 9038
    },
    {
      "epoch": 1.6282819184868274,
      "grad_norm": 1.6969918012619019,
      "learning_rate": 8.295203379898703e-06,
      "loss": 1.539,
      "step": 9039
    },
    {
      "epoch": 1.6284620580950238,
      "grad_norm": 1.599530816078186,
      "learning_rate": 8.28739328081104e-06,
      "loss": 1.4317,
      "step": 9040
    },
    {
      "epoch": 1.62864219770322,
      "grad_norm": 1.5828198194503784,
      "learning_rate": 8.279586527888373e-06,
      "loss": 1.3214,
      "step": 9041
    },
    {
      "epoch": 1.6288223373114163,
      "grad_norm": 1.5459054708480835,
      "learning_rate": 8.271783121756949e-06,
      "loss": 1.4288,
      "step": 9042
    },
    {
      "epoch": 1.6290024769196128,
      "grad_norm": 1.5309839248657227,
      "learning_rate": 8.263983063042757e-06,
      "loss": 1.4071,
      "step": 9043
    },
    {
      "epoch": 1.629182616527809,
      "grad_norm": 1.6980780363082886,
      "learning_rate": 8.256186352371509e-06,
      "loss": 1.3543,
      "step": 9044
    },
    {
      "epoch": 1.6293627561360053,
      "grad_norm": 1.4955205917358398,
      "learning_rate": 8.24839299036867e-06,
      "loss": 1.325,
      "step": 9045
    },
    {
      "epoch": 1.6295428957442017,
      "grad_norm": 1.580418348312378,
      "learning_rate": 8.240602977659406e-06,
      "loss": 1.334,
      "step": 9046
    },
    {
      "epoch": 1.6297230353523982,
      "grad_norm": 1.582728385925293,
      "learning_rate": 8.23281631486863e-06,
      "loss": 1.5868,
      "step": 9047
    },
    {
      "epoch": 1.6299031749605946,
      "grad_norm": 1.5436362028121948,
      "learning_rate": 8.225033002620974e-06,
      "loss": 1.2998,
      "step": 9048
    },
    {
      "epoch": 1.630083314568791,
      "grad_norm": 1.5646473169326782,
      "learning_rate": 8.217253041540845e-06,
      "loss": 1.3856,
      "step": 9049
    },
    {
      "epoch": 1.6302634541769871,
      "grad_norm": 1.6701630353927612,
      "learning_rate": 8.209476432252317e-06,
      "loss": 1.471,
      "step": 9050
    },
    {
      "epoch": 1.6304435937851836,
      "grad_norm": 1.4872112274169922,
      "learning_rate": 8.201703175379234e-06,
      "loss": 1.1637,
      "step": 9051
    },
    {
      "epoch": 1.6306237333933797,
      "grad_norm": 1.4328562021255493,
      "learning_rate": 8.193933271545163e-06,
      "loss": 1.783,
      "step": 9052
    },
    {
      "epoch": 1.6308038730015761,
      "grad_norm": 1.2777864933013916,
      "learning_rate": 8.18616672137339e-06,
      "loss": 1.631,
      "step": 9053
    },
    {
      "epoch": 1.6309840126097725,
      "grad_norm": 1.324245810508728,
      "learning_rate": 8.178403525486966e-06,
      "loss": 1.6477,
      "step": 9054
    },
    {
      "epoch": 1.631164152217969,
      "grad_norm": 1.5102808475494385,
      "learning_rate": 8.170643684508638e-06,
      "loss": 2.1058,
      "step": 9055
    },
    {
      "epoch": 1.6313442918261654,
      "grad_norm": 1.4648336172103882,
      "learning_rate": 8.162887199060903e-06,
      "loss": 1.8486,
      "step": 9056
    },
    {
      "epoch": 1.6315244314343618,
      "grad_norm": 1.5705616474151611,
      "learning_rate": 8.15513406976598e-06,
      "loss": 1.9165,
      "step": 9057
    },
    {
      "epoch": 1.631704571042558,
      "grad_norm": 1.4184564352035522,
      "learning_rate": 8.147384297245803e-06,
      "loss": 1.7502,
      "step": 9058
    },
    {
      "epoch": 1.6318847106507544,
      "grad_norm": 1.6010007858276367,
      "learning_rate": 8.139637882122087e-06,
      "loss": 1.8379,
      "step": 9059
    },
    {
      "epoch": 1.6320648502589505,
      "grad_norm": 1.7304044961929321,
      "learning_rate": 8.131894825016234e-06,
      "loss": 1.7166,
      "step": 9060
    },
    {
      "epoch": 1.632244989867147,
      "grad_norm": 1.8300931453704834,
      "learning_rate": 8.124155126549393e-06,
      "loss": 1.8421,
      "step": 9061
    },
    {
      "epoch": 1.6324251294753434,
      "grad_norm": 1.4604002237319946,
      "learning_rate": 8.116418787342405e-06,
      "loss": 1.5491,
      "step": 9062
    },
    {
      "epoch": 1.6326052690835398,
      "grad_norm": 1.4102141857147217,
      "learning_rate": 8.10868580801592e-06,
      "loss": 1.5145,
      "step": 9063
    },
    {
      "epoch": 1.6327854086917362,
      "grad_norm": 1.4036797285079956,
      "learning_rate": 8.10095618919025e-06,
      "loss": 1.3797,
      "step": 9064
    },
    {
      "epoch": 1.6329655482999326,
      "grad_norm": 1.4088903665542603,
      "learning_rate": 8.093229931485463e-06,
      "loss": 1.4173,
      "step": 9065
    },
    {
      "epoch": 1.6331456879081288,
      "grad_norm": 1.513932704925537,
      "learning_rate": 8.085507035521367e-06,
      "loss": 1.3483,
      "step": 9066
    },
    {
      "epoch": 1.6333258275163252,
      "grad_norm": 1.5356560945510864,
      "learning_rate": 8.07778750191746e-06,
      "loss": 1.4576,
      "step": 9067
    },
    {
      "epoch": 1.6335059671245213,
      "grad_norm": 1.4211124181747437,
      "learning_rate": 8.070071331293039e-06,
      "loss": 1.3563,
      "step": 9068
    },
    {
      "epoch": 1.6336861067327177,
      "grad_norm": 1.4256093502044678,
      "learning_rate": 8.062358524267071e-06,
      "loss": 1.5777,
      "step": 9069
    },
    {
      "epoch": 1.6338662463409142,
      "grad_norm": 1.4369701147079468,
      "learning_rate": 8.05464908145827e-06,
      "loss": 1.3761,
      "step": 9070
    },
    {
      "epoch": 1.6340463859491106,
      "grad_norm": 1.608688235282898,
      "learning_rate": 8.046943003485097e-06,
      "loss": 1.5038,
      "step": 9071
    },
    {
      "epoch": 1.634226525557307,
      "grad_norm": 1.4133819341659546,
      "learning_rate": 8.039240290965716e-06,
      "loss": 1.2728,
      "step": 9072
    },
    {
      "epoch": 1.6344066651655034,
      "grad_norm": 1.513113260269165,
      "learning_rate": 8.031540944518046e-06,
      "loss": 1.4137,
      "step": 9073
    },
    {
      "epoch": 1.6345868047736998,
      "grad_norm": 1.3872692584991455,
      "learning_rate": 8.023844964759713e-06,
      "loss": 1.3946,
      "step": 9074
    },
    {
      "epoch": 1.634766944381896,
      "grad_norm": 1.3328146934509277,
      "learning_rate": 8.016152352308093e-06,
      "loss": 1.2578,
      "step": 9075
    },
    {
      "epoch": 1.6349470839900924,
      "grad_norm": 1.6357322931289673,
      "learning_rate": 8.008463107780274e-06,
      "loss": 1.6611,
      "step": 9076
    },
    {
      "epoch": 1.6351272235982885,
      "grad_norm": 1.5515234470367432,
      "learning_rate": 8.000777231793094e-06,
      "loss": 1.2876,
      "step": 9077
    },
    {
      "epoch": 1.635307363206485,
      "grad_norm": 1.6367772817611694,
      "learning_rate": 7.993094724963113e-06,
      "loss": 1.6553,
      "step": 9078
    },
    {
      "epoch": 1.6354875028146814,
      "grad_norm": 1.5051878690719604,
      "learning_rate": 7.985415587906615e-06,
      "loss": 1.449,
      "step": 9079
    },
    {
      "epoch": 1.6356676424228778,
      "grad_norm": 1.4763249158859253,
      "learning_rate": 7.97773982123961e-06,
      "loss": 1.3077,
      "step": 9080
    },
    {
      "epoch": 1.6358477820310742,
      "grad_norm": 1.245093822479248,
      "learning_rate": 7.970067425577848e-06,
      "loss": 1.1179,
      "step": 9081
    },
    {
      "epoch": 1.6360279216392706,
      "grad_norm": 1.5718320608139038,
      "learning_rate": 7.962398401536809e-06,
      "loss": 1.4774,
      "step": 9082
    },
    {
      "epoch": 1.6362080612474668,
      "grad_norm": 1.5183449983596802,
      "learning_rate": 7.954732749731692e-06,
      "loss": 1.5541,
      "step": 9083
    },
    {
      "epoch": 1.6363882008556632,
      "grad_norm": 1.3619437217712402,
      "learning_rate": 7.94707047077743e-06,
      "loss": 1.2706,
      "step": 9084
    },
    {
      "epoch": 1.6365683404638594,
      "grad_norm": 1.5544508695602417,
      "learning_rate": 7.939411565288695e-06,
      "loss": 1.5961,
      "step": 9085
    },
    {
      "epoch": 1.6367484800720558,
      "grad_norm": 1.4584760665893555,
      "learning_rate": 7.931756033879867e-06,
      "loss": 1.3377,
      "step": 9086
    },
    {
      "epoch": 1.6369286196802522,
      "grad_norm": 1.5793532133102417,
      "learning_rate": 7.92410387716509e-06,
      "loss": 1.3234,
      "step": 9087
    },
    {
      "epoch": 1.6371087592884486,
      "grad_norm": 1.4982151985168457,
      "learning_rate": 7.916455095758208e-06,
      "loss": 1.1766,
      "step": 9088
    },
    {
      "epoch": 1.637288898896645,
      "grad_norm": 1.5931265354156494,
      "learning_rate": 7.908809690272794e-06,
      "loss": 1.4055,
      "step": 9089
    },
    {
      "epoch": 1.6374690385048414,
      "grad_norm": 1.537518858909607,
      "learning_rate": 7.901167661322162e-06,
      "loss": 1.4299,
      "step": 9090
    },
    {
      "epoch": 1.6376491781130376,
      "grad_norm": 1.6215969324111938,
      "learning_rate": 7.893529009519357e-06,
      "loss": 1.6207,
      "step": 9091
    },
    {
      "epoch": 1.637829317721234,
      "grad_norm": 1.5307815074920654,
      "learning_rate": 7.88589373547715e-06,
      "loss": 1.399,
      "step": 9092
    },
    {
      "epoch": 1.6380094573294302,
      "grad_norm": 1.554375410079956,
      "learning_rate": 7.878261839808038e-06,
      "loss": 1.4142,
      "step": 9093
    },
    {
      "epoch": 1.6381895969376266,
      "grad_norm": 1.5103033781051636,
      "learning_rate": 7.870633323124249e-06,
      "loss": 1.2846,
      "step": 9094
    },
    {
      "epoch": 1.638369736545823,
      "grad_norm": 1.5561450719833374,
      "learning_rate": 7.863008186037713e-06,
      "loss": 1.2516,
      "step": 9095
    },
    {
      "epoch": 1.6385498761540194,
      "grad_norm": 1.5411745309829712,
      "learning_rate": 7.85538642916015e-06,
      "loss": 1.3571,
      "step": 9096
    },
    {
      "epoch": 1.6387300157622158,
      "grad_norm": 1.5924540758132935,
      "learning_rate": 7.84776805310296e-06,
      "loss": 1.5043,
      "step": 9097
    },
    {
      "epoch": 1.6389101553704122,
      "grad_norm": 1.6135451793670654,
      "learning_rate": 7.840153058477284e-06,
      "loss": 1.3234,
      "step": 9098
    },
    {
      "epoch": 1.6390902949786086,
      "grad_norm": 1.5516250133514404,
      "learning_rate": 7.832541445893994e-06,
      "loss": 1.3139,
      "step": 9099
    },
    {
      "epoch": 1.6392704345868048,
      "grad_norm": 1.586106538772583,
      "learning_rate": 7.824933215963676e-06,
      "loss": 1.2976,
      "step": 9100
    },
    {
      "epoch": 1.6394505741950012,
      "grad_norm": 1.8480298519134521,
      "learning_rate": 7.817328369296689e-06,
      "loss": 1.3535,
      "step": 9101
    },
    {
      "epoch": 1.6396307138031974,
      "grad_norm": 1.4122203588485718,
      "learning_rate": 7.809726906503073e-06,
      "loss": 1.8084,
      "step": 9102
    },
    {
      "epoch": 1.6398108534113938,
      "grad_norm": 1.3309080600738525,
      "learning_rate": 7.802128828192611e-06,
      "loss": 1.6608,
      "step": 9103
    },
    {
      "epoch": 1.6399909930195902,
      "grad_norm": 1.4233009815216064,
      "learning_rate": 7.794534134974823e-06,
      "loss": 1.8228,
      "step": 9104
    },
    {
      "epoch": 1.6401711326277866,
      "grad_norm": 1.4729584455490112,
      "learning_rate": 7.786942827458949e-06,
      "loss": 1.8481,
      "step": 9105
    },
    {
      "epoch": 1.640351272235983,
      "grad_norm": 1.4456026554107666,
      "learning_rate": 7.77935490625396e-06,
      "loss": 1.7278,
      "step": 9106
    },
    {
      "epoch": 1.6405314118441794,
      "grad_norm": 1.4141380786895752,
      "learning_rate": 7.771770371968556e-06,
      "loss": 1.4236,
      "step": 9107
    },
    {
      "epoch": 1.6407115514523756,
      "grad_norm": 1.4544466733932495,
      "learning_rate": 7.764189225211166e-06,
      "loss": 1.5494,
      "step": 9108
    },
    {
      "epoch": 1.640891691060572,
      "grad_norm": 1.5620163679122925,
      "learning_rate": 7.75661146658993e-06,
      "loss": 1.7708,
      "step": 9109
    },
    {
      "epoch": 1.6410718306687682,
      "grad_norm": 1.7449952363967896,
      "learning_rate": 7.749037096712758e-06,
      "loss": 2.2241,
      "step": 9110
    },
    {
      "epoch": 1.6412519702769646,
      "grad_norm": 1.486367106437683,
      "learning_rate": 7.741466116187246e-06,
      "loss": 1.5255,
      "step": 9111
    },
    {
      "epoch": 1.641432109885161,
      "grad_norm": 1.333660364151001,
      "learning_rate": 7.733898525620747e-06,
      "loss": 1.1848,
      "step": 9112
    },
    {
      "epoch": 1.6416122494933574,
      "grad_norm": 1.441994309425354,
      "learning_rate": 7.726334325620304e-06,
      "loss": 1.5103,
      "step": 9113
    },
    {
      "epoch": 1.6417923891015538,
      "grad_norm": 1.3525357246398926,
      "learning_rate": 7.718773516792754e-06,
      "loss": 1.3706,
      "step": 9114
    },
    {
      "epoch": 1.6419725287097502,
      "grad_norm": 1.3716084957122803,
      "learning_rate": 7.71121609974459e-06,
      "loss": 1.3065,
      "step": 9115
    },
    {
      "epoch": 1.6421526683179464,
      "grad_norm": 1.2797802686691284,
      "learning_rate": 7.70366207508207e-06,
      "loss": 1.2084,
      "step": 9116
    },
    {
      "epoch": 1.6423328079261428,
      "grad_norm": 1.4668563604354858,
      "learning_rate": 7.696111443411174e-06,
      "loss": 1.6296,
      "step": 9117
    },
    {
      "epoch": 1.642512947534339,
      "grad_norm": 1.3905410766601562,
      "learning_rate": 7.688564205337606e-06,
      "loss": 1.3526,
      "step": 9118
    },
    {
      "epoch": 1.6426930871425354,
      "grad_norm": 1.450563907623291,
      "learning_rate": 7.681020361466817e-06,
      "loss": 1.4415,
      "step": 9119
    },
    {
      "epoch": 1.6428732267507318,
      "grad_norm": 1.4846205711364746,
      "learning_rate": 7.673479912403963e-06,
      "loss": 1.4199,
      "step": 9120
    },
    {
      "epoch": 1.6430533663589282,
      "grad_norm": 1.2510420083999634,
      "learning_rate": 7.665942858753932e-06,
      "loss": 1.2963,
      "step": 9121
    },
    {
      "epoch": 1.6432335059671246,
      "grad_norm": 1.455659031867981,
      "learning_rate": 7.658409201121347e-06,
      "loss": 1.4012,
      "step": 9122
    },
    {
      "epoch": 1.643413645575321,
      "grad_norm": 1.512078881263733,
      "learning_rate": 7.65087894011054e-06,
      "loss": 1.478,
      "step": 9123
    },
    {
      "epoch": 1.6435937851835172,
      "grad_norm": 1.5768029689788818,
      "learning_rate": 7.643352076325604e-06,
      "loss": 1.4387,
      "step": 9124
    },
    {
      "epoch": 1.6437739247917136,
      "grad_norm": 1.46713125705719,
      "learning_rate": 7.635828610370338e-06,
      "loss": 1.2989,
      "step": 9125
    },
    {
      "epoch": 1.6439540643999098,
      "grad_norm": 1.411866545677185,
      "learning_rate": 7.628308542848278e-06,
      "loss": 1.4271,
      "step": 9126
    },
    {
      "epoch": 1.6441342040081062,
      "grad_norm": 1.4002352952957153,
      "learning_rate": 7.620791874362643e-06,
      "loss": 1.278,
      "step": 9127
    },
    {
      "epoch": 1.6443143436163026,
      "grad_norm": 1.4821597337722778,
      "learning_rate": 7.613278605516455e-06,
      "loss": 1.3851,
      "step": 9128
    },
    {
      "epoch": 1.644494483224499,
      "grad_norm": 1.4351940155029297,
      "learning_rate": 7.6057687369124105e-06,
      "loss": 1.3596,
      "step": 9129
    },
    {
      "epoch": 1.6446746228326954,
      "grad_norm": 1.3682835102081299,
      "learning_rate": 7.59826226915295e-06,
      "loss": 1.1133,
      "step": 9130
    },
    {
      "epoch": 1.6448547624408918,
      "grad_norm": 1.66762113571167,
      "learning_rate": 7.590759202840237e-06,
      "loss": 1.678,
      "step": 9131
    },
    {
      "epoch": 1.6450349020490882,
      "grad_norm": 1.4181160926818848,
      "learning_rate": 7.583259538576149e-06,
      "loss": 1.2646,
      "step": 9132
    },
    {
      "epoch": 1.6452150416572844,
      "grad_norm": 1.5533076524734497,
      "learning_rate": 7.575763276962333e-06,
      "loss": 1.3039,
      "step": 9133
    },
    {
      "epoch": 1.6453951812654808,
      "grad_norm": 1.5733705759048462,
      "learning_rate": 7.568270418600126e-06,
      "loss": 1.3135,
      "step": 9134
    },
    {
      "epoch": 1.645575320873677,
      "grad_norm": 1.5810977220535278,
      "learning_rate": 7.5607809640905915e-06,
      "loss": 1.6116,
      "step": 9135
    },
    {
      "epoch": 1.6457554604818734,
      "grad_norm": 1.513537049293518,
      "learning_rate": 7.553294914034542e-06,
      "loss": 1.3003,
      "step": 9136
    },
    {
      "epoch": 1.6459356000900698,
      "grad_norm": 1.4928573369979858,
      "learning_rate": 7.545812269032498e-06,
      "loss": 1.199,
      "step": 9137
    },
    {
      "epoch": 1.6461157396982662,
      "grad_norm": 1.3583133220672607,
      "learning_rate": 7.538333029684719e-06,
      "loss": 1.1891,
      "step": 9138
    },
    {
      "epoch": 1.6462958793064626,
      "grad_norm": 1.3966648578643799,
      "learning_rate": 7.530857196591179e-06,
      "loss": 1.2593,
      "step": 9139
    },
    {
      "epoch": 1.646476018914659,
      "grad_norm": 1.447985053062439,
      "learning_rate": 7.523384770351588e-06,
      "loss": 1.3002,
      "step": 9140
    },
    {
      "epoch": 1.6466561585228552,
      "grad_norm": 1.4398032426834106,
      "learning_rate": 7.515915751565367e-06,
      "loss": 1.3891,
      "step": 9141
    },
    {
      "epoch": 1.6468362981310516,
      "grad_norm": 1.6720916032791138,
      "learning_rate": 7.508450140831708e-06,
      "loss": 1.4786,
      "step": 9142
    },
    {
      "epoch": 1.6470164377392478,
      "grad_norm": 1.5649749040603638,
      "learning_rate": 7.500987938749477e-06,
      "loss": 1.5055,
      "step": 9143
    },
    {
      "epoch": 1.6471965773474442,
      "grad_norm": 1.571779727935791,
      "learning_rate": 7.493529145917294e-06,
      "loss": 1.4743,
      "step": 9144
    },
    {
      "epoch": 1.6473767169556406,
      "grad_norm": 1.3726614713668823,
      "learning_rate": 7.4860737629335045e-06,
      "loss": 1.134,
      "step": 9145
    },
    {
      "epoch": 1.647556856563837,
      "grad_norm": 1.498897910118103,
      "learning_rate": 7.478621790396162e-06,
      "loss": 1.2898,
      "step": 9146
    },
    {
      "epoch": 1.6477369961720334,
      "grad_norm": 1.6689149141311646,
      "learning_rate": 7.471173228903072e-06,
      "loss": 1.6573,
      "step": 9147
    },
    {
      "epoch": 1.6479171357802298,
      "grad_norm": 1.5871111154556274,
      "learning_rate": 7.463728079051751e-06,
      "loss": 1.2576,
      "step": 9148
    },
    {
      "epoch": 1.648097275388426,
      "grad_norm": 1.6523500680923462,
      "learning_rate": 7.4562863414394445e-06,
      "loss": 1.3293,
      "step": 9149
    },
    {
      "epoch": 1.6482774149966224,
      "grad_norm": 1.5012227296829224,
      "learning_rate": 7.448848016663124e-06,
      "loss": 1.4425,
      "step": 9150
    },
    {
      "epoch": 1.6484575546048186,
      "grad_norm": 1.4702601432800293,
      "learning_rate": 7.441413105319478e-06,
      "loss": 1.0992,
      "step": 9151
    },
    {
      "epoch": 1.648637694213015,
      "grad_norm": 1.3958467245101929,
      "learning_rate": 7.433981608004958e-06,
      "loss": 2.1669,
      "step": 9152
    },
    {
      "epoch": 1.6488178338212114,
      "grad_norm": 1.3880884647369385,
      "learning_rate": 7.426553525315699e-06,
      "loss": 1.8119,
      "step": 9153
    },
    {
      "epoch": 1.6489979734294078,
      "grad_norm": 1.3903489112854004,
      "learning_rate": 7.419128857847574e-06,
      "loss": 1.6083,
      "step": 9154
    },
    {
      "epoch": 1.6491781130376042,
      "grad_norm": 1.4274431467056274,
      "learning_rate": 7.4117076061961885e-06,
      "loss": 1.6551,
      "step": 9155
    },
    {
      "epoch": 1.6493582526458006,
      "grad_norm": 1.4004285335540771,
      "learning_rate": 7.404289770956879e-06,
      "loss": 1.5025,
      "step": 9156
    },
    {
      "epoch": 1.649538392253997,
      "grad_norm": 1.47087824344635,
      "learning_rate": 7.3968753527247026e-06,
      "loss": 1.7137,
      "step": 9157
    },
    {
      "epoch": 1.6497185318621932,
      "grad_norm": 1.6097174882888794,
      "learning_rate": 7.389464352094428e-06,
      "loss": 1.7584,
      "step": 9158
    },
    {
      "epoch": 1.6498986714703896,
      "grad_norm": 1.7657599449157715,
      "learning_rate": 7.382056769660583e-06,
      "loss": 2.2776,
      "step": 9159
    },
    {
      "epoch": 1.6500788110785858,
      "grad_norm": 1.9096273183822632,
      "learning_rate": 7.374652606017363e-06,
      "loss": 2.0394,
      "step": 9160
    },
    {
      "epoch": 1.6502589506867822,
      "grad_norm": 1.5374268293380737,
      "learning_rate": 7.3672518617587615e-06,
      "loss": 1.5644,
      "step": 9161
    },
    {
      "epoch": 1.6504390902949786,
      "grad_norm": 1.4642930030822754,
      "learning_rate": 7.35985453747845e-06,
      "loss": 1.5732,
      "step": 9162
    },
    {
      "epoch": 1.650619229903175,
      "grad_norm": 1.4503756761550903,
      "learning_rate": 7.352460633769836e-06,
      "loss": 1.6083,
      "step": 9163
    },
    {
      "epoch": 1.6507993695113714,
      "grad_norm": 1.3482158184051514,
      "learning_rate": 7.345070151226058e-06,
      "loss": 1.3422,
      "step": 9164
    },
    {
      "epoch": 1.6509795091195678,
      "grad_norm": 1.4234977960586548,
      "learning_rate": 7.337683090439962e-06,
      "loss": 1.3128,
      "step": 9165
    },
    {
      "epoch": 1.651159648727764,
      "grad_norm": 1.351873755455017,
      "learning_rate": 7.330299452004163e-06,
      "loss": 1.4488,
      "step": 9166
    },
    {
      "epoch": 1.6513397883359604,
      "grad_norm": 1.4494565725326538,
      "learning_rate": 7.322919236510955e-06,
      "loss": 1.3232,
      "step": 9167
    },
    {
      "epoch": 1.6515199279441566,
      "grad_norm": 1.4370148181915283,
      "learning_rate": 7.315542444552381e-06,
      "loss": 1.4195,
      "step": 9168
    },
    {
      "epoch": 1.651700067552353,
      "grad_norm": 1.5724514722824097,
      "learning_rate": 7.308169076720201e-06,
      "loss": 1.5587,
      "step": 9169
    },
    {
      "epoch": 1.6518802071605494,
      "grad_norm": 1.5104765892028809,
      "learning_rate": 7.300799133605907e-06,
      "loss": 1.4954,
      "step": 9170
    },
    {
      "epoch": 1.6520603467687458,
      "grad_norm": 1.5433485507965088,
      "learning_rate": 7.293432615800705e-06,
      "loss": 1.6514,
      "step": 9171
    },
    {
      "epoch": 1.6522404863769422,
      "grad_norm": 1.3862031698226929,
      "learning_rate": 7.286069523895533e-06,
      "loss": 1.3225,
      "step": 9172
    },
    {
      "epoch": 1.6524206259851386,
      "grad_norm": 1.4417142868041992,
      "learning_rate": 7.278709858481064e-06,
      "loss": 1.4194,
      "step": 9173
    },
    {
      "epoch": 1.6526007655933348,
      "grad_norm": 1.4928771257400513,
      "learning_rate": 7.2713536201476695e-06,
      "loss": 1.3339,
      "step": 9174
    },
    {
      "epoch": 1.6527809052015312,
      "grad_norm": 1.5620970726013184,
      "learning_rate": 7.264000809485483e-06,
      "loss": 1.7744,
      "step": 9175
    },
    {
      "epoch": 1.6529610448097274,
      "grad_norm": 1.2913447618484497,
      "learning_rate": 7.256651427084338e-06,
      "loss": 1.1589,
      "step": 9176
    },
    {
      "epoch": 1.6531411844179238,
      "grad_norm": 1.6886887550354004,
      "learning_rate": 7.249305473533796e-06,
      "loss": 1.4549,
      "step": 9177
    },
    {
      "epoch": 1.6533213240261202,
      "grad_norm": 1.4848970174789429,
      "learning_rate": 7.241962949423148e-06,
      "loss": 1.5128,
      "step": 9178
    },
    {
      "epoch": 1.6535014636343166,
      "grad_norm": 1.5131266117095947,
      "learning_rate": 7.234623855341405e-06,
      "loss": 1.5067,
      "step": 9179
    },
    {
      "epoch": 1.653681603242513,
      "grad_norm": 1.5141117572784424,
      "learning_rate": 7.227288191877302e-06,
      "loss": 1.4605,
      "step": 9180
    },
    {
      "epoch": 1.6538617428507094,
      "grad_norm": 1.4768322706222534,
      "learning_rate": 7.219955959619307e-06,
      "loss": 1.3815,
      "step": 9181
    },
    {
      "epoch": 1.6540418824589056,
      "grad_norm": 1.3588227033615112,
      "learning_rate": 7.21262715915561e-06,
      "loss": 1.3051,
      "step": 9182
    },
    {
      "epoch": 1.654222022067102,
      "grad_norm": 1.657265067100525,
      "learning_rate": 7.205301791074109e-06,
      "loss": 1.6501,
      "step": 9183
    },
    {
      "epoch": 1.6544021616752984,
      "grad_norm": 1.4423168897628784,
      "learning_rate": 7.197979855962466e-06,
      "loss": 1.2638,
      "step": 9184
    },
    {
      "epoch": 1.6545823012834946,
      "grad_norm": 1.5935659408569336,
      "learning_rate": 7.190661354408029e-06,
      "loss": 1.4209,
      "step": 9185
    },
    {
      "epoch": 1.654762440891691,
      "grad_norm": 1.3804147243499756,
      "learning_rate": 7.1833462869978885e-06,
      "loss": 1.015,
      "step": 9186
    },
    {
      "epoch": 1.6549425804998874,
      "grad_norm": 1.4230194091796875,
      "learning_rate": 7.176034654318853e-06,
      "loss": 1.497,
      "step": 9187
    },
    {
      "epoch": 1.6551227201080838,
      "grad_norm": 1.480778694152832,
      "learning_rate": 7.168726456957448e-06,
      "loss": 1.286,
      "step": 9188
    },
    {
      "epoch": 1.6553028597162802,
      "grad_norm": 1.4998599290847778,
      "learning_rate": 7.161421695499954e-06,
      "loss": 1.2958,
      "step": 9189
    },
    {
      "epoch": 1.6554829993244766,
      "grad_norm": 1.3281853199005127,
      "learning_rate": 7.1541203705323465e-06,
      "loss": 1.101,
      "step": 9190
    },
    {
      "epoch": 1.6556631389326728,
      "grad_norm": 1.5923738479614258,
      "learning_rate": 7.1468224826403475e-06,
      "loss": 1.4158,
      "step": 9191
    },
    {
      "epoch": 1.6558432785408692,
      "grad_norm": 1.5447237491607666,
      "learning_rate": 7.139528032409359e-06,
      "loss": 1.5122,
      "step": 9192
    },
    {
      "epoch": 1.6560234181490654,
      "grad_norm": 1.6370972394943237,
      "learning_rate": 7.132237020424548e-06,
      "loss": 1.7458,
      "step": 9193
    },
    {
      "epoch": 1.6562035577572618,
      "grad_norm": 1.373033046722412,
      "learning_rate": 7.124949447270812e-06,
      "loss": 1.2028,
      "step": 9194
    },
    {
      "epoch": 1.6563836973654582,
      "grad_norm": 1.6207705736160278,
      "learning_rate": 7.117665313532751e-06,
      "loss": 1.442,
      "step": 9195
    },
    {
      "epoch": 1.6565638369736546,
      "grad_norm": 1.4296326637268066,
      "learning_rate": 7.110384619794685e-06,
      "loss": 1.0397,
      "step": 9196
    },
    {
      "epoch": 1.656743976581851,
      "grad_norm": 1.5252666473388672,
      "learning_rate": 7.103107366640671e-06,
      "loss": 1.1754,
      "step": 9197
    },
    {
      "epoch": 1.6569241161900474,
      "grad_norm": 1.6323976516723633,
      "learning_rate": 7.095833554654497e-06,
      "loss": 1.2867,
      "step": 9198
    },
    {
      "epoch": 1.6571042557982436,
      "grad_norm": 1.6161121129989624,
      "learning_rate": 7.088563184419655e-06,
      "loss": 1.2683,
      "step": 9199
    },
    {
      "epoch": 1.65728439540644,
      "grad_norm": 1.4537461996078491,
      "learning_rate": 7.081296256519376e-06,
      "loss": 1.1056,
      "step": 9200
    },
    {
      "epoch": 1.6574645350146362,
      "grad_norm": 1.4482789039611816,
      "learning_rate": 7.074032771536604e-06,
      "loss": 1.0953,
      "step": 9201
    },
    {
      "epoch": 1.6576446746228326,
      "grad_norm": 1.3574049472808838,
      "learning_rate": 7.066772730054017e-06,
      "loss": 1.6399,
      "step": 9202
    },
    {
      "epoch": 1.657824814231029,
      "grad_norm": 1.353438377380371,
      "learning_rate": 7.059516132654009e-06,
      "loss": 1.9004,
      "step": 9203
    },
    {
      "epoch": 1.6580049538392254,
      "grad_norm": 1.368770956993103,
      "learning_rate": 7.0522629799187e-06,
      "loss": 1.9645,
      "step": 9204
    },
    {
      "epoch": 1.6581850934474218,
      "grad_norm": 1.4041920900344849,
      "learning_rate": 7.045013272429935e-06,
      "loss": 1.7341,
      "step": 9205
    },
    {
      "epoch": 1.6583652330556182,
      "grad_norm": 1.4023265838623047,
      "learning_rate": 7.0377670107692865e-06,
      "loss": 1.6694,
      "step": 9206
    },
    {
      "epoch": 1.6585453726638144,
      "grad_norm": 1.4291603565216064,
      "learning_rate": 7.030524195518024e-06,
      "loss": 1.6597,
      "step": 9207
    },
    {
      "epoch": 1.6587255122720108,
      "grad_norm": 1.6233329772949219,
      "learning_rate": 7.0232848272571985e-06,
      "loss": 1.9106,
      "step": 9208
    },
    {
      "epoch": 1.658905651880207,
      "grad_norm": 1.3920167684555054,
      "learning_rate": 7.0160489065675294e-06,
      "loss": 1.4505,
      "step": 9209
    },
    {
      "epoch": 1.6590857914884034,
      "grad_norm": 1.6637388467788696,
      "learning_rate": 7.008816434029481e-06,
      "loss": 1.8336,
      "step": 9210
    },
    {
      "epoch": 1.6592659310965998,
      "grad_norm": 1.675485610961914,
      "learning_rate": 7.001587410223237e-06,
      "loss": 1.6661,
      "step": 9211
    },
    {
      "epoch": 1.6594460707047962,
      "grad_norm": 1.5309358835220337,
      "learning_rate": 6.994361835728708e-06,
      "loss": 1.4504,
      "step": 9212
    },
    {
      "epoch": 1.6596262103129926,
      "grad_norm": 1.3489881753921509,
      "learning_rate": 6.987139711125524e-06,
      "loss": 1.3182,
      "step": 9213
    },
    {
      "epoch": 1.659806349921189,
      "grad_norm": 1.4475346803665161,
      "learning_rate": 6.979921036993042e-06,
      "loss": 1.4292,
      "step": 9214
    },
    {
      "epoch": 1.6599864895293854,
      "grad_norm": 1.2777899503707886,
      "learning_rate": 6.972705813910341e-06,
      "loss": 1.1309,
      "step": 9215
    },
    {
      "epoch": 1.6601666291375816,
      "grad_norm": 1.556312918663025,
      "learning_rate": 6.965494042456205e-06,
      "loss": 1.654,
      "step": 9216
    },
    {
      "epoch": 1.660346768745778,
      "grad_norm": 1.3114372491836548,
      "learning_rate": 6.958285723209196e-06,
      "loss": 1.396,
      "step": 9217
    },
    {
      "epoch": 1.6605269083539742,
      "grad_norm": 1.3989691734313965,
      "learning_rate": 6.951080856747533e-06,
      "loss": 1.1988,
      "step": 9218
    },
    {
      "epoch": 1.6607070479621706,
      "grad_norm": 1.2972607612609863,
      "learning_rate": 6.943879443649198e-06,
      "loss": 1.1605,
      "step": 9219
    },
    {
      "epoch": 1.660887187570367,
      "grad_norm": 1.4657961130142212,
      "learning_rate": 6.936681484491875e-06,
      "loss": 1.2247,
      "step": 9220
    },
    {
      "epoch": 1.6610673271785634,
      "grad_norm": 1.2845828533172607,
      "learning_rate": 6.929486979852995e-06,
      "loss": 1.0902,
      "step": 9221
    },
    {
      "epoch": 1.6612474667867598,
      "grad_norm": 1.448779821395874,
      "learning_rate": 6.922295930309691e-06,
      "loss": 1.4701,
      "step": 9222
    },
    {
      "epoch": 1.6614276063949562,
      "grad_norm": 1.4207104444503784,
      "learning_rate": 6.915108336438836e-06,
      "loss": 1.3069,
      "step": 9223
    },
    {
      "epoch": 1.6616077460031524,
      "grad_norm": 1.3946537971496582,
      "learning_rate": 6.907924198816995e-06,
      "loss": 1.2854,
      "step": 9224
    },
    {
      "epoch": 1.6617878856113488,
      "grad_norm": 1.581763744354248,
      "learning_rate": 6.900743518020475e-06,
      "loss": 1.5648,
      "step": 9225
    },
    {
      "epoch": 1.661968025219545,
      "grad_norm": 1.4212431907653809,
      "learning_rate": 6.893566294625326e-06,
      "loss": 1.243,
      "step": 9226
    },
    {
      "epoch": 1.6621481648277414,
      "grad_norm": 1.6154202222824097,
      "learning_rate": 6.886392529207292e-06,
      "loss": 1.438,
      "step": 9227
    },
    {
      "epoch": 1.6623283044359378,
      "grad_norm": 1.4189852476119995,
      "learning_rate": 6.879222222341852e-06,
      "loss": 1.4055,
      "step": 9228
    },
    {
      "epoch": 1.6625084440441342,
      "grad_norm": 1.5226151943206787,
      "learning_rate": 6.872055374604197e-06,
      "loss": 1.6372,
      "step": 9229
    },
    {
      "epoch": 1.6626885836523306,
      "grad_norm": 1.573737621307373,
      "learning_rate": 6.86489198656925e-06,
      "loss": 1.4532,
      "step": 9230
    },
    {
      "epoch": 1.662868723260527,
      "grad_norm": 1.535461664199829,
      "learning_rate": 6.857732058811661e-06,
      "loss": 1.3273,
      "step": 9231
    },
    {
      "epoch": 1.6630488628687232,
      "grad_norm": 1.455527663230896,
      "learning_rate": 6.850575591905794e-06,
      "loss": 1.4525,
      "step": 9232
    },
    {
      "epoch": 1.6632290024769196,
      "grad_norm": 1.6086820363998413,
      "learning_rate": 6.843422586425741e-06,
      "loss": 1.5924,
      "step": 9233
    },
    {
      "epoch": 1.6634091420851158,
      "grad_norm": 1.5959326028823853,
      "learning_rate": 6.836273042945302e-06,
      "loss": 1.3925,
      "step": 9234
    },
    {
      "epoch": 1.6635892816933122,
      "grad_norm": 1.4982845783233643,
      "learning_rate": 6.829126962038013e-06,
      "loss": 1.3325,
      "step": 9235
    },
    {
      "epoch": 1.6637694213015086,
      "grad_norm": 1.42143976688385,
      "learning_rate": 6.821984344277138e-06,
      "loss": 1.3187,
      "step": 9236
    },
    {
      "epoch": 1.663949560909705,
      "grad_norm": 1.4057400226593018,
      "learning_rate": 6.814845190235641e-06,
      "loss": 1.307,
      "step": 9237
    },
    {
      "epoch": 1.6641297005179014,
      "grad_norm": 1.5301775932312012,
      "learning_rate": 6.807709500486237e-06,
      "loss": 1.3539,
      "step": 9238
    },
    {
      "epoch": 1.6643098401260978,
      "grad_norm": 1.4692734479904175,
      "learning_rate": 6.800577275601317e-06,
      "loss": 1.3664,
      "step": 9239
    },
    {
      "epoch": 1.664489979734294,
      "grad_norm": 1.64423406124115,
      "learning_rate": 6.793448516153067e-06,
      "loss": 1.5151,
      "step": 9240
    },
    {
      "epoch": 1.6646701193424904,
      "grad_norm": 1.59138822555542,
      "learning_rate": 6.786323222713326e-06,
      "loss": 1.4887,
      "step": 9241
    },
    {
      "epoch": 1.6648502589506868,
      "grad_norm": 1.5972872972488403,
      "learning_rate": 6.779201395853685e-06,
      "loss": 1.3675,
      "step": 9242
    },
    {
      "epoch": 1.665030398558883,
      "grad_norm": 1.5561999082565308,
      "learning_rate": 6.77208303614546e-06,
      "loss": 1.318,
      "step": 9243
    },
    {
      "epoch": 1.6652105381670794,
      "grad_norm": 1.621854543685913,
      "learning_rate": 6.764968144159678e-06,
      "loss": 1.3312,
      "step": 9244
    },
    {
      "epoch": 1.6653906777752758,
      "grad_norm": 1.637001633644104,
      "learning_rate": 6.757856720467093e-06,
      "loss": 1.425,
      "step": 9245
    },
    {
      "epoch": 1.6655708173834722,
      "grad_norm": 1.5609912872314453,
      "learning_rate": 6.750748765638182e-06,
      "loss": 1.3637,
      "step": 9246
    },
    {
      "epoch": 1.6657509569916686,
      "grad_norm": 1.51085364818573,
      "learning_rate": 6.743644280243133e-06,
      "loss": 1.3939,
      "step": 9247
    },
    {
      "epoch": 1.665931096599865,
      "grad_norm": 1.4847136735916138,
      "learning_rate": 6.736543264851864e-06,
      "loss": 1.394,
      "step": 9248
    },
    {
      "epoch": 1.6661112362080612,
      "grad_norm": 1.6096235513687134,
      "learning_rate": 6.729445720034028e-06,
      "loss": 1.3906,
      "step": 9249
    },
    {
      "epoch": 1.6662913758162576,
      "grad_norm": 1.6262269020080566,
      "learning_rate": 6.722351646358987e-06,
      "loss": 1.2838,
      "step": 9250
    },
    {
      "epoch": 1.6664715154244538,
      "grad_norm": 1.3291374444961548,
      "learning_rate": 6.71526104439581e-06,
      "loss": 0.9427,
      "step": 9251
    },
    {
      "epoch": 1.6666516550326502,
      "grad_norm": 1.3149293661117554,
      "learning_rate": 6.7081739147133115e-06,
      "loss": 1.531,
      "step": 9252
    },
    {
      "epoch": 1.6668317946408466,
      "grad_norm": 1.3284752368927002,
      "learning_rate": 6.701090257880005e-06,
      "loss": 1.8974,
      "step": 9253
    },
    {
      "epoch": 1.667011934249043,
      "grad_norm": 1.3059722185134888,
      "learning_rate": 6.694010074464158e-06,
      "loss": 1.7937,
      "step": 9254
    },
    {
      "epoch": 1.6671920738572394,
      "grad_norm": 1.4482176303863525,
      "learning_rate": 6.686933365033726e-06,
      "loss": 1.9599,
      "step": 9255
    },
    {
      "epoch": 1.6673722134654358,
      "grad_norm": 1.4542229175567627,
      "learning_rate": 6.6798601301564175e-06,
      "loss": 1.9155,
      "step": 9256
    },
    {
      "epoch": 1.667552353073632,
      "grad_norm": 1.3910964727401733,
      "learning_rate": 6.672790370399618e-06,
      "loss": 1.6094,
      "step": 9257
    },
    {
      "epoch": 1.6677324926818284,
      "grad_norm": 1.4538975954055786,
      "learning_rate": 6.6657240863304605e-06,
      "loss": 1.6888,
      "step": 9258
    },
    {
      "epoch": 1.6679126322900246,
      "grad_norm": 1.408825397491455,
      "learning_rate": 6.658661278515815e-06,
      "loss": 1.5511,
      "step": 9259
    },
    {
      "epoch": 1.668092771898221,
      "grad_norm": 1.6331695318222046,
      "learning_rate": 6.651601947522257e-06,
      "loss": 1.8638,
      "step": 9260
    },
    {
      "epoch": 1.6682729115064174,
      "grad_norm": 1.7563532590866089,
      "learning_rate": 6.644546093916076e-06,
      "loss": 1.8483,
      "step": 9261
    },
    {
      "epoch": 1.6684530511146138,
      "grad_norm": 1.6008085012435913,
      "learning_rate": 6.637493718263277e-06,
      "loss": 1.9279,
      "step": 9262
    },
    {
      "epoch": 1.6686331907228102,
      "grad_norm": 1.3369431495666504,
      "learning_rate": 6.630444821129622e-06,
      "loss": 1.4001,
      "step": 9263
    },
    {
      "epoch": 1.6688133303310067,
      "grad_norm": 1.3782161474227905,
      "learning_rate": 6.623399403080555e-06,
      "loss": 1.4676,
      "step": 9264
    },
    {
      "epoch": 1.6689934699392028,
      "grad_norm": 1.4244213104248047,
      "learning_rate": 6.616357464681267e-06,
      "loss": 1.4251,
      "step": 9265
    },
    {
      "epoch": 1.6691736095473992,
      "grad_norm": 1.4525614976882935,
      "learning_rate": 6.609319006496645e-06,
      "loss": 1.4063,
      "step": 9266
    },
    {
      "epoch": 1.6693537491555954,
      "grad_norm": 1.5354183912277222,
      "learning_rate": 6.602284029091321e-06,
      "loss": 1.7287,
      "step": 9267
    },
    {
      "epoch": 1.6695338887637918,
      "grad_norm": 1.479678750038147,
      "learning_rate": 6.595252533029633e-06,
      "loss": 1.3172,
      "step": 9268
    },
    {
      "epoch": 1.6697140283719882,
      "grad_norm": 1.5796678066253662,
      "learning_rate": 6.588224518875647e-06,
      "loss": 1.4513,
      "step": 9269
    },
    {
      "epoch": 1.6698941679801846,
      "grad_norm": 1.4105284214019775,
      "learning_rate": 6.581199987193143e-06,
      "loss": 1.3071,
      "step": 9270
    },
    {
      "epoch": 1.670074307588381,
      "grad_norm": 1.412529706954956,
      "learning_rate": 6.574178938545627e-06,
      "loss": 1.3735,
      "step": 9271
    },
    {
      "epoch": 1.6702544471965775,
      "grad_norm": 1.3892160654067993,
      "learning_rate": 6.567161373496316e-06,
      "loss": 1.4019,
      "step": 9272
    },
    {
      "epoch": 1.6704345868047739,
      "grad_norm": 1.3655197620391846,
      "learning_rate": 6.5601472926081766e-06,
      "loss": 1.2421,
      "step": 9273
    },
    {
      "epoch": 1.67061472641297,
      "grad_norm": 1.4209932088851929,
      "learning_rate": 6.553136696443856e-06,
      "loss": 1.4158,
      "step": 9274
    },
    {
      "epoch": 1.6707948660211664,
      "grad_norm": 1.394223690032959,
      "learning_rate": 6.546129585565752e-06,
      "loss": 1.1676,
      "step": 9275
    },
    {
      "epoch": 1.6709750056293626,
      "grad_norm": 1.493454933166504,
      "learning_rate": 6.539125960535969e-06,
      "loss": 1.4004,
      "step": 9276
    },
    {
      "epoch": 1.671155145237559,
      "grad_norm": 1.5593960285186768,
      "learning_rate": 6.53212582191633e-06,
      "loss": 1.4255,
      "step": 9277
    },
    {
      "epoch": 1.6713352848457554,
      "grad_norm": 1.4963762760162354,
      "learning_rate": 6.525129170268385e-06,
      "loss": 1.5255,
      "step": 9278
    },
    {
      "epoch": 1.6715154244539518,
      "grad_norm": 1.4560383558273315,
      "learning_rate": 6.518136006153403e-06,
      "loss": 1.4282,
      "step": 9279
    },
    {
      "epoch": 1.6716955640621483,
      "grad_norm": 1.279828429222107,
      "learning_rate": 6.511146330132367e-06,
      "loss": 1.1918,
      "step": 9280
    },
    {
      "epoch": 1.6718757036703447,
      "grad_norm": 1.5695408582687378,
      "learning_rate": 6.504160142765986e-06,
      "loss": 1.4093,
      "step": 9281
    },
    {
      "epoch": 1.6720558432785408,
      "grad_norm": 1.4600454568862915,
      "learning_rate": 6.4971774446147e-06,
      "loss": 1.2941,
      "step": 9282
    },
    {
      "epoch": 1.6722359828867372,
      "grad_norm": 1.4232903718948364,
      "learning_rate": 6.490198236238648e-06,
      "loss": 1.2589,
      "step": 9283
    },
    {
      "epoch": 1.6724161224949334,
      "grad_norm": 1.5326049327850342,
      "learning_rate": 6.4832225181977015e-06,
      "loss": 1.4691,
      "step": 9284
    },
    {
      "epoch": 1.6725962621031298,
      "grad_norm": 1.533703327178955,
      "learning_rate": 6.47625029105145e-06,
      "loss": 1.4436,
      "step": 9285
    },
    {
      "epoch": 1.6727764017113262,
      "grad_norm": 1.6406062841415405,
      "learning_rate": 6.469281555359185e-06,
      "loss": 1.4567,
      "step": 9286
    },
    {
      "epoch": 1.6729565413195226,
      "grad_norm": 1.5250133275985718,
      "learning_rate": 6.462316311679967e-06,
      "loss": 1.3538,
      "step": 9287
    },
    {
      "epoch": 1.673136680927719,
      "grad_norm": 1.650330662727356,
      "learning_rate": 6.455354560572535e-06,
      "loss": 1.3333,
      "step": 9288
    },
    {
      "epoch": 1.6733168205359155,
      "grad_norm": 1.5261865854263306,
      "learning_rate": 6.448396302595344e-06,
      "loss": 1.321,
      "step": 9289
    },
    {
      "epoch": 1.6734969601441116,
      "grad_norm": 1.5919733047485352,
      "learning_rate": 6.4414415383065755e-06,
      "loss": 1.4509,
      "step": 9290
    },
    {
      "epoch": 1.673677099752308,
      "grad_norm": 1.4564836025238037,
      "learning_rate": 6.43449026826416e-06,
      "loss": 1.328,
      "step": 9291
    },
    {
      "epoch": 1.6738572393605042,
      "grad_norm": 1.517279863357544,
      "learning_rate": 6.427542493025712e-06,
      "loss": 1.4691,
      "step": 9292
    },
    {
      "epoch": 1.6740373789687006,
      "grad_norm": 1.5118247270584106,
      "learning_rate": 6.4205982131485875e-06,
      "loss": 1.1655,
      "step": 9293
    },
    {
      "epoch": 1.674217518576897,
      "grad_norm": 1.6690576076507568,
      "learning_rate": 6.413657429189845e-06,
      "loss": 1.5493,
      "step": 9294
    },
    {
      "epoch": 1.6743976581850935,
      "grad_norm": 1.4392532110214233,
      "learning_rate": 6.406720141706263e-06,
      "loss": 1.1293,
      "step": 9295
    },
    {
      "epoch": 1.6745777977932899,
      "grad_norm": 1.528138518333435,
      "learning_rate": 6.399786351254372e-06,
      "loss": 1.2934,
      "step": 9296
    },
    {
      "epoch": 1.6747579374014863,
      "grad_norm": 1.455649971961975,
      "learning_rate": 6.392856058390379e-06,
      "loss": 1.377,
      "step": 9297
    },
    {
      "epoch": 1.6749380770096827,
      "grad_norm": 1.555851697921753,
      "learning_rate": 6.385929263670232e-06,
      "loss": 1.3059,
      "step": 9298
    },
    {
      "epoch": 1.6751182166178789,
      "grad_norm": 1.9327064752578735,
      "learning_rate": 6.3790059676495984e-06,
      "loss": 1.7124,
      "step": 9299
    },
    {
      "epoch": 1.6752983562260753,
      "grad_norm": 1.417898178100586,
      "learning_rate": 6.372086170883857e-06,
      "loss": 1.2291,
      "step": 9300
    },
    {
      "epoch": 1.6754784958342714,
      "grad_norm": 1.3643193244934082,
      "learning_rate": 6.365169873928112e-06,
      "loss": 1.142,
      "step": 9301
    },
    {
      "epoch": 1.6756586354424678,
      "grad_norm": 1.4310061931610107,
      "learning_rate": 6.3582570773371875e-06,
      "loss": 1.9516,
      "step": 9302
    },
    {
      "epoch": 1.6758387750506643,
      "grad_norm": 1.3903883695602417,
      "learning_rate": 6.351347781665623e-06,
      "loss": 1.7865,
      "step": 9303
    },
    {
      "epoch": 1.6760189146588607,
      "grad_norm": 1.3864984512329102,
      "learning_rate": 6.344441987467664e-06,
      "loss": 1.7112,
      "step": 9304
    },
    {
      "epoch": 1.676199054267057,
      "grad_norm": 1.479129672050476,
      "learning_rate": 6.337539695297323e-06,
      "loss": 1.7727,
      "step": 9305
    },
    {
      "epoch": 1.6763791938752535,
      "grad_norm": 1.475406289100647,
      "learning_rate": 6.330640905708279e-06,
      "loss": 1.6852,
      "step": 9306
    },
    {
      "epoch": 1.6765593334834497,
      "grad_norm": 1.3398133516311646,
      "learning_rate": 6.323745619253946e-06,
      "loss": 1.4649,
      "step": 9307
    },
    {
      "epoch": 1.676739473091646,
      "grad_norm": 1.526917815208435,
      "learning_rate": 6.316853836487469e-06,
      "loss": 1.8893,
      "step": 9308
    },
    {
      "epoch": 1.6769196126998422,
      "grad_norm": 1.5974851846694946,
      "learning_rate": 6.3099655579617e-06,
      "loss": 2.005,
      "step": 9309
    },
    {
      "epoch": 1.6770997523080386,
      "grad_norm": 1.6952717304229736,
      "learning_rate": 6.303080784229215e-06,
      "loss": 1.6756,
      "step": 9310
    },
    {
      "epoch": 1.677279891916235,
      "grad_norm": 2.03130841255188,
      "learning_rate": 6.296199515842299e-06,
      "loss": 2.1392,
      "step": 9311
    },
    {
      "epoch": 1.6774600315244315,
      "grad_norm": 1.6839735507965088,
      "learning_rate": 6.289321753352978e-06,
      "loss": 1.844,
      "step": 9312
    },
    {
      "epoch": 1.6776401711326279,
      "grad_norm": 1.39813232421875,
      "learning_rate": 6.282447497312971e-06,
      "loss": 1.5295,
      "step": 9313
    },
    {
      "epoch": 1.6778203107408243,
      "grad_norm": 1.2228069305419922,
      "learning_rate": 6.275576748273721e-06,
      "loss": 1.1335,
      "step": 9314
    },
    {
      "epoch": 1.6780004503490205,
      "grad_norm": 1.3315496444702148,
      "learning_rate": 6.268709506786419e-06,
      "loss": 1.3848,
      "step": 9315
    },
    {
      "epoch": 1.6781805899572169,
      "grad_norm": 1.4042381048202515,
      "learning_rate": 6.2618457734019364e-06,
      "loss": 1.46,
      "step": 9316
    },
    {
      "epoch": 1.678360729565413,
      "grad_norm": 1.5339460372924805,
      "learning_rate": 6.254985548670889e-06,
      "loss": 1.4393,
      "step": 9317
    },
    {
      "epoch": 1.6785408691736095,
      "grad_norm": 1.3979800939559937,
      "learning_rate": 6.248128833143574e-06,
      "loss": 1.3633,
      "step": 9318
    },
    {
      "epoch": 1.6787210087818059,
      "grad_norm": 1.4021457433700562,
      "learning_rate": 6.241275627370069e-06,
      "loss": 1.4095,
      "step": 9319
    },
    {
      "epoch": 1.6789011483900023,
      "grad_norm": 1.347480297088623,
      "learning_rate": 6.2344259319001175e-06,
      "loss": 1.2728,
      "step": 9320
    },
    {
      "epoch": 1.6790812879981987,
      "grad_norm": 1.5199741125106812,
      "learning_rate": 6.227579747283213e-06,
      "loss": 1.3622,
      "step": 9321
    },
    {
      "epoch": 1.679261427606395,
      "grad_norm": 1.4238966703414917,
      "learning_rate": 6.2207370740685275e-06,
      "loss": 1.3901,
      "step": 9322
    },
    {
      "epoch": 1.6794415672145913,
      "grad_norm": 1.5909440517425537,
      "learning_rate": 6.213897912804984e-06,
      "loss": 1.6959,
      "step": 9323
    },
    {
      "epoch": 1.6796217068227877,
      "grad_norm": 1.3264161348342896,
      "learning_rate": 6.207062264041225e-06,
      "loss": 1.2573,
      "step": 9324
    },
    {
      "epoch": 1.6798018464309838,
      "grad_norm": 1.4370753765106201,
      "learning_rate": 6.200230128325607e-06,
      "loss": 1.2657,
      "step": 9325
    },
    {
      "epoch": 1.6799819860391803,
      "grad_norm": 1.4973523616790771,
      "learning_rate": 6.193401506206192e-06,
      "loss": 1.3223,
      "step": 9326
    },
    {
      "epoch": 1.6801621256473767,
      "grad_norm": 1.409073829650879,
      "learning_rate": 6.18657639823077e-06,
      "loss": 1.2728,
      "step": 9327
    },
    {
      "epoch": 1.680342265255573,
      "grad_norm": 1.5397162437438965,
      "learning_rate": 6.179754804946836e-06,
      "loss": 1.4265,
      "step": 9328
    },
    {
      "epoch": 1.6805224048637695,
      "grad_norm": 1.6442248821258545,
      "learning_rate": 6.1729367269016406e-06,
      "loss": 1.3897,
      "step": 9329
    },
    {
      "epoch": 1.6807025444719659,
      "grad_norm": 1.4820566177368164,
      "learning_rate": 6.166122164642113e-06,
      "loss": 1.3509,
      "step": 9330
    },
    {
      "epoch": 1.6808826840801623,
      "grad_norm": 1.413561224937439,
      "learning_rate": 6.1593111187149165e-06,
      "loss": 1.2368,
      "step": 9331
    },
    {
      "epoch": 1.6810628236883585,
      "grad_norm": 1.5177332162857056,
      "learning_rate": 6.152503589666425e-06,
      "loss": 1.3211,
      "step": 9332
    },
    {
      "epoch": 1.6812429632965549,
      "grad_norm": 1.386770486831665,
      "learning_rate": 6.145699578042741e-06,
      "loss": 1.2337,
      "step": 9333
    },
    {
      "epoch": 1.681423102904751,
      "grad_norm": 1.364485502243042,
      "learning_rate": 6.138899084389677e-06,
      "loss": 1.2659,
      "step": 9334
    },
    {
      "epoch": 1.6816032425129475,
      "grad_norm": 1.471837043762207,
      "learning_rate": 6.132102109252763e-06,
      "loss": 1.1493,
      "step": 9335
    },
    {
      "epoch": 1.6817833821211439,
      "grad_norm": 1.4224473237991333,
      "learning_rate": 6.125308653177247e-06,
      "loss": 1.3251,
      "step": 9336
    },
    {
      "epoch": 1.6819635217293403,
      "grad_norm": 1.3569118976593018,
      "learning_rate": 6.118518716708094e-06,
      "loss": 1.2716,
      "step": 9337
    },
    {
      "epoch": 1.6821436613375367,
      "grad_norm": 1.5761306285858154,
      "learning_rate": 6.1117323003900054e-06,
      "loss": 1.4294,
      "step": 9338
    },
    {
      "epoch": 1.682323800945733,
      "grad_norm": 1.5127991437911987,
      "learning_rate": 6.1049494047673736e-06,
      "loss": 1.3854,
      "step": 9339
    },
    {
      "epoch": 1.6825039405539293,
      "grad_norm": 1.5416299104690552,
      "learning_rate": 6.09817003038432e-06,
      "loss": 1.4102,
      "step": 9340
    },
    {
      "epoch": 1.6826840801621257,
      "grad_norm": 1.5679646730422974,
      "learning_rate": 6.091394177784681e-06,
      "loss": 1.4769,
      "step": 9341
    },
    {
      "epoch": 1.6828642197703219,
      "grad_norm": 1.5497963428497314,
      "learning_rate": 6.084621847512017e-06,
      "loss": 1.2456,
      "step": 9342
    },
    {
      "epoch": 1.6830443593785183,
      "grad_norm": 1.6438498497009277,
      "learning_rate": 6.077853040109593e-06,
      "loss": 1.4494,
      "step": 9343
    },
    {
      "epoch": 1.6832244989867147,
      "grad_norm": 1.6951961517333984,
      "learning_rate": 6.071087756120414e-06,
      "loss": 1.4121,
      "step": 9344
    },
    {
      "epoch": 1.683404638594911,
      "grad_norm": 1.3651118278503418,
      "learning_rate": 6.064325996087172e-06,
      "loss": 1.1641,
      "step": 9345
    },
    {
      "epoch": 1.6835847782031075,
      "grad_norm": 1.4344110488891602,
      "learning_rate": 6.057567760552285e-06,
      "loss": 1.2485,
      "step": 9346
    },
    {
      "epoch": 1.6837649178113039,
      "grad_norm": 1.5804636478424072,
      "learning_rate": 6.050813050057924e-06,
      "loss": 1.3798,
      "step": 9347
    },
    {
      "epoch": 1.6839450574195,
      "grad_norm": 1.6897717714309692,
      "learning_rate": 6.044061865145934e-06,
      "loss": 1.5196,
      "step": 9348
    },
    {
      "epoch": 1.6841251970276965,
      "grad_norm": 1.5246527194976807,
      "learning_rate": 6.037314206357886e-06,
      "loss": 1.2769,
      "step": 9349
    },
    {
      "epoch": 1.6843053366358927,
      "grad_norm": 1.5533305406570435,
      "learning_rate": 6.030570074235081e-06,
      "loss": 1.3791,
      "step": 9350
    },
    {
      "epoch": 1.684485476244089,
      "grad_norm": 1.7750823497772217,
      "learning_rate": 6.023829469318521e-06,
      "loss": 1.4337,
      "step": 9351
    },
    {
      "epoch": 1.6846656158522855,
      "grad_norm": 1.0884023904800415,
      "learning_rate": 6.017092392148943e-06,
      "loss": 1.3946,
      "step": 9352
    },
    {
      "epoch": 1.6848457554604819,
      "grad_norm": 1.3627616167068481,
      "learning_rate": 6.010358843266811e-06,
      "loss": 1.705,
      "step": 9353
    },
    {
      "epoch": 1.6850258950686783,
      "grad_norm": 1.3868818283081055,
      "learning_rate": 6.003628823212248e-06,
      "loss": 1.5316,
      "step": 9354
    },
    {
      "epoch": 1.6852060346768747,
      "grad_norm": 1.3360754251480103,
      "learning_rate": 5.996902332525145e-06,
      "loss": 1.6387,
      "step": 9355
    },
    {
      "epoch": 1.685386174285071,
      "grad_norm": 1.4895849227905273,
      "learning_rate": 5.990179371745108e-06,
      "loss": 1.9594,
      "step": 9356
    },
    {
      "epoch": 1.6855663138932673,
      "grad_norm": 1.4065252542495728,
      "learning_rate": 5.983459941411451e-06,
      "loss": 1.8091,
      "step": 9357
    },
    {
      "epoch": 1.6857464535014637,
      "grad_norm": 1.3718338012695312,
      "learning_rate": 5.976744042063198e-06,
      "loss": 1.5074,
      "step": 9358
    },
    {
      "epoch": 1.6859265931096599,
      "grad_norm": 1.6097397804260254,
      "learning_rate": 5.97003167423909e-06,
      "loss": 1.6376,
      "step": 9359
    },
    {
      "epoch": 1.6861067327178563,
      "grad_norm": 1.5765243768692017,
      "learning_rate": 5.963322838477592e-06,
      "loss": 1.5906,
      "step": 9360
    },
    {
      "epoch": 1.6862868723260527,
      "grad_norm": 1.573461890220642,
      "learning_rate": 5.956617535316888e-06,
      "loss": 1.8628,
      "step": 9361
    },
    {
      "epoch": 1.686467011934249,
      "grad_norm": 1.9458311796188354,
      "learning_rate": 5.9499157652948775e-06,
      "loss": 2.0262,
      "step": 9362
    },
    {
      "epoch": 1.6866471515424455,
      "grad_norm": 1.615006923675537,
      "learning_rate": 5.943217528949168e-06,
      "loss": 1.7074,
      "step": 9363
    },
    {
      "epoch": 1.686827291150642,
      "grad_norm": 1.247019648551941,
      "learning_rate": 5.936522826817092e-06,
      "loss": 1.0771,
      "step": 9364
    },
    {
      "epoch": 1.687007430758838,
      "grad_norm": 1.457051396369934,
      "learning_rate": 5.929831659435686e-06,
      "loss": 1.3708,
      "step": 9365
    },
    {
      "epoch": 1.6871875703670345,
      "grad_norm": 1.389910340309143,
      "learning_rate": 5.923144027341726e-06,
      "loss": 1.436,
      "step": 9366
    },
    {
      "epoch": 1.6873677099752307,
      "grad_norm": 1.5140084028244019,
      "learning_rate": 5.916459931071677e-06,
      "loss": 1.6493,
      "step": 9367
    },
    {
      "epoch": 1.687547849583427,
      "grad_norm": 1.4190173149108887,
      "learning_rate": 5.909779371161739e-06,
      "loss": 1.3909,
      "step": 9368
    },
    {
      "epoch": 1.6877279891916235,
      "grad_norm": 1.3805214166641235,
      "learning_rate": 5.903102348147821e-06,
      "loss": 1.4369,
      "step": 9369
    },
    {
      "epoch": 1.6879081287998199,
      "grad_norm": 1.4165334701538086,
      "learning_rate": 5.896428862565556e-06,
      "loss": 1.4759,
      "step": 9370
    },
    {
      "epoch": 1.6880882684080163,
      "grad_norm": 1.440468192100525,
      "learning_rate": 5.8897589149502886e-06,
      "loss": 1.2986,
      "step": 9371
    },
    {
      "epoch": 1.6882684080162127,
      "grad_norm": 1.5241209268569946,
      "learning_rate": 5.883092505837079e-06,
      "loss": 1.3565,
      "step": 9372
    },
    {
      "epoch": 1.6884485476244089,
      "grad_norm": 1.3354606628417969,
      "learning_rate": 5.876429635760699e-06,
      "loss": 1.3258,
      "step": 9373
    },
    {
      "epoch": 1.6886286872326053,
      "grad_norm": 1.402143120765686,
      "learning_rate": 5.86977030525564e-06,
      "loss": 1.2903,
      "step": 9374
    },
    {
      "epoch": 1.6888088268408015,
      "grad_norm": 1.5762059688568115,
      "learning_rate": 5.86311451485611e-06,
      "loss": 1.4327,
      "step": 9375
    },
    {
      "epoch": 1.6889889664489979,
      "grad_norm": 1.5700163841247559,
      "learning_rate": 5.856462265096041e-06,
      "loss": 1.422,
      "step": 9376
    },
    {
      "epoch": 1.6891691060571943,
      "grad_norm": 1.5477482080459595,
      "learning_rate": 5.849813556509059e-06,
      "loss": 1.429,
      "step": 9377
    },
    {
      "epoch": 1.6893492456653907,
      "grad_norm": 1.3903214931488037,
      "learning_rate": 5.843168389628534e-06,
      "loss": 1.1545,
      "step": 9378
    },
    {
      "epoch": 1.689529385273587,
      "grad_norm": 1.381422758102417,
      "learning_rate": 5.836526764987527e-06,
      "loss": 1.2447,
      "step": 9379
    },
    {
      "epoch": 1.6897095248817835,
      "grad_norm": 1.4002543687820435,
      "learning_rate": 5.829888683118834e-06,
      "loss": 1.3316,
      "step": 9380
    },
    {
      "epoch": 1.6898896644899797,
      "grad_norm": 1.3596376180648804,
      "learning_rate": 5.823254144554963e-06,
      "loss": 1.2525,
      "step": 9381
    },
    {
      "epoch": 1.690069804098176,
      "grad_norm": 1.468553900718689,
      "learning_rate": 5.816623149828121e-06,
      "loss": 1.4356,
      "step": 9382
    },
    {
      "epoch": 1.6902499437063725,
      "grad_norm": 1.5850248336791992,
      "learning_rate": 5.809995699470239e-06,
      "loss": 1.5161,
      "step": 9383
    },
    {
      "epoch": 1.6904300833145687,
      "grad_norm": 1.5036015510559082,
      "learning_rate": 5.80337179401299e-06,
      "loss": 1.4627,
      "step": 9384
    },
    {
      "epoch": 1.690610222922765,
      "grad_norm": 1.4955673217773438,
      "learning_rate": 5.796751433987729e-06,
      "loss": 1.257,
      "step": 9385
    },
    {
      "epoch": 1.6907903625309615,
      "grad_norm": 1.4623759984970093,
      "learning_rate": 5.790134619925547e-06,
      "loss": 1.3062,
      "step": 9386
    },
    {
      "epoch": 1.690970502139158,
      "grad_norm": 1.423148512840271,
      "learning_rate": 5.7835213523572244e-06,
      "loss": 1.3111,
      "step": 9387
    },
    {
      "epoch": 1.6911506417473543,
      "grad_norm": 1.6185672283172607,
      "learning_rate": 5.77691163181327e-06,
      "loss": 1.4842,
      "step": 9388
    },
    {
      "epoch": 1.6913307813555507,
      "grad_norm": 1.4737343788146973,
      "learning_rate": 5.770305458823932e-06,
      "loss": 1.3992,
      "step": 9389
    },
    {
      "epoch": 1.691510920963747,
      "grad_norm": 1.5779510736465454,
      "learning_rate": 5.763702833919149e-06,
      "loss": 1.3094,
      "step": 9390
    },
    {
      "epoch": 1.6916910605719433,
      "grad_norm": 1.5953152179718018,
      "learning_rate": 5.757103757628573e-06,
      "loss": 1.3171,
      "step": 9391
    },
    {
      "epoch": 1.6918712001801395,
      "grad_norm": 1.5416433811187744,
      "learning_rate": 5.750508230481588e-06,
      "loss": 1.14,
      "step": 9392
    },
    {
      "epoch": 1.6920513397883359,
      "grad_norm": 1.4649804830551147,
      "learning_rate": 5.74391625300727e-06,
      "loss": 1.4061,
      "step": 9393
    },
    {
      "epoch": 1.6922314793965323,
      "grad_norm": 1.6010898351669312,
      "learning_rate": 5.737327825734445e-06,
      "loss": 1.3209,
      "step": 9394
    },
    {
      "epoch": 1.6924116190047287,
      "grad_norm": 1.6245843172073364,
      "learning_rate": 5.73074294919162e-06,
      "loss": 1.475,
      "step": 9395
    },
    {
      "epoch": 1.692591758612925,
      "grad_norm": 1.4712828397750854,
      "learning_rate": 5.724161623907032e-06,
      "loss": 1.3641,
      "step": 9396
    },
    {
      "epoch": 1.6927718982211215,
      "grad_norm": 1.6046422719955444,
      "learning_rate": 5.717583850408631e-06,
      "loss": 1.2939,
      "step": 9397
    },
    {
      "epoch": 1.6929520378293177,
      "grad_norm": 1.5541064739227295,
      "learning_rate": 5.71100962922409e-06,
      "loss": 1.4546,
      "step": 9398
    },
    {
      "epoch": 1.693132177437514,
      "grad_norm": 1.495434284210205,
      "learning_rate": 5.704438960880781e-06,
      "loss": 1.3868,
      "step": 9399
    },
    {
      "epoch": 1.6933123170457103,
      "grad_norm": 1.4846793413162231,
      "learning_rate": 5.6978718459058e-06,
      "loss": 1.3019,
      "step": 9400
    },
    {
      "epoch": 1.6934924566539067,
      "grad_norm": 1.448927402496338,
      "learning_rate": 5.691308284825964e-06,
      "loss": 1.0433,
      "step": 9401
    },
    {
      "epoch": 1.693672596262103,
      "grad_norm": 1.3860310316085815,
      "learning_rate": 5.6847482781677845e-06,
      "loss": 1.638,
      "step": 9402
    },
    {
      "epoch": 1.6938527358702995,
      "grad_norm": 1.4743907451629639,
      "learning_rate": 5.678191826457524e-06,
      "loss": 1.8487,
      "step": 9403
    },
    {
      "epoch": 1.694032875478496,
      "grad_norm": 1.6351152658462524,
      "learning_rate": 5.671638930221129e-06,
      "loss": 2.0537,
      "step": 9404
    },
    {
      "epoch": 1.6942130150866923,
      "grad_norm": 1.420829176902771,
      "learning_rate": 5.665089589984263e-06,
      "loss": 1.8793,
      "step": 9405
    },
    {
      "epoch": 1.6943931546948885,
      "grad_norm": 1.4770466089248657,
      "learning_rate": 5.65854380627232e-06,
      "loss": 1.8737,
      "step": 9406
    },
    {
      "epoch": 1.694573294303085,
      "grad_norm": 1.6435444355010986,
      "learning_rate": 5.652001579610389e-06,
      "loss": 1.8561,
      "step": 9407
    },
    {
      "epoch": 1.694753433911281,
      "grad_norm": 1.4760105609893799,
      "learning_rate": 5.645462910523297e-06,
      "loss": 1.558,
      "step": 9408
    },
    {
      "epoch": 1.6949335735194775,
      "grad_norm": 1.5246566534042358,
      "learning_rate": 5.6389277995355595e-06,
      "loss": 1.648,
      "step": 9409
    },
    {
      "epoch": 1.695113713127674,
      "grad_norm": 1.4946370124816895,
      "learning_rate": 5.6323962471714286e-06,
      "loss": 1.5447,
      "step": 9410
    },
    {
      "epoch": 1.6952938527358703,
      "grad_norm": 1.8241276741027832,
      "learning_rate": 5.625868253954852e-06,
      "loss": 1.982,
      "step": 9411
    },
    {
      "epoch": 1.6954739923440667,
      "grad_norm": 1.7427071332931519,
      "learning_rate": 5.619343820409517e-06,
      "loss": 1.5575,
      "step": 9412
    },
    {
      "epoch": 1.695654131952263,
      "grad_norm": 1.602441430091858,
      "learning_rate": 5.612822947058805e-06,
      "loss": 1.6197,
      "step": 9413
    },
    {
      "epoch": 1.6958342715604595,
      "grad_norm": 1.5788562297821045,
      "learning_rate": 5.606305634425818e-06,
      "loss": 1.6529,
      "step": 9414
    },
    {
      "epoch": 1.6960144111686557,
      "grad_norm": 1.4368292093276978,
      "learning_rate": 5.599791883033368e-06,
      "loss": 1.4649,
      "step": 9415
    },
    {
      "epoch": 1.696194550776852,
      "grad_norm": 1.597377896308899,
      "learning_rate": 5.593281693403974e-06,
      "loss": 1.3605,
      "step": 9416
    },
    {
      "epoch": 1.6963746903850483,
      "grad_norm": 1.4312036037445068,
      "learning_rate": 5.586775066059907e-06,
      "loss": 1.4693,
      "step": 9417
    },
    {
      "epoch": 1.6965548299932447,
      "grad_norm": 1.3958052396774292,
      "learning_rate": 5.580272001523118e-06,
      "loss": 1.3418,
      "step": 9418
    },
    {
      "epoch": 1.696734969601441,
      "grad_norm": 1.4030656814575195,
      "learning_rate": 5.573772500315266e-06,
      "loss": 1.2413,
      "step": 9419
    },
    {
      "epoch": 1.6969151092096375,
      "grad_norm": 1.4695589542388916,
      "learning_rate": 5.5672765629577464e-06,
      "loss": 1.5439,
      "step": 9420
    },
    {
      "epoch": 1.697095248817834,
      "grad_norm": 1.5302057266235352,
      "learning_rate": 5.56078418997164e-06,
      "loss": 1.5623,
      "step": 9421
    },
    {
      "epoch": 1.6972753884260303,
      "grad_norm": 1.4400105476379395,
      "learning_rate": 5.5542953818778e-06,
      "loss": 1.2953,
      "step": 9422
    },
    {
      "epoch": 1.6974555280342265,
      "grad_norm": 1.526248574256897,
      "learning_rate": 5.5478101391967294e-06,
      "loss": 1.5861,
      "step": 9423
    },
    {
      "epoch": 1.697635667642423,
      "grad_norm": 1.382607102394104,
      "learning_rate": 5.541328462448681e-06,
      "loss": 1.3512,
      "step": 9424
    },
    {
      "epoch": 1.697815807250619,
      "grad_norm": 1.459639549255371,
      "learning_rate": 5.534850352153598e-06,
      "loss": 1.4039,
      "step": 9425
    },
    {
      "epoch": 1.6979959468588155,
      "grad_norm": 1.4033910036087036,
      "learning_rate": 5.528375808831171e-06,
      "loss": 1.2526,
      "step": 9426
    },
    {
      "epoch": 1.698176086467012,
      "grad_norm": 1.4770731925964355,
      "learning_rate": 5.521904833000774e-06,
      "loss": 1.4194,
      "step": 9427
    },
    {
      "epoch": 1.6983562260752083,
      "grad_norm": 1.5288301706314087,
      "learning_rate": 5.515437425181508e-06,
      "loss": 1.4453,
      "step": 9428
    },
    {
      "epoch": 1.6985363656834047,
      "grad_norm": 1.61703622341156,
      "learning_rate": 5.508973585892185e-06,
      "loss": 1.41,
      "step": 9429
    },
    {
      "epoch": 1.6987165052916011,
      "grad_norm": 1.382460117340088,
      "learning_rate": 5.5025133156513255e-06,
      "loss": 1.3888,
      "step": 9430
    },
    {
      "epoch": 1.6988966448997973,
      "grad_norm": 1.3052409887313843,
      "learning_rate": 5.49605661497718e-06,
      "loss": 1.1898,
      "step": 9431
    },
    {
      "epoch": 1.6990767845079937,
      "grad_norm": 1.4444466829299927,
      "learning_rate": 5.48960348438769e-06,
      "loss": 1.2514,
      "step": 9432
    },
    {
      "epoch": 1.69925692411619,
      "grad_norm": 1.455181360244751,
      "learning_rate": 5.48315392440053e-06,
      "loss": 1.3771,
      "step": 9433
    },
    {
      "epoch": 1.6994370637243863,
      "grad_norm": 1.5011643171310425,
      "learning_rate": 5.476707935533082e-06,
      "loss": 1.3661,
      "step": 9434
    },
    {
      "epoch": 1.6996172033325827,
      "grad_norm": 1.4729206562042236,
      "learning_rate": 5.470265518302425e-06,
      "loss": 1.2943,
      "step": 9435
    },
    {
      "epoch": 1.699797342940779,
      "grad_norm": 1.5996304750442505,
      "learning_rate": 5.463826673225392e-06,
      "loss": 1.524,
      "step": 9436
    },
    {
      "epoch": 1.6999774825489755,
      "grad_norm": 1.5848191976547241,
      "learning_rate": 5.457391400818484e-06,
      "loss": 1.4143,
      "step": 9437
    },
    {
      "epoch": 1.700157622157172,
      "grad_norm": 1.4248039722442627,
      "learning_rate": 5.450959701597946e-06,
      "loss": 1.2437,
      "step": 9438
    },
    {
      "epoch": 1.700337761765368,
      "grad_norm": 1.6933271884918213,
      "learning_rate": 5.44453157607972e-06,
      "loss": 1.3748,
      "step": 9439
    },
    {
      "epoch": 1.7005179013735645,
      "grad_norm": 1.5392844676971436,
      "learning_rate": 5.438107024779471e-06,
      "loss": 1.2576,
      "step": 9440
    },
    {
      "epoch": 1.700698040981761,
      "grad_norm": 1.4799362421035767,
      "learning_rate": 5.43168604821257e-06,
      "loss": 1.3125,
      "step": 9441
    },
    {
      "epoch": 1.700878180589957,
      "grad_norm": 1.6582175493240356,
      "learning_rate": 5.425268646894105e-06,
      "loss": 1.4644,
      "step": 9442
    },
    {
      "epoch": 1.7010583201981535,
      "grad_norm": 1.523781657218933,
      "learning_rate": 5.418854821338876e-06,
      "loss": 1.2672,
      "step": 9443
    },
    {
      "epoch": 1.70123845980635,
      "grad_norm": 1.5922198295593262,
      "learning_rate": 5.412444572061387e-06,
      "loss": 1.5397,
      "step": 9444
    },
    {
      "epoch": 1.7014185994145463,
      "grad_norm": 1.6494568586349487,
      "learning_rate": 5.406037899575894e-06,
      "loss": 1.4975,
      "step": 9445
    },
    {
      "epoch": 1.7015987390227427,
      "grad_norm": 1.7342473268508911,
      "learning_rate": 5.399634804396314e-06,
      "loss": 1.4612,
      "step": 9446
    },
    {
      "epoch": 1.7017788786309391,
      "grad_norm": 1.7269192934036255,
      "learning_rate": 5.393235287036308e-06,
      "loss": 1.4092,
      "step": 9447
    },
    {
      "epoch": 1.7019590182391353,
      "grad_norm": 1.401110053062439,
      "learning_rate": 5.386839348009226e-06,
      "loss": 1.0819,
      "step": 9448
    },
    {
      "epoch": 1.7021391578473317,
      "grad_norm": 1.6218016147613525,
      "learning_rate": 5.38044698782818e-06,
      "loss": 1.463,
      "step": 9449
    },
    {
      "epoch": 1.702319297455528,
      "grad_norm": 1.578056812286377,
      "learning_rate": 5.374058207005944e-06,
      "loss": 1.4303,
      "step": 9450
    },
    {
      "epoch": 1.7024994370637243,
      "grad_norm": 1.610072135925293,
      "learning_rate": 5.367673006055018e-06,
      "loss": 1.5022,
      "step": 9451
    },
    {
      "epoch": 1.7026795766719207,
      "grad_norm": 1.4195905923843384,
      "learning_rate": 5.361291385487621e-06,
      "loss": 1.6489,
      "step": 9452
    },
    {
      "epoch": 1.7028597162801171,
      "grad_norm": 1.4838440418243408,
      "learning_rate": 5.354913345815676e-06,
      "loss": 1.8911,
      "step": 9453
    },
    {
      "epoch": 1.7030398558883135,
      "grad_norm": 1.364926815032959,
      "learning_rate": 5.3485388875508525e-06,
      "loss": 1.7884,
      "step": 9454
    },
    {
      "epoch": 1.70321999549651,
      "grad_norm": 1.4396288394927979,
      "learning_rate": 5.342168011204485e-06,
      "loss": 1.6244,
      "step": 9455
    },
    {
      "epoch": 1.7034001351047061,
      "grad_norm": 1.4937666654586792,
      "learning_rate": 5.335800717287648e-06,
      "loss": 1.7083,
      "step": 9456
    },
    {
      "epoch": 1.7035802747129025,
      "grad_norm": 1.5352312326431274,
      "learning_rate": 5.329437006311122e-06,
      "loss": 1.8709,
      "step": 9457
    },
    {
      "epoch": 1.7037604143210987,
      "grad_norm": 1.511246681213379,
      "learning_rate": 5.323076878785394e-06,
      "loss": 1.943,
      "step": 9458
    },
    {
      "epoch": 1.703940553929295,
      "grad_norm": 1.666136384010315,
      "learning_rate": 5.316720335220682e-06,
      "loss": 1.7991,
      "step": 9459
    },
    {
      "epoch": 1.7041206935374915,
      "grad_norm": 1.6682777404785156,
      "learning_rate": 5.310367376126907e-06,
      "loss": 1.8095,
      "step": 9460
    },
    {
      "epoch": 1.704300833145688,
      "grad_norm": 1.6242525577545166,
      "learning_rate": 5.304018002013688e-06,
      "loss": 1.7541,
      "step": 9461
    },
    {
      "epoch": 1.7044809727538843,
      "grad_norm": 1.6742832660675049,
      "learning_rate": 5.2976722133903755e-06,
      "loss": 1.6623,
      "step": 9462
    },
    {
      "epoch": 1.7046611123620807,
      "grad_norm": 1.5368788242340088,
      "learning_rate": 5.291330010766027e-06,
      "loss": 1.7078,
      "step": 9463
    },
    {
      "epoch": 1.704841251970277,
      "grad_norm": 1.4520397186279297,
      "learning_rate": 5.284991394649408e-06,
      "loss": 1.583,
      "step": 9464
    },
    {
      "epoch": 1.7050213915784733,
      "grad_norm": 1.5874533653259277,
      "learning_rate": 5.278656365548995e-06,
      "loss": 1.6232,
      "step": 9465
    },
    {
      "epoch": 1.7052015311866695,
      "grad_norm": 1.264407753944397,
      "learning_rate": 5.272324923972988e-06,
      "loss": 1.1589,
      "step": 9466
    },
    {
      "epoch": 1.705381670794866,
      "grad_norm": 1.468383550643921,
      "learning_rate": 5.2659970704292795e-06,
      "loss": 1.306,
      "step": 9467
    },
    {
      "epoch": 1.7055618104030623,
      "grad_norm": 1.4647568464279175,
      "learning_rate": 5.259672805425509e-06,
      "loss": 1.5583,
      "step": 9468
    },
    {
      "epoch": 1.7057419500112587,
      "grad_norm": 1.4793637990951538,
      "learning_rate": 5.253352129468991e-06,
      "loss": 1.343,
      "step": 9469
    },
    {
      "epoch": 1.7059220896194551,
      "grad_norm": 1.5027028322219849,
      "learning_rate": 5.2470350430667725e-06,
      "loss": 1.3803,
      "step": 9470
    },
    {
      "epoch": 1.7061022292276515,
      "grad_norm": 1.3048604726791382,
      "learning_rate": 5.240721546725602e-06,
      "loss": 1.1931,
      "step": 9471
    },
    {
      "epoch": 1.706282368835848,
      "grad_norm": 1.4758661985397339,
      "learning_rate": 5.234411640951947e-06,
      "loss": 1.5332,
      "step": 9472
    },
    {
      "epoch": 1.7064625084440441,
      "grad_norm": 1.5689239501953125,
      "learning_rate": 5.228105326251986e-06,
      "loss": 1.4179,
      "step": 9473
    },
    {
      "epoch": 1.7066426480522405,
      "grad_norm": 1.4091473817825317,
      "learning_rate": 5.221802603131609e-06,
      "loss": 1.3525,
      "step": 9474
    },
    {
      "epoch": 1.7068227876604367,
      "grad_norm": 1.368336796760559,
      "learning_rate": 5.215503472096417e-06,
      "loss": 1.2375,
      "step": 9475
    },
    {
      "epoch": 1.7070029272686331,
      "grad_norm": 1.6077932119369507,
      "learning_rate": 5.2092079336517104e-06,
      "loss": 1.4976,
      "step": 9476
    },
    {
      "epoch": 1.7071830668768295,
      "grad_norm": 1.5955305099487305,
      "learning_rate": 5.202915988302542e-06,
      "loss": 1.4345,
      "step": 9477
    },
    {
      "epoch": 1.707363206485026,
      "grad_norm": 1.3216801881790161,
      "learning_rate": 5.19662763655363e-06,
      "loss": 1.3437,
      "step": 9478
    },
    {
      "epoch": 1.7075433460932223,
      "grad_norm": 1.474382758140564,
      "learning_rate": 5.190342878909421e-06,
      "loss": 1.4399,
      "step": 9479
    },
    {
      "epoch": 1.7077234857014187,
      "grad_norm": 1.4468094110488892,
      "learning_rate": 5.184061715874089e-06,
      "loss": 1.3137,
      "step": 9480
    },
    {
      "epoch": 1.707903625309615,
      "grad_norm": 1.406829595565796,
      "learning_rate": 5.177784147951481e-06,
      "loss": 1.3677,
      "step": 9481
    },
    {
      "epoch": 1.7080837649178113,
      "grad_norm": 1.3837640285491943,
      "learning_rate": 5.171510175645211e-06,
      "loss": 1.1095,
      "step": 9482
    },
    {
      "epoch": 1.7082639045260075,
      "grad_norm": 1.5724126100540161,
      "learning_rate": 5.165239799458565e-06,
      "loss": 1.2619,
      "step": 9483
    },
    {
      "epoch": 1.708444044134204,
      "grad_norm": 1.479369878768921,
      "learning_rate": 5.158973019894536e-06,
      "loss": 1.2666,
      "step": 9484
    },
    {
      "epoch": 1.7086241837424003,
      "grad_norm": 1.56902015209198,
      "learning_rate": 5.152709837455849e-06,
      "loss": 1.3841,
      "step": 9485
    },
    {
      "epoch": 1.7088043233505967,
      "grad_norm": 1.6762185096740723,
      "learning_rate": 5.146450252644924e-06,
      "loss": 1.3552,
      "step": 9486
    },
    {
      "epoch": 1.7089844629587931,
      "grad_norm": 1.6517653465270996,
      "learning_rate": 5.140194265963927e-06,
      "loss": 1.3634,
      "step": 9487
    },
    {
      "epoch": 1.7091646025669895,
      "grad_norm": 1.5506644248962402,
      "learning_rate": 5.133941877914694e-06,
      "loss": 1.3462,
      "step": 9488
    },
    {
      "epoch": 1.7093447421751857,
      "grad_norm": 1.483857274055481,
      "learning_rate": 5.127693088998786e-06,
      "loss": 1.5136,
      "step": 9489
    },
    {
      "epoch": 1.7095248817833821,
      "grad_norm": 1.7106719017028809,
      "learning_rate": 5.121447899717474e-06,
      "loss": 1.3607,
      "step": 9490
    },
    {
      "epoch": 1.7097050213915783,
      "grad_norm": 1.4973533153533936,
      "learning_rate": 5.115206310571763e-06,
      "loss": 1.1919,
      "step": 9491
    },
    {
      "epoch": 1.7098851609997747,
      "grad_norm": 1.7351492643356323,
      "learning_rate": 5.1089683220623406e-06,
      "loss": 1.5578,
      "step": 9492
    },
    {
      "epoch": 1.7100653006079711,
      "grad_norm": 1.389843463897705,
      "learning_rate": 5.102733934689613e-06,
      "loss": 1.1202,
      "step": 9493
    },
    {
      "epoch": 1.7102454402161675,
      "grad_norm": 1.5885761976242065,
      "learning_rate": 5.096503148953697e-06,
      "loss": 1.4481,
      "step": 9494
    },
    {
      "epoch": 1.710425579824364,
      "grad_norm": 1.4827508926391602,
      "learning_rate": 5.090275965354429e-06,
      "loss": 1.2605,
      "step": 9495
    },
    {
      "epoch": 1.7106057194325603,
      "grad_norm": 1.606197714805603,
      "learning_rate": 5.084052384391346e-06,
      "loss": 1.5177,
      "step": 9496
    },
    {
      "epoch": 1.7107858590407568,
      "grad_norm": 1.622270107269287,
      "learning_rate": 5.077832406563704e-06,
      "loss": 1.377,
      "step": 9497
    },
    {
      "epoch": 1.710965998648953,
      "grad_norm": 1.5996332168579102,
      "learning_rate": 5.071616032370469e-06,
      "loss": 1.2593,
      "step": 9498
    },
    {
      "epoch": 1.7111461382571493,
      "grad_norm": 1.6517282724380493,
      "learning_rate": 5.065403262310309e-06,
      "loss": 1.5048,
      "step": 9499
    },
    {
      "epoch": 1.7113262778653455,
      "grad_norm": 1.6243906021118164,
      "learning_rate": 5.059194096881603e-06,
      "loss": 1.1542,
      "step": 9500
    },
    {
      "epoch": 1.711506417473542,
      "grad_norm": 1.429052472114563,
      "learning_rate": 5.05298853658247e-06,
      "loss": 1.0496,
      "step": 9501
    },
    {
      "epoch": 1.7116865570817383,
      "grad_norm": 1.4577680826187134,
      "learning_rate": 5.046786581910701e-06,
      "loss": 1.925,
      "step": 9502
    },
    {
      "epoch": 1.7118666966899347,
      "grad_norm": 1.3322657346725464,
      "learning_rate": 5.040588233363813e-06,
      "loss": 1.6145,
      "step": 9503
    },
    {
      "epoch": 1.7120468362981311,
      "grad_norm": 1.3788909912109375,
      "learning_rate": 5.034393491439043e-06,
      "loss": 1.69,
      "step": 9504
    },
    {
      "epoch": 1.7122269759063276,
      "grad_norm": 1.3408071994781494,
      "learning_rate": 5.0282023566333215e-06,
      "loss": 1.5195,
      "step": 9505
    },
    {
      "epoch": 1.7124071155145237,
      "grad_norm": 1.4907881021499634,
      "learning_rate": 5.0220148294433055e-06,
      "loss": 1.7621,
      "step": 9506
    },
    {
      "epoch": 1.7125872551227201,
      "grad_norm": 1.6795850992202759,
      "learning_rate": 5.015830910365349e-06,
      "loss": 1.6156,
      "step": 9507
    },
    {
      "epoch": 1.7127673947309163,
      "grad_norm": 1.6554476022720337,
      "learning_rate": 5.009650599895527e-06,
      "loss": 2.0536,
      "step": 9508
    },
    {
      "epoch": 1.7129475343391127,
      "grad_norm": 1.539005994796753,
      "learning_rate": 5.0034738985296095e-06,
      "loss": 1.6712,
      "step": 9509
    },
    {
      "epoch": 1.7131276739473091,
      "grad_norm": 2.006694793701172,
      "learning_rate": 4.997300806763111e-06,
      "loss": 1.874,
      "step": 9510
    },
    {
      "epoch": 1.7133078135555055,
      "grad_norm": 2.0179882049560547,
      "learning_rate": 4.99113132509122e-06,
      "loss": 1.8106,
      "step": 9511
    },
    {
      "epoch": 1.713487953163702,
      "grad_norm": 1.5218029022216797,
      "learning_rate": 4.98496545400885e-06,
      "loss": 1.3044,
      "step": 9512
    },
    {
      "epoch": 1.7136680927718984,
      "grad_norm": 1.4121038913726807,
      "learning_rate": 4.978803194010623e-06,
      "loss": 1.4674,
      "step": 9513
    },
    {
      "epoch": 1.7138482323800945,
      "grad_norm": 1.510233759880066,
      "learning_rate": 4.972644545590871e-06,
      "loss": 1.5547,
      "step": 9514
    },
    {
      "epoch": 1.714028371988291,
      "grad_norm": 1.3907474279403687,
      "learning_rate": 4.966489509243655e-06,
      "loss": 1.4112,
      "step": 9515
    },
    {
      "epoch": 1.7142085115964871,
      "grad_norm": 1.3813812732696533,
      "learning_rate": 4.960338085462713e-06,
      "loss": 1.3058,
      "step": 9516
    },
    {
      "epoch": 1.7143886512046835,
      "grad_norm": 1.4947965145111084,
      "learning_rate": 4.9541902747415045e-06,
      "loss": 1.521,
      "step": 9517
    },
    {
      "epoch": 1.71456879081288,
      "grad_norm": 1.4388364553451538,
      "learning_rate": 4.948046077573204e-06,
      "loss": 1.3979,
      "step": 9518
    },
    {
      "epoch": 1.7147489304210763,
      "grad_norm": 1.4114837646484375,
      "learning_rate": 4.941905494450716e-06,
      "loss": 1.2735,
      "step": 9519
    },
    {
      "epoch": 1.7149290700292728,
      "grad_norm": 1.451129674911499,
      "learning_rate": 4.935768525866618e-06,
      "loss": 1.3008,
      "step": 9520
    },
    {
      "epoch": 1.7151092096374692,
      "grad_norm": 1.3653805255889893,
      "learning_rate": 4.929635172313219e-06,
      "loss": 1.2658,
      "step": 9521
    },
    {
      "epoch": 1.7152893492456653,
      "grad_norm": 1.4697612524032593,
      "learning_rate": 4.923505434282538e-06,
      "loss": 1.3208,
      "step": 9522
    },
    {
      "epoch": 1.7154694888538617,
      "grad_norm": 1.2516428232192993,
      "learning_rate": 4.917379312266279e-06,
      "loss": 1.103,
      "step": 9523
    },
    {
      "epoch": 1.715649628462058,
      "grad_norm": 1.42966628074646,
      "learning_rate": 4.911256806755909e-06,
      "loss": 1.3999,
      "step": 9524
    },
    {
      "epoch": 1.7158297680702543,
      "grad_norm": 1.522367238998413,
      "learning_rate": 4.9051379182425525e-06,
      "loss": 1.4984,
      "step": 9525
    },
    {
      "epoch": 1.7160099076784507,
      "grad_norm": 1.519760251045227,
      "learning_rate": 4.899022647217066e-06,
      "loss": 1.5008,
      "step": 9526
    },
    {
      "epoch": 1.7161900472866471,
      "grad_norm": 1.4056733846664429,
      "learning_rate": 4.892910994170019e-06,
      "loss": 1.4095,
      "step": 9527
    },
    {
      "epoch": 1.7163701868948436,
      "grad_norm": 1.4832727909088135,
      "learning_rate": 4.886802959591674e-06,
      "loss": 1.5056,
      "step": 9528
    },
    {
      "epoch": 1.71655032650304,
      "grad_norm": 1.3613221645355225,
      "learning_rate": 4.880698543972029e-06,
      "loss": 1.3821,
      "step": 9529
    },
    {
      "epoch": 1.7167304661112364,
      "grad_norm": 1.445778727531433,
      "learning_rate": 4.874597747800769e-06,
      "loss": 1.2269,
      "step": 9530
    },
    {
      "epoch": 1.7169106057194325,
      "grad_norm": 1.4002124071121216,
      "learning_rate": 4.8685005715672915e-06,
      "loss": 1.1958,
      "step": 9531
    },
    {
      "epoch": 1.717090745327629,
      "grad_norm": 1.5762535333633423,
      "learning_rate": 4.862407015760712e-06,
      "loss": 1.6436,
      "step": 9532
    },
    {
      "epoch": 1.7172708849358251,
      "grad_norm": 1.7474557161331177,
      "learning_rate": 4.856317080869865e-06,
      "loss": 1.6265,
      "step": 9533
    },
    {
      "epoch": 1.7174510245440215,
      "grad_norm": 1.5225788354873657,
      "learning_rate": 4.850230767383274e-06,
      "loss": 1.408,
      "step": 9534
    },
    {
      "epoch": 1.717631164152218,
      "grad_norm": 1.4675949811935425,
      "learning_rate": 4.844148075789173e-06,
      "loss": 1.3727,
      "step": 9535
    },
    {
      "epoch": 1.7178113037604144,
      "grad_norm": 1.518534541130066,
      "learning_rate": 4.838069006575524e-06,
      "loss": 1.5101,
      "step": 9536
    },
    {
      "epoch": 1.7179914433686108,
      "grad_norm": 1.394788146018982,
      "learning_rate": 4.8319935602299805e-06,
      "loss": 1.2311,
      "step": 9537
    },
    {
      "epoch": 1.7181715829768072,
      "grad_norm": 1.578537106513977,
      "learning_rate": 4.825921737239914e-06,
      "loss": 1.3598,
      "step": 9538
    },
    {
      "epoch": 1.7183517225850033,
      "grad_norm": 1.6075000762939453,
      "learning_rate": 4.8198535380924e-06,
      "loss": 1.4754,
      "step": 9539
    },
    {
      "epoch": 1.7185318621931998,
      "grad_norm": 1.4809800386428833,
      "learning_rate": 4.81378896327423e-06,
      "loss": 1.4137,
      "step": 9540
    },
    {
      "epoch": 1.718712001801396,
      "grad_norm": 1.6020078659057617,
      "learning_rate": 4.8077280132719006e-06,
      "loss": 1.3505,
      "step": 9541
    },
    {
      "epoch": 1.7188921414095923,
      "grad_norm": 1.5865106582641602,
      "learning_rate": 4.801670688571608e-06,
      "loss": 1.4025,
      "step": 9542
    },
    {
      "epoch": 1.7190722810177888,
      "grad_norm": 1.6301887035369873,
      "learning_rate": 4.795616989659285e-06,
      "loss": 1.4478,
      "step": 9543
    },
    {
      "epoch": 1.7192524206259852,
      "grad_norm": 1.5299969911575317,
      "learning_rate": 4.789566917020549e-06,
      "loss": 1.3583,
      "step": 9544
    },
    {
      "epoch": 1.7194325602341816,
      "grad_norm": 1.5433638095855713,
      "learning_rate": 4.783520471140735e-06,
      "loss": 1.4669,
      "step": 9545
    },
    {
      "epoch": 1.719612699842378,
      "grad_norm": 1.5660542249679565,
      "learning_rate": 4.777477652504875e-06,
      "loss": 1.3841,
      "step": 9546
    },
    {
      "epoch": 1.7197928394505742,
      "grad_norm": 1.6740514039993286,
      "learning_rate": 4.771438461597744e-06,
      "loss": 1.5447,
      "step": 9547
    },
    {
      "epoch": 1.7199729790587706,
      "grad_norm": 1.7518460750579834,
      "learning_rate": 4.765402898903792e-06,
      "loss": 1.6251,
      "step": 9548
    },
    {
      "epoch": 1.7201531186669667,
      "grad_norm": 1.6132197380065918,
      "learning_rate": 4.759370964907184e-06,
      "loss": 1.3366,
      "step": 9549
    },
    {
      "epoch": 1.7203332582751631,
      "grad_norm": 1.5110498666763306,
      "learning_rate": 4.753342660091797e-06,
      "loss": 1.3389,
      "step": 9550
    },
    {
      "epoch": 1.7205133978833596,
      "grad_norm": 1.4261059761047363,
      "learning_rate": 4.747317984941213e-06,
      "loss": 0.8839,
      "step": 9551
    },
    {
      "epoch": 1.720693537491556,
      "grad_norm": 1.5000559091567993,
      "learning_rate": 4.7412969399387495e-06,
      "loss": 1.8664,
      "step": 9552
    },
    {
      "epoch": 1.7208736770997524,
      "grad_norm": 1.3451316356658936,
      "learning_rate": 4.735279525567399e-06,
      "loss": 1.766,
      "step": 9553
    },
    {
      "epoch": 1.7210538167079488,
      "grad_norm": 1.648999571800232,
      "learning_rate": 4.72926574230988e-06,
      "loss": 1.7428,
      "step": 9554
    },
    {
      "epoch": 1.7212339563161452,
      "grad_norm": 1.5035638809204102,
      "learning_rate": 4.723255590648601e-06,
      "loss": 1.9098,
      "step": 9555
    },
    {
      "epoch": 1.7214140959243414,
      "grad_norm": 1.4117939472198486,
      "learning_rate": 4.7172490710657155e-06,
      "loss": 1.8428,
      "step": 9556
    },
    {
      "epoch": 1.7215942355325378,
      "grad_norm": 1.565474033355713,
      "learning_rate": 4.71124618404305e-06,
      "loss": 1.8322,
      "step": 9557
    },
    {
      "epoch": 1.721774375140734,
      "grad_norm": 1.5569462776184082,
      "learning_rate": 4.705246930062163e-06,
      "loss": 2.1899,
      "step": 9558
    },
    {
      "epoch": 1.7219545147489304,
      "grad_norm": 1.4860347509384155,
      "learning_rate": 4.6992513096042965e-06,
      "loss": 1.7084,
      "step": 9559
    },
    {
      "epoch": 1.7221346543571268,
      "grad_norm": 1.6857378482818604,
      "learning_rate": 4.693259323150428e-06,
      "loss": 1.8986,
      "step": 9560
    },
    {
      "epoch": 1.7223147939653232,
      "grad_norm": 1.7689216136932373,
      "learning_rate": 4.687270971181234e-06,
      "loss": 1.7561,
      "step": 9561
    },
    {
      "epoch": 1.7224949335735196,
      "grad_norm": 1.583787202835083,
      "learning_rate": 4.6812862541770865e-06,
      "loss": 1.7573,
      "step": 9562
    },
    {
      "epoch": 1.722675073181716,
      "grad_norm": 1.6196471452713013,
      "learning_rate": 4.675305172618083e-06,
      "loss": 1.7679,
      "step": 9563
    },
    {
      "epoch": 1.7228552127899122,
      "grad_norm": 1.5555570125579834,
      "learning_rate": 4.669327726984019e-06,
      "loss": 1.7591,
      "step": 9564
    },
    {
      "epoch": 1.7230353523981086,
      "grad_norm": 1.3879055976867676,
      "learning_rate": 4.663353917754399e-06,
      "loss": 1.3522,
      "step": 9565
    },
    {
      "epoch": 1.7232154920063047,
      "grad_norm": 1.4087027311325073,
      "learning_rate": 4.65738374540845e-06,
      "loss": 1.4467,
      "step": 9566
    },
    {
      "epoch": 1.7233956316145012,
      "grad_norm": 1.3224717378616333,
      "learning_rate": 4.651417210425097e-06,
      "loss": 1.2774,
      "step": 9567
    },
    {
      "epoch": 1.7235757712226976,
      "grad_norm": 1.3739231824874878,
      "learning_rate": 4.645454313282965e-06,
      "loss": 1.3561,
      "step": 9568
    },
    {
      "epoch": 1.723755910830894,
      "grad_norm": 1.5082818269729614,
      "learning_rate": 4.6394950544603966e-06,
      "loss": 1.5909,
      "step": 9569
    },
    {
      "epoch": 1.7239360504390904,
      "grad_norm": 1.4290140867233276,
      "learning_rate": 4.633539434435441e-06,
      "loss": 1.3721,
      "step": 9570
    },
    {
      "epoch": 1.7241161900472868,
      "grad_norm": 1.546866774559021,
      "learning_rate": 4.62758745368585e-06,
      "loss": 1.658,
      "step": 9571
    },
    {
      "epoch": 1.724296329655483,
      "grad_norm": 1.5424435138702393,
      "learning_rate": 4.621639112689096e-06,
      "loss": 1.409,
      "step": 9572
    },
    {
      "epoch": 1.7244764692636794,
      "grad_norm": 1.443482756614685,
      "learning_rate": 4.615694411922355e-06,
      "loss": 1.5087,
      "step": 9573
    },
    {
      "epoch": 1.7246566088718756,
      "grad_norm": 1.4095914363861084,
      "learning_rate": 4.609753351862489e-06,
      "loss": 1.3148,
      "step": 9574
    },
    {
      "epoch": 1.724836748480072,
      "grad_norm": 1.5599360466003418,
      "learning_rate": 4.603815932986105e-06,
      "loss": 1.6604,
      "step": 9575
    },
    {
      "epoch": 1.7250168880882684,
      "grad_norm": 1.382007360458374,
      "learning_rate": 4.597882155769501e-06,
      "loss": 1.2242,
      "step": 9576
    },
    {
      "epoch": 1.7251970276964648,
      "grad_norm": 1.5322506427764893,
      "learning_rate": 4.591952020688673e-06,
      "loss": 1.406,
      "step": 9577
    },
    {
      "epoch": 1.7253771673046612,
      "grad_norm": 1.4547069072723389,
      "learning_rate": 4.58602552821934e-06,
      "loss": 1.4651,
      "step": 9578
    },
    {
      "epoch": 1.7255573069128576,
      "grad_norm": 1.5488733053207397,
      "learning_rate": 4.580102678836906e-06,
      "loss": 1.4405,
      "step": 9579
    },
    {
      "epoch": 1.7257374465210538,
      "grad_norm": 1.599420189857483,
      "learning_rate": 4.57418347301653e-06,
      "loss": 1.4575,
      "step": 9580
    },
    {
      "epoch": 1.7259175861292502,
      "grad_norm": 1.4575682878494263,
      "learning_rate": 4.568267911233021e-06,
      "loss": 1.3867,
      "step": 9581
    },
    {
      "epoch": 1.7260977257374466,
      "grad_norm": 1.4409619569778442,
      "learning_rate": 4.562355993960931e-06,
      "loss": 1.2522,
      "step": 9582
    },
    {
      "epoch": 1.7262778653456428,
      "grad_norm": 1.6051923036575317,
      "learning_rate": 4.556447721674506e-06,
      "loss": 1.4629,
      "step": 9583
    },
    {
      "epoch": 1.7264580049538392,
      "grad_norm": 1.5101816654205322,
      "learning_rate": 4.5505430948477145e-06,
      "loss": 1.4694,
      "step": 9584
    },
    {
      "epoch": 1.7266381445620356,
      "grad_norm": 1.429427981376648,
      "learning_rate": 4.544642113954223e-06,
      "loss": 1.2766,
      "step": 9585
    },
    {
      "epoch": 1.726818284170232,
      "grad_norm": 1.5331566333770752,
      "learning_rate": 4.538744779467396e-06,
      "loss": 1.6293,
      "step": 9586
    },
    {
      "epoch": 1.7269984237784284,
      "grad_norm": 1.4530586004257202,
      "learning_rate": 4.532851091860324e-06,
      "loss": 1.2788,
      "step": 9587
    },
    {
      "epoch": 1.7271785633866248,
      "grad_norm": 1.541965126991272,
      "learning_rate": 4.526961051605777e-06,
      "loss": 1.3954,
      "step": 9588
    },
    {
      "epoch": 1.727358702994821,
      "grad_norm": 1.6654447317123413,
      "learning_rate": 4.521074659176277e-06,
      "loss": 1.4566,
      "step": 9589
    },
    {
      "epoch": 1.7275388426030174,
      "grad_norm": 1.4552208185195923,
      "learning_rate": 4.515191915044014e-06,
      "loss": 1.2776,
      "step": 9590
    },
    {
      "epoch": 1.7277189822112136,
      "grad_norm": 1.4932048320770264,
      "learning_rate": 4.509312819680905e-06,
      "loss": 1.4921,
      "step": 9591
    },
    {
      "epoch": 1.72789912181941,
      "grad_norm": 1.4901301860809326,
      "learning_rate": 4.5034373735585664e-06,
      "loss": 1.3097,
      "step": 9592
    },
    {
      "epoch": 1.7280792614276064,
      "grad_norm": 1.4690343141555786,
      "learning_rate": 4.49756557714831e-06,
      "loss": 1.2068,
      "step": 9593
    },
    {
      "epoch": 1.7282594010358028,
      "grad_norm": 1.6519129276275635,
      "learning_rate": 4.491697430921182e-06,
      "loss": 1.3556,
      "step": 9594
    },
    {
      "epoch": 1.7284395406439992,
      "grad_norm": 1.4559097290039062,
      "learning_rate": 4.485832935347928e-06,
      "loss": 1.1919,
      "step": 9595
    },
    {
      "epoch": 1.7286196802521956,
      "grad_norm": 1.389897346496582,
      "learning_rate": 4.479972090898982e-06,
      "loss": 1.2374,
      "step": 9596
    },
    {
      "epoch": 1.7287998198603918,
      "grad_norm": 1.7157933712005615,
      "learning_rate": 4.47411489804449e-06,
      "loss": 1.4309,
      "step": 9597
    },
    {
      "epoch": 1.7289799594685882,
      "grad_norm": 1.5302042961120605,
      "learning_rate": 4.468261357254339e-06,
      "loss": 1.3498,
      "step": 9598
    },
    {
      "epoch": 1.7291600990767844,
      "grad_norm": 1.4530069828033447,
      "learning_rate": 4.462411468998085e-06,
      "loss": 1.2727,
      "step": 9599
    },
    {
      "epoch": 1.7293402386849808,
      "grad_norm": 1.5920439958572388,
      "learning_rate": 4.456565233745003e-06,
      "loss": 1.1456,
      "step": 9600
    },
    {
      "epoch": 1.7295203782931772,
      "grad_norm": 1.650061845779419,
      "learning_rate": 4.450722651964073e-06,
      "loss": 1.2532,
      "step": 9601
    },
    {
      "epoch": 1.7297005179013736,
      "grad_norm": 1.317950963973999,
      "learning_rate": 4.444883724123983e-06,
      "loss": 1.7089,
      "step": 9602
    },
    {
      "epoch": 1.72988065750957,
      "grad_norm": 1.2714366912841797,
      "learning_rate": 4.439048450693134e-06,
      "loss": 1.5815,
      "step": 9603
    },
    {
      "epoch": 1.7300607971177664,
      "grad_norm": 1.484217643737793,
      "learning_rate": 4.433216832139625e-06,
      "loss": 2.1427,
      "step": 9604
    },
    {
      "epoch": 1.7302409367259626,
      "grad_norm": 1.3989589214324951,
      "learning_rate": 4.42738886893127e-06,
      "loss": 1.7924,
      "step": 9605
    },
    {
      "epoch": 1.730421076334159,
      "grad_norm": 1.5513818264007568,
      "learning_rate": 4.42156456153558e-06,
      "loss": 1.8296,
      "step": 9606
    },
    {
      "epoch": 1.7306012159423552,
      "grad_norm": 1.5463838577270508,
      "learning_rate": 4.415743910419773e-06,
      "loss": 2.0924,
      "step": 9607
    },
    {
      "epoch": 1.7307813555505516,
      "grad_norm": 1.521953821182251,
      "learning_rate": 4.409926916050794e-06,
      "loss": 1.6124,
      "step": 9608
    },
    {
      "epoch": 1.730961495158748,
      "grad_norm": 1.443172812461853,
      "learning_rate": 4.40411357889527e-06,
      "loss": 1.6189,
      "step": 9609
    },
    {
      "epoch": 1.7311416347669444,
      "grad_norm": 1.6563770771026611,
      "learning_rate": 4.398303899419548e-06,
      "loss": 1.5947,
      "step": 9610
    },
    {
      "epoch": 1.7313217743751408,
      "grad_norm": 1.528943657875061,
      "learning_rate": 4.392497878089669e-06,
      "loss": 1.4543,
      "step": 9611
    },
    {
      "epoch": 1.7315019139833372,
      "grad_norm": 1.5132312774658203,
      "learning_rate": 4.3866955153713975e-06,
      "loss": 1.6889,
      "step": 9612
    },
    {
      "epoch": 1.7316820535915336,
      "grad_norm": 1.5697942972183228,
      "learning_rate": 4.380896811730212e-06,
      "loss": 1.638,
      "step": 9613
    },
    {
      "epoch": 1.7318621931997298,
      "grad_norm": 1.4939794540405273,
      "learning_rate": 4.375101767631251e-06,
      "loss": 1.3965,
      "step": 9614
    },
    {
      "epoch": 1.7320423328079262,
      "grad_norm": 1.364015817642212,
      "learning_rate": 4.369310383539399e-06,
      "loss": 1.4775,
      "step": 9615
    },
    {
      "epoch": 1.7322224724161224,
      "grad_norm": 1.4071285724639893,
      "learning_rate": 4.363522659919239e-06,
      "loss": 1.3233,
      "step": 9616
    },
    {
      "epoch": 1.7324026120243188,
      "grad_norm": 1.4012006521224976,
      "learning_rate": 4.357738597235067e-06,
      "loss": 1.4234,
      "step": 9617
    },
    {
      "epoch": 1.7325827516325152,
      "grad_norm": 1.2294416427612305,
      "learning_rate": 4.351958195950878e-06,
      "loss": 1.1702,
      "step": 9618
    },
    {
      "epoch": 1.7327628912407116,
      "grad_norm": 1.31303870677948,
      "learning_rate": 4.3461814565303625e-06,
      "loss": 1.3742,
      "step": 9619
    },
    {
      "epoch": 1.732943030848908,
      "grad_norm": 1.772634744644165,
      "learning_rate": 4.3404083794369385e-06,
      "loss": 1.4506,
      "step": 9620
    },
    {
      "epoch": 1.7331231704571044,
      "grad_norm": 1.32028067111969,
      "learning_rate": 4.334638965133703e-06,
      "loss": 1.3303,
      "step": 9621
    },
    {
      "epoch": 1.7333033100653006,
      "grad_norm": 1.5191190242767334,
      "learning_rate": 4.328873214083495e-06,
      "loss": 1.5835,
      "step": 9622
    },
    {
      "epoch": 1.733483449673497,
      "grad_norm": 1.4485267400741577,
      "learning_rate": 4.323111126748836e-06,
      "loss": 1.3661,
      "step": 9623
    },
    {
      "epoch": 1.7336635892816932,
      "grad_norm": 1.4925565719604492,
      "learning_rate": 4.317352703591948e-06,
      "loss": 1.384,
      "step": 9624
    },
    {
      "epoch": 1.7338437288898896,
      "grad_norm": 1.5818308591842651,
      "learning_rate": 4.31159794507478e-06,
      "loss": 1.4439,
      "step": 9625
    },
    {
      "epoch": 1.734023868498086,
      "grad_norm": 1.4135690927505493,
      "learning_rate": 4.305846851658968e-06,
      "loss": 1.2722,
      "step": 9626
    },
    {
      "epoch": 1.7342040081062824,
      "grad_norm": 1.4261449575424194,
      "learning_rate": 4.3000994238058644e-06,
      "loss": 1.3275,
      "step": 9627
    },
    {
      "epoch": 1.7343841477144788,
      "grad_norm": 1.3457294702529907,
      "learning_rate": 4.294355661976529e-06,
      "loss": 1.1279,
      "step": 9628
    },
    {
      "epoch": 1.7345642873226752,
      "grad_norm": 1.6347578763961792,
      "learning_rate": 4.288615566631715e-06,
      "loss": 1.6883,
      "step": 9629
    },
    {
      "epoch": 1.7347444269308714,
      "grad_norm": 1.4078067541122437,
      "learning_rate": 4.282879138231888e-06,
      "loss": 1.2738,
      "step": 9630
    },
    {
      "epoch": 1.7349245665390678,
      "grad_norm": 1.4295357465744019,
      "learning_rate": 4.27714637723724e-06,
      "loss": 1.3309,
      "step": 9631
    },
    {
      "epoch": 1.735104706147264,
      "grad_norm": 1.5635004043579102,
      "learning_rate": 4.2714172841076375e-06,
      "loss": 1.4521,
      "step": 9632
    },
    {
      "epoch": 1.7352848457554604,
      "grad_norm": 1.458160161972046,
      "learning_rate": 4.265691859302668e-06,
      "loss": 1.2881,
      "step": 9633
    },
    {
      "epoch": 1.7354649853636568,
      "grad_norm": 1.4123752117156982,
      "learning_rate": 4.2599701032816185e-06,
      "loss": 1.4002,
      "step": 9634
    },
    {
      "epoch": 1.7356451249718532,
      "grad_norm": 1.5402745008468628,
      "learning_rate": 4.2542520165034904e-06,
      "loss": 1.332,
      "step": 9635
    },
    {
      "epoch": 1.7358252645800496,
      "grad_norm": 1.4241961240768433,
      "learning_rate": 4.248537599426982e-06,
      "loss": 1.2178,
      "step": 9636
    },
    {
      "epoch": 1.736005404188246,
      "grad_norm": 1.4660217761993408,
      "learning_rate": 4.242826852510501e-06,
      "loss": 1.3843,
      "step": 9637
    },
    {
      "epoch": 1.7361855437964422,
      "grad_norm": 1.7371903657913208,
      "learning_rate": 4.2371197762121676e-06,
      "loss": 1.4731,
      "step": 9638
    },
    {
      "epoch": 1.7363656834046386,
      "grad_norm": 1.4508026838302612,
      "learning_rate": 4.231416370989782e-06,
      "loss": 1.1948,
      "step": 9639
    },
    {
      "epoch": 1.736545823012835,
      "grad_norm": 1.5844625234603882,
      "learning_rate": 4.225716637300892e-06,
      "loss": 1.4146,
      "step": 9640
    },
    {
      "epoch": 1.7367259626210312,
      "grad_norm": 1.5621311664581299,
      "learning_rate": 4.220020575602718e-06,
      "loss": 1.4425,
      "step": 9641
    },
    {
      "epoch": 1.7369061022292276,
      "grad_norm": 1.711195468902588,
      "learning_rate": 4.2143281863521955e-06,
      "loss": 1.2537,
      "step": 9642
    },
    {
      "epoch": 1.737086241837424,
      "grad_norm": 1.542397379875183,
      "learning_rate": 4.20863947000596e-06,
      "loss": 1.3599,
      "step": 9643
    },
    {
      "epoch": 1.7372663814456204,
      "grad_norm": 1.608813762664795,
      "learning_rate": 4.202954427020356e-06,
      "loss": 1.315,
      "step": 9644
    },
    {
      "epoch": 1.7374465210538168,
      "grad_norm": 1.5510209798812866,
      "learning_rate": 4.197273057851464e-06,
      "loss": 1.2582,
      "step": 9645
    },
    {
      "epoch": 1.7376266606620132,
      "grad_norm": 1.5419559478759766,
      "learning_rate": 4.191595362955003e-06,
      "loss": 1.3826,
      "step": 9646
    },
    {
      "epoch": 1.7378068002702094,
      "grad_norm": 1.4365555047988892,
      "learning_rate": 4.185921342786448e-06,
      "loss": 1.3205,
      "step": 9647
    },
    {
      "epoch": 1.7379869398784058,
      "grad_norm": 1.6133428812026978,
      "learning_rate": 4.180250997800972e-06,
      "loss": 1.3987,
      "step": 9648
    },
    {
      "epoch": 1.738167079486602,
      "grad_norm": 1.4892632961273193,
      "learning_rate": 4.174584328453429e-06,
      "loss": 1.2221,
      "step": 9649
    },
    {
      "epoch": 1.7383472190947984,
      "grad_norm": 1.7542753219604492,
      "learning_rate": 4.168921335198428e-06,
      "loss": 1.641,
      "step": 9650
    },
    {
      "epoch": 1.7385273587029948,
      "grad_norm": 1.3664988279342651,
      "learning_rate": 4.163262018490227e-06,
      "loss": 1.1056,
      "step": 9651
    },
    {
      "epoch": 1.7387074983111912,
      "grad_norm": 1.446403980255127,
      "learning_rate": 4.157606378782819e-06,
      "loss": 1.7099,
      "step": 9652
    },
    {
      "epoch": 1.7388876379193876,
      "grad_norm": 1.3117256164550781,
      "learning_rate": 4.151954416529896e-06,
      "loss": 1.5486,
      "step": 9653
    },
    {
      "epoch": 1.739067777527584,
      "grad_norm": 1.3918155431747437,
      "learning_rate": 4.146306132184863e-06,
      "loss": 1.9564,
      "step": 9654
    },
    {
      "epoch": 1.7392479171357802,
      "grad_norm": 1.4384269714355469,
      "learning_rate": 4.1406615262008175e-06,
      "loss": 2.0072,
      "step": 9655
    },
    {
      "epoch": 1.7394280567439766,
      "grad_norm": 1.4911695718765259,
      "learning_rate": 4.1350205990305644e-06,
      "loss": 1.9314,
      "step": 9656
    },
    {
      "epoch": 1.7396081963521728,
      "grad_norm": 1.3657127618789673,
      "learning_rate": 4.1293833511266315e-06,
      "loss": 1.7937,
      "step": 9657
    },
    {
      "epoch": 1.7397883359603692,
      "grad_norm": 1.3657654523849487,
      "learning_rate": 4.123749782941206e-06,
      "loss": 1.3959,
      "step": 9658
    },
    {
      "epoch": 1.7399684755685656,
      "grad_norm": 1.6948521137237549,
      "learning_rate": 4.118119894926236e-06,
      "loss": 1.5854,
      "step": 9659
    },
    {
      "epoch": 1.740148615176762,
      "grad_norm": 1.6691831350326538,
      "learning_rate": 4.1124936875333365e-06,
      "loss": 1.6065,
      "step": 9660
    },
    {
      "epoch": 1.7403287547849584,
      "grad_norm": 1.6761059761047363,
      "learning_rate": 4.106871161213849e-06,
      "loss": 1.6582,
      "step": 9661
    },
    {
      "epoch": 1.7405088943931548,
      "grad_norm": 1.5670331716537476,
      "learning_rate": 4.101252316418791e-06,
      "loss": 1.8422,
      "step": 9662
    },
    {
      "epoch": 1.740689034001351,
      "grad_norm": 1.4325658082962036,
      "learning_rate": 4.095637153598919e-06,
      "loss": 1.4125,
      "step": 9663
    },
    {
      "epoch": 1.7408691736095474,
      "grad_norm": 1.3561815023422241,
      "learning_rate": 4.090025673204684e-06,
      "loss": 1.3679,
      "step": 9664
    },
    {
      "epoch": 1.7410493132177436,
      "grad_norm": 1.2881895303726196,
      "learning_rate": 4.0844178756862195e-06,
      "loss": 1.2346,
      "step": 9665
    },
    {
      "epoch": 1.74122945282594,
      "grad_norm": 1.5560855865478516,
      "learning_rate": 4.078813761493394e-06,
      "loss": 1.3696,
      "step": 9666
    },
    {
      "epoch": 1.7414095924341364,
      "grad_norm": 1.4385424852371216,
      "learning_rate": 4.073213331075759e-06,
      "loss": 1.4158,
      "step": 9667
    },
    {
      "epoch": 1.7415897320423328,
      "grad_norm": 1.4203969240188599,
      "learning_rate": 4.067616584882583e-06,
      "loss": 1.6062,
      "step": 9668
    },
    {
      "epoch": 1.7417698716505292,
      "grad_norm": 1.4013786315917969,
      "learning_rate": 4.062023523362829e-06,
      "loss": 1.324,
      "step": 9669
    },
    {
      "epoch": 1.7419500112587256,
      "grad_norm": 1.30453622341156,
      "learning_rate": 4.056434146965171e-06,
      "loss": 1.2226,
      "step": 9670
    },
    {
      "epoch": 1.742130150866922,
      "grad_norm": 1.3937655687332153,
      "learning_rate": 4.05084845613799e-06,
      "loss": 1.3982,
      "step": 9671
    },
    {
      "epoch": 1.7423102904751182,
      "grad_norm": 1.4252763986587524,
      "learning_rate": 4.045266451329361e-06,
      "loss": 1.4014,
      "step": 9672
    },
    {
      "epoch": 1.7424904300833146,
      "grad_norm": 1.439408779144287,
      "learning_rate": 4.039688132987074e-06,
      "loss": 1.372,
      "step": 9673
    },
    {
      "epoch": 1.7426705696915108,
      "grad_norm": 1.4185879230499268,
      "learning_rate": 4.034113501558623e-06,
      "loss": 1.3766,
      "step": 9674
    },
    {
      "epoch": 1.7428507092997072,
      "grad_norm": 1.472978115081787,
      "learning_rate": 4.028542557491194e-06,
      "loss": 1.3818,
      "step": 9675
    },
    {
      "epoch": 1.7430308489079036,
      "grad_norm": 1.5642225742340088,
      "learning_rate": 4.022975301231685e-06,
      "loss": 1.5645,
      "step": 9676
    },
    {
      "epoch": 1.7432109885161,
      "grad_norm": 1.5769459009170532,
      "learning_rate": 4.017411733226717e-06,
      "loss": 1.5387,
      "step": 9677
    },
    {
      "epoch": 1.7433911281242964,
      "grad_norm": 1.4561134576797485,
      "learning_rate": 4.011851853922572e-06,
      "loss": 1.3251,
      "step": 9678
    },
    {
      "epoch": 1.7435712677324928,
      "grad_norm": 1.4647002220153809,
      "learning_rate": 4.006295663765275e-06,
      "loss": 1.5141,
      "step": 9679
    },
    {
      "epoch": 1.743751407340689,
      "grad_norm": 1.4577445983886719,
      "learning_rate": 4.000743163200532e-06,
      "loss": 1.472,
      "step": 9680
    },
    {
      "epoch": 1.7439315469488854,
      "grad_norm": 1.4081840515136719,
      "learning_rate": 3.995194352673765e-06,
      "loss": 1.2726,
      "step": 9681
    },
    {
      "epoch": 1.7441116865570816,
      "grad_norm": 1.6844600439071655,
      "learning_rate": 3.989649232630105e-06,
      "loss": 1.5609,
      "step": 9682
    },
    {
      "epoch": 1.744291826165278,
      "grad_norm": 1.575162410736084,
      "learning_rate": 3.984107803514369e-06,
      "loss": 1.4024,
      "step": 9683
    },
    {
      "epoch": 1.7444719657734744,
      "grad_norm": 1.460941195487976,
      "learning_rate": 3.978570065771098e-06,
      "loss": 1.4437,
      "step": 9684
    },
    {
      "epoch": 1.7446521053816708,
      "grad_norm": 1.5462037324905396,
      "learning_rate": 3.9730360198445175e-06,
      "loss": 1.6569,
      "step": 9685
    },
    {
      "epoch": 1.7448322449898672,
      "grad_norm": 1.506374716758728,
      "learning_rate": 3.967505666178556e-06,
      "loss": 1.4948,
      "step": 9686
    },
    {
      "epoch": 1.7450123845980636,
      "grad_norm": 1.4880439043045044,
      "learning_rate": 3.961979005216881e-06,
      "loss": 1.4177,
      "step": 9687
    },
    {
      "epoch": 1.7451925242062598,
      "grad_norm": 1.5648796558380127,
      "learning_rate": 3.956456037402828e-06,
      "loss": 1.5054,
      "step": 9688
    },
    {
      "epoch": 1.7453726638144562,
      "grad_norm": 1.5347487926483154,
      "learning_rate": 3.950936763179442e-06,
      "loss": 1.4358,
      "step": 9689
    },
    {
      "epoch": 1.7455528034226524,
      "grad_norm": 1.480939507484436,
      "learning_rate": 3.945421182989478e-06,
      "loss": 1.3459,
      "step": 9690
    },
    {
      "epoch": 1.7457329430308488,
      "grad_norm": 1.677643895149231,
      "learning_rate": 3.939909297275396e-06,
      "loss": 1.4973,
      "step": 9691
    },
    {
      "epoch": 1.7459130826390452,
      "grad_norm": 1.633876085281372,
      "learning_rate": 3.9344011064793516e-06,
      "loss": 1.4317,
      "step": 9692
    },
    {
      "epoch": 1.7460932222472416,
      "grad_norm": 1.576006293296814,
      "learning_rate": 3.92889661104322e-06,
      "loss": 1.634,
      "step": 9693
    },
    {
      "epoch": 1.746273361855438,
      "grad_norm": 1.6767064332962036,
      "learning_rate": 3.923395811408553e-06,
      "loss": 1.5312,
      "step": 9694
    },
    {
      "epoch": 1.7464535014636344,
      "grad_norm": 1.525294303894043,
      "learning_rate": 3.917898708016627e-06,
      "loss": 1.3689,
      "step": 9695
    },
    {
      "epoch": 1.7466336410718308,
      "grad_norm": 1.742080807685852,
      "learning_rate": 3.912405301308431e-06,
      "loss": 1.662,
      "step": 9696
    },
    {
      "epoch": 1.746813780680027,
      "grad_norm": 1.5430021286010742,
      "learning_rate": 3.90691559172463e-06,
      "loss": 1.057,
      "step": 9697
    },
    {
      "epoch": 1.7469939202882234,
      "grad_norm": 1.7851451635360718,
      "learning_rate": 3.901429579705612e-06,
      "loss": 1.5243,
      "step": 9698
    },
    {
      "epoch": 1.7471740598964196,
      "grad_norm": 1.400574803352356,
      "learning_rate": 3.895947265691458e-06,
      "loss": 1.177,
      "step": 9699
    },
    {
      "epoch": 1.747354199504616,
      "grad_norm": 1.47464919090271,
      "learning_rate": 3.890468650121959e-06,
      "loss": 1.1998,
      "step": 9700
    },
    {
      "epoch": 1.7475343391128124,
      "grad_norm": 1.5411518812179565,
      "learning_rate": 3.884993733436604e-06,
      "loss": 1.4171,
      "step": 9701
    },
    {
      "epoch": 1.7477144787210088,
      "grad_norm": 1.4549037218093872,
      "learning_rate": 3.87952251607459e-06,
      "loss": 1.9851,
      "step": 9702
    },
    {
      "epoch": 1.7478946183292052,
      "grad_norm": 1.2915027141571045,
      "learning_rate": 3.874054998474813e-06,
      "loss": 1.7468,
      "step": 9703
    },
    {
      "epoch": 1.7480747579374016,
      "grad_norm": 1.4986714124679565,
      "learning_rate": 3.86859118107587e-06,
      "loss": 2.1113,
      "step": 9704
    },
    {
      "epoch": 1.7482548975455978,
      "grad_norm": 1.4523217678070068,
      "learning_rate": 3.863131064316083e-06,
      "loss": 1.7455,
      "step": 9705
    },
    {
      "epoch": 1.7484350371537942,
      "grad_norm": 1.4350519180297852,
      "learning_rate": 3.857674648633453e-06,
      "loss": 1.8288,
      "step": 9706
    },
    {
      "epoch": 1.7486151767619904,
      "grad_norm": 1.4179859161376953,
      "learning_rate": 3.852221934465683e-06,
      "loss": 1.6406,
      "step": 9707
    },
    {
      "epoch": 1.7487953163701868,
      "grad_norm": 1.5210756063461304,
      "learning_rate": 3.8467729222501915e-06,
      "loss": 1.7797,
      "step": 9708
    },
    {
      "epoch": 1.7489754559783832,
      "grad_norm": 1.595882773399353,
      "learning_rate": 3.841327612424084e-06,
      "loss": 1.8956,
      "step": 9709
    },
    {
      "epoch": 1.7491555955865796,
      "grad_norm": 1.7114648818969727,
      "learning_rate": 3.835886005424216e-06,
      "loss": 1.8549,
      "step": 9710
    },
    {
      "epoch": 1.749335735194776,
      "grad_norm": 1.6545852422714233,
      "learning_rate": 3.830448101687073e-06,
      "loss": 2.0044,
      "step": 9711
    },
    {
      "epoch": 1.7495158748029724,
      "grad_norm": 1.608636498451233,
      "learning_rate": 3.825013901648899e-06,
      "loss": 1.6402,
      "step": 9712
    },
    {
      "epoch": 1.7496960144111686,
      "grad_norm": 1.3891401290893555,
      "learning_rate": 3.819583405745614e-06,
      "loss": 1.3771,
      "step": 9713
    },
    {
      "epoch": 1.749876154019365,
      "grad_norm": 1.3837445974349976,
      "learning_rate": 3.8141566144128504e-06,
      "loss": 1.5487,
      "step": 9714
    },
    {
      "epoch": 1.7500562936275612,
      "grad_norm": 1.4634850025177002,
      "learning_rate": 3.8087335280859558e-06,
      "loss": 1.4804,
      "step": 9715
    },
    {
      "epoch": 1.7502364332357576,
      "grad_norm": 1.568115234375,
      "learning_rate": 3.8033141471999547e-06,
      "loss": 1.4828,
      "step": 9716
    },
    {
      "epoch": 1.750416572843954,
      "grad_norm": 1.5026984214782715,
      "learning_rate": 3.7978984721895927e-06,
      "loss": 1.5035,
      "step": 9717
    },
    {
      "epoch": 1.7505967124521504,
      "grad_norm": 1.4704281091690063,
      "learning_rate": 3.7924865034893063e-06,
      "loss": 1.5866,
      "step": 9718
    },
    {
      "epoch": 1.7507768520603468,
      "grad_norm": 1.4158943891525269,
      "learning_rate": 3.7870782415332486e-06,
      "loss": 1.384,
      "step": 9719
    },
    {
      "epoch": 1.7509569916685432,
      "grad_norm": 1.3173227310180664,
      "learning_rate": 3.781673686755266e-06,
      "loss": 1.312,
      "step": 9720
    },
    {
      "epoch": 1.7511371312767394,
      "grad_norm": 1.3533389568328857,
      "learning_rate": 3.7762728395889124e-06,
      "loss": 1.2193,
      "step": 9721
    },
    {
      "epoch": 1.7513172708849358,
      "grad_norm": 1.4190833568572998,
      "learning_rate": 3.770875700467441e-06,
      "loss": 1.4416,
      "step": 9722
    },
    {
      "epoch": 1.751497410493132,
      "grad_norm": 1.5404233932495117,
      "learning_rate": 3.7654822698237892e-06,
      "loss": 1.5274,
      "step": 9723
    },
    {
      "epoch": 1.7516775501013284,
      "grad_norm": 1.3749133348464966,
      "learning_rate": 3.7600925480906324e-06,
      "loss": 1.3512,
      "step": 9724
    },
    {
      "epoch": 1.7518576897095248,
      "grad_norm": 1.4603270292282104,
      "learning_rate": 3.7547065357003365e-06,
      "loss": 1.36,
      "step": 9725
    },
    {
      "epoch": 1.7520378293177212,
      "grad_norm": 1.504769206047058,
      "learning_rate": 3.7493242330849552e-06,
      "loss": 1.3747,
      "step": 9726
    },
    {
      "epoch": 1.7522179689259176,
      "grad_norm": 1.4987685680389404,
      "learning_rate": 3.7439456406762497e-06,
      "loss": 1.4548,
      "step": 9727
    },
    {
      "epoch": 1.752398108534114,
      "grad_norm": 1.418268084526062,
      "learning_rate": 3.7385707589056905e-06,
      "loss": 1.2073,
      "step": 9728
    },
    {
      "epoch": 1.7525782481423104,
      "grad_norm": 1.525145411491394,
      "learning_rate": 3.7331995882044614e-06,
      "loss": 1.4641,
      "step": 9729
    },
    {
      "epoch": 1.7527583877505066,
      "grad_norm": 1.651598572731018,
      "learning_rate": 3.727832129003428e-06,
      "loss": 1.3245,
      "step": 9730
    },
    {
      "epoch": 1.752938527358703,
      "grad_norm": 1.4429457187652588,
      "learning_rate": 3.7224683817331573e-06,
      "loss": 1.4994,
      "step": 9731
    },
    {
      "epoch": 1.7531186669668992,
      "grad_norm": 1.4865858554840088,
      "learning_rate": 3.717108346823933e-06,
      "loss": 1.5121,
      "step": 9732
    },
    {
      "epoch": 1.7532988065750956,
      "grad_norm": 1.5316241979599,
      "learning_rate": 3.711752024705739e-06,
      "loss": 1.3634,
      "step": 9733
    },
    {
      "epoch": 1.753478946183292,
      "grad_norm": 1.3886276483535767,
      "learning_rate": 3.7063994158082425e-06,
      "loss": 1.3008,
      "step": 9734
    },
    {
      "epoch": 1.7536590857914884,
      "grad_norm": 1.5707337856292725,
      "learning_rate": 3.7010505205608448e-06,
      "loss": 1.3854,
      "step": 9735
    },
    {
      "epoch": 1.7538392253996848,
      "grad_norm": 1.5333585739135742,
      "learning_rate": 3.695705339392619e-06,
      "loss": 1.3357,
      "step": 9736
    },
    {
      "epoch": 1.7540193650078812,
      "grad_norm": 1.5735903978347778,
      "learning_rate": 3.6903638727323444e-06,
      "loss": 1.4418,
      "step": 9737
    },
    {
      "epoch": 1.7541995046160774,
      "grad_norm": 1.4835413694381714,
      "learning_rate": 3.685026121008539e-06,
      "loss": 1.3883,
      "step": 9738
    },
    {
      "epoch": 1.7543796442242738,
      "grad_norm": 1.624558925628662,
      "learning_rate": 3.679692084649372e-06,
      "loss": 1.5624,
      "step": 9739
    },
    {
      "epoch": 1.75455978383247,
      "grad_norm": 1.594003438949585,
      "learning_rate": 3.674361764082751e-06,
      "loss": 1.567,
      "step": 9740
    },
    {
      "epoch": 1.7547399234406664,
      "grad_norm": 1.5150173902511597,
      "learning_rate": 3.6690351597362615e-06,
      "loss": 1.3894,
      "step": 9741
    },
    {
      "epoch": 1.7549200630488628,
      "grad_norm": 1.5619381666183472,
      "learning_rate": 3.663712272037206e-06,
      "loss": 1.5707,
      "step": 9742
    },
    {
      "epoch": 1.7551002026570592,
      "grad_norm": 1.520041584968567,
      "learning_rate": 3.6583931014125773e-06,
      "loss": 1.303,
      "step": 9743
    },
    {
      "epoch": 1.7552803422652556,
      "grad_norm": 1.6436876058578491,
      "learning_rate": 3.653077648289088e-06,
      "loss": 1.4566,
      "step": 9744
    },
    {
      "epoch": 1.755460481873452,
      "grad_norm": 1.525728702545166,
      "learning_rate": 3.647765913093132e-06,
      "loss": 1.3305,
      "step": 9745
    },
    {
      "epoch": 1.7556406214816482,
      "grad_norm": 1.6755917072296143,
      "learning_rate": 3.6424578962508115e-06,
      "loss": 1.5057,
      "step": 9746
    },
    {
      "epoch": 1.7558207610898446,
      "grad_norm": 1.423240303993225,
      "learning_rate": 3.6371535981879423e-06,
      "loss": 1.05,
      "step": 9747
    },
    {
      "epoch": 1.7560009006980408,
      "grad_norm": 1.746389389038086,
      "learning_rate": 3.631853019330034e-06,
      "loss": 1.4926,
      "step": 9748
    },
    {
      "epoch": 1.7561810403062372,
      "grad_norm": 1.4243775606155396,
      "learning_rate": 3.6265561601022847e-06,
      "loss": 1.1602,
      "step": 9749
    },
    {
      "epoch": 1.7563611799144336,
      "grad_norm": 1.49173104763031,
      "learning_rate": 3.621263020929616e-06,
      "loss": 1.2521,
      "step": 9750
    },
    {
      "epoch": 1.75654131952263,
      "grad_norm": 1.4891126155853271,
      "learning_rate": 3.615973602236633e-06,
      "loss": 1.1817,
      "step": 9751
    },
    {
      "epoch": 1.7567214591308264,
      "grad_norm": 1.4392690658569336,
      "learning_rate": 3.610687904447657e-06,
      "loss": 1.6769,
      "step": 9752
    },
    {
      "epoch": 1.7569015987390229,
      "grad_norm": 1.20699942111969,
      "learning_rate": 3.6054059279867047e-06,
      "loss": 1.4982,
      "step": 9753
    },
    {
      "epoch": 1.7570817383472193,
      "grad_norm": 1.3411917686462402,
      "learning_rate": 3.6001276732774926e-06,
      "loss": 1.7788,
      "step": 9754
    },
    {
      "epoch": 1.7572618779554154,
      "grad_norm": 1.2381129264831543,
      "learning_rate": 3.594853140743443e-06,
      "loss": 1.5767,
      "step": 9755
    },
    {
      "epoch": 1.7574420175636118,
      "grad_norm": 1.5269241333007812,
      "learning_rate": 3.589582330807656e-06,
      "loss": 1.8997,
      "step": 9756
    },
    {
      "epoch": 1.757622157171808,
      "grad_norm": 1.4806115627288818,
      "learning_rate": 3.5843152438929773e-06,
      "loss": 1.7014,
      "step": 9757
    },
    {
      "epoch": 1.7578022967800044,
      "grad_norm": 1.5065656900405884,
      "learning_rate": 3.579051880421924e-06,
      "loss": 1.9056,
      "step": 9758
    },
    {
      "epoch": 1.7579824363882008,
      "grad_norm": 1.7371292114257812,
      "learning_rate": 3.5737922408167145e-06,
      "loss": 2.4525,
      "step": 9759
    },
    {
      "epoch": 1.7581625759963972,
      "grad_norm": 1.6146286725997925,
      "learning_rate": 3.568536325499272e-06,
      "loss": 1.8608,
      "step": 9760
    },
    {
      "epoch": 1.7583427156045937,
      "grad_norm": 1.8243255615234375,
      "learning_rate": 3.563284134891237e-06,
      "loss": 1.8993,
      "step": 9761
    },
    {
      "epoch": 1.75852285521279,
      "grad_norm": 1.4930227994918823,
      "learning_rate": 3.5580356694139282e-06,
      "loss": 1.5501,
      "step": 9762
    },
    {
      "epoch": 1.7587029948209862,
      "grad_norm": 1.514678716659546,
      "learning_rate": 3.5527909294883756e-06,
      "loss": 1.453,
      "step": 9763
    },
    {
      "epoch": 1.7588831344291826,
      "grad_norm": 1.4491084814071655,
      "learning_rate": 3.547549915535314e-06,
      "loss": 1.3982,
      "step": 9764
    },
    {
      "epoch": 1.7590632740373788,
      "grad_norm": 1.4310634136199951,
      "learning_rate": 3.542312627975175e-06,
      "loss": 1.4546,
      "step": 9765
    },
    {
      "epoch": 1.7592434136455752,
      "grad_norm": 1.4227977991104126,
      "learning_rate": 3.537079067228083e-06,
      "loss": 1.3456,
      "step": 9766
    },
    {
      "epoch": 1.7594235532537716,
      "grad_norm": 1.3339054584503174,
      "learning_rate": 3.5318492337138798e-06,
      "loss": 1.2823,
      "step": 9767
    },
    {
      "epoch": 1.759603692861968,
      "grad_norm": 1.4997001886367798,
      "learning_rate": 3.526623127852097e-06,
      "loss": 1.5565,
      "step": 9768
    },
    {
      "epoch": 1.7597838324701645,
      "grad_norm": 1.3836314678192139,
      "learning_rate": 3.5214007500619705e-06,
      "loss": 1.1732,
      "step": 9769
    },
    {
      "epoch": 1.7599639720783609,
      "grad_norm": 1.3238518238067627,
      "learning_rate": 3.5161821007624272e-06,
      "loss": 1.242,
      "step": 9770
    },
    {
      "epoch": 1.760144111686557,
      "grad_norm": 1.3837451934814453,
      "learning_rate": 3.510967180372127e-06,
      "loss": 1.3414,
      "step": 9771
    },
    {
      "epoch": 1.7603242512947535,
      "grad_norm": 1.37803053855896,
      "learning_rate": 3.5057559893093892e-06,
      "loss": 1.3634,
      "step": 9772
    },
    {
      "epoch": 1.7605043909029496,
      "grad_norm": 1.4191161394119263,
      "learning_rate": 3.5005485279922645e-06,
      "loss": 1.2114,
      "step": 9773
    },
    {
      "epoch": 1.760684530511146,
      "grad_norm": 1.3937416076660156,
      "learning_rate": 3.4953447968384734e-06,
      "loss": 1.299,
      "step": 9774
    },
    {
      "epoch": 1.7608646701193424,
      "grad_norm": 1.494080901145935,
      "learning_rate": 3.4901447962654933e-06,
      "loss": 1.5288,
      "step": 9775
    },
    {
      "epoch": 1.7610448097275389,
      "grad_norm": 1.2838844060897827,
      "learning_rate": 3.4849485266904347e-06,
      "loss": 1.0898,
      "step": 9776
    },
    {
      "epoch": 1.7612249493357353,
      "grad_norm": 1.4366638660430908,
      "learning_rate": 3.479755988530148e-06,
      "loss": 1.2903,
      "step": 9777
    },
    {
      "epoch": 1.7614050889439317,
      "grad_norm": 1.4605692625045776,
      "learning_rate": 3.4745671822011715e-06,
      "loss": 1.285,
      "step": 9778
    },
    {
      "epoch": 1.7615852285521278,
      "grad_norm": 1.4743674993515015,
      "learning_rate": 3.4693821081197506e-06,
      "loss": 1.3517,
      "step": 9779
    },
    {
      "epoch": 1.7617653681603243,
      "grad_norm": 1.3900632858276367,
      "learning_rate": 3.4642007667018416e-06,
      "loss": 1.3849,
      "step": 9780
    },
    {
      "epoch": 1.7619455077685207,
      "grad_norm": 1.491556167602539,
      "learning_rate": 3.459023158363084e-06,
      "loss": 1.3595,
      "step": 9781
    },
    {
      "epoch": 1.7621256473767168,
      "grad_norm": 1.4030952453613281,
      "learning_rate": 3.453849283518812e-06,
      "loss": 1.2637,
      "step": 9782
    },
    {
      "epoch": 1.7623057869849132,
      "grad_norm": 1.4194015264511108,
      "learning_rate": 3.4486791425840716e-06,
      "loss": 1.2555,
      "step": 9783
    },
    {
      "epoch": 1.7624859265931097,
      "grad_norm": 1.5971589088439941,
      "learning_rate": 3.4435127359736253e-06,
      "loss": 1.5054,
      "step": 9784
    },
    {
      "epoch": 1.762666066201306,
      "grad_norm": 1.571773648262024,
      "learning_rate": 3.438350064101914e-06,
      "loss": 1.4324,
      "step": 9785
    },
    {
      "epoch": 1.7628462058095025,
      "grad_norm": 1.3381184339523315,
      "learning_rate": 3.4331911273830784e-06,
      "loss": 1.1895,
      "step": 9786
    },
    {
      "epoch": 1.7630263454176989,
      "grad_norm": 1.3946419954299927,
      "learning_rate": 3.4280359262309767e-06,
      "loss": 1.1587,
      "step": 9787
    },
    {
      "epoch": 1.763206485025895,
      "grad_norm": 1.4441640377044678,
      "learning_rate": 3.4228844610591336e-06,
      "loss": 1.1807,
      "step": 9788
    },
    {
      "epoch": 1.7633866246340915,
      "grad_norm": 1.522282361984253,
      "learning_rate": 3.4177367322808183e-06,
      "loss": 1.4291,
      "step": 9789
    },
    {
      "epoch": 1.7635667642422876,
      "grad_norm": 1.5009187459945679,
      "learning_rate": 3.412592740308973e-06,
      "loss": 1.2038,
      "step": 9790
    },
    {
      "epoch": 1.763746903850484,
      "grad_norm": 1.5902810096740723,
      "learning_rate": 3.407452485556245e-06,
      "loss": 1.2673,
      "step": 9791
    },
    {
      "epoch": 1.7639270434586805,
      "grad_norm": 1.555992603302002,
      "learning_rate": 3.402315968434988e-06,
      "loss": 1.3651,
      "step": 9792
    },
    {
      "epoch": 1.7641071830668769,
      "grad_norm": 1.6070942878723145,
      "learning_rate": 3.3971831893572338e-06,
      "loss": 1.3466,
      "step": 9793
    },
    {
      "epoch": 1.7642873226750733,
      "grad_norm": 1.5451864004135132,
      "learning_rate": 3.392054148734758e-06,
      "loss": 1.3503,
      "step": 9794
    },
    {
      "epoch": 1.7644674622832697,
      "grad_norm": 1.509764313697815,
      "learning_rate": 3.386928846978993e-06,
      "loss": 1.3564,
      "step": 9795
    },
    {
      "epoch": 1.7646476018914659,
      "grad_norm": 1.5163006782531738,
      "learning_rate": 3.381807284501093e-06,
      "loss": 1.332,
      "step": 9796
    },
    {
      "epoch": 1.7648277414996623,
      "grad_norm": 1.5303181409835815,
      "learning_rate": 3.376689461711907e-06,
      "loss": 1.255,
      "step": 9797
    },
    {
      "epoch": 1.7650078811078584,
      "grad_norm": 1.6993863582611084,
      "learning_rate": 3.3715753790219794e-06,
      "loss": 1.601,
      "step": 9798
    },
    {
      "epoch": 1.7651880207160549,
      "grad_norm": 1.4489957094192505,
      "learning_rate": 3.366465036841565e-06,
      "loss": 1.1217,
      "step": 9799
    },
    {
      "epoch": 1.7653681603242513,
      "grad_norm": 1.4787137508392334,
      "learning_rate": 3.3613584355806138e-06,
      "loss": 1.1798,
      "step": 9800
    },
    {
      "epoch": 1.7655482999324477,
      "grad_norm": 1.288694977760315,
      "learning_rate": 3.3562555756487703e-06,
      "loss": 1.0289,
      "step": 9801
    },
    {
      "epoch": 1.765728439540644,
      "grad_norm": 1.3893877267837524,
      "learning_rate": 3.3511564574553735e-06,
      "loss": 1.8824,
      "step": 9802
    },
    {
      "epoch": 1.7659085791488405,
      "grad_norm": 1.3039926290512085,
      "learning_rate": 3.346061081409496e-06,
      "loss": 1.7071,
      "step": 9803
    },
    {
      "epoch": 1.7660887187570367,
      "grad_norm": 1.3711354732513428,
      "learning_rate": 3.340969447919873e-06,
      "loss": 1.9659,
      "step": 9804
    },
    {
      "epoch": 1.766268858365233,
      "grad_norm": 1.3592208623886108,
      "learning_rate": 3.335881557394954e-06,
      "loss": 1.764,
      "step": 9805
    },
    {
      "epoch": 1.7664489979734292,
      "grad_norm": 1.447879433631897,
      "learning_rate": 3.3307974102428863e-06,
      "loss": 1.7053,
      "step": 9806
    },
    {
      "epoch": 1.7666291375816257,
      "grad_norm": 1.5441514253616333,
      "learning_rate": 3.3257170068715205e-06,
      "loss": 1.8472,
      "step": 9807
    },
    {
      "epoch": 1.766809277189822,
      "grad_norm": 1.4021772146224976,
      "learning_rate": 3.320640347688397e-06,
      "loss": 1.6305,
      "step": 9808
    },
    {
      "epoch": 1.7669894167980185,
      "grad_norm": 1.5168946981430054,
      "learning_rate": 3.3155674331007745e-06,
      "loss": 1.8199,
      "step": 9809
    },
    {
      "epoch": 1.7671695564062149,
      "grad_norm": 1.6763941049575806,
      "learning_rate": 3.3104982635155932e-06,
      "loss": 2.0727,
      "step": 9810
    },
    {
      "epoch": 1.7673496960144113,
      "grad_norm": 1.576102614402771,
      "learning_rate": 3.3054328393394895e-06,
      "loss": 1.8597,
      "step": 9811
    },
    {
      "epoch": 1.7675298356226077,
      "grad_norm": 1.7728216648101807,
      "learning_rate": 3.300371160978827e-06,
      "loss": 1.6148,
      "step": 9812
    },
    {
      "epoch": 1.7677099752308039,
      "grad_norm": 1.4950711727142334,
      "learning_rate": 3.2953132288396415e-06,
      "loss": 1.6179,
      "step": 9813
    },
    {
      "epoch": 1.7678901148390003,
      "grad_norm": 1.4749361276626587,
      "learning_rate": 3.2902590433276813e-06,
      "loss": 1.5794,
      "step": 9814
    },
    {
      "epoch": 1.7680702544471965,
      "grad_norm": 1.4258922338485718,
      "learning_rate": 3.285208604848389e-06,
      "loss": 1.3797,
      "step": 9815
    },
    {
      "epoch": 1.7682503940553929,
      "grad_norm": 1.352752685546875,
      "learning_rate": 3.2801619138069007e-06,
      "loss": 1.2688,
      "step": 9816
    },
    {
      "epoch": 1.7684305336635893,
      "grad_norm": 1.4454792737960815,
      "learning_rate": 3.2751189706080767e-06,
      "loss": 1.4482,
      "step": 9817
    },
    {
      "epoch": 1.7686106732717857,
      "grad_norm": 1.5224086046218872,
      "learning_rate": 3.2700797756564483e-06,
      "loss": 1.3349,
      "step": 9818
    },
    {
      "epoch": 1.768790812879982,
      "grad_norm": 1.6277973651885986,
      "learning_rate": 3.2650443293562704e-06,
      "loss": 1.8212,
      "step": 9819
    },
    {
      "epoch": 1.7689709524881785,
      "grad_norm": 1.555992603302002,
      "learning_rate": 3.2600126321114587e-06,
      "loss": 1.4072,
      "step": 9820
    },
    {
      "epoch": 1.7691510920963747,
      "grad_norm": 1.6057409048080444,
      "learning_rate": 3.2549846843256627e-06,
      "loss": 1.5905,
      "step": 9821
    },
    {
      "epoch": 1.769331231704571,
      "grad_norm": 1.6142410039901733,
      "learning_rate": 3.249960486402237e-06,
      "loss": 1.3907,
      "step": 9822
    },
    {
      "epoch": 1.7695113713127673,
      "grad_norm": 1.4140647649765015,
      "learning_rate": 3.2449400387442042e-06,
      "loss": 1.2653,
      "step": 9823
    },
    {
      "epoch": 1.7696915109209637,
      "grad_norm": 1.6014339923858643,
      "learning_rate": 3.239923341754314e-06,
      "loss": 1.4371,
      "step": 9824
    },
    {
      "epoch": 1.76987165052916,
      "grad_norm": 1.5181461572647095,
      "learning_rate": 3.2349103958349837e-06,
      "loss": 1.3349,
      "step": 9825
    },
    {
      "epoch": 1.7700517901373565,
      "grad_norm": 1.475053310394287,
      "learning_rate": 3.2299012013883746e-06,
      "loss": 1.3329,
      "step": 9826
    },
    {
      "epoch": 1.7702319297455529,
      "grad_norm": 1.5775713920593262,
      "learning_rate": 3.2248957588163097e-06,
      "loss": 1.4734,
      "step": 9827
    },
    {
      "epoch": 1.7704120693537493,
      "grad_norm": 1.482925295829773,
      "learning_rate": 3.219894068520324e-06,
      "loss": 1.1697,
      "step": 9828
    },
    {
      "epoch": 1.7705922089619455,
      "grad_norm": 1.5949487686157227,
      "learning_rate": 3.2148961309016514e-06,
      "loss": 1.4629,
      "step": 9829
    },
    {
      "epoch": 1.7707723485701419,
      "grad_norm": 1.5618815422058105,
      "learning_rate": 3.209901946361221e-06,
      "loss": 1.4689,
      "step": 9830
    },
    {
      "epoch": 1.770952488178338,
      "grad_norm": 1.6027182340621948,
      "learning_rate": 3.2049115152996633e-06,
      "loss": 1.5968,
      "step": 9831
    },
    {
      "epoch": 1.7711326277865345,
      "grad_norm": 1.5154054164886475,
      "learning_rate": 3.199924838117313e-06,
      "loss": 1.6709,
      "step": 9832
    },
    {
      "epoch": 1.7713127673947309,
      "grad_norm": 1.4988441467285156,
      "learning_rate": 3.1949419152142e-06,
      "loss": 1.4155,
      "step": 9833
    },
    {
      "epoch": 1.7714929070029273,
      "grad_norm": 1.5211819410324097,
      "learning_rate": 3.189962746990044e-06,
      "loss": 1.3442,
      "step": 9834
    },
    {
      "epoch": 1.7716730466111237,
      "grad_norm": 1.5229166746139526,
      "learning_rate": 3.184987333844269e-06,
      "loss": 1.3498,
      "step": 9835
    },
    {
      "epoch": 1.77185318621932,
      "grad_norm": 1.5959737300872803,
      "learning_rate": 3.180015676176018e-06,
      "loss": 1.3395,
      "step": 9836
    },
    {
      "epoch": 1.7720333258275163,
      "grad_norm": 1.5616458654403687,
      "learning_rate": 3.1750477743841044e-06,
      "loss": 1.2583,
      "step": 9837
    },
    {
      "epoch": 1.7722134654357127,
      "grad_norm": 1.6109622716903687,
      "learning_rate": 3.1700836288670486e-06,
      "loss": 1.403,
      "step": 9838
    },
    {
      "epoch": 1.772393605043909,
      "grad_norm": 1.4488798379898071,
      "learning_rate": 3.1651232400230656e-06,
      "loss": 1.0806,
      "step": 9839
    },
    {
      "epoch": 1.7725737446521053,
      "grad_norm": 1.4988574981689453,
      "learning_rate": 3.160166608250098e-06,
      "loss": 1.2333,
      "step": 9840
    },
    {
      "epoch": 1.7727538842603017,
      "grad_norm": 1.6694145202636719,
      "learning_rate": 3.1552137339457445e-06,
      "loss": 1.2971,
      "step": 9841
    },
    {
      "epoch": 1.772934023868498,
      "grad_norm": 1.688083291053772,
      "learning_rate": 3.1502646175073312e-06,
      "loss": 1.5732,
      "step": 9842
    },
    {
      "epoch": 1.7731141634766945,
      "grad_norm": 1.4416602849960327,
      "learning_rate": 3.145319259331869e-06,
      "loss": 1.3582,
      "step": 9843
    },
    {
      "epoch": 1.7732943030848909,
      "grad_norm": 1.5128344297409058,
      "learning_rate": 3.140377659816063e-06,
      "loss": 1.2716,
      "step": 9844
    },
    {
      "epoch": 1.7734744426930873,
      "grad_norm": 1.4940046072006226,
      "learning_rate": 3.135439819356345e-06,
      "loss": 1.3417,
      "step": 9845
    },
    {
      "epoch": 1.7736545823012835,
      "grad_norm": 1.8050631284713745,
      "learning_rate": 3.130505738348821e-06,
      "loss": 1.418,
      "step": 9846
    },
    {
      "epoch": 1.7738347219094799,
      "grad_norm": 1.4687649011611938,
      "learning_rate": 3.1255754171892914e-06,
      "loss": 1.1303,
      "step": 9847
    },
    {
      "epoch": 1.774014861517676,
      "grad_norm": 1.7391098737716675,
      "learning_rate": 3.1206488562732783e-06,
      "loss": 1.3444,
      "step": 9848
    },
    {
      "epoch": 1.7741950011258725,
      "grad_norm": 1.6519453525543213,
      "learning_rate": 3.1157260559959657e-06,
      "loss": 1.311,
      "step": 9849
    },
    {
      "epoch": 1.7743751407340689,
      "grad_norm": 1.5089085102081299,
      "learning_rate": 3.1108070167522817e-06,
      "loss": 1.1043,
      "step": 9850
    },
    {
      "epoch": 1.7745552803422653,
      "grad_norm": 1.464198112487793,
      "learning_rate": 3.1058917389368223e-06,
      "loss": 1.1819,
      "step": 9851
    },
    {
      "epoch": 1.7747354199504617,
      "grad_norm": 1.376810908317566,
      "learning_rate": 3.1009802229438887e-06,
      "loss": 1.634,
      "step": 9852
    },
    {
      "epoch": 1.774915559558658,
      "grad_norm": 1.411827564239502,
      "learning_rate": 3.096072469167466e-06,
      "loss": 1.7541,
      "step": 9853
    },
    {
      "epoch": 1.7750956991668543,
      "grad_norm": 1.490149736404419,
      "learning_rate": 3.0911684780012718e-06,
      "loss": 1.5701,
      "step": 9854
    },
    {
      "epoch": 1.7752758387750507,
      "grad_norm": 1.3167155981063843,
      "learning_rate": 3.0862682498386917e-06,
      "loss": 1.4447,
      "step": 9855
    },
    {
      "epoch": 1.7754559783832469,
      "grad_norm": 1.3508093357086182,
      "learning_rate": 3.081371785072823e-06,
      "loss": 1.5733,
      "step": 9856
    },
    {
      "epoch": 1.7756361179914433,
      "grad_norm": 1.58484947681427,
      "learning_rate": 3.0764790840964563e-06,
      "loss": 1.9759,
      "step": 9857
    },
    {
      "epoch": 1.7758162575996397,
      "grad_norm": 1.5745463371276855,
      "learning_rate": 3.071590147302078e-06,
      "loss": 1.9436,
      "step": 9858
    },
    {
      "epoch": 1.775996397207836,
      "grad_norm": 1.5669615268707275,
      "learning_rate": 3.0667049750818854e-06,
      "loss": 1.3913,
      "step": 9859
    },
    {
      "epoch": 1.7761765368160325,
      "grad_norm": 1.794481635093689,
      "learning_rate": 3.061823567827754e-06,
      "loss": 1.7634,
      "step": 9860
    },
    {
      "epoch": 1.776356676424229,
      "grad_norm": 1.576206922531128,
      "learning_rate": 3.0569459259312818e-06,
      "loss": 1.4121,
      "step": 9861
    },
    {
      "epoch": 1.776536816032425,
      "grad_norm": 1.522792935371399,
      "learning_rate": 3.0520720497837385e-06,
      "loss": 1.4583,
      "step": 9862
    },
    {
      "epoch": 1.7767169556406215,
      "grad_norm": 1.2773865461349487,
      "learning_rate": 3.0472019397761064e-06,
      "loss": 1.2119,
      "step": 9863
    },
    {
      "epoch": 1.7768970952488177,
      "grad_norm": 1.3955366611480713,
      "learning_rate": 3.042335596299062e-06,
      "loss": 1.3323,
      "step": 9864
    },
    {
      "epoch": 1.777077234857014,
      "grad_norm": 1.4555565118789673,
      "learning_rate": 3.037473019742987e-06,
      "loss": 1.4389,
      "step": 9865
    },
    {
      "epoch": 1.7772573744652105,
      "grad_norm": 1.4232789278030396,
      "learning_rate": 3.0326142104979525e-06,
      "loss": 1.445,
      "step": 9866
    },
    {
      "epoch": 1.7774375140734069,
      "grad_norm": 1.341517686843872,
      "learning_rate": 3.0277591689537254e-06,
      "loss": 1.4613,
      "step": 9867
    },
    {
      "epoch": 1.7776176536816033,
      "grad_norm": 1.4803911447525024,
      "learning_rate": 3.0229078954997823e-06,
      "loss": 1.3171,
      "step": 9868
    },
    {
      "epoch": 1.7777977932897997,
      "grad_norm": 1.4366044998168945,
      "learning_rate": 3.0180603905252846e-06,
      "loss": 1.3199,
      "step": 9869
    },
    {
      "epoch": 1.777977932897996,
      "grad_norm": 1.4099029302597046,
      "learning_rate": 3.0132166544190986e-06,
      "loss": 1.3858,
      "step": 9870
    },
    {
      "epoch": 1.7781580725061923,
      "grad_norm": 1.508741021156311,
      "learning_rate": 3.008376687569786e-06,
      "loss": 1.4026,
      "step": 9871
    },
    {
      "epoch": 1.7783382121143887,
      "grad_norm": 1.6278587579727173,
      "learning_rate": 3.0035404903656085e-06,
      "loss": 1.4795,
      "step": 9872
    },
    {
      "epoch": 1.7785183517225849,
      "grad_norm": 1.4682656526565552,
      "learning_rate": 2.998708063194522e-06,
      "loss": 1.178,
      "step": 9873
    },
    {
      "epoch": 1.7786984913307813,
      "grad_norm": 1.4815837144851685,
      "learning_rate": 2.9938794064441776e-06,
      "loss": 1.3455,
      "step": 9874
    },
    {
      "epoch": 1.7788786309389777,
      "grad_norm": 1.4454247951507568,
      "learning_rate": 2.9890545205019328e-06,
      "loss": 1.3776,
      "step": 9875
    },
    {
      "epoch": 1.779058770547174,
      "grad_norm": 1.4900965690612793,
      "learning_rate": 2.984233405754838e-06,
      "loss": 1.3231,
      "step": 9876
    },
    {
      "epoch": 1.7792389101553705,
      "grad_norm": 1.458905816078186,
      "learning_rate": 2.9794160625896296e-06,
      "loss": 1.21,
      "step": 9877
    },
    {
      "epoch": 1.779419049763567,
      "grad_norm": 1.3248931169509888,
      "learning_rate": 2.9746024913927693e-06,
      "loss": 1.2997,
      "step": 9878
    },
    {
      "epoch": 1.779599189371763,
      "grad_norm": 1.495949387550354,
      "learning_rate": 2.9697926925503936e-06,
      "loss": 1.4589,
      "step": 9879
    },
    {
      "epoch": 1.7797793289799595,
      "grad_norm": 1.5799137353897095,
      "learning_rate": 2.9649866664483385e-06,
      "loss": 1.5057,
      "step": 9880
    },
    {
      "epoch": 1.7799594685881557,
      "grad_norm": 1.5541681051254272,
      "learning_rate": 2.960184413472139e-06,
      "loss": 1.3934,
      "step": 9881
    },
    {
      "epoch": 1.780139608196352,
      "grad_norm": 1.615139126777649,
      "learning_rate": 2.9553859340070377e-06,
      "loss": 1.5931,
      "step": 9882
    },
    {
      "epoch": 1.7803197478045485,
      "grad_norm": 1.4375929832458496,
      "learning_rate": 2.9505912284379598e-06,
      "loss": 1.389,
      "step": 9883
    },
    {
      "epoch": 1.780499887412745,
      "grad_norm": 1.4612473249435425,
      "learning_rate": 2.945800297149548e-06,
      "loss": 1.4184,
      "step": 9884
    },
    {
      "epoch": 1.7806800270209413,
      "grad_norm": 1.5776232481002808,
      "learning_rate": 2.941013140526111e-06,
      "loss": 1.3939,
      "step": 9885
    },
    {
      "epoch": 1.7808601666291377,
      "grad_norm": 1.5297393798828125,
      "learning_rate": 2.9362297589516696e-06,
      "loss": 1.2477,
      "step": 9886
    },
    {
      "epoch": 1.781040306237334,
      "grad_norm": 1.4318976402282715,
      "learning_rate": 2.9314501528099615e-06,
      "loss": 1.0685,
      "step": 9887
    },
    {
      "epoch": 1.7812204458455303,
      "grad_norm": 1.5197341442108154,
      "learning_rate": 2.9266743224843907e-06,
      "loss": 1.383,
      "step": 9888
    },
    {
      "epoch": 1.7814005854537265,
      "grad_norm": 1.6641758680343628,
      "learning_rate": 2.9219022683580786e-06,
      "loss": 1.4712,
      "step": 9889
    },
    {
      "epoch": 1.7815807250619229,
      "grad_norm": 1.6749756336212158,
      "learning_rate": 2.9171339908138308e-06,
      "loss": 1.7038,
      "step": 9890
    },
    {
      "epoch": 1.7817608646701193,
      "grad_norm": 1.6026268005371094,
      "learning_rate": 2.912369490234168e-06,
      "loss": 1.5003,
      "step": 9891
    },
    {
      "epoch": 1.7819410042783157,
      "grad_norm": 1.505325436592102,
      "learning_rate": 2.907608767001291e-06,
      "loss": 1.2014,
      "step": 9892
    },
    {
      "epoch": 1.782121143886512,
      "grad_norm": 1.4283524751663208,
      "learning_rate": 2.902851821497099e-06,
      "loss": 1.12,
      "step": 9893
    },
    {
      "epoch": 1.7823012834947085,
      "grad_norm": 1.6709409952163696,
      "learning_rate": 2.898098654103193e-06,
      "loss": 1.5111,
      "step": 9894
    },
    {
      "epoch": 1.782481423102905,
      "grad_norm": 1.493607997894287,
      "learning_rate": 2.8933492652008677e-06,
      "loss": 1.3331,
      "step": 9895
    },
    {
      "epoch": 1.782661562711101,
      "grad_norm": 1.562774658203125,
      "learning_rate": 2.888603655171124e-06,
      "loss": 1.3574,
      "step": 9896
    },
    {
      "epoch": 1.7828417023192975,
      "grad_norm": 1.5309172868728638,
      "learning_rate": 2.883861824394646e-06,
      "loss": 1.1357,
      "step": 9897
    },
    {
      "epoch": 1.7830218419274937,
      "grad_norm": 1.5188308954238892,
      "learning_rate": 2.879123773251824e-06,
      "loss": 1.2806,
      "step": 9898
    },
    {
      "epoch": 1.78320198153569,
      "grad_norm": 1.5923372507095337,
      "learning_rate": 2.874389502122743e-06,
      "loss": 1.2997,
      "step": 9899
    },
    {
      "epoch": 1.7833821211438865,
      "grad_norm": 1.5409752130508423,
      "learning_rate": 2.8696590113871714e-06,
      "loss": 1.2403,
      "step": 9900
    },
    {
      "epoch": 1.783562260752083,
      "grad_norm": 1.587744116783142,
      "learning_rate": 2.8649323014246054e-06,
      "loss": 1.1927,
      "step": 9901
    },
    {
      "epoch": 1.7837424003602793,
      "grad_norm": 1.4742180109024048,
      "learning_rate": 2.860209372614209e-06,
      "loss": 1.7464,
      "step": 9902
    },
    {
      "epoch": 1.7839225399684757,
      "grad_norm": 1.4704670906066895,
      "learning_rate": 2.8554902253348614e-06,
      "loss": 2.0764,
      "step": 9903
    },
    {
      "epoch": 1.784102679576672,
      "grad_norm": 1.3590621948242188,
      "learning_rate": 2.850774859965122e-06,
      "loss": 1.7817,
      "step": 9904
    },
    {
      "epoch": 1.7842828191848683,
      "grad_norm": 1.3859045505523682,
      "learning_rate": 2.8460632768832596e-06,
      "loss": 1.7231,
      "step": 9905
    },
    {
      "epoch": 1.7844629587930645,
      "grad_norm": 1.3658157587051392,
      "learning_rate": 2.841355476467228e-06,
      "loss": 1.6551,
      "step": 9906
    },
    {
      "epoch": 1.784643098401261,
      "grad_norm": 1.5751011371612549,
      "learning_rate": 2.8366514590946967e-06,
      "loss": 1.7875,
      "step": 9907
    },
    {
      "epoch": 1.7848232380094573,
      "grad_norm": 1.6245322227478027,
      "learning_rate": 2.831951225143009e-06,
      "loss": 1.8343,
      "step": 9908
    },
    {
      "epoch": 1.7850033776176537,
      "grad_norm": 1.5365079641342163,
      "learning_rate": 2.827254774989213e-06,
      "loss": 1.5304,
      "step": 9909
    },
    {
      "epoch": 1.7851835172258501,
      "grad_norm": 1.5759741067886353,
      "learning_rate": 2.8225621090100686e-06,
      "loss": 1.7605,
      "step": 9910
    },
    {
      "epoch": 1.7853636568340465,
      "grad_norm": 2.3076226711273193,
      "learning_rate": 2.817873227582013e-06,
      "loss": 2.274,
      "step": 9911
    },
    {
      "epoch": 1.7855437964422427,
      "grad_norm": 1.6789138317108154,
      "learning_rate": 2.8131881310811846e-06,
      "loss": 1.6764,
      "step": 9912
    },
    {
      "epoch": 1.785723936050439,
      "grad_norm": 1.4506316184997559,
      "learning_rate": 2.808506819883422e-06,
      "loss": 1.4689,
      "step": 9913
    },
    {
      "epoch": 1.7859040756586353,
      "grad_norm": 1.386644721031189,
      "learning_rate": 2.8038292943642465e-06,
      "loss": 1.3536,
      "step": 9914
    },
    {
      "epoch": 1.7860842152668317,
      "grad_norm": 1.3460304737091064,
      "learning_rate": 2.7991555548989024e-06,
      "loss": 1.3028,
      "step": 9915
    },
    {
      "epoch": 1.786264354875028,
      "grad_norm": 1.4798442125320435,
      "learning_rate": 2.7944856018623124e-06,
      "loss": 1.4623,
      "step": 9916
    },
    {
      "epoch": 1.7864444944832245,
      "grad_norm": 1.450947642326355,
      "learning_rate": 2.789819435629104e-06,
      "loss": 1.4857,
      "step": 9917
    },
    {
      "epoch": 1.786624634091421,
      "grad_norm": 1.4580761194229126,
      "learning_rate": 2.785157056573567e-06,
      "loss": 1.5624,
      "step": 9918
    },
    {
      "epoch": 1.7868047736996173,
      "grad_norm": 1.5128990411758423,
      "learning_rate": 2.780498465069742e-06,
      "loss": 1.3527,
      "step": 9919
    },
    {
      "epoch": 1.7869849133078135,
      "grad_norm": 1.3524466753005981,
      "learning_rate": 2.7758436614913287e-06,
      "loss": 1.1834,
      "step": 9920
    },
    {
      "epoch": 1.78716505291601,
      "grad_norm": 1.5410993099212646,
      "learning_rate": 2.7711926462117346e-06,
      "loss": 1.4377,
      "step": 9921
    },
    {
      "epoch": 1.787345192524206,
      "grad_norm": 1.4928367137908936,
      "learning_rate": 2.7665454196040664e-06,
      "loss": 1.4857,
      "step": 9922
    },
    {
      "epoch": 1.7875253321324025,
      "grad_norm": 1.6126383543014526,
      "learning_rate": 2.761901982041104e-06,
      "loss": 1.59,
      "step": 9923
    },
    {
      "epoch": 1.787705471740599,
      "grad_norm": 1.371206521987915,
      "learning_rate": 2.7572623338953662e-06,
      "loss": 1.1243,
      "step": 9924
    },
    {
      "epoch": 1.7878856113487953,
      "grad_norm": 1.5020833015441895,
      "learning_rate": 2.752626475539033e-06,
      "loss": 1.573,
      "step": 9925
    },
    {
      "epoch": 1.7880657509569917,
      "grad_norm": 1.4009459018707275,
      "learning_rate": 2.7479944073439957e-06,
      "loss": 1.3558,
      "step": 9926
    },
    {
      "epoch": 1.7882458905651881,
      "grad_norm": 1.4487234354019165,
      "learning_rate": 2.7433661296818235e-06,
      "loss": 1.3079,
      "step": 9927
    },
    {
      "epoch": 1.7884260301733845,
      "grad_norm": 1.7979350090026855,
      "learning_rate": 2.7387416429238086e-06,
      "loss": 1.686,
      "step": 9928
    },
    {
      "epoch": 1.7886061697815807,
      "grad_norm": 1.4197977781295776,
      "learning_rate": 2.734120947440916e-06,
      "loss": 1.283,
      "step": 9929
    },
    {
      "epoch": 1.7887863093897771,
      "grad_norm": 1.463405728340149,
      "learning_rate": 2.7295040436038145e-06,
      "loss": 1.2608,
      "step": 9930
    },
    {
      "epoch": 1.7889664489979733,
      "grad_norm": 1.5794692039489746,
      "learning_rate": 2.724890931782881e-06,
      "loss": 1.4955,
      "step": 9931
    },
    {
      "epoch": 1.7891465886061697,
      "grad_norm": 1.5418800115585327,
      "learning_rate": 2.720281612348158e-06,
      "loss": 1.4654,
      "step": 9932
    },
    {
      "epoch": 1.7893267282143661,
      "grad_norm": 1.467890739440918,
      "learning_rate": 2.7156760856694276e-06,
      "loss": 1.3422,
      "step": 9933
    },
    {
      "epoch": 1.7895068678225625,
      "grad_norm": 1.5787708759307861,
      "learning_rate": 2.7110743521161277e-06,
      "loss": 1.4593,
      "step": 9934
    },
    {
      "epoch": 1.789687007430759,
      "grad_norm": 1.5096412897109985,
      "learning_rate": 2.706476412057407e-06,
      "loss": 1.3807,
      "step": 9935
    },
    {
      "epoch": 1.7898671470389553,
      "grad_norm": 1.6212842464447021,
      "learning_rate": 2.7018822658621155e-06,
      "loss": 1.4962,
      "step": 9936
    },
    {
      "epoch": 1.7900472866471515,
      "grad_norm": 1.4657515287399292,
      "learning_rate": 2.697291913898786e-06,
      "loss": 1.2455,
      "step": 9937
    },
    {
      "epoch": 1.790227426255348,
      "grad_norm": 1.503367304801941,
      "learning_rate": 2.6927053565356632e-06,
      "loss": 1.3998,
      "step": 9938
    },
    {
      "epoch": 1.790407565863544,
      "grad_norm": 1.681787371635437,
      "learning_rate": 2.688122594140674e-06,
      "loss": 1.4904,
      "step": 9939
    },
    {
      "epoch": 1.7905877054717405,
      "grad_norm": 1.5369207859039307,
      "learning_rate": 2.6835436270814475e-06,
      "loss": 1.3873,
      "step": 9940
    },
    {
      "epoch": 1.790767845079937,
      "grad_norm": 1.6265339851379395,
      "learning_rate": 2.6789684557253005e-06,
      "loss": 1.3087,
      "step": 9941
    },
    {
      "epoch": 1.7909479846881333,
      "grad_norm": 1.5250691175460815,
      "learning_rate": 2.6743970804392448e-06,
      "loss": 1.3881,
      "step": 9942
    },
    {
      "epoch": 1.7911281242963297,
      "grad_norm": 1.463868260383606,
      "learning_rate": 2.6698295015900156e-06,
      "loss": 1.2515,
      "step": 9943
    },
    {
      "epoch": 1.7913082639045261,
      "grad_norm": 1.5700215101242065,
      "learning_rate": 2.6652657195440132e-06,
      "loss": 1.3948,
      "step": 9944
    },
    {
      "epoch": 1.7914884035127223,
      "grad_norm": 1.5048481225967407,
      "learning_rate": 2.660705734667335e-06,
      "loss": 1.3173,
      "step": 9945
    },
    {
      "epoch": 1.7916685431209187,
      "grad_norm": 1.688209891319275,
      "learning_rate": 2.6561495473257758e-06,
      "loss": 1.6279,
      "step": 9946
    },
    {
      "epoch": 1.791848682729115,
      "grad_norm": 1.4761959314346313,
      "learning_rate": 2.65159715788485e-06,
      "loss": 1.1893,
      "step": 9947
    },
    {
      "epoch": 1.7920288223373113,
      "grad_norm": 1.6242260932922363,
      "learning_rate": 2.647048566709731e-06,
      "loss": 1.348,
      "step": 9948
    },
    {
      "epoch": 1.7922089619455077,
      "grad_norm": 1.7724230289459229,
      "learning_rate": 2.6425037741653227e-06,
      "loss": 1.6359,
      "step": 9949
    },
    {
      "epoch": 1.7923891015537041,
      "grad_norm": 1.585125207901001,
      "learning_rate": 2.6379627806161877e-06,
      "loss": 1.2213,
      "step": 9950
    },
    {
      "epoch": 1.7925692411619005,
      "grad_norm": 1.6363455057144165,
      "learning_rate": 2.6334255864266076e-06,
      "loss": 1.379,
      "step": 9951
    },
    {
      "epoch": 1.792749380770097,
      "grad_norm": 1.349861979484558,
      "learning_rate": 2.6288921919605524e-06,
      "loss": 1.8176,
      "step": 9952
    },
    {
      "epoch": 1.7929295203782933,
      "grad_norm": 1.3355356454849243,
      "learning_rate": 2.6243625975816977e-06,
      "loss": 1.7971,
      "step": 9953
    },
    {
      "epoch": 1.7931096599864895,
      "grad_norm": 1.287998080253601,
      "learning_rate": 2.6198368036534026e-06,
      "loss": 1.4331,
      "step": 9954
    },
    {
      "epoch": 1.793289799594686,
      "grad_norm": 1.4241411685943604,
      "learning_rate": 2.615314810538716e-06,
      "loss": 1.755,
      "step": 9955
    },
    {
      "epoch": 1.7934699392028821,
      "grad_norm": 1.3556983470916748,
      "learning_rate": 2.610796618600386e-06,
      "loss": 1.5629,
      "step": 9956
    },
    {
      "epoch": 1.7936500788110785,
      "grad_norm": 1.5641840696334839,
      "learning_rate": 2.606282228200879e-06,
      "loss": 1.8697,
      "step": 9957
    },
    {
      "epoch": 1.793830218419275,
      "grad_norm": 1.5936095714569092,
      "learning_rate": 2.6017716397023275e-06,
      "loss": 1.7644,
      "step": 9958
    },
    {
      "epoch": 1.7940103580274713,
      "grad_norm": 1.7188745737075806,
      "learning_rate": 2.597264853466569e-06,
      "loss": 1.717,
      "step": 9959
    },
    {
      "epoch": 1.7941904976356677,
      "grad_norm": 1.7600808143615723,
      "learning_rate": 2.592761869855137e-06,
      "loss": 1.952,
      "step": 9960
    },
    {
      "epoch": 1.7943706372438641,
      "grad_norm": 1.7443944215774536,
      "learning_rate": 2.588262689229254e-06,
      "loss": 1.8312,
      "step": 9961
    },
    {
      "epoch": 1.7945507768520603,
      "grad_norm": 1.5954912900924683,
      "learning_rate": 2.583767311949842e-06,
      "loss": 1.4167,
      "step": 9962
    },
    {
      "epoch": 1.7947309164602567,
      "grad_norm": 1.4277865886688232,
      "learning_rate": 2.5792757383775233e-06,
      "loss": 1.4405,
      "step": 9963
    },
    {
      "epoch": 1.794911056068453,
      "grad_norm": 1.5450223684310913,
      "learning_rate": 2.574787968872605e-06,
      "loss": 1.5595,
      "step": 9964
    },
    {
      "epoch": 1.7950911956766493,
      "grad_norm": 1.4645463228225708,
      "learning_rate": 2.570304003795088e-06,
      "loss": 1.3917,
      "step": 9965
    },
    {
      "epoch": 1.7952713352848457,
      "grad_norm": 1.2948360443115234,
      "learning_rate": 2.5658238435046943e-06,
      "loss": 1.3636,
      "step": 9966
    },
    {
      "epoch": 1.7954514748930421,
      "grad_norm": 1.3741048574447632,
      "learning_rate": 2.5613474883608045e-06,
      "loss": 1.5765,
      "step": 9967
    },
    {
      "epoch": 1.7956316145012385,
      "grad_norm": 1.3922216892242432,
      "learning_rate": 2.556874938722509e-06,
      "loss": 1.4731,
      "step": 9968
    },
    {
      "epoch": 1.795811754109435,
      "grad_norm": 1.344000220298767,
      "learning_rate": 2.5524061949486033e-06,
      "loss": 1.3584,
      "step": 9969
    },
    {
      "epoch": 1.7959918937176311,
      "grad_norm": 1.2065109014511108,
      "learning_rate": 2.547941257397557e-06,
      "loss": 1.097,
      "step": 9970
    },
    {
      "epoch": 1.7961720333258275,
      "grad_norm": 1.4392669200897217,
      "learning_rate": 2.54348012642755e-06,
      "loss": 1.378,
      "step": 9971
    },
    {
      "epoch": 1.7963521729340237,
      "grad_norm": 1.4213449954986572,
      "learning_rate": 2.539022802396457e-06,
      "loss": 1.3661,
      "step": 9972
    },
    {
      "epoch": 1.7965323125422201,
      "grad_norm": 1.3919795751571655,
      "learning_rate": 2.534569285661831e-06,
      "loss": 1.2925,
      "step": 9973
    },
    {
      "epoch": 1.7967124521504165,
      "grad_norm": 1.3523578643798828,
      "learning_rate": 2.530119576580936e-06,
      "loss": 1.264,
      "step": 9974
    },
    {
      "epoch": 1.796892591758613,
      "grad_norm": 1.437835931777954,
      "learning_rate": 2.5256736755107314e-06,
      "loss": 1.3298,
      "step": 9975
    },
    {
      "epoch": 1.7970727313668093,
      "grad_norm": 1.4455056190490723,
      "learning_rate": 2.521231582807859e-06,
      "loss": 1.2947,
      "step": 9976
    },
    {
      "epoch": 1.7972528709750057,
      "grad_norm": 1.5071290731430054,
      "learning_rate": 2.5167932988286623e-06,
      "loss": 1.309,
      "step": 9977
    },
    {
      "epoch": 1.797433010583202,
      "grad_norm": 1.5807194709777832,
      "learning_rate": 2.512358823929184e-06,
      "loss": 1.4594,
      "step": 9978
    },
    {
      "epoch": 1.7976131501913983,
      "grad_norm": 1.6512035131454468,
      "learning_rate": 2.5079281584651393e-06,
      "loss": 1.5128,
      "step": 9979
    },
    {
      "epoch": 1.7977932897995947,
      "grad_norm": 1.501667857170105,
      "learning_rate": 2.5035013027919717e-06,
      "loss": 1.4507,
      "step": 9980
    },
    {
      "epoch": 1.797973429407791,
      "grad_norm": 1.4809303283691406,
      "learning_rate": 2.4990782572647975e-06,
      "loss": 1.3401,
      "step": 9981
    },
    {
      "epoch": 1.7981535690159873,
      "grad_norm": 1.4275723695755005,
      "learning_rate": 2.494659022238438e-06,
      "loss": 1.3783,
      "step": 9982
    },
    {
      "epoch": 1.7983337086241837,
      "grad_norm": 1.6030213832855225,
      "learning_rate": 2.490243598067382e-06,
      "loss": 1.5512,
      "step": 9983
    },
    {
      "epoch": 1.7985138482323801,
      "grad_norm": 1.6979684829711914,
      "learning_rate": 2.4858319851058354e-06,
      "loss": 1.5776,
      "step": 9984
    },
    {
      "epoch": 1.7986939878405765,
      "grad_norm": 1.3408842086791992,
      "learning_rate": 2.4814241837077145e-06,
      "loss": 1.1557,
      "step": 9985
    },
    {
      "epoch": 1.798874127448773,
      "grad_norm": 1.671297311782837,
      "learning_rate": 2.4770201942265927e-06,
      "loss": 1.5044,
      "step": 9986
    },
    {
      "epoch": 1.7990542670569691,
      "grad_norm": 1.49205482006073,
      "learning_rate": 2.47262001701577e-06,
      "loss": 1.3709,
      "step": 9987
    },
    {
      "epoch": 1.7992344066651655,
      "grad_norm": 1.4086471796035767,
      "learning_rate": 2.4682236524282143e-06,
      "loss": 1.3189,
      "step": 9988
    },
    {
      "epoch": 1.7994145462733617,
      "grad_norm": 1.503443956375122,
      "learning_rate": 2.4638311008166047e-06,
      "loss": 1.3862,
      "step": 9989
    },
    {
      "epoch": 1.7995946858815581,
      "grad_norm": 1.35723876953125,
      "learning_rate": 2.459442362533315e-06,
      "loss": 1.2539,
      "step": 9990
    },
    {
      "epoch": 1.7997748254897545,
      "grad_norm": 1.5897128582000732,
      "learning_rate": 2.4550574379304027e-06,
      "loss": 1.5706,
      "step": 9991
    },
    {
      "epoch": 1.799954965097951,
      "grad_norm": 1.6506224870681763,
      "learning_rate": 2.4506763273596247e-06,
      "loss": 1.4168,
      "step": 9992
    },
    {
      "epoch": 1.8001351047061473,
      "grad_norm": 1.6067757606506348,
      "learning_rate": 2.446299031172433e-06,
      "loss": 1.4711,
      "step": 9993
    },
    {
      "epoch": 1.8003152443143438,
      "grad_norm": 1.526239037513733,
      "learning_rate": 2.4419255497199646e-06,
      "loss": 1.3983,
      "step": 9994
    },
    {
      "epoch": 1.80049538392254,
      "grad_norm": 1.463149070739746,
      "learning_rate": 2.4375558833530653e-06,
      "loss": 1.1949,
      "step": 9995
    },
    {
      "epoch": 1.8006755235307363,
      "grad_norm": 1.6321347951889038,
      "learning_rate": 2.4331900324222723e-06,
      "loss": 1.5065,
      "step": 9996
    },
    {
      "epoch": 1.8008556631389325,
      "grad_norm": 1.6577446460723877,
      "learning_rate": 2.428827997277794e-06,
      "loss": 1.4318,
      "step": 9997
    },
    {
      "epoch": 1.801035802747129,
      "grad_norm": 1.56265389919281,
      "learning_rate": 2.4244697782695726e-06,
      "loss": 1.2484,
      "step": 9998
    },
    {
      "epoch": 1.8012159423553253,
      "grad_norm": 1.4934924840927124,
      "learning_rate": 2.420115375747206e-06,
      "loss": 1.2096,
      "step": 9999
    },
    {
      "epoch": 1.8013960819635217,
      "grad_norm": 1.493435025215149,
      "learning_rate": 2.4157647900600154e-06,
      "loss": 1.2484,
      "step": 10000
    },
    {
      "epoch": 1.8015762215717182,
      "grad_norm": 1.6239197254180908,
      "learning_rate": 2.4114180215569936e-06,
      "loss": 1.3808,
      "step": 10001
    },
    {
      "epoch": 1.8017563611799146,
      "grad_norm": 1.2958792448043823,
      "learning_rate": 2.4070750705868394e-06,
      "loss": 1.5937,
      "step": 10002
    },
    {
      "epoch": 1.8019365007881107,
      "grad_norm": 1.387211561203003,
      "learning_rate": 2.4027359374979407e-06,
      "loss": 1.8858,
      "step": 10003
    },
    {
      "epoch": 1.8021166403963071,
      "grad_norm": 1.3409655094146729,
      "learning_rate": 2.3984006226383805e-06,
      "loss": 1.6068,
      "step": 10004
    },
    {
      "epoch": 1.8022967800045033,
      "grad_norm": 1.3563859462738037,
      "learning_rate": 2.3940691263559356e-06,
      "loss": 1.7366,
      "step": 10005
    },
    {
      "epoch": 1.8024769196126997,
      "grad_norm": 1.4528411626815796,
      "learning_rate": 2.3897414489980786e-06,
      "loss": 1.7807,
      "step": 10006
    },
    {
      "epoch": 1.8026570592208961,
      "grad_norm": 1.4837846755981445,
      "learning_rate": 2.3854175909119645e-06,
      "loss": 1.7435,
      "step": 10007
    },
    {
      "epoch": 1.8028371988290925,
      "grad_norm": 1.4679380655288696,
      "learning_rate": 2.381097552444467e-06,
      "loss": 1.647,
      "step": 10008
    },
    {
      "epoch": 1.803017338437289,
      "grad_norm": 1.5496721267700195,
      "learning_rate": 2.3767813339421295e-06,
      "loss": 1.7539,
      "step": 10009
    },
    {
      "epoch": 1.8031974780454854,
      "grad_norm": 1.6995376348495483,
      "learning_rate": 2.3724689357511986e-06,
      "loss": 1.7397,
      "step": 10010
    },
    {
      "epoch": 1.8033776176536818,
      "grad_norm": 1.9499925374984741,
      "learning_rate": 2.368160358217603e-06,
      "loss": 2.0504,
      "step": 10011
    },
    {
      "epoch": 1.803557757261878,
      "grad_norm": 1.675574541091919,
      "learning_rate": 2.3638556016869875e-06,
      "loss": 1.8564,
      "step": 10012
    },
    {
      "epoch": 1.8037378968700744,
      "grad_norm": 1.4830348491668701,
      "learning_rate": 2.3595546665046765e-06,
      "loss": 1.3947,
      "step": 10013
    },
    {
      "epoch": 1.8039180364782705,
      "grad_norm": 1.472398281097412,
      "learning_rate": 2.3552575530156885e-06,
      "loss": 1.4108,
      "step": 10014
    },
    {
      "epoch": 1.804098176086467,
      "grad_norm": 1.3872883319854736,
      "learning_rate": 2.350964261564725e-06,
      "loss": 1.3408,
      "step": 10015
    },
    {
      "epoch": 1.8042783156946633,
      "grad_norm": 1.427467703819275,
      "learning_rate": 2.3466747924961952e-06,
      "loss": 1.1747,
      "step": 10016
    },
    {
      "epoch": 1.8044584553028598,
      "grad_norm": 1.5777684450149536,
      "learning_rate": 2.3423891461542113e-06,
      "loss": 1.463,
      "step": 10017
    },
    {
      "epoch": 1.8046385949110562,
      "grad_norm": 1.277117371559143,
      "learning_rate": 2.3381073228825543e-06,
      "loss": 1.0313,
      "step": 10018
    },
    {
      "epoch": 1.8048187345192526,
      "grad_norm": 1.3496284484863281,
      "learning_rate": 2.3338293230247166e-06,
      "loss": 1.327,
      "step": 10019
    },
    {
      "epoch": 1.8049988741274487,
      "grad_norm": 1.3696247339248657,
      "learning_rate": 2.3295551469238675e-06,
      "loss": 1.2093,
      "step": 10020
    },
    {
      "epoch": 1.8051790137356452,
      "grad_norm": 1.4592262506484985,
      "learning_rate": 2.3252847949228827e-06,
      "loss": 1.3693,
      "step": 10021
    },
    {
      "epoch": 1.8053591533438413,
      "grad_norm": 1.3412666320800781,
      "learning_rate": 2.321018267364333e-06,
      "loss": 1.1342,
      "step": 10022
    },
    {
      "epoch": 1.8055392929520377,
      "grad_norm": 1.7362686395645142,
      "learning_rate": 2.3167555645904783e-06,
      "loss": 1.4592,
      "step": 10023
    },
    {
      "epoch": 1.8057194325602342,
      "grad_norm": 1.5533839464187622,
      "learning_rate": 2.312496686943261e-06,
      "loss": 1.533,
      "step": 10024
    },
    {
      "epoch": 1.8058995721684306,
      "grad_norm": 1.3072752952575684,
      "learning_rate": 2.3082416347643352e-06,
      "loss": 1.3038,
      "step": 10025
    },
    {
      "epoch": 1.806079711776627,
      "grad_norm": 1.2663114070892334,
      "learning_rate": 2.303990408395035e-06,
      "loss": 1.1934,
      "step": 10026
    },
    {
      "epoch": 1.8062598513848234,
      "grad_norm": 1.5046476125717163,
      "learning_rate": 2.2997430081763916e-06,
      "loss": 1.4549,
      "step": 10027
    },
    {
      "epoch": 1.8064399909930196,
      "grad_norm": 1.4995585680007935,
      "learning_rate": 2.295499434449133e-06,
      "loss": 1.3057,
      "step": 10028
    },
    {
      "epoch": 1.806620130601216,
      "grad_norm": 1.4494708776474,
      "learning_rate": 2.2912596875536707e-06,
      "loss": 1.3853,
      "step": 10029
    },
    {
      "epoch": 1.8068002702094121,
      "grad_norm": 1.4352213144302368,
      "learning_rate": 2.2870237678301154e-06,
      "loss": 1.2069,
      "step": 10030
    },
    {
      "epoch": 1.8069804098176085,
      "grad_norm": 1.6235201358795166,
      "learning_rate": 2.282791675618279e-06,
      "loss": 1.6152,
      "step": 10031
    },
    {
      "epoch": 1.807160549425805,
      "grad_norm": 1.4555021524429321,
      "learning_rate": 2.2785634112576505e-06,
      "loss": 1.292,
      "step": 10032
    },
    {
      "epoch": 1.8073406890340014,
      "grad_norm": 1.5760633945465088,
      "learning_rate": 2.27433897508742e-06,
      "loss": 1.3454,
      "step": 10033
    },
    {
      "epoch": 1.8075208286421978,
      "grad_norm": 1.4704110622406006,
      "learning_rate": 2.2701183674464722e-06,
      "loss": 1.4319,
      "step": 10034
    },
    {
      "epoch": 1.8077009682503942,
      "grad_norm": 1.3243273496627808,
      "learning_rate": 2.26590158867338e-06,
      "loss": 1.0766,
      "step": 10035
    },
    {
      "epoch": 1.8078811078585904,
      "grad_norm": 1.3688443899154663,
      "learning_rate": 2.261688639106413e-06,
      "loss": 1.1436,
      "step": 10036
    },
    {
      "epoch": 1.8080612474667868,
      "grad_norm": 1.4448790550231934,
      "learning_rate": 2.2574795190835275e-06,
      "loss": 1.2796,
      "step": 10037
    },
    {
      "epoch": 1.8082413870749832,
      "grad_norm": 1.5203858613967896,
      "learning_rate": 2.2532742289423815e-06,
      "loss": 1.2263,
      "step": 10038
    },
    {
      "epoch": 1.8084215266831793,
      "grad_norm": 1.6407034397125244,
      "learning_rate": 2.249072769020316e-06,
      "loss": 1.4458,
      "step": 10039
    },
    {
      "epoch": 1.8086016662913758,
      "grad_norm": 1.4767647981643677,
      "learning_rate": 2.2448751396543787e-06,
      "loss": 1.3189,
      "step": 10040
    },
    {
      "epoch": 1.8087818058995722,
      "grad_norm": 1.5921157598495483,
      "learning_rate": 2.2406813411812997e-06,
      "loss": 1.3455,
      "step": 10041
    },
    {
      "epoch": 1.8089619455077686,
      "grad_norm": 1.569536566734314,
      "learning_rate": 2.2364913739374995e-06,
      "loss": 1.4725,
      "step": 10042
    },
    {
      "epoch": 1.809142085115965,
      "grad_norm": 1.5300394296646118,
      "learning_rate": 2.232305238259097e-06,
      "loss": 1.2021,
      "step": 10043
    },
    {
      "epoch": 1.8093222247241614,
      "grad_norm": 1.6046299934387207,
      "learning_rate": 2.228122934481891e-06,
      "loss": 1.3953,
      "step": 10044
    },
    {
      "epoch": 1.8095023643323576,
      "grad_norm": 1.498541235923767,
      "learning_rate": 2.223944462941402e-06,
      "loss": 1.3705,
      "step": 10045
    },
    {
      "epoch": 1.809682503940554,
      "grad_norm": 1.7935336828231812,
      "learning_rate": 2.219769823972828e-06,
      "loss": 1.5541,
      "step": 10046
    },
    {
      "epoch": 1.8098626435487501,
      "grad_norm": 1.536406397819519,
      "learning_rate": 2.2155990179110353e-06,
      "loss": 1.409,
      "step": 10047
    },
    {
      "epoch": 1.8100427831569466,
      "grad_norm": 1.6193208694458008,
      "learning_rate": 2.2114320450906166e-06,
      "loss": 1.2533,
      "step": 10048
    },
    {
      "epoch": 1.810222922765143,
      "grad_norm": 1.6491739749908447,
      "learning_rate": 2.2072689058458384e-06,
      "loss": 1.3188,
      "step": 10049
    },
    {
      "epoch": 1.8104030623733394,
      "grad_norm": 1.6443055868148804,
      "learning_rate": 2.203109600510672e-06,
      "loss": 1.3765,
      "step": 10050
    },
    {
      "epoch": 1.8105832019815358,
      "grad_norm": 1.5825101137161255,
      "learning_rate": 2.198954129418773e-06,
      "loss": 1.1676,
      "step": 10051
    },
    {
      "epoch": 1.8107633415897322,
      "grad_norm": 1.3962745666503906,
      "learning_rate": 2.194802492903492e-06,
      "loss": 1.7069,
      "step": 10052
    },
    {
      "epoch": 1.8109434811979284,
      "grad_norm": 1.2918355464935303,
      "learning_rate": 2.190654691297861e-06,
      "loss": 1.7,
      "step": 10053
    },
    {
      "epoch": 1.8111236208061248,
      "grad_norm": 1.3591047525405884,
      "learning_rate": 2.186510724934632e-06,
      "loss": 1.9171,
      "step": 10054
    },
    {
      "epoch": 1.811303760414321,
      "grad_norm": 1.430289626121521,
      "learning_rate": 2.1823705941462215e-06,
      "loss": 1.5948,
      "step": 10055
    },
    {
      "epoch": 1.8114839000225174,
      "grad_norm": 1.529053807258606,
      "learning_rate": 2.178234299264753e-06,
      "loss": 1.8438,
      "step": 10056
    },
    {
      "epoch": 1.8116640396307138,
      "grad_norm": 1.5243395566940308,
      "learning_rate": 2.174101840622034e-06,
      "loss": 1.8101,
      "step": 10057
    },
    {
      "epoch": 1.8118441792389102,
      "grad_norm": 1.5169636011123657,
      "learning_rate": 2.1699732185495757e-06,
      "loss": 1.9676,
      "step": 10058
    },
    {
      "epoch": 1.8120243188471066,
      "grad_norm": 1.5018396377563477,
      "learning_rate": 2.1658484333785643e-06,
      "loss": 1.5021,
      "step": 10059
    },
    {
      "epoch": 1.812204458455303,
      "grad_norm": 1.6437705755233765,
      "learning_rate": 2.1617274854398905e-06,
      "loss": 1.7939,
      "step": 10060
    },
    {
      "epoch": 1.8123845980634992,
      "grad_norm": 1.9435685873031616,
      "learning_rate": 2.15761037506414e-06,
      "loss": 2.1778,
      "step": 10061
    },
    {
      "epoch": 1.8125647376716956,
      "grad_norm": 1.6255899667739868,
      "learning_rate": 2.153497102581581e-06,
      "loss": 1.7574,
      "step": 10062
    },
    {
      "epoch": 1.8127448772798918,
      "grad_norm": 1.5100345611572266,
      "learning_rate": 2.1493876683221727e-06,
      "loss": 1.5662,
      "step": 10063
    },
    {
      "epoch": 1.8129250168880882,
      "grad_norm": 1.451920986175537,
      "learning_rate": 2.1452820726155844e-06,
      "loss": 1.4867,
      "step": 10064
    },
    {
      "epoch": 1.8131051564962846,
      "grad_norm": 1.4553110599517822,
      "learning_rate": 2.141180315791158e-06,
      "loss": 1.5681,
      "step": 10065
    },
    {
      "epoch": 1.813285296104481,
      "grad_norm": 1.444446086883545,
      "learning_rate": 2.1370823981779364e-06,
      "loss": 1.4385,
      "step": 10066
    },
    {
      "epoch": 1.8134654357126774,
      "grad_norm": 1.4607234001159668,
      "learning_rate": 2.1329883201046506e-06,
      "loss": 1.4436,
      "step": 10067
    },
    {
      "epoch": 1.8136455753208738,
      "grad_norm": 1.399939775466919,
      "learning_rate": 2.1288980818997275e-06,
      "loss": 1.2261,
      "step": 10068
    },
    {
      "epoch": 1.8138257149290702,
      "grad_norm": 1.4652578830718994,
      "learning_rate": 2.1248116838912867e-06,
      "loss": 1.4139,
      "step": 10069
    },
    {
      "epoch": 1.8140058545372664,
      "grad_norm": 1.2725590467453003,
      "learning_rate": 2.1207291264071284e-06,
      "loss": 1.2006,
      "step": 10070
    },
    {
      "epoch": 1.8141859941454628,
      "grad_norm": 1.3044129610061646,
      "learning_rate": 2.116650409774762e-06,
      "loss": 1.1308,
      "step": 10071
    },
    {
      "epoch": 1.814366133753659,
      "grad_norm": 1.4853923320770264,
      "learning_rate": 2.112575534321365e-06,
      "loss": 1.4973,
      "step": 10072
    },
    {
      "epoch": 1.8145462733618554,
      "grad_norm": 1.2549855709075928,
      "learning_rate": 2.1085045003738478e-06,
      "loss": 1.1359,
      "step": 10073
    },
    {
      "epoch": 1.8147264129700518,
      "grad_norm": 1.3858438730239868,
      "learning_rate": 2.1044373082587664e-06,
      "loss": 1.334,
      "step": 10074
    },
    {
      "epoch": 1.8149065525782482,
      "grad_norm": 1.4075618982315063,
      "learning_rate": 2.1003739583023984e-06,
      "loss": 1.3103,
      "step": 10075
    },
    {
      "epoch": 1.8150866921864446,
      "grad_norm": 1.3848905563354492,
      "learning_rate": 2.0963144508306944e-06,
      "loss": 1.3577,
      "step": 10076
    },
    {
      "epoch": 1.815266831794641,
      "grad_norm": 1.6225776672363281,
      "learning_rate": 2.09225878616931e-06,
      "loss": 1.6012,
      "step": 10077
    },
    {
      "epoch": 1.8154469714028372,
      "grad_norm": 1.4673770666122437,
      "learning_rate": 2.0882069646435964e-06,
      "loss": 1.2149,
      "step": 10078
    },
    {
      "epoch": 1.8156271110110336,
      "grad_norm": 1.5840201377868652,
      "learning_rate": 2.0841589865785938e-06,
      "loss": 1.4817,
      "step": 10079
    },
    {
      "epoch": 1.8158072506192298,
      "grad_norm": 1.5192190408706665,
      "learning_rate": 2.080114852299009e-06,
      "loss": 1.4626,
      "step": 10080
    },
    {
      "epoch": 1.8159873902274262,
      "grad_norm": 1.4348398447036743,
      "learning_rate": 2.07607456212926e-06,
      "loss": 1.458,
      "step": 10081
    },
    {
      "epoch": 1.8161675298356226,
      "grad_norm": 1.342737078666687,
      "learning_rate": 2.0720381163934765e-06,
      "loss": 1.2706,
      "step": 10082
    },
    {
      "epoch": 1.816347669443819,
      "grad_norm": 1.45393967628479,
      "learning_rate": 2.06800551541545e-06,
      "loss": 1.2374,
      "step": 10083
    },
    {
      "epoch": 1.8165278090520154,
      "grad_norm": 1.415820598602295,
      "learning_rate": 2.0639767595186765e-06,
      "loss": 1.1386,
      "step": 10084
    },
    {
      "epoch": 1.8167079486602118,
      "grad_norm": 1.3049286603927612,
      "learning_rate": 2.0599518490263316e-06,
      "loss": 1.2418,
      "step": 10085
    },
    {
      "epoch": 1.816888088268408,
      "grad_norm": 1.3920007944107056,
      "learning_rate": 2.0559307842613007e-06,
      "loss": 1.1824,
      "step": 10086
    },
    {
      "epoch": 1.8170682278766044,
      "grad_norm": 1.5656129121780396,
      "learning_rate": 2.0519135655461484e-06,
      "loss": 1.242,
      "step": 10087
    },
    {
      "epoch": 1.8172483674848006,
      "grad_norm": 1.453320860862732,
      "learning_rate": 2.0479001932031393e-06,
      "loss": 1.3428,
      "step": 10088
    },
    {
      "epoch": 1.817428507092997,
      "grad_norm": 1.6345810890197754,
      "learning_rate": 2.0438906675542214e-06,
      "loss": 1.4914,
      "step": 10089
    },
    {
      "epoch": 1.8176086467011934,
      "grad_norm": 1.692877173423767,
      "learning_rate": 2.0398849889210313e-06,
      "loss": 1.446,
      "step": 10090
    },
    {
      "epoch": 1.8177887863093898,
      "grad_norm": 1.5764058828353882,
      "learning_rate": 2.0358831576249126e-06,
      "loss": 1.4218,
      "step": 10091
    },
    {
      "epoch": 1.8179689259175862,
      "grad_norm": 1.6342402696609497,
      "learning_rate": 2.0318851739868805e-06,
      "loss": 1.4039,
      "step": 10092
    },
    {
      "epoch": 1.8181490655257826,
      "grad_norm": 1.6368536949157715,
      "learning_rate": 2.027891038327656e-06,
      "loss": 1.4346,
      "step": 10093
    },
    {
      "epoch": 1.818329205133979,
      "grad_norm": 1.5874720811843872,
      "learning_rate": 2.0239007509676443e-06,
      "loss": 1.4866,
      "step": 10094
    },
    {
      "epoch": 1.8185093447421752,
      "grad_norm": 1.6649432182312012,
      "learning_rate": 2.0199143122269394e-06,
      "loss": 1.3545,
      "step": 10095
    },
    {
      "epoch": 1.8186894843503716,
      "grad_norm": 1.6224302053451538,
      "learning_rate": 2.0159317224253517e-06,
      "loss": 1.2969,
      "step": 10096
    },
    {
      "epoch": 1.8188696239585678,
      "grad_norm": 1.5880509614944458,
      "learning_rate": 2.0119529818823423e-06,
      "loss": 1.4174,
      "step": 10097
    },
    {
      "epoch": 1.8190497635667642,
      "grad_norm": 1.609045386314392,
      "learning_rate": 2.00797809091709e-06,
      "loss": 1.4974,
      "step": 10098
    },
    {
      "epoch": 1.8192299031749606,
      "grad_norm": 1.6164910793304443,
      "learning_rate": 2.004007049848461e-06,
      "loss": 1.2302,
      "step": 10099
    },
    {
      "epoch": 1.819410042783157,
      "grad_norm": 1.6428788900375366,
      "learning_rate": 2.000039858995012e-06,
      "loss": 1.3782,
      "step": 10100
    },
    {
      "epoch": 1.8195901823913534,
      "grad_norm": 1.3995789289474487,
      "learning_rate": 1.9960765186749777e-06,
      "loss": 0.9754,
      "step": 10101
    },
    {
      "epoch": 1.8197703219995498,
      "grad_norm": 1.344395637512207,
      "learning_rate": 1.9921170292063086e-06,
      "loss": 1.6348,
      "step": 10102
    },
    {
      "epoch": 1.819950461607746,
      "grad_norm": 1.2843656539916992,
      "learning_rate": 1.9881613909066287e-06,
      "loss": 1.6399,
      "step": 10103
    },
    {
      "epoch": 1.8201306012159424,
      "grad_norm": 1.4430584907531738,
      "learning_rate": 1.9842096040932457e-06,
      "loss": 1.5001,
      "step": 10104
    },
    {
      "epoch": 1.8203107408241386,
      "grad_norm": 1.6491752862930298,
      "learning_rate": 1.980261669083189e-06,
      "loss": 2.1396,
      "step": 10105
    },
    {
      "epoch": 1.820490880432335,
      "grad_norm": 1.5357755422592163,
      "learning_rate": 1.976317586193155e-06,
      "loss": 1.9067,
      "step": 10106
    },
    {
      "epoch": 1.8206710200405314,
      "grad_norm": 1.4185600280761719,
      "learning_rate": 1.9723773557395297e-06,
      "loss": 1.5255,
      "step": 10107
    },
    {
      "epoch": 1.8208511596487278,
      "grad_norm": 1.4339519739151,
      "learning_rate": 1.9684409780383992e-06,
      "loss": 1.6866,
      "step": 10108
    },
    {
      "epoch": 1.8210312992569242,
      "grad_norm": 1.5583972930908203,
      "learning_rate": 1.964508453405528e-06,
      "loss": 1.7289,
      "step": 10109
    },
    {
      "epoch": 1.8212114388651206,
      "grad_norm": 1.7545312643051147,
      "learning_rate": 1.960579782156402e-06,
      "loss": 1.6404,
      "step": 10110
    },
    {
      "epoch": 1.8213915784733168,
      "grad_norm": 1.6488924026489258,
      "learning_rate": 1.956654964606175e-06,
      "loss": 1.8884,
      "step": 10111
    },
    {
      "epoch": 1.8215717180815132,
      "grad_norm": 1.586043357849121,
      "learning_rate": 1.952734001069678e-06,
      "loss": 1.7718,
      "step": 10112
    },
    {
      "epoch": 1.8217518576897094,
      "grad_norm": 1.465941071510315,
      "learning_rate": 1.948816891861455e-06,
      "loss": 1.4703,
      "step": 10113
    },
    {
      "epoch": 1.8219319972979058,
      "grad_norm": 1.3636362552642822,
      "learning_rate": 1.9449036372957307e-06,
      "loss": 1.5068,
      "step": 10114
    },
    {
      "epoch": 1.8221121369061022,
      "grad_norm": 1.4096664190292358,
      "learning_rate": 1.9409942376864333e-06,
      "loss": 1.4697,
      "step": 10115
    },
    {
      "epoch": 1.8222922765142986,
      "grad_norm": 1.3681811094284058,
      "learning_rate": 1.937088693347172e-06,
      "loss": 1.0556,
      "step": 10116
    },
    {
      "epoch": 1.822472416122495,
      "grad_norm": 1.343756079673767,
      "learning_rate": 1.933187004591247e-06,
      "loss": 1.2496,
      "step": 10117
    },
    {
      "epoch": 1.8226525557306914,
      "grad_norm": 1.4849203824996948,
      "learning_rate": 1.929289171731641e-06,
      "loss": 1.2679,
      "step": 10118
    },
    {
      "epoch": 1.8228326953388876,
      "grad_norm": 1.2953969240188599,
      "learning_rate": 1.9253951950810478e-06,
      "loss": 1.2433,
      "step": 10119
    },
    {
      "epoch": 1.823012834947084,
      "grad_norm": 1.337950587272644,
      "learning_rate": 1.9215050749518348e-06,
      "loss": 1.2363,
      "step": 10120
    },
    {
      "epoch": 1.8231929745552802,
      "grad_norm": 1.5039666891098022,
      "learning_rate": 1.9176188116560688e-06,
      "loss": 1.5663,
      "step": 10121
    },
    {
      "epoch": 1.8233731141634766,
      "grad_norm": 1.3163189888000488,
      "learning_rate": 1.9137364055055006e-06,
      "loss": 1.3075,
      "step": 10122
    },
    {
      "epoch": 1.823553253771673,
      "grad_norm": 1.4876776933670044,
      "learning_rate": 1.9098578568115754e-06,
      "loss": 1.3066,
      "step": 10123
    },
    {
      "epoch": 1.8237333933798694,
      "grad_norm": 1.507344126701355,
      "learning_rate": 1.9059831658854278e-06,
      "loss": 1.5504,
      "step": 10124
    },
    {
      "epoch": 1.8239135329880658,
      "grad_norm": 1.416939616203308,
      "learning_rate": 1.9021123330378864e-06,
      "loss": 1.3996,
      "step": 10125
    },
    {
      "epoch": 1.8240936725962622,
      "grad_norm": 1.486570119857788,
      "learning_rate": 1.8982453585794647e-06,
      "loss": 1.4995,
      "step": 10126
    },
    {
      "epoch": 1.8242738122044586,
      "grad_norm": 1.489565134048462,
      "learning_rate": 1.8943822428203751e-06,
      "loss": 1.3586,
      "step": 10127
    },
    {
      "epoch": 1.8244539518126548,
      "grad_norm": 1.5417814254760742,
      "learning_rate": 1.890522986070503e-06,
      "loss": 1.3543,
      "step": 10128
    },
    {
      "epoch": 1.8246340914208512,
      "grad_norm": 1.3175015449523926,
      "learning_rate": 1.8866675886394457e-06,
      "loss": 1.1671,
      "step": 10129
    },
    {
      "epoch": 1.8248142310290474,
      "grad_norm": 1.414724588394165,
      "learning_rate": 1.8828160508364833e-06,
      "loss": 1.2677,
      "step": 10130
    },
    {
      "epoch": 1.8249943706372438,
      "grad_norm": 1.4064226150512695,
      "learning_rate": 1.8789683729705799e-06,
      "loss": 1.4351,
      "step": 10131
    },
    {
      "epoch": 1.8251745102454402,
      "grad_norm": 1.5006777048110962,
      "learning_rate": 1.8751245553503937e-06,
      "loss": 1.2262,
      "step": 10132
    },
    {
      "epoch": 1.8253546498536366,
      "grad_norm": 1.5483022928237915,
      "learning_rate": 1.871284598284273e-06,
      "loss": 1.4874,
      "step": 10133
    },
    {
      "epoch": 1.825534789461833,
      "grad_norm": 1.5291799306869507,
      "learning_rate": 1.86744850208026e-06,
      "loss": 1.3124,
      "step": 10134
    },
    {
      "epoch": 1.8257149290700294,
      "grad_norm": 1.542891502380371,
      "learning_rate": 1.8636162670460867e-06,
      "loss": 1.3712,
      "step": 10135
    },
    {
      "epoch": 1.8258950686782256,
      "grad_norm": 1.6982064247131348,
      "learning_rate": 1.8597878934891677e-06,
      "loss": 1.5616,
      "step": 10136
    },
    {
      "epoch": 1.826075208286422,
      "grad_norm": 1.5469884872436523,
      "learning_rate": 1.8559633817166133e-06,
      "loss": 1.4207,
      "step": 10137
    },
    {
      "epoch": 1.8262553478946182,
      "grad_norm": 1.6189945936203003,
      "learning_rate": 1.8521427320352281e-06,
      "loss": 1.3411,
      "step": 10138
    },
    {
      "epoch": 1.8264354875028146,
      "grad_norm": 1.6197736263275146,
      "learning_rate": 1.8483259447515e-06,
      "loss": 1.4551,
      "step": 10139
    },
    {
      "epoch": 1.826615627111011,
      "grad_norm": 1.43265962600708,
      "learning_rate": 1.8445130201716121e-06,
      "loss": 1.3722,
      "step": 10140
    },
    {
      "epoch": 1.8267957667192074,
      "grad_norm": 1.5799026489257812,
      "learning_rate": 1.8407039586014363e-06,
      "loss": 1.3739,
      "step": 10141
    },
    {
      "epoch": 1.8269759063274038,
      "grad_norm": 1.7282644510269165,
      "learning_rate": 1.8368987603465225e-06,
      "loss": 1.5924,
      "step": 10142
    },
    {
      "epoch": 1.8271560459356002,
      "grad_norm": 1.562292456626892,
      "learning_rate": 1.833097425712138e-06,
      "loss": 1.436,
      "step": 10143
    },
    {
      "epoch": 1.8273361855437964,
      "grad_norm": 1.6173843145370483,
      "learning_rate": 1.8292999550032218e-06,
      "loss": 1.3085,
      "step": 10144
    },
    {
      "epoch": 1.8275163251519928,
      "grad_norm": 1.633520483970642,
      "learning_rate": 1.8255063485243973e-06,
      "loss": 1.3988,
      "step": 10145
    },
    {
      "epoch": 1.827696464760189,
      "grad_norm": 1.6982747316360474,
      "learning_rate": 1.8217166065799818e-06,
      "loss": 1.5243,
      "step": 10146
    },
    {
      "epoch": 1.8278766043683854,
      "grad_norm": 1.6105185747146606,
      "learning_rate": 1.817930729473999e-06,
      "loss": 1.4113,
      "step": 10147
    },
    {
      "epoch": 1.8280567439765818,
      "grad_norm": 1.845295786857605,
      "learning_rate": 1.814148717510139e-06,
      "loss": 1.4898,
      "step": 10148
    },
    {
      "epoch": 1.8282368835847782,
      "grad_norm": 1.5970302820205688,
      "learning_rate": 1.810370570991804e-06,
      "loss": 1.3528,
      "step": 10149
    },
    {
      "epoch": 1.8284170231929746,
      "grad_norm": 1.4034368991851807,
      "learning_rate": 1.806596290222068e-06,
      "loss": 1.0075,
      "step": 10150
    },
    {
      "epoch": 1.828597162801171,
      "grad_norm": 1.4156461954116821,
      "learning_rate": 1.8028258755036997e-06,
      "loss": 1.2236,
      "step": 10151
    },
    {
      "epoch": 1.8287773024093674,
      "grad_norm": 1.4610676765441895,
      "learning_rate": 1.7990593271391688e-06,
      "loss": 2.1319,
      "step": 10152
    },
    {
      "epoch": 1.8289574420175636,
      "grad_norm": 1.227811336517334,
      "learning_rate": 1.7952966454306163e-06,
      "loss": 1.7249,
      "step": 10153
    },
    {
      "epoch": 1.82913758162576,
      "grad_norm": 1.4469199180603027,
      "learning_rate": 1.7915378306798902e-06,
      "loss": 1.9271,
      "step": 10154
    },
    {
      "epoch": 1.8293177212339562,
      "grad_norm": 1.3816676139831543,
      "learning_rate": 1.7877828831885213e-06,
      "loss": 1.6807,
      "step": 10155
    },
    {
      "epoch": 1.8294978608421526,
      "grad_norm": 1.3797568082809448,
      "learning_rate": 1.7840318032577185e-06,
      "loss": 1.7105,
      "step": 10156
    },
    {
      "epoch": 1.829678000450349,
      "grad_norm": 1.4409549236297607,
      "learning_rate": 1.7802845911884081e-06,
      "loss": 1.7099,
      "step": 10157
    },
    {
      "epoch": 1.8298581400585454,
      "grad_norm": 1.6754438877105713,
      "learning_rate": 1.7765412472811771e-06,
      "loss": 1.8905,
      "step": 10158
    },
    {
      "epoch": 1.8300382796667418,
      "grad_norm": 1.5919604301452637,
      "learning_rate": 1.7728017718363133e-06,
      "loss": 1.7661,
      "step": 10159
    },
    {
      "epoch": 1.8302184192749382,
      "grad_norm": 1.6055443286895752,
      "learning_rate": 1.7690661651537989e-06,
      "loss": 1.7528,
      "step": 10160
    },
    {
      "epoch": 1.8303985588831344,
      "grad_norm": 1.9046741724014282,
      "learning_rate": 1.7653344275333105e-06,
      "loss": 1.8019,
      "step": 10161
    },
    {
      "epoch": 1.8305786984913308,
      "grad_norm": 1.4081676006317139,
      "learning_rate": 1.7616065592742038e-06,
      "loss": 1.4979,
      "step": 10162
    },
    {
      "epoch": 1.830758838099527,
      "grad_norm": 1.531377911567688,
      "learning_rate": 1.7578825606755168e-06,
      "loss": 1.5468,
      "step": 10163
    },
    {
      "epoch": 1.8309389777077234,
      "grad_norm": 1.4688129425048828,
      "learning_rate": 1.754162432035994e-06,
      "loss": 1.5188,
      "step": 10164
    },
    {
      "epoch": 1.8311191173159198,
      "grad_norm": 1.3672255277633667,
      "learning_rate": 1.7504461736540635e-06,
      "loss": 1.3315,
      "step": 10165
    },
    {
      "epoch": 1.8312992569241162,
      "grad_norm": 1.5113074779510498,
      "learning_rate": 1.7467337858278421e-06,
      "loss": 1.4907,
      "step": 10166
    },
    {
      "epoch": 1.8314793965323126,
      "grad_norm": 1.4541507959365845,
      "learning_rate": 1.7430252688551251e-06,
      "loss": 1.4352,
      "step": 10167
    },
    {
      "epoch": 1.831659536140509,
      "grad_norm": 1.515793800354004,
      "learning_rate": 1.7393206230334247e-06,
      "loss": 1.298,
      "step": 10168
    },
    {
      "epoch": 1.8318396757487052,
      "grad_norm": 1.4644867181777954,
      "learning_rate": 1.7356198486599085e-06,
      "loss": 1.4508,
      "step": 10169
    },
    {
      "epoch": 1.8320198153569016,
      "grad_norm": 1.5751250982284546,
      "learning_rate": 1.7319229460314612e-06,
      "loss": 1.6029,
      "step": 10170
    },
    {
      "epoch": 1.8321999549650978,
      "grad_norm": 1.658138394355774,
      "learning_rate": 1.7282299154446457e-06,
      "loss": 1.5106,
      "step": 10171
    },
    {
      "epoch": 1.8323800945732942,
      "grad_norm": 1.5127851963043213,
      "learning_rate": 1.7245407571957138e-06,
      "loss": 1.4804,
      "step": 10172
    },
    {
      "epoch": 1.8325602341814906,
      "grad_norm": 1.344985842704773,
      "learning_rate": 1.7208554715806123e-06,
      "loss": 1.3022,
      "step": 10173
    },
    {
      "epoch": 1.832740373789687,
      "grad_norm": 1.5787599086761475,
      "learning_rate": 1.7171740588949604e-06,
      "loss": 1.3923,
      "step": 10174
    },
    {
      "epoch": 1.8329205133978834,
      "grad_norm": 1.4782702922821045,
      "learning_rate": 1.7134965194340935e-06,
      "loss": 1.3885,
      "step": 10175
    },
    {
      "epoch": 1.8331006530060798,
      "grad_norm": 1.4548991918563843,
      "learning_rate": 1.7098228534930205e-06,
      "loss": 1.3111,
      "step": 10176
    },
    {
      "epoch": 1.833280792614276,
      "grad_norm": 1.478531837463379,
      "learning_rate": 1.7061530613664334e-06,
      "loss": 1.4396,
      "step": 10177
    },
    {
      "epoch": 1.8334609322224724,
      "grad_norm": 1.427119493484497,
      "learning_rate": 1.7024871433487244e-06,
      "loss": 1.2653,
      "step": 10178
    },
    {
      "epoch": 1.8336410718306688,
      "grad_norm": 1.5624186992645264,
      "learning_rate": 1.6988250997339638e-06,
      "loss": 1.4843,
      "step": 10179
    },
    {
      "epoch": 1.833821211438865,
      "grad_norm": 1.4301177263259888,
      "learning_rate": 1.6951669308159335e-06,
      "loss": 1.3281,
      "step": 10180
    },
    {
      "epoch": 1.8340013510470614,
      "grad_norm": 1.442164421081543,
      "learning_rate": 1.6915126368880873e-06,
      "loss": 1.243,
      "step": 10181
    },
    {
      "epoch": 1.8341814906552578,
      "grad_norm": 1.5827133655548096,
      "learning_rate": 1.6878622182435632e-06,
      "loss": 1.4054,
      "step": 10182
    },
    {
      "epoch": 1.8343616302634542,
      "grad_norm": 1.5139058828353882,
      "learning_rate": 1.6842156751751991e-06,
      "loss": 1.3846,
      "step": 10183
    },
    {
      "epoch": 1.8345417698716506,
      "grad_norm": 1.407804012298584,
      "learning_rate": 1.6805730079755167e-06,
      "loss": 1.3686,
      "step": 10184
    },
    {
      "epoch": 1.834721909479847,
      "grad_norm": 1.5105146169662476,
      "learning_rate": 1.6769342169367376e-06,
      "loss": 1.3396,
      "step": 10185
    },
    {
      "epoch": 1.8349020490880432,
      "grad_norm": 1.4453729391098022,
      "learning_rate": 1.6732993023507616e-06,
      "loss": 1.2886,
      "step": 10186
    },
    {
      "epoch": 1.8350821886962396,
      "grad_norm": 1.4646527767181396,
      "learning_rate": 1.6696682645091722e-06,
      "loss": 1.2723,
      "step": 10187
    },
    {
      "epoch": 1.8352623283044358,
      "grad_norm": 1.4350448846817017,
      "learning_rate": 1.666041103703253e-06,
      "loss": 1.2644,
      "step": 10188
    },
    {
      "epoch": 1.8354424679126322,
      "grad_norm": 1.661260962486267,
      "learning_rate": 1.6624178202239771e-06,
      "loss": 1.4881,
      "step": 10189
    },
    {
      "epoch": 1.8356226075208286,
      "grad_norm": 1.526147723197937,
      "learning_rate": 1.658798414361995e-06,
      "loss": 1.2691,
      "step": 10190
    },
    {
      "epoch": 1.835802747129025,
      "grad_norm": 1.576967477798462,
      "learning_rate": 1.6551828864076634e-06,
      "loss": 1.3672,
      "step": 10191
    },
    {
      "epoch": 1.8359828867372214,
      "grad_norm": 1.7092487812042236,
      "learning_rate": 1.651571236651006e-06,
      "loss": 1.4823,
      "step": 10192
    },
    {
      "epoch": 1.8361630263454178,
      "grad_norm": 1.584929347038269,
      "learning_rate": 1.647963465381752e-06,
      "loss": 1.4123,
      "step": 10193
    },
    {
      "epoch": 1.836343165953614,
      "grad_norm": 1.4314305782318115,
      "learning_rate": 1.6443595728893257e-06,
      "loss": 1.2829,
      "step": 10194
    },
    {
      "epoch": 1.8365233055618104,
      "grad_norm": 1.5391454696655273,
      "learning_rate": 1.6407595594628177e-06,
      "loss": 1.4597,
      "step": 10195
    },
    {
      "epoch": 1.8367034451700066,
      "grad_norm": 1.4913978576660156,
      "learning_rate": 1.637163425391025e-06,
      "loss": 1.1647,
      "step": 10196
    },
    {
      "epoch": 1.836883584778203,
      "grad_norm": 1.5547447204589844,
      "learning_rate": 1.6335711709624225e-06,
      "loss": 1.3477,
      "step": 10197
    },
    {
      "epoch": 1.8370637243863994,
      "grad_norm": 1.6399352550506592,
      "learning_rate": 1.629982796465185e-06,
      "loss": 1.1574,
      "step": 10198
    },
    {
      "epoch": 1.8372438639945958,
      "grad_norm": 1.5328298807144165,
      "learning_rate": 1.626398302187171e-06,
      "loss": 1.3126,
      "step": 10199
    },
    {
      "epoch": 1.8374240036027922,
      "grad_norm": 1.4650849103927612,
      "learning_rate": 1.6228176884159175e-06,
      "loss": 1.0858,
      "step": 10200
    },
    {
      "epoch": 1.8376041432109886,
      "grad_norm": 1.6660605669021606,
      "learning_rate": 1.6192409554386666e-06,
      "loss": 1.2572,
      "step": 10201
    },
    {
      "epoch": 1.8377842828191848,
      "grad_norm": 1.3632694482803345,
      "learning_rate": 1.6156681035423339e-06,
      "loss": 1.8665,
      "step": 10202
    },
    {
      "epoch": 1.8379644224273812,
      "grad_norm": 1.2494187355041504,
      "learning_rate": 1.6120991330135503e-06,
      "loss": 1.6396,
      "step": 10203
    },
    {
      "epoch": 1.8381445620355774,
      "grad_norm": 1.468569278717041,
      "learning_rate": 1.6085340441385987e-06,
      "loss": 1.7284,
      "step": 10204
    },
    {
      "epoch": 1.8383247016437738,
      "grad_norm": 1.3867123126983643,
      "learning_rate": 1.6049728372034778e-06,
      "loss": 1.7598,
      "step": 10205
    },
    {
      "epoch": 1.8385048412519702,
      "grad_norm": 1.3675591945648193,
      "learning_rate": 1.6014155124938647e-06,
      "loss": 1.4209,
      "step": 10206
    },
    {
      "epoch": 1.8386849808601666,
      "grad_norm": 1.4294353723526,
      "learning_rate": 1.5978620702951197e-06,
      "loss": 1.4595,
      "step": 10207
    },
    {
      "epoch": 1.838865120468363,
      "grad_norm": 1.5362040996551514,
      "learning_rate": 1.594312510892304e-06,
      "loss": 1.8629,
      "step": 10208
    },
    {
      "epoch": 1.8390452600765594,
      "grad_norm": 1.6337419748306274,
      "learning_rate": 1.590766834570173e-06,
      "loss": 1.6611,
      "step": 10209
    },
    {
      "epoch": 1.8392253996847558,
      "grad_norm": 1.7228219509124756,
      "learning_rate": 1.5872250416131385e-06,
      "loss": 1.8913,
      "step": 10210
    },
    {
      "epoch": 1.839405539292952,
      "grad_norm": 1.7560341358184814,
      "learning_rate": 1.5836871323053282e-06,
      "loss": 1.8003,
      "step": 10211
    },
    {
      "epoch": 1.8395856789011484,
      "grad_norm": 1.6732127666473389,
      "learning_rate": 1.5801531069305543e-06,
      "loss": 1.7057,
      "step": 10212
    },
    {
      "epoch": 1.8397658185093446,
      "grad_norm": 1.679484486579895,
      "learning_rate": 1.576622965772312e-06,
      "loss": 1.652,
      "step": 10213
    },
    {
      "epoch": 1.839945958117541,
      "grad_norm": 1.478257417678833,
      "learning_rate": 1.5730967091137972e-06,
      "loss": 1.3543,
      "step": 10214
    },
    {
      "epoch": 1.8401260977257374,
      "grad_norm": 1.4780102968215942,
      "learning_rate": 1.5695743372378724e-06,
      "loss": 1.3114,
      "step": 10215
    },
    {
      "epoch": 1.8403062373339338,
      "grad_norm": 1.3429474830627441,
      "learning_rate": 1.566055850427095e-06,
      "loss": 1.3006,
      "step": 10216
    },
    {
      "epoch": 1.8404863769421302,
      "grad_norm": 1.4642975330352783,
      "learning_rate": 1.5625412489637337e-06,
      "loss": 1.6605,
      "step": 10217
    },
    {
      "epoch": 1.8406665165503266,
      "grad_norm": 1.5364854335784912,
      "learning_rate": 1.5590305331297239e-06,
      "loss": 1.3993,
      "step": 10218
    },
    {
      "epoch": 1.8408466561585228,
      "grad_norm": 1.497288465499878,
      "learning_rate": 1.5555237032066905e-06,
      "loss": 1.5424,
      "step": 10219
    },
    {
      "epoch": 1.8410267957667192,
      "grad_norm": 1.3543816804885864,
      "learning_rate": 1.5520207594759416e-06,
      "loss": 1.2565,
      "step": 10220
    },
    {
      "epoch": 1.8412069353749154,
      "grad_norm": 1.4981905221939087,
      "learning_rate": 1.5485217022184973e-06,
      "loss": 1.4057,
      "step": 10221
    },
    {
      "epoch": 1.8413870749831118,
      "grad_norm": 1.4843984842300415,
      "learning_rate": 1.5450265317150381e-06,
      "loss": 1.5085,
      "step": 10222
    },
    {
      "epoch": 1.8415672145913082,
      "grad_norm": 1.5825599431991577,
      "learning_rate": 1.541535248245951e-06,
      "loss": 1.6376,
      "step": 10223
    },
    {
      "epoch": 1.8417473541995046,
      "grad_norm": 1.3740825653076172,
      "learning_rate": 1.5380478520913011e-06,
      "loss": 1.2755,
      "step": 10224
    },
    {
      "epoch": 1.841927493807701,
      "grad_norm": 1.5060142278671265,
      "learning_rate": 1.5345643435308421e-06,
      "loss": 1.3943,
      "step": 10225
    },
    {
      "epoch": 1.8421076334158974,
      "grad_norm": 1.491155743598938,
      "learning_rate": 1.531084722844034e-06,
      "loss": 1.4939,
      "step": 10226
    },
    {
      "epoch": 1.8422877730240936,
      "grad_norm": 1.6226614713668823,
      "learning_rate": 1.5276089903099977e-06,
      "loss": 1.4909,
      "step": 10227
    },
    {
      "epoch": 1.84246791263229,
      "grad_norm": 1.465377926826477,
      "learning_rate": 1.5241371462075548e-06,
      "loss": 1.3031,
      "step": 10228
    },
    {
      "epoch": 1.8426480522404862,
      "grad_norm": 1.575509786605835,
      "learning_rate": 1.520669190815227e-06,
      "loss": 1.5349,
      "step": 10229
    },
    {
      "epoch": 1.8428281918486826,
      "grad_norm": 1.548424243927002,
      "learning_rate": 1.5172051244111972e-06,
      "loss": 1.3707,
      "step": 10230
    },
    {
      "epoch": 1.843008331456879,
      "grad_norm": 1.5996744632720947,
      "learning_rate": 1.5137449472733546e-06,
      "loss": 1.6673,
      "step": 10231
    },
    {
      "epoch": 1.8431884710650754,
      "grad_norm": 1.638129472732544,
      "learning_rate": 1.5102886596792821e-06,
      "loss": 1.4558,
      "step": 10232
    },
    {
      "epoch": 1.8433686106732718,
      "grad_norm": 1.462013840675354,
      "learning_rate": 1.5068362619062304e-06,
      "loss": 1.421,
      "step": 10233
    },
    {
      "epoch": 1.8435487502814683,
      "grad_norm": 1.371056318283081,
      "learning_rate": 1.5033877542311558e-06,
      "loss": 1.3634,
      "step": 10234
    },
    {
      "epoch": 1.8437288898896644,
      "grad_norm": 1.4632117748260498,
      "learning_rate": 1.4999431369306872e-06,
      "loss": 1.3429,
      "step": 10235
    },
    {
      "epoch": 1.8439090294978608,
      "grad_norm": 1.5581333637237549,
      "learning_rate": 1.4965024102811642e-06,
      "loss": 1.3619,
      "step": 10236
    },
    {
      "epoch": 1.8440891691060572,
      "grad_norm": 1.3980218172073364,
      "learning_rate": 1.4930655745585886e-06,
      "loss": 1.2804,
      "step": 10237
    },
    {
      "epoch": 1.8442693087142534,
      "grad_norm": 1.5534332990646362,
      "learning_rate": 1.4896326300386676e-06,
      "loss": 1.4238,
      "step": 10238
    },
    {
      "epoch": 1.8444494483224498,
      "grad_norm": 1.6386970281600952,
      "learning_rate": 1.4862035769967864e-06,
      "loss": 1.3799,
      "step": 10239
    },
    {
      "epoch": 1.8446295879306462,
      "grad_norm": 1.5318262577056885,
      "learning_rate": 1.4827784157080248e-06,
      "loss": 1.4185,
      "step": 10240
    },
    {
      "epoch": 1.8448097275388426,
      "grad_norm": 1.6147505044937134,
      "learning_rate": 1.4793571464471522e-06,
      "loss": 1.4969,
      "step": 10241
    },
    {
      "epoch": 1.844989867147039,
      "grad_norm": 1.628230333328247,
      "learning_rate": 1.4759397694886157e-06,
      "loss": 1.448,
      "step": 10242
    },
    {
      "epoch": 1.8451700067552355,
      "grad_norm": 1.668972134590149,
      "learning_rate": 1.4725262851065513e-06,
      "loss": 1.4275,
      "step": 10243
    },
    {
      "epoch": 1.8453501463634316,
      "grad_norm": 1.4975521564483643,
      "learning_rate": 1.4691166935747846e-06,
      "loss": 1.3199,
      "step": 10244
    },
    {
      "epoch": 1.845530285971628,
      "grad_norm": 1.5816705226898193,
      "learning_rate": 1.4657109951668468e-06,
      "loss": 1.5044,
      "step": 10245
    },
    {
      "epoch": 1.8457104255798242,
      "grad_norm": 1.6212525367736816,
      "learning_rate": 1.4623091901559306e-06,
      "loss": 1.4703,
      "step": 10246
    },
    {
      "epoch": 1.8458905651880206,
      "grad_norm": 1.731925129890442,
      "learning_rate": 1.458911278814934e-06,
      "loss": 1.3719,
      "step": 10247
    },
    {
      "epoch": 1.846070704796217,
      "grad_norm": 1.4780364036560059,
      "learning_rate": 1.455517261416428e-06,
      "loss": 1.0591,
      "step": 10248
    },
    {
      "epoch": 1.8462508444044134,
      "grad_norm": 1.4438854455947876,
      "learning_rate": 1.452127138232673e-06,
      "loss": 1.1401,
      "step": 10249
    },
    {
      "epoch": 1.8464309840126099,
      "grad_norm": 1.4837759733200073,
      "learning_rate": 1.4487409095356398e-06,
      "loss": 1.2347,
      "step": 10250
    },
    {
      "epoch": 1.8466111236208063,
      "grad_norm": 1.4581249952316284,
      "learning_rate": 1.4453585755969667e-06,
      "loss": 1.2771,
      "step": 10251
    },
    {
      "epoch": 1.8467912632290024,
      "grad_norm": 1.3369414806365967,
      "learning_rate": 1.441980136687976e-06,
      "loss": 1.8201,
      "step": 10252
    },
    {
      "epoch": 1.8469714028371989,
      "grad_norm": 1.4064689874649048,
      "learning_rate": 1.4386055930796837e-06,
      "loss": 1.9002,
      "step": 10253
    },
    {
      "epoch": 1.847151542445395,
      "grad_norm": 1.3935283422470093,
      "learning_rate": 1.4352349450428015e-06,
      "loss": 1.8927,
      "step": 10254
    },
    {
      "epoch": 1.8473316820535914,
      "grad_norm": 1.5412906408309937,
      "learning_rate": 1.431868192847713e-06,
      "loss": 1.9827,
      "step": 10255
    },
    {
      "epoch": 1.8475118216617878,
      "grad_norm": 1.3270776271820068,
      "learning_rate": 1.4285053367645074e-06,
      "loss": 1.682,
      "step": 10256
    },
    {
      "epoch": 1.8476919612699843,
      "grad_norm": 1.4654545783996582,
      "learning_rate": 1.4251463770629414e-06,
      "loss": 1.7095,
      "step": 10257
    },
    {
      "epoch": 1.8478721008781807,
      "grad_norm": 1.601901888847351,
      "learning_rate": 1.4217913140124661e-06,
      "loss": 1.8171,
      "step": 10258
    },
    {
      "epoch": 1.848052240486377,
      "grad_norm": 1.5529544353485107,
      "learning_rate": 1.4184401478822385e-06,
      "loss": 1.6876,
      "step": 10259
    },
    {
      "epoch": 1.8482323800945732,
      "grad_norm": 1.6309280395507812,
      "learning_rate": 1.4150928789410767e-06,
      "loss": 1.7556,
      "step": 10260
    },
    {
      "epoch": 1.8484125197027697,
      "grad_norm": 1.6407078504562378,
      "learning_rate": 1.4117495074574993e-06,
      "loss": 1.817,
      "step": 10261
    },
    {
      "epoch": 1.8485926593109658,
      "grad_norm": 1.434386968612671,
      "learning_rate": 1.4084100336997085e-06,
      "loss": 1.4728,
      "step": 10262
    },
    {
      "epoch": 1.8487727989191622,
      "grad_norm": 1.4617233276367188,
      "learning_rate": 1.4050744579356011e-06,
      "loss": 1.6581,
      "step": 10263
    },
    {
      "epoch": 1.8489529385273586,
      "grad_norm": 1.372172236442566,
      "learning_rate": 1.4017427804327466e-06,
      "loss": 1.4673,
      "step": 10264
    },
    {
      "epoch": 1.849133078135555,
      "grad_norm": 1.4211292266845703,
      "learning_rate": 1.3984150014584197e-06,
      "loss": 1.3121,
      "step": 10265
    },
    {
      "epoch": 1.8493132177437515,
      "grad_norm": 1.3662176132202148,
      "learning_rate": 1.3950911212795682e-06,
      "loss": 1.2187,
      "step": 10266
    },
    {
      "epoch": 1.8494933573519479,
      "grad_norm": 1.4869539737701416,
      "learning_rate": 1.3917711401628231e-06,
      "loss": 1.51,
      "step": 10267
    },
    {
      "epoch": 1.8496734969601443,
      "grad_norm": 1.4698007106781006,
      "learning_rate": 1.388455058374527e-06,
      "loss": 1.5938,
      "step": 10268
    },
    {
      "epoch": 1.8498536365683405,
      "grad_norm": 1.5569158792495728,
      "learning_rate": 1.3851428761806895e-06,
      "loss": 1.6096,
      "step": 10269
    },
    {
      "epoch": 1.8500337761765369,
      "grad_norm": 1.3702348470687866,
      "learning_rate": 1.3818345938470089e-06,
      "loss": 1.1617,
      "step": 10270
    },
    {
      "epoch": 1.850213915784733,
      "grad_norm": 1.372243046760559,
      "learning_rate": 1.3785302116388786e-06,
      "loss": 1.2194,
      "step": 10271
    },
    {
      "epoch": 1.8503940553929294,
      "grad_norm": 1.5003570318222046,
      "learning_rate": 1.3752297298213646e-06,
      "loss": 1.4444,
      "step": 10272
    },
    {
      "epoch": 1.8505741950011259,
      "grad_norm": 1.478769063949585,
      "learning_rate": 1.371933148659249e-06,
      "loss": 1.4254,
      "step": 10273
    },
    {
      "epoch": 1.8507543346093223,
      "grad_norm": 1.5099936723709106,
      "learning_rate": 1.368640468416965e-06,
      "loss": 1.4137,
      "step": 10274
    },
    {
      "epoch": 1.8509344742175187,
      "grad_norm": 1.5282105207443237,
      "learning_rate": 1.3653516893586571e-06,
      "loss": 1.44,
      "step": 10275
    },
    {
      "epoch": 1.851114613825715,
      "grad_norm": 1.4914833307266235,
      "learning_rate": 1.3620668117481472e-06,
      "loss": 1.4503,
      "step": 10276
    },
    {
      "epoch": 1.8512947534339113,
      "grad_norm": 1.5417181253433228,
      "learning_rate": 1.3587858358489413e-06,
      "loss": 1.3782,
      "step": 10277
    },
    {
      "epoch": 1.8514748930421077,
      "grad_norm": 1.473596215248108,
      "learning_rate": 1.3555087619242456e-06,
      "loss": 1.3942,
      "step": 10278
    },
    {
      "epoch": 1.8516550326503038,
      "grad_norm": 1.4994829893112183,
      "learning_rate": 1.3522355902369444e-06,
      "loss": 1.383,
      "step": 10279
    },
    {
      "epoch": 1.8518351722585003,
      "grad_norm": 1.4472023248672485,
      "learning_rate": 1.3489663210496105e-06,
      "loss": 1.3407,
      "step": 10280
    },
    {
      "epoch": 1.8520153118666967,
      "grad_norm": 1.6378897428512573,
      "learning_rate": 1.345700954624496e-06,
      "loss": 1.472,
      "step": 10281
    },
    {
      "epoch": 1.852195451474893,
      "grad_norm": 1.3848503828048706,
      "learning_rate": 1.3424394912235572e-06,
      "loss": 1.3026,
      "step": 10282
    },
    {
      "epoch": 1.8523755910830895,
      "grad_norm": 1.7458934783935547,
      "learning_rate": 1.3391819311084243e-06,
      "loss": 1.6856,
      "step": 10283
    },
    {
      "epoch": 1.8525557306912859,
      "grad_norm": 1.5376888513565063,
      "learning_rate": 1.33592827454041e-06,
      "loss": 1.3225,
      "step": 10284
    },
    {
      "epoch": 1.852735870299482,
      "grad_norm": 1.5704020261764526,
      "learning_rate": 1.3326785217805282e-06,
      "loss": 1.6235,
      "step": 10285
    },
    {
      "epoch": 1.8529160099076785,
      "grad_norm": 1.5847327709197998,
      "learning_rate": 1.3294326730894702e-06,
      "loss": 1.411,
      "step": 10286
    },
    {
      "epoch": 1.8530961495158746,
      "grad_norm": 1.4361884593963623,
      "learning_rate": 1.3261907287276165e-06,
      "loss": 1.2121,
      "step": 10287
    },
    {
      "epoch": 1.853276289124071,
      "grad_norm": 1.5476760864257812,
      "learning_rate": 1.322952688955037e-06,
      "loss": 1.3703,
      "step": 10288
    },
    {
      "epoch": 1.8534564287322675,
      "grad_norm": 1.4903324842453003,
      "learning_rate": 1.3197185540314794e-06,
      "loss": 1.4153,
      "step": 10289
    },
    {
      "epoch": 1.8536365683404639,
      "grad_norm": 1.6558328866958618,
      "learning_rate": 1.3164883242163917e-06,
      "loss": 1.617,
      "step": 10290
    },
    {
      "epoch": 1.8538167079486603,
      "grad_norm": 1.6648749113082886,
      "learning_rate": 1.313261999768889e-06,
      "loss": 1.4665,
      "step": 10291
    },
    {
      "epoch": 1.8539968475568567,
      "grad_norm": 1.6215368509292603,
      "learning_rate": 1.3100395809478027e-06,
      "loss": 1.5125,
      "step": 10292
    },
    {
      "epoch": 1.854176987165053,
      "grad_norm": 1.5777924060821533,
      "learning_rate": 1.306821068011621e-06,
      "loss": 1.4679,
      "step": 10293
    },
    {
      "epoch": 1.8543571267732493,
      "grad_norm": 1.4901269674301147,
      "learning_rate": 1.3036064612185372e-06,
      "loss": 1.2107,
      "step": 10294
    },
    {
      "epoch": 1.8545372663814457,
      "grad_norm": 1.6526262760162354,
      "learning_rate": 1.3003957608264284e-06,
      "loss": 1.6491,
      "step": 10295
    },
    {
      "epoch": 1.8547174059896419,
      "grad_norm": 1.4791837930679321,
      "learning_rate": 1.2971889670928439e-06,
      "loss": 1.2922,
      "step": 10296
    },
    {
      "epoch": 1.8548975455978383,
      "grad_norm": 1.5801347494125366,
      "learning_rate": 1.2939860802750447e-06,
      "loss": 1.4372,
      "step": 10297
    },
    {
      "epoch": 1.8550776852060347,
      "grad_norm": 1.7281008958816528,
      "learning_rate": 1.2907871006299532e-06,
      "loss": 1.3421,
      "step": 10298
    },
    {
      "epoch": 1.855257824814231,
      "grad_norm": 1.4185928106307983,
      "learning_rate": 1.2875920284141973e-06,
      "loss": 1.2514,
      "step": 10299
    },
    {
      "epoch": 1.8554379644224275,
      "grad_norm": 1.4465584754943848,
      "learning_rate": 1.2844008638840777e-06,
      "loss": 1.2245,
      "step": 10300
    },
    {
      "epoch": 1.8556181040306239,
      "grad_norm": 1.5240312814712524,
      "learning_rate": 1.2812136072955948e-06,
      "loss": 1.5414,
      "step": 10301
    },
    {
      "epoch": 1.85579824363882,
      "grad_norm": 1.5047955513000488,
      "learning_rate": 1.278030258904428e-06,
      "loss": 1.7555,
      "step": 10302
    },
    {
      "epoch": 1.8559783832470165,
      "grad_norm": 1.3868755102157593,
      "learning_rate": 1.2748508189659447e-06,
      "loss": 1.8682,
      "step": 10303
    },
    {
      "epoch": 1.8561585228552127,
      "grad_norm": 1.3334591388702393,
      "learning_rate": 1.2716752877351912e-06,
      "loss": 1.6443,
      "step": 10304
    },
    {
      "epoch": 1.856338662463409,
      "grad_norm": 1.424626111984253,
      "learning_rate": 1.268503665466908e-06,
      "loss": 1.8996,
      "step": 10305
    },
    {
      "epoch": 1.8565188020716055,
      "grad_norm": 1.4766196012496948,
      "learning_rate": 1.2653359524155361e-06,
      "loss": 1.8598,
      "step": 10306
    },
    {
      "epoch": 1.8566989416798019,
      "grad_norm": 1.4316827058792114,
      "learning_rate": 1.2621721488351668e-06,
      "loss": 1.4301,
      "step": 10307
    },
    {
      "epoch": 1.8568790812879983,
      "grad_norm": 1.541999340057373,
      "learning_rate": 1.2590122549796134e-06,
      "loss": 1.6946,
      "step": 10308
    },
    {
      "epoch": 1.8570592208961947,
      "grad_norm": 1.4967455863952637,
      "learning_rate": 1.255856271102346e-06,
      "loss": 1.6284,
      "step": 10309
    },
    {
      "epoch": 1.8572393605043909,
      "grad_norm": 1.6613128185272217,
      "learning_rate": 1.2527041974565556e-06,
      "loss": 1.778,
      "step": 10310
    },
    {
      "epoch": 1.8574195001125873,
      "grad_norm": 2.0601134300231934,
      "learning_rate": 1.2495560342950851e-06,
      "loss": 2.1886,
      "step": 10311
    },
    {
      "epoch": 1.8575996397207835,
      "grad_norm": 1.4511455297470093,
      "learning_rate": 1.2464117818704878e-06,
      "loss": 1.3462,
      "step": 10312
    },
    {
      "epoch": 1.8577797793289799,
      "grad_norm": 1.3765714168548584,
      "learning_rate": 1.2432714404349954e-06,
      "loss": 1.4886,
      "step": 10313
    },
    {
      "epoch": 1.8579599189371763,
      "grad_norm": 1.4717071056365967,
      "learning_rate": 1.2401350102405063e-06,
      "loss": 1.4262,
      "step": 10314
    },
    {
      "epoch": 1.8581400585453727,
      "grad_norm": 1.4502239227294922,
      "learning_rate": 1.237002491538647e-06,
      "loss": 1.457,
      "step": 10315
    },
    {
      "epoch": 1.858320198153569,
      "grad_norm": 1.3519161939620972,
      "learning_rate": 1.2338738845806941e-06,
      "loss": 1.3211,
      "step": 10316
    },
    {
      "epoch": 1.8585003377617655,
      "grad_norm": 1.4037913084030151,
      "learning_rate": 1.2307491896176248e-06,
      "loss": 1.3838,
      "step": 10317
    },
    {
      "epoch": 1.8586804773699617,
      "grad_norm": 1.3911057710647583,
      "learning_rate": 1.2276284069001053e-06,
      "loss": 1.4691,
      "step": 10318
    },
    {
      "epoch": 1.858860616978158,
      "grad_norm": 1.5189173221588135,
      "learning_rate": 1.2245115366784687e-06,
      "loss": 1.403,
      "step": 10319
    },
    {
      "epoch": 1.8590407565863543,
      "grad_norm": 1.36824369430542,
      "learning_rate": 1.221398579202765e-06,
      "loss": 1.318,
      "step": 10320
    },
    {
      "epoch": 1.8592208961945507,
      "grad_norm": 1.467315912246704,
      "learning_rate": 1.2182895347227052e-06,
      "loss": 1.4373,
      "step": 10321
    },
    {
      "epoch": 1.859401035802747,
      "grad_norm": 1.5092490911483765,
      "learning_rate": 1.2151844034877013e-06,
      "loss": 1.4128,
      "step": 10322
    },
    {
      "epoch": 1.8595811754109435,
      "grad_norm": 1.3840872049331665,
      "learning_rate": 1.2120831857468373e-06,
      "loss": 1.2558,
      "step": 10323
    },
    {
      "epoch": 1.8597613150191399,
      "grad_norm": 1.416157603263855,
      "learning_rate": 1.208985881748903e-06,
      "loss": 1.2283,
      "step": 10324
    },
    {
      "epoch": 1.8599414546273363,
      "grad_norm": 1.3469048738479614,
      "learning_rate": 1.2058924917423498e-06,
      "loss": 1.3275,
      "step": 10325
    },
    {
      "epoch": 1.8601215942355327,
      "grad_norm": 1.4587441682815552,
      "learning_rate": 1.20280301597534e-06,
      "loss": 1.3504,
      "step": 10326
    },
    {
      "epoch": 1.8603017338437289,
      "grad_norm": 1.4653880596160889,
      "learning_rate": 1.1997174546957034e-06,
      "loss": 1.6289,
      "step": 10327
    },
    {
      "epoch": 1.8604818734519253,
      "grad_norm": 1.6527965068817139,
      "learning_rate": 1.1966358081509588e-06,
      "loss": 1.3499,
      "step": 10328
    },
    {
      "epoch": 1.8606620130601215,
      "grad_norm": 1.6877119541168213,
      "learning_rate": 1.1935580765883248e-06,
      "loss": 1.4594,
      "step": 10329
    },
    {
      "epoch": 1.8608421526683179,
      "grad_norm": 1.4581581354141235,
      "learning_rate": 1.1904842602546873e-06,
      "loss": 1.3978,
      "step": 10330
    },
    {
      "epoch": 1.8610222922765143,
      "grad_norm": 1.4317772388458252,
      "learning_rate": 1.1874143593966214e-06,
      "loss": 1.1962,
      "step": 10331
    },
    {
      "epoch": 1.8612024318847107,
      "grad_norm": 1.6028050184249878,
      "learning_rate": 1.1843483742604022e-06,
      "loss": 1.3752,
      "step": 10332
    },
    {
      "epoch": 1.861382571492907,
      "grad_norm": 1.6119439601898193,
      "learning_rate": 1.1812863050919831e-06,
      "loss": 1.5033,
      "step": 10333
    },
    {
      "epoch": 1.8615627111011035,
      "grad_norm": 1.5995889902114868,
      "learning_rate": 1.1782281521369954e-06,
      "loss": 1.3692,
      "step": 10334
    },
    {
      "epoch": 1.8617428507092997,
      "grad_norm": 1.4851839542388916,
      "learning_rate": 1.1751739156407649e-06,
      "loss": 1.3061,
      "step": 10335
    },
    {
      "epoch": 1.861922990317496,
      "grad_norm": 1.6096994876861572,
      "learning_rate": 1.1721235958483013e-06,
      "loss": 1.5765,
      "step": 10336
    },
    {
      "epoch": 1.8621031299256923,
      "grad_norm": 1.521888256072998,
      "learning_rate": 1.1690771930042975e-06,
      "loss": 1.4017,
      "step": 10337
    },
    {
      "epoch": 1.8622832695338887,
      "grad_norm": 1.445374846458435,
      "learning_rate": 1.1660347073531475e-06,
      "loss": 1.2434,
      "step": 10338
    },
    {
      "epoch": 1.862463409142085,
      "grad_norm": 1.6131882667541504,
      "learning_rate": 1.1629961391388998e-06,
      "loss": 1.6562,
      "step": 10339
    },
    {
      "epoch": 1.8626435487502815,
      "grad_norm": 1.5619280338287354,
      "learning_rate": 1.15996148860531e-06,
      "loss": 1.5297,
      "step": 10340
    },
    {
      "epoch": 1.862823688358478,
      "grad_norm": 1.5652689933776855,
      "learning_rate": 1.1569307559958276e-06,
      "loss": 1.3402,
      "step": 10341
    },
    {
      "epoch": 1.8630038279666743,
      "grad_norm": 1.4430828094482422,
      "learning_rate": 1.153903941553558e-06,
      "loss": 1.288,
      "step": 10342
    },
    {
      "epoch": 1.8631839675748705,
      "grad_norm": 1.5014451742172241,
      "learning_rate": 1.1508810455213348e-06,
      "loss": 1.4052,
      "step": 10343
    },
    {
      "epoch": 1.8633641071830669,
      "grad_norm": 1.6292515993118286,
      "learning_rate": 1.1478620681416363e-06,
      "loss": 1.4215,
      "step": 10344
    },
    {
      "epoch": 1.863544246791263,
      "grad_norm": 1.6442686319351196,
      "learning_rate": 1.1448470096566466e-06,
      "loss": 1.5864,
      "step": 10345
    },
    {
      "epoch": 1.8637243863994595,
      "grad_norm": 1.4768873453140259,
      "learning_rate": 1.1418358703082278e-06,
      "loss": 1.2326,
      "step": 10346
    },
    {
      "epoch": 1.8639045260076559,
      "grad_norm": 1.530089020729065,
      "learning_rate": 1.138828650337942e-06,
      "loss": 1.2744,
      "step": 10347
    },
    {
      "epoch": 1.8640846656158523,
      "grad_norm": 2.0019547939300537,
      "learning_rate": 1.1358253499870241e-06,
      "loss": 1.481,
      "step": 10348
    },
    {
      "epoch": 1.8642648052240487,
      "grad_norm": 1.6834994554519653,
      "learning_rate": 1.1328259694963982e-06,
      "loss": 1.5595,
      "step": 10349
    },
    {
      "epoch": 1.864444944832245,
      "grad_norm": 1.7519543170928955,
      "learning_rate": 1.1298305091066664e-06,
      "loss": 1.4956,
      "step": 10350
    },
    {
      "epoch": 1.8646250844404415,
      "grad_norm": 1.3320525884628296,
      "learning_rate": 1.1268389690581306e-06,
      "loss": 1.073,
      "step": 10351
    },
    {
      "epoch": 1.8648052240486377,
      "grad_norm": 1.2492834329605103,
      "learning_rate": 1.1238513495907654e-06,
      "loss": 1.6528,
      "step": 10352
    },
    {
      "epoch": 1.864985363656834,
      "grad_norm": 1.3353514671325684,
      "learning_rate": 1.1208676509442407e-06,
      "loss": 1.6936,
      "step": 10353
    },
    {
      "epoch": 1.8651655032650303,
      "grad_norm": 1.387160301208496,
      "learning_rate": 1.1178878733579035e-06,
      "loss": 1.7173,
      "step": 10354
    },
    {
      "epoch": 1.8653456428732267,
      "grad_norm": 1.4492627382278442,
      "learning_rate": 1.1149120170707905e-06,
      "loss": 1.81,
      "step": 10355
    },
    {
      "epoch": 1.865525782481423,
      "grad_norm": 1.3958799839019775,
      "learning_rate": 1.111940082321622e-06,
      "loss": 1.7279,
      "step": 10356
    },
    {
      "epoch": 1.8657059220896195,
      "grad_norm": 1.4937512874603271,
      "learning_rate": 1.108972069348807e-06,
      "loss": 1.7007,
      "step": 10357
    },
    {
      "epoch": 1.865886061697816,
      "grad_norm": 1.4598029851913452,
      "learning_rate": 1.1060079783904443e-06,
      "loss": 1.6331,
      "step": 10358
    },
    {
      "epoch": 1.8660662013060123,
      "grad_norm": 1.5105829238891602,
      "learning_rate": 1.1030478096843045e-06,
      "loss": 1.7168,
      "step": 10359
    },
    {
      "epoch": 1.8662463409142085,
      "grad_norm": 1.688501000404358,
      "learning_rate": 1.1000915634678532e-06,
      "loss": 1.7834,
      "step": 10360
    },
    {
      "epoch": 1.866426480522405,
      "grad_norm": 1.9013447761535645,
      "learning_rate": 1.0971392399782343e-06,
      "loss": 2.3872,
      "step": 10361
    },
    {
      "epoch": 1.866606620130601,
      "grad_norm": 1.5750701427459717,
      "learning_rate": 1.0941908394522914e-06,
      "loss": 1.6268,
      "step": 10362
    },
    {
      "epoch": 1.8667867597387975,
      "grad_norm": 1.4531400203704834,
      "learning_rate": 1.0912463621265356e-06,
      "loss": 1.7259,
      "step": 10363
    },
    {
      "epoch": 1.866966899346994,
      "grad_norm": 1.500709056854248,
      "learning_rate": 1.0883058082371721e-06,
      "loss": 1.4662,
      "step": 10364
    },
    {
      "epoch": 1.8671470389551903,
      "grad_norm": 1.5037617683410645,
      "learning_rate": 1.0853691780200904e-06,
      "loss": 1.3406,
      "step": 10365
    },
    {
      "epoch": 1.8673271785633867,
      "grad_norm": 1.3896816968917847,
      "learning_rate": 1.0824364717108737e-06,
      "loss": 1.3752,
      "step": 10366
    },
    {
      "epoch": 1.867507318171583,
      "grad_norm": 1.3719779253005981,
      "learning_rate": 1.079507689544773e-06,
      "loss": 1.3369,
      "step": 10367
    },
    {
      "epoch": 1.8676874577797793,
      "grad_norm": 1.5353895425796509,
      "learning_rate": 1.0765828317567395e-06,
      "loss": 1.6057,
      "step": 10368
    },
    {
      "epoch": 1.8678675973879757,
      "grad_norm": 1.48717200756073,
      "learning_rate": 1.0736618985814018e-06,
      "loss": 1.3854,
      "step": 10369
    },
    {
      "epoch": 1.8680477369961719,
      "grad_norm": 1.3276771306991577,
      "learning_rate": 1.070744890253067e-06,
      "loss": 1.3142,
      "step": 10370
    },
    {
      "epoch": 1.8682278766043683,
      "grad_norm": 1.4071564674377441,
      "learning_rate": 1.0678318070057592e-06,
      "loss": 1.3445,
      "step": 10371
    },
    {
      "epoch": 1.8684080162125647,
      "grad_norm": 1.3767555952072144,
      "learning_rate": 1.0649226490731411e-06,
      "loss": 1.2885,
      "step": 10372
    },
    {
      "epoch": 1.868588155820761,
      "grad_norm": 1.540975570678711,
      "learning_rate": 1.0620174166885931e-06,
      "loss": 1.5272,
      "step": 10373
    },
    {
      "epoch": 1.8687682954289575,
      "grad_norm": 1.39052414894104,
      "learning_rate": 1.0591161100851676e-06,
      "loss": 1.294,
      "step": 10374
    },
    {
      "epoch": 1.868948435037154,
      "grad_norm": 1.4286513328552246,
      "learning_rate": 1.0562187294956172e-06,
      "loss": 1.2455,
      "step": 10375
    },
    {
      "epoch": 1.86912857464535,
      "grad_norm": 1.4543054103851318,
      "learning_rate": 1.0533252751523559e-06,
      "loss": 1.4353,
      "step": 10376
    },
    {
      "epoch": 1.8693087142535465,
      "grad_norm": 1.527773141860962,
      "learning_rate": 1.0504357472874981e-06,
      "loss": 1.3891,
      "step": 10377
    },
    {
      "epoch": 1.869488853861743,
      "grad_norm": 1.5044718980789185,
      "learning_rate": 1.0475501461328473e-06,
      "loss": 1.4894,
      "step": 10378
    },
    {
      "epoch": 1.869668993469939,
      "grad_norm": 1.5600088834762573,
      "learning_rate": 1.044668471919874e-06,
      "loss": 1.5145,
      "step": 10379
    },
    {
      "epoch": 1.8698491330781355,
      "grad_norm": 1.5081301927566528,
      "learning_rate": 1.0417907248797542e-06,
      "loss": 1.4865,
      "step": 10380
    },
    {
      "epoch": 1.870029272686332,
      "grad_norm": 1.356370210647583,
      "learning_rate": 1.0389169052433368e-06,
      "loss": 1.221,
      "step": 10381
    },
    {
      "epoch": 1.8702094122945283,
      "grad_norm": 1.578597903251648,
      "learning_rate": 1.0360470132411593e-06,
      "loss": 1.3449,
      "step": 10382
    },
    {
      "epoch": 1.8703895519027247,
      "grad_norm": 1.7410928010940552,
      "learning_rate": 1.0331810491034434e-06,
      "loss": 1.5618,
      "step": 10383
    },
    {
      "epoch": 1.8705696915109211,
      "grad_norm": 1.689487099647522,
      "learning_rate": 1.0303190130600825e-06,
      "loss": 1.5637,
      "step": 10384
    },
    {
      "epoch": 1.8707498311191173,
      "grad_norm": 1.6730750799179077,
      "learning_rate": 1.0274609053406824e-06,
      "loss": 1.5663,
      "step": 10385
    },
    {
      "epoch": 1.8709299707273137,
      "grad_norm": 1.58219313621521,
      "learning_rate": 1.0246067261745152e-06,
      "loss": 1.424,
      "step": 10386
    },
    {
      "epoch": 1.87111011033551,
      "grad_norm": 1.6409006118774414,
      "learning_rate": 1.0217564757905419e-06,
      "loss": 1.5234,
      "step": 10387
    },
    {
      "epoch": 1.8712902499437063,
      "grad_norm": 1.4316788911819458,
      "learning_rate": 1.0189101544174018e-06,
      "loss": 1.3608,
      "step": 10388
    },
    {
      "epoch": 1.8714703895519027,
      "grad_norm": 1.677889108657837,
      "learning_rate": 1.0160677622834401e-06,
      "loss": 1.5967,
      "step": 10389
    },
    {
      "epoch": 1.871650529160099,
      "grad_norm": 1.6832510232925415,
      "learning_rate": 1.0132292996166582e-06,
      "loss": 1.2584,
      "step": 10390
    },
    {
      "epoch": 1.8718306687682955,
      "grad_norm": 1.5509663820266724,
      "learning_rate": 1.010394766644762e-06,
      "loss": 1.2566,
      "step": 10391
    },
    {
      "epoch": 1.872010808376492,
      "grad_norm": 1.6645439863204956,
      "learning_rate": 1.0075641635951372e-06,
      "loss": 1.6477,
      "step": 10392
    },
    {
      "epoch": 1.872190947984688,
      "grad_norm": 1.3653062582015991,
      "learning_rate": 1.0047374906948516e-06,
      "loss": 1.1367,
      "step": 10393
    },
    {
      "epoch": 1.8723710875928845,
      "grad_norm": 1.8081492185592651,
      "learning_rate": 1.0019147481706625e-06,
      "loss": 1.7334,
      "step": 10394
    },
    {
      "epoch": 1.8725512272010807,
      "grad_norm": 1.5390008687973022,
      "learning_rate": 9.990959362490003e-07,
      "loss": 1.3011,
      "step": 10395
    },
    {
      "epoch": 1.872731366809277,
      "grad_norm": 1.559303641319275,
      "learning_rate": 9.96281055156001e-07,
      "loss": 1.2945,
      "step": 10396
    },
    {
      "epoch": 1.8729115064174735,
      "grad_norm": 1.6995283365249634,
      "learning_rate": 9.93470105117461e-07,
      "loss": 1.269,
      "step": 10397
    },
    {
      "epoch": 1.87309164602567,
      "grad_norm": 1.572171688079834,
      "learning_rate": 9.906630863588784e-07,
      "loss": 1.5208,
      "step": 10398
    },
    {
      "epoch": 1.8732717856338663,
      "grad_norm": 1.4155441522598267,
      "learning_rate": 9.878599991054337e-07,
      "loss": 1.1741,
      "step": 10399
    },
    {
      "epoch": 1.8734519252420627,
      "grad_norm": 1.5453004837036133,
      "learning_rate": 9.850608435819864e-07,
      "loss": 1.189,
      "step": 10400
    },
    {
      "epoch": 1.873632064850259,
      "grad_norm": 1.5081183910369873,
      "learning_rate": 9.822656200130842e-07,
      "loss": 1.2334,
      "step": 10401
    },
    {
      "epoch": 1.8738122044584553,
      "grad_norm": 1.2858686447143555,
      "learning_rate": 9.79474328622959e-07,
      "loss": 1.4632,
      "step": 10402
    },
    {
      "epoch": 1.8739923440666515,
      "grad_norm": 1.393790602684021,
      "learning_rate": 9.76686969635532e-07,
      "loss": 1.8111,
      "step": 10403
    },
    {
      "epoch": 1.874172483674848,
      "grad_norm": 1.4437096118927002,
      "learning_rate": 9.739035432743903e-07,
      "loss": 2.123,
      "step": 10404
    },
    {
      "epoch": 1.8743526232830443,
      "grad_norm": 1.3957126140594482,
      "learning_rate": 9.711240497628281e-07,
      "loss": 1.5462,
      "step": 10405
    },
    {
      "epoch": 1.8745327628912407,
      "grad_norm": 1.5514625310897827,
      "learning_rate": 9.683484893238115e-07,
      "loss": 2.0841,
      "step": 10406
    },
    {
      "epoch": 1.8747129024994371,
      "grad_norm": 1.3577128648757935,
      "learning_rate": 9.655768621799954e-07,
      "loss": 1.4013,
      "step": 10407
    },
    {
      "epoch": 1.8748930421076335,
      "grad_norm": 1.664527416229248,
      "learning_rate": 9.628091685537244e-07,
      "loss": 2.1026,
      "step": 10408
    },
    {
      "epoch": 1.87507318171583,
      "grad_norm": 1.5408542156219482,
      "learning_rate": 9.600454086670152e-07,
      "loss": 1.7735,
      "step": 10409
    },
    {
      "epoch": 1.8752533213240261,
      "grad_norm": 1.6988085508346558,
      "learning_rate": 9.572855827415794e-07,
      "loss": 1.8174,
      "step": 10410
    },
    {
      "epoch": 1.8754334609322225,
      "grad_norm": 1.9394913911819458,
      "learning_rate": 9.54529690998801e-07,
      "loss": 1.761,
      "step": 10411
    },
    {
      "epoch": 1.8756136005404187,
      "grad_norm": 1.610665202140808,
      "learning_rate": 9.517777336597644e-07,
      "loss": 1.6989,
      "step": 10412
    },
    {
      "epoch": 1.875793740148615,
      "grad_norm": 1.3886268138885498,
      "learning_rate": 9.490297109452262e-07,
      "loss": 1.5696,
      "step": 10413
    },
    {
      "epoch": 1.8759738797568115,
      "grad_norm": 1.6117889881134033,
      "learning_rate": 9.462856230756323e-07,
      "loss": 1.4527,
      "step": 10414
    },
    {
      "epoch": 1.876154019365008,
      "grad_norm": 1.267108678817749,
      "learning_rate": 9.435454702711177e-07,
      "loss": 1.2498,
      "step": 10415
    },
    {
      "epoch": 1.8763341589732043,
      "grad_norm": 1.3357423543930054,
      "learning_rate": 9.408092527514845e-07,
      "loss": 1.2348,
      "step": 10416
    },
    {
      "epoch": 1.8765142985814007,
      "grad_norm": 1.5065866708755493,
      "learning_rate": 9.380769707362403e-07,
      "loss": 1.544,
      "step": 10417
    },
    {
      "epoch": 1.876694438189597,
      "grad_norm": 1.4108260869979858,
      "learning_rate": 9.353486244445653e-07,
      "loss": 1.3682,
      "step": 10418
    },
    {
      "epoch": 1.8768745777977933,
      "grad_norm": 1.4301656484603882,
      "learning_rate": 9.326242140953234e-07,
      "loss": 1.2654,
      "step": 10419
    },
    {
      "epoch": 1.8770547174059895,
      "grad_norm": 1.4730911254882812,
      "learning_rate": 9.299037399070676e-07,
      "loss": 1.3764,
      "step": 10420
    },
    {
      "epoch": 1.877234857014186,
      "grad_norm": 1.4049005508422852,
      "learning_rate": 9.271872020980288e-07,
      "loss": 1.5352,
      "step": 10421
    },
    {
      "epoch": 1.8774149966223823,
      "grad_norm": 1.3225533962249756,
      "learning_rate": 9.244746008861327e-07,
      "loss": 1.1391,
      "step": 10422
    },
    {
      "epoch": 1.8775951362305787,
      "grad_norm": 1.3799593448638916,
      "learning_rate": 9.21765936488983e-07,
      "loss": 1.4217,
      "step": 10423
    },
    {
      "epoch": 1.8777752758387751,
      "grad_norm": 1.4268847703933716,
      "learning_rate": 9.19061209123867e-07,
      "loss": 1.3977,
      "step": 10424
    },
    {
      "epoch": 1.8779554154469715,
      "grad_norm": 1.5707372426986694,
      "learning_rate": 9.163604190077502e-07,
      "loss": 1.4816,
      "step": 10425
    },
    {
      "epoch": 1.8781355550551677,
      "grad_norm": 1.4575191736221313,
      "learning_rate": 9.136635663572924e-07,
      "loss": 1.3698,
      "step": 10426
    },
    {
      "epoch": 1.8783156946633641,
      "grad_norm": 1.4714783430099487,
      "learning_rate": 9.109706513888372e-07,
      "loss": 1.4828,
      "step": 10427
    },
    {
      "epoch": 1.8784958342715603,
      "grad_norm": 1.539023756980896,
      "learning_rate": 9.082816743184008e-07,
      "loss": 1.4209,
      "step": 10428
    },
    {
      "epoch": 1.8786759738797567,
      "grad_norm": 1.3515119552612305,
      "learning_rate": 9.055966353616996e-07,
      "loss": 1.0816,
      "step": 10429
    },
    {
      "epoch": 1.8788561134879531,
      "grad_norm": 1.5612529516220093,
      "learning_rate": 9.029155347341223e-07,
      "loss": 1.385,
      "step": 10430
    },
    {
      "epoch": 1.8790362530961495,
      "grad_norm": 1.56065034866333,
      "learning_rate": 9.002383726507469e-07,
      "loss": 1.4927,
      "step": 10431
    },
    {
      "epoch": 1.879216392704346,
      "grad_norm": 1.60688054561615,
      "learning_rate": 8.975651493263349e-07,
      "loss": 1.6076,
      "step": 10432
    },
    {
      "epoch": 1.8793965323125423,
      "grad_norm": 1.6395341157913208,
      "learning_rate": 8.948958649753314e-07,
      "loss": 1.35,
      "step": 10433
    },
    {
      "epoch": 1.8795766719207385,
      "grad_norm": 1.517236590385437,
      "learning_rate": 8.922305198118597e-07,
      "loss": 1.4386,
      "step": 10434
    },
    {
      "epoch": 1.879756811528935,
      "grad_norm": 1.5569759607315063,
      "learning_rate": 8.895691140497376e-07,
      "loss": 1.2799,
      "step": 10435
    },
    {
      "epoch": 1.8799369511371313,
      "grad_norm": 1.5908175706863403,
      "learning_rate": 8.869116479024608e-07,
      "loss": 1.4303,
      "step": 10436
    },
    {
      "epoch": 1.8801170907453275,
      "grad_norm": 1.7033532857894897,
      "learning_rate": 8.84258121583209e-07,
      "loss": 1.5695,
      "step": 10437
    },
    {
      "epoch": 1.880297230353524,
      "grad_norm": 1.4639214277267456,
      "learning_rate": 8.816085353048454e-07,
      "loss": 1.2193,
      "step": 10438
    },
    {
      "epoch": 1.8804773699617203,
      "grad_norm": 1.488935112953186,
      "learning_rate": 8.789628892799218e-07,
      "loss": 1.1333,
      "step": 10439
    },
    {
      "epoch": 1.8806575095699167,
      "grad_norm": 1.5773502588272095,
      "learning_rate": 8.763211837206686e-07,
      "loss": 1.4225,
      "step": 10440
    },
    {
      "epoch": 1.8808376491781131,
      "grad_norm": 1.4575687646865845,
      "learning_rate": 8.736834188390053e-07,
      "loss": 1.2027,
      "step": 10441
    },
    {
      "epoch": 1.8810177887863095,
      "grad_norm": 1.5660375356674194,
      "learning_rate": 8.710495948465292e-07,
      "loss": 1.4276,
      "step": 10442
    },
    {
      "epoch": 1.8811979283945057,
      "grad_norm": 1.5975563526153564,
      "learning_rate": 8.684197119545212e-07,
      "loss": 1.4262,
      "step": 10443
    },
    {
      "epoch": 1.8813780680027021,
      "grad_norm": 1.5464791059494019,
      "learning_rate": 8.657937703739516e-07,
      "loss": 1.4262,
      "step": 10444
    },
    {
      "epoch": 1.8815582076108983,
      "grad_norm": 1.4302457571029663,
      "learning_rate": 8.631717703154796e-07,
      "loss": 1.2234,
      "step": 10445
    },
    {
      "epoch": 1.8817383472190947,
      "grad_norm": 1.5488194227218628,
      "learning_rate": 8.605537119894313e-07,
      "loss": 1.3303,
      "step": 10446
    },
    {
      "epoch": 1.8819184868272911,
      "grad_norm": 1.6429537534713745,
      "learning_rate": 8.579395956058279e-07,
      "loss": 1.3334,
      "step": 10447
    },
    {
      "epoch": 1.8820986264354875,
      "grad_norm": 1.6289007663726807,
      "learning_rate": 8.55329421374379e-07,
      "loss": 1.3364,
      "step": 10448
    },
    {
      "epoch": 1.882278766043684,
      "grad_norm": 1.5279297828674316,
      "learning_rate": 8.527231895044619e-07,
      "loss": 1.3556,
      "step": 10449
    },
    {
      "epoch": 1.8824589056518803,
      "grad_norm": 1.5164117813110352,
      "learning_rate": 8.501209002051535e-07,
      "loss": 1.2232,
      "step": 10450
    },
    {
      "epoch": 1.8826390452600765,
      "grad_norm": 1.638219952583313,
      "learning_rate": 8.475225536852038e-07,
      "loss": 1.3615,
      "step": 10451
    },
    {
      "epoch": 1.882819184868273,
      "grad_norm": 1.413591742515564,
      "learning_rate": 8.44928150153057e-07,
      "loss": 1.8522,
      "step": 10452
    },
    {
      "epoch": 1.8829993244764691,
      "grad_norm": 1.4006863832473755,
      "learning_rate": 8.423376898168245e-07,
      "loss": 1.9277,
      "step": 10453
    },
    {
      "epoch": 1.8831794640846655,
      "grad_norm": 1.4539275169372559,
      "learning_rate": 8.397511728843233e-07,
      "loss": 1.8151,
      "step": 10454
    },
    {
      "epoch": 1.883359603692862,
      "grad_norm": 1.541306972503662,
      "learning_rate": 8.371685995630429e-07,
      "loss": 1.72,
      "step": 10455
    },
    {
      "epoch": 1.8835397433010583,
      "grad_norm": 1.3538320064544678,
      "learning_rate": 8.345899700601456e-07,
      "loss": 1.5999,
      "step": 10456
    },
    {
      "epoch": 1.8837198829092547,
      "grad_norm": 1.625203013420105,
      "learning_rate": 8.320152845824935e-07,
      "loss": 1.9385,
      "step": 10457
    },
    {
      "epoch": 1.8839000225174511,
      "grad_norm": 1.4941275119781494,
      "learning_rate": 8.294445433366327e-07,
      "loss": 1.4806,
      "step": 10458
    },
    {
      "epoch": 1.8840801621256473,
      "grad_norm": 1.6255981922149658,
      "learning_rate": 8.268777465287758e-07,
      "loss": 1.9224,
      "step": 10459
    },
    {
      "epoch": 1.8842603017338437,
      "grad_norm": 1.6820573806762695,
      "learning_rate": 8.243148943648305e-07,
      "loss": 1.428,
      "step": 10460
    },
    {
      "epoch": 1.88444044134204,
      "grad_norm": 1.7153754234313965,
      "learning_rate": 8.217559870503988e-07,
      "loss": 1.9918,
      "step": 10461
    },
    {
      "epoch": 1.8846205809502363,
      "grad_norm": 1.9701718091964722,
      "learning_rate": 8.192010247907445e-07,
      "loss": 1.9317,
      "step": 10462
    },
    {
      "epoch": 1.8848007205584327,
      "grad_norm": 1.584980845451355,
      "learning_rate": 8.166500077908257e-07,
      "loss": 1.7143,
      "step": 10463
    },
    {
      "epoch": 1.8849808601666291,
      "grad_norm": 1.4482595920562744,
      "learning_rate": 8.141029362552899e-07,
      "loss": 1.5345,
      "step": 10464
    },
    {
      "epoch": 1.8851609997748255,
      "grad_norm": 1.411331295967102,
      "learning_rate": 8.115598103884569e-07,
      "loss": 1.3958,
      "step": 10465
    },
    {
      "epoch": 1.885341139383022,
      "grad_norm": 1.4702675342559814,
      "learning_rate": 8.090206303943415e-07,
      "loss": 1.3084,
      "step": 10466
    },
    {
      "epoch": 1.8855212789912184,
      "grad_norm": 1.4231278896331787,
      "learning_rate": 8.06485396476625e-07,
      "loss": 1.4413,
      "step": 10467
    },
    {
      "epoch": 1.8857014185994145,
      "grad_norm": 1.5550378561019897,
      "learning_rate": 8.039541088387004e-07,
      "loss": 1.5643,
      "step": 10468
    },
    {
      "epoch": 1.885881558207611,
      "grad_norm": 1.3691565990447998,
      "learning_rate": 8.014267676836051e-07,
      "loss": 1.3971,
      "step": 10469
    },
    {
      "epoch": 1.8860616978158071,
      "grad_norm": 1.6286625862121582,
      "learning_rate": 7.989033732140938e-07,
      "loss": 1.7423,
      "step": 10470
    },
    {
      "epoch": 1.8862418374240035,
      "grad_norm": 1.3730206489562988,
      "learning_rate": 7.963839256325877e-07,
      "loss": 1.3751,
      "step": 10471
    },
    {
      "epoch": 1.8864219770322,
      "grad_norm": 1.4568886756896973,
      "learning_rate": 7.938684251411921e-07,
      "loss": 1.3351,
      "step": 10472
    },
    {
      "epoch": 1.8866021166403963,
      "grad_norm": 1.4766813516616821,
      "learning_rate": 7.91356871941712e-07,
      "loss": 1.3378,
      "step": 10473
    },
    {
      "epoch": 1.8867822562485927,
      "grad_norm": 1.369766116142273,
      "learning_rate": 7.888492662356084e-07,
      "loss": 1.3716,
      "step": 10474
    },
    {
      "epoch": 1.8869623958567892,
      "grad_norm": 1.5298302173614502,
      "learning_rate": 7.863456082240539e-07,
      "loss": 1.5975,
      "step": 10475
    },
    {
      "epoch": 1.8871425354649853,
      "grad_norm": 1.4681299924850464,
      "learning_rate": 7.838458981078767e-07,
      "loss": 1.453,
      "step": 10476
    },
    {
      "epoch": 1.8873226750731817,
      "grad_norm": 1.540401577949524,
      "learning_rate": 7.813501360876107e-07,
      "loss": 1.347,
      "step": 10477
    },
    {
      "epoch": 1.887502814681378,
      "grad_norm": 1.538756012916565,
      "learning_rate": 7.788583223634627e-07,
      "loss": 1.4703,
      "step": 10478
    },
    {
      "epoch": 1.8876829542895743,
      "grad_norm": 1.3124223947525024,
      "learning_rate": 7.763704571353281e-07,
      "loss": 1.18,
      "step": 10479
    },
    {
      "epoch": 1.8878630938977707,
      "grad_norm": 1.5526539087295532,
      "learning_rate": 7.738865406027862e-07,
      "loss": 1.6838,
      "step": 10480
    },
    {
      "epoch": 1.8880432335059671,
      "grad_norm": 1.5394126176834106,
      "learning_rate": 7.714065729650776e-07,
      "loss": 1.5818,
      "step": 10481
    },
    {
      "epoch": 1.8882233731141636,
      "grad_norm": 1.473955512046814,
      "learning_rate": 7.689305544211601e-07,
      "loss": 1.3791,
      "step": 10482
    },
    {
      "epoch": 1.88840351272236,
      "grad_norm": 1.4336568117141724,
      "learning_rate": 7.664584851696522e-07,
      "loss": 1.3102,
      "step": 10483
    },
    {
      "epoch": 1.8885836523305561,
      "grad_norm": 1.5253362655639648,
      "learning_rate": 7.639903654088676e-07,
      "loss": 1.4434,
      "step": 10484
    },
    {
      "epoch": 1.8887637919387525,
      "grad_norm": 1.4720733165740967,
      "learning_rate": 7.61526195336787e-07,
      "loss": 1.3012,
      "step": 10485
    },
    {
      "epoch": 1.8889439315469487,
      "grad_norm": 1.557500958442688,
      "learning_rate": 7.590659751510965e-07,
      "loss": 1.4302,
      "step": 10486
    },
    {
      "epoch": 1.8891240711551451,
      "grad_norm": 1.4172635078430176,
      "learning_rate": 7.566097050491438e-07,
      "loss": 1.0922,
      "step": 10487
    },
    {
      "epoch": 1.8893042107633415,
      "grad_norm": 1.4982826709747314,
      "learning_rate": 7.541573852279826e-07,
      "loss": 1.4263,
      "step": 10488
    },
    {
      "epoch": 1.889484350371538,
      "grad_norm": 1.4319500923156738,
      "learning_rate": 7.517090158843276e-07,
      "loss": 1.3974,
      "step": 10489
    },
    {
      "epoch": 1.8896644899797344,
      "grad_norm": 1.498268485069275,
      "learning_rate": 7.492645972145829e-07,
      "loss": 1.3361,
      "step": 10490
    },
    {
      "epoch": 1.8898446295879308,
      "grad_norm": 1.580480933189392,
      "learning_rate": 7.46824129414847e-07,
      "loss": 1.3082,
      "step": 10491
    },
    {
      "epoch": 1.8900247691961272,
      "grad_norm": 1.388025164604187,
      "learning_rate": 7.443876126808913e-07,
      "loss": 1.2081,
      "step": 10492
    },
    {
      "epoch": 1.8902049088043233,
      "grad_norm": 1.6182637214660645,
      "learning_rate": 7.419550472081648e-07,
      "loss": 1.502,
      "step": 10493
    },
    {
      "epoch": 1.8903850484125198,
      "grad_norm": 1.5960328578948975,
      "learning_rate": 7.395264331918117e-07,
      "loss": 1.3649,
      "step": 10494
    },
    {
      "epoch": 1.890565188020716,
      "grad_norm": 1.561490774154663,
      "learning_rate": 7.371017708266536e-07,
      "loss": 1.3794,
      "step": 10495
    },
    {
      "epoch": 1.8907453276289123,
      "grad_norm": 1.5892928838729858,
      "learning_rate": 7.346810603072018e-07,
      "loss": 1.2528,
      "step": 10496
    },
    {
      "epoch": 1.8909254672371087,
      "grad_norm": 1.6903246641159058,
      "learning_rate": 7.322643018276343e-07,
      "loss": 1.5753,
      "step": 10497
    },
    {
      "epoch": 1.8911056068453052,
      "grad_norm": 1.6042462587356567,
      "learning_rate": 7.298514955818292e-07,
      "loss": 1.3472,
      "step": 10498
    },
    {
      "epoch": 1.8912857464535016,
      "grad_norm": 1.5658302307128906,
      "learning_rate": 7.274426417633373e-07,
      "loss": 1.3118,
      "step": 10499
    },
    {
      "epoch": 1.891465886061698,
      "grad_norm": 1.4369940757751465,
      "learning_rate": 7.250377405653986e-07,
      "loss": 1.1206,
      "step": 10500
    },
    {
      "epoch": 1.8916460256698941,
      "grad_norm": 1.5997005701065063,
      "learning_rate": 7.226367921809307e-07,
      "loss": 1.4936,
      "step": 10501
    },
    {
      "epoch": 1.8918261652780906,
      "grad_norm": 1.4643075466156006,
      "learning_rate": 7.202397968025354e-07,
      "loss": 1.9299,
      "step": 10502
    },
    {
      "epoch": 1.8920063048862867,
      "grad_norm": 1.4001431465148926,
      "learning_rate": 7.178467546225032e-07,
      "loss": 1.8039,
      "step": 10503
    },
    {
      "epoch": 1.8921864444944831,
      "grad_norm": 1.3389837741851807,
      "learning_rate": 7.154576658327972e-07,
      "loss": 1.8528,
      "step": 10504
    },
    {
      "epoch": 1.8923665841026795,
      "grad_norm": 1.4355262517929077,
      "learning_rate": 7.130725306250696e-07,
      "loss": 1.7639,
      "step": 10505
    },
    {
      "epoch": 1.892546723710876,
      "grad_norm": 1.3411816358566284,
      "learning_rate": 7.106913491906619e-07,
      "loss": 1.524,
      "step": 10506
    },
    {
      "epoch": 1.8927268633190724,
      "grad_norm": 1.4855798482894897,
      "learning_rate": 7.083141217205824e-07,
      "loss": 1.7806,
      "step": 10507
    },
    {
      "epoch": 1.8929070029272688,
      "grad_norm": 1.4194215536117554,
      "learning_rate": 7.059408484055396e-07,
      "loss": 1.7253,
      "step": 10508
    },
    {
      "epoch": 1.893087142535465,
      "grad_norm": 1.7397346496582031,
      "learning_rate": 7.035715294359036e-07,
      "loss": 1.8814,
      "step": 10509
    },
    {
      "epoch": 1.8932672821436614,
      "grad_norm": 1.546053409576416,
      "learning_rate": 7.012061650017555e-07,
      "loss": 1.8013,
      "step": 10510
    },
    {
      "epoch": 1.8934474217518575,
      "grad_norm": 1.7077016830444336,
      "learning_rate": 6.988447552928323e-07,
      "loss": 2.008,
      "step": 10511
    },
    {
      "epoch": 1.893627561360054,
      "grad_norm": 1.927232265472412,
      "learning_rate": 6.964873004985717e-07,
      "loss": 1.986,
      "step": 10512
    },
    {
      "epoch": 1.8938077009682504,
      "grad_norm": 1.558251976966858,
      "learning_rate": 6.941338008080889e-07,
      "loss": 1.5316,
      "step": 10513
    },
    {
      "epoch": 1.8939878405764468,
      "grad_norm": 1.4736559391021729,
      "learning_rate": 6.917842564101662e-07,
      "loss": 1.401,
      "step": 10514
    },
    {
      "epoch": 1.8941679801846432,
      "grad_norm": 1.651094675064087,
      "learning_rate": 6.894386674933029e-07,
      "loss": 1.864,
      "step": 10515
    },
    {
      "epoch": 1.8943481197928396,
      "grad_norm": 1.3020305633544922,
      "learning_rate": 6.870970342456484e-07,
      "loss": 1.2879,
      "step": 10516
    },
    {
      "epoch": 1.8945282594010358,
      "grad_norm": 1.5028489828109741,
      "learning_rate": 6.84759356855047e-07,
      "loss": 1.44,
      "step": 10517
    },
    {
      "epoch": 1.8947083990092322,
      "grad_norm": 1.3594685792922974,
      "learning_rate": 6.824256355090319e-07,
      "loss": 1.2845,
      "step": 10518
    },
    {
      "epoch": 1.8948885386174283,
      "grad_norm": 1.4115666151046753,
      "learning_rate": 6.800958703948091e-07,
      "loss": 1.4618,
      "step": 10519
    },
    {
      "epoch": 1.8950686782256247,
      "grad_norm": 1.3464326858520508,
      "learning_rate": 6.777700616992733e-07,
      "loss": 1.2171,
      "step": 10520
    },
    {
      "epoch": 1.8952488178338212,
      "grad_norm": 1.4653886556625366,
      "learning_rate": 6.754482096089976e-07,
      "loss": 1.4268,
      "step": 10521
    },
    {
      "epoch": 1.8954289574420176,
      "grad_norm": 1.4941991567611694,
      "learning_rate": 6.73130314310244e-07,
      "loss": 1.597,
      "step": 10522
    },
    {
      "epoch": 1.895609097050214,
      "grad_norm": 1.4208091497421265,
      "learning_rate": 6.70816375988953e-07,
      "loss": 1.4384,
      "step": 10523
    },
    {
      "epoch": 1.8957892366584104,
      "grad_norm": 1.5423327684402466,
      "learning_rate": 6.685063948307424e-07,
      "loss": 1.4755,
      "step": 10524
    },
    {
      "epoch": 1.8959693762666068,
      "grad_norm": 1.451150894165039,
      "learning_rate": 6.662003710209197e-07,
      "loss": 1.3439,
      "step": 10525
    },
    {
      "epoch": 1.896149515874803,
      "grad_norm": 1.5733636617660522,
      "learning_rate": 6.638983047444758e-07,
      "loss": 1.4717,
      "step": 10526
    },
    {
      "epoch": 1.8963296554829994,
      "grad_norm": 1.4216341972351074,
      "learning_rate": 6.616001961860741e-07,
      "loss": 1.5663,
      "step": 10527
    },
    {
      "epoch": 1.8965097950911955,
      "grad_norm": 1.428039312362671,
      "learning_rate": 6.593060455300726e-07,
      "loss": 1.3089,
      "step": 10528
    },
    {
      "epoch": 1.896689934699392,
      "grad_norm": 1.3801791667938232,
      "learning_rate": 6.57015852960513e-07,
      "loss": 1.2624,
      "step": 10529
    },
    {
      "epoch": 1.8968700743075884,
      "grad_norm": 1.535874605178833,
      "learning_rate": 6.547296186611041e-07,
      "loss": 1.4809,
      "step": 10530
    },
    {
      "epoch": 1.8970502139157848,
      "grad_norm": 1.6675653457641602,
      "learning_rate": 6.524473428152489e-07,
      "loss": 1.4368,
      "step": 10531
    },
    {
      "epoch": 1.8972303535239812,
      "grad_norm": 1.5514951944351196,
      "learning_rate": 6.501690256060344e-07,
      "loss": 1.3244,
      "step": 10532
    },
    {
      "epoch": 1.8974104931321776,
      "grad_norm": 1.6219271421432495,
      "learning_rate": 6.478946672162256e-07,
      "loss": 1.5808,
      "step": 10533
    },
    {
      "epoch": 1.8975906327403738,
      "grad_norm": 1.5591564178466797,
      "learning_rate": 6.456242678282654e-07,
      "loss": 1.3291,
      "step": 10534
    },
    {
      "epoch": 1.8977707723485702,
      "grad_norm": 1.5878000259399414,
      "learning_rate": 6.433578276242858e-07,
      "loss": 1.4047,
      "step": 10535
    },
    {
      "epoch": 1.8979509119567664,
      "grad_norm": 1.5497514009475708,
      "learning_rate": 6.410953467860969e-07,
      "loss": 1.5151,
      "step": 10536
    },
    {
      "epoch": 1.8981310515649628,
      "grad_norm": 1.6520183086395264,
      "learning_rate": 6.388368254951982e-07,
      "loss": 1.4225,
      "step": 10537
    },
    {
      "epoch": 1.8983111911731592,
      "grad_norm": 1.4138176441192627,
      "learning_rate": 6.365822639327723e-07,
      "loss": 1.3424,
      "step": 10538
    },
    {
      "epoch": 1.8984913307813556,
      "grad_norm": 1.5050724744796753,
      "learning_rate": 6.343316622796691e-07,
      "loss": 1.3083,
      "step": 10539
    },
    {
      "epoch": 1.898671470389552,
      "grad_norm": 1.4484436511993408,
      "learning_rate": 6.320850207164386e-07,
      "loss": 1.3235,
      "step": 10540
    },
    {
      "epoch": 1.8988516099977484,
      "grad_norm": 1.8164695501327515,
      "learning_rate": 6.298423394232978e-07,
      "loss": 1.6193,
      "step": 10541
    },
    {
      "epoch": 1.8990317496059446,
      "grad_norm": 1.5286498069763184,
      "learning_rate": 6.276036185801526e-07,
      "loss": 1.3495,
      "step": 10542
    },
    {
      "epoch": 1.899211889214141,
      "grad_norm": 1.5159372091293335,
      "learning_rate": 6.25368858366604e-07,
      "loss": 1.2539,
      "step": 10543
    },
    {
      "epoch": 1.8993920288223372,
      "grad_norm": 1.628412127494812,
      "learning_rate": 6.23138058961914e-07,
      "loss": 1.444,
      "step": 10544
    },
    {
      "epoch": 1.8995721684305336,
      "grad_norm": 1.5929714441299438,
      "learning_rate": 6.209112205450452e-07,
      "loss": 1.3879,
      "step": 10545
    },
    {
      "epoch": 1.89975230803873,
      "grad_norm": 1.566931128501892,
      "learning_rate": 6.186883432946211e-07,
      "loss": 1.2622,
      "step": 10546
    },
    {
      "epoch": 1.8999324476469264,
      "grad_norm": 1.6450026035308838,
      "learning_rate": 6.164694273889604e-07,
      "loss": 1.4355,
      "step": 10547
    },
    {
      "epoch": 1.9001125872551228,
      "grad_norm": 1.5769636631011963,
      "learning_rate": 6.142544730060706e-07,
      "loss": 1.2818,
      "step": 10548
    },
    {
      "epoch": 1.9002927268633192,
      "grad_norm": 1.4638614654541016,
      "learning_rate": 6.120434803236374e-07,
      "loss": 1.354,
      "step": 10549
    },
    {
      "epoch": 1.9004728664715156,
      "grad_norm": 1.5841487646102905,
      "learning_rate": 6.098364495190135e-07,
      "loss": 1.4453,
      "step": 10550
    },
    {
      "epoch": 1.9006530060797118,
      "grad_norm": 1.3609451055526733,
      "learning_rate": 6.076333807692514e-07,
      "loss": 1.1144,
      "step": 10551
    },
    {
      "epoch": 1.9008331456879082,
      "grad_norm": 1.4796746969223022,
      "learning_rate": 6.054342742510877e-07,
      "loss": 1.8244,
      "step": 10552
    },
    {
      "epoch": 1.9010132852961044,
      "grad_norm": 1.2966715097427368,
      "learning_rate": 6.032391301409257e-07,
      "loss": 1.7389,
      "step": 10553
    },
    {
      "epoch": 1.9011934249043008,
      "grad_norm": 1.3886855840682983,
      "learning_rate": 6.01047948614858e-07,
      "loss": 1.9183,
      "step": 10554
    },
    {
      "epoch": 1.9013735645124972,
      "grad_norm": 1.391626000404358,
      "learning_rate": 5.98860729848666e-07,
      "loss": 1.9345,
      "step": 10555
    },
    {
      "epoch": 1.9015537041206936,
      "grad_norm": 1.4676650762557983,
      "learning_rate": 5.966774740178039e-07,
      "loss": 1.8951,
      "step": 10556
    },
    {
      "epoch": 1.90173384372889,
      "grad_norm": 1.3372501134872437,
      "learning_rate": 5.944981812974093e-07,
      "loss": 1.6935,
      "step": 10557
    },
    {
      "epoch": 1.9019139833370864,
      "grad_norm": 1.4513601064682007,
      "learning_rate": 5.923228518623036e-07,
      "loss": 1.402,
      "step": 10558
    },
    {
      "epoch": 1.9020941229452826,
      "grad_norm": 1.8146361112594604,
      "learning_rate": 5.901514858869972e-07,
      "loss": 2.262,
      "step": 10559
    },
    {
      "epoch": 1.902274262553479,
      "grad_norm": 1.6722593307495117,
      "learning_rate": 5.879840835456674e-07,
      "loss": 1.7593,
      "step": 10560
    },
    {
      "epoch": 1.9024544021616752,
      "grad_norm": 1.9109129905700684,
      "learning_rate": 5.858206450121917e-07,
      "loss": 1.7727,
      "step": 10561
    },
    {
      "epoch": 1.9026345417698716,
      "grad_norm": 1.459641695022583,
      "learning_rate": 5.836611704601147e-07,
      "loss": 1.3479,
      "step": 10562
    },
    {
      "epoch": 1.902814681378068,
      "grad_norm": 1.444750189781189,
      "learning_rate": 5.815056600626645e-07,
      "loss": 1.4824,
      "step": 10563
    },
    {
      "epoch": 1.9029948209862644,
      "grad_norm": 1.3514031171798706,
      "learning_rate": 5.793541139927638e-07,
      "loss": 1.3809,
      "step": 10564
    },
    {
      "epoch": 1.9031749605944608,
      "grad_norm": 1.3343899250030518,
      "learning_rate": 5.772065324230025e-07,
      "loss": 1.3201,
      "step": 10565
    },
    {
      "epoch": 1.9033551002026572,
      "grad_norm": 1.4790526628494263,
      "learning_rate": 5.750629155256593e-07,
      "loss": 1.3902,
      "step": 10566
    },
    {
      "epoch": 1.9035352398108534,
      "grad_norm": 1.3490937948226929,
      "learning_rate": 5.729232634726966e-07,
      "loss": 1.2023,
      "step": 10567
    },
    {
      "epoch": 1.9037153794190498,
      "grad_norm": 1.4462461471557617,
      "learning_rate": 5.707875764357551e-07,
      "loss": 1.3443,
      "step": 10568
    },
    {
      "epoch": 1.903895519027246,
      "grad_norm": 1.5272160768508911,
      "learning_rate": 5.686558545861532e-07,
      "loss": 1.4835,
      "step": 10569
    },
    {
      "epoch": 1.9040756586354424,
      "grad_norm": 1.3934524059295654,
      "learning_rate": 5.66528098094904e-07,
      "loss": 1.176,
      "step": 10570
    },
    {
      "epoch": 1.9042557982436388,
      "grad_norm": 1.2959026098251343,
      "learning_rate": 5.644043071326932e-07,
      "loss": 1.1733,
      "step": 10571
    },
    {
      "epoch": 1.9044359378518352,
      "grad_norm": 1.4108245372772217,
      "learning_rate": 5.622844818698903e-07,
      "loss": 1.2941,
      "step": 10572
    },
    {
      "epoch": 1.9046160774600316,
      "grad_norm": 1.5393667221069336,
      "learning_rate": 5.601686224765479e-07,
      "loss": 1.3567,
      "step": 10573
    },
    {
      "epoch": 1.904796217068228,
      "grad_norm": 1.3832629919052124,
      "learning_rate": 5.580567291223914e-07,
      "loss": 1.2798,
      "step": 10574
    },
    {
      "epoch": 1.9049763566764242,
      "grad_norm": 1.4247702360153198,
      "learning_rate": 5.559488019768466e-07,
      "loss": 1.2418,
      "step": 10575
    },
    {
      "epoch": 1.9051564962846206,
      "grad_norm": 1.3773174285888672,
      "learning_rate": 5.538448412090058e-07,
      "loss": 1.155,
      "step": 10576
    },
    {
      "epoch": 1.905336635892817,
      "grad_norm": 1.4633855819702148,
      "learning_rate": 5.517448469876452e-07,
      "loss": 1.4682,
      "step": 10577
    },
    {
      "epoch": 1.9055167755010132,
      "grad_norm": 1.356995940208435,
      "learning_rate": 5.496488194812355e-07,
      "loss": 1.1869,
      "step": 10578
    },
    {
      "epoch": 1.9056969151092096,
      "grad_norm": 1.4500218629837036,
      "learning_rate": 5.475567588578978e-07,
      "loss": 1.4127,
      "step": 10579
    },
    {
      "epoch": 1.905877054717406,
      "grad_norm": 1.528525710105896,
      "learning_rate": 5.454686652854757e-07,
      "loss": 1.4854,
      "step": 10580
    },
    {
      "epoch": 1.9060571943256024,
      "grad_norm": 1.7621057033538818,
      "learning_rate": 5.433845389314685e-07,
      "loss": 1.5677,
      "step": 10581
    },
    {
      "epoch": 1.9062373339337988,
      "grad_norm": 1.508873701095581,
      "learning_rate": 5.413043799630646e-07,
      "loss": 1.4941,
      "step": 10582
    },
    {
      "epoch": 1.9064174735419952,
      "grad_norm": 1.3634029626846313,
      "learning_rate": 5.392281885471362e-07,
      "loss": 1.2119,
      "step": 10583
    },
    {
      "epoch": 1.9065976131501914,
      "grad_norm": 1.4658923149108887,
      "learning_rate": 5.371559648502223e-07,
      "loss": 1.328,
      "step": 10584
    },
    {
      "epoch": 1.9067777527583878,
      "grad_norm": 1.3000259399414062,
      "learning_rate": 5.350877090385731e-07,
      "loss": 1.0919,
      "step": 10585
    },
    {
      "epoch": 1.906957892366584,
      "grad_norm": 1.559252381324768,
      "learning_rate": 5.330234212780893e-07,
      "loss": 1.616,
      "step": 10586
    },
    {
      "epoch": 1.9071380319747804,
      "grad_norm": 1.4746516942977905,
      "learning_rate": 5.309631017343774e-07,
      "loss": 1.4716,
      "step": 10587
    },
    {
      "epoch": 1.9073181715829768,
      "grad_norm": 1.6650214195251465,
      "learning_rate": 5.289067505727052e-07,
      "loss": 1.4426,
      "step": 10588
    },
    {
      "epoch": 1.9074983111911732,
      "grad_norm": 1.5495333671569824,
      "learning_rate": 5.268543679580407e-07,
      "loss": 1.4142,
      "step": 10589
    },
    {
      "epoch": 1.9076784507993696,
      "grad_norm": 1.5035241842269897,
      "learning_rate": 5.248059540550187e-07,
      "loss": 1.2558,
      "step": 10590
    },
    {
      "epoch": 1.907858590407566,
      "grad_norm": 1.639175295829773,
      "learning_rate": 5.227615090279637e-07,
      "loss": 1.4712,
      "step": 10591
    },
    {
      "epoch": 1.9080387300157622,
      "grad_norm": 1.4025806188583374,
      "learning_rate": 5.207210330408829e-07,
      "loss": 1.164,
      "step": 10592
    },
    {
      "epoch": 1.9082188696239586,
      "grad_norm": 1.6409562826156616,
      "learning_rate": 5.186845262574569e-07,
      "loss": 1.5255,
      "step": 10593
    },
    {
      "epoch": 1.9083990092321548,
      "grad_norm": 1.4403104782104492,
      "learning_rate": 5.166519888410604e-07,
      "loss": 1.3298,
      "step": 10594
    },
    {
      "epoch": 1.9085791488403512,
      "grad_norm": 1.6716806888580322,
      "learning_rate": 5.146234209547351e-07,
      "loss": 1.3915,
      "step": 10595
    },
    {
      "epoch": 1.9087592884485476,
      "grad_norm": 1.6240124702453613,
      "learning_rate": 5.125988227612233e-07,
      "loss": 1.2375,
      "step": 10596
    },
    {
      "epoch": 1.908939428056744,
      "grad_norm": 1.4897117614746094,
      "learning_rate": 5.105781944229226e-07,
      "loss": 1.2471,
      "step": 10597
    },
    {
      "epoch": 1.9091195676649404,
      "grad_norm": 1.5866055488586426,
      "learning_rate": 5.085615361019369e-07,
      "loss": 1.2568,
      "step": 10598
    },
    {
      "epoch": 1.9092997072731368,
      "grad_norm": 1.5033975839614868,
      "learning_rate": 5.065488479600367e-07,
      "loss": 1.3493,
      "step": 10599
    },
    {
      "epoch": 1.909479846881333,
      "grad_norm": 1.587776780128479,
      "learning_rate": 5.045401301586817e-07,
      "loss": 1.5081,
      "step": 10600
    },
    {
      "epoch": 1.9096599864895294,
      "grad_norm": 1.5601608753204346,
      "learning_rate": 5.025353828590095e-07,
      "loss": 1.5215,
      "step": 10601
    },
    {
      "epoch": 1.9098401260977256,
      "grad_norm": 1.330634593963623,
      "learning_rate": 5.005346062218308e-07,
      "loss": 1.8015,
      "step": 10602
    },
    {
      "epoch": 1.910020265705922,
      "grad_norm": 1.224668025970459,
      "learning_rate": 4.985378004076669e-07,
      "loss": 1.5842,
      "step": 10603
    },
    {
      "epoch": 1.9102004053141184,
      "grad_norm": 1.403169870376587,
      "learning_rate": 4.96544965576684e-07,
      "loss": 1.9326,
      "step": 10604
    },
    {
      "epoch": 1.9103805449223148,
      "grad_norm": 1.532300591468811,
      "learning_rate": 4.945561018887546e-07,
      "loss": 1.9527,
      "step": 10605
    },
    {
      "epoch": 1.9105606845305112,
      "grad_norm": 1.553962230682373,
      "learning_rate": 4.925712095034174e-07,
      "loss": 1.7572,
      "step": 10606
    },
    {
      "epoch": 1.9107408241387076,
      "grad_norm": 1.4353445768356323,
      "learning_rate": 4.905902885799063e-07,
      "loss": 1.6705,
      "step": 10607
    },
    {
      "epoch": 1.910920963746904,
      "grad_norm": 1.4835301637649536,
      "learning_rate": 4.886133392771274e-07,
      "loss": 1.5493,
      "step": 10608
    },
    {
      "epoch": 1.9111011033551002,
      "grad_norm": 1.726707100868225,
      "learning_rate": 4.866403617536708e-07,
      "loss": 1.7632,
      "step": 10609
    },
    {
      "epoch": 1.9112812429632966,
      "grad_norm": 1.7077559232711792,
      "learning_rate": 4.846713561678096e-07,
      "loss": 1.8504,
      "step": 10610
    },
    {
      "epoch": 1.9114613825714928,
      "grad_norm": 1.9443711042404175,
      "learning_rate": 4.827063226774897e-07,
      "loss": 1.7915,
      "step": 10611
    },
    {
      "epoch": 1.9116415221796892,
      "grad_norm": 1.7511098384857178,
      "learning_rate": 4.807452614403463e-07,
      "loss": 1.6048,
      "step": 10612
    },
    {
      "epoch": 1.9118216617878856,
      "grad_norm": 1.4773104190826416,
      "learning_rate": 4.787881726137034e-07,
      "loss": 1.3727,
      "step": 10613
    },
    {
      "epoch": 1.912001801396082,
      "grad_norm": 1.495373010635376,
      "learning_rate": 4.7683505635455204e-07,
      "loss": 1.5537,
      "step": 10614
    },
    {
      "epoch": 1.9121819410042784,
      "grad_norm": 1.4386756420135498,
      "learning_rate": 4.748859128195726e-07,
      "loss": 1.4702,
      "step": 10615
    },
    {
      "epoch": 1.9123620806124748,
      "grad_norm": 1.4554927349090576,
      "learning_rate": 4.7294074216511773e-07,
      "loss": 1.5219,
      "step": 10616
    },
    {
      "epoch": 1.912542220220671,
      "grad_norm": 1.428983449935913,
      "learning_rate": 4.709995445472348e-07,
      "loss": 1.3191,
      "step": 10617
    },
    {
      "epoch": 1.9127223598288674,
      "grad_norm": 1.515053153038025,
      "learning_rate": 4.6906232012164377e-07,
      "loss": 1.5178,
      "step": 10618
    },
    {
      "epoch": 1.9129024994370636,
      "grad_norm": 1.345089316368103,
      "learning_rate": 4.6712906904374797e-07,
      "loss": 1.1412,
      "step": 10619
    },
    {
      "epoch": 1.91308263904526,
      "grad_norm": 1.5110143423080444,
      "learning_rate": 4.65199791468629e-07,
      "loss": 1.502,
      "step": 10620
    },
    {
      "epoch": 1.9132627786534564,
      "grad_norm": 1.428783655166626,
      "learning_rate": 4.632744875510631e-07,
      "loss": 1.2869,
      "step": 10621
    },
    {
      "epoch": 1.9134429182616528,
      "grad_norm": 1.4827884435653687,
      "learning_rate": 4.6135315744548216e-07,
      "loss": 1.5696,
      "step": 10622
    },
    {
      "epoch": 1.9136230578698492,
      "grad_norm": 1.4285590648651123,
      "learning_rate": 4.5943580130602406e-07,
      "loss": 1.4427,
      "step": 10623
    },
    {
      "epoch": 1.9138031974780456,
      "grad_norm": 1.4759881496429443,
      "learning_rate": 4.575224192864935e-07,
      "loss": 1.3399,
      "step": 10624
    },
    {
      "epoch": 1.9139833370862418,
      "grad_norm": 1.4492740631103516,
      "learning_rate": 4.5561301154038447e-07,
      "loss": 1.4786,
      "step": 10625
    },
    {
      "epoch": 1.9141634766944382,
      "grad_norm": 1.3105170726776123,
      "learning_rate": 4.537075782208688e-07,
      "loss": 1.1276,
      "step": 10626
    },
    {
      "epoch": 1.9143436163026344,
      "grad_norm": 1.3947352170944214,
      "learning_rate": 4.518061194807965e-07,
      "loss": 1.2279,
      "step": 10627
    },
    {
      "epoch": 1.9145237559108308,
      "grad_norm": 1.4282366037368774,
      "learning_rate": 4.4990863547270114e-07,
      "loss": 1.4077,
      "step": 10628
    },
    {
      "epoch": 1.9147038955190272,
      "grad_norm": 1.5841740369796753,
      "learning_rate": 4.4801512634880527e-07,
      "loss": 1.4595,
      "step": 10629
    },
    {
      "epoch": 1.9148840351272236,
      "grad_norm": 1.5349650382995605,
      "learning_rate": 4.461255922609986e-07,
      "loss": 1.562,
      "step": 10630
    },
    {
      "epoch": 1.91506417473542,
      "grad_norm": 1.4783697128295898,
      "learning_rate": 4.4424003336085427e-07,
      "loss": 1.3324,
      "step": 10631
    },
    {
      "epoch": 1.9152443143436164,
      "grad_norm": 1.5419166088104248,
      "learning_rate": 4.423584497996458e-07,
      "loss": 1.3299,
      "step": 10632
    },
    {
      "epoch": 1.9154244539518126,
      "grad_norm": 1.2100845575332642,
      "learning_rate": 4.4048084172829684e-07,
      "loss": 1.0755,
      "step": 10633
    },
    {
      "epoch": 1.915604593560009,
      "grad_norm": 1.5638476610183716,
      "learning_rate": 4.386072092974369e-07,
      "loss": 1.6335,
      "step": 10634
    },
    {
      "epoch": 1.9157847331682054,
      "grad_norm": 1.4543256759643555,
      "learning_rate": 4.36737552657368e-07,
      "loss": 1.3808,
      "step": 10635
    },
    {
      "epoch": 1.9159648727764016,
      "grad_norm": 1.5768142938613892,
      "learning_rate": 4.348718719580702e-07,
      "loss": 1.3743,
      "step": 10636
    },
    {
      "epoch": 1.916145012384598,
      "grad_norm": 1.5087361335754395,
      "learning_rate": 4.330101673492071e-07,
      "loss": 1.2882,
      "step": 10637
    },
    {
      "epoch": 1.9163251519927944,
      "grad_norm": 1.5756261348724365,
      "learning_rate": 4.311524389801258e-07,
      "loss": 1.5029,
      "step": 10638
    },
    {
      "epoch": 1.9165052916009908,
      "grad_norm": 1.4810669422149658,
      "learning_rate": 4.292986869998517e-07,
      "loss": 1.3864,
      "step": 10639
    },
    {
      "epoch": 1.9166854312091872,
      "grad_norm": 1.7460476160049438,
      "learning_rate": 4.274489115570934e-07,
      "loss": 1.5851,
      "step": 10640
    },
    {
      "epoch": 1.9168655708173836,
      "grad_norm": 1.6554853916168213,
      "learning_rate": 4.256031128002325e-07,
      "loss": 1.5184,
      "step": 10641
    },
    {
      "epoch": 1.9170457104255798,
      "grad_norm": 1.4849661588668823,
      "learning_rate": 4.2376129087735027e-07,
      "loss": 1.3295,
      "step": 10642
    },
    {
      "epoch": 1.9172258500337762,
      "grad_norm": 1.5633811950683594,
      "learning_rate": 4.2192344593618985e-07,
      "loss": 1.5452,
      "step": 10643
    },
    {
      "epoch": 1.9174059896419724,
      "grad_norm": 1.5531846284866333,
      "learning_rate": 4.200895781241776e-07,
      "loss": 1.5133,
      "step": 10644
    },
    {
      "epoch": 1.9175861292501688,
      "grad_norm": 1.3759820461273193,
      "learning_rate": 4.1825968758842925e-07,
      "loss": 1.1063,
      "step": 10645
    },
    {
      "epoch": 1.9177662688583652,
      "grad_norm": 1.5527273416519165,
      "learning_rate": 4.164337744757385e-07,
      "loss": 1.4813,
      "step": 10646
    },
    {
      "epoch": 1.9179464084665616,
      "grad_norm": 1.5540648698806763,
      "learning_rate": 4.146118389325826e-07,
      "loss": 1.3267,
      "step": 10647
    },
    {
      "epoch": 1.918126548074758,
      "grad_norm": 1.63435959815979,
      "learning_rate": 4.127938811051113e-07,
      "loss": 1.6224,
      "step": 10648
    },
    {
      "epoch": 1.9183066876829544,
      "grad_norm": 1.5442606210708618,
      "learning_rate": 4.10979901139158e-07,
      "loss": 1.3054,
      "step": 10649
    },
    {
      "epoch": 1.9184868272911506,
      "grad_norm": 1.5588059425354004,
      "learning_rate": 4.0916989918024507e-07,
      "loss": 1.4591,
      "step": 10650
    },
    {
      "epoch": 1.918666966899347,
      "grad_norm": 1.57216477394104,
      "learning_rate": 4.073638753735731e-07,
      "loss": 1.2147,
      "step": 10651
    },
    {
      "epoch": 1.9188471065075432,
      "grad_norm": 1.2488489151000977,
      "learning_rate": 4.055618298640096e-07,
      "loss": 1.4645,
      "step": 10652
    },
    {
      "epoch": 1.9190272461157396,
      "grad_norm": 1.3164516687393188,
      "learning_rate": 4.0376376279612214e-07,
      "loss": 1.754,
      "step": 10653
    },
    {
      "epoch": 1.919207385723936,
      "grad_norm": 1.4535037279129028,
      "learning_rate": 4.01969674314151e-07,
      "loss": 1.897,
      "step": 10654
    },
    {
      "epoch": 1.9193875253321324,
      "grad_norm": 1.4546953439712524,
      "learning_rate": 4.001795645620088e-07,
      "loss": 1.8098,
      "step": 10655
    },
    {
      "epoch": 1.9195676649403288,
      "grad_norm": 1.3311939239501953,
      "learning_rate": 3.983934336833028e-07,
      "loss": 1.6202,
      "step": 10656
    },
    {
      "epoch": 1.9197478045485252,
      "grad_norm": 1.642439842224121,
      "learning_rate": 3.9661128182131855e-07,
      "loss": 1.9907,
      "step": 10657
    },
    {
      "epoch": 1.9199279441567214,
      "grad_norm": 1.6593586206436157,
      "learning_rate": 3.9483310911901384e-07,
      "loss": 1.625,
      "step": 10658
    },
    {
      "epoch": 1.9201080837649178,
      "grad_norm": 1.6188393831253052,
      "learning_rate": 3.930589157190356e-07,
      "loss": 1.8402,
      "step": 10659
    },
    {
      "epoch": 1.920288223373114,
      "grad_norm": 1.7021390199661255,
      "learning_rate": 3.91288701763709e-07,
      "loss": 2.1461,
      "step": 10660
    },
    {
      "epoch": 1.9204683629813104,
      "grad_norm": 1.8295954465866089,
      "learning_rate": 3.895224673950426e-07,
      "loss": 1.4995,
      "step": 10661
    },
    {
      "epoch": 1.9206485025895068,
      "grad_norm": 1.4977326393127441,
      "learning_rate": 3.8776021275471753e-07,
      "loss": 1.4413,
      "step": 10662
    },
    {
      "epoch": 1.9208286421977032,
      "grad_norm": 1.2695984840393066,
      "learning_rate": 3.8600193798409847e-07,
      "loss": 1.2023,
      "step": 10663
    },
    {
      "epoch": 1.9210087818058996,
      "grad_norm": 1.5357704162597656,
      "learning_rate": 3.8424764322424476e-07,
      "loss": 1.4993,
      "step": 10664
    },
    {
      "epoch": 1.921188921414096,
      "grad_norm": 1.516182541847229,
      "learning_rate": 3.824973286158717e-07,
      "loss": 1.6087,
      "step": 10665
    },
    {
      "epoch": 1.9213690610222924,
      "grad_norm": 1.377296805381775,
      "learning_rate": 3.8075099429940034e-07,
      "loss": 1.3571,
      "step": 10666
    },
    {
      "epoch": 1.9215492006304886,
      "grad_norm": 1.4332109689712524,
      "learning_rate": 3.790086404149129e-07,
      "loss": 1.5036,
      "step": 10667
    },
    {
      "epoch": 1.921729340238685,
      "grad_norm": 1.3980786800384521,
      "learning_rate": 3.7727026710218103e-07,
      "loss": 1.4707,
      "step": 10668
    },
    {
      "epoch": 1.9219094798468812,
      "grad_norm": 1.4621315002441406,
      "learning_rate": 3.7553587450065984e-07,
      "loss": 1.3877,
      "step": 10669
    },
    {
      "epoch": 1.9220896194550776,
      "grad_norm": 1.5376617908477783,
      "learning_rate": 3.738054627494825e-07,
      "loss": 1.42,
      "step": 10670
    },
    {
      "epoch": 1.922269759063274,
      "grad_norm": 1.4621822834014893,
      "learning_rate": 3.720790319874545e-07,
      "loss": 1.2867,
      "step": 10671
    },
    {
      "epoch": 1.9224498986714704,
      "grad_norm": 1.4044430255889893,
      "learning_rate": 3.7035658235307634e-07,
      "loss": 1.3123,
      "step": 10672
    },
    {
      "epoch": 1.9226300382796668,
      "grad_norm": 1.7046087980270386,
      "learning_rate": 3.686381139845152e-07,
      "loss": 1.5982,
      "step": 10673
    },
    {
      "epoch": 1.9228101778878632,
      "grad_norm": 1.3808777332305908,
      "learning_rate": 3.66923627019633e-07,
      "loss": 1.3092,
      "step": 10674
    },
    {
      "epoch": 1.9229903174960594,
      "grad_norm": 1.6141752004623413,
      "learning_rate": 3.652131215959642e-07,
      "loss": 1.4804,
      "step": 10675
    },
    {
      "epoch": 1.9231704571042558,
      "grad_norm": 1.6133005619049072,
      "learning_rate": 3.635065978507213e-07,
      "loss": 1.6464,
      "step": 10676
    },
    {
      "epoch": 1.923350596712452,
      "grad_norm": 1.408409595489502,
      "learning_rate": 3.6180405592080027e-07,
      "loss": 1.3072,
      "step": 10677
    },
    {
      "epoch": 1.9235307363206484,
      "grad_norm": 1.482015609741211,
      "learning_rate": 3.6010549594278073e-07,
      "loss": 1.6211,
      "step": 10678
    },
    {
      "epoch": 1.9237108759288448,
      "grad_norm": 1.5925298929214478,
      "learning_rate": 3.584109180529205e-07,
      "loss": 1.6316,
      "step": 10679
    },
    {
      "epoch": 1.9238910155370412,
      "grad_norm": 1.5678397417068481,
      "learning_rate": 3.567203223871607e-07,
      "loss": 1.2776,
      "step": 10680
    },
    {
      "epoch": 1.9240711551452376,
      "grad_norm": 1.5976370573043823,
      "learning_rate": 3.550337090811096e-07,
      "loss": 1.5719,
      "step": 10681
    },
    {
      "epoch": 1.924251294753434,
      "grad_norm": 1.414941430091858,
      "learning_rate": 3.533510782700755e-07,
      "loss": 1.323,
      "step": 10682
    },
    {
      "epoch": 1.9244314343616302,
      "grad_norm": 1.5755207538604736,
      "learning_rate": 3.5167243008903396e-07,
      "loss": 1.5761,
      "step": 10683
    },
    {
      "epoch": 1.9246115739698266,
      "grad_norm": 1.5255422592163086,
      "learning_rate": 3.499977646726493e-07,
      "loss": 1.5319,
      "step": 10684
    },
    {
      "epoch": 1.9247917135780228,
      "grad_norm": 1.579970121383667,
      "learning_rate": 3.4832708215525865e-07,
      "loss": 1.4748,
      "step": 10685
    },
    {
      "epoch": 1.9249718531862192,
      "grad_norm": 1.544723391532898,
      "learning_rate": 3.4666038267088807e-07,
      "loss": 1.3649,
      "step": 10686
    },
    {
      "epoch": 1.9251519927944156,
      "grad_norm": 1.4438368082046509,
      "learning_rate": 3.449976663532362e-07,
      "loss": 1.0696,
      "step": 10687
    },
    {
      "epoch": 1.925332132402612,
      "grad_norm": 1.4790719747543335,
      "learning_rate": 3.4333893333568514e-07,
      "loss": 1.22,
      "step": 10688
    },
    {
      "epoch": 1.9255122720108084,
      "grad_norm": 1.6829428672790527,
      "learning_rate": 3.416841837512952e-07,
      "loss": 1.6393,
      "step": 10689
    },
    {
      "epoch": 1.9256924116190048,
      "grad_norm": 1.6428983211517334,
      "learning_rate": 3.400334177328157e-07,
      "loss": 1.4522,
      "step": 10690
    },
    {
      "epoch": 1.9258725512272012,
      "grad_norm": 1.5570341348648071,
      "learning_rate": 3.3838663541266304e-07,
      "loss": 1.2209,
      "step": 10691
    },
    {
      "epoch": 1.9260526908353974,
      "grad_norm": 1.540089726448059,
      "learning_rate": 3.367438369229481e-07,
      "loss": 1.3248,
      "step": 10692
    },
    {
      "epoch": 1.9262328304435938,
      "grad_norm": 1.6681678295135498,
      "learning_rate": 3.3510502239544885e-07,
      "loss": 1.5186,
      "step": 10693
    },
    {
      "epoch": 1.92641297005179,
      "grad_norm": 1.493974208831787,
      "learning_rate": 3.3347019196163787e-07,
      "loss": 1.293,
      "step": 10694
    },
    {
      "epoch": 1.9265931096599864,
      "grad_norm": 1.546829104423523,
      "learning_rate": 3.3183934575265473e-07,
      "loss": 1.4679,
      "step": 10695
    },
    {
      "epoch": 1.9267732492681828,
      "grad_norm": 1.701459527015686,
      "learning_rate": 3.302124838993281e-07,
      "loss": 1.5203,
      "step": 10696
    },
    {
      "epoch": 1.9269533888763792,
      "grad_norm": 1.6763008832931519,
      "learning_rate": 3.285896065321592e-07,
      "loss": 1.3289,
      "step": 10697
    },
    {
      "epoch": 1.9271335284845756,
      "grad_norm": 1.5818840265274048,
      "learning_rate": 3.269707137813438e-07,
      "loss": 1.3309,
      "step": 10698
    },
    {
      "epoch": 1.927313668092772,
      "grad_norm": 1.643554925918579,
      "learning_rate": 3.253558057767392e-07,
      "loss": 1.4619,
      "step": 10699
    },
    {
      "epoch": 1.9274938077009682,
      "grad_norm": 1.4544090032577515,
      "learning_rate": 3.237448826478973e-07,
      "loss": 1.1499,
      "step": 10700
    },
    {
      "epoch": 1.9276739473091646,
      "grad_norm": 1.4293798208236694,
      "learning_rate": 3.2213794452404245e-07,
      "loss": 1.2158,
      "step": 10701
    },
    {
      "epoch": 1.9278540869173608,
      "grad_norm": 1.4748966693878174,
      "learning_rate": 3.2053499153408273e-07,
      "loss": 2.0419,
      "step": 10702
    },
    {
      "epoch": 1.9280342265255572,
      "grad_norm": 1.3608410358428955,
      "learning_rate": 3.1893602380661523e-07,
      "loss": 1.7188,
      "step": 10703
    },
    {
      "epoch": 1.9282143661337536,
      "grad_norm": 1.4280856847763062,
      "learning_rate": 3.17341041469893e-07,
      "loss": 1.6843,
      "step": 10704
    },
    {
      "epoch": 1.92839450574195,
      "grad_norm": 1.4475690126419067,
      "learning_rate": 3.1575004465187466e-07,
      "loss": 1.7292,
      "step": 10705
    },
    {
      "epoch": 1.9285746453501464,
      "grad_norm": 1.3864495754241943,
      "learning_rate": 3.141630334801915e-07,
      "loss": 1.7678,
      "step": 10706
    },
    {
      "epoch": 1.9287547849583428,
      "grad_norm": 1.4281917810440063,
      "learning_rate": 3.125800080821528e-07,
      "loss": 1.78,
      "step": 10707
    },
    {
      "epoch": 1.928934924566539,
      "grad_norm": 1.5134031772613525,
      "learning_rate": 3.1100096858473483e-07,
      "loss": 1.6078,
      "step": 10708
    },
    {
      "epoch": 1.9291150641747354,
      "grad_norm": 1.6015766859054565,
      "learning_rate": 3.094259151146195e-07,
      "loss": 1.9317,
      "step": 10709
    },
    {
      "epoch": 1.9292952037829316,
      "grad_norm": 1.6528242826461792,
      "learning_rate": 3.0785484779815576e-07,
      "loss": 1.6737,
      "step": 10710
    },
    {
      "epoch": 1.929475343391128,
      "grad_norm": 1.6990431547164917,
      "learning_rate": 3.062877667613706e-07,
      "loss": 2.1245,
      "step": 10711
    },
    {
      "epoch": 1.9296554829993244,
      "grad_norm": 1.7260316610336304,
      "learning_rate": 3.0472467212997456e-07,
      "loss": 1.7464,
      "step": 10712
    },
    {
      "epoch": 1.9298356226075208,
      "grad_norm": 1.4424936771392822,
      "learning_rate": 3.031655640293618e-07,
      "loss": 1.4541,
      "step": 10713
    },
    {
      "epoch": 1.9300157622157172,
      "grad_norm": 1.5798828601837158,
      "learning_rate": 3.0161044258459337e-07,
      "loss": 1.4402,
      "step": 10714
    },
    {
      "epoch": 1.9301959018239137,
      "grad_norm": 1.3193916082382202,
      "learning_rate": 3.000593079204361e-07,
      "loss": 1.3844,
      "step": 10715
    },
    {
      "epoch": 1.9303760414321098,
      "grad_norm": 1.3552300930023193,
      "learning_rate": 2.9851216016131276e-07,
      "loss": 1.3721,
      "step": 10716
    },
    {
      "epoch": 1.9305561810403062,
      "grad_norm": 1.4207758903503418,
      "learning_rate": 2.969689994313352e-07,
      "loss": 1.2524,
      "step": 10717
    },
    {
      "epoch": 1.9307363206485024,
      "grad_norm": 1.4526771306991577,
      "learning_rate": 2.954298258542931e-07,
      "loss": 1.4771,
      "step": 10718
    },
    {
      "epoch": 1.9309164602566988,
      "grad_norm": 1.4651854038238525,
      "learning_rate": 2.9389463955366016e-07,
      "loss": 1.4377,
      "step": 10719
    },
    {
      "epoch": 1.9310965998648952,
      "grad_norm": 1.4527208805084229,
      "learning_rate": 2.923634406525877e-07,
      "loss": 1.3331,
      "step": 10720
    },
    {
      "epoch": 1.9312767394730916,
      "grad_norm": 1.3058305978775024,
      "learning_rate": 2.908362292739109e-07,
      "loss": 1.1443,
      "step": 10721
    },
    {
      "epoch": 1.931456879081288,
      "grad_norm": 1.3540574312210083,
      "learning_rate": 2.8931300554013717e-07,
      "loss": 1.3873,
      "step": 10722
    },
    {
      "epoch": 1.9316370186894845,
      "grad_norm": 1.4709054231643677,
      "learning_rate": 2.8779376957346335e-07,
      "loss": 1.438,
      "step": 10723
    },
    {
      "epoch": 1.9318171582976809,
      "grad_norm": 1.6403948068618774,
      "learning_rate": 2.8627852149575306e-07,
      "loss": 1.538,
      "step": 10724
    },
    {
      "epoch": 1.931997297905877,
      "grad_norm": 1.4392701387405396,
      "learning_rate": 2.8476726142857013e-07,
      "loss": 1.4715,
      "step": 10725
    },
    {
      "epoch": 1.9321774375140734,
      "grad_norm": 1.5689250230789185,
      "learning_rate": 2.8325998949314536e-07,
      "loss": 1.4759,
      "step": 10726
    },
    {
      "epoch": 1.9323575771222696,
      "grad_norm": 1.5069090127944946,
      "learning_rate": 2.8175670581038206e-07,
      "loss": 1.2367,
      "step": 10727
    },
    {
      "epoch": 1.932537716730466,
      "grad_norm": 1.6274744272232056,
      "learning_rate": 2.8025741050088375e-07,
      "loss": 1.5639,
      "step": 10728
    },
    {
      "epoch": 1.9327178563386624,
      "grad_norm": 1.5826215744018555,
      "learning_rate": 2.787621036849153e-07,
      "loss": 1.4002,
      "step": 10729
    },
    {
      "epoch": 1.9328979959468588,
      "grad_norm": 1.3495070934295654,
      "learning_rate": 2.772707854824308e-07,
      "loss": 1.2874,
      "step": 10730
    },
    {
      "epoch": 1.9330781355550553,
      "grad_norm": 1.550771951675415,
      "learning_rate": 2.7578345601306785e-07,
      "loss": 1.5333,
      "step": 10731
    },
    {
      "epoch": 1.9332582751632517,
      "grad_norm": 1.5069234371185303,
      "learning_rate": 2.7430011539613667e-07,
      "loss": 1.4189,
      "step": 10732
    },
    {
      "epoch": 1.9334384147714478,
      "grad_norm": 1.568824052810669,
      "learning_rate": 2.7282076375062524e-07,
      "loss": 1.2596,
      "step": 10733
    },
    {
      "epoch": 1.9336185543796442,
      "grad_norm": 1.6439555883407593,
      "learning_rate": 2.713454011952166e-07,
      "loss": 1.5865,
      "step": 10734
    },
    {
      "epoch": 1.9337986939878404,
      "grad_norm": 1.3567376136779785,
      "learning_rate": 2.6987402784825476e-07,
      "loss": 1.1757,
      "step": 10735
    },
    {
      "epoch": 1.9339788335960368,
      "grad_norm": 1.6360039710998535,
      "learning_rate": 2.684066438277788e-07,
      "loss": 1.5079,
      "step": 10736
    },
    {
      "epoch": 1.9341589732042332,
      "grad_norm": 1.4489951133728027,
      "learning_rate": 2.669432492514945e-07,
      "loss": 1.3818,
      "step": 10737
    },
    {
      "epoch": 1.9343391128124297,
      "grad_norm": 1.4558089971542358,
      "learning_rate": 2.6548384423679684e-07,
      "loss": 1.2452,
      "step": 10738
    },
    {
      "epoch": 1.934519252420626,
      "grad_norm": 1.5748984813690186,
      "learning_rate": 2.640284289007644e-07,
      "loss": 1.3351,
      "step": 10739
    },
    {
      "epoch": 1.9346993920288225,
      "grad_norm": 1.6168311834335327,
      "learning_rate": 2.6257700336014824e-07,
      "loss": 1.5136,
      "step": 10740
    },
    {
      "epoch": 1.9348795316370186,
      "grad_norm": 1.6417700052261353,
      "learning_rate": 2.611295677313774e-07,
      "loss": 1.5047,
      "step": 10741
    },
    {
      "epoch": 1.935059671245215,
      "grad_norm": 1.5892904996871948,
      "learning_rate": 2.596861221305591e-07,
      "loss": 1.5453,
      "step": 10742
    },
    {
      "epoch": 1.9352398108534112,
      "grad_norm": 1.641786813735962,
      "learning_rate": 2.582466666735006e-07,
      "loss": 1.5017,
      "step": 10743
    },
    {
      "epoch": 1.9354199504616076,
      "grad_norm": 1.4978071451187134,
      "learning_rate": 2.5681120147566517e-07,
      "loss": 1.3077,
      "step": 10744
    },
    {
      "epoch": 1.935600090069804,
      "grad_norm": 1.4038392305374146,
      "learning_rate": 2.5537972665219955e-07,
      "loss": 1.1902,
      "step": 10745
    },
    {
      "epoch": 1.9357802296780005,
      "grad_norm": 1.3287270069122314,
      "learning_rate": 2.5395224231795076e-07,
      "loss": 1.1484,
      "step": 10746
    },
    {
      "epoch": 1.9359603692861969,
      "grad_norm": 1.5975278615951538,
      "learning_rate": 2.5252874858741614e-07,
      "loss": 1.4272,
      "step": 10747
    },
    {
      "epoch": 1.9361405088943933,
      "grad_norm": 1.56769597530365,
      "learning_rate": 2.511092455747932e-07,
      "loss": 1.4345,
      "step": 10748
    },
    {
      "epoch": 1.9363206485025897,
      "grad_norm": 1.6250433921813965,
      "learning_rate": 2.4969373339396307e-07,
      "loss": 1.2983,
      "step": 10749
    },
    {
      "epoch": 1.9365007881107859,
      "grad_norm": 1.6011165380477905,
      "learning_rate": 2.482822121584627e-07,
      "loss": 1.3162,
      "step": 10750
    },
    {
      "epoch": 1.9366809277189823,
      "grad_norm": 1.5798112154006958,
      "learning_rate": 2.4687468198153487e-07,
      "loss": 1.2486,
      "step": 10751
    },
    {
      "epoch": 1.9368610673271784,
      "grad_norm": 1.4361294507980347,
      "learning_rate": 2.454711429760781e-07,
      "loss": 1.9416,
      "step": 10752
    },
    {
      "epoch": 1.9370412069353748,
      "grad_norm": 1.3630291223526,
      "learning_rate": 2.4407159525469683e-07,
      "loss": 1.6923,
      "step": 10753
    },
    {
      "epoch": 1.9372213465435713,
      "grad_norm": 1.381807804107666,
      "learning_rate": 2.426760389296623e-07,
      "loss": 1.8007,
      "step": 10754
    },
    {
      "epoch": 1.9374014861517677,
      "grad_norm": 1.340575098991394,
      "learning_rate": 2.4128447411291275e-07,
      "loss": 1.6226,
      "step": 10755
    },
    {
      "epoch": 1.937581625759964,
      "grad_norm": 1.5154651403427124,
      "learning_rate": 2.3989690091608673e-07,
      "loss": 1.9511,
      "step": 10756
    },
    {
      "epoch": 1.9377617653681605,
      "grad_norm": 1.347575306892395,
      "learning_rate": 2.3851331945049517e-07,
      "loss": 1.4401,
      "step": 10757
    },
    {
      "epoch": 1.9379419049763567,
      "grad_norm": 1.4433164596557617,
      "learning_rate": 2.37133729827127e-07,
      "loss": 1.8063,
      "step": 10758
    },
    {
      "epoch": 1.938122044584553,
      "grad_norm": 1.5673457384109497,
      "learning_rate": 2.3575813215664934e-07,
      "loss": 1.6251,
      "step": 10759
    },
    {
      "epoch": 1.9383021841927492,
      "grad_norm": 1.8343981504440308,
      "learning_rate": 2.3438652654941828e-07,
      "loss": 1.8027,
      "step": 10760
    },
    {
      "epoch": 1.9384823238009457,
      "grad_norm": 1.6909205913543701,
      "learning_rate": 2.3301891311546253e-07,
      "loss": 1.7972,
      "step": 10761
    },
    {
      "epoch": 1.938662463409142,
      "grad_norm": 1.730273962020874,
      "learning_rate": 2.3165529196448877e-07,
      "loss": 1.4655,
      "step": 10762
    },
    {
      "epoch": 1.9388426030173385,
      "grad_norm": 1.3355976343154907,
      "learning_rate": 2.302956632058817e-07,
      "loss": 1.4093,
      "step": 10763
    },
    {
      "epoch": 1.9390227426255349,
      "grad_norm": 1.4256105422973633,
      "learning_rate": 2.289400269487152e-07,
      "loss": 1.4884,
      "step": 10764
    },
    {
      "epoch": 1.9392028822337313,
      "grad_norm": 1.3288540840148926,
      "learning_rate": 2.2758838330174114e-07,
      "loss": 1.181,
      "step": 10765
    },
    {
      "epoch": 1.9393830218419275,
      "grad_norm": 1.348707914352417,
      "learning_rate": 2.262407323733784e-07,
      "loss": 1.1041,
      "step": 10766
    },
    {
      "epoch": 1.9395631614501239,
      "grad_norm": 1.5383409261703491,
      "learning_rate": 2.24897074271746e-07,
      "loss": 1.5553,
      "step": 10767
    },
    {
      "epoch": 1.93974330105832,
      "grad_norm": 1.4523781538009644,
      "learning_rate": 2.2355740910461886e-07,
      "loss": 1.3102,
      "step": 10768
    },
    {
      "epoch": 1.9399234406665165,
      "grad_norm": 1.2538048028945923,
      "learning_rate": 2.2222173697947768e-07,
      "loss": 0.9925,
      "step": 10769
    },
    {
      "epoch": 1.9401035802747129,
      "grad_norm": 1.3417613506317139,
      "learning_rate": 2.2089005800345898e-07,
      "loss": 1.198,
      "step": 10770
    },
    {
      "epoch": 1.9402837198829093,
      "grad_norm": 1.3026797771453857,
      "learning_rate": 2.1956237228339395e-07,
      "loss": 1.1336,
      "step": 10771
    },
    {
      "epoch": 1.9404638594911057,
      "grad_norm": 1.6009228229522705,
      "learning_rate": 2.1823867992579184e-07,
      "loss": 1.5021,
      "step": 10772
    },
    {
      "epoch": 1.940643999099302,
      "grad_norm": 1.5452401638031006,
      "learning_rate": 2.1691898103682883e-07,
      "loss": 1.4621,
      "step": 10773
    },
    {
      "epoch": 1.9408241387074983,
      "grad_norm": 1.6131672859191895,
      "learning_rate": 2.1560327572238138e-07,
      "loss": 1.4616,
      "step": 10774
    },
    {
      "epoch": 1.9410042783156947,
      "grad_norm": 1.4504743814468384,
      "learning_rate": 2.1429156408798722e-07,
      "loss": 1.2417,
      "step": 10775
    },
    {
      "epoch": 1.941184417923891,
      "grad_norm": 1.3496614694595337,
      "learning_rate": 2.1298384623887334e-07,
      "loss": 1.32,
      "step": 10776
    },
    {
      "epoch": 1.9413645575320873,
      "grad_norm": 1.4966177940368652,
      "learning_rate": 2.1168012227995026e-07,
      "loss": 1.4228,
      "step": 10777
    },
    {
      "epoch": 1.9415446971402837,
      "grad_norm": 1.5653752088546753,
      "learning_rate": 2.1038039231578987e-07,
      "loss": 1.505,
      "step": 10778
    },
    {
      "epoch": 1.94172483674848,
      "grad_norm": 1.335074782371521,
      "learning_rate": 2.0908465645066432e-07,
      "loss": 1.2096,
      "step": 10779
    },
    {
      "epoch": 1.9419049763566765,
      "grad_norm": 1.5789923667907715,
      "learning_rate": 2.0779291478851826e-07,
      "loss": 1.3691,
      "step": 10780
    },
    {
      "epoch": 1.9420851159648729,
      "grad_norm": 1.6408206224441528,
      "learning_rate": 2.0650516743297432e-07,
      "loss": 1.573,
      "step": 10781
    },
    {
      "epoch": 1.9422652555730693,
      "grad_norm": 1.667970895767212,
      "learning_rate": 2.0522141448732768e-07,
      "loss": 1.5264,
      "step": 10782
    },
    {
      "epoch": 1.9424453951812655,
      "grad_norm": 1.3787291049957275,
      "learning_rate": 2.0394165605456262e-07,
      "loss": 1.5143,
      "step": 10783
    },
    {
      "epoch": 1.9426255347894619,
      "grad_norm": 1.527496099472046,
      "learning_rate": 2.02665892237347e-07,
      "loss": 1.5217,
      "step": 10784
    },
    {
      "epoch": 1.942805674397658,
      "grad_norm": 1.5242236852645874,
      "learning_rate": 2.013941231380212e-07,
      "loss": 1.2894,
      "step": 10785
    },
    {
      "epoch": 1.9429858140058545,
      "grad_norm": 1.5073187351226807,
      "learning_rate": 2.0012634885859804e-07,
      "loss": 1.48,
      "step": 10786
    },
    {
      "epoch": 1.9431659536140509,
      "grad_norm": 1.4628822803497314,
      "learning_rate": 1.9886256950078507e-07,
      "loss": 1.3702,
      "step": 10787
    },
    {
      "epoch": 1.9433460932222473,
      "grad_norm": 1.5349504947662354,
      "learning_rate": 1.9760278516595677e-07,
      "loss": 1.3916,
      "step": 10788
    },
    {
      "epoch": 1.9435262328304437,
      "grad_norm": 1.3617768287658691,
      "learning_rate": 1.963469959551767e-07,
      "loss": 1.2469,
      "step": 10789
    },
    {
      "epoch": 1.94370637243864,
      "grad_norm": 1.4949501752853394,
      "learning_rate": 1.9509520196918095e-07,
      "loss": 1.3154,
      "step": 10790
    },
    {
      "epoch": 1.9438865120468363,
      "grad_norm": 1.5577994585037231,
      "learning_rate": 1.9384740330838925e-07,
      "loss": 1.4972,
      "step": 10791
    },
    {
      "epoch": 1.9440666516550327,
      "grad_norm": 1.5346155166625977,
      "learning_rate": 1.9260360007289923e-07,
      "loss": 1.3894,
      "step": 10792
    },
    {
      "epoch": 1.9442467912632289,
      "grad_norm": 1.5204501152038574,
      "learning_rate": 1.9136379236249225e-07,
      "loss": 1.3432,
      "step": 10793
    },
    {
      "epoch": 1.9444269308714253,
      "grad_norm": 1.6208473443984985,
      "learning_rate": 1.901279802766165e-07,
      "loss": 1.4404,
      "step": 10794
    },
    {
      "epoch": 1.9446070704796217,
      "grad_norm": 1.6193093061447144,
      "learning_rate": 1.888961639144149e-07,
      "loss": 1.3629,
      "step": 10795
    },
    {
      "epoch": 1.944787210087818,
      "grad_norm": 1.4994977712631226,
      "learning_rate": 1.876683433746973e-07,
      "loss": 1.3601,
      "step": 10796
    },
    {
      "epoch": 1.9449673496960145,
      "grad_norm": 1.4976441860198975,
      "learning_rate": 1.864445187559627e-07,
      "loss": 1.3416,
      "step": 10797
    },
    {
      "epoch": 1.9451474893042109,
      "grad_norm": 1.5912138223648071,
      "learning_rate": 1.852246901563881e-07,
      "loss": 1.2986,
      "step": 10798
    },
    {
      "epoch": 1.945327628912407,
      "grad_norm": 1.664944052696228,
      "learning_rate": 1.8400885767382302e-07,
      "loss": 1.3124,
      "step": 10799
    },
    {
      "epoch": 1.9455077685206035,
      "grad_norm": 1.4616655111312866,
      "learning_rate": 1.8279702140580612e-07,
      "loss": 1.1692,
      "step": 10800
    },
    {
      "epoch": 1.9456879081287997,
      "grad_norm": 1.4720104932785034,
      "learning_rate": 1.8158918144954296e-07,
      "loss": 1.2591,
      "step": 10801
    },
    {
      "epoch": 1.945868047736996,
      "grad_norm": 1.329436182975769,
      "learning_rate": 1.803853379019338e-07,
      "loss": 1.5564,
      "step": 10802
    },
    {
      "epoch": 1.9460481873451925,
      "grad_norm": 1.4360246658325195,
      "learning_rate": 1.7918549085954582e-07,
      "loss": 1.913,
      "step": 10803
    },
    {
      "epoch": 1.9462283269533889,
      "grad_norm": 1.4227567911148071,
      "learning_rate": 1.7798964041862987e-07,
      "loss": 1.7613,
      "step": 10804
    },
    {
      "epoch": 1.9464084665615853,
      "grad_norm": 1.3533034324645996,
      "learning_rate": 1.7679778667512026e-07,
      "loss": 1.7592,
      "step": 10805
    },
    {
      "epoch": 1.9465886061697817,
      "grad_norm": 1.3975611925125122,
      "learning_rate": 1.7560992972461832e-07,
      "loss": 1.7061,
      "step": 10806
    },
    {
      "epoch": 1.946768745777978,
      "grad_norm": 1.4783165454864502,
      "learning_rate": 1.7442606966242004e-07,
      "loss": 1.8563,
      "step": 10807
    },
    {
      "epoch": 1.9469488853861743,
      "grad_norm": 1.3704092502593994,
      "learning_rate": 1.7324620658349945e-07,
      "loss": 1.3216,
      "step": 10808
    },
    {
      "epoch": 1.9471290249943707,
      "grad_norm": 1.3841757774353027,
      "learning_rate": 1.7207034058249195e-07,
      "loss": 1.3249,
      "step": 10809
    },
    {
      "epoch": 1.9473091646025669,
      "grad_norm": 1.4620943069458008,
      "learning_rate": 1.7089847175373874e-07,
      "loss": 1.6252,
      "step": 10810
    },
    {
      "epoch": 1.9474893042107633,
      "grad_norm": 1.6428823471069336,
      "learning_rate": 1.6973060019123132e-07,
      "loss": 1.807,
      "step": 10811
    },
    {
      "epoch": 1.9476694438189597,
      "grad_norm": 1.942167043685913,
      "learning_rate": 1.6856672598866695e-07,
      "loss": 1.6585,
      "step": 10812
    },
    {
      "epoch": 1.947849583427156,
      "grad_norm": 1.476116418838501,
      "learning_rate": 1.6740684923940986e-07,
      "loss": 1.6478,
      "step": 10813
    },
    {
      "epoch": 1.9480297230353525,
      "grad_norm": 1.370652437210083,
      "learning_rate": 1.662509700364967e-07,
      "loss": 1.298,
      "step": 10814
    },
    {
      "epoch": 1.948209862643549,
      "grad_norm": 1.313159704208374,
      "learning_rate": 1.650990884726644e-07,
      "loss": 1.4107,
      "step": 10815
    },
    {
      "epoch": 1.948390002251745,
      "grad_norm": 1.7205506563186646,
      "learning_rate": 1.6395120464030578e-07,
      "loss": 1.6049,
      "step": 10816
    },
    {
      "epoch": 1.9485701418599415,
      "grad_norm": 1.349006175994873,
      "learning_rate": 1.6280731863150822e-07,
      "loss": 1.3423,
      "step": 10817
    },
    {
      "epoch": 1.9487502814681377,
      "grad_norm": 1.4364091157913208,
      "learning_rate": 1.616674305380317e-07,
      "loss": 1.4075,
      "step": 10818
    },
    {
      "epoch": 1.948930421076334,
      "grad_norm": 1.4425793886184692,
      "learning_rate": 1.605315404513197e-07,
      "loss": 1.3551,
      "step": 10819
    },
    {
      "epoch": 1.9491105606845305,
      "grad_norm": 1.3886240720748901,
      "learning_rate": 1.5939964846249378e-07,
      "loss": 1.336,
      "step": 10820
    },
    {
      "epoch": 1.9492907002927269,
      "grad_norm": 1.368329405784607,
      "learning_rate": 1.58271754662348e-07,
      "loss": 1.4743,
      "step": 10821
    },
    {
      "epoch": 1.9494708399009233,
      "grad_norm": 1.42078697681427,
      "learning_rate": 1.5714785914136553e-07,
      "loss": 1.3425,
      "step": 10822
    },
    {
      "epoch": 1.9496509795091197,
      "grad_norm": 1.4186526536941528,
      "learning_rate": 1.560279619897076e-07,
      "loss": 1.2651,
      "step": 10823
    },
    {
      "epoch": 1.9498311191173159,
      "grad_norm": 1.5221070051193237,
      "learning_rate": 1.549120632972023e-07,
      "loss": 1.5366,
      "step": 10824
    },
    {
      "epoch": 1.9500112587255123,
      "grad_norm": 1.567468285560608,
      "learning_rate": 1.5380016315337808e-07,
      "loss": 1.3349,
      "step": 10825
    },
    {
      "epoch": 1.9501913983337085,
      "grad_norm": 1.3771719932556152,
      "learning_rate": 1.5269226164742468e-07,
      "loss": 1.2236,
      "step": 10826
    },
    {
      "epoch": 1.9503715379419049,
      "grad_norm": 1.568207025527954,
      "learning_rate": 1.5158835886821544e-07,
      "loss": 1.4075,
      "step": 10827
    },
    {
      "epoch": 1.9505516775501013,
      "grad_norm": 1.3922921419143677,
      "learning_rate": 1.5048845490431286e-07,
      "loss": 1.2534,
      "step": 10828
    },
    {
      "epoch": 1.9507318171582977,
      "grad_norm": 1.7033252716064453,
      "learning_rate": 1.493925498439408e-07,
      "loss": 1.6314,
      "step": 10829
    },
    {
      "epoch": 1.950911956766494,
      "grad_norm": 1.4187091588974,
      "learning_rate": 1.4830064377501784e-07,
      "loss": 1.0673,
      "step": 10830
    },
    {
      "epoch": 1.9510920963746905,
      "grad_norm": 1.4200152158737183,
      "learning_rate": 1.4721273678513504e-07,
      "loss": 1.4227,
      "step": 10831
    },
    {
      "epoch": 1.9512722359828867,
      "grad_norm": 1.5980957746505737,
      "learning_rate": 1.4612882896156698e-07,
      "loss": 1.367,
      "step": 10832
    },
    {
      "epoch": 1.951452375591083,
      "grad_norm": 1.3979201316833496,
      "learning_rate": 1.4504892039125528e-07,
      "loss": 1.4113,
      "step": 10833
    },
    {
      "epoch": 1.9516325151992795,
      "grad_norm": 1.4458646774291992,
      "learning_rate": 1.439730111608417e-07,
      "loss": 1.2033,
      "step": 10834
    },
    {
      "epoch": 1.9518126548074757,
      "grad_norm": 1.5067696571350098,
      "learning_rate": 1.4290110135662394e-07,
      "loss": 1.377,
      "step": 10835
    },
    {
      "epoch": 1.951992794415672,
      "grad_norm": 1.5077192783355713,
      "learning_rate": 1.418331910645998e-07,
      "loss": 1.3417,
      "step": 10836
    },
    {
      "epoch": 1.9521729340238685,
      "grad_norm": 1.4355510473251343,
      "learning_rate": 1.4076928037043413e-07,
      "loss": 1.4292,
      "step": 10837
    },
    {
      "epoch": 1.952353073632065,
      "grad_norm": 1.4707227945327759,
      "learning_rate": 1.3970936935946976e-07,
      "loss": 1.2702,
      "step": 10838
    },
    {
      "epoch": 1.9525332132402613,
      "grad_norm": 1.6267133951187134,
      "learning_rate": 1.3865345811673314e-07,
      "loss": 1.4466,
      "step": 10839
    },
    {
      "epoch": 1.9527133528484577,
      "grad_norm": 1.3773984909057617,
      "learning_rate": 1.3760154672692316e-07,
      "loss": 1.2227,
      "step": 10840
    },
    {
      "epoch": 1.952893492456654,
      "grad_norm": 1.6703267097473145,
      "learning_rate": 1.3655363527443343e-07,
      "loss": 1.4158,
      "step": 10841
    },
    {
      "epoch": 1.9530736320648503,
      "grad_norm": 1.5142369270324707,
      "learning_rate": 1.3550972384333004e-07,
      "loss": 1.3778,
      "step": 10842
    },
    {
      "epoch": 1.9532537716730465,
      "grad_norm": 1.5401586294174194,
      "learning_rate": 1.3446981251734048e-07,
      "loss": 1.2658,
      "step": 10843
    },
    {
      "epoch": 1.9534339112812429,
      "grad_norm": 1.498421549797058,
      "learning_rate": 1.3343390137989797e-07,
      "loss": 1.2119,
      "step": 10844
    },
    {
      "epoch": 1.9536140508894393,
      "grad_norm": 1.5495822429656982,
      "learning_rate": 1.3240199051409718e-07,
      "loss": 1.4924,
      "step": 10845
    },
    {
      "epoch": 1.9537941904976357,
      "grad_norm": 1.6219462156295776,
      "learning_rate": 1.3137408000272188e-07,
      "loss": 1.196,
      "step": 10846
    },
    {
      "epoch": 1.953974330105832,
      "grad_norm": 1.4840922355651855,
      "learning_rate": 1.303501699282228e-07,
      "loss": 1.1492,
      "step": 10847
    },
    {
      "epoch": 1.9541544697140285,
      "grad_norm": 1.5634247064590454,
      "learning_rate": 1.2933026037274532e-07,
      "loss": 1.3837,
      "step": 10848
    },
    {
      "epoch": 1.9543346093222247,
      "grad_norm": 1.4500490427017212,
      "learning_rate": 1.2831435141810179e-07,
      "loss": 1.1717,
      "step": 10849
    },
    {
      "epoch": 1.954514748930421,
      "grad_norm": 1.3409775495529175,
      "learning_rate": 1.2730244314579364e-07,
      "loss": 1.0463,
      "step": 10850
    },
    {
      "epoch": 1.9546948885386173,
      "grad_norm": 1.3928147554397583,
      "learning_rate": 1.2629453563698933e-07,
      "loss": 0.9688,
      "step": 10851
    },
    {
      "epoch": 1.9548750281468137,
      "grad_norm": 1.4606311321258545,
      "learning_rate": 1.2529062897254084e-07,
      "loss": 1.9747,
      "step": 10852
    },
    {
      "epoch": 1.95505516775501,
      "grad_norm": 1.2397841215133667,
      "learning_rate": 1.242907232329893e-07,
      "loss": 1.5706,
      "step": 10853
    },
    {
      "epoch": 1.9552353073632065,
      "grad_norm": 1.363538146018982,
      "learning_rate": 1.2329481849854273e-07,
      "loss": 1.7423,
      "step": 10854
    },
    {
      "epoch": 1.955415446971403,
      "grad_norm": 1.4047261476516724,
      "learning_rate": 1.2230291484908728e-07,
      "loss": 1.8596,
      "step": 10855
    },
    {
      "epoch": 1.9555955865795993,
      "grad_norm": 1.5757447481155396,
      "learning_rate": 1.213150123641982e-07,
      "loss": 1.9238,
      "step": 10856
    },
    {
      "epoch": 1.9557757261877955,
      "grad_norm": 1.3981974124908447,
      "learning_rate": 1.2033111112312868e-07,
      "loss": 1.5019,
      "step": 10857
    },
    {
      "epoch": 1.955955865795992,
      "grad_norm": 1.4073365926742554,
      "learning_rate": 1.1935121120479898e-07,
      "loss": 1.6148,
      "step": 10858
    },
    {
      "epoch": 1.956136005404188,
      "grad_norm": 1.6653339862823486,
      "learning_rate": 1.1837531268781843e-07,
      "loss": 1.8307,
      "step": 10859
    },
    {
      "epoch": 1.9563161450123845,
      "grad_norm": 1.809057354927063,
      "learning_rate": 1.1740341565047442e-07,
      "loss": 1.9895,
      "step": 10860
    },
    {
      "epoch": 1.956496284620581,
      "grad_norm": 1.9179248809814453,
      "learning_rate": 1.1643552017073233e-07,
      "loss": 2.0714,
      "step": 10861
    },
    {
      "epoch": 1.9566764242287773,
      "grad_norm": 1.498245120048523,
      "learning_rate": 1.1547162632623565e-07,
      "loss": 1.562,
      "step": 10862
    },
    {
      "epoch": 1.9568565638369737,
      "grad_norm": 1.3244699239730835,
      "learning_rate": 1.1451173419430583e-07,
      "loss": 1.4839,
      "step": 10863
    },
    {
      "epoch": 1.9570367034451701,
      "grad_norm": 1.4634143114089966,
      "learning_rate": 1.1355584385194795e-07,
      "loss": 1.4893,
      "step": 10864
    },
    {
      "epoch": 1.9572168430533665,
      "grad_norm": 1.5573850870132446,
      "learning_rate": 1.1260395537583956e-07,
      "loss": 1.7772,
      "step": 10865
    },
    {
      "epoch": 1.9573969826615627,
      "grad_norm": 1.4590612649917603,
      "learning_rate": 1.1165606884234181e-07,
      "loss": 1.3446,
      "step": 10866
    },
    {
      "epoch": 1.957577122269759,
      "grad_norm": 1.2843468189239502,
      "learning_rate": 1.1071218432749941e-07,
      "loss": 1.2392,
      "step": 10867
    },
    {
      "epoch": 1.9577572618779553,
      "grad_norm": 1.4397944211959839,
      "learning_rate": 1.0977230190701848e-07,
      "loss": 1.4551,
      "step": 10868
    },
    {
      "epoch": 1.9579374014861517,
      "grad_norm": 1.4734441041946411,
      "learning_rate": 1.0883642165630536e-07,
      "loss": 1.2826,
      "step": 10869
    },
    {
      "epoch": 1.958117541094348,
      "grad_norm": 1.4606026411056519,
      "learning_rate": 1.079045436504389e-07,
      "loss": 1.3579,
      "step": 10870
    },
    {
      "epoch": 1.9582976807025445,
      "grad_norm": 1.4172842502593994,
      "learning_rate": 1.0697666796415929e-07,
      "loss": 1.3519,
      "step": 10871
    },
    {
      "epoch": 1.958477820310741,
      "grad_norm": 1.3345050811767578,
      "learning_rate": 1.0605279467191254e-07,
      "loss": 1.3285,
      "step": 10872
    },
    {
      "epoch": 1.9586579599189373,
      "grad_norm": 1.2070847749710083,
      "learning_rate": 1.0513292384781159e-07,
      "loss": 1.0915,
      "step": 10873
    },
    {
      "epoch": 1.9588380995271335,
      "grad_norm": 1.311422348022461,
      "learning_rate": 1.0421705556564188e-07,
      "loss": 1.2526,
      "step": 10874
    },
    {
      "epoch": 1.95901823913533,
      "grad_norm": 1.4291080236434937,
      "learning_rate": 1.0330518989887794e-07,
      "loss": 1.2534,
      "step": 10875
    },
    {
      "epoch": 1.959198378743526,
      "grad_norm": 1.427431344985962,
      "learning_rate": 1.0239732692066684e-07,
      "loss": 1.3307,
      "step": 10876
    },
    {
      "epoch": 1.9593785183517225,
      "grad_norm": 1.4616867303848267,
      "learning_rate": 1.0149346670383919e-07,
      "loss": 1.2988,
      "step": 10877
    },
    {
      "epoch": 1.959558657959919,
      "grad_norm": 1.517516851425171,
      "learning_rate": 1.0059360932089812e-07,
      "loss": 1.5038,
      "step": 10878
    },
    {
      "epoch": 1.9597387975681153,
      "grad_norm": 1.4344525337219238,
      "learning_rate": 9.969775484403587e-08,
      "loss": 1.4157,
      "step": 10879
    },
    {
      "epoch": 1.9599189371763117,
      "grad_norm": 1.4902291297912598,
      "learning_rate": 9.880590334511163e-08,
      "loss": 1.3343,
      "step": 10880
    },
    {
      "epoch": 1.9600990767845081,
      "grad_norm": 1.385625958442688,
      "learning_rate": 9.791805489567374e-08,
      "loss": 1.3547,
      "step": 10881
    },
    {
      "epoch": 1.9602792163927043,
      "grad_norm": 1.5179256200790405,
      "learning_rate": 9.703420956694299e-08,
      "loss": 1.4067,
      "step": 10882
    },
    {
      "epoch": 1.9604593560009007,
      "grad_norm": 1.357257604598999,
      "learning_rate": 9.615436742981821e-08,
      "loss": 1.1885,
      "step": 10883
    },
    {
      "epoch": 1.960639495609097,
      "grad_norm": 1.3937796354293823,
      "learning_rate": 9.527852855488739e-08,
      "loss": 1.2516,
      "step": 10884
    },
    {
      "epoch": 1.9608196352172933,
      "grad_norm": 1.3831685781478882,
      "learning_rate": 9.44066930123999e-08,
      "loss": 1.0506,
      "step": 10885
    },
    {
      "epoch": 1.9609997748254897,
      "grad_norm": 1.5415778160095215,
      "learning_rate": 9.353886087230535e-08,
      "loss": 1.5944,
      "step": 10886
    },
    {
      "epoch": 1.9611799144336861,
      "grad_norm": 1.6939843893051147,
      "learning_rate": 9.267503220420915e-08,
      "loss": 1.4236,
      "step": 10887
    },
    {
      "epoch": 1.9613600540418825,
      "grad_norm": 1.6533327102661133,
      "learning_rate": 9.181520707741697e-08,
      "loss": 1.6249,
      "step": 10888
    },
    {
      "epoch": 1.961540193650079,
      "grad_norm": 1.6216720342636108,
      "learning_rate": 9.09593855608959e-08,
      "loss": 1.5197,
      "step": 10889
    },
    {
      "epoch": 1.9617203332582753,
      "grad_norm": 1.5947545766830444,
      "learning_rate": 9.010756772330764e-08,
      "loss": 1.4806,
      "step": 10890
    },
    {
      "epoch": 1.9619004728664715,
      "grad_norm": 1.623610258102417,
      "learning_rate": 8.925975363297534e-08,
      "loss": 1.3992,
      "step": 10891
    },
    {
      "epoch": 1.962080612474668,
      "grad_norm": 1.6965842247009277,
      "learning_rate": 8.841594335791681e-08,
      "loss": 1.4058,
      "step": 10892
    },
    {
      "epoch": 1.962260752082864,
      "grad_norm": 1.4621773958206177,
      "learning_rate": 8.757613696582234e-08,
      "loss": 1.2826,
      "step": 10893
    },
    {
      "epoch": 1.9624408916910605,
      "grad_norm": 1.6303375959396362,
      "learning_rate": 8.674033452405472e-08,
      "loss": 1.371,
      "step": 10894
    },
    {
      "epoch": 1.962621031299257,
      "grad_norm": 1.5197093486785889,
      "learning_rate": 8.590853609966587e-08,
      "loss": 1.3054,
      "step": 10895
    },
    {
      "epoch": 1.9628011709074533,
      "grad_norm": 1.6302917003631592,
      "learning_rate": 8.508074175938574e-08,
      "loss": 1.4609,
      "step": 10896
    },
    {
      "epoch": 1.9629813105156497,
      "grad_norm": 1.446485996246338,
      "learning_rate": 8.425695156961677e-08,
      "loss": 1.1313,
      "step": 10897
    },
    {
      "epoch": 1.9631614501238461,
      "grad_norm": 1.534677505493164,
      "learning_rate": 8.343716559643944e-08,
      "loss": 1.2485,
      "step": 10898
    },
    {
      "epoch": 1.9633415897320423,
      "grad_norm": 1.6634607315063477,
      "learning_rate": 8.26213839056178e-08,
      "loss": 1.3829,
      "step": 10899
    },
    {
      "epoch": 1.9635217293402387,
      "grad_norm": 1.5477266311645508,
      "learning_rate": 8.180960656259395e-08,
      "loss": 1.4037,
      "step": 10900
    },
    {
      "epoch": 1.963701868948435,
      "grad_norm": 1.4062843322753906,
      "learning_rate": 8.100183363249359e-08,
      "loss": 1.2043,
      "step": 10901
    },
    {
      "epoch": 1.9638820085566313,
      "grad_norm": 1.351867914199829,
      "learning_rate": 8.01980651801093e-08,
      "loss": 1.5283,
      "step": 10902
    },
    {
      "epoch": 1.9640621481648277,
      "grad_norm": 1.2529815435409546,
      "learning_rate": 7.939830126992286e-08,
      "loss": 1.6594,
      "step": 10903
    },
    {
      "epoch": 1.9642422877730241,
      "grad_norm": 1.3346374034881592,
      "learning_rate": 7.860254196608296e-08,
      "loss": 1.7968,
      "step": 10904
    },
    {
      "epoch": 1.9644224273812205,
      "grad_norm": 1.3766579627990723,
      "learning_rate": 7.781078733243852e-08,
      "loss": 1.7012,
      "step": 10905
    },
    {
      "epoch": 1.964602566989417,
      "grad_norm": 1.4812413454055786,
      "learning_rate": 7.702303743249428e-08,
      "loss": 1.7673,
      "step": 10906
    },
    {
      "epoch": 1.9647827065976131,
      "grad_norm": 1.4670212268829346,
      "learning_rate": 7.623929232944416e-08,
      "loss": 1.6641,
      "step": 10907
    },
    {
      "epoch": 1.9649628462058095,
      "grad_norm": 1.50018310546875,
      "learning_rate": 7.545955208616006e-08,
      "loss": 1.7005,
      "step": 10908
    },
    {
      "epoch": 1.9651429858140057,
      "grad_norm": 1.6693171262741089,
      "learning_rate": 7.468381676519198e-08,
      "loss": 1.7138,
      "step": 10909
    },
    {
      "epoch": 1.965323125422202,
      "grad_norm": 1.7613017559051514,
      "learning_rate": 7.391208642877345e-08,
      "loss": 1.9698,
      "step": 10910
    },
    {
      "epoch": 1.9655032650303985,
      "grad_norm": 1.6921327114105225,
      "learning_rate": 7.314436113881051e-08,
      "loss": 1.7468,
      "step": 10911
    },
    {
      "epoch": 1.965683404638595,
      "grad_norm": 1.4408345222473145,
      "learning_rate": 7.238064095688724e-08,
      "loss": 1.452,
      "step": 10912
    },
    {
      "epoch": 1.9658635442467913,
      "grad_norm": 1.486197590827942,
      "learning_rate": 7.162092594427128e-08,
      "loss": 1.3393,
      "step": 10913
    },
    {
      "epoch": 1.9660436838549877,
      "grad_norm": 1.571014165878296,
      "learning_rate": 7.086521616190279e-08,
      "loss": 1.5181,
      "step": 10914
    },
    {
      "epoch": 1.966223823463184,
      "grad_norm": 1.4618765115737915,
      "learning_rate": 7.011351167041103e-08,
      "loss": 1.4172,
      "step": 10915
    },
    {
      "epoch": 1.9664039630713803,
      "grad_norm": 1.2717589139938354,
      "learning_rate": 6.936581253008667e-08,
      "loss": 1.3009,
      "step": 10916
    },
    {
      "epoch": 1.9665841026795765,
      "grad_norm": 1.3492579460144043,
      "learning_rate": 6.862211880092061e-08,
      "loss": 1.2506,
      "step": 10917
    },
    {
      "epoch": 1.966764242287773,
      "grad_norm": 1.5550481081008911,
      "learning_rate": 6.788243054257071e-08,
      "loss": 1.5462,
      "step": 10918
    },
    {
      "epoch": 1.9669443818959693,
      "grad_norm": 1.3130022287368774,
      "learning_rate": 6.71467478143617e-08,
      "loss": 1.0724,
      "step": 10919
    },
    {
      "epoch": 1.9671245215041657,
      "grad_norm": 1.4852937459945679,
      "learning_rate": 6.64150706753297e-08,
      "loss": 1.5518,
      "step": 10920
    },
    {
      "epoch": 1.9673046611123621,
      "grad_norm": 1.4843921661376953,
      "learning_rate": 6.568739918415001e-08,
      "loss": 1.3119,
      "step": 10921
    },
    {
      "epoch": 1.9674848007205585,
      "grad_norm": 1.416107177734375,
      "learning_rate": 6.49637333992148e-08,
      "loss": 1.4588,
      "step": 10922
    },
    {
      "epoch": 1.967664940328755,
      "grad_norm": 1.555823802947998,
      "learning_rate": 6.424407337856097e-08,
      "loss": 1.6191,
      "step": 10923
    },
    {
      "epoch": 1.9678450799369511,
      "grad_norm": 1.5154531002044678,
      "learning_rate": 6.352841917992014e-08,
      "loss": 1.619,
      "step": 10924
    },
    {
      "epoch": 1.9680252195451475,
      "grad_norm": 1.5035817623138428,
      "learning_rate": 6.281677086071303e-08,
      "loss": 1.4539,
      "step": 10925
    },
    {
      "epoch": 1.9682053591533437,
      "grad_norm": 1.5245845317840576,
      "learning_rate": 6.210912847802175e-08,
      "loss": 1.3894,
      "step": 10926
    },
    {
      "epoch": 1.9683854987615401,
      "grad_norm": 1.4736533164978027,
      "learning_rate": 6.140549208860646e-08,
      "loss": 1.4583,
      "step": 10927
    },
    {
      "epoch": 1.9685656383697365,
      "grad_norm": 1.3531192541122437,
      "learning_rate": 6.070586174892201e-08,
      "loss": 1.2688,
      "step": 10928
    },
    {
      "epoch": 1.968745777977933,
      "grad_norm": 1.503147840499878,
      "learning_rate": 6.001023751509016e-08,
      "loss": 1.2753,
      "step": 10929
    },
    {
      "epoch": 1.9689259175861293,
      "grad_norm": 1.4450464248657227,
      "learning_rate": 5.9318619442905175e-08,
      "loss": 1.2475,
      "step": 10930
    },
    {
      "epoch": 1.9691060571943257,
      "grad_norm": 1.4868841171264648,
      "learning_rate": 5.8631007587861554e-08,
      "loss": 1.4115,
      "step": 10931
    },
    {
      "epoch": 1.969286196802522,
      "grad_norm": 1.5354737043380737,
      "learning_rate": 5.794740200511517e-08,
      "loss": 1.6124,
      "step": 10932
    },
    {
      "epoch": 1.9694663364107183,
      "grad_norm": 1.5318485498428345,
      "learning_rate": 5.7267802749494394e-08,
      "loss": 1.4552,
      "step": 10933
    },
    {
      "epoch": 1.9696464760189145,
      "grad_norm": 1.6651866436004639,
      "learning_rate": 5.6592209875533374e-08,
      "loss": 1.3626,
      "step": 10934
    },
    {
      "epoch": 1.969826615627111,
      "grad_norm": 1.391055703163147,
      "learning_rate": 5.5920623437411e-08,
      "loss": 1.2536,
      "step": 10935
    },
    {
      "epoch": 1.9700067552353073,
      "grad_norm": 1.3962472677230835,
      "learning_rate": 5.525304348901195e-08,
      "loss": 1.2328,
      "step": 10936
    },
    {
      "epoch": 1.9701868948435037,
      "grad_norm": 1.5940313339233398,
      "learning_rate": 5.458947008388782e-08,
      "loss": 1.4838,
      "step": 10937
    },
    {
      "epoch": 1.9703670344517001,
      "grad_norm": 1.5195205211639404,
      "learning_rate": 5.392990327526826e-08,
      "loss": 1.4395,
      "step": 10938
    },
    {
      "epoch": 1.9705471740598965,
      "grad_norm": 1.5199916362762451,
      "learning_rate": 5.327434311606094e-08,
      "loss": 1.3739,
      "step": 10939
    },
    {
      "epoch": 1.9707273136680927,
      "grad_norm": 1.5955090522766113,
      "learning_rate": 5.2622789658862695e-08,
      "loss": 1.5875,
      "step": 10940
    },
    {
      "epoch": 1.9709074532762891,
      "grad_norm": 1.6120820045471191,
      "learning_rate": 5.19752429559317e-08,
      "loss": 1.3275,
      "step": 10941
    },
    {
      "epoch": 1.9710875928844853,
      "grad_norm": 1.4519851207733154,
      "learning_rate": 5.133170305922086e-08,
      "loss": 1.352,
      "step": 10942
    },
    {
      "epoch": 1.9712677324926817,
      "grad_norm": 1.3888740539550781,
      "learning_rate": 5.069217002034998e-08,
      "loss": 1.1401,
      "step": 10943
    },
    {
      "epoch": 1.9714478721008781,
      "grad_norm": 1.6639899015426636,
      "learning_rate": 5.005664389062248e-08,
      "loss": 1.399,
      "step": 10944
    },
    {
      "epoch": 1.9716280117090745,
      "grad_norm": 1.5957750082015991,
      "learning_rate": 4.942512472102534e-08,
      "loss": 1.6113,
      "step": 10945
    },
    {
      "epoch": 1.971808151317271,
      "grad_norm": 1.5716404914855957,
      "learning_rate": 4.8797612562212494e-08,
      "loss": 1.3791,
      "step": 10946
    },
    {
      "epoch": 1.9719882909254673,
      "grad_norm": 1.6756418943405151,
      "learning_rate": 4.8174107464527e-08,
      "loss": 1.4122,
      "step": 10947
    },
    {
      "epoch": 1.9721684305336638,
      "grad_norm": 1.5390774011611938,
      "learning_rate": 4.755460947797885e-08,
      "loss": 1.255,
      "step": 10948
    },
    {
      "epoch": 1.97234857014186,
      "grad_norm": 1.6439656019210815,
      "learning_rate": 4.693911865227274e-08,
      "loss": 1.3975,
      "step": 10949
    },
    {
      "epoch": 1.9725287097500563,
      "grad_norm": 1.3778516054153442,
      "learning_rate": 4.6327635036780284e-08,
      "loss": 1.0747,
      "step": 10950
    },
    {
      "epoch": 1.9727088493582525,
      "grad_norm": 1.6551183462142944,
      "learning_rate": 4.572015868055113e-08,
      "loss": 1.1972,
      "step": 10951
    },
    {
      "epoch": 1.972888988966449,
      "grad_norm": 1.451918125152588,
      "learning_rate": 4.511668963231852e-08,
      "loss": 1.7403,
      "step": 10952
    },
    {
      "epoch": 1.9730691285746453,
      "grad_norm": 1.4560006856918335,
      "learning_rate": 4.451722794049373e-08,
      "loss": 2.1494,
      "step": 10953
    },
    {
      "epoch": 1.9732492681828417,
      "grad_norm": 1.3630493879318237,
      "learning_rate": 4.3921773653160524e-08,
      "loss": 1.7216,
      "step": 10954
    },
    {
      "epoch": 1.9734294077910381,
      "grad_norm": 1.3320837020874023,
      "learning_rate": 4.3330326818097345e-08,
      "loss": 1.5358,
      "step": 10955
    },
    {
      "epoch": 1.9736095473992346,
      "grad_norm": 1.4620627164840698,
      "learning_rate": 4.274288748273847e-08,
      "loss": 1.7386,
      "step": 10956
    },
    {
      "epoch": 1.9737896870074307,
      "grad_norm": 1.380967378616333,
      "learning_rate": 4.215945569421287e-08,
      "loss": 1.4616,
      "step": 10957
    },
    {
      "epoch": 1.9739698266156271,
      "grad_norm": 1.469442367553711,
      "learning_rate": 4.158003149932199e-08,
      "loss": 1.419,
      "step": 10958
    },
    {
      "epoch": 1.9741499662238233,
      "grad_norm": 1.909218668937683,
      "learning_rate": 4.100461494454533e-08,
      "loss": 1.896,
      "step": 10959
    },
    {
      "epoch": 1.9743301058320197,
      "grad_norm": 1.758531093597412,
      "learning_rate": 4.043320607604595e-08,
      "loss": 1.667,
      "step": 10960
    },
    {
      "epoch": 1.9745102454402161,
      "grad_norm": 2.0069632530212402,
      "learning_rate": 3.986580493965941e-08,
      "loss": 1.969,
      "step": 10961
    },
    {
      "epoch": 1.9746903850484125,
      "grad_norm": 1.6590137481689453,
      "learning_rate": 3.930241158090486e-08,
      "loss": 1.7043,
      "step": 10962
    },
    {
      "epoch": 1.974870524656609,
      "grad_norm": 1.4012058973312378,
      "learning_rate": 3.874302604497393e-08,
      "loss": 1.3647,
      "step": 10963
    },
    {
      "epoch": 1.9750506642648054,
      "grad_norm": 1.4960945844650269,
      "learning_rate": 3.818764837674182e-08,
      "loss": 1.6016,
      "step": 10964
    },
    {
      "epoch": 1.9752308038730015,
      "grad_norm": 1.4604511260986328,
      "learning_rate": 3.76362786207618e-08,
      "loss": 1.5025,
      "step": 10965
    },
    {
      "epoch": 1.975410943481198,
      "grad_norm": 1.5214426517486572,
      "learning_rate": 3.708891682126514e-08,
      "loss": 1.4757,
      "step": 10966
    },
    {
      "epoch": 1.9755910830893941,
      "grad_norm": 1.4050499200820923,
      "learning_rate": 3.6545563022155614e-08,
      "loss": 1.3559,
      "step": 10967
    },
    {
      "epoch": 1.9757712226975905,
      "grad_norm": 1.327239751815796,
      "learning_rate": 3.600621726703168e-08,
      "loss": 1.4349,
      "step": 10968
    },
    {
      "epoch": 1.975951362305787,
      "grad_norm": 1.399746060371399,
      "learning_rate": 3.5470879599147635e-08,
      "loss": 1.2301,
      "step": 10969
    },
    {
      "epoch": 1.9761315019139833,
      "grad_norm": 1.3997142314910889,
      "learning_rate": 3.493955006145244e-08,
      "loss": 1.3044,
      "step": 10970
    },
    {
      "epoch": 1.9763116415221798,
      "grad_norm": 1.5404049158096313,
      "learning_rate": 3.4412228696567573e-08,
      "loss": 1.5043,
      "step": 10971
    },
    {
      "epoch": 1.9764917811303762,
      "grad_norm": 1.6258344650268555,
      "learning_rate": 3.3888915546803624e-08,
      "loss": 1.6682,
      "step": 10972
    },
    {
      "epoch": 1.9766719207385723,
      "grad_norm": 1.5348584651947021,
      "learning_rate": 3.3369610654127025e-08,
      "loss": 1.4256,
      "step": 10973
    },
    {
      "epoch": 1.9768520603467687,
      "grad_norm": 1.3409672975540161,
      "learning_rate": 3.285431406020445e-08,
      "loss": 1.4082,
      "step": 10974
    },
    {
      "epoch": 1.9770321999549652,
      "grad_norm": 1.2668845653533936,
      "learning_rate": 3.23430258063695e-08,
      "loss": 1.2347,
      "step": 10975
    },
    {
      "epoch": 1.9772123395631613,
      "grad_norm": 1.4846363067626953,
      "learning_rate": 3.183574593363936e-08,
      "loss": 1.3976,
      "step": 10976
    },
    {
      "epoch": 1.9773924791713577,
      "grad_norm": 1.5372200012207031,
      "learning_rate": 3.1332474482709265e-08,
      "loss": 1.4191,
      "step": 10977
    },
    {
      "epoch": 1.9775726187795541,
      "grad_norm": 1.5466917753219604,
      "learning_rate": 3.083321149394691e-08,
      "loss": 1.3708,
      "step": 10978
    },
    {
      "epoch": 1.9777527583877506,
      "grad_norm": 1.4514127969741821,
      "learning_rate": 3.0337957007403604e-08,
      "loss": 1.3012,
      "step": 10979
    },
    {
      "epoch": 1.977932897995947,
      "grad_norm": 1.435525894165039,
      "learning_rate": 2.984671106281422e-08,
      "loss": 1.3044,
      "step": 10980
    },
    {
      "epoch": 1.9781130376041434,
      "grad_norm": 1.5968031883239746,
      "learning_rate": 2.9359473699580585e-08,
      "loss": 1.5017,
      "step": 10981
    },
    {
      "epoch": 1.9782931772123395,
      "grad_norm": 1.5171699523925781,
      "learning_rate": 2.887624495678809e-08,
      "loss": 1.469,
      "step": 10982
    },
    {
      "epoch": 1.978473316820536,
      "grad_norm": 1.4652512073516846,
      "learning_rate": 2.8397024873205723e-08,
      "loss": 1.3148,
      "step": 10983
    },
    {
      "epoch": 1.9786534564287321,
      "grad_norm": 1.6118950843811035,
      "learning_rate": 2.792181348726941e-08,
      "loss": 1.3349,
      "step": 10984
    },
    {
      "epoch": 1.9788335960369285,
      "grad_norm": 1.5601909160614014,
      "learning_rate": 2.7450610837109758e-08,
      "loss": 1.4196,
      "step": 10985
    },
    {
      "epoch": 1.979013735645125,
      "grad_norm": 1.4142730236053467,
      "learning_rate": 2.698341696051876e-08,
      "loss": 1.249,
      "step": 10986
    },
    {
      "epoch": 1.9791938752533214,
      "grad_norm": 1.5420595407485962,
      "learning_rate": 2.6520231894977543e-08,
      "loss": 1.3778,
      "step": 10987
    },
    {
      "epoch": 1.9793740148615178,
      "grad_norm": 1.5133724212646484,
      "learning_rate": 2.6061055677639724e-08,
      "loss": 1.4749,
      "step": 10988
    },
    {
      "epoch": 1.9795541544697142,
      "grad_norm": 1.6216070652008057,
      "learning_rate": 2.560588834534805e-08,
      "loss": 1.482,
      "step": 10989
    },
    {
      "epoch": 1.9797342940779104,
      "grad_norm": 1.5707279443740845,
      "learning_rate": 2.5154729934606657e-08,
      "loss": 1.3134,
      "step": 10990
    },
    {
      "epoch": 1.9799144336861068,
      "grad_norm": 1.6374320983886719,
      "learning_rate": 2.4707580481608816e-08,
      "loss": 1.406,
      "step": 10991
    },
    {
      "epoch": 1.980094573294303,
      "grad_norm": 1.7445958852767944,
      "learning_rate": 2.426444002223138e-08,
      "loss": 1.5866,
      "step": 10992
    },
    {
      "epoch": 1.9802747129024993,
      "grad_norm": 1.592361569404602,
      "learning_rate": 2.3825308592018146e-08,
      "loss": 1.4782,
      "step": 10993
    },
    {
      "epoch": 1.9804548525106958,
      "grad_norm": 1.3805537223815918,
      "learning_rate": 2.3390186226190936e-08,
      "loss": 1.1608,
      "step": 10994
    },
    {
      "epoch": 1.9806349921188922,
      "grad_norm": 1.4970101118087769,
      "learning_rate": 2.295907295966626e-08,
      "loss": 1.2956,
      "step": 10995
    },
    {
      "epoch": 1.9808151317270886,
      "grad_norm": 1.5703026056289673,
      "learning_rate": 2.253196882701647e-08,
      "loss": 1.453,
      "step": 10996
    },
    {
      "epoch": 1.980995271335285,
      "grad_norm": 1.7089860439300537,
      "learning_rate": 2.2108873862514144e-08,
      "loss": 1.6542,
      "step": 10997
    },
    {
      "epoch": 1.9811754109434812,
      "grad_norm": 1.4807592630386353,
      "learning_rate": 2.1689788100093254e-08,
      "loss": 1.3048,
      "step": 10998
    },
    {
      "epoch": 1.9813555505516776,
      "grad_norm": 1.4494138956069946,
      "learning_rate": 2.1274711573371353e-08,
      "loss": 1.1937,
      "step": 10999
    },
    {
      "epoch": 1.9815356901598737,
      "grad_norm": 1.5827856063842773,
      "learning_rate": 2.0863644315649577e-08,
      "loss": 1.3806,
      "step": 11000
    },
    {
      "epoch": 1.9817158297680701,
      "grad_norm": 1.4371733665466309,
      "learning_rate": 2.0456586359901553e-08,
      "loss": 1.2542,
      "step": 11001
    },
    {
      "epoch": 1.9818959693762666,
      "grad_norm": 1.4399656057357788,
      "learning_rate": 2.005353773878449e-08,
      "loss": 1.9351,
      "step": 11002
    },
    {
      "epoch": 1.982076108984463,
      "grad_norm": 1.3993446826934814,
      "learning_rate": 1.9654498484628083e-08,
      "loss": 2.0298,
      "step": 11003
    },
    {
      "epoch": 1.9822562485926594,
      "grad_norm": 1.2894805669784546,
      "learning_rate": 1.9259468629440057e-08,
      "loss": 1.8392,
      "step": 11004
    },
    {
      "epoch": 1.9824363882008558,
      "grad_norm": 1.357730507850647,
      "learning_rate": 1.886844820491729e-08,
      "loss": 1.6466,
      "step": 11005
    },
    {
      "epoch": 1.9826165278090522,
      "grad_norm": 1.3873244524002075,
      "learning_rate": 1.848143724241802e-08,
      "loss": 1.6024,
      "step": 11006
    },
    {
      "epoch": 1.9827966674172484,
      "grad_norm": 1.4892712831497192,
      "learning_rate": 1.809843577299519e-08,
      "loss": 1.8916,
      "step": 11007
    },
    {
      "epoch": 1.9829768070254448,
      "grad_norm": 1.4920061826705933,
      "learning_rate": 1.7719443827368677e-08,
      "loss": 1.6953,
      "step": 11008
    },
    {
      "epoch": 1.983156946633641,
      "grad_norm": 1.6392779350280762,
      "learning_rate": 1.7344461435947477e-08,
      "loss": 1.9488,
      "step": 11009
    },
    {
      "epoch": 1.9833370862418374,
      "grad_norm": 1.813663125038147,
      "learning_rate": 1.697348862880199e-08,
      "loss": 2.1107,
      "step": 11010
    },
    {
      "epoch": 1.9835172258500338,
      "grad_norm": 1.9045401811599731,
      "learning_rate": 1.6606525435702848e-08,
      "loss": 1.8947,
      "step": 11011
    },
    {
      "epoch": 1.9836973654582302,
      "grad_norm": 1.6111780405044556,
      "learning_rate": 1.6243571886082055e-08,
      "loss": 1.5044,
      "step": 11012
    },
    {
      "epoch": 1.9838775050664266,
      "grad_norm": 1.3917593955993652,
      "learning_rate": 1.5884628009049664e-08,
      "loss": 1.4241,
      "step": 11013
    },
    {
      "epoch": 1.984057644674623,
      "grad_norm": 1.5415117740631104,
      "learning_rate": 1.552969383341041e-08,
      "loss": 1.5877,
      "step": 11014
    },
    {
      "epoch": 1.9842377842828192,
      "grad_norm": 1.4373112916946411,
      "learning_rate": 1.517876938763041e-08,
      "loss": 1.4897,
      "step": 11015
    },
    {
      "epoch": 1.9844179238910156,
      "grad_norm": 1.382867693901062,
      "learning_rate": 1.4831854699864921e-08,
      "loss": 1.3131,
      "step": 11016
    },
    {
      "epoch": 1.9845980634992118,
      "grad_norm": 1.376491904258728,
      "learning_rate": 1.4488949797941687e-08,
      "loss": 1.2922,
      "step": 11017
    },
    {
      "epoch": 1.9847782031074082,
      "grad_norm": 1.4327082633972168,
      "learning_rate": 1.4150054709366479e-08,
      "loss": 1.3557,
      "step": 11018
    },
    {
      "epoch": 1.9849583427156046,
      "grad_norm": 1.3487615585327148,
      "learning_rate": 1.3815169461323107e-08,
      "loss": 1.2362,
      "step": 11019
    },
    {
      "epoch": 1.985138482323801,
      "grad_norm": 1.5588520765304565,
      "learning_rate": 1.3484294080684523e-08,
      "loss": 1.5735,
      "step": 11020
    },
    {
      "epoch": 1.9853186219319974,
      "grad_norm": 1.3944504261016846,
      "learning_rate": 1.3157428593990607e-08,
      "loss": 1.3834,
      "step": 11021
    },
    {
      "epoch": 1.9854987615401938,
      "grad_norm": 1.5337153673171997,
      "learning_rate": 1.2834573027453723e-08,
      "loss": 1.2569,
      "step": 11022
    },
    {
      "epoch": 1.98567890114839,
      "grad_norm": 1.44857919216156,
      "learning_rate": 1.2515727406980926e-08,
      "loss": 1.3367,
      "step": 11023
    },
    {
      "epoch": 1.9858590407565864,
      "grad_norm": 1.6034716367721558,
      "learning_rate": 1.2200891758151756e-08,
      "loss": 1.6182,
      "step": 11024
    },
    {
      "epoch": 1.9860391803647826,
      "grad_norm": 1.4651128053665161,
      "learning_rate": 1.1890066106212683e-08,
      "loss": 1.3752,
      "step": 11025
    },
    {
      "epoch": 1.986219319972979,
      "grad_norm": 1.527335286140442,
      "learning_rate": 1.1583250476110419e-08,
      "loss": 1.4843,
      "step": 11026
    },
    {
      "epoch": 1.9863994595811754,
      "grad_norm": 1.7838940620422363,
      "learning_rate": 1.1280444892447505e-08,
      "loss": 1.4359,
      "step": 11027
    },
    {
      "epoch": 1.9865795991893718,
      "grad_norm": 1.3547061681747437,
      "learning_rate": 1.0981649379521175e-08,
      "loss": 1.1103,
      "step": 11028
    },
    {
      "epoch": 1.9867597387975682,
      "grad_norm": 1.4999009370803833,
      "learning_rate": 1.068686396129559e-08,
      "loss": 1.4789,
      "step": 11029
    },
    {
      "epoch": 1.9869398784057646,
      "grad_norm": 1.4078631401062012,
      "learning_rate": 1.0396088661424053e-08,
      "loss": 1.2074,
      "step": 11030
    },
    {
      "epoch": 1.9871200180139608,
      "grad_norm": 1.4278786182403564,
      "learning_rate": 1.0109323503226797e-08,
      "loss": 1.2206,
      "step": 11031
    },
    {
      "epoch": 1.9873001576221572,
      "grad_norm": 1.4073715209960938,
      "learning_rate": 9.826568509713197e-09,
      "loss": 1.2757,
      "step": 11032
    },
    {
      "epoch": 1.9874802972303536,
      "grad_norm": 1.5547624826431274,
      "learning_rate": 9.547823703559556e-09,
      "loss": 1.3254,
      "step": 11033
    },
    {
      "epoch": 1.9876604368385498,
      "grad_norm": 1.6270281076431274,
      "learning_rate": 9.273089107131316e-09,
      "loss": 1.6017,
      "step": 11034
    },
    {
      "epoch": 1.9878405764467462,
      "grad_norm": 1.5338398218154907,
      "learning_rate": 9.002364742466408e-09,
      "loss": 1.4097,
      "step": 11035
    },
    {
      "epoch": 1.9880207160549426,
      "grad_norm": 1.7316972017288208,
      "learning_rate": 8.73565063128079e-09,
      "loss": 1.5932,
      "step": 11036
    },
    {
      "epoch": 1.988200855663139,
      "grad_norm": 1.4048669338226318,
      "learning_rate": 8.472946794974013e-09,
      "loss": 1.1715,
      "step": 11037
    },
    {
      "epoch": 1.9883809952713354,
      "grad_norm": 1.526852011680603,
      "learning_rate": 8.21425325461811e-09,
      "loss": 1.2814,
      "step": 11038
    },
    {
      "epoch": 1.9885611348795318,
      "grad_norm": 1.5486758947372437,
      "learning_rate": 7.959570030963149e-09,
      "loss": 1.3398,
      "step": 11039
    },
    {
      "epoch": 1.988741274487728,
      "grad_norm": 1.6917542219161987,
      "learning_rate": 7.708897144437233e-09,
      "loss": 1.7221,
      "step": 11040
    },
    {
      "epoch": 1.9889214140959244,
      "grad_norm": 1.8022210597991943,
      "learning_rate": 7.462234615157605e-09,
      "loss": 1.6295,
      "step": 11041
    },
    {
      "epoch": 1.9891015537041206,
      "grad_norm": 1.5328019857406616,
      "learning_rate": 7.219582462908437e-09,
      "loss": 1.3111,
      "step": 11042
    },
    {
      "epoch": 1.989281693312317,
      "grad_norm": 1.5486127138137817,
      "learning_rate": 6.980940707146389e-09,
      "loss": 1.4873,
      "step": 11043
    },
    {
      "epoch": 1.9894618329205134,
      "grad_norm": 1.4653786420822144,
      "learning_rate": 6.746309367028358e-09,
      "loss": 1.1981,
      "step": 11044
    },
    {
      "epoch": 1.9896419725287098,
      "grad_norm": 1.4437733888626099,
      "learning_rate": 6.515688461372627e-09,
      "loss": 1.2348,
      "step": 11045
    },
    {
      "epoch": 1.9898221121369062,
      "grad_norm": 1.507846474647522,
      "learning_rate": 6.2890780086755085e-09,
      "loss": 1.2631,
      "step": 11046
    },
    {
      "epoch": 1.9900022517451026,
      "grad_norm": 1.671010971069336,
      "learning_rate": 6.066478027116907e-09,
      "loss": 1.5826,
      "step": 11047
    },
    {
      "epoch": 1.9901823913532988,
      "grad_norm": 1.4583747386932373,
      "learning_rate": 5.847888534549206e-09,
      "loss": 1.0864,
      "step": 11048
    },
    {
      "epoch": 1.9903625309614952,
      "grad_norm": 1.508095145225525,
      "learning_rate": 5.633309548519483e-09,
      "loss": 1.2126,
      "step": 11049
    },
    {
      "epoch": 1.9905426705696914,
      "grad_norm": 1.6497952938079834,
      "learning_rate": 5.4227410862306425e-09,
      "loss": 1.2905,
      "step": 11050
    },
    {
      "epoch": 1.9907228101778878,
      "grad_norm": 1.4284565448760986,
      "learning_rate": 5.216183164580279e-09,
      "loss": 1.0046,
      "step": 11051
    },
    {
      "epoch": 1.9909029497860842,
      "grad_norm": 1.3137928247451782,
      "learning_rate": 5.013635800138472e-09,
      "loss": 1.8234,
      "step": 11052
    },
    {
      "epoch": 1.9910830893942806,
      "grad_norm": 1.2813129425048828,
      "learning_rate": 4.8150990091477835e-09,
      "loss": 1.7937,
      "step": 11053
    },
    {
      "epoch": 1.991263229002477,
      "grad_norm": 1.3043104410171509,
      "learning_rate": 4.620572807534362e-09,
      "loss": 1.6349,
      "step": 11054
    },
    {
      "epoch": 1.9914433686106734,
      "grad_norm": 1.4180114269256592,
      "learning_rate": 4.4300572109134966e-09,
      "loss": 1.9602,
      "step": 11055
    },
    {
      "epoch": 1.9916235082188696,
      "grad_norm": 1.4721637964248657,
      "learning_rate": 4.243552234556302e-09,
      "loss": 1.8934,
      "step": 11056
    },
    {
      "epoch": 1.991803647827066,
      "grad_norm": 1.333228349685669,
      "learning_rate": 4.061057893434139e-09,
      "loss": 1.5447,
      "step": 11057
    },
    {
      "epoch": 1.9919837874352622,
      "grad_norm": 1.3397862911224365,
      "learning_rate": 3.882574202174194e-09,
      "loss": 1.5206,
      "step": 11058
    },
    {
      "epoch": 1.9921639270434586,
      "grad_norm": 1.6484938859939575,
      "learning_rate": 3.7081011751094463e-09,
      "loss": 1.7874,
      "step": 11059
    },
    {
      "epoch": 1.992344066651655,
      "grad_norm": 1.6707037687301636,
      "learning_rate": 3.5376388262287063e-09,
      "loss": 1.5398,
      "step": 11060
    },
    {
      "epoch": 1.9925242062598514,
      "grad_norm": 1.6993014812469482,
      "learning_rate": 3.3711871691988196e-09,
      "loss": 1.7502,
      "step": 11061
    },
    {
      "epoch": 1.9927043458680478,
      "grad_norm": 1.7855199575424194,
      "learning_rate": 3.208746217386871e-09,
      "loss": 1.5786,
      "step": 11062
    },
    {
      "epoch": 1.9928844854762442,
      "grad_norm": 1.5815858840942383,
      "learning_rate": 3.050315983815777e-09,
      "loss": 1.5104,
      "step": 11063
    },
    {
      "epoch": 1.9930646250844406,
      "grad_norm": 1.4299613237380981,
      "learning_rate": 2.8958964811920396e-09,
      "loss": 1.3902,
      "step": 11064
    },
    {
      "epoch": 1.9932447646926368,
      "grad_norm": 1.2640552520751953,
      "learning_rate": 2.7454877219112996e-09,
      "loss": 1.1825,
      "step": 11065
    },
    {
      "epoch": 1.9934249043008332,
      "grad_norm": 1.384771704673767,
      "learning_rate": 2.5990897180305785e-09,
      "loss": 1.3792,
      "step": 11066
    },
    {
      "epoch": 1.9936050439090294,
      "grad_norm": 1.379815697669983,
      "learning_rate": 2.456702481301587e-09,
      "loss": 1.2922,
      "step": 11067
    },
    {
      "epoch": 1.9937851835172258,
      "grad_norm": 1.4666544198989868,
      "learning_rate": 2.318326023137418e-09,
      "loss": 1.535,
      "step": 11068
    },
    {
      "epoch": 1.9939653231254222,
      "grad_norm": 1.35880446434021,
      "learning_rate": 2.183960354651404e-09,
      "loss": 1.2615,
      "step": 11069
    },
    {
      "epoch": 1.9941454627336186,
      "grad_norm": 1.5199602842330933,
      "learning_rate": 2.0536054866071573e-09,
      "loss": 1.5071,
      "step": 11070
    },
    {
      "epoch": 1.994325602341815,
      "grad_norm": 1.4640576839447021,
      "learning_rate": 1.927261429474081e-09,
      "loss": 1.3601,
      "step": 11071
    },
    {
      "epoch": 1.9945057419500114,
      "grad_norm": 1.38802170753479,
      "learning_rate": 1.8049281933829597e-09,
      "loss": 1.415,
      "step": 11072
    },
    {
      "epoch": 1.9946858815582076,
      "grad_norm": 1.406002163887024,
      "learning_rate": 1.6866057881481656e-09,
      "loss": 1.2901,
      "step": 11073
    },
    {
      "epoch": 1.994866021166404,
      "grad_norm": 1.3924516439437866,
      "learning_rate": 1.5722942232621052e-09,
      "loss": 1.446,
      "step": 11074
    },
    {
      "epoch": 1.9950461607746002,
      "grad_norm": 1.330531120300293,
      "learning_rate": 1.4619935078896697e-09,
      "loss": 1.1766,
      "step": 11075
    },
    {
      "epoch": 1.9952263003827966,
      "grad_norm": 1.377027153968811,
      "learning_rate": 1.3557036508793363e-09,
      "loss": 1.1435,
      "step": 11076
    },
    {
      "epoch": 1.995406439990993,
      "grad_norm": 1.4500690698623657,
      "learning_rate": 1.2534246607687205e-09,
      "loss": 1.4477,
      "step": 11077
    },
    {
      "epoch": 1.9955865795991894,
      "grad_norm": 1.5224453210830688,
      "learning_rate": 1.1551565457512682e-09,
      "loss": 1.4601,
      "step": 11078
    },
    {
      "epoch": 1.9957667192073858,
      "grad_norm": 1.3258839845657349,
      "learning_rate": 1.0608993137151135e-09,
      "loss": 1.2186,
      "step": 11079
    },
    {
      "epoch": 1.9959468588155822,
      "grad_norm": 1.5319766998291016,
      "learning_rate": 9.706529722153247e-10,
      "loss": 1.6331,
      "step": 11080
    },
    {
      "epoch": 1.9961269984237784,
      "grad_norm": 1.6158783435821533,
      "learning_rate": 8.844175285016576e-10,
      "loss": 1.6333,
      "step": 11081
    },
    {
      "epoch": 1.9963071380319748,
      "grad_norm": 1.5090970993041992,
      "learning_rate": 8.021929894796998e-10,
      "loss": 1.2663,
      "step": 11082
    },
    {
      "epoch": 1.996487277640171,
      "grad_norm": 1.4330544471740723,
      "learning_rate": 7.239793617608293e-10,
      "loss": 1.5099,
      "step": 11083
    },
    {
      "epoch": 1.9966674172483674,
      "grad_norm": 1.5570321083068848,
      "learning_rate": 6.497766516067039e-10,
      "loss": 1.4506,
      "step": 11084
    },
    {
      "epoch": 1.9968475568565638,
      "grad_norm": 1.3722858428955078,
      "learning_rate": 5.79584864968119e-10,
      "loss": 1.21,
      "step": 11085
    },
    {
      "epoch": 1.9970276964647602,
      "grad_norm": 1.6175187826156616,
      "learning_rate": 5.134040074905588e-10,
      "loss": 1.551,
      "step": 11086
    },
    {
      "epoch": 1.9972078360729566,
      "grad_norm": 1.7320308685302734,
      "learning_rate": 4.512340844697871e-10,
      "loss": 1.41,
      "step": 11087
    },
    {
      "epoch": 1.997387975681153,
      "grad_norm": 1.6305747032165527,
      "learning_rate": 3.9307510089625634e-10,
      "loss": 1.4626,
      "step": 11088
    },
    {
      "epoch": 1.9975681152893494,
      "grad_norm": 1.5165183544158936,
      "learning_rate": 3.3892706143845434e-10,
      "loss": 1.3816,
      "step": 11089
    },
    {
      "epoch": 1.9977482548975456,
      "grad_norm": 1.546207308769226,
      "learning_rate": 2.88789970431802e-10,
      "loss": 1.4033,
      "step": 11090
    },
    {
      "epoch": 1.997928394505742,
      "grad_norm": 1.57394278049469,
      "learning_rate": 2.4266383191196007e-10,
      "loss": 1.1819,
      "step": 11091
    },
    {
      "epoch": 1.9981085341139382,
      "grad_norm": 1.4905561208724976,
      "learning_rate": 2.005486495704201e-10,
      "loss": 1.2591,
      "step": 11092
    },
    {
      "epoch": 1.9982886737221346,
      "grad_norm": 1.6398800611495972,
      "learning_rate": 1.6244442678226002e-10,
      "loss": 1.6037,
      "step": 11093
    },
    {
      "epoch": 1.998468813330331,
      "grad_norm": 1.5621079206466675,
      "learning_rate": 1.2835116661169545e-10,
      "loss": 1.2738,
      "step": 11094
    },
    {
      "epoch": 1.9986489529385274,
      "grad_norm": 1.5807408094406128,
      "learning_rate": 9.826887178987499e-11,
      "loss": 1.4298,
      "step": 11095
    },
    {
      "epoch": 1.9988290925467238,
      "grad_norm": 1.3767751455307007,
      "learning_rate": 7.219754473153373e-11,
      "loss": 1.1201,
      "step": 11096
    },
    {
      "epoch": 1.9990092321549202,
      "grad_norm": 1.3734205961227417,
      "learning_rate": 5.0137187529442075e-11,
      "loss": 1.1917,
      "step": 11097
    },
    {
      "epoch": 1.9991893717631164,
      "grad_norm": 1.6411293745040894,
      "learning_rate": 3.2087801948854635e-11,
      "loss": 1.4585,
      "step": 11098
    },
    {
      "epoch": 1.9993695113713128,
      "grad_norm": 1.5668938159942627,
      "learning_rate": 1.8049389438612453e-11,
      "loss": 1.3119,
      "step": 11099
    },
    {
      "epoch": 1.999549650979509,
      "grad_norm": 1.656908392906189,
      "learning_rate": 8.021951131143013e-12,
      "loss": 1.4597,
      "step": 11100
    },
    {
      "epoch": 1.9997297905877054,
      "grad_norm": 1.4120724201202393,
      "learning_rate": 2.0054878202557804e-12,
      "loss": 1.1435,
      "step": 11101
    },
    {
      "epoch": 1.9999099301959018,
      "grad_norm": 1.3890012502670288,
      "learning_rate": 0.0,
      "loss": 1.6596,
      "step": 11102
    },
    {
      "epoch": 1.9999099301959018,
      "eval_loss": 1.5088356733322144,
      "eval_runtime": 184.9011,
      "eval_samples_per_second": 50.568,
      "eval_steps_per_second": 12.645,
      "step": 11102
    }
  ],
  "logging_steps": 1,
  "max_steps": 11102,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.499053016215716e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}