{
  "best_metric": 0.1219615489244461,
  "best_model_checkpoint": "./results/checkpoint-8640",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 8640,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00011574074074074075,
      "grad_norm": 5.692312240600586,
      "learning_rate": 1.9997685185185186e-05,
      "loss": 1.9965,
      "step": 1
    },
    {
      "epoch": 0.0002314814814814815,
      "grad_norm": 4.84286642074585,
      "learning_rate": 1.9995370370370374e-05,
      "loss": 2.216,
      "step": 2
    },
    {
      "epoch": 0.00034722222222222224,
      "grad_norm": 5.341744422912598,
      "learning_rate": 1.9993055555555558e-05,
      "loss": 1.9798,
      "step": 3
    },
    {
      "epoch": 0.000462962962962963,
      "grad_norm": 4.7821736335754395,
      "learning_rate": 1.9990740740740742e-05,
      "loss": 2.0575,
      "step": 4
    },
    {
      "epoch": 0.0005787037037037037,
      "grad_norm": 5.099183559417725,
      "learning_rate": 1.9988425925925926e-05,
      "loss": 2.0466,
      "step": 5
    },
    {
      "epoch": 0.0006944444444444445,
      "grad_norm": 5.1574835777282715,
      "learning_rate": 1.9986111111111114e-05,
      "loss": 2.2455,
      "step": 6
    },
    {
      "epoch": 0.0008101851851851852,
      "grad_norm": 5.093099117279053,
      "learning_rate": 1.9983796296296298e-05,
      "loss": 2.203,
      "step": 7
    },
    {
      "epoch": 0.000925925925925926,
      "grad_norm": 5.105306625366211,
      "learning_rate": 1.9981481481481482e-05,
      "loss": 2.026,
      "step": 8
    },
    {
      "epoch": 0.0010416666666666667,
      "grad_norm": 5.474290370941162,
      "learning_rate": 1.9979166666666667e-05,
      "loss": 2.0294,
      "step": 9
    },
    {
      "epoch": 0.0011574074074074073,
      "grad_norm": 4.85551643371582,
      "learning_rate": 1.9976851851851854e-05,
      "loss": 1.9618,
      "step": 10
    },
    {
      "epoch": 0.0012731481481481483,
      "grad_norm": 4.711113452911377,
      "learning_rate": 1.997453703703704e-05,
      "loss": 2.0421,
      "step": 11
    },
    {
      "epoch": 0.001388888888888889,
      "grad_norm": 4.147070407867432,
      "learning_rate": 1.9972222222222223e-05,
      "loss": 1.958,
      "step": 12
    },
    {
      "epoch": 0.0015046296296296296,
      "grad_norm": 5.110289096832275,
      "learning_rate": 1.996990740740741e-05,
      "loss": 2.0677,
      "step": 13
    },
    {
      "epoch": 0.0016203703703703703,
      "grad_norm": 12.784015655517578,
      "learning_rate": 1.9967592592592595e-05,
      "loss": 2.0563,
      "step": 14
    },
    {
      "epoch": 0.001736111111111111,
      "grad_norm": 3.9455792903900146,
      "learning_rate": 1.996527777777778e-05,
      "loss": 2.187,
      "step": 15
    },
    {
      "epoch": 0.001851851851851852,
      "grad_norm": 4.494223594665527,
      "learning_rate": 1.9962962962962963e-05,
      "loss": 2.0662,
      "step": 16
    },
    {
      "epoch": 0.0019675925925925924,
      "grad_norm": 4.635458469390869,
      "learning_rate": 1.996064814814815e-05,
      "loss": 2.2081,
      "step": 17
    },
    {
      "epoch": 0.0020833333333333333,
      "grad_norm": 4.5755205154418945,
      "learning_rate": 1.9958333333333335e-05,
      "loss": 2.2455,
      "step": 18
    },
    {
      "epoch": 0.002199074074074074,
      "grad_norm": 6.070075988769531,
      "learning_rate": 1.995601851851852e-05,
      "loss": 2.3178,
      "step": 19
    },
    {
      "epoch": 0.0023148148148148147,
      "grad_norm": 5.368930339813232,
      "learning_rate": 1.9953703703703707e-05,
      "loss": 2.1156,
      "step": 20
    },
    {
      "epoch": 0.0024305555555555556,
      "grad_norm": 5.044558048248291,
      "learning_rate": 1.995138888888889e-05,
      "loss": 2.0151,
      "step": 21
    },
    {
      "epoch": 0.0025462962962962965,
      "grad_norm": 4.1191558837890625,
      "learning_rate": 1.9949074074074075e-05,
      "loss": 1.896,
      "step": 22
    },
    {
      "epoch": 0.002662037037037037,
      "grad_norm": 5.0372419357299805,
      "learning_rate": 1.994675925925926e-05,
      "loss": 1.9973,
      "step": 23
    },
    {
      "epoch": 0.002777777777777778,
      "grad_norm": 6.124544620513916,
      "learning_rate": 1.9944444444444447e-05,
      "loss": 2.257,
      "step": 24
    },
    {
      "epoch": 0.0028935185185185184,
      "grad_norm": 4.747463226318359,
      "learning_rate": 1.994212962962963e-05,
      "loss": 2.4314,
      "step": 25
    },
    {
      "epoch": 0.0030092592592592593,
      "grad_norm": 7.170324802398682,
      "learning_rate": 1.9939814814814816e-05,
      "loss": 2.2372,
      "step": 26
    },
    {
      "epoch": 0.003125,
      "grad_norm": 8.67553424835205,
      "learning_rate": 1.99375e-05,
      "loss": 2.2455,
      "step": 27
    },
    {
      "epoch": 0.0032407407407407406,
      "grad_norm": 5.201866149902344,
      "learning_rate": 1.9935185185185188e-05,
      "loss": 2.0796,
      "step": 28
    },
    {
      "epoch": 0.0033564814814814816,
      "grad_norm": 5.467615127563477,
      "learning_rate": 1.9932870370370372e-05,
      "loss": 2.2373,
      "step": 29
    },
    {
      "epoch": 0.003472222222222222,
      "grad_norm": 5.962344169616699,
      "learning_rate": 1.9930555555555556e-05,
      "loss": 2.0194,
      "step": 30
    },
    {
      "epoch": 0.003587962962962963,
      "grad_norm": 4.734037399291992,
      "learning_rate": 1.9928240740740744e-05,
      "loss": 2.1001,
      "step": 31
    },
    {
      "epoch": 0.003703703703703704,
      "grad_norm": 8.052763938903809,
      "learning_rate": 1.9925925925925928e-05,
      "loss": 2.2791,
      "step": 32
    },
    {
      "epoch": 0.0038194444444444443,
      "grad_norm": 5.333888053894043,
      "learning_rate": 1.9923611111111112e-05,
      "loss": 2.1982,
      "step": 33
    },
    {
      "epoch": 0.003935185185185185,
      "grad_norm": 7.304694175720215,
      "learning_rate": 1.9921296296296297e-05,
      "loss": 2.2927,
      "step": 34
    },
    {
      "epoch": 0.004050925925925926,
      "grad_norm": 5.542030334472656,
      "learning_rate": 1.9918981481481484e-05,
      "loss": 1.9012,
      "step": 35
    },
    {
      "epoch": 0.004166666666666667,
      "grad_norm": 6.566902160644531,
      "learning_rate": 1.991666666666667e-05,
      "loss": 2.1674,
      "step": 36
    },
    {
      "epoch": 0.0042824074074074075,
      "grad_norm": 4.752450942993164,
      "learning_rate": 1.9914351851851853e-05,
      "loss": 2.1524,
      "step": 37
    },
    {
      "epoch": 0.004398148148148148,
      "grad_norm": 5.43243932723999,
      "learning_rate": 1.991203703703704e-05,
      "loss": 1.9213,
      "step": 38
    },
    {
      "epoch": 0.0045138888888888885,
      "grad_norm": 5.803874969482422,
      "learning_rate": 1.9909722222222225e-05,
      "loss": 1.7322,
      "step": 39
    },
    {
      "epoch": 0.004629629629629629,
      "grad_norm": 6.247270107269287,
      "learning_rate": 1.990740740740741e-05,
      "loss": 1.9964,
      "step": 40
    },
    {
      "epoch": 0.00474537037037037,
      "grad_norm": 7.485787391662598,
      "learning_rate": 1.9905092592592593e-05,
      "loss": 2.1596,
      "step": 41
    },
    {
      "epoch": 0.004861111111111111,
      "grad_norm": 5.268268585205078,
      "learning_rate": 1.990277777777778e-05,
      "loss": 2.0961,
      "step": 42
    },
    {
      "epoch": 0.004976851851851852,
      "grad_norm": 5.570443153381348,
      "learning_rate": 1.9900462962962965e-05,
      "loss": 1.9833,
      "step": 43
    },
    {
      "epoch": 0.005092592592592593,
      "grad_norm": 7.018050670623779,
      "learning_rate": 1.989814814814815e-05,
      "loss": 2.0031,
      "step": 44
    },
    {
      "epoch": 0.005208333333333333,
      "grad_norm": 6.23184061050415,
      "learning_rate": 1.9895833333333334e-05,
      "loss": 2.1846,
      "step": 45
    },
    {
      "epoch": 0.005324074074074074,
      "grad_norm": 6.5781354904174805,
      "learning_rate": 1.989351851851852e-05,
      "loss": 2.0114,
      "step": 46
    },
    {
      "epoch": 0.005439814814814815,
      "grad_norm": 8.046587944030762,
      "learning_rate": 1.9891203703703705e-05,
      "loss": 1.9729,
      "step": 47
    },
    {
      "epoch": 0.005555555555555556,
      "grad_norm": 5.455897808074951,
      "learning_rate": 1.988888888888889e-05,
      "loss": 2.0953,
      "step": 48
    },
    {
      "epoch": 0.005671296296296297,
      "grad_norm": 7.413969993591309,
      "learning_rate": 1.9886574074074077e-05,
      "loss": 1.9404,
      "step": 49
    },
    {
      "epoch": 0.005787037037037037,
      "grad_norm": 6.355666160583496,
      "learning_rate": 1.9884259259259258e-05,
      "loss": 2.2081,
      "step": 50
    },
    {
      "epoch": 0.005902777777777778,
      "grad_norm": 5.951140880584717,
      "learning_rate": 1.9881944444444446e-05,
      "loss": 1.8962,
      "step": 51
    },
    {
      "epoch": 0.0060185185185185185,
      "grad_norm": 6.399886608123779,
      "learning_rate": 1.987962962962963e-05,
      "loss": 2.178,
      "step": 52
    },
    {
      "epoch": 0.0061342592592592594,
      "grad_norm": 6.410641670227051,
      "learning_rate": 1.9877314814814818e-05,
      "loss": 2.0573,
      "step": 53
    },
    {
      "epoch": 0.00625,
      "grad_norm": 4.3195295333862305,
      "learning_rate": 1.9875000000000002e-05,
      "loss": 1.6384,
      "step": 54
    },
    {
      "epoch": 0.00636574074074074,
      "grad_norm": 7.165890216827393,
      "learning_rate": 1.9872685185185186e-05,
      "loss": 2.0272,
      "step": 55
    },
    {
      "epoch": 0.006481481481481481,
      "grad_norm": 5.0497918128967285,
      "learning_rate": 1.987037037037037e-05,
      "loss": 2.374,
      "step": 56
    },
    {
      "epoch": 0.006597222222222222,
      "grad_norm": 5.465131759643555,
      "learning_rate": 1.9868055555555558e-05,
      "loss": 2.3687,
      "step": 57
    },
    {
      "epoch": 0.006712962962962963,
      "grad_norm": 6.657927989959717,
      "learning_rate": 1.9865740740740742e-05,
      "loss": 1.9926,
      "step": 58
    },
    {
      "epoch": 0.006828703703703704,
      "grad_norm": 6.017234802246094,
      "learning_rate": 1.9863425925925927e-05,
      "loss": 2.1733,
      "step": 59
    },
    {
      "epoch": 0.006944444444444444,
      "grad_norm": 4.933075904846191,
      "learning_rate": 1.9861111111111114e-05,
      "loss": 1.913,
      "step": 60
    },
    {
      "epoch": 0.007060185185185185,
      "grad_norm": 4.504550933837891,
      "learning_rate": 1.98587962962963e-05,
      "loss": 1.6798,
      "step": 61
    },
    {
      "epoch": 0.007175925925925926,
      "grad_norm": 4.959822177886963,
      "learning_rate": 1.9856481481481483e-05,
      "loss": 2.0832,
      "step": 62
    },
    {
      "epoch": 0.007291666666666667,
      "grad_norm": 5.493355751037598,
      "learning_rate": 1.9854166666666667e-05,
      "loss": 1.8379,
      "step": 63
    },
    {
      "epoch": 0.007407407407407408,
      "grad_norm": 5.347026348114014,
      "learning_rate": 1.9851851851851855e-05,
      "loss": 1.9513,
      "step": 64
    },
    {
      "epoch": 0.007523148148148148,
      "grad_norm": 5.869378089904785,
      "learning_rate": 1.984953703703704e-05,
      "loss": 2.2505,
      "step": 65
    },
    {
      "epoch": 0.007638888888888889,
      "grad_norm": 7.80045223236084,
      "learning_rate": 1.9847222222222223e-05,
      "loss": 2.2986,
      "step": 66
    },
    {
      "epoch": 0.0077546296296296295,
      "grad_norm": 5.268881320953369,
      "learning_rate": 1.984490740740741e-05,
      "loss": 1.5781,
      "step": 67
    },
    {
      "epoch": 0.00787037037037037,
      "grad_norm": 6.378146171569824,
      "learning_rate": 1.984259259259259e-05,
      "loss": 2.0254,
      "step": 68
    },
    {
      "epoch": 0.00798611111111111,
      "grad_norm": 5.270776271820068,
      "learning_rate": 1.984027777777778e-05,
      "loss": 1.9587,
      "step": 69
    },
    {
      "epoch": 0.008101851851851851,
      "grad_norm": 5.26032829284668,
      "learning_rate": 1.9837962962962964e-05,
      "loss": 1.7372,
      "step": 70
    },
    {
      "epoch": 0.008217592592592592,
      "grad_norm": 7.769745826721191,
      "learning_rate": 1.983564814814815e-05,
      "loss": 1.9517,
      "step": 71
    },
    {
      "epoch": 0.008333333333333333,
      "grad_norm": 5.212433338165283,
      "learning_rate": 1.9833333333333335e-05,
      "loss": 1.9914,
      "step": 72
    },
    {
      "epoch": 0.008449074074074074,
      "grad_norm": 5.531213760375977,
      "learning_rate": 1.983101851851852e-05,
      "loss": 1.8329,
      "step": 73
    },
    {
      "epoch": 0.008564814814814815,
      "grad_norm": 5.720541954040527,
      "learning_rate": 1.9828703703703704e-05,
      "loss": 2.053,
      "step": 74
    },
    {
      "epoch": 0.008680555555555556,
      "grad_norm": 9.290575981140137,
      "learning_rate": 1.982638888888889e-05,
      "loss": 2.057,
      "step": 75
    },
    {
      "epoch": 0.008796296296296297,
      "grad_norm": 9.961891174316406,
      "learning_rate": 1.9824074074074076e-05,
      "loss": 1.9562,
      "step": 76
    },
    {
      "epoch": 0.008912037037037038,
      "grad_norm": 6.261855125427246,
      "learning_rate": 1.982175925925926e-05,
      "loss": 1.6258,
      "step": 77
    },
    {
      "epoch": 0.009027777777777777,
      "grad_norm": 7.326930046081543,
      "learning_rate": 1.9819444444444448e-05,
      "loss": 2.0667,
      "step": 78
    },
    {
      "epoch": 0.009143518518518518,
      "grad_norm": 5.27416467666626,
      "learning_rate": 1.9817129629629632e-05,
      "loss": 1.677,
      "step": 79
    },
    {
      "epoch": 0.009259259259259259,
      "grad_norm": 13.526907920837402,
      "learning_rate": 1.9814814814814816e-05,
      "loss": 1.9541,
      "step": 80
    },
    {
      "epoch": 0.009375,
      "grad_norm": 5.831089973449707,
      "learning_rate": 1.98125e-05,
      "loss": 1.953,
      "step": 81
    },
    {
      "epoch": 0.00949074074074074,
      "grad_norm": 14.920907974243164,
      "learning_rate": 1.9810185185185188e-05,
      "loss": 1.935,
      "step": 82
    },
    {
      "epoch": 0.009606481481481481,
      "grad_norm": 5.370945930480957,
      "learning_rate": 1.9807870370370372e-05,
      "loss": 2.4397,
      "step": 83
    },
    {
      "epoch": 0.009722222222222222,
      "grad_norm": 5.265706539154053,
      "learning_rate": 1.9805555555555557e-05,
      "loss": 1.7775,
      "step": 84
    },
    {
      "epoch": 0.009837962962962963,
      "grad_norm": 6.790380001068115,
      "learning_rate": 1.9803240740740744e-05,
      "loss": 2.2405,
      "step": 85
    },
    {
      "epoch": 0.009953703703703704,
      "grad_norm": 6.221429347991943,
      "learning_rate": 1.9800925925925925e-05,
      "loss": 1.9849,
      "step": 86
    },
    {
      "epoch": 0.010069444444444445,
      "grad_norm": 8.723067283630371,
      "learning_rate": 1.9798611111111113e-05,
      "loss": 2.1435,
      "step": 87
    },
    {
      "epoch": 0.010185185185185186,
      "grad_norm": 10.16743278503418,
      "learning_rate": 1.9796296296296297e-05,
      "loss": 2.0327,
      "step": 88
    },
    {
      "epoch": 0.010300925925925925,
      "grad_norm": 7.385878086090088,
      "learning_rate": 1.9793981481481485e-05,
      "loss": 1.7866,
      "step": 89
    },
    {
      "epoch": 0.010416666666666666,
      "grad_norm": 8.92994213104248,
      "learning_rate": 1.979166666666667e-05,
      "loss": 1.4474,
      "step": 90
    },
    {
      "epoch": 0.010532407407407407,
      "grad_norm": 5.76594352722168,
      "learning_rate": 1.9789351851851853e-05,
      "loss": 1.8775,
      "step": 91
    },
    {
      "epoch": 0.010648148148148148,
      "grad_norm": 6.122307777404785,
      "learning_rate": 1.9787037037037037e-05,
      "loss": 1.5882,
      "step": 92
    },
    {
      "epoch": 0.010763888888888889,
      "grad_norm": 5.691127300262451,
      "learning_rate": 1.9784722222222225e-05,
      "loss": 2.0978,
      "step": 93
    },
    {
      "epoch": 0.01087962962962963,
      "grad_norm": 6.490463733673096,
      "learning_rate": 1.978240740740741e-05,
      "loss": 1.7224,
      "step": 94
    },
    {
      "epoch": 0.01099537037037037,
      "grad_norm": 5.39813756942749,
      "learning_rate": 1.9780092592592594e-05,
      "loss": 2.0104,
      "step": 95
    },
    {
      "epoch": 0.011111111111111112,
      "grad_norm": 12.42576789855957,
      "learning_rate": 1.977777777777778e-05,
      "loss": 1.9068,
      "step": 96
    },
    {
      "epoch": 0.011226851851851852,
      "grad_norm": 6.4002790451049805,
      "learning_rate": 1.9775462962962962e-05,
      "loss": 2.0448,
      "step": 97
    },
    {
      "epoch": 0.011342592592592593,
      "grad_norm": 4.979708671569824,
      "learning_rate": 1.977314814814815e-05,
      "loss": 1.8255,
      "step": 98
    },
    {
      "epoch": 0.011458333333333333,
      "grad_norm": 5.876491069793701,
      "learning_rate": 1.9770833333333334e-05,
      "loss": 1.9064,
      "step": 99
    },
    {
      "epoch": 0.011574074074074073,
      "grad_norm": 5.78334903717041,
      "learning_rate": 1.976851851851852e-05,
      "loss": 1.7293,
      "step": 100
    },
    {
      "epoch": 0.011689814814814814,
      "grad_norm": 4.841702461242676,
      "learning_rate": 1.9766203703703706e-05,
      "loss": 1.821,
      "step": 101
    },
    {
      "epoch": 0.011805555555555555,
      "grad_norm": 6.424641132354736,
      "learning_rate": 1.976388888888889e-05,
      "loss": 2.0174,
      "step": 102
    },
    {
      "epoch": 0.011921296296296296,
      "grad_norm": 5.661606311798096,
      "learning_rate": 1.9761574074074074e-05,
      "loss": 2.229,
      "step": 103
    },
    {
      "epoch": 0.012037037037037037,
      "grad_norm": 5.802317142486572,
      "learning_rate": 1.975925925925926e-05,
      "loss": 1.6961,
      "step": 104
    },
    {
      "epoch": 0.012152777777777778,
      "grad_norm": 10.739157676696777,
      "learning_rate": 1.9756944444444446e-05,
      "loss": 1.8989,
      "step": 105
    },
    {
      "epoch": 0.012268518518518519,
      "grad_norm": 6.59002685546875,
      "learning_rate": 1.975462962962963e-05,
      "loss": 2.1107,
      "step": 106
    },
    {
      "epoch": 0.01238425925925926,
      "grad_norm": 8.76626205444336,
      "learning_rate": 1.9752314814814818e-05,
      "loss": 1.9276,
      "step": 107
    },
    {
      "epoch": 0.0125,
      "grad_norm": 4.834606170654297,
      "learning_rate": 1.9750000000000002e-05,
      "loss": 1.6849,
      "step": 108
    },
    {
      "epoch": 0.012615740740740742,
      "grad_norm": 6.770847320556641,
      "learning_rate": 1.9747685185185187e-05,
      "loss": 1.6661,
      "step": 109
    },
    {
      "epoch": 0.01273148148148148,
      "grad_norm": 4.638800621032715,
      "learning_rate": 1.974537037037037e-05,
      "loss": 1.3393,
      "step": 110
    },
    {
      "epoch": 0.012847222222222222,
      "grad_norm": 6.085695743560791,
      "learning_rate": 1.974305555555556e-05,
      "loss": 1.907,
      "step": 111
    },
    {
      "epoch": 0.012962962962962963,
      "grad_norm": 5.236244201660156,
      "learning_rate": 1.9740740740740743e-05,
      "loss": 1.6012,
      "step": 112
    },
    {
      "epoch": 0.013078703703703703,
      "grad_norm": 5.095339298248291,
      "learning_rate": 1.9738425925925927e-05,
      "loss": 1.5545,
      "step": 113
    },
    {
      "epoch": 0.013194444444444444,
      "grad_norm": 4.769504547119141,
      "learning_rate": 1.9736111111111115e-05,
      "loss": 1.3849,
      "step": 114
    },
    {
      "epoch": 0.013310185185185185,
      "grad_norm": 9.105463981628418,
      "learning_rate": 1.9733796296296295e-05,
      "loss": 1.978,
      "step": 115
    },
    {
      "epoch": 0.013425925925925926,
      "grad_norm": 7.989753723144531,
      "learning_rate": 1.9731481481481483e-05,
      "loss": 1.7332,
      "step": 116
    },
    {
      "epoch": 0.013541666666666667,
      "grad_norm": 5.6984944343566895,
      "learning_rate": 1.9729166666666667e-05,
      "loss": 1.9904,
      "step": 117
    },
    {
      "epoch": 0.013657407407407408,
      "grad_norm": 4.830297470092773,
      "learning_rate": 1.9726851851851855e-05,
      "loss": 1.7789,
      "step": 118
    },
    {
      "epoch": 0.013773148148148149,
      "grad_norm": 6.466744899749756,
      "learning_rate": 1.972453703703704e-05,
      "loss": 1.5525,
      "step": 119
    },
    {
      "epoch": 0.013888888888888888,
      "grad_norm": 5.596836090087891,
      "learning_rate": 1.9722222222222224e-05,
      "loss": 1.5705,
      "step": 120
    },
    {
      "epoch": 0.014004629629629629,
      "grad_norm": 7.196076393127441,
      "learning_rate": 1.9719907407407408e-05,
      "loss": 1.9581,
      "step": 121
    },
    {
      "epoch": 0.01412037037037037,
      "grad_norm": 6.177059650421143,
      "learning_rate": 1.9717592592592595e-05,
      "loss": 1.8273,
      "step": 122
    },
    {
      "epoch": 0.01423611111111111,
      "grad_norm": 5.619070529937744,
      "learning_rate": 1.971527777777778e-05,
      "loss": 1.803,
      "step": 123
    },
    {
      "epoch": 0.014351851851851852,
      "grad_norm": 5.06971549987793,
      "learning_rate": 1.9712962962962964e-05,
      "loss": 1.8411,
      "step": 124
    },
    {
      "epoch": 0.014467592592592593,
      "grad_norm": 8.163780212402344,
      "learning_rate": 1.971064814814815e-05,
      "loss": 1.4137,
      "step": 125
    },
    {
      "epoch": 0.014583333333333334,
      "grad_norm": 9.463504791259766,
      "learning_rate": 1.9708333333333336e-05,
      "loss": 1.5289,
      "step": 126
    },
    {
      "epoch": 0.014699074074074074,
      "grad_norm": 8.32180404663086,
      "learning_rate": 1.970601851851852e-05,
      "loss": 1.6557,
      "step": 127
    },
    {
      "epoch": 0.014814814814814815,
      "grad_norm": 8.62234115600586,
      "learning_rate": 1.9703703703703704e-05,
      "loss": 1.891,
      "step": 128
    },
    {
      "epoch": 0.014930555555555556,
      "grad_norm": 4.89546537399292,
      "learning_rate": 1.9701388888888892e-05,
      "loss": 1.5483,
      "step": 129
    },
    {
      "epoch": 0.015046296296296295,
      "grad_norm": 7.004753112792969,
      "learning_rate": 1.9699074074074076e-05,
      "loss": 1.6136,
      "step": 130
    },
    {
      "epoch": 0.015162037037037036,
      "grad_norm": 4.789592742919922,
      "learning_rate": 1.969675925925926e-05,
      "loss": 1.2851,
      "step": 131
    },
    {
      "epoch": 0.015277777777777777,
      "grad_norm": 5.328765392303467,
      "learning_rate": 1.9694444444444448e-05,
      "loss": 1.7363,
      "step": 132
    },
    {
      "epoch": 0.015393518518518518,
      "grad_norm": 6.13937520980835,
      "learning_rate": 1.969212962962963e-05,
      "loss": 2.0478,
      "step": 133
    },
    {
      "epoch": 0.015509259259259259,
      "grad_norm": 9.894044876098633,
      "learning_rate": 1.9689814814814817e-05,
      "loss": 1.6396,
      "step": 134
    },
    {
      "epoch": 0.015625,
      "grad_norm": 5.817492485046387,
      "learning_rate": 1.96875e-05,
      "loss": 1.9542,
      "step": 135
    },
    {
      "epoch": 0.01574074074074074,
      "grad_norm": 6.484155178070068,
      "learning_rate": 1.968518518518519e-05,
      "loss": 1.4624,
      "step": 136
    },
    {
      "epoch": 0.015856481481481482,
      "grad_norm": 6.346861362457275,
      "learning_rate": 1.9682870370370373e-05,
      "loss": 1.7305,
      "step": 137
    },
    {
      "epoch": 0.01597222222222222,
      "grad_norm": 6.371883392333984,
      "learning_rate": 1.9680555555555557e-05,
      "loss": 1.843,
      "step": 138
    },
    {
      "epoch": 0.016087962962962964,
      "grad_norm": 6.38097620010376,
      "learning_rate": 1.967824074074074e-05,
      "loss": 1.6992,
      "step": 139
    },
    {
      "epoch": 0.016203703703703703,
      "grad_norm": 7.249087810516357,
      "learning_rate": 1.967592592592593e-05,
      "loss": 1.739,
      "step": 140
    },
    {
      "epoch": 0.016319444444444445,
      "grad_norm": 5.761762619018555,
      "learning_rate": 1.9673611111111113e-05,
      "loss": 1.2483,
      "step": 141
    },
    {
      "epoch": 0.016435185185185185,
      "grad_norm": 7.356204986572266,
      "learning_rate": 1.9671296296296297e-05,
      "loss": 2.135,
      "step": 142
    },
    {
      "epoch": 0.016550925925925927,
      "grad_norm": 6.367476463317871,
      "learning_rate": 1.9668981481481485e-05,
      "loss": 1.3108,
      "step": 143
    },
    {
      "epoch": 0.016666666666666666,
      "grad_norm": 7.598026752471924,
      "learning_rate": 1.9666666666666666e-05,
      "loss": 1.2954,
      "step": 144
    },
    {
      "epoch": 0.01678240740740741,
      "grad_norm": 7.305114269256592,
      "learning_rate": 1.9664351851851853e-05,
      "loss": 2.2732,
      "step": 145
    },
    {
      "epoch": 0.016898148148148148,
      "grad_norm": 7.358803749084473,
      "learning_rate": 1.9662037037037038e-05,
      "loss": 1.9002,
      "step": 146
    },
    {
      "epoch": 0.017013888888888887,
      "grad_norm": 4.93549919128418,
      "learning_rate": 1.9659722222222225e-05,
      "loss": 1.3717,
      "step": 147
    },
    {
      "epoch": 0.01712962962962963,
      "grad_norm": 8.777894020080566,
      "learning_rate": 1.965740740740741e-05,
      "loss": 2.0714,
      "step": 148
    },
    {
      "epoch": 0.01724537037037037,
      "grad_norm": 6.367814064025879,
      "learning_rate": 1.9655092592592594e-05,
      "loss": 1.7371,
      "step": 149
    },
    {
      "epoch": 0.017361111111111112,
      "grad_norm": 10.611685752868652,
      "learning_rate": 1.9652777777777778e-05,
      "loss": 1.4282,
      "step": 150
    },
    {
      "epoch": 0.01747685185185185,
      "grad_norm": 8.853285789489746,
      "learning_rate": 1.9650462962962962e-05,
      "loss": 1.6507,
      "step": 151
    },
    {
      "epoch": 0.017592592592592594,
      "grad_norm": 5.930267333984375,
      "learning_rate": 1.964814814814815e-05,
      "loss": 1.904,
      "step": 152
    },
    {
      "epoch": 0.017708333333333333,
      "grad_norm": 9.275053977966309,
      "learning_rate": 1.9645833333333334e-05,
      "loss": 2.0581,
      "step": 153
    },
    {
      "epoch": 0.017824074074074076,
      "grad_norm": 4.459653377532959,
      "learning_rate": 1.9643518518518522e-05,
      "loss": 1.4549,
      "step": 154
    },
    {
      "epoch": 0.017939814814814815,
      "grad_norm": 8.097359657287598,
      "learning_rate": 1.9641203703703706e-05,
      "loss": 1.81,
      "step": 155
    },
    {
      "epoch": 0.018055555555555554,
      "grad_norm": 4.606447219848633,
      "learning_rate": 1.963888888888889e-05,
      "loss": 1.2985,
      "step": 156
    },
    {
      "epoch": 0.018171296296296297,
      "grad_norm": 7.130870819091797,
      "learning_rate": 1.9636574074074075e-05,
      "loss": 1.8141,
      "step": 157
    },
    {
      "epoch": 0.018287037037037036,
      "grad_norm": 10.379081726074219,
      "learning_rate": 1.9634259259259262e-05,
      "loss": 1.8178,
      "step": 158
    },
    {
      "epoch": 0.01840277777777778,
      "grad_norm": 9.887731552124023,
      "learning_rate": 1.9631944444444447e-05,
      "loss": 1.6474,
      "step": 159
    },
    {
      "epoch": 0.018518518518518517,
      "grad_norm": 7.828157424926758,
      "learning_rate": 1.962962962962963e-05,
      "loss": 1.5967,
      "step": 160
    },
    {
      "epoch": 0.01863425925925926,
      "grad_norm": 8.932489395141602,
      "learning_rate": 1.962731481481482e-05,
      "loss": 1.8475,
      "step": 161
    },
    {
      "epoch": 0.01875,
      "grad_norm": 10.277194023132324,
      "learning_rate": 1.9625e-05,
      "loss": 1.593,
      "step": 162
    },
    {
      "epoch": 0.018865740740740742,
      "grad_norm": 6.23793888092041,
      "learning_rate": 1.9622685185185187e-05,
      "loss": 1.1642,
      "step": 163
    },
    {
      "epoch": 0.01898148148148148,
      "grad_norm": 7.054962158203125,
      "learning_rate": 1.962037037037037e-05,
      "loss": 1.9054,
      "step": 164
    },
    {
      "epoch": 0.019097222222222224,
      "grad_norm": 6.900039196014404,
      "learning_rate": 1.961805555555556e-05,
      "loss": 1.6212,
      "step": 165
    },
    {
      "epoch": 0.019212962962962963,
      "grad_norm": 7.020420074462891,
      "learning_rate": 1.9615740740740743e-05,
      "loss": 1.7771,
      "step": 166
    },
    {
      "epoch": 0.019328703703703702,
      "grad_norm": 7.042113304138184,
      "learning_rate": 1.9613425925925927e-05,
      "loss": 1.4493,
      "step": 167
    },
    {
      "epoch": 0.019444444444444445,
      "grad_norm": 14.36280632019043,
      "learning_rate": 1.961111111111111e-05,
      "loss": 1.5848,
      "step": 168
    },
    {
      "epoch": 0.019560185185185184,
      "grad_norm": 8.361939430236816,
      "learning_rate": 1.9608796296296296e-05,
      "loss": 1.3344,
      "step": 169
    },
    {
      "epoch": 0.019675925925925927,
      "grad_norm": 6.2794108390808105,
      "learning_rate": 1.9606481481481483e-05,
      "loss": 1.2475,
      "step": 170
    },
    {
      "epoch": 0.019791666666666666,
      "grad_norm": 6.323544502258301,
      "learning_rate": 1.9604166666666668e-05,
      "loss": 1.4252,
      "step": 171
    },
    {
      "epoch": 0.01990740740740741,
      "grad_norm": 4.873958110809326,
      "learning_rate": 1.9601851851851855e-05,
      "loss": 1.4045,
      "step": 172
    },
    {
      "epoch": 0.020023148148148148,
      "grad_norm": 5.106293201446533,
      "learning_rate": 1.959953703703704e-05,
      "loss": 1.1248,
      "step": 173
    },
    {
      "epoch": 0.02013888888888889,
      "grad_norm": 5.0389509201049805,
      "learning_rate": 1.9597222222222224e-05,
      "loss": 1.3446,
      "step": 174
    },
    {
      "epoch": 0.02025462962962963,
      "grad_norm": 5.90389347076416,
      "learning_rate": 1.9594907407407408e-05,
      "loss": 1.5427,
      "step": 175
    },
    {
      "epoch": 0.020370370370370372,
      "grad_norm": 5.884286403656006,
      "learning_rate": 1.9592592592592596e-05,
      "loss": 1.2385,
      "step": 176
    },
    {
      "epoch": 0.02048611111111111,
      "grad_norm": 9.325323104858398,
      "learning_rate": 1.959027777777778e-05,
      "loss": 1.5805,
      "step": 177
    },
    {
      "epoch": 0.02060185185185185,
      "grad_norm": 7.621048450469971,
      "learning_rate": 1.9587962962962964e-05,
      "loss": 1.7213,
      "step": 178
    },
    {
      "epoch": 0.020717592592592593,
      "grad_norm": 5.206976413726807,
      "learning_rate": 1.9585648148148152e-05,
      "loss": 1.4996,
      "step": 179
    },
    {
      "epoch": 0.020833333333333332,
      "grad_norm": 7.154023170471191,
      "learning_rate": 1.9583333333333333e-05,
      "loss": 1.724,
      "step": 180
    },
    {
      "epoch": 0.020949074074074075,
      "grad_norm": 6.159602165222168,
      "learning_rate": 1.958101851851852e-05,
      "loss": 1.7832,
      "step": 181
    },
    {
      "epoch": 0.021064814814814814,
      "grad_norm": 7.912835121154785,
      "learning_rate": 1.9578703703703705e-05,
      "loss": 1.5985,
      "step": 182
    },
    {
      "epoch": 0.021180555555555557,
      "grad_norm": 6.675532341003418,
      "learning_rate": 1.9576388888888892e-05,
      "loss": 1.4997,
      "step": 183
    },
    {
      "epoch": 0.021296296296296296,
      "grad_norm": 6.954130172729492,
      "learning_rate": 1.9574074074074077e-05,
      "loss": 1.3619,
      "step": 184
    },
    {
      "epoch": 0.02141203703703704,
      "grad_norm": 6.163214206695557,
      "learning_rate": 1.957175925925926e-05,
      "loss": 1.2695,
      "step": 185
    },
    {
      "epoch": 0.021527777777777778,
      "grad_norm": 5.726155757904053,
      "learning_rate": 1.9569444444444445e-05,
      "loss": 1.1665,
      "step": 186
    },
    {
      "epoch": 0.021643518518518517,
      "grad_norm": 7.195199012756348,
      "learning_rate": 1.956712962962963e-05,
      "loss": 1.8792,
      "step": 187
    },
    {
      "epoch": 0.02175925925925926,
      "grad_norm": 10.322378158569336,
      "learning_rate": 1.9564814814814817e-05,
      "loss": 1.5812,
      "step": 188
    },
    {
      "epoch": 0.021875,
      "grad_norm": 21.261959075927734,
      "learning_rate": 1.95625e-05,
      "loss": 1.0476,
      "step": 189
    },
    {
      "epoch": 0.02199074074074074,
      "grad_norm": 5.806355953216553,
      "learning_rate": 1.956018518518519e-05,
      "loss": 1.3357,
      "step": 190
    },
    {
      "epoch": 0.02210648148148148,
      "grad_norm": 9.688827514648438,
      "learning_rate": 1.955787037037037e-05,
      "loss": 1.9815,
      "step": 191
    },
    {
      "epoch": 0.022222222222222223,
      "grad_norm": 3.4454891681671143,
      "learning_rate": 1.9555555555555557e-05,
      "loss": 0.9469,
      "step": 192
    },
    {
      "epoch": 0.022337962962962962,
      "grad_norm": 7.7801432609558105,
      "learning_rate": 1.955324074074074e-05,
      "loss": 1.5407,
      "step": 193
    },
    {
      "epoch": 0.022453703703703705,
      "grad_norm": 6.090869903564453,
      "learning_rate": 1.955092592592593e-05,
      "loss": 1.4825,
      "step": 194
    },
    {
      "epoch": 0.022569444444444444,
      "grad_norm": 6.800838947296143,
      "learning_rate": 1.9548611111111113e-05,
      "loss": 1.5425,
      "step": 195
    },
    {
      "epoch": 0.022685185185185187,
      "grad_norm": 7.625843524932861,
      "learning_rate": 1.9546296296296298e-05,
      "loss": 1.4722,
      "step": 196
    },
    {
      "epoch": 0.022800925925925926,
      "grad_norm": 6.95320463180542,
      "learning_rate": 1.9543981481481482e-05,
      "loss": 1.7197,
      "step": 197
    },
    {
      "epoch": 0.022916666666666665,
      "grad_norm": 7.990474224090576,
      "learning_rate": 1.9541666666666666e-05,
      "loss": 1.3348,
      "step": 198
    },
    {
      "epoch": 0.023032407407407408,
      "grad_norm": 5.287327766418457,
      "learning_rate": 1.9539351851851854e-05,
      "loss": 1.1539,
      "step": 199
    },
    {
      "epoch": 0.023148148148148147,
      "grad_norm": 5.6665425300598145,
      "learning_rate": 1.9537037037037038e-05,
      "loss": 1.5435,
      "step": 200
    },
    {
      "epoch": 0.02326388888888889,
      "grad_norm": 9.108843803405762,
      "learning_rate": 1.9534722222222226e-05,
      "loss": 2.0715,
      "step": 201
    },
    {
      "epoch": 0.02337962962962963,
      "grad_norm": 7.261268615722656,
      "learning_rate": 1.953240740740741e-05,
      "loss": 1.7833,
      "step": 202
    },
    {
      "epoch": 0.02349537037037037,
      "grad_norm": 5.991909503936768,
      "learning_rate": 1.9530092592592594e-05,
      "loss": 1.4215,
      "step": 203
    },
    {
      "epoch": 0.02361111111111111,
      "grad_norm": 6.697367191314697,
      "learning_rate": 1.952777777777778e-05,
      "loss": 1.7458,
      "step": 204
    },
    {
      "epoch": 0.023726851851851853,
      "grad_norm": 8.323158264160156,
      "learning_rate": 1.9525462962962963e-05,
      "loss": 1.9767,
      "step": 205
    },
    {
      "epoch": 0.023842592592592592,
      "grad_norm": 7.256472587585449,
      "learning_rate": 1.952314814814815e-05,
      "loss": 1.9077,
      "step": 206
    },
    {
      "epoch": 0.023958333333333335,
      "grad_norm": 7.774311065673828,
      "learning_rate": 1.9520833333333335e-05,
      "loss": 1.9299,
      "step": 207
    },
    {
      "epoch": 0.024074074074074074,
      "grad_norm": 4.512392997741699,
      "learning_rate": 1.9518518518518522e-05,
      "loss": 1.4175,
      "step": 208
    },
    {
      "epoch": 0.024189814814814813,
      "grad_norm": 7.975398540496826,
      "learning_rate": 1.9516203703703703e-05,
      "loss": 1.6033,
      "step": 209
    },
    {
      "epoch": 0.024305555555555556,
      "grad_norm": 6.171725273132324,
      "learning_rate": 1.951388888888889e-05,
      "loss": 1.7953,
      "step": 210
    },
    {
      "epoch": 0.024421296296296295,
      "grad_norm": 6.233512878417969,
      "learning_rate": 1.9511574074074075e-05,
      "loss": 1.4768,
      "step": 211
    },
    {
      "epoch": 0.024537037037037038,
      "grad_norm": 8.141648292541504,
      "learning_rate": 1.9509259259259263e-05,
      "loss": 1.6996,
      "step": 212
    },
    {
      "epoch": 0.024652777777777777,
      "grad_norm": 7.830441951751709,
      "learning_rate": 1.9506944444444447e-05,
      "loss": 1.5122,
      "step": 213
    },
    {
      "epoch": 0.02476851851851852,
      "grad_norm": 8.146535873413086,
      "learning_rate": 1.950462962962963e-05,
      "loss": 1.9078,
      "step": 214
    },
    {
      "epoch": 0.02488425925925926,
      "grad_norm": 8.600090980529785,
      "learning_rate": 1.9502314814814815e-05,
      "loss": 2.0327,
      "step": 215
    },
    {
      "epoch": 0.025,
      "grad_norm": 7.676675319671631,
      "learning_rate": 1.95e-05,
      "loss": 1.3123,
      "step": 216
    },
    {
      "epoch": 0.02511574074074074,
      "grad_norm": 6.5683512687683105,
      "learning_rate": 1.9497685185185187e-05,
      "loss": 1.5564,
      "step": 217
    },
    {
      "epoch": 0.025231481481481483,
      "grad_norm": 6.251888275146484,
      "learning_rate": 1.949537037037037e-05,
      "loss": 0.9695,
      "step": 218
    },
    {
      "epoch": 0.025347222222222222,
      "grad_norm": 5.0568976402282715,
      "learning_rate": 1.949305555555556e-05,
      "loss": 1.0132,
      "step": 219
    },
    {
      "epoch": 0.02546296296296296,
      "grad_norm": 6.183098793029785,
      "learning_rate": 1.9490740740740743e-05,
      "loss": 1.2157,
      "step": 220
    },
    {
      "epoch": 0.025578703703703704,
      "grad_norm": 5.073744773864746,
      "learning_rate": 1.9488425925925928e-05,
      "loss": 1.0919,
      "step": 221
    },
    {
      "epoch": 0.025694444444444443,
      "grad_norm": 8.50080394744873,
      "learning_rate": 1.9486111111111112e-05,
      "loss": 1.2699,
      "step": 222
    },
    {
      "epoch": 0.025810185185185186,
      "grad_norm": 8.374058723449707,
      "learning_rate": 1.94837962962963e-05,
      "loss": 1.8355,
      "step": 223
    },
    {
      "epoch": 0.025925925925925925,
      "grad_norm": 6.339303493499756,
      "learning_rate": 1.9481481481481484e-05,
      "loss": 1.3666,
      "step": 224
    },
    {
      "epoch": 0.026041666666666668,
      "grad_norm": 6.665476322174072,
      "learning_rate": 1.9479166666666668e-05,
      "loss": 1.2522,
      "step": 225
    },
    {
      "epoch": 0.026157407407407407,
      "grad_norm": 6.205174446105957,
      "learning_rate": 1.9476851851851852e-05,
      "loss": 1.6379,
      "step": 226
    },
    {
      "epoch": 0.02627314814814815,
      "grad_norm": 5.5422587394714355,
      "learning_rate": 1.9474537037037037e-05,
      "loss": 1.2919,
      "step": 227
    },
    {
      "epoch": 0.02638888888888889,
      "grad_norm": 7.500425338745117,
      "learning_rate": 1.9472222222222224e-05,
      "loss": 1.3033,
      "step": 228
    },
    {
      "epoch": 0.026504629629629628,
      "grad_norm": 7.137895107269287,
      "learning_rate": 1.946990740740741e-05,
      "loss": 1.9308,
      "step": 229
    },
    {
      "epoch": 0.02662037037037037,
      "grad_norm": 4.577306270599365,
      "learning_rate": 1.9467592592592596e-05,
      "loss": 0.9023,
      "step": 230
    },
    {
      "epoch": 0.02673611111111111,
      "grad_norm": 4.196961402893066,
      "learning_rate": 1.946527777777778e-05,
      "loss": 0.9628,
      "step": 231
    },
    {
      "epoch": 0.026851851851851852,
      "grad_norm": 5.287446975708008,
      "learning_rate": 1.9462962962962965e-05,
      "loss": 1.4626,
      "step": 232
    },
    {
      "epoch": 0.02696759259259259,
      "grad_norm": 7.4490132331848145,
      "learning_rate": 1.946064814814815e-05,
      "loss": 1.9243,
      "step": 233
    },
    {
      "epoch": 0.027083333333333334,
      "grad_norm": 5.94210958480835,
      "learning_rate": 1.9458333333333333e-05,
      "loss": 1.3127,
      "step": 234
    },
    {
      "epoch": 0.027199074074074073,
      "grad_norm": 7.2232184410095215,
      "learning_rate": 1.945601851851852e-05,
      "loss": 1.8127,
      "step": 235
    },
    {
      "epoch": 0.027314814814814816,
      "grad_norm": 7.296854496002197,
      "learning_rate": 1.9453703703703705e-05,
      "loss": 1.2617,
      "step": 236
    },
    {
      "epoch": 0.027430555555555555,
      "grad_norm": 6.386005401611328,
      "learning_rate": 1.9451388888888893e-05,
      "loss": 1.159,
      "step": 237
    },
    {
      "epoch": 0.027546296296296298,
      "grad_norm": 6.261192321777344,
      "learning_rate": 1.9449074074074073e-05,
      "loss": 1.5057,
      "step": 238
    },
    {
      "epoch": 0.027662037037037037,
      "grad_norm": 4.987699031829834,
      "learning_rate": 1.944675925925926e-05,
      "loss": 1.0715,
      "step": 239
    },
    {
      "epoch": 0.027777777777777776,
      "grad_norm": 4.940311431884766,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 0.9481,
      "step": 240
    },
    {
      "epoch": 0.02789351851851852,
      "grad_norm": 8.185747146606445,
      "learning_rate": 1.9442129629629633e-05,
      "loss": 1.0468,
      "step": 241
    },
    {
      "epoch": 0.028009259259259258,
      "grad_norm": 4.836315631866455,
      "learning_rate": 1.9439814814814817e-05,
      "loss": 0.9883,
      "step": 242
    },
    {
      "epoch": 0.028125,
      "grad_norm": 7.860507011413574,
      "learning_rate": 1.94375e-05,
      "loss": 1.8121,
      "step": 243
    },
    {
      "epoch": 0.02824074074074074,
      "grad_norm": 5.229879379272461,
      "learning_rate": 1.9435185185185186e-05,
      "loss": 1.0715,
      "step": 244
    },
    {
      "epoch": 0.028356481481481483,
      "grad_norm": 5.848942756652832,
      "learning_rate": 1.943287037037037e-05,
      "loss": 1.2028,
      "step": 245
    },
    {
      "epoch": 0.02847222222222222,
      "grad_norm": 5.929676055908203,
      "learning_rate": 1.9430555555555558e-05,
      "loss": 1.0601,
      "step": 246
    },
    {
      "epoch": 0.028587962962962964,
      "grad_norm": 4.859242916107178,
      "learning_rate": 1.9428240740740742e-05,
      "loss": 0.9811,
      "step": 247
    },
    {
      "epoch": 0.028703703703703703,
      "grad_norm": 14.717595100402832,
      "learning_rate": 1.942592592592593e-05,
      "loss": 1.4499,
      "step": 248
    },
    {
      "epoch": 0.028819444444444446,
      "grad_norm": 5.350089073181152,
      "learning_rate": 1.9423611111111114e-05,
      "loss": 0.9094,
      "step": 249
    },
    {
      "epoch": 0.028935185185185185,
      "grad_norm": 7.415565490722656,
      "learning_rate": 1.9421296296296298e-05,
      "loss": 1.0059,
      "step": 250
    },
    {
      "epoch": 0.029050925925925924,
      "grad_norm": 3.3733887672424316,
      "learning_rate": 1.9418981481481482e-05,
      "loss": 0.7544,
      "step": 251
    },
    {
      "epoch": 0.029166666666666667,
      "grad_norm": 4.1808905601501465,
      "learning_rate": 1.9416666666666667e-05,
      "loss": 0.9214,
      "step": 252
    },
    {
      "epoch": 0.029282407407407406,
      "grad_norm": 4.565878391265869,
      "learning_rate": 1.9414351851851854e-05,
      "loss": 0.8451,
      "step": 253
    },
    {
      "epoch": 0.02939814814814815,
      "grad_norm": 9.220598220825195,
      "learning_rate": 1.941203703703704e-05,
      "loss": 2.0234,
      "step": 254
    },
    {
      "epoch": 0.029513888888888888,
      "grad_norm": 9.103293418884277,
      "learning_rate": 1.9409722222222226e-05,
      "loss": 1.294,
      "step": 255
    },
    {
      "epoch": 0.02962962962962963,
      "grad_norm": 9.048334121704102,
      "learning_rate": 1.9407407407407407e-05,
      "loss": 1.2072,
      "step": 256
    },
    {
      "epoch": 0.02974537037037037,
      "grad_norm": 7.3248162269592285,
      "learning_rate": 1.9405092592592595e-05,
      "loss": 0.7548,
      "step": 257
    },
    {
      "epoch": 0.029861111111111113,
      "grad_norm": 5.165322303771973,
      "learning_rate": 1.940277777777778e-05,
      "loss": 0.7446,
      "step": 258
    },
    {
      "epoch": 0.029976851851851852,
      "grad_norm": 5.025292873382568,
      "learning_rate": 1.9400462962962966e-05,
      "loss": 1.2638,
      "step": 259
    },
    {
      "epoch": 0.03009259259259259,
      "grad_norm": 4.107083797454834,
      "learning_rate": 1.939814814814815e-05,
      "loss": 1.1243,
      "step": 260
    },
    {
      "epoch": 0.030208333333333334,
      "grad_norm": 5.9913716316223145,
      "learning_rate": 1.9395833333333335e-05,
      "loss": 1.2577,
      "step": 261
    },
    {
      "epoch": 0.030324074074074073,
      "grad_norm": 12.437383651733398,
      "learning_rate": 1.939351851851852e-05,
      "loss": 1.5012,
      "step": 262
    },
    {
      "epoch": 0.030439814814814815,
      "grad_norm": 4.384948253631592,
      "learning_rate": 1.9391203703703703e-05,
      "loss": 1.1999,
      "step": 263
    },
    {
      "epoch": 0.030555555555555555,
      "grad_norm": 6.8313307762146,
      "learning_rate": 1.938888888888889e-05,
      "loss": 1.1578,
      "step": 264
    },
    {
      "epoch": 0.030671296296296297,
      "grad_norm": 8.60260009765625,
      "learning_rate": 1.9386574074074075e-05,
      "loss": 1.2838,
      "step": 265
    },
    {
      "epoch": 0.030787037037037036,
      "grad_norm": 8.193117141723633,
      "learning_rate": 1.9384259259259263e-05,
      "loss": 1.4937,
      "step": 266
    },
    {
      "epoch": 0.03090277777777778,
      "grad_norm": 7.519168853759766,
      "learning_rate": 1.9381944444444447e-05,
      "loss": 1.9189,
      "step": 267
    },
    {
      "epoch": 0.031018518518518518,
      "grad_norm": 5.5613226890563965,
      "learning_rate": 1.937962962962963e-05,
      "loss": 1.3865,
      "step": 268
    },
    {
      "epoch": 0.03113425925925926,
      "grad_norm": 9.043374061584473,
      "learning_rate": 1.9377314814814816e-05,
      "loss": 1.5834,
      "step": 269
    },
    {
      "epoch": 0.03125,
      "grad_norm": 7.359215259552002,
      "learning_rate": 1.9375e-05,
      "loss": 1.3825,
      "step": 270
    },
    {
      "epoch": 0.03136574074074074,
      "grad_norm": 15.954339981079102,
      "learning_rate": 1.9372685185185188e-05,
      "loss": 1.403,
      "step": 271
    },
    {
      "epoch": 0.03148148148148148,
      "grad_norm": 6.323440074920654,
      "learning_rate": 1.9370370370370372e-05,
      "loss": 1.2646,
      "step": 272
    },
    {
      "epoch": 0.03159722222222222,
      "grad_norm": 10.689743041992188,
      "learning_rate": 1.9368055555555556e-05,
      "loss": 2.142,
      "step": 273
    },
    {
      "epoch": 0.031712962962962964,
      "grad_norm": 8.381004333496094,
      "learning_rate": 1.936574074074074e-05,
      "loss": 1.5377,
      "step": 274
    },
    {
      "epoch": 0.031828703703703706,
      "grad_norm": 11.411711692810059,
      "learning_rate": 1.9363425925925928e-05,
      "loss": 1.4463,
      "step": 275
    },
    {
      "epoch": 0.03194444444444444,
      "grad_norm": 6.318783283233643,
      "learning_rate": 1.9361111111111112e-05,
      "loss": 0.7493,
      "step": 276
    },
    {
      "epoch": 0.032060185185185185,
      "grad_norm": 4.083025932312012,
      "learning_rate": 1.93587962962963e-05,
      "loss": 0.8446,
      "step": 277
    },
    {
      "epoch": 0.03217592592592593,
      "grad_norm": 15.5701322555542,
      "learning_rate": 1.9356481481481484e-05,
      "loss": 1.1568,
      "step": 278
    },
    {
      "epoch": 0.03229166666666667,
      "grad_norm": 13.549409866333008,
      "learning_rate": 1.935416666666667e-05,
      "loss": 1.5868,
      "step": 279
    },
    {
      "epoch": 0.032407407407407406,
      "grad_norm": 6.466709136962891,
      "learning_rate": 1.9351851851851853e-05,
      "loss": 1.6616,
      "step": 280
    },
    {
      "epoch": 0.03252314814814815,
      "grad_norm": 4.200021743774414,
      "learning_rate": 1.9349537037037037e-05,
      "loss": 0.8147,
      "step": 281
    },
    {
      "epoch": 0.03263888888888889,
      "grad_norm": 8.222716331481934,
      "learning_rate": 1.9347222222222225e-05,
      "loss": 1.3522,
      "step": 282
    },
    {
      "epoch": 0.03275462962962963,
      "grad_norm": 9.077616691589355,
      "learning_rate": 1.934490740740741e-05,
      "loss": 1.0264,
      "step": 283
    },
    {
      "epoch": 0.03287037037037037,
      "grad_norm": 7.3175458908081055,
      "learning_rate": 1.9342592592592596e-05,
      "loss": 1.1299,
      "step": 284
    },
    {
      "epoch": 0.03298611111111111,
      "grad_norm": 4.727084159851074,
      "learning_rate": 1.9340277777777777e-05,
      "loss": 0.8856,
      "step": 285
    },
    {
      "epoch": 0.033101851851851855,
      "grad_norm": 7.67744779586792,
      "learning_rate": 1.9337962962962965e-05,
      "loss": 1.7096,
      "step": 286
    },
    {
      "epoch": 0.03321759259259259,
      "grad_norm": 9.661006927490234,
      "learning_rate": 1.933564814814815e-05,
      "loss": 1.7355,
      "step": 287
    },
    {
      "epoch": 0.03333333333333333,
      "grad_norm": 6.365168571472168,
      "learning_rate": 1.9333333333333333e-05,
      "loss": 1.1784,
      "step": 288
    },
    {
      "epoch": 0.033449074074074076,
      "grad_norm": 5.4882025718688965,
      "learning_rate": 1.933101851851852e-05,
      "loss": 0.8064,
      "step": 289
    },
    {
      "epoch": 0.03356481481481482,
      "grad_norm": 4.8790130615234375,
      "learning_rate": 1.9328703703703705e-05,
      "loss": 1.2365,
      "step": 290
    },
    {
      "epoch": 0.033680555555555554,
      "grad_norm": 5.16702127456665,
      "learning_rate": 1.932638888888889e-05,
      "loss": 1.4345,
      "step": 291
    },
    {
      "epoch": 0.033796296296296297,
      "grad_norm": 10.912044525146484,
      "learning_rate": 1.9324074074074074e-05,
      "loss": 1.1115,
      "step": 292
    },
    {
      "epoch": 0.03391203703703704,
      "grad_norm": 6.227447509765625,
      "learning_rate": 1.932175925925926e-05,
      "loss": 0.9132,
      "step": 293
    },
    {
      "epoch": 0.034027777777777775,
      "grad_norm": 7.244505405426025,
      "learning_rate": 1.9319444444444446e-05,
      "loss": 0.9017,
      "step": 294
    },
    {
      "epoch": 0.03414351851851852,
      "grad_norm": 16.00440788269043,
      "learning_rate": 1.9317129629629633e-05,
      "loss": 1.4258,
      "step": 295
    },
    {
      "epoch": 0.03425925925925926,
      "grad_norm": 7.59874153137207,
      "learning_rate": 1.9314814814814818e-05,
      "loss": 1.0625,
      "step": 296
    },
    {
      "epoch": 0.034375,
      "grad_norm": 11.753899574279785,
      "learning_rate": 1.9312500000000002e-05,
      "loss": 1.3548,
      "step": 297
    },
    {
      "epoch": 0.03449074074074074,
      "grad_norm": 3.5632500648498535,
      "learning_rate": 1.9310185185185186e-05,
      "loss": 0.6993,
      "step": 298
    },
    {
      "epoch": 0.03460648148148148,
      "grad_norm": 9.275835037231445,
      "learning_rate": 1.930787037037037e-05,
      "loss": 2.0841,
      "step": 299
    },
    {
      "epoch": 0.034722222222222224,
      "grad_norm": 3.0309946537017822,
      "learning_rate": 1.9305555555555558e-05,
      "loss": 0.8489,
      "step": 300
    },
    {
      "epoch": 0.034837962962962966,
      "grad_norm": 11.043767929077148,
      "learning_rate": 1.9303240740740742e-05,
      "loss": 1.0868,
      "step": 301
    },
    {
      "epoch": 0.0349537037037037,
      "grad_norm": 7.2854838371276855,
      "learning_rate": 1.930092592592593e-05,
      "loss": 0.954,
      "step": 302
    },
    {
      "epoch": 0.035069444444444445,
      "grad_norm": 5.865585803985596,
      "learning_rate": 1.929861111111111e-05,
      "loss": 0.5829,
      "step": 303
    },
    {
      "epoch": 0.03518518518518519,
      "grad_norm": 10.74530029296875,
      "learning_rate": 1.92962962962963e-05,
      "loss": 1.4048,
      "step": 304
    },
    {
      "epoch": 0.03530092592592592,
      "grad_norm": 5.80052375793457,
      "learning_rate": 1.9293981481481483e-05,
      "loss": 0.8975,
      "step": 305
    },
    {
      "epoch": 0.035416666666666666,
      "grad_norm": 5.900633811950684,
      "learning_rate": 1.9291666666666667e-05,
      "loss": 1.2819,
      "step": 306
    },
    {
      "epoch": 0.03553240740740741,
      "grad_norm": 9.78097915649414,
      "learning_rate": 1.9289351851851855e-05,
      "loss": 1.5355,
      "step": 307
    },
    {
      "epoch": 0.03564814814814815,
      "grad_norm": 3.243520498275757,
      "learning_rate": 1.928703703703704e-05,
      "loss": 0.5418,
      "step": 308
    },
    {
      "epoch": 0.03576388888888889,
      "grad_norm": 6.23818302154541,
      "learning_rate": 1.9284722222222223e-05,
      "loss": 0.8981,
      "step": 309
    },
    {
      "epoch": 0.03587962962962963,
      "grad_norm": 4.0724101066589355,
      "learning_rate": 1.9282407407407407e-05,
      "loss": 0.8857,
      "step": 310
    },
    {
      "epoch": 0.03599537037037037,
      "grad_norm": 4.132826805114746,
      "learning_rate": 1.9280092592592595e-05,
      "loss": 1.0231,
      "step": 311
    },
    {
      "epoch": 0.03611111111111111,
      "grad_norm": 4.152525901794434,
      "learning_rate": 1.927777777777778e-05,
      "loss": 0.844,
      "step": 312
    },
    {
      "epoch": 0.03622685185185185,
      "grad_norm": 4.573829650878906,
      "learning_rate": 1.9275462962962967e-05,
      "loss": 0.785,
      "step": 313
    },
    {
      "epoch": 0.03634259259259259,
      "grad_norm": 8.20241928100586,
      "learning_rate": 1.927314814814815e-05,
      "loss": 1.7184,
      "step": 314
    },
    {
      "epoch": 0.036458333333333336,
      "grad_norm": 4.858456134796143,
      "learning_rate": 1.9270833333333335e-05,
      "loss": 0.9356,
      "step": 315
    },
    {
      "epoch": 0.03657407407407407,
      "grad_norm": 4.32342529296875,
      "learning_rate": 1.926851851851852e-05,
      "loss": 0.8077,
      "step": 316
    },
    {
      "epoch": 0.036689814814814814,
      "grad_norm": 16.301979064941406,
      "learning_rate": 1.9266203703703704e-05,
      "loss": 1.2306,
      "step": 317
    },
    {
      "epoch": 0.03680555555555556,
      "grad_norm": 2.9324851036071777,
      "learning_rate": 1.926388888888889e-05,
      "loss": 0.6331,
      "step": 318
    },
    {
      "epoch": 0.0369212962962963,
      "grad_norm": 3.5670549869537354,
      "learning_rate": 1.9261574074074076e-05,
      "loss": 0.6612,
      "step": 319
    },
    {
      "epoch": 0.037037037037037035,
      "grad_norm": 6.918369770050049,
      "learning_rate": 1.925925925925926e-05,
      "loss": 0.977,
      "step": 320
    },
    {
      "epoch": 0.03715277777777778,
      "grad_norm": 5.931670665740967,
      "learning_rate": 1.9256944444444444e-05,
      "loss": 1.4469,
      "step": 321
    },
    {
      "epoch": 0.03726851851851852,
      "grad_norm": 2.462294816970825,
      "learning_rate": 1.9254629629629632e-05,
      "loss": 0.5262,
      "step": 322
    },
    {
      "epoch": 0.037384259259259256,
      "grad_norm": 4.188912868499756,
      "learning_rate": 1.9252314814814816e-05,
      "loss": 0.9221,
      "step": 323
    },
    {
      "epoch": 0.0375,
      "grad_norm": 7.613870620727539,
      "learning_rate": 1.925e-05,
      "loss": 1.175,
      "step": 324
    },
    {
      "epoch": 0.03761574074074074,
      "grad_norm": 10.249424934387207,
      "learning_rate": 1.9247685185185188e-05,
      "loss": 1.3965,
      "step": 325
    },
    {
      "epoch": 0.037731481481481484,
      "grad_norm": 5.659093379974365,
      "learning_rate": 1.9245370370370372e-05,
      "loss": 0.9823,
      "step": 326
    },
    {
      "epoch": 0.03784722222222222,
      "grad_norm": 19.716928482055664,
      "learning_rate": 1.9243055555555556e-05,
      "loss": 1.1809,
      "step": 327
    },
    {
      "epoch": 0.03796296296296296,
      "grad_norm": 5.755324363708496,
      "learning_rate": 1.924074074074074e-05,
      "loss": 0.7837,
      "step": 328
    },
    {
      "epoch": 0.038078703703703705,
      "grad_norm": 4.666869640350342,
      "learning_rate": 1.923842592592593e-05,
      "loss": 0.7559,
      "step": 329
    },
    {
      "epoch": 0.03819444444444445,
      "grad_norm": 10.11201286315918,
      "learning_rate": 1.9236111111111113e-05,
      "loss": 1.7791,
      "step": 330
    },
    {
      "epoch": 0.03831018518518518,
      "grad_norm": 5.73575496673584,
      "learning_rate": 1.92337962962963e-05,
      "loss": 0.7416,
      "step": 331
    },
    {
      "epoch": 0.038425925925925926,
      "grad_norm": 2.9955828189849854,
      "learning_rate": 1.923148148148148e-05,
      "loss": 0.5694,
      "step": 332
    },
    {
      "epoch": 0.03854166666666667,
      "grad_norm": 12.192222595214844,
      "learning_rate": 1.922916666666667e-05,
      "loss": 1.9405,
      "step": 333
    },
    {
      "epoch": 0.038657407407407404,
      "grad_norm": 4.041222095489502,
      "learning_rate": 1.9226851851851853e-05,
      "loss": 0.6067,
      "step": 334
    },
    {
      "epoch": 0.03877314814814815,
      "grad_norm": 4.060816764831543,
      "learning_rate": 1.9224537037037037e-05,
      "loss": 0.8532,
      "step": 335
    },
    {
      "epoch": 0.03888888888888889,
      "grad_norm": 6.306008815765381,
      "learning_rate": 1.9222222222222225e-05,
      "loss": 0.8002,
      "step": 336
    },
    {
      "epoch": 0.03900462962962963,
      "grad_norm": 4.2312912940979,
      "learning_rate": 1.921990740740741e-05,
      "loss": 0.8084,
      "step": 337
    },
    {
      "epoch": 0.03912037037037037,
      "grad_norm": 10.61928653717041,
      "learning_rate": 1.9217592592592593e-05,
      "loss": 1.5512,
      "step": 338
    },
    {
      "epoch": 0.03923611111111111,
      "grad_norm": 3.9523003101348877,
      "learning_rate": 1.9215277777777778e-05,
      "loss": 1.0123,
      "step": 339
    },
    {
      "epoch": 0.03935185185185185,
      "grad_norm": 5.135763645172119,
      "learning_rate": 1.9212962962962965e-05,
      "loss": 1.0208,
      "step": 340
    },
    {
      "epoch": 0.039467592592592596,
      "grad_norm": 11.017435073852539,
      "learning_rate": 1.921064814814815e-05,
      "loss": 1.0924,
      "step": 341
    },
    {
      "epoch": 0.03958333333333333,
      "grad_norm": 8.260395050048828,
      "learning_rate": 1.9208333333333337e-05,
      "loss": 1.1706,
      "step": 342
    },
    {
      "epoch": 0.039699074074074074,
      "grad_norm": 2.176964044570923,
      "learning_rate": 1.920601851851852e-05,
      "loss": 0.5192,
      "step": 343
    },
    {
      "epoch": 0.03981481481481482,
      "grad_norm": 2.27302622795105,
      "learning_rate": 1.9203703703703706e-05,
      "loss": 0.4236,
      "step": 344
    },
    {
      "epoch": 0.03993055555555555,
      "grad_norm": 4.901488780975342,
      "learning_rate": 1.920138888888889e-05,
      "loss": 0.94,
      "step": 345
    },
    {
      "epoch": 0.040046296296296295,
      "grad_norm": 9.140849113464355,
      "learning_rate": 1.9199074074074074e-05,
      "loss": 1.2403,
      "step": 346
    },
    {
      "epoch": 0.04016203703703704,
      "grad_norm": 4.2225236892700195,
      "learning_rate": 1.9196759259259262e-05,
      "loss": 0.9285,
      "step": 347
    },
    {
      "epoch": 0.04027777777777778,
      "grad_norm": 4.563241958618164,
      "learning_rate": 1.9194444444444446e-05,
      "loss": 0.8565,
      "step": 348
    },
    {
      "epoch": 0.040393518518518516,
      "grad_norm": 4.124519348144531,
      "learning_rate": 1.9192129629629634e-05,
      "loss": 0.5454,
      "step": 349
    },
    {
      "epoch": 0.04050925925925926,
      "grad_norm": 2.397308349609375,
      "learning_rate": 1.9189814814814815e-05,
      "loss": 0.5341,
      "step": 350
    },
    {
      "epoch": 0.040625,
      "grad_norm": 6.335616111755371,
      "learning_rate": 1.9187500000000002e-05,
      "loss": 1.0588,
      "step": 351
    },
    {
      "epoch": 0.040740740740740744,
      "grad_norm": 21.665977478027344,
      "learning_rate": 1.9185185185185186e-05,
      "loss": 1.6975,
      "step": 352
    },
    {
      "epoch": 0.04085648148148148,
      "grad_norm": 2.7691280841827393,
      "learning_rate": 1.918287037037037e-05,
      "loss": 0.5453,
      "step": 353
    },
    {
      "epoch": 0.04097222222222222,
      "grad_norm": 4.051657199859619,
      "learning_rate": 1.918055555555556e-05,
      "loss": 0.5689,
      "step": 354
    },
    {
      "epoch": 0.041087962962962965,
      "grad_norm": 2.1980721950531006,
      "learning_rate": 1.9178240740740743e-05,
      "loss": 0.3681,
      "step": 355
    },
    {
      "epoch": 0.0412037037037037,
      "grad_norm": 3.2632598876953125,
      "learning_rate": 1.9175925925925927e-05,
      "loss": 0.7448,
      "step": 356
    },
    {
      "epoch": 0.04131944444444444,
      "grad_norm": 3.268714666366577,
      "learning_rate": 1.917361111111111e-05,
      "loss": 0.6352,
      "step": 357
    },
    {
      "epoch": 0.041435185185185186,
      "grad_norm": 18.526039123535156,
      "learning_rate": 1.91712962962963e-05,
      "loss": 0.9402,
      "step": 358
    },
    {
      "epoch": 0.04155092592592593,
      "grad_norm": 4.487000465393066,
      "learning_rate": 1.9168981481481483e-05,
      "loss": 0.5426,
      "step": 359
    },
    {
      "epoch": 0.041666666666666664,
      "grad_norm": 9.75832748413086,
      "learning_rate": 1.916666666666667e-05,
      "loss": 1.5201,
      "step": 360
    },
    {
      "epoch": 0.04178240740740741,
      "grad_norm": 2.718775510787964,
      "learning_rate": 1.916435185185185e-05,
      "loss": 0.5835,
      "step": 361
    },
    {
      "epoch": 0.04189814814814815,
      "grad_norm": 6.4704766273498535,
      "learning_rate": 1.916203703703704e-05,
      "loss": 1.5059,
      "step": 362
    },
    {
      "epoch": 0.04201388888888889,
      "grad_norm": 11.4617338180542,
      "learning_rate": 1.9159722222222223e-05,
      "loss": 1.3143,
      "step": 363
    },
    {
      "epoch": 0.04212962962962963,
      "grad_norm": 6.011593341827393,
      "learning_rate": 1.9157407407407408e-05,
      "loss": 0.4466,
      "step": 364
    },
    {
      "epoch": 0.04224537037037037,
      "grad_norm": 4.442085266113281,
      "learning_rate": 1.9155092592592595e-05,
      "loss": 0.6728,
      "step": 365
    },
    {
      "epoch": 0.04236111111111111,
      "grad_norm": 5.328735828399658,
      "learning_rate": 1.915277777777778e-05,
      "loss": 0.9191,
      "step": 366
    },
    {
      "epoch": 0.04247685185185185,
      "grad_norm": 5.239377498626709,
      "learning_rate": 1.9150462962962964e-05,
      "loss": 1.0098,
      "step": 367
    },
    {
      "epoch": 0.04259259259259259,
      "grad_norm": 18.42984962463379,
      "learning_rate": 1.9148148148148148e-05,
      "loss": 1.2879,
      "step": 368
    },
    {
      "epoch": 0.042708333333333334,
      "grad_norm": 13.740267753601074,
      "learning_rate": 1.9145833333333336e-05,
      "loss": 1.1275,
      "step": 369
    },
    {
      "epoch": 0.04282407407407408,
      "grad_norm": 24.619680404663086,
      "learning_rate": 1.914351851851852e-05,
      "loss": 1.4275,
      "step": 370
    },
    {
      "epoch": 0.04293981481481481,
      "grad_norm": 12.735710144042969,
      "learning_rate": 1.9141203703703704e-05,
      "loss": 1.1716,
      "step": 371
    },
    {
      "epoch": 0.043055555555555555,
      "grad_norm": 25.68661117553711,
      "learning_rate": 1.9138888888888892e-05,
      "loss": 1.9037,
      "step": 372
    },
    {
      "epoch": 0.0431712962962963,
      "grad_norm": 2.7418947219848633,
      "learning_rate": 1.9136574074074076e-05,
      "loss": 0.6413,
      "step": 373
    },
    {
      "epoch": 0.043287037037037034,
      "grad_norm": 7.134884357452393,
      "learning_rate": 1.913425925925926e-05,
      "loss": 0.7947,
      "step": 374
    },
    {
      "epoch": 0.043402777777777776,
      "grad_norm": 11.327018737792969,
      "learning_rate": 1.9131944444444445e-05,
      "loss": 1.094,
      "step": 375
    },
    {
      "epoch": 0.04351851851851852,
      "grad_norm": 7.013001441955566,
      "learning_rate": 1.9129629629629632e-05,
      "loss": 0.9158,
      "step": 376
    },
    {
      "epoch": 0.04363425925925926,
      "grad_norm": 2.46071195602417,
      "learning_rate": 1.9127314814814816e-05,
      "loss": 0.4708,
      "step": 377
    },
    {
      "epoch": 0.04375,
      "grad_norm": 6.175307273864746,
      "learning_rate": 1.9125000000000004e-05,
      "loss": 0.729,
      "step": 378
    },
    {
      "epoch": 0.04386574074074074,
      "grad_norm": 5.727567672729492,
      "learning_rate": 1.9122685185185185e-05,
      "loss": 0.4694,
      "step": 379
    },
    {
      "epoch": 0.04398148148148148,
      "grad_norm": 3.6789207458496094,
      "learning_rate": 1.9120370370370373e-05,
      "loss": 0.7661,
      "step": 380
    },
    {
      "epoch": 0.044097222222222225,
      "grad_norm": 7.4703369140625,
      "learning_rate": 1.9118055555555557e-05,
      "loss": 0.8269,
      "step": 381
    },
    {
      "epoch": 0.04421296296296296,
      "grad_norm": 3.8127975463867188,
      "learning_rate": 1.911574074074074e-05,
      "loss": 0.9001,
      "step": 382
    },
    {
      "epoch": 0.044328703703703703,
      "grad_norm": 7.869121074676514,
      "learning_rate": 1.911342592592593e-05,
      "loss": 0.8596,
      "step": 383
    },
    {
      "epoch": 0.044444444444444446,
      "grad_norm": 11.47541332244873,
      "learning_rate": 1.9111111111111113e-05,
      "loss": 2.4993,
      "step": 384
    },
    {
      "epoch": 0.04456018518518518,
      "grad_norm": 5.921472549438477,
      "learning_rate": 1.9108796296296297e-05,
      "loss": 1.0205,
      "step": 385
    },
    {
      "epoch": 0.044675925925925924,
      "grad_norm": 8.509511947631836,
      "learning_rate": 1.910648148148148e-05,
      "loss": 1.1399,
      "step": 386
    },
    {
      "epoch": 0.04479166666666667,
      "grad_norm": 8.219855308532715,
      "learning_rate": 1.910416666666667e-05,
      "loss": 0.7365,
      "step": 387
    },
    {
      "epoch": 0.04490740740740741,
      "grad_norm": 3.1264212131500244,
      "learning_rate": 1.9101851851851853e-05,
      "loss": 0.5239,
      "step": 388
    },
    {
      "epoch": 0.045023148148148145,
      "grad_norm": 2.413195848464966,
      "learning_rate": 1.9099537037037038e-05,
      "loss": 0.3438,
      "step": 389
    },
    {
      "epoch": 0.04513888888888889,
      "grad_norm": 12.205561637878418,
      "learning_rate": 1.9097222222222225e-05,
      "loss": 0.8278,
      "step": 390
    },
    {
      "epoch": 0.04525462962962963,
      "grad_norm": 2.1560986042022705,
      "learning_rate": 1.909490740740741e-05,
      "loss": 0.5178,
      "step": 391
    },
    {
      "epoch": 0.04537037037037037,
      "grad_norm": 5.0577239990234375,
      "learning_rate": 1.9092592592592594e-05,
      "loss": 0.8008,
      "step": 392
    },
    {
      "epoch": 0.04548611111111111,
      "grad_norm": 8.096851348876953,
      "learning_rate": 1.9090277777777778e-05,
      "loss": 0.8098,
      "step": 393
    },
    {
      "epoch": 0.04560185185185185,
      "grad_norm": 13.83551025390625,
      "learning_rate": 1.9087962962962966e-05,
      "loss": 1.0069,
      "step": 394
    },
    {
      "epoch": 0.045717592592592594,
      "grad_norm": 4.150102615356445,
      "learning_rate": 1.908564814814815e-05,
      "loss": 0.5822,
      "step": 395
    },
    {
      "epoch": 0.04583333333333333,
      "grad_norm": 14.241860389709473,
      "learning_rate": 1.9083333333333338e-05,
      "loss": 1.3809,
      "step": 396
    },
    {
      "epoch": 0.04594907407407407,
      "grad_norm": 3.8016982078552246,
      "learning_rate": 1.908101851851852e-05,
      "loss": 0.6821,
      "step": 397
    },
    {
      "epoch": 0.046064814814814815,
      "grad_norm": 9.189355850219727,
      "learning_rate": 1.9078703703703706e-05,
      "loss": 0.6045,
      "step": 398
    },
    {
      "epoch": 0.04618055555555556,
      "grad_norm": 2.708941698074341,
      "learning_rate": 1.907638888888889e-05,
      "loss": 0.5436,
      "step": 399
    },
    {
      "epoch": 0.046296296296296294,
      "grad_norm": 5.903145790100098,
      "learning_rate": 1.9074074074074075e-05,
      "loss": 0.5696,
      "step": 400
    },
    {
      "epoch": 0.046412037037037036,
      "grad_norm": 2.278052806854248,
      "learning_rate": 1.9071759259259262e-05,
      "loss": 0.4412,
      "step": 401
    },
    {
      "epoch": 0.04652777777777778,
      "grad_norm": 2.671617269515991,
      "learning_rate": 1.9069444444444446e-05,
      "loss": 0.5411,
      "step": 402
    },
    {
      "epoch": 0.04664351851851852,
      "grad_norm": 14.654735565185547,
      "learning_rate": 1.906712962962963e-05,
      "loss": 1.313,
      "step": 403
    },
    {
      "epoch": 0.04675925925925926,
      "grad_norm": 22.292829513549805,
      "learning_rate": 1.9064814814814815e-05,
      "loss": 1.0345,
      "step": 404
    },
    {
      "epoch": 0.046875,
      "grad_norm": 3.6748225688934326,
      "learning_rate": 1.9062500000000003e-05,
      "loss": 0.604,
      "step": 405
    },
    {
      "epoch": 0.04699074074074074,
      "grad_norm": 6.079431056976318,
      "learning_rate": 1.9060185185185187e-05,
      "loss": 1.1075,
      "step": 406
    },
    {
      "epoch": 0.04710648148148148,
      "grad_norm": 3.2107656002044678,
      "learning_rate": 1.905787037037037e-05,
      "loss": 0.5804,
      "step": 407
    },
    {
      "epoch": 0.04722222222222222,
      "grad_norm": 8.38014030456543,
      "learning_rate": 1.9055555555555555e-05,
      "loss": 1.1454,
      "step": 408
    },
    {
      "epoch": 0.047337962962962964,
      "grad_norm": 5.288601875305176,
      "learning_rate": 1.9053240740740743e-05,
      "loss": 1.4065,
      "step": 409
    },
    {
      "epoch": 0.047453703703703706,
      "grad_norm": 10.177571296691895,
      "learning_rate": 1.9050925925925927e-05,
      "loss": 0.9182,
      "step": 410
    },
    {
      "epoch": 0.04756944444444444,
      "grad_norm": 10.485347747802734,
      "learning_rate": 1.904861111111111e-05,
      "loss": 1.5781,
      "step": 411
    },
    {
      "epoch": 0.047685185185185185,
      "grad_norm": 2.8951354026794434,
      "learning_rate": 1.90462962962963e-05,
      "loss": 0.5836,
      "step": 412
    },
    {
      "epoch": 0.04780092592592593,
      "grad_norm": 11.012734413146973,
      "learning_rate": 1.9043981481481483e-05,
      "loss": 1.6963,
      "step": 413
    },
    {
      "epoch": 0.04791666666666667,
      "grad_norm": 8.753130912780762,
      "learning_rate": 1.9041666666666668e-05,
      "loss": 1.5091,
      "step": 414
    },
    {
      "epoch": 0.048032407407407406,
      "grad_norm": 11.971151351928711,
      "learning_rate": 1.9039351851851852e-05,
      "loss": 1.3409,
      "step": 415
    },
    {
      "epoch": 0.04814814814814815,
      "grad_norm": 19.40638542175293,
      "learning_rate": 1.903703703703704e-05,
      "loss": 0.8839,
      "step": 416
    },
    {
      "epoch": 0.04826388888888889,
      "grad_norm": 4.912037372589111,
      "learning_rate": 1.9034722222222224e-05,
      "loss": 0.6714,
      "step": 417
    },
    {
      "epoch": 0.04837962962962963,
      "grad_norm": 7.023585319519043,
      "learning_rate": 1.9032407407407408e-05,
      "loss": 0.9907,
      "step": 418
    },
    {
      "epoch": 0.04849537037037037,
      "grad_norm": 9.782857894897461,
      "learning_rate": 1.9030092592592596e-05,
      "loss": 0.7682,
      "step": 419
    },
    {
      "epoch": 0.04861111111111111,
      "grad_norm": 3.040161609649658,
      "learning_rate": 1.902777777777778e-05,
      "loss": 0.5796,
      "step": 420
    },
    {
      "epoch": 0.048726851851851855,
      "grad_norm": 10.07345962524414,
      "learning_rate": 1.9025462962962964e-05,
      "loss": 1.7308,
      "step": 421
    },
    {
      "epoch": 0.04884259259259259,
      "grad_norm": 1.778184413909912,
      "learning_rate": 1.902314814814815e-05,
      "loss": 0.3416,
      "step": 422
    },
    {
      "epoch": 0.04895833333333333,
      "grad_norm": 2.635745048522949,
      "learning_rate": 1.9020833333333336e-05,
      "loss": 0.3642,
      "step": 423
    },
    {
      "epoch": 0.049074074074074076,
      "grad_norm": 2.2524967193603516,
      "learning_rate": 1.901851851851852e-05,
      "loss": 0.4025,
      "step": 424
    },
    {
      "epoch": 0.04918981481481482,
      "grad_norm": 7.712453842163086,
      "learning_rate": 1.9016203703703705e-05,
      "loss": 1.2812,
      "step": 425
    },
    {
      "epoch": 0.049305555555555554,
      "grad_norm": 10.390440940856934,
      "learning_rate": 1.901388888888889e-05,
      "loss": 1.5358,
      "step": 426
    },
    {
      "epoch": 0.049421296296296297,
      "grad_norm": 31.85899543762207,
      "learning_rate": 1.9011574074074076e-05,
      "loss": 1.3092,
      "step": 427
    },
    {
      "epoch": 0.04953703703703704,
      "grad_norm": 13.690132141113281,
      "learning_rate": 1.900925925925926e-05,
      "loss": 1.0456,
      "step": 428
    },
    {
      "epoch": 0.049652777777777775,
      "grad_norm": 24.304391860961914,
      "learning_rate": 1.9006944444444445e-05,
      "loss": 0.9832,
      "step": 429
    },
    {
      "epoch": 0.04976851851851852,
      "grad_norm": 16.698556900024414,
      "learning_rate": 1.9004629629629633e-05,
      "loss": 1.2391,
      "step": 430
    },
    {
      "epoch": 0.04988425925925926,
      "grad_norm": 10.379993438720703,
      "learning_rate": 1.9002314814814817e-05,
      "loss": 1.0863,
      "step": 431
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.817371129989624,
      "learning_rate": 1.9e-05,
      "loss": 0.6397,
      "step": 432
    },
    {
      "epoch": 0.05011574074074074,
      "grad_norm": 15.710813522338867,
      "learning_rate": 1.8997685185185185e-05,
      "loss": 1.6451,
      "step": 433
    },
    {
      "epoch": 0.05023148148148148,
      "grad_norm": 2.11043643951416,
      "learning_rate": 1.8995370370370373e-05,
      "loss": 0.4011,
      "step": 434
    },
    {
      "epoch": 0.050347222222222224,
      "grad_norm": 4.0362420082092285,
      "learning_rate": 1.8993055555555557e-05,
      "loss": 0.8486,
      "step": 435
    },
    {
      "epoch": 0.050462962962962966,
      "grad_norm": 2.7132463455200195,
      "learning_rate": 1.899074074074074e-05,
      "loss": 0.4716,
      "step": 436
    },
    {
      "epoch": 0.0505787037037037,
      "grad_norm": 1.816225290298462,
      "learning_rate": 1.898842592592593e-05,
      "loss": 0.3944,
      "step": 437
    },
    {
      "epoch": 0.050694444444444445,
      "grad_norm": 4.904991626739502,
      "learning_rate": 1.8986111111111113e-05,
      "loss": 0.6799,
      "step": 438
    },
    {
      "epoch": 0.05081018518518519,
      "grad_norm": 6.161055564880371,
      "learning_rate": 1.8983796296296298e-05,
      "loss": 1.4771,
      "step": 439
    },
    {
      "epoch": 0.05092592592592592,
      "grad_norm": 6.8464531898498535,
      "learning_rate": 1.8981481481481482e-05,
      "loss": 0.9684,
      "step": 440
    },
    {
      "epoch": 0.051041666666666666,
      "grad_norm": 10.273117065429688,
      "learning_rate": 1.897916666666667e-05,
      "loss": 1.3223,
      "step": 441
    },
    {
      "epoch": 0.05115740740740741,
      "grad_norm": 2.4379405975341797,
      "learning_rate": 1.8976851851851854e-05,
      "loss": 0.5339,
      "step": 442
    },
    {
      "epoch": 0.05127314814814815,
      "grad_norm": 12.050874710083008,
      "learning_rate": 1.8974537037037038e-05,
      "loss": 1.1819,
      "step": 443
    },
    {
      "epoch": 0.05138888888888889,
      "grad_norm": 2.037050485610962,
      "learning_rate": 1.8972222222222222e-05,
      "loss": 0.4343,
      "step": 444
    },
    {
      "epoch": 0.05150462962962963,
      "grad_norm": 9.743439674377441,
      "learning_rate": 1.896990740740741e-05,
      "loss": 0.7173,
      "step": 445
    },
    {
      "epoch": 0.05162037037037037,
      "grad_norm": 2.3540079593658447,
      "learning_rate": 1.8967592592592594e-05,
      "loss": 0.4613,
      "step": 446
    },
    {
      "epoch": 0.05173611111111111,
      "grad_norm": 19.440902709960938,
      "learning_rate": 1.896527777777778e-05,
      "loss": 1.0609,
      "step": 447
    },
    {
      "epoch": 0.05185185185185185,
      "grad_norm": 4.343188762664795,
      "learning_rate": 1.8962962962962966e-05,
      "loss": 0.6594,
      "step": 448
    },
    {
      "epoch": 0.05196759259259259,
      "grad_norm": 3.2369065284729004,
      "learning_rate": 1.896064814814815e-05,
      "loss": 0.5965,
      "step": 449
    },
    {
      "epoch": 0.052083333333333336,
      "grad_norm": 3.5174801349639893,
      "learning_rate": 1.8958333333333334e-05,
      "loss": 0.5103,
      "step": 450
    },
    {
      "epoch": 0.05219907407407407,
      "grad_norm": 4.331578731536865,
      "learning_rate": 1.895601851851852e-05,
      "loss": 0.5746,
      "step": 451
    },
    {
      "epoch": 0.052314814814814814,
      "grad_norm": 4.195745944976807,
      "learning_rate": 1.8953703703703706e-05,
      "loss": 0.7854,
      "step": 452
    },
    {
      "epoch": 0.05243055555555556,
      "grad_norm": 10.8049955368042,
      "learning_rate": 1.895138888888889e-05,
      "loss": 1.4787,
      "step": 453
    },
    {
      "epoch": 0.0525462962962963,
      "grad_norm": 5.1030683517456055,
      "learning_rate": 1.8949074074074075e-05,
      "loss": 0.7543,
      "step": 454
    },
    {
      "epoch": 0.052662037037037035,
      "grad_norm": 7.142575740814209,
      "learning_rate": 1.894675925925926e-05,
      "loss": 0.8215,
      "step": 455
    },
    {
      "epoch": 0.05277777777777778,
      "grad_norm": 3.410062313079834,
      "learning_rate": 1.8944444444444447e-05,
      "loss": 0.5745,
      "step": 456
    },
    {
      "epoch": 0.05289351851851852,
      "grad_norm": 5.064573287963867,
      "learning_rate": 1.894212962962963e-05,
      "loss": 0.797,
      "step": 457
    },
    {
      "epoch": 0.053009259259259256,
      "grad_norm": 4.0004448890686035,
      "learning_rate": 1.8939814814814815e-05,
      "loss": 0.8638,
      "step": 458
    },
    {
      "epoch": 0.053125,
      "grad_norm": 15.59340763092041,
      "learning_rate": 1.8937500000000003e-05,
      "loss": 1.529,
      "step": 459
    },
    {
      "epoch": 0.05324074074074074,
      "grad_norm": 11.306540489196777,
      "learning_rate": 1.8935185185185187e-05,
      "loss": 1.3314,
      "step": 460
    },
    {
      "epoch": 0.053356481481481484,
      "grad_norm": 3.827939987182617,
      "learning_rate": 1.893287037037037e-05,
      "loss": 0.4695,
      "step": 461
    },
    {
      "epoch": 0.05347222222222222,
      "grad_norm": 9.4883451461792,
      "learning_rate": 1.8930555555555556e-05,
      "loss": 0.8897,
      "step": 462
    },
    {
      "epoch": 0.05358796296296296,
      "grad_norm": 7.51424503326416,
      "learning_rate": 1.8928240740740743e-05,
      "loss": 0.8803,
      "step": 463
    },
    {
      "epoch": 0.053703703703703705,
      "grad_norm": 38.15314865112305,
      "learning_rate": 1.8925925925925928e-05,
      "loss": 1.0275,
      "step": 464
    },
    {
      "epoch": 0.05381944444444445,
      "grad_norm": 12.03460693359375,
      "learning_rate": 1.8923611111111112e-05,
      "loss": 1.762,
      "step": 465
    },
    {
      "epoch": 0.05393518518518518,
      "grad_norm": 3.2524425983428955,
      "learning_rate": 1.89212962962963e-05,
      "loss": 0.4607,
      "step": 466
    },
    {
      "epoch": 0.054050925925925926,
      "grad_norm": 8.265035629272461,
      "learning_rate": 1.8918981481481484e-05,
      "loss": 0.5477,
      "step": 467
    },
    {
      "epoch": 0.05416666666666667,
      "grad_norm": 5.594792366027832,
      "learning_rate": 1.8916666666666668e-05,
      "loss": 0.6092,
      "step": 468
    },
    {
      "epoch": 0.054282407407407404,
      "grad_norm": 2.000802755355835,
      "learning_rate": 1.8914351851851852e-05,
      "loss": 0.4465,
      "step": 469
    },
    {
      "epoch": 0.05439814814814815,
      "grad_norm": 10.605072021484375,
      "learning_rate": 1.891203703703704e-05,
      "loss": 0.9048,
      "step": 470
    },
    {
      "epoch": 0.05451388888888889,
      "grad_norm": 8.087080955505371,
      "learning_rate": 1.8909722222222224e-05,
      "loss": 0.7144,
      "step": 471
    },
    {
      "epoch": 0.05462962962962963,
      "grad_norm": 1.9988552331924438,
      "learning_rate": 1.890740740740741e-05,
      "loss": 0.3844,
      "step": 472
    },
    {
      "epoch": 0.05474537037037037,
      "grad_norm": 1.9251123666763306,
      "learning_rate": 1.8905092592592593e-05,
      "loss": 0.298,
      "step": 473
    },
    {
      "epoch": 0.05486111111111111,
      "grad_norm": 2.9033918380737305,
      "learning_rate": 1.890277777777778e-05,
      "loss": 0.6876,
      "step": 474
    },
    {
      "epoch": 0.05497685185185185,
      "grad_norm": 11.45750617980957,
      "learning_rate": 1.8900462962962964e-05,
      "loss": 1.0964,
      "step": 475
    },
    {
      "epoch": 0.055092592592592596,
      "grad_norm": 12.445656776428223,
      "learning_rate": 1.889814814814815e-05,
      "loss": 0.8342,
      "step": 476
    },
    {
      "epoch": 0.05520833333333333,
      "grad_norm": 10.822380065917969,
      "learning_rate": 1.8895833333333336e-05,
      "loss": 2.0464,
      "step": 477
    },
    {
      "epoch": 0.055324074074074074,
      "grad_norm": 1.81320321559906,
      "learning_rate": 1.889351851851852e-05,
      "loss": 0.3936,
      "step": 478
    },
    {
      "epoch": 0.05543981481481482,
      "grad_norm": 16.738922119140625,
      "learning_rate": 1.8891203703703705e-05,
      "loss": 1.5579,
      "step": 479
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 1.9658023118972778,
      "learning_rate": 1.888888888888889e-05,
      "loss": 0.4304,
      "step": 480
    },
    {
      "epoch": 0.055671296296296295,
      "grad_norm": 2.176849842071533,
      "learning_rate": 1.8886574074074077e-05,
      "loss": 0.4736,
      "step": 481
    },
    {
      "epoch": 0.05578703703703704,
      "grad_norm": 4.4654340744018555,
      "learning_rate": 1.888425925925926e-05,
      "loss": 0.703,
      "step": 482
    },
    {
      "epoch": 0.05590277777777778,
      "grad_norm": 6.407155990600586,
      "learning_rate": 1.8881944444444445e-05,
      "loss": 0.495,
      "step": 483
    },
    {
      "epoch": 0.056018518518518516,
      "grad_norm": 4.380337715148926,
      "learning_rate": 1.8879629629629633e-05,
      "loss": 0.7543,
      "step": 484
    },
    {
      "epoch": 0.05613425925925926,
      "grad_norm": 8.650565147399902,
      "learning_rate": 1.8877314814814817e-05,
      "loss": 0.9067,
      "step": 485
    },
    {
      "epoch": 0.05625,
      "grad_norm": 3.177217483520508,
      "learning_rate": 1.8875e-05,
      "loss": 0.5227,
      "step": 486
    },
    {
      "epoch": 0.056365740740740744,
      "grad_norm": 5.844763278961182,
      "learning_rate": 1.8872685185185186e-05,
      "loss": 0.634,
      "step": 487
    },
    {
      "epoch": 0.05648148148148148,
      "grad_norm": 6.802562236785889,
      "learning_rate": 1.8870370370370373e-05,
      "loss": 0.6121,
      "step": 488
    },
    {
      "epoch": 0.05659722222222222,
      "grad_norm": 3.004882574081421,
      "learning_rate": 1.8868055555555558e-05,
      "loss": 0.5852,
      "step": 489
    },
    {
      "epoch": 0.056712962962962965,
      "grad_norm": 11.463611602783203,
      "learning_rate": 1.8865740740740742e-05,
      "loss": 1.4165,
      "step": 490
    },
    {
      "epoch": 0.0568287037037037,
      "grad_norm": 14.539567947387695,
      "learning_rate": 1.8863425925925926e-05,
      "loss": 2.0958,
      "step": 491
    },
    {
      "epoch": 0.05694444444444444,
      "grad_norm": 2.0142197608947754,
      "learning_rate": 1.8861111111111114e-05,
      "loss": 0.3023,
      "step": 492
    },
    {
      "epoch": 0.057060185185185186,
      "grad_norm": 4.233314037322998,
      "learning_rate": 1.8858796296296298e-05,
      "loss": 0.7922,
      "step": 493
    },
    {
      "epoch": 0.05717592592592593,
      "grad_norm": 2.718778133392334,
      "learning_rate": 1.8856481481481482e-05,
      "loss": 0.4009,
      "step": 494
    },
    {
      "epoch": 0.057291666666666664,
      "grad_norm": 2.7267649173736572,
      "learning_rate": 1.885416666666667e-05,
      "loss": 0.5333,
      "step": 495
    },
    {
      "epoch": 0.05740740740740741,
      "grad_norm": 6.420015335083008,
      "learning_rate": 1.885185185185185e-05,
      "loss": 0.8667,
      "step": 496
    },
    {
      "epoch": 0.05752314814814815,
      "grad_norm": 23.943906784057617,
      "learning_rate": 1.884953703703704e-05,
      "loss": 0.8285,
      "step": 497
    },
    {
      "epoch": 0.05763888888888889,
      "grad_norm": 1.9648361206054688,
      "learning_rate": 1.8847222222222223e-05,
      "loss": 0.4227,
      "step": 498
    },
    {
      "epoch": 0.05775462962962963,
      "grad_norm": 2.9537811279296875,
      "learning_rate": 1.884490740740741e-05,
      "loss": 0.4448,
      "step": 499
    },
    {
      "epoch": 0.05787037037037037,
      "grad_norm": 4.585196495056152,
      "learning_rate": 1.8842592592592594e-05,
      "loss": 0.4971,
      "step": 500
    },
    {
      "epoch": 0.05798611111111111,
      "grad_norm": 7.480199337005615,
      "learning_rate": 1.884027777777778e-05,
      "loss": 1.352,
      "step": 501
    },
    {
      "epoch": 0.05810185185185185,
      "grad_norm": 3.8574037551879883,
      "learning_rate": 1.8837962962962963e-05,
      "loss": 0.7881,
      "step": 502
    },
    {
      "epoch": 0.05821759259259259,
      "grad_norm": 15.990200996398926,
      "learning_rate": 1.883564814814815e-05,
      "loss": 1.1476,
      "step": 503
    },
    {
      "epoch": 0.058333333333333334,
      "grad_norm": 1.7176377773284912,
      "learning_rate": 1.8833333333333335e-05,
      "loss": 0.3368,
      "step": 504
    },
    {
      "epoch": 0.05844907407407408,
      "grad_norm": 14.855551719665527,
      "learning_rate": 1.883101851851852e-05,
      "loss": 1.2218,
      "step": 505
    },
    {
      "epoch": 0.05856481481481481,
      "grad_norm": 2.286942720413208,
      "learning_rate": 1.8828703703703707e-05,
      "loss": 0.3767,
      "step": 506
    },
    {
      "epoch": 0.058680555555555555,
      "grad_norm": 4.476166248321533,
      "learning_rate": 1.882638888888889e-05,
      "loss": 0.7699,
      "step": 507
    },
    {
      "epoch": 0.0587962962962963,
      "grad_norm": 2.1812632083892822,
      "learning_rate": 1.8824074074074075e-05,
      "loss": 0.4752,
      "step": 508
    },
    {
      "epoch": 0.058912037037037034,
      "grad_norm": 3.5457725524902344,
      "learning_rate": 1.882175925925926e-05,
      "loss": 0.6855,
      "step": 509
    },
    {
      "epoch": 0.059027777777777776,
      "grad_norm": 5.701589107513428,
      "learning_rate": 1.8819444444444447e-05,
      "loss": 0.687,
      "step": 510
    },
    {
      "epoch": 0.05914351851851852,
      "grad_norm": 3.990018367767334,
      "learning_rate": 1.881712962962963e-05,
      "loss": 0.4726,
      "step": 511
    },
    {
      "epoch": 0.05925925925925926,
      "grad_norm": 4.628663539886475,
      "learning_rate": 1.8814814814814816e-05,
      "loss": 0.4994,
      "step": 512
    },
    {
      "epoch": 0.059375,
      "grad_norm": 2.7996456623077393,
      "learning_rate": 1.8812500000000003e-05,
      "loss": 0.3625,
      "step": 513
    },
    {
      "epoch": 0.05949074074074074,
      "grad_norm": 12.76262092590332,
      "learning_rate": 1.8810185185185184e-05,
      "loss": 0.8555,
      "step": 514
    },
    {
      "epoch": 0.05960648148148148,
      "grad_norm": 2.5238916873931885,
      "learning_rate": 1.8807870370370372e-05,
      "loss": 0.4708,
      "step": 515
    },
    {
      "epoch": 0.059722222222222225,
      "grad_norm": 10.71958065032959,
      "learning_rate": 1.8805555555555556e-05,
      "loss": 1.4456,
      "step": 516
    },
    {
      "epoch": 0.05983796296296296,
      "grad_norm": 4.174191474914551,
      "learning_rate": 1.8803240740740744e-05,
      "loss": 0.5216,
      "step": 517
    },
    {
      "epoch": 0.059953703703703703,
      "grad_norm": 2.992126226425171,
      "learning_rate": 1.8800925925925928e-05,
      "loss": 0.4376,
      "step": 518
    },
    {
      "epoch": 0.060069444444444446,
      "grad_norm": 5.000021934509277,
      "learning_rate": 1.8798611111111112e-05,
      "loss": 0.5465,
      "step": 519
    },
    {
      "epoch": 0.06018518518518518,
      "grad_norm": 8.156257629394531,
      "learning_rate": 1.8796296296296296e-05,
      "loss": 0.7461,
      "step": 520
    },
    {
      "epoch": 0.060300925925925924,
      "grad_norm": 2.7388558387756348,
      "learning_rate": 1.8793981481481484e-05,
      "loss": 0.6392,
      "step": 521
    },
    {
      "epoch": 0.06041666666666667,
      "grad_norm": 1.7787284851074219,
      "learning_rate": 1.8791666666666668e-05,
      "loss": 0.334,
      "step": 522
    },
    {
      "epoch": 0.06053240740740741,
      "grad_norm": 5.661805629730225,
      "learning_rate": 1.8789351851851853e-05,
      "loss": 0.673,
      "step": 523
    },
    {
      "epoch": 0.060648148148148145,
      "grad_norm": 5.454250812530518,
      "learning_rate": 1.878703703703704e-05,
      "loss": 0.6734,
      "step": 524
    },
    {
      "epoch": 0.06076388888888889,
      "grad_norm": 2.109431028366089,
      "learning_rate": 1.8784722222222224e-05,
      "loss": 0.4718,
      "step": 525
    },
    {
      "epoch": 0.06087962962962963,
      "grad_norm": 1.8317538499832153,
      "learning_rate": 1.878240740740741e-05,
      "loss": 0.3326,
      "step": 526
    },
    {
      "epoch": 0.06099537037037037,
      "grad_norm": 11.806161880493164,
      "learning_rate": 1.8780092592592593e-05,
      "loss": 0.866,
      "step": 527
    },
    {
      "epoch": 0.06111111111111111,
      "grad_norm": 4.236999034881592,
      "learning_rate": 1.877777777777778e-05,
      "loss": 0.5844,
      "step": 528
    },
    {
      "epoch": 0.06122685185185185,
      "grad_norm": 2.9030978679656982,
      "learning_rate": 1.8775462962962965e-05,
      "loss": 0.5278,
      "step": 529
    },
    {
      "epoch": 0.061342592592592594,
      "grad_norm": 2.561445951461792,
      "learning_rate": 1.877314814814815e-05,
      "loss": 0.3466,
      "step": 530
    },
    {
      "epoch": 0.06145833333333333,
      "grad_norm": 4.798981189727783,
      "learning_rate": 1.8770833333333337e-05,
      "loss": 0.3641,
      "step": 531
    },
    {
      "epoch": 0.06157407407407407,
      "grad_norm": 3.0999324321746826,
      "learning_rate": 1.8768518518518518e-05,
      "loss": 0.508,
      "step": 532
    },
    {
      "epoch": 0.061689814814814815,
      "grad_norm": 13.699227333068848,
      "learning_rate": 1.8766203703703705e-05,
      "loss": 0.79,
      "step": 533
    },
    {
      "epoch": 0.06180555555555556,
      "grad_norm": 2.1836295127868652,
      "learning_rate": 1.876388888888889e-05,
      "loss": 0.3565,
      "step": 534
    },
    {
      "epoch": 0.061921296296296294,
      "grad_norm": 1.74143385887146,
      "learning_rate": 1.8761574074074077e-05,
      "loss": 0.3791,
      "step": 535
    },
    {
      "epoch": 0.062037037037037036,
      "grad_norm": 7.160713195800781,
      "learning_rate": 1.875925925925926e-05,
      "loss": 0.9062,
      "step": 536
    },
    {
      "epoch": 0.06215277777777778,
      "grad_norm": 5.9215779304504395,
      "learning_rate": 1.8756944444444446e-05,
      "loss": 0.4836,
      "step": 537
    },
    {
      "epoch": 0.06226851851851852,
      "grad_norm": 18.046831130981445,
      "learning_rate": 1.875462962962963e-05,
      "loss": 0.7003,
      "step": 538
    },
    {
      "epoch": 0.06238425925925926,
      "grad_norm": 8.775941848754883,
      "learning_rate": 1.8752314814814817e-05,
      "loss": 0.8719,
      "step": 539
    },
    {
      "epoch": 0.0625,
      "grad_norm": 4.287759304046631,
      "learning_rate": 1.8750000000000002e-05,
      "loss": 0.6885,
      "step": 540
    },
    {
      "epoch": 0.06261574074074074,
      "grad_norm": 5.9450907707214355,
      "learning_rate": 1.8747685185185186e-05,
      "loss": 0.4202,
      "step": 541
    },
    {
      "epoch": 0.06273148148148149,
      "grad_norm": 17.37308692932129,
      "learning_rate": 1.8745370370370374e-05,
      "loss": 0.8641,
      "step": 542
    },
    {
      "epoch": 0.06284722222222222,
      "grad_norm": 1.526052474975586,
      "learning_rate": 1.8743055555555554e-05,
      "loss": 0.3086,
      "step": 543
    },
    {
      "epoch": 0.06296296296296296,
      "grad_norm": 4.774287223815918,
      "learning_rate": 1.8740740740740742e-05,
      "loss": 1.73,
      "step": 544
    },
    {
      "epoch": 0.0630787037037037,
      "grad_norm": 8.744819641113281,
      "learning_rate": 1.8738425925925926e-05,
      "loss": 0.8301,
      "step": 545
    },
    {
      "epoch": 0.06319444444444444,
      "grad_norm": 12.182914733886719,
      "learning_rate": 1.8736111111111114e-05,
      "loss": 1.1403,
      "step": 546
    },
    {
      "epoch": 0.06331018518518519,
      "grad_norm": 1.4857845306396484,
      "learning_rate": 1.8733796296296298e-05,
      "loss": 0.3237,
      "step": 547
    },
    {
      "epoch": 0.06342592592592593,
      "grad_norm": 1.2507820129394531,
      "learning_rate": 1.8731481481481483e-05,
      "loss": 0.2555,
      "step": 548
    },
    {
      "epoch": 0.06354166666666666,
      "grad_norm": 2.69195818901062,
      "learning_rate": 1.8729166666666667e-05,
      "loss": 0.4129,
      "step": 549
    },
    {
      "epoch": 0.06365740740740741,
      "grad_norm": 11.206866264343262,
      "learning_rate": 1.8726851851851854e-05,
      "loss": 1.5692,
      "step": 550
    },
    {
      "epoch": 0.06377314814814815,
      "grad_norm": 2.4055585861206055,
      "learning_rate": 1.872453703703704e-05,
      "loss": 0.4554,
      "step": 551
    },
    {
      "epoch": 0.06388888888888888,
      "grad_norm": 18.56357765197754,
      "learning_rate": 1.8722222222222223e-05,
      "loss": 1.4522,
      "step": 552
    },
    {
      "epoch": 0.06400462962962963,
      "grad_norm": 11.709128379821777,
      "learning_rate": 1.871990740740741e-05,
      "loss": 0.6042,
      "step": 553
    },
    {
      "epoch": 0.06412037037037037,
      "grad_norm": 7.231555938720703,
      "learning_rate": 1.8717592592592595e-05,
      "loss": 1.381,
      "step": 554
    },
    {
      "epoch": 0.0642361111111111,
      "grad_norm": 16.303125381469727,
      "learning_rate": 1.871527777777778e-05,
      "loss": 0.7707,
      "step": 555
    },
    {
      "epoch": 0.06435185185185185,
      "grad_norm": 2.994880199432373,
      "learning_rate": 1.8712962962962963e-05,
      "loss": 0.5011,
      "step": 556
    },
    {
      "epoch": 0.06446759259259259,
      "grad_norm": 6.065532684326172,
      "learning_rate": 1.871064814814815e-05,
      "loss": 0.4085,
      "step": 557
    },
    {
      "epoch": 0.06458333333333334,
      "grad_norm": 7.289284706115723,
      "learning_rate": 1.8708333333333335e-05,
      "loss": 0.7052,
      "step": 558
    },
    {
      "epoch": 0.06469907407407408,
      "grad_norm": 1.5954686403274536,
      "learning_rate": 1.870601851851852e-05,
      "loss": 0.2466,
      "step": 559
    },
    {
      "epoch": 0.06481481481481481,
      "grad_norm": 13.332711219787598,
      "learning_rate": 1.8703703703703707e-05,
      "loss": 1.0346,
      "step": 560
    },
    {
      "epoch": 0.06493055555555556,
      "grad_norm": 5.00217342376709,
      "learning_rate": 1.8701388888888888e-05,
      "loss": 0.5195,
      "step": 561
    },
    {
      "epoch": 0.0650462962962963,
      "grad_norm": 15.709748268127441,
      "learning_rate": 1.8699074074074076e-05,
      "loss": 1.9908,
      "step": 562
    },
    {
      "epoch": 0.06516203703703703,
      "grad_norm": 3.5664589405059814,
      "learning_rate": 1.869675925925926e-05,
      "loss": 0.5261,
      "step": 563
    },
    {
      "epoch": 0.06527777777777778,
      "grad_norm": 1.5178730487823486,
      "learning_rate": 1.8694444444444447e-05,
      "loss": 0.3001,
      "step": 564
    },
    {
      "epoch": 0.06539351851851852,
      "grad_norm": 1.8938477039337158,
      "learning_rate": 1.8692129629629632e-05,
      "loss": 0.3842,
      "step": 565
    },
    {
      "epoch": 0.06550925925925925,
      "grad_norm": 3.2950503826141357,
      "learning_rate": 1.8689814814814816e-05,
      "loss": 0.4359,
      "step": 566
    },
    {
      "epoch": 0.065625,
      "grad_norm": 8.131829261779785,
      "learning_rate": 1.86875e-05,
      "loss": 0.73,
      "step": 567
    },
    {
      "epoch": 0.06574074074074074,
      "grad_norm": 10.283987998962402,
      "learning_rate": 1.8685185185185188e-05,
      "loss": 1.4469,
      "step": 568
    },
    {
      "epoch": 0.06585648148148149,
      "grad_norm": 5.5157790184021,
      "learning_rate": 1.8682870370370372e-05,
      "loss": 0.4139,
      "step": 569
    },
    {
      "epoch": 0.06597222222222222,
      "grad_norm": 3.5031208992004395,
      "learning_rate": 1.8680555555555556e-05,
      "loss": 0.4513,
      "step": 570
    },
    {
      "epoch": 0.06608796296296296,
      "grad_norm": 1.5179442167282104,
      "learning_rate": 1.8678240740740744e-05,
      "loss": 0.2984,
      "step": 571
    },
    {
      "epoch": 0.06620370370370371,
      "grad_norm": 5.48179817199707,
      "learning_rate": 1.8675925925925928e-05,
      "loss": 0.6466,
      "step": 572
    },
    {
      "epoch": 0.06631944444444444,
      "grad_norm": 27.27899169921875,
      "learning_rate": 1.8673611111111113e-05,
      "loss": 0.927,
      "step": 573
    },
    {
      "epoch": 0.06643518518518518,
      "grad_norm": 2.4272208213806152,
      "learning_rate": 1.8671296296296297e-05,
      "loss": 0.3335,
      "step": 574
    },
    {
      "epoch": 0.06655092592592593,
      "grad_norm": 2.6442556381225586,
      "learning_rate": 1.8668981481481484e-05,
      "loss": 0.4243,
      "step": 575
    },
    {
      "epoch": 0.06666666666666667,
      "grad_norm": 3.3386361598968506,
      "learning_rate": 1.866666666666667e-05,
      "loss": 0.295,
      "step": 576
    },
    {
      "epoch": 0.0667824074074074,
      "grad_norm": 3.960758686065674,
      "learning_rate": 1.8664351851851853e-05,
      "loss": 0.6195,
      "step": 577
    },
    {
      "epoch": 0.06689814814814815,
      "grad_norm": 9.209342956542969,
      "learning_rate": 1.866203703703704e-05,
      "loss": 1.6617,
      "step": 578
    },
    {
      "epoch": 0.06701388888888889,
      "grad_norm": 1.3078620433807373,
      "learning_rate": 1.865972222222222e-05,
      "loss": 0.2816,
      "step": 579
    },
    {
      "epoch": 0.06712962962962964,
      "grad_norm": 5.9715657234191895,
      "learning_rate": 1.865740740740741e-05,
      "loss": 1.3033,
      "step": 580
    },
    {
      "epoch": 0.06724537037037037,
      "grad_norm": 21.546104431152344,
      "learning_rate": 1.8655092592592593e-05,
      "loss": 2.1569,
      "step": 581
    },
    {
      "epoch": 0.06736111111111111,
      "grad_norm": 1.258800983428955,
      "learning_rate": 1.865277777777778e-05,
      "loss": 0.2726,
      "step": 582
    },
    {
      "epoch": 0.06747685185185186,
      "grad_norm": 2.6363277435302734,
      "learning_rate": 1.8650462962962965e-05,
      "loss": 0.3397,
      "step": 583
    },
    {
      "epoch": 0.06759259259259259,
      "grad_norm": 5.348413467407227,
      "learning_rate": 1.864814814814815e-05,
      "loss": 0.7639,
      "step": 584
    },
    {
      "epoch": 0.06770833333333333,
      "grad_norm": 2.9576497077941895,
      "learning_rate": 1.8645833333333334e-05,
      "loss": 0.616,
      "step": 585
    },
    {
      "epoch": 0.06782407407407408,
      "grad_norm": 4.387822151184082,
      "learning_rate": 1.864351851851852e-05,
      "loss": 0.3912,
      "step": 586
    },
    {
      "epoch": 0.06793981481481481,
      "grad_norm": 5.619016647338867,
      "learning_rate": 1.8641203703703706e-05,
      "loss": 0.374,
      "step": 587
    },
    {
      "epoch": 0.06805555555555555,
      "grad_norm": 3.388679027557373,
      "learning_rate": 1.863888888888889e-05,
      "loss": 0.4274,
      "step": 588
    },
    {
      "epoch": 0.0681712962962963,
      "grad_norm": 6.377689361572266,
      "learning_rate": 1.8636574074074077e-05,
      "loss": 0.6763,
      "step": 589
    },
    {
      "epoch": 0.06828703703703703,
      "grad_norm": 1.7015999555587769,
      "learning_rate": 1.863425925925926e-05,
      "loss": 0.3727,
      "step": 590
    },
    {
      "epoch": 0.06840277777777778,
      "grad_norm": 14.86113166809082,
      "learning_rate": 1.8631944444444446e-05,
      "loss": 0.8135,
      "step": 591
    },
    {
      "epoch": 0.06851851851851852,
      "grad_norm": 6.10397481918335,
      "learning_rate": 1.862962962962963e-05,
      "loss": 0.4003,
      "step": 592
    },
    {
      "epoch": 0.06863425925925926,
      "grad_norm": 6.430254936218262,
      "learning_rate": 1.8627314814814818e-05,
      "loss": 1.3286,
      "step": 593
    },
    {
      "epoch": 0.06875,
      "grad_norm": 1.984457015991211,
      "learning_rate": 1.8625000000000002e-05,
      "loss": 0.3226,
      "step": 594
    },
    {
      "epoch": 0.06886574074074074,
      "grad_norm": 1.3867501020431519,
      "learning_rate": 1.8622685185185186e-05,
      "loss": 0.3092,
      "step": 595
    },
    {
      "epoch": 0.06898148148148148,
      "grad_norm": 1.7000905275344849,
      "learning_rate": 1.862037037037037e-05,
      "loss": 0.3523,
      "step": 596
    },
    {
      "epoch": 0.06909722222222223,
      "grad_norm": 9.875489234924316,
      "learning_rate": 1.8618055555555555e-05,
      "loss": 0.9178,
      "step": 597
    },
    {
      "epoch": 0.06921296296296296,
      "grad_norm": 1.6489514112472534,
      "learning_rate": 1.8615740740740742e-05,
      "loss": 0.3268,
      "step": 598
    },
    {
      "epoch": 0.0693287037037037,
      "grad_norm": 15.550714492797852,
      "learning_rate": 1.8613425925925927e-05,
      "loss": 0.8545,
      "step": 599
    },
    {
      "epoch": 0.06944444444444445,
      "grad_norm": 2.059080123901367,
      "learning_rate": 1.8611111111111114e-05,
      "loss": 0.2798,
      "step": 600
    },
    {
      "epoch": 0.06956018518518518,
      "grad_norm": 31.277780532836914,
      "learning_rate": 1.86087962962963e-05,
      "loss": 1.5511,
      "step": 601
    },
    {
      "epoch": 0.06967592592592593,
      "grad_norm": 16.351322174072266,
      "learning_rate": 1.8606481481481483e-05,
      "loss": 1.3945,
      "step": 602
    },
    {
      "epoch": 0.06979166666666667,
      "grad_norm": 35.435089111328125,
      "learning_rate": 1.8604166666666667e-05,
      "loss": 0.9356,
      "step": 603
    },
    {
      "epoch": 0.0699074074074074,
      "grad_norm": 3.797102451324463,
      "learning_rate": 1.8601851851851855e-05,
      "loss": 0.4494,
      "step": 604
    },
    {
      "epoch": 0.07002314814814815,
      "grad_norm": 27.412670135498047,
      "learning_rate": 1.859953703703704e-05,
      "loss": 0.7506,
      "step": 605
    },
    {
      "epoch": 0.07013888888888889,
      "grad_norm": 3.287315845489502,
      "learning_rate": 1.8597222222222223e-05,
      "loss": 0.2568,
      "step": 606
    },
    {
      "epoch": 0.07025462962962963,
      "grad_norm": 2.929816722869873,
      "learning_rate": 1.859490740740741e-05,
      "loss": 0.3507,
      "step": 607
    },
    {
      "epoch": 0.07037037037037037,
      "grad_norm": 1.7495595216751099,
      "learning_rate": 1.8592592592592592e-05,
      "loss": 0.2976,
      "step": 608
    },
    {
      "epoch": 0.07048611111111111,
      "grad_norm": 2.270371675491333,
      "learning_rate": 1.859027777777778e-05,
      "loss": 0.3579,
      "step": 609
    },
    {
      "epoch": 0.07060185185185185,
      "grad_norm": 2.8385260105133057,
      "learning_rate": 1.8587962962962964e-05,
      "loss": 0.4361,
      "step": 610
    },
    {
      "epoch": 0.0707175925925926,
      "grad_norm": 6.028038501739502,
      "learning_rate": 1.858564814814815e-05,
      "loss": 0.6065,
      "step": 611
    },
    {
      "epoch": 0.07083333333333333,
      "grad_norm": 9.06475830078125,
      "learning_rate": 1.8583333333333336e-05,
      "loss": 0.9431,
      "step": 612
    },
    {
      "epoch": 0.07094907407407407,
      "grad_norm": 37.88742446899414,
      "learning_rate": 1.858101851851852e-05,
      "loss": 1.1764,
      "step": 613
    },
    {
      "epoch": 0.07106481481481482,
      "grad_norm": 1.496092438697815,
      "learning_rate": 1.8578703703703704e-05,
      "loss": 0.3286,
      "step": 614
    },
    {
      "epoch": 0.07118055555555555,
      "grad_norm": 2.51096248626709,
      "learning_rate": 1.8576388888888888e-05,
      "loss": 0.3621,
      "step": 615
    },
    {
      "epoch": 0.0712962962962963,
      "grad_norm": 1.2315866947174072,
      "learning_rate": 1.8574074074074076e-05,
      "loss": 0.25,
      "step": 616
    },
    {
      "epoch": 0.07141203703703704,
      "grad_norm": 1.2866612672805786,
      "learning_rate": 1.857175925925926e-05,
      "loss": 0.2014,
      "step": 617
    },
    {
      "epoch": 0.07152777777777777,
      "grad_norm": 1.2066372632980347,
      "learning_rate": 1.8569444444444448e-05,
      "loss": 0.265,
      "step": 618
    },
    {
      "epoch": 0.07164351851851852,
      "grad_norm": 1.3436070680618286,
      "learning_rate": 1.8567129629629632e-05,
      "loss": 0.2622,
      "step": 619
    },
    {
      "epoch": 0.07175925925925926,
      "grad_norm": 1.2528188228607178,
      "learning_rate": 1.8564814814814816e-05,
      "loss": 0.273,
      "step": 620
    },
    {
      "epoch": 0.071875,
      "grad_norm": 1.851069688796997,
      "learning_rate": 1.85625e-05,
      "loss": 0.2992,
      "step": 621
    },
    {
      "epoch": 0.07199074074074074,
      "grad_norm": 2.2942047119140625,
      "learning_rate": 1.8560185185185188e-05,
      "loss": 0.3126,
      "step": 622
    },
    {
      "epoch": 0.07210648148148148,
      "grad_norm": 2.2919859886169434,
      "learning_rate": 1.8557870370370372e-05,
      "loss": 0.5233,
      "step": 623
    },
    {
      "epoch": 0.07222222222222222,
      "grad_norm": 4.3965253829956055,
      "learning_rate": 1.8555555555555557e-05,
      "loss": 0.2926,
      "step": 624
    },
    {
      "epoch": 0.07233796296296297,
      "grad_norm": 10.200839042663574,
      "learning_rate": 1.8553240740740744e-05,
      "loss": 1.2818,
      "step": 625
    },
    {
      "epoch": 0.0724537037037037,
      "grad_norm": 1.2092046737670898,
      "learning_rate": 1.8550925925925925e-05,
      "loss": 0.2429,
      "step": 626
    },
    {
      "epoch": 0.07256944444444445,
      "grad_norm": 9.685284614562988,
      "learning_rate": 1.8548611111111113e-05,
      "loss": 1.6035,
      "step": 627
    },
    {
      "epoch": 0.07268518518518519,
      "grad_norm": 6.131631851196289,
      "learning_rate": 1.8546296296296297e-05,
      "loss": 0.6874,
      "step": 628
    },
    {
      "epoch": 0.07280092592592592,
      "grad_norm": 3.470341444015503,
      "learning_rate": 1.8543981481481485e-05,
      "loss": 0.5595,
      "step": 629
    },
    {
      "epoch": 0.07291666666666667,
      "grad_norm": 2.0267233848571777,
      "learning_rate": 1.854166666666667e-05,
      "loss": 0.3286,
      "step": 630
    },
    {
      "epoch": 0.07303240740740741,
      "grad_norm": 61.270381927490234,
      "learning_rate": 1.8539351851851853e-05,
      "loss": 0.8294,
      "step": 631
    },
    {
      "epoch": 0.07314814814814814,
      "grad_norm": 27.42403793334961,
      "learning_rate": 1.8537037037037037e-05,
      "loss": 2.0552,
      "step": 632
    },
    {
      "epoch": 0.07326388888888889,
      "grad_norm": 5.434396266937256,
      "learning_rate": 1.8534722222222222e-05,
      "loss": 0.5577,
      "step": 633
    },
    {
      "epoch": 0.07337962962962963,
      "grad_norm": 26.29444694519043,
      "learning_rate": 1.853240740740741e-05,
      "loss": 1.2305,
      "step": 634
    },
    {
      "epoch": 0.07349537037037036,
      "grad_norm": 8.014799118041992,
      "learning_rate": 1.8530092592592594e-05,
      "loss": 0.4503,
      "step": 635
    },
    {
      "epoch": 0.07361111111111111,
      "grad_norm": 2.8217363357543945,
      "learning_rate": 1.852777777777778e-05,
      "loss": 0.5988,
      "step": 636
    },
    {
      "epoch": 0.07372685185185185,
      "grad_norm": 10.666057586669922,
      "learning_rate": 1.8525462962962962e-05,
      "loss": 1.8056,
      "step": 637
    },
    {
      "epoch": 0.0738425925925926,
      "grad_norm": 7.24723482131958,
      "learning_rate": 1.852314814814815e-05,
      "loss": 0.6344,
      "step": 638
    },
    {
      "epoch": 0.07395833333333333,
      "grad_norm": 44.84671401977539,
      "learning_rate": 1.8520833333333334e-05,
      "loss": 1.4549,
      "step": 639
    },
    {
      "epoch": 0.07407407407407407,
      "grad_norm": 17.052776336669922,
      "learning_rate": 1.851851851851852e-05,
      "loss": 0.8571,
      "step": 640
    },
    {
      "epoch": 0.07418981481481482,
      "grad_norm": 13.255481719970703,
      "learning_rate": 1.8516203703703706e-05,
      "loss": 1.012,
      "step": 641
    },
    {
      "epoch": 0.07430555555555556,
      "grad_norm": 2.365344762802124,
      "learning_rate": 1.851388888888889e-05,
      "loss": 0.5185,
      "step": 642
    },
    {
      "epoch": 0.07442129629629629,
      "grad_norm": 2.070491313934326,
      "learning_rate": 1.8511574074074074e-05,
      "loss": 0.3072,
      "step": 643
    },
    {
      "epoch": 0.07453703703703704,
      "grad_norm": 7.478021621704102,
      "learning_rate": 1.850925925925926e-05,
      "loss": 0.626,
      "step": 644
    },
    {
      "epoch": 0.07465277777777778,
      "grad_norm": 8.277266502380371,
      "learning_rate": 1.8506944444444446e-05,
      "loss": 0.4403,
      "step": 645
    },
    {
      "epoch": 0.07476851851851851,
      "grad_norm": 3.96532940864563,
      "learning_rate": 1.850462962962963e-05,
      "loss": 0.3583,
      "step": 646
    },
    {
      "epoch": 0.07488425925925926,
      "grad_norm": 1.6336168050765991,
      "learning_rate": 1.8502314814814818e-05,
      "loss": 0.2585,
      "step": 647
    },
    {
      "epoch": 0.075,
      "grad_norm": 2.663788318634033,
      "learning_rate": 1.8500000000000002e-05,
      "loss": 0.5268,
      "step": 648
    },
    {
      "epoch": 0.07511574074074075,
      "grad_norm": 17.390899658203125,
      "learning_rate": 1.8497685185185187e-05,
      "loss": 1.5462,
      "step": 649
    },
    {
      "epoch": 0.07523148148148148,
      "grad_norm": 2.9734928607940674,
      "learning_rate": 1.849537037037037e-05,
      "loss": 0.2417,
      "step": 650
    },
    {
      "epoch": 0.07534722222222222,
      "grad_norm": 28.163671493530273,
      "learning_rate": 1.8493055555555555e-05,
      "loss": 2.4481,
      "step": 651
    },
    {
      "epoch": 0.07546296296296297,
      "grad_norm": 4.3419880867004395,
      "learning_rate": 1.8490740740740743e-05,
      "loss": 0.3784,
      "step": 652
    },
    {
      "epoch": 0.0755787037037037,
      "grad_norm": 11.704846382141113,
      "learning_rate": 1.8488425925925927e-05,
      "loss": 0.4764,
      "step": 653
    },
    {
      "epoch": 0.07569444444444444,
      "grad_norm": 2.2620532512664795,
      "learning_rate": 1.8486111111111115e-05,
      "loss": 0.3079,
      "step": 654
    },
    {
      "epoch": 0.07581018518518519,
      "grad_norm": 4.601413249969482,
      "learning_rate": 1.8483796296296296e-05,
      "loss": 0.4673,
      "step": 655
    },
    {
      "epoch": 0.07592592592592592,
      "grad_norm": 11.353368759155273,
      "learning_rate": 1.8481481481481483e-05,
      "loss": 0.8719,
      "step": 656
    },
    {
      "epoch": 0.07604166666666666,
      "grad_norm": 13.61203670501709,
      "learning_rate": 1.8479166666666667e-05,
      "loss": 1.5811,
      "step": 657
    },
    {
      "epoch": 0.07615740740740741,
      "grad_norm": 5.035242557525635,
      "learning_rate": 1.8476851851851855e-05,
      "loss": 0.3023,
      "step": 658
    },
    {
      "epoch": 0.07627314814814815,
      "grad_norm": 32.47747802734375,
      "learning_rate": 1.847453703703704e-05,
      "loss": 0.8965,
      "step": 659
    },
    {
      "epoch": 0.0763888888888889,
      "grad_norm": 12.823569297790527,
      "learning_rate": 1.8472222222222224e-05,
      "loss": 0.6468,
      "step": 660
    },
    {
      "epoch": 0.07650462962962963,
      "grad_norm": 2.1616978645324707,
      "learning_rate": 1.8469907407407408e-05,
      "loss": 0.3612,
      "step": 661
    },
    {
      "epoch": 0.07662037037037037,
      "grad_norm": 2.09191632270813,
      "learning_rate": 1.8467592592592592e-05,
      "loss": 0.4559,
      "step": 662
    },
    {
      "epoch": 0.07673611111111112,
      "grad_norm": 4.153750419616699,
      "learning_rate": 1.846527777777778e-05,
      "loss": 0.5225,
      "step": 663
    },
    {
      "epoch": 0.07685185185185185,
      "grad_norm": 6.390926361083984,
      "learning_rate": 1.8462962962962964e-05,
      "loss": 0.4142,
      "step": 664
    },
    {
      "epoch": 0.07696759259259259,
      "grad_norm": 1.9868282079696655,
      "learning_rate": 1.846064814814815e-05,
      "loss": 0.4391,
      "step": 665
    },
    {
      "epoch": 0.07708333333333334,
      "grad_norm": 1.9750343561172485,
      "learning_rate": 1.8458333333333336e-05,
      "loss": 0.4298,
      "step": 666
    },
    {
      "epoch": 0.07719907407407407,
      "grad_norm": 1.5916273593902588,
      "learning_rate": 1.845601851851852e-05,
      "loss": 0.2828,
      "step": 667
    },
    {
      "epoch": 0.07731481481481481,
      "grad_norm": 1.2976300716400146,
      "learning_rate": 1.8453703703703704e-05,
      "loss": 0.2058,
      "step": 668
    },
    {
      "epoch": 0.07743055555555556,
      "grad_norm": 20.1488094329834,
      "learning_rate": 1.8451388888888892e-05,
      "loss": 0.6928,
      "step": 669
    },
    {
      "epoch": 0.0775462962962963,
      "grad_norm": 16.391130447387695,
      "learning_rate": 1.8449074074074076e-05,
      "loss": 0.7895,
      "step": 670
    },
    {
      "epoch": 0.07766203703703704,
      "grad_norm": 2.0168867111206055,
      "learning_rate": 1.844675925925926e-05,
      "loss": 0.4627,
      "step": 671
    },
    {
      "epoch": 0.07777777777777778,
      "grad_norm": 1.9429014921188354,
      "learning_rate": 1.8444444444444448e-05,
      "loss": 0.2769,
      "step": 672
    },
    {
      "epoch": 0.07789351851851851,
      "grad_norm": 2.571852922439575,
      "learning_rate": 1.844212962962963e-05,
      "loss": 0.5127,
      "step": 673
    },
    {
      "epoch": 0.07800925925925926,
      "grad_norm": 1.5022858381271362,
      "learning_rate": 1.8439814814814817e-05,
      "loss": 0.3098,
      "step": 674
    },
    {
      "epoch": 0.078125,
      "grad_norm": 3.5266175270080566,
      "learning_rate": 1.84375e-05,
      "loss": 0.5382,
      "step": 675
    },
    {
      "epoch": 0.07824074074074074,
      "grad_norm": 9.865453720092773,
      "learning_rate": 1.843518518518519e-05,
      "loss": 1.6628,
      "step": 676
    },
    {
      "epoch": 0.07835648148148149,
      "grad_norm": 4.175038814544678,
      "learning_rate": 1.8432870370370373e-05,
      "loss": 0.4205,
      "step": 677
    },
    {
      "epoch": 0.07847222222222222,
      "grad_norm": 13.575366973876953,
      "learning_rate": 1.8430555555555557e-05,
      "loss": 0.7763,
      "step": 678
    },
    {
      "epoch": 0.07858796296296296,
      "grad_norm": 1.0936906337738037,
      "learning_rate": 1.842824074074074e-05,
      "loss": 0.2277,
      "step": 679
    },
    {
      "epoch": 0.0787037037037037,
      "grad_norm": 5.471321105957031,
      "learning_rate": 1.8425925925925926e-05,
      "loss": 0.5077,
      "step": 680
    },
    {
      "epoch": 0.07881944444444444,
      "grad_norm": 1.6180864572525024,
      "learning_rate": 1.8423611111111113e-05,
      "loss": 0.3223,
      "step": 681
    },
    {
      "epoch": 0.07893518518518519,
      "grad_norm": 1.4357467889785767,
      "learning_rate": 1.8421296296296297e-05,
      "loss": 0.2484,
      "step": 682
    },
    {
      "epoch": 0.07905092592592593,
      "grad_norm": 2.1454832553863525,
      "learning_rate": 1.8418981481481485e-05,
      "loss": 0.3306,
      "step": 683
    },
    {
      "epoch": 0.07916666666666666,
      "grad_norm": 1.3182541131973267,
      "learning_rate": 1.8416666666666666e-05,
      "loss": 0.2434,
      "step": 684
    },
    {
      "epoch": 0.07928240740740741,
      "grad_norm": 7.374088764190674,
      "learning_rate": 1.8414351851851854e-05,
      "loss": 0.4348,
      "step": 685
    },
    {
      "epoch": 0.07939814814814815,
      "grad_norm": 1.5082213878631592,
      "learning_rate": 1.8412037037037038e-05,
      "loss": 0.2339,
      "step": 686
    },
    {
      "epoch": 0.07951388888888888,
      "grad_norm": 2.2904980182647705,
      "learning_rate": 1.8409722222222225e-05,
      "loss": 0.4076,
      "step": 687
    },
    {
      "epoch": 0.07962962962962963,
      "grad_norm": 24.690933227539062,
      "learning_rate": 1.840740740740741e-05,
      "loss": 0.5545,
      "step": 688
    },
    {
      "epoch": 0.07974537037037037,
      "grad_norm": 19.831527709960938,
      "learning_rate": 1.8405092592592594e-05,
      "loss": 0.5106,
      "step": 689
    },
    {
      "epoch": 0.0798611111111111,
      "grad_norm": 1.5778403282165527,
      "learning_rate": 1.8402777777777778e-05,
      "loss": 0.2748,
      "step": 690
    },
    {
      "epoch": 0.07997685185185185,
      "grad_norm": 5.3340864181518555,
      "learning_rate": 1.8400462962962962e-05,
      "loss": 0.4699,
      "step": 691
    },
    {
      "epoch": 0.08009259259259259,
      "grad_norm": 1.5889049768447876,
      "learning_rate": 1.839814814814815e-05,
      "loss": 0.2377,
      "step": 692
    },
    {
      "epoch": 0.08020833333333334,
      "grad_norm": 2.600437641143799,
      "learning_rate": 1.8395833333333334e-05,
      "loss": 0.3995,
      "step": 693
    },
    {
      "epoch": 0.08032407407407408,
      "grad_norm": 4.323887825012207,
      "learning_rate": 1.8393518518518522e-05,
      "loss": 0.606,
      "step": 694
    },
    {
      "epoch": 0.08043981481481481,
      "grad_norm": 11.878968238830566,
      "learning_rate": 1.8391203703703706e-05,
      "loss": 0.6485,
      "step": 695
    },
    {
      "epoch": 0.08055555555555556,
      "grad_norm": 2.465822219848633,
      "learning_rate": 1.838888888888889e-05,
      "loss": 0.3838,
      "step": 696
    },
    {
      "epoch": 0.0806712962962963,
      "grad_norm": 4.453885555267334,
      "learning_rate": 1.8386574074074075e-05,
      "loss": 0.3711,
      "step": 697
    },
    {
      "epoch": 0.08078703703703703,
      "grad_norm": 2.609605073928833,
      "learning_rate": 1.838425925925926e-05,
      "loss": 0.4264,
      "step": 698
    },
    {
      "epoch": 0.08090277777777778,
      "grad_norm": 3.1978752613067627,
      "learning_rate": 1.8381944444444447e-05,
      "loss": 0.4396,
      "step": 699
    },
    {
      "epoch": 0.08101851851851852,
      "grad_norm": 3.5334255695343018,
      "learning_rate": 1.837962962962963e-05,
      "loss": 0.4286,
      "step": 700
    },
    {
      "epoch": 0.08113425925925925,
      "grad_norm": 2.620798349380493,
      "learning_rate": 1.837731481481482e-05,
      "loss": 0.4127,
      "step": 701
    },
    {
      "epoch": 0.08125,
      "grad_norm": 5.162637710571289,
      "learning_rate": 1.8375e-05,
      "loss": 0.51,
      "step": 702
    },
    {
      "epoch": 0.08136574074074074,
      "grad_norm": 1.0061942338943481,
      "learning_rate": 1.8372685185185187e-05,
      "loss": 0.2245,
      "step": 703
    },
    {
      "epoch": 0.08148148148148149,
      "grad_norm": 2.4655420780181885,
      "learning_rate": 1.837037037037037e-05,
      "loss": 0.3601,
      "step": 704
    },
    {
      "epoch": 0.08159722222222222,
      "grad_norm": 15.819686889648438,
      "learning_rate": 1.836805555555556e-05,
      "loss": 0.4746,
      "step": 705
    },
    {
      "epoch": 0.08171296296296296,
      "grad_norm": 1.6183085441589355,
      "learning_rate": 1.8365740740740743e-05,
      "loss": 0.3452,
      "step": 706
    },
    {
      "epoch": 0.08182870370370371,
      "grad_norm": 3.5601086616516113,
      "learning_rate": 1.8363425925925927e-05,
      "loss": 0.5316,
      "step": 707
    },
    {
      "epoch": 0.08194444444444444,
      "grad_norm": 1.7910844087600708,
      "learning_rate": 1.836111111111111e-05,
      "loss": 0.2788,
      "step": 708
    },
    {
      "epoch": 0.08206018518518518,
      "grad_norm": 4.351409912109375,
      "learning_rate": 1.8358796296296296e-05,
      "loss": 0.4596,
      "step": 709
    },
    {
      "epoch": 0.08217592592592593,
      "grad_norm": 1.4447848796844482,
      "learning_rate": 1.8356481481481484e-05,
      "loss": 0.2924,
      "step": 710
    },
    {
      "epoch": 0.08229166666666667,
      "grad_norm": 24.33026695251465,
      "learning_rate": 1.8354166666666668e-05,
      "loss": 1.569,
      "step": 711
    },
    {
      "epoch": 0.0824074074074074,
      "grad_norm": 4.9158525466918945,
      "learning_rate": 1.8351851851851855e-05,
      "loss": 0.4375,
      "step": 712
    },
    {
      "epoch": 0.08252314814814815,
      "grad_norm": 1.2385739088058472,
      "learning_rate": 1.834953703703704e-05,
      "loss": 0.2699,
      "step": 713
    },
    {
      "epoch": 0.08263888888888889,
      "grad_norm": 1.8682540655136108,
      "learning_rate": 1.8347222222222224e-05,
      "loss": 0.3621,
      "step": 714
    },
    {
      "epoch": 0.08275462962962964,
      "grad_norm": 25.11712646484375,
      "learning_rate": 1.8344907407407408e-05,
      "loss": 0.4731,
      "step": 715
    },
    {
      "epoch": 0.08287037037037037,
      "grad_norm": 2.0941128730773926,
      "learning_rate": 1.8342592592592592e-05,
      "loss": 0.4109,
      "step": 716
    },
    {
      "epoch": 0.08298611111111111,
      "grad_norm": 12.057477951049805,
      "learning_rate": 1.834027777777778e-05,
      "loss": 0.588,
      "step": 717
    },
    {
      "epoch": 0.08310185185185186,
      "grad_norm": 1.4242407083511353,
      "learning_rate": 1.8337962962962964e-05,
      "loss": 0.2377,
      "step": 718
    },
    {
      "epoch": 0.08321759259259259,
      "grad_norm": 1.0462830066680908,
      "learning_rate": 1.8335648148148152e-05,
      "loss": 0.2186,
      "step": 719
    },
    {
      "epoch": 0.08333333333333333,
      "grad_norm": 12.097135543823242,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 0.3124,
      "step": 720
    },
    {
      "epoch": 0.08344907407407408,
      "grad_norm": 6.672476291656494,
      "learning_rate": 1.833101851851852e-05,
      "loss": 0.5051,
      "step": 721
    },
    {
      "epoch": 0.08356481481481481,
      "grad_norm": 9.96355152130127,
      "learning_rate": 1.8328703703703705e-05,
      "loss": 0.4916,
      "step": 722
    },
    {
      "epoch": 0.08368055555555555,
      "grad_norm": 6.513317108154297,
      "learning_rate": 1.8326388888888892e-05,
      "loss": 0.2804,
      "step": 723
    },
    {
      "epoch": 0.0837962962962963,
      "grad_norm": 0.8048027753829956,
      "learning_rate": 1.8324074074074077e-05,
      "loss": 0.1733,
      "step": 724
    },
    {
      "epoch": 0.08391203703703703,
      "grad_norm": 1.181504726409912,
      "learning_rate": 1.832175925925926e-05,
      "loss": 0.2377,
      "step": 725
    },
    {
      "epoch": 0.08402777777777778,
      "grad_norm": 0.7961382269859314,
      "learning_rate": 1.8319444444444445e-05,
      "loss": 0.173,
      "step": 726
    },
    {
      "epoch": 0.08414351851851852,
      "grad_norm": 1.070425033569336,
      "learning_rate": 1.831712962962963e-05,
      "loss": 0.1745,
      "step": 727
    },
    {
      "epoch": 0.08425925925925926,
      "grad_norm": 2.070836067199707,
      "learning_rate": 1.8314814814814817e-05,
      "loss": 0.4492,
      "step": 728
    },
    {
      "epoch": 0.084375,
      "grad_norm": 2.868642807006836,
      "learning_rate": 1.83125e-05,
      "loss": 0.4494,
      "step": 729
    },
    {
      "epoch": 0.08449074074074074,
      "grad_norm": 3.3861284255981445,
      "learning_rate": 1.831018518518519e-05,
      "loss": 0.4281,
      "step": 730
    },
    {
      "epoch": 0.08460648148148148,
      "grad_norm": 1.6100826263427734,
      "learning_rate": 1.830787037037037e-05,
      "loss": 0.3126,
      "step": 731
    },
    {
      "epoch": 0.08472222222222223,
      "grad_norm": 0.999402642250061,
      "learning_rate": 1.8305555555555557e-05,
      "loss": 0.1889,
      "step": 732
    },
    {
      "epoch": 0.08483796296296296,
      "grad_norm": 8.633904457092285,
      "learning_rate": 1.830324074074074e-05,
      "loss": 0.5196,
      "step": 733
    },
    {
      "epoch": 0.0849537037037037,
      "grad_norm": 16.25756072998047,
      "learning_rate": 1.8300925925925926e-05,
      "loss": 1.2407,
      "step": 734
    },
    {
      "epoch": 0.08506944444444445,
      "grad_norm": 2.861893653869629,
      "learning_rate": 1.8298611111111114e-05,
      "loss": 0.5001,
      "step": 735
    },
    {
      "epoch": 0.08518518518518518,
      "grad_norm": 1.1495568752288818,
      "learning_rate": 1.8296296296296298e-05,
      "loss": 0.2419,
      "step": 736
    },
    {
      "epoch": 0.08530092592592593,
      "grad_norm": 1.1629843711853027,
      "learning_rate": 1.8293981481481482e-05,
      "loss": 0.2568,
      "step": 737
    },
    {
      "epoch": 0.08541666666666667,
      "grad_norm": 29.08191680908203,
      "learning_rate": 1.8291666666666666e-05,
      "loss": 0.3518,
      "step": 738
    },
    {
      "epoch": 0.0855324074074074,
      "grad_norm": 7.294058799743652,
      "learning_rate": 1.8289351851851854e-05,
      "loss": 0.4464,
      "step": 739
    },
    {
      "epoch": 0.08564814814814815,
      "grad_norm": 2.2014832496643066,
      "learning_rate": 1.8287037037037038e-05,
      "loss": 0.3216,
      "step": 740
    },
    {
      "epoch": 0.08576388888888889,
      "grad_norm": 2.2298731803894043,
      "learning_rate": 1.8284722222222226e-05,
      "loss": 0.4131,
      "step": 741
    },
    {
      "epoch": 0.08587962962962963,
      "grad_norm": 2.2913970947265625,
      "learning_rate": 1.828240740740741e-05,
      "loss": 0.3618,
      "step": 742
    },
    {
      "epoch": 0.08599537037037037,
      "grad_norm": 1.1317766904830933,
      "learning_rate": 1.8280092592592594e-05,
      "loss": 0.2425,
      "step": 743
    },
    {
      "epoch": 0.08611111111111111,
      "grad_norm": 6.8050665855407715,
      "learning_rate": 1.827777777777778e-05,
      "loss": 0.4938,
      "step": 744
    },
    {
      "epoch": 0.08622685185185185,
      "grad_norm": 1.7571754455566406,
      "learning_rate": 1.8275462962962963e-05,
      "loss": 0.2376,
      "step": 745
    },
    {
      "epoch": 0.0863425925925926,
      "grad_norm": 2.691620349884033,
      "learning_rate": 1.827314814814815e-05,
      "loss": 0.3702,
      "step": 746
    },
    {
      "epoch": 0.08645833333333333,
      "grad_norm": 4.0608720779418945,
      "learning_rate": 1.8270833333333335e-05,
      "loss": 0.4083,
      "step": 747
    },
    {
      "epoch": 0.08657407407407407,
      "grad_norm": 2.5903379917144775,
      "learning_rate": 1.8268518518518522e-05,
      "loss": 0.3606,
      "step": 748
    },
    {
      "epoch": 0.08668981481481482,
      "grad_norm": 4.731637477874756,
      "learning_rate": 1.8266203703703703e-05,
      "loss": 0.352,
      "step": 749
    },
    {
      "epoch": 0.08680555555555555,
      "grad_norm": 4.4854631423950195,
      "learning_rate": 1.826388888888889e-05,
      "loss": 0.4219,
      "step": 750
    },
    {
      "epoch": 0.0869212962962963,
      "grad_norm": 4.1604108810424805,
      "learning_rate": 1.8261574074074075e-05,
      "loss": 0.2764,
      "step": 751
    },
    {
      "epoch": 0.08703703703703704,
      "grad_norm": 2.0252974033355713,
      "learning_rate": 1.825925925925926e-05,
      "loss": 0.3174,
      "step": 752
    },
    {
      "epoch": 0.08715277777777777,
      "grad_norm": 1.1585429906845093,
      "learning_rate": 1.8256944444444447e-05,
      "loss": 0.2374,
      "step": 753
    },
    {
      "epoch": 0.08726851851851852,
      "grad_norm": 1.4990876913070679,
      "learning_rate": 1.825462962962963e-05,
      "loss": 0.2334,
      "step": 754
    },
    {
      "epoch": 0.08738425925925926,
      "grad_norm": 1.8626203536987305,
      "learning_rate": 1.8252314814814816e-05,
      "loss": 0.3114,
      "step": 755
    },
    {
      "epoch": 0.0875,
      "grad_norm": 33.857933044433594,
      "learning_rate": 1.825e-05,
      "loss": 1.3215,
      "step": 756
    },
    {
      "epoch": 0.08761574074074074,
      "grad_norm": 0.8504137396812439,
      "learning_rate": 1.8247685185185187e-05,
      "loss": 0.1824,
      "step": 757
    },
    {
      "epoch": 0.08773148148148148,
      "grad_norm": 1.1738368272781372,
      "learning_rate": 1.824537037037037e-05,
      "loss": 0.2388,
      "step": 758
    },
    {
      "epoch": 0.08784722222222222,
      "grad_norm": 14.415878295898438,
      "learning_rate": 1.824305555555556e-05,
      "loss": 0.5587,
      "step": 759
    },
    {
      "epoch": 0.08796296296296297,
      "grad_norm": 1.5566959381103516,
      "learning_rate": 1.8240740740740744e-05,
      "loss": 0.2146,
      "step": 760
    },
    {
      "epoch": 0.0880787037037037,
      "grad_norm": 1.5733839273452759,
      "learning_rate": 1.8238425925925928e-05,
      "loss": 0.2969,
      "step": 761
    },
    {
      "epoch": 0.08819444444444445,
      "grad_norm": 1.2774697542190552,
      "learning_rate": 1.8236111111111112e-05,
      "loss": 0.2547,
      "step": 762
    },
    {
      "epoch": 0.08831018518518519,
      "grad_norm": 1.7920608520507812,
      "learning_rate": 1.8233796296296296e-05,
      "loss": 0.2928,
      "step": 763
    },
    {
      "epoch": 0.08842592592592592,
      "grad_norm": 23.168718338012695,
      "learning_rate": 1.8231481481481484e-05,
      "loss": 0.7176,
      "step": 764
    },
    {
      "epoch": 0.08854166666666667,
      "grad_norm": 48.395668029785156,
      "learning_rate": 1.8229166666666668e-05,
      "loss": 1.0908,
      "step": 765
    },
    {
      "epoch": 0.08865740740740741,
      "grad_norm": 2.922076940536499,
      "learning_rate": 1.8226851851851852e-05,
      "loss": 0.4479,
      "step": 766
    },
    {
      "epoch": 0.08877314814814814,
      "grad_norm": 1.1360087394714355,
      "learning_rate": 1.8224537037037037e-05,
      "loss": 0.208,
      "step": 767
    },
    {
      "epoch": 0.08888888888888889,
      "grad_norm": 1.3384579420089722,
      "learning_rate": 1.8222222222222224e-05,
      "loss": 0.2142,
      "step": 768
    },
    {
      "epoch": 0.08900462962962963,
      "grad_norm": 1.093686580657959,
      "learning_rate": 1.821990740740741e-05,
      "loss": 0.2294,
      "step": 769
    },
    {
      "epoch": 0.08912037037037036,
      "grad_norm": 43.38526916503906,
      "learning_rate": 1.8217592592592593e-05,
      "loss": 1.0855,
      "step": 770
    },
    {
      "epoch": 0.08923611111111111,
      "grad_norm": 6.733752250671387,
      "learning_rate": 1.821527777777778e-05,
      "loss": 0.3052,
      "step": 771
    },
    {
      "epoch": 0.08935185185185185,
      "grad_norm": 9.114775657653809,
      "learning_rate": 1.8212962962962965e-05,
      "loss": 0.3327,
      "step": 772
    },
    {
      "epoch": 0.0894675925925926,
      "grad_norm": 1.5729091167449951,
      "learning_rate": 1.821064814814815e-05,
      "loss": 0.1953,
      "step": 773
    },
    {
      "epoch": 0.08958333333333333,
      "grad_norm": 35.76393127441406,
      "learning_rate": 1.8208333333333333e-05,
      "loss": 0.8669,
      "step": 774
    },
    {
      "epoch": 0.08969907407407407,
      "grad_norm": 2.7412631511688232,
      "learning_rate": 1.820601851851852e-05,
      "loss": 0.3869,
      "step": 775
    },
    {
      "epoch": 0.08981481481481482,
      "grad_norm": 37.76375198364258,
      "learning_rate": 1.8203703703703705e-05,
      "loss": 0.8918,
      "step": 776
    },
    {
      "epoch": 0.08993055555555556,
      "grad_norm": 14.209734916687012,
      "learning_rate": 1.8201388888888893e-05,
      "loss": 1.6512,
      "step": 777
    },
    {
      "epoch": 0.09004629629629629,
      "grad_norm": 33.7147331237793,
      "learning_rate": 1.8199074074074074e-05,
      "loss": 0.7705,
      "step": 778
    },
    {
      "epoch": 0.09016203703703704,
      "grad_norm": 1.1750916242599487,
      "learning_rate": 1.819675925925926e-05,
      "loss": 0.2225,
      "step": 779
    },
    {
      "epoch": 0.09027777777777778,
      "grad_norm": 8.743062973022461,
      "learning_rate": 1.8194444444444445e-05,
      "loss": 0.5049,
      "step": 780
    },
    {
      "epoch": 0.09039351851851851,
      "grad_norm": 23.683467864990234,
      "learning_rate": 1.819212962962963e-05,
      "loss": 1.6534,
      "step": 781
    },
    {
      "epoch": 0.09050925925925926,
      "grad_norm": 18.2869873046875,
      "learning_rate": 1.8189814814814817e-05,
      "loss": 0.4459,
      "step": 782
    },
    {
      "epoch": 0.090625,
      "grad_norm": 1.222402572631836,
      "learning_rate": 1.81875e-05,
      "loss": 0.2439,
      "step": 783
    },
    {
      "epoch": 0.09074074074074075,
      "grad_norm": 3.1446070671081543,
      "learning_rate": 1.8185185185185186e-05,
      "loss": 0.3987,
      "step": 784
    },
    {
      "epoch": 0.09085648148148148,
      "grad_norm": 5.914869785308838,
      "learning_rate": 1.818287037037037e-05,
      "loss": 0.4744,
      "step": 785
    },
    {
      "epoch": 0.09097222222222222,
      "grad_norm": 1.1648402214050293,
      "learning_rate": 1.8180555555555558e-05,
      "loss": 0.2178,
      "step": 786
    },
    {
      "epoch": 0.09108796296296297,
      "grad_norm": 0.8942012190818787,
      "learning_rate": 1.8178240740740742e-05,
      "loss": 0.1903,
      "step": 787
    },
    {
      "epoch": 0.0912037037037037,
      "grad_norm": 7.825640678405762,
      "learning_rate": 1.817592592592593e-05,
      "loss": 0.4312,
      "step": 788
    },
    {
      "epoch": 0.09131944444444444,
      "grad_norm": 1.0307199954986572,
      "learning_rate": 1.8173611111111114e-05,
      "loss": 0.2099,
      "step": 789
    },
    {
      "epoch": 0.09143518518518519,
      "grad_norm": 31.071001052856445,
      "learning_rate": 1.8171296296296298e-05,
      "loss": 0.8251,
      "step": 790
    },
    {
      "epoch": 0.09155092592592592,
      "grad_norm": 1.9467766284942627,
      "learning_rate": 1.8168981481481482e-05,
      "loss": 0.2279,
      "step": 791
    },
    {
      "epoch": 0.09166666666666666,
      "grad_norm": 3.4483234882354736,
      "learning_rate": 1.8166666666666667e-05,
      "loss": 0.5213,
      "step": 792
    },
    {
      "epoch": 0.09178240740740741,
      "grad_norm": 3.2313406467437744,
      "learning_rate": 1.8164351851851854e-05,
      "loss": 0.2768,
      "step": 793
    },
    {
      "epoch": 0.09189814814814815,
      "grad_norm": 23.558439254760742,
      "learning_rate": 1.816203703703704e-05,
      "loss": 1.2068,
      "step": 794
    },
    {
      "epoch": 0.0920138888888889,
      "grad_norm": 2.668471574783325,
      "learning_rate": 1.8159722222222226e-05,
      "loss": 0.3931,
      "step": 795
    },
    {
      "epoch": 0.09212962962962963,
      "grad_norm": 26.088533401489258,
      "learning_rate": 1.8157407407407407e-05,
      "loss": 1.1321,
      "step": 796
    },
    {
      "epoch": 0.09224537037037037,
      "grad_norm": 11.992645263671875,
      "learning_rate": 1.8155092592592595e-05,
      "loss": 0.5264,
      "step": 797
    },
    {
      "epoch": 0.09236111111111112,
      "grad_norm": 2.262078285217285,
      "learning_rate": 1.815277777777778e-05,
      "loss": 0.2427,
      "step": 798
    },
    {
      "epoch": 0.09247685185185185,
      "grad_norm": 0.8814418315887451,
      "learning_rate": 1.8150462962962963e-05,
      "loss": 0.1755,
      "step": 799
    },
    {
      "epoch": 0.09259259259259259,
      "grad_norm": 0.8799277544021606,
      "learning_rate": 1.814814814814815e-05,
      "loss": 0.1891,
      "step": 800
    },
    {
      "epoch": 0.09270833333333334,
      "grad_norm": 1.2110542058944702,
      "learning_rate": 1.8145833333333335e-05,
      "loss": 0.2731,
      "step": 801
    },
    {
      "epoch": 0.09282407407407407,
      "grad_norm": 3.578211545944214,
      "learning_rate": 1.814351851851852e-05,
      "loss": 0.2841,
      "step": 802
    },
    {
      "epoch": 0.09293981481481481,
      "grad_norm": 5.682281970977783,
      "learning_rate": 1.8141203703703704e-05,
      "loss": 1.611,
      "step": 803
    },
    {
      "epoch": 0.09305555555555556,
      "grad_norm": 14.273393630981445,
      "learning_rate": 1.813888888888889e-05,
      "loss": 1.6447,
      "step": 804
    },
    {
      "epoch": 0.0931712962962963,
      "grad_norm": 2.3659398555755615,
      "learning_rate": 1.8136574074074075e-05,
      "loss": 0.424,
      "step": 805
    },
    {
      "epoch": 0.09328703703703704,
      "grad_norm": 1.611146092414856,
      "learning_rate": 1.8134259259259263e-05,
      "loss": 0.2387,
      "step": 806
    },
    {
      "epoch": 0.09340277777777778,
      "grad_norm": 6.87191915512085,
      "learning_rate": 1.8131944444444447e-05,
      "loss": 1.8251,
      "step": 807
    },
    {
      "epoch": 0.09351851851851851,
      "grad_norm": 1.1196627616882324,
      "learning_rate": 1.812962962962963e-05,
      "loss": 0.2051,
      "step": 808
    },
    {
      "epoch": 0.09363425925925926,
      "grad_norm": 3.4432525634765625,
      "learning_rate": 1.8127314814814816e-05,
      "loss": 0.3024,
      "step": 809
    },
    {
      "epoch": 0.09375,
      "grad_norm": 1.4886088371276855,
      "learning_rate": 1.8125e-05,
      "loss": 0.29,
      "step": 810
    },
    {
      "epoch": 0.09386574074074074,
      "grad_norm": 14.520753860473633,
      "learning_rate": 1.8122685185185188e-05,
      "loss": 0.7265,
      "step": 811
    },
    {
      "epoch": 0.09398148148148149,
      "grad_norm": 4.23621940612793,
      "learning_rate": 1.8120370370370372e-05,
      "loss": 0.3178,
      "step": 812
    },
    {
      "epoch": 0.09409722222222222,
      "grad_norm": 14.405874252319336,
      "learning_rate": 1.8118055555555556e-05,
      "loss": 1.3725,
      "step": 813
    },
    {
      "epoch": 0.09421296296296296,
      "grad_norm": 9.321107864379883,
      "learning_rate": 1.811574074074074e-05,
      "loss": 0.6397,
      "step": 814
    },
    {
      "epoch": 0.0943287037037037,
      "grad_norm": 4.0120744705200195,
      "learning_rate": 1.8113425925925928e-05,
      "loss": 0.3808,
      "step": 815
    },
    {
      "epoch": 0.09444444444444444,
      "grad_norm": 0.8002557158470154,
      "learning_rate": 1.8111111111111112e-05,
      "loss": 0.1727,
      "step": 816
    },
    {
      "epoch": 0.09456018518518519,
      "grad_norm": 1.1554436683654785,
      "learning_rate": 1.8108796296296297e-05,
      "loss": 0.1804,
      "step": 817
    },
    {
      "epoch": 0.09467592592592593,
      "grad_norm": 0.7586098313331604,
      "learning_rate": 1.8106481481481484e-05,
      "loss": 0.1632,
      "step": 818
    },
    {
      "epoch": 0.09479166666666666,
      "grad_norm": 9.18211841583252,
      "learning_rate": 1.810416666666667e-05,
      "loss": 0.4634,
      "step": 819
    },
    {
      "epoch": 0.09490740740740741,
      "grad_norm": 5.4538493156433105,
      "learning_rate": 1.8101851851851853e-05,
      "loss": 0.2825,
      "step": 820
    },
    {
      "epoch": 0.09502314814814815,
      "grad_norm": 3.68876051902771,
      "learning_rate": 1.8099537037037037e-05,
      "loss": 0.4404,
      "step": 821
    },
    {
      "epoch": 0.09513888888888888,
      "grad_norm": 2.143946409225464,
      "learning_rate": 1.8097222222222225e-05,
      "loss": 0.2507,
      "step": 822
    },
    {
      "epoch": 0.09525462962962963,
      "grad_norm": 3.2228193283081055,
      "learning_rate": 1.809490740740741e-05,
      "loss": 0.3472,
      "step": 823
    },
    {
      "epoch": 0.09537037037037037,
      "grad_norm": 3.1841542720794678,
      "learning_rate": 1.8092592592592597e-05,
      "loss": 0.2851,
      "step": 824
    },
    {
      "epoch": 0.0954861111111111,
      "grad_norm": 3.8580315113067627,
      "learning_rate": 1.8090277777777777e-05,
      "loss": 0.4769,
      "step": 825
    },
    {
      "epoch": 0.09560185185185185,
      "grad_norm": 45.17533874511719,
      "learning_rate": 1.8087962962962965e-05,
      "loss": 0.7775,
      "step": 826
    },
    {
      "epoch": 0.09571759259259259,
      "grad_norm": 27.759214401245117,
      "learning_rate": 1.808564814814815e-05,
      "loss": 0.4762,
      "step": 827
    },
    {
      "epoch": 0.09583333333333334,
      "grad_norm": 1.0599350929260254,
      "learning_rate": 1.8083333333333334e-05,
      "loss": 0.186,
      "step": 828
    },
    {
      "epoch": 0.09594907407407408,
      "grad_norm": 14.567631721496582,
      "learning_rate": 1.808101851851852e-05,
      "loss": 0.4504,
      "step": 829
    },
    {
      "epoch": 0.09606481481481481,
      "grad_norm": 2.117802619934082,
      "learning_rate": 1.8078703703703705e-05,
      "loss": 0.1876,
      "step": 830
    },
    {
      "epoch": 0.09618055555555556,
      "grad_norm": 21.024072647094727,
      "learning_rate": 1.807638888888889e-05,
      "loss": 0.7941,
      "step": 831
    },
    {
      "epoch": 0.0962962962962963,
      "grad_norm": 1.640750527381897,
      "learning_rate": 1.8074074074074074e-05,
      "loss": 0.3448,
      "step": 832
    },
    {
      "epoch": 0.09641203703703703,
      "grad_norm": 3.779021978378296,
      "learning_rate": 1.807175925925926e-05,
      "loss": 0.4296,
      "step": 833
    },
    {
      "epoch": 0.09652777777777778,
      "grad_norm": 13.137493133544922,
      "learning_rate": 1.8069444444444446e-05,
      "loss": 1.3459,
      "step": 834
    },
    {
      "epoch": 0.09664351851851852,
      "grad_norm": 2.9277052879333496,
      "learning_rate": 1.806712962962963e-05,
      "loss": 0.4614,
      "step": 835
    },
    {
      "epoch": 0.09675925925925925,
      "grad_norm": 2.311539649963379,
      "learning_rate": 1.8064814814814818e-05,
      "loss": 0.259,
      "step": 836
    },
    {
      "epoch": 0.096875,
      "grad_norm": 4.224998474121094,
      "learning_rate": 1.8062500000000002e-05,
      "loss": 0.3701,
      "step": 837
    },
    {
      "epoch": 0.09699074074074074,
      "grad_norm": 0.725184440612793,
      "learning_rate": 1.8060185185185186e-05,
      "loss": 0.1496,
      "step": 838
    },
    {
      "epoch": 0.09710648148148149,
      "grad_norm": 0.8388811945915222,
      "learning_rate": 1.805787037037037e-05,
      "loss": 0.178,
      "step": 839
    },
    {
      "epoch": 0.09722222222222222,
      "grad_norm": 3.7640902996063232,
      "learning_rate": 1.8055555555555558e-05,
      "loss": 0.3525,
      "step": 840
    },
    {
      "epoch": 0.09733796296296296,
      "grad_norm": 1.720296025276184,
      "learning_rate": 1.8053240740740742e-05,
      "loss": 0.2446,
      "step": 841
    },
    {
      "epoch": 0.09745370370370371,
      "grad_norm": 1.2039039134979248,
      "learning_rate": 1.805092592592593e-05,
      "loss": 0.1936,
      "step": 842
    },
    {
      "epoch": 0.09756944444444444,
      "grad_norm": 1.225277066230774,
      "learning_rate": 1.804861111111111e-05,
      "loss": 0.2232,
      "step": 843
    },
    {
      "epoch": 0.09768518518518518,
      "grad_norm": 1.7095928192138672,
      "learning_rate": 1.80462962962963e-05,
      "loss": 0.283,
      "step": 844
    },
    {
      "epoch": 0.09780092592592593,
      "grad_norm": 26.319799423217773,
      "learning_rate": 1.8043981481481483e-05,
      "loss": 0.5613,
      "step": 845
    },
    {
      "epoch": 0.09791666666666667,
      "grad_norm": 10.87532901763916,
      "learning_rate": 1.8041666666666667e-05,
      "loss": 1.3086,
      "step": 846
    },
    {
      "epoch": 0.0980324074074074,
      "grad_norm": 1.2089107036590576,
      "learning_rate": 1.8039351851851855e-05,
      "loss": 0.1898,
      "step": 847
    },
    {
      "epoch": 0.09814814814814815,
      "grad_norm": 2.052461862564087,
      "learning_rate": 1.803703703703704e-05,
      "loss": 0.2081,
      "step": 848
    },
    {
      "epoch": 0.09826388888888889,
      "grad_norm": 3.3953282833099365,
      "learning_rate": 1.8034722222222223e-05,
      "loss": 0.2757,
      "step": 849
    },
    {
      "epoch": 0.09837962962962964,
      "grad_norm": 1.1815133094787598,
      "learning_rate": 1.8032407407407407e-05,
      "loss": 0.1846,
      "step": 850
    },
    {
      "epoch": 0.09849537037037037,
      "grad_norm": 1.740225911140442,
      "learning_rate": 1.8030092592592595e-05,
      "loss": 0.3272,
      "step": 851
    },
    {
      "epoch": 0.09861111111111111,
      "grad_norm": 1.5746774673461914,
      "learning_rate": 1.802777777777778e-05,
      "loss": 0.2889,
      "step": 852
    },
    {
      "epoch": 0.09872685185185186,
      "grad_norm": 1.429338812828064,
      "learning_rate": 1.8025462962962964e-05,
      "loss": 0.2806,
      "step": 853
    },
    {
      "epoch": 0.09884259259259259,
      "grad_norm": 1.3082226514816284,
      "learning_rate": 1.802314814814815e-05,
      "loss": 0.1844,
      "step": 854
    },
    {
      "epoch": 0.09895833333333333,
      "grad_norm": 1.0798845291137695,
      "learning_rate": 1.8020833333333335e-05,
      "loss": 0.1939,
      "step": 855
    },
    {
      "epoch": 0.09907407407407408,
      "grad_norm": 14.098761558532715,
      "learning_rate": 1.801851851851852e-05,
      "loss": 0.3813,
      "step": 856
    },
    {
      "epoch": 0.09918981481481481,
      "grad_norm": 1.732418417930603,
      "learning_rate": 1.8016203703703704e-05,
      "loss": 0.3419,
      "step": 857
    },
    {
      "epoch": 0.09930555555555555,
      "grad_norm": 1.0655219554901123,
      "learning_rate": 1.801388888888889e-05,
      "loss": 0.2238,
      "step": 858
    },
    {
      "epoch": 0.0994212962962963,
      "grad_norm": 1.336389183998108,
      "learning_rate": 1.8011574074074076e-05,
      "loss": 0.262,
      "step": 859
    },
    {
      "epoch": 0.09953703703703703,
      "grad_norm": 2.534055709838867,
      "learning_rate": 1.800925925925926e-05,
      "loss": 0.2245,
      "step": 860
    },
    {
      "epoch": 0.09965277777777778,
      "grad_norm": 5.373302459716797,
      "learning_rate": 1.8006944444444444e-05,
      "loss": 0.4244,
      "step": 861
    },
    {
      "epoch": 0.09976851851851852,
      "grad_norm": 1.6468784809112549,
      "learning_rate": 1.8004629629629632e-05,
      "loss": 0.2157,
      "step": 862
    },
    {
      "epoch": 0.09988425925925926,
      "grad_norm": 18.360849380493164,
      "learning_rate": 1.8002314814814816e-05,
      "loss": 1.2934,
      "step": 863
    },
    {
      "epoch": 0.1,
      "grad_norm": 18.4736270904541,
      "learning_rate": 1.8e-05,
      "loss": 1.4023,
      "step": 864
    },
    {
      "epoch": 0.10011574074074074,
      "grad_norm": 1.0178676843643188,
      "learning_rate": 1.7997685185185188e-05,
      "loss": 0.2236,
      "step": 865
    },
    {
      "epoch": 0.10023148148148148,
      "grad_norm": 0.8584443926811218,
      "learning_rate": 1.7995370370370372e-05,
      "loss": 0.1516,
      "step": 866
    },
    {
      "epoch": 0.10034722222222223,
      "grad_norm": 41.76673889160156,
      "learning_rate": 1.7993055555555557e-05,
      "loss": 1.1008,
      "step": 867
    },
    {
      "epoch": 0.10046296296296296,
      "grad_norm": 0.9417381286621094,
      "learning_rate": 1.799074074074074e-05,
      "loss": 0.1921,
      "step": 868
    },
    {
      "epoch": 0.1005787037037037,
      "grad_norm": 1.46173894405365,
      "learning_rate": 1.798842592592593e-05,
      "loss": 0.2539,
      "step": 869
    },
    {
      "epoch": 0.10069444444444445,
      "grad_norm": 1.5956535339355469,
      "learning_rate": 1.7986111111111113e-05,
      "loss": 0.2407,
      "step": 870
    },
    {
      "epoch": 0.10081018518518518,
      "grad_norm": 0.9757245182991028,
      "learning_rate": 1.7983796296296297e-05,
      "loss": 0.2047,
      "step": 871
    },
    {
      "epoch": 0.10092592592592593,
      "grad_norm": 11.711851119995117,
      "learning_rate": 1.798148148148148e-05,
      "loss": 1.5145,
      "step": 872
    },
    {
      "epoch": 0.10104166666666667,
      "grad_norm": 16.32665252685547,
      "learning_rate": 1.797916666666667e-05,
      "loss": 0.4417,
      "step": 873
    },
    {
      "epoch": 0.1011574074074074,
      "grad_norm": 20.417192459106445,
      "learning_rate": 1.7976851851851853e-05,
      "loss": 1.1617,
      "step": 874
    },
    {
      "epoch": 0.10127314814814815,
      "grad_norm": 0.8192079663276672,
      "learning_rate": 1.7974537037037037e-05,
      "loss": 0.1699,
      "step": 875
    },
    {
      "epoch": 0.10138888888888889,
      "grad_norm": 1.4257110357284546,
      "learning_rate": 1.7972222222222225e-05,
      "loss": 0.2526,
      "step": 876
    },
    {
      "epoch": 0.10150462962962963,
      "grad_norm": 34.74717712402344,
      "learning_rate": 1.796990740740741e-05,
      "loss": 0.6405,
      "step": 877
    },
    {
      "epoch": 0.10162037037037037,
      "grad_norm": 1.104390025138855,
      "learning_rate": 1.7967592592592594e-05,
      "loss": 0.1918,
      "step": 878
    },
    {
      "epoch": 0.10173611111111111,
      "grad_norm": 0.8120255470275879,
      "learning_rate": 1.7965277777777778e-05,
      "loss": 0.1667,
      "step": 879
    },
    {
      "epoch": 0.10185185185185185,
      "grad_norm": 14.793447494506836,
      "learning_rate": 1.7962962962962965e-05,
      "loss": 0.409,
      "step": 880
    },
    {
      "epoch": 0.1019675925925926,
      "grad_norm": 2.1309096813201904,
      "learning_rate": 1.796064814814815e-05,
      "loss": 0.3499,
      "step": 881
    },
    {
      "epoch": 0.10208333333333333,
      "grad_norm": 2.0536186695098877,
      "learning_rate": 1.7958333333333334e-05,
      "loss": 0.3028,
      "step": 882
    },
    {
      "epoch": 0.10219907407407407,
      "grad_norm": 1.4988479614257812,
      "learning_rate": 1.795601851851852e-05,
      "loss": 0.2657,
      "step": 883
    },
    {
      "epoch": 0.10231481481481482,
      "grad_norm": 2.099879264831543,
      "learning_rate": 1.7953703703703706e-05,
      "loss": 0.1546,
      "step": 884
    },
    {
      "epoch": 0.10243055555555555,
      "grad_norm": 5.231938362121582,
      "learning_rate": 1.795138888888889e-05,
      "loss": 0.4161,
      "step": 885
    },
    {
      "epoch": 0.1025462962962963,
      "grad_norm": 23.75330924987793,
      "learning_rate": 1.7949074074074074e-05,
      "loss": 0.9962,
      "step": 886
    },
    {
      "epoch": 0.10266203703703704,
      "grad_norm": 7.912933826446533,
      "learning_rate": 1.7946759259259262e-05,
      "loss": 0.2153,
      "step": 887
    },
    {
      "epoch": 0.10277777777777777,
      "grad_norm": 13.473678588867188,
      "learning_rate": 1.7944444444444446e-05,
      "loss": 0.2958,
      "step": 888
    },
    {
      "epoch": 0.10289351851851852,
      "grad_norm": 1.5622657537460327,
      "learning_rate": 1.794212962962963e-05,
      "loss": 0.3029,
      "step": 889
    },
    {
      "epoch": 0.10300925925925926,
      "grad_norm": 4.848580837249756,
      "learning_rate": 1.7939814814814815e-05,
      "loss": 0.3472,
      "step": 890
    },
    {
      "epoch": 0.103125,
      "grad_norm": 3.5467615127563477,
      "learning_rate": 1.7937500000000002e-05,
      "loss": 0.3771,
      "step": 891
    },
    {
      "epoch": 0.10324074074074074,
      "grad_norm": 3.466478109359741,
      "learning_rate": 1.7935185185185187e-05,
      "loss": 0.2361,
      "step": 892
    },
    {
      "epoch": 0.10335648148148148,
      "grad_norm": 4.806948184967041,
      "learning_rate": 1.793287037037037e-05,
      "loss": 0.2564,
      "step": 893
    },
    {
      "epoch": 0.10347222222222222,
      "grad_norm": 1.2298638820648193,
      "learning_rate": 1.793055555555556e-05,
      "loss": 0.2645,
      "step": 894
    },
    {
      "epoch": 0.10358796296296297,
      "grad_norm": 14.64049243927002,
      "learning_rate": 1.7928240740740743e-05,
      "loss": 0.4051,
      "step": 895
    },
    {
      "epoch": 0.1037037037037037,
      "grad_norm": 0.9379496574401855,
      "learning_rate": 1.7925925925925927e-05,
      "loss": 0.1507,
      "step": 896
    },
    {
      "epoch": 0.10381944444444445,
      "grad_norm": 0.8267108798027039,
      "learning_rate": 1.792361111111111e-05,
      "loss": 0.1777,
      "step": 897
    },
    {
      "epoch": 0.10393518518518519,
      "grad_norm": 2.224966526031494,
      "learning_rate": 1.79212962962963e-05,
      "loss": 0.2554,
      "step": 898
    },
    {
      "epoch": 0.10405092592592592,
      "grad_norm": 1.433052659034729,
      "learning_rate": 1.7918981481481483e-05,
      "loss": 0.2193,
      "step": 899
    },
    {
      "epoch": 0.10416666666666667,
      "grad_norm": 0.8243880271911621,
      "learning_rate": 1.7916666666666667e-05,
      "loss": 0.164,
      "step": 900
    },
    {
      "epoch": 0.10428240740740741,
      "grad_norm": 1.0491124391555786,
      "learning_rate": 1.791435185185185e-05,
      "loss": 0.1661,
      "step": 901
    },
    {
      "epoch": 0.10439814814814814,
      "grad_norm": 8.241581916809082,
      "learning_rate": 1.791203703703704e-05,
      "loss": 0.3657,
      "step": 902
    },
    {
      "epoch": 0.10451388888888889,
      "grad_norm": 2.1867682933807373,
      "learning_rate": 1.7909722222222223e-05,
      "loss": 0.3093,
      "step": 903
    },
    {
      "epoch": 0.10462962962962963,
      "grad_norm": 1.7824907302856445,
      "learning_rate": 1.7907407407407408e-05,
      "loss": 0.218,
      "step": 904
    },
    {
      "epoch": 0.10474537037037036,
      "grad_norm": 5.8760528564453125,
      "learning_rate": 1.7905092592592595e-05,
      "loss": 0.2615,
      "step": 905
    },
    {
      "epoch": 0.10486111111111111,
      "grad_norm": 19.128740310668945,
      "learning_rate": 1.790277777777778e-05,
      "loss": 0.795,
      "step": 906
    },
    {
      "epoch": 0.10497685185185185,
      "grad_norm": 2.4776546955108643,
      "learning_rate": 1.7900462962962964e-05,
      "loss": 0.2262,
      "step": 907
    },
    {
      "epoch": 0.1050925925925926,
      "grad_norm": 43.7632942199707,
      "learning_rate": 1.7898148148148148e-05,
      "loss": 0.754,
      "step": 908
    },
    {
      "epoch": 0.10520833333333333,
      "grad_norm": 17.09293556213379,
      "learning_rate": 1.7895833333333336e-05,
      "loss": 1.3151,
      "step": 909
    },
    {
      "epoch": 0.10532407407407407,
      "grad_norm": 42.78043746948242,
      "learning_rate": 1.789351851851852e-05,
      "loss": 0.8329,
      "step": 910
    },
    {
      "epoch": 0.10543981481481482,
      "grad_norm": 34.79784393310547,
      "learning_rate": 1.7891203703703704e-05,
      "loss": 0.5147,
      "step": 911
    },
    {
      "epoch": 0.10555555555555556,
      "grad_norm": 11.897608757019043,
      "learning_rate": 1.7888888888888892e-05,
      "loss": 0.2786,
      "step": 912
    },
    {
      "epoch": 0.10567129629629629,
      "grad_norm": 0.8551439046859741,
      "learning_rate": 1.7886574074074076e-05,
      "loss": 0.1785,
      "step": 913
    },
    {
      "epoch": 0.10578703703703704,
      "grad_norm": 1.106663465499878,
      "learning_rate": 1.788425925925926e-05,
      "loss": 0.2272,
      "step": 914
    },
    {
      "epoch": 0.10590277777777778,
      "grad_norm": 6.319456577301025,
      "learning_rate": 1.7881944444444445e-05,
      "loss": 0.2808,
      "step": 915
    },
    {
      "epoch": 0.10601851851851851,
      "grad_norm": 4.523231506347656,
      "learning_rate": 1.7879629629629632e-05,
      "loss": 0.2522,
      "step": 916
    },
    {
      "epoch": 0.10613425925925926,
      "grad_norm": 0.8804287910461426,
      "learning_rate": 1.7877314814814817e-05,
      "loss": 0.1914,
      "step": 917
    },
    {
      "epoch": 0.10625,
      "grad_norm": 0.9426946640014648,
      "learning_rate": 1.7875e-05,
      "loss": 0.1971,
      "step": 918
    },
    {
      "epoch": 0.10636574074074075,
      "grad_norm": 1.3107248544692993,
      "learning_rate": 1.7872685185185185e-05,
      "loss": 0.2318,
      "step": 919
    },
    {
      "epoch": 0.10648148148148148,
      "grad_norm": 11.1814546585083,
      "learning_rate": 1.7870370370370373e-05,
      "loss": 0.3819,
      "step": 920
    },
    {
      "epoch": 0.10659722222222222,
      "grad_norm": 0.6889748573303223,
      "learning_rate": 1.7868055555555557e-05,
      "loss": 0.1458,
      "step": 921
    },
    {
      "epoch": 0.10671296296296297,
      "grad_norm": 1.370671272277832,
      "learning_rate": 1.786574074074074e-05,
      "loss": 0.1704,
      "step": 922
    },
    {
      "epoch": 0.1068287037037037,
      "grad_norm": 1.614721417427063,
      "learning_rate": 1.786342592592593e-05,
      "loss": 0.2346,
      "step": 923
    },
    {
      "epoch": 0.10694444444444444,
      "grad_norm": 10.436413764953613,
      "learning_rate": 1.7861111111111113e-05,
      "loss": 0.2667,
      "step": 924
    },
    {
      "epoch": 0.10706018518518519,
      "grad_norm": 11.948115348815918,
      "learning_rate": 1.7858796296296297e-05,
      "loss": 0.3862,
      "step": 925
    },
    {
      "epoch": 0.10717592592592592,
      "grad_norm": 0.7988684773445129,
      "learning_rate": 1.785648148148148e-05,
      "loss": 0.1245,
      "step": 926
    },
    {
      "epoch": 0.10729166666666666,
      "grad_norm": 1.0606193542480469,
      "learning_rate": 1.785416666666667e-05,
      "loss": 0.1647,
      "step": 927
    },
    {
      "epoch": 0.10740740740740741,
      "grad_norm": 31.93965721130371,
      "learning_rate": 1.7851851851851853e-05,
      "loss": 0.8961,
      "step": 928
    },
    {
      "epoch": 0.10752314814814815,
      "grad_norm": 1.0687813758850098,
      "learning_rate": 1.7849537037037038e-05,
      "loss": 0.1994,
      "step": 929
    },
    {
      "epoch": 0.1076388888888889,
      "grad_norm": 1.7811115980148315,
      "learning_rate": 1.7847222222222225e-05,
      "loss": 0.2271,
      "step": 930
    },
    {
      "epoch": 0.10775462962962963,
      "grad_norm": 15.320669174194336,
      "learning_rate": 1.784490740740741e-05,
      "loss": 0.3667,
      "step": 931
    },
    {
      "epoch": 0.10787037037037037,
      "grad_norm": 1.5281755924224854,
      "learning_rate": 1.7842592592592594e-05,
      "loss": 0.2505,
      "step": 932
    },
    {
      "epoch": 0.10798611111111112,
      "grad_norm": 13.036376953125,
      "learning_rate": 1.7840277777777778e-05,
      "loss": 0.484,
      "step": 933
    },
    {
      "epoch": 0.10810185185185185,
      "grad_norm": 15.992353439331055,
      "learning_rate": 1.7837962962962966e-05,
      "loss": 1.4278,
      "step": 934
    },
    {
      "epoch": 0.10821759259259259,
      "grad_norm": 0.9154366850852966,
      "learning_rate": 1.783564814814815e-05,
      "loss": 0.1851,
      "step": 935
    },
    {
      "epoch": 0.10833333333333334,
      "grad_norm": 0.8447420001029968,
      "learning_rate": 1.7833333333333334e-05,
      "loss": 0.1731,
      "step": 936
    },
    {
      "epoch": 0.10844907407407407,
      "grad_norm": 3.706108808517456,
      "learning_rate": 1.783101851851852e-05,
      "loss": 0.2922,
      "step": 937
    },
    {
      "epoch": 0.10856481481481481,
      "grad_norm": 2.567166805267334,
      "learning_rate": 1.7828703703703706e-05,
      "loss": 0.2478,
      "step": 938
    },
    {
      "epoch": 0.10868055555555556,
      "grad_norm": 28.2506160736084,
      "learning_rate": 1.782638888888889e-05,
      "loss": 1.296,
      "step": 939
    },
    {
      "epoch": 0.1087962962962963,
      "grad_norm": 0.9259400963783264,
      "learning_rate": 1.7824074074074075e-05,
      "loss": 0.1814,
      "step": 940
    },
    {
      "epoch": 0.10891203703703704,
      "grad_norm": 1.5141676664352417,
      "learning_rate": 1.7821759259259262e-05,
      "loss": 0.2343,
      "step": 941
    },
    {
      "epoch": 0.10902777777777778,
      "grad_norm": 26.369823455810547,
      "learning_rate": 1.7819444444444447e-05,
      "loss": 0.5588,
      "step": 942
    },
    {
      "epoch": 0.10914351851851851,
      "grad_norm": 20.515331268310547,
      "learning_rate": 1.781712962962963e-05,
      "loss": 1.4608,
      "step": 943
    },
    {
      "epoch": 0.10925925925925926,
      "grad_norm": 0.6507753133773804,
      "learning_rate": 1.7814814814814815e-05,
      "loss": 0.1389,
      "step": 944
    },
    {
      "epoch": 0.109375,
      "grad_norm": 2.4835691452026367,
      "learning_rate": 1.7812500000000003e-05,
      "loss": 0.215,
      "step": 945
    },
    {
      "epoch": 0.10949074074074074,
      "grad_norm": 7.235436916351318,
      "learning_rate": 1.7810185185185187e-05,
      "loss": 0.2502,
      "step": 946
    },
    {
      "epoch": 0.10960648148148149,
      "grad_norm": 6.284767150878906,
      "learning_rate": 1.780787037037037e-05,
      "loss": 0.2728,
      "step": 947
    },
    {
      "epoch": 0.10972222222222222,
      "grad_norm": 1.7583916187286377,
      "learning_rate": 1.7805555555555555e-05,
      "loss": 0.2355,
      "step": 948
    },
    {
      "epoch": 0.10983796296296296,
      "grad_norm": 4.482048034667969,
      "learning_rate": 1.7803240740740743e-05,
      "loss": 0.2109,
      "step": 949
    },
    {
      "epoch": 0.1099537037037037,
      "grad_norm": 0.7161057591438293,
      "learning_rate": 1.7800925925925927e-05,
      "loss": 0.1456,
      "step": 950
    },
    {
      "epoch": 0.11006944444444444,
      "grad_norm": 1.883136510848999,
      "learning_rate": 1.779861111111111e-05,
      "loss": 0.2948,
      "step": 951
    },
    {
      "epoch": 0.11018518518518519,
      "grad_norm": 0.6609748601913452,
      "learning_rate": 1.77962962962963e-05,
      "loss": 0.1389,
      "step": 952
    },
    {
      "epoch": 0.11030092592592593,
      "grad_norm": 18.12482261657715,
      "learning_rate": 1.7793981481481483e-05,
      "loss": 0.5119,
      "step": 953
    },
    {
      "epoch": 0.11041666666666666,
      "grad_norm": 12.756900787353516,
      "learning_rate": 1.7791666666666668e-05,
      "loss": 0.4831,
      "step": 954
    },
    {
      "epoch": 0.11053240740740741,
      "grad_norm": 1.3796919584274292,
      "learning_rate": 1.7789351851851852e-05,
      "loss": 0.3134,
      "step": 955
    },
    {
      "epoch": 0.11064814814814815,
      "grad_norm": 0.8907844424247742,
      "learning_rate": 1.778703703703704e-05,
      "loss": 0.1912,
      "step": 956
    },
    {
      "epoch": 0.11076388888888888,
      "grad_norm": 26.807775497436523,
      "learning_rate": 1.7784722222222224e-05,
      "loss": 0.5881,
      "step": 957
    },
    {
      "epoch": 0.11087962962962963,
      "grad_norm": 1.2391729354858398,
      "learning_rate": 1.7782407407407408e-05,
      "loss": 0.2361,
      "step": 958
    },
    {
      "epoch": 0.11099537037037037,
      "grad_norm": 0.8769028186798096,
      "learning_rate": 1.7780092592592596e-05,
      "loss": 0.1766,
      "step": 959
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 37.710914611816406,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.9737,
      "step": 960
    },
    {
      "epoch": 0.11122685185185185,
      "grad_norm": 1.7065391540527344,
      "learning_rate": 1.7775462962962964e-05,
      "loss": 0.1651,
      "step": 961
    },
    {
      "epoch": 0.11134259259259259,
      "grad_norm": 1.6461848020553589,
      "learning_rate": 1.777314814814815e-05,
      "loss": 0.2668,
      "step": 962
    },
    {
      "epoch": 0.11145833333333334,
      "grad_norm": 0.6754774451255798,
      "learning_rate": 1.7770833333333336e-05,
      "loss": 0.1443,
      "step": 963
    },
    {
      "epoch": 0.11157407407407408,
      "grad_norm": 1.2552982568740845,
      "learning_rate": 1.776851851851852e-05,
      "loss": 0.218,
      "step": 964
    },
    {
      "epoch": 0.11168981481481481,
      "grad_norm": 1.099179744720459,
      "learning_rate": 1.7766203703703705e-05,
      "loss": 0.1729,
      "step": 965
    },
    {
      "epoch": 0.11180555555555556,
      "grad_norm": 1.3121386766433716,
      "learning_rate": 1.776388888888889e-05,
      "loss": 0.204,
      "step": 966
    },
    {
      "epoch": 0.1119212962962963,
      "grad_norm": 3.5794897079467773,
      "learning_rate": 1.7761574074074077e-05,
      "loss": 0.2877,
      "step": 967
    },
    {
      "epoch": 0.11203703703703703,
      "grad_norm": 1.3536657094955444,
      "learning_rate": 1.775925925925926e-05,
      "loss": 0.1676,
      "step": 968
    },
    {
      "epoch": 0.11215277777777778,
      "grad_norm": 0.8449748754501343,
      "learning_rate": 1.7756944444444445e-05,
      "loss": 0.181,
      "step": 969
    },
    {
      "epoch": 0.11226851851851852,
      "grad_norm": 10.876982688903809,
      "learning_rate": 1.7754629629629633e-05,
      "loss": 0.3503,
      "step": 970
    },
    {
      "epoch": 0.11238425925925925,
      "grad_norm": 1.1542909145355225,
      "learning_rate": 1.7752314814814817e-05,
      "loss": 0.2127,
      "step": 971
    },
    {
      "epoch": 0.1125,
      "grad_norm": 1.7917500734329224,
      "learning_rate": 1.775e-05,
      "loss": 0.2848,
      "step": 972
    },
    {
      "epoch": 0.11261574074074074,
      "grad_norm": 2.6777942180633545,
      "learning_rate": 1.7747685185185185e-05,
      "loss": 0.1629,
      "step": 973
    },
    {
      "epoch": 0.11273148148148149,
      "grad_norm": 1.0787115097045898,
      "learning_rate": 1.7745370370370373e-05,
      "loss": 0.1584,
      "step": 974
    },
    {
      "epoch": 0.11284722222222222,
      "grad_norm": 2.109240770339966,
      "learning_rate": 1.7743055555555557e-05,
      "loss": 0.2538,
      "step": 975
    },
    {
      "epoch": 0.11296296296296296,
      "grad_norm": 5.854368209838867,
      "learning_rate": 1.774074074074074e-05,
      "loss": 0.2437,
      "step": 976
    },
    {
      "epoch": 0.11307870370370371,
      "grad_norm": 74.68434143066406,
      "learning_rate": 1.773842592592593e-05,
      "loss": 0.8574,
      "step": 977
    },
    {
      "epoch": 0.11319444444444444,
      "grad_norm": 7.496344566345215,
      "learning_rate": 1.773611111111111e-05,
      "loss": 0.3015,
      "step": 978
    },
    {
      "epoch": 0.11331018518518518,
      "grad_norm": 0.7161617279052734,
      "learning_rate": 1.7733796296296298e-05,
      "loss": 0.1526,
      "step": 979
    },
    {
      "epoch": 0.11342592592592593,
      "grad_norm": 14.537853240966797,
      "learning_rate": 1.7731481481481482e-05,
      "loss": 1.5137,
      "step": 980
    },
    {
      "epoch": 0.11354166666666667,
      "grad_norm": 33.31151580810547,
      "learning_rate": 1.772916666666667e-05,
      "loss": 1.237,
      "step": 981
    },
    {
      "epoch": 0.1136574074074074,
      "grad_norm": 1.2234388589859009,
      "learning_rate": 1.7726851851851854e-05,
      "loss": 0.1957,
      "step": 982
    },
    {
      "epoch": 0.11377314814814815,
      "grad_norm": 0.7005996108055115,
      "learning_rate": 1.7724537037037038e-05,
      "loss": 0.1362,
      "step": 983
    },
    {
      "epoch": 0.11388888888888889,
      "grad_norm": 6.345552921295166,
      "learning_rate": 1.7722222222222222e-05,
      "loss": 1.8345,
      "step": 984
    },
    {
      "epoch": 0.11400462962962964,
      "grad_norm": 0.7814491391181946,
      "learning_rate": 1.771990740740741e-05,
      "loss": 0.1171,
      "step": 985
    },
    {
      "epoch": 0.11412037037037037,
      "grad_norm": 1.3245612382888794,
      "learning_rate": 1.7717592592592594e-05,
      "loss": 0.2044,
      "step": 986
    },
    {
      "epoch": 0.11423611111111111,
      "grad_norm": 0.8340675830841064,
      "learning_rate": 1.771527777777778e-05,
      "loss": 0.1684,
      "step": 987
    },
    {
      "epoch": 0.11435185185185186,
      "grad_norm": 0.6388011574745178,
      "learning_rate": 1.7712962962962966e-05,
      "loss": 0.1314,
      "step": 988
    },
    {
      "epoch": 0.11446759259259259,
      "grad_norm": 0.6981813907623291,
      "learning_rate": 1.771064814814815e-05,
      "loss": 0.153,
      "step": 989
    },
    {
      "epoch": 0.11458333333333333,
      "grad_norm": 102.60330963134766,
      "learning_rate": 1.7708333333333335e-05,
      "loss": 1.9962,
      "step": 990
    },
    {
      "epoch": 0.11469907407407408,
      "grad_norm": 36.41157531738281,
      "learning_rate": 1.770601851851852e-05,
      "loss": 0.929,
      "step": 991
    },
    {
      "epoch": 0.11481481481481481,
      "grad_norm": 1.091583013534546,
      "learning_rate": 1.7703703703703706e-05,
      "loss": 0.1957,
      "step": 992
    },
    {
      "epoch": 0.11493055555555555,
      "grad_norm": 10.408203125,
      "learning_rate": 1.770138888888889e-05,
      "loss": 0.3898,
      "step": 993
    },
    {
      "epoch": 0.1150462962962963,
      "grad_norm": 1.5154145956039429,
      "learning_rate": 1.7699074074074075e-05,
      "loss": 0.2743,
      "step": 994
    },
    {
      "epoch": 0.11516203703703703,
      "grad_norm": 0.7583009600639343,
      "learning_rate": 1.769675925925926e-05,
      "loss": 0.1617,
      "step": 995
    },
    {
      "epoch": 0.11527777777777778,
      "grad_norm": 41.523067474365234,
      "learning_rate": 1.7694444444444447e-05,
      "loss": 0.9852,
      "step": 996
    },
    {
      "epoch": 0.11539351851851852,
      "grad_norm": 9.493661880493164,
      "learning_rate": 1.769212962962963e-05,
      "loss": 2.1212,
      "step": 997
    },
    {
      "epoch": 0.11550925925925926,
      "grad_norm": 1.0317301750183105,
      "learning_rate": 1.7689814814814815e-05,
      "loss": 0.2115,
      "step": 998
    },
    {
      "epoch": 0.115625,
      "grad_norm": 12.040968894958496,
      "learning_rate": 1.7687500000000003e-05,
      "loss": 0.2271,
      "step": 999
    },
    {
      "epoch": 0.11574074074074074,
      "grad_norm": 0.6992668509483337,
      "learning_rate": 1.7685185185185187e-05,
      "loss": 0.1421,
      "step": 1000
    },
    {
      "epoch": 0.11585648148148148,
      "grad_norm": 7.383852005004883,
      "learning_rate": 1.768287037037037e-05,
      "loss": 1.8453,
      "step": 1001
    },
    {
      "epoch": 0.11597222222222223,
      "grad_norm": 28.72216033935547,
      "learning_rate": 1.7680555555555556e-05,
      "loss": 1.5246,
      "step": 1002
    },
    {
      "epoch": 0.11608796296296296,
      "grad_norm": 0.8981139063835144,
      "learning_rate": 1.7678240740740743e-05,
      "loss": 0.186,
      "step": 1003
    },
    {
      "epoch": 0.1162037037037037,
      "grad_norm": 0.7187681198120117,
      "learning_rate": 1.7675925925925928e-05,
      "loss": 0.1371,
      "step": 1004
    },
    {
      "epoch": 0.11631944444444445,
      "grad_norm": 59.89971160888672,
      "learning_rate": 1.7673611111111112e-05,
      "loss": 0.7025,
      "step": 1005
    },
    {
      "epoch": 0.11643518518518518,
      "grad_norm": 1.3497885465621948,
      "learning_rate": 1.76712962962963e-05,
      "loss": 0.1683,
      "step": 1006
    },
    {
      "epoch": 0.11655092592592593,
      "grad_norm": 0.7541959881782532,
      "learning_rate": 1.766898148148148e-05,
      "loss": 0.1498,
      "step": 1007
    },
    {
      "epoch": 0.11666666666666667,
      "grad_norm": 1.4611858129501343,
      "learning_rate": 1.7666666666666668e-05,
      "loss": 0.1925,
      "step": 1008
    },
    {
      "epoch": 0.1167824074074074,
      "grad_norm": 10.018829345703125,
      "learning_rate": 1.7664351851851852e-05,
      "loss": 0.3497,
      "step": 1009
    },
    {
      "epoch": 0.11689814814814815,
      "grad_norm": 55.49643325805664,
      "learning_rate": 1.766203703703704e-05,
      "loss": 0.9096,
      "step": 1010
    },
    {
      "epoch": 0.11701388888888889,
      "grad_norm": 24.908842086791992,
      "learning_rate": 1.7659722222222224e-05,
      "loss": 1.8517,
      "step": 1011
    },
    {
      "epoch": 0.11712962962962963,
      "grad_norm": 0.6444964408874512,
      "learning_rate": 1.765740740740741e-05,
      "loss": 0.1228,
      "step": 1012
    },
    {
      "epoch": 0.11724537037037037,
      "grad_norm": 23.357995986938477,
      "learning_rate": 1.7655092592592593e-05,
      "loss": 1.6912,
      "step": 1013
    },
    {
      "epoch": 0.11736111111111111,
      "grad_norm": 0.6182068586349487,
      "learning_rate": 1.765277777777778e-05,
      "loss": 0.128,
      "step": 1014
    },
    {
      "epoch": 0.11747685185185185,
      "grad_norm": 1.0707318782806396,
      "learning_rate": 1.7650462962962965e-05,
      "loss": 0.1654,
      "step": 1015
    },
    {
      "epoch": 0.1175925925925926,
      "grad_norm": 4.836459159851074,
      "learning_rate": 1.764814814814815e-05,
      "loss": 0.3692,
      "step": 1016
    },
    {
      "epoch": 0.11770833333333333,
      "grad_norm": 20.33378028869629,
      "learning_rate": 1.7645833333333336e-05,
      "loss": 0.6156,
      "step": 1017
    },
    {
      "epoch": 0.11782407407407407,
      "grad_norm": 0.6897608637809753,
      "learning_rate": 1.764351851851852e-05,
      "loss": 0.146,
      "step": 1018
    },
    {
      "epoch": 0.11793981481481482,
      "grad_norm": 29.58427619934082,
      "learning_rate": 1.7641203703703705e-05,
      "loss": 1.6066,
      "step": 1019
    },
    {
      "epoch": 0.11805555555555555,
      "grad_norm": 0.9625833630561829,
      "learning_rate": 1.763888888888889e-05,
      "loss": 0.19,
      "step": 1020
    },
    {
      "epoch": 0.1181712962962963,
      "grad_norm": 0.7273276448249817,
      "learning_rate": 1.7636574074074077e-05,
      "loss": 0.1416,
      "step": 1021
    },
    {
      "epoch": 0.11828703703703704,
      "grad_norm": 0.9610259532928467,
      "learning_rate": 1.763425925925926e-05,
      "loss": 0.1969,
      "step": 1022
    },
    {
      "epoch": 0.11840277777777777,
      "grad_norm": 1.1077146530151367,
      "learning_rate": 1.7631944444444445e-05,
      "loss": 0.2131,
      "step": 1023
    },
    {
      "epoch": 0.11851851851851852,
      "grad_norm": 0.8702980875968933,
      "learning_rate": 1.7629629629629633e-05,
      "loss": 0.1726,
      "step": 1024
    },
    {
      "epoch": 0.11863425925925926,
      "grad_norm": 3.28367280960083,
      "learning_rate": 1.7627314814814814e-05,
      "loss": 0.2001,
      "step": 1025
    },
    {
      "epoch": 0.11875,
      "grad_norm": 23.270769119262695,
      "learning_rate": 1.7625e-05,
      "loss": 0.7356,
      "step": 1026
    },
    {
      "epoch": 0.11886574074074074,
      "grad_norm": 1.8888009786605835,
      "learning_rate": 1.7622685185185186e-05,
      "loss": 0.2282,
      "step": 1027
    },
    {
      "epoch": 0.11898148148148148,
      "grad_norm": 4.4594197273254395,
      "learning_rate": 1.7620370370370373e-05,
      "loss": 0.1901,
      "step": 1028
    },
    {
      "epoch": 0.11909722222222222,
      "grad_norm": 0.9783742427825928,
      "learning_rate": 1.7618055555555558e-05,
      "loss": 0.1467,
      "step": 1029
    },
    {
      "epoch": 0.11921296296296297,
      "grad_norm": 1.737605333328247,
      "learning_rate": 1.7615740740740742e-05,
      "loss": 0.1832,
      "step": 1030
    },
    {
      "epoch": 0.1193287037037037,
      "grad_norm": 8.629833221435547,
      "learning_rate": 1.7613425925925926e-05,
      "loss": 0.2658,
      "step": 1031
    },
    {
      "epoch": 0.11944444444444445,
      "grad_norm": 0.680389940738678,
      "learning_rate": 1.7611111111111114e-05,
      "loss": 0.1435,
      "step": 1032
    },
    {
      "epoch": 0.11956018518518519,
      "grad_norm": 0.6428085565567017,
      "learning_rate": 1.7608796296296298e-05,
      "loss": 0.1249,
      "step": 1033
    },
    {
      "epoch": 0.11967592592592592,
      "grad_norm": 0.8234307169914246,
      "learning_rate": 1.7606481481481482e-05,
      "loss": 0.1349,
      "step": 1034
    },
    {
      "epoch": 0.11979166666666667,
      "grad_norm": 4.080961227416992,
      "learning_rate": 1.760416666666667e-05,
      "loss": 0.2626,
      "step": 1035
    },
    {
      "epoch": 0.11990740740740741,
      "grad_norm": 1.1891096830368042,
      "learning_rate": 1.760185185185185e-05,
      "loss": 0.2112,
      "step": 1036
    },
    {
      "epoch": 0.12002314814814814,
      "grad_norm": 1.8112859725952148,
      "learning_rate": 1.759953703703704e-05,
      "loss": 0.2109,
      "step": 1037
    },
    {
      "epoch": 0.12013888888888889,
      "grad_norm": 0.8135457038879395,
      "learning_rate": 1.7597222222222223e-05,
      "loss": 0.1387,
      "step": 1038
    },
    {
      "epoch": 0.12025462962962963,
      "grad_norm": 1.248857855796814,
      "learning_rate": 1.759490740740741e-05,
      "loss": 0.1509,
      "step": 1039
    },
    {
      "epoch": 0.12037037037037036,
      "grad_norm": 56.769020080566406,
      "learning_rate": 1.7592592592592595e-05,
      "loss": 1.0055,
      "step": 1040
    },
    {
      "epoch": 0.12048611111111111,
      "grad_norm": 2.472090721130371,
      "learning_rate": 1.759027777777778e-05,
      "loss": 0.1819,
      "step": 1041
    },
    {
      "epoch": 0.12060185185185185,
      "grad_norm": 1.5390905141830444,
      "learning_rate": 1.7587962962962963e-05,
      "loss": 0.2281,
      "step": 1042
    },
    {
      "epoch": 0.1207175925925926,
      "grad_norm": 0.6304051876068115,
      "learning_rate": 1.7585648148148147e-05,
      "loss": 0.1196,
      "step": 1043
    },
    {
      "epoch": 0.12083333333333333,
      "grad_norm": 49.68705368041992,
      "learning_rate": 1.7583333333333335e-05,
      "loss": 0.951,
      "step": 1044
    },
    {
      "epoch": 0.12094907407407407,
      "grad_norm": 1.2513808012008667,
      "learning_rate": 1.758101851851852e-05,
      "loss": 0.1342,
      "step": 1045
    },
    {
      "epoch": 0.12106481481481482,
      "grad_norm": 35.202171325683594,
      "learning_rate": 1.7578703703703707e-05,
      "loss": 1.1069,
      "step": 1046
    },
    {
      "epoch": 0.12118055555555556,
      "grad_norm": 0.6919373869895935,
      "learning_rate": 1.757638888888889e-05,
      "loss": 0.105,
      "step": 1047
    },
    {
      "epoch": 0.12129629629629629,
      "grad_norm": 3.1008121967315674,
      "learning_rate": 1.7574074074074075e-05,
      "loss": 0.236,
      "step": 1048
    },
    {
      "epoch": 0.12141203703703704,
      "grad_norm": 2.3314192295074463,
      "learning_rate": 1.757175925925926e-05,
      "loss": 0.2404,
      "step": 1049
    },
    {
      "epoch": 0.12152777777777778,
      "grad_norm": 0.9047124981880188,
      "learning_rate": 1.7569444444444447e-05,
      "loss": 0.1878,
      "step": 1050
    },
    {
      "epoch": 0.12164351851851851,
      "grad_norm": 6.692124366760254,
      "learning_rate": 1.756712962962963e-05,
      "loss": 0.291,
      "step": 1051
    },
    {
      "epoch": 0.12175925925925926,
      "grad_norm": 1.9813685417175293,
      "learning_rate": 1.7564814814814816e-05,
      "loss": 0.2099,
      "step": 1052
    },
    {
      "epoch": 0.121875,
      "grad_norm": 0.8048036694526672,
      "learning_rate": 1.7562500000000003e-05,
      "loss": 0.147,
      "step": 1053
    },
    {
      "epoch": 0.12199074074074075,
      "grad_norm": 17.90067481994629,
      "learning_rate": 1.7560185185185184e-05,
      "loss": 1.5719,
      "step": 1054
    },
    {
      "epoch": 0.12210648148148148,
      "grad_norm": 1.5028002262115479,
      "learning_rate": 1.7557870370370372e-05,
      "loss": 0.205,
      "step": 1055
    },
    {
      "epoch": 0.12222222222222222,
      "grad_norm": 0.6941428184509277,
      "learning_rate": 1.7555555555555556e-05,
      "loss": 0.1344,
      "step": 1056
    },
    {
      "epoch": 0.12233796296296297,
      "grad_norm": 15.760710716247559,
      "learning_rate": 1.7553240740740744e-05,
      "loss": 1.984,
      "step": 1057
    },
    {
      "epoch": 0.1224537037037037,
      "grad_norm": 11.359654426574707,
      "learning_rate": 1.7550925925925928e-05,
      "loss": 0.3818,
      "step": 1058
    },
    {
      "epoch": 0.12256944444444444,
      "grad_norm": 0.7527684569358826,
      "learning_rate": 1.7548611111111112e-05,
      "loss": 0.1239,
      "step": 1059
    },
    {
      "epoch": 0.12268518518518519,
      "grad_norm": 0.8363924026489258,
      "learning_rate": 1.7546296296296297e-05,
      "loss": 0.166,
      "step": 1060
    },
    {
      "epoch": 0.12280092592592592,
      "grad_norm": 1.2430462837219238,
      "learning_rate": 1.754398148148148e-05,
      "loss": 0.1668,
      "step": 1061
    },
    {
      "epoch": 0.12291666666666666,
      "grad_norm": 1.2350751161575317,
      "learning_rate": 1.754166666666667e-05,
      "loss": 0.2367,
      "step": 1062
    },
    {
      "epoch": 0.12303240740740741,
      "grad_norm": 0.5087425708770752,
      "learning_rate": 1.7539351851851853e-05,
      "loss": 0.1084,
      "step": 1063
    },
    {
      "epoch": 0.12314814814814815,
      "grad_norm": 1.0294736623764038,
      "learning_rate": 1.753703703703704e-05,
      "loss": 0.1863,
      "step": 1064
    },
    {
      "epoch": 0.1232638888888889,
      "grad_norm": 9.900954246520996,
      "learning_rate": 1.7534722222222225e-05,
      "loss": 0.3047,
      "step": 1065
    },
    {
      "epoch": 0.12337962962962963,
      "grad_norm": 0.6668961048126221,
      "learning_rate": 1.753240740740741e-05,
      "loss": 0.1006,
      "step": 1066
    },
    {
      "epoch": 0.12349537037037037,
      "grad_norm": 25.031944274902344,
      "learning_rate": 1.7530092592592593e-05,
      "loss": 0.4371,
      "step": 1067
    },
    {
      "epoch": 0.12361111111111112,
      "grad_norm": 0.5197145938873291,
      "learning_rate": 1.752777777777778e-05,
      "loss": 0.1079,
      "step": 1068
    },
    {
      "epoch": 0.12372685185185185,
      "grad_norm": 27.19144058227539,
      "learning_rate": 1.7525462962962965e-05,
      "loss": 0.6315,
      "step": 1069
    },
    {
      "epoch": 0.12384259259259259,
      "grad_norm": 0.61760014295578,
      "learning_rate": 1.752314814814815e-05,
      "loss": 0.1251,
      "step": 1070
    },
    {
      "epoch": 0.12395833333333334,
      "grad_norm": 1.7324273586273193,
      "learning_rate": 1.7520833333333337e-05,
      "loss": 0.1875,
      "step": 1071
    },
    {
      "epoch": 0.12407407407407407,
      "grad_norm": 0.6102157831192017,
      "learning_rate": 1.7518518518518518e-05,
      "loss": 0.1133,
      "step": 1072
    },
    {
      "epoch": 0.12418981481481481,
      "grad_norm": 0.7440283894538879,
      "learning_rate": 1.7516203703703705e-05,
      "loss": 0.1488,
      "step": 1073
    },
    {
      "epoch": 0.12430555555555556,
      "grad_norm": 1.335437536239624,
      "learning_rate": 1.751388888888889e-05,
      "loss": 0.1273,
      "step": 1074
    },
    {
      "epoch": 0.1244212962962963,
      "grad_norm": 0.8111045956611633,
      "learning_rate": 1.7511574074074077e-05,
      "loss": 0.1029,
      "step": 1075
    },
    {
      "epoch": 0.12453703703703704,
      "grad_norm": 1.2430143356323242,
      "learning_rate": 1.750925925925926e-05,
      "loss": 0.2229,
      "step": 1076
    },
    {
      "epoch": 0.12465277777777778,
      "grad_norm": 3.656252861022949,
      "learning_rate": 1.7506944444444446e-05,
      "loss": 0.1646,
      "step": 1077
    },
    {
      "epoch": 0.12476851851851851,
      "grad_norm": 0.5530129671096802,
      "learning_rate": 1.750462962962963e-05,
      "loss": 0.1164,
      "step": 1078
    },
    {
      "epoch": 0.12488425925925926,
      "grad_norm": 0.8620915412902832,
      "learning_rate": 1.7502314814814814e-05,
      "loss": 0.1632,
      "step": 1079
    },
    {
      "epoch": 0.125,
      "grad_norm": 0.7849047780036926,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 0.1576,
      "step": 1080
    },
    {
      "epoch": 0.12511574074074075,
      "grad_norm": 5.421608924865723,
      "learning_rate": 1.7497685185185186e-05,
      "loss": 0.1944,
      "step": 1081
    },
    {
      "epoch": 0.12523148148148147,
      "grad_norm": 0.7711024880409241,
      "learning_rate": 1.7495370370370374e-05,
      "loss": 0.1577,
      "step": 1082
    },
    {
      "epoch": 0.12534722222222222,
      "grad_norm": 0.8526004552841187,
      "learning_rate": 1.7493055555555555e-05,
      "loss": 0.1731,
      "step": 1083
    },
    {
      "epoch": 0.12546296296296297,
      "grad_norm": 0.9744570255279541,
      "learning_rate": 1.7490740740740742e-05,
      "loss": 0.1604,
      "step": 1084
    },
    {
      "epoch": 0.1255787037037037,
      "grad_norm": 1.0887764692306519,
      "learning_rate": 1.7488425925925926e-05,
      "loss": 0.2087,
      "step": 1085
    },
    {
      "epoch": 0.12569444444444444,
      "grad_norm": 1.1900877952575684,
      "learning_rate": 1.7486111111111114e-05,
      "loss": 0.1894,
      "step": 1086
    },
    {
      "epoch": 0.1258101851851852,
      "grad_norm": 9.446630477905273,
      "learning_rate": 1.74837962962963e-05,
      "loss": 0.2656,
      "step": 1087
    },
    {
      "epoch": 0.1259259259259259,
      "grad_norm": 6.533483028411865,
      "learning_rate": 1.7481481481481483e-05,
      "loss": 0.2676,
      "step": 1088
    },
    {
      "epoch": 0.12604166666666666,
      "grad_norm": 0.6269466876983643,
      "learning_rate": 1.7479166666666667e-05,
      "loss": 0.1342,
      "step": 1089
    },
    {
      "epoch": 0.1261574074074074,
      "grad_norm": 89.48330688476562,
      "learning_rate": 1.747685185185185e-05,
      "loss": 0.9572,
      "step": 1090
    },
    {
      "epoch": 0.12627314814814813,
      "grad_norm": 6.037789344787598,
      "learning_rate": 1.747453703703704e-05,
      "loss": 0.2945,
      "step": 1091
    },
    {
      "epoch": 0.12638888888888888,
      "grad_norm": 8.570237159729004,
      "learning_rate": 1.7472222222222223e-05,
      "loss": 1.6823,
      "step": 1092
    },
    {
      "epoch": 0.12650462962962963,
      "grad_norm": 2.5412750244140625,
      "learning_rate": 1.746990740740741e-05,
      "loss": 0.2365,
      "step": 1093
    },
    {
      "epoch": 0.12662037037037038,
      "grad_norm": 41.61454391479492,
      "learning_rate": 1.7467592592592595e-05,
      "loss": 1.2815,
      "step": 1094
    },
    {
      "epoch": 0.1267361111111111,
      "grad_norm": 0.7507808804512024,
      "learning_rate": 1.746527777777778e-05,
      "loss": 0.1397,
      "step": 1095
    },
    {
      "epoch": 0.12685185185185185,
      "grad_norm": 0.8392742872238159,
      "learning_rate": 1.7462962962962963e-05,
      "loss": 0.1606,
      "step": 1096
    },
    {
      "epoch": 0.1269675925925926,
      "grad_norm": 2.4296152591705322,
      "learning_rate": 1.7460648148148148e-05,
      "loss": 0.1633,
      "step": 1097
    },
    {
      "epoch": 0.12708333333333333,
      "grad_norm": 0.960449755191803,
      "learning_rate": 1.7458333333333335e-05,
      "loss": 0.1618,
      "step": 1098
    },
    {
      "epoch": 0.12719907407407408,
      "grad_norm": 64.61299896240234,
      "learning_rate": 1.745601851851852e-05,
      "loss": 1.0304,
      "step": 1099
    },
    {
      "epoch": 0.12731481481481483,
      "grad_norm": 18.781814575195312,
      "learning_rate": 1.7453703703703707e-05,
      "loss": 1.4153,
      "step": 1100
    },
    {
      "epoch": 0.12743055555555555,
      "grad_norm": 1.572962760925293,
      "learning_rate": 1.7451388888888888e-05,
      "loss": 0.1944,
      "step": 1101
    },
    {
      "epoch": 0.1275462962962963,
      "grad_norm": 1.1104522943496704,
      "learning_rate": 1.7449074074074076e-05,
      "loss": 0.1872,
      "step": 1102
    },
    {
      "epoch": 0.12766203703703705,
      "grad_norm": 53.558040618896484,
      "learning_rate": 1.744675925925926e-05,
      "loss": 0.6045,
      "step": 1103
    },
    {
      "epoch": 0.12777777777777777,
      "grad_norm": 0.6808282732963562,
      "learning_rate": 1.7444444444444448e-05,
      "loss": 0.1312,
      "step": 1104
    },
    {
      "epoch": 0.12789351851851852,
      "grad_norm": 18.28362274169922,
      "learning_rate": 1.7442129629629632e-05,
      "loss": 0.2726,
      "step": 1105
    },
    {
      "epoch": 0.12800925925925927,
      "grad_norm": 0.7585245966911316,
      "learning_rate": 1.7439814814814816e-05,
      "loss": 0.152,
      "step": 1106
    },
    {
      "epoch": 0.128125,
      "grad_norm": 0.61888188123703,
      "learning_rate": 1.74375e-05,
      "loss": 0.1283,
      "step": 1107
    },
    {
      "epoch": 0.12824074074074074,
      "grad_norm": 0.7200965881347656,
      "learning_rate": 1.7435185185185185e-05,
      "loss": 0.1419,
      "step": 1108
    },
    {
      "epoch": 0.1283564814814815,
      "grad_norm": 8.473093032836914,
      "learning_rate": 1.7432870370370372e-05,
      "loss": 0.3031,
      "step": 1109
    },
    {
      "epoch": 0.1284722222222222,
      "grad_norm": 32.95233917236328,
      "learning_rate": 1.7430555555555556e-05,
      "loss": 0.2291,
      "step": 1110
    },
    {
      "epoch": 0.12858796296296296,
      "grad_norm": 0.6575505137443542,
      "learning_rate": 1.7428240740740744e-05,
      "loss": 0.1271,
      "step": 1111
    },
    {
      "epoch": 0.1287037037037037,
      "grad_norm": 0.6776220798492432,
      "learning_rate": 1.742592592592593e-05,
      "loss": 0.1394,
      "step": 1112
    },
    {
      "epoch": 0.12881944444444443,
      "grad_norm": 1.3698734045028687,
      "learning_rate": 1.7423611111111113e-05,
      "loss": 0.2411,
      "step": 1113
    },
    {
      "epoch": 0.12893518518518518,
      "grad_norm": 0.6651082634925842,
      "learning_rate": 1.7421296296296297e-05,
      "loss": 0.1245,
      "step": 1114
    },
    {
      "epoch": 0.12905092592592593,
      "grad_norm": 0.892808735370636,
      "learning_rate": 1.7418981481481485e-05,
      "loss": 0.1867,
      "step": 1115
    },
    {
      "epoch": 0.12916666666666668,
      "grad_norm": 2.2856709957122803,
      "learning_rate": 1.741666666666667e-05,
      "loss": 0.1715,
      "step": 1116
    },
    {
      "epoch": 0.1292824074074074,
      "grad_norm": 0.8710042238235474,
      "learning_rate": 1.7414351851851853e-05,
      "loss": 0.131,
      "step": 1117
    },
    {
      "epoch": 0.12939814814814815,
      "grad_norm": 1.0185273885726929,
      "learning_rate": 1.741203703703704e-05,
      "loss": 0.1605,
      "step": 1118
    },
    {
      "epoch": 0.1295138888888889,
      "grad_norm": 5.007508754730225,
      "learning_rate": 1.740972222222222e-05,
      "loss": 0.1994,
      "step": 1119
    },
    {
      "epoch": 0.12962962962962962,
      "grad_norm": 75.33366394042969,
      "learning_rate": 1.740740740740741e-05,
      "loss": 0.6091,
      "step": 1120
    },
    {
      "epoch": 0.12974537037037037,
      "grad_norm": 2.96757173538208,
      "learning_rate": 1.7405092592592593e-05,
      "loss": 0.2147,
      "step": 1121
    },
    {
      "epoch": 0.12986111111111112,
      "grad_norm": 37.97217559814453,
      "learning_rate": 1.740277777777778e-05,
      "loss": 1.1317,
      "step": 1122
    },
    {
      "epoch": 0.12997685185185184,
      "grad_norm": 0.5739865899085999,
      "learning_rate": 1.7400462962962965e-05,
      "loss": 0.1213,
      "step": 1123
    },
    {
      "epoch": 0.1300925925925926,
      "grad_norm": 39.13033676147461,
      "learning_rate": 1.739814814814815e-05,
      "loss": 0.4327,
      "step": 1124
    },
    {
      "epoch": 0.13020833333333334,
      "grad_norm": 1.1732635498046875,
      "learning_rate": 1.7395833333333334e-05,
      "loss": 0.1588,
      "step": 1125
    },
    {
      "epoch": 0.13032407407407406,
      "grad_norm": 59.69294357299805,
      "learning_rate": 1.7393518518518518e-05,
      "loss": 1.2064,
      "step": 1126
    },
    {
      "epoch": 0.13043981481481481,
      "grad_norm": 0.5496854782104492,
      "learning_rate": 1.7391203703703706e-05,
      "loss": 0.1112,
      "step": 1127
    },
    {
      "epoch": 0.13055555555555556,
      "grad_norm": 63.25535202026367,
      "learning_rate": 1.738888888888889e-05,
      "loss": 1.0994,
      "step": 1128
    },
    {
      "epoch": 0.13067129629629629,
      "grad_norm": 1.875038743019104,
      "learning_rate": 1.7386574074074078e-05,
      "loss": 0.2302,
      "step": 1129
    },
    {
      "epoch": 0.13078703703703703,
      "grad_norm": 17.472986221313477,
      "learning_rate": 1.738425925925926e-05,
      "loss": 0.3419,
      "step": 1130
    },
    {
      "epoch": 0.13090277777777778,
      "grad_norm": 39.98458480834961,
      "learning_rate": 1.7381944444444446e-05,
      "loss": 0.7165,
      "step": 1131
    },
    {
      "epoch": 0.1310185185185185,
      "grad_norm": 0.7868270874023438,
      "learning_rate": 1.737962962962963e-05,
      "loss": 0.1051,
      "step": 1132
    },
    {
      "epoch": 0.13113425925925926,
      "grad_norm": 23.405258178710938,
      "learning_rate": 1.7377314814814818e-05,
      "loss": 0.3957,
      "step": 1133
    },
    {
      "epoch": 0.13125,
      "grad_norm": 39.39350509643555,
      "learning_rate": 1.7375000000000002e-05,
      "loss": 1.6059,
      "step": 1134
    },
    {
      "epoch": 0.13136574074074073,
      "grad_norm": 0.6861304640769958,
      "learning_rate": 1.7372685185185186e-05,
      "loss": 0.1327,
      "step": 1135
    },
    {
      "epoch": 0.13148148148148148,
      "grad_norm": 0.6511560678482056,
      "learning_rate": 1.737037037037037e-05,
      "loss": 0.1329,
      "step": 1136
    },
    {
      "epoch": 0.13159722222222223,
      "grad_norm": 0.7735392451286316,
      "learning_rate": 1.7368055555555555e-05,
      "loss": 0.1557,
      "step": 1137
    },
    {
      "epoch": 0.13171296296296298,
      "grad_norm": 15.086469650268555,
      "learning_rate": 1.7365740740740743e-05,
      "loss": 0.2832,
      "step": 1138
    },
    {
      "epoch": 0.1318287037037037,
      "grad_norm": 1.0147886276245117,
      "learning_rate": 1.7363425925925927e-05,
      "loss": 0.1898,
      "step": 1139
    },
    {
      "epoch": 0.13194444444444445,
      "grad_norm": 54.25403594970703,
      "learning_rate": 1.7361111111111114e-05,
      "loss": 1.3589,
      "step": 1140
    },
    {
      "epoch": 0.1320601851851852,
      "grad_norm": 0.6358903050422668,
      "learning_rate": 1.73587962962963e-05,
      "loss": 0.1224,
      "step": 1141
    },
    {
      "epoch": 0.13217592592592592,
      "grad_norm": 0.5260244607925415,
      "learning_rate": 1.7356481481481483e-05,
      "loss": 0.1093,
      "step": 1142
    },
    {
      "epoch": 0.13229166666666667,
      "grad_norm": 0.9630309343338013,
      "learning_rate": 1.7354166666666667e-05,
      "loss": 0.187,
      "step": 1143
    },
    {
      "epoch": 0.13240740740740742,
      "grad_norm": 2.8700051307678223,
      "learning_rate": 1.735185185185185e-05,
      "loss": 0.283,
      "step": 1144
    },
    {
      "epoch": 0.13252314814814814,
      "grad_norm": 1.6445562839508057,
      "learning_rate": 1.734953703703704e-05,
      "loss": 0.169,
      "step": 1145
    },
    {
      "epoch": 0.1326388888888889,
      "grad_norm": 0.9342140555381775,
      "learning_rate": 1.7347222222222223e-05,
      "loss": 0.1925,
      "step": 1146
    },
    {
      "epoch": 0.13275462962962964,
      "grad_norm": 8.714900016784668,
      "learning_rate": 1.734490740740741e-05,
      "loss": 1.8548,
      "step": 1147
    },
    {
      "epoch": 0.13287037037037036,
      "grad_norm": 0.5227120518684387,
      "learning_rate": 1.7342592592592592e-05,
      "loss": 0.1065,
      "step": 1148
    },
    {
      "epoch": 0.1329861111111111,
      "grad_norm": 2.2272531986236572,
      "learning_rate": 1.734027777777778e-05,
      "loss": 0.2574,
      "step": 1149
    },
    {
      "epoch": 0.13310185185185186,
      "grad_norm": 38.575531005859375,
      "learning_rate": 1.7337962962962964e-05,
      "loss": 0.5921,
      "step": 1150
    },
    {
      "epoch": 0.13321759259259258,
      "grad_norm": 1.1305668354034424,
      "learning_rate": 1.733564814814815e-05,
      "loss": 0.16,
      "step": 1151
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 0.5639537572860718,
      "learning_rate": 1.7333333333333336e-05,
      "loss": 0.1176,
      "step": 1152
    },
    {
      "epoch": 0.13344907407407408,
      "grad_norm": 0.9414640069007874,
      "learning_rate": 1.733101851851852e-05,
      "loss": 0.1971,
      "step": 1153
    },
    {
      "epoch": 0.1335648148148148,
      "grad_norm": 1.5451396703720093,
      "learning_rate": 1.7328703703703704e-05,
      "loss": 0.2189,
      "step": 1154
    },
    {
      "epoch": 0.13368055555555555,
      "grad_norm": 0.7251037955284119,
      "learning_rate": 1.732638888888889e-05,
      "loss": 0.1198,
      "step": 1155
    },
    {
      "epoch": 0.1337962962962963,
      "grad_norm": 42.991573333740234,
      "learning_rate": 1.7324074074074076e-05,
      "loss": 0.4694,
      "step": 1156
    },
    {
      "epoch": 0.13391203703703702,
      "grad_norm": 1.582277774810791,
      "learning_rate": 1.732175925925926e-05,
      "loss": 0.1237,
      "step": 1157
    },
    {
      "epoch": 0.13402777777777777,
      "grad_norm": 0.5505164861679077,
      "learning_rate": 1.7319444444444448e-05,
      "loss": 0.1083,
      "step": 1158
    },
    {
      "epoch": 0.13414351851851852,
      "grad_norm": 7.649942398071289,
      "learning_rate": 1.7317129629629632e-05,
      "loss": 0.1982,
      "step": 1159
    },
    {
      "epoch": 0.13425925925925927,
      "grad_norm": 1.2839220762252808,
      "learning_rate": 1.7314814814814816e-05,
      "loss": 0.2317,
      "step": 1160
    },
    {
      "epoch": 0.134375,
      "grad_norm": 0.8459622859954834,
      "learning_rate": 1.73125e-05,
      "loss": 0.1472,
      "step": 1161
    },
    {
      "epoch": 0.13449074074074074,
      "grad_norm": 0.47924789786338806,
      "learning_rate": 1.7310185185185185e-05,
      "loss": 0.1019,
      "step": 1162
    },
    {
      "epoch": 0.1346064814814815,
      "grad_norm": 21.8752384185791,
      "learning_rate": 1.7307870370370373e-05,
      "loss": 0.4394,
      "step": 1163
    },
    {
      "epoch": 0.13472222222222222,
      "grad_norm": 0.7104778289794922,
      "learning_rate": 1.7305555555555557e-05,
      "loss": 0.126,
      "step": 1164
    },
    {
      "epoch": 0.13483796296296297,
      "grad_norm": 0.8014211058616638,
      "learning_rate": 1.7303240740740744e-05,
      "loss": 0.1439,
      "step": 1165
    },
    {
      "epoch": 0.13495370370370371,
      "grad_norm": 1.1757612228393555,
      "learning_rate": 1.7300925925925925e-05,
      "loss": 0.1619,
      "step": 1166
    },
    {
      "epoch": 0.13506944444444444,
      "grad_norm": 0.9956305623054504,
      "learning_rate": 1.7298611111111113e-05,
      "loss": 0.1529,
      "step": 1167
    },
    {
      "epoch": 0.13518518518518519,
      "grad_norm": 1.4299638271331787,
      "learning_rate": 1.7296296296296297e-05,
      "loss": 0.1651,
      "step": 1168
    },
    {
      "epoch": 0.13530092592592594,
      "grad_norm": 0.8498497605323792,
      "learning_rate": 1.7293981481481485e-05,
      "loss": 0.1482,
      "step": 1169
    },
    {
      "epoch": 0.13541666666666666,
      "grad_norm": 0.7661463618278503,
      "learning_rate": 1.729166666666667e-05,
      "loss": 0.1242,
      "step": 1170
    },
    {
      "epoch": 0.1355324074074074,
      "grad_norm": 0.585026741027832,
      "learning_rate": 1.7289351851851853e-05,
      "loss": 0.1219,
      "step": 1171
    },
    {
      "epoch": 0.13564814814814816,
      "grad_norm": 0.838255763053894,
      "learning_rate": 1.7287037037037038e-05,
      "loss": 0.1757,
      "step": 1172
    },
    {
      "epoch": 0.13576388888888888,
      "grad_norm": 0.6294052600860596,
      "learning_rate": 1.7284722222222222e-05,
      "loss": 0.1089,
      "step": 1173
    },
    {
      "epoch": 0.13587962962962963,
      "grad_norm": 17.322521209716797,
      "learning_rate": 1.728240740740741e-05,
      "loss": 0.389,
      "step": 1174
    },
    {
      "epoch": 0.13599537037037038,
      "grad_norm": 10.8330717086792,
      "learning_rate": 1.7280092592592594e-05,
      "loss": 0.2341,
      "step": 1175
    },
    {
      "epoch": 0.1361111111111111,
      "grad_norm": 0.6412835121154785,
      "learning_rate": 1.727777777777778e-05,
      "loss": 0.1301,
      "step": 1176
    },
    {
      "epoch": 0.13622685185185185,
      "grad_norm": 0.5395072102546692,
      "learning_rate": 1.7275462962962962e-05,
      "loss": 0.1114,
      "step": 1177
    },
    {
      "epoch": 0.1363425925925926,
      "grad_norm": 1.3272861242294312,
      "learning_rate": 1.727314814814815e-05,
      "loss": 0.1924,
      "step": 1178
    },
    {
      "epoch": 0.13645833333333332,
      "grad_norm": 5.110034465789795,
      "learning_rate": 1.7270833333333334e-05,
      "loss": 0.2273,
      "step": 1179
    },
    {
      "epoch": 0.13657407407407407,
      "grad_norm": 0.8734792470932007,
      "learning_rate": 1.726851851851852e-05,
      "loss": 0.1425,
      "step": 1180
    },
    {
      "epoch": 0.13668981481481482,
      "grad_norm": 0.5904064774513245,
      "learning_rate": 1.7266203703703706e-05,
      "loss": 0.1165,
      "step": 1181
    },
    {
      "epoch": 0.13680555555555557,
      "grad_norm": 0.7385920882225037,
      "learning_rate": 1.726388888888889e-05,
      "loss": 0.1342,
      "step": 1182
    },
    {
      "epoch": 0.1369212962962963,
      "grad_norm": 0.5526214838027954,
      "learning_rate": 1.7261574074074075e-05,
      "loss": 0.1008,
      "step": 1183
    },
    {
      "epoch": 0.13703703703703704,
      "grad_norm": 93.6206283569336,
      "learning_rate": 1.725925925925926e-05,
      "loss": 0.9193,
      "step": 1184
    },
    {
      "epoch": 0.1371527777777778,
      "grad_norm": 0.615971565246582,
      "learning_rate": 1.7256944444444446e-05,
      "loss": 0.1143,
      "step": 1185
    },
    {
      "epoch": 0.1372685185185185,
      "grad_norm": 35.83169937133789,
      "learning_rate": 1.725462962962963e-05,
      "loss": 1.8437,
      "step": 1186
    },
    {
      "epoch": 0.13738425925925926,
      "grad_norm": 7.794889450073242,
      "learning_rate": 1.7252314814814818e-05,
      "loss": 0.2845,
      "step": 1187
    },
    {
      "epoch": 0.1375,
      "grad_norm": 0.6259216666221619,
      "learning_rate": 1.7250000000000003e-05,
      "loss": 0.1278,
      "step": 1188
    },
    {
      "epoch": 0.13761574074074073,
      "grad_norm": 9.461052894592285,
      "learning_rate": 1.7247685185185187e-05,
      "loss": 0.3675,
      "step": 1189
    },
    {
      "epoch": 0.13773148148148148,
      "grad_norm": 53.96729278564453,
      "learning_rate": 1.724537037037037e-05,
      "loss": 0.4743,
      "step": 1190
    },
    {
      "epoch": 0.13784722222222223,
      "grad_norm": 0.9211532473564148,
      "learning_rate": 1.7243055555555555e-05,
      "loss": 0.2018,
      "step": 1191
    },
    {
      "epoch": 0.13796296296296295,
      "grad_norm": 24.322851181030273,
      "learning_rate": 1.7240740740740743e-05,
      "loss": 1.3904,
      "step": 1192
    },
    {
      "epoch": 0.1380787037037037,
      "grad_norm": 0.5216641426086426,
      "learning_rate": 1.7238425925925927e-05,
      "loss": 0.1092,
      "step": 1193
    },
    {
      "epoch": 0.13819444444444445,
      "grad_norm": 42.50702667236328,
      "learning_rate": 1.7236111111111115e-05,
      "loss": 0.5253,
      "step": 1194
    },
    {
      "epoch": 0.13831018518518517,
      "grad_norm": 7.830348491668701,
      "learning_rate": 1.7233796296296296e-05,
      "loss": 0.2502,
      "step": 1195
    },
    {
      "epoch": 0.13842592592592592,
      "grad_norm": 0.5971018671989441,
      "learning_rate": 1.7231481481481483e-05,
      "loss": 0.118,
      "step": 1196
    },
    {
      "epoch": 0.13854166666666667,
      "grad_norm": 0.5764947533607483,
      "learning_rate": 1.7229166666666668e-05,
      "loss": 0.0864,
      "step": 1197
    },
    {
      "epoch": 0.1386574074074074,
      "grad_norm": 0.5738339424133301,
      "learning_rate": 1.7226851851851852e-05,
      "loss": 0.1161,
      "step": 1198
    },
    {
      "epoch": 0.13877314814814815,
      "grad_norm": 0.9613949656486511,
      "learning_rate": 1.722453703703704e-05,
      "loss": 0.1238,
      "step": 1199
    },
    {
      "epoch": 0.1388888888888889,
      "grad_norm": 0.7011048197746277,
      "learning_rate": 1.7222222222222224e-05,
      "loss": 0.1405,
      "step": 1200
    },
    {
      "epoch": 0.13900462962962962,
      "grad_norm": 24.007427215576172,
      "learning_rate": 1.7219907407407408e-05,
      "loss": 0.7041,
      "step": 1201
    },
    {
      "epoch": 0.13912037037037037,
      "grad_norm": 6.755678176879883,
      "learning_rate": 1.7217592592592592e-05,
      "loss": 0.1716,
      "step": 1202
    },
    {
      "epoch": 0.13923611111111112,
      "grad_norm": 0.4767834544181824,
      "learning_rate": 1.721527777777778e-05,
      "loss": 0.0993,
      "step": 1203
    },
    {
      "epoch": 0.13935185185185187,
      "grad_norm": 0.7063002586364746,
      "learning_rate": 1.7212962962962964e-05,
      "loss": 0.1383,
      "step": 1204
    },
    {
      "epoch": 0.1394675925925926,
      "grad_norm": 54.57821273803711,
      "learning_rate": 1.7210648148148152e-05,
      "loss": 0.8586,
      "step": 1205
    },
    {
      "epoch": 0.13958333333333334,
      "grad_norm": 0.7503733038902283,
      "learning_rate": 1.7208333333333336e-05,
      "loss": 0.1551,
      "step": 1206
    },
    {
      "epoch": 0.1396990740740741,
      "grad_norm": 19.561660766601562,
      "learning_rate": 1.720601851851852e-05,
      "loss": 1.2992,
      "step": 1207
    },
    {
      "epoch": 0.1398148148148148,
      "grad_norm": 0.6386026740074158,
      "learning_rate": 1.7203703703703705e-05,
      "loss": 0.109,
      "step": 1208
    },
    {
      "epoch": 0.13993055555555556,
      "grad_norm": 0.6222509741783142,
      "learning_rate": 1.720138888888889e-05,
      "loss": 0.1229,
      "step": 1209
    },
    {
      "epoch": 0.1400462962962963,
      "grad_norm": 37.156307220458984,
      "learning_rate": 1.7199074074074076e-05,
      "loss": 0.4488,
      "step": 1210
    },
    {
      "epoch": 0.14016203703703703,
      "grad_norm": 1.9585295915603638,
      "learning_rate": 1.719675925925926e-05,
      "loss": 0.1654,
      "step": 1211
    },
    {
      "epoch": 0.14027777777777778,
      "grad_norm": 1.6132982969284058,
      "learning_rate": 1.7194444444444448e-05,
      "loss": 0.1547,
      "step": 1212
    },
    {
      "epoch": 0.14039351851851853,
      "grad_norm": 0.5380069017410278,
      "learning_rate": 1.719212962962963e-05,
      "loss": 0.109,
      "step": 1213
    },
    {
      "epoch": 0.14050925925925925,
      "grad_norm": 1.34706449508667,
      "learning_rate": 1.7189814814814817e-05,
      "loss": 0.203,
      "step": 1214
    },
    {
      "epoch": 0.140625,
      "grad_norm": 0.9192250370979309,
      "learning_rate": 1.71875e-05,
      "loss": 0.1484,
      "step": 1215
    },
    {
      "epoch": 0.14074074074074075,
      "grad_norm": 39.684234619140625,
      "learning_rate": 1.7185185185185185e-05,
      "loss": 1.2484,
      "step": 1216
    },
    {
      "epoch": 0.14085648148148147,
      "grad_norm": 0.6649503707885742,
      "learning_rate": 1.7182870370370373e-05,
      "loss": 0.1312,
      "step": 1217
    },
    {
      "epoch": 0.14097222222222222,
      "grad_norm": 48.44275665283203,
      "learning_rate": 1.7180555555555557e-05,
      "loss": 0.8814,
      "step": 1218
    },
    {
      "epoch": 0.14108796296296297,
      "grad_norm": 1.4517508745193481,
      "learning_rate": 1.717824074074074e-05,
      "loss": 0.1246,
      "step": 1219
    },
    {
      "epoch": 0.1412037037037037,
      "grad_norm": 4.825052738189697,
      "learning_rate": 1.7175925925925926e-05,
      "loss": 0.1482,
      "step": 1220
    },
    {
      "epoch": 0.14131944444444444,
      "grad_norm": 10.834261894226074,
      "learning_rate": 1.7173611111111113e-05,
      "loss": 2.105,
      "step": 1221
    },
    {
      "epoch": 0.1414351851851852,
      "grad_norm": 0.6140046715736389,
      "learning_rate": 1.7171296296296298e-05,
      "loss": 0.1133,
      "step": 1222
    },
    {
      "epoch": 0.1415509259259259,
      "grad_norm": 30.857219696044922,
      "learning_rate": 1.7168981481481485e-05,
      "loss": 1.1232,
      "step": 1223
    },
    {
      "epoch": 0.14166666666666666,
      "grad_norm": 0.4622340202331543,
      "learning_rate": 1.7166666666666666e-05,
      "loss": 0.0941,
      "step": 1224
    },
    {
      "epoch": 0.1417824074074074,
      "grad_norm": 1.3030242919921875,
      "learning_rate": 1.7164351851851854e-05,
      "loss": 0.1448,
      "step": 1225
    },
    {
      "epoch": 0.14189814814814813,
      "grad_norm": 23.527727127075195,
      "learning_rate": 1.7162037037037038e-05,
      "loss": 1.031,
      "step": 1226
    },
    {
      "epoch": 0.14201388888888888,
      "grad_norm": 0.7631290555000305,
      "learning_rate": 1.7159722222222222e-05,
      "loss": 0.157,
      "step": 1227
    },
    {
      "epoch": 0.14212962962962963,
      "grad_norm": 0.4728264808654785,
      "learning_rate": 1.715740740740741e-05,
      "loss": 0.0974,
      "step": 1228
    },
    {
      "epoch": 0.14224537037037038,
      "grad_norm": 0.6336820125579834,
      "learning_rate": 1.7155092592592594e-05,
      "loss": 0.0952,
      "step": 1229
    },
    {
      "epoch": 0.1423611111111111,
      "grad_norm": 1.1401958465576172,
      "learning_rate": 1.715277777777778e-05,
      "loss": 0.1759,
      "step": 1230
    },
    {
      "epoch": 0.14247685185185185,
      "grad_norm": 2.672837972640991,
      "learning_rate": 1.7150462962962963e-05,
      "loss": 0.1474,
      "step": 1231
    },
    {
      "epoch": 0.1425925925925926,
      "grad_norm": 0.7666226625442505,
      "learning_rate": 1.714814814814815e-05,
      "loss": 0.1582,
      "step": 1232
    },
    {
      "epoch": 0.14270833333333333,
      "grad_norm": 0.5375179052352905,
      "learning_rate": 1.7145833333333334e-05,
      "loss": 0.109,
      "step": 1233
    },
    {
      "epoch": 0.14282407407407408,
      "grad_norm": 0.49982497096061707,
      "learning_rate": 1.7143518518518522e-05,
      "loss": 0.1011,
      "step": 1234
    },
    {
      "epoch": 0.14293981481481483,
      "grad_norm": 0.7097680568695068,
      "learning_rate": 1.7141203703703706e-05,
      "loss": 0.1072,
      "step": 1235
    },
    {
      "epoch": 0.14305555555555555,
      "grad_norm": 45.344329833984375,
      "learning_rate": 1.713888888888889e-05,
      "loss": 0.5824,
      "step": 1236
    },
    {
      "epoch": 0.1431712962962963,
      "grad_norm": 24.024208068847656,
      "learning_rate": 1.7136574074074075e-05,
      "loss": 0.9082,
      "step": 1237
    },
    {
      "epoch": 0.14328703703703705,
      "grad_norm": 2.3156850337982178,
      "learning_rate": 1.713425925925926e-05,
      "loss": 0.1895,
      "step": 1238
    },
    {
      "epoch": 0.14340277777777777,
      "grad_norm": 0.7931790947914124,
      "learning_rate": 1.7131944444444447e-05,
      "loss": 0.1551,
      "step": 1239
    },
    {
      "epoch": 0.14351851851851852,
      "grad_norm": 0.7057933211326599,
      "learning_rate": 1.712962962962963e-05,
      "loss": 0.1459,
      "step": 1240
    },
    {
      "epoch": 0.14363425925925927,
      "grad_norm": 0.7626126408576965,
      "learning_rate": 1.712731481481482e-05,
      "loss": 0.1575,
      "step": 1241
    },
    {
      "epoch": 0.14375,
      "grad_norm": 0.737552285194397,
      "learning_rate": 1.7125e-05,
      "loss": 0.1208,
      "step": 1242
    },
    {
      "epoch": 0.14386574074074074,
      "grad_norm": 4.395932674407959,
      "learning_rate": 1.7122685185185187e-05,
      "loss": 0.1678,
      "step": 1243
    },
    {
      "epoch": 0.1439814814814815,
      "grad_norm": 0.4223648011684418,
      "learning_rate": 1.712037037037037e-05,
      "loss": 0.0881,
      "step": 1244
    },
    {
      "epoch": 0.1440972222222222,
      "grad_norm": 15.68244457244873,
      "learning_rate": 1.7118055555555556e-05,
      "loss": 0.2541,
      "step": 1245
    },
    {
      "epoch": 0.14421296296296296,
      "grad_norm": 2.588805913925171,
      "learning_rate": 1.7115740740740743e-05,
      "loss": 0.1839,
      "step": 1246
    },
    {
      "epoch": 0.1443287037037037,
      "grad_norm": 0.99701327085495,
      "learning_rate": 1.7113425925925928e-05,
      "loss": 0.1408,
      "step": 1247
    },
    {
      "epoch": 0.14444444444444443,
      "grad_norm": 6.291853427886963,
      "learning_rate": 1.7111111111111112e-05,
      "loss": 0.2864,
      "step": 1248
    },
    {
      "epoch": 0.14456018518518518,
      "grad_norm": 68.55278778076172,
      "learning_rate": 1.7108796296296296e-05,
      "loss": 0.9887,
      "step": 1249
    },
    {
      "epoch": 0.14467592592592593,
      "grad_norm": 0.45807331800460815,
      "learning_rate": 1.7106481481481484e-05,
      "loss": 0.0934,
      "step": 1250
    },
    {
      "epoch": 0.14479166666666668,
      "grad_norm": 0.7491921186447144,
      "learning_rate": 1.7104166666666668e-05,
      "loss": 0.1483,
      "step": 1251
    },
    {
      "epoch": 0.1449074074074074,
      "grad_norm": 0.484036922454834,
      "learning_rate": 1.7101851851851856e-05,
      "loss": 0.1023,
      "step": 1252
    },
    {
      "epoch": 0.14502314814814815,
      "grad_norm": 2.084500312805176,
      "learning_rate": 1.709953703703704e-05,
      "loss": 0.1267,
      "step": 1253
    },
    {
      "epoch": 0.1451388888888889,
      "grad_norm": 0.5890926122665405,
      "learning_rate": 1.7097222222222224e-05,
      "loss": 0.1241,
      "step": 1254
    },
    {
      "epoch": 0.14525462962962962,
      "grad_norm": 1.8527265787124634,
      "learning_rate": 1.709490740740741e-05,
      "loss": 0.1605,
      "step": 1255
    },
    {
      "epoch": 0.14537037037037037,
      "grad_norm": 0.5755482912063599,
      "learning_rate": 1.7092592592592593e-05,
      "loss": 0.0828,
      "step": 1256
    },
    {
      "epoch": 0.14548611111111112,
      "grad_norm": 2.5017569065093994,
      "learning_rate": 1.709027777777778e-05,
      "loss": 0.1266,
      "step": 1257
    },
    {
      "epoch": 0.14560185185185184,
      "grad_norm": 0.5581246018409729,
      "learning_rate": 1.7087962962962964e-05,
      "loss": 0.115,
      "step": 1258
    },
    {
      "epoch": 0.1457175925925926,
      "grad_norm": 30.46953582763672,
      "learning_rate": 1.7085648148148152e-05,
      "loss": 0.4228,
      "step": 1259
    },
    {
      "epoch": 0.14583333333333334,
      "grad_norm": 0.5421969294548035,
      "learning_rate": 1.7083333333333333e-05,
      "loss": 0.0805,
      "step": 1260
    },
    {
      "epoch": 0.14594907407407406,
      "grad_norm": 0.5159478783607483,
      "learning_rate": 1.708101851851852e-05,
      "loss": 0.0972,
      "step": 1261
    },
    {
      "epoch": 0.14606481481481481,
      "grad_norm": 42.771644592285156,
      "learning_rate": 1.7078703703703705e-05,
      "loss": 0.7065,
      "step": 1262
    },
    {
      "epoch": 0.14618055555555556,
      "grad_norm": 11.510847091674805,
      "learning_rate": 1.707638888888889e-05,
      "loss": 1.7891,
      "step": 1263
    },
    {
      "epoch": 0.14629629629629629,
      "grad_norm": 0.5356144905090332,
      "learning_rate": 1.7074074074074077e-05,
      "loss": 0.1091,
      "step": 1264
    },
    {
      "epoch": 0.14641203703703703,
      "grad_norm": 0.5569620728492737,
      "learning_rate": 1.707175925925926e-05,
      "loss": 0.117,
      "step": 1265
    },
    {
      "epoch": 0.14652777777777778,
      "grad_norm": 42.08749008178711,
      "learning_rate": 1.7069444444444445e-05,
      "loss": 1.385,
      "step": 1266
    },
    {
      "epoch": 0.1466435185185185,
      "grad_norm": 1.6802467107772827,
      "learning_rate": 1.706712962962963e-05,
      "loss": 0.1418,
      "step": 1267
    },
    {
      "epoch": 0.14675925925925926,
      "grad_norm": 0.7241824865341187,
      "learning_rate": 1.7064814814814817e-05,
      "loss": 0.1106,
      "step": 1268
    },
    {
      "epoch": 0.146875,
      "grad_norm": 60.21817398071289,
      "learning_rate": 1.70625e-05,
      "loss": 1.2655,
      "step": 1269
    },
    {
      "epoch": 0.14699074074074073,
      "grad_norm": 1.4721226692199707,
      "learning_rate": 1.706018518518519e-05,
      "loss": 0.1348,
      "step": 1270
    },
    {
      "epoch": 0.14710648148148148,
      "grad_norm": 3.0593111515045166,
      "learning_rate": 1.705787037037037e-05,
      "loss": 0.2086,
      "step": 1271
    },
    {
      "epoch": 0.14722222222222223,
      "grad_norm": 0.923922598361969,
      "learning_rate": 1.7055555555555558e-05,
      "loss": 0.143,
      "step": 1272
    },
    {
      "epoch": 0.14733796296296298,
      "grad_norm": 0.831537127494812,
      "learning_rate": 1.7053240740740742e-05,
      "loss": 0.1487,
      "step": 1273
    },
    {
      "epoch": 0.1474537037037037,
      "grad_norm": 5.612300872802734,
      "learning_rate": 1.7050925925925926e-05,
      "loss": 0.1955,
      "step": 1274
    },
    {
      "epoch": 0.14756944444444445,
      "grad_norm": 0.6009684801101685,
      "learning_rate": 1.7048611111111114e-05,
      "loss": 0.1153,
      "step": 1275
    },
    {
      "epoch": 0.1476851851851852,
      "grad_norm": 0.5036063194274902,
      "learning_rate": 1.7046296296296298e-05,
      "loss": 0.1024,
      "step": 1276
    },
    {
      "epoch": 0.14780092592592592,
      "grad_norm": 52.63575744628906,
      "learning_rate": 1.7043981481481482e-05,
      "loss": 0.8176,
      "step": 1277
    },
    {
      "epoch": 0.14791666666666667,
      "grad_norm": 5.930103778839111,
      "learning_rate": 1.7041666666666666e-05,
      "loss": 2.3994,
      "step": 1278
    },
    {
      "epoch": 0.14803240740740742,
      "grad_norm": 39.40670394897461,
      "learning_rate": 1.7039351851851854e-05,
      "loss": 0.5479,
      "step": 1279
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 0.5794031620025635,
      "learning_rate": 1.7037037037037038e-05,
      "loss": 0.1222,
      "step": 1280
    },
    {
      "epoch": 0.1482638888888889,
      "grad_norm": 2.4164352416992188,
      "learning_rate": 1.7034722222222223e-05,
      "loss": 0.2085,
      "step": 1281
    },
    {
      "epoch": 0.14837962962962964,
      "grad_norm": 1.2537903785705566,
      "learning_rate": 1.703240740740741e-05,
      "loss": 0.1253,
      "step": 1282
    },
    {
      "epoch": 0.14849537037037036,
      "grad_norm": 30.387502670288086,
      "learning_rate": 1.7030092592592594e-05,
      "loss": 0.5677,
      "step": 1283
    },
    {
      "epoch": 0.1486111111111111,
      "grad_norm": 0.7306548953056335,
      "learning_rate": 1.702777777777778e-05,
      "loss": 0.1317,
      "step": 1284
    },
    {
      "epoch": 0.14872685185185186,
      "grad_norm": 0.6676901578903198,
      "learning_rate": 1.7025462962962963e-05,
      "loss": 0.1022,
      "step": 1285
    },
    {
      "epoch": 0.14884259259259258,
      "grad_norm": 0.7469330430030823,
      "learning_rate": 1.702314814814815e-05,
      "loss": 0.1116,
      "step": 1286
    },
    {
      "epoch": 0.14895833333333333,
      "grad_norm": 0.5669462084770203,
      "learning_rate": 1.7020833333333335e-05,
      "loss": 0.1164,
      "step": 1287
    },
    {
      "epoch": 0.14907407407407408,
      "grad_norm": 7.716516494750977,
      "learning_rate": 1.7018518518518522e-05,
      "loss": 0.282,
      "step": 1288
    },
    {
      "epoch": 0.1491898148148148,
      "grad_norm": 0.9492666125297546,
      "learning_rate": 1.7016203703703703e-05,
      "loss": 0.1408,
      "step": 1289
    },
    {
      "epoch": 0.14930555555555555,
      "grad_norm": 0.995335578918457,
      "learning_rate": 1.701388888888889e-05,
      "loss": 0.1768,
      "step": 1290
    },
    {
      "epoch": 0.1494212962962963,
      "grad_norm": 0.4757225811481476,
      "learning_rate": 1.7011574074074075e-05,
      "loss": 0.0994,
      "step": 1291
    },
    {
      "epoch": 0.14953703703703702,
      "grad_norm": 0.4292718470096588,
      "learning_rate": 1.700925925925926e-05,
      "loss": 0.086,
      "step": 1292
    },
    {
      "epoch": 0.14965277777777777,
      "grad_norm": 6.045912265777588,
      "learning_rate": 1.7006944444444447e-05,
      "loss": 0.1532,
      "step": 1293
    },
    {
      "epoch": 0.14976851851851852,
      "grad_norm": 2.1007325649261475,
      "learning_rate": 1.700462962962963e-05,
      "loss": 0.1658,
      "step": 1294
    },
    {
      "epoch": 0.14988425925925927,
      "grad_norm": 0.4143636226654053,
      "learning_rate": 1.7002314814814816e-05,
      "loss": 0.0851,
      "step": 1295
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.724080502986908,
      "learning_rate": 1.7e-05,
      "loss": 0.1426,
      "step": 1296
    },
    {
      "epoch": 0.15011574074074074,
      "grad_norm": 43.41290283203125,
      "learning_rate": 1.6997685185185188e-05,
      "loss": 0.624,
      "step": 1297
    },
    {
      "epoch": 0.1502314814814815,
      "grad_norm": 0.5964533686637878,
      "learning_rate": 1.6995370370370372e-05,
      "loss": 0.1193,
      "step": 1298
    },
    {
      "epoch": 0.15034722222222222,
      "grad_norm": 0.9164064526557922,
      "learning_rate": 1.6993055555555556e-05,
      "loss": 0.1864,
      "step": 1299
    },
    {
      "epoch": 0.15046296296296297,
      "grad_norm": 0.5305410623550415,
      "learning_rate": 1.6990740740740744e-05,
      "loss": 0.0923,
      "step": 1300
    },
    {
      "epoch": 0.15057870370370371,
      "grad_norm": 0.8621518015861511,
      "learning_rate": 1.6988425925925928e-05,
      "loss": 0.1291,
      "step": 1301
    },
    {
      "epoch": 0.15069444444444444,
      "grad_norm": 1.2982959747314453,
      "learning_rate": 1.6986111111111112e-05,
      "loss": 0.1155,
      "step": 1302
    },
    {
      "epoch": 0.15081018518518519,
      "grad_norm": 0.8502871990203857,
      "learning_rate": 1.6983796296296296e-05,
      "loss": 0.1413,
      "step": 1303
    },
    {
      "epoch": 0.15092592592592594,
      "grad_norm": 2.661189317703247,
      "learning_rate": 1.6981481481481484e-05,
      "loss": 0.1815,
      "step": 1304
    },
    {
      "epoch": 0.15104166666666666,
      "grad_norm": 0.4864727854728699,
      "learning_rate": 1.6979166666666668e-05,
      "loss": 0.101,
      "step": 1305
    },
    {
      "epoch": 0.1511574074074074,
      "grad_norm": 74.1663589477539,
      "learning_rate": 1.6976851851851853e-05,
      "loss": 0.693,
      "step": 1306
    },
    {
      "epoch": 0.15127314814814816,
      "grad_norm": 0.6218557953834534,
      "learning_rate": 1.6974537037037037e-05,
      "loss": 0.1287,
      "step": 1307
    },
    {
      "epoch": 0.15138888888888888,
      "grad_norm": 1.4015690088272095,
      "learning_rate": 1.6972222222222224e-05,
      "loss": 0.2086,
      "step": 1308
    },
    {
      "epoch": 0.15150462962962963,
      "grad_norm": 53.11433410644531,
      "learning_rate": 1.696990740740741e-05,
      "loss": 1.3403,
      "step": 1309
    },
    {
      "epoch": 0.15162037037037038,
      "grad_norm": 46.82775115966797,
      "learning_rate": 1.6967592592592593e-05,
      "loss": 1.0765,
      "step": 1310
    },
    {
      "epoch": 0.1517361111111111,
      "grad_norm": 26.358484268188477,
      "learning_rate": 1.696527777777778e-05,
      "loss": 1.8592,
      "step": 1311
    },
    {
      "epoch": 0.15185185185185185,
      "grad_norm": 35.57199478149414,
      "learning_rate": 1.6962962962962965e-05,
      "loss": 1.9062,
      "step": 1312
    },
    {
      "epoch": 0.1519675925925926,
      "grad_norm": 1.1854439973831177,
      "learning_rate": 1.696064814814815e-05,
      "loss": 0.1723,
      "step": 1313
    },
    {
      "epoch": 0.15208333333333332,
      "grad_norm": 2.141010284423828,
      "learning_rate": 1.6958333333333333e-05,
      "loss": 0.1979,
      "step": 1314
    },
    {
      "epoch": 0.15219907407407407,
      "grad_norm": 0.6604018807411194,
      "learning_rate": 1.695601851851852e-05,
      "loss": 0.0975,
      "step": 1315
    },
    {
      "epoch": 0.15231481481481482,
      "grad_norm": 0.7230814099311829,
      "learning_rate": 1.6953703703703705e-05,
      "loss": 0.1268,
      "step": 1316
    },
    {
      "epoch": 0.15243055555555557,
      "grad_norm": 32.284461975097656,
      "learning_rate": 1.695138888888889e-05,
      "loss": 0.9032,
      "step": 1317
    },
    {
      "epoch": 0.1525462962962963,
      "grad_norm": 50.727821350097656,
      "learning_rate": 1.6949074074074074e-05,
      "loss": 2.3808,
      "step": 1318
    },
    {
      "epoch": 0.15266203703703704,
      "grad_norm": 0.48650261759757996,
      "learning_rate": 1.694675925925926e-05,
      "loss": 0.0962,
      "step": 1319
    },
    {
      "epoch": 0.1527777777777778,
      "grad_norm": 0.6390781998634338,
      "learning_rate": 1.6944444444444446e-05,
      "loss": 0.1276,
      "step": 1320
    },
    {
      "epoch": 0.1528935185185185,
      "grad_norm": 0.8917297124862671,
      "learning_rate": 1.694212962962963e-05,
      "loss": 0.1251,
      "step": 1321
    },
    {
      "epoch": 0.15300925925925926,
      "grad_norm": 0.4928402602672577,
      "learning_rate": 1.6939814814814817e-05,
      "loss": 0.102,
      "step": 1322
    },
    {
      "epoch": 0.153125,
      "grad_norm": 0.5825145840644836,
      "learning_rate": 1.6937500000000002e-05,
      "loss": 0.1114,
      "step": 1323
    },
    {
      "epoch": 0.15324074074074073,
      "grad_norm": 59.117095947265625,
      "learning_rate": 1.6935185185185186e-05,
      "loss": 0.6417,
      "step": 1324
    },
    {
      "epoch": 0.15335648148148148,
      "grad_norm": 7.986137390136719,
      "learning_rate": 1.693287037037037e-05,
      "loss": 2.2117,
      "step": 1325
    },
    {
      "epoch": 0.15347222222222223,
      "grad_norm": 21.285480499267578,
      "learning_rate": 1.6930555555555558e-05,
      "loss": 1.2711,
      "step": 1326
    },
    {
      "epoch": 0.15358796296296295,
      "grad_norm": 5.651910781860352,
      "learning_rate": 1.6928240740740742e-05,
      "loss": 2.0947,
      "step": 1327
    },
    {
      "epoch": 0.1537037037037037,
      "grad_norm": 0.5590596795082092,
      "learning_rate": 1.6925925925925926e-05,
      "loss": 0.1134,
      "step": 1328
    },
    {
      "epoch": 0.15381944444444445,
      "grad_norm": 0.6534568071365356,
      "learning_rate": 1.6923611111111114e-05,
      "loss": 0.1321,
      "step": 1329
    },
    {
      "epoch": 0.15393518518518517,
      "grad_norm": 0.4428369402885437,
      "learning_rate": 1.6921296296296298e-05,
      "loss": 0.0927,
      "step": 1330
    },
    {
      "epoch": 0.15405092592592592,
      "grad_norm": 0.712679386138916,
      "learning_rate": 1.6918981481481483e-05,
      "loss": 0.0939,
      "step": 1331
    },
    {
      "epoch": 0.15416666666666667,
      "grad_norm": 0.7629351019859314,
      "learning_rate": 1.6916666666666667e-05,
      "loss": 0.1301,
      "step": 1332
    },
    {
      "epoch": 0.1542824074074074,
      "grad_norm": 0.5054827928543091,
      "learning_rate": 1.6914351851851854e-05,
      "loss": 0.0917,
      "step": 1333
    },
    {
      "epoch": 0.15439814814814815,
      "grad_norm": 0.5162893533706665,
      "learning_rate": 1.691203703703704e-05,
      "loss": 0.1066,
      "step": 1334
    },
    {
      "epoch": 0.1545138888888889,
      "grad_norm": 1.0447643995285034,
      "learning_rate": 1.6909722222222223e-05,
      "loss": 0.1695,
      "step": 1335
    },
    {
      "epoch": 0.15462962962962962,
      "grad_norm": 0.8138928413391113,
      "learning_rate": 1.6907407407407407e-05,
      "loss": 0.0932,
      "step": 1336
    },
    {
      "epoch": 0.15474537037037037,
      "grad_norm": 0.7940433025360107,
      "learning_rate": 1.6905092592592595e-05,
      "loss": 0.1404,
      "step": 1337
    },
    {
      "epoch": 0.15486111111111112,
      "grad_norm": 0.42261803150177,
      "learning_rate": 1.690277777777778e-05,
      "loss": 0.0874,
      "step": 1338
    },
    {
      "epoch": 0.15497685185185187,
      "grad_norm": 0.5239015817642212,
      "learning_rate": 1.6900462962962963e-05,
      "loss": 0.1069,
      "step": 1339
    },
    {
      "epoch": 0.1550925925925926,
      "grad_norm": 107.35285186767578,
      "learning_rate": 1.689814814814815e-05,
      "loss": 0.6956,
      "step": 1340
    },
    {
      "epoch": 0.15520833333333334,
      "grad_norm": 0.5898963809013367,
      "learning_rate": 1.6895833333333335e-05,
      "loss": 0.1202,
      "step": 1341
    },
    {
      "epoch": 0.1553240740740741,
      "grad_norm": 1.0886255502700806,
      "learning_rate": 1.689351851851852e-05,
      "loss": 0.1854,
      "step": 1342
    },
    {
      "epoch": 0.1554398148148148,
      "grad_norm": 0.6784188747406006,
      "learning_rate": 1.6891203703703704e-05,
      "loss": 0.1185,
      "step": 1343
    },
    {
      "epoch": 0.15555555555555556,
      "grad_norm": 1.1186084747314453,
      "learning_rate": 1.688888888888889e-05,
      "loss": 0.1338,
      "step": 1344
    },
    {
      "epoch": 0.1556712962962963,
      "grad_norm": 0.6771164536476135,
      "learning_rate": 1.6886574074074076e-05,
      "loss": 0.1194,
      "step": 1345
    },
    {
      "epoch": 0.15578703703703703,
      "grad_norm": 0.6811144948005676,
      "learning_rate": 1.688425925925926e-05,
      "loss": 0.1319,
      "step": 1346
    },
    {
      "epoch": 0.15590277777777778,
      "grad_norm": 2.769477605819702,
      "learning_rate": 1.6881944444444447e-05,
      "loss": 0.1354,
      "step": 1347
    },
    {
      "epoch": 0.15601851851851853,
      "grad_norm": 0.4614216387271881,
      "learning_rate": 1.6879629629629632e-05,
      "loss": 0.0939,
      "step": 1348
    },
    {
      "epoch": 0.15613425925925925,
      "grad_norm": 36.19319152832031,
      "learning_rate": 1.6877314814814816e-05,
      "loss": 1.8807,
      "step": 1349
    },
    {
      "epoch": 0.15625,
      "grad_norm": 0.7542181611061096,
      "learning_rate": 1.6875e-05,
      "loss": 0.1557,
      "step": 1350
    },
    {
      "epoch": 0.15636574074074075,
      "grad_norm": 1.1663566827774048,
      "learning_rate": 1.6872685185185188e-05,
      "loss": 0.1521,
      "step": 1351
    },
    {
      "epoch": 0.15648148148148147,
      "grad_norm": 5.96358585357666,
      "learning_rate": 1.6870370370370372e-05,
      "loss": 2.2043,
      "step": 1352
    },
    {
      "epoch": 0.15659722222222222,
      "grad_norm": 32.57609558105469,
      "learning_rate": 1.6868055555555556e-05,
      "loss": 1.516,
      "step": 1353
    },
    {
      "epoch": 0.15671296296296297,
      "grad_norm": 2.8856585025787354,
      "learning_rate": 1.686574074074074e-05,
      "loss": 0.1425,
      "step": 1354
    },
    {
      "epoch": 0.1568287037037037,
      "grad_norm": 0.6864995360374451,
      "learning_rate": 1.6863425925925928e-05,
      "loss": 0.1125,
      "step": 1355
    },
    {
      "epoch": 0.15694444444444444,
      "grad_norm": 9.282934188842773,
      "learning_rate": 1.6861111111111112e-05,
      "loss": 0.1398,
      "step": 1356
    },
    {
      "epoch": 0.1570601851851852,
      "grad_norm": 0.6030538082122803,
      "learning_rate": 1.6858796296296297e-05,
      "loss": 0.1138,
      "step": 1357
    },
    {
      "epoch": 0.1571759259259259,
      "grad_norm": 1.6092357635498047,
      "learning_rate": 1.6856481481481484e-05,
      "loss": 0.1159,
      "step": 1358
    },
    {
      "epoch": 0.15729166666666666,
      "grad_norm": 0.4334900677204132,
      "learning_rate": 1.685416666666667e-05,
      "loss": 0.0892,
      "step": 1359
    },
    {
      "epoch": 0.1574074074074074,
      "grad_norm": 0.38754430413246155,
      "learning_rate": 1.6851851851851853e-05,
      "loss": 0.0787,
      "step": 1360
    },
    {
      "epoch": 0.15752314814814813,
      "grad_norm": 0.5357356071472168,
      "learning_rate": 1.6849537037037037e-05,
      "loss": 0.1075,
      "step": 1361
    },
    {
      "epoch": 0.15763888888888888,
      "grad_norm": 2.279113292694092,
      "learning_rate": 1.6847222222222225e-05,
      "loss": 0.1751,
      "step": 1362
    },
    {
      "epoch": 0.15775462962962963,
      "grad_norm": 0.6328717470169067,
      "learning_rate": 1.684490740740741e-05,
      "loss": 0.1217,
      "step": 1363
    },
    {
      "epoch": 0.15787037037037038,
      "grad_norm": 0.5708826780319214,
      "learning_rate": 1.6842592592592593e-05,
      "loss": 0.0853,
      "step": 1364
    },
    {
      "epoch": 0.1579861111111111,
      "grad_norm": 0.4884289503097534,
      "learning_rate": 1.6840277777777778e-05,
      "loss": 0.0944,
      "step": 1365
    },
    {
      "epoch": 0.15810185185185185,
      "grad_norm": 0.6880401968955994,
      "learning_rate": 1.6837962962962965e-05,
      "loss": 0.1224,
      "step": 1366
    },
    {
      "epoch": 0.1582175925925926,
      "grad_norm": 2.114147186279297,
      "learning_rate": 1.683564814814815e-05,
      "loss": 0.1389,
      "step": 1367
    },
    {
      "epoch": 0.15833333333333333,
      "grad_norm": 0.5973123908042908,
      "learning_rate": 1.6833333333333334e-05,
      "loss": 0.1135,
      "step": 1368
    },
    {
      "epoch": 0.15844907407407408,
      "grad_norm": 1.029417634010315,
      "learning_rate": 1.683101851851852e-05,
      "loss": 0.1551,
      "step": 1369
    },
    {
      "epoch": 0.15856481481481483,
      "grad_norm": 0.5733185410499573,
      "learning_rate": 1.6828703703703706e-05,
      "loss": 0.1135,
      "step": 1370
    },
    {
      "epoch": 0.15868055555555555,
      "grad_norm": 0.37488028407096863,
      "learning_rate": 1.682638888888889e-05,
      "loss": 0.0771,
      "step": 1371
    },
    {
      "epoch": 0.1587962962962963,
      "grad_norm": 0.6590829491615295,
      "learning_rate": 1.6824074074074074e-05,
      "loss": 0.1083,
      "step": 1372
    },
    {
      "epoch": 0.15891203703703705,
      "grad_norm": 47.569984436035156,
      "learning_rate": 1.6821759259259262e-05,
      "loss": 0.5844,
      "step": 1373
    },
    {
      "epoch": 0.15902777777777777,
      "grad_norm": 0.40568745136260986,
      "learning_rate": 1.6819444444444446e-05,
      "loss": 0.0831,
      "step": 1374
    },
    {
      "epoch": 0.15914351851851852,
      "grad_norm": 0.5892946124076843,
      "learning_rate": 1.681712962962963e-05,
      "loss": 0.1079,
      "step": 1375
    },
    {
      "epoch": 0.15925925925925927,
      "grad_norm": 0.8283354043960571,
      "learning_rate": 1.6814814814814818e-05,
      "loss": 0.1235,
      "step": 1376
    },
    {
      "epoch": 0.159375,
      "grad_norm": 39.922019958496094,
      "learning_rate": 1.6812500000000002e-05,
      "loss": 1.4533,
      "step": 1377
    },
    {
      "epoch": 0.15949074074074074,
      "grad_norm": 1.1406495571136475,
      "learning_rate": 1.6810185185185186e-05,
      "loss": 0.1766,
      "step": 1378
    },
    {
      "epoch": 0.1596064814814815,
      "grad_norm": 0.3657708466053009,
      "learning_rate": 1.680787037037037e-05,
      "loss": 0.0761,
      "step": 1379
    },
    {
      "epoch": 0.1597222222222222,
      "grad_norm": 0.694503128528595,
      "learning_rate": 1.6805555555555558e-05,
      "loss": 0.1174,
      "step": 1380
    },
    {
      "epoch": 0.15983796296296296,
      "grad_norm": 4.971059799194336,
      "learning_rate": 1.6803240740740742e-05,
      "loss": 0.1495,
      "step": 1381
    },
    {
      "epoch": 0.1599537037037037,
      "grad_norm": 1.6790143251419067,
      "learning_rate": 1.6800925925925927e-05,
      "loss": 0.154,
      "step": 1382
    },
    {
      "epoch": 0.16006944444444443,
      "grad_norm": 0.5204731822013855,
      "learning_rate": 1.679861111111111e-05,
      "loss": 0.1033,
      "step": 1383
    },
    {
      "epoch": 0.16018518518518518,
      "grad_norm": 0.46919023990631104,
      "learning_rate": 1.67962962962963e-05,
      "loss": 0.0975,
      "step": 1384
    },
    {
      "epoch": 0.16030092592592593,
      "grad_norm": 28.901142120361328,
      "learning_rate": 1.6793981481481483e-05,
      "loss": 0.2672,
      "step": 1385
    },
    {
      "epoch": 0.16041666666666668,
      "grad_norm": 1.862169861793518,
      "learning_rate": 1.6791666666666667e-05,
      "loss": 0.1257,
      "step": 1386
    },
    {
      "epoch": 0.1605324074074074,
      "grad_norm": 0.8388568758964539,
      "learning_rate": 1.6789351851851855e-05,
      "loss": 0.1259,
      "step": 1387
    },
    {
      "epoch": 0.16064814814814815,
      "grad_norm": 1.0328001976013184,
      "learning_rate": 1.678703703703704e-05,
      "loss": 0.1277,
      "step": 1388
    },
    {
      "epoch": 0.1607638888888889,
      "grad_norm": 0.519517719745636,
      "learning_rate": 1.6784722222222223e-05,
      "loss": 0.1025,
      "step": 1389
    },
    {
      "epoch": 0.16087962962962962,
      "grad_norm": 0.48043984174728394,
      "learning_rate": 1.6782407407407408e-05,
      "loss": 0.0906,
      "step": 1390
    },
    {
      "epoch": 0.16099537037037037,
      "grad_norm": 0.4050443172454834,
      "learning_rate": 1.6780092592592595e-05,
      "loss": 0.0846,
      "step": 1391
    },
    {
      "epoch": 0.16111111111111112,
      "grad_norm": 1.21794593334198,
      "learning_rate": 1.677777777777778e-05,
      "loss": 0.1338,
      "step": 1392
    },
    {
      "epoch": 0.16122685185185184,
      "grad_norm": 0.5195878744125366,
      "learning_rate": 1.6775462962962964e-05,
      "loss": 0.1002,
      "step": 1393
    },
    {
      "epoch": 0.1613425925925926,
      "grad_norm": 0.5044280886650085,
      "learning_rate": 1.677314814814815e-05,
      "loss": 0.0906,
      "step": 1394
    },
    {
      "epoch": 0.16145833333333334,
      "grad_norm": 0.5447962284088135,
      "learning_rate": 1.6770833333333336e-05,
      "loss": 0.0802,
      "step": 1395
    },
    {
      "epoch": 0.16157407407407406,
      "grad_norm": 0.48873645067214966,
      "learning_rate": 1.676851851851852e-05,
      "loss": 0.0983,
      "step": 1396
    },
    {
      "epoch": 0.16168981481481481,
      "grad_norm": 0.37002620100975037,
      "learning_rate": 1.6766203703703704e-05,
      "loss": 0.0756,
      "step": 1397
    },
    {
      "epoch": 0.16180555555555556,
      "grad_norm": 0.640963077545166,
      "learning_rate": 1.676388888888889e-05,
      "loss": 0.1099,
      "step": 1398
    },
    {
      "epoch": 0.16192129629629629,
      "grad_norm": 0.4164193868637085,
      "learning_rate": 1.6761574074074076e-05,
      "loss": 0.0857,
      "step": 1399
    },
    {
      "epoch": 0.16203703703703703,
      "grad_norm": 0.7350620627403259,
      "learning_rate": 1.675925925925926e-05,
      "loss": 0.1211,
      "step": 1400
    },
    {
      "epoch": 0.16215277777777778,
      "grad_norm": 61.2726936340332,
      "learning_rate": 1.6756944444444444e-05,
      "loss": 0.8733,
      "step": 1401
    },
    {
      "epoch": 0.1622685185185185,
      "grad_norm": 23.64468002319336,
      "learning_rate": 1.6754629629629632e-05,
      "loss": 1.5479,
      "step": 1402
    },
    {
      "epoch": 0.16238425925925926,
      "grad_norm": 0.3882817029953003,
      "learning_rate": 1.6752314814814816e-05,
      "loss": 0.0791,
      "step": 1403
    },
    {
      "epoch": 0.1625,
      "grad_norm": 1.6730523109436035,
      "learning_rate": 1.675e-05,
      "loss": 0.1517,
      "step": 1404
    },
    {
      "epoch": 0.16261574074074073,
      "grad_norm": 0.40555712580680847,
      "learning_rate": 1.6747685185185188e-05,
      "loss": 0.0825,
      "step": 1405
    },
    {
      "epoch": 0.16273148148148148,
      "grad_norm": 0.7982099056243896,
      "learning_rate": 1.674537037037037e-05,
      "loss": 0.1326,
      "step": 1406
    },
    {
      "epoch": 0.16284722222222223,
      "grad_norm": 0.4566103219985962,
      "learning_rate": 1.6743055555555557e-05,
      "loss": 0.0683,
      "step": 1407
    },
    {
      "epoch": 0.16296296296296298,
      "grad_norm": 0.7768300771713257,
      "learning_rate": 1.674074074074074e-05,
      "loss": 0.0933,
      "step": 1408
    },
    {
      "epoch": 0.1630787037037037,
      "grad_norm": 0.7762245535850525,
      "learning_rate": 1.673842592592593e-05,
      "loss": 0.1557,
      "step": 1409
    },
    {
      "epoch": 0.16319444444444445,
      "grad_norm": 0.7063434720039368,
      "learning_rate": 1.6736111111111113e-05,
      "loss": 0.0872,
      "step": 1410
    },
    {
      "epoch": 0.1633101851851852,
      "grad_norm": 0.43108856678009033,
      "learning_rate": 1.6733796296296297e-05,
      "loss": 0.087,
      "step": 1411
    },
    {
      "epoch": 0.16342592592592592,
      "grad_norm": 0.5863997340202332,
      "learning_rate": 1.673148148148148e-05,
      "loss": 0.1077,
      "step": 1412
    },
    {
      "epoch": 0.16354166666666667,
      "grad_norm": 34.24040222167969,
      "learning_rate": 1.672916666666667e-05,
      "loss": 1.8329,
      "step": 1413
    },
    {
      "epoch": 0.16365740740740742,
      "grad_norm": 1.0271434783935547,
      "learning_rate": 1.6726851851851853e-05,
      "loss": 0.14,
      "step": 1414
    },
    {
      "epoch": 0.16377314814814814,
      "grad_norm": 0.4103582203388214,
      "learning_rate": 1.6724537037037037e-05,
      "loss": 0.0769,
      "step": 1415
    },
    {
      "epoch": 0.1638888888888889,
      "grad_norm": 1.368704915046692,
      "learning_rate": 1.6722222222222225e-05,
      "loss": 0.1285,
      "step": 1416
    },
    {
      "epoch": 0.16400462962962964,
      "grad_norm": 0.40139901638031006,
      "learning_rate": 1.671990740740741e-05,
      "loss": 0.0839,
      "step": 1417
    },
    {
      "epoch": 0.16412037037037036,
      "grad_norm": 6.630648612976074,
      "learning_rate": 1.6717592592592594e-05,
      "loss": 2.4687,
      "step": 1418
    },
    {
      "epoch": 0.1642361111111111,
      "grad_norm": 4.801620006561279,
      "learning_rate": 1.6715277777777778e-05,
      "loss": 0.1522,
      "step": 1419
    },
    {
      "epoch": 0.16435185185185186,
      "grad_norm": 47.87490463256836,
      "learning_rate": 1.6712962962962966e-05,
      "loss": 0.913,
      "step": 1420
    },
    {
      "epoch": 0.16446759259259258,
      "grad_norm": 0.5308407545089722,
      "learning_rate": 1.671064814814815e-05,
      "loss": 0.0786,
      "step": 1421
    },
    {
      "epoch": 0.16458333333333333,
      "grad_norm": 1.3653876781463623,
      "learning_rate": 1.6708333333333334e-05,
      "loss": 0.1179,
      "step": 1422
    },
    {
      "epoch": 0.16469907407407408,
      "grad_norm": 58.47648239135742,
      "learning_rate": 1.670601851851852e-05,
      "loss": 0.7396,
      "step": 1423
    },
    {
      "epoch": 0.1648148148148148,
      "grad_norm": 0.44654762744903564,
      "learning_rate": 1.6703703703703703e-05,
      "loss": 0.0881,
      "step": 1424
    },
    {
      "epoch": 0.16493055555555555,
      "grad_norm": 26.055131912231445,
      "learning_rate": 1.670138888888889e-05,
      "loss": 0.3661,
      "step": 1425
    },
    {
      "epoch": 0.1650462962962963,
      "grad_norm": 0.5852336883544922,
      "learning_rate": 1.6699074074074074e-05,
      "loss": 0.0817,
      "step": 1426
    },
    {
      "epoch": 0.16516203703703702,
      "grad_norm": 0.4257463812828064,
      "learning_rate": 1.6696759259259262e-05,
      "loss": 0.0762,
      "step": 1427
    },
    {
      "epoch": 0.16527777777777777,
      "grad_norm": 0.49610286951065063,
      "learning_rate": 1.6694444444444446e-05,
      "loss": 0.1004,
      "step": 1428
    },
    {
      "epoch": 0.16539351851851852,
      "grad_norm": 0.5400031805038452,
      "learning_rate": 1.669212962962963e-05,
      "loss": 0.1019,
      "step": 1429
    },
    {
      "epoch": 0.16550925925925927,
      "grad_norm": 0.9786663055419922,
      "learning_rate": 1.6689814814814815e-05,
      "loss": 0.1098,
      "step": 1430
    },
    {
      "epoch": 0.165625,
      "grad_norm": 0.6350681185722351,
      "learning_rate": 1.6687500000000002e-05,
      "loss": 0.1234,
      "step": 1431
    },
    {
      "epoch": 0.16574074074074074,
      "grad_norm": 0.8985000848770142,
      "learning_rate": 1.6685185185185187e-05,
      "loss": 0.1246,
      "step": 1432
    },
    {
      "epoch": 0.1658564814814815,
      "grad_norm": 0.708152711391449,
      "learning_rate": 1.668287037037037e-05,
      "loss": 0.133,
      "step": 1433
    },
    {
      "epoch": 0.16597222222222222,
      "grad_norm": 1.9212439060211182,
      "learning_rate": 1.668055555555556e-05,
      "loss": 0.1554,
      "step": 1434
    },
    {
      "epoch": 0.16608796296296297,
      "grad_norm": 0.6953497529029846,
      "learning_rate": 1.6678240740740743e-05,
      "loss": 0.1012,
      "step": 1435
    },
    {
      "epoch": 0.16620370370370371,
      "grad_norm": 0.605236291885376,
      "learning_rate": 1.6675925925925927e-05,
      "loss": 0.1174,
      "step": 1436
    },
    {
      "epoch": 0.16631944444444444,
      "grad_norm": 11.705324172973633,
      "learning_rate": 1.667361111111111e-05,
      "loss": 0.2789,
      "step": 1437
    },
    {
      "epoch": 0.16643518518518519,
      "grad_norm": 0.46408095955848694,
      "learning_rate": 1.66712962962963e-05,
      "loss": 0.0952,
      "step": 1438
    },
    {
      "epoch": 0.16655092592592594,
      "grad_norm": 0.468760222196579,
      "learning_rate": 1.6668981481481483e-05,
      "loss": 0.0976,
      "step": 1439
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 0.637511134147644,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.1288,
      "step": 1440
    },
    {
      "epoch": 0.1667824074074074,
      "grad_norm": 56.25462341308594,
      "learning_rate": 1.6664351851851852e-05,
      "loss": 0.702,
      "step": 1441
    },
    {
      "epoch": 0.16689814814814816,
      "grad_norm": 0.5596020817756653,
      "learning_rate": 1.666203703703704e-05,
      "loss": 0.1014,
      "step": 1442
    },
    {
      "epoch": 0.16701388888888888,
      "grad_norm": 62.17982864379883,
      "learning_rate": 1.6659722222222224e-05,
      "loss": 0.6516,
      "step": 1443
    },
    {
      "epoch": 0.16712962962962963,
      "grad_norm": 14.096769332885742,
      "learning_rate": 1.6657407407407408e-05,
      "loss": 0.1362,
      "step": 1444
    },
    {
      "epoch": 0.16724537037037038,
      "grad_norm": 0.706809401512146,
      "learning_rate": 1.6655092592592595e-05,
      "loss": 0.1062,
      "step": 1445
    },
    {
      "epoch": 0.1673611111111111,
      "grad_norm": 0.5017099380493164,
      "learning_rate": 1.665277777777778e-05,
      "loss": 0.0858,
      "step": 1446
    },
    {
      "epoch": 0.16747685185185185,
      "grad_norm": 0.4746799170970917,
      "learning_rate": 1.6650462962962964e-05,
      "loss": 0.0964,
      "step": 1447
    },
    {
      "epoch": 0.1675925925925926,
      "grad_norm": 25.173954010009766,
      "learning_rate": 1.6648148148148148e-05,
      "loss": 1.9709,
      "step": 1448
    },
    {
      "epoch": 0.16770833333333332,
      "grad_norm": 0.47504547238349915,
      "learning_rate": 1.6645833333333336e-05,
      "loss": 0.0865,
      "step": 1449
    },
    {
      "epoch": 0.16782407407407407,
      "grad_norm": 0.44405367970466614,
      "learning_rate": 1.664351851851852e-05,
      "loss": 0.0873,
      "step": 1450
    },
    {
      "epoch": 0.16793981481481482,
      "grad_norm": 0.45151978731155396,
      "learning_rate": 1.6641203703703704e-05,
      "loss": 0.0889,
      "step": 1451
    },
    {
      "epoch": 0.16805555555555557,
      "grad_norm": 0.3911741375923157,
      "learning_rate": 1.6638888888888892e-05,
      "loss": 0.0797,
      "step": 1452
    },
    {
      "epoch": 0.1681712962962963,
      "grad_norm": 0.43729543685913086,
      "learning_rate": 1.6636574074074073e-05,
      "loss": 0.0896,
      "step": 1453
    },
    {
      "epoch": 0.16828703703703704,
      "grad_norm": 18.16485595703125,
      "learning_rate": 1.663425925925926e-05,
      "loss": 2.0895,
      "step": 1454
    },
    {
      "epoch": 0.1684027777777778,
      "grad_norm": 0.6328445672988892,
      "learning_rate": 1.6631944444444445e-05,
      "loss": 0.1313,
      "step": 1455
    },
    {
      "epoch": 0.1685185185185185,
      "grad_norm": 0.48611828684806824,
      "learning_rate": 1.6629629629629632e-05,
      "loss": 0.0987,
      "step": 1456
    },
    {
      "epoch": 0.16863425925925926,
      "grad_norm": 0.45473599433898926,
      "learning_rate": 1.6627314814814817e-05,
      "loss": 0.0818,
      "step": 1457
    },
    {
      "epoch": 0.16875,
      "grad_norm": 0.4314460754394531,
      "learning_rate": 1.6625e-05,
      "loss": 0.0881,
      "step": 1458
    },
    {
      "epoch": 0.16886574074074073,
      "grad_norm": 7.847153186798096,
      "learning_rate": 1.6622685185185185e-05,
      "loss": 0.1886,
      "step": 1459
    },
    {
      "epoch": 0.16898148148148148,
      "grad_norm": 0.500472366809845,
      "learning_rate": 1.6620370370370373e-05,
      "loss": 0.0741,
      "step": 1460
    },
    {
      "epoch": 0.16909722222222223,
      "grad_norm": 0.3586178719997406,
      "learning_rate": 1.6618055555555557e-05,
      "loss": 0.0708,
      "step": 1461
    },
    {
      "epoch": 0.16921296296296295,
      "grad_norm": 7.472122669219971,
      "learning_rate": 1.661574074074074e-05,
      "loss": 2.1138,
      "step": 1462
    },
    {
      "epoch": 0.1693287037037037,
      "grad_norm": 0.4961380660533905,
      "learning_rate": 1.661342592592593e-05,
      "loss": 0.101,
      "step": 1463
    },
    {
      "epoch": 0.16944444444444445,
      "grad_norm": 0.47905704379081726,
      "learning_rate": 1.6611111111111113e-05,
      "loss": 0.092,
      "step": 1464
    },
    {
      "epoch": 0.16956018518518517,
      "grad_norm": 0.5850228071212769,
      "learning_rate": 1.6608796296296297e-05,
      "loss": 0.1002,
      "step": 1465
    },
    {
      "epoch": 0.16967592592592592,
      "grad_norm": 0.47874313592910767,
      "learning_rate": 1.6606481481481482e-05,
      "loss": 0.0925,
      "step": 1466
    },
    {
      "epoch": 0.16979166666666667,
      "grad_norm": 0.6370421648025513,
      "learning_rate": 1.660416666666667e-05,
      "loss": 0.1356,
      "step": 1467
    },
    {
      "epoch": 0.1699074074074074,
      "grad_norm": 0.8315095901489258,
      "learning_rate": 1.6601851851851854e-05,
      "loss": 0.1181,
      "step": 1468
    },
    {
      "epoch": 0.17002314814814815,
      "grad_norm": 0.4187791347503662,
      "learning_rate": 1.6599537037037038e-05,
      "loss": 0.083,
      "step": 1469
    },
    {
      "epoch": 0.1701388888888889,
      "grad_norm": 74.07523345947266,
      "learning_rate": 1.6597222222222225e-05,
      "loss": 0.4969,
      "step": 1470
    },
    {
      "epoch": 0.17025462962962962,
      "grad_norm": 0.35748180747032166,
      "learning_rate": 1.6594907407407406e-05,
      "loss": 0.0733,
      "step": 1471
    },
    {
      "epoch": 0.17037037037037037,
      "grad_norm": 0.5241669416427612,
      "learning_rate": 1.6592592592592594e-05,
      "loss": 0.0996,
      "step": 1472
    },
    {
      "epoch": 0.17048611111111112,
      "grad_norm": 19.35196304321289,
      "learning_rate": 1.6590277777777778e-05,
      "loss": 2.0103,
      "step": 1473
    },
    {
      "epoch": 0.17060185185185187,
      "grad_norm": 0.580564022064209,
      "learning_rate": 1.6587962962962966e-05,
      "loss": 0.1189,
      "step": 1474
    },
    {
      "epoch": 0.1707175925925926,
      "grad_norm": 0.4901677072048187,
      "learning_rate": 1.658564814814815e-05,
      "loss": 0.0982,
      "step": 1475
    },
    {
      "epoch": 0.17083333333333334,
      "grad_norm": 0.5900912880897522,
      "learning_rate": 1.6583333333333334e-05,
      "loss": 0.0866,
      "step": 1476
    },
    {
      "epoch": 0.1709490740740741,
      "grad_norm": 8.509077072143555,
      "learning_rate": 1.658101851851852e-05,
      "loss": 0.2596,
      "step": 1477
    },
    {
      "epoch": 0.1710648148148148,
      "grad_norm": 0.41581401228904724,
      "learning_rate": 1.6578703703703706e-05,
      "loss": 0.0818,
      "step": 1478
    },
    {
      "epoch": 0.17118055555555556,
      "grad_norm": 0.7975077033042908,
      "learning_rate": 1.657638888888889e-05,
      "loss": 0.1226,
      "step": 1479
    },
    {
      "epoch": 0.1712962962962963,
      "grad_norm": 49.53564453125,
      "learning_rate": 1.6574074074074075e-05,
      "loss": 1.2787,
      "step": 1480
    },
    {
      "epoch": 0.17141203703703703,
      "grad_norm": 0.37311074137687683,
      "learning_rate": 1.6571759259259262e-05,
      "loss": 0.0775,
      "step": 1481
    },
    {
      "epoch": 0.17152777777777778,
      "grad_norm": 0.6138924360275269,
      "learning_rate": 1.6569444444444447e-05,
      "loss": 0.131,
      "step": 1482
    },
    {
      "epoch": 0.17164351851851853,
      "grad_norm": 0.5893270373344421,
      "learning_rate": 1.656712962962963e-05,
      "loss": 0.1245,
      "step": 1483
    },
    {
      "epoch": 0.17175925925925925,
      "grad_norm": 0.4592694044113159,
      "learning_rate": 1.6564814814814815e-05,
      "loss": 0.0806,
      "step": 1484
    },
    {
      "epoch": 0.171875,
      "grad_norm": 0.6285556554794312,
      "learning_rate": 1.6562500000000003e-05,
      "loss": 0.1281,
      "step": 1485
    },
    {
      "epoch": 0.17199074074074075,
      "grad_norm": 0.5605008602142334,
      "learning_rate": 1.6560185185185187e-05,
      "loss": 0.1188,
      "step": 1486
    },
    {
      "epoch": 0.17210648148148147,
      "grad_norm": 0.43049290776252747,
      "learning_rate": 1.655787037037037e-05,
      "loss": 0.0874,
      "step": 1487
    },
    {
      "epoch": 0.17222222222222222,
      "grad_norm": 1.8421382904052734,
      "learning_rate": 1.6555555555555556e-05,
      "loss": 0.114,
      "step": 1488
    },
    {
      "epoch": 0.17233796296296297,
      "grad_norm": 0.5791694521903992,
      "learning_rate": 1.655324074074074e-05,
      "loss": 0.1132,
      "step": 1489
    },
    {
      "epoch": 0.1724537037037037,
      "grad_norm": 0.7689018845558167,
      "learning_rate": 1.6550925925925927e-05,
      "loss": 0.1353,
      "step": 1490
    },
    {
      "epoch": 0.17256944444444444,
      "grad_norm": 0.5107265710830688,
      "learning_rate": 1.654861111111111e-05,
      "loss": 0.0746,
      "step": 1491
    },
    {
      "epoch": 0.1726851851851852,
      "grad_norm": 0.45873576402664185,
      "learning_rate": 1.65462962962963e-05,
      "loss": 0.0911,
      "step": 1492
    },
    {
      "epoch": 0.1728009259259259,
      "grad_norm": 21.477354049682617,
      "learning_rate": 1.6543981481481484e-05,
      "loss": 1.8781,
      "step": 1493
    },
    {
      "epoch": 0.17291666666666666,
      "grad_norm": 0.49497631192207336,
      "learning_rate": 1.6541666666666668e-05,
      "loss": 0.0912,
      "step": 1494
    },
    {
      "epoch": 0.1730324074074074,
      "grad_norm": 0.3576248586177826,
      "learning_rate": 1.6539351851851852e-05,
      "loss": 0.0729,
      "step": 1495
    },
    {
      "epoch": 0.17314814814814813,
      "grad_norm": 74.6554946899414,
      "learning_rate": 1.653703703703704e-05,
      "loss": 0.9072,
      "step": 1496
    },
    {
      "epoch": 0.17326388888888888,
      "grad_norm": 0.4894084334373474,
      "learning_rate": 1.6534722222222224e-05,
      "loss": 0.0988,
      "step": 1497
    },
    {
      "epoch": 0.17337962962962963,
      "grad_norm": 0.5504280924797058,
      "learning_rate": 1.6532407407407408e-05,
      "loss": 0.0841,
      "step": 1498
    },
    {
      "epoch": 0.17349537037037038,
      "grad_norm": 25.755184173583984,
      "learning_rate": 1.6530092592592596e-05,
      "loss": 1.4435,
      "step": 1499
    },
    {
      "epoch": 0.1736111111111111,
      "grad_norm": 0.40763184428215027,
      "learning_rate": 1.6527777777777777e-05,
      "loss": 0.0839,
      "step": 1500
    },
    {
      "epoch": 0.17372685185185185,
      "grad_norm": 17.22577667236328,
      "learning_rate": 1.6525462962962964e-05,
      "loss": 1.9448,
      "step": 1501
    },
    {
      "epoch": 0.1738425925925926,
      "grad_norm": 0.5475454330444336,
      "learning_rate": 1.652314814814815e-05,
      "loss": 0.1138,
      "step": 1502
    },
    {
      "epoch": 0.17395833333333333,
      "grad_norm": 0.3656418025493622,
      "learning_rate": 1.6520833333333336e-05,
      "loss": 0.0696,
      "step": 1503
    },
    {
      "epoch": 0.17407407407407408,
      "grad_norm": 0.5717176795005798,
      "learning_rate": 1.651851851851852e-05,
      "loss": 0.12,
      "step": 1504
    },
    {
      "epoch": 0.17418981481481483,
      "grad_norm": 0.6489041447639465,
      "learning_rate": 1.6516203703703705e-05,
      "loss": 0.1102,
      "step": 1505
    },
    {
      "epoch": 0.17430555555555555,
      "grad_norm": 0.3538295030593872,
      "learning_rate": 1.651388888888889e-05,
      "loss": 0.0727,
      "step": 1506
    },
    {
      "epoch": 0.1744212962962963,
      "grad_norm": 0.4268083870410919,
      "learning_rate": 1.6511574074074073e-05,
      "loss": 0.0855,
      "step": 1507
    },
    {
      "epoch": 0.17453703703703705,
      "grad_norm": 9.088518142700195,
      "learning_rate": 1.650925925925926e-05,
      "loss": 0.193,
      "step": 1508
    },
    {
      "epoch": 0.17465277777777777,
      "grad_norm": 0.40606990456581116,
      "learning_rate": 1.6506944444444445e-05,
      "loss": 0.0824,
      "step": 1509
    },
    {
      "epoch": 0.17476851851851852,
      "grad_norm": 0.4339962899684906,
      "learning_rate": 1.6504629629629633e-05,
      "loss": 0.0875,
      "step": 1510
    },
    {
      "epoch": 0.17488425925925927,
      "grad_norm": 0.3934924602508545,
      "learning_rate": 1.6502314814814817e-05,
      "loss": 0.0751,
      "step": 1511
    },
    {
      "epoch": 0.175,
      "grad_norm": 0.3960837423801422,
      "learning_rate": 1.65e-05,
      "loss": 0.0737,
      "step": 1512
    },
    {
      "epoch": 0.17511574074074074,
      "grad_norm": 0.6814449429512024,
      "learning_rate": 1.6497685185185186e-05,
      "loss": 0.1034,
      "step": 1513
    },
    {
      "epoch": 0.1752314814814815,
      "grad_norm": 45.0511589050293,
      "learning_rate": 1.6495370370370373e-05,
      "loss": 0.3671,
      "step": 1514
    },
    {
      "epoch": 0.1753472222222222,
      "grad_norm": 20.18846893310547,
      "learning_rate": 1.6493055555555557e-05,
      "loss": 1.9398,
      "step": 1515
    },
    {
      "epoch": 0.17546296296296296,
      "grad_norm": 0.46422311663627625,
      "learning_rate": 1.649074074074074e-05,
      "loss": 0.0689,
      "step": 1516
    },
    {
      "epoch": 0.1755787037037037,
      "grad_norm": 0.3574329912662506,
      "learning_rate": 1.648842592592593e-05,
      "loss": 0.0739,
      "step": 1517
    },
    {
      "epoch": 0.17569444444444443,
      "grad_norm": 53.626346588134766,
      "learning_rate": 1.648611111111111e-05,
      "loss": 1.3928,
      "step": 1518
    },
    {
      "epoch": 0.17581018518518518,
      "grad_norm": 0.5554035305976868,
      "learning_rate": 1.6483796296296298e-05,
      "loss": 0.1108,
      "step": 1519
    },
    {
      "epoch": 0.17592592592592593,
      "grad_norm": 0.42129456996917725,
      "learning_rate": 1.6481481481481482e-05,
      "loss": 0.0766,
      "step": 1520
    },
    {
      "epoch": 0.17604166666666668,
      "grad_norm": 0.4432266354560852,
      "learning_rate": 1.647916666666667e-05,
      "loss": 0.0863,
      "step": 1521
    },
    {
      "epoch": 0.1761574074074074,
      "grad_norm": 0.5858285427093506,
      "learning_rate": 1.6476851851851854e-05,
      "loss": 0.1025,
      "step": 1522
    },
    {
      "epoch": 0.17627314814814815,
      "grad_norm": 0.31893762946128845,
      "learning_rate": 1.6474537037037038e-05,
      "loss": 0.0655,
      "step": 1523
    },
    {
      "epoch": 0.1763888888888889,
      "grad_norm": 0.46828046441078186,
      "learning_rate": 1.6472222222222222e-05,
      "loss": 0.0926,
      "step": 1524
    },
    {
      "epoch": 0.17650462962962962,
      "grad_norm": 4.50414514541626,
      "learning_rate": 1.6469907407407407e-05,
      "loss": 0.1407,
      "step": 1525
    },
    {
      "epoch": 0.17662037037037037,
      "grad_norm": 0.4331710636615753,
      "learning_rate": 1.6467592592592594e-05,
      "loss": 0.0808,
      "step": 1526
    },
    {
      "epoch": 0.17673611111111112,
      "grad_norm": 0.5708003640174866,
      "learning_rate": 1.646527777777778e-05,
      "loss": 0.1109,
      "step": 1527
    },
    {
      "epoch": 0.17685185185185184,
      "grad_norm": 0.4819090664386749,
      "learning_rate": 1.6462962962962966e-05,
      "loss": 0.0942,
      "step": 1528
    },
    {
      "epoch": 0.1769675925925926,
      "grad_norm": 66.0534896850586,
      "learning_rate": 1.646064814814815e-05,
      "loss": 0.6318,
      "step": 1529
    },
    {
      "epoch": 0.17708333333333334,
      "grad_norm": 0.542474091053009,
      "learning_rate": 1.6458333333333335e-05,
      "loss": 0.0815,
      "step": 1530
    },
    {
      "epoch": 0.17719907407407406,
      "grad_norm": 2.1436879634857178,
      "learning_rate": 1.645601851851852e-05,
      "loss": 0.1148,
      "step": 1531
    },
    {
      "epoch": 0.17731481481481481,
      "grad_norm": 0.6275568604469299,
      "learning_rate": 1.6453703703703707e-05,
      "loss": 0.0942,
      "step": 1532
    },
    {
      "epoch": 0.17743055555555556,
      "grad_norm": 0.5456817746162415,
      "learning_rate": 1.645138888888889e-05,
      "loss": 0.109,
      "step": 1533
    },
    {
      "epoch": 0.17754629629629629,
      "grad_norm": 0.33097729086875916,
      "learning_rate": 1.6449074074074075e-05,
      "loss": 0.0678,
      "step": 1534
    },
    {
      "epoch": 0.17766203703703703,
      "grad_norm": 4.324406623840332,
      "learning_rate": 1.644675925925926e-05,
      "loss": 0.1191,
      "step": 1535
    },
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 0.44124919176101685,
      "learning_rate": 1.6444444444444444e-05,
      "loss": 0.0915,
      "step": 1536
    },
    {
      "epoch": 0.1778935185185185,
      "grad_norm": 0.7720770835876465,
      "learning_rate": 1.644212962962963e-05,
      "loss": 0.1187,
      "step": 1537
    },
    {
      "epoch": 0.17800925925925926,
      "grad_norm": 0.4745997190475464,
      "learning_rate": 1.6439814814814815e-05,
      "loss": 0.0977,
      "step": 1538
    },
    {
      "epoch": 0.178125,
      "grad_norm": 0.43044281005859375,
      "learning_rate": 1.6437500000000003e-05,
      "loss": 0.0897,
      "step": 1539
    },
    {
      "epoch": 0.17824074074074073,
      "grad_norm": 0.3404152989387512,
      "learning_rate": 1.6435185185185187e-05,
      "loss": 0.0655,
      "step": 1540
    },
    {
      "epoch": 0.17835648148148148,
      "grad_norm": 42.62639236450195,
      "learning_rate": 1.643287037037037e-05,
      "loss": 0.3403,
      "step": 1541
    },
    {
      "epoch": 0.17847222222222223,
      "grad_norm": 0.34366121888160706,
      "learning_rate": 1.6430555555555556e-05,
      "loss": 0.0702,
      "step": 1542
    },
    {
      "epoch": 0.17858796296296298,
      "grad_norm": 0.5398053526878357,
      "learning_rate": 1.642824074074074e-05,
      "loss": 0.0997,
      "step": 1543
    },
    {
      "epoch": 0.1787037037037037,
      "grad_norm": 87.50973510742188,
      "learning_rate": 1.6425925925925928e-05,
      "loss": 1.0697,
      "step": 1544
    },
    {
      "epoch": 0.17881944444444445,
      "grad_norm": 36.640281677246094,
      "learning_rate": 1.6423611111111112e-05,
      "loss": 1.0277,
      "step": 1545
    },
    {
      "epoch": 0.1789351851851852,
      "grad_norm": 30.485416412353516,
      "learning_rate": 1.64212962962963e-05,
      "loss": 0.711,
      "step": 1546
    },
    {
      "epoch": 0.17905092592592592,
      "grad_norm": 0.5483535528182983,
      "learning_rate": 1.641898148148148e-05,
      "loss": 0.1036,
      "step": 1547
    },
    {
      "epoch": 0.17916666666666667,
      "grad_norm": 7.98876428604126,
      "learning_rate": 1.6416666666666668e-05,
      "loss": 0.1604,
      "step": 1548
    },
    {
      "epoch": 0.17928240740740742,
      "grad_norm": 7.288412094116211,
      "learning_rate": 1.6414351851851852e-05,
      "loss": 2.1177,
      "step": 1549
    },
    {
      "epoch": 0.17939814814814814,
      "grad_norm": 0.4559575915336609,
      "learning_rate": 1.641203703703704e-05,
      "loss": 0.0902,
      "step": 1550
    },
    {
      "epoch": 0.1795138888888889,
      "grad_norm": 0.33527156710624695,
      "learning_rate": 1.6409722222222224e-05,
      "loss": 0.069,
      "step": 1551
    },
    {
      "epoch": 0.17962962962962964,
      "grad_norm": 0.4636611342430115,
      "learning_rate": 1.640740740740741e-05,
      "loss": 0.0949,
      "step": 1552
    },
    {
      "epoch": 0.17974537037037036,
      "grad_norm": 0.6725014448165894,
      "learning_rate": 1.6405092592592593e-05,
      "loss": 0.103,
      "step": 1553
    },
    {
      "epoch": 0.1798611111111111,
      "grad_norm": 0.6055047512054443,
      "learning_rate": 1.6402777777777777e-05,
      "loss": 0.1086,
      "step": 1554
    },
    {
      "epoch": 0.17997685185185186,
      "grad_norm": 0.7466843724250793,
      "learning_rate": 1.6400462962962965e-05,
      "loss": 0.1277,
      "step": 1555
    },
    {
      "epoch": 0.18009259259259258,
      "grad_norm": 0.5002665519714355,
      "learning_rate": 1.639814814814815e-05,
      "loss": 0.1029,
      "step": 1556
    },
    {
      "epoch": 0.18020833333333333,
      "grad_norm": 0.41437262296676636,
      "learning_rate": 1.6395833333333337e-05,
      "loss": 0.0609,
      "step": 1557
    },
    {
      "epoch": 0.18032407407407408,
      "grad_norm": 0.8264762163162231,
      "learning_rate": 1.639351851851852e-05,
      "loss": 0.1264,
      "step": 1558
    },
    {
      "epoch": 0.1804398148148148,
      "grad_norm": 0.5813721418380737,
      "learning_rate": 1.6391203703703705e-05,
      "loss": 0.1106,
      "step": 1559
    },
    {
      "epoch": 0.18055555555555555,
      "grad_norm": 3.365448236465454,
      "learning_rate": 1.638888888888889e-05,
      "loss": 0.1395,
      "step": 1560
    },
    {
      "epoch": 0.1806712962962963,
      "grad_norm": 0.4576779007911682,
      "learning_rate": 1.6386574074074077e-05,
      "loss": 0.0921,
      "step": 1561
    },
    {
      "epoch": 0.18078703703703702,
      "grad_norm": 0.43700942397117615,
      "learning_rate": 1.638425925925926e-05,
      "loss": 0.0907,
      "step": 1562
    },
    {
      "epoch": 0.18090277777777777,
      "grad_norm": 0.7327451109886169,
      "learning_rate": 1.6381944444444445e-05,
      "loss": 0.0922,
      "step": 1563
    },
    {
      "epoch": 0.18101851851851852,
      "grad_norm": 0.48275572061538696,
      "learning_rate": 1.6379629629629633e-05,
      "loss": 0.0681,
      "step": 1564
    },
    {
      "epoch": 0.18113425925925927,
      "grad_norm": 0.3306199312210083,
      "learning_rate": 1.6377314814814814e-05,
      "loss": 0.0678,
      "step": 1565
    },
    {
      "epoch": 0.18125,
      "grad_norm": 1.8388688564300537,
      "learning_rate": 1.6375e-05,
      "loss": 0.1099,
      "step": 1566
    },
    {
      "epoch": 0.18136574074074074,
      "grad_norm": 0.5366739630699158,
      "learning_rate": 1.6372685185185186e-05,
      "loss": 0.1064,
      "step": 1567
    },
    {
      "epoch": 0.1814814814814815,
      "grad_norm": 0.7144842743873596,
      "learning_rate": 1.6370370370370374e-05,
      "loss": 0.1041,
      "step": 1568
    },
    {
      "epoch": 0.18159722222222222,
      "grad_norm": 0.33221951127052307,
      "learning_rate": 1.6368055555555558e-05,
      "loss": 0.0671,
      "step": 1569
    },
    {
      "epoch": 0.18171296296296297,
      "grad_norm": 0.4990445077419281,
      "learning_rate": 1.6365740740740742e-05,
      "loss": 0.1019,
      "step": 1570
    },
    {
      "epoch": 0.18182870370370371,
      "grad_norm": 0.4223400056362152,
      "learning_rate": 1.6363425925925926e-05,
      "loss": 0.0842,
      "step": 1571
    },
    {
      "epoch": 0.18194444444444444,
      "grad_norm": 45.19260787963867,
      "learning_rate": 1.636111111111111e-05,
      "loss": 1.0212,
      "step": 1572
    },
    {
      "epoch": 0.18206018518518519,
      "grad_norm": 0.3983538746833801,
      "learning_rate": 1.6358796296296298e-05,
      "loss": 0.0792,
      "step": 1573
    },
    {
      "epoch": 0.18217592592592594,
      "grad_norm": 0.5013889670372009,
      "learning_rate": 1.6356481481481482e-05,
      "loss": 0.0991,
      "step": 1574
    },
    {
      "epoch": 0.18229166666666666,
      "grad_norm": 0.30657413601875305,
      "learning_rate": 1.635416666666667e-05,
      "loss": 0.0624,
      "step": 1575
    },
    {
      "epoch": 0.1824074074074074,
      "grad_norm": 0.33765387535095215,
      "learning_rate": 1.635185185185185e-05,
      "loss": 0.0693,
      "step": 1576
    },
    {
      "epoch": 0.18252314814814816,
      "grad_norm": 0.323378324508667,
      "learning_rate": 1.634953703703704e-05,
      "loss": 0.0635,
      "step": 1577
    },
    {
      "epoch": 0.18263888888888888,
      "grad_norm": 77.54625701904297,
      "learning_rate": 1.6347222222222223e-05,
      "loss": 0.8014,
      "step": 1578
    },
    {
      "epoch": 0.18275462962962963,
      "grad_norm": 0.400113046169281,
      "learning_rate": 1.634490740740741e-05,
      "loss": 0.0786,
      "step": 1579
    },
    {
      "epoch": 0.18287037037037038,
      "grad_norm": 0.5137234926223755,
      "learning_rate": 1.6342592592592595e-05,
      "loss": 0.1034,
      "step": 1580
    },
    {
      "epoch": 0.1829861111111111,
      "grad_norm": 0.43420591950416565,
      "learning_rate": 1.634027777777778e-05,
      "loss": 0.0833,
      "step": 1581
    },
    {
      "epoch": 0.18310185185185185,
      "grad_norm": 1.0966395139694214,
      "learning_rate": 1.6337962962962963e-05,
      "loss": 0.1157,
      "step": 1582
    },
    {
      "epoch": 0.1832175925925926,
      "grad_norm": 0.7326670289039612,
      "learning_rate": 1.6335648148148147e-05,
      "loss": 0.105,
      "step": 1583
    },
    {
      "epoch": 0.18333333333333332,
      "grad_norm": 1.1281712055206299,
      "learning_rate": 1.6333333333333335e-05,
      "loss": 0.1068,
      "step": 1584
    },
    {
      "epoch": 0.18344907407407407,
      "grad_norm": 0.3526194989681244,
      "learning_rate": 1.633101851851852e-05,
      "loss": 0.0712,
      "step": 1585
    },
    {
      "epoch": 0.18356481481481482,
      "grad_norm": 0.5391764044761658,
      "learning_rate": 1.6328703703703707e-05,
      "loss": 0.1066,
      "step": 1586
    },
    {
      "epoch": 0.18368055555555557,
      "grad_norm": 15.00948715209961,
      "learning_rate": 1.632638888888889e-05,
      "loss": 2.1224,
      "step": 1587
    },
    {
      "epoch": 0.1837962962962963,
      "grad_norm": 0.4161316752433777,
      "learning_rate": 1.6324074074074075e-05,
      "loss": 0.0858,
      "step": 1588
    },
    {
      "epoch": 0.18391203703703704,
      "grad_norm": 0.5795010328292847,
      "learning_rate": 1.632175925925926e-05,
      "loss": 0.1111,
      "step": 1589
    },
    {
      "epoch": 0.1840277777777778,
      "grad_norm": 0.3137289583683014,
      "learning_rate": 1.6319444444444444e-05,
      "loss": 0.0643,
      "step": 1590
    },
    {
      "epoch": 0.1841435185185185,
      "grad_norm": 0.33122718334198,
      "learning_rate": 1.631712962962963e-05,
      "loss": 0.0681,
      "step": 1591
    },
    {
      "epoch": 0.18425925925925926,
      "grad_norm": 0.5197389721870422,
      "learning_rate": 1.6314814814814816e-05,
      "loss": 0.0877,
      "step": 1592
    },
    {
      "epoch": 0.184375,
      "grad_norm": 74.83122253417969,
      "learning_rate": 1.6312500000000003e-05,
      "loss": 0.9783,
      "step": 1593
    },
    {
      "epoch": 0.18449074074074073,
      "grad_norm": 38.07498550415039,
      "learning_rate": 1.6310185185185184e-05,
      "loss": 0.259,
      "step": 1594
    },
    {
      "epoch": 0.18460648148148148,
      "grad_norm": 0.4078841507434845,
      "learning_rate": 1.6307870370370372e-05,
      "loss": 0.0786,
      "step": 1595
    },
    {
      "epoch": 0.18472222222222223,
      "grad_norm": 22.889657974243164,
      "learning_rate": 1.6305555555555556e-05,
      "loss": 0.2476,
      "step": 1596
    },
    {
      "epoch": 0.18483796296296295,
      "grad_norm": 0.3760571777820587,
      "learning_rate": 1.6303240740740744e-05,
      "loss": 0.0764,
      "step": 1597
    },
    {
      "epoch": 0.1849537037037037,
      "grad_norm": 0.3761894404888153,
      "learning_rate": 1.6300925925925928e-05,
      "loss": 0.0734,
      "step": 1598
    },
    {
      "epoch": 0.18506944444444445,
      "grad_norm": 16.44077491760254,
      "learning_rate": 1.6298611111111112e-05,
      "loss": 0.2012,
      "step": 1599
    },
    {
      "epoch": 0.18518518518518517,
      "grad_norm": 0.34681305289268494,
      "learning_rate": 1.6296296296296297e-05,
      "loss": 0.0698,
      "step": 1600
    },
    {
      "epoch": 0.18530092592592592,
      "grad_norm": 82.55341339111328,
      "learning_rate": 1.629398148148148e-05,
      "loss": 0.7446,
      "step": 1601
    },
    {
      "epoch": 0.18541666666666667,
      "grad_norm": 0.3412821590900421,
      "learning_rate": 1.629166666666667e-05,
      "loss": 0.0694,
      "step": 1602
    },
    {
      "epoch": 0.1855324074074074,
      "grad_norm": 51.28432846069336,
      "learning_rate": 1.6289351851851853e-05,
      "loss": 0.395,
      "step": 1603
    },
    {
      "epoch": 0.18564814814814815,
      "grad_norm": 0.6723182797431946,
      "learning_rate": 1.628703703703704e-05,
      "loss": 0.1151,
      "step": 1604
    },
    {
      "epoch": 0.1857638888888889,
      "grad_norm": 0.36848023533821106,
      "learning_rate": 1.6284722222222225e-05,
      "loss": 0.0692,
      "step": 1605
    },
    {
      "epoch": 0.18587962962962962,
      "grad_norm": 0.4953189194202423,
      "learning_rate": 1.628240740740741e-05,
      "loss": 0.0981,
      "step": 1606
    },
    {
      "epoch": 0.18599537037037037,
      "grad_norm": 62.809303283691406,
      "learning_rate": 1.6280092592592593e-05,
      "loss": 0.6548,
      "step": 1607
    },
    {
      "epoch": 0.18611111111111112,
      "grad_norm": 0.5684263110160828,
      "learning_rate": 1.6277777777777777e-05,
      "loss": 0.0955,
      "step": 1608
    },
    {
      "epoch": 0.18622685185185187,
      "grad_norm": 0.3626304864883423,
      "learning_rate": 1.6275462962962965e-05,
      "loss": 0.0739,
      "step": 1609
    },
    {
      "epoch": 0.1863425925925926,
      "grad_norm": 54.62129211425781,
      "learning_rate": 1.627314814814815e-05,
      "loss": 0.6794,
      "step": 1610
    },
    {
      "epoch": 0.18645833333333334,
      "grad_norm": 0.8526928424835205,
      "learning_rate": 1.6270833333333337e-05,
      "loss": 0.0999,
      "step": 1611
    },
    {
      "epoch": 0.1865740740740741,
      "grad_norm": 0.35950151085853577,
      "learning_rate": 1.6268518518518518e-05,
      "loss": 0.0706,
      "step": 1612
    },
    {
      "epoch": 0.1866898148148148,
      "grad_norm": 0.763081967830658,
      "learning_rate": 1.6266203703703705e-05,
      "loss": 0.0999,
      "step": 1613
    },
    {
      "epoch": 0.18680555555555556,
      "grad_norm": 0.38439828157424927,
      "learning_rate": 1.626388888888889e-05,
      "loss": 0.0565,
      "step": 1614
    },
    {
      "epoch": 0.1869212962962963,
      "grad_norm": 0.5612685084342957,
      "learning_rate": 1.6261574074074077e-05,
      "loss": 0.1124,
      "step": 1615
    },
    {
      "epoch": 0.18703703703703703,
      "grad_norm": 29.505748748779297,
      "learning_rate": 1.625925925925926e-05,
      "loss": 0.2426,
      "step": 1616
    },
    {
      "epoch": 0.18715277777777778,
      "grad_norm": 0.5829209089279175,
      "learning_rate": 1.6256944444444446e-05,
      "loss": 0.083,
      "step": 1617
    },
    {
      "epoch": 0.18726851851851853,
      "grad_norm": 0.4041840434074402,
      "learning_rate": 1.625462962962963e-05,
      "loss": 0.0832,
      "step": 1618
    },
    {
      "epoch": 0.18738425925925925,
      "grad_norm": 0.367414653301239,
      "learning_rate": 1.6252314814814814e-05,
      "loss": 0.0693,
      "step": 1619
    },
    {
      "epoch": 0.1875,
      "grad_norm": 0.38703253865242004,
      "learning_rate": 1.6250000000000002e-05,
      "loss": 0.0792,
      "step": 1620
    },
    {
      "epoch": 0.18761574074074075,
      "grad_norm": 3.7133545875549316,
      "learning_rate": 1.6247685185185186e-05,
      "loss": 0.1255,
      "step": 1621
    },
    {
      "epoch": 0.18773148148148147,
      "grad_norm": 0.6679739356040955,
      "learning_rate": 1.6245370370370374e-05,
      "loss": 0.0865,
      "step": 1622
    },
    {
      "epoch": 0.18784722222222222,
      "grad_norm": 0.5226842761039734,
      "learning_rate": 1.6243055555555555e-05,
      "loss": 0.0767,
      "step": 1623
    },
    {
      "epoch": 0.18796296296296297,
      "grad_norm": 141.58297729492188,
      "learning_rate": 1.6240740740740742e-05,
      "loss": 1.109,
      "step": 1624
    },
    {
      "epoch": 0.1880787037037037,
      "grad_norm": 0.996178150177002,
      "learning_rate": 1.6238425925925927e-05,
      "loss": 0.076,
      "step": 1625
    },
    {
      "epoch": 0.18819444444444444,
      "grad_norm": 0.39787545800209045,
      "learning_rate": 1.623611111111111e-05,
      "loss": 0.0578,
      "step": 1626
    },
    {
      "epoch": 0.1883101851851852,
      "grad_norm": 0.39882540702819824,
      "learning_rate": 1.62337962962963e-05,
      "loss": 0.0821,
      "step": 1627
    },
    {
      "epoch": 0.1884259259259259,
      "grad_norm": 0.7568697929382324,
      "learning_rate": 1.6231481481481483e-05,
      "loss": 0.0778,
      "step": 1628
    },
    {
      "epoch": 0.18854166666666666,
      "grad_norm": 1.037674903869629,
      "learning_rate": 1.6229166666666667e-05,
      "loss": 0.0892,
      "step": 1629
    },
    {
      "epoch": 0.1886574074074074,
      "grad_norm": 0.3175199031829834,
      "learning_rate": 1.622685185185185e-05,
      "loss": 0.0637,
      "step": 1630
    },
    {
      "epoch": 0.18877314814814813,
      "grad_norm": 0.34380683302879333,
      "learning_rate": 1.622453703703704e-05,
      "loss": 0.0682,
      "step": 1631
    },
    {
      "epoch": 0.18888888888888888,
      "grad_norm": 3.043649911880493,
      "learning_rate": 1.6222222222222223e-05,
      "loss": 0.1315,
      "step": 1632
    },
    {
      "epoch": 0.18900462962962963,
      "grad_norm": 0.5359206199645996,
      "learning_rate": 1.621990740740741e-05,
      "loss": 0.1084,
      "step": 1633
    },
    {
      "epoch": 0.18912037037037038,
      "grad_norm": 0.43528422713279724,
      "learning_rate": 1.6217592592592595e-05,
      "loss": 0.0749,
      "step": 1634
    },
    {
      "epoch": 0.1892361111111111,
      "grad_norm": 0.466022789478302,
      "learning_rate": 1.621527777777778e-05,
      "loss": 0.0812,
      "step": 1635
    },
    {
      "epoch": 0.18935185185185185,
      "grad_norm": 101.93574523925781,
      "learning_rate": 1.6212962962962964e-05,
      "loss": 0.9581,
      "step": 1636
    },
    {
      "epoch": 0.1894675925925926,
      "grad_norm": 0.48114675283432007,
      "learning_rate": 1.6210648148148148e-05,
      "loss": 0.082,
      "step": 1637
    },
    {
      "epoch": 0.18958333333333333,
      "grad_norm": 5.547873497009277,
      "learning_rate": 1.6208333333333335e-05,
      "loss": 2.2479,
      "step": 1638
    },
    {
      "epoch": 0.18969907407407408,
      "grad_norm": 60.43742370605469,
      "learning_rate": 1.620601851851852e-05,
      "loss": 0.3972,
      "step": 1639
    },
    {
      "epoch": 0.18981481481481483,
      "grad_norm": 9.696928977966309,
      "learning_rate": 1.6203703703703707e-05,
      "loss": 0.1733,
      "step": 1640
    },
    {
      "epoch": 0.18993055555555555,
      "grad_norm": 52.26536178588867,
      "learning_rate": 1.6201388888888888e-05,
      "loss": 1.7334,
      "step": 1641
    },
    {
      "epoch": 0.1900462962962963,
      "grad_norm": 15.05859661102295,
      "learning_rate": 1.6199074074074076e-05,
      "loss": 0.1415,
      "step": 1642
    },
    {
      "epoch": 0.19016203703703705,
      "grad_norm": 117.2967758178711,
      "learning_rate": 1.619675925925926e-05,
      "loss": 0.9723,
      "step": 1643
    },
    {
      "epoch": 0.19027777777777777,
      "grad_norm": 0.379097580909729,
      "learning_rate": 1.6194444444444444e-05,
      "loss": 0.0765,
      "step": 1644
    },
    {
      "epoch": 0.19039351851851852,
      "grad_norm": 7.650820732116699,
      "learning_rate": 1.6192129629629632e-05,
      "loss": 0.1465,
      "step": 1645
    },
    {
      "epoch": 0.19050925925925927,
      "grad_norm": 0.3273029923439026,
      "learning_rate": 1.6189814814814816e-05,
      "loss": 0.0667,
      "step": 1646
    },
    {
      "epoch": 0.190625,
      "grad_norm": 0.31114768981933594,
      "learning_rate": 1.61875e-05,
      "loss": 0.063,
      "step": 1647
    },
    {
      "epoch": 0.19074074074074074,
      "grad_norm": 22.551794052124023,
      "learning_rate": 1.6185185185185185e-05,
      "loss": 2.0627,
      "step": 1648
    },
    {
      "epoch": 0.1908564814814815,
      "grad_norm": 0.40096917748451233,
      "learning_rate": 1.6182870370370372e-05,
      "loss": 0.0754,
      "step": 1649
    },
    {
      "epoch": 0.1909722222222222,
      "grad_norm": 0.35591673851013184,
      "learning_rate": 1.6180555555555557e-05,
      "loss": 0.0685,
      "step": 1650
    },
    {
      "epoch": 0.19108796296296296,
      "grad_norm": 0.33211249113082886,
      "learning_rate": 1.6178240740740744e-05,
      "loss": 0.0669,
      "step": 1651
    },
    {
      "epoch": 0.1912037037037037,
      "grad_norm": 0.4840121865272522,
      "learning_rate": 1.617592592592593e-05,
      "loss": 0.0968,
      "step": 1652
    },
    {
      "epoch": 0.19131944444444443,
      "grad_norm": 8.105879783630371,
      "learning_rate": 1.6173611111111113e-05,
      "loss": 0.14,
      "step": 1653
    },
    {
      "epoch": 0.19143518518518518,
      "grad_norm": 61.19565963745117,
      "learning_rate": 1.6171296296296297e-05,
      "loss": 0.9678,
      "step": 1654
    },
    {
      "epoch": 0.19155092592592593,
      "grad_norm": 0.3948344588279724,
      "learning_rate": 1.616898148148148e-05,
      "loss": 0.078,
      "step": 1655
    },
    {
      "epoch": 0.19166666666666668,
      "grad_norm": 0.5101973414421082,
      "learning_rate": 1.616666666666667e-05,
      "loss": 0.0738,
      "step": 1656
    },
    {
      "epoch": 0.1917824074074074,
      "grad_norm": 0.2899615168571472,
      "learning_rate": 1.6164351851851853e-05,
      "loss": 0.0588,
      "step": 1657
    },
    {
      "epoch": 0.19189814814814815,
      "grad_norm": 0.30900290608406067,
      "learning_rate": 1.616203703703704e-05,
      "loss": 0.0595,
      "step": 1658
    },
    {
      "epoch": 0.1920138888888889,
      "grad_norm": 90.2379379272461,
      "learning_rate": 1.615972222222222e-05,
      "loss": 0.9694,
      "step": 1659
    },
    {
      "epoch": 0.19212962962962962,
      "grad_norm": 55.235862731933594,
      "learning_rate": 1.615740740740741e-05,
      "loss": 1.4534,
      "step": 1660
    },
    {
      "epoch": 0.19224537037037037,
      "grad_norm": 0.5429491996765137,
      "learning_rate": 1.6155092592592594e-05,
      "loss": 0.0952,
      "step": 1661
    },
    {
      "epoch": 0.19236111111111112,
      "grad_norm": 0.41663694381713867,
      "learning_rate": 1.6152777777777778e-05,
      "loss": 0.0801,
      "step": 1662
    },
    {
      "epoch": 0.19247685185185184,
      "grad_norm": 59.44485092163086,
      "learning_rate": 1.6150462962962965e-05,
      "loss": 0.4187,
      "step": 1663
    },
    {
      "epoch": 0.1925925925925926,
      "grad_norm": 0.45688581466674805,
      "learning_rate": 1.614814814814815e-05,
      "loss": 0.0915,
      "step": 1664
    },
    {
      "epoch": 0.19270833333333334,
      "grad_norm": 32.38243103027344,
      "learning_rate": 1.6145833333333334e-05,
      "loss": 1.8102,
      "step": 1665
    },
    {
      "epoch": 0.19282407407407406,
      "grad_norm": 67.72091674804688,
      "learning_rate": 1.6143518518518518e-05,
      "loss": 1.2627,
      "step": 1666
    },
    {
      "epoch": 0.19293981481481481,
      "grad_norm": 0.3876035809516907,
      "learning_rate": 1.6141203703703706e-05,
      "loss": 0.0635,
      "step": 1667
    },
    {
      "epoch": 0.19305555555555556,
      "grad_norm": 0.709570050239563,
      "learning_rate": 1.613888888888889e-05,
      "loss": 0.0944,
      "step": 1668
    },
    {
      "epoch": 0.19317129629629629,
      "grad_norm": 0.7571654319763184,
      "learning_rate": 1.6136574074074078e-05,
      "loss": 0.0925,
      "step": 1669
    },
    {
      "epoch": 0.19328703703703703,
      "grad_norm": 15.513598442077637,
      "learning_rate": 1.613425925925926e-05,
      "loss": 0.2493,
      "step": 1670
    },
    {
      "epoch": 0.19340277777777778,
      "grad_norm": 0.3553959131240845,
      "learning_rate": 1.6131944444444446e-05,
      "loss": 0.0627,
      "step": 1671
    },
    {
      "epoch": 0.1935185185185185,
      "grad_norm": 0.4539666175842285,
      "learning_rate": 1.612962962962963e-05,
      "loss": 0.0673,
      "step": 1672
    },
    {
      "epoch": 0.19363425925925926,
      "grad_norm": 0.36994341015815735,
      "learning_rate": 1.6127314814814815e-05,
      "loss": 0.0742,
      "step": 1673
    },
    {
      "epoch": 0.19375,
      "grad_norm": 54.63010025024414,
      "learning_rate": 1.6125000000000002e-05,
      "loss": 0.5204,
      "step": 1674
    },
    {
      "epoch": 0.19386574074074073,
      "grad_norm": 61.109066009521484,
      "learning_rate": 1.6122685185185187e-05,
      "loss": 1.4412,
      "step": 1675
    },
    {
      "epoch": 0.19398148148148148,
      "grad_norm": 0.3395497798919678,
      "learning_rate": 1.612037037037037e-05,
      "loss": 0.0686,
      "step": 1676
    },
    {
      "epoch": 0.19409722222222223,
      "grad_norm": 0.8438761830329895,
      "learning_rate": 1.6118055555555555e-05,
      "loss": 0.1061,
      "step": 1677
    },
    {
      "epoch": 0.19421296296296298,
      "grad_norm": 28.038089752197266,
      "learning_rate": 1.6115740740740743e-05,
      "loss": 1.7397,
      "step": 1678
    },
    {
      "epoch": 0.1943287037037037,
      "grad_norm": 0.509177565574646,
      "learning_rate": 1.6113425925925927e-05,
      "loss": 0.0813,
      "step": 1679
    },
    {
      "epoch": 0.19444444444444445,
      "grad_norm": 11.119701385498047,
      "learning_rate": 1.6111111111111115e-05,
      "loss": 0.1737,
      "step": 1680
    },
    {
      "epoch": 0.1945601851851852,
      "grad_norm": 0.50190669298172,
      "learning_rate": 1.61087962962963e-05,
      "loss": 0.0988,
      "step": 1681
    },
    {
      "epoch": 0.19467592592592592,
      "grad_norm": 0.5026053786277771,
      "learning_rate": 1.6106481481481483e-05,
      "loss": 0.1035,
      "step": 1682
    },
    {
      "epoch": 0.19479166666666667,
      "grad_norm": 5.187301158905029,
      "learning_rate": 1.6104166666666667e-05,
      "loss": 0.1315,
      "step": 1683
    },
    {
      "epoch": 0.19490740740740742,
      "grad_norm": 0.40079188346862793,
      "learning_rate": 1.610185185185185e-05,
      "loss": 0.0755,
      "step": 1684
    },
    {
      "epoch": 0.19502314814814814,
      "grad_norm": 0.32310009002685547,
      "learning_rate": 1.609953703703704e-05,
      "loss": 0.0621,
      "step": 1685
    },
    {
      "epoch": 0.1951388888888889,
      "grad_norm": 0.292239785194397,
      "learning_rate": 1.6097222222222223e-05,
      "loss": 0.0583,
      "step": 1686
    },
    {
      "epoch": 0.19525462962962964,
      "grad_norm": 2.485487699508667,
      "learning_rate": 1.609490740740741e-05,
      "loss": 0.0924,
      "step": 1687
    },
    {
      "epoch": 0.19537037037037036,
      "grad_norm": 0.5291061401367188,
      "learning_rate": 1.6092592592592592e-05,
      "loss": 0.1076,
      "step": 1688
    },
    {
      "epoch": 0.1954861111111111,
      "grad_norm": 0.32386165857315063,
      "learning_rate": 1.609027777777778e-05,
      "loss": 0.0646,
      "step": 1689
    },
    {
      "epoch": 0.19560185185185186,
      "grad_norm": 0.3215344250202179,
      "learning_rate": 1.6087962962962964e-05,
      "loss": 0.0611,
      "step": 1690
    },
    {
      "epoch": 0.19571759259259258,
      "grad_norm": 0.3515690565109253,
      "learning_rate": 1.6085648148148148e-05,
      "loss": 0.0702,
      "step": 1691
    },
    {
      "epoch": 0.19583333333333333,
      "grad_norm": 26.474565505981445,
      "learning_rate": 1.6083333333333336e-05,
      "loss": 1.2793,
      "step": 1692
    },
    {
      "epoch": 0.19594907407407408,
      "grad_norm": 82.43876647949219,
      "learning_rate": 1.608101851851852e-05,
      "loss": 1.269,
      "step": 1693
    },
    {
      "epoch": 0.1960648148148148,
      "grad_norm": 0.5351049900054932,
      "learning_rate": 1.6078703703703704e-05,
      "loss": 0.1042,
      "step": 1694
    },
    {
      "epoch": 0.19618055555555555,
      "grad_norm": 0.3951953649520874,
      "learning_rate": 1.607638888888889e-05,
      "loss": 0.0769,
      "step": 1695
    },
    {
      "epoch": 0.1962962962962963,
      "grad_norm": 0.5229839086532593,
      "learning_rate": 1.6074074074074076e-05,
      "loss": 0.1061,
      "step": 1696
    },
    {
      "epoch": 0.19641203703703702,
      "grad_norm": 0.6618062853813171,
      "learning_rate": 1.607175925925926e-05,
      "loss": 0.1082,
      "step": 1697
    },
    {
      "epoch": 0.19652777777777777,
      "grad_norm": 0.41091570258140564,
      "learning_rate": 1.6069444444444448e-05,
      "loss": 0.0775,
      "step": 1698
    },
    {
      "epoch": 0.19664351851851852,
      "grad_norm": 0.4464718997478485,
      "learning_rate": 1.6067129629629632e-05,
      "loss": 0.0906,
      "step": 1699
    },
    {
      "epoch": 0.19675925925925927,
      "grad_norm": 0.46035486459732056,
      "learning_rate": 1.6064814814814817e-05,
      "loss": 0.0915,
      "step": 1700
    },
    {
      "epoch": 0.196875,
      "grad_norm": 0.39866918325424194,
      "learning_rate": 1.60625e-05,
      "loss": 0.0773,
      "step": 1701
    },
    {
      "epoch": 0.19699074074074074,
      "grad_norm": 1.4900858402252197,
      "learning_rate": 1.6060185185185185e-05,
      "loss": 0.0688,
      "step": 1702
    },
    {
      "epoch": 0.1971064814814815,
      "grad_norm": 0.5320736765861511,
      "learning_rate": 1.6057870370370373e-05,
      "loss": 0.0614,
      "step": 1703
    },
    {
      "epoch": 0.19722222222222222,
      "grad_norm": 0.8086181282997131,
      "learning_rate": 1.6055555555555557e-05,
      "loss": 0.1025,
      "step": 1704
    },
    {
      "epoch": 0.19733796296296297,
      "grad_norm": 0.5532802939414978,
      "learning_rate": 1.6053240740740745e-05,
      "loss": 0.1157,
      "step": 1705
    },
    {
      "epoch": 0.19745370370370371,
      "grad_norm": 23.79521369934082,
      "learning_rate": 1.6050925925925925e-05,
      "loss": 1.9162,
      "step": 1706
    },
    {
      "epoch": 0.19756944444444444,
      "grad_norm": 0.3262244164943695,
      "learning_rate": 1.6048611111111113e-05,
      "loss": 0.0639,
      "step": 1707
    },
    {
      "epoch": 0.19768518518518519,
      "grad_norm": 0.37238049507141113,
      "learning_rate": 1.6046296296296297e-05,
      "loss": 0.0744,
      "step": 1708
    },
    {
      "epoch": 0.19780092592592594,
      "grad_norm": 0.3064058721065521,
      "learning_rate": 1.604398148148148e-05,
      "loss": 0.0598,
      "step": 1709
    },
    {
      "epoch": 0.19791666666666666,
      "grad_norm": 0.40785184502601624,
      "learning_rate": 1.604166666666667e-05,
      "loss": 0.0761,
      "step": 1710
    },
    {
      "epoch": 0.1980324074074074,
      "grad_norm": 0.4869638979434967,
      "learning_rate": 1.6039351851851853e-05,
      "loss": 0.0892,
      "step": 1711
    },
    {
      "epoch": 0.19814814814814816,
      "grad_norm": 0.40465566515922546,
      "learning_rate": 1.6037037037037038e-05,
      "loss": 0.0794,
      "step": 1712
    },
    {
      "epoch": 0.19826388888888888,
      "grad_norm": 0.4769534468650818,
      "learning_rate": 1.6034722222222222e-05,
      "loss": 0.0687,
      "step": 1713
    },
    {
      "epoch": 0.19837962962962963,
      "grad_norm": 30.005413055419922,
      "learning_rate": 1.603240740740741e-05,
      "loss": 0.3335,
      "step": 1714
    },
    {
      "epoch": 0.19849537037037038,
      "grad_norm": 0.4503821134567261,
      "learning_rate": 1.6030092592592594e-05,
      "loss": 0.0627,
      "step": 1715
    },
    {
      "epoch": 0.1986111111111111,
      "grad_norm": 0.40512579679489136,
      "learning_rate": 1.602777777777778e-05,
      "loss": 0.0582,
      "step": 1716
    },
    {
      "epoch": 0.19872685185185185,
      "grad_norm": 0.4783536493778229,
      "learning_rate": 1.6025462962962962e-05,
      "loss": 0.0883,
      "step": 1717
    },
    {
      "epoch": 0.1988425925925926,
      "grad_norm": 0.37266504764556885,
      "learning_rate": 1.602314814814815e-05,
      "loss": 0.0705,
      "step": 1718
    },
    {
      "epoch": 0.19895833333333332,
      "grad_norm": 0.2769918739795685,
      "learning_rate": 1.6020833333333334e-05,
      "loss": 0.0555,
      "step": 1719
    },
    {
      "epoch": 0.19907407407407407,
      "grad_norm": 0.45518580079078674,
      "learning_rate": 1.601851851851852e-05,
      "loss": 0.0664,
      "step": 1720
    },
    {
      "epoch": 0.19918981481481482,
      "grad_norm": 4.94245719909668,
      "learning_rate": 1.6016203703703706e-05,
      "loss": 0.142,
      "step": 1721
    },
    {
      "epoch": 0.19930555555555557,
      "grad_norm": 0.4717915654182434,
      "learning_rate": 1.601388888888889e-05,
      "loss": 0.0919,
      "step": 1722
    },
    {
      "epoch": 0.1994212962962963,
      "grad_norm": 1.0915738344192505,
      "learning_rate": 1.6011574074074075e-05,
      "loss": 0.0933,
      "step": 1723
    },
    {
      "epoch": 0.19953703703703704,
      "grad_norm": 1.7011257410049438,
      "learning_rate": 1.600925925925926e-05,
      "loss": 0.0973,
      "step": 1724
    },
    {
      "epoch": 0.1996527777777778,
      "grad_norm": 0.454412043094635,
      "learning_rate": 1.6006944444444447e-05,
      "loss": 0.0895,
      "step": 1725
    },
    {
      "epoch": 0.1997685185185185,
      "grad_norm": 0.42654845118522644,
      "learning_rate": 1.600462962962963e-05,
      "loss": 0.0802,
      "step": 1726
    },
    {
      "epoch": 0.19988425925925926,
      "grad_norm": 0.6832679510116577,
      "learning_rate": 1.6002314814814815e-05,
      "loss": 0.1041,
      "step": 1727
    },
    {
      "epoch": 0.2,
      "grad_norm": 50.27646255493164,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.5424,
      "step": 1728
    },
    {
      "epoch": 0.20011574074074073,
      "grad_norm": 0.9692054390907288,
      "learning_rate": 1.5997685185185187e-05,
      "loss": 0.074,
      "step": 1729
    },
    {
      "epoch": 0.20023148148148148,
      "grad_norm": 0.5043386220932007,
      "learning_rate": 1.599537037037037e-05,
      "loss": 0.089,
      "step": 1730
    },
    {
      "epoch": 0.20034722222222223,
      "grad_norm": 6.9145121574401855,
      "learning_rate": 1.5993055555555555e-05,
      "loss": 0.1665,
      "step": 1731
    },
    {
      "epoch": 0.20046296296296295,
      "grad_norm": 0.30531492829322815,
      "learning_rate": 1.5990740740740743e-05,
      "loss": 0.0614,
      "step": 1732
    },
    {
      "epoch": 0.2005787037037037,
      "grad_norm": 0.2839508056640625,
      "learning_rate": 1.5988425925925927e-05,
      "loss": 0.0564,
      "step": 1733
    },
    {
      "epoch": 0.20069444444444445,
      "grad_norm": 34.86577606201172,
      "learning_rate": 1.5986111111111115e-05,
      "loss": 1.4827,
      "step": 1734
    },
    {
      "epoch": 0.20081018518518517,
      "grad_norm": 0.45494765043258667,
      "learning_rate": 1.5983796296296296e-05,
      "loss": 0.0887,
      "step": 1735
    },
    {
      "epoch": 0.20092592592592592,
      "grad_norm": 0.47342339158058167,
      "learning_rate": 1.5981481481481483e-05,
      "loss": 0.0878,
      "step": 1736
    },
    {
      "epoch": 0.20104166666666667,
      "grad_norm": 0.43579551577568054,
      "learning_rate": 1.5979166666666668e-05,
      "loss": 0.0838,
      "step": 1737
    },
    {
      "epoch": 0.2011574074074074,
      "grad_norm": 0.45664530992507935,
      "learning_rate": 1.5976851851851852e-05,
      "loss": 0.0804,
      "step": 1738
    },
    {
      "epoch": 0.20127314814814815,
      "grad_norm": 0.33127361536026,
      "learning_rate": 1.597453703703704e-05,
      "loss": 0.0667,
      "step": 1739
    },
    {
      "epoch": 0.2013888888888889,
      "grad_norm": 0.47876396775245667,
      "learning_rate": 1.5972222222222224e-05,
      "loss": 0.0992,
      "step": 1740
    },
    {
      "epoch": 0.20150462962962962,
      "grad_norm": 0.4181203246116638,
      "learning_rate": 1.5969907407407408e-05,
      "loss": 0.0774,
      "step": 1741
    },
    {
      "epoch": 0.20162037037037037,
      "grad_norm": 0.35919347405433655,
      "learning_rate": 1.5967592592592592e-05,
      "loss": 0.0523,
      "step": 1742
    },
    {
      "epoch": 0.20173611111111112,
      "grad_norm": 3.4915273189544678,
      "learning_rate": 1.596527777777778e-05,
      "loss": 0.0851,
      "step": 1743
    },
    {
      "epoch": 0.20185185185185187,
      "grad_norm": 0.33452150225639343,
      "learning_rate": 1.5962962962962964e-05,
      "loss": 0.0679,
      "step": 1744
    },
    {
      "epoch": 0.2019675925925926,
      "grad_norm": 68.94770050048828,
      "learning_rate": 1.596064814814815e-05,
      "loss": 1.0706,
      "step": 1745
    },
    {
      "epoch": 0.20208333333333334,
      "grad_norm": 0.4437830448150635,
      "learning_rate": 1.5958333333333336e-05,
      "loss": 0.0865,
      "step": 1746
    },
    {
      "epoch": 0.2021990740740741,
      "grad_norm": 0.3003343343734741,
      "learning_rate": 1.595601851851852e-05,
      "loss": 0.0598,
      "step": 1747
    },
    {
      "epoch": 0.2023148148148148,
      "grad_norm": 0.32711270451545715,
      "learning_rate": 1.5953703703703705e-05,
      "loss": 0.0666,
      "step": 1748
    },
    {
      "epoch": 0.20243055555555556,
      "grad_norm": 0.4772289991378784,
      "learning_rate": 1.595138888888889e-05,
      "loss": 0.0851,
      "step": 1749
    },
    {
      "epoch": 0.2025462962962963,
      "grad_norm": 1.8838766813278198,
      "learning_rate": 1.5949074074074077e-05,
      "loss": 0.0731,
      "step": 1750
    },
    {
      "epoch": 0.20266203703703703,
      "grad_norm": 0.28393709659576416,
      "learning_rate": 1.594675925925926e-05,
      "loss": 0.0567,
      "step": 1751
    },
    {
      "epoch": 0.20277777777777778,
      "grad_norm": 0.2870407700538635,
      "learning_rate": 1.594444444444445e-05,
      "loss": 0.0582,
      "step": 1752
    },
    {
      "epoch": 0.20289351851851853,
      "grad_norm": 0.3878479599952698,
      "learning_rate": 1.594212962962963e-05,
      "loss": 0.0766,
      "step": 1753
    },
    {
      "epoch": 0.20300925925925925,
      "grad_norm": 92.46924591064453,
      "learning_rate": 1.5939814814814817e-05,
      "loss": 0.9204,
      "step": 1754
    },
    {
      "epoch": 0.203125,
      "grad_norm": 0.4618960916996002,
      "learning_rate": 1.59375e-05,
      "loss": 0.0853,
      "step": 1755
    },
    {
      "epoch": 0.20324074074074075,
      "grad_norm": 0.3590377867221832,
      "learning_rate": 1.5935185185185185e-05,
      "loss": 0.0705,
      "step": 1756
    },
    {
      "epoch": 0.20335648148148147,
      "grad_norm": 35.43350601196289,
      "learning_rate": 1.5932870370370373e-05,
      "loss": 1.5727,
      "step": 1757
    },
    {
      "epoch": 0.20347222222222222,
      "grad_norm": 0.3528125584125519,
      "learning_rate": 1.5930555555555557e-05,
      "loss": 0.0699,
      "step": 1758
    },
    {
      "epoch": 0.20358796296296297,
      "grad_norm": 0.4799154996871948,
      "learning_rate": 1.592824074074074e-05,
      "loss": 0.0978,
      "step": 1759
    },
    {
      "epoch": 0.2037037037037037,
      "grad_norm": 0.28857946395874023,
      "learning_rate": 1.5925925925925926e-05,
      "loss": 0.0572,
      "step": 1760
    },
    {
      "epoch": 0.20381944444444444,
      "grad_norm": 0.2777757942676544,
      "learning_rate": 1.5923611111111113e-05,
      "loss": 0.0563,
      "step": 1761
    },
    {
      "epoch": 0.2039351851851852,
      "grad_norm": 0.6701968908309937,
      "learning_rate": 1.5921296296296298e-05,
      "loss": 0.0794,
      "step": 1762
    },
    {
      "epoch": 0.2040509259259259,
      "grad_norm": 0.35507526993751526,
      "learning_rate": 1.5918981481481482e-05,
      "loss": 0.0619,
      "step": 1763
    },
    {
      "epoch": 0.20416666666666666,
      "grad_norm": 0.43413957953453064,
      "learning_rate": 1.5916666666666666e-05,
      "loss": 0.0633,
      "step": 1764
    },
    {
      "epoch": 0.2042824074074074,
      "grad_norm": 0.34721869230270386,
      "learning_rate": 1.5914351851851854e-05,
      "loss": 0.0684,
      "step": 1765
    },
    {
      "epoch": 0.20439814814814813,
      "grad_norm": 0.3224037289619446,
      "learning_rate": 1.5912037037037038e-05,
      "loss": 0.0601,
      "step": 1766
    },
    {
      "epoch": 0.20451388888888888,
      "grad_norm": 0.6075630187988281,
      "learning_rate": 1.5909722222222222e-05,
      "loss": 0.1016,
      "step": 1767
    },
    {
      "epoch": 0.20462962962962963,
      "grad_norm": 0.4053906202316284,
      "learning_rate": 1.590740740740741e-05,
      "loss": 0.0546,
      "step": 1768
    },
    {
      "epoch": 0.20474537037037038,
      "grad_norm": 11.317258834838867,
      "learning_rate": 1.5905092592592594e-05,
      "loss": 0.1751,
      "step": 1769
    },
    {
      "epoch": 0.2048611111111111,
      "grad_norm": 0.47364532947540283,
      "learning_rate": 1.590277777777778e-05,
      "loss": 0.0953,
      "step": 1770
    },
    {
      "epoch": 0.20497685185185185,
      "grad_norm": 0.37909871339797974,
      "learning_rate": 1.5900462962962963e-05,
      "loss": 0.0554,
      "step": 1771
    },
    {
      "epoch": 0.2050925925925926,
      "grad_norm": 0.4359396696090698,
      "learning_rate": 1.589814814814815e-05,
      "loss": 0.0867,
      "step": 1772
    },
    {
      "epoch": 0.20520833333333333,
      "grad_norm": 0.4451242983341217,
      "learning_rate": 1.5895833333333335e-05,
      "loss": 0.0866,
      "step": 1773
    },
    {
      "epoch": 0.20532407407407408,
      "grad_norm": 2.834028959274292,
      "learning_rate": 1.589351851851852e-05,
      "loss": 0.1251,
      "step": 1774
    },
    {
      "epoch": 0.20543981481481483,
      "grad_norm": 0.495297372341156,
      "learning_rate": 1.5891203703703706e-05,
      "loss": 0.0882,
      "step": 1775
    },
    {
      "epoch": 0.20555555555555555,
      "grad_norm": 0.8966808915138245,
      "learning_rate": 1.588888888888889e-05,
      "loss": 0.0672,
      "step": 1776
    },
    {
      "epoch": 0.2056712962962963,
      "grad_norm": 2.164645195007324,
      "learning_rate": 1.5886574074074075e-05,
      "loss": 0.0997,
      "step": 1777
    },
    {
      "epoch": 0.20578703703703705,
      "grad_norm": 0.4321324825286865,
      "learning_rate": 1.588425925925926e-05,
      "loss": 0.071,
      "step": 1778
    },
    {
      "epoch": 0.20590277777777777,
      "grad_norm": 0.32644525170326233,
      "learning_rate": 1.5881944444444447e-05,
      "loss": 0.0605,
      "step": 1779
    },
    {
      "epoch": 0.20601851851851852,
      "grad_norm": 1.9433269500732422,
      "learning_rate": 1.587962962962963e-05,
      "loss": 0.1046,
      "step": 1780
    },
    {
      "epoch": 0.20613425925925927,
      "grad_norm": 0.35710370540618896,
      "learning_rate": 1.5877314814814815e-05,
      "loss": 0.0717,
      "step": 1781
    },
    {
      "epoch": 0.20625,
      "grad_norm": 0.30629730224609375,
      "learning_rate": 1.5875e-05,
      "loss": 0.0613,
      "step": 1782
    },
    {
      "epoch": 0.20636574074074074,
      "grad_norm": 0.32799994945526123,
      "learning_rate": 1.5872685185185187e-05,
      "loss": 0.0652,
      "step": 1783
    },
    {
      "epoch": 0.2064814814814815,
      "grad_norm": 0.37480250000953674,
      "learning_rate": 1.587037037037037e-05,
      "loss": 0.0702,
      "step": 1784
    },
    {
      "epoch": 0.2065972222222222,
      "grad_norm": 1.237346887588501,
      "learning_rate": 1.5868055555555556e-05,
      "loss": 0.0729,
      "step": 1785
    },
    {
      "epoch": 0.20671296296296296,
      "grad_norm": 0.4432011842727661,
      "learning_rate": 1.5865740740740743e-05,
      "loss": 0.0888,
      "step": 1786
    },
    {
      "epoch": 0.2068287037037037,
      "grad_norm": 0.2642877995967865,
      "learning_rate": 1.5863425925925928e-05,
      "loss": 0.0535,
      "step": 1787
    },
    {
      "epoch": 0.20694444444444443,
      "grad_norm": 0.303975373506546,
      "learning_rate": 1.5861111111111112e-05,
      "loss": 0.0583,
      "step": 1788
    },
    {
      "epoch": 0.20706018518518518,
      "grad_norm": 67.15092468261719,
      "learning_rate": 1.5858796296296296e-05,
      "loss": 1.5992,
      "step": 1789
    },
    {
      "epoch": 0.20717592592592593,
      "grad_norm": 0.38061606884002686,
      "learning_rate": 1.5856481481481484e-05,
      "loss": 0.0747,
      "step": 1790
    },
    {
      "epoch": 0.20729166666666668,
      "grad_norm": 0.3472925126552582,
      "learning_rate": 1.5854166666666668e-05,
      "loss": 0.0686,
      "step": 1791
    },
    {
      "epoch": 0.2074074074074074,
      "grad_norm": 0.2973182797431946,
      "learning_rate": 1.5851851851851852e-05,
      "loss": 0.0599,
      "step": 1792
    },
    {
      "epoch": 0.20752314814814815,
      "grad_norm": 0.40659913420677185,
      "learning_rate": 1.584953703703704e-05,
      "loss": 0.0605,
      "step": 1793
    },
    {
      "epoch": 0.2076388888888889,
      "grad_norm": 0.4248120188713074,
      "learning_rate": 1.5847222222222224e-05,
      "loss": 0.074,
      "step": 1794
    },
    {
      "epoch": 0.20775462962962962,
      "grad_norm": 0.4214966893196106,
      "learning_rate": 1.584490740740741e-05,
      "loss": 0.06,
      "step": 1795
    },
    {
      "epoch": 0.20787037037037037,
      "grad_norm": 0.37065449357032776,
      "learning_rate": 1.5842592592592593e-05,
      "loss": 0.0544,
      "step": 1796
    },
    {
      "epoch": 0.20798611111111112,
      "grad_norm": 0.36738941073417664,
      "learning_rate": 1.584027777777778e-05,
      "loss": 0.0609,
      "step": 1797
    },
    {
      "epoch": 0.20810185185185184,
      "grad_norm": 9.033647537231445,
      "learning_rate": 1.5837962962962965e-05,
      "loss": 2.2229,
      "step": 1798
    },
    {
      "epoch": 0.2082175925925926,
      "grad_norm": 0.3986337184906006,
      "learning_rate": 1.5835648148148152e-05,
      "loss": 0.0692,
      "step": 1799
    },
    {
      "epoch": 0.20833333333333334,
      "grad_norm": 0.5073125958442688,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 0.0916,
      "step": 1800
    },
    {
      "epoch": 0.20844907407407406,
      "grad_norm": 0.3993876576423645,
      "learning_rate": 1.583101851851852e-05,
      "loss": 0.0785,
      "step": 1801
    },
    {
      "epoch": 0.20856481481481481,
      "grad_norm": 0.261325478553772,
      "learning_rate": 1.5828703703703705e-05,
      "loss": 0.0518,
      "step": 1802
    },
    {
      "epoch": 0.20868055555555556,
      "grad_norm": 0.6133962869644165,
      "learning_rate": 1.582638888888889e-05,
      "loss": 0.085,
      "step": 1803
    },
    {
      "epoch": 0.20879629629629629,
      "grad_norm": 0.42893272638320923,
      "learning_rate": 1.5824074074074077e-05,
      "loss": 0.0826,
      "step": 1804
    },
    {
      "epoch": 0.20891203703703703,
      "grad_norm": 0.2870757281780243,
      "learning_rate": 1.582175925925926e-05,
      "loss": 0.0583,
      "step": 1805
    },
    {
      "epoch": 0.20902777777777778,
      "grad_norm": 0.9761524796485901,
      "learning_rate": 1.5819444444444445e-05,
      "loss": 0.0977,
      "step": 1806
    },
    {
      "epoch": 0.2091435185185185,
      "grad_norm": 0.5633363723754883,
      "learning_rate": 1.581712962962963e-05,
      "loss": 0.0827,
      "step": 1807
    },
    {
      "epoch": 0.20925925925925926,
      "grad_norm": 0.8470073342323303,
      "learning_rate": 1.5814814814814817e-05,
      "loss": 0.0946,
      "step": 1808
    },
    {
      "epoch": 0.209375,
      "grad_norm": 0.44411784410476685,
      "learning_rate": 1.58125e-05,
      "loss": 0.0645,
      "step": 1809
    },
    {
      "epoch": 0.20949074074074073,
      "grad_norm": 10.809115409851074,
      "learning_rate": 1.5810185185185186e-05,
      "loss": 2.406,
      "step": 1810
    },
    {
      "epoch": 0.20960648148148148,
      "grad_norm": 39.43190002441406,
      "learning_rate": 1.580787037037037e-05,
      "loss": 0.2354,
      "step": 1811
    },
    {
      "epoch": 0.20972222222222223,
      "grad_norm": 3.4850292205810547,
      "learning_rate": 1.5805555555555558e-05,
      "loss": 0.1093,
      "step": 1812
    },
    {
      "epoch": 0.20983796296296298,
      "grad_norm": 0.2919037938117981,
      "learning_rate": 1.5803240740740742e-05,
      "loss": 0.0588,
      "step": 1813
    },
    {
      "epoch": 0.2099537037037037,
      "grad_norm": 1.6813102960586548,
      "learning_rate": 1.5800925925925926e-05,
      "loss": 0.0897,
      "step": 1814
    },
    {
      "epoch": 0.21006944444444445,
      "grad_norm": 0.35497257113456726,
      "learning_rate": 1.5798611111111114e-05,
      "loss": 0.0683,
      "step": 1815
    },
    {
      "epoch": 0.2101851851851852,
      "grad_norm": 74.0608901977539,
      "learning_rate": 1.5796296296296298e-05,
      "loss": 2.1439,
      "step": 1816
    },
    {
      "epoch": 0.21030092592592592,
      "grad_norm": 0.34534838795661926,
      "learning_rate": 1.5793981481481482e-05,
      "loss": 0.0682,
      "step": 1817
    },
    {
      "epoch": 0.21041666666666667,
      "grad_norm": 0.6394463777542114,
      "learning_rate": 1.5791666666666667e-05,
      "loss": 0.0734,
      "step": 1818
    },
    {
      "epoch": 0.21053240740740742,
      "grad_norm": 0.25171804428100586,
      "learning_rate": 1.5789351851851854e-05,
      "loss": 0.0506,
      "step": 1819
    },
    {
      "epoch": 0.21064814814814814,
      "grad_norm": 0.3756989538669586,
      "learning_rate": 1.578703703703704e-05,
      "loss": 0.0752,
      "step": 1820
    },
    {
      "epoch": 0.2107638888888889,
      "grad_norm": 17.18952178955078,
      "learning_rate": 1.5784722222222223e-05,
      "loss": 0.1794,
      "step": 1821
    },
    {
      "epoch": 0.21087962962962964,
      "grad_norm": 1.6242107152938843,
      "learning_rate": 1.578240740740741e-05,
      "loss": 0.0858,
      "step": 1822
    },
    {
      "epoch": 0.21099537037037036,
      "grad_norm": 0.3381679356098175,
      "learning_rate": 1.5780092592592595e-05,
      "loss": 0.069,
      "step": 1823
    },
    {
      "epoch": 0.2111111111111111,
      "grad_norm": 0.3184848427772522,
      "learning_rate": 1.577777777777778e-05,
      "loss": 0.0624,
      "step": 1824
    },
    {
      "epoch": 0.21122685185185186,
      "grad_norm": 0.33348897099494934,
      "learning_rate": 1.5775462962962963e-05,
      "loss": 0.056,
      "step": 1825
    },
    {
      "epoch": 0.21134259259259258,
      "grad_norm": 1.2514586448669434,
      "learning_rate": 1.577314814814815e-05,
      "loss": 0.0812,
      "step": 1826
    },
    {
      "epoch": 0.21145833333333333,
      "grad_norm": 0.7205459475517273,
      "learning_rate": 1.5770833333333335e-05,
      "loss": 0.0948,
      "step": 1827
    },
    {
      "epoch": 0.21157407407407408,
      "grad_norm": 1.1990761756896973,
      "learning_rate": 1.576851851851852e-05,
      "loss": 0.088,
      "step": 1828
    },
    {
      "epoch": 0.2116898148148148,
      "grad_norm": 6.055995464324951,
      "learning_rate": 1.5766203703703703e-05,
      "loss": 0.1225,
      "step": 1829
    },
    {
      "epoch": 0.21180555555555555,
      "grad_norm": 31.993288040161133,
      "learning_rate": 1.576388888888889e-05,
      "loss": 0.3055,
      "step": 1830
    },
    {
      "epoch": 0.2119212962962963,
      "grad_norm": 1.7046650648117065,
      "learning_rate": 1.5761574074074075e-05,
      "loss": 0.0826,
      "step": 1831
    },
    {
      "epoch": 0.21203703703703702,
      "grad_norm": 60.460514068603516,
      "learning_rate": 1.575925925925926e-05,
      "loss": 1.2761,
      "step": 1832
    },
    {
      "epoch": 0.21215277777777777,
      "grad_norm": 7.888660430908203,
      "learning_rate": 1.5756944444444447e-05,
      "loss": 2.2301,
      "step": 1833
    },
    {
      "epoch": 0.21226851851851852,
      "grad_norm": 0.33722200989723206,
      "learning_rate": 1.575462962962963e-05,
      "loss": 0.0661,
      "step": 1834
    },
    {
      "epoch": 0.21238425925925927,
      "grad_norm": 5.911438941955566,
      "learning_rate": 1.5752314814814816e-05,
      "loss": 0.1037,
      "step": 1835
    },
    {
      "epoch": 0.2125,
      "grad_norm": 0.31203266978263855,
      "learning_rate": 1.575e-05,
      "loss": 0.0535,
      "step": 1836
    },
    {
      "epoch": 0.21261574074074074,
      "grad_norm": 0.2764093279838562,
      "learning_rate": 1.5747685185185188e-05,
      "loss": 0.0555,
      "step": 1837
    },
    {
      "epoch": 0.2127314814814815,
      "grad_norm": 15.590126037597656,
      "learning_rate": 1.5745370370370372e-05,
      "loss": 1.9244,
      "step": 1838
    },
    {
      "epoch": 0.21284722222222222,
      "grad_norm": 26.515043258666992,
      "learning_rate": 1.5743055555555556e-05,
      "loss": 1.7637,
      "step": 1839
    },
    {
      "epoch": 0.21296296296296297,
      "grad_norm": 95.54999542236328,
      "learning_rate": 1.5740740740740744e-05,
      "loss": 0.7031,
      "step": 1840
    },
    {
      "epoch": 0.21307870370370371,
      "grad_norm": 0.27221789956092834,
      "learning_rate": 1.5738425925925928e-05,
      "loss": 0.0517,
      "step": 1841
    },
    {
      "epoch": 0.21319444444444444,
      "grad_norm": 0.6156280040740967,
      "learning_rate": 1.5736111111111112e-05,
      "loss": 0.0686,
      "step": 1842
    },
    {
      "epoch": 0.21331018518518519,
      "grad_norm": 0.407029926776886,
      "learning_rate": 1.5733796296296297e-05,
      "loss": 0.081,
      "step": 1843
    },
    {
      "epoch": 0.21342592592592594,
      "grad_norm": 0.44300970435142517,
      "learning_rate": 1.5731481481481484e-05,
      "loss": 0.0712,
      "step": 1844
    },
    {
      "epoch": 0.21354166666666666,
      "grad_norm": 6.640827178955078,
      "learning_rate": 1.572916666666667e-05,
      "loss": 0.128,
      "step": 1845
    },
    {
      "epoch": 0.2136574074074074,
      "grad_norm": 0.7153841257095337,
      "learning_rate": 1.5726851851851853e-05,
      "loss": 0.1078,
      "step": 1846
    },
    {
      "epoch": 0.21377314814814816,
      "grad_norm": 0.2961915135383606,
      "learning_rate": 1.5724537037037037e-05,
      "loss": 0.0598,
      "step": 1847
    },
    {
      "epoch": 0.21388888888888888,
      "grad_norm": 0.4827454388141632,
      "learning_rate": 1.5722222222222225e-05,
      "loss": 0.0563,
      "step": 1848
    },
    {
      "epoch": 0.21400462962962963,
      "grad_norm": 0.5565133094787598,
      "learning_rate": 1.571990740740741e-05,
      "loss": 0.0844,
      "step": 1849
    },
    {
      "epoch": 0.21412037037037038,
      "grad_norm": 3.1297502517700195,
      "learning_rate": 1.5717592592592593e-05,
      "loss": 0.1488,
      "step": 1850
    },
    {
      "epoch": 0.2142361111111111,
      "grad_norm": 7.795688629150391,
      "learning_rate": 1.571527777777778e-05,
      "loss": 2.3131,
      "step": 1851
    },
    {
      "epoch": 0.21435185185185185,
      "grad_norm": 39.13972854614258,
      "learning_rate": 1.571296296296296e-05,
      "loss": 1.1147,
      "step": 1852
    },
    {
      "epoch": 0.2144675925925926,
      "grad_norm": 0.251143217086792,
      "learning_rate": 1.571064814814815e-05,
      "loss": 0.0501,
      "step": 1853
    },
    {
      "epoch": 0.21458333333333332,
      "grad_norm": 0.40630781650543213,
      "learning_rate": 1.5708333333333333e-05,
      "loss": 0.0783,
      "step": 1854
    },
    {
      "epoch": 0.21469907407407407,
      "grad_norm": 0.4287494719028473,
      "learning_rate": 1.570601851851852e-05,
      "loss": 0.0727,
      "step": 1855
    },
    {
      "epoch": 0.21481481481481482,
      "grad_norm": 53.7556037902832,
      "learning_rate": 1.5703703703703705e-05,
      "loss": 0.5071,
      "step": 1856
    },
    {
      "epoch": 0.21493055555555557,
      "grad_norm": 0.35318705439567566,
      "learning_rate": 1.570138888888889e-05,
      "loss": 0.0637,
      "step": 1857
    },
    {
      "epoch": 0.2150462962962963,
      "grad_norm": 0.42219290137290955,
      "learning_rate": 1.5699074074074074e-05,
      "loss": 0.0817,
      "step": 1858
    },
    {
      "epoch": 0.21516203703703704,
      "grad_norm": 0.4070034325122833,
      "learning_rate": 1.569675925925926e-05,
      "loss": 0.0743,
      "step": 1859
    },
    {
      "epoch": 0.2152777777777778,
      "grad_norm": 0.4019594192504883,
      "learning_rate": 1.5694444444444446e-05,
      "loss": 0.0759,
      "step": 1860
    },
    {
      "epoch": 0.2153935185185185,
      "grad_norm": 0.3002083897590637,
      "learning_rate": 1.569212962962963e-05,
      "loss": 0.0592,
      "step": 1861
    },
    {
      "epoch": 0.21550925925925926,
      "grad_norm": 4.797572135925293,
      "learning_rate": 1.5689814814814818e-05,
      "loss": 0.1321,
      "step": 1862
    },
    {
      "epoch": 0.215625,
      "grad_norm": 32.436920166015625,
      "learning_rate": 1.5687500000000002e-05,
      "loss": 0.455,
      "step": 1863
    },
    {
      "epoch": 0.21574074074074073,
      "grad_norm": 0.4785502254962921,
      "learning_rate": 1.5685185185185186e-05,
      "loss": 0.0585,
      "step": 1864
    },
    {
      "epoch": 0.21585648148148148,
      "grad_norm": 0.24487702548503876,
      "learning_rate": 1.568287037037037e-05,
      "loss": 0.0488,
      "step": 1865
    },
    {
      "epoch": 0.21597222222222223,
      "grad_norm": 0.3526700735092163,
      "learning_rate": 1.5680555555555558e-05,
      "loss": 0.0681,
      "step": 1866
    },
    {
      "epoch": 0.21608796296296295,
      "grad_norm": 36.101158142089844,
      "learning_rate": 1.5678240740740742e-05,
      "loss": 0.2578,
      "step": 1867
    },
    {
      "epoch": 0.2162037037037037,
      "grad_norm": 0.44517818093299866,
      "learning_rate": 1.5675925925925926e-05,
      "loss": 0.0765,
      "step": 1868
    },
    {
      "epoch": 0.21631944444444445,
      "grad_norm": 0.3116911053657532,
      "learning_rate": 1.5673611111111114e-05,
      "loss": 0.0613,
      "step": 1869
    },
    {
      "epoch": 0.21643518518518517,
      "grad_norm": 0.35441824793815613,
      "learning_rate": 1.5671296296296295e-05,
      "loss": 0.071,
      "step": 1870
    },
    {
      "epoch": 0.21655092592592592,
      "grad_norm": 0.9022730588912964,
      "learning_rate": 1.5668981481481483e-05,
      "loss": 0.0548,
      "step": 1871
    },
    {
      "epoch": 0.21666666666666667,
      "grad_norm": 0.9667634963989258,
      "learning_rate": 1.5666666666666667e-05,
      "loss": 0.0708,
      "step": 1872
    },
    {
      "epoch": 0.2167824074074074,
      "grad_norm": 0.3657536506652832,
      "learning_rate": 1.5664351851851855e-05,
      "loss": 0.0738,
      "step": 1873
    },
    {
      "epoch": 0.21689814814814815,
      "grad_norm": 0.2706827223300934,
      "learning_rate": 1.566203703703704e-05,
      "loss": 0.0541,
      "step": 1874
    },
    {
      "epoch": 0.2170138888888889,
      "grad_norm": 77.91734313964844,
      "learning_rate": 1.5659722222222223e-05,
      "loss": 0.5393,
      "step": 1875
    },
    {
      "epoch": 0.21712962962962962,
      "grad_norm": 0.3992881774902344,
      "learning_rate": 1.5657407407407407e-05,
      "loss": 0.0771,
      "step": 1876
    },
    {
      "epoch": 0.21724537037037037,
      "grad_norm": 0.3908001184463501,
      "learning_rate": 1.5655092592592595e-05,
      "loss": 0.0762,
      "step": 1877
    },
    {
      "epoch": 0.21736111111111112,
      "grad_norm": 0.25342822074890137,
      "learning_rate": 1.565277777777778e-05,
      "loss": 0.0516,
      "step": 1878
    },
    {
      "epoch": 0.21747685185185187,
      "grad_norm": 0.3997209966182709,
      "learning_rate": 1.5650462962962963e-05,
      "loss": 0.0802,
      "step": 1879
    },
    {
      "epoch": 0.2175925925925926,
      "grad_norm": 0.36104458570480347,
      "learning_rate": 1.564814814814815e-05,
      "loss": 0.0511,
      "step": 1880
    },
    {
      "epoch": 0.21770833333333334,
      "grad_norm": 0.4271727204322815,
      "learning_rate": 1.5645833333333335e-05,
      "loss": 0.0723,
      "step": 1881
    },
    {
      "epoch": 0.2178240740740741,
      "grad_norm": 0.3073823153972626,
      "learning_rate": 1.564351851851852e-05,
      "loss": 0.0595,
      "step": 1882
    },
    {
      "epoch": 0.2179398148148148,
      "grad_norm": 47.3206901550293,
      "learning_rate": 1.5641203703703704e-05,
      "loss": 1.4244,
      "step": 1883
    },
    {
      "epoch": 0.21805555555555556,
      "grad_norm": 19.61325454711914,
      "learning_rate": 1.563888888888889e-05,
      "loss": 2.0847,
      "step": 1884
    },
    {
      "epoch": 0.2181712962962963,
      "grad_norm": 0.38378769159317017,
      "learning_rate": 1.5636574074074076e-05,
      "loss": 0.0699,
      "step": 1885
    },
    {
      "epoch": 0.21828703703703703,
      "grad_norm": 0.307911217212677,
      "learning_rate": 1.563425925925926e-05,
      "loss": 0.0608,
      "step": 1886
    },
    {
      "epoch": 0.21840277777777778,
      "grad_norm": 0.6625904440879822,
      "learning_rate": 1.5631944444444448e-05,
      "loss": 0.0997,
      "step": 1887
    },
    {
      "epoch": 0.21851851851851853,
      "grad_norm": 0.4435017704963684,
      "learning_rate": 1.5629629629629632e-05,
      "loss": 0.0781,
      "step": 1888
    },
    {
      "epoch": 0.21863425925925925,
      "grad_norm": 0.3157535195350647,
      "learning_rate": 1.5627314814814816e-05,
      "loss": 0.0635,
      "step": 1889
    },
    {
      "epoch": 0.21875,
      "grad_norm": 0.36589232087135315,
      "learning_rate": 1.5625e-05,
      "loss": 0.0684,
      "step": 1890
    },
    {
      "epoch": 0.21886574074074075,
      "grad_norm": 1.2815014123916626,
      "learning_rate": 1.5622685185185188e-05,
      "loss": 0.0986,
      "step": 1891
    },
    {
      "epoch": 0.21898148148148147,
      "grad_norm": 0.4252651035785675,
      "learning_rate": 1.5620370370370372e-05,
      "loss": 0.0859,
      "step": 1892
    },
    {
      "epoch": 0.21909722222222222,
      "grad_norm": 0.2668004035949707,
      "learning_rate": 1.5618055555555556e-05,
      "loss": 0.0528,
      "step": 1893
    },
    {
      "epoch": 0.21921296296296297,
      "grad_norm": 0.409031480550766,
      "learning_rate": 1.561574074074074e-05,
      "loss": 0.0766,
      "step": 1894
    },
    {
      "epoch": 0.2193287037037037,
      "grad_norm": 0.49011996388435364,
      "learning_rate": 1.561342592592593e-05,
      "loss": 0.0723,
      "step": 1895
    },
    {
      "epoch": 0.21944444444444444,
      "grad_norm": 0.4031115174293518,
      "learning_rate": 1.5611111111111113e-05,
      "loss": 0.0586,
      "step": 1896
    },
    {
      "epoch": 0.2195601851851852,
      "grad_norm": 0.2586921155452728,
      "learning_rate": 1.5608796296296297e-05,
      "loss": 0.0513,
      "step": 1897
    },
    {
      "epoch": 0.2196759259259259,
      "grad_norm": 0.5278580784797668,
      "learning_rate": 1.5606481481481484e-05,
      "loss": 0.0909,
      "step": 1898
    },
    {
      "epoch": 0.21979166666666666,
      "grad_norm": 13.196297645568848,
      "learning_rate": 1.5604166666666665e-05,
      "loss": 1.9878,
      "step": 1899
    },
    {
      "epoch": 0.2199074074074074,
      "grad_norm": 0.2559140622615814,
      "learning_rate": 1.5601851851851853e-05,
      "loss": 0.0513,
      "step": 1900
    },
    {
      "epoch": 0.22002314814814813,
      "grad_norm": 25.485490798950195,
      "learning_rate": 1.5599537037037037e-05,
      "loss": 1.6101,
      "step": 1901
    },
    {
      "epoch": 0.22013888888888888,
      "grad_norm": 0.3374015986919403,
      "learning_rate": 1.5597222222222225e-05,
      "loss": 0.0522,
      "step": 1902
    },
    {
      "epoch": 0.22025462962962963,
      "grad_norm": 2.805548906326294,
      "learning_rate": 1.559490740740741e-05,
      "loss": 0.0929,
      "step": 1903
    },
    {
      "epoch": 0.22037037037037038,
      "grad_norm": 0.3112861216068268,
      "learning_rate": 1.5592592592592593e-05,
      "loss": 0.0606,
      "step": 1904
    },
    {
      "epoch": 0.2204861111111111,
      "grad_norm": 0.30733782052993774,
      "learning_rate": 1.5590277777777778e-05,
      "loss": 0.062,
      "step": 1905
    },
    {
      "epoch": 0.22060185185185185,
      "grad_norm": 0.404299259185791,
      "learning_rate": 1.5587962962962965e-05,
      "loss": 0.0735,
      "step": 1906
    },
    {
      "epoch": 0.2207175925925926,
      "grad_norm": 0.25092387199401855,
      "learning_rate": 1.558564814814815e-05,
      "loss": 0.0501,
      "step": 1907
    },
    {
      "epoch": 0.22083333333333333,
      "grad_norm": 3.6766293048858643,
      "learning_rate": 1.5583333333333334e-05,
      "loss": 0.0978,
      "step": 1908
    },
    {
      "epoch": 0.22094907407407408,
      "grad_norm": 0.42017173767089844,
      "learning_rate": 1.558101851851852e-05,
      "loss": 0.0847,
      "step": 1909
    },
    {
      "epoch": 0.22106481481481483,
      "grad_norm": 1.0491766929626465,
      "learning_rate": 1.5578703703703706e-05,
      "loss": 0.0873,
      "step": 1910
    },
    {
      "epoch": 0.22118055555555555,
      "grad_norm": 0.4369349479675293,
      "learning_rate": 1.557638888888889e-05,
      "loss": 0.0867,
      "step": 1911
    },
    {
      "epoch": 0.2212962962962963,
      "grad_norm": 0.33054229617118835,
      "learning_rate": 1.5574074074074074e-05,
      "loss": 0.0476,
      "step": 1912
    },
    {
      "epoch": 0.22141203703703705,
      "grad_norm": 0.3391789495944977,
      "learning_rate": 1.5571759259259262e-05,
      "loss": 0.0644,
      "step": 1913
    },
    {
      "epoch": 0.22152777777777777,
      "grad_norm": 0.4852849245071411,
      "learning_rate": 1.5569444444444446e-05,
      "loss": 0.078,
      "step": 1914
    },
    {
      "epoch": 0.22164351851851852,
      "grad_norm": 0.23568041622638702,
      "learning_rate": 1.556712962962963e-05,
      "loss": 0.0474,
      "step": 1915
    },
    {
      "epoch": 0.22175925925925927,
      "grad_norm": 0.4266975224018097,
      "learning_rate": 1.5564814814814818e-05,
      "loss": 0.0838,
      "step": 1916
    },
    {
      "epoch": 0.221875,
      "grad_norm": 0.32352691888809204,
      "learning_rate": 1.55625e-05,
      "loss": 0.0604,
      "step": 1917
    },
    {
      "epoch": 0.22199074074074074,
      "grad_norm": 0.5791062712669373,
      "learning_rate": 1.5560185185185186e-05,
      "loss": 0.055,
      "step": 1918
    },
    {
      "epoch": 0.2221064814814815,
      "grad_norm": 0.3160233497619629,
      "learning_rate": 1.555787037037037e-05,
      "loss": 0.0605,
      "step": 1919
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 0.37211671471595764,
      "learning_rate": 1.555555555555556e-05,
      "loss": 0.0708,
      "step": 1920
    },
    {
      "epoch": 0.22233796296296296,
      "grad_norm": 10.709263801574707,
      "learning_rate": 1.5553240740740743e-05,
      "loss": 0.1433,
      "step": 1921
    },
    {
      "epoch": 0.2224537037037037,
      "grad_norm": 5.381999969482422,
      "learning_rate": 1.5550925925925927e-05,
      "loss": 2.4671,
      "step": 1922
    },
    {
      "epoch": 0.22256944444444443,
      "grad_norm": 0.521543025970459,
      "learning_rate": 1.554861111111111e-05,
      "loss": 0.0938,
      "step": 1923
    },
    {
      "epoch": 0.22268518518518518,
      "grad_norm": 0.2523115575313568,
      "learning_rate": 1.55462962962963e-05,
      "loss": 0.0508,
      "step": 1924
    },
    {
      "epoch": 0.22280092592592593,
      "grad_norm": 0.5938772559165955,
      "learning_rate": 1.5543981481481483e-05,
      "loss": 0.0983,
      "step": 1925
    },
    {
      "epoch": 0.22291666666666668,
      "grad_norm": 0.3705470561981201,
      "learning_rate": 1.5541666666666667e-05,
      "loss": 0.071,
      "step": 1926
    },
    {
      "epoch": 0.2230324074074074,
      "grad_norm": 0.2931061089038849,
      "learning_rate": 1.5539351851851855e-05,
      "loss": 0.0564,
      "step": 1927
    },
    {
      "epoch": 0.22314814814814815,
      "grad_norm": 0.29363974928855896,
      "learning_rate": 1.553703703703704e-05,
      "loss": 0.058,
      "step": 1928
    },
    {
      "epoch": 0.2232638888888889,
      "grad_norm": 20.67892837524414,
      "learning_rate": 1.5534722222222223e-05,
      "loss": 0.2868,
      "step": 1929
    },
    {
      "epoch": 0.22337962962962962,
      "grad_norm": 0.3539491891860962,
      "learning_rate": 1.5532407407407408e-05,
      "loss": 0.0689,
      "step": 1930
    },
    {
      "epoch": 0.22349537037037037,
      "grad_norm": 0.4086114168167114,
      "learning_rate": 1.5530092592592595e-05,
      "loss": 0.0818,
      "step": 1931
    },
    {
      "epoch": 0.22361111111111112,
      "grad_norm": 0.4375438988208771,
      "learning_rate": 1.552777777777778e-05,
      "loss": 0.0886,
      "step": 1932
    },
    {
      "epoch": 0.22372685185185184,
      "grad_norm": 2.5846986770629883,
      "learning_rate": 1.5525462962962964e-05,
      "loss": 0.124,
      "step": 1933
    },
    {
      "epoch": 0.2238425925925926,
      "grad_norm": 1.3011082410812378,
      "learning_rate": 1.552314814814815e-05,
      "loss": 0.0838,
      "step": 1934
    },
    {
      "epoch": 0.22395833333333334,
      "grad_norm": 8.704487800598145,
      "learning_rate": 1.5520833333333332e-05,
      "loss": 2.2021,
      "step": 1935
    },
    {
      "epoch": 0.22407407407407406,
      "grad_norm": 84.93345642089844,
      "learning_rate": 1.551851851851852e-05,
      "loss": 0.9608,
      "step": 1936
    },
    {
      "epoch": 0.22418981481481481,
      "grad_norm": 0.3384823799133301,
      "learning_rate": 1.5516203703703704e-05,
      "loss": 0.0595,
      "step": 1937
    },
    {
      "epoch": 0.22430555555555556,
      "grad_norm": 0.3348236083984375,
      "learning_rate": 1.5513888888888892e-05,
      "loss": 0.0606,
      "step": 1938
    },
    {
      "epoch": 0.22442129629629629,
      "grad_norm": 0.2840333878993988,
      "learning_rate": 1.5511574074074076e-05,
      "loss": 0.0566,
      "step": 1939
    },
    {
      "epoch": 0.22453703703703703,
      "grad_norm": 7.463122367858887,
      "learning_rate": 1.550925925925926e-05,
      "loss": 0.133,
      "step": 1940
    },
    {
      "epoch": 0.22465277777777778,
      "grad_norm": 0.23084025084972382,
      "learning_rate": 1.5506944444444445e-05,
      "loss": 0.0465,
      "step": 1941
    },
    {
      "epoch": 0.2247685185185185,
      "grad_norm": 0.2681814730167389,
      "learning_rate": 1.5504629629629632e-05,
      "loss": 0.0527,
      "step": 1942
    },
    {
      "epoch": 0.22488425925925926,
      "grad_norm": 0.3985310196876526,
      "learning_rate": 1.5502314814814816e-05,
      "loss": 0.0767,
      "step": 1943
    },
    {
      "epoch": 0.225,
      "grad_norm": 0.2569759786128998,
      "learning_rate": 1.55e-05,
      "loss": 0.0516,
      "step": 1944
    },
    {
      "epoch": 0.22511574074074073,
      "grad_norm": 12.46850872039795,
      "learning_rate": 1.549768518518519e-05,
      "loss": 2.3607,
      "step": 1945
    },
    {
      "epoch": 0.22523148148148148,
      "grad_norm": 0.4469014108181,
      "learning_rate": 1.549537037037037e-05,
      "loss": 0.0506,
      "step": 1946
    },
    {
      "epoch": 0.22534722222222223,
      "grad_norm": 0.2986953556537628,
      "learning_rate": 1.5493055555555557e-05,
      "loss": 0.0517,
      "step": 1947
    },
    {
      "epoch": 0.22546296296296298,
      "grad_norm": 0.30300936102867126,
      "learning_rate": 1.549074074074074e-05,
      "loss": 0.0587,
      "step": 1948
    },
    {
      "epoch": 0.2255787037037037,
      "grad_norm": 0.23745425045490265,
      "learning_rate": 1.548842592592593e-05,
      "loss": 0.0471,
      "step": 1949
    },
    {
      "epoch": 0.22569444444444445,
      "grad_norm": 0.9539612531661987,
      "learning_rate": 1.5486111111111113e-05,
      "loss": 0.0832,
      "step": 1950
    },
    {
      "epoch": 0.2258101851851852,
      "grad_norm": 0.7792121767997742,
      "learning_rate": 1.5483796296296297e-05,
      "loss": 0.0952,
      "step": 1951
    },
    {
      "epoch": 0.22592592592592592,
      "grad_norm": 0.8339198231697083,
      "learning_rate": 1.548148148148148e-05,
      "loss": 0.0917,
      "step": 1952
    },
    {
      "epoch": 0.22604166666666667,
      "grad_norm": 0.5669129490852356,
      "learning_rate": 1.5479166666666666e-05,
      "loss": 0.0508,
      "step": 1953
    },
    {
      "epoch": 0.22615740740740742,
      "grad_norm": 0.25052839517593384,
      "learning_rate": 1.5476851851851853e-05,
      "loss": 0.0496,
      "step": 1954
    },
    {
      "epoch": 0.22627314814814814,
      "grad_norm": 14.205766677856445,
      "learning_rate": 1.5474537037037038e-05,
      "loss": 0.1366,
      "step": 1955
    },
    {
      "epoch": 0.2263888888888889,
      "grad_norm": 0.25508934259414673,
      "learning_rate": 1.5472222222222225e-05,
      "loss": 0.051,
      "step": 1956
    },
    {
      "epoch": 0.22650462962962964,
      "grad_norm": 0.3435778021812439,
      "learning_rate": 1.546990740740741e-05,
      "loss": 0.0592,
      "step": 1957
    },
    {
      "epoch": 0.22662037037037036,
      "grad_norm": 65.27023315429688,
      "learning_rate": 1.5467592592592594e-05,
      "loss": 2.0488,
      "step": 1958
    },
    {
      "epoch": 0.2267361111111111,
      "grad_norm": 0.23312845826148987,
      "learning_rate": 1.5465277777777778e-05,
      "loss": 0.0463,
      "step": 1959
    },
    {
      "epoch": 0.22685185185185186,
      "grad_norm": 0.28551581501960754,
      "learning_rate": 1.5462962962962966e-05,
      "loss": 0.0569,
      "step": 1960
    },
    {
      "epoch": 0.22696759259259258,
      "grad_norm": 0.25021448731422424,
      "learning_rate": 1.546064814814815e-05,
      "loss": 0.0495,
      "step": 1961
    },
    {
      "epoch": 0.22708333333333333,
      "grad_norm": 0.3745468556880951,
      "learning_rate": 1.5458333333333334e-05,
      "loss": 0.0731,
      "step": 1962
    },
    {
      "epoch": 0.22719907407407408,
      "grad_norm": 0.5058594942092896,
      "learning_rate": 1.5456018518518522e-05,
      "loss": 0.0798,
      "step": 1963
    },
    {
      "epoch": 0.2273148148148148,
      "grad_norm": 0.5141664743423462,
      "learning_rate": 1.5453703703703703e-05,
      "loss": 0.0922,
      "step": 1964
    },
    {
      "epoch": 0.22743055555555555,
      "grad_norm": 0.37888050079345703,
      "learning_rate": 1.545138888888889e-05,
      "loss": 0.0748,
      "step": 1965
    },
    {
      "epoch": 0.2275462962962963,
      "grad_norm": 0.7513548135757446,
      "learning_rate": 1.5449074074074075e-05,
      "loss": 0.0988,
      "step": 1966
    },
    {
      "epoch": 0.22766203703703702,
      "grad_norm": 7.82303524017334,
      "learning_rate": 1.5446759259259262e-05,
      "loss": 0.1191,
      "step": 1967
    },
    {
      "epoch": 0.22777777777777777,
      "grad_norm": 0.6288726925849915,
      "learning_rate": 1.5444444444444446e-05,
      "loss": 0.0925,
      "step": 1968
    },
    {
      "epoch": 0.22789351851851852,
      "grad_norm": 0.6215120553970337,
      "learning_rate": 1.544212962962963e-05,
      "loss": 0.0566,
      "step": 1969
    },
    {
      "epoch": 0.22800925925925927,
      "grad_norm": 0.3233203887939453,
      "learning_rate": 1.5439814814814815e-05,
      "loss": 0.0606,
      "step": 1970
    },
    {
      "epoch": 0.228125,
      "grad_norm": 3.308267116546631,
      "learning_rate": 1.54375e-05,
      "loss": 0.087,
      "step": 1971
    },
    {
      "epoch": 0.22824074074074074,
      "grad_norm": 0.2388634979724884,
      "learning_rate": 1.5435185185185187e-05,
      "loss": 0.045,
      "step": 1972
    },
    {
      "epoch": 0.2283564814814815,
      "grad_norm": 0.36520856618881226,
      "learning_rate": 1.543287037037037e-05,
      "loss": 0.072,
      "step": 1973
    },
    {
      "epoch": 0.22847222222222222,
      "grad_norm": 0.3057728707790375,
      "learning_rate": 1.543055555555556e-05,
      "loss": 0.0442,
      "step": 1974
    },
    {
      "epoch": 0.22858796296296297,
      "grad_norm": 28.423446655273438,
      "learning_rate": 1.5428240740740743e-05,
      "loss": 0.2096,
      "step": 1975
    },
    {
      "epoch": 0.22870370370370371,
      "grad_norm": 0.34720176458358765,
      "learning_rate": 1.5425925925925927e-05,
      "loss": 0.0674,
      "step": 1976
    },
    {
      "epoch": 0.22881944444444444,
      "grad_norm": 0.26132696866989136,
      "learning_rate": 1.542361111111111e-05,
      "loss": 0.0505,
      "step": 1977
    },
    {
      "epoch": 0.22893518518518519,
      "grad_norm": 0.36556342244148254,
      "learning_rate": 1.54212962962963e-05,
      "loss": 0.0641,
      "step": 1978
    },
    {
      "epoch": 0.22905092592592594,
      "grad_norm": 0.40538978576660156,
      "learning_rate": 1.5418981481481483e-05,
      "loss": 0.0813,
      "step": 1979
    },
    {
      "epoch": 0.22916666666666666,
      "grad_norm": 0.3417617678642273,
      "learning_rate": 1.5416666666666668e-05,
      "loss": 0.0686,
      "step": 1980
    },
    {
      "epoch": 0.2292824074074074,
      "grad_norm": 39.742191314697266,
      "learning_rate": 1.5414351851851852e-05,
      "loss": 1.8733,
      "step": 1981
    },
    {
      "epoch": 0.22939814814814816,
      "grad_norm": 44.19801330566406,
      "learning_rate": 1.5412037037037036e-05,
      "loss": 0.6256,
      "step": 1982
    },
    {
      "epoch": 0.22951388888888888,
      "grad_norm": 0.2997654676437378,
      "learning_rate": 1.5409722222222224e-05,
      "loss": 0.0433,
      "step": 1983
    },
    {
      "epoch": 0.22962962962962963,
      "grad_norm": 0.261410653591156,
      "learning_rate": 1.5407407407407408e-05,
      "loss": 0.0506,
      "step": 1984
    },
    {
      "epoch": 0.22974537037037038,
      "grad_norm": 0.3411136269569397,
      "learning_rate": 1.5405092592592596e-05,
      "loss": 0.0671,
      "step": 1985
    },
    {
      "epoch": 0.2298611111111111,
      "grad_norm": 0.28295162320137024,
      "learning_rate": 1.540277777777778e-05,
      "loss": 0.0506,
      "step": 1986
    },
    {
      "epoch": 0.22997685185185185,
      "grad_norm": 0.4128044843673706,
      "learning_rate": 1.5400462962962964e-05,
      "loss": 0.0832,
      "step": 1987
    },
    {
      "epoch": 0.2300925925925926,
      "grad_norm": 28.7670841217041,
      "learning_rate": 1.539814814814815e-05,
      "loss": 1.8396,
      "step": 1988
    },
    {
      "epoch": 0.23020833333333332,
      "grad_norm": 0.37482497096061707,
      "learning_rate": 1.5395833333333333e-05,
      "loss": 0.0764,
      "step": 1989
    },
    {
      "epoch": 0.23032407407407407,
      "grad_norm": 54.71498107910156,
      "learning_rate": 1.539351851851852e-05,
      "loss": 1.6195,
      "step": 1990
    },
    {
      "epoch": 0.23043981481481482,
      "grad_norm": 0.4947289526462555,
      "learning_rate": 1.5391203703703704e-05,
      "loss": 0.0722,
      "step": 1991
    },
    {
      "epoch": 0.23055555555555557,
      "grad_norm": 26.775304794311523,
      "learning_rate": 1.5388888888888892e-05,
      "loss": 1.8285,
      "step": 1992
    },
    {
      "epoch": 0.2306712962962963,
      "grad_norm": 0.3156735301017761,
      "learning_rate": 1.5386574074074073e-05,
      "loss": 0.0445,
      "step": 1993
    },
    {
      "epoch": 0.23078703703703704,
      "grad_norm": 0.229491725564003,
      "learning_rate": 1.538425925925926e-05,
      "loss": 0.0452,
      "step": 1994
    },
    {
      "epoch": 0.2309027777777778,
      "grad_norm": 0.38509610295295715,
      "learning_rate": 1.5381944444444445e-05,
      "loss": 0.0701,
      "step": 1995
    },
    {
      "epoch": 0.2310185185185185,
      "grad_norm": 22.171401977539062,
      "learning_rate": 1.5379629629629633e-05,
      "loss": 1.8679,
      "step": 1996
    },
    {
      "epoch": 0.23113425925925926,
      "grad_norm": 0.2920364737510681,
      "learning_rate": 1.5377314814814817e-05,
      "loss": 0.0567,
      "step": 1997
    },
    {
      "epoch": 0.23125,
      "grad_norm": 0.42016634345054626,
      "learning_rate": 1.5375e-05,
      "loss": 0.0818,
      "step": 1998
    },
    {
      "epoch": 0.23136574074074073,
      "grad_norm": 0.48466041684150696,
      "learning_rate": 1.5372685185185185e-05,
      "loss": 0.0703,
      "step": 1999
    },
    {
      "epoch": 0.23148148148148148,
      "grad_norm": 0.3845212161540985,
      "learning_rate": 1.537037037037037e-05,
      "loss": 0.0661,
      "step": 2000
    },
    {
      "epoch": 0.23159722222222223,
      "grad_norm": 0.5700222253799438,
      "learning_rate": 1.5368055555555557e-05,
      "loss": 0.0641,
      "step": 2001
    },
    {
      "epoch": 0.23171296296296295,
      "grad_norm": 0.428099125623703,
      "learning_rate": 1.536574074074074e-05,
      "loss": 0.0551,
      "step": 2002
    },
    {
      "epoch": 0.2318287037037037,
      "grad_norm": 0.24670614302158356,
      "learning_rate": 1.536342592592593e-05,
      "loss": 0.0484,
      "step": 2003
    },
    {
      "epoch": 0.23194444444444445,
      "grad_norm": 0.28436577320098877,
      "learning_rate": 1.5361111111111113e-05,
      "loss": 0.0573,
      "step": 2004
    },
    {
      "epoch": 0.23206018518518517,
      "grad_norm": 0.5583627820014954,
      "learning_rate": 1.5358796296296298e-05,
      "loss": 0.0779,
      "step": 2005
    },
    {
      "epoch": 0.23217592592592592,
      "grad_norm": 0.521853506565094,
      "learning_rate": 1.5356481481481482e-05,
      "loss": 0.0768,
      "step": 2006
    },
    {
      "epoch": 0.23229166666666667,
      "grad_norm": 0.33920618891716003,
      "learning_rate": 1.535416666666667e-05,
      "loss": 0.0664,
      "step": 2007
    },
    {
      "epoch": 0.2324074074074074,
      "grad_norm": 95.15670013427734,
      "learning_rate": 1.5351851851851854e-05,
      "loss": 1.0336,
      "step": 2008
    },
    {
      "epoch": 0.23252314814814815,
      "grad_norm": 0.2396685928106308,
      "learning_rate": 1.5349537037037038e-05,
      "loss": 0.048,
      "step": 2009
    },
    {
      "epoch": 0.2326388888888889,
      "grad_norm": 0.3942984342575073,
      "learning_rate": 1.5347222222222226e-05,
      "loss": 0.0811,
      "step": 2010
    },
    {
      "epoch": 0.23275462962962962,
      "grad_norm": 0.3262227177619934,
      "learning_rate": 1.5344907407407406e-05,
      "loss": 0.0636,
      "step": 2011
    },
    {
      "epoch": 0.23287037037037037,
      "grad_norm": 0.3942925035953522,
      "learning_rate": 1.5342592592592594e-05,
      "loss": 0.0618,
      "step": 2012
    },
    {
      "epoch": 0.23298611111111112,
      "grad_norm": 62.56139373779297,
      "learning_rate": 1.534027777777778e-05,
      "loss": 0.4256,
      "step": 2013
    },
    {
      "epoch": 0.23310185185185187,
      "grad_norm": 82.9926986694336,
      "learning_rate": 1.5337962962962966e-05,
      "loss": 1.565,
      "step": 2014
    },
    {
      "epoch": 0.2332175925925926,
      "grad_norm": 0.4878005385398865,
      "learning_rate": 1.533564814814815e-05,
      "loss": 0.07,
      "step": 2015
    },
    {
      "epoch": 0.23333333333333334,
      "grad_norm": 0.7712349891662598,
      "learning_rate": 1.5333333333333334e-05,
      "loss": 0.0524,
      "step": 2016
    },
    {
      "epoch": 0.2334490740740741,
      "grad_norm": 0.39928311109542847,
      "learning_rate": 1.533101851851852e-05,
      "loss": 0.081,
      "step": 2017
    },
    {
      "epoch": 0.2335648148148148,
      "grad_norm": 0.3739263415336609,
      "learning_rate": 1.5328703703703703e-05,
      "loss": 0.0683,
      "step": 2018
    },
    {
      "epoch": 0.23368055555555556,
      "grad_norm": 0.37802860140800476,
      "learning_rate": 1.532638888888889e-05,
      "loss": 0.076,
      "step": 2019
    },
    {
      "epoch": 0.2337962962962963,
      "grad_norm": 0.7604965567588806,
      "learning_rate": 1.5324074074074075e-05,
      "loss": 0.0769,
      "step": 2020
    },
    {
      "epoch": 0.23391203703703703,
      "grad_norm": 0.22820423543453217,
      "learning_rate": 1.5321759259259263e-05,
      "loss": 0.0458,
      "step": 2021
    },
    {
      "epoch": 0.23402777777777778,
      "grad_norm": 2.5127968788146973,
      "learning_rate": 1.5319444444444447e-05,
      "loss": 0.0744,
      "step": 2022
    },
    {
      "epoch": 0.23414351851851853,
      "grad_norm": 0.5631744265556335,
      "learning_rate": 1.531712962962963e-05,
      "loss": 0.0735,
      "step": 2023
    },
    {
      "epoch": 0.23425925925925925,
      "grad_norm": 0.33266425132751465,
      "learning_rate": 1.5314814814814815e-05,
      "loss": 0.0642,
      "step": 2024
    },
    {
      "epoch": 0.234375,
      "grad_norm": 17.723318099975586,
      "learning_rate": 1.5312500000000003e-05,
      "loss": 2.0415,
      "step": 2025
    },
    {
      "epoch": 0.23449074074074075,
      "grad_norm": 0.30026090145111084,
      "learning_rate": 1.5310185185185187e-05,
      "loss": 0.0431,
      "step": 2026
    },
    {
      "epoch": 0.23460648148148147,
      "grad_norm": 0.3614397943019867,
      "learning_rate": 1.530787037037037e-05,
      "loss": 0.0584,
      "step": 2027
    },
    {
      "epoch": 0.23472222222222222,
      "grad_norm": 6.151769161224365,
      "learning_rate": 1.5305555555555556e-05,
      "loss": 2.2609,
      "step": 2028
    },
    {
      "epoch": 0.23483796296296297,
      "grad_norm": 0.5001289248466492,
      "learning_rate": 1.530324074074074e-05,
      "loss": 0.064,
      "step": 2029
    },
    {
      "epoch": 0.2349537037037037,
      "grad_norm": 0.41330647468566895,
      "learning_rate": 1.5300925925925928e-05,
      "loss": 0.0847,
      "step": 2030
    },
    {
      "epoch": 0.23506944444444444,
      "grad_norm": 0.21929636597633362,
      "learning_rate": 1.5298611111111112e-05,
      "loss": 0.0437,
      "step": 2031
    },
    {
      "epoch": 0.2351851851851852,
      "grad_norm": 1.4801437854766846,
      "learning_rate": 1.52962962962963e-05,
      "loss": 0.0633,
      "step": 2032
    },
    {
      "epoch": 0.2353009259259259,
      "grad_norm": 0.24319112300872803,
      "learning_rate": 1.5293981481481484e-05,
      "loss": 0.0445,
      "step": 2033
    },
    {
      "epoch": 0.23541666666666666,
      "grad_norm": 0.2971060276031494,
      "learning_rate": 1.5291666666666668e-05,
      "loss": 0.0578,
      "step": 2034
    },
    {
      "epoch": 0.2355324074074074,
      "grad_norm": 0.7320089936256409,
      "learning_rate": 1.5289351851851852e-05,
      "loss": 0.059,
      "step": 2035
    },
    {
      "epoch": 0.23564814814814813,
      "grad_norm": 0.3191659450531006,
      "learning_rate": 1.5287037037037036e-05,
      "loss": 0.0581,
      "step": 2036
    },
    {
      "epoch": 0.23576388888888888,
      "grad_norm": 0.3435560166835785,
      "learning_rate": 1.5284722222222224e-05,
      "loss": 0.0597,
      "step": 2037
    },
    {
      "epoch": 0.23587962962962963,
      "grad_norm": 0.2224496603012085,
      "learning_rate": 1.528240740740741e-05,
      "loss": 0.0448,
      "step": 2038
    },
    {
      "epoch": 0.23599537037037038,
      "grad_norm": 0.4196406602859497,
      "learning_rate": 1.5280092592592596e-05,
      "loss": 0.0804,
      "step": 2039
    },
    {
      "epoch": 0.2361111111111111,
      "grad_norm": 0.3265398442745209,
      "learning_rate": 1.5277777777777777e-05,
      "loss": 0.0629,
      "step": 2040
    },
    {
      "epoch": 0.23622685185185185,
      "grad_norm": 27.014665603637695,
      "learning_rate": 1.5275462962962964e-05,
      "loss": 1.9417,
      "step": 2041
    },
    {
      "epoch": 0.2363425925925926,
      "grad_norm": 0.2409350574016571,
      "learning_rate": 1.527314814814815e-05,
      "loss": 0.048,
      "step": 2042
    },
    {
      "epoch": 0.23645833333333333,
      "grad_norm": 8.070516586303711,
      "learning_rate": 1.5270833333333336e-05,
      "loss": 2.5112,
      "step": 2043
    },
    {
      "epoch": 0.23657407407407408,
      "grad_norm": 0.25415003299713135,
      "learning_rate": 1.526851851851852e-05,
      "loss": 0.0514,
      "step": 2044
    },
    {
      "epoch": 0.23668981481481483,
      "grad_norm": 0.31846922636032104,
      "learning_rate": 1.5266203703703705e-05,
      "loss": 0.0643,
      "step": 2045
    },
    {
      "epoch": 0.23680555555555555,
      "grad_norm": 0.8599346280097961,
      "learning_rate": 1.526388888888889e-05,
      "loss": 0.0773,
      "step": 2046
    },
    {
      "epoch": 0.2369212962962963,
      "grad_norm": 73.94255828857422,
      "learning_rate": 1.5261574074074073e-05,
      "loss": 1.5656,
      "step": 2047
    },
    {
      "epoch": 0.23703703703703705,
      "grad_norm": 5.422746181488037,
      "learning_rate": 1.525925925925926e-05,
      "loss": 2.7714,
      "step": 2048
    },
    {
      "epoch": 0.23715277777777777,
      "grad_norm": 0.45575255155563354,
      "learning_rate": 1.5256944444444445e-05,
      "loss": 0.0667,
      "step": 2049
    },
    {
      "epoch": 0.23726851851851852,
      "grad_norm": 16.363473892211914,
      "learning_rate": 1.5254629629629631e-05,
      "loss": 0.1085,
      "step": 2050
    },
    {
      "epoch": 0.23738425925925927,
      "grad_norm": 0.24305522441864014,
      "learning_rate": 1.5252314814814817e-05,
      "loss": 0.045,
      "step": 2051
    },
    {
      "epoch": 0.2375,
      "grad_norm": 0.3041864335536957,
      "learning_rate": 1.525e-05,
      "loss": 0.0438,
      "step": 2052
    },
    {
      "epoch": 0.23761574074074074,
      "grad_norm": 0.40020251274108887,
      "learning_rate": 1.5247685185185186e-05,
      "loss": 0.0694,
      "step": 2053
    },
    {
      "epoch": 0.2377314814814815,
      "grad_norm": 7.355844020843506,
      "learning_rate": 1.5245370370370372e-05,
      "loss": 0.1599,
      "step": 2054
    },
    {
      "epoch": 0.2378472222222222,
      "grad_norm": 0.6053271889686584,
      "learning_rate": 1.5243055555555558e-05,
      "loss": 0.0721,
      "step": 2055
    },
    {
      "epoch": 0.23796296296296296,
      "grad_norm": 0.3205587565898895,
      "learning_rate": 1.5240740740740743e-05,
      "loss": 0.0626,
      "step": 2056
    },
    {
      "epoch": 0.2380787037037037,
      "grad_norm": 0.7733715772628784,
      "learning_rate": 1.5238425925925928e-05,
      "loss": 0.0708,
      "step": 2057
    },
    {
      "epoch": 0.23819444444444443,
      "grad_norm": 0.625145435333252,
      "learning_rate": 1.5236111111111112e-05,
      "loss": 0.0901,
      "step": 2058
    },
    {
      "epoch": 0.23831018518518518,
      "grad_norm": 0.3252159059047699,
      "learning_rate": 1.5233796296296298e-05,
      "loss": 0.0648,
      "step": 2059
    },
    {
      "epoch": 0.23842592592592593,
      "grad_norm": 0.3245202600955963,
      "learning_rate": 1.5231481481481482e-05,
      "loss": 0.0616,
      "step": 2060
    },
    {
      "epoch": 0.23854166666666668,
      "grad_norm": 0.2975468933582306,
      "learning_rate": 1.5229166666666668e-05,
      "loss": 0.0552,
      "step": 2061
    },
    {
      "epoch": 0.2386574074074074,
      "grad_norm": 0.23206521570682526,
      "learning_rate": 1.5226851851851854e-05,
      "loss": 0.0457,
      "step": 2062
    },
    {
      "epoch": 0.23877314814814815,
      "grad_norm": 4.94013786315918,
      "learning_rate": 1.522453703703704e-05,
      "loss": 0.1207,
      "step": 2063
    },
    {
      "epoch": 0.2388888888888889,
      "grad_norm": 0.27487900853157043,
      "learning_rate": 1.5222222222222223e-05,
      "loss": 0.0554,
      "step": 2064
    },
    {
      "epoch": 0.23900462962962962,
      "grad_norm": 0.24222612380981445,
      "learning_rate": 1.5219907407407408e-05,
      "loss": 0.0458,
      "step": 2065
    },
    {
      "epoch": 0.23912037037037037,
      "grad_norm": 0.23143108189105988,
      "learning_rate": 1.5217592592592594e-05,
      "loss": 0.044,
      "step": 2066
    },
    {
      "epoch": 0.23923611111111112,
      "grad_norm": 0.671617329120636,
      "learning_rate": 1.5215277777777779e-05,
      "loss": 0.0876,
      "step": 2067
    },
    {
      "epoch": 0.23935185185185184,
      "grad_norm": 0.33253180980682373,
      "learning_rate": 1.5212962962962965e-05,
      "loss": 0.0655,
      "step": 2068
    },
    {
      "epoch": 0.2394675925925926,
      "grad_norm": 0.26617318391799927,
      "learning_rate": 1.521064814814815e-05,
      "loss": 0.0479,
      "step": 2069
    },
    {
      "epoch": 0.23958333333333334,
      "grad_norm": 0.262959361076355,
      "learning_rate": 1.5208333333333333e-05,
      "loss": 0.0522,
      "step": 2070
    },
    {
      "epoch": 0.23969907407407406,
      "grad_norm": 0.3334309160709381,
      "learning_rate": 1.5206018518518519e-05,
      "loss": 0.064,
      "step": 2071
    },
    {
      "epoch": 0.23981481481481481,
      "grad_norm": 0.36598488688468933,
      "learning_rate": 1.5203703703703705e-05,
      "loss": 0.0733,
      "step": 2072
    },
    {
      "epoch": 0.23993055555555556,
      "grad_norm": 0.2706790566444397,
      "learning_rate": 1.5201388888888891e-05,
      "loss": 0.0529,
      "step": 2073
    },
    {
      "epoch": 0.24004629629629629,
      "grad_norm": 0.5360904932022095,
      "learning_rate": 1.5199074074074077e-05,
      "loss": 0.063,
      "step": 2074
    },
    {
      "epoch": 0.24016203703703703,
      "grad_norm": 0.4136965572834015,
      "learning_rate": 1.519675925925926e-05,
      "loss": 0.0653,
      "step": 2075
    },
    {
      "epoch": 0.24027777777777778,
      "grad_norm": 7.3755269050598145,
      "learning_rate": 1.5194444444444445e-05,
      "loss": 0.0978,
      "step": 2076
    },
    {
      "epoch": 0.2403935185185185,
      "grad_norm": 0.4235052764415741,
      "learning_rate": 1.5192129629629631e-05,
      "loss": 0.0619,
      "step": 2077
    },
    {
      "epoch": 0.24050925925925926,
      "grad_norm": 7.55777645111084,
      "learning_rate": 1.5189814814814816e-05,
      "loss": 2.2393,
      "step": 2078
    },
    {
      "epoch": 0.240625,
      "grad_norm": 0.37456032633781433,
      "learning_rate": 1.5187500000000002e-05,
      "loss": 0.074,
      "step": 2079
    },
    {
      "epoch": 0.24074074074074073,
      "grad_norm": 38.869354248046875,
      "learning_rate": 1.5185185185185187e-05,
      "loss": 0.2619,
      "step": 2080
    },
    {
      "epoch": 0.24085648148148148,
      "grad_norm": 6.271079063415527,
      "learning_rate": 1.518287037037037e-05,
      "loss": 2.4591,
      "step": 2081
    },
    {
      "epoch": 0.24097222222222223,
      "grad_norm": 0.22495633363723755,
      "learning_rate": 1.5180555555555556e-05,
      "loss": 0.0436,
      "step": 2082
    },
    {
      "epoch": 0.24108796296296298,
      "grad_norm": 0.35761135816574097,
      "learning_rate": 1.5178240740740742e-05,
      "loss": 0.0661,
      "step": 2083
    },
    {
      "epoch": 0.2412037037037037,
      "grad_norm": 0.6459215879440308,
      "learning_rate": 1.5175925925925928e-05,
      "loss": 0.0713,
      "step": 2084
    },
    {
      "epoch": 0.24131944444444445,
      "grad_norm": 9.780089378356934,
      "learning_rate": 1.5173611111111112e-05,
      "loss": 0.1168,
      "step": 2085
    },
    {
      "epoch": 0.2414351851851852,
      "grad_norm": 0.28380486369132996,
      "learning_rate": 1.5171296296296298e-05,
      "loss": 0.0532,
      "step": 2086
    },
    {
      "epoch": 0.24155092592592592,
      "grad_norm": 0.22760413587093353,
      "learning_rate": 1.5168981481481482e-05,
      "loss": 0.0453,
      "step": 2087
    },
    {
      "epoch": 0.24166666666666667,
      "grad_norm": 58.50502395629883,
      "learning_rate": 1.5166666666666667e-05,
      "loss": 1.2754,
      "step": 2088
    },
    {
      "epoch": 0.24178240740740742,
      "grad_norm": 0.3128592371940613,
      "learning_rate": 1.5164351851851853e-05,
      "loss": 0.0586,
      "step": 2089
    },
    {
      "epoch": 0.24189814814814814,
      "grad_norm": 0.40509429574012756,
      "learning_rate": 1.5162037037037038e-05,
      "loss": 0.0741,
      "step": 2090
    },
    {
      "epoch": 0.2420138888888889,
      "grad_norm": 0.25533390045166016,
      "learning_rate": 1.5159722222222224e-05,
      "loss": 0.0497,
      "step": 2091
    },
    {
      "epoch": 0.24212962962962964,
      "grad_norm": 0.2069760262966156,
      "learning_rate": 1.515740740740741e-05,
      "loss": 0.0415,
      "step": 2092
    },
    {
      "epoch": 0.24224537037037036,
      "grad_norm": 0.25245243310928345,
      "learning_rate": 1.5155092592592593e-05,
      "loss": 0.0487,
      "step": 2093
    },
    {
      "epoch": 0.2423611111111111,
      "grad_norm": 0.2719798982143402,
      "learning_rate": 1.5152777777777779e-05,
      "loss": 0.0527,
      "step": 2094
    },
    {
      "epoch": 0.24247685185185186,
      "grad_norm": 0.4026345908641815,
      "learning_rate": 1.5150462962962965e-05,
      "loss": 0.0654,
      "step": 2095
    },
    {
      "epoch": 0.24259259259259258,
      "grad_norm": 0.3152810335159302,
      "learning_rate": 1.5148148148148149e-05,
      "loss": 0.0604,
      "step": 2096
    },
    {
      "epoch": 0.24270833333333333,
      "grad_norm": 0.29289212822914124,
      "learning_rate": 1.5145833333333335e-05,
      "loss": 0.0567,
      "step": 2097
    },
    {
      "epoch": 0.24282407407407408,
      "grad_norm": 0.4930521547794342,
      "learning_rate": 1.5143518518518521e-05,
      "loss": 0.068,
      "step": 2098
    },
    {
      "epoch": 0.2429398148148148,
      "grad_norm": 0.26634109020233154,
      "learning_rate": 1.5141203703703704e-05,
      "loss": 0.0535,
      "step": 2099
    },
    {
      "epoch": 0.24305555555555555,
      "grad_norm": 0.28331586718559265,
      "learning_rate": 1.513888888888889e-05,
      "loss": 0.0406,
      "step": 2100
    },
    {
      "epoch": 0.2431712962962963,
      "grad_norm": 0.3466569781303406,
      "learning_rate": 1.5136574074074075e-05,
      "loss": 0.0702,
      "step": 2101
    },
    {
      "epoch": 0.24328703703703702,
      "grad_norm": 0.39569368958473206,
      "learning_rate": 1.5134259259259261e-05,
      "loss": 0.0664,
      "step": 2102
    },
    {
      "epoch": 0.24340277777777777,
      "grad_norm": 0.44586098194122314,
      "learning_rate": 1.5131944444444446e-05,
      "loss": 0.05,
      "step": 2103
    },
    {
      "epoch": 0.24351851851851852,
      "grad_norm": 0.2807145118713379,
      "learning_rate": 1.5129629629629632e-05,
      "loss": 0.0539,
      "step": 2104
    },
    {
      "epoch": 0.24363425925925927,
      "grad_norm": 1.3633129596710205,
      "learning_rate": 1.5127314814814816e-05,
      "loss": 0.0852,
      "step": 2105
    },
    {
      "epoch": 0.24375,
      "grad_norm": 0.3586636781692505,
      "learning_rate": 1.5125e-05,
      "loss": 0.0717,
      "step": 2106
    },
    {
      "epoch": 0.24386574074074074,
      "grad_norm": 0.30918627977371216,
      "learning_rate": 1.5122685185185186e-05,
      "loss": 0.0598,
      "step": 2107
    },
    {
      "epoch": 0.2439814814814815,
      "grad_norm": 25.855735778808594,
      "learning_rate": 1.5120370370370372e-05,
      "loss": 0.2373,
      "step": 2108
    },
    {
      "epoch": 0.24409722222222222,
      "grad_norm": 74.06014251708984,
      "learning_rate": 1.5118055555555558e-05,
      "loss": 1.6118,
      "step": 2109
    },
    {
      "epoch": 0.24421296296296297,
      "grad_norm": 0.4453605115413666,
      "learning_rate": 1.5115740740740744e-05,
      "loss": 0.0527,
      "step": 2110
    },
    {
      "epoch": 0.24432870370370371,
      "grad_norm": 0.23298652470111847,
      "learning_rate": 1.5113425925925926e-05,
      "loss": 0.0452,
      "step": 2111
    },
    {
      "epoch": 0.24444444444444444,
      "grad_norm": 0.26244378089904785,
      "learning_rate": 1.5111111111111112e-05,
      "loss": 0.0526,
      "step": 2112
    },
    {
      "epoch": 0.24456018518518519,
      "grad_norm": 0.5116175413131714,
      "learning_rate": 1.5108796296296298e-05,
      "loss": 0.0659,
      "step": 2113
    },
    {
      "epoch": 0.24467592592592594,
      "grad_norm": 0.23134839534759521,
      "learning_rate": 1.5106481481481483e-05,
      "loss": 0.0461,
      "step": 2114
    },
    {
      "epoch": 0.24479166666666666,
      "grad_norm": 0.29751765727996826,
      "learning_rate": 1.5104166666666668e-05,
      "loss": 0.0552,
      "step": 2115
    },
    {
      "epoch": 0.2449074074074074,
      "grad_norm": 0.3813854455947876,
      "learning_rate": 1.5101851851851853e-05,
      "loss": 0.0571,
      "step": 2116
    },
    {
      "epoch": 0.24502314814814816,
      "grad_norm": 0.31151285767555237,
      "learning_rate": 1.5099537037037037e-05,
      "loss": 0.0527,
      "step": 2117
    },
    {
      "epoch": 0.24513888888888888,
      "grad_norm": 1.2942074537277222,
      "learning_rate": 1.5097222222222223e-05,
      "loss": 0.0934,
      "step": 2118
    },
    {
      "epoch": 0.24525462962962963,
      "grad_norm": 28.755605697631836,
      "learning_rate": 1.5094907407407409e-05,
      "loss": 0.2038,
      "step": 2119
    },
    {
      "epoch": 0.24537037037037038,
      "grad_norm": 0.2948162257671356,
      "learning_rate": 1.5092592592592595e-05,
      "loss": 0.0577,
      "step": 2120
    },
    {
      "epoch": 0.2454861111111111,
      "grad_norm": 0.3022625744342804,
      "learning_rate": 1.5090277777777779e-05,
      "loss": 0.0546,
      "step": 2121
    },
    {
      "epoch": 0.24560185185185185,
      "grad_norm": 0.3237025737762451,
      "learning_rate": 1.5087962962962963e-05,
      "loss": 0.065,
      "step": 2122
    },
    {
      "epoch": 0.2457175925925926,
      "grad_norm": 0.20495747029781342,
      "learning_rate": 1.508564814814815e-05,
      "loss": 0.0407,
      "step": 2123
    },
    {
      "epoch": 0.24583333333333332,
      "grad_norm": 0.22761303186416626,
      "learning_rate": 1.5083333333333333e-05,
      "loss": 0.0428,
      "step": 2124
    },
    {
      "epoch": 0.24594907407407407,
      "grad_norm": 17.020917892456055,
      "learning_rate": 1.508101851851852e-05,
      "loss": 0.1334,
      "step": 2125
    },
    {
      "epoch": 0.24606481481481482,
      "grad_norm": 0.22837552428245544,
      "learning_rate": 1.5078703703703705e-05,
      "loss": 0.0439,
      "step": 2126
    },
    {
      "epoch": 0.24618055555555557,
      "grad_norm": 0.22268104553222656,
      "learning_rate": 1.5076388888888891e-05,
      "loss": 0.0441,
      "step": 2127
    },
    {
      "epoch": 0.2462962962962963,
      "grad_norm": 25.82077980041504,
      "learning_rate": 1.5074074074074074e-05,
      "loss": 1.8684,
      "step": 2128
    },
    {
      "epoch": 0.24641203703703704,
      "grad_norm": 0.4778405725955963,
      "learning_rate": 1.507175925925926e-05,
      "loss": 0.0462,
      "step": 2129
    },
    {
      "epoch": 0.2465277777777778,
      "grad_norm": 0.429836243391037,
      "learning_rate": 1.5069444444444446e-05,
      "loss": 0.0617,
      "step": 2130
    },
    {
      "epoch": 0.2466435185185185,
      "grad_norm": 1.8870829343795776,
      "learning_rate": 1.5067129629629632e-05,
      "loss": 0.0613,
      "step": 2131
    },
    {
      "epoch": 0.24675925925925926,
      "grad_norm": 0.3588796854019165,
      "learning_rate": 1.5064814814814816e-05,
      "loss": 0.0646,
      "step": 2132
    },
    {
      "epoch": 0.246875,
      "grad_norm": 6.092339038848877,
      "learning_rate": 1.5062500000000002e-05,
      "loss": 2.6644,
      "step": 2133
    },
    {
      "epoch": 0.24699074074074073,
      "grad_norm": 0.3141966164112091,
      "learning_rate": 1.5060185185185186e-05,
      "loss": 0.0601,
      "step": 2134
    },
    {
      "epoch": 0.24710648148148148,
      "grad_norm": 0.4318082630634308,
      "learning_rate": 1.505787037037037e-05,
      "loss": 0.0487,
      "step": 2135
    },
    {
      "epoch": 0.24722222222222223,
      "grad_norm": 0.3426856994628906,
      "learning_rate": 1.5055555555555556e-05,
      "loss": 0.0689,
      "step": 2136
    },
    {
      "epoch": 0.24733796296296295,
      "grad_norm": 0.22335954010486603,
      "learning_rate": 1.5053240740740742e-05,
      "loss": 0.0437,
      "step": 2137
    },
    {
      "epoch": 0.2474537037037037,
      "grad_norm": 1.3317984342575073,
      "learning_rate": 1.5050925925925928e-05,
      "loss": 0.0795,
      "step": 2138
    },
    {
      "epoch": 0.24756944444444445,
      "grad_norm": 0.24138060212135315,
      "learning_rate": 1.5048611111111112e-05,
      "loss": 0.0472,
      "step": 2139
    },
    {
      "epoch": 0.24768518518518517,
      "grad_norm": 23.534727096557617,
      "learning_rate": 1.5046296296296297e-05,
      "loss": 2.1598,
      "step": 2140
    },
    {
      "epoch": 0.24780092592592592,
      "grad_norm": 2.1778645515441895,
      "learning_rate": 1.5043981481481483e-05,
      "loss": 0.1069,
      "step": 2141
    },
    {
      "epoch": 0.24791666666666667,
      "grad_norm": 0.274584025144577,
      "learning_rate": 1.5041666666666667e-05,
      "loss": 0.0532,
      "step": 2142
    },
    {
      "epoch": 0.2480324074074074,
      "grad_norm": 18.206390380859375,
      "learning_rate": 1.5039351851851853e-05,
      "loss": 0.18,
      "step": 2143
    },
    {
      "epoch": 0.24814814814814815,
      "grad_norm": 0.502700686454773,
      "learning_rate": 1.5037037037037039e-05,
      "loss": 0.0712,
      "step": 2144
    },
    {
      "epoch": 0.2482638888888889,
      "grad_norm": 0.31456077098846436,
      "learning_rate": 1.5034722222222225e-05,
      "loss": 0.0632,
      "step": 2145
    },
    {
      "epoch": 0.24837962962962962,
      "grad_norm": 0.25838786363601685,
      "learning_rate": 1.5032407407407407e-05,
      "loss": 0.0484,
      "step": 2146
    },
    {
      "epoch": 0.24849537037037037,
      "grad_norm": 1.2213987112045288,
      "learning_rate": 1.5030092592592593e-05,
      "loss": 0.0892,
      "step": 2147
    },
    {
      "epoch": 0.24861111111111112,
      "grad_norm": 0.5672581791877747,
      "learning_rate": 1.502777777777778e-05,
      "loss": 0.0619,
      "step": 2148
    },
    {
      "epoch": 0.24872685185185187,
      "grad_norm": 0.26303625106811523,
      "learning_rate": 1.5025462962962965e-05,
      "loss": 0.0521,
      "step": 2149
    },
    {
      "epoch": 0.2488425925925926,
      "grad_norm": 0.3149748146533966,
      "learning_rate": 1.502314814814815e-05,
      "loss": 0.0589,
      "step": 2150
    },
    {
      "epoch": 0.24895833333333334,
      "grad_norm": 5.532430648803711,
      "learning_rate": 1.5020833333333335e-05,
      "loss": 0.1229,
      "step": 2151
    },
    {
      "epoch": 0.2490740740740741,
      "grad_norm": 0.32270506024360657,
      "learning_rate": 1.501851851851852e-05,
      "loss": 0.0607,
      "step": 2152
    },
    {
      "epoch": 0.2491898148148148,
      "grad_norm": 0.7301768064498901,
      "learning_rate": 1.5016203703703704e-05,
      "loss": 0.0707,
      "step": 2153
    },
    {
      "epoch": 0.24930555555555556,
      "grad_norm": 40.49288558959961,
      "learning_rate": 1.501388888888889e-05,
      "loss": 1.7667,
      "step": 2154
    },
    {
      "epoch": 0.2494212962962963,
      "grad_norm": 0.27332600951194763,
      "learning_rate": 1.5011574074074076e-05,
      "loss": 0.0545,
      "step": 2155
    },
    {
      "epoch": 0.24953703703703703,
      "grad_norm": 15.90131664276123,
      "learning_rate": 1.5009259259259262e-05,
      "loss": 2.156,
      "step": 2156
    },
    {
      "epoch": 0.24965277777777778,
      "grad_norm": 0.44064441323280334,
      "learning_rate": 1.5006944444444446e-05,
      "loss": 0.0688,
      "step": 2157
    },
    {
      "epoch": 0.24976851851851853,
      "grad_norm": 0.32401031255722046,
      "learning_rate": 1.500462962962963e-05,
      "loss": 0.0595,
      "step": 2158
    },
    {
      "epoch": 0.24988425925925925,
      "grad_norm": 3.881854772567749,
      "learning_rate": 1.5002314814814816e-05,
      "loss": 0.104,
      "step": 2159
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.33906298875808716,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.0679,
      "step": 2160
    },
    {
      "epoch": 0.2501157407407407,
      "grad_norm": 0.8548834323883057,
      "learning_rate": 1.4997685185185186e-05,
      "loss": 0.0545,
      "step": 2161
    },
    {
      "epoch": 0.2502314814814815,
      "grad_norm": 0.292080819606781,
      "learning_rate": 1.4995370370370372e-05,
      "loss": 0.0526,
      "step": 2162
    },
    {
      "epoch": 0.2503472222222222,
      "grad_norm": 0.2162952572107315,
      "learning_rate": 1.4993055555555557e-05,
      "loss": 0.042,
      "step": 2163
    },
    {
      "epoch": 0.25046296296296294,
      "grad_norm": 71.44645690917969,
      "learning_rate": 1.499074074074074e-05,
      "loss": 0.4932,
      "step": 2164
    },
    {
      "epoch": 0.2505787037037037,
      "grad_norm": 1.1853704452514648,
      "learning_rate": 1.4988425925925927e-05,
      "loss": 0.0661,
      "step": 2165
    },
    {
      "epoch": 0.25069444444444444,
      "grad_norm": 0.29764387011528015,
      "learning_rate": 1.4986111111111113e-05,
      "loss": 0.0592,
      "step": 2166
    },
    {
      "epoch": 0.25081018518518516,
      "grad_norm": 0.23065400123596191,
      "learning_rate": 1.4983796296296299e-05,
      "loss": 0.0459,
      "step": 2167
    },
    {
      "epoch": 0.25092592592592594,
      "grad_norm": 0.2230556458234787,
      "learning_rate": 1.4981481481481483e-05,
      "loss": 0.0435,
      "step": 2168
    },
    {
      "epoch": 0.25104166666666666,
      "grad_norm": 0.38317328691482544,
      "learning_rate": 1.4979166666666667e-05,
      "loss": 0.0625,
      "step": 2169
    },
    {
      "epoch": 0.2511574074074074,
      "grad_norm": 0.6269640922546387,
      "learning_rate": 1.4976851851851853e-05,
      "loss": 0.0698,
      "step": 2170
    },
    {
      "epoch": 0.25127314814814816,
      "grad_norm": 0.6924633979797363,
      "learning_rate": 1.4974537037037037e-05,
      "loss": 0.0733,
      "step": 2171
    },
    {
      "epoch": 0.2513888888888889,
      "grad_norm": 0.23382489383220673,
      "learning_rate": 1.4972222222222223e-05,
      "loss": 0.0459,
      "step": 2172
    },
    {
      "epoch": 0.2515046296296296,
      "grad_norm": 0.3220977783203125,
      "learning_rate": 1.496990740740741e-05,
      "loss": 0.0615,
      "step": 2173
    },
    {
      "epoch": 0.2516203703703704,
      "grad_norm": 0.706356942653656,
      "learning_rate": 1.4967592592592595e-05,
      "loss": 0.0546,
      "step": 2174
    },
    {
      "epoch": 0.2517361111111111,
      "grad_norm": 2.1324005126953125,
      "learning_rate": 1.4965277777777778e-05,
      "loss": 0.097,
      "step": 2175
    },
    {
      "epoch": 0.2518518518518518,
      "grad_norm": 0.4931703209877014,
      "learning_rate": 1.4962962962962964e-05,
      "loss": 0.0762,
      "step": 2176
    },
    {
      "epoch": 0.2519675925925926,
      "grad_norm": 17.122638702392578,
      "learning_rate": 1.496064814814815e-05,
      "loss": 2.1701,
      "step": 2177
    },
    {
      "epoch": 0.2520833333333333,
      "grad_norm": 1.319451928138733,
      "learning_rate": 1.4958333333333336e-05,
      "loss": 0.0712,
      "step": 2178
    },
    {
      "epoch": 0.25219907407407405,
      "grad_norm": 0.20385582745075226,
      "learning_rate": 1.495601851851852e-05,
      "loss": 0.0405,
      "step": 2179
    },
    {
      "epoch": 0.2523148148148148,
      "grad_norm": 0.341641366481781,
      "learning_rate": 1.4953703703703706e-05,
      "loss": 0.0638,
      "step": 2180
    },
    {
      "epoch": 0.25243055555555555,
      "grad_norm": 0.21014179289340973,
      "learning_rate": 1.495138888888889e-05,
      "loss": 0.0399,
      "step": 2181
    },
    {
      "epoch": 0.25254629629629627,
      "grad_norm": 0.33688223361968994,
      "learning_rate": 1.4949074074074074e-05,
      "loss": 0.0676,
      "step": 2182
    },
    {
      "epoch": 0.25266203703703705,
      "grad_norm": 18.193649291992188,
      "learning_rate": 1.494675925925926e-05,
      "loss": 4.9371,
      "step": 2183
    },
    {
      "epoch": 0.25277777777777777,
      "grad_norm": 0.25019922852516174,
      "learning_rate": 1.4944444444444446e-05,
      "loss": 0.0494,
      "step": 2184
    },
    {
      "epoch": 0.25289351851851855,
      "grad_norm": 1.732836365699768,
      "learning_rate": 1.4942129629629632e-05,
      "loss": 0.0806,
      "step": 2185
    },
    {
      "epoch": 0.25300925925925927,
      "grad_norm": 0.2928297519683838,
      "learning_rate": 1.4939814814814816e-05,
      "loss": 0.0557,
      "step": 2186
    },
    {
      "epoch": 0.253125,
      "grad_norm": 24.284881591796875,
      "learning_rate": 1.49375e-05,
      "loss": 0.1636,
      "step": 2187
    },
    {
      "epoch": 0.25324074074074077,
      "grad_norm": 0.23697452247142792,
      "learning_rate": 1.4935185185185186e-05,
      "loss": 0.0444,
      "step": 2188
    },
    {
      "epoch": 0.2533564814814815,
      "grad_norm": 0.429850697517395,
      "learning_rate": 1.493287037037037e-05,
      "loss": 0.0494,
      "step": 2189
    },
    {
      "epoch": 0.2534722222222222,
      "grad_norm": 0.2645987868309021,
      "learning_rate": 1.4930555555555557e-05,
      "loss": 0.0422,
      "step": 2190
    },
    {
      "epoch": 0.253587962962963,
      "grad_norm": 13.087162017822266,
      "learning_rate": 1.4928240740740743e-05,
      "loss": 0.1141,
      "step": 2191
    },
    {
      "epoch": 0.2537037037037037,
      "grad_norm": 0.35682088136672974,
      "learning_rate": 1.4925925925925929e-05,
      "loss": 0.0509,
      "step": 2192
    },
    {
      "epoch": 0.25381944444444443,
      "grad_norm": 0.37281447649002075,
      "learning_rate": 1.4923611111111111e-05,
      "loss": 0.0481,
      "step": 2193
    },
    {
      "epoch": 0.2539351851851852,
      "grad_norm": 0.19711127877235413,
      "learning_rate": 1.4921296296296297e-05,
      "loss": 0.0385,
      "step": 2194
    },
    {
      "epoch": 0.25405092592592593,
      "grad_norm": 2.8157758712768555,
      "learning_rate": 1.4918981481481483e-05,
      "loss": 0.0759,
      "step": 2195
    },
    {
      "epoch": 0.25416666666666665,
      "grad_norm": 0.35649222135543823,
      "learning_rate": 1.4916666666666669e-05,
      "loss": 0.0608,
      "step": 2196
    },
    {
      "epoch": 0.25428240740740743,
      "grad_norm": 0.9844003915786743,
      "learning_rate": 1.4914351851851853e-05,
      "loss": 0.0674,
      "step": 2197
    },
    {
      "epoch": 0.25439814814814815,
      "grad_norm": 0.20680488646030426,
      "learning_rate": 1.491203703703704e-05,
      "loss": 0.0405,
      "step": 2198
    },
    {
      "epoch": 0.2545138888888889,
      "grad_norm": 0.2626054883003235,
      "learning_rate": 1.4909722222222223e-05,
      "loss": 0.0372,
      "step": 2199
    },
    {
      "epoch": 0.25462962962962965,
      "grad_norm": 0.2578093111515045,
      "learning_rate": 1.4907407407407408e-05,
      "loss": 0.0367,
      "step": 2200
    },
    {
      "epoch": 0.25474537037037037,
      "grad_norm": 0.2444377988576889,
      "learning_rate": 1.4905092592592594e-05,
      "loss": 0.0473,
      "step": 2201
    },
    {
      "epoch": 0.2548611111111111,
      "grad_norm": 0.24931709468364716,
      "learning_rate": 1.490277777777778e-05,
      "loss": 0.047,
      "step": 2202
    },
    {
      "epoch": 0.25497685185185187,
      "grad_norm": 0.1993211805820465,
      "learning_rate": 1.4900462962962966e-05,
      "loss": 0.0381,
      "step": 2203
    },
    {
      "epoch": 0.2550925925925926,
      "grad_norm": 0.19199392199516296,
      "learning_rate": 1.489814814814815e-05,
      "loss": 0.0377,
      "step": 2204
    },
    {
      "epoch": 0.2552083333333333,
      "grad_norm": 0.2625351548194885,
      "learning_rate": 1.4895833333333334e-05,
      "loss": 0.0505,
      "step": 2205
    },
    {
      "epoch": 0.2553240740740741,
      "grad_norm": 0.3266349732875824,
      "learning_rate": 1.489351851851852e-05,
      "loss": 0.0653,
      "step": 2206
    },
    {
      "epoch": 0.2554398148148148,
      "grad_norm": 0.3098796010017395,
      "learning_rate": 1.4891203703703704e-05,
      "loss": 0.0603,
      "step": 2207
    },
    {
      "epoch": 0.25555555555555554,
      "grad_norm": 0.20258677005767822,
      "learning_rate": 1.488888888888889e-05,
      "loss": 0.0398,
      "step": 2208
    },
    {
      "epoch": 0.2556712962962963,
      "grad_norm": 0.2859541177749634,
      "learning_rate": 1.4886574074074076e-05,
      "loss": 0.0556,
      "step": 2209
    },
    {
      "epoch": 0.25578703703703703,
      "grad_norm": 0.24970151484012604,
      "learning_rate": 1.4884259259259259e-05,
      "loss": 0.0468,
      "step": 2210
    },
    {
      "epoch": 0.25590277777777776,
      "grad_norm": 11.361026763916016,
      "learning_rate": 1.4881944444444445e-05,
      "loss": 2.2044,
      "step": 2211
    },
    {
      "epoch": 0.25601851851851853,
      "grad_norm": 0.32665273547172546,
      "learning_rate": 1.487962962962963e-05,
      "loss": 0.0615,
      "step": 2212
    },
    {
      "epoch": 0.25613425925925926,
      "grad_norm": 0.6169847846031189,
      "learning_rate": 1.4877314814814816e-05,
      "loss": 0.079,
      "step": 2213
    },
    {
      "epoch": 0.25625,
      "grad_norm": 0.2849927544593811,
      "learning_rate": 1.4875000000000002e-05,
      "loss": 0.057,
      "step": 2214
    },
    {
      "epoch": 0.25636574074074076,
      "grad_norm": 0.18584738671779633,
      "learning_rate": 1.4872685185185187e-05,
      "loss": 0.0367,
      "step": 2215
    },
    {
      "epoch": 0.2564814814814815,
      "grad_norm": 1.4249423742294312,
      "learning_rate": 1.4870370370370371e-05,
      "loss": 0.0837,
      "step": 2216
    },
    {
      "epoch": 0.2565972222222222,
      "grad_norm": 0.2434905469417572,
      "learning_rate": 1.4868055555555557e-05,
      "loss": 0.0428,
      "step": 2217
    },
    {
      "epoch": 0.256712962962963,
      "grad_norm": 2.7402143478393555,
      "learning_rate": 1.4865740740740741e-05,
      "loss": 0.0563,
      "step": 2218
    },
    {
      "epoch": 0.2568287037037037,
      "grad_norm": 114.30176544189453,
      "learning_rate": 1.4863425925925927e-05,
      "loss": 1.9395,
      "step": 2219
    },
    {
      "epoch": 0.2569444444444444,
      "grad_norm": 35.49263381958008,
      "learning_rate": 1.4861111111111113e-05,
      "loss": 1.8831,
      "step": 2220
    },
    {
      "epoch": 0.2570601851851852,
      "grad_norm": 51.959373474121094,
      "learning_rate": 1.4858796296296299e-05,
      "loss": 1.481,
      "step": 2221
    },
    {
      "epoch": 0.2571759259259259,
      "grad_norm": 0.38038238883018494,
      "learning_rate": 1.4856481481481482e-05,
      "loss": 0.0603,
      "step": 2222
    },
    {
      "epoch": 0.25729166666666664,
      "grad_norm": 33.527183532714844,
      "learning_rate": 1.4854166666666667e-05,
      "loss": 1.9977,
      "step": 2223
    },
    {
      "epoch": 0.2574074074074074,
      "grad_norm": 0.20749010145664215,
      "learning_rate": 1.4851851851851853e-05,
      "loss": 0.0412,
      "step": 2224
    },
    {
      "epoch": 0.25752314814814814,
      "grad_norm": 0.21180979907512665,
      "learning_rate": 1.4849537037037038e-05,
      "loss": 0.0411,
      "step": 2225
    },
    {
      "epoch": 0.25763888888888886,
      "grad_norm": 0.5363352298736572,
      "learning_rate": 1.4847222222222224e-05,
      "loss": 0.0754,
      "step": 2226
    },
    {
      "epoch": 0.25775462962962964,
      "grad_norm": 0.18738317489624023,
      "learning_rate": 1.484490740740741e-05,
      "loss": 0.0369,
      "step": 2227
    },
    {
      "epoch": 0.25787037037037036,
      "grad_norm": 0.3351745903491974,
      "learning_rate": 1.4842592592592592e-05,
      "loss": 0.048,
      "step": 2228
    },
    {
      "epoch": 0.25798611111111114,
      "grad_norm": 0.38372862339019775,
      "learning_rate": 1.4840277777777778e-05,
      "loss": 0.0753,
      "step": 2229
    },
    {
      "epoch": 0.25810185185185186,
      "grad_norm": 0.33172664046287537,
      "learning_rate": 1.4837962962962964e-05,
      "loss": 0.0588,
      "step": 2230
    },
    {
      "epoch": 0.2582175925925926,
      "grad_norm": 0.20306257903575897,
      "learning_rate": 1.483564814814815e-05,
      "loss": 0.0392,
      "step": 2231
    },
    {
      "epoch": 0.25833333333333336,
      "grad_norm": 6.635000705718994,
      "learning_rate": 1.4833333333333336e-05,
      "loss": 2.7987,
      "step": 2232
    },
    {
      "epoch": 0.2584490740740741,
      "grad_norm": 32.561466217041016,
      "learning_rate": 1.483101851851852e-05,
      "loss": 1.8639,
      "step": 2233
    },
    {
      "epoch": 0.2585648148148148,
      "grad_norm": 0.2566065192222595,
      "learning_rate": 1.4828703703703704e-05,
      "loss": 0.0484,
      "step": 2234
    },
    {
      "epoch": 0.2586805555555556,
      "grad_norm": 0.20133167505264282,
      "learning_rate": 1.482638888888889e-05,
      "loss": 0.038,
      "step": 2235
    },
    {
      "epoch": 0.2587962962962963,
      "grad_norm": 0.27608489990234375,
      "learning_rate": 1.4824074074074075e-05,
      "loss": 0.0493,
      "step": 2236
    },
    {
      "epoch": 0.258912037037037,
      "grad_norm": 0.20925799012184143,
      "learning_rate": 1.482175925925926e-05,
      "loss": 0.0406,
      "step": 2237
    },
    {
      "epoch": 0.2590277777777778,
      "grad_norm": 0.4885549545288086,
      "learning_rate": 1.4819444444444446e-05,
      "loss": 0.0612,
      "step": 2238
    },
    {
      "epoch": 0.2591435185185185,
      "grad_norm": 0.3019247353076935,
      "learning_rate": 1.4817129629629632e-05,
      "loss": 0.0498,
      "step": 2239
    },
    {
      "epoch": 0.25925925925925924,
      "grad_norm": 0.3291757106781006,
      "learning_rate": 1.4814814814814815e-05,
      "loss": 0.0511,
      "step": 2240
    },
    {
      "epoch": 0.259375,
      "grad_norm": 16.397075653076172,
      "learning_rate": 1.4812500000000001e-05,
      "loss": 2.3276,
      "step": 2241
    },
    {
      "epoch": 0.25949074074074074,
      "grad_norm": 0.32054388523101807,
      "learning_rate": 1.4810185185185187e-05,
      "loss": 0.0623,
      "step": 2242
    },
    {
      "epoch": 0.25960648148148147,
      "grad_norm": 0.2757337689399719,
      "learning_rate": 1.4807870370370371e-05,
      "loss": 0.0531,
      "step": 2243
    },
    {
      "epoch": 0.25972222222222224,
      "grad_norm": 0.22752690315246582,
      "learning_rate": 1.4805555555555557e-05,
      "loss": 0.0401,
      "step": 2244
    },
    {
      "epoch": 0.25983796296296297,
      "grad_norm": 17.34877586364746,
      "learning_rate": 1.4803240740740743e-05,
      "loss": 2.5003,
      "step": 2245
    },
    {
      "epoch": 0.2599537037037037,
      "grad_norm": 0.29125070571899414,
      "learning_rate": 1.4800925925925926e-05,
      "loss": 0.0531,
      "step": 2246
    },
    {
      "epoch": 0.26006944444444446,
      "grad_norm": 0.29880350828170776,
      "learning_rate": 1.4798611111111111e-05,
      "loss": 0.0428,
      "step": 2247
    },
    {
      "epoch": 0.2601851851851852,
      "grad_norm": 84.29385375976562,
      "learning_rate": 1.4796296296296297e-05,
      "loss": 0.6778,
      "step": 2248
    },
    {
      "epoch": 0.2603009259259259,
      "grad_norm": 0.3140783905982971,
      "learning_rate": 1.4793981481481483e-05,
      "loss": 0.0611,
      "step": 2249
    },
    {
      "epoch": 0.2604166666666667,
      "grad_norm": 0.32438886165618896,
      "learning_rate": 1.479166666666667e-05,
      "loss": 0.064,
      "step": 2250
    },
    {
      "epoch": 0.2605324074074074,
      "grad_norm": 0.2508111000061035,
      "learning_rate": 1.4789351851851852e-05,
      "loss": 0.045,
      "step": 2251
    },
    {
      "epoch": 0.26064814814814813,
      "grad_norm": 0.2883581519126892,
      "learning_rate": 1.4787037037037038e-05,
      "loss": 0.0555,
      "step": 2252
    },
    {
      "epoch": 0.2607638888888889,
      "grad_norm": 20.601259231567383,
      "learning_rate": 1.4784722222222224e-05,
      "loss": 0.1799,
      "step": 2253
    },
    {
      "epoch": 0.26087962962962963,
      "grad_norm": 1.6974979639053345,
      "learning_rate": 1.4782407407407408e-05,
      "loss": 0.0534,
      "step": 2254
    },
    {
      "epoch": 0.26099537037037035,
      "grad_norm": 38.703155517578125,
      "learning_rate": 1.4780092592592594e-05,
      "loss": 0.3213,
      "step": 2255
    },
    {
      "epoch": 0.2611111111111111,
      "grad_norm": 0.2891758978366852,
      "learning_rate": 1.477777777777778e-05,
      "loss": 0.0568,
      "step": 2256
    },
    {
      "epoch": 0.26122685185185185,
      "grad_norm": 100.55636596679688,
      "learning_rate": 1.4775462962962962e-05,
      "loss": 0.6273,
      "step": 2257
    },
    {
      "epoch": 0.26134259259259257,
      "grad_norm": 7.471905708312988,
      "learning_rate": 1.4773148148148148e-05,
      "loss": 0.118,
      "step": 2258
    },
    {
      "epoch": 0.26145833333333335,
      "grad_norm": 6.423748970031738,
      "learning_rate": 1.4770833333333334e-05,
      "loss": 0.1022,
      "step": 2259
    },
    {
      "epoch": 0.26157407407407407,
      "grad_norm": 0.32621413469314575,
      "learning_rate": 1.476851851851852e-05,
      "loss": 0.058,
      "step": 2260
    },
    {
      "epoch": 0.2616898148148148,
      "grad_norm": 0.26286569237709045,
      "learning_rate": 1.4766203703703705e-05,
      "loss": 0.0524,
      "step": 2261
    },
    {
      "epoch": 0.26180555555555557,
      "grad_norm": 38.339778900146484,
      "learning_rate": 1.476388888888889e-05,
      "loss": 1.7942,
      "step": 2262
    },
    {
      "epoch": 0.2619212962962963,
      "grad_norm": 15.838332176208496,
      "learning_rate": 1.4761574074074075e-05,
      "loss": 0.1271,
      "step": 2263
    },
    {
      "epoch": 0.262037037037037,
      "grad_norm": 0.7583422660827637,
      "learning_rate": 1.4759259259259259e-05,
      "loss": 0.0631,
      "step": 2264
    },
    {
      "epoch": 0.2621527777777778,
      "grad_norm": 0.25564366579055786,
      "learning_rate": 1.4756944444444445e-05,
      "loss": 0.0478,
      "step": 2265
    },
    {
      "epoch": 0.2622685185185185,
      "grad_norm": 6.698325157165527,
      "learning_rate": 1.4754629629629631e-05,
      "loss": 0.1337,
      "step": 2266
    },
    {
      "epoch": 0.26238425925925923,
      "grad_norm": 0.19761206209659576,
      "learning_rate": 1.4752314814814817e-05,
      "loss": 0.0378,
      "step": 2267
    },
    {
      "epoch": 0.2625,
      "grad_norm": 0.19793018698692322,
      "learning_rate": 1.4750000000000003e-05,
      "loss": 0.0386,
      "step": 2268
    },
    {
      "epoch": 0.26261574074074073,
      "grad_norm": 12.590265274047852,
      "learning_rate": 1.4747685185185185e-05,
      "loss": 2.4587,
      "step": 2269
    },
    {
      "epoch": 0.26273148148148145,
      "grad_norm": 0.29226478934288025,
      "learning_rate": 1.4745370370370371e-05,
      "loss": 0.0492,
      "step": 2270
    },
    {
      "epoch": 0.26284722222222223,
      "grad_norm": 0.37527820467948914,
      "learning_rate": 1.4743055555555557e-05,
      "loss": 0.0544,
      "step": 2271
    },
    {
      "epoch": 0.26296296296296295,
      "grad_norm": 0.2878415286540985,
      "learning_rate": 1.4740740740740741e-05,
      "loss": 0.0539,
      "step": 2272
    },
    {
      "epoch": 0.26307870370370373,
      "grad_norm": 0.298979789018631,
      "learning_rate": 1.4738425925925927e-05,
      "loss": 0.0512,
      "step": 2273
    },
    {
      "epoch": 0.26319444444444445,
      "grad_norm": 0.3159642517566681,
      "learning_rate": 1.4736111111111113e-05,
      "loss": 0.0494,
      "step": 2274
    },
    {
      "epoch": 0.2633101851851852,
      "grad_norm": 0.930366039276123,
      "learning_rate": 1.4733796296296296e-05,
      "loss": 0.0684,
      "step": 2275
    },
    {
      "epoch": 0.26342592592592595,
      "grad_norm": 0.24216538667678833,
      "learning_rate": 1.4731481481481482e-05,
      "loss": 0.0469,
      "step": 2276
    },
    {
      "epoch": 0.2635416666666667,
      "grad_norm": 0.33353516459465027,
      "learning_rate": 1.4729166666666668e-05,
      "loss": 0.0502,
      "step": 2277
    },
    {
      "epoch": 0.2636574074074074,
      "grad_norm": 0.31088194251060486,
      "learning_rate": 1.4726851851851854e-05,
      "loss": 0.0604,
      "step": 2278
    },
    {
      "epoch": 0.2637731481481482,
      "grad_norm": 9.925065994262695,
      "learning_rate": 1.472453703703704e-05,
      "loss": 0.1322,
      "step": 2279
    },
    {
      "epoch": 0.2638888888888889,
      "grad_norm": 14.249490737915039,
      "learning_rate": 1.4722222222222224e-05,
      "loss": 4.6807,
      "step": 2280
    },
    {
      "epoch": 0.2640046296296296,
      "grad_norm": 0.22511893510818481,
      "learning_rate": 1.4719907407407408e-05,
      "loss": 0.0435,
      "step": 2281
    },
    {
      "epoch": 0.2641203703703704,
      "grad_norm": 0.3236340582370758,
      "learning_rate": 1.4717592592592594e-05,
      "loss": 0.0512,
      "step": 2282
    },
    {
      "epoch": 0.2642361111111111,
      "grad_norm": 0.2343631088733673,
      "learning_rate": 1.4715277777777778e-05,
      "loss": 0.0426,
      "step": 2283
    },
    {
      "epoch": 0.26435185185185184,
      "grad_norm": 0.32970544695854187,
      "learning_rate": 1.4712962962962964e-05,
      "loss": 0.0618,
      "step": 2284
    },
    {
      "epoch": 0.2644675925925926,
      "grad_norm": 0.32508036494255066,
      "learning_rate": 1.471064814814815e-05,
      "loss": 0.0468,
      "step": 2285
    },
    {
      "epoch": 0.26458333333333334,
      "grad_norm": 0.3177044987678528,
      "learning_rate": 1.4708333333333336e-05,
      "loss": 0.063,
      "step": 2286
    },
    {
      "epoch": 0.26469907407407406,
      "grad_norm": 0.9631253480911255,
      "learning_rate": 1.4706018518518519e-05,
      "loss": 0.0633,
      "step": 2287
    },
    {
      "epoch": 0.26481481481481484,
      "grad_norm": 0.23226706683635712,
      "learning_rate": 1.4703703703703705e-05,
      "loss": 0.0405,
      "step": 2288
    },
    {
      "epoch": 0.26493055555555556,
      "grad_norm": 0.30448266863822937,
      "learning_rate": 1.470138888888889e-05,
      "loss": 0.0594,
      "step": 2289
    },
    {
      "epoch": 0.2650462962962963,
      "grad_norm": 89.05289459228516,
      "learning_rate": 1.4699074074074075e-05,
      "loss": 0.3047,
      "step": 2290
    },
    {
      "epoch": 0.26516203703703706,
      "grad_norm": 0.44513848423957825,
      "learning_rate": 1.4696759259259261e-05,
      "loss": 0.0683,
      "step": 2291
    },
    {
      "epoch": 0.2652777777777778,
      "grad_norm": 0.2367095798254013,
      "learning_rate": 1.4694444444444447e-05,
      "loss": 0.046,
      "step": 2292
    },
    {
      "epoch": 0.2653935185185185,
      "grad_norm": 22.58804702758789,
      "learning_rate": 1.469212962962963e-05,
      "loss": 2.1831,
      "step": 2293
    },
    {
      "epoch": 0.2655092592592593,
      "grad_norm": 0.9645591378211975,
      "learning_rate": 1.4689814814814815e-05,
      "loss": 0.0691,
      "step": 2294
    },
    {
      "epoch": 0.265625,
      "grad_norm": 0.2523767650127411,
      "learning_rate": 1.4687500000000001e-05,
      "loss": 0.0353,
      "step": 2295
    },
    {
      "epoch": 0.2657407407407407,
      "grad_norm": 125.46025085449219,
      "learning_rate": 1.4685185185185187e-05,
      "loss": 1.5426,
      "step": 2296
    },
    {
      "epoch": 0.2658564814814815,
      "grad_norm": 7.020386219024658,
      "learning_rate": 1.4682870370370373e-05,
      "loss": 2.334,
      "step": 2297
    },
    {
      "epoch": 0.2659722222222222,
      "grad_norm": 0.245977982878685,
      "learning_rate": 1.4680555555555556e-05,
      "loss": 0.0493,
      "step": 2298
    },
    {
      "epoch": 0.26608796296296294,
      "grad_norm": 0.2481403350830078,
      "learning_rate": 1.4678240740740742e-05,
      "loss": 0.0481,
      "step": 2299
    },
    {
      "epoch": 0.2662037037037037,
      "grad_norm": 30.403282165527344,
      "learning_rate": 1.4675925925925928e-05,
      "loss": 2.0251,
      "step": 2300
    },
    {
      "epoch": 0.26631944444444444,
      "grad_norm": 0.2412661761045456,
      "learning_rate": 1.4673611111111112e-05,
      "loss": 0.0454,
      "step": 2301
    },
    {
      "epoch": 0.26643518518518516,
      "grad_norm": 0.6114447712898254,
      "learning_rate": 1.4671296296296298e-05,
      "loss": 0.0594,
      "step": 2302
    },
    {
      "epoch": 0.26655092592592594,
      "grad_norm": 0.6699556112289429,
      "learning_rate": 1.4668981481481484e-05,
      "loss": 0.0416,
      "step": 2303
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 27.54779052734375,
      "learning_rate": 1.4666666666666666e-05,
      "loss": 2.101,
      "step": 2304
    },
    {
      "epoch": 0.2667824074074074,
      "grad_norm": 0.38442298769950867,
      "learning_rate": 1.4664351851851852e-05,
      "loss": 0.0491,
      "step": 2305
    },
    {
      "epoch": 0.26689814814814816,
      "grad_norm": 0.22752104699611664,
      "learning_rate": 1.4662037037037038e-05,
      "loss": 0.0432,
      "step": 2306
    },
    {
      "epoch": 0.2670138888888889,
      "grad_norm": 0.2789291441440582,
      "learning_rate": 1.4659722222222224e-05,
      "loss": 0.055,
      "step": 2307
    },
    {
      "epoch": 0.2671296296296296,
      "grad_norm": 0.36087921261787415,
      "learning_rate": 1.4657407407407408e-05,
      "loss": 0.0438,
      "step": 2308
    },
    {
      "epoch": 0.2672453703703704,
      "grad_norm": 21.145023345947266,
      "learning_rate": 1.4655092592592594e-05,
      "loss": 2.3334,
      "step": 2309
    },
    {
      "epoch": 0.2673611111111111,
      "grad_norm": 0.5121107697486877,
      "learning_rate": 1.4652777777777779e-05,
      "loss": 0.0721,
      "step": 2310
    },
    {
      "epoch": 0.2674768518518518,
      "grad_norm": 0.2443326860666275,
      "learning_rate": 1.4650462962962963e-05,
      "loss": 0.049,
      "step": 2311
    },
    {
      "epoch": 0.2675925925925926,
      "grad_norm": 0.29986634850502014,
      "learning_rate": 1.4648148148148149e-05,
      "loss": 0.0418,
      "step": 2312
    },
    {
      "epoch": 0.2677083333333333,
      "grad_norm": 1.3164526224136353,
      "learning_rate": 1.4645833333333335e-05,
      "loss": 0.0866,
      "step": 2313
    },
    {
      "epoch": 0.26782407407407405,
      "grad_norm": 0.22554799914360046,
      "learning_rate": 1.464351851851852e-05,
      "loss": 0.0429,
      "step": 2314
    },
    {
      "epoch": 0.2679398148148148,
      "grad_norm": 113.82030487060547,
      "learning_rate": 1.4641203703703707e-05,
      "loss": 1.6301,
      "step": 2315
    },
    {
      "epoch": 0.26805555555555555,
      "grad_norm": 0.22513850033283234,
      "learning_rate": 1.4638888888888889e-05,
      "loss": 0.0438,
      "step": 2316
    },
    {
      "epoch": 0.26817129629629627,
      "grad_norm": 0.4319632351398468,
      "learning_rate": 1.4636574074074075e-05,
      "loss": 0.044,
      "step": 2317
    },
    {
      "epoch": 0.26828703703703705,
      "grad_norm": 0.21002735197544098,
      "learning_rate": 1.4634259259259261e-05,
      "loss": 0.0404,
      "step": 2318
    },
    {
      "epoch": 0.26840277777777777,
      "grad_norm": 0.2632540464401245,
      "learning_rate": 1.4631944444444445e-05,
      "loss": 0.0374,
      "step": 2319
    },
    {
      "epoch": 0.26851851851851855,
      "grad_norm": 30.990068435668945,
      "learning_rate": 1.4629629629629631e-05,
      "loss": 0.2404,
      "step": 2320
    },
    {
      "epoch": 0.26863425925925927,
      "grad_norm": 0.234559565782547,
      "learning_rate": 1.4627314814814817e-05,
      "loss": 0.0456,
      "step": 2321
    },
    {
      "epoch": 0.26875,
      "grad_norm": 39.28299331665039,
      "learning_rate": 1.4625e-05,
      "loss": 1.6216,
      "step": 2322
    },
    {
      "epoch": 0.26886574074074077,
      "grad_norm": 0.32245180010795593,
      "learning_rate": 1.4622685185185186e-05,
      "loss": 0.0508,
      "step": 2323
    },
    {
      "epoch": 0.2689814814814815,
      "grad_norm": 0.27518558502197266,
      "learning_rate": 1.4620370370370372e-05,
      "loss": 0.0528,
      "step": 2324
    },
    {
      "epoch": 0.2690972222222222,
      "grad_norm": 0.20826467871665955,
      "learning_rate": 1.4618055555555558e-05,
      "loss": 0.0384,
      "step": 2325
    },
    {
      "epoch": 0.269212962962963,
      "grad_norm": 0.22747237980365753,
      "learning_rate": 1.4615740740740742e-05,
      "loss": 0.0447,
      "step": 2326
    },
    {
      "epoch": 0.2693287037037037,
      "grad_norm": 0.37928298115730286,
      "learning_rate": 1.4613425925925928e-05,
      "loss": 0.0555,
      "step": 2327
    },
    {
      "epoch": 0.26944444444444443,
      "grad_norm": 0.30382734537124634,
      "learning_rate": 1.4611111111111112e-05,
      "loss": 0.0467,
      "step": 2328
    },
    {
      "epoch": 0.2695601851851852,
      "grad_norm": 0.5398982763290405,
      "learning_rate": 1.4608796296296296e-05,
      "loss": 0.0632,
      "step": 2329
    },
    {
      "epoch": 0.26967592592592593,
      "grad_norm": 0.6594314575195312,
      "learning_rate": 1.4606481481481482e-05,
      "loss": 0.0555,
      "step": 2330
    },
    {
      "epoch": 0.26979166666666665,
      "grad_norm": 29.93090057373047,
      "learning_rate": 1.4604166666666668e-05,
      "loss": 0.28,
      "step": 2331
    },
    {
      "epoch": 0.26990740740740743,
      "grad_norm": 0.2308884561061859,
      "learning_rate": 1.4601851851851854e-05,
      "loss": 0.0449,
      "step": 2332
    },
    {
      "epoch": 0.27002314814814815,
      "grad_norm": 0.27851954102516174,
      "learning_rate": 1.459953703703704e-05,
      "loss": 0.0534,
      "step": 2333
    },
    {
      "epoch": 0.2701388888888889,
      "grad_norm": 0.6442376375198364,
      "learning_rate": 1.4597222222222223e-05,
      "loss": 0.0597,
      "step": 2334
    },
    {
      "epoch": 0.27025462962962965,
      "grad_norm": 0.347487211227417,
      "learning_rate": 1.4594907407407409e-05,
      "loss": 0.06,
      "step": 2335
    },
    {
      "epoch": 0.27037037037037037,
      "grad_norm": 0.2460147738456726,
      "learning_rate": 1.4592592592592594e-05,
      "loss": 0.0489,
      "step": 2336
    },
    {
      "epoch": 0.2704861111111111,
      "grad_norm": 0.25870847702026367,
      "learning_rate": 1.4590277777777779e-05,
      "loss": 0.0505,
      "step": 2337
    },
    {
      "epoch": 0.27060185185185187,
      "grad_norm": 0.19388924539089203,
      "learning_rate": 1.4587962962962965e-05,
      "loss": 0.038,
      "step": 2338
    },
    {
      "epoch": 0.2707175925925926,
      "grad_norm": 0.27417680621147156,
      "learning_rate": 1.458564814814815e-05,
      "loss": 0.0536,
      "step": 2339
    },
    {
      "epoch": 0.2708333333333333,
      "grad_norm": 0.22893279790878296,
      "learning_rate": 1.4583333333333333e-05,
      "loss": 0.0446,
      "step": 2340
    },
    {
      "epoch": 0.2709490740740741,
      "grad_norm": 0.35038232803344727,
      "learning_rate": 1.4581018518518519e-05,
      "loss": 0.0671,
      "step": 2341
    },
    {
      "epoch": 0.2710648148148148,
      "grad_norm": 0.35594454407691956,
      "learning_rate": 1.4578703703703705e-05,
      "loss": 0.0393,
      "step": 2342
    },
    {
      "epoch": 0.27118055555555554,
      "grad_norm": 1.3112651109695435,
      "learning_rate": 1.4576388888888891e-05,
      "loss": 0.0629,
      "step": 2343
    },
    {
      "epoch": 0.2712962962962963,
      "grad_norm": 0.2871062755584717,
      "learning_rate": 1.4574074074074075e-05,
      "loss": 0.0543,
      "step": 2344
    },
    {
      "epoch": 0.27141203703703703,
      "grad_norm": 82.25776672363281,
      "learning_rate": 1.457175925925926e-05,
      "loss": 1.282,
      "step": 2345
    },
    {
      "epoch": 0.27152777777777776,
      "grad_norm": 0.29800429940223694,
      "learning_rate": 1.4569444444444445e-05,
      "loss": 0.0525,
      "step": 2346
    },
    {
      "epoch": 0.27164351851851853,
      "grad_norm": 1.0598782300949097,
      "learning_rate": 1.456712962962963e-05,
      "loss": 0.0516,
      "step": 2347
    },
    {
      "epoch": 0.27175925925925926,
      "grad_norm": 85.11995697021484,
      "learning_rate": 1.4564814814814816e-05,
      "loss": 1.6866,
      "step": 2348
    },
    {
      "epoch": 0.271875,
      "grad_norm": 0.602077066898346,
      "learning_rate": 1.4562500000000002e-05,
      "loss": 0.0508,
      "step": 2349
    },
    {
      "epoch": 0.27199074074074076,
      "grad_norm": 0.2179642617702484,
      "learning_rate": 1.4560185185185188e-05,
      "loss": 0.0429,
      "step": 2350
    },
    {
      "epoch": 0.2721064814814815,
      "grad_norm": 0.21887660026550293,
      "learning_rate": 1.455787037037037e-05,
      "loss": 0.0432,
      "step": 2351
    },
    {
      "epoch": 0.2722222222222222,
      "grad_norm": 3.8221426010131836,
      "learning_rate": 1.4555555555555556e-05,
      "loss": 0.1117,
      "step": 2352
    },
    {
      "epoch": 0.272337962962963,
      "grad_norm": 0.24249309301376343,
      "learning_rate": 1.4553240740740742e-05,
      "loss": 0.0458,
      "step": 2353
    },
    {
      "epoch": 0.2724537037037037,
      "grad_norm": 0.20589490234851837,
      "learning_rate": 1.4550925925925928e-05,
      "loss": 0.0404,
      "step": 2354
    },
    {
      "epoch": 0.2725694444444444,
      "grad_norm": 0.2916598916053772,
      "learning_rate": 1.4548611111111112e-05,
      "loss": 0.0412,
      "step": 2355
    },
    {
      "epoch": 0.2726851851851852,
      "grad_norm": 0.31436771154403687,
      "learning_rate": 1.4546296296296298e-05,
      "loss": 0.0468,
      "step": 2356
    },
    {
      "epoch": 0.2728009259259259,
      "grad_norm": 0.3146117925643921,
      "learning_rate": 1.4543981481481482e-05,
      "loss": 0.047,
      "step": 2357
    },
    {
      "epoch": 0.27291666666666664,
      "grad_norm": 0.245153546333313,
      "learning_rate": 1.4541666666666667e-05,
      "loss": 0.0485,
      "step": 2358
    },
    {
      "epoch": 0.2730324074074074,
      "grad_norm": 0.787811815738678,
      "learning_rate": 1.4539351851851853e-05,
      "loss": 0.074,
      "step": 2359
    },
    {
      "epoch": 0.27314814814814814,
      "grad_norm": 17.0817928314209,
      "learning_rate": 1.4537037037037039e-05,
      "loss": 0.1784,
      "step": 2360
    },
    {
      "epoch": 0.27326388888888886,
      "grad_norm": 2.298743486404419,
      "learning_rate": 1.4534722222222224e-05,
      "loss": 0.0587,
      "step": 2361
    },
    {
      "epoch": 0.27337962962962964,
      "grad_norm": 0.2263183742761612,
      "learning_rate": 1.4532407407407409e-05,
      "loss": 0.0439,
      "step": 2362
    },
    {
      "epoch": 0.27349537037037036,
      "grad_norm": 0.18681205809116364,
      "learning_rate": 1.4530092592592593e-05,
      "loss": 0.0356,
      "step": 2363
    },
    {
      "epoch": 0.27361111111111114,
      "grad_norm": 0.2841378450393677,
      "learning_rate": 1.4527777777777779e-05,
      "loss": 0.0562,
      "step": 2364
    },
    {
      "epoch": 0.27372685185185186,
      "grad_norm": 0.2168474793434143,
      "learning_rate": 1.4525462962962963e-05,
      "loss": 0.0398,
      "step": 2365
    },
    {
      "epoch": 0.2738425925925926,
      "grad_norm": 84.02288818359375,
      "learning_rate": 1.4523148148148149e-05,
      "loss": 0.7782,
      "step": 2366
    },
    {
      "epoch": 0.27395833333333336,
      "grad_norm": 0.21793600916862488,
      "learning_rate": 1.4520833333333335e-05,
      "loss": 0.0429,
      "step": 2367
    },
    {
      "epoch": 0.2740740740740741,
      "grad_norm": 0.29162198305130005,
      "learning_rate": 1.4518518518518521e-05,
      "loss": 0.0458,
      "step": 2368
    },
    {
      "epoch": 0.2741898148148148,
      "grad_norm": 4.886813163757324,
      "learning_rate": 1.4516203703703704e-05,
      "loss": 0.0853,
      "step": 2369
    },
    {
      "epoch": 0.2743055555555556,
      "grad_norm": 0.42726263403892517,
      "learning_rate": 1.451388888888889e-05,
      "loss": 0.0603,
      "step": 2370
    },
    {
      "epoch": 0.2744212962962963,
      "grad_norm": 0.28677231073379517,
      "learning_rate": 1.4511574074074075e-05,
      "loss": 0.056,
      "step": 2371
    },
    {
      "epoch": 0.274537037037037,
      "grad_norm": 0.47070634365081787,
      "learning_rate": 1.4509259259259261e-05,
      "loss": 0.041,
      "step": 2372
    },
    {
      "epoch": 0.2746527777777778,
      "grad_norm": 0.28338882327079773,
      "learning_rate": 1.4506944444444446e-05,
      "loss": 0.0524,
      "step": 2373
    },
    {
      "epoch": 0.2747685185185185,
      "grad_norm": 1.3003710508346558,
      "learning_rate": 1.4504629629629632e-05,
      "loss": 0.0717,
      "step": 2374
    },
    {
      "epoch": 0.27488425925925924,
      "grad_norm": 0.2475651204586029,
      "learning_rate": 1.4502314814814816e-05,
      "loss": 0.0471,
      "step": 2375
    },
    {
      "epoch": 0.275,
      "grad_norm": 0.3807452321052551,
      "learning_rate": 1.45e-05,
      "loss": 0.063,
      "step": 2376
    },
    {
      "epoch": 0.27511574074074074,
      "grad_norm": 0.17289027571678162,
      "learning_rate": 1.4497685185185186e-05,
      "loss": 0.0341,
      "step": 2377
    },
    {
      "epoch": 0.27523148148148147,
      "grad_norm": 0.26376208662986755,
      "learning_rate": 1.4495370370370372e-05,
      "loss": 0.0515,
      "step": 2378
    },
    {
      "epoch": 0.27534722222222224,
      "grad_norm": 0.21790841221809387,
      "learning_rate": 1.4493055555555558e-05,
      "loss": 0.04,
      "step": 2379
    },
    {
      "epoch": 0.27546296296296297,
      "grad_norm": 0.40789252519607544,
      "learning_rate": 1.4490740740740742e-05,
      "loss": 0.0474,
      "step": 2380
    },
    {
      "epoch": 0.2755787037037037,
      "grad_norm": 0.9242231845855713,
      "learning_rate": 1.4488425925925926e-05,
      "loss": 0.0565,
      "step": 2381
    },
    {
      "epoch": 0.27569444444444446,
      "grad_norm": 0.27388569712638855,
      "learning_rate": 1.4486111111111112e-05,
      "loss": 0.0531,
      "step": 2382
    },
    {
      "epoch": 0.2758101851851852,
      "grad_norm": 0.3227585256099701,
      "learning_rate": 1.4483796296296298e-05,
      "loss": 0.0635,
      "step": 2383
    },
    {
      "epoch": 0.2759259259259259,
      "grad_norm": 0.2857602536678314,
      "learning_rate": 1.4481481481481483e-05,
      "loss": 0.0525,
      "step": 2384
    },
    {
      "epoch": 0.2760416666666667,
      "grad_norm": 0.29562869668006897,
      "learning_rate": 1.4479166666666669e-05,
      "loss": 0.058,
      "step": 2385
    },
    {
      "epoch": 0.2761574074074074,
      "grad_norm": 0.18347975611686707,
      "learning_rate": 1.4476851851851853e-05,
      "loss": 0.0353,
      "step": 2386
    },
    {
      "epoch": 0.27627314814814813,
      "grad_norm": 25.402616500854492,
      "learning_rate": 1.4474537037037037e-05,
      "loss": 0.1953,
      "step": 2387
    },
    {
      "epoch": 0.2763888888888889,
      "grad_norm": 0.23536767065525055,
      "learning_rate": 1.4472222222222223e-05,
      "loss": 0.0409,
      "step": 2388
    },
    {
      "epoch": 0.27650462962962963,
      "grad_norm": 0.21391569077968597,
      "learning_rate": 1.4469907407407409e-05,
      "loss": 0.042,
      "step": 2389
    },
    {
      "epoch": 0.27662037037037035,
      "grad_norm": 0.212355375289917,
      "learning_rate": 1.4467592592592595e-05,
      "loss": 0.0415,
      "step": 2390
    },
    {
      "epoch": 0.2767361111111111,
      "grad_norm": 10.214112281799316,
      "learning_rate": 1.4465277777777779e-05,
      "loss": 2.5323,
      "step": 2391
    },
    {
      "epoch": 0.27685185185185185,
      "grad_norm": 113.03467559814453,
      "learning_rate": 1.4462962962962963e-05,
      "loss": 0.8808,
      "step": 2392
    },
    {
      "epoch": 0.27696759259259257,
      "grad_norm": 0.2512751519680023,
      "learning_rate": 1.446064814814815e-05,
      "loss": 0.0499,
      "step": 2393
    },
    {
      "epoch": 0.27708333333333335,
      "grad_norm": 0.3634363114833832,
      "learning_rate": 1.4458333333333334e-05,
      "loss": 0.0518,
      "step": 2394
    },
    {
      "epoch": 0.27719907407407407,
      "grad_norm": 2.0688319206237793,
      "learning_rate": 1.445601851851852e-05,
      "loss": 0.087,
      "step": 2395
    },
    {
      "epoch": 0.2773148148148148,
      "grad_norm": 0.2562752664089203,
      "learning_rate": 1.4453703703703705e-05,
      "loss": 0.0487,
      "step": 2396
    },
    {
      "epoch": 0.27743055555555557,
      "grad_norm": 82.44486999511719,
      "learning_rate": 1.4451388888888891e-05,
      "loss": 1.4087,
      "step": 2397
    },
    {
      "epoch": 0.2775462962962963,
      "grad_norm": 0.3456994593143463,
      "learning_rate": 1.4449074074074074e-05,
      "loss": 0.0617,
      "step": 2398
    },
    {
      "epoch": 0.277662037037037,
      "grad_norm": 44.89563751220703,
      "learning_rate": 1.444675925925926e-05,
      "loss": 0.4233,
      "step": 2399
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 0.2875446081161499,
      "learning_rate": 1.4444444444444446e-05,
      "loss": 0.0524,
      "step": 2400
    },
    {
      "epoch": 0.2778935185185185,
      "grad_norm": 0.28823891282081604,
      "learning_rate": 1.4442129629629632e-05,
      "loss": 0.0573,
      "step": 2401
    },
    {
      "epoch": 0.27800925925925923,
      "grad_norm": 0.26775848865509033,
      "learning_rate": 1.4439814814814816e-05,
      "loss": 0.0361,
      "step": 2402
    },
    {
      "epoch": 0.278125,
      "grad_norm": 0.3473609983921051,
      "learning_rate": 1.4437500000000002e-05,
      "loss": 0.0684,
      "step": 2403
    },
    {
      "epoch": 0.27824074074074073,
      "grad_norm": 0.47809678316116333,
      "learning_rate": 1.4435185185185186e-05,
      "loss": 0.0491,
      "step": 2404
    },
    {
      "epoch": 0.27835648148148145,
      "grad_norm": 0.22667984664440155,
      "learning_rate": 1.443287037037037e-05,
      "loss": 0.0407,
      "step": 2405
    },
    {
      "epoch": 0.27847222222222223,
      "grad_norm": 0.22090861201286316,
      "learning_rate": 1.4430555555555556e-05,
      "loss": 0.0419,
      "step": 2406
    },
    {
      "epoch": 0.27858796296296295,
      "grad_norm": 0.9955098628997803,
      "learning_rate": 1.4428240740740742e-05,
      "loss": 0.0561,
      "step": 2407
    },
    {
      "epoch": 0.27870370370370373,
      "grad_norm": 16.229433059692383,
      "learning_rate": 1.4425925925925928e-05,
      "loss": 2.4578,
      "step": 2408
    },
    {
      "epoch": 0.27881944444444445,
      "grad_norm": 0.32768014073371887,
      "learning_rate": 1.4423611111111113e-05,
      "loss": 0.0639,
      "step": 2409
    },
    {
      "epoch": 0.2789351851851852,
      "grad_norm": 0.21378488838672638,
      "learning_rate": 1.4421296296296297e-05,
      "loss": 0.0417,
      "step": 2410
    },
    {
      "epoch": 0.27905092592592595,
      "grad_norm": 0.31013789772987366,
      "learning_rate": 1.4418981481481483e-05,
      "loss": 0.0618,
      "step": 2411
    },
    {
      "epoch": 0.2791666666666667,
      "grad_norm": 0.9285427927970886,
      "learning_rate": 1.4416666666666667e-05,
      "loss": 0.0595,
      "step": 2412
    },
    {
      "epoch": 0.2792824074074074,
      "grad_norm": 0.5850256085395813,
      "learning_rate": 1.4414351851851853e-05,
      "loss": 0.0591,
      "step": 2413
    },
    {
      "epoch": 0.2793981481481482,
      "grad_norm": 0.882779061794281,
      "learning_rate": 1.4412037037037039e-05,
      "loss": 0.0523,
      "step": 2414
    },
    {
      "epoch": 0.2795138888888889,
      "grad_norm": 0.2976916432380676,
      "learning_rate": 1.4409722222222225e-05,
      "loss": 0.0426,
      "step": 2415
    },
    {
      "epoch": 0.2796296296296296,
      "grad_norm": 0.2167157232761383,
      "learning_rate": 1.4407407407407407e-05,
      "loss": 0.0428,
      "step": 2416
    },
    {
      "epoch": 0.2797453703703704,
      "grad_norm": 0.2757717967033386,
      "learning_rate": 1.4405092592592593e-05,
      "loss": 0.039,
      "step": 2417
    },
    {
      "epoch": 0.2798611111111111,
      "grad_norm": 0.6330306529998779,
      "learning_rate": 1.440277777777778e-05,
      "loss": 0.0587,
      "step": 2418
    },
    {
      "epoch": 0.27997685185185184,
      "grad_norm": 0.3387012779712677,
      "learning_rate": 1.4400462962962965e-05,
      "loss": 0.0486,
      "step": 2419
    },
    {
      "epoch": 0.2800925925925926,
      "grad_norm": 0.1772201806306839,
      "learning_rate": 1.439814814814815e-05,
      "loss": 0.0344,
      "step": 2420
    },
    {
      "epoch": 0.28020833333333334,
      "grad_norm": 9.392631530761719,
      "learning_rate": 1.4395833333333335e-05,
      "loss": 0.1029,
      "step": 2421
    },
    {
      "epoch": 0.28032407407407406,
      "grad_norm": 0.2965809106826782,
      "learning_rate": 1.439351851851852e-05,
      "loss": 0.0578,
      "step": 2422
    },
    {
      "epoch": 0.28043981481481484,
      "grad_norm": 0.2249700278043747,
      "learning_rate": 1.4391203703703704e-05,
      "loss": 0.0398,
      "step": 2423
    },
    {
      "epoch": 0.28055555555555556,
      "grad_norm": 0.2137589305639267,
      "learning_rate": 1.438888888888889e-05,
      "loss": 0.0417,
      "step": 2424
    },
    {
      "epoch": 0.2806712962962963,
      "grad_norm": 0.23079513013362885,
      "learning_rate": 1.4386574074074076e-05,
      "loss": 0.0457,
      "step": 2425
    },
    {
      "epoch": 0.28078703703703706,
      "grad_norm": 0.24284929037094116,
      "learning_rate": 1.4384259259259262e-05,
      "loss": 0.0436,
      "step": 2426
    },
    {
      "epoch": 0.2809027777777778,
      "grad_norm": 0.2516764998435974,
      "learning_rate": 1.4381944444444446e-05,
      "loss": 0.0481,
      "step": 2427
    },
    {
      "epoch": 0.2810185185185185,
      "grad_norm": 0.22496691346168518,
      "learning_rate": 1.437962962962963e-05,
      "loss": 0.0448,
      "step": 2428
    },
    {
      "epoch": 0.2811342592592593,
      "grad_norm": 0.32513588666915894,
      "learning_rate": 1.4377314814814816e-05,
      "loss": 0.065,
      "step": 2429
    },
    {
      "epoch": 0.28125,
      "grad_norm": 0.17846953868865967,
      "learning_rate": 1.4375e-05,
      "loss": 0.0346,
      "step": 2430
    },
    {
      "epoch": 0.2813657407407407,
      "grad_norm": 8.842604637145996,
      "learning_rate": 1.4372685185185186e-05,
      "loss": 2.4026,
      "step": 2431
    },
    {
      "epoch": 0.2814814814814815,
      "grad_norm": 0.3387680649757385,
      "learning_rate": 1.4370370370370372e-05,
      "loss": 0.0483,
      "step": 2432
    },
    {
      "epoch": 0.2815972222222222,
      "grad_norm": 0.24835442006587982,
      "learning_rate": 1.4368055555555555e-05,
      "loss": 0.0342,
      "step": 2433
    },
    {
      "epoch": 0.28171296296296294,
      "grad_norm": 0.28149911761283875,
      "learning_rate": 1.436574074074074e-05,
      "loss": 0.0396,
      "step": 2434
    },
    {
      "epoch": 0.2818287037037037,
      "grad_norm": 0.1909928321838379,
      "learning_rate": 1.4363425925925927e-05,
      "loss": 0.0377,
      "step": 2435
    },
    {
      "epoch": 0.28194444444444444,
      "grad_norm": 0.42583903670310974,
      "learning_rate": 1.4361111111111113e-05,
      "loss": 0.041,
      "step": 2436
    },
    {
      "epoch": 0.28206018518518516,
      "grad_norm": 0.2464781254529953,
      "learning_rate": 1.4358796296296299e-05,
      "loss": 0.0484,
      "step": 2437
    },
    {
      "epoch": 0.28217592592592594,
      "grad_norm": 40.97510528564453,
      "learning_rate": 1.4356481481481483e-05,
      "loss": 2.0437,
      "step": 2438
    },
    {
      "epoch": 0.28229166666666666,
      "grad_norm": 27.449838638305664,
      "learning_rate": 1.4354166666666667e-05,
      "loss": 2.287,
      "step": 2439
    },
    {
      "epoch": 0.2824074074074074,
      "grad_norm": 0.2050974816083908,
      "learning_rate": 1.4351851851851853e-05,
      "loss": 0.0404,
      "step": 2440
    },
    {
      "epoch": 0.28252314814814816,
      "grad_norm": 0.35815349221229553,
      "learning_rate": 1.4349537037037037e-05,
      "loss": 0.0675,
      "step": 2441
    },
    {
      "epoch": 0.2826388888888889,
      "grad_norm": 0.22223204374313354,
      "learning_rate": 1.4347222222222223e-05,
      "loss": 0.0432,
      "step": 2442
    },
    {
      "epoch": 0.2827546296296296,
      "grad_norm": 0.9626216292381287,
      "learning_rate": 1.434490740740741e-05,
      "loss": 0.0789,
      "step": 2443
    },
    {
      "epoch": 0.2828703703703704,
      "grad_norm": 0.1902657300233841,
      "learning_rate": 1.4342592592592595e-05,
      "loss": 0.0373,
      "step": 2444
    },
    {
      "epoch": 0.2829861111111111,
      "grad_norm": 0.20952318608760834,
      "learning_rate": 1.4340277777777778e-05,
      "loss": 0.0405,
      "step": 2445
    },
    {
      "epoch": 0.2831018518518518,
      "grad_norm": 0.23180469870567322,
      "learning_rate": 1.4337962962962964e-05,
      "loss": 0.0414,
      "step": 2446
    },
    {
      "epoch": 0.2832175925925926,
      "grad_norm": 0.2817842364311218,
      "learning_rate": 1.433564814814815e-05,
      "loss": 0.051,
      "step": 2447
    },
    {
      "epoch": 0.2833333333333333,
      "grad_norm": 0.2445952445268631,
      "learning_rate": 1.4333333333333334e-05,
      "loss": 0.0473,
      "step": 2448
    },
    {
      "epoch": 0.28344907407407405,
      "grad_norm": 0.21193227171897888,
      "learning_rate": 1.433101851851852e-05,
      "loss": 0.041,
      "step": 2449
    },
    {
      "epoch": 0.2835648148148148,
      "grad_norm": 0.27886930108070374,
      "learning_rate": 1.4328703703703706e-05,
      "loss": 0.0496,
      "step": 2450
    },
    {
      "epoch": 0.28368055555555555,
      "grad_norm": 0.2315063774585724,
      "learning_rate": 1.4326388888888888e-05,
      "loss": 0.0445,
      "step": 2451
    },
    {
      "epoch": 0.28379629629629627,
      "grad_norm": 0.29566118121147156,
      "learning_rate": 1.4324074074074074e-05,
      "loss": 0.0497,
      "step": 2452
    },
    {
      "epoch": 0.28391203703703705,
      "grad_norm": 0.258598655462265,
      "learning_rate": 1.432175925925926e-05,
      "loss": 0.0504,
      "step": 2453
    },
    {
      "epoch": 0.28402777777777777,
      "grad_norm": 0.30231520533561707,
      "learning_rate": 1.4319444444444446e-05,
      "loss": 0.0573,
      "step": 2454
    },
    {
      "epoch": 0.28414351851851855,
      "grad_norm": 0.2778290808200836,
      "learning_rate": 1.4317129629629632e-05,
      "loss": 0.0553,
      "step": 2455
    },
    {
      "epoch": 0.28425925925925927,
      "grad_norm": 0.41597917675971985,
      "learning_rate": 1.4314814814814816e-05,
      "loss": 0.0406,
      "step": 2456
    },
    {
      "epoch": 0.284375,
      "grad_norm": 0.18380478024482727,
      "learning_rate": 1.43125e-05,
      "loss": 0.0355,
      "step": 2457
    },
    {
      "epoch": 0.28449074074074077,
      "grad_norm": 0.33668243885040283,
      "learning_rate": 1.4310185185185187e-05,
      "loss": 0.0568,
      "step": 2458
    },
    {
      "epoch": 0.2846064814814815,
      "grad_norm": 0.19942116737365723,
      "learning_rate": 1.430787037037037e-05,
      "loss": 0.0393,
      "step": 2459
    },
    {
      "epoch": 0.2847222222222222,
      "grad_norm": 6.072751522064209,
      "learning_rate": 1.4305555555555557e-05,
      "loss": 0.0709,
      "step": 2460
    },
    {
      "epoch": 0.284837962962963,
      "grad_norm": 0.18689565360546112,
      "learning_rate": 1.4303240740740743e-05,
      "loss": 0.0366,
      "step": 2461
    },
    {
      "epoch": 0.2849537037037037,
      "grad_norm": 0.23674090206623077,
      "learning_rate": 1.4300925925925929e-05,
      "loss": 0.0334,
      "step": 2462
    },
    {
      "epoch": 0.28506944444444443,
      "grad_norm": 0.16591931879520416,
      "learning_rate": 1.4298611111111111e-05,
      "loss": 0.0327,
      "step": 2463
    },
    {
      "epoch": 0.2851851851851852,
      "grad_norm": 0.22598223388195038,
      "learning_rate": 1.4296296296296297e-05,
      "loss": 0.0434,
      "step": 2464
    },
    {
      "epoch": 0.28530092592592593,
      "grad_norm": 0.36893230676651,
      "learning_rate": 1.4293981481481483e-05,
      "loss": 0.0474,
      "step": 2465
    },
    {
      "epoch": 0.28541666666666665,
      "grad_norm": 0.16875414550304413,
      "learning_rate": 1.4291666666666667e-05,
      "loss": 0.0324,
      "step": 2466
    },
    {
      "epoch": 0.28553240740740743,
      "grad_norm": 0.25634580850601196,
      "learning_rate": 1.4289351851851853e-05,
      "loss": 0.0501,
      "step": 2467
    },
    {
      "epoch": 0.28564814814814815,
      "grad_norm": 0.39263084530830383,
      "learning_rate": 1.428703703703704e-05,
      "loss": 0.0533,
      "step": 2468
    },
    {
      "epoch": 0.2857638888888889,
      "grad_norm": 0.20230844616889954,
      "learning_rate": 1.4284722222222222e-05,
      "loss": 0.0396,
      "step": 2469
    },
    {
      "epoch": 0.28587962962962965,
      "grad_norm": 0.20945219695568085,
      "learning_rate": 1.4282407407407408e-05,
      "loss": 0.0377,
      "step": 2470
    },
    {
      "epoch": 0.28599537037037037,
      "grad_norm": 0.7979968786239624,
      "learning_rate": 1.4280092592592594e-05,
      "loss": 0.0722,
      "step": 2471
    },
    {
      "epoch": 0.2861111111111111,
      "grad_norm": 0.2080497145652771,
      "learning_rate": 1.427777777777778e-05,
      "loss": 0.037,
      "step": 2472
    },
    {
      "epoch": 0.28622685185185187,
      "grad_norm": 0.4077921211719513,
      "learning_rate": 1.4275462962962966e-05,
      "loss": 0.045,
      "step": 2473
    },
    {
      "epoch": 0.2863425925925926,
      "grad_norm": 0.16342157125473022,
      "learning_rate": 1.427314814814815e-05,
      "loss": 0.0321,
      "step": 2474
    },
    {
      "epoch": 0.2864583333333333,
      "grad_norm": 0.26244044303894043,
      "learning_rate": 1.4270833333333334e-05,
      "loss": 0.045,
      "step": 2475
    },
    {
      "epoch": 0.2865740740740741,
      "grad_norm": 0.9670937657356262,
      "learning_rate": 1.426851851851852e-05,
      "loss": 0.0668,
      "step": 2476
    },
    {
      "epoch": 0.2866898148148148,
      "grad_norm": 95.69551849365234,
      "learning_rate": 1.4266203703703704e-05,
      "loss": 0.603,
      "step": 2477
    },
    {
      "epoch": 0.28680555555555554,
      "grad_norm": 0.17588502168655396,
      "learning_rate": 1.426388888888889e-05,
      "loss": 0.0342,
      "step": 2478
    },
    {
      "epoch": 0.2869212962962963,
      "grad_norm": 0.2540999948978424,
      "learning_rate": 1.4261574074074076e-05,
      "loss": 0.0502,
      "step": 2479
    },
    {
      "epoch": 0.28703703703703703,
      "grad_norm": 28.833776473999023,
      "learning_rate": 1.4259259259259259e-05,
      "loss": 1.6483,
      "step": 2480
    },
    {
      "epoch": 0.28715277777777776,
      "grad_norm": 0.15919972956180573,
      "learning_rate": 1.4256944444444445e-05,
      "loss": 0.0314,
      "step": 2481
    },
    {
      "epoch": 0.28726851851851853,
      "grad_norm": 0.17156697809696198,
      "learning_rate": 1.425462962962963e-05,
      "loss": 0.033,
      "step": 2482
    },
    {
      "epoch": 0.28738425925925926,
      "grad_norm": 9.070643424987793,
      "learning_rate": 1.4252314814814817e-05,
      "loss": 0.0903,
      "step": 2483
    },
    {
      "epoch": 0.2875,
      "grad_norm": 0.21753916144371033,
      "learning_rate": 1.425e-05,
      "loss": 0.0416,
      "step": 2484
    },
    {
      "epoch": 0.28761574074074076,
      "grad_norm": 0.1760084629058838,
      "learning_rate": 1.4247685185185187e-05,
      "loss": 0.0346,
      "step": 2485
    },
    {
      "epoch": 0.2877314814814815,
      "grad_norm": 0.2059558480978012,
      "learning_rate": 1.4245370370370371e-05,
      "loss": 0.0386,
      "step": 2486
    },
    {
      "epoch": 0.2878472222222222,
      "grad_norm": 0.30263447761535645,
      "learning_rate": 1.4243055555555557e-05,
      "loss": 0.0515,
      "step": 2487
    },
    {
      "epoch": 0.287962962962963,
      "grad_norm": 0.2823607921600342,
      "learning_rate": 1.4240740740740741e-05,
      "loss": 0.0526,
      "step": 2488
    },
    {
      "epoch": 0.2880787037037037,
      "grad_norm": 0.20488958060741425,
      "learning_rate": 1.4238425925925927e-05,
      "loss": 0.0392,
      "step": 2489
    },
    {
      "epoch": 0.2881944444444444,
      "grad_norm": 0.2625681459903717,
      "learning_rate": 1.4236111111111113e-05,
      "loss": 0.0346,
      "step": 2490
    },
    {
      "epoch": 0.2883101851851852,
      "grad_norm": 0.249232679605484,
      "learning_rate": 1.4233796296296299e-05,
      "loss": 0.0473,
      "step": 2491
    },
    {
      "epoch": 0.2884259259259259,
      "grad_norm": 0.2041587382555008,
      "learning_rate": 1.4231481481481482e-05,
      "loss": 0.0387,
      "step": 2492
    },
    {
      "epoch": 0.28854166666666664,
      "grad_norm": 0.38522282242774963,
      "learning_rate": 1.4229166666666668e-05,
      "loss": 0.0618,
      "step": 2493
    },
    {
      "epoch": 0.2886574074074074,
      "grad_norm": 0.19938839972019196,
      "learning_rate": 1.4226851851851853e-05,
      "loss": 0.0387,
      "step": 2494
    },
    {
      "epoch": 0.28877314814814814,
      "grad_norm": 0.3810931444168091,
      "learning_rate": 1.4224537037037038e-05,
      "loss": 0.0513,
      "step": 2495
    },
    {
      "epoch": 0.28888888888888886,
      "grad_norm": 13.677824974060059,
      "learning_rate": 1.4222222222222224e-05,
      "loss": 2.4074,
      "step": 2496
    },
    {
      "epoch": 0.28900462962962964,
      "grad_norm": 0.2170128971338272,
      "learning_rate": 1.421990740740741e-05,
      "loss": 0.0418,
      "step": 2497
    },
    {
      "epoch": 0.28912037037037036,
      "grad_norm": 0.290764182806015,
      "learning_rate": 1.4217592592592592e-05,
      "loss": 0.0521,
      "step": 2498
    },
    {
      "epoch": 0.28923611111111114,
      "grad_norm": 0.1967608779668808,
      "learning_rate": 1.4215277777777778e-05,
      "loss": 0.0383,
      "step": 2499
    },
    {
      "epoch": 0.28935185185185186,
      "grad_norm": 0.24017146229743958,
      "learning_rate": 1.4212962962962964e-05,
      "loss": 0.0458,
      "step": 2500
    },
    {
      "epoch": 0.2894675925925926,
      "grad_norm": 0.28099074959754944,
      "learning_rate": 1.421064814814815e-05,
      "loss": 0.0441,
      "step": 2501
    },
    {
      "epoch": 0.28958333333333336,
      "grad_norm": 5.586203098297119,
      "learning_rate": 1.4208333333333336e-05,
      "loss": 2.6053,
      "step": 2502
    },
    {
      "epoch": 0.2896990740740741,
      "grad_norm": 0.23087063431739807,
      "learning_rate": 1.420601851851852e-05,
      "loss": 0.0324,
      "step": 2503
    },
    {
      "epoch": 0.2898148148148148,
      "grad_norm": 0.853594958782196,
      "learning_rate": 1.4203703703703704e-05,
      "loss": 0.0627,
      "step": 2504
    },
    {
      "epoch": 0.2899305555555556,
      "grad_norm": 0.20792417228221893,
      "learning_rate": 1.420138888888889e-05,
      "loss": 0.0407,
      "step": 2505
    },
    {
      "epoch": 0.2900462962962963,
      "grad_norm": 0.6297369003295898,
      "learning_rate": 1.4199074074074075e-05,
      "loss": 0.0512,
      "step": 2506
    },
    {
      "epoch": 0.290162037037037,
      "grad_norm": 0.2655734121799469,
      "learning_rate": 1.419675925925926e-05,
      "loss": 0.0516,
      "step": 2507
    },
    {
      "epoch": 0.2902777777777778,
      "grad_norm": 7.886687278747559,
      "learning_rate": 1.4194444444444447e-05,
      "loss": 0.0738,
      "step": 2508
    },
    {
      "epoch": 0.2903935185185185,
      "grad_norm": 0.529075026512146,
      "learning_rate": 1.4192129629629632e-05,
      "loss": 0.0459,
      "step": 2509
    },
    {
      "epoch": 0.29050925925925924,
      "grad_norm": 0.19286282360553741,
      "learning_rate": 1.4189814814814815e-05,
      "loss": 0.0375,
      "step": 2510
    },
    {
      "epoch": 0.290625,
      "grad_norm": 21.959516525268555,
      "learning_rate": 1.4187500000000001e-05,
      "loss": 0.161,
      "step": 2511
    },
    {
      "epoch": 0.29074074074074074,
      "grad_norm": 0.18887270987033844,
      "learning_rate": 1.4185185185185187e-05,
      "loss": 0.037,
      "step": 2512
    },
    {
      "epoch": 0.29085648148148147,
      "grad_norm": 0.24908073246479034,
      "learning_rate": 1.4182870370370371e-05,
      "loss": 0.047,
      "step": 2513
    },
    {
      "epoch": 0.29097222222222224,
      "grad_norm": 0.37757477164268494,
      "learning_rate": 1.4180555555555557e-05,
      "loss": 0.0452,
      "step": 2514
    },
    {
      "epoch": 0.29108796296296297,
      "grad_norm": 3.1410038471221924,
      "learning_rate": 1.4178240740740743e-05,
      "loss": 0.0654,
      "step": 2515
    },
    {
      "epoch": 0.2912037037037037,
      "grad_norm": 0.16833749413490295,
      "learning_rate": 1.4175925925925926e-05,
      "loss": 0.033,
      "step": 2516
    },
    {
      "epoch": 0.29131944444444446,
      "grad_norm": 0.20836038887500763,
      "learning_rate": 1.4173611111111112e-05,
      "loss": 0.0395,
      "step": 2517
    },
    {
      "epoch": 0.2914351851851852,
      "grad_norm": 0.3526490330696106,
      "learning_rate": 1.4171296296296297e-05,
      "loss": 0.0623,
      "step": 2518
    },
    {
      "epoch": 0.2915509259259259,
      "grad_norm": 0.27337074279785156,
      "learning_rate": 1.4168981481481483e-05,
      "loss": 0.0476,
      "step": 2519
    },
    {
      "epoch": 0.2916666666666667,
      "grad_norm": 0.19127042591571808,
      "learning_rate": 1.416666666666667e-05,
      "loss": 0.0354,
      "step": 2520
    },
    {
      "epoch": 0.2917824074074074,
      "grad_norm": 0.16699515283107758,
      "learning_rate": 1.4164351851851852e-05,
      "loss": 0.0317,
      "step": 2521
    },
    {
      "epoch": 0.29189814814814813,
      "grad_norm": 0.2643101215362549,
      "learning_rate": 1.4162037037037038e-05,
      "loss": 0.0495,
      "step": 2522
    },
    {
      "epoch": 0.2920138888888889,
      "grad_norm": 0.4088514745235443,
      "learning_rate": 1.4159722222222224e-05,
      "loss": 0.048,
      "step": 2523
    },
    {
      "epoch": 0.29212962962962963,
      "grad_norm": 0.22094081342220306,
      "learning_rate": 1.4157407407407408e-05,
      "loss": 0.0423,
      "step": 2524
    },
    {
      "epoch": 0.29224537037037035,
      "grad_norm": 0.2190081775188446,
      "learning_rate": 1.4155092592592594e-05,
      "loss": 0.0307,
      "step": 2525
    },
    {
      "epoch": 0.2923611111111111,
      "grad_norm": 0.25421053171157837,
      "learning_rate": 1.415277777777778e-05,
      "loss": 0.0489,
      "step": 2526
    },
    {
      "epoch": 0.29247685185185185,
      "grad_norm": 0.18190549314022064,
      "learning_rate": 1.4150462962962963e-05,
      "loss": 0.0329,
      "step": 2527
    },
    {
      "epoch": 0.29259259259259257,
      "grad_norm": 0.20165494084358215,
      "learning_rate": 1.4148148148148148e-05,
      "loss": 0.0355,
      "step": 2528
    },
    {
      "epoch": 0.29270833333333335,
      "grad_norm": 0.31545642018318176,
      "learning_rate": 1.4145833333333334e-05,
      "loss": 0.0474,
      "step": 2529
    },
    {
      "epoch": 0.29282407407407407,
      "grad_norm": 0.3034749925136566,
      "learning_rate": 1.414351851851852e-05,
      "loss": 0.0578,
      "step": 2530
    },
    {
      "epoch": 0.2929398148148148,
      "grad_norm": 0.28195783495903015,
      "learning_rate": 1.4141203703703705e-05,
      "loss": 0.0401,
      "step": 2531
    },
    {
      "epoch": 0.29305555555555557,
      "grad_norm": 0.26723766326904297,
      "learning_rate": 1.413888888888889e-05,
      "loss": 0.0504,
      "step": 2532
    },
    {
      "epoch": 0.2931712962962963,
      "grad_norm": 0.19921305775642395,
      "learning_rate": 1.4136574074074075e-05,
      "loss": 0.038,
      "step": 2533
    },
    {
      "epoch": 0.293287037037037,
      "grad_norm": 0.1700761467218399,
      "learning_rate": 1.4134259259259259e-05,
      "loss": 0.0333,
      "step": 2534
    },
    {
      "epoch": 0.2934027777777778,
      "grad_norm": 0.2480795979499817,
      "learning_rate": 1.4131944444444445e-05,
      "loss": 0.0465,
      "step": 2535
    },
    {
      "epoch": 0.2935185185185185,
      "grad_norm": 0.2698211073875427,
      "learning_rate": 1.4129629629629631e-05,
      "loss": 0.0493,
      "step": 2536
    },
    {
      "epoch": 0.29363425925925923,
      "grad_norm": 0.16522620618343353,
      "learning_rate": 1.4127314814814817e-05,
      "loss": 0.0313,
      "step": 2537
    },
    {
      "epoch": 0.29375,
      "grad_norm": 0.19171063601970673,
      "learning_rate": 1.4125000000000003e-05,
      "loss": 0.037,
      "step": 2538
    },
    {
      "epoch": 0.29386574074074073,
      "grad_norm": 0.20052969455718994,
      "learning_rate": 1.4122685185185185e-05,
      "loss": 0.0377,
      "step": 2539
    },
    {
      "epoch": 0.29398148148148145,
      "grad_norm": 0.3352835774421692,
      "learning_rate": 1.4120370370370371e-05,
      "loss": 0.0491,
      "step": 2540
    },
    {
      "epoch": 0.29409722222222223,
      "grad_norm": 0.26013681292533875,
      "learning_rate": 1.4118055555555557e-05,
      "loss": 0.0512,
      "step": 2541
    },
    {
      "epoch": 0.29421296296296295,
      "grad_norm": 0.23040902614593506,
      "learning_rate": 1.4115740740740742e-05,
      "loss": 0.0318,
      "step": 2542
    },
    {
      "epoch": 0.29432870370370373,
      "grad_norm": 0.3082188069820404,
      "learning_rate": 1.4113425925925927e-05,
      "loss": 0.044,
      "step": 2543
    },
    {
      "epoch": 0.29444444444444445,
      "grad_norm": 13.626431465148926,
      "learning_rate": 1.4111111111111113e-05,
      "loss": 2.4897,
      "step": 2544
    },
    {
      "epoch": 0.2945601851851852,
      "grad_norm": 0.186783567070961,
      "learning_rate": 1.4108796296296296e-05,
      "loss": 0.0359,
      "step": 2545
    },
    {
      "epoch": 0.29467592592592595,
      "grad_norm": 29.037273406982422,
      "learning_rate": 1.4106481481481482e-05,
      "loss": 2.0138,
      "step": 2546
    },
    {
      "epoch": 0.2947916666666667,
      "grad_norm": 0.2958577275276184,
      "learning_rate": 1.4104166666666668e-05,
      "loss": 0.0564,
      "step": 2547
    },
    {
      "epoch": 0.2949074074074074,
      "grad_norm": 0.2381100207567215,
      "learning_rate": 1.4101851851851854e-05,
      "loss": 0.0455,
      "step": 2548
    },
    {
      "epoch": 0.2950231481481482,
      "grad_norm": 27.13778305053711,
      "learning_rate": 1.4099537037037038e-05,
      "loss": 0.1206,
      "step": 2549
    },
    {
      "epoch": 0.2951388888888889,
      "grad_norm": 0.24699819087982178,
      "learning_rate": 1.4097222222222224e-05,
      "loss": 0.0483,
      "step": 2550
    },
    {
      "epoch": 0.2952546296296296,
      "grad_norm": 0.30646181106567383,
      "learning_rate": 1.4094907407407408e-05,
      "loss": 0.0516,
      "step": 2551
    },
    {
      "epoch": 0.2953703703703704,
      "grad_norm": 5.151930332183838,
      "learning_rate": 1.4092592592592592e-05,
      "loss": 0.0937,
      "step": 2552
    },
    {
      "epoch": 0.2954861111111111,
      "grad_norm": 0.21877872943878174,
      "learning_rate": 1.4090277777777778e-05,
      "loss": 0.0411,
      "step": 2553
    },
    {
      "epoch": 0.29560185185185184,
      "grad_norm": 0.17451295256614685,
      "learning_rate": 1.4087962962962964e-05,
      "loss": 0.0338,
      "step": 2554
    },
    {
      "epoch": 0.2957175925925926,
      "grad_norm": 0.1877838671207428,
      "learning_rate": 1.408564814814815e-05,
      "loss": 0.0368,
      "step": 2555
    },
    {
      "epoch": 0.29583333333333334,
      "grad_norm": 0.439683735370636,
      "learning_rate": 1.4083333333333336e-05,
      "loss": 0.0387,
      "step": 2556
    },
    {
      "epoch": 0.29594907407407406,
      "grad_norm": 0.2496771514415741,
      "learning_rate": 1.4081018518518519e-05,
      "loss": 0.0478,
      "step": 2557
    },
    {
      "epoch": 0.29606481481481484,
      "grad_norm": 0.3057260513305664,
      "learning_rate": 1.4078703703703705e-05,
      "loss": 0.0481,
      "step": 2558
    },
    {
      "epoch": 0.29618055555555556,
      "grad_norm": 66.86727905273438,
      "learning_rate": 1.407638888888889e-05,
      "loss": 1.4811,
      "step": 2559
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 5.415641784667969,
      "learning_rate": 1.4074074074074075e-05,
      "loss": 0.0704,
      "step": 2560
    },
    {
      "epoch": 0.29641203703703706,
      "grad_norm": 0.42682984471321106,
      "learning_rate": 1.4071759259259261e-05,
      "loss": 0.0495,
      "step": 2561
    },
    {
      "epoch": 0.2965277777777778,
      "grad_norm": 0.2504831850528717,
      "learning_rate": 1.4069444444444447e-05,
      "loss": 0.046,
      "step": 2562
    },
    {
      "epoch": 0.2966435185185185,
      "grad_norm": 0.18761566281318665,
      "learning_rate": 1.406712962962963e-05,
      "loss": 0.0357,
      "step": 2563
    },
    {
      "epoch": 0.2967592592592593,
      "grad_norm": 0.5845277905464172,
      "learning_rate": 1.4064814814814815e-05,
      "loss": 0.0513,
      "step": 2564
    },
    {
      "epoch": 0.296875,
      "grad_norm": 0.1831008344888687,
      "learning_rate": 1.4062500000000001e-05,
      "loss": 0.0336,
      "step": 2565
    },
    {
      "epoch": 0.2969907407407407,
      "grad_norm": 8.844314575195312,
      "learning_rate": 1.4060185185185187e-05,
      "loss": 0.1465,
      "step": 2566
    },
    {
      "epoch": 0.2971064814814815,
      "grad_norm": 0.185603067278862,
      "learning_rate": 1.4057870370370372e-05,
      "loss": 0.0352,
      "step": 2567
    },
    {
      "epoch": 0.2972222222222222,
      "grad_norm": 45.82939529418945,
      "learning_rate": 1.4055555555555556e-05,
      "loss": 1.8395,
      "step": 2568
    },
    {
      "epoch": 0.29733796296296294,
      "grad_norm": 0.4192572236061096,
      "learning_rate": 1.4053240740740742e-05,
      "loss": 0.0596,
      "step": 2569
    },
    {
      "epoch": 0.2974537037037037,
      "grad_norm": 0.2287447452545166,
      "learning_rate": 1.4050925925925926e-05,
      "loss": 0.0437,
      "step": 2570
    },
    {
      "epoch": 0.29756944444444444,
      "grad_norm": 0.23653912544250488,
      "learning_rate": 1.4048611111111112e-05,
      "loss": 0.0442,
      "step": 2571
    },
    {
      "epoch": 0.29768518518518516,
      "grad_norm": 0.21940919756889343,
      "learning_rate": 1.4046296296296298e-05,
      "loss": 0.0385,
      "step": 2572
    },
    {
      "epoch": 0.29780092592592594,
      "grad_norm": 0.21630419790744781,
      "learning_rate": 1.4043981481481484e-05,
      "loss": 0.042,
      "step": 2573
    },
    {
      "epoch": 0.29791666666666666,
      "grad_norm": 81.53306579589844,
      "learning_rate": 1.4041666666666666e-05,
      "loss": 0.5294,
      "step": 2574
    },
    {
      "epoch": 0.2980324074074074,
      "grad_norm": 0.4186966121196747,
      "learning_rate": 1.4039351851851852e-05,
      "loss": 0.0542,
      "step": 2575
    },
    {
      "epoch": 0.29814814814814816,
      "grad_norm": 0.2557004690170288,
      "learning_rate": 1.4037037037037038e-05,
      "loss": 0.0499,
      "step": 2576
    },
    {
      "epoch": 0.2982638888888889,
      "grad_norm": 0.2539660334587097,
      "learning_rate": 1.4034722222222224e-05,
      "loss": 0.0502,
      "step": 2577
    },
    {
      "epoch": 0.2983796296296296,
      "grad_norm": 0.9929208755493164,
      "learning_rate": 1.4032407407407408e-05,
      "loss": 0.0494,
      "step": 2578
    },
    {
      "epoch": 0.2984953703703704,
      "grad_norm": 0.20493613183498383,
      "learning_rate": 1.4030092592592594e-05,
      "loss": 0.0372,
      "step": 2579
    },
    {
      "epoch": 0.2986111111111111,
      "grad_norm": 0.2598350942134857,
      "learning_rate": 1.4027777777777779e-05,
      "loss": 0.051,
      "step": 2580
    },
    {
      "epoch": 0.2987268518518518,
      "grad_norm": 0.24779409170150757,
      "learning_rate": 1.4025462962962963e-05,
      "loss": 0.0479,
      "step": 2581
    },
    {
      "epoch": 0.2988425925925926,
      "grad_norm": 0.28571370244026184,
      "learning_rate": 1.4023148148148149e-05,
      "loss": 0.0575,
      "step": 2582
    },
    {
      "epoch": 0.2989583333333333,
      "grad_norm": 0.2356269359588623,
      "learning_rate": 1.4020833333333335e-05,
      "loss": 0.0373,
      "step": 2583
    },
    {
      "epoch": 0.29907407407407405,
      "grad_norm": 0.18757274746894836,
      "learning_rate": 1.401851851851852e-05,
      "loss": 0.0362,
      "step": 2584
    },
    {
      "epoch": 0.2991898148148148,
      "grad_norm": 0.2568384110927582,
      "learning_rate": 1.4016203703703705e-05,
      "loss": 0.0456,
      "step": 2585
    },
    {
      "epoch": 0.29930555555555555,
      "grad_norm": 0.18653298914432526,
      "learning_rate": 1.401388888888889e-05,
      "loss": 0.0361,
      "step": 2586
    },
    {
      "epoch": 0.29942129629629627,
      "grad_norm": 0.24686220288276672,
      "learning_rate": 1.4011574074074075e-05,
      "loss": 0.045,
      "step": 2587
    },
    {
      "epoch": 0.29953703703703705,
      "grad_norm": 0.17743872106075287,
      "learning_rate": 1.400925925925926e-05,
      "loss": 0.0344,
      "step": 2588
    },
    {
      "epoch": 0.29965277777777777,
      "grad_norm": 0.19579322636127472,
      "learning_rate": 1.4006944444444445e-05,
      "loss": 0.0373,
      "step": 2589
    },
    {
      "epoch": 0.29976851851851855,
      "grad_norm": 0.24796727299690247,
      "learning_rate": 1.4004629629629631e-05,
      "loss": 0.0317,
      "step": 2590
    },
    {
      "epoch": 0.29988425925925927,
      "grad_norm": 0.2210271805524826,
      "learning_rate": 1.4002314814814817e-05,
      "loss": 0.0405,
      "step": 2591
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2797219157218933,
      "learning_rate": 1.4e-05,
      "loss": 0.0548,
      "step": 2592
    },
    {
      "epoch": 0.30011574074074077,
      "grad_norm": 0.1919982135295868,
      "learning_rate": 1.3997685185185186e-05,
      "loss": 0.0372,
      "step": 2593
    },
    {
      "epoch": 0.3002314814814815,
      "grad_norm": 0.4587525725364685,
      "learning_rate": 1.3995370370370372e-05,
      "loss": 0.0381,
      "step": 2594
    },
    {
      "epoch": 0.3003472222222222,
      "grad_norm": 16.707088470458984,
      "learning_rate": 1.3993055555555558e-05,
      "loss": 0.0827,
      "step": 2595
    },
    {
      "epoch": 0.300462962962963,
      "grad_norm": 0.228175550699234,
      "learning_rate": 1.3990740740740742e-05,
      "loss": 0.0433,
      "step": 2596
    },
    {
      "epoch": 0.3005787037037037,
      "grad_norm": 0.24764996767044067,
      "learning_rate": 1.3988425925925928e-05,
      "loss": 0.0455,
      "step": 2597
    },
    {
      "epoch": 0.30069444444444443,
      "grad_norm": 0.7429429292678833,
      "learning_rate": 1.3986111111111112e-05,
      "loss": 0.058,
      "step": 2598
    },
    {
      "epoch": 0.3008101851851852,
      "grad_norm": 7.1310505867004395,
      "learning_rate": 1.3983796296296296e-05,
      "loss": 0.1162,
      "step": 2599
    },
    {
      "epoch": 0.30092592592592593,
      "grad_norm": 0.21695458889007568,
      "learning_rate": 1.3981481481481482e-05,
      "loss": 0.0427,
      "step": 2600
    },
    {
      "epoch": 0.30104166666666665,
      "grad_norm": 0.2472609132528305,
      "learning_rate": 1.3979166666666668e-05,
      "loss": 0.0432,
      "step": 2601
    },
    {
      "epoch": 0.30115740740740743,
      "grad_norm": 0.18358547985553741,
      "learning_rate": 1.3976851851851854e-05,
      "loss": 0.0359,
      "step": 2602
    },
    {
      "epoch": 0.30127314814814815,
      "grad_norm": 29.944406509399414,
      "learning_rate": 1.3974537037037038e-05,
      "loss": 1.8738,
      "step": 2603
    },
    {
      "epoch": 0.3013888888888889,
      "grad_norm": 0.21714437007904053,
      "learning_rate": 1.3972222222222223e-05,
      "loss": 0.0407,
      "step": 2604
    },
    {
      "epoch": 0.30150462962962965,
      "grad_norm": 0.3546592593193054,
      "learning_rate": 1.3969907407407409e-05,
      "loss": 0.0504,
      "step": 2605
    },
    {
      "epoch": 0.30162037037037037,
      "grad_norm": 0.297791063785553,
      "learning_rate": 1.3967592592592595e-05,
      "loss": 0.0423,
      "step": 2606
    },
    {
      "epoch": 0.3017361111111111,
      "grad_norm": 0.4472273290157318,
      "learning_rate": 1.3965277777777779e-05,
      "loss": 0.0484,
      "step": 2607
    },
    {
      "epoch": 0.30185185185185187,
      "grad_norm": 71.72848510742188,
      "learning_rate": 1.3962962962962965e-05,
      "loss": 1.9608,
      "step": 2608
    },
    {
      "epoch": 0.3019675925925926,
      "grad_norm": 0.19449840486049652,
      "learning_rate": 1.396064814814815e-05,
      "loss": 0.0363,
      "step": 2609
    },
    {
      "epoch": 0.3020833333333333,
      "grad_norm": 0.2387940138578415,
      "learning_rate": 1.3958333333333333e-05,
      "loss": 0.0468,
      "step": 2610
    },
    {
      "epoch": 0.3021990740740741,
      "grad_norm": 0.16852131485939026,
      "learning_rate": 1.395601851851852e-05,
      "loss": 0.0322,
      "step": 2611
    },
    {
      "epoch": 0.3023148148148148,
      "grad_norm": 5.962849140167236,
      "learning_rate": 1.3953703703703705e-05,
      "loss": 2.5495,
      "step": 2612
    },
    {
      "epoch": 0.30243055555555554,
      "grad_norm": 9.96354866027832,
      "learning_rate": 1.3951388888888891e-05,
      "loss": 2.7378,
      "step": 2613
    },
    {
      "epoch": 0.3025462962962963,
      "grad_norm": 5.615034103393555,
      "learning_rate": 1.3949074074074075e-05,
      "loss": 0.074,
      "step": 2614
    },
    {
      "epoch": 0.30266203703703703,
      "grad_norm": 0.3281939923763275,
      "learning_rate": 1.394675925925926e-05,
      "loss": 0.0541,
      "step": 2615
    },
    {
      "epoch": 0.30277777777777776,
      "grad_norm": 0.2802217900753021,
      "learning_rate": 1.3944444444444446e-05,
      "loss": 0.051,
      "step": 2616
    },
    {
      "epoch": 0.30289351851851853,
      "grad_norm": 0.20493504405021667,
      "learning_rate": 1.394212962962963e-05,
      "loss": 0.0358,
      "step": 2617
    },
    {
      "epoch": 0.30300925925925926,
      "grad_norm": 0.32843226194381714,
      "learning_rate": 1.3939814814814816e-05,
      "loss": 0.0403,
      "step": 2618
    },
    {
      "epoch": 0.303125,
      "grad_norm": 2.2247354984283447,
      "learning_rate": 1.3937500000000002e-05,
      "loss": 0.061,
      "step": 2619
    },
    {
      "epoch": 0.30324074074074076,
      "grad_norm": 0.21911251544952393,
      "learning_rate": 1.3935185185185188e-05,
      "loss": 0.0421,
      "step": 2620
    },
    {
      "epoch": 0.3033564814814815,
      "grad_norm": 0.3295201063156128,
      "learning_rate": 1.393287037037037e-05,
      "loss": 0.0386,
      "step": 2621
    },
    {
      "epoch": 0.3034722222222222,
      "grad_norm": 0.3706982135772705,
      "learning_rate": 1.3930555555555556e-05,
      "loss": 0.0405,
      "step": 2622
    },
    {
      "epoch": 0.303587962962963,
      "grad_norm": 0.1557956486940384,
      "learning_rate": 1.3928240740740742e-05,
      "loss": 0.0306,
      "step": 2623
    },
    {
      "epoch": 0.3037037037037037,
      "grad_norm": 0.31690990924835205,
      "learning_rate": 1.3925925925925928e-05,
      "loss": 0.0475,
      "step": 2624
    },
    {
      "epoch": 0.3038194444444444,
      "grad_norm": 0.17664484679698944,
      "learning_rate": 1.3923611111111112e-05,
      "loss": 0.0343,
      "step": 2625
    },
    {
      "epoch": 0.3039351851851852,
      "grad_norm": 0.24119363725185394,
      "learning_rate": 1.3921296296296298e-05,
      "loss": 0.0471,
      "step": 2626
    },
    {
      "epoch": 0.3040509259259259,
      "grad_norm": 66.53813934326172,
      "learning_rate": 1.3918981481481482e-05,
      "loss": 0.8904,
      "step": 2627
    },
    {
      "epoch": 0.30416666666666664,
      "grad_norm": 1.134227991104126,
      "learning_rate": 1.3916666666666667e-05,
      "loss": 0.047,
      "step": 2628
    },
    {
      "epoch": 0.3042824074074074,
      "grad_norm": 0.27001139521598816,
      "learning_rate": 1.3914351851851853e-05,
      "loss": 0.0534,
      "step": 2629
    },
    {
      "epoch": 0.30439814814814814,
      "grad_norm": 0.1639089733362198,
      "learning_rate": 1.3912037037037039e-05,
      "loss": 0.0316,
      "step": 2630
    },
    {
      "epoch": 0.30451388888888886,
      "grad_norm": 0.4111159145832062,
      "learning_rate": 1.3909722222222225e-05,
      "loss": 0.0529,
      "step": 2631
    },
    {
      "epoch": 0.30462962962962964,
      "grad_norm": 0.15694472193717957,
      "learning_rate": 1.3907407407407409e-05,
      "loss": 0.0307,
      "step": 2632
    },
    {
      "epoch": 0.30474537037037036,
      "grad_norm": 9.036261558532715,
      "learning_rate": 1.3905092592592593e-05,
      "loss": 0.0872,
      "step": 2633
    },
    {
      "epoch": 0.30486111111111114,
      "grad_norm": 127.74744415283203,
      "learning_rate": 1.3902777777777779e-05,
      "loss": 0.6113,
      "step": 2634
    },
    {
      "epoch": 0.30497685185185186,
      "grad_norm": 100.6592025756836,
      "learning_rate": 1.3900462962962963e-05,
      "loss": 0.9303,
      "step": 2635
    },
    {
      "epoch": 0.3050925925925926,
      "grad_norm": 0.2821598947048187,
      "learning_rate": 1.389814814814815e-05,
      "loss": 0.0556,
      "step": 2636
    },
    {
      "epoch": 0.30520833333333336,
      "grad_norm": 0.3380948603153229,
      "learning_rate": 1.3895833333333335e-05,
      "loss": 0.0578,
      "step": 2637
    },
    {
      "epoch": 0.3053240740740741,
      "grad_norm": 0.22624321281909943,
      "learning_rate": 1.3893518518518521e-05,
      "loss": 0.0302,
      "step": 2638
    },
    {
      "epoch": 0.3054398148148148,
      "grad_norm": 0.6176795959472656,
      "learning_rate": 1.3891203703703704e-05,
      "loss": 0.0617,
      "step": 2639
    },
    {
      "epoch": 0.3055555555555556,
      "grad_norm": 0.2312009334564209,
      "learning_rate": 1.388888888888889e-05,
      "loss": 0.0422,
      "step": 2640
    },
    {
      "epoch": 0.3056712962962963,
      "grad_norm": 0.16042166948318481,
      "learning_rate": 1.3886574074074075e-05,
      "loss": 0.0312,
      "step": 2641
    },
    {
      "epoch": 0.305787037037037,
      "grad_norm": 0.18090824782848358,
      "learning_rate": 1.3884259259259261e-05,
      "loss": 0.0346,
      "step": 2642
    },
    {
      "epoch": 0.3059027777777778,
      "grad_norm": 0.14804118871688843,
      "learning_rate": 1.3881944444444446e-05,
      "loss": 0.029,
      "step": 2643
    },
    {
      "epoch": 0.3060185185185185,
      "grad_norm": 0.21210038661956787,
      "learning_rate": 1.3879629629629632e-05,
      "loss": 0.0396,
      "step": 2644
    },
    {
      "epoch": 0.30613425925925924,
      "grad_norm": 0.22102510929107666,
      "learning_rate": 1.3877314814814816e-05,
      "loss": 0.0415,
      "step": 2645
    },
    {
      "epoch": 0.30625,
      "grad_norm": 0.1580776572227478,
      "learning_rate": 1.3875e-05,
      "loss": 0.0309,
      "step": 2646
    },
    {
      "epoch": 0.30636574074074074,
      "grad_norm": 0.21284040808677673,
      "learning_rate": 1.3872685185185186e-05,
      "loss": 0.0405,
      "step": 2647
    },
    {
      "epoch": 0.30648148148148147,
      "grad_norm": 81.21112823486328,
      "learning_rate": 1.3870370370370372e-05,
      "loss": 3.4562,
      "step": 2648
    },
    {
      "epoch": 0.30659722222222224,
      "grad_norm": 0.1573081612586975,
      "learning_rate": 1.3868055555555558e-05,
      "loss": 0.0308,
      "step": 2649
    },
    {
      "epoch": 0.30671296296296297,
      "grad_norm": 8.236332893371582,
      "learning_rate": 1.3865740740740742e-05,
      "loss": 2.7829,
      "step": 2650
    },
    {
      "epoch": 0.3068287037037037,
      "grad_norm": 1.0840156078338623,
      "learning_rate": 1.3863425925925926e-05,
      "loss": 0.0427,
      "step": 2651
    },
    {
      "epoch": 0.30694444444444446,
      "grad_norm": 0.18468840420246124,
      "learning_rate": 1.3861111111111112e-05,
      "loss": 0.036,
      "step": 2652
    },
    {
      "epoch": 0.3070601851851852,
      "grad_norm": 0.2160429209470749,
      "learning_rate": 1.3858796296296297e-05,
      "loss": 0.0411,
      "step": 2653
    },
    {
      "epoch": 0.3071759259259259,
      "grad_norm": 0.22827352583408356,
      "learning_rate": 1.3856481481481483e-05,
      "loss": 0.043,
      "step": 2654
    },
    {
      "epoch": 0.3072916666666667,
      "grad_norm": 1.4824373722076416,
      "learning_rate": 1.3854166666666669e-05,
      "loss": 0.0526,
      "step": 2655
    },
    {
      "epoch": 0.3074074074074074,
      "grad_norm": 0.2654293179512024,
      "learning_rate": 1.3851851851851851e-05,
      "loss": 0.0359,
      "step": 2656
    },
    {
      "epoch": 0.30752314814814813,
      "grad_norm": 0.5706108808517456,
      "learning_rate": 1.3849537037037037e-05,
      "loss": 0.0522,
      "step": 2657
    },
    {
      "epoch": 0.3076388888888889,
      "grad_norm": 12.276578903198242,
      "learning_rate": 1.3847222222222223e-05,
      "loss": 2.2538,
      "step": 2658
    },
    {
      "epoch": 0.30775462962962963,
      "grad_norm": 2.7688286304473877,
      "learning_rate": 1.3844907407407409e-05,
      "loss": 0.0695,
      "step": 2659
    },
    {
      "epoch": 0.30787037037037035,
      "grad_norm": 0.21639157831668854,
      "learning_rate": 1.3842592592592595e-05,
      "loss": 0.041,
      "step": 2660
    },
    {
      "epoch": 0.3079861111111111,
      "grad_norm": 0.2361118495464325,
      "learning_rate": 1.3840277777777779e-05,
      "loss": 0.0315,
      "step": 2661
    },
    {
      "epoch": 0.30810185185185185,
      "grad_norm": 0.30508503317832947,
      "learning_rate": 1.3837962962962963e-05,
      "loss": 0.0558,
      "step": 2662
    },
    {
      "epoch": 0.30821759259259257,
      "grad_norm": 0.24664001166820526,
      "learning_rate": 1.383564814814815e-05,
      "loss": 0.0484,
      "step": 2663
    },
    {
      "epoch": 0.30833333333333335,
      "grad_norm": 0.8047802448272705,
      "learning_rate": 1.3833333333333334e-05,
      "loss": 0.0363,
      "step": 2664
    },
    {
      "epoch": 0.30844907407407407,
      "grad_norm": 5.307309627532959,
      "learning_rate": 1.383101851851852e-05,
      "loss": 0.0739,
      "step": 2665
    },
    {
      "epoch": 0.3085648148148148,
      "grad_norm": 17.431093215942383,
      "learning_rate": 1.3828703703703705e-05,
      "loss": 2.2068,
      "step": 2666
    },
    {
      "epoch": 0.30868055555555557,
      "grad_norm": 0.2287624478340149,
      "learning_rate": 1.3826388888888891e-05,
      "loss": 0.0416,
      "step": 2667
    },
    {
      "epoch": 0.3087962962962963,
      "grad_norm": 0.25507214665412903,
      "learning_rate": 1.3824074074074074e-05,
      "loss": 0.0472,
      "step": 2668
    },
    {
      "epoch": 0.308912037037037,
      "grad_norm": 0.18782977759838104,
      "learning_rate": 1.382175925925926e-05,
      "loss": 0.0352,
      "step": 2669
    },
    {
      "epoch": 0.3090277777777778,
      "grad_norm": 18.31193733215332,
      "learning_rate": 1.3819444444444446e-05,
      "loss": 4.7581,
      "step": 2670
    },
    {
      "epoch": 0.3091435185185185,
      "grad_norm": 0.5883991122245789,
      "learning_rate": 1.381712962962963e-05,
      "loss": 0.0408,
      "step": 2671
    },
    {
      "epoch": 0.30925925925925923,
      "grad_norm": 114.62813568115234,
      "learning_rate": 1.3814814814814816e-05,
      "loss": 0.5132,
      "step": 2672
    },
    {
      "epoch": 0.309375,
      "grad_norm": 0.21157808601856232,
      "learning_rate": 1.3812500000000002e-05,
      "loss": 0.0378,
      "step": 2673
    },
    {
      "epoch": 0.30949074074074073,
      "grad_norm": 0.23275373876094818,
      "learning_rate": 1.3810185185185185e-05,
      "loss": 0.0422,
      "step": 2674
    },
    {
      "epoch": 0.30960648148148145,
      "grad_norm": 0.2193908840417862,
      "learning_rate": 1.380787037037037e-05,
      "loss": 0.0422,
      "step": 2675
    },
    {
      "epoch": 0.30972222222222223,
      "grad_norm": 0.24353386461734772,
      "learning_rate": 1.3805555555555556e-05,
      "loss": 0.0475,
      "step": 2676
    },
    {
      "epoch": 0.30983796296296295,
      "grad_norm": 0.6354426145553589,
      "learning_rate": 1.3803240740740742e-05,
      "loss": 0.0471,
      "step": 2677
    },
    {
      "epoch": 0.30995370370370373,
      "grad_norm": 0.18660613894462585,
      "learning_rate": 1.3800925925925928e-05,
      "loss": 0.0353,
      "step": 2678
    },
    {
      "epoch": 0.31006944444444445,
      "grad_norm": 0.20736664533615112,
      "learning_rate": 1.3798611111111113e-05,
      "loss": 0.0395,
      "step": 2679
    },
    {
      "epoch": 0.3101851851851852,
      "grad_norm": 0.30495956540107727,
      "learning_rate": 1.3796296296296297e-05,
      "loss": 0.0511,
      "step": 2680
    },
    {
      "epoch": 0.31030092592592595,
      "grad_norm": 0.23304784297943115,
      "learning_rate": 1.3793981481481483e-05,
      "loss": 0.0462,
      "step": 2681
    },
    {
      "epoch": 0.3104166666666667,
      "grad_norm": 44.21629333496094,
      "learning_rate": 1.3791666666666667e-05,
      "loss": 0.2822,
      "step": 2682
    },
    {
      "epoch": 0.3105324074074074,
      "grad_norm": 0.1535705029964447,
      "learning_rate": 1.3789351851851853e-05,
      "loss": 0.029,
      "step": 2683
    },
    {
      "epoch": 0.3106481481481482,
      "grad_norm": 0.259558767080307,
      "learning_rate": 1.3787037037037039e-05,
      "loss": 0.0512,
      "step": 2684
    },
    {
      "epoch": 0.3107638888888889,
      "grad_norm": 0.1587333381175995,
      "learning_rate": 1.3784722222222225e-05,
      "loss": 0.0308,
      "step": 2685
    },
    {
      "epoch": 0.3108796296296296,
      "grad_norm": 0.6954694390296936,
      "learning_rate": 1.3782407407407407e-05,
      "loss": 0.05,
      "step": 2686
    },
    {
      "epoch": 0.3109953703703704,
      "grad_norm": 0.24360395967960358,
      "learning_rate": 1.3780092592592593e-05,
      "loss": 0.0463,
      "step": 2687
    },
    {
      "epoch": 0.3111111111111111,
      "grad_norm": 0.19698721170425415,
      "learning_rate": 1.377777777777778e-05,
      "loss": 0.0368,
      "step": 2688
    },
    {
      "epoch": 0.31122685185185184,
      "grad_norm": 0.840555727481842,
      "learning_rate": 1.3775462962962964e-05,
      "loss": 0.0592,
      "step": 2689
    },
    {
      "epoch": 0.3113425925925926,
      "grad_norm": 0.1878400295972824,
      "learning_rate": 1.377314814814815e-05,
      "loss": 0.0359,
      "step": 2690
    },
    {
      "epoch": 0.31145833333333334,
      "grad_norm": 0.216413676738739,
      "learning_rate": 1.3770833333333335e-05,
      "loss": 0.0419,
      "step": 2691
    },
    {
      "epoch": 0.31157407407407406,
      "grad_norm": 7.488203525543213,
      "learning_rate": 1.3768518518518518e-05,
      "loss": 2.8252,
      "step": 2692
    },
    {
      "epoch": 0.31168981481481484,
      "grad_norm": 0.14880427718162537,
      "learning_rate": 1.3766203703703704e-05,
      "loss": 0.0288,
      "step": 2693
    },
    {
      "epoch": 0.31180555555555556,
      "grad_norm": 0.2829630672931671,
      "learning_rate": 1.376388888888889e-05,
      "loss": 0.0498,
      "step": 2694
    },
    {
      "epoch": 0.3119212962962963,
      "grad_norm": 0.24308520555496216,
      "learning_rate": 1.3761574074074076e-05,
      "loss": 0.043,
      "step": 2695
    },
    {
      "epoch": 0.31203703703703706,
      "grad_norm": 0.2618231177330017,
      "learning_rate": 1.3759259259259262e-05,
      "loss": 0.037,
      "step": 2696
    },
    {
      "epoch": 0.3121527777777778,
      "grad_norm": 0.2433064877986908,
      "learning_rate": 1.3756944444444446e-05,
      "loss": 0.0311,
      "step": 2697
    },
    {
      "epoch": 0.3122685185185185,
      "grad_norm": 0.17428775131702423,
      "learning_rate": 1.375462962962963e-05,
      "loss": 0.0288,
      "step": 2698
    },
    {
      "epoch": 0.3123842592592593,
      "grad_norm": 0.1857498288154602,
      "learning_rate": 1.3752314814814816e-05,
      "loss": 0.0348,
      "step": 2699
    },
    {
      "epoch": 0.3125,
      "grad_norm": 0.18677762150764465,
      "learning_rate": 1.375e-05,
      "loss": 0.0342,
      "step": 2700
    },
    {
      "epoch": 0.3126157407407407,
      "grad_norm": 0.23134204745292664,
      "learning_rate": 1.3747685185185186e-05,
      "loss": 0.0447,
      "step": 2701
    },
    {
      "epoch": 0.3127314814814815,
      "grad_norm": 0.7064937949180603,
      "learning_rate": 1.3745370370370372e-05,
      "loss": 0.0534,
      "step": 2702
    },
    {
      "epoch": 0.3128472222222222,
      "grad_norm": 0.2268686145544052,
      "learning_rate": 1.3743055555555555e-05,
      "loss": 0.0441,
      "step": 2703
    },
    {
      "epoch": 0.31296296296296294,
      "grad_norm": 0.1841631382703781,
      "learning_rate": 1.3740740740740741e-05,
      "loss": 0.0351,
      "step": 2704
    },
    {
      "epoch": 0.3130787037037037,
      "grad_norm": 0.21872851252555847,
      "learning_rate": 1.3738425925925927e-05,
      "loss": 0.0419,
      "step": 2705
    },
    {
      "epoch": 0.31319444444444444,
      "grad_norm": 0.145675927400589,
      "learning_rate": 1.3736111111111113e-05,
      "loss": 0.0281,
      "step": 2706
    },
    {
      "epoch": 0.31331018518518516,
      "grad_norm": 0.18336279690265656,
      "learning_rate": 1.3733796296296297e-05,
      "loss": 0.0352,
      "step": 2707
    },
    {
      "epoch": 0.31342592592592594,
      "grad_norm": 0.2596171200275421,
      "learning_rate": 1.3731481481481483e-05,
      "loss": 0.0517,
      "step": 2708
    },
    {
      "epoch": 0.31354166666666666,
      "grad_norm": 0.22726760804653168,
      "learning_rate": 1.3729166666666667e-05,
      "loss": 0.0414,
      "step": 2709
    },
    {
      "epoch": 0.3136574074074074,
      "grad_norm": 0.40410560369491577,
      "learning_rate": 1.3726851851851853e-05,
      "loss": 0.0462,
      "step": 2710
    },
    {
      "epoch": 0.31377314814814816,
      "grad_norm": 0.1857345998287201,
      "learning_rate": 1.3724537037037037e-05,
      "loss": 0.032,
      "step": 2711
    },
    {
      "epoch": 0.3138888888888889,
      "grad_norm": 0.4532744586467743,
      "learning_rate": 1.3722222222222223e-05,
      "loss": 0.0538,
      "step": 2712
    },
    {
      "epoch": 0.3140046296296296,
      "grad_norm": 0.2486659437417984,
      "learning_rate": 1.371990740740741e-05,
      "loss": 0.0492,
      "step": 2713
    },
    {
      "epoch": 0.3141203703703704,
      "grad_norm": 0.26213693618774414,
      "learning_rate": 1.3717592592592595e-05,
      "loss": 0.0509,
      "step": 2714
    },
    {
      "epoch": 0.3142361111111111,
      "grad_norm": 0.7072455286979675,
      "learning_rate": 1.3715277777777778e-05,
      "loss": 0.0518,
      "step": 2715
    },
    {
      "epoch": 0.3143518518518518,
      "grad_norm": 0.19067327678203583,
      "learning_rate": 1.3712962962962964e-05,
      "loss": 0.0268,
      "step": 2716
    },
    {
      "epoch": 0.3144675925925926,
      "grad_norm": 0.4747524857521057,
      "learning_rate": 1.371064814814815e-05,
      "loss": 0.0471,
      "step": 2717
    },
    {
      "epoch": 0.3145833333333333,
      "grad_norm": 0.23498977720737457,
      "learning_rate": 1.3708333333333334e-05,
      "loss": 0.0452,
      "step": 2718
    },
    {
      "epoch": 0.31469907407407405,
      "grad_norm": 18.367307662963867,
      "learning_rate": 1.370601851851852e-05,
      "loss": 0.0839,
      "step": 2719
    },
    {
      "epoch": 0.3148148148148148,
      "grad_norm": 128.72105407714844,
      "learning_rate": 1.3703703703703706e-05,
      "loss": 1.0181,
      "step": 2720
    },
    {
      "epoch": 0.31493055555555555,
      "grad_norm": 0.1773112714290619,
      "learning_rate": 1.3701388888888888e-05,
      "loss": 0.0336,
      "step": 2721
    },
    {
      "epoch": 0.31504629629629627,
      "grad_norm": 0.22536197304725647,
      "learning_rate": 1.3699074074074074e-05,
      "loss": 0.0441,
      "step": 2722
    },
    {
      "epoch": 0.31516203703703705,
      "grad_norm": 0.23706063628196716,
      "learning_rate": 1.369675925925926e-05,
      "loss": 0.0449,
      "step": 2723
    },
    {
      "epoch": 0.31527777777777777,
      "grad_norm": 0.42417824268341064,
      "learning_rate": 1.3694444444444446e-05,
      "loss": 0.0451,
      "step": 2724
    },
    {
      "epoch": 0.31539351851851855,
      "grad_norm": 0.20926985144615173,
      "learning_rate": 1.3692129629629632e-05,
      "loss": 0.0394,
      "step": 2725
    },
    {
      "epoch": 0.31550925925925927,
      "grad_norm": 5.0850043296813965,
      "learning_rate": 1.3689814814814816e-05,
      "loss": 0.054,
      "step": 2726
    },
    {
      "epoch": 0.315625,
      "grad_norm": 0.20578503608703613,
      "learning_rate": 1.36875e-05,
      "loss": 0.0387,
      "step": 2727
    },
    {
      "epoch": 0.31574074074074077,
      "grad_norm": 0.4626437723636627,
      "learning_rate": 1.3685185185185187e-05,
      "loss": 0.0443,
      "step": 2728
    },
    {
      "epoch": 0.3158564814814815,
      "grad_norm": 0.14495620131492615,
      "learning_rate": 1.3682870370370371e-05,
      "loss": 0.0279,
      "step": 2729
    },
    {
      "epoch": 0.3159722222222222,
      "grad_norm": 0.1861647665500641,
      "learning_rate": 1.3680555555555557e-05,
      "loss": 0.0352,
      "step": 2730
    },
    {
      "epoch": 0.316087962962963,
      "grad_norm": 0.1608540117740631,
      "learning_rate": 1.3678240740740743e-05,
      "loss": 0.0307,
      "step": 2731
    },
    {
      "epoch": 0.3162037037037037,
      "grad_norm": 0.2665579617023468,
      "learning_rate": 1.3675925925925929e-05,
      "loss": 0.0514,
      "step": 2732
    },
    {
      "epoch": 0.31631944444444443,
      "grad_norm": 0.1726182997226715,
      "learning_rate": 1.3673611111111111e-05,
      "loss": 0.0329,
      "step": 2733
    },
    {
      "epoch": 0.3164351851851852,
      "grad_norm": 0.18036334216594696,
      "learning_rate": 1.3671296296296297e-05,
      "loss": 0.0353,
      "step": 2734
    },
    {
      "epoch": 0.31655092592592593,
      "grad_norm": 0.24882228672504425,
      "learning_rate": 1.3668981481481483e-05,
      "loss": 0.0366,
      "step": 2735
    },
    {
      "epoch": 0.31666666666666665,
      "grad_norm": 0.2120024710893631,
      "learning_rate": 1.3666666666666667e-05,
      "loss": 0.0412,
      "step": 2736
    },
    {
      "epoch": 0.31678240740740743,
      "grad_norm": 0.29993754625320435,
      "learning_rate": 1.3664351851851853e-05,
      "loss": 0.0405,
      "step": 2737
    },
    {
      "epoch": 0.31689814814814815,
      "grad_norm": 100.50262451171875,
      "learning_rate": 1.366203703703704e-05,
      "loss": 1.44,
      "step": 2738
    },
    {
      "epoch": 0.3170138888888889,
      "grad_norm": 11.71804141998291,
      "learning_rate": 1.3659722222222222e-05,
      "loss": 2.2413,
      "step": 2739
    },
    {
      "epoch": 0.31712962962962965,
      "grad_norm": 0.15141330659389496,
      "learning_rate": 1.3657407407407408e-05,
      "loss": 0.0287,
      "step": 2740
    },
    {
      "epoch": 0.31724537037037037,
      "grad_norm": 72.021240234375,
      "learning_rate": 1.3655092592592594e-05,
      "loss": 0.4691,
      "step": 2741
    },
    {
      "epoch": 0.3173611111111111,
      "grad_norm": 1.2197861671447754,
      "learning_rate": 1.365277777777778e-05,
      "loss": 0.0493,
      "step": 2742
    },
    {
      "epoch": 0.31747685185185187,
      "grad_norm": 0.22315102815628052,
      "learning_rate": 1.3650462962962966e-05,
      "loss": 0.0398,
      "step": 2743
    },
    {
      "epoch": 0.3175925925925926,
      "grad_norm": 0.19473110139369965,
      "learning_rate": 1.364814814814815e-05,
      "loss": 0.0378,
      "step": 2744
    },
    {
      "epoch": 0.3177083333333333,
      "grad_norm": 0.20980174839496613,
      "learning_rate": 1.3645833333333334e-05,
      "loss": 0.0405,
      "step": 2745
    },
    {
      "epoch": 0.3178240740740741,
      "grad_norm": 0.20314261317253113,
      "learning_rate": 1.364351851851852e-05,
      "loss": 0.0375,
      "step": 2746
    },
    {
      "epoch": 0.3179398148148148,
      "grad_norm": 0.39745020866394043,
      "learning_rate": 1.3641203703703704e-05,
      "loss": 0.039,
      "step": 2747
    },
    {
      "epoch": 0.31805555555555554,
      "grad_norm": 1.0723345279693604,
      "learning_rate": 1.363888888888889e-05,
      "loss": 0.0429,
      "step": 2748
    },
    {
      "epoch": 0.3181712962962963,
      "grad_norm": 0.33473071455955505,
      "learning_rate": 1.3636574074074076e-05,
      "loss": 0.0335,
      "step": 2749
    },
    {
      "epoch": 0.31828703703703703,
      "grad_norm": 0.19416433572769165,
      "learning_rate": 1.3634259259259259e-05,
      "loss": 0.0364,
      "step": 2750
    },
    {
      "epoch": 0.31840277777777776,
      "grad_norm": 0.22225971519947052,
      "learning_rate": 1.3631944444444445e-05,
      "loss": 0.0433,
      "step": 2751
    },
    {
      "epoch": 0.31851851851851853,
      "grad_norm": 5.663860321044922,
      "learning_rate": 1.362962962962963e-05,
      "loss": 2.7259,
      "step": 2752
    },
    {
      "epoch": 0.31863425925925926,
      "grad_norm": 0.3642134368419647,
      "learning_rate": 1.3627314814814817e-05,
      "loss": 0.0505,
      "step": 2753
    },
    {
      "epoch": 0.31875,
      "grad_norm": 0.20593011379241943,
      "learning_rate": 1.3625e-05,
      "loss": 0.0392,
      "step": 2754
    },
    {
      "epoch": 0.31886574074074076,
      "grad_norm": 0.1588936746120453,
      "learning_rate": 1.3622685185185187e-05,
      "loss": 0.0302,
      "step": 2755
    },
    {
      "epoch": 0.3189814814814815,
      "grad_norm": 31.512174606323242,
      "learning_rate": 1.3620370370370371e-05,
      "loss": 1.6372,
      "step": 2756
    },
    {
      "epoch": 0.3190972222222222,
      "grad_norm": 0.21948783099651337,
      "learning_rate": 1.3618055555555555e-05,
      "loss": 0.0433,
      "step": 2757
    },
    {
      "epoch": 0.319212962962963,
      "grad_norm": 0.8021889328956604,
      "learning_rate": 1.3615740740740741e-05,
      "loss": 0.0415,
      "step": 2758
    },
    {
      "epoch": 0.3193287037037037,
      "grad_norm": 0.3463194668292999,
      "learning_rate": 1.3613425925925927e-05,
      "loss": 0.0498,
      "step": 2759
    },
    {
      "epoch": 0.3194444444444444,
      "grad_norm": 0.1790013164281845,
      "learning_rate": 1.3611111111111113e-05,
      "loss": 0.0334,
      "step": 2760
    },
    {
      "epoch": 0.3195601851851852,
      "grad_norm": 0.19173066318035126,
      "learning_rate": 1.3608796296296299e-05,
      "loss": 0.0375,
      "step": 2761
    },
    {
      "epoch": 0.3196759259259259,
      "grad_norm": 0.3603512644767761,
      "learning_rate": 1.3606481481481482e-05,
      "loss": 0.0513,
      "step": 2762
    },
    {
      "epoch": 0.31979166666666664,
      "grad_norm": 0.2814893126487732,
      "learning_rate": 1.3604166666666668e-05,
      "loss": 0.0396,
      "step": 2763
    },
    {
      "epoch": 0.3199074074074074,
      "grad_norm": 0.19215185940265656,
      "learning_rate": 1.3601851851851854e-05,
      "loss": 0.0367,
      "step": 2764
    },
    {
      "epoch": 0.32002314814814814,
      "grad_norm": 2.518784523010254,
      "learning_rate": 1.3599537037037038e-05,
      "loss": 0.0686,
      "step": 2765
    },
    {
      "epoch": 0.32013888888888886,
      "grad_norm": 0.1972881406545639,
      "learning_rate": 1.3597222222222224e-05,
      "loss": 0.0376,
      "step": 2766
    },
    {
      "epoch": 0.32025462962962964,
      "grad_norm": 0.24897009134292603,
      "learning_rate": 1.359490740740741e-05,
      "loss": 0.0464,
      "step": 2767
    },
    {
      "epoch": 0.32037037037037036,
      "grad_norm": 0.20843961834907532,
      "learning_rate": 1.3592592592592592e-05,
      "loss": 0.0392,
      "step": 2768
    },
    {
      "epoch": 0.32048611111111114,
      "grad_norm": 0.1691795438528061,
      "learning_rate": 1.3590277777777778e-05,
      "loss": 0.0327,
      "step": 2769
    },
    {
      "epoch": 0.32060185185185186,
      "grad_norm": 0.16574624180793762,
      "learning_rate": 1.3587962962962964e-05,
      "loss": 0.0304,
      "step": 2770
    },
    {
      "epoch": 0.3207175925925926,
      "grad_norm": 51.59806442260742,
      "learning_rate": 1.358564814814815e-05,
      "loss": 2.1469,
      "step": 2771
    },
    {
      "epoch": 0.32083333333333336,
      "grad_norm": 12.934407234191895,
      "learning_rate": 1.3583333333333334e-05,
      "loss": 2.3879,
      "step": 2772
    },
    {
      "epoch": 0.3209490740740741,
      "grad_norm": 0.22593571245670319,
      "learning_rate": 1.358101851851852e-05,
      "loss": 0.0443,
      "step": 2773
    },
    {
      "epoch": 0.3210648148148148,
      "grad_norm": 0.14057175815105438,
      "learning_rate": 1.3578703703703704e-05,
      "loss": 0.0272,
      "step": 2774
    },
    {
      "epoch": 0.3211805555555556,
      "grad_norm": 0.47051697969436646,
      "learning_rate": 1.3576388888888889e-05,
      "loss": 0.053,
      "step": 2775
    },
    {
      "epoch": 0.3212962962962963,
      "grad_norm": 14.793293952941895,
      "learning_rate": 1.3574074074074075e-05,
      "loss": 2.3935,
      "step": 2776
    },
    {
      "epoch": 0.321412037037037,
      "grad_norm": 0.2410878688097,
      "learning_rate": 1.357175925925926e-05,
      "loss": 0.0475,
      "step": 2777
    },
    {
      "epoch": 0.3215277777777778,
      "grad_norm": 102.62535858154297,
      "learning_rate": 1.3569444444444447e-05,
      "loss": 1.0666,
      "step": 2778
    },
    {
      "epoch": 0.3216435185185185,
      "grad_norm": 0.23183345794677734,
      "learning_rate": 1.3567129629629633e-05,
      "loss": 0.0291,
      "step": 2779
    },
    {
      "epoch": 0.32175925925925924,
      "grad_norm": 0.18583698570728302,
      "learning_rate": 1.3564814814814815e-05,
      "loss": 0.0347,
      "step": 2780
    },
    {
      "epoch": 0.321875,
      "grad_norm": 0.1948646605014801,
      "learning_rate": 1.3562500000000001e-05,
      "loss": 0.0274,
      "step": 2781
    },
    {
      "epoch": 0.32199074074074074,
      "grad_norm": 15.587056159973145,
      "learning_rate": 1.3560185185185187e-05,
      "loss": 2.4942,
      "step": 2782
    },
    {
      "epoch": 0.32210648148148147,
      "grad_norm": 0.21138250827789307,
      "learning_rate": 1.3557870370370371e-05,
      "loss": 0.0394,
      "step": 2783
    },
    {
      "epoch": 0.32222222222222224,
      "grad_norm": 0.1984666883945465,
      "learning_rate": 1.3555555555555557e-05,
      "loss": 0.0375,
      "step": 2784
    },
    {
      "epoch": 0.32233796296296297,
      "grad_norm": 0.19534820318222046,
      "learning_rate": 1.3553240740740743e-05,
      "loss": 0.0369,
      "step": 2785
    },
    {
      "epoch": 0.3224537037037037,
      "grad_norm": 112.81346893310547,
      "learning_rate": 1.3550925925925926e-05,
      "loss": 0.5934,
      "step": 2786
    },
    {
      "epoch": 0.32256944444444446,
      "grad_norm": 0.25889915227890015,
      "learning_rate": 1.3548611111111112e-05,
      "loss": 0.0435,
      "step": 2787
    },
    {
      "epoch": 0.3226851851851852,
      "grad_norm": 0.14968958497047424,
      "learning_rate": 1.3546296296296298e-05,
      "loss": 0.029,
      "step": 2788
    },
    {
      "epoch": 0.3228009259259259,
      "grad_norm": 8.307182312011719,
      "learning_rate": 1.3543981481481483e-05,
      "loss": 2.4741,
      "step": 2789
    },
    {
      "epoch": 0.3229166666666667,
      "grad_norm": 0.1557178795337677,
      "learning_rate": 1.3541666666666668e-05,
      "loss": 0.0296,
      "step": 2790
    },
    {
      "epoch": 0.3230324074074074,
      "grad_norm": 0.7586452960968018,
      "learning_rate": 1.3539351851851852e-05,
      "loss": 0.0518,
      "step": 2791
    },
    {
      "epoch": 0.32314814814814813,
      "grad_norm": 0.2009320855140686,
      "learning_rate": 1.3537037037037038e-05,
      "loss": 0.0376,
      "step": 2792
    },
    {
      "epoch": 0.3232638888888889,
      "grad_norm": 0.22137781977653503,
      "learning_rate": 1.3534722222222222e-05,
      "loss": 0.0429,
      "step": 2793
    },
    {
      "epoch": 0.32337962962962963,
      "grad_norm": 0.20889875292778015,
      "learning_rate": 1.3532407407407408e-05,
      "loss": 0.0385,
      "step": 2794
    },
    {
      "epoch": 0.32349537037037035,
      "grad_norm": 0.14964772760868073,
      "learning_rate": 1.3530092592592594e-05,
      "loss": 0.0291,
      "step": 2795
    },
    {
      "epoch": 0.3236111111111111,
      "grad_norm": 56.183204650878906,
      "learning_rate": 1.352777777777778e-05,
      "loss": 0.3386,
      "step": 2796
    },
    {
      "epoch": 0.32372685185185185,
      "grad_norm": 0.16212214529514313,
      "learning_rate": 1.3525462962962963e-05,
      "loss": 0.0314,
      "step": 2797
    },
    {
      "epoch": 0.32384259259259257,
      "grad_norm": 0.19798503816127777,
      "learning_rate": 1.3523148148148149e-05,
      "loss": 0.038,
      "step": 2798
    },
    {
      "epoch": 0.32395833333333335,
      "grad_norm": 0.2155510038137436,
      "learning_rate": 1.3520833333333334e-05,
      "loss": 0.0411,
      "step": 2799
    },
    {
      "epoch": 0.32407407407407407,
      "grad_norm": 1.0312598943710327,
      "learning_rate": 1.351851851851852e-05,
      "loss": 0.0527,
      "step": 2800
    },
    {
      "epoch": 0.3241898148148148,
      "grad_norm": 0.19083184003829956,
      "learning_rate": 1.3516203703703705e-05,
      "loss": 0.0341,
      "step": 2801
    },
    {
      "epoch": 0.32430555555555557,
      "grad_norm": 0.431916207075119,
      "learning_rate": 1.351388888888889e-05,
      "loss": 0.0463,
      "step": 2802
    },
    {
      "epoch": 0.3244212962962963,
      "grad_norm": 0.38471677899360657,
      "learning_rate": 1.3511574074074075e-05,
      "loss": 0.0337,
      "step": 2803
    },
    {
      "epoch": 0.324537037037037,
      "grad_norm": 0.20201781392097473,
      "learning_rate": 1.3509259259259259e-05,
      "loss": 0.039,
      "step": 2804
    },
    {
      "epoch": 0.3246527777777778,
      "grad_norm": 0.17487205564975739,
      "learning_rate": 1.3506944444444445e-05,
      "loss": 0.0334,
      "step": 2805
    },
    {
      "epoch": 0.3247685185185185,
      "grad_norm": 0.23174597322940826,
      "learning_rate": 1.3504629629629631e-05,
      "loss": 0.0446,
      "step": 2806
    },
    {
      "epoch": 0.32488425925925923,
      "grad_norm": 0.14463499188423157,
      "learning_rate": 1.3502314814814817e-05,
      "loss": 0.0273,
      "step": 2807
    },
    {
      "epoch": 0.325,
      "grad_norm": 0.24931040406227112,
      "learning_rate": 1.3500000000000001e-05,
      "loss": 0.0355,
      "step": 2808
    },
    {
      "epoch": 0.32511574074074073,
      "grad_norm": 0.7449128031730652,
      "learning_rate": 1.3497685185185185e-05,
      "loss": 0.0453,
      "step": 2809
    },
    {
      "epoch": 0.32523148148148145,
      "grad_norm": 0.21280483901500702,
      "learning_rate": 1.3495370370370371e-05,
      "loss": 0.0413,
      "step": 2810
    },
    {
      "epoch": 0.32534722222222223,
      "grad_norm": 0.1984485536813736,
      "learning_rate": 1.3493055555555556e-05,
      "loss": 0.0373,
      "step": 2811
    },
    {
      "epoch": 0.32546296296296295,
      "grad_norm": 0.13573291897773743,
      "learning_rate": 1.3490740740740742e-05,
      "loss": 0.0262,
      "step": 2812
    },
    {
      "epoch": 0.32557870370370373,
      "grad_norm": 32.917972564697266,
      "learning_rate": 1.3488425925925928e-05,
      "loss": 2.0796,
      "step": 2813
    },
    {
      "epoch": 0.32569444444444445,
      "grad_norm": 0.1831064373254776,
      "learning_rate": 1.3486111111111113e-05,
      "loss": 0.0256,
      "step": 2814
    },
    {
      "epoch": 0.3258101851851852,
      "grad_norm": 0.22872324287891388,
      "learning_rate": 1.3483796296296296e-05,
      "loss": 0.0432,
      "step": 2815
    },
    {
      "epoch": 0.32592592592592595,
      "grad_norm": 0.19068492949008942,
      "learning_rate": 1.3481481481481482e-05,
      "loss": 0.0263,
      "step": 2816
    },
    {
      "epoch": 0.3260416666666667,
      "grad_norm": 0.17056497931480408,
      "learning_rate": 1.3479166666666668e-05,
      "loss": 0.0329,
      "step": 2817
    },
    {
      "epoch": 0.3261574074074074,
      "grad_norm": 0.21821478009223938,
      "learning_rate": 1.3476851851851854e-05,
      "loss": 0.0385,
      "step": 2818
    },
    {
      "epoch": 0.3262731481481482,
      "grad_norm": 0.28002768754959106,
      "learning_rate": 1.3474537037037038e-05,
      "loss": 0.0413,
      "step": 2819
    },
    {
      "epoch": 0.3263888888888889,
      "grad_norm": 11.638683319091797,
      "learning_rate": 1.3472222222222224e-05,
      "loss": 0.14,
      "step": 2820
    },
    {
      "epoch": 0.3265046296296296,
      "grad_norm": 7.142297744750977,
      "learning_rate": 1.3469907407407408e-05,
      "loss": 2.5281,
      "step": 2821
    },
    {
      "epoch": 0.3266203703703704,
      "grad_norm": 6.260796546936035,
      "learning_rate": 1.3467592592592593e-05,
      "loss": 0.0825,
      "step": 2822
    },
    {
      "epoch": 0.3267361111111111,
      "grad_norm": 0.3399175703525543,
      "learning_rate": 1.3465277777777778e-05,
      "loss": 0.0474,
      "step": 2823
    },
    {
      "epoch": 0.32685185185185184,
      "grad_norm": 129.81105041503906,
      "learning_rate": 1.3462962962962964e-05,
      "loss": 0.7965,
      "step": 2824
    },
    {
      "epoch": 0.3269675925925926,
      "grad_norm": 0.1706724464893341,
      "learning_rate": 1.346064814814815e-05,
      "loss": 0.0323,
      "step": 2825
    },
    {
      "epoch": 0.32708333333333334,
      "grad_norm": 0.23756149411201477,
      "learning_rate": 1.3458333333333335e-05,
      "loss": 0.0334,
      "step": 2826
    },
    {
      "epoch": 0.32719907407407406,
      "grad_norm": 0.9743481278419495,
      "learning_rate": 1.3456018518518519e-05,
      "loss": 0.0434,
      "step": 2827
    },
    {
      "epoch": 0.32731481481481484,
      "grad_norm": 0.21802733838558197,
      "learning_rate": 1.3453703703703705e-05,
      "loss": 0.0304,
      "step": 2828
    },
    {
      "epoch": 0.32743055555555556,
      "grad_norm": 36.78004455566406,
      "learning_rate": 1.345138888888889e-05,
      "loss": 2.3117,
      "step": 2829
    },
    {
      "epoch": 0.3275462962962963,
      "grad_norm": 0.24378329515457153,
      "learning_rate": 1.3449074074074075e-05,
      "loss": 0.0356,
      "step": 2830
    },
    {
      "epoch": 0.32766203703703706,
      "grad_norm": 1.3217073678970337,
      "learning_rate": 1.3446759259259261e-05,
      "loss": 0.0403,
      "step": 2831
    },
    {
      "epoch": 0.3277777777777778,
      "grad_norm": 0.20207837224006653,
      "learning_rate": 1.3444444444444447e-05,
      "loss": 0.0372,
      "step": 2832
    },
    {
      "epoch": 0.3278935185185185,
      "grad_norm": 1.7192356586456299,
      "learning_rate": 1.344212962962963e-05,
      "loss": 0.039,
      "step": 2833
    },
    {
      "epoch": 0.3280092592592593,
      "grad_norm": 0.2016156017780304,
      "learning_rate": 1.3439814814814815e-05,
      "loss": 0.0377,
      "step": 2834
    },
    {
      "epoch": 0.328125,
      "grad_norm": 0.8604588508605957,
      "learning_rate": 1.3437500000000001e-05,
      "loss": 0.0464,
      "step": 2835
    },
    {
      "epoch": 0.3282407407407407,
      "grad_norm": 0.15332339704036713,
      "learning_rate": 1.3435185185185187e-05,
      "loss": 0.029,
      "step": 2836
    },
    {
      "epoch": 0.3283564814814815,
      "grad_norm": 11.200968742370605,
      "learning_rate": 1.3432870370370372e-05,
      "loss": 0.1548,
      "step": 2837
    },
    {
      "epoch": 0.3284722222222222,
      "grad_norm": 0.17726047337055206,
      "learning_rate": 1.3430555555555556e-05,
      "loss": 0.0341,
      "step": 2838
    },
    {
      "epoch": 0.32858796296296294,
      "grad_norm": 0.17182981967926025,
      "learning_rate": 1.3428240740740742e-05,
      "loss": 0.0322,
      "step": 2839
    },
    {
      "epoch": 0.3287037037037037,
      "grad_norm": 0.13148561120033264,
      "learning_rate": 1.3425925925925926e-05,
      "loss": 0.0256,
      "step": 2840
    },
    {
      "epoch": 0.32881944444444444,
      "grad_norm": 0.1589183360338211,
      "learning_rate": 1.3423611111111112e-05,
      "loss": 0.0307,
      "step": 2841
    },
    {
      "epoch": 0.32893518518518516,
      "grad_norm": 0.16101720929145813,
      "learning_rate": 1.3421296296296298e-05,
      "loss": 0.0305,
      "step": 2842
    },
    {
      "epoch": 0.32905092592592594,
      "grad_norm": 1.8385968208312988,
      "learning_rate": 1.3418981481481484e-05,
      "loss": 0.0541,
      "step": 2843
    },
    {
      "epoch": 0.32916666666666666,
      "grad_norm": 0.21105821430683136,
      "learning_rate": 1.3416666666666666e-05,
      "loss": 0.0403,
      "step": 2844
    },
    {
      "epoch": 0.3292824074074074,
      "grad_norm": 0.19425652921199799,
      "learning_rate": 1.3414351851851852e-05,
      "loss": 0.0334,
      "step": 2845
    },
    {
      "epoch": 0.32939814814814816,
      "grad_norm": 0.20063196122646332,
      "learning_rate": 1.3412037037037038e-05,
      "loss": 0.0274,
      "step": 2846
    },
    {
      "epoch": 0.3295138888888889,
      "grad_norm": 0.19752690196037292,
      "learning_rate": 1.3409722222222224e-05,
      "loss": 0.0373,
      "step": 2847
    },
    {
      "epoch": 0.3296296296296296,
      "grad_norm": 0.727979838848114,
      "learning_rate": 1.3407407407407408e-05,
      "loss": 0.04,
      "step": 2848
    },
    {
      "epoch": 0.3297453703703704,
      "grad_norm": 0.2513992190361023,
      "learning_rate": 1.3405092592592594e-05,
      "loss": 0.0478,
      "step": 2849
    },
    {
      "epoch": 0.3298611111111111,
      "grad_norm": 0.17615906894207,
      "learning_rate": 1.3402777777777779e-05,
      "loss": 0.0331,
      "step": 2850
    },
    {
      "epoch": 0.3299768518518518,
      "grad_norm": 14.874184608459473,
      "learning_rate": 1.3400462962962963e-05,
      "loss": 2.3319,
      "step": 2851
    },
    {
      "epoch": 0.3300925925925926,
      "grad_norm": 4.4967041015625,
      "learning_rate": 1.3398148148148149e-05,
      "loss": 0.061,
      "step": 2852
    },
    {
      "epoch": 0.3302083333333333,
      "grad_norm": 0.16257867217063904,
      "learning_rate": 1.3395833333333335e-05,
      "loss": 0.0316,
      "step": 2853
    },
    {
      "epoch": 0.33032407407407405,
      "grad_norm": 0.2620285451412201,
      "learning_rate": 1.339351851851852e-05,
      "loss": 0.043,
      "step": 2854
    },
    {
      "epoch": 0.3304398148148148,
      "grad_norm": 0.23932021856307983,
      "learning_rate": 1.3391203703703705e-05,
      "loss": 0.0411,
      "step": 2855
    },
    {
      "epoch": 0.33055555555555555,
      "grad_norm": 0.16509023308753967,
      "learning_rate": 1.338888888888889e-05,
      "loss": 0.0318,
      "step": 2856
    },
    {
      "epoch": 0.33067129629629627,
      "grad_norm": 0.19137731194496155,
      "learning_rate": 1.3386574074074075e-05,
      "loss": 0.0353,
      "step": 2857
    },
    {
      "epoch": 0.33078703703703705,
      "grad_norm": 0.20739708840847015,
      "learning_rate": 1.338425925925926e-05,
      "loss": 0.0369,
      "step": 2858
    },
    {
      "epoch": 0.33090277777777777,
      "grad_norm": 0.2957991361618042,
      "learning_rate": 1.3381944444444445e-05,
      "loss": 0.0397,
      "step": 2859
    },
    {
      "epoch": 0.33101851851851855,
      "grad_norm": 0.18946391344070435,
      "learning_rate": 1.3379629629629631e-05,
      "loss": 0.0356,
      "step": 2860
    },
    {
      "epoch": 0.33113425925925927,
      "grad_norm": 4.267814636230469,
      "learning_rate": 1.3377314814814817e-05,
      "loss": 0.0501,
      "step": 2861
    },
    {
      "epoch": 0.33125,
      "grad_norm": 0.15381604433059692,
      "learning_rate": 1.3375e-05,
      "loss": 0.0288,
      "step": 2862
    },
    {
      "epoch": 0.33136574074074077,
      "grad_norm": 0.13250720500946045,
      "learning_rate": 1.3372685185185186e-05,
      "loss": 0.0257,
      "step": 2863
    },
    {
      "epoch": 0.3314814814814815,
      "grad_norm": 0.15133099257946014,
      "learning_rate": 1.3370370370370372e-05,
      "loss": 0.0281,
      "step": 2864
    },
    {
      "epoch": 0.3315972222222222,
      "grad_norm": 0.16547484695911407,
      "learning_rate": 1.3368055555555558e-05,
      "loss": 0.0314,
      "step": 2865
    },
    {
      "epoch": 0.331712962962963,
      "grad_norm": 0.3518719971179962,
      "learning_rate": 1.3365740740740742e-05,
      "loss": 0.0371,
      "step": 2866
    },
    {
      "epoch": 0.3318287037037037,
      "grad_norm": 0.23011396825313568,
      "learning_rate": 1.3363425925925928e-05,
      "loss": 0.0371,
      "step": 2867
    },
    {
      "epoch": 0.33194444444444443,
      "grad_norm": 0.4370724856853485,
      "learning_rate": 1.3361111111111112e-05,
      "loss": 0.0397,
      "step": 2868
    },
    {
      "epoch": 0.3320601851851852,
      "grad_norm": 46.28150939941406,
      "learning_rate": 1.3358796296296296e-05,
      "loss": 0.173,
      "step": 2869
    },
    {
      "epoch": 0.33217592592592593,
      "grad_norm": 0.12847697734832764,
      "learning_rate": 1.3356481481481482e-05,
      "loss": 0.025,
      "step": 2870
    },
    {
      "epoch": 0.33229166666666665,
      "grad_norm": 0.140788733959198,
      "learning_rate": 1.3354166666666668e-05,
      "loss": 0.027,
      "step": 2871
    },
    {
      "epoch": 0.33240740740740743,
      "grad_norm": 0.19951611757278442,
      "learning_rate": 1.3351851851851854e-05,
      "loss": 0.0365,
      "step": 2872
    },
    {
      "epoch": 0.33252314814814815,
      "grad_norm": 0.19592821598052979,
      "learning_rate": 1.3349537037037038e-05,
      "loss": 0.036,
      "step": 2873
    },
    {
      "epoch": 0.3326388888888889,
      "grad_norm": 0.21018409729003906,
      "learning_rate": 1.3347222222222223e-05,
      "loss": 0.0356,
      "step": 2874
    },
    {
      "epoch": 0.33275462962962965,
      "grad_norm": 0.7562466859817505,
      "learning_rate": 1.3344907407407409e-05,
      "loss": 0.0402,
      "step": 2875
    },
    {
      "epoch": 0.33287037037037037,
      "grad_norm": 18.0983943939209,
      "learning_rate": 1.3342592592592593e-05,
      "loss": 0.1045,
      "step": 2876
    },
    {
      "epoch": 0.3329861111111111,
      "grad_norm": 0.17699620127677917,
      "learning_rate": 1.3340277777777779e-05,
      "loss": 0.0338,
      "step": 2877
    },
    {
      "epoch": 0.33310185185185187,
      "grad_norm": 0.2056027352809906,
      "learning_rate": 1.3337962962962965e-05,
      "loss": 0.0404,
      "step": 2878
    },
    {
      "epoch": 0.3332175925925926,
      "grad_norm": 0.19254080951213837,
      "learning_rate": 1.333564814814815e-05,
      "loss": 0.0372,
      "step": 2879
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.1414678990840912,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.0265,
      "step": 2880
    },
    {
      "epoch": 0.3334490740740741,
      "grad_norm": 0.17458967864513397,
      "learning_rate": 1.333101851851852e-05,
      "loss": 0.0243,
      "step": 2881
    },
    {
      "epoch": 0.3335648148148148,
      "grad_norm": 0.2932329773902893,
      "learning_rate": 1.3328703703703705e-05,
      "loss": 0.0355,
      "step": 2882
    },
    {
      "epoch": 0.33368055555555554,
      "grad_norm": 27.117238998413086,
      "learning_rate": 1.3326388888888891e-05,
      "loss": 2.0816,
      "step": 2883
    },
    {
      "epoch": 0.3337962962962963,
      "grad_norm": 0.21088597178459167,
      "learning_rate": 1.3324074074074075e-05,
      "loss": 0.0414,
      "step": 2884
    },
    {
      "epoch": 0.33391203703703703,
      "grad_norm": 0.35960254073143005,
      "learning_rate": 1.332175925925926e-05,
      "loss": 0.0398,
      "step": 2885
    },
    {
      "epoch": 0.33402777777777776,
      "grad_norm": 0.29119354486465454,
      "learning_rate": 1.3319444444444446e-05,
      "loss": 0.0442,
      "step": 2886
    },
    {
      "epoch": 0.33414351851851853,
      "grad_norm": 0.19299529492855072,
      "learning_rate": 1.331712962962963e-05,
      "loss": 0.0264,
      "step": 2887
    },
    {
      "epoch": 0.33425925925925926,
      "grad_norm": 0.2110227346420288,
      "learning_rate": 1.3314814814814816e-05,
      "loss": 0.0414,
      "step": 2888
    },
    {
      "epoch": 0.334375,
      "grad_norm": 0.23921199142932892,
      "learning_rate": 1.3312500000000002e-05,
      "loss": 0.034,
      "step": 2889
    },
    {
      "epoch": 0.33449074074074076,
      "grad_norm": 0.18993884325027466,
      "learning_rate": 1.3310185185185188e-05,
      "loss": 0.0313,
      "step": 2890
    },
    {
      "epoch": 0.3346064814814815,
      "grad_norm": 0.13600005209445953,
      "learning_rate": 1.330787037037037e-05,
      "loss": 0.0264,
      "step": 2891
    },
    {
      "epoch": 0.3347222222222222,
      "grad_norm": 178.11581420898438,
      "learning_rate": 1.3305555555555556e-05,
      "loss": 0.4857,
      "step": 2892
    },
    {
      "epoch": 0.334837962962963,
      "grad_norm": 0.3650864064693451,
      "learning_rate": 1.3303240740740742e-05,
      "loss": 0.0485,
      "step": 2893
    },
    {
      "epoch": 0.3349537037037037,
      "grad_norm": 0.2072441577911377,
      "learning_rate": 1.3300925925925926e-05,
      "loss": 0.0394,
      "step": 2894
    },
    {
      "epoch": 0.3350694444444444,
      "grad_norm": 1.0248384475708008,
      "learning_rate": 1.3298611111111112e-05,
      "loss": 0.0508,
      "step": 2895
    },
    {
      "epoch": 0.3351851851851852,
      "grad_norm": 0.2001778483390808,
      "learning_rate": 1.3296296296296298e-05,
      "loss": 0.0378,
      "step": 2896
    },
    {
      "epoch": 0.3353009259259259,
      "grad_norm": 0.16686461865901947,
      "learning_rate": 1.329398148148148e-05,
      "loss": 0.0314,
      "step": 2897
    },
    {
      "epoch": 0.33541666666666664,
      "grad_norm": 1.225232720375061,
      "learning_rate": 1.3291666666666667e-05,
      "loss": 0.0313,
      "step": 2898
    },
    {
      "epoch": 0.3355324074074074,
      "grad_norm": 2.6426587104797363,
      "learning_rate": 1.3289351851851853e-05,
      "loss": 0.0613,
      "step": 2899
    },
    {
      "epoch": 0.33564814814814814,
      "grad_norm": 0.25887659192085266,
      "learning_rate": 1.3287037037037039e-05,
      "loss": 0.0365,
      "step": 2900
    },
    {
      "epoch": 0.33576388888888886,
      "grad_norm": 0.19311191141605377,
      "learning_rate": 1.3284722222222225e-05,
      "loss": 0.0356,
      "step": 2901
    },
    {
      "epoch": 0.33587962962962964,
      "grad_norm": 0.20710919797420502,
      "learning_rate": 1.3282407407407409e-05,
      "loss": 0.0355,
      "step": 2902
    },
    {
      "epoch": 0.33599537037037036,
      "grad_norm": 0.18408159911632538,
      "learning_rate": 1.3280092592592593e-05,
      "loss": 0.0348,
      "step": 2903
    },
    {
      "epoch": 0.33611111111111114,
      "grad_norm": 0.17200322449207306,
      "learning_rate": 1.3277777777777779e-05,
      "loss": 0.0322,
      "step": 2904
    },
    {
      "epoch": 0.33622685185185186,
      "grad_norm": 0.1674293875694275,
      "learning_rate": 1.3275462962962963e-05,
      "loss": 0.0315,
      "step": 2905
    },
    {
      "epoch": 0.3363425925925926,
      "grad_norm": 0.15924544632434845,
      "learning_rate": 1.327314814814815e-05,
      "loss": 0.0306,
      "step": 2906
    },
    {
      "epoch": 0.33645833333333336,
      "grad_norm": 0.3377203643321991,
      "learning_rate": 1.3270833333333335e-05,
      "loss": 0.0461,
      "step": 2907
    },
    {
      "epoch": 0.3365740740740741,
      "grad_norm": 0.1541406810283661,
      "learning_rate": 1.3268518518518521e-05,
      "loss": 0.0294,
      "step": 2908
    },
    {
      "epoch": 0.3366898148148148,
      "grad_norm": 8.276703834533691,
      "learning_rate": 1.3266203703703704e-05,
      "loss": 0.1067,
      "step": 2909
    },
    {
      "epoch": 0.3368055555555556,
      "grad_norm": 4.459847927093506,
      "learning_rate": 1.326388888888889e-05,
      "loss": 0.0578,
      "step": 2910
    },
    {
      "epoch": 0.3369212962962963,
      "grad_norm": 0.21316863596439362,
      "learning_rate": 1.3261574074074076e-05,
      "loss": 0.0354,
      "step": 2911
    },
    {
      "epoch": 0.337037037037037,
      "grad_norm": 0.16545870900154114,
      "learning_rate": 1.325925925925926e-05,
      "loss": 0.031,
      "step": 2912
    },
    {
      "epoch": 0.3371527777777778,
      "grad_norm": 0.6620042324066162,
      "learning_rate": 1.3256944444444446e-05,
      "loss": 0.0429,
      "step": 2913
    },
    {
      "epoch": 0.3372685185185185,
      "grad_norm": 0.13394774496555328,
      "learning_rate": 1.3254629629629632e-05,
      "loss": 0.026,
      "step": 2914
    },
    {
      "epoch": 0.33738425925925924,
      "grad_norm": 0.18540596961975098,
      "learning_rate": 1.3252314814814814e-05,
      "loss": 0.0347,
      "step": 2915
    },
    {
      "epoch": 0.3375,
      "grad_norm": 135.0394287109375,
      "learning_rate": 1.325e-05,
      "loss": 0.5301,
      "step": 2916
    },
    {
      "epoch": 0.33761574074074074,
      "grad_norm": 0.3060648441314697,
      "learning_rate": 1.3247685185185186e-05,
      "loss": 0.0433,
      "step": 2917
    },
    {
      "epoch": 0.33773148148148147,
      "grad_norm": 0.14859537780284882,
      "learning_rate": 1.3245370370370372e-05,
      "loss": 0.0285,
      "step": 2918
    },
    {
      "epoch": 0.33784722222222224,
      "grad_norm": 0.13451164960861206,
      "learning_rate": 1.3243055555555558e-05,
      "loss": 0.0256,
      "step": 2919
    },
    {
      "epoch": 0.33796296296296297,
      "grad_norm": 0.2467835396528244,
      "learning_rate": 1.3240740740740742e-05,
      "loss": 0.0371,
      "step": 2920
    },
    {
      "epoch": 0.3380787037037037,
      "grad_norm": 12.153977394104004,
      "learning_rate": 1.3238425925925927e-05,
      "loss": 2.3318,
      "step": 2921
    },
    {
      "epoch": 0.33819444444444446,
      "grad_norm": 1.1898304224014282,
      "learning_rate": 1.3236111111111112e-05,
      "loss": 0.058,
      "step": 2922
    },
    {
      "epoch": 0.3383101851851852,
      "grad_norm": 0.17192032933235168,
      "learning_rate": 1.3233796296296297e-05,
      "loss": 0.0239,
      "step": 2923
    },
    {
      "epoch": 0.3384259259259259,
      "grad_norm": 0.17104776203632355,
      "learning_rate": 1.3231481481481483e-05,
      "loss": 0.0313,
      "step": 2924
    },
    {
      "epoch": 0.3385416666666667,
      "grad_norm": 0.17355039715766907,
      "learning_rate": 1.3229166666666669e-05,
      "loss": 0.033,
      "step": 2925
    },
    {
      "epoch": 0.3386574074074074,
      "grad_norm": 0.15703870356082916,
      "learning_rate": 1.3226851851851851e-05,
      "loss": 0.0297,
      "step": 2926
    },
    {
      "epoch": 0.33877314814814813,
      "grad_norm": 16.446264266967773,
      "learning_rate": 1.3224537037037037e-05,
      "loss": 2.3215,
      "step": 2927
    },
    {
      "epoch": 0.3388888888888889,
      "grad_norm": 0.12827855348587036,
      "learning_rate": 1.3222222222222223e-05,
      "loss": 0.0247,
      "step": 2928
    },
    {
      "epoch": 0.33900462962962963,
      "grad_norm": 0.2785281836986542,
      "learning_rate": 1.3219907407407409e-05,
      "loss": 0.0467,
      "step": 2929
    },
    {
      "epoch": 0.33912037037037035,
      "grad_norm": 0.2217300832271576,
      "learning_rate": 1.3217592592592593e-05,
      "loss": 0.0438,
      "step": 2930
    },
    {
      "epoch": 0.3392361111111111,
      "grad_norm": 0.196589395403862,
      "learning_rate": 1.321527777777778e-05,
      "loss": 0.038,
      "step": 2931
    },
    {
      "epoch": 0.33935185185185185,
      "grad_norm": 0.2383737862110138,
      "learning_rate": 1.3212962962962963e-05,
      "loss": 0.0356,
      "step": 2932
    },
    {
      "epoch": 0.33946759259259257,
      "grad_norm": 0.12491947412490845,
      "learning_rate": 1.321064814814815e-05,
      "loss": 0.024,
      "step": 2933
    },
    {
      "epoch": 0.33958333333333335,
      "grad_norm": 0.1630677580833435,
      "learning_rate": 1.3208333333333334e-05,
      "loss": 0.0305,
      "step": 2934
    },
    {
      "epoch": 0.33969907407407407,
      "grad_norm": 0.1892838031053543,
      "learning_rate": 1.320601851851852e-05,
      "loss": 0.036,
      "step": 2935
    },
    {
      "epoch": 0.3398148148148148,
      "grad_norm": 0.16227558255195618,
      "learning_rate": 1.3203703703703706e-05,
      "loss": 0.0307,
      "step": 2936
    },
    {
      "epoch": 0.33993055555555557,
      "grad_norm": 0.12293782085180283,
      "learning_rate": 1.3201388888888891e-05,
      "loss": 0.0238,
      "step": 2937
    },
    {
      "epoch": 0.3400462962962963,
      "grad_norm": 0.20625081658363342,
      "learning_rate": 1.3199074074074074e-05,
      "loss": 0.0404,
      "step": 2938
    },
    {
      "epoch": 0.340162037037037,
      "grad_norm": 0.1828368604183197,
      "learning_rate": 1.319675925925926e-05,
      "loss": 0.0352,
      "step": 2939
    },
    {
      "epoch": 0.3402777777777778,
      "grad_norm": 0.4211386740207672,
      "learning_rate": 1.3194444444444446e-05,
      "loss": 0.0396,
      "step": 2940
    },
    {
      "epoch": 0.3403935185185185,
      "grad_norm": 0.24667900800704956,
      "learning_rate": 1.319212962962963e-05,
      "loss": 0.0454,
      "step": 2941
    },
    {
      "epoch": 0.34050925925925923,
      "grad_norm": 85.13268280029297,
      "learning_rate": 1.3189814814814816e-05,
      "loss": 2.1426,
      "step": 2942
    },
    {
      "epoch": 0.340625,
      "grad_norm": 0.211530864238739,
      "learning_rate": 1.3187500000000002e-05,
      "loss": 0.0383,
      "step": 2943
    },
    {
      "epoch": 0.34074074074074073,
      "grad_norm": 0.18558025360107422,
      "learning_rate": 1.3185185185185185e-05,
      "loss": 0.0355,
      "step": 2944
    },
    {
      "epoch": 0.34085648148148145,
      "grad_norm": 0.6733586192131042,
      "learning_rate": 1.318287037037037e-05,
      "loss": 0.0458,
      "step": 2945
    },
    {
      "epoch": 0.34097222222222223,
      "grad_norm": 0.3079793453216553,
      "learning_rate": 1.3180555555555557e-05,
      "loss": 0.0436,
      "step": 2946
    },
    {
      "epoch": 0.34108796296296295,
      "grad_norm": 0.1935771256685257,
      "learning_rate": 1.3178240740740742e-05,
      "loss": 0.0314,
      "step": 2947
    },
    {
      "epoch": 0.34120370370370373,
      "grad_norm": 0.345854252576828,
      "learning_rate": 1.3175925925925928e-05,
      "loss": 0.0427,
      "step": 2948
    },
    {
      "epoch": 0.34131944444444445,
      "grad_norm": 0.35925793647766113,
      "learning_rate": 1.3173611111111113e-05,
      "loss": 0.0336,
      "step": 2949
    },
    {
      "epoch": 0.3414351851851852,
      "grad_norm": 0.17941229045391083,
      "learning_rate": 1.3171296296296297e-05,
      "loss": 0.0339,
      "step": 2950
    },
    {
      "epoch": 0.34155092592592595,
      "grad_norm": 0.13185641169548035,
      "learning_rate": 1.3168981481481483e-05,
      "loss": 0.0254,
      "step": 2951
    },
    {
      "epoch": 0.3416666666666667,
      "grad_norm": 0.8514050841331482,
      "learning_rate": 1.3166666666666667e-05,
      "loss": 0.0356,
      "step": 2952
    },
    {
      "epoch": 0.3417824074074074,
      "grad_norm": 58.15095520019531,
      "learning_rate": 1.3164351851851853e-05,
      "loss": 1.9172,
      "step": 2953
    },
    {
      "epoch": 0.3418981481481482,
      "grad_norm": 0.24616432189941406,
      "learning_rate": 1.3162037037037039e-05,
      "loss": 0.0366,
      "step": 2954
    },
    {
      "epoch": 0.3420138888888889,
      "grad_norm": 0.12282228469848633,
      "learning_rate": 1.3159722222222225e-05,
      "loss": 0.0237,
      "step": 2955
    },
    {
      "epoch": 0.3421296296296296,
      "grad_norm": 0.9866151213645935,
      "learning_rate": 1.3157407407407407e-05,
      "loss": 0.0521,
      "step": 2956
    },
    {
      "epoch": 0.3422453703703704,
      "grad_norm": 0.22621962428092957,
      "learning_rate": 1.3155092592592593e-05,
      "loss": 0.0306,
      "step": 2957
    },
    {
      "epoch": 0.3423611111111111,
      "grad_norm": 0.14536969363689423,
      "learning_rate": 1.315277777777778e-05,
      "loss": 0.0277,
      "step": 2958
    },
    {
      "epoch": 0.34247685185185184,
      "grad_norm": 0.18325303494930267,
      "learning_rate": 1.3150462962962964e-05,
      "loss": 0.0352,
      "step": 2959
    },
    {
      "epoch": 0.3425925925925926,
      "grad_norm": 0.18305736780166626,
      "learning_rate": 1.314814814814815e-05,
      "loss": 0.035,
      "step": 2960
    },
    {
      "epoch": 0.34270833333333334,
      "grad_norm": 0.17699271440505981,
      "learning_rate": 1.3145833333333336e-05,
      "loss": 0.0332,
      "step": 2961
    },
    {
      "epoch": 0.34282407407407406,
      "grad_norm": 0.39958322048187256,
      "learning_rate": 1.3143518518518518e-05,
      "loss": 0.0353,
      "step": 2962
    },
    {
      "epoch": 0.34293981481481484,
      "grad_norm": 0.17301204800605774,
      "learning_rate": 1.3141203703703704e-05,
      "loss": 0.0314,
      "step": 2963
    },
    {
      "epoch": 0.34305555555555556,
      "grad_norm": 1.156836748123169,
      "learning_rate": 1.313888888888889e-05,
      "loss": 0.0398,
      "step": 2964
    },
    {
      "epoch": 0.3431712962962963,
      "grad_norm": 0.12108328938484192,
      "learning_rate": 1.3136574074074076e-05,
      "loss": 0.0234,
      "step": 2965
    },
    {
      "epoch": 0.34328703703703706,
      "grad_norm": 0.1689060777425766,
      "learning_rate": 1.3134259259259262e-05,
      "loss": 0.031,
      "step": 2966
    },
    {
      "epoch": 0.3434027777777778,
      "grad_norm": 2.236067771911621,
      "learning_rate": 1.3131944444444446e-05,
      "loss": 0.0652,
      "step": 2967
    },
    {
      "epoch": 0.3435185185185185,
      "grad_norm": 0.1642366498708725,
      "learning_rate": 1.312962962962963e-05,
      "loss": 0.0229,
      "step": 2968
    },
    {
      "epoch": 0.3436342592592593,
      "grad_norm": 0.17593945562839508,
      "learning_rate": 1.3127314814814816e-05,
      "loss": 0.0326,
      "step": 2969
    },
    {
      "epoch": 0.34375,
      "grad_norm": 0.3820883333683014,
      "learning_rate": 1.3125e-05,
      "loss": 0.0383,
      "step": 2970
    },
    {
      "epoch": 0.3438657407407407,
      "grad_norm": 0.18169349431991577,
      "learning_rate": 1.3122685185185186e-05,
      "loss": 0.0341,
      "step": 2971
    },
    {
      "epoch": 0.3439814814814815,
      "grad_norm": 0.16634003818035126,
      "learning_rate": 1.3120370370370372e-05,
      "loss": 0.0308,
      "step": 2972
    },
    {
      "epoch": 0.3440972222222222,
      "grad_norm": 0.14689067006111145,
      "learning_rate": 1.3118055555555555e-05,
      "loss": 0.0282,
      "step": 2973
    },
    {
      "epoch": 0.34421296296296294,
      "grad_norm": 68.5843505859375,
      "learning_rate": 1.3115740740740741e-05,
      "loss": 1.2749,
      "step": 2974
    },
    {
      "epoch": 0.3443287037037037,
      "grad_norm": 0.1828153133392334,
      "learning_rate": 1.3113425925925927e-05,
      "loss": 0.0352,
      "step": 2975
    },
    {
      "epoch": 0.34444444444444444,
      "grad_norm": 2.257205009460449,
      "learning_rate": 1.3111111111111113e-05,
      "loss": 0.0514,
      "step": 2976
    },
    {
      "epoch": 0.34456018518518516,
      "grad_norm": 0.2011827975511551,
      "learning_rate": 1.3108796296296297e-05,
      "loss": 0.0261,
      "step": 2977
    },
    {
      "epoch": 0.34467592592592594,
      "grad_norm": 0.17419753968715668,
      "learning_rate": 1.3106481481481483e-05,
      "loss": 0.0323,
      "step": 2978
    },
    {
      "epoch": 0.34479166666666666,
      "grad_norm": 0.25849342346191406,
      "learning_rate": 1.3104166666666667e-05,
      "loss": 0.0358,
      "step": 2979
    },
    {
      "epoch": 0.3449074074074074,
      "grad_norm": 0.139535591006279,
      "learning_rate": 1.3101851851851852e-05,
      "loss": 0.0261,
      "step": 2980
    },
    {
      "epoch": 0.34502314814814816,
      "grad_norm": 0.19247494637966156,
      "learning_rate": 1.3099537037037037e-05,
      "loss": 0.037,
      "step": 2981
    },
    {
      "epoch": 0.3451388888888889,
      "grad_norm": 79.06796264648438,
      "learning_rate": 1.3097222222222223e-05,
      "loss": 1.5436,
      "step": 2982
    },
    {
      "epoch": 0.3452546296296296,
      "grad_norm": 0.15460161864757538,
      "learning_rate": 1.309490740740741e-05,
      "loss": 0.0294,
      "step": 2983
    },
    {
      "epoch": 0.3453703703703704,
      "grad_norm": 8.5569429397583,
      "learning_rate": 1.3092592592592595e-05,
      "loss": 0.0751,
      "step": 2984
    },
    {
      "epoch": 0.3454861111111111,
      "grad_norm": 0.16136138141155243,
      "learning_rate": 1.3090277777777778e-05,
      "loss": 0.0311,
      "step": 2985
    },
    {
      "epoch": 0.3456018518518518,
      "grad_norm": 0.3281351923942566,
      "learning_rate": 1.3087962962962964e-05,
      "loss": 0.0455,
      "step": 2986
    },
    {
      "epoch": 0.3457175925925926,
      "grad_norm": 0.14300329983234406,
      "learning_rate": 1.308564814814815e-05,
      "loss": 0.0273,
      "step": 2987
    },
    {
      "epoch": 0.3458333333333333,
      "grad_norm": 7.058682918548584,
      "learning_rate": 1.3083333333333334e-05,
      "loss": 2.5813,
      "step": 2988
    },
    {
      "epoch": 0.34594907407407405,
      "grad_norm": 0.19775249063968658,
      "learning_rate": 1.308101851851852e-05,
      "loss": 0.035,
      "step": 2989
    },
    {
      "epoch": 0.3460648148148148,
      "grad_norm": 0.12271560728549957,
      "learning_rate": 1.3078703703703706e-05,
      "loss": 0.0237,
      "step": 2990
    },
    {
      "epoch": 0.34618055555555555,
      "grad_norm": 0.14262685179710388,
      "learning_rate": 1.3076388888888888e-05,
      "loss": 0.0273,
      "step": 2991
    },
    {
      "epoch": 0.34629629629629627,
      "grad_norm": 0.20296978950500488,
      "learning_rate": 1.3074074074074074e-05,
      "loss": 0.0391,
      "step": 2992
    },
    {
      "epoch": 0.34641203703703705,
      "grad_norm": 0.15316437184810638,
      "learning_rate": 1.307175925925926e-05,
      "loss": 0.0292,
      "step": 2993
    },
    {
      "epoch": 0.34652777777777777,
      "grad_norm": 5.638631343841553,
      "learning_rate": 1.3069444444444446e-05,
      "loss": 2.9819,
      "step": 2994
    },
    {
      "epoch": 0.34664351851851855,
      "grad_norm": 0.11830326169729233,
      "learning_rate": 1.306712962962963e-05,
      "loss": 0.0229,
      "step": 2995
    },
    {
      "epoch": 0.34675925925925927,
      "grad_norm": 0.1273384392261505,
      "learning_rate": 1.3064814814814816e-05,
      "loss": 0.0246,
      "step": 2996
    },
    {
      "epoch": 0.346875,
      "grad_norm": 0.2306261658668518,
      "learning_rate": 1.30625e-05,
      "loss": 0.0404,
      "step": 2997
    },
    {
      "epoch": 0.34699074074074077,
      "grad_norm": 118.45069885253906,
      "learning_rate": 1.3060185185185185e-05,
      "loss": 1.0843,
      "step": 2998
    },
    {
      "epoch": 0.3471064814814815,
      "grad_norm": 0.6908897161483765,
      "learning_rate": 1.3057870370370371e-05,
      "loss": 0.0315,
      "step": 2999
    },
    {
      "epoch": 0.3472222222222222,
      "grad_norm": 0.1543329358100891,
      "learning_rate": 1.3055555555555557e-05,
      "loss": 0.0289,
      "step": 3000
    },
    {
      "epoch": 0.347337962962963,
      "grad_norm": 0.17223578691482544,
      "learning_rate": 1.3053240740740743e-05,
      "loss": 0.0323,
      "step": 3001
    },
    {
      "epoch": 0.3474537037037037,
      "grad_norm": 0.13185343146324158,
      "learning_rate": 1.3050925925925929e-05,
      "loss": 0.0252,
      "step": 3002
    },
    {
      "epoch": 0.34756944444444443,
      "grad_norm": 0.5026787519454956,
      "learning_rate": 1.3048611111111111e-05,
      "loss": 0.0473,
      "step": 3003
    },
    {
      "epoch": 0.3476851851851852,
      "grad_norm": 17.555835723876953,
      "learning_rate": 1.3046296296296297e-05,
      "loss": 0.0791,
      "step": 3004
    },
    {
      "epoch": 0.34780092592592593,
      "grad_norm": 0.2183459848165512,
      "learning_rate": 1.3043981481481483e-05,
      "loss": 0.0399,
      "step": 3005
    },
    {
      "epoch": 0.34791666666666665,
      "grad_norm": 0.21887612342834473,
      "learning_rate": 1.3041666666666667e-05,
      "loss": 0.0362,
      "step": 3006
    },
    {
      "epoch": 0.34803240740740743,
      "grad_norm": 42.659812927246094,
      "learning_rate": 1.3039351851851853e-05,
      "loss": 0.2244,
      "step": 3007
    },
    {
      "epoch": 0.34814814814814815,
      "grad_norm": 0.18113258481025696,
      "learning_rate": 1.303703703703704e-05,
      "loss": 0.0336,
      "step": 3008
    },
    {
      "epoch": 0.3482638888888889,
      "grad_norm": 0.18748287856578827,
      "learning_rate": 1.3034722222222222e-05,
      "loss": 0.0339,
      "step": 3009
    },
    {
      "epoch": 0.34837962962962965,
      "grad_norm": 27.565082550048828,
      "learning_rate": 1.3032407407407408e-05,
      "loss": 2.1872,
      "step": 3010
    },
    {
      "epoch": 0.34849537037037037,
      "grad_norm": 0.14606577157974243,
      "learning_rate": 1.3030092592592594e-05,
      "loss": 0.0277,
      "step": 3011
    },
    {
      "epoch": 0.3486111111111111,
      "grad_norm": 5.57823371887207,
      "learning_rate": 1.302777777777778e-05,
      "loss": 0.0547,
      "step": 3012
    },
    {
      "epoch": 0.34872685185185187,
      "grad_norm": 0.14785553514957428,
      "learning_rate": 1.3025462962962964e-05,
      "loss": 0.0282,
      "step": 3013
    },
    {
      "epoch": 0.3488425925925926,
      "grad_norm": 0.14297503232955933,
      "learning_rate": 1.302314814814815e-05,
      "loss": 0.0273,
      "step": 3014
    },
    {
      "epoch": 0.3489583333333333,
      "grad_norm": 0.21213728189468384,
      "learning_rate": 1.3020833333333334e-05,
      "loss": 0.0353,
      "step": 3015
    },
    {
      "epoch": 0.3490740740740741,
      "grad_norm": 0.472027987241745,
      "learning_rate": 1.3018518518518518e-05,
      "loss": 0.0404,
      "step": 3016
    },
    {
      "epoch": 0.3491898148148148,
      "grad_norm": 0.14334318041801453,
      "learning_rate": 1.3016203703703704e-05,
      "loss": 0.0273,
      "step": 3017
    },
    {
      "epoch": 0.34930555555555554,
      "grad_norm": 2.7253479957580566,
      "learning_rate": 1.301388888888889e-05,
      "loss": 0.0466,
      "step": 3018
    },
    {
      "epoch": 0.3494212962962963,
      "grad_norm": 0.17634214460849762,
      "learning_rate": 1.3011574074074076e-05,
      "loss": 0.0244,
      "step": 3019
    },
    {
      "epoch": 0.34953703703703703,
      "grad_norm": 0.14418183267116547,
      "learning_rate": 1.3009259259259259e-05,
      "loss": 0.0252,
      "step": 3020
    },
    {
      "epoch": 0.34965277777777776,
      "grad_norm": 34.880001068115234,
      "learning_rate": 1.3006944444444445e-05,
      "loss": 1.3244,
      "step": 3021
    },
    {
      "epoch": 0.34976851851851853,
      "grad_norm": 0.1780836284160614,
      "learning_rate": 1.300462962962963e-05,
      "loss": 0.0337,
      "step": 3022
    },
    {
      "epoch": 0.34988425925925926,
      "grad_norm": 0.15257799625396729,
      "learning_rate": 1.3002314814814817e-05,
      "loss": 0.028,
      "step": 3023
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.12781254947185516,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.0243,
      "step": 3024
    },
    {
      "epoch": 0.35011574074074076,
      "grad_norm": 85.56839752197266,
      "learning_rate": 1.2997685185185187e-05,
      "loss": 0.3252,
      "step": 3025
    },
    {
      "epoch": 0.3502314814814815,
      "grad_norm": 0.25479307770729065,
      "learning_rate": 1.2995370370370371e-05,
      "loss": 0.0414,
      "step": 3026
    },
    {
      "epoch": 0.3503472222222222,
      "grad_norm": 0.3762005567550659,
      "learning_rate": 1.2993055555555555e-05,
      "loss": 0.0473,
      "step": 3027
    },
    {
      "epoch": 0.350462962962963,
      "grad_norm": 0.14072422683238983,
      "learning_rate": 1.2990740740740741e-05,
      "loss": 0.0269,
      "step": 3028
    },
    {
      "epoch": 0.3505787037037037,
      "grad_norm": 93.10458374023438,
      "learning_rate": 1.2988425925925927e-05,
      "loss": 0.9981,
      "step": 3029
    },
    {
      "epoch": 0.3506944444444444,
      "grad_norm": 0.12487436830997467,
      "learning_rate": 1.2986111111111113e-05,
      "loss": 0.0241,
      "step": 3030
    },
    {
      "epoch": 0.3508101851851852,
      "grad_norm": 0.18554885685443878,
      "learning_rate": 1.2983796296296297e-05,
      "loss": 0.0359,
      "step": 3031
    },
    {
      "epoch": 0.3509259259259259,
      "grad_norm": 0.2933078408241272,
      "learning_rate": 1.2981481481481482e-05,
      "loss": 0.033,
      "step": 3032
    },
    {
      "epoch": 0.35104166666666664,
      "grad_norm": 0.14430679380893707,
      "learning_rate": 1.2979166666666668e-05,
      "loss": 0.0272,
      "step": 3033
    },
    {
      "epoch": 0.3511574074074074,
      "grad_norm": 0.4045060873031616,
      "learning_rate": 1.2976851851851852e-05,
      "loss": 0.0333,
      "step": 3034
    },
    {
      "epoch": 0.35127314814814814,
      "grad_norm": 0.4370061159133911,
      "learning_rate": 1.2974537037037038e-05,
      "loss": 0.0438,
      "step": 3035
    },
    {
      "epoch": 0.35138888888888886,
      "grad_norm": 10.374641418457031,
      "learning_rate": 1.2972222222222224e-05,
      "loss": 2.8647,
      "step": 3036
    },
    {
      "epoch": 0.35150462962962964,
      "grad_norm": 0.19216927886009216,
      "learning_rate": 1.296990740740741e-05,
      "loss": 0.0344,
      "step": 3037
    },
    {
      "epoch": 0.35162037037037036,
      "grad_norm": 0.16682450473308563,
      "learning_rate": 1.2967592592592592e-05,
      "loss": 0.0298,
      "step": 3038
    },
    {
      "epoch": 0.35173611111111114,
      "grad_norm": 0.1780572235584259,
      "learning_rate": 1.2965277777777778e-05,
      "loss": 0.0328,
      "step": 3039
    },
    {
      "epoch": 0.35185185185185186,
      "grad_norm": 0.16399091482162476,
      "learning_rate": 1.2962962962962964e-05,
      "loss": 0.0286,
      "step": 3040
    },
    {
      "epoch": 0.3519675925925926,
      "grad_norm": 0.1656022071838379,
      "learning_rate": 1.296064814814815e-05,
      "loss": 0.0313,
      "step": 3041
    },
    {
      "epoch": 0.35208333333333336,
      "grad_norm": 21.195703506469727,
      "learning_rate": 1.2958333333333334e-05,
      "loss": 2.2884,
      "step": 3042
    },
    {
      "epoch": 0.3521990740740741,
      "grad_norm": 0.12400957942008972,
      "learning_rate": 1.295601851851852e-05,
      "loss": 0.0238,
      "step": 3043
    },
    {
      "epoch": 0.3523148148148148,
      "grad_norm": 0.17599256336688995,
      "learning_rate": 1.2953703703703705e-05,
      "loss": 0.0332,
      "step": 3044
    },
    {
      "epoch": 0.3524305555555556,
      "grad_norm": 0.5647238492965698,
      "learning_rate": 1.2951388888888889e-05,
      "loss": 0.0381,
      "step": 3045
    },
    {
      "epoch": 0.3525462962962963,
      "grad_norm": 0.17791083455085754,
      "learning_rate": 1.2949074074074075e-05,
      "loss": 0.0331,
      "step": 3046
    },
    {
      "epoch": 0.352662037037037,
      "grad_norm": 1.414889931678772,
      "learning_rate": 1.294675925925926e-05,
      "loss": 0.0362,
      "step": 3047
    },
    {
      "epoch": 0.3527777777777778,
      "grad_norm": 0.17485816776752472,
      "learning_rate": 1.2944444444444447e-05,
      "loss": 0.0335,
      "step": 3048
    },
    {
      "epoch": 0.3528935185185185,
      "grad_norm": 3.245115041732788,
      "learning_rate": 1.2942129629629631e-05,
      "loss": 0.0589,
      "step": 3049
    },
    {
      "epoch": 0.35300925925925924,
      "grad_norm": 3.012129783630371,
      "learning_rate": 1.2939814814814815e-05,
      "loss": 0.0465,
      "step": 3050
    },
    {
      "epoch": 0.353125,
      "grad_norm": 0.1205899640917778,
      "learning_rate": 1.2937500000000001e-05,
      "loss": 0.0232,
      "step": 3051
    },
    {
      "epoch": 0.35324074074074074,
      "grad_norm": 0.15994977951049805,
      "learning_rate": 1.2935185185185187e-05,
      "loss": 0.0306,
      "step": 3052
    },
    {
      "epoch": 0.35335648148148147,
      "grad_norm": 0.1757587194442749,
      "learning_rate": 1.2932870370370371e-05,
      "loss": 0.0316,
      "step": 3053
    },
    {
      "epoch": 0.35347222222222224,
      "grad_norm": 138.6441192626953,
      "learning_rate": 1.2930555555555557e-05,
      "loss": 0.4972,
      "step": 3054
    },
    {
      "epoch": 0.35358796296296297,
      "grad_norm": 30.807910919189453,
      "learning_rate": 1.2928240740740743e-05,
      "loss": 2.5829,
      "step": 3055
    },
    {
      "epoch": 0.3537037037037037,
      "grad_norm": 0.17428289353847504,
      "learning_rate": 1.2925925925925926e-05,
      "loss": 0.033,
      "step": 3056
    },
    {
      "epoch": 0.35381944444444446,
      "grad_norm": 0.14432355761528015,
      "learning_rate": 1.2923611111111112e-05,
      "loss": 0.0279,
      "step": 3057
    },
    {
      "epoch": 0.3539351851851852,
      "grad_norm": 0.13525442779064178,
      "learning_rate": 1.2921296296296298e-05,
      "loss": 0.0259,
      "step": 3058
    },
    {
      "epoch": 0.3540509259259259,
      "grad_norm": 19.733963012695312,
      "learning_rate": 1.2918981481481484e-05,
      "loss": 2.8274,
      "step": 3059
    },
    {
      "epoch": 0.3541666666666667,
      "grad_norm": 0.14646406471729279,
      "learning_rate": 1.2916666666666668e-05,
      "loss": 0.0282,
      "step": 3060
    },
    {
      "epoch": 0.3542824074074074,
      "grad_norm": 0.21181954443454742,
      "learning_rate": 1.2914351851851852e-05,
      "loss": 0.0415,
      "step": 3061
    },
    {
      "epoch": 0.35439814814814813,
      "grad_norm": 0.16955974698066711,
      "learning_rate": 1.2912037037037038e-05,
      "loss": 0.0314,
      "step": 3062
    },
    {
      "epoch": 0.3545138888888889,
      "grad_norm": 0.2604399025440216,
      "learning_rate": 1.2909722222222222e-05,
      "loss": 0.0356,
      "step": 3063
    },
    {
      "epoch": 0.35462962962962963,
      "grad_norm": 0.18505389988422394,
      "learning_rate": 1.2907407407407408e-05,
      "loss": 0.0327,
      "step": 3064
    },
    {
      "epoch": 0.35474537037037035,
      "grad_norm": 0.13676372170448303,
      "learning_rate": 1.2905092592592594e-05,
      "loss": 0.0261,
      "step": 3065
    },
    {
      "epoch": 0.3548611111111111,
      "grad_norm": 9.613199234008789,
      "learning_rate": 1.290277777777778e-05,
      "loss": 2.7575,
      "step": 3066
    },
    {
      "epoch": 0.35497685185185185,
      "grad_norm": 73.91495513916016,
      "learning_rate": 1.2900462962962963e-05,
      "loss": 0.9681,
      "step": 3067
    },
    {
      "epoch": 0.35509259259259257,
      "grad_norm": 0.8433346748352051,
      "learning_rate": 1.2898148148148149e-05,
      "loss": 0.0501,
      "step": 3068
    },
    {
      "epoch": 0.35520833333333335,
      "grad_norm": 0.2122635692358017,
      "learning_rate": 1.2895833333333335e-05,
      "loss": 0.0296,
      "step": 3069
    },
    {
      "epoch": 0.35532407407407407,
      "grad_norm": 6.0650634765625,
      "learning_rate": 1.289351851851852e-05,
      "loss": 2.8404,
      "step": 3070
    },
    {
      "epoch": 0.3554398148148148,
      "grad_norm": 0.14537163078784943,
      "learning_rate": 1.2891203703703705e-05,
      "loss": 0.0272,
      "step": 3071
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 0.2126806080341339,
      "learning_rate": 1.288888888888889e-05,
      "loss": 0.0394,
      "step": 3072
    },
    {
      "epoch": 0.3556712962962963,
      "grad_norm": 0.15332157909870148,
      "learning_rate": 1.2886574074074075e-05,
      "loss": 0.0268,
      "step": 3073
    },
    {
      "epoch": 0.355787037037037,
      "grad_norm": 0.16733494400978088,
      "learning_rate": 1.288425925925926e-05,
      "loss": 0.0316,
      "step": 3074
    },
    {
      "epoch": 0.3559027777777778,
      "grad_norm": 0.29622945189476013,
      "learning_rate": 1.2881944444444445e-05,
      "loss": 0.0378,
      "step": 3075
    },
    {
      "epoch": 0.3560185185185185,
      "grad_norm": 1.1153525114059448,
      "learning_rate": 1.2879629629629631e-05,
      "loss": 0.0473,
      "step": 3076
    },
    {
      "epoch": 0.35613425925925923,
      "grad_norm": 0.19184231758117676,
      "learning_rate": 1.2877314814814817e-05,
      "loss": 0.0374,
      "step": 3077
    },
    {
      "epoch": 0.35625,
      "grad_norm": 91.96439361572266,
      "learning_rate": 1.2875000000000001e-05,
      "loss": 0.9755,
      "step": 3078
    },
    {
      "epoch": 0.35636574074074073,
      "grad_norm": 0.20266082882881165,
      "learning_rate": 1.2872685185185185e-05,
      "loss": 0.0387,
      "step": 3079
    },
    {
      "epoch": 0.35648148148148145,
      "grad_norm": 177.40325927734375,
      "learning_rate": 1.2870370370370371e-05,
      "loss": 0.5751,
      "step": 3080
    },
    {
      "epoch": 0.35659722222222223,
      "grad_norm": 0.15910187363624573,
      "learning_rate": 1.2868055555555556e-05,
      "loss": 0.0298,
      "step": 3081
    },
    {
      "epoch": 0.35671296296296295,
      "grad_norm": 11.314373016357422,
      "learning_rate": 1.2865740740740742e-05,
      "loss": 2.4718,
      "step": 3082
    },
    {
      "epoch": 0.35682870370370373,
      "grad_norm": 0.17800837755203247,
      "learning_rate": 1.2863425925925928e-05,
      "loss": 0.032,
      "step": 3083
    },
    {
      "epoch": 0.35694444444444445,
      "grad_norm": 0.15854096412658691,
      "learning_rate": 1.2861111111111114e-05,
      "loss": 0.0277,
      "step": 3084
    },
    {
      "epoch": 0.3570601851851852,
      "grad_norm": 0.2931801974773407,
      "learning_rate": 1.2858796296296296e-05,
      "loss": 0.0347,
      "step": 3085
    },
    {
      "epoch": 0.35717592592592595,
      "grad_norm": 0.17220820486545563,
      "learning_rate": 1.2856481481481482e-05,
      "loss": 0.0321,
      "step": 3086
    },
    {
      "epoch": 0.3572916666666667,
      "grad_norm": 0.2126835435628891,
      "learning_rate": 1.2854166666666668e-05,
      "loss": 0.0336,
      "step": 3087
    },
    {
      "epoch": 0.3574074074074074,
      "grad_norm": 0.16580846905708313,
      "learning_rate": 1.2851851851851854e-05,
      "loss": 0.0318,
      "step": 3088
    },
    {
      "epoch": 0.3575231481481482,
      "grad_norm": 0.13664640486240387,
      "learning_rate": 1.2849537037037038e-05,
      "loss": 0.0244,
      "step": 3089
    },
    {
      "epoch": 0.3576388888888889,
      "grad_norm": 0.212082639336586,
      "learning_rate": 1.2847222222222224e-05,
      "loss": 0.0407,
      "step": 3090
    },
    {
      "epoch": 0.3577546296296296,
      "grad_norm": 0.16678836941719055,
      "learning_rate": 1.2844907407407408e-05,
      "loss": 0.0233,
      "step": 3091
    },
    {
      "epoch": 0.3578703703703704,
      "grad_norm": 29.243228912353516,
      "learning_rate": 1.2842592592592593e-05,
      "loss": 1.993,
      "step": 3092
    },
    {
      "epoch": 0.3579861111111111,
      "grad_norm": 0.16365131735801697,
      "learning_rate": 1.2840277777777779e-05,
      "loss": 0.0309,
      "step": 3093
    },
    {
      "epoch": 0.35810185185185184,
      "grad_norm": 0.1647641956806183,
      "learning_rate": 1.2837962962962964e-05,
      "loss": 0.0303,
      "step": 3094
    },
    {
      "epoch": 0.3582175925925926,
      "grad_norm": 0.2178950011730194,
      "learning_rate": 1.283564814814815e-05,
      "loss": 0.0368,
      "step": 3095
    },
    {
      "epoch": 0.35833333333333334,
      "grad_norm": 0.17744357883930206,
      "learning_rate": 1.2833333333333335e-05,
      "loss": 0.0334,
      "step": 3096
    },
    {
      "epoch": 0.35844907407407406,
      "grad_norm": 1.127838134765625,
      "learning_rate": 1.2831018518518519e-05,
      "loss": 0.0459,
      "step": 3097
    },
    {
      "epoch": 0.35856481481481484,
      "grad_norm": 0.2955116033554077,
      "learning_rate": 1.2828703703703705e-05,
      "loss": 0.0347,
      "step": 3098
    },
    {
      "epoch": 0.35868055555555556,
      "grad_norm": 0.6017053127288818,
      "learning_rate": 1.2826388888888889e-05,
      "loss": 0.0391,
      "step": 3099
    },
    {
      "epoch": 0.3587962962962963,
      "grad_norm": 0.1672409176826477,
      "learning_rate": 1.2824074074074075e-05,
      "loss": 0.0312,
      "step": 3100
    },
    {
      "epoch": 0.35891203703703706,
      "grad_norm": 0.27861258387565613,
      "learning_rate": 1.2821759259259261e-05,
      "loss": 0.0449,
      "step": 3101
    },
    {
      "epoch": 0.3590277777777778,
      "grad_norm": 0.17568339407444,
      "learning_rate": 1.2819444444444447e-05,
      "loss": 0.0324,
      "step": 3102
    },
    {
      "epoch": 0.3591435185185185,
      "grad_norm": 0.18631669878959656,
      "learning_rate": 1.281712962962963e-05,
      "loss": 0.0361,
      "step": 3103
    },
    {
      "epoch": 0.3592592592592593,
      "grad_norm": 0.19780763983726501,
      "learning_rate": 1.2814814814814815e-05,
      "loss": 0.0341,
      "step": 3104
    },
    {
      "epoch": 0.359375,
      "grad_norm": 0.13822802901268005,
      "learning_rate": 1.2812500000000001e-05,
      "loss": 0.0233,
      "step": 3105
    },
    {
      "epoch": 0.3594907407407407,
      "grad_norm": 0.21057136356830597,
      "learning_rate": 1.2810185185185187e-05,
      "loss": 0.0411,
      "step": 3106
    },
    {
      "epoch": 0.3596064814814815,
      "grad_norm": 0.20239290595054626,
      "learning_rate": 1.2807870370370372e-05,
      "loss": 0.0387,
      "step": 3107
    },
    {
      "epoch": 0.3597222222222222,
      "grad_norm": 2.0615789890289307,
      "learning_rate": 1.2805555555555556e-05,
      "loss": 0.043,
      "step": 3108
    },
    {
      "epoch": 0.35983796296296294,
      "grad_norm": 0.14620240032672882,
      "learning_rate": 1.2803240740740742e-05,
      "loss": 0.0267,
      "step": 3109
    },
    {
      "epoch": 0.3599537037037037,
      "grad_norm": 0.30129149556159973,
      "learning_rate": 1.2800925925925926e-05,
      "loss": 0.0369,
      "step": 3110
    },
    {
      "epoch": 0.36006944444444444,
      "grad_norm": 8.143372535705566,
      "learning_rate": 1.2798611111111112e-05,
      "loss": 0.065,
      "step": 3111
    },
    {
      "epoch": 0.36018518518518516,
      "grad_norm": 0.45678654313087463,
      "learning_rate": 1.2796296296296298e-05,
      "loss": 0.0386,
      "step": 3112
    },
    {
      "epoch": 0.36030092592592594,
      "grad_norm": 0.16441389918327332,
      "learning_rate": 1.2793981481481484e-05,
      "loss": 0.0225,
      "step": 3113
    },
    {
      "epoch": 0.36041666666666666,
      "grad_norm": 0.42095744609832764,
      "learning_rate": 1.2791666666666666e-05,
      "loss": 0.0271,
      "step": 3114
    },
    {
      "epoch": 0.3605324074074074,
      "grad_norm": 34.1894645690918,
      "learning_rate": 1.2789351851851852e-05,
      "loss": 2.1017,
      "step": 3115
    },
    {
      "epoch": 0.36064814814814816,
      "grad_norm": 6.028829574584961,
      "learning_rate": 1.2787037037037038e-05,
      "loss": 2.662,
      "step": 3116
    },
    {
      "epoch": 0.3607638888888889,
      "grad_norm": 0.15636636316776276,
      "learning_rate": 1.2784722222222223e-05,
      "loss": 0.0302,
      "step": 3117
    },
    {
      "epoch": 0.3608796296296296,
      "grad_norm": 1.4803731441497803,
      "learning_rate": 1.2782407407407409e-05,
      "loss": 0.0328,
      "step": 3118
    },
    {
      "epoch": 0.3609953703703704,
      "grad_norm": 0.13337309658527374,
      "learning_rate": 1.2780092592592594e-05,
      "loss": 0.0255,
      "step": 3119
    },
    {
      "epoch": 0.3611111111111111,
      "grad_norm": 0.16779620945453644,
      "learning_rate": 1.2777777777777777e-05,
      "loss": 0.0228,
      "step": 3120
    },
    {
      "epoch": 0.3612268518518518,
      "grad_norm": 0.23798590898513794,
      "learning_rate": 1.2775462962962963e-05,
      "loss": 0.0397,
      "step": 3121
    },
    {
      "epoch": 0.3613425925925926,
      "grad_norm": 32.60087585449219,
      "learning_rate": 1.2773148148148149e-05,
      "loss": 2.3417,
      "step": 3122
    },
    {
      "epoch": 0.3614583333333333,
      "grad_norm": 0.15279823541641235,
      "learning_rate": 1.2770833333333335e-05,
      "loss": 0.0265,
      "step": 3123
    },
    {
      "epoch": 0.36157407407407405,
      "grad_norm": 0.16278742253780365,
      "learning_rate": 1.276851851851852e-05,
      "loss": 0.0247,
      "step": 3124
    },
    {
      "epoch": 0.3616898148148148,
      "grad_norm": 1.082843542098999,
      "learning_rate": 1.2766203703703705e-05,
      "loss": 0.0436,
      "step": 3125
    },
    {
      "epoch": 0.36180555555555555,
      "grad_norm": 0.21104538440704346,
      "learning_rate": 1.276388888888889e-05,
      "loss": 0.039,
      "step": 3126
    },
    {
      "epoch": 0.36192129629629627,
      "grad_norm": 0.16283534467220306,
      "learning_rate": 1.2761574074074075e-05,
      "loss": 0.0304,
      "step": 3127
    },
    {
      "epoch": 0.36203703703703705,
      "grad_norm": 0.18226557970046997,
      "learning_rate": 1.275925925925926e-05,
      "loss": 0.0321,
      "step": 3128
    },
    {
      "epoch": 0.36215277777777777,
      "grad_norm": 0.44668254256248474,
      "learning_rate": 1.2756944444444445e-05,
      "loss": 0.0363,
      "step": 3129
    },
    {
      "epoch": 0.36226851851851855,
      "grad_norm": 0.1596599668264389,
      "learning_rate": 1.2754629629629631e-05,
      "loss": 0.0291,
      "step": 3130
    },
    {
      "epoch": 0.36238425925925927,
      "grad_norm": 0.12710942327976227,
      "learning_rate": 1.2752314814814817e-05,
      "loss": 0.0244,
      "step": 3131
    },
    {
      "epoch": 0.3625,
      "grad_norm": 0.19151368737220764,
      "learning_rate": 1.275e-05,
      "loss": 0.0339,
      "step": 3132
    },
    {
      "epoch": 0.36261574074074077,
      "grad_norm": 0.1519460380077362,
      "learning_rate": 1.2747685185185186e-05,
      "loss": 0.0279,
      "step": 3133
    },
    {
      "epoch": 0.3627314814814815,
      "grad_norm": 26.686586380004883,
      "learning_rate": 1.2745370370370372e-05,
      "loss": 2.1975,
      "step": 3134
    },
    {
      "epoch": 0.3628472222222222,
      "grad_norm": 0.16271162033081055,
      "learning_rate": 1.2743055555555556e-05,
      "loss": 0.0308,
      "step": 3135
    },
    {
      "epoch": 0.362962962962963,
      "grad_norm": 0.7174069881439209,
      "learning_rate": 1.2740740740740742e-05,
      "loss": 0.0533,
      "step": 3136
    },
    {
      "epoch": 0.3630787037037037,
      "grad_norm": 104.3858413696289,
      "learning_rate": 1.2738425925925928e-05,
      "loss": 2.8797,
      "step": 3137
    },
    {
      "epoch": 0.36319444444444443,
      "grad_norm": 51.0745964050293,
      "learning_rate": 1.273611111111111e-05,
      "loss": 2.1178,
      "step": 3138
    },
    {
      "epoch": 0.3633101851851852,
      "grad_norm": 142.7907257080078,
      "learning_rate": 1.2733796296296296e-05,
      "loss": 1.4908,
      "step": 3139
    },
    {
      "epoch": 0.36342592592592593,
      "grad_norm": 124.48796081542969,
      "learning_rate": 1.2731481481481482e-05,
      "loss": 0.7898,
      "step": 3140
    },
    {
      "epoch": 0.36354166666666665,
      "grad_norm": 0.14527077972888947,
      "learning_rate": 1.2729166666666668e-05,
      "loss": 0.0278,
      "step": 3141
    },
    {
      "epoch": 0.36365740740740743,
      "grad_norm": 0.32082080841064453,
      "learning_rate": 1.2726851851851854e-05,
      "loss": 0.0367,
      "step": 3142
    },
    {
      "epoch": 0.36377314814814815,
      "grad_norm": 3.663534164428711,
      "learning_rate": 1.2724537037037039e-05,
      "loss": 0.0395,
      "step": 3143
    },
    {
      "epoch": 0.3638888888888889,
      "grad_norm": 0.3217402398586273,
      "learning_rate": 1.2722222222222223e-05,
      "loss": 0.039,
      "step": 3144
    },
    {
      "epoch": 0.36400462962962965,
      "grad_norm": 0.17214936017990112,
      "learning_rate": 1.2719907407407409e-05,
      "loss": 0.0309,
      "step": 3145
    },
    {
      "epoch": 0.36412037037037037,
      "grad_norm": 0.11953088641166687,
      "learning_rate": 1.2717592592592593e-05,
      "loss": 0.0226,
      "step": 3146
    },
    {
      "epoch": 0.3642361111111111,
      "grad_norm": 0.15777599811553955,
      "learning_rate": 1.2715277777777779e-05,
      "loss": 0.0295,
      "step": 3147
    },
    {
      "epoch": 0.36435185185185187,
      "grad_norm": 0.1527474969625473,
      "learning_rate": 1.2712962962962965e-05,
      "loss": 0.0281,
      "step": 3148
    },
    {
      "epoch": 0.3644675925925926,
      "grad_norm": 19.5300350189209,
      "learning_rate": 1.271064814814815e-05,
      "loss": 2.3429,
      "step": 3149
    },
    {
      "epoch": 0.3645833333333333,
      "grad_norm": 0.2103492170572281,
      "learning_rate": 1.2708333333333333e-05,
      "loss": 0.0236,
      "step": 3150
    },
    {
      "epoch": 0.3646990740740741,
      "grad_norm": 0.19652605056762695,
      "learning_rate": 1.270601851851852e-05,
      "loss": 0.0372,
      "step": 3151
    },
    {
      "epoch": 0.3648148148148148,
      "grad_norm": 0.12669138610363007,
      "learning_rate": 1.2703703703703705e-05,
      "loss": 0.0229,
      "step": 3152
    },
    {
      "epoch": 0.36493055555555554,
      "grad_norm": 0.11412044614553452,
      "learning_rate": 1.270138888888889e-05,
      "loss": 0.022,
      "step": 3153
    },
    {
      "epoch": 0.3650462962962963,
      "grad_norm": 1.9609637260437012,
      "learning_rate": 1.2699074074074075e-05,
      "loss": 0.0635,
      "step": 3154
    },
    {
      "epoch": 0.36516203703703703,
      "grad_norm": 0.16813695430755615,
      "learning_rate": 1.269675925925926e-05,
      "loss": 0.0282,
      "step": 3155
    },
    {
      "epoch": 0.36527777777777776,
      "grad_norm": 0.15613101422786713,
      "learning_rate": 1.2694444444444446e-05,
      "loss": 0.0293,
      "step": 3156
    },
    {
      "epoch": 0.36539351851851853,
      "grad_norm": 0.2357197403907776,
      "learning_rate": 1.269212962962963e-05,
      "loss": 0.0376,
      "step": 3157
    },
    {
      "epoch": 0.36550925925925926,
      "grad_norm": 0.15922138094902039,
      "learning_rate": 1.2689814814814816e-05,
      "loss": 0.0298,
      "step": 3158
    },
    {
      "epoch": 0.365625,
      "grad_norm": 10.627157211303711,
      "learning_rate": 1.2687500000000002e-05,
      "loss": 2.5653,
      "step": 3159
    },
    {
      "epoch": 0.36574074074074076,
      "grad_norm": 0.16772979497909546,
      "learning_rate": 1.2685185185185188e-05,
      "loss": 0.0312,
      "step": 3160
    },
    {
      "epoch": 0.3658564814814815,
      "grad_norm": 0.11481668055057526,
      "learning_rate": 1.268287037037037e-05,
      "loss": 0.022,
      "step": 3161
    },
    {
      "epoch": 0.3659722222222222,
      "grad_norm": 0.1698460727930069,
      "learning_rate": 1.2680555555555556e-05,
      "loss": 0.0324,
      "step": 3162
    },
    {
      "epoch": 0.366087962962963,
      "grad_norm": 0.1446666717529297,
      "learning_rate": 1.2678240740740742e-05,
      "loss": 0.0272,
      "step": 3163
    },
    {
      "epoch": 0.3662037037037037,
      "grad_norm": 0.35294434428215027,
      "learning_rate": 1.2675925925925926e-05,
      "loss": 0.0436,
      "step": 3164
    },
    {
      "epoch": 0.3663194444444444,
      "grad_norm": 0.1768433004617691,
      "learning_rate": 1.2673611111111112e-05,
      "loss": 0.0336,
      "step": 3165
    },
    {
      "epoch": 0.3664351851851852,
      "grad_norm": 0.182049959897995,
      "learning_rate": 1.2671296296296298e-05,
      "loss": 0.0332,
      "step": 3166
    },
    {
      "epoch": 0.3665509259259259,
      "grad_norm": 0.1560632288455963,
      "learning_rate": 1.2668981481481481e-05,
      "loss": 0.0302,
      "step": 3167
    },
    {
      "epoch": 0.36666666666666664,
      "grad_norm": 0.12522464990615845,
      "learning_rate": 1.2666666666666667e-05,
      "loss": 0.0239,
      "step": 3168
    },
    {
      "epoch": 0.3667824074074074,
      "grad_norm": 0.12708982825279236,
      "learning_rate": 1.2664351851851853e-05,
      "loss": 0.0244,
      "step": 3169
    },
    {
      "epoch": 0.36689814814814814,
      "grad_norm": 0.5128975510597229,
      "learning_rate": 1.2662037037037039e-05,
      "loss": 0.0304,
      "step": 3170
    },
    {
      "epoch": 0.36701388888888886,
      "grad_norm": 0.18692435324192047,
      "learning_rate": 1.2659722222222225e-05,
      "loss": 0.0361,
      "step": 3171
    },
    {
      "epoch": 0.36712962962962964,
      "grad_norm": 0.5671437978744507,
      "learning_rate": 1.2657407407407409e-05,
      "loss": 0.0334,
      "step": 3172
    },
    {
      "epoch": 0.36724537037037036,
      "grad_norm": 0.149915874004364,
      "learning_rate": 1.2655092592592593e-05,
      "loss": 0.0273,
      "step": 3173
    },
    {
      "epoch": 0.36736111111111114,
      "grad_norm": 0.21287640929222107,
      "learning_rate": 1.2652777777777779e-05,
      "loss": 0.04,
      "step": 3174
    },
    {
      "epoch": 0.36747685185185186,
      "grad_norm": 0.18463774025440216,
      "learning_rate": 1.2650462962962963e-05,
      "loss": 0.0304,
      "step": 3175
    },
    {
      "epoch": 0.3675925925925926,
      "grad_norm": 0.16321539878845215,
      "learning_rate": 1.264814814814815e-05,
      "loss": 0.0296,
      "step": 3176
    },
    {
      "epoch": 0.36770833333333336,
      "grad_norm": 0.16397283971309662,
      "learning_rate": 1.2645833333333335e-05,
      "loss": 0.0226,
      "step": 3177
    },
    {
      "epoch": 0.3678240740740741,
      "grad_norm": 0.22176586091518402,
      "learning_rate": 1.2643518518518521e-05,
      "loss": 0.0304,
      "step": 3178
    },
    {
      "epoch": 0.3679398148148148,
      "grad_norm": 0.1522129327058792,
      "learning_rate": 1.2641203703703704e-05,
      "loss": 0.0211,
      "step": 3179
    },
    {
      "epoch": 0.3680555555555556,
      "grad_norm": 67.73489379882812,
      "learning_rate": 1.263888888888889e-05,
      "loss": 1.9422,
      "step": 3180
    },
    {
      "epoch": 0.3681712962962963,
      "grad_norm": 0.11862866580486298,
      "learning_rate": 1.2636574074074076e-05,
      "loss": 0.0223,
      "step": 3181
    },
    {
      "epoch": 0.368287037037037,
      "grad_norm": 5.105484962463379,
      "learning_rate": 1.263425925925926e-05,
      "loss": 0.0485,
      "step": 3182
    },
    {
      "epoch": 0.3684027777777778,
      "grad_norm": 0.14842362701892853,
      "learning_rate": 1.2631944444444446e-05,
      "loss": 0.0276,
      "step": 3183
    },
    {
      "epoch": 0.3685185185185185,
      "grad_norm": 51.88862991333008,
      "learning_rate": 1.2629629629629632e-05,
      "loss": 1.8025,
      "step": 3184
    },
    {
      "epoch": 0.36863425925925924,
      "grad_norm": 0.1881389319896698,
      "learning_rate": 1.2627314814814814e-05,
      "loss": 0.0317,
      "step": 3185
    },
    {
      "epoch": 0.36875,
      "grad_norm": 0.1394190788269043,
      "learning_rate": 1.2625e-05,
      "loss": 0.0268,
      "step": 3186
    },
    {
      "epoch": 0.36886574074074074,
      "grad_norm": 0.17043213546276093,
      "learning_rate": 1.2622685185185186e-05,
      "loss": 0.0312,
      "step": 3187
    },
    {
      "epoch": 0.36898148148148147,
      "grad_norm": 0.1600368469953537,
      "learning_rate": 1.2620370370370372e-05,
      "loss": 0.0301,
      "step": 3188
    },
    {
      "epoch": 0.36909722222222224,
      "grad_norm": 0.17362163960933685,
      "learning_rate": 1.2618055555555558e-05,
      "loss": 0.0322,
      "step": 3189
    },
    {
      "epoch": 0.36921296296296297,
      "grad_norm": 0.14912816882133484,
      "learning_rate": 1.2615740740740742e-05,
      "loss": 0.0273,
      "step": 3190
    },
    {
      "epoch": 0.3693287037037037,
      "grad_norm": 0.17270427942276,
      "learning_rate": 1.2613425925925927e-05,
      "loss": 0.0327,
      "step": 3191
    },
    {
      "epoch": 0.36944444444444446,
      "grad_norm": 0.1849355399608612,
      "learning_rate": 1.2611111111111113e-05,
      "loss": 0.0255,
      "step": 3192
    },
    {
      "epoch": 0.3695601851851852,
      "grad_norm": 0.1658085286617279,
      "learning_rate": 1.2608796296296297e-05,
      "loss": 0.0313,
      "step": 3193
    },
    {
      "epoch": 0.3696759259259259,
      "grad_norm": 139.7681121826172,
      "learning_rate": 1.2606481481481483e-05,
      "loss": 0.8518,
      "step": 3194
    },
    {
      "epoch": 0.3697916666666667,
      "grad_norm": 0.2537861168384552,
      "learning_rate": 1.2604166666666669e-05,
      "loss": 0.0418,
      "step": 3195
    },
    {
      "epoch": 0.3699074074074074,
      "grad_norm": 0.14681781828403473,
      "learning_rate": 1.2601851851851851e-05,
      "loss": 0.0283,
      "step": 3196
    },
    {
      "epoch": 0.37002314814814813,
      "grad_norm": 0.16721512377262115,
      "learning_rate": 1.2599537037037037e-05,
      "loss": 0.0323,
      "step": 3197
    },
    {
      "epoch": 0.3701388888888889,
      "grad_norm": 0.14127041399478912,
      "learning_rate": 1.2597222222222223e-05,
      "loss": 0.0263,
      "step": 3198
    },
    {
      "epoch": 0.37025462962962963,
      "grad_norm": 0.2036186009645462,
      "learning_rate": 1.2594907407407409e-05,
      "loss": 0.0349,
      "step": 3199
    },
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 0.22286547720432281,
      "learning_rate": 1.2592592592592593e-05,
      "loss": 0.0313,
      "step": 3200
    },
    {
      "epoch": 0.3704861111111111,
      "grad_norm": 0.19917891919612885,
      "learning_rate": 1.259027777777778e-05,
      "loss": 0.0389,
      "step": 3201
    },
    {
      "epoch": 0.37060185185185185,
      "grad_norm": 0.1638932079076767,
      "learning_rate": 1.2587962962962964e-05,
      "loss": 0.0307,
      "step": 3202
    },
    {
      "epoch": 0.37071759259259257,
      "grad_norm": 0.11625947058200836,
      "learning_rate": 1.2585648148148148e-05,
      "loss": 0.0215,
      "step": 3203
    },
    {
      "epoch": 0.37083333333333335,
      "grad_norm": 0.13909561932086945,
      "learning_rate": 1.2583333333333334e-05,
      "loss": 0.0266,
      "step": 3204
    },
    {
      "epoch": 0.37094907407407407,
      "grad_norm": 0.39964649081230164,
      "learning_rate": 1.258101851851852e-05,
      "loss": 0.0378,
      "step": 3205
    },
    {
      "epoch": 0.3710648148148148,
      "grad_norm": 0.12655122578144073,
      "learning_rate": 1.2578703703703706e-05,
      "loss": 0.0239,
      "step": 3206
    },
    {
      "epoch": 0.37118055555555557,
      "grad_norm": 0.19175447523593903,
      "learning_rate": 1.2576388888888892e-05,
      "loss": 0.0363,
      "step": 3207
    },
    {
      "epoch": 0.3712962962962963,
      "grad_norm": 0.27147993445396423,
      "learning_rate": 1.2574074074074074e-05,
      "loss": 0.0381,
      "step": 3208
    },
    {
      "epoch": 0.371412037037037,
      "grad_norm": 0.35259678959846497,
      "learning_rate": 1.257175925925926e-05,
      "loss": 0.0388,
      "step": 3209
    },
    {
      "epoch": 0.3715277777777778,
      "grad_norm": 0.1496116667985916,
      "learning_rate": 1.2569444444444446e-05,
      "loss": 0.028,
      "step": 3210
    },
    {
      "epoch": 0.3716435185185185,
      "grad_norm": 0.14975054562091827,
      "learning_rate": 1.256712962962963e-05,
      "loss": 0.0286,
      "step": 3211
    },
    {
      "epoch": 0.37175925925925923,
      "grad_norm": 0.17613434791564941,
      "learning_rate": 1.2564814814814816e-05,
      "loss": 0.0244,
      "step": 3212
    },
    {
      "epoch": 0.371875,
      "grad_norm": 5.97593879699707,
      "learning_rate": 1.2562500000000002e-05,
      "loss": 2.788,
      "step": 3213
    },
    {
      "epoch": 0.37199074074074073,
      "grad_norm": 0.7341095209121704,
      "learning_rate": 1.2560185185185185e-05,
      "loss": 0.0368,
      "step": 3214
    },
    {
      "epoch": 0.37210648148148145,
      "grad_norm": 0.14703483879566193,
      "learning_rate": 1.255787037037037e-05,
      "loss": 0.0282,
      "step": 3215
    },
    {
      "epoch": 0.37222222222222223,
      "grad_norm": 0.14833000302314758,
      "learning_rate": 1.2555555555555557e-05,
      "loss": 0.0278,
      "step": 3216
    },
    {
      "epoch": 0.37233796296296295,
      "grad_norm": 0.13904692232608795,
      "learning_rate": 1.2553240740740743e-05,
      "loss": 0.0265,
      "step": 3217
    },
    {
      "epoch": 0.37245370370370373,
      "grad_norm": 0.13411182165145874,
      "learning_rate": 1.2550925925925927e-05,
      "loss": 0.0255,
      "step": 3218
    },
    {
      "epoch": 0.37256944444444445,
      "grad_norm": 0.19069166481494904,
      "learning_rate": 1.2548611111111113e-05,
      "loss": 0.0354,
      "step": 3219
    },
    {
      "epoch": 0.3726851851851852,
      "grad_norm": 16.547630310058594,
      "learning_rate": 1.2546296296296297e-05,
      "loss": 0.1269,
      "step": 3220
    },
    {
      "epoch": 0.37280092592592595,
      "grad_norm": 0.5782328844070435,
      "learning_rate": 1.2543981481481481e-05,
      "loss": 0.0352,
      "step": 3221
    },
    {
      "epoch": 0.3729166666666667,
      "grad_norm": 0.16888520121574402,
      "learning_rate": 1.2541666666666667e-05,
      "loss": 0.0319,
      "step": 3222
    },
    {
      "epoch": 0.3730324074074074,
      "grad_norm": 8.847467422485352,
      "learning_rate": 1.2539351851851853e-05,
      "loss": 2.6376,
      "step": 3223
    },
    {
      "epoch": 0.3731481481481482,
      "grad_norm": 0.11552399396896362,
      "learning_rate": 1.2537037037037039e-05,
      "loss": 0.0221,
      "step": 3224
    },
    {
      "epoch": 0.3732638888888889,
      "grad_norm": 0.1264313906431198,
      "learning_rate": 1.2534722222222225e-05,
      "loss": 0.0236,
      "step": 3225
    },
    {
      "epoch": 0.3733796296296296,
      "grad_norm": 1.875192403793335,
      "learning_rate": 1.2532407407407408e-05,
      "loss": 0.0381,
      "step": 3226
    },
    {
      "epoch": 0.3734953703703704,
      "grad_norm": 0.13935141265392303,
      "learning_rate": 1.2530092592592593e-05,
      "loss": 0.0247,
      "step": 3227
    },
    {
      "epoch": 0.3736111111111111,
      "grad_norm": 0.1876782774925232,
      "learning_rate": 1.252777777777778e-05,
      "loss": 0.0251,
      "step": 3228
    },
    {
      "epoch": 0.37372685185185184,
      "grad_norm": 0.12619124352931976,
      "learning_rate": 1.2525462962962964e-05,
      "loss": 0.0242,
      "step": 3229
    },
    {
      "epoch": 0.3738425925925926,
      "grad_norm": 130.91493225097656,
      "learning_rate": 1.252314814814815e-05,
      "loss": 0.877,
      "step": 3230
    },
    {
      "epoch": 0.37395833333333334,
      "grad_norm": 0.18426194787025452,
      "learning_rate": 1.2520833333333336e-05,
      "loss": 0.0353,
      "step": 3231
    },
    {
      "epoch": 0.37407407407407406,
      "grad_norm": 0.12176667898893356,
      "learning_rate": 1.2518518518518518e-05,
      "loss": 0.0232,
      "step": 3232
    },
    {
      "epoch": 0.37418981481481484,
      "grad_norm": 3.224433660507202,
      "learning_rate": 1.2516203703703704e-05,
      "loss": 0.0469,
      "step": 3233
    },
    {
      "epoch": 0.37430555555555556,
      "grad_norm": 0.18651799857616425,
      "learning_rate": 1.251388888888889e-05,
      "loss": 0.0286,
      "step": 3234
    },
    {
      "epoch": 0.3744212962962963,
      "grad_norm": 1.7157580852508545,
      "learning_rate": 1.2511574074074076e-05,
      "loss": 0.042,
      "step": 3235
    },
    {
      "epoch": 0.37453703703703706,
      "grad_norm": 0.22430995106697083,
      "learning_rate": 1.250925925925926e-05,
      "loss": 0.0376,
      "step": 3236
    },
    {
      "epoch": 0.3746527777777778,
      "grad_norm": 0.1316552460193634,
      "learning_rate": 1.2506944444444446e-05,
      "loss": 0.0254,
      "step": 3237
    },
    {
      "epoch": 0.3747685185185185,
      "grad_norm": 0.22552482783794403,
      "learning_rate": 1.250462962962963e-05,
      "loss": 0.0323,
      "step": 3238
    },
    {
      "epoch": 0.3748842592592593,
      "grad_norm": 0.12364331632852554,
      "learning_rate": 1.2502314814814815e-05,
      "loss": 0.022,
      "step": 3239
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.20158308744430542,
      "learning_rate": 1.25e-05,
      "loss": 0.0363,
      "step": 3240
    },
    {
      "epoch": 0.3751157407407407,
      "grad_norm": 0.17327594757080078,
      "learning_rate": 1.2497685185185187e-05,
      "loss": 0.0326,
      "step": 3241
    },
    {
      "epoch": 0.3752314814814815,
      "grad_norm": 0.16192908585071564,
      "learning_rate": 1.2495370370370372e-05,
      "loss": 0.0232,
      "step": 3242
    },
    {
      "epoch": 0.3753472222222222,
      "grad_norm": 0.1555384397506714,
      "learning_rate": 1.2493055555555555e-05,
      "loss": 0.0291,
      "step": 3243
    },
    {
      "epoch": 0.37546296296296294,
      "grad_norm": 0.17991141974925995,
      "learning_rate": 1.2490740740740741e-05,
      "loss": 0.0302,
      "step": 3244
    },
    {
      "epoch": 0.3755787037037037,
      "grad_norm": 2.9006896018981934,
      "learning_rate": 1.2488425925925927e-05,
      "loss": 0.0383,
      "step": 3245
    },
    {
      "epoch": 0.37569444444444444,
      "grad_norm": 80.01945495605469,
      "learning_rate": 1.2486111111111113e-05,
      "loss": 1.7233,
      "step": 3246
    },
    {
      "epoch": 0.37581018518518516,
      "grad_norm": 0.18713927268981934,
      "learning_rate": 1.2483796296296297e-05,
      "loss": 0.0353,
      "step": 3247
    },
    {
      "epoch": 0.37592592592592594,
      "grad_norm": 0.16212056577205658,
      "learning_rate": 1.2481481481481483e-05,
      "loss": 0.0295,
      "step": 3248
    },
    {
      "epoch": 0.37604166666666666,
      "grad_norm": 0.16076228022575378,
      "learning_rate": 1.2479166666666667e-05,
      "loss": 0.0301,
      "step": 3249
    },
    {
      "epoch": 0.3761574074074074,
      "grad_norm": 0.15528331696987152,
      "learning_rate": 1.2476851851851852e-05,
      "loss": 0.0288,
      "step": 3250
    },
    {
      "epoch": 0.37627314814814816,
      "grad_norm": 0.10636243969202042,
      "learning_rate": 1.2474537037037038e-05,
      "loss": 0.0205,
      "step": 3251
    },
    {
      "epoch": 0.3763888888888889,
      "grad_norm": 10.167698860168457,
      "learning_rate": 1.2472222222222223e-05,
      "loss": 0.0528,
      "step": 3252
    },
    {
      "epoch": 0.3765046296296296,
      "grad_norm": 0.15522979199886322,
      "learning_rate": 1.246990740740741e-05,
      "loss": 0.0214,
      "step": 3253
    },
    {
      "epoch": 0.3766203703703704,
      "grad_norm": 0.19026461243629456,
      "learning_rate": 1.2467592592592594e-05,
      "loss": 0.031,
      "step": 3254
    },
    {
      "epoch": 0.3767361111111111,
      "grad_norm": 0.11431366950273514,
      "learning_rate": 1.2465277777777778e-05,
      "loss": 0.0216,
      "step": 3255
    },
    {
      "epoch": 0.3768518518518518,
      "grad_norm": 0.18867826461791992,
      "learning_rate": 1.2462962962962964e-05,
      "loss": 0.0341,
      "step": 3256
    },
    {
      "epoch": 0.3769675925925926,
      "grad_norm": 0.3391318917274475,
      "learning_rate": 1.2460648148148148e-05,
      "loss": 0.0364,
      "step": 3257
    },
    {
      "epoch": 0.3770833333333333,
      "grad_norm": 70.27912139892578,
      "learning_rate": 1.2458333333333334e-05,
      "loss": 1.9518,
      "step": 3258
    },
    {
      "epoch": 0.37719907407407405,
      "grad_norm": 0.17473819851875305,
      "learning_rate": 1.245601851851852e-05,
      "loss": 0.0289,
      "step": 3259
    },
    {
      "epoch": 0.3773148148148148,
      "grad_norm": 0.20530153810977936,
      "learning_rate": 1.2453703703703706e-05,
      "loss": 0.0303,
      "step": 3260
    },
    {
      "epoch": 0.37743055555555555,
      "grad_norm": 1.2688590288162231,
      "learning_rate": 1.2451388888888888e-05,
      "loss": 0.0398,
      "step": 3261
    },
    {
      "epoch": 0.37754629629629627,
      "grad_norm": 0.2084856927394867,
      "learning_rate": 1.2449074074074074e-05,
      "loss": 0.0297,
      "step": 3262
    },
    {
      "epoch": 0.37766203703703705,
      "grad_norm": 0.13428179919719696,
      "learning_rate": 1.244675925925926e-05,
      "loss": 0.0256,
      "step": 3263
    },
    {
      "epoch": 0.37777777777777777,
      "grad_norm": 0.12406358867883682,
      "learning_rate": 1.2444444444444446e-05,
      "loss": 0.0238,
      "step": 3264
    },
    {
      "epoch": 0.37789351851851855,
      "grad_norm": 0.11791058629751205,
      "learning_rate": 1.244212962962963e-05,
      "loss": 0.0225,
      "step": 3265
    },
    {
      "epoch": 0.37800925925925927,
      "grad_norm": 0.18544749915599823,
      "learning_rate": 1.2439814814814817e-05,
      "loss": 0.0348,
      "step": 3266
    },
    {
      "epoch": 0.378125,
      "grad_norm": 0.18739968538284302,
      "learning_rate": 1.24375e-05,
      "loss": 0.0365,
      "step": 3267
    },
    {
      "epoch": 0.37824074074074077,
      "grad_norm": 0.15457755327224731,
      "learning_rate": 1.2435185185185185e-05,
      "loss": 0.0287,
      "step": 3268
    },
    {
      "epoch": 0.3783564814814815,
      "grad_norm": 0.14525127410888672,
      "learning_rate": 1.2432870370370371e-05,
      "loss": 0.0268,
      "step": 3269
    },
    {
      "epoch": 0.3784722222222222,
      "grad_norm": 3.1981523036956787,
      "learning_rate": 1.2430555555555557e-05,
      "loss": 0.0473,
      "step": 3270
    },
    {
      "epoch": 0.378587962962963,
      "grad_norm": 1.2267533540725708,
      "learning_rate": 1.2428240740740743e-05,
      "loss": 0.0404,
      "step": 3271
    },
    {
      "epoch": 0.3787037037037037,
      "grad_norm": 0.22906531393527985,
      "learning_rate": 1.2425925925925927e-05,
      "loss": 0.0292,
      "step": 3272
    },
    {
      "epoch": 0.37881944444444443,
      "grad_norm": 0.1647912561893463,
      "learning_rate": 1.2423611111111111e-05,
      "loss": 0.0297,
      "step": 3273
    },
    {
      "epoch": 0.3789351851851852,
      "grad_norm": 0.15153883397579193,
      "learning_rate": 1.2421296296296297e-05,
      "loss": 0.0209,
      "step": 3274
    },
    {
      "epoch": 0.37905092592592593,
      "grad_norm": 0.1635485291481018,
      "learning_rate": 1.2418981481481483e-05,
      "loss": 0.0292,
      "step": 3275
    },
    {
      "epoch": 0.37916666666666665,
      "grad_norm": 0.11527993530035019,
      "learning_rate": 1.2416666666666667e-05,
      "loss": 0.0206,
      "step": 3276
    },
    {
      "epoch": 0.37928240740740743,
      "grad_norm": 1.1006742715835571,
      "learning_rate": 1.2414351851851853e-05,
      "loss": 0.0312,
      "step": 3277
    },
    {
      "epoch": 0.37939814814814815,
      "grad_norm": 0.18555398285388947,
      "learning_rate": 1.241203703703704e-05,
      "loss": 0.0353,
      "step": 3278
    },
    {
      "epoch": 0.3795138888888889,
      "grad_norm": 24.395904541015625,
      "learning_rate": 1.2409722222222222e-05,
      "loss": 2.4837,
      "step": 3279
    },
    {
      "epoch": 0.37962962962962965,
      "grad_norm": 0.16014468669891357,
      "learning_rate": 1.2407407407407408e-05,
      "loss": 0.0265,
      "step": 3280
    },
    {
      "epoch": 0.37974537037037037,
      "grad_norm": 0.188707634806633,
      "learning_rate": 1.2405092592592594e-05,
      "loss": 0.033,
      "step": 3281
    },
    {
      "epoch": 0.3798611111111111,
      "grad_norm": 0.45619556307792664,
      "learning_rate": 1.240277777777778e-05,
      "loss": 0.0367,
      "step": 3282
    },
    {
      "epoch": 0.37997685185185187,
      "grad_norm": 0.21126261353492737,
      "learning_rate": 1.2400462962962964e-05,
      "loss": 0.0271,
      "step": 3283
    },
    {
      "epoch": 0.3800925925925926,
      "grad_norm": 0.14826299250125885,
      "learning_rate": 1.239814814814815e-05,
      "loss": 0.027,
      "step": 3284
    },
    {
      "epoch": 0.3802083333333333,
      "grad_norm": 0.9043852090835571,
      "learning_rate": 1.2395833333333334e-05,
      "loss": 0.0367,
      "step": 3285
    },
    {
      "epoch": 0.3803240740740741,
      "grad_norm": 0.1504049301147461,
      "learning_rate": 1.2393518518518518e-05,
      "loss": 0.0282,
      "step": 3286
    },
    {
      "epoch": 0.3804398148148148,
      "grad_norm": 0.12072136998176575,
      "learning_rate": 1.2391203703703704e-05,
      "loss": 0.0229,
      "step": 3287
    },
    {
      "epoch": 0.38055555555555554,
      "grad_norm": 0.15118396282196045,
      "learning_rate": 1.238888888888889e-05,
      "loss": 0.0209,
      "step": 3288
    },
    {
      "epoch": 0.3806712962962963,
      "grad_norm": 0.5755283236503601,
      "learning_rate": 1.2386574074074076e-05,
      "loss": 0.0361,
      "step": 3289
    },
    {
      "epoch": 0.38078703703703703,
      "grad_norm": 0.17773915827274323,
      "learning_rate": 1.2384259259259259e-05,
      "loss": 0.0297,
      "step": 3290
    },
    {
      "epoch": 0.38090277777777776,
      "grad_norm": 0.13731133937835693,
      "learning_rate": 1.2381944444444445e-05,
      "loss": 0.026,
      "step": 3291
    },
    {
      "epoch": 0.38101851851851853,
      "grad_norm": 0.1169925183057785,
      "learning_rate": 1.237962962962963e-05,
      "loss": 0.0224,
      "step": 3292
    },
    {
      "epoch": 0.38113425925925926,
      "grad_norm": 40.95093536376953,
      "learning_rate": 1.2377314814814817e-05,
      "loss": 1.7119,
      "step": 3293
    },
    {
      "epoch": 0.38125,
      "grad_norm": 0.1575021743774414,
      "learning_rate": 1.2375000000000001e-05,
      "loss": 0.0286,
      "step": 3294
    },
    {
      "epoch": 0.38136574074074076,
      "grad_norm": 0.24495922029018402,
      "learning_rate": 1.2372685185185187e-05,
      "loss": 0.0376,
      "step": 3295
    },
    {
      "epoch": 0.3814814814814815,
      "grad_norm": 0.1527138203382492,
      "learning_rate": 1.2370370370370371e-05,
      "loss": 0.0283,
      "step": 3296
    },
    {
      "epoch": 0.3815972222222222,
      "grad_norm": 0.2527535557746887,
      "learning_rate": 1.2368055555555555e-05,
      "loss": 0.022,
      "step": 3297
    },
    {
      "epoch": 0.381712962962963,
      "grad_norm": 0.12235225737094879,
      "learning_rate": 1.2365740740740741e-05,
      "loss": 0.0218,
      "step": 3298
    },
    {
      "epoch": 0.3818287037037037,
      "grad_norm": 0.16287386417388916,
      "learning_rate": 1.2363425925925927e-05,
      "loss": 0.0297,
      "step": 3299
    },
    {
      "epoch": 0.3819444444444444,
      "grad_norm": 0.14803515374660492,
      "learning_rate": 1.2361111111111113e-05,
      "loss": 0.0277,
      "step": 3300
    },
    {
      "epoch": 0.3820601851851852,
      "grad_norm": 0.14665915071964264,
      "learning_rate": 1.2358796296296297e-05,
      "loss": 0.0275,
      "step": 3301
    },
    {
      "epoch": 0.3821759259259259,
      "grad_norm": 0.14455662667751312,
      "learning_rate": 1.2356481481481482e-05,
      "loss": 0.0279,
      "step": 3302
    },
    {
      "epoch": 0.38229166666666664,
      "grad_norm": 0.33044037222862244,
      "learning_rate": 1.2354166666666668e-05,
      "loss": 0.0312,
      "step": 3303
    },
    {
      "epoch": 0.3824074074074074,
      "grad_norm": 0.10967472195625305,
      "learning_rate": 1.2351851851851852e-05,
      "loss": 0.0208,
      "step": 3304
    },
    {
      "epoch": 0.38252314814814814,
      "grad_norm": 0.18631824851036072,
      "learning_rate": 1.2349537037037038e-05,
      "loss": 0.0348,
      "step": 3305
    },
    {
      "epoch": 0.38263888888888886,
      "grad_norm": 0.15300902724266052,
      "learning_rate": 1.2347222222222224e-05,
      "loss": 0.0268,
      "step": 3306
    },
    {
      "epoch": 0.38275462962962964,
      "grad_norm": 0.2625461220741272,
      "learning_rate": 1.234490740740741e-05,
      "loss": 0.0231,
      "step": 3307
    },
    {
      "epoch": 0.38287037037037036,
      "grad_norm": 0.6342430114746094,
      "learning_rate": 1.2342592592592592e-05,
      "loss": 0.034,
      "step": 3308
    },
    {
      "epoch": 0.38298611111111114,
      "grad_norm": 0.10350073128938675,
      "learning_rate": 1.2340277777777778e-05,
      "loss": 0.0198,
      "step": 3309
    },
    {
      "epoch": 0.38310185185185186,
      "grad_norm": 0.7581238746643066,
      "learning_rate": 1.2337962962962964e-05,
      "loss": 0.0299,
      "step": 3310
    },
    {
      "epoch": 0.3832175925925926,
      "grad_norm": 0.1226256713271141,
      "learning_rate": 1.233564814814815e-05,
      "loss": 0.0233,
      "step": 3311
    },
    {
      "epoch": 0.38333333333333336,
      "grad_norm": 0.16768686473369598,
      "learning_rate": 1.2333333333333334e-05,
      "loss": 0.0322,
      "step": 3312
    },
    {
      "epoch": 0.3834490740740741,
      "grad_norm": 0.3790021538734436,
      "learning_rate": 1.233101851851852e-05,
      "loss": 0.0355,
      "step": 3313
    },
    {
      "epoch": 0.3835648148148148,
      "grad_norm": 95.93492889404297,
      "learning_rate": 1.2328703703703705e-05,
      "loss": 1.6139,
      "step": 3314
    },
    {
      "epoch": 0.3836805555555556,
      "grad_norm": 0.191779226064682,
      "learning_rate": 1.2326388888888889e-05,
      "loss": 0.0369,
      "step": 3315
    },
    {
      "epoch": 0.3837962962962963,
      "grad_norm": 10.092142105102539,
      "learning_rate": 1.2324074074074075e-05,
      "loss": 0.0701,
      "step": 3316
    },
    {
      "epoch": 0.383912037037037,
      "grad_norm": 0.126723051071167,
      "learning_rate": 1.232175925925926e-05,
      "loss": 0.0234,
      "step": 3317
    },
    {
      "epoch": 0.3840277777777778,
      "grad_norm": 0.4848823845386505,
      "learning_rate": 1.2319444444444447e-05,
      "loss": 0.0352,
      "step": 3318
    },
    {
      "epoch": 0.3841435185185185,
      "grad_norm": 0.1697705239057541,
      "learning_rate": 1.2317129629629631e-05,
      "loss": 0.0329,
      "step": 3319
    },
    {
      "epoch": 0.38425925925925924,
      "grad_norm": 0.12061100453138351,
      "learning_rate": 1.2314814814814815e-05,
      "loss": 0.0229,
      "step": 3320
    },
    {
      "epoch": 0.384375,
      "grad_norm": 0.11026040464639664,
      "learning_rate": 1.2312500000000001e-05,
      "loss": 0.0211,
      "step": 3321
    },
    {
      "epoch": 0.38449074074074074,
      "grad_norm": 0.14371436834335327,
      "learning_rate": 1.2310185185185185e-05,
      "loss": 0.0269,
      "step": 3322
    },
    {
      "epoch": 0.38460648148148147,
      "grad_norm": 0.14127343893051147,
      "learning_rate": 1.2307870370370371e-05,
      "loss": 0.0256,
      "step": 3323
    },
    {
      "epoch": 0.38472222222222224,
      "grad_norm": 0.13215860724449158,
      "learning_rate": 1.2305555555555557e-05,
      "loss": 0.0249,
      "step": 3324
    },
    {
      "epoch": 0.38483796296296297,
      "grad_norm": 0.18035681545734406,
      "learning_rate": 1.2303240740740743e-05,
      "loss": 0.0291,
      "step": 3325
    },
    {
      "epoch": 0.3849537037037037,
      "grad_norm": 0.13181860744953156,
      "learning_rate": 1.2300925925925926e-05,
      "loss": 0.0247,
      "step": 3326
    },
    {
      "epoch": 0.38506944444444446,
      "grad_norm": 0.18659067153930664,
      "learning_rate": 1.2298611111111112e-05,
      "loss": 0.0249,
      "step": 3327
    },
    {
      "epoch": 0.3851851851851852,
      "grad_norm": 0.1276492029428482,
      "learning_rate": 1.2296296296296298e-05,
      "loss": 0.0241,
      "step": 3328
    },
    {
      "epoch": 0.3853009259259259,
      "grad_norm": 0.131596177816391,
      "learning_rate": 1.2293981481481484e-05,
      "loss": 0.0247,
      "step": 3329
    },
    {
      "epoch": 0.3854166666666667,
      "grad_norm": 6.140008449554443,
      "learning_rate": 1.2291666666666668e-05,
      "loss": 2.93,
      "step": 3330
    },
    {
      "epoch": 0.3855324074074074,
      "grad_norm": 0.11938472837209702,
      "learning_rate": 1.2289351851851852e-05,
      "loss": 0.0228,
      "step": 3331
    },
    {
      "epoch": 0.38564814814814813,
      "grad_norm": 3.067586660385132,
      "learning_rate": 1.2287037037037038e-05,
      "loss": 0.0474,
      "step": 3332
    },
    {
      "epoch": 0.3857638888888889,
      "grad_norm": 37.508731842041016,
      "learning_rate": 1.2284722222222222e-05,
      "loss": 2.1761,
      "step": 3333
    },
    {
      "epoch": 0.38587962962962963,
      "grad_norm": 95.31217956542969,
      "learning_rate": 1.2282407407407408e-05,
      "loss": 4.3877,
      "step": 3334
    },
    {
      "epoch": 0.38599537037037035,
      "grad_norm": 0.15322135388851166,
      "learning_rate": 1.2280092592592594e-05,
      "loss": 0.0288,
      "step": 3335
    },
    {
      "epoch": 0.3861111111111111,
      "grad_norm": 0.1643434315919876,
      "learning_rate": 1.227777777777778e-05,
      "loss": 0.0309,
      "step": 3336
    },
    {
      "epoch": 0.38622685185185185,
      "grad_norm": 0.17986246943473816,
      "learning_rate": 1.2275462962962963e-05,
      "loss": 0.0343,
      "step": 3337
    },
    {
      "epoch": 0.38634259259259257,
      "grad_norm": 0.1130123883485794,
      "learning_rate": 1.2273148148148149e-05,
      "loss": 0.0217,
      "step": 3338
    },
    {
      "epoch": 0.38645833333333335,
      "grad_norm": 0.14839529991149902,
      "learning_rate": 1.2270833333333335e-05,
      "loss": 0.0283,
      "step": 3339
    },
    {
      "epoch": 0.38657407407407407,
      "grad_norm": 45.79692459106445,
      "learning_rate": 1.2268518518518519e-05,
      "loss": 0.4442,
      "step": 3340
    },
    {
      "epoch": 0.3866898148148148,
      "grad_norm": 46.22412872314453,
      "learning_rate": 1.2266203703703705e-05,
      "loss": 0.2359,
      "step": 3341
    },
    {
      "epoch": 0.38680555555555557,
      "grad_norm": 71.64434051513672,
      "learning_rate": 1.226388888888889e-05,
      "loss": 1.791,
      "step": 3342
    },
    {
      "epoch": 0.3869212962962963,
      "grad_norm": 0.1951625496149063,
      "learning_rate": 1.2261574074074073e-05,
      "loss": 0.0281,
      "step": 3343
    },
    {
      "epoch": 0.387037037037037,
      "grad_norm": 0.11815585941076279,
      "learning_rate": 1.225925925925926e-05,
      "loss": 0.0224,
      "step": 3344
    },
    {
      "epoch": 0.3871527777777778,
      "grad_norm": 0.19489629566669464,
      "learning_rate": 1.2256944444444445e-05,
      "loss": 0.0352,
      "step": 3345
    },
    {
      "epoch": 0.3872685185185185,
      "grad_norm": 0.14421159029006958,
      "learning_rate": 1.2254629629629631e-05,
      "loss": 0.0278,
      "step": 3346
    },
    {
      "epoch": 0.38738425925925923,
      "grad_norm": 10.157666206359863,
      "learning_rate": 1.2252314814814817e-05,
      "loss": 2.5232,
      "step": 3347
    },
    {
      "epoch": 0.3875,
      "grad_norm": 0.16502262651920319,
      "learning_rate": 1.2250000000000001e-05,
      "loss": 0.027,
      "step": 3348
    },
    {
      "epoch": 0.38761574074074073,
      "grad_norm": 0.13224327564239502,
      "learning_rate": 1.2247685185185186e-05,
      "loss": 0.0247,
      "step": 3349
    },
    {
      "epoch": 0.38773148148148145,
      "grad_norm": 0.12471507489681244,
      "learning_rate": 1.2245370370370371e-05,
      "loss": 0.0231,
      "step": 3350
    },
    {
      "epoch": 0.38784722222222223,
      "grad_norm": 41.245452880859375,
      "learning_rate": 1.2243055555555556e-05,
      "loss": 2.2188,
      "step": 3351
    },
    {
      "epoch": 0.38796296296296295,
      "grad_norm": 17.917579650878906,
      "learning_rate": 1.2240740740740742e-05,
      "loss": 2.3693,
      "step": 3352
    },
    {
      "epoch": 0.38807870370370373,
      "grad_norm": 0.14285852015018463,
      "learning_rate": 1.2238425925925928e-05,
      "loss": 0.027,
      "step": 3353
    },
    {
      "epoch": 0.38819444444444445,
      "grad_norm": 0.12858504056930542,
      "learning_rate": 1.2236111111111114e-05,
      "loss": 0.0243,
      "step": 3354
    },
    {
      "epoch": 0.3883101851851852,
      "grad_norm": 0.15741325914859772,
      "learning_rate": 1.2233796296296296e-05,
      "loss": 0.0259,
      "step": 3355
    },
    {
      "epoch": 0.38842592592592595,
      "grad_norm": 0.16407997906208038,
      "learning_rate": 1.2231481481481482e-05,
      "loss": 0.0289,
      "step": 3356
    },
    {
      "epoch": 0.3885416666666667,
      "grad_norm": 0.1501406580209732,
      "learning_rate": 1.2229166666666668e-05,
      "loss": 0.0269,
      "step": 3357
    },
    {
      "epoch": 0.3886574074074074,
      "grad_norm": 0.17166443169116974,
      "learning_rate": 1.2226851851851852e-05,
      "loss": 0.0274,
      "step": 3358
    },
    {
      "epoch": 0.3887731481481482,
      "grad_norm": 0.13369104266166687,
      "learning_rate": 1.2224537037037038e-05,
      "loss": 0.0248,
      "step": 3359
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 0.1122368797659874,
      "learning_rate": 1.2222222222222224e-05,
      "loss": 0.0215,
      "step": 3360
    },
    {
      "epoch": 0.3890046296296296,
      "grad_norm": 0.23434674739837646,
      "learning_rate": 1.2219907407407407e-05,
      "loss": 0.0364,
      "step": 3361
    },
    {
      "epoch": 0.3891203703703704,
      "grad_norm": 0.1010361760854721,
      "learning_rate": 1.2217592592592593e-05,
      "loss": 0.0193,
      "step": 3362
    },
    {
      "epoch": 0.3892361111111111,
      "grad_norm": 0.11965444684028625,
      "learning_rate": 1.2215277777777779e-05,
      "loss": 0.0227,
      "step": 3363
    },
    {
      "epoch": 0.38935185185185184,
      "grad_norm": 38.620880126953125,
      "learning_rate": 1.2212962962962965e-05,
      "loss": 2.1589,
      "step": 3364
    },
    {
      "epoch": 0.3894675925925926,
      "grad_norm": 0.35412147641181946,
      "learning_rate": 1.221064814814815e-05,
      "loss": 0.0331,
      "step": 3365
    },
    {
      "epoch": 0.38958333333333334,
      "grad_norm": 0.1464681476354599,
      "learning_rate": 1.2208333333333335e-05,
      "loss": 0.0217,
      "step": 3366
    },
    {
      "epoch": 0.38969907407407406,
      "grad_norm": 0.1292485147714615,
      "learning_rate": 1.2206018518518519e-05,
      "loss": 0.0233,
      "step": 3367
    },
    {
      "epoch": 0.38981481481481484,
      "grad_norm": 0.11445209383964539,
      "learning_rate": 1.2203703703703705e-05,
      "loss": 0.0215,
      "step": 3368
    },
    {
      "epoch": 0.38993055555555556,
      "grad_norm": 0.25688284635543823,
      "learning_rate": 1.220138888888889e-05,
      "loss": 0.0311,
      "step": 3369
    },
    {
      "epoch": 0.3900462962962963,
      "grad_norm": 0.14820596575737,
      "learning_rate": 1.2199074074074075e-05,
      "loss": 0.0276,
      "step": 3370
    },
    {
      "epoch": 0.39016203703703706,
      "grad_norm": 16.58358383178711,
      "learning_rate": 1.2196759259259261e-05,
      "loss": 2.5316,
      "step": 3371
    },
    {
      "epoch": 0.3902777777777778,
      "grad_norm": 11.58109188079834,
      "learning_rate": 1.2194444444444447e-05,
      "loss": 0.0759,
      "step": 3372
    },
    {
      "epoch": 0.3903935185185185,
      "grad_norm": 31.6570987701416,
      "learning_rate": 1.219212962962963e-05,
      "loss": 2.7098,
      "step": 3373
    },
    {
      "epoch": 0.3905092592592593,
      "grad_norm": 28.633220672607422,
      "learning_rate": 1.2189814814814816e-05,
      "loss": 0.2329,
      "step": 3374
    },
    {
      "epoch": 0.390625,
      "grad_norm": 0.2502628564834595,
      "learning_rate": 1.2187500000000001e-05,
      "loss": 0.0371,
      "step": 3375
    },
    {
      "epoch": 0.3907407407407407,
      "grad_norm": 0.13262976706027985,
      "learning_rate": 1.2185185185185186e-05,
      "loss": 0.0246,
      "step": 3376
    },
    {
      "epoch": 0.3908564814814815,
      "grad_norm": 0.49420368671417236,
      "learning_rate": 1.2182870370370372e-05,
      "loss": 0.0425,
      "step": 3377
    },
    {
      "epoch": 0.3909722222222222,
      "grad_norm": 77.11500549316406,
      "learning_rate": 1.2180555555555556e-05,
      "loss": 0.2725,
      "step": 3378
    },
    {
      "epoch": 0.39108796296296294,
      "grad_norm": 0.1395280957221985,
      "learning_rate": 1.2178240740740742e-05,
      "loss": 0.0258,
      "step": 3379
    },
    {
      "epoch": 0.3912037037037037,
      "grad_norm": 0.1636636108160019,
      "learning_rate": 1.2175925925925926e-05,
      "loss": 0.0315,
      "step": 3380
    },
    {
      "epoch": 0.39131944444444444,
      "grad_norm": 0.1159677505493164,
      "learning_rate": 1.2173611111111112e-05,
      "loss": 0.0218,
      "step": 3381
    },
    {
      "epoch": 0.39143518518518516,
      "grad_norm": 0.13211925327777863,
      "learning_rate": 1.2171296296296298e-05,
      "loss": 0.0247,
      "step": 3382
    },
    {
      "epoch": 0.39155092592592594,
      "grad_norm": 0.14531077444553375,
      "learning_rate": 1.2168981481481484e-05,
      "loss": 0.0199,
      "step": 3383
    },
    {
      "epoch": 0.39166666666666666,
      "grad_norm": 0.22840602695941925,
      "learning_rate": 1.2166666666666667e-05,
      "loss": 0.0328,
      "step": 3384
    },
    {
      "epoch": 0.3917824074074074,
      "grad_norm": 0.19495487213134766,
      "learning_rate": 1.2164351851851852e-05,
      "loss": 0.0309,
      "step": 3385
    },
    {
      "epoch": 0.39189814814814816,
      "grad_norm": 10.628204345703125,
      "learning_rate": 1.2162037037037038e-05,
      "loss": 2.5075,
      "step": 3386
    },
    {
      "epoch": 0.3920138888888889,
      "grad_norm": 0.18777874112129211,
      "learning_rate": 1.2159722222222223e-05,
      "loss": 0.0299,
      "step": 3387
    },
    {
      "epoch": 0.3921296296296296,
      "grad_norm": 0.12180422991514206,
      "learning_rate": 1.2157407407407409e-05,
      "loss": 0.0228,
      "step": 3388
    },
    {
      "epoch": 0.3922453703703704,
      "grad_norm": 0.11236662417650223,
      "learning_rate": 1.2155092592592595e-05,
      "loss": 0.0213,
      "step": 3389
    },
    {
      "epoch": 0.3923611111111111,
      "grad_norm": 154.47735595703125,
      "learning_rate": 1.2152777777777777e-05,
      "loss": 1.1817,
      "step": 3390
    },
    {
      "epoch": 0.3924768518518518,
      "grad_norm": 14.133671760559082,
      "learning_rate": 1.2150462962962963e-05,
      "loss": 0.1021,
      "step": 3391
    },
    {
      "epoch": 0.3925925925925926,
      "grad_norm": 0.1222013458609581,
      "learning_rate": 1.2148148148148149e-05,
      "loss": 0.0235,
      "step": 3392
    },
    {
      "epoch": 0.3927083333333333,
      "grad_norm": 0.12303011119365692,
      "learning_rate": 1.2145833333333335e-05,
      "loss": 0.0228,
      "step": 3393
    },
    {
      "epoch": 0.39282407407407405,
      "grad_norm": 0.11917358636856079,
      "learning_rate": 1.2143518518518521e-05,
      "loss": 0.0225,
      "step": 3394
    },
    {
      "epoch": 0.3929398148148148,
      "grad_norm": 0.18507599830627441,
      "learning_rate": 1.2141203703703705e-05,
      "loss": 0.0358,
      "step": 3395
    },
    {
      "epoch": 0.39305555555555555,
      "grad_norm": 0.19649378955364227,
      "learning_rate": 1.213888888888889e-05,
      "loss": 0.0354,
      "step": 3396
    },
    {
      "epoch": 0.39317129629629627,
      "grad_norm": 0.4703890085220337,
      "learning_rate": 1.2136574074074075e-05,
      "loss": 0.0401,
      "step": 3397
    },
    {
      "epoch": 0.39328703703703705,
      "grad_norm": 19.622743606567383,
      "learning_rate": 1.213425925925926e-05,
      "loss": 2.2922,
      "step": 3398
    },
    {
      "epoch": 0.39340277777777777,
      "grad_norm": 0.24070002138614655,
      "learning_rate": 1.2131944444444446e-05,
      "loss": 0.0207,
      "step": 3399
    },
    {
      "epoch": 0.39351851851851855,
      "grad_norm": 0.18312762677669525,
      "learning_rate": 1.2129629629629631e-05,
      "loss": 0.0344,
      "step": 3400
    },
    {
      "epoch": 0.39363425925925927,
      "grad_norm": 0.1296682059764862,
      "learning_rate": 1.2127314814814817e-05,
      "loss": 0.0246,
      "step": 3401
    },
    {
      "epoch": 0.39375,
      "grad_norm": 0.10362731665372849,
      "learning_rate": 1.2125e-05,
      "loss": 0.0198,
      "step": 3402
    },
    {
      "epoch": 0.39386574074074077,
      "grad_norm": 0.14736215770244598,
      "learning_rate": 1.2122685185185186e-05,
      "loss": 0.0271,
      "step": 3403
    },
    {
      "epoch": 0.3939814814814815,
      "grad_norm": 0.1428939253091812,
      "learning_rate": 1.2120370370370372e-05,
      "loss": 0.0261,
      "step": 3404
    },
    {
      "epoch": 0.3940972222222222,
      "grad_norm": 0.6709968447685242,
      "learning_rate": 1.2118055555555556e-05,
      "loss": 0.0416,
      "step": 3405
    },
    {
      "epoch": 0.394212962962963,
      "grad_norm": 0.2672573924064636,
      "learning_rate": 1.2115740740740742e-05,
      "loss": 0.0309,
      "step": 3406
    },
    {
      "epoch": 0.3943287037037037,
      "grad_norm": 0.16009309887886047,
      "learning_rate": 1.2113425925925928e-05,
      "loss": 0.0302,
      "step": 3407
    },
    {
      "epoch": 0.39444444444444443,
      "grad_norm": 0.15439502894878387,
      "learning_rate": 1.211111111111111e-05,
      "loss": 0.0282,
      "step": 3408
    },
    {
      "epoch": 0.3945601851851852,
      "grad_norm": 0.17885951697826385,
      "learning_rate": 1.2108796296296296e-05,
      "loss": 0.021,
      "step": 3409
    },
    {
      "epoch": 0.39467592592592593,
      "grad_norm": 0.16247153282165527,
      "learning_rate": 1.2106481481481482e-05,
      "loss": 0.0282,
      "step": 3410
    },
    {
      "epoch": 0.39479166666666665,
      "grad_norm": 0.1436731219291687,
      "learning_rate": 1.2104166666666668e-05,
      "loss": 0.0268,
      "step": 3411
    },
    {
      "epoch": 0.39490740740740743,
      "grad_norm": 0.3147934079170227,
      "learning_rate": 1.2101851851851854e-05,
      "loss": 0.0282,
      "step": 3412
    },
    {
      "epoch": 0.39502314814814815,
      "grad_norm": 0.17736636102199554,
      "learning_rate": 1.2099537037037039e-05,
      "loss": 0.0259,
      "step": 3413
    },
    {
      "epoch": 0.3951388888888889,
      "grad_norm": 0.42576706409454346,
      "learning_rate": 1.2097222222222223e-05,
      "loss": 0.0234,
      "step": 3414
    },
    {
      "epoch": 0.39525462962962965,
      "grad_norm": 155.12338256835938,
      "learning_rate": 1.2094907407407409e-05,
      "loss": 0.4693,
      "step": 3415
    },
    {
      "epoch": 0.39537037037037037,
      "grad_norm": 0.23525765538215637,
      "learning_rate": 1.2092592592592593e-05,
      "loss": 0.0374,
      "step": 3416
    },
    {
      "epoch": 0.3954861111111111,
      "grad_norm": 0.16224326193332672,
      "learning_rate": 1.2090277777777779e-05,
      "loss": 0.0224,
      "step": 3417
    },
    {
      "epoch": 0.39560185185185187,
      "grad_norm": 0.10012472420930862,
      "learning_rate": 1.2087962962962965e-05,
      "loss": 0.0188,
      "step": 3418
    },
    {
      "epoch": 0.3957175925925926,
      "grad_norm": 0.8602274060249329,
      "learning_rate": 1.2085648148148151e-05,
      "loss": 0.0356,
      "step": 3419
    },
    {
      "epoch": 0.3958333333333333,
      "grad_norm": 0.17974163591861725,
      "learning_rate": 1.2083333333333333e-05,
      "loss": 0.0331,
      "step": 3420
    },
    {
      "epoch": 0.3959490740740741,
      "grad_norm": 101.69879913330078,
      "learning_rate": 1.208101851851852e-05,
      "loss": 1.5347,
      "step": 3421
    },
    {
      "epoch": 0.3960648148148148,
      "grad_norm": 0.16869738698005676,
      "learning_rate": 1.2078703703703705e-05,
      "loss": 0.0325,
      "step": 3422
    },
    {
      "epoch": 0.39618055555555554,
      "grad_norm": 0.1313008964061737,
      "learning_rate": 1.207638888888889e-05,
      "loss": 0.018,
      "step": 3423
    },
    {
      "epoch": 0.3962962962962963,
      "grad_norm": 0.11113105714321136,
      "learning_rate": 1.2074074074074075e-05,
      "loss": 0.0211,
      "step": 3424
    },
    {
      "epoch": 0.39641203703703703,
      "grad_norm": 1.166203498840332,
      "learning_rate": 1.207175925925926e-05,
      "loss": 0.0398,
      "step": 3425
    },
    {
      "epoch": 0.39652777777777776,
      "grad_norm": 0.10415355861186981,
      "learning_rate": 1.2069444444444444e-05,
      "loss": 0.0188,
      "step": 3426
    },
    {
      "epoch": 0.39664351851851853,
      "grad_norm": 0.17237773537635803,
      "learning_rate": 1.206712962962963e-05,
      "loss": 0.0324,
      "step": 3427
    },
    {
      "epoch": 0.39675925925925926,
      "grad_norm": 1.0652116537094116,
      "learning_rate": 1.2064814814814816e-05,
      "loss": 0.041,
      "step": 3428
    },
    {
      "epoch": 0.396875,
      "grad_norm": 0.10133902728557587,
      "learning_rate": 1.2062500000000002e-05,
      "loss": 0.0194,
      "step": 3429
    },
    {
      "epoch": 0.39699074074074076,
      "grad_norm": 0.14687751233577728,
      "learning_rate": 1.2060185185185188e-05,
      "loss": 0.0272,
      "step": 3430
    },
    {
      "epoch": 0.3971064814814815,
      "grad_norm": 0.16751548647880554,
      "learning_rate": 1.205787037037037e-05,
      "loss": 0.0323,
      "step": 3431
    },
    {
      "epoch": 0.3972222222222222,
      "grad_norm": 0.10676921904087067,
      "learning_rate": 1.2055555555555556e-05,
      "loss": 0.0205,
      "step": 3432
    },
    {
      "epoch": 0.397337962962963,
      "grad_norm": 66.88581848144531,
      "learning_rate": 1.2053240740740742e-05,
      "loss": 0.7456,
      "step": 3433
    },
    {
      "epoch": 0.3974537037037037,
      "grad_norm": 0.10544080287218094,
      "learning_rate": 1.2050925925925926e-05,
      "loss": 0.0201,
      "step": 3434
    },
    {
      "epoch": 0.3975694444444444,
      "grad_norm": 0.16609595715999603,
      "learning_rate": 1.2048611111111112e-05,
      "loss": 0.0304,
      "step": 3435
    },
    {
      "epoch": 0.3976851851851852,
      "grad_norm": 0.13759885728359222,
      "learning_rate": 1.2046296296296298e-05,
      "loss": 0.0247,
      "step": 3436
    },
    {
      "epoch": 0.3978009259259259,
      "grad_norm": 0.1489068567752838,
      "learning_rate": 1.2043981481481481e-05,
      "loss": 0.0273,
      "step": 3437
    },
    {
      "epoch": 0.39791666666666664,
      "grad_norm": 0.16822192072868347,
      "learning_rate": 1.2041666666666667e-05,
      "loss": 0.0298,
      "step": 3438
    },
    {
      "epoch": 0.3980324074074074,
      "grad_norm": 0.1676694005727768,
      "learning_rate": 1.2039351851851853e-05,
      "loss": 0.0321,
      "step": 3439
    },
    {
      "epoch": 0.39814814814814814,
      "grad_norm": 0.11867338418960571,
      "learning_rate": 1.2037037037037039e-05,
      "loss": 0.0222,
      "step": 3440
    },
    {
      "epoch": 0.39826388888888886,
      "grad_norm": 0.15064293146133423,
      "learning_rate": 1.2034722222222223e-05,
      "loss": 0.0228,
      "step": 3441
    },
    {
      "epoch": 0.39837962962962964,
      "grad_norm": 0.10587179660797119,
      "learning_rate": 1.2032407407407409e-05,
      "loss": 0.0203,
      "step": 3442
    },
    {
      "epoch": 0.39849537037037036,
      "grad_norm": 0.25316283106803894,
      "learning_rate": 1.2030092592592593e-05,
      "loss": 0.0225,
      "step": 3443
    },
    {
      "epoch": 0.39861111111111114,
      "grad_norm": 0.5122352838516235,
      "learning_rate": 1.2027777777777777e-05,
      "loss": 0.0309,
      "step": 3444
    },
    {
      "epoch": 0.39872685185185186,
      "grad_norm": 0.1261162906885147,
      "learning_rate": 1.2025462962962963e-05,
      "loss": 0.0238,
      "step": 3445
    },
    {
      "epoch": 0.3988425925925926,
      "grad_norm": 0.10834615677595139,
      "learning_rate": 1.202314814814815e-05,
      "loss": 0.0207,
      "step": 3446
    },
    {
      "epoch": 0.39895833333333336,
      "grad_norm": 0.16590498387813568,
      "learning_rate": 1.2020833333333335e-05,
      "loss": 0.0273,
      "step": 3447
    },
    {
      "epoch": 0.3990740740740741,
      "grad_norm": 0.16609303653240204,
      "learning_rate": 1.2018518518518521e-05,
      "loss": 0.03,
      "step": 3448
    },
    {
      "epoch": 0.3991898148148148,
      "grad_norm": 0.10976354032754898,
      "learning_rate": 1.2016203703703704e-05,
      "loss": 0.0208,
      "step": 3449
    },
    {
      "epoch": 0.3993055555555556,
      "grad_norm": 0.09673161804676056,
      "learning_rate": 1.201388888888889e-05,
      "loss": 0.0185,
      "step": 3450
    },
    {
      "epoch": 0.3994212962962963,
      "grad_norm": 0.09766847640275955,
      "learning_rate": 1.2011574074074076e-05,
      "loss": 0.0186,
      "step": 3451
    },
    {
      "epoch": 0.399537037037037,
      "grad_norm": 0.19251464307308197,
      "learning_rate": 1.200925925925926e-05,
      "loss": 0.0352,
      "step": 3452
    },
    {
      "epoch": 0.3996527777777778,
      "grad_norm": 0.22396418452262878,
      "learning_rate": 1.2006944444444446e-05,
      "loss": 0.0303,
      "step": 3453
    },
    {
      "epoch": 0.3997685185185185,
      "grad_norm": 0.14250585436820984,
      "learning_rate": 1.2004629629629632e-05,
      "loss": 0.0265,
      "step": 3454
    },
    {
      "epoch": 0.39988425925925924,
      "grad_norm": 0.1195259541273117,
      "learning_rate": 1.2002314814814814e-05,
      "loss": 0.0213,
      "step": 3455
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5077177286148071,
      "learning_rate": 1.2e-05,
      "loss": 0.0336,
      "step": 3456
    },
    {
      "epoch": 0.40011574074074074,
      "grad_norm": 8.78145980834961,
      "learning_rate": 1.1997685185185186e-05,
      "loss": 2.877,
      "step": 3457
    },
    {
      "epoch": 0.40023148148148147,
      "grad_norm": 0.20485302805900574,
      "learning_rate": 1.1995370370370372e-05,
      "loss": 0.0358,
      "step": 3458
    },
    {
      "epoch": 0.40034722222222224,
      "grad_norm": 0.11244083195924759,
      "learning_rate": 1.1993055555555556e-05,
      "loss": 0.0215,
      "step": 3459
    },
    {
      "epoch": 0.40046296296296297,
      "grad_norm": 0.12511582672595978,
      "learning_rate": 1.1990740740740742e-05,
      "loss": 0.0236,
      "step": 3460
    },
    {
      "epoch": 0.4005787037037037,
      "grad_norm": 0.14448565244674683,
      "learning_rate": 1.1988425925925927e-05,
      "loss": 0.0262,
      "step": 3461
    },
    {
      "epoch": 0.40069444444444446,
      "grad_norm": 0.22254355251789093,
      "learning_rate": 1.1986111111111111e-05,
      "loss": 0.0296,
      "step": 3462
    },
    {
      "epoch": 0.4008101851851852,
      "grad_norm": 0.12829414010047913,
      "learning_rate": 1.1983796296296297e-05,
      "loss": 0.0239,
      "step": 3463
    },
    {
      "epoch": 0.4009259259259259,
      "grad_norm": 0.18630920350551605,
      "learning_rate": 1.1981481481481483e-05,
      "loss": 0.0286,
      "step": 3464
    },
    {
      "epoch": 0.4010416666666667,
      "grad_norm": 15.403135299682617,
      "learning_rate": 1.1979166666666669e-05,
      "loss": 2.2426,
      "step": 3465
    },
    {
      "epoch": 0.4011574074074074,
      "grad_norm": 0.15627704560756683,
      "learning_rate": 1.1976851851851851e-05,
      "loss": 0.0216,
      "step": 3466
    },
    {
      "epoch": 0.40127314814814813,
      "grad_norm": 37.13679885864258,
      "learning_rate": 1.1974537037037037e-05,
      "loss": 0.3347,
      "step": 3467
    },
    {
      "epoch": 0.4013888888888889,
      "grad_norm": 0.15248212218284607,
      "learning_rate": 1.1972222222222223e-05,
      "loss": 0.0279,
      "step": 3468
    },
    {
      "epoch": 0.40150462962962963,
      "grad_norm": 0.14169202744960785,
      "learning_rate": 1.1969907407407409e-05,
      "loss": 0.0259,
      "step": 3469
    },
    {
      "epoch": 0.40162037037037035,
      "grad_norm": 0.12031707167625427,
      "learning_rate": 1.1967592592592593e-05,
      "loss": 0.0226,
      "step": 3470
    },
    {
      "epoch": 0.4017361111111111,
      "grad_norm": 0.12090405076742172,
      "learning_rate": 1.196527777777778e-05,
      "loss": 0.0222,
      "step": 3471
    },
    {
      "epoch": 0.40185185185185185,
      "grad_norm": 0.09817104041576385,
      "learning_rate": 1.1962962962962964e-05,
      "loss": 0.0188,
      "step": 3472
    },
    {
      "epoch": 0.40196759259259257,
      "grad_norm": 0.1936413198709488,
      "learning_rate": 1.1960648148148148e-05,
      "loss": 0.0348,
      "step": 3473
    },
    {
      "epoch": 0.40208333333333335,
      "grad_norm": 0.17835034430027008,
      "learning_rate": 1.1958333333333334e-05,
      "loss": 0.0269,
      "step": 3474
    },
    {
      "epoch": 0.40219907407407407,
      "grad_norm": 0.1496543288230896,
      "learning_rate": 1.195601851851852e-05,
      "loss": 0.0254,
      "step": 3475
    },
    {
      "epoch": 0.4023148148148148,
      "grad_norm": 0.1705833077430725,
      "learning_rate": 1.1953703703703706e-05,
      "loss": 0.0327,
      "step": 3476
    },
    {
      "epoch": 0.40243055555555557,
      "grad_norm": 0.09376850724220276,
      "learning_rate": 1.195138888888889e-05,
      "loss": 0.0179,
      "step": 3477
    },
    {
      "epoch": 0.4025462962962963,
      "grad_norm": 0.09660722315311432,
      "learning_rate": 1.1949074074074074e-05,
      "loss": 0.0183,
      "step": 3478
    },
    {
      "epoch": 0.402662037037037,
      "grad_norm": 0.10193108022212982,
      "learning_rate": 1.194675925925926e-05,
      "loss": 0.0193,
      "step": 3479
    },
    {
      "epoch": 0.4027777777777778,
      "grad_norm": 0.09342338144779205,
      "learning_rate": 1.1944444444444444e-05,
      "loss": 0.0179,
      "step": 3480
    },
    {
      "epoch": 0.4028935185185185,
      "grad_norm": 0.29991281032562256,
      "learning_rate": 1.194212962962963e-05,
      "loss": 0.0277,
      "step": 3481
    },
    {
      "epoch": 0.40300925925925923,
      "grad_norm": 15.420292854309082,
      "learning_rate": 1.1939814814814816e-05,
      "loss": 2.7211,
      "step": 3482
    },
    {
      "epoch": 0.403125,
      "grad_norm": 0.12775354087352753,
      "learning_rate": 1.1937500000000002e-05,
      "loss": 0.0237,
      "step": 3483
    },
    {
      "epoch": 0.40324074074074073,
      "grad_norm": 0.14465075731277466,
      "learning_rate": 1.1935185185185185e-05,
      "loss": 0.0269,
      "step": 3484
    },
    {
      "epoch": 0.40335648148148145,
      "grad_norm": 0.1185445562005043,
      "learning_rate": 1.193287037037037e-05,
      "loss": 0.0219,
      "step": 3485
    },
    {
      "epoch": 0.40347222222222223,
      "grad_norm": 1.6192775964736938,
      "learning_rate": 1.1930555555555557e-05,
      "loss": 0.0348,
      "step": 3486
    },
    {
      "epoch": 0.40358796296296295,
      "grad_norm": 0.14045919477939606,
      "learning_rate": 1.1928240740740743e-05,
      "loss": 0.0244,
      "step": 3487
    },
    {
      "epoch": 0.40370370370370373,
      "grad_norm": 0.10225389152765274,
      "learning_rate": 1.1925925925925927e-05,
      "loss": 0.0197,
      "step": 3488
    },
    {
      "epoch": 0.40381944444444445,
      "grad_norm": 0.11907748878002167,
      "learning_rate": 1.1923611111111113e-05,
      "loss": 0.0217,
      "step": 3489
    },
    {
      "epoch": 0.4039351851851852,
      "grad_norm": 0.14460362493991852,
      "learning_rate": 1.1921296296296297e-05,
      "loss": 0.0273,
      "step": 3490
    },
    {
      "epoch": 0.40405092592592595,
      "grad_norm": 0.14242178201675415,
      "learning_rate": 1.1918981481481481e-05,
      "loss": 0.0267,
      "step": 3491
    },
    {
      "epoch": 0.4041666666666667,
      "grad_norm": 1.5048742294311523,
      "learning_rate": 1.1916666666666667e-05,
      "loss": 0.0428,
      "step": 3492
    },
    {
      "epoch": 0.4042824074074074,
      "grad_norm": 0.11739566177129745,
      "learning_rate": 1.1914351851851853e-05,
      "loss": 0.0219,
      "step": 3493
    },
    {
      "epoch": 0.4043981481481482,
      "grad_norm": 0.5308757424354553,
      "learning_rate": 1.1912037037037039e-05,
      "loss": 0.031,
      "step": 3494
    },
    {
      "epoch": 0.4045138888888889,
      "grad_norm": 15.7310209274292,
      "learning_rate": 1.1909722222222223e-05,
      "loss": 0.1034,
      "step": 3495
    },
    {
      "epoch": 0.4046296296296296,
      "grad_norm": 0.15645039081573486,
      "learning_rate": 1.1907407407407408e-05,
      "loss": 0.0295,
      "step": 3496
    },
    {
      "epoch": 0.4047453703703704,
      "grad_norm": 2.707245111465454,
      "learning_rate": 1.1905092592592594e-05,
      "loss": 0.0354,
      "step": 3497
    },
    {
      "epoch": 0.4048611111111111,
      "grad_norm": 0.19166269898414612,
      "learning_rate": 1.190277777777778e-05,
      "loss": 0.0321,
      "step": 3498
    },
    {
      "epoch": 0.40497685185185184,
      "grad_norm": 0.19702795147895813,
      "learning_rate": 1.1900462962962964e-05,
      "loss": 0.0322,
      "step": 3499
    },
    {
      "epoch": 0.4050925925925926,
      "grad_norm": 0.12701648473739624,
      "learning_rate": 1.189814814814815e-05,
      "loss": 0.0237,
      "step": 3500
    },
    {
      "epoch": 0.40520833333333334,
      "grad_norm": 0.1270909160375595,
      "learning_rate": 1.1895833333333336e-05,
      "loss": 0.0235,
      "step": 3501
    },
    {
      "epoch": 0.40532407407407406,
      "grad_norm": 0.16477163136005402,
      "learning_rate": 1.1893518518518518e-05,
      "loss": 0.0266,
      "step": 3502
    },
    {
      "epoch": 0.40543981481481484,
      "grad_norm": 0.11447259783744812,
      "learning_rate": 1.1891203703703704e-05,
      "loss": 0.0214,
      "step": 3503
    },
    {
      "epoch": 0.40555555555555556,
      "grad_norm": 0.09640930593013763,
      "learning_rate": 1.188888888888889e-05,
      "loss": 0.0181,
      "step": 3504
    },
    {
      "epoch": 0.4056712962962963,
      "grad_norm": 0.09669245779514313,
      "learning_rate": 1.1886574074074076e-05,
      "loss": 0.0181,
      "step": 3505
    },
    {
      "epoch": 0.40578703703703706,
      "grad_norm": 0.3547993004322052,
      "learning_rate": 1.188425925925926e-05,
      "loss": 0.029,
      "step": 3506
    },
    {
      "epoch": 0.4059027777777778,
      "grad_norm": 0.22452165186405182,
      "learning_rate": 1.1881944444444446e-05,
      "loss": 0.0279,
      "step": 3507
    },
    {
      "epoch": 0.4060185185185185,
      "grad_norm": 0.265412837266922,
      "learning_rate": 1.187962962962963e-05,
      "loss": 0.024,
      "step": 3508
    },
    {
      "epoch": 0.4061342592592593,
      "grad_norm": 0.15431685745716095,
      "learning_rate": 1.1877314814814815e-05,
      "loss": 0.0292,
      "step": 3509
    },
    {
      "epoch": 0.40625,
      "grad_norm": 0.11112255603075027,
      "learning_rate": 1.1875e-05,
      "loss": 0.0205,
      "step": 3510
    },
    {
      "epoch": 0.4063657407407407,
      "grad_norm": 0.13012725114822388,
      "learning_rate": 1.1872685185185187e-05,
      "loss": 0.0247,
      "step": 3511
    },
    {
      "epoch": 0.4064814814814815,
      "grad_norm": 0.16132061183452606,
      "learning_rate": 1.1870370370370373e-05,
      "loss": 0.0248,
      "step": 3512
    },
    {
      "epoch": 0.4065972222222222,
      "grad_norm": 0.1241980642080307,
      "learning_rate": 1.1868055555555555e-05,
      "loss": 0.0233,
      "step": 3513
    },
    {
      "epoch": 0.40671296296296294,
      "grad_norm": 18.085895538330078,
      "learning_rate": 1.1865740740740741e-05,
      "loss": 0.0693,
      "step": 3514
    },
    {
      "epoch": 0.4068287037037037,
      "grad_norm": 0.14150355756282806,
      "learning_rate": 1.1863425925925927e-05,
      "loss": 0.025,
      "step": 3515
    },
    {
      "epoch": 0.40694444444444444,
      "grad_norm": 0.1248411238193512,
      "learning_rate": 1.1861111111111113e-05,
      "loss": 0.0171,
      "step": 3516
    },
    {
      "epoch": 0.40706018518518516,
      "grad_norm": 1.5854175090789795,
      "learning_rate": 1.1858796296296297e-05,
      "loss": 0.0303,
      "step": 3517
    },
    {
      "epoch": 0.40717592592592594,
      "grad_norm": 0.14163966476917267,
      "learning_rate": 1.1856481481481483e-05,
      "loss": 0.0263,
      "step": 3518
    },
    {
      "epoch": 0.40729166666666666,
      "grad_norm": 0.14421434700489044,
      "learning_rate": 1.1854166666666667e-05,
      "loss": 0.0266,
      "step": 3519
    },
    {
      "epoch": 0.4074074074074074,
      "grad_norm": 223.00709533691406,
      "learning_rate": 1.1851851851851852e-05,
      "loss": 1.7604,
      "step": 3520
    },
    {
      "epoch": 0.40752314814814816,
      "grad_norm": 6.027671813964844,
      "learning_rate": 1.1849537037037038e-05,
      "loss": 2.6005,
      "step": 3521
    },
    {
      "epoch": 0.4076388888888889,
      "grad_norm": 0.13765347003936768,
      "learning_rate": 1.1847222222222224e-05,
      "loss": 0.0256,
      "step": 3522
    },
    {
      "epoch": 0.4077546296296296,
      "grad_norm": 0.14039573073387146,
      "learning_rate": 1.184490740740741e-05,
      "loss": 0.026,
      "step": 3523
    },
    {
      "epoch": 0.4078703703703704,
      "grad_norm": 0.14626650512218475,
      "learning_rate": 1.1842592592592594e-05,
      "loss": 0.0279,
      "step": 3524
    },
    {
      "epoch": 0.4079861111111111,
      "grad_norm": 0.1280493438243866,
      "learning_rate": 1.1840277777777778e-05,
      "loss": 0.0232,
      "step": 3525
    },
    {
      "epoch": 0.4081018518518518,
      "grad_norm": 0.12815238535404205,
      "learning_rate": 1.1837962962962964e-05,
      "loss": 0.0244,
      "step": 3526
    },
    {
      "epoch": 0.4082175925925926,
      "grad_norm": 0.09561118483543396,
      "learning_rate": 1.1835648148148148e-05,
      "loss": 0.0179,
      "step": 3527
    },
    {
      "epoch": 0.4083333333333333,
      "grad_norm": 0.10687538981437683,
      "learning_rate": 1.1833333333333334e-05,
      "loss": 0.02,
      "step": 3528
    },
    {
      "epoch": 0.40844907407407405,
      "grad_norm": 0.11969101428985596,
      "learning_rate": 1.183101851851852e-05,
      "loss": 0.0222,
      "step": 3529
    },
    {
      "epoch": 0.4085648148148148,
      "grad_norm": 0.13162349164485931,
      "learning_rate": 1.1828703703703706e-05,
      "loss": 0.0241,
      "step": 3530
    },
    {
      "epoch": 0.40868055555555555,
      "grad_norm": 0.0978696420788765,
      "learning_rate": 1.1826388888888889e-05,
      "loss": 0.0185,
      "step": 3531
    },
    {
      "epoch": 0.40879629629629627,
      "grad_norm": 0.1708829551935196,
      "learning_rate": 1.1824074074074074e-05,
      "loss": 0.0311,
      "step": 3532
    },
    {
      "epoch": 0.40891203703703705,
      "grad_norm": 0.3352418839931488,
      "learning_rate": 1.182175925925926e-05,
      "loss": 0.0337,
      "step": 3533
    },
    {
      "epoch": 0.40902777777777777,
      "grad_norm": 1.884989857673645,
      "learning_rate": 1.1819444444444446e-05,
      "loss": 0.0251,
      "step": 3534
    },
    {
      "epoch": 0.40914351851851855,
      "grad_norm": 0.314730167388916,
      "learning_rate": 1.181712962962963e-05,
      "loss": 0.0369,
      "step": 3535
    },
    {
      "epoch": 0.40925925925925927,
      "grad_norm": 0.16403040289878845,
      "learning_rate": 1.1814814814814817e-05,
      "loss": 0.0259,
      "step": 3536
    },
    {
      "epoch": 0.409375,
      "grad_norm": 0.1329251229763031,
      "learning_rate": 1.18125e-05,
      "loss": 0.0245,
      "step": 3537
    },
    {
      "epoch": 0.40949074074074077,
      "grad_norm": 0.12220767140388489,
      "learning_rate": 1.1810185185185185e-05,
      "loss": 0.023,
      "step": 3538
    },
    {
      "epoch": 0.4096064814814815,
      "grad_norm": 0.1334480494260788,
      "learning_rate": 1.1807870370370371e-05,
      "loss": 0.0248,
      "step": 3539
    },
    {
      "epoch": 0.4097222222222222,
      "grad_norm": 0.1440378874540329,
      "learning_rate": 1.1805555555555557e-05,
      "loss": 0.0277,
      "step": 3540
    },
    {
      "epoch": 0.409837962962963,
      "grad_norm": 0.11035812646150589,
      "learning_rate": 1.1803240740740743e-05,
      "loss": 0.0208,
      "step": 3541
    },
    {
      "epoch": 0.4099537037037037,
      "grad_norm": 0.13277024030685425,
      "learning_rate": 1.1800925925925927e-05,
      "loss": 0.0255,
      "step": 3542
    },
    {
      "epoch": 0.41006944444444443,
      "grad_norm": 0.16508184373378754,
      "learning_rate": 1.1798611111111111e-05,
      "loss": 0.0279,
      "step": 3543
    },
    {
      "epoch": 0.4101851851851852,
      "grad_norm": 5.3796539306640625,
      "learning_rate": 1.1796296296296297e-05,
      "loss": 2.8982,
      "step": 3544
    },
    {
      "epoch": 0.41030092592592593,
      "grad_norm": 0.12973572313785553,
      "learning_rate": 1.1793981481481482e-05,
      "loss": 0.0231,
      "step": 3545
    },
    {
      "epoch": 0.41041666666666665,
      "grad_norm": 0.1467776894569397,
      "learning_rate": 1.1791666666666668e-05,
      "loss": 0.0247,
      "step": 3546
    },
    {
      "epoch": 0.41053240740740743,
      "grad_norm": 0.15564648807048798,
      "learning_rate": 1.1789351851851853e-05,
      "loss": 0.0286,
      "step": 3547
    },
    {
      "epoch": 0.41064814814814815,
      "grad_norm": 0.13806504011154175,
      "learning_rate": 1.178703703703704e-05,
      "loss": 0.023,
      "step": 3548
    },
    {
      "epoch": 0.4107638888888889,
      "grad_norm": 0.19780436158180237,
      "learning_rate": 1.1784722222222222e-05,
      "loss": 0.0275,
      "step": 3549
    },
    {
      "epoch": 0.41087962962962965,
      "grad_norm": 0.1726953685283661,
      "learning_rate": 1.1782407407407408e-05,
      "loss": 0.0321,
      "step": 3550
    },
    {
      "epoch": 0.41099537037037037,
      "grad_norm": 3.547996997833252,
      "learning_rate": 1.1780092592592594e-05,
      "loss": 0.0658,
      "step": 3551
    },
    {
      "epoch": 0.4111111111111111,
      "grad_norm": 0.12506969273090363,
      "learning_rate": 1.177777777777778e-05,
      "loss": 0.0232,
      "step": 3552
    },
    {
      "epoch": 0.41122685185185187,
      "grad_norm": 0.9373804926872253,
      "learning_rate": 1.1775462962962964e-05,
      "loss": 0.0249,
      "step": 3553
    },
    {
      "epoch": 0.4113425925925926,
      "grad_norm": 0.09527339041233063,
      "learning_rate": 1.177314814814815e-05,
      "loss": 0.0178,
      "step": 3554
    },
    {
      "epoch": 0.4114583333333333,
      "grad_norm": 0.2153402864933014,
      "learning_rate": 1.1770833333333334e-05,
      "loss": 0.0289,
      "step": 3555
    },
    {
      "epoch": 0.4115740740740741,
      "grad_norm": 0.12958253920078278,
      "learning_rate": 1.1768518518518519e-05,
      "loss": 0.0249,
      "step": 3556
    },
    {
      "epoch": 0.4116898148148148,
      "grad_norm": 216.65184020996094,
      "learning_rate": 1.1766203703703704e-05,
      "loss": 1.3941,
      "step": 3557
    },
    {
      "epoch": 0.41180555555555554,
      "grad_norm": 0.32728105783462524,
      "learning_rate": 1.176388888888889e-05,
      "loss": 0.0337,
      "step": 3558
    },
    {
      "epoch": 0.4119212962962963,
      "grad_norm": 0.12000906467437744,
      "learning_rate": 1.1761574074074076e-05,
      "loss": 0.0222,
      "step": 3559
    },
    {
      "epoch": 0.41203703703703703,
      "grad_norm": 0.15593139827251434,
      "learning_rate": 1.1759259259259259e-05,
      "loss": 0.0299,
      "step": 3560
    },
    {
      "epoch": 0.41215277777777776,
      "grad_norm": 0.13960519433021545,
      "learning_rate": 1.1756944444444445e-05,
      "loss": 0.0259,
      "step": 3561
    },
    {
      "epoch": 0.41226851851851853,
      "grad_norm": 0.12993542850017548,
      "learning_rate": 1.175462962962963e-05,
      "loss": 0.0233,
      "step": 3562
    },
    {
      "epoch": 0.41238425925925926,
      "grad_norm": 0.14796526730060577,
      "learning_rate": 1.1752314814814815e-05,
      "loss": 0.0266,
      "step": 3563
    },
    {
      "epoch": 0.4125,
      "grad_norm": 0.09183237701654434,
      "learning_rate": 1.1750000000000001e-05,
      "loss": 0.0174,
      "step": 3564
    },
    {
      "epoch": 0.41261574074074076,
      "grad_norm": 0.17025531828403473,
      "learning_rate": 1.1747685185185187e-05,
      "loss": 0.0306,
      "step": 3565
    },
    {
      "epoch": 0.4127314814814815,
      "grad_norm": 0.09965270757675171,
      "learning_rate": 1.174537037037037e-05,
      "loss": 0.019,
      "step": 3566
    },
    {
      "epoch": 0.4128472222222222,
      "grad_norm": 0.20114843547344208,
      "learning_rate": 1.1743055555555555e-05,
      "loss": 0.0263,
      "step": 3567
    },
    {
      "epoch": 0.412962962962963,
      "grad_norm": 0.13894601166248322,
      "learning_rate": 1.1740740740740741e-05,
      "loss": 0.0253,
      "step": 3568
    },
    {
      "epoch": 0.4130787037037037,
      "grad_norm": 0.1328626573085785,
      "learning_rate": 1.1738425925925927e-05,
      "loss": 0.0244,
      "step": 3569
    },
    {
      "epoch": 0.4131944444444444,
      "grad_norm": 0.13910460472106934,
      "learning_rate": 1.1736111111111113e-05,
      "loss": 0.0266,
      "step": 3570
    },
    {
      "epoch": 0.4133101851851852,
      "grad_norm": 0.10660777240991592,
      "learning_rate": 1.1733796296296298e-05,
      "loss": 0.0203,
      "step": 3571
    },
    {
      "epoch": 0.4134259259259259,
      "grad_norm": 1.0694584846496582,
      "learning_rate": 1.1731481481481482e-05,
      "loss": 0.0365,
      "step": 3572
    },
    {
      "epoch": 0.41354166666666664,
      "grad_norm": 0.1019381508231163,
      "learning_rate": 1.1729166666666668e-05,
      "loss": 0.0194,
      "step": 3573
    },
    {
      "epoch": 0.4136574074074074,
      "grad_norm": 0.10981863737106323,
      "learning_rate": 1.1726851851851852e-05,
      "loss": 0.0206,
      "step": 3574
    },
    {
      "epoch": 0.41377314814814814,
      "grad_norm": 0.27061954140663147,
      "learning_rate": 1.1724537037037038e-05,
      "loss": 0.0281,
      "step": 3575
    },
    {
      "epoch": 0.41388888888888886,
      "grad_norm": 0.12097591906785965,
      "learning_rate": 1.1722222222222224e-05,
      "loss": 0.0226,
      "step": 3576
    },
    {
      "epoch": 0.41400462962962964,
      "grad_norm": 3.887836456298828,
      "learning_rate": 1.171990740740741e-05,
      "loss": 0.0506,
      "step": 3577
    },
    {
      "epoch": 0.41412037037037036,
      "grad_norm": 0.1041230782866478,
      "learning_rate": 1.1717592592592592e-05,
      "loss": 0.0192,
      "step": 3578
    },
    {
      "epoch": 0.41423611111111114,
      "grad_norm": 1.5124707221984863,
      "learning_rate": 1.1715277777777778e-05,
      "loss": 0.034,
      "step": 3579
    },
    {
      "epoch": 0.41435185185185186,
      "grad_norm": 0.16914114356040955,
      "learning_rate": 1.1712962962962964e-05,
      "loss": 0.0275,
      "step": 3580
    },
    {
      "epoch": 0.4144675925925926,
      "grad_norm": 0.09282569587230682,
      "learning_rate": 1.1710648148148149e-05,
      "loss": 0.0177,
      "step": 3581
    },
    {
      "epoch": 0.41458333333333336,
      "grad_norm": 0.11932025104761124,
      "learning_rate": 1.1708333333333334e-05,
      "loss": 0.0224,
      "step": 3582
    },
    {
      "epoch": 0.4146990740740741,
      "grad_norm": 0.12721247971057892,
      "learning_rate": 1.170601851851852e-05,
      "loss": 0.0232,
      "step": 3583
    },
    {
      "epoch": 0.4148148148148148,
      "grad_norm": 0.10501819103956223,
      "learning_rate": 1.1703703703703703e-05,
      "loss": 0.02,
      "step": 3584
    },
    {
      "epoch": 0.4149305555555556,
      "grad_norm": 0.10654677450656891,
      "learning_rate": 1.1701388888888889e-05,
      "loss": 0.0201,
      "step": 3585
    },
    {
      "epoch": 0.4150462962962963,
      "grad_norm": 0.16524222493171692,
      "learning_rate": 1.1699074074074075e-05,
      "loss": 0.0227,
      "step": 3586
    },
    {
      "epoch": 0.415162037037037,
      "grad_norm": 7.396703720092773,
      "learning_rate": 1.169675925925926e-05,
      "loss": 0.0646,
      "step": 3587
    },
    {
      "epoch": 0.4152777777777778,
      "grad_norm": 0.09235700219869614,
      "learning_rate": 1.1694444444444447e-05,
      "loss": 0.0173,
      "step": 3588
    },
    {
      "epoch": 0.4153935185185185,
      "grad_norm": 0.0977725237607956,
      "learning_rate": 1.1692129629629631e-05,
      "loss": 0.0182,
      "step": 3589
    },
    {
      "epoch": 0.41550925925925924,
      "grad_norm": 0.16026349365711212,
      "learning_rate": 1.1689814814814815e-05,
      "loss": 0.0309,
      "step": 3590
    },
    {
      "epoch": 0.415625,
      "grad_norm": 0.17524473369121552,
      "learning_rate": 1.1687500000000001e-05,
      "loss": 0.0322,
      "step": 3591
    },
    {
      "epoch": 0.41574074074074074,
      "grad_norm": 0.13707828521728516,
      "learning_rate": 1.1685185185185185e-05,
      "loss": 0.0245,
      "step": 3592
    },
    {
      "epoch": 0.41585648148148147,
      "grad_norm": 0.1301332265138626,
      "learning_rate": 1.1682870370370371e-05,
      "loss": 0.0244,
      "step": 3593
    },
    {
      "epoch": 0.41597222222222224,
      "grad_norm": 0.21149280667304993,
      "learning_rate": 1.1680555555555557e-05,
      "loss": 0.0283,
      "step": 3594
    },
    {
      "epoch": 0.41608796296296297,
      "grad_norm": 0.11764297634363174,
      "learning_rate": 1.1678240740740743e-05,
      "loss": 0.0217,
      "step": 3595
    },
    {
      "epoch": 0.4162037037037037,
      "grad_norm": 0.1393699198961258,
      "learning_rate": 1.1675925925925926e-05,
      "loss": 0.0267,
      "step": 3596
    },
    {
      "epoch": 0.41631944444444446,
      "grad_norm": 0.1457204967737198,
      "learning_rate": 1.1673611111111112e-05,
      "loss": 0.0272,
      "step": 3597
    },
    {
      "epoch": 0.4164351851851852,
      "grad_norm": 0.19825220108032227,
      "learning_rate": 1.1671296296296298e-05,
      "loss": 0.027,
      "step": 3598
    },
    {
      "epoch": 0.4165509259259259,
      "grad_norm": 0.1435527354478836,
      "learning_rate": 1.1668981481481482e-05,
      "loss": 0.0272,
      "step": 3599
    },
    {
      "epoch": 0.4166666666666667,
      "grad_norm": 0.1460312306880951,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 0.0278,
      "step": 3600
    },
    {
      "epoch": 0.4167824074074074,
      "grad_norm": 0.14160871505737305,
      "learning_rate": 1.1664351851851852e-05,
      "loss": 0.0241,
      "step": 3601
    },
    {
      "epoch": 0.41689814814814813,
      "grad_norm": 0.11463755369186401,
      "learning_rate": 1.1662037037037038e-05,
      "loss": 0.0204,
      "step": 3602
    },
    {
      "epoch": 0.4170138888888889,
      "grad_norm": 5.533949851989746,
      "learning_rate": 1.1659722222222222e-05,
      "loss": 0.032,
      "step": 3603
    },
    {
      "epoch": 0.41712962962962963,
      "grad_norm": 5.830709457397461,
      "learning_rate": 1.1657407407407408e-05,
      "loss": 0.0623,
      "step": 3604
    },
    {
      "epoch": 0.41724537037037035,
      "grad_norm": 0.7275494933128357,
      "learning_rate": 1.1655092592592594e-05,
      "loss": 0.0339,
      "step": 3605
    },
    {
      "epoch": 0.4173611111111111,
      "grad_norm": 0.16050830483436584,
      "learning_rate": 1.165277777777778e-05,
      "loss": 0.0209,
      "step": 3606
    },
    {
      "epoch": 0.41747685185185185,
      "grad_norm": 0.1275661289691925,
      "learning_rate": 1.1650462962962963e-05,
      "loss": 0.0175,
      "step": 3607
    },
    {
      "epoch": 0.41759259259259257,
      "grad_norm": 0.12170546501874924,
      "learning_rate": 1.1648148148148149e-05,
      "loss": 0.0225,
      "step": 3608
    },
    {
      "epoch": 0.41770833333333335,
      "grad_norm": 0.13571356236934662,
      "learning_rate": 1.1645833333333335e-05,
      "loss": 0.025,
      "step": 3609
    },
    {
      "epoch": 0.41782407407407407,
      "grad_norm": 78.07569885253906,
      "learning_rate": 1.1643518518518519e-05,
      "loss": 0.1998,
      "step": 3610
    },
    {
      "epoch": 0.4179398148148148,
      "grad_norm": 0.1377391219139099,
      "learning_rate": 1.1641203703703705e-05,
      "loss": 0.0264,
      "step": 3611
    },
    {
      "epoch": 0.41805555555555557,
      "grad_norm": 0.1376713663339615,
      "learning_rate": 1.163888888888889e-05,
      "loss": 0.0251,
      "step": 3612
    },
    {
      "epoch": 0.4181712962962963,
      "grad_norm": 48.38469696044922,
      "learning_rate": 1.1636574074074073e-05,
      "loss": 0.3644,
      "step": 3613
    },
    {
      "epoch": 0.418287037037037,
      "grad_norm": 192.68023681640625,
      "learning_rate": 1.163425925925926e-05,
      "loss": 0.7041,
      "step": 3614
    },
    {
      "epoch": 0.4184027777777778,
      "grad_norm": 0.08918274939060211,
      "learning_rate": 1.1631944444444445e-05,
      "loss": 0.017,
      "step": 3615
    },
    {
      "epoch": 0.4185185185185185,
      "grad_norm": 0.34389254450798035,
      "learning_rate": 1.1629629629629631e-05,
      "loss": 0.0282,
      "step": 3616
    },
    {
      "epoch": 0.41863425925925923,
      "grad_norm": 81.16632080078125,
      "learning_rate": 1.1627314814814817e-05,
      "loss": 2.2809,
      "step": 3617
    },
    {
      "epoch": 0.41875,
      "grad_norm": 0.13954828679561615,
      "learning_rate": 1.1625000000000001e-05,
      "loss": 0.0256,
      "step": 3618
    },
    {
      "epoch": 0.41886574074074073,
      "grad_norm": 0.10671398043632507,
      "learning_rate": 1.1622685185185186e-05,
      "loss": 0.0201,
      "step": 3619
    },
    {
      "epoch": 0.41898148148148145,
      "grad_norm": 25.756248474121094,
      "learning_rate": 1.1620370370370372e-05,
      "loss": 0.3004,
      "step": 3620
    },
    {
      "epoch": 0.41909722222222223,
      "grad_norm": 127.95409393310547,
      "learning_rate": 1.1618055555555556e-05,
      "loss": 1.8746,
      "step": 3621
    },
    {
      "epoch": 0.41921296296296295,
      "grad_norm": 48.24486541748047,
      "learning_rate": 1.1615740740740742e-05,
      "loss": 2.2882,
      "step": 3622
    },
    {
      "epoch": 0.41932870370370373,
      "grad_norm": 0.14067533612251282,
      "learning_rate": 1.1613425925925928e-05,
      "loss": 0.0265,
      "step": 3623
    },
    {
      "epoch": 0.41944444444444445,
      "grad_norm": 0.15782199800014496,
      "learning_rate": 1.1611111111111114e-05,
      "loss": 0.0194,
      "step": 3624
    },
    {
      "epoch": 0.4195601851851852,
      "grad_norm": 0.09367738664150238,
      "learning_rate": 1.1608796296296296e-05,
      "loss": 0.0171,
      "step": 3625
    },
    {
      "epoch": 0.41967592592592595,
      "grad_norm": 0.29612287878990173,
      "learning_rate": 1.1606481481481482e-05,
      "loss": 0.027,
      "step": 3626
    },
    {
      "epoch": 0.4197916666666667,
      "grad_norm": 0.13433770835399628,
      "learning_rate": 1.1604166666666668e-05,
      "loss": 0.025,
      "step": 3627
    },
    {
      "epoch": 0.4199074074074074,
      "grad_norm": 0.098833367228508,
      "learning_rate": 1.1601851851851852e-05,
      "loss": 0.0174,
      "step": 3628
    },
    {
      "epoch": 0.4200231481481482,
      "grad_norm": 0.15415191650390625,
      "learning_rate": 1.1599537037037038e-05,
      "loss": 0.0274,
      "step": 3629
    },
    {
      "epoch": 0.4201388888888889,
      "grad_norm": 0.12288477271795273,
      "learning_rate": 1.1597222222222224e-05,
      "loss": 0.0197,
      "step": 3630
    },
    {
      "epoch": 0.4202546296296296,
      "grad_norm": 0.10123122483491898,
      "learning_rate": 1.1594907407407407e-05,
      "loss": 0.0172,
      "step": 3631
    },
    {
      "epoch": 0.4203703703703704,
      "grad_norm": 0.11700685322284698,
      "learning_rate": 1.1592592592592593e-05,
      "loss": 0.0218,
      "step": 3632
    },
    {
      "epoch": 0.4204861111111111,
      "grad_norm": 0.1500682532787323,
      "learning_rate": 1.1590277777777779e-05,
      "loss": 0.0281,
      "step": 3633
    },
    {
      "epoch": 0.42060185185185184,
      "grad_norm": 0.13722911477088928,
      "learning_rate": 1.1587962962962965e-05,
      "loss": 0.0247,
      "step": 3634
    },
    {
      "epoch": 0.4207175925925926,
      "grad_norm": 1.0997689962387085,
      "learning_rate": 1.158564814814815e-05,
      "loss": 0.0253,
      "step": 3635
    },
    {
      "epoch": 0.42083333333333334,
      "grad_norm": 0.26888641715049744,
      "learning_rate": 1.1583333333333335e-05,
      "loss": 0.0269,
      "step": 3636
    },
    {
      "epoch": 0.42094907407407406,
      "grad_norm": 0.09145552664995193,
      "learning_rate": 1.1581018518518519e-05,
      "loss": 0.0171,
      "step": 3637
    },
    {
      "epoch": 0.42106481481481484,
      "grad_norm": 0.09447401016950607,
      "learning_rate": 1.1578703703703705e-05,
      "loss": 0.0172,
      "step": 3638
    },
    {
      "epoch": 0.42118055555555556,
      "grad_norm": 0.1388733834028244,
      "learning_rate": 1.157638888888889e-05,
      "loss": 0.0252,
      "step": 3639
    },
    {
      "epoch": 0.4212962962962963,
      "grad_norm": 0.12085297703742981,
      "learning_rate": 1.1574074074074075e-05,
      "loss": 0.0221,
      "step": 3640
    },
    {
      "epoch": 0.42141203703703706,
      "grad_norm": 0.19003359973430634,
      "learning_rate": 1.1571759259259261e-05,
      "loss": 0.0339,
      "step": 3641
    },
    {
      "epoch": 0.4215277777777778,
      "grad_norm": 0.149858757853508,
      "learning_rate": 1.1569444444444447e-05,
      "loss": 0.0273,
      "step": 3642
    },
    {
      "epoch": 0.4216435185185185,
      "grad_norm": 0.11363586783409119,
      "learning_rate": 1.156712962962963e-05,
      "loss": 0.0211,
      "step": 3643
    },
    {
      "epoch": 0.4217592592592593,
      "grad_norm": 0.1241905465722084,
      "learning_rate": 1.1564814814814816e-05,
      "loss": 0.0227,
      "step": 3644
    },
    {
      "epoch": 0.421875,
      "grad_norm": 0.10372120141983032,
      "learning_rate": 1.1562500000000002e-05,
      "loss": 0.0196,
      "step": 3645
    },
    {
      "epoch": 0.4219907407407407,
      "grad_norm": 0.12795206904411316,
      "learning_rate": 1.1560185185185186e-05,
      "loss": 0.0242,
      "step": 3646
    },
    {
      "epoch": 0.4221064814814815,
      "grad_norm": 0.15793724358081818,
      "learning_rate": 1.1557870370370372e-05,
      "loss": 0.0301,
      "step": 3647
    },
    {
      "epoch": 0.4222222222222222,
      "grad_norm": 0.1484634280204773,
      "learning_rate": 1.1555555555555556e-05,
      "loss": 0.0249,
      "step": 3648
    },
    {
      "epoch": 0.42233796296296294,
      "grad_norm": 0.1420699656009674,
      "learning_rate": 1.155324074074074e-05,
      "loss": 0.0252,
      "step": 3649
    },
    {
      "epoch": 0.4224537037037037,
      "grad_norm": 1.3388208150863647,
      "learning_rate": 1.1550925925925926e-05,
      "loss": 0.0306,
      "step": 3650
    },
    {
      "epoch": 0.42256944444444444,
      "grad_norm": 0.11044636368751526,
      "learning_rate": 1.1548611111111112e-05,
      "loss": 0.0175,
      "step": 3651
    },
    {
      "epoch": 0.42268518518518516,
      "grad_norm": 0.14189180731773376,
      "learning_rate": 1.1546296296296298e-05,
      "loss": 0.0264,
      "step": 3652
    },
    {
      "epoch": 0.42280092592592594,
      "grad_norm": 0.12596958875656128,
      "learning_rate": 1.1543981481481484e-05,
      "loss": 0.0233,
      "step": 3653
    },
    {
      "epoch": 0.42291666666666666,
      "grad_norm": 242.9991912841797,
      "learning_rate": 1.1541666666666667e-05,
      "loss": 1.2122,
      "step": 3654
    },
    {
      "epoch": 0.4230324074074074,
      "grad_norm": 0.140831857919693,
      "learning_rate": 1.1539351851851853e-05,
      "loss": 0.0265,
      "step": 3655
    },
    {
      "epoch": 0.42314814814814816,
      "grad_norm": 0.32680028676986694,
      "learning_rate": 1.1537037037037038e-05,
      "loss": 0.0335,
      "step": 3656
    },
    {
      "epoch": 0.4232638888888889,
      "grad_norm": 103.95042419433594,
      "learning_rate": 1.1534722222222223e-05,
      "loss": 0.2945,
      "step": 3657
    },
    {
      "epoch": 0.4233796296296296,
      "grad_norm": 97.15840148925781,
      "learning_rate": 1.1532407407407409e-05,
      "loss": 0.3308,
      "step": 3658
    },
    {
      "epoch": 0.4234953703703704,
      "grad_norm": 0.1286538988351822,
      "learning_rate": 1.1530092592592595e-05,
      "loss": 0.0242,
      "step": 3659
    },
    {
      "epoch": 0.4236111111111111,
      "grad_norm": 0.16379742324352264,
      "learning_rate": 1.1527777777777777e-05,
      "loss": 0.0248,
      "step": 3660
    },
    {
      "epoch": 0.4237268518518518,
      "grad_norm": 0.11236386746168137,
      "learning_rate": 1.1525462962962963e-05,
      "loss": 0.0209,
      "step": 3661
    },
    {
      "epoch": 0.4238425925925926,
      "grad_norm": 36.779850006103516,
      "learning_rate": 1.1523148148148149e-05,
      "loss": 0.1453,
      "step": 3662
    },
    {
      "epoch": 0.4239583333333333,
      "grad_norm": 0.13859188556671143,
      "learning_rate": 1.1520833333333335e-05,
      "loss": 0.0263,
      "step": 3663
    },
    {
      "epoch": 0.42407407407407405,
      "grad_norm": 1.0782549381256104,
      "learning_rate": 1.151851851851852e-05,
      "loss": 0.0326,
      "step": 3664
    },
    {
      "epoch": 0.4241898148148148,
      "grad_norm": 0.09038694947957993,
      "learning_rate": 1.1516203703703705e-05,
      "loss": 0.0169,
      "step": 3665
    },
    {
      "epoch": 0.42430555555555555,
      "grad_norm": 0.30127593874931335,
      "learning_rate": 1.151388888888889e-05,
      "loss": 0.019,
      "step": 3666
    },
    {
      "epoch": 0.42442129629629627,
      "grad_norm": 0.1334969401359558,
      "learning_rate": 1.1511574074074074e-05,
      "loss": 0.0247,
      "step": 3667
    },
    {
      "epoch": 0.42453703703703705,
      "grad_norm": 0.08769021928310394,
      "learning_rate": 1.150925925925926e-05,
      "loss": 0.0165,
      "step": 3668
    },
    {
      "epoch": 0.42465277777777777,
      "grad_norm": 0.4682551622390747,
      "learning_rate": 1.1506944444444446e-05,
      "loss": 0.0276,
      "step": 3669
    },
    {
      "epoch": 0.42476851851851855,
      "grad_norm": 0.15083123743534088,
      "learning_rate": 1.1504629629629632e-05,
      "loss": 0.0236,
      "step": 3670
    },
    {
      "epoch": 0.42488425925925927,
      "grad_norm": 0.0961604118347168,
      "learning_rate": 1.1502314814814817e-05,
      "loss": 0.0183,
      "step": 3671
    },
    {
      "epoch": 0.425,
      "grad_norm": 0.12592944502830505,
      "learning_rate": 1.15e-05,
      "loss": 0.0238,
      "step": 3672
    },
    {
      "epoch": 0.42511574074074077,
      "grad_norm": 0.17368562519550323,
      "learning_rate": 1.1497685185185186e-05,
      "loss": 0.0253,
      "step": 3673
    },
    {
      "epoch": 0.4252314814814815,
      "grad_norm": 34.49529266357422,
      "learning_rate": 1.1495370370370372e-05,
      "loss": 2.5228,
      "step": 3674
    },
    {
      "epoch": 0.4253472222222222,
      "grad_norm": 0.12611569464206696,
      "learning_rate": 1.1493055555555556e-05,
      "loss": 0.0237,
      "step": 3675
    },
    {
      "epoch": 0.425462962962963,
      "grad_norm": 0.12923386693000793,
      "learning_rate": 1.1490740740740742e-05,
      "loss": 0.0173,
      "step": 3676
    },
    {
      "epoch": 0.4255787037037037,
      "grad_norm": 0.12611083686351776,
      "learning_rate": 1.1488425925925928e-05,
      "loss": 0.0232,
      "step": 3677
    },
    {
      "epoch": 0.42569444444444443,
      "grad_norm": 12.964755058288574,
      "learning_rate": 1.148611111111111e-05,
      "loss": 2.6667,
      "step": 3678
    },
    {
      "epoch": 0.4258101851851852,
      "grad_norm": 0.1031770408153534,
      "learning_rate": 1.1483796296296297e-05,
      "loss": 0.0193,
      "step": 3679
    },
    {
      "epoch": 0.42592592592592593,
      "grad_norm": 7.947141170501709,
      "learning_rate": 1.1481481481481482e-05,
      "loss": 2.6892,
      "step": 3680
    },
    {
      "epoch": 0.42604166666666665,
      "grad_norm": 0.15465371310710907,
      "learning_rate": 1.1479166666666668e-05,
      "loss": 0.0281,
      "step": 3681
    },
    {
      "epoch": 0.42615740740740743,
      "grad_norm": 0.0870261937379837,
      "learning_rate": 1.1476851851851853e-05,
      "loss": 0.0164,
      "step": 3682
    },
    {
      "epoch": 0.42627314814814815,
      "grad_norm": 0.706671953201294,
      "learning_rate": 1.1474537037037039e-05,
      "loss": 0.0391,
      "step": 3683
    },
    {
      "epoch": 0.4263888888888889,
      "grad_norm": 0.10904386639595032,
      "learning_rate": 1.1472222222222223e-05,
      "loss": 0.0204,
      "step": 3684
    },
    {
      "epoch": 0.42650462962962965,
      "grad_norm": 0.12321996688842773,
      "learning_rate": 1.1469907407407407e-05,
      "loss": 0.0234,
      "step": 3685
    },
    {
      "epoch": 0.42662037037037037,
      "grad_norm": 0.1646163910627365,
      "learning_rate": 1.1467592592592593e-05,
      "loss": 0.0313,
      "step": 3686
    },
    {
      "epoch": 0.4267361111111111,
      "grad_norm": 0.08484850823879242,
      "learning_rate": 1.1465277777777779e-05,
      "loss": 0.0162,
      "step": 3687
    },
    {
      "epoch": 0.42685185185185187,
      "grad_norm": 1.665101170539856,
      "learning_rate": 1.1462962962962965e-05,
      "loss": 0.0323,
      "step": 3688
    },
    {
      "epoch": 0.4269675925925926,
      "grad_norm": 0.15746864676475525,
      "learning_rate": 1.1460648148148151e-05,
      "loss": 0.0278,
      "step": 3689
    },
    {
      "epoch": 0.4270833333333333,
      "grad_norm": 98.84065246582031,
      "learning_rate": 1.1458333333333333e-05,
      "loss": 0.7928,
      "step": 3690
    },
    {
      "epoch": 0.4271990740740741,
      "grad_norm": 16.472728729248047,
      "learning_rate": 1.145601851851852e-05,
      "loss": 0.0721,
      "step": 3691
    },
    {
      "epoch": 0.4273148148148148,
      "grad_norm": 0.1252252459526062,
      "learning_rate": 1.1453703703703705e-05,
      "loss": 0.023,
      "step": 3692
    },
    {
      "epoch": 0.42743055555555554,
      "grad_norm": 0.13915874063968658,
      "learning_rate": 1.145138888888889e-05,
      "loss": 0.0192,
      "step": 3693
    },
    {
      "epoch": 0.4275462962962963,
      "grad_norm": 0.13157141208648682,
      "learning_rate": 1.1449074074074076e-05,
      "loss": 0.0233,
      "step": 3694
    },
    {
      "epoch": 0.42766203703703703,
      "grad_norm": 17.143903732299805,
      "learning_rate": 1.144675925925926e-05,
      "loss": 2.5577,
      "step": 3695
    },
    {
      "epoch": 0.42777777777777776,
      "grad_norm": 2.5442488193511963,
      "learning_rate": 1.1444444444444444e-05,
      "loss": 0.0401,
      "step": 3696
    },
    {
      "epoch": 0.42789351851851853,
      "grad_norm": 2.2974355220794678,
      "learning_rate": 1.144212962962963e-05,
      "loss": 0.0395,
      "step": 3697
    },
    {
      "epoch": 0.42800925925925926,
      "grad_norm": 0.1185598373413086,
      "learning_rate": 1.1439814814814816e-05,
      "loss": 0.0222,
      "step": 3698
    },
    {
      "epoch": 0.428125,
      "grad_norm": 189.8660430908203,
      "learning_rate": 1.1437500000000002e-05,
      "loss": 1.4867,
      "step": 3699
    },
    {
      "epoch": 0.42824074074074076,
      "grad_norm": 0.15976180136203766,
      "learning_rate": 1.1435185185185186e-05,
      "loss": 0.0266,
      "step": 3700
    },
    {
      "epoch": 0.4283564814814815,
      "grad_norm": 0.13605929911136627,
      "learning_rate": 1.143287037037037e-05,
      "loss": 0.0254,
      "step": 3701
    },
    {
      "epoch": 0.4284722222222222,
      "grad_norm": 12.654335021972656,
      "learning_rate": 1.1430555555555556e-05,
      "loss": 2.9783,
      "step": 3702
    },
    {
      "epoch": 0.428587962962963,
      "grad_norm": 0.11795277893543243,
      "learning_rate": 1.142824074074074e-05,
      "loss": 0.0222,
      "step": 3703
    },
    {
      "epoch": 0.4287037037037037,
      "grad_norm": 0.3796110451221466,
      "learning_rate": 1.1425925925925927e-05,
      "loss": 0.0316,
      "step": 3704
    },
    {
      "epoch": 0.4288194444444444,
      "grad_norm": 0.09188617020845413,
      "learning_rate": 1.1423611111111112e-05,
      "loss": 0.0163,
      "step": 3705
    },
    {
      "epoch": 0.4289351851851852,
      "grad_norm": 0.10306905210018158,
      "learning_rate": 1.1421296296296298e-05,
      "loss": 0.0193,
      "step": 3706
    },
    {
      "epoch": 0.4290509259259259,
      "grad_norm": 0.17220176756381989,
      "learning_rate": 1.1418981481481481e-05,
      "loss": 0.0306,
      "step": 3707
    },
    {
      "epoch": 0.42916666666666664,
      "grad_norm": 1.8043930530548096,
      "learning_rate": 1.1416666666666667e-05,
      "loss": 0.032,
      "step": 3708
    },
    {
      "epoch": 0.4292824074074074,
      "grad_norm": 139.00653076171875,
      "learning_rate": 1.1414351851851853e-05,
      "loss": 1.755,
      "step": 3709
    },
    {
      "epoch": 0.42939814814814814,
      "grad_norm": 0.1505538821220398,
      "learning_rate": 1.1412037037037039e-05,
      "loss": 0.0283,
      "step": 3710
    },
    {
      "epoch": 0.42951388888888886,
      "grad_norm": 0.1262822449207306,
      "learning_rate": 1.1409722222222223e-05,
      "loss": 0.023,
      "step": 3711
    },
    {
      "epoch": 0.42962962962962964,
      "grad_norm": 0.11194702982902527,
      "learning_rate": 1.1407407407407409e-05,
      "loss": 0.021,
      "step": 3712
    },
    {
      "epoch": 0.42974537037037036,
      "grad_norm": 0.1501203030347824,
      "learning_rate": 1.1405092592592593e-05,
      "loss": 0.0249,
      "step": 3713
    },
    {
      "epoch": 0.42986111111111114,
      "grad_norm": 0.3026619553565979,
      "learning_rate": 1.1402777777777777e-05,
      "loss": 0.0319,
      "step": 3714
    },
    {
      "epoch": 0.42997685185185186,
      "grad_norm": 0.1330150067806244,
      "learning_rate": 1.1400462962962963e-05,
      "loss": 0.0241,
      "step": 3715
    },
    {
      "epoch": 0.4300925925925926,
      "grad_norm": 0.11294299364089966,
      "learning_rate": 1.139814814814815e-05,
      "loss": 0.0209,
      "step": 3716
    },
    {
      "epoch": 0.43020833333333336,
      "grad_norm": 0.19543826580047607,
      "learning_rate": 1.1395833333333335e-05,
      "loss": 0.0272,
      "step": 3717
    },
    {
      "epoch": 0.4303240740740741,
      "grad_norm": 0.13857188820838928,
      "learning_rate": 1.139351851851852e-05,
      "loss": 0.0237,
      "step": 3718
    },
    {
      "epoch": 0.4304398148148148,
      "grad_norm": 1.0826414823532104,
      "learning_rate": 1.1391203703703704e-05,
      "loss": 0.0315,
      "step": 3719
    },
    {
      "epoch": 0.4305555555555556,
      "grad_norm": 0.09882823377847672,
      "learning_rate": 1.138888888888889e-05,
      "loss": 0.0187,
      "step": 3720
    },
    {
      "epoch": 0.4306712962962963,
      "grad_norm": 0.1284961700439453,
      "learning_rate": 1.1386574074074076e-05,
      "loss": 0.0173,
      "step": 3721
    },
    {
      "epoch": 0.430787037037037,
      "grad_norm": 0.11040431261062622,
      "learning_rate": 1.138425925925926e-05,
      "loss": 0.0207,
      "step": 3722
    },
    {
      "epoch": 0.4309027777777778,
      "grad_norm": 0.08430451899766922,
      "learning_rate": 1.1381944444444446e-05,
      "loss": 0.016,
      "step": 3723
    },
    {
      "epoch": 0.4310185185185185,
      "grad_norm": 0.12157657742500305,
      "learning_rate": 1.1379629629629632e-05,
      "loss": 0.0217,
      "step": 3724
    },
    {
      "epoch": 0.43113425925925924,
      "grad_norm": 0.13789553940296173,
      "learning_rate": 1.1377314814814814e-05,
      "loss": 0.0181,
      "step": 3725
    },
    {
      "epoch": 0.43125,
      "grad_norm": 0.1282753050327301,
      "learning_rate": 1.1375e-05,
      "loss": 0.0235,
      "step": 3726
    },
    {
      "epoch": 0.43136574074074074,
      "grad_norm": 0.09411881119012833,
      "learning_rate": 1.1372685185185186e-05,
      "loss": 0.0172,
      "step": 3727
    },
    {
      "epoch": 0.43148148148148147,
      "grad_norm": 0.11523465067148209,
      "learning_rate": 1.1370370370370372e-05,
      "loss": 0.0205,
      "step": 3728
    },
    {
      "epoch": 0.43159722222222224,
      "grad_norm": 0.08700350672006607,
      "learning_rate": 1.1368055555555556e-05,
      "loss": 0.0163,
      "step": 3729
    },
    {
      "epoch": 0.43171296296296297,
      "grad_norm": 0.11030366271734238,
      "learning_rate": 1.1365740740740742e-05,
      "loss": 0.0206,
      "step": 3730
    },
    {
      "epoch": 0.4318287037037037,
      "grad_norm": 112.70210266113281,
      "learning_rate": 1.1363425925925927e-05,
      "loss": 1.0029,
      "step": 3731
    },
    {
      "epoch": 0.43194444444444446,
      "grad_norm": 0.16775104403495789,
      "learning_rate": 1.1361111111111111e-05,
      "loss": 0.0275,
      "step": 3732
    },
    {
      "epoch": 0.4320601851851852,
      "grad_norm": 0.8114295601844788,
      "learning_rate": 1.1358796296296297e-05,
      "loss": 0.0294,
      "step": 3733
    },
    {
      "epoch": 0.4321759259259259,
      "grad_norm": 0.09705136716365814,
      "learning_rate": 1.1356481481481483e-05,
      "loss": 0.0182,
      "step": 3734
    },
    {
      "epoch": 0.4322916666666667,
      "grad_norm": 0.13296622037887573,
      "learning_rate": 1.1354166666666669e-05,
      "loss": 0.0253,
      "step": 3735
    },
    {
      "epoch": 0.4324074074074074,
      "grad_norm": 0.09757456928491592,
      "learning_rate": 1.1351851851851851e-05,
      "loss": 0.0182,
      "step": 3736
    },
    {
      "epoch": 0.43252314814814813,
      "grad_norm": 0.15579749643802643,
      "learning_rate": 1.1349537037037037e-05,
      "loss": 0.0295,
      "step": 3737
    },
    {
      "epoch": 0.4326388888888889,
      "grad_norm": 0.15493415296077728,
      "learning_rate": 1.1347222222222223e-05,
      "loss": 0.0294,
      "step": 3738
    },
    {
      "epoch": 0.43275462962962963,
      "grad_norm": 1.3228745460510254,
      "learning_rate": 1.134490740740741e-05,
      "loss": 0.0379,
      "step": 3739
    },
    {
      "epoch": 0.43287037037037035,
      "grad_norm": 0.13963551819324493,
      "learning_rate": 1.1342592592592593e-05,
      "loss": 0.0237,
      "step": 3740
    },
    {
      "epoch": 0.4329861111111111,
      "grad_norm": 0.1128176599740982,
      "learning_rate": 1.134027777777778e-05,
      "loss": 0.0208,
      "step": 3741
    },
    {
      "epoch": 0.43310185185185185,
      "grad_norm": 33.74785614013672,
      "learning_rate": 1.1337962962962964e-05,
      "loss": 2.1869,
      "step": 3742
    },
    {
      "epoch": 0.43321759259259257,
      "grad_norm": 0.09858941286802292,
      "learning_rate": 1.1335648148148148e-05,
      "loss": 0.0186,
      "step": 3743
    },
    {
      "epoch": 0.43333333333333335,
      "grad_norm": 0.08340591937303543,
      "learning_rate": 1.1333333333333334e-05,
      "loss": 0.0159,
      "step": 3744
    },
    {
      "epoch": 0.43344907407407407,
      "grad_norm": 0.1251775026321411,
      "learning_rate": 1.133101851851852e-05,
      "loss": 0.0231,
      "step": 3745
    },
    {
      "epoch": 0.4335648148148148,
      "grad_norm": 0.13328927755355835,
      "learning_rate": 1.1328703703703706e-05,
      "loss": 0.0233,
      "step": 3746
    },
    {
      "epoch": 0.43368055555555557,
      "grad_norm": 0.12590225040912628,
      "learning_rate": 1.132638888888889e-05,
      "loss": 0.0231,
      "step": 3747
    },
    {
      "epoch": 0.4337962962962963,
      "grad_norm": 0.14849896728992462,
      "learning_rate": 1.1324074074074074e-05,
      "loss": 0.0259,
      "step": 3748
    },
    {
      "epoch": 0.433912037037037,
      "grad_norm": 0.18191410601139069,
      "learning_rate": 1.132175925925926e-05,
      "loss": 0.0249,
      "step": 3749
    },
    {
      "epoch": 0.4340277777777778,
      "grad_norm": 0.12214507907629013,
      "learning_rate": 1.1319444444444444e-05,
      "loss": 0.0227,
      "step": 3750
    },
    {
      "epoch": 0.4341435185185185,
      "grad_norm": 0.1574803590774536,
      "learning_rate": 1.131712962962963e-05,
      "loss": 0.0296,
      "step": 3751
    },
    {
      "epoch": 0.43425925925925923,
      "grad_norm": 0.1243414506316185,
      "learning_rate": 1.1314814814814816e-05,
      "loss": 0.0231,
      "step": 3752
    },
    {
      "epoch": 0.434375,
      "grad_norm": 27.297901153564453,
      "learning_rate": 1.1312500000000002e-05,
      "loss": 0.1028,
      "step": 3753
    },
    {
      "epoch": 0.43449074074074073,
      "grad_norm": 164.15823364257812,
      "learning_rate": 1.1310185185185185e-05,
      "loss": 0.546,
      "step": 3754
    },
    {
      "epoch": 0.43460648148148145,
      "grad_norm": 0.23969632387161255,
      "learning_rate": 1.130787037037037e-05,
      "loss": 0.0337,
      "step": 3755
    },
    {
      "epoch": 0.43472222222222223,
      "grad_norm": 0.10398803651332855,
      "learning_rate": 1.1305555555555557e-05,
      "loss": 0.0189,
      "step": 3756
    },
    {
      "epoch": 0.43483796296296295,
      "grad_norm": 0.12262008339166641,
      "learning_rate": 1.1303240740740743e-05,
      "loss": 0.0226,
      "step": 3757
    },
    {
      "epoch": 0.43495370370370373,
      "grad_norm": 0.2063758820295334,
      "learning_rate": 1.1300925925925927e-05,
      "loss": 0.0307,
      "step": 3758
    },
    {
      "epoch": 0.43506944444444445,
      "grad_norm": 0.15104179084300995,
      "learning_rate": 1.1298611111111113e-05,
      "loss": 0.0239,
      "step": 3759
    },
    {
      "epoch": 0.4351851851851852,
      "grad_norm": 6.980381488800049,
      "learning_rate": 1.1296296296296297e-05,
      "loss": 2.7881,
      "step": 3760
    },
    {
      "epoch": 0.43530092592592595,
      "grad_norm": 0.08627674728631973,
      "learning_rate": 1.1293981481481481e-05,
      "loss": 0.0161,
      "step": 3761
    },
    {
      "epoch": 0.4354166666666667,
      "grad_norm": 0.12296221405267715,
      "learning_rate": 1.1291666666666667e-05,
      "loss": 0.0227,
      "step": 3762
    },
    {
      "epoch": 0.4355324074074074,
      "grad_norm": 0.19147329032421112,
      "learning_rate": 1.1289351851851853e-05,
      "loss": 0.0225,
      "step": 3763
    },
    {
      "epoch": 0.4356481481481482,
      "grad_norm": 0.13024049997329712,
      "learning_rate": 1.1287037037037039e-05,
      "loss": 0.0245,
      "step": 3764
    },
    {
      "epoch": 0.4357638888888889,
      "grad_norm": 0.12457026541233063,
      "learning_rate": 1.1284722222222223e-05,
      "loss": 0.0166,
      "step": 3765
    },
    {
      "epoch": 0.4358796296296296,
      "grad_norm": 0.09505773335695267,
      "learning_rate": 1.1282407407407408e-05,
      "loss": 0.0179,
      "step": 3766
    },
    {
      "epoch": 0.4359953703703704,
      "grad_norm": 90.29254913330078,
      "learning_rate": 1.1280092592592594e-05,
      "loss": 1.3508,
      "step": 3767
    },
    {
      "epoch": 0.4361111111111111,
      "grad_norm": 0.08539073914289474,
      "learning_rate": 1.1277777777777778e-05,
      "loss": 0.0162,
      "step": 3768
    },
    {
      "epoch": 0.43622685185185184,
      "grad_norm": 0.10800772905349731,
      "learning_rate": 1.1275462962962964e-05,
      "loss": 0.0201,
      "step": 3769
    },
    {
      "epoch": 0.4363425925925926,
      "grad_norm": 0.1197076365351677,
      "learning_rate": 1.127314814814815e-05,
      "loss": 0.0162,
      "step": 3770
    },
    {
      "epoch": 0.43645833333333334,
      "grad_norm": 0.1180795207619667,
      "learning_rate": 1.1270833333333336e-05,
      "loss": 0.0221,
      "step": 3771
    },
    {
      "epoch": 0.43657407407407406,
      "grad_norm": 0.11920957267284393,
      "learning_rate": 1.1268518518518518e-05,
      "loss": 0.021,
      "step": 3772
    },
    {
      "epoch": 0.43668981481481484,
      "grad_norm": 9.554483413696289,
      "learning_rate": 1.1266203703703704e-05,
      "loss": 0.0394,
      "step": 3773
    },
    {
      "epoch": 0.43680555555555556,
      "grad_norm": 0.19490428268909454,
      "learning_rate": 1.126388888888889e-05,
      "loss": 0.0229,
      "step": 3774
    },
    {
      "epoch": 0.4369212962962963,
      "grad_norm": 0.10846329480409622,
      "learning_rate": 1.1261574074074076e-05,
      "loss": 0.0201,
      "step": 3775
    },
    {
      "epoch": 0.43703703703703706,
      "grad_norm": 0.15898606181144714,
      "learning_rate": 1.125925925925926e-05,
      "loss": 0.0295,
      "step": 3776
    },
    {
      "epoch": 0.4371527777777778,
      "grad_norm": 0.12068512290716171,
      "learning_rate": 1.1256944444444446e-05,
      "loss": 0.0219,
      "step": 3777
    },
    {
      "epoch": 0.4372685185185185,
      "grad_norm": 0.10515710711479187,
      "learning_rate": 1.125462962962963e-05,
      "loss": 0.0196,
      "step": 3778
    },
    {
      "epoch": 0.4373842592592593,
      "grad_norm": 0.1464575231075287,
      "learning_rate": 1.1252314814814815e-05,
      "loss": 0.0265,
      "step": 3779
    },
    {
      "epoch": 0.4375,
      "grad_norm": 0.11256590485572815,
      "learning_rate": 1.125e-05,
      "loss": 0.0205,
      "step": 3780
    },
    {
      "epoch": 0.4376157407407407,
      "grad_norm": 0.14575351774692535,
      "learning_rate": 1.1247685185185187e-05,
      "loss": 0.0266,
      "step": 3781
    },
    {
      "epoch": 0.4377314814814815,
      "grad_norm": 0.14566239714622498,
      "learning_rate": 1.1245370370370373e-05,
      "loss": 0.0274,
      "step": 3782
    },
    {
      "epoch": 0.4378472222222222,
      "grad_norm": 110.71240234375,
      "learning_rate": 1.1243055555555555e-05,
      "loss": 0.9757,
      "step": 3783
    },
    {
      "epoch": 0.43796296296296294,
      "grad_norm": 307.049072265625,
      "learning_rate": 1.1240740740740741e-05,
      "loss": 0.4714,
      "step": 3784
    },
    {
      "epoch": 0.4380787037037037,
      "grad_norm": 0.1163276731967926,
      "learning_rate": 1.1238425925925927e-05,
      "loss": 0.0219,
      "step": 3785
    },
    {
      "epoch": 0.43819444444444444,
      "grad_norm": 12.831610679626465,
      "learning_rate": 1.1236111111111111e-05,
      "loss": 2.5436,
      "step": 3786
    },
    {
      "epoch": 0.43831018518518516,
      "grad_norm": 0.10694337636232376,
      "learning_rate": 1.1233796296296297e-05,
      "loss": 0.02,
      "step": 3787
    },
    {
      "epoch": 0.43842592592592594,
      "grad_norm": 0.28671321272850037,
      "learning_rate": 1.1231481481481483e-05,
      "loss": 0.0224,
      "step": 3788
    },
    {
      "epoch": 0.43854166666666666,
      "grad_norm": 0.1460133045911789,
      "learning_rate": 1.1229166666666666e-05,
      "loss": 0.0197,
      "step": 3789
    },
    {
      "epoch": 0.4386574074074074,
      "grad_norm": 0.13592326641082764,
      "learning_rate": 1.1226851851851852e-05,
      "loss": 0.0237,
      "step": 3790
    },
    {
      "epoch": 0.43877314814814816,
      "grad_norm": 0.14968635141849518,
      "learning_rate": 1.1224537037037038e-05,
      "loss": 0.0286,
      "step": 3791
    },
    {
      "epoch": 0.4388888888888889,
      "grad_norm": 0.15027360618114471,
      "learning_rate": 1.1222222222222224e-05,
      "loss": 0.0278,
      "step": 3792
    },
    {
      "epoch": 0.4390046296296296,
      "grad_norm": 0.14440979063510895,
      "learning_rate": 1.121990740740741e-05,
      "loss": 0.0252,
      "step": 3793
    },
    {
      "epoch": 0.4391203703703704,
      "grad_norm": 0.13044090569019318,
      "learning_rate": 1.1217592592592594e-05,
      "loss": 0.023,
      "step": 3794
    },
    {
      "epoch": 0.4392361111111111,
      "grad_norm": 0.09062708914279938,
      "learning_rate": 1.1215277777777778e-05,
      "loss": 0.0171,
      "step": 3795
    },
    {
      "epoch": 0.4393518518518518,
      "grad_norm": 0.09612495452165604,
      "learning_rate": 1.1212962962962964e-05,
      "loss": 0.0179,
      "step": 3796
    },
    {
      "epoch": 0.4394675925925926,
      "grad_norm": 0.9067985415458679,
      "learning_rate": 1.1210648148148148e-05,
      "loss": 0.0253,
      "step": 3797
    },
    {
      "epoch": 0.4395833333333333,
      "grad_norm": 0.15040147304534912,
      "learning_rate": 1.1208333333333334e-05,
      "loss": 0.0206,
      "step": 3798
    },
    {
      "epoch": 0.43969907407407405,
      "grad_norm": 0.08198311179876328,
      "learning_rate": 1.120601851851852e-05,
      "loss": 0.0156,
      "step": 3799
    },
    {
      "epoch": 0.4398148148148148,
      "grad_norm": 0.11410444974899292,
      "learning_rate": 1.1203703703703706e-05,
      "loss": 0.0206,
      "step": 3800
    },
    {
      "epoch": 0.43993055555555555,
      "grad_norm": 0.36601755023002625,
      "learning_rate": 1.1201388888888889e-05,
      "loss": 0.0378,
      "step": 3801
    },
    {
      "epoch": 0.44004629629629627,
      "grad_norm": 0.12536731362342834,
      "learning_rate": 1.1199074074074075e-05,
      "loss": 0.0225,
      "step": 3802
    },
    {
      "epoch": 0.44016203703703705,
      "grad_norm": 0.09839978814125061,
      "learning_rate": 1.119675925925926e-05,
      "loss": 0.0184,
      "step": 3803
    },
    {
      "epoch": 0.44027777777777777,
      "grad_norm": 0.115195132791996,
      "learning_rate": 1.1194444444444445e-05,
      "loss": 0.0217,
      "step": 3804
    },
    {
      "epoch": 0.44039351851851855,
      "grad_norm": 0.3385573923587799,
      "learning_rate": 1.119212962962963e-05,
      "loss": 0.0195,
      "step": 3805
    },
    {
      "epoch": 0.44050925925925927,
      "grad_norm": 14.912522315979004,
      "learning_rate": 1.1189814814814817e-05,
      "loss": 0.12,
      "step": 3806
    },
    {
      "epoch": 0.440625,
      "grad_norm": 0.12279172241687775,
      "learning_rate": 1.11875e-05,
      "loss": 0.0218,
      "step": 3807
    },
    {
      "epoch": 0.44074074074074077,
      "grad_norm": 0.6116386651992798,
      "learning_rate": 1.1185185185185185e-05,
      "loss": 0.0305,
      "step": 3808
    },
    {
      "epoch": 0.4408564814814815,
      "grad_norm": 0.11242762953042984,
      "learning_rate": 1.1182870370370371e-05,
      "loss": 0.0202,
      "step": 3809
    },
    {
      "epoch": 0.4409722222222222,
      "grad_norm": 0.1131284087896347,
      "learning_rate": 1.1180555555555557e-05,
      "loss": 0.0213,
      "step": 3810
    },
    {
      "epoch": 0.441087962962963,
      "grad_norm": 0.13089577853679657,
      "learning_rate": 1.1178240740740743e-05,
      "loss": 0.0245,
      "step": 3811
    },
    {
      "epoch": 0.4412037037037037,
      "grad_norm": 0.12473814934492111,
      "learning_rate": 1.1175925925925927e-05,
      "loss": 0.0224,
      "step": 3812
    },
    {
      "epoch": 0.44131944444444443,
      "grad_norm": 6.304518699645996,
      "learning_rate": 1.1173611111111111e-05,
      "loss": 0.0514,
      "step": 3813
    },
    {
      "epoch": 0.4414351851851852,
      "grad_norm": 0.1072130873799324,
      "learning_rate": 1.1171296296296297e-05,
      "loss": 0.0199,
      "step": 3814
    },
    {
      "epoch": 0.44155092592592593,
      "grad_norm": 0.10849025100469589,
      "learning_rate": 1.1168981481481482e-05,
      "loss": 0.0202,
      "step": 3815
    },
    {
      "epoch": 0.44166666666666665,
      "grad_norm": 0.11898993700742722,
      "learning_rate": 1.1166666666666668e-05,
      "loss": 0.0224,
      "step": 3816
    },
    {
      "epoch": 0.44178240740740743,
      "grad_norm": 0.1462133675813675,
      "learning_rate": 1.1164351851851854e-05,
      "loss": 0.0246,
      "step": 3817
    },
    {
      "epoch": 0.44189814814814815,
      "grad_norm": 0.28196343779563904,
      "learning_rate": 1.116203703703704e-05,
      "loss": 0.0272,
      "step": 3818
    },
    {
      "epoch": 0.4420138888888889,
      "grad_norm": 0.11234883219003677,
      "learning_rate": 1.1159722222222222e-05,
      "loss": 0.0212,
      "step": 3819
    },
    {
      "epoch": 0.44212962962962965,
      "grad_norm": 0.21305972337722778,
      "learning_rate": 1.1157407407407408e-05,
      "loss": 0.0292,
      "step": 3820
    },
    {
      "epoch": 0.44224537037037037,
      "grad_norm": 0.10469334572553635,
      "learning_rate": 1.1155092592592594e-05,
      "loss": 0.0196,
      "step": 3821
    },
    {
      "epoch": 0.4423611111111111,
      "grad_norm": 0.1503657102584839,
      "learning_rate": 1.1152777777777778e-05,
      "loss": 0.0194,
      "step": 3822
    },
    {
      "epoch": 0.44247685185185187,
      "grad_norm": 0.14835791289806366,
      "learning_rate": 1.1150462962962964e-05,
      "loss": 0.0262,
      "step": 3823
    },
    {
      "epoch": 0.4425925925925926,
      "grad_norm": 0.12209124863147736,
      "learning_rate": 1.114814814814815e-05,
      "loss": 0.0221,
      "step": 3824
    },
    {
      "epoch": 0.4427083333333333,
      "grad_norm": 0.13056041300296783,
      "learning_rate": 1.1145833333333334e-05,
      "loss": 0.0222,
      "step": 3825
    },
    {
      "epoch": 0.4428240740740741,
      "grad_norm": 0.08484189957380295,
      "learning_rate": 1.1143518518518519e-05,
      "loss": 0.0159,
      "step": 3826
    },
    {
      "epoch": 0.4429398148148148,
      "grad_norm": 0.15140719711780548,
      "learning_rate": 1.1141203703703705e-05,
      "loss": 0.0227,
      "step": 3827
    },
    {
      "epoch": 0.44305555555555554,
      "grad_norm": 0.13032402098178864,
      "learning_rate": 1.113888888888889e-05,
      "loss": 0.0229,
      "step": 3828
    },
    {
      "epoch": 0.4431712962962963,
      "grad_norm": 0.15488196909427643,
      "learning_rate": 1.1136574074074076e-05,
      "loss": 0.0278,
      "step": 3829
    },
    {
      "epoch": 0.44328703703703703,
      "grad_norm": 0.11540248990058899,
      "learning_rate": 1.1134259259259259e-05,
      "loss": 0.0157,
      "step": 3830
    },
    {
      "epoch": 0.44340277777777776,
      "grad_norm": 0.12676812708377838,
      "learning_rate": 1.1131944444444445e-05,
      "loss": 0.0231,
      "step": 3831
    },
    {
      "epoch": 0.44351851851851853,
      "grad_norm": 0.14834748208522797,
      "learning_rate": 1.1129629629629631e-05,
      "loss": 0.0282,
      "step": 3832
    },
    {
      "epoch": 0.44363425925925926,
      "grad_norm": 0.12020070105791092,
      "learning_rate": 1.1127314814814815e-05,
      "loss": 0.0215,
      "step": 3833
    },
    {
      "epoch": 0.44375,
      "grad_norm": 183.47792053222656,
      "learning_rate": 1.1125000000000001e-05,
      "loss": 1.1833,
      "step": 3834
    },
    {
      "epoch": 0.44386574074074076,
      "grad_norm": 0.11722131818532944,
      "learning_rate": 1.1122685185185187e-05,
      "loss": 0.0192,
      "step": 3835
    },
    {
      "epoch": 0.4439814814814815,
      "grad_norm": 95.47689056396484,
      "learning_rate": 1.112037037037037e-05,
      "loss": 0.6782,
      "step": 3836
    },
    {
      "epoch": 0.4440972222222222,
      "grad_norm": 0.15712803602218628,
      "learning_rate": 1.1118055555555556e-05,
      "loss": 0.0211,
      "step": 3837
    },
    {
      "epoch": 0.444212962962963,
      "grad_norm": 18.109037399291992,
      "learning_rate": 1.1115740740740741e-05,
      "loss": 2.3476,
      "step": 3838
    },
    {
      "epoch": 0.4443287037037037,
      "grad_norm": 0.23212289810180664,
      "learning_rate": 1.1113425925925927e-05,
      "loss": 0.0212,
      "step": 3839
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 0.2128465324640274,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 0.0284,
      "step": 3840
    },
    {
      "epoch": 0.4445601851851852,
      "grad_norm": 10.802033424377441,
      "learning_rate": 1.1108796296296298e-05,
      "loss": 2.5336,
      "step": 3841
    },
    {
      "epoch": 0.4446759259259259,
      "grad_norm": 0.1062355488538742,
      "learning_rate": 1.1106481481481482e-05,
      "loss": 0.0197,
      "step": 3842
    },
    {
      "epoch": 0.44479166666666664,
      "grad_norm": 0.09270496666431427,
      "learning_rate": 1.1104166666666668e-05,
      "loss": 0.0174,
      "step": 3843
    },
    {
      "epoch": 0.4449074074074074,
      "grad_norm": 5.750290393829346,
      "learning_rate": 1.1101851851851852e-05,
      "loss": 0.0477,
      "step": 3844
    },
    {
      "epoch": 0.44502314814814814,
      "grad_norm": 0.11344258487224579,
      "learning_rate": 1.1099537037037038e-05,
      "loss": 0.0179,
      "step": 3845
    },
    {
      "epoch": 0.44513888888888886,
      "grad_norm": 0.17426778376102448,
      "learning_rate": 1.1097222222222224e-05,
      "loss": 0.0238,
      "step": 3846
    },
    {
      "epoch": 0.44525462962962964,
      "grad_norm": 0.14954693615436554,
      "learning_rate": 1.109490740740741e-05,
      "loss": 0.0266,
      "step": 3847
    },
    {
      "epoch": 0.44537037037037036,
      "grad_norm": 0.1555042862892151,
      "learning_rate": 1.1092592592592592e-05,
      "loss": 0.0285,
      "step": 3848
    },
    {
      "epoch": 0.44548611111111114,
      "grad_norm": 0.204475536942482,
      "learning_rate": 1.1090277777777778e-05,
      "loss": 0.0263,
      "step": 3849
    },
    {
      "epoch": 0.44560185185185186,
      "grad_norm": 0.10646437853574753,
      "learning_rate": 1.1087962962962964e-05,
      "loss": 0.0197,
      "step": 3850
    },
    {
      "epoch": 0.4457175925925926,
      "grad_norm": 0.11071562021970749,
      "learning_rate": 1.1085648148148149e-05,
      "loss": 0.0196,
      "step": 3851
    },
    {
      "epoch": 0.44583333333333336,
      "grad_norm": 0.09167350083589554,
      "learning_rate": 1.1083333333333335e-05,
      "loss": 0.0172,
      "step": 3852
    },
    {
      "epoch": 0.4459490740740741,
      "grad_norm": 0.108653724193573,
      "learning_rate": 1.108101851851852e-05,
      "loss": 0.0148,
      "step": 3853
    },
    {
      "epoch": 0.4460648148148148,
      "grad_norm": 0.7989105582237244,
      "learning_rate": 1.1078703703703703e-05,
      "loss": 0.0322,
      "step": 3854
    },
    {
      "epoch": 0.4461805555555556,
      "grad_norm": 0.43386146426200867,
      "learning_rate": 1.1076388888888889e-05,
      "loss": 0.0334,
      "step": 3855
    },
    {
      "epoch": 0.4462962962962963,
      "grad_norm": 0.11539699137210846,
      "learning_rate": 1.1074074074074075e-05,
      "loss": 0.0157,
      "step": 3856
    },
    {
      "epoch": 0.446412037037037,
      "grad_norm": 9.789985656738281,
      "learning_rate": 1.107175925925926e-05,
      "loss": 2.4386,
      "step": 3857
    },
    {
      "epoch": 0.4465277777777778,
      "grad_norm": 0.7477298974990845,
      "learning_rate": 1.1069444444444447e-05,
      "loss": 0.0371,
      "step": 3858
    },
    {
      "epoch": 0.4466435185185185,
      "grad_norm": 7.613979816436768,
      "learning_rate": 1.1067129629629631e-05,
      "loss": 0.046,
      "step": 3859
    },
    {
      "epoch": 0.44675925925925924,
      "grad_norm": 0.09307112544775009,
      "learning_rate": 1.1064814814814815e-05,
      "loss": 0.0176,
      "step": 3860
    },
    {
      "epoch": 0.446875,
      "grad_norm": 0.09972336143255234,
      "learning_rate": 1.1062500000000001e-05,
      "loss": 0.0189,
      "step": 3861
    },
    {
      "epoch": 0.44699074074074074,
      "grad_norm": 0.14041036367416382,
      "learning_rate": 1.1060185185185185e-05,
      "loss": 0.0255,
      "step": 3862
    },
    {
      "epoch": 0.44710648148148147,
      "grad_norm": 0.08202961832284927,
      "learning_rate": 1.1057870370370371e-05,
      "loss": 0.0155,
      "step": 3863
    },
    {
      "epoch": 0.44722222222222224,
      "grad_norm": 0.08876650035381317,
      "learning_rate": 1.1055555555555557e-05,
      "loss": 0.0166,
      "step": 3864
    },
    {
      "epoch": 0.44733796296296297,
      "grad_norm": 0.15839937329292297,
      "learning_rate": 1.1053240740740743e-05,
      "loss": 0.0282,
      "step": 3865
    },
    {
      "epoch": 0.4474537037037037,
      "grad_norm": 1.2064956426620483,
      "learning_rate": 1.1050925925925926e-05,
      "loss": 0.0346,
      "step": 3866
    },
    {
      "epoch": 0.44756944444444446,
      "grad_norm": 0.12997984886169434,
      "learning_rate": 1.1048611111111112e-05,
      "loss": 0.0178,
      "step": 3867
    },
    {
      "epoch": 0.4476851851851852,
      "grad_norm": 0.1940409243106842,
      "learning_rate": 1.1046296296296298e-05,
      "loss": 0.0251,
      "step": 3868
    },
    {
      "epoch": 0.4478009259259259,
      "grad_norm": 0.11549633741378784,
      "learning_rate": 1.1043981481481482e-05,
      "loss": 0.0203,
      "step": 3869
    },
    {
      "epoch": 0.4479166666666667,
      "grad_norm": 0.08890421688556671,
      "learning_rate": 1.1041666666666668e-05,
      "loss": 0.0168,
      "step": 3870
    },
    {
      "epoch": 0.4480324074074074,
      "grad_norm": 0.0808873251080513,
      "learning_rate": 1.1039351851851852e-05,
      "loss": 0.0151,
      "step": 3871
    },
    {
      "epoch": 0.44814814814814813,
      "grad_norm": 0.12900274991989136,
      "learning_rate": 1.1037037037037036e-05,
      "loss": 0.0236,
      "step": 3872
    },
    {
      "epoch": 0.4482638888888889,
      "grad_norm": 0.08964599668979645,
      "learning_rate": 1.1034722222222222e-05,
      "loss": 0.0169,
      "step": 3873
    },
    {
      "epoch": 0.44837962962962963,
      "grad_norm": 0.10630672425031662,
      "learning_rate": 1.1032407407407408e-05,
      "loss": 0.0145,
      "step": 3874
    },
    {
      "epoch": 0.44849537037037035,
      "grad_norm": 0.1420777589082718,
      "learning_rate": 1.1030092592592594e-05,
      "loss": 0.0252,
      "step": 3875
    },
    {
      "epoch": 0.4486111111111111,
      "grad_norm": 56.84318923950195,
      "learning_rate": 1.102777777777778e-05,
      "loss": 2.1921,
      "step": 3876
    },
    {
      "epoch": 0.44872685185185185,
      "grad_norm": 0.4263235032558441,
      "learning_rate": 1.1025462962962963e-05,
      "loss": 0.0176,
      "step": 3877
    },
    {
      "epoch": 0.44884259259259257,
      "grad_norm": 0.13112123310565948,
      "learning_rate": 1.1023148148148149e-05,
      "loss": 0.023,
      "step": 3878
    },
    {
      "epoch": 0.44895833333333335,
      "grad_norm": 0.12457422912120819,
      "learning_rate": 1.1020833333333335e-05,
      "loss": 0.0232,
      "step": 3879
    },
    {
      "epoch": 0.44907407407407407,
      "grad_norm": 0.11619628220796585,
      "learning_rate": 1.1018518518518519e-05,
      "loss": 0.0209,
      "step": 3880
    },
    {
      "epoch": 0.4491898148148148,
      "grad_norm": 0.19892814755439758,
      "learning_rate": 1.1016203703703705e-05,
      "loss": 0.0311,
      "step": 3881
    },
    {
      "epoch": 0.44930555555555557,
      "grad_norm": 0.7383542060852051,
      "learning_rate": 1.101388888888889e-05,
      "loss": 0.0397,
      "step": 3882
    },
    {
      "epoch": 0.4494212962962963,
      "grad_norm": 0.13248249888420105,
      "learning_rate": 1.1011574074074073e-05,
      "loss": 0.0227,
      "step": 3883
    },
    {
      "epoch": 0.449537037037037,
      "grad_norm": 0.14219450950622559,
      "learning_rate": 1.100925925925926e-05,
      "loss": 0.0236,
      "step": 3884
    },
    {
      "epoch": 0.4496527777777778,
      "grad_norm": 0.17249052226543427,
      "learning_rate": 1.1006944444444445e-05,
      "loss": 0.0195,
      "step": 3885
    },
    {
      "epoch": 0.4497685185185185,
      "grad_norm": 0.13230660557746887,
      "learning_rate": 1.1004629629629631e-05,
      "loss": 0.0251,
      "step": 3886
    },
    {
      "epoch": 0.44988425925925923,
      "grad_norm": 0.42603200674057007,
      "learning_rate": 1.1002314814814815e-05,
      "loss": 0.0305,
      "step": 3887
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.08271831274032593,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.0154,
      "step": 3888
    },
    {
      "epoch": 0.45011574074074073,
      "grad_norm": 343.40618896484375,
      "learning_rate": 1.0997685185185186e-05,
      "loss": 0.5291,
      "step": 3889
    },
    {
      "epoch": 0.45023148148148145,
      "grad_norm": 0.43098634481430054,
      "learning_rate": 1.099537037037037e-05,
      "loss": 0.0327,
      "step": 3890
    },
    {
      "epoch": 0.45034722222222223,
      "grad_norm": 0.12421353161334991,
      "learning_rate": 1.0993055555555556e-05,
      "loss": 0.0236,
      "step": 3891
    },
    {
      "epoch": 0.45046296296296295,
      "grad_norm": 2.186102867126465,
      "learning_rate": 1.0990740740740742e-05,
      "loss": 0.0377,
      "step": 3892
    },
    {
      "epoch": 0.45057870370370373,
      "grad_norm": 0.12287448346614838,
      "learning_rate": 1.0988425925925928e-05,
      "loss": 0.0219,
      "step": 3893
    },
    {
      "epoch": 0.45069444444444445,
      "grad_norm": 0.3660595715045929,
      "learning_rate": 1.0986111111111114e-05,
      "loss": 0.0232,
      "step": 3894
    },
    {
      "epoch": 0.4508101851851852,
      "grad_norm": 0.12152888625860214,
      "learning_rate": 1.0983796296296296e-05,
      "loss": 0.022,
      "step": 3895
    },
    {
      "epoch": 0.45092592592592595,
      "grad_norm": 0.11796385049819946,
      "learning_rate": 1.0981481481481482e-05,
      "loss": 0.0221,
      "step": 3896
    },
    {
      "epoch": 0.4510416666666667,
      "grad_norm": 2.348074197769165,
      "learning_rate": 1.0979166666666668e-05,
      "loss": 0.0427,
      "step": 3897
    },
    {
      "epoch": 0.4511574074074074,
      "grad_norm": 0.3719361126422882,
      "learning_rate": 1.0976851851851852e-05,
      "loss": 0.0273,
      "step": 3898
    },
    {
      "epoch": 0.4512731481481482,
      "grad_norm": 0.12574875354766846,
      "learning_rate": 1.0974537037037038e-05,
      "loss": 0.0229,
      "step": 3899
    },
    {
      "epoch": 0.4513888888888889,
      "grad_norm": 0.1267777681350708,
      "learning_rate": 1.0972222222222224e-05,
      "loss": 0.0237,
      "step": 3900
    },
    {
      "epoch": 0.4515046296296296,
      "grad_norm": 0.3703366219997406,
      "learning_rate": 1.0969907407407407e-05,
      "loss": 0.0254,
      "step": 3901
    },
    {
      "epoch": 0.4516203703703704,
      "grad_norm": 0.11602979898452759,
      "learning_rate": 1.0967592592592593e-05,
      "loss": 0.0217,
      "step": 3902
    },
    {
      "epoch": 0.4517361111111111,
      "grad_norm": 0.11711587756872177,
      "learning_rate": 1.0965277777777779e-05,
      "loss": 0.0219,
      "step": 3903
    },
    {
      "epoch": 0.45185185185185184,
      "grad_norm": 0.1524633765220642,
      "learning_rate": 1.0962962962962965e-05,
      "loss": 0.0291,
      "step": 3904
    },
    {
      "epoch": 0.4519675925925926,
      "grad_norm": 0.10621511936187744,
      "learning_rate": 1.0960648148148149e-05,
      "loss": 0.0144,
      "step": 3905
    },
    {
      "epoch": 0.45208333333333334,
      "grad_norm": 0.09494803100824356,
      "learning_rate": 1.0958333333333335e-05,
      "loss": 0.0168,
      "step": 3906
    },
    {
      "epoch": 0.45219907407407406,
      "grad_norm": 0.18627965450286865,
      "learning_rate": 1.0956018518518519e-05,
      "loss": 0.027,
      "step": 3907
    },
    {
      "epoch": 0.45231481481481484,
      "grad_norm": 1.2335065603256226,
      "learning_rate": 1.0953703703703703e-05,
      "loss": 0.0338,
      "step": 3908
    },
    {
      "epoch": 0.45243055555555556,
      "grad_norm": 0.09289966523647308,
      "learning_rate": 1.095138888888889e-05,
      "loss": 0.0174,
      "step": 3909
    },
    {
      "epoch": 0.4525462962962963,
      "grad_norm": 0.11258766055107117,
      "learning_rate": 1.0949074074074075e-05,
      "loss": 0.0207,
      "step": 3910
    },
    {
      "epoch": 0.45266203703703706,
      "grad_norm": 0.1429065614938736,
      "learning_rate": 1.0946759259259261e-05,
      "loss": 0.0235,
      "step": 3911
    },
    {
      "epoch": 0.4527777777777778,
      "grad_norm": 0.10332705080509186,
      "learning_rate": 1.0944444444444447e-05,
      "loss": 0.0191,
      "step": 3912
    },
    {
      "epoch": 0.4528935185185185,
      "grad_norm": 0.12037567049264908,
      "learning_rate": 1.094212962962963e-05,
      "loss": 0.022,
      "step": 3913
    },
    {
      "epoch": 0.4530092592592593,
      "grad_norm": 0.11310096085071564,
      "learning_rate": 1.0939814814814816e-05,
      "loss": 0.0207,
      "step": 3914
    },
    {
      "epoch": 0.453125,
      "grad_norm": 0.1384851485490799,
      "learning_rate": 1.0937500000000002e-05,
      "loss": 0.023,
      "step": 3915
    },
    {
      "epoch": 0.4532407407407407,
      "grad_norm": 0.13994352519512177,
      "learning_rate": 1.0935185185185186e-05,
      "loss": 0.0251,
      "step": 3916
    },
    {
      "epoch": 0.4533564814814815,
      "grad_norm": 0.11241249740123749,
      "learning_rate": 1.0932870370370372e-05,
      "loss": 0.0205,
      "step": 3917
    },
    {
      "epoch": 0.4534722222222222,
      "grad_norm": 0.3597131669521332,
      "learning_rate": 1.0930555555555556e-05,
      "loss": 0.0192,
      "step": 3918
    },
    {
      "epoch": 0.45358796296296294,
      "grad_norm": 0.10469485074281693,
      "learning_rate": 1.092824074074074e-05,
      "loss": 0.019,
      "step": 3919
    },
    {
      "epoch": 0.4537037037037037,
      "grad_norm": 0.08945069462060928,
      "learning_rate": 1.0925925925925926e-05,
      "loss": 0.0168,
      "step": 3920
    },
    {
      "epoch": 0.45381944444444444,
      "grad_norm": 0.10670896619558334,
      "learning_rate": 1.0923611111111112e-05,
      "loss": 0.0201,
      "step": 3921
    },
    {
      "epoch": 0.45393518518518516,
      "grad_norm": 0.09089773148298264,
      "learning_rate": 1.0921296296296298e-05,
      "loss": 0.0167,
      "step": 3922
    },
    {
      "epoch": 0.45405092592592594,
      "grad_norm": 0.1112956628203392,
      "learning_rate": 1.0918981481481482e-05,
      "loss": 0.0152,
      "step": 3923
    },
    {
      "epoch": 0.45416666666666666,
      "grad_norm": 0.11844444274902344,
      "learning_rate": 1.0916666666666667e-05,
      "loss": 0.0216,
      "step": 3924
    },
    {
      "epoch": 0.4542824074074074,
      "grad_norm": 0.13773614168167114,
      "learning_rate": 1.0914351851851853e-05,
      "loss": 0.0232,
      "step": 3925
    },
    {
      "epoch": 0.45439814814814816,
      "grad_norm": 16.088199615478516,
      "learning_rate": 1.0912037037037037e-05,
      "loss": 0.0669,
      "step": 3926
    },
    {
      "epoch": 0.4545138888888889,
      "grad_norm": 0.0931662991642952,
      "learning_rate": 1.0909722222222223e-05,
      "loss": 0.0168,
      "step": 3927
    },
    {
      "epoch": 0.4546296296296296,
      "grad_norm": 8.659886360168457,
      "learning_rate": 1.0907407407407409e-05,
      "loss": 2.8956,
      "step": 3928
    },
    {
      "epoch": 0.4547453703703704,
      "grad_norm": 2.895231008529663,
      "learning_rate": 1.0905092592592595e-05,
      "loss": 0.0313,
      "step": 3929
    },
    {
      "epoch": 0.4548611111111111,
      "grad_norm": 0.24396459758281708,
      "learning_rate": 1.0902777777777777e-05,
      "loss": 0.0282,
      "step": 3930
    },
    {
      "epoch": 0.4549768518518518,
      "grad_norm": 0.11004975438117981,
      "learning_rate": 1.0900462962962963e-05,
      "loss": 0.0209,
      "step": 3931
    },
    {
      "epoch": 0.4550925925925926,
      "grad_norm": 0.11562091112136841,
      "learning_rate": 1.0898148148148149e-05,
      "loss": 0.0217,
      "step": 3932
    },
    {
      "epoch": 0.4552083333333333,
      "grad_norm": 0.10290400683879852,
      "learning_rate": 1.0895833333333335e-05,
      "loss": 0.014,
      "step": 3933
    },
    {
      "epoch": 0.45532407407407405,
      "grad_norm": 0.170152485370636,
      "learning_rate": 1.089351851851852e-05,
      "loss": 0.0207,
      "step": 3934
    },
    {
      "epoch": 0.4554398148148148,
      "grad_norm": 0.09276590496301651,
      "learning_rate": 1.0891203703703705e-05,
      "loss": 0.0157,
      "step": 3935
    },
    {
      "epoch": 0.45555555555555555,
      "grad_norm": 103.87399291992188,
      "learning_rate": 1.088888888888889e-05,
      "loss": 0.8035,
      "step": 3936
    },
    {
      "epoch": 0.45567129629629627,
      "grad_norm": 0.14492687582969666,
      "learning_rate": 1.0886574074074074e-05,
      "loss": 0.0269,
      "step": 3937
    },
    {
      "epoch": 0.45578703703703705,
      "grad_norm": 0.09989020228385925,
      "learning_rate": 1.088425925925926e-05,
      "loss": 0.0186,
      "step": 3938
    },
    {
      "epoch": 0.45590277777777777,
      "grad_norm": 0.6334127187728882,
      "learning_rate": 1.0881944444444446e-05,
      "loss": 0.0311,
      "step": 3939
    },
    {
      "epoch": 0.45601851851851855,
      "grad_norm": 28.56847381591797,
      "learning_rate": 1.0879629629629632e-05,
      "loss": 2.0605,
      "step": 3940
    },
    {
      "epoch": 0.45613425925925927,
      "grad_norm": 0.10832750052213669,
      "learning_rate": 1.0877314814814816e-05,
      "loss": 0.0148,
      "step": 3941
    },
    {
      "epoch": 0.45625,
      "grad_norm": 25.13556480407715,
      "learning_rate": 1.0875e-05,
      "loss": 2.7334,
      "step": 3942
    },
    {
      "epoch": 0.45636574074074077,
      "grad_norm": 0.10356325656175613,
      "learning_rate": 1.0872685185185186e-05,
      "loss": 0.019,
      "step": 3943
    },
    {
      "epoch": 0.4564814814814815,
      "grad_norm": 9.750815391540527,
      "learning_rate": 1.0870370370370372e-05,
      "loss": 2.4891,
      "step": 3944
    },
    {
      "epoch": 0.4565972222222222,
      "grad_norm": 113.10006713867188,
      "learning_rate": 1.0868055555555556e-05,
      "loss": 2.3959,
      "step": 3945
    },
    {
      "epoch": 0.456712962962963,
      "grad_norm": 0.09242449700832367,
      "learning_rate": 1.0865740740740742e-05,
      "loss": 0.0172,
      "step": 3946
    },
    {
      "epoch": 0.4568287037037037,
      "grad_norm": 0.14194893836975098,
      "learning_rate": 1.0863425925925928e-05,
      "loss": 0.0268,
      "step": 3947
    },
    {
      "epoch": 0.45694444444444443,
      "grad_norm": 0.14120891690254211,
      "learning_rate": 1.086111111111111e-05,
      "loss": 0.0232,
      "step": 3948
    },
    {
      "epoch": 0.4570601851851852,
      "grad_norm": 0.1649198830127716,
      "learning_rate": 1.0858796296296297e-05,
      "loss": 0.0221,
      "step": 3949
    },
    {
      "epoch": 0.45717592592592593,
      "grad_norm": 0.08021102845668793,
      "learning_rate": 1.0856481481481483e-05,
      "loss": 0.015,
      "step": 3950
    },
    {
      "epoch": 0.45729166666666665,
      "grad_norm": 0.12596675753593445,
      "learning_rate": 1.0854166666666668e-05,
      "loss": 0.0156,
      "step": 3951
    },
    {
      "epoch": 0.45740740740740743,
      "grad_norm": 0.10832260549068451,
      "learning_rate": 1.0851851851851853e-05,
      "loss": 0.0147,
      "step": 3952
    },
    {
      "epoch": 0.45752314814814815,
      "grad_norm": 0.12150416523218155,
      "learning_rate": 1.0849537037037039e-05,
      "loss": 0.0217,
      "step": 3953
    },
    {
      "epoch": 0.4576388888888889,
      "grad_norm": 0.13465648889541626,
      "learning_rate": 1.0847222222222223e-05,
      "loss": 0.0228,
      "step": 3954
    },
    {
      "epoch": 0.45775462962962965,
      "grad_norm": 0.0895831510424614,
      "learning_rate": 1.0844907407407407e-05,
      "loss": 0.0167,
      "step": 3955
    },
    {
      "epoch": 0.45787037037037037,
      "grad_norm": 0.1313965767621994,
      "learning_rate": 1.0842592592592593e-05,
      "loss": 0.0224,
      "step": 3956
    },
    {
      "epoch": 0.4579861111111111,
      "grad_norm": 0.46947091817855835,
      "learning_rate": 1.0840277777777779e-05,
      "loss": 0.0188,
      "step": 3957
    },
    {
      "epoch": 0.45810185185185187,
      "grad_norm": 0.5774734020233154,
      "learning_rate": 1.0837962962962965e-05,
      "loss": 0.0309,
      "step": 3958
    },
    {
      "epoch": 0.4582175925925926,
      "grad_norm": 0.09540968388319016,
      "learning_rate": 1.0835648148148151e-05,
      "loss": 0.0181,
      "step": 3959
    },
    {
      "epoch": 0.4583333333333333,
      "grad_norm": 0.12212913483381271,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 0.0223,
      "step": 3960
    },
    {
      "epoch": 0.4584490740740741,
      "grad_norm": 26.20662498474121,
      "learning_rate": 1.083101851851852e-05,
      "loss": 2.4556,
      "step": 3961
    },
    {
      "epoch": 0.4585648148148148,
      "grad_norm": 0.17124682664871216,
      "learning_rate": 1.0828703703703705e-05,
      "loss": 0.0285,
      "step": 3962
    },
    {
      "epoch": 0.45868055555555554,
      "grad_norm": 0.1199539378285408,
      "learning_rate": 1.082638888888889e-05,
      "loss": 0.02,
      "step": 3963
    },
    {
      "epoch": 0.4587962962962963,
      "grad_norm": 0.15750467777252197,
      "learning_rate": 1.0824074074074076e-05,
      "loss": 0.0253,
      "step": 3964
    },
    {
      "epoch": 0.45891203703703703,
      "grad_norm": 37.93769836425781,
      "learning_rate": 1.082175925925926e-05,
      "loss": 2.4563,
      "step": 3965
    },
    {
      "epoch": 0.45902777777777776,
      "grad_norm": 0.11263380944728851,
      "learning_rate": 1.0819444444444444e-05,
      "loss": 0.0205,
      "step": 3966
    },
    {
      "epoch": 0.45914351851851853,
      "grad_norm": 0.12110617011785507,
      "learning_rate": 1.081712962962963e-05,
      "loss": 0.0211,
      "step": 3967
    },
    {
      "epoch": 0.45925925925925926,
      "grad_norm": 0.11746494472026825,
      "learning_rate": 1.0814814814814816e-05,
      "loss": 0.0218,
      "step": 3968
    },
    {
      "epoch": 0.459375,
      "grad_norm": 0.10074715316295624,
      "learning_rate": 1.0812500000000002e-05,
      "loss": 0.0185,
      "step": 3969
    },
    {
      "epoch": 0.45949074074074076,
      "grad_norm": 0.13882805407047272,
      "learning_rate": 1.0810185185185186e-05,
      "loss": 0.0222,
      "step": 3970
    },
    {
      "epoch": 0.4596064814814815,
      "grad_norm": 0.10345536470413208,
      "learning_rate": 1.080787037037037e-05,
      "loss": 0.0187,
      "step": 3971
    },
    {
      "epoch": 0.4597222222222222,
      "grad_norm": 0.10673236101865768,
      "learning_rate": 1.0805555555555556e-05,
      "loss": 0.0146,
      "step": 3972
    },
    {
      "epoch": 0.459837962962963,
      "grad_norm": 0.11488202959299088,
      "learning_rate": 1.080324074074074e-05,
      "loss": 0.0205,
      "step": 3973
    },
    {
      "epoch": 0.4599537037037037,
      "grad_norm": 0.1315426528453827,
      "learning_rate": 1.0800925925925927e-05,
      "loss": 0.0248,
      "step": 3974
    },
    {
      "epoch": 0.4600694444444444,
      "grad_norm": 0.12181282788515091,
      "learning_rate": 1.0798611111111113e-05,
      "loss": 0.0219,
      "step": 3975
    },
    {
      "epoch": 0.4601851851851852,
      "grad_norm": 0.1143922284245491,
      "learning_rate": 1.0796296296296298e-05,
      "loss": 0.021,
      "step": 3976
    },
    {
      "epoch": 0.4603009259259259,
      "grad_norm": 0.11158769577741623,
      "learning_rate": 1.0793981481481481e-05,
      "loss": 0.0209,
      "step": 3977
    },
    {
      "epoch": 0.46041666666666664,
      "grad_norm": 0.20308399200439453,
      "learning_rate": 1.0791666666666667e-05,
      "loss": 0.0294,
      "step": 3978
    },
    {
      "epoch": 0.4605324074074074,
      "grad_norm": 0.11834444105625153,
      "learning_rate": 1.0789351851851853e-05,
      "loss": 0.0206,
      "step": 3979
    },
    {
      "epoch": 0.46064814814814814,
      "grad_norm": 38.11184310913086,
      "learning_rate": 1.0787037037037039e-05,
      "loss": 2.2798,
      "step": 3980
    },
    {
      "epoch": 0.46076388888888886,
      "grad_norm": 0.11204198002815247,
      "learning_rate": 1.0784722222222223e-05,
      "loss": 0.0205,
      "step": 3981
    },
    {
      "epoch": 0.46087962962962964,
      "grad_norm": 0.10777705162763596,
      "learning_rate": 1.0782407407407409e-05,
      "loss": 0.0191,
      "step": 3982
    },
    {
      "epoch": 0.46099537037037036,
      "grad_norm": 0.3723234534263611,
      "learning_rate": 1.0780092592592593e-05,
      "loss": 0.023,
      "step": 3983
    },
    {
      "epoch": 0.46111111111111114,
      "grad_norm": 0.11712723970413208,
      "learning_rate": 1.0777777777777778e-05,
      "loss": 0.0221,
      "step": 3984
    },
    {
      "epoch": 0.46122685185185186,
      "grad_norm": 0.16782453656196594,
      "learning_rate": 1.0775462962962963e-05,
      "loss": 0.0267,
      "step": 3985
    },
    {
      "epoch": 0.4613425925925926,
      "grad_norm": 0.0784984827041626,
      "learning_rate": 1.077314814814815e-05,
      "loss": 0.0147,
      "step": 3986
    },
    {
      "epoch": 0.46145833333333336,
      "grad_norm": 0.0771809071302414,
      "learning_rate": 1.0770833333333335e-05,
      "loss": 0.0145,
      "step": 3987
    },
    {
      "epoch": 0.4615740740740741,
      "grad_norm": 0.16228719055652618,
      "learning_rate": 1.076851851851852e-05,
      "loss": 0.026,
      "step": 3988
    },
    {
      "epoch": 0.4616898148148148,
      "grad_norm": 0.07647637277841568,
      "learning_rate": 1.0766203703703704e-05,
      "loss": 0.0143,
      "step": 3989
    },
    {
      "epoch": 0.4618055555555556,
      "grad_norm": 0.1231859028339386,
      "learning_rate": 1.076388888888889e-05,
      "loss": 0.0206,
      "step": 3990
    },
    {
      "epoch": 0.4619212962962963,
      "grad_norm": 0.18111611902713776,
      "learning_rate": 1.0761574074074074e-05,
      "loss": 0.0249,
      "step": 3991
    },
    {
      "epoch": 0.462037037037037,
      "grad_norm": 0.6759688258171082,
      "learning_rate": 1.075925925925926e-05,
      "loss": 0.0279,
      "step": 3992
    },
    {
      "epoch": 0.4621527777777778,
      "grad_norm": 0.10944958031177521,
      "learning_rate": 1.0756944444444446e-05,
      "loss": 0.0199,
      "step": 3993
    },
    {
      "epoch": 0.4622685185185185,
      "grad_norm": 0.1306406557559967,
      "learning_rate": 1.0754629629629632e-05,
      "loss": 0.0227,
      "step": 3994
    },
    {
      "epoch": 0.46238425925925924,
      "grad_norm": 0.1872096061706543,
      "learning_rate": 1.0752314814814814e-05,
      "loss": 0.0245,
      "step": 3995
    },
    {
      "epoch": 0.4625,
      "grad_norm": 0.20925885438919067,
      "learning_rate": 1.075e-05,
      "loss": 0.0233,
      "step": 3996
    },
    {
      "epoch": 0.46261574074074074,
      "grad_norm": 0.09021860361099243,
      "learning_rate": 1.0747685185185186e-05,
      "loss": 0.0166,
      "step": 3997
    },
    {
      "epoch": 0.46273148148148147,
      "grad_norm": 0.07800447940826416,
      "learning_rate": 1.0745370370370372e-05,
      "loss": 0.0146,
      "step": 3998
    },
    {
      "epoch": 0.46284722222222224,
      "grad_norm": 0.12904669344425201,
      "learning_rate": 1.0743055555555557e-05,
      "loss": 0.021,
      "step": 3999
    },
    {
      "epoch": 0.46296296296296297,
      "grad_norm": 0.10505810379981995,
      "learning_rate": 1.0740740740740742e-05,
      "loss": 0.0194,
      "step": 4000
    },
    {
      "epoch": 0.4630787037037037,
      "grad_norm": 0.13671953976154327,
      "learning_rate": 1.0738425925925927e-05,
      "loss": 0.0187,
      "step": 4001
    },
    {
      "epoch": 0.46319444444444446,
      "grad_norm": 0.10016565769910812,
      "learning_rate": 1.0736111111111111e-05,
      "loss": 0.0184,
      "step": 4002
    },
    {
      "epoch": 0.4633101851851852,
      "grad_norm": 0.10470999032258987,
      "learning_rate": 1.0733796296296297e-05,
      "loss": 0.0193,
      "step": 4003
    },
    {
      "epoch": 0.4634259259259259,
      "grad_norm": 0.1505543291568756,
      "learning_rate": 1.0731481481481483e-05,
      "loss": 0.0157,
      "step": 4004
    },
    {
      "epoch": 0.4635416666666667,
      "grad_norm": 0.1291700005531311,
      "learning_rate": 1.0729166666666669e-05,
      "loss": 0.0246,
      "step": 4005
    },
    {
      "epoch": 0.4636574074074074,
      "grad_norm": 0.11869975924491882,
      "learning_rate": 1.0726851851851851e-05,
      "loss": 0.0224,
      "step": 4006
    },
    {
      "epoch": 0.46377314814814813,
      "grad_norm": 0.3200846016407013,
      "learning_rate": 1.0724537037037037e-05,
      "loss": 0.0196,
      "step": 4007
    },
    {
      "epoch": 0.4638888888888889,
      "grad_norm": 0.18895268440246582,
      "learning_rate": 1.0722222222222223e-05,
      "loss": 0.0279,
      "step": 4008
    },
    {
      "epoch": 0.46400462962962963,
      "grad_norm": 0.6467365622520447,
      "learning_rate": 1.0719907407407408e-05,
      "loss": 0.0203,
      "step": 4009
    },
    {
      "epoch": 0.46412037037037035,
      "grad_norm": 0.07503225654363632,
      "learning_rate": 1.0717592592592593e-05,
      "loss": 0.0142,
      "step": 4010
    },
    {
      "epoch": 0.4642361111111111,
      "grad_norm": 0.14200003445148468,
      "learning_rate": 1.071527777777778e-05,
      "loss": 0.0261,
      "step": 4011
    },
    {
      "epoch": 0.46435185185185185,
      "grad_norm": 33.06764221191406,
      "learning_rate": 1.0712962962962962e-05,
      "loss": 0.125,
      "step": 4012
    },
    {
      "epoch": 0.46446759259259257,
      "grad_norm": 0.0975736454129219,
      "learning_rate": 1.0710648148148148e-05,
      "loss": 0.0182,
      "step": 4013
    },
    {
      "epoch": 0.46458333333333335,
      "grad_norm": 0.10998646914958954,
      "learning_rate": 1.0708333333333334e-05,
      "loss": 0.0202,
      "step": 4014
    },
    {
      "epoch": 0.46469907407407407,
      "grad_norm": 0.2132967710494995,
      "learning_rate": 1.070601851851852e-05,
      "loss": 0.0252,
      "step": 4015
    },
    {
      "epoch": 0.4648148148148148,
      "grad_norm": 0.09264962375164032,
      "learning_rate": 1.0703703703703706e-05,
      "loss": 0.0166,
      "step": 4016
    },
    {
      "epoch": 0.46493055555555557,
      "grad_norm": 0.1474384367465973,
      "learning_rate": 1.070138888888889e-05,
      "loss": 0.0238,
      "step": 4017
    },
    {
      "epoch": 0.4650462962962963,
      "grad_norm": 0.10679400712251663,
      "learning_rate": 1.0699074074074074e-05,
      "loss": 0.0183,
      "step": 4018
    },
    {
      "epoch": 0.465162037037037,
      "grad_norm": 0.0999179482460022,
      "learning_rate": 1.069675925925926e-05,
      "loss": 0.0171,
      "step": 4019
    },
    {
      "epoch": 0.4652777777777778,
      "grad_norm": 68.00072479248047,
      "learning_rate": 1.0694444444444444e-05,
      "loss": 2.0223,
      "step": 4020
    },
    {
      "epoch": 0.4653935185185185,
      "grad_norm": 0.11347268521785736,
      "learning_rate": 1.069212962962963e-05,
      "loss": 0.0211,
      "step": 4021
    },
    {
      "epoch": 0.46550925925925923,
      "grad_norm": 0.13783518970012665,
      "learning_rate": 1.0689814814814816e-05,
      "loss": 0.0256,
      "step": 4022
    },
    {
      "epoch": 0.465625,
      "grad_norm": 0.11757403612136841,
      "learning_rate": 1.0687500000000002e-05,
      "loss": 0.0174,
      "step": 4023
    },
    {
      "epoch": 0.46574074074074073,
      "grad_norm": 1.2603366374969482,
      "learning_rate": 1.0685185185185185e-05,
      "loss": 0.0279,
      "step": 4024
    },
    {
      "epoch": 0.46585648148148145,
      "grad_norm": 0.13947542011737823,
      "learning_rate": 1.068287037037037e-05,
      "loss": 0.0215,
      "step": 4025
    },
    {
      "epoch": 0.46597222222222223,
      "grad_norm": 0.20342367887496948,
      "learning_rate": 1.0680555555555557e-05,
      "loss": 0.0275,
      "step": 4026
    },
    {
      "epoch": 0.46608796296296295,
      "grad_norm": 0.11195562034845352,
      "learning_rate": 1.0678240740740741e-05,
      "loss": 0.0208,
      "step": 4027
    },
    {
      "epoch": 0.46620370370370373,
      "grad_norm": 0.11192378401756287,
      "learning_rate": 1.0675925925925927e-05,
      "loss": 0.0203,
      "step": 4028
    },
    {
      "epoch": 0.46631944444444445,
      "grad_norm": 0.0837092399597168,
      "learning_rate": 1.0673611111111113e-05,
      "loss": 0.0156,
      "step": 4029
    },
    {
      "epoch": 0.4664351851851852,
      "grad_norm": 0.11667900532484055,
      "learning_rate": 1.0671296296296295e-05,
      "loss": 0.022,
      "step": 4030
    },
    {
      "epoch": 0.46655092592592595,
      "grad_norm": 3.084756374359131,
      "learning_rate": 1.0668981481481481e-05,
      "loss": 0.0396,
      "step": 4031
    },
    {
      "epoch": 0.4666666666666667,
      "grad_norm": 0.10342114418745041,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 0.014,
      "step": 4032
    },
    {
      "epoch": 0.4667824074074074,
      "grad_norm": 1.07886803150177,
      "learning_rate": 1.0664351851851853e-05,
      "loss": 0.0335,
      "step": 4033
    },
    {
      "epoch": 0.4668981481481482,
      "grad_norm": 0.10614743083715439,
      "learning_rate": 1.066203703703704e-05,
      "loss": 0.0185,
      "step": 4034
    },
    {
      "epoch": 0.4670138888888889,
      "grad_norm": 0.0990927666425705,
      "learning_rate": 1.0659722222222223e-05,
      "loss": 0.0183,
      "step": 4035
    },
    {
      "epoch": 0.4671296296296296,
      "grad_norm": 29.958303451538086,
      "learning_rate": 1.0657407407407408e-05,
      "loss": 2.6551,
      "step": 4036
    },
    {
      "epoch": 0.4672453703703704,
      "grad_norm": 0.20601125061511993,
      "learning_rate": 1.0655092592592594e-05,
      "loss": 0.0167,
      "step": 4037
    },
    {
      "epoch": 0.4673611111111111,
      "grad_norm": 0.08201645314693451,
      "learning_rate": 1.0652777777777778e-05,
      "loss": 0.0154,
      "step": 4038
    },
    {
      "epoch": 0.46747685185185184,
      "grad_norm": 0.1042163223028183,
      "learning_rate": 1.0650462962962964e-05,
      "loss": 0.0185,
      "step": 4039
    },
    {
      "epoch": 0.4675925925925926,
      "grad_norm": 1.6122264862060547,
      "learning_rate": 1.064814814814815e-05,
      "loss": 0.0357,
      "step": 4040
    },
    {
      "epoch": 0.46770833333333334,
      "grad_norm": 0.09652476757764816,
      "learning_rate": 1.0645833333333336e-05,
      "loss": 0.0178,
      "step": 4041
    },
    {
      "epoch": 0.46782407407407406,
      "grad_norm": 0.12210430204868317,
      "learning_rate": 1.0643518518518518e-05,
      "loss": 0.0218,
      "step": 4042
    },
    {
      "epoch": 0.46793981481481484,
      "grad_norm": 0.10606452822685242,
      "learning_rate": 1.0641203703703704e-05,
      "loss": 0.0194,
      "step": 4043
    },
    {
      "epoch": 0.46805555555555556,
      "grad_norm": 28.40017318725586,
      "learning_rate": 1.063888888888889e-05,
      "loss": 3.0332,
      "step": 4044
    },
    {
      "epoch": 0.4681712962962963,
      "grad_norm": 0.08816143870353699,
      "learning_rate": 1.0636574074074074e-05,
      "loss": 0.0159,
      "step": 4045
    },
    {
      "epoch": 0.46828703703703706,
      "grad_norm": 0.11437007039785385,
      "learning_rate": 1.063425925925926e-05,
      "loss": 0.0209,
      "step": 4046
    },
    {
      "epoch": 0.4684027777777778,
      "grad_norm": 0.17604805529117584,
      "learning_rate": 1.0631944444444446e-05,
      "loss": 0.0203,
      "step": 4047
    },
    {
      "epoch": 0.4685185185185185,
      "grad_norm": 0.6065865755081177,
      "learning_rate": 1.062962962962963e-05,
      "loss": 0.0206,
      "step": 4048
    },
    {
      "epoch": 0.4686342592592593,
      "grad_norm": 0.07436545938253403,
      "learning_rate": 1.0627314814814815e-05,
      "loss": 0.0139,
      "step": 4049
    },
    {
      "epoch": 0.46875,
      "grad_norm": 20.823009490966797,
      "learning_rate": 1.0625e-05,
      "loss": 0.0874,
      "step": 4050
    },
    {
      "epoch": 0.4688657407407407,
      "grad_norm": 0.09813568741083145,
      "learning_rate": 1.0622685185185187e-05,
      "loss": 0.0181,
      "step": 4051
    },
    {
      "epoch": 0.4689814814814815,
      "grad_norm": 0.11022523790597916,
      "learning_rate": 1.0620370370370373e-05,
      "loss": 0.0206,
      "step": 4052
    },
    {
      "epoch": 0.4690972222222222,
      "grad_norm": 0.07592269033193588,
      "learning_rate": 1.0618055555555555e-05,
      "loss": 0.0141,
      "step": 4053
    },
    {
      "epoch": 0.46921296296296294,
      "grad_norm": 0.15097729861736298,
      "learning_rate": 1.0615740740740741e-05,
      "loss": 0.0206,
      "step": 4054
    },
    {
      "epoch": 0.4693287037037037,
      "grad_norm": 0.11964628100395203,
      "learning_rate": 1.0613425925925927e-05,
      "loss": 0.0212,
      "step": 4055
    },
    {
      "epoch": 0.46944444444444444,
      "grad_norm": 0.23896963894367218,
      "learning_rate": 1.0611111111111111e-05,
      "loss": 0.0219,
      "step": 4056
    },
    {
      "epoch": 0.46956018518518516,
      "grad_norm": 0.07554970681667328,
      "learning_rate": 1.0608796296296297e-05,
      "loss": 0.0139,
      "step": 4057
    },
    {
      "epoch": 0.46967592592592594,
      "grad_norm": 0.14222325384616852,
      "learning_rate": 1.0606481481481483e-05,
      "loss": 0.0194,
      "step": 4058
    },
    {
      "epoch": 0.46979166666666666,
      "grad_norm": 0.1373981386423111,
      "learning_rate": 1.0604166666666666e-05,
      "loss": 0.0252,
      "step": 4059
    },
    {
      "epoch": 0.4699074074074074,
      "grad_norm": 0.16774892807006836,
      "learning_rate": 1.0601851851851852e-05,
      "loss": 0.0278,
      "step": 4060
    },
    {
      "epoch": 0.47002314814814816,
      "grad_norm": 0.12560710310935974,
      "learning_rate": 1.0599537037037038e-05,
      "loss": 0.0227,
      "step": 4061
    },
    {
      "epoch": 0.4701388888888889,
      "grad_norm": 0.1255662888288498,
      "learning_rate": 1.0597222222222224e-05,
      "loss": 0.0238,
      "step": 4062
    },
    {
      "epoch": 0.4702546296296296,
      "grad_norm": 0.11852926760911942,
      "learning_rate": 1.059490740740741e-05,
      "loss": 0.0215,
      "step": 4063
    },
    {
      "epoch": 0.4703703703703704,
      "grad_norm": 0.09666919708251953,
      "learning_rate": 1.0592592592592594e-05,
      "loss": 0.0177,
      "step": 4064
    },
    {
      "epoch": 0.4704861111111111,
      "grad_norm": 0.12008029222488403,
      "learning_rate": 1.0590277777777778e-05,
      "loss": 0.0223,
      "step": 4065
    },
    {
      "epoch": 0.4706018518518518,
      "grad_norm": 95.04569244384766,
      "learning_rate": 1.0587962962962964e-05,
      "loss": 0.7922,
      "step": 4066
    },
    {
      "epoch": 0.4707175925925926,
      "grad_norm": 0.13246631622314453,
      "learning_rate": 1.0585648148148148e-05,
      "loss": 0.024,
      "step": 4067
    },
    {
      "epoch": 0.4708333333333333,
      "grad_norm": 0.126754492521286,
      "learning_rate": 1.0583333333333334e-05,
      "loss": 0.0164,
      "step": 4068
    },
    {
      "epoch": 0.47094907407407405,
      "grad_norm": 1.5041484832763672,
      "learning_rate": 1.058101851851852e-05,
      "loss": 0.0298,
      "step": 4069
    },
    {
      "epoch": 0.4710648148148148,
      "grad_norm": 0.09528118371963501,
      "learning_rate": 1.0578703703703706e-05,
      "loss": 0.0175,
      "step": 4070
    },
    {
      "epoch": 0.47118055555555555,
      "grad_norm": 93.73954772949219,
      "learning_rate": 1.0576388888888889e-05,
      "loss": 1.7063,
      "step": 4071
    },
    {
      "epoch": 0.47129629629629627,
      "grad_norm": 0.11405787616968155,
      "learning_rate": 1.0574074074074075e-05,
      "loss": 0.0216,
      "step": 4072
    },
    {
      "epoch": 0.47141203703703705,
      "grad_norm": 158.95068359375,
      "learning_rate": 1.057175925925926e-05,
      "loss": 1.3339,
      "step": 4073
    },
    {
      "epoch": 0.47152777777777777,
      "grad_norm": 0.3720417022705078,
      "learning_rate": 1.0569444444444445e-05,
      "loss": 0.0244,
      "step": 4074
    },
    {
      "epoch": 0.47164351851851855,
      "grad_norm": 0.11740004271268845,
      "learning_rate": 1.056712962962963e-05,
      "loss": 0.0211,
      "step": 4075
    },
    {
      "epoch": 0.47175925925925927,
      "grad_norm": 7.851602554321289,
      "learning_rate": 1.0564814814814817e-05,
      "loss": 2.7567,
      "step": 4076
    },
    {
      "epoch": 0.471875,
      "grad_norm": 0.07701098918914795,
      "learning_rate": 1.05625e-05,
      "loss": 0.0139,
      "step": 4077
    },
    {
      "epoch": 0.47199074074074077,
      "grad_norm": 0.08172477781772614,
      "learning_rate": 1.0560185185185185e-05,
      "loss": 0.0153,
      "step": 4078
    },
    {
      "epoch": 0.4721064814814815,
      "grad_norm": 0.15487468242645264,
      "learning_rate": 1.0557870370370371e-05,
      "loss": 0.0227,
      "step": 4079
    },
    {
      "epoch": 0.4722222222222222,
      "grad_norm": 0.13830912113189697,
      "learning_rate": 1.0555555555555557e-05,
      "loss": 0.0247,
      "step": 4080
    },
    {
      "epoch": 0.472337962962963,
      "grad_norm": 0.1571195423603058,
      "learning_rate": 1.0553240740740743e-05,
      "loss": 0.0272,
      "step": 4081
    },
    {
      "epoch": 0.4724537037037037,
      "grad_norm": 0.11442156881093979,
      "learning_rate": 1.0550925925925927e-05,
      "loss": 0.0143,
      "step": 4082
    },
    {
      "epoch": 0.47256944444444443,
      "grad_norm": 1.090802550315857,
      "learning_rate": 1.0548611111111112e-05,
      "loss": 0.0291,
      "step": 4083
    },
    {
      "epoch": 0.4726851851851852,
      "grad_norm": 0.11272617429494858,
      "learning_rate": 1.0546296296296297e-05,
      "loss": 0.0209,
      "step": 4084
    },
    {
      "epoch": 0.47280092592592593,
      "grad_norm": 0.08613356947898865,
      "learning_rate": 1.0543981481481482e-05,
      "loss": 0.016,
      "step": 4085
    },
    {
      "epoch": 0.47291666666666665,
      "grad_norm": 0.10590124875307083,
      "learning_rate": 1.0541666666666668e-05,
      "loss": 0.0141,
      "step": 4086
    },
    {
      "epoch": 0.47303240740740743,
      "grad_norm": 95.40052032470703,
      "learning_rate": 1.0539351851851854e-05,
      "loss": 1.49,
      "step": 4087
    },
    {
      "epoch": 0.47314814814814815,
      "grad_norm": 0.09941788762807846,
      "learning_rate": 1.053703703703704e-05,
      "loss": 0.0186,
      "step": 4088
    },
    {
      "epoch": 0.4732638888888889,
      "grad_norm": 0.08696864545345306,
      "learning_rate": 1.0534722222222222e-05,
      "loss": 0.016,
      "step": 4089
    },
    {
      "epoch": 0.47337962962962965,
      "grad_norm": 0.08658922463655472,
      "learning_rate": 1.0532407407407408e-05,
      "loss": 0.0157,
      "step": 4090
    },
    {
      "epoch": 0.47349537037037037,
      "grad_norm": 0.08388704061508179,
      "learning_rate": 1.0530092592592594e-05,
      "loss": 0.0156,
      "step": 4091
    },
    {
      "epoch": 0.4736111111111111,
      "grad_norm": 0.0858757272362709,
      "learning_rate": 1.0527777777777778e-05,
      "loss": 0.0156,
      "step": 4092
    },
    {
      "epoch": 0.47372685185185187,
      "grad_norm": 0.1009332537651062,
      "learning_rate": 1.0525462962962964e-05,
      "loss": 0.0137,
      "step": 4093
    },
    {
      "epoch": 0.4738425925925926,
      "grad_norm": 19.541461944580078,
      "learning_rate": 1.052314814814815e-05,
      "loss": 2.9241,
      "step": 4094
    },
    {
      "epoch": 0.4739583333333333,
      "grad_norm": 0.7109378576278687,
      "learning_rate": 1.0520833333333333e-05,
      "loss": 0.0299,
      "step": 4095
    },
    {
      "epoch": 0.4740740740740741,
      "grad_norm": 0.07284723967313766,
      "learning_rate": 1.0518518518518519e-05,
      "loss": 0.0137,
      "step": 4096
    },
    {
      "epoch": 0.4741898148148148,
      "grad_norm": 0.09658624976873398,
      "learning_rate": 1.0516203703703705e-05,
      "loss": 0.0177,
      "step": 4097
    },
    {
      "epoch": 0.47430555555555554,
      "grad_norm": 0.4061340093612671,
      "learning_rate": 1.051388888888889e-05,
      "loss": 0.0223,
      "step": 4098
    },
    {
      "epoch": 0.4744212962962963,
      "grad_norm": 0.384358674287796,
      "learning_rate": 1.0511574074074076e-05,
      "loss": 0.0288,
      "step": 4099
    },
    {
      "epoch": 0.47453703703703703,
      "grad_norm": 0.13159826397895813,
      "learning_rate": 1.0509259259259259e-05,
      "loss": 0.024,
      "step": 4100
    },
    {
      "epoch": 0.47465277777777776,
      "grad_norm": 12.355664253234863,
      "learning_rate": 1.0506944444444445e-05,
      "loss": 2.8608,
      "step": 4101
    },
    {
      "epoch": 0.47476851851851853,
      "grad_norm": 0.15057304501533508,
      "learning_rate": 1.0504629629629631e-05,
      "loss": 0.0271,
      "step": 4102
    },
    {
      "epoch": 0.47488425925925926,
      "grad_norm": 0.10225097835063934,
      "learning_rate": 1.0502314814814815e-05,
      "loss": 0.0138,
      "step": 4103
    },
    {
      "epoch": 0.475,
      "grad_norm": 0.11916415393352509,
      "learning_rate": 1.0500000000000001e-05,
      "loss": 0.0224,
      "step": 4104
    },
    {
      "epoch": 0.47511574074074076,
      "grad_norm": 0.09768945723772049,
      "learning_rate": 1.0497685185185187e-05,
      "loss": 0.0133,
      "step": 4105
    },
    {
      "epoch": 0.4752314814814815,
      "grad_norm": 0.10017058998346329,
      "learning_rate": 1.049537037037037e-05,
      "loss": 0.0183,
      "step": 4106
    },
    {
      "epoch": 0.4753472222222222,
      "grad_norm": 0.08164694905281067,
      "learning_rate": 1.0493055555555556e-05,
      "loss": 0.0152,
      "step": 4107
    },
    {
      "epoch": 0.475462962962963,
      "grad_norm": 32.24624252319336,
      "learning_rate": 1.0490740740740742e-05,
      "loss": 0.1002,
      "step": 4108
    },
    {
      "epoch": 0.4755787037037037,
      "grad_norm": 7.395959854125977,
      "learning_rate": 1.0488425925925927e-05,
      "loss": 3.1689,
      "step": 4109
    },
    {
      "epoch": 0.4756944444444444,
      "grad_norm": 0.08567175269126892,
      "learning_rate": 1.0486111111111112e-05,
      "loss": 0.0159,
      "step": 4110
    },
    {
      "epoch": 0.4758101851851852,
      "grad_norm": 0.09713199734687805,
      "learning_rate": 1.0483796296296298e-05,
      "loss": 0.0178,
      "step": 4111
    },
    {
      "epoch": 0.4759259259259259,
      "grad_norm": 3.042421579360962,
      "learning_rate": 1.0481481481481482e-05,
      "loss": 0.0372,
      "step": 4112
    },
    {
      "epoch": 0.47604166666666664,
      "grad_norm": 1.5897252559661865,
      "learning_rate": 1.0479166666666666e-05,
      "loss": 0.0305,
      "step": 4113
    },
    {
      "epoch": 0.4761574074074074,
      "grad_norm": 0.14436165988445282,
      "learning_rate": 1.0476851851851852e-05,
      "loss": 0.0265,
      "step": 4114
    },
    {
      "epoch": 0.47627314814814814,
      "grad_norm": 0.12331680208444595,
      "learning_rate": 1.0474537037037038e-05,
      "loss": 0.0212,
      "step": 4115
    },
    {
      "epoch": 0.47638888888888886,
      "grad_norm": 0.07671651989221573,
      "learning_rate": 1.0472222222222224e-05,
      "loss": 0.014,
      "step": 4116
    },
    {
      "epoch": 0.47650462962962964,
      "grad_norm": 9.48388385772705,
      "learning_rate": 1.046990740740741e-05,
      "loss": 2.6467,
      "step": 4117
    },
    {
      "epoch": 0.47662037037037036,
      "grad_norm": 0.5615820288658142,
      "learning_rate": 1.0467592592592592e-05,
      "loss": 0.0235,
      "step": 4118
    },
    {
      "epoch": 0.47673611111111114,
      "grad_norm": 0.11731255799531937,
      "learning_rate": 1.0465277777777778e-05,
      "loss": 0.0208,
      "step": 4119
    },
    {
      "epoch": 0.47685185185185186,
      "grad_norm": 0.11541464924812317,
      "learning_rate": 1.0462962962962964e-05,
      "loss": 0.0213,
      "step": 4120
    },
    {
      "epoch": 0.4769675925925926,
      "grad_norm": 0.17102785408496857,
      "learning_rate": 1.0460648148148149e-05,
      "loss": 0.0287,
      "step": 4121
    },
    {
      "epoch": 0.47708333333333336,
      "grad_norm": 0.14667555689811707,
      "learning_rate": 1.0458333333333335e-05,
      "loss": 0.0167,
      "step": 4122
    },
    {
      "epoch": 0.4771990740740741,
      "grad_norm": 0.08462481945753098,
      "learning_rate": 1.045601851851852e-05,
      "loss": 0.0149,
      "step": 4123
    },
    {
      "epoch": 0.4773148148148148,
      "grad_norm": 0.10648340731859207,
      "learning_rate": 1.0453703703703703e-05,
      "loss": 0.0195,
      "step": 4124
    },
    {
      "epoch": 0.4774305555555556,
      "grad_norm": 32.14756774902344,
      "learning_rate": 1.0451388888888889e-05,
      "loss": 0.1626,
      "step": 4125
    },
    {
      "epoch": 0.4775462962962963,
      "grad_norm": 0.1161629781126976,
      "learning_rate": 1.0449074074074075e-05,
      "loss": 0.0214,
      "step": 4126
    },
    {
      "epoch": 0.477662037037037,
      "grad_norm": 0.7099695205688477,
      "learning_rate": 1.0446759259259261e-05,
      "loss": 0.0222,
      "step": 4127
    },
    {
      "epoch": 0.4777777777777778,
      "grad_norm": 0.10855316370725632,
      "learning_rate": 1.0444444444444445e-05,
      "loss": 0.0195,
      "step": 4128
    },
    {
      "epoch": 0.4778935185185185,
      "grad_norm": 223.87350463867188,
      "learning_rate": 1.0442129629629631e-05,
      "loss": 1.7702,
      "step": 4129
    },
    {
      "epoch": 0.47800925925925924,
      "grad_norm": 0.12065384536981583,
      "learning_rate": 1.0439814814814815e-05,
      "loss": 0.0217,
      "step": 4130
    },
    {
      "epoch": 0.478125,
      "grad_norm": 0.22023679316043854,
      "learning_rate": 1.04375e-05,
      "loss": 0.0201,
      "step": 4131
    },
    {
      "epoch": 0.47824074074074074,
      "grad_norm": 0.10944956541061401,
      "learning_rate": 1.0435185185185186e-05,
      "loss": 0.0197,
      "step": 4132
    },
    {
      "epoch": 0.47835648148148147,
      "grad_norm": 0.12838374078273773,
      "learning_rate": 1.0432870370370371e-05,
      "loss": 0.0244,
      "step": 4133
    },
    {
      "epoch": 0.47847222222222224,
      "grad_norm": 0.1186666414141655,
      "learning_rate": 1.0430555555555557e-05,
      "loss": 0.0219,
      "step": 4134
    },
    {
      "epoch": 0.47858796296296297,
      "grad_norm": 0.13252362608909607,
      "learning_rate": 1.0428240740740743e-05,
      "loss": 0.0214,
      "step": 4135
    },
    {
      "epoch": 0.4787037037037037,
      "grad_norm": 0.10593701153993607,
      "learning_rate": 1.0425925925925926e-05,
      "loss": 0.0194,
      "step": 4136
    },
    {
      "epoch": 0.47881944444444446,
      "grad_norm": 0.17455466091632843,
      "learning_rate": 1.0423611111111112e-05,
      "loss": 0.0242,
      "step": 4137
    },
    {
      "epoch": 0.4789351851851852,
      "grad_norm": 0.36337339878082275,
      "learning_rate": 1.0421296296296298e-05,
      "loss": 0.0283,
      "step": 4138
    },
    {
      "epoch": 0.4790509259259259,
      "grad_norm": 0.10535754263401031,
      "learning_rate": 1.0418981481481482e-05,
      "loss": 0.0191,
      "step": 4139
    },
    {
      "epoch": 0.4791666666666667,
      "grad_norm": 0.284809410572052,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 0.0227,
      "step": 4140
    },
    {
      "epoch": 0.4792824074074074,
      "grad_norm": 0.09753328561782837,
      "learning_rate": 1.0414351851851852e-05,
      "loss": 0.0184,
      "step": 4141
    },
    {
      "epoch": 0.47939814814814813,
      "grad_norm": 0.12550705671310425,
      "learning_rate": 1.0412037037037037e-05,
      "loss": 0.0168,
      "step": 4142
    },
    {
      "epoch": 0.4795138888888889,
      "grad_norm": 0.11597038805484772,
      "learning_rate": 1.0409722222222222e-05,
      "loss": 0.0201,
      "step": 4143
    },
    {
      "epoch": 0.47962962962962963,
      "grad_norm": 0.0915040373802185,
      "learning_rate": 1.0407407407407408e-05,
      "loss": 0.0169,
      "step": 4144
    },
    {
      "epoch": 0.47974537037037035,
      "grad_norm": 0.15830059349536896,
      "learning_rate": 1.0405092592592594e-05,
      "loss": 0.0253,
      "step": 4145
    },
    {
      "epoch": 0.4798611111111111,
      "grad_norm": 0.12133366614580154,
      "learning_rate": 1.0402777777777779e-05,
      "loss": 0.0211,
      "step": 4146
    },
    {
      "epoch": 0.47997685185185185,
      "grad_norm": 0.11790358275175095,
      "learning_rate": 1.0400462962962963e-05,
      "loss": 0.0219,
      "step": 4147
    },
    {
      "epoch": 0.48009259259259257,
      "grad_norm": 0.07761295139789581,
      "learning_rate": 1.0398148148148149e-05,
      "loss": 0.0141,
      "step": 4148
    },
    {
      "epoch": 0.48020833333333335,
      "grad_norm": 0.1144060492515564,
      "learning_rate": 1.0395833333333333e-05,
      "loss": 0.0206,
      "step": 4149
    },
    {
      "epoch": 0.48032407407407407,
      "grad_norm": 0.11324647814035416,
      "learning_rate": 1.0393518518518519e-05,
      "loss": 0.0208,
      "step": 4150
    },
    {
      "epoch": 0.4804398148148148,
      "grad_norm": 0.12248943746089935,
      "learning_rate": 1.0391203703703705e-05,
      "loss": 0.0221,
      "step": 4151
    },
    {
      "epoch": 0.48055555555555557,
      "grad_norm": 0.13787220418453217,
      "learning_rate": 1.0388888888888891e-05,
      "loss": 0.0199,
      "step": 4152
    },
    {
      "epoch": 0.4806712962962963,
      "grad_norm": 0.11591138690710068,
      "learning_rate": 1.0386574074074073e-05,
      "loss": 0.0205,
      "step": 4153
    },
    {
      "epoch": 0.480787037037037,
      "grad_norm": 7.548290729522705,
      "learning_rate": 1.038425925925926e-05,
      "loss": 0.0449,
      "step": 4154
    },
    {
      "epoch": 0.4809027777777778,
      "grad_norm": 0.4263326823711395,
      "learning_rate": 1.0381944444444445e-05,
      "loss": 0.0197,
      "step": 4155
    },
    {
      "epoch": 0.4810185185185185,
      "grad_norm": 0.08296448737382889,
      "learning_rate": 1.0379629629629631e-05,
      "loss": 0.0152,
      "step": 4156
    },
    {
      "epoch": 0.48113425925925923,
      "grad_norm": 0.11306829750537872,
      "learning_rate": 1.0377314814814816e-05,
      "loss": 0.0191,
      "step": 4157
    },
    {
      "epoch": 0.48125,
      "grad_norm": 0.12877404689788818,
      "learning_rate": 1.0375000000000001e-05,
      "loss": 0.0231,
      "step": 4158
    },
    {
      "epoch": 0.48136574074074073,
      "grad_norm": 0.18649259209632874,
      "learning_rate": 1.0372685185185186e-05,
      "loss": 0.0236,
      "step": 4159
    },
    {
      "epoch": 0.48148148148148145,
      "grad_norm": 0.12545247375965118,
      "learning_rate": 1.037037037037037e-05,
      "loss": 0.0208,
      "step": 4160
    },
    {
      "epoch": 0.48159722222222223,
      "grad_norm": 3.541783094406128,
      "learning_rate": 1.0368055555555556e-05,
      "loss": 0.0387,
      "step": 4161
    },
    {
      "epoch": 0.48171296296296295,
      "grad_norm": 0.7002384662628174,
      "learning_rate": 1.0365740740740742e-05,
      "loss": 0.0288,
      "step": 4162
    },
    {
      "epoch": 0.48182870370370373,
      "grad_norm": 0.09213046729564667,
      "learning_rate": 1.0363425925925928e-05,
      "loss": 0.0169,
      "step": 4163
    },
    {
      "epoch": 0.48194444444444445,
      "grad_norm": 0.1664412021636963,
      "learning_rate": 1.0361111111111114e-05,
      "loss": 0.0228,
      "step": 4164
    },
    {
      "epoch": 0.4820601851851852,
      "grad_norm": 0.08208481967449188,
      "learning_rate": 1.0358796296296296e-05,
      "loss": 0.0154,
      "step": 4165
    },
    {
      "epoch": 0.48217592592592595,
      "grad_norm": 0.09122107177972794,
      "learning_rate": 1.0356481481481482e-05,
      "loss": 0.0161,
      "step": 4166
    },
    {
      "epoch": 0.4822916666666667,
      "grad_norm": 0.11775888502597809,
      "learning_rate": 1.0354166666666668e-05,
      "loss": 0.0221,
      "step": 4167
    },
    {
      "epoch": 0.4824074074074074,
      "grad_norm": 0.09643823653459549,
      "learning_rate": 1.0351851851851852e-05,
      "loss": 0.0176,
      "step": 4168
    },
    {
      "epoch": 0.4825231481481482,
      "grad_norm": 0.10790538042783737,
      "learning_rate": 1.0349537037037038e-05,
      "loss": 0.0157,
      "step": 4169
    },
    {
      "epoch": 0.4826388888888889,
      "grad_norm": 0.313679039478302,
      "learning_rate": 1.0347222222222224e-05,
      "loss": 0.0272,
      "step": 4170
    },
    {
      "epoch": 0.4827546296296296,
      "grad_norm": 0.10596265643835068,
      "learning_rate": 1.0344907407407407e-05,
      "loss": 0.0195,
      "step": 4171
    },
    {
      "epoch": 0.4828703703703704,
      "grad_norm": 0.39268162846565247,
      "learning_rate": 1.0342592592592593e-05,
      "loss": 0.028,
      "step": 4172
    },
    {
      "epoch": 0.4829861111111111,
      "grad_norm": 0.12059903889894485,
      "learning_rate": 1.0340277777777779e-05,
      "loss": 0.0223,
      "step": 4173
    },
    {
      "epoch": 0.48310185185185184,
      "grad_norm": 0.10972151905298233,
      "learning_rate": 1.0337962962962965e-05,
      "loss": 0.0199,
      "step": 4174
    },
    {
      "epoch": 0.4832175925925926,
      "grad_norm": 0.10200131684541702,
      "learning_rate": 1.0335648148148149e-05,
      "loss": 0.0189,
      "step": 4175
    },
    {
      "epoch": 0.48333333333333334,
      "grad_norm": 0.11089625954627991,
      "learning_rate": 1.0333333333333335e-05,
      "loss": 0.0205,
      "step": 4176
    },
    {
      "epoch": 0.48344907407407406,
      "grad_norm": 0.10872781276702881,
      "learning_rate": 1.033101851851852e-05,
      "loss": 0.0195,
      "step": 4177
    },
    {
      "epoch": 0.48356481481481484,
      "grad_norm": 0.08067726343870163,
      "learning_rate": 1.0328703703703703e-05,
      "loss": 0.0151,
      "step": 4178
    },
    {
      "epoch": 0.48368055555555556,
      "grad_norm": 0.0947420671582222,
      "learning_rate": 1.032638888888889e-05,
      "loss": 0.0155,
      "step": 4179
    },
    {
      "epoch": 0.4837962962962963,
      "grad_norm": 25.887521743774414,
      "learning_rate": 1.0324074074074075e-05,
      "loss": 2.2436,
      "step": 4180
    },
    {
      "epoch": 0.48391203703703706,
      "grad_norm": 0.09831897169351578,
      "learning_rate": 1.0321759259259261e-05,
      "loss": 0.015,
      "step": 4181
    },
    {
      "epoch": 0.4840277777777778,
      "grad_norm": 0.07027262449264526,
      "learning_rate": 1.0319444444444447e-05,
      "loss": 0.0132,
      "step": 4182
    },
    {
      "epoch": 0.4841435185185185,
      "grad_norm": 0.11727677285671234,
      "learning_rate": 1.031712962962963e-05,
      "loss": 0.0199,
      "step": 4183
    },
    {
      "epoch": 0.4842592592592593,
      "grad_norm": 0.07759709656238556,
      "learning_rate": 1.0314814814814816e-05,
      "loss": 0.0146,
      "step": 4184
    },
    {
      "epoch": 0.484375,
      "grad_norm": 0.10680066049098969,
      "learning_rate": 1.0312500000000002e-05,
      "loss": 0.0198,
      "step": 4185
    },
    {
      "epoch": 0.4844907407407407,
      "grad_norm": 62.462791442871094,
      "learning_rate": 1.0310185185185186e-05,
      "loss": 2.4875,
      "step": 4186
    },
    {
      "epoch": 0.4846064814814815,
      "grad_norm": 0.09726156294345856,
      "learning_rate": 1.0307870370370372e-05,
      "loss": 0.0176,
      "step": 4187
    },
    {
      "epoch": 0.4847222222222222,
      "grad_norm": 0.1186816617846489,
      "learning_rate": 1.0305555555555556e-05,
      "loss": 0.0159,
      "step": 4188
    },
    {
      "epoch": 0.48483796296296294,
      "grad_norm": 0.11855307221412659,
      "learning_rate": 1.030324074074074e-05,
      "loss": 0.019,
      "step": 4189
    },
    {
      "epoch": 0.4849537037037037,
      "grad_norm": 0.11984767019748688,
      "learning_rate": 1.0300925925925926e-05,
      "loss": 0.0223,
      "step": 4190
    },
    {
      "epoch": 0.48506944444444444,
      "grad_norm": 0.06948143243789673,
      "learning_rate": 1.0298611111111112e-05,
      "loss": 0.0131,
      "step": 4191
    },
    {
      "epoch": 0.48518518518518516,
      "grad_norm": 0.4022732079029083,
      "learning_rate": 1.0296296296296298e-05,
      "loss": 0.0215,
      "step": 4192
    },
    {
      "epoch": 0.48530092592592594,
      "grad_norm": 0.10080153495073318,
      "learning_rate": 1.0293981481481482e-05,
      "loss": 0.0135,
      "step": 4193
    },
    {
      "epoch": 0.48541666666666666,
      "grad_norm": 0.12401031702756882,
      "learning_rate": 1.0291666666666667e-05,
      "loss": 0.0212,
      "step": 4194
    },
    {
      "epoch": 0.4855324074074074,
      "grad_norm": 0.07503242790699005,
      "learning_rate": 1.0289351851851853e-05,
      "loss": 0.0141,
      "step": 4195
    },
    {
      "epoch": 0.48564814814814816,
      "grad_norm": 0.07685462385416031,
      "learning_rate": 1.0287037037037037e-05,
      "loss": 0.0144,
      "step": 4196
    },
    {
      "epoch": 0.4857638888888889,
      "grad_norm": 0.16339941322803497,
      "learning_rate": 1.0284722222222223e-05,
      "loss": 0.0217,
      "step": 4197
    },
    {
      "epoch": 0.4858796296296296,
      "grad_norm": 7.318121910095215,
      "learning_rate": 1.0282407407407409e-05,
      "loss": 2.5677,
      "step": 4198
    },
    {
      "epoch": 0.4859953703703704,
      "grad_norm": 0.0774802640080452,
      "learning_rate": 1.0280092592592595e-05,
      "loss": 0.0146,
      "step": 4199
    },
    {
      "epoch": 0.4861111111111111,
      "grad_norm": 0.08173125982284546,
      "learning_rate": 1.0277777777777777e-05,
      "loss": 0.0146,
      "step": 4200
    },
    {
      "epoch": 0.4862268518518518,
      "grad_norm": 0.1132226511836052,
      "learning_rate": 1.0275462962962963e-05,
      "loss": 0.0196,
      "step": 4201
    },
    {
      "epoch": 0.4863425925925926,
      "grad_norm": 0.10227005183696747,
      "learning_rate": 1.0273148148148149e-05,
      "loss": 0.0187,
      "step": 4202
    },
    {
      "epoch": 0.4864583333333333,
      "grad_norm": 0.10174665600061417,
      "learning_rate": 1.0270833333333335e-05,
      "loss": 0.0179,
      "step": 4203
    },
    {
      "epoch": 0.48657407407407405,
      "grad_norm": 0.07000936567783356,
      "learning_rate": 1.026851851851852e-05,
      "loss": 0.0131,
      "step": 4204
    },
    {
      "epoch": 0.4866898148148148,
      "grad_norm": 0.09348136186599731,
      "learning_rate": 1.0266203703703705e-05,
      "loss": 0.0171,
      "step": 4205
    },
    {
      "epoch": 0.48680555555555555,
      "grad_norm": 0.11216558516025543,
      "learning_rate": 1.026388888888889e-05,
      "loss": 0.0189,
      "step": 4206
    },
    {
      "epoch": 0.48692129629629627,
      "grad_norm": 0.13793759047985077,
      "learning_rate": 1.0261574074074074e-05,
      "loss": 0.0213,
      "step": 4207
    },
    {
      "epoch": 0.48703703703703705,
      "grad_norm": 71.21701049804688,
      "learning_rate": 1.025925925925926e-05,
      "loss": 1.6803,
      "step": 4208
    },
    {
      "epoch": 0.48715277777777777,
      "grad_norm": 0.35071688890457153,
      "learning_rate": 1.0256944444444446e-05,
      "loss": 0.0273,
      "step": 4209
    },
    {
      "epoch": 0.48726851851851855,
      "grad_norm": 0.509878396987915,
      "learning_rate": 1.0254629629629632e-05,
      "loss": 0.0279,
      "step": 4210
    },
    {
      "epoch": 0.48738425925925927,
      "grad_norm": 0.11043407768011093,
      "learning_rate": 1.0252314814814816e-05,
      "loss": 0.0199,
      "step": 4211
    },
    {
      "epoch": 0.4875,
      "grad_norm": 0.09930076450109482,
      "learning_rate": 1.025e-05,
      "loss": 0.0181,
      "step": 4212
    },
    {
      "epoch": 0.48761574074074077,
      "grad_norm": 0.12296969443559647,
      "learning_rate": 1.0247685185185186e-05,
      "loss": 0.0196,
      "step": 4213
    },
    {
      "epoch": 0.4877314814814815,
      "grad_norm": 5.454726219177246,
      "learning_rate": 1.024537037037037e-05,
      "loss": 0.0339,
      "step": 4214
    },
    {
      "epoch": 0.4878472222222222,
      "grad_norm": 0.09422572702169418,
      "learning_rate": 1.0243055555555556e-05,
      "loss": 0.0172,
      "step": 4215
    },
    {
      "epoch": 0.487962962962963,
      "grad_norm": 0.11365517973899841,
      "learning_rate": 1.0240740740740742e-05,
      "loss": 0.0202,
      "step": 4216
    },
    {
      "epoch": 0.4880787037037037,
      "grad_norm": 0.10225965827703476,
      "learning_rate": 1.0238425925925928e-05,
      "loss": 0.0191,
      "step": 4217
    },
    {
      "epoch": 0.48819444444444443,
      "grad_norm": 0.10399355739355087,
      "learning_rate": 1.023611111111111e-05,
      "loss": 0.0179,
      "step": 4218
    },
    {
      "epoch": 0.4883101851851852,
      "grad_norm": 0.10881081223487854,
      "learning_rate": 1.0233796296296297e-05,
      "loss": 0.0198,
      "step": 4219
    },
    {
      "epoch": 0.48842592592592593,
      "grad_norm": 0.11969202011823654,
      "learning_rate": 1.0231481481481483e-05,
      "loss": 0.0201,
      "step": 4220
    },
    {
      "epoch": 0.48854166666666665,
      "grad_norm": 0.07911151647567749,
      "learning_rate": 1.0229166666666669e-05,
      "loss": 0.0148,
      "step": 4221
    },
    {
      "epoch": 0.48865740740740743,
      "grad_norm": 0.13969182968139648,
      "learning_rate": 1.0226851851851853e-05,
      "loss": 0.0209,
      "step": 4222
    },
    {
      "epoch": 0.48877314814814815,
      "grad_norm": 0.08098357915878296,
      "learning_rate": 1.0224537037037039e-05,
      "loss": 0.015,
      "step": 4223
    },
    {
      "epoch": 0.4888888888888889,
      "grad_norm": 0.11738064140081406,
      "learning_rate": 1.0222222222222223e-05,
      "loss": 0.0219,
      "step": 4224
    },
    {
      "epoch": 0.48900462962962965,
      "grad_norm": 0.1611153483390808,
      "learning_rate": 1.0219907407407407e-05,
      "loss": 0.022,
      "step": 4225
    },
    {
      "epoch": 0.48912037037037037,
      "grad_norm": 0.11516734212636948,
      "learning_rate": 1.0217592592592593e-05,
      "loss": 0.0212,
      "step": 4226
    },
    {
      "epoch": 0.4892361111111111,
      "grad_norm": 0.12970471382141113,
      "learning_rate": 1.0215277777777779e-05,
      "loss": 0.0239,
      "step": 4227
    },
    {
      "epoch": 0.48935185185185187,
      "grad_norm": 0.07980431616306305,
      "learning_rate": 1.0212962962962965e-05,
      "loss": 0.0147,
      "step": 4228
    },
    {
      "epoch": 0.4894675925925926,
      "grad_norm": 0.09657671302556992,
      "learning_rate": 1.021064814814815e-05,
      "loss": 0.0181,
      "step": 4229
    },
    {
      "epoch": 0.4895833333333333,
      "grad_norm": 0.5506773591041565,
      "learning_rate": 1.0208333333333334e-05,
      "loss": 0.0289,
      "step": 4230
    },
    {
      "epoch": 0.4896990740740741,
      "grad_norm": 0.11324915289878845,
      "learning_rate": 1.020601851851852e-05,
      "loss": 0.0154,
      "step": 4231
    },
    {
      "epoch": 0.4898148148148148,
      "grad_norm": 0.11480492353439331,
      "learning_rate": 1.0203703703703704e-05,
      "loss": 0.021,
      "step": 4232
    },
    {
      "epoch": 0.48993055555555554,
      "grad_norm": 0.1045541763305664,
      "learning_rate": 1.020138888888889e-05,
      "loss": 0.0178,
      "step": 4233
    },
    {
      "epoch": 0.4900462962962963,
      "grad_norm": 0.06815649569034576,
      "learning_rate": 1.0199074074074076e-05,
      "loss": 0.0128,
      "step": 4234
    },
    {
      "epoch": 0.49016203703703703,
      "grad_norm": 0.13633352518081665,
      "learning_rate": 1.0196759259259258e-05,
      "loss": 0.0257,
      "step": 4235
    },
    {
      "epoch": 0.49027777777777776,
      "grad_norm": 0.10376837849617004,
      "learning_rate": 1.0194444444444444e-05,
      "loss": 0.019,
      "step": 4236
    },
    {
      "epoch": 0.49039351851851853,
      "grad_norm": 20.191503524780273,
      "learning_rate": 1.019212962962963e-05,
      "loss": 0.0949,
      "step": 4237
    },
    {
      "epoch": 0.49050925925925926,
      "grad_norm": 34.204551696777344,
      "learning_rate": 1.0189814814814816e-05,
      "loss": 0.15,
      "step": 4238
    },
    {
      "epoch": 0.490625,
      "grad_norm": 0.10863808542490005,
      "learning_rate": 1.0187500000000002e-05,
      "loss": 0.0199,
      "step": 4239
    },
    {
      "epoch": 0.49074074074074076,
      "grad_norm": 0.1022733822464943,
      "learning_rate": 1.0185185185185186e-05,
      "loss": 0.0179,
      "step": 4240
    },
    {
      "epoch": 0.4908564814814815,
      "grad_norm": 0.12126512825489044,
      "learning_rate": 1.018287037037037e-05,
      "loss": 0.023,
      "step": 4241
    },
    {
      "epoch": 0.4909722222222222,
      "grad_norm": 0.08910859376192093,
      "learning_rate": 1.0180555555555556e-05,
      "loss": 0.0165,
      "step": 4242
    },
    {
      "epoch": 0.491087962962963,
      "grad_norm": 0.1291915327310562,
      "learning_rate": 1.017824074074074e-05,
      "loss": 0.0171,
      "step": 4243
    },
    {
      "epoch": 0.4912037037037037,
      "grad_norm": 0.13290241360664368,
      "learning_rate": 1.0175925925925927e-05,
      "loss": 0.0199,
      "step": 4244
    },
    {
      "epoch": 0.4913194444444444,
      "grad_norm": 84.27334594726562,
      "learning_rate": 1.0173611111111113e-05,
      "loss": 2.2302,
      "step": 4245
    },
    {
      "epoch": 0.4914351851851852,
      "grad_norm": 0.0959346741437912,
      "learning_rate": 1.0171296296296299e-05,
      "loss": 0.0178,
      "step": 4246
    },
    {
      "epoch": 0.4915509259259259,
      "grad_norm": 0.11554481089115143,
      "learning_rate": 1.0168981481481481e-05,
      "loss": 0.0215,
      "step": 4247
    },
    {
      "epoch": 0.49166666666666664,
      "grad_norm": 4.612791061401367,
      "learning_rate": 1.0166666666666667e-05,
      "loss": 0.0334,
      "step": 4248
    },
    {
      "epoch": 0.4917824074074074,
      "grad_norm": 0.10673129558563232,
      "learning_rate": 1.0164351851851853e-05,
      "loss": 0.019,
      "step": 4249
    },
    {
      "epoch": 0.49189814814814814,
      "grad_norm": 0.14993560314178467,
      "learning_rate": 1.0162037037037037e-05,
      "loss": 0.0215,
      "step": 4250
    },
    {
      "epoch": 0.49201388888888886,
      "grad_norm": 0.19228145480155945,
      "learning_rate": 1.0159722222222223e-05,
      "loss": 0.0264,
      "step": 4251
    },
    {
      "epoch": 0.49212962962962964,
      "grad_norm": 0.09249481558799744,
      "learning_rate": 1.0157407407407409e-05,
      "loss": 0.0171,
      "step": 4252
    },
    {
      "epoch": 0.49224537037037036,
      "grad_norm": 0.12902966141700745,
      "learning_rate": 1.0155092592592592e-05,
      "loss": 0.0158,
      "step": 4253
    },
    {
      "epoch": 0.49236111111111114,
      "grad_norm": 234.8494873046875,
      "learning_rate": 1.0152777777777778e-05,
      "loss": 0.9489,
      "step": 4254
    },
    {
      "epoch": 0.49247685185185186,
      "grad_norm": 0.13009826838970184,
      "learning_rate": 1.0150462962962964e-05,
      "loss": 0.0218,
      "step": 4255
    },
    {
      "epoch": 0.4925925925925926,
      "grad_norm": 115.31948852539062,
      "learning_rate": 1.014814814814815e-05,
      "loss": 0.8701,
      "step": 4256
    },
    {
      "epoch": 0.49270833333333336,
      "grad_norm": 0.6229619383811951,
      "learning_rate": 1.0145833333333335e-05,
      "loss": 0.0224,
      "step": 4257
    },
    {
      "epoch": 0.4928240740740741,
      "grad_norm": 0.11547622829675674,
      "learning_rate": 1.014351851851852e-05,
      "loss": 0.021,
      "step": 4258
    },
    {
      "epoch": 0.4929398148148148,
      "grad_norm": 0.8652181029319763,
      "learning_rate": 1.0141203703703704e-05,
      "loss": 0.0212,
      "step": 4259
    },
    {
      "epoch": 0.4930555555555556,
      "grad_norm": 0.1431632936000824,
      "learning_rate": 1.013888888888889e-05,
      "loss": 0.021,
      "step": 4260
    },
    {
      "epoch": 0.4931712962962963,
      "grad_norm": 0.13203401863574982,
      "learning_rate": 1.0136574074074074e-05,
      "loss": 0.022,
      "step": 4261
    },
    {
      "epoch": 0.493287037037037,
      "grad_norm": 0.1292906105518341,
      "learning_rate": 1.013425925925926e-05,
      "loss": 0.0162,
      "step": 4262
    },
    {
      "epoch": 0.4934027777777778,
      "grad_norm": 0.11543025821447372,
      "learning_rate": 1.0131944444444446e-05,
      "loss": 0.0209,
      "step": 4263
    },
    {
      "epoch": 0.4935185185185185,
      "grad_norm": 0.08560308814048767,
      "learning_rate": 1.0129629629629632e-05,
      "loss": 0.0161,
      "step": 4264
    },
    {
      "epoch": 0.49363425925925924,
      "grad_norm": 1.582787036895752,
      "learning_rate": 1.0127314814814815e-05,
      "loss": 0.037,
      "step": 4265
    },
    {
      "epoch": 0.49375,
      "grad_norm": 0.11641930788755417,
      "learning_rate": 1.0125e-05,
      "loss": 0.0219,
      "step": 4266
    },
    {
      "epoch": 0.49386574074074074,
      "grad_norm": 0.06815735250711441,
      "learning_rate": 1.0122685185185186e-05,
      "loss": 0.0127,
      "step": 4267
    },
    {
      "epoch": 0.49398148148148147,
      "grad_norm": 273.2347412109375,
      "learning_rate": 1.0120370370370372e-05,
      "loss": 0.6888,
      "step": 4268
    },
    {
      "epoch": 0.49409722222222224,
      "grad_norm": 0.12575475871562958,
      "learning_rate": 1.0118055555555557e-05,
      "loss": 0.0201,
      "step": 4269
    },
    {
      "epoch": 0.49421296296296297,
      "grad_norm": 0.10342057049274445,
      "learning_rate": 1.0115740740740743e-05,
      "loss": 0.0191,
      "step": 4270
    },
    {
      "epoch": 0.4943287037037037,
      "grad_norm": 0.11180628091096878,
      "learning_rate": 1.0113425925925927e-05,
      "loss": 0.0207,
      "step": 4271
    },
    {
      "epoch": 0.49444444444444446,
      "grad_norm": 0.07495008409023285,
      "learning_rate": 1.0111111111111111e-05,
      "loss": 0.014,
      "step": 4272
    },
    {
      "epoch": 0.4945601851851852,
      "grad_norm": 0.1196717694401741,
      "learning_rate": 1.0108796296296297e-05,
      "loss": 0.0204,
      "step": 4273
    },
    {
      "epoch": 0.4946759259259259,
      "grad_norm": 0.11610850691795349,
      "learning_rate": 1.0106481481481483e-05,
      "loss": 0.0192,
      "step": 4274
    },
    {
      "epoch": 0.4947916666666667,
      "grad_norm": 0.918999433517456,
      "learning_rate": 1.0104166666666669e-05,
      "loss": 0.0234,
      "step": 4275
    },
    {
      "epoch": 0.4949074074074074,
      "grad_norm": 0.5291087031364441,
      "learning_rate": 1.0101851851851851e-05,
      "loss": 0.0239,
      "step": 4276
    },
    {
      "epoch": 0.49502314814814813,
      "grad_norm": 0.42020106315612793,
      "learning_rate": 1.0099537037037037e-05,
      "loss": 0.0206,
      "step": 4277
    },
    {
      "epoch": 0.4951388888888889,
      "grad_norm": 0.10659322142601013,
      "learning_rate": 1.0097222222222223e-05,
      "loss": 0.0194,
      "step": 4278
    },
    {
      "epoch": 0.49525462962962963,
      "grad_norm": 0.13583455979824066,
      "learning_rate": 1.0094907407407408e-05,
      "loss": 0.0252,
      "step": 4279
    },
    {
      "epoch": 0.49537037037037035,
      "grad_norm": 0.08949455618858337,
      "learning_rate": 1.0092592592592594e-05,
      "loss": 0.0165,
      "step": 4280
    },
    {
      "epoch": 0.4954861111111111,
      "grad_norm": 0.12008193880319595,
      "learning_rate": 1.009027777777778e-05,
      "loss": 0.021,
      "step": 4281
    },
    {
      "epoch": 0.49560185185185185,
      "grad_norm": 2.9145398139953613,
      "learning_rate": 1.0087962962962962e-05,
      "loss": 0.029,
      "step": 4282
    },
    {
      "epoch": 0.49571759259259257,
      "grad_norm": 0.17222099006175995,
      "learning_rate": 1.0085648148148148e-05,
      "loss": 0.0231,
      "step": 4283
    },
    {
      "epoch": 0.49583333333333335,
      "grad_norm": 0.06643717736005783,
      "learning_rate": 1.0083333333333334e-05,
      "loss": 0.0125,
      "step": 4284
    },
    {
      "epoch": 0.49594907407407407,
      "grad_norm": 288.5949401855469,
      "learning_rate": 1.008101851851852e-05,
      "loss": 1.2735,
      "step": 4285
    },
    {
      "epoch": 0.4960648148148148,
      "grad_norm": 0.14600931107997894,
      "learning_rate": 1.0078703703703706e-05,
      "loss": 0.0199,
      "step": 4286
    },
    {
      "epoch": 0.49618055555555557,
      "grad_norm": 0.0994894877076149,
      "learning_rate": 1.007638888888889e-05,
      "loss": 0.0173,
      "step": 4287
    },
    {
      "epoch": 0.4962962962962963,
      "grad_norm": 0.12210958451032639,
      "learning_rate": 1.0074074074074074e-05,
      "loss": 0.0231,
      "step": 4288
    },
    {
      "epoch": 0.496412037037037,
      "grad_norm": 21.079179763793945,
      "learning_rate": 1.007175925925926e-05,
      "loss": 2.8431,
      "step": 4289
    },
    {
      "epoch": 0.4965277777777778,
      "grad_norm": 0.15259936451911926,
      "learning_rate": 1.0069444444444445e-05,
      "loss": 0.0216,
      "step": 4290
    },
    {
      "epoch": 0.4966435185185185,
      "grad_norm": 0.09005332738161087,
      "learning_rate": 1.006712962962963e-05,
      "loss": 0.0166,
      "step": 4291
    },
    {
      "epoch": 0.49675925925925923,
      "grad_norm": 0.3123338520526886,
      "learning_rate": 1.0064814814814816e-05,
      "loss": 0.0213,
      "step": 4292
    },
    {
      "epoch": 0.496875,
      "grad_norm": 0.07214809954166412,
      "learning_rate": 1.0062500000000002e-05,
      "loss": 0.0135,
      "step": 4293
    },
    {
      "epoch": 0.49699074074074073,
      "grad_norm": 0.07385062426328659,
      "learning_rate": 1.0060185185185185e-05,
      "loss": 0.0137,
      "step": 4294
    },
    {
      "epoch": 0.49710648148148145,
      "grad_norm": 0.21420785784721375,
      "learning_rate": 1.005787037037037e-05,
      "loss": 0.0204,
      "step": 4295
    },
    {
      "epoch": 0.49722222222222223,
      "grad_norm": 38.366695404052734,
      "learning_rate": 1.0055555555555557e-05,
      "loss": 2.2995,
      "step": 4296
    },
    {
      "epoch": 0.49733796296296295,
      "grad_norm": 7.486630916595459,
      "learning_rate": 1.0053240740740741e-05,
      "loss": 2.939,
      "step": 4297
    },
    {
      "epoch": 0.49745370370370373,
      "grad_norm": 7.690020561218262,
      "learning_rate": 1.0050925925925927e-05,
      "loss": 3.1026,
      "step": 4298
    },
    {
      "epoch": 0.49756944444444445,
      "grad_norm": 0.11431007832288742,
      "learning_rate": 1.0048611111111113e-05,
      "loss": 0.0203,
      "step": 4299
    },
    {
      "epoch": 0.4976851851851852,
      "grad_norm": 0.12105585634708405,
      "learning_rate": 1.0046296296296295e-05,
      "loss": 0.0184,
      "step": 4300
    },
    {
      "epoch": 0.49780092592592595,
      "grad_norm": 0.4680062532424927,
      "learning_rate": 1.0043981481481481e-05,
      "loss": 0.0195,
      "step": 4301
    },
    {
      "epoch": 0.4979166666666667,
      "grad_norm": 0.06905826181173325,
      "learning_rate": 1.0041666666666667e-05,
      "loss": 0.0129,
      "step": 4302
    },
    {
      "epoch": 0.4980324074074074,
      "grad_norm": 0.07587360590696335,
      "learning_rate": 1.0039351851851853e-05,
      "loss": 0.014,
      "step": 4303
    },
    {
      "epoch": 0.4981481481481482,
      "grad_norm": 13.631193161010742,
      "learning_rate": 1.003703703703704e-05,
      "loss": 0.077,
      "step": 4304
    },
    {
      "epoch": 0.4982638888888889,
      "grad_norm": 0.14451900124549866,
      "learning_rate": 1.0034722222222224e-05,
      "loss": 0.0201,
      "step": 4305
    },
    {
      "epoch": 0.4983796296296296,
      "grad_norm": 0.09945820271968842,
      "learning_rate": 1.0032407407407408e-05,
      "loss": 0.0174,
      "step": 4306
    },
    {
      "epoch": 0.4984953703703704,
      "grad_norm": 2.116946220397949,
      "learning_rate": 1.0030092592592594e-05,
      "loss": 0.0323,
      "step": 4307
    },
    {
      "epoch": 0.4986111111111111,
      "grad_norm": 0.10647420585155487,
      "learning_rate": 1.0027777777777778e-05,
      "loss": 0.0189,
      "step": 4308
    },
    {
      "epoch": 0.49872685185185184,
      "grad_norm": 0.10670952498912811,
      "learning_rate": 1.0025462962962964e-05,
      "loss": 0.0195,
      "step": 4309
    },
    {
      "epoch": 0.4988425925925926,
      "grad_norm": 0.09648963063955307,
      "learning_rate": 1.002314814814815e-05,
      "loss": 0.0174,
      "step": 4310
    },
    {
      "epoch": 0.49895833333333334,
      "grad_norm": 0.11656861007213593,
      "learning_rate": 1.0020833333333336e-05,
      "loss": 0.0155,
      "step": 4311
    },
    {
      "epoch": 0.49907407407407406,
      "grad_norm": 0.07713452726602554,
      "learning_rate": 1.0018518518518518e-05,
      "loss": 0.0144,
      "step": 4312
    },
    {
      "epoch": 0.49918981481481484,
      "grad_norm": 0.12912508845329285,
      "learning_rate": 1.0016203703703704e-05,
      "loss": 0.0206,
      "step": 4313
    },
    {
      "epoch": 0.49930555555555556,
      "grad_norm": 0.09979470819234848,
      "learning_rate": 1.001388888888889e-05,
      "loss": 0.0168,
      "step": 4314
    },
    {
      "epoch": 0.4994212962962963,
      "grad_norm": 0.09881029278039932,
      "learning_rate": 1.0011574074074074e-05,
      "loss": 0.018,
      "step": 4315
    },
    {
      "epoch": 0.49953703703703706,
      "grad_norm": 0.2144947052001953,
      "learning_rate": 1.000925925925926e-05,
      "loss": 0.022,
      "step": 4316
    },
    {
      "epoch": 0.4996527777777778,
      "grad_norm": 0.09336841851472855,
      "learning_rate": 1.0006944444444446e-05,
      "loss": 0.0172,
      "step": 4317
    },
    {
      "epoch": 0.4997685185185185,
      "grad_norm": 0.10973543673753738,
      "learning_rate": 1.0004629629629629e-05,
      "loss": 0.0206,
      "step": 4318
    },
    {
      "epoch": 0.4998842592592593,
      "grad_norm": 0.1116490438580513,
      "learning_rate": 1.0002314814814815e-05,
      "loss": 0.0208,
      "step": 4319
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5223750472068787,
      "learning_rate": 1e-05,
      "loss": 0.0218,
      "step": 4320
    },
    {
      "epoch": 0.5001157407407407,
      "grad_norm": 0.09963005781173706,
      "learning_rate": 9.997685185185187e-06,
      "loss": 0.018,
      "step": 4321
    },
    {
      "epoch": 0.5002314814814814,
      "grad_norm": 0.14761993288993835,
      "learning_rate": 9.995370370370371e-06,
      "loss": 0.0243,
      "step": 4322
    },
    {
      "epoch": 0.5003472222222223,
      "grad_norm": 0.10940353572368622,
      "learning_rate": 9.993055555555557e-06,
      "loss": 0.0182,
      "step": 4323
    },
    {
      "epoch": 0.500462962962963,
      "grad_norm": 0.2798956632614136,
      "learning_rate": 9.990740740740741e-06,
      "loss": 0.0144,
      "step": 4324
    },
    {
      "epoch": 0.5005787037037037,
      "grad_norm": 52.425960540771484,
      "learning_rate": 9.988425925925927e-06,
      "loss": 0.6386,
      "step": 4325
    },
    {
      "epoch": 0.5006944444444444,
      "grad_norm": 0.16053138673305511,
      "learning_rate": 9.986111111111111e-06,
      "loss": 0.0214,
      "step": 4326
    },
    {
      "epoch": 0.5008101851851852,
      "grad_norm": 0.14421069622039795,
      "learning_rate": 9.983796296296297e-06,
      "loss": 0.0181,
      "step": 4327
    },
    {
      "epoch": 0.5009259259259259,
      "grad_norm": 0.10294872522354126,
      "learning_rate": 9.981481481481482e-06,
      "loss": 0.0176,
      "step": 4328
    },
    {
      "epoch": 0.5010416666666667,
      "grad_norm": 0.09084917604923248,
      "learning_rate": 9.979166666666668e-06,
      "loss": 0.0122,
      "step": 4329
    },
    {
      "epoch": 0.5011574074074074,
      "grad_norm": 2.6215829849243164,
      "learning_rate": 9.976851851851853e-06,
      "loss": 0.0334,
      "step": 4330
    },
    {
      "epoch": 0.5012731481481482,
      "grad_norm": 0.13218402862548828,
      "learning_rate": 9.974537037037038e-06,
      "loss": 0.0245,
      "step": 4331
    },
    {
      "epoch": 0.5013888888888889,
      "grad_norm": 0.09820962697267532,
      "learning_rate": 9.972222222222224e-06,
      "loss": 0.0177,
      "step": 4332
    },
    {
      "epoch": 0.5015046296296296,
      "grad_norm": 0.09362237900495529,
      "learning_rate": 9.969907407407408e-06,
      "loss": 0.017,
      "step": 4333
    },
    {
      "epoch": 0.5016203703703703,
      "grad_norm": 23.825817108154297,
      "learning_rate": 9.967592592592594e-06,
      "loss": 0.1124,
      "step": 4334
    },
    {
      "epoch": 0.5017361111111112,
      "grad_norm": 0.10635911673307419,
      "learning_rate": 9.965277777777778e-06,
      "loss": 0.0183,
      "step": 4335
    },
    {
      "epoch": 0.5018518518518519,
      "grad_norm": 0.10268048197031021,
      "learning_rate": 9.962962962962964e-06,
      "loss": 0.0185,
      "step": 4336
    },
    {
      "epoch": 0.5019675925925926,
      "grad_norm": 0.07770467549562454,
      "learning_rate": 9.960648148148148e-06,
      "loss": 0.0145,
      "step": 4337
    },
    {
      "epoch": 0.5020833333333333,
      "grad_norm": 0.10286654531955719,
      "learning_rate": 9.958333333333334e-06,
      "loss": 0.0189,
      "step": 4338
    },
    {
      "epoch": 0.502199074074074,
      "grad_norm": 0.0831741988658905,
      "learning_rate": 9.95601851851852e-06,
      "loss": 0.0145,
      "step": 4339
    },
    {
      "epoch": 0.5023148148148148,
      "grad_norm": 0.09492020308971405,
      "learning_rate": 9.953703703703704e-06,
      "loss": 0.0176,
      "step": 4340
    },
    {
      "epoch": 0.5024305555555556,
      "grad_norm": 0.08694713562726974,
      "learning_rate": 9.95138888888889e-06,
      "loss": 0.0158,
      "step": 4341
    },
    {
      "epoch": 0.5025462962962963,
      "grad_norm": 0.10533801466226578,
      "learning_rate": 9.949074074074075e-06,
      "loss": 0.0184,
      "step": 4342
    },
    {
      "epoch": 0.502662037037037,
      "grad_norm": 0.1550842523574829,
      "learning_rate": 9.94675925925926e-06,
      "loss": 0.0211,
      "step": 4343
    },
    {
      "epoch": 0.5027777777777778,
      "grad_norm": 0.06530894339084625,
      "learning_rate": 9.944444444444445e-06,
      "loss": 0.0123,
      "step": 4344
    },
    {
      "epoch": 0.5028935185185185,
      "grad_norm": 0.15966063737869263,
      "learning_rate": 9.942129629629629e-06,
      "loss": 0.0241,
      "step": 4345
    },
    {
      "epoch": 0.5030092592592592,
      "grad_norm": 0.08222011476755142,
      "learning_rate": 9.939814814814815e-06,
      "loss": 0.0154,
      "step": 4346
    },
    {
      "epoch": 0.503125,
      "grad_norm": 0.14843890070915222,
      "learning_rate": 9.937500000000001e-06,
      "loss": 0.022,
      "step": 4347
    },
    {
      "epoch": 0.5032407407407408,
      "grad_norm": 0.11573383212089539,
      "learning_rate": 9.935185185185185e-06,
      "loss": 0.0218,
      "step": 4348
    },
    {
      "epoch": 0.5033564814814815,
      "grad_norm": 0.12020166218280792,
      "learning_rate": 9.932870370370371e-06,
      "loss": 0.0227,
      "step": 4349
    },
    {
      "epoch": 0.5034722222222222,
      "grad_norm": 43.503116607666016,
      "learning_rate": 9.930555555555557e-06,
      "loss": 0.1345,
      "step": 4350
    },
    {
      "epoch": 0.5035879629629629,
      "grad_norm": 0.1816728413105011,
      "learning_rate": 9.928240740740741e-06,
      "loss": 0.0229,
      "step": 4351
    },
    {
      "epoch": 0.5037037037037037,
      "grad_norm": 0.09780776500701904,
      "learning_rate": 9.925925925925927e-06,
      "loss": 0.017,
      "step": 4352
    },
    {
      "epoch": 0.5038194444444445,
      "grad_norm": 0.301448792219162,
      "learning_rate": 9.923611111111112e-06,
      "loss": 0.0266,
      "step": 4353
    },
    {
      "epoch": 0.5039351851851852,
      "grad_norm": 0.0923660546541214,
      "learning_rate": 9.921296296296296e-06,
      "loss": 0.0168,
      "step": 4354
    },
    {
      "epoch": 0.5040509259259259,
      "grad_norm": 0.06671196967363358,
      "learning_rate": 9.918981481481482e-06,
      "loss": 0.0125,
      "step": 4355
    },
    {
      "epoch": 0.5041666666666667,
      "grad_norm": 0.07220214605331421,
      "learning_rate": 9.916666666666668e-06,
      "loss": 0.0135,
      "step": 4356
    },
    {
      "epoch": 0.5042824074074074,
      "grad_norm": 0.06633464992046356,
      "learning_rate": 9.914351851851852e-06,
      "loss": 0.0124,
      "step": 4357
    },
    {
      "epoch": 0.5043981481481481,
      "grad_norm": 0.09623739868402481,
      "learning_rate": 9.912037037037038e-06,
      "loss": 0.0175,
      "step": 4358
    },
    {
      "epoch": 0.5045138888888889,
      "grad_norm": 0.12370017170906067,
      "learning_rate": 9.909722222222224e-06,
      "loss": 0.0236,
      "step": 4359
    },
    {
      "epoch": 0.5046296296296297,
      "grad_norm": 0.07101407647132874,
      "learning_rate": 9.907407407407408e-06,
      "loss": 0.0133,
      "step": 4360
    },
    {
      "epoch": 0.5047453703703704,
      "grad_norm": 0.10415062308311462,
      "learning_rate": 9.905092592592594e-06,
      "loss": 0.0179,
      "step": 4361
    },
    {
      "epoch": 0.5048611111111111,
      "grad_norm": 0.07384742051362991,
      "learning_rate": 9.902777777777778e-06,
      "loss": 0.0131,
      "step": 4362
    },
    {
      "epoch": 0.5049768518518518,
      "grad_norm": 0.09433968365192413,
      "learning_rate": 9.900462962962963e-06,
      "loss": 0.017,
      "step": 4363
    },
    {
      "epoch": 0.5050925925925925,
      "grad_norm": 0.2537499666213989,
      "learning_rate": 9.898148148148148e-06,
      "loss": 0.0273,
      "step": 4364
    },
    {
      "epoch": 0.5052083333333334,
      "grad_norm": 0.10616814345121384,
      "learning_rate": 9.895833333333334e-06,
      "loss": 0.0196,
      "step": 4365
    },
    {
      "epoch": 0.5053240740740741,
      "grad_norm": 0.15818578004837036,
      "learning_rate": 9.893518518518519e-06,
      "loss": 0.0212,
      "step": 4366
    },
    {
      "epoch": 0.5054398148148148,
      "grad_norm": 0.09809623658657074,
      "learning_rate": 9.891203703703705e-06,
      "loss": 0.0178,
      "step": 4367
    },
    {
      "epoch": 0.5055555555555555,
      "grad_norm": 0.9455673098564148,
      "learning_rate": 9.88888888888889e-06,
      "loss": 0.023,
      "step": 4368
    },
    {
      "epoch": 0.5056712962962963,
      "grad_norm": 0.10429325699806213,
      "learning_rate": 9.886574074074075e-06,
      "loss": 0.0196,
      "step": 4369
    },
    {
      "epoch": 0.5057870370370371,
      "grad_norm": 0.07010507583618164,
      "learning_rate": 9.88425925925926e-06,
      "loss": 0.0132,
      "step": 4370
    },
    {
      "epoch": 0.5059027777777778,
      "grad_norm": 0.6285749673843384,
      "learning_rate": 9.881944444444445e-06,
      "loss": 0.0165,
      "step": 4371
    },
    {
      "epoch": 0.5060185185185185,
      "grad_norm": 0.07566045224666595,
      "learning_rate": 9.87962962962963e-06,
      "loss": 0.0138,
      "step": 4372
    },
    {
      "epoch": 0.5061342592592593,
      "grad_norm": 166.30105590820312,
      "learning_rate": 9.877314814814815e-06,
      "loss": 1.4671,
      "step": 4373
    },
    {
      "epoch": 0.50625,
      "grad_norm": 21.784435272216797,
      "learning_rate": 9.875000000000001e-06,
      "loss": 0.0916,
      "step": 4374
    },
    {
      "epoch": 0.5063657407407407,
      "grad_norm": 0.06593046337366104,
      "learning_rate": 9.872685185185185e-06,
      "loss": 0.0123,
      "step": 4375
    },
    {
      "epoch": 0.5064814814814815,
      "grad_norm": 0.10762333124876022,
      "learning_rate": 9.870370370370371e-06,
      "loss": 0.0201,
      "step": 4376
    },
    {
      "epoch": 0.5065972222222223,
      "grad_norm": 2.8448550701141357,
      "learning_rate": 9.868055555555557e-06,
      "loss": 0.0274,
      "step": 4377
    },
    {
      "epoch": 0.506712962962963,
      "grad_norm": 0.06651681661605835,
      "learning_rate": 9.865740740740742e-06,
      "loss": 0.0123,
      "step": 4378
    },
    {
      "epoch": 0.5068287037037037,
      "grad_norm": 0.09668011218309402,
      "learning_rate": 9.863425925925928e-06,
      "loss": 0.0179,
      "step": 4379
    },
    {
      "epoch": 0.5069444444444444,
      "grad_norm": 0.09977137297391891,
      "learning_rate": 9.861111111111112e-06,
      "loss": 0.0179,
      "step": 4380
    },
    {
      "epoch": 0.5070601851851851,
      "grad_norm": 0.08919062465429306,
      "learning_rate": 9.858796296296298e-06,
      "loss": 0.0164,
      "step": 4381
    },
    {
      "epoch": 0.507175925925926,
      "grad_norm": 0.06445953994989395,
      "learning_rate": 9.856481481481482e-06,
      "loss": 0.0121,
      "step": 4382
    },
    {
      "epoch": 0.5072916666666667,
      "grad_norm": 0.09434741735458374,
      "learning_rate": 9.854166666666668e-06,
      "loss": 0.017,
      "step": 4383
    },
    {
      "epoch": 0.5074074074074074,
      "grad_norm": 0.09371201694011688,
      "learning_rate": 9.851851851851852e-06,
      "loss": 0.0173,
      "step": 4384
    },
    {
      "epoch": 0.5075231481481481,
      "grad_norm": 0.09481051564216614,
      "learning_rate": 9.849537037037038e-06,
      "loss": 0.0171,
      "step": 4385
    },
    {
      "epoch": 0.5076388888888889,
      "grad_norm": 1.0984458923339844,
      "learning_rate": 9.847222222222224e-06,
      "loss": 0.0281,
      "step": 4386
    },
    {
      "epoch": 0.5077546296296296,
      "grad_norm": 4.540024280548096,
      "learning_rate": 9.844907407407408e-06,
      "loss": 0.0352,
      "step": 4387
    },
    {
      "epoch": 0.5078703703703704,
      "grad_norm": 0.09025250375270844,
      "learning_rate": 9.842592592592594e-06,
      "loss": 0.0135,
      "step": 4388
    },
    {
      "epoch": 0.5079861111111111,
      "grad_norm": 0.0956776887178421,
      "learning_rate": 9.840277777777778e-06,
      "loss": 0.0176,
      "step": 4389
    },
    {
      "epoch": 0.5081018518518519,
      "grad_norm": 0.10429359972476959,
      "learning_rate": 9.837962962962964e-06,
      "loss": 0.0171,
      "step": 4390
    },
    {
      "epoch": 0.5082175925925926,
      "grad_norm": 0.09225304424762726,
      "learning_rate": 9.835648148148149e-06,
      "loss": 0.015,
      "step": 4391
    },
    {
      "epoch": 0.5083333333333333,
      "grad_norm": 0.1196729987859726,
      "learning_rate": 9.833333333333333e-06,
      "loss": 0.0226,
      "step": 4392
    },
    {
      "epoch": 0.508449074074074,
      "grad_norm": 0.23310984671115875,
      "learning_rate": 9.831018518518519e-06,
      "loss": 0.0183,
      "step": 4393
    },
    {
      "epoch": 0.5085648148148149,
      "grad_norm": 0.12135165929794312,
      "learning_rate": 9.828703703703705e-06,
      "loss": 0.0195,
      "step": 4394
    },
    {
      "epoch": 0.5086805555555556,
      "grad_norm": 1.370367169380188,
      "learning_rate": 9.826388888888889e-06,
      "loss": 0.0264,
      "step": 4395
    },
    {
      "epoch": 0.5087962962962963,
      "grad_norm": 0.18780367076396942,
      "learning_rate": 9.824074074074075e-06,
      "loss": 0.0242,
      "step": 4396
    },
    {
      "epoch": 0.508912037037037,
      "grad_norm": 0.08916567265987396,
      "learning_rate": 9.821759259259261e-06,
      "loss": 0.0161,
      "step": 4397
    },
    {
      "epoch": 0.5090277777777777,
      "grad_norm": 0.15137062966823578,
      "learning_rate": 9.819444444444445e-06,
      "loss": 0.0137,
      "step": 4398
    },
    {
      "epoch": 0.5091435185185185,
      "grad_norm": 0.08410423249006271,
      "learning_rate": 9.817129629629631e-06,
      "loss": 0.0155,
      "step": 4399
    },
    {
      "epoch": 0.5092592592592593,
      "grad_norm": 0.12222544848918915,
      "learning_rate": 9.814814814814815e-06,
      "loss": 0.0233,
      "step": 4400
    },
    {
      "epoch": 0.509375,
      "grad_norm": 0.11846134066581726,
      "learning_rate": 9.8125e-06,
      "loss": 0.0223,
      "step": 4401
    },
    {
      "epoch": 0.5094907407407407,
      "grad_norm": 15.687969207763672,
      "learning_rate": 9.810185185185186e-06,
      "loss": 2.7052,
      "step": 4402
    },
    {
      "epoch": 0.5096064814814815,
      "grad_norm": 0.11475605517625809,
      "learning_rate": 9.807870370370372e-06,
      "loss": 0.0186,
      "step": 4403
    },
    {
      "epoch": 0.5097222222222222,
      "grad_norm": 0.0914638489484787,
      "learning_rate": 9.805555555555556e-06,
      "loss": 0.0123,
      "step": 4404
    },
    {
      "epoch": 0.5098379629629629,
      "grad_norm": 0.07967036962509155,
      "learning_rate": 9.803240740740742e-06,
      "loss": 0.0138,
      "step": 4405
    },
    {
      "epoch": 0.5099537037037037,
      "grad_norm": 0.07263306528329849,
      "learning_rate": 9.800925925925928e-06,
      "loss": 0.0133,
      "step": 4406
    },
    {
      "epoch": 0.5100694444444445,
      "grad_norm": 0.06876514106988907,
      "learning_rate": 9.798611111111112e-06,
      "loss": 0.0128,
      "step": 4407
    },
    {
      "epoch": 0.5101851851851852,
      "grad_norm": 0.11990133672952652,
      "learning_rate": 9.796296296296298e-06,
      "loss": 0.0225,
      "step": 4408
    },
    {
      "epoch": 0.5103009259259259,
      "grad_norm": 0.14604903757572174,
      "learning_rate": 9.793981481481482e-06,
      "loss": 0.0198,
      "step": 4409
    },
    {
      "epoch": 0.5104166666666666,
      "grad_norm": 0.06400252133607864,
      "learning_rate": 9.791666666666666e-06,
      "loss": 0.012,
      "step": 4410
    },
    {
      "epoch": 0.5105324074074075,
      "grad_norm": 0.07545527815818787,
      "learning_rate": 9.789351851851852e-06,
      "loss": 0.0141,
      "step": 4411
    },
    {
      "epoch": 0.5106481481481482,
      "grad_norm": 0.11954018473625183,
      "learning_rate": 9.787037037037038e-06,
      "loss": 0.0198,
      "step": 4412
    },
    {
      "epoch": 0.5107638888888889,
      "grad_norm": 0.06384079903364182,
      "learning_rate": 9.784722222222223e-06,
      "loss": 0.012,
      "step": 4413
    },
    {
      "epoch": 0.5108796296296296,
      "grad_norm": 0.07129539549350739,
      "learning_rate": 9.782407407407408e-06,
      "loss": 0.0121,
      "step": 4414
    },
    {
      "epoch": 0.5109953703703703,
      "grad_norm": 0.09271176904439926,
      "learning_rate": 9.780092592592594e-06,
      "loss": 0.0163,
      "step": 4415
    },
    {
      "epoch": 0.5111111111111111,
      "grad_norm": 0.10455267131328583,
      "learning_rate": 9.777777777777779e-06,
      "loss": 0.0187,
      "step": 4416
    },
    {
      "epoch": 0.5112268518518519,
      "grad_norm": 0.16190700232982635,
      "learning_rate": 9.775462962962965e-06,
      "loss": 0.0221,
      "step": 4417
    },
    {
      "epoch": 0.5113425925925926,
      "grad_norm": 0.2920169532299042,
      "learning_rate": 9.773148148148149e-06,
      "loss": 0.0211,
      "step": 4418
    },
    {
      "epoch": 0.5114583333333333,
      "grad_norm": 0.08603019267320633,
      "learning_rate": 9.770833333333333e-06,
      "loss": 0.0157,
      "step": 4419
    },
    {
      "epoch": 0.5115740740740741,
      "grad_norm": 0.08002546429634094,
      "learning_rate": 9.768518518518519e-06,
      "loss": 0.0147,
      "step": 4420
    },
    {
      "epoch": 0.5116898148148148,
      "grad_norm": 0.13647441565990448,
      "learning_rate": 9.766203703703705e-06,
      "loss": 0.0147,
      "step": 4421
    },
    {
      "epoch": 0.5118055555555555,
      "grad_norm": 0.07694946229457855,
      "learning_rate": 9.76388888888889e-06,
      "loss": 0.0136,
      "step": 4422
    },
    {
      "epoch": 0.5119212962962963,
      "grad_norm": 1.3614610433578491,
      "learning_rate": 9.761574074074075e-06,
      "loss": 0.0285,
      "step": 4423
    },
    {
      "epoch": 0.5120370370370371,
      "grad_norm": 0.10215996950864792,
      "learning_rate": 9.759259259259261e-06,
      "loss": 0.0181,
      "step": 4424
    },
    {
      "epoch": 0.5121527777777778,
      "grad_norm": 0.06549495458602905,
      "learning_rate": 9.756944444444445e-06,
      "loss": 0.0122,
      "step": 4425
    },
    {
      "epoch": 0.5122685185185185,
      "grad_norm": 0.06867540627717972,
      "learning_rate": 9.754629629629631e-06,
      "loss": 0.0128,
      "step": 4426
    },
    {
      "epoch": 0.5123842592592592,
      "grad_norm": 0.08889655023813248,
      "learning_rate": 9.752314814814816e-06,
      "loss": 0.0166,
      "step": 4427
    },
    {
      "epoch": 0.5125,
      "grad_norm": 0.10164599120616913,
      "learning_rate": 9.75e-06,
      "loss": 0.018,
      "step": 4428
    },
    {
      "epoch": 0.5126157407407408,
      "grad_norm": 0.09352459013462067,
      "learning_rate": 9.747685185185186e-06,
      "loss": 0.017,
      "step": 4429
    },
    {
      "epoch": 0.5127314814814815,
      "grad_norm": 0.06927114725112915,
      "learning_rate": 9.745370370370372e-06,
      "loss": 0.0129,
      "step": 4430
    },
    {
      "epoch": 0.5128472222222222,
      "grad_norm": 0.10858044773340225,
      "learning_rate": 9.743055555555556e-06,
      "loss": 0.0195,
      "step": 4431
    },
    {
      "epoch": 0.512962962962963,
      "grad_norm": 0.11380264908075333,
      "learning_rate": 9.740740740740742e-06,
      "loss": 0.0214,
      "step": 4432
    },
    {
      "epoch": 0.5130787037037037,
      "grad_norm": 0.12482629716396332,
      "learning_rate": 9.738425925925926e-06,
      "loss": 0.0188,
      "step": 4433
    },
    {
      "epoch": 0.5131944444444444,
      "grad_norm": 0.10165871679782867,
      "learning_rate": 9.736111111111112e-06,
      "loss": 0.0179,
      "step": 4434
    },
    {
      "epoch": 0.5133101851851852,
      "grad_norm": 0.1195397898554802,
      "learning_rate": 9.733796296296298e-06,
      "loss": 0.0224,
      "step": 4435
    },
    {
      "epoch": 0.513425925925926,
      "grad_norm": 101.93358612060547,
      "learning_rate": 9.731481481481482e-06,
      "loss": 0.1546,
      "step": 4436
    },
    {
      "epoch": 0.5135416666666667,
      "grad_norm": 0.3583647608757019,
      "learning_rate": 9.729166666666667e-06,
      "loss": 0.0208,
      "step": 4437
    },
    {
      "epoch": 0.5136574074074074,
      "grad_norm": 0.11499147117137909,
      "learning_rate": 9.726851851851852e-06,
      "loss": 0.0216,
      "step": 4438
    },
    {
      "epoch": 0.5137731481481481,
      "grad_norm": 0.1537133753299713,
      "learning_rate": 9.724537037037037e-06,
      "loss": 0.0207,
      "step": 4439
    },
    {
      "epoch": 0.5138888888888888,
      "grad_norm": 0.17819343507289886,
      "learning_rate": 9.722222222222223e-06,
      "loss": 0.0176,
      "step": 4440
    },
    {
      "epoch": 0.5140046296296297,
      "grad_norm": 0.08951257914304733,
      "learning_rate": 9.719907407407409e-06,
      "loss": 0.0166,
      "step": 4441
    },
    {
      "epoch": 0.5141203703703704,
      "grad_norm": 26.079761505126953,
      "learning_rate": 9.717592592592593e-06,
      "loss": 2.6025,
      "step": 4442
    },
    {
      "epoch": 0.5142361111111111,
      "grad_norm": 0.06986366957426071,
      "learning_rate": 9.715277777777779e-06,
      "loss": 0.013,
      "step": 4443
    },
    {
      "epoch": 0.5143518518518518,
      "grad_norm": 0.09611429274082184,
      "learning_rate": 9.712962962962965e-06,
      "loss": 0.0171,
      "step": 4444
    },
    {
      "epoch": 0.5144675925925926,
      "grad_norm": 0.07530535757541656,
      "learning_rate": 9.710648148148149e-06,
      "loss": 0.0138,
      "step": 4445
    },
    {
      "epoch": 0.5145833333333333,
      "grad_norm": 12.311969757080078,
      "learning_rate": 9.708333333333333e-06,
      "loss": 3.2006,
      "step": 4446
    },
    {
      "epoch": 0.5146990740740741,
      "grad_norm": 0.07404755800962448,
      "learning_rate": 9.70601851851852e-06,
      "loss": 0.0138,
      "step": 4447
    },
    {
      "epoch": 0.5148148148148148,
      "grad_norm": 0.09355016052722931,
      "learning_rate": 9.703703703703703e-06,
      "loss": 0.0125,
      "step": 4448
    },
    {
      "epoch": 0.5149305555555556,
      "grad_norm": 42.578060150146484,
      "learning_rate": 9.70138888888889e-06,
      "loss": 2.1959,
      "step": 4449
    },
    {
      "epoch": 0.5150462962962963,
      "grad_norm": 16.239192962646484,
      "learning_rate": 9.699074074074075e-06,
      "loss": 2.977,
      "step": 4450
    },
    {
      "epoch": 0.515162037037037,
      "grad_norm": 0.06602300703525543,
      "learning_rate": 9.69675925925926e-06,
      "loss": 0.0122,
      "step": 4451
    },
    {
      "epoch": 0.5152777777777777,
      "grad_norm": 0.08754409849643707,
      "learning_rate": 9.694444444444446e-06,
      "loss": 0.0163,
      "step": 4452
    },
    {
      "epoch": 0.5153935185185186,
      "grad_norm": 0.09534374624490738,
      "learning_rate": 9.692129629629631e-06,
      "loss": 0.0173,
      "step": 4453
    },
    {
      "epoch": 0.5155092592592593,
      "grad_norm": 0.08575144410133362,
      "learning_rate": 9.689814814814816e-06,
      "loss": 0.0156,
      "step": 4454
    },
    {
      "epoch": 0.515625,
      "grad_norm": 1.0007150173187256,
      "learning_rate": 9.6875e-06,
      "loss": 0.0222,
      "step": 4455
    },
    {
      "epoch": 0.5157407407407407,
      "grad_norm": 0.10177857428789139,
      "learning_rate": 9.685185185185186e-06,
      "loss": 0.0181,
      "step": 4456
    },
    {
      "epoch": 0.5158564814814814,
      "grad_norm": 0.07935849577188492,
      "learning_rate": 9.68287037037037e-06,
      "loss": 0.0124,
      "step": 4457
    },
    {
      "epoch": 0.5159722222222223,
      "grad_norm": 0.11183992773294449,
      "learning_rate": 9.680555555555556e-06,
      "loss": 0.0191,
      "step": 4458
    },
    {
      "epoch": 0.516087962962963,
      "grad_norm": 0.09868723154067993,
      "learning_rate": 9.678240740740742e-06,
      "loss": 0.0162,
      "step": 4459
    },
    {
      "epoch": 0.5162037037037037,
      "grad_norm": 0.4063015580177307,
      "learning_rate": 9.675925925925926e-06,
      "loss": 0.0178,
      "step": 4460
    },
    {
      "epoch": 0.5163194444444444,
      "grad_norm": 0.11393914371728897,
      "learning_rate": 9.673611111111112e-06,
      "loss": 0.0214,
      "step": 4461
    },
    {
      "epoch": 0.5164351851851852,
      "grad_norm": 0.10773254185914993,
      "learning_rate": 9.671296296296298e-06,
      "loss": 0.0203,
      "step": 4462
    },
    {
      "epoch": 0.5165509259259259,
      "grad_norm": 0.0891144722700119,
      "learning_rate": 9.668981481481482e-06,
      "loss": 0.0161,
      "step": 4463
    },
    {
      "epoch": 0.5166666666666667,
      "grad_norm": 0.09504406154155731,
      "learning_rate": 9.666666666666667e-06,
      "loss": 0.0172,
      "step": 4464
    },
    {
      "epoch": 0.5167824074074074,
      "grad_norm": 0.09040830284357071,
      "learning_rate": 9.664351851851853e-06,
      "loss": 0.0127,
      "step": 4465
    },
    {
      "epoch": 0.5168981481481482,
      "grad_norm": 0.08687467873096466,
      "learning_rate": 9.662037037037037e-06,
      "loss": 0.0158,
      "step": 4466
    },
    {
      "epoch": 0.5170138888888889,
      "grad_norm": 0.11646826565265656,
      "learning_rate": 9.659722222222223e-06,
      "loss": 0.0158,
      "step": 4467
    },
    {
      "epoch": 0.5171296296296296,
      "grad_norm": 0.10761122405529022,
      "learning_rate": 9.657407407407409e-06,
      "loss": 0.0194,
      "step": 4468
    },
    {
      "epoch": 0.5172453703703703,
      "grad_norm": 0.09150049090385437,
      "learning_rate": 9.655092592592593e-06,
      "loss": 0.0171,
      "step": 4469
    },
    {
      "epoch": 0.5173611111111112,
      "grad_norm": 0.10846849530935287,
      "learning_rate": 9.652777777777779e-06,
      "loss": 0.016,
      "step": 4470
    },
    {
      "epoch": 0.5174768518518519,
      "grad_norm": 0.1492861956357956,
      "learning_rate": 9.650462962962965e-06,
      "loss": 0.023,
      "step": 4471
    },
    {
      "epoch": 0.5175925925925926,
      "grad_norm": 0.1472126543521881,
      "learning_rate": 9.64814814814815e-06,
      "loss": 0.0199,
      "step": 4472
    },
    {
      "epoch": 0.5177083333333333,
      "grad_norm": 0.10755741596221924,
      "learning_rate": 9.645833333333333e-06,
      "loss": 0.0198,
      "step": 4473
    },
    {
      "epoch": 0.517824074074074,
      "grad_norm": 0.13451501727104187,
      "learning_rate": 9.64351851851852e-06,
      "loss": 0.0179,
      "step": 4474
    },
    {
      "epoch": 0.5179398148148148,
      "grad_norm": 67.20111846923828,
      "learning_rate": 9.641203703703704e-06,
      "loss": 1.6245,
      "step": 4475
    },
    {
      "epoch": 0.5180555555555556,
      "grad_norm": 0.16248846054077148,
      "learning_rate": 9.63888888888889e-06,
      "loss": 0.0221,
      "step": 4476
    },
    {
      "epoch": 0.5181712962962963,
      "grad_norm": 0.1577608585357666,
      "learning_rate": 9.636574074074076e-06,
      "loss": 0.0179,
      "step": 4477
    },
    {
      "epoch": 0.518287037037037,
      "grad_norm": 0.08397889882326126,
      "learning_rate": 9.63425925925926e-06,
      "loss": 0.0135,
      "step": 4478
    },
    {
      "epoch": 0.5184027777777778,
      "grad_norm": 0.06761721521615982,
      "learning_rate": 9.631944444444446e-06,
      "loss": 0.0123,
      "step": 4479
    },
    {
      "epoch": 0.5185185185185185,
      "grad_norm": 0.10729465633630753,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.0189,
      "step": 4480
    },
    {
      "epoch": 0.5186342592592592,
      "grad_norm": 0.0734136700630188,
      "learning_rate": 9.627314814814816e-06,
      "loss": 0.0137,
      "step": 4481
    },
    {
      "epoch": 0.51875,
      "grad_norm": 0.09745216369628906,
      "learning_rate": 9.625e-06,
      "loss": 0.0176,
      "step": 4482
    },
    {
      "epoch": 0.5188657407407408,
      "grad_norm": 0.07465845346450806,
      "learning_rate": 9.622685185185186e-06,
      "loss": 0.0122,
      "step": 4483
    },
    {
      "epoch": 0.5189814814814815,
      "grad_norm": 0.09102078527212143,
      "learning_rate": 9.62037037037037e-06,
      "loss": 0.0121,
      "step": 4484
    },
    {
      "epoch": 0.5190972222222222,
      "grad_norm": 0.11284103244543076,
      "learning_rate": 9.618055555555556e-06,
      "loss": 0.0153,
      "step": 4485
    },
    {
      "epoch": 0.5192129629629629,
      "grad_norm": 8.400481224060059,
      "learning_rate": 9.61574074074074e-06,
      "loss": 3.2341,
      "step": 4486
    },
    {
      "epoch": 0.5193287037037037,
      "grad_norm": 0.08843983709812164,
      "learning_rate": 9.613425925925927e-06,
      "loss": 0.0124,
      "step": 4487
    },
    {
      "epoch": 0.5194444444444445,
      "grad_norm": 0.1223728284239769,
      "learning_rate": 9.611111111111112e-06,
      "loss": 0.022,
      "step": 4488
    },
    {
      "epoch": 0.5195601851851852,
      "grad_norm": 0.1254884898662567,
      "learning_rate": 9.608796296296297e-06,
      "loss": 0.0228,
      "step": 4489
    },
    {
      "epoch": 0.5196759259259259,
      "grad_norm": 0.11026574671268463,
      "learning_rate": 9.606481481481483e-06,
      "loss": 0.0202,
      "step": 4490
    },
    {
      "epoch": 0.5197916666666667,
      "grad_norm": 0.08882021903991699,
      "learning_rate": 9.604166666666669e-06,
      "loss": 0.0161,
      "step": 4491
    },
    {
      "epoch": 0.5199074074074074,
      "grad_norm": 0.07393957674503326,
      "learning_rate": 9.601851851851853e-06,
      "loss": 0.0136,
      "step": 4492
    },
    {
      "epoch": 0.5200231481481481,
      "grad_norm": 0.13045816123485565,
      "learning_rate": 9.599537037037037e-06,
      "loss": 0.022,
      "step": 4493
    },
    {
      "epoch": 0.5201388888888889,
      "grad_norm": 0.08820577710866928,
      "learning_rate": 9.597222222222223e-06,
      "loss": 0.0164,
      "step": 4494
    },
    {
      "epoch": 0.5202546296296297,
      "grad_norm": 0.09933329373598099,
      "learning_rate": 9.594907407407407e-06,
      "loss": 0.0179,
      "step": 4495
    },
    {
      "epoch": 0.5203703703703704,
      "grad_norm": 0.10294869542121887,
      "learning_rate": 9.592592592592593e-06,
      "loss": 0.0187,
      "step": 4496
    },
    {
      "epoch": 0.5204861111111111,
      "grad_norm": 0.07392549514770508,
      "learning_rate": 9.59027777777778e-06,
      "loss": 0.0136,
      "step": 4497
    },
    {
      "epoch": 0.5206018518518518,
      "grad_norm": 0.08601830154657364,
      "learning_rate": 9.587962962962963e-06,
      "loss": 0.0155,
      "step": 4498
    },
    {
      "epoch": 0.5207175925925925,
      "grad_norm": 10.861454010009766,
      "learning_rate": 9.58564814814815e-06,
      "loss": 2.8554,
      "step": 4499
    },
    {
      "epoch": 0.5208333333333334,
      "grad_norm": 0.09479045867919922,
      "learning_rate": 9.583333333333335e-06,
      "loss": 0.0169,
      "step": 4500
    },
    {
      "epoch": 0.5209490740740741,
      "grad_norm": 0.07137219607830048,
      "learning_rate": 9.58101851851852e-06,
      "loss": 0.0126,
      "step": 4501
    },
    {
      "epoch": 0.5210648148148148,
      "grad_norm": 0.09369491785764694,
      "learning_rate": 9.578703703703704e-06,
      "loss": 0.017,
      "step": 4502
    },
    {
      "epoch": 0.5211805555555555,
      "grad_norm": 0.09698307514190674,
      "learning_rate": 9.57638888888889e-06,
      "loss": 0.0171,
      "step": 4503
    },
    {
      "epoch": 0.5212962962962963,
      "grad_norm": 0.09501790255308151,
      "learning_rate": 9.574074074074074e-06,
      "loss": 0.0169,
      "step": 4504
    },
    {
      "epoch": 0.5214120370370371,
      "grad_norm": 0.09829278290271759,
      "learning_rate": 9.57175925925926e-06,
      "loss": 0.0175,
      "step": 4505
    },
    {
      "epoch": 0.5215277777777778,
      "grad_norm": 0.10586204379796982,
      "learning_rate": 9.569444444444446e-06,
      "loss": 0.018,
      "step": 4506
    },
    {
      "epoch": 0.5216435185185185,
      "grad_norm": 0.12608572840690613,
      "learning_rate": 9.56712962962963e-06,
      "loss": 0.0141,
      "step": 4507
    },
    {
      "epoch": 0.5217592592592593,
      "grad_norm": 116.72218322753906,
      "learning_rate": 9.564814814814816e-06,
      "loss": 1.9903,
      "step": 4508
    },
    {
      "epoch": 0.521875,
      "grad_norm": 0.0896960124373436,
      "learning_rate": 9.562500000000002e-06,
      "loss": 0.0168,
      "step": 4509
    },
    {
      "epoch": 0.5219907407407407,
      "grad_norm": 0.10097835212945938,
      "learning_rate": 9.560185185185186e-06,
      "loss": 0.018,
      "step": 4510
    },
    {
      "epoch": 0.5221064814814815,
      "grad_norm": 0.10029540956020355,
      "learning_rate": 9.55787037037037e-06,
      "loss": 0.0176,
      "step": 4511
    },
    {
      "epoch": 0.5222222222222223,
      "grad_norm": 0.10636817663908005,
      "learning_rate": 9.555555555555556e-06,
      "loss": 0.0167,
      "step": 4512
    },
    {
      "epoch": 0.522337962962963,
      "grad_norm": 0.07650116086006165,
      "learning_rate": 9.55324074074074e-06,
      "loss": 0.0143,
      "step": 4513
    },
    {
      "epoch": 0.5224537037037037,
      "grad_norm": 0.11394910514354706,
      "learning_rate": 9.550925925925927e-06,
      "loss": 0.02,
      "step": 4514
    },
    {
      "epoch": 0.5225694444444444,
      "grad_norm": 0.12317876517772675,
      "learning_rate": 9.548611111111113e-06,
      "loss": 0.021,
      "step": 4515
    },
    {
      "epoch": 0.5226851851851851,
      "grad_norm": 0.09068374335765839,
      "learning_rate": 9.546296296296297e-06,
      "loss": 0.0165,
      "step": 4516
    },
    {
      "epoch": 0.522800925925926,
      "grad_norm": 0.0668201893568039,
      "learning_rate": 9.543981481481483e-06,
      "loss": 0.0125,
      "step": 4517
    },
    {
      "epoch": 0.5229166666666667,
      "grad_norm": 0.10493148863315582,
      "learning_rate": 9.541666666666669e-06,
      "loss": 0.0194,
      "step": 4518
    },
    {
      "epoch": 0.5230324074074074,
      "grad_norm": 6.051822662353516,
      "learning_rate": 9.539351851851853e-06,
      "loss": 0.0558,
      "step": 4519
    },
    {
      "epoch": 0.5231481481481481,
      "grad_norm": 0.06315378099679947,
      "learning_rate": 9.537037037037037e-06,
      "loss": 0.0117,
      "step": 4520
    },
    {
      "epoch": 0.5232638888888889,
      "grad_norm": 0.11084932833909988,
      "learning_rate": 9.534722222222223e-06,
      "loss": 0.0176,
      "step": 4521
    },
    {
      "epoch": 0.5233796296296296,
      "grad_norm": 0.1253768503665924,
      "learning_rate": 9.532407407407407e-06,
      "loss": 0.0201,
      "step": 4522
    },
    {
      "epoch": 0.5234953703703704,
      "grad_norm": 0.08044243603944778,
      "learning_rate": 9.530092592592593e-06,
      "loss": 0.0149,
      "step": 4523
    },
    {
      "epoch": 0.5236111111111111,
      "grad_norm": 22.629230499267578,
      "learning_rate": 9.527777777777778e-06,
      "loss": 0.0738,
      "step": 4524
    },
    {
      "epoch": 0.5237268518518519,
      "grad_norm": 0.09995785355567932,
      "learning_rate": 9.525462962962964e-06,
      "loss": 0.0176,
      "step": 4525
    },
    {
      "epoch": 0.5238425925925926,
      "grad_norm": 0.08719733357429504,
      "learning_rate": 9.52314814814815e-06,
      "loss": 0.0155,
      "step": 4526
    },
    {
      "epoch": 0.5239583333333333,
      "grad_norm": 0.08504528552293777,
      "learning_rate": 9.520833333333334e-06,
      "loss": 0.0132,
      "step": 4527
    },
    {
      "epoch": 0.524074074074074,
      "grad_norm": 0.10492627322673798,
      "learning_rate": 9.51851851851852e-06,
      "loss": 0.0173,
      "step": 4528
    },
    {
      "epoch": 0.5241898148148149,
      "grad_norm": 29.207321166992188,
      "learning_rate": 9.516203703703704e-06,
      "loss": 2.8233,
      "step": 4529
    },
    {
      "epoch": 0.5243055555555556,
      "grad_norm": 0.1099252924323082,
      "learning_rate": 9.51388888888889e-06,
      "loss": 0.0192,
      "step": 4530
    },
    {
      "epoch": 0.5244212962962963,
      "grad_norm": 0.3136966824531555,
      "learning_rate": 9.511574074074074e-06,
      "loss": 0.0239,
      "step": 4531
    },
    {
      "epoch": 0.524537037037037,
      "grad_norm": 0.08765659481287003,
      "learning_rate": 9.50925925925926e-06,
      "loss": 0.0115,
      "step": 4532
    },
    {
      "epoch": 0.5246527777777777,
      "grad_norm": 0.1404457837343216,
      "learning_rate": 9.506944444444444e-06,
      "loss": 0.0187,
      "step": 4533
    },
    {
      "epoch": 0.5247685185185185,
      "grad_norm": 0.09849481284618378,
      "learning_rate": 9.50462962962963e-06,
      "loss": 0.0186,
      "step": 4534
    },
    {
      "epoch": 0.5248842592592593,
      "grad_norm": 0.07381901890039444,
      "learning_rate": 9.502314814814816e-06,
      "loss": 0.0139,
      "step": 4535
    },
    {
      "epoch": 0.525,
      "grad_norm": 0.08235428482294083,
      "learning_rate": 9.5e-06,
      "loss": 0.0151,
      "step": 4536
    },
    {
      "epoch": 0.5251157407407407,
      "grad_norm": 0.09665881842374802,
      "learning_rate": 9.497685185185186e-06,
      "loss": 0.0168,
      "step": 4537
    },
    {
      "epoch": 0.5252314814814815,
      "grad_norm": 0.17889012396335602,
      "learning_rate": 9.49537037037037e-06,
      "loss": 0.019,
      "step": 4538
    },
    {
      "epoch": 0.5253472222222222,
      "grad_norm": 0.0850224643945694,
      "learning_rate": 9.493055555555557e-06,
      "loss": 0.0152,
      "step": 4539
    },
    {
      "epoch": 0.5254629629629629,
      "grad_norm": 0.09854397922754288,
      "learning_rate": 9.490740740740741e-06,
      "loss": 0.0183,
      "step": 4540
    },
    {
      "epoch": 0.5255787037037037,
      "grad_norm": 0.08320130407810211,
      "learning_rate": 9.488425925925927e-06,
      "loss": 0.0146,
      "step": 4541
    },
    {
      "epoch": 0.5256944444444445,
      "grad_norm": 0.08342457562685013,
      "learning_rate": 9.486111111111111e-06,
      "loss": 0.0151,
      "step": 4542
    },
    {
      "epoch": 0.5258101851851852,
      "grad_norm": 0.16620171070098877,
      "learning_rate": 9.483796296296297e-06,
      "loss": 0.0216,
      "step": 4543
    },
    {
      "epoch": 0.5259259259259259,
      "grad_norm": 0.11450528353452682,
      "learning_rate": 9.481481481481483e-06,
      "loss": 0.0191,
      "step": 4544
    },
    {
      "epoch": 0.5260416666666666,
      "grad_norm": 0.1143980324268341,
      "learning_rate": 9.479166666666667e-06,
      "loss": 0.0184,
      "step": 4545
    },
    {
      "epoch": 0.5261574074074075,
      "grad_norm": 0.0959048941731453,
      "learning_rate": 9.476851851851853e-06,
      "loss": 0.0174,
      "step": 4546
    },
    {
      "epoch": 0.5262731481481482,
      "grad_norm": 0.08883717656135559,
      "learning_rate": 9.474537037037037e-06,
      "loss": 0.016,
      "step": 4547
    },
    {
      "epoch": 0.5263888888888889,
      "grad_norm": 0.09066297858953476,
      "learning_rate": 9.472222222222223e-06,
      "loss": 0.0159,
      "step": 4548
    },
    {
      "epoch": 0.5265046296296296,
      "grad_norm": 0.09034856408834457,
      "learning_rate": 9.469907407407408e-06,
      "loss": 0.0163,
      "step": 4549
    },
    {
      "epoch": 0.5266203703703703,
      "grad_norm": 0.11012589186429977,
      "learning_rate": 9.467592592592594e-06,
      "loss": 0.0146,
      "step": 4550
    },
    {
      "epoch": 0.5267361111111111,
      "grad_norm": 1.075343370437622,
      "learning_rate": 9.465277777777778e-06,
      "loss": 0.0283,
      "step": 4551
    },
    {
      "epoch": 0.5268518518518519,
      "grad_norm": 0.07667776942253113,
      "learning_rate": 9.462962962962964e-06,
      "loss": 0.0133,
      "step": 4552
    },
    {
      "epoch": 0.5269675925925926,
      "grad_norm": 33.53525924682617,
      "learning_rate": 9.46064814814815e-06,
      "loss": 2.7987,
      "step": 4553
    },
    {
      "epoch": 0.5270833333333333,
      "grad_norm": 0.09354043751955032,
      "learning_rate": 9.458333333333334e-06,
      "loss": 0.0119,
      "step": 4554
    },
    {
      "epoch": 0.5271990740740741,
      "grad_norm": 0.1321839541196823,
      "learning_rate": 9.45601851851852e-06,
      "loss": 0.0216,
      "step": 4555
    },
    {
      "epoch": 0.5273148148148148,
      "grad_norm": 0.07130499929189682,
      "learning_rate": 9.453703703703704e-06,
      "loss": 0.0133,
      "step": 4556
    },
    {
      "epoch": 0.5274305555555555,
      "grad_norm": 0.09818943589925766,
      "learning_rate": 9.45138888888889e-06,
      "loss": 0.0183,
      "step": 4557
    },
    {
      "epoch": 0.5275462962962963,
      "grad_norm": 0.10398059338331223,
      "learning_rate": 9.449074074074074e-06,
      "loss": 0.0196,
      "step": 4558
    },
    {
      "epoch": 0.5276620370370371,
      "grad_norm": 0.10235023498535156,
      "learning_rate": 9.44675925925926e-06,
      "loss": 0.0188,
      "step": 4559
    },
    {
      "epoch": 0.5277777777777778,
      "grad_norm": 0.08021371811628342,
      "learning_rate": 9.444444444444445e-06,
      "loss": 0.0148,
      "step": 4560
    },
    {
      "epoch": 0.5278935185185185,
      "grad_norm": 0.07443324476480484,
      "learning_rate": 9.44212962962963e-06,
      "loss": 0.0139,
      "step": 4561
    },
    {
      "epoch": 0.5280092592592592,
      "grad_norm": 28.76931381225586,
      "learning_rate": 9.439814814814816e-06,
      "loss": 3.1935,
      "step": 4562
    },
    {
      "epoch": 0.528125,
      "grad_norm": 0.09686698764562607,
      "learning_rate": 9.4375e-06,
      "loss": 0.0182,
      "step": 4563
    },
    {
      "epoch": 0.5282407407407408,
      "grad_norm": 0.06842397898435593,
      "learning_rate": 9.435185185185187e-06,
      "loss": 0.0127,
      "step": 4564
    },
    {
      "epoch": 0.5283564814814815,
      "grad_norm": 0.09579400718212128,
      "learning_rate": 9.432870370370371e-06,
      "loss": 0.0167,
      "step": 4565
    },
    {
      "epoch": 0.5284722222222222,
      "grad_norm": 0.07048860937356949,
      "learning_rate": 9.430555555555557e-06,
      "loss": 0.0131,
      "step": 4566
    },
    {
      "epoch": 0.528587962962963,
      "grad_norm": 0.06840433925390244,
      "learning_rate": 9.428240740740741e-06,
      "loss": 0.0122,
      "step": 4567
    },
    {
      "epoch": 0.5287037037037037,
      "grad_norm": 0.09569521993398666,
      "learning_rate": 9.425925925925925e-06,
      "loss": 0.0158,
      "step": 4568
    },
    {
      "epoch": 0.5288194444444444,
      "grad_norm": 0.06621576845645905,
      "learning_rate": 9.423611111111111e-06,
      "loss": 0.0121,
      "step": 4569
    },
    {
      "epoch": 0.5289351851851852,
      "grad_norm": 0.08748342096805573,
      "learning_rate": 9.421296296296297e-06,
      "loss": 0.016,
      "step": 4570
    },
    {
      "epoch": 0.529050925925926,
      "grad_norm": 0.08944343775510788,
      "learning_rate": 9.418981481481481e-06,
      "loss": 0.0153,
      "step": 4571
    },
    {
      "epoch": 0.5291666666666667,
      "grad_norm": 0.11880574375391006,
      "learning_rate": 9.416666666666667e-06,
      "loss": 0.019,
      "step": 4572
    },
    {
      "epoch": 0.5292824074074074,
      "grad_norm": 0.07233107089996338,
      "learning_rate": 9.414351851851853e-06,
      "loss": 0.0132,
      "step": 4573
    },
    {
      "epoch": 0.5293981481481481,
      "grad_norm": 0.08974684029817581,
      "learning_rate": 9.412037037037038e-06,
      "loss": 0.0159,
      "step": 4574
    },
    {
      "epoch": 0.5295138888888888,
      "grad_norm": 0.10199834406375885,
      "learning_rate": 9.409722222222224e-06,
      "loss": 0.0176,
      "step": 4575
    },
    {
      "epoch": 0.5296296296296297,
      "grad_norm": 0.06335853040218353,
      "learning_rate": 9.407407407407408e-06,
      "loss": 0.0116,
      "step": 4576
    },
    {
      "epoch": 0.5297453703703704,
      "grad_norm": 0.5273959040641785,
      "learning_rate": 9.405092592592592e-06,
      "loss": 0.0153,
      "step": 4577
    },
    {
      "epoch": 0.5298611111111111,
      "grad_norm": 0.10088775306940079,
      "learning_rate": 9.402777777777778e-06,
      "loss": 0.0178,
      "step": 4578
    },
    {
      "epoch": 0.5299768518518518,
      "grad_norm": 0.11235212534666061,
      "learning_rate": 9.400462962962964e-06,
      "loss": 0.0209,
      "step": 4579
    },
    {
      "epoch": 0.5300925925925926,
      "grad_norm": 0.06665762513875961,
      "learning_rate": 9.398148148148148e-06,
      "loss": 0.0122,
      "step": 4580
    },
    {
      "epoch": 0.5302083333333333,
      "grad_norm": 0.11830222606658936,
      "learning_rate": 9.395833333333334e-06,
      "loss": 0.022,
      "step": 4581
    },
    {
      "epoch": 0.5303240740740741,
      "grad_norm": 0.17457379400730133,
      "learning_rate": 9.39351851851852e-06,
      "loss": 0.018,
      "step": 4582
    },
    {
      "epoch": 0.5304398148148148,
      "grad_norm": 0.09125745296478271,
      "learning_rate": 9.391203703703704e-06,
      "loss": 0.0169,
      "step": 4583
    },
    {
      "epoch": 0.5305555555555556,
      "grad_norm": 2.2244675159454346,
      "learning_rate": 9.38888888888889e-06,
      "loss": 0.0319,
      "step": 4584
    },
    {
      "epoch": 0.5306712962962963,
      "grad_norm": 0.06276007741689682,
      "learning_rate": 9.386574074074075e-06,
      "loss": 0.0116,
      "step": 4585
    },
    {
      "epoch": 0.530787037037037,
      "grad_norm": 53.027793884277344,
      "learning_rate": 9.384259259259259e-06,
      "loss": 2.5735,
      "step": 4586
    },
    {
      "epoch": 0.5309027777777777,
      "grad_norm": 56.78392028808594,
      "learning_rate": 9.381944444444445e-06,
      "loss": 2.1437,
      "step": 4587
    },
    {
      "epoch": 0.5310185185185186,
      "grad_norm": 0.09236814081668854,
      "learning_rate": 9.37962962962963e-06,
      "loss": 0.0165,
      "step": 4588
    },
    {
      "epoch": 0.5311342592592593,
      "grad_norm": 0.09413104504346848,
      "learning_rate": 9.377314814814815e-06,
      "loss": 0.0163,
      "step": 4589
    },
    {
      "epoch": 0.53125,
      "grad_norm": 0.09286193549633026,
      "learning_rate": 9.375000000000001e-06,
      "loss": 0.0121,
      "step": 4590
    },
    {
      "epoch": 0.5313657407407407,
      "grad_norm": 0.13275285065174103,
      "learning_rate": 9.372685185185187e-06,
      "loss": 0.0179,
      "step": 4591
    },
    {
      "epoch": 0.5314814814814814,
      "grad_norm": 1.9557859897613525,
      "learning_rate": 9.370370370370371e-06,
      "loss": 0.0332,
      "step": 4592
    },
    {
      "epoch": 0.5315972222222223,
      "grad_norm": 0.14798130095005035,
      "learning_rate": 9.368055555555557e-06,
      "loss": 0.0201,
      "step": 4593
    },
    {
      "epoch": 0.531712962962963,
      "grad_norm": 0.09450366348028183,
      "learning_rate": 9.365740740740741e-06,
      "loss": 0.0171,
      "step": 4594
    },
    {
      "epoch": 0.5318287037037037,
      "grad_norm": 0.06538072228431702,
      "learning_rate": 9.363425925925927e-06,
      "loss": 0.0122,
      "step": 4595
    },
    {
      "epoch": 0.5319444444444444,
      "grad_norm": 0.06490623950958252,
      "learning_rate": 9.361111111111111e-06,
      "loss": 0.0121,
      "step": 4596
    },
    {
      "epoch": 0.5320601851851852,
      "grad_norm": 0.07150936126708984,
      "learning_rate": 9.358796296296297e-06,
      "loss": 0.0132,
      "step": 4597
    },
    {
      "epoch": 0.5321759259259259,
      "grad_norm": 0.08821636438369751,
      "learning_rate": 9.356481481481482e-06,
      "loss": 0.0163,
      "step": 4598
    },
    {
      "epoch": 0.5322916666666667,
      "grad_norm": 0.08643694221973419,
      "learning_rate": 9.354166666666668e-06,
      "loss": 0.0157,
      "step": 4599
    },
    {
      "epoch": 0.5324074074074074,
      "grad_norm": 0.07232944667339325,
      "learning_rate": 9.351851851851854e-06,
      "loss": 0.0132,
      "step": 4600
    },
    {
      "epoch": 0.5325231481481482,
      "grad_norm": 24.735692977905273,
      "learning_rate": 9.349537037037038e-06,
      "loss": 2.7658,
      "step": 4601
    },
    {
      "epoch": 0.5326388888888889,
      "grad_norm": 0.10240079462528229,
      "learning_rate": 9.347222222222224e-06,
      "loss": 0.0191,
      "step": 4602
    },
    {
      "epoch": 0.5327546296296296,
      "grad_norm": 0.060776274651288986,
      "learning_rate": 9.344907407407408e-06,
      "loss": 0.0112,
      "step": 4603
    },
    {
      "epoch": 0.5328703703703703,
      "grad_norm": 0.1275230199098587,
      "learning_rate": 9.342592592592594e-06,
      "loss": 0.0205,
      "step": 4604
    },
    {
      "epoch": 0.5329861111111112,
      "grad_norm": 0.09415552020072937,
      "learning_rate": 9.340277777777778e-06,
      "loss": 0.0133,
      "step": 4605
    },
    {
      "epoch": 0.5331018518518519,
      "grad_norm": 0.10420901328325272,
      "learning_rate": 9.337962962962964e-06,
      "loss": 0.0191,
      "step": 4606
    },
    {
      "epoch": 0.5332175925925926,
      "grad_norm": 0.08256050944328308,
      "learning_rate": 9.335648148148148e-06,
      "loss": 0.0148,
      "step": 4607
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.6728336215019226,
      "learning_rate": 9.333333333333334e-06,
      "loss": 0.0183,
      "step": 4608
    },
    {
      "epoch": 0.533449074074074,
      "grad_norm": 0.11247982084751129,
      "learning_rate": 9.33101851851852e-06,
      "loss": 0.0198,
      "step": 4609
    },
    {
      "epoch": 0.5335648148148148,
      "grad_norm": 0.06058874353766441,
      "learning_rate": 9.328703703703705e-06,
      "loss": 0.0112,
      "step": 4610
    },
    {
      "epoch": 0.5336805555555556,
      "grad_norm": 0.0742368996143341,
      "learning_rate": 9.32638888888889e-06,
      "loss": 0.0135,
      "step": 4611
    },
    {
      "epoch": 0.5337962962962963,
      "grad_norm": 0.0878075584769249,
      "learning_rate": 9.324074074074075e-06,
      "loss": 0.0116,
      "step": 4612
    },
    {
      "epoch": 0.533912037037037,
      "grad_norm": 0.07249575108289719,
      "learning_rate": 9.32175925925926e-06,
      "loss": 0.0133,
      "step": 4613
    },
    {
      "epoch": 0.5340277777777778,
      "grad_norm": 0.0956396833062172,
      "learning_rate": 9.319444444444445e-06,
      "loss": 0.0179,
      "step": 4614
    },
    {
      "epoch": 0.5341435185185185,
      "grad_norm": 5.184231281280518,
      "learning_rate": 9.31712962962963e-06,
      "loss": 0.0436,
      "step": 4615
    },
    {
      "epoch": 0.5342592592592592,
      "grad_norm": 0.06247773393988609,
      "learning_rate": 9.314814814814815e-06,
      "loss": 0.0116,
      "step": 4616
    },
    {
      "epoch": 0.534375,
      "grad_norm": 44.39402389526367,
      "learning_rate": 9.312500000000001e-06,
      "loss": 2.1948,
      "step": 4617
    },
    {
      "epoch": 0.5344907407407408,
      "grad_norm": 0.062369052320718765,
      "learning_rate": 9.310185185185185e-06,
      "loss": 0.0116,
      "step": 4618
    },
    {
      "epoch": 0.5346064814814815,
      "grad_norm": 0.09526598453521729,
      "learning_rate": 9.307870370370371e-06,
      "loss": 0.0165,
      "step": 4619
    },
    {
      "epoch": 0.5347222222222222,
      "grad_norm": 0.08172345161437988,
      "learning_rate": 9.305555555555557e-06,
      "loss": 0.0152,
      "step": 4620
    },
    {
      "epoch": 0.5348379629629629,
      "grad_norm": 0.08751380443572998,
      "learning_rate": 9.303240740740741e-06,
      "loss": 0.0158,
      "step": 4621
    },
    {
      "epoch": 0.5349537037037037,
      "grad_norm": 0.0875367745757103,
      "learning_rate": 9.300925925925927e-06,
      "loss": 0.0154,
      "step": 4622
    },
    {
      "epoch": 0.5350694444444445,
      "grad_norm": 0.09937398135662079,
      "learning_rate": 9.298611111111112e-06,
      "loss": 0.0187,
      "step": 4623
    },
    {
      "epoch": 0.5351851851851852,
      "grad_norm": 0.06482522934675217,
      "learning_rate": 9.296296296296296e-06,
      "loss": 0.0121,
      "step": 4624
    },
    {
      "epoch": 0.5353009259259259,
      "grad_norm": 42.4340705871582,
      "learning_rate": 9.293981481481482e-06,
      "loss": 2.3304,
      "step": 4625
    },
    {
      "epoch": 0.5354166666666667,
      "grad_norm": 0.0846378356218338,
      "learning_rate": 9.291666666666668e-06,
      "loss": 0.0155,
      "step": 4626
    },
    {
      "epoch": 0.5355324074074074,
      "grad_norm": 14.7741117477417,
      "learning_rate": 9.289351851851852e-06,
      "loss": 2.5082,
      "step": 4627
    },
    {
      "epoch": 0.5356481481481481,
      "grad_norm": 108.00638580322266,
      "learning_rate": 9.287037037037038e-06,
      "loss": 0.4615,
      "step": 4628
    },
    {
      "epoch": 0.5357638888888889,
      "grad_norm": 0.11121147125959396,
      "learning_rate": 9.284722222222224e-06,
      "loss": 0.0148,
      "step": 4629
    },
    {
      "epoch": 0.5358796296296297,
      "grad_norm": 0.06551767885684967,
      "learning_rate": 9.282407407407408e-06,
      "loss": 0.012,
      "step": 4630
    },
    {
      "epoch": 0.5359953703703704,
      "grad_norm": 0.099633127450943,
      "learning_rate": 9.280092592592594e-06,
      "loss": 0.0186,
      "step": 4631
    },
    {
      "epoch": 0.5361111111111111,
      "grad_norm": 48.58491516113281,
      "learning_rate": 9.277777777777778e-06,
      "loss": 2.2271,
      "step": 4632
    },
    {
      "epoch": 0.5362268518518518,
      "grad_norm": 0.17336763441562653,
      "learning_rate": 9.275462962962963e-06,
      "loss": 0.0212,
      "step": 4633
    },
    {
      "epoch": 0.5363425925925925,
      "grad_norm": 0.09395653009414673,
      "learning_rate": 9.273148148148149e-06,
      "loss": 0.0162,
      "step": 4634
    },
    {
      "epoch": 0.5364583333333334,
      "grad_norm": 0.06784985959529877,
      "learning_rate": 9.270833333333334e-06,
      "loss": 0.0126,
      "step": 4635
    },
    {
      "epoch": 0.5365740740740741,
      "grad_norm": 0.08247096091508865,
      "learning_rate": 9.268518518518519e-06,
      "loss": 0.0153,
      "step": 4636
    },
    {
      "epoch": 0.5366898148148148,
      "grad_norm": 0.3056682348251343,
      "learning_rate": 9.266203703703705e-06,
      "loss": 0.0202,
      "step": 4637
    },
    {
      "epoch": 0.5368055555555555,
      "grad_norm": 0.08736994117498398,
      "learning_rate": 9.26388888888889e-06,
      "loss": 0.0156,
      "step": 4638
    },
    {
      "epoch": 0.5369212962962963,
      "grad_norm": 0.41167473793029785,
      "learning_rate": 9.261574074074075e-06,
      "loss": 0.0232,
      "step": 4639
    },
    {
      "epoch": 0.5370370370370371,
      "grad_norm": 0.10685043781995773,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.02,
      "step": 4640
    },
    {
      "epoch": 0.5371527777777778,
      "grad_norm": 0.09115291386842728,
      "learning_rate": 9.256944444444445e-06,
      "loss": 0.0155,
      "step": 4641
    },
    {
      "epoch": 0.5372685185185185,
      "grad_norm": 0.16154982149600983,
      "learning_rate": 9.25462962962963e-06,
      "loss": 0.0138,
      "step": 4642
    },
    {
      "epoch": 0.5373842592592593,
      "grad_norm": 0.29103225469589233,
      "learning_rate": 9.252314814814815e-06,
      "loss": 0.0224,
      "step": 4643
    },
    {
      "epoch": 0.5375,
      "grad_norm": 0.5938490629196167,
      "learning_rate": 9.250000000000001e-06,
      "loss": 0.0206,
      "step": 4644
    },
    {
      "epoch": 0.5376157407407407,
      "grad_norm": 0.09568068385124207,
      "learning_rate": 9.247685185185185e-06,
      "loss": 0.0178,
      "step": 4645
    },
    {
      "epoch": 0.5377314814814815,
      "grad_norm": 0.06516966968774796,
      "learning_rate": 9.245370370370371e-06,
      "loss": 0.0117,
      "step": 4646
    },
    {
      "epoch": 0.5378472222222223,
      "grad_norm": 100.02662658691406,
      "learning_rate": 9.243055555555557e-06,
      "loss": 2.0288,
      "step": 4647
    },
    {
      "epoch": 0.537962962962963,
      "grad_norm": 0.0818789005279541,
      "learning_rate": 9.240740740740742e-06,
      "loss": 0.0147,
      "step": 4648
    },
    {
      "epoch": 0.5380787037037037,
      "grad_norm": 39.86393737792969,
      "learning_rate": 9.238425925925928e-06,
      "loss": 0.0918,
      "step": 4649
    },
    {
      "epoch": 0.5381944444444444,
      "grad_norm": 0.06781242787837982,
      "learning_rate": 9.236111111111112e-06,
      "loss": 0.0126,
      "step": 4650
    },
    {
      "epoch": 0.5383101851851851,
      "grad_norm": 0.08476821333169937,
      "learning_rate": 9.233796296296296e-06,
      "loss": 0.0152,
      "step": 4651
    },
    {
      "epoch": 0.538425925925926,
      "grad_norm": 0.23245565593242645,
      "learning_rate": 9.231481481481482e-06,
      "loss": 0.0222,
      "step": 4652
    },
    {
      "epoch": 0.5385416666666667,
      "grad_norm": 0.05998765304684639,
      "learning_rate": 9.229166666666668e-06,
      "loss": 0.0111,
      "step": 4653
    },
    {
      "epoch": 0.5386574074074074,
      "grad_norm": 0.08929476886987686,
      "learning_rate": 9.226851851851852e-06,
      "loss": 0.015,
      "step": 4654
    },
    {
      "epoch": 0.5387731481481481,
      "grad_norm": 0.08826210349798203,
      "learning_rate": 9.224537037037038e-06,
      "loss": 0.0157,
      "step": 4655
    },
    {
      "epoch": 0.5388888888888889,
      "grad_norm": 0.06551147997379303,
      "learning_rate": 9.222222222222224e-06,
      "loss": 0.0116,
      "step": 4656
    },
    {
      "epoch": 0.5390046296296296,
      "grad_norm": 0.1183481365442276,
      "learning_rate": 9.219907407407408e-06,
      "loss": 0.0217,
      "step": 4657
    },
    {
      "epoch": 0.5391203703703704,
      "grad_norm": 0.09473388642072678,
      "learning_rate": 9.217592592592594e-06,
      "loss": 0.0174,
      "step": 4658
    },
    {
      "epoch": 0.5392361111111111,
      "grad_norm": 12.244282722473145,
      "learning_rate": 9.215277777777779e-06,
      "loss": 0.0718,
      "step": 4659
    },
    {
      "epoch": 0.5393518518518519,
      "grad_norm": 0.2826715111732483,
      "learning_rate": 9.212962962962963e-06,
      "loss": 0.0159,
      "step": 4660
    },
    {
      "epoch": 0.5394675925925926,
      "grad_norm": 0.06672976911067963,
      "learning_rate": 9.210648148148149e-06,
      "loss": 0.0124,
      "step": 4661
    },
    {
      "epoch": 0.5395833333333333,
      "grad_norm": 0.08609326183795929,
      "learning_rate": 9.208333333333333e-06,
      "loss": 0.0155,
      "step": 4662
    },
    {
      "epoch": 0.539699074074074,
      "grad_norm": 0.0906345322728157,
      "learning_rate": 9.206018518518519e-06,
      "loss": 0.0164,
      "step": 4663
    },
    {
      "epoch": 0.5398148148148149,
      "grad_norm": 0.1491333246231079,
      "learning_rate": 9.203703703703705e-06,
      "loss": 0.0183,
      "step": 4664
    },
    {
      "epoch": 0.5399305555555556,
      "grad_norm": 0.06150251626968384,
      "learning_rate": 9.201388888888889e-06,
      "loss": 0.0115,
      "step": 4665
    },
    {
      "epoch": 0.5400462962962963,
      "grad_norm": 0.10439072549343109,
      "learning_rate": 9.199074074074075e-06,
      "loss": 0.019,
      "step": 4666
    },
    {
      "epoch": 0.540162037037037,
      "grad_norm": 55.301361083984375,
      "learning_rate": 9.196759259259261e-06,
      "loss": 2.5178,
      "step": 4667
    },
    {
      "epoch": 0.5402777777777777,
      "grad_norm": 0.5718390941619873,
      "learning_rate": 9.194444444444445e-06,
      "loss": 0.0247,
      "step": 4668
    },
    {
      "epoch": 0.5403935185185185,
      "grad_norm": 0.08369524776935577,
      "learning_rate": 9.19212962962963e-06,
      "loss": 0.015,
      "step": 4669
    },
    {
      "epoch": 0.5405092592592593,
      "grad_norm": 0.1854953020811081,
      "learning_rate": 9.189814814814815e-06,
      "loss": 0.0176,
      "step": 4670
    },
    {
      "epoch": 0.540625,
      "grad_norm": 0.09761025011539459,
      "learning_rate": 9.1875e-06,
      "loss": 0.0182,
      "step": 4671
    },
    {
      "epoch": 0.5407407407407407,
      "grad_norm": 0.11099311709403992,
      "learning_rate": 9.185185185185186e-06,
      "loss": 0.021,
      "step": 4672
    },
    {
      "epoch": 0.5408564814814815,
      "grad_norm": 0.0926145687699318,
      "learning_rate": 9.182870370370372e-06,
      "loss": 0.0167,
      "step": 4673
    },
    {
      "epoch": 0.5409722222222222,
      "grad_norm": 0.0832972526550293,
      "learning_rate": 9.180555555555556e-06,
      "loss": 0.0153,
      "step": 4674
    },
    {
      "epoch": 0.5410879629629629,
      "grad_norm": 0.07150231301784515,
      "learning_rate": 9.178240740740742e-06,
      "loss": 0.012,
      "step": 4675
    },
    {
      "epoch": 0.5412037037037037,
      "grad_norm": 0.09830772876739502,
      "learning_rate": 9.175925925925928e-06,
      "loss": 0.0184,
      "step": 4676
    },
    {
      "epoch": 0.5413194444444445,
      "grad_norm": 0.09075812995433807,
      "learning_rate": 9.173611111111112e-06,
      "loss": 0.0162,
      "step": 4677
    },
    {
      "epoch": 0.5414351851851852,
      "grad_norm": 0.0629039779305458,
      "learning_rate": 9.171296296296296e-06,
      "loss": 0.0118,
      "step": 4678
    },
    {
      "epoch": 0.5415509259259259,
      "grad_norm": 1.705384373664856,
      "learning_rate": 9.168981481481482e-06,
      "loss": 0.0333,
      "step": 4679
    },
    {
      "epoch": 0.5416666666666666,
      "grad_norm": 0.10654783248901367,
      "learning_rate": 9.166666666666666e-06,
      "loss": 0.0143,
      "step": 4680
    },
    {
      "epoch": 0.5417824074074075,
      "grad_norm": 0.0815146416425705,
      "learning_rate": 9.164351851851852e-06,
      "loss": 0.0148,
      "step": 4681
    },
    {
      "epoch": 0.5418981481481482,
      "grad_norm": 0.07458093017339706,
      "learning_rate": 9.162037037037038e-06,
      "loss": 0.0134,
      "step": 4682
    },
    {
      "epoch": 0.5420138888888889,
      "grad_norm": 0.06249675154685974,
      "learning_rate": 9.159722222222223e-06,
      "loss": 0.0117,
      "step": 4683
    },
    {
      "epoch": 0.5421296296296296,
      "grad_norm": 0.1555621325969696,
      "learning_rate": 9.157407407407409e-06,
      "loss": 0.0156,
      "step": 4684
    },
    {
      "epoch": 0.5422453703703703,
      "grad_norm": 0.09066943824291229,
      "learning_rate": 9.155092592592594e-06,
      "loss": 0.0164,
      "step": 4685
    },
    {
      "epoch": 0.5423611111111111,
      "grad_norm": 0.09552644193172455,
      "learning_rate": 9.152777777777779e-06,
      "loss": 0.0173,
      "step": 4686
    },
    {
      "epoch": 0.5424768518518519,
      "grad_norm": 0.07904259860515594,
      "learning_rate": 9.150462962962963e-06,
      "loss": 0.0106,
      "step": 4687
    },
    {
      "epoch": 0.5425925925925926,
      "grad_norm": 0.08624586462974548,
      "learning_rate": 9.148148148148149e-06,
      "loss": 0.016,
      "step": 4688
    },
    {
      "epoch": 0.5427083333333333,
      "grad_norm": 0.12526457011699677,
      "learning_rate": 9.145833333333333e-06,
      "loss": 0.0185,
      "step": 4689
    },
    {
      "epoch": 0.5428240740740741,
      "grad_norm": 0.13300466537475586,
      "learning_rate": 9.143518518518519e-06,
      "loss": 0.0174,
      "step": 4690
    },
    {
      "epoch": 0.5429398148148148,
      "grad_norm": 0.09173101931810379,
      "learning_rate": 9.141203703703705e-06,
      "loss": 0.0165,
      "step": 4691
    },
    {
      "epoch": 0.5430555555555555,
      "grad_norm": 0.08951902389526367,
      "learning_rate": 9.13888888888889e-06,
      "loss": 0.0144,
      "step": 4692
    },
    {
      "epoch": 0.5431712962962963,
      "grad_norm": 0.08300921320915222,
      "learning_rate": 9.136574074074075e-06,
      "loss": 0.013,
      "step": 4693
    },
    {
      "epoch": 0.5432870370370371,
      "grad_norm": 0.0786530151963234,
      "learning_rate": 9.134259259259261e-06,
      "loss": 0.0105,
      "step": 4694
    },
    {
      "epoch": 0.5434027777777778,
      "grad_norm": 0.10967633128166199,
      "learning_rate": 9.131944444444445e-06,
      "loss": 0.0203,
      "step": 4695
    },
    {
      "epoch": 0.5435185185185185,
      "grad_norm": 0.09992974996566772,
      "learning_rate": 9.12962962962963e-06,
      "loss": 0.0167,
      "step": 4696
    },
    {
      "epoch": 0.5436342592592592,
      "grad_norm": 0.11609125882387161,
      "learning_rate": 9.127314814814816e-06,
      "loss": 0.0155,
      "step": 4697
    },
    {
      "epoch": 0.54375,
      "grad_norm": 0.07933623343706131,
      "learning_rate": 9.125e-06,
      "loss": 0.0146,
      "step": 4698
    },
    {
      "epoch": 0.5438657407407408,
      "grad_norm": 0.08750073611736298,
      "learning_rate": 9.122685185185186e-06,
      "loss": 0.0162,
      "step": 4699
    },
    {
      "epoch": 0.5439814814814815,
      "grad_norm": 1.4902945756912231,
      "learning_rate": 9.120370370370372e-06,
      "loss": 0.0215,
      "step": 4700
    },
    {
      "epoch": 0.5440972222222222,
      "grad_norm": 0.9853367209434509,
      "learning_rate": 9.118055555555556e-06,
      "loss": 0.0211,
      "step": 4701
    },
    {
      "epoch": 0.544212962962963,
      "grad_norm": 48.32392501831055,
      "learning_rate": 9.115740740740742e-06,
      "loss": 2.9936,
      "step": 4702
    },
    {
      "epoch": 0.5443287037037037,
      "grad_norm": 0.14227445423603058,
      "learning_rate": 9.113425925925926e-06,
      "loss": 0.0193,
      "step": 4703
    },
    {
      "epoch": 0.5444444444444444,
      "grad_norm": 0.08479562401771545,
      "learning_rate": 9.111111111111112e-06,
      "loss": 0.0153,
      "step": 4704
    },
    {
      "epoch": 0.5445601851851852,
      "grad_norm": 0.08845941722393036,
      "learning_rate": 9.108796296296296e-06,
      "loss": 0.0159,
      "step": 4705
    },
    {
      "epoch": 0.544675925925926,
      "grad_norm": 0.09949402511119843,
      "learning_rate": 9.106481481481482e-06,
      "loss": 0.0117,
      "step": 4706
    },
    {
      "epoch": 0.5447916666666667,
      "grad_norm": 0.09040677547454834,
      "learning_rate": 9.104166666666667e-06,
      "loss": 0.0141,
      "step": 4707
    },
    {
      "epoch": 0.5449074074074074,
      "grad_norm": 0.05712427198886871,
      "learning_rate": 9.101851851851853e-06,
      "loss": 0.0107,
      "step": 4708
    },
    {
      "epoch": 0.5450231481481481,
      "grad_norm": 0.083210289478302,
      "learning_rate": 9.099537037037037e-06,
      "loss": 0.0154,
      "step": 4709
    },
    {
      "epoch": 0.5451388888888888,
      "grad_norm": 0.07973944395780563,
      "learning_rate": 9.097222222222223e-06,
      "loss": 0.0145,
      "step": 4710
    },
    {
      "epoch": 0.5452546296296297,
      "grad_norm": 0.061743393540382385,
      "learning_rate": 9.094907407407409e-06,
      "loss": 0.0112,
      "step": 4711
    },
    {
      "epoch": 0.5453703703703704,
      "grad_norm": 0.0585666224360466,
      "learning_rate": 9.092592592592593e-06,
      "loss": 0.0108,
      "step": 4712
    },
    {
      "epoch": 0.5454861111111111,
      "grad_norm": 0.17857062816619873,
      "learning_rate": 9.090277777777779e-06,
      "loss": 0.0195,
      "step": 4713
    },
    {
      "epoch": 0.5456018518518518,
      "grad_norm": 0.9324958920478821,
      "learning_rate": 9.087962962962965e-06,
      "loss": 0.025,
      "step": 4714
    },
    {
      "epoch": 0.5457175925925926,
      "grad_norm": 0.4719657301902771,
      "learning_rate": 9.085648148148149e-06,
      "loss": 0.0234,
      "step": 4715
    },
    {
      "epoch": 0.5458333333333333,
      "grad_norm": 135.6300811767578,
      "learning_rate": 9.083333333333333e-06,
      "loss": 2.3535,
      "step": 4716
    },
    {
      "epoch": 0.5459490740740741,
      "grad_norm": 0.10863303393125534,
      "learning_rate": 9.08101851851852e-06,
      "loss": 0.02,
      "step": 4717
    },
    {
      "epoch": 0.5460648148148148,
      "grad_norm": 0.08226939290761948,
      "learning_rate": 9.078703703703704e-06,
      "loss": 0.0151,
      "step": 4718
    },
    {
      "epoch": 0.5461805555555556,
      "grad_norm": 0.0760817751288414,
      "learning_rate": 9.07638888888889e-06,
      "loss": 0.0138,
      "step": 4719
    },
    {
      "epoch": 0.5462962962962963,
      "grad_norm": 0.17982842028141022,
      "learning_rate": 9.074074074074075e-06,
      "loss": 0.0191,
      "step": 4720
    },
    {
      "epoch": 0.546412037037037,
      "grad_norm": 3.4405975341796875,
      "learning_rate": 9.07175925925926e-06,
      "loss": 0.043,
      "step": 4721
    },
    {
      "epoch": 0.5465277777777777,
      "grad_norm": 0.1029641330242157,
      "learning_rate": 9.069444444444446e-06,
      "loss": 0.0189,
      "step": 4722
    },
    {
      "epoch": 0.5466435185185186,
      "grad_norm": 0.08907770365476608,
      "learning_rate": 9.067129629629632e-06,
      "loss": 0.0161,
      "step": 4723
    },
    {
      "epoch": 0.5467592592592593,
      "grad_norm": 0.084515780210495,
      "learning_rate": 9.064814814814816e-06,
      "loss": 0.0152,
      "step": 4724
    },
    {
      "epoch": 0.546875,
      "grad_norm": 0.06141634285449982,
      "learning_rate": 9.0625e-06,
      "loss": 0.0114,
      "step": 4725
    },
    {
      "epoch": 0.5469907407407407,
      "grad_norm": 0.10878755152225494,
      "learning_rate": 9.060185185185186e-06,
      "loss": 0.0144,
      "step": 4726
    },
    {
      "epoch": 0.5471064814814814,
      "grad_norm": 0.28286638855934143,
      "learning_rate": 9.05787037037037e-06,
      "loss": 0.0174,
      "step": 4727
    },
    {
      "epoch": 0.5472222222222223,
      "grad_norm": 0.08397246897220612,
      "learning_rate": 9.055555555555556e-06,
      "loss": 0.0151,
      "step": 4728
    },
    {
      "epoch": 0.547337962962963,
      "grad_norm": 0.07970771193504333,
      "learning_rate": 9.053240740740742e-06,
      "loss": 0.0127,
      "step": 4729
    },
    {
      "epoch": 0.5474537037037037,
      "grad_norm": 0.3450596034526825,
      "learning_rate": 9.050925925925926e-06,
      "loss": 0.0159,
      "step": 4730
    },
    {
      "epoch": 0.5475694444444444,
      "grad_norm": 0.0815528854727745,
      "learning_rate": 9.048611111111112e-06,
      "loss": 0.0147,
      "step": 4731
    },
    {
      "epoch": 0.5476851851851852,
      "grad_norm": 0.08397866040468216,
      "learning_rate": 9.046296296296298e-06,
      "loss": 0.0153,
      "step": 4732
    },
    {
      "epoch": 0.5478009259259259,
      "grad_norm": 76.54485321044922,
      "learning_rate": 9.043981481481483e-06,
      "loss": 0.5298,
      "step": 4733
    },
    {
      "epoch": 0.5479166666666667,
      "grad_norm": 0.09663334488868713,
      "learning_rate": 9.041666666666667e-06,
      "loss": 0.018,
      "step": 4734
    },
    {
      "epoch": 0.5480324074074074,
      "grad_norm": 0.07894468307495117,
      "learning_rate": 9.039351851851853e-06,
      "loss": 0.0142,
      "step": 4735
    },
    {
      "epoch": 0.5481481481481482,
      "grad_norm": 132.42030334472656,
      "learning_rate": 9.037037037037037e-06,
      "loss": 1.8237,
      "step": 4736
    },
    {
      "epoch": 0.5482638888888889,
      "grad_norm": 0.08917608857154846,
      "learning_rate": 9.034722222222223e-06,
      "loss": 0.0152,
      "step": 4737
    },
    {
      "epoch": 0.5483796296296296,
      "grad_norm": 5.121174335479736,
      "learning_rate": 9.032407407407409e-06,
      "loss": 0.032,
      "step": 4738
    },
    {
      "epoch": 0.5484953703703703,
      "grad_norm": 0.1454741656780243,
      "learning_rate": 9.030092592592593e-06,
      "loss": 0.0176,
      "step": 4739
    },
    {
      "epoch": 0.5486111111111112,
      "grad_norm": 0.07118858397006989,
      "learning_rate": 9.027777777777779e-06,
      "loss": 0.0126,
      "step": 4740
    },
    {
      "epoch": 0.5487268518518519,
      "grad_norm": 0.07261163741350174,
      "learning_rate": 9.025462962962965e-06,
      "loss": 0.0132,
      "step": 4741
    },
    {
      "epoch": 0.5488425925925926,
      "grad_norm": 0.07218516618013382,
      "learning_rate": 9.02314814814815e-06,
      "loss": 0.0132,
      "step": 4742
    },
    {
      "epoch": 0.5489583333333333,
      "grad_norm": 0.47591593861579895,
      "learning_rate": 9.020833333333334e-06,
      "loss": 0.0252,
      "step": 4743
    },
    {
      "epoch": 0.549074074074074,
      "grad_norm": 0.07521120458841324,
      "learning_rate": 9.01851851851852e-06,
      "loss": 0.0137,
      "step": 4744
    },
    {
      "epoch": 0.5491898148148148,
      "grad_norm": 0.05780857056379318,
      "learning_rate": 9.016203703703704e-06,
      "loss": 0.0107,
      "step": 4745
    },
    {
      "epoch": 0.5493055555555556,
      "grad_norm": 0.06626744568347931,
      "learning_rate": 9.01388888888889e-06,
      "loss": 0.0123,
      "step": 4746
    },
    {
      "epoch": 0.5494212962962963,
      "grad_norm": 121.13703155517578,
      "learning_rate": 9.011574074074076e-06,
      "loss": 1.6946,
      "step": 4747
    },
    {
      "epoch": 0.549537037037037,
      "grad_norm": 0.1342344582080841,
      "learning_rate": 9.00925925925926e-06,
      "loss": 0.0178,
      "step": 4748
    },
    {
      "epoch": 0.5496527777777778,
      "grad_norm": 0.09161585569381714,
      "learning_rate": 9.006944444444446e-06,
      "loss": 0.0171,
      "step": 4749
    },
    {
      "epoch": 0.5497685185185185,
      "grad_norm": 0.08870742470026016,
      "learning_rate": 9.00462962962963e-06,
      "loss": 0.0152,
      "step": 4750
    },
    {
      "epoch": 0.5498842592592592,
      "grad_norm": 14.493951797485352,
      "learning_rate": 9.002314814814816e-06,
      "loss": 3.0332,
      "step": 4751
    },
    {
      "epoch": 0.55,
      "grad_norm": 7.128829002380371,
      "learning_rate": 9e-06,
      "loss": 0.0485,
      "step": 4752
    },
    {
      "epoch": 0.5501157407407408,
      "grad_norm": 0.11659294366836548,
      "learning_rate": 8.997685185185186e-06,
      "loss": 0.0201,
      "step": 4753
    },
    {
      "epoch": 0.5502314814814815,
      "grad_norm": 1.1065585613250732,
      "learning_rate": 8.99537037037037e-06,
      "loss": 0.03,
      "step": 4754
    },
    {
      "epoch": 0.5503472222222222,
      "grad_norm": 1.7844020128250122,
      "learning_rate": 8.993055555555556e-06,
      "loss": 0.0359,
      "step": 4755
    },
    {
      "epoch": 0.5504629629629629,
      "grad_norm": 0.09433342516422272,
      "learning_rate": 8.99074074074074e-06,
      "loss": 0.0164,
      "step": 4756
    },
    {
      "epoch": 0.5505787037037037,
      "grad_norm": 0.057563796639442444,
      "learning_rate": 8.988425925925927e-06,
      "loss": 0.0107,
      "step": 4757
    },
    {
      "epoch": 0.5506944444444445,
      "grad_norm": 0.0823977068066597,
      "learning_rate": 8.986111111111113e-06,
      "loss": 0.011,
      "step": 4758
    },
    {
      "epoch": 0.5508101851851852,
      "grad_norm": 0.09747716039419174,
      "learning_rate": 8.983796296296297e-06,
      "loss": 0.0178,
      "step": 4759
    },
    {
      "epoch": 0.5509259259259259,
      "grad_norm": 0.06848183274269104,
      "learning_rate": 8.981481481481483e-06,
      "loss": 0.0114,
      "step": 4760
    },
    {
      "epoch": 0.5510416666666667,
      "grad_norm": 0.4499553442001343,
      "learning_rate": 8.979166666666667e-06,
      "loss": 0.0132,
      "step": 4761
    },
    {
      "epoch": 0.5511574074074074,
      "grad_norm": 0.07938354462385178,
      "learning_rate": 8.976851851851853e-06,
      "loss": 0.0105,
      "step": 4762
    },
    {
      "epoch": 0.5512731481481481,
      "grad_norm": 0.08838516473770142,
      "learning_rate": 8.974537037037037e-06,
      "loss": 0.0157,
      "step": 4763
    },
    {
      "epoch": 0.5513888888888889,
      "grad_norm": 0.08573738485574722,
      "learning_rate": 8.972222222222223e-06,
      "loss": 0.0157,
      "step": 4764
    },
    {
      "epoch": 0.5515046296296297,
      "grad_norm": 0.5978480577468872,
      "learning_rate": 8.969907407407407e-06,
      "loss": 0.0242,
      "step": 4765
    },
    {
      "epoch": 0.5516203703703704,
      "grad_norm": 0.3599144220352173,
      "learning_rate": 8.967592592592593e-06,
      "loss": 0.0191,
      "step": 4766
    },
    {
      "epoch": 0.5517361111111111,
      "grad_norm": 0.20088446140289307,
      "learning_rate": 8.96527777777778e-06,
      "loss": 0.0179,
      "step": 4767
    },
    {
      "epoch": 0.5518518518518518,
      "grad_norm": 0.1098313182592392,
      "learning_rate": 8.962962962962963e-06,
      "loss": 0.0171,
      "step": 4768
    },
    {
      "epoch": 0.5519675925925925,
      "grad_norm": 0.11607857048511505,
      "learning_rate": 8.96064814814815e-06,
      "loss": 0.0208,
      "step": 4769
    },
    {
      "epoch": 0.5520833333333334,
      "grad_norm": 0.0944160744547844,
      "learning_rate": 8.958333333333334e-06,
      "loss": 0.015,
      "step": 4770
    },
    {
      "epoch": 0.5521990740740741,
      "grad_norm": 6.104227542877197,
      "learning_rate": 8.95601851851852e-06,
      "loss": 0.0472,
      "step": 4771
    },
    {
      "epoch": 0.5523148148148148,
      "grad_norm": 0.19960731267929077,
      "learning_rate": 8.953703703703704e-06,
      "loss": 0.0205,
      "step": 4772
    },
    {
      "epoch": 0.5524305555555555,
      "grad_norm": 0.08745814859867096,
      "learning_rate": 8.95138888888889e-06,
      "loss": 0.0159,
      "step": 4773
    },
    {
      "epoch": 0.5525462962962963,
      "grad_norm": 0.08933119475841522,
      "learning_rate": 8.949074074074074e-06,
      "loss": 0.0157,
      "step": 4774
    },
    {
      "epoch": 0.5526620370370371,
      "grad_norm": 0.0803336501121521,
      "learning_rate": 8.94675925925926e-06,
      "loss": 0.0145,
      "step": 4775
    },
    {
      "epoch": 0.5527777777777778,
      "grad_norm": 0.08165906369686127,
      "learning_rate": 8.944444444444446e-06,
      "loss": 0.0145,
      "step": 4776
    },
    {
      "epoch": 0.5528935185185185,
      "grad_norm": 0.06494845449924469,
      "learning_rate": 8.94212962962963e-06,
      "loss": 0.0119,
      "step": 4777
    },
    {
      "epoch": 0.5530092592592593,
      "grad_norm": 17.368324279785156,
      "learning_rate": 8.939814814814816e-06,
      "loss": 2.5805,
      "step": 4778
    },
    {
      "epoch": 0.553125,
      "grad_norm": 0.11248812079429626,
      "learning_rate": 8.9375e-06,
      "loss": 0.014,
      "step": 4779
    },
    {
      "epoch": 0.5532407407407407,
      "grad_norm": 0.06335840374231339,
      "learning_rate": 8.935185185185186e-06,
      "loss": 0.0112,
      "step": 4780
    },
    {
      "epoch": 0.5533564814814815,
      "grad_norm": 0.07907983660697937,
      "learning_rate": 8.93287037037037e-06,
      "loss": 0.0119,
      "step": 4781
    },
    {
      "epoch": 0.5534722222222223,
      "grad_norm": 0.08286178112030029,
      "learning_rate": 8.930555555555557e-06,
      "loss": 0.0153,
      "step": 4782
    },
    {
      "epoch": 0.553587962962963,
      "grad_norm": 0.08250510692596436,
      "learning_rate": 8.92824074074074e-06,
      "loss": 0.0149,
      "step": 4783
    },
    {
      "epoch": 0.5537037037037037,
      "grad_norm": 0.137563094496727,
      "learning_rate": 8.925925925925927e-06,
      "loss": 0.0185,
      "step": 4784
    },
    {
      "epoch": 0.5538194444444444,
      "grad_norm": 0.1034124493598938,
      "learning_rate": 8.923611111111113e-06,
      "loss": 0.0192,
      "step": 4785
    },
    {
      "epoch": 0.5539351851851851,
      "grad_norm": 0.5238530039787292,
      "learning_rate": 8.921296296296297e-06,
      "loss": 0.0199,
      "step": 4786
    },
    {
      "epoch": 0.554050925925926,
      "grad_norm": 0.08399203419685364,
      "learning_rate": 8.918981481481483e-06,
      "loss": 0.0156,
      "step": 4787
    },
    {
      "epoch": 0.5541666666666667,
      "grad_norm": 0.07721545547246933,
      "learning_rate": 8.916666666666667e-06,
      "loss": 0.0139,
      "step": 4788
    },
    {
      "epoch": 0.5542824074074074,
      "grad_norm": 0.08104636520147324,
      "learning_rate": 8.914351851851853e-06,
      "loss": 0.014,
      "step": 4789
    },
    {
      "epoch": 0.5543981481481481,
      "grad_norm": 40.23684310913086,
      "learning_rate": 8.912037037037037e-06,
      "loss": 2.102,
      "step": 4790
    },
    {
      "epoch": 0.5545138888888889,
      "grad_norm": 0.15274494886398315,
      "learning_rate": 8.909722222222223e-06,
      "loss": 0.0187,
      "step": 4791
    },
    {
      "epoch": 0.5546296296296296,
      "grad_norm": 0.07860587537288666,
      "learning_rate": 8.907407407407408e-06,
      "loss": 0.0139,
      "step": 4792
    },
    {
      "epoch": 0.5547453703703704,
      "grad_norm": 0.10334304720163345,
      "learning_rate": 8.905092592592593e-06,
      "loss": 0.0183,
      "step": 4793
    },
    {
      "epoch": 0.5548611111111111,
      "grad_norm": 0.09065144509077072,
      "learning_rate": 8.902777777777778e-06,
      "loss": 0.0122,
      "step": 4794
    },
    {
      "epoch": 0.5549768518518519,
      "grad_norm": 0.07635296881198883,
      "learning_rate": 8.900462962962964e-06,
      "loss": 0.0137,
      "step": 4795
    },
    {
      "epoch": 0.5550925925925926,
      "grad_norm": 8.545589447021484,
      "learning_rate": 8.89814814814815e-06,
      "loss": 3.1036,
      "step": 4796
    },
    {
      "epoch": 0.5552083333333333,
      "grad_norm": 0.08365099132061005,
      "learning_rate": 8.895833333333334e-06,
      "loss": 0.0105,
      "step": 4797
    },
    {
      "epoch": 0.555324074074074,
      "grad_norm": 0.0753210112452507,
      "learning_rate": 8.89351851851852e-06,
      "loss": 0.0136,
      "step": 4798
    },
    {
      "epoch": 0.5554398148148149,
      "grad_norm": 0.10025232285261154,
      "learning_rate": 8.891203703703704e-06,
      "loss": 0.0155,
      "step": 4799
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 39.72980499267578,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.1353,
      "step": 4800
    },
    {
      "epoch": 0.5556712962962963,
      "grad_norm": 0.07592593133449554,
      "learning_rate": 8.886574074074074e-06,
      "loss": 0.0132,
      "step": 4801
    },
    {
      "epoch": 0.555787037037037,
      "grad_norm": 4.0670084953308105,
      "learning_rate": 8.88425925925926e-06,
      "loss": 0.0293,
      "step": 4802
    },
    {
      "epoch": 0.5559027777777777,
      "grad_norm": 0.07382185012102127,
      "learning_rate": 8.881944444444444e-06,
      "loss": 0.0121,
      "step": 4803
    },
    {
      "epoch": 0.5560185185185185,
      "grad_norm": 0.09296722710132599,
      "learning_rate": 8.87962962962963e-06,
      "loss": 0.0169,
      "step": 4804
    },
    {
      "epoch": 0.5561342592592593,
      "grad_norm": 0.12664690613746643,
      "learning_rate": 8.877314814814816e-06,
      "loss": 0.017,
      "step": 4805
    },
    {
      "epoch": 0.55625,
      "grad_norm": 0.12611818313598633,
      "learning_rate": 8.875e-06,
      "loss": 0.0169,
      "step": 4806
    },
    {
      "epoch": 0.5563657407407407,
      "grad_norm": 0.07932157069444656,
      "learning_rate": 8.872685185185187e-06,
      "loss": 0.0127,
      "step": 4807
    },
    {
      "epoch": 0.5564814814814815,
      "grad_norm": 0.07912632077932358,
      "learning_rate": 8.87037037037037e-06,
      "loss": 0.0143,
      "step": 4808
    },
    {
      "epoch": 0.5565972222222222,
      "grad_norm": 0.28803154826164246,
      "learning_rate": 8.868055555555555e-06,
      "loss": 0.0207,
      "step": 4809
    },
    {
      "epoch": 0.5567129629629629,
      "grad_norm": 16.986391067504883,
      "learning_rate": 8.865740740740741e-06,
      "loss": 2.7055,
      "step": 4810
    },
    {
      "epoch": 0.5568287037037037,
      "grad_norm": 0.09185856580734253,
      "learning_rate": 8.863425925925927e-06,
      "loss": 0.0172,
      "step": 4811
    },
    {
      "epoch": 0.5569444444444445,
      "grad_norm": 0.121390700340271,
      "learning_rate": 8.861111111111111e-06,
      "loss": 0.0208,
      "step": 4812
    },
    {
      "epoch": 0.5570601851851852,
      "grad_norm": 86.5928955078125,
      "learning_rate": 8.858796296296297e-06,
      "loss": 0.4477,
      "step": 4813
    },
    {
      "epoch": 0.5571759259259259,
      "grad_norm": 60.831703186035156,
      "learning_rate": 8.856481481481483e-06,
      "loss": 2.1252,
      "step": 4814
    },
    {
      "epoch": 0.5572916666666666,
      "grad_norm": 0.07797378301620483,
      "learning_rate": 8.854166666666667e-06,
      "loss": 0.014,
      "step": 4815
    },
    {
      "epoch": 0.5574074074074075,
      "grad_norm": 0.0914863869547844,
      "learning_rate": 8.851851851851853e-06,
      "loss": 0.016,
      "step": 4816
    },
    {
      "epoch": 0.5575231481481482,
      "grad_norm": 147.73727416992188,
      "learning_rate": 8.849537037037037e-06,
      "loss": 1.5814,
      "step": 4817
    },
    {
      "epoch": 0.5576388888888889,
      "grad_norm": 0.08813029527664185,
      "learning_rate": 8.847222222222223e-06,
      "loss": 0.0158,
      "step": 4818
    },
    {
      "epoch": 0.5577546296296296,
      "grad_norm": 0.11856541037559509,
      "learning_rate": 8.844907407407408e-06,
      "loss": 0.0154,
      "step": 4819
    },
    {
      "epoch": 0.5578703703703703,
      "grad_norm": 0.09363920986652374,
      "learning_rate": 8.842592592592594e-06,
      "loss": 0.017,
      "step": 4820
    },
    {
      "epoch": 0.5579861111111111,
      "grad_norm": 0.11769657582044601,
      "learning_rate": 8.840277777777778e-06,
      "loss": 0.0151,
      "step": 4821
    },
    {
      "epoch": 0.5581018518518519,
      "grad_norm": 0.08882275223731995,
      "learning_rate": 8.837962962962964e-06,
      "loss": 0.0165,
      "step": 4822
    },
    {
      "epoch": 0.5582175925925926,
      "grad_norm": 0.09593472629785538,
      "learning_rate": 8.83564814814815e-06,
      "loss": 0.0174,
      "step": 4823
    },
    {
      "epoch": 0.5583333333333333,
      "grad_norm": 0.07640406489372253,
      "learning_rate": 8.833333333333334e-06,
      "loss": 0.0136,
      "step": 4824
    },
    {
      "epoch": 0.5584490740740741,
      "grad_norm": 0.0911150574684143,
      "learning_rate": 8.83101851851852e-06,
      "loss": 0.0163,
      "step": 4825
    },
    {
      "epoch": 0.5585648148148148,
      "grad_norm": 0.07345591485500336,
      "learning_rate": 8.828703703703704e-06,
      "loss": 0.0098,
      "step": 4826
    },
    {
      "epoch": 0.5586805555555555,
      "grad_norm": 0.07967626303434372,
      "learning_rate": 8.82638888888889e-06,
      "loss": 0.0142,
      "step": 4827
    },
    {
      "epoch": 0.5587962962962963,
      "grad_norm": 0.15067052841186523,
      "learning_rate": 8.824074074074074e-06,
      "loss": 0.0173,
      "step": 4828
    },
    {
      "epoch": 0.5589120370370371,
      "grad_norm": 0.11447929590940475,
      "learning_rate": 8.82175925925926e-06,
      "loss": 0.0182,
      "step": 4829
    },
    {
      "epoch": 0.5590277777777778,
      "grad_norm": 15.328670501708984,
      "learning_rate": 8.819444444444445e-06,
      "loss": 2.7917,
      "step": 4830
    },
    {
      "epoch": 0.5591435185185185,
      "grad_norm": 0.07807744294404984,
      "learning_rate": 8.81712962962963e-06,
      "loss": 0.0128,
      "step": 4831
    },
    {
      "epoch": 0.5592592592592592,
      "grad_norm": 0.08349834382534027,
      "learning_rate": 8.814814814814817e-06,
      "loss": 0.0151,
      "step": 4832
    },
    {
      "epoch": 0.559375,
      "grad_norm": 0.11049696803092957,
      "learning_rate": 8.8125e-06,
      "loss": 0.0166,
      "step": 4833
    },
    {
      "epoch": 0.5594907407407408,
      "grad_norm": 0.11725234240293503,
      "learning_rate": 8.810185185185187e-06,
      "loss": 0.0207,
      "step": 4834
    },
    {
      "epoch": 0.5596064814814815,
      "grad_norm": 0.30797863006591797,
      "learning_rate": 8.807870370370371e-06,
      "loss": 0.0179,
      "step": 4835
    },
    {
      "epoch": 0.5597222222222222,
      "grad_norm": 0.06375336647033691,
      "learning_rate": 8.805555555555557e-06,
      "loss": 0.0115,
      "step": 4836
    },
    {
      "epoch": 0.559837962962963,
      "grad_norm": 0.08692940324544907,
      "learning_rate": 8.803240740740741e-06,
      "loss": 0.0163,
      "step": 4837
    },
    {
      "epoch": 0.5599537037037037,
      "grad_norm": 0.08067513257265091,
      "learning_rate": 8.800925925925925e-06,
      "loss": 0.0146,
      "step": 4838
    },
    {
      "epoch": 0.5600694444444444,
      "grad_norm": 7.159493923187256,
      "learning_rate": 8.798611111111111e-06,
      "loss": 0.0536,
      "step": 4839
    },
    {
      "epoch": 0.5601851851851852,
      "grad_norm": 0.06497766822576523,
      "learning_rate": 8.796296296296297e-06,
      "loss": 0.0118,
      "step": 4840
    },
    {
      "epoch": 0.560300925925926,
      "grad_norm": 0.06405387818813324,
      "learning_rate": 8.793981481481482e-06,
      "loss": 0.0117,
      "step": 4841
    },
    {
      "epoch": 0.5604166666666667,
      "grad_norm": 0.08135343343019485,
      "learning_rate": 8.791666666666667e-06,
      "loss": 0.0145,
      "step": 4842
    },
    {
      "epoch": 0.5605324074074074,
      "grad_norm": 0.08220788091421127,
      "learning_rate": 8.789351851851853e-06,
      "loss": 0.015,
      "step": 4843
    },
    {
      "epoch": 0.5606481481481481,
      "grad_norm": 0.05676286667585373,
      "learning_rate": 8.787037037037038e-06,
      "loss": 0.0104,
      "step": 4844
    },
    {
      "epoch": 0.5607638888888888,
      "grad_norm": 0.08026568591594696,
      "learning_rate": 8.784722222222224e-06,
      "loss": 0.0141,
      "step": 4845
    },
    {
      "epoch": 0.5608796296296297,
      "grad_norm": 0.06514249742031097,
      "learning_rate": 8.782407407407408e-06,
      "loss": 0.012,
      "step": 4846
    },
    {
      "epoch": 0.5609953703703704,
      "grad_norm": 0.3854827880859375,
      "learning_rate": 8.780092592592592e-06,
      "loss": 0.0213,
      "step": 4847
    },
    {
      "epoch": 0.5611111111111111,
      "grad_norm": 0.09673156589269638,
      "learning_rate": 8.777777777777778e-06,
      "loss": 0.0173,
      "step": 4848
    },
    {
      "epoch": 0.5612268518518518,
      "grad_norm": 0.18235325813293457,
      "learning_rate": 8.775462962962964e-06,
      "loss": 0.0169,
      "step": 4849
    },
    {
      "epoch": 0.5613425925925926,
      "grad_norm": 0.05517248809337616,
      "learning_rate": 8.773148148148148e-06,
      "loss": 0.0102,
      "step": 4850
    },
    {
      "epoch": 0.5614583333333333,
      "grad_norm": 0.0897645577788353,
      "learning_rate": 8.770833333333334e-06,
      "loss": 0.0127,
      "step": 4851
    },
    {
      "epoch": 0.5615740740740741,
      "grad_norm": 0.0747542604804039,
      "learning_rate": 8.76851851851852e-06,
      "loss": 0.0135,
      "step": 4852
    },
    {
      "epoch": 0.5616898148148148,
      "grad_norm": 94.79509735107422,
      "learning_rate": 8.766203703703704e-06,
      "loss": 1.1188,
      "step": 4853
    },
    {
      "epoch": 0.5618055555555556,
      "grad_norm": 0.07486704736948013,
      "learning_rate": 8.76388888888889e-06,
      "loss": 0.0124,
      "step": 4854
    },
    {
      "epoch": 0.5619212962962963,
      "grad_norm": 4.706939697265625,
      "learning_rate": 8.761574074074075e-06,
      "loss": 0.035,
      "step": 4855
    },
    {
      "epoch": 0.562037037037037,
      "grad_norm": 0.1451958417892456,
      "learning_rate": 8.759259259259259e-06,
      "loss": 0.012,
      "step": 4856
    },
    {
      "epoch": 0.5621527777777777,
      "grad_norm": 0.10441131144762039,
      "learning_rate": 8.756944444444445e-06,
      "loss": 0.0189,
      "step": 4857
    },
    {
      "epoch": 0.5622685185185186,
      "grad_norm": 2.075225353240967,
      "learning_rate": 8.75462962962963e-06,
      "loss": 0.0188,
      "step": 4858
    },
    {
      "epoch": 0.5623842592592593,
      "grad_norm": 0.06956347823143005,
      "learning_rate": 8.752314814814815e-06,
      "loss": 0.0124,
      "step": 4859
    },
    {
      "epoch": 0.5625,
      "grad_norm": 0.12542180716991425,
      "learning_rate": 8.750000000000001e-06,
      "loss": 0.0164,
      "step": 4860
    },
    {
      "epoch": 0.5626157407407407,
      "grad_norm": 0.0815785825252533,
      "learning_rate": 8.747685185185187e-06,
      "loss": 0.0149,
      "step": 4861
    },
    {
      "epoch": 0.5627314814814814,
      "grad_norm": 0.08417795598506927,
      "learning_rate": 8.745370370370371e-06,
      "loss": 0.015,
      "step": 4862
    },
    {
      "epoch": 0.5628472222222223,
      "grad_norm": 16.73048210144043,
      "learning_rate": 8.743055555555557e-06,
      "loss": 2.7459,
      "step": 4863
    },
    {
      "epoch": 0.562962962962963,
      "grad_norm": 0.08543401956558228,
      "learning_rate": 8.740740740740741e-06,
      "loss": 0.0152,
      "step": 4864
    },
    {
      "epoch": 0.5630787037037037,
      "grad_norm": 0.08449995517730713,
      "learning_rate": 8.738425925925926e-06,
      "loss": 0.0151,
      "step": 4865
    },
    {
      "epoch": 0.5631944444444444,
      "grad_norm": 0.08476857095956802,
      "learning_rate": 8.736111111111112e-06,
      "loss": 0.0148,
      "step": 4866
    },
    {
      "epoch": 0.5633101851851852,
      "grad_norm": 0.07574005424976349,
      "learning_rate": 8.733796296296297e-06,
      "loss": 0.0136,
      "step": 4867
    },
    {
      "epoch": 0.5634259259259259,
      "grad_norm": 0.10986043512821198,
      "learning_rate": 8.731481481481482e-06,
      "loss": 0.0202,
      "step": 4868
    },
    {
      "epoch": 0.5635416666666667,
      "grad_norm": 0.10116400569677353,
      "learning_rate": 8.729166666666668e-06,
      "loss": 0.0151,
      "step": 4869
    },
    {
      "epoch": 0.5636574074074074,
      "grad_norm": 0.09568554908037186,
      "learning_rate": 8.726851851851854e-06,
      "loss": 0.0176,
      "step": 4870
    },
    {
      "epoch": 0.5637731481481482,
      "grad_norm": 0.082638680934906,
      "learning_rate": 8.724537037037038e-06,
      "loss": 0.0153,
      "step": 4871
    },
    {
      "epoch": 0.5638888888888889,
      "grad_norm": 0.101453498005867,
      "learning_rate": 8.722222222222224e-06,
      "loss": 0.0187,
      "step": 4872
    },
    {
      "epoch": 0.5640046296296296,
      "grad_norm": 0.0722765326499939,
      "learning_rate": 8.719907407407408e-06,
      "loss": 0.013,
      "step": 4873
    },
    {
      "epoch": 0.5641203703703703,
      "grad_norm": 0.09054839611053467,
      "learning_rate": 8.717592592592592e-06,
      "loss": 0.0168,
      "step": 4874
    },
    {
      "epoch": 0.5642361111111112,
      "grad_norm": 0.08539918065071106,
      "learning_rate": 8.715277777777778e-06,
      "loss": 0.0155,
      "step": 4875
    },
    {
      "epoch": 0.5643518518518519,
      "grad_norm": 0.06750104576349258,
      "learning_rate": 8.712962962962964e-06,
      "loss": 0.0126,
      "step": 4876
    },
    {
      "epoch": 0.5644675925925926,
      "grad_norm": 0.10397310554981232,
      "learning_rate": 8.710648148148148e-06,
      "loss": 0.0166,
      "step": 4877
    },
    {
      "epoch": 0.5645833333333333,
      "grad_norm": 1.469758152961731,
      "learning_rate": 8.708333333333334e-06,
      "loss": 0.0238,
      "step": 4878
    },
    {
      "epoch": 0.564699074074074,
      "grad_norm": 0.3066892623901367,
      "learning_rate": 8.70601851851852e-06,
      "loss": 0.0173,
      "step": 4879
    },
    {
      "epoch": 0.5648148148148148,
      "grad_norm": 0.10266993939876556,
      "learning_rate": 8.703703703703705e-06,
      "loss": 0.0182,
      "step": 4880
    },
    {
      "epoch": 0.5649305555555556,
      "grad_norm": 0.06054600700736046,
      "learning_rate": 8.70138888888889e-06,
      "loss": 0.0112,
      "step": 4881
    },
    {
      "epoch": 0.5650462962962963,
      "grad_norm": 0.11685722321271896,
      "learning_rate": 8.699074074074075e-06,
      "loss": 0.0157,
      "step": 4882
    },
    {
      "epoch": 0.565162037037037,
      "grad_norm": 0.07932154089212418,
      "learning_rate": 8.696759259259259e-06,
      "loss": 0.0105,
      "step": 4883
    },
    {
      "epoch": 0.5652777777777778,
      "grad_norm": 0.11274220049381256,
      "learning_rate": 8.694444444444445e-06,
      "loss": 0.0148,
      "step": 4884
    },
    {
      "epoch": 0.5653935185185185,
      "grad_norm": 2.2976009845733643,
      "learning_rate": 8.69212962962963e-06,
      "loss": 0.0308,
      "step": 4885
    },
    {
      "epoch": 0.5655092592592592,
      "grad_norm": 0.09988009929656982,
      "learning_rate": 8.689814814814815e-06,
      "loss": 0.0183,
      "step": 4886
    },
    {
      "epoch": 0.565625,
      "grad_norm": 0.06640974432229996,
      "learning_rate": 8.687500000000001e-06,
      "loss": 0.0109,
      "step": 4887
    },
    {
      "epoch": 0.5657407407407408,
      "grad_norm": 0.08346409350633621,
      "learning_rate": 8.685185185185185e-06,
      "loss": 0.0152,
      "step": 4888
    },
    {
      "epoch": 0.5658564814814815,
      "grad_norm": 0.11609584093093872,
      "learning_rate": 8.682870370370371e-06,
      "loss": 0.0157,
      "step": 4889
    },
    {
      "epoch": 0.5659722222222222,
      "grad_norm": 6.614137649536133,
      "learning_rate": 8.680555555555557e-06,
      "loss": 3.4006,
      "step": 4890
    },
    {
      "epoch": 0.5660879629629629,
      "grad_norm": 167.35342407226562,
      "learning_rate": 8.678240740740741e-06,
      "loss": 0.7269,
      "step": 4891
    },
    {
      "epoch": 0.5662037037037037,
      "grad_norm": 0.08226685225963593,
      "learning_rate": 8.675925925925926e-06,
      "loss": 0.015,
      "step": 4892
    },
    {
      "epoch": 0.5663194444444445,
      "grad_norm": 3.953587293624878,
      "learning_rate": 8.673611111111112e-06,
      "loss": 0.0383,
      "step": 4893
    },
    {
      "epoch": 0.5664351851851852,
      "grad_norm": 9.6538667678833,
      "learning_rate": 8.671296296296296e-06,
      "loss": 2.9534,
      "step": 4894
    },
    {
      "epoch": 0.5665509259259259,
      "grad_norm": 0.3385722041130066,
      "learning_rate": 8.668981481481482e-06,
      "loss": 0.0183,
      "step": 4895
    },
    {
      "epoch": 0.5666666666666667,
      "grad_norm": 0.08411044627428055,
      "learning_rate": 8.666666666666668e-06,
      "loss": 0.0115,
      "step": 4896
    },
    {
      "epoch": 0.5667824074074074,
      "grad_norm": 0.07902631908655167,
      "learning_rate": 8.664351851851852e-06,
      "loss": 0.0144,
      "step": 4897
    },
    {
      "epoch": 0.5668981481481481,
      "grad_norm": 3.0625205039978027,
      "learning_rate": 8.662037037037038e-06,
      "loss": 0.0304,
      "step": 4898
    },
    {
      "epoch": 0.5670138888888889,
      "grad_norm": 0.0702398419380188,
      "learning_rate": 8.659722222222224e-06,
      "loss": 0.0129,
      "step": 4899
    },
    {
      "epoch": 0.5671296296296297,
      "grad_norm": 10.44446849822998,
      "learning_rate": 8.657407407407408e-06,
      "loss": 2.9346,
      "step": 4900
    },
    {
      "epoch": 0.5672453703703704,
      "grad_norm": 0.11522343009710312,
      "learning_rate": 8.655092592592592e-06,
      "loss": 0.0152,
      "step": 4901
    },
    {
      "epoch": 0.5673611111111111,
      "grad_norm": 0.07786451280117035,
      "learning_rate": 8.652777777777778e-06,
      "loss": 0.0142,
      "step": 4902
    },
    {
      "epoch": 0.5674768518518518,
      "grad_norm": 0.07816355675458908,
      "learning_rate": 8.650462962962963e-06,
      "loss": 0.0104,
      "step": 4903
    },
    {
      "epoch": 0.5675925925925925,
      "grad_norm": 0.07902564108371735,
      "learning_rate": 8.648148148148149e-06,
      "loss": 0.0142,
      "step": 4904
    },
    {
      "epoch": 0.5677083333333334,
      "grad_norm": 0.08241133391857147,
      "learning_rate": 8.645833333333335e-06,
      "loss": 0.0146,
      "step": 4905
    },
    {
      "epoch": 0.5678240740740741,
      "grad_norm": 0.14965404570102692,
      "learning_rate": 8.643518518518519e-06,
      "loss": 0.0185,
      "step": 4906
    },
    {
      "epoch": 0.5679398148148148,
      "grad_norm": 0.2802022397518158,
      "learning_rate": 8.641203703703705e-06,
      "loss": 0.0169,
      "step": 4907
    },
    {
      "epoch": 0.5680555555555555,
      "grad_norm": 0.08826786279678345,
      "learning_rate": 8.63888888888889e-06,
      "loss": 0.0163,
      "step": 4908
    },
    {
      "epoch": 0.5681712962962963,
      "grad_norm": 0.07068847119808197,
      "learning_rate": 8.636574074074075e-06,
      "loss": 0.0129,
      "step": 4909
    },
    {
      "epoch": 0.5682870370370371,
      "grad_norm": 0.1329175978899002,
      "learning_rate": 8.63425925925926e-06,
      "loss": 0.0178,
      "step": 4910
    },
    {
      "epoch": 0.5684027777777778,
      "grad_norm": 0.07963795959949493,
      "learning_rate": 8.631944444444445e-06,
      "loss": 0.0144,
      "step": 4911
    },
    {
      "epoch": 0.5685185185185185,
      "grad_norm": 0.06704944372177124,
      "learning_rate": 8.62962962962963e-06,
      "loss": 0.0121,
      "step": 4912
    },
    {
      "epoch": 0.5686342592592593,
      "grad_norm": 0.6560481190681458,
      "learning_rate": 8.627314814814815e-06,
      "loss": 0.0229,
      "step": 4913
    },
    {
      "epoch": 0.56875,
      "grad_norm": 0.10274359583854675,
      "learning_rate": 8.625000000000001e-06,
      "loss": 0.0183,
      "step": 4914
    },
    {
      "epoch": 0.5688657407407407,
      "grad_norm": 0.0848974734544754,
      "learning_rate": 8.622685185185186e-06,
      "loss": 0.0147,
      "step": 4915
    },
    {
      "epoch": 0.5689814814814815,
      "grad_norm": 1.184715747833252,
      "learning_rate": 8.620370370370371e-06,
      "loss": 0.0231,
      "step": 4916
    },
    {
      "epoch": 0.5690972222222223,
      "grad_norm": 0.8812862634658813,
      "learning_rate": 8.618055555555557e-06,
      "loss": 0.0218,
      "step": 4917
    },
    {
      "epoch": 0.569212962962963,
      "grad_norm": 0.23344558477401733,
      "learning_rate": 8.615740740740742e-06,
      "loss": 0.0166,
      "step": 4918
    },
    {
      "epoch": 0.5693287037037037,
      "grad_norm": 0.08035358041524887,
      "learning_rate": 8.613425925925926e-06,
      "loss": 0.0143,
      "step": 4919
    },
    {
      "epoch": 0.5694444444444444,
      "grad_norm": 99.92950439453125,
      "learning_rate": 8.611111111111112e-06,
      "loss": 0.8297,
      "step": 4920
    },
    {
      "epoch": 0.5695601851851851,
      "grad_norm": 0.08808410167694092,
      "learning_rate": 8.608796296296296e-06,
      "loss": 0.0143,
      "step": 4921
    },
    {
      "epoch": 0.569675925925926,
      "grad_norm": 0.07917029410600662,
      "learning_rate": 8.606481481481482e-06,
      "loss": 0.0143,
      "step": 4922
    },
    {
      "epoch": 0.5697916666666667,
      "grad_norm": 0.0779520720243454,
      "learning_rate": 8.604166666666668e-06,
      "loss": 0.0142,
      "step": 4923
    },
    {
      "epoch": 0.5699074074074074,
      "grad_norm": 0.2023729532957077,
      "learning_rate": 8.601851851851852e-06,
      "loss": 0.0156,
      "step": 4924
    },
    {
      "epoch": 0.5700231481481481,
      "grad_norm": 0.09055504202842712,
      "learning_rate": 8.599537037037038e-06,
      "loss": 0.0158,
      "step": 4925
    },
    {
      "epoch": 0.5701388888888889,
      "grad_norm": 0.0766233578324318,
      "learning_rate": 8.597222222222224e-06,
      "loss": 0.0139,
      "step": 4926
    },
    {
      "epoch": 0.5702546296296296,
      "grad_norm": 0.1019303947687149,
      "learning_rate": 8.594907407407408e-06,
      "loss": 0.0188,
      "step": 4927
    },
    {
      "epoch": 0.5703703703703704,
      "grad_norm": 11.977898597717285,
      "learning_rate": 8.592592592592593e-06,
      "loss": 0.1047,
      "step": 4928
    },
    {
      "epoch": 0.5704861111111111,
      "grad_norm": 0.08139326423406601,
      "learning_rate": 8.590277777777779e-06,
      "loss": 0.0142,
      "step": 4929
    },
    {
      "epoch": 0.5706018518518519,
      "grad_norm": 0.21891725063323975,
      "learning_rate": 8.587962962962963e-06,
      "loss": 0.0185,
      "step": 4930
    },
    {
      "epoch": 0.5707175925925926,
      "grad_norm": 0.0855645090341568,
      "learning_rate": 8.585648148148149e-06,
      "loss": 0.0153,
      "step": 4931
    },
    {
      "epoch": 0.5708333333333333,
      "grad_norm": 0.08185861259698868,
      "learning_rate": 8.583333333333333e-06,
      "loss": 0.0146,
      "step": 4932
    },
    {
      "epoch": 0.570949074074074,
      "grad_norm": 29.061769485473633,
      "learning_rate": 8.581018518518519e-06,
      "loss": 0.1334,
      "step": 4933
    },
    {
      "epoch": 0.5710648148148149,
      "grad_norm": 0.09262178838253021,
      "learning_rate": 8.578703703703705e-06,
      "loss": 0.0142,
      "step": 4934
    },
    {
      "epoch": 0.5711805555555556,
      "grad_norm": 0.3872152864933014,
      "learning_rate": 8.57638888888889e-06,
      "loss": 0.0193,
      "step": 4935
    },
    {
      "epoch": 0.5712962962962963,
      "grad_norm": 0.1039581298828125,
      "learning_rate": 8.574074074074075e-06,
      "loss": 0.0127,
      "step": 4936
    },
    {
      "epoch": 0.571412037037037,
      "grad_norm": 0.10176031291484833,
      "learning_rate": 8.571759259259261e-06,
      "loss": 0.0179,
      "step": 4937
    },
    {
      "epoch": 0.5715277777777777,
      "grad_norm": 0.0673348680138588,
      "learning_rate": 8.569444444444445e-06,
      "loss": 0.0119,
      "step": 4938
    },
    {
      "epoch": 0.5716435185185185,
      "grad_norm": 0.2852769196033478,
      "learning_rate": 8.56712962962963e-06,
      "loss": 0.0179,
      "step": 4939
    },
    {
      "epoch": 0.5717592592592593,
      "grad_norm": 0.0792679414153099,
      "learning_rate": 8.564814814814816e-06,
      "loss": 0.0141,
      "step": 4940
    },
    {
      "epoch": 0.571875,
      "grad_norm": 0.06413140147924423,
      "learning_rate": 8.5625e-06,
      "loss": 0.0116,
      "step": 4941
    },
    {
      "epoch": 0.5719907407407407,
      "grad_norm": 0.09810053557157516,
      "learning_rate": 8.560185185185186e-06,
      "loss": 0.0142,
      "step": 4942
    },
    {
      "epoch": 0.5721064814814815,
      "grad_norm": 0.06234261021018028,
      "learning_rate": 8.557870370370372e-06,
      "loss": 0.0115,
      "step": 4943
    },
    {
      "epoch": 0.5722222222222222,
      "grad_norm": 0.07572247833013535,
      "learning_rate": 8.555555555555556e-06,
      "loss": 0.0138,
      "step": 4944
    },
    {
      "epoch": 0.5723379629629629,
      "grad_norm": 172.27723693847656,
      "learning_rate": 8.553240740740742e-06,
      "loss": 1.7418,
      "step": 4945
    },
    {
      "epoch": 0.5724537037037037,
      "grad_norm": 0.10409975051879883,
      "learning_rate": 8.550925925925928e-06,
      "loss": 0.0126,
      "step": 4946
    },
    {
      "epoch": 0.5725694444444445,
      "grad_norm": 122.8727798461914,
      "learning_rate": 8.548611111111112e-06,
      "loss": 2.3224,
      "step": 4947
    },
    {
      "epoch": 0.5726851851851852,
      "grad_norm": 0.0792035460472107,
      "learning_rate": 8.546296296296296e-06,
      "loss": 0.0142,
      "step": 4948
    },
    {
      "epoch": 0.5728009259259259,
      "grad_norm": 0.15762074291706085,
      "learning_rate": 8.543981481481482e-06,
      "loss": 0.0157,
      "step": 4949
    },
    {
      "epoch": 0.5729166666666666,
      "grad_norm": 0.07205988466739655,
      "learning_rate": 8.541666666666666e-06,
      "loss": 0.0125,
      "step": 4950
    },
    {
      "epoch": 0.5730324074074075,
      "grad_norm": 60.008689880371094,
      "learning_rate": 8.539351851851852e-06,
      "loss": 2.1622,
      "step": 4951
    },
    {
      "epoch": 0.5731481481481482,
      "grad_norm": 76.85862731933594,
      "learning_rate": 8.537037037037038e-06,
      "loss": 0.2639,
      "step": 4952
    },
    {
      "epoch": 0.5732638888888889,
      "grad_norm": 0.2557343542575836,
      "learning_rate": 8.534722222222223e-06,
      "loss": 0.0201,
      "step": 4953
    },
    {
      "epoch": 0.5733796296296296,
      "grad_norm": 0.05890050157904625,
      "learning_rate": 8.532407407407409e-06,
      "loss": 0.0106,
      "step": 4954
    },
    {
      "epoch": 0.5734953703703703,
      "grad_norm": 12.716938018798828,
      "learning_rate": 8.530092592592595e-06,
      "loss": 0.0678,
      "step": 4955
    },
    {
      "epoch": 0.5736111111111111,
      "grad_norm": 0.05795247107744217,
      "learning_rate": 8.527777777777779e-06,
      "loss": 0.0105,
      "step": 4956
    },
    {
      "epoch": 0.5737268518518519,
      "grad_norm": 0.08994340896606445,
      "learning_rate": 8.525462962962963e-06,
      "loss": 0.0167,
      "step": 4957
    },
    {
      "epoch": 0.5738425925925926,
      "grad_norm": 0.09332340955734253,
      "learning_rate": 8.523148148148149e-06,
      "loss": 0.0157,
      "step": 4958
    },
    {
      "epoch": 0.5739583333333333,
      "grad_norm": 0.08026787638664246,
      "learning_rate": 8.520833333333333e-06,
      "loss": 0.0141,
      "step": 4959
    },
    {
      "epoch": 0.5740740740740741,
      "grad_norm": 0.05655434727668762,
      "learning_rate": 8.518518518518519e-06,
      "loss": 0.0104,
      "step": 4960
    },
    {
      "epoch": 0.5741898148148148,
      "grad_norm": 0.10133924335241318,
      "learning_rate": 8.516203703703705e-06,
      "loss": 0.0164,
      "step": 4961
    },
    {
      "epoch": 0.5743055555555555,
      "grad_norm": 0.09477009624242783,
      "learning_rate": 8.51388888888889e-06,
      "loss": 0.0176,
      "step": 4962
    },
    {
      "epoch": 0.5744212962962963,
      "grad_norm": 0.07721863687038422,
      "learning_rate": 8.511574074074075e-06,
      "loss": 0.0103,
      "step": 4963
    },
    {
      "epoch": 0.5745370370370371,
      "grad_norm": 0.06770908087491989,
      "learning_rate": 8.509259259259261e-06,
      "loss": 0.012,
      "step": 4964
    },
    {
      "epoch": 0.5746527777777778,
      "grad_norm": 0.19537875056266785,
      "learning_rate": 8.506944444444445e-06,
      "loss": 0.015,
      "step": 4965
    },
    {
      "epoch": 0.5747685185185185,
      "grad_norm": 0.0673782005906105,
      "learning_rate": 8.50462962962963e-06,
      "loss": 0.0117,
      "step": 4966
    },
    {
      "epoch": 0.5748842592592592,
      "grad_norm": 0.0822969600558281,
      "learning_rate": 8.502314814814816e-06,
      "loss": 0.0123,
      "step": 4967
    },
    {
      "epoch": 0.575,
      "grad_norm": 0.11662005633115768,
      "learning_rate": 8.5e-06,
      "loss": 0.0196,
      "step": 4968
    },
    {
      "epoch": 0.5751157407407408,
      "grad_norm": 0.07735206931829453,
      "learning_rate": 8.497685185185186e-06,
      "loss": 0.0133,
      "step": 4969
    },
    {
      "epoch": 0.5752314814814815,
      "grad_norm": 0.36088457703590393,
      "learning_rate": 8.495370370370372e-06,
      "loss": 0.0195,
      "step": 4970
    },
    {
      "epoch": 0.5753472222222222,
      "grad_norm": 0.08946128189563751,
      "learning_rate": 8.493055555555556e-06,
      "loss": 0.0152,
      "step": 4971
    },
    {
      "epoch": 0.575462962962963,
      "grad_norm": 0.056870121508836746,
      "learning_rate": 8.490740740740742e-06,
      "loss": 0.0104,
      "step": 4972
    },
    {
      "epoch": 0.5755787037037037,
      "grad_norm": 0.0557376928627491,
      "learning_rate": 8.488425925925926e-06,
      "loss": 0.0102,
      "step": 4973
    },
    {
      "epoch": 0.5756944444444444,
      "grad_norm": 0.09133756905794144,
      "learning_rate": 8.486111111111112e-06,
      "loss": 0.0168,
      "step": 4974
    },
    {
      "epoch": 0.5758101851851852,
      "grad_norm": 0.07740657776594162,
      "learning_rate": 8.483796296296296e-06,
      "loss": 0.0139,
      "step": 4975
    },
    {
      "epoch": 0.575925925925926,
      "grad_norm": 3.1598639488220215,
      "learning_rate": 8.481481481481482e-06,
      "loss": 0.0347,
      "step": 4976
    },
    {
      "epoch": 0.5760416666666667,
      "grad_norm": 0.06287162750959396,
      "learning_rate": 8.479166666666667e-06,
      "loss": 0.0115,
      "step": 4977
    },
    {
      "epoch": 0.5761574074074074,
      "grad_norm": 0.05854145810008049,
      "learning_rate": 8.476851851851853e-06,
      "loss": 0.0104,
      "step": 4978
    },
    {
      "epoch": 0.5762731481481481,
      "grad_norm": 0.094631627202034,
      "learning_rate": 8.474537037037037e-06,
      "loss": 0.0127,
      "step": 4979
    },
    {
      "epoch": 0.5763888888888888,
      "grad_norm": 0.06732577085494995,
      "learning_rate": 8.472222222222223e-06,
      "loss": 0.0117,
      "step": 4980
    },
    {
      "epoch": 0.5765046296296297,
      "grad_norm": 0.08560580760240555,
      "learning_rate": 8.469907407407409e-06,
      "loss": 0.015,
      "step": 4981
    },
    {
      "epoch": 0.5766203703703704,
      "grad_norm": 0.09778212010860443,
      "learning_rate": 8.467592592592593e-06,
      "loss": 0.0115,
      "step": 4982
    },
    {
      "epoch": 0.5767361111111111,
      "grad_norm": 0.09840280562639236,
      "learning_rate": 8.465277777777779e-06,
      "loss": 0.0156,
      "step": 4983
    },
    {
      "epoch": 0.5768518518518518,
      "grad_norm": 72.3005599975586,
      "learning_rate": 8.462962962962963e-06,
      "loss": 1.7894,
      "step": 4984
    },
    {
      "epoch": 0.5769675925925926,
      "grad_norm": 0.10319672524929047,
      "learning_rate": 8.460648148148149e-06,
      "loss": 0.0188,
      "step": 4985
    },
    {
      "epoch": 0.5770833333333333,
      "grad_norm": 194.99557495117188,
      "learning_rate": 8.458333333333333e-06,
      "loss": 1.3282,
      "step": 4986
    },
    {
      "epoch": 0.5771990740740741,
      "grad_norm": 0.11400160938501358,
      "learning_rate": 8.45601851851852e-06,
      "loss": 0.0169,
      "step": 4987
    },
    {
      "epoch": 0.5773148148148148,
      "grad_norm": 0.08149215579032898,
      "learning_rate": 8.453703703703704e-06,
      "loss": 0.0151,
      "step": 4988
    },
    {
      "epoch": 0.5774305555555556,
      "grad_norm": 78.12894439697266,
      "learning_rate": 8.45138888888889e-06,
      "loss": 2.0978,
      "step": 4989
    },
    {
      "epoch": 0.5775462962962963,
      "grad_norm": 0.054454505443573,
      "learning_rate": 8.449074074074075e-06,
      "loss": 0.0101,
      "step": 4990
    },
    {
      "epoch": 0.577662037037037,
      "grad_norm": 0.06009528040885925,
      "learning_rate": 8.44675925925926e-06,
      "loss": 0.011,
      "step": 4991
    },
    {
      "epoch": 0.5777777777777777,
      "grad_norm": 0.1031046062707901,
      "learning_rate": 8.444444444444446e-06,
      "loss": 0.017,
      "step": 4992
    },
    {
      "epoch": 0.5778935185185186,
      "grad_norm": 0.09009423851966858,
      "learning_rate": 8.44212962962963e-06,
      "loss": 0.0156,
      "step": 4993
    },
    {
      "epoch": 0.5780092592592593,
      "grad_norm": 18.669509887695312,
      "learning_rate": 8.439814814814816e-06,
      "loss": 0.0688,
      "step": 4994
    },
    {
      "epoch": 0.578125,
      "grad_norm": 0.07346255332231522,
      "learning_rate": 8.4375e-06,
      "loss": 0.0132,
      "step": 4995
    },
    {
      "epoch": 0.5782407407407407,
      "grad_norm": 0.8483057022094727,
      "learning_rate": 8.435185185185186e-06,
      "loss": 0.0196,
      "step": 4996
    },
    {
      "epoch": 0.5783564814814814,
      "grad_norm": 0.08004911243915558,
      "learning_rate": 8.43287037037037e-06,
      "loss": 0.0135,
      "step": 4997
    },
    {
      "epoch": 0.5784722222222223,
      "grad_norm": 0.078464575111866,
      "learning_rate": 8.430555555555556e-06,
      "loss": 0.0139,
      "step": 4998
    },
    {
      "epoch": 0.578587962962963,
      "grad_norm": 0.9385637044906616,
      "learning_rate": 8.428240740740742e-06,
      "loss": 0.0163,
      "step": 4999
    },
    {
      "epoch": 0.5787037037037037,
      "grad_norm": 0.08942580223083496,
      "learning_rate": 8.425925925925926e-06,
      "loss": 0.0159,
      "step": 5000
    },
    {
      "epoch": 0.5788194444444444,
      "grad_norm": 0.8000161051750183,
      "learning_rate": 8.423611111111112e-06,
      "loss": 0.023,
      "step": 5001
    },
    {
      "epoch": 0.5789351851851852,
      "grad_norm": 0.09427862614393234,
      "learning_rate": 8.421296296296297e-06,
      "loss": 0.0126,
      "step": 5002
    },
    {
      "epoch": 0.5790509259259259,
      "grad_norm": 0.08962629735469818,
      "learning_rate": 8.418981481481483e-06,
      "loss": 0.0165,
      "step": 5003
    },
    {
      "epoch": 0.5791666666666667,
      "grad_norm": 0.20267577469348907,
      "learning_rate": 8.416666666666667e-06,
      "loss": 0.0149,
      "step": 5004
    },
    {
      "epoch": 0.5792824074074074,
      "grad_norm": 0.05358143523335457,
      "learning_rate": 8.414351851851853e-06,
      "loss": 0.0099,
      "step": 5005
    },
    {
      "epoch": 0.5793981481481482,
      "grad_norm": 0.07546571642160416,
      "learning_rate": 8.412037037037037e-06,
      "loss": 0.0137,
      "step": 5006
    },
    {
      "epoch": 0.5795138888888889,
      "grad_norm": 140.92066955566406,
      "learning_rate": 8.409722222222223e-06,
      "loss": 0.5676,
      "step": 5007
    },
    {
      "epoch": 0.5796296296296296,
      "grad_norm": 0.06083488091826439,
      "learning_rate": 8.407407407407409e-06,
      "loss": 0.0113,
      "step": 5008
    },
    {
      "epoch": 0.5797453703703703,
      "grad_norm": 0.06716986745595932,
      "learning_rate": 8.405092592592593e-06,
      "loss": 0.0124,
      "step": 5009
    },
    {
      "epoch": 0.5798611111111112,
      "grad_norm": 0.06576871126890182,
      "learning_rate": 8.402777777777779e-06,
      "loss": 0.0122,
      "step": 5010
    },
    {
      "epoch": 0.5799768518518519,
      "grad_norm": 0.1264413297176361,
      "learning_rate": 8.400462962962963e-06,
      "loss": 0.0171,
      "step": 5011
    },
    {
      "epoch": 0.5800925925925926,
      "grad_norm": 0.0611894428730011,
      "learning_rate": 8.39814814814815e-06,
      "loss": 0.0113,
      "step": 5012
    },
    {
      "epoch": 0.5802083333333333,
      "grad_norm": 2.144364595413208,
      "learning_rate": 8.395833333333334e-06,
      "loss": 0.0186,
      "step": 5013
    },
    {
      "epoch": 0.580324074074074,
      "grad_norm": 11.246028900146484,
      "learning_rate": 8.39351851851852e-06,
      "loss": 0.0435,
      "step": 5014
    },
    {
      "epoch": 0.5804398148148148,
      "grad_norm": 0.07544773072004318,
      "learning_rate": 8.391203703703704e-06,
      "loss": 0.0136,
      "step": 5015
    },
    {
      "epoch": 0.5805555555555556,
      "grad_norm": 0.05386261269450188,
      "learning_rate": 8.38888888888889e-06,
      "loss": 0.01,
      "step": 5016
    },
    {
      "epoch": 0.5806712962962963,
      "grad_norm": 0.06974498182535172,
      "learning_rate": 8.386574074074076e-06,
      "loss": 0.0127,
      "step": 5017
    },
    {
      "epoch": 0.580787037037037,
      "grad_norm": 0.1479688435792923,
      "learning_rate": 8.38425925925926e-06,
      "loss": 0.02,
      "step": 5018
    },
    {
      "epoch": 0.5809027777777778,
      "grad_norm": 0.060995303094387054,
      "learning_rate": 8.381944444444446e-06,
      "loss": 0.0112,
      "step": 5019
    },
    {
      "epoch": 0.5810185185185185,
      "grad_norm": 0.07284397631883621,
      "learning_rate": 8.37962962962963e-06,
      "loss": 0.0134,
      "step": 5020
    },
    {
      "epoch": 0.5811342592592592,
      "grad_norm": 0.07794120907783508,
      "learning_rate": 8.377314814814816e-06,
      "loss": 0.0139,
      "step": 5021
    },
    {
      "epoch": 0.58125,
      "grad_norm": 0.09496606141328812,
      "learning_rate": 8.375e-06,
      "loss": 0.0176,
      "step": 5022
    },
    {
      "epoch": 0.5813657407407408,
      "grad_norm": 0.09048081934452057,
      "learning_rate": 8.372685185185185e-06,
      "loss": 0.0167,
      "step": 5023
    },
    {
      "epoch": 0.5814814814814815,
      "grad_norm": 0.09085868299007416,
      "learning_rate": 8.37037037037037e-06,
      "loss": 0.0163,
      "step": 5024
    },
    {
      "epoch": 0.5815972222222222,
      "grad_norm": 0.07857711613178253,
      "learning_rate": 8.368055555555556e-06,
      "loss": 0.0145,
      "step": 5025
    },
    {
      "epoch": 0.5817129629629629,
      "grad_norm": 0.06188356131315231,
      "learning_rate": 8.36574074074074e-06,
      "loss": 0.0114,
      "step": 5026
    },
    {
      "epoch": 0.5818287037037037,
      "grad_norm": 0.11501217633485794,
      "learning_rate": 8.363425925925927e-06,
      "loss": 0.0155,
      "step": 5027
    },
    {
      "epoch": 0.5819444444444445,
      "grad_norm": 0.08486035466194153,
      "learning_rate": 8.361111111111113e-06,
      "loss": 0.015,
      "step": 5028
    },
    {
      "epoch": 0.5820601851851852,
      "grad_norm": 0.08940520137548447,
      "learning_rate": 8.358796296296297e-06,
      "loss": 0.0129,
      "step": 5029
    },
    {
      "epoch": 0.5821759259259259,
      "grad_norm": 0.1157534047961235,
      "learning_rate": 8.356481481481483e-06,
      "loss": 0.0156,
      "step": 5030
    },
    {
      "epoch": 0.5822916666666667,
      "grad_norm": 34.941078186035156,
      "learning_rate": 8.354166666666667e-06,
      "loss": 0.0796,
      "step": 5031
    },
    {
      "epoch": 0.5824074074074074,
      "grad_norm": 0.08539564162492752,
      "learning_rate": 8.351851851851851e-06,
      "loss": 0.0158,
      "step": 5032
    },
    {
      "epoch": 0.5825231481481481,
      "grad_norm": 0.09217982739210129,
      "learning_rate": 8.349537037037037e-06,
      "loss": 0.0156,
      "step": 5033
    },
    {
      "epoch": 0.5826388888888889,
      "grad_norm": 0.07337113469839096,
      "learning_rate": 8.347222222222223e-06,
      "loss": 0.0136,
      "step": 5034
    },
    {
      "epoch": 0.5827546296296297,
      "grad_norm": 0.09780307114124298,
      "learning_rate": 8.344907407407407e-06,
      "loss": 0.0167,
      "step": 5035
    },
    {
      "epoch": 0.5828703703703704,
      "grad_norm": 71.17083740234375,
      "learning_rate": 8.342592592592593e-06,
      "loss": 0.2582,
      "step": 5036
    },
    {
      "epoch": 0.5829861111111111,
      "grad_norm": 0.6232799291610718,
      "learning_rate": 8.34027777777778e-06,
      "loss": 0.0237,
      "step": 5037
    },
    {
      "epoch": 0.5831018518518518,
      "grad_norm": 0.07897263765335083,
      "learning_rate": 8.337962962962964e-06,
      "loss": 0.0139,
      "step": 5038
    },
    {
      "epoch": 0.5832175925925925,
      "grad_norm": 0.07370690256357193,
      "learning_rate": 8.33564814814815e-06,
      "loss": 0.0098,
      "step": 5039
    },
    {
      "epoch": 0.5833333333333334,
      "grad_norm": 0.07616563141345978,
      "learning_rate": 8.333333333333334e-06,
      "loss": 0.0138,
      "step": 5040
    },
    {
      "epoch": 0.5834490740740741,
      "grad_norm": 0.06127355992794037,
      "learning_rate": 8.33101851851852e-06,
      "loss": 0.0112,
      "step": 5041
    },
    {
      "epoch": 0.5835648148148148,
      "grad_norm": 0.06117783114314079,
      "learning_rate": 8.328703703703704e-06,
      "loss": 0.0112,
      "step": 5042
    },
    {
      "epoch": 0.5836805555555555,
      "grad_norm": 0.09223921597003937,
      "learning_rate": 8.32638888888889e-06,
      "loss": 0.0164,
      "step": 5043
    },
    {
      "epoch": 0.5837962962962963,
      "grad_norm": 0.08283575624227524,
      "learning_rate": 8.324074074074074e-06,
      "loss": 0.0148,
      "step": 5044
    },
    {
      "epoch": 0.5839120370370371,
      "grad_norm": 0.08757627755403519,
      "learning_rate": 8.32175925925926e-06,
      "loss": 0.0164,
      "step": 5045
    },
    {
      "epoch": 0.5840277777777778,
      "grad_norm": 0.2468290776014328,
      "learning_rate": 8.319444444444446e-06,
      "loss": 0.017,
      "step": 5046
    },
    {
      "epoch": 0.5841435185185185,
      "grad_norm": 0.0542907677590847,
      "learning_rate": 8.31712962962963e-06,
      "loss": 0.0099,
      "step": 5047
    },
    {
      "epoch": 0.5842592592592593,
      "grad_norm": 0.08606105297803879,
      "learning_rate": 8.314814814814816e-06,
      "loss": 0.0159,
      "step": 5048
    },
    {
      "epoch": 0.584375,
      "grad_norm": 5.138966083526611,
      "learning_rate": 8.3125e-06,
      "loss": 0.0312,
      "step": 5049
    },
    {
      "epoch": 0.5844907407407407,
      "grad_norm": 6.437255382537842,
      "learning_rate": 8.310185185185186e-06,
      "loss": 2.942,
      "step": 5050
    },
    {
      "epoch": 0.5846064814814815,
      "grad_norm": 0.07958599179983139,
      "learning_rate": 8.30787037037037e-06,
      "loss": 0.0145,
      "step": 5051
    },
    {
      "epoch": 0.5847222222222223,
      "grad_norm": 0.09082458913326263,
      "learning_rate": 8.305555555555557e-06,
      "loss": 0.0152,
      "step": 5052
    },
    {
      "epoch": 0.584837962962963,
      "grad_norm": 0.06797351688146591,
      "learning_rate": 8.303240740740741e-06,
      "loss": 0.0123,
      "step": 5053
    },
    {
      "epoch": 0.5849537037037037,
      "grad_norm": 23.6524715423584,
      "learning_rate": 8.300925925925927e-06,
      "loss": 2.9614,
      "step": 5054
    },
    {
      "epoch": 0.5850694444444444,
      "grad_norm": 0.07956690341234207,
      "learning_rate": 8.298611111111113e-06,
      "loss": 0.0144,
      "step": 5055
    },
    {
      "epoch": 0.5851851851851851,
      "grad_norm": 0.06458917260169983,
      "learning_rate": 8.296296296296297e-06,
      "loss": 0.0119,
      "step": 5056
    },
    {
      "epoch": 0.585300925925926,
      "grad_norm": 0.09442075341939926,
      "learning_rate": 8.293981481481483e-06,
      "loss": 0.0142,
      "step": 5057
    },
    {
      "epoch": 0.5854166666666667,
      "grad_norm": 0.08201610296964645,
      "learning_rate": 8.291666666666667e-06,
      "loss": 0.0146,
      "step": 5058
    },
    {
      "epoch": 0.5855324074074074,
      "grad_norm": 0.08884494751691818,
      "learning_rate": 8.289351851851853e-06,
      "loss": 0.0153,
      "step": 5059
    },
    {
      "epoch": 0.5856481481481481,
      "grad_norm": 0.05925643444061279,
      "learning_rate": 8.287037037037037e-06,
      "loss": 0.0109,
      "step": 5060
    },
    {
      "epoch": 0.5857638888888889,
      "grad_norm": 0.11027561128139496,
      "learning_rate": 8.284722222222223e-06,
      "loss": 0.0148,
      "step": 5061
    },
    {
      "epoch": 0.5858796296296296,
      "grad_norm": 4.710783004760742,
      "learning_rate": 8.282407407407408e-06,
      "loss": 0.0256,
      "step": 5062
    },
    {
      "epoch": 0.5859953703703704,
      "grad_norm": 0.10522007197141647,
      "learning_rate": 8.280092592592594e-06,
      "loss": 0.0185,
      "step": 5063
    },
    {
      "epoch": 0.5861111111111111,
      "grad_norm": 146.3009490966797,
      "learning_rate": 8.277777777777778e-06,
      "loss": 0.7659,
      "step": 5064
    },
    {
      "epoch": 0.5862268518518519,
      "grad_norm": 0.08308423310518265,
      "learning_rate": 8.275462962962964e-06,
      "loss": 0.0145,
      "step": 5065
    },
    {
      "epoch": 0.5863425925925926,
      "grad_norm": 0.18079306185245514,
      "learning_rate": 8.27314814814815e-06,
      "loss": 0.0156,
      "step": 5066
    },
    {
      "epoch": 0.5864583333333333,
      "grad_norm": 0.10557200014591217,
      "learning_rate": 8.270833333333334e-06,
      "loss": 0.0139,
      "step": 5067
    },
    {
      "epoch": 0.586574074074074,
      "grad_norm": 0.08224194496870041,
      "learning_rate": 8.26851851851852e-06,
      "loss": 0.0133,
      "step": 5068
    },
    {
      "epoch": 0.5866898148148149,
      "grad_norm": 0.05728689953684807,
      "learning_rate": 8.266203703703704e-06,
      "loss": 0.0106,
      "step": 5069
    },
    {
      "epoch": 0.5868055555555556,
      "grad_norm": 72.05599212646484,
      "learning_rate": 8.263888888888888e-06,
      "loss": 0.1738,
      "step": 5070
    },
    {
      "epoch": 0.5869212962962963,
      "grad_norm": 0.13443303108215332,
      "learning_rate": 8.261574074074074e-06,
      "loss": 0.0143,
      "step": 5071
    },
    {
      "epoch": 0.587037037037037,
      "grad_norm": 0.0907445177435875,
      "learning_rate": 8.25925925925926e-06,
      "loss": 0.0169,
      "step": 5072
    },
    {
      "epoch": 0.5871527777777777,
      "grad_norm": 0.4921228885650635,
      "learning_rate": 8.256944444444444e-06,
      "loss": 0.0179,
      "step": 5073
    },
    {
      "epoch": 0.5872685185185185,
      "grad_norm": 0.08084799349308014,
      "learning_rate": 8.25462962962963e-06,
      "loss": 0.0139,
      "step": 5074
    },
    {
      "epoch": 0.5873842592592593,
      "grad_norm": 0.0728394016623497,
      "learning_rate": 8.252314814814816e-06,
      "loss": 0.0133,
      "step": 5075
    },
    {
      "epoch": 0.5875,
      "grad_norm": 0.06978441774845123,
      "learning_rate": 8.25e-06,
      "loss": 0.0125,
      "step": 5076
    },
    {
      "epoch": 0.5876157407407407,
      "grad_norm": 1.459808111190796,
      "learning_rate": 8.247685185185187e-06,
      "loss": 0.02,
      "step": 5077
    },
    {
      "epoch": 0.5877314814814815,
      "grad_norm": 0.057833731174468994,
      "learning_rate": 8.24537037037037e-06,
      "loss": 0.0106,
      "step": 5078
    },
    {
      "epoch": 0.5878472222222222,
      "grad_norm": 0.09083013236522675,
      "learning_rate": 8.243055555555555e-06,
      "loss": 0.016,
      "step": 5079
    },
    {
      "epoch": 0.5879629629629629,
      "grad_norm": 9.185869216918945,
      "learning_rate": 8.240740740740741e-06,
      "loss": 0.0558,
      "step": 5080
    },
    {
      "epoch": 0.5880787037037037,
      "grad_norm": 0.06325549632310867,
      "learning_rate": 8.238425925925927e-06,
      "loss": 0.0118,
      "step": 5081
    },
    {
      "epoch": 0.5881944444444445,
      "grad_norm": 0.07181458175182343,
      "learning_rate": 8.236111111111111e-06,
      "loss": 0.013,
      "step": 5082
    },
    {
      "epoch": 0.5883101851851852,
      "grad_norm": 0.05889203026890755,
      "learning_rate": 8.233796296296297e-06,
      "loss": 0.0109,
      "step": 5083
    },
    {
      "epoch": 0.5884259259259259,
      "grad_norm": 11.28072738647461,
      "learning_rate": 8.231481481481483e-06,
      "loss": 2.9114,
      "step": 5084
    },
    {
      "epoch": 0.5885416666666666,
      "grad_norm": 0.07463046908378601,
      "learning_rate": 8.229166666666667e-06,
      "loss": 0.0137,
      "step": 5085
    },
    {
      "epoch": 0.5886574074074075,
      "grad_norm": 0.05704725533723831,
      "learning_rate": 8.226851851851853e-06,
      "loss": 0.0101,
      "step": 5086
    },
    {
      "epoch": 0.5887731481481482,
      "grad_norm": 0.07374860346317291,
      "learning_rate": 8.224537037037038e-06,
      "loss": 0.0097,
      "step": 5087
    },
    {
      "epoch": 0.5888888888888889,
      "grad_norm": 0.12225107103586197,
      "learning_rate": 8.222222222222222e-06,
      "loss": 0.016,
      "step": 5088
    },
    {
      "epoch": 0.5890046296296296,
      "grad_norm": 0.07426425814628601,
      "learning_rate": 8.219907407407408e-06,
      "loss": 0.0137,
      "step": 5089
    },
    {
      "epoch": 0.5891203703703703,
      "grad_norm": 0.08795121312141418,
      "learning_rate": 8.217592592592594e-06,
      "loss": 0.0163,
      "step": 5090
    },
    {
      "epoch": 0.5892361111111111,
      "grad_norm": 0.05134858563542366,
      "learning_rate": 8.215277777777778e-06,
      "loss": 0.0095,
      "step": 5091
    },
    {
      "epoch": 0.5893518518518519,
      "grad_norm": 0.08186566084623337,
      "learning_rate": 8.212962962962964e-06,
      "loss": 0.0136,
      "step": 5092
    },
    {
      "epoch": 0.5894675925925926,
      "grad_norm": 0.0625663623213768,
      "learning_rate": 8.21064814814815e-06,
      "loss": 0.0111,
      "step": 5093
    },
    {
      "epoch": 0.5895833333333333,
      "grad_norm": 69.59193420410156,
      "learning_rate": 8.208333333333334e-06,
      "loss": 1.6033,
      "step": 5094
    },
    {
      "epoch": 0.5896990740740741,
      "grad_norm": 20.841440200805664,
      "learning_rate": 8.20601851851852e-06,
      "loss": 2.7153,
      "step": 5095
    },
    {
      "epoch": 0.5898148148148148,
      "grad_norm": 4.208897113800049,
      "learning_rate": 8.203703703703704e-06,
      "loss": 0.0252,
      "step": 5096
    },
    {
      "epoch": 0.5899305555555555,
      "grad_norm": 0.08398059010505676,
      "learning_rate": 8.201388888888889e-06,
      "loss": 0.0156,
      "step": 5097
    },
    {
      "epoch": 0.5900462962962963,
      "grad_norm": 0.6888330578804016,
      "learning_rate": 8.199074074074074e-06,
      "loss": 0.0187,
      "step": 5098
    },
    {
      "epoch": 0.5901620370370371,
      "grad_norm": 0.12595690786838531,
      "learning_rate": 8.19675925925926e-06,
      "loss": 0.0169,
      "step": 5099
    },
    {
      "epoch": 0.5902777777777778,
      "grad_norm": 7.1697258949279785,
      "learning_rate": 8.194444444444445e-06,
      "loss": 2.9004,
      "step": 5100
    },
    {
      "epoch": 0.5903935185185185,
      "grad_norm": 0.08595282584428787,
      "learning_rate": 8.19212962962963e-06,
      "loss": 0.0147,
      "step": 5101
    },
    {
      "epoch": 0.5905092592592592,
      "grad_norm": 0.06390831619501114,
      "learning_rate": 8.189814814814817e-06,
      "loss": 0.0118,
      "step": 5102
    },
    {
      "epoch": 0.590625,
      "grad_norm": 0.06971842795610428,
      "learning_rate": 8.1875e-06,
      "loss": 0.0103,
      "step": 5103
    },
    {
      "epoch": 0.5907407407407408,
      "grad_norm": 0.0649695172905922,
      "learning_rate": 8.185185185185187e-06,
      "loss": 0.0119,
      "step": 5104
    },
    {
      "epoch": 0.5908564814814815,
      "grad_norm": 5.902162551879883,
      "learning_rate": 8.182870370370371e-06,
      "loss": 0.0411,
      "step": 5105
    },
    {
      "epoch": 0.5909722222222222,
      "grad_norm": 16.547569274902344,
      "learning_rate": 8.180555555555555e-06,
      "loss": 0.038,
      "step": 5106
    },
    {
      "epoch": 0.591087962962963,
      "grad_norm": 0.08110987395048141,
      "learning_rate": 8.178240740740741e-06,
      "loss": 0.0141,
      "step": 5107
    },
    {
      "epoch": 0.5912037037037037,
      "grad_norm": 0.07222238928079605,
      "learning_rate": 8.175925925925925e-06,
      "loss": 0.013,
      "step": 5108
    },
    {
      "epoch": 0.5913194444444444,
      "grad_norm": 0.07423753291368484,
      "learning_rate": 8.173611111111111e-06,
      "loss": 0.0135,
      "step": 5109
    },
    {
      "epoch": 0.5914351851851852,
      "grad_norm": 0.630388081073761,
      "learning_rate": 8.171296296296297e-06,
      "loss": 0.0233,
      "step": 5110
    },
    {
      "epoch": 0.591550925925926,
      "grad_norm": 0.09810242801904678,
      "learning_rate": 8.168981481481482e-06,
      "loss": 0.0181,
      "step": 5111
    },
    {
      "epoch": 0.5916666666666667,
      "grad_norm": 11.07114315032959,
      "learning_rate": 8.166666666666668e-06,
      "loss": 0.0666,
      "step": 5112
    },
    {
      "epoch": 0.5917824074074074,
      "grad_norm": 0.07152344286441803,
      "learning_rate": 8.164351851851853e-06,
      "loss": 0.0132,
      "step": 5113
    },
    {
      "epoch": 0.5918981481481481,
      "grad_norm": 0.0652805045247078,
      "learning_rate": 8.162037037037038e-06,
      "loss": 0.011,
      "step": 5114
    },
    {
      "epoch": 0.5920138888888888,
      "grad_norm": 0.05928558483719826,
      "learning_rate": 8.159722222222222e-06,
      "loss": 0.0103,
      "step": 5115
    },
    {
      "epoch": 0.5921296296296297,
      "grad_norm": 76.17729187011719,
      "learning_rate": 8.157407407407408e-06,
      "loss": 1.9408,
      "step": 5116
    },
    {
      "epoch": 0.5922453703703704,
      "grad_norm": 0.0719216912984848,
      "learning_rate": 8.155092592592592e-06,
      "loss": 0.0096,
      "step": 5117
    },
    {
      "epoch": 0.5923611111111111,
      "grad_norm": 0.09277387708425522,
      "learning_rate": 8.152777777777778e-06,
      "loss": 0.0142,
      "step": 5118
    },
    {
      "epoch": 0.5924768518518518,
      "grad_norm": 0.058527521789073944,
      "learning_rate": 8.150462962962964e-06,
      "loss": 0.0108,
      "step": 5119
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 0.12428215891122818,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.0162,
      "step": 5120
    },
    {
      "epoch": 0.5927083333333333,
      "grad_norm": 0.07623942196369171,
      "learning_rate": 8.145833333333334e-06,
      "loss": 0.0135,
      "step": 5121
    },
    {
      "epoch": 0.5928240740740741,
      "grad_norm": 0.07597707211971283,
      "learning_rate": 8.14351851851852e-06,
      "loss": 0.0134,
      "step": 5122
    },
    {
      "epoch": 0.5929398148148148,
      "grad_norm": 2.62107515335083,
      "learning_rate": 8.141203703703704e-06,
      "loss": 0.0296,
      "step": 5123
    },
    {
      "epoch": 0.5930555555555556,
      "grad_norm": 0.07483271509408951,
      "learning_rate": 8.138888888888889e-06,
      "loss": 0.0137,
      "step": 5124
    },
    {
      "epoch": 0.5931712962962963,
      "grad_norm": 0.08375795930624008,
      "learning_rate": 8.136574074074075e-06,
      "loss": 0.0155,
      "step": 5125
    },
    {
      "epoch": 0.593287037037037,
      "grad_norm": 4.970132350921631,
      "learning_rate": 8.134259259259259e-06,
      "loss": 0.0262,
      "step": 5126
    },
    {
      "epoch": 0.5934027777777777,
      "grad_norm": 0.06721418350934982,
      "learning_rate": 8.131944444444445e-06,
      "loss": 0.0121,
      "step": 5127
    },
    {
      "epoch": 0.5935185185185186,
      "grad_norm": 123.59413146972656,
      "learning_rate": 8.12962962962963e-06,
      "loss": 1.0801,
      "step": 5128
    },
    {
      "epoch": 0.5936342592592593,
      "grad_norm": 0.08386784791946411,
      "learning_rate": 8.127314814814815e-06,
      "loss": 0.0136,
      "step": 5129
    },
    {
      "epoch": 0.59375,
      "grad_norm": 0.06654047220945358,
      "learning_rate": 8.125000000000001e-06,
      "loss": 0.0121,
      "step": 5130
    },
    {
      "epoch": 0.5938657407407407,
      "grad_norm": 0.07923837751150131,
      "learning_rate": 8.122685185185187e-06,
      "loss": 0.0142,
      "step": 5131
    },
    {
      "epoch": 0.5939814814814814,
      "grad_norm": 0.0817941427230835,
      "learning_rate": 8.120370370370371e-06,
      "loss": 0.0109,
      "step": 5132
    },
    {
      "epoch": 0.5940972222222223,
      "grad_norm": 0.2838073670864105,
      "learning_rate": 8.118055555555555e-06,
      "loss": 0.0201,
      "step": 5133
    },
    {
      "epoch": 0.594212962962963,
      "grad_norm": 0.07065357267856598,
      "learning_rate": 8.115740740740741e-06,
      "loss": 0.0127,
      "step": 5134
    },
    {
      "epoch": 0.5943287037037037,
      "grad_norm": 0.10210438072681427,
      "learning_rate": 8.113425925925926e-06,
      "loss": 0.0166,
      "step": 5135
    },
    {
      "epoch": 0.5944444444444444,
      "grad_norm": 0.07228369265794754,
      "learning_rate": 8.111111111111112e-06,
      "loss": 0.013,
      "step": 5136
    },
    {
      "epoch": 0.5945601851851852,
      "grad_norm": 0.11547987163066864,
      "learning_rate": 8.108796296296298e-06,
      "loss": 0.0171,
      "step": 5137
    },
    {
      "epoch": 0.5946759259259259,
      "grad_norm": 0.051725421100854874,
      "learning_rate": 8.106481481481482e-06,
      "loss": 0.0095,
      "step": 5138
    },
    {
      "epoch": 0.5947916666666667,
      "grad_norm": 0.09508904814720154,
      "learning_rate": 8.104166666666668e-06,
      "loss": 0.0165,
      "step": 5139
    },
    {
      "epoch": 0.5949074074074074,
      "grad_norm": 0.11991589516401291,
      "learning_rate": 8.101851851851854e-06,
      "loss": 0.0133,
      "step": 5140
    },
    {
      "epoch": 0.5950231481481482,
      "grad_norm": 0.08702187240123749,
      "learning_rate": 8.099537037037038e-06,
      "loss": 0.0163,
      "step": 5141
    },
    {
      "epoch": 0.5951388888888889,
      "grad_norm": 0.06028669700026512,
      "learning_rate": 8.097222222222222e-06,
      "loss": 0.011,
      "step": 5142
    },
    {
      "epoch": 0.5952546296296296,
      "grad_norm": 0.07751930505037308,
      "learning_rate": 8.094907407407408e-06,
      "loss": 0.0114,
      "step": 5143
    },
    {
      "epoch": 0.5953703703703703,
      "grad_norm": 0.07206183671951294,
      "learning_rate": 8.092592592592592e-06,
      "loss": 0.0114,
      "step": 5144
    },
    {
      "epoch": 0.5954861111111112,
      "grad_norm": 0.06628736108541489,
      "learning_rate": 8.090277777777778e-06,
      "loss": 0.0113,
      "step": 5145
    },
    {
      "epoch": 0.5956018518518519,
      "grad_norm": 0.08234789967536926,
      "learning_rate": 8.087962962962964e-06,
      "loss": 0.014,
      "step": 5146
    },
    {
      "epoch": 0.5957175925925926,
      "grad_norm": 0.07823047041893005,
      "learning_rate": 8.085648148148148e-06,
      "loss": 0.0143,
      "step": 5147
    },
    {
      "epoch": 0.5958333333333333,
      "grad_norm": 0.33102983236312866,
      "learning_rate": 8.083333333333334e-06,
      "loss": 0.0208,
      "step": 5148
    },
    {
      "epoch": 0.595949074074074,
      "grad_norm": 0.07607421278953552,
      "learning_rate": 8.08101851851852e-06,
      "loss": 0.0132,
      "step": 5149
    },
    {
      "epoch": 0.5960648148148148,
      "grad_norm": 0.09521125257015228,
      "learning_rate": 8.078703703703705e-06,
      "loss": 0.0133,
      "step": 5150
    },
    {
      "epoch": 0.5961805555555556,
      "grad_norm": 0.07982975989580154,
      "learning_rate": 8.076388888888889e-06,
      "loss": 0.0134,
      "step": 5151
    },
    {
      "epoch": 0.5962962962962963,
      "grad_norm": 0.07901564240455627,
      "learning_rate": 8.074074074074075e-06,
      "loss": 0.0134,
      "step": 5152
    },
    {
      "epoch": 0.596412037037037,
      "grad_norm": 0.06505070626735687,
      "learning_rate": 8.071759259259259e-06,
      "loss": 0.0117,
      "step": 5153
    },
    {
      "epoch": 0.5965277777777778,
      "grad_norm": 39.490928649902344,
      "learning_rate": 8.069444444444445e-06,
      "loss": 2.085,
      "step": 5154
    },
    {
      "epoch": 0.5966435185185185,
      "grad_norm": 0.07309369742870331,
      "learning_rate": 8.06712962962963e-06,
      "loss": 0.0097,
      "step": 5155
    },
    {
      "epoch": 0.5967592592592592,
      "grad_norm": 0.05225634202361107,
      "learning_rate": 8.064814814814815e-06,
      "loss": 0.0096,
      "step": 5156
    },
    {
      "epoch": 0.596875,
      "grad_norm": 31.10211944580078,
      "learning_rate": 8.062500000000001e-06,
      "loss": 0.0831,
      "step": 5157
    },
    {
      "epoch": 0.5969907407407408,
      "grad_norm": 0.08051753789186478,
      "learning_rate": 8.060185185185185e-06,
      "loss": 0.015,
      "step": 5158
    },
    {
      "epoch": 0.5971064814814815,
      "grad_norm": 0.07619782537221909,
      "learning_rate": 8.057870370370371e-06,
      "loss": 0.0129,
      "step": 5159
    },
    {
      "epoch": 0.5972222222222222,
      "grad_norm": 0.11951577663421631,
      "learning_rate": 8.055555555555557e-06,
      "loss": 0.0161,
      "step": 5160
    },
    {
      "epoch": 0.5973379629629629,
      "grad_norm": 0.12874211370944977,
      "learning_rate": 8.053240740740742e-06,
      "loss": 0.0143,
      "step": 5161
    },
    {
      "epoch": 0.5974537037037037,
      "grad_norm": 0.06612443923950195,
      "learning_rate": 8.050925925925926e-06,
      "loss": 0.012,
      "step": 5162
    },
    {
      "epoch": 0.5975694444444445,
      "grad_norm": 0.11030644178390503,
      "learning_rate": 8.048611111111112e-06,
      "loss": 0.0121,
      "step": 5163
    },
    {
      "epoch": 0.5976851851851852,
      "grad_norm": 0.059326041489839554,
      "learning_rate": 8.046296296296296e-06,
      "loss": 0.0106,
      "step": 5164
    },
    {
      "epoch": 0.5978009259259259,
      "grad_norm": 0.40990588068962097,
      "learning_rate": 8.043981481481482e-06,
      "loss": 0.0203,
      "step": 5165
    },
    {
      "epoch": 0.5979166666666667,
      "grad_norm": 0.06948923319578171,
      "learning_rate": 8.041666666666668e-06,
      "loss": 0.0124,
      "step": 5166
    },
    {
      "epoch": 0.5980324074074074,
      "grad_norm": 0.05195475369691849,
      "learning_rate": 8.039351851851852e-06,
      "loss": 0.0096,
      "step": 5167
    },
    {
      "epoch": 0.5981481481481481,
      "grad_norm": 0.06559088826179504,
      "learning_rate": 8.037037037037038e-06,
      "loss": 0.0119,
      "step": 5168
    },
    {
      "epoch": 0.5982638888888889,
      "grad_norm": 0.07039543241262436,
      "learning_rate": 8.034722222222224e-06,
      "loss": 0.0127,
      "step": 5169
    },
    {
      "epoch": 0.5983796296296297,
      "grad_norm": 0.11729127913713455,
      "learning_rate": 8.032407407407408e-06,
      "loss": 0.0137,
      "step": 5170
    },
    {
      "epoch": 0.5984953703703704,
      "grad_norm": 0.05483740568161011,
      "learning_rate": 8.030092592592593e-06,
      "loss": 0.0102,
      "step": 5171
    },
    {
      "epoch": 0.5986111111111111,
      "grad_norm": 0.05772695690393448,
      "learning_rate": 8.027777777777778e-06,
      "loss": 0.0106,
      "step": 5172
    },
    {
      "epoch": 0.5987268518518518,
      "grad_norm": 0.09008757770061493,
      "learning_rate": 8.025462962962963e-06,
      "loss": 0.0136,
      "step": 5173
    },
    {
      "epoch": 0.5988425925925925,
      "grad_norm": 107.6123046875,
      "learning_rate": 8.023148148148149e-06,
      "loss": 1.6317,
      "step": 5174
    },
    {
      "epoch": 0.5989583333333334,
      "grad_norm": 0.08222761005163193,
      "learning_rate": 8.020833333333335e-06,
      "loss": 0.0136,
      "step": 5175
    },
    {
      "epoch": 0.5990740740740741,
      "grad_norm": 0.0943697914481163,
      "learning_rate": 8.018518518518519e-06,
      "loss": 0.0159,
      "step": 5176
    },
    {
      "epoch": 0.5991898148148148,
      "grad_norm": 183.45176696777344,
      "learning_rate": 8.016203703703705e-06,
      "loss": 1.1551,
      "step": 5177
    },
    {
      "epoch": 0.5993055555555555,
      "grad_norm": 0.27790918946266174,
      "learning_rate": 8.01388888888889e-06,
      "loss": 0.019,
      "step": 5178
    },
    {
      "epoch": 0.5994212962962963,
      "grad_norm": 0.07348905503749847,
      "learning_rate": 8.011574074074075e-06,
      "loss": 0.0131,
      "step": 5179
    },
    {
      "epoch": 0.5995370370370371,
      "grad_norm": 0.10163021832704544,
      "learning_rate": 8.00925925925926e-06,
      "loss": 0.018,
      "step": 5180
    },
    {
      "epoch": 0.5996527777777778,
      "grad_norm": 0.11031908541917801,
      "learning_rate": 8.006944444444445e-06,
      "loss": 0.014,
      "step": 5181
    },
    {
      "epoch": 0.5997685185185185,
      "grad_norm": 0.07672131806612015,
      "learning_rate": 8.00462962962963e-06,
      "loss": 0.0134,
      "step": 5182
    },
    {
      "epoch": 0.5998842592592593,
      "grad_norm": 0.05137977749109268,
      "learning_rate": 8.002314814814815e-06,
      "loss": 0.0095,
      "step": 5183
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.07145822793245316,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.0128,
      "step": 5184
    },
    {
      "epoch": 0.6001157407407407,
      "grad_norm": 0.06459003686904907,
      "learning_rate": 7.997685185185186e-06,
      "loss": 0.0117,
      "step": 5185
    },
    {
      "epoch": 0.6002314814814815,
      "grad_norm": 0.06994888931512833,
      "learning_rate": 7.995370370370372e-06,
      "loss": 0.0129,
      "step": 5186
    },
    {
      "epoch": 0.6003472222222223,
      "grad_norm": 0.2546600103378296,
      "learning_rate": 7.993055555555557e-06,
      "loss": 0.0198,
      "step": 5187
    },
    {
      "epoch": 0.600462962962963,
      "grad_norm": 0.09570993483066559,
      "learning_rate": 7.990740740740742e-06,
      "loss": 0.0172,
      "step": 5188
    },
    {
      "epoch": 0.6005787037037037,
      "grad_norm": 0.07756046205759048,
      "learning_rate": 7.988425925925926e-06,
      "loss": 0.0133,
      "step": 5189
    },
    {
      "epoch": 0.6006944444444444,
      "grad_norm": 0.07208632677793503,
      "learning_rate": 7.986111111111112e-06,
      "loss": 0.0131,
      "step": 5190
    },
    {
      "epoch": 0.6008101851851851,
      "grad_norm": 0.5948809385299683,
      "learning_rate": 7.983796296296296e-06,
      "loss": 0.0167,
      "step": 5191
    },
    {
      "epoch": 0.600925925925926,
      "grad_norm": 7.851266860961914,
      "learning_rate": 7.981481481481482e-06,
      "loss": 3.3806,
      "step": 5192
    },
    {
      "epoch": 0.6010416666666667,
      "grad_norm": 1.0404832363128662,
      "learning_rate": 7.979166666666668e-06,
      "loss": 0.0214,
      "step": 5193
    },
    {
      "epoch": 0.6011574074074074,
      "grad_norm": 0.05512959137558937,
      "learning_rate": 7.976851851851852e-06,
      "loss": 0.0101,
      "step": 5194
    },
    {
      "epoch": 0.6012731481481481,
      "grad_norm": 0.062403883785009384,
      "learning_rate": 7.974537037037038e-06,
      "loss": 0.0114,
      "step": 5195
    },
    {
      "epoch": 0.6013888888888889,
      "grad_norm": 0.05454881116747856,
      "learning_rate": 7.972222222222224e-06,
      "loss": 0.0101,
      "step": 5196
    },
    {
      "epoch": 0.6015046296296296,
      "grad_norm": 0.06915038079023361,
      "learning_rate": 7.969907407407408e-06,
      "loss": 0.0124,
      "step": 5197
    },
    {
      "epoch": 0.6016203703703704,
      "grad_norm": 0.054067473858594894,
      "learning_rate": 7.967592592592593e-06,
      "loss": 0.01,
      "step": 5198
    },
    {
      "epoch": 0.6017361111111111,
      "grad_norm": 0.08013855665922165,
      "learning_rate": 7.965277777777779e-06,
      "loss": 0.0107,
      "step": 5199
    },
    {
      "epoch": 0.6018518518518519,
      "grad_norm": 0.05302402377128601,
      "learning_rate": 7.962962962962963e-06,
      "loss": 0.0096,
      "step": 5200
    },
    {
      "epoch": 0.6019675925925926,
      "grad_norm": 23.868879318237305,
      "learning_rate": 7.960648148148149e-06,
      "loss": 0.086,
      "step": 5201
    },
    {
      "epoch": 0.6020833333333333,
      "grad_norm": 0.10528773814439774,
      "learning_rate": 7.958333333333333e-06,
      "loss": 0.014,
      "step": 5202
    },
    {
      "epoch": 0.602199074074074,
      "grad_norm": 0.07367366552352905,
      "learning_rate": 7.956018518518519e-06,
      "loss": 0.0135,
      "step": 5203
    },
    {
      "epoch": 0.6023148148148149,
      "grad_norm": 0.05129576101899147,
      "learning_rate": 7.953703703703705e-06,
      "loss": 0.0094,
      "step": 5204
    },
    {
      "epoch": 0.6024305555555556,
      "grad_norm": 0.06656248867511749,
      "learning_rate": 7.95138888888889e-06,
      "loss": 0.012,
      "step": 5205
    },
    {
      "epoch": 0.6025462962962963,
      "grad_norm": 3.3197968006134033,
      "learning_rate": 7.949074074074075e-06,
      "loss": 0.0255,
      "step": 5206
    },
    {
      "epoch": 0.602662037037037,
      "grad_norm": 0.08307456225156784,
      "learning_rate": 7.94675925925926e-06,
      "loss": 0.0155,
      "step": 5207
    },
    {
      "epoch": 0.6027777777777777,
      "grad_norm": 0.0562724769115448,
      "learning_rate": 7.944444444444445e-06,
      "loss": 0.0104,
      "step": 5208
    },
    {
      "epoch": 0.6028935185185185,
      "grad_norm": 0.09300746023654938,
      "learning_rate": 7.94212962962963e-06,
      "loss": 0.0161,
      "step": 5209
    },
    {
      "epoch": 0.6030092592592593,
      "grad_norm": 0.06858766078948975,
      "learning_rate": 7.939814814814816e-06,
      "loss": 0.0123,
      "step": 5210
    },
    {
      "epoch": 0.603125,
      "grad_norm": 0.06578968465328217,
      "learning_rate": 7.9375e-06,
      "loss": 0.0121,
      "step": 5211
    },
    {
      "epoch": 0.6032407407407407,
      "grad_norm": 0.08737123012542725,
      "learning_rate": 7.935185185185186e-06,
      "loss": 0.0149,
      "step": 5212
    },
    {
      "epoch": 0.6033564814814815,
      "grad_norm": 0.07435443252325058,
      "learning_rate": 7.932870370370372e-06,
      "loss": 0.0135,
      "step": 5213
    },
    {
      "epoch": 0.6034722222222222,
      "grad_norm": 0.05114132910966873,
      "learning_rate": 7.930555555555556e-06,
      "loss": 0.0094,
      "step": 5214
    },
    {
      "epoch": 0.6035879629629629,
      "grad_norm": 0.08713383972644806,
      "learning_rate": 7.928240740740742e-06,
      "loss": 0.0158,
      "step": 5215
    },
    {
      "epoch": 0.6037037037037037,
      "grad_norm": 0.5318819880485535,
      "learning_rate": 7.925925925925926e-06,
      "loss": 0.0175,
      "step": 5216
    },
    {
      "epoch": 0.6038194444444445,
      "grad_norm": 0.09206859022378922,
      "learning_rate": 7.923611111111112e-06,
      "loss": 0.0116,
      "step": 5217
    },
    {
      "epoch": 0.6039351851851852,
      "grad_norm": 0.06932748109102249,
      "learning_rate": 7.921296296296296e-06,
      "loss": 0.0125,
      "step": 5218
    },
    {
      "epoch": 0.6040509259259259,
      "grad_norm": 0.07012472301721573,
      "learning_rate": 7.918981481481482e-06,
      "loss": 0.0125,
      "step": 5219
    },
    {
      "epoch": 0.6041666666666666,
      "grad_norm": 0.05950024724006653,
      "learning_rate": 7.916666666666667e-06,
      "loss": 0.0111,
      "step": 5220
    },
    {
      "epoch": 0.6042824074074075,
      "grad_norm": 0.7092527747154236,
      "learning_rate": 7.914351851851852e-06,
      "loss": 0.0196,
      "step": 5221
    },
    {
      "epoch": 0.6043981481481482,
      "grad_norm": 0.06197074428200722,
      "learning_rate": 7.912037037037038e-06,
      "loss": 0.0114,
      "step": 5222
    },
    {
      "epoch": 0.6045138888888889,
      "grad_norm": 0.07341868430376053,
      "learning_rate": 7.909722222222223e-06,
      "loss": 0.0135,
      "step": 5223
    },
    {
      "epoch": 0.6046296296296296,
      "grad_norm": 0.32033878564834595,
      "learning_rate": 7.907407407407409e-06,
      "loss": 0.02,
      "step": 5224
    },
    {
      "epoch": 0.6047453703703703,
      "grad_norm": 0.060451384633779526,
      "learning_rate": 7.905092592592593e-06,
      "loss": 0.0112,
      "step": 5225
    },
    {
      "epoch": 0.6048611111111111,
      "grad_norm": 0.07761730998754501,
      "learning_rate": 7.902777777777779e-06,
      "loss": 0.0144,
      "step": 5226
    },
    {
      "epoch": 0.6049768518518519,
      "grad_norm": 0.2698904871940613,
      "learning_rate": 7.900462962962963e-06,
      "loss": 0.0175,
      "step": 5227
    },
    {
      "epoch": 0.6050925925925926,
      "grad_norm": 0.0851583182811737,
      "learning_rate": 7.898148148148149e-06,
      "loss": 0.0152,
      "step": 5228
    },
    {
      "epoch": 0.6052083333333333,
      "grad_norm": 0.08223149180412292,
      "learning_rate": 7.895833333333333e-06,
      "loss": 0.0151,
      "step": 5229
    },
    {
      "epoch": 0.6053240740740741,
      "grad_norm": 0.05937547981739044,
      "learning_rate": 7.89351851851852e-06,
      "loss": 0.011,
      "step": 5230
    },
    {
      "epoch": 0.6054398148148148,
      "grad_norm": 0.05523522570729256,
      "learning_rate": 7.891203703703705e-06,
      "loss": 0.0101,
      "step": 5231
    },
    {
      "epoch": 0.6055555555555555,
      "grad_norm": 0.08044624328613281,
      "learning_rate": 7.88888888888889e-06,
      "loss": 0.0149,
      "step": 5232
    },
    {
      "epoch": 0.6056712962962963,
      "grad_norm": 0.07330787926912308,
      "learning_rate": 7.886574074074075e-06,
      "loss": 0.0134,
      "step": 5233
    },
    {
      "epoch": 0.6057870370370371,
      "grad_norm": 0.06292083114385605,
      "learning_rate": 7.88425925925926e-06,
      "loss": 0.0111,
      "step": 5234
    },
    {
      "epoch": 0.6059027777777778,
      "grad_norm": 0.06460980325937271,
      "learning_rate": 7.881944444444446e-06,
      "loss": 0.012,
      "step": 5235
    },
    {
      "epoch": 0.6060185185185185,
      "grad_norm": 92.4054183959961,
      "learning_rate": 7.87962962962963e-06,
      "loss": 0.7415,
      "step": 5236
    },
    {
      "epoch": 0.6061342592592592,
      "grad_norm": 3.2692136764526367,
      "learning_rate": 7.877314814814816e-06,
      "loss": 0.0319,
      "step": 5237
    },
    {
      "epoch": 0.60625,
      "grad_norm": 0.08477572351694107,
      "learning_rate": 7.875e-06,
      "loss": 0.0106,
      "step": 5238
    },
    {
      "epoch": 0.6063657407407408,
      "grad_norm": 96.29261779785156,
      "learning_rate": 7.872685185185186e-06,
      "loss": 0.8654,
      "step": 5239
    },
    {
      "epoch": 0.6064814814814815,
      "grad_norm": 0.05809469521045685,
      "learning_rate": 7.870370370370372e-06,
      "loss": 0.0101,
      "step": 5240
    },
    {
      "epoch": 0.6065972222222222,
      "grad_norm": 0.07825624197721481,
      "learning_rate": 7.868055555555556e-06,
      "loss": 0.0144,
      "step": 5241
    },
    {
      "epoch": 0.606712962962963,
      "grad_norm": 0.05256138741970062,
      "learning_rate": 7.865740740740742e-06,
      "loss": 0.0097,
      "step": 5242
    },
    {
      "epoch": 0.6068287037037037,
      "grad_norm": 0.07281959056854248,
      "learning_rate": 7.863425925925926e-06,
      "loss": 0.0133,
      "step": 5243
    },
    {
      "epoch": 0.6069444444444444,
      "grad_norm": 0.05786646530032158,
      "learning_rate": 7.861111111111112e-06,
      "loss": 0.0105,
      "step": 5244
    },
    {
      "epoch": 0.6070601851851852,
      "grad_norm": 0.05181063339114189,
      "learning_rate": 7.858796296296297e-06,
      "loss": 0.0095,
      "step": 5245
    },
    {
      "epoch": 0.607175925925926,
      "grad_norm": 0.05849209427833557,
      "learning_rate": 7.85648148148148e-06,
      "loss": 0.0106,
      "step": 5246
    },
    {
      "epoch": 0.6072916666666667,
      "grad_norm": 0.07516061514616013,
      "learning_rate": 7.854166666666667e-06,
      "loss": 0.0129,
      "step": 5247
    },
    {
      "epoch": 0.6074074074074074,
      "grad_norm": 0.0757318064570427,
      "learning_rate": 7.851851851851853e-06,
      "loss": 0.0132,
      "step": 5248
    },
    {
      "epoch": 0.6075231481481481,
      "grad_norm": 0.09067614376544952,
      "learning_rate": 7.849537037037037e-06,
      "loss": 0.0155,
      "step": 5249
    },
    {
      "epoch": 0.6076388888888888,
      "grad_norm": 2.5614821910858154,
      "learning_rate": 7.847222222222223e-06,
      "loss": 0.027,
      "step": 5250
    },
    {
      "epoch": 0.6077546296296297,
      "grad_norm": 0.10270220041275024,
      "learning_rate": 7.844907407407409e-06,
      "loss": 0.0141,
      "step": 5251
    },
    {
      "epoch": 0.6078703703703704,
      "grad_norm": 0.055343788117170334,
      "learning_rate": 7.842592592592593e-06,
      "loss": 0.0102,
      "step": 5252
    },
    {
      "epoch": 0.6079861111111111,
      "grad_norm": 111.06137084960938,
      "learning_rate": 7.840277777777779e-06,
      "loss": 2.2812,
      "step": 5253
    },
    {
      "epoch": 0.6081018518518518,
      "grad_norm": 0.08311836421489716,
      "learning_rate": 7.837962962962963e-06,
      "loss": 0.0148,
      "step": 5254
    },
    {
      "epoch": 0.6082175925925926,
      "grad_norm": 0.07181457430124283,
      "learning_rate": 7.835648148148147e-06,
      "loss": 0.0129,
      "step": 5255
    },
    {
      "epoch": 0.6083333333333333,
      "grad_norm": 0.5804283022880554,
      "learning_rate": 7.833333333333333e-06,
      "loss": 0.0232,
      "step": 5256
    },
    {
      "epoch": 0.6084490740740741,
      "grad_norm": 0.06947628408670425,
      "learning_rate": 7.83101851851852e-06,
      "loss": 0.0123,
      "step": 5257
    },
    {
      "epoch": 0.6085648148148148,
      "grad_norm": 0.060470059514045715,
      "learning_rate": 7.828703703703704e-06,
      "loss": 0.0111,
      "step": 5258
    },
    {
      "epoch": 0.6086805555555556,
      "grad_norm": 0.09327144175767899,
      "learning_rate": 7.82638888888889e-06,
      "loss": 0.0174,
      "step": 5259
    },
    {
      "epoch": 0.6087962962962963,
      "grad_norm": 0.05986154079437256,
      "learning_rate": 7.824074074074076e-06,
      "loss": 0.011,
      "step": 5260
    },
    {
      "epoch": 0.608912037037037,
      "grad_norm": 0.07182250916957855,
      "learning_rate": 7.82175925925926e-06,
      "loss": 0.0129,
      "step": 5261
    },
    {
      "epoch": 0.6090277777777777,
      "grad_norm": 0.07567574828863144,
      "learning_rate": 7.819444444444446e-06,
      "loss": 0.0141,
      "step": 5262
    },
    {
      "epoch": 0.6091435185185186,
      "grad_norm": 0.06561234593391418,
      "learning_rate": 7.81712962962963e-06,
      "loss": 0.0087,
      "step": 5263
    },
    {
      "epoch": 0.6092592592592593,
      "grad_norm": 0.08014988154172897,
      "learning_rate": 7.814814814814816e-06,
      "loss": 0.0142,
      "step": 5264
    },
    {
      "epoch": 0.609375,
      "grad_norm": 0.05556025728583336,
      "learning_rate": 7.8125e-06,
      "loss": 0.0103,
      "step": 5265
    },
    {
      "epoch": 0.6094907407407407,
      "grad_norm": 0.13138489425182343,
      "learning_rate": 7.810185185185186e-06,
      "loss": 0.0152,
      "step": 5266
    },
    {
      "epoch": 0.6096064814814814,
      "grad_norm": 10.01163387298584,
      "learning_rate": 7.80787037037037e-06,
      "loss": 0.0598,
      "step": 5267
    },
    {
      "epoch": 0.6097222222222223,
      "grad_norm": 0.07749448716640472,
      "learning_rate": 7.805555555555556e-06,
      "loss": 0.0139,
      "step": 5268
    },
    {
      "epoch": 0.609837962962963,
      "grad_norm": 0.07455603033304214,
      "learning_rate": 7.803240740740742e-06,
      "loss": 0.013,
      "step": 5269
    },
    {
      "epoch": 0.6099537037037037,
      "grad_norm": 2.6967828273773193,
      "learning_rate": 7.800925925925926e-06,
      "loss": 0.0287,
      "step": 5270
    },
    {
      "epoch": 0.6100694444444444,
      "grad_norm": 0.07278089225292206,
      "learning_rate": 7.798611111111112e-06,
      "loss": 0.0132,
      "step": 5271
    },
    {
      "epoch": 0.6101851851851852,
      "grad_norm": 0.0683283656835556,
      "learning_rate": 7.796296296296297e-06,
      "loss": 0.0123,
      "step": 5272
    },
    {
      "epoch": 0.6103009259259259,
      "grad_norm": 0.09343715757131577,
      "learning_rate": 7.793981481481483e-06,
      "loss": 0.0173,
      "step": 5273
    },
    {
      "epoch": 0.6104166666666667,
      "grad_norm": 0.1654234081506729,
      "learning_rate": 7.791666666666667e-06,
      "loss": 0.0108,
      "step": 5274
    },
    {
      "epoch": 0.6105324074074074,
      "grad_norm": 0.07503364980220795,
      "learning_rate": 7.789351851851853e-06,
      "loss": 0.0137,
      "step": 5275
    },
    {
      "epoch": 0.6106481481481482,
      "grad_norm": 0.08111165463924408,
      "learning_rate": 7.787037037037037e-06,
      "loss": 0.0146,
      "step": 5276
    },
    {
      "epoch": 0.6107638888888889,
      "grad_norm": 0.4840385913848877,
      "learning_rate": 7.784722222222223e-06,
      "loss": 0.0169,
      "step": 5277
    },
    {
      "epoch": 0.6108796296296296,
      "grad_norm": 0.35807082056999207,
      "learning_rate": 7.782407407407409e-06,
      "loss": 0.0178,
      "step": 5278
    },
    {
      "epoch": 0.6109953703703703,
      "grad_norm": 0.08209634572267532,
      "learning_rate": 7.780092592592593e-06,
      "loss": 0.0151,
      "step": 5279
    },
    {
      "epoch": 0.6111111111111112,
      "grad_norm": 3.1032791137695312,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.0282,
      "step": 5280
    },
    {
      "epoch": 0.6112268518518519,
      "grad_norm": 0.053365059196949005,
      "learning_rate": 7.775462962962963e-06,
      "loss": 0.0097,
      "step": 5281
    },
    {
      "epoch": 0.6113425925925926,
      "grad_norm": 0.08710939437150955,
      "learning_rate": 7.77314814814815e-06,
      "loss": 0.0143,
      "step": 5282
    },
    {
      "epoch": 0.6114583333333333,
      "grad_norm": 0.08731076121330261,
      "learning_rate": 7.770833333333334e-06,
      "loss": 0.0161,
      "step": 5283
    },
    {
      "epoch": 0.611574074074074,
      "grad_norm": 0.07203878462314606,
      "learning_rate": 7.76851851851852e-06,
      "loss": 0.0129,
      "step": 5284
    },
    {
      "epoch": 0.6116898148148148,
      "grad_norm": 0.051320768892765045,
      "learning_rate": 7.766203703703704e-06,
      "loss": 0.0092,
      "step": 5285
    },
    {
      "epoch": 0.6118055555555556,
      "grad_norm": 39.925514221191406,
      "learning_rate": 7.76388888888889e-06,
      "loss": 0.1816,
      "step": 5286
    },
    {
      "epoch": 0.6119212962962963,
      "grad_norm": 0.07907835394144058,
      "learning_rate": 7.761574074074076e-06,
      "loss": 0.0146,
      "step": 5287
    },
    {
      "epoch": 0.612037037037037,
      "grad_norm": 0.07355467230081558,
      "learning_rate": 7.75925925925926e-06,
      "loss": 0.0135,
      "step": 5288
    },
    {
      "epoch": 0.6121527777777778,
      "grad_norm": 0.08621398359537125,
      "learning_rate": 7.756944444444446e-06,
      "loss": 0.0158,
      "step": 5289
    },
    {
      "epoch": 0.6122685185185185,
      "grad_norm": 0.08413472026586533,
      "learning_rate": 7.75462962962963e-06,
      "loss": 0.0112,
      "step": 5290
    },
    {
      "epoch": 0.6123842592592592,
      "grad_norm": 0.07309635728597641,
      "learning_rate": 7.752314814814816e-06,
      "loss": 0.0135,
      "step": 5291
    },
    {
      "epoch": 0.6125,
      "grad_norm": 0.06223541870713234,
      "learning_rate": 7.75e-06,
      "loss": 0.0111,
      "step": 5292
    },
    {
      "epoch": 0.6126157407407408,
      "grad_norm": 0.05408482998609543,
      "learning_rate": 7.747685185185185e-06,
      "loss": 0.0099,
      "step": 5293
    },
    {
      "epoch": 0.6127314814814815,
      "grad_norm": 0.20258860290050507,
      "learning_rate": 7.74537037037037e-06,
      "loss": 0.017,
      "step": 5294
    },
    {
      "epoch": 0.6128472222222222,
      "grad_norm": 0.08322765678167343,
      "learning_rate": 7.743055555555556e-06,
      "loss": 0.0155,
      "step": 5295
    },
    {
      "epoch": 0.6129629629629629,
      "grad_norm": 0.09020829200744629,
      "learning_rate": 7.74074074074074e-06,
      "loss": 0.0149,
      "step": 5296
    },
    {
      "epoch": 0.6130787037037037,
      "grad_norm": 0.049832627177238464,
      "learning_rate": 7.738425925925927e-06,
      "loss": 0.0092,
      "step": 5297
    },
    {
      "epoch": 0.6131944444444445,
      "grad_norm": 0.08990021795034409,
      "learning_rate": 7.736111111111113e-06,
      "loss": 0.0166,
      "step": 5298
    },
    {
      "epoch": 0.6133101851851852,
      "grad_norm": 0.09206229448318481,
      "learning_rate": 7.733796296296297e-06,
      "loss": 0.0113,
      "step": 5299
    },
    {
      "epoch": 0.6134259259259259,
      "grad_norm": 0.06257636845111847,
      "learning_rate": 7.731481481481483e-06,
      "loss": 0.0113,
      "step": 5300
    },
    {
      "epoch": 0.6135416666666667,
      "grad_norm": 40.94144821166992,
      "learning_rate": 7.729166666666667e-06,
      "loss": 0.0955,
      "step": 5301
    },
    {
      "epoch": 0.6136574074074074,
      "grad_norm": 0.06886289268732071,
      "learning_rate": 7.726851851851851e-06,
      "loss": 0.0123,
      "step": 5302
    },
    {
      "epoch": 0.6137731481481481,
      "grad_norm": 30.034849166870117,
      "learning_rate": 7.724537037037037e-06,
      "loss": 2.3689,
      "step": 5303
    },
    {
      "epoch": 0.6138888888888889,
      "grad_norm": 3.1458396911621094,
      "learning_rate": 7.722222222222223e-06,
      "loss": 0.0316,
      "step": 5304
    },
    {
      "epoch": 0.6140046296296297,
      "grad_norm": 0.0776817798614502,
      "learning_rate": 7.719907407407407e-06,
      "loss": 0.0109,
      "step": 5305
    },
    {
      "epoch": 0.6141203703703704,
      "grad_norm": 0.07518685609102249,
      "learning_rate": 7.717592592592593e-06,
      "loss": 0.0134,
      "step": 5306
    },
    {
      "epoch": 0.6142361111111111,
      "grad_norm": 0.07589270174503326,
      "learning_rate": 7.71527777777778e-06,
      "loss": 0.0134,
      "step": 5307
    },
    {
      "epoch": 0.6143518518518518,
      "grad_norm": 0.06309289485216141,
      "learning_rate": 7.712962962962964e-06,
      "loss": 0.0103,
      "step": 5308
    },
    {
      "epoch": 0.6144675925925925,
      "grad_norm": 0.1271905153989792,
      "learning_rate": 7.71064814814815e-06,
      "loss": 0.0108,
      "step": 5309
    },
    {
      "epoch": 0.6145833333333334,
      "grad_norm": 0.051663849502801895,
      "learning_rate": 7.708333333333334e-06,
      "loss": 0.0094,
      "step": 5310
    },
    {
      "epoch": 0.6146990740740741,
      "grad_norm": 0.0698213055729866,
      "learning_rate": 7.706018518518518e-06,
      "loss": 0.0123,
      "step": 5311
    },
    {
      "epoch": 0.6148148148148148,
      "grad_norm": 0.25808700919151306,
      "learning_rate": 7.703703703703704e-06,
      "loss": 0.021,
      "step": 5312
    },
    {
      "epoch": 0.6149305555555555,
      "grad_norm": 0.06452799588441849,
      "learning_rate": 7.70138888888889e-06,
      "loss": 0.0116,
      "step": 5313
    },
    {
      "epoch": 0.6150462962962963,
      "grad_norm": 0.09164370596408844,
      "learning_rate": 7.699074074074074e-06,
      "loss": 0.0171,
      "step": 5314
    },
    {
      "epoch": 0.6151620370370371,
      "grad_norm": 228.70333862304688,
      "learning_rate": 7.69675925925926e-06,
      "loss": 0.5444,
      "step": 5315
    },
    {
      "epoch": 0.6152777777777778,
      "grad_norm": 0.13486525416374207,
      "learning_rate": 7.694444444444446e-06,
      "loss": 0.0182,
      "step": 5316
    },
    {
      "epoch": 0.6153935185185185,
      "grad_norm": 0.09969405829906464,
      "learning_rate": 7.69212962962963e-06,
      "loss": 0.0174,
      "step": 5317
    },
    {
      "epoch": 0.6155092592592593,
      "grad_norm": 0.09319493919610977,
      "learning_rate": 7.689814814814816e-06,
      "loss": 0.0151,
      "step": 5318
    },
    {
      "epoch": 0.615625,
      "grad_norm": 0.07881608605384827,
      "learning_rate": 7.6875e-06,
      "loss": 0.0128,
      "step": 5319
    },
    {
      "epoch": 0.6157407407407407,
      "grad_norm": 0.06498292088508606,
      "learning_rate": 7.685185185185185e-06,
      "loss": 0.0121,
      "step": 5320
    },
    {
      "epoch": 0.6158564814814815,
      "grad_norm": 0.06696918606758118,
      "learning_rate": 7.68287037037037e-06,
      "loss": 0.0121,
      "step": 5321
    },
    {
      "epoch": 0.6159722222222223,
      "grad_norm": 0.0489998534321785,
      "learning_rate": 7.680555555555557e-06,
      "loss": 0.009,
      "step": 5322
    },
    {
      "epoch": 0.616087962962963,
      "grad_norm": 192.67477416992188,
      "learning_rate": 7.678240740740741e-06,
      "loss": 0.4884,
      "step": 5323
    },
    {
      "epoch": 0.6162037037037037,
      "grad_norm": 0.07705892622470856,
      "learning_rate": 7.675925925925927e-06,
      "loss": 0.0099,
      "step": 5324
    },
    {
      "epoch": 0.6163194444444444,
      "grad_norm": 0.06166790425777435,
      "learning_rate": 7.673611111111113e-06,
      "loss": 0.0113,
      "step": 5325
    },
    {
      "epoch": 0.6164351851851851,
      "grad_norm": 0.07819084078073502,
      "learning_rate": 7.671296296296297e-06,
      "loss": 0.0099,
      "step": 5326
    },
    {
      "epoch": 0.616550925925926,
      "grad_norm": 0.07786859571933746,
      "learning_rate": 7.668981481481483e-06,
      "loss": 0.0143,
      "step": 5327
    },
    {
      "epoch": 0.6166666666666667,
      "grad_norm": 0.06935563683509827,
      "learning_rate": 7.666666666666667e-06,
      "loss": 0.0127,
      "step": 5328
    },
    {
      "epoch": 0.6167824074074074,
      "grad_norm": 0.1310204565525055,
      "learning_rate": 7.664351851851851e-06,
      "loss": 0.0176,
      "step": 5329
    },
    {
      "epoch": 0.6168981481481481,
      "grad_norm": 0.1388360857963562,
      "learning_rate": 7.662037037037037e-06,
      "loss": 0.015,
      "step": 5330
    },
    {
      "epoch": 0.6170138888888889,
      "grad_norm": 3.2455596923828125,
      "learning_rate": 7.659722222222223e-06,
      "loss": 0.0288,
      "step": 5331
    },
    {
      "epoch": 0.6171296296296296,
      "grad_norm": 4.102025985717773,
      "learning_rate": 7.657407407407408e-06,
      "loss": 0.0406,
      "step": 5332
    },
    {
      "epoch": 0.6172453703703704,
      "grad_norm": 0.06997858732938766,
      "learning_rate": 7.655092592592594e-06,
      "loss": 0.0125,
      "step": 5333
    },
    {
      "epoch": 0.6173611111111111,
      "grad_norm": 0.08205580711364746,
      "learning_rate": 7.652777777777778e-06,
      "loss": 0.0152,
      "step": 5334
    },
    {
      "epoch": 0.6174768518518519,
      "grad_norm": 0.07791649550199509,
      "learning_rate": 7.650462962962964e-06,
      "loss": 0.0104,
      "step": 5335
    },
    {
      "epoch": 0.6175925925925926,
      "grad_norm": 0.04878618195652962,
      "learning_rate": 7.64814814814815e-06,
      "loss": 0.009,
      "step": 5336
    },
    {
      "epoch": 0.6177083333333333,
      "grad_norm": 0.08699066936969757,
      "learning_rate": 7.645833333333334e-06,
      "loss": 0.0155,
      "step": 5337
    },
    {
      "epoch": 0.617824074074074,
      "grad_norm": 0.07842046767473221,
      "learning_rate": 7.643518518518518e-06,
      "loss": 0.0136,
      "step": 5338
    },
    {
      "epoch": 0.6179398148148149,
      "grad_norm": 0.06515196710824966,
      "learning_rate": 7.641203703703704e-06,
      "loss": 0.0117,
      "step": 5339
    },
    {
      "epoch": 0.6180555555555556,
      "grad_norm": 0.07097096741199493,
      "learning_rate": 7.638888888888888e-06,
      "loss": 0.0131,
      "step": 5340
    },
    {
      "epoch": 0.6181712962962963,
      "grad_norm": 14.953596115112305,
      "learning_rate": 7.636574074074074e-06,
      "loss": 2.9211,
      "step": 5341
    },
    {
      "epoch": 0.618287037037037,
      "grad_norm": 0.08703351020812988,
      "learning_rate": 7.63425925925926e-06,
      "loss": 0.0152,
      "step": 5342
    },
    {
      "epoch": 0.6184027777777777,
      "grad_norm": 0.5423064231872559,
      "learning_rate": 7.631944444444445e-06,
      "loss": 0.0178,
      "step": 5343
    },
    {
      "epoch": 0.6185185185185185,
      "grad_norm": 0.277109831571579,
      "learning_rate": 7.62962962962963e-06,
      "loss": 0.0196,
      "step": 5344
    },
    {
      "epoch": 0.6186342592592593,
      "grad_norm": 21.123966217041016,
      "learning_rate": 7.627314814814816e-06,
      "loss": 0.072,
      "step": 5345
    },
    {
      "epoch": 0.61875,
      "grad_norm": 3.7144923210144043,
      "learning_rate": 7.625e-06,
      "loss": 0.0445,
      "step": 5346
    },
    {
      "epoch": 0.6188657407407407,
      "grad_norm": 0.10370410978794098,
      "learning_rate": 7.622685185185186e-06,
      "loss": 0.0134,
      "step": 5347
    },
    {
      "epoch": 0.6189814814814815,
      "grad_norm": 0.0898149386048317,
      "learning_rate": 7.620370370370372e-06,
      "loss": 0.0165,
      "step": 5348
    },
    {
      "epoch": 0.6190972222222222,
      "grad_norm": 53.77619552612305,
      "learning_rate": 7.618055555555556e-06,
      "loss": 0.1743,
      "step": 5349
    },
    {
      "epoch": 0.6192129629629629,
      "grad_norm": 0.08212178200483322,
      "learning_rate": 7.615740740740741e-06,
      "loss": 0.015,
      "step": 5350
    },
    {
      "epoch": 0.6193287037037037,
      "grad_norm": 0.06874743103981018,
      "learning_rate": 7.613425925925927e-06,
      "loss": 0.0123,
      "step": 5351
    },
    {
      "epoch": 0.6194444444444445,
      "grad_norm": 0.06967861950397491,
      "learning_rate": 7.611111111111111e-06,
      "loss": 0.0121,
      "step": 5352
    },
    {
      "epoch": 0.6195601851851852,
      "grad_norm": 0.06810086965560913,
      "learning_rate": 7.608796296296297e-06,
      "loss": 0.012,
      "step": 5353
    },
    {
      "epoch": 0.6196759259259259,
      "grad_norm": 0.07081454992294312,
      "learning_rate": 7.606481481481482e-06,
      "loss": 0.0129,
      "step": 5354
    },
    {
      "epoch": 0.6197916666666666,
      "grad_norm": 0.05938326567411423,
      "learning_rate": 7.6041666666666666e-06,
      "loss": 0.0106,
      "step": 5355
    },
    {
      "epoch": 0.6199074074074075,
      "grad_norm": 0.055050142109394073,
      "learning_rate": 7.6018518518518525e-06,
      "loss": 0.0094,
      "step": 5356
    },
    {
      "epoch": 0.6200231481481482,
      "grad_norm": 93.35578155517578,
      "learning_rate": 7.5995370370370385e-06,
      "loss": 2.3842,
      "step": 5357
    },
    {
      "epoch": 0.6201388888888889,
      "grad_norm": 10.616626739501953,
      "learning_rate": 7.597222222222223e-06,
      "loss": 0.0397,
      "step": 5358
    },
    {
      "epoch": 0.6202546296296296,
      "grad_norm": 0.07736767083406448,
      "learning_rate": 7.594907407407408e-06,
      "loss": 0.014,
      "step": 5359
    },
    {
      "epoch": 0.6203703703703703,
      "grad_norm": 0.16156446933746338,
      "learning_rate": 7.592592592592594e-06,
      "loss": 0.0164,
      "step": 5360
    },
    {
      "epoch": 0.6204861111111111,
      "grad_norm": 0.05634152144193649,
      "learning_rate": 7.590277777777778e-06,
      "loss": 0.0103,
      "step": 5361
    },
    {
      "epoch": 0.6206018518518519,
      "grad_norm": 0.07945938408374786,
      "learning_rate": 7.587962962962964e-06,
      "loss": 0.014,
      "step": 5362
    },
    {
      "epoch": 0.6207175925925926,
      "grad_norm": 0.057122934609651566,
      "learning_rate": 7.585648148148149e-06,
      "loss": 0.0096,
      "step": 5363
    },
    {
      "epoch": 0.6208333333333333,
      "grad_norm": 0.06432867050170898,
      "learning_rate": 7.583333333333333e-06,
      "loss": 0.0119,
      "step": 5364
    },
    {
      "epoch": 0.6209490740740741,
      "grad_norm": 5.566343307495117,
      "learning_rate": 7.581018518518519e-06,
      "loss": 0.0422,
      "step": 5365
    },
    {
      "epoch": 0.6210648148148148,
      "grad_norm": 185.2796630859375,
      "learning_rate": 7.578703703703705e-06,
      "loss": 0.7495,
      "step": 5366
    },
    {
      "epoch": 0.6211805555555555,
      "grad_norm": 0.21995419263839722,
      "learning_rate": 7.5763888888888894e-06,
      "loss": 0.0137,
      "step": 5367
    },
    {
      "epoch": 0.6212962962962963,
      "grad_norm": 0.06755710393190384,
      "learning_rate": 7.5740740740740745e-06,
      "loss": 0.0108,
      "step": 5368
    },
    {
      "epoch": 0.6214120370370371,
      "grad_norm": 2.1908905506134033,
      "learning_rate": 7.5717592592592605e-06,
      "loss": 0.0268,
      "step": 5369
    },
    {
      "epoch": 0.6215277777777778,
      "grad_norm": 0.11077011376619339,
      "learning_rate": 7.569444444444445e-06,
      "loss": 0.0146,
      "step": 5370
    },
    {
      "epoch": 0.6216435185185185,
      "grad_norm": 0.07932329922914505,
      "learning_rate": 7.567129629629631e-06,
      "loss": 0.0137,
      "step": 5371
    },
    {
      "epoch": 0.6217592592592592,
      "grad_norm": 0.11618097871541977,
      "learning_rate": 7.564814814814816e-06,
      "loss": 0.0154,
      "step": 5372
    },
    {
      "epoch": 0.621875,
      "grad_norm": 15.319809913635254,
      "learning_rate": 7.5625e-06,
      "loss": 2.8302,
      "step": 5373
    },
    {
      "epoch": 0.6219907407407408,
      "grad_norm": 0.11090908199548721,
      "learning_rate": 7.560185185185186e-06,
      "loss": 0.0148,
      "step": 5374
    },
    {
      "epoch": 0.6221064814814815,
      "grad_norm": 139.158935546875,
      "learning_rate": 7.557870370370372e-06,
      "loss": 1.273,
      "step": 5375
    },
    {
      "epoch": 0.6222222222222222,
      "grad_norm": 0.0847323089838028,
      "learning_rate": 7.555555555555556e-06,
      "loss": 0.0148,
      "step": 5376
    },
    {
      "epoch": 0.622337962962963,
      "grad_norm": 0.15913881361484528,
      "learning_rate": 7.553240740740741e-06,
      "loss": 0.01,
      "step": 5377
    },
    {
      "epoch": 0.6224537037037037,
      "grad_norm": 0.06599907577037811,
      "learning_rate": 7.550925925925926e-06,
      "loss": 0.0119,
      "step": 5378
    },
    {
      "epoch": 0.6225694444444444,
      "grad_norm": 0.07446452230215073,
      "learning_rate": 7.5486111111111114e-06,
      "loss": 0.0133,
      "step": 5379
    },
    {
      "epoch": 0.6226851851851852,
      "grad_norm": 0.07053598761558533,
      "learning_rate": 7.546296296296297e-06,
      "loss": 0.0127,
      "step": 5380
    },
    {
      "epoch": 0.622800925925926,
      "grad_norm": 0.07573758065700531,
      "learning_rate": 7.543981481481482e-06,
      "loss": 0.013,
      "step": 5381
    },
    {
      "epoch": 0.6229166666666667,
      "grad_norm": 0.06996962428092957,
      "learning_rate": 7.541666666666667e-06,
      "loss": 0.0128,
      "step": 5382
    },
    {
      "epoch": 0.6230324074074074,
      "grad_norm": 82.83528900146484,
      "learning_rate": 7.539351851851853e-06,
      "loss": 0.6658,
      "step": 5383
    },
    {
      "epoch": 0.6231481481481481,
      "grad_norm": 95.97897338867188,
      "learning_rate": 7.537037037037037e-06,
      "loss": 2.0588,
      "step": 5384
    },
    {
      "epoch": 0.6232638888888888,
      "grad_norm": 0.060441967099905014,
      "learning_rate": 7.534722222222223e-06,
      "loss": 0.0111,
      "step": 5385
    },
    {
      "epoch": 0.6233796296296297,
      "grad_norm": 0.17208433151245117,
      "learning_rate": 7.532407407407408e-06,
      "loss": 0.0152,
      "step": 5386
    },
    {
      "epoch": 0.6234953703703704,
      "grad_norm": 14.025711059570312,
      "learning_rate": 7.530092592592593e-06,
      "loss": 3.2482,
      "step": 5387
    },
    {
      "epoch": 0.6236111111111111,
      "grad_norm": 0.05230384320020676,
      "learning_rate": 7.527777777777778e-06,
      "loss": 0.0094,
      "step": 5388
    },
    {
      "epoch": 0.6237268518518518,
      "grad_norm": 0.06094253063201904,
      "learning_rate": 7.525462962962964e-06,
      "loss": 0.0106,
      "step": 5389
    },
    {
      "epoch": 0.6238425925925926,
      "grad_norm": 0.05812288448214531,
      "learning_rate": 7.523148148148148e-06,
      "loss": 0.0096,
      "step": 5390
    },
    {
      "epoch": 0.6239583333333333,
      "grad_norm": 0.11651875823736191,
      "learning_rate": 7.5208333333333335e-06,
      "loss": 0.0153,
      "step": 5391
    },
    {
      "epoch": 0.6240740740740741,
      "grad_norm": 0.078697569668293,
      "learning_rate": 7.518518518518519e-06,
      "loss": 0.0144,
      "step": 5392
    },
    {
      "epoch": 0.6241898148148148,
      "grad_norm": 0.05200834572315216,
      "learning_rate": 7.516203703703704e-06,
      "loss": 0.0091,
      "step": 5393
    },
    {
      "epoch": 0.6243055555555556,
      "grad_norm": 0.05956161767244339,
      "learning_rate": 7.51388888888889e-06,
      "loss": 0.0108,
      "step": 5394
    },
    {
      "epoch": 0.6244212962962963,
      "grad_norm": 0.07580207288265228,
      "learning_rate": 7.511574074074075e-06,
      "loss": 0.014,
      "step": 5395
    },
    {
      "epoch": 0.624537037037037,
      "grad_norm": 0.09314297139644623,
      "learning_rate": 7.50925925925926e-06,
      "loss": 0.0117,
      "step": 5396
    },
    {
      "epoch": 0.6246527777777777,
      "grad_norm": 0.08303902298212051,
      "learning_rate": 7.506944444444445e-06,
      "loss": 0.0145,
      "step": 5397
    },
    {
      "epoch": 0.6247685185185186,
      "grad_norm": 0.072955422103405,
      "learning_rate": 7.504629629629631e-06,
      "loss": 0.0129,
      "step": 5398
    },
    {
      "epoch": 0.6248842592592593,
      "grad_norm": 0.0690719336271286,
      "learning_rate": 7.502314814814815e-06,
      "loss": 0.0121,
      "step": 5399
    },
    {
      "epoch": 0.625,
      "grad_norm": 132.4497833251953,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.3093,
      "step": 5400
    },
    {
      "epoch": 0.6251157407407407,
      "grad_norm": 0.07245708256959915,
      "learning_rate": 7.497685185185186e-06,
      "loss": 0.0131,
      "step": 5401
    },
    {
      "epoch": 0.6252314814814814,
      "grad_norm": 0.08180329948663712,
      "learning_rate": 7.49537037037037e-06,
      "loss": 0.0136,
      "step": 5402
    },
    {
      "epoch": 0.6253472222222223,
      "grad_norm": 0.07656219601631165,
      "learning_rate": 7.493055555555556e-06,
      "loss": 0.0101,
      "step": 5403
    },
    {
      "epoch": 0.625462962962963,
      "grad_norm": 0.06853035092353821,
      "learning_rate": 7.4907407407407414e-06,
      "loss": 0.0119,
      "step": 5404
    },
    {
      "epoch": 0.6255787037037037,
      "grad_norm": 0.7763587236404419,
      "learning_rate": 7.4884259259259265e-06,
      "loss": 0.0143,
      "step": 5405
    },
    {
      "epoch": 0.6256944444444444,
      "grad_norm": 0.08867047727108002,
      "learning_rate": 7.486111111111112e-06,
      "loss": 0.0158,
      "step": 5406
    },
    {
      "epoch": 0.6258101851851852,
      "grad_norm": 0.07577456533908844,
      "learning_rate": 7.4837962962962976e-06,
      "loss": 0.0131,
      "step": 5407
    },
    {
      "epoch": 0.6259259259259259,
      "grad_norm": 0.07057604193687439,
      "learning_rate": 7.481481481481482e-06,
      "loss": 0.0127,
      "step": 5408
    },
    {
      "epoch": 0.6260416666666667,
      "grad_norm": 0.09363574534654617,
      "learning_rate": 7.479166666666668e-06,
      "loss": 0.0101,
      "step": 5409
    },
    {
      "epoch": 0.6261574074074074,
      "grad_norm": 0.08201594650745392,
      "learning_rate": 7.476851851851853e-06,
      "loss": 0.015,
      "step": 5410
    },
    {
      "epoch": 0.6262731481481482,
      "grad_norm": 0.05273859202861786,
      "learning_rate": 7.474537037037037e-06,
      "loss": 0.0097,
      "step": 5411
    },
    {
      "epoch": 0.6263888888888889,
      "grad_norm": 0.13467587530612946,
      "learning_rate": 7.472222222222223e-06,
      "loss": 0.018,
      "step": 5412
    },
    {
      "epoch": 0.6265046296296296,
      "grad_norm": 0.07739211618900299,
      "learning_rate": 7.469907407407408e-06,
      "loss": 0.0136,
      "step": 5413
    },
    {
      "epoch": 0.6266203703703703,
      "grad_norm": 0.05384138226509094,
      "learning_rate": 7.467592592592593e-06,
      "loss": 0.0097,
      "step": 5414
    },
    {
      "epoch": 0.6267361111111112,
      "grad_norm": 0.04967760667204857,
      "learning_rate": 7.465277777777778e-06,
      "loss": 0.009,
      "step": 5415
    },
    {
      "epoch": 0.6268518518518519,
      "grad_norm": 0.052409783005714417,
      "learning_rate": 7.462962962962964e-06,
      "loss": 0.0096,
      "step": 5416
    },
    {
      "epoch": 0.6269675925925926,
      "grad_norm": 23.480257034301758,
      "learning_rate": 7.4606481481481485e-06,
      "loss": 0.0898,
      "step": 5417
    },
    {
      "epoch": 0.6270833333333333,
      "grad_norm": 0.13084755837917328,
      "learning_rate": 7.4583333333333345e-06,
      "loss": 0.0157,
      "step": 5418
    },
    {
      "epoch": 0.627199074074074,
      "grad_norm": 0.0847916230559349,
      "learning_rate": 7.45601851851852e-06,
      "loss": 0.0142,
      "step": 5419
    },
    {
      "epoch": 0.6273148148148148,
      "grad_norm": 0.07752954959869385,
      "learning_rate": 7.453703703703704e-06,
      "loss": 0.014,
      "step": 5420
    },
    {
      "epoch": 0.6274305555555556,
      "grad_norm": 0.06741341948509216,
      "learning_rate": 7.45138888888889e-06,
      "loss": 0.0121,
      "step": 5421
    },
    {
      "epoch": 0.6275462962962963,
      "grad_norm": 0.055644482374191284,
      "learning_rate": 7.449074074074075e-06,
      "loss": 0.0094,
      "step": 5422
    },
    {
      "epoch": 0.627662037037037,
      "grad_norm": 0.06026425212621689,
      "learning_rate": 7.44675925925926e-06,
      "loss": 0.0101,
      "step": 5423
    },
    {
      "epoch": 0.6277777777777778,
      "grad_norm": 0.10053565353155136,
      "learning_rate": 7.444444444444445e-06,
      "loss": 0.014,
      "step": 5424
    },
    {
      "epoch": 0.6278935185185185,
      "grad_norm": 0.06362728774547577,
      "learning_rate": 7.442129629629629e-06,
      "loss": 0.0118,
      "step": 5425
    },
    {
      "epoch": 0.6280092592592592,
      "grad_norm": 0.072262704372406,
      "learning_rate": 7.439814814814815e-06,
      "loss": 0.0123,
      "step": 5426
    },
    {
      "epoch": 0.628125,
      "grad_norm": 0.2494249790906906,
      "learning_rate": 7.437500000000001e-06,
      "loss": 0.0208,
      "step": 5427
    },
    {
      "epoch": 0.6282407407407408,
      "grad_norm": 0.06409275531768799,
      "learning_rate": 7.4351851851851855e-06,
      "loss": 0.0115,
      "step": 5428
    },
    {
      "epoch": 0.6283564814814815,
      "grad_norm": 0.06946749240159988,
      "learning_rate": 7.4328703703703706e-06,
      "loss": 0.0123,
      "step": 5429
    },
    {
      "epoch": 0.6284722222222222,
      "grad_norm": 0.07717336714267731,
      "learning_rate": 7.4305555555555565e-06,
      "loss": 0.0141,
      "step": 5430
    },
    {
      "epoch": 0.6285879629629629,
      "grad_norm": 0.0664091482758522,
      "learning_rate": 7.428240740740741e-06,
      "loss": 0.0121,
      "step": 5431
    },
    {
      "epoch": 0.6287037037037037,
      "grad_norm": 0.07106873393058777,
      "learning_rate": 7.425925925925927e-06,
      "loss": 0.0122,
      "step": 5432
    },
    {
      "epoch": 0.6288194444444445,
      "grad_norm": 0.07454588264226913,
      "learning_rate": 7.423611111111112e-06,
      "loss": 0.0129,
      "step": 5433
    },
    {
      "epoch": 0.6289351851851852,
      "grad_norm": 0.07329779863357544,
      "learning_rate": 7.421296296296296e-06,
      "loss": 0.0136,
      "step": 5434
    },
    {
      "epoch": 0.6290509259259259,
      "grad_norm": 0.08120168745517731,
      "learning_rate": 7.418981481481482e-06,
      "loss": 0.0128,
      "step": 5435
    },
    {
      "epoch": 0.6291666666666667,
      "grad_norm": 0.07591758668422699,
      "learning_rate": 7.416666666666668e-06,
      "loss": 0.0136,
      "step": 5436
    },
    {
      "epoch": 0.6292824074074074,
      "grad_norm": 0.07729983329772949,
      "learning_rate": 7.414351851851852e-06,
      "loss": 0.0142,
      "step": 5437
    },
    {
      "epoch": 0.6293981481481481,
      "grad_norm": 0.05769734084606171,
      "learning_rate": 7.412037037037037e-06,
      "loss": 0.0105,
      "step": 5438
    },
    {
      "epoch": 0.6295138888888889,
      "grad_norm": 115.94496154785156,
      "learning_rate": 7.409722222222223e-06,
      "loss": 1.9168,
      "step": 5439
    },
    {
      "epoch": 0.6296296296296297,
      "grad_norm": 0.08294026553630829,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 0.0148,
      "step": 5440
    },
    {
      "epoch": 0.6297453703703704,
      "grad_norm": 0.06674454361200333,
      "learning_rate": 7.4050925925925934e-06,
      "loss": 0.0121,
      "step": 5441
    },
    {
      "epoch": 0.6298611111111111,
      "grad_norm": 0.054901134222745895,
      "learning_rate": 7.4027777777777785e-06,
      "loss": 0.0094,
      "step": 5442
    },
    {
      "epoch": 0.6299768518518518,
      "grad_norm": 0.0758921429514885,
      "learning_rate": 7.400462962962963e-06,
      "loss": 0.0099,
      "step": 5443
    },
    {
      "epoch": 0.6300925925925925,
      "grad_norm": 0.0535787008702755,
      "learning_rate": 7.398148148148149e-06,
      "loss": 0.0098,
      "step": 5444
    },
    {
      "epoch": 0.6302083333333334,
      "grad_norm": 0.07650629431009293,
      "learning_rate": 7.395833333333335e-06,
      "loss": 0.0126,
      "step": 5445
    },
    {
      "epoch": 0.6303240740740741,
      "grad_norm": 0.07254374027252197,
      "learning_rate": 7.393518518518519e-06,
      "loss": 0.012,
      "step": 5446
    },
    {
      "epoch": 0.6304398148148148,
      "grad_norm": 0.05656001344323158,
      "learning_rate": 7.391203703703704e-06,
      "loss": 0.0103,
      "step": 5447
    },
    {
      "epoch": 0.6305555555555555,
      "grad_norm": 0.06045514717698097,
      "learning_rate": 7.38888888888889e-06,
      "loss": 0.0107,
      "step": 5448
    },
    {
      "epoch": 0.6306712962962963,
      "grad_norm": 0.061720769852399826,
      "learning_rate": 7.386574074074074e-06,
      "loss": 0.0111,
      "step": 5449
    },
    {
      "epoch": 0.6307870370370371,
      "grad_norm": 0.07238173484802246,
      "learning_rate": 7.38425925925926e-06,
      "loss": 0.0134,
      "step": 5450
    },
    {
      "epoch": 0.6309027777777778,
      "grad_norm": 0.0503377839922905,
      "learning_rate": 7.381944444444445e-06,
      "loss": 0.0092,
      "step": 5451
    },
    {
      "epoch": 0.6310185185185185,
      "grad_norm": 0.06516961753368378,
      "learning_rate": 7.3796296296296295e-06,
      "loss": 0.0116,
      "step": 5452
    },
    {
      "epoch": 0.6311342592592593,
      "grad_norm": 0.07674767822027206,
      "learning_rate": 7.3773148148148154e-06,
      "loss": 0.0142,
      "step": 5453
    },
    {
      "epoch": 0.63125,
      "grad_norm": 0.11002587527036667,
      "learning_rate": 7.375000000000001e-06,
      "loss": 0.013,
      "step": 5454
    },
    {
      "epoch": 0.6313657407407407,
      "grad_norm": 0.07829249650239944,
      "learning_rate": 7.372685185185186e-06,
      "loss": 0.0145,
      "step": 5455
    },
    {
      "epoch": 0.6314814814814815,
      "grad_norm": 0.051389697939157486,
      "learning_rate": 7.370370370370371e-06,
      "loss": 0.0094,
      "step": 5456
    },
    {
      "epoch": 0.6315972222222223,
      "grad_norm": 0.05238255858421326,
      "learning_rate": 7.368055555555557e-06,
      "loss": 0.0091,
      "step": 5457
    },
    {
      "epoch": 0.631712962962963,
      "grad_norm": 0.09517695009708405,
      "learning_rate": 7.365740740740741e-06,
      "loss": 0.0126,
      "step": 5458
    },
    {
      "epoch": 0.6318287037037037,
      "grad_norm": 0.07302561402320862,
      "learning_rate": 7.363425925925927e-06,
      "loss": 0.0135,
      "step": 5459
    },
    {
      "epoch": 0.6319444444444444,
      "grad_norm": 0.08535725623369217,
      "learning_rate": 7.361111111111112e-06,
      "loss": 0.0125,
      "step": 5460
    },
    {
      "epoch": 0.6320601851851851,
      "grad_norm": 0.06443386524915695,
      "learning_rate": 7.358796296296297e-06,
      "loss": 0.0115,
      "step": 5461
    },
    {
      "epoch": 0.632175925925926,
      "grad_norm": 122.9947738647461,
      "learning_rate": 7.356481481481482e-06,
      "loss": 0.8434,
      "step": 5462
    },
    {
      "epoch": 0.6322916666666667,
      "grad_norm": 0.05334276333451271,
      "learning_rate": 7.354166666666668e-06,
      "loss": 0.0098,
      "step": 5463
    },
    {
      "epoch": 0.6324074074074074,
      "grad_norm": 0.055003151297569275,
      "learning_rate": 7.351851851851852e-06,
      "loss": 0.01,
      "step": 5464
    },
    {
      "epoch": 0.6325231481481481,
      "grad_norm": 209.58123779296875,
      "learning_rate": 7.3495370370370375e-06,
      "loss": 4.3984,
      "step": 5465
    },
    {
      "epoch": 0.6326388888888889,
      "grad_norm": 0.06610249727964401,
      "learning_rate": 7.347222222222223e-06,
      "loss": 0.0118,
      "step": 5466
    },
    {
      "epoch": 0.6327546296296296,
      "grad_norm": 0.0701979249715805,
      "learning_rate": 7.344907407407408e-06,
      "loss": 0.0125,
      "step": 5467
    },
    {
      "epoch": 0.6328703703703704,
      "grad_norm": 0.07408131659030914,
      "learning_rate": 7.342592592592594e-06,
      "loss": 0.0135,
      "step": 5468
    },
    {
      "epoch": 0.6329861111111111,
      "grad_norm": 0.12664054334163666,
      "learning_rate": 7.340277777777778e-06,
      "loss": 0.0171,
      "step": 5469
    },
    {
      "epoch": 0.6331018518518519,
      "grad_norm": 0.3886619508266449,
      "learning_rate": 7.337962962962964e-06,
      "loss": 0.0148,
      "step": 5470
    },
    {
      "epoch": 0.6332175925925926,
      "grad_norm": 53.9084587097168,
      "learning_rate": 7.335648148148149e-06,
      "loss": 0.0828,
      "step": 5471
    },
    {
      "epoch": 0.6333333333333333,
      "grad_norm": 7.5817484855651855,
      "learning_rate": 7.333333333333333e-06,
      "loss": 3.0736,
      "step": 5472
    },
    {
      "epoch": 0.633449074074074,
      "grad_norm": 0.06885389983654022,
      "learning_rate": 7.331018518518519e-06,
      "loss": 0.0122,
      "step": 5473
    },
    {
      "epoch": 0.6335648148148149,
      "grad_norm": 0.051670607179403305,
      "learning_rate": 7.328703703703704e-06,
      "loss": 0.0091,
      "step": 5474
    },
    {
      "epoch": 0.6336805555555556,
      "grad_norm": 0.06554612517356873,
      "learning_rate": 7.326388888888889e-06,
      "loss": 0.0118,
      "step": 5475
    },
    {
      "epoch": 0.6337962962962963,
      "grad_norm": 0.044967170804739,
      "learning_rate": 7.324074074074074e-06,
      "loss": 0.0083,
      "step": 5476
    },
    {
      "epoch": 0.633912037037037,
      "grad_norm": 0.6863612532615662,
      "learning_rate": 7.32175925925926e-06,
      "loss": 0.0215,
      "step": 5477
    },
    {
      "epoch": 0.6340277777777777,
      "grad_norm": 0.052965663373470306,
      "learning_rate": 7.3194444444444446e-06,
      "loss": 0.0094,
      "step": 5478
    },
    {
      "epoch": 0.6341435185185185,
      "grad_norm": 0.0811517983675003,
      "learning_rate": 7.3171296296296305e-06,
      "loss": 0.0148,
      "step": 5479
    },
    {
      "epoch": 0.6342592592592593,
      "grad_norm": 0.06609246134757996,
      "learning_rate": 7.314814814814816e-06,
      "loss": 0.0118,
      "step": 5480
    },
    {
      "epoch": 0.634375,
      "grad_norm": 0.09891960024833679,
      "learning_rate": 7.3125e-06,
      "loss": 0.0162,
      "step": 5481
    },
    {
      "epoch": 0.6344907407407407,
      "grad_norm": 0.09704141318798065,
      "learning_rate": 7.310185185185186e-06,
      "loss": 0.0134,
      "step": 5482
    },
    {
      "epoch": 0.6346064814814815,
      "grad_norm": 0.08219286054372787,
      "learning_rate": 7.307870370370371e-06,
      "loss": 0.015,
      "step": 5483
    },
    {
      "epoch": 0.6347222222222222,
      "grad_norm": 0.06436861306428909,
      "learning_rate": 7.305555555555556e-06,
      "loss": 0.0113,
      "step": 5484
    },
    {
      "epoch": 0.6348379629629629,
      "grad_norm": 136.6563262939453,
      "learning_rate": 7.303240740740741e-06,
      "loss": 1.5837,
      "step": 5485
    },
    {
      "epoch": 0.6349537037037037,
      "grad_norm": 0.08034159988164902,
      "learning_rate": 7.300925925925927e-06,
      "loss": 0.0146,
      "step": 5486
    },
    {
      "epoch": 0.6350694444444445,
      "grad_norm": 0.05319271981716156,
      "learning_rate": 7.298611111111111e-06,
      "loss": 0.0098,
      "step": 5487
    },
    {
      "epoch": 0.6351851851851852,
      "grad_norm": 0.1736997663974762,
      "learning_rate": 7.296296296296297e-06,
      "loss": 0.0178,
      "step": 5488
    },
    {
      "epoch": 0.6353009259259259,
      "grad_norm": 0.06593440473079681,
      "learning_rate": 7.293981481481482e-06,
      "loss": 0.0119,
      "step": 5489
    },
    {
      "epoch": 0.6354166666666666,
      "grad_norm": 0.0696755051612854,
      "learning_rate": 7.291666666666667e-06,
      "loss": 0.009,
      "step": 5490
    },
    {
      "epoch": 0.6355324074074075,
      "grad_norm": 0.06261102855205536,
      "learning_rate": 7.2893518518518525e-06,
      "loss": 0.0116,
      "step": 5491
    },
    {
      "epoch": 0.6356481481481482,
      "grad_norm": 16.102678298950195,
      "learning_rate": 7.287037037037038e-06,
      "loss": 2.4835,
      "step": 5492
    },
    {
      "epoch": 0.6357638888888889,
      "grad_norm": 0.07484044879674911,
      "learning_rate": 7.284722222222223e-06,
      "loss": 0.0137,
      "step": 5493
    },
    {
      "epoch": 0.6358796296296296,
      "grad_norm": 0.07147978991270065,
      "learning_rate": 7.282407407407408e-06,
      "loss": 0.0126,
      "step": 5494
    },
    {
      "epoch": 0.6359953703703703,
      "grad_norm": 0.06485084444284439,
      "learning_rate": 7.280092592592594e-06,
      "loss": 0.0111,
      "step": 5495
    },
    {
      "epoch": 0.6361111111111111,
      "grad_norm": 0.07436507940292358,
      "learning_rate": 7.277777777777778e-06,
      "loss": 0.0118,
      "step": 5496
    },
    {
      "epoch": 0.6362268518518519,
      "grad_norm": 94.999755859375,
      "learning_rate": 7.275462962962964e-06,
      "loss": 2.2585,
      "step": 5497
    },
    {
      "epoch": 0.6363425925925926,
      "grad_norm": 27.16388702392578,
      "learning_rate": 7.273148148148149e-06,
      "loss": 2.4629,
      "step": 5498
    },
    {
      "epoch": 0.6364583333333333,
      "grad_norm": 0.09162833541631699,
      "learning_rate": 7.270833333333333e-06,
      "loss": 0.0129,
      "step": 5499
    },
    {
      "epoch": 0.6365740740740741,
      "grad_norm": 176.297119140625,
      "learning_rate": 7.268518518518519e-06,
      "loss": 0.897,
      "step": 5500
    },
    {
      "epoch": 0.6366898148148148,
      "grad_norm": 0.06111333891749382,
      "learning_rate": 7.266203703703704e-06,
      "loss": 0.0096,
      "step": 5501
    },
    {
      "epoch": 0.6368055555555555,
      "grad_norm": 7.9534196853637695,
      "learning_rate": 7.2638888888888895e-06,
      "loss": 0.0362,
      "step": 5502
    },
    {
      "epoch": 0.6369212962962963,
      "grad_norm": 0.08696703612804413,
      "learning_rate": 7.2615740740740746e-06,
      "loss": 0.0094,
      "step": 5503
    },
    {
      "epoch": 0.6370370370370371,
      "grad_norm": 0.5106323957443237,
      "learning_rate": 7.2592592592592605e-06,
      "loss": 0.0184,
      "step": 5504
    },
    {
      "epoch": 0.6371527777777778,
      "grad_norm": 0.08047544211149216,
      "learning_rate": 7.256944444444445e-06,
      "loss": 0.0148,
      "step": 5505
    },
    {
      "epoch": 0.6372685185185185,
      "grad_norm": 0.045819561928510666,
      "learning_rate": 7.254629629629631e-06,
      "loss": 0.0084,
      "step": 5506
    },
    {
      "epoch": 0.6373842592592592,
      "grad_norm": 0.06589845567941666,
      "learning_rate": 7.252314814814816e-06,
      "loss": 0.0087,
      "step": 5507
    },
    {
      "epoch": 0.6375,
      "grad_norm": 0.06584306806325912,
      "learning_rate": 7.25e-06,
      "loss": 0.0087,
      "step": 5508
    },
    {
      "epoch": 0.6376157407407408,
      "grad_norm": 0.052804045379161835,
      "learning_rate": 7.247685185185186e-06,
      "loss": 0.0096,
      "step": 5509
    },
    {
      "epoch": 0.6377314814814815,
      "grad_norm": 0.7693362236022949,
      "learning_rate": 7.245370370370371e-06,
      "loss": 0.0193,
      "step": 5510
    },
    {
      "epoch": 0.6378472222222222,
      "grad_norm": 0.059271737933158875,
      "learning_rate": 7.243055555555556e-06,
      "loss": 0.0089,
      "step": 5511
    },
    {
      "epoch": 0.637962962962963,
      "grad_norm": 0.0711899921298027,
      "learning_rate": 7.240740740740741e-06,
      "loss": 0.0125,
      "step": 5512
    },
    {
      "epoch": 0.6380787037037037,
      "grad_norm": 0.0554809495806694,
      "learning_rate": 7.238425925925926e-06,
      "loss": 0.0092,
      "step": 5513
    },
    {
      "epoch": 0.6381944444444444,
      "grad_norm": 0.11201019585132599,
      "learning_rate": 7.2361111111111115e-06,
      "loss": 0.0111,
      "step": 5514
    },
    {
      "epoch": 0.6383101851851852,
      "grad_norm": 0.06998956948518753,
      "learning_rate": 7.233796296296297e-06,
      "loss": 0.0124,
      "step": 5515
    },
    {
      "epoch": 0.638425925925926,
      "grad_norm": 0.06836527585983276,
      "learning_rate": 7.231481481481482e-06,
      "loss": 0.012,
      "step": 5516
    },
    {
      "epoch": 0.6385416666666667,
      "grad_norm": 0.046402789652347565,
      "learning_rate": 7.229166666666667e-06,
      "loss": 0.0085,
      "step": 5517
    },
    {
      "epoch": 0.6386574074074074,
      "grad_norm": 0.08729822933673859,
      "learning_rate": 7.226851851851853e-06,
      "loss": 0.0161,
      "step": 5518
    },
    {
      "epoch": 0.6387731481481481,
      "grad_norm": 0.06368172913789749,
      "learning_rate": 7.224537037037037e-06,
      "loss": 0.0115,
      "step": 5519
    },
    {
      "epoch": 0.6388888888888888,
      "grad_norm": 0.07493308931589127,
      "learning_rate": 7.222222222222223e-06,
      "loss": 0.0099,
      "step": 5520
    },
    {
      "epoch": 0.6390046296296297,
      "grad_norm": 10.93271541595459,
      "learning_rate": 7.219907407407408e-06,
      "loss": 0.0475,
      "step": 5521
    },
    {
      "epoch": 0.6391203703703704,
      "grad_norm": 0.08452604711055756,
      "learning_rate": 7.217592592592593e-06,
      "loss": 0.0155,
      "step": 5522
    },
    {
      "epoch": 0.6392361111111111,
      "grad_norm": 0.34009283781051636,
      "learning_rate": 7.215277777777778e-06,
      "loss": 0.0168,
      "step": 5523
    },
    {
      "epoch": 0.6393518518518518,
      "grad_norm": 34.04227066040039,
      "learning_rate": 7.212962962962964e-06,
      "loss": 2.7468,
      "step": 5524
    },
    {
      "epoch": 0.6394675925925926,
      "grad_norm": 0.07707761228084564,
      "learning_rate": 7.210648148148148e-06,
      "loss": 0.0122,
      "step": 5525
    },
    {
      "epoch": 0.6395833333333333,
      "grad_norm": 0.0806172639131546,
      "learning_rate": 7.2083333333333335e-06,
      "loss": 0.0148,
      "step": 5526
    },
    {
      "epoch": 0.6396990740740741,
      "grad_norm": 0.07053638249635696,
      "learning_rate": 7.2060185185185194e-06,
      "loss": 0.0123,
      "step": 5527
    },
    {
      "epoch": 0.6398148148148148,
      "grad_norm": 6.263954162597656,
      "learning_rate": 7.203703703703704e-06,
      "loss": 3.1349,
      "step": 5528
    },
    {
      "epoch": 0.6399305555555556,
      "grad_norm": 3.395819902420044,
      "learning_rate": 7.20138888888889e-06,
      "loss": 0.0251,
      "step": 5529
    },
    {
      "epoch": 0.6400462962962963,
      "grad_norm": 0.060255926102399826,
      "learning_rate": 7.199074074074075e-06,
      "loss": 0.0108,
      "step": 5530
    },
    {
      "epoch": 0.640162037037037,
      "grad_norm": 0.08679290115833282,
      "learning_rate": 7.19675925925926e-06,
      "loss": 0.016,
      "step": 5531
    },
    {
      "epoch": 0.6402777777777777,
      "grad_norm": 0.07459874451160431,
      "learning_rate": 7.194444444444445e-06,
      "loss": 0.0129,
      "step": 5532
    },
    {
      "epoch": 0.6403935185185186,
      "grad_norm": 0.06857501715421677,
      "learning_rate": 7.192129629629631e-06,
      "loss": 0.0117,
      "step": 5533
    },
    {
      "epoch": 0.6405092592592593,
      "grad_norm": 0.045469943434000015,
      "learning_rate": 7.189814814814815e-06,
      "loss": 0.0084,
      "step": 5534
    },
    {
      "epoch": 0.640625,
      "grad_norm": 4.743236064910889,
      "learning_rate": 7.1875e-06,
      "loss": 0.0349,
      "step": 5535
    },
    {
      "epoch": 0.6407407407407407,
      "grad_norm": 0.12054309248924255,
      "learning_rate": 7.185185185185186e-06,
      "loss": 0.0155,
      "step": 5536
    },
    {
      "epoch": 0.6408564814814814,
      "grad_norm": 0.06445153057575226,
      "learning_rate": 7.18287037037037e-06,
      "loss": 0.0085,
      "step": 5537
    },
    {
      "epoch": 0.6409722222222223,
      "grad_norm": 0.07568565011024475,
      "learning_rate": 7.180555555555556e-06,
      "loss": 0.0123,
      "step": 5538
    },
    {
      "epoch": 0.641087962962963,
      "grad_norm": 0.06331192702054977,
      "learning_rate": 7.1782407407407415e-06,
      "loss": 0.0115,
      "step": 5539
    },
    {
      "epoch": 0.6412037037037037,
      "grad_norm": 0.046485234051942825,
      "learning_rate": 7.1759259259259266e-06,
      "loss": 0.0083,
      "step": 5540
    },
    {
      "epoch": 0.6413194444444444,
      "grad_norm": 0.06440041214227676,
      "learning_rate": 7.173611111111112e-06,
      "loss": 0.0115,
      "step": 5541
    },
    {
      "epoch": 0.6414351851851852,
      "grad_norm": 0.06059466302394867,
      "learning_rate": 7.171296296296298e-06,
      "loss": 0.0107,
      "step": 5542
    },
    {
      "epoch": 0.6415509259259259,
      "grad_norm": 0.0953751727938652,
      "learning_rate": 7.168981481481482e-06,
      "loss": 0.0166,
      "step": 5543
    },
    {
      "epoch": 0.6416666666666667,
      "grad_norm": 0.08669030666351318,
      "learning_rate": 7.166666666666667e-06,
      "loss": 0.0147,
      "step": 5544
    },
    {
      "epoch": 0.6417824074074074,
      "grad_norm": 0.07669376581907272,
      "learning_rate": 7.164351851851853e-06,
      "loss": 0.0136,
      "step": 5545
    },
    {
      "epoch": 0.6418981481481482,
      "grad_norm": 0.04671502113342285,
      "learning_rate": 7.162037037037037e-06,
      "loss": 0.0085,
      "step": 5546
    },
    {
      "epoch": 0.6420138888888889,
      "grad_norm": 0.09429909288883209,
      "learning_rate": 7.159722222222223e-06,
      "loss": 0.0137,
      "step": 5547
    },
    {
      "epoch": 0.6421296296296296,
      "grad_norm": 0.08795734494924545,
      "learning_rate": 7.157407407407408e-06,
      "loss": 0.0159,
      "step": 5548
    },
    {
      "epoch": 0.6422453703703703,
      "grad_norm": 1.018810749053955,
      "learning_rate": 7.155092592592593e-06,
      "loss": 0.0154,
      "step": 5549
    },
    {
      "epoch": 0.6423611111111112,
      "grad_norm": 0.05087708309292793,
      "learning_rate": 7.152777777777778e-06,
      "loss": 0.0093,
      "step": 5550
    },
    {
      "epoch": 0.6424768518518519,
      "grad_norm": 0.9033306837081909,
      "learning_rate": 7.150462962962964e-06,
      "loss": 0.0132,
      "step": 5551
    },
    {
      "epoch": 0.6425925925925926,
      "grad_norm": 248.7027130126953,
      "learning_rate": 7.1481481481481486e-06,
      "loss": 1.9028,
      "step": 5552
    },
    {
      "epoch": 0.6427083333333333,
      "grad_norm": 0.06686235964298248,
      "learning_rate": 7.145833333333334e-06,
      "loss": 0.012,
      "step": 5553
    },
    {
      "epoch": 0.642824074074074,
      "grad_norm": 0.06558095663785934,
      "learning_rate": 7.14351851851852e-06,
      "loss": 0.0117,
      "step": 5554
    },
    {
      "epoch": 0.6429398148148148,
      "grad_norm": 0.2138713300228119,
      "learning_rate": 7.141203703703704e-06,
      "loss": 0.0139,
      "step": 5555
    },
    {
      "epoch": 0.6430555555555556,
      "grad_norm": 0.06801667809486389,
      "learning_rate": 7.13888888888889e-06,
      "loss": 0.0123,
      "step": 5556
    },
    {
      "epoch": 0.6431712962962963,
      "grad_norm": 0.0643719881772995,
      "learning_rate": 7.136574074074075e-06,
      "loss": 0.0118,
      "step": 5557
    },
    {
      "epoch": 0.643287037037037,
      "grad_norm": 0.047187142074108124,
      "learning_rate": 7.13425925925926e-06,
      "loss": 0.0086,
      "step": 5558
    },
    {
      "epoch": 0.6434027777777778,
      "grad_norm": 0.07920704036951065,
      "learning_rate": 7.131944444444445e-06,
      "loss": 0.0145,
      "step": 5559
    },
    {
      "epoch": 0.6435185185185185,
      "grad_norm": 0.06811879575252533,
      "learning_rate": 7.129629629629629e-06,
      "loss": 0.0089,
      "step": 5560
    },
    {
      "epoch": 0.6436342592592592,
      "grad_norm": 0.08386334776878357,
      "learning_rate": 7.127314814814815e-06,
      "loss": 0.0137,
      "step": 5561
    },
    {
      "epoch": 0.64375,
      "grad_norm": 0.07660465687513351,
      "learning_rate": 7.125e-06,
      "loss": 0.0136,
      "step": 5562
    },
    {
      "epoch": 0.6438657407407408,
      "grad_norm": 0.04582426697015762,
      "learning_rate": 7.1226851851851855e-06,
      "loss": 0.0084,
      "step": 5563
    },
    {
      "epoch": 0.6439814814814815,
      "grad_norm": 0.058397892862558365,
      "learning_rate": 7.120370370370371e-06,
      "loss": 0.01,
      "step": 5564
    },
    {
      "epoch": 0.6440972222222222,
      "grad_norm": 0.04638462886214256,
      "learning_rate": 7.1180555555555565e-06,
      "loss": 0.0083,
      "step": 5565
    },
    {
      "epoch": 0.6442129629629629,
      "grad_norm": 0.0879162847995758,
      "learning_rate": 7.115740740740741e-06,
      "loss": 0.0161,
      "step": 5566
    },
    {
      "epoch": 0.6443287037037037,
      "grad_norm": 0.04750915989279747,
      "learning_rate": 7.113425925925927e-06,
      "loss": 0.0087,
      "step": 5567
    },
    {
      "epoch": 0.6444444444444445,
      "grad_norm": 0.0830477848649025,
      "learning_rate": 7.111111111111112e-06,
      "loss": 0.015,
      "step": 5568
    },
    {
      "epoch": 0.6445601851851852,
      "grad_norm": 0.0790814533829689,
      "learning_rate": 7.108796296296296e-06,
      "loss": 0.0139,
      "step": 5569
    },
    {
      "epoch": 0.6446759259259259,
      "grad_norm": 0.05658203363418579,
      "learning_rate": 7.106481481481482e-06,
      "loss": 0.0103,
      "step": 5570
    },
    {
      "epoch": 0.6447916666666667,
      "grad_norm": 1.3880515098571777,
      "learning_rate": 7.104166666666668e-06,
      "loss": 0.016,
      "step": 5571
    },
    {
      "epoch": 0.6449074074074074,
      "grad_norm": 0.4072161316871643,
      "learning_rate": 7.101851851851852e-06,
      "loss": 0.0139,
      "step": 5572
    },
    {
      "epoch": 0.6450231481481481,
      "grad_norm": 0.06807760149240494,
      "learning_rate": 7.099537037037037e-06,
      "loss": 0.0116,
      "step": 5573
    },
    {
      "epoch": 0.6451388888888889,
      "grad_norm": 0.07863930612802505,
      "learning_rate": 7.097222222222223e-06,
      "loss": 0.0104,
      "step": 5574
    },
    {
      "epoch": 0.6452546296296297,
      "grad_norm": 76.55365753173828,
      "learning_rate": 7.0949074074074075e-06,
      "loss": 2.1221,
      "step": 5575
    },
    {
      "epoch": 0.6453703703703704,
      "grad_norm": 0.3263586163520813,
      "learning_rate": 7.0925925925925935e-06,
      "loss": 0.0184,
      "step": 5576
    },
    {
      "epoch": 0.6454861111111111,
      "grad_norm": 0.08068040013313293,
      "learning_rate": 7.0902777777777785e-06,
      "loss": 0.0146,
      "step": 5577
    },
    {
      "epoch": 0.6456018518518518,
      "grad_norm": 0.07808630168437958,
      "learning_rate": 7.087962962962963e-06,
      "loss": 0.0139,
      "step": 5578
    },
    {
      "epoch": 0.6457175925925925,
      "grad_norm": 0.06378553062677383,
      "learning_rate": 7.085648148148149e-06,
      "loss": 0.0116,
      "step": 5579
    },
    {
      "epoch": 0.6458333333333334,
      "grad_norm": 6.251893043518066,
      "learning_rate": 7.083333333333335e-06,
      "loss": 3.492,
      "step": 5580
    },
    {
      "epoch": 0.6459490740740741,
      "grad_norm": 0.23513299226760864,
      "learning_rate": 7.081018518518519e-06,
      "loss": 0.0139,
      "step": 5581
    },
    {
      "epoch": 0.6460648148148148,
      "grad_norm": 140.3226318359375,
      "learning_rate": 7.078703703703704e-06,
      "loss": 0.7416,
      "step": 5582
    },
    {
      "epoch": 0.6461805555555555,
      "grad_norm": 0.06479990482330322,
      "learning_rate": 7.07638888888889e-06,
      "loss": 0.0115,
      "step": 5583
    },
    {
      "epoch": 0.6462962962962963,
      "grad_norm": 0.0938716009259224,
      "learning_rate": 7.074074074074074e-06,
      "loss": 0.0141,
      "step": 5584
    },
    {
      "epoch": 0.6464120370370371,
      "grad_norm": 0.08845657855272293,
      "learning_rate": 7.07175925925926e-06,
      "loss": 0.0092,
      "step": 5585
    },
    {
      "epoch": 0.6465277777777778,
      "grad_norm": 266.2555847167969,
      "learning_rate": 7.069444444444445e-06,
      "loss": 0.3783,
      "step": 5586
    },
    {
      "epoch": 0.6466435185185185,
      "grad_norm": 0.12906873226165771,
      "learning_rate": 7.0671296296296295e-06,
      "loss": 0.0155,
      "step": 5587
    },
    {
      "epoch": 0.6467592592592593,
      "grad_norm": 0.07679867744445801,
      "learning_rate": 7.0648148148148155e-06,
      "loss": 0.0138,
      "step": 5588
    },
    {
      "epoch": 0.646875,
      "grad_norm": 0.06250967085361481,
      "learning_rate": 7.062500000000001e-06,
      "loss": 0.0113,
      "step": 5589
    },
    {
      "epoch": 0.6469907407407407,
      "grad_norm": 1.4745235443115234,
      "learning_rate": 7.060185185185186e-06,
      "loss": 0.0239,
      "step": 5590
    },
    {
      "epoch": 0.6471064814814815,
      "grad_norm": 0.06284324079751968,
      "learning_rate": 7.057870370370371e-06,
      "loss": 0.0115,
      "step": 5591
    },
    {
      "epoch": 0.6472222222222223,
      "grad_norm": 0.06902594864368439,
      "learning_rate": 7.055555555555557e-06,
      "loss": 0.0125,
      "step": 5592
    },
    {
      "epoch": 0.647337962962963,
      "grad_norm": 15.293631553649902,
      "learning_rate": 7.053240740740741e-06,
      "loss": 0.0652,
      "step": 5593
    },
    {
      "epoch": 0.6474537037037037,
      "grad_norm": 0.14838452637195587,
      "learning_rate": 7.050925925925927e-06,
      "loss": 0.011,
      "step": 5594
    },
    {
      "epoch": 0.6475694444444444,
      "grad_norm": 0.061848461627960205,
      "learning_rate": 7.048611111111112e-06,
      "loss": 0.0111,
      "step": 5595
    },
    {
      "epoch": 0.6476851851851851,
      "grad_norm": 0.07781767845153809,
      "learning_rate": 7.046296296296296e-06,
      "loss": 0.0103,
      "step": 5596
    },
    {
      "epoch": 0.647800925925926,
      "grad_norm": 0.04647459462285042,
      "learning_rate": 7.043981481481482e-06,
      "loss": 0.0085,
      "step": 5597
    },
    {
      "epoch": 0.6479166666666667,
      "grad_norm": 0.14407983422279358,
      "learning_rate": 7.041666666666668e-06,
      "loss": 0.0147,
      "step": 5598
    },
    {
      "epoch": 0.6480324074074074,
      "grad_norm": 0.0643044114112854,
      "learning_rate": 7.039351851851852e-06,
      "loss": 0.0084,
      "step": 5599
    },
    {
      "epoch": 0.6481481481481481,
      "grad_norm": 0.06189335137605667,
      "learning_rate": 7.0370370370370375e-06,
      "loss": 0.0109,
      "step": 5600
    },
    {
      "epoch": 0.6482638888888889,
      "grad_norm": 0.0917472317814827,
      "learning_rate": 7.0347222222222234e-06,
      "loss": 0.0122,
      "step": 5601
    },
    {
      "epoch": 0.6483796296296296,
      "grad_norm": 0.06170905753970146,
      "learning_rate": 7.032407407407408e-06,
      "loss": 0.0082,
      "step": 5602
    },
    {
      "epoch": 0.6484953703703704,
      "grad_norm": 0.07577039301395416,
      "learning_rate": 7.030092592592594e-06,
      "loss": 0.0122,
      "step": 5603
    },
    {
      "epoch": 0.6486111111111111,
      "grad_norm": 0.10597047209739685,
      "learning_rate": 7.027777777777778e-06,
      "loss": 0.0123,
      "step": 5604
    },
    {
      "epoch": 0.6487268518518519,
      "grad_norm": 0.05218036100268364,
      "learning_rate": 7.025462962962963e-06,
      "loss": 0.0091,
      "step": 5605
    },
    {
      "epoch": 0.6488425925925926,
      "grad_norm": 0.079380102455616,
      "learning_rate": 7.023148148148149e-06,
      "loss": 0.0145,
      "step": 5606
    },
    {
      "epoch": 0.6489583333333333,
      "grad_norm": 0.06438931077718735,
      "learning_rate": 7.020833333333333e-06,
      "loss": 0.0115,
      "step": 5607
    },
    {
      "epoch": 0.649074074074074,
      "grad_norm": 0.24780413508415222,
      "learning_rate": 7.018518518518519e-06,
      "loss": 0.0147,
      "step": 5608
    },
    {
      "epoch": 0.6491898148148149,
      "grad_norm": 0.07246649265289307,
      "learning_rate": 7.016203703703704e-06,
      "loss": 0.0131,
      "step": 5609
    },
    {
      "epoch": 0.6493055555555556,
      "grad_norm": 19.127538681030273,
      "learning_rate": 7.013888888888889e-06,
      "loss": 2.7135,
      "step": 5610
    },
    {
      "epoch": 0.6494212962962963,
      "grad_norm": 0.07283186167478561,
      "learning_rate": 7.011574074074074e-06,
      "loss": 0.0117,
      "step": 5611
    },
    {
      "epoch": 0.649537037037037,
      "grad_norm": 0.08565779030323029,
      "learning_rate": 7.00925925925926e-06,
      "loss": 0.0155,
      "step": 5612
    },
    {
      "epoch": 0.6496527777777777,
      "grad_norm": 0.0743861123919487,
      "learning_rate": 7.006944444444445e-06,
      "loss": 0.0138,
      "step": 5613
    },
    {
      "epoch": 0.6497685185185185,
      "grad_norm": 0.06006825715303421,
      "learning_rate": 7.00462962962963e-06,
      "loss": 0.0109,
      "step": 5614
    },
    {
      "epoch": 0.6498842592592593,
      "grad_norm": 12.508211135864258,
      "learning_rate": 7.002314814814816e-06,
      "loss": 2.9652,
      "step": 5615
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.04845240339636803,
      "learning_rate": 7e-06,
      "loss": 0.0085,
      "step": 5616
    },
    {
      "epoch": 0.6501157407407407,
      "grad_norm": 26.158926010131836,
      "learning_rate": 6.997685185185186e-06,
      "loss": 0.0848,
      "step": 5617
    },
    {
      "epoch": 0.6502314814814815,
      "grad_norm": 0.06790537387132645,
      "learning_rate": 6.995370370370371e-06,
      "loss": 0.0118,
      "step": 5618
    },
    {
      "epoch": 0.6503472222222222,
      "grad_norm": 32.48348617553711,
      "learning_rate": 6.993055555555556e-06,
      "loss": 2.835,
      "step": 5619
    },
    {
      "epoch": 0.6504629629629629,
      "grad_norm": 0.058217618614435196,
      "learning_rate": 6.990740740740741e-06,
      "loss": 0.0107,
      "step": 5620
    },
    {
      "epoch": 0.6505787037037037,
      "grad_norm": 6.388825416564941,
      "learning_rate": 6.988425925925927e-06,
      "loss": 0.0341,
      "step": 5621
    },
    {
      "epoch": 0.6506944444444445,
      "grad_norm": 0.4018612504005432,
      "learning_rate": 6.986111111111111e-06,
      "loss": 0.0159,
      "step": 5622
    },
    {
      "epoch": 0.6508101851851852,
      "grad_norm": 0.04938862472772598,
      "learning_rate": 6.983796296296297e-06,
      "loss": 0.009,
      "step": 5623
    },
    {
      "epoch": 0.6509259259259259,
      "grad_norm": 0.10922794044017792,
      "learning_rate": 6.981481481481482e-06,
      "loss": 0.0144,
      "step": 5624
    },
    {
      "epoch": 0.6510416666666666,
      "grad_norm": 0.04957647621631622,
      "learning_rate": 6.979166666666667e-06,
      "loss": 0.0091,
      "step": 5625
    },
    {
      "epoch": 0.6511574074074075,
      "grad_norm": 0.07992157340049744,
      "learning_rate": 6.9768518518518526e-06,
      "loss": 0.0141,
      "step": 5626
    },
    {
      "epoch": 0.6512731481481482,
      "grad_norm": 0.0725012943148613,
      "learning_rate": 6.974537037037038e-06,
      "loss": 0.0128,
      "step": 5627
    },
    {
      "epoch": 0.6513888888888889,
      "grad_norm": 0.05786861851811409,
      "learning_rate": 6.972222222222223e-06,
      "loss": 0.0107,
      "step": 5628
    },
    {
      "epoch": 0.6515046296296296,
      "grad_norm": 0.06164422631263733,
      "learning_rate": 6.969907407407408e-06,
      "loss": 0.0111,
      "step": 5629
    },
    {
      "epoch": 0.6516203703703703,
      "grad_norm": 0.07999315112829208,
      "learning_rate": 6.967592592592594e-06,
      "loss": 0.0146,
      "step": 5630
    },
    {
      "epoch": 0.6517361111111111,
      "grad_norm": 0.056684110313653946,
      "learning_rate": 6.965277777777778e-06,
      "loss": 0.0103,
      "step": 5631
    },
    {
      "epoch": 0.6518518518518519,
      "grad_norm": 0.05212559178471565,
      "learning_rate": 6.962962962962964e-06,
      "loss": 0.0094,
      "step": 5632
    },
    {
      "epoch": 0.6519675925925926,
      "grad_norm": 0.06001989543437958,
      "learning_rate": 6.960648148148149e-06,
      "loss": 0.0109,
      "step": 5633
    },
    {
      "epoch": 0.6520833333333333,
      "grad_norm": 0.051949046552181244,
      "learning_rate": 6.958333333333333e-06,
      "loss": 0.0094,
      "step": 5634
    },
    {
      "epoch": 0.6521990740740741,
      "grad_norm": 0.10619015991687775,
      "learning_rate": 6.956018518518519e-06,
      "loss": 0.0166,
      "step": 5635
    },
    {
      "epoch": 0.6523148148148148,
      "grad_norm": 0.04969280958175659,
      "learning_rate": 6.953703703703704e-06,
      "loss": 0.0091,
      "step": 5636
    },
    {
      "epoch": 0.6524305555555555,
      "grad_norm": 163.6638641357422,
      "learning_rate": 6.9513888888888895e-06,
      "loss": 0.325,
      "step": 5637
    },
    {
      "epoch": 0.6525462962962963,
      "grad_norm": 0.07103899866342545,
      "learning_rate": 6.949074074074075e-06,
      "loss": 0.0125,
      "step": 5638
    },
    {
      "epoch": 0.6526620370370371,
      "grad_norm": 0.05974484235048294,
      "learning_rate": 6.9467592592592605e-06,
      "loss": 0.0109,
      "step": 5639
    },
    {
      "epoch": 0.6527777777777778,
      "grad_norm": 0.07748965173959732,
      "learning_rate": 6.944444444444445e-06,
      "loss": 0.0141,
      "step": 5640
    },
    {
      "epoch": 0.6528935185185185,
      "grad_norm": 0.06855938583612442,
      "learning_rate": 6.942129629629631e-06,
      "loss": 0.0119,
      "step": 5641
    },
    {
      "epoch": 0.6530092592592592,
      "grad_norm": 0.061356380581855774,
      "learning_rate": 6.939814814814816e-06,
      "loss": 0.0081,
      "step": 5642
    },
    {
      "epoch": 0.653125,
      "grad_norm": 0.050095267593860626,
      "learning_rate": 6.9375e-06,
      "loss": 0.0092,
      "step": 5643
    },
    {
      "epoch": 0.6532407407407408,
      "grad_norm": 0.07564178854227066,
      "learning_rate": 6.935185185185186e-06,
      "loss": 0.013,
      "step": 5644
    },
    {
      "epoch": 0.6533564814814815,
      "grad_norm": 0.0640806034207344,
      "learning_rate": 6.932870370370371e-06,
      "loss": 0.0117,
      "step": 5645
    },
    {
      "epoch": 0.6534722222222222,
      "grad_norm": 0.05990441516041756,
      "learning_rate": 6.930555555555556e-06,
      "loss": 0.011,
      "step": 5646
    },
    {
      "epoch": 0.653587962962963,
      "grad_norm": 0.30268594622612,
      "learning_rate": 6.928240740740741e-06,
      "loss": 0.016,
      "step": 5647
    },
    {
      "epoch": 0.6537037037037037,
      "grad_norm": 0.0979585275053978,
      "learning_rate": 6.9259259259259256e-06,
      "loss": 0.0126,
      "step": 5648
    },
    {
      "epoch": 0.6538194444444444,
      "grad_norm": 0.049785587936639786,
      "learning_rate": 6.9236111111111115e-06,
      "loss": 0.009,
      "step": 5649
    },
    {
      "epoch": 0.6539351851851852,
      "grad_norm": 0.06537054479122162,
      "learning_rate": 6.9212962962962974e-06,
      "loss": 0.0115,
      "step": 5650
    },
    {
      "epoch": 0.654050925925926,
      "grad_norm": 5.976502418518066,
      "learning_rate": 6.918981481481482e-06,
      "loss": 0.0276,
      "step": 5651
    },
    {
      "epoch": 0.6541666666666667,
      "grad_norm": 0.07365087419748306,
      "learning_rate": 6.916666666666667e-06,
      "loss": 0.0121,
      "step": 5652
    },
    {
      "epoch": 0.6542824074074074,
      "grad_norm": 0.04887187108397484,
      "learning_rate": 6.914351851851853e-06,
      "loss": 0.0089,
      "step": 5653
    },
    {
      "epoch": 0.6543981481481481,
      "grad_norm": 53.675270080566406,
      "learning_rate": 6.912037037037037e-06,
      "loss": 2.5887,
      "step": 5654
    },
    {
      "epoch": 0.6545138888888888,
      "grad_norm": 0.06858395785093307,
      "learning_rate": 6.909722222222223e-06,
      "loss": 0.0121,
      "step": 5655
    },
    {
      "epoch": 0.6546296296296297,
      "grad_norm": 0.08020824193954468,
      "learning_rate": 6.907407407407408e-06,
      "loss": 0.0116,
      "step": 5656
    },
    {
      "epoch": 0.6547453703703704,
      "grad_norm": 0.14228519797325134,
      "learning_rate": 6.905092592592592e-06,
      "loss": 0.0125,
      "step": 5657
    },
    {
      "epoch": 0.6548611111111111,
      "grad_norm": 11.2146635055542,
      "learning_rate": 6.902777777777778e-06,
      "loss": 2.9912,
      "step": 5658
    },
    {
      "epoch": 0.6549768518518518,
      "grad_norm": 0.0668773353099823,
      "learning_rate": 6.900462962962964e-06,
      "loss": 0.0124,
      "step": 5659
    },
    {
      "epoch": 0.6550925925925926,
      "grad_norm": 0.06793259084224701,
      "learning_rate": 6.898148148148148e-06,
      "loss": 0.0123,
      "step": 5660
    },
    {
      "epoch": 0.6552083333333333,
      "grad_norm": 12.700535774230957,
      "learning_rate": 6.8958333333333335e-06,
      "loss": 2.8062,
      "step": 5661
    },
    {
      "epoch": 0.6553240740740741,
      "grad_norm": 0.06452244520187378,
      "learning_rate": 6.8935185185185195e-06,
      "loss": 0.0112,
      "step": 5662
    },
    {
      "epoch": 0.6554398148148148,
      "grad_norm": 0.05995018780231476,
      "learning_rate": 6.891203703703704e-06,
      "loss": 0.011,
      "step": 5663
    },
    {
      "epoch": 0.6555555555555556,
      "grad_norm": 0.05611143261194229,
      "learning_rate": 6.88888888888889e-06,
      "loss": 0.0096,
      "step": 5664
    },
    {
      "epoch": 0.6556712962962963,
      "grad_norm": 0.0525989830493927,
      "learning_rate": 6.886574074074075e-06,
      "loss": 0.0097,
      "step": 5665
    },
    {
      "epoch": 0.655787037037037,
      "grad_norm": 0.04897473379969597,
      "learning_rate": 6.884259259259259e-06,
      "loss": 0.009,
      "step": 5666
    },
    {
      "epoch": 0.6559027777777777,
      "grad_norm": 82.86630249023438,
      "learning_rate": 6.881944444444445e-06,
      "loss": 1.2938,
      "step": 5667
    },
    {
      "epoch": 0.6560185185185186,
      "grad_norm": 0.061854153871536255,
      "learning_rate": 6.879629629629631e-06,
      "loss": 0.011,
      "step": 5668
    },
    {
      "epoch": 0.6561342592592593,
      "grad_norm": 0.045167915523052216,
      "learning_rate": 6.877314814814815e-06,
      "loss": 0.0083,
      "step": 5669
    },
    {
      "epoch": 0.65625,
      "grad_norm": 0.08070927858352661,
      "learning_rate": 6.875e-06,
      "loss": 0.0134,
      "step": 5670
    },
    {
      "epoch": 0.6563657407407407,
      "grad_norm": 0.0789189264178276,
      "learning_rate": 6.872685185185186e-06,
      "loss": 0.0095,
      "step": 5671
    },
    {
      "epoch": 0.6564814814814814,
      "grad_norm": 0.059650156646966934,
      "learning_rate": 6.8703703703703704e-06,
      "loss": 0.0107,
      "step": 5672
    },
    {
      "epoch": 0.6565972222222223,
      "grad_norm": 0.06864487379789352,
      "learning_rate": 6.868055555555556e-06,
      "loss": 0.0116,
      "step": 5673
    },
    {
      "epoch": 0.656712962962963,
      "grad_norm": 0.07833556830883026,
      "learning_rate": 6.8657407407407415e-06,
      "loss": 0.0102,
      "step": 5674
    },
    {
      "epoch": 0.6568287037037037,
      "grad_norm": 0.07545624673366547,
      "learning_rate": 6.863425925925927e-06,
      "loss": 0.0135,
      "step": 5675
    },
    {
      "epoch": 0.6569444444444444,
      "grad_norm": 0.11373011022806168,
      "learning_rate": 6.861111111111112e-06,
      "loss": 0.0124,
      "step": 5676
    },
    {
      "epoch": 0.6570601851851852,
      "grad_norm": 0.07150730490684509,
      "learning_rate": 6.858796296296298e-06,
      "loss": 0.0125,
      "step": 5677
    },
    {
      "epoch": 0.6571759259259259,
      "grad_norm": 0.07289008051156998,
      "learning_rate": 6.856481481481482e-06,
      "loss": 0.0123,
      "step": 5678
    },
    {
      "epoch": 0.6572916666666667,
      "grad_norm": 0.06928786635398865,
      "learning_rate": 6.854166666666667e-06,
      "loss": 0.0123,
      "step": 5679
    },
    {
      "epoch": 0.6574074074074074,
      "grad_norm": 0.06465476006269455,
      "learning_rate": 6.851851851851853e-06,
      "loss": 0.0118,
      "step": 5680
    },
    {
      "epoch": 0.6575231481481482,
      "grad_norm": 0.06720929592847824,
      "learning_rate": 6.849537037037037e-06,
      "loss": 0.0115,
      "step": 5681
    },
    {
      "epoch": 0.6576388888888889,
      "grad_norm": 0.06287407130002975,
      "learning_rate": 6.847222222222223e-06,
      "loss": 0.0114,
      "step": 5682
    },
    {
      "epoch": 0.6577546296296296,
      "grad_norm": 0.06474670022726059,
      "learning_rate": 6.844907407407408e-06,
      "loss": 0.0095,
      "step": 5683
    },
    {
      "epoch": 0.6578703703703703,
      "grad_norm": 0.08393298834562302,
      "learning_rate": 6.842592592592593e-06,
      "loss": 0.0122,
      "step": 5684
    },
    {
      "epoch": 0.6579861111111112,
      "grad_norm": 0.058147504925727844,
      "learning_rate": 6.840277777777778e-06,
      "loss": 0.0096,
      "step": 5685
    },
    {
      "epoch": 0.6581018518518519,
      "grad_norm": 0.5375710725784302,
      "learning_rate": 6.837962962962964e-06,
      "loss": 0.0122,
      "step": 5686
    },
    {
      "epoch": 0.6582175925925926,
      "grad_norm": 0.08680267632007599,
      "learning_rate": 6.835648148148149e-06,
      "loss": 0.0119,
      "step": 5687
    },
    {
      "epoch": 0.6583333333333333,
      "grad_norm": 0.07436604797840118,
      "learning_rate": 6.833333333333334e-06,
      "loss": 0.0124,
      "step": 5688
    },
    {
      "epoch": 0.658449074074074,
      "grad_norm": 0.06022024527192116,
      "learning_rate": 6.83101851851852e-06,
      "loss": 0.0109,
      "step": 5689
    },
    {
      "epoch": 0.6585648148148148,
      "grad_norm": 40.95987319946289,
      "learning_rate": 6.828703703703704e-06,
      "loss": 0.1081,
      "step": 5690
    },
    {
      "epoch": 0.6586805555555556,
      "grad_norm": 0.04558710381388664,
      "learning_rate": 6.82638888888889e-06,
      "loss": 0.0084,
      "step": 5691
    },
    {
      "epoch": 0.6587962962962963,
      "grad_norm": 0.07501699030399323,
      "learning_rate": 6.824074074074075e-06,
      "loss": 0.0134,
      "step": 5692
    },
    {
      "epoch": 0.658912037037037,
      "grad_norm": 41.544071197509766,
      "learning_rate": 6.82175925925926e-06,
      "loss": 0.1494,
      "step": 5693
    },
    {
      "epoch": 0.6590277777777778,
      "grad_norm": 0.058946188539266586,
      "learning_rate": 6.819444444444445e-06,
      "loss": 0.0106,
      "step": 5694
    },
    {
      "epoch": 0.6591435185185185,
      "grad_norm": 0.04980137571692467,
      "learning_rate": 6.817129629629629e-06,
      "loss": 0.0085,
      "step": 5695
    },
    {
      "epoch": 0.6592592592592592,
      "grad_norm": 0.048609621822834015,
      "learning_rate": 6.814814814814815e-06,
      "loss": 0.0089,
      "step": 5696
    },
    {
      "epoch": 0.659375,
      "grad_norm": 0.0659797340631485,
      "learning_rate": 6.8125e-06,
      "loss": 0.0114,
      "step": 5697
    },
    {
      "epoch": 0.6594907407407408,
      "grad_norm": 0.16727930307388306,
      "learning_rate": 6.8101851851851855e-06,
      "loss": 0.01,
      "step": 5698
    },
    {
      "epoch": 0.6596064814814815,
      "grad_norm": 0.0699441060423851,
      "learning_rate": 6.807870370370371e-06,
      "loss": 0.0128,
      "step": 5699
    },
    {
      "epoch": 0.6597222222222222,
      "grad_norm": 0.0883135125041008,
      "learning_rate": 6.8055555555555566e-06,
      "loss": 0.0136,
      "step": 5700
    },
    {
      "epoch": 0.6598379629629629,
      "grad_norm": 0.08913455903530121,
      "learning_rate": 6.803240740740741e-06,
      "loss": 0.0149,
      "step": 5701
    },
    {
      "epoch": 0.6599537037037037,
      "grad_norm": 0.04819931834936142,
      "learning_rate": 6.800925925925927e-06,
      "loss": 0.0089,
      "step": 5702
    },
    {
      "epoch": 0.6600694444444445,
      "grad_norm": 0.06467490643262863,
      "learning_rate": 6.798611111111112e-06,
      "loss": 0.0118,
      "step": 5703
    },
    {
      "epoch": 0.6601851851851852,
      "grad_norm": 0.04747511446475983,
      "learning_rate": 6.796296296296296e-06,
      "loss": 0.0087,
      "step": 5704
    },
    {
      "epoch": 0.6603009259259259,
      "grad_norm": 0.06383594125509262,
      "learning_rate": 6.793981481481482e-06,
      "loss": 0.0114,
      "step": 5705
    },
    {
      "epoch": 0.6604166666666667,
      "grad_norm": 0.10672423243522644,
      "learning_rate": 6.791666666666667e-06,
      "loss": 0.0137,
      "step": 5706
    },
    {
      "epoch": 0.6605324074074074,
      "grad_norm": 0.060661680996418,
      "learning_rate": 6.789351851851852e-06,
      "loss": 0.0103,
      "step": 5707
    },
    {
      "epoch": 0.6606481481481481,
      "grad_norm": 0.08593567460775375,
      "learning_rate": 6.787037037037037e-06,
      "loss": 0.0146,
      "step": 5708
    },
    {
      "epoch": 0.6607638888888889,
      "grad_norm": 0.06498662382364273,
      "learning_rate": 6.784722222222223e-06,
      "loss": 0.0118,
      "step": 5709
    },
    {
      "epoch": 0.6608796296296297,
      "grad_norm": 0.06428506225347519,
      "learning_rate": 6.7824074074074075e-06,
      "loss": 0.0112,
      "step": 5710
    },
    {
      "epoch": 0.6609953703703704,
      "grad_norm": 0.06598193943500519,
      "learning_rate": 6.7800925925925935e-06,
      "loss": 0.0118,
      "step": 5711
    },
    {
      "epoch": 0.6611111111111111,
      "grad_norm": 0.06690386682748795,
      "learning_rate": 6.777777777777779e-06,
      "loss": 0.0123,
      "step": 5712
    },
    {
      "epoch": 0.6612268518518518,
      "grad_norm": 125.41551971435547,
      "learning_rate": 6.775462962962963e-06,
      "loss": 0.6599,
      "step": 5713
    },
    {
      "epoch": 0.6613425925925925,
      "grad_norm": 0.048592355102300644,
      "learning_rate": 6.773148148148149e-06,
      "loss": 0.0089,
      "step": 5714
    },
    {
      "epoch": 0.6614583333333334,
      "grad_norm": 91.8642578125,
      "learning_rate": 6.770833333333334e-06,
      "loss": 0.7653,
      "step": 5715
    },
    {
      "epoch": 0.6615740740740741,
      "grad_norm": 0.06842809170484543,
      "learning_rate": 6.768518518518519e-06,
      "loss": 0.0116,
      "step": 5716
    },
    {
      "epoch": 0.6616898148148148,
      "grad_norm": 0.06183180585503578,
      "learning_rate": 6.766203703703704e-06,
      "loss": 0.0113,
      "step": 5717
    },
    {
      "epoch": 0.6618055555555555,
      "grad_norm": 0.0733116865158081,
      "learning_rate": 6.76388888888889e-06,
      "loss": 0.0135,
      "step": 5718
    },
    {
      "epoch": 0.6619212962962963,
      "grad_norm": 0.05129490792751312,
      "learning_rate": 6.761574074074074e-06,
      "loss": 0.0095,
      "step": 5719
    },
    {
      "epoch": 0.6620370370370371,
      "grad_norm": 0.05987207964062691,
      "learning_rate": 6.75925925925926e-06,
      "loss": 0.0107,
      "step": 5720
    },
    {
      "epoch": 0.6621527777777778,
      "grad_norm": 0.045251887291669846,
      "learning_rate": 6.756944444444445e-06,
      "loss": 0.0082,
      "step": 5721
    },
    {
      "epoch": 0.6622685185185185,
      "grad_norm": 0.0694396123290062,
      "learning_rate": 6.7546296296296296e-06,
      "loss": 0.0123,
      "step": 5722
    },
    {
      "epoch": 0.6623842592592593,
      "grad_norm": 0.0653909295797348,
      "learning_rate": 6.7523148148148155e-06,
      "loss": 0.0116,
      "step": 5723
    },
    {
      "epoch": 0.6625,
      "grad_norm": 36.71796417236328,
      "learning_rate": 6.750000000000001e-06,
      "loss": 0.1066,
      "step": 5724
    },
    {
      "epoch": 0.6626157407407407,
      "grad_norm": 0.06358359009027481,
      "learning_rate": 6.747685185185186e-06,
      "loss": 0.0111,
      "step": 5725
    },
    {
      "epoch": 0.6627314814814815,
      "grad_norm": 0.048650529235601425,
      "learning_rate": 6.745370370370371e-06,
      "loss": 0.0089,
      "step": 5726
    },
    {
      "epoch": 0.6628472222222223,
      "grad_norm": 48.25823211669922,
      "learning_rate": 6.743055555555557e-06,
      "loss": 2.4844,
      "step": 5727
    },
    {
      "epoch": 0.662962962962963,
      "grad_norm": 0.05110780522227287,
      "learning_rate": 6.740740740740741e-06,
      "loss": 0.0091,
      "step": 5728
    },
    {
      "epoch": 0.6630787037037037,
      "grad_norm": 0.06309425085783005,
      "learning_rate": 6.738425925925927e-06,
      "loss": 0.0112,
      "step": 5729
    },
    {
      "epoch": 0.6631944444444444,
      "grad_norm": 0.3473646342754364,
      "learning_rate": 6.736111111111112e-06,
      "loss": 0.0162,
      "step": 5730
    },
    {
      "epoch": 0.6633101851851851,
      "grad_norm": 0.04503914713859558,
      "learning_rate": 6.733796296296296e-06,
      "loss": 0.0083,
      "step": 5731
    },
    {
      "epoch": 0.663425925925926,
      "grad_norm": 0.049213957041502,
      "learning_rate": 6.731481481481482e-06,
      "loss": 0.0089,
      "step": 5732
    },
    {
      "epoch": 0.6635416666666667,
      "grad_norm": 7.13392972946167,
      "learning_rate": 6.729166666666667e-06,
      "loss": 3.3681,
      "step": 5733
    },
    {
      "epoch": 0.6636574074074074,
      "grad_norm": 6.364194869995117,
      "learning_rate": 6.726851851851852e-06,
      "loss": 0.0337,
      "step": 5734
    },
    {
      "epoch": 0.6637731481481481,
      "grad_norm": 0.09190118312835693,
      "learning_rate": 6.7245370370370375e-06,
      "loss": 0.0131,
      "step": 5735
    },
    {
      "epoch": 0.6638888888888889,
      "grad_norm": 0.061844419687986374,
      "learning_rate": 6.7222222222222235e-06,
      "loss": 0.0081,
      "step": 5736
    },
    {
      "epoch": 0.6640046296296296,
      "grad_norm": 0.0962665006518364,
      "learning_rate": 6.719907407407408e-06,
      "loss": 0.0141,
      "step": 5737
    },
    {
      "epoch": 0.6641203703703704,
      "grad_norm": 0.06501039862632751,
      "learning_rate": 6.717592592592594e-06,
      "loss": 0.0113,
      "step": 5738
    },
    {
      "epoch": 0.6642361111111111,
      "grad_norm": 19.670475006103516,
      "learning_rate": 6.715277777777778e-06,
      "loss": 0.0574,
      "step": 5739
    },
    {
      "epoch": 0.6643518518518519,
      "grad_norm": 0.07076223194599152,
      "learning_rate": 6.712962962962963e-06,
      "loss": 0.0123,
      "step": 5740
    },
    {
      "epoch": 0.6644675925925926,
      "grad_norm": 0.055700019001960754,
      "learning_rate": 6.710648148148149e-06,
      "loss": 0.0101,
      "step": 5741
    },
    {
      "epoch": 0.6645833333333333,
      "grad_norm": 8.99116325378418,
      "learning_rate": 6.708333333333333e-06,
      "loss": 0.0399,
      "step": 5742
    },
    {
      "epoch": 0.664699074074074,
      "grad_norm": 13.872315406799316,
      "learning_rate": 6.706018518518519e-06,
      "loss": 0.0347,
      "step": 5743
    },
    {
      "epoch": 0.6648148148148149,
      "grad_norm": 0.07753648608922958,
      "learning_rate": 6.703703703703704e-06,
      "loss": 0.0133,
      "step": 5744
    },
    {
      "epoch": 0.6649305555555556,
      "grad_norm": 0.04900863766670227,
      "learning_rate": 6.701388888888889e-06,
      "loss": 0.0089,
      "step": 5745
    },
    {
      "epoch": 0.6650462962962963,
      "grad_norm": 0.057242512702941895,
      "learning_rate": 6.6990740740740744e-06,
      "loss": 0.0092,
      "step": 5746
    },
    {
      "epoch": 0.665162037037037,
      "grad_norm": 0.0511678010225296,
      "learning_rate": 6.69675925925926e-06,
      "loss": 0.0087,
      "step": 5747
    },
    {
      "epoch": 0.6652777777777777,
      "grad_norm": 0.044713202863931656,
      "learning_rate": 6.694444444444445e-06,
      "loss": 0.0081,
      "step": 5748
    },
    {
      "epoch": 0.6653935185185185,
      "grad_norm": 0.07308319956064224,
      "learning_rate": 6.69212962962963e-06,
      "loss": 0.0131,
      "step": 5749
    },
    {
      "epoch": 0.6655092592592593,
      "grad_norm": 0.047782015055418015,
      "learning_rate": 6.689814814814816e-06,
      "loss": 0.0086,
      "step": 5750
    },
    {
      "epoch": 0.665625,
      "grad_norm": 0.09031961113214493,
      "learning_rate": 6.6875e-06,
      "loss": 0.0105,
      "step": 5751
    },
    {
      "epoch": 0.6657407407407407,
      "grad_norm": 0.05093945935368538,
      "learning_rate": 6.685185185185186e-06,
      "loss": 0.0092,
      "step": 5752
    },
    {
      "epoch": 0.6658564814814815,
      "grad_norm": 0.060824327170848846,
      "learning_rate": 6.682870370370371e-06,
      "loss": 0.0108,
      "step": 5753
    },
    {
      "epoch": 0.6659722222222222,
      "grad_norm": 0.048399750143289566,
      "learning_rate": 6.680555555555556e-06,
      "loss": 0.0087,
      "step": 5754
    },
    {
      "epoch": 0.6660879629629629,
      "grad_norm": 0.05564882606267929,
      "learning_rate": 6.678240740740741e-06,
      "loss": 0.01,
      "step": 5755
    },
    {
      "epoch": 0.6662037037037037,
      "grad_norm": 0.06413306295871735,
      "learning_rate": 6.675925925925927e-06,
      "loss": 0.0112,
      "step": 5756
    },
    {
      "epoch": 0.6663194444444445,
      "grad_norm": 0.0702221468091011,
      "learning_rate": 6.673611111111111e-06,
      "loss": 0.012,
      "step": 5757
    },
    {
      "epoch": 0.6664351851851852,
      "grad_norm": 0.06361038982868195,
      "learning_rate": 6.6712962962962965e-06,
      "loss": 0.0113,
      "step": 5758
    },
    {
      "epoch": 0.6665509259259259,
      "grad_norm": 0.09878488630056381,
      "learning_rate": 6.668981481481482e-06,
      "loss": 0.0159,
      "step": 5759
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 103.39092254638672,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.5692,
      "step": 5760
    },
    {
      "epoch": 0.6667824074074075,
      "grad_norm": 0.05331568419933319,
      "learning_rate": 6.664351851851853e-06,
      "loss": 0.0097,
      "step": 5761
    },
    {
      "epoch": 0.6668981481481482,
      "grad_norm": 0.04332921653985977,
      "learning_rate": 6.662037037037038e-06,
      "loss": 0.0079,
      "step": 5762
    },
    {
      "epoch": 0.6670138888888889,
      "grad_norm": 0.05725087225437164,
      "learning_rate": 6.659722222222223e-06,
      "loss": 0.0105,
      "step": 5763
    },
    {
      "epoch": 0.6671296296296296,
      "grad_norm": 0.06996943801641464,
      "learning_rate": 6.657407407407408e-06,
      "loss": 0.0128,
      "step": 5764
    },
    {
      "epoch": 0.6672453703703703,
      "grad_norm": 0.0824279636144638,
      "learning_rate": 6.655092592592594e-06,
      "loss": 0.015,
      "step": 5765
    },
    {
      "epoch": 0.6673611111111111,
      "grad_norm": 0.06293774396181107,
      "learning_rate": 6.652777777777778e-06,
      "loss": 0.0081,
      "step": 5766
    },
    {
      "epoch": 0.6674768518518519,
      "grad_norm": 0.062183037400245667,
      "learning_rate": 6.650462962962963e-06,
      "loss": 0.0107,
      "step": 5767
    },
    {
      "epoch": 0.6675925925925926,
      "grad_norm": 0.046971190720796585,
      "learning_rate": 6.648148148148149e-06,
      "loss": 0.0086,
      "step": 5768
    },
    {
      "epoch": 0.6677083333333333,
      "grad_norm": 0.048330143094062805,
      "learning_rate": 6.645833333333333e-06,
      "loss": 0.0084,
      "step": 5769
    },
    {
      "epoch": 0.6678240740740741,
      "grad_norm": 0.059395741671323776,
      "learning_rate": 6.643518518518519e-06,
      "loss": 0.0105,
      "step": 5770
    },
    {
      "epoch": 0.6679398148148148,
      "grad_norm": 0.06074509024620056,
      "learning_rate": 6.641203703703704e-06,
      "loss": 0.0107,
      "step": 5771
    },
    {
      "epoch": 0.6680555555555555,
      "grad_norm": 0.06284889578819275,
      "learning_rate": 6.6388888888888895e-06,
      "loss": 0.0112,
      "step": 5772
    },
    {
      "epoch": 0.6681712962962963,
      "grad_norm": 36.064964294433594,
      "learning_rate": 6.636574074074075e-06,
      "loss": 2.855,
      "step": 5773
    },
    {
      "epoch": 0.6682870370370371,
      "grad_norm": 0.053417183458805084,
      "learning_rate": 6.6342592592592606e-06,
      "loss": 0.0097,
      "step": 5774
    },
    {
      "epoch": 0.6684027777777778,
      "grad_norm": 0.08754716068506241,
      "learning_rate": 6.631944444444445e-06,
      "loss": 0.0123,
      "step": 5775
    },
    {
      "epoch": 0.6685185185185185,
      "grad_norm": 0.05411703884601593,
      "learning_rate": 6.62962962962963e-06,
      "loss": 0.0098,
      "step": 5776
    },
    {
      "epoch": 0.6686342592592592,
      "grad_norm": 0.19783249497413635,
      "learning_rate": 6.627314814814816e-06,
      "loss": 0.0157,
      "step": 5777
    },
    {
      "epoch": 0.66875,
      "grad_norm": 0.04759502038359642,
      "learning_rate": 6.625e-06,
      "loss": 0.0087,
      "step": 5778
    },
    {
      "epoch": 0.6688657407407408,
      "grad_norm": 0.04953637346625328,
      "learning_rate": 6.622685185185186e-06,
      "loss": 0.009,
      "step": 5779
    },
    {
      "epoch": 0.6689814814814815,
      "grad_norm": 0.06827279180288315,
      "learning_rate": 6.620370370370371e-06,
      "loss": 0.0124,
      "step": 5780
    },
    {
      "epoch": 0.6690972222222222,
      "grad_norm": 0.049442075192928314,
      "learning_rate": 6.618055555555556e-06,
      "loss": 0.0089,
      "step": 5781
    },
    {
      "epoch": 0.669212962962963,
      "grad_norm": 0.07847344130277634,
      "learning_rate": 6.615740740740741e-06,
      "loss": 0.0125,
      "step": 5782
    },
    {
      "epoch": 0.6693287037037037,
      "grad_norm": 0.06561229377985,
      "learning_rate": 6.613425925925926e-06,
      "loss": 0.0091,
      "step": 5783
    },
    {
      "epoch": 0.6694444444444444,
      "grad_norm": 140.27601623535156,
      "learning_rate": 6.6111111111111115e-06,
      "loss": 1.8894,
      "step": 5784
    },
    {
      "epoch": 0.6695601851851852,
      "grad_norm": 110.53024291992188,
      "learning_rate": 6.608796296296297e-06,
      "loss": 2.3585,
      "step": 5785
    },
    {
      "epoch": 0.669675925925926,
      "grad_norm": 0.06810438632965088,
      "learning_rate": 6.606481481481482e-06,
      "loss": 0.0113,
      "step": 5786
    },
    {
      "epoch": 0.6697916666666667,
      "grad_norm": 0.052021436393260956,
      "learning_rate": 6.604166666666667e-06,
      "loss": 0.0095,
      "step": 5787
    },
    {
      "epoch": 0.6699074074074074,
      "grad_norm": 0.07460631430149078,
      "learning_rate": 6.601851851851853e-06,
      "loss": 0.0133,
      "step": 5788
    },
    {
      "epoch": 0.6700231481481481,
      "grad_norm": 15.448917388916016,
      "learning_rate": 6.599537037037037e-06,
      "loss": 2.811,
      "step": 5789
    },
    {
      "epoch": 0.6701388888888888,
      "grad_norm": 6.9327473640441895,
      "learning_rate": 6.597222222222223e-06,
      "loss": 3.0459,
      "step": 5790
    },
    {
      "epoch": 0.6702546296296297,
      "grad_norm": 0.06330647319555283,
      "learning_rate": 6.594907407407408e-06,
      "loss": 0.0114,
      "step": 5791
    },
    {
      "epoch": 0.6703703703703704,
      "grad_norm": 0.05592518672347069,
      "learning_rate": 6.592592592592592e-06,
      "loss": 0.0101,
      "step": 5792
    },
    {
      "epoch": 0.6704861111111111,
      "grad_norm": 169.9635009765625,
      "learning_rate": 6.590277777777778e-06,
      "loss": 2.0223,
      "step": 5793
    },
    {
      "epoch": 0.6706018518518518,
      "grad_norm": 0.17450878024101257,
      "learning_rate": 6.587962962962964e-06,
      "loss": 0.015,
      "step": 5794
    },
    {
      "epoch": 0.6707175925925926,
      "grad_norm": 0.048265982419252396,
      "learning_rate": 6.5856481481481484e-06,
      "loss": 0.0086,
      "step": 5795
    },
    {
      "epoch": 0.6708333333333333,
      "grad_norm": 0.0559275783598423,
      "learning_rate": 6.5833333333333335e-06,
      "loss": 0.0103,
      "step": 5796
    },
    {
      "epoch": 0.6709490740740741,
      "grad_norm": 0.06736024469137192,
      "learning_rate": 6.5810185185185195e-06,
      "loss": 0.0119,
      "step": 5797
    },
    {
      "epoch": 0.6710648148148148,
      "grad_norm": 0.07236272096633911,
      "learning_rate": 6.578703703703704e-06,
      "loss": 0.0129,
      "step": 5798
    },
    {
      "epoch": 0.6711805555555556,
      "grad_norm": 138.27513122558594,
      "learning_rate": 6.57638888888889e-06,
      "loss": 0.4339,
      "step": 5799
    },
    {
      "epoch": 0.6712962962962963,
      "grad_norm": 0.07093218713998795,
      "learning_rate": 6.574074074074075e-06,
      "loss": 0.0131,
      "step": 5800
    },
    {
      "epoch": 0.671412037037037,
      "grad_norm": 0.06425048410892487,
      "learning_rate": 6.571759259259259e-06,
      "loss": 0.011,
      "step": 5801
    },
    {
      "epoch": 0.6715277777777777,
      "grad_norm": 0.04554159194231033,
      "learning_rate": 6.569444444444445e-06,
      "loss": 0.0081,
      "step": 5802
    },
    {
      "epoch": 0.6716435185185186,
      "grad_norm": 2.0859436988830566,
      "learning_rate": 6.567129629629631e-06,
      "loss": 0.0209,
      "step": 5803
    },
    {
      "epoch": 0.6717592592592593,
      "grad_norm": 0.24155420064926147,
      "learning_rate": 6.564814814814815e-06,
      "loss": 0.013,
      "step": 5804
    },
    {
      "epoch": 0.671875,
      "grad_norm": 0.04683593288064003,
      "learning_rate": 6.5625e-06,
      "loss": 0.0086,
      "step": 5805
    },
    {
      "epoch": 0.6719907407407407,
      "grad_norm": 0.07770850509405136,
      "learning_rate": 6.560185185185186e-06,
      "loss": 0.014,
      "step": 5806
    },
    {
      "epoch": 0.6721064814814814,
      "grad_norm": 0.10440121591091156,
      "learning_rate": 6.5578703703703705e-06,
      "loss": 0.0136,
      "step": 5807
    },
    {
      "epoch": 0.6722222222222223,
      "grad_norm": 0.10180452466011047,
      "learning_rate": 6.555555555555556e-06,
      "loss": 0.0135,
      "step": 5808
    },
    {
      "epoch": 0.672337962962963,
      "grad_norm": 0.07728168368339539,
      "learning_rate": 6.5532407407407415e-06,
      "loss": 0.0141,
      "step": 5809
    },
    {
      "epoch": 0.6724537037037037,
      "grad_norm": 14.770035743713379,
      "learning_rate": 6.550925925925926e-06,
      "loss": 2.9142,
      "step": 5810
    },
    {
      "epoch": 0.6725694444444444,
      "grad_norm": 0.10392511636018753,
      "learning_rate": 6.548611111111112e-06,
      "loss": 0.0157,
      "step": 5811
    },
    {
      "epoch": 0.6726851851851852,
      "grad_norm": 0.09444601088762283,
      "learning_rate": 6.546296296296298e-06,
      "loss": 0.0151,
      "step": 5812
    },
    {
      "epoch": 0.6728009259259259,
      "grad_norm": 0.055524591356515884,
      "learning_rate": 6.543981481481482e-06,
      "loss": 0.01,
      "step": 5813
    },
    {
      "epoch": 0.6729166666666667,
      "grad_norm": 102.60542297363281,
      "learning_rate": 6.541666666666667e-06,
      "loss": 0.5626,
      "step": 5814
    },
    {
      "epoch": 0.6730324074074074,
      "grad_norm": 0.07677611708641052,
      "learning_rate": 6.539351851851853e-06,
      "loss": 0.013,
      "step": 5815
    },
    {
      "epoch": 0.6731481481481482,
      "grad_norm": 0.05555003881454468,
      "learning_rate": 6.537037037037037e-06,
      "loss": 0.0102,
      "step": 5816
    },
    {
      "epoch": 0.6732638888888889,
      "grad_norm": 0.06310784816741943,
      "learning_rate": 6.534722222222223e-06,
      "loss": 0.0111,
      "step": 5817
    },
    {
      "epoch": 0.6733796296296296,
      "grad_norm": 0.0732155591249466,
      "learning_rate": 6.532407407407408e-06,
      "loss": 0.0134,
      "step": 5818
    },
    {
      "epoch": 0.6734953703703703,
      "grad_norm": 0.04877673089504242,
      "learning_rate": 6.5300925925925925e-06,
      "loss": 0.0089,
      "step": 5819
    },
    {
      "epoch": 0.6736111111111112,
      "grad_norm": 0.11163302510976791,
      "learning_rate": 6.5277777777777784e-06,
      "loss": 0.012,
      "step": 5820
    },
    {
      "epoch": 0.6737268518518519,
      "grad_norm": 0.07782597094774246,
      "learning_rate": 6.525462962962964e-06,
      "loss": 0.0136,
      "step": 5821
    },
    {
      "epoch": 0.6738425925925926,
      "grad_norm": 0.04554855450987816,
      "learning_rate": 6.523148148148149e-06,
      "loss": 0.0082,
      "step": 5822
    },
    {
      "epoch": 0.6739583333333333,
      "grad_norm": 12.72391128540039,
      "learning_rate": 6.520833333333334e-06,
      "loss": 0.038,
      "step": 5823
    },
    {
      "epoch": 0.674074074074074,
      "grad_norm": 0.05356522649526596,
      "learning_rate": 6.51851851851852e-06,
      "loss": 0.0093,
      "step": 5824
    },
    {
      "epoch": 0.6741898148148148,
      "grad_norm": 0.08498717844486237,
      "learning_rate": 6.516203703703704e-06,
      "loss": 0.0099,
      "step": 5825
    },
    {
      "epoch": 0.6743055555555556,
      "grad_norm": 0.046607982367277145,
      "learning_rate": 6.51388888888889e-06,
      "loss": 0.0085,
      "step": 5826
    },
    {
      "epoch": 0.6744212962962963,
      "grad_norm": 0.07681064307689667,
      "learning_rate": 6.511574074074075e-06,
      "loss": 0.0131,
      "step": 5827
    },
    {
      "epoch": 0.674537037037037,
      "grad_norm": 0.05775289982557297,
      "learning_rate": 6.509259259259259e-06,
      "loss": 0.0103,
      "step": 5828
    },
    {
      "epoch": 0.6746527777777778,
      "grad_norm": 0.12166742235422134,
      "learning_rate": 6.506944444444445e-06,
      "loss": 0.014,
      "step": 5829
    },
    {
      "epoch": 0.6747685185185185,
      "grad_norm": 0.0811772346496582,
      "learning_rate": 6.504629629629629e-06,
      "loss": 0.0149,
      "step": 5830
    },
    {
      "epoch": 0.6748842592592592,
      "grad_norm": 0.05983195826411247,
      "learning_rate": 6.502314814814815e-06,
      "loss": 0.0108,
      "step": 5831
    },
    {
      "epoch": 0.675,
      "grad_norm": 0.08237120509147644,
      "learning_rate": 6.5000000000000004e-06,
      "loss": 0.0117,
      "step": 5832
    },
    {
      "epoch": 0.6751157407407408,
      "grad_norm": 0.09580230712890625,
      "learning_rate": 6.4976851851851855e-06,
      "loss": 0.0152,
      "step": 5833
    },
    {
      "epoch": 0.6752314814814815,
      "grad_norm": 0.07055195420980453,
      "learning_rate": 6.495370370370371e-06,
      "loss": 0.0125,
      "step": 5834
    },
    {
      "epoch": 0.6753472222222222,
      "grad_norm": 0.051409557461738586,
      "learning_rate": 6.493055555555557e-06,
      "loss": 0.0091,
      "step": 5835
    },
    {
      "epoch": 0.6754629629629629,
      "grad_norm": 0.07427288591861725,
      "learning_rate": 6.490740740740741e-06,
      "loss": 0.0114,
      "step": 5836
    },
    {
      "epoch": 0.6755787037037037,
      "grad_norm": 0.05978558212518692,
      "learning_rate": 6.488425925925926e-06,
      "loss": 0.0105,
      "step": 5837
    },
    {
      "epoch": 0.6756944444444445,
      "grad_norm": 0.10291679203510284,
      "learning_rate": 6.486111111111112e-06,
      "loss": 0.0136,
      "step": 5838
    },
    {
      "epoch": 0.6758101851851852,
      "grad_norm": 15.754307746887207,
      "learning_rate": 6.483796296296296e-06,
      "loss": 2.9347,
      "step": 5839
    },
    {
      "epoch": 0.6759259259259259,
      "grad_norm": 0.07142677158117294,
      "learning_rate": 6.481481481481482e-06,
      "loss": 0.0126,
      "step": 5840
    },
    {
      "epoch": 0.6760416666666667,
      "grad_norm": 0.05655873566865921,
      "learning_rate": 6.479166666666667e-06,
      "loss": 0.0101,
      "step": 5841
    },
    {
      "epoch": 0.6761574074074074,
      "grad_norm": 0.06103678420186043,
      "learning_rate": 6.476851851851852e-06,
      "loss": 0.0112,
      "step": 5842
    },
    {
      "epoch": 0.6762731481481481,
      "grad_norm": 0.06832677125930786,
      "learning_rate": 6.474537037037037e-06,
      "loss": 0.0119,
      "step": 5843
    },
    {
      "epoch": 0.6763888888888889,
      "grad_norm": 0.06536195427179337,
      "learning_rate": 6.472222222222223e-06,
      "loss": 0.0108,
      "step": 5844
    },
    {
      "epoch": 0.6765046296296297,
      "grad_norm": 0.05739409849047661,
      "learning_rate": 6.4699074074074076e-06,
      "loss": 0.0102,
      "step": 5845
    },
    {
      "epoch": 0.6766203703703704,
      "grad_norm": 0.1096593365073204,
      "learning_rate": 6.4675925925925935e-06,
      "loss": 0.009,
      "step": 5846
    },
    {
      "epoch": 0.6767361111111111,
      "grad_norm": 0.05537416785955429,
      "learning_rate": 6.465277777777779e-06,
      "loss": 0.0089,
      "step": 5847
    },
    {
      "epoch": 0.6768518518518518,
      "grad_norm": 0.08930493891239166,
      "learning_rate": 6.462962962962963e-06,
      "loss": 0.0159,
      "step": 5848
    },
    {
      "epoch": 0.6769675925925925,
      "grad_norm": 0.06158775836229324,
      "learning_rate": 6.460648148148149e-06,
      "loss": 0.0099,
      "step": 5849
    },
    {
      "epoch": 0.6770833333333334,
      "grad_norm": 5.594810962677002,
      "learning_rate": 6.458333333333334e-06,
      "loss": 0.0295,
      "step": 5850
    },
    {
      "epoch": 0.6771990740740741,
      "grad_norm": 0.0615798681974411,
      "learning_rate": 6.456018518518519e-06,
      "loss": 0.0113,
      "step": 5851
    },
    {
      "epoch": 0.6773148148148148,
      "grad_norm": 0.05476563423871994,
      "learning_rate": 6.453703703703704e-06,
      "loss": 0.0096,
      "step": 5852
    },
    {
      "epoch": 0.6774305555555555,
      "grad_norm": 21.58614730834961,
      "learning_rate": 6.45138888888889e-06,
      "loss": 0.0668,
      "step": 5853
    },
    {
      "epoch": 0.6775462962962963,
      "grad_norm": 0.06686391681432724,
      "learning_rate": 6.449074074074074e-06,
      "loss": 0.0116,
      "step": 5854
    },
    {
      "epoch": 0.6776620370370371,
      "grad_norm": 287.9940490722656,
      "learning_rate": 6.44675925925926e-06,
      "loss": 2.0817,
      "step": 5855
    },
    {
      "epoch": 0.6777777777777778,
      "grad_norm": 6.140852451324463,
      "learning_rate": 6.444444444444445e-06,
      "loss": 0.0585,
      "step": 5856
    },
    {
      "epoch": 0.6778935185185185,
      "grad_norm": 0.057546187192201614,
      "learning_rate": 6.44212962962963e-06,
      "loss": 0.0103,
      "step": 5857
    },
    {
      "epoch": 0.6780092592592593,
      "grad_norm": 0.09307616949081421,
      "learning_rate": 6.4398148148148155e-06,
      "loss": 0.0116,
      "step": 5858
    },
    {
      "epoch": 0.678125,
      "grad_norm": 0.07068385183811188,
      "learning_rate": 6.437500000000001e-06,
      "loss": 0.0123,
      "step": 5859
    },
    {
      "epoch": 0.6782407407407407,
      "grad_norm": 0.07074078917503357,
      "learning_rate": 6.435185185185186e-06,
      "loss": 0.0122,
      "step": 5860
    },
    {
      "epoch": 0.6783564814814815,
      "grad_norm": 69.72132110595703,
      "learning_rate": 6.432870370370371e-06,
      "loss": 0.2278,
      "step": 5861
    },
    {
      "epoch": 0.6784722222222223,
      "grad_norm": 19.622116088867188,
      "learning_rate": 6.430555555555557e-06,
      "loss": 3.2222,
      "step": 5862
    },
    {
      "epoch": 0.678587962962963,
      "grad_norm": 0.06090346351265907,
      "learning_rate": 6.428240740740741e-06,
      "loss": 0.0111,
      "step": 5863
    },
    {
      "epoch": 0.6787037037037037,
      "grad_norm": 0.06651324778795242,
      "learning_rate": 6.425925925925927e-06,
      "loss": 0.0117,
      "step": 5864
    },
    {
      "epoch": 0.6788194444444444,
      "grad_norm": 0.049895938485860825,
      "learning_rate": 6.423611111111112e-06,
      "loss": 0.0092,
      "step": 5865
    },
    {
      "epoch": 0.6789351851851851,
      "grad_norm": 224.04388427734375,
      "learning_rate": 6.421296296296296e-06,
      "loss": 0.5608,
      "step": 5866
    },
    {
      "epoch": 0.679050925925926,
      "grad_norm": 0.05857839062809944,
      "learning_rate": 6.418981481481482e-06,
      "loss": 0.0106,
      "step": 5867
    },
    {
      "epoch": 0.6791666666666667,
      "grad_norm": 0.10618285834789276,
      "learning_rate": 6.416666666666667e-06,
      "loss": 0.0137,
      "step": 5868
    },
    {
      "epoch": 0.6792824074074074,
      "grad_norm": 0.08172673732042313,
      "learning_rate": 6.4143518518518524e-06,
      "loss": 0.0139,
      "step": 5869
    },
    {
      "epoch": 0.6793981481481481,
      "grad_norm": 0.06091083586215973,
      "learning_rate": 6.4120370370370375e-06,
      "loss": 0.0108,
      "step": 5870
    },
    {
      "epoch": 0.6795138888888889,
      "grad_norm": 0.07488897442817688,
      "learning_rate": 6.4097222222222235e-06,
      "loss": 0.0088,
      "step": 5871
    },
    {
      "epoch": 0.6796296296296296,
      "grad_norm": 0.058991048485040665,
      "learning_rate": 6.407407407407408e-06,
      "loss": 0.0103,
      "step": 5872
    },
    {
      "epoch": 0.6797453703703704,
      "grad_norm": 0.06887603551149368,
      "learning_rate": 6.405092592592594e-06,
      "loss": 0.0123,
      "step": 5873
    },
    {
      "epoch": 0.6798611111111111,
      "grad_norm": 0.0669533759355545,
      "learning_rate": 6.402777777777778e-06,
      "loss": 0.012,
      "step": 5874
    },
    {
      "epoch": 0.6799768518518519,
      "grad_norm": 0.06394098699092865,
      "learning_rate": 6.400462962962963e-06,
      "loss": 0.0113,
      "step": 5875
    },
    {
      "epoch": 0.6800925925925926,
      "grad_norm": 125.22603607177734,
      "learning_rate": 6.398148148148149e-06,
      "loss": 0.3735,
      "step": 5876
    },
    {
      "epoch": 0.6802083333333333,
      "grad_norm": 0.09481582790613174,
      "learning_rate": 6.395833333333333e-06,
      "loss": 0.0126,
      "step": 5877
    },
    {
      "epoch": 0.680324074074074,
      "grad_norm": 135.07388305664062,
      "learning_rate": 6.393518518518519e-06,
      "loss": 1.8915,
      "step": 5878
    },
    {
      "epoch": 0.6804398148148149,
      "grad_norm": 0.04543549567461014,
      "learning_rate": 6.391203703703704e-06,
      "loss": 0.0082,
      "step": 5879
    },
    {
      "epoch": 0.6805555555555556,
      "grad_norm": 0.1402641236782074,
      "learning_rate": 6.3888888888888885e-06,
      "loss": 0.017,
      "step": 5880
    },
    {
      "epoch": 0.6806712962962963,
      "grad_norm": 0.07322627305984497,
      "learning_rate": 6.3865740740740745e-06,
      "loss": 0.0114,
      "step": 5881
    },
    {
      "epoch": 0.680787037037037,
      "grad_norm": 0.070510134100914,
      "learning_rate": 6.38425925925926e-06,
      "loss": 0.0127,
      "step": 5882
    },
    {
      "epoch": 0.6809027777777777,
      "grad_norm": 49.79607009887695,
      "learning_rate": 6.381944444444445e-06,
      "loss": 2.5669,
      "step": 5883
    },
    {
      "epoch": 0.6810185185185185,
      "grad_norm": 0.059024304151535034,
      "learning_rate": 6.37962962962963e-06,
      "loss": 0.0108,
      "step": 5884
    },
    {
      "epoch": 0.6811342592592593,
      "grad_norm": 0.06535336375236511,
      "learning_rate": 6.377314814814816e-06,
      "loss": 0.0112,
      "step": 5885
    },
    {
      "epoch": 0.68125,
      "grad_norm": 38.55230712890625,
      "learning_rate": 6.375e-06,
      "loss": 2.6278,
      "step": 5886
    },
    {
      "epoch": 0.6813657407407407,
      "grad_norm": 0.06611545383930206,
      "learning_rate": 6.372685185185186e-06,
      "loss": 0.0112,
      "step": 5887
    },
    {
      "epoch": 0.6814814814814815,
      "grad_norm": 0.04481026902794838,
      "learning_rate": 6.370370370370371e-06,
      "loss": 0.0082,
      "step": 5888
    },
    {
      "epoch": 0.6815972222222222,
      "grad_norm": 0.04635719582438469,
      "learning_rate": 6.368055555555555e-06,
      "loss": 0.0085,
      "step": 5889
    },
    {
      "epoch": 0.6817129629629629,
      "grad_norm": 0.06045660004019737,
      "learning_rate": 6.365740740740741e-06,
      "loss": 0.0105,
      "step": 5890
    },
    {
      "epoch": 0.6818287037037037,
      "grad_norm": 0.0887569859623909,
      "learning_rate": 6.363425925925927e-06,
      "loss": 0.0127,
      "step": 5891
    },
    {
      "epoch": 0.6819444444444445,
      "grad_norm": 0.06657096743583679,
      "learning_rate": 6.361111111111111e-06,
      "loss": 0.0113,
      "step": 5892
    },
    {
      "epoch": 0.6820601851851852,
      "grad_norm": 0.061221085488796234,
      "learning_rate": 6.3587962962962965e-06,
      "loss": 0.0106,
      "step": 5893
    },
    {
      "epoch": 0.6821759259259259,
      "grad_norm": 17.20086097717285,
      "learning_rate": 6.3564814814814824e-06,
      "loss": 2.8266,
      "step": 5894
    },
    {
      "epoch": 0.6822916666666666,
      "grad_norm": 0.06975475698709488,
      "learning_rate": 6.354166666666667e-06,
      "loss": 0.0125,
      "step": 5895
    },
    {
      "epoch": 0.6824074074074075,
      "grad_norm": 0.061020489782094955,
      "learning_rate": 6.351851851851853e-06,
      "loss": 0.0104,
      "step": 5896
    },
    {
      "epoch": 0.6825231481481482,
      "grad_norm": 0.25741875171661377,
      "learning_rate": 6.349537037037038e-06,
      "loss": 0.0111,
      "step": 5897
    },
    {
      "epoch": 0.6826388888888889,
      "grad_norm": 0.09102103114128113,
      "learning_rate": 6.347222222222223e-06,
      "loss": 0.015,
      "step": 5898
    },
    {
      "epoch": 0.6827546296296296,
      "grad_norm": 0.2268543690443039,
      "learning_rate": 6.344907407407408e-06,
      "loss": 0.0181,
      "step": 5899
    },
    {
      "epoch": 0.6828703703703703,
      "grad_norm": 0.055044256150722504,
      "learning_rate": 6.342592592592594e-06,
      "loss": 0.0094,
      "step": 5900
    },
    {
      "epoch": 0.6829861111111111,
      "grad_norm": 0.058898501098155975,
      "learning_rate": 6.340277777777778e-06,
      "loss": 0.0104,
      "step": 5901
    },
    {
      "epoch": 0.6831018518518519,
      "grad_norm": 0.07857163995504379,
      "learning_rate": 6.337962962962963e-06,
      "loss": 0.0144,
      "step": 5902
    },
    {
      "epoch": 0.6832175925925926,
      "grad_norm": 0.09010568261146545,
      "learning_rate": 6.335648148148149e-06,
      "loss": 0.0149,
      "step": 5903
    },
    {
      "epoch": 0.6833333333333333,
      "grad_norm": 1.0472760200500488,
      "learning_rate": 6.333333333333333e-06,
      "loss": 0.0154,
      "step": 5904
    },
    {
      "epoch": 0.6834490740740741,
      "grad_norm": 0.05929999798536301,
      "learning_rate": 6.331018518518519e-06,
      "loss": 0.0105,
      "step": 5905
    },
    {
      "epoch": 0.6835648148148148,
      "grad_norm": 0.10207715630531311,
      "learning_rate": 6.3287037037037044e-06,
      "loss": 0.0137,
      "step": 5906
    },
    {
      "epoch": 0.6836805555555555,
      "grad_norm": 0.05780675262212753,
      "learning_rate": 6.3263888888888895e-06,
      "loss": 0.0076,
      "step": 5907
    },
    {
      "epoch": 0.6837962962962963,
      "grad_norm": 0.057964589446783066,
      "learning_rate": 6.324074074074075e-06,
      "loss": 0.0105,
      "step": 5908
    },
    {
      "epoch": 0.6839120370370371,
      "grad_norm": 0.059836551547050476,
      "learning_rate": 6.321759259259261e-06,
      "loss": 0.011,
      "step": 5909
    },
    {
      "epoch": 0.6840277777777778,
      "grad_norm": 0.04870098456740379,
      "learning_rate": 6.319444444444445e-06,
      "loss": 0.0083,
      "step": 5910
    },
    {
      "epoch": 0.6841435185185185,
      "grad_norm": 252.90489196777344,
      "learning_rate": 6.31712962962963e-06,
      "loss": 0.9068,
      "step": 5911
    },
    {
      "epoch": 0.6842592592592592,
      "grad_norm": 0.07411281019449234,
      "learning_rate": 6.314814814814816e-06,
      "loss": 0.0108,
      "step": 5912
    },
    {
      "epoch": 0.684375,
      "grad_norm": 0.22284173965454102,
      "learning_rate": 6.3125e-06,
      "loss": 0.0182,
      "step": 5913
    },
    {
      "epoch": 0.6844907407407408,
      "grad_norm": 0.06594869494438171,
      "learning_rate": 6.310185185185186e-06,
      "loss": 0.0116,
      "step": 5914
    },
    {
      "epoch": 0.6846064814814815,
      "grad_norm": 0.1367354691028595,
      "learning_rate": 6.307870370370371e-06,
      "loss": 0.0119,
      "step": 5915
    },
    {
      "epoch": 0.6847222222222222,
      "grad_norm": 50.34085464477539,
      "learning_rate": 6.305555555555556e-06,
      "loss": 2.1675,
      "step": 5916
    },
    {
      "epoch": 0.684837962962963,
      "grad_norm": 0.0677107647061348,
      "learning_rate": 6.303240740740741e-06,
      "loss": 0.0107,
      "step": 5917
    },
    {
      "epoch": 0.6849537037037037,
      "grad_norm": 0.12676571309566498,
      "learning_rate": 6.300925925925926e-06,
      "loss": 0.0142,
      "step": 5918
    },
    {
      "epoch": 0.6850694444444444,
      "grad_norm": 0.057969506829977036,
      "learning_rate": 6.2986111111111116e-06,
      "loss": 0.0101,
      "step": 5919
    },
    {
      "epoch": 0.6851851851851852,
      "grad_norm": 106.39697265625,
      "learning_rate": 6.296296296296297e-06,
      "loss": 0.5055,
      "step": 5920
    },
    {
      "epoch": 0.685300925925926,
      "grad_norm": 42.378910064697266,
      "learning_rate": 6.293981481481482e-06,
      "loss": 2.6619,
      "step": 5921
    },
    {
      "epoch": 0.6854166666666667,
      "grad_norm": 154.70745849609375,
      "learning_rate": 6.291666666666667e-06,
      "loss": 1.2885,
      "step": 5922
    },
    {
      "epoch": 0.6855324074074074,
      "grad_norm": 0.04482464864850044,
      "learning_rate": 6.289351851851853e-06,
      "loss": 0.0081,
      "step": 5923
    },
    {
      "epoch": 0.6856481481481481,
      "grad_norm": 2.2289273738861084,
      "learning_rate": 6.287037037037037e-06,
      "loss": 0.0203,
      "step": 5924
    },
    {
      "epoch": 0.6857638888888888,
      "grad_norm": 0.04749139025807381,
      "learning_rate": 6.284722222222223e-06,
      "loss": 0.0086,
      "step": 5925
    },
    {
      "epoch": 0.6858796296296297,
      "grad_norm": 0.09677649289369583,
      "learning_rate": 6.282407407407408e-06,
      "loss": 0.0136,
      "step": 5926
    },
    {
      "epoch": 0.6859953703703704,
      "grad_norm": 0.0814153254032135,
      "learning_rate": 6.280092592592592e-06,
      "loss": 0.0149,
      "step": 5927
    },
    {
      "epoch": 0.6861111111111111,
      "grad_norm": 0.049103498458862305,
      "learning_rate": 6.277777777777778e-06,
      "loss": 0.009,
      "step": 5928
    },
    {
      "epoch": 0.6862268518518518,
      "grad_norm": 0.06308738142251968,
      "learning_rate": 6.275462962962963e-06,
      "loss": 0.0113,
      "step": 5929
    },
    {
      "epoch": 0.6863425925925926,
      "grad_norm": 0.13850906491279602,
      "learning_rate": 6.2731481481481485e-06,
      "loss": 0.0107,
      "step": 5930
    },
    {
      "epoch": 0.6864583333333333,
      "grad_norm": 0.06670866161584854,
      "learning_rate": 6.2708333333333336e-06,
      "loss": 0.012,
      "step": 5931
    },
    {
      "epoch": 0.6865740740740741,
      "grad_norm": 0.05309121683239937,
      "learning_rate": 6.2685185185185195e-06,
      "loss": 0.0089,
      "step": 5932
    },
    {
      "epoch": 0.6866898148148148,
      "grad_norm": 0.062403351068496704,
      "learning_rate": 6.266203703703704e-06,
      "loss": 0.0111,
      "step": 5933
    },
    {
      "epoch": 0.6868055555555556,
      "grad_norm": 0.07361616939306259,
      "learning_rate": 6.26388888888889e-06,
      "loss": 0.011,
      "step": 5934
    },
    {
      "epoch": 0.6869212962962963,
      "grad_norm": 0.04995427653193474,
      "learning_rate": 6.261574074074075e-06,
      "loss": 0.0091,
      "step": 5935
    },
    {
      "epoch": 0.687037037037037,
      "grad_norm": 0.09930180758237839,
      "learning_rate": 6.259259259259259e-06,
      "loss": 0.0129,
      "step": 5936
    },
    {
      "epoch": 0.6871527777777777,
      "grad_norm": 0.060580234974622726,
      "learning_rate": 6.256944444444445e-06,
      "loss": 0.0106,
      "step": 5937
    },
    {
      "epoch": 0.6872685185185186,
      "grad_norm": 0.0650462806224823,
      "learning_rate": 6.25462962962963e-06,
      "loss": 0.0111,
      "step": 5938
    },
    {
      "epoch": 0.6873842592592593,
      "grad_norm": 0.06239353120326996,
      "learning_rate": 6.252314814814815e-06,
      "loss": 0.0109,
      "step": 5939
    },
    {
      "epoch": 0.6875,
      "grad_norm": 0.08404618501663208,
      "learning_rate": 6.25e-06,
      "loss": 0.0111,
      "step": 5940
    },
    {
      "epoch": 0.6876157407407407,
      "grad_norm": 0.07211413979530334,
      "learning_rate": 6.247685185185186e-06,
      "loss": 0.0089,
      "step": 5941
    },
    {
      "epoch": 0.6877314814814814,
      "grad_norm": 0.15441039204597473,
      "learning_rate": 6.2453703703703705e-06,
      "loss": 0.0158,
      "step": 5942
    },
    {
      "epoch": 0.6878472222222223,
      "grad_norm": 0.05662447214126587,
      "learning_rate": 6.2430555555555564e-06,
      "loss": 0.0103,
      "step": 5943
    },
    {
      "epoch": 0.687962962962963,
      "grad_norm": 100.66303253173828,
      "learning_rate": 6.2407407407407415e-06,
      "loss": 1.3289,
      "step": 5944
    },
    {
      "epoch": 0.6880787037037037,
      "grad_norm": 0.14672958850860596,
      "learning_rate": 6.238425925925926e-06,
      "loss": 0.0102,
      "step": 5945
    },
    {
      "epoch": 0.6881944444444444,
      "grad_norm": 0.06488870084285736,
      "learning_rate": 6.236111111111112e-06,
      "loss": 0.0119,
      "step": 5946
    },
    {
      "epoch": 0.6883101851851852,
      "grad_norm": 0.1064506322145462,
      "learning_rate": 6.233796296296297e-06,
      "loss": 0.0121,
      "step": 5947
    },
    {
      "epoch": 0.6884259259259259,
      "grad_norm": 101.82066345214844,
      "learning_rate": 6.231481481481482e-06,
      "loss": 0.8046,
      "step": 5948
    },
    {
      "epoch": 0.6885416666666667,
      "grad_norm": 0.07473767548799515,
      "learning_rate": 6.229166666666667e-06,
      "loss": 0.0133,
      "step": 5949
    },
    {
      "epoch": 0.6886574074074074,
      "grad_norm": 0.07737439125776291,
      "learning_rate": 6.226851851851853e-06,
      "loss": 0.0137,
      "step": 5950
    },
    {
      "epoch": 0.6887731481481482,
      "grad_norm": 0.07521620392799377,
      "learning_rate": 6.224537037037037e-06,
      "loss": 0.0139,
      "step": 5951
    },
    {
      "epoch": 0.6888888888888889,
      "grad_norm": 0.06079183891415596,
      "learning_rate": 6.222222222222223e-06,
      "loss": 0.0103,
      "step": 5952
    },
    {
      "epoch": 0.6890046296296296,
      "grad_norm": 0.08276548236608505,
      "learning_rate": 6.219907407407408e-06,
      "loss": 0.011,
      "step": 5953
    },
    {
      "epoch": 0.6891203703703703,
      "grad_norm": 0.06197010353207588,
      "learning_rate": 6.2175925925925925e-06,
      "loss": 0.0108,
      "step": 5954
    },
    {
      "epoch": 0.6892361111111112,
      "grad_norm": 0.055188585072755814,
      "learning_rate": 6.2152777777777785e-06,
      "loss": 0.0099,
      "step": 5955
    },
    {
      "epoch": 0.6893518518518519,
      "grad_norm": 0.05896363779902458,
      "learning_rate": 6.2129629629629636e-06,
      "loss": 0.0107,
      "step": 5956
    },
    {
      "epoch": 0.6894675925925926,
      "grad_norm": 1.3362014293670654,
      "learning_rate": 6.210648148148149e-06,
      "loss": 0.0138,
      "step": 5957
    },
    {
      "epoch": 0.6895833333333333,
      "grad_norm": 0.04557786136865616,
      "learning_rate": 6.208333333333334e-06,
      "loss": 0.0083,
      "step": 5958
    },
    {
      "epoch": 0.689699074074074,
      "grad_norm": 8.714345932006836,
      "learning_rate": 6.20601851851852e-06,
      "loss": 3.2687,
      "step": 5959
    },
    {
      "epoch": 0.6898148148148148,
      "grad_norm": 0.07806319743394852,
      "learning_rate": 6.203703703703704e-06,
      "loss": 0.0142,
      "step": 5960
    },
    {
      "epoch": 0.6899305555555556,
      "grad_norm": 93.74668884277344,
      "learning_rate": 6.20138888888889e-06,
      "loss": 0.6502,
      "step": 5961
    },
    {
      "epoch": 0.6900462962962963,
      "grad_norm": 0.048212677240371704,
      "learning_rate": 6.199074074074075e-06,
      "loss": 0.0087,
      "step": 5962
    },
    {
      "epoch": 0.690162037037037,
      "grad_norm": 0.05913437530398369,
      "learning_rate": 6.196759259259259e-06,
      "loss": 0.0102,
      "step": 5963
    },
    {
      "epoch": 0.6902777777777778,
      "grad_norm": 0.06562261283397675,
      "learning_rate": 6.194444444444445e-06,
      "loss": 0.0118,
      "step": 5964
    },
    {
      "epoch": 0.6903935185185185,
      "grad_norm": 0.048208087682724,
      "learning_rate": 6.1921296296296294e-06,
      "loss": 0.0087,
      "step": 5965
    },
    {
      "epoch": 0.6905092592592592,
      "grad_norm": 0.059795740991830826,
      "learning_rate": 6.189814814814815e-06,
      "loss": 0.0109,
      "step": 5966
    },
    {
      "epoch": 0.690625,
      "grad_norm": 3.3568968772888184,
      "learning_rate": 6.1875000000000005e-06,
      "loss": 0.037,
      "step": 5967
    },
    {
      "epoch": 0.6907407407407408,
      "grad_norm": 0.05026538670063019,
      "learning_rate": 6.1851851851851856e-06,
      "loss": 0.0091,
      "step": 5968
    },
    {
      "epoch": 0.6908564814814815,
      "grad_norm": 0.07396069914102554,
      "learning_rate": 6.182870370370371e-06,
      "loss": 0.0125,
      "step": 5969
    },
    {
      "epoch": 0.6909722222222222,
      "grad_norm": 142.24618530273438,
      "learning_rate": 6.180555555555557e-06,
      "loss": 0.274,
      "step": 5970
    },
    {
      "epoch": 0.6910879629629629,
      "grad_norm": 0.08828379213809967,
      "learning_rate": 6.178240740740741e-06,
      "loss": 0.0143,
      "step": 5971
    },
    {
      "epoch": 0.6912037037037037,
      "grad_norm": 0.06010507419705391,
      "learning_rate": 6.175925925925926e-06,
      "loss": 0.0095,
      "step": 5972
    },
    {
      "epoch": 0.6913194444444445,
      "grad_norm": 0.04370716214179993,
      "learning_rate": 6.173611111111112e-06,
      "loss": 0.0079,
      "step": 5973
    },
    {
      "epoch": 0.6914351851851852,
      "grad_norm": 0.15172792971134186,
      "learning_rate": 6.171296296296296e-06,
      "loss": 0.0119,
      "step": 5974
    },
    {
      "epoch": 0.6915509259259259,
      "grad_norm": 226.14028930664062,
      "learning_rate": 6.168981481481482e-06,
      "loss": 0.9007,
      "step": 5975
    },
    {
      "epoch": 0.6916666666666667,
      "grad_norm": 0.23718726634979248,
      "learning_rate": 6.166666666666667e-06,
      "loss": 0.018,
      "step": 5976
    },
    {
      "epoch": 0.6917824074074074,
      "grad_norm": 0.0888969898223877,
      "learning_rate": 6.164351851851852e-06,
      "loss": 0.0118,
      "step": 5977
    },
    {
      "epoch": 0.6918981481481481,
      "grad_norm": 0.06579954922199249,
      "learning_rate": 6.162037037037037e-06,
      "loss": 0.011,
      "step": 5978
    },
    {
      "epoch": 0.6920138888888889,
      "grad_norm": 0.06063167005777359,
      "learning_rate": 6.159722222222223e-06,
      "loss": 0.0103,
      "step": 5979
    },
    {
      "epoch": 0.6921296296296297,
      "grad_norm": 0.07759792357683182,
      "learning_rate": 6.157407407407408e-06,
      "loss": 0.0107,
      "step": 5980
    },
    {
      "epoch": 0.6922453703703704,
      "grad_norm": 0.05066325515508652,
      "learning_rate": 6.155092592592593e-06,
      "loss": 0.0088,
      "step": 5981
    },
    {
      "epoch": 0.6923611111111111,
      "grad_norm": 0.45297086238861084,
      "learning_rate": 6.152777777777779e-06,
      "loss": 0.0148,
      "step": 5982
    },
    {
      "epoch": 0.6924768518518518,
      "grad_norm": 0.04739179462194443,
      "learning_rate": 6.150462962962963e-06,
      "loss": 0.0086,
      "step": 5983
    },
    {
      "epoch": 0.6925925925925925,
      "grad_norm": 0.06378881633281708,
      "learning_rate": 6.148148148148149e-06,
      "loss": 0.0112,
      "step": 5984
    },
    {
      "epoch": 0.6927083333333334,
      "grad_norm": 0.044129401445388794,
      "learning_rate": 6.145833333333334e-06,
      "loss": 0.0079,
      "step": 5985
    },
    {
      "epoch": 0.6928240740740741,
      "grad_norm": 0.09193482249975204,
      "learning_rate": 6.143518518518519e-06,
      "loss": 0.0129,
      "step": 5986
    },
    {
      "epoch": 0.6929398148148148,
      "grad_norm": 0.06955322623252869,
      "learning_rate": 6.141203703703704e-06,
      "loss": 0.0091,
      "step": 5987
    },
    {
      "epoch": 0.6930555555555555,
      "grad_norm": 0.05732519179582596,
      "learning_rate": 6.13888888888889e-06,
      "loss": 0.0075,
      "step": 5988
    },
    {
      "epoch": 0.6931712962962963,
      "grad_norm": 0.06586787104606628,
      "learning_rate": 6.136574074074074e-06,
      "loss": 0.0113,
      "step": 5989
    },
    {
      "epoch": 0.6932870370370371,
      "grad_norm": 0.06022913381457329,
      "learning_rate": 6.134259259259259e-06,
      "loss": 0.0108,
      "step": 5990
    },
    {
      "epoch": 0.6934027777777778,
      "grad_norm": 0.0601075179874897,
      "learning_rate": 6.131944444444445e-06,
      "loss": 0.0106,
      "step": 5991
    },
    {
      "epoch": 0.6935185185185185,
      "grad_norm": 0.07198692113161087,
      "learning_rate": 6.12962962962963e-06,
      "loss": 0.0114,
      "step": 5992
    },
    {
      "epoch": 0.6936342592592593,
      "grad_norm": 0.09521882236003876,
      "learning_rate": 6.1273148148148156e-06,
      "loss": 0.0108,
      "step": 5993
    },
    {
      "epoch": 0.69375,
      "grad_norm": 0.06757169216871262,
      "learning_rate": 6.125000000000001e-06,
      "loss": 0.0124,
      "step": 5994
    },
    {
      "epoch": 0.6938657407407407,
      "grad_norm": 15.063361167907715,
      "learning_rate": 6.122685185185186e-06,
      "loss": 2.7061,
      "step": 5995
    },
    {
      "epoch": 0.6939814814814815,
      "grad_norm": 1.1312960386276245,
      "learning_rate": 6.120370370370371e-06,
      "loss": 0.0191,
      "step": 5996
    },
    {
      "epoch": 0.6940972222222223,
      "grad_norm": 0.050716131925582886,
      "learning_rate": 6.118055555555557e-06,
      "loss": 0.0092,
      "step": 5997
    },
    {
      "epoch": 0.694212962962963,
      "grad_norm": 0.08723459392786026,
      "learning_rate": 6.115740740740741e-06,
      "loss": 0.0129,
      "step": 5998
    },
    {
      "epoch": 0.6943287037037037,
      "grad_norm": 0.06025458499789238,
      "learning_rate": 6.113425925925926e-06,
      "loss": 0.0109,
      "step": 5999
    },
    {
      "epoch": 0.6944444444444444,
      "grad_norm": 0.06900893896818161,
      "learning_rate": 6.111111111111112e-06,
      "loss": 0.012,
      "step": 6000
    },
    {
      "epoch": 0.6945601851851851,
      "grad_norm": 0.06544019281864166,
      "learning_rate": 6.108796296296296e-06,
      "loss": 0.0119,
      "step": 6001
    },
    {
      "epoch": 0.694675925925926,
      "grad_norm": 65.16226959228516,
      "learning_rate": 6.106481481481482e-06,
      "loss": 0.2373,
      "step": 6002
    },
    {
      "epoch": 0.6947916666666667,
      "grad_norm": 0.07509974390268326,
      "learning_rate": 6.104166666666667e-06,
      "loss": 0.0138,
      "step": 6003
    },
    {
      "epoch": 0.6949074074074074,
      "grad_norm": 0.05234433338046074,
      "learning_rate": 6.1018518518518525e-06,
      "loss": 0.0087,
      "step": 6004
    },
    {
      "epoch": 0.6950231481481481,
      "grad_norm": 0.12286031991243362,
      "learning_rate": 6.0995370370370376e-06,
      "loss": 0.0169,
      "step": 6005
    },
    {
      "epoch": 0.6951388888888889,
      "grad_norm": 0.07895637303590775,
      "learning_rate": 6.0972222222222235e-06,
      "loss": 0.0125,
      "step": 6006
    },
    {
      "epoch": 0.6952546296296296,
      "grad_norm": 0.06516508758068085,
      "learning_rate": 6.094907407407408e-06,
      "loss": 0.0118,
      "step": 6007
    },
    {
      "epoch": 0.6953703703703704,
      "grad_norm": 0.05956685543060303,
      "learning_rate": 6.092592592592593e-06,
      "loss": 0.0108,
      "step": 6008
    },
    {
      "epoch": 0.6954861111111111,
      "grad_norm": 0.07178018987178802,
      "learning_rate": 6.090277777777778e-06,
      "loss": 0.0114,
      "step": 6009
    },
    {
      "epoch": 0.6956018518518519,
      "grad_norm": 39.192203521728516,
      "learning_rate": 6.087962962962963e-06,
      "loss": 2.7348,
      "step": 6010
    },
    {
      "epoch": 0.6957175925925926,
      "grad_norm": 0.05115377902984619,
      "learning_rate": 6.085648148148149e-06,
      "loss": 0.0092,
      "step": 6011
    },
    {
      "epoch": 0.6958333333333333,
      "grad_norm": 0.07661112397909164,
      "learning_rate": 6.083333333333333e-06,
      "loss": 0.0133,
      "step": 6012
    },
    {
      "epoch": 0.695949074074074,
      "grad_norm": 0.057225003838539124,
      "learning_rate": 6.081018518518519e-06,
      "loss": 0.0087,
      "step": 6013
    },
    {
      "epoch": 0.6960648148148149,
      "grad_norm": 0.06124657765030861,
      "learning_rate": 6.078703703703704e-06,
      "loss": 0.0104,
      "step": 6014
    },
    {
      "epoch": 0.6961805555555556,
      "grad_norm": 0.05401258543133736,
      "learning_rate": 6.0763888888888885e-06,
      "loss": 0.0095,
      "step": 6015
    },
    {
      "epoch": 0.6962962962962963,
      "grad_norm": 0.07334597408771515,
      "learning_rate": 6.0740740740740745e-06,
      "loss": 0.0118,
      "step": 6016
    },
    {
      "epoch": 0.696412037037037,
      "grad_norm": 0.05870579555630684,
      "learning_rate": 6.0717592592592604e-06,
      "loss": 0.0102,
      "step": 6017
    },
    {
      "epoch": 0.6965277777777777,
      "grad_norm": 0.068570576608181,
      "learning_rate": 6.069444444444445e-06,
      "loss": 0.0091,
      "step": 6018
    },
    {
      "epoch": 0.6966435185185185,
      "grad_norm": 0.10365083813667297,
      "learning_rate": 6.06712962962963e-06,
      "loss": 0.015,
      "step": 6019
    },
    {
      "epoch": 0.6967592592592593,
      "grad_norm": 0.09031498432159424,
      "learning_rate": 6.064814814814816e-06,
      "loss": 0.0089,
      "step": 6020
    },
    {
      "epoch": 0.696875,
      "grad_norm": 0.04972442612051964,
      "learning_rate": 6.0625e-06,
      "loss": 0.009,
      "step": 6021
    },
    {
      "epoch": 0.6969907407407407,
      "grad_norm": 0.08519422262907028,
      "learning_rate": 6.060185185185186e-06,
      "loss": 0.0097,
      "step": 6022
    },
    {
      "epoch": 0.6971064814814815,
      "grad_norm": 0.07795627415180206,
      "learning_rate": 6.057870370370371e-06,
      "loss": 0.0133,
      "step": 6023
    },
    {
      "epoch": 0.6972222222222222,
      "grad_norm": 0.05069572106003761,
      "learning_rate": 6.055555555555555e-06,
      "loss": 0.009,
      "step": 6024
    },
    {
      "epoch": 0.6973379629629629,
      "grad_norm": 0.06660515069961548,
      "learning_rate": 6.053240740740741e-06,
      "loss": 0.0084,
      "step": 6025
    },
    {
      "epoch": 0.6974537037037037,
      "grad_norm": 0.04313162341713905,
      "learning_rate": 6.050925925925927e-06,
      "loss": 0.0078,
      "step": 6026
    },
    {
      "epoch": 0.6975694444444445,
      "grad_norm": 0.07136432081460953,
      "learning_rate": 6.048611111111111e-06,
      "loss": 0.0092,
      "step": 6027
    },
    {
      "epoch": 0.6976851851851852,
      "grad_norm": 0.053603652864694595,
      "learning_rate": 6.0462962962962965e-06,
      "loss": 0.0098,
      "step": 6028
    },
    {
      "epoch": 0.6978009259259259,
      "grad_norm": 0.05790085345506668,
      "learning_rate": 6.0439814814814825e-06,
      "loss": 0.0104,
      "step": 6029
    },
    {
      "epoch": 0.6979166666666666,
      "grad_norm": 0.06298048049211502,
      "learning_rate": 6.041666666666667e-06,
      "loss": 0.0116,
      "step": 6030
    },
    {
      "epoch": 0.6980324074074075,
      "grad_norm": 0.07394546270370483,
      "learning_rate": 6.039351851851853e-06,
      "loss": 0.0133,
      "step": 6031
    },
    {
      "epoch": 0.6981481481481482,
      "grad_norm": 0.04200766608119011,
      "learning_rate": 6.037037037037038e-06,
      "loss": 0.0076,
      "step": 6032
    },
    {
      "epoch": 0.6982638888888889,
      "grad_norm": 0.07009239494800568,
      "learning_rate": 6.034722222222222e-06,
      "loss": 0.0123,
      "step": 6033
    },
    {
      "epoch": 0.6983796296296296,
      "grad_norm": 0.051310598850250244,
      "learning_rate": 6.032407407407408e-06,
      "loss": 0.0091,
      "step": 6034
    },
    {
      "epoch": 0.6984953703703703,
      "grad_norm": 0.06769533455371857,
      "learning_rate": 6.030092592592594e-06,
      "loss": 0.0112,
      "step": 6035
    },
    {
      "epoch": 0.6986111111111111,
      "grad_norm": 0.04619784280657768,
      "learning_rate": 6.027777777777778e-06,
      "loss": 0.0084,
      "step": 6036
    },
    {
      "epoch": 0.6987268518518519,
      "grad_norm": 0.05583826079964638,
      "learning_rate": 6.025462962962963e-06,
      "loss": 0.0099,
      "step": 6037
    },
    {
      "epoch": 0.6988425925925926,
      "grad_norm": 0.058331649750471115,
      "learning_rate": 6.023148148148149e-06,
      "loss": 0.0107,
      "step": 6038
    },
    {
      "epoch": 0.6989583333333333,
      "grad_norm": 0.06233025714755058,
      "learning_rate": 6.0208333333333334e-06,
      "loss": 0.0115,
      "step": 6039
    },
    {
      "epoch": 0.6990740740740741,
      "grad_norm": 0.06286951154470444,
      "learning_rate": 6.018518518518519e-06,
      "loss": 0.0109,
      "step": 6040
    },
    {
      "epoch": 0.6991898148148148,
      "grad_norm": 0.08075491338968277,
      "learning_rate": 6.0162037037037045e-06,
      "loss": 0.0118,
      "step": 6041
    },
    {
      "epoch": 0.6993055555555555,
      "grad_norm": 66.7684555053711,
      "learning_rate": 6.013888888888889e-06,
      "loss": 0.1793,
      "step": 6042
    },
    {
      "epoch": 0.6994212962962963,
      "grad_norm": 0.1027633547782898,
      "learning_rate": 6.011574074074075e-06,
      "loss": 0.0128,
      "step": 6043
    },
    {
      "epoch": 0.6995370370370371,
      "grad_norm": 0.07866506278514862,
      "learning_rate": 6.009259259259261e-06,
      "loss": 0.0134,
      "step": 6044
    },
    {
      "epoch": 0.6996527777777778,
      "grad_norm": 0.2518162727355957,
      "learning_rate": 6.006944444444445e-06,
      "loss": 0.0116,
      "step": 6045
    },
    {
      "epoch": 0.6997685185185185,
      "grad_norm": 0.08813345432281494,
      "learning_rate": 6.00462962962963e-06,
      "loss": 0.0113,
      "step": 6046
    },
    {
      "epoch": 0.6998842592592592,
      "grad_norm": 0.057041481137275696,
      "learning_rate": 6.002314814814816e-06,
      "loss": 0.0101,
      "step": 6047
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.0669204369187355,
      "learning_rate": 6e-06,
      "loss": 0.0115,
      "step": 6048
    },
    {
      "epoch": 0.7001157407407408,
      "grad_norm": 0.06647197157144547,
      "learning_rate": 5.997685185185186e-06,
      "loss": 0.011,
      "step": 6049
    },
    {
      "epoch": 0.7002314814814815,
      "grad_norm": 0.05101609230041504,
      "learning_rate": 5.995370370370371e-06,
      "loss": 0.0091,
      "step": 6050
    },
    {
      "epoch": 0.7003472222222222,
      "grad_norm": 0.05512618646025658,
      "learning_rate": 5.9930555555555554e-06,
      "loss": 0.0099,
      "step": 6051
    },
    {
      "epoch": 0.700462962962963,
      "grad_norm": 0.054674871265888214,
      "learning_rate": 5.990740740740741e-06,
      "loss": 0.0098,
      "step": 6052
    },
    {
      "epoch": 0.7005787037037037,
      "grad_norm": 0.05543304234743118,
      "learning_rate": 5.988425925925926e-06,
      "loss": 0.0085,
      "step": 6053
    },
    {
      "epoch": 0.7006944444444444,
      "grad_norm": 0.07051464170217514,
      "learning_rate": 5.986111111111112e-06,
      "loss": 0.0129,
      "step": 6054
    },
    {
      "epoch": 0.7008101851851852,
      "grad_norm": 0.05582568794488907,
      "learning_rate": 5.983796296296297e-06,
      "loss": 0.0101,
      "step": 6055
    },
    {
      "epoch": 0.700925925925926,
      "grad_norm": 0.06880651414394379,
      "learning_rate": 5.981481481481482e-06,
      "loss": 0.0121,
      "step": 6056
    },
    {
      "epoch": 0.7010416666666667,
      "grad_norm": 0.04438579082489014,
      "learning_rate": 5.979166666666667e-06,
      "loss": 0.0079,
      "step": 6057
    },
    {
      "epoch": 0.7011574074074074,
      "grad_norm": 0.06843146681785583,
      "learning_rate": 5.976851851851853e-06,
      "loss": 0.0115,
      "step": 6058
    },
    {
      "epoch": 0.7012731481481481,
      "grad_norm": 0.04211997613310814,
      "learning_rate": 5.974537037037037e-06,
      "loss": 0.0077,
      "step": 6059
    },
    {
      "epoch": 0.7013888888888888,
      "grad_norm": 0.06698784977197647,
      "learning_rate": 5.972222222222222e-06,
      "loss": 0.012,
      "step": 6060
    },
    {
      "epoch": 0.7015046296296297,
      "grad_norm": 169.1658935546875,
      "learning_rate": 5.969907407407408e-06,
      "loss": 0.9273,
      "step": 6061
    },
    {
      "epoch": 0.7016203703703704,
      "grad_norm": 0.06578782945871353,
      "learning_rate": 5.967592592592592e-06,
      "loss": 0.0118,
      "step": 6062
    },
    {
      "epoch": 0.7017361111111111,
      "grad_norm": 0.48612985014915466,
      "learning_rate": 5.965277777777778e-06,
      "loss": 0.0191,
      "step": 6063
    },
    {
      "epoch": 0.7018518518518518,
      "grad_norm": 0.06433068215847015,
      "learning_rate": 5.962962962962963e-06,
      "loss": 0.0117,
      "step": 6064
    },
    {
      "epoch": 0.7019675925925926,
      "grad_norm": 0.09083855152130127,
      "learning_rate": 5.9606481481481485e-06,
      "loss": 0.0121,
      "step": 6065
    },
    {
      "epoch": 0.7020833333333333,
      "grad_norm": 0.1913684755563736,
      "learning_rate": 5.958333333333334e-06,
      "loss": 0.0106,
      "step": 6066
    },
    {
      "epoch": 0.7021990740740741,
      "grad_norm": 0.0644315704703331,
      "learning_rate": 5.9560185185185195e-06,
      "loss": 0.0115,
      "step": 6067
    },
    {
      "epoch": 0.7023148148148148,
      "grad_norm": 0.055498745292425156,
      "learning_rate": 5.953703703703704e-06,
      "loss": 0.0073,
      "step": 6068
    },
    {
      "epoch": 0.7024305555555556,
      "grad_norm": 0.03973383828997612,
      "learning_rate": 5.95138888888889e-06,
      "loss": 0.0073,
      "step": 6069
    },
    {
      "epoch": 0.7025462962962963,
      "grad_norm": 0.06478244811296463,
      "learning_rate": 5.949074074074075e-06,
      "loss": 0.0117,
      "step": 6070
    },
    {
      "epoch": 0.702662037037037,
      "grad_norm": 0.05946347117424011,
      "learning_rate": 5.946759259259259e-06,
      "loss": 0.0107,
      "step": 6071
    },
    {
      "epoch": 0.7027777777777777,
      "grad_norm": 0.05514749512076378,
      "learning_rate": 5.944444444444445e-06,
      "loss": 0.0099,
      "step": 6072
    },
    {
      "epoch": 0.7028935185185186,
      "grad_norm": 0.05557941645383835,
      "learning_rate": 5.94212962962963e-06,
      "loss": 0.0098,
      "step": 6073
    },
    {
      "epoch": 0.7030092592592593,
      "grad_norm": 0.05718318000435829,
      "learning_rate": 5.939814814814815e-06,
      "loss": 0.0104,
      "step": 6074
    },
    {
      "epoch": 0.703125,
      "grad_norm": 0.11546587198972702,
      "learning_rate": 5.9375e-06,
      "loss": 0.0143,
      "step": 6075
    },
    {
      "epoch": 0.7032407407407407,
      "grad_norm": 0.056207478046417236,
      "learning_rate": 5.935185185185186e-06,
      "loss": 0.0103,
      "step": 6076
    },
    {
      "epoch": 0.7033564814814814,
      "grad_norm": 0.06969518214464188,
      "learning_rate": 5.9328703703703705e-06,
      "loss": 0.0122,
      "step": 6077
    },
    {
      "epoch": 0.7034722222222223,
      "grad_norm": 0.06034039333462715,
      "learning_rate": 5.9305555555555565e-06,
      "loss": 0.0109,
      "step": 6078
    },
    {
      "epoch": 0.703587962962963,
      "grad_norm": 0.07057183235883713,
      "learning_rate": 5.9282407407407416e-06,
      "loss": 0.0129,
      "step": 6079
    },
    {
      "epoch": 0.7037037037037037,
      "grad_norm": 0.07131408154964447,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.011,
      "step": 6080
    },
    {
      "epoch": 0.7038194444444444,
      "grad_norm": 0.06495439261198044,
      "learning_rate": 5.923611111111112e-06,
      "loss": 0.0107,
      "step": 6081
    },
    {
      "epoch": 0.7039351851851852,
      "grad_norm": 0.06389359384775162,
      "learning_rate": 5.921296296296297e-06,
      "loss": 0.0114,
      "step": 6082
    },
    {
      "epoch": 0.7040509259259259,
      "grad_norm": 0.06684154272079468,
      "learning_rate": 5.918981481481482e-06,
      "loss": 0.0121,
      "step": 6083
    },
    {
      "epoch": 0.7041666666666667,
      "grad_norm": 0.06605520844459534,
      "learning_rate": 5.916666666666667e-06,
      "loss": 0.0118,
      "step": 6084
    },
    {
      "epoch": 0.7042824074074074,
      "grad_norm": 0.6153864860534668,
      "learning_rate": 5.914351851851853e-06,
      "loss": 0.0136,
      "step": 6085
    },
    {
      "epoch": 0.7043981481481482,
      "grad_norm": 0.05397415533661842,
      "learning_rate": 5.912037037037037e-06,
      "loss": 0.0097,
      "step": 6086
    },
    {
      "epoch": 0.7045138888888889,
      "grad_norm": 0.07227268069982529,
      "learning_rate": 5.909722222222223e-06,
      "loss": 0.0121,
      "step": 6087
    },
    {
      "epoch": 0.7046296296296296,
      "grad_norm": 0.06259910762310028,
      "learning_rate": 5.907407407407408e-06,
      "loss": 0.0108,
      "step": 6088
    },
    {
      "epoch": 0.7047453703703703,
      "grad_norm": 0.042549144476652145,
      "learning_rate": 5.9050925925925925e-06,
      "loss": 0.0075,
      "step": 6089
    },
    {
      "epoch": 0.7048611111111112,
      "grad_norm": 0.06676974892616272,
      "learning_rate": 5.9027777777777785e-06,
      "loss": 0.0088,
      "step": 6090
    },
    {
      "epoch": 0.7049768518518519,
      "grad_norm": 0.038971908390522,
      "learning_rate": 5.900462962962964e-06,
      "loss": 0.0072,
      "step": 6091
    },
    {
      "epoch": 0.7050925925925926,
      "grad_norm": 0.07252659648656845,
      "learning_rate": 5.898148148148149e-06,
      "loss": 0.0114,
      "step": 6092
    },
    {
      "epoch": 0.7052083333333333,
      "grad_norm": 0.04397541284561157,
      "learning_rate": 5.895833333333334e-06,
      "loss": 0.008,
      "step": 6093
    },
    {
      "epoch": 0.705324074074074,
      "grad_norm": 0.05059480294585228,
      "learning_rate": 5.89351851851852e-06,
      "loss": 0.0085,
      "step": 6094
    },
    {
      "epoch": 0.7054398148148148,
      "grad_norm": 0.06711678951978683,
      "learning_rate": 5.891203703703704e-06,
      "loss": 0.0088,
      "step": 6095
    },
    {
      "epoch": 0.7055555555555556,
      "grad_norm": 0.05154307931661606,
      "learning_rate": 5.88888888888889e-06,
      "loss": 0.0091,
      "step": 6096
    },
    {
      "epoch": 0.7056712962962963,
      "grad_norm": 0.05827663466334343,
      "learning_rate": 5.886574074074075e-06,
      "loss": 0.0106,
      "step": 6097
    },
    {
      "epoch": 0.705787037037037,
      "grad_norm": 0.055458854883909225,
      "learning_rate": 5.884259259259259e-06,
      "loss": 0.0073,
      "step": 6098
    },
    {
      "epoch": 0.7059027777777778,
      "grad_norm": 0.06058012321591377,
      "learning_rate": 5.881944444444445e-06,
      "loss": 0.0102,
      "step": 6099
    },
    {
      "epoch": 0.7060185185185185,
      "grad_norm": 0.05478394404053688,
      "learning_rate": 5.8796296296296295e-06,
      "loss": 0.0098,
      "step": 6100
    },
    {
      "epoch": 0.7061342592592592,
      "grad_norm": 0.18294356763362885,
      "learning_rate": 5.877314814814815e-06,
      "loss": 0.0097,
      "step": 6101
    },
    {
      "epoch": 0.70625,
      "grad_norm": 12.506284713745117,
      "learning_rate": 5.8750000000000005e-06,
      "loss": 2.9393,
      "step": 6102
    },
    {
      "epoch": 0.7063657407407408,
      "grad_norm": 0.0559241957962513,
      "learning_rate": 5.872685185185185e-06,
      "loss": 0.01,
      "step": 6103
    },
    {
      "epoch": 0.7064814814814815,
      "grad_norm": 0.05902887508273125,
      "learning_rate": 5.870370370370371e-06,
      "loss": 0.0106,
      "step": 6104
    },
    {
      "epoch": 0.7065972222222222,
      "grad_norm": 0.2845039963722229,
      "learning_rate": 5.868055555555557e-06,
      "loss": 0.0103,
      "step": 6105
    },
    {
      "epoch": 0.7067129629629629,
      "grad_norm": 0.04992193728685379,
      "learning_rate": 5.865740740740741e-06,
      "loss": 0.0091,
      "step": 6106
    },
    {
      "epoch": 0.7068287037037037,
      "grad_norm": 0.04321294650435448,
      "learning_rate": 5.863425925925926e-06,
      "loss": 0.0079,
      "step": 6107
    },
    {
      "epoch": 0.7069444444444445,
      "grad_norm": 0.056441228836774826,
      "learning_rate": 5.861111111111112e-06,
      "loss": 0.0102,
      "step": 6108
    },
    {
      "epoch": 0.7070601851851852,
      "grad_norm": 0.07855551689863205,
      "learning_rate": 5.858796296296296e-06,
      "loss": 0.0113,
      "step": 6109
    },
    {
      "epoch": 0.7071759259259259,
      "grad_norm": 0.054561737924814224,
      "learning_rate": 5.856481481481482e-06,
      "loss": 0.0097,
      "step": 6110
    },
    {
      "epoch": 0.7072916666666667,
      "grad_norm": 0.13264407217502594,
      "learning_rate": 5.854166666666667e-06,
      "loss": 0.0103,
      "step": 6111
    },
    {
      "epoch": 0.7074074074074074,
      "grad_norm": 0.039942096918821335,
      "learning_rate": 5.8518518518518515e-06,
      "loss": 0.0072,
      "step": 6112
    },
    {
      "epoch": 0.7075231481481481,
      "grad_norm": 22.66983985900879,
      "learning_rate": 5.849537037037037e-06,
      "loss": 0.0591,
      "step": 6113
    },
    {
      "epoch": 0.7076388888888889,
      "grad_norm": 7.178468704223633,
      "learning_rate": 5.847222222222223e-06,
      "loss": 3.3062,
      "step": 6114
    },
    {
      "epoch": 0.7077546296296297,
      "grad_norm": 0.11636004596948624,
      "learning_rate": 5.844907407407408e-06,
      "loss": 0.0151,
      "step": 6115
    },
    {
      "epoch": 0.7078703703703704,
      "grad_norm": 0.043767500668764114,
      "learning_rate": 5.842592592592593e-06,
      "loss": 0.008,
      "step": 6116
    },
    {
      "epoch": 0.7079861111111111,
      "grad_norm": 0.056977663189172745,
      "learning_rate": 5.840277777777779e-06,
      "loss": 0.0104,
      "step": 6117
    },
    {
      "epoch": 0.7081018518518518,
      "grad_norm": 0.06007632240653038,
      "learning_rate": 5.837962962962963e-06,
      "loss": 0.0111,
      "step": 6118
    },
    {
      "epoch": 0.7082175925925925,
      "grad_norm": 0.07531493157148361,
      "learning_rate": 5.835648148148149e-06,
      "loss": 0.0139,
      "step": 6119
    },
    {
      "epoch": 0.7083333333333334,
      "grad_norm": 0.14442938566207886,
      "learning_rate": 5.833333333333334e-06,
      "loss": 0.0145,
      "step": 6120
    },
    {
      "epoch": 0.7084490740740741,
      "grad_norm": 0.2751260995864868,
      "learning_rate": 5.831018518518519e-06,
      "loss": 0.0116,
      "step": 6121
    },
    {
      "epoch": 0.7085648148148148,
      "grad_norm": 117.03189849853516,
      "learning_rate": 5.828703703703704e-06,
      "loss": 2.7049,
      "step": 6122
    },
    {
      "epoch": 0.7086805555555555,
      "grad_norm": 0.246259406208992,
      "learning_rate": 5.82638888888889e-06,
      "loss": 0.0122,
      "step": 6123
    },
    {
      "epoch": 0.7087962962962963,
      "grad_norm": 0.06136725842952728,
      "learning_rate": 5.824074074074074e-06,
      "loss": 0.0107,
      "step": 6124
    },
    {
      "epoch": 0.7089120370370371,
      "grad_norm": 0.07452265173196793,
      "learning_rate": 5.8217592592592594e-06,
      "loss": 0.0137,
      "step": 6125
    },
    {
      "epoch": 0.7090277777777778,
      "grad_norm": 0.056564632803201675,
      "learning_rate": 5.819444444444445e-06,
      "loss": 0.0098,
      "step": 6126
    },
    {
      "epoch": 0.7091435185185185,
      "grad_norm": 0.043714817613363266,
      "learning_rate": 5.81712962962963e-06,
      "loss": 0.0079,
      "step": 6127
    },
    {
      "epoch": 0.7092592592592593,
      "grad_norm": 0.04866707697510719,
      "learning_rate": 5.814814814814816e-06,
      "loss": 0.0085,
      "step": 6128
    },
    {
      "epoch": 0.709375,
      "grad_norm": 0.1444067358970642,
      "learning_rate": 5.812500000000001e-06,
      "loss": 0.0127,
      "step": 6129
    },
    {
      "epoch": 0.7094907407407407,
      "grad_norm": 0.0565866194665432,
      "learning_rate": 5.810185185185186e-06,
      "loss": 0.0099,
      "step": 6130
    },
    {
      "epoch": 0.7096064814814815,
      "grad_norm": 0.0645405724644661,
      "learning_rate": 5.807870370370371e-06,
      "loss": 0.0097,
      "step": 6131
    },
    {
      "epoch": 0.7097222222222223,
      "grad_norm": 0.9035977721214294,
      "learning_rate": 5.805555555555557e-06,
      "loss": 0.0149,
      "step": 6132
    },
    {
      "epoch": 0.709837962962963,
      "grad_norm": 0.07439408451318741,
      "learning_rate": 5.803240740740741e-06,
      "loss": 0.0132,
      "step": 6133
    },
    {
      "epoch": 0.7099537037037037,
      "grad_norm": 0.05620573088526726,
      "learning_rate": 5.800925925925926e-06,
      "loss": 0.01,
      "step": 6134
    },
    {
      "epoch": 0.7100694444444444,
      "grad_norm": 0.055547140538692474,
      "learning_rate": 5.798611111111112e-06,
      "loss": 0.0073,
      "step": 6135
    },
    {
      "epoch": 0.7101851851851851,
      "grad_norm": 0.07827416062355042,
      "learning_rate": 5.796296296296296e-06,
      "loss": 0.012,
      "step": 6136
    },
    {
      "epoch": 0.710300925925926,
      "grad_norm": 0.05363691970705986,
      "learning_rate": 5.793981481481482e-06,
      "loss": 0.0096,
      "step": 6137
    },
    {
      "epoch": 0.7104166666666667,
      "grad_norm": 0.10370220243930817,
      "learning_rate": 5.791666666666667e-06,
      "loss": 0.0113,
      "step": 6138
    },
    {
      "epoch": 0.7105324074074074,
      "grad_norm": 0.4009060859680176,
      "learning_rate": 5.7893518518518525e-06,
      "loss": 0.0123,
      "step": 6139
    },
    {
      "epoch": 0.7106481481481481,
      "grad_norm": 0.06187206506729126,
      "learning_rate": 5.787037037037038e-06,
      "loss": 0.0108,
      "step": 6140
    },
    {
      "epoch": 0.7107638888888889,
      "grad_norm": 59.460113525390625,
      "learning_rate": 5.7847222222222235e-06,
      "loss": 2.2219,
      "step": 6141
    },
    {
      "epoch": 0.7108796296296296,
      "grad_norm": 0.06281891465187073,
      "learning_rate": 5.782407407407408e-06,
      "loss": 0.0112,
      "step": 6142
    },
    {
      "epoch": 0.7109953703703704,
      "grad_norm": 0.0627121776342392,
      "learning_rate": 5.780092592592593e-06,
      "loss": 0.011,
      "step": 6143
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 0.06376731395721436,
      "learning_rate": 5.777777777777778e-06,
      "loss": 0.0105,
      "step": 6144
    },
    {
      "epoch": 0.7112268518518519,
      "grad_norm": 0.05709221959114075,
      "learning_rate": 5.775462962962963e-06,
      "loss": 0.0103,
      "step": 6145
    },
    {
      "epoch": 0.7113425925925926,
      "grad_norm": 0.07002921402454376,
      "learning_rate": 5.773148148148149e-06,
      "loss": 0.0127,
      "step": 6146
    },
    {
      "epoch": 0.7114583333333333,
      "grad_norm": 0.05023034289479256,
      "learning_rate": 5.770833333333333e-06,
      "loss": 0.0088,
      "step": 6147
    },
    {
      "epoch": 0.711574074074074,
      "grad_norm": 0.0898251011967659,
      "learning_rate": 5.768518518518519e-06,
      "loss": 0.0119,
      "step": 6148
    },
    {
      "epoch": 0.7116898148148149,
      "grad_norm": 0.06338169425725937,
      "learning_rate": 5.766203703703704e-06,
      "loss": 0.0116,
      "step": 6149
    },
    {
      "epoch": 0.7118055555555556,
      "grad_norm": 0.09213190525770187,
      "learning_rate": 5.7638888888888886e-06,
      "loss": 0.0086,
      "step": 6150
    },
    {
      "epoch": 0.7119212962962963,
      "grad_norm": 0.05536244064569473,
      "learning_rate": 5.7615740740740745e-06,
      "loss": 0.0099,
      "step": 6151
    },
    {
      "epoch": 0.712037037037037,
      "grad_norm": 0.04007262736558914,
      "learning_rate": 5.75925925925926e-06,
      "loss": 0.0072,
      "step": 6152
    },
    {
      "epoch": 0.7121527777777777,
      "grad_norm": 0.043783921748399734,
      "learning_rate": 5.756944444444445e-06,
      "loss": 0.008,
      "step": 6153
    },
    {
      "epoch": 0.7122685185185185,
      "grad_norm": 0.0562351755797863,
      "learning_rate": 5.75462962962963e-06,
      "loss": 0.0073,
      "step": 6154
    },
    {
      "epoch": 0.7123842592592593,
      "grad_norm": 0.07166903465986252,
      "learning_rate": 5.752314814814816e-06,
      "loss": 0.0115,
      "step": 6155
    },
    {
      "epoch": 0.7125,
      "grad_norm": 0.053288474678993225,
      "learning_rate": 5.75e-06,
      "loss": 0.007,
      "step": 6156
    },
    {
      "epoch": 0.7126157407407407,
      "grad_norm": 0.054000381380319595,
      "learning_rate": 5.747685185185186e-06,
      "loss": 0.0096,
      "step": 6157
    },
    {
      "epoch": 0.7127314814814815,
      "grad_norm": 0.05563328042626381,
      "learning_rate": 5.745370370370371e-06,
      "loss": 0.0097,
      "step": 6158
    },
    {
      "epoch": 0.7128472222222222,
      "grad_norm": 0.058862969279289246,
      "learning_rate": 5.743055555555555e-06,
      "loss": 0.0102,
      "step": 6159
    },
    {
      "epoch": 0.7129629629629629,
      "grad_norm": 0.07990153878927231,
      "learning_rate": 5.740740740740741e-06,
      "loss": 0.0115,
      "step": 6160
    },
    {
      "epoch": 0.7130787037037037,
      "grad_norm": 0.0651240199804306,
      "learning_rate": 5.738425925925926e-06,
      "loss": 0.0113,
      "step": 6161
    },
    {
      "epoch": 0.7131944444444445,
      "grad_norm": 0.04964461550116539,
      "learning_rate": 5.7361111111111114e-06,
      "loss": 0.0088,
      "step": 6162
    },
    {
      "epoch": 0.7133101851851852,
      "grad_norm": 18.73935317993164,
      "learning_rate": 5.7337962962962965e-06,
      "loss": 2.7079,
      "step": 6163
    },
    {
      "epoch": 0.7134259259259259,
      "grad_norm": 0.04389897733926773,
      "learning_rate": 5.7314814814814825e-06,
      "loss": 0.008,
      "step": 6164
    },
    {
      "epoch": 0.7135416666666666,
      "grad_norm": 0.06396189332008362,
      "learning_rate": 5.729166666666667e-06,
      "loss": 0.0084,
      "step": 6165
    },
    {
      "epoch": 0.7136574074074075,
      "grad_norm": 0.05610709637403488,
      "learning_rate": 5.726851851851853e-06,
      "loss": 0.0098,
      "step": 6166
    },
    {
      "epoch": 0.7137731481481482,
      "grad_norm": 0.06943465024232864,
      "learning_rate": 5.724537037037038e-06,
      "loss": 0.0127,
      "step": 6167
    },
    {
      "epoch": 0.7138888888888889,
      "grad_norm": 0.06023674085736275,
      "learning_rate": 5.722222222222222e-06,
      "loss": 0.0106,
      "step": 6168
    },
    {
      "epoch": 0.7140046296296296,
      "grad_norm": 25.613100051879883,
      "learning_rate": 5.719907407407408e-06,
      "loss": 2.9439,
      "step": 6169
    },
    {
      "epoch": 0.7141203703703703,
      "grad_norm": 0.06815990060567856,
      "learning_rate": 5.717592592592593e-06,
      "loss": 0.0109,
      "step": 6170
    },
    {
      "epoch": 0.7142361111111111,
      "grad_norm": 0.07344380021095276,
      "learning_rate": 5.715277777777778e-06,
      "loss": 0.0134,
      "step": 6171
    },
    {
      "epoch": 0.7143518518518519,
      "grad_norm": 0.04447329789400101,
      "learning_rate": 5.712962962962963e-06,
      "loss": 0.0081,
      "step": 6172
    },
    {
      "epoch": 0.7144675925925926,
      "grad_norm": 0.062060292810201645,
      "learning_rate": 5.710648148148149e-06,
      "loss": 0.011,
      "step": 6173
    },
    {
      "epoch": 0.7145833333333333,
      "grad_norm": 16.01982307434082,
      "learning_rate": 5.7083333333333335e-06,
      "loss": 3.1711,
      "step": 6174
    },
    {
      "epoch": 0.7146990740740741,
      "grad_norm": 0.05317884683609009,
      "learning_rate": 5.706018518518519e-06,
      "loss": 0.0095,
      "step": 6175
    },
    {
      "epoch": 0.7148148148148148,
      "grad_norm": 0.07836000621318817,
      "learning_rate": 5.7037037037037045e-06,
      "loss": 0.0108,
      "step": 6176
    },
    {
      "epoch": 0.7149305555555555,
      "grad_norm": 0.09837736189365387,
      "learning_rate": 5.701388888888889e-06,
      "loss": 0.0089,
      "step": 6177
    },
    {
      "epoch": 0.7150462962962963,
      "grad_norm": 0.053906019777059555,
      "learning_rate": 5.699074074074075e-06,
      "loss": 0.0071,
      "step": 6178
    },
    {
      "epoch": 0.7151620370370371,
      "grad_norm": 0.0432593859732151,
      "learning_rate": 5.69675925925926e-06,
      "loss": 0.0076,
      "step": 6179
    },
    {
      "epoch": 0.7152777777777778,
      "grad_norm": 0.047268837690353394,
      "learning_rate": 5.694444444444445e-06,
      "loss": 0.0083,
      "step": 6180
    },
    {
      "epoch": 0.7153935185185185,
      "grad_norm": 0.06029508635401726,
      "learning_rate": 5.69212962962963e-06,
      "loss": 0.0101,
      "step": 6181
    },
    {
      "epoch": 0.7155092592592592,
      "grad_norm": 0.06508757919073105,
      "learning_rate": 5.689814814814816e-06,
      "loss": 0.0095,
      "step": 6182
    },
    {
      "epoch": 0.715625,
      "grad_norm": 0.06875830143690109,
      "learning_rate": 5.6875e-06,
      "loss": 0.0117,
      "step": 6183
    },
    {
      "epoch": 0.7157407407407408,
      "grad_norm": 0.2511974275112152,
      "learning_rate": 5.685185185185186e-06,
      "loss": 0.013,
      "step": 6184
    },
    {
      "epoch": 0.7158564814814815,
      "grad_norm": 0.047997619956731796,
      "learning_rate": 5.682870370370371e-06,
      "loss": 0.0087,
      "step": 6185
    },
    {
      "epoch": 0.7159722222222222,
      "grad_norm": 0.05854923650622368,
      "learning_rate": 5.6805555555555555e-06,
      "loss": 0.0103,
      "step": 6186
    },
    {
      "epoch": 0.716087962962963,
      "grad_norm": 0.9754601120948792,
      "learning_rate": 5.678240740740741e-06,
      "loss": 0.0198,
      "step": 6187
    },
    {
      "epoch": 0.7162037037037037,
      "grad_norm": 0.0536750927567482,
      "learning_rate": 5.675925925925926e-06,
      "loss": 0.0095,
      "step": 6188
    },
    {
      "epoch": 0.7163194444444444,
      "grad_norm": 0.061769384890794754,
      "learning_rate": 5.673611111111112e-06,
      "loss": 0.0108,
      "step": 6189
    },
    {
      "epoch": 0.7164351851851852,
      "grad_norm": 0.056070223450660706,
      "learning_rate": 5.671296296296297e-06,
      "loss": 0.0099,
      "step": 6190
    },
    {
      "epoch": 0.716550925925926,
      "grad_norm": 0.044390805065631866,
      "learning_rate": 5.668981481481482e-06,
      "loss": 0.0079,
      "step": 6191
    },
    {
      "epoch": 0.7166666666666667,
      "grad_norm": 0.060612913221120834,
      "learning_rate": 5.666666666666667e-06,
      "loss": 0.0111,
      "step": 6192
    },
    {
      "epoch": 0.7167824074074074,
      "grad_norm": 0.09050524234771729,
      "learning_rate": 5.664351851851853e-06,
      "loss": 0.012,
      "step": 6193
    },
    {
      "epoch": 0.7168981481481481,
      "grad_norm": 0.04971607029438019,
      "learning_rate": 5.662037037037037e-06,
      "loss": 0.0088,
      "step": 6194
    },
    {
      "epoch": 0.7170138888888888,
      "grad_norm": 0.05392558127641678,
      "learning_rate": 5.659722222222222e-06,
      "loss": 0.0097,
      "step": 6195
    },
    {
      "epoch": 0.7171296296296297,
      "grad_norm": 0.062187694013118744,
      "learning_rate": 5.657407407407408e-06,
      "loss": 0.0109,
      "step": 6196
    },
    {
      "epoch": 0.7172453703703704,
      "grad_norm": 0.25016170740127563,
      "learning_rate": 5.655092592592592e-06,
      "loss": 0.0113,
      "step": 6197
    },
    {
      "epoch": 0.7173611111111111,
      "grad_norm": 0.06319776922464371,
      "learning_rate": 5.652777777777778e-06,
      "loss": 0.0111,
      "step": 6198
    },
    {
      "epoch": 0.7174768518518518,
      "grad_norm": 0.06575320661067963,
      "learning_rate": 5.6504629629629634e-06,
      "loss": 0.0118,
      "step": 6199
    },
    {
      "epoch": 0.7175925925925926,
      "grad_norm": 1.1258927583694458,
      "learning_rate": 5.6481481481481485e-06,
      "loss": 0.0178,
      "step": 6200
    },
    {
      "epoch": 0.7177083333333333,
      "grad_norm": 58.435646057128906,
      "learning_rate": 5.645833333333334e-06,
      "loss": 0.1578,
      "step": 6201
    },
    {
      "epoch": 0.7178240740740741,
      "grad_norm": 0.055501699447631836,
      "learning_rate": 5.6435185185185196e-06,
      "loss": 0.0099,
      "step": 6202
    },
    {
      "epoch": 0.7179398148148148,
      "grad_norm": 0.049180157482624054,
      "learning_rate": 5.641203703703704e-06,
      "loss": 0.0088,
      "step": 6203
    },
    {
      "epoch": 0.7180555555555556,
      "grad_norm": 0.06760713458061218,
      "learning_rate": 5.638888888888889e-06,
      "loss": 0.0124,
      "step": 6204
    },
    {
      "epoch": 0.7181712962962963,
      "grad_norm": 0.05700937658548355,
      "learning_rate": 5.636574074074075e-06,
      "loss": 0.0103,
      "step": 6205
    },
    {
      "epoch": 0.718287037037037,
      "grad_norm": 0.04374871030449867,
      "learning_rate": 5.634259259259259e-06,
      "loss": 0.0079,
      "step": 6206
    },
    {
      "epoch": 0.7184027777777777,
      "grad_norm": 0.05722634121775627,
      "learning_rate": 5.631944444444445e-06,
      "loss": 0.0104,
      "step": 6207
    },
    {
      "epoch": 0.7185185185185186,
      "grad_norm": 0.0728156715631485,
      "learning_rate": 5.62962962962963e-06,
      "loss": 0.0086,
      "step": 6208
    },
    {
      "epoch": 0.7186342592592593,
      "grad_norm": 0.05629527196288109,
      "learning_rate": 5.627314814814815e-06,
      "loss": 0.0099,
      "step": 6209
    },
    {
      "epoch": 0.71875,
      "grad_norm": 0.07331515848636627,
      "learning_rate": 5.625e-06,
      "loss": 0.0136,
      "step": 6210
    },
    {
      "epoch": 0.7188657407407407,
      "grad_norm": 0.05328478291630745,
      "learning_rate": 5.622685185185186e-06,
      "loss": 0.0095,
      "step": 6211
    },
    {
      "epoch": 0.7189814814814814,
      "grad_norm": 0.045403748750686646,
      "learning_rate": 5.6203703703703705e-06,
      "loss": 0.0081,
      "step": 6212
    },
    {
      "epoch": 0.7190972222222223,
      "grad_norm": 0.06450043618679047,
      "learning_rate": 5.618055555555556e-06,
      "loss": 0.0116,
      "step": 6213
    },
    {
      "epoch": 0.719212962962963,
      "grad_norm": 0.06422268599271774,
      "learning_rate": 5.615740740740742e-06,
      "loss": 0.0102,
      "step": 6214
    },
    {
      "epoch": 0.7193287037037037,
      "grad_norm": 0.03957229480147362,
      "learning_rate": 5.613425925925926e-06,
      "loss": 0.0073,
      "step": 6215
    },
    {
      "epoch": 0.7194444444444444,
      "grad_norm": 0.08049573749303818,
      "learning_rate": 5.611111111111112e-06,
      "loss": 0.0116,
      "step": 6216
    },
    {
      "epoch": 0.7195601851851852,
      "grad_norm": 0.12896688282489777,
      "learning_rate": 5.608796296296297e-06,
      "loss": 0.0105,
      "step": 6217
    },
    {
      "epoch": 0.7196759259259259,
      "grad_norm": 0.14470769464969635,
      "learning_rate": 5.606481481481482e-06,
      "loss": 0.0157,
      "step": 6218
    },
    {
      "epoch": 0.7197916666666667,
      "grad_norm": 38.40679168701172,
      "learning_rate": 5.604166666666667e-06,
      "loss": 2.3833,
      "step": 6219
    },
    {
      "epoch": 0.7199074074074074,
      "grad_norm": 0.05477336794137955,
      "learning_rate": 5.601851851851853e-06,
      "loss": 0.0098,
      "step": 6220
    },
    {
      "epoch": 0.7200231481481482,
      "grad_norm": 0.053030237555503845,
      "learning_rate": 5.599537037037037e-06,
      "loss": 0.0096,
      "step": 6221
    },
    {
      "epoch": 0.7201388888888889,
      "grad_norm": 0.061253681778907776,
      "learning_rate": 5.597222222222222e-06,
      "loss": 0.0111,
      "step": 6222
    },
    {
      "epoch": 0.7202546296296296,
      "grad_norm": 0.06297320127487183,
      "learning_rate": 5.594907407407408e-06,
      "loss": 0.0103,
      "step": 6223
    },
    {
      "epoch": 0.7203703703703703,
      "grad_norm": 0.04776512086391449,
      "learning_rate": 5.5925925925925926e-06,
      "loss": 0.0086,
      "step": 6224
    },
    {
      "epoch": 0.7204861111111112,
      "grad_norm": 0.07143156975507736,
      "learning_rate": 5.5902777777777785e-06,
      "loss": 0.0125,
      "step": 6225
    },
    {
      "epoch": 0.7206018518518519,
      "grad_norm": 0.47601914405822754,
      "learning_rate": 5.587962962962964e-06,
      "loss": 0.0173,
      "step": 6226
    },
    {
      "epoch": 0.7207175925925926,
      "grad_norm": 0.05352991446852684,
      "learning_rate": 5.585648148148149e-06,
      "loss": 0.0097,
      "step": 6227
    },
    {
      "epoch": 0.7208333333333333,
      "grad_norm": 0.053903236985206604,
      "learning_rate": 5.583333333333334e-06,
      "loss": 0.0095,
      "step": 6228
    },
    {
      "epoch": 0.720949074074074,
      "grad_norm": 0.04458015412092209,
      "learning_rate": 5.58101851851852e-06,
      "loss": 0.008,
      "step": 6229
    },
    {
      "epoch": 0.7210648148148148,
      "grad_norm": 0.056017354130744934,
      "learning_rate": 5.578703703703704e-06,
      "loss": 0.0097,
      "step": 6230
    },
    {
      "epoch": 0.7211805555555556,
      "grad_norm": 0.056326497346162796,
      "learning_rate": 5.576388888888889e-06,
      "loss": 0.0099,
      "step": 6231
    },
    {
      "epoch": 0.7212962962962963,
      "grad_norm": 0.06072011590003967,
      "learning_rate": 5.574074074074075e-06,
      "loss": 0.0108,
      "step": 6232
    },
    {
      "epoch": 0.721412037037037,
      "grad_norm": 0.07650792598724365,
      "learning_rate": 5.571759259259259e-06,
      "loss": 0.0101,
      "step": 6233
    },
    {
      "epoch": 0.7215277777777778,
      "grad_norm": 0.04753442853689194,
      "learning_rate": 5.569444444444445e-06,
      "loss": 0.0081,
      "step": 6234
    },
    {
      "epoch": 0.7216435185185185,
      "grad_norm": 9.527275085449219,
      "learning_rate": 5.5671296296296295e-06,
      "loss": 2.8399,
      "step": 6235
    },
    {
      "epoch": 0.7217592592592592,
      "grad_norm": 0.08828800916671753,
      "learning_rate": 5.5648148148148154e-06,
      "loss": 0.0117,
      "step": 6236
    },
    {
      "epoch": 0.721875,
      "grad_norm": 0.5112291574478149,
      "learning_rate": 5.5625000000000005e-06,
      "loss": 0.0133,
      "step": 6237
    },
    {
      "epoch": 0.7219907407407408,
      "grad_norm": 0.05386102572083473,
      "learning_rate": 5.560185185185185e-06,
      "loss": 0.0097,
      "step": 6238
    },
    {
      "epoch": 0.7221064814814815,
      "grad_norm": 0.0643375813961029,
      "learning_rate": 5.557870370370371e-06,
      "loss": 0.0109,
      "step": 6239
    },
    {
      "epoch": 0.7222222222222222,
      "grad_norm": 0.06031975895166397,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.0105,
      "step": 6240
    },
    {
      "epoch": 0.7223379629629629,
      "grad_norm": 0.06679115444421768,
      "learning_rate": 5.553240740740741e-06,
      "loss": 0.0116,
      "step": 6241
    },
    {
      "epoch": 0.7224537037037037,
      "grad_norm": 0.629856288433075,
      "learning_rate": 5.550925925925926e-06,
      "loss": 0.0145,
      "step": 6242
    },
    {
      "epoch": 0.7225694444444445,
      "grad_norm": 16.14238739013672,
      "learning_rate": 5.548611111111112e-06,
      "loss": 2.5009,
      "step": 6243
    },
    {
      "epoch": 0.7226851851851852,
      "grad_norm": 0.05594657361507416,
      "learning_rate": 5.546296296296296e-06,
      "loss": 0.0072,
      "step": 6244
    },
    {
      "epoch": 0.7228009259259259,
      "grad_norm": 0.05963274464011192,
      "learning_rate": 5.543981481481482e-06,
      "loss": 0.0098,
      "step": 6245
    },
    {
      "epoch": 0.7229166666666667,
      "grad_norm": 0.057040221989154816,
      "learning_rate": 5.541666666666667e-06,
      "loss": 0.0102,
      "step": 6246
    },
    {
      "epoch": 0.7230324074074074,
      "grad_norm": 0.38141798973083496,
      "learning_rate": 5.5393518518518515e-06,
      "loss": 0.0142,
      "step": 6247
    },
    {
      "epoch": 0.7231481481481481,
      "grad_norm": 0.06996053457260132,
      "learning_rate": 5.5370370370370374e-06,
      "loss": 0.0126,
      "step": 6248
    },
    {
      "epoch": 0.7232638888888889,
      "grad_norm": 0.07010906934738159,
      "learning_rate": 5.534722222222223e-06,
      "loss": 0.0129,
      "step": 6249
    },
    {
      "epoch": 0.7233796296296297,
      "grad_norm": 0.542590856552124,
      "learning_rate": 5.532407407407408e-06,
      "loss": 0.0179,
      "step": 6250
    },
    {
      "epoch": 0.7234953703703704,
      "grad_norm": 0.03944683074951172,
      "learning_rate": 5.530092592592593e-06,
      "loss": 0.0072,
      "step": 6251
    },
    {
      "epoch": 0.7236111111111111,
      "grad_norm": 0.056655403226614,
      "learning_rate": 5.527777777777779e-06,
      "loss": 0.0096,
      "step": 6252
    },
    {
      "epoch": 0.7237268518518518,
      "grad_norm": 0.0644223615527153,
      "learning_rate": 5.525462962962963e-06,
      "loss": 0.0115,
      "step": 6253
    },
    {
      "epoch": 0.7238425925925925,
      "grad_norm": 0.07045532017946243,
      "learning_rate": 5.523148148148149e-06,
      "loss": 0.0118,
      "step": 6254
    },
    {
      "epoch": 0.7239583333333334,
      "grad_norm": 0.04914041981101036,
      "learning_rate": 5.520833333333334e-06,
      "loss": 0.0089,
      "step": 6255
    },
    {
      "epoch": 0.7240740740740741,
      "grad_norm": 0.050954222679138184,
      "learning_rate": 5.518518518518518e-06,
      "loss": 0.009,
      "step": 6256
    },
    {
      "epoch": 0.7241898148148148,
      "grad_norm": 13.123320579528809,
      "learning_rate": 5.516203703703704e-06,
      "loss": 2.9289,
      "step": 6257
    },
    {
      "epoch": 0.7243055555555555,
      "grad_norm": 0.06925548613071442,
      "learning_rate": 5.51388888888889e-06,
      "loss": 0.0128,
      "step": 6258
    },
    {
      "epoch": 0.7244212962962963,
      "grad_norm": 0.09507531672716141,
      "learning_rate": 5.511574074074074e-06,
      "loss": 0.0121,
      "step": 6259
    },
    {
      "epoch": 0.7245370370370371,
      "grad_norm": 0.07657311111688614,
      "learning_rate": 5.5092592592592595e-06,
      "loss": 0.0135,
      "step": 6260
    },
    {
      "epoch": 0.7246527777777778,
      "grad_norm": 0.06097502261400223,
      "learning_rate": 5.506944444444445e-06,
      "loss": 0.0087,
      "step": 6261
    },
    {
      "epoch": 0.7247685185185185,
      "grad_norm": 0.08907110244035721,
      "learning_rate": 5.50462962962963e-06,
      "loss": 0.0117,
      "step": 6262
    },
    {
      "epoch": 0.7248842592592593,
      "grad_norm": 0.07024263590574265,
      "learning_rate": 5.502314814814816e-06,
      "loss": 0.0112,
      "step": 6263
    },
    {
      "epoch": 0.725,
      "grad_norm": 0.5008041858673096,
      "learning_rate": 5.500000000000001e-06,
      "loss": 0.0113,
      "step": 6264
    },
    {
      "epoch": 0.7251157407407407,
      "grad_norm": 0.040335990488529205,
      "learning_rate": 5.497685185185185e-06,
      "loss": 0.0073,
      "step": 6265
    },
    {
      "epoch": 0.7252314814814815,
      "grad_norm": 0.039285849779844284,
      "learning_rate": 5.495370370370371e-06,
      "loss": 0.0072,
      "step": 6266
    },
    {
      "epoch": 0.7253472222222223,
      "grad_norm": 0.11524078994989395,
      "learning_rate": 5.493055555555557e-06,
      "loss": 0.0142,
      "step": 6267
    },
    {
      "epoch": 0.725462962962963,
      "grad_norm": 0.2023293524980545,
      "learning_rate": 5.490740740740741e-06,
      "loss": 0.0114,
      "step": 6268
    },
    {
      "epoch": 0.7255787037037037,
      "grad_norm": 321.7474670410156,
      "learning_rate": 5.488425925925926e-06,
      "loss": 1.8424,
      "step": 6269
    },
    {
      "epoch": 0.7256944444444444,
      "grad_norm": 0.046705011278390884,
      "learning_rate": 5.486111111111112e-06,
      "loss": 0.0081,
      "step": 6270
    },
    {
      "epoch": 0.7258101851851851,
      "grad_norm": 0.07017499953508377,
      "learning_rate": 5.483796296296296e-06,
      "loss": 0.0129,
      "step": 6271
    },
    {
      "epoch": 0.725925925925926,
      "grad_norm": 0.06787128746509552,
      "learning_rate": 5.481481481481482e-06,
      "loss": 0.0087,
      "step": 6272
    },
    {
      "epoch": 0.7260416666666667,
      "grad_norm": 0.0674331784248352,
      "learning_rate": 5.4791666666666674e-06,
      "loss": 0.0109,
      "step": 6273
    },
    {
      "epoch": 0.7261574074074074,
      "grad_norm": 0.06164063140749931,
      "learning_rate": 5.476851851851852e-06,
      "loss": 0.0104,
      "step": 6274
    },
    {
      "epoch": 0.7262731481481481,
      "grad_norm": 10.259206771850586,
      "learning_rate": 5.474537037037038e-06,
      "loss": 0.0807,
      "step": 6275
    },
    {
      "epoch": 0.7263888888888889,
      "grad_norm": 0.05079318955540657,
      "learning_rate": 5.4722222222222236e-06,
      "loss": 0.0093,
      "step": 6276
    },
    {
      "epoch": 0.7265046296296296,
      "grad_norm": 0.06446629017591476,
      "learning_rate": 5.469907407407408e-06,
      "loss": 0.0085,
      "step": 6277
    },
    {
      "epoch": 0.7266203703703704,
      "grad_norm": 0.0743335708975792,
      "learning_rate": 5.467592592592593e-06,
      "loss": 0.0126,
      "step": 6278
    },
    {
      "epoch": 0.7267361111111111,
      "grad_norm": 0.0459853895008564,
      "learning_rate": 5.465277777777778e-06,
      "loss": 0.0075,
      "step": 6279
    },
    {
      "epoch": 0.7268518518518519,
      "grad_norm": 0.07505105435848236,
      "learning_rate": 5.462962962962963e-06,
      "loss": 0.0111,
      "step": 6280
    },
    {
      "epoch": 0.7269675925925926,
      "grad_norm": 0.09258588403463364,
      "learning_rate": 5.460648148148149e-06,
      "loss": 0.0082,
      "step": 6281
    },
    {
      "epoch": 0.7270833333333333,
      "grad_norm": 0.9696339964866638,
      "learning_rate": 5.458333333333333e-06,
      "loss": 0.0187,
      "step": 6282
    },
    {
      "epoch": 0.727199074074074,
      "grad_norm": 0.04728038236498833,
      "learning_rate": 5.456018518518518e-06,
      "loss": 0.0085,
      "step": 6283
    },
    {
      "epoch": 0.7273148148148149,
      "grad_norm": 0.10294027626514435,
      "learning_rate": 5.453703703703704e-06,
      "loss": 0.0123,
      "step": 6284
    },
    {
      "epoch": 0.7274305555555556,
      "grad_norm": 0.05594291165471077,
      "learning_rate": 5.451388888888889e-06,
      "loss": 0.0101,
      "step": 6285
    },
    {
      "epoch": 0.7275462962962963,
      "grad_norm": 0.059777311980724335,
      "learning_rate": 5.4490740740740745e-06,
      "loss": 0.01,
      "step": 6286
    },
    {
      "epoch": 0.727662037037037,
      "grad_norm": 0.07500474154949188,
      "learning_rate": 5.44675925925926e-06,
      "loss": 0.0099,
      "step": 6287
    },
    {
      "epoch": 0.7277777777777777,
      "grad_norm": 0.03816726431250572,
      "learning_rate": 5.444444444444445e-06,
      "loss": 0.0069,
      "step": 6288
    },
    {
      "epoch": 0.7278935185185185,
      "grad_norm": 0.0679173618555069,
      "learning_rate": 5.44212962962963e-06,
      "loss": 0.0114,
      "step": 6289
    },
    {
      "epoch": 0.7280092592592593,
      "grad_norm": 0.04618511721491814,
      "learning_rate": 5.439814814814816e-06,
      "loss": 0.0084,
      "step": 6290
    },
    {
      "epoch": 0.728125,
      "grad_norm": 0.054584719240665436,
      "learning_rate": 5.4375e-06,
      "loss": 0.0094,
      "step": 6291
    },
    {
      "epoch": 0.7282407407407407,
      "grad_norm": 0.058250218629837036,
      "learning_rate": 5.435185185185186e-06,
      "loss": 0.0103,
      "step": 6292
    },
    {
      "epoch": 0.7283564814814815,
      "grad_norm": 0.06399782001972198,
      "learning_rate": 5.432870370370371e-06,
      "loss": 0.0108,
      "step": 6293
    },
    {
      "epoch": 0.7284722222222222,
      "grad_norm": 0.053299449384212494,
      "learning_rate": 5.430555555555555e-06,
      "loss": 0.0095,
      "step": 6294
    },
    {
      "epoch": 0.7285879629629629,
      "grad_norm": 0.08856146782636642,
      "learning_rate": 5.428240740740741e-06,
      "loss": 0.013,
      "step": 6295
    },
    {
      "epoch": 0.7287037037037037,
      "grad_norm": 0.07442386448383331,
      "learning_rate": 5.425925925925926e-06,
      "loss": 0.0136,
      "step": 6296
    },
    {
      "epoch": 0.7288194444444445,
      "grad_norm": 0.07456019520759583,
      "learning_rate": 5.4236111111111115e-06,
      "loss": 0.0099,
      "step": 6297
    },
    {
      "epoch": 0.7289351851851852,
      "grad_norm": 0.04590623453259468,
      "learning_rate": 5.4212962962962966e-06,
      "loss": 0.0083,
      "step": 6298
    },
    {
      "epoch": 0.7290509259259259,
      "grad_norm": 0.059357982128858566,
      "learning_rate": 5.4189814814814825e-06,
      "loss": 0.0108,
      "step": 6299
    },
    {
      "epoch": 0.7291666666666666,
      "grad_norm": 0.3442530930042267,
      "learning_rate": 5.416666666666667e-06,
      "loss": 0.0133,
      "step": 6300
    },
    {
      "epoch": 0.7292824074074075,
      "grad_norm": 0.04406030476093292,
      "learning_rate": 5.414351851851853e-06,
      "loss": 0.0079,
      "step": 6301
    },
    {
      "epoch": 0.7293981481481482,
      "grad_norm": 0.060994911938905716,
      "learning_rate": 5.412037037037038e-06,
      "loss": 0.011,
      "step": 6302
    },
    {
      "epoch": 0.7295138888888889,
      "grad_norm": 0.0642474889755249,
      "learning_rate": 5.409722222222222e-06,
      "loss": 0.0117,
      "step": 6303
    },
    {
      "epoch": 0.7296296296296296,
      "grad_norm": 0.23331524431705475,
      "learning_rate": 5.407407407407408e-06,
      "loss": 0.0135,
      "step": 6304
    },
    {
      "epoch": 0.7297453703703703,
      "grad_norm": 0.05832415074110031,
      "learning_rate": 5.405092592592593e-06,
      "loss": 0.0107,
      "step": 6305
    },
    {
      "epoch": 0.7298611111111111,
      "grad_norm": 0.052769094705581665,
      "learning_rate": 5.402777777777778e-06,
      "loss": 0.0069,
      "step": 6306
    },
    {
      "epoch": 0.7299768518518519,
      "grad_norm": 0.1472984254360199,
      "learning_rate": 5.400462962962963e-06,
      "loss": 0.0117,
      "step": 6307
    },
    {
      "epoch": 0.7300925925925926,
      "grad_norm": 0.06508582085371017,
      "learning_rate": 5.398148148148149e-06,
      "loss": 0.0119,
      "step": 6308
    },
    {
      "epoch": 0.7302083333333333,
      "grad_norm": 0.04376077279448509,
      "learning_rate": 5.3958333333333335e-06,
      "loss": 0.0078,
      "step": 6309
    },
    {
      "epoch": 0.7303240740740741,
      "grad_norm": 0.20820124447345734,
      "learning_rate": 5.3935185185185194e-06,
      "loss": 0.0163,
      "step": 6310
    },
    {
      "epoch": 0.7304398148148148,
      "grad_norm": 0.18339522182941437,
      "learning_rate": 5.3912037037037045e-06,
      "loss": 0.0115,
      "step": 6311
    },
    {
      "epoch": 0.7305555555555555,
      "grad_norm": 0.12016777694225311,
      "learning_rate": 5.388888888888889e-06,
      "loss": 0.0117,
      "step": 6312
    },
    {
      "epoch": 0.7306712962962963,
      "grad_norm": 0.058983612805604935,
      "learning_rate": 5.386574074074075e-06,
      "loss": 0.0104,
      "step": 6313
    },
    {
      "epoch": 0.7307870370370371,
      "grad_norm": 0.07314909249544144,
      "learning_rate": 5.38425925925926e-06,
      "loss": 0.0131,
      "step": 6314
    },
    {
      "epoch": 0.7309027777777778,
      "grad_norm": 0.06217653304338455,
      "learning_rate": 5.381944444444445e-06,
      "loss": 0.0113,
      "step": 6315
    },
    {
      "epoch": 0.7310185185185185,
      "grad_norm": 0.05348128825426102,
      "learning_rate": 5.37962962962963e-06,
      "loss": 0.0097,
      "step": 6316
    },
    {
      "epoch": 0.7311342592592592,
      "grad_norm": 0.06651048362255096,
      "learning_rate": 5.377314814814816e-06,
      "loss": 0.0115,
      "step": 6317
    },
    {
      "epoch": 0.73125,
      "grad_norm": 0.04582049325108528,
      "learning_rate": 5.375e-06,
      "loss": 0.0082,
      "step": 6318
    },
    {
      "epoch": 0.7313657407407408,
      "grad_norm": 0.03886685520410538,
      "learning_rate": 5.372685185185186e-06,
      "loss": 0.007,
      "step": 6319
    },
    {
      "epoch": 0.7314814814814815,
      "grad_norm": 0.4741683900356293,
      "learning_rate": 5.370370370370371e-06,
      "loss": 0.0128,
      "step": 6320
    },
    {
      "epoch": 0.7315972222222222,
      "grad_norm": 0.048914577811956406,
      "learning_rate": 5.3680555555555555e-06,
      "loss": 0.0087,
      "step": 6321
    },
    {
      "epoch": 0.731712962962963,
      "grad_norm": 0.08496316522359848,
      "learning_rate": 5.3657407407407414e-06,
      "loss": 0.0109,
      "step": 6322
    },
    {
      "epoch": 0.7318287037037037,
      "grad_norm": 0.15144377946853638,
      "learning_rate": 5.363425925925926e-06,
      "loss": 0.0142,
      "step": 6323
    },
    {
      "epoch": 0.7319444444444444,
      "grad_norm": 0.05200062692165375,
      "learning_rate": 5.361111111111112e-06,
      "loss": 0.0092,
      "step": 6324
    },
    {
      "epoch": 0.7320601851851852,
      "grad_norm": 0.04851669818162918,
      "learning_rate": 5.358796296296297e-06,
      "loss": 0.0089,
      "step": 6325
    },
    {
      "epoch": 0.732175925925926,
      "grad_norm": 0.06819086521863937,
      "learning_rate": 5.356481481481481e-06,
      "loss": 0.0112,
      "step": 6326
    },
    {
      "epoch": 0.7322916666666667,
      "grad_norm": 0.07418341189622879,
      "learning_rate": 5.354166666666667e-06,
      "loss": 0.0098,
      "step": 6327
    },
    {
      "epoch": 0.7324074074074074,
      "grad_norm": 98.24082946777344,
      "learning_rate": 5.351851851851853e-06,
      "loss": 0.9261,
      "step": 6328
    },
    {
      "epoch": 0.7325231481481481,
      "grad_norm": 0.058481231331825256,
      "learning_rate": 5.349537037037037e-06,
      "loss": 0.0103,
      "step": 6329
    },
    {
      "epoch": 0.7326388888888888,
      "grad_norm": 6.645513534545898,
      "learning_rate": 5.347222222222222e-06,
      "loss": 3.0522,
      "step": 6330
    },
    {
      "epoch": 0.7327546296296297,
      "grad_norm": 0.04837767779827118,
      "learning_rate": 5.344907407407408e-06,
      "loss": 0.0088,
      "step": 6331
    },
    {
      "epoch": 0.7328703703703704,
      "grad_norm": 27.299320220947266,
      "learning_rate": 5.342592592592592e-06,
      "loss": 0.044,
      "step": 6332
    },
    {
      "epoch": 0.7329861111111111,
      "grad_norm": 270.4356384277344,
      "learning_rate": 5.340277777777778e-06,
      "loss": 1.0026,
      "step": 6333
    },
    {
      "epoch": 0.7331018518518518,
      "grad_norm": 0.043063413351774216,
      "learning_rate": 5.3379629629629635e-06,
      "loss": 0.0078,
      "step": 6334
    },
    {
      "epoch": 0.7332175925925926,
      "grad_norm": 0.07366147637367249,
      "learning_rate": 5.335648148148148e-06,
      "loss": 0.0097,
      "step": 6335
    },
    {
      "epoch": 0.7333333333333333,
      "grad_norm": 0.0644460991024971,
      "learning_rate": 5.333333333333334e-06,
      "loss": 0.0115,
      "step": 6336
    },
    {
      "epoch": 0.7334490740740741,
      "grad_norm": 0.04719460755586624,
      "learning_rate": 5.33101851851852e-06,
      "loss": 0.0085,
      "step": 6337
    },
    {
      "epoch": 0.7335648148148148,
      "grad_norm": 0.06918901205062866,
      "learning_rate": 5.328703703703704e-06,
      "loss": 0.0126,
      "step": 6338
    },
    {
      "epoch": 0.7336805555555556,
      "grad_norm": 0.042711637914180756,
      "learning_rate": 5.326388888888889e-06,
      "loss": 0.0078,
      "step": 6339
    },
    {
      "epoch": 0.7337962962962963,
      "grad_norm": 0.07656502723693848,
      "learning_rate": 5.324074074074075e-06,
      "loss": 0.013,
      "step": 6340
    },
    {
      "epoch": 0.733912037037037,
      "grad_norm": 0.05647321790456772,
      "learning_rate": 5.321759259259259e-06,
      "loss": 0.0097,
      "step": 6341
    },
    {
      "epoch": 0.7340277777777777,
      "grad_norm": 0.03909262642264366,
      "learning_rate": 5.319444444444445e-06,
      "loss": 0.0071,
      "step": 6342
    },
    {
      "epoch": 0.7341435185185186,
      "grad_norm": 0.04508652165532112,
      "learning_rate": 5.31712962962963e-06,
      "loss": 0.0078,
      "step": 6343
    },
    {
      "epoch": 0.7342592592592593,
      "grad_norm": 0.11581381410360336,
      "learning_rate": 5.314814814814815e-06,
      "loss": 0.0137,
      "step": 6344
    },
    {
      "epoch": 0.734375,
      "grad_norm": 0.046043626964092255,
      "learning_rate": 5.3125e-06,
      "loss": 0.0084,
      "step": 6345
    },
    {
      "epoch": 0.7344907407407407,
      "grad_norm": 0.040370386093854904,
      "learning_rate": 5.310185185185186e-06,
      "loss": 0.0074,
      "step": 6346
    },
    {
      "epoch": 0.7346064814814814,
      "grad_norm": 0.05464920401573181,
      "learning_rate": 5.307870370370371e-06,
      "loss": 0.0099,
      "step": 6347
    },
    {
      "epoch": 0.7347222222222223,
      "grad_norm": 0.07037005573511124,
      "learning_rate": 5.305555555555556e-06,
      "loss": 0.0129,
      "step": 6348
    },
    {
      "epoch": 0.734837962962963,
      "grad_norm": 0.05420444533228874,
      "learning_rate": 5.303240740740742e-06,
      "loss": 0.0095,
      "step": 6349
    },
    {
      "epoch": 0.7349537037037037,
      "grad_norm": 0.07242225110530853,
      "learning_rate": 5.300925925925926e-06,
      "loss": 0.0128,
      "step": 6350
    },
    {
      "epoch": 0.7350694444444444,
      "grad_norm": 0.0445401594042778,
      "learning_rate": 5.298611111111112e-06,
      "loss": 0.0082,
      "step": 6351
    },
    {
      "epoch": 0.7351851851851852,
      "grad_norm": 0.04734128713607788,
      "learning_rate": 5.296296296296297e-06,
      "loss": 0.0087,
      "step": 6352
    },
    {
      "epoch": 0.7353009259259259,
      "grad_norm": 0.07232358306646347,
      "learning_rate": 5.293981481481482e-06,
      "loss": 0.0095,
      "step": 6353
    },
    {
      "epoch": 0.7354166666666667,
      "grad_norm": 0.05446912348270416,
      "learning_rate": 5.291666666666667e-06,
      "loss": 0.0099,
      "step": 6354
    },
    {
      "epoch": 0.7355324074074074,
      "grad_norm": 0.059617120772600174,
      "learning_rate": 5.289351851851853e-06,
      "loss": 0.0073,
      "step": 6355
    },
    {
      "epoch": 0.7356481481481482,
      "grad_norm": 0.09024754166603088,
      "learning_rate": 5.287037037037037e-06,
      "loss": 0.0103,
      "step": 6356
    },
    {
      "epoch": 0.7357638888888889,
      "grad_norm": 21.932598114013672,
      "learning_rate": 5.284722222222222e-06,
      "loss": 3.1585,
      "step": 6357
    },
    {
      "epoch": 0.7358796296296296,
      "grad_norm": 0.041976772248744965,
      "learning_rate": 5.282407407407408e-06,
      "loss": 0.0077,
      "step": 6358
    },
    {
      "epoch": 0.7359953703703703,
      "grad_norm": 0.10839608311653137,
      "learning_rate": 5.280092592592593e-06,
      "loss": 0.0114,
      "step": 6359
    },
    {
      "epoch": 0.7361111111111112,
      "grad_norm": 0.05097881332039833,
      "learning_rate": 5.2777777777777785e-06,
      "loss": 0.0091,
      "step": 6360
    },
    {
      "epoch": 0.7362268518518519,
      "grad_norm": 0.03837814927101135,
      "learning_rate": 5.275462962962964e-06,
      "loss": 0.007,
      "step": 6361
    },
    {
      "epoch": 0.7363425925925926,
      "grad_norm": 0.06963461637496948,
      "learning_rate": 5.273148148148149e-06,
      "loss": 0.0103,
      "step": 6362
    },
    {
      "epoch": 0.7364583333333333,
      "grad_norm": 0.0980743020772934,
      "learning_rate": 5.270833333333334e-06,
      "loss": 0.0117,
      "step": 6363
    },
    {
      "epoch": 0.736574074074074,
      "grad_norm": 0.06620069593191147,
      "learning_rate": 5.26851851851852e-06,
      "loss": 0.0121,
      "step": 6364
    },
    {
      "epoch": 0.7366898148148148,
      "grad_norm": 0.04284493997693062,
      "learning_rate": 5.266203703703704e-06,
      "loss": 0.0078,
      "step": 6365
    },
    {
      "epoch": 0.7368055555555556,
      "grad_norm": 0.15077275037765503,
      "learning_rate": 5.263888888888889e-06,
      "loss": 0.0128,
      "step": 6366
    },
    {
      "epoch": 0.7369212962962963,
      "grad_norm": 0.04871503636240959,
      "learning_rate": 5.261574074074075e-06,
      "loss": 0.0088,
      "step": 6367
    },
    {
      "epoch": 0.737037037037037,
      "grad_norm": 0.06600474566221237,
      "learning_rate": 5.259259259259259e-06,
      "loss": 0.0114,
      "step": 6368
    },
    {
      "epoch": 0.7371527777777778,
      "grad_norm": 0.03869680315256119,
      "learning_rate": 5.256944444444445e-06,
      "loss": 0.0071,
      "step": 6369
    },
    {
      "epoch": 0.7372685185185185,
      "grad_norm": 0.055968690663576126,
      "learning_rate": 5.2546296296296295e-06,
      "loss": 0.0073,
      "step": 6370
    },
    {
      "epoch": 0.7373842592592592,
      "grad_norm": 0.05663783848285675,
      "learning_rate": 5.2523148148148155e-06,
      "loss": 0.0098,
      "step": 6371
    },
    {
      "epoch": 0.7375,
      "grad_norm": 0.044403690844774246,
      "learning_rate": 5.2500000000000006e-06,
      "loss": 0.0079,
      "step": 6372
    },
    {
      "epoch": 0.7376157407407408,
      "grad_norm": 0.04456932470202446,
      "learning_rate": 5.247685185185185e-06,
      "loss": 0.0074,
      "step": 6373
    },
    {
      "epoch": 0.7377314814814815,
      "grad_norm": 0.057765018194913864,
      "learning_rate": 5.245370370370371e-06,
      "loss": 0.0101,
      "step": 6374
    },
    {
      "epoch": 0.7378472222222222,
      "grad_norm": 0.06795543432235718,
      "learning_rate": 5.243055555555556e-06,
      "loss": 0.0099,
      "step": 6375
    },
    {
      "epoch": 0.7379629629629629,
      "grad_norm": 0.05250024423003197,
      "learning_rate": 5.240740740740741e-06,
      "loss": 0.0094,
      "step": 6376
    },
    {
      "epoch": 0.7380787037037037,
      "grad_norm": 0.05879323184490204,
      "learning_rate": 5.238425925925926e-06,
      "loss": 0.0105,
      "step": 6377
    },
    {
      "epoch": 0.7381944444444445,
      "grad_norm": 0.05173344910144806,
      "learning_rate": 5.236111111111112e-06,
      "loss": 0.0093,
      "step": 6378
    },
    {
      "epoch": 0.7383101851851852,
      "grad_norm": 0.04766633361577988,
      "learning_rate": 5.233796296296296e-06,
      "loss": 0.0087,
      "step": 6379
    },
    {
      "epoch": 0.7384259259259259,
      "grad_norm": 0.06717780232429504,
      "learning_rate": 5.231481481481482e-06,
      "loss": 0.0119,
      "step": 6380
    },
    {
      "epoch": 0.7385416666666667,
      "grad_norm": 0.1265050619840622,
      "learning_rate": 5.229166666666667e-06,
      "loss": 0.0142,
      "step": 6381
    },
    {
      "epoch": 0.7386574074074074,
      "grad_norm": 0.07750824093818665,
      "learning_rate": 5.2268518518518515e-06,
      "loss": 0.0134,
      "step": 6382
    },
    {
      "epoch": 0.7387731481481481,
      "grad_norm": 0.06304745376110077,
      "learning_rate": 5.2245370370370375e-06,
      "loss": 0.0109,
      "step": 6383
    },
    {
      "epoch": 0.7388888888888889,
      "grad_norm": 0.3251372277736664,
      "learning_rate": 5.2222222222222226e-06,
      "loss": 0.0099,
      "step": 6384
    },
    {
      "epoch": 0.7390046296296297,
      "grad_norm": 0.06487435847520828,
      "learning_rate": 5.219907407407408e-06,
      "loss": 0.0117,
      "step": 6385
    },
    {
      "epoch": 0.7391203703703704,
      "grad_norm": 0.7703679203987122,
      "learning_rate": 5.217592592592593e-06,
      "loss": 0.0119,
      "step": 6386
    },
    {
      "epoch": 0.7392361111111111,
      "grad_norm": 0.06831016391515732,
      "learning_rate": 5.215277777777779e-06,
      "loss": 0.0125,
      "step": 6387
    },
    {
      "epoch": 0.7393518518518518,
      "grad_norm": 0.057421308010816574,
      "learning_rate": 5.212962962962963e-06,
      "loss": 0.01,
      "step": 6388
    },
    {
      "epoch": 0.7394675925925925,
      "grad_norm": 0.0551006905734539,
      "learning_rate": 5.210648148148149e-06,
      "loss": 0.0099,
      "step": 6389
    },
    {
      "epoch": 0.7395833333333334,
      "grad_norm": 45.35782241821289,
      "learning_rate": 5.208333333333334e-06,
      "loss": 2.755,
      "step": 6390
    },
    {
      "epoch": 0.7396990740740741,
      "grad_norm": 0.05134011432528496,
      "learning_rate": 5.206018518518518e-06,
      "loss": 0.0092,
      "step": 6391
    },
    {
      "epoch": 0.7398148148148148,
      "grad_norm": 0.19059014320373535,
      "learning_rate": 5.203703703703704e-06,
      "loss": 0.0122,
      "step": 6392
    },
    {
      "epoch": 0.7399305555555555,
      "grad_norm": 0.04710237681865692,
      "learning_rate": 5.201388888888889e-06,
      "loss": 0.0084,
      "step": 6393
    },
    {
      "epoch": 0.7400462962962963,
      "grad_norm": 0.07129456102848053,
      "learning_rate": 5.199074074074074e-06,
      "loss": 0.0131,
      "step": 6394
    },
    {
      "epoch": 0.7401620370370371,
      "grad_norm": 0.0808461531996727,
      "learning_rate": 5.1967592592592595e-06,
      "loss": 0.013,
      "step": 6395
    },
    {
      "epoch": 0.7402777777777778,
      "grad_norm": 0.057732850313186646,
      "learning_rate": 5.1944444444444454e-06,
      "loss": 0.0105,
      "step": 6396
    },
    {
      "epoch": 0.7403935185185185,
      "grad_norm": 0.047964442521333694,
      "learning_rate": 5.19212962962963e-06,
      "loss": 0.0088,
      "step": 6397
    },
    {
      "epoch": 0.7405092592592593,
      "grad_norm": 0.04009218141436577,
      "learning_rate": 5.189814814814816e-06,
      "loss": 0.0073,
      "step": 6398
    },
    {
      "epoch": 0.740625,
      "grad_norm": 0.07112794369459152,
      "learning_rate": 5.187500000000001e-06,
      "loss": 0.0128,
      "step": 6399
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 0.07812383770942688,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.009,
      "step": 6400
    },
    {
      "epoch": 0.7408564814814815,
      "grad_norm": 0.05499877780675888,
      "learning_rate": 5.182870370370371e-06,
      "loss": 0.0072,
      "step": 6401
    },
    {
      "epoch": 0.7409722222222223,
      "grad_norm": 0.050543904304504395,
      "learning_rate": 5.180555555555557e-06,
      "loss": 0.009,
      "step": 6402
    },
    {
      "epoch": 0.741087962962963,
      "grad_norm": 0.06318943202495575,
      "learning_rate": 5.178240740740741e-06,
      "loss": 0.01,
      "step": 6403
    },
    {
      "epoch": 0.7412037037037037,
      "grad_norm": 0.07636125385761261,
      "learning_rate": 5.175925925925926e-06,
      "loss": 0.0133,
      "step": 6404
    },
    {
      "epoch": 0.7413194444444444,
      "grad_norm": 0.06069927290081978,
      "learning_rate": 5.173611111111112e-06,
      "loss": 0.008,
      "step": 6405
    },
    {
      "epoch": 0.7414351851851851,
      "grad_norm": 0.06418212503194809,
      "learning_rate": 5.171296296296296e-06,
      "loss": 0.0117,
      "step": 6406
    },
    {
      "epoch": 0.741550925925926,
      "grad_norm": 0.057077955454587936,
      "learning_rate": 5.168981481481482e-06,
      "loss": 0.01,
      "step": 6407
    },
    {
      "epoch": 0.7416666666666667,
      "grad_norm": 0.038556043058633804,
      "learning_rate": 5.1666666666666675e-06,
      "loss": 0.007,
      "step": 6408
    },
    {
      "epoch": 0.7417824074074074,
      "grad_norm": 0.36331626772880554,
      "learning_rate": 5.164351851851852e-06,
      "loss": 0.0129,
      "step": 6409
    },
    {
      "epoch": 0.7418981481481481,
      "grad_norm": 0.053668346256017685,
      "learning_rate": 5.162037037037038e-06,
      "loss": 0.0097,
      "step": 6410
    },
    {
      "epoch": 0.7420138888888889,
      "grad_norm": 0.060935474932193756,
      "learning_rate": 5.159722222222224e-06,
      "loss": 0.0099,
      "step": 6411
    },
    {
      "epoch": 0.7421296296296296,
      "grad_norm": 0.05549237132072449,
      "learning_rate": 5.157407407407408e-06,
      "loss": 0.0094,
      "step": 6412
    },
    {
      "epoch": 0.7422453703703704,
      "grad_norm": 0.05187302827835083,
      "learning_rate": 5.155092592592593e-06,
      "loss": 0.0091,
      "step": 6413
    },
    {
      "epoch": 0.7423611111111111,
      "grad_norm": 0.06613942980766296,
      "learning_rate": 5.152777777777778e-06,
      "loss": 0.0091,
      "step": 6414
    },
    {
      "epoch": 0.7424768518518519,
      "grad_norm": 0.05681198835372925,
      "learning_rate": 5.150462962962963e-06,
      "loss": 0.0101,
      "step": 6415
    },
    {
      "epoch": 0.7425925925925926,
      "grad_norm": 0.04052898660302162,
      "learning_rate": 5.148148148148149e-06,
      "loss": 0.0072,
      "step": 6416
    },
    {
      "epoch": 0.7427083333333333,
      "grad_norm": 0.04076999053359032,
      "learning_rate": 5.145833333333333e-06,
      "loss": 0.0074,
      "step": 6417
    },
    {
      "epoch": 0.742824074074074,
      "grad_norm": 0.058612674474716187,
      "learning_rate": 5.1435185185185184e-06,
      "loss": 0.0098,
      "step": 6418
    },
    {
      "epoch": 0.7429398148148149,
      "grad_norm": 0.047589030116796494,
      "learning_rate": 5.141203703703704e-06,
      "loss": 0.0087,
      "step": 6419
    },
    {
      "epoch": 0.7430555555555556,
      "grad_norm": 1.0064918994903564,
      "learning_rate": 5.138888888888889e-06,
      "loss": 0.0171,
      "step": 6420
    },
    {
      "epoch": 0.7431712962962963,
      "grad_norm": 0.08430086076259613,
      "learning_rate": 5.1365740740740746e-06,
      "loss": 0.0132,
      "step": 6421
    },
    {
      "epoch": 0.743287037037037,
      "grad_norm": 0.05195131152868271,
      "learning_rate": 5.13425925925926e-06,
      "loss": 0.0093,
      "step": 6422
    },
    {
      "epoch": 0.7434027777777777,
      "grad_norm": 0.059009719640016556,
      "learning_rate": 5.131944444444445e-06,
      "loss": 0.0104,
      "step": 6423
    },
    {
      "epoch": 0.7435185185185185,
      "grad_norm": 68.21672058105469,
      "learning_rate": 5.12962962962963e-06,
      "loss": 2.3479,
      "step": 6424
    },
    {
      "epoch": 0.7436342592592593,
      "grad_norm": 0.05697057396173477,
      "learning_rate": 5.127314814814816e-06,
      "loss": 0.0103,
      "step": 6425
    },
    {
      "epoch": 0.74375,
      "grad_norm": 0.37722471356391907,
      "learning_rate": 5.125e-06,
      "loss": 0.0167,
      "step": 6426
    },
    {
      "epoch": 0.7438657407407407,
      "grad_norm": 0.20712213218212128,
      "learning_rate": 5.122685185185185e-06,
      "loss": 0.0143,
      "step": 6427
    },
    {
      "epoch": 0.7439814814814815,
      "grad_norm": 38.13880920410156,
      "learning_rate": 5.120370370370371e-06,
      "loss": 2.8201,
      "step": 6428
    },
    {
      "epoch": 0.7440972222222222,
      "grad_norm": 0.06688699126243591,
      "learning_rate": 5.118055555555555e-06,
      "loss": 0.012,
      "step": 6429
    },
    {
      "epoch": 0.7442129629629629,
      "grad_norm": 0.05228932946920395,
      "learning_rate": 5.115740740740741e-06,
      "loss": 0.0091,
      "step": 6430
    },
    {
      "epoch": 0.7443287037037037,
      "grad_norm": 0.05369238927960396,
      "learning_rate": 5.113425925925926e-06,
      "loss": 0.0097,
      "step": 6431
    },
    {
      "epoch": 0.7444444444444445,
      "grad_norm": 0.0409962460398674,
      "learning_rate": 5.1111111111111115e-06,
      "loss": 0.0074,
      "step": 6432
    },
    {
      "epoch": 0.7445601851851852,
      "grad_norm": 0.051640380173921585,
      "learning_rate": 5.108796296296297e-06,
      "loss": 0.0091,
      "step": 6433
    },
    {
      "epoch": 0.7446759259259259,
      "grad_norm": 0.0721840113401413,
      "learning_rate": 5.1064814814814825e-06,
      "loss": 0.0076,
      "step": 6434
    },
    {
      "epoch": 0.7447916666666666,
      "grad_norm": 0.04394948482513428,
      "learning_rate": 5.104166666666667e-06,
      "loss": 0.008,
      "step": 6435
    },
    {
      "epoch": 0.7449074074074075,
      "grad_norm": 0.05937827751040459,
      "learning_rate": 5.101851851851852e-06,
      "loss": 0.0107,
      "step": 6436
    },
    {
      "epoch": 0.7450231481481482,
      "grad_norm": 0.5376483798027039,
      "learning_rate": 5.099537037037038e-06,
      "loss": 0.0151,
      "step": 6437
    },
    {
      "epoch": 0.7451388888888889,
      "grad_norm": 0.04960025101900101,
      "learning_rate": 5.097222222222222e-06,
      "loss": 0.0089,
      "step": 6438
    },
    {
      "epoch": 0.7452546296296296,
      "grad_norm": 0.041515104472637177,
      "learning_rate": 5.094907407407408e-06,
      "loss": 0.0074,
      "step": 6439
    },
    {
      "epoch": 0.7453703703703703,
      "grad_norm": 0.0449124351143837,
      "learning_rate": 5.092592592592593e-06,
      "loss": 0.0081,
      "step": 6440
    },
    {
      "epoch": 0.7454861111111111,
      "grad_norm": 0.06481552869081497,
      "learning_rate": 5.090277777777778e-06,
      "loss": 0.011,
      "step": 6441
    },
    {
      "epoch": 0.7456018518518519,
      "grad_norm": 0.06376414000988007,
      "learning_rate": 5.087962962962963e-06,
      "loss": 0.0117,
      "step": 6442
    },
    {
      "epoch": 0.7457175925925926,
      "grad_norm": 0.06988354027271271,
      "learning_rate": 5.085648148148149e-06,
      "loss": 0.0123,
      "step": 6443
    },
    {
      "epoch": 0.7458333333333333,
      "grad_norm": 0.08199314773082733,
      "learning_rate": 5.0833333333333335e-06,
      "loss": 0.0107,
      "step": 6444
    },
    {
      "epoch": 0.7459490740740741,
      "grad_norm": 0.055884767323732376,
      "learning_rate": 5.081018518518519e-06,
      "loss": 0.0096,
      "step": 6445
    },
    {
      "epoch": 0.7460648148148148,
      "grad_norm": 0.07669779658317566,
      "learning_rate": 5.0787037037037046e-06,
      "loss": 0.0127,
      "step": 6446
    },
    {
      "epoch": 0.7461805555555555,
      "grad_norm": 0.05196176841855049,
      "learning_rate": 5.076388888888889e-06,
      "loss": 0.0068,
      "step": 6447
    },
    {
      "epoch": 0.7462962962962963,
      "grad_norm": 0.05785132572054863,
      "learning_rate": 5.074074074074075e-06,
      "loss": 0.0099,
      "step": 6448
    },
    {
      "epoch": 0.7464120370370371,
      "grad_norm": 0.0553133450448513,
      "learning_rate": 5.07175925925926e-06,
      "loss": 0.0087,
      "step": 6449
    },
    {
      "epoch": 0.7465277777777778,
      "grad_norm": 0.05610691010951996,
      "learning_rate": 5.069444444444445e-06,
      "loss": 0.0099,
      "step": 6450
    },
    {
      "epoch": 0.7466435185185185,
      "grad_norm": 0.07135555893182755,
      "learning_rate": 5.06712962962963e-06,
      "loss": 0.0111,
      "step": 6451
    },
    {
      "epoch": 0.7467592592592592,
      "grad_norm": 0.03972476348280907,
      "learning_rate": 5.064814814814816e-06,
      "loss": 0.0071,
      "step": 6452
    },
    {
      "epoch": 0.746875,
      "grad_norm": 0.13423240184783936,
      "learning_rate": 5.0625e-06,
      "loss": 0.0113,
      "step": 6453
    },
    {
      "epoch": 0.7469907407407408,
      "grad_norm": 0.048028524965047836,
      "learning_rate": 5.060185185185186e-06,
      "loss": 0.0087,
      "step": 6454
    },
    {
      "epoch": 0.7471064814814815,
      "grad_norm": 0.2364647388458252,
      "learning_rate": 5.057870370370371e-06,
      "loss": 0.0118,
      "step": 6455
    },
    {
      "epoch": 0.7472222222222222,
      "grad_norm": 0.037757471203804016,
      "learning_rate": 5.0555555555555555e-06,
      "loss": 0.0069,
      "step": 6456
    },
    {
      "epoch": 0.747337962962963,
      "grad_norm": 0.03966272622346878,
      "learning_rate": 5.0532407407407415e-06,
      "loss": 0.0072,
      "step": 6457
    },
    {
      "epoch": 0.7474537037037037,
      "grad_norm": 0.04368172958493233,
      "learning_rate": 5.050925925925926e-06,
      "loss": 0.0077,
      "step": 6458
    },
    {
      "epoch": 0.7475694444444444,
      "grad_norm": 0.04486166685819626,
      "learning_rate": 5.048611111111112e-06,
      "loss": 0.0081,
      "step": 6459
    },
    {
      "epoch": 0.7476851851851852,
      "grad_norm": 0.05538596212863922,
      "learning_rate": 5.046296296296297e-06,
      "loss": 0.0071,
      "step": 6460
    },
    {
      "epoch": 0.747800925925926,
      "grad_norm": 0.8267145156860352,
      "learning_rate": 5.043981481481481e-06,
      "loss": 0.0122,
      "step": 6461
    },
    {
      "epoch": 0.7479166666666667,
      "grad_norm": 0.057854317128658295,
      "learning_rate": 5.041666666666667e-06,
      "loss": 0.0103,
      "step": 6462
    },
    {
      "epoch": 0.7480324074074074,
      "grad_norm": 0.04324866458773613,
      "learning_rate": 5.039351851851853e-06,
      "loss": 0.0078,
      "step": 6463
    },
    {
      "epoch": 0.7481481481481481,
      "grad_norm": 0.06040603294968605,
      "learning_rate": 5.037037037037037e-06,
      "loss": 0.0109,
      "step": 6464
    },
    {
      "epoch": 0.7482638888888888,
      "grad_norm": 237.07640075683594,
      "learning_rate": 5.034722222222222e-06,
      "loss": 1.5719,
      "step": 6465
    },
    {
      "epoch": 0.7483796296296297,
      "grad_norm": 26.814115524291992,
      "learning_rate": 5.032407407407408e-06,
      "loss": 3.0125,
      "step": 6466
    },
    {
      "epoch": 0.7484953703703704,
      "grad_norm": 0.055195484310388565,
      "learning_rate": 5.0300925925925924e-06,
      "loss": 0.0072,
      "step": 6467
    },
    {
      "epoch": 0.7486111111111111,
      "grad_norm": 0.040327537804841995,
      "learning_rate": 5.027777777777778e-06,
      "loss": 0.0073,
      "step": 6468
    },
    {
      "epoch": 0.7487268518518518,
      "grad_norm": 0.05544775724411011,
      "learning_rate": 5.0254629629629635e-06,
      "loss": 0.0072,
      "step": 6469
    },
    {
      "epoch": 0.7488425925925926,
      "grad_norm": 0.044195663183927536,
      "learning_rate": 5.023148148148148e-06,
      "loss": 0.008,
      "step": 6470
    },
    {
      "epoch": 0.7489583333333333,
      "grad_norm": 0.04277625307440758,
      "learning_rate": 5.020833333333334e-06,
      "loss": 0.0077,
      "step": 6471
    },
    {
      "epoch": 0.7490740740740741,
      "grad_norm": 0.04707964137196541,
      "learning_rate": 5.01851851851852e-06,
      "loss": 0.0086,
      "step": 6472
    },
    {
      "epoch": 0.7491898148148148,
      "grad_norm": 38.24337387084961,
      "learning_rate": 5.016203703703704e-06,
      "loss": 2.7394,
      "step": 6473
    },
    {
      "epoch": 0.7493055555555556,
      "grad_norm": 145.60537719726562,
      "learning_rate": 5.013888888888889e-06,
      "loss": 1.3543,
      "step": 6474
    },
    {
      "epoch": 0.7494212962962963,
      "grad_norm": 0.04113737493753433,
      "learning_rate": 5.011574074074075e-06,
      "loss": 0.0075,
      "step": 6475
    },
    {
      "epoch": 0.749537037037037,
      "grad_norm": 0.0697365254163742,
      "learning_rate": 5.009259259259259e-06,
      "loss": 0.0128,
      "step": 6476
    },
    {
      "epoch": 0.7496527777777777,
      "grad_norm": 0.0495934933423996,
      "learning_rate": 5.006944444444445e-06,
      "loss": 0.0086,
      "step": 6477
    },
    {
      "epoch": 0.7497685185185186,
      "grad_norm": 0.053125832229852676,
      "learning_rate": 5.00462962962963e-06,
      "loss": 0.0093,
      "step": 6478
    },
    {
      "epoch": 0.7498842592592593,
      "grad_norm": 0.052986618131399155,
      "learning_rate": 5.0023148148148145e-06,
      "loss": 0.0096,
      "step": 6479
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.0575164370238781,
      "learning_rate": 5e-06,
      "loss": 0.0101,
      "step": 6480
    },
    {
      "epoch": 0.7501157407407407,
      "grad_norm": 6.877313613891602,
      "learning_rate": 4.9976851851851855e-06,
      "loss": 3.3151,
      "step": 6481
    },
    {
      "epoch": 0.7502314814814814,
      "grad_norm": 0.05427684634923935,
      "learning_rate": 4.995370370370371e-06,
      "loss": 0.0071,
      "step": 6482
    },
    {
      "epoch": 0.7503472222222223,
      "grad_norm": 0.05730461701750755,
      "learning_rate": 4.993055555555556e-06,
      "loss": 0.0102,
      "step": 6483
    },
    {
      "epoch": 0.750462962962963,
      "grad_norm": 0.052348025143146515,
      "learning_rate": 4.990740740740741e-06,
      "loss": 0.0068,
      "step": 6484
    },
    {
      "epoch": 0.7505787037037037,
      "grad_norm": 0.044564537703990936,
      "learning_rate": 4.988425925925927e-06,
      "loss": 0.008,
      "step": 6485
    },
    {
      "epoch": 0.7506944444444444,
      "grad_norm": 41.81822967529297,
      "learning_rate": 4.986111111111112e-06,
      "loss": 2.7035,
      "step": 6486
    },
    {
      "epoch": 0.7508101851851852,
      "grad_norm": 4.579129219055176,
      "learning_rate": 4.983796296296297e-06,
      "loss": 0.0365,
      "step": 6487
    },
    {
      "epoch": 0.7509259259259259,
      "grad_norm": 0.05488703399896622,
      "learning_rate": 4.981481481481482e-06,
      "loss": 0.0097,
      "step": 6488
    },
    {
      "epoch": 0.7510416666666667,
      "grad_norm": 0.04600773751735687,
      "learning_rate": 4.979166666666667e-06,
      "loss": 0.0083,
      "step": 6489
    },
    {
      "epoch": 0.7511574074074074,
      "grad_norm": 0.17841030657291412,
      "learning_rate": 4.976851851851852e-06,
      "loss": 0.0106,
      "step": 6490
    },
    {
      "epoch": 0.7512731481481482,
      "grad_norm": 0.06007164344191551,
      "learning_rate": 4.974537037037037e-06,
      "loss": 0.0103,
      "step": 6491
    },
    {
      "epoch": 0.7513888888888889,
      "grad_norm": 0.08428601920604706,
      "learning_rate": 4.9722222222222224e-06,
      "loss": 0.0111,
      "step": 6492
    },
    {
      "epoch": 0.7515046296296296,
      "grad_norm": 0.9065048694610596,
      "learning_rate": 4.9699074074074075e-06,
      "loss": 0.0147,
      "step": 6493
    },
    {
      "epoch": 0.7516203703703703,
      "grad_norm": 99.89546203613281,
      "learning_rate": 4.967592592592593e-06,
      "loss": 2.2962,
      "step": 6494
    },
    {
      "epoch": 0.7517361111111112,
      "grad_norm": 0.041763562709093094,
      "learning_rate": 4.9652777777777786e-06,
      "loss": 0.0075,
      "step": 6495
    },
    {
      "epoch": 0.7518518518518519,
      "grad_norm": 0.05101956054568291,
      "learning_rate": 4.962962962962964e-06,
      "loss": 0.009,
      "step": 6496
    },
    {
      "epoch": 0.7519675925925926,
      "grad_norm": 0.05958319082856178,
      "learning_rate": 4.960648148148148e-06,
      "loss": 0.0107,
      "step": 6497
    },
    {
      "epoch": 0.7520833333333333,
      "grad_norm": 0.040158141404390335,
      "learning_rate": 4.958333333333334e-06,
      "loss": 0.0073,
      "step": 6498
    },
    {
      "epoch": 0.752199074074074,
      "grad_norm": 0.050599031150341034,
      "learning_rate": 4.956018518518519e-06,
      "loss": 0.009,
      "step": 6499
    },
    {
      "epoch": 0.7523148148148148,
      "grad_norm": 0.06818424165248871,
      "learning_rate": 4.953703703703704e-06,
      "loss": 0.0124,
      "step": 6500
    },
    {
      "epoch": 0.7524305555555556,
      "grad_norm": 0.04527832567691803,
      "learning_rate": 4.951388888888889e-06,
      "loss": 0.0081,
      "step": 6501
    },
    {
      "epoch": 0.7525462962962963,
      "grad_norm": 0.05745985731482506,
      "learning_rate": 4.949074074074074e-06,
      "loss": 0.0102,
      "step": 6502
    },
    {
      "epoch": 0.752662037037037,
      "grad_norm": 0.0403798408806324,
      "learning_rate": 4.946759259259259e-06,
      "loss": 0.0073,
      "step": 6503
    },
    {
      "epoch": 0.7527777777777778,
      "grad_norm": 0.7903213500976562,
      "learning_rate": 4.944444444444445e-06,
      "loss": 0.0174,
      "step": 6504
    },
    {
      "epoch": 0.7528935185185185,
      "grad_norm": 0.05251612886786461,
      "learning_rate": 4.94212962962963e-06,
      "loss": 0.0092,
      "step": 6505
    },
    {
      "epoch": 0.7530092592592592,
      "grad_norm": 0.06795193254947662,
      "learning_rate": 4.939814814814815e-06,
      "loss": 0.0124,
      "step": 6506
    },
    {
      "epoch": 0.753125,
      "grad_norm": 0.05391626060009003,
      "learning_rate": 4.937500000000001e-06,
      "loss": 0.0093,
      "step": 6507
    },
    {
      "epoch": 0.7532407407407408,
      "grad_norm": 154.6907501220703,
      "learning_rate": 4.935185185185186e-06,
      "loss": 1.0154,
      "step": 6508
    },
    {
      "epoch": 0.7533564814814815,
      "grad_norm": 0.05096031725406647,
      "learning_rate": 4.932870370370371e-06,
      "loss": 0.009,
      "step": 6509
    },
    {
      "epoch": 0.7534722222222222,
      "grad_norm": 0.05117608234286308,
      "learning_rate": 4.930555555555556e-06,
      "loss": 0.009,
      "step": 6510
    },
    {
      "epoch": 0.7535879629629629,
      "grad_norm": 0.03765745833516121,
      "learning_rate": 4.928240740740741e-06,
      "loss": 0.0068,
      "step": 6511
    },
    {
      "epoch": 0.7537037037037037,
      "grad_norm": 0.05087639018893242,
      "learning_rate": 4.925925925925926e-06,
      "loss": 0.0089,
      "step": 6512
    },
    {
      "epoch": 0.7538194444444445,
      "grad_norm": 0.055790599435567856,
      "learning_rate": 4.923611111111112e-06,
      "loss": 0.0097,
      "step": 6513
    },
    {
      "epoch": 0.7539351851851852,
      "grad_norm": 0.05458366870880127,
      "learning_rate": 4.921296296296297e-06,
      "loss": 0.0098,
      "step": 6514
    },
    {
      "epoch": 0.7540509259259259,
      "grad_norm": 0.047651708126068115,
      "learning_rate": 4.918981481481482e-06,
      "loss": 0.0087,
      "step": 6515
    },
    {
      "epoch": 0.7541666666666667,
      "grad_norm": 0.04358772933483124,
      "learning_rate": 4.9166666666666665e-06,
      "loss": 0.0079,
      "step": 6516
    },
    {
      "epoch": 0.7542824074074074,
      "grad_norm": 0.04308697208762169,
      "learning_rate": 4.914351851851852e-06,
      "loss": 0.0078,
      "step": 6517
    },
    {
      "epoch": 0.7543981481481481,
      "grad_norm": 0.04010111093521118,
      "learning_rate": 4.9120370370370375e-06,
      "loss": 0.0073,
      "step": 6518
    },
    {
      "epoch": 0.7545138888888889,
      "grad_norm": 0.06088694930076599,
      "learning_rate": 4.909722222222223e-06,
      "loss": 0.0103,
      "step": 6519
    },
    {
      "epoch": 0.7546296296296297,
      "grad_norm": 0.04077048599720001,
      "learning_rate": 4.907407407407408e-06,
      "loss": 0.0074,
      "step": 6520
    },
    {
      "epoch": 0.7547453703703704,
      "grad_norm": 0.0575951524078846,
      "learning_rate": 4.905092592592593e-06,
      "loss": 0.0102,
      "step": 6521
    },
    {
      "epoch": 0.7548611111111111,
      "grad_norm": 127.83574676513672,
      "learning_rate": 4.902777777777778e-06,
      "loss": 2.09,
      "step": 6522
    },
    {
      "epoch": 0.7549768518518518,
      "grad_norm": 0.05132926255464554,
      "learning_rate": 4.900462962962964e-06,
      "loss": 0.0089,
      "step": 6523
    },
    {
      "epoch": 0.7550925925925925,
      "grad_norm": 0.04345853254199028,
      "learning_rate": 4.898148148148149e-06,
      "loss": 0.0075,
      "step": 6524
    },
    {
      "epoch": 0.7552083333333334,
      "grad_norm": 0.06795554608106613,
      "learning_rate": 4.895833333333333e-06,
      "loss": 0.0125,
      "step": 6525
    },
    {
      "epoch": 0.7553240740740741,
      "grad_norm": 0.04526003450155258,
      "learning_rate": 4.893518518518519e-06,
      "loss": 0.0079,
      "step": 6526
    },
    {
      "epoch": 0.7554398148148148,
      "grad_norm": 0.09581568837165833,
      "learning_rate": 4.891203703703704e-06,
      "loss": 0.0119,
      "step": 6527
    },
    {
      "epoch": 0.7555555555555555,
      "grad_norm": 0.045994020998477936,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0084,
      "step": 6528
    },
    {
      "epoch": 0.7556712962962963,
      "grad_norm": 0.05198046937584877,
      "learning_rate": 4.8865740740740744e-06,
      "loss": 0.0084,
      "step": 6529
    },
    {
      "epoch": 0.7557870370370371,
      "grad_norm": 194.9403533935547,
      "learning_rate": 4.8842592592592595e-06,
      "loss": 1.474,
      "step": 6530
    },
    {
      "epoch": 0.7559027777777778,
      "grad_norm": 0.052272725850343704,
      "learning_rate": 4.881944444444445e-06,
      "loss": 0.0094,
      "step": 6531
    },
    {
      "epoch": 0.7560185185185185,
      "grad_norm": 0.050212565809488297,
      "learning_rate": 4.8796296296296306e-06,
      "loss": 0.0066,
      "step": 6532
    },
    {
      "epoch": 0.7561342592592593,
      "grad_norm": 0.3932967483997345,
      "learning_rate": 4.877314814814816e-06,
      "loss": 0.0139,
      "step": 6533
    },
    {
      "epoch": 0.75625,
      "grad_norm": 16.819873809814453,
      "learning_rate": 4.875e-06,
      "loss": 0.0567,
      "step": 6534
    },
    {
      "epoch": 0.7563657407407407,
      "grad_norm": 53.523624420166016,
      "learning_rate": 4.872685185185186e-06,
      "loss": 2.5392,
      "step": 6535
    },
    {
      "epoch": 0.7564814814814815,
      "grad_norm": 0.066654272377491,
      "learning_rate": 4.870370370370371e-06,
      "loss": 0.0109,
      "step": 6536
    },
    {
      "epoch": 0.7565972222222223,
      "grad_norm": 0.04001285880804062,
      "learning_rate": 4.868055555555556e-06,
      "loss": 0.0073,
      "step": 6537
    },
    {
      "epoch": 0.756712962962963,
      "grad_norm": 0.04708142206072807,
      "learning_rate": 4.865740740740741e-06,
      "loss": 0.0086,
      "step": 6538
    },
    {
      "epoch": 0.7568287037037037,
      "grad_norm": 0.045034728944301605,
      "learning_rate": 4.863425925925926e-06,
      "loss": 0.0079,
      "step": 6539
    },
    {
      "epoch": 0.7569444444444444,
      "grad_norm": 0.06090271472930908,
      "learning_rate": 4.861111111111111e-06,
      "loss": 0.0108,
      "step": 6540
    },
    {
      "epoch": 0.7570601851851851,
      "grad_norm": 0.06405393034219742,
      "learning_rate": 4.8587962962962964e-06,
      "loss": 0.0117,
      "step": 6541
    },
    {
      "epoch": 0.757175925925926,
      "grad_norm": 0.04929420351982117,
      "learning_rate": 4.856481481481482e-06,
      "loss": 0.0087,
      "step": 6542
    },
    {
      "epoch": 0.7572916666666667,
      "grad_norm": 0.05721713602542877,
      "learning_rate": 4.854166666666667e-06,
      "loss": 0.0097,
      "step": 6543
    },
    {
      "epoch": 0.7574074074074074,
      "grad_norm": 0.04793304204940796,
      "learning_rate": 4.851851851851852e-06,
      "loss": 0.0082,
      "step": 6544
    },
    {
      "epoch": 0.7575231481481481,
      "grad_norm": 0.03961140662431717,
      "learning_rate": 4.849537037037038e-06,
      "loss": 0.0072,
      "step": 6545
    },
    {
      "epoch": 0.7576388888888889,
      "grad_norm": 0.08748860657215118,
      "learning_rate": 4.847222222222223e-06,
      "loss": 0.0114,
      "step": 6546
    },
    {
      "epoch": 0.7577546296296296,
      "grad_norm": 0.05285952240228653,
      "learning_rate": 4.844907407407408e-06,
      "loss": 0.0096,
      "step": 6547
    },
    {
      "epoch": 0.7578703703703704,
      "grad_norm": 0.054919853806495667,
      "learning_rate": 4.842592592592593e-06,
      "loss": 0.01,
      "step": 6548
    },
    {
      "epoch": 0.7579861111111111,
      "grad_norm": 0.04873022437095642,
      "learning_rate": 4.840277777777778e-06,
      "loss": 0.0087,
      "step": 6549
    },
    {
      "epoch": 0.7581018518518519,
      "grad_norm": 0.055983856320381165,
      "learning_rate": 4.837962962962963e-06,
      "loss": 0.0102,
      "step": 6550
    },
    {
      "epoch": 0.7582175925925926,
      "grad_norm": 0.09397813677787781,
      "learning_rate": 4.835648148148149e-06,
      "loss": 0.0098,
      "step": 6551
    },
    {
      "epoch": 0.7583333333333333,
      "grad_norm": 0.040179137140512466,
      "learning_rate": 4.833333333333333e-06,
      "loss": 0.0073,
      "step": 6552
    },
    {
      "epoch": 0.758449074074074,
      "grad_norm": 0.09207126498222351,
      "learning_rate": 4.8310185185185185e-06,
      "loss": 0.0092,
      "step": 6553
    },
    {
      "epoch": 0.7585648148148149,
      "grad_norm": 0.06608154624700546,
      "learning_rate": 4.828703703703704e-06,
      "loss": 0.0103,
      "step": 6554
    },
    {
      "epoch": 0.7586805555555556,
      "grad_norm": 0.10122731328010559,
      "learning_rate": 4.8263888888888895e-06,
      "loss": 0.0082,
      "step": 6555
    },
    {
      "epoch": 0.7587962962962963,
      "grad_norm": 0.04855832830071449,
      "learning_rate": 4.824074074074075e-06,
      "loss": 0.0086,
      "step": 6556
    },
    {
      "epoch": 0.758912037037037,
      "grad_norm": 0.04476100206375122,
      "learning_rate": 4.82175925925926e-06,
      "loss": 0.0081,
      "step": 6557
    },
    {
      "epoch": 0.7590277777777777,
      "grad_norm": 0.052859097719192505,
      "learning_rate": 4.819444444444445e-06,
      "loss": 0.0095,
      "step": 6558
    },
    {
      "epoch": 0.7591435185185185,
      "grad_norm": 0.12259165197610855,
      "learning_rate": 4.81712962962963e-06,
      "loss": 0.0112,
      "step": 6559
    },
    {
      "epoch": 0.7592592592592593,
      "grad_norm": 0.07182811200618744,
      "learning_rate": 4.814814814814815e-06,
      "loss": 0.0102,
      "step": 6560
    },
    {
      "epoch": 0.759375,
      "grad_norm": 0.049288611859083176,
      "learning_rate": 4.8125e-06,
      "loss": 0.0065,
      "step": 6561
    },
    {
      "epoch": 0.7594907407407407,
      "grad_norm": 0.04536883533000946,
      "learning_rate": 4.810185185185185e-06,
      "loss": 0.0081,
      "step": 6562
    },
    {
      "epoch": 0.7596064814814815,
      "grad_norm": 0.04856086149811745,
      "learning_rate": 4.80787037037037e-06,
      "loss": 0.0087,
      "step": 6563
    },
    {
      "epoch": 0.7597222222222222,
      "grad_norm": 0.04293002933263779,
      "learning_rate": 4.805555555555556e-06,
      "loss": 0.0078,
      "step": 6564
    },
    {
      "epoch": 0.7598379629629629,
      "grad_norm": 0.07419956475496292,
      "learning_rate": 4.803240740740741e-06,
      "loss": 0.0099,
      "step": 6565
    },
    {
      "epoch": 0.7599537037037037,
      "grad_norm": 0.05982745438814163,
      "learning_rate": 4.800925925925926e-06,
      "loss": 0.0091,
      "step": 6566
    },
    {
      "epoch": 0.7600694444444445,
      "grad_norm": 16.44620132446289,
      "learning_rate": 4.7986111111111115e-06,
      "loss": 0.0482,
      "step": 6567
    },
    {
      "epoch": 0.7601851851851852,
      "grad_norm": 0.03590581938624382,
      "learning_rate": 4.796296296296297e-06,
      "loss": 0.0065,
      "step": 6568
    },
    {
      "epoch": 0.7603009259259259,
      "grad_norm": 49.547447204589844,
      "learning_rate": 4.793981481481482e-06,
      "loss": 2.4688,
      "step": 6569
    },
    {
      "epoch": 0.7604166666666666,
      "grad_norm": 0.03875484690070152,
      "learning_rate": 4.791666666666668e-06,
      "loss": 0.0071,
      "step": 6570
    },
    {
      "epoch": 0.7605324074074075,
      "grad_norm": 0.13316519558429718,
      "learning_rate": 4.789351851851852e-06,
      "loss": 0.0085,
      "step": 6571
    },
    {
      "epoch": 0.7606481481481482,
      "grad_norm": 0.05008929967880249,
      "learning_rate": 4.787037037037037e-06,
      "loss": 0.0088,
      "step": 6572
    },
    {
      "epoch": 0.7607638888888889,
      "grad_norm": 0.050536587834358215,
      "learning_rate": 4.784722222222223e-06,
      "loss": 0.009,
      "step": 6573
    },
    {
      "epoch": 0.7608796296296296,
      "grad_norm": 0.07052920013666153,
      "learning_rate": 4.782407407407408e-06,
      "loss": 0.0097,
      "step": 6574
    },
    {
      "epoch": 0.7609953703703703,
      "grad_norm": 0.05078499764204025,
      "learning_rate": 4.780092592592593e-06,
      "loss": 0.0089,
      "step": 6575
    },
    {
      "epoch": 0.7611111111111111,
      "grad_norm": 0.05190352350473404,
      "learning_rate": 4.777777777777778e-06,
      "loss": 0.0094,
      "step": 6576
    },
    {
      "epoch": 0.7612268518518519,
      "grad_norm": 0.05765308067202568,
      "learning_rate": 4.775462962962963e-06,
      "loss": 0.0094,
      "step": 6577
    },
    {
      "epoch": 0.7613425925925926,
      "grad_norm": 0.04074626415967941,
      "learning_rate": 4.7731481481481484e-06,
      "loss": 0.0073,
      "step": 6578
    },
    {
      "epoch": 0.7614583333333333,
      "grad_norm": 0.049845390021800995,
      "learning_rate": 4.770833333333334e-06,
      "loss": 0.0065,
      "step": 6579
    },
    {
      "epoch": 0.7615740740740741,
      "grad_norm": 0.05299181118607521,
      "learning_rate": 4.768518518518519e-06,
      "loss": 0.0069,
      "step": 6580
    },
    {
      "epoch": 0.7616898148148148,
      "grad_norm": 0.05982048064470291,
      "learning_rate": 4.766203703703704e-06,
      "loss": 0.0109,
      "step": 6581
    },
    {
      "epoch": 0.7618055555555555,
      "grad_norm": 0.0649925023317337,
      "learning_rate": 4.763888888888889e-06,
      "loss": 0.012,
      "step": 6582
    },
    {
      "epoch": 0.7619212962962963,
      "grad_norm": 0.08863679319620132,
      "learning_rate": 4.761574074074075e-06,
      "loss": 0.0104,
      "step": 6583
    },
    {
      "epoch": 0.7620370370370371,
      "grad_norm": 0.056317444890737534,
      "learning_rate": 4.75925925925926e-06,
      "loss": 0.0099,
      "step": 6584
    },
    {
      "epoch": 0.7621527777777778,
      "grad_norm": 0.058416180312633514,
      "learning_rate": 4.756944444444445e-06,
      "loss": 0.0105,
      "step": 6585
    },
    {
      "epoch": 0.7622685185185185,
      "grad_norm": 0.035078272223472595,
      "learning_rate": 4.75462962962963e-06,
      "loss": 0.0064,
      "step": 6586
    },
    {
      "epoch": 0.7623842592592592,
      "grad_norm": 0.0657825842499733,
      "learning_rate": 4.752314814814815e-06,
      "loss": 0.0121,
      "step": 6587
    },
    {
      "epoch": 0.7625,
      "grad_norm": 0.6987703442573547,
      "learning_rate": 4.75e-06,
      "loss": 0.0143,
      "step": 6588
    },
    {
      "epoch": 0.7626157407407408,
      "grad_norm": 0.04829777777194977,
      "learning_rate": 4.747685185185185e-06,
      "loss": 0.0086,
      "step": 6589
    },
    {
      "epoch": 0.7627314814814815,
      "grad_norm": 0.04594467207789421,
      "learning_rate": 4.7453703703703705e-06,
      "loss": 0.0084,
      "step": 6590
    },
    {
      "epoch": 0.7628472222222222,
      "grad_norm": 0.054706260561943054,
      "learning_rate": 4.7430555555555556e-06,
      "loss": 0.01,
      "step": 6591
    },
    {
      "epoch": 0.762962962962963,
      "grad_norm": 0.040827322751283646,
      "learning_rate": 4.7407407407407415e-06,
      "loss": 0.0074,
      "step": 6592
    },
    {
      "epoch": 0.7630787037037037,
      "grad_norm": 0.18748782575130463,
      "learning_rate": 4.738425925925927e-06,
      "loss": 0.0126,
      "step": 6593
    },
    {
      "epoch": 0.7631944444444444,
      "grad_norm": 0.040428295731544495,
      "learning_rate": 4.736111111111112e-06,
      "loss": 0.0073,
      "step": 6594
    },
    {
      "epoch": 0.7633101851851852,
      "grad_norm": 0.03822634369134903,
      "learning_rate": 4.733796296296297e-06,
      "loss": 0.007,
      "step": 6595
    },
    {
      "epoch": 0.763425925925926,
      "grad_norm": 0.03537425771355629,
      "learning_rate": 4.731481481481482e-06,
      "loss": 0.0065,
      "step": 6596
    },
    {
      "epoch": 0.7635416666666667,
      "grad_norm": 0.0742182582616806,
      "learning_rate": 4.729166666666667e-06,
      "loss": 0.013,
      "step": 6597
    },
    {
      "epoch": 0.7636574074074074,
      "grad_norm": 0.05175101011991501,
      "learning_rate": 4.726851851851852e-06,
      "loss": 0.0093,
      "step": 6598
    },
    {
      "epoch": 0.7637731481481481,
      "grad_norm": 0.05484582856297493,
      "learning_rate": 4.724537037037037e-06,
      "loss": 0.01,
      "step": 6599
    },
    {
      "epoch": 0.7638888888888888,
      "grad_norm": 0.03507572039961815,
      "learning_rate": 4.722222222222222e-06,
      "loss": 0.0064,
      "step": 6600
    },
    {
      "epoch": 0.7640046296296297,
      "grad_norm": 0.06039813905954361,
      "learning_rate": 4.719907407407408e-06,
      "loss": 0.011,
      "step": 6601
    },
    {
      "epoch": 0.7641203703703704,
      "grad_norm": 0.052371636033058167,
      "learning_rate": 4.717592592592593e-06,
      "loss": 0.0092,
      "step": 6602
    },
    {
      "epoch": 0.7642361111111111,
      "grad_norm": 0.055169276893138885,
      "learning_rate": 4.715277777777778e-06,
      "loss": 0.0101,
      "step": 6603
    },
    {
      "epoch": 0.7643518518518518,
      "grad_norm": 0.3292028307914734,
      "learning_rate": 4.712962962962963e-06,
      "loss": 0.0131,
      "step": 6604
    },
    {
      "epoch": 0.7644675925925926,
      "grad_norm": 0.0748540610074997,
      "learning_rate": 4.710648148148149e-06,
      "loss": 0.013,
      "step": 6605
    },
    {
      "epoch": 0.7645833333333333,
      "grad_norm": 0.3501707911491394,
      "learning_rate": 4.708333333333334e-06,
      "loss": 0.0118,
      "step": 6606
    },
    {
      "epoch": 0.7646990740740741,
      "grad_norm": 0.05070902407169342,
      "learning_rate": 4.706018518518519e-06,
      "loss": 0.0089,
      "step": 6607
    },
    {
      "epoch": 0.7648148148148148,
      "grad_norm": 0.04162399098277092,
      "learning_rate": 4.703703703703704e-06,
      "loss": 0.0068,
      "step": 6608
    },
    {
      "epoch": 0.7649305555555556,
      "grad_norm": 0.039907004684209824,
      "learning_rate": 4.701388888888889e-06,
      "loss": 0.0072,
      "step": 6609
    },
    {
      "epoch": 0.7650462962962963,
      "grad_norm": 0.05182981118559837,
      "learning_rate": 4.699074074074074e-06,
      "loss": 0.0089,
      "step": 6610
    },
    {
      "epoch": 0.765162037037037,
      "grad_norm": 0.1428985595703125,
      "learning_rate": 4.69675925925926e-06,
      "loss": 0.0097,
      "step": 6611
    },
    {
      "epoch": 0.7652777777777777,
      "grad_norm": 0.0589635893702507,
      "learning_rate": 4.694444444444445e-06,
      "loss": 0.0077,
      "step": 6612
    },
    {
      "epoch": 0.7653935185185186,
      "grad_norm": 0.05221617594361305,
      "learning_rate": 4.692129629629629e-06,
      "loss": 0.0091,
      "step": 6613
    },
    {
      "epoch": 0.7655092592592593,
      "grad_norm": 0.050700705498456955,
      "learning_rate": 4.689814814814815e-06,
      "loss": 0.009,
      "step": 6614
    },
    {
      "epoch": 0.765625,
      "grad_norm": 0.049342814832925797,
      "learning_rate": 4.6875000000000004e-06,
      "loss": 0.0064,
      "step": 6615
    },
    {
      "epoch": 0.7657407407407407,
      "grad_norm": 0.04459373280405998,
      "learning_rate": 4.6851851851851855e-06,
      "loss": 0.008,
      "step": 6616
    },
    {
      "epoch": 0.7658564814814814,
      "grad_norm": 0.04193858429789543,
      "learning_rate": 4.682870370370371e-06,
      "loss": 0.0077,
      "step": 6617
    },
    {
      "epoch": 0.7659722222222223,
      "grad_norm": 0.05175374820828438,
      "learning_rate": 4.680555555555556e-06,
      "loss": 0.0093,
      "step": 6618
    },
    {
      "epoch": 0.766087962962963,
      "grad_norm": 0.06030714139342308,
      "learning_rate": 4.678240740740741e-06,
      "loss": 0.011,
      "step": 6619
    },
    {
      "epoch": 0.7662037037037037,
      "grad_norm": 0.0479070246219635,
      "learning_rate": 4.675925925925927e-06,
      "loss": 0.0063,
      "step": 6620
    },
    {
      "epoch": 0.7663194444444444,
      "grad_norm": 0.47410163283348083,
      "learning_rate": 4.673611111111112e-06,
      "loss": 0.0166,
      "step": 6621
    },
    {
      "epoch": 0.7664351851851852,
      "grad_norm": 0.05776611715555191,
      "learning_rate": 4.671296296296297e-06,
      "loss": 0.0104,
      "step": 6622
    },
    {
      "epoch": 0.7665509259259259,
      "grad_norm": 0.08943536132574081,
      "learning_rate": 4.668981481481482e-06,
      "loss": 0.0101,
      "step": 6623
    },
    {
      "epoch": 0.7666666666666667,
      "grad_norm": 0.05836829915642738,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0096,
      "step": 6624
    },
    {
      "epoch": 0.7667824074074074,
      "grad_norm": 0.1609499752521515,
      "learning_rate": 4.664351851851852e-06,
      "loss": 0.0084,
      "step": 6625
    },
    {
      "epoch": 0.7668981481481482,
      "grad_norm": 199.68832397460938,
      "learning_rate": 4.662037037037037e-06,
      "loss": 4.0659,
      "step": 6626
    },
    {
      "epoch": 0.7670138888888889,
      "grad_norm": 0.054175395518541336,
      "learning_rate": 4.6597222222222225e-06,
      "loss": 0.0097,
      "step": 6627
    },
    {
      "epoch": 0.7671296296296296,
      "grad_norm": 0.03979970142245293,
      "learning_rate": 4.6574074074074076e-06,
      "loss": 0.0072,
      "step": 6628
    },
    {
      "epoch": 0.7672453703703703,
      "grad_norm": 0.05169770121574402,
      "learning_rate": 4.655092592592593e-06,
      "loss": 0.0091,
      "step": 6629
    },
    {
      "epoch": 0.7673611111111112,
      "grad_norm": 0.06535446643829346,
      "learning_rate": 4.652777777777779e-06,
      "loss": 0.0119,
      "step": 6630
    },
    {
      "epoch": 0.7674768518518519,
      "grad_norm": 0.045100048184394836,
      "learning_rate": 4.650462962962964e-06,
      "loss": 0.007,
      "step": 6631
    },
    {
      "epoch": 0.7675925925925926,
      "grad_norm": 0.053071364760398865,
      "learning_rate": 4.648148148148148e-06,
      "loss": 0.0094,
      "step": 6632
    },
    {
      "epoch": 0.7677083333333333,
      "grad_norm": 12.587623596191406,
      "learning_rate": 4.645833333333334e-06,
      "loss": 0.0611,
      "step": 6633
    },
    {
      "epoch": 0.767824074074074,
      "grad_norm": 0.04134674742817879,
      "learning_rate": 4.643518518518519e-06,
      "loss": 0.0076,
      "step": 6634
    },
    {
      "epoch": 0.7679398148148148,
      "grad_norm": 0.05537194013595581,
      "learning_rate": 4.641203703703704e-06,
      "loss": 0.0098,
      "step": 6635
    },
    {
      "epoch": 0.7680555555555556,
      "grad_norm": 111.41043090820312,
      "learning_rate": 4.638888888888889e-06,
      "loss": 2.2926,
      "step": 6636
    },
    {
      "epoch": 0.7681712962962963,
      "grad_norm": 0.04817972704768181,
      "learning_rate": 4.636574074074074e-06,
      "loss": 0.0063,
      "step": 6637
    },
    {
      "epoch": 0.768287037037037,
      "grad_norm": 0.0650693029165268,
      "learning_rate": 4.634259259259259e-06,
      "loss": 0.0119,
      "step": 6638
    },
    {
      "epoch": 0.7684027777777778,
      "grad_norm": 0.038556165993213654,
      "learning_rate": 4.631944444444445e-06,
      "loss": 0.007,
      "step": 6639
    },
    {
      "epoch": 0.7685185185185185,
      "grad_norm": 0.05688982084393501,
      "learning_rate": 4.62962962962963e-06,
      "loss": 0.0102,
      "step": 6640
    },
    {
      "epoch": 0.7686342592592592,
      "grad_norm": 0.05872650444507599,
      "learning_rate": 4.627314814814815e-06,
      "loss": 0.0102,
      "step": 6641
    },
    {
      "epoch": 0.76875,
      "grad_norm": 0.06985107064247131,
      "learning_rate": 4.625000000000001e-06,
      "loss": 0.012,
      "step": 6642
    },
    {
      "epoch": 0.7688657407407408,
      "grad_norm": 0.03844340145587921,
      "learning_rate": 4.622685185185186e-06,
      "loss": 0.007,
      "step": 6643
    },
    {
      "epoch": 0.7689814814814815,
      "grad_norm": 96.110595703125,
      "learning_rate": 4.620370370370371e-06,
      "loss": 2.0901,
      "step": 6644
    },
    {
      "epoch": 0.7690972222222222,
      "grad_norm": 0.05119975283741951,
      "learning_rate": 4.618055555555556e-06,
      "loss": 0.0093,
      "step": 6645
    },
    {
      "epoch": 0.7692129629629629,
      "grad_norm": 9.668940544128418,
      "learning_rate": 4.615740740740741e-06,
      "loss": 2.9713,
      "step": 6646
    },
    {
      "epoch": 0.7693287037037037,
      "grad_norm": 0.06001812964677811,
      "learning_rate": 4.613425925925926e-06,
      "loss": 0.0102,
      "step": 6647
    },
    {
      "epoch": 0.7694444444444445,
      "grad_norm": 0.051618821918964386,
      "learning_rate": 4.611111111111112e-06,
      "loss": 0.0093,
      "step": 6648
    },
    {
      "epoch": 0.7695601851851852,
      "grad_norm": 0.08247060328722,
      "learning_rate": 4.608796296296297e-06,
      "loss": 0.0109,
      "step": 6649
    },
    {
      "epoch": 0.7696759259259259,
      "grad_norm": 0.06689513474702835,
      "learning_rate": 4.606481481481481e-06,
      "loss": 0.0117,
      "step": 6650
    },
    {
      "epoch": 0.7697916666666667,
      "grad_norm": 0.06545014679431915,
      "learning_rate": 4.6041666666666665e-06,
      "loss": 0.0099,
      "step": 6651
    },
    {
      "epoch": 0.7699074074074074,
      "grad_norm": 0.03474198654294014,
      "learning_rate": 4.6018518518518524e-06,
      "loss": 0.0063,
      "step": 6652
    },
    {
      "epoch": 0.7700231481481481,
      "grad_norm": 0.06143875792622566,
      "learning_rate": 4.5995370370370375e-06,
      "loss": 0.0112,
      "step": 6653
    },
    {
      "epoch": 0.7701388888888889,
      "grad_norm": 0.06609024107456207,
      "learning_rate": 4.597222222222223e-06,
      "loss": 0.0109,
      "step": 6654
    },
    {
      "epoch": 0.7702546296296297,
      "grad_norm": 0.042430147528648376,
      "learning_rate": 4.594907407407408e-06,
      "loss": 0.0077,
      "step": 6655
    },
    {
      "epoch": 0.7703703703703704,
      "grad_norm": 0.03735155612230301,
      "learning_rate": 4.592592592592593e-06,
      "loss": 0.0067,
      "step": 6656
    },
    {
      "epoch": 0.7704861111111111,
      "grad_norm": 0.05630578100681305,
      "learning_rate": 4.590277777777778e-06,
      "loss": 0.01,
      "step": 6657
    },
    {
      "epoch": 0.7706018518518518,
      "grad_norm": 0.23070888221263885,
      "learning_rate": 4.587962962962964e-06,
      "loss": 0.0147,
      "step": 6658
    },
    {
      "epoch": 0.7707175925925925,
      "grad_norm": 0.06053492799401283,
      "learning_rate": 4.585648148148148e-06,
      "loss": 0.0077,
      "step": 6659
    },
    {
      "epoch": 0.7708333333333334,
      "grad_norm": 0.05742618069052696,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.009,
      "step": 6660
    },
    {
      "epoch": 0.7709490740740741,
      "grad_norm": 0.04776223003864288,
      "learning_rate": 4.581018518518519e-06,
      "loss": 0.0085,
      "step": 6661
    },
    {
      "epoch": 0.7710648148148148,
      "grad_norm": 0.14966268837451935,
      "learning_rate": 4.578703703703704e-06,
      "loss": 0.0101,
      "step": 6662
    },
    {
      "epoch": 0.7711805555555555,
      "grad_norm": 0.06845873594284058,
      "learning_rate": 4.576388888888889e-06,
      "loss": 0.0125,
      "step": 6663
    },
    {
      "epoch": 0.7712962962962963,
      "grad_norm": 0.03981199115514755,
      "learning_rate": 4.5740740740740745e-06,
      "loss": 0.0072,
      "step": 6664
    },
    {
      "epoch": 0.7714120370370371,
      "grad_norm": 0.0687527284026146,
      "learning_rate": 4.5717592592592595e-06,
      "loss": 0.0123,
      "step": 6665
    },
    {
      "epoch": 0.7715277777777778,
      "grad_norm": 0.0407521091401577,
      "learning_rate": 4.569444444444445e-06,
      "loss": 0.0073,
      "step": 6666
    },
    {
      "epoch": 0.7716435185185185,
      "grad_norm": 0.04788772389292717,
      "learning_rate": 4.567129629629631e-06,
      "loss": 0.0085,
      "step": 6667
    },
    {
      "epoch": 0.7717592592592593,
      "grad_norm": 0.06121806427836418,
      "learning_rate": 4.564814814814815e-06,
      "loss": 0.0097,
      "step": 6668
    },
    {
      "epoch": 0.771875,
      "grad_norm": 0.05747801437973976,
      "learning_rate": 4.5625e-06,
      "loss": 0.0098,
      "step": 6669
    },
    {
      "epoch": 0.7719907407407407,
      "grad_norm": 0.04417180269956589,
      "learning_rate": 4.560185185185186e-06,
      "loss": 0.0077,
      "step": 6670
    },
    {
      "epoch": 0.7721064814814815,
      "grad_norm": 0.05937010794878006,
      "learning_rate": 4.557870370370371e-06,
      "loss": 0.0105,
      "step": 6671
    },
    {
      "epoch": 0.7722222222222223,
      "grad_norm": 0.9796828031539917,
      "learning_rate": 4.555555555555556e-06,
      "loss": 0.0212,
      "step": 6672
    },
    {
      "epoch": 0.772337962962963,
      "grad_norm": 0.05560605227947235,
      "learning_rate": 4.553240740740741e-06,
      "loss": 0.0096,
      "step": 6673
    },
    {
      "epoch": 0.7724537037037037,
      "grad_norm": 0.043028514832258224,
      "learning_rate": 4.550925925925926e-06,
      "loss": 0.0078,
      "step": 6674
    },
    {
      "epoch": 0.7725694444444444,
      "grad_norm": 0.10327416658401489,
      "learning_rate": 4.548611111111111e-06,
      "loss": 0.0135,
      "step": 6675
    },
    {
      "epoch": 0.7726851851851851,
      "grad_norm": 0.0580814965069294,
      "learning_rate": 4.5462962962962965e-06,
      "loss": 0.0076,
      "step": 6676
    },
    {
      "epoch": 0.772800925925926,
      "grad_norm": 0.03540094941854477,
      "learning_rate": 4.543981481481482e-06,
      "loss": 0.0064,
      "step": 6677
    },
    {
      "epoch": 0.7729166666666667,
      "grad_norm": 0.03504202142357826,
      "learning_rate": 4.541666666666667e-06,
      "loss": 0.0063,
      "step": 6678
    },
    {
      "epoch": 0.7730324074074074,
      "grad_norm": 0.052386339753866196,
      "learning_rate": 4.539351851851852e-06,
      "loss": 0.0091,
      "step": 6679
    },
    {
      "epoch": 0.7731481481481481,
      "grad_norm": 0.03808993101119995,
      "learning_rate": 4.537037037037038e-06,
      "loss": 0.0068,
      "step": 6680
    },
    {
      "epoch": 0.7732638888888889,
      "grad_norm": 0.050328779965639114,
      "learning_rate": 4.534722222222223e-06,
      "loss": 0.0091,
      "step": 6681
    },
    {
      "epoch": 0.7733796296296296,
      "grad_norm": 0.05696183070540428,
      "learning_rate": 4.532407407407408e-06,
      "loss": 0.01,
      "step": 6682
    },
    {
      "epoch": 0.7734953703703704,
      "grad_norm": 0.0414244644343853,
      "learning_rate": 4.530092592592593e-06,
      "loss": 0.0075,
      "step": 6683
    },
    {
      "epoch": 0.7736111111111111,
      "grad_norm": 0.05567879229784012,
      "learning_rate": 4.527777777777778e-06,
      "loss": 0.0074,
      "step": 6684
    },
    {
      "epoch": 0.7737268518518519,
      "grad_norm": 0.03426656126976013,
      "learning_rate": 4.525462962962963e-06,
      "loss": 0.0063,
      "step": 6685
    },
    {
      "epoch": 0.7738425925925926,
      "grad_norm": 0.038021765649318695,
      "learning_rate": 4.523148148148149e-06,
      "loss": 0.0067,
      "step": 6686
    },
    {
      "epoch": 0.7739583333333333,
      "grad_norm": 244.4409942626953,
      "learning_rate": 4.520833333333333e-06,
      "loss": 1.0352,
      "step": 6687
    },
    {
      "epoch": 0.774074074074074,
      "grad_norm": 0.0659608542919159,
      "learning_rate": 4.5185185185185185e-06,
      "loss": 0.012,
      "step": 6688
    },
    {
      "epoch": 0.7741898148148149,
      "grad_norm": 0.03620794415473938,
      "learning_rate": 4.5162037037037044e-06,
      "loss": 0.0066,
      "step": 6689
    },
    {
      "epoch": 0.7743055555555556,
      "grad_norm": 0.06546394526958466,
      "learning_rate": 4.5138888888888895e-06,
      "loss": 0.0102,
      "step": 6690
    },
    {
      "epoch": 0.7744212962962963,
      "grad_norm": 0.03684345260262489,
      "learning_rate": 4.511574074074075e-06,
      "loss": 0.0066,
      "step": 6691
    },
    {
      "epoch": 0.774537037037037,
      "grad_norm": 12.08862590789795,
      "learning_rate": 4.50925925925926e-06,
      "loss": 0.0433,
      "step": 6692
    },
    {
      "epoch": 0.7746527777777777,
      "grad_norm": 0.09466026723384857,
      "learning_rate": 4.506944444444445e-06,
      "loss": 0.0079,
      "step": 6693
    },
    {
      "epoch": 0.7747685185185185,
      "grad_norm": 0.03797553852200508,
      "learning_rate": 4.50462962962963e-06,
      "loss": 0.0069,
      "step": 6694
    },
    {
      "epoch": 0.7748842592592593,
      "grad_norm": 0.04023759439587593,
      "learning_rate": 4.502314814814815e-06,
      "loss": 0.007,
      "step": 6695
    },
    {
      "epoch": 0.775,
      "grad_norm": 154.99888610839844,
      "learning_rate": 4.5e-06,
      "loss": 0.5259,
      "step": 6696
    },
    {
      "epoch": 0.7751157407407407,
      "grad_norm": 0.04107194021344185,
      "learning_rate": 4.497685185185185e-06,
      "loss": 0.0075,
      "step": 6697
    },
    {
      "epoch": 0.7752314814814815,
      "grad_norm": 0.06999941170215607,
      "learning_rate": 4.49537037037037e-06,
      "loss": 0.009,
      "step": 6698
    },
    {
      "epoch": 0.7753472222222222,
      "grad_norm": 0.12496853619813919,
      "learning_rate": 4.493055555555556e-06,
      "loss": 0.0118,
      "step": 6699
    },
    {
      "epoch": 0.7754629629629629,
      "grad_norm": 0.06052244454622269,
      "learning_rate": 4.490740740740741e-06,
      "loss": 0.0106,
      "step": 6700
    },
    {
      "epoch": 0.7755787037037037,
      "grad_norm": 0.06538309901952744,
      "learning_rate": 4.4884259259259264e-06,
      "loss": 0.0119,
      "step": 6701
    },
    {
      "epoch": 0.7756944444444445,
      "grad_norm": 0.05242575705051422,
      "learning_rate": 4.4861111111111115e-06,
      "loss": 0.0093,
      "step": 6702
    },
    {
      "epoch": 0.7758101851851852,
      "grad_norm": 0.06708259135484695,
      "learning_rate": 4.483796296296297e-06,
      "loss": 0.0122,
      "step": 6703
    },
    {
      "epoch": 0.7759259259259259,
      "grad_norm": 0.06267187744379044,
      "learning_rate": 4.481481481481482e-06,
      "loss": 0.0091,
      "step": 6704
    },
    {
      "epoch": 0.7760416666666666,
      "grad_norm": 0.05415322259068489,
      "learning_rate": 4.479166666666667e-06,
      "loss": 0.0094,
      "step": 6705
    },
    {
      "epoch": 0.7761574074074075,
      "grad_norm": 5.245875835418701,
      "learning_rate": 4.476851851851852e-06,
      "loss": 0.0258,
      "step": 6706
    },
    {
      "epoch": 0.7762731481481482,
      "grad_norm": 0.08752545714378357,
      "learning_rate": 4.474537037037037e-06,
      "loss": 0.0115,
      "step": 6707
    },
    {
      "epoch": 0.7763888888888889,
      "grad_norm": 0.06051236763596535,
      "learning_rate": 4.472222222222223e-06,
      "loss": 0.0105,
      "step": 6708
    },
    {
      "epoch": 0.7765046296296296,
      "grad_norm": 0.041033241897821426,
      "learning_rate": 4.469907407407408e-06,
      "loss": 0.0075,
      "step": 6709
    },
    {
      "epoch": 0.7766203703703703,
      "grad_norm": 0.6147431135177612,
      "learning_rate": 4.467592592592593e-06,
      "loss": 0.0152,
      "step": 6710
    },
    {
      "epoch": 0.7767361111111111,
      "grad_norm": 106.51226806640625,
      "learning_rate": 4.465277777777778e-06,
      "loss": 2.0024,
      "step": 6711
    },
    {
      "epoch": 0.7768518518518519,
      "grad_norm": 0.04760466516017914,
      "learning_rate": 4.462962962962963e-06,
      "loss": 0.0085,
      "step": 6712
    },
    {
      "epoch": 0.7769675925925926,
      "grad_norm": 0.058038730174303055,
      "learning_rate": 4.4606481481481485e-06,
      "loss": 0.0106,
      "step": 6713
    },
    {
      "epoch": 0.7770833333333333,
      "grad_norm": 0.060583893209695816,
      "learning_rate": 4.4583333333333336e-06,
      "loss": 0.0109,
      "step": 6714
    },
    {
      "epoch": 0.7771990740740741,
      "grad_norm": 0.0427621454000473,
      "learning_rate": 4.456018518518519e-06,
      "loss": 0.0076,
      "step": 6715
    },
    {
      "epoch": 0.7773148148148148,
      "grad_norm": 0.0538984090089798,
      "learning_rate": 4.453703703703704e-06,
      "loss": 0.0092,
      "step": 6716
    },
    {
      "epoch": 0.7774305555555555,
      "grad_norm": 0.051109712570905685,
      "learning_rate": 4.451388888888889e-06,
      "loss": 0.0091,
      "step": 6717
    },
    {
      "epoch": 0.7775462962962963,
      "grad_norm": 0.060771819204092026,
      "learning_rate": 4.449074074074075e-06,
      "loss": 0.0105,
      "step": 6718
    },
    {
      "epoch": 0.7776620370370371,
      "grad_norm": 0.04223253205418587,
      "learning_rate": 4.44675925925926e-06,
      "loss": 0.0076,
      "step": 6719
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 0.09235592931509018,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.0094,
      "step": 6720
    },
    {
      "epoch": 0.7778935185185185,
      "grad_norm": 0.050504062324762344,
      "learning_rate": 4.44212962962963e-06,
      "loss": 0.0089,
      "step": 6721
    },
    {
      "epoch": 0.7780092592592592,
      "grad_norm": 0.039137665182352066,
      "learning_rate": 4.439814814814815e-06,
      "loss": 0.007,
      "step": 6722
    },
    {
      "epoch": 0.778125,
      "grad_norm": 0.07467035204172134,
      "learning_rate": 4.4375e-06,
      "loss": 0.0104,
      "step": 6723
    },
    {
      "epoch": 0.7782407407407408,
      "grad_norm": 0.07121534645557404,
      "learning_rate": 4.435185185185185e-06,
      "loss": 0.0076,
      "step": 6724
    },
    {
      "epoch": 0.7783564814814815,
      "grad_norm": 0.034037552773952484,
      "learning_rate": 4.4328703703703705e-06,
      "loss": 0.0062,
      "step": 6725
    },
    {
      "epoch": 0.7784722222222222,
      "grad_norm": 11.984622955322266,
      "learning_rate": 4.430555555555556e-06,
      "loss": 2.8556,
      "step": 6726
    },
    {
      "epoch": 0.778587962962963,
      "grad_norm": 0.8967981338500977,
      "learning_rate": 4.4282407407407415e-06,
      "loss": 0.0114,
      "step": 6727
    },
    {
      "epoch": 0.7787037037037037,
      "grad_norm": 0.09302809089422226,
      "learning_rate": 4.425925925925927e-06,
      "loss": 0.0126,
      "step": 6728
    },
    {
      "epoch": 0.7788194444444444,
      "grad_norm": 1.0588088035583496,
      "learning_rate": 4.423611111111112e-06,
      "loss": 0.0244,
      "step": 6729
    },
    {
      "epoch": 0.7789351851851852,
      "grad_norm": 0.0518685057759285,
      "learning_rate": 4.421296296296297e-06,
      "loss": 0.0093,
      "step": 6730
    },
    {
      "epoch": 0.779050925925926,
      "grad_norm": 141.48207092285156,
      "learning_rate": 4.418981481481482e-06,
      "loss": 1.7091,
      "step": 6731
    },
    {
      "epoch": 0.7791666666666667,
      "grad_norm": 0.038458775728940964,
      "learning_rate": 4.416666666666667e-06,
      "loss": 0.0069,
      "step": 6732
    },
    {
      "epoch": 0.7792824074074074,
      "grad_norm": 0.21107636392116547,
      "learning_rate": 4.414351851851852e-06,
      "loss": 0.0081,
      "step": 6733
    },
    {
      "epoch": 0.7793981481481481,
      "grad_norm": 0.03912162035703659,
      "learning_rate": 4.412037037037037e-06,
      "loss": 0.0071,
      "step": 6734
    },
    {
      "epoch": 0.7795138888888888,
      "grad_norm": 0.038864631205797195,
      "learning_rate": 4.409722222222222e-06,
      "loss": 0.0071,
      "step": 6735
    },
    {
      "epoch": 0.7796296296296297,
      "grad_norm": 0.05569378286600113,
      "learning_rate": 4.407407407407408e-06,
      "loss": 0.0097,
      "step": 6736
    },
    {
      "epoch": 0.7797453703703704,
      "grad_norm": 0.060516487807035446,
      "learning_rate": 4.405092592592593e-06,
      "loss": 0.0091,
      "step": 6737
    },
    {
      "epoch": 0.7798611111111111,
      "grad_norm": 0.08741983026266098,
      "learning_rate": 4.4027777777777784e-06,
      "loss": 0.0115,
      "step": 6738
    },
    {
      "epoch": 0.7799768518518518,
      "grad_norm": 0.20188112556934357,
      "learning_rate": 4.400462962962963e-06,
      "loss": 0.0107,
      "step": 6739
    },
    {
      "epoch": 0.7800925925925926,
      "grad_norm": 0.12705019116401672,
      "learning_rate": 4.398148148148149e-06,
      "loss": 0.0119,
      "step": 6740
    },
    {
      "epoch": 0.7802083333333333,
      "grad_norm": 97.99162292480469,
      "learning_rate": 4.395833333333334e-06,
      "loss": 2.2857,
      "step": 6741
    },
    {
      "epoch": 0.7803240740740741,
      "grad_norm": 0.038904741406440735,
      "learning_rate": 4.393518518518519e-06,
      "loss": 0.007,
      "step": 6742
    },
    {
      "epoch": 0.7804398148148148,
      "grad_norm": 0.05542830377817154,
      "learning_rate": 4.391203703703704e-06,
      "loss": 0.0097,
      "step": 6743
    },
    {
      "epoch": 0.7805555555555556,
      "grad_norm": 0.05302675440907478,
      "learning_rate": 4.388888888888889e-06,
      "loss": 0.0092,
      "step": 6744
    },
    {
      "epoch": 0.7806712962962963,
      "grad_norm": 0.04899488389492035,
      "learning_rate": 4.386574074074074e-06,
      "loss": 0.0084,
      "step": 6745
    },
    {
      "epoch": 0.780787037037037,
      "grad_norm": 0.05452665314078331,
      "learning_rate": 4.38425925925926e-06,
      "loss": 0.0097,
      "step": 6746
    },
    {
      "epoch": 0.7809027777777777,
      "grad_norm": 0.08639226853847504,
      "learning_rate": 4.381944444444445e-06,
      "loss": 0.0127,
      "step": 6747
    },
    {
      "epoch": 0.7810185185185186,
      "grad_norm": 0.05206408351659775,
      "learning_rate": 4.379629629629629e-06,
      "loss": 0.0089,
      "step": 6748
    },
    {
      "epoch": 0.7811342592592593,
      "grad_norm": 0.03870021179318428,
      "learning_rate": 4.377314814814815e-06,
      "loss": 0.0071,
      "step": 6749
    },
    {
      "epoch": 0.78125,
      "grad_norm": 0.04047176241874695,
      "learning_rate": 4.3750000000000005e-06,
      "loss": 0.0074,
      "step": 6750
    },
    {
      "epoch": 0.7813657407407407,
      "grad_norm": 0.035504426807165146,
      "learning_rate": 4.3726851851851856e-06,
      "loss": 0.0064,
      "step": 6751
    },
    {
      "epoch": 0.7814814814814814,
      "grad_norm": 0.06990595161914825,
      "learning_rate": 4.370370370370371e-06,
      "loss": 0.0091,
      "step": 6752
    },
    {
      "epoch": 0.7815972222222223,
      "grad_norm": 0.06132681295275688,
      "learning_rate": 4.368055555555556e-06,
      "loss": 0.009,
      "step": 6753
    },
    {
      "epoch": 0.781712962962963,
      "grad_norm": 0.06365890800952911,
      "learning_rate": 4.365740740740741e-06,
      "loss": 0.0116,
      "step": 6754
    },
    {
      "epoch": 0.7818287037037037,
      "grad_norm": 0.041559189558029175,
      "learning_rate": 4.363425925925927e-06,
      "loss": 0.0075,
      "step": 6755
    },
    {
      "epoch": 0.7819444444444444,
      "grad_norm": 0.06396596133708954,
      "learning_rate": 4.361111111111112e-06,
      "loss": 0.0117,
      "step": 6756
    },
    {
      "epoch": 0.7820601851851852,
      "grad_norm": 0.0358487106859684,
      "learning_rate": 4.358796296296296e-06,
      "loss": 0.0065,
      "step": 6757
    },
    {
      "epoch": 0.7821759259259259,
      "grad_norm": 0.05703434720635414,
      "learning_rate": 4.356481481481482e-06,
      "loss": 0.0098,
      "step": 6758
    },
    {
      "epoch": 0.7822916666666667,
      "grad_norm": 0.052351515740156174,
      "learning_rate": 4.354166666666667e-06,
      "loss": 0.0094,
      "step": 6759
    },
    {
      "epoch": 0.7824074074074074,
      "grad_norm": 0.059317976236343384,
      "learning_rate": 4.351851851851852e-06,
      "loss": 0.0108,
      "step": 6760
    },
    {
      "epoch": 0.7825231481481482,
      "grad_norm": 0.08820808678865433,
      "learning_rate": 4.349537037037037e-06,
      "loss": 0.0128,
      "step": 6761
    },
    {
      "epoch": 0.7826388888888889,
      "grad_norm": 0.049362100660800934,
      "learning_rate": 4.3472222222222225e-06,
      "loss": 0.0084,
      "step": 6762
    },
    {
      "epoch": 0.7827546296296296,
      "grad_norm": 0.03905647620558739,
      "learning_rate": 4.344907407407408e-06,
      "loss": 0.0071,
      "step": 6763
    },
    {
      "epoch": 0.7828703703703703,
      "grad_norm": 0.04382333159446716,
      "learning_rate": 4.342592592592593e-06,
      "loss": 0.0078,
      "step": 6764
    },
    {
      "epoch": 0.7829861111111112,
      "grad_norm": 119.0287857055664,
      "learning_rate": 4.340277777777779e-06,
      "loss": 0.2137,
      "step": 6765
    },
    {
      "epoch": 0.7831018518518519,
      "grad_norm": 0.05221165344119072,
      "learning_rate": 4.337962962962963e-06,
      "loss": 0.0095,
      "step": 6766
    },
    {
      "epoch": 0.7832175925925926,
      "grad_norm": 0.053766295313835144,
      "learning_rate": 4.335648148148148e-06,
      "loss": 0.0096,
      "step": 6767
    },
    {
      "epoch": 0.7833333333333333,
      "grad_norm": 0.1193341463804245,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.011,
      "step": 6768
    },
    {
      "epoch": 0.783449074074074,
      "grad_norm": 0.1145375594496727,
      "learning_rate": 4.331018518518519e-06,
      "loss": 0.0106,
      "step": 6769
    },
    {
      "epoch": 0.7835648148148148,
      "grad_norm": 0.050688568502664566,
      "learning_rate": 4.328703703703704e-06,
      "loss": 0.0091,
      "step": 6770
    },
    {
      "epoch": 0.7836805555555556,
      "grad_norm": 0.08205993473529816,
      "learning_rate": 4.326388888888889e-06,
      "loss": 0.0098,
      "step": 6771
    },
    {
      "epoch": 0.7837962962962963,
      "grad_norm": 0.04783337935805321,
      "learning_rate": 4.324074074074074e-06,
      "loss": 0.0066,
      "step": 6772
    },
    {
      "epoch": 0.783912037037037,
      "grad_norm": 0.04779975116252899,
      "learning_rate": 4.321759259259259e-06,
      "loss": 0.0084,
      "step": 6773
    },
    {
      "epoch": 0.7840277777777778,
      "grad_norm": 0.06803233921527863,
      "learning_rate": 4.319444444444445e-06,
      "loss": 0.0089,
      "step": 6774
    },
    {
      "epoch": 0.7841435185185185,
      "grad_norm": 0.0897766649723053,
      "learning_rate": 4.31712962962963e-06,
      "loss": 0.008,
      "step": 6775
    },
    {
      "epoch": 0.7842592592592592,
      "grad_norm": 0.05252913013100624,
      "learning_rate": 4.314814814814815e-06,
      "loss": 0.0096,
      "step": 6776
    },
    {
      "epoch": 0.784375,
      "grad_norm": 0.046468526124954224,
      "learning_rate": 4.312500000000001e-06,
      "loss": 0.0082,
      "step": 6777
    },
    {
      "epoch": 0.7844907407407408,
      "grad_norm": 0.05315512418746948,
      "learning_rate": 4.310185185185186e-06,
      "loss": 0.0089,
      "step": 6778
    },
    {
      "epoch": 0.7846064814814815,
      "grad_norm": 0.04991280660033226,
      "learning_rate": 4.307870370370371e-06,
      "loss": 0.0087,
      "step": 6779
    },
    {
      "epoch": 0.7847222222222222,
      "grad_norm": 1.4450124502182007,
      "learning_rate": 4.305555555555556e-06,
      "loss": 0.019,
      "step": 6780
    },
    {
      "epoch": 0.7848379629629629,
      "grad_norm": 0.05458441376686096,
      "learning_rate": 4.303240740740741e-06,
      "loss": 0.0085,
      "step": 6781
    },
    {
      "epoch": 0.7849537037037037,
      "grad_norm": 0.04949164763092995,
      "learning_rate": 4.300925925925926e-06,
      "loss": 0.0085,
      "step": 6782
    },
    {
      "epoch": 0.7850694444444445,
      "grad_norm": 0.07002197951078415,
      "learning_rate": 4.298611111111112e-06,
      "loss": 0.0115,
      "step": 6783
    },
    {
      "epoch": 0.7851851851851852,
      "grad_norm": 0.04369253292679787,
      "learning_rate": 4.296296296296296e-06,
      "loss": 0.008,
      "step": 6784
    },
    {
      "epoch": 0.7853009259259259,
      "grad_norm": 0.0522800087928772,
      "learning_rate": 4.293981481481481e-06,
      "loss": 0.0094,
      "step": 6785
    },
    {
      "epoch": 0.7854166666666667,
      "grad_norm": 0.06903252750635147,
      "learning_rate": 4.2916666666666665e-06,
      "loss": 0.0124,
      "step": 6786
    },
    {
      "epoch": 0.7855324074074074,
      "grad_norm": 0.29576075077056885,
      "learning_rate": 4.2893518518518525e-06,
      "loss": 0.013,
      "step": 6787
    },
    {
      "epoch": 0.7856481481481481,
      "grad_norm": 0.04858291521668434,
      "learning_rate": 4.2870370370370376e-06,
      "loss": 0.0086,
      "step": 6788
    },
    {
      "epoch": 0.7857638888888889,
      "grad_norm": 0.0500439815223217,
      "learning_rate": 4.284722222222223e-06,
      "loss": 0.0084,
      "step": 6789
    },
    {
      "epoch": 0.7858796296296297,
      "grad_norm": 0.49295151233673096,
      "learning_rate": 4.282407407407408e-06,
      "loss": 0.0129,
      "step": 6790
    },
    {
      "epoch": 0.7859953703703704,
      "grad_norm": 0.06886748224496841,
      "learning_rate": 4.280092592592593e-06,
      "loss": 0.0091,
      "step": 6791
    },
    {
      "epoch": 0.7861111111111111,
      "grad_norm": 0.05943579599261284,
      "learning_rate": 4.277777777777778e-06,
      "loss": 0.0109,
      "step": 6792
    },
    {
      "epoch": 0.7862268518518518,
      "grad_norm": 0.06883658468723297,
      "learning_rate": 4.275462962962964e-06,
      "loss": 0.0114,
      "step": 6793
    },
    {
      "epoch": 0.7863425925925925,
      "grad_norm": 0.03981082886457443,
      "learning_rate": 4.273148148148148e-06,
      "loss": 0.0071,
      "step": 6794
    },
    {
      "epoch": 0.7864583333333334,
      "grad_norm": 0.09472519904375076,
      "learning_rate": 4.270833333333333e-06,
      "loss": 0.0106,
      "step": 6795
    },
    {
      "epoch": 0.7865740740740741,
      "grad_norm": 135.4068145751953,
      "learning_rate": 4.268518518518519e-06,
      "loss": 0.9426,
      "step": 6796
    },
    {
      "epoch": 0.7866898148148148,
      "grad_norm": 0.052548013627529144,
      "learning_rate": 4.266203703703704e-06,
      "loss": 0.0094,
      "step": 6797
    },
    {
      "epoch": 0.7868055555555555,
      "grad_norm": 0.062458883970975876,
      "learning_rate": 4.263888888888889e-06,
      "loss": 0.0105,
      "step": 6798
    },
    {
      "epoch": 0.7869212962962963,
      "grad_norm": 0.03336630389094353,
      "learning_rate": 4.2615740740740745e-06,
      "loss": 0.0061,
      "step": 6799
    },
    {
      "epoch": 0.7870370370370371,
      "grad_norm": 0.051956240087747574,
      "learning_rate": 4.2592592592592596e-06,
      "loss": 0.0089,
      "step": 6800
    },
    {
      "epoch": 0.7871527777777778,
      "grad_norm": 0.03490357846021652,
      "learning_rate": 4.256944444444445e-06,
      "loss": 0.0064,
      "step": 6801
    },
    {
      "epoch": 0.7872685185185185,
      "grad_norm": 0.12396878004074097,
      "learning_rate": 4.254629629629631e-06,
      "loss": 0.0131,
      "step": 6802
    },
    {
      "epoch": 0.7873842592592593,
      "grad_norm": 0.16530095040798187,
      "learning_rate": 4.252314814814815e-06,
      "loss": 0.011,
      "step": 6803
    },
    {
      "epoch": 0.7875,
      "grad_norm": 0.04431331157684326,
      "learning_rate": 4.25e-06,
      "loss": 0.0072,
      "step": 6804
    },
    {
      "epoch": 0.7876157407407407,
      "grad_norm": 0.03576250746846199,
      "learning_rate": 4.247685185185186e-06,
      "loss": 0.0065,
      "step": 6805
    },
    {
      "epoch": 0.7877314814814815,
      "grad_norm": 0.037814173847436905,
      "learning_rate": 4.245370370370371e-06,
      "loss": 0.0066,
      "step": 6806
    },
    {
      "epoch": 0.7878472222222223,
      "grad_norm": 0.05869524925947189,
      "learning_rate": 4.243055555555556e-06,
      "loss": 0.0107,
      "step": 6807
    },
    {
      "epoch": 0.787962962962963,
      "grad_norm": 0.05165952071547508,
      "learning_rate": 4.240740740740741e-06,
      "loss": 0.0085,
      "step": 6808
    },
    {
      "epoch": 0.7880787037037037,
      "grad_norm": 0.08236818760633469,
      "learning_rate": 4.238425925925926e-06,
      "loss": 0.0096,
      "step": 6809
    },
    {
      "epoch": 0.7881944444444444,
      "grad_norm": 0.04937172681093216,
      "learning_rate": 4.236111111111111e-06,
      "loss": 0.0088,
      "step": 6810
    },
    {
      "epoch": 0.7883101851851851,
      "grad_norm": 0.0496969148516655,
      "learning_rate": 4.2337962962962965e-06,
      "loss": 0.0071,
      "step": 6811
    },
    {
      "epoch": 0.788425925925926,
      "grad_norm": 0.03485452011227608,
      "learning_rate": 4.231481481481482e-06,
      "loss": 0.0063,
      "step": 6812
    },
    {
      "epoch": 0.7885416666666667,
      "grad_norm": 86.81930541992188,
      "learning_rate": 4.229166666666667e-06,
      "loss": 2.3435,
      "step": 6813
    },
    {
      "epoch": 0.7886574074074074,
      "grad_norm": 0.04674265906214714,
      "learning_rate": 4.226851851851852e-06,
      "loss": 0.0061,
      "step": 6814
    },
    {
      "epoch": 0.7887731481481481,
      "grad_norm": 98.81741333007812,
      "learning_rate": 4.224537037037038e-06,
      "loss": 2.679,
      "step": 6815
    },
    {
      "epoch": 0.7888888888888889,
      "grad_norm": 24.088943481445312,
      "learning_rate": 4.222222222222223e-06,
      "loss": 2.7449,
      "step": 6816
    },
    {
      "epoch": 0.7890046296296296,
      "grad_norm": 0.05892353504896164,
      "learning_rate": 4.219907407407408e-06,
      "loss": 0.0105,
      "step": 6817
    },
    {
      "epoch": 0.7891203703703704,
      "grad_norm": 0.036229152232408524,
      "learning_rate": 4.217592592592593e-06,
      "loss": 0.0065,
      "step": 6818
    },
    {
      "epoch": 0.7892361111111111,
      "grad_norm": 0.06018875166773796,
      "learning_rate": 4.215277777777778e-06,
      "loss": 0.0108,
      "step": 6819
    },
    {
      "epoch": 0.7893518518518519,
      "grad_norm": 0.04973419010639191,
      "learning_rate": 4.212962962962963e-06,
      "loss": 0.0084,
      "step": 6820
    },
    {
      "epoch": 0.7894675925925926,
      "grad_norm": 0.047770075500011444,
      "learning_rate": 4.210648148148148e-06,
      "loss": 0.0083,
      "step": 6821
    },
    {
      "epoch": 0.7895833333333333,
      "grad_norm": 0.0655183270573616,
      "learning_rate": 4.208333333333333e-06,
      "loss": 0.0095,
      "step": 6822
    },
    {
      "epoch": 0.789699074074074,
      "grad_norm": 0.04740624502301216,
      "learning_rate": 4.2060185185185185e-06,
      "loss": 0.0083,
      "step": 6823
    },
    {
      "epoch": 0.7898148148148149,
      "grad_norm": 0.056843582540750504,
      "learning_rate": 4.2037037037037045e-06,
      "loss": 0.0097,
      "step": 6824
    },
    {
      "epoch": 0.7899305555555556,
      "grad_norm": 0.4832848012447357,
      "learning_rate": 4.2013888888888896e-06,
      "loss": 0.0127,
      "step": 6825
    },
    {
      "epoch": 0.7900462962962963,
      "grad_norm": 0.04966834932565689,
      "learning_rate": 4.199074074074075e-06,
      "loss": 0.0084,
      "step": 6826
    },
    {
      "epoch": 0.790162037037037,
      "grad_norm": 163.84141540527344,
      "learning_rate": 4.19675925925926e-06,
      "loss": 0.6394,
      "step": 6827
    },
    {
      "epoch": 0.7902777777777777,
      "grad_norm": 1.867048978805542,
      "learning_rate": 4.194444444444445e-06,
      "loss": 0.0204,
      "step": 6828
    },
    {
      "epoch": 0.7903935185185185,
      "grad_norm": 0.0432654470205307,
      "learning_rate": 4.19212962962963e-06,
      "loss": 0.0079,
      "step": 6829
    },
    {
      "epoch": 0.7905092592592593,
      "grad_norm": 0.05327625945210457,
      "learning_rate": 4.189814814814815e-06,
      "loss": 0.0094,
      "step": 6830
    },
    {
      "epoch": 0.790625,
      "grad_norm": 0.0468490906059742,
      "learning_rate": 4.1875e-06,
      "loss": 0.0083,
      "step": 6831
    },
    {
      "epoch": 0.7907407407407407,
      "grad_norm": 0.06553605198860168,
      "learning_rate": 4.185185185185185e-06,
      "loss": 0.0118,
      "step": 6832
    },
    {
      "epoch": 0.7908564814814815,
      "grad_norm": 0.06615772098302841,
      "learning_rate": 4.18287037037037e-06,
      "loss": 0.0118,
      "step": 6833
    },
    {
      "epoch": 0.7909722222222222,
      "grad_norm": 38.054996490478516,
      "learning_rate": 4.180555555555556e-06,
      "loss": 2.7893,
      "step": 6834
    },
    {
      "epoch": 0.7910879629629629,
      "grad_norm": 0.036528002470731735,
      "learning_rate": 4.178240740740741e-06,
      "loss": 0.0066,
      "step": 6835
    },
    {
      "epoch": 0.7912037037037037,
      "grad_norm": 18.89752769470215,
      "learning_rate": 4.175925925925926e-06,
      "loss": 2.869,
      "step": 6836
    },
    {
      "epoch": 0.7913194444444445,
      "grad_norm": 0.0527498833835125,
      "learning_rate": 4.1736111111111116e-06,
      "loss": 0.0093,
      "step": 6837
    },
    {
      "epoch": 0.7914351851851852,
      "grad_norm": 0.055158961564302444,
      "learning_rate": 4.171296296296297e-06,
      "loss": 0.0098,
      "step": 6838
    },
    {
      "epoch": 0.7915509259259259,
      "grad_norm": 0.05109225958585739,
      "learning_rate": 4.168981481481482e-06,
      "loss": 0.0093,
      "step": 6839
    },
    {
      "epoch": 0.7916666666666666,
      "grad_norm": 0.03677545487880707,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.0067,
      "step": 6840
    },
    {
      "epoch": 0.7917824074074075,
      "grad_norm": 0.057421717792749405,
      "learning_rate": 4.164351851851852e-06,
      "loss": 0.0087,
      "step": 6841
    },
    {
      "epoch": 0.7918981481481482,
      "grad_norm": 0.20857033133506775,
      "learning_rate": 4.162037037037037e-06,
      "loss": 0.0109,
      "step": 6842
    },
    {
      "epoch": 0.7920138888888889,
      "grad_norm": 0.04638885334134102,
      "learning_rate": 4.159722222222223e-06,
      "loss": 0.0061,
      "step": 6843
    },
    {
      "epoch": 0.7921296296296296,
      "grad_norm": 0.049040861427783966,
      "learning_rate": 4.157407407407408e-06,
      "loss": 0.0088,
      "step": 6844
    },
    {
      "epoch": 0.7922453703703703,
      "grad_norm": 0.10388550907373428,
      "learning_rate": 4.155092592592593e-06,
      "loss": 0.0094,
      "step": 6845
    },
    {
      "epoch": 0.7923611111111111,
      "grad_norm": 0.06210751086473465,
      "learning_rate": 4.152777777777778e-06,
      "loss": 0.0114,
      "step": 6846
    },
    {
      "epoch": 0.7924768518518519,
      "grad_norm": 0.835191547870636,
      "learning_rate": 4.150462962962963e-06,
      "loss": 0.0148,
      "step": 6847
    },
    {
      "epoch": 0.7925925925925926,
      "grad_norm": 179.90029907226562,
      "learning_rate": 4.1481481481481485e-06,
      "loss": 1.0275,
      "step": 6848
    },
    {
      "epoch": 0.7927083333333333,
      "grad_norm": 1.6769107580184937,
      "learning_rate": 4.145833333333334e-06,
      "loss": 0.0174,
      "step": 6849
    },
    {
      "epoch": 0.7928240740740741,
      "grad_norm": 0.077069491147995,
      "learning_rate": 4.143518518518519e-06,
      "loss": 0.0129,
      "step": 6850
    },
    {
      "epoch": 0.7929398148148148,
      "grad_norm": 0.037284184247255325,
      "learning_rate": 4.141203703703704e-06,
      "loss": 0.0068,
      "step": 6851
    },
    {
      "epoch": 0.7930555555555555,
      "grad_norm": 0.05178701877593994,
      "learning_rate": 4.138888888888889e-06,
      "loss": 0.0094,
      "step": 6852
    },
    {
      "epoch": 0.7931712962962963,
      "grad_norm": 0.05722600966691971,
      "learning_rate": 4.136574074074075e-06,
      "loss": 0.0094,
      "step": 6853
    },
    {
      "epoch": 0.7932870370370371,
      "grad_norm": 0.0372057780623436,
      "learning_rate": 4.13425925925926e-06,
      "loss": 0.0068,
      "step": 6854
    },
    {
      "epoch": 0.7934027777777778,
      "grad_norm": 0.03376344218850136,
      "learning_rate": 4.131944444444444e-06,
      "loss": 0.0061,
      "step": 6855
    },
    {
      "epoch": 0.7935185185185185,
      "grad_norm": 0.05231059342622757,
      "learning_rate": 4.12962962962963e-06,
      "loss": 0.0095,
      "step": 6856
    },
    {
      "epoch": 0.7936342592592592,
      "grad_norm": 0.03500575199723244,
      "learning_rate": 4.127314814814815e-06,
      "loss": 0.0063,
      "step": 6857
    },
    {
      "epoch": 0.79375,
      "grad_norm": 0.043421339243650436,
      "learning_rate": 4.125e-06,
      "loss": 0.0079,
      "step": 6858
    },
    {
      "epoch": 0.7938657407407408,
      "grad_norm": 0.06665138155221939,
      "learning_rate": 4.122685185185185e-06,
      "loss": 0.0105,
      "step": 6859
    },
    {
      "epoch": 0.7939814814814815,
      "grad_norm": 2.965736150741577,
      "learning_rate": 4.1203703703703705e-06,
      "loss": 0.0212,
      "step": 6860
    },
    {
      "epoch": 0.7940972222222222,
      "grad_norm": 0.05374792963266373,
      "learning_rate": 4.118055555555556e-06,
      "loss": 0.0086,
      "step": 6861
    },
    {
      "epoch": 0.794212962962963,
      "grad_norm": 0.06594320386648178,
      "learning_rate": 4.1157407407407416e-06,
      "loss": 0.012,
      "step": 6862
    },
    {
      "epoch": 0.7943287037037037,
      "grad_norm": 0.06360147893428802,
      "learning_rate": 4.113425925925927e-06,
      "loss": 0.0112,
      "step": 6863
    },
    {
      "epoch": 0.7944444444444444,
      "grad_norm": 0.06981105357408524,
      "learning_rate": 4.111111111111111e-06,
      "loss": 0.0101,
      "step": 6864
    },
    {
      "epoch": 0.7945601851851852,
      "grad_norm": 0.0573875792324543,
      "learning_rate": 4.108796296296297e-06,
      "loss": 0.0099,
      "step": 6865
    },
    {
      "epoch": 0.794675925925926,
      "grad_norm": 0.05179809778928757,
      "learning_rate": 4.106481481481482e-06,
      "loss": 0.0094,
      "step": 6866
    },
    {
      "epoch": 0.7947916666666667,
      "grad_norm": 0.04768747463822365,
      "learning_rate": 4.104166666666667e-06,
      "loss": 0.0082,
      "step": 6867
    },
    {
      "epoch": 0.7949074074074074,
      "grad_norm": 34.15016174316406,
      "learning_rate": 4.101851851851852e-06,
      "loss": 2.2974,
      "step": 6868
    },
    {
      "epoch": 0.7950231481481481,
      "grad_norm": 0.08219138532876968,
      "learning_rate": 4.099537037037037e-06,
      "loss": 0.0116,
      "step": 6869
    },
    {
      "epoch": 0.7951388888888888,
      "grad_norm": 0.06094484031200409,
      "learning_rate": 4.097222222222222e-06,
      "loss": 0.0086,
      "step": 6870
    },
    {
      "epoch": 0.7952546296296297,
      "grad_norm": 0.054370373487472534,
      "learning_rate": 4.094907407407408e-06,
      "loss": 0.0099,
      "step": 6871
    },
    {
      "epoch": 0.7953703703703704,
      "grad_norm": 7.044753551483154,
      "learning_rate": 4.092592592592593e-06,
      "loss": 3.4746,
      "step": 6872
    },
    {
      "epoch": 0.7954861111111111,
      "grad_norm": 80.45809173583984,
      "learning_rate": 4.090277777777778e-06,
      "loss": 2.1181,
      "step": 6873
    },
    {
      "epoch": 0.7956018518518518,
      "grad_norm": 30.577999114990234,
      "learning_rate": 4.087962962962963e-06,
      "loss": 0.0992,
      "step": 6874
    },
    {
      "epoch": 0.7957175925925926,
      "grad_norm": 0.05985848233103752,
      "learning_rate": 4.085648148148149e-06,
      "loss": 0.0099,
      "step": 6875
    },
    {
      "epoch": 0.7958333333333333,
      "grad_norm": 0.051983512938022614,
      "learning_rate": 4.083333333333334e-06,
      "loss": 0.0094,
      "step": 6876
    },
    {
      "epoch": 0.7959490740740741,
      "grad_norm": 64.14295196533203,
      "learning_rate": 4.081018518518519e-06,
      "loss": 0.1689,
      "step": 6877
    },
    {
      "epoch": 0.7960648148148148,
      "grad_norm": 69.7774429321289,
      "learning_rate": 4.078703703703704e-06,
      "loss": 2.185,
      "step": 6878
    },
    {
      "epoch": 0.7961805555555556,
      "grad_norm": 0.04768982157111168,
      "learning_rate": 4.076388888888889e-06,
      "loss": 0.0085,
      "step": 6879
    },
    {
      "epoch": 0.7962962962962963,
      "grad_norm": 0.04607245698571205,
      "learning_rate": 4.074074074074074e-06,
      "loss": 0.0082,
      "step": 6880
    },
    {
      "epoch": 0.796412037037037,
      "grad_norm": 0.05873185396194458,
      "learning_rate": 4.07175925925926e-06,
      "loss": 0.0084,
      "step": 6881
    },
    {
      "epoch": 0.7965277777777777,
      "grad_norm": 0.047889649868011475,
      "learning_rate": 4.069444444444444e-06,
      "loss": 0.0062,
      "step": 6882
    },
    {
      "epoch": 0.7966435185185186,
      "grad_norm": 0.04857177287340164,
      "learning_rate": 4.0671296296296294e-06,
      "loss": 0.0085,
      "step": 6883
    },
    {
      "epoch": 0.7967592592592593,
      "grad_norm": 0.06011338531970978,
      "learning_rate": 4.064814814814815e-06,
      "loss": 0.0076,
      "step": 6884
    },
    {
      "epoch": 0.796875,
      "grad_norm": 0.037891581654548645,
      "learning_rate": 4.0625000000000005e-06,
      "loss": 0.0068,
      "step": 6885
    },
    {
      "epoch": 0.7969907407407407,
      "grad_norm": 0.043566666543483734,
      "learning_rate": 4.060185185185186e-06,
      "loss": 0.0076,
      "step": 6886
    },
    {
      "epoch": 0.7971064814814814,
      "grad_norm": 0.05249235779047012,
      "learning_rate": 4.057870370370371e-06,
      "loss": 0.0094,
      "step": 6887
    },
    {
      "epoch": 0.7972222222222223,
      "grad_norm": 0.06355348974466324,
      "learning_rate": 4.055555555555556e-06,
      "loss": 0.0075,
      "step": 6888
    },
    {
      "epoch": 0.797337962962963,
      "grad_norm": 0.04561937600374222,
      "learning_rate": 4.053240740740741e-06,
      "loss": 0.0082,
      "step": 6889
    },
    {
      "epoch": 0.7974537037037037,
      "grad_norm": 0.04316914081573486,
      "learning_rate": 4.050925925925927e-06,
      "loss": 0.0079,
      "step": 6890
    },
    {
      "epoch": 0.7975694444444444,
      "grad_norm": 0.05888764187693596,
      "learning_rate": 4.048611111111111e-06,
      "loss": 0.0097,
      "step": 6891
    },
    {
      "epoch": 0.7976851851851852,
      "grad_norm": 0.049117784947156906,
      "learning_rate": 4.046296296296296e-06,
      "loss": 0.0073,
      "step": 6892
    },
    {
      "epoch": 0.7978009259259259,
      "grad_norm": 6.721508026123047,
      "learning_rate": 4.043981481481482e-06,
      "loss": 3.2787,
      "step": 6893
    },
    {
      "epoch": 0.7979166666666667,
      "grad_norm": 0.23641034960746765,
      "learning_rate": 4.041666666666667e-06,
      "loss": 0.0098,
      "step": 6894
    },
    {
      "epoch": 0.7980324074074074,
      "grad_norm": 0.044972363859415054,
      "learning_rate": 4.039351851851852e-06,
      "loss": 0.0059,
      "step": 6895
    },
    {
      "epoch": 0.7981481481481482,
      "grad_norm": 0.04685090482234955,
      "learning_rate": 4.037037037037037e-06,
      "loss": 0.0082,
      "step": 6896
    },
    {
      "epoch": 0.7982638888888889,
      "grad_norm": 0.059556994587183,
      "learning_rate": 4.0347222222222225e-06,
      "loss": 0.0105,
      "step": 6897
    },
    {
      "epoch": 0.7983796296296296,
      "grad_norm": 0.0398242324590683,
      "learning_rate": 4.032407407407408e-06,
      "loss": 0.0072,
      "step": 6898
    },
    {
      "epoch": 0.7984953703703703,
      "grad_norm": 0.03679494559764862,
      "learning_rate": 4.030092592592593e-06,
      "loss": 0.0067,
      "step": 6899
    },
    {
      "epoch": 0.7986111111111112,
      "grad_norm": 0.05556401610374451,
      "learning_rate": 4.027777777777779e-06,
      "loss": 0.0073,
      "step": 6900
    },
    {
      "epoch": 0.7987268518518519,
      "grad_norm": 0.06358850747346878,
      "learning_rate": 4.025462962962963e-06,
      "loss": 0.0108,
      "step": 6901
    },
    {
      "epoch": 0.7988425925925926,
      "grad_norm": 0.03706667572259903,
      "learning_rate": 4.023148148148148e-06,
      "loss": 0.0063,
      "step": 6902
    },
    {
      "epoch": 0.7989583333333333,
      "grad_norm": 0.045154497027397156,
      "learning_rate": 4.020833333333334e-06,
      "loss": 0.008,
      "step": 6903
    },
    {
      "epoch": 0.799074074074074,
      "grad_norm": 0.053255096077919006,
      "learning_rate": 4.018518518518519e-06,
      "loss": 0.0092,
      "step": 6904
    },
    {
      "epoch": 0.7991898148148148,
      "grad_norm": 0.059378985315561295,
      "learning_rate": 4.016203703703704e-06,
      "loss": 0.0107,
      "step": 6905
    },
    {
      "epoch": 0.7993055555555556,
      "grad_norm": 0.0435805581510067,
      "learning_rate": 4.013888888888889e-06,
      "loss": 0.0079,
      "step": 6906
    },
    {
      "epoch": 0.7994212962962963,
      "grad_norm": 0.05479767173528671,
      "learning_rate": 4.011574074074074e-06,
      "loss": 0.0072,
      "step": 6907
    },
    {
      "epoch": 0.799537037037037,
      "grad_norm": 0.13794194161891937,
      "learning_rate": 4.0092592592592594e-06,
      "loss": 0.0093,
      "step": 6908
    },
    {
      "epoch": 0.7996527777777778,
      "grad_norm": 0.08088470995426178,
      "learning_rate": 4.006944444444445e-06,
      "loss": 0.0105,
      "step": 6909
    },
    {
      "epoch": 0.7997685185185185,
      "grad_norm": 0.06626799702644348,
      "learning_rate": 4.00462962962963e-06,
      "loss": 0.0095,
      "step": 6910
    },
    {
      "epoch": 0.7998842592592592,
      "grad_norm": 0.039862196892499924,
      "learning_rate": 4.002314814814815e-06,
      "loss": 0.0072,
      "step": 6911
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.05454970523715019,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0094,
      "step": 6912
    },
    {
      "epoch": 0.8001157407407408,
      "grad_norm": 0.055030129849910736,
      "learning_rate": 3.997685185185186e-06,
      "loss": 0.0094,
      "step": 6913
    },
    {
      "epoch": 0.8002314814814815,
      "grad_norm": 0.06116222217679024,
      "learning_rate": 3.995370370370371e-06,
      "loss": 0.0076,
      "step": 6914
    },
    {
      "epoch": 0.8003472222222222,
      "grad_norm": 0.40539300441741943,
      "learning_rate": 3.993055555555556e-06,
      "loss": 0.0094,
      "step": 6915
    },
    {
      "epoch": 0.8004629629629629,
      "grad_norm": 0.06738967448472977,
      "learning_rate": 3.990740740740741e-06,
      "loss": 0.0068,
      "step": 6916
    },
    {
      "epoch": 0.8005787037037037,
      "grad_norm": 0.040968626737594604,
      "learning_rate": 3.988425925925926e-06,
      "loss": 0.0073,
      "step": 6917
    },
    {
      "epoch": 0.8006944444444445,
      "grad_norm": 0.058358073234558105,
      "learning_rate": 3.986111111111112e-06,
      "loss": 0.0075,
      "step": 6918
    },
    {
      "epoch": 0.8008101851851852,
      "grad_norm": 0.08820252865552902,
      "learning_rate": 3.983796296296296e-06,
      "loss": 0.0096,
      "step": 6919
    },
    {
      "epoch": 0.8009259259259259,
      "grad_norm": 0.053278952836990356,
      "learning_rate": 3.9814814814814814e-06,
      "loss": 0.009,
      "step": 6920
    },
    {
      "epoch": 0.8010416666666667,
      "grad_norm": 0.056279659271240234,
      "learning_rate": 3.9791666666666665e-06,
      "loss": 0.0103,
      "step": 6921
    },
    {
      "epoch": 0.8011574074074074,
      "grad_norm": 0.07902953028678894,
      "learning_rate": 3.9768518518518525e-06,
      "loss": 0.0112,
      "step": 6922
    },
    {
      "epoch": 0.8012731481481481,
      "grad_norm": 53.924007415771484,
      "learning_rate": 3.974537037037038e-06,
      "loss": 2.8772,
      "step": 6923
    },
    {
      "epoch": 0.8013888888888889,
      "grad_norm": 1.001715064048767,
      "learning_rate": 3.972222222222223e-06,
      "loss": 0.0133,
      "step": 6924
    },
    {
      "epoch": 0.8015046296296297,
      "grad_norm": 57.3524169921875,
      "learning_rate": 3.969907407407408e-06,
      "loss": 2.5001,
      "step": 6925
    },
    {
      "epoch": 0.8016203703703704,
      "grad_norm": 0.03883260115981102,
      "learning_rate": 3.967592592592593e-06,
      "loss": 0.007,
      "step": 6926
    },
    {
      "epoch": 0.8017361111111111,
      "grad_norm": 0.1927405595779419,
      "learning_rate": 3.965277777777778e-06,
      "loss": 0.0106,
      "step": 6927
    },
    {
      "epoch": 0.8018518518518518,
      "grad_norm": 0.05343613773584366,
      "learning_rate": 3.962962962962963e-06,
      "loss": 0.0088,
      "step": 6928
    },
    {
      "epoch": 0.8019675925925925,
      "grad_norm": 0.05138833820819855,
      "learning_rate": 3.960648148148148e-06,
      "loss": 0.0088,
      "step": 6929
    },
    {
      "epoch": 0.8020833333333334,
      "grad_norm": 0.03809218481183052,
      "learning_rate": 3.958333333333333e-06,
      "loss": 0.0068,
      "step": 6930
    },
    {
      "epoch": 0.8021990740740741,
      "grad_norm": 0.05449387803673744,
      "learning_rate": 3.956018518518519e-06,
      "loss": 0.0071,
      "step": 6931
    },
    {
      "epoch": 0.8023148148148148,
      "grad_norm": 0.048420850187540054,
      "learning_rate": 3.953703703703704e-06,
      "loss": 0.0088,
      "step": 6932
    },
    {
      "epoch": 0.8024305555555555,
      "grad_norm": 0.05515475943684578,
      "learning_rate": 3.951388888888889e-06,
      "loss": 0.0096,
      "step": 6933
    },
    {
      "epoch": 0.8025462962962963,
      "grad_norm": 0.056670427322387695,
      "learning_rate": 3.9490740740740745e-06,
      "loss": 0.0094,
      "step": 6934
    },
    {
      "epoch": 0.8026620370370371,
      "grad_norm": 0.05489509552717209,
      "learning_rate": 3.94675925925926e-06,
      "loss": 0.0072,
      "step": 6935
    },
    {
      "epoch": 0.8027777777777778,
      "grad_norm": 1.9575676918029785,
      "learning_rate": 3.944444444444445e-06,
      "loss": 0.0142,
      "step": 6936
    },
    {
      "epoch": 0.8028935185185185,
      "grad_norm": 0.05362575128674507,
      "learning_rate": 3.94212962962963e-06,
      "loss": 0.0093,
      "step": 6937
    },
    {
      "epoch": 0.8030092592592593,
      "grad_norm": 0.058449964970350266,
      "learning_rate": 3.939814814814815e-06,
      "loss": 0.0098,
      "step": 6938
    },
    {
      "epoch": 0.803125,
      "grad_norm": 0.08847574889659882,
      "learning_rate": 3.9375e-06,
      "loss": 0.0091,
      "step": 6939
    },
    {
      "epoch": 0.8032407407407407,
      "grad_norm": 0.05523925647139549,
      "learning_rate": 3.935185185185186e-06,
      "loss": 0.0097,
      "step": 6940
    },
    {
      "epoch": 0.8033564814814815,
      "grad_norm": 0.037902433425188065,
      "learning_rate": 3.932870370370371e-06,
      "loss": 0.0069,
      "step": 6941
    },
    {
      "epoch": 0.8034722222222223,
      "grad_norm": 0.04758899658918381,
      "learning_rate": 3.930555555555556e-06,
      "loss": 0.0083,
      "step": 6942
    },
    {
      "epoch": 0.803587962962963,
      "grad_norm": 1.2220180034637451,
      "learning_rate": 3.92824074074074e-06,
      "loss": 0.0228,
      "step": 6943
    },
    {
      "epoch": 0.8037037037037037,
      "grad_norm": 0.05103102698922157,
      "learning_rate": 3.925925925925926e-06,
      "loss": 0.0066,
      "step": 6944
    },
    {
      "epoch": 0.8038194444444444,
      "grad_norm": 0.0508778840303421,
      "learning_rate": 3.9236111111111114e-06,
      "loss": 0.0089,
      "step": 6945
    },
    {
      "epoch": 0.8039351851851851,
      "grad_norm": 0.03753663972020149,
      "learning_rate": 3.9212962962962965e-06,
      "loss": 0.0067,
      "step": 6946
    },
    {
      "epoch": 0.804050925925926,
      "grad_norm": 0.03885456919670105,
      "learning_rate": 3.918981481481482e-06,
      "loss": 0.0069,
      "step": 6947
    },
    {
      "epoch": 0.8041666666666667,
      "grad_norm": 0.05486895889043808,
      "learning_rate": 3.916666666666667e-06,
      "loss": 0.0088,
      "step": 6948
    },
    {
      "epoch": 0.8042824074074074,
      "grad_norm": 0.08799810707569122,
      "learning_rate": 3.914351851851852e-06,
      "loss": 0.0114,
      "step": 6949
    },
    {
      "epoch": 0.8043981481481481,
      "grad_norm": 0.03580256924033165,
      "learning_rate": 3.912037037037038e-06,
      "loss": 0.0062,
      "step": 6950
    },
    {
      "epoch": 0.8045138888888889,
      "grad_norm": 0.08800531923770905,
      "learning_rate": 3.909722222222223e-06,
      "loss": 0.0096,
      "step": 6951
    },
    {
      "epoch": 0.8046296296296296,
      "grad_norm": 0.05721605196595192,
      "learning_rate": 3.907407407407408e-06,
      "loss": 0.0095,
      "step": 6952
    },
    {
      "epoch": 0.8047453703703704,
      "grad_norm": 0.1366923302412033,
      "learning_rate": 3.905092592592593e-06,
      "loss": 0.0119,
      "step": 6953
    },
    {
      "epoch": 0.8048611111111111,
      "grad_norm": 0.03318089246749878,
      "learning_rate": 3.902777777777778e-06,
      "loss": 0.006,
      "step": 6954
    },
    {
      "epoch": 0.8049768518518519,
      "grad_norm": 0.03457286208868027,
      "learning_rate": 3.900462962962963e-06,
      "loss": 0.0063,
      "step": 6955
    },
    {
      "epoch": 0.8050925925925926,
      "grad_norm": 0.060995738953351974,
      "learning_rate": 3.898148148148148e-06,
      "loss": 0.01,
      "step": 6956
    },
    {
      "epoch": 0.8052083333333333,
      "grad_norm": 0.03920087590813637,
      "learning_rate": 3.8958333333333334e-06,
      "loss": 0.0069,
      "step": 6957
    },
    {
      "epoch": 0.805324074074074,
      "grad_norm": 0.03664299100637436,
      "learning_rate": 3.8935185185185185e-06,
      "loss": 0.0066,
      "step": 6958
    },
    {
      "epoch": 0.8054398148148149,
      "grad_norm": 0.059236057102680206,
      "learning_rate": 3.8912037037037045e-06,
      "loss": 0.01,
      "step": 6959
    },
    {
      "epoch": 0.8055555555555556,
      "grad_norm": 16.91118049621582,
      "learning_rate": 3.88888888888889e-06,
      "loss": 3.1938,
      "step": 6960
    },
    {
      "epoch": 0.8056712962962963,
      "grad_norm": 0.03969230875372887,
      "learning_rate": 3.886574074074075e-06,
      "loss": 0.0068,
      "step": 6961
    },
    {
      "epoch": 0.805787037037037,
      "grad_norm": 0.05937999486923218,
      "learning_rate": 3.88425925925926e-06,
      "loss": 0.0106,
      "step": 6962
    },
    {
      "epoch": 0.8059027777777777,
      "grad_norm": 0.5101921558380127,
      "learning_rate": 3.881944444444445e-06,
      "loss": 0.0107,
      "step": 6963
    },
    {
      "epoch": 0.8060185185185185,
      "grad_norm": 0.0473773330450058,
      "learning_rate": 3.87962962962963e-06,
      "loss": 0.0082,
      "step": 6964
    },
    {
      "epoch": 0.8061342592592593,
      "grad_norm": 10.089170455932617,
      "learning_rate": 3.877314814814815e-06,
      "loss": 3.0969,
      "step": 6965
    },
    {
      "epoch": 0.80625,
      "grad_norm": 0.07161295413970947,
      "learning_rate": 3.875e-06,
      "loss": 0.0088,
      "step": 6966
    },
    {
      "epoch": 0.8063657407407407,
      "grad_norm": 0.2038513571023941,
      "learning_rate": 3.872685185185185e-06,
      "loss": 0.0077,
      "step": 6967
    },
    {
      "epoch": 0.8064814814814815,
      "grad_norm": 0.05250038579106331,
      "learning_rate": 3.87037037037037e-06,
      "loss": 0.0094,
      "step": 6968
    },
    {
      "epoch": 0.8065972222222222,
      "grad_norm": 0.04518391937017441,
      "learning_rate": 3.868055555555556e-06,
      "loss": 0.008,
      "step": 6969
    },
    {
      "epoch": 0.8067129629629629,
      "grad_norm": 0.034521445631980896,
      "learning_rate": 3.865740740740741e-06,
      "loss": 0.0063,
      "step": 6970
    },
    {
      "epoch": 0.8068287037037037,
      "grad_norm": 0.049612611532211304,
      "learning_rate": 3.863425925925926e-06,
      "loss": 0.0085,
      "step": 6971
    },
    {
      "epoch": 0.8069444444444445,
      "grad_norm": 0.04911438375711441,
      "learning_rate": 3.861111111111112e-06,
      "loss": 0.0088,
      "step": 6972
    },
    {
      "epoch": 0.8070601851851852,
      "grad_norm": 26.50895881652832,
      "learning_rate": 3.858796296296297e-06,
      "loss": 2.6667,
      "step": 6973
    },
    {
      "epoch": 0.8071759259259259,
      "grad_norm": 0.03512297943234444,
      "learning_rate": 3.856481481481482e-06,
      "loss": 0.0062,
      "step": 6974
    },
    {
      "epoch": 0.8072916666666666,
      "grad_norm": 0.033924125134944916,
      "learning_rate": 3.854166666666667e-06,
      "loss": 0.0062,
      "step": 6975
    },
    {
      "epoch": 0.8074074074074075,
      "grad_norm": 0.05104244872927666,
      "learning_rate": 3.851851851851852e-06,
      "loss": 0.0087,
      "step": 6976
    },
    {
      "epoch": 0.8075231481481482,
      "grad_norm": 0.0558825246989727,
      "learning_rate": 3.849537037037037e-06,
      "loss": 0.0092,
      "step": 6977
    },
    {
      "epoch": 0.8076388888888889,
      "grad_norm": 0.038224976509809494,
      "learning_rate": 3.847222222222223e-06,
      "loss": 0.0067,
      "step": 6978
    },
    {
      "epoch": 0.8077546296296296,
      "grad_norm": 0.08530004322528839,
      "learning_rate": 3.844907407407408e-06,
      "loss": 0.0111,
      "step": 6979
    },
    {
      "epoch": 0.8078703703703703,
      "grad_norm": 0.034789904952049255,
      "learning_rate": 3.842592592592592e-06,
      "loss": 0.0062,
      "step": 6980
    },
    {
      "epoch": 0.8079861111111111,
      "grad_norm": 94.91361999511719,
      "learning_rate": 3.840277777777778e-06,
      "loss": 0.8925,
      "step": 6981
    },
    {
      "epoch": 0.8081018518518519,
      "grad_norm": 0.049631718546152115,
      "learning_rate": 3.8379629629629634e-06,
      "loss": 0.0088,
      "step": 6982
    },
    {
      "epoch": 0.8082175925925926,
      "grad_norm": 0.07698801904916763,
      "learning_rate": 3.8356481481481485e-06,
      "loss": 0.0101,
      "step": 6983
    },
    {
      "epoch": 0.8083333333333333,
      "grad_norm": 0.044383712112903595,
      "learning_rate": 3.833333333333334e-06,
      "loss": 0.0058,
      "step": 6984
    },
    {
      "epoch": 0.8084490740740741,
      "grad_norm": 0.059206001460552216,
      "learning_rate": 3.831018518518519e-06,
      "loss": 0.009,
      "step": 6985
    },
    {
      "epoch": 0.8085648148148148,
      "grad_norm": 0.10385900735855103,
      "learning_rate": 3.828703703703704e-06,
      "loss": 0.0117,
      "step": 6986
    },
    {
      "epoch": 0.8086805555555555,
      "grad_norm": 0.04069792106747627,
      "learning_rate": 3.826388888888889e-06,
      "loss": 0.0071,
      "step": 6987
    },
    {
      "epoch": 0.8087962962962963,
      "grad_norm": 7.91478157043457,
      "learning_rate": 3.824074074074075e-06,
      "loss": 3.5306,
      "step": 6988
    },
    {
      "epoch": 0.8089120370370371,
      "grad_norm": 0.06268991529941559,
      "learning_rate": 3.821759259259259e-06,
      "loss": 0.0114,
      "step": 6989
    },
    {
      "epoch": 0.8090277777777778,
      "grad_norm": 0.05580223724246025,
      "learning_rate": 3.819444444444444e-06,
      "loss": 0.0099,
      "step": 6990
    },
    {
      "epoch": 0.8091435185185185,
      "grad_norm": 0.03968751057982445,
      "learning_rate": 3.81712962962963e-06,
      "loss": 0.007,
      "step": 6991
    },
    {
      "epoch": 0.8092592592592592,
      "grad_norm": 0.05391889065504074,
      "learning_rate": 3.814814814814815e-06,
      "loss": 0.0097,
      "step": 6992
    },
    {
      "epoch": 0.809375,
      "grad_norm": 0.04015861451625824,
      "learning_rate": 3.8125e-06,
      "loss": 0.007,
      "step": 6993
    },
    {
      "epoch": 0.8094907407407408,
      "grad_norm": 0.04899503290653229,
      "learning_rate": 3.810185185185186e-06,
      "loss": 0.0083,
      "step": 6994
    },
    {
      "epoch": 0.8096064814814815,
      "grad_norm": 0.04738559201359749,
      "learning_rate": 3.8078703703703705e-06,
      "loss": 0.0084,
      "step": 6995
    },
    {
      "epoch": 0.8097222222222222,
      "grad_norm": 0.14608609676361084,
      "learning_rate": 3.8055555555555556e-06,
      "loss": 0.0119,
      "step": 6996
    },
    {
      "epoch": 0.809837962962963,
      "grad_norm": 0.0557195283472538,
      "learning_rate": 3.803240740740741e-06,
      "loss": 0.0097,
      "step": 6997
    },
    {
      "epoch": 0.8099537037037037,
      "grad_norm": 0.05579832196235657,
      "learning_rate": 3.8009259259259263e-06,
      "loss": 0.008,
      "step": 6998
    },
    {
      "epoch": 0.8100694444444444,
      "grad_norm": 0.04889402538537979,
      "learning_rate": 3.7986111111111114e-06,
      "loss": 0.0086,
      "step": 6999
    },
    {
      "epoch": 0.8101851851851852,
      "grad_norm": 0.03614198416471481,
      "learning_rate": 3.796296296296297e-06,
      "loss": 0.0066,
      "step": 7000
    },
    {
      "epoch": 0.810300925925926,
      "grad_norm": 0.05355176702141762,
      "learning_rate": 3.793981481481482e-06,
      "loss": 0.0093,
      "step": 7001
    },
    {
      "epoch": 0.8104166666666667,
      "grad_norm": 0.03892550617456436,
      "learning_rate": 3.7916666666666666e-06,
      "loss": 0.0068,
      "step": 7002
    },
    {
      "epoch": 0.8105324074074074,
      "grad_norm": 0.06181192025542259,
      "learning_rate": 3.7893518518518526e-06,
      "loss": 0.01,
      "step": 7003
    },
    {
      "epoch": 0.8106481481481481,
      "grad_norm": 0.045954667031764984,
      "learning_rate": 3.7870370370370373e-06,
      "loss": 0.006,
      "step": 7004
    },
    {
      "epoch": 0.8107638888888888,
      "grad_norm": 0.06468506157398224,
      "learning_rate": 3.7847222222222224e-06,
      "loss": 0.0118,
      "step": 7005
    },
    {
      "epoch": 0.8108796296296297,
      "grad_norm": 0.09237305819988251,
      "learning_rate": 3.782407407407408e-06,
      "loss": 0.0117,
      "step": 7006
    },
    {
      "epoch": 0.8109953703703704,
      "grad_norm": 0.05367697775363922,
      "learning_rate": 3.780092592592593e-06,
      "loss": 0.0093,
      "step": 7007
    },
    {
      "epoch": 0.8111111111111111,
      "grad_norm": 0.06479023396968842,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.0096,
      "step": 7008
    },
    {
      "epoch": 0.8112268518518518,
      "grad_norm": 0.051983416080474854,
      "learning_rate": 3.775462962962963e-06,
      "loss": 0.0093,
      "step": 7009
    },
    {
      "epoch": 0.8113425925925926,
      "grad_norm": 0.049939315766096115,
      "learning_rate": 3.7731481481481487e-06,
      "loss": 0.0084,
      "step": 7010
    },
    {
      "epoch": 0.8114583333333333,
      "grad_norm": 0.04077500104904175,
      "learning_rate": 3.7708333333333334e-06,
      "loss": 0.0073,
      "step": 7011
    },
    {
      "epoch": 0.8115740740740741,
      "grad_norm": 0.08485516905784607,
      "learning_rate": 3.7685185185185185e-06,
      "loss": 0.0111,
      "step": 7012
    },
    {
      "epoch": 0.8116898148148148,
      "grad_norm": 0.05065424367785454,
      "learning_rate": 3.766203703703704e-06,
      "loss": 0.0086,
      "step": 7013
    },
    {
      "epoch": 0.8118055555555556,
      "grad_norm": 0.10042306780815125,
      "learning_rate": 3.763888888888889e-06,
      "loss": 0.013,
      "step": 7014
    },
    {
      "epoch": 0.8119212962962963,
      "grad_norm": 9.035398483276367,
      "learning_rate": 3.761574074074074e-06,
      "loss": 0.0517,
      "step": 7015
    },
    {
      "epoch": 0.812037037037037,
      "grad_norm": 0.049989111721515656,
      "learning_rate": 3.7592592592592597e-06,
      "loss": 0.0065,
      "step": 7016
    },
    {
      "epoch": 0.8121527777777777,
      "grad_norm": 0.053234729915857315,
      "learning_rate": 3.756944444444445e-06,
      "loss": 0.0095,
      "step": 7017
    },
    {
      "epoch": 0.8122685185185186,
      "grad_norm": 0.045422498136758804,
      "learning_rate": 3.75462962962963e-06,
      "loss": 0.0081,
      "step": 7018
    },
    {
      "epoch": 0.8123842592592593,
      "grad_norm": 0.12969404458999634,
      "learning_rate": 3.7523148148148154e-06,
      "loss": 0.013,
      "step": 7019
    },
    {
      "epoch": 0.8125,
      "grad_norm": 50.99412536621094,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 2.0571,
      "step": 7020
    },
    {
      "epoch": 0.8126157407407407,
      "grad_norm": 0.038620319217443466,
      "learning_rate": 3.747685185185185e-06,
      "loss": 0.0068,
      "step": 7021
    },
    {
      "epoch": 0.8127314814814814,
      "grad_norm": 45.47556686401367,
      "learning_rate": 3.7453703703703707e-06,
      "loss": 0.1206,
      "step": 7022
    },
    {
      "epoch": 0.8128472222222223,
      "grad_norm": 0.04575531184673309,
      "learning_rate": 3.743055555555556e-06,
      "loss": 0.0071,
      "step": 7023
    },
    {
      "epoch": 0.812962962962963,
      "grad_norm": 0.0484439954161644,
      "learning_rate": 3.740740740740741e-06,
      "loss": 0.0087,
      "step": 7024
    },
    {
      "epoch": 0.8130787037037037,
      "grad_norm": 0.04800550267100334,
      "learning_rate": 3.7384259259259264e-06,
      "loss": 0.0086,
      "step": 7025
    },
    {
      "epoch": 0.8131944444444444,
      "grad_norm": 0.03623851388692856,
      "learning_rate": 3.7361111111111115e-06,
      "loss": 0.0066,
      "step": 7026
    },
    {
      "epoch": 0.8133101851851852,
      "grad_norm": 0.04187537729740143,
      "learning_rate": 3.7337962962962966e-06,
      "loss": 0.0075,
      "step": 7027
    },
    {
      "epoch": 0.8134259259259259,
      "grad_norm": 0.10954020917415619,
      "learning_rate": 3.731481481481482e-06,
      "loss": 0.0085,
      "step": 7028
    },
    {
      "epoch": 0.8135416666666667,
      "grad_norm": 0.05842931941151619,
      "learning_rate": 3.7291666666666672e-06,
      "loss": 0.0093,
      "step": 7029
    },
    {
      "epoch": 0.8136574074074074,
      "grad_norm": 0.04865182936191559,
      "learning_rate": 3.726851851851852e-06,
      "loss": 0.0084,
      "step": 7030
    },
    {
      "epoch": 0.8137731481481482,
      "grad_norm": 0.041400834918022156,
      "learning_rate": 3.7245370370370374e-06,
      "loss": 0.0075,
      "step": 7031
    },
    {
      "epoch": 0.8138888888888889,
      "grad_norm": 0.07013605535030365,
      "learning_rate": 3.7222222222222225e-06,
      "loss": 0.009,
      "step": 7032
    },
    {
      "epoch": 0.8140046296296296,
      "grad_norm": 0.055296726524829865,
      "learning_rate": 3.7199074074074076e-06,
      "loss": 0.0072,
      "step": 7033
    },
    {
      "epoch": 0.8141203703703703,
      "grad_norm": 0.046434659510850906,
      "learning_rate": 3.7175925925925927e-06,
      "loss": 0.0076,
      "step": 7034
    },
    {
      "epoch": 0.8142361111111112,
      "grad_norm": 0.06701329350471497,
      "learning_rate": 3.7152777777777783e-06,
      "loss": 0.0109,
      "step": 7035
    },
    {
      "epoch": 0.8143518518518519,
      "grad_norm": 0.03651246801018715,
      "learning_rate": 3.7129629629629633e-06,
      "loss": 0.0066,
      "step": 7036
    },
    {
      "epoch": 0.8144675925925926,
      "grad_norm": 0.050190918147563934,
      "learning_rate": 3.710648148148148e-06,
      "loss": 0.0089,
      "step": 7037
    },
    {
      "epoch": 0.8145833333333333,
      "grad_norm": 0.03623035177588463,
      "learning_rate": 3.708333333333334e-06,
      "loss": 0.0066,
      "step": 7038
    },
    {
      "epoch": 0.814699074074074,
      "grad_norm": 0.048574842512607574,
      "learning_rate": 3.7060185185185186e-06,
      "loss": 0.0083,
      "step": 7039
    },
    {
      "epoch": 0.8148148148148148,
      "grad_norm": 0.05690387636423111,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 0.0104,
      "step": 7040
    },
    {
      "epoch": 0.8149305555555556,
      "grad_norm": 0.03883074223995209,
      "learning_rate": 3.7013888888888893e-06,
      "loss": 0.007,
      "step": 7041
    },
    {
      "epoch": 0.8150462962962963,
      "grad_norm": 0.04712125286459923,
      "learning_rate": 3.6990740740740744e-06,
      "loss": 0.0061,
      "step": 7042
    },
    {
      "epoch": 0.815162037037037,
      "grad_norm": 0.04620899632573128,
      "learning_rate": 3.6967592592592595e-06,
      "loss": 0.008,
      "step": 7043
    },
    {
      "epoch": 0.8152777777777778,
      "grad_norm": 0.045561667531728745,
      "learning_rate": 3.694444444444445e-06,
      "loss": 0.008,
      "step": 7044
    },
    {
      "epoch": 0.8153935185185185,
      "grad_norm": 0.03818809613585472,
      "learning_rate": 3.69212962962963e-06,
      "loss": 0.0069,
      "step": 7045
    },
    {
      "epoch": 0.8155092592592592,
      "grad_norm": 0.036328334361314774,
      "learning_rate": 3.6898148148148147e-06,
      "loss": 0.0066,
      "step": 7046
    },
    {
      "epoch": 0.815625,
      "grad_norm": 0.03353886306285858,
      "learning_rate": 3.6875000000000007e-06,
      "loss": 0.0061,
      "step": 7047
    },
    {
      "epoch": 0.8157407407407408,
      "grad_norm": 0.03960733488202095,
      "learning_rate": 3.6851851851851854e-06,
      "loss": 0.007,
      "step": 7048
    },
    {
      "epoch": 0.8158564814814815,
      "grad_norm": 0.047996990382671356,
      "learning_rate": 3.6828703703703705e-06,
      "loss": 0.0083,
      "step": 7049
    },
    {
      "epoch": 0.8159722222222222,
      "grad_norm": 0.04676694795489311,
      "learning_rate": 3.680555555555556e-06,
      "loss": 0.006,
      "step": 7050
    },
    {
      "epoch": 0.8160879629629629,
      "grad_norm": 0.060631655156612396,
      "learning_rate": 3.678240740740741e-06,
      "loss": 0.0111,
      "step": 7051
    },
    {
      "epoch": 0.8162037037037037,
      "grad_norm": 0.5087856650352478,
      "learning_rate": 3.675925925925926e-06,
      "loss": 0.0137,
      "step": 7052
    },
    {
      "epoch": 0.8163194444444445,
      "grad_norm": 0.05246905982494354,
      "learning_rate": 3.6736111111111117e-06,
      "loss": 0.0094,
      "step": 7053
    },
    {
      "epoch": 0.8164351851851852,
      "grad_norm": 0.9947402477264404,
      "learning_rate": 3.671296296296297e-06,
      "loss": 0.0183,
      "step": 7054
    },
    {
      "epoch": 0.8165509259259259,
      "grad_norm": 0.037047576159238815,
      "learning_rate": 3.668981481481482e-06,
      "loss": 0.0064,
      "step": 7055
    },
    {
      "epoch": 0.8166666666666667,
      "grad_norm": 0.05799795687198639,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.0075,
      "step": 7056
    },
    {
      "epoch": 0.8167824074074074,
      "grad_norm": 7.895653247833252,
      "learning_rate": 3.664351851851852e-06,
      "loss": 3.055,
      "step": 7057
    },
    {
      "epoch": 0.8168981481481481,
      "grad_norm": 0.04430665820837021,
      "learning_rate": 3.662037037037037e-06,
      "loss": 0.0079,
      "step": 7058
    },
    {
      "epoch": 0.8170138888888889,
      "grad_norm": 0.03749814257025719,
      "learning_rate": 3.6597222222222223e-06,
      "loss": 0.0067,
      "step": 7059
    },
    {
      "epoch": 0.8171296296296297,
      "grad_norm": 0.04986238107085228,
      "learning_rate": 3.657407407407408e-06,
      "loss": 0.0085,
      "step": 7060
    },
    {
      "epoch": 0.8172453703703704,
      "grad_norm": 0.05859857425093651,
      "learning_rate": 3.655092592592593e-06,
      "loss": 0.0092,
      "step": 7061
    },
    {
      "epoch": 0.8173611111111111,
      "grad_norm": 0.09391939640045166,
      "learning_rate": 3.652777777777778e-06,
      "loss": 0.0125,
      "step": 7062
    },
    {
      "epoch": 0.8174768518518518,
      "grad_norm": 0.04751455411314964,
      "learning_rate": 3.6504629629629635e-06,
      "loss": 0.0086,
      "step": 7063
    },
    {
      "epoch": 0.8175925925925925,
      "grad_norm": 0.03679288551211357,
      "learning_rate": 3.6481481481481486e-06,
      "loss": 0.0066,
      "step": 7064
    },
    {
      "epoch": 0.8177083333333334,
      "grad_norm": 0.05864532291889191,
      "learning_rate": 3.6458333333333333e-06,
      "loss": 0.0089,
      "step": 7065
    },
    {
      "epoch": 0.8178240740740741,
      "grad_norm": 0.05122259259223938,
      "learning_rate": 3.643518518518519e-06,
      "loss": 0.0088,
      "step": 7066
    },
    {
      "epoch": 0.8179398148148148,
      "grad_norm": 0.03570951521396637,
      "learning_rate": 3.641203703703704e-06,
      "loss": 0.0065,
      "step": 7067
    },
    {
      "epoch": 0.8180555555555555,
      "grad_norm": 0.08239005506038666,
      "learning_rate": 3.638888888888889e-06,
      "loss": 0.0108,
      "step": 7068
    },
    {
      "epoch": 0.8181712962962963,
      "grad_norm": 0.07604750245809555,
      "learning_rate": 3.6365740740740745e-06,
      "loss": 0.01,
      "step": 7069
    },
    {
      "epoch": 0.8182870370370371,
      "grad_norm": 0.05083267018198967,
      "learning_rate": 3.6342592592592596e-06,
      "loss": 0.0092,
      "step": 7070
    },
    {
      "epoch": 0.8184027777777778,
      "grad_norm": 0.037227727472782135,
      "learning_rate": 3.6319444444444447e-06,
      "loss": 0.0068,
      "step": 7071
    },
    {
      "epoch": 0.8185185185185185,
      "grad_norm": 0.05629058554768562,
      "learning_rate": 3.6296296296296302e-06,
      "loss": 0.01,
      "step": 7072
    },
    {
      "epoch": 0.8186342592592593,
      "grad_norm": 0.04234916716814041,
      "learning_rate": 3.6273148148148153e-06,
      "loss": 0.0077,
      "step": 7073
    },
    {
      "epoch": 0.81875,
      "grad_norm": 0.04239951819181442,
      "learning_rate": 3.625e-06,
      "loss": 0.0077,
      "step": 7074
    },
    {
      "epoch": 0.8188657407407407,
      "grad_norm": 0.060669224709272385,
      "learning_rate": 3.6226851851851855e-06,
      "loss": 0.007,
      "step": 7075
    },
    {
      "epoch": 0.8189814814814815,
      "grad_norm": 0.048713166266679764,
      "learning_rate": 3.6203703703703706e-06,
      "loss": 0.0087,
      "step": 7076
    },
    {
      "epoch": 0.8190972222222223,
      "grad_norm": 0.048855822533369064,
      "learning_rate": 3.6180555555555557e-06,
      "loss": 0.0061,
      "step": 7077
    },
    {
      "epoch": 0.819212962962963,
      "grad_norm": 0.03626817837357521,
      "learning_rate": 3.615740740740741e-06,
      "loss": 0.0065,
      "step": 7078
    },
    {
      "epoch": 0.8193287037037037,
      "grad_norm": 0.037619441747665405,
      "learning_rate": 3.6134259259259264e-06,
      "loss": 0.0068,
      "step": 7079
    },
    {
      "epoch": 0.8194444444444444,
      "grad_norm": 0.049764279276132584,
      "learning_rate": 3.6111111111111115e-06,
      "loss": 0.0086,
      "step": 7080
    },
    {
      "epoch": 0.8195601851851851,
      "grad_norm": 14.931952476501465,
      "learning_rate": 3.6087962962962966e-06,
      "loss": 3.2914,
      "step": 7081
    },
    {
      "epoch": 0.819675925925926,
      "grad_norm": 0.07467730343341827,
      "learning_rate": 3.606481481481482e-06,
      "loss": 0.0089,
      "step": 7082
    },
    {
      "epoch": 0.8197916666666667,
      "grad_norm": 0.03788549825549126,
      "learning_rate": 3.6041666666666667e-06,
      "loss": 0.0067,
      "step": 7083
    },
    {
      "epoch": 0.8199074074074074,
      "grad_norm": 0.03392840549349785,
      "learning_rate": 3.601851851851852e-06,
      "loss": 0.0062,
      "step": 7084
    },
    {
      "epoch": 0.8200231481481481,
      "grad_norm": 0.04939186945557594,
      "learning_rate": 3.5995370370370374e-06,
      "loss": 0.0088,
      "step": 7085
    },
    {
      "epoch": 0.8201388888888889,
      "grad_norm": 119.00594329833984,
      "learning_rate": 3.5972222222222225e-06,
      "loss": 1.318,
      "step": 7086
    },
    {
      "epoch": 0.8202546296296296,
      "grad_norm": 0.1548456847667694,
      "learning_rate": 3.5949074074074076e-06,
      "loss": 0.0132,
      "step": 7087
    },
    {
      "epoch": 0.8203703703703704,
      "grad_norm": 0.03539060428738594,
      "learning_rate": 3.592592592592593e-06,
      "loss": 0.0064,
      "step": 7088
    },
    {
      "epoch": 0.8204861111111111,
      "grad_norm": 0.03234616294503212,
      "learning_rate": 3.590277777777778e-06,
      "loss": 0.0059,
      "step": 7089
    },
    {
      "epoch": 0.8206018518518519,
      "grad_norm": 0.04426887258887291,
      "learning_rate": 3.5879629629629633e-06,
      "loss": 0.0078,
      "step": 7090
    },
    {
      "epoch": 0.8207175925925926,
      "grad_norm": 0.03669656068086624,
      "learning_rate": 3.585648148148149e-06,
      "loss": 0.0063,
      "step": 7091
    },
    {
      "epoch": 0.8208333333333333,
      "grad_norm": 0.05548820272088051,
      "learning_rate": 3.5833333333333335e-06,
      "loss": 0.0072,
      "step": 7092
    },
    {
      "epoch": 0.820949074074074,
      "grad_norm": 0.03572402894496918,
      "learning_rate": 3.5810185185185186e-06,
      "loss": 0.0065,
      "step": 7093
    },
    {
      "epoch": 0.8210648148148149,
      "grad_norm": 0.1529269516468048,
      "learning_rate": 3.578703703703704e-06,
      "loss": 0.0098,
      "step": 7094
    },
    {
      "epoch": 0.8211805555555556,
      "grad_norm": 0.04593396931886673,
      "learning_rate": 3.576388888888889e-06,
      "loss": 0.006,
      "step": 7095
    },
    {
      "epoch": 0.8212962962962963,
      "grad_norm": 0.0366051085293293,
      "learning_rate": 3.5740740740740743e-06,
      "loss": 0.0066,
      "step": 7096
    },
    {
      "epoch": 0.821412037037037,
      "grad_norm": 0.04702528566122055,
      "learning_rate": 3.57175925925926e-06,
      "loss": 0.0082,
      "step": 7097
    },
    {
      "epoch": 0.8215277777777777,
      "grad_norm": 0.04938794672489166,
      "learning_rate": 3.569444444444445e-06,
      "loss": 0.0087,
      "step": 7098
    },
    {
      "epoch": 0.8216435185185185,
      "grad_norm": 178.6356964111328,
      "learning_rate": 3.56712962962963e-06,
      "loss": 0.5894,
      "step": 7099
    },
    {
      "epoch": 0.8217592592592593,
      "grad_norm": 0.048715297132730484,
      "learning_rate": 3.5648148148148147e-06,
      "loss": 0.0086,
      "step": 7100
    },
    {
      "epoch": 0.821875,
      "grad_norm": 0.049456462264060974,
      "learning_rate": 3.5625e-06,
      "loss": 0.0085,
      "step": 7101
    },
    {
      "epoch": 0.8219907407407407,
      "grad_norm": 0.03709368780255318,
      "learning_rate": 3.5601851851851853e-06,
      "loss": 0.0063,
      "step": 7102
    },
    {
      "epoch": 0.8221064814814815,
      "grad_norm": 3.389616012573242,
      "learning_rate": 3.5578703703703704e-06,
      "loss": 0.0251,
      "step": 7103
    },
    {
      "epoch": 0.8222222222222222,
      "grad_norm": 0.041921794414520264,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0074,
      "step": 7104
    },
    {
      "epoch": 0.8223379629629629,
      "grad_norm": 0.0495428740978241,
      "learning_rate": 3.553240740740741e-06,
      "loss": 0.0083,
      "step": 7105
    },
    {
      "epoch": 0.8224537037037037,
      "grad_norm": 1.2691820859909058,
      "learning_rate": 3.550925925925926e-06,
      "loss": 0.0158,
      "step": 7106
    },
    {
      "epoch": 0.8225694444444445,
      "grad_norm": 0.317670613527298,
      "learning_rate": 3.5486111111111116e-06,
      "loss": 0.0156,
      "step": 7107
    },
    {
      "epoch": 0.8226851851851852,
      "grad_norm": 0.06314686685800552,
      "learning_rate": 3.5462962962962967e-06,
      "loss": 0.0108,
      "step": 7108
    },
    {
      "epoch": 0.8228009259259259,
      "grad_norm": 0.06586349010467529,
      "learning_rate": 3.5439814814814814e-06,
      "loss": 0.0076,
      "step": 7109
    },
    {
      "epoch": 0.8229166666666666,
      "grad_norm": 0.05947648733854294,
      "learning_rate": 3.5416666666666673e-06,
      "loss": 0.0106,
      "step": 7110
    },
    {
      "epoch": 0.8230324074074075,
      "grad_norm": 0.03429076820611954,
      "learning_rate": 3.539351851851852e-06,
      "loss": 0.006,
      "step": 7111
    },
    {
      "epoch": 0.8231481481481482,
      "grad_norm": 0.04768233746290207,
      "learning_rate": 3.537037037037037e-06,
      "loss": 0.0084,
      "step": 7112
    },
    {
      "epoch": 0.8232638888888889,
      "grad_norm": 0.05229174345731735,
      "learning_rate": 3.5347222222222226e-06,
      "loss": 0.0092,
      "step": 7113
    },
    {
      "epoch": 0.8233796296296296,
      "grad_norm": 0.039233963936567307,
      "learning_rate": 3.5324074074074077e-06,
      "loss": 0.0071,
      "step": 7114
    },
    {
      "epoch": 0.8234953703703703,
      "grad_norm": 0.044447533786296844,
      "learning_rate": 3.530092592592593e-06,
      "loss": 0.0071,
      "step": 7115
    },
    {
      "epoch": 0.8236111111111111,
      "grad_norm": 64.0331802368164,
      "learning_rate": 3.5277777777777784e-06,
      "loss": 2.5634,
      "step": 7116
    },
    {
      "epoch": 0.8237268518518519,
      "grad_norm": 0.03866022825241089,
      "learning_rate": 3.5254629629629635e-06,
      "loss": 0.007,
      "step": 7117
    },
    {
      "epoch": 0.8238425925925926,
      "grad_norm": 0.05428202077746391,
      "learning_rate": 3.523148148148148e-06,
      "loss": 0.0099,
      "step": 7118
    },
    {
      "epoch": 0.8239583333333333,
      "grad_norm": 0.03708099573850632,
      "learning_rate": 3.520833333333334e-06,
      "loss": 0.0067,
      "step": 7119
    },
    {
      "epoch": 0.8240740740740741,
      "grad_norm": 0.05980570241808891,
      "learning_rate": 3.5185185185185187e-06,
      "loss": 0.0085,
      "step": 7120
    },
    {
      "epoch": 0.8241898148148148,
      "grad_norm": 0.3202262222766876,
      "learning_rate": 3.516203703703704e-06,
      "loss": 0.0114,
      "step": 7121
    },
    {
      "epoch": 0.8243055555555555,
      "grad_norm": 0.05879106745123863,
      "learning_rate": 3.513888888888889e-06,
      "loss": 0.0105,
      "step": 7122
    },
    {
      "epoch": 0.8244212962962963,
      "grad_norm": 0.04027228429913521,
      "learning_rate": 3.5115740740740745e-06,
      "loss": 0.0072,
      "step": 7123
    },
    {
      "epoch": 0.8245370370370371,
      "grad_norm": 0.06220337003469467,
      "learning_rate": 3.5092592592592596e-06,
      "loss": 0.0113,
      "step": 7124
    },
    {
      "epoch": 0.8246527777777778,
      "grad_norm": 0.2565576136112213,
      "learning_rate": 3.5069444444444447e-06,
      "loss": 0.0102,
      "step": 7125
    },
    {
      "epoch": 0.8247685185185185,
      "grad_norm": 0.2794451415538788,
      "learning_rate": 3.50462962962963e-06,
      "loss": 0.0089,
      "step": 7126
    },
    {
      "epoch": 0.8248842592592592,
      "grad_norm": 0.04645717516541481,
      "learning_rate": 3.502314814814815e-06,
      "loss": 0.0082,
      "step": 7127
    },
    {
      "epoch": 0.825,
      "grad_norm": 0.04501042142510414,
      "learning_rate": 3.5e-06,
      "loss": 0.0059,
      "step": 7128
    },
    {
      "epoch": 0.8251157407407408,
      "grad_norm": 0.034142401069402695,
      "learning_rate": 3.4976851851851855e-06,
      "loss": 0.0062,
      "step": 7129
    },
    {
      "epoch": 0.8252314814814815,
      "grad_norm": 0.04596811532974243,
      "learning_rate": 3.4953703703703706e-06,
      "loss": 0.0082,
      "step": 7130
    },
    {
      "epoch": 0.8253472222222222,
      "grad_norm": 0.06729091703891754,
      "learning_rate": 3.4930555555555557e-06,
      "loss": 0.0084,
      "step": 7131
    },
    {
      "epoch": 0.825462962962963,
      "grad_norm": 0.054799094796180725,
      "learning_rate": 3.490740740740741e-06,
      "loss": 0.01,
      "step": 7132
    },
    {
      "epoch": 0.8255787037037037,
      "grad_norm": 0.046825818717479706,
      "learning_rate": 3.4884259259259263e-06,
      "loss": 0.0082,
      "step": 7133
    },
    {
      "epoch": 0.8256944444444444,
      "grad_norm": 0.04968700930476189,
      "learning_rate": 3.4861111111111114e-06,
      "loss": 0.0087,
      "step": 7134
    },
    {
      "epoch": 0.8258101851851852,
      "grad_norm": 0.04809396713972092,
      "learning_rate": 3.483796296296297e-06,
      "loss": 0.0086,
      "step": 7135
    },
    {
      "epoch": 0.825925925925926,
      "grad_norm": 0.04218769818544388,
      "learning_rate": 3.481481481481482e-06,
      "loss": 0.0075,
      "step": 7136
    },
    {
      "epoch": 0.8260416666666667,
      "grad_norm": 0.035341326147317886,
      "learning_rate": 3.4791666666666667e-06,
      "loss": 0.0064,
      "step": 7137
    },
    {
      "epoch": 0.8261574074074074,
      "grad_norm": 0.04222048074007034,
      "learning_rate": 3.476851851851852e-06,
      "loss": 0.0076,
      "step": 7138
    },
    {
      "epoch": 0.8262731481481481,
      "grad_norm": 0.17033155262470245,
      "learning_rate": 3.4745370370370373e-06,
      "loss": 0.0085,
      "step": 7139
    },
    {
      "epoch": 0.8263888888888888,
      "grad_norm": 0.04836633801460266,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 0.0071,
      "step": 7140
    },
    {
      "epoch": 0.8265046296296297,
      "grad_norm": 17.613330841064453,
      "learning_rate": 3.469907407407408e-06,
      "loss": 0.0493,
      "step": 7141
    },
    {
      "epoch": 0.8266203703703704,
      "grad_norm": 0.03504501283168793,
      "learning_rate": 3.467592592592593e-06,
      "loss": 0.0062,
      "step": 7142
    },
    {
      "epoch": 0.8267361111111111,
      "grad_norm": 0.034977756440639496,
      "learning_rate": 3.465277777777778e-06,
      "loss": 0.0064,
      "step": 7143
    },
    {
      "epoch": 0.8268518518518518,
      "grad_norm": 0.053872544318437576,
      "learning_rate": 3.4629629629629628e-06,
      "loss": 0.007,
      "step": 7144
    },
    {
      "epoch": 0.8269675925925926,
      "grad_norm": 0.04742908105254173,
      "learning_rate": 3.4606481481481487e-06,
      "loss": 0.0083,
      "step": 7145
    },
    {
      "epoch": 0.8270833333333333,
      "grad_norm": 0.03454688563942909,
      "learning_rate": 3.4583333333333334e-06,
      "loss": 0.0062,
      "step": 7146
    },
    {
      "epoch": 0.8271990740740741,
      "grad_norm": 0.048893216997385025,
      "learning_rate": 3.4560185185185185e-06,
      "loss": 0.0087,
      "step": 7147
    },
    {
      "epoch": 0.8273148148148148,
      "grad_norm": 0.06582006067037582,
      "learning_rate": 3.453703703703704e-06,
      "loss": 0.0085,
      "step": 7148
    },
    {
      "epoch": 0.8274305555555556,
      "grad_norm": 0.04167650267481804,
      "learning_rate": 3.451388888888889e-06,
      "loss": 0.0075,
      "step": 7149
    },
    {
      "epoch": 0.8275462962962963,
      "grad_norm": 0.038377318531274796,
      "learning_rate": 3.449074074074074e-06,
      "loss": 0.0069,
      "step": 7150
    },
    {
      "epoch": 0.827662037037037,
      "grad_norm": 0.044424984604120255,
      "learning_rate": 3.4467592592592597e-06,
      "loss": 0.0079,
      "step": 7151
    },
    {
      "epoch": 0.8277777777777777,
      "grad_norm": 0.06617125868797302,
      "learning_rate": 3.444444444444445e-06,
      "loss": 0.0118,
      "step": 7152
    },
    {
      "epoch": 0.8278935185185186,
      "grad_norm": 0.039102014154195786,
      "learning_rate": 3.4421296296296295e-06,
      "loss": 0.0071,
      "step": 7153
    },
    {
      "epoch": 0.8280092592592593,
      "grad_norm": 0.34687864780426025,
      "learning_rate": 3.4398148148148154e-06,
      "loss": 0.0111,
      "step": 7154
    },
    {
      "epoch": 0.828125,
      "grad_norm": 0.06378418207168579,
      "learning_rate": 3.4375e-06,
      "loss": 0.0115,
      "step": 7155
    },
    {
      "epoch": 0.8282407407407407,
      "grad_norm": 0.048250794410705566,
      "learning_rate": 3.4351851851851852e-06,
      "loss": 0.0086,
      "step": 7156
    },
    {
      "epoch": 0.8283564814814814,
      "grad_norm": 0.03882145509123802,
      "learning_rate": 3.4328703703703707e-06,
      "loss": 0.007,
      "step": 7157
    },
    {
      "epoch": 0.8284722222222223,
      "grad_norm": 0.041584525257349014,
      "learning_rate": 3.430555555555556e-06,
      "loss": 0.0073,
      "step": 7158
    },
    {
      "epoch": 0.828587962962963,
      "grad_norm": 0.0756516084074974,
      "learning_rate": 3.428240740740741e-06,
      "loss": 0.0099,
      "step": 7159
    },
    {
      "epoch": 0.8287037037037037,
      "grad_norm": 0.033325858414173126,
      "learning_rate": 3.4259259259259265e-06,
      "loss": 0.0061,
      "step": 7160
    },
    {
      "epoch": 0.8288194444444444,
      "grad_norm": 0.04697747156023979,
      "learning_rate": 3.4236111111111116e-06,
      "loss": 0.0082,
      "step": 7161
    },
    {
      "epoch": 0.8289351851851852,
      "grad_norm": 1.0439581871032715,
      "learning_rate": 3.4212962962962967e-06,
      "loss": 0.013,
      "step": 7162
    },
    {
      "epoch": 0.8290509259259259,
      "grad_norm": 0.053069066256284714,
      "learning_rate": 3.418981481481482e-06,
      "loss": 0.007,
      "step": 7163
    },
    {
      "epoch": 0.8291666666666667,
      "grad_norm": 0.19623716175556183,
      "learning_rate": 3.416666666666667e-06,
      "loss": 0.0123,
      "step": 7164
    },
    {
      "epoch": 0.8292824074074074,
      "grad_norm": 0.1303500086069107,
      "learning_rate": 3.414351851851852e-06,
      "loss": 0.0139,
      "step": 7165
    },
    {
      "epoch": 0.8293981481481482,
      "grad_norm": 13.019644737243652,
      "learning_rate": 3.4120370370370375e-06,
      "loss": 3.3207,
      "step": 7166
    },
    {
      "epoch": 0.8295138888888889,
      "grad_norm": 0.04582798480987549,
      "learning_rate": 3.4097222222222226e-06,
      "loss": 0.008,
      "step": 7167
    },
    {
      "epoch": 0.8296296296296296,
      "grad_norm": 0.04953210800886154,
      "learning_rate": 3.4074074074074077e-06,
      "loss": 0.009,
      "step": 7168
    },
    {
      "epoch": 0.8297453703703703,
      "grad_norm": 0.05163487792015076,
      "learning_rate": 3.4050925925925928e-06,
      "loss": 0.0066,
      "step": 7169
    },
    {
      "epoch": 0.8298611111111112,
      "grad_norm": 0.05117735639214516,
      "learning_rate": 3.4027777777777783e-06,
      "loss": 0.0093,
      "step": 7170
    },
    {
      "epoch": 0.8299768518518519,
      "grad_norm": 0.04754044488072395,
      "learning_rate": 3.4004629629629634e-06,
      "loss": 0.0084,
      "step": 7171
    },
    {
      "epoch": 0.8300925925925926,
      "grad_norm": 52.667884826660156,
      "learning_rate": 3.398148148148148e-06,
      "loss": 3.1226,
      "step": 7172
    },
    {
      "epoch": 0.8302083333333333,
      "grad_norm": 0.033876579254865646,
      "learning_rate": 3.3958333333333336e-06,
      "loss": 0.0061,
      "step": 7173
    },
    {
      "epoch": 0.830324074074074,
      "grad_norm": 0.23257184028625488,
      "learning_rate": 3.3935185185185187e-06,
      "loss": 0.0079,
      "step": 7174
    },
    {
      "epoch": 0.8304398148148148,
      "grad_norm": 0.045353204011917114,
      "learning_rate": 3.3912037037037038e-06,
      "loss": 0.0081,
      "step": 7175
    },
    {
      "epoch": 0.8305555555555556,
      "grad_norm": 0.05407697707414627,
      "learning_rate": 3.3888888888888893e-06,
      "loss": 0.0098,
      "step": 7176
    },
    {
      "epoch": 0.8306712962962963,
      "grad_norm": 0.05517645925283432,
      "learning_rate": 3.3865740740740744e-06,
      "loss": 0.0093,
      "step": 7177
    },
    {
      "epoch": 0.830787037037037,
      "grad_norm": 0.048573561012744904,
      "learning_rate": 3.3842592592592595e-06,
      "loss": 0.0087,
      "step": 7178
    },
    {
      "epoch": 0.8309027777777778,
      "grad_norm": 0.053214527666568756,
      "learning_rate": 3.381944444444445e-06,
      "loss": 0.0093,
      "step": 7179
    },
    {
      "epoch": 0.8310185185185185,
      "grad_norm": 0.10263411700725555,
      "learning_rate": 3.37962962962963e-06,
      "loss": 0.0128,
      "step": 7180
    },
    {
      "epoch": 0.8311342592592592,
      "grad_norm": 0.32476216554641724,
      "learning_rate": 3.3773148148148148e-06,
      "loss": 0.0143,
      "step": 7181
    },
    {
      "epoch": 0.83125,
      "grad_norm": 0.05807385966181755,
      "learning_rate": 3.3750000000000003e-06,
      "loss": 0.01,
      "step": 7182
    },
    {
      "epoch": 0.8313657407407408,
      "grad_norm": 0.06520136445760727,
      "learning_rate": 3.3726851851851854e-06,
      "loss": 0.0067,
      "step": 7183
    },
    {
      "epoch": 0.8314814814814815,
      "grad_norm": 0.10001537948846817,
      "learning_rate": 3.3703703703703705e-06,
      "loss": 0.0097,
      "step": 7184
    },
    {
      "epoch": 0.8315972222222222,
      "grad_norm": 0.04894355311989784,
      "learning_rate": 3.368055555555556e-06,
      "loss": 0.0083,
      "step": 7185
    },
    {
      "epoch": 0.8317129629629629,
      "grad_norm": 0.031792376190423965,
      "learning_rate": 3.365740740740741e-06,
      "loss": 0.0058,
      "step": 7186
    },
    {
      "epoch": 0.8318287037037037,
      "grad_norm": 0.06930480897426605,
      "learning_rate": 3.363425925925926e-06,
      "loss": 0.009,
      "step": 7187
    },
    {
      "epoch": 0.8319444444444445,
      "grad_norm": 0.05411887541413307,
      "learning_rate": 3.3611111111111117e-06,
      "loss": 0.0071,
      "step": 7188
    },
    {
      "epoch": 0.8320601851851852,
      "grad_norm": 0.05192532390356064,
      "learning_rate": 3.358796296296297e-06,
      "loss": 0.0093,
      "step": 7189
    },
    {
      "epoch": 0.8321759259259259,
      "grad_norm": 0.06686899065971375,
      "learning_rate": 3.3564814814814815e-06,
      "loss": 0.0098,
      "step": 7190
    },
    {
      "epoch": 0.8322916666666667,
      "grad_norm": 0.06451531499624252,
      "learning_rate": 3.3541666666666666e-06,
      "loss": 0.0116,
      "step": 7191
    },
    {
      "epoch": 0.8324074074074074,
      "grad_norm": 0.07508407533168793,
      "learning_rate": 3.351851851851852e-06,
      "loss": 0.0094,
      "step": 7192
    },
    {
      "epoch": 0.8325231481481481,
      "grad_norm": 0.05386405438184738,
      "learning_rate": 3.3495370370370372e-06,
      "loss": 0.0095,
      "step": 7193
    },
    {
      "epoch": 0.8326388888888889,
      "grad_norm": 0.50475013256073,
      "learning_rate": 3.3472222222222223e-06,
      "loss": 0.0157,
      "step": 7194
    },
    {
      "epoch": 0.8327546296296297,
      "grad_norm": 0.05510341003537178,
      "learning_rate": 3.344907407407408e-06,
      "loss": 0.0101,
      "step": 7195
    },
    {
      "epoch": 0.8328703703703704,
      "grad_norm": 0.050126269459724426,
      "learning_rate": 3.342592592592593e-06,
      "loss": 0.009,
      "step": 7196
    },
    {
      "epoch": 0.8329861111111111,
      "grad_norm": 0.04549683630466461,
      "learning_rate": 3.340277777777778e-06,
      "loss": 0.0081,
      "step": 7197
    },
    {
      "epoch": 0.8331018518518518,
      "grad_norm": 0.03578699752688408,
      "learning_rate": 3.3379629629629636e-06,
      "loss": 0.0064,
      "step": 7198
    },
    {
      "epoch": 0.8332175925925925,
      "grad_norm": 0.04844992235302925,
      "learning_rate": 3.3356481481481482e-06,
      "loss": 0.0082,
      "step": 7199
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 0.05350728705525398,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0094,
      "step": 7200
    },
    {
      "epoch": 0.8334490740740741,
      "grad_norm": 0.052444033324718475,
      "learning_rate": 3.331018518518519e-06,
      "loss": 0.0094,
      "step": 7201
    },
    {
      "epoch": 0.8335648148148148,
      "grad_norm": 0.06783249229192734,
      "learning_rate": 3.328703703703704e-06,
      "loss": 0.0117,
      "step": 7202
    },
    {
      "epoch": 0.8336805555555555,
      "grad_norm": 11.321139335632324,
      "learning_rate": 3.326388888888889e-06,
      "loss": 2.7983,
      "step": 7203
    },
    {
      "epoch": 0.8337962962962963,
      "grad_norm": 0.1300651729106903,
      "learning_rate": 3.3240740740740746e-06,
      "loss": 0.0079,
      "step": 7204
    },
    {
      "epoch": 0.8339120370370371,
      "grad_norm": 0.04617404565215111,
      "learning_rate": 3.3217592592592597e-06,
      "loss": 0.008,
      "step": 7205
    },
    {
      "epoch": 0.8340277777777778,
      "grad_norm": 0.032954975962638855,
      "learning_rate": 3.3194444444444448e-06,
      "loss": 0.006,
      "step": 7206
    },
    {
      "epoch": 0.8341435185185185,
      "grad_norm": 0.04357271268963814,
      "learning_rate": 3.3171296296296303e-06,
      "loss": 0.0077,
      "step": 7207
    },
    {
      "epoch": 0.8342592592592593,
      "grad_norm": 0.06805486232042313,
      "learning_rate": 3.314814814814815e-06,
      "loss": 0.0117,
      "step": 7208
    },
    {
      "epoch": 0.834375,
      "grad_norm": 0.03379226475954056,
      "learning_rate": 3.3125e-06,
      "loss": 0.0061,
      "step": 7209
    },
    {
      "epoch": 0.8344907407407407,
      "grad_norm": 0.04600970447063446,
      "learning_rate": 3.3101851851851856e-06,
      "loss": 0.0081,
      "step": 7210
    },
    {
      "epoch": 0.8346064814814815,
      "grad_norm": 0.04625827074050903,
      "learning_rate": 3.3078703703703707e-06,
      "loss": 0.0081,
      "step": 7211
    },
    {
      "epoch": 0.8347222222222223,
      "grad_norm": 0.04399334266781807,
      "learning_rate": 3.3055555555555558e-06,
      "loss": 0.0077,
      "step": 7212
    },
    {
      "epoch": 0.834837962962963,
      "grad_norm": 0.03849223628640175,
      "learning_rate": 3.303240740740741e-06,
      "loss": 0.007,
      "step": 7213
    },
    {
      "epoch": 0.8349537037037037,
      "grad_norm": 0.03750382363796234,
      "learning_rate": 3.3009259259259264e-06,
      "loss": 0.0065,
      "step": 7214
    },
    {
      "epoch": 0.8350694444444444,
      "grad_norm": 0.057211603969335556,
      "learning_rate": 3.2986111111111115e-06,
      "loss": 0.01,
      "step": 7215
    },
    {
      "epoch": 0.8351851851851851,
      "grad_norm": 0.0572076141834259,
      "learning_rate": 3.296296296296296e-06,
      "loss": 0.0103,
      "step": 7216
    },
    {
      "epoch": 0.835300925925926,
      "grad_norm": 0.046343844383955,
      "learning_rate": 3.293981481481482e-06,
      "loss": 0.0081,
      "step": 7217
    },
    {
      "epoch": 0.8354166666666667,
      "grad_norm": 0.05388248711824417,
      "learning_rate": 3.2916666666666668e-06,
      "loss": 0.0093,
      "step": 7218
    },
    {
      "epoch": 0.8355324074074074,
      "grad_norm": 0.08587462455034256,
      "learning_rate": 3.289351851851852e-06,
      "loss": 0.0111,
      "step": 7219
    },
    {
      "epoch": 0.8356481481481481,
      "grad_norm": 0.06397270411252975,
      "learning_rate": 3.2870370370370374e-06,
      "loss": 0.0102,
      "step": 7220
    },
    {
      "epoch": 0.8357638888888889,
      "grad_norm": 0.051793646067380905,
      "learning_rate": 3.2847222222222225e-06,
      "loss": 0.0093,
      "step": 7221
    },
    {
      "epoch": 0.8358796296296296,
      "grad_norm": 0.8444581627845764,
      "learning_rate": 3.2824074074074076e-06,
      "loss": 0.0115,
      "step": 7222
    },
    {
      "epoch": 0.8359953703703704,
      "grad_norm": 0.03787298873066902,
      "learning_rate": 3.280092592592593e-06,
      "loss": 0.0068,
      "step": 7223
    },
    {
      "epoch": 0.8361111111111111,
      "grad_norm": 0.04859667643904686,
      "learning_rate": 3.277777777777778e-06,
      "loss": 0.0081,
      "step": 7224
    },
    {
      "epoch": 0.8362268518518519,
      "grad_norm": 0.05770455300807953,
      "learning_rate": 3.275462962962963e-06,
      "loss": 0.0104,
      "step": 7225
    },
    {
      "epoch": 0.8363425925925926,
      "grad_norm": 0.056715305894613266,
      "learning_rate": 3.273148148148149e-06,
      "loss": 0.0103,
      "step": 7226
    },
    {
      "epoch": 0.8364583333333333,
      "grad_norm": 129.52957153320312,
      "learning_rate": 3.2708333333333335e-06,
      "loss": 0.983,
      "step": 7227
    },
    {
      "epoch": 0.836574074074074,
      "grad_norm": 0.04159543290734291,
      "learning_rate": 3.2685185185185186e-06,
      "loss": 0.0076,
      "step": 7228
    },
    {
      "epoch": 0.8366898148148149,
      "grad_norm": 0.04210857301950455,
      "learning_rate": 3.266203703703704e-06,
      "loss": 0.0076,
      "step": 7229
    },
    {
      "epoch": 0.8368055555555556,
      "grad_norm": 0.048729509115219116,
      "learning_rate": 3.2638888888888892e-06,
      "loss": 0.0071,
      "step": 7230
    },
    {
      "epoch": 0.8369212962962963,
      "grad_norm": 0.4003632068634033,
      "learning_rate": 3.2615740740740743e-06,
      "loss": 0.0179,
      "step": 7231
    },
    {
      "epoch": 0.837037037037037,
      "grad_norm": 0.04498951509594917,
      "learning_rate": 3.25925925925926e-06,
      "loss": 0.008,
      "step": 7232
    },
    {
      "epoch": 0.8371527777777777,
      "grad_norm": 0.03732401132583618,
      "learning_rate": 3.256944444444445e-06,
      "loss": 0.006,
      "step": 7233
    },
    {
      "epoch": 0.8372685185185185,
      "grad_norm": 0.06807859241962433,
      "learning_rate": 3.2546296296296296e-06,
      "loss": 0.0076,
      "step": 7234
    },
    {
      "epoch": 0.8373842592592593,
      "grad_norm": 0.044257014989852905,
      "learning_rate": 3.2523148148148147e-06,
      "loss": 0.0078,
      "step": 7235
    },
    {
      "epoch": 0.8375,
      "grad_norm": 0.05848952755331993,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.0096,
      "step": 7236
    },
    {
      "epoch": 0.8376157407407407,
      "grad_norm": 0.03906857222318649,
      "learning_rate": 3.2476851851851853e-06,
      "loss": 0.007,
      "step": 7237
    },
    {
      "epoch": 0.8377314814814815,
      "grad_norm": 0.12879033386707306,
      "learning_rate": 3.2453703703703704e-06,
      "loss": 0.0127,
      "step": 7238
    },
    {
      "epoch": 0.8378472222222222,
      "grad_norm": 0.04260895773768425,
      "learning_rate": 3.243055555555556e-06,
      "loss": 0.0076,
      "step": 7239
    },
    {
      "epoch": 0.8379629629629629,
      "grad_norm": 1.1603968143463135,
      "learning_rate": 3.240740740740741e-06,
      "loss": 0.0137,
      "step": 7240
    },
    {
      "epoch": 0.8380787037037037,
      "grad_norm": 0.03337813541293144,
      "learning_rate": 3.238425925925926e-06,
      "loss": 0.006,
      "step": 7241
    },
    {
      "epoch": 0.8381944444444445,
      "grad_norm": 8.560651779174805,
      "learning_rate": 3.2361111111111117e-06,
      "loss": 3.1505,
      "step": 7242
    },
    {
      "epoch": 0.8383101851851852,
      "grad_norm": 0.04652533307671547,
      "learning_rate": 3.2337962962962968e-06,
      "loss": 0.0083,
      "step": 7243
    },
    {
      "epoch": 0.8384259259259259,
      "grad_norm": 0.07368391007184982,
      "learning_rate": 3.2314814814814814e-06,
      "loss": 0.008,
      "step": 7244
    },
    {
      "epoch": 0.8385416666666666,
      "grad_norm": 0.03477342054247856,
      "learning_rate": 3.229166666666667e-06,
      "loss": 0.0063,
      "step": 7245
    },
    {
      "epoch": 0.8386574074074075,
      "grad_norm": 0.044053032994270325,
      "learning_rate": 3.226851851851852e-06,
      "loss": 0.0077,
      "step": 7246
    },
    {
      "epoch": 0.8387731481481482,
      "grad_norm": 2.1503915786743164,
      "learning_rate": 3.224537037037037e-06,
      "loss": 0.0187,
      "step": 7247
    },
    {
      "epoch": 0.8388888888888889,
      "grad_norm": 0.05906648561358452,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.0089,
      "step": 7248
    },
    {
      "epoch": 0.8390046296296296,
      "grad_norm": 0.041395142674446106,
      "learning_rate": 3.2199074074074078e-06,
      "loss": 0.0075,
      "step": 7249
    },
    {
      "epoch": 0.8391203703703703,
      "grad_norm": 0.04840517416596413,
      "learning_rate": 3.217592592592593e-06,
      "loss": 0.0063,
      "step": 7250
    },
    {
      "epoch": 0.8392361111111111,
      "grad_norm": 85.53279876708984,
      "learning_rate": 3.2152777777777784e-06,
      "loss": 2.1932,
      "step": 7251
    },
    {
      "epoch": 0.8393518518518519,
      "grad_norm": 0.09859409928321838,
      "learning_rate": 3.2129629629629635e-06,
      "loss": 0.0102,
      "step": 7252
    },
    {
      "epoch": 0.8394675925925926,
      "grad_norm": 0.0654384046792984,
      "learning_rate": 3.210648148148148e-06,
      "loss": 0.0101,
      "step": 7253
    },
    {
      "epoch": 0.8395833333333333,
      "grad_norm": 0.04928455501794815,
      "learning_rate": 3.2083333333333337e-06,
      "loss": 0.0073,
      "step": 7254
    },
    {
      "epoch": 0.8396990740740741,
      "grad_norm": 0.3738107979297638,
      "learning_rate": 3.2060185185185188e-06,
      "loss": 0.0148,
      "step": 7255
    },
    {
      "epoch": 0.8398148148148148,
      "grad_norm": 0.07731080055236816,
      "learning_rate": 3.203703703703704e-06,
      "loss": 0.0099,
      "step": 7256
    },
    {
      "epoch": 0.8399305555555555,
      "grad_norm": 0.04569390416145325,
      "learning_rate": 3.201388888888889e-06,
      "loss": 0.0079,
      "step": 7257
    },
    {
      "epoch": 0.8400462962962963,
      "grad_norm": 0.04253387823700905,
      "learning_rate": 3.1990740740740745e-06,
      "loss": 0.0076,
      "step": 7258
    },
    {
      "epoch": 0.8401620370370371,
      "grad_norm": 4.61631965637207,
      "learning_rate": 3.1967592592592596e-06,
      "loss": 0.0359,
      "step": 7259
    },
    {
      "epoch": 0.8402777777777778,
      "grad_norm": 0.055129315704107285,
      "learning_rate": 3.1944444444444443e-06,
      "loss": 0.01,
      "step": 7260
    },
    {
      "epoch": 0.8403935185185185,
      "grad_norm": 0.1269650161266327,
      "learning_rate": 3.19212962962963e-06,
      "loss": 0.0117,
      "step": 7261
    },
    {
      "epoch": 0.8405092592592592,
      "grad_norm": 0.03286893665790558,
      "learning_rate": 3.189814814814815e-06,
      "loss": 0.006,
      "step": 7262
    },
    {
      "epoch": 0.840625,
      "grad_norm": 14.235132217407227,
      "learning_rate": 3.1875e-06,
      "loss": 3.0742,
      "step": 7263
    },
    {
      "epoch": 0.8407407407407408,
      "grad_norm": 0.052217233926057816,
      "learning_rate": 3.1851851851851855e-06,
      "loss": 0.0068,
      "step": 7264
    },
    {
      "epoch": 0.8408564814814815,
      "grad_norm": 0.036479320377111435,
      "learning_rate": 3.1828703703703706e-06,
      "loss": 0.0065,
      "step": 7265
    },
    {
      "epoch": 0.8409722222222222,
      "grad_norm": 0.04705800861120224,
      "learning_rate": 3.1805555555555557e-06,
      "loss": 0.0082,
      "step": 7266
    },
    {
      "epoch": 0.841087962962963,
      "grad_norm": 0.050635553896427155,
      "learning_rate": 3.1782407407407412e-06,
      "loss": 0.0074,
      "step": 7267
    },
    {
      "epoch": 0.8412037037037037,
      "grad_norm": 0.49112287163734436,
      "learning_rate": 3.1759259259259263e-06,
      "loss": 0.013,
      "step": 7268
    },
    {
      "epoch": 0.8413194444444444,
      "grad_norm": 0.04485657438635826,
      "learning_rate": 3.1736111111111114e-06,
      "loss": 0.008,
      "step": 7269
    },
    {
      "epoch": 0.8414351851851852,
      "grad_norm": 141.7425537109375,
      "learning_rate": 3.171296296296297e-06,
      "loss": 1.4649,
      "step": 7270
    },
    {
      "epoch": 0.841550925925926,
      "grad_norm": 0.049951281398534775,
      "learning_rate": 3.1689814814814816e-06,
      "loss": 0.009,
      "step": 7271
    },
    {
      "epoch": 0.8416666666666667,
      "grad_norm": 0.05555775389075279,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 0.0099,
      "step": 7272
    },
    {
      "epoch": 0.8417824074074074,
      "grad_norm": 0.05691177770495415,
      "learning_rate": 3.1643518518518522e-06,
      "loss": 0.0098,
      "step": 7273
    },
    {
      "epoch": 0.8418981481481481,
      "grad_norm": 0.04293837025761604,
      "learning_rate": 3.1620370370370373e-06,
      "loss": 0.0076,
      "step": 7274
    },
    {
      "epoch": 0.8420138888888888,
      "grad_norm": 0.054354481399059296,
      "learning_rate": 3.1597222222222224e-06,
      "loss": 0.0092,
      "step": 7275
    },
    {
      "epoch": 0.8421296296296297,
      "grad_norm": 0.05044734850525856,
      "learning_rate": 3.157407407407408e-06,
      "loss": 0.0084,
      "step": 7276
    },
    {
      "epoch": 0.8422453703703704,
      "grad_norm": 0.053335562348365784,
      "learning_rate": 3.155092592592593e-06,
      "loss": 0.0093,
      "step": 7277
    },
    {
      "epoch": 0.8423611111111111,
      "grad_norm": 0.046078525483608246,
      "learning_rate": 3.152777777777778e-06,
      "loss": 0.0081,
      "step": 7278
    },
    {
      "epoch": 0.8424768518518518,
      "grad_norm": 0.04395247623324394,
      "learning_rate": 3.150462962962963e-06,
      "loss": 0.0077,
      "step": 7279
    },
    {
      "epoch": 0.8425925925925926,
      "grad_norm": 0.06276178359985352,
      "learning_rate": 3.1481481481481483e-06,
      "loss": 0.0103,
      "step": 7280
    },
    {
      "epoch": 0.8427083333333333,
      "grad_norm": 145.4018096923828,
      "learning_rate": 3.1458333333333334e-06,
      "loss": 0.2576,
      "step": 7281
    },
    {
      "epoch": 0.8428240740740741,
      "grad_norm": 0.0448196679353714,
      "learning_rate": 3.1435185185185185e-06,
      "loss": 0.0058,
      "step": 7282
    },
    {
      "epoch": 0.8429398148148148,
      "grad_norm": 0.03523842245340347,
      "learning_rate": 3.141203703703704e-06,
      "loss": 0.0064,
      "step": 7283
    },
    {
      "epoch": 0.8430555555555556,
      "grad_norm": 0.04712578281760216,
      "learning_rate": 3.138888888888889e-06,
      "loss": 0.0082,
      "step": 7284
    },
    {
      "epoch": 0.8431712962962963,
      "grad_norm": 0.04269786924123764,
      "learning_rate": 3.1365740740740742e-06,
      "loss": 0.0076,
      "step": 7285
    },
    {
      "epoch": 0.843287037037037,
      "grad_norm": 0.07605084031820297,
      "learning_rate": 3.1342592592592598e-06,
      "loss": 0.009,
      "step": 7286
    },
    {
      "epoch": 0.8434027777777777,
      "grad_norm": 0.058596134185791016,
      "learning_rate": 3.131944444444445e-06,
      "loss": 0.007,
      "step": 7287
    },
    {
      "epoch": 0.8435185185185186,
      "grad_norm": 0.0547950454056263,
      "learning_rate": 3.1296296296296295e-06,
      "loss": 0.0092,
      "step": 7288
    },
    {
      "epoch": 0.8436342592592593,
      "grad_norm": 0.04383508488535881,
      "learning_rate": 3.127314814814815e-06,
      "loss": 0.0072,
      "step": 7289
    },
    {
      "epoch": 0.84375,
      "grad_norm": 0.03607353940606117,
      "learning_rate": 3.125e-06,
      "loss": 0.0065,
      "step": 7290
    },
    {
      "epoch": 0.8438657407407407,
      "grad_norm": 0.1467960774898529,
      "learning_rate": 3.1226851851851852e-06,
      "loss": 0.0074,
      "step": 7291
    },
    {
      "epoch": 0.8439814814814814,
      "grad_norm": 0.41325756907463074,
      "learning_rate": 3.1203703703703708e-06,
      "loss": 0.0135,
      "step": 7292
    },
    {
      "epoch": 0.8440972222222223,
      "grad_norm": 0.4559897780418396,
      "learning_rate": 3.118055555555556e-06,
      "loss": 0.0113,
      "step": 7293
    },
    {
      "epoch": 0.844212962962963,
      "grad_norm": 0.04573137313127518,
      "learning_rate": 3.115740740740741e-06,
      "loss": 0.0069,
      "step": 7294
    },
    {
      "epoch": 0.8443287037037037,
      "grad_norm": 0.0411653146147728,
      "learning_rate": 3.1134259259259265e-06,
      "loss": 0.0063,
      "step": 7295
    },
    {
      "epoch": 0.8444444444444444,
      "grad_norm": 0.07036362588405609,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0118,
      "step": 7296
    },
    {
      "epoch": 0.8445601851851852,
      "grad_norm": 0.0421699695289135,
      "learning_rate": 3.1087962962962963e-06,
      "loss": 0.0075,
      "step": 7297
    },
    {
      "epoch": 0.8446759259259259,
      "grad_norm": 0.048898302018642426,
      "learning_rate": 3.1064814814814818e-06,
      "loss": 0.0063,
      "step": 7298
    },
    {
      "epoch": 0.8447916666666667,
      "grad_norm": 0.10418619960546494,
      "learning_rate": 3.104166666666667e-06,
      "loss": 0.0103,
      "step": 7299
    },
    {
      "epoch": 0.8449074074074074,
      "grad_norm": 0.06339020282030106,
      "learning_rate": 3.101851851851852e-06,
      "loss": 0.011,
      "step": 7300
    },
    {
      "epoch": 0.8450231481481482,
      "grad_norm": 0.049453359097242355,
      "learning_rate": 3.0995370370370375e-06,
      "loss": 0.0089,
      "step": 7301
    },
    {
      "epoch": 0.8451388888888889,
      "grad_norm": 133.9278106689453,
      "learning_rate": 3.0972222222222226e-06,
      "loss": 0.8021,
      "step": 7302
    },
    {
      "epoch": 0.8452546296296296,
      "grad_norm": 0.05457098409533501,
      "learning_rate": 3.0949074074074077e-06,
      "loss": 0.0094,
      "step": 7303
    },
    {
      "epoch": 0.8453703703703703,
      "grad_norm": 0.06390602886676788,
      "learning_rate": 3.0925925925925928e-06,
      "loss": 0.0114,
      "step": 7304
    },
    {
      "epoch": 0.8454861111111112,
      "grad_norm": 0.041399307548999786,
      "learning_rate": 3.0902777777777783e-06,
      "loss": 0.0069,
      "step": 7305
    },
    {
      "epoch": 0.8456018518518519,
      "grad_norm": 39.346797943115234,
      "learning_rate": 3.087962962962963e-06,
      "loss": 0.1936,
      "step": 7306
    },
    {
      "epoch": 0.8457175925925926,
      "grad_norm": 0.14014339447021484,
      "learning_rate": 3.085648148148148e-06,
      "loss": 0.0117,
      "step": 7307
    },
    {
      "epoch": 0.8458333333333333,
      "grad_norm": 0.03293323516845703,
      "learning_rate": 3.0833333333333336e-06,
      "loss": 0.006,
      "step": 7308
    },
    {
      "epoch": 0.845949074074074,
      "grad_norm": 0.03598136082291603,
      "learning_rate": 3.0810185185185187e-06,
      "loss": 0.0065,
      "step": 7309
    },
    {
      "epoch": 0.8460648148148148,
      "grad_norm": 0.030688343569636345,
      "learning_rate": 3.078703703703704e-06,
      "loss": 0.0056,
      "step": 7310
    },
    {
      "epoch": 0.8461805555555556,
      "grad_norm": 0.1022840216755867,
      "learning_rate": 3.0763888888888893e-06,
      "loss": 0.009,
      "step": 7311
    },
    {
      "epoch": 0.8462962962962963,
      "grad_norm": 0.0821327194571495,
      "learning_rate": 3.0740740740740744e-06,
      "loss": 0.0084,
      "step": 7312
    },
    {
      "epoch": 0.846412037037037,
      "grad_norm": 0.04454132914543152,
      "learning_rate": 3.0717592592592595e-06,
      "loss": 0.0077,
      "step": 7313
    },
    {
      "epoch": 0.8465277777777778,
      "grad_norm": 0.05960550904273987,
      "learning_rate": 3.069444444444445e-06,
      "loss": 0.0096,
      "step": 7314
    },
    {
      "epoch": 0.8466435185185185,
      "grad_norm": 0.04925640672445297,
      "learning_rate": 3.0671296296296297e-06,
      "loss": 0.0075,
      "step": 7315
    },
    {
      "epoch": 0.8467592592592592,
      "grad_norm": 0.03287368267774582,
      "learning_rate": 3.064814814814815e-06,
      "loss": 0.0059,
      "step": 7316
    },
    {
      "epoch": 0.846875,
      "grad_norm": 0.07679751515388489,
      "learning_rate": 3.0625000000000003e-06,
      "loss": 0.0098,
      "step": 7317
    },
    {
      "epoch": 0.8469907407407408,
      "grad_norm": 0.10861876606941223,
      "learning_rate": 3.0601851851851854e-06,
      "loss": 0.0103,
      "step": 7318
    },
    {
      "epoch": 0.8471064814814815,
      "grad_norm": 0.03336521238088608,
      "learning_rate": 3.0578703703703705e-06,
      "loss": 0.006,
      "step": 7319
    },
    {
      "epoch": 0.8472222222222222,
      "grad_norm": 0.056899428367614746,
      "learning_rate": 3.055555555555556e-06,
      "loss": 0.0101,
      "step": 7320
    },
    {
      "epoch": 0.8473379629629629,
      "grad_norm": 0.03250517696142197,
      "learning_rate": 3.053240740740741e-06,
      "loss": 0.0059,
      "step": 7321
    },
    {
      "epoch": 0.8474537037037037,
      "grad_norm": 0.04293224215507507,
      "learning_rate": 3.0509259259259262e-06,
      "loss": 0.0056,
      "step": 7322
    },
    {
      "epoch": 0.8475694444444445,
      "grad_norm": 0.0955452248454094,
      "learning_rate": 3.0486111111111118e-06,
      "loss": 0.0085,
      "step": 7323
    },
    {
      "epoch": 0.8476851851851852,
      "grad_norm": 0.03051397018134594,
      "learning_rate": 3.0462962962962964e-06,
      "loss": 0.0056,
      "step": 7324
    },
    {
      "epoch": 0.8478009259259259,
      "grad_norm": 0.0777057334780693,
      "learning_rate": 3.0439814814814815e-06,
      "loss": 0.0092,
      "step": 7325
    },
    {
      "epoch": 0.8479166666666667,
      "grad_norm": 0.04434823989868164,
      "learning_rate": 3.0416666666666666e-06,
      "loss": 0.0079,
      "step": 7326
    },
    {
      "epoch": 0.8480324074074074,
      "grad_norm": 0.21324719488620758,
      "learning_rate": 3.039351851851852e-06,
      "loss": 0.0127,
      "step": 7327
    },
    {
      "epoch": 0.8481481481481481,
      "grad_norm": 1.715286135673523,
      "learning_rate": 3.0370370370370372e-06,
      "loss": 0.0181,
      "step": 7328
    },
    {
      "epoch": 0.8482638888888889,
      "grad_norm": 0.04955708608031273,
      "learning_rate": 3.0347222222222223e-06,
      "loss": 0.0085,
      "step": 7329
    },
    {
      "epoch": 0.8483796296296297,
      "grad_norm": 0.15554533898830414,
      "learning_rate": 3.032407407407408e-06,
      "loss": 0.0113,
      "step": 7330
    },
    {
      "epoch": 0.8484953703703704,
      "grad_norm": 0.04885494336485863,
      "learning_rate": 3.030092592592593e-06,
      "loss": 0.0088,
      "step": 7331
    },
    {
      "epoch": 0.8486111111111111,
      "grad_norm": 9.822412490844727,
      "learning_rate": 3.0277777777777776e-06,
      "loss": 3.4735,
      "step": 7332
    },
    {
      "epoch": 0.8487268518518518,
      "grad_norm": 0.037747811526060104,
      "learning_rate": 3.0254629629629636e-06,
      "loss": 0.0067,
      "step": 7333
    },
    {
      "epoch": 0.8488425925925925,
      "grad_norm": 0.038092903792858124,
      "learning_rate": 3.0231481481481483e-06,
      "loss": 0.0068,
      "step": 7334
    },
    {
      "epoch": 0.8489583333333334,
      "grad_norm": 0.17712311446666718,
      "learning_rate": 3.0208333333333334e-06,
      "loss": 0.0126,
      "step": 7335
    },
    {
      "epoch": 0.8490740740740741,
      "grad_norm": 0.03474727272987366,
      "learning_rate": 3.018518518518519e-06,
      "loss": 0.0063,
      "step": 7336
    },
    {
      "epoch": 0.8491898148148148,
      "grad_norm": 0.07547566294670105,
      "learning_rate": 3.016203703703704e-06,
      "loss": 0.011,
      "step": 7337
    },
    {
      "epoch": 0.8493055555555555,
      "grad_norm": 0.042701952159404755,
      "learning_rate": 3.013888888888889e-06,
      "loss": 0.0076,
      "step": 7338
    },
    {
      "epoch": 0.8494212962962963,
      "grad_norm": 0.12951168417930603,
      "learning_rate": 3.0115740740740746e-06,
      "loss": 0.0091,
      "step": 7339
    },
    {
      "epoch": 0.8495370370370371,
      "grad_norm": 0.3028630316257477,
      "learning_rate": 3.0092592592592597e-06,
      "loss": 0.0089,
      "step": 7340
    },
    {
      "epoch": 0.8496527777777778,
      "grad_norm": 0.04239354655146599,
      "learning_rate": 3.0069444444444444e-06,
      "loss": 0.0075,
      "step": 7341
    },
    {
      "epoch": 0.8497685185185185,
      "grad_norm": 0.04825296998023987,
      "learning_rate": 3.0046296296296303e-06,
      "loss": 0.0059,
      "step": 7342
    },
    {
      "epoch": 0.8498842592592593,
      "grad_norm": 0.048778556287288666,
      "learning_rate": 3.002314814814815e-06,
      "loss": 0.0063,
      "step": 7343
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24792391061782837,
      "learning_rate": 3e-06,
      "loss": 0.011,
      "step": 7344
    },
    {
      "epoch": 0.8501157407407407,
      "grad_norm": 0.11490456759929657,
      "learning_rate": 2.9976851851851856e-06,
      "loss": 0.0109,
      "step": 7345
    },
    {
      "epoch": 0.8502314814814815,
      "grad_norm": 0.1412758231163025,
      "learning_rate": 2.9953703703703707e-06,
      "loss": 0.0121,
      "step": 7346
    },
    {
      "epoch": 0.8503472222222223,
      "grad_norm": 0.04771316424012184,
      "learning_rate": 2.993055555555556e-06,
      "loss": 0.0087,
      "step": 7347
    },
    {
      "epoch": 0.850462962962963,
      "grad_norm": 0.049117960035800934,
      "learning_rate": 2.990740740740741e-06,
      "loss": 0.0063,
      "step": 7348
    },
    {
      "epoch": 0.8505787037037037,
      "grad_norm": 0.060628414154052734,
      "learning_rate": 2.9884259259259264e-06,
      "loss": 0.0102,
      "step": 7349
    },
    {
      "epoch": 0.8506944444444444,
      "grad_norm": 0.054332032799720764,
      "learning_rate": 2.986111111111111e-06,
      "loss": 0.0099,
      "step": 7350
    },
    {
      "epoch": 0.8508101851851851,
      "grad_norm": 0.036530498415231705,
      "learning_rate": 2.983796296296296e-06,
      "loss": 0.0066,
      "step": 7351
    },
    {
      "epoch": 0.850925925925926,
      "grad_norm": 4.801314830780029,
      "learning_rate": 2.9814814814814817e-06,
      "loss": 0.0221,
      "step": 7352
    },
    {
      "epoch": 0.8510416666666667,
      "grad_norm": 0.03666551038622856,
      "learning_rate": 2.979166666666667e-06,
      "loss": 0.0066,
      "step": 7353
    },
    {
      "epoch": 0.8511574074074074,
      "grad_norm": 0.049053579568862915,
      "learning_rate": 2.976851851851852e-06,
      "loss": 0.008,
      "step": 7354
    },
    {
      "epoch": 0.8512731481481481,
      "grad_norm": 0.05241665616631508,
      "learning_rate": 2.9745370370370374e-06,
      "loss": 0.0071,
      "step": 7355
    },
    {
      "epoch": 0.8513888888888889,
      "grad_norm": 0.06311319768428802,
      "learning_rate": 2.9722222222222225e-06,
      "loss": 0.0114,
      "step": 7356
    },
    {
      "epoch": 0.8515046296296296,
      "grad_norm": 0.06267161667346954,
      "learning_rate": 2.9699074074074076e-06,
      "loss": 0.0081,
      "step": 7357
    },
    {
      "epoch": 0.8516203703703704,
      "grad_norm": 0.23627085983753204,
      "learning_rate": 2.967592592592593e-06,
      "loss": 0.0119,
      "step": 7358
    },
    {
      "epoch": 0.8517361111111111,
      "grad_norm": 0.06353233754634857,
      "learning_rate": 2.9652777777777782e-06,
      "loss": 0.0112,
      "step": 7359
    },
    {
      "epoch": 0.8518518518518519,
      "grad_norm": 216.93431091308594,
      "learning_rate": 2.962962962962963e-06,
      "loss": 1.7116,
      "step": 7360
    },
    {
      "epoch": 0.8519675925925926,
      "grad_norm": 0.04623414948582649,
      "learning_rate": 2.9606481481481484e-06,
      "loss": 0.0083,
      "step": 7361
    },
    {
      "epoch": 0.8520833333333333,
      "grad_norm": 0.07669933140277863,
      "learning_rate": 2.9583333333333335e-06,
      "loss": 0.0101,
      "step": 7362
    },
    {
      "epoch": 0.852199074074074,
      "grad_norm": 0.03408258408308029,
      "learning_rate": 2.9560185185185186e-06,
      "loss": 0.0062,
      "step": 7363
    },
    {
      "epoch": 0.8523148148148149,
      "grad_norm": 0.033904414623975754,
      "learning_rate": 2.953703703703704e-06,
      "loss": 0.006,
      "step": 7364
    },
    {
      "epoch": 0.8524305555555556,
      "grad_norm": 0.041801419109106064,
      "learning_rate": 2.9513888888888892e-06,
      "loss": 0.0074,
      "step": 7365
    },
    {
      "epoch": 0.8525462962962963,
      "grad_norm": 0.07143433392047882,
      "learning_rate": 2.9490740740740743e-06,
      "loss": 0.0094,
      "step": 7366
    },
    {
      "epoch": 0.852662037037037,
      "grad_norm": 0.06672005355358124,
      "learning_rate": 2.94675925925926e-06,
      "loss": 0.0081,
      "step": 7367
    },
    {
      "epoch": 0.8527777777777777,
      "grad_norm": 203.6136932373047,
      "learning_rate": 2.944444444444445e-06,
      "loss": 0.82,
      "step": 7368
    },
    {
      "epoch": 0.8528935185185185,
      "grad_norm": 0.03498644009232521,
      "learning_rate": 2.9421296296296296e-06,
      "loss": 0.006,
      "step": 7369
    },
    {
      "epoch": 0.8530092592592593,
      "grad_norm": 0.037536315619945526,
      "learning_rate": 2.9398148148148147e-06,
      "loss": 0.0066,
      "step": 7370
    },
    {
      "epoch": 0.853125,
      "grad_norm": 0.05377798154950142,
      "learning_rate": 2.9375000000000003e-06,
      "loss": 0.0069,
      "step": 7371
    },
    {
      "epoch": 0.8532407407407407,
      "grad_norm": 0.049867283552885056,
      "learning_rate": 2.9351851851851853e-06,
      "loss": 0.0086,
      "step": 7372
    },
    {
      "epoch": 0.8533564814814815,
      "grad_norm": 0.044284265488386154,
      "learning_rate": 2.9328703703703704e-06,
      "loss": 0.0079,
      "step": 7373
    },
    {
      "epoch": 0.8534722222222222,
      "grad_norm": 0.054241444915533066,
      "learning_rate": 2.930555555555556e-06,
      "loss": 0.0095,
      "step": 7374
    },
    {
      "epoch": 0.8535879629629629,
      "grad_norm": 0.038010746240615845,
      "learning_rate": 2.928240740740741e-06,
      "loss": 0.0068,
      "step": 7375
    },
    {
      "epoch": 0.8537037037037037,
      "grad_norm": 0.0463653989136219,
      "learning_rate": 2.9259259259259257e-06,
      "loss": 0.0081,
      "step": 7376
    },
    {
      "epoch": 0.8538194444444445,
      "grad_norm": 0.046371180564165115,
      "learning_rate": 2.9236111111111117e-06,
      "loss": 0.0082,
      "step": 7377
    },
    {
      "epoch": 0.8539351851851852,
      "grad_norm": 0.10355276614427567,
      "learning_rate": 2.9212962962962964e-06,
      "loss": 0.0078,
      "step": 7378
    },
    {
      "epoch": 0.8540509259259259,
      "grad_norm": 247.71585083007812,
      "learning_rate": 2.9189814814814815e-06,
      "loss": 2.0557,
      "step": 7379
    },
    {
      "epoch": 0.8541666666666666,
      "grad_norm": 0.03605441004037857,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.0065,
      "step": 7380
    },
    {
      "epoch": 0.8542824074074075,
      "grad_norm": 132.51051330566406,
      "learning_rate": 2.914351851851852e-06,
      "loss": 1.8423,
      "step": 7381
    },
    {
      "epoch": 0.8543981481481482,
      "grad_norm": 0.10111165791749954,
      "learning_rate": 2.912037037037037e-06,
      "loss": 0.01,
      "step": 7382
    },
    {
      "epoch": 0.8545138888888889,
      "grad_norm": 0.996127724647522,
      "learning_rate": 2.9097222222222227e-06,
      "loss": 0.0117,
      "step": 7383
    },
    {
      "epoch": 0.8546296296296296,
      "grad_norm": 0.05338543653488159,
      "learning_rate": 2.907407407407408e-06,
      "loss": 0.0097,
      "step": 7384
    },
    {
      "epoch": 0.8547453703703703,
      "grad_norm": 0.05172204226255417,
      "learning_rate": 2.905092592592593e-06,
      "loss": 0.0084,
      "step": 7385
    },
    {
      "epoch": 0.8548611111111111,
      "grad_norm": 0.04269624873995781,
      "learning_rate": 2.9027777777777784e-06,
      "loss": 0.0056,
      "step": 7386
    },
    {
      "epoch": 0.8549768518518519,
      "grad_norm": 0.04551674425601959,
      "learning_rate": 2.900462962962963e-06,
      "loss": 0.008,
      "step": 7387
    },
    {
      "epoch": 0.8550925925925926,
      "grad_norm": 0.054313044995069504,
      "learning_rate": 2.898148148148148e-06,
      "loss": 0.0096,
      "step": 7388
    },
    {
      "epoch": 0.8552083333333333,
      "grad_norm": 0.052642181515693665,
      "learning_rate": 2.8958333333333337e-06,
      "loss": 0.0069,
      "step": 7389
    },
    {
      "epoch": 0.8553240740740741,
      "grad_norm": 0.047965001314878464,
      "learning_rate": 2.893518518518519e-06,
      "loss": 0.0084,
      "step": 7390
    },
    {
      "epoch": 0.8554398148148148,
      "grad_norm": 126.73005676269531,
      "learning_rate": 2.891203703703704e-06,
      "loss": 1.0647,
      "step": 7391
    },
    {
      "epoch": 0.8555555555555555,
      "grad_norm": 0.03558618202805519,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.0064,
      "step": 7392
    },
    {
      "epoch": 0.8556712962962963,
      "grad_norm": 0.20644451677799225,
      "learning_rate": 2.8865740740740745e-06,
      "loss": 0.0091,
      "step": 7393
    },
    {
      "epoch": 0.8557870370370371,
      "grad_norm": 0.041759733110666275,
      "learning_rate": 2.8842592592592596e-06,
      "loss": 0.0074,
      "step": 7394
    },
    {
      "epoch": 0.8559027777777778,
      "grad_norm": 0.042061157524585724,
      "learning_rate": 2.8819444444444443e-06,
      "loss": 0.0074,
      "step": 7395
    },
    {
      "epoch": 0.8560185185185185,
      "grad_norm": 0.4333832859992981,
      "learning_rate": 2.87962962962963e-06,
      "loss": 0.0103,
      "step": 7396
    },
    {
      "epoch": 0.8561342592592592,
      "grad_norm": 0.05652737244963646,
      "learning_rate": 2.877314814814815e-06,
      "loss": 0.009,
      "step": 7397
    },
    {
      "epoch": 0.85625,
      "grad_norm": 0.0485941506922245,
      "learning_rate": 2.875e-06,
      "loss": 0.0081,
      "step": 7398
    },
    {
      "epoch": 0.8563657407407408,
      "grad_norm": 0.04460814967751503,
      "learning_rate": 2.8726851851851855e-06,
      "loss": 0.0079,
      "step": 7399
    },
    {
      "epoch": 0.8564814814814815,
      "grad_norm": 0.05583363026380539,
      "learning_rate": 2.8703703703703706e-06,
      "loss": 0.0069,
      "step": 7400
    },
    {
      "epoch": 0.8565972222222222,
      "grad_norm": 0.12929467856884003,
      "learning_rate": 2.8680555555555557e-06,
      "loss": 0.0116,
      "step": 7401
    },
    {
      "epoch": 0.856712962962963,
      "grad_norm": 0.07243555039167404,
      "learning_rate": 2.8657407407407412e-06,
      "loss": 0.0115,
      "step": 7402
    },
    {
      "epoch": 0.8568287037037037,
      "grad_norm": 0.04579181969165802,
      "learning_rate": 2.8634259259259263e-06,
      "loss": 0.008,
      "step": 7403
    },
    {
      "epoch": 0.8569444444444444,
      "grad_norm": 0.5597261786460876,
      "learning_rate": 2.861111111111111e-06,
      "loss": 0.0118,
      "step": 7404
    },
    {
      "epoch": 0.8570601851851852,
      "grad_norm": 0.41261690855026245,
      "learning_rate": 2.8587962962962965e-06,
      "loss": 0.0106,
      "step": 7405
    },
    {
      "epoch": 0.857175925925926,
      "grad_norm": 0.06047394126653671,
      "learning_rate": 2.8564814814814816e-06,
      "loss": 0.0111,
      "step": 7406
    },
    {
      "epoch": 0.8572916666666667,
      "grad_norm": 0.046418193727731705,
      "learning_rate": 2.8541666666666667e-06,
      "loss": 0.0082,
      "step": 7407
    },
    {
      "epoch": 0.8574074074074074,
      "grad_norm": 7.515182018280029,
      "learning_rate": 2.8518518518518522e-06,
      "loss": 0.0341,
      "step": 7408
    },
    {
      "epoch": 0.8575231481481481,
      "grad_norm": 0.04874511808156967,
      "learning_rate": 2.8495370370370373e-06,
      "loss": 0.0088,
      "step": 7409
    },
    {
      "epoch": 0.8576388888888888,
      "grad_norm": 0.03452308103442192,
      "learning_rate": 2.8472222222222224e-06,
      "loss": 0.0062,
      "step": 7410
    },
    {
      "epoch": 0.8577546296296297,
      "grad_norm": 0.04657002538442612,
      "learning_rate": 2.844907407407408e-06,
      "loss": 0.0085,
      "step": 7411
    },
    {
      "epoch": 0.8578703703703704,
      "grad_norm": 0.05244837701320648,
      "learning_rate": 2.842592592592593e-06,
      "loss": 0.0083,
      "step": 7412
    },
    {
      "epoch": 0.8579861111111111,
      "grad_norm": 0.07172641903162003,
      "learning_rate": 2.8402777777777777e-06,
      "loss": 0.0117,
      "step": 7413
    },
    {
      "epoch": 0.8581018518518518,
      "grad_norm": 0.08421555906534195,
      "learning_rate": 2.837962962962963e-06,
      "loss": 0.0088,
      "step": 7414
    },
    {
      "epoch": 0.8582175925925926,
      "grad_norm": 0.05106367915868759,
      "learning_rate": 2.8356481481481484e-06,
      "loss": 0.0084,
      "step": 7415
    },
    {
      "epoch": 0.8583333333333333,
      "grad_norm": 0.056674662977457047,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 0.0072,
      "step": 7416
    },
    {
      "epoch": 0.8584490740740741,
      "grad_norm": 0.039771828800439835,
      "learning_rate": 2.8310185185185185e-06,
      "loss": 0.0067,
      "step": 7417
    },
    {
      "epoch": 0.8585648148148148,
      "grad_norm": 0.04264086112380028,
      "learning_rate": 2.828703703703704e-06,
      "loss": 0.0075,
      "step": 7418
    },
    {
      "epoch": 0.8586805555555556,
      "grad_norm": 220.66929626464844,
      "learning_rate": 2.826388888888889e-06,
      "loss": 1.648,
      "step": 7419
    },
    {
      "epoch": 0.8587962962962963,
      "grad_norm": 0.054353050887584686,
      "learning_rate": 2.8240740740740743e-06,
      "loss": 0.0091,
      "step": 7420
    },
    {
      "epoch": 0.858912037037037,
      "grad_norm": 0.07061531394720078,
      "learning_rate": 2.8217592592592598e-06,
      "loss": 0.0076,
      "step": 7421
    },
    {
      "epoch": 0.8590277777777777,
      "grad_norm": 0.04467258229851723,
      "learning_rate": 2.8194444444444445e-06,
      "loss": 0.0081,
      "step": 7422
    },
    {
      "epoch": 0.8591435185185186,
      "grad_norm": 0.04255584999918938,
      "learning_rate": 2.8171296296296296e-06,
      "loss": 0.0056,
      "step": 7423
    },
    {
      "epoch": 0.8592592592592593,
      "grad_norm": 0.0642961785197258,
      "learning_rate": 2.814814814814815e-06,
      "loss": 0.0095,
      "step": 7424
    },
    {
      "epoch": 0.859375,
      "grad_norm": 0.032546818256378174,
      "learning_rate": 2.8125e-06,
      "loss": 0.0059,
      "step": 7425
    },
    {
      "epoch": 0.8594907407407407,
      "grad_norm": 0.054401759058237076,
      "learning_rate": 2.8101851851851853e-06,
      "loss": 0.0093,
      "step": 7426
    },
    {
      "epoch": 0.8596064814814814,
      "grad_norm": 0.05792788788676262,
      "learning_rate": 2.807870370370371e-06,
      "loss": 0.0091,
      "step": 7427
    },
    {
      "epoch": 0.8597222222222223,
      "grad_norm": 209.6029815673828,
      "learning_rate": 2.805555555555556e-06,
      "loss": 1.9349,
      "step": 7428
    },
    {
      "epoch": 0.859837962962963,
      "grad_norm": 0.046880997717380524,
      "learning_rate": 2.803240740740741e-06,
      "loss": 0.0084,
      "step": 7429
    },
    {
      "epoch": 0.8599537037037037,
      "grad_norm": 0.07618749886751175,
      "learning_rate": 2.8009259259259265e-06,
      "loss": 0.0099,
      "step": 7430
    },
    {
      "epoch": 0.8600694444444444,
      "grad_norm": 0.04687248915433884,
      "learning_rate": 2.798611111111111e-06,
      "loss": 0.008,
      "step": 7431
    },
    {
      "epoch": 0.8601851851851852,
      "grad_norm": 0.06757380068302155,
      "learning_rate": 2.7962962962962963e-06,
      "loss": 0.0075,
      "step": 7432
    },
    {
      "epoch": 0.8603009259259259,
      "grad_norm": 0.045429427176713943,
      "learning_rate": 2.793981481481482e-06,
      "loss": 0.0082,
      "step": 7433
    },
    {
      "epoch": 0.8604166666666667,
      "grad_norm": 0.04440193995833397,
      "learning_rate": 2.791666666666667e-06,
      "loss": 0.008,
      "step": 7434
    },
    {
      "epoch": 0.8605324074074074,
      "grad_norm": 0.041942331939935684,
      "learning_rate": 2.789351851851852e-06,
      "loss": 0.0075,
      "step": 7435
    },
    {
      "epoch": 0.8606481481481482,
      "grad_norm": 0.03066861256957054,
      "learning_rate": 2.7870370370370375e-06,
      "loss": 0.0056,
      "step": 7436
    },
    {
      "epoch": 0.8607638888888889,
      "grad_norm": 0.046286240220069885,
      "learning_rate": 2.7847222222222226e-06,
      "loss": 0.008,
      "step": 7437
    },
    {
      "epoch": 0.8608796296296296,
      "grad_norm": 0.05429212376475334,
      "learning_rate": 2.7824074074074077e-06,
      "loss": 0.0096,
      "step": 7438
    },
    {
      "epoch": 0.8609953703703703,
      "grad_norm": 0.03227226808667183,
      "learning_rate": 2.7800925925925924e-06,
      "loss": 0.0059,
      "step": 7439
    },
    {
      "epoch": 0.8611111111111112,
      "grad_norm": 0.0357196219265461,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.0064,
      "step": 7440
    },
    {
      "epoch": 0.8612268518518519,
      "grad_norm": 0.044897813349962234,
      "learning_rate": 2.775462962962963e-06,
      "loss": 0.0079,
      "step": 7441
    },
    {
      "epoch": 0.8613425925925926,
      "grad_norm": 0.04215546324849129,
      "learning_rate": 2.773148148148148e-06,
      "loss": 0.0071,
      "step": 7442
    },
    {
      "epoch": 0.8614583333333333,
      "grad_norm": 0.035378679633140564,
      "learning_rate": 2.7708333333333336e-06,
      "loss": 0.0063,
      "step": 7443
    },
    {
      "epoch": 0.861574074074074,
      "grad_norm": 0.04470188915729523,
      "learning_rate": 2.7685185185185187e-06,
      "loss": 0.008,
      "step": 7444
    },
    {
      "epoch": 0.8616898148148148,
      "grad_norm": 0.03466189652681351,
      "learning_rate": 2.766203703703704e-06,
      "loss": 0.0062,
      "step": 7445
    },
    {
      "epoch": 0.8618055555555556,
      "grad_norm": 0.057001013308763504,
      "learning_rate": 2.7638888888888893e-06,
      "loss": 0.0087,
      "step": 7446
    },
    {
      "epoch": 0.8619212962962963,
      "grad_norm": 0.2428383231163025,
      "learning_rate": 2.7615740740740744e-06,
      "loss": 0.0104,
      "step": 7447
    },
    {
      "epoch": 0.862037037037037,
      "grad_norm": 0.05221923068165779,
      "learning_rate": 2.759259259259259e-06,
      "loss": 0.0078,
      "step": 7448
    },
    {
      "epoch": 0.8621527777777778,
      "grad_norm": 0.30562835931777954,
      "learning_rate": 2.756944444444445e-06,
      "loss": 0.0133,
      "step": 7449
    },
    {
      "epoch": 0.8622685185185185,
      "grad_norm": 0.04176504537463188,
      "learning_rate": 2.7546296296296297e-06,
      "loss": 0.0074,
      "step": 7450
    },
    {
      "epoch": 0.8623842592592592,
      "grad_norm": 0.3248569369316101,
      "learning_rate": 2.752314814814815e-06,
      "loss": 0.0095,
      "step": 7451
    },
    {
      "epoch": 0.8625,
      "grad_norm": 0.2889782786369324,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0114,
      "step": 7452
    },
    {
      "epoch": 0.8626157407407408,
      "grad_norm": 0.04431162029504776,
      "learning_rate": 2.7476851851851854e-06,
      "loss": 0.0078,
      "step": 7453
    },
    {
      "epoch": 0.8627314814814815,
      "grad_norm": 0.054926350712776184,
      "learning_rate": 2.7453703703703705e-06,
      "loss": 0.0097,
      "step": 7454
    },
    {
      "epoch": 0.8628472222222222,
      "grad_norm": 0.054991818964481354,
      "learning_rate": 2.743055555555556e-06,
      "loss": 0.0099,
      "step": 7455
    },
    {
      "epoch": 0.8629629629629629,
      "grad_norm": 180.3927459716797,
      "learning_rate": 2.740740740740741e-06,
      "loss": 0.5006,
      "step": 7456
    },
    {
      "epoch": 0.8630787037037037,
      "grad_norm": 0.04542764276266098,
      "learning_rate": 2.738425925925926e-06,
      "loss": 0.0077,
      "step": 7457
    },
    {
      "epoch": 0.8631944444444445,
      "grad_norm": 0.04392499104142189,
      "learning_rate": 2.7361111111111118e-06,
      "loss": 0.0079,
      "step": 7458
    },
    {
      "epoch": 0.8633101851851852,
      "grad_norm": 0.04250277206301689,
      "learning_rate": 2.7337962962962965e-06,
      "loss": 0.0076,
      "step": 7459
    },
    {
      "epoch": 0.8634259259259259,
      "grad_norm": 0.04563727229833603,
      "learning_rate": 2.7314814814814816e-06,
      "loss": 0.0082,
      "step": 7460
    },
    {
      "epoch": 0.8635416666666667,
      "grad_norm": 0.043961700052022934,
      "learning_rate": 2.7291666666666667e-06,
      "loss": 0.0076,
      "step": 7461
    },
    {
      "epoch": 0.8636574074074074,
      "grad_norm": 0.056115444749593735,
      "learning_rate": 2.726851851851852e-06,
      "loss": 0.0099,
      "step": 7462
    },
    {
      "epoch": 0.8637731481481481,
      "grad_norm": 0.04842087998986244,
      "learning_rate": 2.7245370370370373e-06,
      "loss": 0.0059,
      "step": 7463
    },
    {
      "epoch": 0.8638888888888889,
      "grad_norm": 66.568115234375,
      "learning_rate": 2.7222222222222224e-06,
      "loss": 2.3161,
      "step": 7464
    },
    {
      "epoch": 0.8640046296296297,
      "grad_norm": 0.035835087299346924,
      "learning_rate": 2.719907407407408e-06,
      "loss": 0.0065,
      "step": 7465
    },
    {
      "epoch": 0.8641203703703704,
      "grad_norm": 0.21918344497680664,
      "learning_rate": 2.717592592592593e-06,
      "loss": 0.0112,
      "step": 7466
    },
    {
      "epoch": 0.8642361111111111,
      "grad_norm": 1.1506584882736206,
      "learning_rate": 2.7152777777777777e-06,
      "loss": 0.0138,
      "step": 7467
    },
    {
      "epoch": 0.8643518518518518,
      "grad_norm": 0.034566834568977356,
      "learning_rate": 2.712962962962963e-06,
      "loss": 0.0062,
      "step": 7468
    },
    {
      "epoch": 0.8644675925925925,
      "grad_norm": 0.03178749978542328,
      "learning_rate": 2.7106481481481483e-06,
      "loss": 0.0058,
      "step": 7469
    },
    {
      "epoch": 0.8645833333333334,
      "grad_norm": 0.04269617050886154,
      "learning_rate": 2.7083333333333334e-06,
      "loss": 0.0076,
      "step": 7470
    },
    {
      "epoch": 0.8646990740740741,
      "grad_norm": 0.0440482422709465,
      "learning_rate": 2.706018518518519e-06,
      "loss": 0.0079,
      "step": 7471
    },
    {
      "epoch": 0.8648148148148148,
      "grad_norm": 0.045988310128450394,
      "learning_rate": 2.703703703703704e-06,
      "loss": 0.0078,
      "step": 7472
    },
    {
      "epoch": 0.8649305555555555,
      "grad_norm": 0.033842120319604874,
      "learning_rate": 2.701388888888889e-06,
      "loss": 0.0058,
      "step": 7473
    },
    {
      "epoch": 0.8650462962962963,
      "grad_norm": 0.03271443769335747,
      "learning_rate": 2.6990740740740746e-06,
      "loss": 0.0059,
      "step": 7474
    },
    {
      "epoch": 0.8651620370370371,
      "grad_norm": 0.05248947814106941,
      "learning_rate": 2.6967592592592597e-06,
      "loss": 0.0093,
      "step": 7475
    },
    {
      "epoch": 0.8652777777777778,
      "grad_norm": 0.047689296305179596,
      "learning_rate": 2.6944444444444444e-06,
      "loss": 0.0083,
      "step": 7476
    },
    {
      "epoch": 0.8653935185185185,
      "grad_norm": 0.0481901541352272,
      "learning_rate": 2.69212962962963e-06,
      "loss": 0.0083,
      "step": 7477
    },
    {
      "epoch": 0.8655092592592593,
      "grad_norm": 0.1388193815946579,
      "learning_rate": 2.689814814814815e-06,
      "loss": 0.0119,
      "step": 7478
    },
    {
      "epoch": 0.865625,
      "grad_norm": 0.05320068821310997,
      "learning_rate": 2.6875e-06,
      "loss": 0.0068,
      "step": 7479
    },
    {
      "epoch": 0.8657407407407407,
      "grad_norm": 0.048798613250255585,
      "learning_rate": 2.6851851851851856e-06,
      "loss": 0.0063,
      "step": 7480
    },
    {
      "epoch": 0.8658564814814815,
      "grad_norm": 0.049596842378377914,
      "learning_rate": 2.6828703703703707e-06,
      "loss": 0.0066,
      "step": 7481
    },
    {
      "epoch": 0.8659722222222223,
      "grad_norm": 0.04802614450454712,
      "learning_rate": 2.680555555555556e-06,
      "loss": 0.0086,
      "step": 7482
    },
    {
      "epoch": 0.866087962962963,
      "grad_norm": 0.04211676865816116,
      "learning_rate": 2.6782407407407405e-06,
      "loss": 0.0055,
      "step": 7483
    },
    {
      "epoch": 0.8662037037037037,
      "grad_norm": 0.033509910106658936,
      "learning_rate": 2.6759259259259264e-06,
      "loss": 0.0061,
      "step": 7484
    },
    {
      "epoch": 0.8663194444444444,
      "grad_norm": 0.03548341989517212,
      "learning_rate": 2.673611111111111e-06,
      "loss": 0.0064,
      "step": 7485
    },
    {
      "epoch": 0.8664351851851851,
      "grad_norm": 0.04990030825138092,
      "learning_rate": 2.671296296296296e-06,
      "loss": 0.0078,
      "step": 7486
    },
    {
      "epoch": 0.866550925925926,
      "grad_norm": 0.06076187267899513,
      "learning_rate": 2.6689814814814817e-06,
      "loss": 0.011,
      "step": 7487
    },
    {
      "epoch": 0.8666666666666667,
      "grad_norm": 0.31748640537261963,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.008,
      "step": 7488
    },
    {
      "epoch": 0.8667824074074074,
      "grad_norm": 2.685809373855591,
      "learning_rate": 2.664351851851852e-06,
      "loss": 0.023,
      "step": 7489
    },
    {
      "epoch": 0.8668981481481481,
      "grad_norm": 0.03384145721793175,
      "learning_rate": 2.6620370370370374e-06,
      "loss": 0.0061,
      "step": 7490
    },
    {
      "epoch": 0.8670138888888889,
      "grad_norm": 0.04334819316864014,
      "learning_rate": 2.6597222222222225e-06,
      "loss": 0.0056,
      "step": 7491
    },
    {
      "epoch": 0.8671296296296296,
      "grad_norm": 0.04711582139134407,
      "learning_rate": 2.6574074074074076e-06,
      "loss": 0.0085,
      "step": 7492
    },
    {
      "epoch": 0.8672453703703704,
      "grad_norm": 0.0348396897315979,
      "learning_rate": 2.655092592592593e-06,
      "loss": 0.0062,
      "step": 7493
    },
    {
      "epoch": 0.8673611111111111,
      "grad_norm": 0.05234384164214134,
      "learning_rate": 2.652777777777778e-06,
      "loss": 0.0095,
      "step": 7494
    },
    {
      "epoch": 0.8674768518518519,
      "grad_norm": 0.03415178507566452,
      "learning_rate": 2.650462962962963e-06,
      "loss": 0.0061,
      "step": 7495
    },
    {
      "epoch": 0.8675925925925926,
      "grad_norm": 0.050637997686862946,
      "learning_rate": 2.6481481481481485e-06,
      "loss": 0.0087,
      "step": 7496
    },
    {
      "epoch": 0.8677083333333333,
      "grad_norm": 0.05396920070052147,
      "learning_rate": 2.6458333333333336e-06,
      "loss": 0.0092,
      "step": 7497
    },
    {
      "epoch": 0.867824074074074,
      "grad_norm": 0.06154797598719597,
      "learning_rate": 2.6435185185185187e-06,
      "loss": 0.0091,
      "step": 7498
    },
    {
      "epoch": 0.8679398148148149,
      "grad_norm": 0.034175626933574677,
      "learning_rate": 2.641203703703704e-06,
      "loss": 0.0062,
      "step": 7499
    },
    {
      "epoch": 0.8680555555555556,
      "grad_norm": 0.0325409434735775,
      "learning_rate": 2.6388888888888893e-06,
      "loss": 0.0059,
      "step": 7500
    },
    {
      "epoch": 0.8681712962962963,
      "grad_norm": 15.50744915008545,
      "learning_rate": 2.6365740740740744e-06,
      "loss": 3.2554,
      "step": 7501
    },
    {
      "epoch": 0.868287037037037,
      "grad_norm": 0.04828628897666931,
      "learning_rate": 2.63425925925926e-06,
      "loss": 0.0078,
      "step": 7502
    },
    {
      "epoch": 0.8684027777777777,
      "grad_norm": 0.051613856106996536,
      "learning_rate": 2.6319444444444446e-06,
      "loss": 0.0068,
      "step": 7503
    },
    {
      "epoch": 0.8685185185185185,
      "grad_norm": 0.03839590772986412,
      "learning_rate": 2.6296296296296297e-06,
      "loss": 0.0069,
      "step": 7504
    },
    {
      "epoch": 0.8686342592592593,
      "grad_norm": 0.09355118125677109,
      "learning_rate": 2.6273148148148148e-06,
      "loss": 0.0122,
      "step": 7505
    },
    {
      "epoch": 0.86875,
      "grad_norm": 0.03361363708972931,
      "learning_rate": 2.6250000000000003e-06,
      "loss": 0.0061,
      "step": 7506
    },
    {
      "epoch": 0.8688657407407407,
      "grad_norm": 0.29160693287849426,
      "learning_rate": 2.6226851851851854e-06,
      "loss": 0.0091,
      "step": 7507
    },
    {
      "epoch": 0.8689814814814815,
      "grad_norm": 0.034103963524103165,
      "learning_rate": 2.6203703703703705e-06,
      "loss": 0.0059,
      "step": 7508
    },
    {
      "epoch": 0.8690972222222222,
      "grad_norm": 0.05184976011514664,
      "learning_rate": 2.618055555555556e-06,
      "loss": 0.0088,
      "step": 7509
    },
    {
      "epoch": 0.8692129629629629,
      "grad_norm": 61.29854202270508,
      "learning_rate": 2.615740740740741e-06,
      "loss": 2.3909,
      "step": 7510
    },
    {
      "epoch": 0.8693287037037037,
      "grad_norm": 0.07137849181890488,
      "learning_rate": 2.6134259259259258e-06,
      "loss": 0.0104,
      "step": 7511
    },
    {
      "epoch": 0.8694444444444445,
      "grad_norm": 0.051410380750894547,
      "learning_rate": 2.6111111111111113e-06,
      "loss": 0.0091,
      "step": 7512
    },
    {
      "epoch": 0.8695601851851852,
      "grad_norm": 72.64493560791016,
      "learning_rate": 2.6087962962962964e-06,
      "loss": 2.6118,
      "step": 7513
    },
    {
      "epoch": 0.8696759259259259,
      "grad_norm": 0.03643321990966797,
      "learning_rate": 2.6064814814814815e-06,
      "loss": 0.0066,
      "step": 7514
    },
    {
      "epoch": 0.8697916666666666,
      "grad_norm": 0.05673643574118614,
      "learning_rate": 2.604166666666667e-06,
      "loss": 0.0089,
      "step": 7515
    },
    {
      "epoch": 0.8699074074074075,
      "grad_norm": 0.05732590705156326,
      "learning_rate": 2.601851851851852e-06,
      "loss": 0.0104,
      "step": 7516
    },
    {
      "epoch": 0.8700231481481482,
      "grad_norm": 0.05555841326713562,
      "learning_rate": 2.599537037037037e-06,
      "loss": 0.009,
      "step": 7517
    },
    {
      "epoch": 0.8701388888888889,
      "grad_norm": 0.05805547907948494,
      "learning_rate": 2.5972222222222227e-06,
      "loss": 0.0085,
      "step": 7518
    },
    {
      "epoch": 0.8702546296296296,
      "grad_norm": 47.08285140991211,
      "learning_rate": 2.594907407407408e-06,
      "loss": 0.1182,
      "step": 7519
    },
    {
      "epoch": 0.8703703703703703,
      "grad_norm": 0.030038833618164062,
      "learning_rate": 2.5925925925925925e-06,
      "loss": 0.0055,
      "step": 7520
    },
    {
      "epoch": 0.8704861111111111,
      "grad_norm": 0.05293568968772888,
      "learning_rate": 2.5902777777777784e-06,
      "loss": 0.0092,
      "step": 7521
    },
    {
      "epoch": 0.8706018518518519,
      "grad_norm": 0.5647469758987427,
      "learning_rate": 2.587962962962963e-06,
      "loss": 0.0133,
      "step": 7522
    },
    {
      "epoch": 0.8707175925925926,
      "grad_norm": 0.08508367091417313,
      "learning_rate": 2.585648148148148e-06,
      "loss": 0.0093,
      "step": 7523
    },
    {
      "epoch": 0.8708333333333333,
      "grad_norm": 0.05620889365673065,
      "learning_rate": 2.5833333333333337e-06,
      "loss": 0.0099,
      "step": 7524
    },
    {
      "epoch": 0.8709490740740741,
      "grad_norm": 0.04443662613630295,
      "learning_rate": 2.581018518518519e-06,
      "loss": 0.0066,
      "step": 7525
    },
    {
      "epoch": 0.8710648148148148,
      "grad_norm": 0.05298031121492386,
      "learning_rate": 2.578703703703704e-06,
      "loss": 0.0092,
      "step": 7526
    },
    {
      "epoch": 0.8711805555555555,
      "grad_norm": 0.03301694244146347,
      "learning_rate": 2.576388888888889e-06,
      "loss": 0.006,
      "step": 7527
    },
    {
      "epoch": 0.8712962962962963,
      "grad_norm": 166.2525177001953,
      "learning_rate": 2.5740740740740745e-06,
      "loss": 2.0809,
      "step": 7528
    },
    {
      "epoch": 0.8714120370370371,
      "grad_norm": 0.05121699348092079,
      "learning_rate": 2.5717592592592592e-06,
      "loss": 0.0089,
      "step": 7529
    },
    {
      "epoch": 0.8715277777777778,
      "grad_norm": 67.4268569946289,
      "learning_rate": 2.5694444444444443e-06,
      "loss": 1.9583,
      "step": 7530
    },
    {
      "epoch": 0.8716435185185185,
      "grad_norm": 111.7411880493164,
      "learning_rate": 2.56712962962963e-06,
      "loss": 1.9218,
      "step": 7531
    },
    {
      "epoch": 0.8717592592592592,
      "grad_norm": 0.04589506983757019,
      "learning_rate": 2.564814814814815e-06,
      "loss": 0.0081,
      "step": 7532
    },
    {
      "epoch": 0.871875,
      "grad_norm": 0.04889281094074249,
      "learning_rate": 2.5625e-06,
      "loss": 0.0061,
      "step": 7533
    },
    {
      "epoch": 0.8719907407407408,
      "grad_norm": 0.04321230575442314,
      "learning_rate": 2.5601851851851856e-06,
      "loss": 0.0076,
      "step": 7534
    },
    {
      "epoch": 0.8721064814814815,
      "grad_norm": 0.034834202378988266,
      "learning_rate": 2.5578703703703706e-06,
      "loss": 0.0058,
      "step": 7535
    },
    {
      "epoch": 0.8722222222222222,
      "grad_norm": 0.04505719244480133,
      "learning_rate": 2.5555555555555557e-06,
      "loss": 0.0077,
      "step": 7536
    },
    {
      "epoch": 0.872337962962963,
      "grad_norm": 0.09565893560647964,
      "learning_rate": 2.5532407407407413e-06,
      "loss": 0.0089,
      "step": 7537
    },
    {
      "epoch": 0.8724537037037037,
      "grad_norm": 1.7073312997817993,
      "learning_rate": 2.550925925925926e-06,
      "loss": 0.0152,
      "step": 7538
    },
    {
      "epoch": 0.8725694444444444,
      "grad_norm": 0.04418086260557175,
      "learning_rate": 2.548611111111111e-06,
      "loss": 0.008,
      "step": 7539
    },
    {
      "epoch": 0.8726851851851852,
      "grad_norm": 1.1543694734573364,
      "learning_rate": 2.5462962962962966e-06,
      "loss": 0.0165,
      "step": 7540
    },
    {
      "epoch": 0.872800925925926,
      "grad_norm": 0.03684375062584877,
      "learning_rate": 2.5439814814814817e-06,
      "loss": 0.0066,
      "step": 7541
    },
    {
      "epoch": 0.8729166666666667,
      "grad_norm": 0.03614678606390953,
      "learning_rate": 2.5416666666666668e-06,
      "loss": 0.0062,
      "step": 7542
    },
    {
      "epoch": 0.8730324074074074,
      "grad_norm": 355.8469543457031,
      "learning_rate": 2.5393518518518523e-06,
      "loss": 0.6649,
      "step": 7543
    },
    {
      "epoch": 0.8731481481481481,
      "grad_norm": 9.333847045898438,
      "learning_rate": 2.5370370370370374e-06,
      "loss": 3.0526,
      "step": 7544
    },
    {
      "epoch": 0.8732638888888888,
      "grad_norm": 0.05780456215143204,
      "learning_rate": 2.5347222222222225e-06,
      "loss": 0.0092,
      "step": 7545
    },
    {
      "epoch": 0.8733796296296297,
      "grad_norm": 0.053904782980680466,
      "learning_rate": 2.532407407407408e-06,
      "loss": 0.0084,
      "step": 7546
    },
    {
      "epoch": 0.8734953703703704,
      "grad_norm": 0.04030265659093857,
      "learning_rate": 2.530092592592593e-06,
      "loss": 0.0073,
      "step": 7547
    },
    {
      "epoch": 0.8736111111111111,
      "grad_norm": 0.0493585467338562,
      "learning_rate": 2.5277777777777778e-06,
      "loss": 0.0061,
      "step": 7548
    },
    {
      "epoch": 0.8737268518518518,
      "grad_norm": 0.052590589970350266,
      "learning_rate": 2.525462962962963e-06,
      "loss": 0.0091,
      "step": 7549
    },
    {
      "epoch": 0.8738425925925926,
      "grad_norm": 0.07644769549369812,
      "learning_rate": 2.5231481481481484e-06,
      "loss": 0.0088,
      "step": 7550
    },
    {
      "epoch": 0.8739583333333333,
      "grad_norm": 0.045839376747608185,
      "learning_rate": 2.5208333333333335e-06,
      "loss": 0.0081,
      "step": 7551
    },
    {
      "epoch": 0.8740740740740741,
      "grad_norm": 0.05218082293868065,
      "learning_rate": 2.5185185185185186e-06,
      "loss": 0.0068,
      "step": 7552
    },
    {
      "epoch": 0.8741898148148148,
      "grad_norm": 0.04603925347328186,
      "learning_rate": 2.516203703703704e-06,
      "loss": 0.0077,
      "step": 7553
    },
    {
      "epoch": 0.8743055555555556,
      "grad_norm": 0.05280503258109093,
      "learning_rate": 2.513888888888889e-06,
      "loss": 0.0088,
      "step": 7554
    },
    {
      "epoch": 0.8744212962962963,
      "grad_norm": 1.2263998985290527,
      "learning_rate": 2.511574074074074e-06,
      "loss": 0.0149,
      "step": 7555
    },
    {
      "epoch": 0.874537037037037,
      "grad_norm": 0.044494569301605225,
      "learning_rate": 2.50925925925926e-06,
      "loss": 0.008,
      "step": 7556
    },
    {
      "epoch": 0.8746527777777777,
      "grad_norm": 0.0429067462682724,
      "learning_rate": 2.5069444444444445e-06,
      "loss": 0.0076,
      "step": 7557
    },
    {
      "epoch": 0.8747685185185186,
      "grad_norm": 0.045177243649959564,
      "learning_rate": 2.5046296296296296e-06,
      "loss": 0.0077,
      "step": 7558
    },
    {
      "epoch": 0.8748842592592593,
      "grad_norm": 9.99702262878418,
      "learning_rate": 2.502314814814815e-06,
      "loss": 3.294,
      "step": 7559
    },
    {
      "epoch": 0.875,
      "grad_norm": 0.038460660725831985,
      "learning_rate": 2.5e-06,
      "loss": 0.0069,
      "step": 7560
    },
    {
      "epoch": 0.8751157407407407,
      "grad_norm": 0.09421678632497787,
      "learning_rate": 2.4976851851851853e-06,
      "loss": 0.009,
      "step": 7561
    },
    {
      "epoch": 0.8752314814814814,
      "grad_norm": 0.04245968163013458,
      "learning_rate": 2.4953703703703704e-06,
      "loss": 0.0074,
      "step": 7562
    },
    {
      "epoch": 0.8753472222222223,
      "grad_norm": 0.03802905231714249,
      "learning_rate": 2.493055555555556e-06,
      "loss": 0.0064,
      "step": 7563
    },
    {
      "epoch": 0.875462962962963,
      "grad_norm": 0.29878973960876465,
      "learning_rate": 2.490740740740741e-06,
      "loss": 0.012,
      "step": 7564
    },
    {
      "epoch": 0.8755787037037037,
      "grad_norm": 0.042763274163007736,
      "learning_rate": 2.488425925925926e-06,
      "loss": 0.0076,
      "step": 7565
    },
    {
      "epoch": 0.8756944444444444,
      "grad_norm": 35.19607162475586,
      "learning_rate": 2.4861111111111112e-06,
      "loss": 2.8286,
      "step": 7566
    },
    {
      "epoch": 0.8758101851851852,
      "grad_norm": 0.03985799476504326,
      "learning_rate": 2.4837962962962963e-06,
      "loss": 0.0071,
      "step": 7567
    },
    {
      "epoch": 0.8759259259259259,
      "grad_norm": 0.04026192054152489,
      "learning_rate": 2.481481481481482e-06,
      "loss": 0.0073,
      "step": 7568
    },
    {
      "epoch": 0.8760416666666667,
      "grad_norm": 0.059342481195926666,
      "learning_rate": 2.479166666666667e-06,
      "loss": 0.0108,
      "step": 7569
    },
    {
      "epoch": 0.8761574074074074,
      "grad_norm": 9.948508262634277,
      "learning_rate": 2.476851851851852e-06,
      "loss": 3.0548,
      "step": 7570
    },
    {
      "epoch": 0.8762731481481482,
      "grad_norm": 0.059820763766765594,
      "learning_rate": 2.474537037037037e-06,
      "loss": 0.0094,
      "step": 7571
    },
    {
      "epoch": 0.8763888888888889,
      "grad_norm": 0.0963149145245552,
      "learning_rate": 2.4722222222222226e-06,
      "loss": 0.0093,
      "step": 7572
    },
    {
      "epoch": 0.8765046296296296,
      "grad_norm": 0.19277088344097137,
      "learning_rate": 2.4699074074074073e-06,
      "loss": 0.0105,
      "step": 7573
    },
    {
      "epoch": 0.8766203703703703,
      "grad_norm": 0.03338410332798958,
      "learning_rate": 2.467592592592593e-06,
      "loss": 0.0059,
      "step": 7574
    },
    {
      "epoch": 0.8767361111111112,
      "grad_norm": 0.03995607793331146,
      "learning_rate": 2.465277777777778e-06,
      "loss": 0.0065,
      "step": 7575
    },
    {
      "epoch": 0.8768518518518519,
      "grad_norm": 0.04089367017149925,
      "learning_rate": 2.462962962962963e-06,
      "loss": 0.0072,
      "step": 7576
    },
    {
      "epoch": 0.8769675925925926,
      "grad_norm": 0.03908724710345268,
      "learning_rate": 2.4606481481481486e-06,
      "loss": 0.0067,
      "step": 7577
    },
    {
      "epoch": 0.8770833333333333,
      "grad_norm": 0.04396381974220276,
      "learning_rate": 2.4583333333333332e-06,
      "loss": 0.0079,
      "step": 7578
    },
    {
      "epoch": 0.877199074074074,
      "grad_norm": 0.03855201601982117,
      "learning_rate": 2.4560185185185188e-06,
      "loss": 0.0063,
      "step": 7579
    },
    {
      "epoch": 0.8773148148148148,
      "grad_norm": 0.05656144767999649,
      "learning_rate": 2.453703703703704e-06,
      "loss": 0.0098,
      "step": 7580
    },
    {
      "epoch": 0.8774305555555556,
      "grad_norm": 0.04192974045872688,
      "learning_rate": 2.451388888888889e-06,
      "loss": 0.0074,
      "step": 7581
    },
    {
      "epoch": 0.8775462962962963,
      "grad_norm": 0.06143917888402939,
      "learning_rate": 2.4490740740740745e-06,
      "loss": 0.0093,
      "step": 7582
    },
    {
      "epoch": 0.877662037037037,
      "grad_norm": 21.913375854492188,
      "learning_rate": 2.4467592592592596e-06,
      "loss": 0.0574,
      "step": 7583
    },
    {
      "epoch": 0.8777777777777778,
      "grad_norm": 0.05109883099794388,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.0067,
      "step": 7584
    },
    {
      "epoch": 0.8778935185185185,
      "grad_norm": 0.06914269179105759,
      "learning_rate": 2.4421296296296298e-06,
      "loss": 0.0111,
      "step": 7585
    },
    {
      "epoch": 0.8780092592592592,
      "grad_norm": 0.0549309141933918,
      "learning_rate": 2.4398148148148153e-06,
      "loss": 0.008,
      "step": 7586
    },
    {
      "epoch": 0.878125,
      "grad_norm": 0.13964606821537018,
      "learning_rate": 2.4375e-06,
      "loss": 0.01,
      "step": 7587
    },
    {
      "epoch": 0.8782407407407408,
      "grad_norm": 0.04605128988623619,
      "learning_rate": 2.4351851851851855e-06,
      "loss": 0.0076,
      "step": 7588
    },
    {
      "epoch": 0.8783564814814815,
      "grad_norm": 0.33355948328971863,
      "learning_rate": 2.4328703703703706e-06,
      "loss": 0.0082,
      "step": 7589
    },
    {
      "epoch": 0.8784722222222222,
      "grad_norm": 0.05074051767587662,
      "learning_rate": 2.4305555555555557e-06,
      "loss": 0.0087,
      "step": 7590
    },
    {
      "epoch": 0.8785879629629629,
      "grad_norm": 0.04903150349855423,
      "learning_rate": 2.428240740740741e-06,
      "loss": 0.0085,
      "step": 7591
    },
    {
      "epoch": 0.8787037037037037,
      "grad_norm": 0.06535635888576508,
      "learning_rate": 2.425925925925926e-06,
      "loss": 0.0087,
      "step": 7592
    },
    {
      "epoch": 0.8788194444444445,
      "grad_norm": 0.037063729017972946,
      "learning_rate": 2.4236111111111114e-06,
      "loss": 0.0067,
      "step": 7593
    },
    {
      "epoch": 0.8789351851851852,
      "grad_norm": 0.045648764818906784,
      "learning_rate": 2.4212962962962965e-06,
      "loss": 0.0083,
      "step": 7594
    },
    {
      "epoch": 0.8790509259259259,
      "grad_norm": 0.04076870530843735,
      "learning_rate": 2.4189814814814816e-06,
      "loss": 0.0072,
      "step": 7595
    },
    {
      "epoch": 0.8791666666666667,
      "grad_norm": 0.05239581689238548,
      "learning_rate": 2.4166666666666667e-06,
      "loss": 0.0091,
      "step": 7596
    },
    {
      "epoch": 0.8792824074074074,
      "grad_norm": 0.07034104317426682,
      "learning_rate": 2.414351851851852e-06,
      "loss": 0.0109,
      "step": 7597
    },
    {
      "epoch": 0.8793981481481481,
      "grad_norm": 0.05638258159160614,
      "learning_rate": 2.4120370370370373e-06,
      "loss": 0.0097,
      "step": 7598
    },
    {
      "epoch": 0.8795138888888889,
      "grad_norm": 0.04092481732368469,
      "learning_rate": 2.4097222222222224e-06,
      "loss": 0.0073,
      "step": 7599
    },
    {
      "epoch": 0.8796296296296297,
      "grad_norm": 0.041793495416641235,
      "learning_rate": 2.4074074074074075e-06,
      "loss": 0.0073,
      "step": 7600
    },
    {
      "epoch": 0.8797453703703704,
      "grad_norm": 0.05238577350974083,
      "learning_rate": 2.4050925925925926e-06,
      "loss": 0.0093,
      "step": 7601
    },
    {
      "epoch": 0.8798611111111111,
      "grad_norm": 0.03616182133555412,
      "learning_rate": 2.402777777777778e-06,
      "loss": 0.0063,
      "step": 7602
    },
    {
      "epoch": 0.8799768518518518,
      "grad_norm": 0.049182359129190445,
      "learning_rate": 2.400462962962963e-06,
      "loss": 0.0088,
      "step": 7603
    },
    {
      "epoch": 0.8800925925925925,
      "grad_norm": 0.03243998438119888,
      "learning_rate": 2.3981481481481483e-06,
      "loss": 0.0058,
      "step": 7604
    },
    {
      "epoch": 0.8802083333333334,
      "grad_norm": 0.04126739129424095,
      "learning_rate": 2.395833333333334e-06,
      "loss": 0.0064,
      "step": 7605
    },
    {
      "epoch": 0.8803240740740741,
      "grad_norm": 0.4120098650455475,
      "learning_rate": 2.3935185185185185e-06,
      "loss": 0.0105,
      "step": 7606
    },
    {
      "epoch": 0.8804398148148148,
      "grad_norm": 0.03320463374257088,
      "learning_rate": 2.391203703703704e-06,
      "loss": 0.0059,
      "step": 7607
    },
    {
      "epoch": 0.8805555555555555,
      "grad_norm": 0.04011211171746254,
      "learning_rate": 2.388888888888889e-06,
      "loss": 0.0072,
      "step": 7608
    },
    {
      "epoch": 0.8806712962962963,
      "grad_norm": 0.04900700971484184,
      "learning_rate": 2.3865740740740742e-06,
      "loss": 0.0087,
      "step": 7609
    },
    {
      "epoch": 0.8807870370370371,
      "grad_norm": 0.03678342327475548,
      "learning_rate": 2.3842592592592593e-06,
      "loss": 0.0067,
      "step": 7610
    },
    {
      "epoch": 0.8809027777777778,
      "grad_norm": 0.040809568017721176,
      "learning_rate": 2.3819444444444444e-06,
      "loss": 0.0053,
      "step": 7611
    },
    {
      "epoch": 0.8810185185185185,
      "grad_norm": 0.039707623422145844,
      "learning_rate": 2.37962962962963e-06,
      "loss": 0.0064,
      "step": 7612
    },
    {
      "epoch": 0.8811342592592593,
      "grad_norm": 0.051458872854709625,
      "learning_rate": 2.377314814814815e-06,
      "loss": 0.0088,
      "step": 7613
    },
    {
      "epoch": 0.88125,
      "grad_norm": 20.007648468017578,
      "learning_rate": 2.375e-06,
      "loss": 3.0458,
      "step": 7614
    },
    {
      "epoch": 0.8813657407407407,
      "grad_norm": 132.65567016601562,
      "learning_rate": 2.3726851851851852e-06,
      "loss": 0.5716,
      "step": 7615
    },
    {
      "epoch": 0.8814814814814815,
      "grad_norm": 0.05369499698281288,
      "learning_rate": 2.3703703703703707e-06,
      "loss": 0.0072,
      "step": 7616
    },
    {
      "epoch": 0.8815972222222223,
      "grad_norm": 0.03962525725364685,
      "learning_rate": 2.368055555555556e-06,
      "loss": 0.0072,
      "step": 7617
    },
    {
      "epoch": 0.881712962962963,
      "grad_norm": 0.039462894201278687,
      "learning_rate": 2.365740740740741e-06,
      "loss": 0.0072,
      "step": 7618
    },
    {
      "epoch": 0.8818287037037037,
      "grad_norm": 0.0477721244096756,
      "learning_rate": 2.363425925925926e-06,
      "loss": 0.0085,
      "step": 7619
    },
    {
      "epoch": 0.8819444444444444,
      "grad_norm": 0.043560903519392014,
      "learning_rate": 2.361111111111111e-06,
      "loss": 0.0079,
      "step": 7620
    },
    {
      "epoch": 0.8820601851851851,
      "grad_norm": 0.04068661481142044,
      "learning_rate": 2.3587962962962967e-06,
      "loss": 0.0072,
      "step": 7621
    },
    {
      "epoch": 0.882175925925926,
      "grad_norm": 0.1571948230266571,
      "learning_rate": 2.3564814814814813e-06,
      "loss": 0.0091,
      "step": 7622
    },
    {
      "epoch": 0.8822916666666667,
      "grad_norm": 0.05087598040699959,
      "learning_rate": 2.354166666666667e-06,
      "loss": 0.0086,
      "step": 7623
    },
    {
      "epoch": 0.8824074074074074,
      "grad_norm": 0.057947736233472824,
      "learning_rate": 2.351851851851852e-06,
      "loss": 0.0067,
      "step": 7624
    },
    {
      "epoch": 0.8825231481481481,
      "grad_norm": 0.03638170659542084,
      "learning_rate": 2.349537037037037e-06,
      "loss": 0.0065,
      "step": 7625
    },
    {
      "epoch": 0.8826388888888889,
      "grad_norm": 0.054392654448747635,
      "learning_rate": 2.3472222222222226e-06,
      "loss": 0.0082,
      "step": 7626
    },
    {
      "epoch": 0.8827546296296296,
      "grad_norm": 0.059831757098436356,
      "learning_rate": 2.3449074074074077e-06,
      "loss": 0.0085,
      "step": 7627
    },
    {
      "epoch": 0.8828703703703704,
      "grad_norm": 0.12058688700199127,
      "learning_rate": 2.3425925925925928e-06,
      "loss": 0.0112,
      "step": 7628
    },
    {
      "epoch": 0.8829861111111111,
      "grad_norm": 0.05178437754511833,
      "learning_rate": 2.340277777777778e-06,
      "loss": 0.0091,
      "step": 7629
    },
    {
      "epoch": 0.8831018518518519,
      "grad_norm": 0.043276168406009674,
      "learning_rate": 2.3379629629629634e-06,
      "loss": 0.0078,
      "step": 7630
    },
    {
      "epoch": 0.8832175925925926,
      "grad_norm": 0.04529291391372681,
      "learning_rate": 2.3356481481481485e-06,
      "loss": 0.0075,
      "step": 7631
    },
    {
      "epoch": 0.8833333333333333,
      "grad_norm": 0.05074283853173256,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.0079,
      "step": 7632
    },
    {
      "epoch": 0.883449074074074,
      "grad_norm": 0.0462041012942791,
      "learning_rate": 2.3310185185185187e-06,
      "loss": 0.0083,
      "step": 7633
    },
    {
      "epoch": 0.8835648148148149,
      "grad_norm": 0.06978558003902435,
      "learning_rate": 2.3287037037037038e-06,
      "loss": 0.0091,
      "step": 7634
    },
    {
      "epoch": 0.8836805555555556,
      "grad_norm": 0.06182139366865158,
      "learning_rate": 2.3263888888888893e-06,
      "loss": 0.0098,
      "step": 7635
    },
    {
      "epoch": 0.8837962962962963,
      "grad_norm": 0.041280876845121384,
      "learning_rate": 2.324074074074074e-06,
      "loss": 0.0071,
      "step": 7636
    },
    {
      "epoch": 0.883912037037037,
      "grad_norm": 0.04544208571314812,
      "learning_rate": 2.3217592592592595e-06,
      "loss": 0.0079,
      "step": 7637
    },
    {
      "epoch": 0.8840277777777777,
      "grad_norm": 0.07393404096364975,
      "learning_rate": 2.3194444444444446e-06,
      "loss": 0.0114,
      "step": 7638
    },
    {
      "epoch": 0.8841435185185185,
      "grad_norm": 0.04634932801127434,
      "learning_rate": 2.3171296296296297e-06,
      "loss": 0.0082,
      "step": 7639
    },
    {
      "epoch": 0.8842592592592593,
      "grad_norm": 0.04872024059295654,
      "learning_rate": 2.314814814814815e-06,
      "loss": 0.0068,
      "step": 7640
    },
    {
      "epoch": 0.884375,
      "grad_norm": 13.972636222839355,
      "learning_rate": 2.3125000000000003e-06,
      "loss": 3.1227,
      "step": 7641
    },
    {
      "epoch": 0.8844907407407407,
      "grad_norm": 0.030333060771226883,
      "learning_rate": 2.3101851851851854e-06,
      "loss": 0.0055,
      "step": 7642
    },
    {
      "epoch": 0.8846064814814815,
      "grad_norm": 0.07653740048408508,
      "learning_rate": 2.3078703703703705e-06,
      "loss": 0.0077,
      "step": 7643
    },
    {
      "epoch": 0.8847222222222222,
      "grad_norm": 0.13864201307296753,
      "learning_rate": 2.305555555555556e-06,
      "loss": 0.0113,
      "step": 7644
    },
    {
      "epoch": 0.8848379629629629,
      "grad_norm": 0.0721459910273552,
      "learning_rate": 2.3032407407407407e-06,
      "loss": 0.0082,
      "step": 7645
    },
    {
      "epoch": 0.8849537037037037,
      "grad_norm": 0.05802810192108154,
      "learning_rate": 2.3009259259259262e-06,
      "loss": 0.0105,
      "step": 7646
    },
    {
      "epoch": 0.8850694444444445,
      "grad_norm": 0.03441809490323067,
      "learning_rate": 2.2986111111111113e-06,
      "loss": 0.0059,
      "step": 7647
    },
    {
      "epoch": 0.8851851851851852,
      "grad_norm": 0.05111556127667427,
      "learning_rate": 2.2962962962962964e-06,
      "loss": 0.0089,
      "step": 7648
    },
    {
      "epoch": 0.8853009259259259,
      "grad_norm": 0.03518597409129143,
      "learning_rate": 2.293981481481482e-06,
      "loss": 0.0064,
      "step": 7649
    },
    {
      "epoch": 0.8854166666666666,
      "grad_norm": 0.06827805191278458,
      "learning_rate": 2.2916666666666666e-06,
      "loss": 0.0106,
      "step": 7650
    },
    {
      "epoch": 0.8855324074074075,
      "grad_norm": 0.05442047864198685,
      "learning_rate": 2.289351851851852e-06,
      "loss": 0.0084,
      "step": 7651
    },
    {
      "epoch": 0.8856481481481482,
      "grad_norm": 0.05029755458235741,
      "learning_rate": 2.2870370370370372e-06,
      "loss": 0.0088,
      "step": 7652
    },
    {
      "epoch": 0.8857638888888889,
      "grad_norm": 0.03666740655899048,
      "learning_rate": 2.2847222222222223e-06,
      "loss": 0.0065,
      "step": 7653
    },
    {
      "epoch": 0.8858796296296296,
      "grad_norm": 0.037128008902072906,
      "learning_rate": 2.2824074074074074e-06,
      "loss": 0.0066,
      "step": 7654
    },
    {
      "epoch": 0.8859953703703703,
      "grad_norm": 0.8195858001708984,
      "learning_rate": 2.280092592592593e-06,
      "loss": 0.0148,
      "step": 7655
    },
    {
      "epoch": 0.8861111111111111,
      "grad_norm": 0.04607964679598808,
      "learning_rate": 2.277777777777778e-06,
      "loss": 0.0084,
      "step": 7656
    },
    {
      "epoch": 0.8862268518518519,
      "grad_norm": 0.036098286509513855,
      "learning_rate": 2.275462962962963e-06,
      "loss": 0.0066,
      "step": 7657
    },
    {
      "epoch": 0.8863425925925926,
      "grad_norm": 40.324649810791016,
      "learning_rate": 2.2731481481481482e-06,
      "loss": 2.5108,
      "step": 7658
    },
    {
      "epoch": 0.8864583333333333,
      "grad_norm": 0.08606165647506714,
      "learning_rate": 2.2708333333333333e-06,
      "loss": 0.0071,
      "step": 7659
    },
    {
      "epoch": 0.8865740740740741,
      "grad_norm": 0.06582268327474594,
      "learning_rate": 2.268518518518519e-06,
      "loss": 0.0114,
      "step": 7660
    },
    {
      "epoch": 0.8866898148148148,
      "grad_norm": 0.052275944501161575,
      "learning_rate": 2.266203703703704e-06,
      "loss": 0.0071,
      "step": 7661
    },
    {
      "epoch": 0.8868055555555555,
      "grad_norm": 0.039189308881759644,
      "learning_rate": 2.263888888888889e-06,
      "loss": 0.007,
      "step": 7662
    },
    {
      "epoch": 0.8869212962962963,
      "grad_norm": 8.064152717590332,
      "learning_rate": 2.2615740740740746e-06,
      "loss": 0.0456,
      "step": 7663
    },
    {
      "epoch": 0.8870370370370371,
      "grad_norm": 0.04879142343997955,
      "learning_rate": 2.2592592592592592e-06,
      "loss": 0.0062,
      "step": 7664
    },
    {
      "epoch": 0.8871527777777778,
      "grad_norm": 0.03491196036338806,
      "learning_rate": 2.2569444444444448e-06,
      "loss": 0.0063,
      "step": 7665
    },
    {
      "epoch": 0.8872685185185185,
      "grad_norm": 0.053039662539958954,
      "learning_rate": 2.25462962962963e-06,
      "loss": 0.0095,
      "step": 7666
    },
    {
      "epoch": 0.8873842592592592,
      "grad_norm": 0.03460662066936493,
      "learning_rate": 2.252314814814815e-06,
      "loss": 0.0062,
      "step": 7667
    },
    {
      "epoch": 0.8875,
      "grad_norm": 120.58345031738281,
      "learning_rate": 2.25e-06,
      "loss": 0.4289,
      "step": 7668
    },
    {
      "epoch": 0.8876157407407408,
      "grad_norm": 0.04706849157810211,
      "learning_rate": 2.247685185185185e-06,
      "loss": 0.0082,
      "step": 7669
    },
    {
      "epoch": 0.8877314814814815,
      "grad_norm": 0.03652266785502434,
      "learning_rate": 2.2453703703703707e-06,
      "loss": 0.0064,
      "step": 7670
    },
    {
      "epoch": 0.8878472222222222,
      "grad_norm": 0.3236945867538452,
      "learning_rate": 2.2430555555555558e-06,
      "loss": 0.0093,
      "step": 7671
    },
    {
      "epoch": 0.887962962962963,
      "grad_norm": 7.113489627838135,
      "learning_rate": 2.240740740740741e-06,
      "loss": 0.0347,
      "step": 7672
    },
    {
      "epoch": 0.8880787037037037,
      "grad_norm": 0.03772803395986557,
      "learning_rate": 2.238425925925926e-06,
      "loss": 0.0067,
      "step": 7673
    },
    {
      "epoch": 0.8881944444444444,
      "grad_norm": 0.044753365218639374,
      "learning_rate": 2.2361111111111115e-06,
      "loss": 0.0077,
      "step": 7674
    },
    {
      "epoch": 0.8883101851851852,
      "grad_norm": 0.035560112446546555,
      "learning_rate": 2.2337962962962966e-06,
      "loss": 0.0064,
      "step": 7675
    },
    {
      "epoch": 0.888425925925926,
      "grad_norm": 0.04020658880472183,
      "learning_rate": 2.2314814814814817e-06,
      "loss": 0.0072,
      "step": 7676
    },
    {
      "epoch": 0.8885416666666667,
      "grad_norm": 0.11570808291435242,
      "learning_rate": 2.2291666666666668e-06,
      "loss": 0.0123,
      "step": 7677
    },
    {
      "epoch": 0.8886574074074074,
      "grad_norm": 0.037638984620571136,
      "learning_rate": 2.226851851851852e-06,
      "loss": 0.0067,
      "step": 7678
    },
    {
      "epoch": 0.8887731481481481,
      "grad_norm": 0.05404683202505112,
      "learning_rate": 2.2245370370370374e-06,
      "loss": 0.008,
      "step": 7679
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 0.038931477814912796,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.007,
      "step": 7680
    },
    {
      "epoch": 0.8890046296296297,
      "grad_norm": 0.04131060093641281,
      "learning_rate": 2.2199074074074076e-06,
      "loss": 0.0071,
      "step": 7681
    },
    {
      "epoch": 0.8891203703703704,
      "grad_norm": 0.0319422110915184,
      "learning_rate": 2.2175925925925927e-06,
      "loss": 0.0056,
      "step": 7682
    },
    {
      "epoch": 0.8892361111111111,
      "grad_norm": 0.10566605627536774,
      "learning_rate": 2.215277777777778e-06,
      "loss": 0.0106,
      "step": 7683
    },
    {
      "epoch": 0.8893518518518518,
      "grad_norm": 0.6868130564689636,
      "learning_rate": 2.2129629629629633e-06,
      "loss": 0.0142,
      "step": 7684
    },
    {
      "epoch": 0.8894675925925926,
      "grad_norm": 0.04686631262302399,
      "learning_rate": 2.2106481481481484e-06,
      "loss": 0.0082,
      "step": 7685
    },
    {
      "epoch": 0.8895833333333333,
      "grad_norm": 0.04802665114402771,
      "learning_rate": 2.2083333333333335e-06,
      "loss": 0.0086,
      "step": 7686
    },
    {
      "epoch": 0.8896990740740741,
      "grad_norm": 0.0320076048374176,
      "learning_rate": 2.2060185185185186e-06,
      "loss": 0.0058,
      "step": 7687
    },
    {
      "epoch": 0.8898148148148148,
      "grad_norm": 0.03704237565398216,
      "learning_rate": 2.203703703703704e-06,
      "loss": 0.0066,
      "step": 7688
    },
    {
      "epoch": 0.8899305555555556,
      "grad_norm": 0.04213671386241913,
      "learning_rate": 2.2013888888888892e-06,
      "loss": 0.0075,
      "step": 7689
    },
    {
      "epoch": 0.8900462962962963,
      "grad_norm": 0.047963663935661316,
      "learning_rate": 2.1990740740740743e-06,
      "loss": 0.0082,
      "step": 7690
    },
    {
      "epoch": 0.890162037037037,
      "grad_norm": 0.041053347289562225,
      "learning_rate": 2.1967592592592594e-06,
      "loss": 0.0072,
      "step": 7691
    },
    {
      "epoch": 0.8902777777777777,
      "grad_norm": 0.033587969839572906,
      "learning_rate": 2.1944444444444445e-06,
      "loss": 0.0061,
      "step": 7692
    },
    {
      "epoch": 0.8903935185185186,
      "grad_norm": 0.03274274989962578,
      "learning_rate": 2.19212962962963e-06,
      "loss": 0.0059,
      "step": 7693
    },
    {
      "epoch": 0.8905092592592593,
      "grad_norm": 0.04355912283062935,
      "learning_rate": 2.1898148148148147e-06,
      "loss": 0.0079,
      "step": 7694
    },
    {
      "epoch": 0.890625,
      "grad_norm": 0.04353741928935051,
      "learning_rate": 2.1875000000000002e-06,
      "loss": 0.0076,
      "step": 7695
    },
    {
      "epoch": 0.8907407407407407,
      "grad_norm": 0.03807017579674721,
      "learning_rate": 2.1851851851851853e-06,
      "loss": 0.0066,
      "step": 7696
    },
    {
      "epoch": 0.8908564814814814,
      "grad_norm": 0.04194498062133789,
      "learning_rate": 2.1828703703703704e-06,
      "loss": 0.0074,
      "step": 7697
    },
    {
      "epoch": 0.8909722222222223,
      "grad_norm": 0.05211224406957626,
      "learning_rate": 2.180555555555556e-06,
      "loss": 0.0091,
      "step": 7698
    },
    {
      "epoch": 0.891087962962963,
      "grad_norm": 0.0754937008023262,
      "learning_rate": 2.178240740740741e-06,
      "loss": 0.0098,
      "step": 7699
    },
    {
      "epoch": 0.8912037037037037,
      "grad_norm": 0.03672119602560997,
      "learning_rate": 2.175925925925926e-06,
      "loss": 0.0065,
      "step": 7700
    },
    {
      "epoch": 0.8913194444444444,
      "grad_norm": 0.03529078885912895,
      "learning_rate": 2.1736111111111112e-06,
      "loss": 0.0061,
      "step": 7701
    },
    {
      "epoch": 0.8914351851851852,
      "grad_norm": 0.07672569155693054,
      "learning_rate": 2.1712962962962963e-06,
      "loss": 0.0094,
      "step": 7702
    },
    {
      "epoch": 0.8915509259259259,
      "grad_norm": 0.031979482620954514,
      "learning_rate": 2.1689814814814814e-06,
      "loss": 0.0057,
      "step": 7703
    },
    {
      "epoch": 0.8916666666666667,
      "grad_norm": 0.04721665009856224,
      "learning_rate": 2.166666666666667e-06,
      "loss": 0.0083,
      "step": 7704
    },
    {
      "epoch": 0.8917824074074074,
      "grad_norm": 0.046010822057724,
      "learning_rate": 2.164351851851852e-06,
      "loss": 0.0081,
      "step": 7705
    },
    {
      "epoch": 0.8918981481481482,
      "grad_norm": 3.5533511638641357,
      "learning_rate": 2.162037037037037e-06,
      "loss": 0.0236,
      "step": 7706
    },
    {
      "epoch": 0.8920138888888889,
      "grad_norm": 0.033609602600336075,
      "learning_rate": 2.1597222222222227e-06,
      "loss": 0.0061,
      "step": 7707
    },
    {
      "epoch": 0.8921296296296296,
      "grad_norm": 0.03370700031518936,
      "learning_rate": 2.1574074074074073e-06,
      "loss": 0.0061,
      "step": 7708
    },
    {
      "epoch": 0.8922453703703703,
      "grad_norm": 0.04191838204860687,
      "learning_rate": 2.155092592592593e-06,
      "loss": 0.0073,
      "step": 7709
    },
    {
      "epoch": 0.8923611111111112,
      "grad_norm": 0.07304082065820694,
      "learning_rate": 2.152777777777778e-06,
      "loss": 0.0067,
      "step": 7710
    },
    {
      "epoch": 0.8924768518518519,
      "grad_norm": 0.04013567790389061,
      "learning_rate": 2.150462962962963e-06,
      "loss": 0.0071,
      "step": 7711
    },
    {
      "epoch": 0.8925925925925926,
      "grad_norm": 0.056725163012742996,
      "learning_rate": 2.148148148148148e-06,
      "loss": 0.0103,
      "step": 7712
    },
    {
      "epoch": 0.8927083333333333,
      "grad_norm": 0.03595554083585739,
      "learning_rate": 2.1458333333333333e-06,
      "loss": 0.0065,
      "step": 7713
    },
    {
      "epoch": 0.892824074074074,
      "grad_norm": 0.037887636572122574,
      "learning_rate": 2.1435185185185188e-06,
      "loss": 0.0068,
      "step": 7714
    },
    {
      "epoch": 0.8929398148148148,
      "grad_norm": 0.051685355603694916,
      "learning_rate": 2.141203703703704e-06,
      "loss": 0.0083,
      "step": 7715
    },
    {
      "epoch": 0.8930555555555556,
      "grad_norm": 0.05343610420823097,
      "learning_rate": 2.138888888888889e-06,
      "loss": 0.0089,
      "step": 7716
    },
    {
      "epoch": 0.8931712962962963,
      "grad_norm": 0.050687119364738464,
      "learning_rate": 2.136574074074074e-06,
      "loss": 0.0088,
      "step": 7717
    },
    {
      "epoch": 0.893287037037037,
      "grad_norm": 0.045876745134592056,
      "learning_rate": 2.1342592592592596e-06,
      "loss": 0.0081,
      "step": 7718
    },
    {
      "epoch": 0.8934027777777778,
      "grad_norm": 0.04573351889848709,
      "learning_rate": 2.1319444444444447e-06,
      "loss": 0.0079,
      "step": 7719
    },
    {
      "epoch": 0.8935185185185185,
      "grad_norm": 0.04219011962413788,
      "learning_rate": 2.1296296296296298e-06,
      "loss": 0.0075,
      "step": 7720
    },
    {
      "epoch": 0.8936342592592592,
      "grad_norm": 0.05123169347643852,
      "learning_rate": 2.1273148148148153e-06,
      "loss": 0.0093,
      "step": 7721
    },
    {
      "epoch": 0.89375,
      "grad_norm": 0.048820845782756805,
      "learning_rate": 2.125e-06,
      "loss": 0.0084,
      "step": 7722
    },
    {
      "epoch": 0.8938657407407408,
      "grad_norm": 0.029864724725484848,
      "learning_rate": 2.1226851851851855e-06,
      "loss": 0.0054,
      "step": 7723
    },
    {
      "epoch": 0.8939814814814815,
      "grad_norm": 0.0292192455381155,
      "learning_rate": 2.1203703703703706e-06,
      "loss": 0.0053,
      "step": 7724
    },
    {
      "epoch": 0.8940972222222222,
      "grad_norm": 0.040744200348854065,
      "learning_rate": 2.1180555555555557e-06,
      "loss": 0.0072,
      "step": 7725
    },
    {
      "epoch": 0.8942129629629629,
      "grad_norm": 0.04874954745173454,
      "learning_rate": 2.115740740740741e-06,
      "loss": 0.0062,
      "step": 7726
    },
    {
      "epoch": 0.8943287037037037,
      "grad_norm": 0.4200775921344757,
      "learning_rate": 2.113425925925926e-06,
      "loss": 0.014,
      "step": 7727
    },
    {
      "epoch": 0.8944444444444445,
      "grad_norm": 0.0473468042910099,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.0082,
      "step": 7728
    },
    {
      "epoch": 0.8945601851851852,
      "grad_norm": 74.05290222167969,
      "learning_rate": 2.1087962962962965e-06,
      "loss": 2.6069,
      "step": 7729
    },
    {
      "epoch": 0.8946759259259259,
      "grad_norm": 0.038149766623973846,
      "learning_rate": 2.1064814814814816e-06,
      "loss": 0.0066,
      "step": 7730
    },
    {
      "epoch": 0.8947916666666667,
      "grad_norm": 0.05053050443530083,
      "learning_rate": 2.1041666666666667e-06,
      "loss": 0.0078,
      "step": 7731
    },
    {
      "epoch": 0.8949074074074074,
      "grad_norm": 0.03362428396940231,
      "learning_rate": 2.1018518518518522e-06,
      "loss": 0.0061,
      "step": 7732
    },
    {
      "epoch": 0.8950231481481481,
      "grad_norm": 0.03748869523406029,
      "learning_rate": 2.0995370370370373e-06,
      "loss": 0.0067,
      "step": 7733
    },
    {
      "epoch": 0.8951388888888889,
      "grad_norm": 0.037995584309101105,
      "learning_rate": 2.0972222222222224e-06,
      "loss": 0.0069,
      "step": 7734
    },
    {
      "epoch": 0.8952546296296297,
      "grad_norm": 0.06076602637767792,
      "learning_rate": 2.0949074074074075e-06,
      "loss": 0.0109,
      "step": 7735
    },
    {
      "epoch": 0.8953703703703704,
      "grad_norm": 7.6808342933654785,
      "learning_rate": 2.0925925925925926e-06,
      "loss": 3.2309,
      "step": 7736
    },
    {
      "epoch": 0.8954861111111111,
      "grad_norm": 0.04324036464095116,
      "learning_rate": 2.090277777777778e-06,
      "loss": 0.0074,
      "step": 7737
    },
    {
      "epoch": 0.8956018518518518,
      "grad_norm": 0.04540887102484703,
      "learning_rate": 2.087962962962963e-06,
      "loss": 0.0081,
      "step": 7738
    },
    {
      "epoch": 0.8957175925925925,
      "grad_norm": 13.50031852722168,
      "learning_rate": 2.0856481481481483e-06,
      "loss": 2.9197,
      "step": 7739
    },
    {
      "epoch": 0.8958333333333334,
      "grad_norm": 0.0465724878013134,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.0079,
      "step": 7740
    },
    {
      "epoch": 0.8959490740740741,
      "grad_norm": 0.05913109704852104,
      "learning_rate": 2.0810185185185185e-06,
      "loss": 0.0075,
      "step": 7741
    },
    {
      "epoch": 0.8960648148148148,
      "grad_norm": 0.05147572234272957,
      "learning_rate": 2.078703703703704e-06,
      "loss": 0.0094,
      "step": 7742
    },
    {
      "epoch": 0.8961805555555555,
      "grad_norm": 0.04217379912734032,
      "learning_rate": 2.076388888888889e-06,
      "loss": 0.0075,
      "step": 7743
    },
    {
      "epoch": 0.8962962962962963,
      "grad_norm": 0.18134522438049316,
      "learning_rate": 2.0740740740740742e-06,
      "loss": 0.0123,
      "step": 7744
    },
    {
      "epoch": 0.8964120370370371,
      "grad_norm": 0.040543247014284134,
      "learning_rate": 2.0717592592592593e-06,
      "loss": 0.0071,
      "step": 7745
    },
    {
      "epoch": 0.8965277777777778,
      "grad_norm": 0.04691831022500992,
      "learning_rate": 2.0694444444444444e-06,
      "loss": 0.0083,
      "step": 7746
    },
    {
      "epoch": 0.8966435185185185,
      "grad_norm": 0.04583580046892166,
      "learning_rate": 2.06712962962963e-06,
      "loss": 0.0081,
      "step": 7747
    },
    {
      "epoch": 0.8967592592592593,
      "grad_norm": 0.05471804738044739,
      "learning_rate": 2.064814814814815e-06,
      "loss": 0.0088,
      "step": 7748
    },
    {
      "epoch": 0.896875,
      "grad_norm": 0.03292204067111015,
      "learning_rate": 2.0625e-06,
      "loss": 0.0059,
      "step": 7749
    },
    {
      "epoch": 0.8969907407407407,
      "grad_norm": 0.05831136181950569,
      "learning_rate": 2.0601851851851853e-06,
      "loss": 0.0095,
      "step": 7750
    },
    {
      "epoch": 0.8971064814814815,
      "grad_norm": 0.04911952465772629,
      "learning_rate": 2.0578703703703708e-06,
      "loss": 0.0084,
      "step": 7751
    },
    {
      "epoch": 0.8972222222222223,
      "grad_norm": 0.029862787574529648,
      "learning_rate": 2.0555555555555555e-06,
      "loss": 0.0054,
      "step": 7752
    },
    {
      "epoch": 0.897337962962963,
      "grad_norm": 0.050804171711206436,
      "learning_rate": 2.053240740740741e-06,
      "loss": 0.0089,
      "step": 7753
    },
    {
      "epoch": 0.8974537037037037,
      "grad_norm": 0.03674919158220291,
      "learning_rate": 2.050925925925926e-06,
      "loss": 0.0061,
      "step": 7754
    },
    {
      "epoch": 0.8975694444444444,
      "grad_norm": 0.04604983329772949,
      "learning_rate": 2.048611111111111e-06,
      "loss": 0.0082,
      "step": 7755
    },
    {
      "epoch": 0.8976851851851851,
      "grad_norm": 0.05429510399699211,
      "learning_rate": 2.0462962962962967e-06,
      "loss": 0.0085,
      "step": 7756
    },
    {
      "epoch": 0.897800925925926,
      "grad_norm": 0.03584929183125496,
      "learning_rate": 2.0439814814814814e-06,
      "loss": 0.0065,
      "step": 7757
    },
    {
      "epoch": 0.8979166666666667,
      "grad_norm": 0.04501045495271683,
      "learning_rate": 2.041666666666667e-06,
      "loss": 0.0066,
      "step": 7758
    },
    {
      "epoch": 0.8980324074074074,
      "grad_norm": 0.0427616648375988,
      "learning_rate": 2.039351851851852e-06,
      "loss": 0.0077,
      "step": 7759
    },
    {
      "epoch": 0.8981481481481481,
      "grad_norm": 0.03482702374458313,
      "learning_rate": 2.037037037037037e-06,
      "loss": 0.0061,
      "step": 7760
    },
    {
      "epoch": 0.8982638888888889,
      "grad_norm": 0.044852934777736664,
      "learning_rate": 2.034722222222222e-06,
      "loss": 0.008,
      "step": 7761
    },
    {
      "epoch": 0.8983796296296296,
      "grad_norm": 0.04280447959899902,
      "learning_rate": 2.0324074074074077e-06,
      "loss": 0.0077,
      "step": 7762
    },
    {
      "epoch": 0.8984953703703704,
      "grad_norm": 0.04399183392524719,
      "learning_rate": 2.030092592592593e-06,
      "loss": 0.0076,
      "step": 7763
    },
    {
      "epoch": 0.8986111111111111,
      "grad_norm": 0.0335196778178215,
      "learning_rate": 2.027777777777778e-06,
      "loss": 0.006,
      "step": 7764
    },
    {
      "epoch": 0.8987268518518519,
      "grad_norm": 0.07056239247322083,
      "learning_rate": 2.0254629629629634e-06,
      "loss": 0.0094,
      "step": 7765
    },
    {
      "epoch": 0.8988425925925926,
      "grad_norm": 0.05961861461400986,
      "learning_rate": 2.023148148148148e-06,
      "loss": 0.0077,
      "step": 7766
    },
    {
      "epoch": 0.8989583333333333,
      "grad_norm": 0.035505227744579315,
      "learning_rate": 2.0208333333333336e-06,
      "loss": 0.0064,
      "step": 7767
    },
    {
      "epoch": 0.899074074074074,
      "grad_norm": 0.07608389854431152,
      "learning_rate": 2.0185185185185187e-06,
      "loss": 0.0086,
      "step": 7768
    },
    {
      "epoch": 0.8991898148148149,
      "grad_norm": 0.04414580389857292,
      "learning_rate": 2.016203703703704e-06,
      "loss": 0.0076,
      "step": 7769
    },
    {
      "epoch": 0.8993055555555556,
      "grad_norm": 0.0501890629529953,
      "learning_rate": 2.0138888888888893e-06,
      "loss": 0.0085,
      "step": 7770
    },
    {
      "epoch": 0.8994212962962963,
      "grad_norm": 0.07716409862041473,
      "learning_rate": 2.011574074074074e-06,
      "loss": 0.0098,
      "step": 7771
    },
    {
      "epoch": 0.899537037037037,
      "grad_norm": 14.364479064941406,
      "learning_rate": 2.0092592592592595e-06,
      "loss": 3.0897,
      "step": 7772
    },
    {
      "epoch": 0.8996527777777777,
      "grad_norm": 0.03868158534169197,
      "learning_rate": 2.0069444444444446e-06,
      "loss": 0.007,
      "step": 7773
    },
    {
      "epoch": 0.8997685185185185,
      "grad_norm": 1.0167516469955444,
      "learning_rate": 2.0046296296296297e-06,
      "loss": 0.0119,
      "step": 7774
    },
    {
      "epoch": 0.8998842592592593,
      "grad_norm": 0.04402390867471695,
      "learning_rate": 2.002314814814815e-06,
      "loss": 0.0076,
      "step": 7775
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.03605072200298309,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0065,
      "step": 7776
    },
    {
      "epoch": 0.9001157407407407,
      "grad_norm": 0.048824358731508255,
      "learning_rate": 1.9976851851851854e-06,
      "loss": 0.0085,
      "step": 7777
    },
    {
      "epoch": 0.9002314814814815,
      "grad_norm": 0.07137925922870636,
      "learning_rate": 1.9953703703703705e-06,
      "loss": 0.0068,
      "step": 7778
    },
    {
      "epoch": 0.9003472222222222,
      "grad_norm": 0.054170142859220505,
      "learning_rate": 1.993055555555556e-06,
      "loss": 0.0072,
      "step": 7779
    },
    {
      "epoch": 0.9004629629629629,
      "grad_norm": 0.095502108335495,
      "learning_rate": 1.9907407407407407e-06,
      "loss": 0.0107,
      "step": 7780
    },
    {
      "epoch": 0.9005787037037037,
      "grad_norm": 0.04014464095234871,
      "learning_rate": 1.9884259259259262e-06,
      "loss": 0.0071,
      "step": 7781
    },
    {
      "epoch": 0.9006944444444445,
      "grad_norm": 0.04239508882164955,
      "learning_rate": 1.9861111111111113e-06,
      "loss": 0.0066,
      "step": 7782
    },
    {
      "epoch": 0.9008101851851852,
      "grad_norm": 0.05406760051846504,
      "learning_rate": 1.9837962962962964e-06,
      "loss": 0.0068,
      "step": 7783
    },
    {
      "epoch": 0.9009259259259259,
      "grad_norm": 0.3393630087375641,
      "learning_rate": 1.9814814814814815e-06,
      "loss": 0.0091,
      "step": 7784
    },
    {
      "epoch": 0.9010416666666666,
      "grad_norm": 0.044681623578071594,
      "learning_rate": 1.9791666666666666e-06,
      "loss": 0.0076,
      "step": 7785
    },
    {
      "epoch": 0.9011574074074075,
      "grad_norm": 0.05889028310775757,
      "learning_rate": 1.976851851851852e-06,
      "loss": 0.0102,
      "step": 7786
    },
    {
      "epoch": 0.9012731481481482,
      "grad_norm": 0.05159700661897659,
      "learning_rate": 1.9745370370370373e-06,
      "loss": 0.0091,
      "step": 7787
    },
    {
      "epoch": 0.9013888888888889,
      "grad_norm": 0.032380688935518265,
      "learning_rate": 1.9722222222222224e-06,
      "loss": 0.0057,
      "step": 7788
    },
    {
      "epoch": 0.9015046296296296,
      "grad_norm": 0.03337721899151802,
      "learning_rate": 1.9699074074074074e-06,
      "loss": 0.006,
      "step": 7789
    },
    {
      "epoch": 0.9016203703703703,
      "grad_norm": 0.054586198180913925,
      "learning_rate": 1.967592592592593e-06,
      "loss": 0.0087,
      "step": 7790
    },
    {
      "epoch": 0.9017361111111111,
      "grad_norm": 0.03277912363409996,
      "learning_rate": 1.965277777777778e-06,
      "loss": 0.0057,
      "step": 7791
    },
    {
      "epoch": 0.9018518518518519,
      "grad_norm": 0.03320659324526787,
      "learning_rate": 1.962962962962963e-06,
      "loss": 0.006,
      "step": 7792
    },
    {
      "epoch": 0.9019675925925926,
      "grad_norm": 0.0369175486266613,
      "learning_rate": 1.9606481481481483e-06,
      "loss": 0.0067,
      "step": 7793
    },
    {
      "epoch": 0.9020833333333333,
      "grad_norm": 0.05552149936556816,
      "learning_rate": 1.9583333333333334e-06,
      "loss": 0.0096,
      "step": 7794
    },
    {
      "epoch": 0.9021990740740741,
      "grad_norm": 0.03184597194194794,
      "learning_rate": 1.956018518518519e-06,
      "loss": 0.0057,
      "step": 7795
    },
    {
      "epoch": 0.9023148148148148,
      "grad_norm": 0.7985448241233826,
      "learning_rate": 1.953703703703704e-06,
      "loss": 0.0129,
      "step": 7796
    },
    {
      "epoch": 0.9024305555555555,
      "grad_norm": 0.055112600326538086,
      "learning_rate": 1.951388888888889e-06,
      "loss": 0.0099,
      "step": 7797
    },
    {
      "epoch": 0.9025462962962963,
      "grad_norm": 0.05344971641898155,
      "learning_rate": 1.949074074074074e-06,
      "loss": 0.009,
      "step": 7798
    },
    {
      "epoch": 0.9026620370370371,
      "grad_norm": 0.11896377056837082,
      "learning_rate": 1.9467592592592593e-06,
      "loss": 0.0111,
      "step": 7799
    },
    {
      "epoch": 0.9027777777777778,
      "grad_norm": 0.042701296508312225,
      "learning_rate": 1.944444444444445e-06,
      "loss": 0.0075,
      "step": 7800
    },
    {
      "epoch": 0.9028935185185185,
      "grad_norm": 0.07659336179494858,
      "learning_rate": 1.94212962962963e-06,
      "loss": 0.0087,
      "step": 7801
    },
    {
      "epoch": 0.9030092592592592,
      "grad_norm": 0.588111937046051,
      "learning_rate": 1.939814814814815e-06,
      "loss": 0.0136,
      "step": 7802
    },
    {
      "epoch": 0.903125,
      "grad_norm": 0.0432235524058342,
      "learning_rate": 1.9375e-06,
      "loss": 0.0075,
      "step": 7803
    },
    {
      "epoch": 0.9032407407407408,
      "grad_norm": 0.04770923778414726,
      "learning_rate": 1.935185185185185e-06,
      "loss": 0.0077,
      "step": 7804
    },
    {
      "epoch": 0.9033564814814815,
      "grad_norm": 0.03402473032474518,
      "learning_rate": 1.9328703703703707e-06,
      "loss": 0.0058,
      "step": 7805
    },
    {
      "epoch": 0.9034722222222222,
      "grad_norm": 0.45379629731178284,
      "learning_rate": 1.930555555555556e-06,
      "loss": 0.0085,
      "step": 7806
    },
    {
      "epoch": 0.903587962962963,
      "grad_norm": 0.05062020942568779,
      "learning_rate": 1.928240740740741e-06,
      "loss": 0.0088,
      "step": 7807
    },
    {
      "epoch": 0.9037037037037037,
      "grad_norm": 0.03258758410811424,
      "learning_rate": 1.925925925925926e-06,
      "loss": 0.0059,
      "step": 7808
    },
    {
      "epoch": 0.9038194444444444,
      "grad_norm": 0.1790517419576645,
      "learning_rate": 1.9236111111111115e-06,
      "loss": 0.0101,
      "step": 7809
    },
    {
      "epoch": 0.9039351851851852,
      "grad_norm": 0.04360410198569298,
      "learning_rate": 1.921296296296296e-06,
      "loss": 0.0077,
      "step": 7810
    },
    {
      "epoch": 0.904050925925926,
      "grad_norm": 0.06491483747959137,
      "learning_rate": 1.9189814814814817e-06,
      "loss": 0.0113,
      "step": 7811
    },
    {
      "epoch": 0.9041666666666667,
      "grad_norm": 0.078898124396801,
      "learning_rate": 1.916666666666667e-06,
      "loss": 0.0094,
      "step": 7812
    },
    {
      "epoch": 0.9042824074074074,
      "grad_norm": 0.05055541545152664,
      "learning_rate": 1.914351851851852e-06,
      "loss": 0.0086,
      "step": 7813
    },
    {
      "epoch": 0.9043981481481481,
      "grad_norm": 0.21743613481521606,
      "learning_rate": 1.9120370370370374e-06,
      "loss": 0.0117,
      "step": 7814
    },
    {
      "epoch": 0.9045138888888888,
      "grad_norm": 0.061897262930870056,
      "learning_rate": 1.909722222222222e-06,
      "loss": 0.0099,
      "step": 7815
    },
    {
      "epoch": 0.9046296296296297,
      "grad_norm": 0.03790393844246864,
      "learning_rate": 1.9074074074074076e-06,
      "loss": 0.0068,
      "step": 7816
    },
    {
      "epoch": 0.9047453703703704,
      "grad_norm": 0.038809072226285934,
      "learning_rate": 1.905092592592593e-06,
      "loss": 0.007,
      "step": 7817
    },
    {
      "epoch": 0.9048611111111111,
      "grad_norm": 0.04562908411026001,
      "learning_rate": 1.9027777777777778e-06,
      "loss": 0.0082,
      "step": 7818
    },
    {
      "epoch": 0.9049768518518518,
      "grad_norm": 0.05449262633919716,
      "learning_rate": 1.9004629629629631e-06,
      "loss": 0.0098,
      "step": 7819
    },
    {
      "epoch": 0.9050925925925926,
      "grad_norm": 0.042689066380262375,
      "learning_rate": 1.8981481481481484e-06,
      "loss": 0.0075,
      "step": 7820
    },
    {
      "epoch": 0.9052083333333333,
      "grad_norm": 0.04250313714146614,
      "learning_rate": 1.8958333333333333e-06,
      "loss": 0.0074,
      "step": 7821
    },
    {
      "epoch": 0.9053240740740741,
      "grad_norm": 0.046569664031267166,
      "learning_rate": 1.8935185185185186e-06,
      "loss": 0.0082,
      "step": 7822
    },
    {
      "epoch": 0.9054398148148148,
      "grad_norm": 0.03492434322834015,
      "learning_rate": 1.891203703703704e-06,
      "loss": 0.0063,
      "step": 7823
    },
    {
      "epoch": 0.9055555555555556,
      "grad_norm": 0.03283624351024628,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.0059,
      "step": 7824
    },
    {
      "epoch": 0.9056712962962963,
      "grad_norm": 0.05433674156665802,
      "learning_rate": 1.8865740740740743e-06,
      "loss": 0.0083,
      "step": 7825
    },
    {
      "epoch": 0.905787037037037,
      "grad_norm": 0.057609401643276215,
      "learning_rate": 1.8842592592592592e-06,
      "loss": 0.0099,
      "step": 7826
    },
    {
      "epoch": 0.9059027777777777,
      "grad_norm": 0.03863880783319473,
      "learning_rate": 1.8819444444444445e-06,
      "loss": 0.007,
      "step": 7827
    },
    {
      "epoch": 0.9060185185185186,
      "grad_norm": 0.06326716393232346,
      "learning_rate": 1.8796296296296299e-06,
      "loss": 0.0078,
      "step": 7828
    },
    {
      "epoch": 0.9061342592592593,
      "grad_norm": 0.03468671813607216,
      "learning_rate": 1.877314814814815e-06,
      "loss": 0.0062,
      "step": 7829
    },
    {
      "epoch": 0.90625,
      "grad_norm": 0.035577442497015,
      "learning_rate": 1.8750000000000003e-06,
      "loss": 0.0063,
      "step": 7830
    },
    {
      "epoch": 0.9063657407407407,
      "grad_norm": 0.042144209146499634,
      "learning_rate": 1.8726851851851854e-06,
      "loss": 0.0054,
      "step": 7831
    },
    {
      "epoch": 0.9064814814814814,
      "grad_norm": 0.03966345638036728,
      "learning_rate": 1.8703703703703705e-06,
      "loss": 0.0068,
      "step": 7832
    },
    {
      "epoch": 0.9065972222222223,
      "grad_norm": 0.05668322741985321,
      "learning_rate": 1.8680555555555558e-06,
      "loss": 0.0074,
      "step": 7833
    },
    {
      "epoch": 0.906712962962963,
      "grad_norm": 0.03289678692817688,
      "learning_rate": 1.865740740740741e-06,
      "loss": 0.0059,
      "step": 7834
    },
    {
      "epoch": 0.9068287037037037,
      "grad_norm": 0.05175674706697464,
      "learning_rate": 1.863425925925926e-06,
      "loss": 0.0095,
      "step": 7835
    },
    {
      "epoch": 0.9069444444444444,
      "grad_norm": 0.0467425175011158,
      "learning_rate": 1.8611111111111113e-06,
      "loss": 0.0078,
      "step": 7836
    },
    {
      "epoch": 0.9070601851851852,
      "grad_norm": 0.04373638331890106,
      "learning_rate": 1.8587962962962964e-06,
      "loss": 0.0075,
      "step": 7837
    },
    {
      "epoch": 0.9071759259259259,
      "grad_norm": 0.0432494655251503,
      "learning_rate": 1.8564814814814817e-06,
      "loss": 0.0074,
      "step": 7838
    },
    {
      "epoch": 0.9072916666666667,
      "grad_norm": 0.04272031411528587,
      "learning_rate": 1.854166666666667e-06,
      "loss": 0.0073,
      "step": 7839
    },
    {
      "epoch": 0.9074074074074074,
      "grad_norm": 0.04296282306313515,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.0076,
      "step": 7840
    },
    {
      "epoch": 0.9075231481481482,
      "grad_norm": 0.02900470793247223,
      "learning_rate": 1.8495370370370372e-06,
      "loss": 0.0053,
      "step": 7841
    },
    {
      "epoch": 0.9076388888888889,
      "grad_norm": 0.04300219938158989,
      "learning_rate": 1.8472222222222225e-06,
      "loss": 0.0075,
      "step": 7842
    },
    {
      "epoch": 0.9077546296296296,
      "grad_norm": 0.044832874089479446,
      "learning_rate": 1.8449074074074074e-06,
      "loss": 0.0077,
      "step": 7843
    },
    {
      "epoch": 0.9078703703703703,
      "grad_norm": 0.05367109924554825,
      "learning_rate": 1.8425925925925927e-06,
      "loss": 0.0094,
      "step": 7844
    },
    {
      "epoch": 0.9079861111111112,
      "grad_norm": 0.07454854995012283,
      "learning_rate": 1.840277777777778e-06,
      "loss": 0.0073,
      "step": 7845
    },
    {
      "epoch": 0.9081018518518519,
      "grad_norm": 0.050535429269075394,
      "learning_rate": 1.837962962962963e-06,
      "loss": 0.0089,
      "step": 7846
    },
    {
      "epoch": 0.9082175925925926,
      "grad_norm": 0.057123132050037384,
      "learning_rate": 1.8356481481481484e-06,
      "loss": 0.0102,
      "step": 7847
    },
    {
      "epoch": 0.9083333333333333,
      "grad_norm": 0.06458690017461777,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 0.0096,
      "step": 7848
    },
    {
      "epoch": 0.908449074074074,
      "grad_norm": 0.040646571666002274,
      "learning_rate": 1.8310185185185186e-06,
      "loss": 0.0072,
      "step": 7849
    },
    {
      "epoch": 0.9085648148148148,
      "grad_norm": 0.0363687165081501,
      "learning_rate": 1.828703703703704e-06,
      "loss": 0.0065,
      "step": 7850
    },
    {
      "epoch": 0.9086805555555556,
      "grad_norm": 0.04922572523355484,
      "learning_rate": 1.826388888888889e-06,
      "loss": 0.0083,
      "step": 7851
    },
    {
      "epoch": 0.9087962962962963,
      "grad_norm": 0.048211194574832916,
      "learning_rate": 1.8240740740740743e-06,
      "loss": 0.0061,
      "step": 7852
    },
    {
      "epoch": 0.908912037037037,
      "grad_norm": 0.03634215518832207,
      "learning_rate": 1.8217592592592594e-06,
      "loss": 0.0062,
      "step": 7853
    },
    {
      "epoch": 0.9090277777777778,
      "grad_norm": 0.04739901050925255,
      "learning_rate": 1.8194444444444445e-06,
      "loss": 0.0083,
      "step": 7854
    },
    {
      "epoch": 0.9091435185185185,
      "grad_norm": 0.05663163587450981,
      "learning_rate": 1.8171296296296298e-06,
      "loss": 0.0091,
      "step": 7855
    },
    {
      "epoch": 0.9092592592592592,
      "grad_norm": 0.06964210420846939,
      "learning_rate": 1.8148148148148151e-06,
      "loss": 0.0091,
      "step": 7856
    },
    {
      "epoch": 0.909375,
      "grad_norm": 0.02969738282263279,
      "learning_rate": 1.8125e-06,
      "loss": 0.0053,
      "step": 7857
    },
    {
      "epoch": 0.9094907407407408,
      "grad_norm": 0.0737355425953865,
      "learning_rate": 1.8101851851851853e-06,
      "loss": 0.0096,
      "step": 7858
    },
    {
      "epoch": 0.9096064814814815,
      "grad_norm": 0.03178365156054497,
      "learning_rate": 1.8078703703703704e-06,
      "loss": 0.0058,
      "step": 7859
    },
    {
      "epoch": 0.9097222222222222,
      "grad_norm": 8.493189811706543,
      "learning_rate": 1.8055555555555557e-06,
      "loss": 3.663,
      "step": 7860
    },
    {
      "epoch": 0.9098379629629629,
      "grad_norm": 0.043899718672037125,
      "learning_rate": 1.803240740740741e-06,
      "loss": 0.0057,
      "step": 7861
    },
    {
      "epoch": 0.9099537037037037,
      "grad_norm": 0.04087714105844498,
      "learning_rate": 1.800925925925926e-06,
      "loss": 0.0073,
      "step": 7862
    },
    {
      "epoch": 0.9100694444444445,
      "grad_norm": 0.04388356953859329,
      "learning_rate": 1.7986111111111112e-06,
      "loss": 0.0078,
      "step": 7863
    },
    {
      "epoch": 0.9101851851851852,
      "grad_norm": 10.070025444030762,
      "learning_rate": 1.7962962962962965e-06,
      "loss": 0.0371,
      "step": 7864
    },
    {
      "epoch": 0.9103009259259259,
      "grad_norm": 0.06778857111930847,
      "learning_rate": 1.7939814814814816e-06,
      "loss": 0.0112,
      "step": 7865
    },
    {
      "epoch": 0.9104166666666667,
      "grad_norm": 2.8849785327911377,
      "learning_rate": 1.7916666666666667e-06,
      "loss": 0.0207,
      "step": 7866
    },
    {
      "epoch": 0.9105324074074074,
      "grad_norm": 0.06641587615013123,
      "learning_rate": 1.789351851851852e-06,
      "loss": 0.0083,
      "step": 7867
    },
    {
      "epoch": 0.9106481481481481,
      "grad_norm": 0.035292331129312515,
      "learning_rate": 1.7870370370370371e-06,
      "loss": 0.0063,
      "step": 7868
    },
    {
      "epoch": 0.9107638888888889,
      "grad_norm": 0.02897886000573635,
      "learning_rate": 1.7847222222222225e-06,
      "loss": 0.0053,
      "step": 7869
    },
    {
      "epoch": 0.9108796296296297,
      "grad_norm": 0.15650209784507751,
      "learning_rate": 1.7824074074074073e-06,
      "loss": 0.0083,
      "step": 7870
    },
    {
      "epoch": 0.9109953703703704,
      "grad_norm": 0.047566551715135574,
      "learning_rate": 1.7800925925925926e-06,
      "loss": 0.0083,
      "step": 7871
    },
    {
      "epoch": 0.9111111111111111,
      "grad_norm": 0.04535403475165367,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.0078,
      "step": 7872
    },
    {
      "epoch": 0.9112268518518518,
      "grad_norm": 0.03227509930729866,
      "learning_rate": 1.775462962962963e-06,
      "loss": 0.0057,
      "step": 7873
    },
    {
      "epoch": 0.9113425925925925,
      "grad_norm": 0.03885728120803833,
      "learning_rate": 1.7731481481481484e-06,
      "loss": 0.007,
      "step": 7874
    },
    {
      "epoch": 0.9114583333333334,
      "grad_norm": 0.04938401281833649,
      "learning_rate": 1.7708333333333337e-06,
      "loss": 0.0085,
      "step": 7875
    },
    {
      "epoch": 0.9115740740740741,
      "grad_norm": 0.04719718173146248,
      "learning_rate": 1.7685185185185186e-06,
      "loss": 0.0083,
      "step": 7876
    },
    {
      "epoch": 0.9116898148148148,
      "grad_norm": 0.04188227280974388,
      "learning_rate": 1.7662037037037039e-06,
      "loss": 0.0073,
      "step": 7877
    },
    {
      "epoch": 0.9118055555555555,
      "grad_norm": 0.042261894792318344,
      "learning_rate": 1.7638888888888892e-06,
      "loss": 0.0064,
      "step": 7878
    },
    {
      "epoch": 0.9119212962962963,
      "grad_norm": 0.04489195719361305,
      "learning_rate": 1.761574074074074e-06,
      "loss": 0.008,
      "step": 7879
    },
    {
      "epoch": 0.9120370370370371,
      "grad_norm": 0.046308983117341995,
      "learning_rate": 1.7592592592592594e-06,
      "loss": 0.0078,
      "step": 7880
    },
    {
      "epoch": 0.9121527777777778,
      "grad_norm": 0.053161561489105225,
      "learning_rate": 1.7569444444444445e-06,
      "loss": 0.0064,
      "step": 7881
    },
    {
      "epoch": 0.9122685185185185,
      "grad_norm": 0.038633644580841064,
      "learning_rate": 1.7546296296296298e-06,
      "loss": 0.0062,
      "step": 7882
    },
    {
      "epoch": 0.9123842592592593,
      "grad_norm": 0.05332568287849426,
      "learning_rate": 1.752314814814815e-06,
      "loss": 0.0074,
      "step": 7883
    },
    {
      "epoch": 0.9125,
      "grad_norm": 0.043247971683740616,
      "learning_rate": 1.75e-06,
      "loss": 0.0067,
      "step": 7884
    },
    {
      "epoch": 0.9126157407407407,
      "grad_norm": 0.03760272264480591,
      "learning_rate": 1.7476851851851853e-06,
      "loss": 0.0068,
      "step": 7885
    },
    {
      "epoch": 0.9127314814814815,
      "grad_norm": 0.05875160172581673,
      "learning_rate": 1.7453703703703706e-06,
      "loss": 0.0059,
      "step": 7886
    },
    {
      "epoch": 0.9128472222222223,
      "grad_norm": 0.049807846546173096,
      "learning_rate": 1.7430555555555557e-06,
      "loss": 0.0081,
      "step": 7887
    },
    {
      "epoch": 0.912962962962963,
      "grad_norm": 0.058855101466178894,
      "learning_rate": 1.740740740740741e-06,
      "loss": 0.0102,
      "step": 7888
    },
    {
      "epoch": 0.9130787037037037,
      "grad_norm": 0.04994690790772438,
      "learning_rate": 1.738425925925926e-06,
      "loss": 0.0088,
      "step": 7889
    },
    {
      "epoch": 0.9131944444444444,
      "grad_norm": 0.07219581305980682,
      "learning_rate": 1.7361111111111112e-06,
      "loss": 0.0115,
      "step": 7890
    },
    {
      "epoch": 0.9133101851851851,
      "grad_norm": 0.057738929986953735,
      "learning_rate": 1.7337962962962965e-06,
      "loss": 0.0105,
      "step": 7891
    },
    {
      "epoch": 0.913425925925926,
      "grad_norm": 0.03224050998687744,
      "learning_rate": 1.7314814814814814e-06,
      "loss": 0.0058,
      "step": 7892
    },
    {
      "epoch": 0.9135416666666667,
      "grad_norm": 0.032460786402225494,
      "learning_rate": 1.7291666666666667e-06,
      "loss": 0.0059,
      "step": 7893
    },
    {
      "epoch": 0.9136574074074074,
      "grad_norm": 7.199158191680908,
      "learning_rate": 1.726851851851852e-06,
      "loss": 3.3949,
      "step": 7894
    },
    {
      "epoch": 0.9137731481481481,
      "grad_norm": 34.13240432739258,
      "learning_rate": 1.724537037037037e-06,
      "loss": 0.0738,
      "step": 7895
    },
    {
      "epoch": 0.9138888888888889,
      "grad_norm": 0.08856478333473206,
      "learning_rate": 1.7222222222222224e-06,
      "loss": 0.0103,
      "step": 7896
    },
    {
      "epoch": 0.9140046296296296,
      "grad_norm": 0.032055798918008804,
      "learning_rate": 1.7199074074074077e-06,
      "loss": 0.0058,
      "step": 7897
    },
    {
      "epoch": 0.9141203703703704,
      "grad_norm": 0.04883159324526787,
      "learning_rate": 1.7175925925925926e-06,
      "loss": 0.0079,
      "step": 7898
    },
    {
      "epoch": 0.9142361111111111,
      "grad_norm": 0.060633186250925064,
      "learning_rate": 1.715277777777778e-06,
      "loss": 0.0091,
      "step": 7899
    },
    {
      "epoch": 0.9143518518518519,
      "grad_norm": 0.03252457454800606,
      "learning_rate": 1.7129629629629632e-06,
      "loss": 0.0057,
      "step": 7900
    },
    {
      "epoch": 0.9144675925925926,
      "grad_norm": 0.04392702132463455,
      "learning_rate": 1.7106481481481483e-06,
      "loss": 0.0078,
      "step": 7901
    },
    {
      "epoch": 0.9145833333333333,
      "grad_norm": 0.04392701014876366,
      "learning_rate": 1.7083333333333334e-06,
      "loss": 0.0077,
      "step": 7902
    },
    {
      "epoch": 0.914699074074074,
      "grad_norm": 0.04854021966457367,
      "learning_rate": 1.7060185185185187e-06,
      "loss": 0.0079,
      "step": 7903
    },
    {
      "epoch": 0.9148148148148149,
      "grad_norm": 0.04883507639169693,
      "learning_rate": 1.7037037037037038e-06,
      "loss": 0.0083,
      "step": 7904
    },
    {
      "epoch": 0.9149305555555556,
      "grad_norm": 0.04336908832192421,
      "learning_rate": 1.7013888888888891e-06,
      "loss": 0.0077,
      "step": 7905
    },
    {
      "epoch": 0.9150462962962963,
      "grad_norm": 0.07136163115501404,
      "learning_rate": 1.699074074074074e-06,
      "loss": 0.0082,
      "step": 7906
    },
    {
      "epoch": 0.915162037037037,
      "grad_norm": 0.04481067508459091,
      "learning_rate": 1.6967592592592593e-06,
      "loss": 0.0055,
      "step": 7907
    },
    {
      "epoch": 0.9152777777777777,
      "grad_norm": 0.04555663838982582,
      "learning_rate": 1.6944444444444446e-06,
      "loss": 0.0083,
      "step": 7908
    },
    {
      "epoch": 0.9153935185185185,
      "grad_norm": 23.599451065063477,
      "learning_rate": 1.6921296296296297e-06,
      "loss": 2.8923,
      "step": 7909
    },
    {
      "epoch": 0.9155092592592593,
      "grad_norm": 0.0775023028254509,
      "learning_rate": 1.689814814814815e-06,
      "loss": 0.01,
      "step": 7910
    },
    {
      "epoch": 0.915625,
      "grad_norm": 0.046542394906282425,
      "learning_rate": 1.6875000000000001e-06,
      "loss": 0.0079,
      "step": 7911
    },
    {
      "epoch": 0.9157407407407407,
      "grad_norm": 0.05074705183506012,
      "learning_rate": 1.6851851851851852e-06,
      "loss": 0.0084,
      "step": 7912
    },
    {
      "epoch": 0.9158564814814815,
      "grad_norm": 0.04484666511416435,
      "learning_rate": 1.6828703703703706e-06,
      "loss": 0.0078,
      "step": 7913
    },
    {
      "epoch": 0.9159722222222222,
      "grad_norm": 0.03280862793326378,
      "learning_rate": 1.6805555555555559e-06,
      "loss": 0.0059,
      "step": 7914
    },
    {
      "epoch": 0.9160879629629629,
      "grad_norm": 0.056511662900447845,
      "learning_rate": 1.6782407407407408e-06,
      "loss": 0.0082,
      "step": 7915
    },
    {
      "epoch": 0.9162037037037037,
      "grad_norm": 87.29227447509766,
      "learning_rate": 1.675925925925926e-06,
      "loss": 0.9396,
      "step": 7916
    },
    {
      "epoch": 0.9163194444444445,
      "grad_norm": 0.052119385451078415,
      "learning_rate": 1.6736111111111112e-06,
      "loss": 0.0093,
      "step": 7917
    },
    {
      "epoch": 0.9164351851851852,
      "grad_norm": 0.046668604016304016,
      "learning_rate": 1.6712962962962965e-06,
      "loss": 0.0082,
      "step": 7918
    },
    {
      "epoch": 0.9165509259259259,
      "grad_norm": 0.03901301324367523,
      "learning_rate": 1.6689814814814818e-06,
      "loss": 0.0065,
      "step": 7919
    },
    {
      "epoch": 0.9166666666666666,
      "grad_norm": 0.041583314538002014,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.0068,
      "step": 7920
    },
    {
      "epoch": 0.9167824074074075,
      "grad_norm": 0.04513128474354744,
      "learning_rate": 1.664351851851852e-06,
      "loss": 0.0079,
      "step": 7921
    },
    {
      "epoch": 0.9168981481481482,
      "grad_norm": 0.033454541116952896,
      "learning_rate": 1.6620370370370373e-06,
      "loss": 0.006,
      "step": 7922
    },
    {
      "epoch": 0.9170138888888889,
      "grad_norm": 0.03939398378133774,
      "learning_rate": 1.6597222222222224e-06,
      "loss": 0.0069,
      "step": 7923
    },
    {
      "epoch": 0.9171296296296296,
      "grad_norm": 3.15142822265625,
      "learning_rate": 1.6574074074074075e-06,
      "loss": 0.0263,
      "step": 7924
    },
    {
      "epoch": 0.9172453703703703,
      "grad_norm": 0.04822326824069023,
      "learning_rate": 1.6550925925925928e-06,
      "loss": 0.0083,
      "step": 7925
    },
    {
      "epoch": 0.9173611111111111,
      "grad_norm": 0.044108085334300995,
      "learning_rate": 1.6527777777777779e-06,
      "loss": 0.0075,
      "step": 7926
    },
    {
      "epoch": 0.9174768518518519,
      "grad_norm": 0.03169795870780945,
      "learning_rate": 1.6504629629629632e-06,
      "loss": 0.0057,
      "step": 7927
    },
    {
      "epoch": 0.9175925925925926,
      "grad_norm": 0.067936971783638,
      "learning_rate": 1.648148148148148e-06,
      "loss": 0.0091,
      "step": 7928
    },
    {
      "epoch": 0.9177083333333333,
      "grad_norm": 0.07534895837306976,
      "learning_rate": 1.6458333333333334e-06,
      "loss": 0.0103,
      "step": 7929
    },
    {
      "epoch": 0.9178240740740741,
      "grad_norm": 0.05451353266835213,
      "learning_rate": 1.6435185185185187e-06,
      "loss": 0.01,
      "step": 7930
    },
    {
      "epoch": 0.9179398148148148,
      "grad_norm": 0.057643722742795944,
      "learning_rate": 1.6412037037037038e-06,
      "loss": 0.0098,
      "step": 7931
    },
    {
      "epoch": 0.9180555555555555,
      "grad_norm": 0.03774186968803406,
      "learning_rate": 1.638888888888889e-06,
      "loss": 0.0065,
      "step": 7932
    },
    {
      "epoch": 0.9181712962962963,
      "grad_norm": 0.07607339322566986,
      "learning_rate": 1.6365740740740744e-06,
      "loss": 0.0098,
      "step": 7933
    },
    {
      "epoch": 0.9182870370370371,
      "grad_norm": 0.041888073086738586,
      "learning_rate": 1.6342592592592593e-06,
      "loss": 0.0075,
      "step": 7934
    },
    {
      "epoch": 0.9184027777777778,
      "grad_norm": 0.04567943140864372,
      "learning_rate": 1.6319444444444446e-06,
      "loss": 0.0079,
      "step": 7935
    },
    {
      "epoch": 0.9185185185185185,
      "grad_norm": 0.04958627000451088,
      "learning_rate": 1.62962962962963e-06,
      "loss": 0.0087,
      "step": 7936
    },
    {
      "epoch": 0.9186342592592592,
      "grad_norm": 0.03494562953710556,
      "learning_rate": 1.6273148148148148e-06,
      "loss": 0.0063,
      "step": 7937
    },
    {
      "epoch": 0.91875,
      "grad_norm": 0.05567748844623566,
      "learning_rate": 1.6250000000000001e-06,
      "loss": 0.0084,
      "step": 7938
    },
    {
      "epoch": 0.9188657407407408,
      "grad_norm": 0.06009401008486748,
      "learning_rate": 1.6226851851851852e-06,
      "loss": 0.0084,
      "step": 7939
    },
    {
      "epoch": 0.9189814814814815,
      "grad_norm": 0.05074501037597656,
      "learning_rate": 1.6203703703703705e-06,
      "loss": 0.0083,
      "step": 7940
    },
    {
      "epoch": 0.9190972222222222,
      "grad_norm": 0.055763985961675644,
      "learning_rate": 1.6180555555555558e-06,
      "loss": 0.0098,
      "step": 7941
    },
    {
      "epoch": 0.919212962962963,
      "grad_norm": 0.04159694164991379,
      "learning_rate": 1.6157407407407407e-06,
      "loss": 0.0072,
      "step": 7942
    },
    {
      "epoch": 0.9193287037037037,
      "grad_norm": 0.047615379095077515,
      "learning_rate": 1.613425925925926e-06,
      "loss": 0.0082,
      "step": 7943
    },
    {
      "epoch": 0.9194444444444444,
      "grad_norm": 0.041097648441791534,
      "learning_rate": 1.6111111111111113e-06,
      "loss": 0.0071,
      "step": 7944
    },
    {
      "epoch": 0.9195601851851852,
      "grad_norm": 0.05231446772813797,
      "learning_rate": 1.6087962962962964e-06,
      "loss": 0.0093,
      "step": 7945
    },
    {
      "epoch": 0.919675925925926,
      "grad_norm": 0.031225424259901047,
      "learning_rate": 1.6064814814814817e-06,
      "loss": 0.0056,
      "step": 7946
    },
    {
      "epoch": 0.9197916666666667,
      "grad_norm": 0.04950704425573349,
      "learning_rate": 1.6041666666666668e-06,
      "loss": 0.0088,
      "step": 7947
    },
    {
      "epoch": 0.9199074074074074,
      "grad_norm": 0.04303058981895447,
      "learning_rate": 1.601851851851852e-06,
      "loss": 0.0074,
      "step": 7948
    },
    {
      "epoch": 0.9200231481481481,
      "grad_norm": 0.031922318041324615,
      "learning_rate": 1.5995370370370372e-06,
      "loss": 0.0056,
      "step": 7949
    },
    {
      "epoch": 0.9201388888888888,
      "grad_norm": 0.04191374033689499,
      "learning_rate": 1.5972222222222221e-06,
      "loss": 0.0074,
      "step": 7950
    },
    {
      "epoch": 0.9202546296296297,
      "grad_norm": 0.04747382923960686,
      "learning_rate": 1.5949074074074074e-06,
      "loss": 0.0085,
      "step": 7951
    },
    {
      "epoch": 0.9203703703703704,
      "grad_norm": 0.12265541404485703,
      "learning_rate": 1.5925925925925927e-06,
      "loss": 0.0107,
      "step": 7952
    },
    {
      "epoch": 0.9204861111111111,
      "grad_norm": 0.04164673015475273,
      "learning_rate": 1.5902777777777778e-06,
      "loss": 0.0072,
      "step": 7953
    },
    {
      "epoch": 0.9206018518518518,
      "grad_norm": 0.08035333454608917,
      "learning_rate": 1.5879629629629632e-06,
      "loss": 0.0103,
      "step": 7954
    },
    {
      "epoch": 0.9207175925925926,
      "grad_norm": 0.07454076409339905,
      "learning_rate": 1.5856481481481485e-06,
      "loss": 0.0095,
      "step": 7955
    },
    {
      "epoch": 0.9208333333333333,
      "grad_norm": 0.033715005964040756,
      "learning_rate": 1.5833333333333333e-06,
      "loss": 0.006,
      "step": 7956
    },
    {
      "epoch": 0.9209490740740741,
      "grad_norm": 0.060052886605262756,
      "learning_rate": 1.5810185185185187e-06,
      "loss": 0.0094,
      "step": 7957
    },
    {
      "epoch": 0.9210648148148148,
      "grad_norm": 0.03928948566317558,
      "learning_rate": 1.578703703703704e-06,
      "loss": 0.0068,
      "step": 7958
    },
    {
      "epoch": 0.9211805555555556,
      "grad_norm": 0.03639419376850128,
      "learning_rate": 1.576388888888889e-06,
      "loss": 0.0064,
      "step": 7959
    },
    {
      "epoch": 0.9212962962962963,
      "grad_norm": 0.04751845449209213,
      "learning_rate": 1.5740740740740742e-06,
      "loss": 0.0082,
      "step": 7960
    },
    {
      "epoch": 0.921412037037037,
      "grad_norm": 0.040856484323740005,
      "learning_rate": 1.5717592592592593e-06,
      "loss": 0.0072,
      "step": 7961
    },
    {
      "epoch": 0.9215277777777777,
      "grad_norm": 0.04279632121324539,
      "learning_rate": 1.5694444444444446e-06,
      "loss": 0.0076,
      "step": 7962
    },
    {
      "epoch": 0.9216435185185186,
      "grad_norm": 0.05194546654820442,
      "learning_rate": 1.5671296296296299e-06,
      "loss": 0.0067,
      "step": 7963
    },
    {
      "epoch": 0.9217592592592593,
      "grad_norm": 0.0778445228934288,
      "learning_rate": 1.5648148148148148e-06,
      "loss": 0.0091,
      "step": 7964
    },
    {
      "epoch": 0.921875,
      "grad_norm": 0.05776482820510864,
      "learning_rate": 1.5625e-06,
      "loss": 0.0091,
      "step": 7965
    },
    {
      "epoch": 0.9219907407407407,
      "grad_norm": 0.5354434251785278,
      "learning_rate": 1.5601851851851854e-06,
      "loss": 0.0119,
      "step": 7966
    },
    {
      "epoch": 0.9221064814814814,
      "grad_norm": 0.047156404703855515,
      "learning_rate": 1.5578703703703705e-06,
      "loss": 0.0065,
      "step": 7967
    },
    {
      "epoch": 0.9222222222222223,
      "grad_norm": 0.06139934062957764,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.0098,
      "step": 7968
    },
    {
      "epoch": 0.922337962962963,
      "grad_norm": 0.29762065410614014,
      "learning_rate": 1.5532407407407409e-06,
      "loss": 0.0089,
      "step": 7969
    },
    {
      "epoch": 0.9224537037037037,
      "grad_norm": 0.05110229179263115,
      "learning_rate": 1.550925925925926e-06,
      "loss": 0.0076,
      "step": 7970
    },
    {
      "epoch": 0.9225694444444444,
      "grad_norm": 0.04459325969219208,
      "learning_rate": 1.5486111111111113e-06,
      "loss": 0.0075,
      "step": 7971
    },
    {
      "epoch": 0.9226851851851852,
      "grad_norm": 0.03182804957032204,
      "learning_rate": 1.5462962962962964e-06,
      "loss": 0.0057,
      "step": 7972
    },
    {
      "epoch": 0.9228009259259259,
      "grad_norm": 0.050145432353019714,
      "learning_rate": 1.5439814814814815e-06,
      "loss": 0.0074,
      "step": 7973
    },
    {
      "epoch": 0.9229166666666667,
      "grad_norm": 160.32798767089844,
      "learning_rate": 1.5416666666666668e-06,
      "loss": 0.7032,
      "step": 7974
    },
    {
      "epoch": 0.9230324074074074,
      "grad_norm": 0.035113375633955,
      "learning_rate": 1.539351851851852e-06,
      "loss": 0.0063,
      "step": 7975
    },
    {
      "epoch": 0.9231481481481482,
      "grad_norm": 0.03657441586256027,
      "learning_rate": 1.5370370370370372e-06,
      "loss": 0.0062,
      "step": 7976
    },
    {
      "epoch": 0.9232638888888889,
      "grad_norm": 0.0428561307489872,
      "learning_rate": 1.5347222222222225e-06,
      "loss": 0.0075,
      "step": 7977
    },
    {
      "epoch": 0.9233796296296296,
      "grad_norm": 0.04874766245484352,
      "learning_rate": 1.5324074074074074e-06,
      "loss": 0.0087,
      "step": 7978
    },
    {
      "epoch": 0.9234953703703703,
      "grad_norm": 0.05359295383095741,
      "learning_rate": 1.5300925925925927e-06,
      "loss": 0.0094,
      "step": 7979
    },
    {
      "epoch": 0.9236111111111112,
      "grad_norm": 0.03809152543544769,
      "learning_rate": 1.527777777777778e-06,
      "loss": 0.0069,
      "step": 7980
    },
    {
      "epoch": 0.9237268518518519,
      "grad_norm": 0.04605831205844879,
      "learning_rate": 1.5254629629629631e-06,
      "loss": 0.0083,
      "step": 7981
    },
    {
      "epoch": 0.9238425925925926,
      "grad_norm": 0.04898565262556076,
      "learning_rate": 1.5231481481481482e-06,
      "loss": 0.0083,
      "step": 7982
    },
    {
      "epoch": 0.9239583333333333,
      "grad_norm": 0.05675278976559639,
      "learning_rate": 1.5208333333333333e-06,
      "loss": 0.0086,
      "step": 7983
    },
    {
      "epoch": 0.924074074074074,
      "grad_norm": 0.04267251119017601,
      "learning_rate": 1.5185185185185186e-06,
      "loss": 0.0074,
      "step": 7984
    },
    {
      "epoch": 0.9241898148148148,
      "grad_norm": 0.03460592404007912,
      "learning_rate": 1.516203703703704e-06,
      "loss": 0.0062,
      "step": 7985
    },
    {
      "epoch": 0.9243055555555556,
      "grad_norm": 0.04459543153643608,
      "learning_rate": 1.5138888888888888e-06,
      "loss": 0.008,
      "step": 7986
    },
    {
      "epoch": 0.9244212962962963,
      "grad_norm": 0.03287769854068756,
      "learning_rate": 1.5115740740740741e-06,
      "loss": 0.0058,
      "step": 7987
    },
    {
      "epoch": 0.924537037037037,
      "grad_norm": 0.044473908841609955,
      "learning_rate": 1.5092592592592594e-06,
      "loss": 0.0081,
      "step": 7988
    },
    {
      "epoch": 0.9246527777777778,
      "grad_norm": 0.03509360924363136,
      "learning_rate": 1.5069444444444445e-06,
      "loss": 0.0063,
      "step": 7989
    },
    {
      "epoch": 0.9247685185185185,
      "grad_norm": 0.043504536151885986,
      "learning_rate": 1.5046296296296298e-06,
      "loss": 0.0075,
      "step": 7990
    },
    {
      "epoch": 0.9248842592592592,
      "grad_norm": 174.01895141601562,
      "learning_rate": 1.5023148148148152e-06,
      "loss": 0.8309,
      "step": 7991
    },
    {
      "epoch": 0.925,
      "grad_norm": 0.02854466810822487,
      "learning_rate": 1.5e-06,
      "loss": 0.0052,
      "step": 7992
    },
    {
      "epoch": 0.9251157407407408,
      "grad_norm": 15.244916915893555,
      "learning_rate": 1.4976851851851853e-06,
      "loss": 0.0607,
      "step": 7993
    },
    {
      "epoch": 0.9252314814814815,
      "grad_norm": 16.299175262451172,
      "learning_rate": 1.4953703703703704e-06,
      "loss": 3.1258,
      "step": 7994
    },
    {
      "epoch": 0.9253472222222222,
      "grad_norm": 0.05611031502485275,
      "learning_rate": 1.4930555555555555e-06,
      "loss": 0.0073,
      "step": 7995
    },
    {
      "epoch": 0.9254629629629629,
      "grad_norm": 0.040495000779628754,
      "learning_rate": 1.4907407407407409e-06,
      "loss": 0.0067,
      "step": 7996
    },
    {
      "epoch": 0.9255787037037037,
      "grad_norm": 0.051569968461990356,
      "learning_rate": 1.488425925925926e-06,
      "loss": 0.0092,
      "step": 7997
    },
    {
      "epoch": 0.9256944444444445,
      "grad_norm": 204.3791961669922,
      "learning_rate": 1.4861111111111113e-06,
      "loss": 0.7059,
      "step": 7998
    },
    {
      "epoch": 0.9258101851851852,
      "grad_norm": 0.03815023973584175,
      "learning_rate": 1.4837962962962966e-06,
      "loss": 0.0068,
      "step": 7999
    },
    {
      "epoch": 0.9259259259259259,
      "grad_norm": 0.06280530244112015,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.0095,
      "step": 8000
    },
    {
      "epoch": 0.9260416666666667,
      "grad_norm": 0.03600270301103592,
      "learning_rate": 1.4791666666666668e-06,
      "loss": 0.0065,
      "step": 8001
    },
    {
      "epoch": 0.9261574074074074,
      "grad_norm": 0.08786170929670334,
      "learning_rate": 1.476851851851852e-06,
      "loss": 0.0094,
      "step": 8002
    },
    {
      "epoch": 0.9262731481481481,
      "grad_norm": 0.054775647819042206,
      "learning_rate": 1.4745370370370372e-06,
      "loss": 0.0089,
      "step": 8003
    },
    {
      "epoch": 0.9263888888888889,
      "grad_norm": 0.036956362426280975,
      "learning_rate": 1.4722222222222225e-06,
      "loss": 0.006,
      "step": 8004
    },
    {
      "epoch": 0.9265046296296297,
      "grad_norm": 0.04184100776910782,
      "learning_rate": 1.4699074074074074e-06,
      "loss": 0.0075,
      "step": 8005
    },
    {
      "epoch": 0.9266203703703704,
      "grad_norm": 0.03376943990588188,
      "learning_rate": 1.4675925925925927e-06,
      "loss": 0.006,
      "step": 8006
    },
    {
      "epoch": 0.9267361111111111,
      "grad_norm": 0.030653195455670357,
      "learning_rate": 1.465277777777778e-06,
      "loss": 0.0056,
      "step": 8007
    },
    {
      "epoch": 0.9268518518518518,
      "grad_norm": 0.1429537981748581,
      "learning_rate": 1.4629629629629629e-06,
      "loss": 0.0097,
      "step": 8008
    },
    {
      "epoch": 0.9269675925925925,
      "grad_norm": 0.041894588619470596,
      "learning_rate": 1.4606481481481482e-06,
      "loss": 0.0069,
      "step": 8009
    },
    {
      "epoch": 0.9270833333333334,
      "grad_norm": 0.04515566676855087,
      "learning_rate": 1.4583333333333335e-06,
      "loss": 0.0081,
      "step": 8010
    },
    {
      "epoch": 0.9271990740740741,
      "grad_norm": 0.03164801374077797,
      "learning_rate": 1.4560185185185186e-06,
      "loss": 0.0056,
      "step": 8011
    },
    {
      "epoch": 0.9273148148148148,
      "grad_norm": 0.05372222512960434,
      "learning_rate": 1.453703703703704e-06,
      "loss": 0.0095,
      "step": 8012
    },
    {
      "epoch": 0.9274305555555555,
      "grad_norm": 0.03569519519805908,
      "learning_rate": 1.4513888888888892e-06,
      "loss": 0.0062,
      "step": 8013
    },
    {
      "epoch": 0.9275462962962963,
      "grad_norm": 159.80982971191406,
      "learning_rate": 1.449074074074074e-06,
      "loss": 0.4123,
      "step": 8014
    },
    {
      "epoch": 0.9276620370370371,
      "grad_norm": 0.07087194919586182,
      "learning_rate": 1.4467592592592594e-06,
      "loss": 0.0096,
      "step": 8015
    },
    {
      "epoch": 0.9277777777777778,
      "grad_norm": 0.035795703530311584,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.0064,
      "step": 8016
    },
    {
      "epoch": 0.9278935185185185,
      "grad_norm": 0.10698067396879196,
      "learning_rate": 1.4421296296296298e-06,
      "loss": 0.007,
      "step": 8017
    },
    {
      "epoch": 0.9280092592592593,
      "grad_norm": 0.07256752997636795,
      "learning_rate": 1.439814814814815e-06,
      "loss": 0.0096,
      "step": 8018
    },
    {
      "epoch": 0.928125,
      "grad_norm": 0.0341336764395237,
      "learning_rate": 1.4375e-06,
      "loss": 0.0058,
      "step": 8019
    },
    {
      "epoch": 0.9282407407407407,
      "grad_norm": 0.039920318871736526,
      "learning_rate": 1.4351851851851853e-06,
      "loss": 0.0071,
      "step": 8020
    },
    {
      "epoch": 0.9283564814814815,
      "grad_norm": 0.06278174370527267,
      "learning_rate": 1.4328703703703706e-06,
      "loss": 0.0099,
      "step": 8021
    },
    {
      "epoch": 0.9284722222222223,
      "grad_norm": 0.044831328094005585,
      "learning_rate": 1.4305555555555555e-06,
      "loss": 0.0078,
      "step": 8022
    },
    {
      "epoch": 0.928587962962963,
      "grad_norm": 0.062284328043460846,
      "learning_rate": 1.4282407407407408e-06,
      "loss": 0.0109,
      "step": 8023
    },
    {
      "epoch": 0.9287037037037037,
      "grad_norm": 0.039872460067272186,
      "learning_rate": 1.4259259259259261e-06,
      "loss": 0.0052,
      "step": 8024
    },
    {
      "epoch": 0.9288194444444444,
      "grad_norm": 0.038973618298769,
      "learning_rate": 1.4236111111111112e-06,
      "loss": 0.0069,
      "step": 8025
    },
    {
      "epoch": 0.9289351851851851,
      "grad_norm": 0.05470157042145729,
      "learning_rate": 1.4212962962962965e-06,
      "loss": 0.0078,
      "step": 8026
    },
    {
      "epoch": 0.929050925925926,
      "grad_norm": 0.11593107879161835,
      "learning_rate": 1.4189814814814814e-06,
      "loss": 0.0107,
      "step": 8027
    },
    {
      "epoch": 0.9291666666666667,
      "grad_norm": 0.045624397695064545,
      "learning_rate": 1.4166666666666667e-06,
      "loss": 0.0082,
      "step": 8028
    },
    {
      "epoch": 0.9292824074074074,
      "grad_norm": 0.0360373817384243,
      "learning_rate": 1.414351851851852e-06,
      "loss": 0.0064,
      "step": 8029
    },
    {
      "epoch": 0.9293981481481481,
      "grad_norm": 0.036253027617931366,
      "learning_rate": 1.4120370370370371e-06,
      "loss": 0.0062,
      "step": 8030
    },
    {
      "epoch": 0.9295138888888889,
      "grad_norm": 0.04833802208304405,
      "learning_rate": 1.4097222222222222e-06,
      "loss": 0.0061,
      "step": 8031
    },
    {
      "epoch": 0.9296296296296296,
      "grad_norm": 0.058718111366033554,
      "learning_rate": 1.4074074074074075e-06,
      "loss": 0.0076,
      "step": 8032
    },
    {
      "epoch": 0.9297453703703704,
      "grad_norm": 0.03631036356091499,
      "learning_rate": 1.4050925925925926e-06,
      "loss": 0.0064,
      "step": 8033
    },
    {
      "epoch": 0.9298611111111111,
      "grad_norm": 0.042829181998968124,
      "learning_rate": 1.402777777777778e-06,
      "loss": 0.0074,
      "step": 8034
    },
    {
      "epoch": 0.9299768518518519,
      "grad_norm": 0.0362321212887764,
      "learning_rate": 1.4004629629629633e-06,
      "loss": 0.0064,
      "step": 8035
    },
    {
      "epoch": 0.9300925925925926,
      "grad_norm": 0.0493033267557621,
      "learning_rate": 1.3981481481481481e-06,
      "loss": 0.0078,
      "step": 8036
    },
    {
      "epoch": 0.9302083333333333,
      "grad_norm": 0.04511990770697594,
      "learning_rate": 1.3958333333333335e-06,
      "loss": 0.0058,
      "step": 8037
    },
    {
      "epoch": 0.930324074074074,
      "grad_norm": 0.037654466927051544,
      "learning_rate": 1.3935185185185188e-06,
      "loss": 0.0068,
      "step": 8038
    },
    {
      "epoch": 0.9304398148148149,
      "grad_norm": 0.044518645852804184,
      "learning_rate": 1.3912037037037039e-06,
      "loss": 0.0075,
      "step": 8039
    },
    {
      "epoch": 0.9305555555555556,
      "grad_norm": 0.032437924295663834,
      "learning_rate": 1.3888888888888892e-06,
      "loss": 0.0058,
      "step": 8040
    },
    {
      "epoch": 0.9306712962962963,
      "grad_norm": 0.03681188449263573,
      "learning_rate": 1.386574074074074e-06,
      "loss": 0.0066,
      "step": 8041
    },
    {
      "epoch": 0.930787037037037,
      "grad_norm": 0.0659627690911293,
      "learning_rate": 1.3842592592592594e-06,
      "loss": 0.0096,
      "step": 8042
    },
    {
      "epoch": 0.9309027777777777,
      "grad_norm": 0.05234047397971153,
      "learning_rate": 1.3819444444444447e-06,
      "loss": 0.0092,
      "step": 8043
    },
    {
      "epoch": 0.9310185185185185,
      "grad_norm": 0.07653114944696426,
      "learning_rate": 1.3796296296296296e-06,
      "loss": 0.0098,
      "step": 8044
    },
    {
      "epoch": 0.9311342592592593,
      "grad_norm": 0.080706387758255,
      "learning_rate": 1.3773148148148149e-06,
      "loss": 0.008,
      "step": 8045
    },
    {
      "epoch": 0.93125,
      "grad_norm": 0.04272380843758583,
      "learning_rate": 1.3750000000000002e-06,
      "loss": 0.0074,
      "step": 8046
    },
    {
      "epoch": 0.9313657407407407,
      "grad_norm": 0.045416925102472305,
      "learning_rate": 1.3726851851851853e-06,
      "loss": 0.0081,
      "step": 8047
    },
    {
      "epoch": 0.9314814814814815,
      "grad_norm": 0.04291477054357529,
      "learning_rate": 1.3703703703703706e-06,
      "loss": 0.0076,
      "step": 8048
    },
    {
      "epoch": 0.9315972222222222,
      "grad_norm": 0.040087662637233734,
      "learning_rate": 1.3680555555555559e-06,
      "loss": 0.007,
      "step": 8049
    },
    {
      "epoch": 0.9317129629629629,
      "grad_norm": 0.03494901582598686,
      "learning_rate": 1.3657407407407408e-06,
      "loss": 0.0062,
      "step": 8050
    },
    {
      "epoch": 0.9318287037037037,
      "grad_norm": 0.04626864567399025,
      "learning_rate": 1.363425925925926e-06,
      "loss": 0.0081,
      "step": 8051
    },
    {
      "epoch": 0.9319444444444445,
      "grad_norm": 0.0851207822561264,
      "learning_rate": 1.3611111111111112e-06,
      "loss": 0.0085,
      "step": 8052
    },
    {
      "epoch": 0.9320601851851852,
      "grad_norm": 0.051995500922203064,
      "learning_rate": 1.3587962962962965e-06,
      "loss": 0.0093,
      "step": 8053
    },
    {
      "epoch": 0.9321759259259259,
      "grad_norm": 0.13298183679580688,
      "learning_rate": 1.3564814814814816e-06,
      "loss": 0.0094,
      "step": 8054
    },
    {
      "epoch": 0.9322916666666666,
      "grad_norm": 0.04891672730445862,
      "learning_rate": 1.3541666666666667e-06,
      "loss": 0.0081,
      "step": 8055
    },
    {
      "epoch": 0.9324074074074075,
      "grad_norm": 0.0557192824780941,
      "learning_rate": 1.351851851851852e-06,
      "loss": 0.0102,
      "step": 8056
    },
    {
      "epoch": 0.9325231481481482,
      "grad_norm": 0.030500218272209167,
      "learning_rate": 1.3495370370370373e-06,
      "loss": 0.0055,
      "step": 8057
    },
    {
      "epoch": 0.9326388888888889,
      "grad_norm": 0.030750306323170662,
      "learning_rate": 1.3472222222222222e-06,
      "loss": 0.0056,
      "step": 8058
    },
    {
      "epoch": 0.9327546296296296,
      "grad_norm": 0.03537612780928612,
      "learning_rate": 1.3449074074074075e-06,
      "loss": 0.0062,
      "step": 8059
    },
    {
      "epoch": 0.9328703703703703,
      "grad_norm": 0.03308820724487305,
      "learning_rate": 1.3425925925925928e-06,
      "loss": 0.0056,
      "step": 8060
    },
    {
      "epoch": 0.9329861111111111,
      "grad_norm": 0.03413864225149155,
      "learning_rate": 1.340277777777778e-06,
      "loss": 0.0061,
      "step": 8061
    },
    {
      "epoch": 0.9331018518518519,
      "grad_norm": 0.050211165100336075,
      "learning_rate": 1.3379629629629632e-06,
      "loss": 0.0089,
      "step": 8062
    },
    {
      "epoch": 0.9332175925925926,
      "grad_norm": 0.031918201595544815,
      "learning_rate": 1.335648148148148e-06,
      "loss": 0.0057,
      "step": 8063
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 0.04048813506960869,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0053,
      "step": 8064
    },
    {
      "epoch": 0.9334490740740741,
      "grad_norm": 0.04072058945894241,
      "learning_rate": 1.3310185185185187e-06,
      "loss": 0.0073,
      "step": 8065
    },
    {
      "epoch": 0.9335648148148148,
      "grad_norm": 0.03918559104204178,
      "learning_rate": 1.3287037037037038e-06,
      "loss": 0.0069,
      "step": 8066
    },
    {
      "epoch": 0.9336805555555555,
      "grad_norm": 0.05621382221579552,
      "learning_rate": 1.326388888888889e-06,
      "loss": 0.0092,
      "step": 8067
    },
    {
      "epoch": 0.9337962962962963,
      "grad_norm": 0.04237990081310272,
      "learning_rate": 1.3240740740740742e-06,
      "loss": 0.0076,
      "step": 8068
    },
    {
      "epoch": 0.9339120370370371,
      "grad_norm": 0.046888697892427444,
      "learning_rate": 1.3217592592592593e-06,
      "loss": 0.0085,
      "step": 8069
    },
    {
      "epoch": 0.9340277777777778,
      "grad_norm": 0.039652105420827866,
      "learning_rate": 1.3194444444444446e-06,
      "loss": 0.007,
      "step": 8070
    },
    {
      "epoch": 0.9341435185185185,
      "grad_norm": 0.032527513802051544,
      "learning_rate": 1.31712962962963e-06,
      "loss": 0.0058,
      "step": 8071
    },
    {
      "epoch": 0.9342592592592592,
      "grad_norm": 0.0406927689909935,
      "learning_rate": 1.3148148148148148e-06,
      "loss": 0.0072,
      "step": 8072
    },
    {
      "epoch": 0.934375,
      "grad_norm": 0.4701458811759949,
      "learning_rate": 1.3125000000000001e-06,
      "loss": 0.0099,
      "step": 8073
    },
    {
      "epoch": 0.9344907407407408,
      "grad_norm": 0.26219454407691956,
      "learning_rate": 1.3101851851851852e-06,
      "loss": 0.0096,
      "step": 8074
    },
    {
      "epoch": 0.9346064814814815,
      "grad_norm": 0.07983727753162384,
      "learning_rate": 1.3078703703703705e-06,
      "loss": 0.0114,
      "step": 8075
    },
    {
      "epoch": 0.9347222222222222,
      "grad_norm": 0.059177424758672714,
      "learning_rate": 1.3055555555555556e-06,
      "loss": 0.0087,
      "step": 8076
    },
    {
      "epoch": 0.934837962962963,
      "grad_norm": 0.04503477364778519,
      "learning_rate": 1.3032407407407407e-06,
      "loss": 0.0079,
      "step": 8077
    },
    {
      "epoch": 0.9349537037037037,
      "grad_norm": 0.06547987461090088,
      "learning_rate": 1.300925925925926e-06,
      "loss": 0.0086,
      "step": 8078
    },
    {
      "epoch": 0.9350694444444444,
      "grad_norm": 0.049652326852083206,
      "learning_rate": 1.2986111111111114e-06,
      "loss": 0.0088,
      "step": 8079
    },
    {
      "epoch": 0.9351851851851852,
      "grad_norm": 0.05315178632736206,
      "learning_rate": 1.2962962962962962e-06,
      "loss": 0.0082,
      "step": 8080
    },
    {
      "epoch": 0.935300925925926,
      "grad_norm": 0.037504781037569046,
      "learning_rate": 1.2939814814814816e-06,
      "loss": 0.0066,
      "step": 8081
    },
    {
      "epoch": 0.9354166666666667,
      "grad_norm": 0.04625433310866356,
      "learning_rate": 1.2916666666666669e-06,
      "loss": 0.0079,
      "step": 8082
    },
    {
      "epoch": 0.9355324074074074,
      "grad_norm": 0.0705309733748436,
      "learning_rate": 1.289351851851852e-06,
      "loss": 0.0082,
      "step": 8083
    },
    {
      "epoch": 0.9356481481481481,
      "grad_norm": 0.04294692724943161,
      "learning_rate": 1.2870370370370373e-06,
      "loss": 0.0074,
      "step": 8084
    },
    {
      "epoch": 0.9357638888888888,
      "grad_norm": 0.03088567592203617,
      "learning_rate": 1.2847222222222222e-06,
      "loss": 0.0056,
      "step": 8085
    },
    {
      "epoch": 0.9358796296296297,
      "grad_norm": 0.03782088682055473,
      "learning_rate": 1.2824074074074075e-06,
      "loss": 0.0067,
      "step": 8086
    },
    {
      "epoch": 0.9359953703703704,
      "grad_norm": 0.04864975064992905,
      "learning_rate": 1.2800925925925928e-06,
      "loss": 0.0085,
      "step": 8087
    },
    {
      "epoch": 0.9361111111111111,
      "grad_norm": 0.03498414158821106,
      "learning_rate": 1.2777777777777779e-06,
      "loss": 0.0062,
      "step": 8088
    },
    {
      "epoch": 0.9362268518518518,
      "grad_norm": 0.055210597813129425,
      "learning_rate": 1.275462962962963e-06,
      "loss": 0.0096,
      "step": 8089
    },
    {
      "epoch": 0.9363425925925926,
      "grad_norm": 0.035410452634096146,
      "learning_rate": 1.2731481481481483e-06,
      "loss": 0.0063,
      "step": 8090
    },
    {
      "epoch": 0.9364583333333333,
      "grad_norm": 0.05141863226890564,
      "learning_rate": 1.2708333333333334e-06,
      "loss": 0.009,
      "step": 8091
    },
    {
      "epoch": 0.9365740740740741,
      "grad_norm": 0.04672470688819885,
      "learning_rate": 1.2685185185185187e-06,
      "loss": 0.0084,
      "step": 8092
    },
    {
      "epoch": 0.9366898148148148,
      "grad_norm": 0.046809010207653046,
      "learning_rate": 1.266203703703704e-06,
      "loss": 0.0082,
      "step": 8093
    },
    {
      "epoch": 0.9368055555555556,
      "grad_norm": 115.48377227783203,
      "learning_rate": 1.2638888888888889e-06,
      "loss": 2.1518,
      "step": 8094
    },
    {
      "epoch": 0.9369212962962963,
      "grad_norm": 0.19800333678722382,
      "learning_rate": 1.2615740740740742e-06,
      "loss": 0.0088,
      "step": 8095
    },
    {
      "epoch": 0.937037037037037,
      "grad_norm": 0.045831210911273956,
      "learning_rate": 1.2592592592592593e-06,
      "loss": 0.0083,
      "step": 8096
    },
    {
      "epoch": 0.9371527777777777,
      "grad_norm": 0.04376426711678505,
      "learning_rate": 1.2569444444444446e-06,
      "loss": 0.0078,
      "step": 8097
    },
    {
      "epoch": 0.9372685185185186,
      "grad_norm": 0.04547553509473801,
      "learning_rate": 1.25462962962963e-06,
      "loss": 0.0074,
      "step": 8098
    },
    {
      "epoch": 0.9373842592592593,
      "grad_norm": 0.039106905460357666,
      "learning_rate": 1.2523148148148148e-06,
      "loss": 0.007,
      "step": 8099
    },
    {
      "epoch": 0.9375,
      "grad_norm": 0.31813257932662964,
      "learning_rate": 1.25e-06,
      "loss": 0.0131,
      "step": 8100
    },
    {
      "epoch": 0.9376157407407407,
      "grad_norm": 0.050188060849905014,
      "learning_rate": 1.2476851851851852e-06,
      "loss": 0.009,
      "step": 8101
    },
    {
      "epoch": 0.9377314814814814,
      "grad_norm": 0.030659915879368782,
      "learning_rate": 1.2453703703703705e-06,
      "loss": 0.0055,
      "step": 8102
    },
    {
      "epoch": 0.9378472222222223,
      "grad_norm": 0.045933742076158524,
      "learning_rate": 1.2430555555555556e-06,
      "loss": 0.008,
      "step": 8103
    },
    {
      "epoch": 0.937962962962963,
      "grad_norm": 0.047376297414302826,
      "learning_rate": 1.240740740740741e-06,
      "loss": 0.0083,
      "step": 8104
    },
    {
      "epoch": 0.9380787037037037,
      "grad_norm": 0.04158911481499672,
      "learning_rate": 1.238425925925926e-06,
      "loss": 0.0075,
      "step": 8105
    },
    {
      "epoch": 0.9381944444444444,
      "grad_norm": 0.03284257650375366,
      "learning_rate": 1.2361111111111113e-06,
      "loss": 0.0059,
      "step": 8106
    },
    {
      "epoch": 0.9383101851851852,
      "grad_norm": 0.14030788838863373,
      "learning_rate": 1.2337962962962964e-06,
      "loss": 0.0098,
      "step": 8107
    },
    {
      "epoch": 0.9384259259259259,
      "grad_norm": 0.05206404626369476,
      "learning_rate": 1.2314814814814815e-06,
      "loss": 0.0086,
      "step": 8108
    },
    {
      "epoch": 0.9385416666666667,
      "grad_norm": 0.03437306731939316,
      "learning_rate": 1.2291666666666666e-06,
      "loss": 0.0061,
      "step": 8109
    },
    {
      "epoch": 0.9386574074074074,
      "grad_norm": 0.07585122436285019,
      "learning_rate": 1.226851851851852e-06,
      "loss": 0.0095,
      "step": 8110
    },
    {
      "epoch": 0.9387731481481482,
      "grad_norm": 0.05990252271294594,
      "learning_rate": 1.2245370370370372e-06,
      "loss": 0.0088,
      "step": 8111
    },
    {
      "epoch": 0.9388888888888889,
      "grad_norm": 0.03862154483795166,
      "learning_rate": 1.2222222222222223e-06,
      "loss": 0.0067,
      "step": 8112
    },
    {
      "epoch": 0.9390046296296296,
      "grad_norm": 0.05808322876691818,
      "learning_rate": 1.2199074074074076e-06,
      "loss": 0.0083,
      "step": 8113
    },
    {
      "epoch": 0.9391203703703703,
      "grad_norm": 0.035208359360694885,
      "learning_rate": 1.2175925925925927e-06,
      "loss": 0.0063,
      "step": 8114
    },
    {
      "epoch": 0.9392361111111112,
      "grad_norm": 120.0123062133789,
      "learning_rate": 1.2152777777777778e-06,
      "loss": 0.587,
      "step": 8115
    },
    {
      "epoch": 0.9393518518518519,
      "grad_norm": 0.04240266606211662,
      "learning_rate": 1.212962962962963e-06,
      "loss": 0.0074,
      "step": 8116
    },
    {
      "epoch": 0.9394675925925926,
      "grad_norm": 0.18273909389972687,
      "learning_rate": 1.2106481481481482e-06,
      "loss": 0.01,
      "step": 8117
    },
    {
      "epoch": 0.9395833333333333,
      "grad_norm": 0.04834263399243355,
      "learning_rate": 1.2083333333333333e-06,
      "loss": 0.0085,
      "step": 8118
    },
    {
      "epoch": 0.939699074074074,
      "grad_norm": 0.04381047189235687,
      "learning_rate": 1.2060185185185186e-06,
      "loss": 0.0077,
      "step": 8119
    },
    {
      "epoch": 0.9398148148148148,
      "grad_norm": 0.038742534816265106,
      "learning_rate": 1.2037037037037037e-06,
      "loss": 0.0068,
      "step": 8120
    },
    {
      "epoch": 0.9399305555555556,
      "grad_norm": 0.15525028109550476,
      "learning_rate": 1.201388888888889e-06,
      "loss": 0.01,
      "step": 8121
    },
    {
      "epoch": 0.9400462962962963,
      "grad_norm": 0.07788681983947754,
      "learning_rate": 1.1990740740740742e-06,
      "loss": 0.0109,
      "step": 8122
    },
    {
      "epoch": 0.940162037037037,
      "grad_norm": 0.4480229318141937,
      "learning_rate": 1.1967592592592593e-06,
      "loss": 0.0119,
      "step": 8123
    },
    {
      "epoch": 0.9402777777777778,
      "grad_norm": 0.0436796098947525,
      "learning_rate": 1.1944444444444446e-06,
      "loss": 0.0077,
      "step": 8124
    },
    {
      "epoch": 0.9403935185185185,
      "grad_norm": 0.03446150943636894,
      "learning_rate": 1.1921296296296297e-06,
      "loss": 0.0062,
      "step": 8125
    },
    {
      "epoch": 0.9405092592592592,
      "grad_norm": 98.34534454345703,
      "learning_rate": 1.189814814814815e-06,
      "loss": 1.8319,
      "step": 8126
    },
    {
      "epoch": 0.940625,
      "grad_norm": 0.10892033576965332,
      "learning_rate": 1.1875e-06,
      "loss": 0.0078,
      "step": 8127
    },
    {
      "epoch": 0.9407407407407408,
      "grad_norm": 0.04619520902633667,
      "learning_rate": 1.1851851851851854e-06,
      "loss": 0.008,
      "step": 8128
    },
    {
      "epoch": 0.9408564814814815,
      "grad_norm": 0.05933007597923279,
      "learning_rate": 1.1828703703703705e-06,
      "loss": 0.0105,
      "step": 8129
    },
    {
      "epoch": 0.9409722222222222,
      "grad_norm": 0.05494268983602524,
      "learning_rate": 1.1805555555555556e-06,
      "loss": 0.0068,
      "step": 8130
    },
    {
      "epoch": 0.9410879629629629,
      "grad_norm": 0.037868745625019073,
      "learning_rate": 1.1782407407407407e-06,
      "loss": 0.0066,
      "step": 8131
    },
    {
      "epoch": 0.9412037037037037,
      "grad_norm": 77.96990203857422,
      "learning_rate": 1.175925925925926e-06,
      "loss": 0.1805,
      "step": 8132
    },
    {
      "epoch": 0.9413194444444445,
      "grad_norm": 0.04256569221615791,
      "learning_rate": 1.1736111111111113e-06,
      "loss": 0.0075,
      "step": 8133
    },
    {
      "epoch": 0.9414351851851852,
      "grad_norm": 0.04896625503897667,
      "learning_rate": 1.1712962962962964e-06,
      "loss": 0.0086,
      "step": 8134
    },
    {
      "epoch": 0.9415509259259259,
      "grad_norm": 0.034400783479213715,
      "learning_rate": 1.1689814814814817e-06,
      "loss": 0.0062,
      "step": 8135
    },
    {
      "epoch": 0.9416666666666667,
      "grad_norm": 0.04332819581031799,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 0.0073,
      "step": 8136
    },
    {
      "epoch": 0.9417824074074074,
      "grad_norm": 0.03478693589568138,
      "learning_rate": 1.1643518518518519e-06,
      "loss": 0.0062,
      "step": 8137
    },
    {
      "epoch": 0.9418981481481481,
      "grad_norm": 0.08825667947530746,
      "learning_rate": 1.162037037037037e-06,
      "loss": 0.0108,
      "step": 8138
    },
    {
      "epoch": 0.9420138888888889,
      "grad_norm": 0.07347846031188965,
      "learning_rate": 1.1597222222222223e-06,
      "loss": 0.0086,
      "step": 8139
    },
    {
      "epoch": 0.9421296296296297,
      "grad_norm": 0.04955270141363144,
      "learning_rate": 1.1574074074074076e-06,
      "loss": 0.0084,
      "step": 8140
    },
    {
      "epoch": 0.9422453703703704,
      "grad_norm": 0.03179949149489403,
      "learning_rate": 1.1550925925925927e-06,
      "loss": 0.0055,
      "step": 8141
    },
    {
      "epoch": 0.9423611111111111,
      "grad_norm": 0.041085079312324524,
      "learning_rate": 1.152777777777778e-06,
      "loss": 0.0074,
      "step": 8142
    },
    {
      "epoch": 0.9424768518518518,
      "grad_norm": 0.04321593791246414,
      "learning_rate": 1.1504629629629631e-06,
      "loss": 0.0075,
      "step": 8143
    },
    {
      "epoch": 0.9425925925925925,
      "grad_norm": 0.03620999678969383,
      "learning_rate": 1.1481481481481482e-06,
      "loss": 0.0061,
      "step": 8144
    },
    {
      "epoch": 0.9427083333333334,
      "grad_norm": 3.334834337234497,
      "learning_rate": 1.1458333333333333e-06,
      "loss": 0.0175,
      "step": 8145
    },
    {
      "epoch": 0.9428240740740741,
      "grad_norm": 0.14210133254528046,
      "learning_rate": 1.1435185185185186e-06,
      "loss": 0.009,
      "step": 8146
    },
    {
      "epoch": 0.9429398148148148,
      "grad_norm": 0.03452899679541588,
      "learning_rate": 1.1412037037037037e-06,
      "loss": 0.0062,
      "step": 8147
    },
    {
      "epoch": 0.9430555555555555,
      "grad_norm": 0.0353049673140049,
      "learning_rate": 1.138888888888889e-06,
      "loss": 0.0063,
      "step": 8148
    },
    {
      "epoch": 0.9431712962962963,
      "grad_norm": 0.7087665796279907,
      "learning_rate": 1.1365740740740741e-06,
      "loss": 0.0081,
      "step": 8149
    },
    {
      "epoch": 0.9432870370370371,
      "grad_norm": 0.050701212137937546,
      "learning_rate": 1.1342592592592594e-06,
      "loss": 0.0091,
      "step": 8150
    },
    {
      "epoch": 0.9434027777777778,
      "grad_norm": 0.0481921024620533,
      "learning_rate": 1.1319444444444445e-06,
      "loss": 0.0088,
      "step": 8151
    },
    {
      "epoch": 0.9435185185185185,
      "grad_norm": 0.04201119393110275,
      "learning_rate": 1.1296296296296296e-06,
      "loss": 0.0076,
      "step": 8152
    },
    {
      "epoch": 0.9436342592592593,
      "grad_norm": 0.12486241012811661,
      "learning_rate": 1.127314814814815e-06,
      "loss": 0.0072,
      "step": 8153
    },
    {
      "epoch": 0.94375,
      "grad_norm": 0.043409623205661774,
      "learning_rate": 1.125e-06,
      "loss": 0.0079,
      "step": 8154
    },
    {
      "epoch": 0.9438657407407407,
      "grad_norm": 0.040809743106365204,
      "learning_rate": 1.1226851851851853e-06,
      "loss": 0.0071,
      "step": 8155
    },
    {
      "epoch": 0.9439814814814815,
      "grad_norm": 0.4137842357158661,
      "learning_rate": 1.1203703703703704e-06,
      "loss": 0.0128,
      "step": 8156
    },
    {
      "epoch": 0.9440972222222223,
      "grad_norm": 0.05122223496437073,
      "learning_rate": 1.1180555555555557e-06,
      "loss": 0.0085,
      "step": 8157
    },
    {
      "epoch": 0.944212962962963,
      "grad_norm": 0.033276114612817764,
      "learning_rate": 1.1157407407407408e-06,
      "loss": 0.0059,
      "step": 8158
    },
    {
      "epoch": 0.9443287037037037,
      "grad_norm": 0.03357474133372307,
      "learning_rate": 1.113425925925926e-06,
      "loss": 0.006,
      "step": 8159
    },
    {
      "epoch": 0.9444444444444444,
      "grad_norm": 0.03282049670815468,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0059,
      "step": 8160
    },
    {
      "epoch": 0.9445601851851851,
      "grad_norm": 0.04413912072777748,
      "learning_rate": 1.1087962962962963e-06,
      "loss": 0.0076,
      "step": 8161
    },
    {
      "epoch": 0.944675925925926,
      "grad_norm": 0.0409933440387249,
      "learning_rate": 1.1064814814814817e-06,
      "loss": 0.0053,
      "step": 8162
    },
    {
      "epoch": 0.9447916666666667,
      "grad_norm": 12.421631813049316,
      "learning_rate": 1.1041666666666668e-06,
      "loss": 2.9954,
      "step": 8163
    },
    {
      "epoch": 0.9449074074074074,
      "grad_norm": 0.05234033241868019,
      "learning_rate": 1.101851851851852e-06,
      "loss": 0.0083,
      "step": 8164
    },
    {
      "epoch": 0.9450231481481481,
      "grad_norm": 0.04052349179983139,
      "learning_rate": 1.0995370370370372e-06,
      "loss": 0.007,
      "step": 8165
    },
    {
      "epoch": 0.9451388888888889,
      "grad_norm": 0.04282239452004433,
      "learning_rate": 1.0972222222222223e-06,
      "loss": 0.0076,
      "step": 8166
    },
    {
      "epoch": 0.9452546296296296,
      "grad_norm": 0.06195690110325813,
      "learning_rate": 1.0949074074074074e-06,
      "loss": 0.0076,
      "step": 8167
    },
    {
      "epoch": 0.9453703703703704,
      "grad_norm": 0.035115282982587814,
      "learning_rate": 1.0925925925925927e-06,
      "loss": 0.0064,
      "step": 8168
    },
    {
      "epoch": 0.9454861111111111,
      "grad_norm": 0.03448045626282692,
      "learning_rate": 1.090277777777778e-06,
      "loss": 0.006,
      "step": 8169
    },
    {
      "epoch": 0.9456018518518519,
      "grad_norm": 0.06150558218359947,
      "learning_rate": 1.087962962962963e-06,
      "loss": 0.0076,
      "step": 8170
    },
    {
      "epoch": 0.9457175925925926,
      "grad_norm": 0.03384535014629364,
      "learning_rate": 1.0856481481481482e-06,
      "loss": 0.0061,
      "step": 8171
    },
    {
      "epoch": 0.9458333333333333,
      "grad_norm": 0.03175437077879906,
      "learning_rate": 1.0833333333333335e-06,
      "loss": 0.0057,
      "step": 8172
    },
    {
      "epoch": 0.945949074074074,
      "grad_norm": 0.031164921820163727,
      "learning_rate": 1.0810185185185186e-06,
      "loss": 0.0055,
      "step": 8173
    },
    {
      "epoch": 0.9460648148148149,
      "grad_norm": 0.03459278866648674,
      "learning_rate": 1.0787037037037037e-06,
      "loss": 0.0057,
      "step": 8174
    },
    {
      "epoch": 0.9461805555555556,
      "grad_norm": 0.05095906928181648,
      "learning_rate": 1.076388888888889e-06,
      "loss": 0.0089,
      "step": 8175
    },
    {
      "epoch": 0.9462962962962963,
      "grad_norm": 0.040827151387929916,
      "learning_rate": 1.074074074074074e-06,
      "loss": 0.0067,
      "step": 8176
    },
    {
      "epoch": 0.946412037037037,
      "grad_norm": 0.03566853329539299,
      "learning_rate": 1.0717592592592594e-06,
      "loss": 0.0061,
      "step": 8177
    },
    {
      "epoch": 0.9465277777777777,
      "grad_norm": 0.053060226142406464,
      "learning_rate": 1.0694444444444445e-06,
      "loss": 0.0092,
      "step": 8178
    },
    {
      "epoch": 0.9466435185185185,
      "grad_norm": 0.041771192103624344,
      "learning_rate": 1.0671296296296298e-06,
      "loss": 0.0075,
      "step": 8179
    },
    {
      "epoch": 0.9467592592592593,
      "grad_norm": 0.047453686594963074,
      "learning_rate": 1.0648148148148149e-06,
      "loss": 0.0082,
      "step": 8180
    },
    {
      "epoch": 0.946875,
      "grad_norm": 0.03556579351425171,
      "learning_rate": 1.0625e-06,
      "loss": 0.0062,
      "step": 8181
    },
    {
      "epoch": 0.9469907407407407,
      "grad_norm": 0.047764554619789124,
      "learning_rate": 1.0601851851851853e-06,
      "loss": 0.0082,
      "step": 8182
    },
    {
      "epoch": 0.9471064814814815,
      "grad_norm": 0.04960167035460472,
      "learning_rate": 1.0578703703703704e-06,
      "loss": 0.0064,
      "step": 8183
    },
    {
      "epoch": 0.9472222222222222,
      "grad_norm": 0.05129818990826607,
      "learning_rate": 1.0555555555555557e-06,
      "loss": 0.009,
      "step": 8184
    },
    {
      "epoch": 0.9473379629629629,
      "grad_norm": 9.023669242858887,
      "learning_rate": 1.0532407407407408e-06,
      "loss": 2.8669,
      "step": 8185
    },
    {
      "epoch": 0.9474537037037037,
      "grad_norm": 0.04194221645593643,
      "learning_rate": 1.0509259259259261e-06,
      "loss": 0.0075,
      "step": 8186
    },
    {
      "epoch": 0.9475694444444445,
      "grad_norm": 0.030916964635252953,
      "learning_rate": 1.0486111111111112e-06,
      "loss": 0.0055,
      "step": 8187
    },
    {
      "epoch": 0.9476851851851852,
      "grad_norm": 2.0570313930511475,
      "learning_rate": 1.0462962962962963e-06,
      "loss": 0.0171,
      "step": 8188
    },
    {
      "epoch": 0.9478009259259259,
      "grad_norm": 0.031397681683301926,
      "learning_rate": 1.0439814814814814e-06,
      "loss": 0.0055,
      "step": 8189
    },
    {
      "epoch": 0.9479166666666666,
      "grad_norm": 0.06696432828903198,
      "learning_rate": 1.0416666666666667e-06,
      "loss": 0.0097,
      "step": 8190
    },
    {
      "epoch": 0.9480324074074075,
      "grad_norm": 0.03223119303584099,
      "learning_rate": 1.039351851851852e-06,
      "loss": 0.0058,
      "step": 8191
    },
    {
      "epoch": 0.9481481481481482,
      "grad_norm": 0.03798380494117737,
      "learning_rate": 1.0370370370370371e-06,
      "loss": 0.0062,
      "step": 8192
    },
    {
      "epoch": 0.9482638888888889,
      "grad_norm": 0.05679183080792427,
      "learning_rate": 1.0347222222222222e-06,
      "loss": 0.0074,
      "step": 8193
    },
    {
      "epoch": 0.9483796296296296,
      "grad_norm": 0.030807064846158028,
      "learning_rate": 1.0324074074074075e-06,
      "loss": 0.0056,
      "step": 8194
    },
    {
      "epoch": 0.9484953703703703,
      "grad_norm": 0.032966334372758865,
      "learning_rate": 1.0300925925925926e-06,
      "loss": 0.0059,
      "step": 8195
    },
    {
      "epoch": 0.9486111111111111,
      "grad_norm": 0.042174894362688065,
      "learning_rate": 1.0277777777777777e-06,
      "loss": 0.0073,
      "step": 8196
    },
    {
      "epoch": 0.9487268518518519,
      "grad_norm": 0.058714237064123154,
      "learning_rate": 1.025462962962963e-06,
      "loss": 0.0102,
      "step": 8197
    },
    {
      "epoch": 0.9488425925925926,
      "grad_norm": 0.06129659339785576,
      "learning_rate": 1.0231481481481483e-06,
      "loss": 0.0089,
      "step": 8198
    },
    {
      "epoch": 0.9489583333333333,
      "grad_norm": 0.04808470234274864,
      "learning_rate": 1.0208333333333334e-06,
      "loss": 0.0085,
      "step": 8199
    },
    {
      "epoch": 0.9490740740740741,
      "grad_norm": 0.04844563081860542,
      "learning_rate": 1.0185185185185185e-06,
      "loss": 0.0085,
      "step": 8200
    },
    {
      "epoch": 0.9491898148148148,
      "grad_norm": 0.07239806652069092,
      "learning_rate": 1.0162037037037038e-06,
      "loss": 0.0095,
      "step": 8201
    },
    {
      "epoch": 0.9493055555555555,
      "grad_norm": 0.04509960487484932,
      "learning_rate": 1.013888888888889e-06,
      "loss": 0.0081,
      "step": 8202
    },
    {
      "epoch": 0.9494212962962963,
      "grad_norm": 0.052495382726192474,
      "learning_rate": 1.011574074074074e-06,
      "loss": 0.0091,
      "step": 8203
    },
    {
      "epoch": 0.9495370370370371,
      "grad_norm": 0.08873393386602402,
      "learning_rate": 1.0092592592592594e-06,
      "loss": 0.0094,
      "step": 8204
    },
    {
      "epoch": 0.9496527777777778,
      "grad_norm": 0.05237030237913132,
      "learning_rate": 1.0069444444444447e-06,
      "loss": 0.0085,
      "step": 8205
    },
    {
      "epoch": 0.9497685185185185,
      "grad_norm": 0.06529327481985092,
      "learning_rate": 1.0046296296296298e-06,
      "loss": 0.0085,
      "step": 8206
    },
    {
      "epoch": 0.9498842592592592,
      "grad_norm": 0.03251013532280922,
      "learning_rate": 1.0023148148148149e-06,
      "loss": 0.0058,
      "step": 8207
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.044469937682151794,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0079,
      "step": 8208
    },
    {
      "epoch": 0.9501157407407408,
      "grad_norm": 0.0487905852496624,
      "learning_rate": 9.976851851851853e-07,
      "loss": 0.0077,
      "step": 8209
    },
    {
      "epoch": 0.9502314814814815,
      "grad_norm": 0.07525423914194107,
      "learning_rate": 9.953703703703704e-07,
      "loss": 0.0095,
      "step": 8210
    },
    {
      "epoch": 0.9503472222222222,
      "grad_norm": 0.034183185547590256,
      "learning_rate": 9.930555555555557e-07,
      "loss": 0.0061,
      "step": 8211
    },
    {
      "epoch": 0.950462962962963,
      "grad_norm": 128.90374755859375,
      "learning_rate": 9.907407407407408e-07,
      "loss": 0.3518,
      "step": 8212
    },
    {
      "epoch": 0.9505787037037037,
      "grad_norm": 0.05607857555150986,
      "learning_rate": 9.88425925925926e-07,
      "loss": 0.0097,
      "step": 8213
    },
    {
      "epoch": 0.9506944444444444,
      "grad_norm": 0.03397873416543007,
      "learning_rate": 9.861111111111112e-07,
      "loss": 0.0061,
      "step": 8214
    },
    {
      "epoch": 0.9508101851851852,
      "grad_norm": 0.035536862909793854,
      "learning_rate": 9.837962962962965e-07,
      "loss": 0.0064,
      "step": 8215
    },
    {
      "epoch": 0.950925925925926,
      "grad_norm": 0.03146373853087425,
      "learning_rate": 9.814814814814816e-07,
      "loss": 0.0056,
      "step": 8216
    },
    {
      "epoch": 0.9510416666666667,
      "grad_norm": 0.05555451288819313,
      "learning_rate": 9.791666666666667e-07,
      "loss": 0.0102,
      "step": 8217
    },
    {
      "epoch": 0.9511574074074074,
      "grad_norm": 0.04910699650645256,
      "learning_rate": 9.76851851851852e-07,
      "loss": 0.0089,
      "step": 8218
    },
    {
      "epoch": 0.9512731481481481,
      "grad_norm": 0.06507011502981186,
      "learning_rate": 9.74537037037037e-07,
      "loss": 0.0085,
      "step": 8219
    },
    {
      "epoch": 0.9513888888888888,
      "grad_norm": 0.09302808344364166,
      "learning_rate": 9.722222222222224e-07,
      "loss": 0.0093,
      "step": 8220
    },
    {
      "epoch": 0.9515046296296297,
      "grad_norm": 0.04729054868221283,
      "learning_rate": 9.699074074074075e-07,
      "loss": 0.0079,
      "step": 8221
    },
    {
      "epoch": 0.9516203703703704,
      "grad_norm": 0.0519287995994091,
      "learning_rate": 9.675925925925926e-07,
      "loss": 0.0083,
      "step": 8222
    },
    {
      "epoch": 0.9517361111111111,
      "grad_norm": 0.031602319329977036,
      "learning_rate": 9.65277777777778e-07,
      "loss": 0.0057,
      "step": 8223
    },
    {
      "epoch": 0.9518518518518518,
      "grad_norm": 0.03619558364152908,
      "learning_rate": 9.62962962962963e-07,
      "loss": 0.0065,
      "step": 8224
    },
    {
      "epoch": 0.9519675925925926,
      "grad_norm": 0.05723598971962929,
      "learning_rate": 9.60648148148148e-07,
      "loss": 0.0074,
      "step": 8225
    },
    {
      "epoch": 0.9520833333333333,
      "grad_norm": 0.038406625390052795,
      "learning_rate": 9.583333333333334e-07,
      "loss": 0.0068,
      "step": 8226
    },
    {
      "epoch": 0.9521990740740741,
      "grad_norm": 0.04346223920583725,
      "learning_rate": 9.560185185185187e-07,
      "loss": 0.0079,
      "step": 8227
    },
    {
      "epoch": 0.9523148148148148,
      "grad_norm": 0.04859317094087601,
      "learning_rate": 9.537037037037038e-07,
      "loss": 0.0082,
      "step": 8228
    },
    {
      "epoch": 0.9524305555555556,
      "grad_norm": 0.03299672529101372,
      "learning_rate": 9.513888888888889e-07,
      "loss": 0.0057,
      "step": 8229
    },
    {
      "epoch": 0.9525462962962963,
      "grad_norm": 6.999440670013428,
      "learning_rate": 9.490740740740742e-07,
      "loss": 0.0303,
      "step": 8230
    },
    {
      "epoch": 0.952662037037037,
      "grad_norm": 0.04352245107293129,
      "learning_rate": 9.467592592592593e-07,
      "loss": 0.0071,
      "step": 8231
    },
    {
      "epoch": 0.9527777777777777,
      "grad_norm": 0.11986110359430313,
      "learning_rate": 9.444444444444445e-07,
      "loss": 0.0104,
      "step": 8232
    },
    {
      "epoch": 0.9528935185185186,
      "grad_norm": 0.0394819937646389,
      "learning_rate": 9.421296296296296e-07,
      "loss": 0.0069,
      "step": 8233
    },
    {
      "epoch": 0.9530092592592593,
      "grad_norm": 0.07559458166360855,
      "learning_rate": 9.398148148148149e-07,
      "loss": 0.0073,
      "step": 8234
    },
    {
      "epoch": 0.953125,
      "grad_norm": 0.05551354959607124,
      "learning_rate": 9.375000000000001e-07,
      "loss": 0.0063,
      "step": 8235
    },
    {
      "epoch": 0.9532407407407407,
      "grad_norm": 0.04891026020050049,
      "learning_rate": 9.351851851851852e-07,
      "loss": 0.0089,
      "step": 8236
    },
    {
      "epoch": 0.9533564814814814,
      "grad_norm": 0.03285185247659683,
      "learning_rate": 9.328703703703705e-07,
      "loss": 0.0059,
      "step": 8237
    },
    {
      "epoch": 0.9534722222222223,
      "grad_norm": 0.08162957429885864,
      "learning_rate": 9.305555555555556e-07,
      "loss": 0.0099,
      "step": 8238
    },
    {
      "epoch": 0.953587962962963,
      "grad_norm": 0.04436888173222542,
      "learning_rate": 9.282407407407408e-07,
      "loss": 0.0058,
      "step": 8239
    },
    {
      "epoch": 0.9537037037037037,
      "grad_norm": 0.05766421556472778,
      "learning_rate": 9.259259259259259e-07,
      "loss": 0.0104,
      "step": 8240
    },
    {
      "epoch": 0.9538194444444444,
      "grad_norm": 0.046913452446460724,
      "learning_rate": 9.236111111111112e-07,
      "loss": 0.0084,
      "step": 8241
    },
    {
      "epoch": 0.9539351851851852,
      "grad_norm": 0.0338384248316288,
      "learning_rate": 9.212962962962963e-07,
      "loss": 0.0061,
      "step": 8242
    },
    {
      "epoch": 0.9540509259259259,
      "grad_norm": 0.07393302023410797,
      "learning_rate": 9.189814814814815e-07,
      "loss": 0.0065,
      "step": 8243
    },
    {
      "epoch": 0.9541666666666667,
      "grad_norm": 0.071597158908844,
      "learning_rate": 9.166666666666666e-07,
      "loss": 0.0068,
      "step": 8244
    },
    {
      "epoch": 0.9542824074074074,
      "grad_norm": 0.041415806859731674,
      "learning_rate": 9.14351851851852e-07,
      "loss": 0.0071,
      "step": 8245
    },
    {
      "epoch": 0.9543981481481482,
      "grad_norm": 0.045910682529211044,
      "learning_rate": 9.120370370370372e-07,
      "loss": 0.0083,
      "step": 8246
    },
    {
      "epoch": 0.9545138888888889,
      "grad_norm": 0.033381570130586624,
      "learning_rate": 9.097222222222223e-07,
      "loss": 0.006,
      "step": 8247
    },
    {
      "epoch": 0.9546296296296296,
      "grad_norm": 0.051960594952106476,
      "learning_rate": 9.074074074074076e-07,
      "loss": 0.0067,
      "step": 8248
    },
    {
      "epoch": 0.9547453703703703,
      "grad_norm": 0.039294756948947906,
      "learning_rate": 9.050925925925927e-07,
      "loss": 0.0069,
      "step": 8249
    },
    {
      "epoch": 0.9548611111111112,
      "grad_norm": 0.05305085331201553,
      "learning_rate": 9.027777777777779e-07,
      "loss": 0.0095,
      "step": 8250
    },
    {
      "epoch": 0.9549768518518519,
      "grad_norm": 0.07383108884096146,
      "learning_rate": 9.00462962962963e-07,
      "loss": 0.0094,
      "step": 8251
    },
    {
      "epoch": 0.9550925925925926,
      "grad_norm": 0.04747297242283821,
      "learning_rate": 8.981481481481483e-07,
      "loss": 0.0087,
      "step": 8252
    },
    {
      "epoch": 0.9552083333333333,
      "grad_norm": 0.040001511573791504,
      "learning_rate": 8.958333333333334e-07,
      "loss": 0.0071,
      "step": 8253
    },
    {
      "epoch": 0.955324074074074,
      "grad_norm": 0.049036238342523575,
      "learning_rate": 8.935185185185186e-07,
      "loss": 0.0089,
      "step": 8254
    },
    {
      "epoch": 0.9554398148148148,
      "grad_norm": 0.13067275285720825,
      "learning_rate": 8.912037037037037e-07,
      "loss": 0.0081,
      "step": 8255
    },
    {
      "epoch": 0.9555555555555556,
      "grad_norm": 0.045385003089904785,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0082,
      "step": 8256
    },
    {
      "epoch": 0.9556712962962963,
      "grad_norm": 0.03233950212597847,
      "learning_rate": 8.865740740740742e-07,
      "loss": 0.0058,
      "step": 8257
    },
    {
      "epoch": 0.955787037037037,
      "grad_norm": 0.057259928435087204,
      "learning_rate": 8.842592592592593e-07,
      "loss": 0.0097,
      "step": 8258
    },
    {
      "epoch": 0.9559027777777778,
      "grad_norm": 93.2154312133789,
      "learning_rate": 8.819444444444446e-07,
      "loss": 2.4836,
      "step": 8259
    },
    {
      "epoch": 0.9560185185185185,
      "grad_norm": 0.0394473597407341,
      "learning_rate": 8.796296296296297e-07,
      "loss": 0.0051,
      "step": 8260
    },
    {
      "epoch": 0.9561342592592592,
      "grad_norm": 0.04338586702942848,
      "learning_rate": 8.773148148148149e-07,
      "loss": 0.0077,
      "step": 8261
    },
    {
      "epoch": 0.95625,
      "grad_norm": 0.03484727814793587,
      "learning_rate": 8.75e-07,
      "loss": 0.006,
      "step": 8262
    },
    {
      "epoch": 0.9563657407407408,
      "grad_norm": 0.03769050911068916,
      "learning_rate": 8.726851851851853e-07,
      "loss": 0.0067,
      "step": 8263
    },
    {
      "epoch": 0.9564814814814815,
      "grad_norm": 0.034426674246788025,
      "learning_rate": 8.703703703703705e-07,
      "loss": 0.0062,
      "step": 8264
    },
    {
      "epoch": 0.9565972222222222,
      "grad_norm": 0.048274844884872437,
      "learning_rate": 8.680555555555556e-07,
      "loss": 0.0088,
      "step": 8265
    },
    {
      "epoch": 0.9567129629629629,
      "grad_norm": 0.06266511976718903,
      "learning_rate": 8.657407407407407e-07,
      "loss": 0.0106,
      "step": 8266
    },
    {
      "epoch": 0.9568287037037037,
      "grad_norm": 0.031321294605731964,
      "learning_rate": 8.63425925925926e-07,
      "loss": 0.0056,
      "step": 8267
    },
    {
      "epoch": 0.9569444444444445,
      "grad_norm": 0.042840030044317245,
      "learning_rate": 8.611111111111112e-07,
      "loss": 0.0074,
      "step": 8268
    },
    {
      "epoch": 0.9570601851851852,
      "grad_norm": 0.04523220658302307,
      "learning_rate": 8.587962962962963e-07,
      "loss": 0.0062,
      "step": 8269
    },
    {
      "epoch": 0.9571759259259259,
      "grad_norm": 0.04111603647470474,
      "learning_rate": 8.564814814814816e-07,
      "loss": 0.0072,
      "step": 8270
    },
    {
      "epoch": 0.9572916666666667,
      "grad_norm": 0.0419297032058239,
      "learning_rate": 8.541666666666667e-07,
      "loss": 0.0075,
      "step": 8271
    },
    {
      "epoch": 0.9574074074074074,
      "grad_norm": 0.03148185461759567,
      "learning_rate": 8.518518518518519e-07,
      "loss": 0.0057,
      "step": 8272
    },
    {
      "epoch": 0.9575231481481481,
      "grad_norm": 0.04667475074529648,
      "learning_rate": 8.49537037037037e-07,
      "loss": 0.0079,
      "step": 8273
    },
    {
      "epoch": 0.9576388888888889,
      "grad_norm": 0.04010686278343201,
      "learning_rate": 8.472222222222223e-07,
      "loss": 0.0071,
      "step": 8274
    },
    {
      "epoch": 0.9577546296296297,
      "grad_norm": 0.04830631986260414,
      "learning_rate": 8.449074074074075e-07,
      "loss": 0.0084,
      "step": 8275
    },
    {
      "epoch": 0.9578703703703704,
      "grad_norm": 0.03317475691437721,
      "learning_rate": 8.425925925925926e-07,
      "loss": 0.006,
      "step": 8276
    },
    {
      "epoch": 0.9579861111111111,
      "grad_norm": 0.04874303191900253,
      "learning_rate": 8.402777777777779e-07,
      "loss": 0.0088,
      "step": 8277
    },
    {
      "epoch": 0.9581018518518518,
      "grad_norm": 0.03137604147195816,
      "learning_rate": 8.37962962962963e-07,
      "loss": 0.0056,
      "step": 8278
    },
    {
      "epoch": 0.9582175925925925,
      "grad_norm": 0.029753396287560463,
      "learning_rate": 8.356481481481482e-07,
      "loss": 0.0054,
      "step": 8279
    },
    {
      "epoch": 0.9583333333333334,
      "grad_norm": 0.047044966369867325,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0077,
      "step": 8280
    },
    {
      "epoch": 0.9584490740740741,
      "grad_norm": 0.05717247724533081,
      "learning_rate": 8.310185185185186e-07,
      "loss": 0.0103,
      "step": 8281
    },
    {
      "epoch": 0.9585648148148148,
      "grad_norm": 0.0396527461707592,
      "learning_rate": 8.287037037037037e-07,
      "loss": 0.007,
      "step": 8282
    },
    {
      "epoch": 0.9586805555555555,
      "grad_norm": 0.09015388786792755,
      "learning_rate": 8.263888888888889e-07,
      "loss": 0.0101,
      "step": 8283
    },
    {
      "epoch": 0.9587962962962963,
      "grad_norm": 0.038274671882390976,
      "learning_rate": 8.24074074074074e-07,
      "loss": 0.0069,
      "step": 8284
    },
    {
      "epoch": 0.9589120370370371,
      "grad_norm": 0.05374132841825485,
      "learning_rate": 8.217592592592593e-07,
      "loss": 0.0098,
      "step": 8285
    },
    {
      "epoch": 0.9590277777777778,
      "grad_norm": 0.042005110532045364,
      "learning_rate": 8.194444444444446e-07,
      "loss": 0.0073,
      "step": 8286
    },
    {
      "epoch": 0.9591435185185185,
      "grad_norm": 0.027963610365986824,
      "learning_rate": 8.171296296296296e-07,
      "loss": 0.0051,
      "step": 8287
    },
    {
      "epoch": 0.9592592592592593,
      "grad_norm": 0.042162369936704636,
      "learning_rate": 8.14814814814815e-07,
      "loss": 0.0075,
      "step": 8288
    },
    {
      "epoch": 0.959375,
      "grad_norm": 0.04360808804631233,
      "learning_rate": 8.125000000000001e-07,
      "loss": 0.0079,
      "step": 8289
    },
    {
      "epoch": 0.9594907407407407,
      "grad_norm": 0.029505044221878052,
      "learning_rate": 8.101851851851853e-07,
      "loss": 0.0053,
      "step": 8290
    },
    {
      "epoch": 0.9596064814814815,
      "grad_norm": 0.04372907429933548,
      "learning_rate": 8.078703703703704e-07,
      "loss": 0.0079,
      "step": 8291
    },
    {
      "epoch": 0.9597222222222223,
      "grad_norm": 0.49128904938697815,
      "learning_rate": 8.055555555555557e-07,
      "loss": 0.0132,
      "step": 8292
    },
    {
      "epoch": 0.959837962962963,
      "grad_norm": 0.0422176867723465,
      "learning_rate": 8.032407407407409e-07,
      "loss": 0.0066,
      "step": 8293
    },
    {
      "epoch": 0.9599537037037037,
      "grad_norm": 0.09104559570550919,
      "learning_rate": 8.00925925925926e-07,
      "loss": 0.0082,
      "step": 8294
    },
    {
      "epoch": 0.9600694444444444,
      "grad_norm": 0.03505704179406166,
      "learning_rate": 7.986111111111111e-07,
      "loss": 0.0063,
      "step": 8295
    },
    {
      "epoch": 0.9601851851851851,
      "grad_norm": 0.030465763062238693,
      "learning_rate": 7.962962962962964e-07,
      "loss": 0.0054,
      "step": 8296
    },
    {
      "epoch": 0.960300925925926,
      "grad_norm": 0.03811672329902649,
      "learning_rate": 7.939814814814816e-07,
      "loss": 0.0066,
      "step": 8297
    },
    {
      "epoch": 0.9604166666666667,
      "grad_norm": 0.0582103431224823,
      "learning_rate": 7.916666666666667e-07,
      "loss": 0.0078,
      "step": 8298
    },
    {
      "epoch": 0.9605324074074074,
      "grad_norm": 0.05931193009018898,
      "learning_rate": 7.89351851851852e-07,
      "loss": 0.0066,
      "step": 8299
    },
    {
      "epoch": 0.9606481481481481,
      "grad_norm": 69.53038024902344,
      "learning_rate": 7.870370370370371e-07,
      "loss": 0.2034,
      "step": 8300
    },
    {
      "epoch": 0.9607638888888889,
      "grad_norm": 0.04291914030909538,
      "learning_rate": 7.847222222222223e-07,
      "loss": 0.0078,
      "step": 8301
    },
    {
      "epoch": 0.9608796296296296,
      "grad_norm": 0.047849275171756744,
      "learning_rate": 7.824074074074074e-07,
      "loss": 0.0082,
      "step": 8302
    },
    {
      "epoch": 0.9609953703703704,
      "grad_norm": 28.42102813720703,
      "learning_rate": 7.800925925925927e-07,
      "loss": 2.535,
      "step": 8303
    },
    {
      "epoch": 0.9611111111111111,
      "grad_norm": 0.05024884268641472,
      "learning_rate": 7.777777777777779e-07,
      "loss": 0.0086,
      "step": 8304
    },
    {
      "epoch": 0.9612268518518519,
      "grad_norm": 0.038180701434612274,
      "learning_rate": 7.75462962962963e-07,
      "loss": 0.0068,
      "step": 8305
    },
    {
      "epoch": 0.9613425925925926,
      "grad_norm": 0.17020659148693085,
      "learning_rate": 7.731481481481482e-07,
      "loss": 0.0101,
      "step": 8306
    },
    {
      "epoch": 0.9614583333333333,
      "grad_norm": 0.03331896662712097,
      "learning_rate": 7.708333333333334e-07,
      "loss": 0.006,
      "step": 8307
    },
    {
      "epoch": 0.961574074074074,
      "grad_norm": 0.04652216285467148,
      "learning_rate": 7.685185185185186e-07,
      "loss": 0.0081,
      "step": 8308
    },
    {
      "epoch": 0.9616898148148149,
      "grad_norm": 0.04213671386241913,
      "learning_rate": 7.662037037037037e-07,
      "loss": 0.0072,
      "step": 8309
    },
    {
      "epoch": 0.9618055555555556,
      "grad_norm": 0.031239161267876625,
      "learning_rate": 7.63888888888889e-07,
      "loss": 0.0056,
      "step": 8310
    },
    {
      "epoch": 0.9619212962962963,
      "grad_norm": 0.04821588471531868,
      "learning_rate": 7.615740740740741e-07,
      "loss": 0.0088,
      "step": 8311
    },
    {
      "epoch": 0.962037037037037,
      "grad_norm": 0.07115010172128677,
      "learning_rate": 7.592592592592593e-07,
      "loss": 0.0093,
      "step": 8312
    },
    {
      "epoch": 0.9621527777777777,
      "grad_norm": 0.03881041333079338,
      "learning_rate": 7.569444444444444e-07,
      "loss": 0.0062,
      "step": 8313
    },
    {
      "epoch": 0.9622685185185185,
      "grad_norm": 0.042933352291584015,
      "learning_rate": 7.546296296296297e-07,
      "loss": 0.0076,
      "step": 8314
    },
    {
      "epoch": 0.9623842592592593,
      "grad_norm": 0.09343316406011581,
      "learning_rate": 7.523148148148149e-07,
      "loss": 0.0115,
      "step": 8315
    },
    {
      "epoch": 0.9625,
      "grad_norm": 0.04806487634778023,
      "learning_rate": 7.5e-07,
      "loss": 0.0085,
      "step": 8316
    },
    {
      "epoch": 0.9626157407407407,
      "grad_norm": 0.0680573433637619,
      "learning_rate": 7.476851851851852e-07,
      "loss": 0.0079,
      "step": 8317
    },
    {
      "epoch": 0.9627314814814815,
      "grad_norm": 0.04731408506631851,
      "learning_rate": 7.453703703703704e-07,
      "loss": 0.0081,
      "step": 8318
    },
    {
      "epoch": 0.9628472222222222,
      "grad_norm": 0.04459869861602783,
      "learning_rate": 7.430555555555556e-07,
      "loss": 0.0078,
      "step": 8319
    },
    {
      "epoch": 0.9629629629629629,
      "grad_norm": 0.03480131924152374,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.0062,
      "step": 8320
    },
    {
      "epoch": 0.9630787037037037,
      "grad_norm": 100.510986328125,
      "learning_rate": 7.38425925925926e-07,
      "loss": 2.3059,
      "step": 8321
    },
    {
      "epoch": 0.9631944444444445,
      "grad_norm": 0.04704653471708298,
      "learning_rate": 7.361111111111112e-07,
      "loss": 0.0084,
      "step": 8322
    },
    {
      "epoch": 0.9633101851851852,
      "grad_norm": 0.040563639253377914,
      "learning_rate": 7.337962962962963e-07,
      "loss": 0.0071,
      "step": 8323
    },
    {
      "epoch": 0.9634259259259259,
      "grad_norm": 0.03260820358991623,
      "learning_rate": 7.314814814814814e-07,
      "loss": 0.0058,
      "step": 8324
    },
    {
      "epoch": 0.9635416666666666,
      "grad_norm": 0.03522292524576187,
      "learning_rate": 7.291666666666667e-07,
      "loss": 0.0064,
      "step": 8325
    },
    {
      "epoch": 0.9636574074074075,
      "grad_norm": 0.0614810548722744,
      "learning_rate": 7.26851851851852e-07,
      "loss": 0.008,
      "step": 8326
    },
    {
      "epoch": 0.9637731481481482,
      "grad_norm": 0.05238870158791542,
      "learning_rate": 7.24537037037037e-07,
      "loss": 0.0092,
      "step": 8327
    },
    {
      "epoch": 0.9638888888888889,
      "grad_norm": 0.03791162371635437,
      "learning_rate": 7.222222222222222e-07,
      "loss": 0.0068,
      "step": 8328
    },
    {
      "epoch": 0.9640046296296296,
      "grad_norm": 0.08195522427558899,
      "learning_rate": 7.199074074074075e-07,
      "loss": 0.0083,
      "step": 8329
    },
    {
      "epoch": 0.9641203703703703,
      "grad_norm": 0.04482664167881012,
      "learning_rate": 7.175925925925927e-07,
      "loss": 0.0069,
      "step": 8330
    },
    {
      "epoch": 0.9642361111111111,
      "grad_norm": 0.05040524899959564,
      "learning_rate": 7.152777777777778e-07,
      "loss": 0.0084,
      "step": 8331
    },
    {
      "epoch": 0.9643518518518519,
      "grad_norm": 0.041463226079940796,
      "learning_rate": 7.129629629629631e-07,
      "loss": 0.0074,
      "step": 8332
    },
    {
      "epoch": 0.9644675925925926,
      "grad_norm": 0.04382581636309624,
      "learning_rate": 7.106481481481483e-07,
      "loss": 0.0077,
      "step": 8333
    },
    {
      "epoch": 0.9645833333333333,
      "grad_norm": 0.035030800849199295,
      "learning_rate": 7.083333333333334e-07,
      "loss": 0.0062,
      "step": 8334
    },
    {
      "epoch": 0.9646990740740741,
      "grad_norm": 0.052139464765787125,
      "learning_rate": 7.060185185185186e-07,
      "loss": 0.0086,
      "step": 8335
    },
    {
      "epoch": 0.9648148148148148,
      "grad_norm": 17.993757247924805,
      "learning_rate": 7.037037037037038e-07,
      "loss": 0.0722,
      "step": 8336
    },
    {
      "epoch": 0.9649305555555555,
      "grad_norm": 0.3819003105163574,
      "learning_rate": 7.01388888888889e-07,
      "loss": 0.0131,
      "step": 8337
    },
    {
      "epoch": 0.9650462962962963,
      "grad_norm": 0.06573113054037094,
      "learning_rate": 6.990740740740741e-07,
      "loss": 0.0099,
      "step": 8338
    },
    {
      "epoch": 0.9651620370370371,
      "grad_norm": 0.057118918746709824,
      "learning_rate": 6.967592592592594e-07,
      "loss": 0.0102,
      "step": 8339
    },
    {
      "epoch": 0.9652777777777778,
      "grad_norm": 0.05194597318768501,
      "learning_rate": 6.944444444444446e-07,
      "loss": 0.0067,
      "step": 8340
    },
    {
      "epoch": 0.9653935185185185,
      "grad_norm": 0.03238454461097717,
      "learning_rate": 6.921296296296297e-07,
      "loss": 0.0058,
      "step": 8341
    },
    {
      "epoch": 0.9655092592592592,
      "grad_norm": 0.06340181827545166,
      "learning_rate": 6.898148148148148e-07,
      "loss": 0.0072,
      "step": 8342
    },
    {
      "epoch": 0.965625,
      "grad_norm": 0.045844584703445435,
      "learning_rate": 6.875000000000001e-07,
      "loss": 0.0083,
      "step": 8343
    },
    {
      "epoch": 0.9657407407407408,
      "grad_norm": 0.04133826494216919,
      "learning_rate": 6.851851851851853e-07,
      "loss": 0.0074,
      "step": 8344
    },
    {
      "epoch": 0.9658564814814815,
      "grad_norm": 0.058094557374715805,
      "learning_rate": 6.828703703703704e-07,
      "loss": 0.0104,
      "step": 8345
    },
    {
      "epoch": 0.9659722222222222,
      "grad_norm": 1.0135647058486938,
      "learning_rate": 6.805555555555556e-07,
      "loss": 0.0166,
      "step": 8346
    },
    {
      "epoch": 0.966087962962963,
      "grad_norm": 0.05035032704472542,
      "learning_rate": 6.782407407407408e-07,
      "loss": 0.0086,
      "step": 8347
    },
    {
      "epoch": 0.9662037037037037,
      "grad_norm": 0.041105665266513824,
      "learning_rate": 6.75925925925926e-07,
      "loss": 0.0061,
      "step": 8348
    },
    {
      "epoch": 0.9663194444444444,
      "grad_norm": 0.043008316308259964,
      "learning_rate": 6.736111111111111e-07,
      "loss": 0.0078,
      "step": 8349
    },
    {
      "epoch": 0.9664351851851852,
      "grad_norm": 0.04320894554257393,
      "learning_rate": 6.712962962962964e-07,
      "loss": 0.0078,
      "step": 8350
    },
    {
      "epoch": 0.966550925925926,
      "grad_norm": 0.0781964659690857,
      "learning_rate": 6.689814814814816e-07,
      "loss": 0.0082,
      "step": 8351
    },
    {
      "epoch": 0.9666666666666667,
      "grad_norm": 0.05448819324374199,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0073,
      "step": 8352
    },
    {
      "epoch": 0.9667824074074074,
      "grad_norm": 0.046172428876161575,
      "learning_rate": 6.643518518518519e-07,
      "loss": 0.0075,
      "step": 8353
    },
    {
      "epoch": 0.9668981481481481,
      "grad_norm": 0.0396362766623497,
      "learning_rate": 6.620370370370371e-07,
      "loss": 0.0071,
      "step": 8354
    },
    {
      "epoch": 0.9670138888888888,
      "grad_norm": 0.05131206288933754,
      "learning_rate": 6.597222222222223e-07,
      "loss": 0.0093,
      "step": 8355
    },
    {
      "epoch": 0.9671296296296297,
      "grad_norm": 0.10038334876298904,
      "learning_rate": 6.574074074074074e-07,
      "loss": 0.0101,
      "step": 8356
    },
    {
      "epoch": 0.9672453703703704,
      "grad_norm": 0.03968598321080208,
      "learning_rate": 6.550925925925926e-07,
      "loss": 0.0056,
      "step": 8357
    },
    {
      "epoch": 0.9673611111111111,
      "grad_norm": 0.0711396336555481,
      "learning_rate": 6.527777777777778e-07,
      "loss": 0.0092,
      "step": 8358
    },
    {
      "epoch": 0.9674768518518518,
      "grad_norm": 0.036415036767721176,
      "learning_rate": 6.50462962962963e-07,
      "loss": 0.0065,
      "step": 8359
    },
    {
      "epoch": 0.9675925925925926,
      "grad_norm": 0.03602802753448486,
      "learning_rate": 6.481481481481481e-07,
      "loss": 0.0065,
      "step": 8360
    },
    {
      "epoch": 0.9677083333333333,
      "grad_norm": 0.03898433968424797,
      "learning_rate": 6.458333333333334e-07,
      "loss": 0.0059,
      "step": 8361
    },
    {
      "epoch": 0.9678240740740741,
      "grad_norm": 0.05529742315411568,
      "learning_rate": 6.435185185185186e-07,
      "loss": 0.009,
      "step": 8362
    },
    {
      "epoch": 0.9679398148148148,
      "grad_norm": 0.0563715435564518,
      "learning_rate": 6.412037037037037e-07,
      "loss": 0.0102,
      "step": 8363
    },
    {
      "epoch": 0.9680555555555556,
      "grad_norm": 0.059782449156045914,
      "learning_rate": 6.388888888888889e-07,
      "loss": 0.0106,
      "step": 8364
    },
    {
      "epoch": 0.9681712962962963,
      "grad_norm": 0.04251319169998169,
      "learning_rate": 6.365740740740741e-07,
      "loss": 0.0075,
      "step": 8365
    },
    {
      "epoch": 0.968287037037037,
      "grad_norm": 0.050788864493370056,
      "learning_rate": 6.342592592592593e-07,
      "loss": 0.0091,
      "step": 8366
    },
    {
      "epoch": 0.9684027777777777,
      "grad_norm": 0.05127236992120743,
      "learning_rate": 6.319444444444444e-07,
      "loss": 0.0092,
      "step": 8367
    },
    {
      "epoch": 0.9685185185185186,
      "grad_norm": 0.05522892251610756,
      "learning_rate": 6.296296296296296e-07,
      "loss": 0.0077,
      "step": 8368
    },
    {
      "epoch": 0.9686342592592593,
      "grad_norm": 0.035915665328502655,
      "learning_rate": 6.27314814814815e-07,
      "loss": 0.0065,
      "step": 8369
    },
    {
      "epoch": 0.96875,
      "grad_norm": 0.8105081915855408,
      "learning_rate": 6.25e-07,
      "loss": 0.0127,
      "step": 8370
    },
    {
      "epoch": 0.9688657407407407,
      "grad_norm": 0.031167425215244293,
      "learning_rate": 6.226851851851853e-07,
      "loss": 0.0055,
      "step": 8371
    },
    {
      "epoch": 0.9689814814814814,
      "grad_norm": 0.04978622496128082,
      "learning_rate": 6.203703703703705e-07,
      "loss": 0.0088,
      "step": 8372
    },
    {
      "epoch": 0.9690972222222223,
      "grad_norm": 0.03402505815029144,
      "learning_rate": 6.180555555555557e-07,
      "loss": 0.0061,
      "step": 8373
    },
    {
      "epoch": 0.969212962962963,
      "grad_norm": 0.06281981617212296,
      "learning_rate": 6.157407407407408e-07,
      "loss": 0.0078,
      "step": 8374
    },
    {
      "epoch": 0.9693287037037037,
      "grad_norm": 97.3980712890625,
      "learning_rate": 6.13425925925926e-07,
      "loss": 1.4855,
      "step": 8375
    },
    {
      "epoch": 0.9694444444444444,
      "grad_norm": 0.19317209720611572,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.0101,
      "step": 8376
    },
    {
      "epoch": 0.9695601851851852,
      "grad_norm": 0.043321382254362106,
      "learning_rate": 6.087962962962964e-07,
      "loss": 0.0076,
      "step": 8377
    },
    {
      "epoch": 0.9696759259259259,
      "grad_norm": 0.03304092213511467,
      "learning_rate": 6.064814814814815e-07,
      "loss": 0.0059,
      "step": 8378
    },
    {
      "epoch": 0.9697916666666667,
      "grad_norm": 0.051890525966882706,
      "learning_rate": 6.041666666666667e-07,
      "loss": 0.0063,
      "step": 8379
    },
    {
      "epoch": 0.9699074074074074,
      "grad_norm": 0.031957756727933884,
      "learning_rate": 6.018518518518519e-07,
      "loss": 0.0057,
      "step": 8380
    },
    {
      "epoch": 0.9700231481481482,
      "grad_norm": 0.03354531526565552,
      "learning_rate": 5.995370370370371e-07,
      "loss": 0.006,
      "step": 8381
    },
    {
      "epoch": 0.9701388888888889,
      "grad_norm": 0.06595509499311447,
      "learning_rate": 5.972222222222223e-07,
      "loss": 0.0096,
      "step": 8382
    },
    {
      "epoch": 0.9702546296296296,
      "grad_norm": 0.03711314871907234,
      "learning_rate": 5.949074074074075e-07,
      "loss": 0.0067,
      "step": 8383
    },
    {
      "epoch": 0.9703703703703703,
      "grad_norm": 0.02979155257344246,
      "learning_rate": 5.925925925925927e-07,
      "loss": 0.0054,
      "step": 8384
    },
    {
      "epoch": 0.9704861111111112,
      "grad_norm": 0.04126082360744476,
      "learning_rate": 5.902777777777778e-07,
      "loss": 0.0071,
      "step": 8385
    },
    {
      "epoch": 0.9706018518518519,
      "grad_norm": 0.02968195080757141,
      "learning_rate": 5.87962962962963e-07,
      "loss": 0.0054,
      "step": 8386
    },
    {
      "epoch": 0.9707175925925926,
      "grad_norm": 0.04188719764351845,
      "learning_rate": 5.856481481481482e-07,
      "loss": 0.0054,
      "step": 8387
    },
    {
      "epoch": 0.9708333333333333,
      "grad_norm": 1.2190935611724854,
      "learning_rate": 5.833333333333334e-07,
      "loss": 0.0136,
      "step": 8388
    },
    {
      "epoch": 0.970949074074074,
      "grad_norm": 0.04550323635339737,
      "learning_rate": 5.810185185185185e-07,
      "loss": 0.0059,
      "step": 8389
    },
    {
      "epoch": 0.9710648148148148,
      "grad_norm": 0.03168429434299469,
      "learning_rate": 5.787037037037038e-07,
      "loss": 0.0057,
      "step": 8390
    },
    {
      "epoch": 0.9711805555555556,
      "grad_norm": 0.04030515253543854,
      "learning_rate": 5.76388888888889e-07,
      "loss": 0.0072,
      "step": 8391
    },
    {
      "epoch": 0.9712962962962963,
      "grad_norm": 0.032106392085552216,
      "learning_rate": 5.740740740740741e-07,
      "loss": 0.0058,
      "step": 8392
    },
    {
      "epoch": 0.971412037037037,
      "grad_norm": 0.046175289899110794,
      "learning_rate": 5.717592592592593e-07,
      "loss": 0.0062,
      "step": 8393
    },
    {
      "epoch": 0.9715277777777778,
      "grad_norm": 0.06394441425800323,
      "learning_rate": 5.694444444444445e-07,
      "loss": 0.0084,
      "step": 8394
    },
    {
      "epoch": 0.9716435185185185,
      "grad_norm": 0.060238681733608246,
      "learning_rate": 5.671296296296297e-07,
      "loss": 0.0088,
      "step": 8395
    },
    {
      "epoch": 0.9717592592592592,
      "grad_norm": 0.03497278317809105,
      "learning_rate": 5.648148148148148e-07,
      "loss": 0.0062,
      "step": 8396
    },
    {
      "epoch": 0.971875,
      "grad_norm": 0.05149718374013901,
      "learning_rate": 5.625e-07,
      "loss": 0.009,
      "step": 8397
    },
    {
      "epoch": 0.9719907407407408,
      "grad_norm": 125.48455047607422,
      "learning_rate": 5.601851851851852e-07,
      "loss": 1.062,
      "step": 8398
    },
    {
      "epoch": 0.9721064814814815,
      "grad_norm": 154.17677307128906,
      "learning_rate": 5.578703703703704e-07,
      "loss": 1.1168,
      "step": 8399
    },
    {
      "epoch": 0.9722222222222222,
      "grad_norm": 0.21462777256965637,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.01,
      "step": 8400
    },
    {
      "epoch": 0.9723379629629629,
      "grad_norm": 0.033520009368658066,
      "learning_rate": 5.532407407407408e-07,
      "loss": 0.0061,
      "step": 8401
    },
    {
      "epoch": 0.9724537037037037,
      "grad_norm": 0.028689809143543243,
      "learning_rate": 5.50925925925926e-07,
      "loss": 0.0052,
      "step": 8402
    },
    {
      "epoch": 0.9725694444444445,
      "grad_norm": 0.045614343136548996,
      "learning_rate": 5.486111111111111e-07,
      "loss": 0.0072,
      "step": 8403
    },
    {
      "epoch": 0.9726851851851852,
      "grad_norm": 0.03264656662940979,
      "learning_rate": 5.462962962962963e-07,
      "loss": 0.0058,
      "step": 8404
    },
    {
      "epoch": 0.9728009259259259,
      "grad_norm": 0.12089547514915466,
      "learning_rate": 5.439814814814815e-07,
      "loss": 0.007,
      "step": 8405
    },
    {
      "epoch": 0.9729166666666667,
      "grad_norm": 0.0531386137008667,
      "learning_rate": 5.416666666666667e-07,
      "loss": 0.0094,
      "step": 8406
    },
    {
      "epoch": 0.9730324074074074,
      "grad_norm": 0.037110764533281326,
      "learning_rate": 5.393518518518518e-07,
      "loss": 0.0059,
      "step": 8407
    },
    {
      "epoch": 0.9731481481481481,
      "grad_norm": 0.042553722858428955,
      "learning_rate": 5.37037037037037e-07,
      "loss": 0.0072,
      "step": 8408
    },
    {
      "epoch": 0.9732638888888889,
      "grad_norm": 0.05500292778015137,
      "learning_rate": 5.347222222222222e-07,
      "loss": 0.01,
      "step": 8409
    },
    {
      "epoch": 0.9733796296296297,
      "grad_norm": 0.04242468997836113,
      "learning_rate": 5.324074074074074e-07,
      "loss": 0.0077,
      "step": 8410
    },
    {
      "epoch": 0.9734953703703704,
      "grad_norm": 0.13786299526691437,
      "learning_rate": 5.300925925925927e-07,
      "loss": 0.0114,
      "step": 8411
    },
    {
      "epoch": 0.9736111111111111,
      "grad_norm": 0.04574298858642578,
      "learning_rate": 5.277777777777779e-07,
      "loss": 0.0082,
      "step": 8412
    },
    {
      "epoch": 0.9737268518518518,
      "grad_norm": 0.1362677365541458,
      "learning_rate": 5.254629629629631e-07,
      "loss": 0.0102,
      "step": 8413
    },
    {
      "epoch": 0.9738425925925925,
      "grad_norm": 0.04158995673060417,
      "learning_rate": 5.231481481481482e-07,
      "loss": 0.0074,
      "step": 8414
    },
    {
      "epoch": 0.9739583333333334,
      "grad_norm": 0.040046483278274536,
      "learning_rate": 5.208333333333334e-07,
      "loss": 0.0071,
      "step": 8415
    },
    {
      "epoch": 0.9740740740740741,
      "grad_norm": 0.03899870067834854,
      "learning_rate": 5.185185185185186e-07,
      "loss": 0.0067,
      "step": 8416
    },
    {
      "epoch": 0.9741898148148148,
      "grad_norm": 0.05180148780345917,
      "learning_rate": 5.162037037037038e-07,
      "loss": 0.0091,
      "step": 8417
    },
    {
      "epoch": 0.9743055555555555,
      "grad_norm": 0.04532499611377716,
      "learning_rate": 5.138888888888889e-07,
      "loss": 0.0081,
      "step": 8418
    },
    {
      "epoch": 0.9744212962962963,
      "grad_norm": 0.04538749158382416,
      "learning_rate": 5.115740740740742e-07,
      "loss": 0.0074,
      "step": 8419
    },
    {
      "epoch": 0.9745370370370371,
      "grad_norm": 0.04156912490725517,
      "learning_rate": 5.092592592592593e-07,
      "loss": 0.0072,
      "step": 8420
    },
    {
      "epoch": 0.9746527777777778,
      "grad_norm": 0.07311306893825531,
      "learning_rate": 5.069444444444445e-07,
      "loss": 0.0083,
      "step": 8421
    },
    {
      "epoch": 0.9747685185185185,
      "grad_norm": 0.03429728373885155,
      "learning_rate": 5.046296296296297e-07,
      "loss": 0.0062,
      "step": 8422
    },
    {
      "epoch": 0.9748842592592593,
      "grad_norm": 0.032200343906879425,
      "learning_rate": 5.023148148148149e-07,
      "loss": 0.0057,
      "step": 8423
    },
    {
      "epoch": 0.975,
      "grad_norm": 0.04575158655643463,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0073,
      "step": 8424
    },
    {
      "epoch": 0.9751157407407407,
      "grad_norm": 0.050519708544015884,
      "learning_rate": 4.976851851851852e-07,
      "loss": 0.0091,
      "step": 8425
    },
    {
      "epoch": 0.9752314814814815,
      "grad_norm": 0.0519326888024807,
      "learning_rate": 4.953703703703704e-07,
      "loss": 0.0092,
      "step": 8426
    },
    {
      "epoch": 0.9753472222222223,
      "grad_norm": 0.0582905150949955,
      "learning_rate": 4.930555555555556e-07,
      "loss": 0.0105,
      "step": 8427
    },
    {
      "epoch": 0.975462962962963,
      "grad_norm": 0.03954492509365082,
      "learning_rate": 4.907407407407408e-07,
      "loss": 0.0051,
      "step": 8428
    },
    {
      "epoch": 0.9755787037037037,
      "grad_norm": 0.048410650342702866,
      "learning_rate": 4.88425925925926e-07,
      "loss": 0.0084,
      "step": 8429
    },
    {
      "epoch": 0.9756944444444444,
      "grad_norm": 0.03229331597685814,
      "learning_rate": 4.861111111111112e-07,
      "loss": 0.0057,
      "step": 8430
    },
    {
      "epoch": 0.9758101851851851,
      "grad_norm": 0.050966549664735794,
      "learning_rate": 4.837962962962963e-07,
      "loss": 0.0091,
      "step": 8431
    },
    {
      "epoch": 0.975925925925926,
      "grad_norm": 0.03389359265565872,
      "learning_rate": 4.814814814814815e-07,
      "loss": 0.0058,
      "step": 8432
    },
    {
      "epoch": 0.9760416666666667,
      "grad_norm": 0.04779202863574028,
      "learning_rate": 4.791666666666667e-07,
      "loss": 0.0062,
      "step": 8433
    },
    {
      "epoch": 0.9761574074074074,
      "grad_norm": 0.040332261472940445,
      "learning_rate": 4.768518518518519e-07,
      "loss": 0.0073,
      "step": 8434
    },
    {
      "epoch": 0.9762731481481481,
      "grad_norm": 0.13293786346912384,
      "learning_rate": 4.745370370370371e-07,
      "loss": 0.0104,
      "step": 8435
    },
    {
      "epoch": 0.9763888888888889,
      "grad_norm": 0.039872702211141586,
      "learning_rate": 4.7222222222222226e-07,
      "loss": 0.0069,
      "step": 8436
    },
    {
      "epoch": 0.9765046296296296,
      "grad_norm": 0.04640136659145355,
      "learning_rate": 4.6990740740740746e-07,
      "loss": 0.008,
      "step": 8437
    },
    {
      "epoch": 0.9766203703703704,
      "grad_norm": 0.37188729643821716,
      "learning_rate": 4.675925925925926e-07,
      "loss": 0.0105,
      "step": 8438
    },
    {
      "epoch": 0.9767361111111111,
      "grad_norm": 0.04906943440437317,
      "learning_rate": 4.652777777777778e-07,
      "loss": 0.0086,
      "step": 8439
    },
    {
      "epoch": 0.9768518518518519,
      "grad_norm": 0.03828519955277443,
      "learning_rate": 4.6296296296296297e-07,
      "loss": 0.0068,
      "step": 8440
    },
    {
      "epoch": 0.9769675925925926,
      "grad_norm": 0.02866736426949501,
      "learning_rate": 4.6064814814814817e-07,
      "loss": 0.0051,
      "step": 8441
    },
    {
      "epoch": 0.9770833333333333,
      "grad_norm": 0.030619975179433823,
      "learning_rate": 4.583333333333333e-07,
      "loss": 0.0055,
      "step": 8442
    },
    {
      "epoch": 0.977199074074074,
      "grad_norm": 0.03825092315673828,
      "learning_rate": 4.560185185185186e-07,
      "loss": 0.0067,
      "step": 8443
    },
    {
      "epoch": 0.9773148148148149,
      "grad_norm": 139.33871459960938,
      "learning_rate": 4.537037037037038e-07,
      "loss": 0.7163,
      "step": 8444
    },
    {
      "epoch": 0.9774305555555556,
      "grad_norm": 0.03275391459465027,
      "learning_rate": 4.5138888888888893e-07,
      "loss": 0.0059,
      "step": 8445
    },
    {
      "epoch": 0.9775462962962963,
      "grad_norm": 0.07632274925708771,
      "learning_rate": 4.4907407407407414e-07,
      "loss": 0.0096,
      "step": 8446
    },
    {
      "epoch": 0.977662037037037,
      "grad_norm": 0.04009715095162392,
      "learning_rate": 4.467592592592593e-07,
      "loss": 0.007,
      "step": 8447
    },
    {
      "epoch": 0.9777777777777777,
      "grad_norm": 0.04563542827963829,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0076,
      "step": 8448
    },
    {
      "epoch": 0.9778935185185185,
      "grad_norm": 0.03393710032105446,
      "learning_rate": 4.4212962962962964e-07,
      "loss": 0.0061,
      "step": 8449
    },
    {
      "epoch": 0.9780092592592593,
      "grad_norm": 0.10012899339199066,
      "learning_rate": 4.3981481481481484e-07,
      "loss": 0.0109,
      "step": 8450
    },
    {
      "epoch": 0.978125,
      "grad_norm": 0.03289307281374931,
      "learning_rate": 4.375e-07,
      "loss": 0.0059,
      "step": 8451
    },
    {
      "epoch": 0.9782407407407407,
      "grad_norm": 0.1533019244670868,
      "learning_rate": 4.3518518518518525e-07,
      "loss": 0.0097,
      "step": 8452
    },
    {
      "epoch": 0.9783564814814815,
      "grad_norm": 0.4744795858860016,
      "learning_rate": 4.3287037037037035e-07,
      "loss": 0.0117,
      "step": 8453
    },
    {
      "epoch": 0.9784722222222222,
      "grad_norm": 0.04915035516023636,
      "learning_rate": 4.305555555555556e-07,
      "loss": 0.0087,
      "step": 8454
    },
    {
      "epoch": 0.9785879629629629,
      "grad_norm": 0.04839989170432091,
      "learning_rate": 4.282407407407408e-07,
      "loss": 0.0082,
      "step": 8455
    },
    {
      "epoch": 0.9787037037037037,
      "grad_norm": 0.04508787393569946,
      "learning_rate": 4.2592592592592596e-07,
      "loss": 0.0081,
      "step": 8456
    },
    {
      "epoch": 0.9788194444444445,
      "grad_norm": 0.03915814682841301,
      "learning_rate": 4.2361111111111116e-07,
      "loss": 0.0069,
      "step": 8457
    },
    {
      "epoch": 0.9789351851851852,
      "grad_norm": 0.038449354469776154,
      "learning_rate": 4.212962962962963e-07,
      "loss": 0.0068,
      "step": 8458
    },
    {
      "epoch": 0.9790509259259259,
      "grad_norm": 0.043923426419496536,
      "learning_rate": 4.189814814814815e-07,
      "loss": 0.0079,
      "step": 8459
    },
    {
      "epoch": 0.9791666666666666,
      "grad_norm": 0.032189007848501205,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.0058,
      "step": 8460
    },
    {
      "epoch": 0.9792824074074075,
      "grad_norm": 0.17534761130809784,
      "learning_rate": 4.1435185185185187e-07,
      "loss": 0.0096,
      "step": 8461
    },
    {
      "epoch": 0.9793981481481482,
      "grad_norm": 0.04198930040001869,
      "learning_rate": 4.12037037037037e-07,
      "loss": 0.0073,
      "step": 8462
    },
    {
      "epoch": 0.9795138888888889,
      "grad_norm": 0.07294014096260071,
      "learning_rate": 4.097222222222223e-07,
      "loss": 0.0069,
      "step": 8463
    },
    {
      "epoch": 0.9796296296296296,
      "grad_norm": 0.07476595044136047,
      "learning_rate": 4.074074074074075e-07,
      "loss": 0.0099,
      "step": 8464
    },
    {
      "epoch": 0.9797453703703703,
      "grad_norm": 0.04863090440630913,
      "learning_rate": 4.0509259259259263e-07,
      "loss": 0.0088,
      "step": 8465
    },
    {
      "epoch": 0.9798611111111111,
      "grad_norm": 0.05662897974252701,
      "learning_rate": 4.0277777777777783e-07,
      "loss": 0.0103,
      "step": 8466
    },
    {
      "epoch": 0.9799768518518519,
      "grad_norm": 0.04954441264271736,
      "learning_rate": 4.00462962962963e-07,
      "loss": 0.0076,
      "step": 8467
    },
    {
      "epoch": 0.9800925925925926,
      "grad_norm": 6.2601704597473145,
      "learning_rate": 3.981481481481482e-07,
      "loss": 0.0269,
      "step": 8468
    },
    {
      "epoch": 0.9802083333333333,
      "grad_norm": 0.08113276958465576,
      "learning_rate": 3.9583333333333334e-07,
      "loss": 0.0095,
      "step": 8469
    },
    {
      "epoch": 0.9803240740740741,
      "grad_norm": 0.03869038447737694,
      "learning_rate": 3.9351851851851854e-07,
      "loss": 0.0068,
      "step": 8470
    },
    {
      "epoch": 0.9804398148148148,
      "grad_norm": 0.03363322839140892,
      "learning_rate": 3.912037037037037e-07,
      "loss": 0.006,
      "step": 8471
    },
    {
      "epoch": 0.9805555555555555,
      "grad_norm": 0.03867146000266075,
      "learning_rate": 3.8888888888888895e-07,
      "loss": 0.0061,
      "step": 8472
    },
    {
      "epoch": 0.9806712962962963,
      "grad_norm": 0.0464123971760273,
      "learning_rate": 3.865740740740741e-07,
      "loss": 0.0083,
      "step": 8473
    },
    {
      "epoch": 0.9807870370370371,
      "grad_norm": 0.030652042478322983,
      "learning_rate": 3.842592592592593e-07,
      "loss": 0.0055,
      "step": 8474
    },
    {
      "epoch": 0.9809027777777778,
      "grad_norm": 0.040726903825998306,
      "learning_rate": 3.819444444444445e-07,
      "loss": 0.0072,
      "step": 8475
    },
    {
      "epoch": 0.9810185185185185,
      "grad_norm": 0.07744278013706207,
      "learning_rate": 3.7962962962962966e-07,
      "loss": 0.0097,
      "step": 8476
    },
    {
      "epoch": 0.9811342592592592,
      "grad_norm": 0.07138651609420776,
      "learning_rate": 3.7731481481481486e-07,
      "loss": 0.0092,
      "step": 8477
    },
    {
      "epoch": 0.98125,
      "grad_norm": 0.03293727710843086,
      "learning_rate": 3.75e-07,
      "loss": 0.0059,
      "step": 8478
    },
    {
      "epoch": 0.9813657407407408,
      "grad_norm": 62.618743896484375,
      "learning_rate": 3.726851851851852e-07,
      "loss": 0.1266,
      "step": 8479
    },
    {
      "epoch": 0.9814814814814815,
      "grad_norm": 0.04334770143032074,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.0078,
      "step": 8480
    },
    {
      "epoch": 0.9815972222222222,
      "grad_norm": 0.03260050341486931,
      "learning_rate": 3.680555555555556e-07,
      "loss": 0.0059,
      "step": 8481
    },
    {
      "epoch": 0.981712962962963,
      "grad_norm": 0.3647201657295227,
      "learning_rate": 3.657407407407407e-07,
      "loss": 0.0118,
      "step": 8482
    },
    {
      "epoch": 0.9818287037037037,
      "grad_norm": 0.050961777567863464,
      "learning_rate": 3.63425925925926e-07,
      "loss": 0.008,
      "step": 8483
    },
    {
      "epoch": 0.9819444444444444,
      "grad_norm": 0.049860239028930664,
      "learning_rate": 3.611111111111111e-07,
      "loss": 0.0091,
      "step": 8484
    },
    {
      "epoch": 0.9820601851851852,
      "grad_norm": 0.22088199853897095,
      "learning_rate": 3.5879629629629633e-07,
      "loss": 0.009,
      "step": 8485
    },
    {
      "epoch": 0.982175925925926,
      "grad_norm": 0.041272230446338654,
      "learning_rate": 3.5648148148148153e-07,
      "loss": 0.0069,
      "step": 8486
    },
    {
      "epoch": 0.9822916666666667,
      "grad_norm": 0.03553846478462219,
      "learning_rate": 3.541666666666667e-07,
      "loss": 0.0064,
      "step": 8487
    },
    {
      "epoch": 0.9824074074074074,
      "grad_norm": 0.027905358001589775,
      "learning_rate": 3.518518518518519e-07,
      "loss": 0.0051,
      "step": 8488
    },
    {
      "epoch": 0.9825231481481481,
      "grad_norm": 0.03572041541337967,
      "learning_rate": 3.4953703703703704e-07,
      "loss": 0.0064,
      "step": 8489
    },
    {
      "epoch": 0.9826388888888888,
      "grad_norm": 0.15610229969024658,
      "learning_rate": 3.472222222222223e-07,
      "loss": 0.0117,
      "step": 8490
    },
    {
      "epoch": 0.9827546296296297,
      "grad_norm": 0.04682266712188721,
      "learning_rate": 3.449074074074074e-07,
      "loss": 0.0084,
      "step": 8491
    },
    {
      "epoch": 0.9828703703703704,
      "grad_norm": 0.1965199112892151,
      "learning_rate": 3.4259259259259265e-07,
      "loss": 0.0085,
      "step": 8492
    },
    {
      "epoch": 0.9829861111111111,
      "grad_norm": 0.17759191989898682,
      "learning_rate": 3.402777777777778e-07,
      "loss": 0.0091,
      "step": 8493
    },
    {
      "epoch": 0.9831018518518518,
      "grad_norm": 0.037710901349782944,
      "learning_rate": 3.37962962962963e-07,
      "loss": 0.0067,
      "step": 8494
    },
    {
      "epoch": 0.9832175925925926,
      "grad_norm": 0.03572957590222359,
      "learning_rate": 3.356481481481482e-07,
      "loss": 0.0064,
      "step": 8495
    },
    {
      "epoch": 0.9833333333333333,
      "grad_norm": 0.04418567568063736,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.0076,
      "step": 8496
    },
    {
      "epoch": 0.9834490740740741,
      "grad_norm": 0.04911224916577339,
      "learning_rate": 3.3101851851851856e-07,
      "loss": 0.0064,
      "step": 8497
    },
    {
      "epoch": 0.9835648148148148,
      "grad_norm": 0.04051355645060539,
      "learning_rate": 3.287037037037037e-07,
      "loss": 0.0073,
      "step": 8498
    },
    {
      "epoch": 0.9836805555555556,
      "grad_norm": 0.046849820762872696,
      "learning_rate": 3.263888888888889e-07,
      "loss": 0.0081,
      "step": 8499
    },
    {
      "epoch": 0.9837962962962963,
      "grad_norm": 0.05106686055660248,
      "learning_rate": 3.2407407407407406e-07,
      "loss": 0.0079,
      "step": 8500
    },
    {
      "epoch": 0.983912037037037,
      "grad_norm": 0.07269396632909775,
      "learning_rate": 3.217592592592593e-07,
      "loss": 0.0095,
      "step": 8501
    },
    {
      "epoch": 0.9840277777777777,
      "grad_norm": 0.032581303268671036,
      "learning_rate": 3.1944444444444447e-07,
      "loss": 0.0056,
      "step": 8502
    },
    {
      "epoch": 0.9841435185185186,
      "grad_norm": 0.0697474405169487,
      "learning_rate": 3.1712962962962967e-07,
      "loss": 0.0091,
      "step": 8503
    },
    {
      "epoch": 0.9842592592592593,
      "grad_norm": 0.03271952643990517,
      "learning_rate": 3.148148148148148e-07,
      "loss": 0.0058,
      "step": 8504
    },
    {
      "epoch": 0.984375,
      "grad_norm": 0.0518437996506691,
      "learning_rate": 3.125e-07,
      "loss": 0.0091,
      "step": 8505
    },
    {
      "epoch": 0.9844907407407407,
      "grad_norm": 0.05013054609298706,
      "learning_rate": 3.1018518518518523e-07,
      "loss": 0.0088,
      "step": 8506
    },
    {
      "epoch": 0.9846064814814814,
      "grad_norm": 0.052543021738529205,
      "learning_rate": 3.078703703703704e-07,
      "loss": 0.0067,
      "step": 8507
    },
    {
      "epoch": 0.9847222222222223,
      "grad_norm": 0.05004999414086342,
      "learning_rate": 3.055555555555556e-07,
      "loss": 0.009,
      "step": 8508
    },
    {
      "epoch": 0.984837962962963,
      "grad_norm": 0.20469507575035095,
      "learning_rate": 3.0324074074074073e-07,
      "loss": 0.0086,
      "step": 8509
    },
    {
      "epoch": 0.9849537037037037,
      "grad_norm": 0.042275018990039825,
      "learning_rate": 3.0092592592592594e-07,
      "loss": 0.0075,
      "step": 8510
    },
    {
      "epoch": 0.9850694444444444,
      "grad_norm": 0.039106421172618866,
      "learning_rate": 2.9861111111111114e-07,
      "loss": 0.0051,
      "step": 8511
    },
    {
      "epoch": 0.9851851851851852,
      "grad_norm": 0.03258245065808296,
      "learning_rate": 2.9629629629629634e-07,
      "loss": 0.0058,
      "step": 8512
    },
    {
      "epoch": 0.9853009259259259,
      "grad_norm": 0.05399410426616669,
      "learning_rate": 2.939814814814815e-07,
      "loss": 0.0074,
      "step": 8513
    },
    {
      "epoch": 0.9854166666666667,
      "grad_norm": 0.04048997908830643,
      "learning_rate": 2.916666666666667e-07,
      "loss": 0.0073,
      "step": 8514
    },
    {
      "epoch": 0.9855324074074074,
      "grad_norm": 0.040885064750909805,
      "learning_rate": 2.893518518518519e-07,
      "loss": 0.0071,
      "step": 8515
    },
    {
      "epoch": 0.9856481481481482,
      "grad_norm": 0.04943612217903137,
      "learning_rate": 2.8703703703703705e-07,
      "loss": 0.0063,
      "step": 8516
    },
    {
      "epoch": 0.9857638888888889,
      "grad_norm": 0.043159157037734985,
      "learning_rate": 2.8472222222222225e-07,
      "loss": 0.0076,
      "step": 8517
    },
    {
      "epoch": 0.9858796296296296,
      "grad_norm": 0.03280705213546753,
      "learning_rate": 2.824074074074074e-07,
      "loss": 0.0059,
      "step": 8518
    },
    {
      "epoch": 0.9859953703703703,
      "grad_norm": 0.0416310653090477,
      "learning_rate": 2.800925925925926e-07,
      "loss": 0.0053,
      "step": 8519
    },
    {
      "epoch": 0.9861111111111112,
      "grad_norm": 0.04833924025297165,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 0.0063,
      "step": 8520
    },
    {
      "epoch": 0.9862268518518519,
      "grad_norm": 0.036234013736248016,
      "learning_rate": 2.75462962962963e-07,
      "loss": 0.0065,
      "step": 8521
    },
    {
      "epoch": 0.9863425925925926,
      "grad_norm": 0.03845756873488426,
      "learning_rate": 2.7314814814814817e-07,
      "loss": 0.005,
      "step": 8522
    },
    {
      "epoch": 0.9864583333333333,
      "grad_norm": 14.849140167236328,
      "learning_rate": 2.7083333333333337e-07,
      "loss": 3.1868,
      "step": 8523
    },
    {
      "epoch": 0.986574074074074,
      "grad_norm": 75.91948699951172,
      "learning_rate": 2.685185185185185e-07,
      "loss": 0.2986,
      "step": 8524
    },
    {
      "epoch": 0.9866898148148148,
      "grad_norm": 0.03969741240143776,
      "learning_rate": 2.662037037037037e-07,
      "loss": 0.0071,
      "step": 8525
    },
    {
      "epoch": 0.9868055555555556,
      "grad_norm": 0.04093853384256363,
      "learning_rate": 2.6388888888888893e-07,
      "loss": 0.0074,
      "step": 8526
    },
    {
      "epoch": 0.9869212962962963,
      "grad_norm": 1.106554627418518,
      "learning_rate": 2.615740740740741e-07,
      "loss": 0.0114,
      "step": 8527
    },
    {
      "epoch": 0.987037037037037,
      "grad_norm": 0.033961694687604904,
      "learning_rate": 2.592592592592593e-07,
      "loss": 0.006,
      "step": 8528
    },
    {
      "epoch": 0.9871527777777778,
      "grad_norm": 0.02794300764799118,
      "learning_rate": 2.5694444444444443e-07,
      "loss": 0.0051,
      "step": 8529
    },
    {
      "epoch": 0.9872685185185185,
      "grad_norm": 0.03042500466108322,
      "learning_rate": 2.5462962962962963e-07,
      "loss": 0.0054,
      "step": 8530
    },
    {
      "epoch": 0.9873842592592592,
      "grad_norm": 0.03436707705259323,
      "learning_rate": 2.5231481481481484e-07,
      "loss": 0.0062,
      "step": 8531
    },
    {
      "epoch": 0.9875,
      "grad_norm": 0.03461512178182602,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.0063,
      "step": 8532
    },
    {
      "epoch": 0.9876157407407408,
      "grad_norm": 0.10801012068986893,
      "learning_rate": 2.476851851851852e-07,
      "loss": 0.0078,
      "step": 8533
    },
    {
      "epoch": 0.9877314814814815,
      "grad_norm": 0.11132451891899109,
      "learning_rate": 2.453703703703704e-07,
      "loss": 0.0124,
      "step": 8534
    },
    {
      "epoch": 0.9878472222222222,
      "grad_norm": 0.04344279691576958,
      "learning_rate": 2.430555555555556e-07,
      "loss": 0.0078,
      "step": 8535
    },
    {
      "epoch": 0.9879629629629629,
      "grad_norm": 0.07580671459436417,
      "learning_rate": 2.4074074074074075e-07,
      "loss": 0.0089,
      "step": 8536
    },
    {
      "epoch": 0.9880787037037037,
      "grad_norm": 0.0424896776676178,
      "learning_rate": 2.3842592592592595e-07,
      "loss": 0.0077,
      "step": 8537
    },
    {
      "epoch": 0.9881944444444445,
      "grad_norm": 0.04147704690694809,
      "learning_rate": 2.3611111111111113e-07,
      "loss": 0.0073,
      "step": 8538
    },
    {
      "epoch": 0.9883101851851852,
      "grad_norm": 0.037925273180007935,
      "learning_rate": 2.337962962962963e-07,
      "loss": 0.0067,
      "step": 8539
    },
    {
      "epoch": 0.9884259259259259,
      "grad_norm": 0.040910571813583374,
      "learning_rate": 2.3148148148148148e-07,
      "loss": 0.0073,
      "step": 8540
    },
    {
      "epoch": 0.9885416666666667,
      "grad_norm": 0.03963904827833176,
      "learning_rate": 2.2916666666666666e-07,
      "loss": 0.007,
      "step": 8541
    },
    {
      "epoch": 0.9886574074074074,
      "grad_norm": 0.04672487452626228,
      "learning_rate": 2.268518518518519e-07,
      "loss": 0.0075,
      "step": 8542
    },
    {
      "epoch": 0.9887731481481481,
      "grad_norm": 0.036843616515398026,
      "learning_rate": 2.2453703703703707e-07,
      "loss": 0.0067,
      "step": 8543
    },
    {
      "epoch": 0.9888888888888889,
      "grad_norm": 0.060771770775318146,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.0069,
      "step": 8544
    },
    {
      "epoch": 0.9890046296296297,
      "grad_norm": 0.04500812664628029,
      "learning_rate": 2.1990740740740742e-07,
      "loss": 0.0056,
      "step": 8545
    },
    {
      "epoch": 0.9891203703703704,
      "grad_norm": 0.05200362578034401,
      "learning_rate": 2.1759259259259262e-07,
      "loss": 0.0091,
      "step": 8546
    },
    {
      "epoch": 0.9892361111111111,
      "grad_norm": 0.03254631906747818,
      "learning_rate": 2.152777777777778e-07,
      "loss": 0.0058,
      "step": 8547
    },
    {
      "epoch": 0.9893518518518518,
      "grad_norm": 0.04728212207555771,
      "learning_rate": 2.1296296296296298e-07,
      "loss": 0.0083,
      "step": 8548
    },
    {
      "epoch": 0.9894675925925925,
      "grad_norm": 0.04600485786795616,
      "learning_rate": 2.1064814814814816e-07,
      "loss": 0.0071,
      "step": 8549
    },
    {
      "epoch": 0.9895833333333334,
      "grad_norm": 0.06962046772241592,
      "learning_rate": 2.0833333333333333e-07,
      "loss": 0.0089,
      "step": 8550
    },
    {
      "epoch": 0.9896990740740741,
      "grad_norm": 0.04415929317474365,
      "learning_rate": 2.060185185185185e-07,
      "loss": 0.0057,
      "step": 8551
    },
    {
      "epoch": 0.9898148148148148,
      "grad_norm": 0.10728110373020172,
      "learning_rate": 2.0370370370370374e-07,
      "loss": 0.0069,
      "step": 8552
    },
    {
      "epoch": 0.9899305555555555,
      "grad_norm": 0.053193870931863785,
      "learning_rate": 2.0138888888888892e-07,
      "loss": 0.0097,
      "step": 8553
    },
    {
      "epoch": 0.9900462962962963,
      "grad_norm": 0.042989663779735565,
      "learning_rate": 1.990740740740741e-07,
      "loss": 0.0077,
      "step": 8554
    },
    {
      "epoch": 0.9901620370370371,
      "grad_norm": 0.034738343209028244,
      "learning_rate": 1.9675925925925927e-07,
      "loss": 0.0062,
      "step": 8555
    },
    {
      "epoch": 0.9902777777777778,
      "grad_norm": 0.031816206872463226,
      "learning_rate": 1.9444444444444447e-07,
      "loss": 0.0057,
      "step": 8556
    },
    {
      "epoch": 0.9903935185185185,
      "grad_norm": 0.039682261645793915,
      "learning_rate": 1.9212962962962965e-07,
      "loss": 0.0068,
      "step": 8557
    },
    {
      "epoch": 0.9905092592592593,
      "grad_norm": 0.05085286498069763,
      "learning_rate": 1.8981481481481483e-07,
      "loss": 0.0082,
      "step": 8558
    },
    {
      "epoch": 0.990625,
      "grad_norm": 0.0472385473549366,
      "learning_rate": 1.875e-07,
      "loss": 0.0084,
      "step": 8559
    },
    {
      "epoch": 0.9907407407407407,
      "grad_norm": 0.05938674136996269,
      "learning_rate": 1.8518518518518518e-07,
      "loss": 0.0068,
      "step": 8560
    },
    {
      "epoch": 0.9908564814814815,
      "grad_norm": 0.05368654057383537,
      "learning_rate": 1.8287037037037036e-07,
      "loss": 0.0085,
      "step": 8561
    },
    {
      "epoch": 0.9909722222222223,
      "grad_norm": 9.338436126708984,
      "learning_rate": 1.8055555555555556e-07,
      "loss": 0.0294,
      "step": 8562
    },
    {
      "epoch": 0.991087962962963,
      "grad_norm": 0.04282490536570549,
      "learning_rate": 1.7824074074074077e-07,
      "loss": 0.0076,
      "step": 8563
    },
    {
      "epoch": 0.9912037037037037,
      "grad_norm": 0.03726855292916298,
      "learning_rate": 1.7592592592592594e-07,
      "loss": 0.0068,
      "step": 8564
    },
    {
      "epoch": 0.9913194444444444,
      "grad_norm": 0.06348847597837448,
      "learning_rate": 1.7361111111111115e-07,
      "loss": 0.0087,
      "step": 8565
    },
    {
      "epoch": 0.9914351851851851,
      "grad_norm": 0.04658232256770134,
      "learning_rate": 1.7129629629629632e-07,
      "loss": 0.0064,
      "step": 8566
    },
    {
      "epoch": 0.991550925925926,
      "grad_norm": 0.04870889335870743,
      "learning_rate": 1.689814814814815e-07,
      "loss": 0.0088,
      "step": 8567
    },
    {
      "epoch": 0.9916666666666667,
      "grad_norm": 0.05316809192299843,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 0.0079,
      "step": 8568
    },
    {
      "epoch": 0.9917824074074074,
      "grad_norm": 0.03124709241092205,
      "learning_rate": 1.6435185185185185e-07,
      "loss": 0.0056,
      "step": 8569
    },
    {
      "epoch": 0.9918981481481481,
      "grad_norm": 0.042538903653621674,
      "learning_rate": 1.6203703703703703e-07,
      "loss": 0.0071,
      "step": 8570
    },
    {
      "epoch": 0.9920138888888889,
      "grad_norm": 0.044016312807798386,
      "learning_rate": 1.5972222222222223e-07,
      "loss": 0.0078,
      "step": 8571
    },
    {
      "epoch": 0.9921296296296296,
      "grad_norm": 0.03416658565402031,
      "learning_rate": 1.574074074074074e-07,
      "loss": 0.0061,
      "step": 8572
    },
    {
      "epoch": 0.9922453703703704,
      "grad_norm": 0.05388482287526131,
      "learning_rate": 1.5509259259259261e-07,
      "loss": 0.009,
      "step": 8573
    },
    {
      "epoch": 0.9923611111111111,
      "grad_norm": 0.04906708374619484,
      "learning_rate": 1.527777777777778e-07,
      "loss": 0.0085,
      "step": 8574
    },
    {
      "epoch": 0.9924768518518519,
      "grad_norm": 0.04797270894050598,
      "learning_rate": 1.5046296296296297e-07,
      "loss": 0.0085,
      "step": 8575
    },
    {
      "epoch": 0.9925925925925926,
      "grad_norm": 0.033773958683013916,
      "learning_rate": 1.4814814814814817e-07,
      "loss": 0.0061,
      "step": 8576
    },
    {
      "epoch": 0.9927083333333333,
      "grad_norm": 0.04434015601873398,
      "learning_rate": 1.4583333333333335e-07,
      "loss": 0.0058,
      "step": 8577
    },
    {
      "epoch": 0.992824074074074,
      "grad_norm": 0.035032592713832855,
      "learning_rate": 1.4351851851851853e-07,
      "loss": 0.0059,
      "step": 8578
    },
    {
      "epoch": 0.9929398148148149,
      "grad_norm": 0.04932721331715584,
      "learning_rate": 1.412037037037037e-07,
      "loss": 0.0078,
      "step": 8579
    },
    {
      "epoch": 0.9930555555555556,
      "grad_norm": 0.039811696857213974,
      "learning_rate": 1.3888888888888888e-07,
      "loss": 0.0066,
      "step": 8580
    },
    {
      "epoch": 0.9931712962962963,
      "grad_norm": 0.05014076083898544,
      "learning_rate": 1.3657407407407408e-07,
      "loss": 0.0065,
      "step": 8581
    },
    {
      "epoch": 0.993287037037037,
      "grad_norm": 0.057708751410245895,
      "learning_rate": 1.3425925925925926e-07,
      "loss": 0.0097,
      "step": 8582
    },
    {
      "epoch": 0.9934027777777777,
      "grad_norm": 0.04384395852684975,
      "learning_rate": 1.3194444444444446e-07,
      "loss": 0.0079,
      "step": 8583
    },
    {
      "epoch": 0.9935185185185185,
      "grad_norm": 0.038530416786670685,
      "learning_rate": 1.2962962962962964e-07,
      "loss": 0.0063,
      "step": 8584
    },
    {
      "epoch": 0.9936342592592593,
      "grad_norm": 0.038996148854494095,
      "learning_rate": 1.2731481481481482e-07,
      "loss": 0.0058,
      "step": 8585
    },
    {
      "epoch": 0.99375,
      "grad_norm": 1.2403970956802368,
      "learning_rate": 1.2500000000000002e-07,
      "loss": 0.0158,
      "step": 8586
    },
    {
      "epoch": 0.9938657407407407,
      "grad_norm": 0.02827265113592148,
      "learning_rate": 1.226851851851852e-07,
      "loss": 0.0051,
      "step": 8587
    },
    {
      "epoch": 0.9939814814814815,
      "grad_norm": 0.04924679920077324,
      "learning_rate": 1.2037037037037037e-07,
      "loss": 0.0089,
      "step": 8588
    },
    {
      "epoch": 0.9940972222222222,
      "grad_norm": 0.03578106313943863,
      "learning_rate": 1.1805555555555556e-07,
      "loss": 0.0064,
      "step": 8589
    },
    {
      "epoch": 0.9942129629629629,
      "grad_norm": 6.09435510635376,
      "learning_rate": 1.1574074074074074e-07,
      "loss": 0.0269,
      "step": 8590
    },
    {
      "epoch": 0.9943287037037037,
      "grad_norm": 0.02990189753472805,
      "learning_rate": 1.1342592592592595e-07,
      "loss": 0.0054,
      "step": 8591
    },
    {
      "epoch": 0.9944444444444445,
      "grad_norm": 0.03940217196941376,
      "learning_rate": 1.1111111111111112e-07,
      "loss": 0.007,
      "step": 8592
    },
    {
      "epoch": 0.9945601851851852,
      "grad_norm": 0.04265856370329857,
      "learning_rate": 1.0879629629629631e-07,
      "loss": 0.0075,
      "step": 8593
    },
    {
      "epoch": 0.9946759259259259,
      "grad_norm": 1.6351850032806396,
      "learning_rate": 1.0648148148148149e-07,
      "loss": 0.0204,
      "step": 8594
    },
    {
      "epoch": 0.9947916666666666,
      "grad_norm": 0.031547706574201584,
      "learning_rate": 1.0416666666666667e-07,
      "loss": 0.0054,
      "step": 8595
    },
    {
      "epoch": 0.9949074074074075,
      "grad_norm": 0.04786114767193794,
      "learning_rate": 1.0185185185185187e-07,
      "loss": 0.0062,
      "step": 8596
    },
    {
      "epoch": 0.9950231481481482,
      "grad_norm": 0.029407845810055733,
      "learning_rate": 9.953703703703705e-08,
      "loss": 0.0053,
      "step": 8597
    },
    {
      "epoch": 0.9951388888888889,
      "grad_norm": 0.04074941948056221,
      "learning_rate": 9.722222222222224e-08,
      "loss": 0.0071,
      "step": 8598
    },
    {
      "epoch": 0.9952546296296296,
      "grad_norm": 0.03635404258966446,
      "learning_rate": 9.490740740740741e-08,
      "loss": 0.0066,
      "step": 8599
    },
    {
      "epoch": 0.9953703703703703,
      "grad_norm": 0.04548713192343712,
      "learning_rate": 9.259259259259259e-08,
      "loss": 0.0079,
      "step": 8600
    },
    {
      "epoch": 0.9954861111111111,
      "grad_norm": 0.03225128352642059,
      "learning_rate": 9.027777777777778e-08,
      "loss": 0.0057,
      "step": 8601
    },
    {
      "epoch": 0.9956018518518519,
      "grad_norm": 0.029938921332359314,
      "learning_rate": 8.796296296296297e-08,
      "loss": 0.0054,
      "step": 8602
    },
    {
      "epoch": 0.9957175925925926,
      "grad_norm": 0.2244551181793213,
      "learning_rate": 8.564814814814816e-08,
      "loss": 0.0113,
      "step": 8603
    },
    {
      "epoch": 0.9958333333333333,
      "grad_norm": 0.049832701683044434,
      "learning_rate": 8.333333333333334e-08,
      "loss": 0.0086,
      "step": 8604
    },
    {
      "epoch": 0.9959490740740741,
      "grad_norm": 0.038136985152959824,
      "learning_rate": 8.101851851851852e-08,
      "loss": 0.0067,
      "step": 8605
    },
    {
      "epoch": 0.9960648148148148,
      "grad_norm": 0.05117850378155708,
      "learning_rate": 7.87037037037037e-08,
      "loss": 0.0074,
      "step": 8606
    },
    {
      "epoch": 0.9961805555555555,
      "grad_norm": 0.049130477011203766,
      "learning_rate": 7.63888888888889e-08,
      "loss": 0.0086,
      "step": 8607
    },
    {
      "epoch": 0.9962962962962963,
      "grad_norm": 0.04478546977043152,
      "learning_rate": 7.407407407407409e-08,
      "loss": 0.0079,
      "step": 8608
    },
    {
      "epoch": 0.9964120370370371,
      "grad_norm": 3.264068365097046,
      "learning_rate": 7.175925925925926e-08,
      "loss": 0.02,
      "step": 8609
    },
    {
      "epoch": 0.9965277777777778,
      "grad_norm": 0.13377954065799713,
      "learning_rate": 6.944444444444444e-08,
      "loss": 0.0103,
      "step": 8610
    },
    {
      "epoch": 0.9966435185185185,
      "grad_norm": 0.040175631642341614,
      "learning_rate": 6.712962962962963e-08,
      "loss": 0.0071,
      "step": 8611
    },
    {
      "epoch": 0.9967592592592592,
      "grad_norm": 0.03186540678143501,
      "learning_rate": 6.481481481481482e-08,
      "loss": 0.0057,
      "step": 8612
    },
    {
      "epoch": 0.996875,
      "grad_norm": 0.3143482804298401,
      "learning_rate": 6.250000000000001e-08,
      "loss": 0.007,
      "step": 8613
    },
    {
      "epoch": 0.9969907407407408,
      "grad_norm": 0.08309095352888107,
      "learning_rate": 6.018518518518519e-08,
      "loss": 0.011,
      "step": 8614
    },
    {
      "epoch": 0.9971064814814815,
      "grad_norm": 0.032895516604185104,
      "learning_rate": 5.787037037037037e-08,
      "loss": 0.0058,
      "step": 8615
    },
    {
      "epoch": 0.9972222222222222,
      "grad_norm": 0.5038349628448486,
      "learning_rate": 5.555555555555556e-08,
      "loss": 0.0125,
      "step": 8616
    },
    {
      "epoch": 0.997337962962963,
      "grad_norm": 0.049336910247802734,
      "learning_rate": 5.3240740740740745e-08,
      "loss": 0.0088,
      "step": 8617
    },
    {
      "epoch": 0.9974537037037037,
      "grad_norm": 0.047813523560762405,
      "learning_rate": 5.0925925925925935e-08,
      "loss": 0.0077,
      "step": 8618
    },
    {
      "epoch": 0.9975694444444444,
      "grad_norm": 0.06908373534679413,
      "learning_rate": 4.861111111111112e-08,
      "loss": 0.0091,
      "step": 8619
    },
    {
      "epoch": 0.9976851851851852,
      "grad_norm": 96.96414184570312,
      "learning_rate": 4.6296296296296295e-08,
      "loss": 2.4084,
      "step": 8620
    },
    {
      "epoch": 0.997800925925926,
      "grad_norm": 0.042384181171655655,
      "learning_rate": 4.3981481481481486e-08,
      "loss": 0.0076,
      "step": 8621
    },
    {
      "epoch": 0.9979166666666667,
      "grad_norm": 0.037463512271642685,
      "learning_rate": 4.166666666666667e-08,
      "loss": 0.0066,
      "step": 8622
    },
    {
      "epoch": 0.9980324074074074,
      "grad_norm": 0.04008089378476143,
      "learning_rate": 3.935185185185185e-08,
      "loss": 0.0071,
      "step": 8623
    },
    {
      "epoch": 0.9981481481481481,
      "grad_norm": 46.03755569458008,
      "learning_rate": 3.703703703703704e-08,
      "loss": 2.5998,
      "step": 8624
    },
    {
      "epoch": 0.9982638888888888,
      "grad_norm": 0.11792168021202087,
      "learning_rate": 3.472222222222222e-08,
      "loss": 0.008,
      "step": 8625
    },
    {
      "epoch": 0.9983796296296297,
      "grad_norm": 0.03389938175678253,
      "learning_rate": 3.240740740740741e-08,
      "loss": 0.0061,
      "step": 8626
    },
    {
      "epoch": 0.9984953703703704,
      "grad_norm": 0.04070289060473442,
      "learning_rate": 3.0092592592592594e-08,
      "loss": 0.0073,
      "step": 8627
    },
    {
      "epoch": 0.9986111111111111,
      "grad_norm": 0.03275568038225174,
      "learning_rate": 2.777777777777778e-08,
      "loss": 0.0058,
      "step": 8628
    },
    {
      "epoch": 0.9987268518518518,
      "grad_norm": 0.04893122613430023,
      "learning_rate": 2.5462962962962967e-08,
      "loss": 0.0087,
      "step": 8629
    },
    {
      "epoch": 0.9988425925925926,
      "grad_norm": 0.04104945808649063,
      "learning_rate": 2.3148148148148148e-08,
      "loss": 0.0065,
      "step": 8630
    },
    {
      "epoch": 0.9989583333333333,
      "grad_norm": 0.04078352451324463,
      "learning_rate": 2.0833333333333335e-08,
      "loss": 0.007,
      "step": 8631
    },
    {
      "epoch": 0.9990740740740741,
      "grad_norm": 0.048695359379053116,
      "learning_rate": 1.851851851851852e-08,
      "loss": 0.0085,
      "step": 8632
    },
    {
      "epoch": 0.9991898148148148,
      "grad_norm": 0.03867856785655022,
      "learning_rate": 1.6203703703703705e-08,
      "loss": 0.0067,
      "step": 8633
    },
    {
      "epoch": 0.9993055555555556,
      "grad_norm": 0.03215945512056351,
      "learning_rate": 1.388888888888889e-08,
      "loss": 0.0055,
      "step": 8634
    },
    {
      "epoch": 0.9994212962962963,
      "grad_norm": 0.04823770001530647,
      "learning_rate": 1.1574074074074074e-08,
      "loss": 0.0083,
      "step": 8635
    },
    {
      "epoch": 0.999537037037037,
      "grad_norm": 0.04391235113143921,
      "learning_rate": 9.25925925925926e-09,
      "loss": 0.0072,
      "step": 8636
    },
    {
      "epoch": 0.9996527777777777,
      "grad_norm": 0.04062962159514427,
      "learning_rate": 6.944444444444445e-09,
      "loss": 0.0071,
      "step": 8637
    },
    {
      "epoch": 0.9997685185185186,
      "grad_norm": 190.11463928222656,
      "learning_rate": 4.62962962962963e-09,
      "loss": 1.4428,
      "step": 8638
    },
    {
      "epoch": 0.9998842592592593,
      "grad_norm": 0.04771370068192482,
      "learning_rate": 2.314814814814815e-09,
      "loss": 0.0062,
      "step": 8639
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.040885426104068756,
      "learning_rate": 0.0,
      "loss": 0.0052,
      "step": 8640
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.1219615489244461,
      "eval_runtime": 669.7415,
      "eval_samples_per_second": 6.45,
      "eval_steps_per_second": 3.225,
      "step": 8640
    }
  ],
  "logging_steps": 1,
  "max_steps": 8640,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "EarlyStoppingCallback": {
      "args": {
        "early_stopping_patience": 1,
        "early_stopping_threshold": 0.0
      },
      "attributes": {
        "early_stopping_patience_counter": 0
      }
    },
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.3391575932233318e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}