{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 406,
  "global_step": 1624,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0006157635467980296,
      "grad_norm": 36.35367202758789,
      "learning_rate": 0.0,
      "loss": 1.0477,
      "step": 1
    },
    {
      "epoch": 0.0012315270935960591,
      "grad_norm": 14.517570495605469,
      "learning_rate": 1.0686215613240665e-06,
      "loss": 0.9934,
      "step": 2
    },
    {
      "epoch": 0.0018472906403940886,
      "grad_norm": 11.640591621398926,
      "learning_rate": 1.693725102160739e-06,
      "loss": 1.002,
      "step": 3
    },
    {
      "epoch": 0.0024630541871921183,
      "grad_norm": 20.654525756835938,
      "learning_rate": 2.137243122648133e-06,
      "loss": 1.0092,
      "step": 4
    },
    {
      "epoch": 0.003078817733990148,
      "grad_norm": 7.632515907287598,
      "learning_rate": 2.4812624260407488e-06,
      "loss": 1.0149,
      "step": 5
    },
    {
      "epoch": 0.003694581280788177,
      "grad_norm": 9.955108642578125,
      "learning_rate": 2.7623466634848054e-06,
      "loss": 0.9808,
      "step": 6
    },
    {
      "epoch": 0.004310344827586207,
      "grad_norm": 15.791479110717773,
      "learning_rate": 3e-06,
      "loss": 1.003,
      "step": 7
    },
    {
      "epoch": 0.0049261083743842365,
      "grad_norm": 5.557374477386475,
      "learning_rate": 3.2058646839722e-06,
      "loss": 0.9222,
      "step": 8
    },
    {
      "epoch": 0.005541871921182266,
      "grad_norm": 7.952128887176514,
      "learning_rate": 3.387450204321478e-06,
      "loss": 0.9575,
      "step": 9
    },
    {
      "epoch": 0.006157635467980296,
      "grad_norm": 4.459101676940918,
      "learning_rate": 3.5498839873648155e-06,
      "loss": 0.9328,
      "step": 10
    },
    {
      "epoch": 0.0067733990147783255,
      "grad_norm": 3.5258655548095703,
      "learning_rate": 3.6968232176020314e-06,
      "loss": 0.9137,
      "step": 11
    },
    {
      "epoch": 0.007389162561576354,
      "grad_norm": 2.4829490184783936,
      "learning_rate": 3.830968224808873e-06,
      "loss": 0.9308,
      "step": 12
    },
    {
      "epoch": 0.008004926108374385,
      "grad_norm": 4.386059284210205,
      "learning_rate": 3.954369669185523e-06,
      "loss": 0.925,
      "step": 13
    },
    {
      "epoch": 0.008620689655172414,
      "grad_norm": 1.8689730167388916,
      "learning_rate": 4.068621561324066e-06,
      "loss": 0.9062,
      "step": 14
    },
    {
      "epoch": 0.009236453201970444,
      "grad_norm": 8.052454948425293,
      "learning_rate": 4.174987528201488e-06,
      "loss": 0.9206,
      "step": 15
    },
    {
      "epoch": 0.009852216748768473,
      "grad_norm": 17.698686599731445,
      "learning_rate": 4.274486245296266e-06,
      "loss": 0.9546,
      "step": 16
    },
    {
      "epoch": 0.010467980295566502,
      "grad_norm": 3.5210695266723633,
      "learning_rate": 4.367950923271043e-06,
      "loss": 0.914,
      "step": 17
    },
    {
      "epoch": 0.011083743842364532,
      "grad_norm": 1.9735019207000732,
      "learning_rate": 4.4560717656455445e-06,
      "loss": 0.9316,
      "step": 18
    },
    {
      "epoch": 0.011699507389162561,
      "grad_norm": 1.9073392152786255,
      "learning_rate": 4.539426931807544e-06,
      "loss": 0.961,
      "step": 19
    },
    {
      "epoch": 0.012315270935960592,
      "grad_norm": 1.5129183530807495,
      "learning_rate": 4.618505548688882e-06,
      "loss": 0.9023,
      "step": 20
    },
    {
      "epoch": 0.01293103448275862,
      "grad_norm": 1.8761876821517944,
      "learning_rate": 4.693725102160739e-06,
      "loss": 0.9107,
      "step": 21
    },
    {
      "epoch": 0.013546798029556651,
      "grad_norm": 1.8950828313827515,
      "learning_rate": 4.765444778926099e-06,
      "loss": 0.9433,
      "step": 22
    },
    {
      "epoch": 0.01416256157635468,
      "grad_norm": 1.926555871963501,
      "learning_rate": 4.833975840227794e-06,
      "loss": 0.9605,
      "step": 23
    },
    {
      "epoch": 0.014778325123152709,
      "grad_norm": 1.726905107498169,
      "learning_rate": 4.899589786132939e-06,
      "loss": 0.9187,
      "step": 24
    },
    {
      "epoch": 0.01539408866995074,
      "grad_norm": 1.498301386833191,
      "learning_rate": 4.9625248520814976e-06,
      "loss": 0.903,
      "step": 25
    },
    {
      "epoch": 0.01600985221674877,
      "grad_norm": 2.279438018798828,
      "learning_rate": 5.02299123050959e-06,
      "loss": 0.9321,
      "step": 26
    },
    {
      "epoch": 0.0166256157635468,
      "grad_norm": 1.6443735361099243,
      "learning_rate": 5.0811753064822164e-06,
      "loss": 0.8848,
      "step": 27
    },
    {
      "epoch": 0.017241379310344827,
      "grad_norm": 1.5222970247268677,
      "learning_rate": 5.137243122648134e-06,
      "loss": 0.9094,
      "step": 28
    },
    {
      "epoch": 0.017857142857142856,
      "grad_norm": 1.63950514793396,
      "learning_rate": 5.1913432358958695e-06,
      "loss": 0.9491,
      "step": 29
    },
    {
      "epoch": 0.01847290640394089,
      "grad_norm": 1.4933983087539673,
      "learning_rate": 5.243609089525555e-06,
      "loss": 0.9357,
      "step": 30
    },
    {
      "epoch": 0.019088669950738917,
      "grad_norm": 1.6717708110809326,
      "learning_rate": 5.294160996311553e-06,
      "loss": 0.9066,
      "step": 31
    },
    {
      "epoch": 0.019704433497536946,
      "grad_norm": 1.5995877981185913,
      "learning_rate": 5.343107806620333e-06,
      "loss": 0.9405,
      "step": 32
    },
    {
      "epoch": 0.020320197044334975,
      "grad_norm": 1.4515255689620972,
      "learning_rate": 5.390548319762771e-06,
      "loss": 0.9235,
      "step": 33
    },
    {
      "epoch": 0.020935960591133004,
      "grad_norm": 1.7589017152786255,
      "learning_rate": 5.43657248459511e-06,
      "loss": 0.9382,
      "step": 34
    },
    {
      "epoch": 0.021551724137931036,
      "grad_norm": 1.682663917541504,
      "learning_rate": 5.4812624260407484e-06,
      "loss": 0.9016,
      "step": 35
    },
    {
      "epoch": 0.022167487684729065,
      "grad_norm": 1.502163290977478,
      "learning_rate": 5.524693326969611e-06,
      "loss": 0.9146,
      "step": 36
    },
    {
      "epoch": 0.022783251231527094,
      "grad_norm": 1.9817074537277222,
      "learning_rate": 5.566934189223322e-06,
      "loss": 0.9132,
      "step": 37
    },
    {
      "epoch": 0.023399014778325122,
      "grad_norm": 1.7520332336425781,
      "learning_rate": 5.608048493131611e-06,
      "loss": 0.9192,
      "step": 38
    },
    {
      "epoch": 0.02401477832512315,
      "grad_norm": 1.6490265130996704,
      "learning_rate": 5.648094771346262e-06,
      "loss": 0.9062,
      "step": 39
    },
    {
      "epoch": 0.024630541871921183,
      "grad_norm": 1.5501534938812256,
      "learning_rate": 5.687127110012948e-06,
      "loss": 0.8859,
      "step": 40
    },
    {
      "epoch": 0.025246305418719212,
      "grad_norm": 1.271614670753479,
      "learning_rate": 5.72519558804986e-06,
      "loss": 0.9457,
      "step": 41
    },
    {
      "epoch": 0.02586206896551724,
      "grad_norm": 1.3739732503890991,
      "learning_rate": 5.7623466634848055e-06,
      "loss": 0.8791,
      "step": 42
    },
    {
      "epoch": 0.02647783251231527,
      "grad_norm": 1.4129613637924194,
      "learning_rate": 5.79862351432751e-06,
      "loss": 0.8953,
      "step": 43
    },
    {
      "epoch": 0.027093596059113302,
      "grad_norm": 1.7332401275634766,
      "learning_rate": 5.834066340250165e-06,
      "loss": 0.8788,
      "step": 44
    },
    {
      "epoch": 0.02770935960591133,
      "grad_norm": 1.4359893798828125,
      "learning_rate": 5.8687126303622265e-06,
      "loss": 0.9116,
      "step": 45
    },
    {
      "epoch": 0.02832512315270936,
      "grad_norm": 1.384950041770935,
      "learning_rate": 5.902597401551861e-06,
      "loss": 0.9223,
      "step": 46
    },
    {
      "epoch": 0.02894088669950739,
      "grad_norm": 1.6219000816345215,
      "learning_rate": 5.935753411193011e-06,
      "loss": 0.8719,
      "step": 47
    },
    {
      "epoch": 0.029556650246305417,
      "grad_norm": 1.4559160470962524,
      "learning_rate": 5.968211347457006e-06,
      "loss": 0.8746,
      "step": 48
    },
    {
      "epoch": 0.03017241379310345,
      "grad_norm": 1.4369477033615112,
      "learning_rate": 6e-06,
      "loss": 0.9112,
      "step": 49
    },
    {
      "epoch": 0.03078817733990148,
      "grad_norm": 1.4078340530395508,
      "learning_rate": 6e-06,
      "loss": 0.8984,
      "step": 50
    },
    {
      "epoch": 0.03140394088669951,
      "grad_norm": 1.4506973028182983,
      "learning_rate": 5.996190476190477e-06,
      "loss": 0.9142,
      "step": 51
    },
    {
      "epoch": 0.03201970443349754,
      "grad_norm": 1.4609692096710205,
      "learning_rate": 5.992380952380952e-06,
      "loss": 0.8947,
      "step": 52
    },
    {
      "epoch": 0.03263546798029557,
      "grad_norm": 1.3819327354431152,
      "learning_rate": 5.9885714285714294e-06,
      "loss": 0.9226,
      "step": 53
    },
    {
      "epoch": 0.0332512315270936,
      "grad_norm": 1.5369493961334229,
      "learning_rate": 5.984761904761905e-06,
      "loss": 0.9021,
      "step": 54
    },
    {
      "epoch": 0.033866995073891626,
      "grad_norm": 1.3379381895065308,
      "learning_rate": 5.9809523809523814e-06,
      "loss": 0.9224,
      "step": 55
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 1.4000790119171143,
      "learning_rate": 5.977142857142857e-06,
      "loss": 0.9063,
      "step": 56
    },
    {
      "epoch": 0.035098522167487683,
      "grad_norm": 1.410876750946045,
      "learning_rate": 5.9733333333333335e-06,
      "loss": 0.9008,
      "step": 57
    },
    {
      "epoch": 0.03571428571428571,
      "grad_norm": 1.3722566366195679,
      "learning_rate": 5.969523809523809e-06,
      "loss": 0.8492,
      "step": 58
    },
    {
      "epoch": 0.03633004926108374,
      "grad_norm": 1.6557255983352661,
      "learning_rate": 5.965714285714286e-06,
      "loss": 0.8862,
      "step": 59
    },
    {
      "epoch": 0.03694581280788178,
      "grad_norm": 1.3114638328552246,
      "learning_rate": 5.961904761904762e-06,
      "loss": 0.9188,
      "step": 60
    },
    {
      "epoch": 0.037561576354679806,
      "grad_norm": 1.4468998908996582,
      "learning_rate": 5.958095238095238e-06,
      "loss": 0.9668,
      "step": 61
    },
    {
      "epoch": 0.038177339901477834,
      "grad_norm": 1.4321094751358032,
      "learning_rate": 5.954285714285715e-06,
      "loss": 0.9095,
      "step": 62
    },
    {
      "epoch": 0.03879310344827586,
      "grad_norm": 1.4419448375701904,
      "learning_rate": 5.95047619047619e-06,
      "loss": 0.894,
      "step": 63
    },
    {
      "epoch": 0.03940886699507389,
      "grad_norm": 1.5440068244934082,
      "learning_rate": 5.946666666666667e-06,
      "loss": 0.8869,
      "step": 64
    },
    {
      "epoch": 0.04002463054187192,
      "grad_norm": 1.5087432861328125,
      "learning_rate": 5.942857142857143e-06,
      "loss": 0.8716,
      "step": 65
    },
    {
      "epoch": 0.04064039408866995,
      "grad_norm": 1.5499300956726074,
      "learning_rate": 5.93904761904762e-06,
      "loss": 0.8643,
      "step": 66
    },
    {
      "epoch": 0.04125615763546798,
      "grad_norm": 1.4838142395019531,
      "learning_rate": 5.935238095238095e-06,
      "loss": 0.846,
      "step": 67
    },
    {
      "epoch": 0.04187192118226601,
      "grad_norm": 1.5086207389831543,
      "learning_rate": 5.931428571428572e-06,
      "loss": 0.8869,
      "step": 68
    },
    {
      "epoch": 0.042487684729064036,
      "grad_norm": 1.2491344213485718,
      "learning_rate": 5.927619047619047e-06,
      "loss": 0.879,
      "step": 69
    },
    {
      "epoch": 0.04310344827586207,
      "grad_norm": 1.4236856698989868,
      "learning_rate": 5.9238095238095245e-06,
      "loss": 0.8971,
      "step": 70
    },
    {
      "epoch": 0.0437192118226601,
      "grad_norm": 1.4545221328735352,
      "learning_rate": 5.92e-06,
      "loss": 0.8977,
      "step": 71
    },
    {
      "epoch": 0.04433497536945813,
      "grad_norm": 1.3406494855880737,
      "learning_rate": 5.9161904761904765e-06,
      "loss": 0.9051,
      "step": 72
    },
    {
      "epoch": 0.04495073891625616,
      "grad_norm": 1.7517465353012085,
      "learning_rate": 5.912380952380952e-06,
      "loss": 0.8916,
      "step": 73
    },
    {
      "epoch": 0.04556650246305419,
      "grad_norm": 1.3098891973495483,
      "learning_rate": 5.9085714285714285e-06,
      "loss": 0.8734,
      "step": 74
    },
    {
      "epoch": 0.046182266009852216,
      "grad_norm": 1.458084225654602,
      "learning_rate": 5.904761904761905e-06,
      "loss": 0.8671,
      "step": 75
    },
    {
      "epoch": 0.046798029556650245,
      "grad_norm": 1.3404593467712402,
      "learning_rate": 5.900952380952381e-06,
      "loss": 0.8794,
      "step": 76
    },
    {
      "epoch": 0.04741379310344827,
      "grad_norm": 1.3528378009796143,
      "learning_rate": 5.897142857142858e-06,
      "loss": 0.8974,
      "step": 77
    },
    {
      "epoch": 0.0480295566502463,
      "grad_norm": 1.60671865940094,
      "learning_rate": 5.893333333333333e-06,
      "loss": 0.921,
      "step": 78
    },
    {
      "epoch": 0.04864532019704434,
      "grad_norm": 1.4191659688949585,
      "learning_rate": 5.88952380952381e-06,
      "loss": 0.8913,
      "step": 79
    },
    {
      "epoch": 0.04926108374384237,
      "grad_norm": 1.3047839403152466,
      "learning_rate": 5.885714285714285e-06,
      "loss": 0.8478,
      "step": 80
    },
    {
      "epoch": 0.049876847290640396,
      "grad_norm": 1.2785903215408325,
      "learning_rate": 5.881904761904763e-06,
      "loss": 0.8254,
      "step": 81
    },
    {
      "epoch": 0.050492610837438424,
      "grad_norm": 1.3927315473556519,
      "learning_rate": 5.878095238095238e-06,
      "loss": 0.8564,
      "step": 82
    },
    {
      "epoch": 0.05110837438423645,
      "grad_norm": 1.3651491403579712,
      "learning_rate": 5.874285714285715e-06,
      "loss": 0.8866,
      "step": 83
    },
    {
      "epoch": 0.05172413793103448,
      "grad_norm": 1.245163083076477,
      "learning_rate": 5.87047619047619e-06,
      "loss": 0.8848,
      "step": 84
    },
    {
      "epoch": 0.05233990147783251,
      "grad_norm": 1.4479676485061646,
      "learning_rate": 5.866666666666667e-06,
      "loss": 0.8764,
      "step": 85
    },
    {
      "epoch": 0.05295566502463054,
      "grad_norm": 1.3517879247665405,
      "learning_rate": 5.862857142857143e-06,
      "loss": 0.8596,
      "step": 86
    },
    {
      "epoch": 0.05357142857142857,
      "grad_norm": 1.4050191640853882,
      "learning_rate": 5.8590476190476195e-06,
      "loss": 0.899,
      "step": 87
    },
    {
      "epoch": 0.054187192118226604,
      "grad_norm": 1.0798320770263672,
      "learning_rate": 5.855238095238095e-06,
      "loss": 0.8836,
      "step": 88
    },
    {
      "epoch": 0.05480295566502463,
      "grad_norm": 1.2569149732589722,
      "learning_rate": 5.8514285714285716e-06,
      "loss": 0.8552,
      "step": 89
    },
    {
      "epoch": 0.05541871921182266,
      "grad_norm": 1.4904707670211792,
      "learning_rate": 5.847619047619048e-06,
      "loss": 0.8883,
      "step": 90
    },
    {
      "epoch": 0.05603448275862069,
      "grad_norm": 1.2472726106643677,
      "learning_rate": 5.8438095238095236e-06,
      "loss": 0.8945,
      "step": 91
    },
    {
      "epoch": 0.05665024630541872,
      "grad_norm": 1.3093783855438232,
      "learning_rate": 5.840000000000001e-06,
      "loss": 0.8401,
      "step": 92
    },
    {
      "epoch": 0.05726600985221675,
      "grad_norm": 1.163570523262024,
      "learning_rate": 5.836190476190476e-06,
      "loss": 0.882,
      "step": 93
    },
    {
      "epoch": 0.05788177339901478,
      "grad_norm": 1.3225419521331787,
      "learning_rate": 5.832380952380953e-06,
      "loss": 0.8916,
      "step": 94
    },
    {
      "epoch": 0.058497536945812806,
      "grad_norm": 1.4216433763504028,
      "learning_rate": 5.8285714285714284e-06,
      "loss": 0.8777,
      "step": 95
    },
    {
      "epoch": 0.059113300492610835,
      "grad_norm": 1.1359666585922241,
      "learning_rate": 5.824761904761905e-06,
      "loss": 0.8488,
      "step": 96
    },
    {
      "epoch": 0.05972906403940887,
      "grad_norm": 1.3381409645080566,
      "learning_rate": 5.8209523809523804e-06,
      "loss": 0.8724,
      "step": 97
    },
    {
      "epoch": 0.0603448275862069,
      "grad_norm": 1.3888726234436035,
      "learning_rate": 5.817142857142858e-06,
      "loss": 0.8501,
      "step": 98
    },
    {
      "epoch": 0.06096059113300493,
      "grad_norm": 1.161588191986084,
      "learning_rate": 5.813333333333333e-06,
      "loss": 0.8785,
      "step": 99
    },
    {
      "epoch": 0.06157635467980296,
      "grad_norm": 1.2433103322982788,
      "learning_rate": 5.80952380952381e-06,
      "loss": 0.8909,
      "step": 100
    },
    {
      "epoch": 0.062192118226600986,
      "grad_norm": 1.3087376356124878,
      "learning_rate": 5.805714285714286e-06,
      "loss": 0.8278,
      "step": 101
    },
    {
      "epoch": 0.06280788177339902,
      "grad_norm": 1.3910329341888428,
      "learning_rate": 5.801904761904762e-06,
      "loss": 0.8713,
      "step": 102
    },
    {
      "epoch": 0.06342364532019705,
      "grad_norm": 1.2319529056549072,
      "learning_rate": 5.798095238095238e-06,
      "loss": 0.8706,
      "step": 103
    },
    {
      "epoch": 0.06403940886699508,
      "grad_norm": 1.2095369100570679,
      "learning_rate": 5.794285714285715e-06,
      "loss": 0.8683,
      "step": 104
    },
    {
      "epoch": 0.06465517241379311,
      "grad_norm": 1.1708300113677979,
      "learning_rate": 5.790476190476191e-06,
      "loss": 0.8619,
      "step": 105
    },
    {
      "epoch": 0.06527093596059114,
      "grad_norm": 1.2331058979034424,
      "learning_rate": 5.786666666666667e-06,
      "loss": 0.8938,
      "step": 106
    },
    {
      "epoch": 0.06588669950738917,
      "grad_norm": 1.1638177633285522,
      "learning_rate": 5.782857142857143e-06,
      "loss": 0.8571,
      "step": 107
    },
    {
      "epoch": 0.0665024630541872,
      "grad_norm": 1.1121563911437988,
      "learning_rate": 5.779047619047619e-06,
      "loss": 0.8476,
      "step": 108
    },
    {
      "epoch": 0.06711822660098522,
      "grad_norm": 1.1893773078918457,
      "learning_rate": 5.775238095238096e-06,
      "loss": 0.8897,
      "step": 109
    },
    {
      "epoch": 0.06773399014778325,
      "grad_norm": 1.2557870149612427,
      "learning_rate": 5.7714285714285715e-06,
      "loss": 0.9003,
      "step": 110
    },
    {
      "epoch": 0.06834975369458128,
      "grad_norm": 1.1182951927185059,
      "learning_rate": 5.767619047619048e-06,
      "loss": 0.8872,
      "step": 111
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 1.1956005096435547,
      "learning_rate": 5.7638095238095235e-06,
      "loss": 0.9075,
      "step": 112
    },
    {
      "epoch": 0.06958128078817734,
      "grad_norm": 1.1626731157302856,
      "learning_rate": 5.76e-06,
      "loss": 0.9424,
      "step": 113
    },
    {
      "epoch": 0.07019704433497537,
      "grad_norm": 1.1537976264953613,
      "learning_rate": 5.756190476190476e-06,
      "loss": 0.8739,
      "step": 114
    },
    {
      "epoch": 0.0708128078817734,
      "grad_norm": 1.177437663078308,
      "learning_rate": 5.752380952380953e-06,
      "loss": 0.8312,
      "step": 115
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 1.1597214937210083,
      "learning_rate": 5.748571428571429e-06,
      "loss": 0.9234,
      "step": 116
    },
    {
      "epoch": 0.07204433497536945,
      "grad_norm": 1.1179101467132568,
      "learning_rate": 5.744761904761905e-06,
      "loss": 0.8994,
      "step": 117
    },
    {
      "epoch": 0.07266009852216748,
      "grad_norm": 1.6248124837875366,
      "learning_rate": 5.740952380952381e-06,
      "loss": 0.8911,
      "step": 118
    },
    {
      "epoch": 0.07327586206896551,
      "grad_norm": 1.1523460149765015,
      "learning_rate": 5.737142857142857e-06,
      "loss": 0.8693,
      "step": 119
    },
    {
      "epoch": 0.07389162561576355,
      "grad_norm": 1.3294626474380493,
      "learning_rate": 5.733333333333334e-06,
      "loss": 0.8969,
      "step": 120
    },
    {
      "epoch": 0.07450738916256158,
      "grad_norm": 1.1428542137145996,
      "learning_rate": 5.72952380952381e-06,
      "loss": 0.8875,
      "step": 121
    },
    {
      "epoch": 0.07512315270935961,
      "grad_norm": 1.1817996501922607,
      "learning_rate": 5.725714285714286e-06,
      "loss": 0.888,
      "step": 122
    },
    {
      "epoch": 0.07573891625615764,
      "grad_norm": 1.1294951438903809,
      "learning_rate": 5.721904761904762e-06,
      "loss": 0.8529,
      "step": 123
    },
    {
      "epoch": 0.07635467980295567,
      "grad_norm": 1.15193510055542,
      "learning_rate": 5.718095238095238e-06,
      "loss": 0.881,
      "step": 124
    },
    {
      "epoch": 0.0769704433497537,
      "grad_norm": 1.2102628946304321,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.926,
      "step": 125
    },
    {
      "epoch": 0.07758620689655173,
      "grad_norm": 1.1487412452697754,
      "learning_rate": 5.710476190476191e-06,
      "loss": 0.7985,
      "step": 126
    },
    {
      "epoch": 0.07820197044334976,
      "grad_norm": 1.2155689001083374,
      "learning_rate": 5.7066666666666665e-06,
      "loss": 0.8953,
      "step": 127
    },
    {
      "epoch": 0.07881773399014778,
      "grad_norm": 1.2132171392440796,
      "learning_rate": 5.702857142857143e-06,
      "loss": 0.8841,
      "step": 128
    },
    {
      "epoch": 0.07943349753694581,
      "grad_norm": 1.28113853931427,
      "learning_rate": 5.699047619047619e-06,
      "loss": 0.9125,
      "step": 129
    },
    {
      "epoch": 0.08004926108374384,
      "grad_norm": 1.1806972026824951,
      "learning_rate": 5.695238095238095e-06,
      "loss": 0.898,
      "step": 130
    },
    {
      "epoch": 0.08066502463054187,
      "grad_norm": 1.1579880714416504,
      "learning_rate": 5.691428571428572e-06,
      "loss": 0.8475,
      "step": 131
    },
    {
      "epoch": 0.0812807881773399,
      "grad_norm": 1.1883717775344849,
      "learning_rate": 5.687619047619048e-06,
      "loss": 0.821,
      "step": 132
    },
    {
      "epoch": 0.08189655172413793,
      "grad_norm": 1.2056052684783936,
      "learning_rate": 5.683809523809524e-06,
      "loss": 0.8923,
      "step": 133
    },
    {
      "epoch": 0.08251231527093596,
      "grad_norm": 1.1401325464248657,
      "learning_rate": 5.68e-06,
      "loss": 0.8391,
      "step": 134
    },
    {
      "epoch": 0.08312807881773399,
      "grad_norm": 1.137695074081421,
      "learning_rate": 5.676190476190476e-06,
      "loss": 0.8331,
      "step": 135
    },
    {
      "epoch": 0.08374384236453201,
      "grad_norm": 1.2140518426895142,
      "learning_rate": 5.672380952380952e-06,
      "loss": 0.8672,
      "step": 136
    },
    {
      "epoch": 0.08435960591133004,
      "grad_norm": 1.3218492269515991,
      "learning_rate": 5.668571428571429e-06,
      "loss": 0.8523,
      "step": 137
    },
    {
      "epoch": 0.08497536945812807,
      "grad_norm": 1.20176100730896,
      "learning_rate": 5.664761904761905e-06,
      "loss": 0.8455,
      "step": 138
    },
    {
      "epoch": 0.08559113300492611,
      "grad_norm": 1.1329739093780518,
      "learning_rate": 5.660952380952381e-06,
      "loss": 0.8937,
      "step": 139
    },
    {
      "epoch": 0.08620689655172414,
      "grad_norm": 1.2568690776824951,
      "learning_rate": 5.6571428571428576e-06,
      "loss": 0.8759,
      "step": 140
    },
    {
      "epoch": 0.08682266009852217,
      "grad_norm": 1.2636555433273315,
      "learning_rate": 5.653333333333333e-06,
      "loss": 0.8914,
      "step": 141
    },
    {
      "epoch": 0.0874384236453202,
      "grad_norm": 1.1239299774169922,
      "learning_rate": 5.6495238095238096e-06,
      "loss": 0.8935,
      "step": 142
    },
    {
      "epoch": 0.08805418719211823,
      "grad_norm": 1.1039814949035645,
      "learning_rate": 5.645714285714286e-06,
      "loss": 0.8585,
      "step": 143
    },
    {
      "epoch": 0.08866995073891626,
      "grad_norm": 1.1381434202194214,
      "learning_rate": 5.641904761904762e-06,
      "loss": 0.8793,
      "step": 144
    },
    {
      "epoch": 0.08928571428571429,
      "grad_norm": 1.1313204765319824,
      "learning_rate": 5.638095238095238e-06,
      "loss": 0.9302,
      "step": 145
    },
    {
      "epoch": 0.08990147783251232,
      "grad_norm": 1.1036722660064697,
      "learning_rate": 5.6342857142857144e-06,
      "loss": 0.8829,
      "step": 146
    },
    {
      "epoch": 0.09051724137931035,
      "grad_norm": 1.3259217739105225,
      "learning_rate": 5.63047619047619e-06,
      "loss": 0.8644,
      "step": 147
    },
    {
      "epoch": 0.09113300492610837,
      "grad_norm": 1.1917400360107422,
      "learning_rate": 5.626666666666667e-06,
      "loss": 0.8703,
      "step": 148
    },
    {
      "epoch": 0.0917487684729064,
      "grad_norm": 1.189653754234314,
      "learning_rate": 5.622857142857143e-06,
      "loss": 0.865,
      "step": 149
    },
    {
      "epoch": 0.09236453201970443,
      "grad_norm": 1.168756365776062,
      "learning_rate": 5.619047619047619e-06,
      "loss": 0.859,
      "step": 150
    },
    {
      "epoch": 0.09298029556650246,
      "grad_norm": 1.1856180429458618,
      "learning_rate": 5.615238095238095e-06,
      "loss": 0.9022,
      "step": 151
    },
    {
      "epoch": 0.09359605911330049,
      "grad_norm": 1.1183501482009888,
      "learning_rate": 5.611428571428571e-06,
      "loss": 0.8572,
      "step": 152
    },
    {
      "epoch": 0.09421182266009852,
      "grad_norm": 1.2948905229568481,
      "learning_rate": 5.607619047619048e-06,
      "loss": 0.877,
      "step": 153
    },
    {
      "epoch": 0.09482758620689655,
      "grad_norm": 1.1875604391098022,
      "learning_rate": 5.603809523809524e-06,
      "loss": 0.8781,
      "step": 154
    },
    {
      "epoch": 0.09544334975369458,
      "grad_norm": 1.3480019569396973,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.8664,
      "step": 155
    },
    {
      "epoch": 0.0960591133004926,
      "grad_norm": 1.1254591941833496,
      "learning_rate": 5.596190476190476e-06,
      "loss": 0.8481,
      "step": 156
    },
    {
      "epoch": 0.09667487684729065,
      "grad_norm": 1.1160861253738403,
      "learning_rate": 5.592380952380953e-06,
      "loss": 0.8609,
      "step": 157
    },
    {
      "epoch": 0.09729064039408868,
      "grad_norm": 1.1722326278686523,
      "learning_rate": 5.588571428571428e-06,
      "loss": 0.8672,
      "step": 158
    },
    {
      "epoch": 0.0979064039408867,
      "grad_norm": 1.0427522659301758,
      "learning_rate": 5.5847619047619055e-06,
      "loss": 0.8396,
      "step": 159
    },
    {
      "epoch": 0.09852216748768473,
      "grad_norm": 1.1800332069396973,
      "learning_rate": 5.580952380952381e-06,
      "loss": 0.8591,
      "step": 160
    },
    {
      "epoch": 0.09913793103448276,
      "grad_norm": 1.2190465927124023,
      "learning_rate": 5.5771428571428575e-06,
      "loss": 0.8305,
      "step": 161
    },
    {
      "epoch": 0.09975369458128079,
      "grad_norm": 1.1807329654693604,
      "learning_rate": 5.573333333333333e-06,
      "loss": 0.8957,
      "step": 162
    },
    {
      "epoch": 0.10036945812807882,
      "grad_norm": 1.4590678215026855,
      "learning_rate": 5.5695238095238095e-06,
      "loss": 0.847,
      "step": 163
    },
    {
      "epoch": 0.10098522167487685,
      "grad_norm": 1.1271862983703613,
      "learning_rate": 5.565714285714286e-06,
      "loss": 0.8972,
      "step": 164
    },
    {
      "epoch": 0.10160098522167488,
      "grad_norm": 1.078830599784851,
      "learning_rate": 5.561904761904762e-06,
      "loss": 0.8888,
      "step": 165
    },
    {
      "epoch": 0.1022167487684729,
      "grad_norm": 1.0548816919326782,
      "learning_rate": 5.558095238095238e-06,
      "loss": 0.8661,
      "step": 166
    },
    {
      "epoch": 0.10283251231527094,
      "grad_norm": 1.3490275144577026,
      "learning_rate": 5.554285714285714e-06,
      "loss": 0.8847,
      "step": 167
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 1.1454274654388428,
      "learning_rate": 5.550476190476191e-06,
      "loss": 0.8416,
      "step": 168
    },
    {
      "epoch": 0.10406403940886699,
      "grad_norm": 1.121837854385376,
      "learning_rate": 5.546666666666666e-06,
      "loss": 0.8951,
      "step": 169
    },
    {
      "epoch": 0.10467980295566502,
      "grad_norm": 1.260864496231079,
      "learning_rate": 5.542857142857144e-06,
      "loss": 0.8963,
      "step": 170
    },
    {
      "epoch": 0.10529556650246305,
      "grad_norm": 1.0481185913085938,
      "learning_rate": 5.539047619047619e-06,
      "loss": 0.888,
      "step": 171
    },
    {
      "epoch": 0.10591133004926108,
      "grad_norm": 1.2206023931503296,
      "learning_rate": 5.535238095238096e-06,
      "loss": 0.8864,
      "step": 172
    },
    {
      "epoch": 0.10652709359605911,
      "grad_norm": 1.2968778610229492,
      "learning_rate": 5.531428571428571e-06,
      "loss": 0.8416,
      "step": 173
    },
    {
      "epoch": 0.10714285714285714,
      "grad_norm": 1.0612694025039673,
      "learning_rate": 5.527619047619048e-06,
      "loss": 0.8809,
      "step": 174
    },
    {
      "epoch": 0.10775862068965517,
      "grad_norm": 1.1105479001998901,
      "learning_rate": 5.523809523809523e-06,
      "loss": 0.8979,
      "step": 175
    },
    {
      "epoch": 0.10837438423645321,
      "grad_norm": 1.1283435821533203,
      "learning_rate": 5.5200000000000005e-06,
      "loss": 0.852,
      "step": 176
    },
    {
      "epoch": 0.10899014778325124,
      "grad_norm": 1.1292593479156494,
      "learning_rate": 5.516190476190476e-06,
      "loss": 0.8788,
      "step": 177
    },
    {
      "epoch": 0.10960591133004927,
      "grad_norm": 1.0862294435501099,
      "learning_rate": 5.5123809523809525e-06,
      "loss": 0.8505,
      "step": 178
    },
    {
      "epoch": 0.1102216748768473,
      "grad_norm": 1.1376081705093384,
      "learning_rate": 5.508571428571429e-06,
      "loss": 0.8464,
      "step": 179
    },
    {
      "epoch": 0.11083743842364532,
      "grad_norm": 1.4304242134094238,
      "learning_rate": 5.5047619047619045e-06,
      "loss": 0.8992,
      "step": 180
    },
    {
      "epoch": 0.11145320197044335,
      "grad_norm": 1.1022571325302124,
      "learning_rate": 5.500952380952382e-06,
      "loss": 0.8681,
      "step": 181
    },
    {
      "epoch": 0.11206896551724138,
      "grad_norm": 1.1266021728515625,
      "learning_rate": 5.497142857142857e-06,
      "loss": 0.8737,
      "step": 182
    },
    {
      "epoch": 0.11268472906403941,
      "grad_norm": 1.077064871788025,
      "learning_rate": 5.493333333333334e-06,
      "loss": 0.8578,
      "step": 183
    },
    {
      "epoch": 0.11330049261083744,
      "grad_norm": 1.230276346206665,
      "learning_rate": 5.489523809523809e-06,
      "loss": 0.8659,
      "step": 184
    },
    {
      "epoch": 0.11391625615763547,
      "grad_norm": 1.0161148309707642,
      "learning_rate": 5.485714285714286e-06,
      "loss": 0.8574,
      "step": 185
    },
    {
      "epoch": 0.1145320197044335,
      "grad_norm": 1.2032387256622314,
      "learning_rate": 5.481904761904761e-06,
      "loss": 0.8392,
      "step": 186
    },
    {
      "epoch": 0.11514778325123153,
      "grad_norm": 1.3450112342834473,
      "learning_rate": 5.478095238095239e-06,
      "loss": 0.8642,
      "step": 187
    },
    {
      "epoch": 0.11576354679802955,
      "grad_norm": 1.5081307888031006,
      "learning_rate": 5.474285714285714e-06,
      "loss": 0.8645,
      "step": 188
    },
    {
      "epoch": 0.11637931034482758,
      "grad_norm": 1.0227556228637695,
      "learning_rate": 5.470476190476191e-06,
      "loss": 0.8364,
      "step": 189
    },
    {
      "epoch": 0.11699507389162561,
      "grad_norm": 1.118567705154419,
      "learning_rate": 5.466666666666666e-06,
      "loss": 0.8307,
      "step": 190
    },
    {
      "epoch": 0.11761083743842364,
      "grad_norm": 1.198087215423584,
      "learning_rate": 5.462857142857143e-06,
      "loss": 0.8788,
      "step": 191
    },
    {
      "epoch": 0.11822660098522167,
      "grad_norm": 1.1396293640136719,
      "learning_rate": 5.459047619047619e-06,
      "loss": 0.8403,
      "step": 192
    },
    {
      "epoch": 0.1188423645320197,
      "grad_norm": 1.2122846841812134,
      "learning_rate": 5.4552380952380956e-06,
      "loss": 0.8198,
      "step": 193
    },
    {
      "epoch": 0.11945812807881774,
      "grad_norm": 1.1449072360992432,
      "learning_rate": 5.451428571428572e-06,
      "loss": 0.8831,
      "step": 194
    },
    {
      "epoch": 0.12007389162561577,
      "grad_norm": 1.2797808647155762,
      "learning_rate": 5.4476190476190476e-06,
      "loss": 0.9013,
      "step": 195
    },
    {
      "epoch": 0.1206896551724138,
      "grad_norm": 1.2525681257247925,
      "learning_rate": 5.443809523809524e-06,
      "loss": 0.8879,
      "step": 196
    },
    {
      "epoch": 0.12130541871921183,
      "grad_norm": 1.146490216255188,
      "learning_rate": 5.44e-06,
      "loss": 0.892,
      "step": 197
    },
    {
      "epoch": 0.12192118226600986,
      "grad_norm": 1.4946634769439697,
      "learning_rate": 5.436190476190477e-06,
      "loss": 0.9361,
      "step": 198
    },
    {
      "epoch": 0.12253694581280788,
      "grad_norm": 1.083890438079834,
      "learning_rate": 5.4323809523809524e-06,
      "loss": 0.8512,
      "step": 199
    },
    {
      "epoch": 0.12315270935960591,
      "grad_norm": 1.224650263786316,
      "learning_rate": 5.428571428571429e-06,
      "loss": 0.8581,
      "step": 200
    },
    {
      "epoch": 0.12376847290640394,
      "grad_norm": 1.1192318201065063,
      "learning_rate": 5.4247619047619045e-06,
      "loss": 0.8409,
      "step": 201
    },
    {
      "epoch": 0.12438423645320197,
      "grad_norm": 1.1843795776367188,
      "learning_rate": 5.420952380952381e-06,
      "loss": 0.8388,
      "step": 202
    },
    {
      "epoch": 0.125,
      "grad_norm": 1.2829216718673706,
      "learning_rate": 5.417142857142857e-06,
      "loss": 0.838,
      "step": 203
    },
    {
      "epoch": 0.12561576354679804,
      "grad_norm": 1.1019855737686157,
      "learning_rate": 5.413333333333334e-06,
      "loss": 0.8448,
      "step": 204
    },
    {
      "epoch": 0.12623152709359606,
      "grad_norm": 1.21596360206604,
      "learning_rate": 5.40952380952381e-06,
      "loss": 0.8409,
      "step": 205
    },
    {
      "epoch": 0.1268472906403941,
      "grad_norm": 1.1130937337875366,
      "learning_rate": 5.405714285714286e-06,
      "loss": 0.8277,
      "step": 206
    },
    {
      "epoch": 0.12746305418719212,
      "grad_norm": 1.0802650451660156,
      "learning_rate": 5.401904761904762e-06,
      "loss": 0.856,
      "step": 207
    },
    {
      "epoch": 0.12807881773399016,
      "grad_norm": 1.130344271659851,
      "learning_rate": 5.398095238095238e-06,
      "loss": 0.8458,
      "step": 208
    },
    {
      "epoch": 0.12869458128078817,
      "grad_norm": 1.2739241123199463,
      "learning_rate": 5.394285714285715e-06,
      "loss": 0.9047,
      "step": 209
    },
    {
      "epoch": 0.12931034482758622,
      "grad_norm": 1.266101360321045,
      "learning_rate": 5.390476190476191e-06,
      "loss": 0.8574,
      "step": 210
    },
    {
      "epoch": 0.12992610837438423,
      "grad_norm": 1.1921247243881226,
      "learning_rate": 5.386666666666667e-06,
      "loss": 0.8687,
      "step": 211
    },
    {
      "epoch": 0.13054187192118227,
      "grad_norm": 1.1276596784591675,
      "learning_rate": 5.382857142857143e-06,
      "loss": 0.8844,
      "step": 212
    },
    {
      "epoch": 0.1311576354679803,
      "grad_norm": 1.0927220582962036,
      "learning_rate": 5.379047619047619e-06,
      "loss": 0.8276,
      "step": 213
    },
    {
      "epoch": 0.13177339901477833,
      "grad_norm": 1.1888712644577026,
      "learning_rate": 5.3752380952380955e-06,
      "loss": 0.8712,
      "step": 214
    },
    {
      "epoch": 0.13238916256157635,
      "grad_norm": 1.2157961130142212,
      "learning_rate": 5.371428571428572e-06,
      "loss": 0.8787,
      "step": 215
    },
    {
      "epoch": 0.1330049261083744,
      "grad_norm": 1.0661420822143555,
      "learning_rate": 5.3676190476190475e-06,
      "loss": 0.8965,
      "step": 216
    },
    {
      "epoch": 0.1336206896551724,
      "grad_norm": 1.649373173713684,
      "learning_rate": 5.363809523809524e-06,
      "loss": 0.8593,
      "step": 217
    },
    {
      "epoch": 0.13423645320197045,
      "grad_norm": 1.1910051107406616,
      "learning_rate": 5.36e-06,
      "loss": 0.8693,
      "step": 218
    },
    {
      "epoch": 0.13485221674876846,
      "grad_norm": 1.099755883216858,
      "learning_rate": 5.356190476190476e-06,
      "loss": 0.8876,
      "step": 219
    },
    {
      "epoch": 0.1354679802955665,
      "grad_norm": 1.3697484731674194,
      "learning_rate": 5.352380952380953e-06,
      "loss": 0.8821,
      "step": 220
    },
    {
      "epoch": 0.13608374384236452,
      "grad_norm": 1.0995042324066162,
      "learning_rate": 5.348571428571429e-06,
      "loss": 0.917,
      "step": 221
    },
    {
      "epoch": 0.13669950738916256,
      "grad_norm": 1.1404619216918945,
      "learning_rate": 5.344761904761905e-06,
      "loss": 0.8981,
      "step": 222
    },
    {
      "epoch": 0.1373152709359606,
      "grad_norm": 2.2940824031829834,
      "learning_rate": 5.340952380952381e-06,
      "loss": 0.8402,
      "step": 223
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 5.520970344543457,
      "learning_rate": 5.337142857142857e-06,
      "loss": 0.9038,
      "step": 224
    },
    {
      "epoch": 0.13854679802955666,
      "grad_norm": 2.44114351272583,
      "learning_rate": 5.333333333333333e-06,
      "loss": 0.8491,
      "step": 225
    },
    {
      "epoch": 0.13916256157635468,
      "grad_norm": 4.566291332244873,
      "learning_rate": 5.32952380952381e-06,
      "loss": 0.8658,
      "step": 226
    },
    {
      "epoch": 0.13977832512315272,
      "grad_norm": 1.0625686645507812,
      "learning_rate": 5.325714285714286e-06,
      "loss": 0.8653,
      "step": 227
    },
    {
      "epoch": 0.14039408866995073,
      "grad_norm": 3.374521255493164,
      "learning_rate": 5.321904761904762e-06,
      "loss": 0.8871,
      "step": 228
    },
    {
      "epoch": 0.14100985221674878,
      "grad_norm": 29.291587829589844,
      "learning_rate": 5.3180952380952385e-06,
      "loss": 0.8782,
      "step": 229
    },
    {
      "epoch": 0.1416256157635468,
      "grad_norm": 4.579706192016602,
      "learning_rate": 5.314285714285714e-06,
      "loss": 0.8575,
      "step": 230
    },
    {
      "epoch": 0.14224137931034483,
      "grad_norm": 5.344178199768066,
      "learning_rate": 5.3104761904761905e-06,
      "loss": 0.8698,
      "step": 231
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 10.386034965515137,
      "learning_rate": 5.306666666666667e-06,
      "loss": 0.8234,
      "step": 232
    },
    {
      "epoch": 0.1434729064039409,
      "grad_norm": 17.308822631835938,
      "learning_rate": 5.302857142857143e-06,
      "loss": 0.9205,
      "step": 233
    },
    {
      "epoch": 0.1440886699507389,
      "grad_norm": 10.811816215515137,
      "learning_rate": 5.299047619047619e-06,
      "loss": 0.9053,
      "step": 234
    },
    {
      "epoch": 0.14470443349753695,
      "grad_norm": 4.283424377441406,
      "learning_rate": 5.295238095238095e-06,
      "loss": 0.8762,
      "step": 235
    },
    {
      "epoch": 0.14532019704433496,
      "grad_norm": 2.517817497253418,
      "learning_rate": 5.291428571428571e-06,
      "loss": 0.897,
      "step": 236
    },
    {
      "epoch": 0.145935960591133,
      "grad_norm": 2.168529987335205,
      "learning_rate": 5.287619047619048e-06,
      "loss": 0.8953,
      "step": 237
    },
    {
      "epoch": 0.14655172413793102,
      "grad_norm": 5.304393768310547,
      "learning_rate": 5.283809523809524e-06,
      "loss": 0.8599,
      "step": 238
    },
    {
      "epoch": 0.14716748768472906,
      "grad_norm": 5.690064907073975,
      "learning_rate": 5.28e-06,
      "loss": 0.8319,
      "step": 239
    },
    {
      "epoch": 0.1477832512315271,
      "grad_norm": 8.950540542602539,
      "learning_rate": 5.276190476190476e-06,
      "loss": 0.8799,
      "step": 240
    },
    {
      "epoch": 0.14839901477832512,
      "grad_norm": 7.783779144287109,
      "learning_rate": 5.272380952380952e-06,
      "loss": 0.8771,
      "step": 241
    },
    {
      "epoch": 0.14901477832512317,
      "grad_norm": 8.56550121307373,
      "learning_rate": 5.268571428571429e-06,
      "loss": 0.8868,
      "step": 242
    },
    {
      "epoch": 0.14963054187192118,
      "grad_norm": 28.451763153076172,
      "learning_rate": 5.264761904761905e-06,
      "loss": 0.823,
      "step": 243
    },
    {
      "epoch": 0.15024630541871922,
      "grad_norm": 6.663080215454102,
      "learning_rate": 5.2609523809523816e-06,
      "loss": 0.8723,
      "step": 244
    },
    {
      "epoch": 0.15086206896551724,
      "grad_norm": 1.3278969526290894,
      "learning_rate": 5.257142857142857e-06,
      "loss": 0.881,
      "step": 245
    },
    {
      "epoch": 0.15147783251231528,
      "grad_norm": 3.209476947784424,
      "learning_rate": 5.2533333333333336e-06,
      "loss": 0.8492,
      "step": 246
    },
    {
      "epoch": 0.1520935960591133,
      "grad_norm": 1.4075722694396973,
      "learning_rate": 5.249523809523809e-06,
      "loss": 0.8581,
      "step": 247
    },
    {
      "epoch": 0.15270935960591134,
      "grad_norm": 1.322488784790039,
      "learning_rate": 5.2457142857142864e-06,
      "loss": 0.8317,
      "step": 248
    },
    {
      "epoch": 0.15332512315270935,
      "grad_norm": 1.4066427946090698,
      "learning_rate": 5.241904761904762e-06,
      "loss": 0.8856,
      "step": 249
    },
    {
      "epoch": 0.1539408866995074,
      "grad_norm": 2.7206406593322754,
      "learning_rate": 5.2380952380952384e-06,
      "loss": 0.8898,
      "step": 250
    },
    {
      "epoch": 0.1545566502463054,
      "grad_norm": 2.6765048503875732,
      "learning_rate": 5.234285714285714e-06,
      "loss": 0.834,
      "step": 251
    },
    {
      "epoch": 0.15517241379310345,
      "grad_norm": 1.6989318132400513,
      "learning_rate": 5.2304761904761905e-06,
      "loss": 0.8868,
      "step": 252
    },
    {
      "epoch": 0.15578817733990147,
      "grad_norm": 1.2456961870193481,
      "learning_rate": 5.226666666666667e-06,
      "loss": 0.8616,
      "step": 253
    },
    {
      "epoch": 0.1564039408866995,
      "grad_norm": 1.1997747421264648,
      "learning_rate": 5.222857142857143e-06,
      "loss": 0.8864,
      "step": 254
    },
    {
      "epoch": 0.15701970443349753,
      "grad_norm": 1.2236082553863525,
      "learning_rate": 5.219047619047619e-06,
      "loss": 0.835,
      "step": 255
    },
    {
      "epoch": 0.15763546798029557,
      "grad_norm": 1.2350659370422363,
      "learning_rate": 5.215238095238095e-06,
      "loss": 0.9246,
      "step": 256
    },
    {
      "epoch": 0.15825123152709358,
      "grad_norm": 1.8595610857009888,
      "learning_rate": 5.211428571428572e-06,
      "loss": 0.864,
      "step": 257
    },
    {
      "epoch": 0.15886699507389163,
      "grad_norm": 1.1133977174758911,
      "learning_rate": 5.207619047619047e-06,
      "loss": 0.8953,
      "step": 258
    },
    {
      "epoch": 0.15948275862068967,
      "grad_norm": 1.2005869150161743,
      "learning_rate": 5.203809523809525e-06,
      "loss": 0.8619,
      "step": 259
    },
    {
      "epoch": 0.16009852216748768,
      "grad_norm": 1.1968201398849487,
      "learning_rate": 5.2e-06,
      "loss": 0.8765,
      "step": 260
    },
    {
      "epoch": 0.16071428571428573,
      "grad_norm": 1.1928001642227173,
      "learning_rate": 5.196190476190477e-06,
      "loss": 0.8334,
      "step": 261
    },
    {
      "epoch": 0.16133004926108374,
      "grad_norm": 1.1945537328720093,
      "learning_rate": 5.192380952380952e-06,
      "loss": 0.8721,
      "step": 262
    },
    {
      "epoch": 0.16194581280788178,
      "grad_norm": 1.184354305267334,
      "learning_rate": 5.188571428571429e-06,
      "loss": 0.8926,
      "step": 263
    },
    {
      "epoch": 0.1625615763546798,
      "grad_norm": 1.1372910737991333,
      "learning_rate": 5.184761904761904e-06,
      "loss": 0.8689,
      "step": 264
    },
    {
      "epoch": 0.16317733990147784,
      "grad_norm": 1.2706125974655151,
      "learning_rate": 5.1809523809523815e-06,
      "loss": 0.8347,
      "step": 265
    },
    {
      "epoch": 0.16379310344827586,
      "grad_norm": 1.2233738899230957,
      "learning_rate": 5.177142857142857e-06,
      "loss": 0.8834,
      "step": 266
    },
    {
      "epoch": 0.1644088669950739,
      "grad_norm": 1.0843745470046997,
      "learning_rate": 5.1733333333333335e-06,
      "loss": 0.866,
      "step": 267
    },
    {
      "epoch": 0.16502463054187191,
      "grad_norm": 2.2695066928863525,
      "learning_rate": 5.16952380952381e-06,
      "loss": 0.8964,
      "step": 268
    },
    {
      "epoch": 0.16564039408866996,
      "grad_norm": 1.2098004817962646,
      "learning_rate": 5.1657142857142855e-06,
      "loss": 0.8809,
      "step": 269
    },
    {
      "epoch": 0.16625615763546797,
      "grad_norm": 1.175437331199646,
      "learning_rate": 5.161904761904762e-06,
      "loss": 0.8858,
      "step": 270
    },
    {
      "epoch": 0.16687192118226601,
      "grad_norm": 1.841796636581421,
      "learning_rate": 5.158095238095238e-06,
      "loss": 0.8427,
      "step": 271
    },
    {
      "epoch": 0.16748768472906403,
      "grad_norm": 1.0769678354263306,
      "learning_rate": 5.154285714285715e-06,
      "loss": 0.9028,
      "step": 272
    },
    {
      "epoch": 0.16810344827586207,
      "grad_norm": 1.0605462789535522,
      "learning_rate": 5.15047619047619e-06,
      "loss": 0.8595,
      "step": 273
    },
    {
      "epoch": 0.1687192118226601,
      "grad_norm": 1.2110742330551147,
      "learning_rate": 5.146666666666667e-06,
      "loss": 0.8808,
      "step": 274
    },
    {
      "epoch": 0.16933497536945813,
      "grad_norm": 1.52138352394104,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.8781,
      "step": 275
    },
    {
      "epoch": 0.16995073891625614,
      "grad_norm": 1.3455487489700317,
      "learning_rate": 5.13904761904762e-06,
      "loss": 0.854,
      "step": 276
    },
    {
      "epoch": 0.1705665024630542,
      "grad_norm": 1.151937484741211,
      "learning_rate": 5.135238095238095e-06,
      "loss": 0.9077,
      "step": 277
    },
    {
      "epoch": 0.17118226600985223,
      "grad_norm": 1.8301992416381836,
      "learning_rate": 5.131428571428572e-06,
      "loss": 0.8792,
      "step": 278
    },
    {
      "epoch": 0.17179802955665024,
      "grad_norm": 1.3835124969482422,
      "learning_rate": 5.127619047619047e-06,
      "loss": 0.8952,
      "step": 279
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 2.209676504135132,
      "learning_rate": 5.123809523809524e-06,
      "loss": 0.8415,
      "step": 280
    },
    {
      "epoch": 0.1730295566502463,
      "grad_norm": 1.7914087772369385,
      "learning_rate": 5.12e-06,
      "loss": 0.8633,
      "step": 281
    },
    {
      "epoch": 0.17364532019704434,
      "grad_norm": 1.3271392583847046,
      "learning_rate": 5.1161904761904765e-06,
      "loss": 0.8459,
      "step": 282
    },
    {
      "epoch": 0.17426108374384236,
      "grad_norm": 1.4130874872207642,
      "learning_rate": 5.112380952380953e-06,
      "loss": 0.9145,
      "step": 283
    },
    {
      "epoch": 0.1748768472906404,
      "grad_norm": 1.2915335893630981,
      "learning_rate": 5.1085714285714285e-06,
      "loss": 0.8475,
      "step": 284
    },
    {
      "epoch": 0.17549261083743842,
      "grad_norm": 1.1510486602783203,
      "learning_rate": 5.104761904761905e-06,
      "loss": 0.8763,
      "step": 285
    },
    {
      "epoch": 0.17610837438423646,
      "grad_norm": 1.4610087871551514,
      "learning_rate": 5.1009523809523806e-06,
      "loss": 0.8644,
      "step": 286
    },
    {
      "epoch": 0.17672413793103448,
      "grad_norm": 1.1213489770889282,
      "learning_rate": 5.097142857142858e-06,
      "loss": 0.9181,
      "step": 287
    },
    {
      "epoch": 0.17733990147783252,
      "grad_norm": 1.3746322393417358,
      "learning_rate": 5.093333333333333e-06,
      "loss": 0.8693,
      "step": 288
    },
    {
      "epoch": 0.17795566502463053,
      "grad_norm": 1.1176527738571167,
      "learning_rate": 5.08952380952381e-06,
      "loss": 0.8331,
      "step": 289
    },
    {
      "epoch": 0.17857142857142858,
      "grad_norm": 1.1112719774246216,
      "learning_rate": 5.0857142857142854e-06,
      "loss": 0.8478,
      "step": 290
    },
    {
      "epoch": 0.1791871921182266,
      "grad_norm": 1.0972497463226318,
      "learning_rate": 5.081904761904762e-06,
      "loss": 0.8772,
      "step": 291
    },
    {
      "epoch": 0.17980295566502463,
      "grad_norm": 1.2696088552474976,
      "learning_rate": 5.078095238095238e-06,
      "loss": 0.867,
      "step": 292
    },
    {
      "epoch": 0.18041871921182265,
      "grad_norm": 1.1089624166488647,
      "learning_rate": 5.074285714285715e-06,
      "loss": 0.8631,
      "step": 293
    },
    {
      "epoch": 0.1810344827586207,
      "grad_norm": 1.2776823043823242,
      "learning_rate": 5.07047619047619e-06,
      "loss": 0.9205,
      "step": 294
    },
    {
      "epoch": 0.1816502463054187,
      "grad_norm": 1.2953342199325562,
      "learning_rate": 5.066666666666667e-06,
      "loss": 0.8545,
      "step": 295
    },
    {
      "epoch": 0.18226600985221675,
      "grad_norm": 1.180152177810669,
      "learning_rate": 5.062857142857143e-06,
      "loss": 0.9203,
      "step": 296
    },
    {
      "epoch": 0.1828817733990148,
      "grad_norm": 1.0510919094085693,
      "learning_rate": 5.059047619047619e-06,
      "loss": 0.8908,
      "step": 297
    },
    {
      "epoch": 0.1834975369458128,
      "grad_norm": 1.2766355276107788,
      "learning_rate": 5.055238095238096e-06,
      "loss": 0.8572,
      "step": 298
    },
    {
      "epoch": 0.18411330049261085,
      "grad_norm": 1.0861619710922241,
      "learning_rate": 5.051428571428572e-06,
      "loss": 0.8836,
      "step": 299
    },
    {
      "epoch": 0.18472906403940886,
      "grad_norm": 1.0663539171218872,
      "learning_rate": 5.047619047619048e-06,
      "loss": 0.8873,
      "step": 300
    },
    {
      "epoch": 0.1853448275862069,
      "grad_norm": 1.032915472984314,
      "learning_rate": 5.043809523809524e-06,
      "loss": 0.8464,
      "step": 301
    },
    {
      "epoch": 0.18596059113300492,
      "grad_norm": 1.1256338357925415,
      "learning_rate": 5.04e-06,
      "loss": 0.8657,
      "step": 302
    },
    {
      "epoch": 0.18657635467980296,
      "grad_norm": 1.1169484853744507,
      "learning_rate": 5.036190476190476e-06,
      "loss": 0.909,
      "step": 303
    },
    {
      "epoch": 0.18719211822660098,
      "grad_norm": 1.2397226095199585,
      "learning_rate": 5.032380952380953e-06,
      "loss": 0.8477,
      "step": 304
    },
    {
      "epoch": 0.18780788177339902,
      "grad_norm": 1.1459541320800781,
      "learning_rate": 5.0285714285714285e-06,
      "loss": 0.8565,
      "step": 305
    },
    {
      "epoch": 0.18842364532019704,
      "grad_norm": 1.2260910272598267,
      "learning_rate": 5.024761904761905e-06,
      "loss": 0.8472,
      "step": 306
    },
    {
      "epoch": 0.18903940886699508,
      "grad_norm": 1.3628312349319458,
      "learning_rate": 5.020952380952381e-06,
      "loss": 0.8839,
      "step": 307
    },
    {
      "epoch": 0.1896551724137931,
      "grad_norm": 1.1895159482955933,
      "learning_rate": 5.017142857142857e-06,
      "loss": 0.8628,
      "step": 308
    },
    {
      "epoch": 0.19027093596059114,
      "grad_norm": 1.223396897315979,
      "learning_rate": 5.013333333333334e-06,
      "loss": 0.8532,
      "step": 309
    },
    {
      "epoch": 0.19088669950738915,
      "grad_norm": 1.1213815212249756,
      "learning_rate": 5.00952380952381e-06,
      "loss": 0.8477,
      "step": 310
    },
    {
      "epoch": 0.1915024630541872,
      "grad_norm": 1.0548524856567383,
      "learning_rate": 5.005714285714286e-06,
      "loss": 0.8482,
      "step": 311
    },
    {
      "epoch": 0.1921182266009852,
      "grad_norm": 1.202810525894165,
      "learning_rate": 5.001904761904762e-06,
      "loss": 0.8272,
      "step": 312
    },
    {
      "epoch": 0.19273399014778325,
      "grad_norm": 1.1839394569396973,
      "learning_rate": 4.998095238095238e-06,
      "loss": 0.902,
      "step": 313
    },
    {
      "epoch": 0.1933497536945813,
      "grad_norm": 1.115837812423706,
      "learning_rate": 4.994285714285714e-06,
      "loss": 0.8511,
      "step": 314
    },
    {
      "epoch": 0.1939655172413793,
      "grad_norm": 1.1370697021484375,
      "learning_rate": 4.990476190476191e-06,
      "loss": 0.9024,
      "step": 315
    },
    {
      "epoch": 0.19458128078817735,
      "grad_norm": 1.059780240058899,
      "learning_rate": 4.986666666666667e-06,
      "loss": 0.8581,
      "step": 316
    },
    {
      "epoch": 0.19519704433497537,
      "grad_norm": 1.181428074836731,
      "learning_rate": 4.982857142857143e-06,
      "loss": 0.8811,
      "step": 317
    },
    {
      "epoch": 0.1958128078817734,
      "grad_norm": 1.2130671739578247,
      "learning_rate": 4.979047619047619e-06,
      "loss": 0.8802,
      "step": 318
    },
    {
      "epoch": 0.19642857142857142,
      "grad_norm": 1.1750119924545288,
      "learning_rate": 4.975238095238095e-06,
      "loss": 0.87,
      "step": 319
    },
    {
      "epoch": 0.19704433497536947,
      "grad_norm": 1.1327788829803467,
      "learning_rate": 4.9714285714285715e-06,
      "loss": 0.8805,
      "step": 320
    },
    {
      "epoch": 0.19766009852216748,
      "grad_norm": 1.0986442565917969,
      "learning_rate": 4.967619047619048e-06,
      "loss": 0.8515,
      "step": 321
    },
    {
      "epoch": 0.19827586206896552,
      "grad_norm": 1.4404698610305786,
      "learning_rate": 4.963809523809524e-06,
      "loss": 0.8894,
      "step": 322
    },
    {
      "epoch": 0.19889162561576354,
      "grad_norm": 1.2088637351989746,
      "learning_rate": 4.96e-06,
      "loss": 0.8522,
      "step": 323
    },
    {
      "epoch": 0.19950738916256158,
      "grad_norm": 1.0706119537353516,
      "learning_rate": 4.956190476190476e-06,
      "loss": 0.8594,
      "step": 324
    },
    {
      "epoch": 0.2001231527093596,
      "grad_norm": 1.0410555601119995,
      "learning_rate": 4.952380952380952e-06,
      "loss": 0.8621,
      "step": 325
    },
    {
      "epoch": 0.20073891625615764,
      "grad_norm": 1.4728288650512695,
      "learning_rate": 4.948571428571429e-06,
      "loss": 0.8355,
      "step": 326
    },
    {
      "epoch": 0.20135467980295566,
      "grad_norm": 1.2625833749771118,
      "learning_rate": 4.944761904761905e-06,
      "loss": 0.8892,
      "step": 327
    },
    {
      "epoch": 0.2019704433497537,
      "grad_norm": 1.085358738899231,
      "learning_rate": 4.940952380952381e-06,
      "loss": 0.8248,
      "step": 328
    },
    {
      "epoch": 0.2025862068965517,
      "grad_norm": 1.1912381649017334,
      "learning_rate": 4.937142857142857e-06,
      "loss": 0.8516,
      "step": 329
    },
    {
      "epoch": 0.20320197044334976,
      "grad_norm": 1.1776145696640015,
      "learning_rate": 4.933333333333333e-06,
      "loss": 0.8845,
      "step": 330
    },
    {
      "epoch": 0.20381773399014777,
      "grad_norm": 1.0765786170959473,
      "learning_rate": 4.92952380952381e-06,
      "loss": 0.8335,
      "step": 331
    },
    {
      "epoch": 0.2044334975369458,
      "grad_norm": 1.11163330078125,
      "learning_rate": 4.925714285714286e-06,
      "loss": 0.8586,
      "step": 332
    },
    {
      "epoch": 0.20504926108374386,
      "grad_norm": 1.4313175678253174,
      "learning_rate": 4.9219047619047625e-06,
      "loss": 0.8345,
      "step": 333
    },
    {
      "epoch": 0.20566502463054187,
      "grad_norm": 1.1111764907836914,
      "learning_rate": 4.918095238095238e-06,
      "loss": 0.8595,
      "step": 334
    },
    {
      "epoch": 0.2062807881773399,
      "grad_norm": 1.0591881275177002,
      "learning_rate": 4.9142857142857145e-06,
      "loss": 0.8522,
      "step": 335
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 1.126545786857605,
      "learning_rate": 4.91047619047619e-06,
      "loss": 0.8121,
      "step": 336
    },
    {
      "epoch": 0.20751231527093597,
      "grad_norm": 1.0603841543197632,
      "learning_rate": 4.906666666666667e-06,
      "loss": 0.8694,
      "step": 337
    },
    {
      "epoch": 0.20812807881773399,
      "grad_norm": 1.0510146617889404,
      "learning_rate": 4.902857142857143e-06,
      "loss": 0.8462,
      "step": 338
    },
    {
      "epoch": 0.20874384236453203,
      "grad_norm": 1.223373293876648,
      "learning_rate": 4.899047619047619e-06,
      "loss": 0.8619,
      "step": 339
    },
    {
      "epoch": 0.20935960591133004,
      "grad_norm": 1.1268737316131592,
      "learning_rate": 4.895238095238095e-06,
      "loss": 0.8688,
      "step": 340
    },
    {
      "epoch": 0.20997536945812809,
      "grad_norm": 1.1189382076263428,
      "learning_rate": 4.8914285714285714e-06,
      "loss": 0.8511,
      "step": 341
    },
    {
      "epoch": 0.2105911330049261,
      "grad_norm": 1.0459853410720825,
      "learning_rate": 4.887619047619048e-06,
      "loss": 0.8659,
      "step": 342
    },
    {
      "epoch": 0.21120689655172414,
      "grad_norm": 1.0760074853897095,
      "learning_rate": 4.883809523809524e-06,
      "loss": 0.9042,
      "step": 343
    },
    {
      "epoch": 0.21182266009852216,
      "grad_norm": 1.1302834749221802,
      "learning_rate": 4.88e-06,
      "loss": 0.8339,
      "step": 344
    },
    {
      "epoch": 0.2124384236453202,
      "grad_norm": 1.0517975091934204,
      "learning_rate": 4.876190476190476e-06,
      "loss": 0.8584,
      "step": 345
    },
    {
      "epoch": 0.21305418719211822,
      "grad_norm": 1.0268917083740234,
      "learning_rate": 4.872380952380953e-06,
      "loss": 0.8914,
      "step": 346
    },
    {
      "epoch": 0.21366995073891626,
      "grad_norm": 1.1653934717178345,
      "learning_rate": 4.868571428571428e-06,
      "loss": 0.8169,
      "step": 347
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 1.0825992822647095,
      "learning_rate": 4.864761904761906e-06,
      "loss": 0.8564,
      "step": 348
    },
    {
      "epoch": 0.21490147783251232,
      "grad_norm": 1.1065682172775269,
      "learning_rate": 4.860952380952381e-06,
      "loss": 0.8692,
      "step": 349
    },
    {
      "epoch": 0.21551724137931033,
      "grad_norm": 1.1529004573822021,
      "learning_rate": 4.857142857142858e-06,
      "loss": 0.8622,
      "step": 350
    },
    {
      "epoch": 0.21613300492610837,
      "grad_norm": 1.086659550666809,
      "learning_rate": 4.853333333333333e-06,
      "loss": 0.8393,
      "step": 351
    },
    {
      "epoch": 0.21674876847290642,
      "grad_norm": 1.0905269384384155,
      "learning_rate": 4.84952380952381e-06,
      "loss": 0.8647,
      "step": 352
    },
    {
      "epoch": 0.21736453201970443,
      "grad_norm": 1.0725101232528687,
      "learning_rate": 4.845714285714285e-06,
      "loss": 0.8674,
      "step": 353
    },
    {
      "epoch": 0.21798029556650247,
      "grad_norm": 1.3235262632369995,
      "learning_rate": 4.8419047619047625e-06,
      "loss": 0.8802,
      "step": 354
    },
    {
      "epoch": 0.2185960591133005,
      "grad_norm": 1.2218725681304932,
      "learning_rate": 4.838095238095238e-06,
      "loss": 0.8583,
      "step": 355
    },
    {
      "epoch": 0.21921182266009853,
      "grad_norm": 1.113556146621704,
      "learning_rate": 4.8342857142857145e-06,
      "loss": 0.8379,
      "step": 356
    },
    {
      "epoch": 0.21982758620689655,
      "grad_norm": 1.1689013242721558,
      "learning_rate": 4.830476190476191e-06,
      "loss": 0.8752,
      "step": 357
    },
    {
      "epoch": 0.2204433497536946,
      "grad_norm": 1.145027756690979,
      "learning_rate": 4.8266666666666665e-06,
      "loss": 0.8956,
      "step": 358
    },
    {
      "epoch": 0.2210591133004926,
      "grad_norm": 1.199471354484558,
      "learning_rate": 4.822857142857143e-06,
      "loss": 0.922,
      "step": 359
    },
    {
      "epoch": 0.22167487684729065,
      "grad_norm": 1.102174162864685,
      "learning_rate": 4.819047619047619e-06,
      "loss": 0.8452,
      "step": 360
    },
    {
      "epoch": 0.22229064039408866,
      "grad_norm": 1.084350347518921,
      "learning_rate": 4.815238095238096e-06,
      "loss": 0.8735,
      "step": 361
    },
    {
      "epoch": 0.2229064039408867,
      "grad_norm": 0.9981065392494202,
      "learning_rate": 4.811428571428571e-06,
      "loss": 0.8896,
      "step": 362
    },
    {
      "epoch": 0.22352216748768472,
      "grad_norm": 1.131090521812439,
      "learning_rate": 4.807619047619048e-06,
      "loss": 0.8797,
      "step": 363
    },
    {
      "epoch": 0.22413793103448276,
      "grad_norm": 1.13848876953125,
      "learning_rate": 4.803809523809523e-06,
      "loss": 0.8579,
      "step": 364
    },
    {
      "epoch": 0.22475369458128078,
      "grad_norm": 1.1223580837249756,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.8224,
      "step": 365
    },
    {
      "epoch": 0.22536945812807882,
      "grad_norm": 1.2132917642593384,
      "learning_rate": 4.796190476190476e-06,
      "loss": 0.8746,
      "step": 366
    },
    {
      "epoch": 0.22598522167487683,
      "grad_norm": 1.1261727809906006,
      "learning_rate": 4.792380952380953e-06,
      "loss": 0.8936,
      "step": 367
    },
    {
      "epoch": 0.22660098522167488,
      "grad_norm": 1.255142092704773,
      "learning_rate": 4.788571428571428e-06,
      "loss": 0.8568,
      "step": 368
    },
    {
      "epoch": 0.2272167487684729,
      "grad_norm": 1.0664314031600952,
      "learning_rate": 4.784761904761905e-06,
      "loss": 0.8877,
      "step": 369
    },
    {
      "epoch": 0.22783251231527094,
      "grad_norm": 1.0640528202056885,
      "learning_rate": 4.780952380952381e-06,
      "loss": 0.859,
      "step": 370
    },
    {
      "epoch": 0.22844827586206898,
      "grad_norm": 1.1176449060440063,
      "learning_rate": 4.7771428571428575e-06,
      "loss": 0.8524,
      "step": 371
    },
    {
      "epoch": 0.229064039408867,
      "grad_norm": 1.240795612335205,
      "learning_rate": 4.773333333333334e-06,
      "loss": 0.8754,
      "step": 372
    },
    {
      "epoch": 0.22967980295566504,
      "grad_norm": 1.1183545589447021,
      "learning_rate": 4.7695238095238095e-06,
      "loss": 0.8783,
      "step": 373
    },
    {
      "epoch": 0.23029556650246305,
      "grad_norm": 1.0474815368652344,
      "learning_rate": 4.765714285714286e-06,
      "loss": 0.8467,
      "step": 374
    },
    {
      "epoch": 0.2309113300492611,
      "grad_norm": 1.180295467376709,
      "learning_rate": 4.7619047619047615e-06,
      "loss": 0.8462,
      "step": 375
    },
    {
      "epoch": 0.2315270935960591,
      "grad_norm": 1.0439234972000122,
      "learning_rate": 4.758095238095239e-06,
      "loss": 0.8559,
      "step": 376
    },
    {
      "epoch": 0.23214285714285715,
      "grad_norm": 1.0324128866195679,
      "learning_rate": 4.754285714285714e-06,
      "loss": 0.886,
      "step": 377
    },
    {
      "epoch": 0.23275862068965517,
      "grad_norm": 1.220220685005188,
      "learning_rate": 4.750476190476191e-06,
      "loss": 0.8715,
      "step": 378
    },
    {
      "epoch": 0.2333743842364532,
      "grad_norm": 1.0494054555892944,
      "learning_rate": 4.746666666666666e-06,
      "loss": 0.8732,
      "step": 379
    },
    {
      "epoch": 0.23399014778325122,
      "grad_norm": 0.994168758392334,
      "learning_rate": 4.742857142857143e-06,
      "loss": 0.8305,
      "step": 380
    },
    {
      "epoch": 0.23460591133004927,
      "grad_norm": 1.0505030155181885,
      "learning_rate": 4.739047619047619e-06,
      "loss": 0.8569,
      "step": 381
    },
    {
      "epoch": 0.23522167487684728,
      "grad_norm": 1.038307547569275,
      "learning_rate": 4.735238095238096e-06,
      "loss": 0.8896,
      "step": 382
    },
    {
      "epoch": 0.23583743842364532,
      "grad_norm": 1.1059892177581787,
      "learning_rate": 4.731428571428571e-06,
      "loss": 0.8789,
      "step": 383
    },
    {
      "epoch": 0.23645320197044334,
      "grad_norm": 1.1170305013656616,
      "learning_rate": 4.727619047619048e-06,
      "loss": 0.8689,
      "step": 384
    },
    {
      "epoch": 0.23706896551724138,
      "grad_norm": 1.1822565793991089,
      "learning_rate": 4.723809523809524e-06,
      "loss": 0.8799,
      "step": 385
    },
    {
      "epoch": 0.2376847290640394,
      "grad_norm": 1.178815245628357,
      "learning_rate": 4.72e-06,
      "loss": 0.8757,
      "step": 386
    },
    {
      "epoch": 0.23830049261083744,
      "grad_norm": 1.1188043355941772,
      "learning_rate": 4.716190476190477e-06,
      "loss": 0.8765,
      "step": 387
    },
    {
      "epoch": 0.23891625615763548,
      "grad_norm": 1.1529953479766846,
      "learning_rate": 4.7123809523809526e-06,
      "loss": 0.8669,
      "step": 388
    },
    {
      "epoch": 0.2395320197044335,
      "grad_norm": 1.0818513631820679,
      "learning_rate": 4.708571428571429e-06,
      "loss": 0.8639,
      "step": 389
    },
    {
      "epoch": 0.24014778325123154,
      "grad_norm": 1.0507328510284424,
      "learning_rate": 4.7047619047619046e-06,
      "loss": 0.8497,
      "step": 390
    },
    {
      "epoch": 0.24076354679802955,
      "grad_norm": 1.150769829750061,
      "learning_rate": 4.700952380952381e-06,
      "loss": 0.8377,
      "step": 391
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 1.1131374835968018,
      "learning_rate": 4.697142857142857e-06,
      "loss": 0.8321,
      "step": 392
    },
    {
      "epoch": 0.2419950738916256,
      "grad_norm": 1.0651566982269287,
      "learning_rate": 4.693333333333334e-06,
      "loss": 0.8884,
      "step": 393
    },
    {
      "epoch": 0.24261083743842365,
      "grad_norm": 1.302112102508545,
      "learning_rate": 4.6895238095238094e-06,
      "loss": 0.8449,
      "step": 394
    },
    {
      "epoch": 0.24322660098522167,
      "grad_norm": 1.0977972745895386,
      "learning_rate": 4.685714285714286e-06,
      "loss": 0.8316,
      "step": 395
    },
    {
      "epoch": 0.2438423645320197,
      "grad_norm": 1.113629698753357,
      "learning_rate": 4.681904761904762e-06,
      "loss": 0.8316,
      "step": 396
    },
    {
      "epoch": 0.24445812807881773,
      "grad_norm": 1.2946683168411255,
      "learning_rate": 4.678095238095238e-06,
      "loss": 0.8845,
      "step": 397
    },
    {
      "epoch": 0.24507389162561577,
      "grad_norm": 1.015247106552124,
      "learning_rate": 4.674285714285714e-06,
      "loss": 0.8576,
      "step": 398
    },
    {
      "epoch": 0.24568965517241378,
      "grad_norm": 1.1030054092407227,
      "learning_rate": 4.670476190476191e-06,
      "loss": 0.8979,
      "step": 399
    },
    {
      "epoch": 0.24630541871921183,
      "grad_norm": 1.0925713777542114,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.888,
      "step": 400
    },
    {
      "epoch": 0.24692118226600984,
      "grad_norm": 1.1895160675048828,
      "learning_rate": 4.662857142857143e-06,
      "loss": 0.8297,
      "step": 401
    },
    {
      "epoch": 0.24753694581280788,
      "grad_norm": 1.1218611001968384,
      "learning_rate": 4.659047619047619e-06,
      "loss": 0.8395,
      "step": 402
    },
    {
      "epoch": 0.2481527093596059,
      "grad_norm": 1.0012036561965942,
      "learning_rate": 4.655238095238095e-06,
      "loss": 0.8906,
      "step": 403
    },
    {
      "epoch": 0.24876847290640394,
      "grad_norm": 1.1159744262695312,
      "learning_rate": 4.651428571428572e-06,
      "loss": 0.8517,
      "step": 404
    },
    {
      "epoch": 0.24938423645320196,
      "grad_norm": 1.1211782693862915,
      "learning_rate": 4.647619047619048e-06,
      "loss": 0.8705,
      "step": 405
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1720571517944336,
      "learning_rate": 4.643809523809524e-06,
      "loss": 0.8453,
      "step": 406
    },
    {
      "epoch": 0.25,
      "eval_loss": 0.8883407711982727,
      "eval_runtime": 269.9361,
      "eval_samples_per_second": 85.613,
      "eval_steps_per_second": 2.678,
      "step": 406
    },
    {
      "epoch": 0.25061576354679804,
      "grad_norm": 1.326387643814087,
      "learning_rate": 4.64e-06,
      "loss": 0.7992,
      "step": 407
    },
    {
      "epoch": 0.2512315270935961,
      "grad_norm": 1.0635367631912231,
      "learning_rate": 4.636190476190476e-06,
      "loss": 0.8584,
      "step": 408
    },
    {
      "epoch": 0.2518472906403941,
      "grad_norm": 0.9910187125205994,
      "learning_rate": 4.6323809523809525e-06,
      "loss": 0.8958,
      "step": 409
    },
    {
      "epoch": 0.2524630541871921,
      "grad_norm": 1.2217634916305542,
      "learning_rate": 4.628571428571429e-06,
      "loss": 0.8247,
      "step": 410
    },
    {
      "epoch": 0.25307881773399016,
      "grad_norm": 1.0566860437393188,
      "learning_rate": 4.624761904761905e-06,
      "loss": 0.875,
      "step": 411
    },
    {
      "epoch": 0.2536945812807882,
      "grad_norm": 1.1172093152999878,
      "learning_rate": 4.620952380952381e-06,
      "loss": 0.8617,
      "step": 412
    },
    {
      "epoch": 0.2543103448275862,
      "grad_norm": 1.0818588733673096,
      "learning_rate": 4.617142857142857e-06,
      "loss": 0.8559,
      "step": 413
    },
    {
      "epoch": 0.25492610837438423,
      "grad_norm": 1.075353741645813,
      "learning_rate": 4.613333333333333e-06,
      "loss": 0.8378,
      "step": 414
    },
    {
      "epoch": 0.2555418719211823,
      "grad_norm": 1.0603846311569214,
      "learning_rate": 4.60952380952381e-06,
      "loss": 0.8344,
      "step": 415
    },
    {
      "epoch": 0.2561576354679803,
      "grad_norm": 1.1812664270401,
      "learning_rate": 4.605714285714286e-06,
      "loss": 0.8587,
      "step": 416
    },
    {
      "epoch": 0.2567733990147783,
      "grad_norm": 1.1849968433380127,
      "learning_rate": 4.601904761904762e-06,
      "loss": 0.8285,
      "step": 417
    },
    {
      "epoch": 0.25738916256157635,
      "grad_norm": 1.0550254583358765,
      "learning_rate": 4.598095238095238e-06,
      "loss": 0.8528,
      "step": 418
    },
    {
      "epoch": 0.2580049261083744,
      "grad_norm": 1.0451703071594238,
      "learning_rate": 4.594285714285714e-06,
      "loss": 0.8801,
      "step": 419
    },
    {
      "epoch": 0.25862068965517243,
      "grad_norm": 1.0322027206420898,
      "learning_rate": 4.590476190476191e-06,
      "loss": 0.8508,
      "step": 420
    },
    {
      "epoch": 0.2592364532019704,
      "grad_norm": 1.0139656066894531,
      "learning_rate": 4.586666666666667e-06,
      "loss": 0.8328,
      "step": 421
    },
    {
      "epoch": 0.25985221674876846,
      "grad_norm": 1.1668692827224731,
      "learning_rate": 4.582857142857143e-06,
      "loss": 0.8646,
      "step": 422
    },
    {
      "epoch": 0.2604679802955665,
      "grad_norm": 1.1138207912445068,
      "learning_rate": 4.579047619047619e-06,
      "loss": 0.8858,
      "step": 423
    },
    {
      "epoch": 0.26108374384236455,
      "grad_norm": 1.1367518901824951,
      "learning_rate": 4.5752380952380955e-06,
      "loss": 0.8709,
      "step": 424
    },
    {
      "epoch": 0.2616995073891626,
      "grad_norm": 1.062653660774231,
      "learning_rate": 4.571428571428571e-06,
      "loss": 0.8777,
      "step": 425
    },
    {
      "epoch": 0.2623152709359606,
      "grad_norm": 1.181408405303955,
      "learning_rate": 4.567619047619048e-06,
      "loss": 0.87,
      "step": 426
    },
    {
      "epoch": 0.2629310344827586,
      "grad_norm": 1.0364530086517334,
      "learning_rate": 4.563809523809524e-06,
      "loss": 0.8341,
      "step": 427
    },
    {
      "epoch": 0.26354679802955666,
      "grad_norm": 1.0678155422210693,
      "learning_rate": 4.56e-06,
      "loss": 0.8401,
      "step": 428
    },
    {
      "epoch": 0.2641625615763547,
      "grad_norm": 1.0431066751480103,
      "learning_rate": 4.556190476190476e-06,
      "loss": 0.9057,
      "step": 429
    },
    {
      "epoch": 0.2647783251231527,
      "grad_norm": 0.9909984469413757,
      "learning_rate": 4.552380952380952e-06,
      "loss": 0.8464,
      "step": 430
    },
    {
      "epoch": 0.26539408866995073,
      "grad_norm": 1.073660135269165,
      "learning_rate": 4.548571428571428e-06,
      "loss": 0.8414,
      "step": 431
    },
    {
      "epoch": 0.2660098522167488,
      "grad_norm": 1.016649603843689,
      "learning_rate": 4.544761904761905e-06,
      "loss": 0.8287,
      "step": 432
    },
    {
      "epoch": 0.2666256157635468,
      "grad_norm": 1.091370940208435,
      "learning_rate": 4.540952380952381e-06,
      "loss": 0.8495,
      "step": 433
    },
    {
      "epoch": 0.2672413793103448,
      "grad_norm": 1.1625758409500122,
      "learning_rate": 4.537142857142857e-06,
      "loss": 0.8358,
      "step": 434
    },
    {
      "epoch": 0.26785714285714285,
      "grad_norm": 1.1378281116485596,
      "learning_rate": 4.533333333333334e-06,
      "loss": 0.8637,
      "step": 435
    },
    {
      "epoch": 0.2684729064039409,
      "grad_norm": 1.1528282165527344,
      "learning_rate": 4.529523809523809e-06,
      "loss": 0.875,
      "step": 436
    },
    {
      "epoch": 0.26908866995073893,
      "grad_norm": 1.0709673166275024,
      "learning_rate": 4.5257142857142866e-06,
      "loss": 0.8754,
      "step": 437
    },
    {
      "epoch": 0.2697044334975369,
      "grad_norm": 1.111246943473816,
      "learning_rate": 4.521904761904762e-06,
      "loss": 0.8798,
      "step": 438
    },
    {
      "epoch": 0.27032019704433496,
      "grad_norm": 1.1035542488098145,
      "learning_rate": 4.5180952380952386e-06,
      "loss": 0.8727,
      "step": 439
    },
    {
      "epoch": 0.270935960591133,
      "grad_norm": 1.0389037132263184,
      "learning_rate": 4.514285714285714e-06,
      "loss": 0.8243,
      "step": 440
    },
    {
      "epoch": 0.27155172413793105,
      "grad_norm": 1.0198010206222534,
      "learning_rate": 4.5104761904761906e-06,
      "loss": 0.8361,
      "step": 441
    },
    {
      "epoch": 0.27216748768472904,
      "grad_norm": 1.0104531049728394,
      "learning_rate": 4.506666666666666e-06,
      "loss": 0.8355,
      "step": 442
    },
    {
      "epoch": 0.2727832512315271,
      "grad_norm": 1.0556972026824951,
      "learning_rate": 4.5028571428571434e-06,
      "loss": 0.8626,
      "step": 443
    },
    {
      "epoch": 0.2733990147783251,
      "grad_norm": 1.0757490396499634,
      "learning_rate": 4.499047619047619e-06,
      "loss": 0.8836,
      "step": 444
    },
    {
      "epoch": 0.27401477832512317,
      "grad_norm": 1.070380687713623,
      "learning_rate": 4.4952380952380954e-06,
      "loss": 0.8764,
      "step": 445
    },
    {
      "epoch": 0.2746305418719212,
      "grad_norm": 1.0719302892684937,
      "learning_rate": 4.491428571428571e-06,
      "loss": 0.8657,
      "step": 446
    },
    {
      "epoch": 0.2752463054187192,
      "grad_norm": 1.0640945434570312,
      "learning_rate": 4.4876190476190474e-06,
      "loss": 0.8354,
      "step": 447
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 1.0978323221206665,
      "learning_rate": 4.483809523809524e-06,
      "loss": 0.8564,
      "step": 448
    },
    {
      "epoch": 0.2764778325123153,
      "grad_norm": 1.0938689708709717,
      "learning_rate": 4.48e-06,
      "loss": 0.832,
      "step": 449
    },
    {
      "epoch": 0.2770935960591133,
      "grad_norm": 1.0815638303756714,
      "learning_rate": 4.476190476190477e-06,
      "loss": 0.8637,
      "step": 450
    },
    {
      "epoch": 0.2777093596059113,
      "grad_norm": 1.0313066244125366,
      "learning_rate": 4.472380952380952e-06,
      "loss": 0.8323,
      "step": 451
    },
    {
      "epoch": 0.27832512315270935,
      "grad_norm": 1.0554368495941162,
      "learning_rate": 4.468571428571429e-06,
      "loss": 0.8753,
      "step": 452
    },
    {
      "epoch": 0.2789408866995074,
      "grad_norm": 1.1731836795806885,
      "learning_rate": 4.464761904761904e-06,
      "loss": 0.8446,
      "step": 453
    },
    {
      "epoch": 0.27955665024630544,
      "grad_norm": 1.0367320775985718,
      "learning_rate": 4.460952380952382e-06,
      "loss": 0.8714,
      "step": 454
    },
    {
      "epoch": 0.2801724137931034,
      "grad_norm": 1.0686653852462769,
      "learning_rate": 4.457142857142857e-06,
      "loss": 0.8794,
      "step": 455
    },
    {
      "epoch": 0.28078817733990147,
      "grad_norm": 1.1889142990112305,
      "learning_rate": 4.453333333333334e-06,
      "loss": 0.8531,
      "step": 456
    },
    {
      "epoch": 0.2814039408866995,
      "grad_norm": 0.9977126121520996,
      "learning_rate": 4.449523809523809e-06,
      "loss": 0.8568,
      "step": 457
    },
    {
      "epoch": 0.28201970443349755,
      "grad_norm": 1.0368129014968872,
      "learning_rate": 4.445714285714286e-06,
      "loss": 0.8794,
      "step": 458
    },
    {
      "epoch": 0.28263546798029554,
      "grad_norm": 1.0818475484848022,
      "learning_rate": 4.441904761904762e-06,
      "loss": 0.8649,
      "step": 459
    },
    {
      "epoch": 0.2832512315270936,
      "grad_norm": 1.0848298072814941,
      "learning_rate": 4.4380952380952385e-06,
      "loss": 0.8607,
      "step": 460
    },
    {
      "epoch": 0.2838669950738916,
      "grad_norm": 1.1932364702224731,
      "learning_rate": 4.434285714285715e-06,
      "loss": 0.8794,
      "step": 461
    },
    {
      "epoch": 0.28448275862068967,
      "grad_norm": 1.1388568878173828,
      "learning_rate": 4.4304761904761905e-06,
      "loss": 0.904,
      "step": 462
    },
    {
      "epoch": 0.2850985221674877,
      "grad_norm": 1.0724272727966309,
      "learning_rate": 4.426666666666667e-06,
      "loss": 0.8611,
      "step": 463
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 1.1187509298324585,
      "learning_rate": 4.4228571428571425e-06,
      "loss": 0.8753,
      "step": 464
    },
    {
      "epoch": 0.28633004926108374,
      "grad_norm": 1.097156286239624,
      "learning_rate": 4.41904761904762e-06,
      "loss": 0.8454,
      "step": 465
    },
    {
      "epoch": 0.2869458128078818,
      "grad_norm": 1.111629843711853,
      "learning_rate": 4.415238095238095e-06,
      "loss": 0.8727,
      "step": 466
    },
    {
      "epoch": 0.2875615763546798,
      "grad_norm": 1.230038046836853,
      "learning_rate": 4.411428571428572e-06,
      "loss": 0.8937,
      "step": 467
    },
    {
      "epoch": 0.2881773399014778,
      "grad_norm": 0.964840829372406,
      "learning_rate": 4.407619047619047e-06,
      "loss": 0.8995,
      "step": 468
    },
    {
      "epoch": 0.28879310344827586,
      "grad_norm": 1.1765974760055542,
      "learning_rate": 4.403809523809524e-06,
      "loss": 0.8549,
      "step": 469
    },
    {
      "epoch": 0.2894088669950739,
      "grad_norm": 1.0168225765228271,
      "learning_rate": 4.399999999999999e-06,
      "loss": 0.8676,
      "step": 470
    },
    {
      "epoch": 0.29002463054187194,
      "grad_norm": 1.1367756128311157,
      "learning_rate": 4.396190476190477e-06,
      "loss": 0.8796,
      "step": 471
    },
    {
      "epoch": 0.29064039408866993,
      "grad_norm": 1.015526533126831,
      "learning_rate": 4.392380952380952e-06,
      "loss": 0.8783,
      "step": 472
    },
    {
      "epoch": 0.29125615763546797,
      "grad_norm": 1.0942308902740479,
      "learning_rate": 4.388571428571429e-06,
      "loss": 0.8218,
      "step": 473
    },
    {
      "epoch": 0.291871921182266,
      "grad_norm": 1.1513761281967163,
      "learning_rate": 4.384761904761905e-06,
      "loss": 0.8368,
      "step": 474
    },
    {
      "epoch": 0.29248768472906406,
      "grad_norm": 1.0758631229400635,
      "learning_rate": 4.380952380952381e-06,
      "loss": 0.8369,
      "step": 475
    },
    {
      "epoch": 0.29310344827586204,
      "grad_norm": 1.248110294342041,
      "learning_rate": 4.377142857142858e-06,
      "loss": 0.893,
      "step": 476
    },
    {
      "epoch": 0.2937192118226601,
      "grad_norm": 1.018654465675354,
      "learning_rate": 4.3733333333333335e-06,
      "loss": 0.8313,
      "step": 477
    },
    {
      "epoch": 0.29433497536945813,
      "grad_norm": 1.0988887548446655,
      "learning_rate": 4.36952380952381e-06,
      "loss": 0.9047,
      "step": 478
    },
    {
      "epoch": 0.29495073891625617,
      "grad_norm": 1.1662428379058838,
      "learning_rate": 4.3657142857142855e-06,
      "loss": 0.8534,
      "step": 479
    },
    {
      "epoch": 0.2955665024630542,
      "grad_norm": 1.1017979383468628,
      "learning_rate": 4.361904761904762e-06,
      "loss": 0.8259,
      "step": 480
    },
    {
      "epoch": 0.2961822660098522,
      "grad_norm": 1.1759361028671265,
      "learning_rate": 4.3580952380952376e-06,
      "loss": 0.8634,
      "step": 481
    },
    {
      "epoch": 0.29679802955665024,
      "grad_norm": 1.0892645120620728,
      "learning_rate": 4.354285714285715e-06,
      "loss": 0.8791,
      "step": 482
    },
    {
      "epoch": 0.2974137931034483,
      "grad_norm": 1.2275850772857666,
      "learning_rate": 4.35047619047619e-06,
      "loss": 0.8728,
      "step": 483
    },
    {
      "epoch": 0.29802955665024633,
      "grad_norm": 1.0316951274871826,
      "learning_rate": 4.346666666666667e-06,
      "loss": 0.8638,
      "step": 484
    },
    {
      "epoch": 0.2986453201970443,
      "grad_norm": 1.042384147644043,
      "learning_rate": 4.342857142857143e-06,
      "loss": 0.8453,
      "step": 485
    },
    {
      "epoch": 0.29926108374384236,
      "grad_norm": 1.1943469047546387,
      "learning_rate": 4.339047619047619e-06,
      "loss": 0.8921,
      "step": 486
    },
    {
      "epoch": 0.2998768472906404,
      "grad_norm": 1.185974359512329,
      "learning_rate": 4.335238095238095e-06,
      "loss": 0.8184,
      "step": 487
    },
    {
      "epoch": 0.30049261083743845,
      "grad_norm": 1.0955842733383179,
      "learning_rate": 4.331428571428572e-06,
      "loss": 0.8433,
      "step": 488
    },
    {
      "epoch": 0.30110837438423643,
      "grad_norm": 1.0485440492630005,
      "learning_rate": 4.327619047619048e-06,
      "loss": 0.8243,
      "step": 489
    },
    {
      "epoch": 0.3017241379310345,
      "grad_norm": 2.9736337661743164,
      "learning_rate": 4.323809523809524e-06,
      "loss": 0.8537,
      "step": 490
    },
    {
      "epoch": 0.3023399014778325,
      "grad_norm": 1.0855962038040161,
      "learning_rate": 4.32e-06,
      "loss": 0.8689,
      "step": 491
    },
    {
      "epoch": 0.30295566502463056,
      "grad_norm": 1.0700222253799438,
      "learning_rate": 4.316190476190476e-06,
      "loss": 0.8342,
      "step": 492
    },
    {
      "epoch": 0.30357142857142855,
      "grad_norm": 1.0470479726791382,
      "learning_rate": 4.312380952380953e-06,
      "loss": 0.8923,
      "step": 493
    },
    {
      "epoch": 0.3041871921182266,
      "grad_norm": 1.094109058380127,
      "learning_rate": 4.308571428571429e-06,
      "loss": 0.881,
      "step": 494
    },
    {
      "epoch": 0.30480295566502463,
      "grad_norm": 1.2364366054534912,
      "learning_rate": 4.304761904761905e-06,
      "loss": 0.8526,
      "step": 495
    },
    {
      "epoch": 0.3054187192118227,
      "grad_norm": 1.1311712265014648,
      "learning_rate": 4.300952380952381e-06,
      "loss": 0.8529,
      "step": 496
    },
    {
      "epoch": 0.30603448275862066,
      "grad_norm": 1.1069450378417969,
      "learning_rate": 4.297142857142857e-06,
      "loss": 0.8476,
      "step": 497
    },
    {
      "epoch": 0.3066502463054187,
      "grad_norm": 0.9692174792289734,
      "learning_rate": 4.2933333333333334e-06,
      "loss": 0.8587,
      "step": 498
    },
    {
      "epoch": 0.30726600985221675,
      "grad_norm": 1.0376008749008179,
      "learning_rate": 4.28952380952381e-06,
      "loss": 0.8969,
      "step": 499
    },
    {
      "epoch": 0.3078817733990148,
      "grad_norm": 1.0374581813812256,
      "learning_rate": 4.285714285714286e-06,
      "loss": 0.8394,
      "step": 500
    },
    {
      "epoch": 0.30849753694581283,
      "grad_norm": 1.11138916015625,
      "learning_rate": 4.281904761904762e-06,
      "loss": 0.8599,
      "step": 501
    },
    {
      "epoch": 0.3091133004926108,
      "grad_norm": 1.156724452972412,
      "learning_rate": 4.278095238095238e-06,
      "loss": 0.8495,
      "step": 502
    },
    {
      "epoch": 0.30972906403940886,
      "grad_norm": 1.1152230501174927,
      "learning_rate": 4.274285714285714e-06,
      "loss": 0.8576,
      "step": 503
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 1.064919114112854,
      "learning_rate": 4.270476190476191e-06,
      "loss": 0.8229,
      "step": 504
    },
    {
      "epoch": 0.31096059113300495,
      "grad_norm": 1.0233497619628906,
      "learning_rate": 4.266666666666667e-06,
      "loss": 0.8534,
      "step": 505
    },
    {
      "epoch": 0.31157635467980294,
      "grad_norm": 1.1392590999603271,
      "learning_rate": 4.262857142857143e-06,
      "loss": 0.8301,
      "step": 506
    },
    {
      "epoch": 0.312192118226601,
      "grad_norm": 1.0085766315460205,
      "learning_rate": 4.259047619047619e-06,
      "loss": 0.844,
      "step": 507
    },
    {
      "epoch": 0.312807881773399,
      "grad_norm": 1.018964409828186,
      "learning_rate": 4.255238095238095e-06,
      "loss": 0.8272,
      "step": 508
    },
    {
      "epoch": 0.31342364532019706,
      "grad_norm": 1.003261685371399,
      "learning_rate": 4.251428571428572e-06,
      "loss": 0.8414,
      "step": 509
    },
    {
      "epoch": 0.31403940886699505,
      "grad_norm": 1.1973103284835815,
      "learning_rate": 4.247619047619048e-06,
      "loss": 0.8358,
      "step": 510
    },
    {
      "epoch": 0.3146551724137931,
      "grad_norm": 1.0562916994094849,
      "learning_rate": 4.243809523809524e-06,
      "loss": 0.8396,
      "step": 511
    },
    {
      "epoch": 0.31527093596059114,
      "grad_norm": 1.044811487197876,
      "learning_rate": 4.24e-06,
      "loss": 0.8599,
      "step": 512
    },
    {
      "epoch": 0.3158866995073892,
      "grad_norm": 1.0390619039535522,
      "learning_rate": 4.2361904761904765e-06,
      "loss": 0.8294,
      "step": 513
    },
    {
      "epoch": 0.31650246305418717,
      "grad_norm": 1.3433046340942383,
      "learning_rate": 4.232380952380952e-06,
      "loss": 0.8529,
      "step": 514
    },
    {
      "epoch": 0.3171182266009852,
      "grad_norm": 1.0149775743484497,
      "learning_rate": 4.228571428571429e-06,
      "loss": 0.9224,
      "step": 515
    },
    {
      "epoch": 0.31773399014778325,
      "grad_norm": 1.1930155754089355,
      "learning_rate": 4.224761904761905e-06,
      "loss": 0.8373,
      "step": 516
    },
    {
      "epoch": 0.3183497536945813,
      "grad_norm": 1.002420425415039,
      "learning_rate": 4.220952380952381e-06,
      "loss": 0.8853,
      "step": 517
    },
    {
      "epoch": 0.31896551724137934,
      "grad_norm": 1.1205101013183594,
      "learning_rate": 4.217142857142857e-06,
      "loss": 0.8601,
      "step": 518
    },
    {
      "epoch": 0.3195812807881773,
      "grad_norm": 1.0698479413986206,
      "learning_rate": 4.213333333333333e-06,
      "loss": 0.8918,
      "step": 519
    },
    {
      "epoch": 0.32019704433497537,
      "grad_norm": 1.1270798444747925,
      "learning_rate": 4.209523809523809e-06,
      "loss": 0.8525,
      "step": 520
    },
    {
      "epoch": 0.3208128078817734,
      "grad_norm": 1.01841402053833,
      "learning_rate": 4.205714285714286e-06,
      "loss": 0.8022,
      "step": 521
    },
    {
      "epoch": 0.32142857142857145,
      "grad_norm": 1.0746996402740479,
      "learning_rate": 4.201904761904762e-06,
      "loss": 0.8318,
      "step": 522
    },
    {
      "epoch": 0.32204433497536944,
      "grad_norm": 1.0839848518371582,
      "learning_rate": 4.198095238095238e-06,
      "loss": 0.8429,
      "step": 523
    },
    {
      "epoch": 0.3226600985221675,
      "grad_norm": 1.060539960861206,
      "learning_rate": 4.194285714285715e-06,
      "loss": 0.8818,
      "step": 524
    },
    {
      "epoch": 0.3232758620689655,
      "grad_norm": 1.075338363647461,
      "learning_rate": 4.19047619047619e-06,
      "loss": 0.8661,
      "step": 525
    },
    {
      "epoch": 0.32389162561576357,
      "grad_norm": 1.0940507650375366,
      "learning_rate": 4.186666666666667e-06,
      "loss": 0.8203,
      "step": 526
    },
    {
      "epoch": 0.32450738916256155,
      "grad_norm": 1.0409997701644897,
      "learning_rate": 4.182857142857143e-06,
      "loss": 0.8556,
      "step": 527
    },
    {
      "epoch": 0.3251231527093596,
      "grad_norm": 1.2078025341033936,
      "learning_rate": 4.1790476190476195e-06,
      "loss": 0.8753,
      "step": 528
    },
    {
      "epoch": 0.32573891625615764,
      "grad_norm": 1.0868655443191528,
      "learning_rate": 4.175238095238095e-06,
      "loss": 0.85,
      "step": 529
    },
    {
      "epoch": 0.3263546798029557,
      "grad_norm": 1.0788785219192505,
      "learning_rate": 4.1714285714285715e-06,
      "loss": 0.8638,
      "step": 530
    },
    {
      "epoch": 0.32697044334975367,
      "grad_norm": 1.1450470685958862,
      "learning_rate": 4.167619047619047e-06,
      "loss": 0.8544,
      "step": 531
    },
    {
      "epoch": 0.3275862068965517,
      "grad_norm": 1.1287654638290405,
      "learning_rate": 4.163809523809524e-06,
      "loss": 0.8806,
      "step": 532
    },
    {
      "epoch": 0.32820197044334976,
      "grad_norm": 1.0987231731414795,
      "learning_rate": 4.16e-06,
      "loss": 0.811,
      "step": 533
    },
    {
      "epoch": 0.3288177339901478,
      "grad_norm": 1.122533917427063,
      "learning_rate": 4.156190476190476e-06,
      "loss": 0.8565,
      "step": 534
    },
    {
      "epoch": 0.3294334975369458,
      "grad_norm": 1.0209048986434937,
      "learning_rate": 4.152380952380952e-06,
      "loss": 0.873,
      "step": 535
    },
    {
      "epoch": 0.33004926108374383,
      "grad_norm": 1.0634294748306274,
      "learning_rate": 4.148571428571428e-06,
      "loss": 0.8976,
      "step": 536
    },
    {
      "epoch": 0.33066502463054187,
      "grad_norm": 1.0890878438949585,
      "learning_rate": 4.144761904761905e-06,
      "loss": 0.8541,
      "step": 537
    },
    {
      "epoch": 0.3312807881773399,
      "grad_norm": 1.0536800622940063,
      "learning_rate": 4.140952380952381e-06,
      "loss": 0.8586,
      "step": 538
    },
    {
      "epoch": 0.33189655172413796,
      "grad_norm": 1.0224428176879883,
      "learning_rate": 4.137142857142858e-06,
      "loss": 0.8551,
      "step": 539
    },
    {
      "epoch": 0.33251231527093594,
      "grad_norm": 1.1248220205307007,
      "learning_rate": 4.133333333333333e-06,
      "loss": 0.8286,
      "step": 540
    },
    {
      "epoch": 0.333128078817734,
      "grad_norm": 1.1711807250976562,
      "learning_rate": 4.12952380952381e-06,
      "loss": 0.8517,
      "step": 541
    },
    {
      "epoch": 0.33374384236453203,
      "grad_norm": 1.0651068687438965,
      "learning_rate": 4.125714285714285e-06,
      "loss": 0.8584,
      "step": 542
    },
    {
      "epoch": 0.33435960591133007,
      "grad_norm": 1.0584664344787598,
      "learning_rate": 4.1219047619047626e-06,
      "loss": 0.8397,
      "step": 543
    },
    {
      "epoch": 0.33497536945812806,
      "grad_norm": 1.0545891523361206,
      "learning_rate": 4.118095238095238e-06,
      "loss": 0.8371,
      "step": 544
    },
    {
      "epoch": 0.3355911330049261,
      "grad_norm": 1.1821256875991821,
      "learning_rate": 4.114285714285715e-06,
      "loss": 0.9384,
      "step": 545
    },
    {
      "epoch": 0.33620689655172414,
      "grad_norm": 1.0218020677566528,
      "learning_rate": 4.11047619047619e-06,
      "loss": 0.8302,
      "step": 546
    },
    {
      "epoch": 0.3368226600985222,
      "grad_norm": 1.0062767267227173,
      "learning_rate": 4.106666666666667e-06,
      "loss": 0.8804,
      "step": 547
    },
    {
      "epoch": 0.3374384236453202,
      "grad_norm": 1.0590101480484009,
      "learning_rate": 4.102857142857143e-06,
      "loss": 0.8296,
      "step": 548
    },
    {
      "epoch": 0.3380541871921182,
      "grad_norm": 1.6194275617599487,
      "learning_rate": 4.0990476190476195e-06,
      "loss": 0.8417,
      "step": 549
    },
    {
      "epoch": 0.33866995073891626,
      "grad_norm": 1.010354995727539,
      "learning_rate": 4.095238095238095e-06,
      "loss": 0.8768,
      "step": 550
    },
    {
      "epoch": 0.3392857142857143,
      "grad_norm": 1.1116127967834473,
      "learning_rate": 4.0914285714285715e-06,
      "loss": 0.8085,
      "step": 551
    },
    {
      "epoch": 0.3399014778325123,
      "grad_norm": 0.9940347075462341,
      "learning_rate": 4.087619047619048e-06,
      "loss": 0.8453,
      "step": 552
    },
    {
      "epoch": 0.34051724137931033,
      "grad_norm": 1.1591161489486694,
      "learning_rate": 4.0838095238095235e-06,
      "loss": 0.8516,
      "step": 553
    },
    {
      "epoch": 0.3411330049261084,
      "grad_norm": 1.0663765668869019,
      "learning_rate": 4.080000000000001e-06,
      "loss": 0.8212,
      "step": 554
    },
    {
      "epoch": 0.3417487684729064,
      "grad_norm": 1.0029891729354858,
      "learning_rate": 4.076190476190476e-06,
      "loss": 0.8472,
      "step": 555
    },
    {
      "epoch": 0.34236453201970446,
      "grad_norm": 1.136018991470337,
      "learning_rate": 4.072380952380953e-06,
      "loss": 0.8568,
      "step": 556
    },
    {
      "epoch": 0.34298029556650245,
      "grad_norm": 1.0130082368850708,
      "learning_rate": 4.068571428571428e-06,
      "loss": 0.8736,
      "step": 557
    },
    {
      "epoch": 0.3435960591133005,
      "grad_norm": 1.1266756057739258,
      "learning_rate": 4.064761904761905e-06,
      "loss": 0.8255,
      "step": 558
    },
    {
      "epoch": 0.34421182266009853,
      "grad_norm": 1.0904688835144043,
      "learning_rate": 4.06095238095238e-06,
      "loss": 0.881,
      "step": 559
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 1.1094493865966797,
      "learning_rate": 4.057142857142858e-06,
      "loss": 0.8931,
      "step": 560
    },
    {
      "epoch": 0.34544334975369456,
      "grad_norm": 0.9849966168403625,
      "learning_rate": 4.053333333333333e-06,
      "loss": 0.83,
      "step": 561
    },
    {
      "epoch": 0.3460591133004926,
      "grad_norm": 0.9797407388687134,
      "learning_rate": 4.04952380952381e-06,
      "loss": 0.8667,
      "step": 562
    },
    {
      "epoch": 0.34667487684729065,
      "grad_norm": 1.2623845338821411,
      "learning_rate": 4.045714285714286e-06,
      "loss": 0.8651,
      "step": 563
    },
    {
      "epoch": 0.3472906403940887,
      "grad_norm": 1.143247127532959,
      "learning_rate": 4.041904761904762e-06,
      "loss": 0.867,
      "step": 564
    },
    {
      "epoch": 0.3479064039408867,
      "grad_norm": 1.1282062530517578,
      "learning_rate": 4.038095238095239e-06,
      "loss": 0.8899,
      "step": 565
    },
    {
      "epoch": 0.3485221674876847,
      "grad_norm": 0.9871397614479065,
      "learning_rate": 4.0342857142857145e-06,
      "loss": 0.8472,
      "step": 566
    },
    {
      "epoch": 0.34913793103448276,
      "grad_norm": 1.0571969747543335,
      "learning_rate": 4.030476190476191e-06,
      "loss": 0.8915,
      "step": 567
    },
    {
      "epoch": 0.3497536945812808,
      "grad_norm": 0.9758468866348267,
      "learning_rate": 4.0266666666666665e-06,
      "loss": 0.8537,
      "step": 568
    },
    {
      "epoch": 0.3503694581280788,
      "grad_norm": 1.0342894792556763,
      "learning_rate": 4.022857142857143e-06,
      "loss": 0.8505,
      "step": 569
    },
    {
      "epoch": 0.35098522167487683,
      "grad_norm": 1.0493879318237305,
      "learning_rate": 4.0190476190476185e-06,
      "loss": 0.8784,
      "step": 570
    },
    {
      "epoch": 0.3516009852216749,
      "grad_norm": 1.1192538738250732,
      "learning_rate": 4.015238095238096e-06,
      "loss": 0.8616,
      "step": 571
    },
    {
      "epoch": 0.3522167487684729,
      "grad_norm": 1.0433403253555298,
      "learning_rate": 4.011428571428571e-06,
      "loss": 0.8563,
      "step": 572
    },
    {
      "epoch": 0.35283251231527096,
      "grad_norm": 1.0584360361099243,
      "learning_rate": 4.007619047619048e-06,
      "loss": 0.8569,
      "step": 573
    },
    {
      "epoch": 0.35344827586206895,
      "grad_norm": 1.1038928031921387,
      "learning_rate": 4.003809523809523e-06,
      "loss": 0.8814,
      "step": 574
    },
    {
      "epoch": 0.354064039408867,
      "grad_norm": 1.0374562740325928,
      "learning_rate": 4e-06,
      "loss": 0.8447,
      "step": 575
    },
    {
      "epoch": 0.35467980295566504,
      "grad_norm": 1.0578666925430298,
      "learning_rate": 3.996190476190476e-06,
      "loss": 0.8262,
      "step": 576
    },
    {
      "epoch": 0.3552955665024631,
      "grad_norm": 1.0964778661727905,
      "learning_rate": 3.992380952380953e-06,
      "loss": 0.8746,
      "step": 577
    },
    {
      "epoch": 0.35591133004926107,
      "grad_norm": 1.0103867053985596,
      "learning_rate": 3.988571428571429e-06,
      "loss": 0.8405,
      "step": 578
    },
    {
      "epoch": 0.3565270935960591,
      "grad_norm": 1.076891303062439,
      "learning_rate": 3.984761904761905e-06,
      "loss": 0.8433,
      "step": 579
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 0.978365957736969,
      "learning_rate": 3.980952380952381e-06,
      "loss": 0.8658,
      "step": 580
    },
    {
      "epoch": 0.3577586206896552,
      "grad_norm": 1.0098029375076294,
      "learning_rate": 3.977142857142857e-06,
      "loss": 0.8462,
      "step": 581
    },
    {
      "epoch": 0.3583743842364532,
      "grad_norm": 1.063058614730835,
      "learning_rate": 3.973333333333334e-06,
      "loss": 0.838,
      "step": 582
    },
    {
      "epoch": 0.3589901477832512,
      "grad_norm": 1.0049047470092773,
      "learning_rate": 3.9695238095238096e-06,
      "loss": 0.8745,
      "step": 583
    },
    {
      "epoch": 0.35960591133004927,
      "grad_norm": 1.0930806398391724,
      "learning_rate": 3.965714285714286e-06,
      "loss": 0.8494,
      "step": 584
    },
    {
      "epoch": 0.3602216748768473,
      "grad_norm": 1.0398629903793335,
      "learning_rate": 3.9619047619047616e-06,
      "loss": 0.8582,
      "step": 585
    },
    {
      "epoch": 0.3608374384236453,
      "grad_norm": 1.0607738494873047,
      "learning_rate": 3.958095238095238e-06,
      "loss": 0.8182,
      "step": 586
    },
    {
      "epoch": 0.36145320197044334,
      "grad_norm": 1.109592080116272,
      "learning_rate": 3.954285714285714e-06,
      "loss": 0.8663,
      "step": 587
    },
    {
      "epoch": 0.3620689655172414,
      "grad_norm": 1.1596324443817139,
      "learning_rate": 3.950476190476191e-06,
      "loss": 0.8686,
      "step": 588
    },
    {
      "epoch": 0.3626847290640394,
      "grad_norm": 1.0123991966247559,
      "learning_rate": 3.946666666666667e-06,
      "loss": 0.8701,
      "step": 589
    },
    {
      "epoch": 0.3633004926108374,
      "grad_norm": 1.1007795333862305,
      "learning_rate": 3.942857142857143e-06,
      "loss": 0.8161,
      "step": 590
    },
    {
      "epoch": 0.36391625615763545,
      "grad_norm": 1.0757731199264526,
      "learning_rate": 3.939047619047619e-06,
      "loss": 0.8232,
      "step": 591
    },
    {
      "epoch": 0.3645320197044335,
      "grad_norm": 1.0279558897018433,
      "learning_rate": 3.935238095238095e-06,
      "loss": 0.8693,
      "step": 592
    },
    {
      "epoch": 0.36514778325123154,
      "grad_norm": 1.0190210342407227,
      "learning_rate": 3.931428571428572e-06,
      "loss": 0.8506,
      "step": 593
    },
    {
      "epoch": 0.3657635467980296,
      "grad_norm": 1.0958889722824097,
      "learning_rate": 3.927619047619048e-06,
      "loss": 0.8709,
      "step": 594
    },
    {
      "epoch": 0.36637931034482757,
      "grad_norm": 1.104931354522705,
      "learning_rate": 3.923809523809524e-06,
      "loss": 0.8323,
      "step": 595
    },
    {
      "epoch": 0.3669950738916256,
      "grad_norm": 0.9790024161338806,
      "learning_rate": 3.92e-06,
      "loss": 0.8205,
      "step": 596
    },
    {
      "epoch": 0.36761083743842365,
      "grad_norm": 1.0219544172286987,
      "learning_rate": 3.916190476190476e-06,
      "loss": 0.8722,
      "step": 597
    },
    {
      "epoch": 0.3682266009852217,
      "grad_norm": 1.0553905963897705,
      "learning_rate": 3.912380952380952e-06,
      "loss": 0.8691,
      "step": 598
    },
    {
      "epoch": 0.3688423645320197,
      "grad_norm": 0.9860970973968506,
      "learning_rate": 3.908571428571429e-06,
      "loss": 0.8463,
      "step": 599
    },
    {
      "epoch": 0.3694581280788177,
      "grad_norm": 1.0598429441452026,
      "learning_rate": 3.904761904761905e-06,
      "loss": 0.904,
      "step": 600
    },
    {
      "epoch": 0.37007389162561577,
      "grad_norm": 0.9757109880447388,
      "learning_rate": 3.900952380952381e-06,
      "loss": 0.8869,
      "step": 601
    },
    {
      "epoch": 0.3706896551724138,
      "grad_norm": 0.9772175550460815,
      "learning_rate": 3.8971428571428575e-06,
      "loss": 0.8805,
      "step": 602
    },
    {
      "epoch": 0.3713054187192118,
      "grad_norm": 1.0357502698898315,
      "learning_rate": 3.893333333333333e-06,
      "loss": 0.8217,
      "step": 603
    },
    {
      "epoch": 0.37192118226600984,
      "grad_norm": 0.9697862267494202,
      "learning_rate": 3.88952380952381e-06,
      "loss": 0.8482,
      "step": 604
    },
    {
      "epoch": 0.3725369458128079,
      "grad_norm": 1.0220838785171509,
      "learning_rate": 3.885714285714286e-06,
      "loss": 0.8779,
      "step": 605
    },
    {
      "epoch": 0.3731527093596059,
      "grad_norm": 1.0987836122512817,
      "learning_rate": 3.881904761904762e-06,
      "loss": 0.8946,
      "step": 606
    },
    {
      "epoch": 0.3737684729064039,
      "grad_norm": 2.2758753299713135,
      "learning_rate": 3.878095238095238e-06,
      "loss": 0.8298,
      "step": 607
    },
    {
      "epoch": 0.37438423645320196,
      "grad_norm": 1.1647089719772339,
      "learning_rate": 3.874285714285714e-06,
      "loss": 0.9049,
      "step": 608
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.9756439328193665,
      "learning_rate": 3.87047619047619e-06,
      "loss": 0.8822,
      "step": 609
    },
    {
      "epoch": 0.37561576354679804,
      "grad_norm": 1.0878584384918213,
      "learning_rate": 3.866666666666667e-06,
      "loss": 0.8211,
      "step": 610
    },
    {
      "epoch": 0.3762315270935961,
      "grad_norm": 1.2449827194213867,
      "learning_rate": 3.862857142857143e-06,
      "loss": 0.8974,
      "step": 611
    },
    {
      "epoch": 0.3768472906403941,
      "grad_norm": 0.9817798137664795,
      "learning_rate": 3.859047619047619e-06,
      "loss": 0.8496,
      "step": 612
    },
    {
      "epoch": 0.3774630541871921,
      "grad_norm": 1.222293496131897,
      "learning_rate": 3.855238095238096e-06,
      "loss": 0.8886,
      "step": 613
    },
    {
      "epoch": 0.37807881773399016,
      "grad_norm": 1.151087999343872,
      "learning_rate": 3.851428571428571e-06,
      "loss": 0.8456,
      "step": 614
    },
    {
      "epoch": 0.3786945812807882,
      "grad_norm": 1.110823392868042,
      "learning_rate": 3.847619047619048e-06,
      "loss": 0.8786,
      "step": 615
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 1.1524778604507446,
      "learning_rate": 3.843809523809524e-06,
      "loss": 0.839,
      "step": 616
    },
    {
      "epoch": 0.37992610837438423,
      "grad_norm": 1.076412558555603,
      "learning_rate": 3.8400000000000005e-06,
      "loss": 0.8656,
      "step": 617
    },
    {
      "epoch": 0.3805418719211823,
      "grad_norm": 0.9950555562973022,
      "learning_rate": 3.836190476190476e-06,
      "loss": 0.8468,
      "step": 618
    },
    {
      "epoch": 0.3811576354679803,
      "grad_norm": 0.9644134044647217,
      "learning_rate": 3.8323809523809525e-06,
      "loss": 0.8389,
      "step": 619
    },
    {
      "epoch": 0.3817733990147783,
      "grad_norm": 0.9292529821395874,
      "learning_rate": 3.828571428571428e-06,
      "loss": 0.87,
      "step": 620
    },
    {
      "epoch": 0.38238916256157635,
      "grad_norm": 1.1063586473464966,
      "learning_rate": 3.824761904761905e-06,
      "loss": 0.7936,
      "step": 621
    },
    {
      "epoch": 0.3830049261083744,
      "grad_norm": 1.0533337593078613,
      "learning_rate": 3.820952380952381e-06,
      "loss": 0.8801,
      "step": 622
    },
    {
      "epoch": 0.38362068965517243,
      "grad_norm": 1.399061918258667,
      "learning_rate": 3.817142857142857e-06,
      "loss": 0.8312,
      "step": 623
    },
    {
      "epoch": 0.3842364532019704,
      "grad_norm": 1.0913255214691162,
      "learning_rate": 3.8133333333333334e-06,
      "loss": 0.8465,
      "step": 624
    },
    {
      "epoch": 0.38485221674876846,
      "grad_norm": 1.0454483032226562,
      "learning_rate": 3.8095238095238094e-06,
      "loss": 0.8402,
      "step": 625
    },
    {
      "epoch": 0.3854679802955665,
      "grad_norm": 1.101060152053833,
      "learning_rate": 3.8057142857142854e-06,
      "loss": 0.8682,
      "step": 626
    },
    {
      "epoch": 0.38608374384236455,
      "grad_norm": 1.0114318132400513,
      "learning_rate": 3.8019047619047622e-06,
      "loss": 0.8779,
      "step": 627
    },
    {
      "epoch": 0.3866995073891626,
      "grad_norm": 1.0350960493087769,
      "learning_rate": 3.7980952380952383e-06,
      "loss": 0.9325,
      "step": 628
    },
    {
      "epoch": 0.3873152709359606,
      "grad_norm": 1.044208288192749,
      "learning_rate": 3.7942857142857143e-06,
      "loss": 0.8223,
      "step": 629
    },
    {
      "epoch": 0.3879310344827586,
      "grad_norm": 1.0012304782867432,
      "learning_rate": 3.7904761904761903e-06,
      "loss": 0.8302,
      "step": 630
    },
    {
      "epoch": 0.38854679802955666,
      "grad_norm": 1.011075496673584,
      "learning_rate": 3.7866666666666667e-06,
      "loss": 0.8362,
      "step": 631
    },
    {
      "epoch": 0.3891625615763547,
      "grad_norm": 1.112684965133667,
      "learning_rate": 3.782857142857143e-06,
      "loss": 0.8716,
      "step": 632
    },
    {
      "epoch": 0.3897783251231527,
      "grad_norm": 1.112701177597046,
      "learning_rate": 3.779047619047619e-06,
      "loss": 0.8315,
      "step": 633
    },
    {
      "epoch": 0.39039408866995073,
      "grad_norm": 1.0335479974746704,
      "learning_rate": 3.7752380952380956e-06,
      "loss": 0.8208,
      "step": 634
    },
    {
      "epoch": 0.3910098522167488,
      "grad_norm": 1.0139485597610474,
      "learning_rate": 3.7714285714285716e-06,
      "loss": 0.8636,
      "step": 635
    },
    {
      "epoch": 0.3916256157635468,
      "grad_norm": 1.2703145742416382,
      "learning_rate": 3.7676190476190476e-06,
      "loss": 0.8545,
      "step": 636
    },
    {
      "epoch": 0.3922413793103448,
      "grad_norm": 1.0527417659759521,
      "learning_rate": 3.7638095238095236e-06,
      "loss": 0.7899,
      "step": 637
    },
    {
      "epoch": 0.39285714285714285,
      "grad_norm": 1.1055347919464111,
      "learning_rate": 3.7600000000000004e-06,
      "loss": 0.8214,
      "step": 638
    },
    {
      "epoch": 0.3934729064039409,
      "grad_norm": 0.9779605865478516,
      "learning_rate": 3.7561904761904764e-06,
      "loss": 0.8671,
      "step": 639
    },
    {
      "epoch": 0.39408866995073893,
      "grad_norm": 1.0249648094177246,
      "learning_rate": 3.7523809523809524e-06,
      "loss": 0.8318,
      "step": 640
    },
    {
      "epoch": 0.3947044334975369,
      "grad_norm": 1.1067818403244019,
      "learning_rate": 3.7485714285714284e-06,
      "loss": 0.8528,
      "step": 641
    },
    {
      "epoch": 0.39532019704433496,
      "grad_norm": 1.5875294208526611,
      "learning_rate": 3.7447619047619044e-06,
      "loss": 0.8428,
      "step": 642
    },
    {
      "epoch": 0.395935960591133,
      "grad_norm": 0.9768543839454651,
      "learning_rate": 3.7409523809523813e-06,
      "loss": 0.8894,
      "step": 643
    },
    {
      "epoch": 0.39655172413793105,
      "grad_norm": 1.019620418548584,
      "learning_rate": 3.7371428571428573e-06,
      "loss": 0.8477,
      "step": 644
    },
    {
      "epoch": 0.39716748768472904,
      "grad_norm": 1.0284451246261597,
      "learning_rate": 3.7333333333333333e-06,
      "loss": 0.8732,
      "step": 645
    },
    {
      "epoch": 0.3977832512315271,
      "grad_norm": 1.0007133483886719,
      "learning_rate": 3.7295238095238097e-06,
      "loss": 0.8526,
      "step": 646
    },
    {
      "epoch": 0.3983990147783251,
      "grad_norm": 1.0182842016220093,
      "learning_rate": 3.7257142857142857e-06,
      "loss": 0.8113,
      "step": 647
    },
    {
      "epoch": 0.39901477832512317,
      "grad_norm": 1.0092148780822754,
      "learning_rate": 3.7219047619047617e-06,
      "loss": 0.8786,
      "step": 648
    },
    {
      "epoch": 0.3996305418719212,
      "grad_norm": 1.0278863906860352,
      "learning_rate": 3.7180952380952386e-06,
      "loss": 0.7931,
      "step": 649
    },
    {
      "epoch": 0.4002463054187192,
      "grad_norm": 1.114086627960205,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 0.8272,
      "step": 650
    },
    {
      "epoch": 0.40086206896551724,
      "grad_norm": 1.0934114456176758,
      "learning_rate": 3.7104761904761906e-06,
      "loss": 0.8425,
      "step": 651
    },
    {
      "epoch": 0.4014778325123153,
      "grad_norm": 1.0394742488861084,
      "learning_rate": 3.7066666666666666e-06,
      "loss": 0.849,
      "step": 652
    },
    {
      "epoch": 0.4020935960591133,
      "grad_norm": 0.9983641505241394,
      "learning_rate": 3.7028571428571426e-06,
      "loss": 0.8583,
      "step": 653
    },
    {
      "epoch": 0.4027093596059113,
      "grad_norm": 1.1975606679916382,
      "learning_rate": 3.6990476190476195e-06,
      "loss": 0.8361,
      "step": 654
    },
    {
      "epoch": 0.40332512315270935,
      "grad_norm": 0.9864385724067688,
      "learning_rate": 3.6952380952380955e-06,
      "loss": 0.8364,
      "step": 655
    },
    {
      "epoch": 0.4039408866995074,
      "grad_norm": 1.0290805101394653,
      "learning_rate": 3.6914285714285715e-06,
      "loss": 0.8461,
      "step": 656
    },
    {
      "epoch": 0.40455665024630544,
      "grad_norm": 1.1939847469329834,
      "learning_rate": 3.6876190476190475e-06,
      "loss": 0.8682,
      "step": 657
    },
    {
      "epoch": 0.4051724137931034,
      "grad_norm": 1.019196629524231,
      "learning_rate": 3.683809523809524e-06,
      "loss": 0.8743,
      "step": 658
    },
    {
      "epoch": 0.40578817733990147,
      "grad_norm": 1.0902228355407715,
      "learning_rate": 3.68e-06,
      "loss": 0.8262,
      "step": 659
    },
    {
      "epoch": 0.4064039408866995,
      "grad_norm": 1.112337589263916,
      "learning_rate": 3.6761904761904763e-06,
      "loss": 0.8227,
      "step": 660
    },
    {
      "epoch": 0.40701970443349755,
      "grad_norm": 1.1501730680465698,
      "learning_rate": 3.6723809523809528e-06,
      "loss": 0.8368,
      "step": 661
    },
    {
      "epoch": 0.40763546798029554,
      "grad_norm": 1.0485265254974365,
      "learning_rate": 3.6685714285714288e-06,
      "loss": 0.847,
      "step": 662
    },
    {
      "epoch": 0.4082512315270936,
      "grad_norm": 1.1670284271240234,
      "learning_rate": 3.6647619047619048e-06,
      "loss": 0.8762,
      "step": 663
    },
    {
      "epoch": 0.4088669950738916,
      "grad_norm": 1.0823757648468018,
      "learning_rate": 3.660952380952381e-06,
      "loss": 0.8513,
      "step": 664
    },
    {
      "epoch": 0.40948275862068967,
      "grad_norm": 1.1498258113861084,
      "learning_rate": 3.6571428571428576e-06,
      "loss": 0.8528,
      "step": 665
    },
    {
      "epoch": 0.4100985221674877,
      "grad_norm": 1.0984798669815063,
      "learning_rate": 3.6533333333333336e-06,
      "loss": 0.84,
      "step": 666
    },
    {
      "epoch": 0.4107142857142857,
      "grad_norm": 1.0659737586975098,
      "learning_rate": 3.6495238095238097e-06,
      "loss": 0.8927,
      "step": 667
    },
    {
      "epoch": 0.41133004926108374,
      "grad_norm": 1.000779390335083,
      "learning_rate": 3.6457142857142857e-06,
      "loss": 0.8307,
      "step": 668
    },
    {
      "epoch": 0.4119458128078818,
      "grad_norm": 1.1248339414596558,
      "learning_rate": 3.6419047619047617e-06,
      "loss": 0.8974,
      "step": 669
    },
    {
      "epoch": 0.4125615763546798,
      "grad_norm": 1.0574941635131836,
      "learning_rate": 3.638095238095238e-06,
      "loss": 0.8534,
      "step": 670
    },
    {
      "epoch": 0.4131773399014778,
      "grad_norm": 0.9692283868789673,
      "learning_rate": 3.6342857142857145e-06,
      "loss": 0.8495,
      "step": 671
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 1.0863795280456543,
      "learning_rate": 3.6304761904761905e-06,
      "loss": 0.8617,
      "step": 672
    },
    {
      "epoch": 0.4144088669950739,
      "grad_norm": 1.1028242111206055,
      "learning_rate": 3.626666666666667e-06,
      "loss": 0.8604,
      "step": 673
    },
    {
      "epoch": 0.41502463054187194,
      "grad_norm": 1.018739938735962,
      "learning_rate": 3.622857142857143e-06,
      "loss": 0.8094,
      "step": 674
    },
    {
      "epoch": 0.41564039408866993,
      "grad_norm": 1.030184268951416,
      "learning_rate": 3.619047619047619e-06,
      "loss": 0.8125,
      "step": 675
    },
    {
      "epoch": 0.41625615763546797,
      "grad_norm": 1.0946388244628906,
      "learning_rate": 3.615238095238095e-06,
      "loss": 0.8482,
      "step": 676
    },
    {
      "epoch": 0.416871921182266,
      "grad_norm": 1.1192251443862915,
      "learning_rate": 3.611428571428572e-06,
      "loss": 0.881,
      "step": 677
    },
    {
      "epoch": 0.41748768472906406,
      "grad_norm": 0.9766618609428406,
      "learning_rate": 3.607619047619048e-06,
      "loss": 0.8889,
      "step": 678
    },
    {
      "epoch": 0.41810344827586204,
      "grad_norm": 1.0749006271362305,
      "learning_rate": 3.603809523809524e-06,
      "loss": 0.8272,
      "step": 679
    },
    {
      "epoch": 0.4187192118226601,
      "grad_norm": 1.029495358467102,
      "learning_rate": 3.6e-06,
      "loss": 0.8692,
      "step": 680
    },
    {
      "epoch": 0.41933497536945813,
      "grad_norm": 1.0686229467391968,
      "learning_rate": 3.596190476190476e-06,
      "loss": 0.8726,
      "step": 681
    },
    {
      "epoch": 0.41995073891625617,
      "grad_norm": 1.1688812971115112,
      "learning_rate": 3.5923809523809527e-06,
      "loss": 0.8774,
      "step": 682
    },
    {
      "epoch": 0.4205665024630542,
      "grad_norm": 1.0339192152023315,
      "learning_rate": 3.5885714285714287e-06,
      "loss": 0.8569,
      "step": 683
    },
    {
      "epoch": 0.4211822660098522,
      "grad_norm": 1.458644986152649,
      "learning_rate": 3.5847619047619047e-06,
      "loss": 0.8547,
      "step": 684
    },
    {
      "epoch": 0.42179802955665024,
      "grad_norm": 1.0142347812652588,
      "learning_rate": 3.580952380952381e-06,
      "loss": 0.8067,
      "step": 685
    },
    {
      "epoch": 0.4224137931034483,
      "grad_norm": 1.104541301727295,
      "learning_rate": 3.577142857142857e-06,
      "loss": 0.8637,
      "step": 686
    },
    {
      "epoch": 0.42302955665024633,
      "grad_norm": 1.122173547744751,
      "learning_rate": 3.573333333333333e-06,
      "loss": 0.9318,
      "step": 687
    },
    {
      "epoch": 0.4236453201970443,
      "grad_norm": 1.0356078147888184,
      "learning_rate": 3.56952380952381e-06,
      "loss": 0.8524,
      "step": 688
    },
    {
      "epoch": 0.42426108374384236,
      "grad_norm": 1.0378117561340332,
      "learning_rate": 3.565714285714286e-06,
      "loss": 0.8441,
      "step": 689
    },
    {
      "epoch": 0.4248768472906404,
      "grad_norm": 1.0197079181671143,
      "learning_rate": 3.561904761904762e-06,
      "loss": 0.814,
      "step": 690
    },
    {
      "epoch": 0.42549261083743845,
      "grad_norm": 1.0916557312011719,
      "learning_rate": 3.558095238095238e-06,
      "loss": 0.8096,
      "step": 691
    },
    {
      "epoch": 0.42610837438423643,
      "grad_norm": 1.0934640169143677,
      "learning_rate": 3.554285714285714e-06,
      "loss": 0.8992,
      "step": 692
    },
    {
      "epoch": 0.4267241379310345,
      "grad_norm": 1.0444587469100952,
      "learning_rate": 3.550476190476191e-06,
      "loss": 0.851,
      "step": 693
    },
    {
      "epoch": 0.4273399014778325,
      "grad_norm": 1.1221433877944946,
      "learning_rate": 3.546666666666667e-06,
      "loss": 0.8861,
      "step": 694
    },
    {
      "epoch": 0.42795566502463056,
      "grad_norm": 0.9346603155136108,
      "learning_rate": 3.542857142857143e-06,
      "loss": 0.8652,
      "step": 695
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 1.051917314529419,
      "learning_rate": 3.539047619047619e-06,
      "loss": 0.8716,
      "step": 696
    },
    {
      "epoch": 0.4291871921182266,
      "grad_norm": 1.0202256441116333,
      "learning_rate": 3.5352380952380953e-06,
      "loss": 0.8495,
      "step": 697
    },
    {
      "epoch": 0.42980295566502463,
      "grad_norm": 1.1154778003692627,
      "learning_rate": 3.5314285714285713e-06,
      "loss": 0.8365,
      "step": 698
    },
    {
      "epoch": 0.4304187192118227,
      "grad_norm": 1.0536282062530518,
      "learning_rate": 3.527619047619048e-06,
      "loss": 0.8237,
      "step": 699
    },
    {
      "epoch": 0.43103448275862066,
      "grad_norm": 1.0644415616989136,
      "learning_rate": 3.523809523809524e-06,
      "loss": 0.8117,
      "step": 700
    },
    {
      "epoch": 0.4316502463054187,
      "grad_norm": 1.0446839332580566,
      "learning_rate": 3.52e-06,
      "loss": 0.8704,
      "step": 701
    },
    {
      "epoch": 0.43226600985221675,
      "grad_norm": 1.1385530233383179,
      "learning_rate": 3.516190476190476e-06,
      "loss": 0.8477,
      "step": 702
    },
    {
      "epoch": 0.4328817733990148,
      "grad_norm": 1.1551405191421509,
      "learning_rate": 3.512380952380952e-06,
      "loss": 0.8715,
      "step": 703
    },
    {
      "epoch": 0.43349753694581283,
      "grad_norm": 1.1821926832199097,
      "learning_rate": 3.508571428571429e-06,
      "loss": 0.8791,
      "step": 704
    },
    {
      "epoch": 0.4341133004926108,
      "grad_norm": 1.0931240320205688,
      "learning_rate": 3.504761904761905e-06,
      "loss": 0.8333,
      "step": 705
    },
    {
      "epoch": 0.43472906403940886,
      "grad_norm": 1.0749316215515137,
      "learning_rate": 3.500952380952381e-06,
      "loss": 0.8585,
      "step": 706
    },
    {
      "epoch": 0.4353448275862069,
      "grad_norm": 1.0534937381744385,
      "learning_rate": 3.497142857142857e-06,
      "loss": 0.9004,
      "step": 707
    },
    {
      "epoch": 0.43596059113300495,
      "grad_norm": 1.0187636613845825,
      "learning_rate": 3.493333333333333e-06,
      "loss": 0.8303,
      "step": 708
    },
    {
      "epoch": 0.43657635467980294,
      "grad_norm": 0.9848130941390991,
      "learning_rate": 3.4895238095238095e-06,
      "loss": 0.8773,
      "step": 709
    },
    {
      "epoch": 0.437192118226601,
      "grad_norm": 1.1343554258346558,
      "learning_rate": 3.485714285714286e-06,
      "loss": 0.8653,
      "step": 710
    },
    {
      "epoch": 0.437807881773399,
      "grad_norm": 1.1065356731414795,
      "learning_rate": 3.4819047619047623e-06,
      "loss": 0.8664,
      "step": 711
    },
    {
      "epoch": 0.43842364532019706,
      "grad_norm": 1.0179922580718994,
      "learning_rate": 3.4780952380952384e-06,
      "loss": 0.8671,
      "step": 712
    },
    {
      "epoch": 0.43903940886699505,
      "grad_norm": 1.05556321144104,
      "learning_rate": 3.4742857142857144e-06,
      "loss": 0.8434,
      "step": 713
    },
    {
      "epoch": 0.4396551724137931,
      "grad_norm": 0.9762723445892334,
      "learning_rate": 3.4704761904761904e-06,
      "loss": 0.8714,
      "step": 714
    },
    {
      "epoch": 0.44027093596059114,
      "grad_norm": 1.0573965311050415,
      "learning_rate": 3.4666666666666664e-06,
      "loss": 0.8588,
      "step": 715
    },
    {
      "epoch": 0.4408866995073892,
      "grad_norm": 1.0761876106262207,
      "learning_rate": 3.4628571428571432e-06,
      "loss": 0.8467,
      "step": 716
    },
    {
      "epoch": 0.44150246305418717,
      "grad_norm": 1.076746940612793,
      "learning_rate": 3.4590476190476192e-06,
      "loss": 0.8756,
      "step": 717
    },
    {
      "epoch": 0.4421182266009852,
      "grad_norm": 1.0588257312774658,
      "learning_rate": 3.4552380952380952e-06,
      "loss": 0.8809,
      "step": 718
    },
    {
      "epoch": 0.44273399014778325,
      "grad_norm": 1.0639344453811646,
      "learning_rate": 3.4514285714285712e-06,
      "loss": 0.8036,
      "step": 719
    },
    {
      "epoch": 0.4433497536945813,
      "grad_norm": 1.0324206352233887,
      "learning_rate": 3.4476190476190472e-06,
      "loss": 0.8573,
      "step": 720
    },
    {
      "epoch": 0.44396551724137934,
      "grad_norm": 1.0766931772232056,
      "learning_rate": 3.443809523809524e-06,
      "loss": 0.838,
      "step": 721
    },
    {
      "epoch": 0.4445812807881773,
      "grad_norm": 1.1031827926635742,
      "learning_rate": 3.44e-06,
      "loss": 0.8577,
      "step": 722
    },
    {
      "epoch": 0.44519704433497537,
      "grad_norm": 1.142377257347107,
      "learning_rate": 3.4361904761904765e-06,
      "loss": 0.8407,
      "step": 723
    },
    {
      "epoch": 0.4458128078817734,
      "grad_norm": 1.0569690465927124,
      "learning_rate": 3.4323809523809525e-06,
      "loss": 0.8407,
      "step": 724
    },
    {
      "epoch": 0.44642857142857145,
      "grad_norm": 1.0069644451141357,
      "learning_rate": 3.4285714285714285e-06,
      "loss": 0.8849,
      "step": 725
    },
    {
      "epoch": 0.44704433497536944,
      "grad_norm": 1.023883581161499,
      "learning_rate": 3.4247619047619045e-06,
      "loss": 0.8464,
      "step": 726
    },
    {
      "epoch": 0.4476600985221675,
      "grad_norm": 1.0199867486953735,
      "learning_rate": 3.4209523809523814e-06,
      "loss": 0.857,
      "step": 727
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 1.0830762386322021,
      "learning_rate": 3.4171428571428574e-06,
      "loss": 0.8588,
      "step": 728
    },
    {
      "epoch": 0.44889162561576357,
      "grad_norm": 1.1007062196731567,
      "learning_rate": 3.4133333333333334e-06,
      "loss": 0.8443,
      "step": 729
    },
    {
      "epoch": 0.44950738916256155,
      "grad_norm": 0.9997031092643738,
      "learning_rate": 3.4095238095238094e-06,
      "loss": 0.8665,
      "step": 730
    },
    {
      "epoch": 0.4501231527093596,
      "grad_norm": 1.1003788709640503,
      "learning_rate": 3.4057142857142854e-06,
      "loss": 0.8618,
      "step": 731
    },
    {
      "epoch": 0.45073891625615764,
      "grad_norm": 1.0441685914993286,
      "learning_rate": 3.4019047619047623e-06,
      "loss": 0.8719,
      "step": 732
    },
    {
      "epoch": 0.4513546798029557,
      "grad_norm": 1.0472732782363892,
      "learning_rate": 3.3980952380952383e-06,
      "loss": 0.8631,
      "step": 733
    },
    {
      "epoch": 0.45197044334975367,
      "grad_norm": 1.1157914400100708,
      "learning_rate": 3.3942857142857143e-06,
      "loss": 0.8458,
      "step": 734
    },
    {
      "epoch": 0.4525862068965517,
      "grad_norm": 1.0497125387191772,
      "learning_rate": 3.3904761904761907e-06,
      "loss": 0.8766,
      "step": 735
    },
    {
      "epoch": 0.45320197044334976,
      "grad_norm": 1.1049699783325195,
      "learning_rate": 3.3866666666666667e-06,
      "loss": 0.8481,
      "step": 736
    },
    {
      "epoch": 0.4538177339901478,
      "grad_norm": 1.0551190376281738,
      "learning_rate": 3.3828571428571427e-06,
      "loss": 0.8336,
      "step": 737
    },
    {
      "epoch": 0.4544334975369458,
      "grad_norm": 1.0839823484420776,
      "learning_rate": 3.3790476190476196e-06,
      "loss": 0.8195,
      "step": 738
    },
    {
      "epoch": 0.45504926108374383,
      "grad_norm": 1.135833978652954,
      "learning_rate": 3.3752380952380956e-06,
      "loss": 0.8588,
      "step": 739
    },
    {
      "epoch": 0.45566502463054187,
      "grad_norm": 1.0465127229690552,
      "learning_rate": 3.3714285714285716e-06,
      "loss": 0.862,
      "step": 740
    },
    {
      "epoch": 0.4562807881773399,
      "grad_norm": 1.1093614101409912,
      "learning_rate": 3.3676190476190476e-06,
      "loss": 0.8268,
      "step": 741
    },
    {
      "epoch": 0.45689655172413796,
      "grad_norm": 1.1063597202301025,
      "learning_rate": 3.3638095238095236e-06,
      "loss": 0.835,
      "step": 742
    },
    {
      "epoch": 0.45751231527093594,
      "grad_norm": 0.9655569791793823,
      "learning_rate": 3.3600000000000004e-06,
      "loss": 0.8269,
      "step": 743
    },
    {
      "epoch": 0.458128078817734,
      "grad_norm": 1.0791382789611816,
      "learning_rate": 3.3561904761904764e-06,
      "loss": 0.8167,
      "step": 744
    },
    {
      "epoch": 0.45874384236453203,
      "grad_norm": 1.1727373600006104,
      "learning_rate": 3.3523809523809525e-06,
      "loss": 0.8856,
      "step": 745
    },
    {
      "epoch": 0.45935960591133007,
      "grad_norm": 1.0931748151779175,
      "learning_rate": 3.3485714285714285e-06,
      "loss": 0.8321,
      "step": 746
    },
    {
      "epoch": 0.45997536945812806,
      "grad_norm": 0.9513764977455139,
      "learning_rate": 3.344761904761905e-06,
      "loss": 0.8557,
      "step": 747
    },
    {
      "epoch": 0.4605911330049261,
      "grad_norm": 0.9480428099632263,
      "learning_rate": 3.340952380952381e-06,
      "loss": 0.8603,
      "step": 748
    },
    {
      "epoch": 0.46120689655172414,
      "grad_norm": 1.0395598411560059,
      "learning_rate": 3.3371428571428573e-06,
      "loss": 0.9026,
      "step": 749
    },
    {
      "epoch": 0.4618226600985222,
      "grad_norm": 1.0375797748565674,
      "learning_rate": 3.3333333333333337e-06,
      "loss": 0.7981,
      "step": 750
    },
    {
      "epoch": 0.4624384236453202,
      "grad_norm": 2.086379051208496,
      "learning_rate": 3.3295238095238098e-06,
      "loss": 0.8201,
      "step": 751
    },
    {
      "epoch": 0.4630541871921182,
      "grad_norm": 1.0373835563659668,
      "learning_rate": 3.3257142857142858e-06,
      "loss": 0.7902,
      "step": 752
    },
    {
      "epoch": 0.46366995073891626,
      "grad_norm": 1.1491589546203613,
      "learning_rate": 3.3219047619047618e-06,
      "loss": 0.8376,
      "step": 753
    },
    {
      "epoch": 0.4642857142857143,
      "grad_norm": 1.0457746982574463,
      "learning_rate": 3.3180952380952378e-06,
      "loss": 0.8146,
      "step": 754
    },
    {
      "epoch": 0.4649014778325123,
      "grad_norm": 1.0036338567733765,
      "learning_rate": 3.3142857142857146e-06,
      "loss": 0.8729,
      "step": 755
    },
    {
      "epoch": 0.46551724137931033,
      "grad_norm": 1.1319177150726318,
      "learning_rate": 3.3104761904761906e-06,
      "loss": 0.8256,
      "step": 756
    },
    {
      "epoch": 0.4661330049261084,
      "grad_norm": 1.2203702926635742,
      "learning_rate": 3.3066666666666666e-06,
      "loss": 0.8891,
      "step": 757
    },
    {
      "epoch": 0.4667487684729064,
      "grad_norm": 1.0205421447753906,
      "learning_rate": 3.3028571428571426e-06,
      "loss": 0.8686,
      "step": 758
    },
    {
      "epoch": 0.46736453201970446,
      "grad_norm": 1.047969937324524,
      "learning_rate": 3.299047619047619e-06,
      "loss": 0.8446,
      "step": 759
    },
    {
      "epoch": 0.46798029556650245,
      "grad_norm": 1.065043568611145,
      "learning_rate": 3.2952380952380955e-06,
      "loss": 0.8228,
      "step": 760
    },
    {
      "epoch": 0.4685960591133005,
      "grad_norm": 1.0562912225723267,
      "learning_rate": 3.2914285714285715e-06,
      "loss": 0.8229,
      "step": 761
    },
    {
      "epoch": 0.46921182266009853,
      "grad_norm": 1.1790891885757446,
      "learning_rate": 3.287619047619048e-06,
      "loss": 0.8791,
      "step": 762
    },
    {
      "epoch": 0.4698275862068966,
      "grad_norm": 1.101363182067871,
      "learning_rate": 3.283809523809524e-06,
      "loss": 0.8475,
      "step": 763
    },
    {
      "epoch": 0.47044334975369456,
      "grad_norm": 1.0715439319610596,
      "learning_rate": 3.28e-06,
      "loss": 0.8051,
      "step": 764
    },
    {
      "epoch": 0.4710591133004926,
      "grad_norm": 0.9670822024345398,
      "learning_rate": 3.276190476190476e-06,
      "loss": 0.8438,
      "step": 765
    },
    {
      "epoch": 0.47167487684729065,
      "grad_norm": 1.0746891498565674,
      "learning_rate": 3.272380952380953e-06,
      "loss": 0.8838,
      "step": 766
    },
    {
      "epoch": 0.4722906403940887,
      "grad_norm": 1.0948419570922852,
      "learning_rate": 3.268571428571429e-06,
      "loss": 0.8208,
      "step": 767
    },
    {
      "epoch": 0.4729064039408867,
      "grad_norm": 1.0149691104888916,
      "learning_rate": 3.264761904761905e-06,
      "loss": 0.8268,
      "step": 768
    },
    {
      "epoch": 0.4735221674876847,
      "grad_norm": 1.074938416481018,
      "learning_rate": 3.260952380952381e-06,
      "loss": 0.8744,
      "step": 769
    },
    {
      "epoch": 0.47413793103448276,
      "grad_norm": 1.073037028312683,
      "learning_rate": 3.257142857142857e-06,
      "loss": 0.8665,
      "step": 770
    },
    {
      "epoch": 0.4747536945812808,
      "grad_norm": 1.0629936456680298,
      "learning_rate": 3.2533333333333337e-06,
      "loss": 0.8382,
      "step": 771
    },
    {
      "epoch": 0.4753694581280788,
      "grad_norm": 1.1785749197006226,
      "learning_rate": 3.2495238095238097e-06,
      "loss": 0.8437,
      "step": 772
    },
    {
      "epoch": 0.47598522167487683,
      "grad_norm": 1.0811126232147217,
      "learning_rate": 3.2457142857142857e-06,
      "loss": 0.8378,
      "step": 773
    },
    {
      "epoch": 0.4766009852216749,
      "grad_norm": 1.0233824253082275,
      "learning_rate": 3.241904761904762e-06,
      "loss": 0.8659,
      "step": 774
    },
    {
      "epoch": 0.4772167487684729,
      "grad_norm": 1.113381266593933,
      "learning_rate": 3.238095238095238e-06,
      "loss": 0.9059,
      "step": 775
    },
    {
      "epoch": 0.47783251231527096,
      "grad_norm": 1.110066533088684,
      "learning_rate": 3.234285714285714e-06,
      "loss": 0.8456,
      "step": 776
    },
    {
      "epoch": 0.47844827586206895,
      "grad_norm": 1.160791277885437,
      "learning_rate": 3.230476190476191e-06,
      "loss": 0.8611,
      "step": 777
    },
    {
      "epoch": 0.479064039408867,
      "grad_norm": 1.1111358404159546,
      "learning_rate": 3.226666666666667e-06,
      "loss": 0.8706,
      "step": 778
    },
    {
      "epoch": 0.47967980295566504,
      "grad_norm": 1.0240800380706787,
      "learning_rate": 3.222857142857143e-06,
      "loss": 0.829,
      "step": 779
    },
    {
      "epoch": 0.4802955665024631,
      "grad_norm": 1.088821530342102,
      "learning_rate": 3.219047619047619e-06,
      "loss": 0.8356,
      "step": 780
    },
    {
      "epoch": 0.48091133004926107,
      "grad_norm": 1.0253431797027588,
      "learning_rate": 3.215238095238095e-06,
      "loss": 0.831,
      "step": 781
    },
    {
      "epoch": 0.4815270935960591,
      "grad_norm": 1.1937546730041504,
      "learning_rate": 3.211428571428572e-06,
      "loss": 0.8947,
      "step": 782
    },
    {
      "epoch": 0.48214285714285715,
      "grad_norm": 1.00270414352417,
      "learning_rate": 3.207619047619048e-06,
      "loss": 0.8298,
      "step": 783
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 1.0591520071029663,
      "learning_rate": 3.203809523809524e-06,
      "loss": 0.8363,
      "step": 784
    },
    {
      "epoch": 0.4833743842364532,
      "grad_norm": 1.0177632570266724,
      "learning_rate": 3.2e-06,
      "loss": 0.84,
      "step": 785
    },
    {
      "epoch": 0.4839901477832512,
      "grad_norm": 1.070509672164917,
      "learning_rate": 3.1961904761904763e-06,
      "loss": 0.9002,
      "step": 786
    },
    {
      "epoch": 0.48460591133004927,
      "grad_norm": 1.1278901100158691,
      "learning_rate": 3.1923809523809523e-06,
      "loss": 0.8271,
      "step": 787
    },
    {
      "epoch": 0.4852216748768473,
      "grad_norm": 0.9520243406295776,
      "learning_rate": 3.1885714285714287e-06,
      "loss": 0.814,
      "step": 788
    },
    {
      "epoch": 0.4858374384236453,
      "grad_norm": 1.0565284490585327,
      "learning_rate": 3.184761904761905e-06,
      "loss": 0.8386,
      "step": 789
    },
    {
      "epoch": 0.48645320197044334,
      "grad_norm": 0.9762368202209473,
      "learning_rate": 3.180952380952381e-06,
      "loss": 0.8547,
      "step": 790
    },
    {
      "epoch": 0.4870689655172414,
      "grad_norm": 1.1626652479171753,
      "learning_rate": 3.177142857142857e-06,
      "loss": 0.8511,
      "step": 791
    },
    {
      "epoch": 0.4876847290640394,
      "grad_norm": 1.0576441287994385,
      "learning_rate": 3.173333333333333e-06,
      "loss": 0.914,
      "step": 792
    },
    {
      "epoch": 0.4883004926108374,
      "grad_norm": 1.0417972803115845,
      "learning_rate": 3.16952380952381e-06,
      "loss": 0.8252,
      "step": 793
    },
    {
      "epoch": 0.48891625615763545,
      "grad_norm": 1.0664061307907104,
      "learning_rate": 3.165714285714286e-06,
      "loss": 0.8682,
      "step": 794
    },
    {
      "epoch": 0.4895320197044335,
      "grad_norm": 1.0283581018447876,
      "learning_rate": 3.161904761904762e-06,
      "loss": 0.8818,
      "step": 795
    },
    {
      "epoch": 0.49014778325123154,
      "grad_norm": 1.0520821809768677,
      "learning_rate": 3.158095238095238e-06,
      "loss": 0.8588,
      "step": 796
    },
    {
      "epoch": 0.4907635467980296,
      "grad_norm": 0.9605892300605774,
      "learning_rate": 3.154285714285714e-06,
      "loss": 0.8222,
      "step": 797
    },
    {
      "epoch": 0.49137931034482757,
      "grad_norm": 1.0501517057418823,
      "learning_rate": 3.1504761904761905e-06,
      "loss": 0.8494,
      "step": 798
    },
    {
      "epoch": 0.4919950738916256,
      "grad_norm": 1.0612753629684448,
      "learning_rate": 3.146666666666667e-06,
      "loss": 0.8511,
      "step": 799
    },
    {
      "epoch": 0.49261083743842365,
      "grad_norm": 0.9731685519218445,
      "learning_rate": 3.142857142857143e-06,
      "loss": 0.8604,
      "step": 800
    },
    {
      "epoch": 0.4932266009852217,
      "grad_norm": 0.9665550589561462,
      "learning_rate": 3.1390476190476193e-06,
      "loss": 0.8949,
      "step": 801
    },
    {
      "epoch": 0.4938423645320197,
      "grad_norm": 1.094528079032898,
      "learning_rate": 3.1352380952380953e-06,
      "loss": 0.8505,
      "step": 802
    },
    {
      "epoch": 0.4944581280788177,
      "grad_norm": 1.056065320968628,
      "learning_rate": 3.1314285714285713e-06,
      "loss": 0.8512,
      "step": 803
    },
    {
      "epoch": 0.49507389162561577,
      "grad_norm": 1.0397143363952637,
      "learning_rate": 3.1276190476190473e-06,
      "loss": 0.8537,
      "step": 804
    },
    {
      "epoch": 0.4956896551724138,
      "grad_norm": 0.984984815120697,
      "learning_rate": 3.123809523809524e-06,
      "loss": 0.893,
      "step": 805
    },
    {
      "epoch": 0.4963054187192118,
      "grad_norm": 1.0411816835403442,
      "learning_rate": 3.12e-06,
      "loss": 0.8413,
      "step": 806
    },
    {
      "epoch": 0.49692118226600984,
      "grad_norm": 1.021840214729309,
      "learning_rate": 3.116190476190476e-06,
      "loss": 0.8202,
      "step": 807
    },
    {
      "epoch": 0.4975369458128079,
      "grad_norm": 1.0008939504623413,
      "learning_rate": 3.112380952380952e-06,
      "loss": 0.8353,
      "step": 808
    },
    {
      "epoch": 0.4981527093596059,
      "grad_norm": 1.0451992750167847,
      "learning_rate": 3.1085714285714282e-06,
      "loss": 0.8515,
      "step": 809
    },
    {
      "epoch": 0.4987684729064039,
      "grad_norm": 1.0350042581558228,
      "learning_rate": 3.104761904761905e-06,
      "loss": 0.8781,
      "step": 810
    },
    {
      "epoch": 0.49938423645320196,
      "grad_norm": 1.0541698932647705,
      "learning_rate": 3.100952380952381e-06,
      "loss": 0.7967,
      "step": 811
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1662479639053345,
      "learning_rate": 3.097142857142857e-06,
      "loss": 0.8544,
      "step": 812
    },
    {
      "epoch": 0.5,
      "eval_loss": 0.8787217736244202,
      "eval_runtime": 268.3309,
      "eval_samples_per_second": 86.125,
      "eval_steps_per_second": 2.694,
      "step": 812
    },
    {
      "epoch": 0.500615763546798,
      "grad_norm": 1.0439629554748535,
      "learning_rate": 3.0933333333333335e-06,
      "loss": 0.8874,
      "step": 813
    },
    {
      "epoch": 0.5012315270935961,
      "grad_norm": 1.0212533473968506,
      "learning_rate": 3.0895238095238095e-06,
      "loss": 0.8559,
      "step": 814
    },
    {
      "epoch": 0.5018472906403941,
      "grad_norm": 0.9897382855415344,
      "learning_rate": 3.0857142857142855e-06,
      "loss": 0.8293,
      "step": 815
    },
    {
      "epoch": 0.5024630541871922,
      "grad_norm": 0.9111842513084412,
      "learning_rate": 3.0819047619047624e-06,
      "loss": 0.8349,
      "step": 816
    },
    {
      "epoch": 0.5030788177339901,
      "grad_norm": 1.0861130952835083,
      "learning_rate": 3.0780952380952384e-06,
      "loss": 0.816,
      "step": 817
    },
    {
      "epoch": 0.5036945812807881,
      "grad_norm": 1.1129403114318848,
      "learning_rate": 3.0742857142857144e-06,
      "loss": 0.8476,
      "step": 818
    },
    {
      "epoch": 0.5043103448275862,
      "grad_norm": 1.0409799814224243,
      "learning_rate": 3.0704761904761904e-06,
      "loss": 0.8396,
      "step": 819
    },
    {
      "epoch": 0.5049261083743842,
      "grad_norm": 1.0135536193847656,
      "learning_rate": 3.0666666666666664e-06,
      "loss": 0.8969,
      "step": 820
    },
    {
      "epoch": 0.5055418719211823,
      "grad_norm": 1.051710605621338,
      "learning_rate": 3.0628571428571432e-06,
      "loss": 0.8237,
      "step": 821
    },
    {
      "epoch": 0.5061576354679803,
      "grad_norm": 1.0663784742355347,
      "learning_rate": 3.0590476190476192e-06,
      "loss": 0.8098,
      "step": 822
    },
    {
      "epoch": 0.5067733990147784,
      "grad_norm": 1.096837043762207,
      "learning_rate": 3.0552380952380952e-06,
      "loss": 0.8535,
      "step": 823
    },
    {
      "epoch": 0.5073891625615764,
      "grad_norm": 0.9402517080307007,
      "learning_rate": 3.0514285714285713e-06,
      "loss": 0.8714,
      "step": 824
    },
    {
      "epoch": 0.5080049261083743,
      "grad_norm": 1.033381462097168,
      "learning_rate": 3.0476190476190477e-06,
      "loss": 0.8707,
      "step": 825
    },
    {
      "epoch": 0.5086206896551724,
      "grad_norm": 0.9798709154129028,
      "learning_rate": 3.0438095238095237e-06,
      "loss": 0.8562,
      "step": 826
    },
    {
      "epoch": 0.5092364532019704,
      "grad_norm": 1.0839507579803467,
      "learning_rate": 3.0400000000000005e-06,
      "loss": 0.82,
      "step": 827
    },
    {
      "epoch": 0.5098522167487685,
      "grad_norm": 1.0627403259277344,
      "learning_rate": 3.0361904761904765e-06,
      "loss": 0.8496,
      "step": 828
    },
    {
      "epoch": 0.5104679802955665,
      "grad_norm": 0.9222130179405212,
      "learning_rate": 3.0323809523809526e-06,
      "loss": 0.831,
      "step": 829
    },
    {
      "epoch": 0.5110837438423645,
      "grad_norm": 0.9685607552528381,
      "learning_rate": 3.0285714285714286e-06,
      "loss": 0.8541,
      "step": 830
    },
    {
      "epoch": 0.5116995073891626,
      "grad_norm": 1.1866211891174316,
      "learning_rate": 3.0247619047619046e-06,
      "loss": 0.8437,
      "step": 831
    },
    {
      "epoch": 0.5123152709359606,
      "grad_norm": 1.0553767681121826,
      "learning_rate": 3.0209523809523814e-06,
      "loss": 0.8297,
      "step": 832
    },
    {
      "epoch": 0.5129310344827587,
      "grad_norm": 1.0751044750213623,
      "learning_rate": 3.0171428571428574e-06,
      "loss": 0.8386,
      "step": 833
    },
    {
      "epoch": 0.5135467980295566,
      "grad_norm": 1.002668857574463,
      "learning_rate": 3.0133333333333334e-06,
      "loss": 0.8181,
      "step": 834
    },
    {
      "epoch": 0.5141625615763546,
      "grad_norm": 1.1063146591186523,
      "learning_rate": 3.0095238095238094e-06,
      "loss": 0.8297,
      "step": 835
    },
    {
      "epoch": 0.5147783251231527,
      "grad_norm": 1.0365573167800903,
      "learning_rate": 3.0057142857142854e-06,
      "loss": 0.8545,
      "step": 836
    },
    {
      "epoch": 0.5153940886699507,
      "grad_norm": 0.9341920614242554,
      "learning_rate": 3.001904761904762e-06,
      "loss": 0.8734,
      "step": 837
    },
    {
      "epoch": 0.5160098522167488,
      "grad_norm": 1.0286774635314941,
      "learning_rate": 2.9980952380952383e-06,
      "loss": 0.8316,
      "step": 838
    },
    {
      "epoch": 0.5166256157635468,
      "grad_norm": 1.1247801780700684,
      "learning_rate": 2.9942857142857147e-06,
      "loss": 0.8583,
      "step": 839
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 1.0262326002120972,
      "learning_rate": 2.9904761904761907e-06,
      "loss": 0.887,
      "step": 840
    },
    {
      "epoch": 0.5178571428571429,
      "grad_norm": 1.1662551164627075,
      "learning_rate": 2.9866666666666667e-06,
      "loss": 0.8769,
      "step": 841
    },
    {
      "epoch": 0.5184729064039408,
      "grad_norm": 0.9902069568634033,
      "learning_rate": 2.982857142857143e-06,
      "loss": 0.8544,
      "step": 842
    },
    {
      "epoch": 0.5190886699507389,
      "grad_norm": 1.037928581237793,
      "learning_rate": 2.979047619047619e-06,
      "loss": 0.8717,
      "step": 843
    },
    {
      "epoch": 0.5197044334975369,
      "grad_norm": 1.0197633504867554,
      "learning_rate": 2.975238095238095e-06,
      "loss": 0.8188,
      "step": 844
    },
    {
      "epoch": 0.520320197044335,
      "grad_norm": 1.0952281951904297,
      "learning_rate": 2.9714285714285716e-06,
      "loss": 0.8009,
      "step": 845
    },
    {
      "epoch": 0.520935960591133,
      "grad_norm": 1.065794587135315,
      "learning_rate": 2.9676190476190476e-06,
      "loss": 0.8574,
      "step": 846
    },
    {
      "epoch": 0.521551724137931,
      "grad_norm": 0.9816839098930359,
      "learning_rate": 2.9638095238095236e-06,
      "loss": 0.8338,
      "step": 847
    },
    {
      "epoch": 0.5221674876847291,
      "grad_norm": 0.9973509907722473,
      "learning_rate": 2.96e-06,
      "loss": 0.8463,
      "step": 848
    },
    {
      "epoch": 0.5227832512315271,
      "grad_norm": 0.9568844437599182,
      "learning_rate": 2.956190476190476e-06,
      "loss": 0.8523,
      "step": 849
    },
    {
      "epoch": 0.5233990147783252,
      "grad_norm": 0.9187963604927063,
      "learning_rate": 2.9523809523809525e-06,
      "loss": 0.8031,
      "step": 850
    },
    {
      "epoch": 0.5240147783251231,
      "grad_norm": 0.9772159457206726,
      "learning_rate": 2.948571428571429e-06,
      "loss": 0.8389,
      "step": 851
    },
    {
      "epoch": 0.5246305418719212,
      "grad_norm": 1.0649089813232422,
      "learning_rate": 2.944761904761905e-06,
      "loss": 0.8959,
      "step": 852
    },
    {
      "epoch": 0.5252463054187192,
      "grad_norm": 1.1267566680908203,
      "learning_rate": 2.9409523809523813e-06,
      "loss": 0.9156,
      "step": 853
    },
    {
      "epoch": 0.5258620689655172,
      "grad_norm": 1.0313090085983276,
      "learning_rate": 2.9371428571428573e-06,
      "loss": 0.8688,
      "step": 854
    },
    {
      "epoch": 0.5264778325123153,
      "grad_norm": 1.0342624187469482,
      "learning_rate": 2.9333333333333333e-06,
      "loss": 0.8641,
      "step": 855
    },
    {
      "epoch": 0.5270935960591133,
      "grad_norm": 1.0217267274856567,
      "learning_rate": 2.9295238095238098e-06,
      "loss": 0.8446,
      "step": 856
    },
    {
      "epoch": 0.5277093596059114,
      "grad_norm": 0.9402487874031067,
      "learning_rate": 2.9257142857142858e-06,
      "loss": 0.8286,
      "step": 857
    },
    {
      "epoch": 0.5283251231527094,
      "grad_norm": 0.9692648649215698,
      "learning_rate": 2.9219047619047618e-06,
      "loss": 0.8242,
      "step": 858
    },
    {
      "epoch": 0.5289408866995073,
      "grad_norm": 1.133242130279541,
      "learning_rate": 2.918095238095238e-06,
      "loss": 0.8375,
      "step": 859
    },
    {
      "epoch": 0.5295566502463054,
      "grad_norm": 0.9990291595458984,
      "learning_rate": 2.9142857142857142e-06,
      "loss": 0.8538,
      "step": 860
    },
    {
      "epoch": 0.5301724137931034,
      "grad_norm": 0.9985235929489136,
      "learning_rate": 2.9104761904761902e-06,
      "loss": 0.8767,
      "step": 861
    },
    {
      "epoch": 0.5307881773399015,
      "grad_norm": 1.035129427909851,
      "learning_rate": 2.9066666666666666e-06,
      "loss": 0.8396,
      "step": 862
    },
    {
      "epoch": 0.5314039408866995,
      "grad_norm": 1.0639305114746094,
      "learning_rate": 2.902857142857143e-06,
      "loss": 0.8478,
      "step": 863
    },
    {
      "epoch": 0.5320197044334976,
      "grad_norm": 1.0382306575775146,
      "learning_rate": 2.899047619047619e-06,
      "loss": 0.8152,
      "step": 864
    },
    {
      "epoch": 0.5326354679802956,
      "grad_norm": 1.0769697427749634,
      "learning_rate": 2.8952380952380955e-06,
      "loss": 0.8618,
      "step": 865
    },
    {
      "epoch": 0.5332512315270936,
      "grad_norm": 1.0110454559326172,
      "learning_rate": 2.8914285714285715e-06,
      "loss": 0.8538,
      "step": 866
    },
    {
      "epoch": 0.5338669950738916,
      "grad_norm": 1.0384002923965454,
      "learning_rate": 2.887619047619048e-06,
      "loss": 0.8425,
      "step": 867
    },
    {
      "epoch": 0.5344827586206896,
      "grad_norm": 0.9902776479721069,
      "learning_rate": 2.883809523809524e-06,
      "loss": 0.8717,
      "step": 868
    },
    {
      "epoch": 0.5350985221674877,
      "grad_norm": 1.038712978363037,
      "learning_rate": 2.88e-06,
      "loss": 0.8348,
      "step": 869
    },
    {
      "epoch": 0.5357142857142857,
      "grad_norm": 1.1228604316711426,
      "learning_rate": 2.8761904761904764e-06,
      "loss": 0.8309,
      "step": 870
    },
    {
      "epoch": 0.5363300492610837,
      "grad_norm": 0.9380071759223938,
      "learning_rate": 2.8723809523809524e-06,
      "loss": 0.8566,
      "step": 871
    },
    {
      "epoch": 0.5369458128078818,
      "grad_norm": 1.1962231397628784,
      "learning_rate": 2.8685714285714284e-06,
      "loss": 0.8498,
      "step": 872
    },
    {
      "epoch": 0.5375615763546798,
      "grad_norm": 1.0597563982009888,
      "learning_rate": 2.864761904761905e-06,
      "loss": 0.8487,
      "step": 873
    },
    {
      "epoch": 0.5381773399014779,
      "grad_norm": 0.9821975231170654,
      "learning_rate": 2.860952380952381e-06,
      "loss": 0.8436,
      "step": 874
    },
    {
      "epoch": 0.5387931034482759,
      "grad_norm": 0.9423992037773132,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.853,
      "step": 875
    },
    {
      "epoch": 0.5394088669950738,
      "grad_norm": 1.0471032857894897,
      "learning_rate": 2.8533333333333333e-06,
      "loss": 0.8576,
      "step": 876
    },
    {
      "epoch": 0.5400246305418719,
      "grad_norm": 1.0625181198120117,
      "learning_rate": 2.8495238095238097e-06,
      "loss": 0.8308,
      "step": 877
    },
    {
      "epoch": 0.5406403940886699,
      "grad_norm": 1.0755056142807007,
      "learning_rate": 2.845714285714286e-06,
      "loss": 0.8622,
      "step": 878
    },
    {
      "epoch": 0.541256157635468,
      "grad_norm": 1.059847354888916,
      "learning_rate": 2.841904761904762e-06,
      "loss": 0.8936,
      "step": 879
    },
    {
      "epoch": 0.541871921182266,
      "grad_norm": 1.1054593324661255,
      "learning_rate": 2.838095238095238e-06,
      "loss": 0.863,
      "step": 880
    },
    {
      "epoch": 0.5424876847290641,
      "grad_norm": 1.0562098026275635,
      "learning_rate": 2.8342857142857146e-06,
      "loss": 0.8505,
      "step": 881
    },
    {
      "epoch": 0.5431034482758621,
      "grad_norm": 1.057468056678772,
      "learning_rate": 2.8304761904761906e-06,
      "loss": 0.8339,
      "step": 882
    },
    {
      "epoch": 0.5437192118226601,
      "grad_norm": 1.0910779237747192,
      "learning_rate": 2.8266666666666666e-06,
      "loss": 0.8745,
      "step": 883
    },
    {
      "epoch": 0.5443349753694581,
      "grad_norm": 0.9054524302482605,
      "learning_rate": 2.822857142857143e-06,
      "loss": 0.8653,
      "step": 884
    },
    {
      "epoch": 0.5449507389162561,
      "grad_norm": 0.9769971370697021,
      "learning_rate": 2.819047619047619e-06,
      "loss": 0.839,
      "step": 885
    },
    {
      "epoch": 0.5455665024630542,
      "grad_norm": 0.977857768535614,
      "learning_rate": 2.815238095238095e-06,
      "loss": 0.8867,
      "step": 886
    },
    {
      "epoch": 0.5461822660098522,
      "grad_norm": 1.0129292011260986,
      "learning_rate": 2.8114285714285714e-06,
      "loss": 0.789,
      "step": 887
    },
    {
      "epoch": 0.5467980295566502,
      "grad_norm": 0.9860545992851257,
      "learning_rate": 2.8076190476190474e-06,
      "loss": 0.8198,
      "step": 888
    },
    {
      "epoch": 0.5474137931034483,
      "grad_norm": 1.0507599115371704,
      "learning_rate": 2.803809523809524e-06,
      "loss": 0.8408,
      "step": 889
    },
    {
      "epoch": 0.5480295566502463,
      "grad_norm": 1.2953475713729858,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.8296,
      "step": 890
    },
    {
      "epoch": 0.5486453201970444,
      "grad_norm": 1.0075750350952148,
      "learning_rate": 2.7961904761904763e-06,
      "loss": 0.8416,
      "step": 891
    },
    {
      "epoch": 0.5492610837438424,
      "grad_norm": 1.0231502056121826,
      "learning_rate": 2.7923809523809527e-06,
      "loss": 0.8823,
      "step": 892
    },
    {
      "epoch": 0.5498768472906403,
      "grad_norm": 0.9733885526657104,
      "learning_rate": 2.7885714285714287e-06,
      "loss": 0.8489,
      "step": 893
    },
    {
      "epoch": 0.5504926108374384,
      "grad_norm": 1.0107492208480835,
      "learning_rate": 2.7847619047619047e-06,
      "loss": 0.871,
      "step": 894
    },
    {
      "epoch": 0.5511083743842364,
      "grad_norm": 1.0766600370407104,
      "learning_rate": 2.780952380952381e-06,
      "loss": 0.8272,
      "step": 895
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 1.0034263134002686,
      "learning_rate": 2.777142857142857e-06,
      "loss": 0.8385,
      "step": 896
    },
    {
      "epoch": 0.5523399014778325,
      "grad_norm": 1.0722439289093018,
      "learning_rate": 2.773333333333333e-06,
      "loss": 0.8575,
      "step": 897
    },
    {
      "epoch": 0.5529556650246306,
      "grad_norm": 0.9719768762588501,
      "learning_rate": 2.7695238095238096e-06,
      "loss": 0.8413,
      "step": 898
    },
    {
      "epoch": 0.5535714285714286,
      "grad_norm": 0.9946536421775818,
      "learning_rate": 2.7657142857142856e-06,
      "loss": 0.8523,
      "step": 899
    },
    {
      "epoch": 0.5541871921182266,
      "grad_norm": 1.0527000427246094,
      "learning_rate": 2.7619047619047616e-06,
      "loss": 0.8488,
      "step": 900
    },
    {
      "epoch": 0.5548029556650246,
      "grad_norm": 1.0085550546646118,
      "learning_rate": 2.758095238095238e-06,
      "loss": 0.8284,
      "step": 901
    },
    {
      "epoch": 0.5554187192118226,
      "grad_norm": 1.0342648029327393,
      "learning_rate": 2.7542857142857145e-06,
      "loss": 0.8166,
      "step": 902
    },
    {
      "epoch": 0.5560344827586207,
      "grad_norm": 1.0739760398864746,
      "learning_rate": 2.750476190476191e-06,
      "loss": 0.8646,
      "step": 903
    },
    {
      "epoch": 0.5566502463054187,
      "grad_norm": 1.0582547187805176,
      "learning_rate": 2.746666666666667e-06,
      "loss": 0.8239,
      "step": 904
    },
    {
      "epoch": 0.5572660098522167,
      "grad_norm": 1.0479904413223267,
      "learning_rate": 2.742857142857143e-06,
      "loss": 0.8556,
      "step": 905
    },
    {
      "epoch": 0.5578817733990148,
      "grad_norm": 0.9984835386276245,
      "learning_rate": 2.7390476190476193e-06,
      "loss": 0.8887,
      "step": 906
    },
    {
      "epoch": 0.5584975369458128,
      "grad_norm": 0.9534016251564026,
      "learning_rate": 2.7352380952380953e-06,
      "loss": 0.8054,
      "step": 907
    },
    {
      "epoch": 0.5591133004926109,
      "grad_norm": 1.0325086116790771,
      "learning_rate": 2.7314285714285714e-06,
      "loss": 0.8498,
      "step": 908
    },
    {
      "epoch": 0.5597290640394089,
      "grad_norm": 1.0531162023544312,
      "learning_rate": 2.7276190476190478e-06,
      "loss": 0.7864,
      "step": 909
    },
    {
      "epoch": 0.5603448275862069,
      "grad_norm": 0.9445928335189819,
      "learning_rate": 2.7238095238095238e-06,
      "loss": 0.8547,
      "step": 910
    },
    {
      "epoch": 0.5609605911330049,
      "grad_norm": 1.069180965423584,
      "learning_rate": 2.72e-06,
      "loss": 0.8456,
      "step": 911
    },
    {
      "epoch": 0.5615763546798029,
      "grad_norm": 1.0739282369613647,
      "learning_rate": 2.7161904761904762e-06,
      "loss": 0.8392,
      "step": 912
    },
    {
      "epoch": 0.562192118226601,
      "grad_norm": 1.1287260055541992,
      "learning_rate": 2.7123809523809522e-06,
      "loss": 0.8793,
      "step": 913
    },
    {
      "epoch": 0.562807881773399,
      "grad_norm": 1.0170884132385254,
      "learning_rate": 2.7085714285714287e-06,
      "loss": 0.8889,
      "step": 914
    },
    {
      "epoch": 0.5634236453201971,
      "grad_norm": 0.9465327262878418,
      "learning_rate": 2.704761904761905e-06,
      "loss": 0.8033,
      "step": 915
    },
    {
      "epoch": 0.5640394088669951,
      "grad_norm": 0.9384053349494934,
      "learning_rate": 2.700952380952381e-06,
      "loss": 0.8162,
      "step": 916
    },
    {
      "epoch": 0.5646551724137931,
      "grad_norm": 0.9899068474769592,
      "learning_rate": 2.6971428571428575e-06,
      "loss": 0.802,
      "step": 917
    },
    {
      "epoch": 0.5652709359605911,
      "grad_norm": 1.072324514389038,
      "learning_rate": 2.6933333333333335e-06,
      "loss": 0.8631,
      "step": 918
    },
    {
      "epoch": 0.5658866995073891,
      "grad_norm": 1.0542510747909546,
      "learning_rate": 2.6895238095238095e-06,
      "loss": 0.8312,
      "step": 919
    },
    {
      "epoch": 0.5665024630541872,
      "grad_norm": 1.0148500204086304,
      "learning_rate": 2.685714285714286e-06,
      "loss": 0.8405,
      "step": 920
    },
    {
      "epoch": 0.5671182266009852,
      "grad_norm": 0.9555147886276245,
      "learning_rate": 2.681904761904762e-06,
      "loss": 0.8116,
      "step": 921
    },
    {
      "epoch": 0.5677339901477833,
      "grad_norm": 0.908726692199707,
      "learning_rate": 2.678095238095238e-06,
      "loss": 0.8213,
      "step": 922
    },
    {
      "epoch": 0.5683497536945813,
      "grad_norm": 1.07468581199646,
      "learning_rate": 2.6742857142857144e-06,
      "loss": 0.8526,
      "step": 923
    },
    {
      "epoch": 0.5689655172413793,
      "grad_norm": 0.9721816182136536,
      "learning_rate": 2.6704761904761904e-06,
      "loss": 0.908,
      "step": 924
    },
    {
      "epoch": 0.5695812807881774,
      "grad_norm": 1.07399582862854,
      "learning_rate": 2.6666666666666664e-06,
      "loss": 0.8468,
      "step": 925
    },
    {
      "epoch": 0.5701970443349754,
      "grad_norm": 0.9420062303543091,
      "learning_rate": 2.662857142857143e-06,
      "loss": 0.8292,
      "step": 926
    },
    {
      "epoch": 0.5708128078817734,
      "grad_norm": 1.0710134506225586,
      "learning_rate": 2.6590476190476193e-06,
      "loss": 0.8461,
      "step": 927
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 1.2411563396453857,
      "learning_rate": 2.6552380952380953e-06,
      "loss": 0.851,
      "step": 928
    },
    {
      "epoch": 0.5720443349753694,
      "grad_norm": 1.0767879486083984,
      "learning_rate": 2.6514285714285717e-06,
      "loss": 0.841,
      "step": 929
    },
    {
      "epoch": 0.5726600985221675,
      "grad_norm": 0.9807257652282715,
      "learning_rate": 2.6476190476190477e-06,
      "loss": 0.8323,
      "step": 930
    },
    {
      "epoch": 0.5732758620689655,
      "grad_norm": 1.0455483198165894,
      "learning_rate": 2.643809523809524e-06,
      "loss": 0.8326,
      "step": 931
    },
    {
      "epoch": 0.5738916256157636,
      "grad_norm": 1.1034036874771118,
      "learning_rate": 2.64e-06,
      "loss": 0.8381,
      "step": 932
    },
    {
      "epoch": 0.5745073891625616,
      "grad_norm": 1.167769432067871,
      "learning_rate": 2.636190476190476e-06,
      "loss": 0.8434,
      "step": 933
    },
    {
      "epoch": 0.5751231527093597,
      "grad_norm": 1.0086592435836792,
      "learning_rate": 2.6323809523809526e-06,
      "loss": 0.8702,
      "step": 934
    },
    {
      "epoch": 0.5757389162561576,
      "grad_norm": 1.0377849340438843,
      "learning_rate": 2.6285714285714286e-06,
      "loss": 0.8394,
      "step": 935
    },
    {
      "epoch": 0.5763546798029556,
      "grad_norm": 1.0401242971420288,
      "learning_rate": 2.6247619047619046e-06,
      "loss": 0.8717,
      "step": 936
    },
    {
      "epoch": 0.5769704433497537,
      "grad_norm": 0.9536492228507996,
      "learning_rate": 2.620952380952381e-06,
      "loss": 0.8003,
      "step": 937
    },
    {
      "epoch": 0.5775862068965517,
      "grad_norm": 1.0148124694824219,
      "learning_rate": 2.617142857142857e-06,
      "loss": 0.8399,
      "step": 938
    },
    {
      "epoch": 0.5782019704433498,
      "grad_norm": 1.0109648704528809,
      "learning_rate": 2.6133333333333334e-06,
      "loss": 0.8593,
      "step": 939
    },
    {
      "epoch": 0.5788177339901478,
      "grad_norm": 1.0172138214111328,
      "learning_rate": 2.6095238095238094e-06,
      "loss": 0.8822,
      "step": 940
    },
    {
      "epoch": 0.5794334975369458,
      "grad_norm": 1.0163543224334717,
      "learning_rate": 2.605714285714286e-06,
      "loss": 0.8423,
      "step": 941
    },
    {
      "epoch": 0.5800492610837439,
      "grad_norm": 0.9883476495742798,
      "learning_rate": 2.6019047619047623e-06,
      "loss": 0.836,
      "step": 942
    },
    {
      "epoch": 0.5806650246305419,
      "grad_norm": 0.9698370695114136,
      "learning_rate": 2.5980952380952383e-06,
      "loss": 0.8758,
      "step": 943
    },
    {
      "epoch": 0.5812807881773399,
      "grad_norm": 0.9637258052825928,
      "learning_rate": 2.5942857142857143e-06,
      "loss": 0.8231,
      "step": 944
    },
    {
      "epoch": 0.5818965517241379,
      "grad_norm": 1.218523621559143,
      "learning_rate": 2.5904761904761907e-06,
      "loss": 0.8394,
      "step": 945
    },
    {
      "epoch": 0.5825123152709359,
      "grad_norm": 0.9497475624084473,
      "learning_rate": 2.5866666666666667e-06,
      "loss": 0.8165,
      "step": 946
    },
    {
      "epoch": 0.583128078817734,
      "grad_norm": 0.9990437626838684,
      "learning_rate": 2.5828571428571428e-06,
      "loss": 0.8306,
      "step": 947
    },
    {
      "epoch": 0.583743842364532,
      "grad_norm": 0.9409560561180115,
      "learning_rate": 2.579047619047619e-06,
      "loss": 0.8604,
      "step": 948
    },
    {
      "epoch": 0.5843596059113301,
      "grad_norm": 1.0152831077575684,
      "learning_rate": 2.575238095238095e-06,
      "loss": 0.8329,
      "step": 949
    },
    {
      "epoch": 0.5849753694581281,
      "grad_norm": 1.0241119861602783,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.8368,
      "step": 950
    },
    {
      "epoch": 0.5855911330049262,
      "grad_norm": 1.0195608139038086,
      "learning_rate": 2.5676190476190476e-06,
      "loss": 0.8445,
      "step": 951
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.997439444065094,
      "learning_rate": 2.5638095238095236e-06,
      "loss": 0.8515,
      "step": 952
    },
    {
      "epoch": 0.5868226600985221,
      "grad_norm": 1.087146520614624,
      "learning_rate": 2.56e-06,
      "loss": 0.8244,
      "step": 953
    },
    {
      "epoch": 0.5874384236453202,
      "grad_norm": 1.0732909440994263,
      "learning_rate": 2.5561904761904765e-06,
      "loss": 0.8543,
      "step": 954
    },
    {
      "epoch": 0.5880541871921182,
      "grad_norm": 0.915617048740387,
      "learning_rate": 2.5523809523809525e-06,
      "loss": 0.8133,
      "step": 955
    },
    {
      "epoch": 0.5886699507389163,
      "grad_norm": 1.0086543560028076,
      "learning_rate": 2.548571428571429e-06,
      "loss": 0.8585,
      "step": 956
    },
    {
      "epoch": 0.5892857142857143,
      "grad_norm": 1.0065165758132935,
      "learning_rate": 2.544761904761905e-06,
      "loss": 0.8062,
      "step": 957
    },
    {
      "epoch": 0.5899014778325123,
      "grad_norm": 1.0158047676086426,
      "learning_rate": 2.540952380952381e-06,
      "loss": 0.8286,
      "step": 958
    },
    {
      "epoch": 0.5905172413793104,
      "grad_norm": 1.0015265941619873,
      "learning_rate": 2.5371428571428574e-06,
      "loss": 0.8354,
      "step": 959
    },
    {
      "epoch": 0.5911330049261084,
      "grad_norm": 0.9662439823150635,
      "learning_rate": 2.5333333333333334e-06,
      "loss": 0.8205,
      "step": 960
    },
    {
      "epoch": 0.5917487684729064,
      "grad_norm": 0.9409492611885071,
      "learning_rate": 2.5295238095238094e-06,
      "loss": 0.8567,
      "step": 961
    },
    {
      "epoch": 0.5923645320197044,
      "grad_norm": 1.086824655532837,
      "learning_rate": 2.525714285714286e-06,
      "loss": 0.8271,
      "step": 962
    },
    {
      "epoch": 0.5929802955665024,
      "grad_norm": 1.092983365058899,
      "learning_rate": 2.521904761904762e-06,
      "loss": 0.817,
      "step": 963
    },
    {
      "epoch": 0.5935960591133005,
      "grad_norm": 0.9979318976402283,
      "learning_rate": 2.518095238095238e-06,
      "loss": 0.8076,
      "step": 964
    },
    {
      "epoch": 0.5942118226600985,
      "grad_norm": 1.0546082258224487,
      "learning_rate": 2.5142857142857142e-06,
      "loss": 0.8901,
      "step": 965
    },
    {
      "epoch": 0.5948275862068966,
      "grad_norm": 1.0841916799545288,
      "learning_rate": 2.5104761904761907e-06,
      "loss": 0.8271,
      "step": 966
    },
    {
      "epoch": 0.5954433497536946,
      "grad_norm": 1.0196930170059204,
      "learning_rate": 2.506666666666667e-06,
      "loss": 0.8192,
      "step": 967
    },
    {
      "epoch": 0.5960591133004927,
      "grad_norm": 1.1608837842941284,
      "learning_rate": 2.502857142857143e-06,
      "loss": 0.8079,
      "step": 968
    },
    {
      "epoch": 0.5966748768472906,
      "grad_norm": 0.9882367253303528,
      "learning_rate": 2.499047619047619e-06,
      "loss": 0.8722,
      "step": 969
    },
    {
      "epoch": 0.5972906403940886,
      "grad_norm": 1.1385109424591064,
      "learning_rate": 2.4952380952380955e-06,
      "loss": 0.8383,
      "step": 970
    },
    {
      "epoch": 0.5979064039408867,
      "grad_norm": 0.9962518215179443,
      "learning_rate": 2.4914285714285715e-06,
      "loss": 0.8644,
      "step": 971
    },
    {
      "epoch": 0.5985221674876847,
      "grad_norm": 1.0181434154510498,
      "learning_rate": 2.4876190476190475e-06,
      "loss": 0.8431,
      "step": 972
    },
    {
      "epoch": 0.5991379310344828,
      "grad_norm": 0.9908741116523743,
      "learning_rate": 2.483809523809524e-06,
      "loss": 0.802,
      "step": 973
    },
    {
      "epoch": 0.5997536945812808,
      "grad_norm": 0.9893283247947693,
      "learning_rate": 2.48e-06,
      "loss": 0.8393,
      "step": 974
    },
    {
      "epoch": 0.6003694581280788,
      "grad_norm": 1.0618805885314941,
      "learning_rate": 2.476190476190476e-06,
      "loss": 0.8103,
      "step": 975
    },
    {
      "epoch": 0.6009852216748769,
      "grad_norm": 1.044451117515564,
      "learning_rate": 2.4723809523809524e-06,
      "loss": 0.8658,
      "step": 976
    },
    {
      "epoch": 0.6016009852216748,
      "grad_norm": 1.0128687620162964,
      "learning_rate": 2.4685714285714284e-06,
      "loss": 0.8201,
      "step": 977
    },
    {
      "epoch": 0.6022167487684729,
      "grad_norm": 1.0137152671813965,
      "learning_rate": 2.464761904761905e-06,
      "loss": 0.8244,
      "step": 978
    },
    {
      "epoch": 0.6028325123152709,
      "grad_norm": 0.922044038772583,
      "learning_rate": 2.4609523809523813e-06,
      "loss": 0.8432,
      "step": 979
    },
    {
      "epoch": 0.603448275862069,
      "grad_norm": 1.1059908866882324,
      "learning_rate": 2.4571428571428573e-06,
      "loss": 0.882,
      "step": 980
    },
    {
      "epoch": 0.604064039408867,
      "grad_norm": 1.0738403797149658,
      "learning_rate": 2.4533333333333337e-06,
      "loss": 0.8488,
      "step": 981
    },
    {
      "epoch": 0.604679802955665,
      "grad_norm": 1.0751841068267822,
      "learning_rate": 2.4495238095238097e-06,
      "loss": 0.8744,
      "step": 982
    },
    {
      "epoch": 0.6052955665024631,
      "grad_norm": 0.9943671226501465,
      "learning_rate": 2.4457142857142857e-06,
      "loss": 0.7988,
      "step": 983
    },
    {
      "epoch": 0.6059113300492611,
      "grad_norm": 1.018326759338379,
      "learning_rate": 2.441904761904762e-06,
      "loss": 0.8497,
      "step": 984
    },
    {
      "epoch": 0.6065270935960592,
      "grad_norm": 1.0432591438293457,
      "learning_rate": 2.438095238095238e-06,
      "loss": 0.8354,
      "step": 985
    },
    {
      "epoch": 0.6071428571428571,
      "grad_norm": 1.1030921936035156,
      "learning_rate": 2.434285714285714e-06,
      "loss": 0.8526,
      "step": 986
    },
    {
      "epoch": 0.6077586206896551,
      "grad_norm": 0.9995616674423218,
      "learning_rate": 2.4304761904761906e-06,
      "loss": 0.8193,
      "step": 987
    },
    {
      "epoch": 0.6083743842364532,
      "grad_norm": 1.0428574085235596,
      "learning_rate": 2.4266666666666666e-06,
      "loss": 0.8301,
      "step": 988
    },
    {
      "epoch": 0.6089901477832512,
      "grad_norm": 0.9831702709197998,
      "learning_rate": 2.4228571428571426e-06,
      "loss": 0.8351,
      "step": 989
    },
    {
      "epoch": 0.6096059113300493,
      "grad_norm": 1.0399752855300903,
      "learning_rate": 2.419047619047619e-06,
      "loss": 0.8523,
      "step": 990
    },
    {
      "epoch": 0.6102216748768473,
      "grad_norm": 1.0548717975616455,
      "learning_rate": 2.4152380952380954e-06,
      "loss": 0.8484,
      "step": 991
    },
    {
      "epoch": 0.6108374384236454,
      "grad_norm": 1.0277878046035767,
      "learning_rate": 2.4114285714285715e-06,
      "loss": 0.8355,
      "step": 992
    },
    {
      "epoch": 0.6114532019704434,
      "grad_norm": 1.0546752214431763,
      "learning_rate": 2.407619047619048e-06,
      "loss": 0.8652,
      "step": 993
    },
    {
      "epoch": 0.6120689655172413,
      "grad_norm": 0.9474188089370728,
      "learning_rate": 2.403809523809524e-06,
      "loss": 0.8543,
      "step": 994
    },
    {
      "epoch": 0.6126847290640394,
      "grad_norm": 0.9541683793067932,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.8386,
      "step": 995
    },
    {
      "epoch": 0.6133004926108374,
      "grad_norm": 1.0309746265411377,
      "learning_rate": 2.3961904761904763e-06,
      "loss": 0.8676,
      "step": 996
    },
    {
      "epoch": 0.6139162561576355,
      "grad_norm": 1.1148124933242798,
      "learning_rate": 2.3923809523809523e-06,
      "loss": 0.8621,
      "step": 997
    },
    {
      "epoch": 0.6145320197044335,
      "grad_norm": 1.0955052375793457,
      "learning_rate": 2.3885714285714288e-06,
      "loss": 0.8478,
      "step": 998
    },
    {
      "epoch": 0.6151477832512315,
      "grad_norm": 1.0075651407241821,
      "learning_rate": 2.3847619047619048e-06,
      "loss": 0.877,
      "step": 999
    },
    {
      "epoch": 0.6157635467980296,
      "grad_norm": 1.0625154972076416,
      "learning_rate": 2.3809523809523808e-06,
      "loss": 0.889,
      "step": 1000
    },
    {
      "epoch": 0.6163793103448276,
      "grad_norm": 1.072076439857483,
      "learning_rate": 2.377142857142857e-06,
      "loss": 0.8308,
      "step": 1001
    },
    {
      "epoch": 0.6169950738916257,
      "grad_norm": 1.075554609298706,
      "learning_rate": 2.373333333333333e-06,
      "loss": 0.8317,
      "step": 1002
    },
    {
      "epoch": 0.6176108374384236,
      "grad_norm": 1.1705986261367798,
      "learning_rate": 2.3695238095238096e-06,
      "loss": 0.8262,
      "step": 1003
    },
    {
      "epoch": 0.6182266009852216,
      "grad_norm": 1.021462321281433,
      "learning_rate": 2.3657142857142856e-06,
      "loss": 0.8344,
      "step": 1004
    },
    {
      "epoch": 0.6188423645320197,
      "grad_norm": 0.9518441557884216,
      "learning_rate": 2.361904761904762e-06,
      "loss": 0.8866,
      "step": 1005
    },
    {
      "epoch": 0.6194581280788177,
      "grad_norm": 1.0836522579193115,
      "learning_rate": 2.3580952380952385e-06,
      "loss": 0.7963,
      "step": 1006
    },
    {
      "epoch": 0.6200738916256158,
      "grad_norm": 1.027862548828125,
      "learning_rate": 2.3542857142857145e-06,
      "loss": 0.8561,
      "step": 1007
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 1.0748722553253174,
      "learning_rate": 2.3504761904761905e-06,
      "loss": 0.9251,
      "step": 1008
    },
    {
      "epoch": 0.6213054187192119,
      "grad_norm": 1.034463882446289,
      "learning_rate": 2.346666666666667e-06,
      "loss": 0.8803,
      "step": 1009
    },
    {
      "epoch": 0.6219211822660099,
      "grad_norm": 1.0238631963729858,
      "learning_rate": 2.342857142857143e-06,
      "loss": 0.8318,
      "step": 1010
    },
    {
      "epoch": 0.6225369458128078,
      "grad_norm": 1.0112851858139038,
      "learning_rate": 2.339047619047619e-06,
      "loss": 0.8272,
      "step": 1011
    },
    {
      "epoch": 0.6231527093596059,
      "grad_norm": 0.9703636765480042,
      "learning_rate": 2.3352380952380954e-06,
      "loss": 0.8186,
      "step": 1012
    },
    {
      "epoch": 0.6237684729064039,
      "grad_norm": 1.0441447496414185,
      "learning_rate": 2.3314285714285714e-06,
      "loss": 0.8303,
      "step": 1013
    },
    {
      "epoch": 0.624384236453202,
      "grad_norm": 1.0281224250793457,
      "learning_rate": 2.3276190476190474e-06,
      "loss": 0.813,
      "step": 1014
    },
    {
      "epoch": 0.625,
      "grad_norm": 1.0130330324172974,
      "learning_rate": 2.323809523809524e-06,
      "loss": 0.8159,
      "step": 1015
    },
    {
      "epoch": 0.625615763546798,
      "grad_norm": 1.022853136062622,
      "learning_rate": 2.32e-06,
      "loss": 0.8512,
      "step": 1016
    },
    {
      "epoch": 0.6262315270935961,
      "grad_norm": 1.021051287651062,
      "learning_rate": 2.3161904761904762e-06,
      "loss": 0.8112,
      "step": 1017
    },
    {
      "epoch": 0.6268472906403941,
      "grad_norm": 1.0904004573822021,
      "learning_rate": 2.3123809523809527e-06,
      "loss": 0.8445,
      "step": 1018
    },
    {
      "epoch": 0.6274630541871922,
      "grad_norm": 1.0242350101470947,
      "learning_rate": 2.3085714285714287e-06,
      "loss": 0.8558,
      "step": 1019
    },
    {
      "epoch": 0.6280788177339901,
      "grad_norm": 1.0237197875976562,
      "learning_rate": 2.304761904761905e-06,
      "loss": 0.8653,
      "step": 1020
    },
    {
      "epoch": 0.6286945812807881,
      "grad_norm": 1.0433472394943237,
      "learning_rate": 2.300952380952381e-06,
      "loss": 0.8482,
      "step": 1021
    },
    {
      "epoch": 0.6293103448275862,
      "grad_norm": 1.0790034532546997,
      "learning_rate": 2.297142857142857e-06,
      "loss": 0.8644,
      "step": 1022
    },
    {
      "epoch": 0.6299261083743842,
      "grad_norm": 0.9393789768218994,
      "learning_rate": 2.2933333333333335e-06,
      "loss": 0.7944,
      "step": 1023
    },
    {
      "epoch": 0.6305418719211823,
      "grad_norm": 1.0704666376113892,
      "learning_rate": 2.2895238095238095e-06,
      "loss": 0.8202,
      "step": 1024
    },
    {
      "epoch": 0.6311576354679803,
      "grad_norm": 1.0453557968139648,
      "learning_rate": 2.2857142857142856e-06,
      "loss": 0.8705,
      "step": 1025
    },
    {
      "epoch": 0.6317733990147784,
      "grad_norm": 1.0100388526916504,
      "learning_rate": 2.281904761904762e-06,
      "loss": 0.8459,
      "step": 1026
    },
    {
      "epoch": 0.6323891625615764,
      "grad_norm": 0.9309197068214417,
      "learning_rate": 2.278095238095238e-06,
      "loss": 0.8684,
      "step": 1027
    },
    {
      "epoch": 0.6330049261083743,
      "grad_norm": 1.005162239074707,
      "learning_rate": 2.274285714285714e-06,
      "loss": 0.8424,
      "step": 1028
    },
    {
      "epoch": 0.6336206896551724,
      "grad_norm": 1.0321189165115356,
      "learning_rate": 2.2704761904761904e-06,
      "loss": 0.7849,
      "step": 1029
    },
    {
      "epoch": 0.6342364532019704,
      "grad_norm": 0.9888390302658081,
      "learning_rate": 2.266666666666667e-06,
      "loss": 0.8146,
      "step": 1030
    },
    {
      "epoch": 0.6348522167487685,
      "grad_norm": 1.0745997428894043,
      "learning_rate": 2.2628571428571433e-06,
      "loss": 0.8602,
      "step": 1031
    },
    {
      "epoch": 0.6354679802955665,
      "grad_norm": 1.0374040603637695,
      "learning_rate": 2.2590476190476193e-06,
      "loss": 0.8476,
      "step": 1032
    },
    {
      "epoch": 0.6360837438423645,
      "grad_norm": 1.0994529724121094,
      "learning_rate": 2.2552380952380953e-06,
      "loss": 0.8587,
      "step": 1033
    },
    {
      "epoch": 0.6366995073891626,
      "grad_norm": 1.0467356443405151,
      "learning_rate": 2.2514285714285717e-06,
      "loss": 0.8472,
      "step": 1034
    },
    {
      "epoch": 0.6373152709359606,
      "grad_norm": 1.0655418634414673,
      "learning_rate": 2.2476190476190477e-06,
      "loss": 0.8379,
      "step": 1035
    },
    {
      "epoch": 0.6379310344827587,
      "grad_norm": 0.984703540802002,
      "learning_rate": 2.2438095238095237e-06,
      "loss": 0.844,
      "step": 1036
    },
    {
      "epoch": 0.6385467980295566,
      "grad_norm": 1.1105161905288696,
      "learning_rate": 2.24e-06,
      "loss": 0.8752,
      "step": 1037
    },
    {
      "epoch": 0.6391625615763546,
      "grad_norm": 0.9944372773170471,
      "learning_rate": 2.236190476190476e-06,
      "loss": 0.9221,
      "step": 1038
    },
    {
      "epoch": 0.6397783251231527,
      "grad_norm": 1.0068645477294922,
      "learning_rate": 2.232380952380952e-06,
      "loss": 0.9048,
      "step": 1039
    },
    {
      "epoch": 0.6403940886699507,
      "grad_norm": 1.225646734237671,
      "learning_rate": 2.2285714285714286e-06,
      "loss": 0.8328,
      "step": 1040
    },
    {
      "epoch": 0.6410098522167488,
      "grad_norm": 1.032476544380188,
      "learning_rate": 2.2247619047619046e-06,
      "loss": 0.8146,
      "step": 1041
    },
    {
      "epoch": 0.6416256157635468,
      "grad_norm": 1.0454856157302856,
      "learning_rate": 2.220952380952381e-06,
      "loss": 0.868,
      "step": 1042
    },
    {
      "epoch": 0.6422413793103449,
      "grad_norm": 1.0167503356933594,
      "learning_rate": 2.2171428571428575e-06,
      "loss": 0.8383,
      "step": 1043
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 1.0770487785339355,
      "learning_rate": 2.2133333333333335e-06,
      "loss": 0.8118,
      "step": 1044
    },
    {
      "epoch": 0.6434729064039408,
      "grad_norm": 1.021677851676941,
      "learning_rate": 2.20952380952381e-06,
      "loss": 0.8554,
      "step": 1045
    },
    {
      "epoch": 0.6440886699507389,
      "grad_norm": 0.9594687223434448,
      "learning_rate": 2.205714285714286e-06,
      "loss": 0.8744,
      "step": 1046
    },
    {
      "epoch": 0.6447044334975369,
      "grad_norm": 1.0049660205841064,
      "learning_rate": 2.201904761904762e-06,
      "loss": 0.8869,
      "step": 1047
    },
    {
      "epoch": 0.645320197044335,
      "grad_norm": 1.0204987525939941,
      "learning_rate": 2.1980952380952383e-06,
      "loss": 0.8804,
      "step": 1048
    },
    {
      "epoch": 0.645935960591133,
      "grad_norm": 1.0086159706115723,
      "learning_rate": 2.1942857142857143e-06,
      "loss": 0.8106,
      "step": 1049
    },
    {
      "epoch": 0.646551724137931,
      "grad_norm": 1.0101432800292969,
      "learning_rate": 2.1904761904761903e-06,
      "loss": 0.8352,
      "step": 1050
    },
    {
      "epoch": 0.6471674876847291,
      "grad_norm": 0.9669128656387329,
      "learning_rate": 2.1866666666666668e-06,
      "loss": 0.8416,
      "step": 1051
    },
    {
      "epoch": 0.6477832512315271,
      "grad_norm": 1.149319052696228,
      "learning_rate": 2.1828571428571428e-06,
      "loss": 0.8463,
      "step": 1052
    },
    {
      "epoch": 0.6483990147783252,
      "grad_norm": 1.0524758100509644,
      "learning_rate": 2.1790476190476188e-06,
      "loss": 0.806,
      "step": 1053
    },
    {
      "epoch": 0.6490147783251231,
      "grad_norm": 1.0459415912628174,
      "learning_rate": 2.175238095238095e-06,
      "loss": 0.8663,
      "step": 1054
    },
    {
      "epoch": 0.6496305418719212,
      "grad_norm": 1.0448567867279053,
      "learning_rate": 2.1714285714285716e-06,
      "loss": 0.8358,
      "step": 1055
    },
    {
      "epoch": 0.6502463054187192,
      "grad_norm": 1.032975673675537,
      "learning_rate": 2.1676190476190476e-06,
      "loss": 0.878,
      "step": 1056
    },
    {
      "epoch": 0.6508620689655172,
      "grad_norm": 1.0608699321746826,
      "learning_rate": 2.163809523809524e-06,
      "loss": 0.8731,
      "step": 1057
    },
    {
      "epoch": 0.6514778325123153,
      "grad_norm": 0.9903609752655029,
      "learning_rate": 2.16e-06,
      "loss": 0.8627,
      "step": 1058
    },
    {
      "epoch": 0.6520935960591133,
      "grad_norm": 1.0215083360671997,
      "learning_rate": 2.1561904761904765e-06,
      "loss": 0.8282,
      "step": 1059
    },
    {
      "epoch": 0.6527093596059114,
      "grad_norm": 1.0043853521347046,
      "learning_rate": 2.1523809523809525e-06,
      "loss": 0.8573,
      "step": 1060
    },
    {
      "epoch": 0.6533251231527094,
      "grad_norm": 0.9875706434249878,
      "learning_rate": 2.1485714285714285e-06,
      "loss": 0.8384,
      "step": 1061
    },
    {
      "epoch": 0.6539408866995073,
      "grad_norm": 1.0353132486343384,
      "learning_rate": 2.144761904761905e-06,
      "loss": 0.845,
      "step": 1062
    },
    {
      "epoch": 0.6545566502463054,
      "grad_norm": 1.0169804096221924,
      "learning_rate": 2.140952380952381e-06,
      "loss": 0.8318,
      "step": 1063
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 1.0403574705123901,
      "learning_rate": 2.137142857142857e-06,
      "loss": 0.8893,
      "step": 1064
    },
    {
      "epoch": 0.6557881773399015,
      "grad_norm": 1.1282763481140137,
      "learning_rate": 2.1333333333333334e-06,
      "loss": 0.8677,
      "step": 1065
    },
    {
      "epoch": 0.6564039408866995,
      "grad_norm": 0.9803518056869507,
      "learning_rate": 2.1295238095238094e-06,
      "loss": 0.8596,
      "step": 1066
    },
    {
      "epoch": 0.6570197044334976,
      "grad_norm": 1.024086356163025,
      "learning_rate": 2.125714285714286e-06,
      "loss": 0.8417,
      "step": 1067
    },
    {
      "epoch": 0.6576354679802956,
      "grad_norm": 1.0866937637329102,
      "learning_rate": 2.121904761904762e-06,
      "loss": 0.8359,
      "step": 1068
    },
    {
      "epoch": 0.6582512315270936,
      "grad_norm": 1.0231972932815552,
      "learning_rate": 2.1180952380952382e-06,
      "loss": 0.8393,
      "step": 1069
    },
    {
      "epoch": 0.6588669950738916,
      "grad_norm": 1.0619103908538818,
      "learning_rate": 2.1142857142857147e-06,
      "loss": 0.9142,
      "step": 1070
    },
    {
      "epoch": 0.6594827586206896,
      "grad_norm": 1.0807710886001587,
      "learning_rate": 2.1104761904761907e-06,
      "loss": 0.8509,
      "step": 1071
    },
    {
      "epoch": 0.6600985221674877,
      "grad_norm": 1.0495434999465942,
      "learning_rate": 2.1066666666666667e-06,
      "loss": 0.8201,
      "step": 1072
    },
    {
      "epoch": 0.6607142857142857,
      "grad_norm": 1.0145084857940674,
      "learning_rate": 2.102857142857143e-06,
      "loss": 0.8698,
      "step": 1073
    },
    {
      "epoch": 0.6613300492610837,
      "grad_norm": 1.04902184009552,
      "learning_rate": 2.099047619047619e-06,
      "loss": 0.8546,
      "step": 1074
    },
    {
      "epoch": 0.6619458128078818,
      "grad_norm": 0.9580238461494446,
      "learning_rate": 2.095238095238095e-06,
      "loss": 0.8566,
      "step": 1075
    },
    {
      "epoch": 0.6625615763546798,
      "grad_norm": 1.0222722291946411,
      "learning_rate": 2.0914285714285716e-06,
      "loss": 0.8351,
      "step": 1076
    },
    {
      "epoch": 0.6631773399014779,
      "grad_norm": 1.0343191623687744,
      "learning_rate": 2.0876190476190476e-06,
      "loss": 0.8865,
      "step": 1077
    },
    {
      "epoch": 0.6637931034482759,
      "grad_norm": 0.9986064434051514,
      "learning_rate": 2.0838095238095236e-06,
      "loss": 0.8459,
      "step": 1078
    },
    {
      "epoch": 0.6644088669950738,
      "grad_norm": 1.0529553890228271,
      "learning_rate": 2.08e-06,
      "loss": 0.8297,
      "step": 1079
    },
    {
      "epoch": 0.6650246305418719,
      "grad_norm": 0.9728953838348389,
      "learning_rate": 2.076190476190476e-06,
      "loss": 0.8566,
      "step": 1080
    },
    {
      "epoch": 0.6656403940886699,
      "grad_norm": 1.1703251600265503,
      "learning_rate": 2.0723809523809524e-06,
      "loss": 0.8505,
      "step": 1081
    },
    {
      "epoch": 0.666256157635468,
      "grad_norm": 0.9675374627113342,
      "learning_rate": 2.068571428571429e-06,
      "loss": 0.8309,
      "step": 1082
    },
    {
      "epoch": 0.666871921182266,
      "grad_norm": 0.9819311499595642,
      "learning_rate": 2.064761904761905e-06,
      "loss": 0.8282,
      "step": 1083
    },
    {
      "epoch": 0.6674876847290641,
      "grad_norm": 0.9409564733505249,
      "learning_rate": 2.0609523809523813e-06,
      "loss": 0.8512,
      "step": 1084
    },
    {
      "epoch": 0.6681034482758621,
      "grad_norm": 0.9291539788246155,
      "learning_rate": 2.0571428571428573e-06,
      "loss": 0.8406,
      "step": 1085
    },
    {
      "epoch": 0.6687192118226601,
      "grad_norm": 1.0949643850326538,
      "learning_rate": 2.0533333333333333e-06,
      "loss": 0.8396,
      "step": 1086
    },
    {
      "epoch": 0.6693349753694581,
      "grad_norm": 1.0173063278198242,
      "learning_rate": 2.0495238095238097e-06,
      "loss": 0.8511,
      "step": 1087
    },
    {
      "epoch": 0.6699507389162561,
      "grad_norm": 1.0233629941940308,
      "learning_rate": 2.0457142857142857e-06,
      "loss": 0.8623,
      "step": 1088
    },
    {
      "epoch": 0.6705665024630542,
      "grad_norm": 0.9350053071975708,
      "learning_rate": 2.0419047619047617e-06,
      "loss": 0.8556,
      "step": 1089
    },
    {
      "epoch": 0.6711822660098522,
      "grad_norm": 0.9917851090431213,
      "learning_rate": 2.038095238095238e-06,
      "loss": 0.839,
      "step": 1090
    },
    {
      "epoch": 0.6717980295566502,
      "grad_norm": 0.9782205820083618,
      "learning_rate": 2.034285714285714e-06,
      "loss": 0.7944,
      "step": 1091
    },
    {
      "epoch": 0.6724137931034483,
      "grad_norm": 0.9626043438911438,
      "learning_rate": 2.03047619047619e-06,
      "loss": 0.8466,
      "step": 1092
    },
    {
      "epoch": 0.6730295566502463,
      "grad_norm": 1.0404011011123657,
      "learning_rate": 2.0266666666666666e-06,
      "loss": 0.8667,
      "step": 1093
    },
    {
      "epoch": 0.6736453201970444,
      "grad_norm": 0.9642599821090698,
      "learning_rate": 2.022857142857143e-06,
      "loss": 0.8336,
      "step": 1094
    },
    {
      "epoch": 0.6742610837438424,
      "grad_norm": 0.996139407157898,
      "learning_rate": 2.0190476190476195e-06,
      "loss": 0.8333,
      "step": 1095
    },
    {
      "epoch": 0.6748768472906403,
      "grad_norm": 1.092588186264038,
      "learning_rate": 2.0152380952380955e-06,
      "loss": 0.8497,
      "step": 1096
    },
    {
      "epoch": 0.6754926108374384,
      "grad_norm": 1.1350131034851074,
      "learning_rate": 2.0114285714285715e-06,
      "loss": 0.815,
      "step": 1097
    },
    {
      "epoch": 0.6761083743842364,
      "grad_norm": 1.0161532163619995,
      "learning_rate": 2.007619047619048e-06,
      "loss": 0.8136,
      "step": 1098
    },
    {
      "epoch": 0.6767241379310345,
      "grad_norm": 0.9644890427589417,
      "learning_rate": 2.003809523809524e-06,
      "loss": 0.8353,
      "step": 1099
    },
    {
      "epoch": 0.6773399014778325,
      "grad_norm": 1.0205581188201904,
      "learning_rate": 2e-06,
      "loss": 0.8473,
      "step": 1100
    },
    {
      "epoch": 0.6779556650246306,
      "grad_norm": 0.9401888847351074,
      "learning_rate": 1.9961904761904763e-06,
      "loss": 0.8422,
      "step": 1101
    },
    {
      "epoch": 0.6785714285714286,
      "grad_norm": 1.1366227865219116,
      "learning_rate": 1.9923809523809523e-06,
      "loss": 0.8316,
      "step": 1102
    },
    {
      "epoch": 0.6791871921182266,
      "grad_norm": 1.0479120016098022,
      "learning_rate": 1.9885714285714283e-06,
      "loss": 0.9002,
      "step": 1103
    },
    {
      "epoch": 0.6798029556650246,
      "grad_norm": 0.9822748303413391,
      "learning_rate": 1.9847619047619048e-06,
      "loss": 0.8584,
      "step": 1104
    },
    {
      "epoch": 0.6804187192118226,
      "grad_norm": 0.9746829867362976,
      "learning_rate": 1.9809523809523808e-06,
      "loss": 0.8757,
      "step": 1105
    },
    {
      "epoch": 0.6810344827586207,
      "grad_norm": 1.05191171169281,
      "learning_rate": 1.977142857142857e-06,
      "loss": 0.8161,
      "step": 1106
    },
    {
      "epoch": 0.6816502463054187,
      "grad_norm": 0.9775899052619934,
      "learning_rate": 1.9733333333333336e-06,
      "loss": 0.8717,
      "step": 1107
    },
    {
      "epoch": 0.6822660098522167,
      "grad_norm": 1.0329792499542236,
      "learning_rate": 1.9695238095238096e-06,
      "loss": 0.8292,
      "step": 1108
    },
    {
      "epoch": 0.6828817733990148,
      "grad_norm": 1.087691068649292,
      "learning_rate": 1.965714285714286e-06,
      "loss": 0.8269,
      "step": 1109
    },
    {
      "epoch": 0.6834975369458128,
      "grad_norm": 0.9976260662078857,
      "learning_rate": 1.961904761904762e-06,
      "loss": 0.7991,
      "step": 1110
    },
    {
      "epoch": 0.6841133004926109,
      "grad_norm": 1.046807885169983,
      "learning_rate": 1.958095238095238e-06,
      "loss": 0.8395,
      "step": 1111
    },
    {
      "epoch": 0.6847290640394089,
      "grad_norm": 0.993168830871582,
      "learning_rate": 1.9542857142857145e-06,
      "loss": 0.8268,
      "step": 1112
    },
    {
      "epoch": 0.6853448275862069,
      "grad_norm": 0.9414188265800476,
      "learning_rate": 1.9504761904761905e-06,
      "loss": 0.8366,
      "step": 1113
    },
    {
      "epoch": 0.6859605911330049,
      "grad_norm": 1.0948630571365356,
      "learning_rate": 1.9466666666666665e-06,
      "loss": 0.8587,
      "step": 1114
    },
    {
      "epoch": 0.6865763546798029,
      "grad_norm": 0.9887826442718506,
      "learning_rate": 1.942857142857143e-06,
      "loss": 0.8097,
      "step": 1115
    },
    {
      "epoch": 0.687192118226601,
      "grad_norm": 1.181016445159912,
      "learning_rate": 1.939047619047619e-06,
      "loss": 0.8557,
      "step": 1116
    },
    {
      "epoch": 0.687807881773399,
      "grad_norm": 1.0811787843704224,
      "learning_rate": 1.935238095238095e-06,
      "loss": 0.8532,
      "step": 1117
    },
    {
      "epoch": 0.6884236453201971,
      "grad_norm": 1.081970453262329,
      "learning_rate": 1.9314285714285714e-06,
      "loss": 0.7986,
      "step": 1118
    },
    {
      "epoch": 0.6890394088669951,
      "grad_norm": 1.028104305267334,
      "learning_rate": 1.927619047619048e-06,
      "loss": 0.8339,
      "step": 1119
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.9617072939872742,
      "learning_rate": 1.923809523809524e-06,
      "loss": 0.8658,
      "step": 1120
    },
    {
      "epoch": 0.6902709359605911,
      "grad_norm": 0.9445987939834595,
      "learning_rate": 1.9200000000000003e-06,
      "loss": 0.8733,
      "step": 1121
    },
    {
      "epoch": 0.6908866995073891,
      "grad_norm": 0.9881397485733032,
      "learning_rate": 1.9161904761904763e-06,
      "loss": 0.7946,
      "step": 1122
    },
    {
      "epoch": 0.6915024630541872,
      "grad_norm": 0.9458835124969482,
      "learning_rate": 1.9123809523809527e-06,
      "loss": 0.8413,
      "step": 1123
    },
    {
      "epoch": 0.6921182266009852,
      "grad_norm": 0.9880066514015198,
      "learning_rate": 1.9085714285714287e-06,
      "loss": 0.8815,
      "step": 1124
    },
    {
      "epoch": 0.6927339901477833,
      "grad_norm": 1.0077669620513916,
      "learning_rate": 1.9047619047619047e-06,
      "loss": 0.8391,
      "step": 1125
    },
    {
      "epoch": 0.6933497536945813,
      "grad_norm": 0.9830943942070007,
      "learning_rate": 1.9009523809523811e-06,
      "loss": 0.8641,
      "step": 1126
    },
    {
      "epoch": 0.6939655172413793,
      "grad_norm": 1.03627347946167,
      "learning_rate": 1.8971428571428571e-06,
      "loss": 0.8145,
      "step": 1127
    },
    {
      "epoch": 0.6945812807881774,
      "grad_norm": 1.0804024934768677,
      "learning_rate": 1.8933333333333333e-06,
      "loss": 0.8912,
      "step": 1128
    },
    {
      "epoch": 0.6951970443349754,
      "grad_norm": 1.0266863107681274,
      "learning_rate": 1.8895238095238096e-06,
      "loss": 0.8492,
      "step": 1129
    },
    {
      "epoch": 0.6958128078817734,
      "grad_norm": 0.9590283632278442,
      "learning_rate": 1.8857142857142858e-06,
      "loss": 0.8417,
      "step": 1130
    },
    {
      "epoch": 0.6964285714285714,
      "grad_norm": 1.0092190504074097,
      "learning_rate": 1.8819047619047618e-06,
      "loss": 0.904,
      "step": 1131
    },
    {
      "epoch": 0.6970443349753694,
      "grad_norm": 0.993207573890686,
      "learning_rate": 1.8780952380952382e-06,
      "loss": 0.8487,
      "step": 1132
    },
    {
      "epoch": 0.6976600985221675,
      "grad_norm": 0.9587389826774597,
      "learning_rate": 1.8742857142857142e-06,
      "loss": 0.8428,
      "step": 1133
    },
    {
      "epoch": 0.6982758620689655,
      "grad_norm": 1.0457615852355957,
      "learning_rate": 1.8704761904761906e-06,
      "loss": 0.8585,
      "step": 1134
    },
    {
      "epoch": 0.6988916256157636,
      "grad_norm": 1.0807713270187378,
      "learning_rate": 1.8666666666666667e-06,
      "loss": 0.8462,
      "step": 1135
    },
    {
      "epoch": 0.6995073891625616,
      "grad_norm": 0.9385108351707458,
      "learning_rate": 1.8628571428571429e-06,
      "loss": 0.8495,
      "step": 1136
    },
    {
      "epoch": 0.7001231527093597,
      "grad_norm": 0.9597833156585693,
      "learning_rate": 1.8590476190476193e-06,
      "loss": 0.8331,
      "step": 1137
    },
    {
      "epoch": 0.7007389162561576,
      "grad_norm": 1.0728925466537476,
      "learning_rate": 1.8552380952380953e-06,
      "loss": 0.8705,
      "step": 1138
    },
    {
      "epoch": 0.7013546798029556,
      "grad_norm": 1.0033833980560303,
      "learning_rate": 1.8514285714285713e-06,
      "loss": 0.8758,
      "step": 1139
    },
    {
      "epoch": 0.7019704433497537,
      "grad_norm": 1.0606648921966553,
      "learning_rate": 1.8476190476190477e-06,
      "loss": 0.8682,
      "step": 1140
    },
    {
      "epoch": 0.7025862068965517,
      "grad_norm": 1.000316858291626,
      "learning_rate": 1.8438095238095237e-06,
      "loss": 0.8503,
      "step": 1141
    },
    {
      "epoch": 0.7032019704433498,
      "grad_norm": 1.0293192863464355,
      "learning_rate": 1.84e-06,
      "loss": 0.8064,
      "step": 1142
    },
    {
      "epoch": 0.7038177339901478,
      "grad_norm": 1.059357762336731,
      "learning_rate": 1.8361904761904764e-06,
      "loss": 0.8454,
      "step": 1143
    },
    {
      "epoch": 0.7044334975369458,
      "grad_norm": 1.0345875024795532,
      "learning_rate": 1.8323809523809524e-06,
      "loss": 0.8854,
      "step": 1144
    },
    {
      "epoch": 0.7050492610837439,
      "grad_norm": 0.9679014682769775,
      "learning_rate": 1.8285714285714288e-06,
      "loss": 0.8481,
      "step": 1145
    },
    {
      "epoch": 0.7056650246305419,
      "grad_norm": 0.991224467754364,
      "learning_rate": 1.8247619047619048e-06,
      "loss": 0.8064,
      "step": 1146
    },
    {
      "epoch": 0.7062807881773399,
      "grad_norm": 1.062353253364563,
      "learning_rate": 1.8209523809523808e-06,
      "loss": 0.836,
      "step": 1147
    },
    {
      "epoch": 0.7068965517241379,
      "grad_norm": 1.1197363138198853,
      "learning_rate": 1.8171428571428573e-06,
      "loss": 0.8886,
      "step": 1148
    },
    {
      "epoch": 0.7075123152709359,
      "grad_norm": 1.0739158391952515,
      "learning_rate": 1.8133333333333335e-06,
      "loss": 0.8381,
      "step": 1149
    },
    {
      "epoch": 0.708128078817734,
      "grad_norm": 1.0280282497406006,
      "learning_rate": 1.8095238095238095e-06,
      "loss": 0.8578,
      "step": 1150
    },
    {
      "epoch": 0.708743842364532,
      "grad_norm": 1.0029007196426392,
      "learning_rate": 1.805714285714286e-06,
      "loss": 0.8399,
      "step": 1151
    },
    {
      "epoch": 0.7093596059113301,
      "grad_norm": 1.024682641029358,
      "learning_rate": 1.801904761904762e-06,
      "loss": 0.8864,
      "step": 1152
    },
    {
      "epoch": 0.7099753694581281,
      "grad_norm": 1.1962937116622925,
      "learning_rate": 1.798095238095238e-06,
      "loss": 0.8225,
      "step": 1153
    },
    {
      "epoch": 0.7105911330049262,
      "grad_norm": 1.0528719425201416,
      "learning_rate": 1.7942857142857143e-06,
      "loss": 0.8586,
      "step": 1154
    },
    {
      "epoch": 0.7112068965517241,
      "grad_norm": 1.0293055772781372,
      "learning_rate": 1.7904761904761906e-06,
      "loss": 0.8851,
      "step": 1155
    },
    {
      "epoch": 0.7118226600985221,
      "grad_norm": 1.0167309045791626,
      "learning_rate": 1.7866666666666666e-06,
      "loss": 0.8682,
      "step": 1156
    },
    {
      "epoch": 0.7124384236453202,
      "grad_norm": 1.1520717144012451,
      "learning_rate": 1.782857142857143e-06,
      "loss": 0.8429,
      "step": 1157
    },
    {
      "epoch": 0.7130541871921182,
      "grad_norm": 0.9263671636581421,
      "learning_rate": 1.779047619047619e-06,
      "loss": 0.8609,
      "step": 1158
    },
    {
      "epoch": 0.7136699507389163,
      "grad_norm": 0.9577580094337463,
      "learning_rate": 1.7752380952380954e-06,
      "loss": 0.8437,
      "step": 1159
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 0.9856265783309937,
      "learning_rate": 1.7714285714285714e-06,
      "loss": 0.867,
      "step": 1160
    },
    {
      "epoch": 0.7149014778325123,
      "grad_norm": 1.0278570652008057,
      "learning_rate": 1.7676190476190477e-06,
      "loss": 0.8841,
      "step": 1161
    },
    {
      "epoch": 0.7155172413793104,
      "grad_norm": 0.9736598134040833,
      "learning_rate": 1.763809523809524e-06,
      "loss": 0.8003,
      "step": 1162
    },
    {
      "epoch": 0.7161330049261084,
      "grad_norm": 0.9977447390556335,
      "learning_rate": 1.76e-06,
      "loss": 0.841,
      "step": 1163
    },
    {
      "epoch": 0.7167487684729064,
      "grad_norm": 0.9542884230613708,
      "learning_rate": 1.756190476190476e-06,
      "loss": 0.8574,
      "step": 1164
    },
    {
      "epoch": 0.7173645320197044,
      "grad_norm": 1.0224093198776245,
      "learning_rate": 1.7523809523809525e-06,
      "loss": 0.8829,
      "step": 1165
    },
    {
      "epoch": 0.7179802955665024,
      "grad_norm": 0.9454525709152222,
      "learning_rate": 1.7485714285714285e-06,
      "loss": 0.8528,
      "step": 1166
    },
    {
      "epoch": 0.7185960591133005,
      "grad_norm": 0.9981720447540283,
      "learning_rate": 1.7447619047619047e-06,
      "loss": 0.8307,
      "step": 1167
    },
    {
      "epoch": 0.7192118226600985,
      "grad_norm": 1.0600336790084839,
      "learning_rate": 1.7409523809523812e-06,
      "loss": 0.866,
      "step": 1168
    },
    {
      "epoch": 0.7198275862068966,
      "grad_norm": 1.1206040382385254,
      "learning_rate": 1.7371428571428572e-06,
      "loss": 0.8769,
      "step": 1169
    },
    {
      "epoch": 0.7204433497536946,
      "grad_norm": 1.0140055418014526,
      "learning_rate": 1.7333333333333332e-06,
      "loss": 0.7913,
      "step": 1170
    },
    {
      "epoch": 0.7210591133004927,
      "grad_norm": 0.9890464544296265,
      "learning_rate": 1.7295238095238096e-06,
      "loss": 0.8215,
      "step": 1171
    },
    {
      "epoch": 0.7216748768472906,
      "grad_norm": 1.0246636867523193,
      "learning_rate": 1.7257142857142856e-06,
      "loss": 0.8599,
      "step": 1172
    },
    {
      "epoch": 0.7222906403940886,
      "grad_norm": 1.0513180494308472,
      "learning_rate": 1.721904761904762e-06,
      "loss": 0.8235,
      "step": 1173
    },
    {
      "epoch": 0.7229064039408867,
      "grad_norm": 0.9971263408660889,
      "learning_rate": 1.7180952380952383e-06,
      "loss": 0.8436,
      "step": 1174
    },
    {
      "epoch": 0.7235221674876847,
      "grad_norm": 1.095039963722229,
      "learning_rate": 1.7142857142857143e-06,
      "loss": 0.8492,
      "step": 1175
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 1.0824326276779175,
      "learning_rate": 1.7104761904761907e-06,
      "loss": 0.8659,
      "step": 1176
    },
    {
      "epoch": 0.7247536945812808,
      "grad_norm": 1.0211081504821777,
      "learning_rate": 1.7066666666666667e-06,
      "loss": 0.8409,
      "step": 1177
    },
    {
      "epoch": 0.7253694581280788,
      "grad_norm": 1.0088194608688354,
      "learning_rate": 1.7028571428571427e-06,
      "loss": 0.8238,
      "step": 1178
    },
    {
      "epoch": 0.7259852216748769,
      "grad_norm": 1.0489200353622437,
      "learning_rate": 1.6990476190476191e-06,
      "loss": 0.843,
      "step": 1179
    },
    {
      "epoch": 0.7266009852216748,
      "grad_norm": 0.9931618571281433,
      "learning_rate": 1.6952380952380954e-06,
      "loss": 0.8467,
      "step": 1180
    },
    {
      "epoch": 0.7272167487684729,
      "grad_norm": 0.9898736476898193,
      "learning_rate": 1.6914285714285714e-06,
      "loss": 0.8509,
      "step": 1181
    },
    {
      "epoch": 0.7278325123152709,
      "grad_norm": 1.0575085878372192,
      "learning_rate": 1.6876190476190478e-06,
      "loss": 0.8175,
      "step": 1182
    },
    {
      "epoch": 0.728448275862069,
      "grad_norm": 1.0016741752624512,
      "learning_rate": 1.6838095238095238e-06,
      "loss": 0.838,
      "step": 1183
    },
    {
      "epoch": 0.729064039408867,
      "grad_norm": 1.0701466798782349,
      "learning_rate": 1.6800000000000002e-06,
      "loss": 0.8212,
      "step": 1184
    },
    {
      "epoch": 0.729679802955665,
      "grad_norm": 0.9654819965362549,
      "learning_rate": 1.6761904761904762e-06,
      "loss": 0.8246,
      "step": 1185
    },
    {
      "epoch": 0.7302955665024631,
      "grad_norm": 0.9202854633331299,
      "learning_rate": 1.6723809523809524e-06,
      "loss": 0.8416,
      "step": 1186
    },
    {
      "epoch": 0.7309113300492611,
      "grad_norm": 1.1107659339904785,
      "learning_rate": 1.6685714285714287e-06,
      "loss": 0.8107,
      "step": 1187
    },
    {
      "epoch": 0.7315270935960592,
      "grad_norm": 0.956035315990448,
      "learning_rate": 1.6647619047619049e-06,
      "loss": 0.8462,
      "step": 1188
    },
    {
      "epoch": 0.7321428571428571,
      "grad_norm": 1.1575995683670044,
      "learning_rate": 1.6609523809523809e-06,
      "loss": 0.8664,
      "step": 1189
    },
    {
      "epoch": 0.7327586206896551,
      "grad_norm": 0.9892578125,
      "learning_rate": 1.6571428571428573e-06,
      "loss": 0.8375,
      "step": 1190
    },
    {
      "epoch": 0.7333743842364532,
      "grad_norm": 1.0138314962387085,
      "learning_rate": 1.6533333333333333e-06,
      "loss": 0.8437,
      "step": 1191
    },
    {
      "epoch": 0.7339901477832512,
      "grad_norm": 1.2278560400009155,
      "learning_rate": 1.6495238095238095e-06,
      "loss": 0.8493,
      "step": 1192
    },
    {
      "epoch": 0.7346059113300493,
      "grad_norm": 0.9439584612846375,
      "learning_rate": 1.6457142857142857e-06,
      "loss": 0.8185,
      "step": 1193
    },
    {
      "epoch": 0.7352216748768473,
      "grad_norm": 0.9655652046203613,
      "learning_rate": 1.641904761904762e-06,
      "loss": 0.8439,
      "step": 1194
    },
    {
      "epoch": 0.7358374384236454,
      "grad_norm": 0.9829930067062378,
      "learning_rate": 1.638095238095238e-06,
      "loss": 0.8044,
      "step": 1195
    },
    {
      "epoch": 0.7364532019704434,
      "grad_norm": 1.049998164176941,
      "learning_rate": 1.6342857142857144e-06,
      "loss": 0.813,
      "step": 1196
    },
    {
      "epoch": 0.7370689655172413,
      "grad_norm": 1.0639232397079468,
      "learning_rate": 1.6304761904761904e-06,
      "loss": 0.8583,
      "step": 1197
    },
    {
      "epoch": 0.7376847290640394,
      "grad_norm": 1.3533313274383545,
      "learning_rate": 1.6266666666666668e-06,
      "loss": 0.8608,
      "step": 1198
    },
    {
      "epoch": 0.7383004926108374,
      "grad_norm": 1.0437766313552856,
      "learning_rate": 1.6228571428571428e-06,
      "loss": 0.9128,
      "step": 1199
    },
    {
      "epoch": 0.7389162561576355,
      "grad_norm": 0.9621192216873169,
      "learning_rate": 1.619047619047619e-06,
      "loss": 0.8222,
      "step": 1200
    },
    {
      "epoch": 0.7395320197044335,
      "grad_norm": 0.9232994914054871,
      "learning_rate": 1.6152380952380955e-06,
      "loss": 0.8258,
      "step": 1201
    },
    {
      "epoch": 0.7401477832512315,
      "grad_norm": 0.9646934270858765,
      "learning_rate": 1.6114285714285715e-06,
      "loss": 0.8954,
      "step": 1202
    },
    {
      "epoch": 0.7407635467980296,
      "grad_norm": 0.9813653826713562,
      "learning_rate": 1.6076190476190475e-06,
      "loss": 0.8139,
      "step": 1203
    },
    {
      "epoch": 0.7413793103448276,
      "grad_norm": 0.9908918738365173,
      "learning_rate": 1.603809523809524e-06,
      "loss": 0.8696,
      "step": 1204
    },
    {
      "epoch": 0.7419950738916257,
      "grad_norm": 1.0804203748703003,
      "learning_rate": 1.6e-06,
      "loss": 0.8208,
      "step": 1205
    },
    {
      "epoch": 0.7426108374384236,
      "grad_norm": 0.9613355994224548,
      "learning_rate": 1.5961904761904761e-06,
      "loss": 0.8338,
      "step": 1206
    },
    {
      "epoch": 0.7432266009852216,
      "grad_norm": 0.9932162761688232,
      "learning_rate": 1.5923809523809526e-06,
      "loss": 0.8034,
      "step": 1207
    },
    {
      "epoch": 0.7438423645320197,
      "grad_norm": 1.1329025030136108,
      "learning_rate": 1.5885714285714286e-06,
      "loss": 0.8219,
      "step": 1208
    },
    {
      "epoch": 0.7444581280788177,
      "grad_norm": 1.032566785812378,
      "learning_rate": 1.584761904761905e-06,
      "loss": 0.8134,
      "step": 1209
    },
    {
      "epoch": 0.7450738916256158,
      "grad_norm": 1.0814666748046875,
      "learning_rate": 1.580952380952381e-06,
      "loss": 0.881,
      "step": 1210
    },
    {
      "epoch": 0.7456896551724138,
      "grad_norm": 0.9292041063308716,
      "learning_rate": 1.577142857142857e-06,
      "loss": 0.8635,
      "step": 1211
    },
    {
      "epoch": 0.7463054187192119,
      "grad_norm": 1.037340521812439,
      "learning_rate": 1.5733333333333334e-06,
      "loss": 0.7921,
      "step": 1212
    },
    {
      "epoch": 0.7469211822660099,
      "grad_norm": 1.1076730489730835,
      "learning_rate": 1.5695238095238097e-06,
      "loss": 0.7968,
      "step": 1213
    },
    {
      "epoch": 0.7475369458128078,
      "grad_norm": 1.0760498046875,
      "learning_rate": 1.5657142857142857e-06,
      "loss": 0.8783,
      "step": 1214
    },
    {
      "epoch": 0.7481527093596059,
      "grad_norm": 1.0263563394546509,
      "learning_rate": 1.561904761904762e-06,
      "loss": 0.804,
      "step": 1215
    },
    {
      "epoch": 0.7487684729064039,
      "grad_norm": 1.0080331563949585,
      "learning_rate": 1.558095238095238e-06,
      "loss": 0.8224,
      "step": 1216
    },
    {
      "epoch": 0.749384236453202,
      "grad_norm": 0.9433367252349854,
      "learning_rate": 1.5542857142857141e-06,
      "loss": 0.8932,
      "step": 1217
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0507266521453857,
      "learning_rate": 1.5504761904761905e-06,
      "loss": 0.8108,
      "step": 1218
    },
    {
      "epoch": 0.75,
      "eval_loss": 0.8714036345481873,
      "eval_runtime": 268.7086,
      "eval_samples_per_second": 86.004,
      "eval_steps_per_second": 2.691,
      "step": 1218
    },
    {
      "epoch": 0.750615763546798,
      "grad_norm": 0.924110472202301,
      "learning_rate": 1.5466666666666668e-06,
      "loss": 0.8151,
      "step": 1219
    },
    {
      "epoch": 0.7512315270935961,
      "grad_norm": 0.9699553847312927,
      "learning_rate": 1.5428571428571428e-06,
      "loss": 0.8573,
      "step": 1220
    },
    {
      "epoch": 0.7518472906403941,
      "grad_norm": 1.0396736860275269,
      "learning_rate": 1.5390476190476192e-06,
      "loss": 0.8108,
      "step": 1221
    },
    {
      "epoch": 0.7524630541871922,
      "grad_norm": 1.0152760744094849,
      "learning_rate": 1.5352380952380952e-06,
      "loss": 0.8489,
      "step": 1222
    },
    {
      "epoch": 0.7530788177339901,
      "grad_norm": 1.0237215757369995,
      "learning_rate": 1.5314285714285716e-06,
      "loss": 0.8251,
      "step": 1223
    },
    {
      "epoch": 0.7536945812807881,
      "grad_norm": 0.9972196221351624,
      "learning_rate": 1.5276190476190476e-06,
      "loss": 0.8348,
      "step": 1224
    },
    {
      "epoch": 0.7543103448275862,
      "grad_norm": 0.9697617888450623,
      "learning_rate": 1.5238095238095238e-06,
      "loss": 0.8326,
      "step": 1225
    },
    {
      "epoch": 0.7549261083743842,
      "grad_norm": 0.9744370579719543,
      "learning_rate": 1.5200000000000003e-06,
      "loss": 0.8045,
      "step": 1226
    },
    {
      "epoch": 0.7555418719211823,
      "grad_norm": 1.0161895751953125,
      "learning_rate": 1.5161904761904763e-06,
      "loss": 0.855,
      "step": 1227
    },
    {
      "epoch": 0.7561576354679803,
      "grad_norm": 0.9612058997154236,
      "learning_rate": 1.5123809523809523e-06,
      "loss": 0.8507,
      "step": 1228
    },
    {
      "epoch": 0.7567733990147784,
      "grad_norm": 0.9412856698036194,
      "learning_rate": 1.5085714285714287e-06,
      "loss": 0.8534,
      "step": 1229
    },
    {
      "epoch": 0.7573891625615764,
      "grad_norm": 1.0300320386886597,
      "learning_rate": 1.5047619047619047e-06,
      "loss": 0.8535,
      "step": 1230
    },
    {
      "epoch": 0.7580049261083743,
      "grad_norm": 0.9363780617713928,
      "learning_rate": 1.500952380952381e-06,
      "loss": 0.8559,
      "step": 1231
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 1.0474114418029785,
      "learning_rate": 1.4971428571428574e-06,
      "loss": 0.8933,
      "step": 1232
    },
    {
      "epoch": 0.7592364532019704,
      "grad_norm": 1.1414858102798462,
      "learning_rate": 1.4933333333333334e-06,
      "loss": 0.868,
      "step": 1233
    },
    {
      "epoch": 0.7598522167487685,
      "grad_norm": 1.0124553442001343,
      "learning_rate": 1.4895238095238096e-06,
      "loss": 0.8461,
      "step": 1234
    },
    {
      "epoch": 0.7604679802955665,
      "grad_norm": 1.0843786001205444,
      "learning_rate": 1.4857142857142858e-06,
      "loss": 0.8409,
      "step": 1235
    },
    {
      "epoch": 0.7610837438423645,
      "grad_norm": 0.9127656817436218,
      "learning_rate": 1.4819047619047618e-06,
      "loss": 0.8779,
      "step": 1236
    },
    {
      "epoch": 0.7616995073891626,
      "grad_norm": 0.9803403615951538,
      "learning_rate": 1.478095238095238e-06,
      "loss": 0.8625,
      "step": 1237
    },
    {
      "epoch": 0.7623152709359606,
      "grad_norm": 1.021360993385315,
      "learning_rate": 1.4742857142857144e-06,
      "loss": 0.8644,
      "step": 1238
    },
    {
      "epoch": 0.7629310344827587,
      "grad_norm": 1.0375638008117676,
      "learning_rate": 1.4704761904761907e-06,
      "loss": 0.8669,
      "step": 1239
    },
    {
      "epoch": 0.7635467980295566,
      "grad_norm": 1.06625235080719,
      "learning_rate": 1.4666666666666667e-06,
      "loss": 0.8529,
      "step": 1240
    },
    {
      "epoch": 0.7641625615763546,
      "grad_norm": 0.9876265525817871,
      "learning_rate": 1.4628571428571429e-06,
      "loss": 0.872,
      "step": 1241
    },
    {
      "epoch": 0.7647783251231527,
      "grad_norm": 1.0873311758041382,
      "learning_rate": 1.459047619047619e-06,
      "loss": 0.8745,
      "step": 1242
    },
    {
      "epoch": 0.7653940886699507,
      "grad_norm": 0.96006840467453,
      "learning_rate": 1.4552380952380951e-06,
      "loss": 0.8269,
      "step": 1243
    },
    {
      "epoch": 0.7660098522167488,
      "grad_norm": 1.0414228439331055,
      "learning_rate": 1.4514285714285715e-06,
      "loss": 0.8362,
      "step": 1244
    },
    {
      "epoch": 0.7666256157635468,
      "grad_norm": 1.05013906955719,
      "learning_rate": 1.4476190476190478e-06,
      "loss": 0.8294,
      "step": 1245
    },
    {
      "epoch": 0.7672413793103449,
      "grad_norm": 1.016183853149414,
      "learning_rate": 1.443809523809524e-06,
      "loss": 0.8336,
      "step": 1246
    },
    {
      "epoch": 0.7678571428571429,
      "grad_norm": 0.9257124066352844,
      "learning_rate": 1.44e-06,
      "loss": 0.8509,
      "step": 1247
    },
    {
      "epoch": 0.7684729064039408,
      "grad_norm": 1.001266598701477,
      "learning_rate": 1.4361904761904762e-06,
      "loss": 0.8477,
      "step": 1248
    },
    {
      "epoch": 0.7690886699507389,
      "grad_norm": 0.9803850650787354,
      "learning_rate": 1.4323809523809524e-06,
      "loss": 0.827,
      "step": 1249
    },
    {
      "epoch": 0.7697044334975369,
      "grad_norm": 0.97281414270401,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.8716,
      "step": 1250
    },
    {
      "epoch": 0.770320197044335,
      "grad_norm": 1.0978363752365112,
      "learning_rate": 1.4247619047619048e-06,
      "loss": 0.852,
      "step": 1251
    },
    {
      "epoch": 0.770935960591133,
      "grad_norm": 1.0431822538375854,
      "learning_rate": 1.420952380952381e-06,
      "loss": 0.8479,
      "step": 1252
    },
    {
      "epoch": 0.771551724137931,
      "grad_norm": 1.059956669807434,
      "learning_rate": 1.4171428571428573e-06,
      "loss": 0.8602,
      "step": 1253
    },
    {
      "epoch": 0.7721674876847291,
      "grad_norm": 1.105189561843872,
      "learning_rate": 1.4133333333333333e-06,
      "loss": 0.8811,
      "step": 1254
    },
    {
      "epoch": 0.7727832512315271,
      "grad_norm": 0.9794146418571472,
      "learning_rate": 1.4095238095238095e-06,
      "loss": 0.7944,
      "step": 1255
    },
    {
      "epoch": 0.7733990147783252,
      "grad_norm": 0.9872285723686218,
      "learning_rate": 1.4057142857142857e-06,
      "loss": 0.8112,
      "step": 1256
    },
    {
      "epoch": 0.7740147783251231,
      "grad_norm": 1.062320351600647,
      "learning_rate": 1.401904761904762e-06,
      "loss": 0.8359,
      "step": 1257
    },
    {
      "epoch": 0.7746305418719212,
      "grad_norm": 0.9300822615623474,
      "learning_rate": 1.3980952380952382e-06,
      "loss": 0.8021,
      "step": 1258
    },
    {
      "epoch": 0.7752463054187192,
      "grad_norm": 0.9956058263778687,
      "learning_rate": 1.3942857142857144e-06,
      "loss": 0.8965,
      "step": 1259
    },
    {
      "epoch": 0.7758620689655172,
      "grad_norm": 1.0262537002563477,
      "learning_rate": 1.3904761904761906e-06,
      "loss": 0.7731,
      "step": 1260
    },
    {
      "epoch": 0.7764778325123153,
      "grad_norm": 1.0473467111587524,
      "learning_rate": 1.3866666666666666e-06,
      "loss": 0.8702,
      "step": 1261
    },
    {
      "epoch": 0.7770935960591133,
      "grad_norm": 0.9722353219985962,
      "learning_rate": 1.3828571428571428e-06,
      "loss": 0.8383,
      "step": 1262
    },
    {
      "epoch": 0.7777093596059114,
      "grad_norm": 1.03757905960083,
      "learning_rate": 1.379047619047619e-06,
      "loss": 0.854,
      "step": 1263
    },
    {
      "epoch": 0.7783251231527094,
      "grad_norm": 0.9498728513717651,
      "learning_rate": 1.3752380952380955e-06,
      "loss": 0.8203,
      "step": 1264
    },
    {
      "epoch": 0.7789408866995073,
      "grad_norm": 1.0383034944534302,
      "learning_rate": 1.3714285714285715e-06,
      "loss": 0.8023,
      "step": 1265
    },
    {
      "epoch": 0.7795566502463054,
      "grad_norm": 0.9646442532539368,
      "learning_rate": 1.3676190476190477e-06,
      "loss": 0.8454,
      "step": 1266
    },
    {
      "epoch": 0.7801724137931034,
      "grad_norm": 0.949752926826477,
      "learning_rate": 1.3638095238095239e-06,
      "loss": 0.8003,
      "step": 1267
    },
    {
      "epoch": 0.7807881773399015,
      "grad_norm": 0.9708337783813477,
      "learning_rate": 1.36e-06,
      "loss": 0.8299,
      "step": 1268
    },
    {
      "epoch": 0.7814039408866995,
      "grad_norm": 0.9988180994987488,
      "learning_rate": 1.3561904761904761e-06,
      "loss": 0.8636,
      "step": 1269
    },
    {
      "epoch": 0.7820197044334976,
      "grad_norm": 0.9207890033721924,
      "learning_rate": 1.3523809523809525e-06,
      "loss": 0.8425,
      "step": 1270
    },
    {
      "epoch": 0.7826354679802956,
      "grad_norm": 1.053898572921753,
      "learning_rate": 1.3485714285714288e-06,
      "loss": 0.8486,
      "step": 1271
    },
    {
      "epoch": 0.7832512315270936,
      "grad_norm": 0.9591391086578369,
      "learning_rate": 1.3447619047619048e-06,
      "loss": 0.8075,
      "step": 1272
    },
    {
      "epoch": 0.7838669950738916,
      "grad_norm": 1.0271767377853394,
      "learning_rate": 1.340952380952381e-06,
      "loss": 0.8378,
      "step": 1273
    },
    {
      "epoch": 0.7844827586206896,
      "grad_norm": 1.0015199184417725,
      "learning_rate": 1.3371428571428572e-06,
      "loss": 0.8419,
      "step": 1274
    },
    {
      "epoch": 0.7850985221674877,
      "grad_norm": 0.9617854356765747,
      "learning_rate": 1.3333333333333332e-06,
      "loss": 0.87,
      "step": 1275
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 1.0176334381103516,
      "learning_rate": 1.3295238095238096e-06,
      "loss": 0.8236,
      "step": 1276
    },
    {
      "epoch": 0.7863300492610837,
      "grad_norm": 0.9947876930236816,
      "learning_rate": 1.3257142857142858e-06,
      "loss": 0.8143,
      "step": 1277
    },
    {
      "epoch": 0.7869458128078818,
      "grad_norm": 0.9810738563537598,
      "learning_rate": 1.321904761904762e-06,
      "loss": 0.8396,
      "step": 1278
    },
    {
      "epoch": 0.7875615763546798,
      "grad_norm": 0.9892545938491821,
      "learning_rate": 1.318095238095238e-06,
      "loss": 0.8762,
      "step": 1279
    },
    {
      "epoch": 0.7881773399014779,
      "grad_norm": 1.0067731142044067,
      "learning_rate": 1.3142857142857143e-06,
      "loss": 0.8311,
      "step": 1280
    },
    {
      "epoch": 0.7887931034482759,
      "grad_norm": 0.9148168563842773,
      "learning_rate": 1.3104761904761905e-06,
      "loss": 0.8799,
      "step": 1281
    },
    {
      "epoch": 0.7894088669950738,
      "grad_norm": 0.9745421409606934,
      "learning_rate": 1.3066666666666667e-06,
      "loss": 0.8535,
      "step": 1282
    },
    {
      "epoch": 0.7900246305418719,
      "grad_norm": 0.9474236965179443,
      "learning_rate": 1.302857142857143e-06,
      "loss": 0.7963,
      "step": 1283
    },
    {
      "epoch": 0.7906403940886699,
      "grad_norm": 1.1620914936065674,
      "learning_rate": 1.2990476190476192e-06,
      "loss": 0.7875,
      "step": 1284
    },
    {
      "epoch": 0.791256157635468,
      "grad_norm": 1.0521327257156372,
      "learning_rate": 1.2952380952380954e-06,
      "loss": 0.8079,
      "step": 1285
    },
    {
      "epoch": 0.791871921182266,
      "grad_norm": 0.9932060837745667,
      "learning_rate": 1.2914285714285714e-06,
      "loss": 0.8709,
      "step": 1286
    },
    {
      "epoch": 0.7924876847290641,
      "grad_norm": 1.0757747888565063,
      "learning_rate": 1.2876190476190476e-06,
      "loss": 0.8071,
      "step": 1287
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 1.1082212924957275,
      "learning_rate": 1.2838095238095238e-06,
      "loss": 0.886,
      "step": 1288
    },
    {
      "epoch": 0.7937192118226601,
      "grad_norm": 1.0677675008773804,
      "learning_rate": 1.28e-06,
      "loss": 0.8583,
      "step": 1289
    },
    {
      "epoch": 0.7943349753694581,
      "grad_norm": 0.991081178188324,
      "learning_rate": 1.2761904761904762e-06,
      "loss": 0.8384,
      "step": 1290
    },
    {
      "epoch": 0.7949507389162561,
      "grad_norm": 1.0262521505355835,
      "learning_rate": 1.2723809523809525e-06,
      "loss": 0.8295,
      "step": 1291
    },
    {
      "epoch": 0.7955665024630542,
      "grad_norm": 0.9731985926628113,
      "learning_rate": 1.2685714285714287e-06,
      "loss": 0.8469,
      "step": 1292
    },
    {
      "epoch": 0.7961822660098522,
      "grad_norm": 1.072624921798706,
      "learning_rate": 1.2647619047619047e-06,
      "loss": 0.8199,
      "step": 1293
    },
    {
      "epoch": 0.7967980295566502,
      "grad_norm": 1.273785948753357,
      "learning_rate": 1.260952380952381e-06,
      "loss": 0.823,
      "step": 1294
    },
    {
      "epoch": 0.7974137931034483,
      "grad_norm": 0.9685826301574707,
      "learning_rate": 1.2571428571428571e-06,
      "loss": 0.8211,
      "step": 1295
    },
    {
      "epoch": 0.7980295566502463,
      "grad_norm": 0.9549744129180908,
      "learning_rate": 1.2533333333333335e-06,
      "loss": 0.8495,
      "step": 1296
    },
    {
      "epoch": 0.7986453201970444,
      "grad_norm": 0.9146218299865723,
      "learning_rate": 1.2495238095238095e-06,
      "loss": 0.8583,
      "step": 1297
    },
    {
      "epoch": 0.7992610837438424,
      "grad_norm": 0.9252996444702148,
      "learning_rate": 1.2457142857142858e-06,
      "loss": 0.8775,
      "step": 1298
    },
    {
      "epoch": 0.7998768472906403,
      "grad_norm": 0.893285870552063,
      "learning_rate": 1.241904761904762e-06,
      "loss": 0.8197,
      "step": 1299
    },
    {
      "epoch": 0.8004926108374384,
      "grad_norm": 0.9788442850112915,
      "learning_rate": 1.238095238095238e-06,
      "loss": 0.7987,
      "step": 1300
    },
    {
      "epoch": 0.8011083743842364,
      "grad_norm": 1.0793761014938354,
      "learning_rate": 1.2342857142857142e-06,
      "loss": 0.8373,
      "step": 1301
    },
    {
      "epoch": 0.8017241379310345,
      "grad_norm": 1.0503733158111572,
      "learning_rate": 1.2304761904761906e-06,
      "loss": 0.8771,
      "step": 1302
    },
    {
      "epoch": 0.8023399014778325,
      "grad_norm": 0.9789201617240906,
      "learning_rate": 1.2266666666666669e-06,
      "loss": 0.8543,
      "step": 1303
    },
    {
      "epoch": 0.8029556650246306,
      "grad_norm": 0.9756903052330017,
      "learning_rate": 1.2228571428571429e-06,
      "loss": 0.8314,
      "step": 1304
    },
    {
      "epoch": 0.8035714285714286,
      "grad_norm": 0.9995084404945374,
      "learning_rate": 1.219047619047619e-06,
      "loss": 0.8268,
      "step": 1305
    },
    {
      "epoch": 0.8041871921182266,
      "grad_norm": 0.9603376388549805,
      "learning_rate": 1.2152380952380953e-06,
      "loss": 0.8247,
      "step": 1306
    },
    {
      "epoch": 0.8048029556650246,
      "grad_norm": 0.8635754585266113,
      "learning_rate": 1.2114285714285713e-06,
      "loss": 0.8552,
      "step": 1307
    },
    {
      "epoch": 0.8054187192118226,
      "grad_norm": 1.0747708082199097,
      "learning_rate": 1.2076190476190477e-06,
      "loss": 0.8459,
      "step": 1308
    },
    {
      "epoch": 0.8060344827586207,
      "grad_norm": 0.9503686428070068,
      "learning_rate": 1.203809523809524e-06,
      "loss": 0.8579,
      "step": 1309
    },
    {
      "epoch": 0.8066502463054187,
      "grad_norm": 1.0363603830337524,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.8874,
      "step": 1310
    },
    {
      "epoch": 0.8072660098522167,
      "grad_norm": 0.9794528484344482,
      "learning_rate": 1.1961904761904762e-06,
      "loss": 0.8102,
      "step": 1311
    },
    {
      "epoch": 0.8078817733990148,
      "grad_norm": 0.9862688183784485,
      "learning_rate": 1.1923809523809524e-06,
      "loss": 0.8209,
      "step": 1312
    },
    {
      "epoch": 0.8084975369458128,
      "grad_norm": 1.0676308870315552,
      "learning_rate": 1.1885714285714286e-06,
      "loss": 0.8431,
      "step": 1313
    },
    {
      "epoch": 0.8091133004926109,
      "grad_norm": 1.0343494415283203,
      "learning_rate": 1.1847619047619048e-06,
      "loss": 0.7977,
      "step": 1314
    },
    {
      "epoch": 0.8097290640394089,
      "grad_norm": 1.05465829372406,
      "learning_rate": 1.180952380952381e-06,
      "loss": 0.8442,
      "step": 1315
    },
    {
      "epoch": 0.8103448275862069,
      "grad_norm": 1.0937306880950928,
      "learning_rate": 1.1771428571428572e-06,
      "loss": 0.8024,
      "step": 1316
    },
    {
      "epoch": 0.8109605911330049,
      "grad_norm": 1.0709527730941772,
      "learning_rate": 1.1733333333333335e-06,
      "loss": 0.8015,
      "step": 1317
    },
    {
      "epoch": 0.8115763546798029,
      "grad_norm": 1.0188068151474,
      "learning_rate": 1.1695238095238095e-06,
      "loss": 0.8615,
      "step": 1318
    },
    {
      "epoch": 0.812192118226601,
      "grad_norm": 0.9486567378044128,
      "learning_rate": 1.1657142857142857e-06,
      "loss": 0.8205,
      "step": 1319
    },
    {
      "epoch": 0.812807881773399,
      "grad_norm": 1.0669846534729004,
      "learning_rate": 1.161904761904762e-06,
      "loss": 0.809,
      "step": 1320
    },
    {
      "epoch": 0.8134236453201971,
      "grad_norm": 1.0672231912612915,
      "learning_rate": 1.1580952380952381e-06,
      "loss": 0.8575,
      "step": 1321
    },
    {
      "epoch": 0.8140394088669951,
      "grad_norm": 1.0085557699203491,
      "learning_rate": 1.1542857142857143e-06,
      "loss": 0.8559,
      "step": 1322
    },
    {
      "epoch": 0.8146551724137931,
      "grad_norm": 0.9712440371513367,
      "learning_rate": 1.1504761904761906e-06,
      "loss": 0.8159,
      "step": 1323
    },
    {
      "epoch": 0.8152709359605911,
      "grad_norm": 1.1090574264526367,
      "learning_rate": 1.1466666666666668e-06,
      "loss": 0.7954,
      "step": 1324
    },
    {
      "epoch": 0.8158866995073891,
      "grad_norm": 0.9814254641532898,
      "learning_rate": 1.1428571428571428e-06,
      "loss": 0.8186,
      "step": 1325
    },
    {
      "epoch": 0.8165024630541872,
      "grad_norm": 1.0256515741348267,
      "learning_rate": 1.139047619047619e-06,
      "loss": 0.8758,
      "step": 1326
    },
    {
      "epoch": 0.8171182266009852,
      "grad_norm": 0.9528703689575195,
      "learning_rate": 1.1352380952380952e-06,
      "loss": 0.8158,
      "step": 1327
    },
    {
      "epoch": 0.8177339901477833,
      "grad_norm": 0.9408023357391357,
      "learning_rate": 1.1314285714285716e-06,
      "loss": 0.8732,
      "step": 1328
    },
    {
      "epoch": 0.8183497536945813,
      "grad_norm": 0.981395959854126,
      "learning_rate": 1.1276190476190476e-06,
      "loss": 0.8442,
      "step": 1329
    },
    {
      "epoch": 0.8189655172413793,
      "grad_norm": 0.9333049058914185,
      "learning_rate": 1.1238095238095239e-06,
      "loss": 0.8293,
      "step": 1330
    },
    {
      "epoch": 0.8195812807881774,
      "grad_norm": 1.0591691732406616,
      "learning_rate": 1.12e-06,
      "loss": 0.8381,
      "step": 1331
    },
    {
      "epoch": 0.8201970443349754,
      "grad_norm": 1.0735316276550293,
      "learning_rate": 1.116190476190476e-06,
      "loss": 0.8222,
      "step": 1332
    },
    {
      "epoch": 0.8208128078817734,
      "grad_norm": 1.0112611055374146,
      "learning_rate": 1.1123809523809523e-06,
      "loss": 0.8251,
      "step": 1333
    },
    {
      "epoch": 0.8214285714285714,
      "grad_norm": 0.9757543206214905,
      "learning_rate": 1.1085714285714287e-06,
      "loss": 0.7681,
      "step": 1334
    },
    {
      "epoch": 0.8220443349753694,
      "grad_norm": 0.9355512261390686,
      "learning_rate": 1.104761904761905e-06,
      "loss": 0.8612,
      "step": 1335
    },
    {
      "epoch": 0.8226600985221675,
      "grad_norm": 1.0624736547470093,
      "learning_rate": 1.100952380952381e-06,
      "loss": 0.8264,
      "step": 1336
    },
    {
      "epoch": 0.8232758620689655,
      "grad_norm": 1.1398824453353882,
      "learning_rate": 1.0971428571428572e-06,
      "loss": 0.8333,
      "step": 1337
    },
    {
      "epoch": 0.8238916256157636,
      "grad_norm": 1.0618537664413452,
      "learning_rate": 1.0933333333333334e-06,
      "loss": 0.8497,
      "step": 1338
    },
    {
      "epoch": 0.8245073891625616,
      "grad_norm": 0.9265748262405396,
      "learning_rate": 1.0895238095238094e-06,
      "loss": 0.8398,
      "step": 1339
    },
    {
      "epoch": 0.8251231527093597,
      "grad_norm": 1.1633543968200684,
      "learning_rate": 1.0857142857142858e-06,
      "loss": 0.8226,
      "step": 1340
    },
    {
      "epoch": 0.8257389162561576,
      "grad_norm": 0.9384273290634155,
      "learning_rate": 1.081904761904762e-06,
      "loss": 0.8538,
      "step": 1341
    },
    {
      "epoch": 0.8263546798029556,
      "grad_norm": 1.067854642868042,
      "learning_rate": 1.0780952380952383e-06,
      "loss": 0.8198,
      "step": 1342
    },
    {
      "epoch": 0.8269704433497537,
      "grad_norm": 1.009758472442627,
      "learning_rate": 1.0742857142857143e-06,
      "loss": 0.8696,
      "step": 1343
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 1.0183945894241333,
      "learning_rate": 1.0704761904761905e-06,
      "loss": 0.8609,
      "step": 1344
    },
    {
      "epoch": 0.8282019704433498,
      "grad_norm": 1.0237149000167847,
      "learning_rate": 1.0666666666666667e-06,
      "loss": 0.8581,
      "step": 1345
    },
    {
      "epoch": 0.8288177339901478,
      "grad_norm": 1.0161064863204956,
      "learning_rate": 1.062857142857143e-06,
      "loss": 0.8398,
      "step": 1346
    },
    {
      "epoch": 0.8294334975369458,
      "grad_norm": 0.9370342493057251,
      "learning_rate": 1.0590476190476191e-06,
      "loss": 0.823,
      "step": 1347
    },
    {
      "epoch": 0.8300492610837439,
      "grad_norm": 0.9428346753120422,
      "learning_rate": 1.0552380952380953e-06,
      "loss": 0.8726,
      "step": 1348
    },
    {
      "epoch": 0.8306650246305419,
      "grad_norm": 1.0794907808303833,
      "learning_rate": 1.0514285714285716e-06,
      "loss": 0.8568,
      "step": 1349
    },
    {
      "epoch": 0.8312807881773399,
      "grad_norm": 1.0499330759048462,
      "learning_rate": 1.0476190476190476e-06,
      "loss": 0.8838,
      "step": 1350
    },
    {
      "epoch": 0.8318965517241379,
      "grad_norm": 1.0859465599060059,
      "learning_rate": 1.0438095238095238e-06,
      "loss": 0.8602,
      "step": 1351
    },
    {
      "epoch": 0.8325123152709359,
      "grad_norm": 0.9156026244163513,
      "learning_rate": 1.04e-06,
      "loss": 0.8098,
      "step": 1352
    },
    {
      "epoch": 0.833128078817734,
      "grad_norm": 1.0595169067382812,
      "learning_rate": 1.0361904761904762e-06,
      "loss": 0.8448,
      "step": 1353
    },
    {
      "epoch": 0.833743842364532,
      "grad_norm": 0.9834839105606079,
      "learning_rate": 1.0323809523809524e-06,
      "loss": 0.8443,
      "step": 1354
    },
    {
      "epoch": 0.8343596059113301,
      "grad_norm": 1.0067179203033447,
      "learning_rate": 1.0285714285714286e-06,
      "loss": 0.8085,
      "step": 1355
    },
    {
      "epoch": 0.8349753694581281,
      "grad_norm": 0.9367899894714355,
      "learning_rate": 1.0247619047619049e-06,
      "loss": 0.8243,
      "step": 1356
    },
    {
      "epoch": 0.8355911330049262,
      "grad_norm": 1.0154826641082764,
      "learning_rate": 1.0209523809523809e-06,
      "loss": 0.8181,
      "step": 1357
    },
    {
      "epoch": 0.8362068965517241,
      "grad_norm": 0.9664872884750366,
      "learning_rate": 1.017142857142857e-06,
      "loss": 0.8339,
      "step": 1358
    },
    {
      "epoch": 0.8368226600985221,
      "grad_norm": 1.0317003726959229,
      "learning_rate": 1.0133333333333333e-06,
      "loss": 0.8576,
      "step": 1359
    },
    {
      "epoch": 0.8374384236453202,
      "grad_norm": 0.9899953007698059,
      "learning_rate": 1.0095238095238097e-06,
      "loss": 0.8345,
      "step": 1360
    },
    {
      "epoch": 0.8380541871921182,
      "grad_norm": 0.9646126627922058,
      "learning_rate": 1.0057142857142857e-06,
      "loss": 0.7806,
      "step": 1361
    },
    {
      "epoch": 0.8386699507389163,
      "grad_norm": 1.014291524887085,
      "learning_rate": 1.001904761904762e-06,
      "loss": 0.7874,
      "step": 1362
    },
    {
      "epoch": 0.8392857142857143,
      "grad_norm": 0.9843891263008118,
      "learning_rate": 9.980952380952382e-07,
      "loss": 0.8837,
      "step": 1363
    },
    {
      "epoch": 0.8399014778325123,
      "grad_norm": 1.1121692657470703,
      "learning_rate": 9.942857142857142e-07,
      "loss": 0.8286,
      "step": 1364
    },
    {
      "epoch": 0.8405172413793104,
      "grad_norm": 0.9821648597717285,
      "learning_rate": 9.904761904761904e-07,
      "loss": 0.8169,
      "step": 1365
    },
    {
      "epoch": 0.8411330049261084,
      "grad_norm": 1.0404927730560303,
      "learning_rate": 9.866666666666668e-07,
      "loss": 0.8414,
      "step": 1366
    },
    {
      "epoch": 0.8417487684729064,
      "grad_norm": 1.027420997619629,
      "learning_rate": 9.82857142857143e-07,
      "loss": 0.8255,
      "step": 1367
    },
    {
      "epoch": 0.8423645320197044,
      "grad_norm": 0.9649743437767029,
      "learning_rate": 9.79047619047619e-07,
      "loss": 0.834,
      "step": 1368
    },
    {
      "epoch": 0.8429802955665024,
      "grad_norm": 0.932175874710083,
      "learning_rate": 9.752380952380953e-07,
      "loss": 0.8762,
      "step": 1369
    },
    {
      "epoch": 0.8435960591133005,
      "grad_norm": 0.9331344366073608,
      "learning_rate": 9.714285714285715e-07,
      "loss": 0.8331,
      "step": 1370
    },
    {
      "epoch": 0.8442118226600985,
      "grad_norm": 0.9597103595733643,
      "learning_rate": 9.676190476190475e-07,
      "loss": 0.7841,
      "step": 1371
    },
    {
      "epoch": 0.8448275862068966,
      "grad_norm": 0.8880664110183716,
      "learning_rate": 9.63809523809524e-07,
      "loss": 0.8272,
      "step": 1372
    },
    {
      "epoch": 0.8454433497536946,
      "grad_norm": 0.908309280872345,
      "learning_rate": 9.600000000000001e-07,
      "loss": 0.8303,
      "step": 1373
    },
    {
      "epoch": 0.8460591133004927,
      "grad_norm": 0.9371874332427979,
      "learning_rate": 9.561904761904763e-07,
      "loss": 0.8459,
      "step": 1374
    },
    {
      "epoch": 0.8466748768472906,
      "grad_norm": 0.9983593225479126,
      "learning_rate": 9.523809523809523e-07,
      "loss": 0.8475,
      "step": 1375
    },
    {
      "epoch": 0.8472906403940886,
      "grad_norm": 1.0846425294876099,
      "learning_rate": 9.485714285714286e-07,
      "loss": 0.8167,
      "step": 1376
    },
    {
      "epoch": 0.8479064039408867,
      "grad_norm": 0.9565457701683044,
      "learning_rate": 9.447619047619048e-07,
      "loss": 0.8851,
      "step": 1377
    },
    {
      "epoch": 0.8485221674876847,
      "grad_norm": 1.0812933444976807,
      "learning_rate": 9.409523809523809e-07,
      "loss": 0.8705,
      "step": 1378
    },
    {
      "epoch": 0.8491379310344828,
      "grad_norm": 0.9405567049980164,
      "learning_rate": 9.371428571428571e-07,
      "loss": 0.8668,
      "step": 1379
    },
    {
      "epoch": 0.8497536945812808,
      "grad_norm": 0.9964160323143005,
      "learning_rate": 9.333333333333333e-07,
      "loss": 0.8313,
      "step": 1380
    },
    {
      "epoch": 0.8503694581280788,
      "grad_norm": 0.9516879916191101,
      "learning_rate": 9.295238095238096e-07,
      "loss": 0.7873,
      "step": 1381
    },
    {
      "epoch": 0.8509852216748769,
      "grad_norm": 1.0105963945388794,
      "learning_rate": 9.257142857142857e-07,
      "loss": 0.8552,
      "step": 1382
    },
    {
      "epoch": 0.8516009852216748,
      "grad_norm": 0.9788857698440552,
      "learning_rate": 9.219047619047619e-07,
      "loss": 0.8636,
      "step": 1383
    },
    {
      "epoch": 0.8522167487684729,
      "grad_norm": 1.0309922695159912,
      "learning_rate": 9.180952380952382e-07,
      "loss": 0.8382,
      "step": 1384
    },
    {
      "epoch": 0.8528325123152709,
      "grad_norm": 1.0769717693328857,
      "learning_rate": 9.142857142857144e-07,
      "loss": 0.8293,
      "step": 1385
    },
    {
      "epoch": 0.853448275862069,
      "grad_norm": 1.0496612787246704,
      "learning_rate": 9.104761904761904e-07,
      "loss": 0.8059,
      "step": 1386
    },
    {
      "epoch": 0.854064039408867,
      "grad_norm": 1.0084871053695679,
      "learning_rate": 9.066666666666667e-07,
      "loss": 0.8184,
      "step": 1387
    },
    {
      "epoch": 0.854679802955665,
      "grad_norm": 1.1086969375610352,
      "learning_rate": 9.02857142857143e-07,
      "loss": 0.8347,
      "step": 1388
    },
    {
      "epoch": 0.8552955665024631,
      "grad_norm": 0.9990587830543518,
      "learning_rate": 8.99047619047619e-07,
      "loss": 0.8457,
      "step": 1389
    },
    {
      "epoch": 0.8559113300492611,
      "grad_norm": 1.1133555173873901,
      "learning_rate": 8.952380952380953e-07,
      "loss": 0.8118,
      "step": 1390
    },
    {
      "epoch": 0.8565270935960592,
      "grad_norm": 0.9758033156394958,
      "learning_rate": 8.914285714285715e-07,
      "loss": 0.8594,
      "step": 1391
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 1.0623552799224854,
      "learning_rate": 8.876190476190477e-07,
      "loss": 0.8909,
      "step": 1392
    },
    {
      "epoch": 0.8577586206896551,
      "grad_norm": 0.9761639833450317,
      "learning_rate": 8.838095238095238e-07,
      "loss": 0.8158,
      "step": 1393
    },
    {
      "epoch": 0.8583743842364532,
      "grad_norm": 1.1358237266540527,
      "learning_rate": 8.8e-07,
      "loss": 0.8307,
      "step": 1394
    },
    {
      "epoch": 0.8589901477832512,
      "grad_norm": 0.99006187915802,
      "learning_rate": 8.761904761904763e-07,
      "loss": 0.815,
      "step": 1395
    },
    {
      "epoch": 0.8596059113300493,
      "grad_norm": 1.030927300453186,
      "learning_rate": 8.723809523809524e-07,
      "loss": 0.8494,
      "step": 1396
    },
    {
      "epoch": 0.8602216748768473,
      "grad_norm": 1.000778079032898,
      "learning_rate": 8.685714285714286e-07,
      "loss": 0.85,
      "step": 1397
    },
    {
      "epoch": 0.8608374384236454,
      "grad_norm": 1.002195954322815,
      "learning_rate": 8.647619047619048e-07,
      "loss": 0.8254,
      "step": 1398
    },
    {
      "epoch": 0.8614532019704434,
      "grad_norm": 0.9813644886016846,
      "learning_rate": 8.60952380952381e-07,
      "loss": 0.8026,
      "step": 1399
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 1.0646271705627441,
      "learning_rate": 8.571428571428571e-07,
      "loss": 0.8414,
      "step": 1400
    },
    {
      "epoch": 0.8626847290640394,
      "grad_norm": 0.9839184284210205,
      "learning_rate": 8.533333333333334e-07,
      "loss": 0.8316,
      "step": 1401
    },
    {
      "epoch": 0.8633004926108374,
      "grad_norm": 0.9630414843559265,
      "learning_rate": 8.495238095238096e-07,
      "loss": 0.78,
      "step": 1402
    },
    {
      "epoch": 0.8639162561576355,
      "grad_norm": 1.094186782836914,
      "learning_rate": 8.457142857142857e-07,
      "loss": 0.8276,
      "step": 1403
    },
    {
      "epoch": 0.8645320197044335,
      "grad_norm": 0.9484034776687622,
      "learning_rate": 8.419047619047619e-07,
      "loss": 0.8559,
      "step": 1404
    },
    {
      "epoch": 0.8651477832512315,
      "grad_norm": 1.0230942964553833,
      "learning_rate": 8.380952380952381e-07,
      "loss": 0.8296,
      "step": 1405
    },
    {
      "epoch": 0.8657635467980296,
      "grad_norm": 1.0030949115753174,
      "learning_rate": 8.342857142857143e-07,
      "loss": 0.7987,
      "step": 1406
    },
    {
      "epoch": 0.8663793103448276,
      "grad_norm": 0.9881190061569214,
      "learning_rate": 8.304761904761904e-07,
      "loss": 0.8203,
      "step": 1407
    },
    {
      "epoch": 0.8669950738916257,
      "grad_norm": 1.0378164052963257,
      "learning_rate": 8.266666666666667e-07,
      "loss": 0.8688,
      "step": 1408
    },
    {
      "epoch": 0.8676108374384236,
      "grad_norm": 1.009438157081604,
      "learning_rate": 8.228571428571429e-07,
      "loss": 0.8435,
      "step": 1409
    },
    {
      "epoch": 0.8682266009852216,
      "grad_norm": 1.0257142782211304,
      "learning_rate": 8.19047619047619e-07,
      "loss": 0.9059,
      "step": 1410
    },
    {
      "epoch": 0.8688423645320197,
      "grad_norm": 0.95546555519104,
      "learning_rate": 8.152380952380952e-07,
      "loss": 0.8502,
      "step": 1411
    },
    {
      "epoch": 0.8694581280788177,
      "grad_norm": 1.0171270370483398,
      "learning_rate": 8.114285714285714e-07,
      "loss": 0.8343,
      "step": 1412
    },
    {
      "epoch": 0.8700738916256158,
      "grad_norm": 0.9699042439460754,
      "learning_rate": 8.076190476190477e-07,
      "loss": 0.8187,
      "step": 1413
    },
    {
      "epoch": 0.8706896551724138,
      "grad_norm": 0.9271160364151001,
      "learning_rate": 8.038095238095237e-07,
      "loss": 0.8244,
      "step": 1414
    },
    {
      "epoch": 0.8713054187192119,
      "grad_norm": 1.0655497312545776,
      "learning_rate": 8e-07,
      "loss": 0.8771,
      "step": 1415
    },
    {
      "epoch": 0.8719211822660099,
      "grad_norm": 1.086912751197815,
      "learning_rate": 7.961904761904763e-07,
      "loss": 0.8833,
      "step": 1416
    },
    {
      "epoch": 0.8725369458128078,
      "grad_norm": 0.9758270382881165,
      "learning_rate": 7.923809523809525e-07,
      "loss": 0.8301,
      "step": 1417
    },
    {
      "epoch": 0.8731527093596059,
      "grad_norm": 1.0449674129486084,
      "learning_rate": 7.885714285714285e-07,
      "loss": 0.7992,
      "step": 1418
    },
    {
      "epoch": 0.8737684729064039,
      "grad_norm": 1.0238860845565796,
      "learning_rate": 7.847619047619048e-07,
      "loss": 0.8298,
      "step": 1419
    },
    {
      "epoch": 0.874384236453202,
      "grad_norm": 1.065205693244934,
      "learning_rate": 7.80952380952381e-07,
      "loss": 0.832,
      "step": 1420
    },
    {
      "epoch": 0.875,
      "grad_norm": 1.0967708826065063,
      "learning_rate": 7.771428571428571e-07,
      "loss": 0.8251,
      "step": 1421
    },
    {
      "epoch": 0.875615763546798,
      "grad_norm": 0.9871619343757629,
      "learning_rate": 7.733333333333334e-07,
      "loss": 0.8651,
      "step": 1422
    },
    {
      "epoch": 0.8762315270935961,
      "grad_norm": 1.0236549377441406,
      "learning_rate": 7.695238095238096e-07,
      "loss": 0.8325,
      "step": 1423
    },
    {
      "epoch": 0.8768472906403941,
      "grad_norm": 1.020857572555542,
      "learning_rate": 7.657142857142858e-07,
      "loss": 0.8503,
      "step": 1424
    },
    {
      "epoch": 0.8774630541871922,
      "grad_norm": 0.9735039472579956,
      "learning_rate": 7.619047619047619e-07,
      "loss": 0.8419,
      "step": 1425
    },
    {
      "epoch": 0.8780788177339901,
      "grad_norm": 1.0333142280578613,
      "learning_rate": 7.580952380952381e-07,
      "loss": 0.8143,
      "step": 1426
    },
    {
      "epoch": 0.8786945812807881,
      "grad_norm": 0.9771987199783325,
      "learning_rate": 7.542857142857144e-07,
      "loss": 0.8044,
      "step": 1427
    },
    {
      "epoch": 0.8793103448275862,
      "grad_norm": 1.0015151500701904,
      "learning_rate": 7.504761904761905e-07,
      "loss": 0.853,
      "step": 1428
    },
    {
      "epoch": 0.8799261083743842,
      "grad_norm": 1.0181846618652344,
      "learning_rate": 7.466666666666667e-07,
      "loss": 0.8489,
      "step": 1429
    },
    {
      "epoch": 0.8805418719211823,
      "grad_norm": 1.0579944849014282,
      "learning_rate": 7.428571428571429e-07,
      "loss": 0.8922,
      "step": 1430
    },
    {
      "epoch": 0.8811576354679803,
      "grad_norm": 1.0034675598144531,
      "learning_rate": 7.39047619047619e-07,
      "loss": 0.8085,
      "step": 1431
    },
    {
      "epoch": 0.8817733990147784,
      "grad_norm": 1.0354599952697754,
      "learning_rate": 7.352380952380953e-07,
      "loss": 0.826,
      "step": 1432
    },
    {
      "epoch": 0.8823891625615764,
      "grad_norm": 0.9339838624000549,
      "learning_rate": 7.314285714285714e-07,
      "loss": 0.8656,
      "step": 1433
    },
    {
      "epoch": 0.8830049261083743,
      "grad_norm": 1.0193158388137817,
      "learning_rate": 7.276190476190476e-07,
      "loss": 0.8239,
      "step": 1434
    },
    {
      "epoch": 0.8836206896551724,
      "grad_norm": 0.9804522395133972,
      "learning_rate": 7.238095238095239e-07,
      "loss": 0.8258,
      "step": 1435
    },
    {
      "epoch": 0.8842364532019704,
      "grad_norm": 1.0258877277374268,
      "learning_rate": 7.2e-07,
      "loss": 0.811,
      "step": 1436
    },
    {
      "epoch": 0.8848522167487685,
      "grad_norm": 0.9578343629837036,
      "learning_rate": 7.161904761904762e-07,
      "loss": 0.831,
      "step": 1437
    },
    {
      "epoch": 0.8854679802955665,
      "grad_norm": 1.0085080862045288,
      "learning_rate": 7.123809523809524e-07,
      "loss": 0.8161,
      "step": 1438
    },
    {
      "epoch": 0.8860837438423645,
      "grad_norm": 0.8980692028999329,
      "learning_rate": 7.085714285714286e-07,
      "loss": 0.8823,
      "step": 1439
    },
    {
      "epoch": 0.8866995073891626,
      "grad_norm": 1.0166064500808716,
      "learning_rate": 7.047619047619048e-07,
      "loss": 0.9004,
      "step": 1440
    },
    {
      "epoch": 0.8873152709359606,
      "grad_norm": 0.9729240536689758,
      "learning_rate": 7.00952380952381e-07,
      "loss": 0.8855,
      "step": 1441
    },
    {
      "epoch": 0.8879310344827587,
      "grad_norm": 1.004325270652771,
      "learning_rate": 6.971428571428572e-07,
      "loss": 0.859,
      "step": 1442
    },
    {
      "epoch": 0.8885467980295566,
      "grad_norm": 0.9958060383796692,
      "learning_rate": 6.933333333333333e-07,
      "loss": 0.8527,
      "step": 1443
    },
    {
      "epoch": 0.8891625615763546,
      "grad_norm": 0.9030126333236694,
      "learning_rate": 6.895238095238095e-07,
      "loss": 0.8148,
      "step": 1444
    },
    {
      "epoch": 0.8897783251231527,
      "grad_norm": 0.9447693228721619,
      "learning_rate": 6.857142857142857e-07,
      "loss": 0.8192,
      "step": 1445
    },
    {
      "epoch": 0.8903940886699507,
      "grad_norm": 0.9637632966041565,
      "learning_rate": 6.819047619047619e-07,
      "loss": 0.8329,
      "step": 1446
    },
    {
      "epoch": 0.8910098522167488,
      "grad_norm": 0.9550942182540894,
      "learning_rate": 6.780952380952381e-07,
      "loss": 0.8493,
      "step": 1447
    },
    {
      "epoch": 0.8916256157635468,
      "grad_norm": 0.9747555255889893,
      "learning_rate": 6.742857142857144e-07,
      "loss": 0.7917,
      "step": 1448
    },
    {
      "epoch": 0.8922413793103449,
      "grad_norm": 1.0164076089859009,
      "learning_rate": 6.704761904761905e-07,
      "loss": 0.8401,
      "step": 1449
    },
    {
      "epoch": 0.8928571428571429,
      "grad_norm": 1.1487065553665161,
      "learning_rate": 6.666666666666666e-07,
      "loss": 0.8748,
      "step": 1450
    },
    {
      "epoch": 0.8934729064039408,
      "grad_norm": 0.9546734094619751,
      "learning_rate": 6.628571428571429e-07,
      "loss": 0.8632,
      "step": 1451
    },
    {
      "epoch": 0.8940886699507389,
      "grad_norm": 0.9366591572761536,
      "learning_rate": 6.59047619047619e-07,
      "loss": 0.8291,
      "step": 1452
    },
    {
      "epoch": 0.8947044334975369,
      "grad_norm": 0.9759460687637329,
      "learning_rate": 6.552380952380953e-07,
      "loss": 0.8286,
      "step": 1453
    },
    {
      "epoch": 0.895320197044335,
      "grad_norm": 1.0152314901351929,
      "learning_rate": 6.514285714285715e-07,
      "loss": 0.8658,
      "step": 1454
    },
    {
      "epoch": 0.895935960591133,
      "grad_norm": 1.0651804208755493,
      "learning_rate": 6.476190476190477e-07,
      "loss": 0.8507,
      "step": 1455
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.9924448728561401,
      "learning_rate": 6.438095238095238e-07,
      "loss": 0.8746,
      "step": 1456
    },
    {
      "epoch": 0.8971674876847291,
      "grad_norm": 0.9351373314857483,
      "learning_rate": 6.4e-07,
      "loss": 0.7978,
      "step": 1457
    },
    {
      "epoch": 0.8977832512315271,
      "grad_norm": 0.9441096782684326,
      "learning_rate": 6.361904761904762e-07,
      "loss": 0.8624,
      "step": 1458
    },
    {
      "epoch": 0.8983990147783252,
      "grad_norm": 1.0685787200927734,
      "learning_rate": 6.323809523809523e-07,
      "loss": 0.7873,
      "step": 1459
    },
    {
      "epoch": 0.8990147783251231,
      "grad_norm": 0.9563184976577759,
      "learning_rate": 6.285714285714286e-07,
      "loss": 0.8288,
      "step": 1460
    },
    {
      "epoch": 0.8996305418719212,
      "grad_norm": 1.0187715291976929,
      "learning_rate": 6.247619047619048e-07,
      "loss": 0.8126,
      "step": 1461
    },
    {
      "epoch": 0.9002463054187192,
      "grad_norm": 0.9516103267669678,
      "learning_rate": 6.20952380952381e-07,
      "loss": 0.8549,
      "step": 1462
    },
    {
      "epoch": 0.9008620689655172,
      "grad_norm": 1.0149345397949219,
      "learning_rate": 6.171428571428571e-07,
      "loss": 0.8223,
      "step": 1463
    },
    {
      "epoch": 0.9014778325123153,
      "grad_norm": 0.9856411218643188,
      "learning_rate": 6.133333333333334e-07,
      "loss": 0.8346,
      "step": 1464
    },
    {
      "epoch": 0.9020935960591133,
      "grad_norm": 0.9852429628372192,
      "learning_rate": 6.095238095238095e-07,
      "loss": 0.8344,
      "step": 1465
    },
    {
      "epoch": 0.9027093596059114,
      "grad_norm": 1.0091029405593872,
      "learning_rate": 6.057142857142856e-07,
      "loss": 0.864,
      "step": 1466
    },
    {
      "epoch": 0.9033251231527094,
      "grad_norm": 0.9601311683654785,
      "learning_rate": 6.01904761904762e-07,
      "loss": 0.8079,
      "step": 1467
    },
    {
      "epoch": 0.9039408866995073,
      "grad_norm": 1.0159615278244019,
      "learning_rate": 5.980952380952381e-07,
      "loss": 0.8492,
      "step": 1468
    },
    {
      "epoch": 0.9045566502463054,
      "grad_norm": 1.1116135120391846,
      "learning_rate": 5.942857142857143e-07,
      "loss": 0.8328,
      "step": 1469
    },
    {
      "epoch": 0.9051724137931034,
      "grad_norm": 0.9594138264656067,
      "learning_rate": 5.904761904761905e-07,
      "loss": 0.8302,
      "step": 1470
    },
    {
      "epoch": 0.9057881773399015,
      "grad_norm": 0.9953395128250122,
      "learning_rate": 5.866666666666667e-07,
      "loss": 0.8347,
      "step": 1471
    },
    {
      "epoch": 0.9064039408866995,
      "grad_norm": 0.9971722364425659,
      "learning_rate": 5.828571428571428e-07,
      "loss": 0.8801,
      "step": 1472
    },
    {
      "epoch": 0.9070197044334976,
      "grad_norm": 1.0096867084503174,
      "learning_rate": 5.790476190476191e-07,
      "loss": 0.8331,
      "step": 1473
    },
    {
      "epoch": 0.9076354679802956,
      "grad_norm": 1.020545244216919,
      "learning_rate": 5.752380952380953e-07,
      "loss": 0.8096,
      "step": 1474
    },
    {
      "epoch": 0.9082512315270936,
      "grad_norm": 0.9185134768486023,
      "learning_rate": 5.714285714285714e-07,
      "loss": 0.817,
      "step": 1475
    },
    {
      "epoch": 0.9088669950738916,
      "grad_norm": 0.9891690015792847,
      "learning_rate": 5.676190476190476e-07,
      "loss": 0.8195,
      "step": 1476
    },
    {
      "epoch": 0.9094827586206896,
      "grad_norm": 1.0691343545913696,
      "learning_rate": 5.638095238095238e-07,
      "loss": 0.8511,
      "step": 1477
    },
    {
      "epoch": 0.9100985221674877,
      "grad_norm": 1.0056654214859009,
      "learning_rate": 5.6e-07,
      "loss": 0.8287,
      "step": 1478
    },
    {
      "epoch": 0.9107142857142857,
      "grad_norm": 1.0931408405303955,
      "learning_rate": 5.561904761904761e-07,
      "loss": 0.8862,
      "step": 1479
    },
    {
      "epoch": 0.9113300492610837,
      "grad_norm": 1.0248433351516724,
      "learning_rate": 5.523809523809525e-07,
      "loss": 0.8355,
      "step": 1480
    },
    {
      "epoch": 0.9119458128078818,
      "grad_norm": 0.9726777672767639,
      "learning_rate": 5.485714285714286e-07,
      "loss": 0.8374,
      "step": 1481
    },
    {
      "epoch": 0.9125615763546798,
      "grad_norm": 1.0030945539474487,
      "learning_rate": 5.447619047619047e-07,
      "loss": 0.8704,
      "step": 1482
    },
    {
      "epoch": 0.9131773399014779,
      "grad_norm": 1.0118154287338257,
      "learning_rate": 5.40952380952381e-07,
      "loss": 0.8532,
      "step": 1483
    },
    {
      "epoch": 0.9137931034482759,
      "grad_norm": 1.0708106756210327,
      "learning_rate": 5.371428571428571e-07,
      "loss": 0.8104,
      "step": 1484
    },
    {
      "epoch": 0.9144088669950738,
      "grad_norm": 1.0759670734405518,
      "learning_rate": 5.333333333333333e-07,
      "loss": 0.8446,
      "step": 1485
    },
    {
      "epoch": 0.9150246305418719,
      "grad_norm": 0.9399509429931641,
      "learning_rate": 5.295238095238096e-07,
      "loss": 0.8489,
      "step": 1486
    },
    {
      "epoch": 0.9156403940886699,
      "grad_norm": 0.9708300232887268,
      "learning_rate": 5.257142857142858e-07,
      "loss": 0.8288,
      "step": 1487
    },
    {
      "epoch": 0.916256157635468,
      "grad_norm": 0.9911633133888245,
      "learning_rate": 5.219047619047619e-07,
      "loss": 0.8656,
      "step": 1488
    },
    {
      "epoch": 0.916871921182266,
      "grad_norm": 1.0183771848678589,
      "learning_rate": 5.180952380952381e-07,
      "loss": 0.8578,
      "step": 1489
    },
    {
      "epoch": 0.9174876847290641,
      "grad_norm": 0.9071673154830933,
      "learning_rate": 5.142857142857143e-07,
      "loss": 0.8519,
      "step": 1490
    },
    {
      "epoch": 0.9181034482758621,
      "grad_norm": 1.0171922445297241,
      "learning_rate": 5.104761904761904e-07,
      "loss": 0.8763,
      "step": 1491
    },
    {
      "epoch": 0.9187192118226601,
      "grad_norm": 1.0001027584075928,
      "learning_rate": 5.066666666666667e-07,
      "loss": 0.8585,
      "step": 1492
    },
    {
      "epoch": 0.9193349753694581,
      "grad_norm": 0.9800043106079102,
      "learning_rate": 5.028571428571429e-07,
      "loss": 0.8311,
      "step": 1493
    },
    {
      "epoch": 0.9199507389162561,
      "grad_norm": 1.2396092414855957,
      "learning_rate": 4.990476190476191e-07,
      "loss": 0.838,
      "step": 1494
    },
    {
      "epoch": 0.9205665024630542,
      "grad_norm": 1.0795276165008545,
      "learning_rate": 4.952380952380952e-07,
      "loss": 0.8529,
      "step": 1495
    },
    {
      "epoch": 0.9211822660098522,
      "grad_norm": 1.0199109315872192,
      "learning_rate": 4.914285714285715e-07,
      "loss": 0.7906,
      "step": 1496
    },
    {
      "epoch": 0.9217980295566502,
      "grad_norm": 0.9677224159240723,
      "learning_rate": 4.876190476190476e-07,
      "loss": 0.8714,
      "step": 1497
    },
    {
      "epoch": 0.9224137931034483,
      "grad_norm": 1.028145670890808,
      "learning_rate": 4.838095238095237e-07,
      "loss": 0.7924,
      "step": 1498
    },
    {
      "epoch": 0.9230295566502463,
      "grad_norm": 1.0566893815994263,
      "learning_rate": 4.800000000000001e-07,
      "loss": 0.8291,
      "step": 1499
    },
    {
      "epoch": 0.9236453201970444,
      "grad_norm": 1.0697981119155884,
      "learning_rate": 4.761904761904762e-07,
      "loss": 0.8243,
      "step": 1500
    },
    {
      "epoch": 0.9242610837438424,
      "grad_norm": 0.9703925251960754,
      "learning_rate": 4.723809523809524e-07,
      "loss": 0.8171,
      "step": 1501
    },
    {
      "epoch": 0.9248768472906403,
      "grad_norm": 0.9718853235244751,
      "learning_rate": 4.6857142857142855e-07,
      "loss": 0.8183,
      "step": 1502
    },
    {
      "epoch": 0.9254926108374384,
      "grad_norm": 1.0200927257537842,
      "learning_rate": 4.647619047619048e-07,
      "loss": 0.8717,
      "step": 1503
    },
    {
      "epoch": 0.9261083743842364,
      "grad_norm": 1.028759479522705,
      "learning_rate": 4.6095238095238094e-07,
      "loss": 0.8435,
      "step": 1504
    },
    {
      "epoch": 0.9267241379310345,
      "grad_norm": 0.895150363445282,
      "learning_rate": 4.571428571428572e-07,
      "loss": 0.8386,
      "step": 1505
    },
    {
      "epoch": 0.9273399014778325,
      "grad_norm": 1.1226285696029663,
      "learning_rate": 4.5333333333333337e-07,
      "loss": 0.8512,
      "step": 1506
    },
    {
      "epoch": 0.9279556650246306,
      "grad_norm": 0.9795399308204651,
      "learning_rate": 4.495238095238095e-07,
      "loss": 0.8129,
      "step": 1507
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 1.0409235954284668,
      "learning_rate": 4.4571428571428575e-07,
      "loss": 0.8632,
      "step": 1508
    },
    {
      "epoch": 0.9291871921182266,
      "grad_norm": 0.9996502995491028,
      "learning_rate": 4.419047619047619e-07,
      "loss": 0.8426,
      "step": 1509
    },
    {
      "epoch": 0.9298029556650246,
      "grad_norm": 1.2089818716049194,
      "learning_rate": 4.3809523809523813e-07,
      "loss": 0.8688,
      "step": 1510
    },
    {
      "epoch": 0.9304187192118226,
      "grad_norm": 1.0056438446044922,
      "learning_rate": 4.342857142857143e-07,
      "loss": 0.8286,
      "step": 1511
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 1.002866506576538,
      "learning_rate": 4.304761904761905e-07,
      "loss": 0.8891,
      "step": 1512
    },
    {
      "epoch": 0.9316502463054187,
      "grad_norm": 0.9476414322853088,
      "learning_rate": 4.266666666666667e-07,
      "loss": 0.88,
      "step": 1513
    },
    {
      "epoch": 0.9322660098522167,
      "grad_norm": 0.9077138304710388,
      "learning_rate": 4.2285714285714284e-07,
      "loss": 0.883,
      "step": 1514
    },
    {
      "epoch": 0.9328817733990148,
      "grad_norm": 1.0710996389389038,
      "learning_rate": 4.1904761904761906e-07,
      "loss": 0.789,
      "step": 1515
    },
    {
      "epoch": 0.9334975369458128,
      "grad_norm": 1.056732177734375,
      "learning_rate": 4.152380952380952e-07,
      "loss": 0.8636,
      "step": 1516
    },
    {
      "epoch": 0.9341133004926109,
      "grad_norm": 1.0192607641220093,
      "learning_rate": 4.1142857142857144e-07,
      "loss": 0.8342,
      "step": 1517
    },
    {
      "epoch": 0.9347290640394089,
      "grad_norm": 0.9550378322601318,
      "learning_rate": 4.076190476190476e-07,
      "loss": 0.809,
      "step": 1518
    },
    {
      "epoch": 0.9353448275862069,
      "grad_norm": 0.9770001173019409,
      "learning_rate": 4.0380952380952387e-07,
      "loss": 0.8442,
      "step": 1519
    },
    {
      "epoch": 0.9359605911330049,
      "grad_norm": 0.9796767830848694,
      "learning_rate": 4e-07,
      "loss": 0.8035,
      "step": 1520
    },
    {
      "epoch": 0.9365763546798029,
      "grad_norm": 0.9813427925109863,
      "learning_rate": 3.9619047619047625e-07,
      "loss": 0.8475,
      "step": 1521
    },
    {
      "epoch": 0.937192118226601,
      "grad_norm": 0.9813137650489807,
      "learning_rate": 3.923809523809524e-07,
      "loss": 0.8357,
      "step": 1522
    },
    {
      "epoch": 0.937807881773399,
      "grad_norm": 0.9302571415901184,
      "learning_rate": 3.8857142857142853e-07,
      "loss": 0.8441,
      "step": 1523
    },
    {
      "epoch": 0.9384236453201971,
      "grad_norm": 0.9432770609855652,
      "learning_rate": 3.847619047619048e-07,
      "loss": 0.8307,
      "step": 1524
    },
    {
      "epoch": 0.9390394088669951,
      "grad_norm": 1.0462896823883057,
      "learning_rate": 3.8095238095238096e-07,
      "loss": 0.8003,
      "step": 1525
    },
    {
      "epoch": 0.9396551724137931,
      "grad_norm": 1.0406652688980103,
      "learning_rate": 3.771428571428572e-07,
      "loss": 0.8637,
      "step": 1526
    },
    {
      "epoch": 0.9402709359605911,
      "grad_norm": 0.9877525568008423,
      "learning_rate": 3.7333333333333334e-07,
      "loss": 0.818,
      "step": 1527
    },
    {
      "epoch": 0.9408866995073891,
      "grad_norm": 0.9676465392112732,
      "learning_rate": 3.695238095238095e-07,
      "loss": 0.846,
      "step": 1528
    },
    {
      "epoch": 0.9415024630541872,
      "grad_norm": 0.9289883375167847,
      "learning_rate": 3.657142857142857e-07,
      "loss": 0.8437,
      "step": 1529
    },
    {
      "epoch": 0.9421182266009852,
      "grad_norm": 0.9581652283668518,
      "learning_rate": 3.6190476190476194e-07,
      "loss": 0.8198,
      "step": 1530
    },
    {
      "epoch": 0.9427339901477833,
      "grad_norm": 0.9894691705703735,
      "learning_rate": 3.580952380952381e-07,
      "loss": 0.8258,
      "step": 1531
    },
    {
      "epoch": 0.9433497536945813,
      "grad_norm": 0.9627754092216492,
      "learning_rate": 3.542857142857143e-07,
      "loss": 0.8187,
      "step": 1532
    },
    {
      "epoch": 0.9439655172413793,
      "grad_norm": 0.9886759519577026,
      "learning_rate": 3.504761904761905e-07,
      "loss": 0.8158,
      "step": 1533
    },
    {
      "epoch": 0.9445812807881774,
      "grad_norm": 1.0085524320602417,
      "learning_rate": 3.4666666666666665e-07,
      "loss": 0.8101,
      "step": 1534
    },
    {
      "epoch": 0.9451970443349754,
      "grad_norm": 0.9755401611328125,
      "learning_rate": 3.4285714285714286e-07,
      "loss": 0.8315,
      "step": 1535
    },
    {
      "epoch": 0.9458128078817734,
      "grad_norm": 0.985993504524231,
      "learning_rate": 3.3904761904761903e-07,
      "loss": 0.8099,
      "step": 1536
    },
    {
      "epoch": 0.9464285714285714,
      "grad_norm": 0.9407062530517578,
      "learning_rate": 3.3523809523809525e-07,
      "loss": 0.8412,
      "step": 1537
    },
    {
      "epoch": 0.9470443349753694,
      "grad_norm": 0.951041579246521,
      "learning_rate": 3.3142857142857146e-07,
      "loss": 0.8704,
      "step": 1538
    },
    {
      "epoch": 0.9476600985221675,
      "grad_norm": 0.970804750919342,
      "learning_rate": 3.276190476190476e-07,
      "loss": 0.8329,
      "step": 1539
    },
    {
      "epoch": 0.9482758620689655,
      "grad_norm": 1.0332529544830322,
      "learning_rate": 3.2380952380952384e-07,
      "loss": 0.8722,
      "step": 1540
    },
    {
      "epoch": 0.9488916256157636,
      "grad_norm": 0.9267615079879761,
      "learning_rate": 3.2e-07,
      "loss": 0.8423,
      "step": 1541
    },
    {
      "epoch": 0.9495073891625616,
      "grad_norm": 1.0128666162490845,
      "learning_rate": 3.1619047619047617e-07,
      "loss": 0.8284,
      "step": 1542
    },
    {
      "epoch": 0.9501231527093597,
      "grad_norm": 0.911386251449585,
      "learning_rate": 3.123809523809524e-07,
      "loss": 0.8446,
      "step": 1543
    },
    {
      "epoch": 0.9507389162561576,
      "grad_norm": 0.978055477142334,
      "learning_rate": 3.0857142857142855e-07,
      "loss": 0.8062,
      "step": 1544
    },
    {
      "epoch": 0.9513546798029556,
      "grad_norm": 0.9671547412872314,
      "learning_rate": 3.0476190476190477e-07,
      "loss": 0.8723,
      "step": 1545
    },
    {
      "epoch": 0.9519704433497537,
      "grad_norm": 0.9915730953216553,
      "learning_rate": 3.00952380952381e-07,
      "loss": 0.8377,
      "step": 1546
    },
    {
      "epoch": 0.9525862068965517,
      "grad_norm": 0.9849810004234314,
      "learning_rate": 2.9714285714285715e-07,
      "loss": 0.7931,
      "step": 1547
    },
    {
      "epoch": 0.9532019704433498,
      "grad_norm": 1.0310399532318115,
      "learning_rate": 2.9333333333333337e-07,
      "loss": 0.8706,
      "step": 1548
    },
    {
      "epoch": 0.9538177339901478,
      "grad_norm": 0.997661828994751,
      "learning_rate": 2.8952380952380953e-07,
      "loss": 0.9039,
      "step": 1549
    },
    {
      "epoch": 0.9544334975369458,
      "grad_norm": 1.0511738061904907,
      "learning_rate": 2.857142857142857e-07,
      "loss": 0.8456,
      "step": 1550
    },
    {
      "epoch": 0.9550492610837439,
      "grad_norm": 0.9806724786758423,
      "learning_rate": 2.819047619047619e-07,
      "loss": 0.8396,
      "step": 1551
    },
    {
      "epoch": 0.9556650246305419,
      "grad_norm": 0.9931904077529907,
      "learning_rate": 2.780952380952381e-07,
      "loss": 0.8095,
      "step": 1552
    },
    {
      "epoch": 0.9562807881773399,
      "grad_norm": 1.011623740196228,
      "learning_rate": 2.742857142857143e-07,
      "loss": 0.8254,
      "step": 1553
    },
    {
      "epoch": 0.9568965517241379,
      "grad_norm": 1.0123281478881836,
      "learning_rate": 2.704761904761905e-07,
      "loss": 0.8654,
      "step": 1554
    },
    {
      "epoch": 0.9575123152709359,
      "grad_norm": 0.9169091582298279,
      "learning_rate": 2.6666666666666667e-07,
      "loss": 0.851,
      "step": 1555
    },
    {
      "epoch": 0.958128078817734,
      "grad_norm": 0.9753339886665344,
      "learning_rate": 2.628571428571429e-07,
      "loss": 0.8676,
      "step": 1556
    },
    {
      "epoch": 0.958743842364532,
      "grad_norm": 1.0122783184051514,
      "learning_rate": 2.5904761904761905e-07,
      "loss": 0.7832,
      "step": 1557
    },
    {
      "epoch": 0.9593596059113301,
      "grad_norm": 0.9614569544792175,
      "learning_rate": 2.552380952380952e-07,
      "loss": 0.7836,
      "step": 1558
    },
    {
      "epoch": 0.9599753694581281,
      "grad_norm": 1.0054634809494019,
      "learning_rate": 2.5142857142857143e-07,
      "loss": 0.891,
      "step": 1559
    },
    {
      "epoch": 0.9605911330049262,
      "grad_norm": 1.0321553945541382,
      "learning_rate": 2.476190476190476e-07,
      "loss": 0.8451,
      "step": 1560
    },
    {
      "epoch": 0.9612068965517241,
      "grad_norm": 0.936842679977417,
      "learning_rate": 2.438095238095238e-07,
      "loss": 0.8225,
      "step": 1561
    },
    {
      "epoch": 0.9618226600985221,
      "grad_norm": 0.9043599963188171,
      "learning_rate": 2.4000000000000003e-07,
      "loss": 0.8276,
      "step": 1562
    },
    {
      "epoch": 0.9624384236453202,
      "grad_norm": 0.9764125347137451,
      "learning_rate": 2.361904761904762e-07,
      "loss": 0.8175,
      "step": 1563
    },
    {
      "epoch": 0.9630541871921182,
      "grad_norm": 0.9205788969993591,
      "learning_rate": 2.323809523809524e-07,
      "loss": 0.8566,
      "step": 1564
    },
    {
      "epoch": 0.9636699507389163,
      "grad_norm": 1.0700900554656982,
      "learning_rate": 2.285714285714286e-07,
      "loss": 0.8357,
      "step": 1565
    },
    {
      "epoch": 0.9642857142857143,
      "grad_norm": 1.0012506246566772,
      "learning_rate": 2.2476190476190474e-07,
      "loss": 0.7931,
      "step": 1566
    },
    {
      "epoch": 0.9649014778325123,
      "grad_norm": 1.0079331398010254,
      "learning_rate": 2.2095238095238096e-07,
      "loss": 0.8744,
      "step": 1567
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.979428768157959,
      "learning_rate": 2.1714285714285715e-07,
      "loss": 0.8058,
      "step": 1568
    },
    {
      "epoch": 0.9661330049261084,
      "grad_norm": 1.1462762355804443,
      "learning_rate": 2.1333333333333334e-07,
      "loss": 0.83,
      "step": 1569
    },
    {
      "epoch": 0.9667487684729064,
      "grad_norm": 0.9788756370544434,
      "learning_rate": 2.0952380952380953e-07,
      "loss": 0.8715,
      "step": 1570
    },
    {
      "epoch": 0.9673645320197044,
      "grad_norm": 1.0784056186676025,
      "learning_rate": 2.0571428571428572e-07,
      "loss": 0.8736,
      "step": 1571
    },
    {
      "epoch": 0.9679802955665024,
      "grad_norm": 1.0514867305755615,
      "learning_rate": 2.0190476190476194e-07,
      "loss": 0.7661,
      "step": 1572
    },
    {
      "epoch": 0.9685960591133005,
      "grad_norm": 1.0291898250579834,
      "learning_rate": 1.9809523809523813e-07,
      "loss": 0.906,
      "step": 1573
    },
    {
      "epoch": 0.9692118226600985,
      "grad_norm": 0.9371717572212219,
      "learning_rate": 1.9428571428571426e-07,
      "loss": 0.8541,
      "step": 1574
    },
    {
      "epoch": 0.9698275862068966,
      "grad_norm": 1.0080207586288452,
      "learning_rate": 1.9047619047619048e-07,
      "loss": 0.8217,
      "step": 1575
    },
    {
      "epoch": 0.9704433497536946,
      "grad_norm": 0.9528869390487671,
      "learning_rate": 1.8666666666666667e-07,
      "loss": 0.8201,
      "step": 1576
    },
    {
      "epoch": 0.9710591133004927,
      "grad_norm": 1.0154980421066284,
      "learning_rate": 1.8285714285714286e-07,
      "loss": 0.8373,
      "step": 1577
    },
    {
      "epoch": 0.9716748768472906,
      "grad_norm": 0.9053487777709961,
      "learning_rate": 1.7904761904761905e-07,
      "loss": 0.8051,
      "step": 1578
    },
    {
      "epoch": 0.9722906403940886,
      "grad_norm": 1.1615312099456787,
      "learning_rate": 1.7523809523809524e-07,
      "loss": 0.8168,
      "step": 1579
    },
    {
      "epoch": 0.9729064039408867,
      "grad_norm": 0.9866144061088562,
      "learning_rate": 1.7142857142857143e-07,
      "loss": 0.8589,
      "step": 1580
    },
    {
      "epoch": 0.9735221674876847,
      "grad_norm": 0.922502338886261,
      "learning_rate": 1.6761904761904762e-07,
      "loss": 0.8454,
      "step": 1581
    },
    {
      "epoch": 0.9741379310344828,
      "grad_norm": 1.0707663297653198,
      "learning_rate": 1.638095238095238e-07,
      "loss": 0.7837,
      "step": 1582
    },
    {
      "epoch": 0.9747536945812808,
      "grad_norm": 1.005251169204712,
      "learning_rate": 1.6e-07,
      "loss": 0.8159,
      "step": 1583
    },
    {
      "epoch": 0.9753694581280788,
      "grad_norm": 1.0226916074752808,
      "learning_rate": 1.561904761904762e-07,
      "loss": 0.8417,
      "step": 1584
    },
    {
      "epoch": 0.9759852216748769,
      "grad_norm": 0.9404586553573608,
      "learning_rate": 1.5238095238095238e-07,
      "loss": 0.7962,
      "step": 1585
    },
    {
      "epoch": 0.9766009852216748,
      "grad_norm": 0.9079810976982117,
      "learning_rate": 1.4857142857142857e-07,
      "loss": 0.8251,
      "step": 1586
    },
    {
      "epoch": 0.9772167487684729,
      "grad_norm": 0.938382625579834,
      "learning_rate": 1.4476190476190476e-07,
      "loss": 0.8841,
      "step": 1587
    },
    {
      "epoch": 0.9778325123152709,
      "grad_norm": 1.066123604774475,
      "learning_rate": 1.4095238095238096e-07,
      "loss": 0.865,
      "step": 1588
    },
    {
      "epoch": 0.978448275862069,
      "grad_norm": 1.038551688194275,
      "learning_rate": 1.3714285714285715e-07,
      "loss": 0.7971,
      "step": 1589
    },
    {
      "epoch": 0.979064039408867,
      "grad_norm": 1.1643422842025757,
      "learning_rate": 1.3333333333333334e-07,
      "loss": 0.8547,
      "step": 1590
    },
    {
      "epoch": 0.979679802955665,
      "grad_norm": 1.015071988105774,
      "learning_rate": 1.2952380952380953e-07,
      "loss": 0.8146,
      "step": 1591
    },
    {
      "epoch": 0.9802955665024631,
      "grad_norm": 1.1664942502975464,
      "learning_rate": 1.2571428571428572e-07,
      "loss": 0.7958,
      "step": 1592
    },
    {
      "epoch": 0.9809113300492611,
      "grad_norm": 0.9968972206115723,
      "learning_rate": 1.219047619047619e-07,
      "loss": 0.8244,
      "step": 1593
    },
    {
      "epoch": 0.9815270935960592,
      "grad_norm": 0.9502185583114624,
      "learning_rate": 1.180952380952381e-07,
      "loss": 0.8157,
      "step": 1594
    },
    {
      "epoch": 0.9821428571428571,
      "grad_norm": 1.0035356283187866,
      "learning_rate": 1.142857142857143e-07,
      "loss": 0.8314,
      "step": 1595
    },
    {
      "epoch": 0.9827586206896551,
      "grad_norm": 1.0809407234191895,
      "learning_rate": 1.1047619047619048e-07,
      "loss": 0.8562,
      "step": 1596
    },
    {
      "epoch": 0.9833743842364532,
      "grad_norm": 0.9987855553627014,
      "learning_rate": 1.0666666666666667e-07,
      "loss": 0.8545,
      "step": 1597
    },
    {
      "epoch": 0.9839901477832512,
      "grad_norm": 0.9769032597541809,
      "learning_rate": 1.0285714285714286e-07,
      "loss": 0.8103,
      "step": 1598
    },
    {
      "epoch": 0.9846059113300493,
      "grad_norm": 1.0050386190414429,
      "learning_rate": 9.904761904761906e-08,
      "loss": 0.8313,
      "step": 1599
    },
    {
      "epoch": 0.9852216748768473,
      "grad_norm": 1.0013020038604736,
      "learning_rate": 9.523809523809524e-08,
      "loss": 0.8774,
      "step": 1600
    },
    {
      "epoch": 0.9858374384236454,
      "grad_norm": 1.0689412355422974,
      "learning_rate": 9.142857142857143e-08,
      "loss": 0.8191,
      "step": 1601
    },
    {
      "epoch": 0.9864532019704434,
      "grad_norm": 0.9286015033721924,
      "learning_rate": 8.761904761904762e-08,
      "loss": 0.8168,
      "step": 1602
    },
    {
      "epoch": 0.9870689655172413,
      "grad_norm": 0.9589946866035461,
      "learning_rate": 8.380952380952381e-08,
      "loss": 0.8595,
      "step": 1603
    },
    {
      "epoch": 0.9876847290640394,
      "grad_norm": 0.9680138826370239,
      "learning_rate": 8e-08,
      "loss": 0.8143,
      "step": 1604
    },
    {
      "epoch": 0.9883004926108374,
      "grad_norm": 1.0145773887634277,
      "learning_rate": 7.619047619047619e-08,
      "loss": 0.8354,
      "step": 1605
    },
    {
      "epoch": 0.9889162561576355,
      "grad_norm": 0.9854435324668884,
      "learning_rate": 7.238095238095238e-08,
      "loss": 0.8397,
      "step": 1606
    },
    {
      "epoch": 0.9895320197044335,
      "grad_norm": 1.0392318964004517,
      "learning_rate": 6.857142857142857e-08,
      "loss": 0.8397,
      "step": 1607
    },
    {
      "epoch": 0.9901477832512315,
      "grad_norm": 1.0826729536056519,
      "learning_rate": 6.476190476190476e-08,
      "loss": 0.827,
      "step": 1608
    },
    {
      "epoch": 0.9907635467980296,
      "grad_norm": 1.0190343856811523,
      "learning_rate": 6.095238095238095e-08,
      "loss": 0.8772,
      "step": 1609
    },
    {
      "epoch": 0.9913793103448276,
      "grad_norm": 1.0687893629074097,
      "learning_rate": 5.714285714285715e-08,
      "loss": 0.8248,
      "step": 1610
    },
    {
      "epoch": 0.9919950738916257,
      "grad_norm": 1.037475347518921,
      "learning_rate": 5.3333333333333334e-08,
      "loss": 0.8227,
      "step": 1611
    },
    {
      "epoch": 0.9926108374384236,
      "grad_norm": 0.9208523631095886,
      "learning_rate": 4.952380952380953e-08,
      "loss": 0.8498,
      "step": 1612
    },
    {
      "epoch": 0.9932266009852216,
      "grad_norm": 0.9159571528434753,
      "learning_rate": 4.5714285714285715e-08,
      "loss": 0.8176,
      "step": 1613
    },
    {
      "epoch": 0.9938423645320197,
      "grad_norm": 1.0408233404159546,
      "learning_rate": 4.1904761904761906e-08,
      "loss": 0.8781,
      "step": 1614
    },
    {
      "epoch": 0.9944581280788177,
      "grad_norm": 0.932273268699646,
      "learning_rate": 3.8095238095238096e-08,
      "loss": 0.8159,
      "step": 1615
    },
    {
      "epoch": 0.9950738916256158,
      "grad_norm": 0.9656065106391907,
      "learning_rate": 3.4285714285714286e-08,
      "loss": 0.7901,
      "step": 1616
    },
    {
      "epoch": 0.9956896551724138,
      "grad_norm": 1.0278900861740112,
      "learning_rate": 3.047619047619048e-08,
      "loss": 0.8495,
      "step": 1617
    },
    {
      "epoch": 0.9963054187192119,
      "grad_norm": 1.027269959449768,
      "learning_rate": 2.6666666666666667e-08,
      "loss": 0.8685,
      "step": 1618
    },
    {
      "epoch": 0.9969211822660099,
      "grad_norm": 0.9474722146987915,
      "learning_rate": 2.2857142857142858e-08,
      "loss": 0.8523,
      "step": 1619
    },
    {
      "epoch": 0.9975369458128078,
      "grad_norm": 0.9887866973876953,
      "learning_rate": 1.9047619047619048e-08,
      "loss": 0.8299,
      "step": 1620
    },
    {
      "epoch": 0.9981527093596059,
      "grad_norm": 1.0065726041793823,
      "learning_rate": 1.523809523809524e-08,
      "loss": 0.822,
      "step": 1621
    },
    {
      "epoch": 0.9987684729064039,
      "grad_norm": 1.1150017976760864,
      "learning_rate": 1.1428571428571429e-08,
      "loss": 0.7918,
      "step": 1622
    },
    {
      "epoch": 0.999384236453202,
      "grad_norm": 0.999965250492096,
      "learning_rate": 7.61904761904762e-09,
      "loss": 0.8422,
      "step": 1623
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9817562699317932,
      "learning_rate": 3.80952380952381e-09,
      "loss": 0.8181,
      "step": 1624
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.8671314716339111,
      "eval_runtime": 267.4037,
      "eval_samples_per_second": 86.424,
      "eval_steps_per_second": 2.704,
      "step": 1624
    }
  ],
  "logging_steps": 1,
  "max_steps": 1624,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 82,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.1074127547867333e+19,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}