{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 11142,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 8.975049362771496e-05,
      "grad_norm": 26.152917861938477,
      "learning_rate": 0.0,
      "loss": 1.6695,
      "step": 1
    },
    {
      "epoch": 0.0001795009872554299,
      "grad_norm": 33.55382537841797,
      "learning_rate": 0.0,
      "loss": 1.6218,
      "step": 2
    },
    {
      "epoch": 0.00026925148088314486,
      "grad_norm": 32.87733840942383,
      "learning_rate": 0.0,
      "loss": 1.579,
      "step": 3
    },
    {
      "epoch": 0.0003590019745108598,
      "grad_norm": 26.170360565185547,
      "learning_rate": 0.0,
      "loss": 1.6536,
      "step": 4
    },
    {
      "epoch": 0.00044875246813857474,
      "grad_norm": 24.22742462158203,
      "learning_rate": 0.0,
      "loss": 1.6274,
      "step": 5
    },
    {
      "epoch": 0.0005385029617662897,
      "grad_norm": 28.0578670501709,
      "learning_rate": 0.0,
      "loss": 1.6091,
      "step": 6
    },
    {
      "epoch": 0.0006282534553940047,
      "grad_norm": 23.102970123291016,
      "learning_rate": 0.0,
      "loss": 1.5881,
      "step": 7
    },
    {
      "epoch": 0.0007180039490217197,
      "grad_norm": 26.690277099609375,
      "learning_rate": 0.0,
      "loss": 1.6145,
      "step": 8
    },
    {
      "epoch": 0.0008077544426494346,
      "grad_norm": 26.564645767211914,
      "learning_rate": 0.0,
      "loss": 1.5641,
      "step": 9
    },
    {
      "epoch": 0.0008975049362771495,
      "grad_norm": 24.34095001220703,
      "learning_rate": 0.0,
      "loss": 1.6134,
      "step": 10
    },
    {
      "epoch": 0.0009872554299048646,
      "grad_norm": 31.507919311523438,
      "learning_rate": 0.0,
      "loss": 1.5898,
      "step": 11
    },
    {
      "epoch": 0.0010770059235325794,
      "grad_norm": 23.533363342285156,
      "learning_rate": 0.0,
      "loss": 1.5584,
      "step": 12
    },
    {
      "epoch": 0.0011667564171602943,
      "grad_norm": 27.943809509277344,
      "learning_rate": 0.0,
      "loss": 1.617,
      "step": 13
    },
    {
      "epoch": 0.0012565069107880094,
      "grad_norm": 27.70842170715332,
      "learning_rate": 0.0,
      "loss": 1.6454,
      "step": 14
    },
    {
      "epoch": 0.0013462574044157242,
      "grad_norm": 21.269529342651367,
      "learning_rate": 0.0,
      "loss": 1.5171,
      "step": 15
    },
    {
      "epoch": 0.0014360078980434393,
      "grad_norm": 24.082069396972656,
      "learning_rate": 0.0,
      "loss": 1.4827,
      "step": 16
    },
    {
      "epoch": 0.0015257583916711542,
      "grad_norm": 27.07015609741211,
      "learning_rate": 0.0,
      "loss": 1.6411,
      "step": 17
    },
    {
      "epoch": 0.0016155088852988692,
      "grad_norm": 23.609163284301758,
      "learning_rate": 0.0,
      "loss": 1.6642,
      "step": 18
    },
    {
      "epoch": 0.001705259378926584,
      "grad_norm": 26.63987159729004,
      "learning_rate": 0.0,
      "loss": 1.5756,
      "step": 19
    },
    {
      "epoch": 0.001795009872554299,
      "grad_norm": 25.38481330871582,
      "learning_rate": 0.0,
      "loss": 1.6368,
      "step": 20
    },
    {
      "epoch": 0.001884760366182014,
      "grad_norm": 28.967275619506836,
      "learning_rate": 0.0,
      "loss": 1.6673,
      "step": 21
    },
    {
      "epoch": 0.001974510859809729,
      "grad_norm": 33.9649658203125,
      "learning_rate": 0.0,
      "loss": 1.6652,
      "step": 22
    },
    {
      "epoch": 0.0020642613534374438,
      "grad_norm": 27.510683059692383,
      "learning_rate": 0.0,
      "loss": 1.6686,
      "step": 23
    },
    {
      "epoch": 0.002154011847065159,
      "grad_norm": 23.20203399658203,
      "learning_rate": 0.0,
      "loss": 1.5556,
      "step": 24
    },
    {
      "epoch": 0.002243762340692874,
      "grad_norm": 26.175939559936523,
      "learning_rate": 0.0,
      "loss": 1.6128,
      "step": 25
    },
    {
      "epoch": 0.0023335128343205886,
      "grad_norm": 73.48287963867188,
      "learning_rate": 0.0,
      "loss": 1.7186,
      "step": 26
    },
    {
      "epoch": 0.0024232633279483036,
      "grad_norm": 25.64657211303711,
      "learning_rate": 0.0,
      "loss": 1.5596,
      "step": 27
    },
    {
      "epoch": 0.0025130138215760187,
      "grad_norm": 52.773250579833984,
      "learning_rate": 0.0,
      "loss": 1.7305,
      "step": 28
    },
    {
      "epoch": 0.002602764315203734,
      "grad_norm": 26.533517837524414,
      "learning_rate": 0.0,
      "loss": 1.6064,
      "step": 29
    },
    {
      "epoch": 0.0026925148088314485,
      "grad_norm": 29.962785720825195,
      "learning_rate": 0.0,
      "loss": 1.7194,
      "step": 30
    },
    {
      "epoch": 0.0027822653024591635,
      "grad_norm": 22.8751277923584,
      "learning_rate": 0.0,
      "loss": 1.5366,
      "step": 31
    },
    {
      "epoch": 0.0028720157960868786,
      "grad_norm": 26.486433029174805,
      "learning_rate": 0.0,
      "loss": 1.594,
      "step": 32
    },
    {
      "epoch": 0.0029617662897145933,
      "grad_norm": 25.95654296875,
      "learning_rate": 0.0,
      "loss": 1.5879,
      "step": 33
    },
    {
      "epoch": 0.0030515167833423083,
      "grad_norm": 22.153379440307617,
      "learning_rate": 0.0,
      "loss": 1.5508,
      "step": 34
    },
    {
      "epoch": 0.0031412672769700234,
      "grad_norm": 27.099489212036133,
      "learning_rate": 0.0,
      "loss": 1.6024,
      "step": 35
    },
    {
      "epoch": 0.0032310177705977385,
      "grad_norm": 27.113630294799805,
      "learning_rate": 0.0,
      "loss": 1.5974,
      "step": 36
    },
    {
      "epoch": 0.003320768264225453,
      "grad_norm": 25.12582778930664,
      "learning_rate": 0.0,
      "loss": 1.6613,
      "step": 37
    },
    {
      "epoch": 0.003410518757853168,
      "grad_norm": 27.99002456665039,
      "learning_rate": 0.0,
      "loss": 1.5746,
      "step": 38
    },
    {
      "epoch": 0.0035002692514808833,
      "grad_norm": 27.57147216796875,
      "learning_rate": 0.0,
      "loss": 1.5153,
      "step": 39
    },
    {
      "epoch": 0.003590019745108598,
      "grad_norm": 92.59241485595703,
      "learning_rate": 0.0,
      "loss": 1.6292,
      "step": 40
    },
    {
      "epoch": 0.003679770238736313,
      "grad_norm": 26.960996627807617,
      "learning_rate": 0.0,
      "loss": 1.6147,
      "step": 41
    },
    {
      "epoch": 0.003769520732364028,
      "grad_norm": 24.15026092529297,
      "learning_rate": 0.0,
      "loss": 1.5116,
      "step": 42
    },
    {
      "epoch": 0.003859271225991743,
      "grad_norm": 24.157381057739258,
      "learning_rate": 0.0,
      "loss": 1.5702,
      "step": 43
    },
    {
      "epoch": 0.003949021719619458,
      "grad_norm": 25.433080673217773,
      "learning_rate": 0.0,
      "loss": 1.6111,
      "step": 44
    },
    {
      "epoch": 0.004038772213247173,
      "grad_norm": 23.479412078857422,
      "learning_rate": 0.0,
      "loss": 1.6307,
      "step": 45
    },
    {
      "epoch": 0.0041285227068748875,
      "grad_norm": 26.914522171020508,
      "learning_rate": 0.0,
      "loss": 1.6109,
      "step": 46
    },
    {
      "epoch": 0.004218273200502603,
      "grad_norm": 26.717313766479492,
      "learning_rate": 0.0,
      "loss": 1.6538,
      "step": 47
    },
    {
      "epoch": 0.004308023694130318,
      "grad_norm": 26.396909713745117,
      "learning_rate": 0.0,
      "loss": 1.598,
      "step": 48
    },
    {
      "epoch": 0.004397774187758032,
      "grad_norm": 30.399221420288086,
      "learning_rate": 0.0,
      "loss": 1.5709,
      "step": 49
    },
    {
      "epoch": 0.004487524681385748,
      "grad_norm": 26.00916862487793,
      "learning_rate": 0.0,
      "loss": 1.6653,
      "step": 50
    },
    {
      "epoch": 0.0045772751750134625,
      "grad_norm": 23.317018508911133,
      "learning_rate": 0.0,
      "loss": 1.6389,
      "step": 51
    },
    {
      "epoch": 0.004667025668641177,
      "grad_norm": 24.416353225708008,
      "learning_rate": 0.0,
      "loss": 1.5295,
      "step": 52
    },
    {
      "epoch": 0.004756776162268893,
      "grad_norm": 24.018569946289062,
      "learning_rate": 0.0,
      "loss": 1.544,
      "step": 53
    },
    {
      "epoch": 0.004846526655896607,
      "grad_norm": 23.266204833984375,
      "learning_rate": 0.0,
      "loss": 1.5967,
      "step": 54
    },
    {
      "epoch": 0.004936277149524323,
      "grad_norm": 30.091249465942383,
      "learning_rate": 0.0,
      "loss": 1.5841,
      "step": 55
    },
    {
      "epoch": 0.0050260276431520375,
      "grad_norm": 24.314411163330078,
      "learning_rate": 0.0,
      "loss": 1.6143,
      "step": 56
    },
    {
      "epoch": 0.005115778136779752,
      "grad_norm": 27.68694305419922,
      "learning_rate": 0.0,
      "loss": 1.7155,
      "step": 57
    },
    {
      "epoch": 0.005205528630407468,
      "grad_norm": 28.809463500976562,
      "learning_rate": 0.0,
      "loss": 1.5775,
      "step": 58
    },
    {
      "epoch": 0.005295279124035182,
      "grad_norm": 38.93540954589844,
      "learning_rate": 0.0,
      "loss": 1.5131,
      "step": 59
    },
    {
      "epoch": 0.005385029617662897,
      "grad_norm": 24.41024398803711,
      "learning_rate": 0.0,
      "loss": 1.5764,
      "step": 60
    },
    {
      "epoch": 0.005474780111290612,
      "grad_norm": 27.596847534179688,
      "learning_rate": 0.0,
      "loss": 1.7181,
      "step": 61
    },
    {
      "epoch": 0.005564530604918327,
      "grad_norm": 22.657678604125977,
      "learning_rate": 0.0,
      "loss": 1.5277,
      "step": 62
    },
    {
      "epoch": 0.005654281098546042,
      "grad_norm": 26.26171875,
      "learning_rate": 0.0,
      "loss": 1.6109,
      "step": 63
    },
    {
      "epoch": 0.005744031592173757,
      "grad_norm": 22.27169418334961,
      "learning_rate": 0.0,
      "loss": 1.5355,
      "step": 64
    },
    {
      "epoch": 0.005833782085801472,
      "grad_norm": 26.672792434692383,
      "learning_rate": 0.0,
      "loss": 1.5215,
      "step": 65
    },
    {
      "epoch": 0.0059235325794291865,
      "grad_norm": 25.087617874145508,
      "learning_rate": 0.0,
      "loss": 1.5149,
      "step": 66
    },
    {
      "epoch": 0.006013283073056902,
      "grad_norm": 25.29945945739746,
      "learning_rate": 0.0,
      "loss": 1.6263,
      "step": 67
    },
    {
      "epoch": 0.006103033566684617,
      "grad_norm": 33.656768798828125,
      "learning_rate": 0.0,
      "loss": 1.6631,
      "step": 68
    },
    {
      "epoch": 0.006192784060312331,
      "grad_norm": 38.10761642456055,
      "learning_rate": 0.0,
      "loss": 1.5804,
      "step": 69
    },
    {
      "epoch": 0.006282534553940047,
      "grad_norm": 20.548892974853516,
      "learning_rate": 0.0,
      "loss": 1.5841,
      "step": 70
    },
    {
      "epoch": 0.0063722850475677615,
      "grad_norm": 26.25657081604004,
      "learning_rate": 0.0,
      "loss": 1.6744,
      "step": 71
    },
    {
      "epoch": 0.006462035541195477,
      "grad_norm": 23.553552627563477,
      "learning_rate": 0.0,
      "loss": 1.6444,
      "step": 72
    },
    {
      "epoch": 0.006551786034823192,
      "grad_norm": 23.829336166381836,
      "learning_rate": 0.0,
      "loss": 1.5952,
      "step": 73
    },
    {
      "epoch": 0.006641536528450906,
      "grad_norm": 24.45437240600586,
      "learning_rate": 0.0,
      "loss": 1.5847,
      "step": 74
    },
    {
      "epoch": 0.006731287022078622,
      "grad_norm": 19.292957305908203,
      "learning_rate": 0.0,
      "loss": 1.6479,
      "step": 75
    },
    {
      "epoch": 0.006821037515706336,
      "grad_norm": 19.125526428222656,
      "learning_rate": 0.0,
      "loss": 1.5795,
      "step": 76
    },
    {
      "epoch": 0.006910788009334051,
      "grad_norm": 25.871789932250977,
      "learning_rate": 0.0,
      "loss": 1.5979,
      "step": 77
    },
    {
      "epoch": 0.007000538502961767,
      "grad_norm": 39.549434661865234,
      "learning_rate": 0.0,
      "loss": 1.5767,
      "step": 78
    },
    {
      "epoch": 0.007090288996589481,
      "grad_norm": 21.010318756103516,
      "learning_rate": 0.0,
      "loss": 1.6035,
      "step": 79
    },
    {
      "epoch": 0.007180039490217196,
      "grad_norm": 25.413124084472656,
      "learning_rate": 0.0,
      "loss": 1.6114,
      "step": 80
    },
    {
      "epoch": 0.007269789983844911,
      "grad_norm": 29.6253719329834,
      "learning_rate": 0.0,
      "loss": 1.7096,
      "step": 81
    },
    {
      "epoch": 0.007359540477472626,
      "grad_norm": 23.309053421020508,
      "learning_rate": 0.0,
      "loss": 1.642,
      "step": 82
    },
    {
      "epoch": 0.007449290971100341,
      "grad_norm": 23.22365951538086,
      "learning_rate": 0.0,
      "loss": 1.6737,
      "step": 83
    },
    {
      "epoch": 0.007539041464728056,
      "grad_norm": 21.74359893798828,
      "learning_rate": 0.0,
      "loss": 1.6555,
      "step": 84
    },
    {
      "epoch": 0.007628791958355771,
      "grad_norm": 28.64274024963379,
      "learning_rate": 0.0,
      "loss": 1.6442,
      "step": 85
    },
    {
      "epoch": 0.007718542451983486,
      "grad_norm": 32.727115631103516,
      "learning_rate": 0.0,
      "loss": 1.5269,
      "step": 86
    },
    {
      "epoch": 0.007808292945611201,
      "grad_norm": 22.13539695739746,
      "learning_rate": 0.0,
      "loss": 1.5711,
      "step": 87
    },
    {
      "epoch": 0.007898043439238916,
      "grad_norm": 28.659156799316406,
      "learning_rate": 0.0,
      "loss": 1.5603,
      "step": 88
    },
    {
      "epoch": 0.007987793932866631,
      "grad_norm": 29.97081756591797,
      "learning_rate": 0.0,
      "loss": 1.6082,
      "step": 89
    },
    {
      "epoch": 0.008077544426494346,
      "grad_norm": 25.177602767944336,
      "learning_rate": 0.0,
      "loss": 1.568,
      "step": 90
    },
    {
      "epoch": 0.00816729492012206,
      "grad_norm": 24.704565048217773,
      "learning_rate": 0.0,
      "loss": 1.6136,
      "step": 91
    },
    {
      "epoch": 0.008257045413749775,
      "grad_norm": 23.20800018310547,
      "learning_rate": 0.0,
      "loss": 1.6748,
      "step": 92
    },
    {
      "epoch": 0.00834679590737749,
      "grad_norm": 25.28188133239746,
      "learning_rate": 0.0,
      "loss": 1.5711,
      "step": 93
    },
    {
      "epoch": 0.008436546401005206,
      "grad_norm": 26.439477920532227,
      "learning_rate": 0.0,
      "loss": 1.566,
      "step": 94
    },
    {
      "epoch": 0.00852629689463292,
      "grad_norm": 20.247591018676758,
      "learning_rate": 0.0,
      "loss": 1.619,
      "step": 95
    },
    {
      "epoch": 0.008616047388260635,
      "grad_norm": 28.66176414489746,
      "learning_rate": 0.0,
      "loss": 1.5528,
      "step": 96
    },
    {
      "epoch": 0.00870579788188835,
      "grad_norm": 46.857242584228516,
      "learning_rate": 0.0,
      "loss": 1.7713,
      "step": 97
    },
    {
      "epoch": 0.008795548375516065,
      "grad_norm": 23.869800567626953,
      "learning_rate": 0.0,
      "loss": 1.5416,
      "step": 98
    },
    {
      "epoch": 0.008885298869143781,
      "grad_norm": 25.9635009765625,
      "learning_rate": 0.0,
      "loss": 1.6153,
      "step": 99
    },
    {
      "epoch": 0.008975049362771496,
      "grad_norm": 26.848220825195312,
      "learning_rate": 0.0,
      "loss": 1.6189,
      "step": 100
    },
    {
      "epoch": 0.00906479985639921,
      "grad_norm": 25.54722785949707,
      "learning_rate": 0.0,
      "loss": 1.5982,
      "step": 101
    },
    {
      "epoch": 0.009154550350026925,
      "grad_norm": 29.275510787963867,
      "learning_rate": 0.0,
      "loss": 1.5512,
      "step": 102
    },
    {
      "epoch": 0.00924430084365464,
      "grad_norm": 25.808996200561523,
      "learning_rate": 0.0,
      "loss": 1.6208,
      "step": 103
    },
    {
      "epoch": 0.009334051337282354,
      "grad_norm": 51.99929428100586,
      "learning_rate": 0.0,
      "loss": 1.5724,
      "step": 104
    },
    {
      "epoch": 0.00942380183091007,
      "grad_norm": 26.994720458984375,
      "learning_rate": 0.0,
      "loss": 1.6614,
      "step": 105
    },
    {
      "epoch": 0.009513552324537785,
      "grad_norm": 29.01144027709961,
      "learning_rate": 0.0,
      "loss": 1.6809,
      "step": 106
    },
    {
      "epoch": 0.0096033028181655,
      "grad_norm": 24.549089431762695,
      "learning_rate": 0.0,
      "loss": 1.6267,
      "step": 107
    },
    {
      "epoch": 0.009693053311793215,
      "grad_norm": 32.655853271484375,
      "learning_rate": 0.0,
      "loss": 1.6218,
      "step": 108
    },
    {
      "epoch": 0.00978280380542093,
      "grad_norm": 28.411935806274414,
      "learning_rate": 0.0,
      "loss": 1.6394,
      "step": 109
    },
    {
      "epoch": 0.009872554299048646,
      "grad_norm": 25.139461517333984,
      "learning_rate": 0.0,
      "loss": 1.588,
      "step": 110
    },
    {
      "epoch": 0.00996230479267636,
      "grad_norm": 26.537015914916992,
      "learning_rate": 0.0,
      "loss": 1.606,
      "step": 111
    },
    {
      "epoch": 0.010052055286304075,
      "grad_norm": 24.64919090270996,
      "learning_rate": 0.0,
      "loss": 1.5055,
      "step": 112
    },
    {
      "epoch": 0.01014180577993179,
      "grad_norm": 23.487625122070312,
      "learning_rate": 0.0,
      "loss": 1.6501,
      "step": 113
    },
    {
      "epoch": 0.010231556273559504,
      "grad_norm": 21.49905014038086,
      "learning_rate": 0.0,
      "loss": 1.5984,
      "step": 114
    },
    {
      "epoch": 0.010321306767187219,
      "grad_norm": 25.80181884765625,
      "learning_rate": 0.0,
      "loss": 1.6862,
      "step": 115
    },
    {
      "epoch": 0.010411057260814935,
      "grad_norm": 30.55630874633789,
      "learning_rate": 0.0,
      "loss": 1.6326,
      "step": 116
    },
    {
      "epoch": 0.01050080775444265,
      "grad_norm": 22.829389572143555,
      "learning_rate": 0.0,
      "loss": 1.7004,
      "step": 117
    },
    {
      "epoch": 0.010590558248070365,
      "grad_norm": 23.39740562438965,
      "learning_rate": 0.0,
      "loss": 1.5444,
      "step": 118
    },
    {
      "epoch": 0.01068030874169808,
      "grad_norm": 27.882919311523438,
      "learning_rate": 0.0,
      "loss": 1.5899,
      "step": 119
    },
    {
      "epoch": 0.010770059235325794,
      "grad_norm": 24.941267013549805,
      "learning_rate": 0.0,
      "loss": 1.624,
      "step": 120
    },
    {
      "epoch": 0.010859809728953508,
      "grad_norm": 26.025423049926758,
      "learning_rate": 0.0,
      "loss": 1.6079,
      "step": 121
    },
    {
      "epoch": 0.010949560222581225,
      "grad_norm": 29.621849060058594,
      "learning_rate": 0.0,
      "loss": 1.6209,
      "step": 122
    },
    {
      "epoch": 0.01103931071620894,
      "grad_norm": 24.678693771362305,
      "learning_rate": 0.0,
      "loss": 1.571,
      "step": 123
    },
    {
      "epoch": 0.011129061209836654,
      "grad_norm": 25.787193298339844,
      "learning_rate": 0.0,
      "loss": 1.6492,
      "step": 124
    },
    {
      "epoch": 0.011218811703464369,
      "grad_norm": 27.79629135131836,
      "learning_rate": 0.0,
      "loss": 1.7211,
      "step": 125
    },
    {
      "epoch": 0.011308562197092083,
      "grad_norm": 28.35909652709961,
      "learning_rate": 0.0,
      "loss": 1.7211,
      "step": 126
    },
    {
      "epoch": 0.0113983126907198,
      "grad_norm": 33.9471435546875,
      "learning_rate": 0.0,
      "loss": 1.7715,
      "step": 127
    },
    {
      "epoch": 0.011488063184347514,
      "grad_norm": 24.056140899658203,
      "learning_rate": 0.0,
      "loss": 1.4952,
      "step": 128
    },
    {
      "epoch": 0.011577813677975229,
      "grad_norm": 26.650482177734375,
      "learning_rate": 0.0,
      "loss": 1.6498,
      "step": 129
    },
    {
      "epoch": 0.011667564171602944,
      "grad_norm": 26.58772087097168,
      "learning_rate": 0.0,
      "loss": 1.6648,
      "step": 130
    },
    {
      "epoch": 0.011757314665230658,
      "grad_norm": 33.411041259765625,
      "learning_rate": 0.0,
      "loss": 1.6284,
      "step": 131
    },
    {
      "epoch": 0.011847065158858373,
      "grad_norm": 24.956775665283203,
      "learning_rate": 0.0,
      "loss": 1.6099,
      "step": 132
    },
    {
      "epoch": 0.01193681565248609,
      "grad_norm": 25.417930603027344,
      "learning_rate": 0.0,
      "loss": 1.6513,
      "step": 133
    },
    {
      "epoch": 0.012026566146113804,
      "grad_norm": 68.38418579101562,
      "learning_rate": 0.0,
      "loss": 1.6477,
      "step": 134
    },
    {
      "epoch": 0.012116316639741519,
      "grad_norm": 26.54812240600586,
      "learning_rate": 0.0,
      "loss": 1.6521,
      "step": 135
    },
    {
      "epoch": 0.012206067133369233,
      "grad_norm": 24.520673751831055,
      "learning_rate": 0.0,
      "loss": 1.6225,
      "step": 136
    },
    {
      "epoch": 0.012295817626996948,
      "grad_norm": 23.843868255615234,
      "learning_rate": 0.0,
      "loss": 1.6823,
      "step": 137
    },
    {
      "epoch": 0.012385568120624663,
      "grad_norm": 25.00153160095215,
      "learning_rate": 0.0,
      "loss": 1.677,
      "step": 138
    },
    {
      "epoch": 0.012475318614252379,
      "grad_norm": 36.94251251220703,
      "learning_rate": 0.0,
      "loss": 1.7056,
      "step": 139
    },
    {
      "epoch": 0.012565069107880094,
      "grad_norm": 32.466678619384766,
      "learning_rate": 0.0,
      "loss": 1.566,
      "step": 140
    },
    {
      "epoch": 0.012654819601507808,
      "grad_norm": 27.899810791015625,
      "learning_rate": 0.0,
      "loss": 1.7187,
      "step": 141
    },
    {
      "epoch": 0.012744570095135523,
      "grad_norm": 27.21063995361328,
      "learning_rate": 0.0,
      "loss": 1.5697,
      "step": 142
    },
    {
      "epoch": 0.012834320588763238,
      "grad_norm": 28.91971206665039,
      "learning_rate": 0.0,
      "loss": 1.699,
      "step": 143
    },
    {
      "epoch": 0.012924071082390954,
      "grad_norm": 23.28516960144043,
      "learning_rate": 0.0,
      "loss": 1.6727,
      "step": 144
    },
    {
      "epoch": 0.013013821576018669,
      "grad_norm": 30.023029327392578,
      "learning_rate": 0.0,
      "loss": 1.7205,
      "step": 145
    },
    {
      "epoch": 0.013103572069646383,
      "grad_norm": 29.90929412841797,
      "learning_rate": 0.0,
      "loss": 1.7044,
      "step": 146
    },
    {
      "epoch": 0.013193322563274098,
      "grad_norm": 28.362335205078125,
      "learning_rate": 0.0,
      "loss": 1.5933,
      "step": 147
    },
    {
      "epoch": 0.013283073056901813,
      "grad_norm": 28.555744171142578,
      "learning_rate": 0.0,
      "loss": 1.5938,
      "step": 148
    },
    {
      "epoch": 0.013372823550529527,
      "grad_norm": 30.524290084838867,
      "learning_rate": 0.0,
      "loss": 1.6475,
      "step": 149
    },
    {
      "epoch": 0.013462574044157244,
      "grad_norm": 29.631439208984375,
      "learning_rate": 0.0,
      "loss": 1.5917,
      "step": 150
    },
    {
      "epoch": 0.013552324537784958,
      "grad_norm": 23.193885803222656,
      "learning_rate": 0.0,
      "loss": 1.6172,
      "step": 151
    },
    {
      "epoch": 0.013642075031412673,
      "grad_norm": 29.485763549804688,
      "learning_rate": 0.0,
      "loss": 1.5973,
      "step": 152
    },
    {
      "epoch": 0.013731825525040387,
      "grad_norm": 25.890954971313477,
      "learning_rate": 0.0,
      "loss": 1.6118,
      "step": 153
    },
    {
      "epoch": 0.013821576018668102,
      "grad_norm": 26.54442596435547,
      "learning_rate": 0.0,
      "loss": 1.5711,
      "step": 154
    },
    {
      "epoch": 0.013911326512295817,
      "grad_norm": 26.887123107910156,
      "learning_rate": 0.0,
      "loss": 1.6367,
      "step": 155
    },
    {
      "epoch": 0.014001077005923533,
      "grad_norm": 27.34661102294922,
      "learning_rate": 0.0,
      "loss": 1.5905,
      "step": 156
    },
    {
      "epoch": 0.014090827499551248,
      "grad_norm": 27.807676315307617,
      "learning_rate": 0.0,
      "loss": 1.6057,
      "step": 157
    },
    {
      "epoch": 0.014180577993178962,
      "grad_norm": 27.144161224365234,
      "learning_rate": 0.0,
      "loss": 1.6829,
      "step": 158
    },
    {
      "epoch": 0.014270328486806677,
      "grad_norm": 68.18128967285156,
      "learning_rate": 0.0,
      "loss": 1.7234,
      "step": 159
    },
    {
      "epoch": 0.014360078980434392,
      "grad_norm": 25.117521286010742,
      "learning_rate": 0.0,
      "loss": 1.6,
      "step": 160
    },
    {
      "epoch": 0.014449829474062108,
      "grad_norm": 36.553646087646484,
      "learning_rate": 0.0,
      "loss": 1.7443,
      "step": 161
    },
    {
      "epoch": 0.014539579967689823,
      "grad_norm": 24.447864532470703,
      "learning_rate": 0.0,
      "loss": 1.6482,
      "step": 162
    },
    {
      "epoch": 0.014629330461317537,
      "grad_norm": 53.75963592529297,
      "learning_rate": 0.0,
      "loss": 1.6293,
      "step": 163
    },
    {
      "epoch": 0.014719080954945252,
      "grad_norm": 30.21377944946289,
      "learning_rate": 0.0,
      "loss": 1.6984,
      "step": 164
    },
    {
      "epoch": 0.014808831448572967,
      "grad_norm": 26.268896102905273,
      "learning_rate": 0.0,
      "loss": 1.7175,
      "step": 165
    },
    {
      "epoch": 0.014898581942200681,
      "grad_norm": 32.049495697021484,
      "learning_rate": 0.0,
      "loss": 1.6024,
      "step": 166
    },
    {
      "epoch": 0.014988332435828398,
      "grad_norm": 26.462749481201172,
      "learning_rate": 0.0,
      "loss": 1.6297,
      "step": 167
    },
    {
      "epoch": 0.015078082929456112,
      "grad_norm": 54.01758575439453,
      "learning_rate": 0.0,
      "loss": 1.6495,
      "step": 168
    },
    {
      "epoch": 0.015167833423083827,
      "grad_norm": 30.429370880126953,
      "learning_rate": 0.0,
      "loss": 1.7645,
      "step": 169
    },
    {
      "epoch": 0.015257583916711542,
      "grad_norm": 24.266489028930664,
      "learning_rate": 0.0,
      "loss": 1.5891,
      "step": 170
    },
    {
      "epoch": 0.015347334410339256,
      "grad_norm": 27.328699111938477,
      "learning_rate": 0.0,
      "loss": 1.6977,
      "step": 171
    },
    {
      "epoch": 0.015437084903966973,
      "grad_norm": 31.681076049804688,
      "learning_rate": 0.0,
      "loss": 1.5349,
      "step": 172
    },
    {
      "epoch": 0.015526835397594687,
      "grad_norm": 116.81834411621094,
      "learning_rate": 0.0,
      "loss": 1.6725,
      "step": 173
    },
    {
      "epoch": 0.015616585891222402,
      "grad_norm": 22.820323944091797,
      "learning_rate": 0.0,
      "loss": 1.591,
      "step": 174
    },
    {
      "epoch": 0.01570633638485012,
      "grad_norm": 25.30412483215332,
      "learning_rate": 0.0,
      "loss": 1.6166,
      "step": 175
    },
    {
      "epoch": 0.015796086878477833,
      "grad_norm": 24.341304779052734,
      "learning_rate": 0.0,
      "loss": 1.6305,
      "step": 176
    },
    {
      "epoch": 0.015885837372105548,
      "grad_norm": 30.764705657958984,
      "learning_rate": 0.0,
      "loss": 1.708,
      "step": 177
    },
    {
      "epoch": 0.015975587865733262,
      "grad_norm": 23.94184112548828,
      "learning_rate": 0.0,
      "loss": 1.5733,
      "step": 178
    },
    {
      "epoch": 0.016065338359360977,
      "grad_norm": 21.23302459716797,
      "learning_rate": 0.0,
      "loss": 1.47,
      "step": 179
    },
    {
      "epoch": 0.01615508885298869,
      "grad_norm": 25.454572677612305,
      "learning_rate": 0.0,
      "loss": 1.499,
      "step": 180
    },
    {
      "epoch": 0.016244839346616406,
      "grad_norm": 24.63162612915039,
      "learning_rate": 0.0,
      "loss": 1.6077,
      "step": 181
    },
    {
      "epoch": 0.01633458984024412,
      "grad_norm": 26.149456024169922,
      "learning_rate": 0.0,
      "loss": 1.7249,
      "step": 182
    },
    {
      "epoch": 0.016424340333871836,
      "grad_norm": 24.15140151977539,
      "learning_rate": 0.0,
      "loss": 1.7018,
      "step": 183
    },
    {
      "epoch": 0.01651409082749955,
      "grad_norm": 24.022693634033203,
      "learning_rate": 0.0,
      "loss": 1.5258,
      "step": 184
    },
    {
      "epoch": 0.016603841321127265,
      "grad_norm": 23.04458236694336,
      "learning_rate": 0.0,
      "loss": 1.6135,
      "step": 185
    },
    {
      "epoch": 0.01669359181475498,
      "grad_norm": 24.860586166381836,
      "learning_rate": 0.0,
      "loss": 1.5738,
      "step": 186
    },
    {
      "epoch": 0.016783342308382698,
      "grad_norm": 24.965675354003906,
      "learning_rate": 0.0,
      "loss": 1.7075,
      "step": 187
    },
    {
      "epoch": 0.016873092802010412,
      "grad_norm": 24.327255249023438,
      "learning_rate": 0.0,
      "loss": 1.6868,
      "step": 188
    },
    {
      "epoch": 0.016962843295638127,
      "grad_norm": 27.1917667388916,
      "learning_rate": 0.0,
      "loss": 1.5698,
      "step": 189
    },
    {
      "epoch": 0.01705259378926584,
      "grad_norm": 33.853885650634766,
      "learning_rate": 0.0,
      "loss": 1.5572,
      "step": 190
    },
    {
      "epoch": 0.017142344282893556,
      "grad_norm": 27.237010955810547,
      "learning_rate": 0.0,
      "loss": 1.6761,
      "step": 191
    },
    {
      "epoch": 0.01723209477652127,
      "grad_norm": 28.328807830810547,
      "learning_rate": 0.0,
      "loss": 1.6093,
      "step": 192
    },
    {
      "epoch": 0.017321845270148985,
      "grad_norm": 26.07171630859375,
      "learning_rate": 0.0,
      "loss": 1.599,
      "step": 193
    },
    {
      "epoch": 0.0174115957637767,
      "grad_norm": 22.59205436706543,
      "learning_rate": 0.0,
      "loss": 1.5664,
      "step": 194
    },
    {
      "epoch": 0.017501346257404415,
      "grad_norm": 26.775779724121094,
      "learning_rate": 0.0,
      "loss": 1.5313,
      "step": 195
    },
    {
      "epoch": 0.01759109675103213,
      "grad_norm": 20.262203216552734,
      "learning_rate": 0.0,
      "loss": 1.5133,
      "step": 196
    },
    {
      "epoch": 0.017680847244659844,
      "grad_norm": 28.25116539001465,
      "learning_rate": 0.0,
      "loss": 1.6056,
      "step": 197
    },
    {
      "epoch": 0.017770597738287562,
      "grad_norm": 24.36191749572754,
      "learning_rate": 0.0,
      "loss": 1.5654,
      "step": 198
    },
    {
      "epoch": 0.017860348231915277,
      "grad_norm": 17.876996994018555,
      "learning_rate": 0.0,
      "loss": 1.5093,
      "step": 199
    },
    {
      "epoch": 0.01795009872554299,
      "grad_norm": 21.389055252075195,
      "learning_rate": 0.0,
      "loss": 1.6105,
      "step": 200
    },
    {
      "epoch": 0.018039849219170706,
      "grad_norm": 26.412988662719727,
      "learning_rate": 0.0,
      "loss": 1.6527,
      "step": 201
    },
    {
      "epoch": 0.01812959971279842,
      "grad_norm": 27.351022720336914,
      "learning_rate": 0.0,
      "loss": 1.5626,
      "step": 202
    },
    {
      "epoch": 0.018219350206426135,
      "grad_norm": 28.44472312927246,
      "learning_rate": 0.0,
      "loss": 1.6959,
      "step": 203
    },
    {
      "epoch": 0.01830910070005385,
      "grad_norm": 26.055984497070312,
      "learning_rate": 0.0,
      "loss": 1.5782,
      "step": 204
    },
    {
      "epoch": 0.018398851193681565,
      "grad_norm": 42.78719711303711,
      "learning_rate": 0.0,
      "loss": 1.6766,
      "step": 205
    },
    {
      "epoch": 0.01848860168730928,
      "grad_norm": 27.76035499572754,
      "learning_rate": 0.0,
      "loss": 1.5721,
      "step": 206
    },
    {
      "epoch": 0.018578352180936994,
      "grad_norm": 24.954858779907227,
      "learning_rate": 0.0,
      "loss": 1.5336,
      "step": 207
    },
    {
      "epoch": 0.01866810267456471,
      "grad_norm": 39.77936935424805,
      "learning_rate": 0.0,
      "loss": 1.7171,
      "step": 208
    },
    {
      "epoch": 0.018757853168192427,
      "grad_norm": 24.20283317565918,
      "learning_rate": 0.0,
      "loss": 1.5991,
      "step": 209
    },
    {
      "epoch": 0.01884760366182014,
      "grad_norm": 25.497106552124023,
      "learning_rate": 0.0,
      "loss": 1.7055,
      "step": 210
    },
    {
      "epoch": 0.018937354155447856,
      "grad_norm": 27.420166015625,
      "learning_rate": 0.0,
      "loss": 1.6766,
      "step": 211
    },
    {
      "epoch": 0.01902710464907557,
      "grad_norm": 23.779706954956055,
      "learning_rate": 0.0,
      "loss": 1.53,
      "step": 212
    },
    {
      "epoch": 0.019116855142703285,
      "grad_norm": 27.148134231567383,
      "learning_rate": 0.0,
      "loss": 1.6545,
      "step": 213
    },
    {
      "epoch": 0.019206605636331,
      "grad_norm": 30.76536750793457,
      "learning_rate": 0.0,
      "loss": 1.6346,
      "step": 214
    },
    {
      "epoch": 0.019296356129958715,
      "grad_norm": 25.275840759277344,
      "learning_rate": 0.0,
      "loss": 1.6218,
      "step": 215
    },
    {
      "epoch": 0.01938610662358643,
      "grad_norm": 28.269084930419922,
      "learning_rate": 0.0,
      "loss": 1.6902,
      "step": 216
    },
    {
      "epoch": 0.019475857117214144,
      "grad_norm": 24.121854782104492,
      "learning_rate": 0.0,
      "loss": 1.6043,
      "step": 217
    },
    {
      "epoch": 0.01956560761084186,
      "grad_norm": 24.281679153442383,
      "learning_rate": 0.0,
      "loss": 1.5488,
      "step": 218
    },
    {
      "epoch": 0.019655358104469573,
      "grad_norm": 28.247377395629883,
      "learning_rate": 0.0,
      "loss": 1.6445,
      "step": 219
    },
    {
      "epoch": 0.01974510859809729,
      "grad_norm": 25.967994689941406,
      "learning_rate": 0.0,
      "loss": 1.5751,
      "step": 220
    },
    {
      "epoch": 0.019834859091725006,
      "grad_norm": 27.00880241394043,
      "learning_rate": 0.0,
      "loss": 1.5751,
      "step": 221
    },
    {
      "epoch": 0.01992460958535272,
      "grad_norm": 22.650463104248047,
      "learning_rate": 0.0,
      "loss": 1.5714,
      "step": 222
    },
    {
      "epoch": 0.020014360078980435,
      "grad_norm": 24.539714813232422,
      "learning_rate": 0.0,
      "loss": 1.5104,
      "step": 223
    },
    {
      "epoch": 0.02010411057260815,
      "grad_norm": 27.062562942504883,
      "learning_rate": 0.0,
      "loss": 1.5842,
      "step": 224
    },
    {
      "epoch": 0.020193861066235864,
      "grad_norm": 19.6573543548584,
      "learning_rate": 0.0,
      "loss": 1.5134,
      "step": 225
    },
    {
      "epoch": 0.02028361155986358,
      "grad_norm": 27.575824737548828,
      "learning_rate": 0.0,
      "loss": 1.5735,
      "step": 226
    },
    {
      "epoch": 0.020373362053491294,
      "grad_norm": 26.58475112915039,
      "learning_rate": 0.0,
      "loss": 1.5477,
      "step": 227
    },
    {
      "epoch": 0.02046311254711901,
      "grad_norm": 25.03936195373535,
      "learning_rate": 0.0,
      "loss": 1.6324,
      "step": 228
    },
    {
      "epoch": 0.020552863040746723,
      "grad_norm": 25.633708953857422,
      "learning_rate": 0.0,
      "loss": 1.5691,
      "step": 229
    },
    {
      "epoch": 0.020642613534374438,
      "grad_norm": 22.177526473999023,
      "learning_rate": 0.0,
      "loss": 1.5946,
      "step": 230
    },
    {
      "epoch": 0.020732364028002152,
      "grad_norm": 29.579116821289062,
      "learning_rate": 0.0,
      "loss": 1.6804,
      "step": 231
    },
    {
      "epoch": 0.02082211452162987,
      "grad_norm": 26.702564239501953,
      "learning_rate": 0.0,
      "loss": 1.7167,
      "step": 232
    },
    {
      "epoch": 0.020911865015257585,
      "grad_norm": 25.58138084411621,
      "learning_rate": 0.0,
      "loss": 1.5962,
      "step": 233
    },
    {
      "epoch": 0.0210016155088853,
      "grad_norm": 24.141603469848633,
      "learning_rate": 0.0,
      "loss": 1.5964,
      "step": 234
    },
    {
      "epoch": 0.021091366002513014,
      "grad_norm": 31.886924743652344,
      "learning_rate": 0.0,
      "loss": 1.719,
      "step": 235
    },
    {
      "epoch": 0.02118111649614073,
      "grad_norm": 20.47477149963379,
      "learning_rate": 0.0,
      "loss": 1.5834,
      "step": 236
    },
    {
      "epoch": 0.021270866989768444,
      "grad_norm": 23.516626358032227,
      "learning_rate": 0.0,
      "loss": 1.5703,
      "step": 237
    },
    {
      "epoch": 0.02136061748339616,
      "grad_norm": 28.095256805419922,
      "learning_rate": 0.0,
      "loss": 1.5908,
      "step": 238
    },
    {
      "epoch": 0.021450367977023873,
      "grad_norm": 25.03912925720215,
      "learning_rate": 0.0,
      "loss": 1.549,
      "step": 239
    },
    {
      "epoch": 0.021540118470651588,
      "grad_norm": 22.026443481445312,
      "learning_rate": 0.0,
      "loss": 1.561,
      "step": 240
    },
    {
      "epoch": 0.021629868964279302,
      "grad_norm": 26.729867935180664,
      "learning_rate": 0.0,
      "loss": 1.627,
      "step": 241
    },
    {
      "epoch": 0.021719619457907017,
      "grad_norm": 25.978038787841797,
      "learning_rate": 0.0,
      "loss": 1.6472,
      "step": 242
    },
    {
      "epoch": 0.021809369951534735,
      "grad_norm": 29.23665428161621,
      "learning_rate": 0.0,
      "loss": 1.6,
      "step": 243
    },
    {
      "epoch": 0.02189912044516245,
      "grad_norm": 25.264646530151367,
      "learning_rate": 0.0,
      "loss": 1.6023,
      "step": 244
    },
    {
      "epoch": 0.021988870938790164,
      "grad_norm": 22.278661727905273,
      "learning_rate": 0.0,
      "loss": 1.5996,
      "step": 245
    },
    {
      "epoch": 0.02207862143241788,
      "grad_norm": 30.503536224365234,
      "learning_rate": 0.0,
      "loss": 1.6657,
      "step": 246
    },
    {
      "epoch": 0.022168371926045594,
      "grad_norm": 21.31138801574707,
      "learning_rate": 0.0,
      "loss": 1.6813,
      "step": 247
    },
    {
      "epoch": 0.022258122419673308,
      "grad_norm": 26.60215950012207,
      "learning_rate": 0.0,
      "loss": 1.6075,
      "step": 248
    },
    {
      "epoch": 0.022347872913301023,
      "grad_norm": 26.944780349731445,
      "learning_rate": 0.0,
      "loss": 1.6953,
      "step": 249
    },
    {
      "epoch": 0.022437623406928738,
      "grad_norm": 22.51814842224121,
      "learning_rate": 0.0,
      "loss": 1.5178,
      "step": 250
    },
    {
      "epoch": 0.022527373900556452,
      "grad_norm": 26.46634292602539,
      "learning_rate": 0.0,
      "loss": 1.6438,
      "step": 251
    },
    {
      "epoch": 0.022617124394184167,
      "grad_norm": 25.726228713989258,
      "learning_rate": 0.0,
      "loss": 1.6175,
      "step": 252
    },
    {
      "epoch": 0.02270687488781188,
      "grad_norm": 32.07600021362305,
      "learning_rate": 0.0,
      "loss": 1.7334,
      "step": 253
    },
    {
      "epoch": 0.0227966253814396,
      "grad_norm": 26.297969818115234,
      "learning_rate": 0.0,
      "loss": 1.5661,
      "step": 254
    },
    {
      "epoch": 0.022886375875067314,
      "grad_norm": 28.597881317138672,
      "learning_rate": 0.0,
      "loss": 1.6242,
      "step": 255
    },
    {
      "epoch": 0.02297612636869503,
      "grad_norm": 24.996328353881836,
      "learning_rate": 0.0,
      "loss": 1.5939,
      "step": 256
    },
    {
      "epoch": 0.023065876862322744,
      "grad_norm": 26.335237503051758,
      "learning_rate": 0.0,
      "loss": 1.6679,
      "step": 257
    },
    {
      "epoch": 0.023155627355950458,
      "grad_norm": 22.724302291870117,
      "learning_rate": 0.0,
      "loss": 1.4937,
      "step": 258
    },
    {
      "epoch": 0.023245377849578173,
      "grad_norm": 30.7111759185791,
      "learning_rate": 0.0,
      "loss": 1.5591,
      "step": 259
    },
    {
      "epoch": 0.023335128343205887,
      "grad_norm": 30.8575496673584,
      "learning_rate": 0.0,
      "loss": 1.6862,
      "step": 260
    },
    {
      "epoch": 0.023424878836833602,
      "grad_norm": 25.213157653808594,
      "learning_rate": 0.0,
      "loss": 1.5594,
      "step": 261
    },
    {
      "epoch": 0.023514629330461317,
      "grad_norm": 23.897432327270508,
      "learning_rate": 0.0,
      "loss": 1.5936,
      "step": 262
    },
    {
      "epoch": 0.02360437982408903,
      "grad_norm": 26.779218673706055,
      "learning_rate": 0.0,
      "loss": 1.6129,
      "step": 263
    },
    {
      "epoch": 0.023694130317716746,
      "grad_norm": 20.31458282470703,
      "learning_rate": 0.0,
      "loss": 1.6005,
      "step": 264
    },
    {
      "epoch": 0.02378388081134446,
      "grad_norm": 22.858308792114258,
      "learning_rate": 0.0,
      "loss": 1.5863,
      "step": 265
    },
    {
      "epoch": 0.02387363130497218,
      "grad_norm": 23.582307815551758,
      "learning_rate": 0.0,
      "loss": 1.6883,
      "step": 266
    },
    {
      "epoch": 0.023963381798599893,
      "grad_norm": 22.910091400146484,
      "learning_rate": 0.0,
      "loss": 1.739,
      "step": 267
    },
    {
      "epoch": 0.024053132292227608,
      "grad_norm": 28.596759796142578,
      "learning_rate": 0.0,
      "loss": 1.5729,
      "step": 268
    },
    {
      "epoch": 0.024142882785855323,
      "grad_norm": 25.23520851135254,
      "learning_rate": 0.0,
      "loss": 1.6306,
      "step": 269
    },
    {
      "epoch": 0.024232633279483037,
      "grad_norm": 30.236614227294922,
      "learning_rate": 0.0,
      "loss": 1.6001,
      "step": 270
    },
    {
      "epoch": 0.024322383773110752,
      "grad_norm": 27.570302963256836,
      "learning_rate": 0.0,
      "loss": 1.6553,
      "step": 271
    },
    {
      "epoch": 0.024412134266738467,
      "grad_norm": 25.421794891357422,
      "learning_rate": 0.0,
      "loss": 1.7755,
      "step": 272
    },
    {
      "epoch": 0.02450188476036618,
      "grad_norm": 31.59478759765625,
      "learning_rate": 0.0,
      "loss": 1.5431,
      "step": 273
    },
    {
      "epoch": 0.024591635253993896,
      "grad_norm": 24.42188835144043,
      "learning_rate": 0.0,
      "loss": 1.606,
      "step": 274
    },
    {
      "epoch": 0.02468138574762161,
      "grad_norm": 28.0344295501709,
      "learning_rate": 0.0,
      "loss": 1.5973,
      "step": 275
    },
    {
      "epoch": 0.024771136241249325,
      "grad_norm": 24.979063034057617,
      "learning_rate": 0.0,
      "loss": 1.6649,
      "step": 276
    },
    {
      "epoch": 0.024860886734877043,
      "grad_norm": 20.05938148498535,
      "learning_rate": 0.0,
      "loss": 1.5179,
      "step": 277
    },
    {
      "epoch": 0.024950637228504758,
      "grad_norm": 66.5769271850586,
      "learning_rate": 0.0,
      "loss": 1.6402,
      "step": 278
    },
    {
      "epoch": 0.025040387722132473,
      "grad_norm": 25.526540756225586,
      "learning_rate": 0.0,
      "loss": 1.5098,
      "step": 279
    },
    {
      "epoch": 0.025130138215760187,
      "grad_norm": 26.7034969329834,
      "learning_rate": 0.0,
      "loss": 1.6898,
      "step": 280
    },
    {
      "epoch": 0.025219888709387902,
      "grad_norm": 25.066532135009766,
      "learning_rate": 0.0,
      "loss": 1.5882,
      "step": 281
    },
    {
      "epoch": 0.025309639203015617,
      "grad_norm": 34.833709716796875,
      "learning_rate": 0.0,
      "loss": 1.6992,
      "step": 282
    },
    {
      "epoch": 0.02539938969664333,
      "grad_norm": 25.075712203979492,
      "learning_rate": 0.0,
      "loss": 1.5229,
      "step": 283
    },
    {
      "epoch": 0.025489140190271046,
      "grad_norm": 22.574996948242188,
      "learning_rate": 0.0,
      "loss": 1.6599,
      "step": 284
    },
    {
      "epoch": 0.02557889068389876,
      "grad_norm": 21.028139114379883,
      "learning_rate": 0.0,
      "loss": 1.623,
      "step": 285
    },
    {
      "epoch": 0.025668641177526475,
      "grad_norm": 26.134349822998047,
      "learning_rate": 0.0,
      "loss": 1.6421,
      "step": 286
    },
    {
      "epoch": 0.02575839167115419,
      "grad_norm": 24.375328063964844,
      "learning_rate": 0.0,
      "loss": 1.5637,
      "step": 287
    },
    {
      "epoch": 0.025848142164781908,
      "grad_norm": 26.15487289428711,
      "learning_rate": 0.0,
      "loss": 1.6251,
      "step": 288
    },
    {
      "epoch": 0.025937892658409623,
      "grad_norm": 26.484331130981445,
      "learning_rate": 0.0,
      "loss": 1.5786,
      "step": 289
    },
    {
      "epoch": 0.026027643152037337,
      "grad_norm": 24.671823501586914,
      "learning_rate": 0.0,
      "loss": 1.536,
      "step": 290
    },
    {
      "epoch": 0.026117393645665052,
      "grad_norm": 31.600759506225586,
      "learning_rate": 0.0,
      "loss": 1.6108,
      "step": 291
    },
    {
      "epoch": 0.026207144139292766,
      "grad_norm": 28.073986053466797,
      "learning_rate": 0.0,
      "loss": 1.6098,
      "step": 292
    },
    {
      "epoch": 0.02629689463292048,
      "grad_norm": 26.04122543334961,
      "learning_rate": 0.0,
      "loss": 1.7105,
      "step": 293
    },
    {
      "epoch": 0.026386645126548196,
      "grad_norm": 27.867136001586914,
      "learning_rate": 0.0,
      "loss": 1.7104,
      "step": 294
    },
    {
      "epoch": 0.02647639562017591,
      "grad_norm": 30.232696533203125,
      "learning_rate": 0.0,
      "loss": 1.608,
      "step": 295
    },
    {
      "epoch": 0.026566146113803625,
      "grad_norm": 25.630510330200195,
      "learning_rate": 0.0,
      "loss": 1.5972,
      "step": 296
    },
    {
      "epoch": 0.02665589660743134,
      "grad_norm": 28.827566146850586,
      "learning_rate": 0.0,
      "loss": 1.5898,
      "step": 297
    },
    {
      "epoch": 0.026745647101059054,
      "grad_norm": 24.100324630737305,
      "learning_rate": 0.0,
      "loss": 1.5283,
      "step": 298
    },
    {
      "epoch": 0.026835397594686772,
      "grad_norm": 31.816282272338867,
      "learning_rate": 0.0,
      "loss": 1.6896,
      "step": 299
    },
    {
      "epoch": 0.026925148088314487,
      "grad_norm": 22.669816970825195,
      "learning_rate": 0.0,
      "loss": 1.5716,
      "step": 300
    },
    {
      "epoch": 0.027014898581942202,
      "grad_norm": 22.20302963256836,
      "learning_rate": 0.0,
      "loss": 1.6077,
      "step": 301
    },
    {
      "epoch": 0.027104649075569916,
      "grad_norm": 25.468597412109375,
      "learning_rate": 0.0,
      "loss": 1.62,
      "step": 302
    },
    {
      "epoch": 0.02719439956919763,
      "grad_norm": 27.734203338623047,
      "learning_rate": 0.0,
      "loss": 1.6182,
      "step": 303
    },
    {
      "epoch": 0.027284150062825346,
      "grad_norm": 24.819644927978516,
      "learning_rate": 0.0,
      "loss": 1.6996,
      "step": 304
    },
    {
      "epoch": 0.02737390055645306,
      "grad_norm": 25.217906951904297,
      "learning_rate": 0.0,
      "loss": 1.6123,
      "step": 305
    },
    {
      "epoch": 0.027463651050080775,
      "grad_norm": 23.867185592651367,
      "learning_rate": 0.0,
      "loss": 1.586,
      "step": 306
    },
    {
      "epoch": 0.02755340154370849,
      "grad_norm": 22.893999099731445,
      "learning_rate": 0.0,
      "loss": 1.6318,
      "step": 307
    },
    {
      "epoch": 0.027643152037336204,
      "grad_norm": 25.554393768310547,
      "learning_rate": 0.0,
      "loss": 1.6432,
      "step": 308
    },
    {
      "epoch": 0.02773290253096392,
      "grad_norm": 76.14978790283203,
      "learning_rate": 0.0,
      "loss": 1.6417,
      "step": 309
    },
    {
      "epoch": 0.027822653024591634,
      "grad_norm": 25.462505340576172,
      "learning_rate": 0.0,
      "loss": 1.6446,
      "step": 310
    },
    {
      "epoch": 0.02791240351821935,
      "grad_norm": 26.524450302124023,
      "learning_rate": 0.0,
      "loss": 1.6047,
      "step": 311
    },
    {
      "epoch": 0.028002154011847066,
      "grad_norm": 22.739015579223633,
      "learning_rate": 0.0,
      "loss": 1.5656,
      "step": 312
    },
    {
      "epoch": 0.02809190450547478,
      "grad_norm": 25.717512130737305,
      "learning_rate": 0.0,
      "loss": 1.5787,
      "step": 313
    },
    {
      "epoch": 0.028181654999102496,
      "grad_norm": 28.861087799072266,
      "learning_rate": 0.0,
      "loss": 1.6341,
      "step": 314
    },
    {
      "epoch": 0.02827140549273021,
      "grad_norm": 30.106767654418945,
      "learning_rate": 0.0,
      "loss": 1.6578,
      "step": 315
    },
    {
      "epoch": 0.028361155986357925,
      "grad_norm": 26.54692840576172,
      "learning_rate": 0.0,
      "loss": 1.6003,
      "step": 316
    },
    {
      "epoch": 0.02845090647998564,
      "grad_norm": 27.09174156188965,
      "learning_rate": 0.0,
      "loss": 1.7339,
      "step": 317
    },
    {
      "epoch": 0.028540656973613354,
      "grad_norm": 26.497228622436523,
      "learning_rate": 0.0,
      "loss": 1.5796,
      "step": 318
    },
    {
      "epoch": 0.02863040746724107,
      "grad_norm": 27.239587783813477,
      "learning_rate": 0.0,
      "loss": 1.6022,
      "step": 319
    },
    {
      "epoch": 0.028720157960868783,
      "grad_norm": 23.474882125854492,
      "learning_rate": 0.0,
      "loss": 1.53,
      "step": 320
    },
    {
      "epoch": 0.028809908454496498,
      "grad_norm": 33.201419830322266,
      "learning_rate": 0.0,
      "loss": 1.6432,
      "step": 321
    },
    {
      "epoch": 0.028899658948124216,
      "grad_norm": 26.489097595214844,
      "learning_rate": 0.0,
      "loss": 1.5573,
      "step": 322
    },
    {
      "epoch": 0.02898940944175193,
      "grad_norm": 73.55593872070312,
      "learning_rate": 0.0,
      "loss": 1.6328,
      "step": 323
    },
    {
      "epoch": 0.029079159935379646,
      "grad_norm": 25.8769588470459,
      "learning_rate": 0.0,
      "loss": 1.5787,
      "step": 324
    },
    {
      "epoch": 0.02916891042900736,
      "grad_norm": 25.740829467773438,
      "learning_rate": 0.0,
      "loss": 1.5816,
      "step": 325
    },
    {
      "epoch": 0.029258660922635075,
      "grad_norm": 22.46575355529785,
      "learning_rate": 0.0,
      "loss": 1.7337,
      "step": 326
    },
    {
      "epoch": 0.02934841141626279,
      "grad_norm": 26.186201095581055,
      "learning_rate": 0.0,
      "loss": 1.6313,
      "step": 327
    },
    {
      "epoch": 0.029438161909890504,
      "grad_norm": 24.705419540405273,
      "learning_rate": 0.0,
      "loss": 1.6996,
      "step": 328
    },
    {
      "epoch": 0.02952791240351822,
      "grad_norm": 25.121294021606445,
      "learning_rate": 0.0,
      "loss": 1.6446,
      "step": 329
    },
    {
      "epoch": 0.029617662897145933,
      "grad_norm": 27.419153213500977,
      "learning_rate": 0.0,
      "loss": 1.6037,
      "step": 330
    },
    {
      "epoch": 0.029707413390773648,
      "grad_norm": 26.58218765258789,
      "learning_rate": 0.0,
      "loss": 1.6135,
      "step": 331
    },
    {
      "epoch": 0.029797163884401363,
      "grad_norm": 23.73853874206543,
      "learning_rate": 0.0,
      "loss": 1.6782,
      "step": 332
    },
    {
      "epoch": 0.02988691437802908,
      "grad_norm": 21.760229110717773,
      "learning_rate": 0.0,
      "loss": 1.5563,
      "step": 333
    },
    {
      "epoch": 0.029976664871656795,
      "grad_norm": 23.958532333374023,
      "learning_rate": 0.0,
      "loss": 1.5816,
      "step": 334
    },
    {
      "epoch": 0.03006641536528451,
      "grad_norm": 36.571533203125,
      "learning_rate": 0.0,
      "loss": 1.7503,
      "step": 335
    },
    {
      "epoch": 0.030156165858912225,
      "grad_norm": 28.630107879638672,
      "learning_rate": 0.0,
      "loss": 1.5798,
      "step": 336
    },
    {
      "epoch": 0.03024591635253994,
      "grad_norm": 28.234304428100586,
      "learning_rate": 0.0,
      "loss": 1.6686,
      "step": 337
    },
    {
      "epoch": 0.030335666846167654,
      "grad_norm": 24.230236053466797,
      "learning_rate": 0.0,
      "loss": 1.5987,
      "step": 338
    },
    {
      "epoch": 0.03042541733979537,
      "grad_norm": 27.191011428833008,
      "learning_rate": 0.0,
      "loss": 1.6578,
      "step": 339
    },
    {
      "epoch": 0.030515167833423083,
      "grad_norm": 24.090852737426758,
      "learning_rate": 0.0,
      "loss": 1.5616,
      "step": 340
    },
    {
      "epoch": 0.030604918327050798,
      "grad_norm": 170.0489501953125,
      "learning_rate": 0.0,
      "loss": 1.6181,
      "step": 341
    },
    {
      "epoch": 0.030694668820678513,
      "grad_norm": 32.364532470703125,
      "learning_rate": 0.0,
      "loss": 1.6184,
      "step": 342
    },
    {
      "epoch": 0.030784419314306227,
      "grad_norm": 23.356056213378906,
      "learning_rate": 0.0,
      "loss": 1.5284,
      "step": 343
    },
    {
      "epoch": 0.030874169807933945,
      "grad_norm": 23.349782943725586,
      "learning_rate": 0.0,
      "loss": 1.5202,
      "step": 344
    },
    {
      "epoch": 0.03096392030156166,
      "grad_norm": 27.446836471557617,
      "learning_rate": 0.0,
      "loss": 1.6094,
      "step": 345
    },
    {
      "epoch": 0.031053670795189375,
      "grad_norm": 23.3684024810791,
      "learning_rate": 0.0,
      "loss": 1.5684,
      "step": 346
    },
    {
      "epoch": 0.03114342128881709,
      "grad_norm": 25.44284439086914,
      "learning_rate": 0.0,
      "loss": 1.5568,
      "step": 347
    },
    {
      "epoch": 0.031233171782444804,
      "grad_norm": 39.80449676513672,
      "learning_rate": 0.0,
      "loss": 1.6324,
      "step": 348
    },
    {
      "epoch": 0.031322922276072515,
      "grad_norm": 76.73066711425781,
      "learning_rate": 0.0,
      "loss": 1.6911,
      "step": 349
    },
    {
      "epoch": 0.03141267276970024,
      "grad_norm": 26.279577255249023,
      "learning_rate": 0.0,
      "loss": 1.6963,
      "step": 350
    },
    {
      "epoch": 0.03150242326332795,
      "grad_norm": 25.08281898498535,
      "learning_rate": 0.0,
      "loss": 1.6541,
      "step": 351
    },
    {
      "epoch": 0.031592173756955666,
      "grad_norm": 25.01738929748535,
      "learning_rate": 0.0,
      "loss": 1.6798,
      "step": 352
    },
    {
      "epoch": 0.03168192425058338,
      "grad_norm": 23.344980239868164,
      "learning_rate": 0.0,
      "loss": 1.6509,
      "step": 353
    },
    {
      "epoch": 0.031771674744211095,
      "grad_norm": 30.11532974243164,
      "learning_rate": 0.0,
      "loss": 1.6869,
      "step": 354
    },
    {
      "epoch": 0.03186142523783881,
      "grad_norm": 25.262859344482422,
      "learning_rate": 0.0,
      "loss": 1.643,
      "step": 355
    },
    {
      "epoch": 0.031951175731466525,
      "grad_norm": 25.43334197998047,
      "learning_rate": 0.0,
      "loss": 1.7172,
      "step": 356
    },
    {
      "epoch": 0.03204092622509424,
      "grad_norm": 31.938138961791992,
      "learning_rate": 0.0,
      "loss": 1.5338,
      "step": 357
    },
    {
      "epoch": 0.032130676718721954,
      "grad_norm": 27.60070037841797,
      "learning_rate": 0.0,
      "loss": 1.6662,
      "step": 358
    },
    {
      "epoch": 0.03222042721234967,
      "grad_norm": 32.54337692260742,
      "learning_rate": 0.0,
      "loss": 1.653,
      "step": 359
    },
    {
      "epoch": 0.03231017770597738,
      "grad_norm": 76.92144775390625,
      "learning_rate": 0.0,
      "loss": 1.7284,
      "step": 360
    },
    {
      "epoch": 0.0323999281996051,
      "grad_norm": 28.603872299194336,
      "learning_rate": 0.0,
      "loss": 1.5689,
      "step": 361
    },
    {
      "epoch": 0.03248967869323281,
      "grad_norm": 24.63780403137207,
      "learning_rate": 0.0,
      "loss": 1.5471,
      "step": 362
    },
    {
      "epoch": 0.03257942918686053,
      "grad_norm": 28.631961822509766,
      "learning_rate": 0.0,
      "loss": 1.5816,
      "step": 363
    },
    {
      "epoch": 0.03266917968048824,
      "grad_norm": 22.750459671020508,
      "learning_rate": 0.0,
      "loss": 1.6141,
      "step": 364
    },
    {
      "epoch": 0.032758930174115956,
      "grad_norm": 48.41012954711914,
      "learning_rate": 0.0,
      "loss": 1.5807,
      "step": 365
    },
    {
      "epoch": 0.03284868066774367,
      "grad_norm": 25.967153549194336,
      "learning_rate": 0.0,
      "loss": 1.599,
      "step": 366
    },
    {
      "epoch": 0.032938431161371386,
      "grad_norm": 31.44416618347168,
      "learning_rate": 0.0,
      "loss": 1.7873,
      "step": 367
    },
    {
      "epoch": 0.0330281816549991,
      "grad_norm": 25.39376449584961,
      "learning_rate": 0.0,
      "loss": 1.6572,
      "step": 368
    },
    {
      "epoch": 0.033117932148626815,
      "grad_norm": 24.9439697265625,
      "learning_rate": 0.0,
      "loss": 1.5556,
      "step": 369
    },
    {
      "epoch": 0.03320768264225453,
      "grad_norm": 23.749431610107422,
      "learning_rate": 0.0,
      "loss": 1.5688,
      "step": 370
    },
    {
      "epoch": 0.033297433135882244,
      "grad_norm": 25.821504592895508,
      "learning_rate": 0.0,
      "loss": 1.5662,
      "step": 371
    },
    {
      "epoch": 0.03338718362950996,
      "grad_norm": 22.38866424560547,
      "learning_rate": 0.0,
      "loss": 1.6258,
      "step": 372
    },
    {
      "epoch": 0.03347693412313768,
      "grad_norm": 24.602458953857422,
      "learning_rate": 0.0,
      "loss": 1.6263,
      "step": 373
    },
    {
      "epoch": 0.033566684616765395,
      "grad_norm": 23.2358455657959,
      "learning_rate": 0.0,
      "loss": 1.5418,
      "step": 374
    },
    {
      "epoch": 0.03365643511039311,
      "grad_norm": 27.544225692749023,
      "learning_rate": 0.0,
      "loss": 1.7851,
      "step": 375
    },
    {
      "epoch": 0.033746185604020824,
      "grad_norm": 24.287410736083984,
      "learning_rate": 0.0,
      "loss": 1.6187,
      "step": 376
    },
    {
      "epoch": 0.03383593609764854,
      "grad_norm": 24.302461624145508,
      "learning_rate": 0.0,
      "loss": 1.5563,
      "step": 377
    },
    {
      "epoch": 0.033925686591276254,
      "grad_norm": 23.87409210205078,
      "learning_rate": 0.0,
      "loss": 1.5351,
      "step": 378
    },
    {
      "epoch": 0.03401543708490397,
      "grad_norm": 25.233537673950195,
      "learning_rate": 0.0,
      "loss": 1.6496,
      "step": 379
    },
    {
      "epoch": 0.03410518757853168,
      "grad_norm": 22.641372680664062,
      "learning_rate": 0.0,
      "loss": 1.5354,
      "step": 380
    },
    {
      "epoch": 0.0341949380721594,
      "grad_norm": 22.940265655517578,
      "learning_rate": 0.0,
      "loss": 1.6269,
      "step": 381
    },
    {
      "epoch": 0.03428468856578711,
      "grad_norm": 27.857929229736328,
      "learning_rate": 0.0,
      "loss": 1.5981,
      "step": 382
    },
    {
      "epoch": 0.03437443905941483,
      "grad_norm": 29.505083084106445,
      "learning_rate": 0.0,
      "loss": 1.5916,
      "step": 383
    },
    {
      "epoch": 0.03446418955304254,
      "grad_norm": 33.077205657958984,
      "learning_rate": 0.0,
      "loss": 1.6464,
      "step": 384
    },
    {
      "epoch": 0.034553940046670256,
      "grad_norm": 26.038761138916016,
      "learning_rate": 0.0,
      "loss": 1.6173,
      "step": 385
    },
    {
      "epoch": 0.03464369054029797,
      "grad_norm": 27.208982467651367,
      "learning_rate": 0.0,
      "loss": 1.525,
      "step": 386
    },
    {
      "epoch": 0.034733441033925685,
      "grad_norm": 24.797483444213867,
      "learning_rate": 0.0,
      "loss": 1.6391,
      "step": 387
    },
    {
      "epoch": 0.0348231915275534,
      "grad_norm": 60.42938232421875,
      "learning_rate": 0.0,
      "loss": 1.5385,
      "step": 388
    },
    {
      "epoch": 0.034912942021181115,
      "grad_norm": 27.842405319213867,
      "learning_rate": 0.0,
      "loss": 1.5639,
      "step": 389
    },
    {
      "epoch": 0.03500269251480883,
      "grad_norm": 24.914331436157227,
      "learning_rate": 0.0,
      "loss": 1.6232,
      "step": 390
    },
    {
      "epoch": 0.035092443008436544,
      "grad_norm": 25.619596481323242,
      "learning_rate": 0.0,
      "loss": 1.5703,
      "step": 391
    },
    {
      "epoch": 0.03518219350206426,
      "grad_norm": 31.650463104248047,
      "learning_rate": 0.0,
      "loss": 1.794,
      "step": 392
    },
    {
      "epoch": 0.03527194399569197,
      "grad_norm": 24.623437881469727,
      "learning_rate": 0.0,
      "loss": 1.6386,
      "step": 393
    },
    {
      "epoch": 0.03536169448931969,
      "grad_norm": 24.08147621154785,
      "learning_rate": 0.0,
      "loss": 1.6179,
      "step": 394
    },
    {
      "epoch": 0.03545144498294741,
      "grad_norm": 27.81768226623535,
      "learning_rate": 0.0,
      "loss": 1.6481,
      "step": 395
    },
    {
      "epoch": 0.035541195476575124,
      "grad_norm": 24.931640625,
      "learning_rate": 0.0,
      "loss": 1.5817,
      "step": 396
    },
    {
      "epoch": 0.03563094597020284,
      "grad_norm": 21.290895462036133,
      "learning_rate": 0.0,
      "loss": 1.532,
      "step": 397
    },
    {
      "epoch": 0.035720696463830554,
      "grad_norm": 21.668094635009766,
      "learning_rate": 0.0,
      "loss": 1.6786,
      "step": 398
    },
    {
      "epoch": 0.03581044695745827,
      "grad_norm": 21.58544921875,
      "learning_rate": 0.0,
      "loss": 1.5062,
      "step": 399
    },
    {
      "epoch": 0.03590019745108598,
      "grad_norm": 24.441425323486328,
      "learning_rate": 0.0,
      "loss": 1.5723,
      "step": 400
    },
    {
      "epoch": 0.0359899479447137,
      "grad_norm": 27.13036346435547,
      "learning_rate": 0.0,
      "loss": 1.5934,
      "step": 401
    },
    {
      "epoch": 0.03607969843834141,
      "grad_norm": 25.879701614379883,
      "learning_rate": 0.0,
      "loss": 1.5412,
      "step": 402
    },
    {
      "epoch": 0.03616944893196913,
      "grad_norm": 27.007822036743164,
      "learning_rate": 0.0,
      "loss": 1.6752,
      "step": 403
    },
    {
      "epoch": 0.03625919942559684,
      "grad_norm": 23.36930274963379,
      "learning_rate": 0.0,
      "loss": 1.5974,
      "step": 404
    },
    {
      "epoch": 0.036348949919224556,
      "grad_norm": 24.40472412109375,
      "learning_rate": 0.0,
      "loss": 1.5482,
      "step": 405
    },
    {
      "epoch": 0.03643870041285227,
      "grad_norm": 25.661640167236328,
      "learning_rate": 0.0,
      "loss": 1.4868,
      "step": 406
    },
    {
      "epoch": 0.036528450906479985,
      "grad_norm": 24.316558837890625,
      "learning_rate": 0.0,
      "loss": 1.5434,
      "step": 407
    },
    {
      "epoch": 0.0366182014001077,
      "grad_norm": 27.80588722229004,
      "learning_rate": 0.0,
      "loss": 1.5433,
      "step": 408
    },
    {
      "epoch": 0.036707951893735415,
      "grad_norm": 67.76631164550781,
      "learning_rate": 0.0,
      "loss": 1.6565,
      "step": 409
    },
    {
      "epoch": 0.03679770238736313,
      "grad_norm": 31.515621185302734,
      "learning_rate": 0.0,
      "loss": 1.6408,
      "step": 410
    },
    {
      "epoch": 0.036887452880990844,
      "grad_norm": 23.472991943359375,
      "learning_rate": 0.0,
      "loss": 1.5335,
      "step": 411
    },
    {
      "epoch": 0.03697720337461856,
      "grad_norm": 24.109458923339844,
      "learning_rate": 0.0,
      "loss": 1.6122,
      "step": 412
    },
    {
      "epoch": 0.03706695386824627,
      "grad_norm": 23.99498748779297,
      "learning_rate": 0.0,
      "loss": 1.574,
      "step": 413
    },
    {
      "epoch": 0.03715670436187399,
      "grad_norm": 26.71710205078125,
      "learning_rate": 0.0,
      "loss": 1.5406,
      "step": 414
    },
    {
      "epoch": 0.0372464548555017,
      "grad_norm": 19.64081382751465,
      "learning_rate": 0.0,
      "loss": 1.5417,
      "step": 415
    },
    {
      "epoch": 0.03733620534912942,
      "grad_norm": 25.554500579833984,
      "learning_rate": 0.0,
      "loss": 1.5783,
      "step": 416
    },
    {
      "epoch": 0.03742595584275713,
      "grad_norm": 31.52701187133789,
      "learning_rate": 0.0,
      "loss": 1.5821,
      "step": 417
    },
    {
      "epoch": 0.03751570633638485,
      "grad_norm": 24.54152488708496,
      "learning_rate": 0.0,
      "loss": 1.6568,
      "step": 418
    },
    {
      "epoch": 0.03760545683001257,
      "grad_norm": 26.523780822753906,
      "learning_rate": 0.0,
      "loss": 1.6011,
      "step": 419
    },
    {
      "epoch": 0.03769520732364028,
      "grad_norm": 26.633161544799805,
      "learning_rate": 0.0,
      "loss": 1.6673,
      "step": 420
    },
    {
      "epoch": 0.037784957817268,
      "grad_norm": 30.206876754760742,
      "learning_rate": 0.0,
      "loss": 1.5616,
      "step": 421
    },
    {
      "epoch": 0.03787470831089571,
      "grad_norm": 29.922121047973633,
      "learning_rate": 0.0,
      "loss": 1.6448,
      "step": 422
    },
    {
      "epoch": 0.03796445880452343,
      "grad_norm": 23.138280868530273,
      "learning_rate": 0.0,
      "loss": 1.6109,
      "step": 423
    },
    {
      "epoch": 0.03805420929815114,
      "grad_norm": 26.025236129760742,
      "learning_rate": 0.0,
      "loss": 1.6016,
      "step": 424
    },
    {
      "epoch": 0.038143959791778856,
      "grad_norm": 27.89295196533203,
      "learning_rate": 0.0,
      "loss": 1.7487,
      "step": 425
    },
    {
      "epoch": 0.03823371028540657,
      "grad_norm": 22.79923439025879,
      "learning_rate": 0.0,
      "loss": 1.5873,
      "step": 426
    },
    {
      "epoch": 0.038323460779034285,
      "grad_norm": 45.7028694152832,
      "learning_rate": 0.0,
      "loss": 1.5676,
      "step": 427
    },
    {
      "epoch": 0.038413211272662,
      "grad_norm": 27.57731056213379,
      "learning_rate": 0.0,
      "loss": 1.6597,
      "step": 428
    },
    {
      "epoch": 0.038502961766289714,
      "grad_norm": 26.045190811157227,
      "learning_rate": 0.0,
      "loss": 1.7383,
      "step": 429
    },
    {
      "epoch": 0.03859271225991743,
      "grad_norm": 47.3829345703125,
      "learning_rate": 0.0,
      "loss": 1.6118,
      "step": 430
    },
    {
      "epoch": 0.038682462753545144,
      "grad_norm": 22.132427215576172,
      "learning_rate": 0.0,
      "loss": 1.6212,
      "step": 431
    },
    {
      "epoch": 0.03877221324717286,
      "grad_norm": 31.766321182250977,
      "learning_rate": 0.0,
      "loss": 1.7773,
      "step": 432
    },
    {
      "epoch": 0.03886196374080057,
      "grad_norm": 25.597797393798828,
      "learning_rate": 0.0,
      "loss": 1.7044,
      "step": 433
    },
    {
      "epoch": 0.03895171423442829,
      "grad_norm": 29.94695281982422,
      "learning_rate": 0.0,
      "loss": 1.6549,
      "step": 434
    },
    {
      "epoch": 0.039041464728056,
      "grad_norm": 22.95456886291504,
      "learning_rate": 0.0,
      "loss": 1.6213,
      "step": 435
    },
    {
      "epoch": 0.03913121522168372,
      "grad_norm": 27.0970458984375,
      "learning_rate": 0.0,
      "loss": 1.6519,
      "step": 436
    },
    {
      "epoch": 0.03922096571531143,
      "grad_norm": 25.56778335571289,
      "learning_rate": 0.0,
      "loss": 1.6399,
      "step": 437
    },
    {
      "epoch": 0.039310716208939146,
      "grad_norm": 27.69911003112793,
      "learning_rate": 0.0,
      "loss": 1.6612,
      "step": 438
    },
    {
      "epoch": 0.03940046670256686,
      "grad_norm": 29.995811462402344,
      "learning_rate": 0.0,
      "loss": 1.6198,
      "step": 439
    },
    {
      "epoch": 0.03949021719619458,
      "grad_norm": 24.612417221069336,
      "learning_rate": 0.0,
      "loss": 1.6124,
      "step": 440
    },
    {
      "epoch": 0.0395799676898223,
      "grad_norm": 22.709463119506836,
      "learning_rate": 0.0,
      "loss": 1.594,
      "step": 441
    },
    {
      "epoch": 0.03966971818345001,
      "grad_norm": 24.08415412902832,
      "learning_rate": 0.0,
      "loss": 1.5612,
      "step": 442
    },
    {
      "epoch": 0.039759468677077726,
      "grad_norm": 34.43753433227539,
      "learning_rate": 0.0,
      "loss": 1.6772,
      "step": 443
    },
    {
      "epoch": 0.03984921917070544,
      "grad_norm": 23.083345413208008,
      "learning_rate": 0.0,
      "loss": 1.5927,
      "step": 444
    },
    {
      "epoch": 0.039938969664333156,
      "grad_norm": 26.555131912231445,
      "learning_rate": 0.0,
      "loss": 1.6728,
      "step": 445
    },
    {
      "epoch": 0.04002872015796087,
      "grad_norm": 24.78805923461914,
      "learning_rate": 0.0,
      "loss": 1.6257,
      "step": 446
    },
    {
      "epoch": 0.040118470651588585,
      "grad_norm": 23.406076431274414,
      "learning_rate": 0.0,
      "loss": 1.5851,
      "step": 447
    },
    {
      "epoch": 0.0402082211452163,
      "grad_norm": 25.304908752441406,
      "learning_rate": 0.0,
      "loss": 1.727,
      "step": 448
    },
    {
      "epoch": 0.040297971638844014,
      "grad_norm": 21.90865707397461,
      "learning_rate": 0.0,
      "loss": 1.5577,
      "step": 449
    },
    {
      "epoch": 0.04038772213247173,
      "grad_norm": 27.377927780151367,
      "learning_rate": 0.0,
      "loss": 1.5923,
      "step": 450
    },
    {
      "epoch": 0.040477472626099444,
      "grad_norm": 22.06640625,
      "learning_rate": 0.0,
      "loss": 1.4775,
      "step": 451
    },
    {
      "epoch": 0.04056722311972716,
      "grad_norm": 22.04629898071289,
      "learning_rate": 0.0,
      "loss": 1.554,
      "step": 452
    },
    {
      "epoch": 0.04065697361335487,
      "grad_norm": 24.46075439453125,
      "learning_rate": 0.0,
      "loss": 1.5762,
      "step": 453
    },
    {
      "epoch": 0.04074672410698259,
      "grad_norm": 27.826444625854492,
      "learning_rate": 0.0,
      "loss": 1.6745,
      "step": 454
    },
    {
      "epoch": 0.0408364746006103,
      "grad_norm": 22.71664810180664,
      "learning_rate": 0.0,
      "loss": 1.7156,
      "step": 455
    },
    {
      "epoch": 0.04092622509423802,
      "grad_norm": 23.23731231689453,
      "learning_rate": 0.0,
      "loss": 1.6572,
      "step": 456
    },
    {
      "epoch": 0.04101597558786573,
      "grad_norm": 21.788658142089844,
      "learning_rate": 0.0,
      "loss": 1.6156,
      "step": 457
    },
    {
      "epoch": 0.041105726081493446,
      "grad_norm": 33.04486083984375,
      "learning_rate": 0.0,
      "loss": 1.6413,
      "step": 458
    },
    {
      "epoch": 0.04119547657512116,
      "grad_norm": 21.623489379882812,
      "learning_rate": 0.0,
      "loss": 1.6369,
      "step": 459
    },
    {
      "epoch": 0.041285227068748875,
      "grad_norm": 23.40774154663086,
      "learning_rate": 0.0,
      "loss": 1.5386,
      "step": 460
    },
    {
      "epoch": 0.04137497756237659,
      "grad_norm": 23.00798225402832,
      "learning_rate": 0.0,
      "loss": 1.5513,
      "step": 461
    },
    {
      "epoch": 0.041464728056004305,
      "grad_norm": 22.54790687561035,
      "learning_rate": 0.0,
      "loss": 1.6393,
      "step": 462
    },
    {
      "epoch": 0.041554478549632026,
      "grad_norm": 26.70391082763672,
      "learning_rate": 0.0,
      "loss": 1.5905,
      "step": 463
    },
    {
      "epoch": 0.04164422904325974,
      "grad_norm": 34.53096389770508,
      "learning_rate": 0.0,
      "loss": 1.667,
      "step": 464
    },
    {
      "epoch": 0.041733979536887456,
      "grad_norm": 29.551815032958984,
      "learning_rate": 0.0,
      "loss": 1.6028,
      "step": 465
    },
    {
      "epoch": 0.04182373003051517,
      "grad_norm": 25.20648765563965,
      "learning_rate": 0.0,
      "loss": 1.5213,
      "step": 466
    },
    {
      "epoch": 0.041913480524142885,
      "grad_norm": 27.24469566345215,
      "learning_rate": 0.0,
      "loss": 1.6696,
      "step": 467
    },
    {
      "epoch": 0.0420032310177706,
      "grad_norm": 46.18013000488281,
      "learning_rate": 0.0,
      "loss": 1.5923,
      "step": 468
    },
    {
      "epoch": 0.042092981511398314,
      "grad_norm": 27.608497619628906,
      "learning_rate": 0.0,
      "loss": 1.5803,
      "step": 469
    },
    {
      "epoch": 0.04218273200502603,
      "grad_norm": 21.77648162841797,
      "learning_rate": 0.0,
      "loss": 1.5468,
      "step": 470
    },
    {
      "epoch": 0.04227248249865374,
      "grad_norm": 26.540550231933594,
      "learning_rate": 0.0,
      "loss": 1.5386,
      "step": 471
    },
    {
      "epoch": 0.04236223299228146,
      "grad_norm": 134.0621337890625,
      "learning_rate": 0.0,
      "loss": 1.6238,
      "step": 472
    },
    {
      "epoch": 0.04245198348590917,
      "grad_norm": 30.857833862304688,
      "learning_rate": 0.0,
      "loss": 1.7864,
      "step": 473
    },
    {
      "epoch": 0.04254173397953689,
      "grad_norm": 20.282381057739258,
      "learning_rate": 0.0,
      "loss": 1.5435,
      "step": 474
    },
    {
      "epoch": 0.0426314844731646,
      "grad_norm": 65.5844955444336,
      "learning_rate": 0.0,
      "loss": 1.6207,
      "step": 475
    },
    {
      "epoch": 0.04272123496679232,
      "grad_norm": 21.210601806640625,
      "learning_rate": 0.0,
      "loss": 1.5272,
      "step": 476
    },
    {
      "epoch": 0.04281098546042003,
      "grad_norm": 19.504074096679688,
      "learning_rate": 0.0,
      "loss": 1.4755,
      "step": 477
    },
    {
      "epoch": 0.042900735954047746,
      "grad_norm": 29.61644172668457,
      "learning_rate": 0.0,
      "loss": 1.6803,
      "step": 478
    },
    {
      "epoch": 0.04299048644767546,
      "grad_norm": 24.086214065551758,
      "learning_rate": 0.0,
      "loss": 1.7724,
      "step": 479
    },
    {
      "epoch": 0.043080236941303175,
      "grad_norm": 25.336809158325195,
      "learning_rate": 0.0,
      "loss": 1.5607,
      "step": 480
    },
    {
      "epoch": 0.04316998743493089,
      "grad_norm": 25.05587387084961,
      "learning_rate": 0.0,
      "loss": 1.6162,
      "step": 481
    },
    {
      "epoch": 0.043259737928558605,
      "grad_norm": 21.807941436767578,
      "learning_rate": 0.0,
      "loss": 1.5663,
      "step": 482
    },
    {
      "epoch": 0.04334948842218632,
      "grad_norm": 22.81028175354004,
      "learning_rate": 0.0,
      "loss": 1.5216,
      "step": 483
    },
    {
      "epoch": 0.043439238915814034,
      "grad_norm": 23.255178451538086,
      "learning_rate": 0.0,
      "loss": 1.6603,
      "step": 484
    },
    {
      "epoch": 0.043528989409441755,
      "grad_norm": 24.487049102783203,
      "learning_rate": 0.0,
      "loss": 1.5395,
      "step": 485
    },
    {
      "epoch": 0.04361873990306947,
      "grad_norm": 26.129932403564453,
      "learning_rate": 0.0,
      "loss": 1.6422,
      "step": 486
    },
    {
      "epoch": 0.043708490396697185,
      "grad_norm": 23.978517532348633,
      "learning_rate": 0.0,
      "loss": 1.6231,
      "step": 487
    },
    {
      "epoch": 0.0437982408903249,
      "grad_norm": 27.978872299194336,
      "learning_rate": 0.0,
      "loss": 1.5966,
      "step": 488
    },
    {
      "epoch": 0.043887991383952614,
      "grad_norm": 27.805313110351562,
      "learning_rate": 0.0,
      "loss": 1.7036,
      "step": 489
    },
    {
      "epoch": 0.04397774187758033,
      "grad_norm": 22.483850479125977,
      "learning_rate": 0.0,
      "loss": 1.498,
      "step": 490
    },
    {
      "epoch": 0.04406749237120804,
      "grad_norm": 22.03746223449707,
      "learning_rate": 0.0,
      "loss": 1.5726,
      "step": 491
    },
    {
      "epoch": 0.04415724286483576,
      "grad_norm": 24.17377281188965,
      "learning_rate": 0.0,
      "loss": 1.6591,
      "step": 492
    },
    {
      "epoch": 0.04424699335846347,
      "grad_norm": 25.36363410949707,
      "learning_rate": 0.0,
      "loss": 1.6081,
      "step": 493
    },
    {
      "epoch": 0.04433674385209119,
      "grad_norm": 23.997949600219727,
      "learning_rate": 0.0,
      "loss": 1.5834,
      "step": 494
    },
    {
      "epoch": 0.0444264943457189,
      "grad_norm": 23.587141036987305,
      "learning_rate": 0.0,
      "loss": 1.5658,
      "step": 495
    },
    {
      "epoch": 0.044516244839346616,
      "grad_norm": 23.249549865722656,
      "learning_rate": 0.0,
      "loss": 1.5466,
      "step": 496
    },
    {
      "epoch": 0.04460599533297433,
      "grad_norm": 27.516902923583984,
      "learning_rate": 0.0,
      "loss": 1.7587,
      "step": 497
    },
    {
      "epoch": 0.044695745826602046,
      "grad_norm": 27.991125106811523,
      "learning_rate": 0.0,
      "loss": 1.6351,
      "step": 498
    },
    {
      "epoch": 0.04478549632022976,
      "grad_norm": 22.307523727416992,
      "learning_rate": 0.0,
      "loss": 1.5497,
      "step": 499
    },
    {
      "epoch": 0.044875246813857475,
      "grad_norm": 32.228397369384766,
      "learning_rate": 0.0,
      "loss": 1.5584,
      "step": 500
    },
    {
      "epoch": 0.04496499730748519,
      "grad_norm": 28.38634490966797,
      "learning_rate": 0.0,
      "loss": 1.6304,
      "step": 501
    },
    {
      "epoch": 0.045054747801112904,
      "grad_norm": 12.462540626525879,
      "learning_rate": 0.0,
      "loss": 1.1968,
      "step": 502
    },
    {
      "epoch": 0.04514449829474062,
      "grad_norm": 8.557329177856445,
      "learning_rate": 0.0,
      "loss": 1.3272,
      "step": 503
    },
    {
      "epoch": 0.045234248788368334,
      "grad_norm": 7.487260818481445,
      "learning_rate": 0.0,
      "loss": 1.258,
      "step": 504
    },
    {
      "epoch": 0.04532399928199605,
      "grad_norm": 8.043220520019531,
      "learning_rate": 0.0,
      "loss": 1.1687,
      "step": 505
    },
    {
      "epoch": 0.04541374977562376,
      "grad_norm": 7.339669704437256,
      "learning_rate": 0.0,
      "loss": 1.1619,
      "step": 506
    },
    {
      "epoch": 0.04550350026925148,
      "grad_norm": 6.700636863708496,
      "learning_rate": 0.0,
      "loss": 1.246,
      "step": 507
    },
    {
      "epoch": 0.0455932507628792,
      "grad_norm": 8.79631519317627,
      "learning_rate": 0.0,
      "loss": 1.24,
      "step": 508
    },
    {
      "epoch": 0.045683001256506914,
      "grad_norm": 7.4711012840271,
      "learning_rate": 0.0,
      "loss": 1.2296,
      "step": 509
    },
    {
      "epoch": 0.04577275175013463,
      "grad_norm": 13.172078132629395,
      "learning_rate": 0.0,
      "loss": 1.2168,
      "step": 510
    },
    {
      "epoch": 0.04586250224376234,
      "grad_norm": 7.665863037109375,
      "learning_rate": 0.0,
      "loss": 1.1841,
      "step": 511
    },
    {
      "epoch": 0.04595225273739006,
      "grad_norm": 8.955485343933105,
      "learning_rate": 0.0,
      "loss": 1.2759,
      "step": 512
    },
    {
      "epoch": 0.04604200323101777,
      "grad_norm": 8.225412368774414,
      "learning_rate": 0.0,
      "loss": 1.2236,
      "step": 513
    },
    {
      "epoch": 0.04613175372464549,
      "grad_norm": 6.404390811920166,
      "learning_rate": 0.0,
      "loss": 1.2652,
      "step": 514
    },
    {
      "epoch": 0.0462215042182732,
      "grad_norm": 7.02842378616333,
      "learning_rate": 0.0,
      "loss": 1.103,
      "step": 515
    },
    {
      "epoch": 0.046311254711900916,
      "grad_norm": 9.250580787658691,
      "learning_rate": 0.0,
      "loss": 1.1944,
      "step": 516
    },
    {
      "epoch": 0.04640100520552863,
      "grad_norm": 7.9946465492248535,
      "learning_rate": 0.0,
      "loss": 1.2196,
      "step": 517
    },
    {
      "epoch": 0.046490755699156346,
      "grad_norm": 7.921901226043701,
      "learning_rate": 0.0,
      "loss": 1.2494,
      "step": 518
    },
    {
      "epoch": 0.04658050619278406,
      "grad_norm": 7.786976337432861,
      "learning_rate": 0.0,
      "loss": 1.2247,
      "step": 519
    },
    {
      "epoch": 0.046670256686411775,
      "grad_norm": 7.181194305419922,
      "learning_rate": 0.0,
      "loss": 1.2484,
      "step": 520
    },
    {
      "epoch": 0.04676000718003949,
      "grad_norm": 17.27275848388672,
      "learning_rate": 0.0,
      "loss": 1.2834,
      "step": 521
    },
    {
      "epoch": 0.046849757673667204,
      "grad_norm": 7.811083793640137,
      "learning_rate": 0.0,
      "loss": 1.2928,
      "step": 522
    },
    {
      "epoch": 0.04693950816729492,
      "grad_norm": 7.165767669677734,
      "learning_rate": 0.0,
      "loss": 1.2034,
      "step": 523
    },
    {
      "epoch": 0.04702925866092263,
      "grad_norm": 7.449375629425049,
      "learning_rate": 0.0,
      "loss": 1.2531,
      "step": 524
    },
    {
      "epoch": 0.04711900915455035,
      "grad_norm": 7.4443535804748535,
      "learning_rate": 0.0,
      "loss": 1.1247,
      "step": 525
    },
    {
      "epoch": 0.04720875964817806,
      "grad_norm": 7.05775785446167,
      "learning_rate": 0.0,
      "loss": 1.264,
      "step": 526
    },
    {
      "epoch": 0.04729851014180578,
      "grad_norm": 8.000849723815918,
      "learning_rate": 0.0,
      "loss": 1.2453,
      "step": 527
    },
    {
      "epoch": 0.04738826063543349,
      "grad_norm": 8.304893493652344,
      "learning_rate": 0.0,
      "loss": 1.1757,
      "step": 528
    },
    {
      "epoch": 0.04747801112906121,
      "grad_norm": 8.254316329956055,
      "learning_rate": 0.0,
      "loss": 1.1572,
      "step": 529
    },
    {
      "epoch": 0.04756776162268892,
      "grad_norm": 6.874651908874512,
      "learning_rate": 0.0,
      "loss": 1.2315,
      "step": 530
    },
    {
      "epoch": 0.04765751211631664,
      "grad_norm": 9.247281074523926,
      "learning_rate": 0.0,
      "loss": 1.3573,
      "step": 531
    },
    {
      "epoch": 0.04774726260994436,
      "grad_norm": 6.6874775886535645,
      "learning_rate": 0.0,
      "loss": 1.2564,
      "step": 532
    },
    {
      "epoch": 0.04783701310357207,
      "grad_norm": 7.556613445281982,
      "learning_rate": 0.0,
      "loss": 1.3433,
      "step": 533
    },
    {
      "epoch": 0.04792676359719979,
      "grad_norm": 6.175554275512695,
      "learning_rate": 0.0,
      "loss": 1.1903,
      "step": 534
    },
    {
      "epoch": 0.0480165140908275,
      "grad_norm": 38.7926139831543,
      "learning_rate": 0.0,
      "loss": 1.1899,
      "step": 535
    },
    {
      "epoch": 0.048106264584455216,
      "grad_norm": 8.043590545654297,
      "learning_rate": 0.0,
      "loss": 1.329,
      "step": 536
    },
    {
      "epoch": 0.04819601507808293,
      "grad_norm": 8.044459342956543,
      "learning_rate": 0.0,
      "loss": 1.2454,
      "step": 537
    },
    {
      "epoch": 0.048285765571710645,
      "grad_norm": 8.423386573791504,
      "learning_rate": 0.0,
      "loss": 1.1461,
      "step": 538
    },
    {
      "epoch": 0.04837551606533836,
      "grad_norm": 9.978515625,
      "learning_rate": 0.0,
      "loss": 1.1595,
      "step": 539
    },
    {
      "epoch": 0.048465266558966075,
      "grad_norm": 8.674606323242188,
      "learning_rate": 0.0,
      "loss": 1.1628,
      "step": 540
    },
    {
      "epoch": 0.04855501705259379,
      "grad_norm": 7.62788724899292,
      "learning_rate": 0.0,
      "loss": 1.2278,
      "step": 541
    },
    {
      "epoch": 0.048644767546221504,
      "grad_norm": 8.474723815917969,
      "learning_rate": 0.0,
      "loss": 1.1553,
      "step": 542
    },
    {
      "epoch": 0.04873451803984922,
      "grad_norm": 8.686922073364258,
      "learning_rate": 0.0,
      "loss": 1.175,
      "step": 543
    },
    {
      "epoch": 0.04882426853347693,
      "grad_norm": 14.65805435180664,
      "learning_rate": 0.0,
      "loss": 1.1683,
      "step": 544
    },
    {
      "epoch": 0.04891401902710465,
      "grad_norm": 9.46110725402832,
      "learning_rate": 0.0,
      "loss": 1.3294,
      "step": 545
    },
    {
      "epoch": 0.04900376952073236,
      "grad_norm": 7.26961088180542,
      "learning_rate": 0.0,
      "loss": 1.1923,
      "step": 546
    },
    {
      "epoch": 0.04909352001436008,
      "grad_norm": 9.631993293762207,
      "learning_rate": 0.0,
      "loss": 1.16,
      "step": 547
    },
    {
      "epoch": 0.04918327050798779,
      "grad_norm": 31.307453155517578,
      "learning_rate": 0.0,
      "loss": 1.4242,
      "step": 548
    },
    {
      "epoch": 0.049273021001615507,
      "grad_norm": 10.786386489868164,
      "learning_rate": 0.0,
      "loss": 1.2764,
      "step": 549
    },
    {
      "epoch": 0.04936277149524322,
      "grad_norm": 16.542280197143555,
      "learning_rate": 0.0,
      "loss": 1.2368,
      "step": 550
    },
    {
      "epoch": 0.049452521988870936,
      "grad_norm": 7.045132637023926,
      "learning_rate": 0.0,
      "loss": 1.1694,
      "step": 551
    },
    {
      "epoch": 0.04954227248249865,
      "grad_norm": 7.616021633148193,
      "learning_rate": 0.0,
      "loss": 1.2293,
      "step": 552
    },
    {
      "epoch": 0.04963202297612637,
      "grad_norm": 7.384084224700928,
      "learning_rate": 0.0,
      "loss": 1.2079,
      "step": 553
    },
    {
      "epoch": 0.04972177346975409,
      "grad_norm": 7.186660289764404,
      "learning_rate": 0.0,
      "loss": 1.2036,
      "step": 554
    },
    {
      "epoch": 0.0498115239633818,
      "grad_norm": 6.416055679321289,
      "learning_rate": 0.0,
      "loss": 1.2323,
      "step": 555
    },
    {
      "epoch": 0.049901274457009516,
      "grad_norm": 7.121069431304932,
      "learning_rate": 0.0,
      "loss": 1.2214,
      "step": 556
    },
    {
      "epoch": 0.04999102495063723,
      "grad_norm": 7.030943393707275,
      "learning_rate": 0.0,
      "loss": 1.234,
      "step": 557
    },
    {
      "epoch": 0.050080775444264945,
      "grad_norm": 9.27779769897461,
      "learning_rate": 0.0,
      "loss": 1.1693,
      "step": 558
    },
    {
      "epoch": 0.05017052593789266,
      "grad_norm": 7.307068824768066,
      "learning_rate": 0.0,
      "loss": 1.2245,
      "step": 559
    },
    {
      "epoch": 0.050260276431520375,
      "grad_norm": 8.212464332580566,
      "learning_rate": 0.0,
      "loss": 1.1824,
      "step": 560
    },
    {
      "epoch": 0.05035002692514809,
      "grad_norm": 7.624232769012451,
      "learning_rate": 0.0,
      "loss": 1.2141,
      "step": 561
    },
    {
      "epoch": 0.050439777418775804,
      "grad_norm": 7.447394371032715,
      "learning_rate": 0.0,
      "loss": 1.2278,
      "step": 562
    },
    {
      "epoch": 0.05052952791240352,
      "grad_norm": 6.880248069763184,
      "learning_rate": 0.0,
      "loss": 1.1558,
      "step": 563
    },
    {
      "epoch": 0.05061927840603123,
      "grad_norm": 7.166173934936523,
      "learning_rate": 0.0,
      "loss": 1.2685,
      "step": 564
    },
    {
      "epoch": 0.05070902889965895,
      "grad_norm": 7.550762176513672,
      "learning_rate": 0.0,
      "loss": 1.2318,
      "step": 565
    },
    {
      "epoch": 0.05079877939328666,
      "grad_norm": 8.093884468078613,
      "learning_rate": 0.0,
      "loss": 1.2626,
      "step": 566
    },
    {
      "epoch": 0.05088852988691438,
      "grad_norm": 22.270755767822266,
      "learning_rate": 0.0,
      "loss": 1.2216,
      "step": 567
    },
    {
      "epoch": 0.05097828038054209,
      "grad_norm": 6.761216163635254,
      "learning_rate": 0.0,
      "loss": 1.2182,
      "step": 568
    },
    {
      "epoch": 0.051068030874169806,
      "grad_norm": 12.393269538879395,
      "learning_rate": 0.0,
      "loss": 1.2553,
      "step": 569
    },
    {
      "epoch": 0.05115778136779752,
      "grad_norm": 9.736340522766113,
      "learning_rate": 0.0,
      "loss": 1.2913,
      "step": 570
    },
    {
      "epoch": 0.051247531861425236,
      "grad_norm": 14.820860862731934,
      "learning_rate": 0.0,
      "loss": 1.262,
      "step": 571
    },
    {
      "epoch": 0.05133728235505295,
      "grad_norm": 8.833534240722656,
      "learning_rate": 0.0,
      "loss": 1.2533,
      "step": 572
    },
    {
      "epoch": 0.051427032848680665,
      "grad_norm": 8.18061351776123,
      "learning_rate": 0.0,
      "loss": 1.2069,
      "step": 573
    },
    {
      "epoch": 0.05151678334230838,
      "grad_norm": 9.255142211914062,
      "learning_rate": 0.0,
      "loss": 1.1742,
      "step": 574
    },
    {
      "epoch": 0.051606533835936094,
      "grad_norm": 8.671813011169434,
      "learning_rate": 0.0,
      "loss": 1.2381,
      "step": 575
    },
    {
      "epoch": 0.051696284329563816,
      "grad_norm": 6.6900482177734375,
      "learning_rate": 0.0,
      "loss": 1.2145,
      "step": 576
    },
    {
      "epoch": 0.05178603482319153,
      "grad_norm": 6.686525344848633,
      "learning_rate": 0.0,
      "loss": 1.2297,
      "step": 577
    },
    {
      "epoch": 0.051875785316819245,
      "grad_norm": 7.588764190673828,
      "learning_rate": 0.0,
      "loss": 1.2335,
      "step": 578
    },
    {
      "epoch": 0.05196553581044696,
      "grad_norm": 7.254631996154785,
      "learning_rate": 0.0,
      "loss": 1.2821,
      "step": 579
    },
    {
      "epoch": 0.052055286304074674,
      "grad_norm": 8.32887077331543,
      "learning_rate": 0.0,
      "loss": 1.2399,
      "step": 580
    },
    {
      "epoch": 0.05214503679770239,
      "grad_norm": 12.360590934753418,
      "learning_rate": 0.0,
      "loss": 1.3235,
      "step": 581
    },
    {
      "epoch": 0.052234787291330104,
      "grad_norm": 19.48790740966797,
      "learning_rate": 0.0,
      "loss": 1.3614,
      "step": 582
    },
    {
      "epoch": 0.05232453778495782,
      "grad_norm": 6.700620174407959,
      "learning_rate": 0.0,
      "loss": 1.2162,
      "step": 583
    },
    {
      "epoch": 0.05241428827858553,
      "grad_norm": 10.199204444885254,
      "learning_rate": 0.0,
      "loss": 1.2148,
      "step": 584
    },
    {
      "epoch": 0.05250403877221325,
      "grad_norm": 5.965697765350342,
      "learning_rate": 0.0,
      "loss": 1.1401,
      "step": 585
    },
    {
      "epoch": 0.05259378926584096,
      "grad_norm": 7.305534362792969,
      "learning_rate": 0.0,
      "loss": 1.1347,
      "step": 586
    },
    {
      "epoch": 0.05268353975946868,
      "grad_norm": 32.069984436035156,
      "learning_rate": 0.0,
      "loss": 1.2261,
      "step": 587
    },
    {
      "epoch": 0.05277329025309639,
      "grad_norm": 7.393807411193848,
      "learning_rate": 0.0,
      "loss": 1.2062,
      "step": 588
    },
    {
      "epoch": 0.052863040746724106,
      "grad_norm": 8.67479133605957,
      "learning_rate": 0.0,
      "loss": 1.1958,
      "step": 589
    },
    {
      "epoch": 0.05295279124035182,
      "grad_norm": 7.9644670486450195,
      "learning_rate": 0.0,
      "loss": 1.2807,
      "step": 590
    },
    {
      "epoch": 0.053042541733979535,
      "grad_norm": 7.029698848724365,
      "learning_rate": 0.0,
      "loss": 1.2098,
      "step": 591
    },
    {
      "epoch": 0.05313229222760725,
      "grad_norm": 14.976594924926758,
      "learning_rate": 0.0,
      "loss": 1.1867,
      "step": 592
    },
    {
      "epoch": 0.053222042721234965,
      "grad_norm": 9.729155540466309,
      "learning_rate": 0.0,
      "loss": 1.3291,
      "step": 593
    },
    {
      "epoch": 0.05331179321486268,
      "grad_norm": 9.147235870361328,
      "learning_rate": 0.0,
      "loss": 1.2415,
      "step": 594
    },
    {
      "epoch": 0.053401543708490394,
      "grad_norm": 6.881488800048828,
      "learning_rate": 0.0,
      "loss": 1.218,
      "step": 595
    },
    {
      "epoch": 0.05349129420211811,
      "grad_norm": 75.4875717163086,
      "learning_rate": 0.0,
      "loss": 1.197,
      "step": 596
    },
    {
      "epoch": 0.05358104469574582,
      "grad_norm": 17.253501892089844,
      "learning_rate": 0.0,
      "loss": 1.1288,
      "step": 597
    },
    {
      "epoch": 0.053670795189373545,
      "grad_norm": 6.951292037963867,
      "learning_rate": 0.0,
      "loss": 1.1769,
      "step": 598
    },
    {
      "epoch": 0.05376054568300126,
      "grad_norm": 8.280523300170898,
      "learning_rate": 0.0,
      "loss": 1.3221,
      "step": 599
    },
    {
      "epoch": 0.053850296176628974,
      "grad_norm": 9.713400840759277,
      "learning_rate": 0.0,
      "loss": 1.1963,
      "step": 600
    },
    {
      "epoch": 0.05394004667025669,
      "grad_norm": 20.052734375,
      "learning_rate": 0.0,
      "loss": 1.172,
      "step": 601
    },
    {
      "epoch": 0.054029797163884404,
      "grad_norm": 6.893290996551514,
      "learning_rate": 0.0,
      "loss": 1.1635,
      "step": 602
    },
    {
      "epoch": 0.05411954765751212,
      "grad_norm": 16.207963943481445,
      "learning_rate": 0.0,
      "loss": 1.2865,
      "step": 603
    },
    {
      "epoch": 0.05420929815113983,
      "grad_norm": 7.283064365386963,
      "learning_rate": 0.0,
      "loss": 1.214,
      "step": 604
    },
    {
      "epoch": 0.05429904864476755,
      "grad_norm": 13.490817070007324,
      "learning_rate": 0.0,
      "loss": 1.2751,
      "step": 605
    },
    {
      "epoch": 0.05438879913839526,
      "grad_norm": 6.900382041931152,
      "learning_rate": 0.0,
      "loss": 1.2526,
      "step": 606
    },
    {
      "epoch": 0.05447854963202298,
      "grad_norm": 9.643807411193848,
      "learning_rate": 0.0,
      "loss": 1.1949,
      "step": 607
    },
    {
      "epoch": 0.05456830012565069,
      "grad_norm": 6.906343936920166,
      "learning_rate": 0.0,
      "loss": 1.1849,
      "step": 608
    },
    {
      "epoch": 0.054658050619278406,
      "grad_norm": 8.88301944732666,
      "learning_rate": 0.0,
      "loss": 1.2777,
      "step": 609
    },
    {
      "epoch": 0.05474780111290612,
      "grad_norm": 7.51396369934082,
      "learning_rate": 0.0,
      "loss": 1.2631,
      "step": 610
    },
    {
      "epoch": 0.054837551606533835,
      "grad_norm": 11.990108489990234,
      "learning_rate": 0.0,
      "loss": 1.2343,
      "step": 611
    },
    {
      "epoch": 0.05492730210016155,
      "grad_norm": 8.776200294494629,
      "learning_rate": 0.0,
      "loss": 1.2654,
      "step": 612
    },
    {
      "epoch": 0.055017052593789265,
      "grad_norm": 7.022143363952637,
      "learning_rate": 0.0,
      "loss": 1.2332,
      "step": 613
    },
    {
      "epoch": 0.05510680308741698,
      "grad_norm": 12.353822708129883,
      "learning_rate": 0.0,
      "loss": 1.2427,
      "step": 614
    },
    {
      "epoch": 0.055196553581044694,
      "grad_norm": 8.070234298706055,
      "learning_rate": 0.0,
      "loss": 1.1562,
      "step": 615
    },
    {
      "epoch": 0.05528630407467241,
      "grad_norm": 10.936837196350098,
      "learning_rate": 0.0,
      "loss": 1.2065,
      "step": 616
    },
    {
      "epoch": 0.05537605456830012,
      "grad_norm": 7.178462028503418,
      "learning_rate": 0.0,
      "loss": 1.162,
      "step": 617
    },
    {
      "epoch": 0.05546580506192784,
      "grad_norm": 6.843204975128174,
      "learning_rate": 0.0,
      "loss": 1.1471,
      "step": 618
    },
    {
      "epoch": 0.05555555555555555,
      "grad_norm": 8.265068054199219,
      "learning_rate": 0.0,
      "loss": 1.2182,
      "step": 619
    },
    {
      "epoch": 0.05564530604918327,
      "grad_norm": 7.270506858825684,
      "learning_rate": 0.0,
      "loss": 1.3063,
      "step": 620
    },
    {
      "epoch": 0.05573505654281099,
      "grad_norm": 8.815617561340332,
      "learning_rate": 0.0,
      "loss": 1.2305,
      "step": 621
    },
    {
      "epoch": 0.0558248070364387,
      "grad_norm": 11.062637329101562,
      "learning_rate": 0.0,
      "loss": 1.2016,
      "step": 622
    },
    {
      "epoch": 0.05591455753006642,
      "grad_norm": 6.960347652435303,
      "learning_rate": 0.0,
      "loss": 1.2072,
      "step": 623
    },
    {
      "epoch": 0.05600430802369413,
      "grad_norm": 8.873368263244629,
      "learning_rate": 0.0,
      "loss": 1.3113,
      "step": 624
    },
    {
      "epoch": 0.05609405851732185,
      "grad_norm": 7.329237937927246,
      "learning_rate": 0.0,
      "loss": 1.1942,
      "step": 625
    },
    {
      "epoch": 0.05618380901094956,
      "grad_norm": 20.20212745666504,
      "learning_rate": 0.0,
      "loss": 1.191,
      "step": 626
    },
    {
      "epoch": 0.05627355950457728,
      "grad_norm": 6.955742359161377,
      "learning_rate": 0.0,
      "loss": 1.2068,
      "step": 627
    },
    {
      "epoch": 0.05636330999820499,
      "grad_norm": 9.083464622497559,
      "learning_rate": 0.0,
      "loss": 1.3062,
      "step": 628
    },
    {
      "epoch": 0.056453060491832706,
      "grad_norm": 6.877889633178711,
      "learning_rate": 0.0,
      "loss": 1.1886,
      "step": 629
    },
    {
      "epoch": 0.05654281098546042,
      "grad_norm": 7.675149917602539,
      "learning_rate": 0.0,
      "loss": 1.2335,
      "step": 630
    },
    {
      "epoch": 0.056632561479088135,
      "grad_norm": 9.677528381347656,
      "learning_rate": 0.0,
      "loss": 1.2463,
      "step": 631
    },
    {
      "epoch": 0.05672231197271585,
      "grad_norm": 6.8806352615356445,
      "learning_rate": 0.0,
      "loss": 1.1518,
      "step": 632
    },
    {
      "epoch": 0.056812062466343564,
      "grad_norm": 7.337514400482178,
      "learning_rate": 0.0,
      "loss": 1.2217,
      "step": 633
    },
    {
      "epoch": 0.05690181295997128,
      "grad_norm": 7.778676509857178,
      "learning_rate": 0.0,
      "loss": 1.1822,
      "step": 634
    },
    {
      "epoch": 0.056991563453598994,
      "grad_norm": 7.280935764312744,
      "learning_rate": 0.0,
      "loss": 1.1922,
      "step": 635
    },
    {
      "epoch": 0.05708131394722671,
      "grad_norm": 6.542960166931152,
      "learning_rate": 0.0,
      "loss": 1.2014,
      "step": 636
    },
    {
      "epoch": 0.05717106444085442,
      "grad_norm": 8.952205657958984,
      "learning_rate": 0.0,
      "loss": 1.2482,
      "step": 637
    },
    {
      "epoch": 0.05726081493448214,
      "grad_norm": 7.9217729568481445,
      "learning_rate": 0.0,
      "loss": 1.2072,
      "step": 638
    },
    {
      "epoch": 0.05735056542810985,
      "grad_norm": 8.59414005279541,
      "learning_rate": 0.0,
      "loss": 1.2476,
      "step": 639
    },
    {
      "epoch": 0.05744031592173757,
      "grad_norm": 7.807615280151367,
      "learning_rate": 0.0,
      "loss": 1.2368,
      "step": 640
    },
    {
      "epoch": 0.05753006641536528,
      "grad_norm": 30.15920066833496,
      "learning_rate": 0.0,
      "loss": 1.2331,
      "step": 641
    },
    {
      "epoch": 0.057619816908992996,
      "grad_norm": 9.299492835998535,
      "learning_rate": 0.0,
      "loss": 1.1741,
      "step": 642
    },
    {
      "epoch": 0.05770956740262072,
      "grad_norm": 6.654836177825928,
      "learning_rate": 0.0,
      "loss": 1.1941,
      "step": 643
    },
    {
      "epoch": 0.05779931789624843,
      "grad_norm": 6.687337875366211,
      "learning_rate": 0.0,
      "loss": 1.2062,
      "step": 644
    },
    {
      "epoch": 0.05788906838987615,
      "grad_norm": 8.019832611083984,
      "learning_rate": 0.0,
      "loss": 1.2086,
      "step": 645
    },
    {
      "epoch": 0.05797881888350386,
      "grad_norm": 9.348161697387695,
      "learning_rate": 0.0,
      "loss": 1.232,
      "step": 646
    },
    {
      "epoch": 0.058068569377131576,
      "grad_norm": 8.219810485839844,
      "learning_rate": 0.0,
      "loss": 1.1854,
      "step": 647
    },
    {
      "epoch": 0.05815831987075929,
      "grad_norm": 24.81812286376953,
      "learning_rate": 0.0,
      "loss": 1.2593,
      "step": 648
    },
    {
      "epoch": 0.058248070364387006,
      "grad_norm": 6.953103542327881,
      "learning_rate": 0.0,
      "loss": 1.1526,
      "step": 649
    },
    {
      "epoch": 0.05833782085801472,
      "grad_norm": 6.597920894622803,
      "learning_rate": 0.0,
      "loss": 1.2262,
      "step": 650
    },
    {
      "epoch": 0.058427571351642435,
      "grad_norm": 19.543954849243164,
      "learning_rate": 0.0,
      "loss": 1.2433,
      "step": 651
    },
    {
      "epoch": 0.05851732184527015,
      "grad_norm": 9.918458938598633,
      "learning_rate": 0.0,
      "loss": 1.2157,
      "step": 652
    },
    {
      "epoch": 0.058607072338897864,
      "grad_norm": 6.157138347625732,
      "learning_rate": 0.0,
      "loss": 1.1979,
      "step": 653
    },
    {
      "epoch": 0.05869682283252558,
      "grad_norm": 8.914946556091309,
      "learning_rate": 0.0,
      "loss": 1.2015,
      "step": 654
    },
    {
      "epoch": 0.058786573326153294,
      "grad_norm": 14.045143127441406,
      "learning_rate": 0.0,
      "loss": 1.1881,
      "step": 655
    },
    {
      "epoch": 0.05887632381978101,
      "grad_norm": 7.792648792266846,
      "learning_rate": 0.0,
      "loss": 1.1935,
      "step": 656
    },
    {
      "epoch": 0.05896607431340872,
      "grad_norm": 7.095909595489502,
      "learning_rate": 0.0,
      "loss": 1.2025,
      "step": 657
    },
    {
      "epoch": 0.05905582480703644,
      "grad_norm": 8.077377319335938,
      "learning_rate": 0.0,
      "loss": 1.1505,
      "step": 658
    },
    {
      "epoch": 0.05914557530066415,
      "grad_norm": 7.325658321380615,
      "learning_rate": 0.0,
      "loss": 1.2144,
      "step": 659
    },
    {
      "epoch": 0.05923532579429187,
      "grad_norm": 12.802783966064453,
      "learning_rate": 0.0,
      "loss": 1.2913,
      "step": 660
    },
    {
      "epoch": 0.05932507628791958,
      "grad_norm": 9.203788757324219,
      "learning_rate": 0.0,
      "loss": 1.2665,
      "step": 661
    },
    {
      "epoch": 0.059414826781547296,
      "grad_norm": 7.43321418762207,
      "learning_rate": 0.0,
      "loss": 1.2484,
      "step": 662
    },
    {
      "epoch": 0.05950457727517501,
      "grad_norm": 7.608431339263916,
      "learning_rate": 0.0,
      "loss": 1.1957,
      "step": 663
    },
    {
      "epoch": 0.059594327768802725,
      "grad_norm": 8.455257415771484,
      "learning_rate": 0.0,
      "loss": 1.2208,
      "step": 664
    },
    {
      "epoch": 0.05968407826243044,
      "grad_norm": 6.9590535163879395,
      "learning_rate": 0.0,
      "loss": 1.1836,
      "step": 665
    },
    {
      "epoch": 0.05977382875605816,
      "grad_norm": 10.64202880859375,
      "learning_rate": 0.0,
      "loss": 1.2432,
      "step": 666
    },
    {
      "epoch": 0.059863579249685876,
      "grad_norm": 7.102421760559082,
      "learning_rate": 0.0,
      "loss": 1.235,
      "step": 667
    },
    {
      "epoch": 0.05995332974331359,
      "grad_norm": 31.316539764404297,
      "learning_rate": 0.0,
      "loss": 1.2634,
      "step": 668
    },
    {
      "epoch": 0.060043080236941306,
      "grad_norm": 8.165168762207031,
      "learning_rate": 0.0,
      "loss": 1.1974,
      "step": 669
    },
    {
      "epoch": 0.06013283073056902,
      "grad_norm": 9.815034866333008,
      "learning_rate": 0.0,
      "loss": 1.2369,
      "step": 670
    },
    {
      "epoch": 0.060222581224196735,
      "grad_norm": 7.9807353019714355,
      "learning_rate": 0.0,
      "loss": 1.2778,
      "step": 671
    },
    {
      "epoch": 0.06031233171782445,
      "grad_norm": 9.183271408081055,
      "learning_rate": 0.0,
      "loss": 1.1847,
      "step": 672
    },
    {
      "epoch": 0.060402082211452164,
      "grad_norm": 8.369861602783203,
      "learning_rate": 0.0,
      "loss": 1.2805,
      "step": 673
    },
    {
      "epoch": 0.06049183270507988,
      "grad_norm": 14.489574432373047,
      "learning_rate": 0.0,
      "loss": 1.1713,
      "step": 674
    },
    {
      "epoch": 0.06058158319870759,
      "grad_norm": 8.623817443847656,
      "learning_rate": 0.0,
      "loss": 1.2475,
      "step": 675
    },
    {
      "epoch": 0.06067133369233531,
      "grad_norm": 6.541132926940918,
      "learning_rate": 0.0,
      "loss": 1.1757,
      "step": 676
    },
    {
      "epoch": 0.06076108418596302,
      "grad_norm": 9.851337432861328,
      "learning_rate": 0.0,
      "loss": 1.183,
      "step": 677
    },
    {
      "epoch": 0.06085083467959074,
      "grad_norm": 8.141185760498047,
      "learning_rate": 0.0,
      "loss": 1.2971,
      "step": 678
    },
    {
      "epoch": 0.06094058517321845,
      "grad_norm": 6.3403120040893555,
      "learning_rate": 0.0,
      "loss": 1.1375,
      "step": 679
    },
    {
      "epoch": 0.06103033566684617,
      "grad_norm": 11.845658302307129,
      "learning_rate": 0.0,
      "loss": 1.2006,
      "step": 680
    },
    {
      "epoch": 0.06112008616047388,
      "grad_norm": 8.099393844604492,
      "learning_rate": 0.0,
      "loss": 1.1793,
      "step": 681
    },
    {
      "epoch": 0.061209836654101596,
      "grad_norm": 8.642646789550781,
      "learning_rate": 0.0,
      "loss": 1.2629,
      "step": 682
    },
    {
      "epoch": 0.06129958714772931,
      "grad_norm": 7.250520706176758,
      "learning_rate": 0.0,
      "loss": 1.2954,
      "step": 683
    },
    {
      "epoch": 0.061389337641357025,
      "grad_norm": 17.114681243896484,
      "learning_rate": 0.0,
      "loss": 1.2471,
      "step": 684
    },
    {
      "epoch": 0.06147908813498474,
      "grad_norm": 6.20037317276001,
      "learning_rate": 0.0,
      "loss": 1.1761,
      "step": 685
    },
    {
      "epoch": 0.061568838628612454,
      "grad_norm": 8.657028198242188,
      "learning_rate": 0.0,
      "loss": 1.3193,
      "step": 686
    },
    {
      "epoch": 0.06165858912224017,
      "grad_norm": 11.428947448730469,
      "learning_rate": 0.0,
      "loss": 1.1263,
      "step": 687
    },
    {
      "epoch": 0.06174833961586789,
      "grad_norm": 7.955925941467285,
      "learning_rate": 0.0,
      "loss": 1.2406,
      "step": 688
    },
    {
      "epoch": 0.061838090109495605,
      "grad_norm": 8.518975257873535,
      "learning_rate": 0.0,
      "loss": 1.2952,
      "step": 689
    },
    {
      "epoch": 0.06192784060312332,
      "grad_norm": 8.126190185546875,
      "learning_rate": 0.0,
      "loss": 1.3257,
      "step": 690
    },
    {
      "epoch": 0.062017591096751035,
      "grad_norm": 9.248212814331055,
      "learning_rate": 0.0,
      "loss": 1.1431,
      "step": 691
    },
    {
      "epoch": 0.06210734159037875,
      "grad_norm": 12.316169738769531,
      "learning_rate": 0.0,
      "loss": 1.2077,
      "step": 692
    },
    {
      "epoch": 0.062197092084006464,
      "grad_norm": 8.50390625,
      "learning_rate": 0.0,
      "loss": 1.2981,
      "step": 693
    },
    {
      "epoch": 0.06228684257763418,
      "grad_norm": 6.9215779304504395,
      "learning_rate": 0.0,
      "loss": 1.262,
      "step": 694
    },
    {
      "epoch": 0.06237659307126189,
      "grad_norm": 7.897719383239746,
      "learning_rate": 0.0,
      "loss": 1.2435,
      "step": 695
    },
    {
      "epoch": 0.06246634356488961,
      "grad_norm": 6.160327911376953,
      "learning_rate": 0.0,
      "loss": 1.1483,
      "step": 696
    },
    {
      "epoch": 0.06255609405851732,
      "grad_norm": 7.75071907043457,
      "learning_rate": 0.0,
      "loss": 1.236,
      "step": 697
    },
    {
      "epoch": 0.06264584455214503,
      "grad_norm": 14.363978385925293,
      "learning_rate": 0.0,
      "loss": 1.2955,
      "step": 698
    },
    {
      "epoch": 0.06273559504577275,
      "grad_norm": 6.927734851837158,
      "learning_rate": 0.0,
      "loss": 1.2957,
      "step": 699
    },
    {
      "epoch": 0.06282534553940047,
      "grad_norm": 9.828089714050293,
      "learning_rate": 0.0,
      "loss": 1.1499,
      "step": 700
    },
    {
      "epoch": 0.06291509603302818,
      "grad_norm": 18.66272735595703,
      "learning_rate": 0.0,
      "loss": 1.2152,
      "step": 701
    },
    {
      "epoch": 0.0630048465266559,
      "grad_norm": 7.887050628662109,
      "learning_rate": 0.0,
      "loss": 1.2922,
      "step": 702
    },
    {
      "epoch": 0.06309459702028361,
      "grad_norm": 7.400357723236084,
      "learning_rate": 0.0,
      "loss": 1.2064,
      "step": 703
    },
    {
      "epoch": 0.06318434751391133,
      "grad_norm": 17.585416793823242,
      "learning_rate": 0.0,
      "loss": 1.1403,
      "step": 704
    },
    {
      "epoch": 0.06327409800753904,
      "grad_norm": 13.975406646728516,
      "learning_rate": 0.0,
      "loss": 1.193,
      "step": 705
    },
    {
      "epoch": 0.06336384850116676,
      "grad_norm": 8.205892562866211,
      "learning_rate": 0.0,
      "loss": 1.2238,
      "step": 706
    },
    {
      "epoch": 0.06345359899479447,
      "grad_norm": 7.143798828125,
      "learning_rate": 0.0,
      "loss": 1.1935,
      "step": 707
    },
    {
      "epoch": 0.06354334948842219,
      "grad_norm": 8.79647159576416,
      "learning_rate": 0.0,
      "loss": 1.2642,
      "step": 708
    },
    {
      "epoch": 0.0636330999820499,
      "grad_norm": 7.727572917938232,
      "learning_rate": 0.0,
      "loss": 1.1495,
      "step": 709
    },
    {
      "epoch": 0.06372285047567762,
      "grad_norm": 29.4732666015625,
      "learning_rate": 0.0,
      "loss": 1.2508,
      "step": 710
    },
    {
      "epoch": 0.06381260096930533,
      "grad_norm": 9.474916458129883,
      "learning_rate": 0.0,
      "loss": 1.1442,
      "step": 711
    },
    {
      "epoch": 0.06390235146293305,
      "grad_norm": 7.3754987716674805,
      "learning_rate": 0.0,
      "loss": 1.241,
      "step": 712
    },
    {
      "epoch": 0.06399210195656076,
      "grad_norm": 9.256865501403809,
      "learning_rate": 0.0,
      "loss": 1.1059,
      "step": 713
    },
    {
      "epoch": 0.06408185245018848,
      "grad_norm": 7.675616264343262,
      "learning_rate": 0.0,
      "loss": 1.1497,
      "step": 714
    },
    {
      "epoch": 0.06417160294381619,
      "grad_norm": 6.839668273925781,
      "learning_rate": 0.0,
      "loss": 1.2714,
      "step": 715
    },
    {
      "epoch": 0.06426135343744391,
      "grad_norm": 7.9349446296691895,
      "learning_rate": 0.0,
      "loss": 1.1685,
      "step": 716
    },
    {
      "epoch": 0.06435110393107162,
      "grad_norm": 9.680684089660645,
      "learning_rate": 0.0,
      "loss": 1.1754,
      "step": 717
    },
    {
      "epoch": 0.06444085442469934,
      "grad_norm": 7.260528087615967,
      "learning_rate": 0.0,
      "loss": 1.259,
      "step": 718
    },
    {
      "epoch": 0.06453060491832704,
      "grad_norm": 7.959448337554932,
      "learning_rate": 0.0,
      "loss": 1.2582,
      "step": 719
    },
    {
      "epoch": 0.06462035541195477,
      "grad_norm": 7.1109724044799805,
      "learning_rate": 0.0,
      "loss": 1.2664,
      "step": 720
    },
    {
      "epoch": 0.06471010590558247,
      "grad_norm": 8.753149032592773,
      "learning_rate": 0.0,
      "loss": 1.3466,
      "step": 721
    },
    {
      "epoch": 0.0647998563992102,
      "grad_norm": 7.312036514282227,
      "learning_rate": 0.0,
      "loss": 1.18,
      "step": 722
    },
    {
      "epoch": 0.06488960689283792,
      "grad_norm": 7.33585786819458,
      "learning_rate": 0.0,
      "loss": 1.2324,
      "step": 723
    },
    {
      "epoch": 0.06497935738646562,
      "grad_norm": 7.0310378074646,
      "learning_rate": 0.0,
      "loss": 1.1905,
      "step": 724
    },
    {
      "epoch": 0.06506910788009335,
      "grad_norm": 6.732304096221924,
      "learning_rate": 0.0,
      "loss": 1.1754,
      "step": 725
    },
    {
      "epoch": 0.06515885837372105,
      "grad_norm": 10.951607704162598,
      "learning_rate": 0.0,
      "loss": 1.3251,
      "step": 726
    },
    {
      "epoch": 0.06524860886734878,
      "grad_norm": 6.842125415802002,
      "learning_rate": 0.0,
      "loss": 1.1679,
      "step": 727
    },
    {
      "epoch": 0.06533835936097648,
      "grad_norm": 31.24333381652832,
      "learning_rate": 0.0,
      "loss": 1.2842,
      "step": 728
    },
    {
      "epoch": 0.0654281098546042,
      "grad_norm": 20.60365867614746,
      "learning_rate": 0.0,
      "loss": 1.2619,
      "step": 729
    },
    {
      "epoch": 0.06551786034823191,
      "grad_norm": 8.163527488708496,
      "learning_rate": 0.0,
      "loss": 1.2426,
      "step": 730
    },
    {
      "epoch": 0.06560761084185963,
      "grad_norm": 6.971711158752441,
      "learning_rate": 0.0,
      "loss": 1.2153,
      "step": 731
    },
    {
      "epoch": 0.06569736133548734,
      "grad_norm": 8.44365119934082,
      "learning_rate": 0.0,
      "loss": 1.2742,
      "step": 732
    },
    {
      "epoch": 0.06578711182911506,
      "grad_norm": 8.128055572509766,
      "learning_rate": 0.0,
      "loss": 1.2085,
      "step": 733
    },
    {
      "epoch": 0.06587686232274277,
      "grad_norm": 8.607390403747559,
      "learning_rate": 0.0,
      "loss": 1.1534,
      "step": 734
    },
    {
      "epoch": 0.06596661281637049,
      "grad_norm": 7.442856788635254,
      "learning_rate": 0.0,
      "loss": 1.1697,
      "step": 735
    },
    {
      "epoch": 0.0660563633099982,
      "grad_norm": 8.28091812133789,
      "learning_rate": 0.0,
      "loss": 1.2566,
      "step": 736
    },
    {
      "epoch": 0.06614611380362592,
      "grad_norm": 7.47618293762207,
      "learning_rate": 0.0,
      "loss": 1.2075,
      "step": 737
    },
    {
      "epoch": 0.06623586429725363,
      "grad_norm": 10.73886489868164,
      "learning_rate": 0.0,
      "loss": 1.1912,
      "step": 738
    },
    {
      "epoch": 0.06632561479088135,
      "grad_norm": 7.2000932693481445,
      "learning_rate": 0.0,
      "loss": 1.1645,
      "step": 739
    },
    {
      "epoch": 0.06641536528450906,
      "grad_norm": 7.536135196685791,
      "learning_rate": 0.0,
      "loss": 1.2597,
      "step": 740
    },
    {
      "epoch": 0.06650511577813678,
      "grad_norm": 7.574118137359619,
      "learning_rate": 0.0,
      "loss": 1.1671,
      "step": 741
    },
    {
      "epoch": 0.06659486627176449,
      "grad_norm": 8.812167167663574,
      "learning_rate": 0.0,
      "loss": 1.233,
      "step": 742
    },
    {
      "epoch": 0.06668461676539221,
      "grad_norm": 6.989040851593018,
      "learning_rate": 0.0,
      "loss": 1.1851,
      "step": 743
    },
    {
      "epoch": 0.06677436725901992,
      "grad_norm": 6.955749034881592,
      "learning_rate": 0.0,
      "loss": 1.2196,
      "step": 744
    },
    {
      "epoch": 0.06686411775264764,
      "grad_norm": 15.332763671875,
      "learning_rate": 0.0,
      "loss": 1.202,
      "step": 745
    },
    {
      "epoch": 0.06695386824627536,
      "grad_norm": 6.480578899383545,
      "learning_rate": 0.0,
      "loss": 1.238,
      "step": 746
    },
    {
      "epoch": 0.06704361873990307,
      "grad_norm": 16.174081802368164,
      "learning_rate": 0.0,
      "loss": 1.2326,
      "step": 747
    },
    {
      "epoch": 0.06713336923353079,
      "grad_norm": 8.442909240722656,
      "learning_rate": 0.0,
      "loss": 1.2064,
      "step": 748
    },
    {
      "epoch": 0.0672231197271585,
      "grad_norm": 8.169200897216797,
      "learning_rate": 0.0,
      "loss": 1.1466,
      "step": 749
    },
    {
      "epoch": 0.06731287022078622,
      "grad_norm": 9.890436172485352,
      "learning_rate": 0.0,
      "loss": 1.2553,
      "step": 750
    },
    {
      "epoch": 0.06740262071441393,
      "grad_norm": 6.694606304168701,
      "learning_rate": 0.0,
      "loss": 1.1477,
      "step": 751
    },
    {
      "epoch": 0.06749237120804165,
      "grad_norm": 11.647445678710938,
      "learning_rate": 0.0,
      "loss": 1.2627,
      "step": 752
    },
    {
      "epoch": 0.06758212170166936,
      "grad_norm": 8.215876579284668,
      "learning_rate": 0.0,
      "loss": 1.2552,
      "step": 753
    },
    {
      "epoch": 0.06767187219529708,
      "grad_norm": 8.456825256347656,
      "learning_rate": 0.0,
      "loss": 1.3136,
      "step": 754
    },
    {
      "epoch": 0.06776162268892479,
      "grad_norm": 10.277746200561523,
      "learning_rate": 0.0,
      "loss": 1.1373,
      "step": 755
    },
    {
      "epoch": 0.06785137318255251,
      "grad_norm": 8.996195793151855,
      "learning_rate": 0.0,
      "loss": 1.1579,
      "step": 756
    },
    {
      "epoch": 0.06794112367618022,
      "grad_norm": 9.645790100097656,
      "learning_rate": 0.0,
      "loss": 1.1362,
      "step": 757
    },
    {
      "epoch": 0.06803087416980794,
      "grad_norm": 6.626172065734863,
      "learning_rate": 0.0,
      "loss": 1.1957,
      "step": 758
    },
    {
      "epoch": 0.06812062466343564,
      "grad_norm": 8.773826599121094,
      "learning_rate": 0.0,
      "loss": 1.2414,
      "step": 759
    },
    {
      "epoch": 0.06821037515706337,
      "grad_norm": 6.706480979919434,
      "learning_rate": 0.0,
      "loss": 1.3325,
      "step": 760
    },
    {
      "epoch": 0.06830012565069107,
      "grad_norm": 10.221118927001953,
      "learning_rate": 0.0,
      "loss": 1.1824,
      "step": 761
    },
    {
      "epoch": 0.0683898761443188,
      "grad_norm": 7.728327751159668,
      "learning_rate": 0.0,
      "loss": 1.2389,
      "step": 762
    },
    {
      "epoch": 0.0684796266379465,
      "grad_norm": 7.103193283081055,
      "learning_rate": 0.0,
      "loss": 1.276,
      "step": 763
    },
    {
      "epoch": 0.06856937713157422,
      "grad_norm": 8.242243766784668,
      "learning_rate": 0.0,
      "loss": 1.206,
      "step": 764
    },
    {
      "epoch": 0.06865912762520193,
      "grad_norm": 8.32974910736084,
      "learning_rate": 0.0,
      "loss": 1.177,
      "step": 765
    },
    {
      "epoch": 0.06874887811882965,
      "grad_norm": 6.5688676834106445,
      "learning_rate": 0.0,
      "loss": 1.1296,
      "step": 766
    },
    {
      "epoch": 0.06883862861245738,
      "grad_norm": 8.473599433898926,
      "learning_rate": 0.0,
      "loss": 1.2119,
      "step": 767
    },
    {
      "epoch": 0.06892837910608508,
      "grad_norm": 8.427823066711426,
      "learning_rate": 0.0,
      "loss": 1.3997,
      "step": 768
    },
    {
      "epoch": 0.0690181295997128,
      "grad_norm": 7.408858776092529,
      "learning_rate": 0.0,
      "loss": 1.276,
      "step": 769
    },
    {
      "epoch": 0.06910788009334051,
      "grad_norm": 8.283830642700195,
      "learning_rate": 0.0,
      "loss": 1.2164,
      "step": 770
    },
    {
      "epoch": 0.06919763058696823,
      "grad_norm": 7.299758434295654,
      "learning_rate": 0.0,
      "loss": 1.2306,
      "step": 771
    },
    {
      "epoch": 0.06928738108059594,
      "grad_norm": 7.612678527832031,
      "learning_rate": 0.0,
      "loss": 1.263,
      "step": 772
    },
    {
      "epoch": 0.06937713157422366,
      "grad_norm": 8.479561805725098,
      "learning_rate": 0.0,
      "loss": 1.2367,
      "step": 773
    },
    {
      "epoch": 0.06946688206785137,
      "grad_norm": 6.482140064239502,
      "learning_rate": 0.0,
      "loss": 1.2293,
      "step": 774
    },
    {
      "epoch": 0.06955663256147909,
      "grad_norm": 7.932343482971191,
      "learning_rate": 0.0,
      "loss": 1.1915,
      "step": 775
    },
    {
      "epoch": 0.0696463830551068,
      "grad_norm": 9.7037935256958,
      "learning_rate": 0.0,
      "loss": 1.2876,
      "step": 776
    },
    {
      "epoch": 0.06973613354873452,
      "grad_norm": 22.56230354309082,
      "learning_rate": 0.0,
      "loss": 1.3031,
      "step": 777
    },
    {
      "epoch": 0.06982588404236223,
      "grad_norm": 7.3355865478515625,
      "learning_rate": 0.0,
      "loss": 1.2482,
      "step": 778
    },
    {
      "epoch": 0.06991563453598995,
      "grad_norm": 9.720284461975098,
      "learning_rate": 0.0,
      "loss": 1.2158,
      "step": 779
    },
    {
      "epoch": 0.07000538502961766,
      "grad_norm": 9.861888885498047,
      "learning_rate": 0.0,
      "loss": 1.1461,
      "step": 780
    },
    {
      "epoch": 0.07009513552324538,
      "grad_norm": 11.222338676452637,
      "learning_rate": 0.0,
      "loss": 1.3537,
      "step": 781
    },
    {
      "epoch": 0.07018488601687309,
      "grad_norm": 16.92066192626953,
      "learning_rate": 0.0,
      "loss": 1.1685,
      "step": 782
    },
    {
      "epoch": 0.07027463651050081,
      "grad_norm": 8.89771556854248,
      "learning_rate": 0.0,
      "loss": 1.1227,
      "step": 783
    },
    {
      "epoch": 0.07036438700412852,
      "grad_norm": 7.397648811340332,
      "learning_rate": 0.0,
      "loss": 1.3218,
      "step": 784
    },
    {
      "epoch": 0.07045413749775624,
      "grad_norm": 9.256004333496094,
      "learning_rate": 0.0,
      "loss": 1.2827,
      "step": 785
    },
    {
      "epoch": 0.07054388799138395,
      "grad_norm": 7.107243537902832,
      "learning_rate": 0.0,
      "loss": 1.2058,
      "step": 786
    },
    {
      "epoch": 0.07063363848501167,
      "grad_norm": 6.642272472381592,
      "learning_rate": 0.0,
      "loss": 1.1541,
      "step": 787
    },
    {
      "epoch": 0.07072338897863938,
      "grad_norm": 6.554023265838623,
      "learning_rate": 0.0,
      "loss": 1.1982,
      "step": 788
    },
    {
      "epoch": 0.0708131394722671,
      "grad_norm": 6.255526542663574,
      "learning_rate": 0.0,
      "loss": 1.2017,
      "step": 789
    },
    {
      "epoch": 0.07090288996589482,
      "grad_norm": 7.690706253051758,
      "learning_rate": 0.0,
      "loss": 1.2003,
      "step": 790
    },
    {
      "epoch": 0.07099264045952253,
      "grad_norm": 6.711826324462891,
      "learning_rate": 0.0,
      "loss": 1.1787,
      "step": 791
    },
    {
      "epoch": 0.07108239095315025,
      "grad_norm": 8.048186302185059,
      "learning_rate": 0.0,
      "loss": 1.2578,
      "step": 792
    },
    {
      "epoch": 0.07117214144677796,
      "grad_norm": 7.217451572418213,
      "learning_rate": 0.0,
      "loss": 1.1829,
      "step": 793
    },
    {
      "epoch": 0.07126189194040568,
      "grad_norm": 7.6622772216796875,
      "learning_rate": 0.0,
      "loss": 1.1761,
      "step": 794
    },
    {
      "epoch": 0.07135164243403339,
      "grad_norm": 8.135239601135254,
      "learning_rate": 0.0,
      "loss": 1.1712,
      "step": 795
    },
    {
      "epoch": 0.07144139292766111,
      "grad_norm": 7.48563289642334,
      "learning_rate": 0.0,
      "loss": 1.1957,
      "step": 796
    },
    {
      "epoch": 0.07153114342128881,
      "grad_norm": 16.007081985473633,
      "learning_rate": 0.0,
      "loss": 1.2894,
      "step": 797
    },
    {
      "epoch": 0.07162089391491654,
      "grad_norm": 8.43378734588623,
      "learning_rate": 0.0,
      "loss": 1.1919,
      "step": 798
    },
    {
      "epoch": 0.07171064440854424,
      "grad_norm": 9.972838401794434,
      "learning_rate": 0.0,
      "loss": 1.214,
      "step": 799
    },
    {
      "epoch": 0.07180039490217197,
      "grad_norm": 6.840358257293701,
      "learning_rate": 0.0,
      "loss": 1.1581,
      "step": 800
    },
    {
      "epoch": 0.07189014539579967,
      "grad_norm": 8.057051658630371,
      "learning_rate": 0.0,
      "loss": 1.223,
      "step": 801
    },
    {
      "epoch": 0.0719798958894274,
      "grad_norm": 6.591375350952148,
      "learning_rate": 0.0,
      "loss": 1.1954,
      "step": 802
    },
    {
      "epoch": 0.0720696463830551,
      "grad_norm": 9.266378402709961,
      "learning_rate": 0.0,
      "loss": 1.2944,
      "step": 803
    },
    {
      "epoch": 0.07215939687668282,
      "grad_norm": 6.905794143676758,
      "learning_rate": 0.0,
      "loss": 1.1577,
      "step": 804
    },
    {
      "epoch": 0.07224914737031053,
      "grad_norm": 6.85727596282959,
      "learning_rate": 0.0,
      "loss": 1.1671,
      "step": 805
    },
    {
      "epoch": 0.07233889786393825,
      "grad_norm": 17.60844612121582,
      "learning_rate": 0.0,
      "loss": 1.2222,
      "step": 806
    },
    {
      "epoch": 0.07242864835756596,
      "grad_norm": 9.453246116638184,
      "learning_rate": 0.0,
      "loss": 1.2915,
      "step": 807
    },
    {
      "epoch": 0.07251839885119368,
      "grad_norm": 7.449007034301758,
      "learning_rate": 0.0,
      "loss": 1.2018,
      "step": 808
    },
    {
      "epoch": 0.07260814934482139,
      "grad_norm": 8.543807983398438,
      "learning_rate": 0.0,
      "loss": 1.2592,
      "step": 809
    },
    {
      "epoch": 0.07269789983844911,
      "grad_norm": 8.940797805786133,
      "learning_rate": 0.0,
      "loss": 1.1754,
      "step": 810
    },
    {
      "epoch": 0.07278765033207682,
      "grad_norm": 6.00596809387207,
      "learning_rate": 0.0,
      "loss": 1.1449,
      "step": 811
    },
    {
      "epoch": 0.07287740082570454,
      "grad_norm": 7.4050068855285645,
      "learning_rate": 0.0,
      "loss": 1.2063,
      "step": 812
    },
    {
      "epoch": 0.07296715131933226,
      "grad_norm": 6.327250957489014,
      "learning_rate": 0.0,
      "loss": 1.1869,
      "step": 813
    },
    {
      "epoch": 0.07305690181295997,
      "grad_norm": 13.234457015991211,
      "learning_rate": 0.0,
      "loss": 1.1681,
      "step": 814
    },
    {
      "epoch": 0.07314665230658769,
      "grad_norm": 6.509833812713623,
      "learning_rate": 0.0,
      "loss": 1.2233,
      "step": 815
    },
    {
      "epoch": 0.0732364028002154,
      "grad_norm": 6.728079795837402,
      "learning_rate": 0.0,
      "loss": 1.1512,
      "step": 816
    },
    {
      "epoch": 0.07332615329384312,
      "grad_norm": 7.254768371582031,
      "learning_rate": 0.0,
      "loss": 1.2025,
      "step": 817
    },
    {
      "epoch": 0.07341590378747083,
      "grad_norm": 6.7629241943359375,
      "learning_rate": 0.0,
      "loss": 1.2392,
      "step": 818
    },
    {
      "epoch": 0.07350565428109855,
      "grad_norm": 8.764588356018066,
      "learning_rate": 0.0,
      "loss": 1.1976,
      "step": 819
    },
    {
      "epoch": 0.07359540477472626,
      "grad_norm": 11.947595596313477,
      "learning_rate": 0.0,
      "loss": 1.2089,
      "step": 820
    },
    {
      "epoch": 0.07368515526835398,
      "grad_norm": 8.044597625732422,
      "learning_rate": 0.0,
      "loss": 1.2972,
      "step": 821
    },
    {
      "epoch": 0.07377490576198169,
      "grad_norm": 8.550481796264648,
      "learning_rate": 0.0,
      "loss": 1.1777,
      "step": 822
    },
    {
      "epoch": 0.07386465625560941,
      "grad_norm": 7.809882640838623,
      "learning_rate": 0.0,
      "loss": 1.2485,
      "step": 823
    },
    {
      "epoch": 0.07395440674923712,
      "grad_norm": 6.6445631980896,
      "learning_rate": 0.0,
      "loss": 1.2586,
      "step": 824
    },
    {
      "epoch": 0.07404415724286484,
      "grad_norm": 7.329726696014404,
      "learning_rate": 0.0,
      "loss": 1.2288,
      "step": 825
    },
    {
      "epoch": 0.07413390773649255,
      "grad_norm": 6.452029228210449,
      "learning_rate": 0.0,
      "loss": 1.2172,
      "step": 826
    },
    {
      "epoch": 0.07422365823012027,
      "grad_norm": 7.021517753601074,
      "learning_rate": 0.0,
      "loss": 1.2317,
      "step": 827
    },
    {
      "epoch": 0.07431340872374798,
      "grad_norm": 7.776928901672363,
      "learning_rate": 0.0,
      "loss": 1.2248,
      "step": 828
    },
    {
      "epoch": 0.0744031592173757,
      "grad_norm": 7.489571571350098,
      "learning_rate": 0.0,
      "loss": 1.2276,
      "step": 829
    },
    {
      "epoch": 0.0744929097110034,
      "grad_norm": 7.420658588409424,
      "learning_rate": 0.0,
      "loss": 1.2086,
      "step": 830
    },
    {
      "epoch": 0.07458266020463113,
      "grad_norm": 7.517425537109375,
      "learning_rate": 0.0,
      "loss": 1.1848,
      "step": 831
    },
    {
      "epoch": 0.07467241069825883,
      "grad_norm": 8.9600248336792,
      "learning_rate": 0.0,
      "loss": 1.1772,
      "step": 832
    },
    {
      "epoch": 0.07476216119188656,
      "grad_norm": 13.469948768615723,
      "learning_rate": 0.0,
      "loss": 1.2426,
      "step": 833
    },
    {
      "epoch": 0.07485191168551426,
      "grad_norm": 7.874142646789551,
      "learning_rate": 0.0,
      "loss": 1.1254,
      "step": 834
    },
    {
      "epoch": 0.07494166217914199,
      "grad_norm": 10.567255973815918,
      "learning_rate": 0.0,
      "loss": 1.1291,
      "step": 835
    },
    {
      "epoch": 0.0750314126727697,
      "grad_norm": 8.008703231811523,
      "learning_rate": 0.0,
      "loss": 1.1805,
      "step": 836
    },
    {
      "epoch": 0.07512116316639741,
      "grad_norm": 6.580844879150391,
      "learning_rate": 0.0,
      "loss": 1.1471,
      "step": 837
    },
    {
      "epoch": 0.07521091366002514,
      "grad_norm": 6.4846014976501465,
      "learning_rate": 0.0,
      "loss": 1.168,
      "step": 838
    },
    {
      "epoch": 0.07530066415365284,
      "grad_norm": 7.446361541748047,
      "learning_rate": 0.0,
      "loss": 1.1823,
      "step": 839
    },
    {
      "epoch": 0.07539041464728057,
      "grad_norm": 8.074203491210938,
      "learning_rate": 0.0,
      "loss": 1.1567,
      "step": 840
    },
    {
      "epoch": 0.07548016514090827,
      "grad_norm": 8.219217300415039,
      "learning_rate": 0.0,
      "loss": 1.2107,
      "step": 841
    },
    {
      "epoch": 0.075569915634536,
      "grad_norm": 7.227478981018066,
      "learning_rate": 0.0,
      "loss": 1.2357,
      "step": 842
    },
    {
      "epoch": 0.0756596661281637,
      "grad_norm": 11.660436630249023,
      "learning_rate": 0.0,
      "loss": 1.2308,
      "step": 843
    },
    {
      "epoch": 0.07574941662179142,
      "grad_norm": 18.282194137573242,
      "learning_rate": 0.0,
      "loss": 1.2279,
      "step": 844
    },
    {
      "epoch": 0.07583916711541913,
      "grad_norm": 7.76088285446167,
      "learning_rate": 0.0,
      "loss": 1.2486,
      "step": 845
    },
    {
      "epoch": 0.07592891760904685,
      "grad_norm": 7.343393325805664,
      "learning_rate": 0.0,
      "loss": 1.1407,
      "step": 846
    },
    {
      "epoch": 0.07601866810267456,
      "grad_norm": 6.921496391296387,
      "learning_rate": 0.0,
      "loss": 1.273,
      "step": 847
    },
    {
      "epoch": 0.07610841859630228,
      "grad_norm": 7.236971855163574,
      "learning_rate": 0.0,
      "loss": 1.1668,
      "step": 848
    },
    {
      "epoch": 0.07619816908992999,
      "grad_norm": 9.196698188781738,
      "learning_rate": 0.0,
      "loss": 1.2189,
      "step": 849
    },
    {
      "epoch": 0.07628791958355771,
      "grad_norm": 9.17573356628418,
      "learning_rate": 0.0,
      "loss": 1.2022,
      "step": 850
    },
    {
      "epoch": 0.07637767007718542,
      "grad_norm": 7.845760345458984,
      "learning_rate": 0.0,
      "loss": 1.1425,
      "step": 851
    },
    {
      "epoch": 0.07646742057081314,
      "grad_norm": 7.151289463043213,
      "learning_rate": 0.0,
      "loss": 1.2213,
      "step": 852
    },
    {
      "epoch": 0.07655717106444085,
      "grad_norm": 16.760568618774414,
      "learning_rate": 0.0,
      "loss": 1.2619,
      "step": 853
    },
    {
      "epoch": 0.07664692155806857,
      "grad_norm": 9.059422492980957,
      "learning_rate": 0.0,
      "loss": 1.1828,
      "step": 854
    },
    {
      "epoch": 0.07673667205169628,
      "grad_norm": 6.598907947540283,
      "learning_rate": 0.0,
      "loss": 1.1872,
      "step": 855
    },
    {
      "epoch": 0.076826422545324,
      "grad_norm": 9.278754234313965,
      "learning_rate": 0.0,
      "loss": 1.2031,
      "step": 856
    },
    {
      "epoch": 0.07691617303895171,
      "grad_norm": 9.145720481872559,
      "learning_rate": 0.0,
      "loss": 1.1318,
      "step": 857
    },
    {
      "epoch": 0.07700592353257943,
      "grad_norm": 6.089593410491943,
      "learning_rate": 0.0,
      "loss": 1.1413,
      "step": 858
    },
    {
      "epoch": 0.07709567402620715,
      "grad_norm": 8.665871620178223,
      "learning_rate": 0.0,
      "loss": 1.1806,
      "step": 859
    },
    {
      "epoch": 0.07718542451983486,
      "grad_norm": 7.694634437561035,
      "learning_rate": 0.0,
      "loss": 1.2056,
      "step": 860
    },
    {
      "epoch": 0.07727517501346258,
      "grad_norm": 7.201373100280762,
      "learning_rate": 0.0,
      "loss": 1.1764,
      "step": 861
    },
    {
      "epoch": 0.07736492550709029,
      "grad_norm": 12.837779998779297,
      "learning_rate": 0.0,
      "loss": 1.1922,
      "step": 862
    },
    {
      "epoch": 0.07745467600071801,
      "grad_norm": 6.5688371658325195,
      "learning_rate": 0.0,
      "loss": 1.1373,
      "step": 863
    },
    {
      "epoch": 0.07754442649434572,
      "grad_norm": 8.22211742401123,
      "learning_rate": 0.0,
      "loss": 1.2037,
      "step": 864
    },
    {
      "epoch": 0.07763417698797344,
      "grad_norm": 7.601479530334473,
      "learning_rate": 0.0,
      "loss": 1.275,
      "step": 865
    },
    {
      "epoch": 0.07772392748160115,
      "grad_norm": 5.90705680847168,
      "learning_rate": 0.0,
      "loss": 1.185,
      "step": 866
    },
    {
      "epoch": 0.07781367797522887,
      "grad_norm": 7.438308238983154,
      "learning_rate": 0.0,
      "loss": 1.2045,
      "step": 867
    },
    {
      "epoch": 0.07790342846885658,
      "grad_norm": 7.955481052398682,
      "learning_rate": 0.0,
      "loss": 1.172,
      "step": 868
    },
    {
      "epoch": 0.0779931789624843,
      "grad_norm": 6.435699939727783,
      "learning_rate": 0.0,
      "loss": 1.1254,
      "step": 869
    },
    {
      "epoch": 0.078082929456112,
      "grad_norm": 7.300787448883057,
      "learning_rate": 0.0,
      "loss": 1.1982,
      "step": 870
    },
    {
      "epoch": 0.07817267994973973,
      "grad_norm": 6.790376663208008,
      "learning_rate": 0.0,
      "loss": 1.1484,
      "step": 871
    },
    {
      "epoch": 0.07826243044336743,
      "grad_norm": 7.453437805175781,
      "learning_rate": 0.0,
      "loss": 1.1461,
      "step": 872
    },
    {
      "epoch": 0.07835218093699516,
      "grad_norm": 7.476863384246826,
      "learning_rate": 0.0,
      "loss": 1.2733,
      "step": 873
    },
    {
      "epoch": 0.07844193143062286,
      "grad_norm": 6.835644245147705,
      "learning_rate": 0.0,
      "loss": 1.1912,
      "step": 874
    },
    {
      "epoch": 0.07853168192425058,
      "grad_norm": 10.508700370788574,
      "learning_rate": 0.0,
      "loss": 1.2674,
      "step": 875
    },
    {
      "epoch": 0.07862143241787829,
      "grad_norm": 11.131389617919922,
      "learning_rate": 0.0,
      "loss": 1.209,
      "step": 876
    },
    {
      "epoch": 0.07871118291150601,
      "grad_norm": 8.191415786743164,
      "learning_rate": 0.0,
      "loss": 1.2053,
      "step": 877
    },
    {
      "epoch": 0.07880093340513372,
      "grad_norm": 8.031414985656738,
      "learning_rate": 0.0,
      "loss": 1.2613,
      "step": 878
    },
    {
      "epoch": 0.07889068389876144,
      "grad_norm": 10.458081245422363,
      "learning_rate": 0.0,
      "loss": 1.1804,
      "step": 879
    },
    {
      "epoch": 0.07898043439238916,
      "grad_norm": 6.822637557983398,
      "learning_rate": 0.0,
      "loss": 1.1961,
      "step": 880
    },
    {
      "epoch": 0.07907018488601687,
      "grad_norm": 11.033238410949707,
      "learning_rate": 0.0,
      "loss": 1.1805,
      "step": 881
    },
    {
      "epoch": 0.0791599353796446,
      "grad_norm": 6.670080661773682,
      "learning_rate": 0.0,
      "loss": 1.1458,
      "step": 882
    },
    {
      "epoch": 0.0792496858732723,
      "grad_norm": 6.130035400390625,
      "learning_rate": 0.0,
      "loss": 1.2301,
      "step": 883
    },
    {
      "epoch": 0.07933943636690002,
      "grad_norm": 9.61927318572998,
      "learning_rate": 0.0,
      "loss": 1.2024,
      "step": 884
    },
    {
      "epoch": 0.07942918686052773,
      "grad_norm": 9.9964599609375,
      "learning_rate": 0.0,
      "loss": 1.1345,
      "step": 885
    },
    {
      "epoch": 0.07951893735415545,
      "grad_norm": 8.449503898620605,
      "learning_rate": 0.0,
      "loss": 1.1587,
      "step": 886
    },
    {
      "epoch": 0.07960868784778316,
      "grad_norm": 7.383955478668213,
      "learning_rate": 0.0,
      "loss": 1.1869,
      "step": 887
    },
    {
      "epoch": 0.07969843834141088,
      "grad_norm": 8.026031494140625,
      "learning_rate": 0.0,
      "loss": 1.1783,
      "step": 888
    },
    {
      "epoch": 0.07978818883503859,
      "grad_norm": 9.84389591217041,
      "learning_rate": 0.0,
      "loss": 1.2407,
      "step": 889
    },
    {
      "epoch": 0.07987793932866631,
      "grad_norm": 9.95412540435791,
      "learning_rate": 0.0,
      "loss": 1.2071,
      "step": 890
    },
    {
      "epoch": 0.07996768982229402,
      "grad_norm": 19.523942947387695,
      "learning_rate": 0.0,
      "loss": 1.2209,
      "step": 891
    },
    {
      "epoch": 0.08005744031592174,
      "grad_norm": 64.311767578125,
      "learning_rate": 0.0,
      "loss": 1.1337,
      "step": 892
    },
    {
      "epoch": 0.08014719080954945,
      "grad_norm": 19.176645278930664,
      "learning_rate": 0.0,
      "loss": 1.228,
      "step": 893
    },
    {
      "epoch": 0.08023694130317717,
      "grad_norm": 13.19560718536377,
      "learning_rate": 0.0,
      "loss": 1.1719,
      "step": 894
    },
    {
      "epoch": 0.08032669179680488,
      "grad_norm": 9.690608024597168,
      "learning_rate": 0.0,
      "loss": 1.1782,
      "step": 895
    },
    {
      "epoch": 0.0804164422904326,
      "grad_norm": 6.4091877937316895,
      "learning_rate": 0.0,
      "loss": 1.1861,
      "step": 896
    },
    {
      "epoch": 0.08050619278406031,
      "grad_norm": 11.37458610534668,
      "learning_rate": 0.0,
      "loss": 1.1727,
      "step": 897
    },
    {
      "epoch": 0.08059594327768803,
      "grad_norm": 7.963089466094971,
      "learning_rate": 0.0,
      "loss": 1.1556,
      "step": 898
    },
    {
      "epoch": 0.08068569377131574,
      "grad_norm": 9.752516746520996,
      "learning_rate": 0.0,
      "loss": 1.1714,
      "step": 899
    },
    {
      "epoch": 0.08077544426494346,
      "grad_norm": 24.19694709777832,
      "learning_rate": 0.0,
      "loss": 1.2245,
      "step": 900
    },
    {
      "epoch": 0.08086519475857117,
      "grad_norm": 5.99568510055542,
      "learning_rate": 0.0,
      "loss": 1.2018,
      "step": 901
    },
    {
      "epoch": 0.08095494525219889,
      "grad_norm": 11.29816722869873,
      "learning_rate": 0.0,
      "loss": 1.1552,
      "step": 902
    },
    {
      "epoch": 0.08104469574582661,
      "grad_norm": 7.459784030914307,
      "learning_rate": 0.0,
      "loss": 1.2536,
      "step": 903
    },
    {
      "epoch": 0.08113444623945432,
      "grad_norm": 6.2434000968933105,
      "learning_rate": 0.0,
      "loss": 1.1681,
      "step": 904
    },
    {
      "epoch": 0.08122419673308204,
      "grad_norm": 7.285932540893555,
      "learning_rate": 0.0,
      "loss": 1.2273,
      "step": 905
    },
    {
      "epoch": 0.08131394722670975,
      "grad_norm": 7.426870346069336,
      "learning_rate": 0.0,
      "loss": 1.1229,
      "step": 906
    },
    {
      "epoch": 0.08140369772033747,
      "grad_norm": 11.44782543182373,
      "learning_rate": 0.0,
      "loss": 1.159,
      "step": 907
    },
    {
      "epoch": 0.08149344821396518,
      "grad_norm": 6.112159729003906,
      "learning_rate": 0.0,
      "loss": 1.2834,
      "step": 908
    },
    {
      "epoch": 0.0815831987075929,
      "grad_norm": 6.262115955352783,
      "learning_rate": 0.0,
      "loss": 1.2446,
      "step": 909
    },
    {
      "epoch": 0.0816729492012206,
      "grad_norm": 8.489212036132812,
      "learning_rate": 0.0,
      "loss": 1.2081,
      "step": 910
    },
    {
      "epoch": 0.08176269969484833,
      "grad_norm": 10.171384811401367,
      "learning_rate": 0.0,
      "loss": 1.2394,
      "step": 911
    },
    {
      "epoch": 0.08185245018847603,
      "grad_norm": 7.40484619140625,
      "learning_rate": 0.0,
      "loss": 1.1582,
      "step": 912
    },
    {
      "epoch": 0.08194220068210376,
      "grad_norm": 6.700619220733643,
      "learning_rate": 0.0,
      "loss": 1.1624,
      "step": 913
    },
    {
      "epoch": 0.08203195117573146,
      "grad_norm": 6.758608341217041,
      "learning_rate": 0.0,
      "loss": 1.2023,
      "step": 914
    },
    {
      "epoch": 0.08212170166935918,
      "grad_norm": 12.899343490600586,
      "learning_rate": 0.0,
      "loss": 1.2104,
      "step": 915
    },
    {
      "epoch": 0.08221145216298689,
      "grad_norm": 7.650911331176758,
      "learning_rate": 0.0,
      "loss": 1.2044,
      "step": 916
    },
    {
      "epoch": 0.08230120265661461,
      "grad_norm": 31.004640579223633,
      "learning_rate": 0.0,
      "loss": 1.2753,
      "step": 917
    },
    {
      "epoch": 0.08239095315024232,
      "grad_norm": 6.285728454589844,
      "learning_rate": 0.0,
      "loss": 1.2001,
      "step": 918
    },
    {
      "epoch": 0.08248070364387004,
      "grad_norm": 7.181550979614258,
      "learning_rate": 0.0,
      "loss": 1.1725,
      "step": 919
    },
    {
      "epoch": 0.08257045413749775,
      "grad_norm": 9.516425132751465,
      "learning_rate": 0.0,
      "loss": 1.2668,
      "step": 920
    },
    {
      "epoch": 0.08266020463112547,
      "grad_norm": 7.79632043838501,
      "learning_rate": 0.0,
      "loss": 1.1711,
      "step": 921
    },
    {
      "epoch": 0.08274995512475318,
      "grad_norm": 7.173058032989502,
      "learning_rate": 0.0,
      "loss": 1.2199,
      "step": 922
    },
    {
      "epoch": 0.0828397056183809,
      "grad_norm": 8.771575927734375,
      "learning_rate": 0.0,
      "loss": 1.2218,
      "step": 923
    },
    {
      "epoch": 0.08292945611200861,
      "grad_norm": 6.734317302703857,
      "learning_rate": 0.0,
      "loss": 1.1246,
      "step": 924
    },
    {
      "epoch": 0.08301920660563633,
      "grad_norm": 7.835805892944336,
      "learning_rate": 0.0,
      "loss": 1.1908,
      "step": 925
    },
    {
      "epoch": 0.08310895709926405,
      "grad_norm": 56.917022705078125,
      "learning_rate": 0.0,
      "loss": 1.1626,
      "step": 926
    },
    {
      "epoch": 0.08319870759289176,
      "grad_norm": 7.821000099182129,
      "learning_rate": 0.0,
      "loss": 1.1406,
      "step": 927
    },
    {
      "epoch": 0.08328845808651948,
      "grad_norm": 7.602083206176758,
      "learning_rate": 0.0,
      "loss": 1.2185,
      "step": 928
    },
    {
      "epoch": 0.08337820858014719,
      "grad_norm": 30.258586883544922,
      "learning_rate": 0.0,
      "loss": 1.2115,
      "step": 929
    },
    {
      "epoch": 0.08346795907377491,
      "grad_norm": 13.010772705078125,
      "learning_rate": 0.0,
      "loss": 1.2156,
      "step": 930
    },
    {
      "epoch": 0.08355770956740262,
      "grad_norm": 7.9814276695251465,
      "learning_rate": 0.0,
      "loss": 1.1966,
      "step": 931
    },
    {
      "epoch": 0.08364746006103034,
      "grad_norm": 7.482748031616211,
      "learning_rate": 0.0,
      "loss": 1.1359,
      "step": 932
    },
    {
      "epoch": 0.08373721055465805,
      "grad_norm": 8.304885864257812,
      "learning_rate": 0.0,
      "loss": 1.147,
      "step": 933
    },
    {
      "epoch": 0.08382696104828577,
      "grad_norm": 7.896909236907959,
      "learning_rate": 0.0,
      "loss": 1.1972,
      "step": 934
    },
    {
      "epoch": 0.08391671154191348,
      "grad_norm": 7.343756675720215,
      "learning_rate": 0.0,
      "loss": 1.1657,
      "step": 935
    },
    {
      "epoch": 0.0840064620355412,
      "grad_norm": 7.052554130554199,
      "learning_rate": 0.0,
      "loss": 1.1932,
      "step": 936
    },
    {
      "epoch": 0.0840962125291689,
      "grad_norm": 7.366707801818848,
      "learning_rate": 0.0,
      "loss": 1.1482,
      "step": 937
    },
    {
      "epoch": 0.08418596302279663,
      "grad_norm": 62.67080307006836,
      "learning_rate": 0.0,
      "loss": 1.401,
      "step": 938
    },
    {
      "epoch": 0.08427571351642434,
      "grad_norm": 8.108113288879395,
      "learning_rate": 0.0,
      "loss": 1.2752,
      "step": 939
    },
    {
      "epoch": 0.08436546401005206,
      "grad_norm": 7.313398838043213,
      "learning_rate": 0.0,
      "loss": 1.2014,
      "step": 940
    },
    {
      "epoch": 0.08445521450367977,
      "grad_norm": 9.765104293823242,
      "learning_rate": 0.0,
      "loss": 1.1418,
      "step": 941
    },
    {
      "epoch": 0.08454496499730749,
      "grad_norm": 6.430360794067383,
      "learning_rate": 0.0,
      "loss": 1.1899,
      "step": 942
    },
    {
      "epoch": 0.0846347154909352,
      "grad_norm": 7.832334518432617,
      "learning_rate": 0.0,
      "loss": 1.2548,
      "step": 943
    },
    {
      "epoch": 0.08472446598456292,
      "grad_norm": 25.304805755615234,
      "learning_rate": 0.0,
      "loss": 1.2709,
      "step": 944
    },
    {
      "epoch": 0.08481421647819062,
      "grad_norm": 10.725024223327637,
      "learning_rate": 0.0,
      "loss": 1.2024,
      "step": 945
    },
    {
      "epoch": 0.08490396697181835,
      "grad_norm": 7.58568811416626,
      "learning_rate": 0.0,
      "loss": 1.2406,
      "step": 946
    },
    {
      "epoch": 0.08499371746544605,
      "grad_norm": 16.245206832885742,
      "learning_rate": 0.0,
      "loss": 1.1533,
      "step": 947
    },
    {
      "epoch": 0.08508346795907377,
      "grad_norm": 7.28816556930542,
      "learning_rate": 0.0,
      "loss": 1.2571,
      "step": 948
    },
    {
      "epoch": 0.0851732184527015,
      "grad_norm": 8.235501289367676,
      "learning_rate": 0.0,
      "loss": 1.1505,
      "step": 949
    },
    {
      "epoch": 0.0852629689463292,
      "grad_norm": 7.623768329620361,
      "learning_rate": 0.0,
      "loss": 1.2267,
      "step": 950
    },
    {
      "epoch": 0.08535271943995693,
      "grad_norm": 6.568101406097412,
      "learning_rate": 0.0,
      "loss": 1.1806,
      "step": 951
    },
    {
      "epoch": 0.08544246993358463,
      "grad_norm": 7.296595573425293,
      "learning_rate": 0.0,
      "loss": 1.3279,
      "step": 952
    },
    {
      "epoch": 0.08553222042721235,
      "grad_norm": 8.479077339172363,
      "learning_rate": 0.0,
      "loss": 1.181,
      "step": 953
    },
    {
      "epoch": 0.08562197092084006,
      "grad_norm": 8.264373779296875,
      "learning_rate": 0.0,
      "loss": 1.2282,
      "step": 954
    },
    {
      "epoch": 0.08571172141446778,
      "grad_norm": 8.98880672454834,
      "learning_rate": 0.0,
      "loss": 1.1253,
      "step": 955
    },
    {
      "epoch": 0.08580147190809549,
      "grad_norm": 13.230514526367188,
      "learning_rate": 0.0,
      "loss": 1.1469,
      "step": 956
    },
    {
      "epoch": 0.08589122240172321,
      "grad_norm": 7.374499797821045,
      "learning_rate": 0.0,
      "loss": 1.2272,
      "step": 957
    },
    {
      "epoch": 0.08598097289535092,
      "grad_norm": 6.780921936035156,
      "learning_rate": 0.0,
      "loss": 1.1705,
      "step": 958
    },
    {
      "epoch": 0.08607072338897864,
      "grad_norm": 8.501517295837402,
      "learning_rate": 0.0,
      "loss": 1.2173,
      "step": 959
    },
    {
      "epoch": 0.08616047388260635,
      "grad_norm": 8.65261459350586,
      "learning_rate": 0.0,
      "loss": 1.2381,
      "step": 960
    },
    {
      "epoch": 0.08625022437623407,
      "grad_norm": 6.876990795135498,
      "learning_rate": 0.0,
      "loss": 1.1723,
      "step": 961
    },
    {
      "epoch": 0.08633997486986178,
      "grad_norm": 7.882692337036133,
      "learning_rate": 0.0,
      "loss": 1.1878,
      "step": 962
    },
    {
      "epoch": 0.0864297253634895,
      "grad_norm": 8.460224151611328,
      "learning_rate": 0.0,
      "loss": 1.2266,
      "step": 963
    },
    {
      "epoch": 0.08651947585711721,
      "grad_norm": 10.098305702209473,
      "learning_rate": 0.0,
      "loss": 1.1498,
      "step": 964
    },
    {
      "epoch": 0.08660922635074493,
      "grad_norm": 8.386086463928223,
      "learning_rate": 0.0,
      "loss": 1.2173,
      "step": 965
    },
    {
      "epoch": 0.08669897684437264,
      "grad_norm": 6.781382083892822,
      "learning_rate": 0.0,
      "loss": 1.1845,
      "step": 966
    },
    {
      "epoch": 0.08678872733800036,
      "grad_norm": 12.016521453857422,
      "learning_rate": 0.0,
      "loss": 1.1608,
      "step": 967
    },
    {
      "epoch": 0.08687847783162807,
      "grad_norm": 7.807769775390625,
      "learning_rate": 0.0,
      "loss": 1.1666,
      "step": 968
    },
    {
      "epoch": 0.08696822832525579,
      "grad_norm": 6.243021488189697,
      "learning_rate": 0.0,
      "loss": 1.1987,
      "step": 969
    },
    {
      "epoch": 0.08705797881888351,
      "grad_norm": 7.656095504760742,
      "learning_rate": 0.0,
      "loss": 1.2406,
      "step": 970
    },
    {
      "epoch": 0.08714772931251122,
      "grad_norm": 13.959967613220215,
      "learning_rate": 0.0,
      "loss": 1.2826,
      "step": 971
    },
    {
      "epoch": 0.08723747980613894,
      "grad_norm": 6.384106636047363,
      "learning_rate": 0.0,
      "loss": 1.1636,
      "step": 972
    },
    {
      "epoch": 0.08732723029976665,
      "grad_norm": 7.330763816833496,
      "learning_rate": 0.0,
      "loss": 1.2248,
      "step": 973
    },
    {
      "epoch": 0.08741698079339437,
      "grad_norm": 8.058364868164062,
      "learning_rate": 0.0,
      "loss": 1.244,
      "step": 974
    },
    {
      "epoch": 0.08750673128702208,
      "grad_norm": 6.976603984832764,
      "learning_rate": 0.0,
      "loss": 1.2324,
      "step": 975
    },
    {
      "epoch": 0.0875964817806498,
      "grad_norm": 6.300326347351074,
      "learning_rate": 0.0,
      "loss": 1.2352,
      "step": 976
    },
    {
      "epoch": 0.0876862322742775,
      "grad_norm": 9.719828605651855,
      "learning_rate": 0.0,
      "loss": 1.231,
      "step": 977
    },
    {
      "epoch": 0.08777598276790523,
      "grad_norm": 8.221780776977539,
      "learning_rate": 0.0,
      "loss": 1.1936,
      "step": 978
    },
    {
      "epoch": 0.08786573326153294,
      "grad_norm": 6.7104268074035645,
      "learning_rate": 0.0,
      "loss": 1.157,
      "step": 979
    },
    {
      "epoch": 0.08795548375516066,
      "grad_norm": 6.825798988342285,
      "learning_rate": 0.0,
      "loss": 1.1555,
      "step": 980
    },
    {
      "epoch": 0.08804523424878836,
      "grad_norm": 7.958093643188477,
      "learning_rate": 0.0,
      "loss": 1.2087,
      "step": 981
    },
    {
      "epoch": 0.08813498474241609,
      "grad_norm": 6.486677169799805,
      "learning_rate": 0.0,
      "loss": 1.1969,
      "step": 982
    },
    {
      "epoch": 0.0882247352360438,
      "grad_norm": 8.5619478225708,
      "learning_rate": 0.0,
      "loss": 1.2335,
      "step": 983
    },
    {
      "epoch": 0.08831448572967152,
      "grad_norm": 7.4012064933776855,
      "learning_rate": 0.0,
      "loss": 1.1909,
      "step": 984
    },
    {
      "epoch": 0.08840423622329922,
      "grad_norm": 8.785893440246582,
      "learning_rate": 0.0,
      "loss": 1.2308,
      "step": 985
    },
    {
      "epoch": 0.08849398671692695,
      "grad_norm": 8.705587387084961,
      "learning_rate": 0.0,
      "loss": 1.2288,
      "step": 986
    },
    {
      "epoch": 0.08858373721055465,
      "grad_norm": 8.409089088439941,
      "learning_rate": 0.0,
      "loss": 1.2164,
      "step": 987
    },
    {
      "epoch": 0.08867348770418237,
      "grad_norm": 7.804633140563965,
      "learning_rate": 0.0,
      "loss": 1.2088,
      "step": 988
    },
    {
      "epoch": 0.08876323819781008,
      "grad_norm": 8.13038444519043,
      "learning_rate": 0.0,
      "loss": 1.1775,
      "step": 989
    },
    {
      "epoch": 0.0888529886914378,
      "grad_norm": 8.071677207946777,
      "learning_rate": 0.0,
      "loss": 1.1818,
      "step": 990
    },
    {
      "epoch": 0.08894273918506551,
      "grad_norm": 9.226482391357422,
      "learning_rate": 0.0,
      "loss": 1.2042,
      "step": 991
    },
    {
      "epoch": 0.08903248967869323,
      "grad_norm": 8.742788314819336,
      "learning_rate": 0.0,
      "loss": 1.3114,
      "step": 992
    },
    {
      "epoch": 0.08912224017232095,
      "grad_norm": 7.577439308166504,
      "learning_rate": 0.0,
      "loss": 1.1627,
      "step": 993
    },
    {
      "epoch": 0.08921199066594866,
      "grad_norm": 29.25438117980957,
      "learning_rate": 0.0,
      "loss": 1.2896,
      "step": 994
    },
    {
      "epoch": 0.08930174115957638,
      "grad_norm": 7.112527847290039,
      "learning_rate": 0.0,
      "loss": 1.1919,
      "step": 995
    },
    {
      "epoch": 0.08939149165320409,
      "grad_norm": 13.069058418273926,
      "learning_rate": 0.0,
      "loss": 1.1538,
      "step": 996
    },
    {
      "epoch": 0.08948124214683181,
      "grad_norm": 7.554577827453613,
      "learning_rate": 0.0,
      "loss": 1.2123,
      "step": 997
    },
    {
      "epoch": 0.08957099264045952,
      "grad_norm": 6.4764180183410645,
      "learning_rate": 0.0,
      "loss": 1.2047,
      "step": 998
    },
    {
      "epoch": 0.08966074313408724,
      "grad_norm": 10.657317161560059,
      "learning_rate": 0.0,
      "loss": 1.2164,
      "step": 999
    },
    {
      "epoch": 0.08975049362771495,
      "grad_norm": 6.7847723960876465,
      "learning_rate": 0.0,
      "loss": 1.1721,
      "step": 1000
    },
    {
      "epoch": 0.08984024412134267,
      "grad_norm": 8.81544017791748,
      "learning_rate": 0.0,
      "loss": 1.204,
      "step": 1001
    },
    {
      "epoch": 0.08992999461497038,
      "grad_norm": 7.987589359283447,
      "learning_rate": 0.0,
      "loss": 1.2381,
      "step": 1002
    },
    {
      "epoch": 0.0900197451085981,
      "grad_norm": 8.345020294189453,
      "learning_rate": 0.0,
      "loss": 1.1364,
      "step": 1003
    },
    {
      "epoch": 0.09010949560222581,
      "grad_norm": 7.249290466308594,
      "learning_rate": 0.0,
      "loss": 1.2183,
      "step": 1004
    },
    {
      "epoch": 0.09019924609585353,
      "grad_norm": 7.609748840332031,
      "learning_rate": 0.0,
      "loss": 1.2016,
      "step": 1005
    },
    {
      "epoch": 0.09028899658948124,
      "grad_norm": 8.23070240020752,
      "learning_rate": 0.0,
      "loss": 1.2327,
      "step": 1006
    },
    {
      "epoch": 0.09037874708310896,
      "grad_norm": 7.701525688171387,
      "learning_rate": 0.0,
      "loss": 1.1684,
      "step": 1007
    },
    {
      "epoch": 0.09046849757673667,
      "grad_norm": 7.316336154937744,
      "learning_rate": 0.0,
      "loss": 1.2167,
      "step": 1008
    },
    {
      "epoch": 0.09055824807036439,
      "grad_norm": 6.755519866943359,
      "learning_rate": 0.0,
      "loss": 1.115,
      "step": 1009
    },
    {
      "epoch": 0.0906479985639921,
      "grad_norm": 7.2630109786987305,
      "learning_rate": 0.0,
      "loss": 1.2154,
      "step": 1010
    },
    {
      "epoch": 0.09073774905761982,
      "grad_norm": 7.091452598571777,
      "learning_rate": 0.0,
      "loss": 1.1428,
      "step": 1011
    },
    {
      "epoch": 0.09082749955124753,
      "grad_norm": 8.992631912231445,
      "learning_rate": 0.0,
      "loss": 1.1851,
      "step": 1012
    },
    {
      "epoch": 0.09091725004487525,
      "grad_norm": 9.187089920043945,
      "learning_rate": 0.0,
      "loss": 1.1913,
      "step": 1013
    },
    {
      "epoch": 0.09100700053850296,
      "grad_norm": 11.573824882507324,
      "learning_rate": 0.0,
      "loss": 1.221,
      "step": 1014
    },
    {
      "epoch": 0.09109675103213068,
      "grad_norm": 16.81496810913086,
      "learning_rate": 0.0,
      "loss": 1.2203,
      "step": 1015
    },
    {
      "epoch": 0.0911865015257584,
      "grad_norm": 20.906421661376953,
      "learning_rate": 0.0,
      "loss": 1.18,
      "step": 1016
    },
    {
      "epoch": 0.0912762520193861,
      "grad_norm": 7.72597599029541,
      "learning_rate": 0.0,
      "loss": 1.2332,
      "step": 1017
    },
    {
      "epoch": 0.09136600251301383,
      "grad_norm": 7.024073600769043,
      "learning_rate": 0.0,
      "loss": 1.1134,
      "step": 1018
    },
    {
      "epoch": 0.09145575300664154,
      "grad_norm": 7.716778755187988,
      "learning_rate": 0.0,
      "loss": 1.2028,
      "step": 1019
    },
    {
      "epoch": 0.09154550350026926,
      "grad_norm": 8.525825500488281,
      "learning_rate": 0.0,
      "loss": 1.1715,
      "step": 1020
    },
    {
      "epoch": 0.09163525399389696,
      "grad_norm": 39.13383865356445,
      "learning_rate": 0.0,
      "loss": 1.4517,
      "step": 1021
    },
    {
      "epoch": 0.09172500448752469,
      "grad_norm": 7.450038433074951,
      "learning_rate": 0.0,
      "loss": 1.2235,
      "step": 1022
    },
    {
      "epoch": 0.0918147549811524,
      "grad_norm": 10.805953025817871,
      "learning_rate": 0.0,
      "loss": 1.2698,
      "step": 1023
    },
    {
      "epoch": 0.09190450547478012,
      "grad_norm": 10.11421012878418,
      "learning_rate": 0.0,
      "loss": 1.1629,
      "step": 1024
    },
    {
      "epoch": 0.09199425596840782,
      "grad_norm": 7.650753498077393,
      "learning_rate": 0.0,
      "loss": 1.2342,
      "step": 1025
    },
    {
      "epoch": 0.09208400646203554,
      "grad_norm": 6.714000701904297,
      "learning_rate": 0.0,
      "loss": 1.2539,
      "step": 1026
    },
    {
      "epoch": 0.09217375695566325,
      "grad_norm": 8.242430686950684,
      "learning_rate": 0.0,
      "loss": 1.2458,
      "step": 1027
    },
    {
      "epoch": 0.09226350744929097,
      "grad_norm": 7.132680892944336,
      "learning_rate": 0.0,
      "loss": 1.1616,
      "step": 1028
    },
    {
      "epoch": 0.09235325794291868,
      "grad_norm": 7.191984176635742,
      "learning_rate": 0.0,
      "loss": 1.1639,
      "step": 1029
    },
    {
      "epoch": 0.0924430084365464,
      "grad_norm": 7.372185707092285,
      "learning_rate": 0.0,
      "loss": 1.1906,
      "step": 1030
    },
    {
      "epoch": 0.09253275893017411,
      "grad_norm": 8.40042781829834,
      "learning_rate": 0.0,
      "loss": 1.2246,
      "step": 1031
    },
    {
      "epoch": 0.09262250942380183,
      "grad_norm": 7.149620532989502,
      "learning_rate": 0.0,
      "loss": 1.1975,
      "step": 1032
    },
    {
      "epoch": 0.09271225991742954,
      "grad_norm": 6.077856063842773,
      "learning_rate": 0.0,
      "loss": 1.1908,
      "step": 1033
    },
    {
      "epoch": 0.09280201041105726,
      "grad_norm": 7.0159454345703125,
      "learning_rate": 0.0,
      "loss": 1.2235,
      "step": 1034
    },
    {
      "epoch": 0.09289176090468497,
      "grad_norm": 12.143881797790527,
      "learning_rate": 0.0,
      "loss": 1.1594,
      "step": 1035
    },
    {
      "epoch": 0.09298151139831269,
      "grad_norm": 8.499753952026367,
      "learning_rate": 0.0,
      "loss": 1.3009,
      "step": 1036
    },
    {
      "epoch": 0.0930712618919404,
      "grad_norm": 6.738633632659912,
      "learning_rate": 0.0,
      "loss": 1.1149,
      "step": 1037
    },
    {
      "epoch": 0.09316101238556812,
      "grad_norm": 28.070127487182617,
      "learning_rate": 0.0,
      "loss": 1.1523,
      "step": 1038
    },
    {
      "epoch": 0.09325076287919584,
      "grad_norm": 7.119271755218506,
      "learning_rate": 0.0,
      "loss": 1.2341,
      "step": 1039
    },
    {
      "epoch": 0.09334051337282355,
      "grad_norm": 7.0336151123046875,
      "learning_rate": 0.0,
      "loss": 1.267,
      "step": 1040
    },
    {
      "epoch": 0.09343026386645127,
      "grad_norm": 8.380191802978516,
      "learning_rate": 0.0,
      "loss": 1.2208,
      "step": 1041
    },
    {
      "epoch": 0.09352001436007898,
      "grad_norm": 7.190478801727295,
      "learning_rate": 0.0,
      "loss": 1.1849,
      "step": 1042
    },
    {
      "epoch": 0.0936097648537067,
      "grad_norm": 8.558664321899414,
      "learning_rate": 0.0,
      "loss": 1.2074,
      "step": 1043
    },
    {
      "epoch": 0.09369951534733441,
      "grad_norm": 14.82476806640625,
      "learning_rate": 0.0,
      "loss": 1.1731,
      "step": 1044
    },
    {
      "epoch": 0.09378926584096213,
      "grad_norm": 9.171021461486816,
      "learning_rate": 0.0,
      "loss": 1.2699,
      "step": 1045
    },
    {
      "epoch": 0.09387901633458984,
      "grad_norm": 7.656778335571289,
      "learning_rate": 0.0,
      "loss": 1.1649,
      "step": 1046
    },
    {
      "epoch": 0.09396876682821756,
      "grad_norm": 7.704495429992676,
      "learning_rate": 0.0,
      "loss": 1.1294,
      "step": 1047
    },
    {
      "epoch": 0.09405851732184527,
      "grad_norm": 7.554352760314941,
      "learning_rate": 0.0,
      "loss": 1.2205,
      "step": 1048
    },
    {
      "epoch": 0.09414826781547299,
      "grad_norm": 13.986176490783691,
      "learning_rate": 0.0,
      "loss": 1.3016,
      "step": 1049
    },
    {
      "epoch": 0.0942380183091007,
      "grad_norm": 6.887753486633301,
      "learning_rate": 0.0,
      "loss": 1.1321,
      "step": 1050
    },
    {
      "epoch": 0.09432776880272842,
      "grad_norm": 7.105311870574951,
      "learning_rate": 0.0,
      "loss": 1.1954,
      "step": 1051
    },
    {
      "epoch": 0.09441751929635613,
      "grad_norm": 6.423686981201172,
      "learning_rate": 0.0,
      "loss": 1.1674,
      "step": 1052
    },
    {
      "epoch": 0.09450726978998385,
      "grad_norm": 5.860947608947754,
      "learning_rate": 0.0,
      "loss": 1.1527,
      "step": 1053
    },
    {
      "epoch": 0.09459702028361155,
      "grad_norm": 7.901552677154541,
      "learning_rate": 0.0,
      "loss": 1.1682,
      "step": 1054
    },
    {
      "epoch": 0.09468677077723928,
      "grad_norm": 6.713497161865234,
      "learning_rate": 0.0,
      "loss": 1.2257,
      "step": 1055
    },
    {
      "epoch": 0.09477652127086698,
      "grad_norm": 10.425678253173828,
      "learning_rate": 0.0,
      "loss": 1.3025,
      "step": 1056
    },
    {
      "epoch": 0.0948662717644947,
      "grad_norm": 8.308570861816406,
      "learning_rate": 0.0,
      "loss": 1.1788,
      "step": 1057
    },
    {
      "epoch": 0.09495602225812241,
      "grad_norm": 7.5866265296936035,
      "learning_rate": 0.0,
      "loss": 1.1704,
      "step": 1058
    },
    {
      "epoch": 0.09504577275175013,
      "grad_norm": 8.22181224822998,
      "learning_rate": 0.0,
      "loss": 1.2032,
      "step": 1059
    },
    {
      "epoch": 0.09513552324537784,
      "grad_norm": 6.028190612792969,
      "learning_rate": 0.0,
      "loss": 1.1864,
      "step": 1060
    },
    {
      "epoch": 0.09522527373900556,
      "grad_norm": 7.760622501373291,
      "learning_rate": 0.0,
      "loss": 1.2379,
      "step": 1061
    },
    {
      "epoch": 0.09531502423263329,
      "grad_norm": 7.838149070739746,
      "learning_rate": 0.0,
      "loss": 1.2213,
      "step": 1062
    },
    {
      "epoch": 0.095404774726261,
      "grad_norm": 7.923225402832031,
      "learning_rate": 0.0,
      "loss": 1.1719,
      "step": 1063
    },
    {
      "epoch": 0.09549452521988872,
      "grad_norm": 6.666560173034668,
      "learning_rate": 0.0,
      "loss": 1.2167,
      "step": 1064
    },
    {
      "epoch": 0.09558427571351642,
      "grad_norm": 6.945178031921387,
      "learning_rate": 0.0,
      "loss": 1.2086,
      "step": 1065
    },
    {
      "epoch": 0.09567402620714414,
      "grad_norm": 11.147552490234375,
      "learning_rate": 0.0,
      "loss": 1.2249,
      "step": 1066
    },
    {
      "epoch": 0.09576377670077185,
      "grad_norm": 7.183469295501709,
      "learning_rate": 0.0,
      "loss": 1.1911,
      "step": 1067
    },
    {
      "epoch": 0.09585352719439957,
      "grad_norm": 7.347174644470215,
      "learning_rate": 0.0,
      "loss": 1.185,
      "step": 1068
    },
    {
      "epoch": 0.09594327768802728,
      "grad_norm": 7.885452747344971,
      "learning_rate": 0.0,
      "loss": 1.2111,
      "step": 1069
    },
    {
      "epoch": 0.096033028181655,
      "grad_norm": 9.465923309326172,
      "learning_rate": 0.0,
      "loss": 1.184,
      "step": 1070
    },
    {
      "epoch": 0.09612277867528271,
      "grad_norm": 9.962908744812012,
      "learning_rate": 0.0,
      "loss": 1.2218,
      "step": 1071
    },
    {
      "epoch": 0.09621252916891043,
      "grad_norm": 7.010715007781982,
      "learning_rate": 0.0,
      "loss": 1.268,
      "step": 1072
    },
    {
      "epoch": 0.09630227966253814,
      "grad_norm": 7.461204528808594,
      "learning_rate": 0.0,
      "loss": 1.2095,
      "step": 1073
    },
    {
      "epoch": 0.09639203015616586,
      "grad_norm": 9.385540008544922,
      "learning_rate": 0.0,
      "loss": 1.2301,
      "step": 1074
    },
    {
      "epoch": 0.09648178064979357,
      "grad_norm": 7.466344356536865,
      "learning_rate": 0.0,
      "loss": 1.3218,
      "step": 1075
    },
    {
      "epoch": 0.09657153114342129,
      "grad_norm": 16.121244430541992,
      "learning_rate": 0.0,
      "loss": 1.2441,
      "step": 1076
    },
    {
      "epoch": 0.096661281637049,
      "grad_norm": 11.131217956542969,
      "learning_rate": 0.0,
      "loss": 1.2057,
      "step": 1077
    },
    {
      "epoch": 0.09675103213067672,
      "grad_norm": 7.170916557312012,
      "learning_rate": 0.0,
      "loss": 1.2543,
      "step": 1078
    },
    {
      "epoch": 0.09684078262430443,
      "grad_norm": 7.106405258178711,
      "learning_rate": 0.0,
      "loss": 1.2126,
      "step": 1079
    },
    {
      "epoch": 0.09693053311793215,
      "grad_norm": 8.156275749206543,
      "learning_rate": 0.0,
      "loss": 1.128,
      "step": 1080
    },
    {
      "epoch": 0.09702028361155986,
      "grad_norm": 18.32164764404297,
      "learning_rate": 0.0,
      "loss": 1.1921,
      "step": 1081
    },
    {
      "epoch": 0.09711003410518758,
      "grad_norm": 7.691723823547363,
      "learning_rate": 0.0,
      "loss": 1.2137,
      "step": 1082
    },
    {
      "epoch": 0.0971997845988153,
      "grad_norm": 7.61482048034668,
      "learning_rate": 0.0,
      "loss": 1.087,
      "step": 1083
    },
    {
      "epoch": 0.09728953509244301,
      "grad_norm": 9.740516662597656,
      "learning_rate": 0.0,
      "loss": 1.1985,
      "step": 1084
    },
    {
      "epoch": 0.09737928558607073,
      "grad_norm": 6.97592830657959,
      "learning_rate": 0.0,
      "loss": 1.1608,
      "step": 1085
    },
    {
      "epoch": 0.09746903607969844,
      "grad_norm": 7.257114887237549,
      "learning_rate": 0.0,
      "loss": 1.1958,
      "step": 1086
    },
    {
      "epoch": 0.09755878657332616,
      "grad_norm": 6.802114486694336,
      "learning_rate": 0.0,
      "loss": 1.2661,
      "step": 1087
    },
    {
      "epoch": 0.09764853706695387,
      "grad_norm": 6.79412841796875,
      "learning_rate": 0.0,
      "loss": 1.1666,
      "step": 1088
    },
    {
      "epoch": 0.09773828756058159,
      "grad_norm": 8.418318748474121,
      "learning_rate": 0.0,
      "loss": 1.1478,
      "step": 1089
    },
    {
      "epoch": 0.0978280380542093,
      "grad_norm": 11.184213638305664,
      "learning_rate": 0.0,
      "loss": 1.241,
      "step": 1090
    },
    {
      "epoch": 0.09791778854783702,
      "grad_norm": 7.953930377960205,
      "learning_rate": 0.0,
      "loss": 1.1326,
      "step": 1091
    },
    {
      "epoch": 0.09800753904146473,
      "grad_norm": 6.688270092010498,
      "learning_rate": 0.0,
      "loss": 1.1922,
      "step": 1092
    },
    {
      "epoch": 0.09809728953509245,
      "grad_norm": 9.664051055908203,
      "learning_rate": 0.0,
      "loss": 1.1816,
      "step": 1093
    },
    {
      "epoch": 0.09818704002872015,
      "grad_norm": 6.781725883483887,
      "learning_rate": 0.0,
      "loss": 1.1621,
      "step": 1094
    },
    {
      "epoch": 0.09827679052234788,
      "grad_norm": 8.232401847839355,
      "learning_rate": 0.0,
      "loss": 1.1617,
      "step": 1095
    },
    {
      "epoch": 0.09836654101597558,
      "grad_norm": 14.146578788757324,
      "learning_rate": 0.0,
      "loss": 1.2535,
      "step": 1096
    },
    {
      "epoch": 0.0984562915096033,
      "grad_norm": 5.7434916496276855,
      "learning_rate": 0.0,
      "loss": 1.0956,
      "step": 1097
    },
    {
      "epoch": 0.09854604200323101,
      "grad_norm": 15.86760139465332,
      "learning_rate": 0.0,
      "loss": 1.2137,
      "step": 1098
    },
    {
      "epoch": 0.09863579249685873,
      "grad_norm": 6.815981388092041,
      "learning_rate": 0.0,
      "loss": 1.1681,
      "step": 1099
    },
    {
      "epoch": 0.09872554299048644,
      "grad_norm": 7.5769429206848145,
      "learning_rate": 0.0,
      "loss": 1.2042,
      "step": 1100
    },
    {
      "epoch": 0.09881529348411416,
      "grad_norm": 7.806396961212158,
      "learning_rate": 0.0,
      "loss": 1.2259,
      "step": 1101
    },
    {
      "epoch": 0.09890504397774187,
      "grad_norm": 9.628486633300781,
      "learning_rate": 0.0,
      "loss": 1.1425,
      "step": 1102
    },
    {
      "epoch": 0.0989947944713696,
      "grad_norm": 6.4671244621276855,
      "learning_rate": 0.0,
      "loss": 1.1468,
      "step": 1103
    },
    {
      "epoch": 0.0990845449649973,
      "grad_norm": 9.074005126953125,
      "learning_rate": 0.0,
      "loss": 1.1791,
      "step": 1104
    },
    {
      "epoch": 0.09917429545862502,
      "grad_norm": 6.762948036193848,
      "learning_rate": 0.0,
      "loss": 1.1353,
      "step": 1105
    },
    {
      "epoch": 0.09926404595225274,
      "grad_norm": 7.4968342781066895,
      "learning_rate": 0.0,
      "loss": 1.3087,
      "step": 1106
    },
    {
      "epoch": 0.09935379644588045,
      "grad_norm": 8.800630569458008,
      "learning_rate": 0.0,
      "loss": 1.2166,
      "step": 1107
    },
    {
      "epoch": 0.09944354693950817,
      "grad_norm": 6.234934329986572,
      "learning_rate": 0.0,
      "loss": 1.1579,
      "step": 1108
    },
    {
      "epoch": 0.09953329743313588,
      "grad_norm": 7.589239120483398,
      "learning_rate": 0.0,
      "loss": 1.2964,
      "step": 1109
    },
    {
      "epoch": 0.0996230479267636,
      "grad_norm": 8.383721351623535,
      "learning_rate": 0.0,
      "loss": 1.2263,
      "step": 1110
    },
    {
      "epoch": 0.09971279842039131,
      "grad_norm": 7.533040523529053,
      "learning_rate": 0.0,
      "loss": 1.1713,
      "step": 1111
    },
    {
      "epoch": 0.09980254891401903,
      "grad_norm": 7.470641136169434,
      "learning_rate": 0.0,
      "loss": 1.1988,
      "step": 1112
    },
    {
      "epoch": 0.09989229940764674,
      "grad_norm": 7.397196292877197,
      "learning_rate": 0.0,
      "loss": 1.3076,
      "step": 1113
    },
    {
      "epoch": 0.09998204990127446,
      "grad_norm": 7.4580488204956055,
      "learning_rate": 0.0,
      "loss": 1.2523,
      "step": 1114
    },
    {
      "epoch": 0.10007180039490217,
      "grad_norm": 11.449862480163574,
      "learning_rate": 0.0,
      "loss": 1.1994,
      "step": 1115
    },
    {
      "epoch": 0.10016155088852989,
      "grad_norm": 7.228401184082031,
      "learning_rate": 0.0,
      "loss": 1.1936,
      "step": 1116
    },
    {
      "epoch": 0.1002513013821576,
      "grad_norm": 7.141993999481201,
      "learning_rate": 0.0,
      "loss": 1.1697,
      "step": 1117
    },
    {
      "epoch": 0.10034105187578532,
      "grad_norm": 7.64440393447876,
      "learning_rate": 0.0,
      "loss": 1.1562,
      "step": 1118
    },
    {
      "epoch": 0.10043080236941303,
      "grad_norm": 7.4717607498168945,
      "learning_rate": 0.0,
      "loss": 1.1821,
      "step": 1119
    },
    {
      "epoch": 0.10052055286304075,
      "grad_norm": 6.6784467697143555,
      "learning_rate": 0.0,
      "loss": 1.2241,
      "step": 1120
    },
    {
      "epoch": 0.10061030335666846,
      "grad_norm": 9.989487648010254,
      "learning_rate": 0.0,
      "loss": 1.1551,
      "step": 1121
    },
    {
      "epoch": 0.10070005385029618,
      "grad_norm": 8.639439582824707,
      "learning_rate": 0.0,
      "loss": 1.1477,
      "step": 1122
    },
    {
      "epoch": 0.10078980434392389,
      "grad_norm": 9.454732894897461,
      "learning_rate": 0.0,
      "loss": 1.2101,
      "step": 1123
    },
    {
      "epoch": 0.10087955483755161,
      "grad_norm": 12.935420036315918,
      "learning_rate": 0.0,
      "loss": 1.1519,
      "step": 1124
    },
    {
      "epoch": 0.10096930533117932,
      "grad_norm": 9.94426155090332,
      "learning_rate": 0.0,
      "loss": 1.2232,
      "step": 1125
    },
    {
      "epoch": 0.10105905582480704,
      "grad_norm": 7.8799147605896,
      "learning_rate": 0.0,
      "loss": 1.1866,
      "step": 1126
    },
    {
      "epoch": 0.10114880631843474,
      "grad_norm": 9.740560531616211,
      "learning_rate": 0.0,
      "loss": 1.335,
      "step": 1127
    },
    {
      "epoch": 0.10123855681206247,
      "grad_norm": 7.76869010925293,
      "learning_rate": 0.0,
      "loss": 1.1101,
      "step": 1128
    },
    {
      "epoch": 0.10132830730569019,
      "grad_norm": 9.693552017211914,
      "learning_rate": 0.0,
      "loss": 1.1944,
      "step": 1129
    },
    {
      "epoch": 0.1014180577993179,
      "grad_norm": 6.828721523284912,
      "learning_rate": 0.0,
      "loss": 1.1681,
      "step": 1130
    },
    {
      "epoch": 0.10150780829294562,
      "grad_norm": 7.398787021636963,
      "learning_rate": 0.0,
      "loss": 1.2466,
      "step": 1131
    },
    {
      "epoch": 0.10159755878657332,
      "grad_norm": 9.31094741821289,
      "learning_rate": 0.0,
      "loss": 1.1516,
      "step": 1132
    },
    {
      "epoch": 0.10168730928020105,
      "grad_norm": 6.057366371154785,
      "learning_rate": 0.0,
      "loss": 1.145,
      "step": 1133
    },
    {
      "epoch": 0.10177705977382875,
      "grad_norm": 13.921158790588379,
      "learning_rate": 0.0,
      "loss": 1.1419,
      "step": 1134
    },
    {
      "epoch": 0.10186681026745648,
      "grad_norm": 7.978126049041748,
      "learning_rate": 0.0,
      "loss": 1.223,
      "step": 1135
    },
    {
      "epoch": 0.10195656076108418,
      "grad_norm": 8.267828941345215,
      "learning_rate": 0.0,
      "loss": 1.1787,
      "step": 1136
    },
    {
      "epoch": 0.1020463112547119,
      "grad_norm": 11.160860061645508,
      "learning_rate": 0.0,
      "loss": 1.1678,
      "step": 1137
    },
    {
      "epoch": 0.10213606174833961,
      "grad_norm": 13.161173820495605,
      "learning_rate": 0.0,
      "loss": 1.2982,
      "step": 1138
    },
    {
      "epoch": 0.10222581224196733,
      "grad_norm": 12.908098220825195,
      "learning_rate": 0.0,
      "loss": 1.1194,
      "step": 1139
    },
    {
      "epoch": 0.10231556273559504,
      "grad_norm": 41.393165588378906,
      "learning_rate": 0.0,
      "loss": 1.2687,
      "step": 1140
    },
    {
      "epoch": 0.10240531322922276,
      "grad_norm": 7.518251419067383,
      "learning_rate": 0.0,
      "loss": 1.2058,
      "step": 1141
    },
    {
      "epoch": 0.10249506372285047,
      "grad_norm": 8.484843254089355,
      "learning_rate": 0.0,
      "loss": 1.2453,
      "step": 1142
    },
    {
      "epoch": 0.10258481421647819,
      "grad_norm": 7.06842565536499,
      "learning_rate": 0.0,
      "loss": 1.2475,
      "step": 1143
    },
    {
      "epoch": 0.1026745647101059,
      "grad_norm": 8.05817985534668,
      "learning_rate": 0.0,
      "loss": 1.1546,
      "step": 1144
    },
    {
      "epoch": 0.10276431520373362,
      "grad_norm": 9.937104225158691,
      "learning_rate": 0.0,
      "loss": 1.2235,
      "step": 1145
    },
    {
      "epoch": 0.10285406569736133,
      "grad_norm": 7.52036714553833,
      "learning_rate": 0.0,
      "loss": 1.2378,
      "step": 1146
    },
    {
      "epoch": 0.10294381619098905,
      "grad_norm": 7.523068428039551,
      "learning_rate": 0.0,
      "loss": 1.1963,
      "step": 1147
    },
    {
      "epoch": 0.10303356668461676,
      "grad_norm": 9.28679084777832,
      "learning_rate": 0.0,
      "loss": 1.1909,
      "step": 1148
    },
    {
      "epoch": 0.10312331717824448,
      "grad_norm": 6.317113876342773,
      "learning_rate": 0.0,
      "loss": 1.2038,
      "step": 1149
    },
    {
      "epoch": 0.10321306767187219,
      "grad_norm": 9.567991256713867,
      "learning_rate": 0.0,
      "loss": 1.1881,
      "step": 1150
    },
    {
      "epoch": 0.10330281816549991,
      "grad_norm": 7.057479381561279,
      "learning_rate": 0.0,
      "loss": 1.1755,
      "step": 1151
    },
    {
      "epoch": 0.10339256865912763,
      "grad_norm": 7.312747955322266,
      "learning_rate": 0.0,
      "loss": 1.1891,
      "step": 1152
    },
    {
      "epoch": 0.10348231915275534,
      "grad_norm": 8.523305892944336,
      "learning_rate": 0.0,
      "loss": 1.2635,
      "step": 1153
    },
    {
      "epoch": 0.10357206964638306,
      "grad_norm": 6.951488971710205,
      "learning_rate": 0.0,
      "loss": 1.175,
      "step": 1154
    },
    {
      "epoch": 0.10366182014001077,
      "grad_norm": 10.06381893157959,
      "learning_rate": 0.0,
      "loss": 1.2014,
      "step": 1155
    },
    {
      "epoch": 0.10375157063363849,
      "grad_norm": 29.39043426513672,
      "learning_rate": 0.0,
      "loss": 1.2198,
      "step": 1156
    },
    {
      "epoch": 0.1038413211272662,
      "grad_norm": 6.807641983032227,
      "learning_rate": 0.0,
      "loss": 1.1332,
      "step": 1157
    },
    {
      "epoch": 0.10393107162089392,
      "grad_norm": 6.1219353675842285,
      "learning_rate": 0.0,
      "loss": 1.1704,
      "step": 1158
    },
    {
      "epoch": 0.10402082211452163,
      "grad_norm": 6.646277904510498,
      "learning_rate": 0.0,
      "loss": 1.1765,
      "step": 1159
    },
    {
      "epoch": 0.10411057260814935,
      "grad_norm": 7.604135990142822,
      "learning_rate": 0.0,
      "loss": 1.1831,
      "step": 1160
    },
    {
      "epoch": 0.10420032310177706,
      "grad_norm": 6.851521968841553,
      "learning_rate": 0.0,
      "loss": 1.137,
      "step": 1161
    },
    {
      "epoch": 0.10429007359540478,
      "grad_norm": 7.401684284210205,
      "learning_rate": 0.0,
      "loss": 1.2448,
      "step": 1162
    },
    {
      "epoch": 0.10437982408903249,
      "grad_norm": 9.533721923828125,
      "learning_rate": 0.0,
      "loss": 1.1765,
      "step": 1163
    },
    {
      "epoch": 0.10446957458266021,
      "grad_norm": 12.370539665222168,
      "learning_rate": 0.0,
      "loss": 1.261,
      "step": 1164
    },
    {
      "epoch": 0.10455932507628792,
      "grad_norm": 8.27772045135498,
      "learning_rate": 0.0,
      "loss": 1.2678,
      "step": 1165
    },
    {
      "epoch": 0.10464907556991564,
      "grad_norm": 6.8677473068237305,
      "learning_rate": 0.0,
      "loss": 1.2283,
      "step": 1166
    },
    {
      "epoch": 0.10473882606354334,
      "grad_norm": 6.228326797485352,
      "learning_rate": 0.0,
      "loss": 1.1782,
      "step": 1167
    },
    {
      "epoch": 0.10482857655717107,
      "grad_norm": 42.65775680541992,
      "learning_rate": 0.0,
      "loss": 1.2156,
      "step": 1168
    },
    {
      "epoch": 0.10491832705079877,
      "grad_norm": 6.6877336502075195,
      "learning_rate": 0.0,
      "loss": 1.1231,
      "step": 1169
    },
    {
      "epoch": 0.1050080775444265,
      "grad_norm": 7.48610782623291,
      "learning_rate": 0.0,
      "loss": 1.2343,
      "step": 1170
    },
    {
      "epoch": 0.1050978280380542,
      "grad_norm": 7.622134685516357,
      "learning_rate": 0.0,
      "loss": 1.1842,
      "step": 1171
    },
    {
      "epoch": 0.10518757853168192,
      "grad_norm": 12.213566780090332,
      "learning_rate": 0.0,
      "loss": 1.1795,
      "step": 1172
    },
    {
      "epoch": 0.10527732902530965,
      "grad_norm": 14.935593605041504,
      "learning_rate": 0.0,
      "loss": 1.1833,
      "step": 1173
    },
    {
      "epoch": 0.10536707951893735,
      "grad_norm": 10.504748344421387,
      "learning_rate": 0.0,
      "loss": 1.2031,
      "step": 1174
    },
    {
      "epoch": 0.10545683001256508,
      "grad_norm": 9.865992546081543,
      "learning_rate": 0.0,
      "loss": 1.2949,
      "step": 1175
    },
    {
      "epoch": 0.10554658050619278,
      "grad_norm": 7.784281253814697,
      "learning_rate": 0.0,
      "loss": 1.175,
      "step": 1176
    },
    {
      "epoch": 0.1056363309998205,
      "grad_norm": 7.295712947845459,
      "learning_rate": 0.0,
      "loss": 1.1724,
      "step": 1177
    },
    {
      "epoch": 0.10572608149344821,
      "grad_norm": 20.28819465637207,
      "learning_rate": 0.0,
      "loss": 1.1864,
      "step": 1178
    },
    {
      "epoch": 0.10581583198707593,
      "grad_norm": 7.999528408050537,
      "learning_rate": 0.0,
      "loss": 1.2233,
      "step": 1179
    },
    {
      "epoch": 0.10590558248070364,
      "grad_norm": 7.084168910980225,
      "learning_rate": 0.0,
      "loss": 1.1653,
      "step": 1180
    },
    {
      "epoch": 0.10599533297433136,
      "grad_norm": 10.972064018249512,
      "learning_rate": 0.0,
      "loss": 1.1388,
      "step": 1181
    },
    {
      "epoch": 0.10608508346795907,
      "grad_norm": 27.590476989746094,
      "learning_rate": 0.0,
      "loss": 1.1839,
      "step": 1182
    },
    {
      "epoch": 0.10617483396158679,
      "grad_norm": 6.125041961669922,
      "learning_rate": 0.0,
      "loss": 1.1927,
      "step": 1183
    },
    {
      "epoch": 0.1062645844552145,
      "grad_norm": 9.475578308105469,
      "learning_rate": 0.0,
      "loss": 1.2921,
      "step": 1184
    },
    {
      "epoch": 0.10635433494884222,
      "grad_norm": 9.484760284423828,
      "learning_rate": 0.0,
      "loss": 1.1845,
      "step": 1185
    },
    {
      "epoch": 0.10644408544246993,
      "grad_norm": 7.627954959869385,
      "learning_rate": 0.0,
      "loss": 1.1367,
      "step": 1186
    },
    {
      "epoch": 0.10653383593609765,
      "grad_norm": 6.552107810974121,
      "learning_rate": 0.0,
      "loss": 1.1818,
      "step": 1187
    },
    {
      "epoch": 0.10662358642972536,
      "grad_norm": 7.146627426147461,
      "learning_rate": 0.0,
      "loss": 1.1986,
      "step": 1188
    },
    {
      "epoch": 0.10671333692335308,
      "grad_norm": 8.372563362121582,
      "learning_rate": 0.0,
      "loss": 1.2938,
      "step": 1189
    },
    {
      "epoch": 0.10680308741698079,
      "grad_norm": 11.690475463867188,
      "learning_rate": 0.0,
      "loss": 1.205,
      "step": 1190
    },
    {
      "epoch": 0.10689283791060851,
      "grad_norm": 6.831663131713867,
      "learning_rate": 0.0,
      "loss": 1.2051,
      "step": 1191
    },
    {
      "epoch": 0.10698258840423622,
      "grad_norm": 10.426613807678223,
      "learning_rate": 0.0,
      "loss": 1.1605,
      "step": 1192
    },
    {
      "epoch": 0.10707233889786394,
      "grad_norm": 7.646509647369385,
      "learning_rate": 0.0,
      "loss": 1.197,
      "step": 1193
    },
    {
      "epoch": 0.10716208939149165,
      "grad_norm": 10.784693717956543,
      "learning_rate": 0.0,
      "loss": 1.123,
      "step": 1194
    },
    {
      "epoch": 0.10725183988511937,
      "grad_norm": 6.83977746963501,
      "learning_rate": 0.0,
      "loss": 1.165,
      "step": 1195
    },
    {
      "epoch": 0.10734159037874709,
      "grad_norm": 8.040549278259277,
      "learning_rate": 0.0,
      "loss": 1.2104,
      "step": 1196
    },
    {
      "epoch": 0.1074313408723748,
      "grad_norm": 7.9545698165893555,
      "learning_rate": 0.0,
      "loss": 1.1783,
      "step": 1197
    },
    {
      "epoch": 0.10752109136600252,
      "grad_norm": 15.189764022827148,
      "learning_rate": 0.0,
      "loss": 1.1546,
      "step": 1198
    },
    {
      "epoch": 0.10761084185963023,
      "grad_norm": 15.712651252746582,
      "learning_rate": 0.0,
      "loss": 1.2264,
      "step": 1199
    },
    {
      "epoch": 0.10770059235325795,
      "grad_norm": 7.610531806945801,
      "learning_rate": 0.0,
      "loss": 1.1805,
      "step": 1200
    },
    {
      "epoch": 0.10779034284688566,
      "grad_norm": 7.907209396362305,
      "learning_rate": 0.0,
      "loss": 1.1881,
      "step": 1201
    },
    {
      "epoch": 0.10788009334051338,
      "grad_norm": 7.268981456756592,
      "learning_rate": 0.0,
      "loss": 1.1872,
      "step": 1202
    },
    {
      "epoch": 0.10796984383414109,
      "grad_norm": 14.515198707580566,
      "learning_rate": 0.0,
      "loss": 1.1266,
      "step": 1203
    },
    {
      "epoch": 0.10805959432776881,
      "grad_norm": 7.6044721603393555,
      "learning_rate": 0.0,
      "loss": 1.2831,
      "step": 1204
    },
    {
      "epoch": 0.10814934482139651,
      "grad_norm": 10.502046585083008,
      "learning_rate": 0.0,
      "loss": 1.168,
      "step": 1205
    },
    {
      "epoch": 0.10823909531502424,
      "grad_norm": 7.369009494781494,
      "learning_rate": 0.0,
      "loss": 1.1684,
      "step": 1206
    },
    {
      "epoch": 0.10832884580865194,
      "grad_norm": 9.222478866577148,
      "learning_rate": 0.0,
      "loss": 1.1925,
      "step": 1207
    },
    {
      "epoch": 0.10841859630227967,
      "grad_norm": 13.179414749145508,
      "learning_rate": 0.0,
      "loss": 1.2306,
      "step": 1208
    },
    {
      "epoch": 0.10850834679590737,
      "grad_norm": 5.9077277183532715,
      "learning_rate": 0.0,
      "loss": 1.1929,
      "step": 1209
    },
    {
      "epoch": 0.1085980972895351,
      "grad_norm": 7.279188632965088,
      "learning_rate": 0.0,
      "loss": 1.2295,
      "step": 1210
    },
    {
      "epoch": 0.1086878477831628,
      "grad_norm": 14.352042198181152,
      "learning_rate": 0.0,
      "loss": 1.2297,
      "step": 1211
    },
    {
      "epoch": 0.10877759827679052,
      "grad_norm": 6.4306135177612305,
      "learning_rate": 0.0,
      "loss": 1.1486,
      "step": 1212
    },
    {
      "epoch": 0.10886734877041823,
      "grad_norm": 7.60858154296875,
      "learning_rate": 0.0,
      "loss": 1.1665,
      "step": 1213
    },
    {
      "epoch": 0.10895709926404595,
      "grad_norm": 8.519367218017578,
      "learning_rate": 0.0,
      "loss": 1.1571,
      "step": 1214
    },
    {
      "epoch": 0.10904684975767366,
      "grad_norm": 6.887984275817871,
      "learning_rate": 0.0,
      "loss": 1.2868,
      "step": 1215
    },
    {
      "epoch": 0.10913660025130138,
      "grad_norm": 8.734406471252441,
      "learning_rate": 0.0,
      "loss": 1.1681,
      "step": 1216
    },
    {
      "epoch": 0.10922635074492909,
      "grad_norm": 9.151483535766602,
      "learning_rate": 0.0,
      "loss": 1.1644,
      "step": 1217
    },
    {
      "epoch": 0.10931610123855681,
      "grad_norm": 28.47056770324707,
      "learning_rate": 0.0,
      "loss": 1.3091,
      "step": 1218
    },
    {
      "epoch": 0.10940585173218453,
      "grad_norm": 11.348506927490234,
      "learning_rate": 0.0,
      "loss": 1.188,
      "step": 1219
    },
    {
      "epoch": 0.10949560222581224,
      "grad_norm": 6.243442535400391,
      "learning_rate": 0.0,
      "loss": 1.2218,
      "step": 1220
    },
    {
      "epoch": 0.10958535271943996,
      "grad_norm": 7.549234867095947,
      "learning_rate": 0.0,
      "loss": 1.1617,
      "step": 1221
    },
    {
      "epoch": 0.10967510321306767,
      "grad_norm": 8.665948867797852,
      "learning_rate": 0.0,
      "loss": 1.1903,
      "step": 1222
    },
    {
      "epoch": 0.10976485370669539,
      "grad_norm": 9.804648399353027,
      "learning_rate": 0.0,
      "loss": 1.2067,
      "step": 1223
    },
    {
      "epoch": 0.1098546042003231,
      "grad_norm": 6.668400764465332,
      "learning_rate": 0.0,
      "loss": 1.1248,
      "step": 1224
    },
    {
      "epoch": 0.10994435469395082,
      "grad_norm": 7.75205135345459,
      "learning_rate": 0.0,
      "loss": 1.081,
      "step": 1225
    },
    {
      "epoch": 0.11003410518757853,
      "grad_norm": 7.177562236785889,
      "learning_rate": 0.0,
      "loss": 1.2148,
      "step": 1226
    },
    {
      "epoch": 0.11012385568120625,
      "grad_norm": 7.418159008026123,
      "learning_rate": 0.0,
      "loss": 1.2114,
      "step": 1227
    },
    {
      "epoch": 0.11021360617483396,
      "grad_norm": 7.565596103668213,
      "learning_rate": 0.0,
      "loss": 1.2769,
      "step": 1228
    },
    {
      "epoch": 0.11030335666846168,
      "grad_norm": 9.702741622924805,
      "learning_rate": 0.0,
      "loss": 1.1807,
      "step": 1229
    },
    {
      "epoch": 0.11039310716208939,
      "grad_norm": 7.837614059448242,
      "learning_rate": 0.0,
      "loss": 1.2898,
      "step": 1230
    },
    {
      "epoch": 0.11048285765571711,
      "grad_norm": 7.053057670593262,
      "learning_rate": 0.0,
      "loss": 1.1927,
      "step": 1231
    },
    {
      "epoch": 0.11057260814934482,
      "grad_norm": 9.62713623046875,
      "learning_rate": 0.0,
      "loss": 1.1858,
      "step": 1232
    },
    {
      "epoch": 0.11066235864297254,
      "grad_norm": 8.093538284301758,
      "learning_rate": 0.0,
      "loss": 1.2785,
      "step": 1233
    },
    {
      "epoch": 0.11075210913660025,
      "grad_norm": 6.607880592346191,
      "learning_rate": 0.0,
      "loss": 1.1747,
      "step": 1234
    },
    {
      "epoch": 0.11084185963022797,
      "grad_norm": 7.070347309112549,
      "learning_rate": 0.0,
      "loss": 1.1497,
      "step": 1235
    },
    {
      "epoch": 0.11093161012385568,
      "grad_norm": 7.810986518859863,
      "learning_rate": 0.0,
      "loss": 1.265,
      "step": 1236
    },
    {
      "epoch": 0.1110213606174834,
      "grad_norm": 7.272270202636719,
      "learning_rate": 0.0,
      "loss": 1.1991,
      "step": 1237
    },
    {
      "epoch": 0.1111111111111111,
      "grad_norm": 10.324978828430176,
      "learning_rate": 0.0,
      "loss": 1.1962,
      "step": 1238
    },
    {
      "epoch": 0.11120086160473883,
      "grad_norm": 8.946743965148926,
      "learning_rate": 0.0,
      "loss": 1.2623,
      "step": 1239
    },
    {
      "epoch": 0.11129061209836653,
      "grad_norm": 18.4953670501709,
      "learning_rate": 0.0,
      "loss": 1.1854,
      "step": 1240
    },
    {
      "epoch": 0.11138036259199426,
      "grad_norm": 8.365201950073242,
      "learning_rate": 0.0,
      "loss": 1.2416,
      "step": 1241
    },
    {
      "epoch": 0.11147011308562198,
      "grad_norm": 7.022207260131836,
      "learning_rate": 0.0,
      "loss": 1.1749,
      "step": 1242
    },
    {
      "epoch": 0.11155986357924969,
      "grad_norm": 6.526654243469238,
      "learning_rate": 0.0,
      "loss": 1.227,
      "step": 1243
    },
    {
      "epoch": 0.1116496140728774,
      "grad_norm": 28.715490341186523,
      "learning_rate": 0.0,
      "loss": 1.2161,
      "step": 1244
    },
    {
      "epoch": 0.11173936456650511,
      "grad_norm": 8.134570121765137,
      "learning_rate": 0.0,
      "loss": 1.1748,
      "step": 1245
    },
    {
      "epoch": 0.11182911506013284,
      "grad_norm": 10.845027923583984,
      "learning_rate": 0.0,
      "loss": 1.2627,
      "step": 1246
    },
    {
      "epoch": 0.11191886555376054,
      "grad_norm": 16.823156356811523,
      "learning_rate": 0.0,
      "loss": 1.2812,
      "step": 1247
    },
    {
      "epoch": 0.11200861604738827,
      "grad_norm": 6.451925754547119,
      "learning_rate": 0.0,
      "loss": 1.2575,
      "step": 1248
    },
    {
      "epoch": 0.11209836654101597,
      "grad_norm": 7.778298854827881,
      "learning_rate": 0.0,
      "loss": 1.1216,
      "step": 1249
    },
    {
      "epoch": 0.1121881170346437,
      "grad_norm": 8.172855377197266,
      "learning_rate": 0.0,
      "loss": 1.1467,
      "step": 1250
    },
    {
      "epoch": 0.1122778675282714,
      "grad_norm": 9.794872283935547,
      "learning_rate": 0.0,
      "loss": 1.2224,
      "step": 1251
    },
    {
      "epoch": 0.11236761802189912,
      "grad_norm": 7.459484100341797,
      "learning_rate": 0.0,
      "loss": 1.1655,
      "step": 1252
    },
    {
      "epoch": 0.11245736851552683,
      "grad_norm": 9.003609657287598,
      "learning_rate": 0.0,
      "loss": 1.2182,
      "step": 1253
    },
    {
      "epoch": 0.11254711900915455,
      "grad_norm": 8.609299659729004,
      "learning_rate": 0.0,
      "loss": 1.2523,
      "step": 1254
    },
    {
      "epoch": 0.11263686950278226,
      "grad_norm": 7.021795749664307,
      "learning_rate": 0.0,
      "loss": 1.1427,
      "step": 1255
    },
    {
      "epoch": 0.11272661999640998,
      "grad_norm": 6.084628582000732,
      "learning_rate": 0.0,
      "loss": 1.1996,
      "step": 1256
    },
    {
      "epoch": 0.11281637049003769,
      "grad_norm": 7.359714031219482,
      "learning_rate": 0.0,
      "loss": 1.1433,
      "step": 1257
    },
    {
      "epoch": 0.11290612098366541,
      "grad_norm": 7.1031012535095215,
      "learning_rate": 0.0,
      "loss": 1.2138,
      "step": 1258
    },
    {
      "epoch": 0.11299587147729312,
      "grad_norm": 7.928866863250732,
      "learning_rate": 0.0,
      "loss": 1.1837,
      "step": 1259
    },
    {
      "epoch": 0.11308562197092084,
      "grad_norm": 6.584926605224609,
      "learning_rate": 0.0,
      "loss": 1.2305,
      "step": 1260
    },
    {
      "epoch": 0.11317537246454855,
      "grad_norm": 7.438521385192871,
      "learning_rate": 0.0,
      "loss": 1.2366,
      "step": 1261
    },
    {
      "epoch": 0.11326512295817627,
      "grad_norm": 16.52115821838379,
      "learning_rate": 0.0,
      "loss": 1.2261,
      "step": 1262
    },
    {
      "epoch": 0.11335487345180398,
      "grad_norm": 6.690235614776611,
      "learning_rate": 0.0,
      "loss": 1.1615,
      "step": 1263
    },
    {
      "epoch": 0.1134446239454317,
      "grad_norm": 7.955246925354004,
      "learning_rate": 0.0,
      "loss": 1.2468,
      "step": 1264
    },
    {
      "epoch": 0.11353437443905942,
      "grad_norm": 7.648593902587891,
      "learning_rate": 0.0,
      "loss": 1.15,
      "step": 1265
    },
    {
      "epoch": 0.11362412493268713,
      "grad_norm": 7.6822943687438965,
      "learning_rate": 0.0,
      "loss": 1.2701,
      "step": 1266
    },
    {
      "epoch": 0.11371387542631485,
      "grad_norm": 6.7660298347473145,
      "learning_rate": 0.0,
      "loss": 1.2215,
      "step": 1267
    },
    {
      "epoch": 0.11380362591994256,
      "grad_norm": 7.505252361297607,
      "learning_rate": 0.0,
      "loss": 1.2972,
      "step": 1268
    },
    {
      "epoch": 0.11389337641357028,
      "grad_norm": 7.5392327308654785,
      "learning_rate": 0.0,
      "loss": 1.1587,
      "step": 1269
    },
    {
      "epoch": 0.11398312690719799,
      "grad_norm": 6.35374641418457,
      "learning_rate": 0.0,
      "loss": 1.2116,
      "step": 1270
    },
    {
      "epoch": 0.11407287740082571,
      "grad_norm": 8.672408103942871,
      "learning_rate": 0.0,
      "loss": 1.2199,
      "step": 1271
    },
    {
      "epoch": 0.11416262789445342,
      "grad_norm": 9.403782844543457,
      "learning_rate": 0.0,
      "loss": 1.2533,
      "step": 1272
    },
    {
      "epoch": 0.11425237838808114,
      "grad_norm": 7.117609977722168,
      "learning_rate": 0.0,
      "loss": 1.1937,
      "step": 1273
    },
    {
      "epoch": 0.11434212888170885,
      "grad_norm": 13.920687675476074,
      "learning_rate": 0.0,
      "loss": 1.1877,
      "step": 1274
    },
    {
      "epoch": 0.11443187937533657,
      "grad_norm": 11.61279582977295,
      "learning_rate": 0.0,
      "loss": 1.2113,
      "step": 1275
    },
    {
      "epoch": 0.11452162986896428,
      "grad_norm": 18.209335327148438,
      "learning_rate": 0.0,
      "loss": 1.2275,
      "step": 1276
    },
    {
      "epoch": 0.114611380362592,
      "grad_norm": 38.255008697509766,
      "learning_rate": 0.0,
      "loss": 1.1793,
      "step": 1277
    },
    {
      "epoch": 0.1147011308562197,
      "grad_norm": 26.93497657775879,
      "learning_rate": 0.0,
      "loss": 1.2581,
      "step": 1278
    },
    {
      "epoch": 0.11479088134984743,
      "grad_norm": 7.25846529006958,
      "learning_rate": 0.0,
      "loss": 1.164,
      "step": 1279
    },
    {
      "epoch": 0.11488063184347513,
      "grad_norm": 8.052148818969727,
      "learning_rate": 0.0,
      "loss": 1.1642,
      "step": 1280
    },
    {
      "epoch": 0.11497038233710286,
      "grad_norm": 6.493648052215576,
      "learning_rate": 0.0,
      "loss": 1.2078,
      "step": 1281
    },
    {
      "epoch": 0.11506013283073056,
      "grad_norm": 8.22099494934082,
      "learning_rate": 0.0,
      "loss": 1.206,
      "step": 1282
    },
    {
      "epoch": 0.11514988332435828,
      "grad_norm": 11.645425796508789,
      "learning_rate": 0.0,
      "loss": 1.2098,
      "step": 1283
    },
    {
      "epoch": 0.11523963381798599,
      "grad_norm": 25.124202728271484,
      "learning_rate": 0.0,
      "loss": 1.1661,
      "step": 1284
    },
    {
      "epoch": 0.11532938431161371,
      "grad_norm": 8.770313262939453,
      "learning_rate": 0.0,
      "loss": 1.1563,
      "step": 1285
    },
    {
      "epoch": 0.11541913480524144,
      "grad_norm": 7.769540309906006,
      "learning_rate": 0.0,
      "loss": 1.2407,
      "step": 1286
    },
    {
      "epoch": 0.11550888529886914,
      "grad_norm": 17.53595542907715,
      "learning_rate": 0.0,
      "loss": 1.1392,
      "step": 1287
    },
    {
      "epoch": 0.11559863579249686,
      "grad_norm": 7.700291156768799,
      "learning_rate": 0.0,
      "loss": 1.1904,
      "step": 1288
    },
    {
      "epoch": 0.11568838628612457,
      "grad_norm": 8.194092750549316,
      "learning_rate": 0.0,
      "loss": 1.2079,
      "step": 1289
    },
    {
      "epoch": 0.1157781367797523,
      "grad_norm": 7.172759532928467,
      "learning_rate": 0.0,
      "loss": 1.1823,
      "step": 1290
    },
    {
      "epoch": 0.11586788727338,
      "grad_norm": 7.9571027755737305,
      "learning_rate": 0.0,
      "loss": 1.1998,
      "step": 1291
    },
    {
      "epoch": 0.11595763776700772,
      "grad_norm": 6.519827842712402,
      "learning_rate": 0.0,
      "loss": 1.1465,
      "step": 1292
    },
    {
      "epoch": 0.11604738826063543,
      "grad_norm": 24.02733039855957,
      "learning_rate": 0.0,
      "loss": 1.2162,
      "step": 1293
    },
    {
      "epoch": 0.11613713875426315,
      "grad_norm": 7.202793121337891,
      "learning_rate": 0.0,
      "loss": 1.1319,
      "step": 1294
    },
    {
      "epoch": 0.11622688924789086,
      "grad_norm": 28.088790893554688,
      "learning_rate": 0.0,
      "loss": 1.3219,
      "step": 1295
    },
    {
      "epoch": 0.11631663974151858,
      "grad_norm": 6.829746723175049,
      "learning_rate": 0.0,
      "loss": 1.2202,
      "step": 1296
    },
    {
      "epoch": 0.11640639023514629,
      "grad_norm": 8.281753540039062,
      "learning_rate": 0.0,
      "loss": 1.1605,
      "step": 1297
    },
    {
      "epoch": 0.11649614072877401,
      "grad_norm": 21.075531005859375,
      "learning_rate": 0.0,
      "loss": 1.2397,
      "step": 1298
    },
    {
      "epoch": 0.11658589122240172,
      "grad_norm": 7.879082202911377,
      "learning_rate": 0.0,
      "loss": 1.2482,
      "step": 1299
    },
    {
      "epoch": 0.11667564171602944,
      "grad_norm": 6.843628883361816,
      "learning_rate": 0.0,
      "loss": 1.1522,
      "step": 1300
    },
    {
      "epoch": 0.11676539220965715,
      "grad_norm": 6.45770263671875,
      "learning_rate": 0.0,
      "loss": 1.1666,
      "step": 1301
    },
    {
      "epoch": 0.11685514270328487,
      "grad_norm": 16.390039443969727,
      "learning_rate": 0.0,
      "loss": 1.2726,
      "step": 1302
    },
    {
      "epoch": 0.11694489319691258,
      "grad_norm": 11.887619972229004,
      "learning_rate": 0.0,
      "loss": 1.3305,
      "step": 1303
    },
    {
      "epoch": 0.1170346436905403,
      "grad_norm": 33.935546875,
      "learning_rate": 0.0,
      "loss": 1.1884,
      "step": 1304
    },
    {
      "epoch": 0.11712439418416801,
      "grad_norm": 12.05514907836914,
      "learning_rate": 0.0,
      "loss": 1.202,
      "step": 1305
    },
    {
      "epoch": 0.11721414467779573,
      "grad_norm": 7.869248867034912,
      "learning_rate": 0.0,
      "loss": 1.107,
      "step": 1306
    },
    {
      "epoch": 0.11730389517142344,
      "grad_norm": 9.288777351379395,
      "learning_rate": 0.0,
      "loss": 1.0902,
      "step": 1307
    },
    {
      "epoch": 0.11739364566505116,
      "grad_norm": 11.053153038024902,
      "learning_rate": 0.0,
      "loss": 1.0952,
      "step": 1308
    },
    {
      "epoch": 0.11748339615867888,
      "grad_norm": 4.840173244476318,
      "learning_rate": 0.0,
      "loss": 1.0416,
      "step": 1309
    },
    {
      "epoch": 0.11757314665230659,
      "grad_norm": 4.625016212463379,
      "learning_rate": 0.0,
      "loss": 0.9445,
      "step": 1310
    },
    {
      "epoch": 0.11766289714593431,
      "grad_norm": 8.185973167419434,
      "learning_rate": 0.0,
      "loss": 0.9467,
      "step": 1311
    },
    {
      "epoch": 0.11775264763956202,
      "grad_norm": 5.689120292663574,
      "learning_rate": 0.0,
      "loss": 0.9931,
      "step": 1312
    },
    {
      "epoch": 0.11784239813318974,
      "grad_norm": 6.271623134613037,
      "learning_rate": 0.0,
      "loss": 0.9479,
      "step": 1313
    },
    {
      "epoch": 0.11793214862681745,
      "grad_norm": 2.591674327850342,
      "learning_rate": 0.0,
      "loss": 0.9285,
      "step": 1314
    },
    {
      "epoch": 0.11802189912044517,
      "grad_norm": 2.572460174560547,
      "learning_rate": 0.0,
      "loss": 0.901,
      "step": 1315
    },
    {
      "epoch": 0.11811164961407287,
      "grad_norm": 8.5349702835083,
      "learning_rate": 0.0,
      "loss": 0.9255,
      "step": 1316
    },
    {
      "epoch": 0.1182014001077006,
      "grad_norm": 3.5618927478790283,
      "learning_rate": 0.0,
      "loss": 0.9116,
      "step": 1317
    },
    {
      "epoch": 0.1182911506013283,
      "grad_norm": 4.149507999420166,
      "learning_rate": 0.0,
      "loss": 0.8659,
      "step": 1318
    },
    {
      "epoch": 0.11838090109495603,
      "grad_norm": 3.054419994354248,
      "learning_rate": 0.0,
      "loss": 0.9241,
      "step": 1319
    },
    {
      "epoch": 0.11847065158858373,
      "grad_norm": 3.1219875812530518,
      "learning_rate": 0.0,
      "loss": 0.9541,
      "step": 1320
    },
    {
      "epoch": 0.11856040208221146,
      "grad_norm": 7.282851696014404,
      "learning_rate": 0.0,
      "loss": 0.8978,
      "step": 1321
    },
    {
      "epoch": 0.11865015257583916,
      "grad_norm": 3.055021047592163,
      "learning_rate": 0.0,
      "loss": 0.8873,
      "step": 1322
    },
    {
      "epoch": 0.11873990306946688,
      "grad_norm": 6.2146687507629395,
      "learning_rate": 0.0,
      "loss": 0.9363,
      "step": 1323
    },
    {
      "epoch": 0.11882965356309459,
      "grad_norm": 3.9281272888183594,
      "learning_rate": 0.0,
      "loss": 0.8496,
      "step": 1324
    },
    {
      "epoch": 0.11891940405672231,
      "grad_norm": 3.495316743850708,
      "learning_rate": 0.0,
      "loss": 0.8813,
      "step": 1325
    },
    {
      "epoch": 0.11900915455035002,
      "grad_norm": 3.6599321365356445,
      "learning_rate": 0.0,
      "loss": 0.8674,
      "step": 1326
    },
    {
      "epoch": 0.11909890504397774,
      "grad_norm": 4.056681156158447,
      "learning_rate": 0.0,
      "loss": 0.8975,
      "step": 1327
    },
    {
      "epoch": 0.11918865553760545,
      "grad_norm": 3.427947998046875,
      "learning_rate": 0.0,
      "loss": 0.8824,
      "step": 1328
    },
    {
      "epoch": 0.11927840603123317,
      "grad_norm": 2.955547571182251,
      "learning_rate": 0.0,
      "loss": 0.8802,
      "step": 1329
    },
    {
      "epoch": 0.11936815652486088,
      "grad_norm": 6.7525529861450195,
      "learning_rate": 0.0,
      "loss": 0.8262,
      "step": 1330
    },
    {
      "epoch": 0.1194579070184886,
      "grad_norm": 2.9729349613189697,
      "learning_rate": 0.0,
      "loss": 0.8988,
      "step": 1331
    },
    {
      "epoch": 0.11954765751211632,
      "grad_norm": 3.6296300888061523,
      "learning_rate": 0.0,
      "loss": 0.8919,
      "step": 1332
    },
    {
      "epoch": 0.11963740800574403,
      "grad_norm": 8.945056915283203,
      "learning_rate": 0.0,
      "loss": 0.8048,
      "step": 1333
    },
    {
      "epoch": 0.11972715849937175,
      "grad_norm": 3.169689655303955,
      "learning_rate": 0.0,
      "loss": 0.884,
      "step": 1334
    },
    {
      "epoch": 0.11981690899299946,
      "grad_norm": 5.0558929443359375,
      "learning_rate": 0.0,
      "loss": 0.8784,
      "step": 1335
    },
    {
      "epoch": 0.11990665948662718,
      "grad_norm": 3.1115188598632812,
      "learning_rate": 0.0,
      "loss": 0.8925,
      "step": 1336
    },
    {
      "epoch": 0.11999640998025489,
      "grad_norm": 53.6694221496582,
      "learning_rate": 0.0,
      "loss": 0.886,
      "step": 1337
    },
    {
      "epoch": 0.12008616047388261,
      "grad_norm": 4.775752544403076,
      "learning_rate": 0.0,
      "loss": 0.8872,
      "step": 1338
    },
    {
      "epoch": 0.12017591096751032,
      "grad_norm": 16.484338760375977,
      "learning_rate": 0.0,
      "loss": 0.8863,
      "step": 1339
    },
    {
      "epoch": 0.12026566146113804,
      "grad_norm": 13.74631404876709,
      "learning_rate": 0.0,
      "loss": 0.9139,
      "step": 1340
    },
    {
      "epoch": 0.12035541195476575,
      "grad_norm": 7.539771556854248,
      "learning_rate": 0.0,
      "loss": 0.8938,
      "step": 1341
    },
    {
      "epoch": 0.12044516244839347,
      "grad_norm": 4.201356887817383,
      "learning_rate": 0.0,
      "loss": 0.8602,
      "step": 1342
    },
    {
      "epoch": 0.12053491294202118,
      "grad_norm": 2.961987257003784,
      "learning_rate": 0.0,
      "loss": 0.8562,
      "step": 1343
    },
    {
      "epoch": 0.1206246634356489,
      "grad_norm": 4.72053337097168,
      "learning_rate": 0.0,
      "loss": 0.8408,
      "step": 1344
    },
    {
      "epoch": 0.1207144139292766,
      "grad_norm": 4.231282711029053,
      "learning_rate": 0.0,
      "loss": 0.85,
      "step": 1345
    },
    {
      "epoch": 0.12080416442290433,
      "grad_norm": 2.923861026763916,
      "learning_rate": 0.0,
      "loss": 0.8575,
      "step": 1346
    },
    {
      "epoch": 0.12089391491653204,
      "grad_norm": 2.2491421699523926,
      "learning_rate": 0.0,
      "loss": 0.8919,
      "step": 1347
    },
    {
      "epoch": 0.12098366541015976,
      "grad_norm": 2.543591022491455,
      "learning_rate": 0.0,
      "loss": 0.8794,
      "step": 1348
    },
    {
      "epoch": 0.12107341590378747,
      "grad_norm": 2.6642165184020996,
      "learning_rate": 0.0,
      "loss": 0.8226,
      "step": 1349
    },
    {
      "epoch": 0.12116316639741519,
      "grad_norm": 3.7759668827056885,
      "learning_rate": 0.0,
      "loss": 0.7821,
      "step": 1350
    },
    {
      "epoch": 0.1212529168910429,
      "grad_norm": 9.212008476257324,
      "learning_rate": 0.0,
      "loss": 0.8565,
      "step": 1351
    },
    {
      "epoch": 0.12134266738467062,
      "grad_norm": 3.5141794681549072,
      "learning_rate": 0.0,
      "loss": 0.831,
      "step": 1352
    },
    {
      "epoch": 0.12143241787829832,
      "grad_norm": 3.7751424312591553,
      "learning_rate": 0.0,
      "loss": 0.8418,
      "step": 1353
    },
    {
      "epoch": 0.12152216837192605,
      "grad_norm": 3.5769803524017334,
      "learning_rate": 0.0,
      "loss": 0.8704,
      "step": 1354
    },
    {
      "epoch": 0.12161191886555377,
      "grad_norm": 3.1551246643066406,
      "learning_rate": 0.0,
      "loss": 0.905,
      "step": 1355
    },
    {
      "epoch": 0.12170166935918147,
      "grad_norm": 2.9197802543640137,
      "learning_rate": 0.0,
      "loss": 0.8966,
      "step": 1356
    },
    {
      "epoch": 0.1217914198528092,
      "grad_norm": 2.1754260063171387,
      "learning_rate": 0.0,
      "loss": 0.8403,
      "step": 1357
    },
    {
      "epoch": 0.1218811703464369,
      "grad_norm": 3.233304977416992,
      "learning_rate": 0.0,
      "loss": 0.8768,
      "step": 1358
    },
    {
      "epoch": 0.12197092084006463,
      "grad_norm": 3.8422610759735107,
      "learning_rate": 0.0,
      "loss": 0.8159,
      "step": 1359
    },
    {
      "epoch": 0.12206067133369233,
      "grad_norm": 27.34998321533203,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 1360
    },
    {
      "epoch": 0.12215042182732005,
      "grad_norm": 10.579364776611328,
      "learning_rate": 0.0,
      "loss": 0.8114,
      "step": 1361
    },
    {
      "epoch": 0.12224017232094776,
      "grad_norm": 3.0684635639190674,
      "learning_rate": 0.0,
      "loss": 0.8439,
      "step": 1362
    },
    {
      "epoch": 0.12232992281457548,
      "grad_norm": 2.6055803298950195,
      "learning_rate": 0.0,
      "loss": 0.8315,
      "step": 1363
    },
    {
      "epoch": 0.12241967330820319,
      "grad_norm": 2.9728775024414062,
      "learning_rate": 0.0,
      "loss": 0.8464,
      "step": 1364
    },
    {
      "epoch": 0.12250942380183091,
      "grad_norm": 3.1382853984832764,
      "learning_rate": 0.0,
      "loss": 0.8426,
      "step": 1365
    },
    {
      "epoch": 0.12259917429545862,
      "grad_norm": 7.65596866607666,
      "learning_rate": 0.0,
      "loss": 0.7664,
      "step": 1366
    },
    {
      "epoch": 0.12268892478908634,
      "grad_norm": 3.3511533737182617,
      "learning_rate": 0.0,
      "loss": 0.7829,
      "step": 1367
    },
    {
      "epoch": 0.12277867528271405,
      "grad_norm": 3.417457103729248,
      "learning_rate": 0.0,
      "loss": 0.8374,
      "step": 1368
    },
    {
      "epoch": 0.12286842577634177,
      "grad_norm": 3.5268969535827637,
      "learning_rate": 0.0,
      "loss": 0.8058,
      "step": 1369
    },
    {
      "epoch": 0.12295817626996948,
      "grad_norm": 4.202220916748047,
      "learning_rate": 0.0,
      "loss": 0.8125,
      "step": 1370
    },
    {
      "epoch": 0.1230479267635972,
      "grad_norm": 5.48383903503418,
      "learning_rate": 0.0,
      "loss": 0.8108,
      "step": 1371
    },
    {
      "epoch": 0.12313767725722491,
      "grad_norm": 3.3315043449401855,
      "learning_rate": 0.0,
      "loss": 0.8972,
      "step": 1372
    },
    {
      "epoch": 0.12322742775085263,
      "grad_norm": 5.842352867126465,
      "learning_rate": 0.0,
      "loss": 0.8383,
      "step": 1373
    },
    {
      "epoch": 0.12331717824448034,
      "grad_norm": 3.0852839946746826,
      "learning_rate": 0.0,
      "loss": 0.8286,
      "step": 1374
    },
    {
      "epoch": 0.12340692873810806,
      "grad_norm": 4.278652667999268,
      "learning_rate": 0.0,
      "loss": 0.9369,
      "step": 1375
    },
    {
      "epoch": 0.12349667923173578,
      "grad_norm": 3.437389373779297,
      "learning_rate": 0.0,
      "loss": 0.8221,
      "step": 1376
    },
    {
      "epoch": 0.12358642972536349,
      "grad_norm": 4.829852104187012,
      "learning_rate": 0.0,
      "loss": 0.8638,
      "step": 1377
    },
    {
      "epoch": 0.12367618021899121,
      "grad_norm": 4.801817893981934,
      "learning_rate": 0.0,
      "loss": 0.8636,
      "step": 1378
    },
    {
      "epoch": 0.12376593071261892,
      "grad_norm": 2.8580024242401123,
      "learning_rate": 0.0,
      "loss": 0.8409,
      "step": 1379
    },
    {
      "epoch": 0.12385568120624664,
      "grad_norm": 2.943033456802368,
      "learning_rate": 0.0,
      "loss": 0.7981,
      "step": 1380
    },
    {
      "epoch": 0.12394543169987435,
      "grad_norm": 3.327155113220215,
      "learning_rate": 0.0,
      "loss": 0.8137,
      "step": 1381
    },
    {
      "epoch": 0.12403518219350207,
      "grad_norm": 3.209129810333252,
      "learning_rate": 0.0,
      "loss": 0.8257,
      "step": 1382
    },
    {
      "epoch": 0.12412493268712978,
      "grad_norm": 3.4587008953094482,
      "learning_rate": 0.0,
      "loss": 0.8699,
      "step": 1383
    },
    {
      "epoch": 0.1242146831807575,
      "grad_norm": 2.7336111068725586,
      "learning_rate": 0.0,
      "loss": 0.8138,
      "step": 1384
    },
    {
      "epoch": 0.1243044336743852,
      "grad_norm": 5.170899391174316,
      "learning_rate": 0.0,
      "loss": 0.8362,
      "step": 1385
    },
    {
      "epoch": 0.12439418416801293,
      "grad_norm": 2.695998430252075,
      "learning_rate": 0.0,
      "loss": 0.8244,
      "step": 1386
    },
    {
      "epoch": 0.12448393466164064,
      "grad_norm": 2.6565158367156982,
      "learning_rate": 0.0,
      "loss": 0.8263,
      "step": 1387
    },
    {
      "epoch": 0.12457368515526836,
      "grad_norm": 3.30728816986084,
      "learning_rate": 0.0,
      "loss": 0.7929,
      "step": 1388
    },
    {
      "epoch": 0.12466343564889606,
      "grad_norm": 2.8895537853240967,
      "learning_rate": 0.0,
      "loss": 0.835,
      "step": 1389
    },
    {
      "epoch": 0.12475318614252379,
      "grad_norm": 4.486013889312744,
      "learning_rate": 0.0,
      "loss": 0.7316,
      "step": 1390
    },
    {
      "epoch": 0.1248429366361515,
      "grad_norm": 3.220323324203491,
      "learning_rate": 0.0,
      "loss": 0.836,
      "step": 1391
    },
    {
      "epoch": 0.12493268712977922,
      "grad_norm": 3.681502103805542,
      "learning_rate": 0.0,
      "loss": 0.826,
      "step": 1392
    },
    {
      "epoch": 0.12502243762340692,
      "grad_norm": 3.2954819202423096,
      "learning_rate": 0.0,
      "loss": 0.8346,
      "step": 1393
    },
    {
      "epoch": 0.12511218811703465,
      "grad_norm": 3.3995254039764404,
      "learning_rate": 0.0,
      "loss": 0.8264,
      "step": 1394
    },
    {
      "epoch": 0.12520193861066237,
      "grad_norm": 3.293337345123291,
      "learning_rate": 0.0,
      "loss": 0.8494,
      "step": 1395
    },
    {
      "epoch": 0.12529168910429006,
      "grad_norm": 3.1651837825775146,
      "learning_rate": 0.0,
      "loss": 0.7669,
      "step": 1396
    },
    {
      "epoch": 0.12538143959791778,
      "grad_norm": 4.2832865715026855,
      "learning_rate": 0.0,
      "loss": 0.8312,
      "step": 1397
    },
    {
      "epoch": 0.1254711900915455,
      "grad_norm": 4.65566349029541,
      "learning_rate": 0.0,
      "loss": 0.7932,
      "step": 1398
    },
    {
      "epoch": 0.12556094058517323,
      "grad_norm": 4.833236217498779,
      "learning_rate": 0.0,
      "loss": 0.8386,
      "step": 1399
    },
    {
      "epoch": 0.12565069107880095,
      "grad_norm": 3.736387252807617,
      "learning_rate": 0.0,
      "loss": 0.8092,
      "step": 1400
    },
    {
      "epoch": 0.12574044157242864,
      "grad_norm": 28.42026710510254,
      "learning_rate": 0.0,
      "loss": 0.7876,
      "step": 1401
    },
    {
      "epoch": 0.12583019206605636,
      "grad_norm": 3.2829086780548096,
      "learning_rate": 0.0,
      "loss": 0.769,
      "step": 1402
    },
    {
      "epoch": 0.12591994255968408,
      "grad_norm": 3.3568921089172363,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 1403
    },
    {
      "epoch": 0.1260096930533118,
      "grad_norm": 5.384433269500732,
      "learning_rate": 0.0,
      "loss": 0.8559,
      "step": 1404
    },
    {
      "epoch": 0.1260994435469395,
      "grad_norm": 4.942105293273926,
      "learning_rate": 0.0,
      "loss": 0.8611,
      "step": 1405
    },
    {
      "epoch": 0.12618919404056722,
      "grad_norm": 3.296809673309326,
      "learning_rate": 0.0,
      "loss": 0.8499,
      "step": 1406
    },
    {
      "epoch": 0.12627894453419494,
      "grad_norm": 6.000167369842529,
      "learning_rate": 0.0,
      "loss": 0.8587,
      "step": 1407
    },
    {
      "epoch": 0.12636869502782266,
      "grad_norm": 3.370332956314087,
      "learning_rate": 0.0,
      "loss": 0.8319,
      "step": 1408
    },
    {
      "epoch": 0.12645844552145036,
      "grad_norm": 3.380084753036499,
      "learning_rate": 0.0,
      "loss": 0.852,
      "step": 1409
    },
    {
      "epoch": 0.12654819601507808,
      "grad_norm": 4.914562225341797,
      "learning_rate": 0.0,
      "loss": 0.8121,
      "step": 1410
    },
    {
      "epoch": 0.1266379465087058,
      "grad_norm": 4.047183036804199,
      "learning_rate": 0.0,
      "loss": 0.8428,
      "step": 1411
    },
    {
      "epoch": 0.12672769700233352,
      "grad_norm": 3.277526378631592,
      "learning_rate": 0.0,
      "loss": 0.8387,
      "step": 1412
    },
    {
      "epoch": 0.12681744749596122,
      "grad_norm": 4.395352363586426,
      "learning_rate": 0.0,
      "loss": 0.7993,
      "step": 1413
    },
    {
      "epoch": 0.12690719798958894,
      "grad_norm": 7.212756156921387,
      "learning_rate": 0.0,
      "loss": 0.8076,
      "step": 1414
    },
    {
      "epoch": 0.12699694848321666,
      "grad_norm": 4.71928596496582,
      "learning_rate": 0.0,
      "loss": 0.822,
      "step": 1415
    },
    {
      "epoch": 0.12708669897684438,
      "grad_norm": 2.953191041946411,
      "learning_rate": 0.0,
      "loss": 0.8347,
      "step": 1416
    },
    {
      "epoch": 0.12717644947047207,
      "grad_norm": 3.4680428504943848,
      "learning_rate": 0.0,
      "loss": 0.8459,
      "step": 1417
    },
    {
      "epoch": 0.1272661999640998,
      "grad_norm": 2.9385268688201904,
      "learning_rate": 0.0,
      "loss": 0.8073,
      "step": 1418
    },
    {
      "epoch": 0.12735595045772752,
      "grad_norm": 3.0528273582458496,
      "learning_rate": 0.0,
      "loss": 0.8007,
      "step": 1419
    },
    {
      "epoch": 0.12744570095135524,
      "grad_norm": 15.826151847839355,
      "learning_rate": 0.0,
      "loss": 0.783,
      "step": 1420
    },
    {
      "epoch": 0.12753545144498296,
      "grad_norm": 3.290626049041748,
      "learning_rate": 0.0,
      "loss": 0.7874,
      "step": 1421
    },
    {
      "epoch": 0.12762520193861066,
      "grad_norm": 3.1153650283813477,
      "learning_rate": 0.0,
      "loss": 0.8332,
      "step": 1422
    },
    {
      "epoch": 0.12771495243223838,
      "grad_norm": 28.003129959106445,
      "learning_rate": 0.0,
      "loss": 0.8426,
      "step": 1423
    },
    {
      "epoch": 0.1278047029258661,
      "grad_norm": 4.164175510406494,
      "learning_rate": 0.0,
      "loss": 0.8486,
      "step": 1424
    },
    {
      "epoch": 0.12789445341949382,
      "grad_norm": 4.136649131774902,
      "learning_rate": 0.0,
      "loss": 0.8511,
      "step": 1425
    },
    {
      "epoch": 0.1279842039131215,
      "grad_norm": 2.930567741394043,
      "learning_rate": 0.0,
      "loss": 0.8207,
      "step": 1426
    },
    {
      "epoch": 0.12807395440674924,
      "grad_norm": 3.742475986480713,
      "learning_rate": 0.0,
      "loss": 0.8411,
      "step": 1427
    },
    {
      "epoch": 0.12816370490037696,
      "grad_norm": 3.2726218700408936,
      "learning_rate": 0.0,
      "loss": 0.8466,
      "step": 1428
    },
    {
      "epoch": 0.12825345539400468,
      "grad_norm": 3.855350971221924,
      "learning_rate": 0.0,
      "loss": 0.8432,
      "step": 1429
    },
    {
      "epoch": 0.12834320588763237,
      "grad_norm": 2.975270986557007,
      "learning_rate": 0.0,
      "loss": 0.7992,
      "step": 1430
    },
    {
      "epoch": 0.1284329563812601,
      "grad_norm": 5.116308212280273,
      "learning_rate": 0.0,
      "loss": 0.7977,
      "step": 1431
    },
    {
      "epoch": 0.12852270687488782,
      "grad_norm": 5.25217342376709,
      "learning_rate": 0.0,
      "loss": 0.8369,
      "step": 1432
    },
    {
      "epoch": 0.12861245736851554,
      "grad_norm": 5.213943958282471,
      "learning_rate": 0.0,
      "loss": 0.8059,
      "step": 1433
    },
    {
      "epoch": 0.12870220786214323,
      "grad_norm": 2.9914968013763428,
      "learning_rate": 0.0,
      "loss": 0.8293,
      "step": 1434
    },
    {
      "epoch": 0.12879195835577095,
      "grad_norm": 3.257394313812256,
      "learning_rate": 0.0,
      "loss": 0.7828,
      "step": 1435
    },
    {
      "epoch": 0.12888170884939867,
      "grad_norm": 11.995745658874512,
      "learning_rate": 0.0,
      "loss": 0.8376,
      "step": 1436
    },
    {
      "epoch": 0.1289714593430264,
      "grad_norm": 7.033924579620361,
      "learning_rate": 0.0,
      "loss": 0.8602,
      "step": 1437
    },
    {
      "epoch": 0.1290612098366541,
      "grad_norm": 3.1522083282470703,
      "learning_rate": 0.0,
      "loss": 0.7923,
      "step": 1438
    },
    {
      "epoch": 0.1291509603302818,
      "grad_norm": 4.362265586853027,
      "learning_rate": 0.0,
      "loss": 0.7678,
      "step": 1439
    },
    {
      "epoch": 0.12924071082390953,
      "grad_norm": 3.1200053691864014,
      "learning_rate": 0.0,
      "loss": 0.803,
      "step": 1440
    },
    {
      "epoch": 0.12933046131753725,
      "grad_norm": 2.5968031883239746,
      "learning_rate": 0.0,
      "loss": 0.8834,
      "step": 1441
    },
    {
      "epoch": 0.12942021181116495,
      "grad_norm": 4.015469074249268,
      "learning_rate": 0.0,
      "loss": 0.8211,
      "step": 1442
    },
    {
      "epoch": 0.12950996230479267,
      "grad_norm": 3.4720048904418945,
      "learning_rate": 0.0,
      "loss": 0.7981,
      "step": 1443
    },
    {
      "epoch": 0.1295997127984204,
      "grad_norm": 5.314981937408447,
      "learning_rate": 0.0,
      "loss": 0.8098,
      "step": 1444
    },
    {
      "epoch": 0.1296894632920481,
      "grad_norm": 3.337761878967285,
      "learning_rate": 0.0,
      "loss": 0.809,
      "step": 1445
    },
    {
      "epoch": 0.12977921378567583,
      "grad_norm": 3.876595973968506,
      "learning_rate": 0.0,
      "loss": 0.8075,
      "step": 1446
    },
    {
      "epoch": 0.12986896427930353,
      "grad_norm": 4.335240840911865,
      "learning_rate": 0.0,
      "loss": 0.7976,
      "step": 1447
    },
    {
      "epoch": 0.12995871477293125,
      "grad_norm": 3.3347103595733643,
      "learning_rate": 0.0,
      "loss": 0.8221,
      "step": 1448
    },
    {
      "epoch": 0.13004846526655897,
      "grad_norm": 3.7204883098602295,
      "learning_rate": 0.0,
      "loss": 0.8072,
      "step": 1449
    },
    {
      "epoch": 0.1301382157601867,
      "grad_norm": 6.366868019104004,
      "learning_rate": 0.0,
      "loss": 0.801,
      "step": 1450
    },
    {
      "epoch": 0.1302279662538144,
      "grad_norm": 3.9813666343688965,
      "learning_rate": 0.0,
      "loss": 0.828,
      "step": 1451
    },
    {
      "epoch": 0.1303177167474421,
      "grad_norm": 6.814449310302734,
      "learning_rate": 0.0,
      "loss": 0.7566,
      "step": 1452
    },
    {
      "epoch": 0.13040746724106983,
      "grad_norm": 3.555870771408081,
      "learning_rate": 0.0,
      "loss": 0.8267,
      "step": 1453
    },
    {
      "epoch": 0.13049721773469755,
      "grad_norm": 5.159738540649414,
      "learning_rate": 0.0,
      "loss": 0.7992,
      "step": 1454
    },
    {
      "epoch": 0.13058696822832525,
      "grad_norm": 4.987093925476074,
      "learning_rate": 0.0,
      "loss": 0.8585,
      "step": 1455
    },
    {
      "epoch": 0.13067671872195297,
      "grad_norm": 3.26273250579834,
      "learning_rate": 0.0,
      "loss": 0.7894,
      "step": 1456
    },
    {
      "epoch": 0.1307664692155807,
      "grad_norm": 23.634620666503906,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 1457
    },
    {
      "epoch": 0.1308562197092084,
      "grad_norm": 3.5267012119293213,
      "learning_rate": 0.0,
      "loss": 0.8022,
      "step": 1458
    },
    {
      "epoch": 0.1309459702028361,
      "grad_norm": 14.212865829467773,
      "learning_rate": 0.0,
      "loss": 0.8493,
      "step": 1459
    },
    {
      "epoch": 0.13103572069646383,
      "grad_norm": 3.117931604385376,
      "learning_rate": 0.0,
      "loss": 0.8136,
      "step": 1460
    },
    {
      "epoch": 0.13112547119009155,
      "grad_norm": 3.611910343170166,
      "learning_rate": 0.0,
      "loss": 0.8182,
      "step": 1461
    },
    {
      "epoch": 0.13121522168371927,
      "grad_norm": 3.1421852111816406,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 1462
    },
    {
      "epoch": 0.13130497217734696,
      "grad_norm": 7.689531326293945,
      "learning_rate": 0.0,
      "loss": 0.8024,
      "step": 1463
    },
    {
      "epoch": 0.13139472267097468,
      "grad_norm": 4.756537914276123,
      "learning_rate": 0.0,
      "loss": 0.7698,
      "step": 1464
    },
    {
      "epoch": 0.1314844731646024,
      "grad_norm": 2.747042179107666,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 1465
    },
    {
      "epoch": 0.13157422365823013,
      "grad_norm": 5.083803653717041,
      "learning_rate": 0.0,
      "loss": 0.8097,
      "step": 1466
    },
    {
      "epoch": 0.13166397415185785,
      "grad_norm": 3.5092403888702393,
      "learning_rate": 0.0,
      "loss": 0.7886,
      "step": 1467
    },
    {
      "epoch": 0.13175372464548554,
      "grad_norm": 4.208290100097656,
      "learning_rate": 0.0,
      "loss": 0.8415,
      "step": 1468
    },
    {
      "epoch": 0.13184347513911326,
      "grad_norm": 3.375845193862915,
      "learning_rate": 0.0,
      "loss": 0.8397,
      "step": 1469
    },
    {
      "epoch": 0.13193322563274099,
      "grad_norm": 3.564755439758301,
      "learning_rate": 0.0,
      "loss": 0.7884,
      "step": 1470
    },
    {
      "epoch": 0.1320229761263687,
      "grad_norm": 4.49355936050415,
      "learning_rate": 0.0,
      "loss": 0.861,
      "step": 1471
    },
    {
      "epoch": 0.1321127266199964,
      "grad_norm": 5.399920463562012,
      "learning_rate": 0.0,
      "loss": 0.7722,
      "step": 1472
    },
    {
      "epoch": 0.13220247711362412,
      "grad_norm": 5.038569450378418,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 1473
    },
    {
      "epoch": 0.13229222760725184,
      "grad_norm": 4.007917881011963,
      "learning_rate": 0.0,
      "loss": 0.8287,
      "step": 1474
    },
    {
      "epoch": 0.13238197810087957,
      "grad_norm": 31.763458251953125,
      "learning_rate": 0.0,
      "loss": 0.8037,
      "step": 1475
    },
    {
      "epoch": 0.13247172859450726,
      "grad_norm": 6.95653772354126,
      "learning_rate": 0.0,
      "loss": 0.7835,
      "step": 1476
    },
    {
      "epoch": 0.13256147908813498,
      "grad_norm": 5.09358024597168,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 1477
    },
    {
      "epoch": 0.1326512295817627,
      "grad_norm": 4.794590473175049,
      "learning_rate": 0.0,
      "loss": 0.8035,
      "step": 1478
    },
    {
      "epoch": 0.13274098007539042,
      "grad_norm": 2.948798179626465,
      "learning_rate": 0.0,
      "loss": 0.8339,
      "step": 1479
    },
    {
      "epoch": 0.13283073056901812,
      "grad_norm": 3.881348133087158,
      "learning_rate": 0.0,
      "loss": 0.8142,
      "step": 1480
    },
    {
      "epoch": 0.13292048106264584,
      "grad_norm": 4.243941307067871,
      "learning_rate": 0.0,
      "loss": 0.8117,
      "step": 1481
    },
    {
      "epoch": 0.13301023155627356,
      "grad_norm": 3.6955807209014893,
      "learning_rate": 0.0,
      "loss": 0.8671,
      "step": 1482
    },
    {
      "epoch": 0.13309998204990128,
      "grad_norm": 12.307872772216797,
      "learning_rate": 0.0,
      "loss": 0.798,
      "step": 1483
    },
    {
      "epoch": 0.13318973254352898,
      "grad_norm": 137.09596252441406,
      "learning_rate": 0.0,
      "loss": 0.8355,
      "step": 1484
    },
    {
      "epoch": 0.1332794830371567,
      "grad_norm": 8.175546646118164,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 1485
    },
    {
      "epoch": 0.13336923353078442,
      "grad_norm": 3.095238447189331,
      "learning_rate": 0.0,
      "loss": 0.8439,
      "step": 1486
    },
    {
      "epoch": 0.13345898402441214,
      "grad_norm": 5.654092311859131,
      "learning_rate": 0.0,
      "loss": 0.8151,
      "step": 1487
    },
    {
      "epoch": 0.13354873451803984,
      "grad_norm": 4.622614860534668,
      "learning_rate": 0.0,
      "loss": 0.7901,
      "step": 1488
    },
    {
      "epoch": 0.13363848501166756,
      "grad_norm": 3.0681564807891846,
      "learning_rate": 0.0,
      "loss": 0.8727,
      "step": 1489
    },
    {
      "epoch": 0.13372823550529528,
      "grad_norm": 7.82111930847168,
      "learning_rate": 0.0,
      "loss": 0.8321,
      "step": 1490
    },
    {
      "epoch": 0.133817985998923,
      "grad_norm": 9.011972427368164,
      "learning_rate": 0.0,
      "loss": 0.7908,
      "step": 1491
    },
    {
      "epoch": 0.13390773649255072,
      "grad_norm": 4.244862079620361,
      "learning_rate": 0.0,
      "loss": 0.8256,
      "step": 1492
    },
    {
      "epoch": 0.13399748698617842,
      "grad_norm": 6.555444240570068,
      "learning_rate": 0.0,
      "loss": 0.8041,
      "step": 1493
    },
    {
      "epoch": 0.13408723747980614,
      "grad_norm": 4.925446510314941,
      "learning_rate": 0.0,
      "loss": 0.8415,
      "step": 1494
    },
    {
      "epoch": 0.13417698797343386,
      "grad_norm": 3.199561357498169,
      "learning_rate": 0.0,
      "loss": 0.785,
      "step": 1495
    },
    {
      "epoch": 0.13426673846706158,
      "grad_norm": 3.3326032161712646,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 1496
    },
    {
      "epoch": 0.13435648896068927,
      "grad_norm": 6.204670429229736,
      "learning_rate": 0.0,
      "loss": 0.849,
      "step": 1497
    },
    {
      "epoch": 0.134446239454317,
      "grad_norm": 4.512786388397217,
      "learning_rate": 0.0,
      "loss": 0.7636,
      "step": 1498
    },
    {
      "epoch": 0.13453598994794472,
      "grad_norm": 28.2576961517334,
      "learning_rate": 0.0,
      "loss": 0.7826,
      "step": 1499
    },
    {
      "epoch": 0.13462574044157244,
      "grad_norm": 3.1445059776306152,
      "learning_rate": 0.0,
      "loss": 0.8336,
      "step": 1500
    },
    {
      "epoch": 0.13471549093520013,
      "grad_norm": 10.936219215393066,
      "learning_rate": 0.0,
      "loss": 0.8068,
      "step": 1501
    },
    {
      "epoch": 0.13480524142882785,
      "grad_norm": 6.0204315185546875,
      "learning_rate": 0.0,
      "loss": 0.7957,
      "step": 1502
    },
    {
      "epoch": 0.13489499192245558,
      "grad_norm": 5.517419338226318,
      "learning_rate": 0.0,
      "loss": 0.8295,
      "step": 1503
    },
    {
      "epoch": 0.1349847424160833,
      "grad_norm": 4.203054428100586,
      "learning_rate": 0.0,
      "loss": 0.8035,
      "step": 1504
    },
    {
      "epoch": 0.135074492909711,
      "grad_norm": 3.3641905784606934,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 1505
    },
    {
      "epoch": 0.1351642434033387,
      "grad_norm": 21.117454528808594,
      "learning_rate": 0.0,
      "loss": 0.7842,
      "step": 1506
    },
    {
      "epoch": 0.13525399389696643,
      "grad_norm": 3.7183563709259033,
      "learning_rate": 0.0,
      "loss": 0.844,
      "step": 1507
    },
    {
      "epoch": 0.13534374439059416,
      "grad_norm": 4.892834186553955,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 1508
    },
    {
      "epoch": 0.13543349488422185,
      "grad_norm": 7.216326713562012,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 1509
    },
    {
      "epoch": 0.13552324537784957,
      "grad_norm": 6.195159912109375,
      "learning_rate": 0.0,
      "loss": 0.8056,
      "step": 1510
    },
    {
      "epoch": 0.1356129958714773,
      "grad_norm": 4.509872913360596,
      "learning_rate": 0.0,
      "loss": 0.7888,
      "step": 1511
    },
    {
      "epoch": 0.13570274636510501,
      "grad_norm": 9.1488618850708,
      "learning_rate": 0.0,
      "loss": 0.8315,
      "step": 1512
    },
    {
      "epoch": 0.13579249685873274,
      "grad_norm": 3.903238534927368,
      "learning_rate": 0.0,
      "loss": 0.8045,
      "step": 1513
    },
    {
      "epoch": 0.13588224735236043,
      "grad_norm": 6.0497260093688965,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 1514
    },
    {
      "epoch": 0.13597199784598815,
      "grad_norm": 5.9053425788879395,
      "learning_rate": 0.0,
      "loss": 0.7941,
      "step": 1515
    },
    {
      "epoch": 0.13606174833961587,
      "grad_norm": 8.562419891357422,
      "learning_rate": 0.0,
      "loss": 0.8037,
      "step": 1516
    },
    {
      "epoch": 0.1361514988332436,
      "grad_norm": 21.54397964477539,
      "learning_rate": 0.0,
      "loss": 0.7845,
      "step": 1517
    },
    {
      "epoch": 0.1362412493268713,
      "grad_norm": 13.83799934387207,
      "learning_rate": 0.0,
      "loss": 0.7664,
      "step": 1518
    },
    {
      "epoch": 0.136330999820499,
      "grad_norm": 2.7262790203094482,
      "learning_rate": 0.0,
      "loss": 0.7897,
      "step": 1519
    },
    {
      "epoch": 0.13642075031412673,
      "grad_norm": 4.8440117835998535,
      "learning_rate": 0.0,
      "loss": 0.9098,
      "step": 1520
    },
    {
      "epoch": 0.13651050080775445,
      "grad_norm": 5.991893768310547,
      "learning_rate": 0.0,
      "loss": 0.8016,
      "step": 1521
    },
    {
      "epoch": 0.13660025130138215,
      "grad_norm": 5.69284725189209,
      "learning_rate": 0.0,
      "loss": 0.8495,
      "step": 1522
    },
    {
      "epoch": 0.13669000179500987,
      "grad_norm": 5.677691459655762,
      "learning_rate": 0.0,
      "loss": 0.8369,
      "step": 1523
    },
    {
      "epoch": 0.1367797522886376,
      "grad_norm": 6.228060245513916,
      "learning_rate": 0.0,
      "loss": 0.791,
      "step": 1524
    },
    {
      "epoch": 0.1368695027822653,
      "grad_norm": 3.256396532058716,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 1525
    },
    {
      "epoch": 0.136959253275893,
      "grad_norm": 3.101088285446167,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 1526
    },
    {
      "epoch": 0.13704900376952073,
      "grad_norm": 4.024650573730469,
      "learning_rate": 0.0,
      "loss": 0.8037,
      "step": 1527
    },
    {
      "epoch": 0.13713875426314845,
      "grad_norm": 2.6009371280670166,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 1528
    },
    {
      "epoch": 0.13722850475677617,
      "grad_norm": 2.999730348587036,
      "learning_rate": 0.0,
      "loss": 0.8354,
      "step": 1529
    },
    {
      "epoch": 0.13731825525040386,
      "grad_norm": 3.3506112098693848,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 1530
    },
    {
      "epoch": 0.13740800574403159,
      "grad_norm": 2.7669737339019775,
      "learning_rate": 0.0,
      "loss": 0.8254,
      "step": 1531
    },
    {
      "epoch": 0.1374977562376593,
      "grad_norm": 4.088271141052246,
      "learning_rate": 0.0,
      "loss": 0.8136,
      "step": 1532
    },
    {
      "epoch": 0.13758750673128703,
      "grad_norm": 3.149228811264038,
      "learning_rate": 0.0,
      "loss": 0.8095,
      "step": 1533
    },
    {
      "epoch": 0.13767725722491475,
      "grad_norm": 3.2076098918914795,
      "learning_rate": 0.0,
      "loss": 0.7899,
      "step": 1534
    },
    {
      "epoch": 0.13776700771854244,
      "grad_norm": 2.985337734222412,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 1535
    },
    {
      "epoch": 0.13785675821217017,
      "grad_norm": 3.5015008449554443,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 1536
    },
    {
      "epoch": 0.1379465087057979,
      "grad_norm": 4.3768391609191895,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 1537
    },
    {
      "epoch": 0.1380362591994256,
      "grad_norm": 3.974320650100708,
      "learning_rate": 0.0,
      "loss": 0.7858,
      "step": 1538
    },
    {
      "epoch": 0.1381260096930533,
      "grad_norm": 4.427311420440674,
      "learning_rate": 0.0,
      "loss": 0.8337,
      "step": 1539
    },
    {
      "epoch": 0.13821576018668102,
      "grad_norm": 5.264533996582031,
      "learning_rate": 0.0,
      "loss": 0.7968,
      "step": 1540
    },
    {
      "epoch": 0.13830551068030875,
      "grad_norm": 3.1967344284057617,
      "learning_rate": 0.0,
      "loss": 0.8617,
      "step": 1541
    },
    {
      "epoch": 0.13839526117393647,
      "grad_norm": 3.666820764541626,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 1542
    },
    {
      "epoch": 0.13848501166756416,
      "grad_norm": 7.557955265045166,
      "learning_rate": 0.0,
      "loss": 0.773,
      "step": 1543
    },
    {
      "epoch": 0.13857476216119188,
      "grad_norm": 3.749234199523926,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 1544
    },
    {
      "epoch": 0.1386645126548196,
      "grad_norm": 2.710585594177246,
      "learning_rate": 0.0,
      "loss": 0.7901,
      "step": 1545
    },
    {
      "epoch": 0.13875426314844733,
      "grad_norm": 4.290912628173828,
      "learning_rate": 0.0,
      "loss": 0.7697,
      "step": 1546
    },
    {
      "epoch": 0.13884401364207502,
      "grad_norm": 6.459543704986572,
      "learning_rate": 0.0,
      "loss": 0.8594,
      "step": 1547
    },
    {
      "epoch": 0.13893376413570274,
      "grad_norm": 4.909173488616943,
      "learning_rate": 0.0,
      "loss": 0.784,
      "step": 1548
    },
    {
      "epoch": 0.13902351462933046,
      "grad_norm": 7.295698165893555,
      "learning_rate": 0.0,
      "loss": 0.831,
      "step": 1549
    },
    {
      "epoch": 0.13911326512295819,
      "grad_norm": 4.963049411773682,
      "learning_rate": 0.0,
      "loss": 0.8167,
      "step": 1550
    },
    {
      "epoch": 0.13920301561658588,
      "grad_norm": 3.6755576133728027,
      "learning_rate": 0.0,
      "loss": 0.7941,
      "step": 1551
    },
    {
      "epoch": 0.1392927661102136,
      "grad_norm": 4.396621227264404,
      "learning_rate": 0.0,
      "loss": 0.833,
      "step": 1552
    },
    {
      "epoch": 0.13938251660384132,
      "grad_norm": 11.51919174194336,
      "learning_rate": 0.0,
      "loss": 0.8095,
      "step": 1553
    },
    {
      "epoch": 0.13947226709746904,
      "grad_norm": 4.979683876037598,
      "learning_rate": 0.0,
      "loss": 0.8107,
      "step": 1554
    },
    {
      "epoch": 0.13956201759109674,
      "grad_norm": 7.393730640411377,
      "learning_rate": 0.0,
      "loss": 0.7834,
      "step": 1555
    },
    {
      "epoch": 0.13965176808472446,
      "grad_norm": 3.478435516357422,
      "learning_rate": 0.0,
      "loss": 0.831,
      "step": 1556
    },
    {
      "epoch": 0.13974151857835218,
      "grad_norm": 2.964731454849243,
      "learning_rate": 0.0,
      "loss": 0.8591,
      "step": 1557
    },
    {
      "epoch": 0.1398312690719799,
      "grad_norm": 5.179037570953369,
      "learning_rate": 0.0,
      "loss": 0.8021,
      "step": 1558
    },
    {
      "epoch": 0.13992101956560762,
      "grad_norm": 4.303788661956787,
      "learning_rate": 0.0,
      "loss": 0.8272,
      "step": 1559
    },
    {
      "epoch": 0.14001077005923532,
      "grad_norm": 4.74716329574585,
      "learning_rate": 0.0,
      "loss": 0.7867,
      "step": 1560
    },
    {
      "epoch": 0.14010052055286304,
      "grad_norm": 4.038660526275635,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 1561
    },
    {
      "epoch": 0.14019027104649076,
      "grad_norm": 18.817529678344727,
      "learning_rate": 0.0,
      "loss": 0.7648,
      "step": 1562
    },
    {
      "epoch": 0.14028002154011848,
      "grad_norm": 9.704695701599121,
      "learning_rate": 0.0,
      "loss": 0.8023,
      "step": 1563
    },
    {
      "epoch": 0.14036977203374618,
      "grad_norm": 3.5573246479034424,
      "learning_rate": 0.0,
      "loss": 0.8059,
      "step": 1564
    },
    {
      "epoch": 0.1404595225273739,
      "grad_norm": 3.6688570976257324,
      "learning_rate": 0.0,
      "loss": 0.8054,
      "step": 1565
    },
    {
      "epoch": 0.14054927302100162,
      "grad_norm": 4.711482048034668,
      "learning_rate": 0.0,
      "loss": 0.773,
      "step": 1566
    },
    {
      "epoch": 0.14063902351462934,
      "grad_norm": 3.2644646167755127,
      "learning_rate": 0.0,
      "loss": 0.778,
      "step": 1567
    },
    {
      "epoch": 0.14072877400825703,
      "grad_norm": 4.48269510269165,
      "learning_rate": 0.0,
      "loss": 0.8172,
      "step": 1568
    },
    {
      "epoch": 0.14081852450188476,
      "grad_norm": 4.646212100982666,
      "learning_rate": 0.0,
      "loss": 0.8927,
      "step": 1569
    },
    {
      "epoch": 0.14090827499551248,
      "grad_norm": 9.597685813903809,
      "learning_rate": 0.0,
      "loss": 0.8108,
      "step": 1570
    },
    {
      "epoch": 0.1409980254891402,
      "grad_norm": 3.2871968746185303,
      "learning_rate": 0.0,
      "loss": 0.8152,
      "step": 1571
    },
    {
      "epoch": 0.1410877759827679,
      "grad_norm": 11.082829475402832,
      "learning_rate": 0.0,
      "loss": 0.7495,
      "step": 1572
    },
    {
      "epoch": 0.14117752647639562,
      "grad_norm": 4.732816696166992,
      "learning_rate": 0.0,
      "loss": 0.7956,
      "step": 1573
    },
    {
      "epoch": 0.14126727697002334,
      "grad_norm": 3.8509035110473633,
      "learning_rate": 0.0,
      "loss": 0.8568,
      "step": 1574
    },
    {
      "epoch": 0.14135702746365106,
      "grad_norm": 4.12294864654541,
      "learning_rate": 0.0,
      "loss": 0.8358,
      "step": 1575
    },
    {
      "epoch": 0.14144677795727875,
      "grad_norm": 4.781147003173828,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 1576
    },
    {
      "epoch": 0.14153652845090647,
      "grad_norm": 6.953753471374512,
      "learning_rate": 0.0,
      "loss": 0.8345,
      "step": 1577
    },
    {
      "epoch": 0.1416262789445342,
      "grad_norm": 3.165726900100708,
      "learning_rate": 0.0,
      "loss": 0.8188,
      "step": 1578
    },
    {
      "epoch": 0.14171602943816192,
      "grad_norm": 3.806959867477417,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 1579
    },
    {
      "epoch": 0.14180577993178964,
      "grad_norm": 3.8087971210479736,
      "learning_rate": 0.0,
      "loss": 0.7572,
      "step": 1580
    },
    {
      "epoch": 0.14189553042541733,
      "grad_norm": 7.095966815948486,
      "learning_rate": 0.0,
      "loss": 0.842,
      "step": 1581
    },
    {
      "epoch": 0.14198528091904505,
      "grad_norm": 2.5141048431396484,
      "learning_rate": 0.0,
      "loss": 0.7958,
      "step": 1582
    },
    {
      "epoch": 0.14207503141267278,
      "grad_norm": 5.915368556976318,
      "learning_rate": 0.0,
      "loss": 0.7891,
      "step": 1583
    },
    {
      "epoch": 0.1421647819063005,
      "grad_norm": 3.038588047027588,
      "learning_rate": 0.0,
      "loss": 0.8449,
      "step": 1584
    },
    {
      "epoch": 0.1422545323999282,
      "grad_norm": 7.7038750648498535,
      "learning_rate": 0.0,
      "loss": 0.8051,
      "step": 1585
    },
    {
      "epoch": 0.1423442828935559,
      "grad_norm": 6.105831623077393,
      "learning_rate": 0.0,
      "loss": 0.8241,
      "step": 1586
    },
    {
      "epoch": 0.14243403338718363,
      "grad_norm": 7.283270835876465,
      "learning_rate": 0.0,
      "loss": 0.7822,
      "step": 1587
    },
    {
      "epoch": 0.14252378388081136,
      "grad_norm": 5.497828483581543,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 1588
    },
    {
      "epoch": 0.14261353437443905,
      "grad_norm": 3.043389320373535,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 1589
    },
    {
      "epoch": 0.14270328486806677,
      "grad_norm": 4.6543426513671875,
      "learning_rate": 0.0,
      "loss": 0.8189,
      "step": 1590
    },
    {
      "epoch": 0.1427930353616945,
      "grad_norm": 3.5306951999664307,
      "learning_rate": 0.0,
      "loss": 0.8592,
      "step": 1591
    },
    {
      "epoch": 0.14288278585532221,
      "grad_norm": 5.446950435638428,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 1592
    },
    {
      "epoch": 0.1429725363489499,
      "grad_norm": 3.276156187057495,
      "learning_rate": 0.0,
      "loss": 0.8167,
      "step": 1593
    },
    {
      "epoch": 0.14306228684257763,
      "grad_norm": 2.741182327270508,
      "learning_rate": 0.0,
      "loss": 0.8274,
      "step": 1594
    },
    {
      "epoch": 0.14315203733620535,
      "grad_norm": 4.154709339141846,
      "learning_rate": 0.0,
      "loss": 0.7726,
      "step": 1595
    },
    {
      "epoch": 0.14324178782983307,
      "grad_norm": 4.167494297027588,
      "learning_rate": 0.0,
      "loss": 0.8857,
      "step": 1596
    },
    {
      "epoch": 0.14333153832346077,
      "grad_norm": 4.41543436050415,
      "learning_rate": 0.0,
      "loss": 0.7846,
      "step": 1597
    },
    {
      "epoch": 0.1434212888170885,
      "grad_norm": 4.045876979827881,
      "learning_rate": 0.0,
      "loss": 0.8597,
      "step": 1598
    },
    {
      "epoch": 0.1435110393107162,
      "grad_norm": 7.652373313903809,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 1599
    },
    {
      "epoch": 0.14360078980434393,
      "grad_norm": 25.536605834960938,
      "learning_rate": 0.0,
      "loss": 0.7401,
      "step": 1600
    },
    {
      "epoch": 0.14369054029797163,
      "grad_norm": 5.893154144287109,
      "learning_rate": 0.0,
      "loss": 0.8071,
      "step": 1601
    },
    {
      "epoch": 0.14378029079159935,
      "grad_norm": 3.6258296966552734,
      "learning_rate": 0.0,
      "loss": 0.8024,
      "step": 1602
    },
    {
      "epoch": 0.14387004128522707,
      "grad_norm": 4.279819011688232,
      "learning_rate": 0.0,
      "loss": 0.7667,
      "step": 1603
    },
    {
      "epoch": 0.1439597917788548,
      "grad_norm": 3.9120054244995117,
      "learning_rate": 0.0,
      "loss": 0.8397,
      "step": 1604
    },
    {
      "epoch": 0.1440495422724825,
      "grad_norm": 3.741197109222412,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 1605
    },
    {
      "epoch": 0.1441392927661102,
      "grad_norm": 3.7420387268066406,
      "learning_rate": 0.0,
      "loss": 0.8229,
      "step": 1606
    },
    {
      "epoch": 0.14422904325973793,
      "grad_norm": 3.9783883094787598,
      "learning_rate": 0.0,
      "loss": 0.7656,
      "step": 1607
    },
    {
      "epoch": 0.14431879375336565,
      "grad_norm": 4.039865016937256,
      "learning_rate": 0.0,
      "loss": 0.7814,
      "step": 1608
    },
    {
      "epoch": 0.14440854424699337,
      "grad_norm": 3.5531134605407715,
      "learning_rate": 0.0,
      "loss": 0.781,
      "step": 1609
    },
    {
      "epoch": 0.14449829474062106,
      "grad_norm": 12.271679878234863,
      "learning_rate": 0.0,
      "loss": 0.7929,
      "step": 1610
    },
    {
      "epoch": 0.14458804523424879,
      "grad_norm": 6.685312747955322,
      "learning_rate": 0.0,
      "loss": 0.8026,
      "step": 1611
    },
    {
      "epoch": 0.1446777957278765,
      "grad_norm": 3.839932918548584,
      "learning_rate": 0.0,
      "loss": 0.7935,
      "step": 1612
    },
    {
      "epoch": 0.14476754622150423,
      "grad_norm": 8.116487503051758,
      "learning_rate": 0.0,
      "loss": 0.8065,
      "step": 1613
    },
    {
      "epoch": 0.14485729671513192,
      "grad_norm": 4.6377177238464355,
      "learning_rate": 0.0,
      "loss": 0.789,
      "step": 1614
    },
    {
      "epoch": 0.14494704720875964,
      "grad_norm": 3.6983823776245117,
      "learning_rate": 0.0,
      "loss": 0.8154,
      "step": 1615
    },
    {
      "epoch": 0.14503679770238737,
      "grad_norm": 7.846090316772461,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 1616
    },
    {
      "epoch": 0.1451265481960151,
      "grad_norm": 3.534348964691162,
      "learning_rate": 0.0,
      "loss": 0.8378,
      "step": 1617
    },
    {
      "epoch": 0.14521629868964278,
      "grad_norm": 3.287564516067505,
      "learning_rate": 0.0,
      "loss": 0.7879,
      "step": 1618
    },
    {
      "epoch": 0.1453060491832705,
      "grad_norm": 8.37126350402832,
      "learning_rate": 0.0,
      "loss": 0.7965,
      "step": 1619
    },
    {
      "epoch": 0.14539579967689822,
      "grad_norm": 3.3157992362976074,
      "learning_rate": 0.0,
      "loss": 0.8367,
      "step": 1620
    },
    {
      "epoch": 0.14548555017052595,
      "grad_norm": 4.5830888748168945,
      "learning_rate": 0.0,
      "loss": 0.8196,
      "step": 1621
    },
    {
      "epoch": 0.14557530066415364,
      "grad_norm": 3.3094964027404785,
      "learning_rate": 0.0,
      "loss": 0.778,
      "step": 1622
    },
    {
      "epoch": 0.14566505115778136,
      "grad_norm": 3.47428035736084,
      "learning_rate": 0.0,
      "loss": 0.8201,
      "step": 1623
    },
    {
      "epoch": 0.14575480165140908,
      "grad_norm": 4.353351593017578,
      "learning_rate": 0.0,
      "loss": 0.851,
      "step": 1624
    },
    {
      "epoch": 0.1458445521450368,
      "grad_norm": 3.721020460128784,
      "learning_rate": 0.0,
      "loss": 0.8122,
      "step": 1625
    },
    {
      "epoch": 0.14593430263866453,
      "grad_norm": 2.730405569076538,
      "learning_rate": 0.0,
      "loss": 0.7513,
      "step": 1626
    },
    {
      "epoch": 0.14602405313229222,
      "grad_norm": 4.519694805145264,
      "learning_rate": 0.0,
      "loss": 0.789,
      "step": 1627
    },
    {
      "epoch": 0.14611380362591994,
      "grad_norm": 10.715428352355957,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 1628
    },
    {
      "epoch": 0.14620355411954766,
      "grad_norm": 5.629361629486084,
      "learning_rate": 0.0,
      "loss": 0.8321,
      "step": 1629
    },
    {
      "epoch": 0.14629330461317538,
      "grad_norm": 3.036839246749878,
      "learning_rate": 0.0,
      "loss": 0.7756,
      "step": 1630
    },
    {
      "epoch": 0.14638305510680308,
      "grad_norm": 7.28476095199585,
      "learning_rate": 0.0,
      "loss": 0.8051,
      "step": 1631
    },
    {
      "epoch": 0.1464728056004308,
      "grad_norm": 25.96481704711914,
      "learning_rate": 0.0,
      "loss": 0.8092,
      "step": 1632
    },
    {
      "epoch": 0.14656255609405852,
      "grad_norm": 7.1737961769104,
      "learning_rate": 0.0,
      "loss": 0.8766,
      "step": 1633
    },
    {
      "epoch": 0.14665230658768624,
      "grad_norm": 4.369621753692627,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 1634
    },
    {
      "epoch": 0.14674205708131394,
      "grad_norm": 15.297351837158203,
      "learning_rate": 0.0,
      "loss": 0.7681,
      "step": 1635
    },
    {
      "epoch": 0.14683180757494166,
      "grad_norm": 3.2803452014923096,
      "learning_rate": 0.0,
      "loss": 0.8314,
      "step": 1636
    },
    {
      "epoch": 0.14692155806856938,
      "grad_norm": 4.384207725524902,
      "learning_rate": 0.0,
      "loss": 0.8129,
      "step": 1637
    },
    {
      "epoch": 0.1470113085621971,
      "grad_norm": 7.746286392211914,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 1638
    },
    {
      "epoch": 0.1471010590558248,
      "grad_norm": 4.421990394592285,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 1639
    },
    {
      "epoch": 0.14719080954945252,
      "grad_norm": 7.1072869300842285,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 1640
    },
    {
      "epoch": 0.14728056004308024,
      "grad_norm": 3.1918044090270996,
      "learning_rate": 0.0,
      "loss": 0.7859,
      "step": 1641
    },
    {
      "epoch": 0.14737031053670796,
      "grad_norm": 3.207775354385376,
      "learning_rate": 0.0,
      "loss": 0.8469,
      "step": 1642
    },
    {
      "epoch": 0.14746006103033565,
      "grad_norm": 3.5287821292877197,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 1643
    },
    {
      "epoch": 0.14754981152396338,
      "grad_norm": 4.227137565612793,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 1644
    },
    {
      "epoch": 0.1476395620175911,
      "grad_norm": 4.0505242347717285,
      "learning_rate": 0.0,
      "loss": 0.7946,
      "step": 1645
    },
    {
      "epoch": 0.14772931251121882,
      "grad_norm": 6.667533874511719,
      "learning_rate": 0.0,
      "loss": 0.7898,
      "step": 1646
    },
    {
      "epoch": 0.14781906300484654,
      "grad_norm": 3.24300479888916,
      "learning_rate": 0.0,
      "loss": 0.7911,
      "step": 1647
    },
    {
      "epoch": 0.14790881349847423,
      "grad_norm": 7.192837715148926,
      "learning_rate": 0.0,
      "loss": 0.8262,
      "step": 1648
    },
    {
      "epoch": 0.14799856399210196,
      "grad_norm": 3.827972173690796,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 1649
    },
    {
      "epoch": 0.14808831448572968,
      "grad_norm": 3.321739435195923,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 1650
    },
    {
      "epoch": 0.1481780649793574,
      "grad_norm": 4.558859825134277,
      "learning_rate": 0.0,
      "loss": 0.831,
      "step": 1651
    },
    {
      "epoch": 0.1482678154729851,
      "grad_norm": 3.2938287258148193,
      "learning_rate": 0.0,
      "loss": 0.7828,
      "step": 1652
    },
    {
      "epoch": 0.14835756596661281,
      "grad_norm": 5.152710914611816,
      "learning_rate": 0.0,
      "loss": 0.7841,
      "step": 1653
    },
    {
      "epoch": 0.14844731646024054,
      "grad_norm": 3.855210542678833,
      "learning_rate": 0.0,
      "loss": 0.789,
      "step": 1654
    },
    {
      "epoch": 0.14853706695386826,
      "grad_norm": 3.0344510078430176,
      "learning_rate": 0.0,
      "loss": 0.8083,
      "step": 1655
    },
    {
      "epoch": 0.14862681744749595,
      "grad_norm": 3.883857250213623,
      "learning_rate": 0.0,
      "loss": 0.8168,
      "step": 1656
    },
    {
      "epoch": 0.14871656794112367,
      "grad_norm": 4.290075778961182,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 1657
    },
    {
      "epoch": 0.1488063184347514,
      "grad_norm": 3.861083984375,
      "learning_rate": 0.0,
      "loss": 0.7808,
      "step": 1658
    },
    {
      "epoch": 0.14889606892837912,
      "grad_norm": 7.683670520782471,
      "learning_rate": 0.0,
      "loss": 0.7868,
      "step": 1659
    },
    {
      "epoch": 0.1489858194220068,
      "grad_norm": 4.2978315353393555,
      "learning_rate": 0.0,
      "loss": 0.8114,
      "step": 1660
    },
    {
      "epoch": 0.14907556991563453,
      "grad_norm": 7.791370391845703,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 1661
    },
    {
      "epoch": 0.14916532040926225,
      "grad_norm": 2.822822332382202,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 1662
    },
    {
      "epoch": 0.14925507090288997,
      "grad_norm": 2.7333691120147705,
      "learning_rate": 0.0,
      "loss": 0.786,
      "step": 1663
    },
    {
      "epoch": 0.14934482139651767,
      "grad_norm": 3.9663589000701904,
      "learning_rate": 0.0,
      "loss": 0.8161,
      "step": 1664
    },
    {
      "epoch": 0.1494345718901454,
      "grad_norm": 4.890150547027588,
      "learning_rate": 0.0,
      "loss": 0.7988,
      "step": 1665
    },
    {
      "epoch": 0.1495243223837731,
      "grad_norm": 3.2683117389678955,
      "learning_rate": 0.0,
      "loss": 0.8101,
      "step": 1666
    },
    {
      "epoch": 0.14961407287740083,
      "grad_norm": 5.072534084320068,
      "learning_rate": 0.0,
      "loss": 0.7899,
      "step": 1667
    },
    {
      "epoch": 0.14970382337102853,
      "grad_norm": 3.1176247596740723,
      "learning_rate": 0.0,
      "loss": 0.8513,
      "step": 1668
    },
    {
      "epoch": 0.14979357386465625,
      "grad_norm": 5.667816162109375,
      "learning_rate": 0.0,
      "loss": 0.7958,
      "step": 1669
    },
    {
      "epoch": 0.14988332435828397,
      "grad_norm": 2.958109140396118,
      "learning_rate": 0.0,
      "loss": 0.8574,
      "step": 1670
    },
    {
      "epoch": 0.1499730748519117,
      "grad_norm": 3.3279457092285156,
      "learning_rate": 0.0,
      "loss": 0.7891,
      "step": 1671
    },
    {
      "epoch": 0.1500628253455394,
      "grad_norm": 3.854609489440918,
      "learning_rate": 0.0,
      "loss": 0.7935,
      "step": 1672
    },
    {
      "epoch": 0.1501525758391671,
      "grad_norm": 4.96049165725708,
      "learning_rate": 0.0,
      "loss": 0.79,
      "step": 1673
    },
    {
      "epoch": 0.15024232633279483,
      "grad_norm": 3.576798677444458,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 1674
    },
    {
      "epoch": 0.15033207682642255,
      "grad_norm": 4.145639419555664,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 1675
    },
    {
      "epoch": 0.15042182732005027,
      "grad_norm": 3.5421721935272217,
      "learning_rate": 0.0,
      "loss": 0.8181,
      "step": 1676
    },
    {
      "epoch": 0.15051157781367797,
      "grad_norm": 4.432629585266113,
      "learning_rate": 0.0,
      "loss": 0.7993,
      "step": 1677
    },
    {
      "epoch": 0.1506013283073057,
      "grad_norm": 3.9923694133758545,
      "learning_rate": 0.0,
      "loss": 0.7959,
      "step": 1678
    },
    {
      "epoch": 0.1506910788009334,
      "grad_norm": 3.7088916301727295,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 1679
    },
    {
      "epoch": 0.15078082929456113,
      "grad_norm": 29.222665786743164,
      "learning_rate": 0.0,
      "loss": 0.7843,
      "step": 1680
    },
    {
      "epoch": 0.15087057978818882,
      "grad_norm": 5.949617385864258,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 1681
    },
    {
      "epoch": 0.15096033028181655,
      "grad_norm": 3.8560678958892822,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 1682
    },
    {
      "epoch": 0.15105008077544427,
      "grad_norm": 6.5609211921691895,
      "learning_rate": 0.0,
      "loss": 0.8162,
      "step": 1683
    },
    {
      "epoch": 0.151139831269072,
      "grad_norm": 2.7586677074432373,
      "learning_rate": 0.0,
      "loss": 0.769,
      "step": 1684
    },
    {
      "epoch": 0.15122958176269968,
      "grad_norm": 2.6363415718078613,
      "learning_rate": 0.0,
      "loss": 0.7964,
      "step": 1685
    },
    {
      "epoch": 0.1513193322563274,
      "grad_norm": 5.460452079772949,
      "learning_rate": 0.0,
      "loss": 0.8244,
      "step": 1686
    },
    {
      "epoch": 0.15140908274995513,
      "grad_norm": 3.328599691390991,
      "learning_rate": 0.0,
      "loss": 0.8299,
      "step": 1687
    },
    {
      "epoch": 0.15149883324358285,
      "grad_norm": 11.345171928405762,
      "learning_rate": 0.0,
      "loss": 0.8269,
      "step": 1688
    },
    {
      "epoch": 0.15158858373721054,
      "grad_norm": 3.9467780590057373,
      "learning_rate": 0.0,
      "loss": 0.8425,
      "step": 1689
    },
    {
      "epoch": 0.15167833423083826,
      "grad_norm": 2.852726697921753,
      "learning_rate": 0.0,
      "loss": 0.7774,
      "step": 1690
    },
    {
      "epoch": 0.15176808472446598,
      "grad_norm": 3.3252484798431396,
      "learning_rate": 0.0,
      "loss": 0.7955,
      "step": 1691
    },
    {
      "epoch": 0.1518578352180937,
      "grad_norm": 3.3755686283111572,
      "learning_rate": 0.0,
      "loss": 0.8117,
      "step": 1692
    },
    {
      "epoch": 0.15194758571172143,
      "grad_norm": 3.765089511871338,
      "learning_rate": 0.0,
      "loss": 0.8136,
      "step": 1693
    },
    {
      "epoch": 0.15203733620534912,
      "grad_norm": 2.6562349796295166,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 1694
    },
    {
      "epoch": 0.15212708669897684,
      "grad_norm": 2.8298285007476807,
      "learning_rate": 0.0,
      "loss": 0.7797,
      "step": 1695
    },
    {
      "epoch": 0.15221683719260456,
      "grad_norm": 5.467107772827148,
      "learning_rate": 0.0,
      "loss": 0.7672,
      "step": 1696
    },
    {
      "epoch": 0.1523065876862323,
      "grad_norm": 6.168618202209473,
      "learning_rate": 0.0,
      "loss": 0.8481,
      "step": 1697
    },
    {
      "epoch": 0.15239633817985998,
      "grad_norm": 5.551270961761475,
      "learning_rate": 0.0,
      "loss": 0.8512,
      "step": 1698
    },
    {
      "epoch": 0.1524860886734877,
      "grad_norm": 2.9026052951812744,
      "learning_rate": 0.0,
      "loss": 0.8411,
      "step": 1699
    },
    {
      "epoch": 0.15257583916711542,
      "grad_norm": 3.093459367752075,
      "learning_rate": 0.0,
      "loss": 0.8593,
      "step": 1700
    },
    {
      "epoch": 0.15266558966074315,
      "grad_norm": 5.89598274230957,
      "learning_rate": 0.0,
      "loss": 0.8331,
      "step": 1701
    },
    {
      "epoch": 0.15275534015437084,
      "grad_norm": 4.613770961761475,
      "learning_rate": 0.0,
      "loss": 0.7751,
      "step": 1702
    },
    {
      "epoch": 0.15284509064799856,
      "grad_norm": 4.591054439544678,
      "learning_rate": 0.0,
      "loss": 0.8208,
      "step": 1703
    },
    {
      "epoch": 0.15293484114162628,
      "grad_norm": 3.612384080886841,
      "learning_rate": 0.0,
      "loss": 0.9018,
      "step": 1704
    },
    {
      "epoch": 0.153024591635254,
      "grad_norm": 3.015645742416382,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 1705
    },
    {
      "epoch": 0.1531143421288817,
      "grad_norm": 3.4195637702941895,
      "learning_rate": 0.0,
      "loss": 0.8045,
      "step": 1706
    },
    {
      "epoch": 0.15320409262250942,
      "grad_norm": 6.175956726074219,
      "learning_rate": 0.0,
      "loss": 0.7788,
      "step": 1707
    },
    {
      "epoch": 0.15329384311613714,
      "grad_norm": 3.3825433254241943,
      "learning_rate": 0.0,
      "loss": 0.8478,
      "step": 1708
    },
    {
      "epoch": 0.15338359360976486,
      "grad_norm": 5.388416767120361,
      "learning_rate": 0.0,
      "loss": 0.8216,
      "step": 1709
    },
    {
      "epoch": 0.15347334410339256,
      "grad_norm": 4.798441410064697,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 1710
    },
    {
      "epoch": 0.15356309459702028,
      "grad_norm": 4.211033821105957,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 1711
    },
    {
      "epoch": 0.153652845090648,
      "grad_norm": 4.276221752166748,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 1712
    },
    {
      "epoch": 0.15374259558427572,
      "grad_norm": 7.0431694984436035,
      "learning_rate": 0.0,
      "loss": 0.7821,
      "step": 1713
    },
    {
      "epoch": 0.15383234607790341,
      "grad_norm": 3.2654035091400146,
      "learning_rate": 0.0,
      "loss": 0.7869,
      "step": 1714
    },
    {
      "epoch": 0.15392209657153114,
      "grad_norm": 7.159914493560791,
      "learning_rate": 0.0,
      "loss": 0.8537,
      "step": 1715
    },
    {
      "epoch": 0.15401184706515886,
      "grad_norm": 10.235418319702148,
      "learning_rate": 0.0,
      "loss": 0.7573,
      "step": 1716
    },
    {
      "epoch": 0.15410159755878658,
      "grad_norm": 3.8433969020843506,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 1717
    },
    {
      "epoch": 0.1541913480524143,
      "grad_norm": 2.8981969356536865,
      "learning_rate": 0.0,
      "loss": 0.8086,
      "step": 1718
    },
    {
      "epoch": 0.154281098546042,
      "grad_norm": 4.174559593200684,
      "learning_rate": 0.0,
      "loss": 0.7401,
      "step": 1719
    },
    {
      "epoch": 0.15437084903966972,
      "grad_norm": 5.771835803985596,
      "learning_rate": 0.0,
      "loss": 0.7279,
      "step": 1720
    },
    {
      "epoch": 0.15446059953329744,
      "grad_norm": 3.0991029739379883,
      "learning_rate": 0.0,
      "loss": 0.7851,
      "step": 1721
    },
    {
      "epoch": 0.15455035002692516,
      "grad_norm": 6.551607131958008,
      "learning_rate": 0.0,
      "loss": 0.799,
      "step": 1722
    },
    {
      "epoch": 0.15464010052055285,
      "grad_norm": 2.7035865783691406,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 1723
    },
    {
      "epoch": 0.15472985101418057,
      "grad_norm": 6.7690863609313965,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 1724
    },
    {
      "epoch": 0.1548196015078083,
      "grad_norm": 2.7554290294647217,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 1725
    },
    {
      "epoch": 0.15490935200143602,
      "grad_norm": 5.31464958190918,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 1726
    },
    {
      "epoch": 0.1549991024950637,
      "grad_norm": 3.380007028579712,
      "learning_rate": 0.0,
      "loss": 0.7994,
      "step": 1727
    },
    {
      "epoch": 0.15508885298869143,
      "grad_norm": 3.0180323123931885,
      "learning_rate": 0.0,
      "loss": 0.8025,
      "step": 1728
    },
    {
      "epoch": 0.15517860348231916,
      "grad_norm": 4.291436672210693,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 1729
    },
    {
      "epoch": 0.15526835397594688,
      "grad_norm": 2.6152877807617188,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 1730
    },
    {
      "epoch": 0.15535810446957457,
      "grad_norm": 2.84256911277771,
      "learning_rate": 0.0,
      "loss": 0.7897,
      "step": 1731
    },
    {
      "epoch": 0.1554478549632023,
      "grad_norm": 2.889802932739258,
      "learning_rate": 0.0,
      "loss": 0.8085,
      "step": 1732
    },
    {
      "epoch": 0.15553760545683,
      "grad_norm": 3.4670422077178955,
      "learning_rate": 0.0,
      "loss": 0.8141,
      "step": 1733
    },
    {
      "epoch": 0.15562735595045774,
      "grad_norm": 3.9354429244995117,
      "learning_rate": 0.0,
      "loss": 0.8492,
      "step": 1734
    },
    {
      "epoch": 0.15571710644408543,
      "grad_norm": 3.8121209144592285,
      "learning_rate": 0.0,
      "loss": 0.7942,
      "step": 1735
    },
    {
      "epoch": 0.15580685693771315,
      "grad_norm": 3.2494168281555176,
      "learning_rate": 0.0,
      "loss": 0.8071,
      "step": 1736
    },
    {
      "epoch": 0.15589660743134087,
      "grad_norm": 2.994842529296875,
      "learning_rate": 0.0,
      "loss": 0.8044,
      "step": 1737
    },
    {
      "epoch": 0.1559863579249686,
      "grad_norm": 3.1958913803100586,
      "learning_rate": 0.0,
      "loss": 0.7951,
      "step": 1738
    },
    {
      "epoch": 0.15607610841859632,
      "grad_norm": 3.0610580444335938,
      "learning_rate": 0.0,
      "loss": 0.7557,
      "step": 1739
    },
    {
      "epoch": 0.156165858912224,
      "grad_norm": 3.7901902198791504,
      "learning_rate": 0.0,
      "loss": 0.7432,
      "step": 1740
    },
    {
      "epoch": 0.15625560940585173,
      "grad_norm": 4.628288269042969,
      "learning_rate": 0.0,
      "loss": 0.7855,
      "step": 1741
    },
    {
      "epoch": 0.15634535989947945,
      "grad_norm": 2.7967779636383057,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 1742
    },
    {
      "epoch": 0.15643511039310717,
      "grad_norm": 3.8926477432250977,
      "learning_rate": 0.0,
      "loss": 0.8156,
      "step": 1743
    },
    {
      "epoch": 0.15652486088673487,
      "grad_norm": 4.160140037536621,
      "learning_rate": 0.0,
      "loss": 0.7709,
      "step": 1744
    },
    {
      "epoch": 0.1566146113803626,
      "grad_norm": 3.5279860496520996,
      "learning_rate": 0.0,
      "loss": 0.8058,
      "step": 1745
    },
    {
      "epoch": 0.1567043618739903,
      "grad_norm": 3.3276641368865967,
      "learning_rate": 0.0,
      "loss": 0.7989,
      "step": 1746
    },
    {
      "epoch": 0.15679411236761803,
      "grad_norm": 3.2731785774230957,
      "learning_rate": 0.0,
      "loss": 0.7768,
      "step": 1747
    },
    {
      "epoch": 0.15688386286124573,
      "grad_norm": 15.429272651672363,
      "learning_rate": 0.0,
      "loss": 0.8032,
      "step": 1748
    },
    {
      "epoch": 0.15697361335487345,
      "grad_norm": 3.645705461502075,
      "learning_rate": 0.0,
      "loss": 0.8441,
      "step": 1749
    },
    {
      "epoch": 0.15706336384850117,
      "grad_norm": 9.721328735351562,
      "learning_rate": 0.0,
      "loss": 0.7602,
      "step": 1750
    },
    {
      "epoch": 0.1571531143421289,
      "grad_norm": 4.891539096832275,
      "learning_rate": 0.0,
      "loss": 0.8175,
      "step": 1751
    },
    {
      "epoch": 0.15724286483575659,
      "grad_norm": 3.524484634399414,
      "learning_rate": 0.0,
      "loss": 0.7961,
      "step": 1752
    },
    {
      "epoch": 0.1573326153293843,
      "grad_norm": 4.091954708099365,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 1753
    },
    {
      "epoch": 0.15742236582301203,
      "grad_norm": 5.298373222351074,
      "learning_rate": 0.0,
      "loss": 0.8001,
      "step": 1754
    },
    {
      "epoch": 0.15751211631663975,
      "grad_norm": 3.4641168117523193,
      "learning_rate": 0.0,
      "loss": 0.7956,
      "step": 1755
    },
    {
      "epoch": 0.15760186681026744,
      "grad_norm": 3.671964645385742,
      "learning_rate": 0.0,
      "loss": 0.8198,
      "step": 1756
    },
    {
      "epoch": 0.15769161730389517,
      "grad_norm": 10.806663513183594,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 1757
    },
    {
      "epoch": 0.1577813677975229,
      "grad_norm": 6.513108730316162,
      "learning_rate": 0.0,
      "loss": 0.8236,
      "step": 1758
    },
    {
      "epoch": 0.1578711182911506,
      "grad_norm": 5.007513046264648,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 1759
    },
    {
      "epoch": 0.15796086878477833,
      "grad_norm": 4.955020427703857,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 1760
    },
    {
      "epoch": 0.15805061927840602,
      "grad_norm": 5.552761554718018,
      "learning_rate": 0.0,
      "loss": 0.7935,
      "step": 1761
    },
    {
      "epoch": 0.15814036977203375,
      "grad_norm": 2.6310245990753174,
      "learning_rate": 0.0,
      "loss": 0.8184,
      "step": 1762
    },
    {
      "epoch": 0.15823012026566147,
      "grad_norm": 3.124467611312866,
      "learning_rate": 0.0,
      "loss": 0.7848,
      "step": 1763
    },
    {
      "epoch": 0.1583198707592892,
      "grad_norm": 3.116652250289917,
      "learning_rate": 0.0,
      "loss": 0.8164,
      "step": 1764
    },
    {
      "epoch": 0.15840962125291688,
      "grad_norm": 3.940619945526123,
      "learning_rate": 0.0,
      "loss": 0.7941,
      "step": 1765
    },
    {
      "epoch": 0.1584993717465446,
      "grad_norm": 2.6745495796203613,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 1766
    },
    {
      "epoch": 0.15858912224017233,
      "grad_norm": 3.083089590072632,
      "learning_rate": 0.0,
      "loss": 0.8298,
      "step": 1767
    },
    {
      "epoch": 0.15867887273380005,
      "grad_norm": 4.7042317390441895,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 1768
    },
    {
      "epoch": 0.15876862322742774,
      "grad_norm": 5.493862628936768,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 1769
    },
    {
      "epoch": 0.15885837372105546,
      "grad_norm": 3.794808864593506,
      "learning_rate": 0.0,
      "loss": 0.8016,
      "step": 1770
    },
    {
      "epoch": 0.15894812421468318,
      "grad_norm": 2.9495465755462646,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 1771
    },
    {
      "epoch": 0.1590378747083109,
      "grad_norm": 7.252605438232422,
      "learning_rate": 0.0,
      "loss": 0.7784,
      "step": 1772
    },
    {
      "epoch": 0.1591276252019386,
      "grad_norm": 6.2665252685546875,
      "learning_rate": 0.0,
      "loss": 0.7344,
      "step": 1773
    },
    {
      "epoch": 0.15921737569556632,
      "grad_norm": 4.35286283493042,
      "learning_rate": 0.0,
      "loss": 0.7868,
      "step": 1774
    },
    {
      "epoch": 0.15930712618919404,
      "grad_norm": 7.175297260284424,
      "learning_rate": 0.0,
      "loss": 0.7565,
      "step": 1775
    },
    {
      "epoch": 0.15939687668282176,
      "grad_norm": 12.949705123901367,
      "learning_rate": 0.0,
      "loss": 0.7983,
      "step": 1776
    },
    {
      "epoch": 0.15948662717644946,
      "grad_norm": 4.243785858154297,
      "learning_rate": 0.0,
      "loss": 0.7891,
      "step": 1777
    },
    {
      "epoch": 0.15957637767007718,
      "grad_norm": 4.409718990325928,
      "learning_rate": 0.0,
      "loss": 0.8151,
      "step": 1778
    },
    {
      "epoch": 0.1596661281637049,
      "grad_norm": 3.1254074573516846,
      "learning_rate": 0.0,
      "loss": 0.7832,
      "step": 1779
    },
    {
      "epoch": 0.15975587865733262,
      "grad_norm": 3.7440946102142334,
      "learning_rate": 0.0,
      "loss": 0.8046,
      "step": 1780
    },
    {
      "epoch": 0.15984562915096032,
      "grad_norm": 8.800244331359863,
      "learning_rate": 0.0,
      "loss": 0.7677,
      "step": 1781
    },
    {
      "epoch": 0.15993537964458804,
      "grad_norm": 9.960772514343262,
      "learning_rate": 0.0,
      "loss": 0.8187,
      "step": 1782
    },
    {
      "epoch": 0.16002513013821576,
      "grad_norm": 3.207315444946289,
      "learning_rate": 0.0,
      "loss": 0.7868,
      "step": 1783
    },
    {
      "epoch": 0.16011488063184348,
      "grad_norm": 3.2560150623321533,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 1784
    },
    {
      "epoch": 0.1602046311254712,
      "grad_norm": 3.3774914741516113,
      "learning_rate": 0.0,
      "loss": 0.8537,
      "step": 1785
    },
    {
      "epoch": 0.1602943816190989,
      "grad_norm": 5.568129539489746,
      "learning_rate": 0.0,
      "loss": 0.8231,
      "step": 1786
    },
    {
      "epoch": 0.16038413211272662,
      "grad_norm": 2.655782699584961,
      "learning_rate": 0.0,
      "loss": 0.7883,
      "step": 1787
    },
    {
      "epoch": 0.16047388260635434,
      "grad_norm": 3.935978651046753,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 1788
    },
    {
      "epoch": 0.16056363309998206,
      "grad_norm": 4.343422889709473,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 1789
    },
    {
      "epoch": 0.16065338359360976,
      "grad_norm": 3.120788812637329,
      "learning_rate": 0.0,
      "loss": 0.784,
      "step": 1790
    },
    {
      "epoch": 0.16074313408723748,
      "grad_norm": 4.087812423706055,
      "learning_rate": 0.0,
      "loss": 0.8206,
      "step": 1791
    },
    {
      "epoch": 0.1608328845808652,
      "grad_norm": 4.224339485168457,
      "learning_rate": 0.0,
      "loss": 0.7843,
      "step": 1792
    },
    {
      "epoch": 0.16092263507449292,
      "grad_norm": 3.3979742527008057,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 1793
    },
    {
      "epoch": 0.16101238556812061,
      "grad_norm": 5.035244464874268,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 1794
    },
    {
      "epoch": 0.16110213606174834,
      "grad_norm": 12.330953598022461,
      "learning_rate": 0.0,
      "loss": 0.751,
      "step": 1795
    },
    {
      "epoch": 0.16119188655537606,
      "grad_norm": 5.232416152954102,
      "learning_rate": 0.0,
      "loss": 0.793,
      "step": 1796
    },
    {
      "epoch": 0.16128163704900378,
      "grad_norm": 6.4269561767578125,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 1797
    },
    {
      "epoch": 0.16137138754263147,
      "grad_norm": 3.748102903366089,
      "learning_rate": 0.0,
      "loss": 0.793,
      "step": 1798
    },
    {
      "epoch": 0.1614611380362592,
      "grad_norm": 5.09131383895874,
      "learning_rate": 0.0,
      "loss": 0.8054,
      "step": 1799
    },
    {
      "epoch": 0.16155088852988692,
      "grad_norm": 3.6300160884857178,
      "learning_rate": 0.0,
      "loss": 0.7893,
      "step": 1800
    },
    {
      "epoch": 0.16164063902351464,
      "grad_norm": 9.288633346557617,
      "learning_rate": 0.0,
      "loss": 0.7939,
      "step": 1801
    },
    {
      "epoch": 0.16173038951714233,
      "grad_norm": 5.205272674560547,
      "learning_rate": 0.0,
      "loss": 0.813,
      "step": 1802
    },
    {
      "epoch": 0.16182014001077005,
      "grad_norm": 3.859053373336792,
      "learning_rate": 0.0,
      "loss": 0.838,
      "step": 1803
    },
    {
      "epoch": 0.16190989050439777,
      "grad_norm": 4.700564861297607,
      "learning_rate": 0.0,
      "loss": 0.7874,
      "step": 1804
    },
    {
      "epoch": 0.1619996409980255,
      "grad_norm": 2.8008580207824707,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 1805
    },
    {
      "epoch": 0.16208939149165322,
      "grad_norm": 6.118667125701904,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 1806
    },
    {
      "epoch": 0.1621791419852809,
      "grad_norm": 8.038151741027832,
      "learning_rate": 0.0,
      "loss": 0.7738,
      "step": 1807
    },
    {
      "epoch": 0.16226889247890863,
      "grad_norm": 5.364910125732422,
      "learning_rate": 0.0,
      "loss": 0.7658,
      "step": 1808
    },
    {
      "epoch": 0.16235864297253635,
      "grad_norm": 2.448786735534668,
      "learning_rate": 0.0,
      "loss": 0.7798,
      "step": 1809
    },
    {
      "epoch": 0.16244839346616408,
      "grad_norm": 30.005014419555664,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 1810
    },
    {
      "epoch": 0.16253814395979177,
      "grad_norm": 4.796141147613525,
      "learning_rate": 0.0,
      "loss": 0.831,
      "step": 1811
    },
    {
      "epoch": 0.1626278944534195,
      "grad_norm": 3.368201494216919,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 1812
    },
    {
      "epoch": 0.1627176449470472,
      "grad_norm": 4.619268417358398,
      "learning_rate": 0.0,
      "loss": 0.8154,
      "step": 1813
    },
    {
      "epoch": 0.16280739544067493,
      "grad_norm": 2.8208041191101074,
      "learning_rate": 0.0,
      "loss": 0.8118,
      "step": 1814
    },
    {
      "epoch": 0.16289714593430263,
      "grad_norm": 5.52925968170166,
      "learning_rate": 0.0,
      "loss": 0.7661,
      "step": 1815
    },
    {
      "epoch": 0.16298689642793035,
      "grad_norm": 4.328476428985596,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 1816
    },
    {
      "epoch": 0.16307664692155807,
      "grad_norm": 2.3119418621063232,
      "learning_rate": 0.0,
      "loss": 0.8211,
      "step": 1817
    },
    {
      "epoch": 0.1631663974151858,
      "grad_norm": 4.263214111328125,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 1818
    },
    {
      "epoch": 0.1632561479088135,
      "grad_norm": 7.012444972991943,
      "learning_rate": 0.0,
      "loss": 0.7794,
      "step": 1819
    },
    {
      "epoch": 0.1633458984024412,
      "grad_norm": 3.35213303565979,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 1820
    },
    {
      "epoch": 0.16343564889606893,
      "grad_norm": 15.713327407836914,
      "learning_rate": 0.0,
      "loss": 0.7618,
      "step": 1821
    },
    {
      "epoch": 0.16352539938969665,
      "grad_norm": 4.89967679977417,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 1822
    },
    {
      "epoch": 0.16361514988332435,
      "grad_norm": 4.489142417907715,
      "learning_rate": 0.0,
      "loss": 0.8248,
      "step": 1823
    },
    {
      "epoch": 0.16370490037695207,
      "grad_norm": 3.3560681343078613,
      "learning_rate": 0.0,
      "loss": 0.8116,
      "step": 1824
    },
    {
      "epoch": 0.1637946508705798,
      "grad_norm": 2.568716287612915,
      "learning_rate": 0.0,
      "loss": 0.7877,
      "step": 1825
    },
    {
      "epoch": 0.1638844013642075,
      "grad_norm": 2.4145524501800537,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 1826
    },
    {
      "epoch": 0.16397415185783523,
      "grad_norm": 2.5573930740356445,
      "learning_rate": 0.0,
      "loss": 0.7845,
      "step": 1827
    },
    {
      "epoch": 0.16406390235146293,
      "grad_norm": 120.91207122802734,
      "learning_rate": 0.0,
      "loss": 0.8197,
      "step": 1828
    },
    {
      "epoch": 0.16415365284509065,
      "grad_norm": 2.981259822845459,
      "learning_rate": 0.0,
      "loss": 0.7882,
      "step": 1829
    },
    {
      "epoch": 0.16424340333871837,
      "grad_norm": 2.5592024326324463,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 1830
    },
    {
      "epoch": 0.1643331538323461,
      "grad_norm": 44.58485794067383,
      "learning_rate": 0.0,
      "loss": 0.8099,
      "step": 1831
    },
    {
      "epoch": 0.16442290432597378,
      "grad_norm": 3.63891339302063,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 1832
    },
    {
      "epoch": 0.1645126548196015,
      "grad_norm": 5.275019645690918,
      "learning_rate": 0.0,
      "loss": 0.7692,
      "step": 1833
    },
    {
      "epoch": 0.16460240531322923,
      "grad_norm": 3.1583025455474854,
      "learning_rate": 0.0,
      "loss": 0.7728,
      "step": 1834
    },
    {
      "epoch": 0.16469215580685695,
      "grad_norm": 2.911226511001587,
      "learning_rate": 0.0,
      "loss": 0.8158,
      "step": 1835
    },
    {
      "epoch": 0.16478190630048464,
      "grad_norm": 3.957606315612793,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 1836
    },
    {
      "epoch": 0.16487165679411236,
      "grad_norm": 3.828129529953003,
      "learning_rate": 0.0,
      "loss": 0.8495,
      "step": 1837
    },
    {
      "epoch": 0.16496140728774009,
      "grad_norm": 3.223177671432495,
      "learning_rate": 0.0,
      "loss": 0.7787,
      "step": 1838
    },
    {
      "epoch": 0.1650511577813678,
      "grad_norm": 4.2502617835998535,
      "learning_rate": 0.0,
      "loss": 0.7564,
      "step": 1839
    },
    {
      "epoch": 0.1651409082749955,
      "grad_norm": 9.311372756958008,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 1840
    },
    {
      "epoch": 0.16523065876862322,
      "grad_norm": 3.587541103363037,
      "learning_rate": 0.0,
      "loss": 0.7808,
      "step": 1841
    },
    {
      "epoch": 0.16532040926225094,
      "grad_norm": 4.0654120445251465,
      "learning_rate": 0.0,
      "loss": 0.8028,
      "step": 1842
    },
    {
      "epoch": 0.16541015975587867,
      "grad_norm": 4.279010772705078,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 1843
    },
    {
      "epoch": 0.16549991024950636,
      "grad_norm": 2.704824447631836,
      "learning_rate": 0.0,
      "loss": 0.8208,
      "step": 1844
    },
    {
      "epoch": 0.16558966074313408,
      "grad_norm": 2.4772143363952637,
      "learning_rate": 0.0,
      "loss": 0.8084,
      "step": 1845
    },
    {
      "epoch": 0.1656794112367618,
      "grad_norm": 2.899184465408325,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 1846
    },
    {
      "epoch": 0.16576916173038952,
      "grad_norm": 3.1223573684692383,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 1847
    },
    {
      "epoch": 0.16585891222401722,
      "grad_norm": 2.989936590194702,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 1848
    },
    {
      "epoch": 0.16594866271764494,
      "grad_norm": 2.973684549331665,
      "learning_rate": 0.0,
      "loss": 0.7828,
      "step": 1849
    },
    {
      "epoch": 0.16603841321127266,
      "grad_norm": 11.461050987243652,
      "learning_rate": 0.0,
      "loss": 0.8023,
      "step": 1850
    },
    {
      "epoch": 0.16612816370490038,
      "grad_norm": 3.2302637100219727,
      "learning_rate": 0.0,
      "loss": 0.7789,
      "step": 1851
    },
    {
      "epoch": 0.1662179141985281,
      "grad_norm": 4.154444217681885,
      "learning_rate": 0.0,
      "loss": 0.832,
      "step": 1852
    },
    {
      "epoch": 0.1663076646921558,
      "grad_norm": 4.866849422454834,
      "learning_rate": 0.0,
      "loss": 0.7796,
      "step": 1853
    },
    {
      "epoch": 0.16639741518578352,
      "grad_norm": 2.3121321201324463,
      "learning_rate": 0.0,
      "loss": 0.7893,
      "step": 1854
    },
    {
      "epoch": 0.16648716567941124,
      "grad_norm": 2.8289382457733154,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 1855
    },
    {
      "epoch": 0.16657691617303896,
      "grad_norm": 19.892005920410156,
      "learning_rate": 0.0,
      "loss": 0.7594,
      "step": 1856
    },
    {
      "epoch": 0.16666666666666666,
      "grad_norm": 5.965771675109863,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 1857
    },
    {
      "epoch": 0.16675641716029438,
      "grad_norm": 2.622466564178467,
      "learning_rate": 0.0,
      "loss": 0.8133,
      "step": 1858
    },
    {
      "epoch": 0.1668461676539221,
      "grad_norm": 8.798226356506348,
      "learning_rate": 0.0,
      "loss": 0.8167,
      "step": 1859
    },
    {
      "epoch": 0.16693591814754982,
      "grad_norm": 3.1406946182250977,
      "learning_rate": 0.0,
      "loss": 0.7894,
      "step": 1860
    },
    {
      "epoch": 0.16702566864117752,
      "grad_norm": 3.3449511528015137,
      "learning_rate": 0.0,
      "loss": 0.7851,
      "step": 1861
    },
    {
      "epoch": 0.16711541913480524,
      "grad_norm": 8.03305721282959,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 1862
    },
    {
      "epoch": 0.16720516962843296,
      "grad_norm": 4.1484904289245605,
      "learning_rate": 0.0,
      "loss": 0.7681,
      "step": 1863
    },
    {
      "epoch": 0.16729492012206068,
      "grad_norm": 2.9048707485198975,
      "learning_rate": 0.0,
      "loss": 0.8191,
      "step": 1864
    },
    {
      "epoch": 0.16738467061568837,
      "grad_norm": 3.744856357574463,
      "learning_rate": 0.0,
      "loss": 0.7818,
      "step": 1865
    },
    {
      "epoch": 0.1674744211093161,
      "grad_norm": 6.859939098358154,
      "learning_rate": 0.0,
      "loss": 0.7896,
      "step": 1866
    },
    {
      "epoch": 0.16756417160294382,
      "grad_norm": 3.4023823738098145,
      "learning_rate": 0.0,
      "loss": 0.8145,
      "step": 1867
    },
    {
      "epoch": 0.16765392209657154,
      "grad_norm": 12.4747314453125,
      "learning_rate": 0.0,
      "loss": 0.8148,
      "step": 1868
    },
    {
      "epoch": 0.16774367259019923,
      "grad_norm": 3.765449047088623,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 1869
    },
    {
      "epoch": 0.16783342308382695,
      "grad_norm": 2.792480707168579,
      "learning_rate": 0.0,
      "loss": 0.7822,
      "step": 1870
    },
    {
      "epoch": 0.16792317357745468,
      "grad_norm": 3.471836566925049,
      "learning_rate": 0.0,
      "loss": 0.811,
      "step": 1871
    },
    {
      "epoch": 0.1680129240710824,
      "grad_norm": 3.0053787231445312,
      "learning_rate": 0.0,
      "loss": 0.8057,
      "step": 1872
    },
    {
      "epoch": 0.16810267456471012,
      "grad_norm": 2.807140588760376,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 1873
    },
    {
      "epoch": 0.1681924250583378,
      "grad_norm": 2.920656681060791,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 1874
    },
    {
      "epoch": 0.16828217555196553,
      "grad_norm": 3.236095428466797,
      "learning_rate": 0.0,
      "loss": 0.7542,
      "step": 1875
    },
    {
      "epoch": 0.16837192604559326,
      "grad_norm": 3.0518689155578613,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 1876
    },
    {
      "epoch": 0.16846167653922098,
      "grad_norm": 2.888227701187134,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 1877
    },
    {
      "epoch": 0.16855142703284867,
      "grad_norm": 2.8387184143066406,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 1878
    },
    {
      "epoch": 0.1686411775264764,
      "grad_norm": 4.105151653289795,
      "learning_rate": 0.0,
      "loss": 0.8178,
      "step": 1879
    },
    {
      "epoch": 0.16873092802010412,
      "grad_norm": 2.9738025665283203,
      "learning_rate": 0.0,
      "loss": 0.8205,
      "step": 1880
    },
    {
      "epoch": 0.16882067851373184,
      "grad_norm": 2.9246160984039307,
      "learning_rate": 0.0,
      "loss": 0.8281,
      "step": 1881
    },
    {
      "epoch": 0.16891042900735953,
      "grad_norm": 2.6673712730407715,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 1882
    },
    {
      "epoch": 0.16900017950098725,
      "grad_norm": 2.2988197803497314,
      "learning_rate": 0.0,
      "loss": 0.7912,
      "step": 1883
    },
    {
      "epoch": 0.16908992999461497,
      "grad_norm": 3.6867659091949463,
      "learning_rate": 0.0,
      "loss": 0.7982,
      "step": 1884
    },
    {
      "epoch": 0.1691796804882427,
      "grad_norm": 2.4449846744537354,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 1885
    },
    {
      "epoch": 0.1692694309818704,
      "grad_norm": 3.024057626724243,
      "learning_rate": 0.0,
      "loss": 0.7895,
      "step": 1886
    },
    {
      "epoch": 0.1693591814754981,
      "grad_norm": 3.4656505584716797,
      "learning_rate": 0.0,
      "loss": 0.7776,
      "step": 1887
    },
    {
      "epoch": 0.16944893196912583,
      "grad_norm": 3.206376314163208,
      "learning_rate": 0.0,
      "loss": 0.8331,
      "step": 1888
    },
    {
      "epoch": 0.16953868246275355,
      "grad_norm": 3.8423211574554443,
      "learning_rate": 0.0,
      "loss": 0.817,
      "step": 1889
    },
    {
      "epoch": 0.16962843295638125,
      "grad_norm": 2.647106409072876,
      "learning_rate": 0.0,
      "loss": 0.8715,
      "step": 1890
    },
    {
      "epoch": 0.16971818345000897,
      "grad_norm": 4.022595405578613,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 1891
    },
    {
      "epoch": 0.1698079339436367,
      "grad_norm": 4.315979480743408,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 1892
    },
    {
      "epoch": 0.1698976844372644,
      "grad_norm": 4.548421859741211,
      "learning_rate": 0.0,
      "loss": 0.7869,
      "step": 1893
    },
    {
      "epoch": 0.1699874349308921,
      "grad_norm": 5.1182637214660645,
      "learning_rate": 0.0,
      "loss": 0.8081,
      "step": 1894
    },
    {
      "epoch": 0.17007718542451983,
      "grad_norm": 4.1747145652771,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 1895
    },
    {
      "epoch": 0.17016693591814755,
      "grad_norm": 2.702096939086914,
      "learning_rate": 0.0,
      "loss": 0.8328,
      "step": 1896
    },
    {
      "epoch": 0.17025668641177527,
      "grad_norm": 9.68958854675293,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 1897
    },
    {
      "epoch": 0.170346436905403,
      "grad_norm": 2.388967752456665,
      "learning_rate": 0.0,
      "loss": 0.8085,
      "step": 1898
    },
    {
      "epoch": 0.1704361873990307,
      "grad_norm": 8.635712623596191,
      "learning_rate": 0.0,
      "loss": 0.7905,
      "step": 1899
    },
    {
      "epoch": 0.1705259378926584,
      "grad_norm": 6.557069301605225,
      "learning_rate": 0.0,
      "loss": 0.7887,
      "step": 1900
    },
    {
      "epoch": 0.17061568838628613,
      "grad_norm": 2.600689172744751,
      "learning_rate": 0.0,
      "loss": 0.8158,
      "step": 1901
    },
    {
      "epoch": 0.17070543887991385,
      "grad_norm": 3.036071300506592,
      "learning_rate": 0.0,
      "loss": 0.79,
      "step": 1902
    },
    {
      "epoch": 0.17079518937354154,
      "grad_norm": 2.5613670349121094,
      "learning_rate": 0.0,
      "loss": 0.7609,
      "step": 1903
    },
    {
      "epoch": 0.17088493986716927,
      "grad_norm": 2.1275761127471924,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 1904
    },
    {
      "epoch": 0.170974690360797,
      "grad_norm": 2.8839499950408936,
      "learning_rate": 0.0,
      "loss": 0.8155,
      "step": 1905
    },
    {
      "epoch": 0.1710644408544247,
      "grad_norm": 2.8903489112854004,
      "learning_rate": 0.0,
      "loss": 0.7776,
      "step": 1906
    },
    {
      "epoch": 0.1711541913480524,
      "grad_norm": 2.599720001220703,
      "learning_rate": 0.0,
      "loss": 0.7892,
      "step": 1907
    },
    {
      "epoch": 0.17124394184168013,
      "grad_norm": 4.382303714752197,
      "learning_rate": 0.0,
      "loss": 0.7686,
      "step": 1908
    },
    {
      "epoch": 0.17133369233530785,
      "grad_norm": 2.963690757751465,
      "learning_rate": 0.0,
      "loss": 0.7843,
      "step": 1909
    },
    {
      "epoch": 0.17142344282893557,
      "grad_norm": 2.9333415031433105,
      "learning_rate": 0.0,
      "loss": 0.8034,
      "step": 1910
    },
    {
      "epoch": 0.17151319332256326,
      "grad_norm": 3.9559578895568848,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 1911
    },
    {
      "epoch": 0.17160294381619098,
      "grad_norm": 7.814508438110352,
      "learning_rate": 0.0,
      "loss": 0.8104,
      "step": 1912
    },
    {
      "epoch": 0.1716926943098187,
      "grad_norm": 2.927375078201294,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 1913
    },
    {
      "epoch": 0.17178244480344643,
      "grad_norm": 4.677928447723389,
      "learning_rate": 0.0,
      "loss": 0.7985,
      "step": 1914
    },
    {
      "epoch": 0.17187219529707412,
      "grad_norm": 2.627298355102539,
      "learning_rate": 0.0,
      "loss": 0.7774,
      "step": 1915
    },
    {
      "epoch": 0.17196194579070184,
      "grad_norm": 2.5387821197509766,
      "learning_rate": 0.0,
      "loss": 0.8029,
      "step": 1916
    },
    {
      "epoch": 0.17205169628432956,
      "grad_norm": 3.3446717262268066,
      "learning_rate": 0.0,
      "loss": 0.7667,
      "step": 1917
    },
    {
      "epoch": 0.17214144677795729,
      "grad_norm": 4.1028218269348145,
      "learning_rate": 0.0,
      "loss": 0.8026,
      "step": 1918
    },
    {
      "epoch": 0.172231197271585,
      "grad_norm": 3.2087349891662598,
      "learning_rate": 0.0,
      "loss": 0.7854,
      "step": 1919
    },
    {
      "epoch": 0.1723209477652127,
      "grad_norm": 2.6978516578674316,
      "learning_rate": 0.0,
      "loss": 0.8004,
      "step": 1920
    },
    {
      "epoch": 0.17241069825884042,
      "grad_norm": 3.515211582183838,
      "learning_rate": 0.0,
      "loss": 0.8014,
      "step": 1921
    },
    {
      "epoch": 0.17250044875246814,
      "grad_norm": 2.993617057800293,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 1922
    },
    {
      "epoch": 0.17259019924609587,
      "grad_norm": 3.0095982551574707,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 1923
    },
    {
      "epoch": 0.17267994973972356,
      "grad_norm": 6.588475227355957,
      "learning_rate": 0.0,
      "loss": 0.8134,
      "step": 1924
    },
    {
      "epoch": 0.17276970023335128,
      "grad_norm": 2.8648345470428467,
      "learning_rate": 0.0,
      "loss": 0.7843,
      "step": 1925
    },
    {
      "epoch": 0.172859450726979,
      "grad_norm": 3.29571533203125,
      "learning_rate": 0.0,
      "loss": 0.7836,
      "step": 1926
    },
    {
      "epoch": 0.17294920122060672,
      "grad_norm": 3.2108194828033447,
      "learning_rate": 0.0,
      "loss": 0.7923,
      "step": 1927
    },
    {
      "epoch": 0.17303895171423442,
      "grad_norm": 2.873892068862915,
      "learning_rate": 0.0,
      "loss": 0.7955,
      "step": 1928
    },
    {
      "epoch": 0.17312870220786214,
      "grad_norm": 4.591238021850586,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 1929
    },
    {
      "epoch": 0.17321845270148986,
      "grad_norm": 15.69715404510498,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 1930
    },
    {
      "epoch": 0.17330820319511758,
      "grad_norm": 6.049088001251221,
      "learning_rate": 0.0,
      "loss": 0.7752,
      "step": 1931
    },
    {
      "epoch": 0.17339795368874528,
      "grad_norm": 3.2544150352478027,
      "learning_rate": 0.0,
      "loss": 0.839,
      "step": 1932
    },
    {
      "epoch": 0.173487704182373,
      "grad_norm": 2.877520799636841,
      "learning_rate": 0.0,
      "loss": 0.7826,
      "step": 1933
    },
    {
      "epoch": 0.17357745467600072,
      "grad_norm": 3.2471611499786377,
      "learning_rate": 0.0,
      "loss": 0.7043,
      "step": 1934
    },
    {
      "epoch": 0.17366720516962844,
      "grad_norm": 8.729570388793945,
      "learning_rate": 0.0,
      "loss": 0.8309,
      "step": 1935
    },
    {
      "epoch": 0.17375695566325614,
      "grad_norm": 3.757359504699707,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 1936
    },
    {
      "epoch": 0.17384670615688386,
      "grad_norm": 3.7297275066375732,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 1937
    },
    {
      "epoch": 0.17393645665051158,
      "grad_norm": 2.7728047370910645,
      "learning_rate": 0.0,
      "loss": 0.7733,
      "step": 1938
    },
    {
      "epoch": 0.1740262071441393,
      "grad_norm": 3.3328428268432617,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 1939
    },
    {
      "epoch": 0.17411595763776702,
      "grad_norm": 5.837542533874512,
      "learning_rate": 0.0,
      "loss": 0.8401,
      "step": 1940
    },
    {
      "epoch": 0.17420570813139472,
      "grad_norm": 3.8262269496917725,
      "learning_rate": 0.0,
      "loss": 0.8038,
      "step": 1941
    },
    {
      "epoch": 0.17429545862502244,
      "grad_norm": 3.7265257835388184,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 1942
    },
    {
      "epoch": 0.17438520911865016,
      "grad_norm": 3.9500908851623535,
      "learning_rate": 0.0,
      "loss": 0.7516,
      "step": 1943
    },
    {
      "epoch": 0.17447495961227788,
      "grad_norm": 3.3231873512268066,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 1944
    },
    {
      "epoch": 0.17456471010590557,
      "grad_norm": 5.612390518188477,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 1945
    },
    {
      "epoch": 0.1746544605995333,
      "grad_norm": 5.924403190612793,
      "learning_rate": 0.0,
      "loss": 0.8136,
      "step": 1946
    },
    {
      "epoch": 0.17474421109316102,
      "grad_norm": 2.8484644889831543,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 1947
    },
    {
      "epoch": 0.17483396158678874,
      "grad_norm": 3.058985948562622,
      "learning_rate": 0.0,
      "loss": 0.6972,
      "step": 1948
    },
    {
      "epoch": 0.17492371208041643,
      "grad_norm": 3.799738883972168,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 1949
    },
    {
      "epoch": 0.17501346257404415,
      "grad_norm": 38.6955451965332,
      "learning_rate": 0.0,
      "loss": 0.7941,
      "step": 1950
    },
    {
      "epoch": 0.17510321306767188,
      "grad_norm": 4.979582786560059,
      "learning_rate": 0.0,
      "loss": 0.8239,
      "step": 1951
    },
    {
      "epoch": 0.1751929635612996,
      "grad_norm": 4.422044277191162,
      "learning_rate": 0.0,
      "loss": 0.7565,
      "step": 1952
    },
    {
      "epoch": 0.1752827140549273,
      "grad_norm": 2.8183817863464355,
      "learning_rate": 0.0,
      "loss": 0.8243,
      "step": 1953
    },
    {
      "epoch": 0.175372464548555,
      "grad_norm": 2.816498279571533,
      "learning_rate": 0.0,
      "loss": 0.7905,
      "step": 1954
    },
    {
      "epoch": 0.17546221504218273,
      "grad_norm": 3.93930721282959,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 1955
    },
    {
      "epoch": 0.17555196553581046,
      "grad_norm": 3.5149929523468018,
      "learning_rate": 0.0,
      "loss": 0.8444,
      "step": 1956
    },
    {
      "epoch": 0.17564171602943815,
      "grad_norm": 2.9916837215423584,
      "learning_rate": 0.0,
      "loss": 0.7979,
      "step": 1957
    },
    {
      "epoch": 0.17573146652306587,
      "grad_norm": 2.594783067703247,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 1958
    },
    {
      "epoch": 0.1758212170166936,
      "grad_norm": 3.6245598793029785,
      "learning_rate": 0.0,
      "loss": 0.8053,
      "step": 1959
    },
    {
      "epoch": 0.17591096751032131,
      "grad_norm": 3.736419916152954,
      "learning_rate": 0.0,
      "loss": 0.7791,
      "step": 1960
    },
    {
      "epoch": 0.176000718003949,
      "grad_norm": 3.143141269683838,
      "learning_rate": 0.0,
      "loss": 0.7667,
      "step": 1961
    },
    {
      "epoch": 0.17609046849757673,
      "grad_norm": 3.909679651260376,
      "learning_rate": 0.0,
      "loss": 0.776,
      "step": 1962
    },
    {
      "epoch": 0.17618021899120445,
      "grad_norm": 3.108642339706421,
      "learning_rate": 0.0,
      "loss": 0.7609,
      "step": 1963
    },
    {
      "epoch": 0.17626996948483217,
      "grad_norm": 2.5354418754577637,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 1964
    },
    {
      "epoch": 0.1763597199784599,
      "grad_norm": 2.8805665969848633,
      "learning_rate": 0.0,
      "loss": 0.8245,
      "step": 1965
    },
    {
      "epoch": 0.1764494704720876,
      "grad_norm": 3.1180062294006348,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 1966
    },
    {
      "epoch": 0.1765392209657153,
      "grad_norm": 2.828800916671753,
      "learning_rate": 0.0,
      "loss": 0.7951,
      "step": 1967
    },
    {
      "epoch": 0.17662897145934303,
      "grad_norm": 3.2005491256713867,
      "learning_rate": 0.0,
      "loss": 0.8064,
      "step": 1968
    },
    {
      "epoch": 0.17671872195297075,
      "grad_norm": 2.9093713760375977,
      "learning_rate": 0.0,
      "loss": 0.795,
      "step": 1969
    },
    {
      "epoch": 0.17680847244659845,
      "grad_norm": 2.5594069957733154,
      "learning_rate": 0.0,
      "loss": 0.8592,
      "step": 1970
    },
    {
      "epoch": 0.17689822294022617,
      "grad_norm": 4.7033281326293945,
      "learning_rate": 0.0,
      "loss": 0.8164,
      "step": 1971
    },
    {
      "epoch": 0.1769879734338539,
      "grad_norm": 4.695409297943115,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 1972
    },
    {
      "epoch": 0.1770777239274816,
      "grad_norm": 3.0215981006622314,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 1973
    },
    {
      "epoch": 0.1771674744211093,
      "grad_norm": 3.950201988220215,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 1974
    },
    {
      "epoch": 0.17725722491473703,
      "grad_norm": 2.687495231628418,
      "learning_rate": 0.0,
      "loss": 0.8264,
      "step": 1975
    },
    {
      "epoch": 0.17734697540836475,
      "grad_norm": 2.866382122039795,
      "learning_rate": 0.0,
      "loss": 0.8476,
      "step": 1976
    },
    {
      "epoch": 0.17743672590199247,
      "grad_norm": 4.649231433868408,
      "learning_rate": 0.0,
      "loss": 0.7745,
      "step": 1977
    },
    {
      "epoch": 0.17752647639562016,
      "grad_norm": 5.172459602355957,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 1978
    },
    {
      "epoch": 0.17761622688924789,
      "grad_norm": 2.912344217300415,
      "learning_rate": 0.0,
      "loss": 0.7983,
      "step": 1979
    },
    {
      "epoch": 0.1777059773828756,
      "grad_norm": 5.884006977081299,
      "learning_rate": 0.0,
      "loss": 0.8322,
      "step": 1980
    },
    {
      "epoch": 0.17779572787650333,
      "grad_norm": 2.9417777061462402,
      "learning_rate": 0.0,
      "loss": 0.8003,
      "step": 1981
    },
    {
      "epoch": 0.17788547837013102,
      "grad_norm": 3.3736300468444824,
      "learning_rate": 0.0,
      "loss": 0.7986,
      "step": 1982
    },
    {
      "epoch": 0.17797522886375874,
      "grad_norm": 2.7400383949279785,
      "learning_rate": 0.0,
      "loss": 0.8178,
      "step": 1983
    },
    {
      "epoch": 0.17806497935738647,
      "grad_norm": 7.261805057525635,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 1984
    },
    {
      "epoch": 0.1781547298510142,
      "grad_norm": 3.6134421825408936,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 1985
    },
    {
      "epoch": 0.1782444803446419,
      "grad_norm": 2.7496416568756104,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 1986
    },
    {
      "epoch": 0.1783342308382696,
      "grad_norm": 2.912829637527466,
      "learning_rate": 0.0,
      "loss": 0.7697,
      "step": 1987
    },
    {
      "epoch": 0.17842398133189732,
      "grad_norm": 3.006103277206421,
      "learning_rate": 0.0,
      "loss": 0.7523,
      "step": 1988
    },
    {
      "epoch": 0.17851373182552505,
      "grad_norm": 2.523318290710449,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 1989
    },
    {
      "epoch": 0.17860348231915277,
      "grad_norm": 4.604520797729492,
      "learning_rate": 0.0,
      "loss": 0.858,
      "step": 1990
    },
    {
      "epoch": 0.17869323281278046,
      "grad_norm": 3.2361066341400146,
      "learning_rate": 0.0,
      "loss": 0.813,
      "step": 1991
    },
    {
      "epoch": 0.17878298330640818,
      "grad_norm": 5.974128723144531,
      "learning_rate": 0.0,
      "loss": 0.7542,
      "step": 1992
    },
    {
      "epoch": 0.1788727338000359,
      "grad_norm": 3.1734158992767334,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 1993
    },
    {
      "epoch": 0.17896248429366363,
      "grad_norm": 5.261252403259277,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 1994
    },
    {
      "epoch": 0.17905223478729132,
      "grad_norm": 5.264753818511963,
      "learning_rate": 0.0,
      "loss": 0.7527,
      "step": 1995
    },
    {
      "epoch": 0.17914198528091904,
      "grad_norm": 3.456843852996826,
      "learning_rate": 0.0,
      "loss": 0.7825,
      "step": 1996
    },
    {
      "epoch": 0.17923173577454676,
      "grad_norm": 4.708425521850586,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 1997
    },
    {
      "epoch": 0.17932148626817448,
      "grad_norm": 3.140080690383911,
      "learning_rate": 0.0,
      "loss": 0.7833,
      "step": 1998
    },
    {
      "epoch": 0.17941123676180218,
      "grad_norm": 5.146877765655518,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 1999
    },
    {
      "epoch": 0.1795009872554299,
      "grad_norm": 2.636868953704834,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 2000
    },
    {
      "epoch": 0.17959073774905762,
      "grad_norm": 6.203399181365967,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 2001
    },
    {
      "epoch": 0.17968048824268534,
      "grad_norm": 4.279970169067383,
      "learning_rate": 0.0,
      "loss": 0.7889,
      "step": 2002
    },
    {
      "epoch": 0.17977023873631304,
      "grad_norm": 3.8593337535858154,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 2003
    },
    {
      "epoch": 0.17985998922994076,
      "grad_norm": 2.124532461166382,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 2004
    },
    {
      "epoch": 0.17994973972356848,
      "grad_norm": 2.6296613216400146,
      "learning_rate": 0.0,
      "loss": 0.8273,
      "step": 2005
    },
    {
      "epoch": 0.1800394902171962,
      "grad_norm": 2.7547900676727295,
      "learning_rate": 0.0,
      "loss": 0.7894,
      "step": 2006
    },
    {
      "epoch": 0.1801292407108239,
      "grad_norm": 16.25506591796875,
      "learning_rate": 0.0,
      "loss": 0.7841,
      "step": 2007
    },
    {
      "epoch": 0.18021899120445162,
      "grad_norm": 3.521644353866577,
      "learning_rate": 0.0,
      "loss": 0.806,
      "step": 2008
    },
    {
      "epoch": 0.18030874169807934,
      "grad_norm": 2.7314298152923584,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 2009
    },
    {
      "epoch": 0.18039849219170706,
      "grad_norm": 1.7458616495132446,
      "learning_rate": 0.0,
      "loss": 0.8369,
      "step": 2010
    },
    {
      "epoch": 0.18048824268533478,
      "grad_norm": 3.7835874557495117,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 2011
    },
    {
      "epoch": 0.18057799317896248,
      "grad_norm": 2.9204118251800537,
      "learning_rate": 0.0,
      "loss": 0.8107,
      "step": 2012
    },
    {
      "epoch": 0.1806677436725902,
      "grad_norm": 2.821840763092041,
      "learning_rate": 0.0,
      "loss": 0.8006,
      "step": 2013
    },
    {
      "epoch": 0.18075749416621792,
      "grad_norm": 2.615201950073242,
      "learning_rate": 0.0,
      "loss": 0.8117,
      "step": 2014
    },
    {
      "epoch": 0.18084724465984564,
      "grad_norm": 3.659149169921875,
      "learning_rate": 0.0,
      "loss": 0.814,
      "step": 2015
    },
    {
      "epoch": 0.18093699515347333,
      "grad_norm": 2.7303452491760254,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 2016
    },
    {
      "epoch": 0.18102674564710106,
      "grad_norm": 3.270292282104492,
      "learning_rate": 0.0,
      "loss": 0.7872,
      "step": 2017
    },
    {
      "epoch": 0.18111649614072878,
      "grad_norm": 2.855132579803467,
      "learning_rate": 0.0,
      "loss": 0.771,
      "step": 2018
    },
    {
      "epoch": 0.1812062466343565,
      "grad_norm": 3.018416404724121,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 2019
    },
    {
      "epoch": 0.1812959971279842,
      "grad_norm": 5.503456115722656,
      "learning_rate": 0.0,
      "loss": 0.8171,
      "step": 2020
    },
    {
      "epoch": 0.18138574762161191,
      "grad_norm": 9.075661659240723,
      "learning_rate": 0.0,
      "loss": 0.8321,
      "step": 2021
    },
    {
      "epoch": 0.18147549811523964,
      "grad_norm": 7.409101963043213,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 2022
    },
    {
      "epoch": 0.18156524860886736,
      "grad_norm": 6.413672924041748,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 2023
    },
    {
      "epoch": 0.18165499910249505,
      "grad_norm": 2.356698989868164,
      "learning_rate": 0.0,
      "loss": 0.7772,
      "step": 2024
    },
    {
      "epoch": 0.18174474959612277,
      "grad_norm": 2.0250661373138428,
      "learning_rate": 0.0,
      "loss": 0.8129,
      "step": 2025
    },
    {
      "epoch": 0.1818345000897505,
      "grad_norm": 4.761077404022217,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 2026
    },
    {
      "epoch": 0.18192425058337822,
      "grad_norm": 3.2036030292510986,
      "learning_rate": 0.0,
      "loss": 0.8265,
      "step": 2027
    },
    {
      "epoch": 0.1820140010770059,
      "grad_norm": 2.8882222175598145,
      "learning_rate": 0.0,
      "loss": 0.8099,
      "step": 2028
    },
    {
      "epoch": 0.18210375157063363,
      "grad_norm": 2.8108463287353516,
      "learning_rate": 0.0,
      "loss": 0.8016,
      "step": 2029
    },
    {
      "epoch": 0.18219350206426135,
      "grad_norm": 9.561469078063965,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 2030
    },
    {
      "epoch": 0.18228325255788907,
      "grad_norm": 3.276238203048706,
      "learning_rate": 0.0,
      "loss": 0.7894,
      "step": 2031
    },
    {
      "epoch": 0.1823730030515168,
      "grad_norm": 2.926227569580078,
      "learning_rate": 0.0,
      "loss": 0.8103,
      "step": 2032
    },
    {
      "epoch": 0.1824627535451445,
      "grad_norm": 10.29720401763916,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 2033
    },
    {
      "epoch": 0.1825525040387722,
      "grad_norm": 3.335228443145752,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 2034
    },
    {
      "epoch": 0.18264225453239993,
      "grad_norm": 4.8559088706970215,
      "learning_rate": 0.0,
      "loss": 0.7861,
      "step": 2035
    },
    {
      "epoch": 0.18273200502602766,
      "grad_norm": 3.1208760738372803,
      "learning_rate": 0.0,
      "loss": 0.8349,
      "step": 2036
    },
    {
      "epoch": 0.18282175551965535,
      "grad_norm": 5.263544082641602,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 2037
    },
    {
      "epoch": 0.18291150601328307,
      "grad_norm": 5.74838399887085,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 2038
    },
    {
      "epoch": 0.1830012565069108,
      "grad_norm": 5.427893161773682,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 2039
    },
    {
      "epoch": 0.1830910070005385,
      "grad_norm": 2.7397072315216064,
      "learning_rate": 0.0,
      "loss": 0.7977,
      "step": 2040
    },
    {
      "epoch": 0.1831807574941662,
      "grad_norm": 3.205724000930786,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 2041
    },
    {
      "epoch": 0.18327050798779393,
      "grad_norm": 2.9206223487854004,
      "learning_rate": 0.0,
      "loss": 0.7076,
      "step": 2042
    },
    {
      "epoch": 0.18336025848142165,
      "grad_norm": 4.843723297119141,
      "learning_rate": 0.0,
      "loss": 0.8266,
      "step": 2043
    },
    {
      "epoch": 0.18345000897504937,
      "grad_norm": 2.189509153366089,
      "learning_rate": 0.0,
      "loss": 0.7737,
      "step": 2044
    },
    {
      "epoch": 0.18353975946867707,
      "grad_norm": 2.4964957237243652,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 2045
    },
    {
      "epoch": 0.1836295099623048,
      "grad_norm": 2.479132652282715,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 2046
    },
    {
      "epoch": 0.1837192604559325,
      "grad_norm": 3.8490734100341797,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 2047
    },
    {
      "epoch": 0.18380901094956023,
      "grad_norm": 5.177379608154297,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 2048
    },
    {
      "epoch": 0.18389876144318792,
      "grad_norm": 4.003405570983887,
      "learning_rate": 0.0,
      "loss": 0.854,
      "step": 2049
    },
    {
      "epoch": 0.18398851193681565,
      "grad_norm": 2.158135175704956,
      "learning_rate": 0.0,
      "loss": 0.794,
      "step": 2050
    },
    {
      "epoch": 0.18407826243044337,
      "grad_norm": 2.766845226287842,
      "learning_rate": 0.0,
      "loss": 0.7677,
      "step": 2051
    },
    {
      "epoch": 0.1841680129240711,
      "grad_norm": 4.071296691894531,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "step": 2052
    },
    {
      "epoch": 0.1842577634176988,
      "grad_norm": 2.6029770374298096,
      "learning_rate": 0.0,
      "loss": 0.8032,
      "step": 2053
    },
    {
      "epoch": 0.1843475139113265,
      "grad_norm": 3.3414905071258545,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 2054
    },
    {
      "epoch": 0.18443726440495423,
      "grad_norm": 3.8755810260772705,
      "learning_rate": 0.0,
      "loss": 0.7683,
      "step": 2055
    },
    {
      "epoch": 0.18452701489858195,
      "grad_norm": 3.2605934143066406,
      "learning_rate": 0.0,
      "loss": 0.8022,
      "step": 2056
    },
    {
      "epoch": 0.18461676539220967,
      "grad_norm": 4.542389392852783,
      "learning_rate": 0.0,
      "loss": 0.7663,
      "step": 2057
    },
    {
      "epoch": 0.18470651588583736,
      "grad_norm": 3.6797778606414795,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 2058
    },
    {
      "epoch": 0.18479626637946509,
      "grad_norm": 3.5152385234832764,
      "learning_rate": 0.0,
      "loss": 0.778,
      "step": 2059
    },
    {
      "epoch": 0.1848860168730928,
      "grad_norm": 2.378183364868164,
      "learning_rate": 0.0,
      "loss": 0.8405,
      "step": 2060
    },
    {
      "epoch": 0.18497576736672053,
      "grad_norm": 2.448702812194824,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 2061
    },
    {
      "epoch": 0.18506551786034822,
      "grad_norm": 3.8072900772094727,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 2062
    },
    {
      "epoch": 0.18515526835397594,
      "grad_norm": 4.170334815979004,
      "learning_rate": 0.0,
      "loss": 0.8277,
      "step": 2063
    },
    {
      "epoch": 0.18524501884760367,
      "grad_norm": 9.252466201782227,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 2064
    },
    {
      "epoch": 0.1853347693412314,
      "grad_norm": 3.0477075576782227,
      "learning_rate": 0.0,
      "loss": 0.7811,
      "step": 2065
    },
    {
      "epoch": 0.18542451983485908,
      "grad_norm": 2.5305068492889404,
      "learning_rate": 0.0,
      "loss": 0.8193,
      "step": 2066
    },
    {
      "epoch": 0.1855142703284868,
      "grad_norm": 3.8154232501983643,
      "learning_rate": 0.0,
      "loss": 0.7699,
      "step": 2067
    },
    {
      "epoch": 0.18560402082211452,
      "grad_norm": 3.781637668609619,
      "learning_rate": 0.0,
      "loss": 0.792,
      "step": 2068
    },
    {
      "epoch": 0.18569377131574225,
      "grad_norm": 2.2859959602355957,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 2069
    },
    {
      "epoch": 0.18578352180936994,
      "grad_norm": 2.8799219131469727,
      "learning_rate": 0.0,
      "loss": 0.7784,
      "step": 2070
    },
    {
      "epoch": 0.18587327230299766,
      "grad_norm": 3.1649558544158936,
      "learning_rate": 0.0,
      "loss": 0.7346,
      "step": 2071
    },
    {
      "epoch": 0.18596302279662538,
      "grad_norm": 3.149545192718506,
      "learning_rate": 0.0,
      "loss": 0.7912,
      "step": 2072
    },
    {
      "epoch": 0.1860527732902531,
      "grad_norm": 2.9764933586120605,
      "learning_rate": 0.0,
      "loss": 0.7966,
      "step": 2073
    },
    {
      "epoch": 0.1861425237838808,
      "grad_norm": 2.855283498764038,
      "learning_rate": 0.0,
      "loss": 0.8,
      "step": 2074
    },
    {
      "epoch": 0.18623227427750852,
      "grad_norm": 2.2555229663848877,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 2075
    },
    {
      "epoch": 0.18632202477113624,
      "grad_norm": 6.128325462341309,
      "learning_rate": 0.0,
      "loss": 0.7883,
      "step": 2076
    },
    {
      "epoch": 0.18641177526476396,
      "grad_norm": 2.4759371280670166,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 2077
    },
    {
      "epoch": 0.18650152575839168,
      "grad_norm": 3.6937034130096436,
      "learning_rate": 0.0,
      "loss": 0.8096,
      "step": 2078
    },
    {
      "epoch": 0.18659127625201938,
      "grad_norm": 4.982515811920166,
      "learning_rate": 0.0,
      "loss": 0.7411,
      "step": 2079
    },
    {
      "epoch": 0.1866810267456471,
      "grad_norm": 3.1872127056121826,
      "learning_rate": 0.0,
      "loss": 0.7952,
      "step": 2080
    },
    {
      "epoch": 0.18677077723927482,
      "grad_norm": 3.1807777881622314,
      "learning_rate": 0.0,
      "loss": 0.7736,
      "step": 2081
    },
    {
      "epoch": 0.18686052773290254,
      "grad_norm": 2.2402889728546143,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 2082
    },
    {
      "epoch": 0.18695027822653024,
      "grad_norm": 3.503127098083496,
      "learning_rate": 0.0,
      "loss": 0.8144,
      "step": 2083
    },
    {
      "epoch": 0.18704002872015796,
      "grad_norm": 55.674407958984375,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 2084
    },
    {
      "epoch": 0.18712977921378568,
      "grad_norm": 2.4345407485961914,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 2085
    },
    {
      "epoch": 0.1872195297074134,
      "grad_norm": 2.693645477294922,
      "learning_rate": 0.0,
      "loss": 0.7846,
      "step": 2086
    },
    {
      "epoch": 0.1873092802010411,
      "grad_norm": 2.6529335975646973,
      "learning_rate": 0.0,
      "loss": 0.7885,
      "step": 2087
    },
    {
      "epoch": 0.18739903069466882,
      "grad_norm": 2.66207218170166,
      "learning_rate": 0.0,
      "loss": 0.8014,
      "step": 2088
    },
    {
      "epoch": 0.18748878118829654,
      "grad_norm": 3.668745517730713,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 2089
    },
    {
      "epoch": 0.18757853168192426,
      "grad_norm": 2.2885589599609375,
      "learning_rate": 0.0,
      "loss": 0.7737,
      "step": 2090
    },
    {
      "epoch": 0.18766828217555195,
      "grad_norm": 13.482264518737793,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 2091
    },
    {
      "epoch": 0.18775803266917968,
      "grad_norm": 2.6618785858154297,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 2092
    },
    {
      "epoch": 0.1878477831628074,
      "grad_norm": 7.541994094848633,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 2093
    },
    {
      "epoch": 0.18793753365643512,
      "grad_norm": 2.3926992416381836,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 2094
    },
    {
      "epoch": 0.1880272841500628,
      "grad_norm": 1.8733867406845093,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 2095
    },
    {
      "epoch": 0.18811703464369053,
      "grad_norm": 2.2566258907318115,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 2096
    },
    {
      "epoch": 0.18820678513731826,
      "grad_norm": 2.713268280029297,
      "learning_rate": 0.0,
      "loss": 0.7977,
      "step": 2097
    },
    {
      "epoch": 0.18829653563094598,
      "grad_norm": 2.318246841430664,
      "learning_rate": 0.0,
      "loss": 0.81,
      "step": 2098
    },
    {
      "epoch": 0.1883862861245737,
      "grad_norm": 2.038891077041626,
      "learning_rate": 0.0,
      "loss": 0.8069,
      "step": 2099
    },
    {
      "epoch": 0.1884760366182014,
      "grad_norm": 2.5899150371551514,
      "learning_rate": 0.0,
      "loss": 0.8048,
      "step": 2100
    },
    {
      "epoch": 0.18856578711182911,
      "grad_norm": 2.4862794876098633,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 2101
    },
    {
      "epoch": 0.18865553760545684,
      "grad_norm": 3.712924003601074,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 2102
    },
    {
      "epoch": 0.18874528809908456,
      "grad_norm": 3.078996181488037,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 2103
    },
    {
      "epoch": 0.18883503859271225,
      "grad_norm": 3.4977128505706787,
      "learning_rate": 0.0,
      "loss": 0.7854,
      "step": 2104
    },
    {
      "epoch": 0.18892478908633997,
      "grad_norm": 2.975634813308716,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 2105
    },
    {
      "epoch": 0.1890145395799677,
      "grad_norm": 2.93076491355896,
      "learning_rate": 0.0,
      "loss": 0.8141,
      "step": 2106
    },
    {
      "epoch": 0.18910429007359542,
      "grad_norm": 2.7336621284484863,
      "learning_rate": 0.0,
      "loss": 0.8121,
      "step": 2107
    },
    {
      "epoch": 0.1891940405672231,
      "grad_norm": 5.529118537902832,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 2108
    },
    {
      "epoch": 0.18928379106085083,
      "grad_norm": 2.7730939388275146,
      "learning_rate": 0.0,
      "loss": 0.8184,
      "step": 2109
    },
    {
      "epoch": 0.18937354155447855,
      "grad_norm": 4.360902309417725,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 2110
    },
    {
      "epoch": 0.18946329204810627,
      "grad_norm": 3.659036874771118,
      "learning_rate": 0.0,
      "loss": 0.8,
      "step": 2111
    },
    {
      "epoch": 0.18955304254173397,
      "grad_norm": 30.578144073486328,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 2112
    },
    {
      "epoch": 0.1896427930353617,
      "grad_norm": 4.803082466125488,
      "learning_rate": 0.0,
      "loss": 0.7857,
      "step": 2113
    },
    {
      "epoch": 0.1897325435289894,
      "grad_norm": 2.7154734134674072,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 2114
    },
    {
      "epoch": 0.18982229402261713,
      "grad_norm": 2.474644899368286,
      "learning_rate": 0.0,
      "loss": 0.8443,
      "step": 2115
    },
    {
      "epoch": 0.18991204451624483,
      "grad_norm": 2.890716791152954,
      "learning_rate": 0.0,
      "loss": 0.7715,
      "step": 2116
    },
    {
      "epoch": 0.19000179500987255,
      "grad_norm": 5.083712100982666,
      "learning_rate": 0.0,
      "loss": 0.8641,
      "step": 2117
    },
    {
      "epoch": 0.19009154550350027,
      "grad_norm": 3.621692180633545,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 2118
    },
    {
      "epoch": 0.190181295997128,
      "grad_norm": 2.6743223667144775,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 2119
    },
    {
      "epoch": 0.19027104649075569,
      "grad_norm": 2.55065655708313,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 2120
    },
    {
      "epoch": 0.1903607969843834,
      "grad_norm": 2.643667459487915,
      "learning_rate": 0.0,
      "loss": 0.8545,
      "step": 2121
    },
    {
      "epoch": 0.19045054747801113,
      "grad_norm": 2.293030261993408,
      "learning_rate": 0.0,
      "loss": 0.7572,
      "step": 2122
    },
    {
      "epoch": 0.19054029797163885,
      "grad_norm": 2.9595115184783936,
      "learning_rate": 0.0,
      "loss": 0.7817,
      "step": 2123
    },
    {
      "epoch": 0.19063004846526657,
      "grad_norm": 5.84333610534668,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 2124
    },
    {
      "epoch": 0.19071979895889427,
      "grad_norm": 3.106106758117676,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 2125
    },
    {
      "epoch": 0.190809549452522,
      "grad_norm": 2.6483817100524902,
      "learning_rate": 0.0,
      "loss": 0.7782,
      "step": 2126
    },
    {
      "epoch": 0.1908992999461497,
      "grad_norm": 2.673717737197876,
      "learning_rate": 0.0,
      "loss": 0.7866,
      "step": 2127
    },
    {
      "epoch": 0.19098905043977743,
      "grad_norm": 2.6296775341033936,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 2128
    },
    {
      "epoch": 0.19107880093340512,
      "grad_norm": 3.43328857421875,
      "learning_rate": 0.0,
      "loss": 0.7887,
      "step": 2129
    },
    {
      "epoch": 0.19116855142703285,
      "grad_norm": 2.797461986541748,
      "learning_rate": 0.0,
      "loss": 0.8244,
      "step": 2130
    },
    {
      "epoch": 0.19125830192066057,
      "grad_norm": 3.1063971519470215,
      "learning_rate": 0.0,
      "loss": 0.7964,
      "step": 2131
    },
    {
      "epoch": 0.1913480524142883,
      "grad_norm": 2.521998405456543,
      "learning_rate": 0.0,
      "loss": 0.8072,
      "step": 2132
    },
    {
      "epoch": 0.19143780290791598,
      "grad_norm": 6.628070831298828,
      "learning_rate": 0.0,
      "loss": 0.82,
      "step": 2133
    },
    {
      "epoch": 0.1915275534015437,
      "grad_norm": 32.32122802734375,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 2134
    },
    {
      "epoch": 0.19161730389517143,
      "grad_norm": 2.8353843688964844,
      "learning_rate": 0.0,
      "loss": 0.7897,
      "step": 2135
    },
    {
      "epoch": 0.19170705438879915,
      "grad_norm": 3.748140811920166,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 2136
    },
    {
      "epoch": 0.19179680488242684,
      "grad_norm": 3.8445591926574707,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 2137
    },
    {
      "epoch": 0.19188655537605456,
      "grad_norm": 2.8632330894470215,
      "learning_rate": 0.0,
      "loss": 0.7675,
      "step": 2138
    },
    {
      "epoch": 0.19197630586968228,
      "grad_norm": 4.464412212371826,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 2139
    },
    {
      "epoch": 0.19206605636331,
      "grad_norm": 3.941152572631836,
      "learning_rate": 0.0,
      "loss": 0.806,
      "step": 2140
    },
    {
      "epoch": 0.1921558068569377,
      "grad_norm": 3.337491989135742,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 2141
    },
    {
      "epoch": 0.19224555735056542,
      "grad_norm": 3.3862686157226562,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 2142
    },
    {
      "epoch": 0.19233530784419314,
      "grad_norm": 3.0434083938598633,
      "learning_rate": 0.0,
      "loss": 0.7906,
      "step": 2143
    },
    {
      "epoch": 0.19242505833782086,
      "grad_norm": 2.950594425201416,
      "learning_rate": 0.0,
      "loss": 0.8041,
      "step": 2144
    },
    {
      "epoch": 0.1925148088314486,
      "grad_norm": 3.919257164001465,
      "learning_rate": 0.0,
      "loss": 0.8539,
      "step": 2145
    },
    {
      "epoch": 0.19260455932507628,
      "grad_norm": 3.2604153156280518,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 2146
    },
    {
      "epoch": 0.192694309818704,
      "grad_norm": 2.5294277667999268,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 2147
    },
    {
      "epoch": 0.19278406031233172,
      "grad_norm": 6.633860111236572,
      "learning_rate": 0.0,
      "loss": 0.7715,
      "step": 2148
    },
    {
      "epoch": 0.19287381080595944,
      "grad_norm": 3.746814250946045,
      "learning_rate": 0.0,
      "loss": 0.8233,
      "step": 2149
    },
    {
      "epoch": 0.19296356129958714,
      "grad_norm": 3.0332868099212646,
      "learning_rate": 0.0,
      "loss": 0.783,
      "step": 2150
    },
    {
      "epoch": 0.19305331179321486,
      "grad_norm": 7.397636890411377,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 2151
    },
    {
      "epoch": 0.19314306228684258,
      "grad_norm": 4.379924297332764,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 2152
    },
    {
      "epoch": 0.1932328127804703,
      "grad_norm": 6.193148612976074,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 2153
    },
    {
      "epoch": 0.193322563274098,
      "grad_norm": 3.3220129013061523,
      "learning_rate": 0.0,
      "loss": 0.6855,
      "step": 2154
    },
    {
      "epoch": 0.19341231376772572,
      "grad_norm": 3.8579816818237305,
      "learning_rate": 0.0,
      "loss": 0.7812,
      "step": 2155
    },
    {
      "epoch": 0.19350206426135344,
      "grad_norm": 3.5646047592163086,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 2156
    },
    {
      "epoch": 0.19359181475498116,
      "grad_norm": 3.1229934692382812,
      "learning_rate": 0.0,
      "loss": 0.808,
      "step": 2157
    },
    {
      "epoch": 0.19368156524860886,
      "grad_norm": 3.9430923461914062,
      "learning_rate": 0.0,
      "loss": 0.7712,
      "step": 2158
    },
    {
      "epoch": 0.19377131574223658,
      "grad_norm": 2.9541282653808594,
      "learning_rate": 0.0,
      "loss": 0.8113,
      "step": 2159
    },
    {
      "epoch": 0.1938610662358643,
      "grad_norm": 2.896094560623169,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 2160
    },
    {
      "epoch": 0.19395081672949202,
      "grad_norm": 4.653855323791504,
      "learning_rate": 0.0,
      "loss": 0.7916,
      "step": 2161
    },
    {
      "epoch": 0.19404056722311971,
      "grad_norm": 3.0450215339660645,
      "learning_rate": 0.0,
      "loss": 0.7465,
      "step": 2162
    },
    {
      "epoch": 0.19413031771674744,
      "grad_norm": 2.2856407165527344,
      "learning_rate": 0.0,
      "loss": 0.819,
      "step": 2163
    },
    {
      "epoch": 0.19422006821037516,
      "grad_norm": 2.7457737922668457,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 2164
    },
    {
      "epoch": 0.19430981870400288,
      "grad_norm": 2.725738525390625,
      "learning_rate": 0.0,
      "loss": 0.7796,
      "step": 2165
    },
    {
      "epoch": 0.1943995691976306,
      "grad_norm": 2.7631049156188965,
      "learning_rate": 0.0,
      "loss": 0.7906,
      "step": 2166
    },
    {
      "epoch": 0.1944893196912583,
      "grad_norm": 4.111721515655518,
      "learning_rate": 0.0,
      "loss": 0.8014,
      "step": 2167
    },
    {
      "epoch": 0.19457907018488602,
      "grad_norm": 3.019632577896118,
      "learning_rate": 0.0,
      "loss": 0.7986,
      "step": 2168
    },
    {
      "epoch": 0.19466882067851374,
      "grad_norm": 3.801276922225952,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 2169
    },
    {
      "epoch": 0.19475857117214146,
      "grad_norm": 2.5375635623931885,
      "learning_rate": 0.0,
      "loss": 0.7951,
      "step": 2170
    },
    {
      "epoch": 0.19484832166576915,
      "grad_norm": 3.5002992153167725,
      "learning_rate": 0.0,
      "loss": 0.8067,
      "step": 2171
    },
    {
      "epoch": 0.19493807215939687,
      "grad_norm": 2.785271406173706,
      "learning_rate": 0.0,
      "loss": 0.8157,
      "step": 2172
    },
    {
      "epoch": 0.1950278226530246,
      "grad_norm": 2.4730021953582764,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 2173
    },
    {
      "epoch": 0.19511757314665232,
      "grad_norm": 3.0004031658172607,
      "learning_rate": 0.0,
      "loss": 0.8148,
      "step": 2174
    },
    {
      "epoch": 0.19520732364028,
      "grad_norm": 4.4836297035217285,
      "learning_rate": 0.0,
      "loss": 0.797,
      "step": 2175
    },
    {
      "epoch": 0.19529707413390773,
      "grad_norm": 3.1909220218658447,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 2176
    },
    {
      "epoch": 0.19538682462753545,
      "grad_norm": 4.967761039733887,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 2177
    },
    {
      "epoch": 0.19547657512116318,
      "grad_norm": 4.082630634307861,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 2178
    },
    {
      "epoch": 0.19556632561479087,
      "grad_norm": 4.82273006439209,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 2179
    },
    {
      "epoch": 0.1956560761084186,
      "grad_norm": 3.2472116947174072,
      "learning_rate": 0.0,
      "loss": 0.7991,
      "step": 2180
    },
    {
      "epoch": 0.1957458266020463,
      "grad_norm": 6.849954605102539,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 2181
    },
    {
      "epoch": 0.19583557709567403,
      "grad_norm": 2.85833477973938,
      "learning_rate": 0.0,
      "loss": 0.8162,
      "step": 2182
    },
    {
      "epoch": 0.19592532758930173,
      "grad_norm": 2.5800294876098633,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 2183
    },
    {
      "epoch": 0.19601507808292945,
      "grad_norm": 7.440226078033447,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 2184
    },
    {
      "epoch": 0.19610482857655717,
      "grad_norm": 15.369662284851074,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 2185
    },
    {
      "epoch": 0.1961945790701849,
      "grad_norm": 3.531437873840332,
      "learning_rate": 0.0,
      "loss": 0.789,
      "step": 2186
    },
    {
      "epoch": 0.1962843295638126,
      "grad_norm": 3.007258415222168,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 2187
    },
    {
      "epoch": 0.1963740800574403,
      "grad_norm": 3.2347774505615234,
      "learning_rate": 0.0,
      "loss": 0.7574,
      "step": 2188
    },
    {
      "epoch": 0.19646383055106803,
      "grad_norm": 2.5135159492492676,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 2189
    },
    {
      "epoch": 0.19655358104469575,
      "grad_norm": 2.719676971435547,
      "learning_rate": 0.0,
      "loss": 0.7728,
      "step": 2190
    },
    {
      "epoch": 0.19664333153832347,
      "grad_norm": 3.2171695232391357,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 2191
    },
    {
      "epoch": 0.19673308203195117,
      "grad_norm": 7.568337917327881,
      "learning_rate": 0.0,
      "loss": 0.7835,
      "step": 2192
    },
    {
      "epoch": 0.1968228325255789,
      "grad_norm": 4.18361234664917,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 2193
    },
    {
      "epoch": 0.1969125830192066,
      "grad_norm": 3.0256845951080322,
      "learning_rate": 0.0,
      "loss": 0.7845,
      "step": 2194
    },
    {
      "epoch": 0.19700233351283433,
      "grad_norm": 3.448547124862671,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 2195
    },
    {
      "epoch": 0.19709208400646203,
      "grad_norm": 15.603425979614258,
      "learning_rate": 0.0,
      "loss": 0.7669,
      "step": 2196
    },
    {
      "epoch": 0.19718183450008975,
      "grad_norm": 4.009857177734375,
      "learning_rate": 0.0,
      "loss": 0.7829,
      "step": 2197
    },
    {
      "epoch": 0.19727158499371747,
      "grad_norm": 3.371084213256836,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 2198
    },
    {
      "epoch": 0.1973613354873452,
      "grad_norm": 20.923635482788086,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 2199
    },
    {
      "epoch": 0.19745108598097288,
      "grad_norm": 3.0773987770080566,
      "learning_rate": 0.0,
      "loss": 0.7744,
      "step": 2200
    },
    {
      "epoch": 0.1975408364746006,
      "grad_norm": 3.3840203285217285,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 2201
    },
    {
      "epoch": 0.19763058696822833,
      "grad_norm": 3.236560583114624,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 2202
    },
    {
      "epoch": 0.19772033746185605,
      "grad_norm": 3.6688032150268555,
      "learning_rate": 0.0,
      "loss": 0.7274,
      "step": 2203
    },
    {
      "epoch": 0.19781008795548374,
      "grad_norm": 3.996459484100342,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 2204
    },
    {
      "epoch": 0.19789983844911146,
      "grad_norm": 3.489039659500122,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 2205
    },
    {
      "epoch": 0.1979895889427392,
      "grad_norm": 4.537407398223877,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 2206
    },
    {
      "epoch": 0.1980793394363669,
      "grad_norm": 3.3163270950317383,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 2207
    },
    {
      "epoch": 0.1981690899299946,
      "grad_norm": 3.592301607131958,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 2208
    },
    {
      "epoch": 0.19825884042362232,
      "grad_norm": 3.908378839492798,
      "learning_rate": 0.0,
      "loss": 0.7955,
      "step": 2209
    },
    {
      "epoch": 0.19834859091725004,
      "grad_norm": 11.618204116821289,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 2210
    },
    {
      "epoch": 0.19843834141087777,
      "grad_norm": 6.146525859832764,
      "learning_rate": 0.0,
      "loss": 0.7737,
      "step": 2211
    },
    {
      "epoch": 0.1985280919045055,
      "grad_norm": 2.8847670555114746,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 2212
    },
    {
      "epoch": 0.19861784239813318,
      "grad_norm": 5.337097644805908,
      "learning_rate": 0.0,
      "loss": 0.7806,
      "step": 2213
    },
    {
      "epoch": 0.1987075928917609,
      "grad_norm": 4.432526588439941,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 2214
    },
    {
      "epoch": 0.19879734338538863,
      "grad_norm": 4.932809829711914,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 2215
    },
    {
      "epoch": 0.19888709387901635,
      "grad_norm": 2.909632444381714,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 2216
    },
    {
      "epoch": 0.19897684437264404,
      "grad_norm": 3.657130718231201,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 2217
    },
    {
      "epoch": 0.19906659486627176,
      "grad_norm": 3.1289212703704834,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 2218
    },
    {
      "epoch": 0.19915634535989948,
      "grad_norm": 6.046507835388184,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 2219
    },
    {
      "epoch": 0.1992460958535272,
      "grad_norm": 7.11259651184082,
      "learning_rate": 0.0,
      "loss": 0.8016,
      "step": 2220
    },
    {
      "epoch": 0.1993358463471549,
      "grad_norm": 4.895362377166748,
      "learning_rate": 0.0,
      "loss": 0.8146,
      "step": 2221
    },
    {
      "epoch": 0.19942559684078262,
      "grad_norm": 9.995196342468262,
      "learning_rate": 0.0,
      "loss": 0.8548,
      "step": 2222
    },
    {
      "epoch": 0.19951534733441034,
      "grad_norm": 5.751162052154541,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 2223
    },
    {
      "epoch": 0.19960509782803806,
      "grad_norm": 3.8694095611572266,
      "learning_rate": 0.0,
      "loss": 0.8116,
      "step": 2224
    },
    {
      "epoch": 0.19969484832166576,
      "grad_norm": 6.027638912200928,
      "learning_rate": 0.0,
      "loss": 0.7496,
      "step": 2225
    },
    {
      "epoch": 0.19978459881529348,
      "grad_norm": 3.3180720806121826,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 2226
    },
    {
      "epoch": 0.1998743493089212,
      "grad_norm": 17.143556594848633,
      "learning_rate": 0.0,
      "loss": 0.7534,
      "step": 2227
    },
    {
      "epoch": 0.19996409980254892,
      "grad_norm": 4.952237129211426,
      "learning_rate": 0.0,
      "loss": 0.7782,
      "step": 2228
    },
    {
      "epoch": 0.20005385029617662,
      "grad_norm": 3.6299448013305664,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 2229
    },
    {
      "epoch": 0.20014360078980434,
      "grad_norm": 9.118568420410156,
      "learning_rate": 0.0,
      "loss": 0.7582,
      "step": 2230
    },
    {
      "epoch": 0.20023335128343206,
      "grad_norm": 3.1203341484069824,
      "learning_rate": 0.0,
      "loss": 0.7941,
      "step": 2231
    },
    {
      "epoch": 0.20032310177705978,
      "grad_norm": 4.751515865325928,
      "learning_rate": 0.0,
      "loss": 0.7913,
      "step": 2232
    },
    {
      "epoch": 0.2004128522706875,
      "grad_norm": 4.358302593231201,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 2233
    },
    {
      "epoch": 0.2005026027643152,
      "grad_norm": 4.0257439613342285,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 2234
    },
    {
      "epoch": 0.20059235325794292,
      "grad_norm": 2.78872013092041,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 2235
    },
    {
      "epoch": 0.20068210375157064,
      "grad_norm": 7.276110649108887,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 2236
    },
    {
      "epoch": 0.20077185424519836,
      "grad_norm": 8.273983001708984,
      "learning_rate": 0.0,
      "loss": 0.7801,
      "step": 2237
    },
    {
      "epoch": 0.20086160473882606,
      "grad_norm": 5.317710876464844,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 2238
    },
    {
      "epoch": 0.20095135523245378,
      "grad_norm": 3.5138888359069824,
      "learning_rate": 0.0,
      "loss": 0.7797,
      "step": 2239
    },
    {
      "epoch": 0.2010411057260815,
      "grad_norm": 8.838828086853027,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 2240
    },
    {
      "epoch": 0.20113085621970922,
      "grad_norm": 3.499608278274536,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 2241
    },
    {
      "epoch": 0.2012206067133369,
      "grad_norm": 3.8159103393554688,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 2242
    },
    {
      "epoch": 0.20131035720696464,
      "grad_norm": 2.8563475608825684,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 2243
    },
    {
      "epoch": 0.20140010770059236,
      "grad_norm": 2.5088772773742676,
      "learning_rate": 0.0,
      "loss": 0.8034,
      "step": 2244
    },
    {
      "epoch": 0.20148985819422008,
      "grad_norm": 4.5910210609436035,
      "learning_rate": 0.0,
      "loss": 0.8619,
      "step": 2245
    },
    {
      "epoch": 0.20157960868784777,
      "grad_norm": 3.40151309967041,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 2246
    },
    {
      "epoch": 0.2016693591814755,
      "grad_norm": 2.8891186714172363,
      "learning_rate": 0.0,
      "loss": 0.8019,
      "step": 2247
    },
    {
      "epoch": 0.20175910967510322,
      "grad_norm": 7.444084167480469,
      "learning_rate": 0.0,
      "loss": 0.8064,
      "step": 2248
    },
    {
      "epoch": 0.20184886016873094,
      "grad_norm": 2.9873290061950684,
      "learning_rate": 0.0,
      "loss": 0.82,
      "step": 2249
    },
    {
      "epoch": 0.20193861066235863,
      "grad_norm": 3.3589162826538086,
      "learning_rate": 0.0,
      "loss": 0.7825,
      "step": 2250
    },
    {
      "epoch": 0.20202836115598635,
      "grad_norm": 3.388178825378418,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 2251
    },
    {
      "epoch": 0.20211811164961407,
      "grad_norm": 2.9608705043792725,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 2252
    },
    {
      "epoch": 0.2022078621432418,
      "grad_norm": 2.392763376235962,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 2253
    },
    {
      "epoch": 0.2022976126368695,
      "grad_norm": 5.377963542938232,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 2254
    },
    {
      "epoch": 0.2023873631304972,
      "grad_norm": 8.600228309631348,
      "learning_rate": 0.0,
      "loss": 0.78,
      "step": 2255
    },
    {
      "epoch": 0.20247711362412493,
      "grad_norm": 3.466026782989502,
      "learning_rate": 0.0,
      "loss": 0.7858,
      "step": 2256
    },
    {
      "epoch": 0.20256686411775265,
      "grad_norm": 3.0194101333618164,
      "learning_rate": 0.0,
      "loss": 0.7649,
      "step": 2257
    },
    {
      "epoch": 0.20265661461138038,
      "grad_norm": 4.014604568481445,
      "learning_rate": 0.0,
      "loss": 0.7998,
      "step": 2258
    },
    {
      "epoch": 0.20274636510500807,
      "grad_norm": 38.8354606628418,
      "learning_rate": 0.0,
      "loss": 0.7999,
      "step": 2259
    },
    {
      "epoch": 0.2028361155986358,
      "grad_norm": 4.568796157836914,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 2260
    },
    {
      "epoch": 0.2029258660922635,
      "grad_norm": 11.49007511138916,
      "learning_rate": 0.0,
      "loss": 0.7866,
      "step": 2261
    },
    {
      "epoch": 0.20301561658589123,
      "grad_norm": 5.680341720581055,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 2262
    },
    {
      "epoch": 0.20310536707951893,
      "grad_norm": 4.765652656555176,
      "learning_rate": 0.0,
      "loss": 0.8226,
      "step": 2263
    },
    {
      "epoch": 0.20319511757314665,
      "grad_norm": 5.609388828277588,
      "learning_rate": 0.0,
      "loss": 0.827,
      "step": 2264
    },
    {
      "epoch": 0.20328486806677437,
      "grad_norm": 4.02333402633667,
      "learning_rate": 0.0,
      "loss": 0.8149,
      "step": 2265
    },
    {
      "epoch": 0.2033746185604021,
      "grad_norm": 4.058172702789307,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 2266
    },
    {
      "epoch": 0.2034643690540298,
      "grad_norm": 10.032562255859375,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 2267
    },
    {
      "epoch": 0.2035541195476575,
      "grad_norm": 2.763127088546753,
      "learning_rate": 0.0,
      "loss": 0.7599,
      "step": 2268
    },
    {
      "epoch": 0.20364387004128523,
      "grad_norm": 20.986709594726562,
      "learning_rate": 0.0,
      "loss": 0.8058,
      "step": 2269
    },
    {
      "epoch": 0.20373362053491295,
      "grad_norm": 6.026888370513916,
      "learning_rate": 0.0,
      "loss": 0.7572,
      "step": 2270
    },
    {
      "epoch": 0.20382337102854065,
      "grad_norm": 7.107997417449951,
      "learning_rate": 0.0,
      "loss": 0.7908,
      "step": 2271
    },
    {
      "epoch": 0.20391312152216837,
      "grad_norm": 4.227787017822266,
      "learning_rate": 0.0,
      "loss": 0.76,
      "step": 2272
    },
    {
      "epoch": 0.2040028720157961,
      "grad_norm": 6.404241561889648,
      "learning_rate": 0.0,
      "loss": 0.7636,
      "step": 2273
    },
    {
      "epoch": 0.2040926225094238,
      "grad_norm": 3.733548879623413,
      "learning_rate": 0.0,
      "loss": 0.7739,
      "step": 2274
    },
    {
      "epoch": 0.2041823730030515,
      "grad_norm": 2.6638402938842773,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 2275
    },
    {
      "epoch": 0.20427212349667923,
      "grad_norm": 4.5178961753845215,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 2276
    },
    {
      "epoch": 0.20436187399030695,
      "grad_norm": 3.3895018100738525,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 2277
    },
    {
      "epoch": 0.20445162448393467,
      "grad_norm": 7.648850917816162,
      "learning_rate": 0.0,
      "loss": 0.8281,
      "step": 2278
    },
    {
      "epoch": 0.2045413749775624,
      "grad_norm": 3.1186273097991943,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 2279
    },
    {
      "epoch": 0.20463112547119008,
      "grad_norm": 3.6941542625427246,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 2280
    },
    {
      "epoch": 0.2047208759648178,
      "grad_norm": 3.118858814239502,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 2281
    },
    {
      "epoch": 0.20481062645844553,
      "grad_norm": 9.242000579833984,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 2282
    },
    {
      "epoch": 0.20490037695207325,
      "grad_norm": 8.45477294921875,
      "learning_rate": 0.0,
      "loss": 0.788,
      "step": 2283
    },
    {
      "epoch": 0.20499012744570094,
      "grad_norm": 5.4878339767456055,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 2284
    },
    {
      "epoch": 0.20507987793932866,
      "grad_norm": 4.091027736663818,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 2285
    },
    {
      "epoch": 0.20516962843295639,
      "grad_norm": 2.70306134223938,
      "learning_rate": 0.0,
      "loss": 0.7533,
      "step": 2286
    },
    {
      "epoch": 0.2052593789265841,
      "grad_norm": 3.744185209274292,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 2287
    },
    {
      "epoch": 0.2053491294202118,
      "grad_norm": 4.159204959869385,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 2288
    },
    {
      "epoch": 0.20543887991383952,
      "grad_norm": 5.261654376983643,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 2289
    },
    {
      "epoch": 0.20552863040746724,
      "grad_norm": 3.7271440029144287,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 2290
    },
    {
      "epoch": 0.20561838090109497,
      "grad_norm": 2.9320454597473145,
      "learning_rate": 0.0,
      "loss": 0.7973,
      "step": 2291
    },
    {
      "epoch": 0.20570813139472266,
      "grad_norm": 2.394409656524658,
      "learning_rate": 0.0,
      "loss": 0.8398,
      "step": 2292
    },
    {
      "epoch": 0.20579788188835038,
      "grad_norm": 2.1305248737335205,
      "learning_rate": 0.0,
      "loss": 0.7913,
      "step": 2293
    },
    {
      "epoch": 0.2058876323819781,
      "grad_norm": 2.3739492893218994,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 2294
    },
    {
      "epoch": 0.20597738287560582,
      "grad_norm": 3.0037217140197754,
      "learning_rate": 0.0,
      "loss": 0.7661,
      "step": 2295
    },
    {
      "epoch": 0.20606713336923352,
      "grad_norm": 2.9854679107666016,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 2296
    },
    {
      "epoch": 0.20615688386286124,
      "grad_norm": 2.906658411026001,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 2297
    },
    {
      "epoch": 0.20624663435648896,
      "grad_norm": 4.949425220489502,
      "learning_rate": 0.0,
      "loss": 0.7978,
      "step": 2298
    },
    {
      "epoch": 0.20633638485011668,
      "grad_norm": 2.9641456604003906,
      "learning_rate": 0.0,
      "loss": 0.768,
      "step": 2299
    },
    {
      "epoch": 0.20642613534374438,
      "grad_norm": 2.2009410858154297,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 2300
    },
    {
      "epoch": 0.2065158858373721,
      "grad_norm": 4.257469654083252,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 2301
    },
    {
      "epoch": 0.20660563633099982,
      "grad_norm": 2.618215322494507,
      "learning_rate": 0.0,
      "loss": 0.7888,
      "step": 2302
    },
    {
      "epoch": 0.20669538682462754,
      "grad_norm": 5.625435829162598,
      "learning_rate": 0.0,
      "loss": 0.7655,
      "step": 2303
    },
    {
      "epoch": 0.20678513731825526,
      "grad_norm": 3.658951759338379,
      "learning_rate": 0.0,
      "loss": 0.8032,
      "step": 2304
    },
    {
      "epoch": 0.20687488781188296,
      "grad_norm": 2.9706430435180664,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 2305
    },
    {
      "epoch": 0.20696463830551068,
      "grad_norm": 2.9511635303497314,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 2306
    },
    {
      "epoch": 0.2070543887991384,
      "grad_norm": 5.163003921508789,
      "learning_rate": 0.0,
      "loss": 0.8105,
      "step": 2307
    },
    {
      "epoch": 0.20714413929276612,
      "grad_norm": 3.3167715072631836,
      "learning_rate": 0.0,
      "loss": 0.8234,
      "step": 2308
    },
    {
      "epoch": 0.20723388978639382,
      "grad_norm": 3.402282953262329,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 2309
    },
    {
      "epoch": 0.20732364028002154,
      "grad_norm": 8.23836898803711,
      "learning_rate": 0.0,
      "loss": 0.781,
      "step": 2310
    },
    {
      "epoch": 0.20741339077364926,
      "grad_norm": 4.491726875305176,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 2311
    },
    {
      "epoch": 0.20750314126727698,
      "grad_norm": 3.791276454925537,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 2312
    },
    {
      "epoch": 0.20759289176090467,
      "grad_norm": 4.894859313964844,
      "learning_rate": 0.0,
      "loss": 0.8015,
      "step": 2313
    },
    {
      "epoch": 0.2076826422545324,
      "grad_norm": 5.294313430786133,
      "learning_rate": 0.0,
      "loss": 0.7968,
      "step": 2314
    },
    {
      "epoch": 0.20777239274816012,
      "grad_norm": 4.730602264404297,
      "learning_rate": 0.0,
      "loss": 0.8072,
      "step": 2315
    },
    {
      "epoch": 0.20786214324178784,
      "grad_norm": 2.943932056427002,
      "learning_rate": 0.0,
      "loss": 0.8128,
      "step": 2316
    },
    {
      "epoch": 0.20795189373541553,
      "grad_norm": 2.574319362640381,
      "learning_rate": 0.0,
      "loss": 0.7729,
      "step": 2317
    },
    {
      "epoch": 0.20804164422904325,
      "grad_norm": 3.496338129043579,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 2318
    },
    {
      "epoch": 0.20813139472267098,
      "grad_norm": 3.7651774883270264,
      "learning_rate": 0.0,
      "loss": 0.8004,
      "step": 2319
    },
    {
      "epoch": 0.2082211452162987,
      "grad_norm": 3.534339666366577,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 2320
    },
    {
      "epoch": 0.2083108957099264,
      "grad_norm": 2.600792169570923,
      "learning_rate": 0.0,
      "loss": 0.7924,
      "step": 2321
    },
    {
      "epoch": 0.2084006462035541,
      "grad_norm": 4.318081378936768,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 2322
    },
    {
      "epoch": 0.20849039669718183,
      "grad_norm": 2.8172996044158936,
      "learning_rate": 0.0,
      "loss": 0.8365,
      "step": 2323
    },
    {
      "epoch": 0.20858014719080956,
      "grad_norm": 3.327082395553589,
      "learning_rate": 0.0,
      "loss": 0.7871,
      "step": 2324
    },
    {
      "epoch": 0.20866989768443728,
      "grad_norm": 3.899965524673462,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 2325
    },
    {
      "epoch": 0.20875964817806497,
      "grad_norm": 3.9202866554260254,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 2326
    },
    {
      "epoch": 0.2088493986716927,
      "grad_norm": 3.67667555809021,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 2327
    },
    {
      "epoch": 0.20893914916532041,
      "grad_norm": 2.5708608627319336,
      "learning_rate": 0.0,
      "loss": 0.7894,
      "step": 2328
    },
    {
      "epoch": 0.20902889965894814,
      "grad_norm": 3.7453718185424805,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 2329
    },
    {
      "epoch": 0.20911865015257583,
      "grad_norm": 4.234838962554932,
      "learning_rate": 0.0,
      "loss": 0.8005,
      "step": 2330
    },
    {
      "epoch": 0.20920840064620355,
      "grad_norm": 2.725801467895508,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 2331
    },
    {
      "epoch": 0.20929815113983127,
      "grad_norm": 2.4908599853515625,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 2332
    },
    {
      "epoch": 0.209387901633459,
      "grad_norm": 2.9338319301605225,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 2333
    },
    {
      "epoch": 0.2094776521270867,
      "grad_norm": 11.847864151000977,
      "learning_rate": 0.0,
      "loss": 0.7855,
      "step": 2334
    },
    {
      "epoch": 0.2095674026207144,
      "grad_norm": 3.2491352558135986,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 2335
    },
    {
      "epoch": 0.20965715311434213,
      "grad_norm": 6.994534492492676,
      "learning_rate": 0.0,
      "loss": 0.8753,
      "step": 2336
    },
    {
      "epoch": 0.20974690360796985,
      "grad_norm": 2.8317794799804688,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 2337
    },
    {
      "epoch": 0.20983665410159755,
      "grad_norm": 5.993319034576416,
      "learning_rate": 0.0,
      "loss": 0.8233,
      "step": 2338
    },
    {
      "epoch": 0.20992640459522527,
      "grad_norm": 3.786970376968384,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 2339
    },
    {
      "epoch": 0.210016155088853,
      "grad_norm": 3.096719980239868,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 2340
    },
    {
      "epoch": 0.2101059055824807,
      "grad_norm": 2.658482074737549,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 2341
    },
    {
      "epoch": 0.2101956560761084,
      "grad_norm": 3.753089427947998,
      "learning_rate": 0.0,
      "loss": 0.7929,
      "step": 2342
    },
    {
      "epoch": 0.21028540656973613,
      "grad_norm": 3.2012743949890137,
      "learning_rate": 0.0,
      "loss": 0.7951,
      "step": 2343
    },
    {
      "epoch": 0.21037515706336385,
      "grad_norm": 2.3110156059265137,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 2344
    },
    {
      "epoch": 0.21046490755699157,
      "grad_norm": 2.8872525691986084,
      "learning_rate": 0.0,
      "loss": 0.7618,
      "step": 2345
    },
    {
      "epoch": 0.2105546580506193,
      "grad_norm": 2.9268319606781006,
      "learning_rate": 0.0,
      "loss": 0.7977,
      "step": 2346
    },
    {
      "epoch": 0.21064440854424699,
      "grad_norm": 2.403604030609131,
      "learning_rate": 0.0,
      "loss": 0.8026,
      "step": 2347
    },
    {
      "epoch": 0.2107341590378747,
      "grad_norm": 3.3782145977020264,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 2348
    },
    {
      "epoch": 0.21082390953150243,
      "grad_norm": 2.9228856563568115,
      "learning_rate": 0.0,
      "loss": 0.7978,
      "step": 2349
    },
    {
      "epoch": 0.21091366002513015,
      "grad_norm": 3.0889322757720947,
      "learning_rate": 0.0,
      "loss": 0.7495,
      "step": 2350
    },
    {
      "epoch": 0.21100341051875784,
      "grad_norm": 4.357146263122559,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "step": 2351
    },
    {
      "epoch": 0.21109316101238557,
      "grad_norm": 7.35590934753418,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 2352
    },
    {
      "epoch": 0.2111829115060133,
      "grad_norm": 4.1772050857543945,
      "learning_rate": 0.0,
      "loss": 0.7988,
      "step": 2353
    },
    {
      "epoch": 0.211272661999641,
      "grad_norm": 2.6700408458709717,
      "learning_rate": 0.0,
      "loss": 0.8005,
      "step": 2354
    },
    {
      "epoch": 0.2113624124932687,
      "grad_norm": 2.4982144832611084,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 2355
    },
    {
      "epoch": 0.21145216298689642,
      "grad_norm": 4.560385227203369,
      "learning_rate": 0.0,
      "loss": 0.8295,
      "step": 2356
    },
    {
      "epoch": 0.21154191348052415,
      "grad_norm": 2.564913034439087,
      "learning_rate": 0.0,
      "loss": 0.7932,
      "step": 2357
    },
    {
      "epoch": 0.21163166397415187,
      "grad_norm": 2.6057450771331787,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 2358
    },
    {
      "epoch": 0.21172141446777956,
      "grad_norm": 2.581486225128174,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 2359
    },
    {
      "epoch": 0.21181116496140728,
      "grad_norm": 2.422912836074829,
      "learning_rate": 0.0,
      "loss": 0.7969,
      "step": 2360
    },
    {
      "epoch": 0.211900915455035,
      "grad_norm": 3.973947048187256,
      "learning_rate": 0.0,
      "loss": 0.7808,
      "step": 2361
    },
    {
      "epoch": 0.21199066594866273,
      "grad_norm": 2.2911458015441895,
      "learning_rate": 0.0,
      "loss": 0.7901,
      "step": 2362
    },
    {
      "epoch": 0.21208041644229042,
      "grad_norm": 2.7770025730133057,
      "learning_rate": 0.0,
      "loss": 0.6944,
      "step": 2363
    },
    {
      "epoch": 0.21217016693591814,
      "grad_norm": 3.7323505878448486,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 2364
    },
    {
      "epoch": 0.21225991742954586,
      "grad_norm": 4.774266242980957,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 2365
    },
    {
      "epoch": 0.21234966792317359,
      "grad_norm": 3.4827475547790527,
      "learning_rate": 0.0,
      "loss": 0.7814,
      "step": 2366
    },
    {
      "epoch": 0.21243941841680128,
      "grad_norm": 3.9826109409332275,
      "learning_rate": 0.0,
      "loss": 0.8135,
      "step": 2367
    },
    {
      "epoch": 0.212529168910429,
      "grad_norm": 3.9140217304229736,
      "learning_rate": 0.0,
      "loss": 0.7675,
      "step": 2368
    },
    {
      "epoch": 0.21261891940405672,
      "grad_norm": 3.0464611053466797,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 2369
    },
    {
      "epoch": 0.21270866989768444,
      "grad_norm": 2.645737409591675,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 2370
    },
    {
      "epoch": 0.21279842039131217,
      "grad_norm": 2.693915367126465,
      "learning_rate": 0.0,
      "loss": 0.785,
      "step": 2371
    },
    {
      "epoch": 0.21288817088493986,
      "grad_norm": 6.421985626220703,
      "learning_rate": 0.0,
      "loss": 0.786,
      "step": 2372
    },
    {
      "epoch": 0.21297792137856758,
      "grad_norm": 3.562808036804199,
      "learning_rate": 0.0,
      "loss": 0.8271,
      "step": 2373
    },
    {
      "epoch": 0.2130676718721953,
      "grad_norm": 4.179168701171875,
      "learning_rate": 0.0,
      "loss": 0.806,
      "step": 2374
    },
    {
      "epoch": 0.21315742236582302,
      "grad_norm": 2.8255252838134766,
      "learning_rate": 0.0,
      "loss": 0.8334,
      "step": 2375
    },
    {
      "epoch": 0.21324717285945072,
      "grad_norm": 2.861527442932129,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 2376
    },
    {
      "epoch": 0.21333692335307844,
      "grad_norm": 3.8970558643341064,
      "learning_rate": 0.0,
      "loss": 0.7684,
      "step": 2377
    },
    {
      "epoch": 0.21342667384670616,
      "grad_norm": 2.8964803218841553,
      "learning_rate": 0.0,
      "loss": 0.7655,
      "step": 2378
    },
    {
      "epoch": 0.21351642434033388,
      "grad_norm": 2.06477689743042,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 2379
    },
    {
      "epoch": 0.21360617483396158,
      "grad_norm": 12.519848823547363,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 2380
    },
    {
      "epoch": 0.2136959253275893,
      "grad_norm": 3.2735612392425537,
      "learning_rate": 0.0,
      "loss": 0.7823,
      "step": 2381
    },
    {
      "epoch": 0.21378567582121702,
      "grad_norm": 3.125627279281616,
      "learning_rate": 0.0,
      "loss": 0.7792,
      "step": 2382
    },
    {
      "epoch": 0.21387542631484474,
      "grad_norm": 3.1069493293762207,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 2383
    },
    {
      "epoch": 0.21396517680847243,
      "grad_norm": 2.691816568374634,
      "learning_rate": 0.0,
      "loss": 0.7976,
      "step": 2384
    },
    {
      "epoch": 0.21405492730210016,
      "grad_norm": 2.3670578002929688,
      "learning_rate": 0.0,
      "loss": 0.7873,
      "step": 2385
    },
    {
      "epoch": 0.21414467779572788,
      "grad_norm": 2.5833990573883057,
      "learning_rate": 0.0,
      "loss": 0.8456,
      "step": 2386
    },
    {
      "epoch": 0.2142344282893556,
      "grad_norm": 2.4893558025360107,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 2387
    },
    {
      "epoch": 0.2143241787829833,
      "grad_norm": 2.767611265182495,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 2388
    },
    {
      "epoch": 0.21441392927661101,
      "grad_norm": 2.59330677986145,
      "learning_rate": 0.0,
      "loss": 0.7999,
      "step": 2389
    },
    {
      "epoch": 0.21450367977023874,
      "grad_norm": 3.1483733654022217,
      "learning_rate": 0.0,
      "loss": 0.8028,
      "step": 2390
    },
    {
      "epoch": 0.21459343026386646,
      "grad_norm": 4.206470012664795,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 2391
    },
    {
      "epoch": 0.21468318075749418,
      "grad_norm": 2.806701421737671,
      "learning_rate": 0.0,
      "loss": 0.8147,
      "step": 2392
    },
    {
      "epoch": 0.21477293125112187,
      "grad_norm": 2.716560125350952,
      "learning_rate": 0.0,
      "loss": 0.8855,
      "step": 2393
    },
    {
      "epoch": 0.2148626817447496,
      "grad_norm": 2.971086025238037,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 2394
    },
    {
      "epoch": 0.21495243223837732,
      "grad_norm": 3.3203983306884766,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 2395
    },
    {
      "epoch": 0.21504218273200504,
      "grad_norm": 2.6537461280822754,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 2396
    },
    {
      "epoch": 0.21513193322563273,
      "grad_norm": 6.8133111000061035,
      "learning_rate": 0.0,
      "loss": 0.8153,
      "step": 2397
    },
    {
      "epoch": 0.21522168371926045,
      "grad_norm": 2.738568067550659,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 2398
    },
    {
      "epoch": 0.21531143421288818,
      "grad_norm": 3.6319658756256104,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 2399
    },
    {
      "epoch": 0.2154011847065159,
      "grad_norm": 3.6068387031555176,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 2400
    },
    {
      "epoch": 0.2154909352001436,
      "grad_norm": 4.867395877838135,
      "learning_rate": 0.0,
      "loss": 0.787,
      "step": 2401
    },
    {
      "epoch": 0.2155806856937713,
      "grad_norm": 2.2239503860473633,
      "learning_rate": 0.0,
      "loss": 0.7747,
      "step": 2402
    },
    {
      "epoch": 0.21567043618739903,
      "grad_norm": 2.4911792278289795,
      "learning_rate": 0.0,
      "loss": 0.8013,
      "step": 2403
    },
    {
      "epoch": 0.21576018668102676,
      "grad_norm": 3.197016954421997,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 2404
    },
    {
      "epoch": 0.21584993717465445,
      "grad_norm": 3.691425323486328,
      "learning_rate": 0.0,
      "loss": 0.8175,
      "step": 2405
    },
    {
      "epoch": 0.21593968766828217,
      "grad_norm": 2.4766311645507812,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 2406
    },
    {
      "epoch": 0.2160294381619099,
      "grad_norm": 5.055393218994141,
      "learning_rate": 0.0,
      "loss": 0.8128,
      "step": 2407
    },
    {
      "epoch": 0.21611918865553761,
      "grad_norm": 2.1710965633392334,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 2408
    },
    {
      "epoch": 0.2162089391491653,
      "grad_norm": 2.498918294906616,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 2409
    },
    {
      "epoch": 0.21629868964279303,
      "grad_norm": 2.5697104930877686,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 2410
    },
    {
      "epoch": 0.21638844013642075,
      "grad_norm": 4.161758899688721,
      "learning_rate": 0.0,
      "loss": 0.7972,
      "step": 2411
    },
    {
      "epoch": 0.21647819063004847,
      "grad_norm": 2.2940680980682373,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 2412
    },
    {
      "epoch": 0.21656794112367617,
      "grad_norm": 3.773338794708252,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 2413
    },
    {
      "epoch": 0.2166576916173039,
      "grad_norm": 4.400176048278809,
      "learning_rate": 0.0,
      "loss": 0.7579,
      "step": 2414
    },
    {
      "epoch": 0.2167474421109316,
      "grad_norm": 2.6389641761779785,
      "learning_rate": 0.0,
      "loss": 0.8028,
      "step": 2415
    },
    {
      "epoch": 0.21683719260455933,
      "grad_norm": 2.3536086082458496,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 2416
    },
    {
      "epoch": 0.21692694309818705,
      "grad_norm": 2.259079933166504,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 2417
    },
    {
      "epoch": 0.21701669359181475,
      "grad_norm": 2.258855104446411,
      "learning_rate": 0.0,
      "loss": 0.7866,
      "step": 2418
    },
    {
      "epoch": 0.21710644408544247,
      "grad_norm": 3.1993162631988525,
      "learning_rate": 0.0,
      "loss": 0.7596,
      "step": 2419
    },
    {
      "epoch": 0.2171961945790702,
      "grad_norm": 2.5555825233459473,
      "learning_rate": 0.0,
      "loss": 0.7659,
      "step": 2420
    },
    {
      "epoch": 0.2172859450726979,
      "grad_norm": 2.6634140014648438,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 2421
    },
    {
      "epoch": 0.2173756955663256,
      "grad_norm": 2.512340545654297,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 2422
    },
    {
      "epoch": 0.21746544605995333,
      "grad_norm": 2.700496196746826,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 2423
    },
    {
      "epoch": 0.21755519655358105,
      "grad_norm": 2.9188284873962402,
      "learning_rate": 0.0,
      "loss": 0.7463,
      "step": 2424
    },
    {
      "epoch": 0.21764494704720877,
      "grad_norm": 3.2389731407165527,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 2425
    },
    {
      "epoch": 0.21773469754083646,
      "grad_norm": 2.440244197845459,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 2426
    },
    {
      "epoch": 0.21782444803446419,
      "grad_norm": 3.0864524841308594,
      "learning_rate": 0.0,
      "loss": 0.7728,
      "step": 2427
    },
    {
      "epoch": 0.2179141985280919,
      "grad_norm": 2.220752716064453,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 2428
    },
    {
      "epoch": 0.21800394902171963,
      "grad_norm": 2.95298171043396,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 2429
    },
    {
      "epoch": 0.21809369951534732,
      "grad_norm": 2.625582695007324,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 2430
    },
    {
      "epoch": 0.21818345000897504,
      "grad_norm": 2.9593987464904785,
      "learning_rate": 0.0,
      "loss": 0.7796,
      "step": 2431
    },
    {
      "epoch": 0.21827320050260277,
      "grad_norm": 3.8305604457855225,
      "learning_rate": 0.0,
      "loss": 0.8127,
      "step": 2432
    },
    {
      "epoch": 0.2183629509962305,
      "grad_norm": 2.4033491611480713,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 2433
    },
    {
      "epoch": 0.21845270148985818,
      "grad_norm": 2.3949038982391357,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 2434
    },
    {
      "epoch": 0.2185424519834859,
      "grad_norm": 2.2684240341186523,
      "learning_rate": 0.0,
      "loss": 0.7886,
      "step": 2435
    },
    {
      "epoch": 0.21863220247711362,
      "grad_norm": 3.006037950515747,
      "learning_rate": 0.0,
      "loss": 0.7728,
      "step": 2436
    },
    {
      "epoch": 0.21872195297074135,
      "grad_norm": 2.610511302947998,
      "learning_rate": 0.0,
      "loss": 0.7687,
      "step": 2437
    },
    {
      "epoch": 0.21881170346436907,
      "grad_norm": 2.4069066047668457,
      "learning_rate": 0.0,
      "loss": 0.7599,
      "step": 2438
    },
    {
      "epoch": 0.21890145395799676,
      "grad_norm": 3.3453142642974854,
      "learning_rate": 0.0,
      "loss": 0.7861,
      "step": 2439
    },
    {
      "epoch": 0.21899120445162448,
      "grad_norm": 4.049151420593262,
      "learning_rate": 0.0,
      "loss": 0.831,
      "step": 2440
    },
    {
      "epoch": 0.2190809549452522,
      "grad_norm": 2.3896584510803223,
      "learning_rate": 0.0,
      "loss": 0.7524,
      "step": 2441
    },
    {
      "epoch": 0.21917070543887993,
      "grad_norm": 3.575253963470459,
      "learning_rate": 0.0,
      "loss": 0.799,
      "step": 2442
    },
    {
      "epoch": 0.21926045593250762,
      "grad_norm": 3.015415906906128,
      "learning_rate": 0.0,
      "loss": 0.7873,
      "step": 2443
    },
    {
      "epoch": 0.21935020642613534,
      "grad_norm": 2.6575145721435547,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 2444
    },
    {
      "epoch": 0.21943995691976306,
      "grad_norm": 2.0574963092803955,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 2445
    },
    {
      "epoch": 0.21952970741339078,
      "grad_norm": 2.675781726837158,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 2446
    },
    {
      "epoch": 0.21961945790701848,
      "grad_norm": 2.3459603786468506,
      "learning_rate": 0.0,
      "loss": 0.8056,
      "step": 2447
    },
    {
      "epoch": 0.2197092084006462,
      "grad_norm": 3.5193116664886475,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 2448
    },
    {
      "epoch": 0.21979895889427392,
      "grad_norm": 4.225274562835693,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 2449
    },
    {
      "epoch": 0.21988870938790164,
      "grad_norm": 3.415813446044922,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 2450
    },
    {
      "epoch": 0.21997845988152934,
      "grad_norm": 2.7275288105010986,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 2451
    },
    {
      "epoch": 0.22006821037515706,
      "grad_norm": 2.9943180084228516,
      "learning_rate": 0.0,
      "loss": 0.85,
      "step": 2452
    },
    {
      "epoch": 0.22015796086878478,
      "grad_norm": 3.549811363220215,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 2453
    },
    {
      "epoch": 0.2202477113624125,
      "grad_norm": 3.564969062805176,
      "learning_rate": 0.0,
      "loss": 0.7895,
      "step": 2454
    },
    {
      "epoch": 0.2203374618560402,
      "grad_norm": 3.106318473815918,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 2455
    },
    {
      "epoch": 0.22042721234966792,
      "grad_norm": 6.591579914093018,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 2456
    },
    {
      "epoch": 0.22051696284329564,
      "grad_norm": 3.300989866256714,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 2457
    },
    {
      "epoch": 0.22060671333692336,
      "grad_norm": 20.55245018005371,
      "learning_rate": 0.0,
      "loss": 0.8138,
      "step": 2458
    },
    {
      "epoch": 0.22069646383055108,
      "grad_norm": 4.004583835601807,
      "learning_rate": 0.0,
      "loss": 0.7657,
      "step": 2459
    },
    {
      "epoch": 0.22078621432417878,
      "grad_norm": 2.5851891040802,
      "learning_rate": 0.0,
      "loss": 0.7749,
      "step": 2460
    },
    {
      "epoch": 0.2208759648178065,
      "grad_norm": 8.947073936462402,
      "learning_rate": 0.0,
      "loss": 0.8052,
      "step": 2461
    },
    {
      "epoch": 0.22096571531143422,
      "grad_norm": 2.886998414993286,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 2462
    },
    {
      "epoch": 0.22105546580506194,
      "grad_norm": 2.633591413497925,
      "learning_rate": 0.0,
      "loss": 0.7999,
      "step": 2463
    },
    {
      "epoch": 0.22114521629868963,
      "grad_norm": 2.1769680976867676,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 2464
    },
    {
      "epoch": 0.22123496679231736,
      "grad_norm": 2.5424931049346924,
      "learning_rate": 0.0,
      "loss": 0.7964,
      "step": 2465
    },
    {
      "epoch": 0.22132471728594508,
      "grad_norm": 2.2740774154663086,
      "learning_rate": 0.0,
      "loss": 0.7689,
      "step": 2466
    },
    {
      "epoch": 0.2214144677795728,
      "grad_norm": 4.820224285125732,
      "learning_rate": 0.0,
      "loss": 0.7831,
      "step": 2467
    },
    {
      "epoch": 0.2215042182732005,
      "grad_norm": 2.655632495880127,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 2468
    },
    {
      "epoch": 0.22159396876682821,
      "grad_norm": 2.277208089828491,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 2469
    },
    {
      "epoch": 0.22168371926045594,
      "grad_norm": 2.8266525268554688,
      "learning_rate": 0.0,
      "loss": 0.8201,
      "step": 2470
    },
    {
      "epoch": 0.22177346975408366,
      "grad_norm": 2.978227376937866,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 2471
    },
    {
      "epoch": 0.22186322024771135,
      "grad_norm": 2.9209184646606445,
      "learning_rate": 0.0,
      "loss": 0.7658,
      "step": 2472
    },
    {
      "epoch": 0.22195297074133907,
      "grad_norm": 3.104570150375366,
      "learning_rate": 0.0,
      "loss": 0.7506,
      "step": 2473
    },
    {
      "epoch": 0.2220427212349668,
      "grad_norm": 3.4836249351501465,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 2474
    },
    {
      "epoch": 0.22213247172859452,
      "grad_norm": 3.526700019836426,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 2475
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 3.366877555847168,
      "learning_rate": 0.0,
      "loss": 0.7729,
      "step": 2476
    },
    {
      "epoch": 0.22231197271584993,
      "grad_norm": 1.9973125457763672,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 2477
    },
    {
      "epoch": 0.22240172320947765,
      "grad_norm": 2.236971616744995,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 2478
    },
    {
      "epoch": 0.22249147370310537,
      "grad_norm": 2.533696174621582,
      "learning_rate": 0.0,
      "loss": 0.7814,
      "step": 2479
    },
    {
      "epoch": 0.22258122419673307,
      "grad_norm": 2.212574005126953,
      "learning_rate": 0.0,
      "loss": 0.7683,
      "step": 2480
    },
    {
      "epoch": 0.2226709746903608,
      "grad_norm": 4.2778239250183105,
      "learning_rate": 0.0,
      "loss": 0.7533,
      "step": 2481
    },
    {
      "epoch": 0.2227607251839885,
      "grad_norm": 3.226684331893921,
      "learning_rate": 0.0,
      "loss": 0.8351,
      "step": 2482
    },
    {
      "epoch": 0.22285047567761623,
      "grad_norm": 3.8231704235076904,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 2483
    },
    {
      "epoch": 0.22294022617124395,
      "grad_norm": 2.1872172355651855,
      "learning_rate": 0.0,
      "loss": 0.7785,
      "step": 2484
    },
    {
      "epoch": 0.22302997666487165,
      "grad_norm": 3.1376028060913086,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 2485
    },
    {
      "epoch": 0.22311972715849937,
      "grad_norm": 2.594778537750244,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 2486
    },
    {
      "epoch": 0.2232094776521271,
      "grad_norm": 2.627040386199951,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 2487
    },
    {
      "epoch": 0.2232992281457548,
      "grad_norm": 2.5430352687835693,
      "learning_rate": 0.0,
      "loss": 0.7488,
      "step": 2488
    },
    {
      "epoch": 0.2233889786393825,
      "grad_norm": 2.804898977279663,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 2489
    },
    {
      "epoch": 0.22347872913301023,
      "grad_norm": 2.0076613426208496,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 2490
    },
    {
      "epoch": 0.22356847962663795,
      "grad_norm": 2.1854634284973145,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 2491
    },
    {
      "epoch": 0.22365823012026567,
      "grad_norm": 3.882246255874634,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 2492
    },
    {
      "epoch": 0.22374798061389337,
      "grad_norm": 3.6460659503936768,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 2493
    },
    {
      "epoch": 0.2238377311075211,
      "grad_norm": 2.25848388671875,
      "learning_rate": 0.0,
      "loss": 0.7924,
      "step": 2494
    },
    {
      "epoch": 0.2239274816011488,
      "grad_norm": 2.411686420440674,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 2495
    },
    {
      "epoch": 0.22401723209477653,
      "grad_norm": 2.2503349781036377,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 2496
    },
    {
      "epoch": 0.22410698258840422,
      "grad_norm": 3.322716474533081,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 2497
    },
    {
      "epoch": 0.22419673308203195,
      "grad_norm": 3.2651820182800293,
      "learning_rate": 0.0,
      "loss": 0.7881,
      "step": 2498
    },
    {
      "epoch": 0.22428648357565967,
      "grad_norm": 2.943805694580078,
      "learning_rate": 0.0,
      "loss": 0.7759,
      "step": 2499
    },
    {
      "epoch": 0.2243762340692874,
      "grad_norm": 3.0842247009277344,
      "learning_rate": 0.0,
      "loss": 0.7795,
      "step": 2500
    },
    {
      "epoch": 0.22446598456291508,
      "grad_norm": 2.1293158531188965,
      "learning_rate": 0.0,
      "loss": 0.8069,
      "step": 2501
    },
    {
      "epoch": 0.2245557350565428,
      "grad_norm": 3.199291944503784,
      "learning_rate": 0.0,
      "loss": 0.7444,
      "step": 2502
    },
    {
      "epoch": 0.22464548555017053,
      "grad_norm": 2.627279758453369,
      "learning_rate": 0.0,
      "loss": 0.807,
      "step": 2503
    },
    {
      "epoch": 0.22473523604379825,
      "grad_norm": 4.174624443054199,
      "learning_rate": 0.0,
      "loss": 0.8119,
      "step": 2504
    },
    {
      "epoch": 0.22482498653742597,
      "grad_norm": 7.982093811035156,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 2505
    },
    {
      "epoch": 0.22491473703105366,
      "grad_norm": 2.571608543395996,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 2506
    },
    {
      "epoch": 0.22500448752468138,
      "grad_norm": 2.7088735103607178,
      "learning_rate": 0.0,
      "loss": 0.7624,
      "step": 2507
    },
    {
      "epoch": 0.2250942380183091,
      "grad_norm": 2.0395357608795166,
      "learning_rate": 0.0,
      "loss": 0.8293,
      "step": 2508
    },
    {
      "epoch": 0.22518398851193683,
      "grad_norm": 3.7475974559783936,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 2509
    },
    {
      "epoch": 0.22527373900556452,
      "grad_norm": 3.5164031982421875,
      "learning_rate": 0.0,
      "loss": 0.8058,
      "step": 2510
    },
    {
      "epoch": 0.22536348949919224,
      "grad_norm": 2.5949971675872803,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 2511
    },
    {
      "epoch": 0.22545323999281996,
      "grad_norm": 3.5724904537200928,
      "learning_rate": 0.0,
      "loss": 0.7657,
      "step": 2512
    },
    {
      "epoch": 0.2255429904864477,
      "grad_norm": 2.2425856590270996,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 2513
    },
    {
      "epoch": 0.22563274098007538,
      "grad_norm": 1.9077208042144775,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 2514
    },
    {
      "epoch": 0.2257224914737031,
      "grad_norm": 5.101880073547363,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 2515
    },
    {
      "epoch": 0.22581224196733082,
      "grad_norm": 3.559190034866333,
      "learning_rate": 0.0,
      "loss": 0.829,
      "step": 2516
    },
    {
      "epoch": 0.22590199246095855,
      "grad_norm": 3.1963179111480713,
      "learning_rate": 0.0,
      "loss": 0.7862,
      "step": 2517
    },
    {
      "epoch": 0.22599174295458624,
      "grad_norm": 2.43125057220459,
      "learning_rate": 0.0,
      "loss": 0.8138,
      "step": 2518
    },
    {
      "epoch": 0.22608149344821396,
      "grad_norm": 2.8705315589904785,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 2519
    },
    {
      "epoch": 0.22617124394184168,
      "grad_norm": 4.718075275421143,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 2520
    },
    {
      "epoch": 0.2262609944354694,
      "grad_norm": 2.6880910396575928,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 2521
    },
    {
      "epoch": 0.2263507449290971,
      "grad_norm": 3.1191956996917725,
      "learning_rate": 0.0,
      "loss": 0.8202,
      "step": 2522
    },
    {
      "epoch": 0.22644049542272482,
      "grad_norm": 2.451798439025879,
      "learning_rate": 0.0,
      "loss": 0.8254,
      "step": 2523
    },
    {
      "epoch": 0.22653024591635254,
      "grad_norm": 3.11800217628479,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 2524
    },
    {
      "epoch": 0.22661999640998026,
      "grad_norm": 3.4800875186920166,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 2525
    },
    {
      "epoch": 0.22670974690360796,
      "grad_norm": 2.947606325149536,
      "learning_rate": 0.0,
      "loss": 0.7519,
      "step": 2526
    },
    {
      "epoch": 0.22679949739723568,
      "grad_norm": 4.142229080200195,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 2527
    },
    {
      "epoch": 0.2268892478908634,
      "grad_norm": 7.437318325042725,
      "learning_rate": 0.0,
      "loss": 0.8171,
      "step": 2528
    },
    {
      "epoch": 0.22697899838449112,
      "grad_norm": 5.28177547454834,
      "learning_rate": 0.0,
      "loss": 0.7815,
      "step": 2529
    },
    {
      "epoch": 0.22706874887811884,
      "grad_norm": 2.706165075302124,
      "learning_rate": 0.0,
      "loss": 0.8095,
      "step": 2530
    },
    {
      "epoch": 0.22715849937174654,
      "grad_norm": 3.375112533569336,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 2531
    },
    {
      "epoch": 0.22724824986537426,
      "grad_norm": 2.5171868801116943,
      "learning_rate": 0.0,
      "loss": 0.8188,
      "step": 2532
    },
    {
      "epoch": 0.22733800035900198,
      "grad_norm": 3.6131491661071777,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 2533
    },
    {
      "epoch": 0.2274277508526297,
      "grad_norm": 3.8755509853363037,
      "learning_rate": 0.0,
      "loss": 0.7718,
      "step": 2534
    },
    {
      "epoch": 0.2275175013462574,
      "grad_norm": 3.2243525981903076,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 2535
    },
    {
      "epoch": 0.22760725183988512,
      "grad_norm": 3.5483102798461914,
      "learning_rate": 0.0,
      "loss": 0.8045,
      "step": 2536
    },
    {
      "epoch": 0.22769700233351284,
      "grad_norm": 2.353606700897217,
      "learning_rate": 0.0,
      "loss": 0.7718,
      "step": 2537
    },
    {
      "epoch": 0.22778675282714056,
      "grad_norm": 2.413778781890869,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 2538
    },
    {
      "epoch": 0.22787650332076825,
      "grad_norm": 18.486303329467773,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 2539
    },
    {
      "epoch": 0.22796625381439597,
      "grad_norm": 2.8163516521453857,
      "learning_rate": 0.0,
      "loss": 0.8093,
      "step": 2540
    },
    {
      "epoch": 0.2280560043080237,
      "grad_norm": 2.6742401123046875,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 2541
    },
    {
      "epoch": 0.22814575480165142,
      "grad_norm": 3.5363481044769287,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 2542
    },
    {
      "epoch": 0.2282355052952791,
      "grad_norm": 3.9724032878875732,
      "learning_rate": 0.0,
      "loss": 0.8072,
      "step": 2543
    },
    {
      "epoch": 0.22832525578890683,
      "grad_norm": 7.107311248779297,
      "learning_rate": 0.0,
      "loss": 0.7021,
      "step": 2544
    },
    {
      "epoch": 0.22841500628253456,
      "grad_norm": 13.002609252929688,
      "learning_rate": 0.0,
      "loss": 0.7705,
      "step": 2545
    },
    {
      "epoch": 0.22850475677616228,
      "grad_norm": 2.851902961730957,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 2546
    },
    {
      "epoch": 0.22859450726978997,
      "grad_norm": 2.8904526233673096,
      "learning_rate": 0.0,
      "loss": 0.8392,
      "step": 2547
    },
    {
      "epoch": 0.2286842577634177,
      "grad_norm": 2.626406192779541,
      "learning_rate": 0.0,
      "loss": 0.826,
      "step": 2548
    },
    {
      "epoch": 0.2287740082570454,
      "grad_norm": 3.1606860160827637,
      "learning_rate": 0.0,
      "loss": 0.7984,
      "step": 2549
    },
    {
      "epoch": 0.22886375875067314,
      "grad_norm": 2.6607043743133545,
      "learning_rate": 0.0,
      "loss": 0.7759,
      "step": 2550
    },
    {
      "epoch": 0.22895350924430086,
      "grad_norm": 3.648134231567383,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 2551
    },
    {
      "epoch": 0.22904325973792855,
      "grad_norm": 3.0335333347320557,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 2552
    },
    {
      "epoch": 0.22913301023155627,
      "grad_norm": 2.7948009967803955,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 2553
    },
    {
      "epoch": 0.229222760725184,
      "grad_norm": 3.7545034885406494,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 2554
    },
    {
      "epoch": 0.22931251121881172,
      "grad_norm": 3.183861494064331,
      "learning_rate": 0.0,
      "loss": 0.7752,
      "step": 2555
    },
    {
      "epoch": 0.2294022617124394,
      "grad_norm": 2.8218607902526855,
      "learning_rate": 0.0,
      "loss": 0.7685,
      "step": 2556
    },
    {
      "epoch": 0.22949201220606713,
      "grad_norm": 3.582902431488037,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 2557
    },
    {
      "epoch": 0.22958176269969485,
      "grad_norm": 3.2467527389526367,
      "learning_rate": 0.0,
      "loss": 0.7969,
      "step": 2558
    },
    {
      "epoch": 0.22967151319332257,
      "grad_norm": 3.979952335357666,
      "learning_rate": 0.0,
      "loss": 0.815,
      "step": 2559
    },
    {
      "epoch": 0.22976126368695027,
      "grad_norm": 2.339305877685547,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 2560
    },
    {
      "epoch": 0.229851014180578,
      "grad_norm": 2.199140787124634,
      "learning_rate": 0.0,
      "loss": 0.7669,
      "step": 2561
    },
    {
      "epoch": 0.2299407646742057,
      "grad_norm": 2.6780405044555664,
      "learning_rate": 0.0,
      "loss": 0.7714,
      "step": 2562
    },
    {
      "epoch": 0.23003051516783343,
      "grad_norm": 2.7131547927856445,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 2563
    },
    {
      "epoch": 0.23012026566146113,
      "grad_norm": 3.061944007873535,
      "learning_rate": 0.0,
      "loss": 0.8454,
      "step": 2564
    },
    {
      "epoch": 0.23021001615508885,
      "grad_norm": 2.98865008354187,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 2565
    },
    {
      "epoch": 0.23029976664871657,
      "grad_norm": 4.6824846267700195,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 2566
    },
    {
      "epoch": 0.2303895171423443,
      "grad_norm": 2.882554054260254,
      "learning_rate": 0.0,
      "loss": 0.7932,
      "step": 2567
    },
    {
      "epoch": 0.23047926763597198,
      "grad_norm": 3.023163080215454,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 2568
    },
    {
      "epoch": 0.2305690181295997,
      "grad_norm": 3.3330326080322266,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 2569
    },
    {
      "epoch": 0.23065876862322743,
      "grad_norm": 2.856457233428955,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 2570
    },
    {
      "epoch": 0.23074851911685515,
      "grad_norm": 3.247657537460327,
      "learning_rate": 0.0,
      "loss": 0.8078,
      "step": 2571
    },
    {
      "epoch": 0.23083826961048287,
      "grad_norm": 3.789109230041504,
      "learning_rate": 0.0,
      "loss": 0.7627,
      "step": 2572
    },
    {
      "epoch": 0.23092802010411057,
      "grad_norm": 5.962863922119141,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 2573
    },
    {
      "epoch": 0.2310177705977383,
      "grad_norm": 3.272930145263672,
      "learning_rate": 0.0,
      "loss": 0.7776,
      "step": 2574
    },
    {
      "epoch": 0.231107521091366,
      "grad_norm": 5.4781928062438965,
      "learning_rate": 0.0,
      "loss": 0.7984,
      "step": 2575
    },
    {
      "epoch": 0.23119727158499373,
      "grad_norm": 2.3591554164886475,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 2576
    },
    {
      "epoch": 0.23128702207862142,
      "grad_norm": 3.3828165531158447,
      "learning_rate": 0.0,
      "loss": 0.8323,
      "step": 2577
    },
    {
      "epoch": 0.23137677257224915,
      "grad_norm": 2.7113914489746094,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 2578
    },
    {
      "epoch": 0.23146652306587687,
      "grad_norm": 4.192108631134033,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 2579
    },
    {
      "epoch": 0.2315562735595046,
      "grad_norm": 2.558758497238159,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 2580
    },
    {
      "epoch": 0.23164602405313228,
      "grad_norm": 2.8485493659973145,
      "learning_rate": 0.0,
      "loss": 0.798,
      "step": 2581
    },
    {
      "epoch": 0.23173577454676,
      "grad_norm": 2.792419672012329,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 2582
    },
    {
      "epoch": 0.23182552504038773,
      "grad_norm": 5.118008613586426,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 2583
    },
    {
      "epoch": 0.23191527553401545,
      "grad_norm": 2.9383885860443115,
      "learning_rate": 0.0,
      "loss": 0.8052,
      "step": 2584
    },
    {
      "epoch": 0.23200502602764314,
      "grad_norm": 3.1523642539978027,
      "learning_rate": 0.0,
      "loss": 0.8214,
      "step": 2585
    },
    {
      "epoch": 0.23209477652127086,
      "grad_norm": 3.002697229385376,
      "learning_rate": 0.0,
      "loss": 0.8184,
      "step": 2586
    },
    {
      "epoch": 0.23218452701489858,
      "grad_norm": 2.406299114227295,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 2587
    },
    {
      "epoch": 0.2322742775085263,
      "grad_norm": 2.7915642261505127,
      "learning_rate": 0.0,
      "loss": 0.7912,
      "step": 2588
    },
    {
      "epoch": 0.232364028002154,
      "grad_norm": 3.7669169902801514,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 2589
    },
    {
      "epoch": 0.23245377849578172,
      "grad_norm": 3.1564781665802,
      "learning_rate": 0.0,
      "loss": 0.7939,
      "step": 2590
    },
    {
      "epoch": 0.23254352898940944,
      "grad_norm": 2.9602999687194824,
      "learning_rate": 0.0,
      "loss": 0.7961,
      "step": 2591
    },
    {
      "epoch": 0.23263327948303716,
      "grad_norm": 2.6939191818237305,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 2592
    },
    {
      "epoch": 0.23272302997666486,
      "grad_norm": 2.2462587356567383,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 2593
    },
    {
      "epoch": 0.23281278047029258,
      "grad_norm": 5.645567893981934,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 2594
    },
    {
      "epoch": 0.2329025309639203,
      "grad_norm": 14.17728042602539,
      "learning_rate": 0.0,
      "loss": 0.8311,
      "step": 2595
    },
    {
      "epoch": 0.23299228145754802,
      "grad_norm": 2.539900779724121,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 2596
    },
    {
      "epoch": 0.23308203195117574,
      "grad_norm": 2.6247365474700928,
      "learning_rate": 0.0,
      "loss": 0.7594,
      "step": 2597
    },
    {
      "epoch": 0.23317178244480344,
      "grad_norm": 2.703984022140503,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 2598
    },
    {
      "epoch": 0.23326153293843116,
      "grad_norm": 2.8859317302703857,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 2599
    },
    {
      "epoch": 0.23335128343205888,
      "grad_norm": 3.4619240760803223,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 2600
    },
    {
      "epoch": 0.2334410339256866,
      "grad_norm": 3.008312225341797,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 2601
    },
    {
      "epoch": 0.2335307844193143,
      "grad_norm": 3.3769803047180176,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 2602
    },
    {
      "epoch": 0.23362053491294202,
      "grad_norm": 5.021580219268799,
      "learning_rate": 0.0,
      "loss": 0.8238,
      "step": 2603
    },
    {
      "epoch": 0.23371028540656974,
      "grad_norm": 6.616533279418945,
      "learning_rate": 0.0,
      "loss": 0.7653,
      "step": 2604
    },
    {
      "epoch": 0.23380003590019746,
      "grad_norm": 3.5754945278167725,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 2605
    },
    {
      "epoch": 0.23388978639382516,
      "grad_norm": 2.650332450866699,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 2606
    },
    {
      "epoch": 0.23397953688745288,
      "grad_norm": 5.0248870849609375,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 2607
    },
    {
      "epoch": 0.2340692873810806,
      "grad_norm": 10.943927764892578,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 2608
    },
    {
      "epoch": 0.23415903787470832,
      "grad_norm": 3.849748373031616,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 2609
    },
    {
      "epoch": 0.23424878836833601,
      "grad_norm": 4.221120834350586,
      "learning_rate": 0.0,
      "loss": 0.7411,
      "step": 2610
    },
    {
      "epoch": 0.23433853886196374,
      "grad_norm": 3.0881149768829346,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 2611
    },
    {
      "epoch": 0.23442828935559146,
      "grad_norm": 5.180844783782959,
      "learning_rate": 0.0,
      "loss": 0.8247,
      "step": 2612
    },
    {
      "epoch": 0.23451803984921918,
      "grad_norm": 4.177338600158691,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 2613
    },
    {
      "epoch": 0.23460779034284687,
      "grad_norm": 3.5870113372802734,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 2614
    },
    {
      "epoch": 0.2346975408364746,
      "grad_norm": 26.884706497192383,
      "learning_rate": 0.0,
      "loss": 0.8056,
      "step": 2615
    },
    {
      "epoch": 0.23478729133010232,
      "grad_norm": 4.422726631164551,
      "learning_rate": 0.0,
      "loss": 0.7811,
      "step": 2616
    },
    {
      "epoch": 0.23487704182373004,
      "grad_norm": 3.2072553634643555,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 2617
    },
    {
      "epoch": 0.23496679231735776,
      "grad_norm": 2.336865186691284,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 2618
    },
    {
      "epoch": 0.23505654281098545,
      "grad_norm": 3.247734785079956,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 2619
    },
    {
      "epoch": 0.23514629330461317,
      "grad_norm": 4.1466193199157715,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 2620
    },
    {
      "epoch": 0.2352360437982409,
      "grad_norm": 3.903317451477051,
      "learning_rate": 0.0,
      "loss": 0.7993,
      "step": 2621
    },
    {
      "epoch": 0.23532579429186862,
      "grad_norm": 2.718787908554077,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 2622
    },
    {
      "epoch": 0.2354155447854963,
      "grad_norm": 3.4793660640716553,
      "learning_rate": 0.0,
      "loss": 0.7768,
      "step": 2623
    },
    {
      "epoch": 0.23550529527912403,
      "grad_norm": 5.242769241333008,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 2624
    },
    {
      "epoch": 0.23559504577275175,
      "grad_norm": 3.779453754425049,
      "learning_rate": 0.0,
      "loss": 0.7877,
      "step": 2625
    },
    {
      "epoch": 0.23568479626637948,
      "grad_norm": 3.3049514293670654,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 2626
    },
    {
      "epoch": 0.23577454676000717,
      "grad_norm": 3.317781448364258,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 2627
    },
    {
      "epoch": 0.2358642972536349,
      "grad_norm": 4.516298294067383,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "step": 2628
    },
    {
      "epoch": 0.2359540477472626,
      "grad_norm": 2.676950693130493,
      "learning_rate": 0.0,
      "loss": 0.7246,
      "step": 2629
    },
    {
      "epoch": 0.23604379824089033,
      "grad_norm": 3.2756595611572266,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 2630
    },
    {
      "epoch": 0.23613354873451803,
      "grad_norm": 2.7866525650024414,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 2631
    },
    {
      "epoch": 0.23622329922814575,
      "grad_norm": 3.448927879333496,
      "learning_rate": 0.0,
      "loss": 0.8027,
      "step": 2632
    },
    {
      "epoch": 0.23631304972177347,
      "grad_norm": 3.8497159481048584,
      "learning_rate": 0.0,
      "loss": 0.7458,
      "step": 2633
    },
    {
      "epoch": 0.2364028002154012,
      "grad_norm": 2.3089635372161865,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 2634
    },
    {
      "epoch": 0.2364925507090289,
      "grad_norm": 2.3689687252044678,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 2635
    },
    {
      "epoch": 0.2365823012026566,
      "grad_norm": 3.472668409347534,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 2636
    },
    {
      "epoch": 0.23667205169628433,
      "grad_norm": 2.9378559589385986,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 2637
    },
    {
      "epoch": 0.23676180218991205,
      "grad_norm": 3.9870522022247314,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 2638
    },
    {
      "epoch": 0.23685155268353977,
      "grad_norm": 2.769562244415283,
      "learning_rate": 0.0,
      "loss": 0.8123,
      "step": 2639
    },
    {
      "epoch": 0.23694130317716747,
      "grad_norm": 2.493232011795044,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 2640
    },
    {
      "epoch": 0.2370310536707952,
      "grad_norm": 3.56325101852417,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 2641
    },
    {
      "epoch": 0.2371208041644229,
      "grad_norm": 3.2984721660614014,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 2642
    },
    {
      "epoch": 0.23721055465805063,
      "grad_norm": 3.417233467102051,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 2643
    },
    {
      "epoch": 0.23730030515167833,
      "grad_norm": 3.0609073638916016,
      "learning_rate": 0.0,
      "loss": 0.8194,
      "step": 2644
    },
    {
      "epoch": 0.23739005564530605,
      "grad_norm": 2.333522081375122,
      "learning_rate": 0.0,
      "loss": 0.7602,
      "step": 2645
    },
    {
      "epoch": 0.23747980613893377,
      "grad_norm": 2.9573373794555664,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 2646
    },
    {
      "epoch": 0.2375695566325615,
      "grad_norm": 3.96238112449646,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 2647
    },
    {
      "epoch": 0.23765930712618918,
      "grad_norm": 2.236558675765991,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 2648
    },
    {
      "epoch": 0.2377490576198169,
      "grad_norm": 2.718076229095459,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 2649
    },
    {
      "epoch": 0.23783880811344463,
      "grad_norm": 3.5413060188293457,
      "learning_rate": 0.0,
      "loss": 0.8359,
      "step": 2650
    },
    {
      "epoch": 0.23792855860707235,
      "grad_norm": 2.913134813308716,
      "learning_rate": 0.0,
      "loss": 0.7927,
      "step": 2651
    },
    {
      "epoch": 0.23801830910070004,
      "grad_norm": 2.852630138397217,
      "learning_rate": 0.0,
      "loss": 0.786,
      "step": 2652
    },
    {
      "epoch": 0.23810805959432776,
      "grad_norm": 2.688832998275757,
      "learning_rate": 0.0,
      "loss": 0.7968,
      "step": 2653
    },
    {
      "epoch": 0.23819781008795549,
      "grad_norm": 3.00339937210083,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 2654
    },
    {
      "epoch": 0.2382875605815832,
      "grad_norm": 2.230088710784912,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 2655
    },
    {
      "epoch": 0.2383773110752109,
      "grad_norm": 2.8593015670776367,
      "learning_rate": 0.0,
      "loss": 0.8044,
      "step": 2656
    },
    {
      "epoch": 0.23846706156883862,
      "grad_norm": 2.7447292804718018,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 2657
    },
    {
      "epoch": 0.23855681206246634,
      "grad_norm": 2.196688652038574,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 2658
    },
    {
      "epoch": 0.23864656255609407,
      "grad_norm": 3.410684585571289,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 2659
    },
    {
      "epoch": 0.23873631304972176,
      "grad_norm": 7.282494068145752,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 2660
    },
    {
      "epoch": 0.23882606354334948,
      "grad_norm": 2.768435001373291,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 2661
    },
    {
      "epoch": 0.2389158140369772,
      "grad_norm": 2.590304374694824,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 2662
    },
    {
      "epoch": 0.23900556453060492,
      "grad_norm": 3.696108341217041,
      "learning_rate": 0.0,
      "loss": 0.7659,
      "step": 2663
    },
    {
      "epoch": 0.23909531502423265,
      "grad_norm": 2.6330184936523438,
      "learning_rate": 0.0,
      "loss": 0.7991,
      "step": 2664
    },
    {
      "epoch": 0.23918506551786034,
      "grad_norm": 4.158567905426025,
      "learning_rate": 0.0,
      "loss": 0.7632,
      "step": 2665
    },
    {
      "epoch": 0.23927481601148806,
      "grad_norm": 3.2465522289276123,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 2666
    },
    {
      "epoch": 0.23936456650511578,
      "grad_norm": 2.7827720642089844,
      "learning_rate": 0.0,
      "loss": 0.6846,
      "step": 2667
    },
    {
      "epoch": 0.2394543169987435,
      "grad_norm": 2.63869571685791,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 2668
    },
    {
      "epoch": 0.2395440674923712,
      "grad_norm": 2.754249095916748,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 2669
    },
    {
      "epoch": 0.23963381798599892,
      "grad_norm": 2.2574946880340576,
      "learning_rate": 0.0,
      "loss": 0.7773,
      "step": 2670
    },
    {
      "epoch": 0.23972356847962664,
      "grad_norm": 3.781512975692749,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 2671
    },
    {
      "epoch": 0.23981331897325436,
      "grad_norm": 2.3682138919830322,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 2672
    },
    {
      "epoch": 0.23990306946688206,
      "grad_norm": 2.6847987174987793,
      "learning_rate": 0.0,
      "loss": 0.7729,
      "step": 2673
    },
    {
      "epoch": 0.23999281996050978,
      "grad_norm": 2.6801559925079346,
      "learning_rate": 0.0,
      "loss": 0.7939,
      "step": 2674
    },
    {
      "epoch": 0.2400825704541375,
      "grad_norm": 3.868884325027466,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 2675
    },
    {
      "epoch": 0.24017232094776522,
      "grad_norm": 5.096828937530518,
      "learning_rate": 0.0,
      "loss": 0.8068,
      "step": 2676
    },
    {
      "epoch": 0.24026207144139292,
      "grad_norm": 2.9540538787841797,
      "learning_rate": 0.0,
      "loss": 0.7802,
      "step": 2677
    },
    {
      "epoch": 0.24035182193502064,
      "grad_norm": 3.391916036605835,
      "learning_rate": 0.0,
      "loss": 0.7822,
      "step": 2678
    },
    {
      "epoch": 0.24044157242864836,
      "grad_norm": 2.184026002883911,
      "learning_rate": 0.0,
      "loss": 0.7904,
      "step": 2679
    },
    {
      "epoch": 0.24053132292227608,
      "grad_norm": 2.768725633621216,
      "learning_rate": 0.0,
      "loss": 0.7907,
      "step": 2680
    },
    {
      "epoch": 0.24062107341590377,
      "grad_norm": 6.089725494384766,
      "learning_rate": 0.0,
      "loss": 0.7822,
      "step": 2681
    },
    {
      "epoch": 0.2407108239095315,
      "grad_norm": 4.4809980392456055,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 2682
    },
    {
      "epoch": 0.24080057440315922,
      "grad_norm": 3.2535417079925537,
      "learning_rate": 0.0,
      "loss": 0.7906,
      "step": 2683
    },
    {
      "epoch": 0.24089032489678694,
      "grad_norm": 2.808511257171631,
      "learning_rate": 0.0,
      "loss": 0.7779,
      "step": 2684
    },
    {
      "epoch": 0.24098007539041466,
      "grad_norm": 3.1495656967163086,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 2685
    },
    {
      "epoch": 0.24106982588404235,
      "grad_norm": 2.8280463218688965,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 2686
    },
    {
      "epoch": 0.24115957637767008,
      "grad_norm": 5.073254585266113,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 2687
    },
    {
      "epoch": 0.2412493268712978,
      "grad_norm": 2.3983232975006104,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 2688
    },
    {
      "epoch": 0.24133907736492552,
      "grad_norm": 2.3271396160125732,
      "learning_rate": 0.0,
      "loss": 0.7684,
      "step": 2689
    },
    {
      "epoch": 0.2414288278585532,
      "grad_norm": 2.8648529052734375,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 2690
    },
    {
      "epoch": 0.24151857835218093,
      "grad_norm": 3.0402674674987793,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 2691
    },
    {
      "epoch": 0.24160832884580866,
      "grad_norm": 2.41239070892334,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 2692
    },
    {
      "epoch": 0.24169807933943638,
      "grad_norm": 2.532374620437622,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 2693
    },
    {
      "epoch": 0.24178782983306407,
      "grad_norm": 4.542538166046143,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 2694
    },
    {
      "epoch": 0.2418775803266918,
      "grad_norm": 2.562730073928833,
      "learning_rate": 0.0,
      "loss": 0.8128,
      "step": 2695
    },
    {
      "epoch": 0.24196733082031952,
      "grad_norm": 4.168868541717529,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 2696
    },
    {
      "epoch": 0.24205708131394724,
      "grad_norm": 2.6728622913360596,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 2697
    },
    {
      "epoch": 0.24214683180757493,
      "grad_norm": 4.469545364379883,
      "learning_rate": 0.0,
      "loss": 0.7808,
      "step": 2698
    },
    {
      "epoch": 0.24223658230120265,
      "grad_norm": 3.747481346130371,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 2699
    },
    {
      "epoch": 0.24232633279483037,
      "grad_norm": 4.384591102600098,
      "learning_rate": 0.0,
      "loss": 0.7842,
      "step": 2700
    },
    {
      "epoch": 0.2424160832884581,
      "grad_norm": 2.260054349899292,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 2701
    },
    {
      "epoch": 0.2425058337820858,
      "grad_norm": 3.9848361015319824,
      "learning_rate": 0.0,
      "loss": 0.7831,
      "step": 2702
    },
    {
      "epoch": 0.2425955842757135,
      "grad_norm": 54.09084701538086,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 2703
    },
    {
      "epoch": 0.24268533476934123,
      "grad_norm": 2.7946648597717285,
      "learning_rate": 0.0,
      "loss": 0.8267,
      "step": 2704
    },
    {
      "epoch": 0.24277508526296895,
      "grad_norm": 4.14868688583374,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 2705
    },
    {
      "epoch": 0.24286483575659665,
      "grad_norm": 3.540985345840454,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 2706
    },
    {
      "epoch": 0.24295458625022437,
      "grad_norm": 2.8037447929382324,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 2707
    },
    {
      "epoch": 0.2430443367438521,
      "grad_norm": 4.2091383934021,
      "learning_rate": 0.0,
      "loss": 0.8006,
      "step": 2708
    },
    {
      "epoch": 0.2431340872374798,
      "grad_norm": 2.7019267082214355,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 2709
    },
    {
      "epoch": 0.24322383773110753,
      "grad_norm": 9.596160888671875,
      "learning_rate": 0.0,
      "loss": 0.8219,
      "step": 2710
    },
    {
      "epoch": 0.24331358822473523,
      "grad_norm": 2.36549711227417,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 2711
    },
    {
      "epoch": 0.24340333871836295,
      "grad_norm": 3.0540201663970947,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 2712
    },
    {
      "epoch": 0.24349308921199067,
      "grad_norm": 6.753158092498779,
      "learning_rate": 0.0,
      "loss": 0.7776,
      "step": 2713
    },
    {
      "epoch": 0.2435828397056184,
      "grad_norm": 5.0619940757751465,
      "learning_rate": 0.0,
      "loss": 0.7444,
      "step": 2714
    },
    {
      "epoch": 0.2436725901992461,
      "grad_norm": 2.9733896255493164,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 2715
    },
    {
      "epoch": 0.2437623406928738,
      "grad_norm": 3.1975669860839844,
      "learning_rate": 0.0,
      "loss": 0.8346,
      "step": 2716
    },
    {
      "epoch": 0.24385209118650153,
      "grad_norm": 2.6225407123565674,
      "learning_rate": 0.0,
      "loss": 0.7524,
      "step": 2717
    },
    {
      "epoch": 0.24394184168012925,
      "grad_norm": 50.85346603393555,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 2718
    },
    {
      "epoch": 0.24403159217375694,
      "grad_norm": 3.633218765258789,
      "learning_rate": 0.0,
      "loss": 0.7432,
      "step": 2719
    },
    {
      "epoch": 0.24412134266738467,
      "grad_norm": 7.046574592590332,
      "learning_rate": 0.0,
      "loss": 0.805,
      "step": 2720
    },
    {
      "epoch": 0.2442110931610124,
      "grad_norm": 4.029664993286133,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 2721
    },
    {
      "epoch": 0.2443008436546401,
      "grad_norm": 2.967362642288208,
      "learning_rate": 0.0,
      "loss": 0.8333,
      "step": 2722
    },
    {
      "epoch": 0.2443905941482678,
      "grad_norm": 2.666170358657837,
      "learning_rate": 0.0,
      "loss": 0.7837,
      "step": 2723
    },
    {
      "epoch": 0.24448034464189553,
      "grad_norm": 3.1004202365875244,
      "learning_rate": 0.0,
      "loss": 0.7872,
      "step": 2724
    },
    {
      "epoch": 0.24457009513552325,
      "grad_norm": 2.7209560871124268,
      "learning_rate": 0.0,
      "loss": 0.7363,
      "step": 2725
    },
    {
      "epoch": 0.24465984562915097,
      "grad_norm": 3.2950363159179688,
      "learning_rate": 0.0,
      "loss": 0.758,
      "step": 2726
    },
    {
      "epoch": 0.24474959612277866,
      "grad_norm": 2.615196704864502,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 2727
    },
    {
      "epoch": 0.24483934661640638,
      "grad_norm": 3.0308051109313965,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 2728
    },
    {
      "epoch": 0.2449290971100341,
      "grad_norm": 3.2444651126861572,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 2729
    },
    {
      "epoch": 0.24501884760366183,
      "grad_norm": 3.949897527694702,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 2730
    },
    {
      "epoch": 0.24510859809728955,
      "grad_norm": 3.621312141418457,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 2731
    },
    {
      "epoch": 0.24519834859091724,
      "grad_norm": 3.1679365634918213,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 2732
    },
    {
      "epoch": 0.24528809908454496,
      "grad_norm": 4.1228837966918945,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 2733
    },
    {
      "epoch": 0.24537784957817269,
      "grad_norm": 4.198512554168701,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 2734
    },
    {
      "epoch": 0.2454676000718004,
      "grad_norm": 5.181012153625488,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 2735
    },
    {
      "epoch": 0.2455573505654281,
      "grad_norm": 2.758977174758911,
      "learning_rate": 0.0,
      "loss": 0.774,
      "step": 2736
    },
    {
      "epoch": 0.24564710105905582,
      "grad_norm": 4.3806047439575195,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 2737
    },
    {
      "epoch": 0.24573685155268354,
      "grad_norm": 3.605814218521118,
      "learning_rate": 0.0,
      "loss": 0.8224,
      "step": 2738
    },
    {
      "epoch": 0.24582660204631127,
      "grad_norm": 3.4015824794769287,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 2739
    },
    {
      "epoch": 0.24591635253993896,
      "grad_norm": 5.872452259063721,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 2740
    },
    {
      "epoch": 0.24600610303356668,
      "grad_norm": 2.8339176177978516,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 2741
    },
    {
      "epoch": 0.2460958535271944,
      "grad_norm": 2.342910051345825,
      "learning_rate": 0.0,
      "loss": 0.7906,
      "step": 2742
    },
    {
      "epoch": 0.24618560402082212,
      "grad_norm": 2.8830668926239014,
      "learning_rate": 0.0,
      "loss": 0.7414,
      "step": 2743
    },
    {
      "epoch": 0.24627535451444982,
      "grad_norm": 2.9103829860687256,
      "learning_rate": 0.0,
      "loss": 0.779,
      "step": 2744
    },
    {
      "epoch": 0.24636510500807754,
      "grad_norm": 3.4515764713287354,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 2745
    },
    {
      "epoch": 0.24645485550170526,
      "grad_norm": 17.516456604003906,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 2746
    },
    {
      "epoch": 0.24654460599533298,
      "grad_norm": 6.930692672729492,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 2747
    },
    {
      "epoch": 0.24663435648896068,
      "grad_norm": 16.20273208618164,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 2748
    },
    {
      "epoch": 0.2467241069825884,
      "grad_norm": 7.291207790374756,
      "learning_rate": 0.0,
      "loss": 0.7801,
      "step": 2749
    },
    {
      "epoch": 0.24681385747621612,
      "grad_norm": 3.693009376525879,
      "learning_rate": 0.0,
      "loss": 0.8159,
      "step": 2750
    },
    {
      "epoch": 0.24690360796984384,
      "grad_norm": 2.9326844215393066,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 2751
    },
    {
      "epoch": 0.24699335846347156,
      "grad_norm": 3.3861403465270996,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 2752
    },
    {
      "epoch": 0.24708310895709926,
      "grad_norm": 3.6497724056243896,
      "learning_rate": 0.0,
      "loss": 0.7691,
      "step": 2753
    },
    {
      "epoch": 0.24717285945072698,
      "grad_norm": 4.795495986938477,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 2754
    },
    {
      "epoch": 0.2472626099443547,
      "grad_norm": 3.87129545211792,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 2755
    },
    {
      "epoch": 0.24735236043798242,
      "grad_norm": 3.8296961784362793,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 2756
    },
    {
      "epoch": 0.24744211093161012,
      "grad_norm": 3.9095029830932617,
      "learning_rate": 0.0,
      "loss": 0.7843,
      "step": 2757
    },
    {
      "epoch": 0.24753186142523784,
      "grad_norm": 3.717120885848999,
      "learning_rate": 0.0,
      "loss": 0.701,
      "step": 2758
    },
    {
      "epoch": 0.24762161191886556,
      "grad_norm": 15.445329666137695,
      "learning_rate": 0.0,
      "loss": 0.803,
      "step": 2759
    },
    {
      "epoch": 0.24771136241249328,
      "grad_norm": 2.8211019039154053,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 2760
    },
    {
      "epoch": 0.24780111290612097,
      "grad_norm": 4.759279251098633,
      "learning_rate": 0.0,
      "loss": 0.8009,
      "step": 2761
    },
    {
      "epoch": 0.2478908633997487,
      "grad_norm": 4.27324914932251,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 2762
    },
    {
      "epoch": 0.24798061389337642,
      "grad_norm": 2.839752197265625,
      "learning_rate": 0.0,
      "loss": 0.7564,
      "step": 2763
    },
    {
      "epoch": 0.24807036438700414,
      "grad_norm": 3.5981216430664062,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 2764
    },
    {
      "epoch": 0.24816011488063183,
      "grad_norm": 4.770797252655029,
      "learning_rate": 0.0,
      "loss": 0.7712,
      "step": 2765
    },
    {
      "epoch": 0.24824986537425955,
      "grad_norm": 5.851487159729004,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 2766
    },
    {
      "epoch": 0.24833961586788728,
      "grad_norm": 2.893533706665039,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 2767
    },
    {
      "epoch": 0.248429366361515,
      "grad_norm": 4.033637523651123,
      "learning_rate": 0.0,
      "loss": 0.7898,
      "step": 2768
    },
    {
      "epoch": 0.2485191168551427,
      "grad_norm": 3.524198293685913,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 2769
    },
    {
      "epoch": 0.2486088673487704,
      "grad_norm": 4.282919883728027,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 2770
    },
    {
      "epoch": 0.24869861784239813,
      "grad_norm": 2.856600046157837,
      "learning_rate": 0.0,
      "loss": 0.7797,
      "step": 2771
    },
    {
      "epoch": 0.24878836833602586,
      "grad_norm": 3.220116376876831,
      "learning_rate": 0.0,
      "loss": 0.832,
      "step": 2772
    },
    {
      "epoch": 0.24887811882965355,
      "grad_norm": 3.7673022747039795,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 2773
    },
    {
      "epoch": 0.24896786932328127,
      "grad_norm": 2.784898042678833,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 2774
    },
    {
      "epoch": 0.249057619816909,
      "grad_norm": 4.126042366027832,
      "learning_rate": 0.0,
      "loss": 0.7705,
      "step": 2775
    },
    {
      "epoch": 0.24914737031053671,
      "grad_norm": 5.956042289733887,
      "learning_rate": 0.0,
      "loss": 0.7686,
      "step": 2776
    },
    {
      "epoch": 0.24923712080416444,
      "grad_norm": 3.529531478881836,
      "learning_rate": 0.0,
      "loss": 0.743,
      "step": 2777
    },
    {
      "epoch": 0.24932687129779213,
      "grad_norm": 4.04985237121582,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 2778
    },
    {
      "epoch": 0.24941662179141985,
      "grad_norm": 2.522146224975586,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 2779
    },
    {
      "epoch": 0.24950637228504757,
      "grad_norm": 3.3043930530548096,
      "learning_rate": 0.0,
      "loss": 0.7818,
      "step": 2780
    },
    {
      "epoch": 0.2495961227786753,
      "grad_norm": 2.4486825466156006,
      "learning_rate": 0.0,
      "loss": 0.8134,
      "step": 2781
    },
    {
      "epoch": 0.249685873272303,
      "grad_norm": 3.218104362487793,
      "learning_rate": 0.0,
      "loss": 0.7991,
      "step": 2782
    },
    {
      "epoch": 0.2497756237659307,
      "grad_norm": 2.752164602279663,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 2783
    },
    {
      "epoch": 0.24986537425955843,
      "grad_norm": 2.9848742485046387,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 2784
    },
    {
      "epoch": 0.24995512475318615,
      "grad_norm": 3.7878029346466064,
      "learning_rate": 0.0,
      "loss": 0.804,
      "step": 2785
    },
    {
      "epoch": 0.25004487524681385,
      "grad_norm": 3.3684942722320557,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 2786
    },
    {
      "epoch": 0.2501346257404416,
      "grad_norm": 2.3722822666168213,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 2787
    },
    {
      "epoch": 0.2502243762340693,
      "grad_norm": 4.52009391784668,
      "learning_rate": 0.0,
      "loss": 0.7708,
      "step": 2788
    },
    {
      "epoch": 0.250314126727697,
      "grad_norm": 3.1273906230926514,
      "learning_rate": 0.0,
      "loss": 0.7589,
      "step": 2789
    },
    {
      "epoch": 0.25040387722132473,
      "grad_norm": 4.887368679046631,
      "learning_rate": 0.0,
      "loss": 0.768,
      "step": 2790
    },
    {
      "epoch": 0.2504936277149524,
      "grad_norm": 2.8263933658599854,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 2791
    },
    {
      "epoch": 0.2505833782085801,
      "grad_norm": 2.906697988510132,
      "learning_rate": 0.0,
      "loss": 0.7708,
      "step": 2792
    },
    {
      "epoch": 0.25067312870220787,
      "grad_norm": 4.810269832611084,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 2793
    },
    {
      "epoch": 0.25076287919583556,
      "grad_norm": 3.201359272003174,
      "learning_rate": 0.0,
      "loss": 0.8108,
      "step": 2794
    },
    {
      "epoch": 0.2508526296894633,
      "grad_norm": 3.9152379035949707,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 2795
    },
    {
      "epoch": 0.250942380183091,
      "grad_norm": 413.7352600097656,
      "learning_rate": 0.0,
      "loss": 0.8885,
      "step": 2796
    },
    {
      "epoch": 0.2510321306767187,
      "grad_norm": 3.8939695358276367,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 2797
    },
    {
      "epoch": 0.25112188117034645,
      "grad_norm": 2.797492504119873,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 2798
    },
    {
      "epoch": 0.25121163166397414,
      "grad_norm": 3.272709846496582,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 2799
    },
    {
      "epoch": 0.2513013821576019,
      "grad_norm": 2.785701274871826,
      "learning_rate": 0.0,
      "loss": 0.7749,
      "step": 2800
    },
    {
      "epoch": 0.2513911326512296,
      "grad_norm": 3.992577075958252,
      "learning_rate": 0.0,
      "loss": 0.7957,
      "step": 2801
    },
    {
      "epoch": 0.2514808831448573,
      "grad_norm": 3.354031801223755,
      "learning_rate": 0.0,
      "loss": 0.8006,
      "step": 2802
    },
    {
      "epoch": 0.25157063363848503,
      "grad_norm": 2.880790948867798,
      "learning_rate": 0.0,
      "loss": 0.7807,
      "step": 2803
    },
    {
      "epoch": 0.2516603841321127,
      "grad_norm": 4.090303897857666,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 2804
    },
    {
      "epoch": 0.2517501346257404,
      "grad_norm": 3.467247247695923,
      "learning_rate": 0.0,
      "loss": 0.803,
      "step": 2805
    },
    {
      "epoch": 0.25183988511936817,
      "grad_norm": 3.36556077003479,
      "learning_rate": 0.0,
      "loss": 0.7897,
      "step": 2806
    },
    {
      "epoch": 0.25192963561299586,
      "grad_norm": 2.8132619857788086,
      "learning_rate": 0.0,
      "loss": 0.7874,
      "step": 2807
    },
    {
      "epoch": 0.2520193861066236,
      "grad_norm": 3.296696424484253,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 2808
    },
    {
      "epoch": 0.2521091366002513,
      "grad_norm": 2.8119418621063232,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 2809
    },
    {
      "epoch": 0.252198887093879,
      "grad_norm": 4.17502498626709,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 2810
    },
    {
      "epoch": 0.25228863758750675,
      "grad_norm": 2.9547994136810303,
      "learning_rate": 0.0,
      "loss": 0.799,
      "step": 2811
    },
    {
      "epoch": 0.25237838808113444,
      "grad_norm": 14.606375694274902,
      "learning_rate": 0.0,
      "loss": 0.7925,
      "step": 2812
    },
    {
      "epoch": 0.25246813857476214,
      "grad_norm": 2.5967602729797363,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 2813
    },
    {
      "epoch": 0.2525578890683899,
      "grad_norm": 3.0838632583618164,
      "learning_rate": 0.0,
      "loss": 0.8075,
      "step": 2814
    },
    {
      "epoch": 0.2526476395620176,
      "grad_norm": 2.789149522781372,
      "learning_rate": 0.0,
      "loss": 0.7913,
      "step": 2815
    },
    {
      "epoch": 0.25273739005564533,
      "grad_norm": 3.5498058795928955,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 2816
    },
    {
      "epoch": 0.252827140549273,
      "grad_norm": 2.628063678741455,
      "learning_rate": 0.0,
      "loss": 0.7826,
      "step": 2817
    },
    {
      "epoch": 0.2529168910429007,
      "grad_norm": 2.969433307647705,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 2818
    },
    {
      "epoch": 0.25300664153652846,
      "grad_norm": 2.711655616760254,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 2819
    },
    {
      "epoch": 0.25309639203015616,
      "grad_norm": 3.6263339519500732,
      "learning_rate": 0.0,
      "loss": 0.773,
      "step": 2820
    },
    {
      "epoch": 0.2531861425237839,
      "grad_norm": 172.96441650390625,
      "learning_rate": 0.0,
      "loss": 0.8475,
      "step": 2821
    },
    {
      "epoch": 0.2532758930174116,
      "grad_norm": 3.468242645263672,
      "learning_rate": 0.0,
      "loss": 0.7698,
      "step": 2822
    },
    {
      "epoch": 0.2533656435110393,
      "grad_norm": 3.1004340648651123,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 2823
    },
    {
      "epoch": 0.25345539400466705,
      "grad_norm": 3.0372822284698486,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 2824
    },
    {
      "epoch": 0.25354514449829474,
      "grad_norm": 3.75642728805542,
      "learning_rate": 0.0,
      "loss": 0.7882,
      "step": 2825
    },
    {
      "epoch": 0.25363489499192243,
      "grad_norm": 3.291093349456787,
      "learning_rate": 0.0,
      "loss": 0.7564,
      "step": 2826
    },
    {
      "epoch": 0.2537246454855502,
      "grad_norm": 4.5267720222473145,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 2827
    },
    {
      "epoch": 0.2538143959791779,
      "grad_norm": 2.9779109954833984,
      "learning_rate": 0.0,
      "loss": 0.7709,
      "step": 2828
    },
    {
      "epoch": 0.2539041464728056,
      "grad_norm": 2.7585065364837646,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 2829
    },
    {
      "epoch": 0.2539938969664333,
      "grad_norm": 4.117290019989014,
      "learning_rate": 0.0,
      "loss": 0.7196,
      "step": 2830
    },
    {
      "epoch": 0.254083647460061,
      "grad_norm": 9.211283683776855,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 2831
    },
    {
      "epoch": 0.25417339795368876,
      "grad_norm": 2.9148855209350586,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 2832
    },
    {
      "epoch": 0.25426314844731646,
      "grad_norm": 4.149166107177734,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 2833
    },
    {
      "epoch": 0.25435289894094415,
      "grad_norm": 5.582327365875244,
      "learning_rate": 0.0,
      "loss": 0.7879,
      "step": 2834
    },
    {
      "epoch": 0.2544426494345719,
      "grad_norm": 2.5733580589294434,
      "learning_rate": 0.0,
      "loss": 0.8353,
      "step": 2835
    },
    {
      "epoch": 0.2545323999281996,
      "grad_norm": 2.8653576374053955,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 2836
    },
    {
      "epoch": 0.25462215042182734,
      "grad_norm": 3.793436288833618,
      "learning_rate": 0.0,
      "loss": 0.7486,
      "step": 2837
    },
    {
      "epoch": 0.25471190091545504,
      "grad_norm": 2.812389373779297,
      "learning_rate": 0.0,
      "loss": 0.7344,
      "step": 2838
    },
    {
      "epoch": 0.25480165140908273,
      "grad_norm": 2.790384531021118,
      "learning_rate": 0.0,
      "loss": 0.7381,
      "step": 2839
    },
    {
      "epoch": 0.2548914019027105,
      "grad_norm": 2.3973681926727295,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 2840
    },
    {
      "epoch": 0.2549811523963382,
      "grad_norm": 2.4393787384033203,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 2841
    },
    {
      "epoch": 0.2550709028899659,
      "grad_norm": 5.538821220397949,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 2842
    },
    {
      "epoch": 0.2551606533835936,
      "grad_norm": 20.704748153686523,
      "learning_rate": 0.0,
      "loss": 0.858,
      "step": 2843
    },
    {
      "epoch": 0.2552504038772213,
      "grad_norm": 3.8307576179504395,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 2844
    },
    {
      "epoch": 0.25534015437084906,
      "grad_norm": 3.238501787185669,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 2845
    },
    {
      "epoch": 0.25542990486447675,
      "grad_norm": 3.2502121925354004,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 2846
    },
    {
      "epoch": 0.25551965535810445,
      "grad_norm": 3.598841428756714,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 2847
    },
    {
      "epoch": 0.2556094058517322,
      "grad_norm": 3.116950273513794,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 2848
    },
    {
      "epoch": 0.2556991563453599,
      "grad_norm": 3.3590707778930664,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 2849
    },
    {
      "epoch": 0.25578890683898764,
      "grad_norm": 5.971126556396484,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 2850
    },
    {
      "epoch": 0.25587865733261533,
      "grad_norm": 2.6225571632385254,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 2851
    },
    {
      "epoch": 0.255968407826243,
      "grad_norm": 4.123240947723389,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 2852
    },
    {
      "epoch": 0.2560581583198708,
      "grad_norm": 2.8067827224731445,
      "learning_rate": 0.0,
      "loss": 0.8011,
      "step": 2853
    },
    {
      "epoch": 0.25614790881349847,
      "grad_norm": 4.085751533508301,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 2854
    },
    {
      "epoch": 0.25623765930712616,
      "grad_norm": 6.700647354125977,
      "learning_rate": 0.0,
      "loss": 0.8207,
      "step": 2855
    },
    {
      "epoch": 0.2563274098007539,
      "grad_norm": 4.072305202484131,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 2856
    },
    {
      "epoch": 0.2564171602943816,
      "grad_norm": 15.537158966064453,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 2857
    },
    {
      "epoch": 0.25650691078800936,
      "grad_norm": 2.9159038066864014,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 2858
    },
    {
      "epoch": 0.25659666128163705,
      "grad_norm": 2.5765433311462402,
      "learning_rate": 0.0,
      "loss": 0.7861,
      "step": 2859
    },
    {
      "epoch": 0.25668641177526474,
      "grad_norm": 3.1061863899230957,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 2860
    },
    {
      "epoch": 0.2567761622688925,
      "grad_norm": 5.5295891761779785,
      "learning_rate": 0.0,
      "loss": 0.7987,
      "step": 2861
    },
    {
      "epoch": 0.2568659127625202,
      "grad_norm": 3.996772527694702,
      "learning_rate": 0.0,
      "loss": 0.8293,
      "step": 2862
    },
    {
      "epoch": 0.2569556632561479,
      "grad_norm": 20.105995178222656,
      "learning_rate": 0.0,
      "loss": 0.8321,
      "step": 2863
    },
    {
      "epoch": 0.25704541374977563,
      "grad_norm": 3.3899309635162354,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 2864
    },
    {
      "epoch": 0.2571351642434033,
      "grad_norm": 2.2509312629699707,
      "learning_rate": 0.0,
      "loss": 0.7739,
      "step": 2865
    },
    {
      "epoch": 0.2572249147370311,
      "grad_norm": 2.7472658157348633,
      "learning_rate": 0.0,
      "loss": 0.7702,
      "step": 2866
    },
    {
      "epoch": 0.25731466523065877,
      "grad_norm": 2.6764237880706787,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 2867
    },
    {
      "epoch": 0.25740441572428646,
      "grad_norm": 3.761321544647217,
      "learning_rate": 0.0,
      "loss": 0.7801,
      "step": 2868
    },
    {
      "epoch": 0.2574941662179142,
      "grad_norm": 3.1296777725219727,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 2869
    },
    {
      "epoch": 0.2575839167115419,
      "grad_norm": 3.453274726867676,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 2870
    },
    {
      "epoch": 0.25767366720516965,
      "grad_norm": 7.439834117889404,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 2871
    },
    {
      "epoch": 0.25776341769879735,
      "grad_norm": 3.444230556488037,
      "learning_rate": 0.0,
      "loss": 0.7667,
      "step": 2872
    },
    {
      "epoch": 0.25785316819242504,
      "grad_norm": 2.415195941925049,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 2873
    },
    {
      "epoch": 0.2579429186860528,
      "grad_norm": 2.9824018478393555,
      "learning_rate": 0.0,
      "loss": 0.7624,
      "step": 2874
    },
    {
      "epoch": 0.2580326691796805,
      "grad_norm": 3.179471015930176,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 2875
    },
    {
      "epoch": 0.2581224196733082,
      "grad_norm": 2.8923346996307373,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 2876
    },
    {
      "epoch": 0.25821217016693593,
      "grad_norm": 7.942513465881348,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 2877
    },
    {
      "epoch": 0.2583019206605636,
      "grad_norm": 4.05005407333374,
      "learning_rate": 0.0,
      "loss": 0.8109,
      "step": 2878
    },
    {
      "epoch": 0.25839167115419137,
      "grad_norm": 2.720015048980713,
      "learning_rate": 0.0,
      "loss": 0.8101,
      "step": 2879
    },
    {
      "epoch": 0.25848142164781907,
      "grad_norm": 4.00491189956665,
      "learning_rate": 0.0,
      "loss": 0.7669,
      "step": 2880
    },
    {
      "epoch": 0.25857117214144676,
      "grad_norm": 4.902667999267578,
      "learning_rate": 0.0,
      "loss": 0.8146,
      "step": 2881
    },
    {
      "epoch": 0.2586609226350745,
      "grad_norm": 2.926370620727539,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 2882
    },
    {
      "epoch": 0.2587506731287022,
      "grad_norm": 2.630582332611084,
      "learning_rate": 0.0,
      "loss": 0.7496,
      "step": 2883
    },
    {
      "epoch": 0.2588404236223299,
      "grad_norm": 3.800987482070923,
      "learning_rate": 0.0,
      "loss": 0.8219,
      "step": 2884
    },
    {
      "epoch": 0.25893017411595765,
      "grad_norm": 4.342880725860596,
      "learning_rate": 0.0,
      "loss": 0.7821,
      "step": 2885
    },
    {
      "epoch": 0.25901992460958534,
      "grad_norm": 3.13832950592041,
      "learning_rate": 0.0,
      "loss": 0.7654,
      "step": 2886
    },
    {
      "epoch": 0.2591096751032131,
      "grad_norm": 3.035426378250122,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 2887
    },
    {
      "epoch": 0.2591994255968408,
      "grad_norm": 3.0700390338897705,
      "learning_rate": 0.0,
      "loss": 0.8006,
      "step": 2888
    },
    {
      "epoch": 0.2592891760904685,
      "grad_norm": 2.6162033081054688,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 2889
    },
    {
      "epoch": 0.2593789265840962,
      "grad_norm": 2.7651779651641846,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 2890
    },
    {
      "epoch": 0.2594686770777239,
      "grad_norm": 4.024365425109863,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 2891
    },
    {
      "epoch": 0.25955842757135167,
      "grad_norm": 5.086237907409668,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 2892
    },
    {
      "epoch": 0.25964817806497936,
      "grad_norm": 4.119685173034668,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 2893
    },
    {
      "epoch": 0.25973792855860706,
      "grad_norm": 26.650983810424805,
      "learning_rate": 0.0,
      "loss": 0.7911,
      "step": 2894
    },
    {
      "epoch": 0.2598276790522348,
      "grad_norm": 3.8131563663482666,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 2895
    },
    {
      "epoch": 0.2599174295458625,
      "grad_norm": 3.1717917919158936,
      "learning_rate": 0.0,
      "loss": 0.7643,
      "step": 2896
    },
    {
      "epoch": 0.2600071800394902,
      "grad_norm": 2.6862025260925293,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 2897
    },
    {
      "epoch": 0.26009693053311794,
      "grad_norm": 2.8595705032348633,
      "learning_rate": 0.0,
      "loss": 0.7143,
      "step": 2898
    },
    {
      "epoch": 0.26018668102674564,
      "grad_norm": 6.262414932250977,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 2899
    },
    {
      "epoch": 0.2602764315203734,
      "grad_norm": 3.4734106063842773,
      "learning_rate": 0.0,
      "loss": 0.7935,
      "step": 2900
    },
    {
      "epoch": 0.2603661820140011,
      "grad_norm": 2.79128098487854,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 2901
    },
    {
      "epoch": 0.2604559325076288,
      "grad_norm": 3.170499324798584,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 2902
    },
    {
      "epoch": 0.2605456830012565,
      "grad_norm": 2.987823009490967,
      "learning_rate": 0.0,
      "loss": 0.6854,
      "step": 2903
    },
    {
      "epoch": 0.2606354334948842,
      "grad_norm": 2.698300361633301,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 2904
    },
    {
      "epoch": 0.2607251839885119,
      "grad_norm": 2.8180904388427734,
      "learning_rate": 0.0,
      "loss": 0.7546,
      "step": 2905
    },
    {
      "epoch": 0.26081493448213966,
      "grad_norm": 3.7628965377807617,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 2906
    },
    {
      "epoch": 0.26090468497576735,
      "grad_norm": 3.0157883167266846,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 2907
    },
    {
      "epoch": 0.2609944354693951,
      "grad_norm": 2.793706178665161,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 2908
    },
    {
      "epoch": 0.2610841859630228,
      "grad_norm": 2.53049898147583,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 2909
    },
    {
      "epoch": 0.2611739364566505,
      "grad_norm": 3.2014763355255127,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 2910
    },
    {
      "epoch": 0.26126368695027824,
      "grad_norm": 2.627936601638794,
      "learning_rate": 0.0,
      "loss": 0.808,
      "step": 2911
    },
    {
      "epoch": 0.26135343744390593,
      "grad_norm": 2.7048826217651367,
      "learning_rate": 0.0,
      "loss": 0.697,
      "step": 2912
    },
    {
      "epoch": 0.2614431879375337,
      "grad_norm": 2.743967294692993,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 2913
    },
    {
      "epoch": 0.2615329384311614,
      "grad_norm": 3.6487069129943848,
      "learning_rate": 0.0,
      "loss": 0.7653,
      "step": 2914
    },
    {
      "epoch": 0.26162268892478907,
      "grad_norm": 4.770095348358154,
      "learning_rate": 0.0,
      "loss": 0.796,
      "step": 2915
    },
    {
      "epoch": 0.2617124394184168,
      "grad_norm": 2.6619949340820312,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 2916
    },
    {
      "epoch": 0.2618021899120445,
      "grad_norm": 3.4179129600524902,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 2917
    },
    {
      "epoch": 0.2618919404056722,
      "grad_norm": 3.2393527030944824,
      "learning_rate": 0.0,
      "loss": 0.779,
      "step": 2918
    },
    {
      "epoch": 0.26198169089929996,
      "grad_norm": 3.885338544845581,
      "learning_rate": 0.0,
      "loss": 0.7724,
      "step": 2919
    },
    {
      "epoch": 0.26207144139292765,
      "grad_norm": 3.4817709922790527,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 2920
    },
    {
      "epoch": 0.2621611918865554,
      "grad_norm": 3.2577738761901855,
      "learning_rate": 0.0,
      "loss": 0.7858,
      "step": 2921
    },
    {
      "epoch": 0.2622509423801831,
      "grad_norm": 3.130004644393921,
      "learning_rate": 0.0,
      "loss": 0.7731,
      "step": 2922
    },
    {
      "epoch": 0.2623406928738108,
      "grad_norm": 4.540986061096191,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 2923
    },
    {
      "epoch": 0.26243044336743854,
      "grad_norm": 3.6105432510375977,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 2924
    },
    {
      "epoch": 0.26252019386106623,
      "grad_norm": 6.99323034286499,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 2925
    },
    {
      "epoch": 0.2626099443546939,
      "grad_norm": 4.001212120056152,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 2926
    },
    {
      "epoch": 0.2626996948483217,
      "grad_norm": 3.4812989234924316,
      "learning_rate": 0.0,
      "loss": 0.7564,
      "step": 2927
    },
    {
      "epoch": 0.26278944534194937,
      "grad_norm": 4.610288143157959,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 2928
    },
    {
      "epoch": 0.2628791958355771,
      "grad_norm": 2.526365041732788,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 2929
    },
    {
      "epoch": 0.2629689463292048,
      "grad_norm": 2.513017416000366,
      "learning_rate": 0.0,
      "loss": 0.8371,
      "step": 2930
    },
    {
      "epoch": 0.2630586968228325,
      "grad_norm": 4.9943928718566895,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 2931
    },
    {
      "epoch": 0.26314844731646025,
      "grad_norm": 3.897066354751587,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 2932
    },
    {
      "epoch": 0.26323819781008795,
      "grad_norm": 2.323688268661499,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 2933
    },
    {
      "epoch": 0.2633279483037157,
      "grad_norm": 2.7901833057403564,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 2934
    },
    {
      "epoch": 0.2634176987973434,
      "grad_norm": 2.350621223449707,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 2935
    },
    {
      "epoch": 0.2635074492909711,
      "grad_norm": 13.34140682220459,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 2936
    },
    {
      "epoch": 0.26359719978459883,
      "grad_norm": 7.013286590576172,
      "learning_rate": 0.0,
      "loss": 0.7781,
      "step": 2937
    },
    {
      "epoch": 0.26368695027822653,
      "grad_norm": 2.5963327884674072,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 2938
    },
    {
      "epoch": 0.2637767007718542,
      "grad_norm": 2.3206801414489746,
      "learning_rate": 0.0,
      "loss": 0.7886,
      "step": 2939
    },
    {
      "epoch": 0.26386645126548197,
      "grad_norm": 3.5948991775512695,
      "learning_rate": 0.0,
      "loss": 0.7849,
      "step": 2940
    },
    {
      "epoch": 0.26395620175910967,
      "grad_norm": 3.0094244480133057,
      "learning_rate": 0.0,
      "loss": 0.8011,
      "step": 2941
    },
    {
      "epoch": 0.2640459522527374,
      "grad_norm": 3.0205202102661133,
      "learning_rate": 0.0,
      "loss": 0.7844,
      "step": 2942
    },
    {
      "epoch": 0.2641357027463651,
      "grad_norm": 4.134900093078613,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 2943
    },
    {
      "epoch": 0.2642254532399928,
      "grad_norm": 2.56685209274292,
      "learning_rate": 0.0,
      "loss": 0.752,
      "step": 2944
    },
    {
      "epoch": 0.26431520373362055,
      "grad_norm": 2.662954330444336,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 2945
    },
    {
      "epoch": 0.26440495422724825,
      "grad_norm": 2.333239793777466,
      "learning_rate": 0.0,
      "loss": 0.793,
      "step": 2946
    },
    {
      "epoch": 0.26449470472087594,
      "grad_norm": 3.5880753993988037,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 2947
    },
    {
      "epoch": 0.2645844552145037,
      "grad_norm": 2.806663751602173,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 2948
    },
    {
      "epoch": 0.2646742057081314,
      "grad_norm": 2.6795334815979004,
      "learning_rate": 0.0,
      "loss": 0.7711,
      "step": 2949
    },
    {
      "epoch": 0.26476395620175913,
      "grad_norm": 3.1217803955078125,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 2950
    },
    {
      "epoch": 0.2648537066953868,
      "grad_norm": 4.180622577667236,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 2951
    },
    {
      "epoch": 0.2649434571890145,
      "grad_norm": 3.5035078525543213,
      "learning_rate": 0.0,
      "loss": 0.758,
      "step": 2952
    },
    {
      "epoch": 0.26503320768264227,
      "grad_norm": 3.1055777072906494,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 2953
    },
    {
      "epoch": 0.26512295817626996,
      "grad_norm": 3.2774147987365723,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 2954
    },
    {
      "epoch": 0.2652127086698977,
      "grad_norm": 3.5005476474761963,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 2955
    },
    {
      "epoch": 0.2653024591635254,
      "grad_norm": 2.845492124557495,
      "learning_rate": 0.0,
      "loss": 0.787,
      "step": 2956
    },
    {
      "epoch": 0.2653922096571531,
      "grad_norm": 53.49597930908203,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 2957
    },
    {
      "epoch": 0.26548196015078085,
      "grad_norm": 3.0870375633239746,
      "learning_rate": 0.0,
      "loss": 0.7465,
      "step": 2958
    },
    {
      "epoch": 0.26557171064440854,
      "grad_norm": 3.1960678100585938,
      "learning_rate": 0.0,
      "loss": 0.7932,
      "step": 2959
    },
    {
      "epoch": 0.26566146113803624,
      "grad_norm": 2.96126127243042,
      "learning_rate": 0.0,
      "loss": 0.7759,
      "step": 2960
    },
    {
      "epoch": 0.265751211631664,
      "grad_norm": 2.751762866973877,
      "learning_rate": 0.0,
      "loss": 0.7044,
      "step": 2961
    },
    {
      "epoch": 0.2658409621252917,
      "grad_norm": 2.9390337467193604,
      "learning_rate": 0.0,
      "loss": 0.7719,
      "step": 2962
    },
    {
      "epoch": 0.26593071261891943,
      "grad_norm": 3.418501853942871,
      "learning_rate": 0.0,
      "loss": 0.799,
      "step": 2963
    },
    {
      "epoch": 0.2660204631125471,
      "grad_norm": 2.3739030361175537,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 2964
    },
    {
      "epoch": 0.2661102136061748,
      "grad_norm": 2.472694158554077,
      "learning_rate": 0.0,
      "loss": 0.7998,
      "step": 2965
    },
    {
      "epoch": 0.26619996409980257,
      "grad_norm": 2.725851535797119,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 2966
    },
    {
      "epoch": 0.26628971459343026,
      "grad_norm": 4.695065498352051,
      "learning_rate": 0.0,
      "loss": 0.8017,
      "step": 2967
    },
    {
      "epoch": 0.26637946508705795,
      "grad_norm": 3.3947558403015137,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 2968
    },
    {
      "epoch": 0.2664692155806857,
      "grad_norm": 4.366701126098633,
      "learning_rate": 0.0,
      "loss": 0.7677,
      "step": 2969
    },
    {
      "epoch": 0.2665589660743134,
      "grad_norm": 3.2546918392181396,
      "learning_rate": 0.0,
      "loss": 0.7855,
      "step": 2970
    },
    {
      "epoch": 0.26664871656794115,
      "grad_norm": 2.6016783714294434,
      "learning_rate": 0.0,
      "loss": 0.7718,
      "step": 2971
    },
    {
      "epoch": 0.26673846706156884,
      "grad_norm": 4.892971992492676,
      "learning_rate": 0.0,
      "loss": 0.7892,
      "step": 2972
    },
    {
      "epoch": 0.26682821755519653,
      "grad_norm": 3.0029728412628174,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 2973
    },
    {
      "epoch": 0.2669179680488243,
      "grad_norm": 13.094743728637695,
      "learning_rate": 0.0,
      "loss": 0.7984,
      "step": 2974
    },
    {
      "epoch": 0.267007718542452,
      "grad_norm": 2.6509766578674316,
      "learning_rate": 0.0,
      "loss": 0.7741,
      "step": 2975
    },
    {
      "epoch": 0.26709746903607967,
      "grad_norm": 3.3759894371032715,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 2976
    },
    {
      "epoch": 0.2671872195297074,
      "grad_norm": 3.187668800354004,
      "learning_rate": 0.0,
      "loss": 0.808,
      "step": 2977
    },
    {
      "epoch": 0.2672769700233351,
      "grad_norm": 3.602109670639038,
      "learning_rate": 0.0,
      "loss": 0.8234,
      "step": 2978
    },
    {
      "epoch": 0.26736672051696286,
      "grad_norm": 4.051023960113525,
      "learning_rate": 0.0,
      "loss": 0.803,
      "step": 2979
    },
    {
      "epoch": 0.26745647101059056,
      "grad_norm": 3.336874485015869,
      "learning_rate": 0.0,
      "loss": 0.7788,
      "step": 2980
    },
    {
      "epoch": 0.26754622150421825,
      "grad_norm": 3.2903289794921875,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 2981
    },
    {
      "epoch": 0.267635971997846,
      "grad_norm": 2.561540126800537,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 2982
    },
    {
      "epoch": 0.2677257224914737,
      "grad_norm": 4.458677291870117,
      "learning_rate": 0.0,
      "loss": 0.7687,
      "step": 2983
    },
    {
      "epoch": 0.26781547298510144,
      "grad_norm": 4.996741771697998,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 2984
    },
    {
      "epoch": 0.26790522347872914,
      "grad_norm": 3.1721575260162354,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 2985
    },
    {
      "epoch": 0.26799497397235683,
      "grad_norm": 6.288721084594727,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 2986
    },
    {
      "epoch": 0.2680847244659846,
      "grad_norm": 8.134604454040527,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 2987
    },
    {
      "epoch": 0.2681744749596123,
      "grad_norm": 4.504687309265137,
      "learning_rate": 0.0,
      "loss": 0.8165,
      "step": 2988
    },
    {
      "epoch": 0.26826422545323997,
      "grad_norm": 4.978982448577881,
      "learning_rate": 0.0,
      "loss": 0.8395,
      "step": 2989
    },
    {
      "epoch": 0.2683539759468677,
      "grad_norm": 3.468486785888672,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 2990
    },
    {
      "epoch": 0.2684437264404954,
      "grad_norm": 3.1994285583496094,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 2991
    },
    {
      "epoch": 0.26853347693412316,
      "grad_norm": 4.334839344024658,
      "learning_rate": 0.0,
      "loss": 0.8067,
      "step": 2992
    },
    {
      "epoch": 0.26862322742775085,
      "grad_norm": 3.7653515338897705,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 2993
    },
    {
      "epoch": 0.26871297792137855,
      "grad_norm": 2.876997470855713,
      "learning_rate": 0.0,
      "loss": 0.7803,
      "step": 2994
    },
    {
      "epoch": 0.2688027284150063,
      "grad_norm": 3.5320072174072266,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 2995
    },
    {
      "epoch": 0.268892478908634,
      "grad_norm": 2.777146577835083,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 2996
    },
    {
      "epoch": 0.2689822294022617,
      "grad_norm": 3.6711642742156982,
      "learning_rate": 0.0,
      "loss": 0.7681,
      "step": 2997
    },
    {
      "epoch": 0.26907197989588943,
      "grad_norm": 4.102054119110107,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 2998
    },
    {
      "epoch": 0.26916173038951713,
      "grad_norm": 2.8111629486083984,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 2999
    },
    {
      "epoch": 0.2692514808831449,
      "grad_norm": 3.1452527046203613,
      "learning_rate": 0.0,
      "loss": 0.7517,
      "step": 3000
    },
    {
      "epoch": 0.26934123137677257,
      "grad_norm": 3.9498932361602783,
      "learning_rate": 0.0,
      "loss": 0.7864,
      "step": 3001
    },
    {
      "epoch": 0.26943098187040027,
      "grad_norm": 2.5715293884277344,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 3002
    },
    {
      "epoch": 0.269520732364028,
      "grad_norm": 2.972480297088623,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 3003
    },
    {
      "epoch": 0.2696104828576557,
      "grad_norm": 2.5581297874450684,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 3004
    },
    {
      "epoch": 0.26970023335128346,
      "grad_norm": 3.5248701572418213,
      "learning_rate": 0.0,
      "loss": 0.8021,
      "step": 3005
    },
    {
      "epoch": 0.26978998384491115,
      "grad_norm": 3.03531551361084,
      "learning_rate": 0.0,
      "loss": 0.6758,
      "step": 3006
    },
    {
      "epoch": 0.26987973433853885,
      "grad_norm": 2.17771053314209,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 3007
    },
    {
      "epoch": 0.2699694848321666,
      "grad_norm": 2.50476336479187,
      "learning_rate": 0.0,
      "loss": 0.8196,
      "step": 3008
    },
    {
      "epoch": 0.2700592353257943,
      "grad_norm": 3.217233180999756,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 3009
    },
    {
      "epoch": 0.270148985819422,
      "grad_norm": 5.206796169281006,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 3010
    },
    {
      "epoch": 0.27023873631304973,
      "grad_norm": 3.3194310665130615,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 3011
    },
    {
      "epoch": 0.2703284868066774,
      "grad_norm": 6.2397050857543945,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 3012
    },
    {
      "epoch": 0.2704182373003052,
      "grad_norm": 4.282526969909668,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 3013
    },
    {
      "epoch": 0.27050798779393287,
      "grad_norm": 2.9632949829101562,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 3014
    },
    {
      "epoch": 0.27059773828756056,
      "grad_norm": 2.372457504272461,
      "learning_rate": 0.0,
      "loss": 0.7819,
      "step": 3015
    },
    {
      "epoch": 0.2706874887811883,
      "grad_norm": 2.4241626262664795,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 3016
    },
    {
      "epoch": 0.270777239274816,
      "grad_norm": 2.48685359954834,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 3017
    },
    {
      "epoch": 0.2708669897684437,
      "grad_norm": 3.3482751846313477,
      "learning_rate": 0.0,
      "loss": 0.7712,
      "step": 3018
    },
    {
      "epoch": 0.27095674026207145,
      "grad_norm": 3.2201147079467773,
      "learning_rate": 0.0,
      "loss": 0.8017,
      "step": 3019
    },
    {
      "epoch": 0.27104649075569914,
      "grad_norm": 2.877253532409668,
      "learning_rate": 0.0,
      "loss": 0.7967,
      "step": 3020
    },
    {
      "epoch": 0.2711362412493269,
      "grad_norm": 5.885582447052002,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 3021
    },
    {
      "epoch": 0.2712259917429546,
      "grad_norm": 3.6243515014648438,
      "learning_rate": 0.0,
      "loss": 0.7933,
      "step": 3022
    },
    {
      "epoch": 0.2713157422365823,
      "grad_norm": 2.4325602054595947,
      "learning_rate": 0.0,
      "loss": 0.7676,
      "step": 3023
    },
    {
      "epoch": 0.27140549273021003,
      "grad_norm": 3.0253591537475586,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 3024
    },
    {
      "epoch": 0.2714952432238377,
      "grad_norm": 2.630872964859009,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 3025
    },
    {
      "epoch": 0.2715849937174655,
      "grad_norm": 3.350635528564453,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 3026
    },
    {
      "epoch": 0.27167474421109317,
      "grad_norm": 4.468926429748535,
      "learning_rate": 0.0,
      "loss": 0.7848,
      "step": 3027
    },
    {
      "epoch": 0.27176449470472086,
      "grad_norm": 8.822941780090332,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 3028
    },
    {
      "epoch": 0.2718542451983486,
      "grad_norm": 5.007615566253662,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 3029
    },
    {
      "epoch": 0.2719439956919763,
      "grad_norm": 3.064795970916748,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 3030
    },
    {
      "epoch": 0.272033746185604,
      "grad_norm": 4.005752086639404,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 3031
    },
    {
      "epoch": 0.27212349667923175,
      "grad_norm": 12.423362731933594,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 3032
    },
    {
      "epoch": 0.27221324717285944,
      "grad_norm": 5.236153602600098,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 3033
    },
    {
      "epoch": 0.2723029976664872,
      "grad_norm": 3.561709403991699,
      "learning_rate": 0.0,
      "loss": 0.7377,
      "step": 3034
    },
    {
      "epoch": 0.2723927481601149,
      "grad_norm": 4.069181442260742,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 3035
    },
    {
      "epoch": 0.2724824986537426,
      "grad_norm": 3.247835874557495,
      "learning_rate": 0.0,
      "loss": 0.7641,
      "step": 3036
    },
    {
      "epoch": 0.2725722491473703,
      "grad_norm": 3.3407418727874756,
      "learning_rate": 0.0,
      "loss": 0.747,
      "step": 3037
    },
    {
      "epoch": 0.272661999640998,
      "grad_norm": 4.648486137390137,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 3038
    },
    {
      "epoch": 0.2727517501346257,
      "grad_norm": 2.8156495094299316,
      "learning_rate": 0.0,
      "loss": 0.7899,
      "step": 3039
    },
    {
      "epoch": 0.27284150062825346,
      "grad_norm": 2.5997493267059326,
      "learning_rate": 0.0,
      "loss": 0.7336,
      "step": 3040
    },
    {
      "epoch": 0.27293125112188116,
      "grad_norm": 14.465076446533203,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 3041
    },
    {
      "epoch": 0.2730210016155089,
      "grad_norm": 5.0420660972595215,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 3042
    },
    {
      "epoch": 0.2731107521091366,
      "grad_norm": 3.405757427215576,
      "learning_rate": 0.0,
      "loss": 0.7631,
      "step": 3043
    },
    {
      "epoch": 0.2732005026027643,
      "grad_norm": 3.4181833267211914,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 3044
    },
    {
      "epoch": 0.27329025309639204,
      "grad_norm": 4.09797477722168,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 3045
    },
    {
      "epoch": 0.27338000359001974,
      "grad_norm": 3.2584917545318604,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 3046
    },
    {
      "epoch": 0.2734697540836475,
      "grad_norm": 2.6965482234954834,
      "learning_rate": 0.0,
      "loss": 0.7547,
      "step": 3047
    },
    {
      "epoch": 0.2735595045772752,
      "grad_norm": 2.327826499938965,
      "learning_rate": 0.0,
      "loss": 0.7211,
      "step": 3048
    },
    {
      "epoch": 0.2736492550709029,
      "grad_norm": 3.326425552368164,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 3049
    },
    {
      "epoch": 0.2737390055645306,
      "grad_norm": 2.640353202819824,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 3050
    },
    {
      "epoch": 0.2738287560581583,
      "grad_norm": 3.0357611179351807,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 3051
    },
    {
      "epoch": 0.273918506551786,
      "grad_norm": 5.4405107498168945,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 3052
    },
    {
      "epoch": 0.27400825704541376,
      "grad_norm": 2.927001953125,
      "learning_rate": 0.0,
      "loss": 0.7934,
      "step": 3053
    },
    {
      "epoch": 0.27409800753904145,
      "grad_norm": 8.687049865722656,
      "learning_rate": 0.0,
      "loss": 0.8445,
      "step": 3054
    },
    {
      "epoch": 0.2741877580326692,
      "grad_norm": 2.4808928966522217,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 3055
    },
    {
      "epoch": 0.2742775085262969,
      "grad_norm": 2.847161293029785,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 3056
    },
    {
      "epoch": 0.2743672590199246,
      "grad_norm": 2.2419211864471436,
      "learning_rate": 0.0,
      "loss": 0.7353,
      "step": 3057
    },
    {
      "epoch": 0.27445700951355234,
      "grad_norm": 4.268082618713379,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 3058
    },
    {
      "epoch": 0.27454676000718004,
      "grad_norm": 3.5765132904052734,
      "learning_rate": 0.0,
      "loss": 0.7433,
      "step": 3059
    },
    {
      "epoch": 0.27463651050080773,
      "grad_norm": 2.461768865585327,
      "learning_rate": 0.0,
      "loss": 0.7621,
      "step": 3060
    },
    {
      "epoch": 0.2747262609944355,
      "grad_norm": 2.976308584213257,
      "learning_rate": 0.0,
      "loss": 0.7806,
      "step": 3061
    },
    {
      "epoch": 0.27481601148806317,
      "grad_norm": 2.558361053466797,
      "learning_rate": 0.0,
      "loss": 0.8169,
      "step": 3062
    },
    {
      "epoch": 0.2749057619816909,
      "grad_norm": 3.1069562435150146,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 3063
    },
    {
      "epoch": 0.2749955124753186,
      "grad_norm": 2.5255467891693115,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 3064
    },
    {
      "epoch": 0.2750852629689463,
      "grad_norm": 2.494544744491577,
      "learning_rate": 0.0,
      "loss": 0.74,
      "step": 3065
    },
    {
      "epoch": 0.27517501346257406,
      "grad_norm": 3.6364002227783203,
      "learning_rate": 0.0,
      "loss": 0.7961,
      "step": 3066
    },
    {
      "epoch": 0.27526476395620175,
      "grad_norm": 19.300233840942383,
      "learning_rate": 0.0,
      "loss": 0.7925,
      "step": 3067
    },
    {
      "epoch": 0.2753545144498295,
      "grad_norm": 2.8891146183013916,
      "learning_rate": 0.0,
      "loss": 0.752,
      "step": 3068
    },
    {
      "epoch": 0.2754442649434572,
      "grad_norm": 2.4875688552856445,
      "learning_rate": 0.0,
      "loss": 0.7865,
      "step": 3069
    },
    {
      "epoch": 0.2755340154370849,
      "grad_norm": 2.8835201263427734,
      "learning_rate": 0.0,
      "loss": 0.764,
      "step": 3070
    },
    {
      "epoch": 0.27562376593071264,
      "grad_norm": 3.4896979331970215,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 3071
    },
    {
      "epoch": 0.27571351642434033,
      "grad_norm": 6.99737548828125,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 3072
    },
    {
      "epoch": 0.275803266917968,
      "grad_norm": 3.1063950061798096,
      "learning_rate": 0.0,
      "loss": 0.8123,
      "step": 3073
    },
    {
      "epoch": 0.2758930174115958,
      "grad_norm": 3.1148364543914795,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 3074
    },
    {
      "epoch": 0.27598276790522347,
      "grad_norm": 2.4155895709991455,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 3075
    },
    {
      "epoch": 0.2760725183988512,
      "grad_norm": 2.5293352603912354,
      "learning_rate": 0.0,
      "loss": 0.7626,
      "step": 3076
    },
    {
      "epoch": 0.2761622688924789,
      "grad_norm": 2.628899335861206,
      "learning_rate": 0.0,
      "loss": 0.7641,
      "step": 3077
    },
    {
      "epoch": 0.2762520193861066,
      "grad_norm": 2.453782320022583,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 3078
    },
    {
      "epoch": 0.27634176987973436,
      "grad_norm": 2.8784584999084473,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 3079
    },
    {
      "epoch": 0.27643152037336205,
      "grad_norm": 2.9512014389038086,
      "learning_rate": 0.0,
      "loss": 0.7336,
      "step": 3080
    },
    {
      "epoch": 0.27652127086698974,
      "grad_norm": 2.66017746925354,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 3081
    },
    {
      "epoch": 0.2766110213606175,
      "grad_norm": 2.780503988265991,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 3082
    },
    {
      "epoch": 0.2767007718542452,
      "grad_norm": 9.43493366241455,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 3083
    },
    {
      "epoch": 0.27679052234787294,
      "grad_norm": 3.0544445514678955,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 3084
    },
    {
      "epoch": 0.27688027284150063,
      "grad_norm": 3.4463858604431152,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 3085
    },
    {
      "epoch": 0.2769700233351283,
      "grad_norm": 2.4818053245544434,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 3086
    },
    {
      "epoch": 0.2770597738287561,
      "grad_norm": 2.431753158569336,
      "learning_rate": 0.0,
      "loss": 0.7855,
      "step": 3087
    },
    {
      "epoch": 0.27714952432238377,
      "grad_norm": 2.842661142349243,
      "learning_rate": 0.0,
      "loss": 0.7983,
      "step": 3088
    },
    {
      "epoch": 0.27723927481601146,
      "grad_norm": 2.7223024368286133,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 3089
    },
    {
      "epoch": 0.2773290253096392,
      "grad_norm": 2.474120855331421,
      "learning_rate": 0.0,
      "loss": 0.7456,
      "step": 3090
    },
    {
      "epoch": 0.2774187758032669,
      "grad_norm": 3.200927972793579,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 3091
    },
    {
      "epoch": 0.27750852629689465,
      "grad_norm": 7.1222076416015625,
      "learning_rate": 0.0,
      "loss": 0.8283,
      "step": 3092
    },
    {
      "epoch": 0.27759827679052235,
      "grad_norm": 2.8973474502563477,
      "learning_rate": 0.0,
      "loss": 0.7811,
      "step": 3093
    },
    {
      "epoch": 0.27768802728415004,
      "grad_norm": 2.424743890762329,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 3094
    },
    {
      "epoch": 0.2777777777777778,
      "grad_norm": 2.3682079315185547,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 3095
    },
    {
      "epoch": 0.2778675282714055,
      "grad_norm": 2.3820133209228516,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 3096
    },
    {
      "epoch": 0.27795727876503323,
      "grad_norm": 2.8475351333618164,
      "learning_rate": 0.0,
      "loss": 0.7411,
      "step": 3097
    },
    {
      "epoch": 0.2780470292586609,
      "grad_norm": 2.7324440479278564,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 3098
    },
    {
      "epoch": 0.2781367797522886,
      "grad_norm": 3.0259761810302734,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 3099
    },
    {
      "epoch": 0.27822653024591637,
      "grad_norm": 3.762911081314087,
      "learning_rate": 0.0,
      "loss": 0.7847,
      "step": 3100
    },
    {
      "epoch": 0.27831628073954406,
      "grad_norm": 2.453549385070801,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 3101
    },
    {
      "epoch": 0.27840603123317176,
      "grad_norm": 6.839095115661621,
      "learning_rate": 0.0,
      "loss": 0.7699,
      "step": 3102
    },
    {
      "epoch": 0.2784957817267995,
      "grad_norm": 2.945545196533203,
      "learning_rate": 0.0,
      "loss": 0.7868,
      "step": 3103
    },
    {
      "epoch": 0.2785855322204272,
      "grad_norm": 2.590049982070923,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 3104
    },
    {
      "epoch": 0.27867528271405495,
      "grad_norm": 3.274937391281128,
      "learning_rate": 0.0,
      "loss": 0.7835,
      "step": 3105
    },
    {
      "epoch": 0.27876503320768264,
      "grad_norm": 4.601589202880859,
      "learning_rate": 0.0,
      "loss": 0.7653,
      "step": 3106
    },
    {
      "epoch": 0.27885478370131034,
      "grad_norm": 3.525275707244873,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 3107
    },
    {
      "epoch": 0.2789445341949381,
      "grad_norm": 19.87906265258789,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 3108
    },
    {
      "epoch": 0.2790342846885658,
      "grad_norm": 5.118961811065674,
      "learning_rate": 0.0,
      "loss": 0.7812,
      "step": 3109
    },
    {
      "epoch": 0.2791240351821935,
      "grad_norm": 2.8912575244903564,
      "learning_rate": 0.0,
      "loss": 0.7842,
      "step": 3110
    },
    {
      "epoch": 0.2792137856758212,
      "grad_norm": 3.9728052616119385,
      "learning_rate": 0.0,
      "loss": 0.7594,
      "step": 3111
    },
    {
      "epoch": 0.2793035361694489,
      "grad_norm": 2.6268694400787354,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 3112
    },
    {
      "epoch": 0.27939328666307667,
      "grad_norm": 1.826672911643982,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 3113
    },
    {
      "epoch": 0.27948303715670436,
      "grad_norm": 2.2713658809661865,
      "learning_rate": 0.0,
      "loss": 0.764,
      "step": 3114
    },
    {
      "epoch": 0.27957278765033206,
      "grad_norm": 3.178407907485962,
      "learning_rate": 0.0,
      "loss": 0.7912,
      "step": 3115
    },
    {
      "epoch": 0.2796625381439598,
      "grad_norm": 2.7901062965393066,
      "learning_rate": 0.0,
      "loss": 0.8048,
      "step": 3116
    },
    {
      "epoch": 0.2797522886375875,
      "grad_norm": 2.3170573711395264,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 3117
    },
    {
      "epoch": 0.27984203913121525,
      "grad_norm": 3.0817806720733643,
      "learning_rate": 0.0,
      "loss": 0.7931,
      "step": 3118
    },
    {
      "epoch": 0.27993178962484294,
      "grad_norm": 2.5453481674194336,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 3119
    },
    {
      "epoch": 0.28002154011847064,
      "grad_norm": 2.0330705642700195,
      "learning_rate": 0.0,
      "loss": 0.7486,
      "step": 3120
    },
    {
      "epoch": 0.2801112906120984,
      "grad_norm": 2.42362904548645,
      "learning_rate": 0.0,
      "loss": 0.6941,
      "step": 3121
    },
    {
      "epoch": 0.2802010411057261,
      "grad_norm": 2.572549343109131,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 3122
    },
    {
      "epoch": 0.2802907915993538,
      "grad_norm": 2.813570022583008,
      "learning_rate": 0.0,
      "loss": 0.7733,
      "step": 3123
    },
    {
      "epoch": 0.2803805420929815,
      "grad_norm": 2.1362617015838623,
      "learning_rate": 0.0,
      "loss": 0.7618,
      "step": 3124
    },
    {
      "epoch": 0.2804702925866092,
      "grad_norm": 2.9723496437072754,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 3125
    },
    {
      "epoch": 0.28056004308023696,
      "grad_norm": 2.5074429512023926,
      "learning_rate": 0.0,
      "loss": 0.8083,
      "step": 3126
    },
    {
      "epoch": 0.28064979357386466,
      "grad_norm": 1.8898944854736328,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 3127
    },
    {
      "epoch": 0.28073954406749235,
      "grad_norm": 2.0742275714874268,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 3128
    },
    {
      "epoch": 0.2808292945611201,
      "grad_norm": 16.8712100982666,
      "learning_rate": 0.0,
      "loss": 0.8802,
      "step": 3129
    },
    {
      "epoch": 0.2809190450547478,
      "grad_norm": 2.191329002380371,
      "learning_rate": 0.0,
      "loss": 0.7734,
      "step": 3130
    },
    {
      "epoch": 0.2810087955483755,
      "grad_norm": 2.270350933074951,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 3131
    },
    {
      "epoch": 0.28109854604200324,
      "grad_norm": 3.242093563079834,
      "learning_rate": 0.0,
      "loss": 0.6936,
      "step": 3132
    },
    {
      "epoch": 0.28118829653563093,
      "grad_norm": 3.2242095470428467,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 3133
    },
    {
      "epoch": 0.2812780470292587,
      "grad_norm": 2.6647212505340576,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 3134
    },
    {
      "epoch": 0.2813677975228864,
      "grad_norm": 1.9186745882034302,
      "learning_rate": 0.0,
      "loss": 0.7832,
      "step": 3135
    },
    {
      "epoch": 0.28145754801651407,
      "grad_norm": 4.712666988372803,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 3136
    },
    {
      "epoch": 0.2815472985101418,
      "grad_norm": 2.5478599071502686,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 3137
    },
    {
      "epoch": 0.2816370490037695,
      "grad_norm": 1.988351821899414,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 3138
    },
    {
      "epoch": 0.28172679949739726,
      "grad_norm": 3.1278390884399414,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 3139
    },
    {
      "epoch": 0.28181654999102496,
      "grad_norm": 2.680809736251831,
      "learning_rate": 0.0,
      "loss": 0.7523,
      "step": 3140
    },
    {
      "epoch": 0.28190630048465265,
      "grad_norm": 2.2393405437469482,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 3141
    },
    {
      "epoch": 0.2819960509782804,
      "grad_norm": 2.263326644897461,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 3142
    },
    {
      "epoch": 0.2820858014719081,
      "grad_norm": 2.260289192199707,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 3143
    },
    {
      "epoch": 0.2821755519655358,
      "grad_norm": 2.3571856021881104,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 3144
    },
    {
      "epoch": 0.28226530245916354,
      "grad_norm": 3.9311575889587402,
      "learning_rate": 0.0,
      "loss": 0.8061,
      "step": 3145
    },
    {
      "epoch": 0.28235505295279123,
      "grad_norm": 2.8943800926208496,
      "learning_rate": 0.0,
      "loss": 0.7977,
      "step": 3146
    },
    {
      "epoch": 0.282444803446419,
      "grad_norm": 2.354642152786255,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 3147
    },
    {
      "epoch": 0.2825345539400467,
      "grad_norm": 2.1855030059814453,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 3148
    },
    {
      "epoch": 0.28262430443367437,
      "grad_norm": 2.4596872329711914,
      "learning_rate": 0.0,
      "loss": 0.8462,
      "step": 3149
    },
    {
      "epoch": 0.2827140549273021,
      "grad_norm": 2.6265981197357178,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 3150
    },
    {
      "epoch": 0.2828038054209298,
      "grad_norm": 2.5311014652252197,
      "learning_rate": 0.0,
      "loss": 0.6916,
      "step": 3151
    },
    {
      "epoch": 0.2828935559145575,
      "grad_norm": 2.8012161254882812,
      "learning_rate": 0.0,
      "loss": 0.7872,
      "step": 3152
    },
    {
      "epoch": 0.28298330640818525,
      "grad_norm": 3.001336097717285,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 3153
    },
    {
      "epoch": 0.28307305690181295,
      "grad_norm": 3.316885232925415,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 3154
    },
    {
      "epoch": 0.2831628073954407,
      "grad_norm": 3.2401185035705566,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 3155
    },
    {
      "epoch": 0.2832525578890684,
      "grad_norm": 2.358863592147827,
      "learning_rate": 0.0,
      "loss": 0.7825,
      "step": 3156
    },
    {
      "epoch": 0.2833423083826961,
      "grad_norm": 2.679739236831665,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 3157
    },
    {
      "epoch": 0.28343205887632383,
      "grad_norm": 3.504223108291626,
      "learning_rate": 0.0,
      "loss": 0.8035,
      "step": 3158
    },
    {
      "epoch": 0.2835218093699515,
      "grad_norm": 2.471379280090332,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 3159
    },
    {
      "epoch": 0.2836115598635793,
      "grad_norm": 2.54811954498291,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 3160
    },
    {
      "epoch": 0.28370131035720697,
      "grad_norm": 2.969754457473755,
      "learning_rate": 0.0,
      "loss": 0.7849,
      "step": 3161
    },
    {
      "epoch": 0.28379106085083466,
      "grad_norm": 2.4375228881835938,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 3162
    },
    {
      "epoch": 0.2838808113444624,
      "grad_norm": 2.3033626079559326,
      "learning_rate": 0.0,
      "loss": 0.7944,
      "step": 3163
    },
    {
      "epoch": 0.2839705618380901,
      "grad_norm": 2.2522029876708984,
      "learning_rate": 0.0,
      "loss": 0.7845,
      "step": 3164
    },
    {
      "epoch": 0.2840603123317178,
      "grad_norm": 3.2139315605163574,
      "learning_rate": 0.0,
      "loss": 0.6899,
      "step": 3165
    },
    {
      "epoch": 0.28415006282534555,
      "grad_norm": 2.650352716445923,
      "learning_rate": 0.0,
      "loss": 0.7915,
      "step": 3166
    },
    {
      "epoch": 0.28423981331897324,
      "grad_norm": 2.988407850265503,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 3167
    },
    {
      "epoch": 0.284329563812601,
      "grad_norm": 2.6142866611480713,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 3168
    },
    {
      "epoch": 0.2844193143062287,
      "grad_norm": 2.4181838035583496,
      "learning_rate": 0.0,
      "loss": 0.7496,
      "step": 3169
    },
    {
      "epoch": 0.2845090647998564,
      "grad_norm": 3.372699499130249,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 3170
    },
    {
      "epoch": 0.28459881529348413,
      "grad_norm": 2.0091664791107178,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 3171
    },
    {
      "epoch": 0.2846885657871118,
      "grad_norm": 2.7133443355560303,
      "learning_rate": 0.0,
      "loss": 0.7495,
      "step": 3172
    },
    {
      "epoch": 0.2847783162807395,
      "grad_norm": 2.4655821323394775,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 3173
    },
    {
      "epoch": 0.28486806677436727,
      "grad_norm": 7.173502445220947,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 3174
    },
    {
      "epoch": 0.28495781726799496,
      "grad_norm": 2.7168548107147217,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 3175
    },
    {
      "epoch": 0.2850475677616227,
      "grad_norm": 2.3107595443725586,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 3176
    },
    {
      "epoch": 0.2851373182552504,
      "grad_norm": 3.5204617977142334,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 3177
    },
    {
      "epoch": 0.2852270687488781,
      "grad_norm": 2.9219484329223633,
      "learning_rate": 0.0,
      "loss": 0.7964,
      "step": 3178
    },
    {
      "epoch": 0.28531681924250585,
      "grad_norm": 3.5810635089874268,
      "learning_rate": 0.0,
      "loss": 0.8414,
      "step": 3179
    },
    {
      "epoch": 0.28540656973613354,
      "grad_norm": 2.3555092811584473,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 3180
    },
    {
      "epoch": 0.2854963202297613,
      "grad_norm": 2.4183743000030518,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 3181
    },
    {
      "epoch": 0.285586070723389,
      "grad_norm": 3.0294010639190674,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 3182
    },
    {
      "epoch": 0.2856758212170167,
      "grad_norm": 2.8741397857666016,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 3183
    },
    {
      "epoch": 0.28576557171064443,
      "grad_norm": 3.830036163330078,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 3184
    },
    {
      "epoch": 0.2858553222042721,
      "grad_norm": 3.44088077545166,
      "learning_rate": 0.0,
      "loss": 0.7758,
      "step": 3185
    },
    {
      "epoch": 0.2859450726978998,
      "grad_norm": 4.092803478240967,
      "learning_rate": 0.0,
      "loss": 0.7658,
      "step": 3186
    },
    {
      "epoch": 0.28603482319152757,
      "grad_norm": 2.8026788234710693,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 3187
    },
    {
      "epoch": 0.28612457368515526,
      "grad_norm": 3.602047920227051,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 3188
    },
    {
      "epoch": 0.286214324178783,
      "grad_norm": 20.525291442871094,
      "learning_rate": 0.0,
      "loss": 0.7751,
      "step": 3189
    },
    {
      "epoch": 0.2863040746724107,
      "grad_norm": 3.1188414096832275,
      "learning_rate": 0.0,
      "loss": 0.8052,
      "step": 3190
    },
    {
      "epoch": 0.2863938251660384,
      "grad_norm": 3.5607645511627197,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 3191
    },
    {
      "epoch": 0.28648357565966615,
      "grad_norm": 3.7113420963287354,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 3192
    },
    {
      "epoch": 0.28657332615329384,
      "grad_norm": 2.548909902572632,
      "learning_rate": 0.0,
      "loss": 0.7911,
      "step": 3193
    },
    {
      "epoch": 0.28666307664692153,
      "grad_norm": 3.1383135318756104,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 3194
    },
    {
      "epoch": 0.2867528271405493,
      "grad_norm": 2.2179222106933594,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 3195
    },
    {
      "epoch": 0.286842577634177,
      "grad_norm": 5.352376937866211,
      "learning_rate": 0.0,
      "loss": 0.7709,
      "step": 3196
    },
    {
      "epoch": 0.2869323281278047,
      "grad_norm": 6.265097141265869,
      "learning_rate": 0.0,
      "loss": 0.7775,
      "step": 3197
    },
    {
      "epoch": 0.2870220786214324,
      "grad_norm": 5.206716060638428,
      "learning_rate": 0.0,
      "loss": 0.7792,
      "step": 3198
    },
    {
      "epoch": 0.2871118291150601,
      "grad_norm": 2.4884612560272217,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 3199
    },
    {
      "epoch": 0.28720157960868786,
      "grad_norm": 2.664881944656372,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 3200
    },
    {
      "epoch": 0.28729133010231556,
      "grad_norm": 2.51527738571167,
      "learning_rate": 0.0,
      "loss": 0.8038,
      "step": 3201
    },
    {
      "epoch": 0.28738108059594325,
      "grad_norm": 3.5070250034332275,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 3202
    },
    {
      "epoch": 0.287470831089571,
      "grad_norm": 3.552710771560669,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 3203
    },
    {
      "epoch": 0.2875605815831987,
      "grad_norm": 2.4320948123931885,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 3204
    },
    {
      "epoch": 0.28765033207682644,
      "grad_norm": 3.8336429595947266,
      "learning_rate": 0.0,
      "loss": 0.8068,
      "step": 3205
    },
    {
      "epoch": 0.28774008257045414,
      "grad_norm": 2.6038851737976074,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 3206
    },
    {
      "epoch": 0.28782983306408183,
      "grad_norm": 2.3759145736694336,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 3207
    },
    {
      "epoch": 0.2879195835577096,
      "grad_norm": 4.415555000305176,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 3208
    },
    {
      "epoch": 0.2880093340513373,
      "grad_norm": 4.864240646362305,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 3209
    },
    {
      "epoch": 0.288099084544965,
      "grad_norm": 4.093867778778076,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 3210
    },
    {
      "epoch": 0.2881888350385927,
      "grad_norm": 2.7487096786499023,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 3211
    },
    {
      "epoch": 0.2882785855322204,
      "grad_norm": 8.23219108581543,
      "learning_rate": 0.0,
      "loss": 0.825,
      "step": 3212
    },
    {
      "epoch": 0.28836833602584816,
      "grad_norm": 2.3360085487365723,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 3213
    },
    {
      "epoch": 0.28845808651947585,
      "grad_norm": 2.531531810760498,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 3214
    },
    {
      "epoch": 0.28854783701310355,
      "grad_norm": 3.3794994354248047,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 3215
    },
    {
      "epoch": 0.2886375875067313,
      "grad_norm": 2.7746307849884033,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 3216
    },
    {
      "epoch": 0.288727338000359,
      "grad_norm": 2.6083614826202393,
      "learning_rate": 0.0,
      "loss": 0.7074,
      "step": 3217
    },
    {
      "epoch": 0.28881708849398674,
      "grad_norm": 6.042762756347656,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 3218
    },
    {
      "epoch": 0.28890683898761443,
      "grad_norm": 2.8378541469573975,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 3219
    },
    {
      "epoch": 0.2889965894812421,
      "grad_norm": 2.3531064987182617,
      "learning_rate": 0.0,
      "loss": 0.7924,
      "step": 3220
    },
    {
      "epoch": 0.2890863399748699,
      "grad_norm": 4.4082136154174805,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 3221
    },
    {
      "epoch": 0.28917609046849757,
      "grad_norm": 3.4961748123168945,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 3222
    },
    {
      "epoch": 0.28926584096212526,
      "grad_norm": 3.116132974624634,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 3223
    },
    {
      "epoch": 0.289355591455753,
      "grad_norm": 3.987368106842041,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 3224
    },
    {
      "epoch": 0.2894453419493807,
      "grad_norm": 3.111659049987793,
      "learning_rate": 0.0,
      "loss": 0.6819,
      "step": 3225
    },
    {
      "epoch": 0.28953509244300846,
      "grad_norm": 9.48481273651123,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 3226
    },
    {
      "epoch": 0.28962484293663615,
      "grad_norm": 4.149248123168945,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 3227
    },
    {
      "epoch": 0.28971459343026384,
      "grad_norm": 4.043421268463135,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 3228
    },
    {
      "epoch": 0.2898043439238916,
      "grad_norm": 2.380697250366211,
      "learning_rate": 0.0,
      "loss": 0.7663,
      "step": 3229
    },
    {
      "epoch": 0.2898940944175193,
      "grad_norm": 2.9399466514587402,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 3230
    },
    {
      "epoch": 0.28998384491114704,
      "grad_norm": 2.819126844406128,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 3231
    },
    {
      "epoch": 0.29007359540477473,
      "grad_norm": 2.62589955329895,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 3232
    },
    {
      "epoch": 0.2901633458984024,
      "grad_norm": 2.698881149291992,
      "learning_rate": 0.0,
      "loss": 0.7991,
      "step": 3233
    },
    {
      "epoch": 0.2902530963920302,
      "grad_norm": 2.940203905105591,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 3234
    },
    {
      "epoch": 0.29034284688565787,
      "grad_norm": 6.5030436515808105,
      "learning_rate": 0.0,
      "loss": 0.8381,
      "step": 3235
    },
    {
      "epoch": 0.29043259737928556,
      "grad_norm": 2.9299051761627197,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 3236
    },
    {
      "epoch": 0.2905223478729133,
      "grad_norm": 5.907791614532471,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 3237
    },
    {
      "epoch": 0.290612098366541,
      "grad_norm": 2.7444865703582764,
      "learning_rate": 0.0,
      "loss": 0.7825,
      "step": 3238
    },
    {
      "epoch": 0.29070184886016875,
      "grad_norm": 2.23034405708313,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 3239
    },
    {
      "epoch": 0.29079159935379645,
      "grad_norm": 2.368540048599243,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 3240
    },
    {
      "epoch": 0.29088134984742414,
      "grad_norm": 3.6543915271759033,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 3241
    },
    {
      "epoch": 0.2909711003410519,
      "grad_norm": 2.6752328872680664,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 3242
    },
    {
      "epoch": 0.2910608508346796,
      "grad_norm": 2.158547878265381,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 3243
    },
    {
      "epoch": 0.2911506013283073,
      "grad_norm": 3.12668514251709,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 3244
    },
    {
      "epoch": 0.29124035182193503,
      "grad_norm": 3.4913618564605713,
      "learning_rate": 0.0,
      "loss": 0.7491,
      "step": 3245
    },
    {
      "epoch": 0.2913301023155627,
      "grad_norm": 2.050858736038208,
      "learning_rate": 0.0,
      "loss": 0.8009,
      "step": 3246
    },
    {
      "epoch": 0.29141985280919047,
      "grad_norm": 2.794640064239502,
      "learning_rate": 0.0,
      "loss": 0.7959,
      "step": 3247
    },
    {
      "epoch": 0.29150960330281817,
      "grad_norm": 2.39992356300354,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 3248
    },
    {
      "epoch": 0.29159935379644586,
      "grad_norm": 2.9400339126586914,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 3249
    },
    {
      "epoch": 0.2916891042900736,
      "grad_norm": 6.005091190338135,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 3250
    },
    {
      "epoch": 0.2917788547837013,
      "grad_norm": 4.633561134338379,
      "learning_rate": 0.0,
      "loss": 0.66,
      "step": 3251
    },
    {
      "epoch": 0.29186860527732905,
      "grad_norm": 2.3208415508270264,
      "learning_rate": 0.0,
      "loss": 0.8044,
      "step": 3252
    },
    {
      "epoch": 0.29195835577095675,
      "grad_norm": 2.286983013153076,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 3253
    },
    {
      "epoch": 0.29204810626458444,
      "grad_norm": 6.507967472076416,
      "learning_rate": 0.0,
      "loss": 0.808,
      "step": 3254
    },
    {
      "epoch": 0.2921378567582122,
      "grad_norm": 5.793437480926514,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 3255
    },
    {
      "epoch": 0.2922276072518399,
      "grad_norm": 10.422958374023438,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 3256
    },
    {
      "epoch": 0.2923173577454676,
      "grad_norm": 3.4073071479797363,
      "learning_rate": 0.0,
      "loss": 0.7808,
      "step": 3257
    },
    {
      "epoch": 0.2924071082390953,
      "grad_norm": 2.5486044883728027,
      "learning_rate": 0.0,
      "loss": 0.7861,
      "step": 3258
    },
    {
      "epoch": 0.292496858732723,
      "grad_norm": 2.343425750732422,
      "learning_rate": 0.0,
      "loss": 0.7715,
      "step": 3259
    },
    {
      "epoch": 0.29258660922635077,
      "grad_norm": 3.7561638355255127,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 3260
    },
    {
      "epoch": 0.29267635971997846,
      "grad_norm": 6.562443733215332,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 3261
    },
    {
      "epoch": 0.29276611021360616,
      "grad_norm": 3.1866931915283203,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 3262
    },
    {
      "epoch": 0.2928558607072339,
      "grad_norm": 2.845090866088867,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 3263
    },
    {
      "epoch": 0.2929456112008616,
      "grad_norm": 2.6856391429901123,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 3264
    },
    {
      "epoch": 0.2930353616944893,
      "grad_norm": 3.253800630569458,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 3265
    },
    {
      "epoch": 0.29312511218811704,
      "grad_norm": 3.2109668254852295,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 3266
    },
    {
      "epoch": 0.29321486268174474,
      "grad_norm": 2.793928384780884,
      "learning_rate": 0.0,
      "loss": 0.7621,
      "step": 3267
    },
    {
      "epoch": 0.2933046131753725,
      "grad_norm": 3.060110330581665,
      "learning_rate": 0.0,
      "loss": 0.7632,
      "step": 3268
    },
    {
      "epoch": 0.2933943636690002,
      "grad_norm": 2.405719041824341,
      "learning_rate": 0.0,
      "loss": 0.7853,
      "step": 3269
    },
    {
      "epoch": 0.2934841141626279,
      "grad_norm": 2.4290428161621094,
      "learning_rate": 0.0,
      "loss": 0.783,
      "step": 3270
    },
    {
      "epoch": 0.2935738646562556,
      "grad_norm": 2.7468838691711426,
      "learning_rate": 0.0,
      "loss": 0.7831,
      "step": 3271
    },
    {
      "epoch": 0.2936636151498833,
      "grad_norm": 6.251927375793457,
      "learning_rate": 0.0,
      "loss": 0.7365,
      "step": 3272
    },
    {
      "epoch": 0.29375336564351107,
      "grad_norm": 4.004001140594482,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 3273
    },
    {
      "epoch": 0.29384311613713876,
      "grad_norm": 2.9474189281463623,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 3274
    },
    {
      "epoch": 0.29393286663076645,
      "grad_norm": 2.9195375442504883,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 3275
    },
    {
      "epoch": 0.2940226171243942,
      "grad_norm": 4.940101146697998,
      "learning_rate": 0.0,
      "loss": 0.7872,
      "step": 3276
    },
    {
      "epoch": 0.2941123676180219,
      "grad_norm": 2.603558301925659,
      "learning_rate": 0.0,
      "loss": 0.8351,
      "step": 3277
    },
    {
      "epoch": 0.2942021181116496,
      "grad_norm": 8.531452178955078,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 3278
    },
    {
      "epoch": 0.29429186860527734,
      "grad_norm": 3.037644624710083,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 3279
    },
    {
      "epoch": 0.29438161909890503,
      "grad_norm": 2.017843008041382,
      "learning_rate": 0.0,
      "loss": 0.7411,
      "step": 3280
    },
    {
      "epoch": 0.2944713695925328,
      "grad_norm": 2.596099615097046,
      "learning_rate": 0.0,
      "loss": 0.7551,
      "step": 3281
    },
    {
      "epoch": 0.2945611200861605,
      "grad_norm": 8.153803825378418,
      "learning_rate": 0.0,
      "loss": 0.8378,
      "step": 3282
    },
    {
      "epoch": 0.29465087057978817,
      "grad_norm": 3.3486740589141846,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 3283
    },
    {
      "epoch": 0.2947406210734159,
      "grad_norm": 2.6698288917541504,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 3284
    },
    {
      "epoch": 0.2948303715670436,
      "grad_norm": 156.8665313720703,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 3285
    },
    {
      "epoch": 0.2949201220606713,
      "grad_norm": 3.780041217803955,
      "learning_rate": 0.0,
      "loss": 0.7694,
      "step": 3286
    },
    {
      "epoch": 0.29500987255429906,
      "grad_norm": 3.456122398376465,
      "learning_rate": 0.0,
      "loss": 0.79,
      "step": 3287
    },
    {
      "epoch": 0.29509962304792675,
      "grad_norm": 3.0442287921905518,
      "learning_rate": 0.0,
      "loss": 0.7714,
      "step": 3288
    },
    {
      "epoch": 0.2951893735415545,
      "grad_norm": 5.148378849029541,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 3289
    },
    {
      "epoch": 0.2952791240351822,
      "grad_norm": 3.820666551589966,
      "learning_rate": 0.0,
      "loss": 0.7967,
      "step": 3290
    },
    {
      "epoch": 0.2953688745288099,
      "grad_norm": 2.3073031902313232,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 3291
    },
    {
      "epoch": 0.29545862502243764,
      "grad_norm": 3.008422374725342,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 3292
    },
    {
      "epoch": 0.29554837551606533,
      "grad_norm": 3.916043996810913,
      "learning_rate": 0.0,
      "loss": 0.776,
      "step": 3293
    },
    {
      "epoch": 0.2956381260096931,
      "grad_norm": 4.058054447174072,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 3294
    },
    {
      "epoch": 0.2957278765033208,
      "grad_norm": 3.504024028778076,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 3295
    },
    {
      "epoch": 0.29581762699694847,
      "grad_norm": 2.7149598598480225,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 3296
    },
    {
      "epoch": 0.2959073774905762,
      "grad_norm": 2.472548246383667,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 3297
    },
    {
      "epoch": 0.2959971279842039,
      "grad_norm": 2.5331358909606934,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 3298
    },
    {
      "epoch": 0.2960868784778316,
      "grad_norm": 2.4252517223358154,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 3299
    },
    {
      "epoch": 0.29617662897145935,
      "grad_norm": 2.8773245811462402,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 3300
    },
    {
      "epoch": 0.29626637946508705,
      "grad_norm": 2.8338828086853027,
      "learning_rate": 0.0,
      "loss": 0.7019,
      "step": 3301
    },
    {
      "epoch": 0.2963561299587148,
      "grad_norm": 2.3563876152038574,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 3302
    },
    {
      "epoch": 0.2964458804523425,
      "grad_norm": 3.002753973007202,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 3303
    },
    {
      "epoch": 0.2965356309459702,
      "grad_norm": 4.410477638244629,
      "learning_rate": 0.0,
      "loss": 0.8571,
      "step": 3304
    },
    {
      "epoch": 0.29662538143959793,
      "grad_norm": 3.194958448410034,
      "learning_rate": 0.0,
      "loss": 0.832,
      "step": 3305
    },
    {
      "epoch": 0.29671513193322563,
      "grad_norm": 2.9890670776367188,
      "learning_rate": 0.0,
      "loss": 0.7344,
      "step": 3306
    },
    {
      "epoch": 0.2968048824268533,
      "grad_norm": 2.8148529529571533,
      "learning_rate": 0.0,
      "loss": 0.6949,
      "step": 3307
    },
    {
      "epoch": 0.29689463292048107,
      "grad_norm": 2.8459413051605225,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 3308
    },
    {
      "epoch": 0.29698438341410877,
      "grad_norm": 6.046570777893066,
      "learning_rate": 0.0,
      "loss": 0.7976,
      "step": 3309
    },
    {
      "epoch": 0.2970741339077365,
      "grad_norm": 2.687666416168213,
      "learning_rate": 0.0,
      "loss": 0.77,
      "step": 3310
    },
    {
      "epoch": 0.2971638844013642,
      "grad_norm": 2.482041835784912,
      "learning_rate": 0.0,
      "loss": 0.7589,
      "step": 3311
    },
    {
      "epoch": 0.2972536348949919,
      "grad_norm": 2.6822640895843506,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 3312
    },
    {
      "epoch": 0.29734338538861965,
      "grad_norm": 2.4136526584625244,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 3313
    },
    {
      "epoch": 0.29743313588224735,
      "grad_norm": 2.340816020965576,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 3314
    },
    {
      "epoch": 0.29752288637587504,
      "grad_norm": 2.8195323944091797,
      "learning_rate": 0.0,
      "loss": 0.815,
      "step": 3315
    },
    {
      "epoch": 0.2976126368695028,
      "grad_norm": 2.4691498279571533,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 3316
    },
    {
      "epoch": 0.2977023873631305,
      "grad_norm": 2.828012704849243,
      "learning_rate": 0.0,
      "loss": 0.8018,
      "step": 3317
    },
    {
      "epoch": 0.29779213785675823,
      "grad_norm": 4.922490119934082,
      "learning_rate": 0.0,
      "loss": 0.7888,
      "step": 3318
    },
    {
      "epoch": 0.2978818883503859,
      "grad_norm": 2.2524452209472656,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 3319
    },
    {
      "epoch": 0.2979716388440136,
      "grad_norm": 4.271609306335449,
      "learning_rate": 0.0,
      "loss": 0.7596,
      "step": 3320
    },
    {
      "epoch": 0.29806138933764137,
      "grad_norm": 2.4812605381011963,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 3321
    },
    {
      "epoch": 0.29815113983126906,
      "grad_norm": 2.096179246902466,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 3322
    },
    {
      "epoch": 0.2982408903248968,
      "grad_norm": 2.4720592498779297,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 3323
    },
    {
      "epoch": 0.2983306408185245,
      "grad_norm": 3.4945812225341797,
      "learning_rate": 0.0,
      "loss": 0.7842,
      "step": 3324
    },
    {
      "epoch": 0.2984203913121522,
      "grad_norm": 2.7808496952056885,
      "learning_rate": 0.0,
      "loss": 0.7884,
      "step": 3325
    },
    {
      "epoch": 0.29851014180577995,
      "grad_norm": 2.6115052700042725,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 3326
    },
    {
      "epoch": 0.29859989229940764,
      "grad_norm": 10.780652046203613,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 3327
    },
    {
      "epoch": 0.29868964279303534,
      "grad_norm": 3.36793851852417,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 3328
    },
    {
      "epoch": 0.2987793932866631,
      "grad_norm": 2.937420606613159,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 3329
    },
    {
      "epoch": 0.2988691437802908,
      "grad_norm": 2.9895153045654297,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 3330
    },
    {
      "epoch": 0.29895889427391853,
      "grad_norm": 2.3829612731933594,
      "learning_rate": 0.0,
      "loss": 0.7794,
      "step": 3331
    },
    {
      "epoch": 0.2990486447675462,
      "grad_norm": 3.4882712364196777,
      "learning_rate": 0.0,
      "loss": 0.7963,
      "step": 3332
    },
    {
      "epoch": 0.2991383952611739,
      "grad_norm": 2.463926315307617,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 3333
    },
    {
      "epoch": 0.29922814575480167,
      "grad_norm": 2.366399049758911,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 3334
    },
    {
      "epoch": 0.29931789624842936,
      "grad_norm": 39.45337677001953,
      "learning_rate": 0.0,
      "loss": 0.7855,
      "step": 3335
    },
    {
      "epoch": 0.29940764674205705,
      "grad_norm": 4.954985618591309,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 3336
    },
    {
      "epoch": 0.2994973972356848,
      "grad_norm": 3.5789337158203125,
      "learning_rate": 0.0,
      "loss": 0.8532,
      "step": 3337
    },
    {
      "epoch": 0.2995871477293125,
      "grad_norm": 3.3839664459228516,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 3338
    },
    {
      "epoch": 0.29967689822294025,
      "grad_norm": 3.0279226303100586,
      "learning_rate": 0.0,
      "loss": 0.8095,
      "step": 3339
    },
    {
      "epoch": 0.29976664871656794,
      "grad_norm": 9.805441856384277,
      "learning_rate": 0.0,
      "loss": 0.659,
      "step": 3340
    },
    {
      "epoch": 0.29985639921019563,
      "grad_norm": 3.4821951389312744,
      "learning_rate": 0.0,
      "loss": 0.7172,
      "step": 3341
    },
    {
      "epoch": 0.2999461497038234,
      "grad_norm": 3.515862226486206,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 3342
    },
    {
      "epoch": 0.3000359001974511,
      "grad_norm": 3.8450639247894287,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 3343
    },
    {
      "epoch": 0.3001256506910788,
      "grad_norm": 2.445298671722412,
      "learning_rate": 0.0,
      "loss": 0.7937,
      "step": 3344
    },
    {
      "epoch": 0.3002154011847065,
      "grad_norm": 5.379225254058838,
      "learning_rate": 0.0,
      "loss": 0.6916,
      "step": 3345
    },
    {
      "epoch": 0.3003051516783342,
      "grad_norm": 4.114489555358887,
      "learning_rate": 0.0,
      "loss": 0.7773,
      "step": 3346
    },
    {
      "epoch": 0.30039490217196196,
      "grad_norm": 6.925131320953369,
      "learning_rate": 0.0,
      "loss": 0.7789,
      "step": 3347
    },
    {
      "epoch": 0.30048465266558966,
      "grad_norm": 4.271319389343262,
      "learning_rate": 0.0,
      "loss": 0.7698,
      "step": 3348
    },
    {
      "epoch": 0.30057440315921735,
      "grad_norm": 2.292698383331299,
      "learning_rate": 0.0,
      "loss": 0.7821,
      "step": 3349
    },
    {
      "epoch": 0.3006641536528451,
      "grad_norm": 2.4872453212738037,
      "learning_rate": 0.0,
      "loss": 0.7813,
      "step": 3350
    },
    {
      "epoch": 0.3007539041464728,
      "grad_norm": 7.481935977935791,
      "learning_rate": 0.0,
      "loss": 0.7587,
      "step": 3351
    },
    {
      "epoch": 0.30084365464010054,
      "grad_norm": 3.019829273223877,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 3352
    },
    {
      "epoch": 0.30093340513372824,
      "grad_norm": 3.3018839359283447,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 3353
    },
    {
      "epoch": 0.30102315562735593,
      "grad_norm": 4.3864569664001465,
      "learning_rate": 0.0,
      "loss": 0.7702,
      "step": 3354
    },
    {
      "epoch": 0.3011129061209837,
      "grad_norm": 2.7218117713928223,
      "learning_rate": 0.0,
      "loss": 0.7314,
      "step": 3355
    },
    {
      "epoch": 0.3012026566146114,
      "grad_norm": 2.3361825942993164,
      "learning_rate": 0.0,
      "loss": 0.7996,
      "step": 3356
    },
    {
      "epoch": 0.30129240710823907,
      "grad_norm": 2.1207547187805176,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 3357
    },
    {
      "epoch": 0.3013821576018668,
      "grad_norm": 2.2158093452453613,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 3358
    },
    {
      "epoch": 0.3014719080954945,
      "grad_norm": 4.678260326385498,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 3359
    },
    {
      "epoch": 0.30156165858912226,
      "grad_norm": 2.4955990314483643,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 3360
    },
    {
      "epoch": 0.30165140908274996,
      "grad_norm": 6.160301685333252,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 3361
    },
    {
      "epoch": 0.30174115957637765,
      "grad_norm": 2.84566330909729,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 3362
    },
    {
      "epoch": 0.3018309100700054,
      "grad_norm": 2.2709107398986816,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 3363
    },
    {
      "epoch": 0.3019206605636331,
      "grad_norm": 5.906916618347168,
      "learning_rate": 0.0,
      "loss": 0.796,
      "step": 3364
    },
    {
      "epoch": 0.30201041105726084,
      "grad_norm": 2.746537446975708,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 3365
    },
    {
      "epoch": 0.30210016155088854,
      "grad_norm": 10.181690216064453,
      "learning_rate": 0.0,
      "loss": 0.7933,
      "step": 3366
    },
    {
      "epoch": 0.30218991204451623,
      "grad_norm": 2.9898135662078857,
      "learning_rate": 0.0,
      "loss": 0.7776,
      "step": 3367
    },
    {
      "epoch": 0.302279662538144,
      "grad_norm": 2.1418497562408447,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 3368
    },
    {
      "epoch": 0.30236941303177167,
      "grad_norm": 4.235442161560059,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 3369
    },
    {
      "epoch": 0.30245916352539937,
      "grad_norm": 4.2356414794921875,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 3370
    },
    {
      "epoch": 0.3025489140190271,
      "grad_norm": 2.410191535949707,
      "learning_rate": 0.0,
      "loss": 0.8264,
      "step": 3371
    },
    {
      "epoch": 0.3026386645126548,
      "grad_norm": 3.3816030025482178,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 3372
    },
    {
      "epoch": 0.30272841500628256,
      "grad_norm": 4.818566799163818,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 3373
    },
    {
      "epoch": 0.30281816549991025,
      "grad_norm": 3.178201675415039,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 3374
    },
    {
      "epoch": 0.30290791599353795,
      "grad_norm": 2.363670587539673,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 3375
    },
    {
      "epoch": 0.3029976664871657,
      "grad_norm": 2.351930618286133,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 3376
    },
    {
      "epoch": 0.3030874169807934,
      "grad_norm": 7.305445194244385,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 3377
    },
    {
      "epoch": 0.3031771674744211,
      "grad_norm": 2.455270767211914,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 3378
    },
    {
      "epoch": 0.30326691796804883,
      "grad_norm": 3.1829376220703125,
      "learning_rate": 0.0,
      "loss": 0.7749,
      "step": 3379
    },
    {
      "epoch": 0.3033566684616765,
      "grad_norm": 8.023244857788086,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 3380
    },
    {
      "epoch": 0.3034464189553043,
      "grad_norm": 2.084376335144043,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 3381
    },
    {
      "epoch": 0.30353616944893197,
      "grad_norm": 2.720452308654785,
      "learning_rate": 0.0,
      "loss": 0.7432,
      "step": 3382
    },
    {
      "epoch": 0.30362591994255966,
      "grad_norm": 4.5543532371521,
      "learning_rate": 0.0,
      "loss": 0.7673,
      "step": 3383
    },
    {
      "epoch": 0.3037156704361874,
      "grad_norm": 2.3302929401397705,
      "learning_rate": 0.0,
      "loss": 0.7751,
      "step": 3384
    },
    {
      "epoch": 0.3038054209298151,
      "grad_norm": 2.3573920726776123,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 3385
    },
    {
      "epoch": 0.30389517142344286,
      "grad_norm": 2.3133811950683594,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 3386
    },
    {
      "epoch": 0.30398492191707055,
      "grad_norm": 4.408329486846924,
      "learning_rate": 0.0,
      "loss": 0.7697,
      "step": 3387
    },
    {
      "epoch": 0.30407467241069824,
      "grad_norm": 2.721728801727295,
      "learning_rate": 0.0,
      "loss": 0.7566,
      "step": 3388
    },
    {
      "epoch": 0.304164422904326,
      "grad_norm": 2.320600986480713,
      "learning_rate": 0.0,
      "loss": 0.8144,
      "step": 3389
    },
    {
      "epoch": 0.3042541733979537,
      "grad_norm": 2.840991735458374,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 3390
    },
    {
      "epoch": 0.3043439238915814,
      "grad_norm": 2.598659038543701,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 3391
    },
    {
      "epoch": 0.30443367438520913,
      "grad_norm": 2.9733998775482178,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 3392
    },
    {
      "epoch": 0.3045234248788368,
      "grad_norm": 2.511190176010132,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 3393
    },
    {
      "epoch": 0.3046131753724646,
      "grad_norm": 2.3995866775512695,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 3394
    },
    {
      "epoch": 0.30470292586609227,
      "grad_norm": 2.5880978107452393,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 3395
    },
    {
      "epoch": 0.30479267635971996,
      "grad_norm": 2.3065786361694336,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 3396
    },
    {
      "epoch": 0.3048824268533477,
      "grad_norm": 3.3081068992614746,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 3397
    },
    {
      "epoch": 0.3049721773469754,
      "grad_norm": 2.479341983795166,
      "learning_rate": 0.0,
      "loss": 0.7705,
      "step": 3398
    },
    {
      "epoch": 0.3050619278406031,
      "grad_norm": 7.528953552246094,
      "learning_rate": 0.0,
      "loss": 0.7357,
      "step": 3399
    },
    {
      "epoch": 0.30515167833423085,
      "grad_norm": 4.113193988800049,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 3400
    },
    {
      "epoch": 0.30524142882785854,
      "grad_norm": 3.999906063079834,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 3401
    },
    {
      "epoch": 0.3053311793214863,
      "grad_norm": 3.0626182556152344,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 3402
    },
    {
      "epoch": 0.305420929815114,
      "grad_norm": 5.925898551940918,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 3403
    },
    {
      "epoch": 0.3055106803087417,
      "grad_norm": 2.86272931098938,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 3404
    },
    {
      "epoch": 0.3056004308023694,
      "grad_norm": 2.8379414081573486,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 3405
    },
    {
      "epoch": 0.3056901812959971,
      "grad_norm": 3.1908583641052246,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 3406
    },
    {
      "epoch": 0.30577993178962487,
      "grad_norm": 3.1348936557769775,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 3407
    },
    {
      "epoch": 0.30586968228325256,
      "grad_norm": 9.946269035339355,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 3408
    },
    {
      "epoch": 0.30595943277688026,
      "grad_norm": 2.706038236618042,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 3409
    },
    {
      "epoch": 0.306049183270508,
      "grad_norm": 6.0060319900512695,
      "learning_rate": 0.0,
      "loss": 0.7675,
      "step": 3410
    },
    {
      "epoch": 0.3061389337641357,
      "grad_norm": 2.5201714038848877,
      "learning_rate": 0.0,
      "loss": 0.793,
      "step": 3411
    },
    {
      "epoch": 0.3062286842577634,
      "grad_norm": 2.2621753215789795,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 3412
    },
    {
      "epoch": 0.30631843475139114,
      "grad_norm": 2.383518934249878,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 3413
    },
    {
      "epoch": 0.30640818524501884,
      "grad_norm": 3.0302064418792725,
      "learning_rate": 0.0,
      "loss": 0.7621,
      "step": 3414
    },
    {
      "epoch": 0.3064979357386466,
      "grad_norm": 2.4921720027923584,
      "learning_rate": 0.0,
      "loss": 0.7675,
      "step": 3415
    },
    {
      "epoch": 0.3065876862322743,
      "grad_norm": 3.0883569717407227,
      "learning_rate": 0.0,
      "loss": 0.8122,
      "step": 3416
    },
    {
      "epoch": 0.306677436725902,
      "grad_norm": 2.9047818183898926,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 3417
    },
    {
      "epoch": 0.3067671872195297,
      "grad_norm": 3.268554449081421,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 3418
    },
    {
      "epoch": 0.3068569377131574,
      "grad_norm": 3.337970018386841,
      "learning_rate": 0.0,
      "loss": 0.7565,
      "step": 3419
    },
    {
      "epoch": 0.3069466882067851,
      "grad_norm": 4.047457218170166,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 3420
    },
    {
      "epoch": 0.30703643870041286,
      "grad_norm": 34.90424346923828,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 3421
    },
    {
      "epoch": 0.30712618919404056,
      "grad_norm": 3.552442789077759,
      "learning_rate": 0.0,
      "loss": 0.7826,
      "step": 3422
    },
    {
      "epoch": 0.3072159396876683,
      "grad_norm": 3.902230739593506,
      "learning_rate": 0.0,
      "loss": 0.7463,
      "step": 3423
    },
    {
      "epoch": 0.307305690181296,
      "grad_norm": 5.9380574226379395,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 3424
    },
    {
      "epoch": 0.3073954406749237,
      "grad_norm": 3.501776933670044,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 3425
    },
    {
      "epoch": 0.30748519116855144,
      "grad_norm": 3.9462242126464844,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 3426
    },
    {
      "epoch": 0.30757494166217914,
      "grad_norm": 2.480584144592285,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 3427
    },
    {
      "epoch": 0.30766469215580683,
      "grad_norm": 2.557138442993164,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 3428
    },
    {
      "epoch": 0.3077544426494346,
      "grad_norm": 5.095679759979248,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 3429
    },
    {
      "epoch": 0.3078441931430623,
      "grad_norm": 2.934051275253296,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 3430
    },
    {
      "epoch": 0.30793394363669,
      "grad_norm": 2.8175156116485596,
      "learning_rate": 0.0,
      "loss": 0.7709,
      "step": 3431
    },
    {
      "epoch": 0.3080236941303177,
      "grad_norm": 2.908222198486328,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 3432
    },
    {
      "epoch": 0.3081134446239454,
      "grad_norm": 7.4234938621521,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 3433
    },
    {
      "epoch": 0.30820319511757316,
      "grad_norm": 2.991509437561035,
      "learning_rate": 0.0,
      "loss": 0.7001,
      "step": 3434
    },
    {
      "epoch": 0.30829294561120085,
      "grad_norm": 3.1128408908843994,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 3435
    },
    {
      "epoch": 0.3083826961048286,
      "grad_norm": 2.250509262084961,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 3436
    },
    {
      "epoch": 0.3084724465984563,
      "grad_norm": 3.3695662021636963,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 3437
    },
    {
      "epoch": 0.308562197092084,
      "grad_norm": 2.7275710105895996,
      "learning_rate": 0.0,
      "loss": 0.7377,
      "step": 3438
    },
    {
      "epoch": 0.30865194758571174,
      "grad_norm": 2.4241058826446533,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 3439
    },
    {
      "epoch": 0.30874169807933943,
      "grad_norm": 6.974220275878906,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 3440
    },
    {
      "epoch": 0.3088314485729671,
      "grad_norm": 2.413090705871582,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 3441
    },
    {
      "epoch": 0.3089211990665949,
      "grad_norm": 3.5414977073669434,
      "learning_rate": 0.0,
      "loss": 0.6814,
      "step": 3442
    },
    {
      "epoch": 0.30901094956022257,
      "grad_norm": 2.98303484916687,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 3443
    },
    {
      "epoch": 0.3091007000538503,
      "grad_norm": 3.3119900226593018,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 3444
    },
    {
      "epoch": 0.309190450547478,
      "grad_norm": 2.6036901473999023,
      "learning_rate": 0.0,
      "loss": 0.7976,
      "step": 3445
    },
    {
      "epoch": 0.3092802010411057,
      "grad_norm": 2.643230438232422,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 3446
    },
    {
      "epoch": 0.30936995153473346,
      "grad_norm": 2.8085684776306152,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 3447
    },
    {
      "epoch": 0.30945970202836115,
      "grad_norm": 2.865036964416504,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 3448
    },
    {
      "epoch": 0.30954945252198884,
      "grad_norm": 3.010291576385498,
      "learning_rate": 0.0,
      "loss": 0.7774,
      "step": 3449
    },
    {
      "epoch": 0.3096392030156166,
      "grad_norm": 2.097679615020752,
      "learning_rate": 0.0,
      "loss": 0.771,
      "step": 3450
    },
    {
      "epoch": 0.3097289535092443,
      "grad_norm": 3.004213333129883,
      "learning_rate": 0.0,
      "loss": 0.7759,
      "step": 3451
    },
    {
      "epoch": 0.30981870400287204,
      "grad_norm": 3.2803220748901367,
      "learning_rate": 0.0,
      "loss": 0.681,
      "step": 3452
    },
    {
      "epoch": 0.30990845449649973,
      "grad_norm": 2.695162296295166,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 3453
    },
    {
      "epoch": 0.3099982049901274,
      "grad_norm": 2.357545852661133,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 3454
    },
    {
      "epoch": 0.3100879554837552,
      "grad_norm": 3.9851489067077637,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 3455
    },
    {
      "epoch": 0.31017770597738287,
      "grad_norm": 2.5381219387054443,
      "learning_rate": 0.0,
      "loss": 0.7006,
      "step": 3456
    },
    {
      "epoch": 0.3102674564710106,
      "grad_norm": 2.081207752227783,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 3457
    },
    {
      "epoch": 0.3103572069646383,
      "grad_norm": 2.4502782821655273,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 3458
    },
    {
      "epoch": 0.310446957458266,
      "grad_norm": 3.1314518451690674,
      "learning_rate": 0.0,
      "loss": 0.7655,
      "step": 3459
    },
    {
      "epoch": 0.31053670795189375,
      "grad_norm": 2.7667295932769775,
      "learning_rate": 0.0,
      "loss": 0.7798,
      "step": 3460
    },
    {
      "epoch": 0.31062645844552145,
      "grad_norm": 2.772664785385132,
      "learning_rate": 0.0,
      "loss": 0.794,
      "step": 3461
    },
    {
      "epoch": 0.31071620893914914,
      "grad_norm": 2.489006996154785,
      "learning_rate": 0.0,
      "loss": 0.7846,
      "step": 3462
    },
    {
      "epoch": 0.3108059594327769,
      "grad_norm": 2.179539918899536,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 3463
    },
    {
      "epoch": 0.3108957099264046,
      "grad_norm": 2.6203370094299316,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 3464
    },
    {
      "epoch": 0.31098546042003233,
      "grad_norm": 2.575352430343628,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 3465
    },
    {
      "epoch": 0.31107521091366,
      "grad_norm": 2.714920997619629,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 3466
    },
    {
      "epoch": 0.3111649614072877,
      "grad_norm": 3.5518813133239746,
      "learning_rate": 0.0,
      "loss": 0.7684,
      "step": 3467
    },
    {
      "epoch": 0.31125471190091547,
      "grad_norm": 2.5692131519317627,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 3468
    },
    {
      "epoch": 0.31134446239454316,
      "grad_norm": 5.5558085441589355,
      "learning_rate": 0.0,
      "loss": 0.7751,
      "step": 3469
    },
    {
      "epoch": 0.31143421288817086,
      "grad_norm": 3.2983527183532715,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 3470
    },
    {
      "epoch": 0.3115239633817986,
      "grad_norm": 2.4536705017089844,
      "learning_rate": 0.0,
      "loss": 0.8286,
      "step": 3471
    },
    {
      "epoch": 0.3116137138754263,
      "grad_norm": 2.6220226287841797,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 3472
    },
    {
      "epoch": 0.31170346436905405,
      "grad_norm": 2.436413526535034,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 3473
    },
    {
      "epoch": 0.31179321486268174,
      "grad_norm": 2.3421573638916016,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 3474
    },
    {
      "epoch": 0.31188296535630944,
      "grad_norm": 2.6975831985473633,
      "learning_rate": 0.0,
      "loss": 0.7724,
      "step": 3475
    },
    {
      "epoch": 0.3119727158499372,
      "grad_norm": 2.6550474166870117,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 3476
    },
    {
      "epoch": 0.3120624663435649,
      "grad_norm": 2.438326358795166,
      "learning_rate": 0.0,
      "loss": 0.8619,
      "step": 3477
    },
    {
      "epoch": 0.31215221683719263,
      "grad_norm": 3.4289634227752686,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 3478
    },
    {
      "epoch": 0.3122419673308203,
      "grad_norm": 1.8633430004119873,
      "learning_rate": 0.0,
      "loss": 0.7486,
      "step": 3479
    },
    {
      "epoch": 0.312331717824448,
      "grad_norm": 2.658123731613159,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 3480
    },
    {
      "epoch": 0.31242146831807577,
      "grad_norm": 3.607524871826172,
      "learning_rate": 0.0,
      "loss": 0.8065,
      "step": 3481
    },
    {
      "epoch": 0.31251121881170346,
      "grad_norm": 2.887742280960083,
      "learning_rate": 0.0,
      "loss": 0.7672,
      "step": 3482
    },
    {
      "epoch": 0.31260096930533116,
      "grad_norm": 2.364748477935791,
      "learning_rate": 0.0,
      "loss": 0.814,
      "step": 3483
    },
    {
      "epoch": 0.3126907197989589,
      "grad_norm": 3.153057813644409,
      "learning_rate": 0.0,
      "loss": 0.811,
      "step": 3484
    },
    {
      "epoch": 0.3127804702925866,
      "grad_norm": 3.6595935821533203,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 3485
    },
    {
      "epoch": 0.31287022078621435,
      "grad_norm": 2.6370933055877686,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 3486
    },
    {
      "epoch": 0.31295997127984204,
      "grad_norm": 2.6750898361206055,
      "learning_rate": 0.0,
      "loss": 0.764,
      "step": 3487
    },
    {
      "epoch": 0.31304972177346974,
      "grad_norm": 3.9276747703552246,
      "learning_rate": 0.0,
      "loss": 0.7141,
      "step": 3488
    },
    {
      "epoch": 0.3131394722670975,
      "grad_norm": 2.839348554611206,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 3489
    },
    {
      "epoch": 0.3132292227607252,
      "grad_norm": 2.42018723487854,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 3490
    },
    {
      "epoch": 0.3133189732543529,
      "grad_norm": 2.5772693157196045,
      "learning_rate": 0.0,
      "loss": 0.7714,
      "step": 3491
    },
    {
      "epoch": 0.3134087237479806,
      "grad_norm": 3.234478712081909,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 3492
    },
    {
      "epoch": 0.3134984742416083,
      "grad_norm": 2.9086878299713135,
      "learning_rate": 0.0,
      "loss": 0.7869,
      "step": 3493
    },
    {
      "epoch": 0.31358822473523607,
      "grad_norm": 2.7760324478149414,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 3494
    },
    {
      "epoch": 0.31367797522886376,
      "grad_norm": 3.380709409713745,
      "learning_rate": 0.0,
      "loss": 0.7692,
      "step": 3495
    },
    {
      "epoch": 0.31376772572249145,
      "grad_norm": 3.3043744564056396,
      "learning_rate": 0.0,
      "loss": 0.7784,
      "step": 3496
    },
    {
      "epoch": 0.3138574762161192,
      "grad_norm": 2.76690411567688,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 3497
    },
    {
      "epoch": 0.3139472267097469,
      "grad_norm": 2.5959815979003906,
      "learning_rate": 0.0,
      "loss": 0.7828,
      "step": 3498
    },
    {
      "epoch": 0.31403697720337465,
      "grad_norm": 2.854436159133911,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 3499
    },
    {
      "epoch": 0.31412672769700234,
      "grad_norm": 3.0868635177612305,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 3500
    },
    {
      "epoch": 0.31421647819063003,
      "grad_norm": 1.9527231454849243,
      "learning_rate": 0.0,
      "loss": 0.7951,
      "step": 3501
    },
    {
      "epoch": 0.3143062286842578,
      "grad_norm": 2.8922605514526367,
      "learning_rate": 0.0,
      "loss": 0.7453,
      "step": 3502
    },
    {
      "epoch": 0.3143959791778855,
      "grad_norm": 2.399869680404663,
      "learning_rate": 0.0,
      "loss": 0.7747,
      "step": 3503
    },
    {
      "epoch": 0.31448572967151317,
      "grad_norm": 2.459707260131836,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 3504
    },
    {
      "epoch": 0.3145754801651409,
      "grad_norm": 3.2524425983428955,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 3505
    },
    {
      "epoch": 0.3146652306587686,
      "grad_norm": 2.5324010848999023,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 3506
    },
    {
      "epoch": 0.31475498115239636,
      "grad_norm": 3.889664649963379,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 3507
    },
    {
      "epoch": 0.31484473164602406,
      "grad_norm": 6.995002269744873,
      "learning_rate": 0.0,
      "loss": 0.8014,
      "step": 3508
    },
    {
      "epoch": 0.31493448213965175,
      "grad_norm": 4.135746479034424,
      "learning_rate": 0.0,
      "loss": 0.7737,
      "step": 3509
    },
    {
      "epoch": 0.3150242326332795,
      "grad_norm": 2.0820977687835693,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 3510
    },
    {
      "epoch": 0.3151139831269072,
      "grad_norm": 2.197641611099243,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 3511
    },
    {
      "epoch": 0.3152037336205349,
      "grad_norm": 2.268799066543579,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 3512
    },
    {
      "epoch": 0.31529348411416264,
      "grad_norm": 2.658827304840088,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 3513
    },
    {
      "epoch": 0.31538323460779033,
      "grad_norm": 2.908545732498169,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 3514
    },
    {
      "epoch": 0.3154729851014181,
      "grad_norm": 2.613591194152832,
      "learning_rate": 0.0,
      "loss": 0.7697,
      "step": 3515
    },
    {
      "epoch": 0.3155627355950458,
      "grad_norm": 2.270446300506592,
      "learning_rate": 0.0,
      "loss": 0.7733,
      "step": 3516
    },
    {
      "epoch": 0.31565248608867347,
      "grad_norm": 5.918171405792236,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 3517
    },
    {
      "epoch": 0.3157422365823012,
      "grad_norm": 2.694406032562256,
      "learning_rate": 0.0,
      "loss": 0.7734,
      "step": 3518
    },
    {
      "epoch": 0.3158319870759289,
      "grad_norm": 2.4291884899139404,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 3519
    },
    {
      "epoch": 0.31592173756955666,
      "grad_norm": 3.017221450805664,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 3520
    },
    {
      "epoch": 0.31601148806318435,
      "grad_norm": 2.4903998374938965,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 3521
    },
    {
      "epoch": 0.31610123855681205,
      "grad_norm": 2.47900652885437,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 3522
    },
    {
      "epoch": 0.3161909890504398,
      "grad_norm": 2.5709891319274902,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 3523
    },
    {
      "epoch": 0.3162807395440675,
      "grad_norm": 3.107649087905884,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 3524
    },
    {
      "epoch": 0.3163704900376952,
      "grad_norm": 2.291349411010742,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 3525
    },
    {
      "epoch": 0.31646024053132293,
      "grad_norm": 3.3424272537231445,
      "learning_rate": 0.0,
      "loss": 0.7773,
      "step": 3526
    },
    {
      "epoch": 0.3165499910249506,
      "grad_norm": 2.489602565765381,
      "learning_rate": 0.0,
      "loss": 0.7881,
      "step": 3527
    },
    {
      "epoch": 0.3166397415185784,
      "grad_norm": 2.744074821472168,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 3528
    },
    {
      "epoch": 0.31672949201220607,
      "grad_norm": 4.728106498718262,
      "learning_rate": 0.0,
      "loss": 0.7715,
      "step": 3529
    },
    {
      "epoch": 0.31681924250583376,
      "grad_norm": 4.146826267242432,
      "learning_rate": 0.0,
      "loss": 0.752,
      "step": 3530
    },
    {
      "epoch": 0.3169089929994615,
      "grad_norm": 2.8725216388702393,
      "learning_rate": 0.0,
      "loss": 0.7572,
      "step": 3531
    },
    {
      "epoch": 0.3169987434930892,
      "grad_norm": 2.3329994678497314,
      "learning_rate": 0.0,
      "loss": 0.74,
      "step": 3532
    },
    {
      "epoch": 0.3170884939867169,
      "grad_norm": 2.3422324657440186,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 3533
    },
    {
      "epoch": 0.31717824448034465,
      "grad_norm": 3.0072832107543945,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 3534
    },
    {
      "epoch": 0.31726799497397234,
      "grad_norm": 3.0609309673309326,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 3535
    },
    {
      "epoch": 0.3173577454676001,
      "grad_norm": 10.042546272277832,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 3536
    },
    {
      "epoch": 0.3174474959612278,
      "grad_norm": 3.549469232559204,
      "learning_rate": 0.0,
      "loss": 0.7927,
      "step": 3537
    },
    {
      "epoch": 0.3175372464548555,
      "grad_norm": 3.012324333190918,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 3538
    },
    {
      "epoch": 0.31762699694848323,
      "grad_norm": 2.399292469024658,
      "learning_rate": 0.0,
      "loss": 0.6941,
      "step": 3539
    },
    {
      "epoch": 0.3177167474421109,
      "grad_norm": 2.9154105186462402,
      "learning_rate": 0.0,
      "loss": 0.7446,
      "step": 3540
    },
    {
      "epoch": 0.3178064979357386,
      "grad_norm": 3.068695545196533,
      "learning_rate": 0.0,
      "loss": 0.7965,
      "step": 3541
    },
    {
      "epoch": 0.31789624842936637,
      "grad_norm": 2.4318437576293945,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 3542
    },
    {
      "epoch": 0.31798599892299406,
      "grad_norm": 2.3047478199005127,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 3543
    },
    {
      "epoch": 0.3180757494166218,
      "grad_norm": 8.819601058959961,
      "learning_rate": 0.0,
      "loss": 0.7505,
      "step": 3544
    },
    {
      "epoch": 0.3181654999102495,
      "grad_norm": 2.7842116355895996,
      "learning_rate": 0.0,
      "loss": 0.7523,
      "step": 3545
    },
    {
      "epoch": 0.3182552504038772,
      "grad_norm": 2.9113643169403076,
      "learning_rate": 0.0,
      "loss": 0.7906,
      "step": 3546
    },
    {
      "epoch": 0.31834500089750495,
      "grad_norm": 2.308912992477417,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 3547
    },
    {
      "epoch": 0.31843475139113264,
      "grad_norm": 10.990090370178223,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 3548
    },
    {
      "epoch": 0.3185245018847604,
      "grad_norm": 2.810410737991333,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 3549
    },
    {
      "epoch": 0.3186142523783881,
      "grad_norm": 2.4496829509735107,
      "learning_rate": 0.0,
      "loss": 0.7547,
      "step": 3550
    },
    {
      "epoch": 0.3187040028720158,
      "grad_norm": 2.712261199951172,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 3551
    },
    {
      "epoch": 0.31879375336564353,
      "grad_norm": 3.186424970626831,
      "learning_rate": 0.0,
      "loss": 0.7635,
      "step": 3552
    },
    {
      "epoch": 0.3188835038592712,
      "grad_norm": 2.489611864089966,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 3553
    },
    {
      "epoch": 0.3189732543528989,
      "grad_norm": 2.075565814971924,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 3554
    },
    {
      "epoch": 0.31906300484652667,
      "grad_norm": 2.7822136878967285,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 3555
    },
    {
      "epoch": 0.31915275534015436,
      "grad_norm": 3.582958459854126,
      "learning_rate": 0.0,
      "loss": 0.7862,
      "step": 3556
    },
    {
      "epoch": 0.3192425058337821,
      "grad_norm": 4.322345733642578,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 3557
    },
    {
      "epoch": 0.3193322563274098,
      "grad_norm": 2.5169949531555176,
      "learning_rate": 0.0,
      "loss": 0.761,
      "step": 3558
    },
    {
      "epoch": 0.3194220068210375,
      "grad_norm": 6.567562103271484,
      "learning_rate": 0.0,
      "loss": 0.824,
      "step": 3559
    },
    {
      "epoch": 0.31951175731466525,
      "grad_norm": 2.6547932624816895,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 3560
    },
    {
      "epoch": 0.31960150780829294,
      "grad_norm": 3.067861557006836,
      "learning_rate": 0.0,
      "loss": 0.7902,
      "step": 3561
    },
    {
      "epoch": 0.31969125830192063,
      "grad_norm": 3.9389491081237793,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 3562
    },
    {
      "epoch": 0.3197810087955484,
      "grad_norm": 4.329052925109863,
      "learning_rate": 0.0,
      "loss": 0.7631,
      "step": 3563
    },
    {
      "epoch": 0.3198707592891761,
      "grad_norm": 2.850700855255127,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 3564
    },
    {
      "epoch": 0.3199605097828038,
      "grad_norm": 5.770508766174316,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 3565
    },
    {
      "epoch": 0.3200502602764315,
      "grad_norm": 2.1252002716064453,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 3566
    },
    {
      "epoch": 0.3201400107700592,
      "grad_norm": 2.2572078704833984,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 3567
    },
    {
      "epoch": 0.32022976126368696,
      "grad_norm": 3.1054556369781494,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 3568
    },
    {
      "epoch": 0.32031951175731466,
      "grad_norm": 2.697270154953003,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 3569
    },
    {
      "epoch": 0.3204092622509424,
      "grad_norm": 3.441537380218506,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 3570
    },
    {
      "epoch": 0.3204990127445701,
      "grad_norm": 6.146100044250488,
      "learning_rate": 0.0,
      "loss": 0.7491,
      "step": 3571
    },
    {
      "epoch": 0.3205887632381978,
      "grad_norm": 4.979922294616699,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 3572
    },
    {
      "epoch": 0.32067851373182554,
      "grad_norm": 3.9074504375457764,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 3573
    },
    {
      "epoch": 0.32076826422545324,
      "grad_norm": 2.5184898376464844,
      "learning_rate": 0.0,
      "loss": 0.7599,
      "step": 3574
    },
    {
      "epoch": 0.32085801471908093,
      "grad_norm": 2.3483073711395264,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 3575
    },
    {
      "epoch": 0.3209477652127087,
      "grad_norm": 2.7644593715667725,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 3576
    },
    {
      "epoch": 0.3210375157063364,
      "grad_norm": 3.902261257171631,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 3577
    },
    {
      "epoch": 0.3211272661999641,
      "grad_norm": 2.746631383895874,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 3578
    },
    {
      "epoch": 0.3212170166935918,
      "grad_norm": 2.2342417240142822,
      "learning_rate": 0.0,
      "loss": 0.7656,
      "step": 3579
    },
    {
      "epoch": 0.3213067671872195,
      "grad_norm": 5.031963348388672,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 3580
    },
    {
      "epoch": 0.32139651768084726,
      "grad_norm": 2.6664528846740723,
      "learning_rate": 0.0,
      "loss": 0.7808,
      "step": 3581
    },
    {
      "epoch": 0.32148626817447495,
      "grad_norm": 2.5472476482391357,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 3582
    },
    {
      "epoch": 0.32157601866810265,
      "grad_norm": 2.114907741546631,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 3583
    },
    {
      "epoch": 0.3216657691617304,
      "grad_norm": 4.039572715759277,
      "learning_rate": 0.0,
      "loss": 0.7937,
      "step": 3584
    },
    {
      "epoch": 0.3217555196553581,
      "grad_norm": 2.1973633766174316,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 3585
    },
    {
      "epoch": 0.32184527014898584,
      "grad_norm": 4.997013568878174,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 3586
    },
    {
      "epoch": 0.32193502064261353,
      "grad_norm": 2.660247802734375,
      "learning_rate": 0.0,
      "loss": 0.8104,
      "step": 3587
    },
    {
      "epoch": 0.32202477113624123,
      "grad_norm": 5.020429611206055,
      "learning_rate": 0.0,
      "loss": 0.8022,
      "step": 3588
    },
    {
      "epoch": 0.322114521629869,
      "grad_norm": 2.342054843902588,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 3589
    },
    {
      "epoch": 0.32220427212349667,
      "grad_norm": 2.2904396057128906,
      "learning_rate": 0.0,
      "loss": 0.7986,
      "step": 3590
    },
    {
      "epoch": 0.3222940226171244,
      "grad_norm": 2.793574094772339,
      "learning_rate": 0.0,
      "loss": 0.7685,
      "step": 3591
    },
    {
      "epoch": 0.3223837731107521,
      "grad_norm": 2.2025883197784424,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 3592
    },
    {
      "epoch": 0.3224735236043798,
      "grad_norm": 2.3751578330993652,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 3593
    },
    {
      "epoch": 0.32256327409800756,
      "grad_norm": 2.5957324504852295,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 3594
    },
    {
      "epoch": 0.32265302459163525,
      "grad_norm": 2.745973825454712,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 3595
    },
    {
      "epoch": 0.32274277508526295,
      "grad_norm": 3.029388666152954,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 3596
    },
    {
      "epoch": 0.3228325255788907,
      "grad_norm": 2.6829984188079834,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 3597
    },
    {
      "epoch": 0.3229222760725184,
      "grad_norm": 6.769768714904785,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 3598
    },
    {
      "epoch": 0.32301202656614614,
      "grad_norm": 3.6949641704559326,
      "learning_rate": 0.0,
      "loss": 0.817,
      "step": 3599
    },
    {
      "epoch": 0.32310177705977383,
      "grad_norm": 3.024223566055298,
      "learning_rate": 0.0,
      "loss": 0.7889,
      "step": 3600
    },
    {
      "epoch": 0.3231915275534015,
      "grad_norm": 3.6694109439849854,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 3601
    },
    {
      "epoch": 0.3232812780470293,
      "grad_norm": 2.8964974880218506,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 3602
    },
    {
      "epoch": 0.32337102854065697,
      "grad_norm": 3.166834592819214,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 3603
    },
    {
      "epoch": 0.32346077903428466,
      "grad_norm": 2.1283819675445557,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 3604
    },
    {
      "epoch": 0.3235505295279124,
      "grad_norm": 3.3279011249542236,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 3605
    },
    {
      "epoch": 0.3236402800215401,
      "grad_norm": 2.993839740753174,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 3606
    },
    {
      "epoch": 0.32373003051516785,
      "grad_norm": 2.4889814853668213,
      "learning_rate": 0.0,
      "loss": 0.7867,
      "step": 3607
    },
    {
      "epoch": 0.32381978100879555,
      "grad_norm": 3.248995304107666,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 3608
    },
    {
      "epoch": 0.32390953150242324,
      "grad_norm": 4.251481533050537,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 3609
    },
    {
      "epoch": 0.323999281996051,
      "grad_norm": 2.6972997188568115,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 3610
    },
    {
      "epoch": 0.3240890324896787,
      "grad_norm": 2.15704083442688,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 3611
    },
    {
      "epoch": 0.32417878298330643,
      "grad_norm": 4.130475044250488,
      "learning_rate": 0.0,
      "loss": 0.7944,
      "step": 3612
    },
    {
      "epoch": 0.32426853347693413,
      "grad_norm": 3.2495925426483154,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 3613
    },
    {
      "epoch": 0.3243582839705618,
      "grad_norm": 4.10734748840332,
      "learning_rate": 0.0,
      "loss": 0.7488,
      "step": 3614
    },
    {
      "epoch": 0.32444803446418957,
      "grad_norm": 2.039011240005493,
      "learning_rate": 0.0,
      "loss": 0.7972,
      "step": 3615
    },
    {
      "epoch": 0.32453778495781727,
      "grad_norm": 2.695482015609741,
      "learning_rate": 0.0,
      "loss": 0.7663,
      "step": 3616
    },
    {
      "epoch": 0.32462753545144496,
      "grad_norm": 3.913702964782715,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 3617
    },
    {
      "epoch": 0.3247172859450727,
      "grad_norm": 2.187875747680664,
      "learning_rate": 0.0,
      "loss": 0.785,
      "step": 3618
    },
    {
      "epoch": 0.3248070364387004,
      "grad_norm": 2.8393030166625977,
      "learning_rate": 0.0,
      "loss": 0.7357,
      "step": 3619
    },
    {
      "epoch": 0.32489678693232815,
      "grad_norm": 2.3922412395477295,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 3620
    },
    {
      "epoch": 0.32498653742595585,
      "grad_norm": 2.8632590770721436,
      "learning_rate": 0.0,
      "loss": 0.8026,
      "step": 3621
    },
    {
      "epoch": 0.32507628791958354,
      "grad_norm": 2.5060596466064453,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 3622
    },
    {
      "epoch": 0.3251660384132113,
      "grad_norm": 2.6796317100524902,
      "learning_rate": 0.0,
      "loss": 0.719,
      "step": 3623
    },
    {
      "epoch": 0.325255788906839,
      "grad_norm": 2.845477819442749,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 3624
    },
    {
      "epoch": 0.3253455394004667,
      "grad_norm": 2.8407089710235596,
      "learning_rate": 0.0,
      "loss": 0.7956,
      "step": 3625
    },
    {
      "epoch": 0.3254352898940944,
      "grad_norm": 2.479645252227783,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 3626
    },
    {
      "epoch": 0.3255250403877221,
      "grad_norm": 2.4994592666625977,
      "learning_rate": 0.0,
      "loss": 0.8072,
      "step": 3627
    },
    {
      "epoch": 0.32561479088134987,
      "grad_norm": 2.6520166397094727,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 3628
    },
    {
      "epoch": 0.32570454137497756,
      "grad_norm": 1.9966636896133423,
      "learning_rate": 0.0,
      "loss": 0.7867,
      "step": 3629
    },
    {
      "epoch": 0.32579429186860526,
      "grad_norm": 2.6143360137939453,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 3630
    },
    {
      "epoch": 0.325884042362233,
      "grad_norm": 2.073608160018921,
      "learning_rate": 0.0,
      "loss": 0.7738,
      "step": 3631
    },
    {
      "epoch": 0.3259737928558607,
      "grad_norm": 2.2534420490264893,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 3632
    },
    {
      "epoch": 0.32606354334948845,
      "grad_norm": 3.039116859436035,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 3633
    },
    {
      "epoch": 0.32615329384311614,
      "grad_norm": 2.8152952194213867,
      "learning_rate": 0.0,
      "loss": 0.7656,
      "step": 3634
    },
    {
      "epoch": 0.32624304433674384,
      "grad_norm": 2.246623992919922,
      "learning_rate": 0.0,
      "loss": 0.7414,
      "step": 3635
    },
    {
      "epoch": 0.3263327948303716,
      "grad_norm": 2.5539474487304688,
      "learning_rate": 0.0,
      "loss": 0.7724,
      "step": 3636
    },
    {
      "epoch": 0.3264225453239993,
      "grad_norm": 2.285466194152832,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 3637
    },
    {
      "epoch": 0.326512295817627,
      "grad_norm": 3.027921199798584,
      "learning_rate": 0.0,
      "loss": 0.7878,
      "step": 3638
    },
    {
      "epoch": 0.3266020463112547,
      "grad_norm": 3.031008005142212,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 3639
    },
    {
      "epoch": 0.3266917968048824,
      "grad_norm": 2.5542495250701904,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 3640
    },
    {
      "epoch": 0.32678154729851017,
      "grad_norm": 2.598315954208374,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 3641
    },
    {
      "epoch": 0.32687129779213786,
      "grad_norm": 1.9772343635559082,
      "learning_rate": 0.0,
      "loss": 0.8164,
      "step": 3642
    },
    {
      "epoch": 0.32696104828576555,
      "grad_norm": 2.6633548736572266,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 3643
    },
    {
      "epoch": 0.3270507987793933,
      "grad_norm": 2.969058036804199,
      "learning_rate": 0.0,
      "loss": 0.7935,
      "step": 3644
    },
    {
      "epoch": 0.327140549273021,
      "grad_norm": 3.4262306690216064,
      "learning_rate": 0.0,
      "loss": 0.7733,
      "step": 3645
    },
    {
      "epoch": 0.3272302997666487,
      "grad_norm": 2.590071201324463,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 3646
    },
    {
      "epoch": 0.32732005026027644,
      "grad_norm": 2.5874874591827393,
      "learning_rate": 0.0,
      "loss": 0.7076,
      "step": 3647
    },
    {
      "epoch": 0.32740980075390413,
      "grad_norm": 6.122093200683594,
      "learning_rate": 0.0,
      "loss": 0.7888,
      "step": 3648
    },
    {
      "epoch": 0.3274995512475319,
      "grad_norm": 2.4225239753723145,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 3649
    },
    {
      "epoch": 0.3275893017411596,
      "grad_norm": 2.467050552368164,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 3650
    },
    {
      "epoch": 0.32767905223478727,
      "grad_norm": 2.3636980056762695,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 3651
    },
    {
      "epoch": 0.327768802728415,
      "grad_norm": 2.0352256298065186,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 3652
    },
    {
      "epoch": 0.3278585532220427,
      "grad_norm": 13.669916152954102,
      "learning_rate": 0.0,
      "loss": 0.7684,
      "step": 3653
    },
    {
      "epoch": 0.32794830371567046,
      "grad_norm": 2.450991630554199,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 3654
    },
    {
      "epoch": 0.32803805420929816,
      "grad_norm": 2.6418190002441406,
      "learning_rate": 0.0,
      "loss": 0.7579,
      "step": 3655
    },
    {
      "epoch": 0.32812780470292585,
      "grad_norm": 3.1900033950805664,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 3656
    },
    {
      "epoch": 0.3282175551965536,
      "grad_norm": 2.388535261154175,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 3657
    },
    {
      "epoch": 0.3283073056901813,
      "grad_norm": 2.327495813369751,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 3658
    },
    {
      "epoch": 0.328397056183809,
      "grad_norm": 2.2440247535705566,
      "learning_rate": 0.0,
      "loss": 0.818,
      "step": 3659
    },
    {
      "epoch": 0.32848680667743674,
      "grad_norm": 1.9965755939483643,
      "learning_rate": 0.0,
      "loss": 0.7302,
      "step": 3660
    },
    {
      "epoch": 0.32857655717106443,
      "grad_norm": 2.1627848148345947,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 3661
    },
    {
      "epoch": 0.3286663076646922,
      "grad_norm": 2.1566288471221924,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 3662
    },
    {
      "epoch": 0.3287560581583199,
      "grad_norm": 2.016190767288208,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 3663
    },
    {
      "epoch": 0.32884580865194757,
      "grad_norm": 2.3707640171051025,
      "learning_rate": 0.0,
      "loss": 0.7844,
      "step": 3664
    },
    {
      "epoch": 0.3289355591455753,
      "grad_norm": 2.972404956817627,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 3665
    },
    {
      "epoch": 0.329025309639203,
      "grad_norm": 2.8976736068725586,
      "learning_rate": 0.0,
      "loss": 0.8051,
      "step": 3666
    },
    {
      "epoch": 0.3291150601328307,
      "grad_norm": 3.9935824871063232,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 3667
    },
    {
      "epoch": 0.32920481062645846,
      "grad_norm": 2.584465980529785,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 3668
    },
    {
      "epoch": 0.32929456112008615,
      "grad_norm": 2.2836103439331055,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 3669
    },
    {
      "epoch": 0.3293843116137139,
      "grad_norm": 2.2001962661743164,
      "learning_rate": 0.0,
      "loss": 0.7647,
      "step": 3670
    },
    {
      "epoch": 0.3294740621073416,
      "grad_norm": 1.875247597694397,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 3671
    },
    {
      "epoch": 0.3295638126009693,
      "grad_norm": 2.1462621688842773,
      "learning_rate": 0.0,
      "loss": 0.7806,
      "step": 3672
    },
    {
      "epoch": 0.32965356309459704,
      "grad_norm": 2.9012722969055176,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 3673
    },
    {
      "epoch": 0.32974331358822473,
      "grad_norm": 2.714778423309326,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 3674
    },
    {
      "epoch": 0.3298330640818524,
      "grad_norm": 2.1844398975372314,
      "learning_rate": 0.0,
      "loss": 0.7965,
      "step": 3675
    },
    {
      "epoch": 0.32992281457548017,
      "grad_norm": 2.317162275314331,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 3676
    },
    {
      "epoch": 0.33001256506910787,
      "grad_norm": 1.8342828750610352,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 3677
    },
    {
      "epoch": 0.3301023155627356,
      "grad_norm": 2.07392954826355,
      "learning_rate": 0.0,
      "loss": 0.7093,
      "step": 3678
    },
    {
      "epoch": 0.3301920660563633,
      "grad_norm": 2.284430980682373,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 3679
    },
    {
      "epoch": 0.330281816549991,
      "grad_norm": 3.114215612411499,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 3680
    },
    {
      "epoch": 0.33037156704361875,
      "grad_norm": 2.9833250045776367,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 3681
    },
    {
      "epoch": 0.33046131753724645,
      "grad_norm": 2.2562663555145264,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 3682
    },
    {
      "epoch": 0.3305510680308742,
      "grad_norm": 3.1671764850616455,
      "learning_rate": 0.0,
      "loss": 0.794,
      "step": 3683
    },
    {
      "epoch": 0.3306408185245019,
      "grad_norm": 2.6032474040985107,
      "learning_rate": 0.0,
      "loss": 0.8083,
      "step": 3684
    },
    {
      "epoch": 0.3307305690181296,
      "grad_norm": 3.9299426078796387,
      "learning_rate": 0.0,
      "loss": 0.7941,
      "step": 3685
    },
    {
      "epoch": 0.33082031951175733,
      "grad_norm": 2.1243093013763428,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 3686
    },
    {
      "epoch": 0.330910070005385,
      "grad_norm": 2.4097166061401367,
      "learning_rate": 0.0,
      "loss": 0.7431,
      "step": 3687
    },
    {
      "epoch": 0.3309998204990127,
      "grad_norm": 1.8866312503814697,
      "learning_rate": 0.0,
      "loss": 0.7742,
      "step": 3688
    },
    {
      "epoch": 0.33108957099264047,
      "grad_norm": 2.790360689163208,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 3689
    },
    {
      "epoch": 0.33117932148626816,
      "grad_norm": 2.5931124687194824,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 3690
    },
    {
      "epoch": 0.3312690719798959,
      "grad_norm": 1.8992540836334229,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 3691
    },
    {
      "epoch": 0.3313588224735236,
      "grad_norm": 1.8558533191680908,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 3692
    },
    {
      "epoch": 0.3314485729671513,
      "grad_norm": 2.2267098426818848,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 3693
    },
    {
      "epoch": 0.33153832346077905,
      "grad_norm": 2.1259639263153076,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 3694
    },
    {
      "epoch": 0.33162807395440674,
      "grad_norm": 7.737873077392578,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 3695
    },
    {
      "epoch": 0.33171782444803444,
      "grad_norm": 3.344320774078369,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 3696
    },
    {
      "epoch": 0.3318075749416622,
      "grad_norm": 2.846458673477173,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 3697
    },
    {
      "epoch": 0.3318973254352899,
      "grad_norm": 8.686773300170898,
      "learning_rate": 0.0,
      "loss": 0.6599,
      "step": 3698
    },
    {
      "epoch": 0.33198707592891763,
      "grad_norm": 4.17483377456665,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 3699
    },
    {
      "epoch": 0.3320768264225453,
      "grad_norm": 3.512169122695923,
      "learning_rate": 0.0,
      "loss": 0.8117,
      "step": 3700
    },
    {
      "epoch": 0.332166576916173,
      "grad_norm": 2.5851895809173584,
      "learning_rate": 0.0,
      "loss": 0.7804,
      "step": 3701
    },
    {
      "epoch": 0.33225632740980077,
      "grad_norm": 2.649144172668457,
      "learning_rate": 0.0,
      "loss": 0.7955,
      "step": 3702
    },
    {
      "epoch": 0.33234607790342846,
      "grad_norm": 3.5362424850463867,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 3703
    },
    {
      "epoch": 0.3324358283970562,
      "grad_norm": 9.65973949432373,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 3704
    },
    {
      "epoch": 0.3325255788906839,
      "grad_norm": 3.726018190383911,
      "learning_rate": 0.0,
      "loss": 0.7702,
      "step": 3705
    },
    {
      "epoch": 0.3326153293843116,
      "grad_norm": 2.07865834236145,
      "learning_rate": 0.0,
      "loss": 0.7812,
      "step": 3706
    },
    {
      "epoch": 0.33270507987793935,
      "grad_norm": 7.41755485534668,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 3707
    },
    {
      "epoch": 0.33279483037156704,
      "grad_norm": 3.205648422241211,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 3708
    },
    {
      "epoch": 0.33288458086519473,
      "grad_norm": 2.498051881790161,
      "learning_rate": 0.0,
      "loss": 0.7739,
      "step": 3709
    },
    {
      "epoch": 0.3329743313588225,
      "grad_norm": 2.558765411376953,
      "learning_rate": 0.0,
      "loss": 0.7101,
      "step": 3710
    },
    {
      "epoch": 0.3330640818524502,
      "grad_norm": 2.487760066986084,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 3711
    },
    {
      "epoch": 0.3331538323460779,
      "grad_norm": 2.5188345909118652,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 3712
    },
    {
      "epoch": 0.3332435828397056,
      "grad_norm": 2.316378116607666,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 3713
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 3.628774881362915,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 3714
    },
    {
      "epoch": 0.33342308382696106,
      "grad_norm": 3.127673387527466,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 3715
    },
    {
      "epoch": 0.33351283432058876,
      "grad_norm": 3.5044870376586914,
      "learning_rate": 0.0,
      "loss": 0.8141,
      "step": 3716
    },
    {
      "epoch": 0.33360258481421645,
      "grad_norm": 10.807241439819336,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 3717
    },
    {
      "epoch": 0.3336923353078442,
      "grad_norm": 2.829218626022339,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 3718
    },
    {
      "epoch": 0.3337820858014719,
      "grad_norm": 3.5542333126068115,
      "learning_rate": 0.0,
      "loss": 0.7444,
      "step": 3719
    },
    {
      "epoch": 0.33387183629509964,
      "grad_norm": 2.312485456466675,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 3720
    },
    {
      "epoch": 0.33396158678872734,
      "grad_norm": 2.4468939304351807,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 3721
    },
    {
      "epoch": 0.33405133728235503,
      "grad_norm": 3.8463666439056396,
      "learning_rate": 0.0,
      "loss": 0.7811,
      "step": 3722
    },
    {
      "epoch": 0.3341410877759828,
      "grad_norm": 2.938505172729492,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 3723
    },
    {
      "epoch": 0.3342308382696105,
      "grad_norm": 2.871190071105957,
      "learning_rate": 0.0,
      "loss": 0.6866,
      "step": 3724
    },
    {
      "epoch": 0.3343205887632382,
      "grad_norm": 2.405233383178711,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 3725
    },
    {
      "epoch": 0.3344103392568659,
      "grad_norm": 5.506438732147217,
      "learning_rate": 0.0,
      "loss": 0.7204,
      "step": 3726
    },
    {
      "epoch": 0.3345000897504936,
      "grad_norm": 2.981912851333618,
      "learning_rate": 0.0,
      "loss": 0.7143,
      "step": 3727
    },
    {
      "epoch": 0.33458984024412136,
      "grad_norm": 2.246183156967163,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 3728
    },
    {
      "epoch": 0.33467959073774906,
      "grad_norm": 3.4915759563446045,
      "learning_rate": 0.0,
      "loss": 0.8336,
      "step": 3729
    },
    {
      "epoch": 0.33476934123137675,
      "grad_norm": 2.564641237258911,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 3730
    },
    {
      "epoch": 0.3348590917250045,
      "grad_norm": 2.5077803134918213,
      "learning_rate": 0.0,
      "loss": 0.7546,
      "step": 3731
    },
    {
      "epoch": 0.3349488422186322,
      "grad_norm": 2.3594439029693604,
      "learning_rate": 0.0,
      "loss": 0.8115,
      "step": 3732
    },
    {
      "epoch": 0.33503859271225994,
      "grad_norm": 3.3318305015563965,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 3733
    },
    {
      "epoch": 0.33512834320588764,
      "grad_norm": 3.0235157012939453,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 3734
    },
    {
      "epoch": 0.33521809369951533,
      "grad_norm": 5.371245384216309,
      "learning_rate": 0.0,
      "loss": 0.7801,
      "step": 3735
    },
    {
      "epoch": 0.3353078441931431,
      "grad_norm": 3.4123284816741943,
      "learning_rate": 0.0,
      "loss": 0.7061,
      "step": 3736
    },
    {
      "epoch": 0.3353975946867708,
      "grad_norm": 2.578936815261841,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 3737
    },
    {
      "epoch": 0.33548734518039847,
      "grad_norm": 1.8261159658432007,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 3738
    },
    {
      "epoch": 0.3355770956740262,
      "grad_norm": 2.444910764694214,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 3739
    },
    {
      "epoch": 0.3356668461676539,
      "grad_norm": 2.638505220413208,
      "learning_rate": 0.0,
      "loss": 0.7971,
      "step": 3740
    },
    {
      "epoch": 0.33575659666128166,
      "grad_norm": 5.163928031921387,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 3741
    },
    {
      "epoch": 0.33584634715490935,
      "grad_norm": 3.151447296142578,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 3742
    },
    {
      "epoch": 0.33593609764853705,
      "grad_norm": 2.4534778594970703,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 3743
    },
    {
      "epoch": 0.3360258481421648,
      "grad_norm": 2.2678775787353516,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 3744
    },
    {
      "epoch": 0.3361155986357925,
      "grad_norm": 3.1695690155029297,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 3745
    },
    {
      "epoch": 0.33620534912942024,
      "grad_norm": 4.726334095001221,
      "learning_rate": 0.0,
      "loss": 0.7443,
      "step": 3746
    },
    {
      "epoch": 0.33629509962304793,
      "grad_norm": 2.397021532058716,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 3747
    },
    {
      "epoch": 0.3363848501166756,
      "grad_norm": 2.4446780681610107,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 3748
    },
    {
      "epoch": 0.3364746006103034,
      "grad_norm": 5.943296432495117,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 3749
    },
    {
      "epoch": 0.33656435110393107,
      "grad_norm": 2.255293130874634,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 3750
    },
    {
      "epoch": 0.33665410159755876,
      "grad_norm": 1.9378037452697754,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 3751
    },
    {
      "epoch": 0.3367438520911865,
      "grad_norm": 2.3329691886901855,
      "learning_rate": 0.0,
      "loss": 0.7496,
      "step": 3752
    },
    {
      "epoch": 0.3368336025848142,
      "grad_norm": 6.155669212341309,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 3753
    },
    {
      "epoch": 0.33692335307844196,
      "grad_norm": 3.1943624019622803,
      "learning_rate": 0.0,
      "loss": 0.7143,
      "step": 3754
    },
    {
      "epoch": 0.33701310357206965,
      "grad_norm": 2.4273629188537598,
      "learning_rate": 0.0,
      "loss": 0.7818,
      "step": 3755
    },
    {
      "epoch": 0.33710285406569734,
      "grad_norm": 2.0261499881744385,
      "learning_rate": 0.0,
      "loss": 0.754,
      "step": 3756
    },
    {
      "epoch": 0.3371926045593251,
      "grad_norm": 2.6590805053710938,
      "learning_rate": 0.0,
      "loss": 0.7961,
      "step": 3757
    },
    {
      "epoch": 0.3372823550529528,
      "grad_norm": 3.212094783782959,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 3758
    },
    {
      "epoch": 0.3373721055465805,
      "grad_norm": 2.6966140270233154,
      "learning_rate": 0.0,
      "loss": 0.7447,
      "step": 3759
    },
    {
      "epoch": 0.33746185604020823,
      "grad_norm": 17.402896881103516,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 3760
    },
    {
      "epoch": 0.3375516065338359,
      "grad_norm": 4.022500991821289,
      "learning_rate": 0.0,
      "loss": 0.823,
      "step": 3761
    },
    {
      "epoch": 0.3376413570274637,
      "grad_norm": 3.610912799835205,
      "learning_rate": 0.0,
      "loss": 0.7862,
      "step": 3762
    },
    {
      "epoch": 0.33773110752109137,
      "grad_norm": 2.1089439392089844,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 3763
    },
    {
      "epoch": 0.33782085801471906,
      "grad_norm": 2.5666022300720215,
      "learning_rate": 0.0,
      "loss": 0.8257,
      "step": 3764
    },
    {
      "epoch": 0.3379106085083468,
      "grad_norm": 2.6036782264709473,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 3765
    },
    {
      "epoch": 0.3380003590019745,
      "grad_norm": 2.5003502368927,
      "learning_rate": 0.0,
      "loss": 0.8332,
      "step": 3766
    },
    {
      "epoch": 0.33809010949560225,
      "grad_norm": 2.633589029312134,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 3767
    },
    {
      "epoch": 0.33817985998922995,
      "grad_norm": 2.6364238262176514,
      "learning_rate": 0.0,
      "loss": 0.7922,
      "step": 3768
    },
    {
      "epoch": 0.33826961048285764,
      "grad_norm": 2.9552369117736816,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 3769
    },
    {
      "epoch": 0.3383593609764854,
      "grad_norm": 2.2922170162200928,
      "learning_rate": 0.0,
      "loss": 0.743,
      "step": 3770
    },
    {
      "epoch": 0.3384491114701131,
      "grad_norm": 2.8691422939300537,
      "learning_rate": 0.0,
      "loss": 0.6886,
      "step": 3771
    },
    {
      "epoch": 0.3385388619637408,
      "grad_norm": 2.4760866165161133,
      "learning_rate": 0.0,
      "loss": 0.7659,
      "step": 3772
    },
    {
      "epoch": 0.3386286124573685,
      "grad_norm": 2.7605512142181396,
      "learning_rate": 0.0,
      "loss": 0.8065,
      "step": 3773
    },
    {
      "epoch": 0.3387183629509962,
      "grad_norm": 3.3585216999053955,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 3774
    },
    {
      "epoch": 0.33880811344462397,
      "grad_norm": 2.6474900245666504,
      "learning_rate": 0.0,
      "loss": 0.7865,
      "step": 3775
    },
    {
      "epoch": 0.33889786393825166,
      "grad_norm": 2.349158763885498,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 3776
    },
    {
      "epoch": 0.33898761443187936,
      "grad_norm": 2.25651478767395,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 3777
    },
    {
      "epoch": 0.3390773649255071,
      "grad_norm": 3.8893301486968994,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 3778
    },
    {
      "epoch": 0.3391671154191348,
      "grad_norm": 3.5470776557922363,
      "learning_rate": 0.0,
      "loss": 0.6765,
      "step": 3779
    },
    {
      "epoch": 0.3392568659127625,
      "grad_norm": 3.7677693367004395,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 3780
    },
    {
      "epoch": 0.33934661640639024,
      "grad_norm": 3.3774282932281494,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 3781
    },
    {
      "epoch": 0.33943636690001794,
      "grad_norm": 2.1395628452301025,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 3782
    },
    {
      "epoch": 0.3395261173936457,
      "grad_norm": 2.052700996398926,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 3783
    },
    {
      "epoch": 0.3396158678872734,
      "grad_norm": 3.087742328643799,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 3784
    },
    {
      "epoch": 0.3397056183809011,
      "grad_norm": 2.6382453441619873,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 3785
    },
    {
      "epoch": 0.3397953688745288,
      "grad_norm": 2.7291781902313232,
      "learning_rate": 0.0,
      "loss": 0.8065,
      "step": 3786
    },
    {
      "epoch": 0.3398851193681565,
      "grad_norm": 2.6997010707855225,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 3787
    },
    {
      "epoch": 0.3399748698617842,
      "grad_norm": 3.344409227371216,
      "learning_rate": 0.0,
      "loss": 0.7846,
      "step": 3788
    },
    {
      "epoch": 0.34006462035541196,
      "grad_norm": 4.86781120300293,
      "learning_rate": 0.0,
      "loss": 0.7527,
      "step": 3789
    },
    {
      "epoch": 0.34015437084903966,
      "grad_norm": 2.744920492172241,
      "learning_rate": 0.0,
      "loss": 0.7961,
      "step": 3790
    },
    {
      "epoch": 0.3402441213426674,
      "grad_norm": 2.3539576530456543,
      "learning_rate": 0.0,
      "loss": 0.7128,
      "step": 3791
    },
    {
      "epoch": 0.3403338718362951,
      "grad_norm": 1.8350707292556763,
      "learning_rate": 0.0,
      "loss": 0.7886,
      "step": 3792
    },
    {
      "epoch": 0.3404236223299228,
      "grad_norm": 2.654176712036133,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 3793
    },
    {
      "epoch": 0.34051337282355054,
      "grad_norm": 2.390258312225342,
      "learning_rate": 0.0,
      "loss": 0.8523,
      "step": 3794
    },
    {
      "epoch": 0.34060312331717824,
      "grad_norm": 2.47281813621521,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 3795
    },
    {
      "epoch": 0.340692873810806,
      "grad_norm": 2.9771461486816406,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 3796
    },
    {
      "epoch": 0.3407826243044337,
      "grad_norm": 2.799539804458618,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 3797
    },
    {
      "epoch": 0.3408723747980614,
      "grad_norm": 3.28277325630188,
      "learning_rate": 0.0,
      "loss": 0.7737,
      "step": 3798
    },
    {
      "epoch": 0.3409621252916891,
      "grad_norm": 16.993085861206055,
      "learning_rate": 0.0,
      "loss": 0.7982,
      "step": 3799
    },
    {
      "epoch": 0.3410518757853168,
      "grad_norm": 4.766755104064941,
      "learning_rate": 0.0,
      "loss": 0.7624,
      "step": 3800
    },
    {
      "epoch": 0.3411416262789445,
      "grad_norm": 3.29343843460083,
      "learning_rate": 0.0,
      "loss": 0.7825,
      "step": 3801
    },
    {
      "epoch": 0.34123137677257226,
      "grad_norm": 2.7121424674987793,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 3802
    },
    {
      "epoch": 0.34132112726619995,
      "grad_norm": 2.753265380859375,
      "learning_rate": 0.0,
      "loss": 0.7463,
      "step": 3803
    },
    {
      "epoch": 0.3414108777598277,
      "grad_norm": 10.541501998901367,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 3804
    },
    {
      "epoch": 0.3415006282534554,
      "grad_norm": 2.507638931274414,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 3805
    },
    {
      "epoch": 0.3415903787470831,
      "grad_norm": 2.790059804916382,
      "learning_rate": 0.0,
      "loss": 0.7381,
      "step": 3806
    },
    {
      "epoch": 0.34168012924071084,
      "grad_norm": 2.6095614433288574,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 3807
    },
    {
      "epoch": 0.34176987973433853,
      "grad_norm": 2.2348480224609375,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 3808
    },
    {
      "epoch": 0.3418596302279662,
      "grad_norm": 2.5915234088897705,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 3809
    },
    {
      "epoch": 0.341949380721594,
      "grad_norm": 2.2082643508911133,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 3810
    },
    {
      "epoch": 0.34203913121522167,
      "grad_norm": 2.261639356613159,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 3811
    },
    {
      "epoch": 0.3421288817088494,
      "grad_norm": 2.884423017501831,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 3812
    },
    {
      "epoch": 0.3422186322024771,
      "grad_norm": 3.4297337532043457,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 3813
    },
    {
      "epoch": 0.3423083826961048,
      "grad_norm": 2.3693060874938965,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 3814
    },
    {
      "epoch": 0.34239813318973256,
      "grad_norm": 2.6526029109954834,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 3815
    },
    {
      "epoch": 0.34248788368336025,
      "grad_norm": 2.28245210647583,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 3816
    },
    {
      "epoch": 0.342577634176988,
      "grad_norm": 2.375445604324341,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 3817
    },
    {
      "epoch": 0.3426673846706157,
      "grad_norm": 2.5223214626312256,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 3818
    },
    {
      "epoch": 0.3427571351642434,
      "grad_norm": 2.5352394580841064,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 3819
    },
    {
      "epoch": 0.34284688565787114,
      "grad_norm": 3.153636932373047,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 3820
    },
    {
      "epoch": 0.34293663615149883,
      "grad_norm": 2.460136651992798,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 3821
    },
    {
      "epoch": 0.3430263866451265,
      "grad_norm": 2.2751755714416504,
      "learning_rate": 0.0,
      "loss": 0.7534,
      "step": 3822
    },
    {
      "epoch": 0.3431161371387543,
      "grad_norm": 4.188752174377441,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 3823
    },
    {
      "epoch": 0.34320588763238197,
      "grad_norm": 3.2708041667938232,
      "learning_rate": 0.0,
      "loss": 0.768,
      "step": 3824
    },
    {
      "epoch": 0.3432956381260097,
      "grad_norm": 2.473100423812866,
      "learning_rate": 0.0,
      "loss": 0.8051,
      "step": 3825
    },
    {
      "epoch": 0.3433853886196374,
      "grad_norm": 3.288597583770752,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 3826
    },
    {
      "epoch": 0.3434751391132651,
      "grad_norm": 3.508375406265259,
      "learning_rate": 0.0,
      "loss": 0.7486,
      "step": 3827
    },
    {
      "epoch": 0.34356488960689285,
      "grad_norm": 3.579336404800415,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 3828
    },
    {
      "epoch": 0.34365464010052055,
      "grad_norm": 2.186129093170166,
      "learning_rate": 0.0,
      "loss": 0.8454,
      "step": 3829
    },
    {
      "epoch": 0.34374439059414824,
      "grad_norm": 2.1273179054260254,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 3830
    },
    {
      "epoch": 0.343834141087776,
      "grad_norm": 2.6397817134857178,
      "learning_rate": 0.0,
      "loss": 0.7886,
      "step": 3831
    },
    {
      "epoch": 0.3439238915814037,
      "grad_norm": 3.05181622505188,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 3832
    },
    {
      "epoch": 0.34401364207503143,
      "grad_norm": 2.867161989212036,
      "learning_rate": 0.0,
      "loss": 0.76,
      "step": 3833
    },
    {
      "epoch": 0.3441033925686591,
      "grad_norm": 2.8402562141418457,
      "learning_rate": 0.0,
      "loss": 0.6839,
      "step": 3834
    },
    {
      "epoch": 0.3441931430622868,
      "grad_norm": 2.795520305633545,
      "learning_rate": 0.0,
      "loss": 0.8353,
      "step": 3835
    },
    {
      "epoch": 0.34428289355591457,
      "grad_norm": 4.485491752624512,
      "learning_rate": 0.0,
      "loss": 0.7663,
      "step": 3836
    },
    {
      "epoch": 0.34437264404954226,
      "grad_norm": 3.577268600463867,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 3837
    },
    {
      "epoch": 0.34446239454317,
      "grad_norm": 3.9089887142181396,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 3838
    },
    {
      "epoch": 0.3445521450367977,
      "grad_norm": 2.342320442199707,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 3839
    },
    {
      "epoch": 0.3446418955304254,
      "grad_norm": 2.2552571296691895,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 3840
    },
    {
      "epoch": 0.34473164602405315,
      "grad_norm": 5.872867584228516,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 3841
    },
    {
      "epoch": 0.34482139651768084,
      "grad_norm": 4.0185747146606445,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 3842
    },
    {
      "epoch": 0.34491114701130854,
      "grad_norm": 11.198029518127441,
      "learning_rate": 0.0,
      "loss": 0.7152,
      "step": 3843
    },
    {
      "epoch": 0.3450008975049363,
      "grad_norm": 11.728644371032715,
      "learning_rate": 0.0,
      "loss": 0.7825,
      "step": 3844
    },
    {
      "epoch": 0.345090647998564,
      "grad_norm": 2.828930377960205,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 3845
    },
    {
      "epoch": 0.34518039849219173,
      "grad_norm": 2.1912548542022705,
      "learning_rate": 0.0,
      "loss": 0.8087,
      "step": 3846
    },
    {
      "epoch": 0.3452701489858194,
      "grad_norm": 2.5441293716430664,
      "learning_rate": 0.0,
      "loss": 0.802,
      "step": 3847
    },
    {
      "epoch": 0.3453598994794471,
      "grad_norm": 2.460350275039673,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 3848
    },
    {
      "epoch": 0.34544964997307487,
      "grad_norm": 2.1569671630859375,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 3849
    },
    {
      "epoch": 0.34553940046670256,
      "grad_norm": 6.664294719696045,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 3850
    },
    {
      "epoch": 0.34562915096033026,
      "grad_norm": 3.4589383602142334,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 3851
    },
    {
      "epoch": 0.345718901453958,
      "grad_norm": 2.5171172618865967,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 3852
    },
    {
      "epoch": 0.3458086519475857,
      "grad_norm": 4.037029266357422,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 3853
    },
    {
      "epoch": 0.34589840244121345,
      "grad_norm": 5.592763423919678,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 3854
    },
    {
      "epoch": 0.34598815293484114,
      "grad_norm": 3.084948778152466,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 3855
    },
    {
      "epoch": 0.34607790342846884,
      "grad_norm": 3.3226394653320312,
      "learning_rate": 0.0,
      "loss": 0.7782,
      "step": 3856
    },
    {
      "epoch": 0.3461676539220966,
      "grad_norm": 2.313924551010132,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 3857
    },
    {
      "epoch": 0.3462574044157243,
      "grad_norm": 2.7761199474334717,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 3858
    },
    {
      "epoch": 0.34634715490935203,
      "grad_norm": 2.3900063037872314,
      "learning_rate": 0.0,
      "loss": 0.7943,
      "step": 3859
    },
    {
      "epoch": 0.3464369054029797,
      "grad_norm": 30.313615798950195,
      "learning_rate": 0.0,
      "loss": 0.7443,
      "step": 3860
    },
    {
      "epoch": 0.3465266558966074,
      "grad_norm": 2.612368106842041,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 3861
    },
    {
      "epoch": 0.34661640639023517,
      "grad_norm": 2.423156261444092,
      "learning_rate": 0.0,
      "loss": 0.7446,
      "step": 3862
    },
    {
      "epoch": 0.34670615688386286,
      "grad_norm": 2.3584890365600586,
      "learning_rate": 0.0,
      "loss": 0.7815,
      "step": 3863
    },
    {
      "epoch": 0.34679590737749055,
      "grad_norm": 2.6010210514068604,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 3864
    },
    {
      "epoch": 0.3468856578711183,
      "grad_norm": 12.719188690185547,
      "learning_rate": 0.0,
      "loss": 0.7856,
      "step": 3865
    },
    {
      "epoch": 0.346975408364746,
      "grad_norm": 2.6062984466552734,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 3866
    },
    {
      "epoch": 0.34706515885837375,
      "grad_norm": 3.070509672164917,
      "learning_rate": 0.0,
      "loss": 0.8189,
      "step": 3867
    },
    {
      "epoch": 0.34715490935200144,
      "grad_norm": 3.3608503341674805,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 3868
    },
    {
      "epoch": 0.34724465984562913,
      "grad_norm": 3.294468402862549,
      "learning_rate": 0.0,
      "loss": 0.7782,
      "step": 3869
    },
    {
      "epoch": 0.3473344103392569,
      "grad_norm": 2.574526786804199,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 3870
    },
    {
      "epoch": 0.3474241608328846,
      "grad_norm": 2.2792866230010986,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 3871
    },
    {
      "epoch": 0.34751391132651227,
      "grad_norm": 3.0295515060424805,
      "learning_rate": 0.0,
      "loss": 0.7694,
      "step": 3872
    },
    {
      "epoch": 0.34760366182014,
      "grad_norm": 2.7644691467285156,
      "learning_rate": 0.0,
      "loss": 0.6966,
      "step": 3873
    },
    {
      "epoch": 0.3476934123137677,
      "grad_norm": 2.1823551654815674,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 3874
    },
    {
      "epoch": 0.34778316280739546,
      "grad_norm": 2.6550071239471436,
      "learning_rate": 0.0,
      "loss": 0.8073,
      "step": 3875
    },
    {
      "epoch": 0.34787291330102316,
      "grad_norm": 3.4107959270477295,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 3876
    },
    {
      "epoch": 0.34796266379465085,
      "grad_norm": 2.8057587146759033,
      "learning_rate": 0.0,
      "loss": 0.7762,
      "step": 3877
    },
    {
      "epoch": 0.3480524142882786,
      "grad_norm": 2.8614845275878906,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 3878
    },
    {
      "epoch": 0.3481421647819063,
      "grad_norm": 3.143876314163208,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 3879
    },
    {
      "epoch": 0.34823191527553404,
      "grad_norm": 4.371637344360352,
      "learning_rate": 0.0,
      "loss": 0.758,
      "step": 3880
    },
    {
      "epoch": 0.34832166576916174,
      "grad_norm": 2.631239652633667,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 3881
    },
    {
      "epoch": 0.34841141626278943,
      "grad_norm": 3.307485818862915,
      "learning_rate": 0.0,
      "loss": 0.7719,
      "step": 3882
    },
    {
      "epoch": 0.3485011667564172,
      "grad_norm": 3.600123405456543,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 3883
    },
    {
      "epoch": 0.3485909172500449,
      "grad_norm": 2.723226547241211,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 3884
    },
    {
      "epoch": 0.34868066774367257,
      "grad_norm": 5.562067031860352,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 3885
    },
    {
      "epoch": 0.3487704182373003,
      "grad_norm": 5.284936904907227,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 3886
    },
    {
      "epoch": 0.348860168730928,
      "grad_norm": 11.82543659210205,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 3887
    },
    {
      "epoch": 0.34894991922455576,
      "grad_norm": 3.2339987754821777,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 3888
    },
    {
      "epoch": 0.34903966971818345,
      "grad_norm": 2.6041030883789062,
      "learning_rate": 0.0,
      "loss": 0.7851,
      "step": 3889
    },
    {
      "epoch": 0.34912942021181115,
      "grad_norm": 5.518514156341553,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 3890
    },
    {
      "epoch": 0.3492191707054389,
      "grad_norm": 2.9822819232940674,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 3891
    },
    {
      "epoch": 0.3493089211990666,
      "grad_norm": 2.9588632583618164,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 3892
    },
    {
      "epoch": 0.3493986716926943,
      "grad_norm": 2.803727388381958,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 3893
    },
    {
      "epoch": 0.34948842218632203,
      "grad_norm": 2.386671781539917,
      "learning_rate": 0.0,
      "loss": 0.7948,
      "step": 3894
    },
    {
      "epoch": 0.34957817267994973,
      "grad_norm": 5.54005241394043,
      "learning_rate": 0.0,
      "loss": 0.7357,
      "step": 3895
    },
    {
      "epoch": 0.3496679231735775,
      "grad_norm": 3.136922597885132,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 3896
    },
    {
      "epoch": 0.34975767366720517,
      "grad_norm": 2.685089588165283,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 3897
    },
    {
      "epoch": 0.34984742416083286,
      "grad_norm": 3.8935914039611816,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 3898
    },
    {
      "epoch": 0.3499371746544606,
      "grad_norm": 2.515160322189331,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 3899
    },
    {
      "epoch": 0.3500269251480883,
      "grad_norm": 2.4591104984283447,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 3900
    },
    {
      "epoch": 0.350116675641716,
      "grad_norm": 4.249281406402588,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 3901
    },
    {
      "epoch": 0.35020642613534375,
      "grad_norm": 2.6792101860046387,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 3902
    },
    {
      "epoch": 0.35029617662897145,
      "grad_norm": 3.0178022384643555,
      "learning_rate": 0.0,
      "loss": 0.7673,
      "step": 3903
    },
    {
      "epoch": 0.3503859271225992,
      "grad_norm": 3.0858466625213623,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 3904
    },
    {
      "epoch": 0.3504756776162269,
      "grad_norm": 4.388747215270996,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 3905
    },
    {
      "epoch": 0.3505654281098546,
      "grad_norm": 2.4331164360046387,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 3906
    },
    {
      "epoch": 0.35065517860348233,
      "grad_norm": 2.406078815460205,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 3907
    },
    {
      "epoch": 0.35074492909711,
      "grad_norm": 3.2150285243988037,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 3908
    },
    {
      "epoch": 0.3508346795907378,
      "grad_norm": 2.4144182205200195,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 3909
    },
    {
      "epoch": 0.35092443008436547,
      "grad_norm": 2.7829103469848633,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 3910
    },
    {
      "epoch": 0.35101418057799316,
      "grad_norm": 3.400871515274048,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 3911
    },
    {
      "epoch": 0.3511039310716209,
      "grad_norm": 2.661442756652832,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 3912
    },
    {
      "epoch": 0.3511936815652486,
      "grad_norm": 2.9481372833251953,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 3913
    },
    {
      "epoch": 0.3512834320588763,
      "grad_norm": 2.2862610816955566,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 3914
    },
    {
      "epoch": 0.35137318255250405,
      "grad_norm": 2.9264683723449707,
      "learning_rate": 0.0,
      "loss": 0.7691,
      "step": 3915
    },
    {
      "epoch": 0.35146293304613174,
      "grad_norm": 3.19697904586792,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 3916
    },
    {
      "epoch": 0.3515526835397595,
      "grad_norm": 2.8699920177459717,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 3917
    },
    {
      "epoch": 0.3516424340333872,
      "grad_norm": 3.9376518726348877,
      "learning_rate": 0.0,
      "loss": 0.7336,
      "step": 3918
    },
    {
      "epoch": 0.3517321845270149,
      "grad_norm": 3.889991044998169,
      "learning_rate": 0.0,
      "loss": 0.7624,
      "step": 3919
    },
    {
      "epoch": 0.35182193502064263,
      "grad_norm": 2.645228862762451,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 3920
    },
    {
      "epoch": 0.3519116855142703,
      "grad_norm": 2.354391574859619,
      "learning_rate": 0.0,
      "loss": 0.7375,
      "step": 3921
    },
    {
      "epoch": 0.352001436007898,
      "grad_norm": 5.0260210037231445,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 3922
    },
    {
      "epoch": 0.35209118650152577,
      "grad_norm": 2.9737236499786377,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 3923
    },
    {
      "epoch": 0.35218093699515346,
      "grad_norm": 2.844294309616089,
      "learning_rate": 0.0,
      "loss": 0.7542,
      "step": 3924
    },
    {
      "epoch": 0.3522706874887812,
      "grad_norm": 14.113019943237305,
      "learning_rate": 0.0,
      "loss": 0.7848,
      "step": 3925
    },
    {
      "epoch": 0.3523604379824089,
      "grad_norm": 8.542920112609863,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 3926
    },
    {
      "epoch": 0.3524501884760366,
      "grad_norm": 2.5644428730010986,
      "learning_rate": 0.0,
      "loss": 0.7986,
      "step": 3927
    },
    {
      "epoch": 0.35253993896966435,
      "grad_norm": 2.247605562210083,
      "learning_rate": 0.0,
      "loss": 0.7683,
      "step": 3928
    },
    {
      "epoch": 0.35262968946329204,
      "grad_norm": 2.4800219535827637,
      "learning_rate": 0.0,
      "loss": 0.6722,
      "step": 3929
    },
    {
      "epoch": 0.3527194399569198,
      "grad_norm": 4.16140604019165,
      "learning_rate": 0.0,
      "loss": 0.7882,
      "step": 3930
    },
    {
      "epoch": 0.3528091904505475,
      "grad_norm": 6.227364540100098,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 3931
    },
    {
      "epoch": 0.3528989409441752,
      "grad_norm": 2.986375093460083,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 3932
    },
    {
      "epoch": 0.3529886914378029,
      "grad_norm": 3.1623592376708984,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 3933
    },
    {
      "epoch": 0.3530784419314306,
      "grad_norm": 2.989670753479004,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 3934
    },
    {
      "epoch": 0.3531681924250583,
      "grad_norm": 2.887967824935913,
      "learning_rate": 0.0,
      "loss": 0.724,
      "step": 3935
    },
    {
      "epoch": 0.35325794291868606,
      "grad_norm": 2.952007532119751,
      "learning_rate": 0.0,
      "loss": 0.688,
      "step": 3936
    },
    {
      "epoch": 0.35334769341231376,
      "grad_norm": 3.350532293319702,
      "learning_rate": 0.0,
      "loss": 0.802,
      "step": 3937
    },
    {
      "epoch": 0.3534374439059415,
      "grad_norm": 2.4319496154785156,
      "learning_rate": 0.0,
      "loss": 0.8003,
      "step": 3938
    },
    {
      "epoch": 0.3535271943995692,
      "grad_norm": 2.7821786403656006,
      "learning_rate": 0.0,
      "loss": 0.7632,
      "step": 3939
    },
    {
      "epoch": 0.3536169448931969,
      "grad_norm": 2.373610258102417,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 3940
    },
    {
      "epoch": 0.35370669538682464,
      "grad_norm": 3.076857566833496,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 3941
    },
    {
      "epoch": 0.35379644588045234,
      "grad_norm": 3.005786180496216,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 3942
    },
    {
      "epoch": 0.35388619637408003,
      "grad_norm": 4.578231334686279,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 3943
    },
    {
      "epoch": 0.3539759468677078,
      "grad_norm": 2.4722793102264404,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 3944
    },
    {
      "epoch": 0.3540656973613355,
      "grad_norm": 8.002705574035645,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 3945
    },
    {
      "epoch": 0.3541554478549632,
      "grad_norm": 2.7164695262908936,
      "learning_rate": 0.0,
      "loss": 0.7626,
      "step": 3946
    },
    {
      "epoch": 0.3542451983485909,
      "grad_norm": 4.639617443084717,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 3947
    },
    {
      "epoch": 0.3543349488422186,
      "grad_norm": 3.832247734069824,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 3948
    },
    {
      "epoch": 0.35442469933584636,
      "grad_norm": 3.3647027015686035,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 3949
    },
    {
      "epoch": 0.35451444982947405,
      "grad_norm": 3.053637981414795,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 3950
    },
    {
      "epoch": 0.3546042003231018,
      "grad_norm": 4.192505359649658,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 3951
    },
    {
      "epoch": 0.3546939508167295,
      "grad_norm": 3.4051482677459717,
      "learning_rate": 0.0,
      "loss": 0.7506,
      "step": 3952
    },
    {
      "epoch": 0.3547837013103572,
      "grad_norm": 5.607193470001221,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 3953
    },
    {
      "epoch": 0.35487345180398494,
      "grad_norm": 4.598918914794922,
      "learning_rate": 0.0,
      "loss": 0.7784,
      "step": 3954
    },
    {
      "epoch": 0.35496320229761263,
      "grad_norm": 3.1046345233917236,
      "learning_rate": 0.0,
      "loss": 0.7643,
      "step": 3955
    },
    {
      "epoch": 0.35505295279124033,
      "grad_norm": 4.017154216766357,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 3956
    },
    {
      "epoch": 0.3551427032848681,
      "grad_norm": 4.242393970489502,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 3957
    },
    {
      "epoch": 0.35523245377849577,
      "grad_norm": 4.566004276275635,
      "learning_rate": 0.0,
      "loss": 0.8056,
      "step": 3958
    },
    {
      "epoch": 0.3553222042721235,
      "grad_norm": 4.155086517333984,
      "learning_rate": 0.0,
      "loss": 0.7716,
      "step": 3959
    },
    {
      "epoch": 0.3554119547657512,
      "grad_norm": 2.2778167724609375,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 3960
    },
    {
      "epoch": 0.3555017052593789,
      "grad_norm": 6.352217674255371,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 3961
    },
    {
      "epoch": 0.35559145575300666,
      "grad_norm": 2.598068952560425,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 3962
    },
    {
      "epoch": 0.35568120624663435,
      "grad_norm": 2.648681640625,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 3963
    },
    {
      "epoch": 0.35577095674026205,
      "grad_norm": 3.1971094608306885,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 3964
    },
    {
      "epoch": 0.3558607072338898,
      "grad_norm": 2.671776294708252,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 3965
    },
    {
      "epoch": 0.3559504577275175,
      "grad_norm": 6.0930585861206055,
      "learning_rate": 0.0,
      "loss": 0.7836,
      "step": 3966
    },
    {
      "epoch": 0.35604020822114524,
      "grad_norm": 3.3946685791015625,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 3967
    },
    {
      "epoch": 0.35612995871477293,
      "grad_norm": 3.254997730255127,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 3968
    },
    {
      "epoch": 0.3562197092084006,
      "grad_norm": 4.071510314941406,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 3969
    },
    {
      "epoch": 0.3563094597020284,
      "grad_norm": 3.592888116836548,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 3970
    },
    {
      "epoch": 0.35639921019565607,
      "grad_norm": 3.5415103435516357,
      "learning_rate": 0.0,
      "loss": 0.8014,
      "step": 3971
    },
    {
      "epoch": 0.3564889606892838,
      "grad_norm": 3.525873899459839,
      "learning_rate": 0.0,
      "loss": 0.7887,
      "step": 3972
    },
    {
      "epoch": 0.3565787111829115,
      "grad_norm": 2.396860361099243,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 3973
    },
    {
      "epoch": 0.3566684616765392,
      "grad_norm": 3.456982135772705,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 3974
    },
    {
      "epoch": 0.35675821217016696,
      "grad_norm": 4.79865837097168,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 3975
    },
    {
      "epoch": 0.35684796266379465,
      "grad_norm": 4.66523551940918,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 3976
    },
    {
      "epoch": 0.35693771315742234,
      "grad_norm": 3.091684341430664,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 3977
    },
    {
      "epoch": 0.3570274636510501,
      "grad_norm": 3.6927132606506348,
      "learning_rate": 0.0,
      "loss": 0.6795,
      "step": 3978
    },
    {
      "epoch": 0.3571172141446778,
      "grad_norm": 4.052168369293213,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 3979
    },
    {
      "epoch": 0.35720696463830554,
      "grad_norm": 6.219174861907959,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 3980
    },
    {
      "epoch": 0.35729671513193323,
      "grad_norm": 3.1624677181243896,
      "learning_rate": 0.0,
      "loss": 0.7803,
      "step": 3981
    },
    {
      "epoch": 0.3573864656255609,
      "grad_norm": 2.9921324253082275,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 3982
    },
    {
      "epoch": 0.35747621611918867,
      "grad_norm": 2.6581196784973145,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 3983
    },
    {
      "epoch": 0.35756596661281637,
      "grad_norm": 2.5942764282226562,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 3984
    },
    {
      "epoch": 0.35765571710644406,
      "grad_norm": 2.927396535873413,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 3985
    },
    {
      "epoch": 0.3577454676000718,
      "grad_norm": 3.3797683715820312,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 3986
    },
    {
      "epoch": 0.3578352180936995,
      "grad_norm": 2.948282480239868,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 3987
    },
    {
      "epoch": 0.35792496858732725,
      "grad_norm": 166.8498992919922,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 3988
    },
    {
      "epoch": 0.35801471908095495,
      "grad_norm": 3.768223524093628,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 3989
    },
    {
      "epoch": 0.35810446957458264,
      "grad_norm": 2.8795807361602783,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 3990
    },
    {
      "epoch": 0.3581942200682104,
      "grad_norm": 5.211670875549316,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 3991
    },
    {
      "epoch": 0.3582839705618381,
      "grad_norm": 3.5962769985198975,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 3992
    },
    {
      "epoch": 0.35837372105546583,
      "grad_norm": 3.3194944858551025,
      "learning_rate": 0.0,
      "loss": 0.7072,
      "step": 3993
    },
    {
      "epoch": 0.3584634715490935,
      "grad_norm": 2.520526647567749,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 3994
    },
    {
      "epoch": 0.3585532220427212,
      "grad_norm": 4.208664417266846,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 3995
    },
    {
      "epoch": 0.35864297253634897,
      "grad_norm": 46.75336837768555,
      "learning_rate": 0.0,
      "loss": 0.7881,
      "step": 3996
    },
    {
      "epoch": 0.35873272302997666,
      "grad_norm": 6.628696918487549,
      "learning_rate": 0.0,
      "loss": 0.8086,
      "step": 3997
    },
    {
      "epoch": 0.35882247352360436,
      "grad_norm": 6.219925403594971,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 3998
    },
    {
      "epoch": 0.3589122240172321,
      "grad_norm": 4.145536422729492,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 3999
    },
    {
      "epoch": 0.3590019745108598,
      "grad_norm": 5.082260608673096,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 4000
    },
    {
      "epoch": 0.35909172500448755,
      "grad_norm": 8.130023002624512,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 4001
    },
    {
      "epoch": 0.35918147549811524,
      "grad_norm": 2.8760111331939697,
      "learning_rate": 0.0,
      "loss": 0.7203,
      "step": 4002
    },
    {
      "epoch": 0.35927122599174294,
      "grad_norm": 5.995151042938232,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 4003
    },
    {
      "epoch": 0.3593609764853707,
      "grad_norm": 3.761934518814087,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 4004
    },
    {
      "epoch": 0.3594507269789984,
      "grad_norm": 2.8610620498657227,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 4005
    },
    {
      "epoch": 0.3595404774726261,
      "grad_norm": 2.511990547180176,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 4006
    },
    {
      "epoch": 0.3596302279662538,
      "grad_norm": 3.12895131111145,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 4007
    },
    {
      "epoch": 0.3597199784598815,
      "grad_norm": 4.259524345397949,
      "learning_rate": 0.0,
      "loss": 0.7052,
      "step": 4008
    },
    {
      "epoch": 0.35980972895350927,
      "grad_norm": 7.929801940917969,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 4009
    },
    {
      "epoch": 0.35989947944713696,
      "grad_norm": 5.641621112823486,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 4010
    },
    {
      "epoch": 0.35998922994076465,
      "grad_norm": 2.8719162940979004,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 4011
    },
    {
      "epoch": 0.3600789804343924,
      "grad_norm": 3.274338483810425,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 4012
    },
    {
      "epoch": 0.3601687309280201,
      "grad_norm": 12.582616806030273,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 4013
    },
    {
      "epoch": 0.3602584814216478,
      "grad_norm": 4.744102954864502,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 4014
    },
    {
      "epoch": 0.36034823191527554,
      "grad_norm": 2.7414743900299072,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 4015
    },
    {
      "epoch": 0.36043798240890323,
      "grad_norm": 2.6503233909606934,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 4016
    },
    {
      "epoch": 0.360527732902531,
      "grad_norm": 5.1666364669799805,
      "learning_rate": 0.0,
      "loss": 0.792,
      "step": 4017
    },
    {
      "epoch": 0.3606174833961587,
      "grad_norm": 3.6909775733947754,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 4018
    },
    {
      "epoch": 0.36070723388978637,
      "grad_norm": 3.5224173069000244,
      "learning_rate": 0.0,
      "loss": 0.773,
      "step": 4019
    },
    {
      "epoch": 0.3607969843834141,
      "grad_norm": 4.375161647796631,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 4020
    },
    {
      "epoch": 0.3608867348770418,
      "grad_norm": 3.6165030002593994,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 4021
    },
    {
      "epoch": 0.36097648537066956,
      "grad_norm": 3.2627525329589844,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 4022
    },
    {
      "epoch": 0.36106623586429726,
      "grad_norm": 2.3979291915893555,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 4023
    },
    {
      "epoch": 0.36115598635792495,
      "grad_norm": 2.4285407066345215,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 4024
    },
    {
      "epoch": 0.3612457368515527,
      "grad_norm": 2.2866122722625732,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 4025
    },
    {
      "epoch": 0.3613354873451804,
      "grad_norm": 3.3000478744506836,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 4026
    },
    {
      "epoch": 0.3614252378388081,
      "grad_norm": 2.8896989822387695,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 4027
    },
    {
      "epoch": 0.36151498833243584,
      "grad_norm": 2.5817606449127197,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 4028
    },
    {
      "epoch": 0.36160473882606353,
      "grad_norm": 2.3514580726623535,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 4029
    },
    {
      "epoch": 0.3616944893196913,
      "grad_norm": 2.6858608722686768,
      "learning_rate": 0.0,
      "loss": 0.7929,
      "step": 4030
    },
    {
      "epoch": 0.361784239813319,
      "grad_norm": 3.6804075241088867,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 4031
    },
    {
      "epoch": 0.36187399030694667,
      "grad_norm": 7.578358173370361,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 4032
    },
    {
      "epoch": 0.3619637408005744,
      "grad_norm": 3.571117401123047,
      "learning_rate": 0.0,
      "loss": 0.7124,
      "step": 4033
    },
    {
      "epoch": 0.3620534912942021,
      "grad_norm": 4.946100234985352,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 4034
    },
    {
      "epoch": 0.3621432417878298,
      "grad_norm": 5.654266357421875,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 4035
    },
    {
      "epoch": 0.36223299228145756,
      "grad_norm": 4.0446319580078125,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 4036
    },
    {
      "epoch": 0.36232274277508525,
      "grad_norm": 4.087086200714111,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 4037
    },
    {
      "epoch": 0.362412493268713,
      "grad_norm": 4.421653747558594,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 4038
    },
    {
      "epoch": 0.3625022437623407,
      "grad_norm": 2.873933792114258,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 4039
    },
    {
      "epoch": 0.3625919942559684,
      "grad_norm": 4.803135395050049,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 4040
    },
    {
      "epoch": 0.36268174474959614,
      "grad_norm": 4.405425071716309,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 4041
    },
    {
      "epoch": 0.36277149524322383,
      "grad_norm": 3.46020245552063,
      "learning_rate": 0.0,
      "loss": 0.6942,
      "step": 4042
    },
    {
      "epoch": 0.3628612457368516,
      "grad_norm": 7.713572025299072,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 4043
    },
    {
      "epoch": 0.3629509962304793,
      "grad_norm": 3.3291614055633545,
      "learning_rate": 0.0,
      "loss": 0.7186,
      "step": 4044
    },
    {
      "epoch": 0.36304074672410697,
      "grad_norm": 2.3436360359191895,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 4045
    },
    {
      "epoch": 0.3631304972177347,
      "grad_norm": 3.863940715789795,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 4046
    },
    {
      "epoch": 0.3632202477113624,
      "grad_norm": 3.1084372997283936,
      "learning_rate": 0.0,
      "loss": 0.746,
      "step": 4047
    },
    {
      "epoch": 0.3633099982049901,
      "grad_norm": 7.558996677398682,
      "learning_rate": 0.0,
      "loss": 0.7842,
      "step": 4048
    },
    {
      "epoch": 0.36339974869861785,
      "grad_norm": 3.0749595165252686,
      "learning_rate": 0.0,
      "loss": 0.8114,
      "step": 4049
    },
    {
      "epoch": 0.36348949919224555,
      "grad_norm": 4.035207748413086,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 4050
    },
    {
      "epoch": 0.3635792496858733,
      "grad_norm": 3.150583028793335,
      "learning_rate": 0.0,
      "loss": 0.8196,
      "step": 4051
    },
    {
      "epoch": 0.363669000179501,
      "grad_norm": 2.7387335300445557,
      "learning_rate": 0.0,
      "loss": 0.7972,
      "step": 4052
    },
    {
      "epoch": 0.3637587506731287,
      "grad_norm": 2.870577573776245,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 4053
    },
    {
      "epoch": 0.36384850116675643,
      "grad_norm": 3.2029061317443848,
      "learning_rate": 0.0,
      "loss": 0.7491,
      "step": 4054
    },
    {
      "epoch": 0.3639382516603841,
      "grad_norm": 2.6806094646453857,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 4055
    },
    {
      "epoch": 0.3640280021540118,
      "grad_norm": 2.4508438110351562,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 4056
    },
    {
      "epoch": 0.36411775264763957,
      "grad_norm": 4.11589241027832,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 4057
    },
    {
      "epoch": 0.36420750314126726,
      "grad_norm": 3.145226001739502,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 4058
    },
    {
      "epoch": 0.364297253634895,
      "grad_norm": 3.0677742958068848,
      "learning_rate": 0.0,
      "loss": 0.7989,
      "step": 4059
    },
    {
      "epoch": 0.3643870041285227,
      "grad_norm": 3.006720542907715,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 4060
    },
    {
      "epoch": 0.3644767546221504,
      "grad_norm": 3.145158529281616,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 4061
    },
    {
      "epoch": 0.36456650511577815,
      "grad_norm": 2.561706066131592,
      "learning_rate": 0.0,
      "loss": 0.724,
      "step": 4062
    },
    {
      "epoch": 0.36465625560940584,
      "grad_norm": 7.335243225097656,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 4063
    },
    {
      "epoch": 0.3647460061030336,
      "grad_norm": 3.2538657188415527,
      "learning_rate": 0.0,
      "loss": 0.7456,
      "step": 4064
    },
    {
      "epoch": 0.3648357565966613,
      "grad_norm": 2.91081166267395,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 4065
    },
    {
      "epoch": 0.364925507090289,
      "grad_norm": 2.4191343784332275,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 4066
    },
    {
      "epoch": 0.36501525758391673,
      "grad_norm": 2.115067481994629,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 4067
    },
    {
      "epoch": 0.3651050080775444,
      "grad_norm": 3.2483038902282715,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 4068
    },
    {
      "epoch": 0.3651947585711721,
      "grad_norm": 3.864767074584961,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 4069
    },
    {
      "epoch": 0.36528450906479987,
      "grad_norm": 3.459073543548584,
      "learning_rate": 0.0,
      "loss": 0.8373,
      "step": 4070
    },
    {
      "epoch": 0.36537425955842756,
      "grad_norm": 3.1222646236419678,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 4071
    },
    {
      "epoch": 0.3654640100520553,
      "grad_norm": 2.704010248184204,
      "learning_rate": 0.0,
      "loss": 0.6642,
      "step": 4072
    },
    {
      "epoch": 0.365553760545683,
      "grad_norm": 14.453357696533203,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 4073
    },
    {
      "epoch": 0.3656435110393107,
      "grad_norm": 3.98840069770813,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 4074
    },
    {
      "epoch": 0.36573326153293845,
      "grad_norm": 3.0006206035614014,
      "learning_rate": 0.0,
      "loss": 0.7867,
      "step": 4075
    },
    {
      "epoch": 0.36582301202656614,
      "grad_norm": 24.557958602905273,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 4076
    },
    {
      "epoch": 0.36591276252019383,
      "grad_norm": 3.2586989402770996,
      "learning_rate": 0.0,
      "loss": 0.7764,
      "step": 4077
    },
    {
      "epoch": 0.3660025130138216,
      "grad_norm": 3.555568218231201,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 4078
    },
    {
      "epoch": 0.3660922635074493,
      "grad_norm": 4.290005207061768,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 4079
    },
    {
      "epoch": 0.366182014001077,
      "grad_norm": 2.354013442993164,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 4080
    },
    {
      "epoch": 0.3662717644947047,
      "grad_norm": 2.5877251625061035,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 4081
    },
    {
      "epoch": 0.3663615149883324,
      "grad_norm": 2.71943736076355,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 4082
    },
    {
      "epoch": 0.36645126548196016,
      "grad_norm": 2.2782680988311768,
      "learning_rate": 0.0,
      "loss": 0.8193,
      "step": 4083
    },
    {
      "epoch": 0.36654101597558786,
      "grad_norm": 2.971252202987671,
      "learning_rate": 0.0,
      "loss": 0.7668,
      "step": 4084
    },
    {
      "epoch": 0.3666307664692156,
      "grad_norm": 3.2092981338500977,
      "learning_rate": 0.0,
      "loss": 0.7891,
      "step": 4085
    },
    {
      "epoch": 0.3667205169628433,
      "grad_norm": 2.8686399459838867,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 4086
    },
    {
      "epoch": 0.366810267456471,
      "grad_norm": 2.5421385765075684,
      "learning_rate": 0.0,
      "loss": 0.7781,
      "step": 4087
    },
    {
      "epoch": 0.36690001795009874,
      "grad_norm": 4.0678253173828125,
      "learning_rate": 0.0,
      "loss": 0.7819,
      "step": 4088
    },
    {
      "epoch": 0.36698976844372644,
      "grad_norm": 2.237323760986328,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 4089
    },
    {
      "epoch": 0.36707951893735413,
      "grad_norm": 2.2226860523223877,
      "learning_rate": 0.0,
      "loss": 0.6682,
      "step": 4090
    },
    {
      "epoch": 0.3671692694309819,
      "grad_norm": 2.496812105178833,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 4091
    },
    {
      "epoch": 0.3672590199246096,
      "grad_norm": 2.386274576187134,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 4092
    },
    {
      "epoch": 0.3673487704182373,
      "grad_norm": 2.2436416149139404,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 4093
    },
    {
      "epoch": 0.367438520911865,
      "grad_norm": 3.126401662826538,
      "learning_rate": 0.0,
      "loss": 0.784,
      "step": 4094
    },
    {
      "epoch": 0.3675282714054927,
      "grad_norm": 2.1637542247772217,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 4095
    },
    {
      "epoch": 0.36761802189912046,
      "grad_norm": 2.178635835647583,
      "learning_rate": 0.0,
      "loss": 0.7764,
      "step": 4096
    },
    {
      "epoch": 0.36770777239274816,
      "grad_norm": 4.072469711303711,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 4097
    },
    {
      "epoch": 0.36779752288637585,
      "grad_norm": 3.314361810684204,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 4098
    },
    {
      "epoch": 0.3678872733800036,
      "grad_norm": 2.6602439880371094,
      "learning_rate": 0.0,
      "loss": 0.7319,
      "step": 4099
    },
    {
      "epoch": 0.3679770238736313,
      "grad_norm": 2.426347494125366,
      "learning_rate": 0.0,
      "loss": 0.8423,
      "step": 4100
    },
    {
      "epoch": 0.36806677436725904,
      "grad_norm": 3.0595312118530273,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 4101
    },
    {
      "epoch": 0.36815652486088674,
      "grad_norm": 6.1867594718933105,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 4102
    },
    {
      "epoch": 0.36824627535451443,
      "grad_norm": 2.6272714138031006,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 4103
    },
    {
      "epoch": 0.3683360258481422,
      "grad_norm": 4.1953864097595215,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 4104
    },
    {
      "epoch": 0.3684257763417699,
      "grad_norm": 3.3269476890563965,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 4105
    },
    {
      "epoch": 0.3685155268353976,
      "grad_norm": 3.489020586013794,
      "learning_rate": 0.0,
      "loss": 0.7707,
      "step": 4106
    },
    {
      "epoch": 0.3686052773290253,
      "grad_norm": 3.6251471042633057,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 4107
    },
    {
      "epoch": 0.368695027822653,
      "grad_norm": 1.9421911239624023,
      "learning_rate": 0.0,
      "loss": 0.8355,
      "step": 4108
    },
    {
      "epoch": 0.36878477831628076,
      "grad_norm": 2.906099319458008,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 4109
    },
    {
      "epoch": 0.36887452880990845,
      "grad_norm": 2.5108673572540283,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 4110
    },
    {
      "epoch": 0.36896427930353615,
      "grad_norm": 3.4162440299987793,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 4111
    },
    {
      "epoch": 0.3690540297971639,
      "grad_norm": 2.994908094406128,
      "learning_rate": 0.0,
      "loss": 0.7659,
      "step": 4112
    },
    {
      "epoch": 0.3691437802907916,
      "grad_norm": 2.5500423908233643,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 4113
    },
    {
      "epoch": 0.36923353078441934,
      "grad_norm": 2.365849494934082,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 4114
    },
    {
      "epoch": 0.36932328127804703,
      "grad_norm": 2.941493272781372,
      "learning_rate": 0.0,
      "loss": 0.6744,
      "step": 4115
    },
    {
      "epoch": 0.3694130317716747,
      "grad_norm": 2.2207846641540527,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 4116
    },
    {
      "epoch": 0.3695027822653025,
      "grad_norm": 3.0288448333740234,
      "learning_rate": 0.0,
      "loss": 0.7844,
      "step": 4117
    },
    {
      "epoch": 0.36959253275893017,
      "grad_norm": 2.587132692337036,
      "learning_rate": 0.0,
      "loss": 0.7749,
      "step": 4118
    },
    {
      "epoch": 0.36968228325255786,
      "grad_norm": 3.324530601501465,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 4119
    },
    {
      "epoch": 0.3697720337461856,
      "grad_norm": 3.347477436065674,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 4120
    },
    {
      "epoch": 0.3698617842398133,
      "grad_norm": 2.799748182296753,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 4121
    },
    {
      "epoch": 0.36995153473344106,
      "grad_norm": 2.9251909255981445,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 4122
    },
    {
      "epoch": 0.37004128522706875,
      "grad_norm": 4.5349297523498535,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 4123
    },
    {
      "epoch": 0.37013103572069644,
      "grad_norm": 2.839547872543335,
      "learning_rate": 0.0,
      "loss": 0.7745,
      "step": 4124
    },
    {
      "epoch": 0.3702207862143242,
      "grad_norm": 3.0373330116271973,
      "learning_rate": 0.0,
      "loss": 0.7329,
      "step": 4125
    },
    {
      "epoch": 0.3703105367079519,
      "grad_norm": 2.900956392288208,
      "learning_rate": 0.0,
      "loss": 0.7975,
      "step": 4126
    },
    {
      "epoch": 0.3704002872015796,
      "grad_norm": 3.4125897884368896,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 4127
    },
    {
      "epoch": 0.37049003769520733,
      "grad_norm": 3.442542314529419,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 4128
    },
    {
      "epoch": 0.370579788188835,
      "grad_norm": 2.7797625064849854,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 4129
    },
    {
      "epoch": 0.3706695386824628,
      "grad_norm": 2.273184061050415,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 4130
    },
    {
      "epoch": 0.37075928917609047,
      "grad_norm": 4.051140308380127,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 4131
    },
    {
      "epoch": 0.37084903966971816,
      "grad_norm": 3.097888946533203,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 4132
    },
    {
      "epoch": 0.3709387901633459,
      "grad_norm": 2.6766088008880615,
      "learning_rate": 0.0,
      "loss": 0.8462,
      "step": 4133
    },
    {
      "epoch": 0.3710285406569736,
      "grad_norm": 3.760366916656494,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 4134
    },
    {
      "epoch": 0.37111829115060135,
      "grad_norm": 2.561539888381958,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 4135
    },
    {
      "epoch": 0.37120804164422905,
      "grad_norm": 9.19952392578125,
      "learning_rate": 0.0,
      "loss": 0.7944,
      "step": 4136
    },
    {
      "epoch": 0.37129779213785674,
      "grad_norm": 3.514894962310791,
      "learning_rate": 0.0,
      "loss": 0.7353,
      "step": 4137
    },
    {
      "epoch": 0.3713875426314845,
      "grad_norm": 2.5905492305755615,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 4138
    },
    {
      "epoch": 0.3714772931251122,
      "grad_norm": 2.2230544090270996,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 4139
    },
    {
      "epoch": 0.3715670436187399,
      "grad_norm": 2.79305362701416,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 4140
    },
    {
      "epoch": 0.3716567941123676,
      "grad_norm": 8.116561889648438,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 4141
    },
    {
      "epoch": 0.3717465446059953,
      "grad_norm": 2.638566732406616,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 4142
    },
    {
      "epoch": 0.37183629509962307,
      "grad_norm": 2.7591121196746826,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 4143
    },
    {
      "epoch": 0.37192604559325076,
      "grad_norm": 3.55488920211792,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 4144
    },
    {
      "epoch": 0.37201579608687846,
      "grad_norm": 13.884735107421875,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 4145
    },
    {
      "epoch": 0.3721055465805062,
      "grad_norm": 3.5068511962890625,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 4146
    },
    {
      "epoch": 0.3721952970741339,
      "grad_norm": 5.475210666656494,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 4147
    },
    {
      "epoch": 0.3722850475677616,
      "grad_norm": 3.492537260055542,
      "learning_rate": 0.0,
      "loss": 0.6833,
      "step": 4148
    },
    {
      "epoch": 0.37237479806138934,
      "grad_norm": 2.642179012298584,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 4149
    },
    {
      "epoch": 0.37246454855501704,
      "grad_norm": 5.068558216094971,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 4150
    },
    {
      "epoch": 0.3725542990486448,
      "grad_norm": 3.1490371227264404,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 4151
    },
    {
      "epoch": 0.3726440495422725,
      "grad_norm": 2.856815814971924,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 4152
    },
    {
      "epoch": 0.3727338000359002,
      "grad_norm": 4.568470478057861,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 4153
    },
    {
      "epoch": 0.3728235505295279,
      "grad_norm": 2.83486008644104,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 4154
    },
    {
      "epoch": 0.3729133010231556,
      "grad_norm": 2.681440591812134,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 4155
    },
    {
      "epoch": 0.37300305151678337,
      "grad_norm": 3.34025239944458,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 4156
    },
    {
      "epoch": 0.37309280201041106,
      "grad_norm": 5.324084281921387,
      "learning_rate": 0.0,
      "loss": 0.7365,
      "step": 4157
    },
    {
      "epoch": 0.37318255250403876,
      "grad_norm": 3.1854536533355713,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 4158
    },
    {
      "epoch": 0.3732723029976665,
      "grad_norm": 3.6061925888061523,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 4159
    },
    {
      "epoch": 0.3733620534912942,
      "grad_norm": 2.796485424041748,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 4160
    },
    {
      "epoch": 0.3734518039849219,
      "grad_norm": 5.223779678344727,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 4161
    },
    {
      "epoch": 0.37354155447854964,
      "grad_norm": 3.973818302154541,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 4162
    },
    {
      "epoch": 0.37363130497217734,
      "grad_norm": 5.215452671051025,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 4163
    },
    {
      "epoch": 0.3737210554658051,
      "grad_norm": 6.604392051696777,
      "learning_rate": 0.0,
      "loss": 0.7967,
      "step": 4164
    },
    {
      "epoch": 0.3738108059594328,
      "grad_norm": 4.054443836212158,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 4165
    },
    {
      "epoch": 0.3739005564530605,
      "grad_norm": 6.2731122970581055,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 4166
    },
    {
      "epoch": 0.3739903069466882,
      "grad_norm": 2.324378728866577,
      "learning_rate": 0.0,
      "loss": 0.7731,
      "step": 4167
    },
    {
      "epoch": 0.3740800574403159,
      "grad_norm": 1.9919815063476562,
      "learning_rate": 0.0,
      "loss": 0.7861,
      "step": 4168
    },
    {
      "epoch": 0.3741698079339436,
      "grad_norm": 2.562452793121338,
      "learning_rate": 0.0,
      "loss": 0.7314,
      "step": 4169
    },
    {
      "epoch": 0.37425955842757136,
      "grad_norm": 4.096845626831055,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 4170
    },
    {
      "epoch": 0.37434930892119905,
      "grad_norm": 3.5341098308563232,
      "learning_rate": 0.0,
      "loss": 0.7375,
      "step": 4171
    },
    {
      "epoch": 0.3744390594148268,
      "grad_norm": 2.866173267364502,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 4172
    },
    {
      "epoch": 0.3745288099084545,
      "grad_norm": 3.583132266998291,
      "learning_rate": 0.0,
      "loss": 0.7971,
      "step": 4173
    },
    {
      "epoch": 0.3746185604020822,
      "grad_norm": 2.9852027893066406,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 4174
    },
    {
      "epoch": 0.37470831089570994,
      "grad_norm": 2.7071762084960938,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 4175
    },
    {
      "epoch": 0.37479806138933763,
      "grad_norm": 2.0691699981689453,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 4176
    },
    {
      "epoch": 0.3748878118829654,
      "grad_norm": 6.966726303100586,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 4177
    },
    {
      "epoch": 0.3749775623765931,
      "grad_norm": 3.817864418029785,
      "learning_rate": 0.0,
      "loss": 0.7204,
      "step": 4178
    },
    {
      "epoch": 0.37506731287022077,
      "grad_norm": 3.0389175415039062,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 4179
    },
    {
      "epoch": 0.3751570633638485,
      "grad_norm": 2.5067214965820312,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 4180
    },
    {
      "epoch": 0.3752468138574762,
      "grad_norm": 3.261449098587036,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 4181
    },
    {
      "epoch": 0.3753365643511039,
      "grad_norm": 6.918036937713623,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 4182
    },
    {
      "epoch": 0.37542631484473166,
      "grad_norm": 2.499399423599243,
      "learning_rate": 0.0,
      "loss": 0.7635,
      "step": 4183
    },
    {
      "epoch": 0.37551606533835935,
      "grad_norm": 3.336848497390747,
      "learning_rate": 0.0,
      "loss": 0.6992,
      "step": 4184
    },
    {
      "epoch": 0.3756058158319871,
      "grad_norm": 2.809485912322998,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 4185
    },
    {
      "epoch": 0.3756955663256148,
      "grad_norm": 4.627539157867432,
      "learning_rate": 0.0,
      "loss": 0.7583,
      "step": 4186
    },
    {
      "epoch": 0.3757853168192425,
      "grad_norm": 2.5772786140441895,
      "learning_rate": 0.0,
      "loss": 0.7837,
      "step": 4187
    },
    {
      "epoch": 0.37587506731287024,
      "grad_norm": 2.8750107288360596,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 4188
    },
    {
      "epoch": 0.37596481780649793,
      "grad_norm": 17.09433364868164,
      "learning_rate": 0.0,
      "loss": 0.8143,
      "step": 4189
    },
    {
      "epoch": 0.3760545683001256,
      "grad_norm": 3.3000714778900146,
      "learning_rate": 0.0,
      "loss": 0.808,
      "step": 4190
    },
    {
      "epoch": 0.3761443187937534,
      "grad_norm": 3.0454154014587402,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 4191
    },
    {
      "epoch": 0.37623406928738107,
      "grad_norm": 3.075129508972168,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 4192
    },
    {
      "epoch": 0.3763238197810088,
      "grad_norm": 4.0164008140563965,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 4193
    },
    {
      "epoch": 0.3764135702746365,
      "grad_norm": 3.1716465950012207,
      "learning_rate": 0.0,
      "loss": 0.8038,
      "step": 4194
    },
    {
      "epoch": 0.3765033207682642,
      "grad_norm": 2.1963069438934326,
      "learning_rate": 0.0,
      "loss": 0.7931,
      "step": 4195
    },
    {
      "epoch": 0.37659307126189195,
      "grad_norm": 3.745415210723877,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 4196
    },
    {
      "epoch": 0.37668282175551965,
      "grad_norm": 4.268096923828125,
      "learning_rate": 0.0,
      "loss": 0.7779,
      "step": 4197
    },
    {
      "epoch": 0.3767725722491474,
      "grad_norm": 3.1303555965423584,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 4198
    },
    {
      "epoch": 0.3768623227427751,
      "grad_norm": 2.4470293521881104,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 4199
    },
    {
      "epoch": 0.3769520732364028,
      "grad_norm": 2.73380446434021,
      "learning_rate": 0.0,
      "loss": 0.7152,
      "step": 4200
    },
    {
      "epoch": 0.37704182373003053,
      "grad_norm": 2.789876699447632,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 4201
    },
    {
      "epoch": 0.37713157422365823,
      "grad_norm": 6.807017803192139,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 4202
    },
    {
      "epoch": 0.3772213247172859,
      "grad_norm": 2.939229965209961,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 4203
    },
    {
      "epoch": 0.37731107521091367,
      "grad_norm": 4.245284080505371,
      "learning_rate": 0.0,
      "loss": 0.6846,
      "step": 4204
    },
    {
      "epoch": 0.37740082570454137,
      "grad_norm": 3.217223644256592,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 4205
    },
    {
      "epoch": 0.3774905761981691,
      "grad_norm": 3.43088436126709,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 4206
    },
    {
      "epoch": 0.3775803266917968,
      "grad_norm": 4.643762111663818,
      "learning_rate": 0.0,
      "loss": 0.7759,
      "step": 4207
    },
    {
      "epoch": 0.3776700771854245,
      "grad_norm": 2.4383292198181152,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 4208
    },
    {
      "epoch": 0.37775982767905225,
      "grad_norm": 5.572000503540039,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 4209
    },
    {
      "epoch": 0.37784957817267995,
      "grad_norm": 7.39697790145874,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 4210
    },
    {
      "epoch": 0.37793932866630764,
      "grad_norm": 2.684159755706787,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 4211
    },
    {
      "epoch": 0.3780290791599354,
      "grad_norm": 2.385037899017334,
      "learning_rate": 0.0,
      "loss": 0.7301,
      "step": 4212
    },
    {
      "epoch": 0.3781188296535631,
      "grad_norm": 3.014155864715576,
      "learning_rate": 0.0,
      "loss": 0.6825,
      "step": 4213
    },
    {
      "epoch": 0.37820858014719083,
      "grad_norm": 2.525604486465454,
      "learning_rate": 0.0,
      "loss": 0.8053,
      "step": 4214
    },
    {
      "epoch": 0.3782983306408185,
      "grad_norm": 3.4938597679138184,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 4215
    },
    {
      "epoch": 0.3783880811344462,
      "grad_norm": 4.8034515380859375,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 4216
    },
    {
      "epoch": 0.37847783162807397,
      "grad_norm": 3.012988805770874,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 4217
    },
    {
      "epoch": 0.37856758212170166,
      "grad_norm": 2.548684597015381,
      "learning_rate": 0.0,
      "loss": 0.7927,
      "step": 4218
    },
    {
      "epoch": 0.3786573326153294,
      "grad_norm": 3.0363075733184814,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 4219
    },
    {
      "epoch": 0.3787470831089571,
      "grad_norm": 2.9158153533935547,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 4220
    },
    {
      "epoch": 0.3788368336025848,
      "grad_norm": 3.349374771118164,
      "learning_rate": 0.0,
      "loss": 0.7694,
      "step": 4221
    },
    {
      "epoch": 0.37892658409621255,
      "grad_norm": 3.867198944091797,
      "learning_rate": 0.0,
      "loss": 0.7643,
      "step": 4222
    },
    {
      "epoch": 0.37901633458984024,
      "grad_norm": 2.617659091949463,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 4223
    },
    {
      "epoch": 0.37910608508346794,
      "grad_norm": 2.725959062576294,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 4224
    },
    {
      "epoch": 0.3791958355770957,
      "grad_norm": 3.1156234741210938,
      "learning_rate": 0.0,
      "loss": 0.6685,
      "step": 4225
    },
    {
      "epoch": 0.3792855860707234,
      "grad_norm": 2.8908004760742188,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 4226
    },
    {
      "epoch": 0.37937533656435113,
      "grad_norm": 2.5412895679473877,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 4227
    },
    {
      "epoch": 0.3794650870579788,
      "grad_norm": 3.352648973464966,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 4228
    },
    {
      "epoch": 0.3795548375516065,
      "grad_norm": 2.909868001937866,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 4229
    },
    {
      "epoch": 0.37964458804523427,
      "grad_norm": 2.665865182876587,
      "learning_rate": 0.0,
      "loss": 0.787,
      "step": 4230
    },
    {
      "epoch": 0.37973433853886196,
      "grad_norm": 2.735593318939209,
      "learning_rate": 0.0,
      "loss": 0.747,
      "step": 4231
    },
    {
      "epoch": 0.37982408903248965,
      "grad_norm": 4.0260701179504395,
      "learning_rate": 0.0,
      "loss": 0.7627,
      "step": 4232
    },
    {
      "epoch": 0.3799138395261174,
      "grad_norm": 2.5754497051239014,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 4233
    },
    {
      "epoch": 0.3800035900197451,
      "grad_norm": 2.994349718093872,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 4234
    },
    {
      "epoch": 0.38009334051337285,
      "grad_norm": 3.3300979137420654,
      "learning_rate": 0.0,
      "loss": 0.7788,
      "step": 4235
    },
    {
      "epoch": 0.38018309100700054,
      "grad_norm": 2.430638313293457,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 4236
    },
    {
      "epoch": 0.38027284150062823,
      "grad_norm": 2.9890081882476807,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 4237
    },
    {
      "epoch": 0.380362591994256,
      "grad_norm": 2.2647600173950195,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 4238
    },
    {
      "epoch": 0.3804523424878837,
      "grad_norm": 2.619868755340576,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 4239
    },
    {
      "epoch": 0.38054209298151137,
      "grad_norm": 2.7152748107910156,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 4240
    },
    {
      "epoch": 0.3806318434751391,
      "grad_norm": 4.347195625305176,
      "learning_rate": 0.0,
      "loss": 0.7418,
      "step": 4241
    },
    {
      "epoch": 0.3807215939687668,
      "grad_norm": 8.050092697143555,
      "learning_rate": 0.0,
      "loss": 0.8048,
      "step": 4242
    },
    {
      "epoch": 0.38081134446239456,
      "grad_norm": 2.8086462020874023,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 4243
    },
    {
      "epoch": 0.38090109495602226,
      "grad_norm": 3.2169604301452637,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 4244
    },
    {
      "epoch": 0.38099084544964995,
      "grad_norm": 16.866315841674805,
      "learning_rate": 0.0,
      "loss": 0.769,
      "step": 4245
    },
    {
      "epoch": 0.3810805959432777,
      "grad_norm": 3.6649656295776367,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 4246
    },
    {
      "epoch": 0.3811703464369054,
      "grad_norm": 3.232058048248291,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 4247
    },
    {
      "epoch": 0.38126009693053314,
      "grad_norm": 2.2037220001220703,
      "learning_rate": 0.0,
      "loss": 0.7431,
      "step": 4248
    },
    {
      "epoch": 0.38134984742416084,
      "grad_norm": 2.562006950378418,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 4249
    },
    {
      "epoch": 0.38143959791778853,
      "grad_norm": 2.3992090225219727,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 4250
    },
    {
      "epoch": 0.3815293484114163,
      "grad_norm": 2.5226104259490967,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 4251
    },
    {
      "epoch": 0.381619098905044,
      "grad_norm": 3.0306105613708496,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 4252
    },
    {
      "epoch": 0.38170884939867167,
      "grad_norm": 2.9352571964263916,
      "learning_rate": 0.0,
      "loss": 0.7649,
      "step": 4253
    },
    {
      "epoch": 0.3817985998922994,
      "grad_norm": 2.0859177112579346,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 4254
    },
    {
      "epoch": 0.3818883503859271,
      "grad_norm": 2.925706386566162,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 4255
    },
    {
      "epoch": 0.38197810087955486,
      "grad_norm": 2.778313636779785,
      "learning_rate": 0.0,
      "loss": 0.797,
      "step": 4256
    },
    {
      "epoch": 0.38206785137318255,
      "grad_norm": 2.407869338989258,
      "learning_rate": 0.0,
      "loss": 0.7546,
      "step": 4257
    },
    {
      "epoch": 0.38215760186681025,
      "grad_norm": 3.5911853313446045,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 4258
    },
    {
      "epoch": 0.382247352360438,
      "grad_norm": 3.300525188446045,
      "learning_rate": 0.0,
      "loss": 0.7779,
      "step": 4259
    },
    {
      "epoch": 0.3823371028540657,
      "grad_norm": 3.257171154022217,
      "learning_rate": 0.0,
      "loss": 0.781,
      "step": 4260
    },
    {
      "epoch": 0.3824268533476934,
      "grad_norm": 2.7412126064300537,
      "learning_rate": 0.0,
      "loss": 0.7951,
      "step": 4261
    },
    {
      "epoch": 0.38251660384132113,
      "grad_norm": 3.0891075134277344,
      "learning_rate": 0.0,
      "loss": 0.7657,
      "step": 4262
    },
    {
      "epoch": 0.38260635433494883,
      "grad_norm": 2.304084062576294,
      "learning_rate": 0.0,
      "loss": 0.839,
      "step": 4263
    },
    {
      "epoch": 0.3826961048285766,
      "grad_norm": 3.287632942199707,
      "learning_rate": 0.0,
      "loss": 0.7676,
      "step": 4264
    },
    {
      "epoch": 0.38278585532220427,
      "grad_norm": 2.409454822540283,
      "learning_rate": 0.0,
      "loss": 0.7034,
      "step": 4265
    },
    {
      "epoch": 0.38287560581583197,
      "grad_norm": 3.023944854736328,
      "learning_rate": 0.0,
      "loss": 0.6752,
      "step": 4266
    },
    {
      "epoch": 0.3829653563094597,
      "grad_norm": 2.7871737480163574,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 4267
    },
    {
      "epoch": 0.3830551068030874,
      "grad_norm": 2.92545747756958,
      "learning_rate": 0.0,
      "loss": 0.7566,
      "step": 4268
    },
    {
      "epoch": 0.38314485729671516,
      "grad_norm": 2.419022560119629,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 4269
    },
    {
      "epoch": 0.38323460779034285,
      "grad_norm": 2.396472930908203,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 4270
    },
    {
      "epoch": 0.38332435828397055,
      "grad_norm": 2.73392391204834,
      "learning_rate": 0.0,
      "loss": 0.7823,
      "step": 4271
    },
    {
      "epoch": 0.3834141087775983,
      "grad_norm": 2.5734126567840576,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 4272
    },
    {
      "epoch": 0.383503859271226,
      "grad_norm": 2.9525983333587646,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 4273
    },
    {
      "epoch": 0.3835936097648537,
      "grad_norm": 2.4364607334136963,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 4274
    },
    {
      "epoch": 0.38368336025848143,
      "grad_norm": 6.222799777984619,
      "learning_rate": 0.0,
      "loss": 0.8009,
      "step": 4275
    },
    {
      "epoch": 0.3837731107521091,
      "grad_norm": 2.8843610286712646,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 4276
    },
    {
      "epoch": 0.3838628612457369,
      "grad_norm": 2.8685243129730225,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 4277
    },
    {
      "epoch": 0.38395261173936457,
      "grad_norm": 10.040005683898926,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 4278
    },
    {
      "epoch": 0.38404236223299226,
      "grad_norm": 2.2321362495422363,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 4279
    },
    {
      "epoch": 0.38413211272662,
      "grad_norm": 3.510401487350464,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 4280
    },
    {
      "epoch": 0.3842218632202477,
      "grad_norm": 2.3055660724639893,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 4281
    },
    {
      "epoch": 0.3843116137138754,
      "grad_norm": 2.8452718257904053,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 4282
    },
    {
      "epoch": 0.38440136420750315,
      "grad_norm": 2.4355294704437256,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 4283
    },
    {
      "epoch": 0.38449111470113084,
      "grad_norm": 2.5270867347717285,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 4284
    },
    {
      "epoch": 0.3845808651947586,
      "grad_norm": 2.880706548690796,
      "learning_rate": 0.0,
      "loss": 0.8347,
      "step": 4285
    },
    {
      "epoch": 0.3846706156883863,
      "grad_norm": 2.1683452129364014,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 4286
    },
    {
      "epoch": 0.384760366182014,
      "grad_norm": 2.382930278778076,
      "learning_rate": 0.0,
      "loss": 0.7664,
      "step": 4287
    },
    {
      "epoch": 0.38485011667564173,
      "grad_norm": 3.4447357654571533,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 4288
    },
    {
      "epoch": 0.3849398671692694,
      "grad_norm": 3.0932064056396484,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 4289
    },
    {
      "epoch": 0.3850296176628972,
      "grad_norm": 2.329555034637451,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 4290
    },
    {
      "epoch": 0.38511936815652487,
      "grad_norm": 3.0183119773864746,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 4291
    },
    {
      "epoch": 0.38520911865015256,
      "grad_norm": 3.0006823539733887,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 4292
    },
    {
      "epoch": 0.3852988691437803,
      "grad_norm": 2.188930034637451,
      "learning_rate": 0.0,
      "loss": 0.7393,
      "step": 4293
    },
    {
      "epoch": 0.385388619637408,
      "grad_norm": 2.558058261871338,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 4294
    },
    {
      "epoch": 0.3854783701310357,
      "grad_norm": 3.457643985748291,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 4295
    },
    {
      "epoch": 0.38556812062466345,
      "grad_norm": 4.0653181076049805,
      "learning_rate": 0.0,
      "loss": 0.7124,
      "step": 4296
    },
    {
      "epoch": 0.38565787111829114,
      "grad_norm": 2.828890562057495,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 4297
    },
    {
      "epoch": 0.3857476216119189,
      "grad_norm": 6.676628589630127,
      "learning_rate": 0.0,
      "loss": 0.7929,
      "step": 4298
    },
    {
      "epoch": 0.3858373721055466,
      "grad_norm": 5.941256046295166,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 4299
    },
    {
      "epoch": 0.3859271225991743,
      "grad_norm": 3.352492332458496,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 4300
    },
    {
      "epoch": 0.386016873092802,
      "grad_norm": 3.3846216201782227,
      "learning_rate": 0.0,
      "loss": 0.7375,
      "step": 4301
    },
    {
      "epoch": 0.3861066235864297,
      "grad_norm": 3.50854229927063,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 4302
    },
    {
      "epoch": 0.3861963740800574,
      "grad_norm": 3.3473329544067383,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 4303
    },
    {
      "epoch": 0.38628612457368516,
      "grad_norm": 2.2782859802246094,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 4304
    },
    {
      "epoch": 0.38637587506731286,
      "grad_norm": 2.1643574237823486,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 4305
    },
    {
      "epoch": 0.3864656255609406,
      "grad_norm": 3.204310655593872,
      "learning_rate": 0.0,
      "loss": 0.6865,
      "step": 4306
    },
    {
      "epoch": 0.3865553760545683,
      "grad_norm": 3.288001775741577,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 4307
    },
    {
      "epoch": 0.386645126548196,
      "grad_norm": 5.336968421936035,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 4308
    },
    {
      "epoch": 0.38673487704182374,
      "grad_norm": 2.309150218963623,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 4309
    },
    {
      "epoch": 0.38682462753545144,
      "grad_norm": 2.592663049697876,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 4310
    },
    {
      "epoch": 0.3869143780290792,
      "grad_norm": 2.4958460330963135,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 4311
    },
    {
      "epoch": 0.3870041285227069,
      "grad_norm": 3.775502920150757,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 4312
    },
    {
      "epoch": 0.3870938790163346,
      "grad_norm": 3.1693859100341797,
      "learning_rate": 0.0,
      "loss": 0.7937,
      "step": 4313
    },
    {
      "epoch": 0.3871836295099623,
      "grad_norm": 2.9845941066741943,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 4314
    },
    {
      "epoch": 0.38727338000359,
      "grad_norm": 2.6074273586273193,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 4315
    },
    {
      "epoch": 0.3873631304972177,
      "grad_norm": 3.670971632003784,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 4316
    },
    {
      "epoch": 0.38745288099084546,
      "grad_norm": 3.1913836002349854,
      "learning_rate": 0.0,
      "loss": 0.8256,
      "step": 4317
    },
    {
      "epoch": 0.38754263148447315,
      "grad_norm": 3.258234739303589,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 4318
    },
    {
      "epoch": 0.3876323819781009,
      "grad_norm": 3.575927495956421,
      "learning_rate": 0.0,
      "loss": 0.7876,
      "step": 4319
    },
    {
      "epoch": 0.3877221324717286,
      "grad_norm": 3.0491690635681152,
      "learning_rate": 0.0,
      "loss": 0.6755,
      "step": 4320
    },
    {
      "epoch": 0.3878118829653563,
      "grad_norm": 2.9719057083129883,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 4321
    },
    {
      "epoch": 0.38790163345898404,
      "grad_norm": 10.562131881713867,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 4322
    },
    {
      "epoch": 0.38799138395261173,
      "grad_norm": 3.6497323513031006,
      "learning_rate": 0.0,
      "loss": 0.7867,
      "step": 4323
    },
    {
      "epoch": 0.38808113444623943,
      "grad_norm": 2.8713767528533936,
      "learning_rate": 0.0,
      "loss": 0.801,
      "step": 4324
    },
    {
      "epoch": 0.3881708849398672,
      "grad_norm": 2.9219603538513184,
      "learning_rate": 0.0,
      "loss": 0.6709,
      "step": 4325
    },
    {
      "epoch": 0.38826063543349487,
      "grad_norm": 3.051041603088379,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 4326
    },
    {
      "epoch": 0.3883503859271226,
      "grad_norm": 2.385606527328491,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 4327
    },
    {
      "epoch": 0.3884401364207503,
      "grad_norm": 2.6807000637054443,
      "learning_rate": 0.0,
      "loss": 0.8286,
      "step": 4328
    },
    {
      "epoch": 0.388529886914378,
      "grad_norm": 5.475252628326416,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 4329
    },
    {
      "epoch": 0.38861963740800576,
      "grad_norm": 4.028135776519775,
      "learning_rate": 0.0,
      "loss": 0.752,
      "step": 4330
    },
    {
      "epoch": 0.38870938790163345,
      "grad_norm": 3.060178756713867,
      "learning_rate": 0.0,
      "loss": 0.7587,
      "step": 4331
    },
    {
      "epoch": 0.3887991383952612,
      "grad_norm": 6.423725128173828,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 4332
    },
    {
      "epoch": 0.3888888888888889,
      "grad_norm": 5.732402801513672,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 4333
    },
    {
      "epoch": 0.3889786393825166,
      "grad_norm": 3.191288948059082,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 4334
    },
    {
      "epoch": 0.38906838987614434,
      "grad_norm": 2.711653470993042,
      "learning_rate": 0.0,
      "loss": 0.743,
      "step": 4335
    },
    {
      "epoch": 0.38915814036977203,
      "grad_norm": 3.044377326965332,
      "learning_rate": 0.0,
      "loss": 0.7815,
      "step": 4336
    },
    {
      "epoch": 0.3892478908633997,
      "grad_norm": 3.3241806030273438,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 4337
    },
    {
      "epoch": 0.3893376413570275,
      "grad_norm": 2.666095018386841,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 4338
    },
    {
      "epoch": 0.38942739185065517,
      "grad_norm": 3.128605365753174,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 4339
    },
    {
      "epoch": 0.3895171423442829,
      "grad_norm": 2.535521984100342,
      "learning_rate": 0.0,
      "loss": 0.7093,
      "step": 4340
    },
    {
      "epoch": 0.3896068928379106,
      "grad_norm": 3.2472732067108154,
      "learning_rate": 0.0,
      "loss": 0.8146,
      "step": 4341
    },
    {
      "epoch": 0.3896966433315383,
      "grad_norm": 3.4320216178894043,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 4342
    },
    {
      "epoch": 0.38978639382516606,
      "grad_norm": 2.2840442657470703,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 4343
    },
    {
      "epoch": 0.38987614431879375,
      "grad_norm": 3.233227014541626,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 4344
    },
    {
      "epoch": 0.38996589481242144,
      "grad_norm": 2.705575942993164,
      "learning_rate": 0.0,
      "loss": 0.7702,
      "step": 4345
    },
    {
      "epoch": 0.3900556453060492,
      "grad_norm": 2.946366310119629,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 4346
    },
    {
      "epoch": 0.3901453957996769,
      "grad_norm": 2.673099994659424,
      "learning_rate": 0.0,
      "loss": 0.6806,
      "step": 4347
    },
    {
      "epoch": 0.39023514629330464,
      "grad_norm": 3.1085824966430664,
      "learning_rate": 0.0,
      "loss": 0.7691,
      "step": 4348
    },
    {
      "epoch": 0.39032489678693233,
      "grad_norm": 3.5932252407073975,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 4349
    },
    {
      "epoch": 0.39041464728056,
      "grad_norm": 2.759251117706299,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 4350
    },
    {
      "epoch": 0.3905043977741878,
      "grad_norm": 4.714598655700684,
      "learning_rate": 0.0,
      "loss": 0.8084,
      "step": 4351
    },
    {
      "epoch": 0.39059414826781547,
      "grad_norm": 3.2901744842529297,
      "learning_rate": 0.0,
      "loss": 0.8113,
      "step": 4352
    },
    {
      "epoch": 0.39068389876144316,
      "grad_norm": 3.039677858352661,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 4353
    },
    {
      "epoch": 0.3907736492550709,
      "grad_norm": 2.8478844165802,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 4354
    },
    {
      "epoch": 0.3908633997486986,
      "grad_norm": 4.660661697387695,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 4355
    },
    {
      "epoch": 0.39095315024232635,
      "grad_norm": 7.198916912078857,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 4356
    },
    {
      "epoch": 0.39104290073595405,
      "grad_norm": 3.55400013923645,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 4357
    },
    {
      "epoch": 0.39113265122958174,
      "grad_norm": 3.247267007827759,
      "learning_rate": 0.0,
      "loss": 0.7186,
      "step": 4358
    },
    {
      "epoch": 0.3912224017232095,
      "grad_norm": 2.8601224422454834,
      "learning_rate": 0.0,
      "loss": 0.8776,
      "step": 4359
    },
    {
      "epoch": 0.3913121522168372,
      "grad_norm": 13.655306816101074,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 4360
    },
    {
      "epoch": 0.39140190271046493,
      "grad_norm": 3.9300308227539062,
      "learning_rate": 0.0,
      "loss": 0.8268,
      "step": 4361
    },
    {
      "epoch": 0.3914916532040926,
      "grad_norm": 14.133074760437012,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 4362
    },
    {
      "epoch": 0.3915814036977203,
      "grad_norm": 5.136984348297119,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 4363
    },
    {
      "epoch": 0.39167115419134807,
      "grad_norm": 3.8831138610839844,
      "learning_rate": 0.0,
      "loss": 0.6911,
      "step": 4364
    },
    {
      "epoch": 0.39176090468497576,
      "grad_norm": 3.027834892272949,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 4365
    },
    {
      "epoch": 0.39185065517860346,
      "grad_norm": 2.2808425426483154,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 4366
    },
    {
      "epoch": 0.3919404056722312,
      "grad_norm": 2.8524184226989746,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 4367
    },
    {
      "epoch": 0.3920301561658589,
      "grad_norm": 2.6394765377044678,
      "learning_rate": 0.0,
      "loss": 0.6774,
      "step": 4368
    },
    {
      "epoch": 0.39211990665948665,
      "grad_norm": 3.5528178215026855,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 4369
    },
    {
      "epoch": 0.39220965715311434,
      "grad_norm": 2.32232928276062,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 4370
    },
    {
      "epoch": 0.39229940764674204,
      "grad_norm": 4.463796615600586,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 4371
    },
    {
      "epoch": 0.3923891581403698,
      "grad_norm": 3.9800548553466797,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 4372
    },
    {
      "epoch": 0.3924789086339975,
      "grad_norm": 3.986450433731079,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 4373
    },
    {
      "epoch": 0.3925686591276252,
      "grad_norm": 2.730882406234741,
      "learning_rate": 0.0,
      "loss": 0.7952,
      "step": 4374
    },
    {
      "epoch": 0.3926584096212529,
      "grad_norm": 3.1310973167419434,
      "learning_rate": 0.0,
      "loss": 0.7204,
      "step": 4375
    },
    {
      "epoch": 0.3927481601148806,
      "grad_norm": 10.26452350616455,
      "learning_rate": 0.0,
      "loss": 0.6871,
      "step": 4376
    },
    {
      "epoch": 0.39283791060850837,
      "grad_norm": 3.075605869293213,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 4377
    },
    {
      "epoch": 0.39292766110213606,
      "grad_norm": 2.9730632305145264,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 4378
    },
    {
      "epoch": 0.39301741159576375,
      "grad_norm": 2.701338291168213,
      "learning_rate": 0.0,
      "loss": 0.7468,
      "step": 4379
    },
    {
      "epoch": 0.3931071620893915,
      "grad_norm": 4.150375843048096,
      "learning_rate": 0.0,
      "loss": 0.817,
      "step": 4380
    },
    {
      "epoch": 0.3931969125830192,
      "grad_norm": 5.530220031738281,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 4381
    },
    {
      "epoch": 0.39328666307664695,
      "grad_norm": 4.289727210998535,
      "learning_rate": 0.0,
      "loss": 0.747,
      "step": 4382
    },
    {
      "epoch": 0.39337641357027464,
      "grad_norm": 2.366551399230957,
      "learning_rate": 0.0,
      "loss": 0.8161,
      "step": 4383
    },
    {
      "epoch": 0.39346616406390234,
      "grad_norm": 3.42604660987854,
      "learning_rate": 0.0,
      "loss": 0.7199,
      "step": 4384
    },
    {
      "epoch": 0.3935559145575301,
      "grad_norm": 6.014654636383057,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 4385
    },
    {
      "epoch": 0.3936456650511578,
      "grad_norm": 3.03908109664917,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 4386
    },
    {
      "epoch": 0.39373541554478547,
      "grad_norm": 3.1789684295654297,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 4387
    },
    {
      "epoch": 0.3938251660384132,
      "grad_norm": 3.1979148387908936,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 4388
    },
    {
      "epoch": 0.3939149165320409,
      "grad_norm": 2.5242762565612793,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 4389
    },
    {
      "epoch": 0.39400466702566866,
      "grad_norm": 2.3029003143310547,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 4390
    },
    {
      "epoch": 0.39409441751929636,
      "grad_norm": 3.7107341289520264,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 4391
    },
    {
      "epoch": 0.39418416801292405,
      "grad_norm": 3.564723253250122,
      "learning_rate": 0.0,
      "loss": 0.6656,
      "step": 4392
    },
    {
      "epoch": 0.3942739185065518,
      "grad_norm": 3.430293083190918,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 4393
    },
    {
      "epoch": 0.3943636690001795,
      "grad_norm": 24.348182678222656,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 4394
    },
    {
      "epoch": 0.3944534194938072,
      "grad_norm": 4.241549491882324,
      "learning_rate": 0.0,
      "loss": 0.7648,
      "step": 4395
    },
    {
      "epoch": 0.39454316998743494,
      "grad_norm": 4.450930118560791,
      "learning_rate": 0.0,
      "loss": 0.8061,
      "step": 4396
    },
    {
      "epoch": 0.39463292048106263,
      "grad_norm": 8.048430442810059,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 4397
    },
    {
      "epoch": 0.3947226709746904,
      "grad_norm": 3.709648370742798,
      "learning_rate": 0.0,
      "loss": 0.768,
      "step": 4398
    },
    {
      "epoch": 0.3948124214683181,
      "grad_norm": 2.8124115467071533,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 4399
    },
    {
      "epoch": 0.39490217196194577,
      "grad_norm": 3.0592238903045654,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 4400
    },
    {
      "epoch": 0.3949919224555735,
      "grad_norm": 8.068060874938965,
      "learning_rate": 0.0,
      "loss": 0.7991,
      "step": 4401
    },
    {
      "epoch": 0.3950816729492012,
      "grad_norm": 3.58500599861145,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 4402
    },
    {
      "epoch": 0.39517142344282896,
      "grad_norm": 4.851542949676514,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 4403
    },
    {
      "epoch": 0.39526117393645666,
      "grad_norm": 2.059957265853882,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 4404
    },
    {
      "epoch": 0.39535092443008435,
      "grad_norm": 3.494601011276245,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 4405
    },
    {
      "epoch": 0.3954406749237121,
      "grad_norm": 3.40470552444458,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 4406
    },
    {
      "epoch": 0.3955304254173398,
      "grad_norm": 4.47633695602417,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 4407
    },
    {
      "epoch": 0.3956201759109675,
      "grad_norm": 3.2383511066436768,
      "learning_rate": 0.0,
      "loss": 0.6892,
      "step": 4408
    },
    {
      "epoch": 0.39570992640459524,
      "grad_norm": 3.0123555660247803,
      "learning_rate": 0.0,
      "loss": 0.781,
      "step": 4409
    },
    {
      "epoch": 0.39579967689822293,
      "grad_norm": 3.1638975143432617,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 4410
    },
    {
      "epoch": 0.3958894273918507,
      "grad_norm": 2.7256152629852295,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 4411
    },
    {
      "epoch": 0.3959791778854784,
      "grad_norm": 2.3436639308929443,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 4412
    },
    {
      "epoch": 0.39606892837910607,
      "grad_norm": 3.1596477031707764,
      "learning_rate": 0.0,
      "loss": 0.6738,
      "step": 4413
    },
    {
      "epoch": 0.3961586788727338,
      "grad_norm": 3.006770133972168,
      "learning_rate": 0.0,
      "loss": 0.7736,
      "step": 4414
    },
    {
      "epoch": 0.3962484293663615,
      "grad_norm": 6.626667499542236,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 4415
    },
    {
      "epoch": 0.3963381798599892,
      "grad_norm": 3.0629308223724365,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 4416
    },
    {
      "epoch": 0.39642793035361695,
      "grad_norm": 2.646792411804199,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 4417
    },
    {
      "epoch": 0.39651768084724465,
      "grad_norm": 2.8734540939331055,
      "learning_rate": 0.0,
      "loss": 0.8259,
      "step": 4418
    },
    {
      "epoch": 0.3966074313408724,
      "grad_norm": 7.339529991149902,
      "learning_rate": 0.0,
      "loss": 0.7896,
      "step": 4419
    },
    {
      "epoch": 0.3966971818345001,
      "grad_norm": 3.7184269428253174,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 4420
    },
    {
      "epoch": 0.3967869323281278,
      "grad_norm": 2.8760530948638916,
      "learning_rate": 0.0,
      "loss": 0.7661,
      "step": 4421
    },
    {
      "epoch": 0.39687668282175553,
      "grad_norm": 4.545596122741699,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 4422
    },
    {
      "epoch": 0.3969664333153832,
      "grad_norm": 3.6825342178344727,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 4423
    },
    {
      "epoch": 0.397056183809011,
      "grad_norm": 2.695895195007324,
      "learning_rate": 0.0,
      "loss": 0.788,
      "step": 4424
    },
    {
      "epoch": 0.39714593430263867,
      "grad_norm": 2.609131097793579,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 4425
    },
    {
      "epoch": 0.39723568479626636,
      "grad_norm": 3.965385913848877,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 4426
    },
    {
      "epoch": 0.3973254352898941,
      "grad_norm": 4.152858734130859,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 4427
    },
    {
      "epoch": 0.3974151857835218,
      "grad_norm": 3.553884983062744,
      "learning_rate": 0.0,
      "loss": 0.834,
      "step": 4428
    },
    {
      "epoch": 0.3975049362771495,
      "grad_norm": 3.5016331672668457,
      "learning_rate": 0.0,
      "loss": 0.754,
      "step": 4429
    },
    {
      "epoch": 0.39759468677077725,
      "grad_norm": 8.581611633300781,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 4430
    },
    {
      "epoch": 0.39768443726440494,
      "grad_norm": 3.3468711376190186,
      "learning_rate": 0.0,
      "loss": 0.6972,
      "step": 4431
    },
    {
      "epoch": 0.3977741877580327,
      "grad_norm": 3.8485770225524902,
      "learning_rate": 0.0,
      "loss": 0.8155,
      "step": 4432
    },
    {
      "epoch": 0.3978639382516604,
      "grad_norm": 3.1569414138793945,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 4433
    },
    {
      "epoch": 0.3979536887452881,
      "grad_norm": 2.1122682094573975,
      "learning_rate": 0.0,
      "loss": 0.7003,
      "step": 4434
    },
    {
      "epoch": 0.39804343923891583,
      "grad_norm": 3.115788221359253,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 4435
    },
    {
      "epoch": 0.3981331897325435,
      "grad_norm": 3.707653522491455,
      "learning_rate": 0.0,
      "loss": 0.7199,
      "step": 4436
    },
    {
      "epoch": 0.3982229402261712,
      "grad_norm": 4.684394359588623,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 4437
    },
    {
      "epoch": 0.39831269071979897,
      "grad_norm": 3.150604248046875,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 4438
    },
    {
      "epoch": 0.39840244121342666,
      "grad_norm": 2.9987411499023438,
      "learning_rate": 0.0,
      "loss": 0.733,
      "step": 4439
    },
    {
      "epoch": 0.3984921917070544,
      "grad_norm": 25.839065551757812,
      "learning_rate": 0.0,
      "loss": 0.7456,
      "step": 4440
    },
    {
      "epoch": 0.3985819422006821,
      "grad_norm": 3.516300678253174,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 4441
    },
    {
      "epoch": 0.3986716926943098,
      "grad_norm": 4.111924171447754,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 4442
    },
    {
      "epoch": 0.39876144318793755,
      "grad_norm": 3.8155934810638428,
      "learning_rate": 0.0,
      "loss": 0.7101,
      "step": 4443
    },
    {
      "epoch": 0.39885119368156524,
      "grad_norm": 2.1618709564208984,
      "learning_rate": 0.0,
      "loss": 0.7627,
      "step": 4444
    },
    {
      "epoch": 0.398940944175193,
      "grad_norm": 2.6861255168914795,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 4445
    },
    {
      "epoch": 0.3990306946688207,
      "grad_norm": 2.215489149093628,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 4446
    },
    {
      "epoch": 0.3991204451624484,
      "grad_norm": 3.476433277130127,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 4447
    },
    {
      "epoch": 0.39921019565607613,
      "grad_norm": 2.5902698040008545,
      "learning_rate": 0.0,
      "loss": 0.8029,
      "step": 4448
    },
    {
      "epoch": 0.3992999461497038,
      "grad_norm": 3.028240919113159,
      "learning_rate": 0.0,
      "loss": 0.7418,
      "step": 4449
    },
    {
      "epoch": 0.3993896966433315,
      "grad_norm": 3.2896857261657715,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 4450
    },
    {
      "epoch": 0.39947944713695926,
      "grad_norm": 2.6060400009155273,
      "learning_rate": 0.0,
      "loss": 0.8389,
      "step": 4451
    },
    {
      "epoch": 0.39956919763058696,
      "grad_norm": 2.6506221294403076,
      "learning_rate": 0.0,
      "loss": 0.8033,
      "step": 4452
    },
    {
      "epoch": 0.3996589481242147,
      "grad_norm": 2.6541616916656494,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 4453
    },
    {
      "epoch": 0.3997486986178424,
      "grad_norm": 2.6585357189178467,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 4454
    },
    {
      "epoch": 0.3998384491114701,
      "grad_norm": 2.8786115646362305,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 4455
    },
    {
      "epoch": 0.39992819960509784,
      "grad_norm": 4.320756435394287,
      "learning_rate": 0.0,
      "loss": 0.8033,
      "step": 4456
    },
    {
      "epoch": 0.40001795009872554,
      "grad_norm": 3.920658588409424,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 4457
    },
    {
      "epoch": 0.40010770059235323,
      "grad_norm": 3.1609854698181152,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 4458
    },
    {
      "epoch": 0.400197451085981,
      "grad_norm": 2.2217676639556885,
      "learning_rate": 0.0,
      "loss": 0.6901,
      "step": 4459
    },
    {
      "epoch": 0.4002872015796087,
      "grad_norm": 3.55155873298645,
      "learning_rate": 0.0,
      "loss": 0.663,
      "step": 4460
    },
    {
      "epoch": 0.4003769520732364,
      "grad_norm": 2.478585958480835,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 4461
    },
    {
      "epoch": 0.4004667025668641,
      "grad_norm": 2.850553512573242,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 4462
    },
    {
      "epoch": 0.4005564530604918,
      "grad_norm": 2.0565688610076904,
      "learning_rate": 0.0,
      "loss": 0.7627,
      "step": 4463
    },
    {
      "epoch": 0.40064620355411956,
      "grad_norm": 7.879218578338623,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 4464
    },
    {
      "epoch": 0.40073595404774726,
      "grad_norm": 3.362013578414917,
      "learning_rate": 0.0,
      "loss": 0.7882,
      "step": 4465
    },
    {
      "epoch": 0.400825704541375,
      "grad_norm": 3.051710367202759,
      "learning_rate": 0.0,
      "loss": 0.7672,
      "step": 4466
    },
    {
      "epoch": 0.4009154550350027,
      "grad_norm": 2.084153652191162,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 4467
    },
    {
      "epoch": 0.4010052055286304,
      "grad_norm": 3.5488176345825195,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 4468
    },
    {
      "epoch": 0.40109495602225814,
      "grad_norm": 3.3822991847991943,
      "learning_rate": 0.0,
      "loss": 0.785,
      "step": 4469
    },
    {
      "epoch": 0.40118470651588584,
      "grad_norm": 3.061586618423462,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 4470
    },
    {
      "epoch": 0.40127445700951353,
      "grad_norm": 2.9645285606384277,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 4471
    },
    {
      "epoch": 0.4013642075031413,
      "grad_norm": 3.2317278385162354,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 4472
    },
    {
      "epoch": 0.401453957996769,
      "grad_norm": 4.171561241149902,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 4473
    },
    {
      "epoch": 0.4015437084903967,
      "grad_norm": 4.6605658531188965,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 4474
    },
    {
      "epoch": 0.4016334589840244,
      "grad_norm": 2.916160821914673,
      "learning_rate": 0.0,
      "loss": 0.7922,
      "step": 4475
    },
    {
      "epoch": 0.4017232094776521,
      "grad_norm": 2.6584548950195312,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 4476
    },
    {
      "epoch": 0.40181295997127986,
      "grad_norm": 2.280945301055908,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 4477
    },
    {
      "epoch": 0.40190271046490755,
      "grad_norm": 4.107948303222656,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 4478
    },
    {
      "epoch": 0.40199246095853525,
      "grad_norm": 2.7213168144226074,
      "learning_rate": 0.0,
      "loss": 0.674,
      "step": 4479
    },
    {
      "epoch": 0.402082211452163,
      "grad_norm": 3.2491614818573,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 4480
    },
    {
      "epoch": 0.4021719619457907,
      "grad_norm": 2.7627408504486084,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 4481
    },
    {
      "epoch": 0.40226171243941844,
      "grad_norm": 3.2267847061157227,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 4482
    },
    {
      "epoch": 0.40235146293304613,
      "grad_norm": 2.4833860397338867,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 4483
    },
    {
      "epoch": 0.4024412134266738,
      "grad_norm": 2.6536107063293457,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 4484
    },
    {
      "epoch": 0.4025309639203016,
      "grad_norm": 3.0197622776031494,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 4485
    },
    {
      "epoch": 0.40262071441392927,
      "grad_norm": 4.254358291625977,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 4486
    },
    {
      "epoch": 0.40271046490755696,
      "grad_norm": 2.936708450317383,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 4487
    },
    {
      "epoch": 0.4028002154011847,
      "grad_norm": 3.66414475440979,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 4488
    },
    {
      "epoch": 0.4028899658948124,
      "grad_norm": 2.4572887420654297,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 4489
    },
    {
      "epoch": 0.40297971638844016,
      "grad_norm": 2.698845624923706,
      "learning_rate": 0.0,
      "loss": 0.7211,
      "step": 4490
    },
    {
      "epoch": 0.40306946688206785,
      "grad_norm": 2.6198573112487793,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 4491
    },
    {
      "epoch": 0.40315921737569554,
      "grad_norm": 3.1821253299713135,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 4492
    },
    {
      "epoch": 0.4032489678693233,
      "grad_norm": 2.767359495162964,
      "learning_rate": 0.0,
      "loss": 0.7719,
      "step": 4493
    },
    {
      "epoch": 0.403338718362951,
      "grad_norm": 2.7146148681640625,
      "learning_rate": 0.0,
      "loss": 0.6664,
      "step": 4494
    },
    {
      "epoch": 0.40342846885657874,
      "grad_norm": 2.434262752532959,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 4495
    },
    {
      "epoch": 0.40351821935020643,
      "grad_norm": 28.99799919128418,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 4496
    },
    {
      "epoch": 0.4036079698438341,
      "grad_norm": 2.9908008575439453,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 4497
    },
    {
      "epoch": 0.4036977203374619,
      "grad_norm": 8.00250244140625,
      "learning_rate": 0.0,
      "loss": 0.7916,
      "step": 4498
    },
    {
      "epoch": 0.40378747083108957,
      "grad_norm": 2.9938442707061768,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 4499
    },
    {
      "epoch": 0.40387722132471726,
      "grad_norm": 2.709251880645752,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 4500
    },
    {
      "epoch": 0.403966971818345,
      "grad_norm": 4.252358913421631,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 4501
    },
    {
      "epoch": 0.4040567223119727,
      "grad_norm": 2.905817985534668,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 4502
    },
    {
      "epoch": 0.40414647280560045,
      "grad_norm": 2.8453898429870605,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 4503
    },
    {
      "epoch": 0.40423622329922815,
      "grad_norm": 4.430549621582031,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 4504
    },
    {
      "epoch": 0.40432597379285584,
      "grad_norm": 5.456242561340332,
      "learning_rate": 0.0,
      "loss": 0.8504,
      "step": 4505
    },
    {
      "epoch": 0.4044157242864836,
      "grad_norm": 4.0429253578186035,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 4506
    },
    {
      "epoch": 0.4045054747801113,
      "grad_norm": 2.734612226486206,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 4507
    },
    {
      "epoch": 0.404595225273739,
      "grad_norm": 4.94826078414917,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 4508
    },
    {
      "epoch": 0.40468497576736673,
      "grad_norm": 3.3600013256073,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 4509
    },
    {
      "epoch": 0.4047747262609944,
      "grad_norm": 2.4850237369537354,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 4510
    },
    {
      "epoch": 0.40486447675462217,
      "grad_norm": 2.921846866607666,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 4511
    },
    {
      "epoch": 0.40495422724824987,
      "grad_norm": 2.556548595428467,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 4512
    },
    {
      "epoch": 0.40504397774187756,
      "grad_norm": 4.128906726837158,
      "learning_rate": 0.0,
      "loss": 0.7796,
      "step": 4513
    },
    {
      "epoch": 0.4051337282355053,
      "grad_norm": 3.181273937225342,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 4514
    },
    {
      "epoch": 0.405223478729133,
      "grad_norm": 5.197795391082764,
      "learning_rate": 0.0,
      "loss": 0.6647,
      "step": 4515
    },
    {
      "epoch": 0.40531322922276075,
      "grad_norm": 4.526033878326416,
      "learning_rate": 0.0,
      "loss": 0.8159,
      "step": 4516
    },
    {
      "epoch": 0.40540297971638845,
      "grad_norm": 2.819815158843994,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 4517
    },
    {
      "epoch": 0.40549273021001614,
      "grad_norm": 2.3223884105682373,
      "learning_rate": 0.0,
      "loss": 0.7647,
      "step": 4518
    },
    {
      "epoch": 0.4055824807036439,
      "grad_norm": 4.7129998207092285,
      "learning_rate": 0.0,
      "loss": 0.7973,
      "step": 4519
    },
    {
      "epoch": 0.4056722311972716,
      "grad_norm": 2.9775149822235107,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 4520
    },
    {
      "epoch": 0.4057619816908993,
      "grad_norm": 3.589698314666748,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 4521
    },
    {
      "epoch": 0.405851732184527,
      "grad_norm": 2.385450601577759,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 4522
    },
    {
      "epoch": 0.4059414826781547,
      "grad_norm": 2.698014259338379,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 4523
    },
    {
      "epoch": 0.40603123317178247,
      "grad_norm": 3.1440489292144775,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 4524
    },
    {
      "epoch": 0.40612098366541016,
      "grad_norm": 3.4748406410217285,
      "learning_rate": 0.0,
      "loss": 0.7872,
      "step": 4525
    },
    {
      "epoch": 0.40621073415903786,
      "grad_norm": 2.829108715057373,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 4526
    },
    {
      "epoch": 0.4063004846526656,
      "grad_norm": 3.398463010787964,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 4527
    },
    {
      "epoch": 0.4063902351462933,
      "grad_norm": 2.294696092605591,
      "learning_rate": 0.0,
      "loss": 0.7667,
      "step": 4528
    },
    {
      "epoch": 0.406479985639921,
      "grad_norm": 4.882735729217529,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 4529
    },
    {
      "epoch": 0.40656973613354874,
      "grad_norm": 2.6374073028564453,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 4530
    },
    {
      "epoch": 0.40665948662717644,
      "grad_norm": 2.7257041931152344,
      "learning_rate": 0.0,
      "loss": 0.7431,
      "step": 4531
    },
    {
      "epoch": 0.4067492371208042,
      "grad_norm": 3.5345587730407715,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 4532
    },
    {
      "epoch": 0.4068389876144319,
      "grad_norm": 2.9464898109436035,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 4533
    },
    {
      "epoch": 0.4069287381080596,
      "grad_norm": 3.155205726623535,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 4534
    },
    {
      "epoch": 0.4070184886016873,
      "grad_norm": 2.2817513942718506,
      "learning_rate": 0.0,
      "loss": 0.7346,
      "step": 4535
    },
    {
      "epoch": 0.407108239095315,
      "grad_norm": 2.6624040603637695,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 4536
    },
    {
      "epoch": 0.40719798958894277,
      "grad_norm": 7.045925140380859,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 4537
    },
    {
      "epoch": 0.40728774008257046,
      "grad_norm": 6.303225040435791,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 4538
    },
    {
      "epoch": 0.40737749057619815,
      "grad_norm": 3.0392403602600098,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 4539
    },
    {
      "epoch": 0.4074672410698259,
      "grad_norm": 2.6297695636749268,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 4540
    },
    {
      "epoch": 0.4075569915634536,
      "grad_norm": 4.83586311340332,
      "learning_rate": 0.0,
      "loss": 0.743,
      "step": 4541
    },
    {
      "epoch": 0.4076467420570813,
      "grad_norm": 7.224246501922607,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 4542
    },
    {
      "epoch": 0.40773649255070904,
      "grad_norm": 4.060956954956055,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 4543
    },
    {
      "epoch": 0.40782624304433673,
      "grad_norm": 2.9225573539733887,
      "learning_rate": 0.0,
      "loss": 0.7983,
      "step": 4544
    },
    {
      "epoch": 0.4079159935379645,
      "grad_norm": 2.4090254306793213,
      "learning_rate": 0.0,
      "loss": 0.817,
      "step": 4545
    },
    {
      "epoch": 0.4080057440315922,
      "grad_norm": 2.5245654582977295,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 4546
    },
    {
      "epoch": 0.40809549452521987,
      "grad_norm": 3.4535276889801025,
      "learning_rate": 0.0,
      "loss": 0.8182,
      "step": 4547
    },
    {
      "epoch": 0.4081852450188476,
      "grad_norm": 3.093197822570801,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 4548
    },
    {
      "epoch": 0.4082749955124753,
      "grad_norm": 3.216169595718384,
      "learning_rate": 0.0,
      "loss": 0.7702,
      "step": 4549
    },
    {
      "epoch": 0.408364746006103,
      "grad_norm": 4.727292060852051,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 4550
    },
    {
      "epoch": 0.40845449649973076,
      "grad_norm": 11.203668594360352,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 4551
    },
    {
      "epoch": 0.40854424699335845,
      "grad_norm": 3.930664539337158,
      "learning_rate": 0.0,
      "loss": 0.8078,
      "step": 4552
    },
    {
      "epoch": 0.4086339974869862,
      "grad_norm": 4.183338165283203,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 4553
    },
    {
      "epoch": 0.4087237479806139,
      "grad_norm": 2.864929676055908,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 4554
    },
    {
      "epoch": 0.4088134984742416,
      "grad_norm": 9.023148536682129,
      "learning_rate": 0.0,
      "loss": 0.6989,
      "step": 4555
    },
    {
      "epoch": 0.40890324896786934,
      "grad_norm": 2.6608211994171143,
      "learning_rate": 0.0,
      "loss": 0.7203,
      "step": 4556
    },
    {
      "epoch": 0.40899299946149703,
      "grad_norm": 4.226413249969482,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 4557
    },
    {
      "epoch": 0.4090827499551248,
      "grad_norm": 2.4952359199523926,
      "learning_rate": 0.0,
      "loss": 0.7838,
      "step": 4558
    },
    {
      "epoch": 0.4091725004487525,
      "grad_norm": 2.4086482524871826,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 4559
    },
    {
      "epoch": 0.40926225094238017,
      "grad_norm": 5.201790809631348,
      "learning_rate": 0.0,
      "loss": 0.7878,
      "step": 4560
    },
    {
      "epoch": 0.4093520014360079,
      "grad_norm": 3.71368670463562,
      "learning_rate": 0.0,
      "loss": 0.801,
      "step": 4561
    },
    {
      "epoch": 0.4094417519296356,
      "grad_norm": 4.586146831512451,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 4562
    },
    {
      "epoch": 0.4095315024232633,
      "grad_norm": 3.240851640701294,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 4563
    },
    {
      "epoch": 0.40962125291689105,
      "grad_norm": 3.7034525871276855,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 4564
    },
    {
      "epoch": 0.40971100341051875,
      "grad_norm": 3.8411195278167725,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 4565
    },
    {
      "epoch": 0.4098007539041465,
      "grad_norm": 3.1710057258605957,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 4566
    },
    {
      "epoch": 0.4098905043977742,
      "grad_norm": 2.684418201446533,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 4567
    },
    {
      "epoch": 0.4099802548914019,
      "grad_norm": 5.0154500007629395,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 4568
    },
    {
      "epoch": 0.41007000538502963,
      "grad_norm": 3.9481005668640137,
      "learning_rate": 0.0,
      "loss": 0.7788,
      "step": 4569
    },
    {
      "epoch": 0.41015975587865733,
      "grad_norm": 3.8097925186157227,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 4570
    },
    {
      "epoch": 0.410249506372285,
      "grad_norm": 6.817379951477051,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 4571
    },
    {
      "epoch": 0.41033925686591277,
      "grad_norm": 2.707376480102539,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 4572
    },
    {
      "epoch": 0.41042900735954047,
      "grad_norm": 2.2185542583465576,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 4573
    },
    {
      "epoch": 0.4105187578531682,
      "grad_norm": 6.630904674530029,
      "learning_rate": 0.0,
      "loss": 0.7401,
      "step": 4574
    },
    {
      "epoch": 0.4106085083467959,
      "grad_norm": 3.918616771697998,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 4575
    },
    {
      "epoch": 0.4106982588404236,
      "grad_norm": 2.8290212154388428,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 4576
    },
    {
      "epoch": 0.41078800933405135,
      "grad_norm": 2.4980781078338623,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 4577
    },
    {
      "epoch": 0.41087775982767905,
      "grad_norm": 4.274794578552246,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 4578
    },
    {
      "epoch": 0.4109675103213068,
      "grad_norm": 2.4824540615081787,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 4579
    },
    {
      "epoch": 0.4110572608149345,
      "grad_norm": 9.428016662597656,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 4580
    },
    {
      "epoch": 0.4111470113085622,
      "grad_norm": 3.430753231048584,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 4581
    },
    {
      "epoch": 0.41123676180218993,
      "grad_norm": 2.648365020751953,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 4582
    },
    {
      "epoch": 0.4113265122958176,
      "grad_norm": 3.379300355911255,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 4583
    },
    {
      "epoch": 0.4114162627894453,
      "grad_norm": 5.037209987640381,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 4584
    },
    {
      "epoch": 0.41150601328307307,
      "grad_norm": 3.742125988006592,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 4585
    },
    {
      "epoch": 0.41159576377670076,
      "grad_norm": 2.790785312652588,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 4586
    },
    {
      "epoch": 0.4116855142703285,
      "grad_norm": 8.150823593139648,
      "learning_rate": 0.0,
      "loss": 0.6899,
      "step": 4587
    },
    {
      "epoch": 0.4117752647639562,
      "grad_norm": 3.2841291427612305,
      "learning_rate": 0.0,
      "loss": 0.7826,
      "step": 4588
    },
    {
      "epoch": 0.4118650152575839,
      "grad_norm": 2.600065231323242,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 4589
    },
    {
      "epoch": 0.41195476575121165,
      "grad_norm": 4.0937089920043945,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 4590
    },
    {
      "epoch": 0.41204451624483934,
      "grad_norm": 5.179574966430664,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 4591
    },
    {
      "epoch": 0.41213426673846704,
      "grad_norm": 2.491969585418701,
      "learning_rate": 0.0,
      "loss": 0.7047,
      "step": 4592
    },
    {
      "epoch": 0.4122240172320948,
      "grad_norm": 2.7253870964050293,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 4593
    },
    {
      "epoch": 0.4123137677257225,
      "grad_norm": 2.39034104347229,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 4594
    },
    {
      "epoch": 0.41240351821935023,
      "grad_norm": 3.129270076751709,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 4595
    },
    {
      "epoch": 0.4124932687129779,
      "grad_norm": 11.969392776489258,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 4596
    },
    {
      "epoch": 0.4125830192066056,
      "grad_norm": 5.3397016525268555,
      "learning_rate": 0.0,
      "loss": 0.7764,
      "step": 4597
    },
    {
      "epoch": 0.41267276970023337,
      "grad_norm": 3.191225290298462,
      "learning_rate": 0.0,
      "loss": 0.7881,
      "step": 4598
    },
    {
      "epoch": 0.41276252019386106,
      "grad_norm": 2.87041974067688,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 4599
    },
    {
      "epoch": 0.41285227068748875,
      "grad_norm": 2.6509108543395996,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 4600
    },
    {
      "epoch": 0.4129420211811165,
      "grad_norm": 2.9357335567474365,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 4601
    },
    {
      "epoch": 0.4130317716747442,
      "grad_norm": 3.3795487880706787,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 4602
    },
    {
      "epoch": 0.41312152216837195,
      "grad_norm": 2.8542542457580566,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 4603
    },
    {
      "epoch": 0.41321127266199964,
      "grad_norm": 4.313778400421143,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 4604
    },
    {
      "epoch": 0.41330102315562733,
      "grad_norm": 3.1989352703094482,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 4605
    },
    {
      "epoch": 0.4133907736492551,
      "grad_norm": 4.258549690246582,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 4606
    },
    {
      "epoch": 0.4134805241428828,
      "grad_norm": 2.877761125564575,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 4607
    },
    {
      "epoch": 0.4135702746365105,
      "grad_norm": 2.0798428058624268,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 4608
    },
    {
      "epoch": 0.4136600251301382,
      "grad_norm": 3.1025242805480957,
      "learning_rate": 0.0,
      "loss": 0.7804,
      "step": 4609
    },
    {
      "epoch": 0.4137497756237659,
      "grad_norm": 2.8582911491394043,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 4610
    },
    {
      "epoch": 0.41383952611739366,
      "grad_norm": 2.783365249633789,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 4611
    },
    {
      "epoch": 0.41392927661102136,
      "grad_norm": 4.299332618713379,
      "learning_rate": 0.0,
      "loss": 0.7871,
      "step": 4612
    },
    {
      "epoch": 0.41401902710464905,
      "grad_norm": 5.897436141967773,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 4613
    },
    {
      "epoch": 0.4141087775982768,
      "grad_norm": 3.726391315460205,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 4614
    },
    {
      "epoch": 0.4141985280919045,
      "grad_norm": 2.360872745513916,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 4615
    },
    {
      "epoch": 0.41428827858553224,
      "grad_norm": 5.226831436157227,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 4616
    },
    {
      "epoch": 0.41437802907915994,
      "grad_norm": 2.4645793437957764,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 4617
    },
    {
      "epoch": 0.41446777957278763,
      "grad_norm": 2.4248759746551514,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 4618
    },
    {
      "epoch": 0.4145575300664154,
      "grad_norm": 3.709669828414917,
      "learning_rate": 0.0,
      "loss": 0.651,
      "step": 4619
    },
    {
      "epoch": 0.4146472805600431,
      "grad_norm": 2.5753228664398193,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 4620
    },
    {
      "epoch": 0.41473703105367077,
      "grad_norm": 3.1199734210968018,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 4621
    },
    {
      "epoch": 0.4148267815472985,
      "grad_norm": 4.730615139007568,
      "learning_rate": 0.0,
      "loss": 0.7656,
      "step": 4622
    },
    {
      "epoch": 0.4149165320409262,
      "grad_norm": 4.036073207855225,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 4623
    },
    {
      "epoch": 0.41500628253455396,
      "grad_norm": 4.42507266998291,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 4624
    },
    {
      "epoch": 0.41509603302818165,
      "grad_norm": 3.0953686237335205,
      "learning_rate": 0.0,
      "loss": 0.7301,
      "step": 4625
    },
    {
      "epoch": 0.41518578352180935,
      "grad_norm": 2.9459989070892334,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 4626
    },
    {
      "epoch": 0.4152755340154371,
      "grad_norm": 3.2209599018096924,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 4627
    },
    {
      "epoch": 0.4153652845090648,
      "grad_norm": 3.043057680130005,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 4628
    },
    {
      "epoch": 0.41545503500269254,
      "grad_norm": 2.4483275413513184,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 4629
    },
    {
      "epoch": 0.41554478549632023,
      "grad_norm": 2.98022198677063,
      "learning_rate": 0.0,
      "loss": 0.7772,
      "step": 4630
    },
    {
      "epoch": 0.41563453598994793,
      "grad_norm": 3.9764208793640137,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 4631
    },
    {
      "epoch": 0.4157242864835757,
      "grad_norm": 6.614599227905273,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 4632
    },
    {
      "epoch": 0.41581403697720337,
      "grad_norm": 3.1067609786987305,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 4633
    },
    {
      "epoch": 0.41590378747083107,
      "grad_norm": 2.543384313583374,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 4634
    },
    {
      "epoch": 0.4159935379644588,
      "grad_norm": 2.3653366565704346,
      "learning_rate": 0.0,
      "loss": 0.6811,
      "step": 4635
    },
    {
      "epoch": 0.4160832884580865,
      "grad_norm": 3.151543140411377,
      "learning_rate": 0.0,
      "loss": 0.764,
      "step": 4636
    },
    {
      "epoch": 0.41617303895171426,
      "grad_norm": 2.707998514175415,
      "learning_rate": 0.0,
      "loss": 0.7662,
      "step": 4637
    },
    {
      "epoch": 0.41626278944534195,
      "grad_norm": 2.9654173851013184,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 4638
    },
    {
      "epoch": 0.41635253993896965,
      "grad_norm": 2.9043641090393066,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 4639
    },
    {
      "epoch": 0.4164422904325974,
      "grad_norm": 2.696390151977539,
      "learning_rate": 0.0,
      "loss": 0.6763,
      "step": 4640
    },
    {
      "epoch": 0.4165320409262251,
      "grad_norm": 2.826489210128784,
      "learning_rate": 0.0,
      "loss": 0.7689,
      "step": 4641
    },
    {
      "epoch": 0.4166217914198528,
      "grad_norm": 3.357429265975952,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 4642
    },
    {
      "epoch": 0.41671154191348053,
      "grad_norm": 2.7443089485168457,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 4643
    },
    {
      "epoch": 0.4168012924071082,
      "grad_norm": 3.6130499839782715,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 4644
    },
    {
      "epoch": 0.416891042900736,
      "grad_norm": 4.418353080749512,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 4645
    },
    {
      "epoch": 0.41698079339436367,
      "grad_norm": 3.355736255645752,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 4646
    },
    {
      "epoch": 0.41707054388799136,
      "grad_norm": 4.907280445098877,
      "learning_rate": 0.0,
      "loss": 0.7176,
      "step": 4647
    },
    {
      "epoch": 0.4171602943816191,
      "grad_norm": 2.9501450061798096,
      "learning_rate": 0.0,
      "loss": 0.7505,
      "step": 4648
    },
    {
      "epoch": 0.4172500448752468,
      "grad_norm": 3.626511812210083,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 4649
    },
    {
      "epoch": 0.41733979536887456,
      "grad_norm": 2.656386613845825,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 4650
    },
    {
      "epoch": 0.41742954586250225,
      "grad_norm": 2.7957937717437744,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 4651
    },
    {
      "epoch": 0.41751929635612994,
      "grad_norm": 2.693166494369507,
      "learning_rate": 0.0,
      "loss": 0.7656,
      "step": 4652
    },
    {
      "epoch": 0.4176090468497577,
      "grad_norm": 3.2673981189727783,
      "learning_rate": 0.0,
      "loss": 0.7128,
      "step": 4653
    },
    {
      "epoch": 0.4176987973433854,
      "grad_norm": 2.500319242477417,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 4654
    },
    {
      "epoch": 0.4177885478370131,
      "grad_norm": 2.2547295093536377,
      "learning_rate": 0.0,
      "loss": 0.7664,
      "step": 4655
    },
    {
      "epoch": 0.41787829833064083,
      "grad_norm": 2.4227917194366455,
      "learning_rate": 0.0,
      "loss": 0.688,
      "step": 4656
    },
    {
      "epoch": 0.4179680488242685,
      "grad_norm": 3.192523717880249,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 4657
    },
    {
      "epoch": 0.4180577993178963,
      "grad_norm": 4.230035781860352,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 4658
    },
    {
      "epoch": 0.41814754981152397,
      "grad_norm": 7.634725093841553,
      "learning_rate": 0.0,
      "loss": 0.8006,
      "step": 4659
    },
    {
      "epoch": 0.41823730030515166,
      "grad_norm": 2.572012424468994,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 4660
    },
    {
      "epoch": 0.4183270507987794,
      "grad_norm": 2.8521828651428223,
      "learning_rate": 0.0,
      "loss": 0.6774,
      "step": 4661
    },
    {
      "epoch": 0.4184168012924071,
      "grad_norm": 2.7080142498016357,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 4662
    },
    {
      "epoch": 0.4185065517860348,
      "grad_norm": 3.3463876247406006,
      "learning_rate": 0.0,
      "loss": 0.7574,
      "step": 4663
    },
    {
      "epoch": 0.41859630227966255,
      "grad_norm": 2.8896284103393555,
      "learning_rate": 0.0,
      "loss": 0.751,
      "step": 4664
    },
    {
      "epoch": 0.41868605277329024,
      "grad_norm": 2.8489792346954346,
      "learning_rate": 0.0,
      "loss": 0.7599,
      "step": 4665
    },
    {
      "epoch": 0.418775803266918,
      "grad_norm": 3.7571845054626465,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 4666
    },
    {
      "epoch": 0.4188655537605457,
      "grad_norm": 2.7108547687530518,
      "learning_rate": 0.0,
      "loss": 0.6922,
      "step": 4667
    },
    {
      "epoch": 0.4189553042541734,
      "grad_norm": 4.310579299926758,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 4668
    },
    {
      "epoch": 0.4190450547478011,
      "grad_norm": 2.8241376876831055,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 4669
    },
    {
      "epoch": 0.4191348052414288,
      "grad_norm": 2.826162576675415,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 4670
    },
    {
      "epoch": 0.41922455573505657,
      "grad_norm": 2.788464307785034,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 4671
    },
    {
      "epoch": 0.41931430622868426,
      "grad_norm": 2.8503055572509766,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 4672
    },
    {
      "epoch": 0.41940405672231196,
      "grad_norm": 3.192253589630127,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 4673
    },
    {
      "epoch": 0.4194938072159397,
      "grad_norm": 2.685898542404175,
      "learning_rate": 0.0,
      "loss": 0.7756,
      "step": 4674
    },
    {
      "epoch": 0.4195835577095674,
      "grad_norm": 3.6203808784484863,
      "learning_rate": 0.0,
      "loss": 0.7465,
      "step": 4675
    },
    {
      "epoch": 0.4196733082031951,
      "grad_norm": 2.8361666202545166,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 4676
    },
    {
      "epoch": 0.41976305869682284,
      "grad_norm": 4.312153339385986,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 4677
    },
    {
      "epoch": 0.41985280919045054,
      "grad_norm": 3.7737948894500732,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 4678
    },
    {
      "epoch": 0.4199425596840783,
      "grad_norm": 2.791755199432373,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 4679
    },
    {
      "epoch": 0.420032310177706,
      "grad_norm": 2.8584957122802734,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 4680
    },
    {
      "epoch": 0.4201220606713337,
      "grad_norm": 2.3603298664093018,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 4681
    },
    {
      "epoch": 0.4202118111649614,
      "grad_norm": 3.1872057914733887,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 4682
    },
    {
      "epoch": 0.4203015616585891,
      "grad_norm": 2.4288928508758545,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 4683
    },
    {
      "epoch": 0.4203913121522168,
      "grad_norm": 3.0766780376434326,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 4684
    },
    {
      "epoch": 0.42048106264584456,
      "grad_norm": 4.976428985595703,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 4685
    },
    {
      "epoch": 0.42057081313947225,
      "grad_norm": 3.7541980743408203,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 4686
    },
    {
      "epoch": 0.4206605636331,
      "grad_norm": 3.3752188682556152,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 4687
    },
    {
      "epoch": 0.4207503141267277,
      "grad_norm": 2.932340145111084,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 4688
    },
    {
      "epoch": 0.4208400646203554,
      "grad_norm": 3.9830143451690674,
      "learning_rate": 0.0,
      "loss": 0.7199,
      "step": 4689
    },
    {
      "epoch": 0.42092981511398314,
      "grad_norm": 3.0450024604797363,
      "learning_rate": 0.0,
      "loss": 0.6742,
      "step": 4690
    },
    {
      "epoch": 0.42101956560761084,
      "grad_norm": 2.624971866607666,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 4691
    },
    {
      "epoch": 0.4211093161012386,
      "grad_norm": 15.20697021484375,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 4692
    },
    {
      "epoch": 0.4211990665948663,
      "grad_norm": 4.0113325119018555,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 4693
    },
    {
      "epoch": 0.42128881708849397,
      "grad_norm": 2.875786304473877,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 4694
    },
    {
      "epoch": 0.4213785675821217,
      "grad_norm": 3.124654769897461,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 4695
    },
    {
      "epoch": 0.4214683180757494,
      "grad_norm": 3.8283073902130127,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 4696
    },
    {
      "epoch": 0.4215580685693771,
      "grad_norm": 2.703998565673828,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 4697
    },
    {
      "epoch": 0.42164781906300486,
      "grad_norm": 4.575442790985107,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 4698
    },
    {
      "epoch": 0.42173756955663255,
      "grad_norm": 4.172669410705566,
      "learning_rate": 0.0,
      "loss": 0.8315,
      "step": 4699
    },
    {
      "epoch": 0.4218273200502603,
      "grad_norm": 3.305692434310913,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 4700
    },
    {
      "epoch": 0.421917070543888,
      "grad_norm": 2.4682092666625977,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 4701
    },
    {
      "epoch": 0.4220068210375157,
      "grad_norm": 9.845563888549805,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 4702
    },
    {
      "epoch": 0.42209657153114344,
      "grad_norm": 3.7486767768859863,
      "learning_rate": 0.0,
      "loss": 0.7319,
      "step": 4703
    },
    {
      "epoch": 0.42218632202477113,
      "grad_norm": 3.536635160446167,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 4704
    },
    {
      "epoch": 0.4222760725183988,
      "grad_norm": 4.075616836547852,
      "learning_rate": 0.0,
      "loss": 0.7006,
      "step": 4705
    },
    {
      "epoch": 0.4223658230120266,
      "grad_norm": 4.8801703453063965,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 4706
    },
    {
      "epoch": 0.42245557350565427,
      "grad_norm": 3.720611095428467,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 4707
    },
    {
      "epoch": 0.422545323999282,
      "grad_norm": 5.319940567016602,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 4708
    },
    {
      "epoch": 0.4226350744929097,
      "grad_norm": 2.77380633354187,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 4709
    },
    {
      "epoch": 0.4227248249865374,
      "grad_norm": 2.3984904289245605,
      "learning_rate": 0.0,
      "loss": 0.6432,
      "step": 4710
    },
    {
      "epoch": 0.42281457548016516,
      "grad_norm": 3.6718709468841553,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 4711
    },
    {
      "epoch": 0.42290432597379285,
      "grad_norm": 2.692570447921753,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 4712
    },
    {
      "epoch": 0.42299407646742054,
      "grad_norm": 2.2816288471221924,
      "learning_rate": 0.0,
      "loss": 0.8274,
      "step": 4713
    },
    {
      "epoch": 0.4230838269610483,
      "grad_norm": 4.1232428550720215,
      "learning_rate": 0.0,
      "loss": 0.7196,
      "step": 4714
    },
    {
      "epoch": 0.423173577454676,
      "grad_norm": 9.486380577087402,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 4715
    },
    {
      "epoch": 0.42326332794830374,
      "grad_norm": 4.9112162590026855,
      "learning_rate": 0.0,
      "loss": 0.8511,
      "step": 4716
    },
    {
      "epoch": 0.42335307844193143,
      "grad_norm": 3.5644097328186035,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 4717
    },
    {
      "epoch": 0.4234428289355591,
      "grad_norm": 2.6680872440338135,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 4718
    },
    {
      "epoch": 0.4235325794291869,
      "grad_norm": 3.5265138149261475,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 4719
    },
    {
      "epoch": 0.42362232992281457,
      "grad_norm": 2.9448235034942627,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 4720
    },
    {
      "epoch": 0.4237120804164423,
      "grad_norm": 3.857142686843872,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 4721
    },
    {
      "epoch": 0.42380183091007,
      "grad_norm": 13.213078498840332,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 4722
    },
    {
      "epoch": 0.4238915814036977,
      "grad_norm": 3.450326442718506,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 4723
    },
    {
      "epoch": 0.42398133189732545,
      "grad_norm": 3.8737902641296387,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 4724
    },
    {
      "epoch": 0.42407108239095315,
      "grad_norm": 2.8643813133239746,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 4725
    },
    {
      "epoch": 0.42416083288458084,
      "grad_norm": 2.8557193279266357,
      "learning_rate": 0.0,
      "loss": 0.7795,
      "step": 4726
    },
    {
      "epoch": 0.4242505833782086,
      "grad_norm": 2.4496984481811523,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 4727
    },
    {
      "epoch": 0.4243403338718363,
      "grad_norm": 3.8329427242279053,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 4728
    },
    {
      "epoch": 0.42443008436546403,
      "grad_norm": 29.402542114257812,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 4729
    },
    {
      "epoch": 0.4245198348590917,
      "grad_norm": 2.7326126098632812,
      "learning_rate": 0.0,
      "loss": 0.7857,
      "step": 4730
    },
    {
      "epoch": 0.4246095853527194,
      "grad_norm": 3.0108325481414795,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 4731
    },
    {
      "epoch": 0.42469933584634717,
      "grad_norm": 3.737396240234375,
      "learning_rate": 0.0,
      "loss": 0.7741,
      "step": 4732
    },
    {
      "epoch": 0.42478908633997486,
      "grad_norm": 4.652256965637207,
      "learning_rate": 0.0,
      "loss": 0.7801,
      "step": 4733
    },
    {
      "epoch": 0.42487883683360256,
      "grad_norm": 2.95754337310791,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 4734
    },
    {
      "epoch": 0.4249685873272303,
      "grad_norm": 3.0618069171905518,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 4735
    },
    {
      "epoch": 0.425058337820858,
      "grad_norm": 4.141149997711182,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 4736
    },
    {
      "epoch": 0.42514808831448575,
      "grad_norm": 3.3577871322631836,
      "learning_rate": 0.0,
      "loss": 0.7446,
      "step": 4737
    },
    {
      "epoch": 0.42523783880811344,
      "grad_norm": 3.170358896255493,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 4738
    },
    {
      "epoch": 0.42532758930174114,
      "grad_norm": 2.771730661392212,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 4739
    },
    {
      "epoch": 0.4254173397953689,
      "grad_norm": 3.3659706115722656,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 4740
    },
    {
      "epoch": 0.4255070902889966,
      "grad_norm": 2.8652729988098145,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 4741
    },
    {
      "epoch": 0.42559684078262433,
      "grad_norm": 2.770185947418213,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 4742
    },
    {
      "epoch": 0.425686591276252,
      "grad_norm": 3.2766919136047363,
      "learning_rate": 0.0,
      "loss": 0.782,
      "step": 4743
    },
    {
      "epoch": 0.4257763417698797,
      "grad_norm": 3.2721457481384277,
      "learning_rate": 0.0,
      "loss": 0.7839,
      "step": 4744
    },
    {
      "epoch": 0.42586609226350747,
      "grad_norm": 3.5115201473236084,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 4745
    },
    {
      "epoch": 0.42595584275713516,
      "grad_norm": 3.841991424560547,
      "learning_rate": 0.0,
      "loss": 0.6703,
      "step": 4746
    },
    {
      "epoch": 0.42604559325076286,
      "grad_norm": 3.587327480316162,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 4747
    },
    {
      "epoch": 0.4261353437443906,
      "grad_norm": 2.607776641845703,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 4748
    },
    {
      "epoch": 0.4262250942380183,
      "grad_norm": 2.2138407230377197,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 4749
    },
    {
      "epoch": 0.42631484473164605,
      "grad_norm": 2.5189454555511475,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 4750
    },
    {
      "epoch": 0.42640459522527374,
      "grad_norm": 3.5206458568573,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 4751
    },
    {
      "epoch": 0.42649434571890144,
      "grad_norm": 2.680375814437866,
      "learning_rate": 0.0,
      "loss": 0.7747,
      "step": 4752
    },
    {
      "epoch": 0.4265840962125292,
      "grad_norm": 2.802309274673462,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 4753
    },
    {
      "epoch": 0.4266738467061569,
      "grad_norm": 2.4506239891052246,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 4754
    },
    {
      "epoch": 0.4267635971997846,
      "grad_norm": 2.7510855197906494,
      "learning_rate": 0.0,
      "loss": 0.7418,
      "step": 4755
    },
    {
      "epoch": 0.4268533476934123,
      "grad_norm": 3.765601873397827,
      "learning_rate": 0.0,
      "loss": 0.7989,
      "step": 4756
    },
    {
      "epoch": 0.42694309818704,
      "grad_norm": 3.5723516941070557,
      "learning_rate": 0.0,
      "loss": 0.7833,
      "step": 4757
    },
    {
      "epoch": 0.42703284868066776,
      "grad_norm": 12.764555931091309,
      "learning_rate": 0.0,
      "loss": 0.7662,
      "step": 4758
    },
    {
      "epoch": 0.42712259917429546,
      "grad_norm": 4.23858118057251,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 4759
    },
    {
      "epoch": 0.42721234966792315,
      "grad_norm": 2.791533946990967,
      "learning_rate": 0.0,
      "loss": 0.697,
      "step": 4760
    },
    {
      "epoch": 0.4273021001615509,
      "grad_norm": 8.12319278717041,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 4761
    },
    {
      "epoch": 0.4273918506551786,
      "grad_norm": 3.9399192333221436,
      "learning_rate": 0.0,
      "loss": 0.8421,
      "step": 4762
    },
    {
      "epoch": 0.42748160114880634,
      "grad_norm": 3.369940996170044,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 4763
    },
    {
      "epoch": 0.42757135164243404,
      "grad_norm": 2.4668662548065186,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 4764
    },
    {
      "epoch": 0.42766110213606173,
      "grad_norm": 2.8369739055633545,
      "learning_rate": 0.0,
      "loss": 0.7432,
      "step": 4765
    },
    {
      "epoch": 0.4277508526296895,
      "grad_norm": 3.2943832874298096,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 4766
    },
    {
      "epoch": 0.4278406031233172,
      "grad_norm": 3.438533306121826,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 4767
    },
    {
      "epoch": 0.42793035361694487,
      "grad_norm": 2.705146551132202,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 4768
    },
    {
      "epoch": 0.4280201041105726,
      "grad_norm": 2.704803466796875,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 4769
    },
    {
      "epoch": 0.4281098546042003,
      "grad_norm": 2.603569269180298,
      "learning_rate": 0.0,
      "loss": 0.7306,
      "step": 4770
    },
    {
      "epoch": 0.42819960509782806,
      "grad_norm": 3.9186222553253174,
      "learning_rate": 0.0,
      "loss": 0.7381,
      "step": 4771
    },
    {
      "epoch": 0.42828935559145576,
      "grad_norm": 2.9373679161071777,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 4772
    },
    {
      "epoch": 0.42837910608508345,
      "grad_norm": 2.8916566371917725,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 4773
    },
    {
      "epoch": 0.4284688565787112,
      "grad_norm": 5.351807594299316,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 4774
    },
    {
      "epoch": 0.4285586070723389,
      "grad_norm": 6.057129383087158,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 4775
    },
    {
      "epoch": 0.4286483575659666,
      "grad_norm": 3.6040070056915283,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 4776
    },
    {
      "epoch": 0.42873810805959434,
      "grad_norm": 3.322500467300415,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 4777
    },
    {
      "epoch": 0.42882785855322203,
      "grad_norm": 2.804110288619995,
      "learning_rate": 0.0,
      "loss": 0.7788,
      "step": 4778
    },
    {
      "epoch": 0.4289176090468498,
      "grad_norm": 2.749316930770874,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 4779
    },
    {
      "epoch": 0.4290073595404775,
      "grad_norm": 5.237757682800293,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 4780
    },
    {
      "epoch": 0.42909711003410517,
      "grad_norm": 2.370323896408081,
      "learning_rate": 0.0,
      "loss": 0.6992,
      "step": 4781
    },
    {
      "epoch": 0.4291868605277329,
      "grad_norm": 2.526334047317505,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 4782
    },
    {
      "epoch": 0.4292766110213606,
      "grad_norm": 2.999622344970703,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 4783
    },
    {
      "epoch": 0.42936636151498836,
      "grad_norm": 5.767947673797607,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 4784
    },
    {
      "epoch": 0.42945611200861605,
      "grad_norm": 4.006098747253418,
      "learning_rate": 0.0,
      "loss": 0.7714,
      "step": 4785
    },
    {
      "epoch": 0.42954586250224375,
      "grad_norm": 9.09935474395752,
      "learning_rate": 0.0,
      "loss": 0.768,
      "step": 4786
    },
    {
      "epoch": 0.4296356129958715,
      "grad_norm": 3.2352778911590576,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 4787
    },
    {
      "epoch": 0.4297253634894992,
      "grad_norm": 2.986809730529785,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 4788
    },
    {
      "epoch": 0.4298151139831269,
      "grad_norm": 3.335681438446045,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 4789
    },
    {
      "epoch": 0.42990486447675463,
      "grad_norm": 2.4888625144958496,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 4790
    },
    {
      "epoch": 0.4299946149703823,
      "grad_norm": 2.5808990001678467,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 4791
    },
    {
      "epoch": 0.4300843654640101,
      "grad_norm": 3.491469621658325,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 4792
    },
    {
      "epoch": 0.43017411595763777,
      "grad_norm": 2.9121174812316895,
      "learning_rate": 0.0,
      "loss": 0.7885,
      "step": 4793
    },
    {
      "epoch": 0.43026386645126546,
      "grad_norm": 2.5912528038024902,
      "learning_rate": 0.0,
      "loss": 0.743,
      "step": 4794
    },
    {
      "epoch": 0.4303536169448932,
      "grad_norm": 3.0640392303466797,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 4795
    },
    {
      "epoch": 0.4304433674385209,
      "grad_norm": 2.798907518386841,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 4796
    },
    {
      "epoch": 0.4305331179321486,
      "grad_norm": 2.879103422164917,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 4797
    },
    {
      "epoch": 0.43062286842577635,
      "grad_norm": 5.893740653991699,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 4798
    },
    {
      "epoch": 0.43071261891940404,
      "grad_norm": 3.2195756435394287,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 4799
    },
    {
      "epoch": 0.4308023694130318,
      "grad_norm": 2.953819751739502,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 4800
    },
    {
      "epoch": 0.4308921199066595,
      "grad_norm": 2.9046552181243896,
      "learning_rate": 0.0,
      "loss": 0.7924,
      "step": 4801
    },
    {
      "epoch": 0.4309818704002872,
      "grad_norm": 2.8710057735443115,
      "learning_rate": 0.0,
      "loss": 0.8373,
      "step": 4802
    },
    {
      "epoch": 0.43107162089391493,
      "grad_norm": 2.4950637817382812,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 4803
    },
    {
      "epoch": 0.4311613713875426,
      "grad_norm": 3.734285593032837,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 4804
    },
    {
      "epoch": 0.4312511218811704,
      "grad_norm": 5.235177993774414,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 4805
    },
    {
      "epoch": 0.43134087237479807,
      "grad_norm": 3.1455488204956055,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 4806
    },
    {
      "epoch": 0.43143062286842576,
      "grad_norm": 4.35747766494751,
      "learning_rate": 0.0,
      "loss": 0.6755,
      "step": 4807
    },
    {
      "epoch": 0.4315203733620535,
      "grad_norm": 5.708597183227539,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 4808
    },
    {
      "epoch": 0.4316101238556812,
      "grad_norm": 4.534907817840576,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 4809
    },
    {
      "epoch": 0.4316998743493089,
      "grad_norm": 2.8288142681121826,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 4810
    },
    {
      "epoch": 0.43178962484293665,
      "grad_norm": 21.752426147460938,
      "learning_rate": 0.0,
      "loss": 0.6854,
      "step": 4811
    },
    {
      "epoch": 0.43187937533656434,
      "grad_norm": 4.220375061035156,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 4812
    },
    {
      "epoch": 0.4319691258301921,
      "grad_norm": 2.5227251052856445,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 4813
    },
    {
      "epoch": 0.4320588763238198,
      "grad_norm": 2.6455130577087402,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 4814
    },
    {
      "epoch": 0.4321486268174475,
      "grad_norm": 3.2034950256347656,
      "learning_rate": 0.0,
      "loss": 0.7204,
      "step": 4815
    },
    {
      "epoch": 0.43223837731107523,
      "grad_norm": 2.3445067405700684,
      "learning_rate": 0.0,
      "loss": 0.7001,
      "step": 4816
    },
    {
      "epoch": 0.4323281278047029,
      "grad_norm": 2.6470041275024414,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 4817
    },
    {
      "epoch": 0.4324178782983306,
      "grad_norm": 3.226376533508301,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 4818
    },
    {
      "epoch": 0.43250762879195837,
      "grad_norm": 3.742851495742798,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 4819
    },
    {
      "epoch": 0.43259737928558606,
      "grad_norm": 3.5871689319610596,
      "learning_rate": 0.0,
      "loss": 0.6849,
      "step": 4820
    },
    {
      "epoch": 0.4326871297792138,
      "grad_norm": 3.1482062339782715,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 4821
    },
    {
      "epoch": 0.4327768802728415,
      "grad_norm": 2.9344005584716797,
      "learning_rate": 0.0,
      "loss": 0.7458,
      "step": 4822
    },
    {
      "epoch": 0.4328666307664692,
      "grad_norm": 3.0645499229431152,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 4823
    },
    {
      "epoch": 0.43295638126009695,
      "grad_norm": 3.6520535945892334,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 4824
    },
    {
      "epoch": 0.43304613175372464,
      "grad_norm": 10.522553443908691,
      "learning_rate": 0.0,
      "loss": 0.794,
      "step": 4825
    },
    {
      "epoch": 0.43313588224735233,
      "grad_norm": 5.812234401702881,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 4826
    },
    {
      "epoch": 0.4332256327409801,
      "grad_norm": 6.237459182739258,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 4827
    },
    {
      "epoch": 0.4333153832346078,
      "grad_norm": 3.971855878829956,
      "learning_rate": 0.0,
      "loss": 0.6686,
      "step": 4828
    },
    {
      "epoch": 0.4334051337282355,
      "grad_norm": 7.182819366455078,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 4829
    },
    {
      "epoch": 0.4334948842218632,
      "grad_norm": 2.3101375102996826,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 4830
    },
    {
      "epoch": 0.4335846347154909,
      "grad_norm": 2.7746176719665527,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 4831
    },
    {
      "epoch": 0.43367438520911866,
      "grad_norm": 65.69673156738281,
      "learning_rate": 0.0,
      "loss": 0.7794,
      "step": 4832
    },
    {
      "epoch": 0.43376413570274636,
      "grad_norm": 6.100004196166992,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 4833
    },
    {
      "epoch": 0.4338538861963741,
      "grad_norm": 4.165258407592773,
      "learning_rate": 0.0,
      "loss": 0.6785,
      "step": 4834
    },
    {
      "epoch": 0.4339436366900018,
      "grad_norm": 3.4696872234344482,
      "learning_rate": 0.0,
      "loss": 0.6925,
      "step": 4835
    },
    {
      "epoch": 0.4340333871836295,
      "grad_norm": 2.44360613822937,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 4836
    },
    {
      "epoch": 0.43412313767725724,
      "grad_norm": 4.0581440925598145,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 4837
    },
    {
      "epoch": 0.43421288817088494,
      "grad_norm": 2.748633623123169,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 4838
    },
    {
      "epoch": 0.43430263866451263,
      "grad_norm": 2.9905288219451904,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 4839
    },
    {
      "epoch": 0.4343923891581404,
      "grad_norm": 3.26027512550354,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 4840
    },
    {
      "epoch": 0.4344821396517681,
      "grad_norm": 3.3889904022216797,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 4841
    },
    {
      "epoch": 0.4345718901453958,
      "grad_norm": 2.8393895626068115,
      "learning_rate": 0.0,
      "loss": 0.7203,
      "step": 4842
    },
    {
      "epoch": 0.4346616406390235,
      "grad_norm": 2.665336847305298,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 4843
    },
    {
      "epoch": 0.4347513911326512,
      "grad_norm": 5.733819007873535,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 4844
    },
    {
      "epoch": 0.43484114162627896,
      "grad_norm": 3.290189027786255,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 4845
    },
    {
      "epoch": 0.43493089211990665,
      "grad_norm": 2.860020399093628,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 4846
    },
    {
      "epoch": 0.43502064261353435,
      "grad_norm": 5.235310077667236,
      "learning_rate": 0.0,
      "loss": 0.7872,
      "step": 4847
    },
    {
      "epoch": 0.4351103931071621,
      "grad_norm": 3.4746713638305664,
      "learning_rate": 0.0,
      "loss": 0.7635,
      "step": 4848
    },
    {
      "epoch": 0.4352001436007898,
      "grad_norm": 4.9758219718933105,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 4849
    },
    {
      "epoch": 0.43528989409441754,
      "grad_norm": 3.365600109100342,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 4850
    },
    {
      "epoch": 0.43537964458804523,
      "grad_norm": 2.404998540878296,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 4851
    },
    {
      "epoch": 0.4354693950816729,
      "grad_norm": 3.131873607635498,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 4852
    },
    {
      "epoch": 0.4355591455753007,
      "grad_norm": 3.48376727104187,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 4853
    },
    {
      "epoch": 0.43564889606892837,
      "grad_norm": 4.257928371429443,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 4854
    },
    {
      "epoch": 0.4357386465625561,
      "grad_norm": 3.9874632358551025,
      "learning_rate": 0.0,
      "loss": 0.6741,
      "step": 4855
    },
    {
      "epoch": 0.4358283970561838,
      "grad_norm": 4.4688615798950195,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 4856
    },
    {
      "epoch": 0.4359181475498115,
      "grad_norm": 2.6603329181671143,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 4857
    },
    {
      "epoch": 0.43600789804343926,
      "grad_norm": 3.119901180267334,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 4858
    },
    {
      "epoch": 0.43609764853706695,
      "grad_norm": 8.738306045532227,
      "learning_rate": 0.0,
      "loss": 0.7527,
      "step": 4859
    },
    {
      "epoch": 0.43618739903069464,
      "grad_norm": 3.3960750102996826,
      "learning_rate": 0.0,
      "loss": 0.6925,
      "step": 4860
    },
    {
      "epoch": 0.4362771495243224,
      "grad_norm": 2.612494945526123,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 4861
    },
    {
      "epoch": 0.4363669000179501,
      "grad_norm": 2.4404799938201904,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 4862
    },
    {
      "epoch": 0.43645665051157784,
      "grad_norm": 10.14219856262207,
      "learning_rate": 0.0,
      "loss": 0.8037,
      "step": 4863
    },
    {
      "epoch": 0.43654640100520553,
      "grad_norm": 2.802030324935913,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 4864
    },
    {
      "epoch": 0.4366361514988332,
      "grad_norm": 3.3231234550476074,
      "learning_rate": 0.0,
      "loss": 0.6947,
      "step": 4865
    },
    {
      "epoch": 0.436725901992461,
      "grad_norm": 2.3027803897857666,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 4866
    },
    {
      "epoch": 0.43681565248608867,
      "grad_norm": 3.0976765155792236,
      "learning_rate": 0.0,
      "loss": 0.7057,
      "step": 4867
    },
    {
      "epoch": 0.43690540297971636,
      "grad_norm": 2.6609065532684326,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 4868
    },
    {
      "epoch": 0.4369951534733441,
      "grad_norm": 2.4403433799743652,
      "learning_rate": 0.0,
      "loss": 0.7596,
      "step": 4869
    },
    {
      "epoch": 0.4370849039669718,
      "grad_norm": 2.5947864055633545,
      "learning_rate": 0.0,
      "loss": 0.774,
      "step": 4870
    },
    {
      "epoch": 0.43717465446059955,
      "grad_norm": 5.307293891906738,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 4871
    },
    {
      "epoch": 0.43726440495422725,
      "grad_norm": 3.801018238067627,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 4872
    },
    {
      "epoch": 0.43735415544785494,
      "grad_norm": 3.0706984996795654,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 4873
    },
    {
      "epoch": 0.4374439059414827,
      "grad_norm": 3.5638351440429688,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 4874
    },
    {
      "epoch": 0.4375336564351104,
      "grad_norm": 3.84413743019104,
      "learning_rate": 0.0,
      "loss": 0.7884,
      "step": 4875
    },
    {
      "epoch": 0.43762340692873813,
      "grad_norm": 4.257320880889893,
      "learning_rate": 0.0,
      "loss": 0.76,
      "step": 4876
    },
    {
      "epoch": 0.43771315742236583,
      "grad_norm": 3.717466354370117,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 4877
    },
    {
      "epoch": 0.4378029079159935,
      "grad_norm": 3.714552640914917,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 4878
    },
    {
      "epoch": 0.43789265840962127,
      "grad_norm": 5.799731731414795,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 4879
    },
    {
      "epoch": 0.43798240890324897,
      "grad_norm": 5.529098033905029,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 4880
    },
    {
      "epoch": 0.43807215939687666,
      "grad_norm": 3.0452964305877686,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 4881
    },
    {
      "epoch": 0.4381619098905044,
      "grad_norm": 3.8129312992095947,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 4882
    },
    {
      "epoch": 0.4382516603841321,
      "grad_norm": 5.682473659515381,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 4883
    },
    {
      "epoch": 0.43834141087775985,
      "grad_norm": 3.9464144706726074,
      "learning_rate": 0.0,
      "loss": 0.7632,
      "step": 4884
    },
    {
      "epoch": 0.43843116137138755,
      "grad_norm": 2.6495862007141113,
      "learning_rate": 0.0,
      "loss": 0.7648,
      "step": 4885
    },
    {
      "epoch": 0.43852091186501524,
      "grad_norm": 2.9125843048095703,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 4886
    },
    {
      "epoch": 0.438610662358643,
      "grad_norm": 3.208237409591675,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 4887
    },
    {
      "epoch": 0.4387004128522707,
      "grad_norm": 2.8052127361297607,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 4888
    },
    {
      "epoch": 0.4387901633458984,
      "grad_norm": 9.115546226501465,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 4889
    },
    {
      "epoch": 0.4388799138395261,
      "grad_norm": 4.958300590515137,
      "learning_rate": 0.0,
      "loss": 0.663,
      "step": 4890
    },
    {
      "epoch": 0.4389696643331538,
      "grad_norm": 4.753262042999268,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 4891
    },
    {
      "epoch": 0.43905941482678157,
      "grad_norm": 3.1042346954345703,
      "learning_rate": 0.0,
      "loss": 0.8182,
      "step": 4892
    },
    {
      "epoch": 0.43914916532040926,
      "grad_norm": 2.433298110961914,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 4893
    },
    {
      "epoch": 0.43923891581403696,
      "grad_norm": 3.503404378890991,
      "learning_rate": 0.0,
      "loss": 0.6692,
      "step": 4894
    },
    {
      "epoch": 0.4393286663076647,
      "grad_norm": 10.322394371032715,
      "learning_rate": 0.0,
      "loss": 0.7488,
      "step": 4895
    },
    {
      "epoch": 0.4394184168012924,
      "grad_norm": 3.925555467605591,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 4896
    },
    {
      "epoch": 0.43950816729492015,
      "grad_norm": 2.967653512954712,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 4897
    },
    {
      "epoch": 0.43959791778854784,
      "grad_norm": 3.2255911827087402,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 4898
    },
    {
      "epoch": 0.43968766828217554,
      "grad_norm": 2.7814362049102783,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 4899
    },
    {
      "epoch": 0.4397774187758033,
      "grad_norm": 3.2126190662384033,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 4900
    },
    {
      "epoch": 0.439867169269431,
      "grad_norm": 3.315443277359009,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 4901
    },
    {
      "epoch": 0.4399569197630587,
      "grad_norm": 2.4869251251220703,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 4902
    },
    {
      "epoch": 0.4400466702566864,
      "grad_norm": 11.627095222473145,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 4903
    },
    {
      "epoch": 0.4401364207503141,
      "grad_norm": 4.134926795959473,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 4904
    },
    {
      "epoch": 0.44022617124394187,
      "grad_norm": 2.948913812637329,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 4905
    },
    {
      "epoch": 0.44031592173756956,
      "grad_norm": 2.326563596725464,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 4906
    },
    {
      "epoch": 0.44040567223119725,
      "grad_norm": 6.56026029586792,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 4907
    },
    {
      "epoch": 0.440495422724825,
      "grad_norm": 3.33663272857666,
      "learning_rate": 0.0,
      "loss": 0.7863,
      "step": 4908
    },
    {
      "epoch": 0.4405851732184527,
      "grad_norm": 2.86498761177063,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 4909
    },
    {
      "epoch": 0.4406749237120804,
      "grad_norm": 2.702416181564331,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 4910
    },
    {
      "epoch": 0.44076467420570814,
      "grad_norm": 2.6223630905151367,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 4911
    },
    {
      "epoch": 0.44085442469933583,
      "grad_norm": 3.419802665710449,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 4912
    },
    {
      "epoch": 0.4409441751929636,
      "grad_norm": 3.4016900062561035,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 4913
    },
    {
      "epoch": 0.4410339256865913,
      "grad_norm": 2.5002524852752686,
      "learning_rate": 0.0,
      "loss": 0.7602,
      "step": 4914
    },
    {
      "epoch": 0.44112367618021897,
      "grad_norm": 4.135557651519775,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 4915
    },
    {
      "epoch": 0.4412134266738467,
      "grad_norm": 5.0262651443481445,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 4916
    },
    {
      "epoch": 0.4413031771674744,
      "grad_norm": 4.149956226348877,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 4917
    },
    {
      "epoch": 0.44139292766110216,
      "grad_norm": 3.1758267879486084,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 4918
    },
    {
      "epoch": 0.44148267815472986,
      "grad_norm": 3.324871778488159,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 4919
    },
    {
      "epoch": 0.44157242864835755,
      "grad_norm": 2.797675371170044,
      "learning_rate": 0.0,
      "loss": 0.6977,
      "step": 4920
    },
    {
      "epoch": 0.4416621791419853,
      "grad_norm": 2.7934601306915283,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 4921
    },
    {
      "epoch": 0.441751929635613,
      "grad_norm": 2.524312973022461,
      "learning_rate": 0.0,
      "loss": 0.8031,
      "step": 4922
    },
    {
      "epoch": 0.4418416801292407,
      "grad_norm": 4.716702938079834,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 4923
    },
    {
      "epoch": 0.44193143062286844,
      "grad_norm": 2.5589823722839355,
      "learning_rate": 0.0,
      "loss": 0.7691,
      "step": 4924
    },
    {
      "epoch": 0.44202118111649613,
      "grad_norm": 4.562977313995361,
      "learning_rate": 0.0,
      "loss": 0.7937,
      "step": 4925
    },
    {
      "epoch": 0.4421109316101239,
      "grad_norm": 8.47123908996582,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 4926
    },
    {
      "epoch": 0.4422006821037516,
      "grad_norm": 3.4237539768218994,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 4927
    },
    {
      "epoch": 0.44229043259737927,
      "grad_norm": 5.526391506195068,
      "learning_rate": 0.0,
      "loss": 0.7587,
      "step": 4928
    },
    {
      "epoch": 0.442380183091007,
      "grad_norm": 3.0177528858184814,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 4929
    },
    {
      "epoch": 0.4424699335846347,
      "grad_norm": 2.7152464389801025,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 4930
    },
    {
      "epoch": 0.4425596840782624,
      "grad_norm": 2.661816358566284,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 4931
    },
    {
      "epoch": 0.44264943457189015,
      "grad_norm": 2.7051048278808594,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 4932
    },
    {
      "epoch": 0.44273918506551785,
      "grad_norm": 2.7761142253875732,
      "learning_rate": 0.0,
      "loss": 0.6949,
      "step": 4933
    },
    {
      "epoch": 0.4428289355591456,
      "grad_norm": 3.364607810974121,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 4934
    },
    {
      "epoch": 0.4429186860527733,
      "grad_norm": 3.743010997772217,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 4935
    },
    {
      "epoch": 0.443008436546401,
      "grad_norm": 3.2606301307678223,
      "learning_rate": 0.0,
      "loss": 0.7527,
      "step": 4936
    },
    {
      "epoch": 0.44309818704002873,
      "grad_norm": 2.6966793537139893,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 4937
    },
    {
      "epoch": 0.44318793753365643,
      "grad_norm": 6.22252893447876,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 4938
    },
    {
      "epoch": 0.4432776880272841,
      "grad_norm": 6.304412364959717,
      "learning_rate": 0.0,
      "loss": 0.8041,
      "step": 4939
    },
    {
      "epoch": 0.44336743852091187,
      "grad_norm": 3.9474165439605713,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 4940
    },
    {
      "epoch": 0.44345718901453957,
      "grad_norm": 2.0605509281158447,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 4941
    },
    {
      "epoch": 0.4435469395081673,
      "grad_norm": 3.363510847091675,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 4942
    },
    {
      "epoch": 0.443636690001795,
      "grad_norm": 4.360350131988525,
      "learning_rate": 0.0,
      "loss": 0.6561,
      "step": 4943
    },
    {
      "epoch": 0.4437264404954227,
      "grad_norm": 4.94108772277832,
      "learning_rate": 0.0,
      "loss": 0.8147,
      "step": 4944
    },
    {
      "epoch": 0.44381619098905045,
      "grad_norm": 2.407731533050537,
      "learning_rate": 0.0,
      "loss": 0.746,
      "step": 4945
    },
    {
      "epoch": 0.44390594148267815,
      "grad_norm": 3.718047618865967,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 4946
    },
    {
      "epoch": 0.4439956919763059,
      "grad_norm": 6.990278244018555,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 4947
    },
    {
      "epoch": 0.4440854424699336,
      "grad_norm": 4.423760414123535,
      "learning_rate": 0.0,
      "loss": 0.6881,
      "step": 4948
    },
    {
      "epoch": 0.4441751929635613,
      "grad_norm": 2.9290525913238525,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 4949
    },
    {
      "epoch": 0.44426494345718903,
      "grad_norm": 5.265776634216309,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 4950
    },
    {
      "epoch": 0.4443546939508167,
      "grad_norm": 2.324584484100342,
      "learning_rate": 0.0,
      "loss": 0.7688,
      "step": 4951
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 3.0445151329040527,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 4952
    },
    {
      "epoch": 0.44453419493807217,
      "grad_norm": 2.8998794555664062,
      "learning_rate": 0.0,
      "loss": 0.7707,
      "step": 4953
    },
    {
      "epoch": 0.44462394543169986,
      "grad_norm": 2.2287888526916504,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 4954
    },
    {
      "epoch": 0.4447136959253276,
      "grad_norm": 2.6966664791107178,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 4955
    },
    {
      "epoch": 0.4448034464189553,
      "grad_norm": 7.789462089538574,
      "learning_rate": 0.0,
      "loss": 0.7829,
      "step": 4956
    },
    {
      "epoch": 0.444893196912583,
      "grad_norm": 4.229278087615967,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 4957
    },
    {
      "epoch": 0.44498294740621075,
      "grad_norm": 3.2974305152893066,
      "learning_rate": 0.0,
      "loss": 0.8029,
      "step": 4958
    },
    {
      "epoch": 0.44507269789983844,
      "grad_norm": 2.8243961334228516,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 4959
    },
    {
      "epoch": 0.44516244839346614,
      "grad_norm": 3.1675028800964355,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 4960
    },
    {
      "epoch": 0.4452521988870939,
      "grad_norm": 3.199986457824707,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 4961
    },
    {
      "epoch": 0.4453419493807216,
      "grad_norm": 2.978828191757202,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 4962
    },
    {
      "epoch": 0.44543169987434933,
      "grad_norm": 2.296938180923462,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 4963
    },
    {
      "epoch": 0.445521450367977,
      "grad_norm": 2.679232358932495,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 4964
    },
    {
      "epoch": 0.4456112008616047,
      "grad_norm": 4.402146816253662,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 4965
    },
    {
      "epoch": 0.44570095135523247,
      "grad_norm": 5.5495476722717285,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 4966
    },
    {
      "epoch": 0.44579070184886016,
      "grad_norm": 5.108234405517578,
      "learning_rate": 0.0,
      "loss": 0.7753,
      "step": 4967
    },
    {
      "epoch": 0.4458804523424879,
      "grad_norm": 2.605208396911621,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 4968
    },
    {
      "epoch": 0.4459702028361156,
      "grad_norm": 2.6673829555511475,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 4969
    },
    {
      "epoch": 0.4460599533297433,
      "grad_norm": 3.3521478176116943,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 4970
    },
    {
      "epoch": 0.44614970382337105,
      "grad_norm": 4.907279968261719,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 4971
    },
    {
      "epoch": 0.44623945431699874,
      "grad_norm": 3.2963287830352783,
      "learning_rate": 0.0,
      "loss": 0.827,
      "step": 4972
    },
    {
      "epoch": 0.44632920481062643,
      "grad_norm": 2.4787652492523193,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 4973
    },
    {
      "epoch": 0.4464189553042542,
      "grad_norm": 3.129103183746338,
      "learning_rate": 0.0,
      "loss": 0.7992,
      "step": 4974
    },
    {
      "epoch": 0.4465087057978819,
      "grad_norm": 5.283010005950928,
      "learning_rate": 0.0,
      "loss": 0.7657,
      "step": 4975
    },
    {
      "epoch": 0.4465984562915096,
      "grad_norm": 2.5952954292297363,
      "learning_rate": 0.0,
      "loss": 0.7506,
      "step": 4976
    },
    {
      "epoch": 0.4466882067851373,
      "grad_norm": 3.544804096221924,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 4977
    },
    {
      "epoch": 0.446777957278765,
      "grad_norm": 3.782029867172241,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 4978
    },
    {
      "epoch": 0.44686770777239276,
      "grad_norm": 2.707247734069824,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 4979
    },
    {
      "epoch": 0.44695745826602046,
      "grad_norm": 3.157101631164551,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 4980
    },
    {
      "epoch": 0.44704720875964815,
      "grad_norm": 3.934380054473877,
      "learning_rate": 0.0,
      "loss": 0.7249,
      "step": 4981
    },
    {
      "epoch": 0.4471369592532759,
      "grad_norm": 3.6321616172790527,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 4982
    },
    {
      "epoch": 0.4472267097469036,
      "grad_norm": 2.595299005508423,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 4983
    },
    {
      "epoch": 0.44731646024053134,
      "grad_norm": 3.4195146560668945,
      "learning_rate": 0.0,
      "loss": 0.7854,
      "step": 4984
    },
    {
      "epoch": 0.44740621073415904,
      "grad_norm": 3.0853517055511475,
      "learning_rate": 0.0,
      "loss": 0.774,
      "step": 4985
    },
    {
      "epoch": 0.44749596122778673,
      "grad_norm": 2.759904623031616,
      "learning_rate": 0.0,
      "loss": 0.763,
      "step": 4986
    },
    {
      "epoch": 0.4475857117214145,
      "grad_norm": 2.9196107387542725,
      "learning_rate": 0.0,
      "loss": 0.7641,
      "step": 4987
    },
    {
      "epoch": 0.4476754622150422,
      "grad_norm": 6.295373439788818,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 4988
    },
    {
      "epoch": 0.4477652127086699,
      "grad_norm": 2.881964683532715,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 4989
    },
    {
      "epoch": 0.4478549632022976,
      "grad_norm": 5.246219158172607,
      "learning_rate": 0.0,
      "loss": 0.6763,
      "step": 4990
    },
    {
      "epoch": 0.4479447136959253,
      "grad_norm": 4.552048683166504,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 4991
    },
    {
      "epoch": 0.44803446418955306,
      "grad_norm": 7.627259731292725,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 4992
    },
    {
      "epoch": 0.44812421468318075,
      "grad_norm": 8.572175979614258,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 4993
    },
    {
      "epoch": 0.44821396517680845,
      "grad_norm": 3.410062789916992,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 4994
    },
    {
      "epoch": 0.4483037156704362,
      "grad_norm": 3.480731725692749,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 4995
    },
    {
      "epoch": 0.4483934661640639,
      "grad_norm": 2.33640193939209,
      "learning_rate": 0.0,
      "loss": 0.8093,
      "step": 4996
    },
    {
      "epoch": 0.44848321665769164,
      "grad_norm": 2.9196908473968506,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 4997
    },
    {
      "epoch": 0.44857296715131934,
      "grad_norm": 2.8089890480041504,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 4998
    },
    {
      "epoch": 0.44866271764494703,
      "grad_norm": 2.624238967895508,
      "learning_rate": 0.0,
      "loss": 0.7902,
      "step": 4999
    },
    {
      "epoch": 0.4487524681385748,
      "grad_norm": 3.086230516433716,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 5000
    },
    {
      "epoch": 0.44884221863220247,
      "grad_norm": 4.474336624145508,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 5001
    },
    {
      "epoch": 0.44893196912583017,
      "grad_norm": 4.945656776428223,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 5002
    },
    {
      "epoch": 0.4490217196194579,
      "grad_norm": 6.285930156707764,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 5003
    },
    {
      "epoch": 0.4491114701130856,
      "grad_norm": 4.246393203735352,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 5004
    },
    {
      "epoch": 0.44920122060671336,
      "grad_norm": 3.5259857177734375,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 5005
    },
    {
      "epoch": 0.44929097110034105,
      "grad_norm": 3.08233904838562,
      "learning_rate": 0.0,
      "loss": 0.7468,
      "step": 5006
    },
    {
      "epoch": 0.44938072159396875,
      "grad_norm": 2.659432888031006,
      "learning_rate": 0.0,
      "loss": 0.7853,
      "step": 5007
    },
    {
      "epoch": 0.4494704720875965,
      "grad_norm": 2.8040177822113037,
      "learning_rate": 0.0,
      "loss": 0.6866,
      "step": 5008
    },
    {
      "epoch": 0.4495602225812242,
      "grad_norm": 2.496065378189087,
      "learning_rate": 0.0,
      "loss": 0.8179,
      "step": 5009
    },
    {
      "epoch": 0.44964997307485194,
      "grad_norm": 2.7276437282562256,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 5010
    },
    {
      "epoch": 0.44973972356847963,
      "grad_norm": 2.9758598804473877,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 5011
    },
    {
      "epoch": 0.4498294740621073,
      "grad_norm": 3.0431325435638428,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 5012
    },
    {
      "epoch": 0.4499192245557351,
      "grad_norm": 3.03598952293396,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 5013
    },
    {
      "epoch": 0.45000897504936277,
      "grad_norm": 3.0235846042633057,
      "learning_rate": 0.0,
      "loss": 0.7446,
      "step": 5014
    },
    {
      "epoch": 0.45009872554299046,
      "grad_norm": 2.639235496520996,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 5015
    },
    {
      "epoch": 0.4501884760366182,
      "grad_norm": 2.208017110824585,
      "learning_rate": 0.0,
      "loss": 0.8218,
      "step": 5016
    },
    {
      "epoch": 0.4502782265302459,
      "grad_norm": 18.294246673583984,
      "learning_rate": 0.0,
      "loss": 0.7447,
      "step": 5017
    },
    {
      "epoch": 0.45036797702387366,
      "grad_norm": 8.751993179321289,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 5018
    },
    {
      "epoch": 0.45045772751750135,
      "grad_norm": 4.042441368103027,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 5019
    },
    {
      "epoch": 0.45054747801112904,
      "grad_norm": 3.1411588191986084,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 5020
    },
    {
      "epoch": 0.4506372285047568,
      "grad_norm": 8.912809371948242,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 5021
    },
    {
      "epoch": 0.4507269789983845,
      "grad_norm": 4.996775150299072,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 5022
    },
    {
      "epoch": 0.4508167294920122,
      "grad_norm": 6.412389278411865,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 5023
    },
    {
      "epoch": 0.45090647998563993,
      "grad_norm": 3.8120672702789307,
      "learning_rate": 0.0,
      "loss": 0.747,
      "step": 5024
    },
    {
      "epoch": 0.4509962304792676,
      "grad_norm": 2.859426259994507,
      "learning_rate": 0.0,
      "loss": 0.7667,
      "step": 5025
    },
    {
      "epoch": 0.4510859809728954,
      "grad_norm": 17.229001998901367,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 5026
    },
    {
      "epoch": 0.45117573146652307,
      "grad_norm": 4.095582008361816,
      "learning_rate": 0.0,
      "loss": 0.7802,
      "step": 5027
    },
    {
      "epoch": 0.45126548196015076,
      "grad_norm": 3.7324161529541016,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 5028
    },
    {
      "epoch": 0.4513552324537785,
      "grad_norm": 7.117912292480469,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 5029
    },
    {
      "epoch": 0.4514449829474062,
      "grad_norm": 3.2193500995635986,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 5030
    },
    {
      "epoch": 0.45153473344103395,
      "grad_norm": 3.638960838317871,
      "learning_rate": 0.0,
      "loss": 0.7369,
      "step": 5031
    },
    {
      "epoch": 0.45162448393466165,
      "grad_norm": 2.462293863296509,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 5032
    },
    {
      "epoch": 0.45171423442828934,
      "grad_norm": 3.3873302936553955,
      "learning_rate": 0.0,
      "loss": 0.6743,
      "step": 5033
    },
    {
      "epoch": 0.4518039849219171,
      "grad_norm": 3.226142644882202,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 5034
    },
    {
      "epoch": 0.4518937354155448,
      "grad_norm": 3.697866916656494,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 5035
    },
    {
      "epoch": 0.4519834859091725,
      "grad_norm": 2.252838134765625,
      "learning_rate": 0.0,
      "loss": 0.7344,
      "step": 5036
    },
    {
      "epoch": 0.4520732364028002,
      "grad_norm": 3.1603496074676514,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 5037
    },
    {
      "epoch": 0.4521629868964279,
      "grad_norm": 2.755946397781372,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 5038
    },
    {
      "epoch": 0.45225273739005567,
      "grad_norm": 2.9210007190704346,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 5039
    },
    {
      "epoch": 0.45234248788368336,
      "grad_norm": 3.1495602130889893,
      "learning_rate": 0.0,
      "loss": 0.6918,
      "step": 5040
    },
    {
      "epoch": 0.45243223837731106,
      "grad_norm": 2.681871175765991,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 5041
    },
    {
      "epoch": 0.4525219888709388,
      "grad_norm": 2.310364007949829,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 5042
    },
    {
      "epoch": 0.4526117393645665,
      "grad_norm": 2.8664824962615967,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 5043
    },
    {
      "epoch": 0.4527014898581942,
      "grad_norm": 3.9192395210266113,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 5044
    },
    {
      "epoch": 0.45279124035182194,
      "grad_norm": 2.7180025577545166,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 5045
    },
    {
      "epoch": 0.45288099084544964,
      "grad_norm": 2.7515511512756348,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 5046
    },
    {
      "epoch": 0.4529707413390774,
      "grad_norm": 3.206298589706421,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 5047
    },
    {
      "epoch": 0.4530604918327051,
      "grad_norm": 2.76956844329834,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 5048
    },
    {
      "epoch": 0.4531502423263328,
      "grad_norm": 4.273536205291748,
      "learning_rate": 0.0,
      "loss": 0.7491,
      "step": 5049
    },
    {
      "epoch": 0.4532399928199605,
      "grad_norm": 3.4994826316833496,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 5050
    },
    {
      "epoch": 0.4533297433135882,
      "grad_norm": 3.5473592281341553,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 5051
    },
    {
      "epoch": 0.4534194938072159,
      "grad_norm": 2.945935010910034,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 5052
    },
    {
      "epoch": 0.45350924430084366,
      "grad_norm": 4.095198631286621,
      "learning_rate": 0.0,
      "loss": 0.7976,
      "step": 5053
    },
    {
      "epoch": 0.45359899479447136,
      "grad_norm": 4.322175979614258,
      "learning_rate": 0.0,
      "loss": 0.793,
      "step": 5054
    },
    {
      "epoch": 0.4536887452880991,
      "grad_norm": 3.101440191268921,
      "learning_rate": 0.0,
      "loss": 0.724,
      "step": 5055
    },
    {
      "epoch": 0.4537784957817268,
      "grad_norm": 3.0348258018493652,
      "learning_rate": 0.0,
      "loss": 0.7196,
      "step": 5056
    },
    {
      "epoch": 0.4538682462753545,
      "grad_norm": 2.840466022491455,
      "learning_rate": 0.0,
      "loss": 0.6747,
      "step": 5057
    },
    {
      "epoch": 0.45395799676898224,
      "grad_norm": 4.3641581535339355,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 5058
    },
    {
      "epoch": 0.45404774726260994,
      "grad_norm": 3.210805654525757,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 5059
    },
    {
      "epoch": 0.4541374977562377,
      "grad_norm": 2.791942596435547,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 5060
    },
    {
      "epoch": 0.4542272482498654,
      "grad_norm": 2.397966146469116,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 5061
    },
    {
      "epoch": 0.4543169987434931,
      "grad_norm": 2.909604072570801,
      "learning_rate": 0.0,
      "loss": 0.6722,
      "step": 5062
    },
    {
      "epoch": 0.4544067492371208,
      "grad_norm": 3.2979023456573486,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 5063
    },
    {
      "epoch": 0.4544964997307485,
      "grad_norm": 3.61669659614563,
      "learning_rate": 0.0,
      "loss": 0.6841,
      "step": 5064
    },
    {
      "epoch": 0.4545862502243762,
      "grad_norm": 2.2457776069641113,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 5065
    },
    {
      "epoch": 0.45467600071800396,
      "grad_norm": 3.2252914905548096,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 5066
    },
    {
      "epoch": 0.45476575121163165,
      "grad_norm": 3.6484858989715576,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 5067
    },
    {
      "epoch": 0.4548555017052594,
      "grad_norm": 2.384270191192627,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 5068
    },
    {
      "epoch": 0.4549452521988871,
      "grad_norm": 3.2756245136260986,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 5069
    },
    {
      "epoch": 0.4550350026925148,
      "grad_norm": 2.8843469619750977,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 5070
    },
    {
      "epoch": 0.45512475318614254,
      "grad_norm": 2.065133571624756,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 5071
    },
    {
      "epoch": 0.45521450367977023,
      "grad_norm": 2.4670145511627197,
      "learning_rate": 0.0,
      "loss": 0.6808,
      "step": 5072
    },
    {
      "epoch": 0.4553042541733979,
      "grad_norm": 2.986736297607422,
      "learning_rate": 0.0,
      "loss": 0.6887,
      "step": 5073
    },
    {
      "epoch": 0.4553940046670257,
      "grad_norm": 2.4328927993774414,
      "learning_rate": 0.0,
      "loss": 0.7676,
      "step": 5074
    },
    {
      "epoch": 0.45548375516065337,
      "grad_norm": 3.3781898021698,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 5075
    },
    {
      "epoch": 0.4555735056542811,
      "grad_norm": 3.5852975845336914,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 5076
    },
    {
      "epoch": 0.4556632561479088,
      "grad_norm": 3.159106969833374,
      "learning_rate": 0.0,
      "loss": 0.7533,
      "step": 5077
    },
    {
      "epoch": 0.4557530066415365,
      "grad_norm": 3.2570552825927734,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 5078
    },
    {
      "epoch": 0.45584275713516426,
      "grad_norm": 2.917376756668091,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 5079
    },
    {
      "epoch": 0.45593250762879195,
      "grad_norm": 2.659196138381958,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 5080
    },
    {
      "epoch": 0.4560222581224197,
      "grad_norm": 5.221518516540527,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 5081
    },
    {
      "epoch": 0.4561120086160474,
      "grad_norm": 2.444531202316284,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 5082
    },
    {
      "epoch": 0.4562017591096751,
      "grad_norm": 2.5825603008270264,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 5083
    },
    {
      "epoch": 0.45629150960330284,
      "grad_norm": 3.189615249633789,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 5084
    },
    {
      "epoch": 0.45638126009693053,
      "grad_norm": 3.1204588413238525,
      "learning_rate": 0.0,
      "loss": 0.8071,
      "step": 5085
    },
    {
      "epoch": 0.4564710105905582,
      "grad_norm": 2.4448108673095703,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 5086
    },
    {
      "epoch": 0.456560761084186,
      "grad_norm": 2.9946610927581787,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 5087
    },
    {
      "epoch": 0.45665051157781367,
      "grad_norm": 3.688863515853882,
      "learning_rate": 0.0,
      "loss": 0.7356,
      "step": 5088
    },
    {
      "epoch": 0.4567402620714414,
      "grad_norm": 3.7720844745635986,
      "learning_rate": 0.0,
      "loss": 0.6725,
      "step": 5089
    },
    {
      "epoch": 0.4568300125650691,
      "grad_norm": 4.136844635009766,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 5090
    },
    {
      "epoch": 0.4569197630586968,
      "grad_norm": 3.31613826751709,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 5091
    },
    {
      "epoch": 0.45700951355232455,
      "grad_norm": 3.1433238983154297,
      "learning_rate": 0.0,
      "loss": 0.7519,
      "step": 5092
    },
    {
      "epoch": 0.45709926404595225,
      "grad_norm": 3.3560080528259277,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 5093
    },
    {
      "epoch": 0.45718901453957994,
      "grad_norm": 3.809455394744873,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 5094
    },
    {
      "epoch": 0.4572787650332077,
      "grad_norm": 2.526275157928467,
      "learning_rate": 0.0,
      "loss": 0.7356,
      "step": 5095
    },
    {
      "epoch": 0.4573685155268354,
      "grad_norm": 2.387366771697998,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 5096
    },
    {
      "epoch": 0.45745826602046313,
      "grad_norm": 2.697589874267578,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 5097
    },
    {
      "epoch": 0.4575480165140908,
      "grad_norm": 3.4920268058776855,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 5098
    },
    {
      "epoch": 0.4576377670077185,
      "grad_norm": 2.9593684673309326,
      "learning_rate": 0.0,
      "loss": 0.798,
      "step": 5099
    },
    {
      "epoch": 0.45772751750134627,
      "grad_norm": 2.6578338146209717,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 5100
    },
    {
      "epoch": 0.45781726799497396,
      "grad_norm": 3.6170873641967773,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 5101
    },
    {
      "epoch": 0.4579070184886017,
      "grad_norm": 3.3407647609710693,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 5102
    },
    {
      "epoch": 0.4579967689822294,
      "grad_norm": 2.8941986560821533,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 5103
    },
    {
      "epoch": 0.4580865194758571,
      "grad_norm": 2.4906296730041504,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 5104
    },
    {
      "epoch": 0.45817626996948485,
      "grad_norm": 2.6305272579193115,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 5105
    },
    {
      "epoch": 0.45826602046311254,
      "grad_norm": 2.800523042678833,
      "learning_rate": 0.0,
      "loss": 0.6332,
      "step": 5106
    },
    {
      "epoch": 0.45835577095674024,
      "grad_norm": 2.489527463912964,
      "learning_rate": 0.0,
      "loss": 0.6447,
      "step": 5107
    },
    {
      "epoch": 0.458445521450368,
      "grad_norm": 3.2491748332977295,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 5108
    },
    {
      "epoch": 0.4585352719439957,
      "grad_norm": 2.9684746265411377,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 5109
    },
    {
      "epoch": 0.45862502243762343,
      "grad_norm": 4.518491744995117,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 5110
    },
    {
      "epoch": 0.4587147729312511,
      "grad_norm": 3.2910614013671875,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 5111
    },
    {
      "epoch": 0.4588045234248788,
      "grad_norm": 2.716648817062378,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 5112
    },
    {
      "epoch": 0.45889427391850657,
      "grad_norm": 2.4816150665283203,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 5113
    },
    {
      "epoch": 0.45898402441213426,
      "grad_norm": 2.070204019546509,
      "learning_rate": 0.0,
      "loss": 0.724,
      "step": 5114
    },
    {
      "epoch": 0.45907377490576196,
      "grad_norm": 3.598036766052246,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 5115
    },
    {
      "epoch": 0.4591635253993897,
      "grad_norm": 2.889512538909912,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 5116
    },
    {
      "epoch": 0.4592532758930174,
      "grad_norm": 2.7770440578460693,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 5117
    },
    {
      "epoch": 0.45934302638664515,
      "grad_norm": 9.131109237670898,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 5118
    },
    {
      "epoch": 0.45943277688027284,
      "grad_norm": 3.8483264446258545,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 5119
    },
    {
      "epoch": 0.45952252737390054,
      "grad_norm": 3.4493634700775146,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 5120
    },
    {
      "epoch": 0.4596122778675283,
      "grad_norm": 2.2760915756225586,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 5121
    },
    {
      "epoch": 0.459702028361156,
      "grad_norm": 2.61147403717041,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 5122
    },
    {
      "epoch": 0.45979177885478373,
      "grad_norm": 4.920708179473877,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 5123
    },
    {
      "epoch": 0.4598815293484114,
      "grad_norm": 4.692398548126221,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 5124
    },
    {
      "epoch": 0.4599712798420391,
      "grad_norm": 15.472633361816406,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 5125
    },
    {
      "epoch": 0.46006103033566687,
      "grad_norm": 3.713942527770996,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 5126
    },
    {
      "epoch": 0.46015078082929456,
      "grad_norm": 2.6840507984161377,
      "learning_rate": 0.0,
      "loss": 0.6694,
      "step": 5127
    },
    {
      "epoch": 0.46024053132292225,
      "grad_norm": 2.0855207443237305,
      "learning_rate": 0.0,
      "loss": 0.7128,
      "step": 5128
    },
    {
      "epoch": 0.46033028181655,
      "grad_norm": 2.832771062850952,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 5129
    },
    {
      "epoch": 0.4604200323101777,
      "grad_norm": 1.999880313873291,
      "learning_rate": 0.0,
      "loss": 0.6682,
      "step": 5130
    },
    {
      "epoch": 0.46050978280380545,
      "grad_norm": 3.4416184425354004,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 5131
    },
    {
      "epoch": 0.46059953329743314,
      "grad_norm": 2.3249168395996094,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 5132
    },
    {
      "epoch": 0.46068928379106083,
      "grad_norm": 4.030741214752197,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 5133
    },
    {
      "epoch": 0.4607790342846886,
      "grad_norm": 3.282503843307495,
      "learning_rate": 0.0,
      "loss": 0.7609,
      "step": 5134
    },
    {
      "epoch": 0.4608687847783163,
      "grad_norm": 2.7492430210113525,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 5135
    },
    {
      "epoch": 0.46095853527194397,
      "grad_norm": 3.217477560043335,
      "learning_rate": 0.0,
      "loss": 0.7659,
      "step": 5136
    },
    {
      "epoch": 0.4610482857655717,
      "grad_norm": 2.6594655513763428,
      "learning_rate": 0.0,
      "loss": 0.7626,
      "step": 5137
    },
    {
      "epoch": 0.4611380362591994,
      "grad_norm": 4.0408101081848145,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 5138
    },
    {
      "epoch": 0.46122778675282716,
      "grad_norm": 2.8091044425964355,
      "learning_rate": 0.0,
      "loss": 0.694,
      "step": 5139
    },
    {
      "epoch": 0.46131753724645486,
      "grad_norm": 3.401927947998047,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 5140
    },
    {
      "epoch": 0.46140728774008255,
      "grad_norm": 2.72277569770813,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 5141
    },
    {
      "epoch": 0.4614970382337103,
      "grad_norm": 2.6211376190185547,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 5142
    },
    {
      "epoch": 0.461586788727338,
      "grad_norm": 2.6837751865386963,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 5143
    },
    {
      "epoch": 0.46167653922096574,
      "grad_norm": 2.914430856704712,
      "learning_rate": 0.0,
      "loss": 0.6664,
      "step": 5144
    },
    {
      "epoch": 0.46176628971459344,
      "grad_norm": 2.604182481765747,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 5145
    },
    {
      "epoch": 0.46185604020822113,
      "grad_norm": 3.232377529144287,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 5146
    },
    {
      "epoch": 0.4619457907018489,
      "grad_norm": 4.318670749664307,
      "learning_rate": 0.0,
      "loss": 0.6792,
      "step": 5147
    },
    {
      "epoch": 0.4620355411954766,
      "grad_norm": 3.83831787109375,
      "learning_rate": 0.0,
      "loss": 0.7203,
      "step": 5148
    },
    {
      "epoch": 0.46212529168910427,
      "grad_norm": 3.348841905593872,
      "learning_rate": 0.0,
      "loss": 0.7871,
      "step": 5149
    },
    {
      "epoch": 0.462215042182732,
      "grad_norm": 2.51025390625,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 5150
    },
    {
      "epoch": 0.4623047926763597,
      "grad_norm": 3.3780171871185303,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 5151
    },
    {
      "epoch": 0.46239454316998746,
      "grad_norm": 5.498173713684082,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 5152
    },
    {
      "epoch": 0.46248429366361515,
      "grad_norm": 3.590632438659668,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 5153
    },
    {
      "epoch": 0.46257404415724285,
      "grad_norm": 2.6699774265289307,
      "learning_rate": 0.0,
      "loss": 0.8016,
      "step": 5154
    },
    {
      "epoch": 0.4626637946508706,
      "grad_norm": 3.143749952316284,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 5155
    },
    {
      "epoch": 0.4627535451444983,
      "grad_norm": 2.8667311668395996,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 5156
    },
    {
      "epoch": 0.462843295638126,
      "grad_norm": 3.057385206222534,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 5157
    },
    {
      "epoch": 0.46293304613175373,
      "grad_norm": 2.252390146255493,
      "learning_rate": 0.0,
      "loss": 0.7547,
      "step": 5158
    },
    {
      "epoch": 0.4630227966253814,
      "grad_norm": 5.700392723083496,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 5159
    },
    {
      "epoch": 0.4631125471190092,
      "grad_norm": 5.119565963745117,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 5160
    },
    {
      "epoch": 0.46320229761263687,
      "grad_norm": 3.147559642791748,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 5161
    },
    {
      "epoch": 0.46329204810626456,
      "grad_norm": 2.6265957355499268,
      "learning_rate": 0.0,
      "loss": 0.6754,
      "step": 5162
    },
    {
      "epoch": 0.4633817985998923,
      "grad_norm": 2.6840310096740723,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 5163
    },
    {
      "epoch": 0.46347154909352,
      "grad_norm": 3.174769878387451,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 5164
    },
    {
      "epoch": 0.4635612995871477,
      "grad_norm": 3.2374048233032227,
      "learning_rate": 0.0,
      "loss": 0.7119,
      "step": 5165
    },
    {
      "epoch": 0.46365105008077545,
      "grad_norm": 4.237705707550049,
      "learning_rate": 0.0,
      "loss": 0.8201,
      "step": 5166
    },
    {
      "epoch": 0.46374080057440314,
      "grad_norm": 3.671229839324951,
      "learning_rate": 0.0,
      "loss": 0.6947,
      "step": 5167
    },
    {
      "epoch": 0.4638305510680309,
      "grad_norm": 2.765622615814209,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 5168
    },
    {
      "epoch": 0.4639203015616586,
      "grad_norm": 3.8885719776153564,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 5169
    },
    {
      "epoch": 0.4640100520552863,
      "grad_norm": 3.520707130432129,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 5170
    },
    {
      "epoch": 0.46409980254891403,
      "grad_norm": 21.645288467407227,
      "learning_rate": 0.0,
      "loss": 0.7774,
      "step": 5171
    },
    {
      "epoch": 0.4641895530425417,
      "grad_norm": 2.909587860107422,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 5172
    },
    {
      "epoch": 0.4642793035361695,
      "grad_norm": 4.731481552124023,
      "learning_rate": 0.0,
      "loss": 0.7418,
      "step": 5173
    },
    {
      "epoch": 0.46436905402979717,
      "grad_norm": 4.415790557861328,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 5174
    },
    {
      "epoch": 0.46445880452342486,
      "grad_norm": 3.7141904830932617,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 5175
    },
    {
      "epoch": 0.4645485550170526,
      "grad_norm": 2.765007495880127,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 5176
    },
    {
      "epoch": 0.4646383055106803,
      "grad_norm": 2.283210277557373,
      "learning_rate": 0.0,
      "loss": 0.7722,
      "step": 5177
    },
    {
      "epoch": 0.464728056004308,
      "grad_norm": 2.8801491260528564,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 5178
    },
    {
      "epoch": 0.46481780649793575,
      "grad_norm": 2.6160919666290283,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 5179
    },
    {
      "epoch": 0.46490755699156344,
      "grad_norm": 2.628105640411377,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 5180
    },
    {
      "epoch": 0.4649973074851912,
      "grad_norm": 2.4481394290924072,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 5181
    },
    {
      "epoch": 0.4650870579788189,
      "grad_norm": 2.8003172874450684,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 5182
    },
    {
      "epoch": 0.4651768084724466,
      "grad_norm": 3.3413150310516357,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 5183
    },
    {
      "epoch": 0.46526655896607433,
      "grad_norm": 2.592599391937256,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 5184
    },
    {
      "epoch": 0.465356309459702,
      "grad_norm": 2.8533313274383545,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 5185
    },
    {
      "epoch": 0.4654460599533297,
      "grad_norm": 8.79692268371582,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 5186
    },
    {
      "epoch": 0.46553581044695747,
      "grad_norm": 5.2210235595703125,
      "learning_rate": 0.0,
      "loss": 0.7662,
      "step": 5187
    },
    {
      "epoch": 0.46562556094058516,
      "grad_norm": 3.0521600246429443,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 5188
    },
    {
      "epoch": 0.4657153114342129,
      "grad_norm": 2.9429738521575928,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 5189
    },
    {
      "epoch": 0.4658050619278406,
      "grad_norm": 5.072442054748535,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 5190
    },
    {
      "epoch": 0.4658948124214683,
      "grad_norm": 8.290590286254883,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 5191
    },
    {
      "epoch": 0.46598456291509605,
      "grad_norm": 4.1067423820495605,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 5192
    },
    {
      "epoch": 0.46607431340872374,
      "grad_norm": 2.8078646659851074,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 5193
    },
    {
      "epoch": 0.4661640639023515,
      "grad_norm": 2.242408514022827,
      "learning_rate": 0.0,
      "loss": 0.6745,
      "step": 5194
    },
    {
      "epoch": 0.4662538143959792,
      "grad_norm": 4.063770771026611,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 5195
    },
    {
      "epoch": 0.4663435648896069,
      "grad_norm": 8.577680587768555,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 5196
    },
    {
      "epoch": 0.4664333153832346,
      "grad_norm": 3.464888095855713,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 5197
    },
    {
      "epoch": 0.4665230658768623,
      "grad_norm": 21.731124877929688,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 5198
    },
    {
      "epoch": 0.46661281637049,
      "grad_norm": 3.3321661949157715,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 5199
    },
    {
      "epoch": 0.46670256686411776,
      "grad_norm": 3.5773472785949707,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 5200
    },
    {
      "epoch": 0.46679231735774546,
      "grad_norm": 2.4049267768859863,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 5201
    },
    {
      "epoch": 0.4668820678513732,
      "grad_norm": 2.7304670810699463,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 5202
    },
    {
      "epoch": 0.4669718183450009,
      "grad_norm": 2.4594414234161377,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 5203
    },
    {
      "epoch": 0.4670615688386286,
      "grad_norm": 3.0757973194122314,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 5204
    },
    {
      "epoch": 0.46715131933225634,
      "grad_norm": 3.3875327110290527,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 5205
    },
    {
      "epoch": 0.46724106982588404,
      "grad_norm": 2.4908902645111084,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 5206
    },
    {
      "epoch": 0.46733082031951173,
      "grad_norm": 4.147172927856445,
      "learning_rate": 0.0,
      "loss": 0.7867,
      "step": 5207
    },
    {
      "epoch": 0.4674205708131395,
      "grad_norm": 3.31754469871521,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 5208
    },
    {
      "epoch": 0.4675103213067672,
      "grad_norm": 3.4395480155944824,
      "learning_rate": 0.0,
      "loss": 0.761,
      "step": 5209
    },
    {
      "epoch": 0.4676000718003949,
      "grad_norm": 3.5587687492370605,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 5210
    },
    {
      "epoch": 0.4676898222940226,
      "grad_norm": 7.551653861999512,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 5211
    },
    {
      "epoch": 0.4677795727876503,
      "grad_norm": 4.418292045593262,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 5212
    },
    {
      "epoch": 0.46786932328127806,
      "grad_norm": 4.601284027099609,
      "learning_rate": 0.0,
      "loss": 0.7534,
      "step": 5213
    },
    {
      "epoch": 0.46795907377490575,
      "grad_norm": 3.257441759109497,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 5214
    },
    {
      "epoch": 0.4680488242685335,
      "grad_norm": 3.227104425430298,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 5215
    },
    {
      "epoch": 0.4681385747621612,
      "grad_norm": 3.2907116413116455,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 5216
    },
    {
      "epoch": 0.4682283252557889,
      "grad_norm": 2.930926561355591,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 5217
    },
    {
      "epoch": 0.46831807574941664,
      "grad_norm": 3.0837619304656982,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 5218
    },
    {
      "epoch": 0.46840782624304433,
      "grad_norm": 2.6636099815368652,
      "learning_rate": 0.0,
      "loss": 0.7356,
      "step": 5219
    },
    {
      "epoch": 0.46849757673667203,
      "grad_norm": 5.809447765350342,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 5220
    },
    {
      "epoch": 0.4685873272302998,
      "grad_norm": 8.471138954162598,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 5221
    },
    {
      "epoch": 0.46867707772392747,
      "grad_norm": 3.3930504322052,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 5222
    },
    {
      "epoch": 0.4687668282175552,
      "grad_norm": 2.58798885345459,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 5223
    },
    {
      "epoch": 0.4688565787111829,
      "grad_norm": 4.271694660186768,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 5224
    },
    {
      "epoch": 0.4689463292048106,
      "grad_norm": 3.4815549850463867,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 5225
    },
    {
      "epoch": 0.46903607969843836,
      "grad_norm": 6.949792385101318,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 5226
    },
    {
      "epoch": 0.46912583019206605,
      "grad_norm": 5.168853282928467,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 5227
    },
    {
      "epoch": 0.46921558068569375,
      "grad_norm": 4.041531562805176,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 5228
    },
    {
      "epoch": 0.4693053311793215,
      "grad_norm": 4.072425365447998,
      "learning_rate": 0.0,
      "loss": 0.7631,
      "step": 5229
    },
    {
      "epoch": 0.4693950816729492,
      "grad_norm": 3.7030246257781982,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 5230
    },
    {
      "epoch": 0.46948483216657694,
      "grad_norm": 4.04222297668457,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 5231
    },
    {
      "epoch": 0.46957458266020463,
      "grad_norm": 3.407548427581787,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 5232
    },
    {
      "epoch": 0.4696643331538323,
      "grad_norm": 2.674443483352661,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 5233
    },
    {
      "epoch": 0.4697540836474601,
      "grad_norm": 1.8232043981552124,
      "learning_rate": 0.0,
      "loss": 0.8042,
      "step": 5234
    },
    {
      "epoch": 0.46984383414108777,
      "grad_norm": 3.033463478088379,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 5235
    },
    {
      "epoch": 0.4699335846347155,
      "grad_norm": 3.9679808616638184,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 5236
    },
    {
      "epoch": 0.4700233351283432,
      "grad_norm": 2.6265618801116943,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 5237
    },
    {
      "epoch": 0.4701130856219709,
      "grad_norm": 3.485875129699707,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 5238
    },
    {
      "epoch": 0.47020283611559865,
      "grad_norm": 2.288625955581665,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 5239
    },
    {
      "epoch": 0.47029258660922635,
      "grad_norm": 3.777310609817505,
      "learning_rate": 0.0,
      "loss": 0.7067,
      "step": 5240
    },
    {
      "epoch": 0.47038233710285404,
      "grad_norm": 3.3023338317871094,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 5241
    },
    {
      "epoch": 0.4704720875964818,
      "grad_norm": 2.8444325923919678,
      "learning_rate": 0.0,
      "loss": 0.7642,
      "step": 5242
    },
    {
      "epoch": 0.4705618380901095,
      "grad_norm": 3.321080207824707,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 5243
    },
    {
      "epoch": 0.47065158858373723,
      "grad_norm": 3.7712414264678955,
      "learning_rate": 0.0,
      "loss": 0.8163,
      "step": 5244
    },
    {
      "epoch": 0.47074133907736493,
      "grad_norm": 2.654762029647827,
      "learning_rate": 0.0,
      "loss": 0.7043,
      "step": 5245
    },
    {
      "epoch": 0.4708310895709926,
      "grad_norm": 2.737752676010132,
      "learning_rate": 0.0,
      "loss": 0.7021,
      "step": 5246
    },
    {
      "epoch": 0.47092084006462037,
      "grad_norm": 2.4074594974517822,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 5247
    },
    {
      "epoch": 0.47101059055824807,
      "grad_norm": 4.395518779754639,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 5248
    },
    {
      "epoch": 0.47110034105187576,
      "grad_norm": 2.639402389526367,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 5249
    },
    {
      "epoch": 0.4711900915455035,
      "grad_norm": 3.7004106044769287,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "step": 5250
    },
    {
      "epoch": 0.4712798420391312,
      "grad_norm": 2.707744598388672,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 5251
    },
    {
      "epoch": 0.47136959253275895,
      "grad_norm": 2.7953407764434814,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 5252
    },
    {
      "epoch": 0.47145934302638665,
      "grad_norm": 3.1652534008026123,
      "learning_rate": 0.0,
      "loss": 0.6665,
      "step": 5253
    },
    {
      "epoch": 0.47154909352001434,
      "grad_norm": 5.861434459686279,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 5254
    },
    {
      "epoch": 0.4716388440136421,
      "grad_norm": 2.7367289066314697,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 5255
    },
    {
      "epoch": 0.4717285945072698,
      "grad_norm": 2.9646241664886475,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 5256
    },
    {
      "epoch": 0.47181834500089753,
      "grad_norm": 4.383591651916504,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 5257
    },
    {
      "epoch": 0.4719080954945252,
      "grad_norm": 3.030771493911743,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 5258
    },
    {
      "epoch": 0.4719978459881529,
      "grad_norm": 3.742990493774414,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 5259
    },
    {
      "epoch": 0.47208759648178067,
      "grad_norm": 2.9632232189178467,
      "learning_rate": 0.0,
      "loss": 0.6859,
      "step": 5260
    },
    {
      "epoch": 0.47217734697540836,
      "grad_norm": 4.225264072418213,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 5261
    },
    {
      "epoch": 0.47226709746903606,
      "grad_norm": 2.6797258853912354,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 5262
    },
    {
      "epoch": 0.4723568479626638,
      "grad_norm": 7.391566753387451,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 5263
    },
    {
      "epoch": 0.4724465984562915,
      "grad_norm": 7.038003444671631,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 5264
    },
    {
      "epoch": 0.47253634894991925,
      "grad_norm": 2.9369418621063232,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 5265
    },
    {
      "epoch": 0.47262609944354694,
      "grad_norm": 9.157232284545898,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 5266
    },
    {
      "epoch": 0.47271584993717464,
      "grad_norm": 2.9253528118133545,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 5267
    },
    {
      "epoch": 0.4728056004308024,
      "grad_norm": 4.304656505584717,
      "learning_rate": 0.0,
      "loss": 0.8048,
      "step": 5268
    },
    {
      "epoch": 0.4728953509244301,
      "grad_norm": 3.921505928039551,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 5269
    },
    {
      "epoch": 0.4729851014180578,
      "grad_norm": 2.4602649211883545,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 5270
    },
    {
      "epoch": 0.4730748519116855,
      "grad_norm": 8.764986991882324,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 5271
    },
    {
      "epoch": 0.4731646024053132,
      "grad_norm": 2.872912883758545,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 5272
    },
    {
      "epoch": 0.47325435289894097,
      "grad_norm": 2.4410147666931152,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 5273
    },
    {
      "epoch": 0.47334410339256866,
      "grad_norm": 3.206137180328369,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 5274
    },
    {
      "epoch": 0.47343385388619635,
      "grad_norm": 3.2031209468841553,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 5275
    },
    {
      "epoch": 0.4735236043798241,
      "grad_norm": 2.623764753341675,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 5276
    },
    {
      "epoch": 0.4736133548734518,
      "grad_norm": 2.6143627166748047,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 5277
    },
    {
      "epoch": 0.47370310536707955,
      "grad_norm": 2.0620601177215576,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 5278
    },
    {
      "epoch": 0.47379285586070724,
      "grad_norm": 3.2035481929779053,
      "learning_rate": 0.0,
      "loss": 0.6862,
      "step": 5279
    },
    {
      "epoch": 0.47388260635433493,
      "grad_norm": 3.30072283744812,
      "learning_rate": 0.0,
      "loss": 0.7468,
      "step": 5280
    },
    {
      "epoch": 0.4739723568479627,
      "grad_norm": 11.438508987426758,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 5281
    },
    {
      "epoch": 0.4740621073415904,
      "grad_norm": 6.248138904571533,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 5282
    },
    {
      "epoch": 0.47415185783521807,
      "grad_norm": 3.0010035037994385,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 5283
    },
    {
      "epoch": 0.4742416083288458,
      "grad_norm": 2.677743911743164,
      "learning_rate": 0.0,
      "loss": 0.719,
      "step": 5284
    },
    {
      "epoch": 0.4743313588224735,
      "grad_norm": 3.4025893211364746,
      "learning_rate": 0.0,
      "loss": 0.6882,
      "step": 5285
    },
    {
      "epoch": 0.47442110931610126,
      "grad_norm": 2.9272570610046387,
      "learning_rate": 0.0,
      "loss": 0.6809,
      "step": 5286
    },
    {
      "epoch": 0.47451085980972896,
      "grad_norm": 3.239398956298828,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 5287
    },
    {
      "epoch": 0.47460061030335665,
      "grad_norm": 2.586606502532959,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 5288
    },
    {
      "epoch": 0.4746903607969844,
      "grad_norm": 2.5175118446350098,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 5289
    },
    {
      "epoch": 0.4747801112906121,
      "grad_norm": 2.921165943145752,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 5290
    },
    {
      "epoch": 0.4748698617842398,
      "grad_norm": 2.8915371894836426,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 5291
    },
    {
      "epoch": 0.47495961227786754,
      "grad_norm": 3.320449113845825,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 5292
    },
    {
      "epoch": 0.47504936277149523,
      "grad_norm": 3.3938963413238525,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 5293
    },
    {
      "epoch": 0.475139113265123,
      "grad_norm": 26.413509368896484,
      "learning_rate": 0.0,
      "loss": 0.6995,
      "step": 5294
    },
    {
      "epoch": 0.4752288637587507,
      "grad_norm": 4.4699177742004395,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 5295
    },
    {
      "epoch": 0.47531861425237837,
      "grad_norm": 8.302247047424316,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 5296
    },
    {
      "epoch": 0.4754083647460061,
      "grad_norm": 16.574384689331055,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 5297
    },
    {
      "epoch": 0.4754981152396338,
      "grad_norm": 3.4945459365844727,
      "learning_rate": 0.0,
      "loss": 0.7453,
      "step": 5298
    },
    {
      "epoch": 0.4755878657332615,
      "grad_norm": 2.5979154109954834,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 5299
    },
    {
      "epoch": 0.47567761622688925,
      "grad_norm": 2.454394817352295,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 5300
    },
    {
      "epoch": 0.47576736672051695,
      "grad_norm": 3.268425941467285,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 5301
    },
    {
      "epoch": 0.4758571172141447,
      "grad_norm": 2.8658039569854736,
      "learning_rate": 0.0,
      "loss": 0.807,
      "step": 5302
    },
    {
      "epoch": 0.4759468677077724,
      "grad_norm": 3.162914276123047,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 5303
    },
    {
      "epoch": 0.4760366182014001,
      "grad_norm": 2.6599907875061035,
      "learning_rate": 0.0,
      "loss": 0.7505,
      "step": 5304
    },
    {
      "epoch": 0.47612636869502784,
      "grad_norm": 3.8251774311065674,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 5305
    },
    {
      "epoch": 0.47621611918865553,
      "grad_norm": 2.6986966133117676,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 5306
    },
    {
      "epoch": 0.4763058696822833,
      "grad_norm": 4.4316864013671875,
      "learning_rate": 0.0,
      "loss": 0.6892,
      "step": 5307
    },
    {
      "epoch": 0.47639562017591097,
      "grad_norm": 2.2419188022613525,
      "learning_rate": 0.0,
      "loss": 0.7819,
      "step": 5308
    },
    {
      "epoch": 0.47648537066953867,
      "grad_norm": 3.261449098587036,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 5309
    },
    {
      "epoch": 0.4765751211631664,
      "grad_norm": 3.489377021789551,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 5310
    },
    {
      "epoch": 0.4766648716567941,
      "grad_norm": 5.663569927215576,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 5311
    },
    {
      "epoch": 0.4767546221504218,
      "grad_norm": 3.6683390140533447,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 5312
    },
    {
      "epoch": 0.47684437264404955,
      "grad_norm": 3.485819101333618,
      "learning_rate": 0.0,
      "loss": 0.7547,
      "step": 5313
    },
    {
      "epoch": 0.47693412313767725,
      "grad_norm": 3.061870813369751,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 5314
    },
    {
      "epoch": 0.477023873631305,
      "grad_norm": 2.881490707397461,
      "learning_rate": 0.0,
      "loss": 0.6754,
      "step": 5315
    },
    {
      "epoch": 0.4771136241249327,
      "grad_norm": 3.093437671661377,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 5316
    },
    {
      "epoch": 0.4772033746185604,
      "grad_norm": 3.2997195720672607,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 5317
    },
    {
      "epoch": 0.47729312511218813,
      "grad_norm": 2.958798885345459,
      "learning_rate": 0.0,
      "loss": 0.679,
      "step": 5318
    },
    {
      "epoch": 0.4773828756058158,
      "grad_norm": 3.48769211769104,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 5319
    },
    {
      "epoch": 0.4774726260994435,
      "grad_norm": 3.364591598510742,
      "learning_rate": 0.0,
      "loss": 0.7762,
      "step": 5320
    },
    {
      "epoch": 0.47756237659307127,
      "grad_norm": 2.844841241836548,
      "learning_rate": 0.0,
      "loss": 0.6849,
      "step": 5321
    },
    {
      "epoch": 0.47765212708669896,
      "grad_norm": 3.279818058013916,
      "learning_rate": 0.0,
      "loss": 0.771,
      "step": 5322
    },
    {
      "epoch": 0.4777418775803267,
      "grad_norm": 4.043025493621826,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 5323
    },
    {
      "epoch": 0.4778316280739544,
      "grad_norm": 2.905280113220215,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 5324
    },
    {
      "epoch": 0.4779213785675821,
      "grad_norm": 2.8187577724456787,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 5325
    },
    {
      "epoch": 0.47801112906120985,
      "grad_norm": 2.8893072605133057,
      "learning_rate": 0.0,
      "loss": 0.7316,
      "step": 5326
    },
    {
      "epoch": 0.47810087955483754,
      "grad_norm": 3.9647908210754395,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 5327
    },
    {
      "epoch": 0.4781906300484653,
      "grad_norm": 4.134607315063477,
      "learning_rate": 0.0,
      "loss": 0.667,
      "step": 5328
    },
    {
      "epoch": 0.478280380542093,
      "grad_norm": 2.4967305660247803,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 5329
    },
    {
      "epoch": 0.4783701310357207,
      "grad_norm": 4.526304244995117,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 5330
    },
    {
      "epoch": 0.47845988152934843,
      "grad_norm": 3.57324481010437,
      "learning_rate": 0.0,
      "loss": 0.7506,
      "step": 5331
    },
    {
      "epoch": 0.4785496320229761,
      "grad_norm": 4.074728965759277,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 5332
    },
    {
      "epoch": 0.4786393825166038,
      "grad_norm": 10.629186630249023,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 5333
    },
    {
      "epoch": 0.47872913301023157,
      "grad_norm": 2.604708194732666,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 5334
    },
    {
      "epoch": 0.47881888350385926,
      "grad_norm": 4.209428310394287,
      "learning_rate": 0.0,
      "loss": 0.7997,
      "step": 5335
    },
    {
      "epoch": 0.478908633997487,
      "grad_norm": 2.475950241088867,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 5336
    },
    {
      "epoch": 0.4789983844911147,
      "grad_norm": 2.895468235015869,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 5337
    },
    {
      "epoch": 0.4790881349847424,
      "grad_norm": 3.4204673767089844,
      "learning_rate": 0.0,
      "loss": 0.7842,
      "step": 5338
    },
    {
      "epoch": 0.47917788547837015,
      "grad_norm": 2.853864908218384,
      "learning_rate": 0.0,
      "loss": 0.6745,
      "step": 5339
    },
    {
      "epoch": 0.47926763597199784,
      "grad_norm": 2.3948569297790527,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 5340
    },
    {
      "epoch": 0.47935738646562553,
      "grad_norm": 2.302764415740967,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 5341
    },
    {
      "epoch": 0.4794471369592533,
      "grad_norm": 4.244433879852295,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 5342
    },
    {
      "epoch": 0.479536887452881,
      "grad_norm": 3.2947239875793457,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 5343
    },
    {
      "epoch": 0.4796266379465087,
      "grad_norm": 2.577545642852783,
      "learning_rate": 0.0,
      "loss": 0.6848,
      "step": 5344
    },
    {
      "epoch": 0.4797163884401364,
      "grad_norm": 2.527050256729126,
      "learning_rate": 0.0,
      "loss": 0.786,
      "step": 5345
    },
    {
      "epoch": 0.4798061389337641,
      "grad_norm": 3.864680290222168,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 5346
    },
    {
      "epoch": 0.47989588942739186,
      "grad_norm": 2.289869785308838,
      "learning_rate": 0.0,
      "loss": 0.8459,
      "step": 5347
    },
    {
      "epoch": 0.47998563992101956,
      "grad_norm": 2.62465500831604,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 5348
    },
    {
      "epoch": 0.4800753904146473,
      "grad_norm": 2.6138083934783936,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 5349
    },
    {
      "epoch": 0.480165140908275,
      "grad_norm": 2.7935590744018555,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 5350
    },
    {
      "epoch": 0.4802548914019027,
      "grad_norm": 3.140228033065796,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 5351
    },
    {
      "epoch": 0.48034464189553044,
      "grad_norm": 9.498401641845703,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 5352
    },
    {
      "epoch": 0.48043439238915814,
      "grad_norm": 4.794589519500732,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 5353
    },
    {
      "epoch": 0.48052414288278583,
      "grad_norm": 5.172739505767822,
      "learning_rate": 0.0,
      "loss": 0.676,
      "step": 5354
    },
    {
      "epoch": 0.4806138933764136,
      "grad_norm": 2.551839590072632,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 5355
    },
    {
      "epoch": 0.4807036438700413,
      "grad_norm": 14.662843704223633,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 5356
    },
    {
      "epoch": 0.480793394363669,
      "grad_norm": 3.3402559757232666,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 5357
    },
    {
      "epoch": 0.4808831448572967,
      "grad_norm": 3.9516184329986572,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 5358
    },
    {
      "epoch": 0.4809728953509244,
      "grad_norm": 6.931956768035889,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 5359
    },
    {
      "epoch": 0.48106264584455216,
      "grad_norm": 4.016912937164307,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 5360
    },
    {
      "epoch": 0.48115239633817986,
      "grad_norm": 4.578266620635986,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 5361
    },
    {
      "epoch": 0.48124214683180755,
      "grad_norm": 4.604530334472656,
      "learning_rate": 0.0,
      "loss": 0.7329,
      "step": 5362
    },
    {
      "epoch": 0.4813318973254353,
      "grad_norm": 3.310682773590088,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 5363
    },
    {
      "epoch": 0.481421647819063,
      "grad_norm": 4.497254848480225,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 5364
    },
    {
      "epoch": 0.48151139831269074,
      "grad_norm": 2.851560592651367,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 5365
    },
    {
      "epoch": 0.48160114880631844,
      "grad_norm": 3.357314109802246,
      "learning_rate": 0.0,
      "loss": 0.6718,
      "step": 5366
    },
    {
      "epoch": 0.48169089929994613,
      "grad_norm": 3.696995973587036,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 5367
    },
    {
      "epoch": 0.4817806497935739,
      "grad_norm": 2.763596534729004,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 5368
    },
    {
      "epoch": 0.4818704002872016,
      "grad_norm": 3.1452465057373047,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 5369
    },
    {
      "epoch": 0.4819601507808293,
      "grad_norm": 2.9500794410705566,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 5370
    },
    {
      "epoch": 0.482049901274457,
      "grad_norm": 2.771355390548706,
      "learning_rate": 0.0,
      "loss": 0.6669,
      "step": 5371
    },
    {
      "epoch": 0.4821396517680847,
      "grad_norm": 9.021576881408691,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 5372
    },
    {
      "epoch": 0.48222940226171246,
      "grad_norm": 4.678915023803711,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 5373
    },
    {
      "epoch": 0.48231915275534015,
      "grad_norm": 4.560102939605713,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 5374
    },
    {
      "epoch": 0.48240890324896785,
      "grad_norm": 3.535076856613159,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 5375
    },
    {
      "epoch": 0.4824986537425956,
      "grad_norm": 2.5696282386779785,
      "learning_rate": 0.0,
      "loss": 0.7193,
      "step": 5376
    },
    {
      "epoch": 0.4825884042362233,
      "grad_norm": 48.16871643066406,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 5377
    },
    {
      "epoch": 0.48267815472985104,
      "grad_norm": 4.272710800170898,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 5378
    },
    {
      "epoch": 0.48276790522347873,
      "grad_norm": 3.1249916553497314,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 5379
    },
    {
      "epoch": 0.4828576557171064,
      "grad_norm": 3.3014979362487793,
      "learning_rate": 0.0,
      "loss": 0.7662,
      "step": 5380
    },
    {
      "epoch": 0.4829474062107342,
      "grad_norm": 3.645822525024414,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 5381
    },
    {
      "epoch": 0.48303715670436187,
      "grad_norm": 3.478684425354004,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 5382
    },
    {
      "epoch": 0.48312690719798956,
      "grad_norm": 2.7432188987731934,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 5383
    },
    {
      "epoch": 0.4832166576916173,
      "grad_norm": 3.3936359882354736,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 5384
    },
    {
      "epoch": 0.483306408185245,
      "grad_norm": 2.496168851852417,
      "learning_rate": 0.0,
      "loss": 0.7845,
      "step": 5385
    },
    {
      "epoch": 0.48339615867887276,
      "grad_norm": 3.1335816383361816,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 5386
    },
    {
      "epoch": 0.48348590917250045,
      "grad_norm": 3.2963056564331055,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 5387
    },
    {
      "epoch": 0.48357565966612814,
      "grad_norm": 2.7384819984436035,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 5388
    },
    {
      "epoch": 0.4836654101597559,
      "grad_norm": 7.862074851989746,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 5389
    },
    {
      "epoch": 0.4837551606533836,
      "grad_norm": 5.353908538818359,
      "learning_rate": 0.0,
      "loss": 0.7353,
      "step": 5390
    },
    {
      "epoch": 0.48384491114701134,
      "grad_norm": 4.377378463745117,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 5391
    },
    {
      "epoch": 0.48393466164063903,
      "grad_norm": 2.738168478012085,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 5392
    },
    {
      "epoch": 0.4840244121342667,
      "grad_norm": 4.357721328735352,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 5393
    },
    {
      "epoch": 0.4841141626278945,
      "grad_norm": 3.2833306789398193,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 5394
    },
    {
      "epoch": 0.48420391312152217,
      "grad_norm": 3.9923055171966553,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 5395
    },
    {
      "epoch": 0.48429366361514986,
      "grad_norm": 3.576113224029541,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 5396
    },
    {
      "epoch": 0.4843834141087776,
      "grad_norm": 2.896968364715576,
      "learning_rate": 0.0,
      "loss": 0.7853,
      "step": 5397
    },
    {
      "epoch": 0.4844731646024053,
      "grad_norm": 2.851698875427246,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 5398
    },
    {
      "epoch": 0.48456291509603305,
      "grad_norm": 3.889266014099121,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 5399
    },
    {
      "epoch": 0.48465266558966075,
      "grad_norm": 3.647599458694458,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 5400
    },
    {
      "epoch": 0.48474241608328844,
      "grad_norm": 3.089772939682007,
      "learning_rate": 0.0,
      "loss": 0.7863,
      "step": 5401
    },
    {
      "epoch": 0.4848321665769162,
      "grad_norm": 5.501049995422363,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 5402
    },
    {
      "epoch": 0.4849219170705439,
      "grad_norm": 3.2880606651306152,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 5403
    },
    {
      "epoch": 0.4850116675641716,
      "grad_norm": 2.9533793926239014,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 5404
    },
    {
      "epoch": 0.4851014180577993,
      "grad_norm": 3.776560068130493,
      "learning_rate": 0.0,
      "loss": 0.6891,
      "step": 5405
    },
    {
      "epoch": 0.485191168551427,
      "grad_norm": 3.257643461227417,
      "learning_rate": 0.0,
      "loss": 0.781,
      "step": 5406
    },
    {
      "epoch": 0.48528091904505477,
      "grad_norm": 2.7997491359710693,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 5407
    },
    {
      "epoch": 0.48537066953868246,
      "grad_norm": 3.3062005043029785,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 5408
    },
    {
      "epoch": 0.48546042003231016,
      "grad_norm": 3.014458417892456,
      "learning_rate": 0.0,
      "loss": 0.795,
      "step": 5409
    },
    {
      "epoch": 0.4855501705259379,
      "grad_norm": 4.4628214836120605,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 5410
    },
    {
      "epoch": 0.4856399210195656,
      "grad_norm": 2.9111461639404297,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 5411
    },
    {
      "epoch": 0.4857296715131933,
      "grad_norm": 2.575895309448242,
      "learning_rate": 0.0,
      "loss": 0.7749,
      "step": 5412
    },
    {
      "epoch": 0.48581942200682104,
      "grad_norm": 4.107630252838135,
      "learning_rate": 0.0,
      "loss": 0.7346,
      "step": 5413
    },
    {
      "epoch": 0.48590917250044874,
      "grad_norm": 3.2184054851531982,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 5414
    },
    {
      "epoch": 0.4859989229940765,
      "grad_norm": 3.395689010620117,
      "learning_rate": 0.0,
      "loss": 0.6492,
      "step": 5415
    },
    {
      "epoch": 0.4860886734877042,
      "grad_norm": 5.7687811851501465,
      "learning_rate": 0.0,
      "loss": 0.7193,
      "step": 5416
    },
    {
      "epoch": 0.4861784239813319,
      "grad_norm": 3.6209652423858643,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 5417
    },
    {
      "epoch": 0.4862681744749596,
      "grad_norm": 2.948364734649658,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 5418
    },
    {
      "epoch": 0.4863579249685873,
      "grad_norm": 3.403362989425659,
      "learning_rate": 0.0,
      "loss": 0.7844,
      "step": 5419
    },
    {
      "epoch": 0.48644767546221507,
      "grad_norm": 2.592698097229004,
      "learning_rate": 0.0,
      "loss": 0.6836,
      "step": 5420
    },
    {
      "epoch": 0.48653742595584276,
      "grad_norm": 2.9757673740386963,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 5421
    },
    {
      "epoch": 0.48662717644947046,
      "grad_norm": 5.178384304046631,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 5422
    },
    {
      "epoch": 0.4867169269430982,
      "grad_norm": 5.2330217361450195,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 5423
    },
    {
      "epoch": 0.4868066774367259,
      "grad_norm": 3.2393672466278076,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 5424
    },
    {
      "epoch": 0.4868964279303536,
      "grad_norm": 2.869197368621826,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 5425
    },
    {
      "epoch": 0.48698617842398134,
      "grad_norm": 2.425304651260376,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 5426
    },
    {
      "epoch": 0.48707592891760904,
      "grad_norm": 3.3953170776367188,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 5427
    },
    {
      "epoch": 0.4871656794112368,
      "grad_norm": 4.48322868347168,
      "learning_rate": 0.0,
      "loss": 0.7741,
      "step": 5428
    },
    {
      "epoch": 0.4872554299048645,
      "grad_norm": 6.987168312072754,
      "learning_rate": 0.0,
      "loss": 0.7433,
      "step": 5429
    },
    {
      "epoch": 0.4873451803984922,
      "grad_norm": 4.061686038970947,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 5430
    },
    {
      "epoch": 0.4874349308921199,
      "grad_norm": 3.1769754886627197,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 5431
    },
    {
      "epoch": 0.4875246813857476,
      "grad_norm": 5.106956958770752,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 5432
    },
    {
      "epoch": 0.4876144318793753,
      "grad_norm": 2.8753457069396973,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 5433
    },
    {
      "epoch": 0.48770418237300306,
      "grad_norm": 2.7935826778411865,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 5434
    },
    {
      "epoch": 0.48779393286663075,
      "grad_norm": 3.1545753479003906,
      "learning_rate": 0.0,
      "loss": 0.7687,
      "step": 5435
    },
    {
      "epoch": 0.4878836833602585,
      "grad_norm": 3.6904916763305664,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 5436
    },
    {
      "epoch": 0.4879734338538862,
      "grad_norm": 3.0613389015197754,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 5437
    },
    {
      "epoch": 0.4880631843475139,
      "grad_norm": 2.793703317642212,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 5438
    },
    {
      "epoch": 0.48815293484114164,
      "grad_norm": 2.799373149871826,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 5439
    },
    {
      "epoch": 0.48824268533476933,
      "grad_norm": 4.412750244140625,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 5440
    },
    {
      "epoch": 0.4883324358283971,
      "grad_norm": 7.251389026641846,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 5441
    },
    {
      "epoch": 0.4884221863220248,
      "grad_norm": 3.111560106277466,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 5442
    },
    {
      "epoch": 0.48851193681565247,
      "grad_norm": 2.951634168624878,
      "learning_rate": 0.0,
      "loss": 0.8056,
      "step": 5443
    },
    {
      "epoch": 0.4886016873092802,
      "grad_norm": 2.973473310470581,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 5444
    },
    {
      "epoch": 0.4886914378029079,
      "grad_norm": 3.2283756732940674,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 5445
    },
    {
      "epoch": 0.4887811882965356,
      "grad_norm": 4.842960834503174,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 5446
    },
    {
      "epoch": 0.48887093879016336,
      "grad_norm": 3.067254066467285,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 5447
    },
    {
      "epoch": 0.48896068928379105,
      "grad_norm": 3.3611397743225098,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 5448
    },
    {
      "epoch": 0.4890504397774188,
      "grad_norm": 3.616403818130493,
      "learning_rate": 0.0,
      "loss": 0.6679,
      "step": 5449
    },
    {
      "epoch": 0.4891401902710465,
      "grad_norm": 4.291398525238037,
      "learning_rate": 0.0,
      "loss": 0.6806,
      "step": 5450
    },
    {
      "epoch": 0.4892299407646742,
      "grad_norm": 3.4640867710113525,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 5451
    },
    {
      "epoch": 0.48931969125830194,
      "grad_norm": 3.147475481033325,
      "learning_rate": 0.0,
      "loss": 0.789,
      "step": 5452
    },
    {
      "epoch": 0.48940944175192963,
      "grad_norm": 3.276567220687866,
      "learning_rate": 0.0,
      "loss": 0.6826,
      "step": 5453
    },
    {
      "epoch": 0.4894991922455573,
      "grad_norm": 3.8483355045318604,
      "learning_rate": 0.0,
      "loss": 0.6728,
      "step": 5454
    },
    {
      "epoch": 0.4895889427391851,
      "grad_norm": 3.3983092308044434,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 5455
    },
    {
      "epoch": 0.48967869323281277,
      "grad_norm": 3.057630777359009,
      "learning_rate": 0.0,
      "loss": 0.7594,
      "step": 5456
    },
    {
      "epoch": 0.4897684437264405,
      "grad_norm": 3.3482155799865723,
      "learning_rate": 0.0,
      "loss": 0.7724,
      "step": 5457
    },
    {
      "epoch": 0.4898581942200682,
      "grad_norm": 5.362768650054932,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 5458
    },
    {
      "epoch": 0.4899479447136959,
      "grad_norm": 3.289989471435547,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 5459
    },
    {
      "epoch": 0.49003769520732365,
      "grad_norm": 2.2494394779205322,
      "learning_rate": 0.0,
      "loss": 0.7533,
      "step": 5460
    },
    {
      "epoch": 0.49012744570095135,
      "grad_norm": 2.155620813369751,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 5461
    },
    {
      "epoch": 0.4902171961945791,
      "grad_norm": 3.377568006515503,
      "learning_rate": 0.0,
      "loss": 0.7731,
      "step": 5462
    },
    {
      "epoch": 0.4903069466882068,
      "grad_norm": 3.2278175354003906,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 5463
    },
    {
      "epoch": 0.4903966971818345,
      "grad_norm": 2.6173176765441895,
      "learning_rate": 0.0,
      "loss": 0.7916,
      "step": 5464
    },
    {
      "epoch": 0.49048644767546223,
      "grad_norm": 3.7326080799102783,
      "learning_rate": 0.0,
      "loss": 0.6801,
      "step": 5465
    },
    {
      "epoch": 0.4905761981690899,
      "grad_norm": 4.08545446395874,
      "learning_rate": 0.0,
      "loss": 0.6755,
      "step": 5466
    },
    {
      "epoch": 0.4906659486627176,
      "grad_norm": 3.932448387145996,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 5467
    },
    {
      "epoch": 0.49075569915634537,
      "grad_norm": 2.292006254196167,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 5468
    },
    {
      "epoch": 0.49084544964997306,
      "grad_norm": 4.4998345375061035,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 5469
    },
    {
      "epoch": 0.4909352001436008,
      "grad_norm": 3.724824905395508,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 5470
    },
    {
      "epoch": 0.4910249506372285,
      "grad_norm": 2.8912556171417236,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 5471
    },
    {
      "epoch": 0.4911147011308562,
      "grad_norm": 2.1787054538726807,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 5472
    },
    {
      "epoch": 0.49120445162448395,
      "grad_norm": 2.712634801864624,
      "learning_rate": 0.0,
      "loss": 0.7119,
      "step": 5473
    },
    {
      "epoch": 0.49129420211811164,
      "grad_norm": 3.0338451862335205,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 5474
    },
    {
      "epoch": 0.49138395261173934,
      "grad_norm": 2.8975188732147217,
      "learning_rate": 0.0,
      "loss": 0.7954,
      "step": 5475
    },
    {
      "epoch": 0.4914737031053671,
      "grad_norm": 2.8024041652679443,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 5476
    },
    {
      "epoch": 0.4915634535989948,
      "grad_norm": 2.933929443359375,
      "learning_rate": 0.0,
      "loss": 0.7047,
      "step": 5477
    },
    {
      "epoch": 0.49165320409262253,
      "grad_norm": 3.193756580352783,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 5478
    },
    {
      "epoch": 0.4917429545862502,
      "grad_norm": 9.705582618713379,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 5479
    },
    {
      "epoch": 0.4918327050798779,
      "grad_norm": 3.5417704582214355,
      "learning_rate": 0.0,
      "loss": 0.6815,
      "step": 5480
    },
    {
      "epoch": 0.49192245557350567,
      "grad_norm": 2.435805320739746,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 5481
    },
    {
      "epoch": 0.49201220606713336,
      "grad_norm": 2.2821290493011475,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 5482
    },
    {
      "epoch": 0.4921019565607611,
      "grad_norm": 2.952878952026367,
      "learning_rate": 0.0,
      "loss": 0.6698,
      "step": 5483
    },
    {
      "epoch": 0.4921917070543888,
      "grad_norm": 11.298006057739258,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 5484
    },
    {
      "epoch": 0.4922814575480165,
      "grad_norm": 2.7810068130493164,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 5485
    },
    {
      "epoch": 0.49237120804164425,
      "grad_norm": 2.4718687534332275,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 5486
    },
    {
      "epoch": 0.49246095853527194,
      "grad_norm": 2.824737310409546,
      "learning_rate": 0.0,
      "loss": 0.7357,
      "step": 5487
    },
    {
      "epoch": 0.49255070902889964,
      "grad_norm": 2.420851945877075,
      "learning_rate": 0.0,
      "loss": 0.7019,
      "step": 5488
    },
    {
      "epoch": 0.4926404595225274,
      "grad_norm": 2.68345308303833,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 5489
    },
    {
      "epoch": 0.4927302100161551,
      "grad_norm": 2.9994101524353027,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 5490
    },
    {
      "epoch": 0.49281996050978283,
      "grad_norm": 2.4974400997161865,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 5491
    },
    {
      "epoch": 0.4929097110034105,
      "grad_norm": 2.885051965713501,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 5492
    },
    {
      "epoch": 0.4929994614970382,
      "grad_norm": 3.090998411178589,
      "learning_rate": 0.0,
      "loss": 0.7486,
      "step": 5493
    },
    {
      "epoch": 0.49308921199066597,
      "grad_norm": 2.251006603240967,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 5494
    },
    {
      "epoch": 0.49317896248429366,
      "grad_norm": 2.74462890625,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 5495
    },
    {
      "epoch": 0.49326871297792135,
      "grad_norm": 2.641993522644043,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 5496
    },
    {
      "epoch": 0.4933584634715491,
      "grad_norm": 4.205852508544922,
      "learning_rate": 0.0,
      "loss": 0.752,
      "step": 5497
    },
    {
      "epoch": 0.4934482139651768,
      "grad_norm": 6.971463680267334,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 5498
    },
    {
      "epoch": 0.49353796445880455,
      "grad_norm": 3.692344903945923,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 5499
    },
    {
      "epoch": 0.49362771495243224,
      "grad_norm": 2.8084723949432373,
      "learning_rate": 0.0,
      "loss": 0.7369,
      "step": 5500
    },
    {
      "epoch": 0.49371746544605993,
      "grad_norm": 3.592193603515625,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 5501
    },
    {
      "epoch": 0.4938072159396877,
      "grad_norm": 2.5385537147521973,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 5502
    },
    {
      "epoch": 0.4938969664333154,
      "grad_norm": 4.435826778411865,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 5503
    },
    {
      "epoch": 0.4939867169269431,
      "grad_norm": 3.2026498317718506,
      "learning_rate": 0.0,
      "loss": 0.6656,
      "step": 5504
    },
    {
      "epoch": 0.4940764674205708,
      "grad_norm": 2.534742832183838,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 5505
    },
    {
      "epoch": 0.4941662179141985,
      "grad_norm": 10.694374084472656,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 5506
    },
    {
      "epoch": 0.49425596840782626,
      "grad_norm": 3.215198278427124,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 5507
    },
    {
      "epoch": 0.49434571890145396,
      "grad_norm": 3.482814311981201,
      "learning_rate": 0.0,
      "loss": 0.7369,
      "step": 5508
    },
    {
      "epoch": 0.49443546939508165,
      "grad_norm": 3.4249229431152344,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 5509
    },
    {
      "epoch": 0.4945252198887094,
      "grad_norm": 4.2708740234375,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 5510
    },
    {
      "epoch": 0.4946149703823371,
      "grad_norm": 2.727522134780884,
      "learning_rate": 0.0,
      "loss": 0.7756,
      "step": 5511
    },
    {
      "epoch": 0.49470472087596484,
      "grad_norm": 2.637437582015991,
      "learning_rate": 0.0,
      "loss": 0.7523,
      "step": 5512
    },
    {
      "epoch": 0.49479447136959254,
      "grad_norm": 3.0473275184631348,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 5513
    },
    {
      "epoch": 0.49488422186322023,
      "grad_norm": 3.243195056915283,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 5514
    },
    {
      "epoch": 0.494973972356848,
      "grad_norm": 3.266842842102051,
      "learning_rate": 0.0,
      "loss": 0.7517,
      "step": 5515
    },
    {
      "epoch": 0.4950637228504757,
      "grad_norm": 3.5020346641540527,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 5516
    },
    {
      "epoch": 0.49515347334410337,
      "grad_norm": 3.666775941848755,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 5517
    },
    {
      "epoch": 0.4952432238377311,
      "grad_norm": 3.841749668121338,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 5518
    },
    {
      "epoch": 0.4953329743313588,
      "grad_norm": 3.351827383041382,
      "learning_rate": 0.0,
      "loss": 0.7056,
      "step": 5519
    },
    {
      "epoch": 0.49542272482498656,
      "grad_norm": 11.902767181396484,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 5520
    },
    {
      "epoch": 0.49551247531861425,
      "grad_norm": 5.9481282234191895,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 5521
    },
    {
      "epoch": 0.49560222581224195,
      "grad_norm": 4.2834343910217285,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 5522
    },
    {
      "epoch": 0.4956919763058697,
      "grad_norm": 3.1627283096313477,
      "learning_rate": 0.0,
      "loss": 0.7813,
      "step": 5523
    },
    {
      "epoch": 0.4957817267994974,
      "grad_norm": 2.679586172103882,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 5524
    },
    {
      "epoch": 0.4958714772931251,
      "grad_norm": 3.024726152420044,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 5525
    },
    {
      "epoch": 0.49596122778675283,
      "grad_norm": 3.277355909347534,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 5526
    },
    {
      "epoch": 0.49605097828038053,
      "grad_norm": 4.353637218475342,
      "learning_rate": 0.0,
      "loss": 0.6709,
      "step": 5527
    },
    {
      "epoch": 0.4961407287740083,
      "grad_norm": 2.9635696411132812,
      "learning_rate": 0.0,
      "loss": 0.6687,
      "step": 5528
    },
    {
      "epoch": 0.49623047926763597,
      "grad_norm": 2.6927075386047363,
      "learning_rate": 0.0,
      "loss": 0.684,
      "step": 5529
    },
    {
      "epoch": 0.49632022976126366,
      "grad_norm": 4.548245906829834,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 5530
    },
    {
      "epoch": 0.4964099802548914,
      "grad_norm": 4.567256927490234,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 5531
    },
    {
      "epoch": 0.4964997307485191,
      "grad_norm": 3.81915545463562,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 5532
    },
    {
      "epoch": 0.49658948124214686,
      "grad_norm": 2.4700894355773926,
      "learning_rate": 0.0,
      "loss": 0.7919,
      "step": 5533
    },
    {
      "epoch": 0.49667923173577455,
      "grad_norm": 2.9991683959960938,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 5534
    },
    {
      "epoch": 0.49676898222940225,
      "grad_norm": 4.486911773681641,
      "learning_rate": 0.0,
      "loss": 0.7848,
      "step": 5535
    },
    {
      "epoch": 0.49685873272303,
      "grad_norm": 3.6127769947052,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 5536
    },
    {
      "epoch": 0.4969484832166577,
      "grad_norm": 2.974031686782837,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 5537
    },
    {
      "epoch": 0.4970382337102854,
      "grad_norm": 2.2202060222625732,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 5538
    },
    {
      "epoch": 0.49712798420391313,
      "grad_norm": 2.280447244644165,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 5539
    },
    {
      "epoch": 0.4972177346975408,
      "grad_norm": 3.327054738998413,
      "learning_rate": 0.0,
      "loss": 0.712,
      "step": 5540
    },
    {
      "epoch": 0.4973074851911686,
      "grad_norm": 2.3061397075653076,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 5541
    },
    {
      "epoch": 0.49739723568479627,
      "grad_norm": 4.572428226470947,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 5542
    },
    {
      "epoch": 0.49748698617842396,
      "grad_norm": 2.6886987686157227,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 5543
    },
    {
      "epoch": 0.4975767366720517,
      "grad_norm": 2.8561885356903076,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 5544
    },
    {
      "epoch": 0.4976664871656794,
      "grad_norm": 3.1338858604431152,
      "learning_rate": 0.0,
      "loss": 0.7621,
      "step": 5545
    },
    {
      "epoch": 0.4977562376593071,
      "grad_norm": 2.466776132583618,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 5546
    },
    {
      "epoch": 0.49784598815293485,
      "grad_norm": 2.576611280441284,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 5547
    },
    {
      "epoch": 0.49793573864656254,
      "grad_norm": 2.5248467922210693,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 5548
    },
    {
      "epoch": 0.4980254891401903,
      "grad_norm": 3.1016645431518555,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 5549
    },
    {
      "epoch": 0.498115239633818,
      "grad_norm": 2.329049825668335,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 5550
    },
    {
      "epoch": 0.4982049901274457,
      "grad_norm": 4.010648250579834,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 5551
    },
    {
      "epoch": 0.49829474062107343,
      "grad_norm": 2.9679744243621826,
      "learning_rate": 0.0,
      "loss": 0.712,
      "step": 5552
    },
    {
      "epoch": 0.4983844911147011,
      "grad_norm": 4.356034278869629,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 5553
    },
    {
      "epoch": 0.49847424160832887,
      "grad_norm": 2.9949240684509277,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 5554
    },
    {
      "epoch": 0.49856399210195657,
      "grad_norm": 3.111196994781494,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 5555
    },
    {
      "epoch": 0.49865374259558426,
      "grad_norm": 2.7806284427642822,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 5556
    },
    {
      "epoch": 0.498743493089212,
      "grad_norm": 3.528890609741211,
      "learning_rate": 0.0,
      "loss": 0.7729,
      "step": 5557
    },
    {
      "epoch": 0.4988332435828397,
      "grad_norm": 2.9359545707702637,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 5558
    },
    {
      "epoch": 0.4989229940764674,
      "grad_norm": 2.89656925201416,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 5559
    },
    {
      "epoch": 0.49901274457009515,
      "grad_norm": 6.290491580963135,
      "learning_rate": 0.0,
      "loss": 0.7905,
      "step": 5560
    },
    {
      "epoch": 0.49910249506372284,
      "grad_norm": 5.342245578765869,
      "learning_rate": 0.0,
      "loss": 0.719,
      "step": 5561
    },
    {
      "epoch": 0.4991922455573506,
      "grad_norm": 2.8559725284576416,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 5562
    },
    {
      "epoch": 0.4992819960509783,
      "grad_norm": 3.4973485469818115,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 5563
    },
    {
      "epoch": 0.499371746544606,
      "grad_norm": 2.699908971786499,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 5564
    },
    {
      "epoch": 0.4994614970382337,
      "grad_norm": 3.1570026874542236,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 5565
    },
    {
      "epoch": 0.4995512475318614,
      "grad_norm": 3.2392361164093018,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 5566
    },
    {
      "epoch": 0.4996409980254891,
      "grad_norm": 2.6923539638519287,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 5567
    },
    {
      "epoch": 0.49973074851911686,
      "grad_norm": 3.4526784420013428,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 5568
    },
    {
      "epoch": 0.49982049901274456,
      "grad_norm": 16.245031356811523,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 5569
    },
    {
      "epoch": 0.4999102495063723,
      "grad_norm": 3.181816816329956,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 5570
    },
    {
      "epoch": 0.5,
      "grad_norm": 3.6029858589172363,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 5571
    },
    {
      "epoch": 0.5000897504936277,
      "grad_norm": 2.9286115169525146,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 5572
    },
    {
      "epoch": 0.5001795009872554,
      "grad_norm": 4.234097957611084,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 5573
    },
    {
      "epoch": 0.5002692514808832,
      "grad_norm": 3.3417551517486572,
      "learning_rate": 0.0,
      "loss": 0.7547,
      "step": 5574
    },
    {
      "epoch": 0.5003590019745109,
      "grad_norm": 2.960099697113037,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 5575
    },
    {
      "epoch": 0.5004487524681386,
      "grad_norm": 3.0281546115875244,
      "learning_rate": 0.0,
      "loss": 0.659,
      "step": 5576
    },
    {
      "epoch": 0.5005385029617663,
      "grad_norm": 10.048186302185059,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 5577
    },
    {
      "epoch": 0.500628253455394,
      "grad_norm": 4.712998867034912,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 5578
    },
    {
      "epoch": 0.5007180039490218,
      "grad_norm": 3.5657308101654053,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 5579
    },
    {
      "epoch": 0.5008077544426495,
      "grad_norm": 3.0276145935058594,
      "learning_rate": 0.0,
      "loss": 0.7788,
      "step": 5580
    },
    {
      "epoch": 0.5008975049362772,
      "grad_norm": 2.708142042160034,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 5581
    },
    {
      "epoch": 0.5009872554299049,
      "grad_norm": 3.014324903488159,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 5582
    },
    {
      "epoch": 0.5010770059235325,
      "grad_norm": 3.5369842052459717,
      "learning_rate": 0.0,
      "loss": 0.6715,
      "step": 5583
    },
    {
      "epoch": 0.5011667564171602,
      "grad_norm": 3.415623903274536,
      "learning_rate": 0.0,
      "loss": 0.7249,
      "step": 5584
    },
    {
      "epoch": 0.501256506910788,
      "grad_norm": 3.2185590267181396,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 5585
    },
    {
      "epoch": 0.5013462574044157,
      "grad_norm": 5.60941219329834,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 5586
    },
    {
      "epoch": 0.5014360078980434,
      "grad_norm": 3.3726468086242676,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 5587
    },
    {
      "epoch": 0.5015257583916711,
      "grad_norm": 3.799530029296875,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 5588
    },
    {
      "epoch": 0.5016155088852988,
      "grad_norm": 4.717813968658447,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 5589
    },
    {
      "epoch": 0.5017052593789266,
      "grad_norm": 3.1303577423095703,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 5590
    },
    {
      "epoch": 0.5017950098725543,
      "grad_norm": 2.3933286666870117,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 5591
    },
    {
      "epoch": 0.501884760366182,
      "grad_norm": 2.3264951705932617,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 5592
    },
    {
      "epoch": 0.5019745108598097,
      "grad_norm": 2.2316319942474365,
      "learning_rate": 0.0,
      "loss": 0.8013,
      "step": 5593
    },
    {
      "epoch": 0.5020642613534374,
      "grad_norm": 2.5156898498535156,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 5594
    },
    {
      "epoch": 0.5021540118470652,
      "grad_norm": 2.864474296569824,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 5595
    },
    {
      "epoch": 0.5022437623406929,
      "grad_norm": 2.781437397003174,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 5596
    },
    {
      "epoch": 0.5023335128343206,
      "grad_norm": 3.6491243839263916,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 5597
    },
    {
      "epoch": 0.5024232633279483,
      "grad_norm": 3.3548660278320312,
      "learning_rate": 0.0,
      "loss": 0.7355,
      "step": 5598
    },
    {
      "epoch": 0.502513013821576,
      "grad_norm": 2.859339714050293,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 5599
    },
    {
      "epoch": 0.5026027643152038,
      "grad_norm": 3.219099760055542,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 5600
    },
    {
      "epoch": 0.5026925148088315,
      "grad_norm": 2.813981294631958,
      "learning_rate": 0.0,
      "loss": 0.6676,
      "step": 5601
    },
    {
      "epoch": 0.5027822653024592,
      "grad_norm": 2.626330852508545,
      "learning_rate": 0.0,
      "loss": 0.7103,
      "step": 5602
    },
    {
      "epoch": 0.5028720157960869,
      "grad_norm": 3.220329523086548,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 5603
    },
    {
      "epoch": 0.5029617662897146,
      "grad_norm": 2.730215072631836,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 5604
    },
    {
      "epoch": 0.5030515167833423,
      "grad_norm": 2.6512351036071777,
      "learning_rate": 0.0,
      "loss": 0.7661,
      "step": 5605
    },
    {
      "epoch": 0.5031412672769701,
      "grad_norm": 2.8884763717651367,
      "learning_rate": 0.0,
      "loss": 0.701,
      "step": 5606
    },
    {
      "epoch": 0.5032310177705978,
      "grad_norm": 2.8264894485473633,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 5607
    },
    {
      "epoch": 0.5033207682642254,
      "grad_norm": 2.05538272857666,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 5608
    },
    {
      "epoch": 0.5034105187578531,
      "grad_norm": 5.565760612487793,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 5609
    },
    {
      "epoch": 0.5035002692514808,
      "grad_norm": 3.6176259517669678,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 5610
    },
    {
      "epoch": 0.5035900197451086,
      "grad_norm": 2.944798469543457,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 5611
    },
    {
      "epoch": 0.5036797702387363,
      "grad_norm": 6.840074062347412,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 5612
    },
    {
      "epoch": 0.503769520732364,
      "grad_norm": 2.902770519256592,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 5613
    },
    {
      "epoch": 0.5038592712259917,
      "grad_norm": 4.3233160972595215,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 5614
    },
    {
      "epoch": 0.5039490217196194,
      "grad_norm": 3.1700456142425537,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 5615
    },
    {
      "epoch": 0.5040387722132472,
      "grad_norm": 2.513490676879883,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 5616
    },
    {
      "epoch": 0.5041285227068749,
      "grad_norm": 3.107250213623047,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 5617
    },
    {
      "epoch": 0.5042182732005026,
      "grad_norm": 2.244008779525757,
      "learning_rate": 0.0,
      "loss": 0.6563,
      "step": 5618
    },
    {
      "epoch": 0.5043080236941303,
      "grad_norm": 2.5478663444519043,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 5619
    },
    {
      "epoch": 0.504397774187758,
      "grad_norm": 3.014935255050659,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 5620
    },
    {
      "epoch": 0.5044875246813858,
      "grad_norm": 3.925893783569336,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 5621
    },
    {
      "epoch": 0.5045772751750135,
      "grad_norm": 2.4783875942230225,
      "learning_rate": 0.0,
      "loss": 0.7103,
      "step": 5622
    },
    {
      "epoch": 0.5046670256686412,
      "grad_norm": 4.301263809204102,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 5623
    },
    {
      "epoch": 0.5047567761622689,
      "grad_norm": 3.076084613800049,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 5624
    },
    {
      "epoch": 0.5048465266558966,
      "grad_norm": 3.710777759552002,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 5625
    },
    {
      "epoch": 0.5049362771495243,
      "grad_norm": 3.094172477722168,
      "learning_rate": 0.0,
      "loss": 0.7062,
      "step": 5626
    },
    {
      "epoch": 0.5050260276431521,
      "grad_norm": 3.5331811904907227,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 5627
    },
    {
      "epoch": 0.5051157781367798,
      "grad_norm": 3.2356350421905518,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 5628
    },
    {
      "epoch": 0.5052055286304075,
      "grad_norm": 2.9598751068115234,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 5629
    },
    {
      "epoch": 0.5052952791240352,
      "grad_norm": 2.2113802433013916,
      "learning_rate": 0.0,
      "loss": 0.7824,
      "step": 5630
    },
    {
      "epoch": 0.5053850296176629,
      "grad_norm": 4.639435291290283,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 5631
    },
    {
      "epoch": 0.5054747801112907,
      "grad_norm": 5.170487403869629,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 5632
    },
    {
      "epoch": 0.5055645306049183,
      "grad_norm": 3.7018120288848877,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 5633
    },
    {
      "epoch": 0.505654281098546,
      "grad_norm": 3.108022928237915,
      "learning_rate": 0.0,
      "loss": 0.7249,
      "step": 5634
    },
    {
      "epoch": 0.5057440315921737,
      "grad_norm": 2.8837509155273438,
      "learning_rate": 0.0,
      "loss": 0.7697,
      "step": 5635
    },
    {
      "epoch": 0.5058337820858014,
      "grad_norm": 3.1315391063690186,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 5636
    },
    {
      "epoch": 0.5059235325794292,
      "grad_norm": 2.759686231613159,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 5637
    },
    {
      "epoch": 0.5060132830730569,
      "grad_norm": 2.533491611480713,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 5638
    },
    {
      "epoch": 0.5061030335666846,
      "grad_norm": 2.606847047805786,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 5639
    },
    {
      "epoch": 0.5061927840603123,
      "grad_norm": 2.6177706718444824,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 5640
    },
    {
      "epoch": 0.50628253455394,
      "grad_norm": 2.5253777503967285,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 5641
    },
    {
      "epoch": 0.5063722850475678,
      "grad_norm": 2.448190927505493,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 5642
    },
    {
      "epoch": 0.5064620355411955,
      "grad_norm": 2.217331886291504,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 5643
    },
    {
      "epoch": 0.5065517860348232,
      "grad_norm": 3.2444748878479004,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 5644
    },
    {
      "epoch": 0.5066415365284509,
      "grad_norm": 4.445455551147461,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 5645
    },
    {
      "epoch": 0.5067312870220786,
      "grad_norm": 2.846240997314453,
      "learning_rate": 0.0,
      "loss": 0.7524,
      "step": 5646
    },
    {
      "epoch": 0.5068210375157063,
      "grad_norm": 4.525406837463379,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 5647
    },
    {
      "epoch": 0.5069107880093341,
      "grad_norm": 12.764179229736328,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 5648
    },
    {
      "epoch": 0.5070005385029618,
      "grad_norm": 5.274633407592773,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 5649
    },
    {
      "epoch": 0.5070902889965895,
      "grad_norm": 3.8940696716308594,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 5650
    },
    {
      "epoch": 0.5071800394902172,
      "grad_norm": 3.386969804763794,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 5651
    },
    {
      "epoch": 0.5072697899838449,
      "grad_norm": 3.7504944801330566,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 5652
    },
    {
      "epoch": 0.5073595404774727,
      "grad_norm": 2.4955544471740723,
      "learning_rate": 0.0,
      "loss": 0.8098,
      "step": 5653
    },
    {
      "epoch": 0.5074492909711004,
      "grad_norm": 3.344463348388672,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 5654
    },
    {
      "epoch": 0.5075390414647281,
      "grad_norm": 5.881726264953613,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 5655
    },
    {
      "epoch": 0.5076287919583558,
      "grad_norm": 4.965535640716553,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 5656
    },
    {
      "epoch": 0.5077185424519834,
      "grad_norm": 2.3348228931427,
      "learning_rate": 0.0,
      "loss": 0.6597,
      "step": 5657
    },
    {
      "epoch": 0.5078082929456113,
      "grad_norm": 2.4837119579315186,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 5658
    },
    {
      "epoch": 0.5078980434392389,
      "grad_norm": 2.9551331996917725,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 5659
    },
    {
      "epoch": 0.5079877939328666,
      "grad_norm": 2.584817409515381,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 5660
    },
    {
      "epoch": 0.5080775444264943,
      "grad_norm": 4.5498504638671875,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 5661
    },
    {
      "epoch": 0.508167294920122,
      "grad_norm": 2.802919864654541,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 5662
    },
    {
      "epoch": 0.5082570454137498,
      "grad_norm": 2.910038948059082,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 5663
    },
    {
      "epoch": 0.5083467959073775,
      "grad_norm": 3.3885936737060547,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 5664
    },
    {
      "epoch": 0.5084365464010052,
      "grad_norm": 4.8903326988220215,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 5665
    },
    {
      "epoch": 0.5085262968946329,
      "grad_norm": 2.969261884689331,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 5666
    },
    {
      "epoch": 0.5086160473882606,
      "grad_norm": 2.78764009475708,
      "learning_rate": 0.0,
      "loss": 0.7734,
      "step": 5667
    },
    {
      "epoch": 0.5087057978818883,
      "grad_norm": 3.5696699619293213,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 5668
    },
    {
      "epoch": 0.5087955483755161,
      "grad_norm": 3.7876179218292236,
      "learning_rate": 0.0,
      "loss": 0.7296,
      "step": 5669
    },
    {
      "epoch": 0.5088852988691438,
      "grad_norm": 2.9754796028137207,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 5670
    },
    {
      "epoch": 0.5089750493627715,
      "grad_norm": 2.253588914871216,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 5671
    },
    {
      "epoch": 0.5090647998563992,
      "grad_norm": 7.173533916473389,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 5672
    },
    {
      "epoch": 0.5091545503500269,
      "grad_norm": 10.923269271850586,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 5673
    },
    {
      "epoch": 0.5092443008436547,
      "grad_norm": 5.400153636932373,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 5674
    },
    {
      "epoch": 0.5093340513372824,
      "grad_norm": 3.130808115005493,
      "learning_rate": 0.0,
      "loss": 0.6698,
      "step": 5675
    },
    {
      "epoch": 0.5094238018309101,
      "grad_norm": 2.4803762435913086,
      "learning_rate": 0.0,
      "loss": 0.7322,
      "step": 5676
    },
    {
      "epoch": 0.5095135523245378,
      "grad_norm": 3.2706148624420166,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 5677
    },
    {
      "epoch": 0.5096033028181655,
      "grad_norm": 5.136148452758789,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 5678
    },
    {
      "epoch": 0.5096930533117933,
      "grad_norm": 2.558222770690918,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 5679
    },
    {
      "epoch": 0.509782803805421,
      "grad_norm": 3.902467966079712,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 5680
    },
    {
      "epoch": 0.5098725542990487,
      "grad_norm": 3.7149713039398193,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 5681
    },
    {
      "epoch": 0.5099623047926763,
      "grad_norm": 2.8320975303649902,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 5682
    },
    {
      "epoch": 0.510052055286304,
      "grad_norm": 3.3577730655670166,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 5683
    },
    {
      "epoch": 0.5101418057799318,
      "grad_norm": 2.205932855606079,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 5684
    },
    {
      "epoch": 0.5102315562735595,
      "grad_norm": 2.466715097427368,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 5685
    },
    {
      "epoch": 0.5103213067671872,
      "grad_norm": 3.580178737640381,
      "learning_rate": 0.0,
      "loss": 0.7047,
      "step": 5686
    },
    {
      "epoch": 0.5104110572608149,
      "grad_norm": 3.7798688411712646,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 5687
    },
    {
      "epoch": 0.5105008077544426,
      "grad_norm": 4.443334579467773,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 5688
    },
    {
      "epoch": 0.5105905582480703,
      "grad_norm": 3.3860559463500977,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 5689
    },
    {
      "epoch": 0.5106803087416981,
      "grad_norm": 4.870700836181641,
      "learning_rate": 0.0,
      "loss": 0.7699,
      "step": 5690
    },
    {
      "epoch": 0.5107700592353258,
      "grad_norm": 2.1440258026123047,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 5691
    },
    {
      "epoch": 0.5108598097289535,
      "grad_norm": 3.134065866470337,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 5692
    },
    {
      "epoch": 0.5109495602225812,
      "grad_norm": 2.1542727947235107,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 5693
    },
    {
      "epoch": 0.5110393107162089,
      "grad_norm": 2.1298787593841553,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 5694
    },
    {
      "epoch": 0.5111290612098367,
      "grad_norm": 3.0584933757781982,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 5695
    },
    {
      "epoch": 0.5112188117034644,
      "grad_norm": 2.064055919647217,
      "learning_rate": 0.0,
      "loss": 0.7587,
      "step": 5696
    },
    {
      "epoch": 0.5113085621970921,
      "grad_norm": 1.9492255449295044,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 5697
    },
    {
      "epoch": 0.5113983126907198,
      "grad_norm": 2.1714491844177246,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 5698
    },
    {
      "epoch": 0.5114880631843475,
      "grad_norm": 3.0714547634124756,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 5699
    },
    {
      "epoch": 0.5115778136779753,
      "grad_norm": 2.5405080318450928,
      "learning_rate": 0.0,
      "loss": 0.746,
      "step": 5700
    },
    {
      "epoch": 0.511667564171603,
      "grad_norm": 2.568223714828491,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 5701
    },
    {
      "epoch": 0.5117573146652307,
      "grad_norm": 2.2400805950164795,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 5702
    },
    {
      "epoch": 0.5118470651588584,
      "grad_norm": 2.767636775970459,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 5703
    },
    {
      "epoch": 0.511936815652486,
      "grad_norm": 2.6079659461975098,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 5704
    },
    {
      "epoch": 0.5120265661461137,
      "grad_norm": 2.2551522254943848,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 5705
    },
    {
      "epoch": 0.5121163166397416,
      "grad_norm": 3.4885475635528564,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 5706
    },
    {
      "epoch": 0.5122060671333692,
      "grad_norm": 2.744973659515381,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 5707
    },
    {
      "epoch": 0.5122958176269969,
      "grad_norm": 3.6199638843536377,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 5708
    },
    {
      "epoch": 0.5123855681206246,
      "grad_norm": 2.643249750137329,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 5709
    },
    {
      "epoch": 0.5124753186142523,
      "grad_norm": 4.557456016540527,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 5710
    },
    {
      "epoch": 0.5125650691078801,
      "grad_norm": 2.520902156829834,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 5711
    },
    {
      "epoch": 0.5126548196015078,
      "grad_norm": 2.829089879989624,
      "learning_rate": 0.0,
      "loss": 0.8227,
      "step": 5712
    },
    {
      "epoch": 0.5127445700951355,
      "grad_norm": 9.44403076171875,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 5713
    },
    {
      "epoch": 0.5128343205887632,
      "grad_norm": 2.867966890335083,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 5714
    },
    {
      "epoch": 0.5129240710823909,
      "grad_norm": 3.208965301513672,
      "learning_rate": 0.0,
      "loss": 0.7768,
      "step": 5715
    },
    {
      "epoch": 0.5130138215760187,
      "grad_norm": 2.548508882522583,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 5716
    },
    {
      "epoch": 0.5131035720696464,
      "grad_norm": 3.151139974594116,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 5717
    },
    {
      "epoch": 0.5131933225632741,
      "grad_norm": 2.6321346759796143,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 5718
    },
    {
      "epoch": 0.5132830730569018,
      "grad_norm": 2.244508743286133,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 5719
    },
    {
      "epoch": 0.5133728235505295,
      "grad_norm": 3.717700481414795,
      "learning_rate": 0.0,
      "loss": 0.7496,
      "step": 5720
    },
    {
      "epoch": 0.5134625740441573,
      "grad_norm": 3.0539700984954834,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 5721
    },
    {
      "epoch": 0.513552324537785,
      "grad_norm": 2.5530176162719727,
      "learning_rate": 0.0,
      "loss": 0.8135,
      "step": 5722
    },
    {
      "epoch": 0.5136420750314127,
      "grad_norm": 2.7161760330200195,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 5723
    },
    {
      "epoch": 0.5137318255250404,
      "grad_norm": 2.813805103302002,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 5724
    },
    {
      "epoch": 0.5138215760186681,
      "grad_norm": 2.5202553272247314,
      "learning_rate": 0.0,
      "loss": 0.7707,
      "step": 5725
    },
    {
      "epoch": 0.5139113265122958,
      "grad_norm": 2.642113208770752,
      "learning_rate": 0.0,
      "loss": 0.6615,
      "step": 5726
    },
    {
      "epoch": 0.5140010770059236,
      "grad_norm": 3.514594078063965,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 5727
    },
    {
      "epoch": 0.5140908274995513,
      "grad_norm": 2.365358352661133,
      "learning_rate": 0.0,
      "loss": 0.7336,
      "step": 5728
    },
    {
      "epoch": 0.514180577993179,
      "grad_norm": 2.5781803131103516,
      "learning_rate": 0.0,
      "loss": 0.7599,
      "step": 5729
    },
    {
      "epoch": 0.5142703284868066,
      "grad_norm": 4.309101581573486,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 5730
    },
    {
      "epoch": 0.5143600789804343,
      "grad_norm": 4.231478691101074,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 5731
    },
    {
      "epoch": 0.5144498294740621,
      "grad_norm": 4.440852642059326,
      "learning_rate": 0.0,
      "loss": 0.7316,
      "step": 5732
    },
    {
      "epoch": 0.5145395799676898,
      "grad_norm": 2.7253623008728027,
      "learning_rate": 0.0,
      "loss": 0.7851,
      "step": 5733
    },
    {
      "epoch": 0.5146293304613175,
      "grad_norm": 3.7843399047851562,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 5734
    },
    {
      "epoch": 0.5147190809549452,
      "grad_norm": 3.2164878845214844,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 5735
    },
    {
      "epoch": 0.5148088314485729,
      "grad_norm": 2.5760810375213623,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 5736
    },
    {
      "epoch": 0.5148985819422007,
      "grad_norm": 3.324904203414917,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 5737
    },
    {
      "epoch": 0.5149883324358284,
      "grad_norm": 2.624817132949829,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 5738
    },
    {
      "epoch": 0.5150780829294561,
      "grad_norm": 2.4665987491607666,
      "learning_rate": 0.0,
      "loss": 0.6992,
      "step": 5739
    },
    {
      "epoch": 0.5151678334230838,
      "grad_norm": 2.5432019233703613,
      "learning_rate": 0.0,
      "loss": 0.6944,
      "step": 5740
    },
    {
      "epoch": 0.5152575839167115,
      "grad_norm": 2.440197706222534,
      "learning_rate": 0.0,
      "loss": 0.6459,
      "step": 5741
    },
    {
      "epoch": 0.5153473344103393,
      "grad_norm": 2.970829725265503,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 5742
    },
    {
      "epoch": 0.515437084903967,
      "grad_norm": 2.636803150177002,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 5743
    },
    {
      "epoch": 0.5155268353975947,
      "grad_norm": 2.896658182144165,
      "learning_rate": 0.0,
      "loss": 0.7505,
      "step": 5744
    },
    {
      "epoch": 0.5156165858912224,
      "grad_norm": 7.111575603485107,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 5745
    },
    {
      "epoch": 0.5157063363848501,
      "grad_norm": 3.0169591903686523,
      "learning_rate": 0.0,
      "loss": 0.7752,
      "step": 5746
    },
    {
      "epoch": 0.5157960868784778,
      "grad_norm": 4.2411274909973145,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 5747
    },
    {
      "epoch": 0.5158858373721056,
      "grad_norm": 2.3612217903137207,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 5748
    },
    {
      "epoch": 0.5159755878657333,
      "grad_norm": 10.90182876586914,
      "learning_rate": 0.0,
      "loss": 0.6852,
      "step": 5749
    },
    {
      "epoch": 0.516065338359361,
      "grad_norm": 2.771251916885376,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 5750
    },
    {
      "epoch": 0.5161550888529887,
      "grad_norm": 4.435730934143066,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 5751
    },
    {
      "epoch": 0.5162448393466164,
      "grad_norm": 3.020397186279297,
      "learning_rate": 0.0,
      "loss": 0.6868,
      "step": 5752
    },
    {
      "epoch": 0.5163345898402442,
      "grad_norm": 2.2191085815429688,
      "learning_rate": 0.0,
      "loss": 0.8185,
      "step": 5753
    },
    {
      "epoch": 0.5164243403338719,
      "grad_norm": 2.97770619392395,
      "learning_rate": 0.0,
      "loss": 0.7648,
      "step": 5754
    },
    {
      "epoch": 0.5165140908274996,
      "grad_norm": 2.1802308559417725,
      "learning_rate": 0.0,
      "loss": 0.7784,
      "step": 5755
    },
    {
      "epoch": 0.5166038413211272,
      "grad_norm": 2.241281270980835,
      "learning_rate": 0.0,
      "loss": 0.7346,
      "step": 5756
    },
    {
      "epoch": 0.5166935918147549,
      "grad_norm": 3.636138439178467,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 5757
    },
    {
      "epoch": 0.5167833423083827,
      "grad_norm": 3.398118019104004,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 5758
    },
    {
      "epoch": 0.5168730928020104,
      "grad_norm": 2.535381555557251,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 5759
    },
    {
      "epoch": 0.5169628432956381,
      "grad_norm": 3.9943482875823975,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 5760
    },
    {
      "epoch": 0.5170525937892658,
      "grad_norm": 3.481325626373291,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 5761
    },
    {
      "epoch": 0.5171423442828935,
      "grad_norm": 3.0455307960510254,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 5762
    },
    {
      "epoch": 0.5172320947765213,
      "grad_norm": 2.5071446895599365,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 5763
    },
    {
      "epoch": 0.517321845270149,
      "grad_norm": 3.0922203063964844,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 5764
    },
    {
      "epoch": 0.5174115957637767,
      "grad_norm": 3.9676618576049805,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 5765
    },
    {
      "epoch": 0.5175013462574044,
      "grad_norm": 7.493227958679199,
      "learning_rate": 0.0,
      "loss": 0.6262,
      "step": 5766
    },
    {
      "epoch": 0.5175910967510321,
      "grad_norm": 4.354401111602783,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 5767
    },
    {
      "epoch": 0.5176808472446598,
      "grad_norm": 2.539713144302368,
      "learning_rate": 0.0,
      "loss": 0.6628,
      "step": 5768
    },
    {
      "epoch": 0.5177705977382876,
      "grad_norm": 3.3249576091766357,
      "learning_rate": 0.0,
      "loss": 0.761,
      "step": 5769
    },
    {
      "epoch": 0.5178603482319153,
      "grad_norm": 4.344575881958008,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 5770
    },
    {
      "epoch": 0.517950098725543,
      "grad_norm": 2.544890880584717,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 5771
    },
    {
      "epoch": 0.5180398492191707,
      "grad_norm": 2.2413370609283447,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 5772
    },
    {
      "epoch": 0.5181295997127984,
      "grad_norm": 3.1833250522613525,
      "learning_rate": 0.0,
      "loss": 0.6901,
      "step": 5773
    },
    {
      "epoch": 0.5182193502064262,
      "grad_norm": 2.6670966148376465,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 5774
    },
    {
      "epoch": 0.5183091007000539,
      "grad_norm": 3.277794122695923,
      "learning_rate": 0.0,
      "loss": 0.733,
      "step": 5775
    },
    {
      "epoch": 0.5183988511936816,
      "grad_norm": 2.7633888721466064,
      "learning_rate": 0.0,
      "loss": 0.7363,
      "step": 5776
    },
    {
      "epoch": 0.5184886016873093,
      "grad_norm": 2.225334405899048,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 5777
    },
    {
      "epoch": 0.518578352180937,
      "grad_norm": 3.0601391792297363,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 5778
    },
    {
      "epoch": 0.5186681026745648,
      "grad_norm": 3.8743696212768555,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 5779
    },
    {
      "epoch": 0.5187578531681925,
      "grad_norm": 2.5718493461608887,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 5780
    },
    {
      "epoch": 0.5188476036618201,
      "grad_norm": 3.3254740238189697,
      "learning_rate": 0.0,
      "loss": 0.8116,
      "step": 5781
    },
    {
      "epoch": 0.5189373541554478,
      "grad_norm": 2.3275787830352783,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 5782
    },
    {
      "epoch": 0.5190271046490755,
      "grad_norm": 2.412865400314331,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 5783
    },
    {
      "epoch": 0.5191168551427033,
      "grad_norm": 2.590259552001953,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 5784
    },
    {
      "epoch": 0.519206605636331,
      "grad_norm": 2.538238525390625,
      "learning_rate": 0.0,
      "loss": 0.664,
      "step": 5785
    },
    {
      "epoch": 0.5192963561299587,
      "grad_norm": 3.0838522911071777,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 5786
    },
    {
      "epoch": 0.5193861066235864,
      "grad_norm": 2.747746706008911,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 5787
    },
    {
      "epoch": 0.5194758571172141,
      "grad_norm": 8.555669784545898,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 5788
    },
    {
      "epoch": 0.5195656076108418,
      "grad_norm": 5.405815124511719,
      "learning_rate": 0.0,
      "loss": 0.7643,
      "step": 5789
    },
    {
      "epoch": 0.5196553581044696,
      "grad_norm": 2.927751064300537,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 5790
    },
    {
      "epoch": 0.5197451085980973,
      "grad_norm": 5.667635917663574,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 5791
    },
    {
      "epoch": 0.519834859091725,
      "grad_norm": 3.162316083908081,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 5792
    },
    {
      "epoch": 0.5199246095853527,
      "grad_norm": 3.1031761169433594,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 5793
    },
    {
      "epoch": 0.5200143600789804,
      "grad_norm": 2.558809995651245,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 5794
    },
    {
      "epoch": 0.5201041105726082,
      "grad_norm": 2.6288692951202393,
      "learning_rate": 0.0,
      "loss": 0.774,
      "step": 5795
    },
    {
      "epoch": 0.5201938610662359,
      "grad_norm": 2.413297176361084,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 5796
    },
    {
      "epoch": 0.5202836115598636,
      "grad_norm": 2.8487257957458496,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 5797
    },
    {
      "epoch": 0.5203733620534913,
      "grad_norm": 2.615196466445923,
      "learning_rate": 0.0,
      "loss": 0.701,
      "step": 5798
    },
    {
      "epoch": 0.520463112547119,
      "grad_norm": 4.077789783477783,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 5799
    },
    {
      "epoch": 0.5205528630407468,
      "grad_norm": 3.3673737049102783,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 5800
    },
    {
      "epoch": 0.5206426135343745,
      "grad_norm": 2.5258913040161133,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 5801
    },
    {
      "epoch": 0.5207323640280022,
      "grad_norm": 4.431833267211914,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 5802
    },
    {
      "epoch": 0.5208221145216299,
      "grad_norm": 2.5311617851257324,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 5803
    },
    {
      "epoch": 0.5209118650152575,
      "grad_norm": 2.441704750061035,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 5804
    },
    {
      "epoch": 0.5210016155088854,
      "grad_norm": 2.5223751068115234,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 5805
    },
    {
      "epoch": 0.521091366002513,
      "grad_norm": 2.487598419189453,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 5806
    },
    {
      "epoch": 0.5211811164961407,
      "grad_norm": 2.729417085647583,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 5807
    },
    {
      "epoch": 0.5212708669897684,
      "grad_norm": 2.0017752647399902,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 5808
    },
    {
      "epoch": 0.5213606174833961,
      "grad_norm": 2.8184640407562256,
      "learning_rate": 0.0,
      "loss": 0.7761,
      "step": 5809
    },
    {
      "epoch": 0.5214503679770238,
      "grad_norm": 2.8982436656951904,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 5810
    },
    {
      "epoch": 0.5215401184706516,
      "grad_norm": 2.9582202434539795,
      "learning_rate": 0.0,
      "loss": 0.7996,
      "step": 5811
    },
    {
      "epoch": 0.5216298689642793,
      "grad_norm": 2.997985601425171,
      "learning_rate": 0.0,
      "loss": 0.7714,
      "step": 5812
    },
    {
      "epoch": 0.521719619457907,
      "grad_norm": 2.5793914794921875,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 5813
    },
    {
      "epoch": 0.5218093699515347,
      "grad_norm": 3.5094189643859863,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 5814
    },
    {
      "epoch": 0.5218991204451624,
      "grad_norm": 3.0980491638183594,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 5815
    },
    {
      "epoch": 0.5219888709387902,
      "grad_norm": 2.920499086380005,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 5816
    },
    {
      "epoch": 0.5220786214324179,
      "grad_norm": 2.1040830612182617,
      "learning_rate": 0.0,
      "loss": 0.6809,
      "step": 5817
    },
    {
      "epoch": 0.5221683719260456,
      "grad_norm": 10.018972396850586,
      "learning_rate": 0.0,
      "loss": 0.7914,
      "step": 5818
    },
    {
      "epoch": 0.5222581224196733,
      "grad_norm": 3.3611371517181396,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 5819
    },
    {
      "epoch": 0.522347872913301,
      "grad_norm": 2.4474642276763916,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 5820
    },
    {
      "epoch": 0.5224376234069288,
      "grad_norm": 2.8389413356781006,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 5821
    },
    {
      "epoch": 0.5225273739005565,
      "grad_norm": 3.1407384872436523,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 5822
    },
    {
      "epoch": 0.5226171243941842,
      "grad_norm": 3.4750521183013916,
      "learning_rate": 0.0,
      "loss": 0.712,
      "step": 5823
    },
    {
      "epoch": 0.5227068748878119,
      "grad_norm": 3.0778937339782715,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 5824
    },
    {
      "epoch": 0.5227966253814396,
      "grad_norm": 8.243975639343262,
      "learning_rate": 0.0,
      "loss": 0.8004,
      "step": 5825
    },
    {
      "epoch": 0.5228863758750674,
      "grad_norm": 4.0863142013549805,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 5826
    },
    {
      "epoch": 0.5229761263686951,
      "grad_norm": 3.7344324588775635,
      "learning_rate": 0.0,
      "loss": 0.761,
      "step": 5827
    },
    {
      "epoch": 0.5230658768623228,
      "grad_norm": 3.054271697998047,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 5828
    },
    {
      "epoch": 0.5231556273559504,
      "grad_norm": 2.1809375286102295,
      "learning_rate": 0.0,
      "loss": 0.6545,
      "step": 5829
    },
    {
      "epoch": 0.5232453778495781,
      "grad_norm": 3.1917593479156494,
      "learning_rate": 0.0,
      "loss": 0.7301,
      "step": 5830
    },
    {
      "epoch": 0.5233351283432058,
      "grad_norm": 4.3188018798828125,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 5831
    },
    {
      "epoch": 0.5234248788368336,
      "grad_norm": 2.588695764541626,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 5832
    },
    {
      "epoch": 0.5235146293304613,
      "grad_norm": 2.087683916091919,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 5833
    },
    {
      "epoch": 0.523604379824089,
      "grad_norm": 2.621269941329956,
      "learning_rate": 0.0,
      "loss": 0.681,
      "step": 5834
    },
    {
      "epoch": 0.5236941303177167,
      "grad_norm": 4.05034875869751,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 5835
    },
    {
      "epoch": 0.5237838808113444,
      "grad_norm": 2.7906887531280518,
      "learning_rate": 0.0,
      "loss": 0.6792,
      "step": 5836
    },
    {
      "epoch": 0.5238736313049722,
      "grad_norm": 2.091574192047119,
      "learning_rate": 0.0,
      "loss": 0.6603,
      "step": 5837
    },
    {
      "epoch": 0.5239633817985999,
      "grad_norm": 2.715758800506592,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 5838
    },
    {
      "epoch": 0.5240531322922276,
      "grad_norm": 2.5659027099609375,
      "learning_rate": 0.0,
      "loss": 0.7301,
      "step": 5839
    },
    {
      "epoch": 0.5241428827858553,
      "grad_norm": 2.5323617458343506,
      "learning_rate": 0.0,
      "loss": 0.7976,
      "step": 5840
    },
    {
      "epoch": 0.524232633279483,
      "grad_norm": 2.789510726928711,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 5841
    },
    {
      "epoch": 0.5243223837731108,
      "grad_norm": 2.892838954925537,
      "learning_rate": 0.0,
      "loss": 0.7775,
      "step": 5842
    },
    {
      "epoch": 0.5244121342667385,
      "grad_norm": 2.918790340423584,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 5843
    },
    {
      "epoch": 0.5245018847603662,
      "grad_norm": 2.5426526069641113,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 5844
    },
    {
      "epoch": 0.5245916352539939,
      "grad_norm": 2.790628671646118,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 5845
    },
    {
      "epoch": 0.5246813857476216,
      "grad_norm": 3.1266350746154785,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 5846
    },
    {
      "epoch": 0.5247711362412494,
      "grad_norm": 2.8488616943359375,
      "learning_rate": 0.0,
      "loss": 0.6723,
      "step": 5847
    },
    {
      "epoch": 0.5248608867348771,
      "grad_norm": 2.7426538467407227,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 5848
    },
    {
      "epoch": 0.5249506372285048,
      "grad_norm": 2.271960735321045,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 5849
    },
    {
      "epoch": 0.5250403877221325,
      "grad_norm": 2.7690505981445312,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 5850
    },
    {
      "epoch": 0.5251301382157602,
      "grad_norm": 2.7490322589874268,
      "learning_rate": 0.0,
      "loss": 0.7711,
      "step": 5851
    },
    {
      "epoch": 0.5252198887093878,
      "grad_norm": 2.336915969848633,
      "learning_rate": 0.0,
      "loss": 0.7143,
      "step": 5852
    },
    {
      "epoch": 0.5253096392030157,
      "grad_norm": 5.251223087310791,
      "learning_rate": 0.0,
      "loss": 0.6646,
      "step": 5853
    },
    {
      "epoch": 0.5253993896966433,
      "grad_norm": 2.317131519317627,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 5854
    },
    {
      "epoch": 0.525489140190271,
      "grad_norm": 2.6732025146484375,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 5855
    },
    {
      "epoch": 0.5255788906838987,
      "grad_norm": 3.1881887912750244,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 5856
    },
    {
      "epoch": 0.5256686411775264,
      "grad_norm": 2.685276985168457,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 5857
    },
    {
      "epoch": 0.5257583916711542,
      "grad_norm": 2.2527382373809814,
      "learning_rate": 0.0,
      "loss": 0.7019,
      "step": 5858
    },
    {
      "epoch": 0.5258481421647819,
      "grad_norm": 3.826613426208496,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 5859
    },
    {
      "epoch": 0.5259378926584096,
      "grad_norm": 3.8565855026245117,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 5860
    },
    {
      "epoch": 0.5260276431520373,
      "grad_norm": 4.339319229125977,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 5861
    },
    {
      "epoch": 0.526117393645665,
      "grad_norm": 3.05204701423645,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 5862
    },
    {
      "epoch": 0.5262071441392928,
      "grad_norm": 6.637218952178955,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 5863
    },
    {
      "epoch": 0.5262968946329205,
      "grad_norm": 3.531813859939575,
      "learning_rate": 0.0,
      "loss": 0.7627,
      "step": 5864
    },
    {
      "epoch": 0.5263866451265482,
      "grad_norm": 3.91115665435791,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 5865
    },
    {
      "epoch": 0.5264763956201759,
      "grad_norm": 3.0084402561187744,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 5866
    },
    {
      "epoch": 0.5265661461138036,
      "grad_norm": 2.875014543533325,
      "learning_rate": 0.0,
      "loss": 0.771,
      "step": 5867
    },
    {
      "epoch": 0.5266558966074314,
      "grad_norm": 12.35693073272705,
      "learning_rate": 0.0,
      "loss": 0.7546,
      "step": 5868
    },
    {
      "epoch": 0.5267456471010591,
      "grad_norm": 5.179283142089844,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 5869
    },
    {
      "epoch": 0.5268353975946868,
      "grad_norm": 2.435389280319214,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 5870
    },
    {
      "epoch": 0.5269251480883145,
      "grad_norm": 2.676199197769165,
      "learning_rate": 0.0,
      "loss": 0.682,
      "step": 5871
    },
    {
      "epoch": 0.5270148985819422,
      "grad_norm": 2.3163790702819824,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 5872
    },
    {
      "epoch": 0.5271046490755699,
      "grad_norm": 2.8105556964874268,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 5873
    },
    {
      "epoch": 0.5271943995691977,
      "grad_norm": 3.3706467151641846,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 5874
    },
    {
      "epoch": 0.5272841500628254,
      "grad_norm": 5.6547722816467285,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 5875
    },
    {
      "epoch": 0.5273739005564531,
      "grad_norm": 4.912181854248047,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 5876
    },
    {
      "epoch": 0.5274636510500808,
      "grad_norm": 2.8946008682250977,
      "learning_rate": 0.0,
      "loss": 0.693,
      "step": 5877
    },
    {
      "epoch": 0.5275534015437084,
      "grad_norm": 3.4520294666290283,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 5878
    },
    {
      "epoch": 0.5276431520373362,
      "grad_norm": 1.929120421409607,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 5879
    },
    {
      "epoch": 0.5277329025309639,
      "grad_norm": 13.119569778442383,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 5880
    },
    {
      "epoch": 0.5278226530245916,
      "grad_norm": 4.577392101287842,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 5881
    },
    {
      "epoch": 0.5279124035182193,
      "grad_norm": 4.278448581695557,
      "learning_rate": 0.0,
      "loss": 0.6728,
      "step": 5882
    },
    {
      "epoch": 0.528002154011847,
      "grad_norm": 2.6034772396087646,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 5883
    },
    {
      "epoch": 0.5280919045054748,
      "grad_norm": 6.118347644805908,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 5884
    },
    {
      "epoch": 0.5281816549991025,
      "grad_norm": 5.291409492492676,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 5885
    },
    {
      "epoch": 0.5282714054927302,
      "grad_norm": 3.481011390686035,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 5886
    },
    {
      "epoch": 0.5283611559863579,
      "grad_norm": 3.623910665512085,
      "learning_rate": 0.0,
      "loss": 0.8086,
      "step": 5887
    },
    {
      "epoch": 0.5284509064799856,
      "grad_norm": 3.2844936847686768,
      "learning_rate": 0.0,
      "loss": 0.7034,
      "step": 5888
    },
    {
      "epoch": 0.5285406569736134,
      "grad_norm": 4.863206386566162,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 5889
    },
    {
      "epoch": 0.5286304074672411,
      "grad_norm": 4.758115768432617,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 5890
    },
    {
      "epoch": 0.5287201579608688,
      "grad_norm": 3.9639387130737305,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 5891
    },
    {
      "epoch": 0.5288099084544965,
      "grad_norm": 3.813624382019043,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 5892
    },
    {
      "epoch": 0.5288996589481242,
      "grad_norm": 2.8040614128112793,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 5893
    },
    {
      "epoch": 0.5289894094417519,
      "grad_norm": 3.9524195194244385,
      "learning_rate": 0.0,
      "loss": 0.7574,
      "step": 5894
    },
    {
      "epoch": 0.5290791599353797,
      "grad_norm": 2.9192659854888916,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 5895
    },
    {
      "epoch": 0.5291689104290074,
      "grad_norm": 3.3881771564483643,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 5896
    },
    {
      "epoch": 0.5292586609226351,
      "grad_norm": 3.6264705657958984,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 5897
    },
    {
      "epoch": 0.5293484114162628,
      "grad_norm": 11.028596878051758,
      "learning_rate": 0.0,
      "loss": 0.6914,
      "step": 5898
    },
    {
      "epoch": 0.5294381619098905,
      "grad_norm": 4.711150646209717,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 5899
    },
    {
      "epoch": 0.5295279124035183,
      "grad_norm": 6.660467147827148,
      "learning_rate": 0.0,
      "loss": 0.7524,
      "step": 5900
    },
    {
      "epoch": 0.529617662897146,
      "grad_norm": 2.8115410804748535,
      "learning_rate": 0.0,
      "loss": 0.7676,
      "step": 5901
    },
    {
      "epoch": 0.5297074133907737,
      "grad_norm": 2.393653154373169,
      "learning_rate": 0.0,
      "loss": 0.6709,
      "step": 5902
    },
    {
      "epoch": 0.5297971638844013,
      "grad_norm": 2.697307586669922,
      "learning_rate": 0.0,
      "loss": 0.656,
      "step": 5903
    },
    {
      "epoch": 0.529886914378029,
      "grad_norm": 3.720872640609741,
      "learning_rate": 0.0,
      "loss": 0.7596,
      "step": 5904
    },
    {
      "epoch": 0.5299766648716568,
      "grad_norm": 3.0196428298950195,
      "learning_rate": 0.0,
      "loss": 0.6859,
      "step": 5905
    },
    {
      "epoch": 0.5300664153652845,
      "grad_norm": 2.587860345840454,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 5906
    },
    {
      "epoch": 0.5301561658589122,
      "grad_norm": 2.8421452045440674,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 5907
    },
    {
      "epoch": 0.5302459163525399,
      "grad_norm": 2.8546743392944336,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 5908
    },
    {
      "epoch": 0.5303356668461676,
      "grad_norm": 2.7420156002044678,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 5909
    },
    {
      "epoch": 0.5304254173397954,
      "grad_norm": 2.423912763595581,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 5910
    },
    {
      "epoch": 0.5305151678334231,
      "grad_norm": 2.741713285446167,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 5911
    },
    {
      "epoch": 0.5306049183270508,
      "grad_norm": 2.3993465900421143,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 5912
    },
    {
      "epoch": 0.5306946688206785,
      "grad_norm": 2.664350748062134,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 5913
    },
    {
      "epoch": 0.5307844193143062,
      "grad_norm": 2.5106637477874756,
      "learning_rate": 0.0,
      "loss": 0.7381,
      "step": 5914
    },
    {
      "epoch": 0.5308741698079339,
      "grad_norm": 2.6859378814697266,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 5915
    },
    {
      "epoch": 0.5309639203015617,
      "grad_norm": 7.70554780960083,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 5916
    },
    {
      "epoch": 0.5310536707951894,
      "grad_norm": 3.8739824295043945,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 5917
    },
    {
      "epoch": 0.5311434212888171,
      "grad_norm": 2.827432632446289,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 5918
    },
    {
      "epoch": 0.5312331717824448,
      "grad_norm": 2.206404209136963,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 5919
    },
    {
      "epoch": 0.5313229222760725,
      "grad_norm": 6.688103675842285,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 5920
    },
    {
      "epoch": 0.5314126727697003,
      "grad_norm": 3.589921712875366,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 5921
    },
    {
      "epoch": 0.531502423263328,
      "grad_norm": 3.3049445152282715,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 5922
    },
    {
      "epoch": 0.5315921737569557,
      "grad_norm": 1.8522173166275024,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 5923
    },
    {
      "epoch": 0.5316819242505834,
      "grad_norm": 3.439136505126953,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 5924
    },
    {
      "epoch": 0.531771674744211,
      "grad_norm": 3.2188785076141357,
      "learning_rate": 0.0,
      "loss": 0.7447,
      "step": 5925
    },
    {
      "epoch": 0.5318614252378389,
      "grad_norm": 3.7664506435394287,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 5926
    },
    {
      "epoch": 0.5319511757314666,
      "grad_norm": 2.628183603286743,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 5927
    },
    {
      "epoch": 0.5320409262250942,
      "grad_norm": 2.9363057613372803,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 5928
    },
    {
      "epoch": 0.5321306767187219,
      "grad_norm": 3.0734353065490723,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 5929
    },
    {
      "epoch": 0.5322204272123496,
      "grad_norm": 2.936821937561035,
      "learning_rate": 0.0,
      "loss": 0.7827,
      "step": 5930
    },
    {
      "epoch": 0.5323101777059773,
      "grad_norm": 5.36340856552124,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 5931
    },
    {
      "epoch": 0.5323999281996051,
      "grad_norm": 3.3159022331237793,
      "learning_rate": 0.0,
      "loss": 0.7052,
      "step": 5932
    },
    {
      "epoch": 0.5324896786932328,
      "grad_norm": 2.4427366256713867,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "step": 5933
    },
    {
      "epoch": 0.5325794291868605,
      "grad_norm": 7.395838260650635,
      "learning_rate": 0.0,
      "loss": 0.768,
      "step": 5934
    },
    {
      "epoch": 0.5326691796804882,
      "grad_norm": 4.731066703796387,
      "learning_rate": 0.0,
      "loss": 0.6944,
      "step": 5935
    },
    {
      "epoch": 0.5327589301741159,
      "grad_norm": 4.310279369354248,
      "learning_rate": 0.0,
      "loss": 0.7369,
      "step": 5936
    },
    {
      "epoch": 0.5328486806677437,
      "grad_norm": 2.3767998218536377,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 5937
    },
    {
      "epoch": 0.5329384311613714,
      "grad_norm": 2.8604543209075928,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 5938
    },
    {
      "epoch": 0.5330281816549991,
      "grad_norm": 2.92582368850708,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 5939
    },
    {
      "epoch": 0.5331179321486268,
      "grad_norm": 2.8958840370178223,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 5940
    },
    {
      "epoch": 0.5332076826422545,
      "grad_norm": 3.1261563301086426,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 5941
    },
    {
      "epoch": 0.5332974331358823,
      "grad_norm": 3.4063656330108643,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 5942
    },
    {
      "epoch": 0.53338718362951,
      "grad_norm": 3.5692923069000244,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 5943
    },
    {
      "epoch": 0.5334769341231377,
      "grad_norm": 3.57643461227417,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 5944
    },
    {
      "epoch": 0.5335666846167654,
      "grad_norm": 2.595506191253662,
      "learning_rate": 0.0,
      "loss": 0.7823,
      "step": 5945
    },
    {
      "epoch": 0.5336564351103931,
      "grad_norm": 2.514141321182251,
      "learning_rate": 0.0,
      "loss": 0.7807,
      "step": 5946
    },
    {
      "epoch": 0.5337461856040209,
      "grad_norm": 2.338850975036621,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 5947
    },
    {
      "epoch": 0.5338359360976486,
      "grad_norm": 2.8170478343963623,
      "learning_rate": 0.0,
      "loss": 0.7447,
      "step": 5948
    },
    {
      "epoch": 0.5339256865912763,
      "grad_norm": 2.409801959991455,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 5949
    },
    {
      "epoch": 0.534015437084904,
      "grad_norm": 2.1509854793548584,
      "learning_rate": 0.0,
      "loss": 0.7401,
      "step": 5950
    },
    {
      "epoch": 0.5341051875785316,
      "grad_norm": 3.071483850479126,
      "learning_rate": 0.0,
      "loss": 0.712,
      "step": 5951
    },
    {
      "epoch": 0.5341949380721593,
      "grad_norm": 25.408634185791016,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 5952
    },
    {
      "epoch": 0.5342846885657871,
      "grad_norm": 2.7911477088928223,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 5953
    },
    {
      "epoch": 0.5343744390594148,
      "grad_norm": 3.1093454360961914,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 5954
    },
    {
      "epoch": 0.5344641895530425,
      "grad_norm": 2.7659060955047607,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 5955
    },
    {
      "epoch": 0.5345539400466702,
      "grad_norm": 3.5420124530792236,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 5956
    },
    {
      "epoch": 0.5346436905402979,
      "grad_norm": 2.4008946418762207,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 5957
    },
    {
      "epoch": 0.5347334410339257,
      "grad_norm": 2.38053297996521,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 5958
    },
    {
      "epoch": 0.5348231915275534,
      "grad_norm": 2.4322338104248047,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 5959
    },
    {
      "epoch": 0.5349129420211811,
      "grad_norm": 6.2931132316589355,
      "learning_rate": 0.0,
      "loss": 0.7794,
      "step": 5960
    },
    {
      "epoch": 0.5350026925148088,
      "grad_norm": 3.106276750564575,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 5961
    },
    {
      "epoch": 0.5350924430084365,
      "grad_norm": 2.957310914993286,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 5962
    },
    {
      "epoch": 0.5351821935020643,
      "grad_norm": 3.8779850006103516,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 5963
    },
    {
      "epoch": 0.535271943995692,
      "grad_norm": 2.659289598464966,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 5964
    },
    {
      "epoch": 0.5353616944893197,
      "grad_norm": 3.6728756427764893,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 5965
    },
    {
      "epoch": 0.5354514449829474,
      "grad_norm": 2.4838263988494873,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 5966
    },
    {
      "epoch": 0.5355411954765751,
      "grad_norm": 2.608527660369873,
      "learning_rate": 0.0,
      "loss": 0.6849,
      "step": 5967
    },
    {
      "epoch": 0.5356309459702029,
      "grad_norm": 3.6987221240997314,
      "learning_rate": 0.0,
      "loss": 0.6794,
      "step": 5968
    },
    {
      "epoch": 0.5357206964638306,
      "grad_norm": 2.8134775161743164,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 5969
    },
    {
      "epoch": 0.5358104469574583,
      "grad_norm": 2.8231515884399414,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 5970
    },
    {
      "epoch": 0.535900197451086,
      "grad_norm": 2.3947417736053467,
      "learning_rate": 0.0,
      "loss": 0.7056,
      "step": 5971
    },
    {
      "epoch": 0.5359899479447137,
      "grad_norm": 3.569627046585083,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 5972
    },
    {
      "epoch": 0.5360796984383414,
      "grad_norm": 2.7627451419830322,
      "learning_rate": 0.0,
      "loss": 0.7736,
      "step": 5973
    },
    {
      "epoch": 0.5361694489319692,
      "grad_norm": 2.537297487258911,
      "learning_rate": 0.0,
      "loss": 0.7468,
      "step": 5974
    },
    {
      "epoch": 0.5362591994255969,
      "grad_norm": 2.9270503520965576,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 5975
    },
    {
      "epoch": 0.5363489499192245,
      "grad_norm": 7.462170600891113,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 5976
    },
    {
      "epoch": 0.5364387004128522,
      "grad_norm": 3.271120071411133,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 5977
    },
    {
      "epoch": 0.5365284509064799,
      "grad_norm": 2.5905838012695312,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 5978
    },
    {
      "epoch": 0.5366182014001077,
      "grad_norm": 4.055696487426758,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 5979
    },
    {
      "epoch": 0.5367079518937354,
      "grad_norm": 4.297749996185303,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 5980
    },
    {
      "epoch": 0.5367977023873631,
      "grad_norm": 2.68928861618042,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 5981
    },
    {
      "epoch": 0.5368874528809908,
      "grad_norm": 2.2829065322875977,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 5982
    },
    {
      "epoch": 0.5369772033746185,
      "grad_norm": 3.386813163757324,
      "learning_rate": 0.0,
      "loss": 0.8145,
      "step": 5983
    },
    {
      "epoch": 0.5370669538682463,
      "grad_norm": 3.1731250286102295,
      "learning_rate": 0.0,
      "loss": 0.6503,
      "step": 5984
    },
    {
      "epoch": 0.537156704361874,
      "grad_norm": 2.890249013900757,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 5985
    },
    {
      "epoch": 0.5372464548555017,
      "grad_norm": 2.8286750316619873,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 5986
    },
    {
      "epoch": 0.5373362053491294,
      "grad_norm": 2.5693092346191406,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 5987
    },
    {
      "epoch": 0.5374259558427571,
      "grad_norm": 2.747572898864746,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 5988
    },
    {
      "epoch": 0.5375157063363849,
      "grad_norm": 2.4260032176971436,
      "learning_rate": 0.0,
      "loss": 0.6796,
      "step": 5989
    },
    {
      "epoch": 0.5376054568300126,
      "grad_norm": 3.006650447845459,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 5990
    },
    {
      "epoch": 0.5376952073236403,
      "grad_norm": 2.6214513778686523,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 5991
    },
    {
      "epoch": 0.537784957817268,
      "grad_norm": 2.9087748527526855,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 5992
    },
    {
      "epoch": 0.5378747083108957,
      "grad_norm": 4.542593955993652,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 5993
    },
    {
      "epoch": 0.5379644588045234,
      "grad_norm": 2.8318676948547363,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 5994
    },
    {
      "epoch": 0.5380542092981512,
      "grad_norm": 1.9652175903320312,
      "learning_rate": 0.0,
      "loss": 0.7863,
      "step": 5995
    },
    {
      "epoch": 0.5381439597917789,
      "grad_norm": 3.116731882095337,
      "learning_rate": 0.0,
      "loss": 0.7566,
      "step": 5996
    },
    {
      "epoch": 0.5382337102854066,
      "grad_norm": 3.057793378829956,
      "learning_rate": 0.0,
      "loss": 0.7653,
      "step": 5997
    },
    {
      "epoch": 0.5383234607790343,
      "grad_norm": 4.1389946937561035,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 5998
    },
    {
      "epoch": 0.538413211272662,
      "grad_norm": 5.702837944030762,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 5999
    },
    {
      "epoch": 0.5385029617662898,
      "grad_norm": 2.877441644668579,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 6000
    },
    {
      "epoch": 0.5385927122599174,
      "grad_norm": 2.4765961170196533,
      "learning_rate": 0.0,
      "loss": 0.7014,
      "step": 6001
    },
    {
      "epoch": 0.5386824627535451,
      "grad_norm": 2.2418997287750244,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 6002
    },
    {
      "epoch": 0.5387722132471728,
      "grad_norm": 2.920841693878174,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 6003
    },
    {
      "epoch": 0.5388619637408005,
      "grad_norm": 2.615950107574463,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 6004
    },
    {
      "epoch": 0.5389517142344283,
      "grad_norm": 2.503391981124878,
      "learning_rate": 0.0,
      "loss": 0.669,
      "step": 6005
    },
    {
      "epoch": 0.539041464728056,
      "grad_norm": 2.6435766220092773,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 6006
    },
    {
      "epoch": 0.5391312152216837,
      "grad_norm": 2.2871437072753906,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 6007
    },
    {
      "epoch": 0.5392209657153114,
      "grad_norm": 6.853264808654785,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 6008
    },
    {
      "epoch": 0.5393107162089391,
      "grad_norm": 2.7757625579833984,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 6009
    },
    {
      "epoch": 0.5394004667025669,
      "grad_norm": 2.361551284790039,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 6010
    },
    {
      "epoch": 0.5394902171961946,
      "grad_norm": 6.399425506591797,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 6011
    },
    {
      "epoch": 0.5395799676898223,
      "grad_norm": 3.565762996673584,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 6012
    },
    {
      "epoch": 0.53966971818345,
      "grad_norm": 7.086522579193115,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 6013
    },
    {
      "epoch": 0.5397594686770777,
      "grad_norm": 3.3930275440216064,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 6014
    },
    {
      "epoch": 0.5398492191707054,
      "grad_norm": 2.542009115219116,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 6015
    },
    {
      "epoch": 0.5399389696643332,
      "grad_norm": 4.099908351898193,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 6016
    },
    {
      "epoch": 0.5400287201579609,
      "grad_norm": 2.046060085296631,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 6017
    },
    {
      "epoch": 0.5401184706515886,
      "grad_norm": 6.205502986907959,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 6018
    },
    {
      "epoch": 0.5402082211452163,
      "grad_norm": 2.65448260307312,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 6019
    },
    {
      "epoch": 0.540297971638844,
      "grad_norm": 2.3525264263153076,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 6020
    },
    {
      "epoch": 0.5403877221324718,
      "grad_norm": 2.458564043045044,
      "learning_rate": 0.0,
      "loss": 0.7823,
      "step": 6021
    },
    {
      "epoch": 0.5404774726260995,
      "grad_norm": 2.6262593269348145,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 6022
    },
    {
      "epoch": 0.5405672231197272,
      "grad_norm": 4.749284744262695,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 6023
    },
    {
      "epoch": 0.5406569736133549,
      "grad_norm": 8.675591468811035,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 6024
    },
    {
      "epoch": 0.5407467241069825,
      "grad_norm": 2.763986110687256,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 6025
    },
    {
      "epoch": 0.5408364746006104,
      "grad_norm": 2.438784122467041,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 6026
    },
    {
      "epoch": 0.540926225094238,
      "grad_norm": 2.2888400554656982,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 6027
    },
    {
      "epoch": 0.5410159755878657,
      "grad_norm": 2.7368946075439453,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 6028
    },
    {
      "epoch": 0.5411057260814934,
      "grad_norm": 2.584459066390991,
      "learning_rate": 0.0,
      "loss": 0.7939,
      "step": 6029
    },
    {
      "epoch": 0.5411954765751211,
      "grad_norm": 4.235699653625488,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 6030
    },
    {
      "epoch": 0.5412852270687489,
      "grad_norm": 2.873614549636841,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 6031
    },
    {
      "epoch": 0.5413749775623766,
      "grad_norm": 1.8535387516021729,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 6032
    },
    {
      "epoch": 0.5414647280560043,
      "grad_norm": 5.845251083374023,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 6033
    },
    {
      "epoch": 0.541554478549632,
      "grad_norm": 3.1332015991210938,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 6034
    },
    {
      "epoch": 0.5416442290432597,
      "grad_norm": 2.7682385444641113,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 6035
    },
    {
      "epoch": 0.5417339795368874,
      "grad_norm": 2.3248374462127686,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 6036
    },
    {
      "epoch": 0.5418237300305152,
      "grad_norm": 3.0894439220428467,
      "learning_rate": 0.0,
      "loss": 0.6507,
      "step": 6037
    },
    {
      "epoch": 0.5419134805241429,
      "grad_norm": 2.4430699348449707,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 6038
    },
    {
      "epoch": 0.5420032310177706,
      "grad_norm": 5.543582916259766,
      "learning_rate": 0.0,
      "loss": 0.7533,
      "step": 6039
    },
    {
      "epoch": 0.5420929815113983,
      "grad_norm": 3.2882697582244873,
      "learning_rate": 0.0,
      "loss": 0.6972,
      "step": 6040
    },
    {
      "epoch": 0.542182732005026,
      "grad_norm": 4.7928690910339355,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 6041
    },
    {
      "epoch": 0.5422724824986538,
      "grad_norm": 2.35451078414917,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 6042
    },
    {
      "epoch": 0.5423622329922815,
      "grad_norm": 2.172975778579712,
      "learning_rate": 0.0,
      "loss": 0.754,
      "step": 6043
    },
    {
      "epoch": 0.5424519834859092,
      "grad_norm": 5.992407321929932,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 6044
    },
    {
      "epoch": 0.5425417339795369,
      "grad_norm": 3.4620988368988037,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 6045
    },
    {
      "epoch": 0.5426314844731646,
      "grad_norm": 2.3603904247283936,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 6046
    },
    {
      "epoch": 0.5427212349667924,
      "grad_norm": 2.5064566135406494,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 6047
    },
    {
      "epoch": 0.5428109854604201,
      "grad_norm": 2.3517541885375977,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 6048
    },
    {
      "epoch": 0.5429007359540478,
      "grad_norm": 3.2446117401123047,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 6049
    },
    {
      "epoch": 0.5429904864476754,
      "grad_norm": 2.9620487689971924,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 6050
    },
    {
      "epoch": 0.5430802369413031,
      "grad_norm": 2.7427585124969482,
      "learning_rate": 0.0,
      "loss": 0.7395,
      "step": 6051
    },
    {
      "epoch": 0.543169987434931,
      "grad_norm": 2.957423210144043,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 6052
    },
    {
      "epoch": 0.5432597379285586,
      "grad_norm": 2.930253505706787,
      "learning_rate": 0.0,
      "loss": 0.7375,
      "step": 6053
    },
    {
      "epoch": 0.5433494884221863,
      "grad_norm": 4.732675075531006,
      "learning_rate": 0.0,
      "loss": 0.6789,
      "step": 6054
    },
    {
      "epoch": 0.543439238915814,
      "grad_norm": 3.1945371627807617,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 6055
    },
    {
      "epoch": 0.5435289894094417,
      "grad_norm": 5.3909735679626465,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 6056
    },
    {
      "epoch": 0.5436187399030694,
      "grad_norm": 3.041245937347412,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 6057
    },
    {
      "epoch": 0.5437084903966972,
      "grad_norm": 3.149477243423462,
      "learning_rate": 0.0,
      "loss": 0.7669,
      "step": 6058
    },
    {
      "epoch": 0.5437982408903249,
      "grad_norm": 2.8753368854522705,
      "learning_rate": 0.0,
      "loss": 0.6839,
      "step": 6059
    },
    {
      "epoch": 0.5438879913839526,
      "grad_norm": 3.7769596576690674,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 6060
    },
    {
      "epoch": 0.5439777418775803,
      "grad_norm": 3.3914246559143066,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 6061
    },
    {
      "epoch": 0.544067492371208,
      "grad_norm": 2.9518983364105225,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 6062
    },
    {
      "epoch": 0.5441572428648358,
      "grad_norm": 2.4697461128234863,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 6063
    },
    {
      "epoch": 0.5442469933584635,
      "grad_norm": 4.463583946228027,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 6064
    },
    {
      "epoch": 0.5443367438520912,
      "grad_norm": 2.302297830581665,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 6065
    },
    {
      "epoch": 0.5444264943457189,
      "grad_norm": 2.003127098083496,
      "learning_rate": 0.0,
      "loss": 0.7811,
      "step": 6066
    },
    {
      "epoch": 0.5445162448393466,
      "grad_norm": 3.0907936096191406,
      "learning_rate": 0.0,
      "loss": 0.6681,
      "step": 6067
    },
    {
      "epoch": 0.5446059953329744,
      "grad_norm": 2.566391706466675,
      "learning_rate": 0.0,
      "loss": 0.7759,
      "step": 6068
    },
    {
      "epoch": 0.5446957458266021,
      "grad_norm": 2.739065408706665,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 6069
    },
    {
      "epoch": 0.5447854963202298,
      "grad_norm": 2.770885705947876,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 6070
    },
    {
      "epoch": 0.5448752468138575,
      "grad_norm": 2.8169686794281006,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 6071
    },
    {
      "epoch": 0.5449649973074852,
      "grad_norm": 9.69361686706543,
      "learning_rate": 0.0,
      "loss": 0.689,
      "step": 6072
    },
    {
      "epoch": 0.545054747801113,
      "grad_norm": 2.63798189163208,
      "learning_rate": 0.0,
      "loss": 0.6974,
      "step": 6073
    },
    {
      "epoch": 0.5451444982947407,
      "grad_norm": 3.4279346466064453,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 6074
    },
    {
      "epoch": 0.5452342487883683,
      "grad_norm": 2.299152374267578,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 6075
    },
    {
      "epoch": 0.545323999281996,
      "grad_norm": 3.7583324909210205,
      "learning_rate": 0.0,
      "loss": 0.773,
      "step": 6076
    },
    {
      "epoch": 0.5454137497756237,
      "grad_norm": 4.456445693969727,
      "learning_rate": 0.0,
      "loss": 0.7786,
      "step": 6077
    },
    {
      "epoch": 0.5455035002692514,
      "grad_norm": 2.7812795639038086,
      "learning_rate": 0.0,
      "loss": 0.695,
      "step": 6078
    },
    {
      "epoch": 0.5455932507628792,
      "grad_norm": 3.864267587661743,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 6079
    },
    {
      "epoch": 0.5456830012565069,
      "grad_norm": 3.262972354888916,
      "learning_rate": 0.0,
      "loss": 0.6767,
      "step": 6080
    },
    {
      "epoch": 0.5457727517501346,
      "grad_norm": 3.6298749446868896,
      "learning_rate": 0.0,
      "loss": 0.7061,
      "step": 6081
    },
    {
      "epoch": 0.5458625022437623,
      "grad_norm": 2.7061548233032227,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 6082
    },
    {
      "epoch": 0.54595225273739,
      "grad_norm": 2.2902121543884277,
      "learning_rate": 0.0,
      "loss": 0.6535,
      "step": 6083
    },
    {
      "epoch": 0.5460420032310178,
      "grad_norm": 5.325524806976318,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 6084
    },
    {
      "epoch": 0.5461317537246455,
      "grad_norm": 3.568262815475464,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 6085
    },
    {
      "epoch": 0.5462215042182732,
      "grad_norm": 2.830739974975586,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 6086
    },
    {
      "epoch": 0.5463112547119009,
      "grad_norm": 2.8359270095825195,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 6087
    },
    {
      "epoch": 0.5464010052055286,
      "grad_norm": 3.9540481567382812,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 6088
    },
    {
      "epoch": 0.5464907556991564,
      "grad_norm": 3.211210250854492,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 6089
    },
    {
      "epoch": 0.5465805061927841,
      "grad_norm": 2.5457048416137695,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 6090
    },
    {
      "epoch": 0.5466702566864118,
      "grad_norm": 5.321426868438721,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 6091
    },
    {
      "epoch": 0.5467600071800395,
      "grad_norm": 3.5537736415863037,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 6092
    },
    {
      "epoch": 0.5468497576736672,
      "grad_norm": 5.211171627044678,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 6093
    },
    {
      "epoch": 0.546939508167295,
      "grad_norm": 2.8463032245635986,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 6094
    },
    {
      "epoch": 0.5470292586609227,
      "grad_norm": 2.1291513442993164,
      "learning_rate": 0.0,
      "loss": 0.6917,
      "step": 6095
    },
    {
      "epoch": 0.5471190091545504,
      "grad_norm": 2.8658506870269775,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 6096
    },
    {
      "epoch": 0.5472087596481781,
      "grad_norm": 3.2058746814727783,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 6097
    },
    {
      "epoch": 0.5472985101418057,
      "grad_norm": 4.169866561889648,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 6098
    },
    {
      "epoch": 0.5473882606354334,
      "grad_norm": 2.8266334533691406,
      "learning_rate": 0.0,
      "loss": 0.7596,
      "step": 6099
    },
    {
      "epoch": 0.5474780111290612,
      "grad_norm": 3.386946201324463,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 6100
    },
    {
      "epoch": 0.5475677616226889,
      "grad_norm": 3.297971248626709,
      "learning_rate": 0.0,
      "loss": 0.7906,
      "step": 6101
    },
    {
      "epoch": 0.5476575121163166,
      "grad_norm": 3.285914659500122,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 6102
    },
    {
      "epoch": 0.5477472626099443,
      "grad_norm": 2.6422743797302246,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 6103
    },
    {
      "epoch": 0.547837013103572,
      "grad_norm": 2.7744648456573486,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 6104
    },
    {
      "epoch": 0.5479267635971998,
      "grad_norm": 2.9534263610839844,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 6105
    },
    {
      "epoch": 0.5480165140908275,
      "grad_norm": 13.186691284179688,
      "learning_rate": 0.0,
      "loss": 0.7864,
      "step": 6106
    },
    {
      "epoch": 0.5481062645844552,
      "grad_norm": 3.467602252960205,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 6107
    },
    {
      "epoch": 0.5481960150780829,
      "grad_norm": 2.6410512924194336,
      "learning_rate": 0.0,
      "loss": 0.6595,
      "step": 6108
    },
    {
      "epoch": 0.5482857655717106,
      "grad_norm": 2.33762526512146,
      "learning_rate": 0.0,
      "loss": 0.7505,
      "step": 6109
    },
    {
      "epoch": 0.5483755160653384,
      "grad_norm": 2.763355016708374,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 6110
    },
    {
      "epoch": 0.5484652665589661,
      "grad_norm": 3.9591805934906006,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 6111
    },
    {
      "epoch": 0.5485550170525938,
      "grad_norm": 4.146978855133057,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 6112
    },
    {
      "epoch": 0.5486447675462215,
      "grad_norm": 2.6397833824157715,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 6113
    },
    {
      "epoch": 0.5487345180398492,
      "grad_norm": 2.0944983959198,
      "learning_rate": 0.0,
      "loss": 0.6975,
      "step": 6114
    },
    {
      "epoch": 0.548824268533477,
      "grad_norm": 2.442495346069336,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 6115
    },
    {
      "epoch": 0.5489140190271047,
      "grad_norm": 3.0515544414520264,
      "learning_rate": 0.0,
      "loss": 0.7669,
      "step": 6116
    },
    {
      "epoch": 0.5490037695207324,
      "grad_norm": 2.597273349761963,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 6117
    },
    {
      "epoch": 0.5490935200143601,
      "grad_norm": 3.4052388668060303,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 6118
    },
    {
      "epoch": 0.5491832705079878,
      "grad_norm": 2.8130416870117188,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 6119
    },
    {
      "epoch": 0.5492730210016155,
      "grad_norm": 3.602710008621216,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 6120
    },
    {
      "epoch": 0.5493627714952433,
      "grad_norm": 2.404344320297241,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 6121
    },
    {
      "epoch": 0.549452521988871,
      "grad_norm": 3.7401459217071533,
      "learning_rate": 0.0,
      "loss": 0.8053,
      "step": 6122
    },
    {
      "epoch": 0.5495422724824987,
      "grad_norm": 3.249572277069092,
      "learning_rate": 0.0,
      "loss": 0.6966,
      "step": 6123
    },
    {
      "epoch": 0.5496320229761263,
      "grad_norm": 2.197211980819702,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 6124
    },
    {
      "epoch": 0.549721773469754,
      "grad_norm": 2.7669622898101807,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 6125
    },
    {
      "epoch": 0.5498115239633818,
      "grad_norm": 5.226823806762695,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 6126
    },
    {
      "epoch": 0.5499012744570095,
      "grad_norm": 5.406783580780029,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 6127
    },
    {
      "epoch": 0.5499910249506372,
      "grad_norm": 2.6192996501922607,
      "learning_rate": 0.0,
      "loss": 0.771,
      "step": 6128
    },
    {
      "epoch": 0.5500807754442649,
      "grad_norm": 10.180231094360352,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 6129
    },
    {
      "epoch": 0.5501705259378926,
      "grad_norm": 2.433406114578247,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 6130
    },
    {
      "epoch": 0.5502602764315204,
      "grad_norm": 3.1003284454345703,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 6131
    },
    {
      "epoch": 0.5503500269251481,
      "grad_norm": 3.3349878787994385,
      "learning_rate": 0.0,
      "loss": 0.7363,
      "step": 6132
    },
    {
      "epoch": 0.5504397774187758,
      "grad_norm": 2.2013189792633057,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 6133
    },
    {
      "epoch": 0.5505295279124035,
      "grad_norm": 2.3200736045837402,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 6134
    },
    {
      "epoch": 0.5506192784060312,
      "grad_norm": 3.111576795578003,
      "learning_rate": 0.0,
      "loss": 0.7781,
      "step": 6135
    },
    {
      "epoch": 0.550709028899659,
      "grad_norm": 3.7465896606445312,
      "learning_rate": 0.0,
      "loss": 0.6589,
      "step": 6136
    },
    {
      "epoch": 0.5507987793932867,
      "grad_norm": 2.742300033569336,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 6137
    },
    {
      "epoch": 0.5508885298869144,
      "grad_norm": 2.0346908569335938,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 6138
    },
    {
      "epoch": 0.5509782803805421,
      "grad_norm": 2.5432016849517822,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 6139
    },
    {
      "epoch": 0.5510680308741698,
      "grad_norm": 3.9417502880096436,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 6140
    },
    {
      "epoch": 0.5511577813677975,
      "grad_norm": 3.0354557037353516,
      "learning_rate": 0.0,
      "loss": 0.788,
      "step": 6141
    },
    {
      "epoch": 0.5512475318614253,
      "grad_norm": 2.0865421295166016,
      "learning_rate": 0.0,
      "loss": 0.8114,
      "step": 6142
    },
    {
      "epoch": 0.551337282355053,
      "grad_norm": 2.3962748050689697,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 6143
    },
    {
      "epoch": 0.5514270328486807,
      "grad_norm": 2.92214298248291,
      "learning_rate": 0.0,
      "loss": 0.6966,
      "step": 6144
    },
    {
      "epoch": 0.5515167833423084,
      "grad_norm": 2.8245790004730225,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 6145
    },
    {
      "epoch": 0.551606533835936,
      "grad_norm": 4.060076713562012,
      "learning_rate": 0.0,
      "loss": 0.6967,
      "step": 6146
    },
    {
      "epoch": 0.5516962843295639,
      "grad_norm": 2.7633891105651855,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 6147
    },
    {
      "epoch": 0.5517860348231916,
      "grad_norm": 2.672606945037842,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 6148
    },
    {
      "epoch": 0.5518757853168192,
      "grad_norm": 1.9765950441360474,
      "learning_rate": 0.0,
      "loss": 0.6963,
      "step": 6149
    },
    {
      "epoch": 0.5519655358104469,
      "grad_norm": 1.9480071067810059,
      "learning_rate": 0.0,
      "loss": 0.7047,
      "step": 6150
    },
    {
      "epoch": 0.5520552863040746,
      "grad_norm": 3.290787696838379,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 6151
    },
    {
      "epoch": 0.5521450367977024,
      "grad_norm": 2.15651273727417,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 6152
    },
    {
      "epoch": 0.5522347872913301,
      "grad_norm": 2.8198182582855225,
      "learning_rate": 0.0,
      "loss": 0.7967,
      "step": 6153
    },
    {
      "epoch": 0.5523245377849578,
      "grad_norm": 4.783196926116943,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 6154
    },
    {
      "epoch": 0.5524142882785855,
      "grad_norm": 6.705075740814209,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 6155
    },
    {
      "epoch": 0.5525040387722132,
      "grad_norm": 2.3266377449035645,
      "learning_rate": 0.0,
      "loss": 0.7656,
      "step": 6156
    },
    {
      "epoch": 0.5525937892658409,
      "grad_norm": 2.1803057193756104,
      "learning_rate": 0.0,
      "loss": 0.6708,
      "step": 6157
    },
    {
      "epoch": 0.5526835397594687,
      "grad_norm": 2.6434783935546875,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 6158
    },
    {
      "epoch": 0.5527732902530964,
      "grad_norm": 2.11747670173645,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 6159
    },
    {
      "epoch": 0.5528630407467241,
      "grad_norm": 2.324526786804199,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 6160
    },
    {
      "epoch": 0.5529527912403518,
      "grad_norm": 2.9071106910705566,
      "learning_rate": 0.0,
      "loss": 0.7585,
      "step": 6161
    },
    {
      "epoch": 0.5530425417339795,
      "grad_norm": 2.50877046585083,
      "learning_rate": 0.0,
      "loss": 0.694,
      "step": 6162
    },
    {
      "epoch": 0.5531322922276073,
      "grad_norm": 2.6252799034118652,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 6163
    },
    {
      "epoch": 0.553222042721235,
      "grad_norm": 2.705735683441162,
      "learning_rate": 0.0,
      "loss": 0.8479,
      "step": 6164
    },
    {
      "epoch": 0.5533117932148627,
      "grad_norm": 2.4741170406341553,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 6165
    },
    {
      "epoch": 0.5534015437084904,
      "grad_norm": 4.178325176239014,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 6166
    },
    {
      "epoch": 0.5534912942021181,
      "grad_norm": 3.734400510787964,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 6167
    },
    {
      "epoch": 0.5535810446957459,
      "grad_norm": 2.744551658630371,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 6168
    },
    {
      "epoch": 0.5536707951893736,
      "grad_norm": 4.453043460845947,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 6169
    },
    {
      "epoch": 0.5537605456830013,
      "grad_norm": 2.7821598052978516,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 6170
    },
    {
      "epoch": 0.553850296176629,
      "grad_norm": 2.980715036392212,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 6171
    },
    {
      "epoch": 0.5539400466702566,
      "grad_norm": 5.165645122528076,
      "learning_rate": 0.0,
      "loss": 0.6944,
      "step": 6172
    },
    {
      "epoch": 0.5540297971638845,
      "grad_norm": 2.4001410007476807,
      "learning_rate": 0.0,
      "loss": 0.6837,
      "step": 6173
    },
    {
      "epoch": 0.5541195476575121,
      "grad_norm": 3.5437235832214355,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 6174
    },
    {
      "epoch": 0.5542092981511398,
      "grad_norm": 2.5245790481567383,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 6175
    },
    {
      "epoch": 0.5542990486447675,
      "grad_norm": 3.7008118629455566,
      "learning_rate": 0.0,
      "loss": 0.7507,
      "step": 6176
    },
    {
      "epoch": 0.5543887991383952,
      "grad_norm": 3.358393430709839,
      "learning_rate": 0.0,
      "loss": 0.7741,
      "step": 6177
    },
    {
      "epoch": 0.5544785496320229,
      "grad_norm": 2.220231294631958,
      "learning_rate": 0.0,
      "loss": 0.7573,
      "step": 6178
    },
    {
      "epoch": 0.5545683001256507,
      "grad_norm": 2.232879400253296,
      "learning_rate": 0.0,
      "loss": 0.7125,
      "step": 6179
    },
    {
      "epoch": 0.5546580506192784,
      "grad_norm": 2.2182140350341797,
      "learning_rate": 0.0,
      "loss": 0.7034,
      "step": 6180
    },
    {
      "epoch": 0.5547478011129061,
      "grad_norm": 2.3086163997650146,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 6181
    },
    {
      "epoch": 0.5548375516065338,
      "grad_norm": 3.325686454772949,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 6182
    },
    {
      "epoch": 0.5549273021001615,
      "grad_norm": 2.5628983974456787,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 6183
    },
    {
      "epoch": 0.5550170525937893,
      "grad_norm": 3.210663080215454,
      "learning_rate": 0.0,
      "loss": 0.7207,
      "step": 6184
    },
    {
      "epoch": 0.555106803087417,
      "grad_norm": 2.8539373874664307,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 6185
    },
    {
      "epoch": 0.5551965535810447,
      "grad_norm": 3.329158306121826,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 6186
    },
    {
      "epoch": 0.5552863040746724,
      "grad_norm": 6.730790615081787,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 6187
    },
    {
      "epoch": 0.5553760545683001,
      "grad_norm": 3.075068712234497,
      "learning_rate": 0.0,
      "loss": 0.7893,
      "step": 6188
    },
    {
      "epoch": 0.5554658050619279,
      "grad_norm": 2.9889402389526367,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 6189
    },
    {
      "epoch": 0.5555555555555556,
      "grad_norm": 5.567690849304199,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 6190
    },
    {
      "epoch": 0.5556453060491833,
      "grad_norm": 3.6670284271240234,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 6191
    },
    {
      "epoch": 0.555735056542811,
      "grad_norm": 2.7121260166168213,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 6192
    },
    {
      "epoch": 0.5558248070364387,
      "grad_norm": 2.050419569015503,
      "learning_rate": 0.0,
      "loss": 0.6553,
      "step": 6193
    },
    {
      "epoch": 0.5559145575300665,
      "grad_norm": 3.1699163913726807,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 6194
    },
    {
      "epoch": 0.5560043080236942,
      "grad_norm": 3.4073548316955566,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 6195
    },
    {
      "epoch": 0.5560940585173219,
      "grad_norm": 2.6466758251190186,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 6196
    },
    {
      "epoch": 0.5561838090109495,
      "grad_norm": 2.1622331142425537,
      "learning_rate": 0.0,
      "loss": 0.7767,
      "step": 6197
    },
    {
      "epoch": 0.5562735595045772,
      "grad_norm": 2.215467691421509,
      "learning_rate": 0.0,
      "loss": 0.7957,
      "step": 6198
    },
    {
      "epoch": 0.5563633099982049,
      "grad_norm": 2.445812940597534,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 6199
    },
    {
      "epoch": 0.5564530604918327,
      "grad_norm": 2.561129570007324,
      "learning_rate": 0.0,
      "loss": 0.794,
      "step": 6200
    },
    {
      "epoch": 0.5565428109854604,
      "grad_norm": 2.6618404388427734,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 6201
    },
    {
      "epoch": 0.5566325614790881,
      "grad_norm": 4.042046546936035,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 6202
    },
    {
      "epoch": 0.5567223119727158,
      "grad_norm": 2.731781482696533,
      "learning_rate": 0.0,
      "loss": 0.6891,
      "step": 6203
    },
    {
      "epoch": 0.5568120624663435,
      "grad_norm": 2.649921178817749,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 6204
    },
    {
      "epoch": 0.5569018129599713,
      "grad_norm": 2.432805299758911,
      "learning_rate": 0.0,
      "loss": 0.7967,
      "step": 6205
    },
    {
      "epoch": 0.556991563453599,
      "grad_norm": 2.698108673095703,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 6206
    },
    {
      "epoch": 0.5570813139472267,
      "grad_norm": 2.7404401302337646,
      "learning_rate": 0.0,
      "loss": 0.7702,
      "step": 6207
    },
    {
      "epoch": 0.5571710644408544,
      "grad_norm": 2.5610814094543457,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 6208
    },
    {
      "epoch": 0.5572608149344821,
      "grad_norm": 2.5635647773742676,
      "learning_rate": 0.0,
      "loss": 0.7838,
      "step": 6209
    },
    {
      "epoch": 0.5573505654281099,
      "grad_norm": 2.2035977840423584,
      "learning_rate": 0.0,
      "loss": 0.6814,
      "step": 6210
    },
    {
      "epoch": 0.5574403159217376,
      "grad_norm": 2.9992408752441406,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 6211
    },
    {
      "epoch": 0.5575300664153653,
      "grad_norm": 3.058586597442627,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 6212
    },
    {
      "epoch": 0.557619816908993,
      "grad_norm": 2.794956684112549,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 6213
    },
    {
      "epoch": 0.5577095674026207,
      "grad_norm": 2.4325499534606934,
      "learning_rate": 0.0,
      "loss": 0.695,
      "step": 6214
    },
    {
      "epoch": 0.5577993178962485,
      "grad_norm": 2.4534599781036377,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 6215
    },
    {
      "epoch": 0.5578890683898762,
      "grad_norm": 2.52707839012146,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 6216
    },
    {
      "epoch": 0.5579788188835039,
      "grad_norm": 4.707418918609619,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 6217
    },
    {
      "epoch": 0.5580685693771316,
      "grad_norm": 3.9128479957580566,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 6218
    },
    {
      "epoch": 0.5581583198707593,
      "grad_norm": 2.7165675163269043,
      "learning_rate": 0.0,
      "loss": 0.69,
      "step": 6219
    },
    {
      "epoch": 0.558248070364387,
      "grad_norm": 2.6990771293640137,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 6220
    },
    {
      "epoch": 0.5583378208580148,
      "grad_norm": 2.552581787109375,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 6221
    },
    {
      "epoch": 0.5584275713516424,
      "grad_norm": 3.2020225524902344,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 6222
    },
    {
      "epoch": 0.5585173218452701,
      "grad_norm": 2.591627836227417,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 6223
    },
    {
      "epoch": 0.5586070723388978,
      "grad_norm": 2.750150203704834,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 6224
    },
    {
      "epoch": 0.5586968228325255,
      "grad_norm": 2.8327953815460205,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 6225
    },
    {
      "epoch": 0.5587865733261533,
      "grad_norm": 2.193049669265747,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 6226
    },
    {
      "epoch": 0.558876323819781,
      "grad_norm": 2.5947694778442383,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 6227
    },
    {
      "epoch": 0.5589660743134087,
      "grad_norm": 3.136061429977417,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 6228
    },
    {
      "epoch": 0.5590558248070364,
      "grad_norm": 2.9114561080932617,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 6229
    },
    {
      "epoch": 0.5591455753006641,
      "grad_norm": 2.3563361167907715,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 6230
    },
    {
      "epoch": 0.5592353257942919,
      "grad_norm": 2.883892774581909,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 6231
    },
    {
      "epoch": 0.5593250762879196,
      "grad_norm": 5.84513521194458,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 6232
    },
    {
      "epoch": 0.5594148267815473,
      "grad_norm": 3.6802725791931152,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 6233
    },
    {
      "epoch": 0.559504577275175,
      "grad_norm": 3.2062339782714844,
      "learning_rate": 0.0,
      "loss": 0.6703,
      "step": 6234
    },
    {
      "epoch": 0.5595943277688027,
      "grad_norm": 2.35896635055542,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 6235
    },
    {
      "epoch": 0.5596840782624305,
      "grad_norm": 2.678760528564453,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 6236
    },
    {
      "epoch": 0.5597738287560582,
      "grad_norm": 3.089815139770508,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 6237
    },
    {
      "epoch": 0.5598635792496859,
      "grad_norm": 2.7215611934661865,
      "learning_rate": 0.0,
      "loss": 0.7734,
      "step": 6238
    },
    {
      "epoch": 0.5599533297433136,
      "grad_norm": 4.269233226776123,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 6239
    },
    {
      "epoch": 0.5600430802369413,
      "grad_norm": 2.4448063373565674,
      "learning_rate": 0.0,
      "loss": 0.7618,
      "step": 6240
    },
    {
      "epoch": 0.560132830730569,
      "grad_norm": 2.0021727085113525,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 6241
    },
    {
      "epoch": 0.5602225812241968,
      "grad_norm": 2.652488946914673,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 6242
    },
    {
      "epoch": 0.5603123317178245,
      "grad_norm": 2.4332995414733887,
      "learning_rate": 0.0,
      "loss": 0.7375,
      "step": 6243
    },
    {
      "epoch": 0.5604020822114522,
      "grad_norm": 3.0527377128601074,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 6244
    },
    {
      "epoch": 0.5604918327050799,
      "grad_norm": 7.13448429107666,
      "learning_rate": 0.0,
      "loss": 0.6512,
      "step": 6245
    },
    {
      "epoch": 0.5605815831987075,
      "grad_norm": 2.825651168823242,
      "learning_rate": 0.0,
      "loss": 0.6724,
      "step": 6246
    },
    {
      "epoch": 0.5606713336923353,
      "grad_norm": 2.1484012603759766,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 6247
    },
    {
      "epoch": 0.560761084185963,
      "grad_norm": 2.9704151153564453,
      "learning_rate": 0.0,
      "loss": 0.7653,
      "step": 6248
    },
    {
      "epoch": 0.5608508346795907,
      "grad_norm": 2.4635047912597656,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 6249
    },
    {
      "epoch": 0.5609405851732184,
      "grad_norm": 2.325615882873535,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 6250
    },
    {
      "epoch": 0.5610303356668461,
      "grad_norm": 2.1665537357330322,
      "learning_rate": 0.0,
      "loss": 0.6915,
      "step": 6251
    },
    {
      "epoch": 0.5611200861604739,
      "grad_norm": 2.796600818634033,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 6252
    },
    {
      "epoch": 0.5612098366541016,
      "grad_norm": 2.2312428951263428,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 6253
    },
    {
      "epoch": 0.5612995871477293,
      "grad_norm": 2.8748908042907715,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 6254
    },
    {
      "epoch": 0.561389337641357,
      "grad_norm": 2.514401912689209,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 6255
    },
    {
      "epoch": 0.5614790881349847,
      "grad_norm": 2.3188838958740234,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 6256
    },
    {
      "epoch": 0.5615688386286125,
      "grad_norm": 2.5680253505706787,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 6257
    },
    {
      "epoch": 0.5616585891222402,
      "grad_norm": 2.624392032623291,
      "learning_rate": 0.0,
      "loss": 0.7432,
      "step": 6258
    },
    {
      "epoch": 0.5617483396158679,
      "grad_norm": 2.46639084815979,
      "learning_rate": 0.0,
      "loss": 0.7925,
      "step": 6259
    },
    {
      "epoch": 0.5618380901094956,
      "grad_norm": 3.0554003715515137,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 6260
    },
    {
      "epoch": 0.5619278406031233,
      "grad_norm": 3.2699623107910156,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 6261
    },
    {
      "epoch": 0.562017591096751,
      "grad_norm": 2.7007246017456055,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 6262
    },
    {
      "epoch": 0.5621073415903788,
      "grad_norm": 2.2222890853881836,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 6263
    },
    {
      "epoch": 0.5621970920840065,
      "grad_norm": 2.056328773498535,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 6264
    },
    {
      "epoch": 0.5622868425776342,
      "grad_norm": 2.4622013568878174,
      "learning_rate": 0.0,
      "loss": 0.6554,
      "step": 6265
    },
    {
      "epoch": 0.5623765930712619,
      "grad_norm": 3.047398805618286,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 6266
    },
    {
      "epoch": 0.5624663435648896,
      "grad_norm": 2.669174909591675,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 6267
    },
    {
      "epoch": 0.5625560940585174,
      "grad_norm": 2.803656578063965,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 6268
    },
    {
      "epoch": 0.5626458445521451,
      "grad_norm": 2.5758440494537354,
      "learning_rate": 0.0,
      "loss": 0.7172,
      "step": 6269
    },
    {
      "epoch": 0.5627355950457728,
      "grad_norm": 6.009243965148926,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 6270
    },
    {
      "epoch": 0.5628253455394004,
      "grad_norm": 2.473064661026001,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 6271
    },
    {
      "epoch": 0.5629150960330281,
      "grad_norm": 2.63642954826355,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 6272
    },
    {
      "epoch": 0.5630048465266559,
      "grad_norm": 4.012826442718506,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 6273
    },
    {
      "epoch": 0.5630945970202836,
      "grad_norm": 2.850473165512085,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 6274
    },
    {
      "epoch": 0.5631843475139113,
      "grad_norm": 2.4488742351531982,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 6275
    },
    {
      "epoch": 0.563274098007539,
      "grad_norm": 7.503546714782715,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 6276
    },
    {
      "epoch": 0.5633638485011667,
      "grad_norm": 3.282505989074707,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 6277
    },
    {
      "epoch": 0.5634535989947945,
      "grad_norm": 4.003122806549072,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 6278
    },
    {
      "epoch": 0.5635433494884222,
      "grad_norm": 2.803772449493408,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 6279
    },
    {
      "epoch": 0.5636330999820499,
      "grad_norm": 2.253434658050537,
      "learning_rate": 0.0,
      "loss": 0.6538,
      "step": 6280
    },
    {
      "epoch": 0.5637228504756776,
      "grad_norm": 2.7065556049346924,
      "learning_rate": 0.0,
      "loss": 0.7888,
      "step": 6281
    },
    {
      "epoch": 0.5638126009693053,
      "grad_norm": 3.5953192710876465,
      "learning_rate": 0.0,
      "loss": 0.6784,
      "step": 6282
    },
    {
      "epoch": 0.563902351462933,
      "grad_norm": 3.3463282585144043,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 6283
    },
    {
      "epoch": 0.5639921019565608,
      "grad_norm": 2.754713535308838,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 6284
    },
    {
      "epoch": 0.5640818524501885,
      "grad_norm": 3.6446449756622314,
      "learning_rate": 0.0,
      "loss": 0.7678,
      "step": 6285
    },
    {
      "epoch": 0.5641716029438162,
      "grad_norm": 2.8126413822174072,
      "learning_rate": 0.0,
      "loss": 0.8348,
      "step": 6286
    },
    {
      "epoch": 0.5642613534374439,
      "grad_norm": 2.514221668243408,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 6287
    },
    {
      "epoch": 0.5643511039310716,
      "grad_norm": 2.384793519973755,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 6288
    },
    {
      "epoch": 0.5644408544246994,
      "grad_norm": 3.1560072898864746,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 6289
    },
    {
      "epoch": 0.5645306049183271,
      "grad_norm": 2.6988396644592285,
      "learning_rate": 0.0,
      "loss": 0.7056,
      "step": 6290
    },
    {
      "epoch": 0.5646203554119548,
      "grad_norm": 2.7877731323242188,
      "learning_rate": 0.0,
      "loss": 0.7891,
      "step": 6291
    },
    {
      "epoch": 0.5647101059055825,
      "grad_norm": 2.8255138397216797,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 6292
    },
    {
      "epoch": 0.5647998563992102,
      "grad_norm": 2.8149425983428955,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 6293
    },
    {
      "epoch": 0.564889606892838,
      "grad_norm": 3.560914993286133,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 6294
    },
    {
      "epoch": 0.5649793573864657,
      "grad_norm": 2.7693684101104736,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 6295
    },
    {
      "epoch": 0.5650691078800933,
      "grad_norm": 3.642225980758667,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 6296
    },
    {
      "epoch": 0.565158858373721,
      "grad_norm": 2.863889217376709,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 6297
    },
    {
      "epoch": 0.5652486088673487,
      "grad_norm": 2.8051934242248535,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 6298
    },
    {
      "epoch": 0.5653383593609765,
      "grad_norm": 3.6144652366638184,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 6299
    },
    {
      "epoch": 0.5654281098546042,
      "grad_norm": 2.5175836086273193,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 6300
    },
    {
      "epoch": 0.5655178603482319,
      "grad_norm": 2.6727969646453857,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 6301
    },
    {
      "epoch": 0.5656076108418596,
      "grad_norm": 2.875631809234619,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 6302
    },
    {
      "epoch": 0.5656973613354873,
      "grad_norm": 2.922457456588745,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 6303
    },
    {
      "epoch": 0.565787111829115,
      "grad_norm": 3.4368066787719727,
      "learning_rate": 0.0,
      "loss": 0.7193,
      "step": 6304
    },
    {
      "epoch": 0.5658768623227428,
      "grad_norm": 3.161959409713745,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 6305
    },
    {
      "epoch": 0.5659666128163705,
      "grad_norm": 18.638097763061523,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 6306
    },
    {
      "epoch": 0.5660563633099982,
      "grad_norm": 3.4703516960144043,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 6307
    },
    {
      "epoch": 0.5661461138036259,
      "grad_norm": 3.8781232833862305,
      "learning_rate": 0.0,
      "loss": 0.7516,
      "step": 6308
    },
    {
      "epoch": 0.5662358642972536,
      "grad_norm": 2.59521222114563,
      "learning_rate": 0.0,
      "loss": 0.6773,
      "step": 6309
    },
    {
      "epoch": 0.5663256147908814,
      "grad_norm": 2.909571647644043,
      "learning_rate": 0.0,
      "loss": 0.7447,
      "step": 6310
    },
    {
      "epoch": 0.5664153652845091,
      "grad_norm": 3.2637221813201904,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 6311
    },
    {
      "epoch": 0.5665051157781368,
      "grad_norm": 2.6595399379730225,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 6312
    },
    {
      "epoch": 0.5665948662717645,
      "grad_norm": 3.6945383548736572,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 6313
    },
    {
      "epoch": 0.5666846167653922,
      "grad_norm": 2.460404634475708,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 6314
    },
    {
      "epoch": 0.56677436725902,
      "grad_norm": 2.626826763153076,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 6315
    },
    {
      "epoch": 0.5668641177526477,
      "grad_norm": 2.969428062438965,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 6316
    },
    {
      "epoch": 0.5669538682462754,
      "grad_norm": 4.644039154052734,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 6317
    },
    {
      "epoch": 0.567043618739903,
      "grad_norm": 2.920997381210327,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 6318
    },
    {
      "epoch": 0.5671333692335307,
      "grad_norm": 3.1806745529174805,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 6319
    },
    {
      "epoch": 0.5672231197271586,
      "grad_norm": 3.1284658908843994,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 6320
    },
    {
      "epoch": 0.5673128702207862,
      "grad_norm": 3.034148931503296,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 6321
    },
    {
      "epoch": 0.5674026207144139,
      "grad_norm": 2.611992597579956,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 6322
    },
    {
      "epoch": 0.5674923712080416,
      "grad_norm": 4.208182334899902,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 6323
    },
    {
      "epoch": 0.5675821217016693,
      "grad_norm": 6.113923072814941,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 6324
    },
    {
      "epoch": 0.567671872195297,
      "grad_norm": 3.676654815673828,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 6325
    },
    {
      "epoch": 0.5677616226889248,
      "grad_norm": 2.7024080753326416,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 6326
    },
    {
      "epoch": 0.5678513731825525,
      "grad_norm": 2.3681795597076416,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 6327
    },
    {
      "epoch": 0.5679411236761802,
      "grad_norm": 2.6633381843566895,
      "learning_rate": 0.0,
      "loss": 0.7683,
      "step": 6328
    },
    {
      "epoch": 0.5680308741698079,
      "grad_norm": 3.365102529525757,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 6329
    },
    {
      "epoch": 0.5681206246634356,
      "grad_norm": 5.895088195800781,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 6330
    },
    {
      "epoch": 0.5682103751570634,
      "grad_norm": 3.633753776550293,
      "learning_rate": 0.0,
      "loss": 0.7128,
      "step": 6331
    },
    {
      "epoch": 0.5683001256506911,
      "grad_norm": 3.269329786300659,
      "learning_rate": 0.0,
      "loss": 0.7645,
      "step": 6332
    },
    {
      "epoch": 0.5683898761443188,
      "grad_norm": 2.253084182739258,
      "learning_rate": 0.0,
      "loss": 0.7922,
      "step": 6333
    },
    {
      "epoch": 0.5684796266379465,
      "grad_norm": 1.9545801877975464,
      "learning_rate": 0.0,
      "loss": 0.7463,
      "step": 6334
    },
    {
      "epoch": 0.5685693771315742,
      "grad_norm": 2.9061484336853027,
      "learning_rate": 0.0,
      "loss": 0.7458,
      "step": 6335
    },
    {
      "epoch": 0.568659127625202,
      "grad_norm": 2.4270730018615723,
      "learning_rate": 0.0,
      "loss": 0.6531,
      "step": 6336
    },
    {
      "epoch": 0.5687488781188297,
      "grad_norm": 3.6893234252929688,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 6337
    },
    {
      "epoch": 0.5688386286124574,
      "grad_norm": 3.2402024269104004,
      "learning_rate": 0.0,
      "loss": 0.6668,
      "step": 6338
    },
    {
      "epoch": 0.5689283791060851,
      "grad_norm": 2.9095633029937744,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 6339
    },
    {
      "epoch": 0.5690181295997128,
      "grad_norm": 3.929690361022949,
      "learning_rate": 0.0,
      "loss": 0.6797,
      "step": 6340
    },
    {
      "epoch": 0.5691078800933406,
      "grad_norm": 3.9892117977142334,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 6341
    },
    {
      "epoch": 0.5691976305869683,
      "grad_norm": 2.88747239112854,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 6342
    },
    {
      "epoch": 0.569287381080596,
      "grad_norm": 3.3603296279907227,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 6343
    },
    {
      "epoch": 0.5693771315742236,
      "grad_norm": 3.6015753746032715,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 6344
    },
    {
      "epoch": 0.5694668820678513,
      "grad_norm": 14.280179023742676,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 6345
    },
    {
      "epoch": 0.569556632561479,
      "grad_norm": 3.2080657482147217,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 6346
    },
    {
      "epoch": 0.5696463830551068,
      "grad_norm": 3.436967134475708,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 6347
    },
    {
      "epoch": 0.5697361335487345,
      "grad_norm": 3.0686988830566406,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 6348
    },
    {
      "epoch": 0.5698258840423622,
      "grad_norm": 3.610941171646118,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 6349
    },
    {
      "epoch": 0.5699156345359899,
      "grad_norm": 2.563300132751465,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 6350
    },
    {
      "epoch": 0.5700053850296176,
      "grad_norm": 3.2262649536132812,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 6351
    },
    {
      "epoch": 0.5700951355232454,
      "grad_norm": 4.336414337158203,
      "learning_rate": 0.0,
      "loss": 0.7728,
      "step": 6352
    },
    {
      "epoch": 0.5701848860168731,
      "grad_norm": 3.9029595851898193,
      "learning_rate": 0.0,
      "loss": 0.7003,
      "step": 6353
    },
    {
      "epoch": 0.5702746365105008,
      "grad_norm": 3.2718634605407715,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 6354
    },
    {
      "epoch": 0.5703643870041285,
      "grad_norm": 2.703645944595337,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 6355
    },
    {
      "epoch": 0.5704541374977562,
      "grad_norm": 2.5259289741516113,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 6356
    },
    {
      "epoch": 0.570543887991384,
      "grad_norm": 3.7406961917877197,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 6357
    },
    {
      "epoch": 0.5706336384850117,
      "grad_norm": 2.500594139099121,
      "learning_rate": 0.0,
      "loss": 0.6803,
      "step": 6358
    },
    {
      "epoch": 0.5707233889786394,
      "grad_norm": 2.5721447467803955,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 6359
    },
    {
      "epoch": 0.5708131394722671,
      "grad_norm": 7.136885166168213,
      "learning_rate": 0.0,
      "loss": 0.7796,
      "step": 6360
    },
    {
      "epoch": 0.5709028899658948,
      "grad_norm": 3.514664649963379,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 6361
    },
    {
      "epoch": 0.5709926404595226,
      "grad_norm": 2.720379114151001,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 6362
    },
    {
      "epoch": 0.5710823909531503,
      "grad_norm": 2.6391379833221436,
      "learning_rate": 0.0,
      "loss": 0.69,
      "step": 6363
    },
    {
      "epoch": 0.571172141446778,
      "grad_norm": 2.94912052154541,
      "learning_rate": 0.0,
      "loss": 0.6464,
      "step": 6364
    },
    {
      "epoch": 0.5712618919404057,
      "grad_norm": 3.8604044914245605,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 6365
    },
    {
      "epoch": 0.5713516424340334,
      "grad_norm": 2.4311001300811768,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 6366
    },
    {
      "epoch": 0.571441392927661,
      "grad_norm": 2.3993852138519287,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 6367
    },
    {
      "epoch": 0.5715311434212889,
      "grad_norm": 2.608206033706665,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 6368
    },
    {
      "epoch": 0.5716208939149166,
      "grad_norm": 2.6794559955596924,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 6369
    },
    {
      "epoch": 0.5717106444085442,
      "grad_norm": 2.736140727996826,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 6370
    },
    {
      "epoch": 0.5718003949021719,
      "grad_norm": 2.2023775577545166,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 6371
    },
    {
      "epoch": 0.5718901453957996,
      "grad_norm": 2.699930429458618,
      "learning_rate": 0.0,
      "loss": 0.7718,
      "step": 6372
    },
    {
      "epoch": 0.5719798958894274,
      "grad_norm": 3.217470169067383,
      "learning_rate": 0.0,
      "loss": 0.6771,
      "step": 6373
    },
    {
      "epoch": 0.5720696463830551,
      "grad_norm": 9.029288291931152,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 6374
    },
    {
      "epoch": 0.5721593968766828,
      "grad_norm": 3.644439458847046,
      "learning_rate": 0.0,
      "loss": 0.6746,
      "step": 6375
    },
    {
      "epoch": 0.5722491473703105,
      "grad_norm": 3.1176059246063232,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 6376
    },
    {
      "epoch": 0.5723388978639382,
      "grad_norm": 3.6284070014953613,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 6377
    },
    {
      "epoch": 0.572428648357566,
      "grad_norm": 2.933868408203125,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 6378
    },
    {
      "epoch": 0.5725183988511937,
      "grad_norm": 2.5367910861968994,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 6379
    },
    {
      "epoch": 0.5726081493448214,
      "grad_norm": 2.6870369911193848,
      "learning_rate": 0.0,
      "loss": 0.6464,
      "step": 6380
    },
    {
      "epoch": 0.5726978998384491,
      "grad_norm": 5.275073051452637,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 6381
    },
    {
      "epoch": 0.5727876503320768,
      "grad_norm": 3.2070772647857666,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 6382
    },
    {
      "epoch": 0.5728774008257046,
      "grad_norm": 2.5038580894470215,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 6383
    },
    {
      "epoch": 0.5729671513193323,
      "grad_norm": 2.6357555389404297,
      "learning_rate": 0.0,
      "loss": 0.704,
      "step": 6384
    },
    {
      "epoch": 0.57305690181296,
      "grad_norm": 3.07999849319458,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 6385
    },
    {
      "epoch": 0.5731466523065877,
      "grad_norm": 2.8278822898864746,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 6386
    },
    {
      "epoch": 0.5732364028002154,
      "grad_norm": 2.7598209381103516,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 6387
    },
    {
      "epoch": 0.5733261532938431,
      "grad_norm": 3.274214267730713,
      "learning_rate": 0.0,
      "loss": 0.6616,
      "step": 6388
    },
    {
      "epoch": 0.5734159037874709,
      "grad_norm": 1.899751901626587,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 6389
    },
    {
      "epoch": 0.5735056542810986,
      "grad_norm": 2.5624945163726807,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 6390
    },
    {
      "epoch": 0.5735954047747263,
      "grad_norm": 3.3107082843780518,
      "learning_rate": 0.0,
      "loss": 0.8015,
      "step": 6391
    },
    {
      "epoch": 0.573685155268354,
      "grad_norm": 3.4353880882263184,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 6392
    },
    {
      "epoch": 0.5737749057619816,
      "grad_norm": 2.754840850830078,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 6393
    },
    {
      "epoch": 0.5738646562556095,
      "grad_norm": 3.755596876144409,
      "learning_rate": 0.0,
      "loss": 0.7336,
      "step": 6394
    },
    {
      "epoch": 0.5739544067492371,
      "grad_norm": 2.5567941665649414,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 6395
    },
    {
      "epoch": 0.5740441572428648,
      "grad_norm": 2.347656488418579,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 6396
    },
    {
      "epoch": 0.5741339077364925,
      "grad_norm": 3.669341802597046,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 6397
    },
    {
      "epoch": 0.5742236582301202,
      "grad_norm": 4.221528053283691,
      "learning_rate": 0.0,
      "loss": 0.7819,
      "step": 6398
    },
    {
      "epoch": 0.574313408723748,
      "grad_norm": 2.7649590969085693,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 6399
    },
    {
      "epoch": 0.5744031592173757,
      "grad_norm": 2.8460206985473633,
      "learning_rate": 0.0,
      "loss": 0.7047,
      "step": 6400
    },
    {
      "epoch": 0.5744929097110034,
      "grad_norm": 2.501058340072632,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 6401
    },
    {
      "epoch": 0.5745826602046311,
      "grad_norm": 2.0841405391693115,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 6402
    },
    {
      "epoch": 0.5746724106982588,
      "grad_norm": 2.3641536235809326,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 6403
    },
    {
      "epoch": 0.5747621611918865,
      "grad_norm": 2.5599217414855957,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 6404
    },
    {
      "epoch": 0.5748519116855143,
      "grad_norm": 4.434748649597168,
      "learning_rate": 0.0,
      "loss": 0.6713,
      "step": 6405
    },
    {
      "epoch": 0.574941662179142,
      "grad_norm": 2.543595314025879,
      "learning_rate": 0.0,
      "loss": 0.8322,
      "step": 6406
    },
    {
      "epoch": 0.5750314126727697,
      "grad_norm": 3.144266366958618,
      "learning_rate": 0.0,
      "loss": 0.6806,
      "step": 6407
    },
    {
      "epoch": 0.5751211631663974,
      "grad_norm": 2.6898844242095947,
      "learning_rate": 0.0,
      "loss": 0.7659,
      "step": 6408
    },
    {
      "epoch": 0.5752109136600251,
      "grad_norm": 61.53994369506836,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 6409
    },
    {
      "epoch": 0.5753006641536529,
      "grad_norm": 2.86690354347229,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 6410
    },
    {
      "epoch": 0.5753904146472806,
      "grad_norm": 2.753398895263672,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 6411
    },
    {
      "epoch": 0.5754801651409083,
      "grad_norm": 2.339163064956665,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 6412
    },
    {
      "epoch": 0.575569915634536,
      "grad_norm": 7.151888847351074,
      "learning_rate": 0.0,
      "loss": 0.681,
      "step": 6413
    },
    {
      "epoch": 0.5756596661281637,
      "grad_norm": 2.987740993499756,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 6414
    },
    {
      "epoch": 0.5757494166217915,
      "grad_norm": 3.4919636249542236,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 6415
    },
    {
      "epoch": 0.5758391671154192,
      "grad_norm": 2.6833205223083496,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 6416
    },
    {
      "epoch": 0.5759289176090469,
      "grad_norm": 2.79756498336792,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 6417
    },
    {
      "epoch": 0.5760186681026745,
      "grad_norm": 3.5777828693389893,
      "learning_rate": 0.0,
      "loss": 0.7695,
      "step": 6418
    },
    {
      "epoch": 0.5761084185963022,
      "grad_norm": 3.163081169128418,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 6419
    },
    {
      "epoch": 0.57619816908993,
      "grad_norm": 2.5694265365600586,
      "learning_rate": 0.0,
      "loss": 0.7554,
      "step": 6420
    },
    {
      "epoch": 0.5762879195835577,
      "grad_norm": 3.2760560512542725,
      "learning_rate": 0.0,
      "loss": 0.7003,
      "step": 6421
    },
    {
      "epoch": 0.5763776700771854,
      "grad_norm": 4.081149101257324,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 6422
    },
    {
      "epoch": 0.5764674205708131,
      "grad_norm": 2.7815256118774414,
      "learning_rate": 0.0,
      "loss": 0.7649,
      "step": 6423
    },
    {
      "epoch": 0.5765571710644408,
      "grad_norm": 3.911992073059082,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 6424
    },
    {
      "epoch": 0.5766469215580685,
      "grad_norm": 3.1936755180358887,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 6425
    },
    {
      "epoch": 0.5767366720516963,
      "grad_norm": 2.3054893016815186,
      "learning_rate": 0.0,
      "loss": 0.7076,
      "step": 6426
    },
    {
      "epoch": 0.576826422545324,
      "grad_norm": 2.006126880645752,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 6427
    },
    {
      "epoch": 0.5769161730389517,
      "grad_norm": 2.674222707748413,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 6428
    },
    {
      "epoch": 0.5770059235325794,
      "grad_norm": 2.2990610599517822,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 6429
    },
    {
      "epoch": 0.5770956740262071,
      "grad_norm": 2.3680520057678223,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 6430
    },
    {
      "epoch": 0.5771854245198349,
      "grad_norm": 2.281073570251465,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 6431
    },
    {
      "epoch": 0.5772751750134626,
      "grad_norm": 2.718914270401001,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 6432
    },
    {
      "epoch": 0.5773649255070903,
      "grad_norm": 3.066058874130249,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 6433
    },
    {
      "epoch": 0.577454676000718,
      "grad_norm": 2.596745491027832,
      "learning_rate": 0.0,
      "loss": 0.6458,
      "step": 6434
    },
    {
      "epoch": 0.5775444264943457,
      "grad_norm": 2.95184063911438,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 6435
    },
    {
      "epoch": 0.5776341769879735,
      "grad_norm": 4.609899044036865,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 6436
    },
    {
      "epoch": 0.5777239274816012,
      "grad_norm": 3.2325689792633057,
      "learning_rate": 0.0,
      "loss": 0.6837,
      "step": 6437
    },
    {
      "epoch": 0.5778136779752289,
      "grad_norm": 3.4840915203094482,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 6438
    },
    {
      "epoch": 0.5779034284688566,
      "grad_norm": 8.091032981872559,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 6439
    },
    {
      "epoch": 0.5779931789624843,
      "grad_norm": 2.516083002090454,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 6440
    },
    {
      "epoch": 0.5780829294561121,
      "grad_norm": 3.0060532093048096,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 6441
    },
    {
      "epoch": 0.5781726799497398,
      "grad_norm": 2.496182918548584,
      "learning_rate": 0.0,
      "loss": 0.6704,
      "step": 6442
    },
    {
      "epoch": 0.5782624304433674,
      "grad_norm": 2.9563682079315186,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 6443
    },
    {
      "epoch": 0.5783521809369951,
      "grad_norm": 2.61348557472229,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 6444
    },
    {
      "epoch": 0.5784419314306228,
      "grad_norm": 3.436399459838867,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 6445
    },
    {
      "epoch": 0.5785316819242505,
      "grad_norm": 3.288844108581543,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 6446
    },
    {
      "epoch": 0.5786214324178783,
      "grad_norm": 2.958871603012085,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 6447
    },
    {
      "epoch": 0.578711182911506,
      "grad_norm": 3.213150978088379,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 6448
    },
    {
      "epoch": 0.5788009334051337,
      "grad_norm": 2.707709789276123,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 6449
    },
    {
      "epoch": 0.5788906838987614,
      "grad_norm": 2.3677539825439453,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 6450
    },
    {
      "epoch": 0.5789804343923891,
      "grad_norm": 3.5618274211883545,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 6451
    },
    {
      "epoch": 0.5790701848860169,
      "grad_norm": 3.2228407859802246,
      "learning_rate": 0.0,
      "loss": 0.7602,
      "step": 6452
    },
    {
      "epoch": 0.5791599353796446,
      "grad_norm": 3.470423460006714,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 6453
    },
    {
      "epoch": 0.5792496858732723,
      "grad_norm": 3.5353455543518066,
      "learning_rate": 0.0,
      "loss": 0.7592,
      "step": 6454
    },
    {
      "epoch": 0.5793394363669,
      "grad_norm": 2.5427141189575195,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 6455
    },
    {
      "epoch": 0.5794291868605277,
      "grad_norm": 3.271618127822876,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 6456
    },
    {
      "epoch": 0.5795189373541555,
      "grad_norm": 2.9531307220458984,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 6457
    },
    {
      "epoch": 0.5796086878477832,
      "grad_norm": 2.6808531284332275,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 6458
    },
    {
      "epoch": 0.5796984383414109,
      "grad_norm": 15.071293830871582,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 6459
    },
    {
      "epoch": 0.5797881888350386,
      "grad_norm": 3.7371091842651367,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 6460
    },
    {
      "epoch": 0.5798779393286663,
      "grad_norm": 4.3813676834106445,
      "learning_rate": 0.0,
      "loss": 0.6882,
      "step": 6461
    },
    {
      "epoch": 0.5799676898222941,
      "grad_norm": 3.325131416320801,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 6462
    },
    {
      "epoch": 0.5800574403159218,
      "grad_norm": 3.2448489665985107,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 6463
    },
    {
      "epoch": 0.5801471908095495,
      "grad_norm": 3.739025831222534,
      "learning_rate": 0.0,
      "loss": 0.7546,
      "step": 6464
    },
    {
      "epoch": 0.5802369413031772,
      "grad_norm": 3.3921501636505127,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 6465
    },
    {
      "epoch": 0.5803266917968048,
      "grad_norm": 4.403278827667236,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 6466
    },
    {
      "epoch": 0.5804164422904325,
      "grad_norm": 2.713456153869629,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 6467
    },
    {
      "epoch": 0.5805061927840603,
      "grad_norm": 2.567330837249756,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 6468
    },
    {
      "epoch": 0.580595943277688,
      "grad_norm": 3.3365771770477295,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 6469
    },
    {
      "epoch": 0.5806856937713157,
      "grad_norm": 2.8687644004821777,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 6470
    },
    {
      "epoch": 0.5807754442649434,
      "grad_norm": 2.127187490463257,
      "learning_rate": 0.0,
      "loss": 0.7692,
      "step": 6471
    },
    {
      "epoch": 0.5808651947585711,
      "grad_norm": 5.233154773712158,
      "learning_rate": 0.0,
      "loss": 0.6843,
      "step": 6472
    },
    {
      "epoch": 0.5809549452521989,
      "grad_norm": 2.3844492435455322,
      "learning_rate": 0.0,
      "loss": 0.6747,
      "step": 6473
    },
    {
      "epoch": 0.5810446957458266,
      "grad_norm": 2.763850450515747,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 6474
    },
    {
      "epoch": 0.5811344462394543,
      "grad_norm": 2.8102548122406006,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 6475
    },
    {
      "epoch": 0.581224196733082,
      "grad_norm": 2.209425449371338,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 6476
    },
    {
      "epoch": 0.5813139472267097,
      "grad_norm": 2.4225552082061768,
      "learning_rate": 0.0,
      "loss": 0.6629,
      "step": 6477
    },
    {
      "epoch": 0.5814036977203375,
      "grad_norm": 2.598889112472534,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 6478
    },
    {
      "epoch": 0.5814934482139652,
      "grad_norm": 2.403890609741211,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 6479
    },
    {
      "epoch": 0.5815831987075929,
      "grad_norm": 2.2466328144073486,
      "learning_rate": 0.0,
      "loss": 0.688,
      "step": 6480
    },
    {
      "epoch": 0.5816729492012206,
      "grad_norm": 2.177940845489502,
      "learning_rate": 0.0,
      "loss": 0.7816,
      "step": 6481
    },
    {
      "epoch": 0.5817626996948483,
      "grad_norm": 2.0680789947509766,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 6482
    },
    {
      "epoch": 0.5818524501884761,
      "grad_norm": 2.4831204414367676,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 6483
    },
    {
      "epoch": 0.5819422006821038,
      "grad_norm": 3.440770149230957,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 6484
    },
    {
      "epoch": 0.5820319511757315,
      "grad_norm": 3.8183910846710205,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 6485
    },
    {
      "epoch": 0.5821217016693592,
      "grad_norm": 2.8432300090789795,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 6486
    },
    {
      "epoch": 0.5822114521629869,
      "grad_norm": 3.5051276683807373,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 6487
    },
    {
      "epoch": 0.5823012026566146,
      "grad_norm": 2.404533624649048,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 6488
    },
    {
      "epoch": 0.5823909531502424,
      "grad_norm": 3.5708401203155518,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 6489
    },
    {
      "epoch": 0.5824807036438701,
      "grad_norm": 2.962615489959717,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 6490
    },
    {
      "epoch": 0.5825704541374978,
      "grad_norm": 2.6113016605377197,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 6491
    },
    {
      "epoch": 0.5826602046311254,
      "grad_norm": 2.7589738368988037,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 6492
    },
    {
      "epoch": 0.5827499551247531,
      "grad_norm": 3.9672350883483887,
      "learning_rate": 0.0,
      "loss": 0.6612,
      "step": 6493
    },
    {
      "epoch": 0.5828397056183809,
      "grad_norm": 2.9537813663482666,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 6494
    },
    {
      "epoch": 0.5829294561120086,
      "grad_norm": 2.316772937774658,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 6495
    },
    {
      "epoch": 0.5830192066056363,
      "grad_norm": 3.243717670440674,
      "learning_rate": 0.0,
      "loss": 0.6881,
      "step": 6496
    },
    {
      "epoch": 0.583108957099264,
      "grad_norm": 2.969177007675171,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 6497
    },
    {
      "epoch": 0.5831987075928917,
      "grad_norm": 2.4243245124816895,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 6498
    },
    {
      "epoch": 0.5832884580865195,
      "grad_norm": 2.8848624229431152,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 6499
    },
    {
      "epoch": 0.5833782085801472,
      "grad_norm": 3.1991631984710693,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 6500
    },
    {
      "epoch": 0.5834679590737749,
      "grad_norm": 2.7561023235321045,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 6501
    },
    {
      "epoch": 0.5835577095674026,
      "grad_norm": 2.4783542156219482,
      "learning_rate": 0.0,
      "loss": 0.7141,
      "step": 6502
    },
    {
      "epoch": 0.5836474600610303,
      "grad_norm": 3.8842856884002686,
      "learning_rate": 0.0,
      "loss": 0.7074,
      "step": 6503
    },
    {
      "epoch": 0.5837372105546581,
      "grad_norm": 2.7387819290161133,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 6504
    },
    {
      "epoch": 0.5838269610482858,
      "grad_norm": 2.953795909881592,
      "learning_rate": 0.0,
      "loss": 0.78,
      "step": 6505
    },
    {
      "epoch": 0.5839167115419135,
      "grad_norm": 2.778183698654175,
      "learning_rate": 0.0,
      "loss": 0.6555,
      "step": 6506
    },
    {
      "epoch": 0.5840064620355412,
      "grad_norm": 3.46077561378479,
      "learning_rate": 0.0,
      "loss": 0.6919,
      "step": 6507
    },
    {
      "epoch": 0.5840962125291689,
      "grad_norm": 2.894125461578369,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 6508
    },
    {
      "epoch": 0.5841859630227966,
      "grad_norm": 3.7386910915374756,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 6509
    },
    {
      "epoch": 0.5842757135164244,
      "grad_norm": 2.2845163345336914,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 6510
    },
    {
      "epoch": 0.5843654640100521,
      "grad_norm": 3.4546923637390137,
      "learning_rate": 0.0,
      "loss": 0.6773,
      "step": 6511
    },
    {
      "epoch": 0.5844552145036798,
      "grad_norm": 2.797370195388794,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 6512
    },
    {
      "epoch": 0.5845449649973075,
      "grad_norm": 2.7526652812957764,
      "learning_rate": 0.0,
      "loss": 0.8047,
      "step": 6513
    },
    {
      "epoch": 0.5846347154909352,
      "grad_norm": 2.786691665649414,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 6514
    },
    {
      "epoch": 0.584724465984563,
      "grad_norm": 3.3940680027008057,
      "learning_rate": 0.0,
      "loss": 0.8085,
      "step": 6515
    },
    {
      "epoch": 0.5848142164781907,
      "grad_norm": 3.252929210662842,
      "learning_rate": 0.0,
      "loss": 0.6872,
      "step": 6516
    },
    {
      "epoch": 0.5849039669718183,
      "grad_norm": 2.40166974067688,
      "learning_rate": 0.0,
      "loss": 0.6629,
      "step": 6517
    },
    {
      "epoch": 0.584993717465446,
      "grad_norm": 2.927630662918091,
      "learning_rate": 0.0,
      "loss": 0.7458,
      "step": 6518
    },
    {
      "epoch": 0.5850834679590737,
      "grad_norm": 2.5396063327789307,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 6519
    },
    {
      "epoch": 0.5851732184527015,
      "grad_norm": 4.998879909515381,
      "learning_rate": 0.0,
      "loss": 0.8073,
      "step": 6520
    },
    {
      "epoch": 0.5852629689463292,
      "grad_norm": 2.877700090408325,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 6521
    },
    {
      "epoch": 0.5853527194399569,
      "grad_norm": 2.677152633666992,
      "learning_rate": 0.0,
      "loss": 0.747,
      "step": 6522
    },
    {
      "epoch": 0.5854424699335846,
      "grad_norm": 2.7027361392974854,
      "learning_rate": 0.0,
      "loss": 0.7774,
      "step": 6523
    },
    {
      "epoch": 0.5855322204272123,
      "grad_norm": 3.0870161056518555,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 6524
    },
    {
      "epoch": 0.5856219709208401,
      "grad_norm": 2.475947856903076,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 6525
    },
    {
      "epoch": 0.5857117214144678,
      "grad_norm": 2.611591339111328,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 6526
    },
    {
      "epoch": 0.5858014719080955,
      "grad_norm": 2.4370815753936768,
      "learning_rate": 0.0,
      "loss": 0.7739,
      "step": 6527
    },
    {
      "epoch": 0.5858912224017232,
      "grad_norm": 3.2388715744018555,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 6528
    },
    {
      "epoch": 0.5859809728953509,
      "grad_norm": 3.1072282791137695,
      "learning_rate": 0.0,
      "loss": 0.7069,
      "step": 6529
    },
    {
      "epoch": 0.5860707233889786,
      "grad_norm": 3.433907985687256,
      "learning_rate": 0.0,
      "loss": 0.677,
      "step": 6530
    },
    {
      "epoch": 0.5861604738826064,
      "grad_norm": 2.4952080249786377,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 6531
    },
    {
      "epoch": 0.5862502243762341,
      "grad_norm": 2.5494306087493896,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 6532
    },
    {
      "epoch": 0.5863399748698618,
      "grad_norm": 2.778538465499878,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 6533
    },
    {
      "epoch": 0.5864297253634895,
      "grad_norm": 2.8098533153533936,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 6534
    },
    {
      "epoch": 0.5865194758571172,
      "grad_norm": 2.76122784614563,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 6535
    },
    {
      "epoch": 0.586609226350745,
      "grad_norm": 2.474318027496338,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 6536
    },
    {
      "epoch": 0.5866989768443727,
      "grad_norm": 2.4361178874969482,
      "learning_rate": 0.0,
      "loss": 0.7573,
      "step": 6537
    },
    {
      "epoch": 0.5867887273380004,
      "grad_norm": 2.339198112487793,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 6538
    },
    {
      "epoch": 0.586878477831628,
      "grad_norm": 2.822927951812744,
      "learning_rate": 0.0,
      "loss": 0.7141,
      "step": 6539
    },
    {
      "epoch": 0.5869682283252557,
      "grad_norm": 2.861532688140869,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 6540
    },
    {
      "epoch": 0.5870579788188836,
      "grad_norm": 2.396162748336792,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 6541
    },
    {
      "epoch": 0.5871477293125112,
      "grad_norm": 2.7721147537231445,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 6542
    },
    {
      "epoch": 0.5872374798061389,
      "grad_norm": 2.766667127609253,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 6543
    },
    {
      "epoch": 0.5873272302997666,
      "grad_norm": 2.4883575439453125,
      "learning_rate": 0.0,
      "loss": 0.667,
      "step": 6544
    },
    {
      "epoch": 0.5874169807933943,
      "grad_norm": 5.130420207977295,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 6545
    },
    {
      "epoch": 0.5875067312870221,
      "grad_norm": 3.5254642963409424,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 6546
    },
    {
      "epoch": 0.5875964817806498,
      "grad_norm": 2.7964518070220947,
      "learning_rate": 0.0,
      "loss": 0.7319,
      "step": 6547
    },
    {
      "epoch": 0.5876862322742775,
      "grad_norm": 3.1757161617279053,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 6548
    },
    {
      "epoch": 0.5877759827679052,
      "grad_norm": 2.980761766433716,
      "learning_rate": 0.0,
      "loss": 0.6666,
      "step": 6549
    },
    {
      "epoch": 0.5878657332615329,
      "grad_norm": 3.4716053009033203,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 6550
    },
    {
      "epoch": 0.5879554837551606,
      "grad_norm": 2.516024112701416,
      "learning_rate": 0.0,
      "loss": 0.7021,
      "step": 6551
    },
    {
      "epoch": 0.5880452342487884,
      "grad_norm": 2.801241397857666,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 6552
    },
    {
      "epoch": 0.5881349847424161,
      "grad_norm": 3.246217966079712,
      "learning_rate": 0.0,
      "loss": 0.7542,
      "step": 6553
    },
    {
      "epoch": 0.5882247352360438,
      "grad_norm": 2.7877252101898193,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 6554
    },
    {
      "epoch": 0.5883144857296715,
      "grad_norm": 2.6649534702301025,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 6555
    },
    {
      "epoch": 0.5884042362232992,
      "grad_norm": 4.361511707305908,
      "learning_rate": 0.0,
      "loss": 0.6848,
      "step": 6556
    },
    {
      "epoch": 0.588493986716927,
      "grad_norm": 3.277029514312744,
      "learning_rate": 0.0,
      "loss": 0.7124,
      "step": 6557
    },
    {
      "epoch": 0.5885837372105547,
      "grad_norm": 6.089234352111816,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 6558
    },
    {
      "epoch": 0.5886734877041824,
      "grad_norm": 4.085573196411133,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 6559
    },
    {
      "epoch": 0.5887632381978101,
      "grad_norm": 3.5702176094055176,
      "learning_rate": 0.0,
      "loss": 0.7302,
      "step": 6560
    },
    {
      "epoch": 0.5888529886914378,
      "grad_norm": 2.8258585929870605,
      "learning_rate": 0.0,
      "loss": 0.7729,
      "step": 6561
    },
    {
      "epoch": 0.5889427391850656,
      "grad_norm": 2.4999613761901855,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 6562
    },
    {
      "epoch": 0.5890324896786933,
      "grad_norm": 3.95841121673584,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 6563
    },
    {
      "epoch": 0.589122240172321,
      "grad_norm": 2.871744394302368,
      "learning_rate": 0.0,
      "loss": 0.665,
      "step": 6564
    },
    {
      "epoch": 0.5892119906659486,
      "grad_norm": 3.8606128692626953,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 6565
    },
    {
      "epoch": 0.5893017411595763,
      "grad_norm": 3.602898120880127,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 6566
    },
    {
      "epoch": 0.5893914916532041,
      "grad_norm": 3.1120734214782715,
      "learning_rate": 0.0,
      "loss": 0.6826,
      "step": 6567
    },
    {
      "epoch": 0.5894812421468318,
      "grad_norm": 4.404734134674072,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 6568
    },
    {
      "epoch": 0.5895709926404595,
      "grad_norm": 2.3831892013549805,
      "learning_rate": 0.0,
      "loss": 0.8375,
      "step": 6569
    },
    {
      "epoch": 0.5896607431340872,
      "grad_norm": 2.6838889122009277,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 6570
    },
    {
      "epoch": 0.5897504936277149,
      "grad_norm": 2.8848233222961426,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 6571
    },
    {
      "epoch": 0.5898402441213426,
      "grad_norm": 3.1662235260009766,
      "learning_rate": 0.0,
      "loss": 0.6958,
      "step": 6572
    },
    {
      "epoch": 0.5899299946149704,
      "grad_norm": 3.3645153045654297,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 6573
    },
    {
      "epoch": 0.5900197451085981,
      "grad_norm": 2.773322343826294,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 6574
    },
    {
      "epoch": 0.5901094956022258,
      "grad_norm": 7.958773136138916,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 6575
    },
    {
      "epoch": 0.5901992460958535,
      "grad_norm": 2.5207860469818115,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 6576
    },
    {
      "epoch": 0.5902889965894812,
      "grad_norm": 4.013160705566406,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 6577
    },
    {
      "epoch": 0.590378747083109,
      "grad_norm": 3.5327274799346924,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 6578
    },
    {
      "epoch": 0.5904684975767367,
      "grad_norm": 3.8615260124206543,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 6579
    },
    {
      "epoch": 0.5905582480703644,
      "grad_norm": 2.3533010482788086,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 6580
    },
    {
      "epoch": 0.5906479985639921,
      "grad_norm": 3.2353270053863525,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 6581
    },
    {
      "epoch": 0.5907377490576198,
      "grad_norm": 4.14102840423584,
      "learning_rate": 0.0,
      "loss": 0.7715,
      "step": 6582
    },
    {
      "epoch": 0.5908274995512476,
      "grad_norm": 5.191098213195801,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 6583
    },
    {
      "epoch": 0.5909172500448753,
      "grad_norm": 2.596719264984131,
      "learning_rate": 0.0,
      "loss": 0.7093,
      "step": 6584
    },
    {
      "epoch": 0.591007000538503,
      "grad_norm": 2.6252174377441406,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 6585
    },
    {
      "epoch": 0.5910967510321307,
      "grad_norm": 5.299757480621338,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 6586
    },
    {
      "epoch": 0.5911865015257584,
      "grad_norm": 2.9068892002105713,
      "learning_rate": 0.0,
      "loss": 0.6737,
      "step": 6587
    },
    {
      "epoch": 0.5912762520193862,
      "grad_norm": 3.732154369354248,
      "learning_rate": 0.0,
      "loss": 0.7302,
      "step": 6588
    },
    {
      "epoch": 0.5913660025130139,
      "grad_norm": 4.602272987365723,
      "learning_rate": 0.0,
      "loss": 0.7274,
      "step": 6589
    },
    {
      "epoch": 0.5914557530066415,
      "grad_norm": 2.572672128677368,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 6590
    },
    {
      "epoch": 0.5915455035002692,
      "grad_norm": 2.6054153442382812,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 6591
    },
    {
      "epoch": 0.5916352539938969,
      "grad_norm": 2.59419322013855,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 6592
    },
    {
      "epoch": 0.5917250044875246,
      "grad_norm": 2.5131711959838867,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 6593
    },
    {
      "epoch": 0.5918147549811524,
      "grad_norm": 3.4552721977233887,
      "learning_rate": 0.0,
      "loss": 0.7642,
      "step": 6594
    },
    {
      "epoch": 0.5919045054747801,
      "grad_norm": 2.687666416168213,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 6595
    },
    {
      "epoch": 0.5919942559684078,
      "grad_norm": 2.634873867034912,
      "learning_rate": 0.0,
      "loss": 0.7388,
      "step": 6596
    },
    {
      "epoch": 0.5920840064620355,
      "grad_norm": 3.25321364402771,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 6597
    },
    {
      "epoch": 0.5921737569556632,
      "grad_norm": 2.9790737628936768,
      "learning_rate": 0.0,
      "loss": 0.7657,
      "step": 6598
    },
    {
      "epoch": 0.592263507449291,
      "grad_norm": 3.26918625831604,
      "learning_rate": 0.0,
      "loss": 0.686,
      "step": 6599
    },
    {
      "epoch": 0.5923532579429187,
      "grad_norm": 3.3040833473205566,
      "learning_rate": 0.0,
      "loss": 0.7822,
      "step": 6600
    },
    {
      "epoch": 0.5924430084365464,
      "grad_norm": 2.7438673973083496,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 6601
    },
    {
      "epoch": 0.5925327589301741,
      "grad_norm": 3.246891975402832,
      "learning_rate": 0.0,
      "loss": 0.6723,
      "step": 6602
    },
    {
      "epoch": 0.5926225094238018,
      "grad_norm": 2.6536998748779297,
      "learning_rate": 0.0,
      "loss": 0.7081,
      "step": 6603
    },
    {
      "epoch": 0.5927122599174296,
      "grad_norm": 3.003932476043701,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 6604
    },
    {
      "epoch": 0.5928020104110573,
      "grad_norm": 3.375438690185547,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 6605
    },
    {
      "epoch": 0.592891760904685,
      "grad_norm": 8.905045509338379,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 6606
    },
    {
      "epoch": 0.5929815113983127,
      "grad_norm": 2.4163014888763428,
      "learning_rate": 0.0,
      "loss": 0.7701,
      "step": 6607
    },
    {
      "epoch": 0.5930712618919404,
      "grad_norm": 2.629364490509033,
      "learning_rate": 0.0,
      "loss": 0.6687,
      "step": 6608
    },
    {
      "epoch": 0.5931610123855682,
      "grad_norm": 2.9083802700042725,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 6609
    },
    {
      "epoch": 0.5932507628791959,
      "grad_norm": 4.6430768966674805,
      "learning_rate": 0.0,
      "loss": 0.697,
      "step": 6610
    },
    {
      "epoch": 0.5933405133728236,
      "grad_norm": 2.8119349479675293,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 6611
    },
    {
      "epoch": 0.5934302638664513,
      "grad_norm": 2.8806376457214355,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 6612
    },
    {
      "epoch": 0.593520014360079,
      "grad_norm": 2.5014965534210205,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 6613
    },
    {
      "epoch": 0.5936097648537066,
      "grad_norm": 3.6152966022491455,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 6614
    },
    {
      "epoch": 0.5936995153473344,
      "grad_norm": 2.8749289512634277,
      "learning_rate": 0.0,
      "loss": 0.8553,
      "step": 6615
    },
    {
      "epoch": 0.5937892658409621,
      "grad_norm": 2.421463966369629,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 6616
    },
    {
      "epoch": 0.5938790163345898,
      "grad_norm": 3.53794002532959,
      "learning_rate": 0.0,
      "loss": 0.7411,
      "step": 6617
    },
    {
      "epoch": 0.5939687668282175,
      "grad_norm": 2.5823559761047363,
      "learning_rate": 0.0,
      "loss": 0.7006,
      "step": 6618
    },
    {
      "epoch": 0.5940585173218452,
      "grad_norm": 3.204232692718506,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 6619
    },
    {
      "epoch": 0.594148267815473,
      "grad_norm": 3.254847288131714,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 6620
    },
    {
      "epoch": 0.5942380183091007,
      "grad_norm": 4.651917457580566,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 6621
    },
    {
      "epoch": 0.5943277688027284,
      "grad_norm": 3.1119046211242676,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 6622
    },
    {
      "epoch": 0.5944175192963561,
      "grad_norm": 3.0531296730041504,
      "learning_rate": 0.0,
      "loss": 0.6733,
      "step": 6623
    },
    {
      "epoch": 0.5945072697899838,
      "grad_norm": 4.2307305335998535,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 6624
    },
    {
      "epoch": 0.5945970202836116,
      "grad_norm": 4.415794372558594,
      "learning_rate": 0.0,
      "loss": 0.8163,
      "step": 6625
    },
    {
      "epoch": 0.5946867707772393,
      "grad_norm": 3.1268861293792725,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 6626
    },
    {
      "epoch": 0.594776521270867,
      "grad_norm": 2.758089780807495,
      "learning_rate": 0.0,
      "loss": 0.8434,
      "step": 6627
    },
    {
      "epoch": 0.5948662717644947,
      "grad_norm": 3.3602280616760254,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 6628
    },
    {
      "epoch": 0.5949560222581224,
      "grad_norm": 3.431572198867798,
      "learning_rate": 0.0,
      "loss": 0.6705,
      "step": 6629
    },
    {
      "epoch": 0.5950457727517501,
      "grad_norm": 2.751168966293335,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 6630
    },
    {
      "epoch": 0.5951355232453779,
      "grad_norm": 2.7438459396362305,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 6631
    },
    {
      "epoch": 0.5952252737390056,
      "grad_norm": 3.197706937789917,
      "learning_rate": 0.0,
      "loss": 0.7026,
      "step": 6632
    },
    {
      "epoch": 0.5953150242326333,
      "grad_norm": 2.4694058895111084,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 6633
    },
    {
      "epoch": 0.595404774726261,
      "grad_norm": 2.5036380290985107,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 6634
    },
    {
      "epoch": 0.5954945252198887,
      "grad_norm": 2.6223220825195312,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 6635
    },
    {
      "epoch": 0.5955842757135165,
      "grad_norm": 2.227471351623535,
      "learning_rate": 0.0,
      "loss": 0.7677,
      "step": 6636
    },
    {
      "epoch": 0.5956740262071442,
      "grad_norm": 3.270280599594116,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 6637
    },
    {
      "epoch": 0.5957637767007719,
      "grad_norm": 2.931396007537842,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 6638
    },
    {
      "epoch": 0.5958535271943995,
      "grad_norm": 2.9552721977233887,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 6639
    },
    {
      "epoch": 0.5959432776880272,
      "grad_norm": 2.3702547550201416,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 6640
    },
    {
      "epoch": 0.596033028181655,
      "grad_norm": 2.703803777694702,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 6641
    },
    {
      "epoch": 0.5961227786752827,
      "grad_norm": 2.9264159202575684,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 6642
    },
    {
      "epoch": 0.5962125291689104,
      "grad_norm": 3.5340576171875,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 6643
    },
    {
      "epoch": 0.5963022796625381,
      "grad_norm": 2.780132532119751,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 6644
    },
    {
      "epoch": 0.5963920301561658,
      "grad_norm": 2.1968488693237305,
      "learning_rate": 0.0,
      "loss": 0.7863,
      "step": 6645
    },
    {
      "epoch": 0.5964817806497936,
      "grad_norm": 2.156794309616089,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 6646
    },
    {
      "epoch": 0.5965715311434213,
      "grad_norm": 2.6637990474700928,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 6647
    },
    {
      "epoch": 0.596661281637049,
      "grad_norm": 2.471681833267212,
      "learning_rate": 0.0,
      "loss": 0.6896,
      "step": 6648
    },
    {
      "epoch": 0.5967510321306767,
      "grad_norm": 2.5716073513031006,
      "learning_rate": 0.0,
      "loss": 0.7773,
      "step": 6649
    },
    {
      "epoch": 0.5968407826243044,
      "grad_norm": 3.843644618988037,
      "learning_rate": 0.0,
      "loss": 0.787,
      "step": 6650
    },
    {
      "epoch": 0.5969305331179321,
      "grad_norm": 2.725660562515259,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 6651
    },
    {
      "epoch": 0.5970202836115599,
      "grad_norm": 2.807435989379883,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 6652
    },
    {
      "epoch": 0.5971100341051876,
      "grad_norm": 2.85347843170166,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 6653
    },
    {
      "epoch": 0.5971997845988153,
      "grad_norm": 3.5942227840423584,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 6654
    },
    {
      "epoch": 0.597289535092443,
      "grad_norm": 2.5594189167022705,
      "learning_rate": 0.0,
      "loss": 0.7677,
      "step": 6655
    },
    {
      "epoch": 0.5973792855860707,
      "grad_norm": 4.216099262237549,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 6656
    },
    {
      "epoch": 0.5974690360796985,
      "grad_norm": 2.9115705490112305,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 6657
    },
    {
      "epoch": 0.5975587865733262,
      "grad_norm": 2.5873730182647705,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 6658
    },
    {
      "epoch": 0.5976485370669539,
      "grad_norm": 2.743262529373169,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 6659
    },
    {
      "epoch": 0.5977382875605816,
      "grad_norm": 2.2710793018341064,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 6660
    },
    {
      "epoch": 0.5978280380542093,
      "grad_norm": 2.801774740219116,
      "learning_rate": 0.0,
      "loss": 0.7682,
      "step": 6661
    },
    {
      "epoch": 0.5979177885478371,
      "grad_norm": 3.5413925647735596,
      "learning_rate": 0.0,
      "loss": 0.8221,
      "step": 6662
    },
    {
      "epoch": 0.5980075390414648,
      "grad_norm": 2.6169474124908447,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 6663
    },
    {
      "epoch": 0.5980972895350924,
      "grad_norm": 2.5025932788848877,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 6664
    },
    {
      "epoch": 0.5981870400287201,
      "grad_norm": 2.740645170211792,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 6665
    },
    {
      "epoch": 0.5982767905223478,
      "grad_norm": 3.3753576278686523,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 6666
    },
    {
      "epoch": 0.5983665410159756,
      "grad_norm": 2.376281261444092,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 6667
    },
    {
      "epoch": 0.5984562915096033,
      "grad_norm": 2.8382883071899414,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 6668
    },
    {
      "epoch": 0.598546042003231,
      "grad_norm": 2.5278563499450684,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 6669
    },
    {
      "epoch": 0.5986357924968587,
      "grad_norm": 2.9783315658569336,
      "learning_rate": 0.0,
      "loss": 0.7813,
      "step": 6670
    },
    {
      "epoch": 0.5987255429904864,
      "grad_norm": 2.68961238861084,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 6671
    },
    {
      "epoch": 0.5988152934841141,
      "grad_norm": 2.7778849601745605,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 6672
    },
    {
      "epoch": 0.5989050439777419,
      "grad_norm": 2.2228894233703613,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 6673
    },
    {
      "epoch": 0.5989947944713696,
      "grad_norm": 2.9993059635162354,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 6674
    },
    {
      "epoch": 0.5990845449649973,
      "grad_norm": 3.728689670562744,
      "learning_rate": 0.0,
      "loss": 0.7061,
      "step": 6675
    },
    {
      "epoch": 0.599174295458625,
      "grad_norm": 3.112302303314209,
      "learning_rate": 0.0,
      "loss": 0.693,
      "step": 6676
    },
    {
      "epoch": 0.5992640459522527,
      "grad_norm": 3.3248753547668457,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 6677
    },
    {
      "epoch": 0.5993537964458805,
      "grad_norm": 2.6433358192443848,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 6678
    },
    {
      "epoch": 0.5994435469395082,
      "grad_norm": 2.879507541656494,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 6679
    },
    {
      "epoch": 0.5995332974331359,
      "grad_norm": 4.322277545928955,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 6680
    },
    {
      "epoch": 0.5996230479267636,
      "grad_norm": 2.9399962425231934,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 6681
    },
    {
      "epoch": 0.5997127984203913,
      "grad_norm": 3.240422248840332,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 6682
    },
    {
      "epoch": 0.5998025489140191,
      "grad_norm": 3.3920586109161377,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 6683
    },
    {
      "epoch": 0.5998922994076468,
      "grad_norm": 3.6971559524536133,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 6684
    },
    {
      "epoch": 0.5999820499012745,
      "grad_norm": 2.582751512527466,
      "learning_rate": 0.0,
      "loss": 0.6872,
      "step": 6685
    },
    {
      "epoch": 0.6000718003949022,
      "grad_norm": 2.684087038040161,
      "learning_rate": 0.0,
      "loss": 0.7891,
      "step": 6686
    },
    {
      "epoch": 0.6001615508885298,
      "grad_norm": 4.373727321624756,
      "learning_rate": 0.0,
      "loss": 0.7393,
      "step": 6687
    },
    {
      "epoch": 0.6002513013821577,
      "grad_norm": 2.3595287799835205,
      "learning_rate": 0.0,
      "loss": 0.6577,
      "step": 6688
    },
    {
      "epoch": 0.6003410518757853,
      "grad_norm": 3.209247350692749,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 6689
    },
    {
      "epoch": 0.600430802369413,
      "grad_norm": 2.8131964206695557,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 6690
    },
    {
      "epoch": 0.6005205528630407,
      "grad_norm": 2.259828567504883,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 6691
    },
    {
      "epoch": 0.6006103033566684,
      "grad_norm": 2.6785027980804443,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 6692
    },
    {
      "epoch": 0.6007000538502961,
      "grad_norm": 3.3088490962982178,
      "learning_rate": 0.0,
      "loss": 0.7365,
      "step": 6693
    },
    {
      "epoch": 0.6007898043439239,
      "grad_norm": 2.9060323238372803,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 6694
    },
    {
      "epoch": 0.6008795548375516,
      "grad_norm": 3.252208948135376,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 6695
    },
    {
      "epoch": 0.6009693053311793,
      "grad_norm": 3.5500380992889404,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 6696
    },
    {
      "epoch": 0.601059055824807,
      "grad_norm": 30.489715576171875,
      "learning_rate": 0.0,
      "loss": 0.6738,
      "step": 6697
    },
    {
      "epoch": 0.6011488063184347,
      "grad_norm": 3.587110996246338,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 6698
    },
    {
      "epoch": 0.6012385568120625,
      "grad_norm": 3.7027618885040283,
      "learning_rate": 0.0,
      "loss": 0.7756,
      "step": 6699
    },
    {
      "epoch": 0.6013283073056902,
      "grad_norm": 3.344331741333008,
      "learning_rate": 0.0,
      "loss": 0.793,
      "step": 6700
    },
    {
      "epoch": 0.6014180577993179,
      "grad_norm": 2.5151166915893555,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 6701
    },
    {
      "epoch": 0.6015078082929456,
      "grad_norm": 4.045135498046875,
      "learning_rate": 0.0,
      "loss": 0.6789,
      "step": 6702
    },
    {
      "epoch": 0.6015975587865733,
      "grad_norm": 2.8798086643218994,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 6703
    },
    {
      "epoch": 0.6016873092802011,
      "grad_norm": 4.062810897827148,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 6704
    },
    {
      "epoch": 0.6017770597738288,
      "grad_norm": 7.92051887512207,
      "learning_rate": 0.0,
      "loss": 0.7431,
      "step": 6705
    },
    {
      "epoch": 0.6018668102674565,
      "grad_norm": 3.5556273460388184,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 6706
    },
    {
      "epoch": 0.6019565607610842,
      "grad_norm": 2.8973262310028076,
      "learning_rate": 0.0,
      "loss": 0.7979,
      "step": 6707
    },
    {
      "epoch": 0.6020463112547119,
      "grad_norm": 3.5627224445343018,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 6708
    },
    {
      "epoch": 0.6021360617483397,
      "grad_norm": 2.9106037616729736,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 6709
    },
    {
      "epoch": 0.6022258122419674,
      "grad_norm": 2.7758686542510986,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 6710
    },
    {
      "epoch": 0.6023155627355951,
      "grad_norm": 3.630706548690796,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 6711
    },
    {
      "epoch": 0.6024053132292227,
      "grad_norm": 4.51489782333374,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 6712
    },
    {
      "epoch": 0.6024950637228504,
      "grad_norm": 3.0623817443847656,
      "learning_rate": 0.0,
      "loss": 0.6859,
      "step": 6713
    },
    {
      "epoch": 0.6025848142164781,
      "grad_norm": 2.5996696949005127,
      "learning_rate": 0.0,
      "loss": 0.7606,
      "step": 6714
    },
    {
      "epoch": 0.6026745647101059,
      "grad_norm": 3.895840644836426,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 6715
    },
    {
      "epoch": 0.6027643152037336,
      "grad_norm": 4.9076828956604,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 6716
    },
    {
      "epoch": 0.6028540656973613,
      "grad_norm": 3.59494686126709,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 6717
    },
    {
      "epoch": 0.602943816190989,
      "grad_norm": 2.7337985038757324,
      "learning_rate": 0.0,
      "loss": 0.7594,
      "step": 6718
    },
    {
      "epoch": 0.6030335666846167,
      "grad_norm": 3.3109130859375,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 6719
    },
    {
      "epoch": 0.6031233171782445,
      "grad_norm": 4.387004375457764,
      "learning_rate": 0.0,
      "loss": 0.7047,
      "step": 6720
    },
    {
      "epoch": 0.6032130676718722,
      "grad_norm": 2.2784676551818848,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 6721
    },
    {
      "epoch": 0.6033028181654999,
      "grad_norm": 2.7335150241851807,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 6722
    },
    {
      "epoch": 0.6033925686591276,
      "grad_norm": 3.5345239639282227,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 6723
    },
    {
      "epoch": 0.6034823191527553,
      "grad_norm": 3.7375223636627197,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 6724
    },
    {
      "epoch": 0.6035720696463831,
      "grad_norm": 3.110840082168579,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 6725
    },
    {
      "epoch": 0.6036618201400108,
      "grad_norm": 3.442918062210083,
      "learning_rate": 0.0,
      "loss": 0.7432,
      "step": 6726
    },
    {
      "epoch": 0.6037515706336385,
      "grad_norm": 3.0991547107696533,
      "learning_rate": 0.0,
      "loss": 0.7538,
      "step": 6727
    },
    {
      "epoch": 0.6038413211272662,
      "grad_norm": 3.561635732650757,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 6728
    },
    {
      "epoch": 0.6039310716208939,
      "grad_norm": 3.196265697479248,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 6729
    },
    {
      "epoch": 0.6040208221145217,
      "grad_norm": 2.617546319961548,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 6730
    },
    {
      "epoch": 0.6041105726081494,
      "grad_norm": 2.527561664581299,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 6731
    },
    {
      "epoch": 0.6042003231017771,
      "grad_norm": 2.356989860534668,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 6732
    },
    {
      "epoch": 0.6042900735954048,
      "grad_norm": 3.8690710067749023,
      "learning_rate": 0.0,
      "loss": 0.7057,
      "step": 6733
    },
    {
      "epoch": 0.6043798240890325,
      "grad_norm": 2.9691176414489746,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 6734
    },
    {
      "epoch": 0.6044695745826602,
      "grad_norm": 2.722073793411255,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 6735
    },
    {
      "epoch": 0.604559325076288,
      "grad_norm": 2.5753285884857178,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 6736
    },
    {
      "epoch": 0.6046490755699157,
      "grad_norm": 6.5374345779418945,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 6737
    },
    {
      "epoch": 0.6047388260635433,
      "grad_norm": 2.9130940437316895,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 6738
    },
    {
      "epoch": 0.604828576557171,
      "grad_norm": 3.5435400009155273,
      "learning_rate": 0.0,
      "loss": 0.7957,
      "step": 6739
    },
    {
      "epoch": 0.6049183270507987,
      "grad_norm": 2.7809057235717773,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 6740
    },
    {
      "epoch": 0.6050080775444265,
      "grad_norm": 2.693814754486084,
      "learning_rate": 0.0,
      "loss": 0.758,
      "step": 6741
    },
    {
      "epoch": 0.6050978280380542,
      "grad_norm": 3.4242725372314453,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 6742
    },
    {
      "epoch": 0.6051875785316819,
      "grad_norm": 2.4303417205810547,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 6743
    },
    {
      "epoch": 0.6052773290253096,
      "grad_norm": 2.6253321170806885,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 6744
    },
    {
      "epoch": 0.6053670795189373,
      "grad_norm": 3.1915321350097656,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 6745
    },
    {
      "epoch": 0.6054568300125651,
      "grad_norm": 2.8239076137542725,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 6746
    },
    {
      "epoch": 0.6055465805061928,
      "grad_norm": 3.8024401664733887,
      "learning_rate": 0.0,
      "loss": 0.6917,
      "step": 6747
    },
    {
      "epoch": 0.6056363309998205,
      "grad_norm": 2.4377596378326416,
      "learning_rate": 0.0,
      "loss": 0.7052,
      "step": 6748
    },
    {
      "epoch": 0.6057260814934482,
      "grad_norm": 2.7990245819091797,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 6749
    },
    {
      "epoch": 0.6058158319870759,
      "grad_norm": 2.4235713481903076,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 6750
    },
    {
      "epoch": 0.6059055824807037,
      "grad_norm": 2.3090221881866455,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 6751
    },
    {
      "epoch": 0.6059953329743314,
      "grad_norm": 5.517598628997803,
      "learning_rate": 0.0,
      "loss": 0.6861,
      "step": 6752
    },
    {
      "epoch": 0.6060850834679591,
      "grad_norm": 3.6966350078582764,
      "learning_rate": 0.0,
      "loss": 0.7631,
      "step": 6753
    },
    {
      "epoch": 0.6061748339615868,
      "grad_norm": 2.9847421646118164,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 6754
    },
    {
      "epoch": 0.6062645844552145,
      "grad_norm": 2.6082887649536133,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 6755
    },
    {
      "epoch": 0.6063543349488422,
      "grad_norm": 3.938667058944702,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 6756
    },
    {
      "epoch": 0.60644408544247,
      "grad_norm": 3.043579578399658,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 6757
    },
    {
      "epoch": 0.6065338359360977,
      "grad_norm": 2.26233172416687,
      "learning_rate": 0.0,
      "loss": 0.7125,
      "step": 6758
    },
    {
      "epoch": 0.6066235864297254,
      "grad_norm": 2.596874952316284,
      "learning_rate": 0.0,
      "loss": 0.6862,
      "step": 6759
    },
    {
      "epoch": 0.606713336923353,
      "grad_norm": 3.777522325515747,
      "learning_rate": 0.0,
      "loss": 0.7329,
      "step": 6760
    },
    {
      "epoch": 0.6068030874169807,
      "grad_norm": 2.797203302383423,
      "learning_rate": 0.0,
      "loss": 0.6551,
      "step": 6761
    },
    {
      "epoch": 0.6068928379106086,
      "grad_norm": 3.518038272857666,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 6762
    },
    {
      "epoch": 0.6069825884042362,
      "grad_norm": 2.769991159439087,
      "learning_rate": 0.0,
      "loss": 0.7026,
      "step": 6763
    },
    {
      "epoch": 0.6070723388978639,
      "grad_norm": 4.079473495483398,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 6764
    },
    {
      "epoch": 0.6071620893914916,
      "grad_norm": 4.035571098327637,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 6765
    },
    {
      "epoch": 0.6072518398851193,
      "grad_norm": 3.9071364402770996,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 6766
    },
    {
      "epoch": 0.6073415903787471,
      "grad_norm": 2.396143913269043,
      "learning_rate": 0.0,
      "loss": 0.7927,
      "step": 6767
    },
    {
      "epoch": 0.6074313408723748,
      "grad_norm": 2.764721155166626,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 6768
    },
    {
      "epoch": 0.6075210913660025,
      "grad_norm": 11.136245727539062,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 6769
    },
    {
      "epoch": 0.6076108418596302,
      "grad_norm": 3.4062671661376953,
      "learning_rate": 0.0,
      "loss": 0.8166,
      "step": 6770
    },
    {
      "epoch": 0.6077005923532579,
      "grad_norm": 3.0655879974365234,
      "learning_rate": 0.0,
      "loss": 0.7983,
      "step": 6771
    },
    {
      "epoch": 0.6077903428468857,
      "grad_norm": 3.492607355117798,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 6772
    },
    {
      "epoch": 0.6078800933405134,
      "grad_norm": 3.4421515464782715,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 6773
    },
    {
      "epoch": 0.6079698438341411,
      "grad_norm": 2.712271213531494,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 6774
    },
    {
      "epoch": 0.6080595943277688,
      "grad_norm": 3.002405881881714,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 6775
    },
    {
      "epoch": 0.6081493448213965,
      "grad_norm": 2.7629125118255615,
      "learning_rate": 0.0,
      "loss": 0.6567,
      "step": 6776
    },
    {
      "epoch": 0.6082390953150242,
      "grad_norm": 3.22497296333313,
      "learning_rate": 0.0,
      "loss": 0.6794,
      "step": 6777
    },
    {
      "epoch": 0.608328845808652,
      "grad_norm": 2.966830253601074,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 6778
    },
    {
      "epoch": 0.6084185963022797,
      "grad_norm": 4.1293864250183105,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 6779
    },
    {
      "epoch": 0.6085083467959074,
      "grad_norm": 3.1450185775756836,
      "learning_rate": 0.0,
      "loss": 0.7524,
      "step": 6780
    },
    {
      "epoch": 0.6085980972895351,
      "grad_norm": 2.3788747787475586,
      "learning_rate": 0.0,
      "loss": 0.6418,
      "step": 6781
    },
    {
      "epoch": 0.6086878477831628,
      "grad_norm": 2.81850266456604,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 6782
    },
    {
      "epoch": 0.6087775982767906,
      "grad_norm": 3.0913009643554688,
      "learning_rate": 0.0,
      "loss": 0.7772,
      "step": 6783
    },
    {
      "epoch": 0.6088673487704183,
      "grad_norm": 3.221283197402954,
      "learning_rate": 0.0,
      "loss": 0.7733,
      "step": 6784
    },
    {
      "epoch": 0.608957099264046,
      "grad_norm": 3.678867816925049,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 6785
    },
    {
      "epoch": 0.6090468497576736,
      "grad_norm": 3.2290194034576416,
      "learning_rate": 0.0,
      "loss": 0.6779,
      "step": 6786
    },
    {
      "epoch": 0.6091366002513013,
      "grad_norm": 3.2170143127441406,
      "learning_rate": 0.0,
      "loss": 0.6887,
      "step": 6787
    },
    {
      "epoch": 0.6092263507449291,
      "grad_norm": 2.706040382385254,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 6788
    },
    {
      "epoch": 0.6093161012385568,
      "grad_norm": 3.8803021907806396,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 6789
    },
    {
      "epoch": 0.6094058517321845,
      "grad_norm": 2.782127857208252,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 6790
    },
    {
      "epoch": 0.6094956022258122,
      "grad_norm": 2.515798807144165,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 6791
    },
    {
      "epoch": 0.6095853527194399,
      "grad_norm": 2.1809380054473877,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 6792
    },
    {
      "epoch": 0.6096751032130677,
      "grad_norm": 3.8434717655181885,
      "learning_rate": 0.0,
      "loss": 0.6974,
      "step": 6793
    },
    {
      "epoch": 0.6097648537066954,
      "grad_norm": 2.4415175914764404,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 6794
    },
    {
      "epoch": 0.6098546042003231,
      "grad_norm": 7.906202793121338,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 6795
    },
    {
      "epoch": 0.6099443546939508,
      "grad_norm": 2.3453826904296875,
      "learning_rate": 0.0,
      "loss": 0.6705,
      "step": 6796
    },
    {
      "epoch": 0.6100341051875785,
      "grad_norm": 2.5897715091705322,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 6797
    },
    {
      "epoch": 0.6101238556812062,
      "grad_norm": 2.2389180660247803,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 6798
    },
    {
      "epoch": 0.610213606174834,
      "grad_norm": 2.5185458660125732,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 6799
    },
    {
      "epoch": 0.6103033566684617,
      "grad_norm": 3.451202154159546,
      "learning_rate": 0.0,
      "loss": 0.6975,
      "step": 6800
    },
    {
      "epoch": 0.6103931071620894,
      "grad_norm": 2.6699628829956055,
      "learning_rate": 0.0,
      "loss": 0.6795,
      "step": 6801
    },
    {
      "epoch": 0.6104828576557171,
      "grad_norm": 12.376317024230957,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 6802
    },
    {
      "epoch": 0.6105726081493448,
      "grad_norm": 3.68176007270813,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 6803
    },
    {
      "epoch": 0.6106623586429726,
      "grad_norm": 2.888050079345703,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 6804
    },
    {
      "epoch": 0.6107521091366003,
      "grad_norm": 2.366265296936035,
      "learning_rate": 0.0,
      "loss": 0.6824,
      "step": 6805
    },
    {
      "epoch": 0.610841859630228,
      "grad_norm": 4.887862205505371,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 6806
    },
    {
      "epoch": 0.6109316101238557,
      "grad_norm": 3.437857151031494,
      "learning_rate": 0.0,
      "loss": 0.6784,
      "step": 6807
    },
    {
      "epoch": 0.6110213606174834,
      "grad_norm": 3.6364073753356934,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 6808
    },
    {
      "epoch": 0.6111111111111112,
      "grad_norm": 2.228116035461426,
      "learning_rate": 0.0,
      "loss": 0.6763,
      "step": 6809
    },
    {
      "epoch": 0.6112008616047389,
      "grad_norm": 3.9557974338531494,
      "learning_rate": 0.0,
      "loss": 0.6577,
      "step": 6810
    },
    {
      "epoch": 0.6112906120983665,
      "grad_norm": 2.7921087741851807,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 6811
    },
    {
      "epoch": 0.6113803625919942,
      "grad_norm": 2.5271058082580566,
      "learning_rate": 0.0,
      "loss": 0.6806,
      "step": 6812
    },
    {
      "epoch": 0.6114701130856219,
      "grad_norm": 1.9402564764022827,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 6813
    },
    {
      "epoch": 0.6115598635792497,
      "grad_norm": 2.4791152477264404,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 6814
    },
    {
      "epoch": 0.6116496140728774,
      "grad_norm": 4.073365688323975,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 6815
    },
    {
      "epoch": 0.6117393645665051,
      "grad_norm": 3.0388340950012207,
      "learning_rate": 0.0,
      "loss": 0.7519,
      "step": 6816
    },
    {
      "epoch": 0.6118291150601328,
      "grad_norm": 2.846808433532715,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 6817
    },
    {
      "epoch": 0.6119188655537605,
      "grad_norm": 2.8698980808258057,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 6818
    },
    {
      "epoch": 0.6120086160473882,
      "grad_norm": 2.8604700565338135,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 6819
    },
    {
      "epoch": 0.612098366541016,
      "grad_norm": 2.4458069801330566,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 6820
    },
    {
      "epoch": 0.6121881170346437,
      "grad_norm": 2.508618116378784,
      "learning_rate": 0.0,
      "loss": 0.7703,
      "step": 6821
    },
    {
      "epoch": 0.6122778675282714,
      "grad_norm": 2.308265447616577,
      "learning_rate": 0.0,
      "loss": 0.6612,
      "step": 6822
    },
    {
      "epoch": 0.6123676180218991,
      "grad_norm": 2.6497087478637695,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 6823
    },
    {
      "epoch": 0.6124573685155268,
      "grad_norm": 2.301257371902466,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 6824
    },
    {
      "epoch": 0.6125471190091546,
      "grad_norm": 2.4617433547973633,
      "learning_rate": 0.0,
      "loss": 0.8401,
      "step": 6825
    },
    {
      "epoch": 0.6126368695027823,
      "grad_norm": 2.6720023155212402,
      "learning_rate": 0.0,
      "loss": 0.6865,
      "step": 6826
    },
    {
      "epoch": 0.61272661999641,
      "grad_norm": 3.021407127380371,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 6827
    },
    {
      "epoch": 0.6128163704900377,
      "grad_norm": 2.2180066108703613,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 6828
    },
    {
      "epoch": 0.6129061209836654,
      "grad_norm": 2.93314790725708,
      "learning_rate": 0.0,
      "loss": 0.7488,
      "step": 6829
    },
    {
      "epoch": 0.6129958714772932,
      "grad_norm": 2.325551748275757,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 6830
    },
    {
      "epoch": 0.6130856219709209,
      "grad_norm": 2.22179913520813,
      "learning_rate": 0.0,
      "loss": 0.6941,
      "step": 6831
    },
    {
      "epoch": 0.6131753724645486,
      "grad_norm": 2.772691011428833,
      "learning_rate": 0.0,
      "loss": 0.7026,
      "step": 6832
    },
    {
      "epoch": 0.6132651229581763,
      "grad_norm": 2.5396249294281006,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 6833
    },
    {
      "epoch": 0.613354873451804,
      "grad_norm": 2.096209764480591,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 6834
    },
    {
      "epoch": 0.6134446239454318,
      "grad_norm": 3.448253870010376,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 6835
    },
    {
      "epoch": 0.6135343744390594,
      "grad_norm": 2.862713575363159,
      "learning_rate": 0.0,
      "loss": 0.6767,
      "step": 6836
    },
    {
      "epoch": 0.6136241249326871,
      "grad_norm": 3.095808744430542,
      "learning_rate": 0.0,
      "loss": 0.6804,
      "step": 6837
    },
    {
      "epoch": 0.6137138754263148,
      "grad_norm": 1.90697181224823,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 6838
    },
    {
      "epoch": 0.6138036259199425,
      "grad_norm": 2.639281749725342,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 6839
    },
    {
      "epoch": 0.6138933764135702,
      "grad_norm": 2.9630589485168457,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 6840
    },
    {
      "epoch": 0.613983126907198,
      "grad_norm": 2.544273853302002,
      "learning_rate": 0.0,
      "loss": 0.772,
      "step": 6841
    },
    {
      "epoch": 0.6140728774008257,
      "grad_norm": 2.614130973815918,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 6842
    },
    {
      "epoch": 0.6141626278944534,
      "grad_norm": 2.3325021266937256,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 6843
    },
    {
      "epoch": 0.6142523783880811,
      "grad_norm": 2.7519633769989014,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 6844
    },
    {
      "epoch": 0.6143421288817088,
      "grad_norm": 2.7533223628997803,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 6845
    },
    {
      "epoch": 0.6144318793753366,
      "grad_norm": 2.4374353885650635,
      "learning_rate": 0.0,
      "loss": 0.7395,
      "step": 6846
    },
    {
      "epoch": 0.6145216298689643,
      "grad_norm": 2.769507646560669,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 6847
    },
    {
      "epoch": 0.614611380362592,
      "grad_norm": 2.708585262298584,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 6848
    },
    {
      "epoch": 0.6147011308562197,
      "grad_norm": 4.275646686553955,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 6849
    },
    {
      "epoch": 0.6147908813498474,
      "grad_norm": 2.14320707321167,
      "learning_rate": 0.0,
      "loss": 0.651,
      "step": 6850
    },
    {
      "epoch": 0.6148806318434752,
      "grad_norm": 6.398077487945557,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 6851
    },
    {
      "epoch": 0.6149703823371029,
      "grad_norm": 2.7917678356170654,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 6852
    },
    {
      "epoch": 0.6150601328307306,
      "grad_norm": 2.2837018966674805,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 6853
    },
    {
      "epoch": 0.6151498833243583,
      "grad_norm": 3.5441770553588867,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 6854
    },
    {
      "epoch": 0.615239633817986,
      "grad_norm": 2.834543228149414,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 6855
    },
    {
      "epoch": 0.6153293843116137,
      "grad_norm": 2.4638421535491943,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 6856
    },
    {
      "epoch": 0.6154191348052415,
      "grad_norm": 2.3863449096679688,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 6857
    },
    {
      "epoch": 0.6155088852988692,
      "grad_norm": 2.8200433254241943,
      "learning_rate": 0.0,
      "loss": 0.6834,
      "step": 6858
    },
    {
      "epoch": 0.6155986357924969,
      "grad_norm": 5.982085704803467,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 6859
    },
    {
      "epoch": 0.6156883862861245,
      "grad_norm": 3.477257251739502,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 6860
    },
    {
      "epoch": 0.6157781367797522,
      "grad_norm": 2.3448965549468994,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 6861
    },
    {
      "epoch": 0.61586788727338,
      "grad_norm": 2.825078248977661,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 6862
    },
    {
      "epoch": 0.6159576377670077,
      "grad_norm": 3.194270610809326,
      "learning_rate": 0.0,
      "loss": 0.6975,
      "step": 6863
    },
    {
      "epoch": 0.6160473882606354,
      "grad_norm": 3.0125768184661865,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 6864
    },
    {
      "epoch": 0.6161371387542631,
      "grad_norm": 2.9122204780578613,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 6865
    },
    {
      "epoch": 0.6162268892478908,
      "grad_norm": 3.180145263671875,
      "learning_rate": 0.0,
      "loss": 0.6641,
      "step": 6866
    },
    {
      "epoch": 0.6163166397415186,
      "grad_norm": 3.0103282928466797,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 6867
    },
    {
      "epoch": 0.6164063902351463,
      "grad_norm": 2.7334742546081543,
      "learning_rate": 0.0,
      "loss": 0.7987,
      "step": 6868
    },
    {
      "epoch": 0.616496140728774,
      "grad_norm": 15.231287956237793,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 6869
    },
    {
      "epoch": 0.6165858912224017,
      "grad_norm": 4.531485557556152,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 6870
    },
    {
      "epoch": 0.6166756417160294,
      "grad_norm": 4.375033378601074,
      "learning_rate": 0.0,
      "loss": 0.6707,
      "step": 6871
    },
    {
      "epoch": 0.6167653922096572,
      "grad_norm": 2.8602242469787598,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 6872
    },
    {
      "epoch": 0.6168551427032849,
      "grad_norm": 3.087181806564331,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 6873
    },
    {
      "epoch": 0.6169448931969126,
      "grad_norm": 3.7931249141693115,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 6874
    },
    {
      "epoch": 0.6170346436905403,
      "grad_norm": 3.050652503967285,
      "learning_rate": 0.0,
      "loss": 0.663,
      "step": 6875
    },
    {
      "epoch": 0.617124394184168,
      "grad_norm": 2.5600783824920654,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 6876
    },
    {
      "epoch": 0.6172141446777957,
      "grad_norm": 2.4481759071350098,
      "learning_rate": 0.0,
      "loss": 0.7565,
      "step": 6877
    },
    {
      "epoch": 0.6173038951714235,
      "grad_norm": 7.4952778816223145,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 6878
    },
    {
      "epoch": 0.6173936456650512,
      "grad_norm": 2.883906126022339,
      "learning_rate": 0.0,
      "loss": 0.7172,
      "step": 6879
    },
    {
      "epoch": 0.6174833961586789,
      "grad_norm": 3.1110785007476807,
      "learning_rate": 0.0,
      "loss": 0.7838,
      "step": 6880
    },
    {
      "epoch": 0.6175731466523066,
      "grad_norm": 2.1065213680267334,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 6881
    },
    {
      "epoch": 0.6176628971459343,
      "grad_norm": 2.2146494388580322,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 6882
    },
    {
      "epoch": 0.6177526476395621,
      "grad_norm": 3.0197649002075195,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 6883
    },
    {
      "epoch": 0.6178423981331898,
      "grad_norm": 2.8851301670074463,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 6884
    },
    {
      "epoch": 0.6179321486268174,
      "grad_norm": 2.6798555850982666,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 6885
    },
    {
      "epoch": 0.6180218991204451,
      "grad_norm": 2.927868366241455,
      "learning_rate": 0.0,
      "loss": 0.7925,
      "step": 6886
    },
    {
      "epoch": 0.6181116496140728,
      "grad_norm": 2.676297426223755,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 6887
    },
    {
      "epoch": 0.6182014001077006,
      "grad_norm": 3.161553382873535,
      "learning_rate": 0.0,
      "loss": 0.6927,
      "step": 6888
    },
    {
      "epoch": 0.6182911506013283,
      "grad_norm": 2.1132259368896484,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 6889
    },
    {
      "epoch": 0.618380901094956,
      "grad_norm": 2.596925973892212,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 6890
    },
    {
      "epoch": 0.6184706515885837,
      "grad_norm": 3.204761505126953,
      "learning_rate": 0.0,
      "loss": 0.6866,
      "step": 6891
    },
    {
      "epoch": 0.6185604020822114,
      "grad_norm": 6.567229747772217,
      "learning_rate": 0.0,
      "loss": 0.6627,
      "step": 6892
    },
    {
      "epoch": 0.6186501525758392,
      "grad_norm": 2.0616273880004883,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 6893
    },
    {
      "epoch": 0.6187399030694669,
      "grad_norm": 2.6210825443267822,
      "learning_rate": 0.0,
      "loss": 0.6671,
      "step": 6894
    },
    {
      "epoch": 0.6188296535630946,
      "grad_norm": 2.8404741287231445,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 6895
    },
    {
      "epoch": 0.6189194040567223,
      "grad_norm": 5.641280174255371,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 6896
    },
    {
      "epoch": 0.61900915455035,
      "grad_norm": 3.4090752601623535,
      "learning_rate": 0.0,
      "loss": 0.6871,
      "step": 6897
    },
    {
      "epoch": 0.6190989050439777,
      "grad_norm": 6.024676322937012,
      "learning_rate": 0.0,
      "loss": 0.7568,
      "step": 6898
    },
    {
      "epoch": 0.6191886555376055,
      "grad_norm": 2.595102071762085,
      "learning_rate": 0.0,
      "loss": 0.6569,
      "step": 6899
    },
    {
      "epoch": 0.6192784060312332,
      "grad_norm": 2.415175199508667,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 6900
    },
    {
      "epoch": 0.6193681565248609,
      "grad_norm": 2.276618480682373,
      "learning_rate": 0.0,
      "loss": 0.6828,
      "step": 6901
    },
    {
      "epoch": 0.6194579070184886,
      "grad_norm": 2.490570306777954,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 6902
    },
    {
      "epoch": 0.6195476575121163,
      "grad_norm": 2.129807472229004,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 6903
    },
    {
      "epoch": 0.6196374080057441,
      "grad_norm": 2.3921570777893066,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 6904
    },
    {
      "epoch": 0.6197271584993718,
      "grad_norm": 2.409085750579834,
      "learning_rate": 0.0,
      "loss": 0.6859,
      "step": 6905
    },
    {
      "epoch": 0.6198169089929995,
      "grad_norm": 2.7687039375305176,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 6906
    },
    {
      "epoch": 0.6199066594866272,
      "grad_norm": 2.6268367767333984,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 6907
    },
    {
      "epoch": 0.6199964099802548,
      "grad_norm": 2.8787953853607178,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 6908
    },
    {
      "epoch": 0.6200861604738827,
      "grad_norm": 5.123167037963867,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 6909
    },
    {
      "epoch": 0.6201759109675103,
      "grad_norm": 2.711430549621582,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 6910
    },
    {
      "epoch": 0.620265661461138,
      "grad_norm": 6.455804824829102,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 6911
    },
    {
      "epoch": 0.6203554119547657,
      "grad_norm": 2.860481023788452,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 6912
    },
    {
      "epoch": 0.6204451624483934,
      "grad_norm": 2.6211626529693604,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 6913
    },
    {
      "epoch": 0.6205349129420212,
      "grad_norm": 2.417405843734741,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 6914
    },
    {
      "epoch": 0.6206246634356489,
      "grad_norm": 2.9136197566986084,
      "learning_rate": 0.0,
      "loss": 0.8133,
      "step": 6915
    },
    {
      "epoch": 0.6207144139292766,
      "grad_norm": 2.993051767349243,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 6916
    },
    {
      "epoch": 0.6208041644229043,
      "grad_norm": 2.1469430923461914,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 6917
    },
    {
      "epoch": 0.620893914916532,
      "grad_norm": 3.6207759380340576,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 6918
    },
    {
      "epoch": 0.6209836654101597,
      "grad_norm": 2.7299578189849854,
      "learning_rate": 0.0,
      "loss": 0.6861,
      "step": 6919
    },
    {
      "epoch": 0.6210734159037875,
      "grad_norm": 2.547899007797241,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 6920
    },
    {
      "epoch": 0.6211631663974152,
      "grad_norm": 2.694835662841797,
      "learning_rate": 0.0,
      "loss": 0.84,
      "step": 6921
    },
    {
      "epoch": 0.6212529168910429,
      "grad_norm": 2.759165048599243,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 6922
    },
    {
      "epoch": 0.6213426673846706,
      "grad_norm": 2.6685807704925537,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 6923
    },
    {
      "epoch": 0.6214324178782983,
      "grad_norm": 4.372408866882324,
      "learning_rate": 0.0,
      "loss": 0.6724,
      "step": 6924
    },
    {
      "epoch": 0.6215221683719261,
      "grad_norm": 2.6634886264801025,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 6925
    },
    {
      "epoch": 0.6216119188655538,
      "grad_norm": 4.083906173706055,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 6926
    },
    {
      "epoch": 0.6217016693591815,
      "grad_norm": 2.592482328414917,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 6927
    },
    {
      "epoch": 0.6217914198528092,
      "grad_norm": 7.425328254699707,
      "learning_rate": 0.0,
      "loss": 0.7119,
      "step": 6928
    },
    {
      "epoch": 0.6218811703464369,
      "grad_norm": 2.765928030014038,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 6929
    },
    {
      "epoch": 0.6219709208400647,
      "grad_norm": 2.6631925106048584,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 6930
    },
    {
      "epoch": 0.6220606713336924,
      "grad_norm": 3.3673908710479736,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 6931
    },
    {
      "epoch": 0.62215042182732,
      "grad_norm": 4.725871562957764,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 6932
    },
    {
      "epoch": 0.6222401723209477,
      "grad_norm": 2.6560542583465576,
      "learning_rate": 0.0,
      "loss": 0.6915,
      "step": 6933
    },
    {
      "epoch": 0.6223299228145754,
      "grad_norm": 3.2775306701660156,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 6934
    },
    {
      "epoch": 0.6224196733082032,
      "grad_norm": 3.088104486465454,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 6935
    },
    {
      "epoch": 0.6225094238018309,
      "grad_norm": 2.4077975749969482,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 6936
    },
    {
      "epoch": 0.6225991742954586,
      "grad_norm": 2.712099313735962,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 6937
    },
    {
      "epoch": 0.6226889247890863,
      "grad_norm": 2.4539871215820312,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 6938
    },
    {
      "epoch": 0.622778675282714,
      "grad_norm": 13.233684539794922,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 6939
    },
    {
      "epoch": 0.6228684257763417,
      "grad_norm": 3.2376251220703125,
      "learning_rate": 0.0,
      "loss": 0.8051,
      "step": 6940
    },
    {
      "epoch": 0.6229581762699695,
      "grad_norm": 2.772871494293213,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 6941
    },
    {
      "epoch": 0.6230479267635972,
      "grad_norm": 2.794057846069336,
      "learning_rate": 0.0,
      "loss": 0.6979,
      "step": 6942
    },
    {
      "epoch": 0.6231376772572249,
      "grad_norm": 3.852111577987671,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 6943
    },
    {
      "epoch": 0.6232274277508526,
      "grad_norm": 4.899245738983154,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 6944
    },
    {
      "epoch": 0.6233171782444803,
      "grad_norm": 3.5496582984924316,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 6945
    },
    {
      "epoch": 0.6234069287381081,
      "grad_norm": 2.486849784851074,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 6946
    },
    {
      "epoch": 0.6234966792317358,
      "grad_norm": 2.7996785640716553,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 6947
    },
    {
      "epoch": 0.6235864297253635,
      "grad_norm": 4.121232032775879,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 6948
    },
    {
      "epoch": 0.6236761802189912,
      "grad_norm": 4.471390247344971,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 6949
    },
    {
      "epoch": 0.6237659307126189,
      "grad_norm": 3.2027902603149414,
      "learning_rate": 0.0,
      "loss": 0.7675,
      "step": 6950
    },
    {
      "epoch": 0.6238556812062467,
      "grad_norm": 3.7932870388031006,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 6951
    },
    {
      "epoch": 0.6239454316998744,
      "grad_norm": 7.516088485717773,
      "learning_rate": 0.0,
      "loss": 0.6835,
      "step": 6952
    },
    {
      "epoch": 0.6240351821935021,
      "grad_norm": 3.5748674869537354,
      "learning_rate": 0.0,
      "loss": 0.6641,
      "step": 6953
    },
    {
      "epoch": 0.6241249326871298,
      "grad_norm": 7.573915958404541,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 6954
    },
    {
      "epoch": 0.6242146831807575,
      "grad_norm": 9.3552827835083,
      "learning_rate": 0.0,
      "loss": 0.6925,
      "step": 6955
    },
    {
      "epoch": 0.6243044336743853,
      "grad_norm": 2.6556057929992676,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 6956
    },
    {
      "epoch": 0.624394184168013,
      "grad_norm": 2.4654898643493652,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 6957
    },
    {
      "epoch": 0.6244839346616406,
      "grad_norm": 3.25215220451355,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 6958
    },
    {
      "epoch": 0.6245736851552683,
      "grad_norm": 2.421700954437256,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 6959
    },
    {
      "epoch": 0.624663435648896,
      "grad_norm": 3.1726107597351074,
      "learning_rate": 0.0,
      "loss": 0.6618,
      "step": 6960
    },
    {
      "epoch": 0.6247531861425237,
      "grad_norm": 3.495330572128296,
      "learning_rate": 0.0,
      "loss": 0.6658,
      "step": 6961
    },
    {
      "epoch": 0.6248429366361515,
      "grad_norm": 2.6391782760620117,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 6962
    },
    {
      "epoch": 0.6249326871297792,
      "grad_norm": 2.5553107261657715,
      "learning_rate": 0.0,
      "loss": 0.6642,
      "step": 6963
    },
    {
      "epoch": 0.6250224376234069,
      "grad_norm": 3.250960350036621,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 6964
    },
    {
      "epoch": 0.6251121881170346,
      "grad_norm": 3.321915864944458,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 6965
    },
    {
      "epoch": 0.6252019386106623,
      "grad_norm": 3.115278959274292,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 6966
    },
    {
      "epoch": 0.6252916891042901,
      "grad_norm": 3.3310463428497314,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 6967
    },
    {
      "epoch": 0.6253814395979178,
      "grad_norm": 2.85292387008667,
      "learning_rate": 0.0,
      "loss": 0.695,
      "step": 6968
    },
    {
      "epoch": 0.6254711900915455,
      "grad_norm": 2.900524854660034,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 6969
    },
    {
      "epoch": 0.6255609405851732,
      "grad_norm": 2.818829298019409,
      "learning_rate": 0.0,
      "loss": 0.7632,
      "step": 6970
    },
    {
      "epoch": 0.6256506910788009,
      "grad_norm": 2.732919216156006,
      "learning_rate": 0.0,
      "loss": 0.6678,
      "step": 6971
    },
    {
      "epoch": 0.6257404415724287,
      "grad_norm": 4.022541046142578,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 6972
    },
    {
      "epoch": 0.6258301920660564,
      "grad_norm": 2.963160753250122,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 6973
    },
    {
      "epoch": 0.6259199425596841,
      "grad_norm": 2.380629301071167,
      "learning_rate": 0.0,
      "loss": 0.7939,
      "step": 6974
    },
    {
      "epoch": 0.6260096930533118,
      "grad_norm": 2.223033905029297,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 6975
    },
    {
      "epoch": 0.6260994435469395,
      "grad_norm": 2.4841256141662598,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 6976
    },
    {
      "epoch": 0.6261891940405673,
      "grad_norm": 2.455813407897949,
      "learning_rate": 0.0,
      "loss": 0.6891,
      "step": 6977
    },
    {
      "epoch": 0.626278944534195,
      "grad_norm": 2.501190662384033,
      "learning_rate": 0.0,
      "loss": 0.7773,
      "step": 6978
    },
    {
      "epoch": 0.6263686950278227,
      "grad_norm": 4.343144416809082,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 6979
    },
    {
      "epoch": 0.6264584455214504,
      "grad_norm": 3.773585796356201,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 6980
    },
    {
      "epoch": 0.626548196015078,
      "grad_norm": 4.182703495025635,
      "learning_rate": 0.0,
      "loss": 0.6871,
      "step": 6981
    },
    {
      "epoch": 0.6266379465087057,
      "grad_norm": 2.7777886390686035,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 6982
    },
    {
      "epoch": 0.6267276970023336,
      "grad_norm": 1.9885683059692383,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 6983
    },
    {
      "epoch": 0.6268174474959612,
      "grad_norm": 2.7907590866088867,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 6984
    },
    {
      "epoch": 0.6269071979895889,
      "grad_norm": 2.838407516479492,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 6985
    },
    {
      "epoch": 0.6269969484832166,
      "grad_norm": 6.007766246795654,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 6986
    },
    {
      "epoch": 0.6270866989768443,
      "grad_norm": 2.767641544342041,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 6987
    },
    {
      "epoch": 0.6271764494704721,
      "grad_norm": 3.311077356338501,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 6988
    },
    {
      "epoch": 0.6272661999640998,
      "grad_norm": 3.060600996017456,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 6989
    },
    {
      "epoch": 0.6273559504577275,
      "grad_norm": 2.410841226577759,
      "learning_rate": 0.0,
      "loss": 0.6604,
      "step": 6990
    },
    {
      "epoch": 0.6274457009513552,
      "grad_norm": 3.1630494594573975,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 6991
    },
    {
      "epoch": 0.6275354514449829,
      "grad_norm": 2.950667142868042,
      "learning_rate": 0.0,
      "loss": 0.6911,
      "step": 6992
    },
    {
      "epoch": 0.6276252019386107,
      "grad_norm": 4.060629844665527,
      "learning_rate": 0.0,
      "loss": 0.7486,
      "step": 6993
    },
    {
      "epoch": 0.6277149524322384,
      "grad_norm": 2.328026056289673,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 6994
    },
    {
      "epoch": 0.6278047029258661,
      "grad_norm": 3.0515661239624023,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 6995
    },
    {
      "epoch": 0.6278944534194938,
      "grad_norm": 2.650000810623169,
      "learning_rate": 0.0,
      "loss": 0.6612,
      "step": 6996
    },
    {
      "epoch": 0.6279842039131215,
      "grad_norm": 2.8778321743011475,
      "learning_rate": 0.0,
      "loss": 0.6905,
      "step": 6997
    },
    {
      "epoch": 0.6280739544067493,
      "grad_norm": 2.838364839553833,
      "learning_rate": 0.0,
      "loss": 0.7418,
      "step": 6998
    },
    {
      "epoch": 0.628163704900377,
      "grad_norm": 2.6659255027770996,
      "learning_rate": 0.0,
      "loss": 0.6673,
      "step": 6999
    },
    {
      "epoch": 0.6282534553940047,
      "grad_norm": 2.922743320465088,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 7000
    },
    {
      "epoch": 0.6283432058876324,
      "grad_norm": 3.0805811882019043,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 7001
    },
    {
      "epoch": 0.6284329563812601,
      "grad_norm": 3.0655853748321533,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 7002
    },
    {
      "epoch": 0.6285227068748878,
      "grad_norm": 2.390671491622925,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 7003
    },
    {
      "epoch": 0.6286124573685156,
      "grad_norm": 2.9645049571990967,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 7004
    },
    {
      "epoch": 0.6287022078621433,
      "grad_norm": 4.862165927886963,
      "learning_rate": 0.0,
      "loss": 0.6556,
      "step": 7005
    },
    {
      "epoch": 0.628791958355771,
      "grad_norm": 2.7723004817962646,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 7006
    },
    {
      "epoch": 0.6288817088493986,
      "grad_norm": 2.378807783126831,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 7007
    },
    {
      "epoch": 0.6289714593430263,
      "grad_norm": 2.157033920288086,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 7008
    },
    {
      "epoch": 0.6290612098366541,
      "grad_norm": 2.512026071548462,
      "learning_rate": 0.0,
      "loss": 0.7866,
      "step": 7009
    },
    {
      "epoch": 0.6291509603302818,
      "grad_norm": 2.143437385559082,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 7010
    },
    {
      "epoch": 0.6292407108239095,
      "grad_norm": 3.0442581176757812,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 7011
    },
    {
      "epoch": 0.6293304613175372,
      "grad_norm": 2.37919020652771,
      "learning_rate": 0.0,
      "loss": 0.7062,
      "step": 7012
    },
    {
      "epoch": 0.6294202118111649,
      "grad_norm": 3.2250070571899414,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 7013
    },
    {
      "epoch": 0.6295099623047927,
      "grad_norm": 2.378282070159912,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 7014
    },
    {
      "epoch": 0.6295997127984204,
      "grad_norm": 2.286288022994995,
      "learning_rate": 0.0,
      "loss": 0.7607,
      "step": 7015
    },
    {
      "epoch": 0.6296894632920481,
      "grad_norm": 2.941185474395752,
      "learning_rate": 0.0,
      "loss": 0.6806,
      "step": 7016
    },
    {
      "epoch": 0.6297792137856758,
      "grad_norm": 5.789433002471924,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 7017
    },
    {
      "epoch": 0.6298689642793035,
      "grad_norm": 3.6889281272888184,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 7018
    },
    {
      "epoch": 0.6299587147729313,
      "grad_norm": 5.376245975494385,
      "learning_rate": 0.0,
      "loss": 0.7602,
      "step": 7019
    },
    {
      "epoch": 0.630048465266559,
      "grad_norm": 2.4891843795776367,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 7020
    },
    {
      "epoch": 0.6301382157601867,
      "grad_norm": 7.482235908508301,
      "learning_rate": 0.0,
      "loss": 0.6771,
      "step": 7021
    },
    {
      "epoch": 0.6302279662538144,
      "grad_norm": 2.502006769180298,
      "learning_rate": 0.0,
      "loss": 0.7945,
      "step": 7022
    },
    {
      "epoch": 0.6303177167474421,
      "grad_norm": 2.6141600608825684,
      "learning_rate": 0.0,
      "loss": 0.7688,
      "step": 7023
    },
    {
      "epoch": 0.6304074672410698,
      "grad_norm": 3.336099863052368,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 7024
    },
    {
      "epoch": 0.6304972177346976,
      "grad_norm": 4.872405529022217,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 7025
    },
    {
      "epoch": 0.6305869682283253,
      "grad_norm": 4.8317060470581055,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 7026
    },
    {
      "epoch": 0.630676718721953,
      "grad_norm": 2.9921817779541016,
      "learning_rate": 0.0,
      "loss": 0.7044,
      "step": 7027
    },
    {
      "epoch": 0.6307664692155807,
      "grad_norm": 2.5351152420043945,
      "learning_rate": 0.0,
      "loss": 0.7488,
      "step": 7028
    },
    {
      "epoch": 0.6308562197092084,
      "grad_norm": 2.8910064697265625,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 7029
    },
    {
      "epoch": 0.6309459702028362,
      "grad_norm": 6.12619686126709,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 7030
    },
    {
      "epoch": 0.6310357206964639,
      "grad_norm": 2.5096864700317383,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 7031
    },
    {
      "epoch": 0.6311254711900915,
      "grad_norm": 2.5973362922668457,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 7032
    },
    {
      "epoch": 0.6312152216837192,
      "grad_norm": 2.554614543914795,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 7033
    },
    {
      "epoch": 0.6313049721773469,
      "grad_norm": 2.7530417442321777,
      "learning_rate": 0.0,
      "loss": 0.6869,
      "step": 7034
    },
    {
      "epoch": 0.6313947226709747,
      "grad_norm": 10.622456550598145,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 7035
    },
    {
      "epoch": 0.6314844731646024,
      "grad_norm": 3.7630631923675537,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 7036
    },
    {
      "epoch": 0.6315742236582301,
      "grad_norm": 2.760488271713257,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 7037
    },
    {
      "epoch": 0.6316639741518578,
      "grad_norm": 2.6024837493896484,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 7038
    },
    {
      "epoch": 0.6317537246454855,
      "grad_norm": 3.2458579540252686,
      "learning_rate": 0.0,
      "loss": 0.7196,
      "step": 7039
    },
    {
      "epoch": 0.6318434751391133,
      "grad_norm": 2.9527499675750732,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 7040
    },
    {
      "epoch": 0.631933225632741,
      "grad_norm": 2.3727471828460693,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 7041
    },
    {
      "epoch": 0.6320229761263687,
      "grad_norm": 2.970393419265747,
      "learning_rate": 0.0,
      "loss": 0.7021,
      "step": 7042
    },
    {
      "epoch": 0.6321127266199964,
      "grad_norm": 2.3887104988098145,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 7043
    },
    {
      "epoch": 0.6322024771136241,
      "grad_norm": 3.5270111560821533,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 7044
    },
    {
      "epoch": 0.6322922276072518,
      "grad_norm": 2.50482177734375,
      "learning_rate": 0.0,
      "loss": 0.6694,
      "step": 7045
    },
    {
      "epoch": 0.6323819781008796,
      "grad_norm": 4.21469259262085,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 7046
    },
    {
      "epoch": 0.6324717285945073,
      "grad_norm": 3.762328624725342,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 7047
    },
    {
      "epoch": 0.632561479088135,
      "grad_norm": 3.1914446353912354,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 7048
    },
    {
      "epoch": 0.6326512295817627,
      "grad_norm": 2.8165931701660156,
      "learning_rate": 0.0,
      "loss": 0.7684,
      "step": 7049
    },
    {
      "epoch": 0.6327409800753904,
      "grad_norm": 2.488528251647949,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 7050
    },
    {
      "epoch": 0.6328307305690182,
      "grad_norm": 3.66178035736084,
      "learning_rate": 0.0,
      "loss": 0.6958,
      "step": 7051
    },
    {
      "epoch": 0.6329204810626459,
      "grad_norm": 2.8435871601104736,
      "learning_rate": 0.0,
      "loss": 0.7393,
      "step": 7052
    },
    {
      "epoch": 0.6330102315562736,
      "grad_norm": 2.4086720943450928,
      "learning_rate": 0.0,
      "loss": 0.6807,
      "step": 7053
    },
    {
      "epoch": 0.6330999820499013,
      "grad_norm": 2.5799331665039062,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 7054
    },
    {
      "epoch": 0.633189732543529,
      "grad_norm": 4.145565509796143,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 7055
    },
    {
      "epoch": 0.6332794830371568,
      "grad_norm": 2.986801862716675,
      "learning_rate": 0.0,
      "loss": 0.7609,
      "step": 7056
    },
    {
      "epoch": 0.6333692335307844,
      "grad_norm": 2.9047183990478516,
      "learning_rate": 0.0,
      "loss": 0.6861,
      "step": 7057
    },
    {
      "epoch": 0.6334589840244121,
      "grad_norm": 2.8018553256988525,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 7058
    },
    {
      "epoch": 0.6335487345180398,
      "grad_norm": 3.643308162689209,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 7059
    },
    {
      "epoch": 0.6336384850116675,
      "grad_norm": 4.374095439910889,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 7060
    },
    {
      "epoch": 0.6337282355052953,
      "grad_norm": 2.4896976947784424,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 7061
    },
    {
      "epoch": 0.633817985998923,
      "grad_norm": 3.223684310913086,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 7062
    },
    {
      "epoch": 0.6339077364925507,
      "grad_norm": 3.348238706588745,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 7063
    },
    {
      "epoch": 0.6339974869861784,
      "grad_norm": 2.894711971282959,
      "learning_rate": 0.0,
      "loss": 0.6874,
      "step": 7064
    },
    {
      "epoch": 0.6340872374798061,
      "grad_norm": 2.9451332092285156,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 7065
    },
    {
      "epoch": 0.6341769879734338,
      "grad_norm": 3.2298038005828857,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 7066
    },
    {
      "epoch": 0.6342667384670616,
      "grad_norm": 4.58267068862915,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 7067
    },
    {
      "epoch": 0.6343564889606893,
      "grad_norm": 3.101910352706909,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 7068
    },
    {
      "epoch": 0.634446239454317,
      "grad_norm": 3.2782838344573975,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 7069
    },
    {
      "epoch": 0.6345359899479447,
      "grad_norm": 3.2420992851257324,
      "learning_rate": 0.0,
      "loss": 0.7624,
      "step": 7070
    },
    {
      "epoch": 0.6346257404415724,
      "grad_norm": 2.9506590366363525,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 7071
    },
    {
      "epoch": 0.6347154909352002,
      "grad_norm": 7.039891242980957,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 7072
    },
    {
      "epoch": 0.6348052414288279,
      "grad_norm": 3.0262701511383057,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 7073
    },
    {
      "epoch": 0.6348949919224556,
      "grad_norm": 2.6952576637268066,
      "learning_rate": 0.0,
      "loss": 0.6709,
      "step": 7074
    },
    {
      "epoch": 0.6349847424160833,
      "grad_norm": 2.771843433380127,
      "learning_rate": 0.0,
      "loss": 0.7831,
      "step": 7075
    },
    {
      "epoch": 0.635074492909711,
      "grad_norm": 2.7132604122161865,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 7076
    },
    {
      "epoch": 0.6351642434033388,
      "grad_norm": 3.570519208908081,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 7077
    },
    {
      "epoch": 0.6352539938969665,
      "grad_norm": 2.740496873855591,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 7078
    },
    {
      "epoch": 0.6353437443905942,
      "grad_norm": 2.4736406803131104,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 7079
    },
    {
      "epoch": 0.6354334948842219,
      "grad_norm": 4.395361423492432,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 7080
    },
    {
      "epoch": 0.6355232453778495,
      "grad_norm": 4.271203517913818,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 7081
    },
    {
      "epoch": 0.6356129958714772,
      "grad_norm": 3.229334592819214,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 7082
    },
    {
      "epoch": 0.635702746365105,
      "grad_norm": 3.325711488723755,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 7083
    },
    {
      "epoch": 0.6357924968587327,
      "grad_norm": 2.66805362701416,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 7084
    },
    {
      "epoch": 0.6358822473523604,
      "grad_norm": 2.482255697250366,
      "learning_rate": 0.0,
      "loss": 0.7856,
      "step": 7085
    },
    {
      "epoch": 0.6359719978459881,
      "grad_norm": 2.2712323665618896,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 7086
    },
    {
      "epoch": 0.6360617483396158,
      "grad_norm": 2.519599437713623,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 7087
    },
    {
      "epoch": 0.6361514988332436,
      "grad_norm": 2.749089479446411,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 7088
    },
    {
      "epoch": 0.6362412493268713,
      "grad_norm": 2.5427589416503906,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 7089
    },
    {
      "epoch": 0.636330999820499,
      "grad_norm": 2.403498888015747,
      "learning_rate": 0.0,
      "loss": 0.6755,
      "step": 7090
    },
    {
      "epoch": 0.6364207503141267,
      "grad_norm": 2.477832794189453,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 7091
    },
    {
      "epoch": 0.6365105008077544,
      "grad_norm": 2.1987929344177246,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 7092
    },
    {
      "epoch": 0.6366002513013822,
      "grad_norm": 8.992246627807617,
      "learning_rate": 0.0,
      "loss": 0.7952,
      "step": 7093
    },
    {
      "epoch": 0.6366900017950099,
      "grad_norm": 2.304625988006592,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 7094
    },
    {
      "epoch": 0.6367797522886376,
      "grad_norm": 2.7594265937805176,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 7095
    },
    {
      "epoch": 0.6368695027822653,
      "grad_norm": 2.943967819213867,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 7096
    },
    {
      "epoch": 0.636959253275893,
      "grad_norm": 2.7911627292633057,
      "learning_rate": 0.0,
      "loss": 0.7519,
      "step": 7097
    },
    {
      "epoch": 0.6370490037695208,
      "grad_norm": 2.3787200450897217,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 7098
    },
    {
      "epoch": 0.6371387542631485,
      "grad_norm": 2.3413655757904053,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 7099
    },
    {
      "epoch": 0.6372285047567762,
      "grad_norm": 2.6885318756103516,
      "learning_rate": 0.0,
      "loss": 0.6718,
      "step": 7100
    },
    {
      "epoch": 0.6373182552504039,
      "grad_norm": 2.4657199382781982,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 7101
    },
    {
      "epoch": 0.6374080057440316,
      "grad_norm": 2.0196895599365234,
      "learning_rate": 0.0,
      "loss": 0.6751,
      "step": 7102
    },
    {
      "epoch": 0.6374977562376593,
      "grad_norm": 2.0974667072296143,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 7103
    },
    {
      "epoch": 0.6375875067312871,
      "grad_norm": 2.4645636081695557,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 7104
    },
    {
      "epoch": 0.6376772572249148,
      "grad_norm": 2.1424756050109863,
      "learning_rate": 0.0,
      "loss": 0.7871,
      "step": 7105
    },
    {
      "epoch": 0.6377670077185424,
      "grad_norm": 4.249545574188232,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 7106
    },
    {
      "epoch": 0.6378567582121701,
      "grad_norm": 2.656764507293701,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 7107
    },
    {
      "epoch": 0.6379465087057978,
      "grad_norm": 3.03136944770813,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 7108
    },
    {
      "epoch": 0.6380362591994256,
      "grad_norm": 3.744960308074951,
      "learning_rate": 0.0,
      "loss": 0.7519,
      "step": 7109
    },
    {
      "epoch": 0.6381260096930533,
      "grad_norm": 2.642050266265869,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 7110
    },
    {
      "epoch": 0.638215760186681,
      "grad_norm": 2.9858298301696777,
      "learning_rate": 0.0,
      "loss": 0.6488,
      "step": 7111
    },
    {
      "epoch": 0.6383055106803087,
      "grad_norm": 2.4556236267089844,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7112
    },
    {
      "epoch": 0.6383952611739364,
      "grad_norm": 3.64119291305542,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 7113
    },
    {
      "epoch": 0.6384850116675642,
      "grad_norm": 2.5684659481048584,
      "learning_rate": 0.0,
      "loss": 0.6662,
      "step": 7114
    },
    {
      "epoch": 0.6385747621611919,
      "grad_norm": 3.2533485889434814,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 7115
    },
    {
      "epoch": 0.6386645126548196,
      "grad_norm": 3.268191337585449,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 7116
    },
    {
      "epoch": 0.6387542631484473,
      "grad_norm": 3.4647133350372314,
      "learning_rate": 0.0,
      "loss": 0.7691,
      "step": 7117
    },
    {
      "epoch": 0.638844013642075,
      "grad_norm": 2.616920232772827,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 7118
    },
    {
      "epoch": 0.6389337641357028,
      "grad_norm": 3.073124647140503,
      "learning_rate": 0.0,
      "loss": 0.6641,
      "step": 7119
    },
    {
      "epoch": 0.6390235146293305,
      "grad_norm": 2.1829960346221924,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 7120
    },
    {
      "epoch": 0.6391132651229582,
      "grad_norm": 2.66255521774292,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 7121
    },
    {
      "epoch": 0.6392030156165859,
      "grad_norm": 2.7936131954193115,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 7122
    },
    {
      "epoch": 0.6392927661102136,
      "grad_norm": 2.0835659503936768,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 7123
    },
    {
      "epoch": 0.6393825166038413,
      "grad_norm": 2.003800392150879,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 7124
    },
    {
      "epoch": 0.6394722670974691,
      "grad_norm": 3.3782880306243896,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 7125
    },
    {
      "epoch": 0.6395620175910968,
      "grad_norm": 2.0996384620666504,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 7126
    },
    {
      "epoch": 0.6396517680847245,
      "grad_norm": 2.098288059234619,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 7127
    },
    {
      "epoch": 0.6397415185783522,
      "grad_norm": 2.633216619491577,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 7128
    },
    {
      "epoch": 0.6398312690719798,
      "grad_norm": 2.561566114425659,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 7129
    },
    {
      "epoch": 0.6399210195656077,
      "grad_norm": 3.320003032684326,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 7130
    },
    {
      "epoch": 0.6400107700592353,
      "grad_norm": 2.8442418575286865,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 7131
    },
    {
      "epoch": 0.640100520552863,
      "grad_norm": 2.3149635791778564,
      "learning_rate": 0.0,
      "loss": 0.6911,
      "step": 7132
    },
    {
      "epoch": 0.6401902710464907,
      "grad_norm": 2.7561681270599365,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 7133
    },
    {
      "epoch": 0.6402800215401184,
      "grad_norm": 2.4706549644470215,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 7134
    },
    {
      "epoch": 0.6403697720337462,
      "grad_norm": 2.9105637073516846,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 7135
    },
    {
      "epoch": 0.6404595225273739,
      "grad_norm": 2.3908514976501465,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 7136
    },
    {
      "epoch": 0.6405492730210016,
      "grad_norm": 2.4535768032073975,
      "learning_rate": 0.0,
      "loss": 0.7612,
      "step": 7137
    },
    {
      "epoch": 0.6406390235146293,
      "grad_norm": 2.2796595096588135,
      "learning_rate": 0.0,
      "loss": 0.6818,
      "step": 7138
    },
    {
      "epoch": 0.640728774008257,
      "grad_norm": 2.835094690322876,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 7139
    },
    {
      "epoch": 0.6408185245018848,
      "grad_norm": 3.1783909797668457,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 7140
    },
    {
      "epoch": 0.6409082749955125,
      "grad_norm": 2.1973249912261963,
      "learning_rate": 0.0,
      "loss": 0.6919,
      "step": 7141
    },
    {
      "epoch": 0.6409980254891402,
      "grad_norm": 2.4487462043762207,
      "learning_rate": 0.0,
      "loss": 0.6562,
      "step": 7142
    },
    {
      "epoch": 0.6410877759827679,
      "grad_norm": 2.84051775932312,
      "learning_rate": 0.0,
      "loss": 0.7633,
      "step": 7143
    },
    {
      "epoch": 0.6411775264763956,
      "grad_norm": 2.4980642795562744,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 7144
    },
    {
      "epoch": 0.6412672769700233,
      "grad_norm": 2.369161367416382,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 7145
    },
    {
      "epoch": 0.6413570274636511,
      "grad_norm": 2.601698398590088,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 7146
    },
    {
      "epoch": 0.6414467779572788,
      "grad_norm": 2.5845413208007812,
      "learning_rate": 0.0,
      "loss": 0.7072,
      "step": 7147
    },
    {
      "epoch": 0.6415365284509065,
      "grad_norm": 3.2835755348205566,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 7148
    },
    {
      "epoch": 0.6416262789445342,
      "grad_norm": 3.3743844032287598,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 7149
    },
    {
      "epoch": 0.6417160294381619,
      "grad_norm": 4.885948181152344,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 7150
    },
    {
      "epoch": 0.6418057799317897,
      "grad_norm": 4.775932788848877,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 7151
    },
    {
      "epoch": 0.6418955304254174,
      "grad_norm": 2.4353160858154297,
      "learning_rate": 0.0,
      "loss": 0.7582,
      "step": 7152
    },
    {
      "epoch": 0.641985280919045,
      "grad_norm": 3.0437989234924316,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 7153
    },
    {
      "epoch": 0.6420750314126727,
      "grad_norm": 2.981780767440796,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 7154
    },
    {
      "epoch": 0.6421647819063004,
      "grad_norm": 3.3072316646575928,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 7155
    },
    {
      "epoch": 0.6422545323999282,
      "grad_norm": 3.6027543544769287,
      "learning_rate": 0.0,
      "loss": 0.7337,
      "step": 7156
    },
    {
      "epoch": 0.6423442828935559,
      "grad_norm": 3.0616648197174072,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 7157
    },
    {
      "epoch": 0.6424340333871836,
      "grad_norm": 3.0677130222320557,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 7158
    },
    {
      "epoch": 0.6425237838808113,
      "grad_norm": 3.4752116203308105,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 7159
    },
    {
      "epoch": 0.642613534374439,
      "grad_norm": 2.613983631134033,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 7160
    },
    {
      "epoch": 0.6427032848680668,
      "grad_norm": 4.161665439605713,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 7161
    },
    {
      "epoch": 0.6427930353616945,
      "grad_norm": 2.784311056137085,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 7162
    },
    {
      "epoch": 0.6428827858553222,
      "grad_norm": 2.8259143829345703,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 7163
    },
    {
      "epoch": 0.6429725363489499,
      "grad_norm": 6.123075485229492,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 7164
    },
    {
      "epoch": 0.6430622868425776,
      "grad_norm": 3.408677816390991,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 7165
    },
    {
      "epoch": 0.6431520373362053,
      "grad_norm": 4.475594997406006,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 7166
    },
    {
      "epoch": 0.6432417878298331,
      "grad_norm": 2.8023040294647217,
      "learning_rate": 0.0,
      "loss": 0.6751,
      "step": 7167
    },
    {
      "epoch": 0.6433315383234608,
      "grad_norm": 2.7312238216400146,
      "learning_rate": 0.0,
      "loss": 0.6824,
      "step": 7168
    },
    {
      "epoch": 0.6434212888170885,
      "grad_norm": 4.04923677444458,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 7169
    },
    {
      "epoch": 0.6435110393107162,
      "grad_norm": 3.6022181510925293,
      "learning_rate": 0.0,
      "loss": 0.7728,
      "step": 7170
    },
    {
      "epoch": 0.6436007898043439,
      "grad_norm": 2.7336313724517822,
      "learning_rate": 0.0,
      "loss": 0.6389,
      "step": 7171
    },
    {
      "epoch": 0.6436905402979717,
      "grad_norm": 2.925386905670166,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 7172
    },
    {
      "epoch": 0.6437802907915994,
      "grad_norm": 2.713660478591919,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 7173
    },
    {
      "epoch": 0.6438700412852271,
      "grad_norm": 2.9334218502044678,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 7174
    },
    {
      "epoch": 0.6439597917788548,
      "grad_norm": 2.267019033432007,
      "learning_rate": 0.0,
      "loss": 0.6805,
      "step": 7175
    },
    {
      "epoch": 0.6440495422724825,
      "grad_norm": 2.695991039276123,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 7176
    },
    {
      "epoch": 0.6441392927661103,
      "grad_norm": 4.5022196769714355,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 7177
    },
    {
      "epoch": 0.644229043259738,
      "grad_norm": 5.238496780395508,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 7178
    },
    {
      "epoch": 0.6443187937533656,
      "grad_norm": 3.8506438732147217,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 7179
    },
    {
      "epoch": 0.6444085442469933,
      "grad_norm": 2.2392971515655518,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 7180
    },
    {
      "epoch": 0.644498294740621,
      "grad_norm": 2.6366302967071533,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 7181
    },
    {
      "epoch": 0.6445880452342488,
      "grad_norm": 2.7831010818481445,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 7182
    },
    {
      "epoch": 0.6446777957278765,
      "grad_norm": 2.62290620803833,
      "learning_rate": 0.0,
      "loss": 0.6901,
      "step": 7183
    },
    {
      "epoch": 0.6447675462215042,
      "grad_norm": 2.332365036010742,
      "learning_rate": 0.0,
      "loss": 0.7727,
      "step": 7184
    },
    {
      "epoch": 0.6448572967151319,
      "grad_norm": 2.8138318061828613,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 7185
    },
    {
      "epoch": 0.6449470472087596,
      "grad_norm": 3.65425968170166,
      "learning_rate": 0.0,
      "loss": 0.7336,
      "step": 7186
    },
    {
      "epoch": 0.6450367977023873,
      "grad_norm": 3.4180402755737305,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 7187
    },
    {
      "epoch": 0.6451265481960151,
      "grad_norm": 2.743685483932495,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 7188
    },
    {
      "epoch": 0.6452162986896428,
      "grad_norm": 3.128237724304199,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 7189
    },
    {
      "epoch": 0.6453060491832705,
      "grad_norm": 2.9731788635253906,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 7190
    },
    {
      "epoch": 0.6453957996768982,
      "grad_norm": 2.864598512649536,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 7191
    },
    {
      "epoch": 0.6454855501705259,
      "grad_norm": 12.431108474731445,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 7192
    },
    {
      "epoch": 0.6455753006641537,
      "grad_norm": 3.5353360176086426,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 7193
    },
    {
      "epoch": 0.6456650511577814,
      "grad_norm": 3.304945230484009,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 7194
    },
    {
      "epoch": 0.6457548016514091,
      "grad_norm": 5.637900352478027,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 7195
    },
    {
      "epoch": 0.6458445521450368,
      "grad_norm": 3.041341543197632,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 7196
    },
    {
      "epoch": 0.6459343026386645,
      "grad_norm": 6.385684967041016,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 7197
    },
    {
      "epoch": 0.6460240531322923,
      "grad_norm": 5.178423881530762,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 7198
    },
    {
      "epoch": 0.64611380362592,
      "grad_norm": 2.818721055984497,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 7199
    },
    {
      "epoch": 0.6462035541195477,
      "grad_norm": 2.8085439205169678,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 7200
    },
    {
      "epoch": 0.6462933046131754,
      "grad_norm": 3.633373737335205,
      "learning_rate": 0.0,
      "loss": 0.7014,
      "step": 7201
    },
    {
      "epoch": 0.646383055106803,
      "grad_norm": 3.0209414958953857,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 7202
    },
    {
      "epoch": 0.6464728056004309,
      "grad_norm": 3.007110357284546,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 7203
    },
    {
      "epoch": 0.6465625560940585,
      "grad_norm": 3.201486110687256,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7204
    },
    {
      "epoch": 0.6466523065876862,
      "grad_norm": 2.7895185947418213,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 7205
    },
    {
      "epoch": 0.6467420570813139,
      "grad_norm": 3.7741262912750244,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 7206
    },
    {
      "epoch": 0.6468318075749416,
      "grad_norm": 3.6893997192382812,
      "learning_rate": 0.0,
      "loss": 0.6989,
      "step": 7207
    },
    {
      "epoch": 0.6469215580685693,
      "grad_norm": 2.7274932861328125,
      "learning_rate": 0.0,
      "loss": 0.74,
      "step": 7208
    },
    {
      "epoch": 0.6470113085621971,
      "grad_norm": 2.9894192218780518,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 7209
    },
    {
      "epoch": 0.6471010590558248,
      "grad_norm": 10.370125770568848,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 7210
    },
    {
      "epoch": 0.6471908095494525,
      "grad_norm": 3.3338592052459717,
      "learning_rate": 0.0,
      "loss": 0.6537,
      "step": 7211
    },
    {
      "epoch": 0.6472805600430802,
      "grad_norm": 34.1961555480957,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 7212
    },
    {
      "epoch": 0.6473703105367079,
      "grad_norm": 3.3027477264404297,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 7213
    },
    {
      "epoch": 0.6474600610303357,
      "grad_norm": 3.1320130825042725,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 7214
    },
    {
      "epoch": 0.6475498115239634,
      "grad_norm": 3.859299421310425,
      "learning_rate": 0.0,
      "loss": 0.848,
      "step": 7215
    },
    {
      "epoch": 0.6476395620175911,
      "grad_norm": 2.6606266498565674,
      "learning_rate": 0.0,
      "loss": 0.6597,
      "step": 7216
    },
    {
      "epoch": 0.6477293125112188,
      "grad_norm": 2.2754390239715576,
      "learning_rate": 0.0,
      "loss": 0.7974,
      "step": 7217
    },
    {
      "epoch": 0.6478190630048465,
      "grad_norm": 2.15962553024292,
      "learning_rate": 0.0,
      "loss": 0.7619,
      "step": 7218
    },
    {
      "epoch": 0.6479088134984743,
      "grad_norm": 5.42412805557251,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7219
    },
    {
      "epoch": 0.647998563992102,
      "grad_norm": 2.6067090034484863,
      "learning_rate": 0.0,
      "loss": 0.7709,
      "step": 7220
    },
    {
      "epoch": 0.6480883144857297,
      "grad_norm": 2.39558744430542,
      "learning_rate": 0.0,
      "loss": 0.8072,
      "step": 7221
    },
    {
      "epoch": 0.6481780649793574,
      "grad_norm": 2.224372386932373,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 7222
    },
    {
      "epoch": 0.6482678154729851,
      "grad_norm": 2.1806206703186035,
      "learning_rate": 0.0,
      "loss": 0.6947,
      "step": 7223
    },
    {
      "epoch": 0.6483575659666129,
      "grad_norm": 2.472092390060425,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 7224
    },
    {
      "epoch": 0.6484473164602406,
      "grad_norm": 3.2635085582733154,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 7225
    },
    {
      "epoch": 0.6485370669538683,
      "grad_norm": 3.209360361099243,
      "learning_rate": 0.0,
      "loss": 0.7878,
      "step": 7226
    },
    {
      "epoch": 0.648626817447496,
      "grad_norm": 3.725297451019287,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 7227
    },
    {
      "epoch": 0.6487165679411236,
      "grad_norm": 2.9757871627807617,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 7228
    },
    {
      "epoch": 0.6488063184347513,
      "grad_norm": 3.019202470779419,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 7229
    },
    {
      "epoch": 0.6488960689283791,
      "grad_norm": 2.320435047149658,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 7230
    },
    {
      "epoch": 0.6489858194220068,
      "grad_norm": 2.7701313495635986,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 7231
    },
    {
      "epoch": 0.6490755699156345,
      "grad_norm": 2.707324504852295,
      "learning_rate": 0.0,
      "loss": 0.6722,
      "step": 7232
    },
    {
      "epoch": 0.6491653204092622,
      "grad_norm": 2.5964596271514893,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 7233
    },
    {
      "epoch": 0.6492550709028899,
      "grad_norm": 2.05905818939209,
      "learning_rate": 0.0,
      "loss": 0.7207,
      "step": 7234
    },
    {
      "epoch": 0.6493448213965177,
      "grad_norm": 7.513299465179443,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 7235
    },
    {
      "epoch": 0.6494345718901454,
      "grad_norm": 3.885087251663208,
      "learning_rate": 0.0,
      "loss": 0.7465,
      "step": 7236
    },
    {
      "epoch": 0.6495243223837731,
      "grad_norm": 2.886432409286499,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 7237
    },
    {
      "epoch": 0.6496140728774008,
      "grad_norm": 2.207913637161255,
      "learning_rate": 0.0,
      "loss": 0.666,
      "step": 7238
    },
    {
      "epoch": 0.6497038233710285,
      "grad_norm": 3.0165340900421143,
      "learning_rate": 0.0,
      "loss": 0.6716,
      "step": 7239
    },
    {
      "epoch": 0.6497935738646563,
      "grad_norm": 3.8279693126678467,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 7240
    },
    {
      "epoch": 0.649883324358284,
      "grad_norm": 2.7278876304626465,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 7241
    },
    {
      "epoch": 0.6499730748519117,
      "grad_norm": 2.9313266277313232,
      "learning_rate": 0.0,
      "loss": 0.6958,
      "step": 7242
    },
    {
      "epoch": 0.6500628253455394,
      "grad_norm": 3.000473976135254,
      "learning_rate": 0.0,
      "loss": 0.7176,
      "step": 7243
    },
    {
      "epoch": 0.6501525758391671,
      "grad_norm": 3.616934061050415,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 7244
    },
    {
      "epoch": 0.6502423263327949,
      "grad_norm": 2.2087996006011963,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 7245
    },
    {
      "epoch": 0.6503320768264226,
      "grad_norm": 2.186915159225464,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 7246
    },
    {
      "epoch": 0.6504218273200503,
      "grad_norm": 3.6917364597320557,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 7247
    },
    {
      "epoch": 0.650511577813678,
      "grad_norm": 2.7924396991729736,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 7248
    },
    {
      "epoch": 0.6506013283073057,
      "grad_norm": 2.9675705432891846,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 7249
    },
    {
      "epoch": 0.6506910788009334,
      "grad_norm": 2.2017621994018555,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 7250
    },
    {
      "epoch": 0.6507808292945612,
      "grad_norm": 2.1991567611694336,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 7251
    },
    {
      "epoch": 0.6508705797881889,
      "grad_norm": 2.428194522857666,
      "learning_rate": 0.0,
      "loss": 0.7741,
      "step": 7252
    },
    {
      "epoch": 0.6509603302818165,
      "grad_norm": 2.112151622772217,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 7253
    },
    {
      "epoch": 0.6510500807754442,
      "grad_norm": 3.4863905906677246,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 7254
    },
    {
      "epoch": 0.6511398312690719,
      "grad_norm": 2.868086338043213,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 7255
    },
    {
      "epoch": 0.6512295817626997,
      "grad_norm": 4.0825371742248535,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 7256
    },
    {
      "epoch": 0.6513193322563274,
      "grad_norm": 3.122235059738159,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 7257
    },
    {
      "epoch": 0.6514090827499551,
      "grad_norm": 2.503136396408081,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 7258
    },
    {
      "epoch": 0.6514988332435828,
      "grad_norm": 2.6937007904052734,
      "learning_rate": 0.0,
      "loss": 0.7476,
      "step": 7259
    },
    {
      "epoch": 0.6515885837372105,
      "grad_norm": 3.389299154281616,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 7260
    },
    {
      "epoch": 0.6516783342308383,
      "grad_norm": 3.395704746246338,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 7261
    },
    {
      "epoch": 0.651768084724466,
      "grad_norm": 2.8957808017730713,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 7262
    },
    {
      "epoch": 0.6518578352180937,
      "grad_norm": 2.285480499267578,
      "learning_rate": 0.0,
      "loss": 0.7593,
      "step": 7263
    },
    {
      "epoch": 0.6519475857117214,
      "grad_norm": 3.4971091747283936,
      "learning_rate": 0.0,
      "loss": 0.6598,
      "step": 7264
    },
    {
      "epoch": 0.6520373362053491,
      "grad_norm": 3.6081554889678955,
      "learning_rate": 0.0,
      "loss": 0.7681,
      "step": 7265
    },
    {
      "epoch": 0.6521270866989769,
      "grad_norm": 2.5995328426361084,
      "learning_rate": 0.0,
      "loss": 0.7062,
      "step": 7266
    },
    {
      "epoch": 0.6522168371926046,
      "grad_norm": 4.144601821899414,
      "learning_rate": 0.0,
      "loss": 0.6859,
      "step": 7267
    },
    {
      "epoch": 0.6523065876862323,
      "grad_norm": 2.723515748977661,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 7268
    },
    {
      "epoch": 0.65239633817986,
      "grad_norm": 2.3202803134918213,
      "learning_rate": 0.0,
      "loss": 0.7848,
      "step": 7269
    },
    {
      "epoch": 0.6524860886734877,
      "grad_norm": 2.5862700939178467,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 7270
    },
    {
      "epoch": 0.6525758391671154,
      "grad_norm": 2.609866142272949,
      "learning_rate": 0.0,
      "loss": 0.7782,
      "step": 7271
    },
    {
      "epoch": 0.6526655896607432,
      "grad_norm": 3.2436437606811523,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 7272
    },
    {
      "epoch": 0.6527553401543709,
      "grad_norm": 4.701066493988037,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 7273
    },
    {
      "epoch": 0.6528450906479986,
      "grad_norm": 25.16753387451172,
      "learning_rate": 0.0,
      "loss": 0.7296,
      "step": 7274
    },
    {
      "epoch": 0.6529348411416263,
      "grad_norm": 2.790351152420044,
      "learning_rate": 0.0,
      "loss": 0.7744,
      "step": 7275
    },
    {
      "epoch": 0.653024591635254,
      "grad_norm": 3.366513729095459,
      "learning_rate": 0.0,
      "loss": 0.6727,
      "step": 7276
    },
    {
      "epoch": 0.6531143421288818,
      "grad_norm": 2.8892884254455566,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 7277
    },
    {
      "epoch": 0.6532040926225094,
      "grad_norm": 6.380153179168701,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 7278
    },
    {
      "epoch": 0.6532938431161371,
      "grad_norm": 3.1897459030151367,
      "learning_rate": 0.0,
      "loss": 0.7592,
      "step": 7279
    },
    {
      "epoch": 0.6533835936097648,
      "grad_norm": 3.147310495376587,
      "learning_rate": 0.0,
      "loss": 0.7519,
      "step": 7280
    },
    {
      "epoch": 0.6534733441033925,
      "grad_norm": 2.527846336364746,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 7281
    },
    {
      "epoch": 0.6535630945970203,
      "grad_norm": 3.1337687969207764,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 7282
    },
    {
      "epoch": 0.653652845090648,
      "grad_norm": 2.6182632446289062,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 7283
    },
    {
      "epoch": 0.6537425955842757,
      "grad_norm": 2.6405107975006104,
      "learning_rate": 0.0,
      "loss": 0.6766,
      "step": 7284
    },
    {
      "epoch": 0.6538323460779034,
      "grad_norm": 2.116971731185913,
      "learning_rate": 0.0,
      "loss": 0.6781,
      "step": 7285
    },
    {
      "epoch": 0.6539220965715311,
      "grad_norm": 2.268190622329712,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 7286
    },
    {
      "epoch": 0.6540118470651589,
      "grad_norm": 6.288534164428711,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 7287
    },
    {
      "epoch": 0.6541015975587866,
      "grad_norm": 4.238353729248047,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 7288
    },
    {
      "epoch": 0.6541913480524143,
      "grad_norm": 2.302438735961914,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 7289
    },
    {
      "epoch": 0.654281098546042,
      "grad_norm": 2.752917528152466,
      "learning_rate": 0.0,
      "loss": 0.6766,
      "step": 7290
    },
    {
      "epoch": 0.6543708490396697,
      "grad_norm": 2.247159481048584,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 7291
    },
    {
      "epoch": 0.6544605995332974,
      "grad_norm": 3.14030122756958,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 7292
    },
    {
      "epoch": 0.6545503500269252,
      "grad_norm": 2.399671792984009,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 7293
    },
    {
      "epoch": 0.6546401005205529,
      "grad_norm": 2.608414888381958,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 7294
    },
    {
      "epoch": 0.6547298510141806,
      "grad_norm": 2.317948818206787,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 7295
    },
    {
      "epoch": 0.6548196015078083,
      "grad_norm": 3.403123140335083,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 7296
    },
    {
      "epoch": 0.654909352001436,
      "grad_norm": 2.6372950077056885,
      "learning_rate": 0.0,
      "loss": 0.7639,
      "step": 7297
    },
    {
      "epoch": 0.6549991024950638,
      "grad_norm": 5.5348920822143555,
      "learning_rate": 0.0,
      "loss": 0.7686,
      "step": 7298
    },
    {
      "epoch": 0.6550888529886915,
      "grad_norm": 2.5892744064331055,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 7299
    },
    {
      "epoch": 0.6551786034823192,
      "grad_norm": 2.664428234100342,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 7300
    },
    {
      "epoch": 0.6552683539759468,
      "grad_norm": 2.0178213119506836,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 7301
    },
    {
      "epoch": 0.6553581044695745,
      "grad_norm": 12.872247695922852,
      "learning_rate": 0.0,
      "loss": 0.7813,
      "step": 7302
    },
    {
      "epoch": 0.6554478549632023,
      "grad_norm": 3.443101167678833,
      "learning_rate": 0.0,
      "loss": 0.7538,
      "step": 7303
    },
    {
      "epoch": 0.65553760545683,
      "grad_norm": 3.0492265224456787,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 7304
    },
    {
      "epoch": 0.6556273559504577,
      "grad_norm": 2.6077682971954346,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 7305
    },
    {
      "epoch": 0.6557171064440854,
      "grad_norm": 3.0159542560577393,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 7306
    },
    {
      "epoch": 0.6558068569377131,
      "grad_norm": 3.6509621143341064,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 7307
    },
    {
      "epoch": 0.6558966074313409,
      "grad_norm": 2.98874831199646,
      "learning_rate": 0.0,
      "loss": 0.6848,
      "step": 7308
    },
    {
      "epoch": 0.6559863579249686,
      "grad_norm": 2.8193013668060303,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 7309
    },
    {
      "epoch": 0.6560761084185963,
      "grad_norm": 3.534341812133789,
      "learning_rate": 0.0,
      "loss": 0.7587,
      "step": 7310
    },
    {
      "epoch": 0.656165858912224,
      "grad_norm": 3.200162887573242,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 7311
    },
    {
      "epoch": 0.6562556094058517,
      "grad_norm": 2.5223121643066406,
      "learning_rate": 0.0,
      "loss": 0.704,
      "step": 7312
    },
    {
      "epoch": 0.6563453598994794,
      "grad_norm": 2.664008855819702,
      "learning_rate": 0.0,
      "loss": 0.6658,
      "step": 7313
    },
    {
      "epoch": 0.6564351103931072,
      "grad_norm": 2.527669906616211,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 7314
    },
    {
      "epoch": 0.6565248608867349,
      "grad_norm": 3.2261428833007812,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 7315
    },
    {
      "epoch": 0.6566146113803626,
      "grad_norm": 2.488076686859131,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 7316
    },
    {
      "epoch": 0.6567043618739903,
      "grad_norm": 2.8369383811950684,
      "learning_rate": 0.0,
      "loss": 0.7708,
      "step": 7317
    },
    {
      "epoch": 0.656794112367618,
      "grad_norm": 2.819239616394043,
      "learning_rate": 0.0,
      "loss": 0.7836,
      "step": 7318
    },
    {
      "epoch": 0.6568838628612458,
      "grad_norm": 2.1175010204315186,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 7319
    },
    {
      "epoch": 0.6569736133548735,
      "grad_norm": 3.7163949012756348,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 7320
    },
    {
      "epoch": 0.6570633638485012,
      "grad_norm": 2.705934762954712,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 7321
    },
    {
      "epoch": 0.6571531143421289,
      "grad_norm": 6.73972225189209,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 7322
    },
    {
      "epoch": 0.6572428648357566,
      "grad_norm": 2.915966272354126,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 7323
    },
    {
      "epoch": 0.6573326153293844,
      "grad_norm": 3.5485987663269043,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 7324
    },
    {
      "epoch": 0.6574223658230121,
      "grad_norm": 2.870558500289917,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 7325
    },
    {
      "epoch": 0.6575121163166397,
      "grad_norm": 2.6952247619628906,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 7326
    },
    {
      "epoch": 0.6576018668102674,
      "grad_norm": 2.614903688430786,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 7327
    },
    {
      "epoch": 0.6576916173038951,
      "grad_norm": 3.195343494415283,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 7328
    },
    {
      "epoch": 0.6577813677975228,
      "grad_norm": 5.7511091232299805,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 7329
    },
    {
      "epoch": 0.6578711182911506,
      "grad_norm": 2.9798481464385986,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 7330
    },
    {
      "epoch": 0.6579608687847783,
      "grad_norm": 2.6666553020477295,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 7331
    },
    {
      "epoch": 0.658050619278406,
      "grad_norm": 3.536102771759033,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 7332
    },
    {
      "epoch": 0.6581403697720337,
      "grad_norm": 4.50950813293457,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 7333
    },
    {
      "epoch": 0.6582301202656614,
      "grad_norm": 2.4933927059173584,
      "learning_rate": 0.0,
      "loss": 0.7414,
      "step": 7334
    },
    {
      "epoch": 0.6583198707592892,
      "grad_norm": 2.414320707321167,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 7335
    },
    {
      "epoch": 0.6584096212529169,
      "grad_norm": 2.6061899662017822,
      "learning_rate": 0.0,
      "loss": 0.6789,
      "step": 7336
    },
    {
      "epoch": 0.6584993717465446,
      "grad_norm": 3.5939924716949463,
      "learning_rate": 0.0,
      "loss": 0.6815,
      "step": 7337
    },
    {
      "epoch": 0.6585891222401723,
      "grad_norm": 3.7187349796295166,
      "learning_rate": 0.0,
      "loss": 0.6922,
      "step": 7338
    },
    {
      "epoch": 0.6586788727338,
      "grad_norm": 3.1075618267059326,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 7339
    },
    {
      "epoch": 0.6587686232274278,
      "grad_norm": 2.228606700897217,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 7340
    },
    {
      "epoch": 0.6588583737210555,
      "grad_norm": 2.3287551403045654,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 7341
    },
    {
      "epoch": 0.6589481242146832,
      "grad_norm": 3.2140908241271973,
      "learning_rate": 0.0,
      "loss": 0.7915,
      "step": 7342
    },
    {
      "epoch": 0.6590378747083109,
      "grad_norm": 3.352555751800537,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 7343
    },
    {
      "epoch": 0.6591276252019386,
      "grad_norm": 3.3425636291503906,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 7344
    },
    {
      "epoch": 0.6592173756955664,
      "grad_norm": 3.83072566986084,
      "learning_rate": 0.0,
      "loss": 0.6607,
      "step": 7345
    },
    {
      "epoch": 0.6593071261891941,
      "grad_norm": 2.668250799179077,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 7346
    },
    {
      "epoch": 0.6593968766828218,
      "grad_norm": 2.750342845916748,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 7347
    },
    {
      "epoch": 0.6594866271764495,
      "grad_norm": 2.2790589332580566,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 7348
    },
    {
      "epoch": 0.6595763776700772,
      "grad_norm": 2.319772243499756,
      "learning_rate": 0.0,
      "loss": 0.6896,
      "step": 7349
    },
    {
      "epoch": 0.6596661281637048,
      "grad_norm": 3.157015800476074,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 7350
    },
    {
      "epoch": 0.6597558786573327,
      "grad_norm": 3.0792267322540283,
      "learning_rate": 0.0,
      "loss": 0.7026,
      "step": 7351
    },
    {
      "epoch": 0.6598456291509603,
      "grad_norm": 2.397486448287964,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 7352
    },
    {
      "epoch": 0.659935379644588,
      "grad_norm": 2.7114527225494385,
      "learning_rate": 0.0,
      "loss": 0.7666,
      "step": 7353
    },
    {
      "epoch": 0.6600251301382157,
      "grad_norm": 2.611905336380005,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 7354
    },
    {
      "epoch": 0.6601148806318434,
      "grad_norm": 2.57853102684021,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 7355
    },
    {
      "epoch": 0.6602046311254712,
      "grad_norm": 3.4673736095428467,
      "learning_rate": 0.0,
      "loss": 0.7649,
      "step": 7356
    },
    {
      "epoch": 0.6602943816190989,
      "grad_norm": 5.88952112197876,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 7357
    },
    {
      "epoch": 0.6603841321127266,
      "grad_norm": 2.834033250808716,
      "learning_rate": 0.0,
      "loss": 0.7622,
      "step": 7358
    },
    {
      "epoch": 0.6604738826063543,
      "grad_norm": 2.6810569763183594,
      "learning_rate": 0.0,
      "loss": 0.7862,
      "step": 7359
    },
    {
      "epoch": 0.660563633099982,
      "grad_norm": 3.932321548461914,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 7360
    },
    {
      "epoch": 0.6606533835936098,
      "grad_norm": 3.1212449073791504,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 7361
    },
    {
      "epoch": 0.6607431340872375,
      "grad_norm": 2.569887399673462,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 7362
    },
    {
      "epoch": 0.6608328845808652,
      "grad_norm": 3.474611759185791,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 7363
    },
    {
      "epoch": 0.6609226350744929,
      "grad_norm": 2.498011350631714,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 7364
    },
    {
      "epoch": 0.6610123855681206,
      "grad_norm": 2.598905086517334,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 7365
    },
    {
      "epoch": 0.6611021360617484,
      "grad_norm": 2.906644344329834,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 7366
    },
    {
      "epoch": 0.6611918865553761,
      "grad_norm": 2.129781484603882,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 7367
    },
    {
      "epoch": 0.6612816370490038,
      "grad_norm": 2.6507160663604736,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 7368
    },
    {
      "epoch": 0.6613713875426315,
      "grad_norm": 3.735948085784912,
      "learning_rate": 0.0,
      "loss": 0.746,
      "step": 7369
    },
    {
      "epoch": 0.6614611380362592,
      "grad_norm": 2.9311306476593018,
      "learning_rate": 0.0,
      "loss": 0.7946,
      "step": 7370
    },
    {
      "epoch": 0.6615508885298869,
      "grad_norm": 2.987431526184082,
      "learning_rate": 0.0,
      "loss": 0.6728,
      "step": 7371
    },
    {
      "epoch": 0.6616406390235147,
      "grad_norm": 2.9374570846557617,
      "learning_rate": 0.0,
      "loss": 0.6667,
      "step": 7372
    },
    {
      "epoch": 0.6617303895171424,
      "grad_norm": 3.381573438644409,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 7373
    },
    {
      "epoch": 0.66182014001077,
      "grad_norm": 2.7969701290130615,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 7374
    },
    {
      "epoch": 0.6619098905043977,
      "grad_norm": 2.1233294010162354,
      "learning_rate": 0.0,
      "loss": 0.7538,
      "step": 7375
    },
    {
      "epoch": 0.6619996409980254,
      "grad_norm": 2.357848644256592,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 7376
    },
    {
      "epoch": 0.6620893914916532,
      "grad_norm": 3.05393385887146,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 7377
    },
    {
      "epoch": 0.6621791419852809,
      "grad_norm": 2.3827900886535645,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 7378
    },
    {
      "epoch": 0.6622688924789086,
      "grad_norm": 2.4606456756591797,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 7379
    },
    {
      "epoch": 0.6623586429725363,
      "grad_norm": 4.486944675445557,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 7380
    },
    {
      "epoch": 0.662448393466164,
      "grad_norm": 2.7220709323883057,
      "learning_rate": 0.0,
      "loss": 0.6631,
      "step": 7381
    },
    {
      "epoch": 0.6625381439597918,
      "grad_norm": 2.7369916439056396,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 7382
    },
    {
      "epoch": 0.6626278944534195,
      "grad_norm": 2.6900041103363037,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 7383
    },
    {
      "epoch": 0.6627176449470472,
      "grad_norm": 3.1623125076293945,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 7384
    },
    {
      "epoch": 0.6628073954406749,
      "grad_norm": 3.253995418548584,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7385
    },
    {
      "epoch": 0.6628971459343026,
      "grad_norm": 2.9998934268951416,
      "learning_rate": 0.0,
      "loss": 0.6778,
      "step": 7386
    },
    {
      "epoch": 0.6629868964279304,
      "grad_norm": 3.1058273315429688,
      "learning_rate": 0.0,
      "loss": 0.7478,
      "step": 7387
    },
    {
      "epoch": 0.6630766469215581,
      "grad_norm": 3.2055835723876953,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 7388
    },
    {
      "epoch": 0.6631663974151858,
      "grad_norm": 3.390859603881836,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 7389
    },
    {
      "epoch": 0.6632561479088135,
      "grad_norm": 4.60214900970459,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 7390
    },
    {
      "epoch": 0.6633458984024412,
      "grad_norm": 2.8449018001556396,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 7391
    },
    {
      "epoch": 0.6634356488960689,
      "grad_norm": 2.609583854675293,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 7392
    },
    {
      "epoch": 0.6635253993896967,
      "grad_norm": 2.515352249145508,
      "learning_rate": 0.0,
      "loss": 0.7034,
      "step": 7393
    },
    {
      "epoch": 0.6636151498833244,
      "grad_norm": 2.84344744682312,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 7394
    },
    {
      "epoch": 0.6637049003769521,
      "grad_norm": 2.876420259475708,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 7395
    },
    {
      "epoch": 0.6637946508705798,
      "grad_norm": 2.3094468116760254,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 7396
    },
    {
      "epoch": 0.6638844013642075,
      "grad_norm": 3.544166326522827,
      "learning_rate": 0.0,
      "loss": 0.704,
      "step": 7397
    },
    {
      "epoch": 0.6639741518578353,
      "grad_norm": 2.968043088912964,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 7398
    },
    {
      "epoch": 0.664063902351463,
      "grad_norm": 2.418647050857544,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 7399
    },
    {
      "epoch": 0.6641536528450906,
      "grad_norm": 6.604562759399414,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 7400
    },
    {
      "epoch": 0.6642434033387183,
      "grad_norm": 2.7596893310546875,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 7401
    },
    {
      "epoch": 0.664333153832346,
      "grad_norm": 3.5314998626708984,
      "learning_rate": 0.0,
      "loss": 0.6765,
      "step": 7402
    },
    {
      "epoch": 0.6644229043259738,
      "grad_norm": 3.6909403800964355,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 7403
    },
    {
      "epoch": 0.6645126548196015,
      "grad_norm": 3.8473682403564453,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 7404
    },
    {
      "epoch": 0.6646024053132292,
      "grad_norm": 2.321988582611084,
      "learning_rate": 0.0,
      "loss": 0.6658,
      "step": 7405
    },
    {
      "epoch": 0.6646921558068569,
      "grad_norm": 3.172614574432373,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 7406
    },
    {
      "epoch": 0.6647819063004846,
      "grad_norm": 2.9555771350860596,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 7407
    },
    {
      "epoch": 0.6648716567941124,
      "grad_norm": 2.840010404586792,
      "learning_rate": 0.0,
      "loss": 0.7454,
      "step": 7408
    },
    {
      "epoch": 0.6649614072877401,
      "grad_norm": 4.327831268310547,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 7409
    },
    {
      "epoch": 0.6650511577813678,
      "grad_norm": 2.9908430576324463,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 7410
    },
    {
      "epoch": 0.6651409082749955,
      "grad_norm": 2.412686347961426,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 7411
    },
    {
      "epoch": 0.6652306587686232,
      "grad_norm": 2.8995137214660645,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 7412
    },
    {
      "epoch": 0.6653204092622509,
      "grad_norm": 3.2508325576782227,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 7413
    },
    {
      "epoch": 0.6654101597558787,
      "grad_norm": 3.3290064334869385,
      "learning_rate": 0.0,
      "loss": 0.7273,
      "step": 7414
    },
    {
      "epoch": 0.6654999102495064,
      "grad_norm": 2.3704166412353516,
      "learning_rate": 0.0,
      "loss": 0.6876,
      "step": 7415
    },
    {
      "epoch": 0.6655896607431341,
      "grad_norm": 3.6969387531280518,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 7416
    },
    {
      "epoch": 0.6656794112367618,
      "grad_norm": 3.1660025119781494,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 7417
    },
    {
      "epoch": 0.6657691617303895,
      "grad_norm": 4.292861461639404,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 7418
    },
    {
      "epoch": 0.6658589122240173,
      "grad_norm": 2.8852009773254395,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 7419
    },
    {
      "epoch": 0.665948662717645,
      "grad_norm": 2.901726484298706,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 7420
    },
    {
      "epoch": 0.6660384132112727,
      "grad_norm": 2.4173038005828857,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 7421
    },
    {
      "epoch": 0.6661281637049004,
      "grad_norm": 2.2629294395446777,
      "learning_rate": 0.0,
      "loss": 0.8245,
      "step": 7422
    },
    {
      "epoch": 0.666217914198528,
      "grad_norm": 2.7482504844665527,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 7423
    },
    {
      "epoch": 0.6663076646921559,
      "grad_norm": 2.7116539478302,
      "learning_rate": 0.0,
      "loss": 0.6899,
      "step": 7424
    },
    {
      "epoch": 0.6663974151857835,
      "grad_norm": 3.9467039108276367,
      "learning_rate": 0.0,
      "loss": 0.7498,
      "step": 7425
    },
    {
      "epoch": 0.6664871656794112,
      "grad_norm": 2.9172558784484863,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 7426
    },
    {
      "epoch": 0.6665769161730389,
      "grad_norm": 3.0617258548736572,
      "learning_rate": 0.0,
      "loss": 0.7583,
      "step": 7427
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 2.3725383281707764,
      "learning_rate": 0.0,
      "loss": 0.6417,
      "step": 7428
    },
    {
      "epoch": 0.6667564171602944,
      "grad_norm": 2.952148199081421,
      "learning_rate": 0.0,
      "loss": 0.744,
      "step": 7429
    },
    {
      "epoch": 0.6668461676539221,
      "grad_norm": 3.4233312606811523,
      "learning_rate": 0.0,
      "loss": 0.7249,
      "step": 7430
    },
    {
      "epoch": 0.6669359181475498,
      "grad_norm": 4.4586100578308105,
      "learning_rate": 0.0,
      "loss": 0.6948,
      "step": 7431
    },
    {
      "epoch": 0.6670256686411775,
      "grad_norm": 2.359572649002075,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 7432
    },
    {
      "epoch": 0.6671154191348052,
      "grad_norm": 2.802811622619629,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 7433
    },
    {
      "epoch": 0.6672051696284329,
      "grad_norm": 2.4496278762817383,
      "learning_rate": 0.0,
      "loss": 0.7809,
      "step": 7434
    },
    {
      "epoch": 0.6672949201220607,
      "grad_norm": 2.3023033142089844,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 7435
    },
    {
      "epoch": 0.6673846706156884,
      "grad_norm": 2.396899700164795,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 7436
    },
    {
      "epoch": 0.6674744211093161,
      "grad_norm": 3.3337461948394775,
      "learning_rate": 0.0,
      "loss": 0.6852,
      "step": 7437
    },
    {
      "epoch": 0.6675641716029438,
      "grad_norm": 2.6991066932678223,
      "learning_rate": 0.0,
      "loss": 0.6781,
      "step": 7438
    },
    {
      "epoch": 0.6676539220965715,
      "grad_norm": 2.465559244155884,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 7439
    },
    {
      "epoch": 0.6677436725901993,
      "grad_norm": 4.170602321624756,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 7440
    },
    {
      "epoch": 0.667833423083827,
      "grad_norm": 2.3841397762298584,
      "learning_rate": 0.0,
      "loss": 0.6918,
      "step": 7441
    },
    {
      "epoch": 0.6679231735774547,
      "grad_norm": 2.6586413383483887,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 7442
    },
    {
      "epoch": 0.6680129240710824,
      "grad_norm": 9.786980628967285,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 7443
    },
    {
      "epoch": 0.6681026745647101,
      "grad_norm": 3.6344082355499268,
      "learning_rate": 0.0,
      "loss": 0.6708,
      "step": 7444
    },
    {
      "epoch": 0.6681924250583379,
      "grad_norm": 2.63130521774292,
      "learning_rate": 0.0,
      "loss": 0.6888,
      "step": 7445
    },
    {
      "epoch": 0.6682821755519656,
      "grad_norm": 3.3605315685272217,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 7446
    },
    {
      "epoch": 0.6683719260455933,
      "grad_norm": 6.381349086761475,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 7447
    },
    {
      "epoch": 0.668461676539221,
      "grad_norm": 3.159729242324829,
      "learning_rate": 0.0,
      "loss": 0.6775,
      "step": 7448
    },
    {
      "epoch": 0.6685514270328486,
      "grad_norm": 2.8483524322509766,
      "learning_rate": 0.0,
      "loss": 0.6662,
      "step": 7449
    },
    {
      "epoch": 0.6686411775264764,
      "grad_norm": 2.363070487976074,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 7450
    },
    {
      "epoch": 0.6687309280201041,
      "grad_norm": 2.536578416824341,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 7451
    },
    {
      "epoch": 0.6688206785137318,
      "grad_norm": 3.0130844116210938,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 7452
    },
    {
      "epoch": 0.6689104290073595,
      "grad_norm": 2.6339025497436523,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 7453
    },
    {
      "epoch": 0.6690001795009872,
      "grad_norm": 2.941282272338867,
      "learning_rate": 0.0,
      "loss": 0.6846,
      "step": 7454
    },
    {
      "epoch": 0.6690899299946149,
      "grad_norm": 3.180521249771118,
      "learning_rate": 0.0,
      "loss": 0.6651,
      "step": 7455
    },
    {
      "epoch": 0.6691796804882427,
      "grad_norm": 2.633338689804077,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 7456
    },
    {
      "epoch": 0.6692694309818704,
      "grad_norm": 2.746150493621826,
      "learning_rate": 0.0,
      "loss": 0.6558,
      "step": 7457
    },
    {
      "epoch": 0.6693591814754981,
      "grad_norm": 2.34936785697937,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 7458
    },
    {
      "epoch": 0.6694489319691258,
      "grad_norm": 2.450793743133545,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 7459
    },
    {
      "epoch": 0.6695386824627535,
      "grad_norm": 2.7195630073547363,
      "learning_rate": 0.0,
      "loss": 0.6841,
      "step": 7460
    },
    {
      "epoch": 0.6696284329563813,
      "grad_norm": 2.332984209060669,
      "learning_rate": 0.0,
      "loss": 0.6215,
      "step": 7461
    },
    {
      "epoch": 0.669718183450009,
      "grad_norm": 2.544079065322876,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 7462
    },
    {
      "epoch": 0.6698079339436367,
      "grad_norm": 2.3014004230499268,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 7463
    },
    {
      "epoch": 0.6698976844372644,
      "grad_norm": 2.1890385150909424,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 7464
    },
    {
      "epoch": 0.6699874349308921,
      "grad_norm": 3.75138783454895,
      "learning_rate": 0.0,
      "loss": 0.6764,
      "step": 7465
    },
    {
      "epoch": 0.6700771854245199,
      "grad_norm": 2.982990264892578,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 7466
    },
    {
      "epoch": 0.6701669359181476,
      "grad_norm": 2.6226816177368164,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 7467
    },
    {
      "epoch": 0.6702566864117753,
      "grad_norm": 2.0841898918151855,
      "learning_rate": 0.0,
      "loss": 0.6843,
      "step": 7468
    },
    {
      "epoch": 0.670346436905403,
      "grad_norm": 2.340796947479248,
      "learning_rate": 0.0,
      "loss": 0.6848,
      "step": 7469
    },
    {
      "epoch": 0.6704361873990307,
      "grad_norm": 3.1620612144470215,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 7470
    },
    {
      "epoch": 0.6705259378926585,
      "grad_norm": 3.2206807136535645,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 7471
    },
    {
      "epoch": 0.6706156883862862,
      "grad_norm": 3.2232770919799805,
      "learning_rate": 0.0,
      "loss": 0.6872,
      "step": 7472
    },
    {
      "epoch": 0.6707054388799139,
      "grad_norm": 2.987278938293457,
      "learning_rate": 0.0,
      "loss": 0.8022,
      "step": 7473
    },
    {
      "epoch": 0.6707951893735415,
      "grad_norm": 2.5164458751678467,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 7474
    },
    {
      "epoch": 0.6708849398671692,
      "grad_norm": 2.9951963424682617,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 7475
    },
    {
      "epoch": 0.6709746903607969,
      "grad_norm": 2.8448545932769775,
      "learning_rate": 0.0,
      "loss": 0.6735,
      "step": 7476
    },
    {
      "epoch": 0.6710644408544247,
      "grad_norm": 2.1943836212158203,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 7477
    },
    {
      "epoch": 0.6711541913480524,
      "grad_norm": 3.9532010555267334,
      "learning_rate": 0.0,
      "loss": 0.7847,
      "step": 7478
    },
    {
      "epoch": 0.6712439418416801,
      "grad_norm": 2.6472299098968506,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 7479
    },
    {
      "epoch": 0.6713336923353078,
      "grad_norm": 3.426260471343994,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 7480
    },
    {
      "epoch": 0.6714234428289355,
      "grad_norm": 2.242678642272949,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 7481
    },
    {
      "epoch": 0.6715131933225633,
      "grad_norm": 3.142461061477661,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 7482
    },
    {
      "epoch": 0.671602943816191,
      "grad_norm": 2.9126174449920654,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 7483
    },
    {
      "epoch": 0.6716926943098187,
      "grad_norm": 3.5969135761260986,
      "learning_rate": 0.0,
      "loss": 0.6896,
      "step": 7484
    },
    {
      "epoch": 0.6717824448034464,
      "grad_norm": 2.896454095840454,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 7485
    },
    {
      "epoch": 0.6718721952970741,
      "grad_norm": 2.877042770385742,
      "learning_rate": 0.0,
      "loss": 0.6447,
      "step": 7486
    },
    {
      "epoch": 0.6719619457907019,
      "grad_norm": 2.7966227531433105,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 7487
    },
    {
      "epoch": 0.6720516962843296,
      "grad_norm": 2.865138292312622,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 7488
    },
    {
      "epoch": 0.6721414467779573,
      "grad_norm": 2.6836581230163574,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 7489
    },
    {
      "epoch": 0.672231197271585,
      "grad_norm": 2.7903189659118652,
      "learning_rate": 0.0,
      "loss": 0.6905,
      "step": 7490
    },
    {
      "epoch": 0.6723209477652127,
      "grad_norm": 3.5178234577178955,
      "learning_rate": 0.0,
      "loss": 0.672,
      "step": 7491
    },
    {
      "epoch": 0.6724106982588405,
      "grad_norm": 3.2172648906707764,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 7492
    },
    {
      "epoch": 0.6725004487524682,
      "grad_norm": 2.9955456256866455,
      "learning_rate": 0.0,
      "loss": 0.7505,
      "step": 7493
    },
    {
      "epoch": 0.6725901992460959,
      "grad_norm": 2.608281135559082,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 7494
    },
    {
      "epoch": 0.6726799497397236,
      "grad_norm": 2.8201587200164795,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 7495
    },
    {
      "epoch": 0.6727697002333513,
      "grad_norm": 2.772507905960083,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 7496
    },
    {
      "epoch": 0.672859450726979,
      "grad_norm": 3.2456212043762207,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 7497
    },
    {
      "epoch": 0.6729492012206068,
      "grad_norm": 4.051540851593018,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 7498
    },
    {
      "epoch": 0.6730389517142344,
      "grad_norm": 3.7970287799835205,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 7499
    },
    {
      "epoch": 0.6731287022078621,
      "grad_norm": 3.2724530696868896,
      "learning_rate": 0.0,
      "loss": 0.6743,
      "step": 7500
    },
    {
      "epoch": 0.6732184527014898,
      "grad_norm": 3.5460095405578613,
      "learning_rate": 0.0,
      "loss": 0.7658,
      "step": 7501
    },
    {
      "epoch": 0.6733082031951175,
      "grad_norm": 2.896235227584839,
      "learning_rate": 0.0,
      "loss": 0.7875,
      "step": 7502
    },
    {
      "epoch": 0.6733979536887453,
      "grad_norm": 3.084540843963623,
      "learning_rate": 0.0,
      "loss": 0.7564,
      "step": 7503
    },
    {
      "epoch": 0.673487704182373,
      "grad_norm": 3.2654473781585693,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 7504
    },
    {
      "epoch": 0.6735774546760007,
      "grad_norm": 3.3313145637512207,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 7505
    },
    {
      "epoch": 0.6736672051696284,
      "grad_norm": 3.0341572761535645,
      "learning_rate": 0.0,
      "loss": 0.7558,
      "step": 7506
    },
    {
      "epoch": 0.6737569556632561,
      "grad_norm": 3.7174158096313477,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 7507
    },
    {
      "epoch": 0.6738467061568839,
      "grad_norm": 2.8661224842071533,
      "learning_rate": 0.0,
      "loss": 0.7646,
      "step": 7508
    },
    {
      "epoch": 0.6739364566505116,
      "grad_norm": 2.5411300659179688,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 7509
    },
    {
      "epoch": 0.6740262071441393,
      "grad_norm": 4.153045177459717,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 7510
    },
    {
      "epoch": 0.674115957637767,
      "grad_norm": 2.9814300537109375,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 7511
    },
    {
      "epoch": 0.6742057081313947,
      "grad_norm": 2.4332895278930664,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 7512
    },
    {
      "epoch": 0.6742954586250225,
      "grad_norm": 2.905893087387085,
      "learning_rate": 0.0,
      "loss": 0.6614,
      "step": 7513
    },
    {
      "epoch": 0.6743852091186502,
      "grad_norm": 3.792850971221924,
      "learning_rate": 0.0,
      "loss": 0.7962,
      "step": 7514
    },
    {
      "epoch": 0.6744749596122779,
      "grad_norm": 3.7941181659698486,
      "learning_rate": 0.0,
      "loss": 0.6975,
      "step": 7515
    },
    {
      "epoch": 0.6745647101059056,
      "grad_norm": 6.739884853363037,
      "learning_rate": 0.0,
      "loss": 0.6765,
      "step": 7516
    },
    {
      "epoch": 0.6746544605995333,
      "grad_norm": 2.60776948928833,
      "learning_rate": 0.0,
      "loss": 0.6924,
      "step": 7517
    },
    {
      "epoch": 0.674744211093161,
      "grad_norm": 3.4335553646087646,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 7518
    },
    {
      "epoch": 0.6748339615867888,
      "grad_norm": 2.8614232540130615,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 7519
    },
    {
      "epoch": 0.6749237120804165,
      "grad_norm": 2.7433290481567383,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 7520
    },
    {
      "epoch": 0.6750134625740442,
      "grad_norm": 2.9615697860717773,
      "learning_rate": 0.0,
      "loss": 0.6808,
      "step": 7521
    },
    {
      "epoch": 0.6751032130676718,
      "grad_norm": 2.455846071243286,
      "learning_rate": 0.0,
      "loss": 0.6314,
      "step": 7522
    },
    {
      "epoch": 0.6751929635612995,
      "grad_norm": 9.056118965148926,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 7523
    },
    {
      "epoch": 0.6752827140549273,
      "grad_norm": 2.656398057937622,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 7524
    },
    {
      "epoch": 0.675372464548555,
      "grad_norm": 2.035905361175537,
      "learning_rate": 0.0,
      "loss": 0.6359,
      "step": 7525
    },
    {
      "epoch": 0.6754622150421827,
      "grad_norm": 2.593928575515747,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 7526
    },
    {
      "epoch": 0.6755519655358104,
      "grad_norm": 3.4778854846954346,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 7527
    },
    {
      "epoch": 0.6756417160294381,
      "grad_norm": 2.3902199268341064,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 7528
    },
    {
      "epoch": 0.6757314665230659,
      "grad_norm": 2.5345661640167236,
      "learning_rate": 0.0,
      "loss": 0.6846,
      "step": 7529
    },
    {
      "epoch": 0.6758212170166936,
      "grad_norm": 10.24842643737793,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 7530
    },
    {
      "epoch": 0.6759109675103213,
      "grad_norm": 3.2786149978637695,
      "learning_rate": 0.0,
      "loss": 0.7014,
      "step": 7531
    },
    {
      "epoch": 0.676000718003949,
      "grad_norm": 3.0125010013580322,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 7532
    },
    {
      "epoch": 0.6760904684975767,
      "grad_norm": 2.4056291580200195,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 7533
    },
    {
      "epoch": 0.6761802189912045,
      "grad_norm": 2.5189261436462402,
      "learning_rate": 0.0,
      "loss": 0.6685,
      "step": 7534
    },
    {
      "epoch": 0.6762699694848322,
      "grad_norm": 3.3575501441955566,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 7535
    },
    {
      "epoch": 0.6763597199784599,
      "grad_norm": 2.583778142929077,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 7536
    },
    {
      "epoch": 0.6764494704720876,
      "grad_norm": 2.667769432067871,
      "learning_rate": 0.0,
      "loss": 0.6561,
      "step": 7537
    },
    {
      "epoch": 0.6765392209657153,
      "grad_norm": 2.6252784729003906,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 7538
    },
    {
      "epoch": 0.676628971459343,
      "grad_norm": 2.6638505458831787,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 7539
    },
    {
      "epoch": 0.6767187219529708,
      "grad_norm": 2.9990170001983643,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 7540
    },
    {
      "epoch": 0.6768084724465985,
      "grad_norm": 2.9330596923828125,
      "learning_rate": 0.0,
      "loss": 0.747,
      "step": 7541
    },
    {
      "epoch": 0.6768982229402262,
      "grad_norm": 2.9230167865753174,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 7542
    },
    {
      "epoch": 0.6769879734338539,
      "grad_norm": 3.4150924682617188,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 7543
    },
    {
      "epoch": 0.6770777239274816,
      "grad_norm": 2.305314302444458,
      "learning_rate": 0.0,
      "loss": 0.6638,
      "step": 7544
    },
    {
      "epoch": 0.6771674744211094,
      "grad_norm": 2.5354502201080322,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 7545
    },
    {
      "epoch": 0.677257224914737,
      "grad_norm": 2.7775826454162598,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 7546
    },
    {
      "epoch": 0.6773469754083647,
      "grad_norm": 2.426976442337036,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 7547
    },
    {
      "epoch": 0.6774367259019924,
      "grad_norm": 5.556073188781738,
      "learning_rate": 0.0,
      "loss": 0.6885,
      "step": 7548
    },
    {
      "epoch": 0.6775264763956201,
      "grad_norm": 2.4901623725891113,
      "learning_rate": 0.0,
      "loss": 0.676,
      "step": 7549
    },
    {
      "epoch": 0.6776162268892479,
      "grad_norm": 3.094658136367798,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 7550
    },
    {
      "epoch": 0.6777059773828756,
      "grad_norm": 2.697313070297241,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 7551
    },
    {
      "epoch": 0.6777957278765033,
      "grad_norm": 2.6016252040863037,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 7552
    },
    {
      "epoch": 0.677885478370131,
      "grad_norm": 3.1973812580108643,
      "learning_rate": 0.0,
      "loss": 0.8158,
      "step": 7553
    },
    {
      "epoch": 0.6779752288637587,
      "grad_norm": 2.873652219772339,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 7554
    },
    {
      "epoch": 0.6780649793573864,
      "grad_norm": 2.654153823852539,
      "learning_rate": 0.0,
      "loss": 0.6421,
      "step": 7555
    },
    {
      "epoch": 0.6781547298510142,
      "grad_norm": 4.963978290557861,
      "learning_rate": 0.0,
      "loss": 0.7401,
      "step": 7556
    },
    {
      "epoch": 0.6782444803446419,
      "grad_norm": 2.4861855506896973,
      "learning_rate": 0.0,
      "loss": 0.7061,
      "step": 7557
    },
    {
      "epoch": 0.6783342308382696,
      "grad_norm": 2.728172779083252,
      "learning_rate": 0.0,
      "loss": 0.7679,
      "step": 7558
    },
    {
      "epoch": 0.6784239813318973,
      "grad_norm": 2.8712539672851562,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 7559
    },
    {
      "epoch": 0.678513731825525,
      "grad_norm": 5.084611892700195,
      "learning_rate": 0.0,
      "loss": 0.6739,
      "step": 7560
    },
    {
      "epoch": 0.6786034823191528,
      "grad_norm": 4.2306437492370605,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 7561
    },
    {
      "epoch": 0.6786932328127805,
      "grad_norm": 2.3845901489257812,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 7562
    },
    {
      "epoch": 0.6787829833064082,
      "grad_norm": 3.65588116645813,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 7563
    },
    {
      "epoch": 0.6788727338000359,
      "grad_norm": 2.480412721633911,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 7564
    },
    {
      "epoch": 0.6789624842936636,
      "grad_norm": 3.3453259468078613,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 7565
    },
    {
      "epoch": 0.6790522347872914,
      "grad_norm": 3.1979081630706787,
      "learning_rate": 0.0,
      "loss": 0.7671,
      "step": 7566
    },
    {
      "epoch": 0.6791419852809191,
      "grad_norm": 3.1346676349639893,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 7567
    },
    {
      "epoch": 0.6792317357745468,
      "grad_norm": 5.650232791900635,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 7568
    },
    {
      "epoch": 0.6793214862681745,
      "grad_norm": 3.3905107975006104,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 7569
    },
    {
      "epoch": 0.6794112367618022,
      "grad_norm": 2.4156394004821777,
      "learning_rate": 0.0,
      "loss": 0.7405,
      "step": 7570
    },
    {
      "epoch": 0.67950098725543,
      "grad_norm": 5.128472805023193,
      "learning_rate": 0.0,
      "loss": 0.7652,
      "step": 7571
    },
    {
      "epoch": 0.6795907377490576,
      "grad_norm": 2.5470128059387207,
      "learning_rate": 0.0,
      "loss": 0.6241,
      "step": 7572
    },
    {
      "epoch": 0.6796804882426853,
      "grad_norm": 3.3592636585235596,
      "learning_rate": 0.0,
      "loss": 0.6673,
      "step": 7573
    },
    {
      "epoch": 0.679770238736313,
      "grad_norm": 2.7343270778656006,
      "learning_rate": 0.0,
      "loss": 0.6569,
      "step": 7574
    },
    {
      "epoch": 0.6798599892299407,
      "grad_norm": 2.6684272289276123,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 7575
    },
    {
      "epoch": 0.6799497397235684,
      "grad_norm": 3.067899465560913,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 7576
    },
    {
      "epoch": 0.6800394902171962,
      "grad_norm": 2.9493093490600586,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 7577
    },
    {
      "epoch": 0.6801292407108239,
      "grad_norm": 2.196254014968872,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 7578
    },
    {
      "epoch": 0.6802189912044516,
      "grad_norm": 2.7025201320648193,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 7579
    },
    {
      "epoch": 0.6803087416980793,
      "grad_norm": 2.427215814590454,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 7580
    },
    {
      "epoch": 0.680398492191707,
      "grad_norm": 3.1323702335357666,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 7581
    },
    {
      "epoch": 0.6804882426853348,
      "grad_norm": 2.555325746536255,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 7582
    },
    {
      "epoch": 0.6805779931789625,
      "grad_norm": 2.7407779693603516,
      "learning_rate": 0.0,
      "loss": 0.6898,
      "step": 7583
    },
    {
      "epoch": 0.6806677436725902,
      "grad_norm": 2.8649299144744873,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 7584
    },
    {
      "epoch": 0.6807574941662179,
      "grad_norm": 3.153944969177246,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 7585
    },
    {
      "epoch": 0.6808472446598456,
      "grad_norm": 3.5497887134552,
      "learning_rate": 0.0,
      "loss": 0.7203,
      "step": 7586
    },
    {
      "epoch": 0.6809369951534734,
      "grad_norm": 6.168327331542969,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 7587
    },
    {
      "epoch": 0.6810267456471011,
      "grad_norm": 2.9662868976593018,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 7588
    },
    {
      "epoch": 0.6811164961407288,
      "grad_norm": 5.757750988006592,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 7589
    },
    {
      "epoch": 0.6812062466343565,
      "grad_norm": 3.2658021450042725,
      "learning_rate": 0.0,
      "loss": 0.7768,
      "step": 7590
    },
    {
      "epoch": 0.6812959971279842,
      "grad_norm": 2.5477068424224854,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 7591
    },
    {
      "epoch": 0.681385747621612,
      "grad_norm": 2.7732303142547607,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 7592
    },
    {
      "epoch": 0.6814754981152397,
      "grad_norm": 2.009382486343384,
      "learning_rate": 0.0,
      "loss": 0.6866,
      "step": 7593
    },
    {
      "epoch": 0.6815652486088674,
      "grad_norm": 2.4880895614624023,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 7594
    },
    {
      "epoch": 0.681654999102495,
      "grad_norm": 2.933039426803589,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 7595
    },
    {
      "epoch": 0.6817447495961227,
      "grad_norm": 2.1316614151000977,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 7596
    },
    {
      "epoch": 0.6818345000897504,
      "grad_norm": 2.781174898147583,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 7597
    },
    {
      "epoch": 0.6819242505833782,
      "grad_norm": 3.635824203491211,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 7598
    },
    {
      "epoch": 0.6820140010770059,
      "grad_norm": 3.02596378326416,
      "learning_rate": 0.0,
      "loss": 0.7211,
      "step": 7599
    },
    {
      "epoch": 0.6821037515706336,
      "grad_norm": 2.4604060649871826,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 7600
    },
    {
      "epoch": 0.6821935020642613,
      "grad_norm": 2.3817968368530273,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 7601
    },
    {
      "epoch": 0.682283252557889,
      "grad_norm": 2.199655055999756,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 7602
    },
    {
      "epoch": 0.6823730030515168,
      "grad_norm": 2.9009571075439453,
      "learning_rate": 0.0,
      "loss": 0.679,
      "step": 7603
    },
    {
      "epoch": 0.6824627535451445,
      "grad_norm": 4.895392894744873,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 7604
    },
    {
      "epoch": 0.6825525040387722,
      "grad_norm": 6.4032793045043945,
      "learning_rate": 0.0,
      "loss": 0.6688,
      "step": 7605
    },
    {
      "epoch": 0.6826422545323999,
      "grad_norm": 4.244372844696045,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 7606
    },
    {
      "epoch": 0.6827320050260276,
      "grad_norm": 2.761312246322632,
      "learning_rate": 0.0,
      "loss": 0.7575,
      "step": 7607
    },
    {
      "epoch": 0.6828217555196554,
      "grad_norm": 3.119351387023926,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 7608
    },
    {
      "epoch": 0.6829115060132831,
      "grad_norm": 3.304267644882202,
      "learning_rate": 0.0,
      "loss": 0.7698,
      "step": 7609
    },
    {
      "epoch": 0.6830012565069108,
      "grad_norm": 2.687201976776123,
      "learning_rate": 0.0,
      "loss": 0.655,
      "step": 7610
    },
    {
      "epoch": 0.6830910070005385,
      "grad_norm": 3.8492674827575684,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 7611
    },
    {
      "epoch": 0.6831807574941662,
      "grad_norm": 2.577244758605957,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 7612
    },
    {
      "epoch": 0.683270507987794,
      "grad_norm": 2.326186180114746,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 7613
    },
    {
      "epoch": 0.6833602584814217,
      "grad_norm": 2.9133810997009277,
      "learning_rate": 0.0,
      "loss": 0.7775,
      "step": 7614
    },
    {
      "epoch": 0.6834500089750494,
      "grad_norm": 2.47379994392395,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 7615
    },
    {
      "epoch": 0.6835397594686771,
      "grad_norm": 5.696069240570068,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 7616
    },
    {
      "epoch": 0.6836295099623048,
      "grad_norm": 2.768066644668579,
      "learning_rate": 0.0,
      "loss": 0.7034,
      "step": 7617
    },
    {
      "epoch": 0.6837192604559325,
      "grad_norm": 3.1107466220855713,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 7618
    },
    {
      "epoch": 0.6838090109495603,
      "grad_norm": 5.380439758300781,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 7619
    },
    {
      "epoch": 0.683898761443188,
      "grad_norm": 2.9883198738098145,
      "learning_rate": 0.0,
      "loss": 0.6777,
      "step": 7620
    },
    {
      "epoch": 0.6839885119368156,
      "grad_norm": 3.343508005142212,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 7621
    },
    {
      "epoch": 0.6840782624304433,
      "grad_norm": 2.6869399547576904,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 7622
    },
    {
      "epoch": 0.684168012924071,
      "grad_norm": 2.4300363063812256,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 7623
    },
    {
      "epoch": 0.6842577634176988,
      "grad_norm": 2.482334852218628,
      "learning_rate": 0.0,
      "loss": 0.8019,
      "step": 7624
    },
    {
      "epoch": 0.6843475139113265,
      "grad_norm": 2.536017894744873,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 7625
    },
    {
      "epoch": 0.6844372644049542,
      "grad_norm": 2.6686480045318604,
      "learning_rate": 0.0,
      "loss": 0.7893,
      "step": 7626
    },
    {
      "epoch": 0.6845270148985819,
      "grad_norm": 2.7360939979553223,
      "learning_rate": 0.0,
      "loss": 0.6927,
      "step": 7627
    },
    {
      "epoch": 0.6846167653922096,
      "grad_norm": 3.1038241386413574,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7628
    },
    {
      "epoch": 0.6847065158858374,
      "grad_norm": 3.5918455123901367,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 7629
    },
    {
      "epoch": 0.6847962663794651,
      "grad_norm": 2.725978374481201,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 7630
    },
    {
      "epoch": 0.6848860168730928,
      "grad_norm": 2.8350539207458496,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 7631
    },
    {
      "epoch": 0.6849757673667205,
      "grad_norm": 3.305448055267334,
      "learning_rate": 0.0,
      "loss": 0.6979,
      "step": 7632
    },
    {
      "epoch": 0.6850655178603482,
      "grad_norm": 3.178535223007202,
      "learning_rate": 0.0,
      "loss": 0.7093,
      "step": 7633
    },
    {
      "epoch": 0.685155268353976,
      "grad_norm": 3.1851749420166016,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 7634
    },
    {
      "epoch": 0.6852450188476037,
      "grad_norm": 3.096799373626709,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 7635
    },
    {
      "epoch": 0.6853347693412314,
      "grad_norm": 3.0649967193603516,
      "learning_rate": 0.0,
      "loss": 0.6455,
      "step": 7636
    },
    {
      "epoch": 0.6854245198348591,
      "grad_norm": 2.9605257511138916,
      "learning_rate": 0.0,
      "loss": 0.7596,
      "step": 7637
    },
    {
      "epoch": 0.6855142703284868,
      "grad_norm": 2.2626044750213623,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 7638
    },
    {
      "epoch": 0.6856040208221145,
      "grad_norm": 2.426564931869507,
      "learning_rate": 0.0,
      "loss": 0.7937,
      "step": 7639
    },
    {
      "epoch": 0.6856937713157423,
      "grad_norm": 5.847156047821045,
      "learning_rate": 0.0,
      "loss": 0.7006,
      "step": 7640
    },
    {
      "epoch": 0.68578352180937,
      "grad_norm": 3.1782846450805664,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 7641
    },
    {
      "epoch": 0.6858732723029977,
      "grad_norm": 2.4198102951049805,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 7642
    },
    {
      "epoch": 0.6859630227966254,
      "grad_norm": 3.002025842666626,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 7643
    },
    {
      "epoch": 0.686052773290253,
      "grad_norm": 3.1748344898223877,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 7644
    },
    {
      "epoch": 0.6861425237838809,
      "grad_norm": 3.4147605895996094,
      "learning_rate": 0.0,
      "loss": 0.7885,
      "step": 7645
    },
    {
      "epoch": 0.6862322742775085,
      "grad_norm": 2.8856077194213867,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 7646
    },
    {
      "epoch": 0.6863220247711362,
      "grad_norm": 3.547452688217163,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 7647
    },
    {
      "epoch": 0.6864117752647639,
      "grad_norm": 2.8836662769317627,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 7648
    },
    {
      "epoch": 0.6865015257583916,
      "grad_norm": 3.106956720352173,
      "learning_rate": 0.0,
      "loss": 0.7207,
      "step": 7649
    },
    {
      "epoch": 0.6865912762520194,
      "grad_norm": 2.46144962310791,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 7650
    },
    {
      "epoch": 0.6866810267456471,
      "grad_norm": 2.469440221786499,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 7651
    },
    {
      "epoch": 0.6867707772392748,
      "grad_norm": 2.841259002685547,
      "learning_rate": 0.0,
      "loss": 0.7571,
      "step": 7652
    },
    {
      "epoch": 0.6868605277329025,
      "grad_norm": 2.449408531188965,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 7653
    },
    {
      "epoch": 0.6869502782265302,
      "grad_norm": 2.8640050888061523,
      "learning_rate": 0.0,
      "loss": 0.6972,
      "step": 7654
    },
    {
      "epoch": 0.687040028720158,
      "grad_norm": 8.91486930847168,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 7655
    },
    {
      "epoch": 0.6871297792137857,
      "grad_norm": 2.273566722869873,
      "learning_rate": 0.0,
      "loss": 0.6695,
      "step": 7656
    },
    {
      "epoch": 0.6872195297074134,
      "grad_norm": 3.065617322921753,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 7657
    },
    {
      "epoch": 0.6873092802010411,
      "grad_norm": 2.520885467529297,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 7658
    },
    {
      "epoch": 0.6873990306946688,
      "grad_norm": 2.941011905670166,
      "learning_rate": 0.0,
      "loss": 0.7506,
      "step": 7659
    },
    {
      "epoch": 0.6874887811882965,
      "grad_norm": 2.3195958137512207,
      "learning_rate": 0.0,
      "loss": 0.8172,
      "step": 7660
    },
    {
      "epoch": 0.6875785316819243,
      "grad_norm": 3.211956024169922,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 7661
    },
    {
      "epoch": 0.687668282175552,
      "grad_norm": 2.3188207149505615,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 7662
    },
    {
      "epoch": 0.6877580326691797,
      "grad_norm": 2.803910970687866,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 7663
    },
    {
      "epoch": 0.6878477831628074,
      "grad_norm": 9.533297538757324,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 7664
    },
    {
      "epoch": 0.6879375336564351,
      "grad_norm": 2.802708625793457,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 7665
    },
    {
      "epoch": 0.6880272841500629,
      "grad_norm": 2.3422791957855225,
      "learning_rate": 0.0,
      "loss": 0.7344,
      "step": 7666
    },
    {
      "epoch": 0.6881170346436906,
      "grad_norm": 2.4482133388519287,
      "learning_rate": 0.0,
      "loss": 0.6376,
      "step": 7667
    },
    {
      "epoch": 0.6882067851373183,
      "grad_norm": 3.514986515045166,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 7668
    },
    {
      "epoch": 0.688296535630946,
      "grad_norm": 2.948284149169922,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 7669
    },
    {
      "epoch": 0.6883862861245736,
      "grad_norm": 3.0799715518951416,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 7670
    },
    {
      "epoch": 0.6884760366182014,
      "grad_norm": 4.929605960845947,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 7671
    },
    {
      "epoch": 0.6885657871118291,
      "grad_norm": 4.176370143890381,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 7672
    },
    {
      "epoch": 0.6886555376054568,
      "grad_norm": 3.4013731479644775,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 7673
    },
    {
      "epoch": 0.6887452880990845,
      "grad_norm": 2.399778366088867,
      "learning_rate": 0.0,
      "loss": 0.6433,
      "step": 7674
    },
    {
      "epoch": 0.6888350385927122,
      "grad_norm": 3.1863648891448975,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 7675
    },
    {
      "epoch": 0.68892478908634,
      "grad_norm": 2.9419939517974854,
      "learning_rate": 0.0,
      "loss": 0.7211,
      "step": 7676
    },
    {
      "epoch": 0.6890145395799677,
      "grad_norm": 2.2633628845214844,
      "learning_rate": 0.0,
      "loss": 0.7719,
      "step": 7677
    },
    {
      "epoch": 0.6891042900735954,
      "grad_norm": 3.302708625793457,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 7678
    },
    {
      "epoch": 0.6891940405672231,
      "grad_norm": 10.032313346862793,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 7679
    },
    {
      "epoch": 0.6892837910608508,
      "grad_norm": 3.2876572608947754,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 7680
    },
    {
      "epoch": 0.6893735415544785,
      "grad_norm": 3.2441723346710205,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 7681
    },
    {
      "epoch": 0.6894632920481063,
      "grad_norm": 4.014781951904297,
      "learning_rate": 0.0,
      "loss": 0.746,
      "step": 7682
    },
    {
      "epoch": 0.689553042541734,
      "grad_norm": 3.180541753768921,
      "learning_rate": 0.0,
      "loss": 0.6819,
      "step": 7683
    },
    {
      "epoch": 0.6896427930353617,
      "grad_norm": 2.9769909381866455,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 7684
    },
    {
      "epoch": 0.6897325435289894,
      "grad_norm": 2.4641337394714355,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 7685
    },
    {
      "epoch": 0.6898222940226171,
      "grad_norm": 2.937962055206299,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 7686
    },
    {
      "epoch": 0.6899120445162449,
      "grad_norm": 3.151041269302368,
      "learning_rate": 0.0,
      "loss": 0.6589,
      "step": 7687
    },
    {
      "epoch": 0.6900017950098726,
      "grad_norm": 2.5841565132141113,
      "learning_rate": 0.0,
      "loss": 0.7716,
      "step": 7688
    },
    {
      "epoch": 0.6900915455035003,
      "grad_norm": 3.483264923095703,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 7689
    },
    {
      "epoch": 0.690181295997128,
      "grad_norm": 2.603800058364868,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 7690
    },
    {
      "epoch": 0.6902710464907557,
      "grad_norm": 2.9422101974487305,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 7691
    },
    {
      "epoch": 0.6903607969843835,
      "grad_norm": 4.540676593780518,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 7692
    },
    {
      "epoch": 0.6904505474780112,
      "grad_norm": 3.299673318862915,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 7693
    },
    {
      "epoch": 0.6905402979716389,
      "grad_norm": 3.355238199234009,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 7694
    },
    {
      "epoch": 0.6906300484652665,
      "grad_norm": 3.406975269317627,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 7695
    },
    {
      "epoch": 0.6907197989588942,
      "grad_norm": 5.061450958251953,
      "learning_rate": 0.0,
      "loss": 0.6421,
      "step": 7696
    },
    {
      "epoch": 0.690809549452522,
      "grad_norm": 3.898071527481079,
      "learning_rate": 0.0,
      "loss": 0.6987,
      "step": 7697
    },
    {
      "epoch": 0.6908992999461497,
      "grad_norm": 2.534050464630127,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 7698
    },
    {
      "epoch": 0.6909890504397774,
      "grad_norm": 2.551281690597534,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 7699
    },
    {
      "epoch": 0.6910788009334051,
      "grad_norm": 1.9854179620742798,
      "learning_rate": 0.0,
      "loss": 0.7626,
      "step": 7700
    },
    {
      "epoch": 0.6911685514270328,
      "grad_norm": 3.3152639865875244,
      "learning_rate": 0.0,
      "loss": 0.6635,
      "step": 7701
    },
    {
      "epoch": 0.6912583019206605,
      "grad_norm": 2.6709682941436768,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 7702
    },
    {
      "epoch": 0.6913480524142883,
      "grad_norm": 2.65161395072937,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 7703
    },
    {
      "epoch": 0.691437802907916,
      "grad_norm": 2.5956127643585205,
      "learning_rate": 0.0,
      "loss": 0.7433,
      "step": 7704
    },
    {
      "epoch": 0.6915275534015437,
      "grad_norm": 3.167447090148926,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 7705
    },
    {
      "epoch": 0.6916173038951714,
      "grad_norm": 2.8933403491973877,
      "learning_rate": 0.0,
      "loss": 0.6905,
      "step": 7706
    },
    {
      "epoch": 0.6917070543887991,
      "grad_norm": 3.895777940750122,
      "learning_rate": 0.0,
      "loss": 0.7317,
      "step": 7707
    },
    {
      "epoch": 0.6917968048824269,
      "grad_norm": 2.900472640991211,
      "learning_rate": 0.0,
      "loss": 0.6646,
      "step": 7708
    },
    {
      "epoch": 0.6918865553760546,
      "grad_norm": 3.7499091625213623,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 7709
    },
    {
      "epoch": 0.6919763058696823,
      "grad_norm": 2.5782623291015625,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 7710
    },
    {
      "epoch": 0.69206605636331,
      "grad_norm": 2.361366033554077,
      "learning_rate": 0.0,
      "loss": 0.6809,
      "step": 7711
    },
    {
      "epoch": 0.6921558068569377,
      "grad_norm": 4.640886306762695,
      "learning_rate": 0.0,
      "loss": 0.7828,
      "step": 7712
    },
    {
      "epoch": 0.6922455573505655,
      "grad_norm": 3.146530866622925,
      "learning_rate": 0.0,
      "loss": 0.7878,
      "step": 7713
    },
    {
      "epoch": 0.6923353078441932,
      "grad_norm": 2.685436725616455,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 7714
    },
    {
      "epoch": 0.6924250583378209,
      "grad_norm": 3.98968768119812,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 7715
    },
    {
      "epoch": 0.6925148088314486,
      "grad_norm": 3.4001429080963135,
      "learning_rate": 0.0,
      "loss": 0.6925,
      "step": 7716
    },
    {
      "epoch": 0.6926045593250763,
      "grad_norm": 2.9938549995422363,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 7717
    },
    {
      "epoch": 0.6926943098187041,
      "grad_norm": 2.7443206310272217,
      "learning_rate": 0.0,
      "loss": 0.6793,
      "step": 7718
    },
    {
      "epoch": 0.6927840603123318,
      "grad_norm": 2.7758052349090576,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 7719
    },
    {
      "epoch": 0.6928738108059594,
      "grad_norm": 2.803997755050659,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 7720
    },
    {
      "epoch": 0.6929635612995871,
      "grad_norm": 3.781006336212158,
      "learning_rate": 0.0,
      "loss": 0.6721,
      "step": 7721
    },
    {
      "epoch": 0.6930533117932148,
      "grad_norm": 14.74190902709961,
      "learning_rate": 0.0,
      "loss": 0.6871,
      "step": 7722
    },
    {
      "epoch": 0.6931430622868425,
      "grad_norm": 4.090052604675293,
      "learning_rate": 0.0,
      "loss": 0.7553,
      "step": 7723
    },
    {
      "epoch": 0.6932328127804703,
      "grad_norm": 2.69901704788208,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 7724
    },
    {
      "epoch": 0.693322563274098,
      "grad_norm": 4.838626384735107,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 7725
    },
    {
      "epoch": 0.6934123137677257,
      "grad_norm": 3.1213111877441406,
      "learning_rate": 0.0,
      "loss": 0.7086,
      "step": 7726
    },
    {
      "epoch": 0.6935020642613534,
      "grad_norm": 2.7235682010650635,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 7727
    },
    {
      "epoch": 0.6935918147549811,
      "grad_norm": 3.690293788909912,
      "learning_rate": 0.0,
      "loss": 0.7907,
      "step": 7728
    },
    {
      "epoch": 0.6936815652486089,
      "grad_norm": 3.0140278339385986,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 7729
    },
    {
      "epoch": 0.6937713157422366,
      "grad_norm": 2.3968958854675293,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 7730
    },
    {
      "epoch": 0.6938610662358643,
      "grad_norm": 3.055255174636841,
      "learning_rate": 0.0,
      "loss": 0.781,
      "step": 7731
    },
    {
      "epoch": 0.693950816729492,
      "grad_norm": 2.9936437606811523,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 7732
    },
    {
      "epoch": 0.6940405672231197,
      "grad_norm": 3.3339459896087646,
      "learning_rate": 0.0,
      "loss": 0.6974,
      "step": 7733
    },
    {
      "epoch": 0.6941303177167475,
      "grad_norm": 4.079405784606934,
      "learning_rate": 0.0,
      "loss": 0.7065,
      "step": 7734
    },
    {
      "epoch": 0.6942200682103752,
      "grad_norm": 2.7814857959747314,
      "learning_rate": 0.0,
      "loss": 0.746,
      "step": 7735
    },
    {
      "epoch": 0.6943098187040029,
      "grad_norm": 2.900345802307129,
      "learning_rate": 0.0,
      "loss": 0.6927,
      "step": 7736
    },
    {
      "epoch": 0.6943995691976306,
      "grad_norm": 13.095170021057129,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 7737
    },
    {
      "epoch": 0.6944893196912583,
      "grad_norm": 3.4050540924072266,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 7738
    },
    {
      "epoch": 0.6945790701848861,
      "grad_norm": 9.157209396362305,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 7739
    },
    {
      "epoch": 0.6946688206785138,
      "grad_norm": 2.4629650115966797,
      "learning_rate": 0.0,
      "loss": 0.8073,
      "step": 7740
    },
    {
      "epoch": 0.6947585711721415,
      "grad_norm": 2.9621665477752686,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 7741
    },
    {
      "epoch": 0.6948483216657692,
      "grad_norm": 3.42069411277771,
      "learning_rate": 0.0,
      "loss": 0.7847,
      "step": 7742
    },
    {
      "epoch": 0.6949380721593968,
      "grad_norm": 3.184279441833496,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 7743
    },
    {
      "epoch": 0.6950278226530245,
      "grad_norm": 2.132857322692871,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 7744
    },
    {
      "epoch": 0.6951175731466523,
      "grad_norm": 2.6368730068206787,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 7745
    },
    {
      "epoch": 0.69520732364028,
      "grad_norm": 2.6499650478363037,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 7746
    },
    {
      "epoch": 0.6952970741339077,
      "grad_norm": 3.2967703342437744,
      "learning_rate": 0.0,
      "loss": 0.789,
      "step": 7747
    },
    {
      "epoch": 0.6953868246275354,
      "grad_norm": 3.018541097640991,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7748
    },
    {
      "epoch": 0.6954765751211631,
      "grad_norm": 2.5656168460845947,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 7749
    },
    {
      "epoch": 0.6955663256147909,
      "grad_norm": 2.4557864665985107,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 7750
    },
    {
      "epoch": 0.6956560761084186,
      "grad_norm": 2.98274827003479,
      "learning_rate": 0.0,
      "loss": 0.7395,
      "step": 7751
    },
    {
      "epoch": 0.6957458266020463,
      "grad_norm": 3.3789753913879395,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 7752
    },
    {
      "epoch": 0.695835577095674,
      "grad_norm": 3.0554516315460205,
      "learning_rate": 0.0,
      "loss": 0.7329,
      "step": 7753
    },
    {
      "epoch": 0.6959253275893017,
      "grad_norm": 2.850105047225952,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 7754
    },
    {
      "epoch": 0.6960150780829295,
      "grad_norm": 2.974663257598877,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 7755
    },
    {
      "epoch": 0.6961048285765572,
      "grad_norm": 3.2665154933929443,
      "learning_rate": 0.0,
      "loss": 0.7693,
      "step": 7756
    },
    {
      "epoch": 0.6961945790701849,
      "grad_norm": 6.643617153167725,
      "learning_rate": 0.0,
      "loss": 0.7743,
      "step": 7757
    },
    {
      "epoch": 0.6962843295638126,
      "grad_norm": 3.5807061195373535,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 7758
    },
    {
      "epoch": 0.6963740800574403,
      "grad_norm": 4.200815200805664,
      "learning_rate": 0.0,
      "loss": 0.6989,
      "step": 7759
    },
    {
      "epoch": 0.6964638305510681,
      "grad_norm": 4.455434799194336,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 7760
    },
    {
      "epoch": 0.6965535810446958,
      "grad_norm": 2.9542713165283203,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 7761
    },
    {
      "epoch": 0.6966433315383235,
      "grad_norm": 5.097710132598877,
      "learning_rate": 0.0,
      "loss": 0.7061,
      "step": 7762
    },
    {
      "epoch": 0.6967330820319512,
      "grad_norm": 2.332122802734375,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 7763
    },
    {
      "epoch": 0.6968228325255789,
      "grad_norm": 3.4852123260498047,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 7764
    },
    {
      "epoch": 0.6969125830192066,
      "grad_norm": 2.8467841148376465,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 7765
    },
    {
      "epoch": 0.6970023335128344,
      "grad_norm": 2.4006903171539307,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 7766
    },
    {
      "epoch": 0.697092084006462,
      "grad_norm": 2.9461417198181152,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 7767
    },
    {
      "epoch": 0.6971818345000897,
      "grad_norm": 3.1215505599975586,
      "learning_rate": 0.0,
      "loss": 0.6632,
      "step": 7768
    },
    {
      "epoch": 0.6972715849937174,
      "grad_norm": 4.856113910675049,
      "learning_rate": 0.0,
      "loss": 0.7809,
      "step": 7769
    },
    {
      "epoch": 0.6973613354873451,
      "grad_norm": 3.04170823097229,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 7770
    },
    {
      "epoch": 0.6974510859809729,
      "grad_norm": 3.109091281890869,
      "learning_rate": 0.0,
      "loss": 0.6772,
      "step": 7771
    },
    {
      "epoch": 0.6975408364746006,
      "grad_norm": 2.921452760696411,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 7772
    },
    {
      "epoch": 0.6976305869682283,
      "grad_norm": 4.589075088500977,
      "learning_rate": 0.0,
      "loss": 0.6742,
      "step": 7773
    },
    {
      "epoch": 0.697720337461856,
      "grad_norm": 2.9233291149139404,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 7774
    },
    {
      "epoch": 0.6978100879554837,
      "grad_norm": 2.442634344100952,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 7775
    },
    {
      "epoch": 0.6978998384491115,
      "grad_norm": 3.14431095123291,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 7776
    },
    {
      "epoch": 0.6979895889427392,
      "grad_norm": 3.1071248054504395,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 7777
    },
    {
      "epoch": 0.6980793394363669,
      "grad_norm": 2.9817092418670654,
      "learning_rate": 0.0,
      "loss": 0.7962,
      "step": 7778
    },
    {
      "epoch": 0.6981690899299946,
      "grad_norm": 2.6432297229766846,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 7779
    },
    {
      "epoch": 0.6982588404236223,
      "grad_norm": 2.2865066528320312,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 7780
    },
    {
      "epoch": 0.69834859091725,
      "grad_norm": 3.0851781368255615,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 7781
    },
    {
      "epoch": 0.6984383414108778,
      "grad_norm": 2.809804916381836,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 7782
    },
    {
      "epoch": 0.6985280919045055,
      "grad_norm": 2.6564764976501465,
      "learning_rate": 0.0,
      "loss": 0.7196,
      "step": 7783
    },
    {
      "epoch": 0.6986178423981332,
      "grad_norm": 2.7552642822265625,
      "learning_rate": 0.0,
      "loss": 0.7044,
      "step": 7784
    },
    {
      "epoch": 0.6987075928917609,
      "grad_norm": 3.126801013946533,
      "learning_rate": 0.0,
      "loss": 0.6683,
      "step": 7785
    },
    {
      "epoch": 0.6987973433853886,
      "grad_norm": 3.531001567840576,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 7786
    },
    {
      "epoch": 0.6988870938790164,
      "grad_norm": 3.4663166999816895,
      "learning_rate": 0.0,
      "loss": 0.6916,
      "step": 7787
    },
    {
      "epoch": 0.6989768443726441,
      "grad_norm": 5.098100185394287,
      "learning_rate": 0.0,
      "loss": 0.6714,
      "step": 7788
    },
    {
      "epoch": 0.6990665948662718,
      "grad_norm": 2.6426069736480713,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 7789
    },
    {
      "epoch": 0.6991563453598995,
      "grad_norm": 2.9197468757629395,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 7790
    },
    {
      "epoch": 0.6992460958535271,
      "grad_norm": 4.555596351623535,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 7791
    },
    {
      "epoch": 0.699335846347155,
      "grad_norm": 3.4281561374664307,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 7792
    },
    {
      "epoch": 0.6994255968407826,
      "grad_norm": 4.475138187408447,
      "learning_rate": 0.0,
      "loss": 0.7846,
      "step": 7793
    },
    {
      "epoch": 0.6995153473344103,
      "grad_norm": 2.6709327697753906,
      "learning_rate": 0.0,
      "loss": 0.7081,
      "step": 7794
    },
    {
      "epoch": 0.699605097828038,
      "grad_norm": 3.027298927307129,
      "learning_rate": 0.0,
      "loss": 0.6731,
      "step": 7795
    },
    {
      "epoch": 0.6996948483216657,
      "grad_norm": 3.9747109413146973,
      "learning_rate": 0.0,
      "loss": 0.7725,
      "step": 7796
    },
    {
      "epoch": 0.6997845988152935,
      "grad_norm": 2.743759870529175,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 7797
    },
    {
      "epoch": 0.6998743493089212,
      "grad_norm": 3.548950433731079,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 7798
    },
    {
      "epoch": 0.6999640998025489,
      "grad_norm": 3.1486501693725586,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 7799
    },
    {
      "epoch": 0.7000538502961766,
      "grad_norm": 2.864565372467041,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 7800
    },
    {
      "epoch": 0.7001436007898043,
      "grad_norm": 2.6132631301879883,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 7801
    },
    {
      "epoch": 0.700233351283432,
      "grad_norm": 3.0392584800720215,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 7802
    },
    {
      "epoch": 0.7003231017770598,
      "grad_norm": 3.8505072593688965,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 7803
    },
    {
      "epoch": 0.7004128522706875,
      "grad_norm": 3.303500175476074,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 7804
    },
    {
      "epoch": 0.7005026027643152,
      "grad_norm": 3.1969082355499268,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 7805
    },
    {
      "epoch": 0.7005923532579429,
      "grad_norm": 2.73659610748291,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 7806
    },
    {
      "epoch": 0.7006821037515706,
      "grad_norm": 2.5181102752685547,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 7807
    },
    {
      "epoch": 0.7007718542451984,
      "grad_norm": 2.61932110786438,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 7808
    },
    {
      "epoch": 0.7008616047388261,
      "grad_norm": 2.5911667346954346,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 7809
    },
    {
      "epoch": 0.7009513552324538,
      "grad_norm": 3.3921384811401367,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 7810
    },
    {
      "epoch": 0.7010411057260815,
      "grad_norm": 6.043535232543945,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 7811
    },
    {
      "epoch": 0.7011308562197092,
      "grad_norm": 2.3112499713897705,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 7812
    },
    {
      "epoch": 0.701220606713337,
      "grad_norm": 3.3283495903015137,
      "learning_rate": 0.0,
      "loss": 0.6732,
      "step": 7813
    },
    {
      "epoch": 0.7013103572069647,
      "grad_norm": 3.168011426925659,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 7814
    },
    {
      "epoch": 0.7014001077005924,
      "grad_norm": 2.268024206161499,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 7815
    },
    {
      "epoch": 0.70148985819422,
      "grad_norm": 3.3047256469726562,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 7816
    },
    {
      "epoch": 0.7015796086878477,
      "grad_norm": 2.3402903079986572,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 7817
    },
    {
      "epoch": 0.7016693591814755,
      "grad_norm": 3.2909021377563477,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 7818
    },
    {
      "epoch": 0.7017591096751032,
      "grad_norm": 4.40167236328125,
      "learning_rate": 0.0,
      "loss": 0.6974,
      "step": 7819
    },
    {
      "epoch": 0.7018488601687309,
      "grad_norm": 2.8194892406463623,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 7820
    },
    {
      "epoch": 0.7019386106623586,
      "grad_norm": 2.573996067047119,
      "learning_rate": 0.0,
      "loss": 0.7026,
      "step": 7821
    },
    {
      "epoch": 0.7020283611559863,
      "grad_norm": 2.780510902404785,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 7822
    },
    {
      "epoch": 0.702118111649614,
      "grad_norm": 3.5375823974609375,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 7823
    },
    {
      "epoch": 0.7022078621432418,
      "grad_norm": 3.6038811206817627,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 7824
    },
    {
      "epoch": 0.7022976126368695,
      "grad_norm": 2.7398219108581543,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 7825
    },
    {
      "epoch": 0.7023873631304972,
      "grad_norm": 3.1031301021575928,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 7826
    },
    {
      "epoch": 0.7024771136241249,
      "grad_norm": 2.7431018352508545,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 7827
    },
    {
      "epoch": 0.7025668641177526,
      "grad_norm": 2.1963441371917725,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 7828
    },
    {
      "epoch": 0.7026566146113804,
      "grad_norm": 2.8643765449523926,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 7829
    },
    {
      "epoch": 0.7027463651050081,
      "grad_norm": 2.4735333919525146,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 7830
    },
    {
      "epoch": 0.7028361155986358,
      "grad_norm": 2.454962968826294,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 7831
    },
    {
      "epoch": 0.7029258660922635,
      "grad_norm": 3.5633578300476074,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 7832
    },
    {
      "epoch": 0.7030156165858912,
      "grad_norm": 3.53857421875,
      "learning_rate": 0.0,
      "loss": 0.763,
      "step": 7833
    },
    {
      "epoch": 0.703105367079519,
      "grad_norm": 3.436436891555786,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 7834
    },
    {
      "epoch": 0.7031951175731467,
      "grad_norm": 4.9551310539245605,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 7835
    },
    {
      "epoch": 0.7032848680667744,
      "grad_norm": 2.715813398361206,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 7836
    },
    {
      "epoch": 0.7033746185604021,
      "grad_norm": 3.049160957336426,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 7837
    },
    {
      "epoch": 0.7034643690540298,
      "grad_norm": 4.363902568817139,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 7838
    },
    {
      "epoch": 0.7035541195476576,
      "grad_norm": 2.4873509407043457,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 7839
    },
    {
      "epoch": 0.7036438700412853,
      "grad_norm": 5.1042160987854,
      "learning_rate": 0.0,
      "loss": 0.7805,
      "step": 7840
    },
    {
      "epoch": 0.703733620534913,
      "grad_norm": 2.6650633811950684,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 7841
    },
    {
      "epoch": 0.7038233710285406,
      "grad_norm": 1.8689663410186768,
      "learning_rate": 0.0,
      "loss": 0.8155,
      "step": 7842
    },
    {
      "epoch": 0.7039131215221683,
      "grad_norm": 2.581653356552124,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 7843
    },
    {
      "epoch": 0.704002872015796,
      "grad_norm": 3.65631365776062,
      "learning_rate": 0.0,
      "loss": 0.7819,
      "step": 7844
    },
    {
      "epoch": 0.7040926225094238,
      "grad_norm": 2.9495463371276855,
      "learning_rate": 0.0,
      "loss": 0.7006,
      "step": 7845
    },
    {
      "epoch": 0.7041823730030515,
      "grad_norm": 3.4424757957458496,
      "learning_rate": 0.0,
      "loss": 0.6876,
      "step": 7846
    },
    {
      "epoch": 0.7042721234966792,
      "grad_norm": 7.208250045776367,
      "learning_rate": 0.0,
      "loss": 0.6799,
      "step": 7847
    },
    {
      "epoch": 0.7043618739903069,
      "grad_norm": 2.5300862789154053,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 7848
    },
    {
      "epoch": 0.7044516244839346,
      "grad_norm": 7.725045680999756,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 7849
    },
    {
      "epoch": 0.7045413749775624,
      "grad_norm": 2.807070255279541,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 7850
    },
    {
      "epoch": 0.7046311254711901,
      "grad_norm": 2.2561495304107666,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 7851
    },
    {
      "epoch": 0.7047208759648178,
      "grad_norm": 3.4761605262756348,
      "learning_rate": 0.0,
      "loss": 0.7835,
      "step": 7852
    },
    {
      "epoch": 0.7048106264584455,
      "grad_norm": 3.043144702911377,
      "learning_rate": 0.0,
      "loss": 0.7559,
      "step": 7853
    },
    {
      "epoch": 0.7049003769520732,
      "grad_norm": 2.4997830390930176,
      "learning_rate": 0.0,
      "loss": 0.6764,
      "step": 7854
    },
    {
      "epoch": 0.704990127445701,
      "grad_norm": 2.9578301906585693,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 7855
    },
    {
      "epoch": 0.7050798779393287,
      "grad_norm": 3.319697141647339,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 7856
    },
    {
      "epoch": 0.7051696284329564,
      "grad_norm": 2.7665460109710693,
      "learning_rate": 0.0,
      "loss": 0.7495,
      "step": 7857
    },
    {
      "epoch": 0.7052593789265841,
      "grad_norm": 2.608982801437378,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 7858
    },
    {
      "epoch": 0.7053491294202118,
      "grad_norm": 3.4470136165618896,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 7859
    },
    {
      "epoch": 0.7054388799138396,
      "grad_norm": 2.527114152908325,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 7860
    },
    {
      "epoch": 0.7055286304074673,
      "grad_norm": 2.642611503601074,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 7861
    },
    {
      "epoch": 0.705618380901095,
      "grad_norm": 2.263490915298462,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 7862
    },
    {
      "epoch": 0.7057081313947227,
      "grad_norm": 3.728419542312622,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 7863
    },
    {
      "epoch": 0.7057978818883504,
      "grad_norm": 2.6656904220581055,
      "learning_rate": 0.0,
      "loss": 0.6663,
      "step": 7864
    },
    {
      "epoch": 0.705887632381978,
      "grad_norm": 3.066011905670166,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 7865
    },
    {
      "epoch": 0.7059773828756059,
      "grad_norm": 3.4068198204040527,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 7866
    },
    {
      "epoch": 0.7060671333692335,
      "grad_norm": 3.270235061645508,
      "learning_rate": 0.0,
      "loss": 0.6609,
      "step": 7867
    },
    {
      "epoch": 0.7061568838628612,
      "grad_norm": 4.7444748878479,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 7868
    },
    {
      "epoch": 0.7062466343564889,
      "grad_norm": 3.5851025581359863,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 7869
    },
    {
      "epoch": 0.7063363848501166,
      "grad_norm": 3.0092968940734863,
      "learning_rate": 0.0,
      "loss": 0.6822,
      "step": 7870
    },
    {
      "epoch": 0.7064261353437444,
      "grad_norm": 3.2923781871795654,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 7871
    },
    {
      "epoch": 0.7065158858373721,
      "grad_norm": 9.579663276672363,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 7872
    },
    {
      "epoch": 0.7066056363309998,
      "grad_norm": 3.266432762145996,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 7873
    },
    {
      "epoch": 0.7066953868246275,
      "grad_norm": 2.988766670227051,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 7874
    },
    {
      "epoch": 0.7067851373182552,
      "grad_norm": 6.288182735443115,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 7875
    },
    {
      "epoch": 0.706874887811883,
      "grad_norm": 3.346639394760132,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 7876
    },
    {
      "epoch": 0.7069646383055107,
      "grad_norm": 2.922122001647949,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 7877
    },
    {
      "epoch": 0.7070543887991384,
      "grad_norm": 2.357703685760498,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 7878
    },
    {
      "epoch": 0.7071441392927661,
      "grad_norm": 2.741591453552246,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 7879
    },
    {
      "epoch": 0.7072338897863938,
      "grad_norm": 3.1154072284698486,
      "learning_rate": 0.0,
      "loss": 0.693,
      "step": 7880
    },
    {
      "epoch": 0.7073236402800216,
      "grad_norm": 3.880831003189087,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 7881
    },
    {
      "epoch": 0.7074133907736493,
      "grad_norm": 4.161078929901123,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 7882
    },
    {
      "epoch": 0.707503141267277,
      "grad_norm": 2.9215707778930664,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 7883
    },
    {
      "epoch": 0.7075928917609047,
      "grad_norm": 2.4651596546173096,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 7884
    },
    {
      "epoch": 0.7076826422545324,
      "grad_norm": 2.3443188667297363,
      "learning_rate": 0.0,
      "loss": 0.7074,
      "step": 7885
    },
    {
      "epoch": 0.7077723927481601,
      "grad_norm": 2.4058568477630615,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 7886
    },
    {
      "epoch": 0.7078621432417879,
      "grad_norm": 3.0778627395629883,
      "learning_rate": 0.0,
      "loss": 0.6838,
      "step": 7887
    },
    {
      "epoch": 0.7079518937354156,
      "grad_norm": 2.7689208984375,
      "learning_rate": 0.0,
      "loss": 0.6675,
      "step": 7888
    },
    {
      "epoch": 0.7080416442290433,
      "grad_norm": 3.4546821117401123,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 7889
    },
    {
      "epoch": 0.708131394722671,
      "grad_norm": 3.0953495502471924,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 7890
    },
    {
      "epoch": 0.7082211452162986,
      "grad_norm": 2.9452881813049316,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 7891
    },
    {
      "epoch": 0.7083108957099264,
      "grad_norm": 3.1003637313842773,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 7892
    },
    {
      "epoch": 0.7084006462035541,
      "grad_norm": 4.384395122528076,
      "learning_rate": 0.0,
      "loss": 0.6995,
      "step": 7893
    },
    {
      "epoch": 0.7084903966971818,
      "grad_norm": 3.0202159881591797,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 7894
    },
    {
      "epoch": 0.7085801471908095,
      "grad_norm": 3.406865119934082,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 7895
    },
    {
      "epoch": 0.7086698976844372,
      "grad_norm": 2.827791929244995,
      "learning_rate": 0.0,
      "loss": 0.6886,
      "step": 7896
    },
    {
      "epoch": 0.708759648178065,
      "grad_norm": 2.4888288974761963,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 7897
    },
    {
      "epoch": 0.7088493986716927,
      "grad_norm": 3.451802968978882,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 7898
    },
    {
      "epoch": 0.7089391491653204,
      "grad_norm": 9.000911712646484,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 7899
    },
    {
      "epoch": 0.7090288996589481,
      "grad_norm": 11.251934051513672,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 7900
    },
    {
      "epoch": 0.7091186501525758,
      "grad_norm": 2.4115347862243652,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 7901
    },
    {
      "epoch": 0.7092084006462036,
      "grad_norm": 2.789829730987549,
      "learning_rate": 0.0,
      "loss": 0.8499,
      "step": 7902
    },
    {
      "epoch": 0.7092981511398313,
      "grad_norm": 1.9486762285232544,
      "learning_rate": 0.0,
      "loss": 0.7199,
      "step": 7903
    },
    {
      "epoch": 0.709387901633459,
      "grad_norm": 3.7596888542175293,
      "learning_rate": 0.0,
      "loss": 0.6789,
      "step": 7904
    },
    {
      "epoch": 0.7094776521270867,
      "grad_norm": 3.2280569076538086,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 7905
    },
    {
      "epoch": 0.7095674026207144,
      "grad_norm": 3.5247483253479004,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 7906
    },
    {
      "epoch": 0.7096571531143421,
      "grad_norm": 2.3976662158966064,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 7907
    },
    {
      "epoch": 0.7097469036079699,
      "grad_norm": 2.6224331855773926,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 7908
    },
    {
      "epoch": 0.7098366541015976,
      "grad_norm": 3.6383743286132812,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 7909
    },
    {
      "epoch": 0.7099264045952253,
      "grad_norm": 3.125530958175659,
      "learning_rate": 0.0,
      "loss": 0.7847,
      "step": 7910
    },
    {
      "epoch": 0.710016155088853,
      "grad_norm": 6.1143622398376465,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 7911
    },
    {
      "epoch": 0.7101059055824807,
      "grad_norm": 3.096419095993042,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 7912
    },
    {
      "epoch": 0.7101956560761085,
      "grad_norm": 3.0963633060455322,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 7913
    },
    {
      "epoch": 0.7102854065697362,
      "grad_norm": 3.0657029151916504,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 7914
    },
    {
      "epoch": 0.7103751570633638,
      "grad_norm": 3.0823917388916016,
      "learning_rate": 0.0,
      "loss": 0.6691,
      "step": 7915
    },
    {
      "epoch": 0.7104649075569915,
      "grad_norm": 2.6080901622772217,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 7916
    },
    {
      "epoch": 0.7105546580506192,
      "grad_norm": 3.4431347846984863,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 7917
    },
    {
      "epoch": 0.710644408544247,
      "grad_norm": 3.4775149822235107,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 7918
    },
    {
      "epoch": 0.7107341590378747,
      "grad_norm": 9.454357147216797,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 7919
    },
    {
      "epoch": 0.7108239095315024,
      "grad_norm": 3.0362889766693115,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 7920
    },
    {
      "epoch": 0.7109136600251301,
      "grad_norm": 2.94600248336792,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 7921
    },
    {
      "epoch": 0.7110034105187578,
      "grad_norm": 5.085402488708496,
      "learning_rate": 0.0,
      "loss": 0.6384,
      "step": 7922
    },
    {
      "epoch": 0.7110931610123856,
      "grad_norm": 4.311812400817871,
      "learning_rate": 0.0,
      "loss": 0.7716,
      "step": 7923
    },
    {
      "epoch": 0.7111829115060133,
      "grad_norm": 3.3279051780700684,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 7924
    },
    {
      "epoch": 0.711272661999641,
      "grad_norm": 3.449850082397461,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 7925
    },
    {
      "epoch": 0.7113624124932687,
      "grad_norm": 3.4915647506713867,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 7926
    },
    {
      "epoch": 0.7114521629868964,
      "grad_norm": 2.7004663944244385,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 7927
    },
    {
      "epoch": 0.7115419134805241,
      "grad_norm": 2.838254928588867,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 7928
    },
    {
      "epoch": 0.7116316639741519,
      "grad_norm": 2.9247987270355225,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 7929
    },
    {
      "epoch": 0.7117214144677796,
      "grad_norm": 4.907057762145996,
      "learning_rate": 0.0,
      "loss": 0.6773,
      "step": 7930
    },
    {
      "epoch": 0.7118111649614073,
      "grad_norm": 3.0099949836730957,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 7931
    },
    {
      "epoch": 0.711900915455035,
      "grad_norm": 2.4507577419281006,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 7932
    },
    {
      "epoch": 0.7119906659486627,
      "grad_norm": 4.835752487182617,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 7933
    },
    {
      "epoch": 0.7120804164422905,
      "grad_norm": 3.002028465270996,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 7934
    },
    {
      "epoch": 0.7121701669359182,
      "grad_norm": 3.0448195934295654,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 7935
    },
    {
      "epoch": 0.7122599174295459,
      "grad_norm": 2.729896306991577,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 7936
    },
    {
      "epoch": 0.7123496679231736,
      "grad_norm": 3.749730348587036,
      "learning_rate": 0.0,
      "loss": 0.799,
      "step": 7937
    },
    {
      "epoch": 0.7124394184168013,
      "grad_norm": 2.661717653274536,
      "learning_rate": 0.0,
      "loss": 0.7865,
      "step": 7938
    },
    {
      "epoch": 0.7125291689104291,
      "grad_norm": 2.8651599884033203,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 7939
    },
    {
      "epoch": 0.7126189194040567,
      "grad_norm": 3.7269129753112793,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 7940
    },
    {
      "epoch": 0.7127086698976844,
      "grad_norm": 7.523043632507324,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 7941
    },
    {
      "epoch": 0.7127984203913121,
      "grad_norm": 6.8165998458862305,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 7942
    },
    {
      "epoch": 0.7128881708849398,
      "grad_norm": 2.9636752605438232,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 7943
    },
    {
      "epoch": 0.7129779213785676,
      "grad_norm": 3.0616273880004883,
      "learning_rate": 0.0,
      "loss": 0.7637,
      "step": 7944
    },
    {
      "epoch": 0.7130676718721953,
      "grad_norm": 2.3180105686187744,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 7945
    },
    {
      "epoch": 0.713157422365823,
      "grad_norm": 6.2073774337768555,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 7946
    },
    {
      "epoch": 0.7132471728594507,
      "grad_norm": 4.347837448120117,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 7947
    },
    {
      "epoch": 0.7133369233530784,
      "grad_norm": 6.599672794342041,
      "learning_rate": 0.0,
      "loss": 0.6659,
      "step": 7948
    },
    {
      "epoch": 0.7134266738467061,
      "grad_norm": 3.3662285804748535,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 7949
    },
    {
      "epoch": 0.7135164243403339,
      "grad_norm": 2.893030881881714,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 7950
    },
    {
      "epoch": 0.7136061748339616,
      "grad_norm": 4.021646976470947,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 7951
    },
    {
      "epoch": 0.7136959253275893,
      "grad_norm": 3.338681697845459,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 7952
    },
    {
      "epoch": 0.713785675821217,
      "grad_norm": 2.557771921157837,
      "learning_rate": 0.0,
      "loss": 0.6817,
      "step": 7953
    },
    {
      "epoch": 0.7138754263148447,
      "grad_norm": 3.070178985595703,
      "learning_rate": 0.0,
      "loss": 0.7355,
      "step": 7954
    },
    {
      "epoch": 0.7139651768084725,
      "grad_norm": 2.460313558578491,
      "learning_rate": 0.0,
      "loss": 0.631,
      "step": 7955
    },
    {
      "epoch": 0.7140549273021002,
      "grad_norm": 3.3840198516845703,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 7956
    },
    {
      "epoch": 0.7141446777957279,
      "grad_norm": 3.1881144046783447,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 7957
    },
    {
      "epoch": 0.7142344282893556,
      "grad_norm": 2.677976131439209,
      "learning_rate": 0.0,
      "loss": 0.6614,
      "step": 7958
    },
    {
      "epoch": 0.7143241787829833,
      "grad_norm": 2.9867024421691895,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 7959
    },
    {
      "epoch": 0.7144139292766111,
      "grad_norm": 2.840749979019165,
      "learning_rate": 0.0,
      "loss": 0.7329,
      "step": 7960
    },
    {
      "epoch": 0.7145036797702388,
      "grad_norm": 3.3705759048461914,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 7961
    },
    {
      "epoch": 0.7145934302638665,
      "grad_norm": 2.9659574031829834,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 7962
    },
    {
      "epoch": 0.7146831807574942,
      "grad_norm": 16.452741622924805,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 7963
    },
    {
      "epoch": 0.7147729312511218,
      "grad_norm": 3.4380269050598145,
      "learning_rate": 0.0,
      "loss": 0.6767,
      "step": 7964
    },
    {
      "epoch": 0.7148626817447497,
      "grad_norm": 3.089252233505249,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 7965
    },
    {
      "epoch": 0.7149524322383773,
      "grad_norm": 2.94146728515625,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 7966
    },
    {
      "epoch": 0.715042182732005,
      "grad_norm": 2.6280369758605957,
      "learning_rate": 0.0,
      "loss": 0.786,
      "step": 7967
    },
    {
      "epoch": 0.7151319332256327,
      "grad_norm": 2.9069011211395264,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 7968
    },
    {
      "epoch": 0.7152216837192604,
      "grad_norm": 3.7442209720611572,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 7969
    },
    {
      "epoch": 0.7153114342128881,
      "grad_norm": 4.1620564460754395,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 7970
    },
    {
      "epoch": 0.7154011847065159,
      "grad_norm": 5.771120548248291,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 7971
    },
    {
      "epoch": 0.7154909352001436,
      "grad_norm": 3.638486862182617,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 7972
    },
    {
      "epoch": 0.7155806856937713,
      "grad_norm": 3.1670145988464355,
      "learning_rate": 0.0,
      "loss": 0.6782,
      "step": 7973
    },
    {
      "epoch": 0.715670436187399,
      "grad_norm": 3.968355417251587,
      "learning_rate": 0.0,
      "loss": 0.7778,
      "step": 7974
    },
    {
      "epoch": 0.7157601866810267,
      "grad_norm": 4.266495704650879,
      "learning_rate": 0.0,
      "loss": 0.7769,
      "step": 7975
    },
    {
      "epoch": 0.7158499371746545,
      "grad_norm": 2.6796553134918213,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 7976
    },
    {
      "epoch": 0.7159396876682822,
      "grad_norm": 2.99729061126709,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 7977
    },
    {
      "epoch": 0.7160294381619099,
      "grad_norm": 2.6839723587036133,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 7978
    },
    {
      "epoch": 0.7161191886555376,
      "grad_norm": 2.7069859504699707,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 7979
    },
    {
      "epoch": 0.7162089391491653,
      "grad_norm": 2.665750741958618,
      "learning_rate": 0.0,
      "loss": 0.7525,
      "step": 7980
    },
    {
      "epoch": 0.7162986896427931,
      "grad_norm": 4.3939595222473145,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 7981
    },
    {
      "epoch": 0.7163884401364208,
      "grad_norm": 3.9871888160705566,
      "learning_rate": 0.0,
      "loss": 0.6753,
      "step": 7982
    },
    {
      "epoch": 0.7164781906300485,
      "grad_norm": 2.6671361923217773,
      "learning_rate": 0.0,
      "loss": 0.6629,
      "step": 7983
    },
    {
      "epoch": 0.7165679411236762,
      "grad_norm": 2.859607219696045,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 7984
    },
    {
      "epoch": 0.7166576916173039,
      "grad_norm": 3.3010032176971436,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 7985
    },
    {
      "epoch": 0.7167474421109317,
      "grad_norm": 4.891960620880127,
      "learning_rate": 0.0,
      "loss": 0.6821,
      "step": 7986
    },
    {
      "epoch": 0.7168371926045594,
      "grad_norm": 2.902770757675171,
      "learning_rate": 0.0,
      "loss": 0.7992,
      "step": 7987
    },
    {
      "epoch": 0.716926943098187,
      "grad_norm": 3.294919490814209,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 7988
    },
    {
      "epoch": 0.7170166935918147,
      "grad_norm": 2.5028140544891357,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 7989
    },
    {
      "epoch": 0.7171064440854424,
      "grad_norm": 6.1974101066589355,
      "learning_rate": 0.0,
      "loss": 0.6721,
      "step": 7990
    },
    {
      "epoch": 0.7171961945790701,
      "grad_norm": 4.1154465675354,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 7991
    },
    {
      "epoch": 0.7172859450726979,
      "grad_norm": 5.911594867706299,
      "learning_rate": 0.0,
      "loss": 0.6913,
      "step": 7992
    },
    {
      "epoch": 0.7173756955663256,
      "grad_norm": 2.9318323135375977,
      "learning_rate": 0.0,
      "loss": 0.7143,
      "step": 7993
    },
    {
      "epoch": 0.7174654460599533,
      "grad_norm": 2.2546896934509277,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 7994
    },
    {
      "epoch": 0.717555196553581,
      "grad_norm": 6.680614471435547,
      "learning_rate": 0.0,
      "loss": 0.6752,
      "step": 7995
    },
    {
      "epoch": 0.7176449470472087,
      "grad_norm": 2.5360333919525146,
      "learning_rate": 0.0,
      "loss": 0.6657,
      "step": 7996
    },
    {
      "epoch": 0.7177346975408365,
      "grad_norm": 2.5371551513671875,
      "learning_rate": 0.0,
      "loss": 0.6622,
      "step": 7997
    },
    {
      "epoch": 0.7178244480344642,
      "grad_norm": 2.558952569961548,
      "learning_rate": 0.0,
      "loss": 0.7686,
      "step": 7998
    },
    {
      "epoch": 0.7179141985280919,
      "grad_norm": 8.596976280212402,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 7999
    },
    {
      "epoch": 0.7180039490217196,
      "grad_norm": 3.349405288696289,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 8000
    },
    {
      "epoch": 0.7180936995153473,
      "grad_norm": 3.5509932041168213,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 8001
    },
    {
      "epoch": 0.7181834500089751,
      "grad_norm": 2.5087406635284424,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 8002
    },
    {
      "epoch": 0.7182732005026028,
      "grad_norm": 6.710418701171875,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 8003
    },
    {
      "epoch": 0.7183629509962305,
      "grad_norm": 2.9013733863830566,
      "learning_rate": 0.0,
      "loss": 0.7381,
      "step": 8004
    },
    {
      "epoch": 0.7184527014898582,
      "grad_norm": 3.1360738277435303,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 8005
    },
    {
      "epoch": 0.7185424519834859,
      "grad_norm": 3.3450400829315186,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 8006
    },
    {
      "epoch": 0.7186322024771137,
      "grad_norm": 2.78021502494812,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 8007
    },
    {
      "epoch": 0.7187219529707414,
      "grad_norm": 3.1053051948547363,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 8008
    },
    {
      "epoch": 0.7188117034643691,
      "grad_norm": 3.281512498855591,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 8009
    },
    {
      "epoch": 0.7189014539579968,
      "grad_norm": 2.2870919704437256,
      "learning_rate": 0.0,
      "loss": 0.6654,
      "step": 8010
    },
    {
      "epoch": 0.7189912044516245,
      "grad_norm": 5.624756336212158,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 8011
    },
    {
      "epoch": 0.7190809549452521,
      "grad_norm": 2.8352482318878174,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 8012
    },
    {
      "epoch": 0.71917070543888,
      "grad_norm": 3.570247173309326,
      "learning_rate": 0.0,
      "loss": 0.6706,
      "step": 8013
    },
    {
      "epoch": 0.7192604559325076,
      "grad_norm": 2.9469218254089355,
      "learning_rate": 0.0,
      "loss": 0.6746,
      "step": 8014
    },
    {
      "epoch": 0.7193502064261353,
      "grad_norm": 2.7950901985168457,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 8015
    },
    {
      "epoch": 0.719439956919763,
      "grad_norm": 3.223053455352783,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 8016
    },
    {
      "epoch": 0.7195297074133907,
      "grad_norm": 3.035325527191162,
      "learning_rate": 0.0,
      "loss": 0.6759,
      "step": 8017
    },
    {
      "epoch": 0.7196194579070185,
      "grad_norm": 3.9583630561828613,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 8018
    },
    {
      "epoch": 0.7197092084006462,
      "grad_norm": 2.9880259037017822,
      "learning_rate": 0.0,
      "loss": 0.7279,
      "step": 8019
    },
    {
      "epoch": 0.7197989588942739,
      "grad_norm": 5.2681097984313965,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 8020
    },
    {
      "epoch": 0.7198887093879016,
      "grad_norm": 2.8799595832824707,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 8021
    },
    {
      "epoch": 0.7199784598815293,
      "grad_norm": 3.968092918395996,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 8022
    },
    {
      "epoch": 0.7200682103751571,
      "grad_norm": 2.949436664581299,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 8023
    },
    {
      "epoch": 0.7201579608687848,
      "grad_norm": 4.195887565612793,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 8024
    },
    {
      "epoch": 0.7202477113624125,
      "grad_norm": 3.9965412616729736,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 8025
    },
    {
      "epoch": 0.7203374618560402,
      "grad_norm": 3.040968179702759,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 8026
    },
    {
      "epoch": 0.7204272123496679,
      "grad_norm": 3.793804883956909,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 8027
    },
    {
      "epoch": 0.7205169628432956,
      "grad_norm": 4.4727911949157715,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 8028
    },
    {
      "epoch": 0.7206067133369234,
      "grad_norm": 3.99355411529541,
      "learning_rate": 0.0,
      "loss": 0.6718,
      "step": 8029
    },
    {
      "epoch": 0.7206964638305511,
      "grad_norm": 5.378203868865967,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 8030
    },
    {
      "epoch": 0.7207862143241788,
      "grad_norm": 3.164611339569092,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 8031
    },
    {
      "epoch": 0.7208759648178065,
      "grad_norm": 2.9410576820373535,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 8032
    },
    {
      "epoch": 0.7209657153114342,
      "grad_norm": 2.8474414348602295,
      "learning_rate": 0.0,
      "loss": 0.6978,
      "step": 8033
    },
    {
      "epoch": 0.721055465805062,
      "grad_norm": 2.2986505031585693,
      "learning_rate": 0.0,
      "loss": 0.6922,
      "step": 8034
    },
    {
      "epoch": 0.7211452162986897,
      "grad_norm": 4.860374927520752,
      "learning_rate": 0.0,
      "loss": 0.6781,
      "step": 8035
    },
    {
      "epoch": 0.7212349667923174,
      "grad_norm": 11.400650978088379,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 8036
    },
    {
      "epoch": 0.721324717285945,
      "grad_norm": 3.4769444465637207,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 8037
    },
    {
      "epoch": 0.7214144677795727,
      "grad_norm": 3.2254698276519775,
      "learning_rate": 0.0,
      "loss": 0.7551,
      "step": 8038
    },
    {
      "epoch": 0.7215042182732005,
      "grad_norm": 3.1022841930389404,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 8039
    },
    {
      "epoch": 0.7215939687668282,
      "grad_norm": 7.352534294128418,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 8040
    },
    {
      "epoch": 0.7216837192604559,
      "grad_norm": 3.3201117515563965,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 8041
    },
    {
      "epoch": 0.7217734697540836,
      "grad_norm": 4.301034450531006,
      "learning_rate": 0.0,
      "loss": 0.7431,
      "step": 8042
    },
    {
      "epoch": 0.7218632202477113,
      "grad_norm": 2.535445213317871,
      "learning_rate": 0.0,
      "loss": 0.652,
      "step": 8043
    },
    {
      "epoch": 0.7219529707413391,
      "grad_norm": 4.795647144317627,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 8044
    },
    {
      "epoch": 0.7220427212349668,
      "grad_norm": 10.435782432556152,
      "learning_rate": 0.0,
      "loss": 0.6673,
      "step": 8045
    },
    {
      "epoch": 0.7221324717285945,
      "grad_norm": 4.975407600402832,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 8046
    },
    {
      "epoch": 0.7222222222222222,
      "grad_norm": 6.099831581115723,
      "learning_rate": 0.0,
      "loss": 0.7279,
      "step": 8047
    },
    {
      "epoch": 0.7223119727158499,
      "grad_norm": 2.866530418395996,
      "learning_rate": 0.0,
      "loss": 0.6795,
      "step": 8048
    },
    {
      "epoch": 0.7224017232094776,
      "grad_norm": 3.129220724105835,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 8049
    },
    {
      "epoch": 0.7224914737031054,
      "grad_norm": 4.743639945983887,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 8050
    },
    {
      "epoch": 0.7225812241967331,
      "grad_norm": 5.039282321929932,
      "learning_rate": 0.0,
      "loss": 0.7685,
      "step": 8051
    },
    {
      "epoch": 0.7226709746903608,
      "grad_norm": 3.9893858432769775,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 8052
    },
    {
      "epoch": 0.7227607251839885,
      "grad_norm": 2.6273422241210938,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 8053
    },
    {
      "epoch": 0.7228504756776162,
      "grad_norm": 2.6816298961639404,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 8054
    },
    {
      "epoch": 0.722940226171244,
      "grad_norm": 3.515953302383423,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 8055
    },
    {
      "epoch": 0.7230299766648717,
      "grad_norm": 5.090493679046631,
      "learning_rate": 0.0,
      "loss": 0.7393,
      "step": 8056
    },
    {
      "epoch": 0.7231197271584994,
      "grad_norm": 3.13333797454834,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 8057
    },
    {
      "epoch": 0.7232094776521271,
      "grad_norm": 3.1513450145721436,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 8058
    },
    {
      "epoch": 0.7232992281457548,
      "grad_norm": 2.566103458404541,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 8059
    },
    {
      "epoch": 0.7233889786393826,
      "grad_norm": 2.606715202331543,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 8060
    },
    {
      "epoch": 0.7234787291330103,
      "grad_norm": 4.327167510986328,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 8061
    },
    {
      "epoch": 0.723568479626638,
      "grad_norm": 4.4875359535217285,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 8062
    },
    {
      "epoch": 0.7236582301202656,
      "grad_norm": 4.11500358581543,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 8063
    },
    {
      "epoch": 0.7237479806138933,
      "grad_norm": 5.137627601623535,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 8064
    },
    {
      "epoch": 0.7238377311075211,
      "grad_norm": 4.895646572113037,
      "learning_rate": 0.0,
      "loss": 0.7635,
      "step": 8065
    },
    {
      "epoch": 0.7239274816011488,
      "grad_norm": 2.581975221633911,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 8066
    },
    {
      "epoch": 0.7240172320947765,
      "grad_norm": 2.4145281314849854,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 8067
    },
    {
      "epoch": 0.7241069825884042,
      "grad_norm": 3.169689416885376,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 8068
    },
    {
      "epoch": 0.7241967330820319,
      "grad_norm": 2.6143863201141357,
      "learning_rate": 0.0,
      "loss": 0.6748,
      "step": 8069
    },
    {
      "epoch": 0.7242864835756596,
      "grad_norm": 2.649984121322632,
      "learning_rate": 0.0,
      "loss": 0.7339,
      "step": 8070
    },
    {
      "epoch": 0.7243762340692874,
      "grad_norm": 3.196129560470581,
      "learning_rate": 0.0,
      "loss": 0.6741,
      "step": 8071
    },
    {
      "epoch": 0.7244659845629151,
      "grad_norm": 3.2117679119110107,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 8072
    },
    {
      "epoch": 0.7245557350565428,
      "grad_norm": 20.970754623413086,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 8073
    },
    {
      "epoch": 0.7246454855501705,
      "grad_norm": 3.663331985473633,
      "learning_rate": 0.0,
      "loss": 0.751,
      "step": 8074
    },
    {
      "epoch": 0.7247352360437982,
      "grad_norm": 3.3890469074249268,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 8075
    },
    {
      "epoch": 0.724824986537426,
      "grad_norm": 2.592588186264038,
      "learning_rate": 0.0,
      "loss": 0.7322,
      "step": 8076
    },
    {
      "epoch": 0.7249147370310537,
      "grad_norm": 5.445084095001221,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 8077
    },
    {
      "epoch": 0.7250044875246814,
      "grad_norm": 2.9720191955566406,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 8078
    },
    {
      "epoch": 0.7250942380183091,
      "grad_norm": 3.2454957962036133,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 8079
    },
    {
      "epoch": 0.7251839885119368,
      "grad_norm": 3.2736661434173584,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 8080
    },
    {
      "epoch": 0.7252737390055646,
      "grad_norm": 2.809904098510742,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 8081
    },
    {
      "epoch": 0.7253634894991923,
      "grad_norm": 2.462618827819824,
      "learning_rate": 0.0,
      "loss": 0.7787,
      "step": 8082
    },
    {
      "epoch": 0.72545323999282,
      "grad_norm": 2.3270516395568848,
      "learning_rate": 0.0,
      "loss": 0.7877,
      "step": 8083
    },
    {
      "epoch": 0.7255429904864477,
      "grad_norm": 3.380253553390503,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 8084
    },
    {
      "epoch": 0.7256327409800754,
      "grad_norm": 2.3835322856903076,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 8085
    },
    {
      "epoch": 0.7257224914737032,
      "grad_norm": 2.9149324893951416,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 8086
    },
    {
      "epoch": 0.7258122419673309,
      "grad_norm": 2.531022310256958,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 8087
    },
    {
      "epoch": 0.7259019924609585,
      "grad_norm": 2.7115724086761475,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 8088
    },
    {
      "epoch": 0.7259917429545862,
      "grad_norm": 2.9470977783203125,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 8089
    },
    {
      "epoch": 0.7260814934482139,
      "grad_norm": 13.15540599822998,
      "learning_rate": 0.0,
      "loss": 0.7799,
      "step": 8090
    },
    {
      "epoch": 0.7261712439418416,
      "grad_norm": 3.373187780380249,
      "learning_rate": 0.0,
      "loss": 0.6837,
      "step": 8091
    },
    {
      "epoch": 0.7262609944354694,
      "grad_norm": 3.633284330368042,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 8092
    },
    {
      "epoch": 0.7263507449290971,
      "grad_norm": 2.3856990337371826,
      "learning_rate": 0.0,
      "loss": 0.7849,
      "step": 8093
    },
    {
      "epoch": 0.7264404954227248,
      "grad_norm": 3.322362184524536,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 8094
    },
    {
      "epoch": 0.7265302459163525,
      "grad_norm": 7.342915058135986,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 8095
    },
    {
      "epoch": 0.7266199964099802,
      "grad_norm": 4.028507232666016,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 8096
    },
    {
      "epoch": 0.726709746903608,
      "grad_norm": 3.1605756282806396,
      "learning_rate": 0.0,
      "loss": 0.6693,
      "step": 8097
    },
    {
      "epoch": 0.7267994973972357,
      "grad_norm": 2.4883124828338623,
      "learning_rate": 0.0,
      "loss": 0.6938,
      "step": 8098
    },
    {
      "epoch": 0.7268892478908634,
      "grad_norm": 2.6999502182006836,
      "learning_rate": 0.0,
      "loss": 0.6766,
      "step": 8099
    },
    {
      "epoch": 0.7269789983844911,
      "grad_norm": 3.054741859436035,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 8100
    },
    {
      "epoch": 0.7270687488781188,
      "grad_norm": 3.064068078994751,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 8101
    },
    {
      "epoch": 0.7271584993717466,
      "grad_norm": 3.0470852851867676,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 8102
    },
    {
      "epoch": 0.7272482498653743,
      "grad_norm": 27.734861373901367,
      "learning_rate": 0.0,
      "loss": 0.7609,
      "step": 8103
    },
    {
      "epoch": 0.727338000359002,
      "grad_norm": 4.132726669311523,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 8104
    },
    {
      "epoch": 0.7274277508526297,
      "grad_norm": 3.353201150894165,
      "learning_rate": 0.0,
      "loss": 0.7696,
      "step": 8105
    },
    {
      "epoch": 0.7275175013462574,
      "grad_norm": 3.0656535625457764,
      "learning_rate": 0.0,
      "loss": 0.6893,
      "step": 8106
    },
    {
      "epoch": 0.7276072518398852,
      "grad_norm": 3.7565953731536865,
      "learning_rate": 0.0,
      "loss": 0.6917,
      "step": 8107
    },
    {
      "epoch": 0.7276970023335129,
      "grad_norm": 2.8427698612213135,
      "learning_rate": 0.0,
      "loss": 0.8594,
      "step": 8108
    },
    {
      "epoch": 0.7277867528271406,
      "grad_norm": 3.6495213508605957,
      "learning_rate": 0.0,
      "loss": 0.7322,
      "step": 8109
    },
    {
      "epoch": 0.7278765033207683,
      "grad_norm": 3.4472744464874268,
      "learning_rate": 0.0,
      "loss": 0.8134,
      "step": 8110
    },
    {
      "epoch": 0.727966253814396,
      "grad_norm": 2.88051700592041,
      "learning_rate": 0.0,
      "loss": 0.6775,
      "step": 8111
    },
    {
      "epoch": 0.7280560043080236,
      "grad_norm": 2.590524673461914,
      "learning_rate": 0.0,
      "loss": 0.7582,
      "step": 8112
    },
    {
      "epoch": 0.7281457548016514,
      "grad_norm": 3.629701852798462,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 8113
    },
    {
      "epoch": 0.7282355052952791,
      "grad_norm": 3.181385040283203,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 8114
    },
    {
      "epoch": 0.7283252557889068,
      "grad_norm": 3.1736905574798584,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 8115
    },
    {
      "epoch": 0.7284150062825345,
      "grad_norm": 2.939603567123413,
      "learning_rate": 0.0,
      "loss": 0.6751,
      "step": 8116
    },
    {
      "epoch": 0.7285047567761622,
      "grad_norm": 4.052356243133545,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 8117
    },
    {
      "epoch": 0.72859450726979,
      "grad_norm": 2.821934700012207,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 8118
    },
    {
      "epoch": 0.7286842577634177,
      "grad_norm": 3.420724391937256,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 8119
    },
    {
      "epoch": 0.7287740082570454,
      "grad_norm": 4.727854251861572,
      "learning_rate": 0.0,
      "loss": 0.7043,
      "step": 8120
    },
    {
      "epoch": 0.7288637587506731,
      "grad_norm": 4.84345817565918,
      "learning_rate": 0.0,
      "loss": 0.7363,
      "step": 8121
    },
    {
      "epoch": 0.7289535092443008,
      "grad_norm": 6.328706741333008,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 8122
    },
    {
      "epoch": 0.7290432597379286,
      "grad_norm": 3.0373952388763428,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 8123
    },
    {
      "epoch": 0.7291330102315563,
      "grad_norm": 3.966876745223999,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 8124
    },
    {
      "epoch": 0.729222760725184,
      "grad_norm": 3.172173023223877,
      "learning_rate": 0.0,
      "loss": 0.6867,
      "step": 8125
    },
    {
      "epoch": 0.7293125112188117,
      "grad_norm": 2.4008333683013916,
      "learning_rate": 0.0,
      "loss": 0.7069,
      "step": 8126
    },
    {
      "epoch": 0.7294022617124394,
      "grad_norm": 2.9789247512817383,
      "learning_rate": 0.0,
      "loss": 0.7296,
      "step": 8127
    },
    {
      "epoch": 0.7294920122060672,
      "grad_norm": 2.919377565383911,
      "learning_rate": 0.0,
      "loss": 0.6607,
      "step": 8128
    },
    {
      "epoch": 0.7295817626996949,
      "grad_norm": 4.115672588348389,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 8129
    },
    {
      "epoch": 0.7296715131933226,
      "grad_norm": 2.7409520149230957,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 8130
    },
    {
      "epoch": 0.7297612636869503,
      "grad_norm": 2.7353615760803223,
      "learning_rate": 0.0,
      "loss": 0.689,
      "step": 8131
    },
    {
      "epoch": 0.729851014180578,
      "grad_norm": 2.4997918605804443,
      "learning_rate": 0.0,
      "loss": 0.6748,
      "step": 8132
    },
    {
      "epoch": 0.7299407646742057,
      "grad_norm": 2.7972497940063477,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 8133
    },
    {
      "epoch": 0.7300305151678335,
      "grad_norm": 2.919175863265991,
      "learning_rate": 0.0,
      "loss": 0.6577,
      "step": 8134
    },
    {
      "epoch": 0.7301202656614612,
      "grad_norm": 2.9646482467651367,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 8135
    },
    {
      "epoch": 0.7302100161550888,
      "grad_norm": 2.9874956607818604,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 8136
    },
    {
      "epoch": 0.7302997666487165,
      "grad_norm": 3.0714964866638184,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 8137
    },
    {
      "epoch": 0.7303895171423442,
      "grad_norm": 2.9541122913360596,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 8138
    },
    {
      "epoch": 0.730479267635972,
      "grad_norm": 3.6073756217956543,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 8139
    },
    {
      "epoch": 0.7305690181295997,
      "grad_norm": 4.5491461753845215,
      "learning_rate": 0.0,
      "loss": 0.7821,
      "step": 8140
    },
    {
      "epoch": 0.7306587686232274,
      "grad_norm": 2.9843358993530273,
      "learning_rate": 0.0,
      "loss": 0.6782,
      "step": 8141
    },
    {
      "epoch": 0.7307485191168551,
      "grad_norm": 3.307905435562134,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 8142
    },
    {
      "epoch": 0.7308382696104828,
      "grad_norm": 2.7397117614746094,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 8143
    },
    {
      "epoch": 0.7309280201041106,
      "grad_norm": 4.377048969268799,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 8144
    },
    {
      "epoch": 0.7310177705977383,
      "grad_norm": 2.2704989910125732,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 8145
    },
    {
      "epoch": 0.731107521091366,
      "grad_norm": 3.5167484283447266,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 8146
    },
    {
      "epoch": 0.7311972715849937,
      "grad_norm": 3.097141742706299,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 8147
    },
    {
      "epoch": 0.7312870220786214,
      "grad_norm": 2.86474871635437,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 8148
    },
    {
      "epoch": 0.7313767725722492,
      "grad_norm": 2.3839356899261475,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 8149
    },
    {
      "epoch": 0.7314665230658769,
      "grad_norm": 3.9195151329040527,
      "learning_rate": 0.0,
      "loss": 0.7141,
      "step": 8150
    },
    {
      "epoch": 0.7315562735595046,
      "grad_norm": 6.744812488555908,
      "learning_rate": 0.0,
      "loss": 0.6671,
      "step": 8151
    },
    {
      "epoch": 0.7316460240531323,
      "grad_norm": 2.77436900138855,
      "learning_rate": 0.0,
      "loss": 0.7835,
      "step": 8152
    },
    {
      "epoch": 0.73173577454676,
      "grad_norm": 2.711205244064331,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 8153
    },
    {
      "epoch": 0.7318255250403877,
      "grad_norm": 3.5297744274139404,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 8154
    },
    {
      "epoch": 0.7319152755340155,
      "grad_norm": 2.4119932651519775,
      "learning_rate": 0.0,
      "loss": 0.7447,
      "step": 8155
    },
    {
      "epoch": 0.7320050260276432,
      "grad_norm": 2.3875110149383545,
      "learning_rate": 0.0,
      "loss": 0.6768,
      "step": 8156
    },
    {
      "epoch": 0.7320947765212709,
      "grad_norm": 3.639822006225586,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 8157
    },
    {
      "epoch": 0.7321845270148986,
      "grad_norm": 3.980539083480835,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 8158
    },
    {
      "epoch": 0.7322742775085263,
      "grad_norm": 2.6639602184295654,
      "learning_rate": 0.0,
      "loss": 0.6764,
      "step": 8159
    },
    {
      "epoch": 0.732364028002154,
      "grad_norm": 2.6785542964935303,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 8160
    },
    {
      "epoch": 0.7324537784957817,
      "grad_norm": 3.8187572956085205,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 8161
    },
    {
      "epoch": 0.7325435289894094,
      "grad_norm": 4.337026119232178,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 8162
    },
    {
      "epoch": 0.7326332794830371,
      "grad_norm": 2.6551930904388428,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 8163
    },
    {
      "epoch": 0.7327230299766648,
      "grad_norm": 4.01829195022583,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 8164
    },
    {
      "epoch": 0.7328127804702926,
      "grad_norm": 3.44771146774292,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 8165
    },
    {
      "epoch": 0.7329025309639203,
      "grad_norm": 2.7249808311462402,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 8166
    },
    {
      "epoch": 0.732992281457548,
      "grad_norm": 3.1101152896881104,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 8167
    },
    {
      "epoch": 0.7330820319511757,
      "grad_norm": 3.532484769821167,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 8168
    },
    {
      "epoch": 0.7331717824448034,
      "grad_norm": 3.1197402477264404,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 8169
    },
    {
      "epoch": 0.7332615329384312,
      "grad_norm": 2.309152603149414,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 8170
    },
    {
      "epoch": 0.7333512834320589,
      "grad_norm": 2.701652765274048,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 8171
    },
    {
      "epoch": 0.7334410339256866,
      "grad_norm": 3.7687511444091797,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 8172
    },
    {
      "epoch": 0.7335307844193143,
      "grad_norm": 2.7492058277130127,
      "learning_rate": 0.0,
      "loss": 0.6839,
      "step": 8173
    },
    {
      "epoch": 0.733620534912942,
      "grad_norm": 2.7110037803649902,
      "learning_rate": 0.0,
      "loss": 0.7036,
      "step": 8174
    },
    {
      "epoch": 0.7337102854065697,
      "grad_norm": 4.218677043914795,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 8175
    },
    {
      "epoch": 0.7338000359001975,
      "grad_norm": 4.476298809051514,
      "learning_rate": 0.0,
      "loss": 0.7765,
      "step": 8176
    },
    {
      "epoch": 0.7338897863938252,
      "grad_norm": 2.7620160579681396,
      "learning_rate": 0.0,
      "loss": 0.7983,
      "step": 8177
    },
    {
      "epoch": 0.7339795368874529,
      "grad_norm": 2.797774076461792,
      "learning_rate": 0.0,
      "loss": 0.6645,
      "step": 8178
    },
    {
      "epoch": 0.7340692873810806,
      "grad_norm": 2.338076591491699,
      "learning_rate": 0.0,
      "loss": 0.7686,
      "step": 8179
    },
    {
      "epoch": 0.7341590378747083,
      "grad_norm": 2.606471538543701,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 8180
    },
    {
      "epoch": 0.7342487883683361,
      "grad_norm": 3.634997844696045,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 8181
    },
    {
      "epoch": 0.7343385388619638,
      "grad_norm": 2.996821641921997,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 8182
    },
    {
      "epoch": 0.7344282893555915,
      "grad_norm": 2.9417402744293213,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 8183
    },
    {
      "epoch": 0.7345180398492192,
      "grad_norm": 4.5442304611206055,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 8184
    },
    {
      "epoch": 0.7346077903428468,
      "grad_norm": 12.71968936920166,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 8185
    },
    {
      "epoch": 0.7346975408364746,
      "grad_norm": 4.090799331665039,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 8186
    },
    {
      "epoch": 0.7347872913301023,
      "grad_norm": 4.451990604400635,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 8187
    },
    {
      "epoch": 0.73487704182373,
      "grad_norm": 3.2293996810913086,
      "learning_rate": 0.0,
      "loss": 0.6833,
      "step": 8188
    },
    {
      "epoch": 0.7349667923173577,
      "grad_norm": 3.3601877689361572,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 8189
    },
    {
      "epoch": 0.7350565428109854,
      "grad_norm": 2.9525146484375,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 8190
    },
    {
      "epoch": 0.7351462933046132,
      "grad_norm": 3.5927035808563232,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 8191
    },
    {
      "epoch": 0.7352360437982409,
      "grad_norm": 2.6307969093322754,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 8192
    },
    {
      "epoch": 0.7353257942918686,
      "grad_norm": 2.8198776245117188,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 8193
    },
    {
      "epoch": 0.7354155447854963,
      "grad_norm": 2.561823844909668,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 8194
    },
    {
      "epoch": 0.735505295279124,
      "grad_norm": 3.3734498023986816,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 8195
    },
    {
      "epoch": 0.7355950457727517,
      "grad_norm": 3.0896520614624023,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 8196
    },
    {
      "epoch": 0.7356847962663795,
      "grad_norm": 2.691679000854492,
      "learning_rate": 0.0,
      "loss": 0.705,
      "step": 8197
    },
    {
      "epoch": 0.7357745467600072,
      "grad_norm": 3.7309868335723877,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 8198
    },
    {
      "epoch": 0.7358642972536349,
      "grad_norm": 2.790957450866699,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 8199
    },
    {
      "epoch": 0.7359540477472626,
      "grad_norm": 2.7641937732696533,
      "learning_rate": 0.0,
      "loss": 0.6703,
      "step": 8200
    },
    {
      "epoch": 0.7360437982408903,
      "grad_norm": 3.003951072692871,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 8201
    },
    {
      "epoch": 0.7361335487345181,
      "grad_norm": 2.6914143562316895,
      "learning_rate": 0.0,
      "loss": 0.6584,
      "step": 8202
    },
    {
      "epoch": 0.7362232992281458,
      "grad_norm": 3.110915184020996,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 8203
    },
    {
      "epoch": 0.7363130497217735,
      "grad_norm": 3.2344658374786377,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 8204
    },
    {
      "epoch": 0.7364028002154012,
      "grad_norm": 3.198432207107544,
      "learning_rate": 0.0,
      "loss": 0.7904,
      "step": 8205
    },
    {
      "epoch": 0.7364925507090289,
      "grad_norm": 2.447341203689575,
      "learning_rate": 0.0,
      "loss": 0.7585,
      "step": 8206
    },
    {
      "epoch": 0.7365823012026567,
      "grad_norm": 3.6243109703063965,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 8207
    },
    {
      "epoch": 0.7366720516962844,
      "grad_norm": 2.6117310523986816,
      "learning_rate": 0.0,
      "loss": 0.8879,
      "step": 8208
    },
    {
      "epoch": 0.736761802189912,
      "grad_norm": 5.591320037841797,
      "learning_rate": 0.0,
      "loss": 0.7618,
      "step": 8209
    },
    {
      "epoch": 0.7368515526835397,
      "grad_norm": 2.801409959793091,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 8210
    },
    {
      "epoch": 0.7369413031771674,
      "grad_norm": 3.74090576171875,
      "learning_rate": 0.0,
      "loss": 0.6721,
      "step": 8211
    },
    {
      "epoch": 0.7370310536707952,
      "grad_norm": 2.3639450073242188,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 8212
    },
    {
      "epoch": 0.7371208041644229,
      "grad_norm": 3.253495693206787,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 8213
    },
    {
      "epoch": 0.7372105546580506,
      "grad_norm": 2.9150564670562744,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 8214
    },
    {
      "epoch": 0.7373003051516783,
      "grad_norm": 3.308318853378296,
      "learning_rate": 0.0,
      "loss": 0.7457,
      "step": 8215
    },
    {
      "epoch": 0.737390055645306,
      "grad_norm": 3.844870090484619,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 8216
    },
    {
      "epoch": 0.7374798061389337,
      "grad_norm": 6.155542373657227,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 8217
    },
    {
      "epoch": 0.7375695566325615,
      "grad_norm": 3.4268155097961426,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 8218
    },
    {
      "epoch": 0.7376593071261892,
      "grad_norm": 3.6593708992004395,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 8219
    },
    {
      "epoch": 0.7377490576198169,
      "grad_norm": 2.7144577503204346,
      "learning_rate": 0.0,
      "loss": 0.6936,
      "step": 8220
    },
    {
      "epoch": 0.7378388081134446,
      "grad_norm": 3.5833828449249268,
      "learning_rate": 0.0,
      "loss": 0.6966,
      "step": 8221
    },
    {
      "epoch": 0.7379285586070723,
      "grad_norm": 3.592447519302368,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 8222
    },
    {
      "epoch": 0.7380183091007001,
      "grad_norm": 2.65354323387146,
      "learning_rate": 0.0,
      "loss": 0.7956,
      "step": 8223
    },
    {
      "epoch": 0.7381080595943278,
      "grad_norm": 3.0171396732330322,
      "learning_rate": 0.0,
      "loss": 0.6938,
      "step": 8224
    },
    {
      "epoch": 0.7381978100879555,
      "grad_norm": 3.1196837425231934,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 8225
    },
    {
      "epoch": 0.7382875605815832,
      "grad_norm": 2.9203052520751953,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 8226
    },
    {
      "epoch": 0.7383773110752109,
      "grad_norm": 3.3515467643737793,
      "learning_rate": 0.0,
      "loss": 0.6646,
      "step": 8227
    },
    {
      "epoch": 0.7384670615688387,
      "grad_norm": 2.3222851753234863,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 8228
    },
    {
      "epoch": 0.7385568120624664,
      "grad_norm": 4.777186393737793,
      "learning_rate": 0.0,
      "loss": 0.7837,
      "step": 8229
    },
    {
      "epoch": 0.7386465625560941,
      "grad_norm": 3.0155482292175293,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 8230
    },
    {
      "epoch": 0.7387363130497218,
      "grad_norm": 2.9734911918640137,
      "learning_rate": 0.0,
      "loss": 0.7938,
      "step": 8231
    },
    {
      "epoch": 0.7388260635433495,
      "grad_norm": 3.0909736156463623,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 8232
    },
    {
      "epoch": 0.7389158140369773,
      "grad_norm": 2.6503989696502686,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 8233
    },
    {
      "epoch": 0.739005564530605,
      "grad_norm": 2.947322368621826,
      "learning_rate": 0.0,
      "loss": 0.7706,
      "step": 8234
    },
    {
      "epoch": 0.7390953150242326,
      "grad_norm": 3.1209075450897217,
      "learning_rate": 0.0,
      "loss": 0.6712,
      "step": 8235
    },
    {
      "epoch": 0.7391850655178603,
      "grad_norm": 2.294346570968628,
      "learning_rate": 0.0,
      "loss": 0.6588,
      "step": 8236
    },
    {
      "epoch": 0.739274816011488,
      "grad_norm": 2.358947515487671,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 8237
    },
    {
      "epoch": 0.7393645665051157,
      "grad_norm": 3.0505292415618896,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 8238
    },
    {
      "epoch": 0.7394543169987435,
      "grad_norm": 2.6817126274108887,
      "learning_rate": 0.0,
      "loss": 0.6761,
      "step": 8239
    },
    {
      "epoch": 0.7395440674923712,
      "grad_norm": 2.815343141555786,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 8240
    },
    {
      "epoch": 0.7396338179859989,
      "grad_norm": 3.400564432144165,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 8241
    },
    {
      "epoch": 0.7397235684796266,
      "grad_norm": 3.118427276611328,
      "learning_rate": 0.0,
      "loss": 0.765,
      "step": 8242
    },
    {
      "epoch": 0.7398133189732543,
      "grad_norm": 3.1177539825439453,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 8243
    },
    {
      "epoch": 0.7399030694668821,
      "grad_norm": 2.638352632522583,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 8244
    },
    {
      "epoch": 0.7399928199605098,
      "grad_norm": 2.5512948036193848,
      "learning_rate": 0.0,
      "loss": 0.7052,
      "step": 8245
    },
    {
      "epoch": 0.7400825704541375,
      "grad_norm": 2.783942222595215,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 8246
    },
    {
      "epoch": 0.7401723209477652,
      "grad_norm": 4.314289093017578,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 8247
    },
    {
      "epoch": 0.7402620714413929,
      "grad_norm": 2.3909013271331787,
      "learning_rate": 0.0,
      "loss": 0.7631,
      "step": 8248
    },
    {
      "epoch": 0.7403518219350207,
      "grad_norm": 3.044003963470459,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 8249
    },
    {
      "epoch": 0.7404415724286484,
      "grad_norm": 4.995213031768799,
      "learning_rate": 0.0,
      "loss": 0.7406,
      "step": 8250
    },
    {
      "epoch": 0.7405313229222761,
      "grad_norm": 3.6566851139068604,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 8251
    },
    {
      "epoch": 0.7406210734159038,
      "grad_norm": 3.7082016468048096,
      "learning_rate": 0.0,
      "loss": 0.733,
      "step": 8252
    },
    {
      "epoch": 0.7407108239095315,
      "grad_norm": 2.900940418243408,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 8253
    },
    {
      "epoch": 0.7408005744031592,
      "grad_norm": 2.657762050628662,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 8254
    },
    {
      "epoch": 0.740890324896787,
      "grad_norm": 4.3275675773620605,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 8255
    },
    {
      "epoch": 0.7409800753904147,
      "grad_norm": 3.6661322116851807,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 8256
    },
    {
      "epoch": 0.7410698258840424,
      "grad_norm": 2.957205057144165,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 8257
    },
    {
      "epoch": 0.74115957637767,
      "grad_norm": 2.6568171977996826,
      "learning_rate": 0.0,
      "loss": 0.6641,
      "step": 8258
    },
    {
      "epoch": 0.7412493268712977,
      "grad_norm": 2.8010032176971436,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 8259
    },
    {
      "epoch": 0.7413390773649255,
      "grad_norm": 2.9204864501953125,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 8260
    },
    {
      "epoch": 0.7414288278585532,
      "grad_norm": 2.578639030456543,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 8261
    },
    {
      "epoch": 0.7415185783521809,
      "grad_norm": 2.8266828060150146,
      "learning_rate": 0.0,
      "loss": 0.6944,
      "step": 8262
    },
    {
      "epoch": 0.7416083288458086,
      "grad_norm": 3.756776809692383,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 8263
    },
    {
      "epoch": 0.7416980793394363,
      "grad_norm": 3.3439881801605225,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 8264
    },
    {
      "epoch": 0.7417878298330641,
      "grad_norm": 2.7979722023010254,
      "learning_rate": 0.0,
      "loss": 0.614,
      "step": 8265
    },
    {
      "epoch": 0.7418775803266918,
      "grad_norm": 2.2630934715270996,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 8266
    },
    {
      "epoch": 0.7419673308203195,
      "grad_norm": 2.6778504848480225,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 8267
    },
    {
      "epoch": 0.7420570813139472,
      "grad_norm": 2.753413438796997,
      "learning_rate": 0.0,
      "loss": 0.6821,
      "step": 8268
    },
    {
      "epoch": 0.7421468318075749,
      "grad_norm": 3.9681951999664307,
      "learning_rate": 0.0,
      "loss": 0.6748,
      "step": 8269
    },
    {
      "epoch": 0.7422365823012027,
      "grad_norm": 3.1474690437316895,
      "learning_rate": 0.0,
      "loss": 0.682,
      "step": 8270
    },
    {
      "epoch": 0.7423263327948304,
      "grad_norm": 2.6012086868286133,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 8271
    },
    {
      "epoch": 0.7424160832884581,
      "grad_norm": 2.4591755867004395,
      "learning_rate": 0.0,
      "loss": 0.7061,
      "step": 8272
    },
    {
      "epoch": 0.7425058337820858,
      "grad_norm": 3.044235944747925,
      "learning_rate": 0.0,
      "loss": 0.6803,
      "step": 8273
    },
    {
      "epoch": 0.7425955842757135,
      "grad_norm": 3.201937675476074,
      "learning_rate": 0.0,
      "loss": 0.6708,
      "step": 8274
    },
    {
      "epoch": 0.7426853347693412,
      "grad_norm": 2.3023862838745117,
      "learning_rate": 0.0,
      "loss": 0.6796,
      "step": 8275
    },
    {
      "epoch": 0.742775085262969,
      "grad_norm": 2.4061949253082275,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 8276
    },
    {
      "epoch": 0.7428648357565967,
      "grad_norm": 5.569538116455078,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 8277
    },
    {
      "epoch": 0.7429545862502244,
      "grad_norm": 4.517435550689697,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 8278
    },
    {
      "epoch": 0.7430443367438521,
      "grad_norm": 3.929344892501831,
      "learning_rate": 0.0,
      "loss": 0.656,
      "step": 8279
    },
    {
      "epoch": 0.7431340872374798,
      "grad_norm": 2.9695160388946533,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 8280
    },
    {
      "epoch": 0.7432238377311076,
      "grad_norm": 4.64859676361084,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 8281
    },
    {
      "epoch": 0.7433135882247353,
      "grad_norm": 2.740659236907959,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 8282
    },
    {
      "epoch": 0.743403338718363,
      "grad_norm": 2.856995105743408,
      "learning_rate": 0.0,
      "loss": 0.6848,
      "step": 8283
    },
    {
      "epoch": 0.7434930892119906,
      "grad_norm": 2.436558961868286,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 8284
    },
    {
      "epoch": 0.7435828397056183,
      "grad_norm": 3.0253565311431885,
      "learning_rate": 0.0,
      "loss": 0.7377,
      "step": 8285
    },
    {
      "epoch": 0.7436725901992461,
      "grad_norm": 6.10073709487915,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 8286
    },
    {
      "epoch": 0.7437623406928738,
      "grad_norm": 2.992950677871704,
      "learning_rate": 0.0,
      "loss": 0.6821,
      "step": 8287
    },
    {
      "epoch": 0.7438520911865015,
      "grad_norm": 5.653541088104248,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 8288
    },
    {
      "epoch": 0.7439418416801292,
      "grad_norm": 9.615885734558105,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 8289
    },
    {
      "epoch": 0.7440315921737569,
      "grad_norm": 2.8662898540496826,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 8290
    },
    {
      "epoch": 0.7441213426673847,
      "grad_norm": 2.504051446914673,
      "learning_rate": 0.0,
      "loss": 0.7653,
      "step": 8291
    },
    {
      "epoch": 0.7442110931610124,
      "grad_norm": 3.6911087036132812,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 8292
    },
    {
      "epoch": 0.7443008436546401,
      "grad_norm": 2.9777987003326416,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 8293
    },
    {
      "epoch": 0.7443905941482678,
      "grad_norm": 2.543466329574585,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 8294
    },
    {
      "epoch": 0.7444803446418955,
      "grad_norm": 2.500335693359375,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 8295
    },
    {
      "epoch": 0.7445700951355232,
      "grad_norm": 2.6115522384643555,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 8296
    },
    {
      "epoch": 0.744659845629151,
      "grad_norm": 3.967689037322998,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 8297
    },
    {
      "epoch": 0.7447495961227787,
      "grad_norm": 2.4279212951660156,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 8298
    },
    {
      "epoch": 0.7448393466164064,
      "grad_norm": 2.8565118312835693,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 8299
    },
    {
      "epoch": 0.7449290971100341,
      "grad_norm": 3.3519041538238525,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 8300
    },
    {
      "epoch": 0.7450188476036618,
      "grad_norm": 3.982125997543335,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 8301
    },
    {
      "epoch": 0.7451085980972896,
      "grad_norm": 3.136671543121338,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 8302
    },
    {
      "epoch": 0.7451983485909173,
      "grad_norm": 2.2147202491760254,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 8303
    },
    {
      "epoch": 0.745288099084545,
      "grad_norm": 2.535313844680786,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 8304
    },
    {
      "epoch": 0.7453778495781727,
      "grad_norm": 2.5009689331054688,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 8305
    },
    {
      "epoch": 0.7454676000718004,
      "grad_norm": 2.596683979034424,
      "learning_rate": 0.0,
      "loss": 0.6914,
      "step": 8306
    },
    {
      "epoch": 0.7455573505654282,
      "grad_norm": 4.210161209106445,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 8307
    },
    {
      "epoch": 0.7456471010590559,
      "grad_norm": 3.249544620513916,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 8308
    },
    {
      "epoch": 0.7457368515526835,
      "grad_norm": 2.3271782398223877,
      "learning_rate": 0.0,
      "loss": 0.7579,
      "step": 8309
    },
    {
      "epoch": 0.7458266020463112,
      "grad_norm": 4.266336441040039,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 8310
    },
    {
      "epoch": 0.7459163525399389,
      "grad_norm": 3.3204691410064697,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 8311
    },
    {
      "epoch": 0.7460061030335667,
      "grad_norm": 2.70443058013916,
      "learning_rate": 0.0,
      "loss": 0.6869,
      "step": 8312
    },
    {
      "epoch": 0.7460958535271944,
      "grad_norm": 3.2091429233551025,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 8313
    },
    {
      "epoch": 0.7461856040208221,
      "grad_norm": 2.808995246887207,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 8314
    },
    {
      "epoch": 0.7462753545144498,
      "grad_norm": 3.4610297679901123,
      "learning_rate": 0.0,
      "loss": 0.7119,
      "step": 8315
    },
    {
      "epoch": 0.7463651050080775,
      "grad_norm": 2.635528326034546,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 8316
    },
    {
      "epoch": 0.7464548555017052,
      "grad_norm": 2.9413046836853027,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 8317
    },
    {
      "epoch": 0.746544605995333,
      "grad_norm": 3.0725626945495605,
      "learning_rate": 0.0,
      "loss": 0.6725,
      "step": 8318
    },
    {
      "epoch": 0.7466343564889607,
      "grad_norm": 3.60530424118042,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 8319
    },
    {
      "epoch": 0.7467241069825884,
      "grad_norm": 5.953869342803955,
      "learning_rate": 0.0,
      "loss": 0.6821,
      "step": 8320
    },
    {
      "epoch": 0.7468138574762161,
      "grad_norm": 4.2872233390808105,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 8321
    },
    {
      "epoch": 0.7469036079698438,
      "grad_norm": 2.797107696533203,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 8322
    },
    {
      "epoch": 0.7469933584634716,
      "grad_norm": 2.93640398979187,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 8323
    },
    {
      "epoch": 0.7470831089570993,
      "grad_norm": 3.8940274715423584,
      "learning_rate": 0.0,
      "loss": 0.6686,
      "step": 8324
    },
    {
      "epoch": 0.747172859450727,
      "grad_norm": 4.344468593597412,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 8325
    },
    {
      "epoch": 0.7472626099443547,
      "grad_norm": 2.989114761352539,
      "learning_rate": 0.0,
      "loss": 0.6782,
      "step": 8326
    },
    {
      "epoch": 0.7473523604379824,
      "grad_norm": 55.91313934326172,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 8327
    },
    {
      "epoch": 0.7474421109316102,
      "grad_norm": 3.515658378601074,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 8328
    },
    {
      "epoch": 0.7475318614252379,
      "grad_norm": 7.0695061683654785,
      "learning_rate": 0.0,
      "loss": 0.802,
      "step": 8329
    },
    {
      "epoch": 0.7476216119188656,
      "grad_norm": 8.83704948425293,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 8330
    },
    {
      "epoch": 0.7477113624124933,
      "grad_norm": 3.1537623405456543,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 8331
    },
    {
      "epoch": 0.747801112906121,
      "grad_norm": 3.36639404296875,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 8332
    },
    {
      "epoch": 0.7478908633997488,
      "grad_norm": 2.6717283725738525,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 8333
    },
    {
      "epoch": 0.7479806138933764,
      "grad_norm": 2.485050678253174,
      "learning_rate": 0.0,
      "loss": 0.7199,
      "step": 8334
    },
    {
      "epoch": 0.7480703643870041,
      "grad_norm": 3.245203971862793,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 8335
    },
    {
      "epoch": 0.7481601148806318,
      "grad_norm": 3.5364153385162354,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 8336
    },
    {
      "epoch": 0.7482498653742595,
      "grad_norm": 4.03292989730835,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 8337
    },
    {
      "epoch": 0.7483396158678872,
      "grad_norm": 2.3846726417541504,
      "learning_rate": 0.0,
      "loss": 0.6899,
      "step": 8338
    },
    {
      "epoch": 0.748429366361515,
      "grad_norm": 3.039889335632324,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 8339
    },
    {
      "epoch": 0.7485191168551427,
      "grad_norm": 6.225986480712891,
      "learning_rate": 0.0,
      "loss": 0.8163,
      "step": 8340
    },
    {
      "epoch": 0.7486088673487704,
      "grad_norm": 2.9302690029144287,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 8341
    },
    {
      "epoch": 0.7486986178423981,
      "grad_norm": 4.188964366912842,
      "learning_rate": 0.0,
      "loss": 0.7635,
      "step": 8342
    },
    {
      "epoch": 0.7487883683360258,
      "grad_norm": 3.075173854827881,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 8343
    },
    {
      "epoch": 0.7488781188296536,
      "grad_norm": 2.385859966278076,
      "learning_rate": 0.0,
      "loss": 0.7154,
      "step": 8344
    },
    {
      "epoch": 0.7489678693232813,
      "grad_norm": 4.84087610244751,
      "learning_rate": 0.0,
      "loss": 0.6918,
      "step": 8345
    },
    {
      "epoch": 0.749057619816909,
      "grad_norm": 3.03262996673584,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 8346
    },
    {
      "epoch": 0.7491473703105367,
      "grad_norm": 2.8062820434570312,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 8347
    },
    {
      "epoch": 0.7492371208041644,
      "grad_norm": 3.0244393348693848,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 8348
    },
    {
      "epoch": 0.7493268712977922,
      "grad_norm": 2.904611349105835,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 8349
    },
    {
      "epoch": 0.7494166217914199,
      "grad_norm": 3.528838872909546,
      "learning_rate": 0.0,
      "loss": 0.6734,
      "step": 8350
    },
    {
      "epoch": 0.7495063722850476,
      "grad_norm": 3.0027377605438232,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 8351
    },
    {
      "epoch": 0.7495961227786753,
      "grad_norm": 3.722785711288452,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 8352
    },
    {
      "epoch": 0.749685873272303,
      "grad_norm": 5.031952857971191,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 8353
    },
    {
      "epoch": 0.7497756237659308,
      "grad_norm": 3.5564301013946533,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 8354
    },
    {
      "epoch": 0.7498653742595585,
      "grad_norm": 2.1973791122436523,
      "learning_rate": 0.0,
      "loss": 0.7001,
      "step": 8355
    },
    {
      "epoch": 0.7499551247531862,
      "grad_norm": 3.0129199028015137,
      "learning_rate": 0.0,
      "loss": 0.6674,
      "step": 8356
    },
    {
      "epoch": 0.7500448752468138,
      "grad_norm": 2.425133466720581,
      "learning_rate": 0.0,
      "loss": 0.6675,
      "step": 8357
    },
    {
      "epoch": 0.7501346257404415,
      "grad_norm": 2.9445536136627197,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 8358
    },
    {
      "epoch": 0.7502243762340692,
      "grad_norm": 2.468834400177002,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 8359
    },
    {
      "epoch": 0.750314126727697,
      "grad_norm": 2.7717299461364746,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 8360
    },
    {
      "epoch": 0.7504038772213247,
      "grad_norm": 2.360192060470581,
      "learning_rate": 0.0,
      "loss": 0.6663,
      "step": 8361
    },
    {
      "epoch": 0.7504936277149524,
      "grad_norm": 2.792344093322754,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 8362
    },
    {
      "epoch": 0.7505833782085801,
      "grad_norm": 2.848275661468506,
      "learning_rate": 0.0,
      "loss": 0.7044,
      "step": 8363
    },
    {
      "epoch": 0.7506731287022078,
      "grad_norm": 2.6936886310577393,
      "learning_rate": 0.0,
      "loss": 0.6815,
      "step": 8364
    },
    {
      "epoch": 0.7507628791958356,
      "grad_norm": 18.314273834228516,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 8365
    },
    {
      "epoch": 0.7508526296894633,
      "grad_norm": 2.9431564807891846,
      "learning_rate": 0.0,
      "loss": 0.7757,
      "step": 8366
    },
    {
      "epoch": 0.750942380183091,
      "grad_norm": 3.067095994949341,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 8367
    },
    {
      "epoch": 0.7510321306767187,
      "grad_norm": 3.519758939743042,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 8368
    },
    {
      "epoch": 0.7511218811703464,
      "grad_norm": 2.9393270015716553,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 8369
    },
    {
      "epoch": 0.7512116316639742,
      "grad_norm": 2.6071879863739014,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 8370
    },
    {
      "epoch": 0.7513013821576019,
      "grad_norm": 2.711324691772461,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 8371
    },
    {
      "epoch": 0.7513911326512296,
      "grad_norm": 2.5402684211730957,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 8372
    },
    {
      "epoch": 0.7514808831448573,
      "grad_norm": 2.4127488136291504,
      "learning_rate": 0.0,
      "loss": 0.6808,
      "step": 8373
    },
    {
      "epoch": 0.751570633638485,
      "grad_norm": 2.5233263969421387,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 8374
    },
    {
      "epoch": 0.7516603841321128,
      "grad_norm": 3.039777994155884,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 8375
    },
    {
      "epoch": 0.7517501346257405,
      "grad_norm": 3.0412299633026123,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 8376
    },
    {
      "epoch": 0.7518398851193682,
      "grad_norm": 2.1539320945739746,
      "learning_rate": 0.0,
      "loss": 0.7955,
      "step": 8377
    },
    {
      "epoch": 0.7519296356129959,
      "grad_norm": 2.9724717140197754,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 8378
    },
    {
      "epoch": 0.7520193861066236,
      "grad_norm": 2.476222515106201,
      "learning_rate": 0.0,
      "loss": 0.7193,
      "step": 8379
    },
    {
      "epoch": 0.7521091366002512,
      "grad_norm": 2.5583581924438477,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 8380
    },
    {
      "epoch": 0.752198887093879,
      "grad_norm": 2.745737314224243,
      "learning_rate": 0.0,
      "loss": 0.6714,
      "step": 8381
    },
    {
      "epoch": 0.7522886375875067,
      "grad_norm": 2.1829187870025635,
      "learning_rate": 0.0,
      "loss": 0.6535,
      "step": 8382
    },
    {
      "epoch": 0.7523783880811344,
      "grad_norm": 1.9319270849227905,
      "learning_rate": 0.0,
      "loss": 0.6947,
      "step": 8383
    },
    {
      "epoch": 0.7524681385747621,
      "grad_norm": 3.279237747192383,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 8384
    },
    {
      "epoch": 0.7525578890683898,
      "grad_norm": 2.146920919418335,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 8385
    },
    {
      "epoch": 0.7526476395620176,
      "grad_norm": 2.3618075847625732,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 8386
    },
    {
      "epoch": 0.7527373900556453,
      "grad_norm": 3.58316969871521,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 8387
    },
    {
      "epoch": 0.752827140549273,
      "grad_norm": 3.8222527503967285,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 8388
    },
    {
      "epoch": 0.7529168910429007,
      "grad_norm": 3.0932321548461914,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 8389
    },
    {
      "epoch": 0.7530066415365284,
      "grad_norm": 3.530789613723755,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 8390
    },
    {
      "epoch": 0.7530963920301562,
      "grad_norm": 3.5927274227142334,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 8391
    },
    {
      "epoch": 0.7531861425237839,
      "grad_norm": 2.806518316268921,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 8392
    },
    {
      "epoch": 0.7532758930174116,
      "grad_norm": 2.0840609073638916,
      "learning_rate": 0.0,
      "loss": 0.676,
      "step": 8393
    },
    {
      "epoch": 0.7533656435110393,
      "grad_norm": 3.1542725563049316,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 8394
    },
    {
      "epoch": 0.753455394004667,
      "grad_norm": 4.100200176239014,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 8395
    },
    {
      "epoch": 0.7535451444982948,
      "grad_norm": 2.5150716304779053,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 8396
    },
    {
      "epoch": 0.7536348949919225,
      "grad_norm": 2.4405126571655273,
      "learning_rate": 0.0,
      "loss": 0.7057,
      "step": 8397
    },
    {
      "epoch": 0.7537246454855502,
      "grad_norm": 2.6879594326019287,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 8398
    },
    {
      "epoch": 0.7538143959791779,
      "grad_norm": 37.82470703125,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 8399
    },
    {
      "epoch": 0.7539041464728056,
      "grad_norm": 3.6331684589385986,
      "learning_rate": 0.0,
      "loss": 0.7766,
      "step": 8400
    },
    {
      "epoch": 0.7539938969664333,
      "grad_norm": 2.6938316822052,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 8401
    },
    {
      "epoch": 0.7540836474600611,
      "grad_norm": 2.514549493789673,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 8402
    },
    {
      "epoch": 0.7541733979536888,
      "grad_norm": 7.197820663452148,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 8403
    },
    {
      "epoch": 0.7542631484473165,
      "grad_norm": 4.175751209259033,
      "learning_rate": 0.0,
      "loss": 0.78,
      "step": 8404
    },
    {
      "epoch": 0.7543528989409441,
      "grad_norm": 2.4891207218170166,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 8405
    },
    {
      "epoch": 0.7544426494345718,
      "grad_norm": 25.214733123779297,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 8406
    },
    {
      "epoch": 0.7545323999281996,
      "grad_norm": 4.2320556640625,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 8407
    },
    {
      "epoch": 0.7546221504218273,
      "grad_norm": 4.789122581481934,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 8408
    },
    {
      "epoch": 0.754711900915455,
      "grad_norm": 2.7714576721191406,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 8409
    },
    {
      "epoch": 0.7548016514090827,
      "grad_norm": 2.6197845935821533,
      "learning_rate": 0.0,
      "loss": 0.6989,
      "step": 8410
    },
    {
      "epoch": 0.7548914019027104,
      "grad_norm": 4.429442405700684,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 8411
    },
    {
      "epoch": 0.7549811523963382,
      "grad_norm": 3.529977798461914,
      "learning_rate": 0.0,
      "loss": 0.7152,
      "step": 8412
    },
    {
      "epoch": 0.7550709028899659,
      "grad_norm": 2.8385651111602783,
      "learning_rate": 0.0,
      "loss": 0.7592,
      "step": 8413
    },
    {
      "epoch": 0.7551606533835936,
      "grad_norm": 4.610324859619141,
      "learning_rate": 0.0,
      "loss": 0.6745,
      "step": 8414
    },
    {
      "epoch": 0.7552504038772213,
      "grad_norm": 3.305422782897949,
      "learning_rate": 0.0,
      "loss": 0.6876,
      "step": 8415
    },
    {
      "epoch": 0.755340154370849,
      "grad_norm": 3.252314329147339,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 8416
    },
    {
      "epoch": 0.7554299048644768,
      "grad_norm": 2.4322030544281006,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 8417
    },
    {
      "epoch": 0.7555196553581045,
      "grad_norm": 2.269611358642578,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 8418
    },
    {
      "epoch": 0.7556094058517322,
      "grad_norm": 2.7865281105041504,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 8419
    },
    {
      "epoch": 0.7556991563453599,
      "grad_norm": 2.498847723007202,
      "learning_rate": 0.0,
      "loss": 0.7614,
      "step": 8420
    },
    {
      "epoch": 0.7557889068389876,
      "grad_norm": 2.740171194076538,
      "learning_rate": 0.0,
      "loss": 0.6945,
      "step": 8421
    },
    {
      "epoch": 0.7558786573326153,
      "grad_norm": 2.7729172706604004,
      "learning_rate": 0.0,
      "loss": 0.6781,
      "step": 8422
    },
    {
      "epoch": 0.7559684078262431,
      "grad_norm": 4.993068695068359,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 8423
    },
    {
      "epoch": 0.7560581583198708,
      "grad_norm": 2.6547932624816895,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 8424
    },
    {
      "epoch": 0.7561479088134985,
      "grad_norm": 3.3591582775115967,
      "learning_rate": 0.0,
      "loss": 0.8053,
      "step": 8425
    },
    {
      "epoch": 0.7562376593071262,
      "grad_norm": 3.2725982666015625,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 8426
    },
    {
      "epoch": 0.7563274098007539,
      "grad_norm": 4.238455295562744,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 8427
    },
    {
      "epoch": 0.7564171602943817,
      "grad_norm": 3.4452362060546875,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 8428
    },
    {
      "epoch": 0.7565069107880094,
      "grad_norm": 2.8736860752105713,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 8429
    },
    {
      "epoch": 0.756596661281637,
      "grad_norm": 3.2099294662475586,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 8430
    },
    {
      "epoch": 0.7566864117752647,
      "grad_norm": 2.92771315574646,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 8431
    },
    {
      "epoch": 0.7567761622688924,
      "grad_norm": 2.523317337036133,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 8432
    },
    {
      "epoch": 0.7568659127625202,
      "grad_norm": 3.479408025741577,
      "learning_rate": 0.0,
      "loss": 0.6817,
      "step": 8433
    },
    {
      "epoch": 0.7569556632561479,
      "grad_norm": 2.883798599243164,
      "learning_rate": 0.0,
      "loss": 0.7393,
      "step": 8434
    },
    {
      "epoch": 0.7570454137497756,
      "grad_norm": 6.9450602531433105,
      "learning_rate": 0.0,
      "loss": 0.6759,
      "step": 8435
    },
    {
      "epoch": 0.7571351642434033,
      "grad_norm": 2.7092013359069824,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 8436
    },
    {
      "epoch": 0.757224914737031,
      "grad_norm": 3.044834852218628,
      "learning_rate": 0.0,
      "loss": 0.7199,
      "step": 8437
    },
    {
      "epoch": 0.7573146652306588,
      "grad_norm": 2.6570489406585693,
      "learning_rate": 0.0,
      "loss": 0.711,
      "step": 8438
    },
    {
      "epoch": 0.7574044157242865,
      "grad_norm": 3.1887171268463135,
      "learning_rate": 0.0,
      "loss": 0.7074,
      "step": 8439
    },
    {
      "epoch": 0.7574941662179142,
      "grad_norm": 3.154982089996338,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 8440
    },
    {
      "epoch": 0.7575839167115419,
      "grad_norm": 2.5965700149536133,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 8441
    },
    {
      "epoch": 0.7576736672051696,
      "grad_norm": 3.3885557651519775,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 8442
    },
    {
      "epoch": 0.7577634176987973,
      "grad_norm": 2.566079616546631,
      "learning_rate": 0.0,
      "loss": 0.6672,
      "step": 8443
    },
    {
      "epoch": 0.7578531681924251,
      "grad_norm": 2.526268243789673,
      "learning_rate": 0.0,
      "loss": 0.6638,
      "step": 8444
    },
    {
      "epoch": 0.7579429186860528,
      "grad_norm": 2.2167229652404785,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 8445
    },
    {
      "epoch": 0.7580326691796805,
      "grad_norm": 7.0052385330200195,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 8446
    },
    {
      "epoch": 0.7581224196733082,
      "grad_norm": 3.1818199157714844,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 8447
    },
    {
      "epoch": 0.7582121701669359,
      "grad_norm": 3.738246202468872,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 8448
    },
    {
      "epoch": 0.7583019206605637,
      "grad_norm": 2.7488815784454346,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 8449
    },
    {
      "epoch": 0.7583916711541914,
      "grad_norm": 3.08933162689209,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 8450
    },
    {
      "epoch": 0.7584814216478191,
      "grad_norm": 3.1737539768218994,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 8451
    },
    {
      "epoch": 0.7585711721414468,
      "grad_norm": 2.6451330184936523,
      "learning_rate": 0.0,
      "loss": 0.6901,
      "step": 8452
    },
    {
      "epoch": 0.7586609226350745,
      "grad_norm": 2.879204511642456,
      "learning_rate": 0.0,
      "loss": 0.6808,
      "step": 8453
    },
    {
      "epoch": 0.7587506731287023,
      "grad_norm": 3.2883551120758057,
      "learning_rate": 0.0,
      "loss": 0.6525,
      "step": 8454
    },
    {
      "epoch": 0.75884042362233,
      "grad_norm": 30.337392807006836,
      "learning_rate": 0.0,
      "loss": 0.6949,
      "step": 8455
    },
    {
      "epoch": 0.7589301741159576,
      "grad_norm": 3.0308663845062256,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 8456
    },
    {
      "epoch": 0.7590199246095853,
      "grad_norm": 3.9367127418518066,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 8457
    },
    {
      "epoch": 0.759109675103213,
      "grad_norm": 2.364612102508545,
      "learning_rate": 0.0,
      "loss": 0.6978,
      "step": 8458
    },
    {
      "epoch": 0.7591994255968408,
      "grad_norm": 2.5914978981018066,
      "learning_rate": 0.0,
      "loss": 0.6993,
      "step": 8459
    },
    {
      "epoch": 0.7592891760904685,
      "grad_norm": 4.380197525024414,
      "learning_rate": 0.0,
      "loss": 0.7445,
      "step": 8460
    },
    {
      "epoch": 0.7593789265840962,
      "grad_norm": 3.1703152656555176,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 8461
    },
    {
      "epoch": 0.7594686770777239,
      "grad_norm": 3.0721163749694824,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 8462
    },
    {
      "epoch": 0.7595584275713516,
      "grad_norm": 2.571702718734741,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 8463
    },
    {
      "epoch": 0.7596481780649793,
      "grad_norm": 2.970379114151001,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 8464
    },
    {
      "epoch": 0.7597379285586071,
      "grad_norm": 3.099717378616333,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 8465
    },
    {
      "epoch": 0.7598276790522348,
      "grad_norm": 2.679034471511841,
      "learning_rate": 0.0,
      "loss": 0.7034,
      "step": 8466
    },
    {
      "epoch": 0.7599174295458625,
      "grad_norm": 3.375856876373291,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 8467
    },
    {
      "epoch": 0.7600071800394902,
      "grad_norm": 4.3202338218688965,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 8468
    },
    {
      "epoch": 0.7600969305331179,
      "grad_norm": 2.621509075164795,
      "learning_rate": 0.0,
      "loss": 0.7196,
      "step": 8469
    },
    {
      "epoch": 0.7601866810267457,
      "grad_norm": 2.467311382293701,
      "learning_rate": 0.0,
      "loss": 0.6927,
      "step": 8470
    },
    {
      "epoch": 0.7602764315203734,
      "grad_norm": 5.877490520477295,
      "learning_rate": 0.0,
      "loss": 0.8035,
      "step": 8471
    },
    {
      "epoch": 0.7603661820140011,
      "grad_norm": 3.8299412727355957,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 8472
    },
    {
      "epoch": 0.7604559325076288,
      "grad_norm": 2.4806180000305176,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 8473
    },
    {
      "epoch": 0.7605456830012565,
      "grad_norm": 10.448936462402344,
      "learning_rate": 0.0,
      "loss": 0.7019,
      "step": 8474
    },
    {
      "epoch": 0.7606354334948843,
      "grad_norm": 3.814561128616333,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 8475
    },
    {
      "epoch": 0.760725183988512,
      "grad_norm": 3.3365232944488525,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 8476
    },
    {
      "epoch": 0.7608149344821397,
      "grad_norm": 17.787317276000977,
      "learning_rate": 0.0,
      "loss": 0.6787,
      "step": 8477
    },
    {
      "epoch": 0.7609046849757674,
      "grad_norm": 3.85237193107605,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 8478
    },
    {
      "epoch": 0.760994435469395,
      "grad_norm": 2.539241075515747,
      "learning_rate": 0.0,
      "loss": 0.7176,
      "step": 8479
    },
    {
      "epoch": 0.7610841859630227,
      "grad_norm": 2.854335069656372,
      "learning_rate": 0.0,
      "loss": 0.6508,
      "step": 8480
    },
    {
      "epoch": 0.7611739364566505,
      "grad_norm": 3.5440664291381836,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 8481
    },
    {
      "epoch": 0.7612636869502782,
      "grad_norm": 2.5034713745117188,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 8482
    },
    {
      "epoch": 0.7613534374439059,
      "grad_norm": 3.2164173126220703,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 8483
    },
    {
      "epoch": 0.7614431879375336,
      "grad_norm": 4.084323406219482,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 8484
    },
    {
      "epoch": 0.7615329384311613,
      "grad_norm": 2.910663604736328,
      "learning_rate": 0.0,
      "loss": 0.8106,
      "step": 8485
    },
    {
      "epoch": 0.7616226889247891,
      "grad_norm": 2.89502215385437,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 8486
    },
    {
      "epoch": 0.7617124394184168,
      "grad_norm": 3.17616868019104,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 8487
    },
    {
      "epoch": 0.7618021899120445,
      "grad_norm": 3.929844379425049,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 8488
    },
    {
      "epoch": 0.7618919404056722,
      "grad_norm": 3.1337475776672363,
      "learning_rate": 0.0,
      "loss": 0.6745,
      "step": 8489
    },
    {
      "epoch": 0.7619816908992999,
      "grad_norm": 3.793344497680664,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 8490
    },
    {
      "epoch": 0.7620714413929277,
      "grad_norm": 2.386958122253418,
      "learning_rate": 0.0,
      "loss": 0.7824,
      "step": 8491
    },
    {
      "epoch": 0.7621611918865554,
      "grad_norm": 15.378960609436035,
      "learning_rate": 0.0,
      "loss": 0.7611,
      "step": 8492
    },
    {
      "epoch": 0.7622509423801831,
      "grad_norm": 4.568644046783447,
      "learning_rate": 0.0,
      "loss": 0.6815,
      "step": 8493
    },
    {
      "epoch": 0.7623406928738108,
      "grad_norm": 3.366788625717163,
      "learning_rate": 0.0,
      "loss": 0.6829,
      "step": 8494
    },
    {
      "epoch": 0.7624304433674385,
      "grad_norm": 5.203916072845459,
      "learning_rate": 0.0,
      "loss": 0.7763,
      "step": 8495
    },
    {
      "epoch": 0.7625201938610663,
      "grad_norm": 2.868425130844116,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 8496
    },
    {
      "epoch": 0.762609944354694,
      "grad_norm": 3.3123931884765625,
      "learning_rate": 0.0,
      "loss": 0.6613,
      "step": 8497
    },
    {
      "epoch": 0.7626996948483217,
      "grad_norm": 2.7697293758392334,
      "learning_rate": 0.0,
      "loss": 0.7355,
      "step": 8498
    },
    {
      "epoch": 0.7627894453419494,
      "grad_norm": 2.9546895027160645,
      "learning_rate": 0.0,
      "loss": 0.7546,
      "step": 8499
    },
    {
      "epoch": 0.7628791958355771,
      "grad_norm": 3.2194976806640625,
      "learning_rate": 0.0,
      "loss": 0.7246,
      "step": 8500
    },
    {
      "epoch": 0.7629689463292048,
      "grad_norm": 2.5908524990081787,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 8501
    },
    {
      "epoch": 0.7630586968228326,
      "grad_norm": 3.560638666152954,
      "learning_rate": 0.0,
      "loss": 0.6948,
      "step": 8502
    },
    {
      "epoch": 0.7631484473164603,
      "grad_norm": 3.026384115219116,
      "learning_rate": 0.0,
      "loss": 0.6538,
      "step": 8503
    },
    {
      "epoch": 0.763238197810088,
      "grad_norm": 2.4552063941955566,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 8504
    },
    {
      "epoch": 0.7633279483037156,
      "grad_norm": 4.063220024108887,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 8505
    },
    {
      "epoch": 0.7634176987973433,
      "grad_norm": 3.201164960861206,
      "learning_rate": 0.0,
      "loss": 0.7926,
      "step": 8506
    },
    {
      "epoch": 0.7635074492909711,
      "grad_norm": 5.615516662597656,
      "learning_rate": 0.0,
      "loss": 0.6732,
      "step": 8507
    },
    {
      "epoch": 0.7635971997845988,
      "grad_norm": 3.199855327606201,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 8508
    },
    {
      "epoch": 0.7636869502782265,
      "grad_norm": 2.9093685150146484,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 8509
    },
    {
      "epoch": 0.7637767007718542,
      "grad_norm": 3.0170352458953857,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 8510
    },
    {
      "epoch": 0.7638664512654819,
      "grad_norm": 3.0473546981811523,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 8511
    },
    {
      "epoch": 0.7639562017591097,
      "grad_norm": 4.94525671005249,
      "learning_rate": 0.0,
      "loss": 0.719,
      "step": 8512
    },
    {
      "epoch": 0.7640459522527374,
      "grad_norm": 3.530574321746826,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 8513
    },
    {
      "epoch": 0.7641357027463651,
      "grad_norm": 2.7967729568481445,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 8514
    },
    {
      "epoch": 0.7642254532399928,
      "grad_norm": 6.9239959716796875,
      "learning_rate": 0.0,
      "loss": 0.6861,
      "step": 8515
    },
    {
      "epoch": 0.7643152037336205,
      "grad_norm": 3.3419039249420166,
      "learning_rate": 0.0,
      "loss": 0.6778,
      "step": 8516
    },
    {
      "epoch": 0.7644049542272483,
      "grad_norm": 3.4917733669281006,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 8517
    },
    {
      "epoch": 0.764494704720876,
      "grad_norm": 2.308237075805664,
      "learning_rate": 0.0,
      "loss": 0.6753,
      "step": 8518
    },
    {
      "epoch": 0.7645844552145037,
      "grad_norm": 4.329983711242676,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 8519
    },
    {
      "epoch": 0.7646742057081314,
      "grad_norm": 2.9330170154571533,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 8520
    },
    {
      "epoch": 0.7647639562017591,
      "grad_norm": 3.3285281658172607,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 8521
    },
    {
      "epoch": 0.7648537066953868,
      "grad_norm": 3.4407360553741455,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 8522
    },
    {
      "epoch": 0.7649434571890146,
      "grad_norm": 2.212886095046997,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 8523
    },
    {
      "epoch": 0.7650332076826423,
      "grad_norm": 3.090176820755005,
      "learning_rate": 0.0,
      "loss": 0.6503,
      "step": 8524
    },
    {
      "epoch": 0.76512295817627,
      "grad_norm": 2.7114994525909424,
      "learning_rate": 0.0,
      "loss": 0.6803,
      "step": 8525
    },
    {
      "epoch": 0.7652127086698977,
      "grad_norm": 2.7532858848571777,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 8526
    },
    {
      "epoch": 0.7653024591635254,
      "grad_norm": 3.397036552429199,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 8527
    },
    {
      "epoch": 0.7653922096571532,
      "grad_norm": 2.876610517501831,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 8528
    },
    {
      "epoch": 0.7654819601507808,
      "grad_norm": 3.483708620071411,
      "learning_rate": 0.0,
      "loss": 0.6811,
      "step": 8529
    },
    {
      "epoch": 0.7655717106444085,
      "grad_norm": 2.394228458404541,
      "learning_rate": 0.0,
      "loss": 0.6898,
      "step": 8530
    },
    {
      "epoch": 0.7656614611380362,
      "grad_norm": 1.8946818113327026,
      "learning_rate": 0.0,
      "loss": 0.7837,
      "step": 8531
    },
    {
      "epoch": 0.7657512116316639,
      "grad_norm": 3.452540636062622,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 8532
    },
    {
      "epoch": 0.7658409621252917,
      "grad_norm": 3.615567445755005,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 8533
    },
    {
      "epoch": 0.7659307126189194,
      "grad_norm": 3.026975393295288,
      "learning_rate": 0.0,
      "loss": 0.7397,
      "step": 8534
    },
    {
      "epoch": 0.7660204631125471,
      "grad_norm": 3.1768369674682617,
      "learning_rate": 0.0,
      "loss": 0.7844,
      "step": 8535
    },
    {
      "epoch": 0.7661102136061748,
      "grad_norm": 6.73657751083374,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 8536
    },
    {
      "epoch": 0.7661999640998025,
      "grad_norm": 4.01094388961792,
      "learning_rate": 0.0,
      "loss": 0.7859,
      "step": 8537
    },
    {
      "epoch": 0.7662897145934303,
      "grad_norm": 3.6422712802886963,
      "learning_rate": 0.0,
      "loss": 0.7572,
      "step": 8538
    },
    {
      "epoch": 0.766379465087058,
      "grad_norm": 2.2589685916900635,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 8539
    },
    {
      "epoch": 0.7664692155806857,
      "grad_norm": 2.5941991806030273,
      "learning_rate": 0.0,
      "loss": 0.6834,
      "step": 8540
    },
    {
      "epoch": 0.7665589660743134,
      "grad_norm": 2.3780407905578613,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 8541
    },
    {
      "epoch": 0.7666487165679411,
      "grad_norm": 2.65389084815979,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 8542
    },
    {
      "epoch": 0.7667384670615688,
      "grad_norm": 2.1083717346191406,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 8543
    },
    {
      "epoch": 0.7668282175551966,
      "grad_norm": 3.46514892578125,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 8544
    },
    {
      "epoch": 0.7669179680488243,
      "grad_norm": 2.804453134536743,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 8545
    },
    {
      "epoch": 0.767007718542452,
      "grad_norm": 3.308454990386963,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 8546
    },
    {
      "epoch": 0.7670974690360797,
      "grad_norm": 2.366037130355835,
      "learning_rate": 0.0,
      "loss": 0.778,
      "step": 8547
    },
    {
      "epoch": 0.7671872195297074,
      "grad_norm": 3.266077756881714,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 8548
    },
    {
      "epoch": 0.7672769700233352,
      "grad_norm": 2.6982460021972656,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 8549
    },
    {
      "epoch": 0.7673667205169629,
      "grad_norm": 5.860673427581787,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 8550
    },
    {
      "epoch": 0.7674564710105906,
      "grad_norm": 3.2984375953674316,
      "learning_rate": 0.0,
      "loss": 0.6663,
      "step": 8551
    },
    {
      "epoch": 0.7675462215042183,
      "grad_norm": 2.678062915802002,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 8552
    },
    {
      "epoch": 0.767635971997846,
      "grad_norm": 2.643416166305542,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 8553
    },
    {
      "epoch": 0.7677257224914737,
      "grad_norm": 2.3809425830841064,
      "learning_rate": 0.0,
      "loss": 0.7344,
      "step": 8554
    },
    {
      "epoch": 0.7678154729851014,
      "grad_norm": 2.6486289501190186,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 8555
    },
    {
      "epoch": 0.7679052234787291,
      "grad_norm": 3.0155625343322754,
      "learning_rate": 0.0,
      "loss": 0.6778,
      "step": 8556
    },
    {
      "epoch": 0.7679949739723568,
      "grad_norm": 2.8358535766601562,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 8557
    },
    {
      "epoch": 0.7680847244659845,
      "grad_norm": 3.759270668029785,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 8558
    },
    {
      "epoch": 0.7681744749596123,
      "grad_norm": 2.79000186920166,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 8559
    },
    {
      "epoch": 0.76826422545324,
      "grad_norm": 3.2265560626983643,
      "learning_rate": 0.0,
      "loss": 0.6783,
      "step": 8560
    },
    {
      "epoch": 0.7683539759468677,
      "grad_norm": 3.838083028793335,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 8561
    },
    {
      "epoch": 0.7684437264404954,
      "grad_norm": 2.67165207862854,
      "learning_rate": 0.0,
      "loss": 0.7491,
      "step": 8562
    },
    {
      "epoch": 0.7685334769341231,
      "grad_norm": 3.2886545658111572,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 8563
    },
    {
      "epoch": 0.7686232274277508,
      "grad_norm": 2.9990487098693848,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 8564
    },
    {
      "epoch": 0.7687129779213786,
      "grad_norm": 2.2234854698181152,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 8565
    },
    {
      "epoch": 0.7688027284150063,
      "grad_norm": 2.4204976558685303,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 8566
    },
    {
      "epoch": 0.768892478908634,
      "grad_norm": 4.462011337280273,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 8567
    },
    {
      "epoch": 0.7689822294022617,
      "grad_norm": 2.5688319206237793,
      "learning_rate": 0.0,
      "loss": 0.6447,
      "step": 8568
    },
    {
      "epoch": 0.7690719798958894,
      "grad_norm": 3.852952241897583,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 8569
    },
    {
      "epoch": 0.7691617303895172,
      "grad_norm": 3.5471267700195312,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 8570
    },
    {
      "epoch": 0.7692514808831449,
      "grad_norm": 3.9282848834991455,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 8571
    },
    {
      "epoch": 0.7693412313767726,
      "grad_norm": 2.2871146202087402,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 8572
    },
    {
      "epoch": 0.7694309818704003,
      "grad_norm": 2.552166223526001,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 8573
    },
    {
      "epoch": 0.769520732364028,
      "grad_norm": 2.7904624938964844,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 8574
    },
    {
      "epoch": 0.7696104828576558,
      "grad_norm": 2.9674270153045654,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 8575
    },
    {
      "epoch": 0.7697002333512835,
      "grad_norm": 2.871434450149536,
      "learning_rate": 0.0,
      "loss": 0.6637,
      "step": 8576
    },
    {
      "epoch": 0.7697899838449112,
      "grad_norm": 2.62839674949646,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 8577
    },
    {
      "epoch": 0.7698797343385388,
      "grad_norm": 3.61348295211792,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 8578
    },
    {
      "epoch": 0.7699694848321665,
      "grad_norm": 3.359156608581543,
      "learning_rate": 0.0,
      "loss": 0.734,
      "step": 8579
    },
    {
      "epoch": 0.7700592353257943,
      "grad_norm": 2.3430352210998535,
      "learning_rate": 0.0,
      "loss": 0.7654,
      "step": 8580
    },
    {
      "epoch": 0.770148985819422,
      "grad_norm": 2.410752296447754,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 8581
    },
    {
      "epoch": 0.7702387363130497,
      "grad_norm": 3.058877468109131,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 8582
    },
    {
      "epoch": 0.7703284868066774,
      "grad_norm": 3.253957748413086,
      "learning_rate": 0.0,
      "loss": 0.7211,
      "step": 8583
    },
    {
      "epoch": 0.7704182373003051,
      "grad_norm": 2.74841046333313,
      "learning_rate": 0.0,
      "loss": 0.6603,
      "step": 8584
    },
    {
      "epoch": 0.7705079877939328,
      "grad_norm": 3.1904451847076416,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 8585
    },
    {
      "epoch": 0.7705977382875606,
      "grad_norm": 3.291152000427246,
      "learning_rate": 0.0,
      "loss": 0.7589,
      "step": 8586
    },
    {
      "epoch": 0.7706874887811883,
      "grad_norm": 2.4954988956451416,
      "learning_rate": 0.0,
      "loss": 0.6893,
      "step": 8587
    },
    {
      "epoch": 0.770777239274816,
      "grad_norm": 8.165182113647461,
      "learning_rate": 0.0,
      "loss": 0.6618,
      "step": 8588
    },
    {
      "epoch": 0.7708669897684437,
      "grad_norm": 6.586930274963379,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 8589
    },
    {
      "epoch": 0.7709567402620714,
      "grad_norm": 4.268600940704346,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 8590
    },
    {
      "epoch": 0.7710464907556992,
      "grad_norm": 3.154841899871826,
      "learning_rate": 0.0,
      "loss": 0.7125,
      "step": 8591
    },
    {
      "epoch": 0.7711362412493269,
      "grad_norm": 3.1290860176086426,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 8592
    },
    {
      "epoch": 0.7712259917429546,
      "grad_norm": 3.3779993057250977,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 8593
    },
    {
      "epoch": 0.7713157422365823,
      "grad_norm": 3.96693754196167,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 8594
    },
    {
      "epoch": 0.77140549273021,
      "grad_norm": 2.9248762130737305,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 8595
    },
    {
      "epoch": 0.7714952432238378,
      "grad_norm": 2.70495343208313,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 8596
    },
    {
      "epoch": 0.7715849937174655,
      "grad_norm": 2.8605399131774902,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 8597
    },
    {
      "epoch": 0.7716747442110932,
      "grad_norm": 2.771503448486328,
      "learning_rate": 0.0,
      "loss": 0.7827,
      "step": 8598
    },
    {
      "epoch": 0.7717644947047209,
      "grad_norm": 2.7367889881134033,
      "learning_rate": 0.0,
      "loss": 0.6615,
      "step": 8599
    },
    {
      "epoch": 0.7718542451983486,
      "grad_norm": 3.196565628051758,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 8600
    },
    {
      "epoch": 0.7719439956919764,
      "grad_norm": 5.901310443878174,
      "learning_rate": 0.0,
      "loss": 0.6876,
      "step": 8601
    },
    {
      "epoch": 0.772033746185604,
      "grad_norm": 3.167280673980713,
      "learning_rate": 0.0,
      "loss": 0.6718,
      "step": 8602
    },
    {
      "epoch": 0.7721234966792317,
      "grad_norm": 2.7116286754608154,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 8603
    },
    {
      "epoch": 0.7722132471728594,
      "grad_norm": 2.3029305934906006,
      "learning_rate": 0.0,
      "loss": 0.6532,
      "step": 8604
    },
    {
      "epoch": 0.7723029976664871,
      "grad_norm": 2.419724225997925,
      "learning_rate": 0.0,
      "loss": 0.678,
      "step": 8605
    },
    {
      "epoch": 0.7723927481601148,
      "grad_norm": 2.528183937072754,
      "learning_rate": 0.0,
      "loss": 0.6787,
      "step": 8606
    },
    {
      "epoch": 0.7724824986537426,
      "grad_norm": 4.016151428222656,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 8607
    },
    {
      "epoch": 0.7725722491473703,
      "grad_norm": 3.4362144470214844,
      "learning_rate": 0.0,
      "loss": 0.7601,
      "step": 8608
    },
    {
      "epoch": 0.772661999640998,
      "grad_norm": 2.830886125564575,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 8609
    },
    {
      "epoch": 0.7727517501346257,
      "grad_norm": 1.9458352327346802,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 8610
    },
    {
      "epoch": 0.7728415006282534,
      "grad_norm": 2.4839446544647217,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 8611
    },
    {
      "epoch": 0.7729312511218812,
      "grad_norm": 2.2223942279815674,
      "learning_rate": 0.0,
      "loss": 0.7823,
      "step": 8612
    },
    {
      "epoch": 0.7730210016155089,
      "grad_norm": 2.5880160331726074,
      "learning_rate": 0.0,
      "loss": 0.6856,
      "step": 8613
    },
    {
      "epoch": 0.7731107521091366,
      "grad_norm": 2.6692891120910645,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 8614
    },
    {
      "epoch": 0.7732005026027643,
      "grad_norm": 2.4515931606292725,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 8615
    },
    {
      "epoch": 0.773290253096392,
      "grad_norm": 2.0496931076049805,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 8616
    },
    {
      "epoch": 0.7733800035900198,
      "grad_norm": 3.2805283069610596,
      "learning_rate": 0.0,
      "loss": 0.6827,
      "step": 8617
    },
    {
      "epoch": 0.7734697540836475,
      "grad_norm": 2.6324679851531982,
      "learning_rate": 0.0,
      "loss": 0.6524,
      "step": 8618
    },
    {
      "epoch": 0.7735595045772752,
      "grad_norm": 2.1721198558807373,
      "learning_rate": 0.0,
      "loss": 0.7301,
      "step": 8619
    },
    {
      "epoch": 0.7736492550709029,
      "grad_norm": 2.962388038635254,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 8620
    },
    {
      "epoch": 0.7737390055645306,
      "grad_norm": 2.51765513420105,
      "learning_rate": 0.0,
      "loss": 0.7704,
      "step": 8621
    },
    {
      "epoch": 0.7738287560581584,
      "grad_norm": 2.5906407833099365,
      "learning_rate": 0.0,
      "loss": 0.6748,
      "step": 8622
    },
    {
      "epoch": 0.7739185065517861,
      "grad_norm": 7.96431827545166,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 8623
    },
    {
      "epoch": 0.7740082570454138,
      "grad_norm": 3.4938082695007324,
      "learning_rate": 0.0,
      "loss": 0.7389,
      "step": 8624
    },
    {
      "epoch": 0.7740980075390415,
      "grad_norm": 2.745987892150879,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 8625
    },
    {
      "epoch": 0.7741877580326691,
      "grad_norm": 27.461429595947266,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 8626
    },
    {
      "epoch": 0.7742775085262968,
      "grad_norm": 4.548468589782715,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 8627
    },
    {
      "epoch": 0.7743672590199246,
      "grad_norm": 4.375185966491699,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 8628
    },
    {
      "epoch": 0.7744570095135523,
      "grad_norm": 2.8577311038970947,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 8629
    },
    {
      "epoch": 0.77454676000718,
      "grad_norm": 3.1603400707244873,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 8630
    },
    {
      "epoch": 0.7746365105008077,
      "grad_norm": 5.372519493103027,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 8631
    },
    {
      "epoch": 0.7747262609944354,
      "grad_norm": 3.401320457458496,
      "learning_rate": 0.0,
      "loss": 0.7579,
      "step": 8632
    },
    {
      "epoch": 0.7748160114880632,
      "grad_norm": 2.9440784454345703,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 8633
    },
    {
      "epoch": 0.7749057619816909,
      "grad_norm": 2.5039875507354736,
      "learning_rate": 0.0,
      "loss": 0.7602,
      "step": 8634
    },
    {
      "epoch": 0.7749955124753186,
      "grad_norm": 3.573277235031128,
      "learning_rate": 0.0,
      "loss": 0.7141,
      "step": 8635
    },
    {
      "epoch": 0.7750852629689463,
      "grad_norm": 3.453497886657715,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 8636
    },
    {
      "epoch": 0.775175013462574,
      "grad_norm": 3.3072657585144043,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 8637
    },
    {
      "epoch": 0.7752647639562018,
      "grad_norm": 2.6268112659454346,
      "learning_rate": 0.0,
      "loss": 0.6838,
      "step": 8638
    },
    {
      "epoch": 0.7753545144498295,
      "grad_norm": 2.5318727493286133,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 8639
    },
    {
      "epoch": 0.7754442649434572,
      "grad_norm": 2.9127633571624756,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 8640
    },
    {
      "epoch": 0.7755340154370849,
      "grad_norm": 3.4991097450256348,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 8641
    },
    {
      "epoch": 0.7756237659307126,
      "grad_norm": 3.292235851287842,
      "learning_rate": 0.0,
      "loss": 0.7306,
      "step": 8642
    },
    {
      "epoch": 0.7757135164243404,
      "grad_norm": 2.6151530742645264,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 8643
    },
    {
      "epoch": 0.7758032669179681,
      "grad_norm": 3.1015398502349854,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 8644
    },
    {
      "epoch": 0.7758930174115958,
      "grad_norm": 2.6960978507995605,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 8645
    },
    {
      "epoch": 0.7759827679052235,
      "grad_norm": 2.845569133758545,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 8646
    },
    {
      "epoch": 0.7760725183988512,
      "grad_norm": 2.7494630813598633,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 8647
    },
    {
      "epoch": 0.7761622688924789,
      "grad_norm": 2.609717607498169,
      "learning_rate": 0.0,
      "loss": 0.6599,
      "step": 8648
    },
    {
      "epoch": 0.7762520193861067,
      "grad_norm": 3.9841833114624023,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 8649
    },
    {
      "epoch": 0.7763417698797344,
      "grad_norm": 2.896108388900757,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 8650
    },
    {
      "epoch": 0.776431520373362,
      "grad_norm": 2.6534125804901123,
      "learning_rate": 0.0,
      "loss": 0.6804,
      "step": 8651
    },
    {
      "epoch": 0.7765212708669897,
      "grad_norm": 2.9025332927703857,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 8652
    },
    {
      "epoch": 0.7766110213606174,
      "grad_norm": 3.23038649559021,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 8653
    },
    {
      "epoch": 0.7767007718542452,
      "grad_norm": 2.5714595317840576,
      "learning_rate": 0.0,
      "loss": 0.6918,
      "step": 8654
    },
    {
      "epoch": 0.7767905223478729,
      "grad_norm": 3.4325850009918213,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 8655
    },
    {
      "epoch": 0.7768802728415006,
      "grad_norm": 3.6462717056274414,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 8656
    },
    {
      "epoch": 0.7769700233351283,
      "grad_norm": 3.1480090618133545,
      "learning_rate": 0.0,
      "loss": 0.6391,
      "step": 8657
    },
    {
      "epoch": 0.777059773828756,
      "grad_norm": 2.9030778408050537,
      "learning_rate": 0.0,
      "loss": 0.6827,
      "step": 8658
    },
    {
      "epoch": 0.7771495243223838,
      "grad_norm": 3.62548565864563,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 8659
    },
    {
      "epoch": 0.7772392748160115,
      "grad_norm": 3.0431578159332275,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 8660
    },
    {
      "epoch": 0.7773290253096392,
      "grad_norm": 2.8604044914245605,
      "learning_rate": 0.0,
      "loss": 0.7373,
      "step": 8661
    },
    {
      "epoch": 0.7774187758032669,
      "grad_norm": 2.5471129417419434,
      "learning_rate": 0.0,
      "loss": 0.7603,
      "step": 8662
    },
    {
      "epoch": 0.7775085262968946,
      "grad_norm": 3.63875412940979,
      "learning_rate": 0.0,
      "loss": 0.7435,
      "step": 8663
    },
    {
      "epoch": 0.7775982767905224,
      "grad_norm": 2.7681357860565186,
      "learning_rate": 0.0,
      "loss": 0.7499,
      "step": 8664
    },
    {
      "epoch": 0.7776880272841501,
      "grad_norm": 2.810177803039551,
      "learning_rate": 0.0,
      "loss": 0.6259,
      "step": 8665
    },
    {
      "epoch": 0.7777777777777778,
      "grad_norm": 3.4640910625457764,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 8666
    },
    {
      "epoch": 0.7778675282714055,
      "grad_norm": 3.1637401580810547,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 8667
    },
    {
      "epoch": 0.7779572787650332,
      "grad_norm": 2.7394561767578125,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 8668
    },
    {
      "epoch": 0.7780470292586609,
      "grad_norm": 3.091217041015625,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 8669
    },
    {
      "epoch": 0.7781367797522887,
      "grad_norm": 3.223771810531616,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 8670
    },
    {
      "epoch": 0.7782265302459164,
      "grad_norm": 3.067845582962036,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 8671
    },
    {
      "epoch": 0.7783162807395441,
      "grad_norm": 2.6595547199249268,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 8672
    },
    {
      "epoch": 0.7784060312331718,
      "grad_norm": 2.8917582035064697,
      "learning_rate": 0.0,
      "loss": 0.6627,
      "step": 8673
    },
    {
      "epoch": 0.7784957817267995,
      "grad_norm": 2.057813882827759,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 8674
    },
    {
      "epoch": 0.7785855322204273,
      "grad_norm": 2.737352132797241,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 8675
    },
    {
      "epoch": 0.778675282714055,
      "grad_norm": 3.4147963523864746,
      "learning_rate": 0.0,
      "loss": 0.7369,
      "step": 8676
    },
    {
      "epoch": 0.7787650332076826,
      "grad_norm": 2.4052224159240723,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 8677
    },
    {
      "epoch": 0.7788547837013103,
      "grad_norm": 3.2958710193634033,
      "learning_rate": 0.0,
      "loss": 0.6579,
      "step": 8678
    },
    {
      "epoch": 0.778944534194938,
      "grad_norm": 3.0197715759277344,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 8679
    },
    {
      "epoch": 0.7790342846885658,
      "grad_norm": 4.256289005279541,
      "learning_rate": 0.0,
      "loss": 0.7968,
      "step": 8680
    },
    {
      "epoch": 0.7791240351821935,
      "grad_norm": 2.8114936351776123,
      "learning_rate": 0.0,
      "loss": 0.719,
      "step": 8681
    },
    {
      "epoch": 0.7792137856758212,
      "grad_norm": 2.7633469104766846,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 8682
    },
    {
      "epoch": 0.7793035361694489,
      "grad_norm": 3.545943260192871,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 8683
    },
    {
      "epoch": 0.7793932866630766,
      "grad_norm": 3.3474109172821045,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 8684
    },
    {
      "epoch": 0.7794830371567044,
      "grad_norm": 4.783896446228027,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 8685
    },
    {
      "epoch": 0.7795727876503321,
      "grad_norm": 2.6675617694854736,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 8686
    },
    {
      "epoch": 0.7796625381439598,
      "grad_norm": 2.414456605911255,
      "learning_rate": 0.0,
      "loss": 0.7737,
      "step": 8687
    },
    {
      "epoch": 0.7797522886375875,
      "grad_norm": 2.45784592628479,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 8688
    },
    {
      "epoch": 0.7798420391312152,
      "grad_norm": 2.6619713306427,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 8689
    },
    {
      "epoch": 0.7799317896248429,
      "grad_norm": 6.002089977264404,
      "learning_rate": 0.0,
      "loss": 0.6924,
      "step": 8690
    },
    {
      "epoch": 0.7800215401184707,
      "grad_norm": 2.9122314453125,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 8691
    },
    {
      "epoch": 0.7801112906120984,
      "grad_norm": 2.735201120376587,
      "learning_rate": 0.0,
      "loss": 0.6793,
      "step": 8692
    },
    {
      "epoch": 0.7802010411057261,
      "grad_norm": 2.7943453788757324,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 8693
    },
    {
      "epoch": 0.7802907915993538,
      "grad_norm": 3.248548984527588,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 8694
    },
    {
      "epoch": 0.7803805420929815,
      "grad_norm": 2.7276439666748047,
      "learning_rate": 0.0,
      "loss": 0.6784,
      "step": 8695
    },
    {
      "epoch": 0.7804702925866093,
      "grad_norm": 5.796113967895508,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 8696
    },
    {
      "epoch": 0.780560043080237,
      "grad_norm": 3.369791269302368,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 8697
    },
    {
      "epoch": 0.7806497935738647,
      "grad_norm": 2.380099058151245,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 8698
    },
    {
      "epoch": 0.7807395440674924,
      "grad_norm": 2.9399807453155518,
      "learning_rate": 0.0,
      "loss": 0.6814,
      "step": 8699
    },
    {
      "epoch": 0.78082929456112,
      "grad_norm": 2.3009729385375977,
      "learning_rate": 0.0,
      "loss": 0.704,
      "step": 8700
    },
    {
      "epoch": 0.7809190450547479,
      "grad_norm": 2.50018572807312,
      "learning_rate": 0.0,
      "loss": 0.682,
      "step": 8701
    },
    {
      "epoch": 0.7810087955483755,
      "grad_norm": 2.7179837226867676,
      "learning_rate": 0.0,
      "loss": 0.6758,
      "step": 8702
    },
    {
      "epoch": 0.7810985460420032,
      "grad_norm": 2.095242977142334,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 8703
    },
    {
      "epoch": 0.7811882965356309,
      "grad_norm": 2.407302141189575,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 8704
    },
    {
      "epoch": 0.7812780470292586,
      "grad_norm": 4.6706862449646,
      "learning_rate": 0.0,
      "loss": 0.761,
      "step": 8705
    },
    {
      "epoch": 0.7813677975228863,
      "grad_norm": 2.363656759262085,
      "learning_rate": 0.0,
      "loss": 0.6824,
      "step": 8706
    },
    {
      "epoch": 0.7814575480165141,
      "grad_norm": 3.186635971069336,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 8707
    },
    {
      "epoch": 0.7815472985101418,
      "grad_norm": 2.789442777633667,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 8708
    },
    {
      "epoch": 0.7816370490037695,
      "grad_norm": 3.529365062713623,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 8709
    },
    {
      "epoch": 0.7817267994973972,
      "grad_norm": 3.0595903396606445,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 8710
    },
    {
      "epoch": 0.7818165499910249,
      "grad_norm": 2.4418997764587402,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 8711
    },
    {
      "epoch": 0.7819063004846527,
      "grad_norm": 2.5707578659057617,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 8712
    },
    {
      "epoch": 0.7819960509782804,
      "grad_norm": 2.8352015018463135,
      "learning_rate": 0.0,
      "loss": 0.8068,
      "step": 8713
    },
    {
      "epoch": 0.7820858014719081,
      "grad_norm": 2.732055902481079,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 8714
    },
    {
      "epoch": 0.7821755519655358,
      "grad_norm": 3.102976083755493,
      "learning_rate": 0.0,
      "loss": 0.6843,
      "step": 8715
    },
    {
      "epoch": 0.7822653024591635,
      "grad_norm": 3.0020203590393066,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 8716
    },
    {
      "epoch": 0.7823550529527913,
      "grad_norm": 2.5041024684906006,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 8717
    },
    {
      "epoch": 0.782444803446419,
      "grad_norm": 2.9652650356292725,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 8718
    },
    {
      "epoch": 0.7825345539400467,
      "grad_norm": 2.913214683532715,
      "learning_rate": 0.0,
      "loss": 0.7124,
      "step": 8719
    },
    {
      "epoch": 0.7826243044336744,
      "grad_norm": 4.212339878082275,
      "learning_rate": 0.0,
      "loss": 0.7504,
      "step": 8720
    },
    {
      "epoch": 0.7827140549273021,
      "grad_norm": 3.026837110519409,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 8721
    },
    {
      "epoch": 0.7828038054209299,
      "grad_norm": 5.449391841888428,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 8722
    },
    {
      "epoch": 0.7828935559145576,
      "grad_norm": 2.993471622467041,
      "learning_rate": 0.0,
      "loss": 0.7591,
      "step": 8723
    },
    {
      "epoch": 0.7829833064081853,
      "grad_norm": 2.7496230602264404,
      "learning_rate": 0.0,
      "loss": 0.6758,
      "step": 8724
    },
    {
      "epoch": 0.783073056901813,
      "grad_norm": 2.8172802925109863,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 8725
    },
    {
      "epoch": 0.7831628073954406,
      "grad_norm": 14.916919708251953,
      "learning_rate": 0.0,
      "loss": 0.6792,
      "step": 8726
    },
    {
      "epoch": 0.7832525578890683,
      "grad_norm": 3.9528918266296387,
      "learning_rate": 0.0,
      "loss": 0.7204,
      "step": 8727
    },
    {
      "epoch": 0.7833423083826961,
      "grad_norm": 2.9975779056549072,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 8728
    },
    {
      "epoch": 0.7834320588763238,
      "grad_norm": 2.515538454055786,
      "learning_rate": 0.0,
      "loss": 0.7426,
      "step": 8729
    },
    {
      "epoch": 0.7835218093699515,
      "grad_norm": 2.539496660232544,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 8730
    },
    {
      "epoch": 0.7836115598635792,
      "grad_norm": 3.6292641162872314,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 8731
    },
    {
      "epoch": 0.7837013103572069,
      "grad_norm": 3.0712571144104004,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 8732
    },
    {
      "epoch": 0.7837910608508347,
      "grad_norm": 1.9848616123199463,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 8733
    },
    {
      "epoch": 0.7838808113444624,
      "grad_norm": 2.5933215618133545,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 8734
    },
    {
      "epoch": 0.7839705618380901,
      "grad_norm": 2.984388828277588,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 8735
    },
    {
      "epoch": 0.7840603123317178,
      "grad_norm": 3.5487220287323,
      "learning_rate": 0.0,
      "loss": 0.7428,
      "step": 8736
    },
    {
      "epoch": 0.7841500628253455,
      "grad_norm": 2.3507158756256104,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 8737
    },
    {
      "epoch": 0.7842398133189733,
      "grad_norm": 2.520083427429199,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 8738
    },
    {
      "epoch": 0.784329563812601,
      "grad_norm": 3.175095558166504,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 8739
    },
    {
      "epoch": 0.7844193143062287,
      "grad_norm": 2.733407497406006,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 8740
    },
    {
      "epoch": 0.7845090647998564,
      "grad_norm": 2.806501865386963,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 8741
    },
    {
      "epoch": 0.7845988152934841,
      "grad_norm": 3.4103012084960938,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 8742
    },
    {
      "epoch": 0.7846885657871119,
      "grad_norm": 3.7632029056549072,
      "learning_rate": 0.0,
      "loss": 0.6735,
      "step": 8743
    },
    {
      "epoch": 0.7847783162807396,
      "grad_norm": 3.2799007892608643,
      "learning_rate": 0.0,
      "loss": 0.7924,
      "step": 8744
    },
    {
      "epoch": 0.7848680667743673,
      "grad_norm": 2.5924794673919678,
      "learning_rate": 0.0,
      "loss": 0.6797,
      "step": 8745
    },
    {
      "epoch": 0.784957817267995,
      "grad_norm": 2.1715126037597656,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 8746
    },
    {
      "epoch": 0.7850475677616227,
      "grad_norm": 3.268730878829956,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 8747
    },
    {
      "epoch": 0.7851373182552503,
      "grad_norm": 2.4304351806640625,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 8748
    },
    {
      "epoch": 0.7852270687488782,
      "grad_norm": 2.0446107387542725,
      "learning_rate": 0.0,
      "loss": 0.6855,
      "step": 8749
    },
    {
      "epoch": 0.7853168192425058,
      "grad_norm": 3.719456195831299,
      "learning_rate": 0.0,
      "loss": 0.6801,
      "step": 8750
    },
    {
      "epoch": 0.7854065697361335,
      "grad_norm": 2.4208290576934814,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 8751
    },
    {
      "epoch": 0.7854963202297612,
      "grad_norm": 2.9825758934020996,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 8752
    },
    {
      "epoch": 0.7855860707233889,
      "grad_norm": 2.626086950302124,
      "learning_rate": 0.0,
      "loss": 0.7233,
      "step": 8753
    },
    {
      "epoch": 0.7856758212170167,
      "grad_norm": 2.887136697769165,
      "learning_rate": 0.0,
      "loss": 0.6876,
      "step": 8754
    },
    {
      "epoch": 0.7857655717106444,
      "grad_norm": 2.9146289825439453,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 8755
    },
    {
      "epoch": 0.7858553222042721,
      "grad_norm": 2.4306037425994873,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 8756
    },
    {
      "epoch": 0.7859450726978998,
      "grad_norm": 2.6368777751922607,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 8757
    },
    {
      "epoch": 0.7860348231915275,
      "grad_norm": 2.3856241703033447,
      "learning_rate": 0.0,
      "loss": 0.7152,
      "step": 8758
    },
    {
      "epoch": 0.7861245736851553,
      "grad_norm": 3.0686655044555664,
      "learning_rate": 0.0,
      "loss": 0.6502,
      "step": 8759
    },
    {
      "epoch": 0.786214324178783,
      "grad_norm": 2.5627830028533936,
      "learning_rate": 0.0,
      "loss": 0.7417,
      "step": 8760
    },
    {
      "epoch": 0.7863040746724107,
      "grad_norm": 2.3380661010742188,
      "learning_rate": 0.0,
      "loss": 0.7152,
      "step": 8761
    },
    {
      "epoch": 0.7863938251660384,
      "grad_norm": 3.8991048336029053,
      "learning_rate": 0.0,
      "loss": 0.6721,
      "step": 8762
    },
    {
      "epoch": 0.7864835756596661,
      "grad_norm": 2.8035104274749756,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 8763
    },
    {
      "epoch": 0.7865733261532939,
      "grad_norm": 2.6370348930358887,
      "learning_rate": 0.0,
      "loss": 0.7443,
      "step": 8764
    },
    {
      "epoch": 0.7866630766469216,
      "grad_norm": 2.389239549636841,
      "learning_rate": 0.0,
      "loss": 0.7587,
      "step": 8765
    },
    {
      "epoch": 0.7867528271405493,
      "grad_norm": 2.7156567573547363,
      "learning_rate": 0.0,
      "loss": 0.7829,
      "step": 8766
    },
    {
      "epoch": 0.786842577634177,
      "grad_norm": 4.486902713775635,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 8767
    },
    {
      "epoch": 0.7869323281278047,
      "grad_norm": 3.078594446182251,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 8768
    },
    {
      "epoch": 0.7870220786214324,
      "grad_norm": 2.7245421409606934,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 8769
    },
    {
      "epoch": 0.7871118291150602,
      "grad_norm": 2.9302239418029785,
      "learning_rate": 0.0,
      "loss": 0.6834,
      "step": 8770
    },
    {
      "epoch": 0.7872015796086879,
      "grad_norm": 4.487969398498535,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 8771
    },
    {
      "epoch": 0.7872913301023156,
      "grad_norm": 3.2768595218658447,
      "learning_rate": 0.0,
      "loss": 0.7834,
      "step": 8772
    },
    {
      "epoch": 0.7873810805959433,
      "grad_norm": 2.5229697227478027,
      "learning_rate": 0.0,
      "loss": 0.6809,
      "step": 8773
    },
    {
      "epoch": 0.7874708310895709,
      "grad_norm": 4.126455783843994,
      "learning_rate": 0.0,
      "loss": 0.7355,
      "step": 8774
    },
    {
      "epoch": 0.7875605815831987,
      "grad_norm": 3.3916714191436768,
      "learning_rate": 0.0,
      "loss": 0.6519,
      "step": 8775
    },
    {
      "epoch": 0.7876503320768264,
      "grad_norm": 3.5899884700775146,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 8776
    },
    {
      "epoch": 0.7877400825704541,
      "grad_norm": 4.025026798248291,
      "learning_rate": 0.0,
      "loss": 0.6948,
      "step": 8777
    },
    {
      "epoch": 0.7878298330640818,
      "grad_norm": 27.863765716552734,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 8778
    },
    {
      "epoch": 0.7879195835577095,
      "grad_norm": 4.910008907318115,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 8779
    },
    {
      "epoch": 0.7880093340513373,
      "grad_norm": 2.5662031173706055,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 8780
    },
    {
      "epoch": 0.788099084544965,
      "grad_norm": 2.480309009552002,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 8781
    },
    {
      "epoch": 0.7881888350385927,
      "grad_norm": 3.143725872039795,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 8782
    },
    {
      "epoch": 0.7882785855322204,
      "grad_norm": 3.0512218475341797,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 8783
    },
    {
      "epoch": 0.7883683360258481,
      "grad_norm": 3.0525918006896973,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 8784
    },
    {
      "epoch": 0.7884580865194759,
      "grad_norm": 8.647151947021484,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 8785
    },
    {
      "epoch": 0.7885478370131036,
      "grad_norm": 3.0876853466033936,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 8786
    },
    {
      "epoch": 0.7886375875067313,
      "grad_norm": 2.331327438354492,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 8787
    },
    {
      "epoch": 0.788727338000359,
      "grad_norm": 2.889578104019165,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 8788
    },
    {
      "epoch": 0.7888170884939867,
      "grad_norm": 2.315678358078003,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 8789
    },
    {
      "epoch": 0.7889068389876144,
      "grad_norm": 2.7114317417144775,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 8790
    },
    {
      "epoch": 0.7889965894812422,
      "grad_norm": 2.9652342796325684,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 8791
    },
    {
      "epoch": 0.7890863399748699,
      "grad_norm": 2.765986442565918,
      "learning_rate": 0.0,
      "loss": 0.7694,
      "step": 8792
    },
    {
      "epoch": 0.7891760904684976,
      "grad_norm": 2.267462730407715,
      "learning_rate": 0.0,
      "loss": 0.7433,
      "step": 8793
    },
    {
      "epoch": 0.7892658409621253,
      "grad_norm": 1.8684715032577515,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 8794
    },
    {
      "epoch": 0.789355591455753,
      "grad_norm": 3.7238805294036865,
      "learning_rate": 0.0,
      "loss": 0.6913,
      "step": 8795
    },
    {
      "epoch": 0.7894453419493808,
      "grad_norm": 8.92642593383789,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 8796
    },
    {
      "epoch": 0.7895350924430085,
      "grad_norm": 2.712388753890991,
      "learning_rate": 0.0,
      "loss": 0.6402,
      "step": 8797
    },
    {
      "epoch": 0.7896248429366362,
      "grad_norm": 2.2461557388305664,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 8798
    },
    {
      "epoch": 0.7897145934302638,
      "grad_norm": 2.675537109375,
      "learning_rate": 0.0,
      "loss": 0.6619,
      "step": 8799
    },
    {
      "epoch": 0.7898043439238915,
      "grad_norm": 2.229417324066162,
      "learning_rate": 0.0,
      "loss": 0.7001,
      "step": 8800
    },
    {
      "epoch": 0.7898940944175193,
      "grad_norm": 2.850837469100952,
      "learning_rate": 0.0,
      "loss": 0.7674,
      "step": 8801
    },
    {
      "epoch": 0.789983844911147,
      "grad_norm": 3.329800605773926,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 8802
    },
    {
      "epoch": 0.7900735954047747,
      "grad_norm": 4.684251308441162,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 8803
    },
    {
      "epoch": 0.7901633458984024,
      "grad_norm": 2.8805885314941406,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 8804
    },
    {
      "epoch": 0.7902530963920301,
      "grad_norm": 6.036820888519287,
      "learning_rate": 0.0,
      "loss": 0.6585,
      "step": 8805
    },
    {
      "epoch": 0.7903428468856579,
      "grad_norm": 2.5340700149536133,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 8806
    },
    {
      "epoch": 0.7904325973792856,
      "grad_norm": 2.479938507080078,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 8807
    },
    {
      "epoch": 0.7905223478729133,
      "grad_norm": 7.461958885192871,
      "learning_rate": 0.0,
      "loss": 0.6534,
      "step": 8808
    },
    {
      "epoch": 0.790612098366541,
      "grad_norm": 2.9779133796691895,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 8809
    },
    {
      "epoch": 0.7907018488601687,
      "grad_norm": 2.7058420181274414,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 8810
    },
    {
      "epoch": 0.7907915993537964,
      "grad_norm": 2.6747474670410156,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 8811
    },
    {
      "epoch": 0.7908813498474242,
      "grad_norm": 2.593700647354126,
      "learning_rate": 0.0,
      "loss": 0.643,
      "step": 8812
    },
    {
      "epoch": 0.7909711003410519,
      "grad_norm": 4.712214469909668,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 8813
    },
    {
      "epoch": 0.7910608508346796,
      "grad_norm": 3.843194007873535,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 8814
    },
    {
      "epoch": 0.7911506013283073,
      "grad_norm": 2.8924198150634766,
      "learning_rate": 0.0,
      "loss": 0.6687,
      "step": 8815
    },
    {
      "epoch": 0.791240351821935,
      "grad_norm": 2.857971429824829,
      "learning_rate": 0.0,
      "loss": 0.6732,
      "step": 8816
    },
    {
      "epoch": 0.7913301023155628,
      "grad_norm": 2.371211051940918,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 8817
    },
    {
      "epoch": 0.7914198528091905,
      "grad_norm": 2.757270574569702,
      "learning_rate": 0.0,
      "loss": 0.7495,
      "step": 8818
    },
    {
      "epoch": 0.7915096033028182,
      "grad_norm": 2.467580556869507,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 8819
    },
    {
      "epoch": 0.7915993537964459,
      "grad_norm": 36.2558479309082,
      "learning_rate": 0.0,
      "loss": 0.7067,
      "step": 8820
    },
    {
      "epoch": 0.7916891042900736,
      "grad_norm": 5.529475688934326,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 8821
    },
    {
      "epoch": 0.7917788547837014,
      "grad_norm": 2.8402810096740723,
      "learning_rate": 0.0,
      "loss": 0.6078,
      "step": 8822
    },
    {
      "epoch": 0.791868605277329,
      "grad_norm": 3.820413112640381,
      "learning_rate": 0.0,
      "loss": 0.7557,
      "step": 8823
    },
    {
      "epoch": 0.7919583557709567,
      "grad_norm": 10.705148696899414,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 8824
    },
    {
      "epoch": 0.7920481062645844,
      "grad_norm": 2.8438141345977783,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 8825
    },
    {
      "epoch": 0.7921378567582121,
      "grad_norm": 3.4900848865509033,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 8826
    },
    {
      "epoch": 0.7922276072518399,
      "grad_norm": 1.9574650526046753,
      "learning_rate": 0.0,
      "loss": 0.7186,
      "step": 8827
    },
    {
      "epoch": 0.7923173577454676,
      "grad_norm": 2.5400137901306152,
      "learning_rate": 0.0,
      "loss": 0.7636,
      "step": 8828
    },
    {
      "epoch": 0.7924071082390953,
      "grad_norm": 2.629647970199585,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 8829
    },
    {
      "epoch": 0.792496858732723,
      "grad_norm": 2.667473077774048,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 8830
    },
    {
      "epoch": 0.7925866092263507,
      "grad_norm": 2.5362462997436523,
      "learning_rate": 0.0,
      "loss": 0.6927,
      "step": 8831
    },
    {
      "epoch": 0.7926763597199784,
      "grad_norm": 2.971738815307617,
      "learning_rate": 0.0,
      "loss": 0.8396,
      "step": 8832
    },
    {
      "epoch": 0.7927661102136062,
      "grad_norm": 2.885418653488159,
      "learning_rate": 0.0,
      "loss": 0.6807,
      "step": 8833
    },
    {
      "epoch": 0.7928558607072339,
      "grad_norm": 4.194342613220215,
      "learning_rate": 0.0,
      "loss": 0.7503,
      "step": 8834
    },
    {
      "epoch": 0.7929456112008616,
      "grad_norm": 34.933570861816406,
      "learning_rate": 0.0,
      "loss": 0.7069,
      "step": 8835
    },
    {
      "epoch": 0.7930353616944893,
      "grad_norm": 4.878201961517334,
      "learning_rate": 0.0,
      "loss": 0.6827,
      "step": 8836
    },
    {
      "epoch": 0.793125112188117,
      "grad_norm": 2.989901542663574,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 8837
    },
    {
      "epoch": 0.7932148626817448,
      "grad_norm": 3.0559656620025635,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 8838
    },
    {
      "epoch": 0.7933046131753725,
      "grad_norm": 3.6231095790863037,
      "learning_rate": 0.0,
      "loss": 0.6799,
      "step": 8839
    },
    {
      "epoch": 0.7933943636690002,
      "grad_norm": 5.754875183105469,
      "learning_rate": 0.0,
      "loss": 0.6914,
      "step": 8840
    },
    {
      "epoch": 0.7934841141626279,
      "grad_norm": 3.4794130325317383,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 8841
    },
    {
      "epoch": 0.7935738646562556,
      "grad_norm": 2.345273017883301,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 8842
    },
    {
      "epoch": 0.7936636151498834,
      "grad_norm": 2.8564417362213135,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 8843
    },
    {
      "epoch": 0.7937533656435111,
      "grad_norm": 2.813136339187622,
      "learning_rate": 0.0,
      "loss": 0.7203,
      "step": 8844
    },
    {
      "epoch": 0.7938431161371388,
      "grad_norm": 2.6965532302856445,
      "learning_rate": 0.0,
      "loss": 0.6868,
      "step": 8845
    },
    {
      "epoch": 0.7939328666307665,
      "grad_norm": 2.498096466064453,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 8846
    },
    {
      "epoch": 0.7940226171243941,
      "grad_norm": 2.7335383892059326,
      "learning_rate": 0.0,
      "loss": 0.6583,
      "step": 8847
    },
    {
      "epoch": 0.794112367618022,
      "grad_norm": 3.068840742111206,
      "learning_rate": 0.0,
      "loss": 0.7705,
      "step": 8848
    },
    {
      "epoch": 0.7942021181116496,
      "grad_norm": 2.7275078296661377,
      "learning_rate": 0.0,
      "loss": 0.7274,
      "step": 8849
    },
    {
      "epoch": 0.7942918686052773,
      "grad_norm": 3.6873037815093994,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 8850
    },
    {
      "epoch": 0.794381619098905,
      "grad_norm": 2.0750961303710938,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 8851
    },
    {
      "epoch": 0.7944713695925327,
      "grad_norm": 2.5540411472320557,
      "learning_rate": 0.0,
      "loss": 0.8054,
      "step": 8852
    },
    {
      "epoch": 0.7945611200861604,
      "grad_norm": 2.7181313037872314,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 8853
    },
    {
      "epoch": 0.7946508705797882,
      "grad_norm": 2.6028025150299072,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 8854
    },
    {
      "epoch": 0.7947406210734159,
      "grad_norm": 3.082031488418579,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 8855
    },
    {
      "epoch": 0.7948303715670436,
      "grad_norm": 2.9807348251342773,
      "learning_rate": 0.0,
      "loss": 0.712,
      "step": 8856
    },
    {
      "epoch": 0.7949201220606713,
      "grad_norm": 3.133643388748169,
      "learning_rate": 0.0,
      "loss": 0.7954,
      "step": 8857
    },
    {
      "epoch": 0.795009872554299,
      "grad_norm": 2.9241943359375,
      "learning_rate": 0.0,
      "loss": 0.7062,
      "step": 8858
    },
    {
      "epoch": 0.7950996230479268,
      "grad_norm": 2.5907375812530518,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 8859
    },
    {
      "epoch": 0.7951893735415545,
      "grad_norm": 4.669808387756348,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 8860
    },
    {
      "epoch": 0.7952791240351822,
      "grad_norm": 3.328996181488037,
      "learning_rate": 0.0,
      "loss": 0.6704,
      "step": 8861
    },
    {
      "epoch": 0.7953688745288099,
      "grad_norm": 2.353452205657959,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 8862
    },
    {
      "epoch": 0.7954586250224376,
      "grad_norm": 2.3218348026275635,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 8863
    },
    {
      "epoch": 0.7955483755160654,
      "grad_norm": 2.599551200866699,
      "learning_rate": 0.0,
      "loss": 0.6893,
      "step": 8864
    },
    {
      "epoch": 0.7956381260096931,
      "grad_norm": 2.9750936031341553,
      "learning_rate": 0.0,
      "loss": 0.682,
      "step": 8865
    },
    {
      "epoch": 0.7957278765033208,
      "grad_norm": 2.5116958618164062,
      "learning_rate": 0.0,
      "loss": 0.6673,
      "step": 8866
    },
    {
      "epoch": 0.7958176269969485,
      "grad_norm": 2.3118064403533936,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 8867
    },
    {
      "epoch": 0.7959073774905762,
      "grad_norm": 2.561772346496582,
      "learning_rate": 0.0,
      "loss": 0.7936,
      "step": 8868
    },
    {
      "epoch": 0.795997127984204,
      "grad_norm": 2.390153408050537,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 8869
    },
    {
      "epoch": 0.7960868784778317,
      "grad_norm": 2.5359392166137695,
      "learning_rate": 0.0,
      "loss": 0.7255,
      "step": 8870
    },
    {
      "epoch": 0.7961766289714594,
      "grad_norm": 2.7520172595977783,
      "learning_rate": 0.0,
      "loss": 0.6787,
      "step": 8871
    },
    {
      "epoch": 0.796266379465087,
      "grad_norm": 2.83198618888855,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 8872
    },
    {
      "epoch": 0.7963561299587147,
      "grad_norm": 2.5315299034118652,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 8873
    },
    {
      "epoch": 0.7964458804523424,
      "grad_norm": 3.015920400619507,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 8874
    },
    {
      "epoch": 0.7965356309459702,
      "grad_norm": 3.206110715866089,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 8875
    },
    {
      "epoch": 0.7966253814395979,
      "grad_norm": 3.069538116455078,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 8876
    },
    {
      "epoch": 0.7967151319332256,
      "grad_norm": 3.110469102859497,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 8877
    },
    {
      "epoch": 0.7968048824268533,
      "grad_norm": 2.7330803871154785,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 8878
    },
    {
      "epoch": 0.796894632920481,
      "grad_norm": 3.1196553707122803,
      "learning_rate": 0.0,
      "loss": 0.6796,
      "step": 8879
    },
    {
      "epoch": 0.7969843834141088,
      "grad_norm": 3.655508518218994,
      "learning_rate": 0.0,
      "loss": 0.6597,
      "step": 8880
    },
    {
      "epoch": 0.7970741339077365,
      "grad_norm": 3.9565141201019287,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 8881
    },
    {
      "epoch": 0.7971638844013642,
      "grad_norm": 3.5835959911346436,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 8882
    },
    {
      "epoch": 0.7972536348949919,
      "grad_norm": 2.961648464202881,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 8883
    },
    {
      "epoch": 0.7973433853886196,
      "grad_norm": 2.4611356258392334,
      "learning_rate": 0.0,
      "loss": 0.6832,
      "step": 8884
    },
    {
      "epoch": 0.7974331358822474,
      "grad_norm": 2.427618980407715,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 8885
    },
    {
      "epoch": 0.7975228863758751,
      "grad_norm": 3.088397264480591,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 8886
    },
    {
      "epoch": 0.7976126368695028,
      "grad_norm": 3.9356298446655273,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 8887
    },
    {
      "epoch": 0.7977023873631305,
      "grad_norm": 2.775197982788086,
      "learning_rate": 0.0,
      "loss": 0.6565,
      "step": 8888
    },
    {
      "epoch": 0.7977921378567582,
      "grad_norm": 3.1404857635498047,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 8889
    },
    {
      "epoch": 0.797881888350386,
      "grad_norm": 5.461668491363525,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 8890
    },
    {
      "epoch": 0.7979716388440137,
      "grad_norm": 2.334808826446533,
      "learning_rate": 0.0,
      "loss": 0.7186,
      "step": 8891
    },
    {
      "epoch": 0.7980613893376414,
      "grad_norm": 2.7621803283691406,
      "learning_rate": 0.0,
      "loss": 0.7539,
      "step": 8892
    },
    {
      "epoch": 0.7981511398312691,
      "grad_norm": 2.691469669342041,
      "learning_rate": 0.0,
      "loss": 0.7542,
      "step": 8893
    },
    {
      "epoch": 0.7982408903248968,
      "grad_norm": 2.8786685466766357,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 8894
    },
    {
      "epoch": 0.7983306408185245,
      "grad_norm": 2.6135787963867188,
      "learning_rate": 0.0,
      "loss": 0.6867,
      "step": 8895
    },
    {
      "epoch": 0.7984203913121523,
      "grad_norm": 3.4727139472961426,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 8896
    },
    {
      "epoch": 0.79851014180578,
      "grad_norm": 2.5353753566741943,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 8897
    },
    {
      "epoch": 0.7985998922994076,
      "grad_norm": 3.026193141937256,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 8898
    },
    {
      "epoch": 0.7986896427930353,
      "grad_norm": 3.577324867248535,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 8899
    },
    {
      "epoch": 0.798779393286663,
      "grad_norm": 2.6265528202056885,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 8900
    },
    {
      "epoch": 0.7988691437802908,
      "grad_norm": 2.884441375732422,
      "learning_rate": 0.0,
      "loss": 0.7267,
      "step": 8901
    },
    {
      "epoch": 0.7989588942739185,
      "grad_norm": 2.389324188232422,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 8902
    },
    {
      "epoch": 0.7990486447675462,
      "grad_norm": 2.91264009475708,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 8903
    },
    {
      "epoch": 0.7991383952611739,
      "grad_norm": 2.7543485164642334,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 8904
    },
    {
      "epoch": 0.7992281457548016,
      "grad_norm": 2.4505982398986816,
      "learning_rate": 0.0,
      "loss": 0.7093,
      "step": 8905
    },
    {
      "epoch": 0.7993178962484294,
      "grad_norm": 2.262659788131714,
      "learning_rate": 0.0,
      "loss": 0.6442,
      "step": 8906
    },
    {
      "epoch": 0.7994076467420571,
      "grad_norm": 2.655465841293335,
      "learning_rate": 0.0,
      "loss": 0.6681,
      "step": 8907
    },
    {
      "epoch": 0.7994973972356848,
      "grad_norm": 2.8694052696228027,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 8908
    },
    {
      "epoch": 0.7995871477293125,
      "grad_norm": 3.2418582439422607,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 8909
    },
    {
      "epoch": 0.7996768982229402,
      "grad_norm": 2.1511518955230713,
      "learning_rate": 0.0,
      "loss": 0.6629,
      "step": 8910
    },
    {
      "epoch": 0.799766648716568,
      "grad_norm": 2.9662744998931885,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 8911
    },
    {
      "epoch": 0.7998563992101957,
      "grad_norm": 2.464754819869995,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 8912
    },
    {
      "epoch": 0.7999461497038234,
      "grad_norm": 2.366781711578369,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 8913
    },
    {
      "epoch": 0.8000359001974511,
      "grad_norm": 2.439488410949707,
      "learning_rate": 0.0,
      "loss": 0.6604,
      "step": 8914
    },
    {
      "epoch": 0.8001256506910788,
      "grad_norm": 4.654123306274414,
      "learning_rate": 0.0,
      "loss": 0.7301,
      "step": 8915
    },
    {
      "epoch": 0.8002154011847065,
      "grad_norm": 3.1212048530578613,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 8916
    },
    {
      "epoch": 0.8003051516783343,
      "grad_norm": 3.1859192848205566,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 8917
    },
    {
      "epoch": 0.800394902171962,
      "grad_norm": 2.3026154041290283,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 8918
    },
    {
      "epoch": 0.8004846526655897,
      "grad_norm": 3.4412105083465576,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 8919
    },
    {
      "epoch": 0.8005744031592174,
      "grad_norm": 2.5344908237457275,
      "learning_rate": 0.0,
      "loss": 0.6891,
      "step": 8920
    },
    {
      "epoch": 0.800664153652845,
      "grad_norm": 2.604151725769043,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 8921
    },
    {
      "epoch": 0.8007539041464729,
      "grad_norm": 3.840975522994995,
      "learning_rate": 0.0,
      "loss": 0.7515,
      "step": 8922
    },
    {
      "epoch": 0.8008436546401005,
      "grad_norm": 3.8762755393981934,
      "learning_rate": 0.0,
      "loss": 0.7359,
      "step": 8923
    },
    {
      "epoch": 0.8009334051337282,
      "grad_norm": 2.4131412506103516,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 8924
    },
    {
      "epoch": 0.8010231556273559,
      "grad_norm": 2.6421475410461426,
      "learning_rate": 0.0,
      "loss": 0.7474,
      "step": 8925
    },
    {
      "epoch": 0.8011129061209836,
      "grad_norm": 2.811976909637451,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 8926
    },
    {
      "epoch": 0.8012026566146114,
      "grad_norm": 2.978926181793213,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 8927
    },
    {
      "epoch": 0.8012924071082391,
      "grad_norm": 2.6947460174560547,
      "learning_rate": 0.0,
      "loss": 0.6593,
      "step": 8928
    },
    {
      "epoch": 0.8013821576018668,
      "grad_norm": 5.898365497589111,
      "learning_rate": 0.0,
      "loss": 0.6898,
      "step": 8929
    },
    {
      "epoch": 0.8014719080954945,
      "grad_norm": 2.968482255935669,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 8930
    },
    {
      "epoch": 0.8015616585891222,
      "grad_norm": 2.5917208194732666,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 8931
    },
    {
      "epoch": 0.80165140908275,
      "grad_norm": 2.3168423175811768,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 8932
    },
    {
      "epoch": 0.8017411595763777,
      "grad_norm": 2.9868245124816895,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 8933
    },
    {
      "epoch": 0.8018309100700054,
      "grad_norm": 2.4478461742401123,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 8934
    },
    {
      "epoch": 0.8019206605636331,
      "grad_norm": 2.7042348384857178,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 8935
    },
    {
      "epoch": 0.8020104110572608,
      "grad_norm": 2.1432557106018066,
      "learning_rate": 0.0,
      "loss": 0.6849,
      "step": 8936
    },
    {
      "epoch": 0.8021001615508885,
      "grad_norm": 3.6468746662139893,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 8937
    },
    {
      "epoch": 0.8021899120445163,
      "grad_norm": 2.56156587600708,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 8938
    },
    {
      "epoch": 0.802279662538144,
      "grad_norm": 1.9641176462173462,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 8939
    },
    {
      "epoch": 0.8023694130317717,
      "grad_norm": 2.827688694000244,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 8940
    },
    {
      "epoch": 0.8024591635253994,
      "grad_norm": 2.596895933151245,
      "learning_rate": 0.0,
      "loss": 0.7625,
      "step": 8941
    },
    {
      "epoch": 0.8025489140190271,
      "grad_norm": 3.2147674560546875,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 8942
    },
    {
      "epoch": 0.8026386645126549,
      "grad_norm": 2.8366217613220215,
      "learning_rate": 0.0,
      "loss": 0.6789,
      "step": 8943
    },
    {
      "epoch": 0.8027284150062826,
      "grad_norm": 3.012453556060791,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 8944
    },
    {
      "epoch": 0.8028181654999103,
      "grad_norm": 2.929635524749756,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 8945
    },
    {
      "epoch": 0.802907915993538,
      "grad_norm": 3.058089256286621,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 8946
    },
    {
      "epoch": 0.8029976664871656,
      "grad_norm": 3.8271071910858154,
      "learning_rate": 0.0,
      "loss": 0.7278,
      "step": 8947
    },
    {
      "epoch": 0.8030874169807934,
      "grad_norm": 3.1167988777160645,
      "learning_rate": 0.0,
      "loss": 0.7263,
      "step": 8948
    },
    {
      "epoch": 0.8031771674744211,
      "grad_norm": 3.5650975704193115,
      "learning_rate": 0.0,
      "loss": 0.7321,
      "step": 8949
    },
    {
      "epoch": 0.8032669179680488,
      "grad_norm": 2.862542152404785,
      "learning_rate": 0.0,
      "loss": 0.6872,
      "step": 8950
    },
    {
      "epoch": 0.8033566684616765,
      "grad_norm": 2.5918710231781006,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 8951
    },
    {
      "epoch": 0.8034464189553042,
      "grad_norm": 3.2065165042877197,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 8952
    },
    {
      "epoch": 0.8035361694489319,
      "grad_norm": 4.9401373863220215,
      "learning_rate": 0.0,
      "loss": 0.7076,
      "step": 8953
    },
    {
      "epoch": 0.8036259199425597,
      "grad_norm": 2.9777209758758545,
      "learning_rate": 0.0,
      "loss": 0.7086,
      "step": 8954
    },
    {
      "epoch": 0.8037156704361874,
      "grad_norm": 1.9710443019866943,
      "learning_rate": 0.0,
      "loss": 0.7247,
      "step": 8955
    },
    {
      "epoch": 0.8038054209298151,
      "grad_norm": 3.1767284870147705,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 8956
    },
    {
      "epoch": 0.8038951714234428,
      "grad_norm": 2.5323727130889893,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 8957
    },
    {
      "epoch": 0.8039849219170705,
      "grad_norm": 3.964578151702881,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 8958
    },
    {
      "epoch": 0.8040746724106983,
      "grad_norm": 4.0215888023376465,
      "learning_rate": 0.0,
      "loss": 0.6461,
      "step": 8959
    },
    {
      "epoch": 0.804164422904326,
      "grad_norm": 2.225168228149414,
      "learning_rate": 0.0,
      "loss": 0.6633,
      "step": 8960
    },
    {
      "epoch": 0.8042541733979537,
      "grad_norm": 4.22755241394043,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 8961
    },
    {
      "epoch": 0.8043439238915814,
      "grad_norm": 2.7088701725006104,
      "learning_rate": 0.0,
      "loss": 0.6865,
      "step": 8962
    },
    {
      "epoch": 0.8044336743852091,
      "grad_norm": 5.753403186798096,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 8963
    },
    {
      "epoch": 0.8045234248788369,
      "grad_norm": 3.6736154556274414,
      "learning_rate": 0.0,
      "loss": 0.6719,
      "step": 8964
    },
    {
      "epoch": 0.8046131753724646,
      "grad_norm": 12.118819236755371,
      "learning_rate": 0.0,
      "loss": 0.7479,
      "step": 8965
    },
    {
      "epoch": 0.8047029258660923,
      "grad_norm": 2.8476831912994385,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 8966
    },
    {
      "epoch": 0.80479267635972,
      "grad_norm": 1.9640872478485107,
      "learning_rate": 0.0,
      "loss": 0.6975,
      "step": 8967
    },
    {
      "epoch": 0.8048824268533477,
      "grad_norm": 3.4488961696624756,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 8968
    },
    {
      "epoch": 0.8049721773469755,
      "grad_norm": 3.1189866065979004,
      "learning_rate": 0.0,
      "loss": 0.6653,
      "step": 8969
    },
    {
      "epoch": 0.8050619278406032,
      "grad_norm": 3.083113431930542,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 8970
    },
    {
      "epoch": 0.8051516783342308,
      "grad_norm": 3.2261221408843994,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 8971
    },
    {
      "epoch": 0.8052414288278585,
      "grad_norm": 2.9027650356292725,
      "learning_rate": 0.0,
      "loss": 0.679,
      "step": 8972
    },
    {
      "epoch": 0.8053311793214862,
      "grad_norm": 3.3906760215759277,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 8973
    },
    {
      "epoch": 0.8054209298151139,
      "grad_norm": 3.3413889408111572,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 8974
    },
    {
      "epoch": 0.8055106803087417,
      "grad_norm": 2.977782726287842,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 8975
    },
    {
      "epoch": 0.8056004308023694,
      "grad_norm": 3.0301990509033203,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 8976
    },
    {
      "epoch": 0.8056901812959971,
      "grad_norm": 4.307399749755859,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 8977
    },
    {
      "epoch": 0.8057799317896248,
      "grad_norm": 4.773717880249023,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 8978
    },
    {
      "epoch": 0.8058696822832525,
      "grad_norm": 3.087993860244751,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 8979
    },
    {
      "epoch": 0.8059594327768803,
      "grad_norm": 3.311141014099121,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 8980
    },
    {
      "epoch": 0.806049183270508,
      "grad_norm": 2.774996280670166,
      "learning_rate": 0.0,
      "loss": 0.74,
      "step": 8981
    },
    {
      "epoch": 0.8061389337641357,
      "grad_norm": 3.049596071243286,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 8982
    },
    {
      "epoch": 0.8062286842577634,
      "grad_norm": 2.276028633117676,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 8983
    },
    {
      "epoch": 0.8063184347513911,
      "grad_norm": 6.385080814361572,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 8984
    },
    {
      "epoch": 0.8064081852450189,
      "grad_norm": 3.4816641807556152,
      "learning_rate": 0.0,
      "loss": 0.6832,
      "step": 8985
    },
    {
      "epoch": 0.8064979357386466,
      "grad_norm": 3.4484403133392334,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 8986
    },
    {
      "epoch": 0.8065876862322743,
      "grad_norm": 4.366994380950928,
      "learning_rate": 0.0,
      "loss": 0.7425,
      "step": 8987
    },
    {
      "epoch": 0.806677436725902,
      "grad_norm": 3.123591184616089,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 8988
    },
    {
      "epoch": 0.8067671872195297,
      "grad_norm": 2.4241583347320557,
      "learning_rate": 0.0,
      "loss": 0.8165,
      "step": 8989
    },
    {
      "epoch": 0.8068569377131575,
      "grad_norm": 2.6294243335723877,
      "learning_rate": 0.0,
      "loss": 0.7056,
      "step": 8990
    },
    {
      "epoch": 0.8069466882067852,
      "grad_norm": 3.0484893321990967,
      "learning_rate": 0.0,
      "loss": 0.6766,
      "step": 8991
    },
    {
      "epoch": 0.8070364387004129,
      "grad_norm": 3.0260732173919678,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 8992
    },
    {
      "epoch": 0.8071261891940406,
      "grad_norm": 3.408273696899414,
      "learning_rate": 0.0,
      "loss": 0.7991,
      "step": 8993
    },
    {
      "epoch": 0.8072159396876682,
      "grad_norm": 3.1908254623413086,
      "learning_rate": 0.0,
      "loss": 0.6838,
      "step": 8994
    },
    {
      "epoch": 0.8073056901812959,
      "grad_norm": 4.064762115478516,
      "learning_rate": 0.0,
      "loss": 0.6518,
      "step": 8995
    },
    {
      "epoch": 0.8073954406749237,
      "grad_norm": 3.623396635055542,
      "learning_rate": 0.0,
      "loss": 0.6868,
      "step": 8996
    },
    {
      "epoch": 0.8074851911685514,
      "grad_norm": 4.088142395019531,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 8997
    },
    {
      "epoch": 0.8075749416621791,
      "grad_norm": 3.351809024810791,
      "learning_rate": 0.0,
      "loss": 0.6559,
      "step": 8998
    },
    {
      "epoch": 0.8076646921558068,
      "grad_norm": 4.14436674118042,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 8999
    },
    {
      "epoch": 0.8077544426494345,
      "grad_norm": 3.015687942504883,
      "learning_rate": 0.0,
      "loss": 0.66,
      "step": 9000
    },
    {
      "epoch": 0.8078441931430623,
      "grad_norm": 2.9020888805389404,
      "learning_rate": 0.0,
      "loss": 0.6774,
      "step": 9001
    },
    {
      "epoch": 0.80793394363669,
      "grad_norm": 2.7946279048919678,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 9002
    },
    {
      "epoch": 0.8080236941303177,
      "grad_norm": 6.793879985809326,
      "learning_rate": 0.0,
      "loss": 0.7681,
      "step": 9003
    },
    {
      "epoch": 0.8081134446239454,
      "grad_norm": 2.791360378265381,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 9004
    },
    {
      "epoch": 0.8082031951175731,
      "grad_norm": 4.9200334548950195,
      "learning_rate": 0.0,
      "loss": 0.7133,
      "step": 9005
    },
    {
      "epoch": 0.8082929456112009,
      "grad_norm": 2.4291625022888184,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 9006
    },
    {
      "epoch": 0.8083826961048286,
      "grad_norm": 2.4277541637420654,
      "learning_rate": 0.0,
      "loss": 0.7809,
      "step": 9007
    },
    {
      "epoch": 0.8084724465984563,
      "grad_norm": 2.3432815074920654,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 9008
    },
    {
      "epoch": 0.808562197092084,
      "grad_norm": 2.751720905303955,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 9009
    },
    {
      "epoch": 0.8086519475857117,
      "grad_norm": 3.457775115966797,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 9010
    },
    {
      "epoch": 0.8087416980793395,
      "grad_norm": 2.917555809020996,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 9011
    },
    {
      "epoch": 0.8088314485729672,
      "grad_norm": 3.6606290340423584,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 9012
    },
    {
      "epoch": 0.8089211990665949,
      "grad_norm": 3.45869779586792,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 9013
    },
    {
      "epoch": 0.8090109495602226,
      "grad_norm": 2.623383045196533,
      "learning_rate": 0.0,
      "loss": 0.7754,
      "step": 9014
    },
    {
      "epoch": 0.8091007000538503,
      "grad_norm": 2.87666916847229,
      "learning_rate": 0.0,
      "loss": 0.697,
      "step": 9015
    },
    {
      "epoch": 0.809190450547478,
      "grad_norm": 3.6762349605560303,
      "learning_rate": 0.0,
      "loss": 0.7365,
      "step": 9016
    },
    {
      "epoch": 0.8092802010411058,
      "grad_norm": 3.0734243392944336,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 9017
    },
    {
      "epoch": 0.8093699515347335,
      "grad_norm": 6.611454486846924,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 9018
    },
    {
      "epoch": 0.8094597020283611,
      "grad_norm": 3.5374393463134766,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 9019
    },
    {
      "epoch": 0.8095494525219888,
      "grad_norm": 3.4809892177581787,
      "learning_rate": 0.0,
      "loss": 0.7357,
      "step": 9020
    },
    {
      "epoch": 0.8096392030156165,
      "grad_norm": 2.2897708415985107,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 9021
    },
    {
      "epoch": 0.8097289535092443,
      "grad_norm": 3.1667909622192383,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 9022
    },
    {
      "epoch": 0.809818704002872,
      "grad_norm": 2.7613701820373535,
      "learning_rate": 0.0,
      "loss": 0.6787,
      "step": 9023
    },
    {
      "epoch": 0.8099084544964997,
      "grad_norm": 3.4777822494506836,
      "learning_rate": 0.0,
      "loss": 0.6841,
      "step": 9024
    },
    {
      "epoch": 0.8099982049901274,
      "grad_norm": 2.991180181503296,
      "learning_rate": 0.0,
      "loss": 0.6291,
      "step": 9025
    },
    {
      "epoch": 0.8100879554837551,
      "grad_norm": 2.6769018173217773,
      "learning_rate": 0.0,
      "loss": 0.7302,
      "step": 9026
    },
    {
      "epoch": 0.8101777059773829,
      "grad_norm": 4.784292697906494,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 9027
    },
    {
      "epoch": 0.8102674564710106,
      "grad_norm": 3.086137294769287,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 9028
    },
    {
      "epoch": 0.8103572069646383,
      "grad_norm": 2.7499520778656006,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 9029
    },
    {
      "epoch": 0.810446957458266,
      "grad_norm": 2.6460211277008057,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 9030
    },
    {
      "epoch": 0.8105367079518937,
      "grad_norm": 3.088130235671997,
      "learning_rate": 0.0,
      "loss": 0.6758,
      "step": 9031
    },
    {
      "epoch": 0.8106264584455215,
      "grad_norm": 2.8989217281341553,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 9032
    },
    {
      "epoch": 0.8107162089391492,
      "grad_norm": 2.8608055114746094,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 9033
    },
    {
      "epoch": 0.8108059594327769,
      "grad_norm": 2.7467572689056396,
      "learning_rate": 0.0,
      "loss": 0.6825,
      "step": 9034
    },
    {
      "epoch": 0.8108957099264046,
      "grad_norm": 3.283160448074341,
      "learning_rate": 0.0,
      "loss": 0.6828,
      "step": 9035
    },
    {
      "epoch": 0.8109854604200323,
      "grad_norm": 4.075192928314209,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 9036
    },
    {
      "epoch": 0.81107521091366,
      "grad_norm": 6.287801742553711,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 9037
    },
    {
      "epoch": 0.8111649614072878,
      "grad_norm": 3.5127785205841064,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 9038
    },
    {
      "epoch": 0.8112547119009155,
      "grad_norm": 3.176445245742798,
      "learning_rate": 0.0,
      "loss": 0.6973,
      "step": 9039
    },
    {
      "epoch": 0.8113444623945432,
      "grad_norm": 3.2392711639404297,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 9040
    },
    {
      "epoch": 0.8114342128881709,
      "grad_norm": 2.7447195053100586,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 9041
    },
    {
      "epoch": 0.8115239633817986,
      "grad_norm": 2.809519052505493,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 9042
    },
    {
      "epoch": 0.8116137138754264,
      "grad_norm": 2.892054319381714,
      "learning_rate": 0.0,
      "loss": 0.6736,
      "step": 9043
    },
    {
      "epoch": 0.811703464369054,
      "grad_norm": 3.110790967941284,
      "learning_rate": 0.0,
      "loss": 0.6949,
      "step": 9044
    },
    {
      "epoch": 0.8117932148626817,
      "grad_norm": 2.3142595291137695,
      "learning_rate": 0.0,
      "loss": 0.6655,
      "step": 9045
    },
    {
      "epoch": 0.8118829653563094,
      "grad_norm": 3.1801035404205322,
      "learning_rate": 0.0,
      "loss": 0.693,
      "step": 9046
    },
    {
      "epoch": 0.8119727158499371,
      "grad_norm": 2.78584623336792,
      "learning_rate": 0.0,
      "loss": 0.7746,
      "step": 9047
    },
    {
      "epoch": 0.8120624663435649,
      "grad_norm": 2.6471872329711914,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 9048
    },
    {
      "epoch": 0.8121522168371926,
      "grad_norm": 3.350461006164551,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 9049
    },
    {
      "epoch": 0.8122419673308203,
      "grad_norm": 3.15454363822937,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 9050
    },
    {
      "epoch": 0.812331717824448,
      "grad_norm": 2.400197982788086,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 9051
    },
    {
      "epoch": 0.8124214683180757,
      "grad_norm": 2.8461899757385254,
      "learning_rate": 0.0,
      "loss": 0.7818,
      "step": 9052
    },
    {
      "epoch": 0.8125112188117035,
      "grad_norm": 3.8974852561950684,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 9053
    },
    {
      "epoch": 0.8126009693053312,
      "grad_norm": 2.9701452255249023,
      "learning_rate": 0.0,
      "loss": 0.7003,
      "step": 9054
    },
    {
      "epoch": 0.8126907197989589,
      "grad_norm": 7.066159725189209,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 9055
    },
    {
      "epoch": 0.8127804702925866,
      "grad_norm": 2.8243143558502197,
      "learning_rate": 0.0,
      "loss": 0.6774,
      "step": 9056
    },
    {
      "epoch": 0.8128702207862143,
      "grad_norm": 2.778634786605835,
      "learning_rate": 0.0,
      "loss": 0.7236,
      "step": 9057
    },
    {
      "epoch": 0.812959971279842,
      "grad_norm": 2.610172986984253,
      "learning_rate": 0.0,
      "loss": 0.7621,
      "step": 9058
    },
    {
      "epoch": 0.8130497217734698,
      "grad_norm": 3.050136089324951,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 9059
    },
    {
      "epoch": 0.8131394722670975,
      "grad_norm": 3.982510805130005,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 9060
    },
    {
      "epoch": 0.8132292227607252,
      "grad_norm": 3.2739098072052,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 9061
    },
    {
      "epoch": 0.8133189732543529,
      "grad_norm": 6.112913608551025,
      "learning_rate": 0.0,
      "loss": 0.6854,
      "step": 9062
    },
    {
      "epoch": 0.8134087237479806,
      "grad_norm": 2.4228878021240234,
      "learning_rate": 0.0,
      "loss": 0.7351,
      "step": 9063
    },
    {
      "epoch": 0.8134984742416084,
      "grad_norm": 2.7707884311676025,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 9064
    },
    {
      "epoch": 0.8135882247352361,
      "grad_norm": 2.8042140007019043,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 9065
    },
    {
      "epoch": 0.8136779752288638,
      "grad_norm": 3.158045768737793,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 9066
    },
    {
      "epoch": 0.8137677257224915,
      "grad_norm": 2.687164068222046,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 9067
    },
    {
      "epoch": 0.8138574762161191,
      "grad_norm": 2.8563220500946045,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 9068
    },
    {
      "epoch": 0.813947226709747,
      "grad_norm": 3.3456690311431885,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 9069
    },
    {
      "epoch": 0.8140369772033746,
      "grad_norm": 2.4679954051971436,
      "learning_rate": 0.0,
      "loss": 0.6739,
      "step": 9070
    },
    {
      "epoch": 0.8141267276970023,
      "grad_norm": 3.2719249725341797,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 9071
    },
    {
      "epoch": 0.81421647819063,
      "grad_norm": 2.922137975692749,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 9072
    },
    {
      "epoch": 0.8143062286842577,
      "grad_norm": 3.290524959564209,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 9073
    },
    {
      "epoch": 0.8143959791778855,
      "grad_norm": 2.8948519229888916,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 9074
    },
    {
      "epoch": 0.8144857296715132,
      "grad_norm": 2.6718006134033203,
      "learning_rate": 0.0,
      "loss": 0.7346,
      "step": 9075
    },
    {
      "epoch": 0.8145754801651409,
      "grad_norm": 3.1121726036071777,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 9076
    },
    {
      "epoch": 0.8146652306587686,
      "grad_norm": 5.114002227783203,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 9077
    },
    {
      "epoch": 0.8147549811523963,
      "grad_norm": 3.1596367359161377,
      "learning_rate": 0.0,
      "loss": 0.7062,
      "step": 9078
    },
    {
      "epoch": 0.814844731646024,
      "grad_norm": 3.2523975372314453,
      "learning_rate": 0.0,
      "loss": 0.7691,
      "step": 9079
    },
    {
      "epoch": 0.8149344821396518,
      "grad_norm": 3.2793424129486084,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 9080
    },
    {
      "epoch": 0.8150242326332795,
      "grad_norm": 2.873873472213745,
      "learning_rate": 0.0,
      "loss": 0.6917,
      "step": 9081
    },
    {
      "epoch": 0.8151139831269072,
      "grad_norm": 3.22894287109375,
      "learning_rate": 0.0,
      "loss": 0.6879,
      "step": 9082
    },
    {
      "epoch": 0.8152037336205349,
      "grad_norm": 3.416206121444702,
      "learning_rate": 0.0,
      "loss": 0.7641,
      "step": 9083
    },
    {
      "epoch": 0.8152934841141626,
      "grad_norm": 3.1343460083007812,
      "learning_rate": 0.0,
      "loss": 0.6828,
      "step": 9084
    },
    {
      "epoch": 0.8153832346077904,
      "grad_norm": 2.650895118713379,
      "learning_rate": 0.0,
      "loss": 0.7518,
      "step": 9085
    },
    {
      "epoch": 0.8154729851014181,
      "grad_norm": 3.7187581062316895,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 9086
    },
    {
      "epoch": 0.8155627355950458,
      "grad_norm": 3.492868185043335,
      "learning_rate": 0.0,
      "loss": 0.6905,
      "step": 9087
    },
    {
      "epoch": 0.8156524860886735,
      "grad_norm": 4.160069942474365,
      "learning_rate": 0.0,
      "loss": 0.6657,
      "step": 9088
    },
    {
      "epoch": 0.8157422365823012,
      "grad_norm": 2.6900155544281006,
      "learning_rate": 0.0,
      "loss": 0.7402,
      "step": 9089
    },
    {
      "epoch": 0.815831987075929,
      "grad_norm": 4.1578192710876465,
      "learning_rate": 0.0,
      "loss": 0.7306,
      "step": 9090
    },
    {
      "epoch": 0.8159217375695567,
      "grad_norm": 3.3593311309814453,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 9091
    },
    {
      "epoch": 0.8160114880631844,
      "grad_norm": 3.299741506576538,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 9092
    },
    {
      "epoch": 0.816101238556812,
      "grad_norm": 3.1887571811676025,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 9093
    },
    {
      "epoch": 0.8161909890504397,
      "grad_norm": 2.6013505458831787,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 9094
    },
    {
      "epoch": 0.8162807395440675,
      "grad_norm": 2.9609928131103516,
      "learning_rate": 0.0,
      "loss": 0.6758,
      "step": 9095
    },
    {
      "epoch": 0.8163704900376952,
      "grad_norm": 3.0883288383483887,
      "learning_rate": 0.0,
      "loss": 0.6978,
      "step": 9096
    },
    {
      "epoch": 0.8164602405313229,
      "grad_norm": 2.299881935119629,
      "learning_rate": 0.0,
      "loss": 0.6584,
      "step": 9097
    },
    {
      "epoch": 0.8165499910249506,
      "grad_norm": 2.992509126663208,
      "learning_rate": 0.0,
      "loss": 0.664,
      "step": 9098
    },
    {
      "epoch": 0.8166397415185783,
      "grad_norm": 2.486297845840454,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 9099
    },
    {
      "epoch": 0.816729492012206,
      "grad_norm": 2.453399419784546,
      "learning_rate": 0.0,
      "loss": 0.8038,
      "step": 9100
    },
    {
      "epoch": 0.8168192425058338,
      "grad_norm": 4.351288795471191,
      "learning_rate": 0.0,
      "loss": 0.753,
      "step": 9101
    },
    {
      "epoch": 0.8169089929994615,
      "grad_norm": 2.738333225250244,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 9102
    },
    {
      "epoch": 0.8169987434930892,
      "grad_norm": 3.161996603012085,
      "learning_rate": 0.0,
      "loss": 0.7797,
      "step": 9103
    },
    {
      "epoch": 0.8170884939867169,
      "grad_norm": 3.043389320373535,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 9104
    },
    {
      "epoch": 0.8171782444803446,
      "grad_norm": 3.5797157287597656,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 9105
    },
    {
      "epoch": 0.8172679949739724,
      "grad_norm": 2.569511890411377,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 9106
    },
    {
      "epoch": 0.8173577454676001,
      "grad_norm": 4.066756725311279,
      "learning_rate": 0.0,
      "loss": 0.7567,
      "step": 9107
    },
    {
      "epoch": 0.8174474959612278,
      "grad_norm": 2.9159626960754395,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 9108
    },
    {
      "epoch": 0.8175372464548555,
      "grad_norm": 4.3207316398620605,
      "learning_rate": 0.0,
      "loss": 0.6668,
      "step": 9109
    },
    {
      "epoch": 0.8176269969484832,
      "grad_norm": 4.852136135101318,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 9110
    },
    {
      "epoch": 0.817716747442111,
      "grad_norm": 2.9058454036712646,
      "learning_rate": 0.0,
      "loss": 0.7297,
      "step": 9111
    },
    {
      "epoch": 0.8178064979357387,
      "grad_norm": 2.938164234161377,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 9112
    },
    {
      "epoch": 0.8178962484293664,
      "grad_norm": 2.552666664123535,
      "learning_rate": 0.0,
      "loss": 0.6699,
      "step": 9113
    },
    {
      "epoch": 0.8179859989229941,
      "grad_norm": 3.030426025390625,
      "learning_rate": 0.0,
      "loss": 0.7893,
      "step": 9114
    },
    {
      "epoch": 0.8180757494166218,
      "grad_norm": 3.864659309387207,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 9115
    },
    {
      "epoch": 0.8181654999102496,
      "grad_norm": 2.4896976947784424,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 9116
    },
    {
      "epoch": 0.8182552504038773,
      "grad_norm": 3.0432591438293457,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 9117
    },
    {
      "epoch": 0.818345000897505,
      "grad_norm": 3.671678066253662,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 9118
    },
    {
      "epoch": 0.8184347513911326,
      "grad_norm": 4.001484394073486,
      "learning_rate": 0.0,
      "loss": 0.6861,
      "step": 9119
    },
    {
      "epoch": 0.8185245018847603,
      "grad_norm": 3.214029312133789,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 9120
    },
    {
      "epoch": 0.818614252378388,
      "grad_norm": 2.4763662815093994,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 9121
    },
    {
      "epoch": 0.8187040028720158,
      "grad_norm": 2.2358152866363525,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 9122
    },
    {
      "epoch": 0.8187937533656435,
      "grad_norm": 2.474774122238159,
      "learning_rate": 0.0,
      "loss": 0.7226,
      "step": 9123
    },
    {
      "epoch": 0.8188835038592712,
      "grad_norm": 4.5772857666015625,
      "learning_rate": 0.0,
      "loss": 0.7523,
      "step": 9124
    },
    {
      "epoch": 0.8189732543528989,
      "grad_norm": 3.739551305770874,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 9125
    },
    {
      "epoch": 0.8190630048465266,
      "grad_norm": 3.103353500366211,
      "learning_rate": 0.0,
      "loss": 0.6648,
      "step": 9126
    },
    {
      "epoch": 0.8191527553401544,
      "grad_norm": 2.4304749965667725,
      "learning_rate": 0.0,
      "loss": 0.7341,
      "step": 9127
    },
    {
      "epoch": 0.8192425058337821,
      "grad_norm": 2.6649703979492188,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 9128
    },
    {
      "epoch": 0.8193322563274098,
      "grad_norm": 2.7541048526763916,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 9129
    },
    {
      "epoch": 0.8194220068210375,
      "grad_norm": 4.021118640899658,
      "learning_rate": 0.0,
      "loss": 0.6329,
      "step": 9130
    },
    {
      "epoch": 0.8195117573146652,
      "grad_norm": 4.070761203765869,
      "learning_rate": 0.0,
      "loss": 0.6965,
      "step": 9131
    },
    {
      "epoch": 0.819601507808293,
      "grad_norm": 3.4881560802459717,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 9132
    },
    {
      "epoch": 0.8196912583019207,
      "grad_norm": 2.7494101524353027,
      "learning_rate": 0.0,
      "loss": 0.739,
      "step": 9133
    },
    {
      "epoch": 0.8197810087955484,
      "grad_norm": 3.4868502616882324,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 9134
    },
    {
      "epoch": 0.8198707592891761,
      "grad_norm": 2.581942558288574,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 9135
    },
    {
      "epoch": 0.8199605097828038,
      "grad_norm": 2.860830307006836,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 9136
    },
    {
      "epoch": 0.8200502602764316,
      "grad_norm": 2.8927786350250244,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 9137
    },
    {
      "epoch": 0.8201400107700593,
      "grad_norm": 2.2404024600982666,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 9138
    },
    {
      "epoch": 0.820229761263687,
      "grad_norm": 3.162602186203003,
      "learning_rate": 0.0,
      "loss": 0.7168,
      "step": 9139
    },
    {
      "epoch": 0.8203195117573147,
      "grad_norm": 2.9550557136535645,
      "learning_rate": 0.0,
      "loss": 0.6737,
      "step": 9140
    },
    {
      "epoch": 0.8204092622509424,
      "grad_norm": 2.840688467025757,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 9141
    },
    {
      "epoch": 0.82049901274457,
      "grad_norm": 3.75162672996521,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 9142
    },
    {
      "epoch": 0.8205887632381978,
      "grad_norm": 2.9122023582458496,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 9143
    },
    {
      "epoch": 0.8206785137318255,
      "grad_norm": 2.7347867488861084,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 9144
    },
    {
      "epoch": 0.8207682642254532,
      "grad_norm": 2.4384777545928955,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 9145
    },
    {
      "epoch": 0.8208580147190809,
      "grad_norm": 1.9741564989089966,
      "learning_rate": 0.0,
      "loss": 0.6833,
      "step": 9146
    },
    {
      "epoch": 0.8209477652127086,
      "grad_norm": 2.746583938598633,
      "learning_rate": 0.0,
      "loss": 0.7675,
      "step": 9147
    },
    {
      "epoch": 0.8210375157063364,
      "grad_norm": 2.63417649269104,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 9148
    },
    {
      "epoch": 0.8211272661999641,
      "grad_norm": 2.4405715465545654,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 9149
    },
    {
      "epoch": 0.8212170166935918,
      "grad_norm": 2.746408224105835,
      "learning_rate": 0.0,
      "loss": 0.7281,
      "step": 9150
    },
    {
      "epoch": 0.8213067671872195,
      "grad_norm": 2.8863112926483154,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 9151
    },
    {
      "epoch": 0.8213965176808472,
      "grad_norm": 2.6330316066741943,
      "learning_rate": 0.0,
      "loss": 0.6833,
      "step": 9152
    },
    {
      "epoch": 0.821486268174475,
      "grad_norm": 3.074981689453125,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 9153
    },
    {
      "epoch": 0.8215760186681027,
      "grad_norm": 3.3869433403015137,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 9154
    },
    {
      "epoch": 0.8216657691617304,
      "grad_norm": 3.03947114944458,
      "learning_rate": 0.0,
      "loss": 0.6538,
      "step": 9155
    },
    {
      "epoch": 0.8217555196553581,
      "grad_norm": 3.5201098918914795,
      "learning_rate": 0.0,
      "loss": 0.7062,
      "step": 9156
    },
    {
      "epoch": 0.8218452701489858,
      "grad_norm": 4.050161361694336,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 9157
    },
    {
      "epoch": 0.8219350206426136,
      "grad_norm": 3.5607099533081055,
      "learning_rate": 0.0,
      "loss": 0.6869,
      "step": 9158
    },
    {
      "epoch": 0.8220247711362413,
      "grad_norm": 3.8925888538360596,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 9159
    },
    {
      "epoch": 0.822114521629869,
      "grad_norm": 2.5919597148895264,
      "learning_rate": 0.0,
      "loss": 0.7714,
      "step": 9160
    },
    {
      "epoch": 0.8222042721234967,
      "grad_norm": 5.842674255371094,
      "learning_rate": 0.0,
      "loss": 0.7709,
      "step": 9161
    },
    {
      "epoch": 0.8222940226171244,
      "grad_norm": 3.431234836578369,
      "learning_rate": 0.0,
      "loss": 0.6874,
      "step": 9162
    },
    {
      "epoch": 0.8223837731107521,
      "grad_norm": 3.7945897579193115,
      "learning_rate": 0.0,
      "loss": 0.7513,
      "step": 9163
    },
    {
      "epoch": 0.8224735236043799,
      "grad_norm": 2.970813751220703,
      "learning_rate": 0.0,
      "loss": 0.6755,
      "step": 9164
    },
    {
      "epoch": 0.8225632740980076,
      "grad_norm": 2.4104340076446533,
      "learning_rate": 0.0,
      "loss": 0.6974,
      "step": 9165
    },
    {
      "epoch": 0.8226530245916353,
      "grad_norm": 4.720928192138672,
      "learning_rate": 0.0,
      "loss": 0.6773,
      "step": 9166
    },
    {
      "epoch": 0.822742775085263,
      "grad_norm": 6.158131122589111,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 9167
    },
    {
      "epoch": 0.8228325255788906,
      "grad_norm": 3.418928384780884,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 9168
    },
    {
      "epoch": 0.8229222760725184,
      "grad_norm": 3.3261375427246094,
      "learning_rate": 0.0,
      "loss": 0.7152,
      "step": 9169
    },
    {
      "epoch": 0.8230120265661461,
      "grad_norm": 3.3589255809783936,
      "learning_rate": 0.0,
      "loss": 0.6622,
      "step": 9170
    },
    {
      "epoch": 0.8231017770597738,
      "grad_norm": 2.7690236568450928,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 9171
    },
    {
      "epoch": 0.8231915275534015,
      "grad_norm": 2.4150404930114746,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 9172
    },
    {
      "epoch": 0.8232812780470292,
      "grad_norm": 3.5136401653289795,
      "learning_rate": 0.0,
      "loss": 0.6407,
      "step": 9173
    },
    {
      "epoch": 0.823371028540657,
      "grad_norm": 2.8251025676727295,
      "learning_rate": 0.0,
      "loss": 0.6924,
      "step": 9174
    },
    {
      "epoch": 0.8234607790342847,
      "grad_norm": 3.0932934284210205,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 9175
    },
    {
      "epoch": 0.8235505295279124,
      "grad_norm": 3.1267001628875732,
      "learning_rate": 0.0,
      "loss": 0.7093,
      "step": 9176
    },
    {
      "epoch": 0.8236402800215401,
      "grad_norm": 3.860457420349121,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 9177
    },
    {
      "epoch": 0.8237300305151678,
      "grad_norm": 3.402707099914551,
      "learning_rate": 0.0,
      "loss": 0.7655,
      "step": 9178
    },
    {
      "epoch": 0.8238197810087955,
      "grad_norm": 2.4296395778656006,
      "learning_rate": 0.0,
      "loss": 0.6224,
      "step": 9179
    },
    {
      "epoch": 0.8239095315024233,
      "grad_norm": 3.4515888690948486,
      "learning_rate": 0.0,
      "loss": 0.7548,
      "step": 9180
    },
    {
      "epoch": 0.823999281996051,
      "grad_norm": 2.7706339359283447,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 9181
    },
    {
      "epoch": 0.8240890324896787,
      "grad_norm": 3.8180251121520996,
      "learning_rate": 0.0,
      "loss": 0.7206,
      "step": 9182
    },
    {
      "epoch": 0.8241787829833064,
      "grad_norm": 3.163214921951294,
      "learning_rate": 0.0,
      "loss": 0.6704,
      "step": 9183
    },
    {
      "epoch": 0.8242685334769341,
      "grad_norm": 2.5621509552001953,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 9184
    },
    {
      "epoch": 0.8243582839705619,
      "grad_norm": 2.8314173221588135,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 9185
    },
    {
      "epoch": 0.8244480344641896,
      "grad_norm": 6.152703762054443,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 9186
    },
    {
      "epoch": 0.8245377849578173,
      "grad_norm": 2.720694065093994,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 9187
    },
    {
      "epoch": 0.824627535451445,
      "grad_norm": 3.083798885345459,
      "learning_rate": 0.0,
      "loss": 0.7579,
      "step": 9188
    },
    {
      "epoch": 0.8247172859450727,
      "grad_norm": 5.0119404792785645,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 9189
    },
    {
      "epoch": 0.8248070364387005,
      "grad_norm": 4.516861915588379,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 9190
    },
    {
      "epoch": 0.8248967869323282,
      "grad_norm": 2.822258710861206,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 9191
    },
    {
      "epoch": 0.8249865374259558,
      "grad_norm": 2.910081624984741,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 9192
    },
    {
      "epoch": 0.8250762879195835,
      "grad_norm": 2.7697017192840576,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 9193
    },
    {
      "epoch": 0.8251660384132112,
      "grad_norm": 4.5820231437683105,
      "learning_rate": 0.0,
      "loss": 0.6489,
      "step": 9194
    },
    {
      "epoch": 0.825255788906839,
      "grad_norm": 3.1288037300109863,
      "learning_rate": 0.0,
      "loss": 0.7353,
      "step": 9195
    },
    {
      "epoch": 0.8253455394004667,
      "grad_norm": 3.260397434234619,
      "learning_rate": 0.0,
      "loss": 0.7302,
      "step": 9196
    },
    {
      "epoch": 0.8254352898940944,
      "grad_norm": 3.088308095932007,
      "learning_rate": 0.0,
      "loss": 0.6826,
      "step": 9197
    },
    {
      "epoch": 0.8255250403877221,
      "grad_norm": 3.339477777481079,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 9198
    },
    {
      "epoch": 0.8256147908813498,
      "grad_norm": 2.48756742477417,
      "learning_rate": 0.0,
      "loss": 0.6874,
      "step": 9199
    },
    {
      "epoch": 0.8257045413749775,
      "grad_norm": 4.803395748138428,
      "learning_rate": 0.0,
      "loss": 0.7672,
      "step": 9200
    },
    {
      "epoch": 0.8257942918686053,
      "grad_norm": 3.8611209392547607,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 9201
    },
    {
      "epoch": 0.825884042362233,
      "grad_norm": 3.518172264099121,
      "learning_rate": 0.0,
      "loss": 0.7912,
      "step": 9202
    },
    {
      "epoch": 0.8259737928558607,
      "grad_norm": 2.481177806854248,
      "learning_rate": 0.0,
      "loss": 0.7535,
      "step": 9203
    },
    {
      "epoch": 0.8260635433494884,
      "grad_norm": 4.212523460388184,
      "learning_rate": 0.0,
      "loss": 0.7149,
      "step": 9204
    },
    {
      "epoch": 0.8261532938431161,
      "grad_norm": 6.724275588989258,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 9205
    },
    {
      "epoch": 0.8262430443367439,
      "grad_norm": 4.266756057739258,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 9206
    },
    {
      "epoch": 0.8263327948303716,
      "grad_norm": 2.793126106262207,
      "learning_rate": 0.0,
      "loss": 0.7651,
      "step": 9207
    },
    {
      "epoch": 0.8264225453239993,
      "grad_norm": 3.722208023071289,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 9208
    },
    {
      "epoch": 0.826512295817627,
      "grad_norm": 3.682833671569824,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 9209
    },
    {
      "epoch": 0.8266020463112547,
      "grad_norm": 4.17730188369751,
      "learning_rate": 0.0,
      "loss": 0.6695,
      "step": 9210
    },
    {
      "epoch": 0.8266917968048825,
      "grad_norm": 3.2643024921417236,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 9211
    },
    {
      "epoch": 0.8267815472985102,
      "grad_norm": 2.8653523921966553,
      "learning_rate": 0.0,
      "loss": 0.6694,
      "step": 9212
    },
    {
      "epoch": 0.8268712977921379,
      "grad_norm": 3.3913626670837402,
      "learning_rate": 0.0,
      "loss": 0.7658,
      "step": 9213
    },
    {
      "epoch": 0.8269610482857656,
      "grad_norm": 3.1588807106018066,
      "learning_rate": 0.0,
      "loss": 0.6958,
      "step": 9214
    },
    {
      "epoch": 0.8270507987793932,
      "grad_norm": 2.9300224781036377,
      "learning_rate": 0.0,
      "loss": 0.7202,
      "step": 9215
    },
    {
      "epoch": 0.827140549273021,
      "grad_norm": 6.082269668579102,
      "learning_rate": 0.0,
      "loss": 0.7019,
      "step": 9216
    },
    {
      "epoch": 0.8272302997666487,
      "grad_norm": 2.8263769149780273,
      "learning_rate": 0.0,
      "loss": 0.7105,
      "step": 9217
    },
    {
      "epoch": 0.8273200502602764,
      "grad_norm": 2.6800804138183594,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 9218
    },
    {
      "epoch": 0.8274098007539041,
      "grad_norm": 2.1376497745513916,
      "learning_rate": 0.0,
      "loss": 0.7533,
      "step": 9219
    },
    {
      "epoch": 0.8274995512475318,
      "grad_norm": 3.46220064163208,
      "learning_rate": 0.0,
      "loss": 0.6591,
      "step": 9220
    },
    {
      "epoch": 0.8275893017411595,
      "grad_norm": 2.270469903945923,
      "learning_rate": 0.0,
      "loss": 0.6678,
      "step": 9221
    },
    {
      "epoch": 0.8276790522347873,
      "grad_norm": 3.7859532833099365,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 9222
    },
    {
      "epoch": 0.827768802728415,
      "grad_norm": 8.039159774780273,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 9223
    },
    {
      "epoch": 0.8278585532220427,
      "grad_norm": 5.621329307556152,
      "learning_rate": 0.0,
      "loss": 0.6648,
      "step": 9224
    },
    {
      "epoch": 0.8279483037156704,
      "grad_norm": 8.987083435058594,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 9225
    },
    {
      "epoch": 0.8280380542092981,
      "grad_norm": 3.217514753341675,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 9226
    },
    {
      "epoch": 0.8281278047029259,
      "grad_norm": 3.274693727493286,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 9227
    },
    {
      "epoch": 0.8282175551965536,
      "grad_norm": 2.501263380050659,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 9228
    },
    {
      "epoch": 0.8283073056901813,
      "grad_norm": 2.448014497756958,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 9229
    },
    {
      "epoch": 0.828397056183809,
      "grad_norm": 4.559122562408447,
      "learning_rate": 0.0,
      "loss": 0.7566,
      "step": 9230
    },
    {
      "epoch": 0.8284868066774367,
      "grad_norm": 2.207136631011963,
      "learning_rate": 0.0,
      "loss": 0.7717,
      "step": 9231
    },
    {
      "epoch": 0.8285765571710645,
      "grad_norm": 4.435221195220947,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 9232
    },
    {
      "epoch": 0.8286663076646922,
      "grad_norm": 3.696906328201294,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 9233
    },
    {
      "epoch": 0.8287560581583199,
      "grad_norm": 2.7009763717651367,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 9234
    },
    {
      "epoch": 0.8288458086519476,
      "grad_norm": 2.2817060947418213,
      "learning_rate": 0.0,
      "loss": 0.7081,
      "step": 9235
    },
    {
      "epoch": 0.8289355591455753,
      "grad_norm": 4.32763147354126,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 9236
    },
    {
      "epoch": 0.8290253096392031,
      "grad_norm": 2.9774138927459717,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 9237
    },
    {
      "epoch": 0.8291150601328308,
      "grad_norm": 3.2438297271728516,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 9238
    },
    {
      "epoch": 0.8292048106264585,
      "grad_norm": 2.069312810897827,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 9239
    },
    {
      "epoch": 0.8292945611200861,
      "grad_norm": 2.4588961601257324,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 9240
    },
    {
      "epoch": 0.8293843116137138,
      "grad_norm": 3.251668691635132,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 9241
    },
    {
      "epoch": 0.8294740621073415,
      "grad_norm": 3.0488367080688477,
      "learning_rate": 0.0,
      "loss": 0.6717,
      "step": 9242
    },
    {
      "epoch": 0.8295638126009693,
      "grad_norm": 3.27601957321167,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 9243
    },
    {
      "epoch": 0.829653563094597,
      "grad_norm": 3.4855148792266846,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 9244
    },
    {
      "epoch": 0.8297433135882247,
      "grad_norm": 3.697603702545166,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 9245
    },
    {
      "epoch": 0.8298330640818524,
      "grad_norm": 2.904975175857544,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 9246
    },
    {
      "epoch": 0.8299228145754801,
      "grad_norm": 3.2787957191467285,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 9247
    },
    {
      "epoch": 0.8300125650691079,
      "grad_norm": 2.920651912689209,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 9248
    },
    {
      "epoch": 0.8301023155627356,
      "grad_norm": 3.4343793392181396,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 9249
    },
    {
      "epoch": 0.8301920660563633,
      "grad_norm": 4.038681507110596,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 9250
    },
    {
      "epoch": 0.830281816549991,
      "grad_norm": 3.4973561763763428,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 9251
    },
    {
      "epoch": 0.8303715670436187,
      "grad_norm": 2.877037763595581,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 9252
    },
    {
      "epoch": 0.8304613175372465,
      "grad_norm": 5.56553316116333,
      "learning_rate": 0.0,
      "loss": 0.8057,
      "step": 9253
    },
    {
      "epoch": 0.8305510680308742,
      "grad_norm": 3.075103521347046,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 9254
    },
    {
      "epoch": 0.8306408185245019,
      "grad_norm": 2.9424054622650146,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 9255
    },
    {
      "epoch": 0.8307305690181296,
      "grad_norm": 2.7290961742401123,
      "learning_rate": 0.0,
      "loss": 0.7222,
      "step": 9256
    },
    {
      "epoch": 0.8308203195117573,
      "grad_norm": 4.175123691558838,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 9257
    },
    {
      "epoch": 0.8309100700053851,
      "grad_norm": 3.6428091526031494,
      "learning_rate": 0.0,
      "loss": 0.6947,
      "step": 9258
    },
    {
      "epoch": 0.8309998204990128,
      "grad_norm": 2.2944893836975098,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 9259
    },
    {
      "epoch": 0.8310895709926405,
      "grad_norm": 2.9538605213165283,
      "learning_rate": 0.0,
      "loss": 0.6287,
      "step": 9260
    },
    {
      "epoch": 0.8311793214862682,
      "grad_norm": 3.1409027576446533,
      "learning_rate": 0.0,
      "loss": 0.7229,
      "step": 9261
    },
    {
      "epoch": 0.8312690719798959,
      "grad_norm": 2.700943946838379,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 9262
    },
    {
      "epoch": 0.8313588224735236,
      "grad_norm": 1.9322360754013062,
      "learning_rate": 0.0,
      "loss": 0.6421,
      "step": 9263
    },
    {
      "epoch": 0.8314485729671514,
      "grad_norm": 10.193344116210938,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 9264
    },
    {
      "epoch": 0.831538323460779,
      "grad_norm": 3.4730160236358643,
      "learning_rate": 0.0,
      "loss": 0.7207,
      "step": 9265
    },
    {
      "epoch": 0.8316280739544067,
      "grad_norm": 3.253767251968384,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 9266
    },
    {
      "epoch": 0.8317178244480344,
      "grad_norm": 2.517749547958374,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 9267
    },
    {
      "epoch": 0.8318075749416621,
      "grad_norm": 3.1883387565612793,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 9268
    },
    {
      "epoch": 0.8318973254352899,
      "grad_norm": 3.4884939193725586,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 9269
    },
    {
      "epoch": 0.8319870759289176,
      "grad_norm": 2.9986002445220947,
      "learning_rate": 0.0,
      "loss": 0.7638,
      "step": 9270
    },
    {
      "epoch": 0.8320768264225453,
      "grad_norm": 13.415056228637695,
      "learning_rate": 0.0,
      "loss": 0.7306,
      "step": 9271
    },
    {
      "epoch": 0.832166576916173,
      "grad_norm": 3.584360361099243,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 9272
    },
    {
      "epoch": 0.8322563274098007,
      "grad_norm": 2.7891857624053955,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 9273
    },
    {
      "epoch": 0.8323460779034285,
      "grad_norm": 3.784477472305298,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 9274
    },
    {
      "epoch": 0.8324358283970562,
      "grad_norm": 2.9850013256073,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 9275
    },
    {
      "epoch": 0.8325255788906839,
      "grad_norm": 2.9608020782470703,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 9276
    },
    {
      "epoch": 0.8326153293843116,
      "grad_norm": 3.074530601501465,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 9277
    },
    {
      "epoch": 0.8327050798779393,
      "grad_norm": 8.085829734802246,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 9278
    },
    {
      "epoch": 0.8327948303715671,
      "grad_norm": 3.7317771911621094,
      "learning_rate": 0.0,
      "loss": 0.7286,
      "step": 9279
    },
    {
      "epoch": 0.8328845808651948,
      "grad_norm": 3.4410274028778076,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 9280
    },
    {
      "epoch": 0.8329743313588225,
      "grad_norm": 2.5737106800079346,
      "learning_rate": 0.0,
      "loss": 0.6977,
      "step": 9281
    },
    {
      "epoch": 0.8330640818524502,
      "grad_norm": 2.3543570041656494,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 9282
    },
    {
      "epoch": 0.8331538323460779,
      "grad_norm": 5.160919189453125,
      "learning_rate": 0.0,
      "loss": 0.7411,
      "step": 9283
    },
    {
      "epoch": 0.8332435828397056,
      "grad_norm": 3.8074960708618164,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 9284
    },
    {
      "epoch": 0.8333333333333334,
      "grad_norm": 3.095898389816284,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 9285
    },
    {
      "epoch": 0.8334230838269611,
      "grad_norm": 2.71758770942688,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 9286
    },
    {
      "epoch": 0.8335128343205888,
      "grad_norm": 2.849787950515747,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 9287
    },
    {
      "epoch": 0.8336025848142165,
      "grad_norm": 2.949995279312134,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 9288
    },
    {
      "epoch": 0.8336923353078441,
      "grad_norm": 3.0435070991516113,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 9289
    },
    {
      "epoch": 0.833782085801472,
      "grad_norm": 4.705258369445801,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 9290
    },
    {
      "epoch": 0.8338718362950996,
      "grad_norm": 2.034029483795166,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 9291
    },
    {
      "epoch": 0.8339615867887273,
      "grad_norm": 5.161596298217773,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 9292
    },
    {
      "epoch": 0.834051337282355,
      "grad_norm": 3.8997385501861572,
      "learning_rate": 0.0,
      "loss": 0.6783,
      "step": 9293
    },
    {
      "epoch": 0.8341410877759827,
      "grad_norm": 2.641045093536377,
      "learning_rate": 0.0,
      "loss": 0.7182,
      "step": 9294
    },
    {
      "epoch": 0.8342308382696105,
      "grad_norm": 2.24434757232666,
      "learning_rate": 0.0,
      "loss": 0.7283,
      "step": 9295
    },
    {
      "epoch": 0.8343205887632382,
      "grad_norm": 2.837909460067749,
      "learning_rate": 0.0,
      "loss": 0.689,
      "step": 9296
    },
    {
      "epoch": 0.8344103392568659,
      "grad_norm": 3.1038169860839844,
      "learning_rate": 0.0,
      "loss": 0.6905,
      "step": 9297
    },
    {
      "epoch": 0.8345000897504936,
      "grad_norm": 2.7451059818267822,
      "learning_rate": 0.0,
      "loss": 0.8035,
      "step": 9298
    },
    {
      "epoch": 0.8345898402441213,
      "grad_norm": 3.229158639907837,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 9299
    },
    {
      "epoch": 0.8346795907377491,
      "grad_norm": 2.527869701385498,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 9300
    },
    {
      "epoch": 0.8347693412313768,
      "grad_norm": 2.68137788772583,
      "learning_rate": 0.0,
      "loss": 0.6802,
      "step": 9301
    },
    {
      "epoch": 0.8348590917250045,
      "grad_norm": 3.489283800125122,
      "learning_rate": 0.0,
      "loss": 0.6713,
      "step": 9302
    },
    {
      "epoch": 0.8349488422186322,
      "grad_norm": 3.1880459785461426,
      "learning_rate": 0.0,
      "loss": 0.7511,
      "step": 9303
    },
    {
      "epoch": 0.8350385927122599,
      "grad_norm": 2.580446243286133,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 9304
    },
    {
      "epoch": 0.8351283432058876,
      "grad_norm": 3.4750335216522217,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 9305
    },
    {
      "epoch": 0.8352180936995154,
      "grad_norm": 3.3877484798431396,
      "learning_rate": 0.0,
      "loss": 0.7796,
      "step": 9306
    },
    {
      "epoch": 0.8353078441931431,
      "grad_norm": 2.9676077365875244,
      "learning_rate": 0.0,
      "loss": 0.7677,
      "step": 9307
    },
    {
      "epoch": 0.8353975946867708,
      "grad_norm": 3.4124391078948975,
      "learning_rate": 0.0,
      "loss": 0.6743,
      "step": 9308
    },
    {
      "epoch": 0.8354873451803985,
      "grad_norm": 2.3360209465026855,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 9309
    },
    {
      "epoch": 0.8355770956740262,
      "grad_norm": 2.602430820465088,
      "learning_rate": 0.0,
      "loss": 0.6819,
      "step": 9310
    },
    {
      "epoch": 0.835666846167654,
      "grad_norm": 2.644319772720337,
      "learning_rate": 0.0,
      "loss": 0.6958,
      "step": 9311
    },
    {
      "epoch": 0.8357565966612817,
      "grad_norm": 2.484872817993164,
      "learning_rate": 0.0,
      "loss": 0.7056,
      "step": 9312
    },
    {
      "epoch": 0.8358463471549094,
      "grad_norm": 2.436136245727539,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 9313
    },
    {
      "epoch": 0.835936097648537,
      "grad_norm": 2.3835396766662598,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 9314
    },
    {
      "epoch": 0.8360258481421647,
      "grad_norm": 2.8186283111572266,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 9315
    },
    {
      "epoch": 0.8361155986357925,
      "grad_norm": 2.8719210624694824,
      "learning_rate": 0.0,
      "loss": 0.8065,
      "step": 9316
    },
    {
      "epoch": 0.8362053491294202,
      "grad_norm": 3.0818793773651123,
      "learning_rate": 0.0,
      "loss": 0.7489,
      "step": 9317
    },
    {
      "epoch": 0.8362950996230479,
      "grad_norm": 2.5483875274658203,
      "learning_rate": 0.0,
      "loss": 0.6713,
      "step": 9318
    },
    {
      "epoch": 0.8363848501166756,
      "grad_norm": 2.717550039291382,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 9319
    },
    {
      "epoch": 0.8364746006103033,
      "grad_norm": 2.4690489768981934,
      "learning_rate": 0.0,
      "loss": 0.7631,
      "step": 9320
    },
    {
      "epoch": 0.8365643511039311,
      "grad_norm": 2.763699769973755,
      "learning_rate": 0.0,
      "loss": 0.7192,
      "step": 9321
    },
    {
      "epoch": 0.8366541015975588,
      "grad_norm": 2.881042718887329,
      "learning_rate": 0.0,
      "loss": 0.7502,
      "step": 9322
    },
    {
      "epoch": 0.8367438520911865,
      "grad_norm": 3.267526388168335,
      "learning_rate": 0.0,
      "loss": 0.7869,
      "step": 9323
    },
    {
      "epoch": 0.8368336025848142,
      "grad_norm": 3.2185604572296143,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 9324
    },
    {
      "epoch": 0.8369233530784419,
      "grad_norm": 3.001513957977295,
      "learning_rate": 0.0,
      "loss": 0.6645,
      "step": 9325
    },
    {
      "epoch": 0.8370131035720696,
      "grad_norm": 2.9741666316986084,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 9326
    },
    {
      "epoch": 0.8371028540656974,
      "grad_norm": 2.1002132892608643,
      "learning_rate": 0.0,
      "loss": 0.682,
      "step": 9327
    },
    {
      "epoch": 0.8371926045593251,
      "grad_norm": 2.585198163986206,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 9328
    },
    {
      "epoch": 0.8372823550529528,
      "grad_norm": 2.3853871822357178,
      "learning_rate": 0.0,
      "loss": 0.6542,
      "step": 9329
    },
    {
      "epoch": 0.8373721055465805,
      "grad_norm": 3.4734864234924316,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 9330
    },
    {
      "epoch": 0.8374618560402082,
      "grad_norm": 3.1147873401641846,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 9331
    },
    {
      "epoch": 0.837551606533836,
      "grad_norm": 2.9149439334869385,
      "learning_rate": 0.0,
      "loss": 0.7007,
      "step": 9332
    },
    {
      "epoch": 0.8376413570274637,
      "grad_norm": 3.233760118484497,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 9333
    },
    {
      "epoch": 0.8377311075210914,
      "grad_norm": 2.532660961151123,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 9334
    },
    {
      "epoch": 0.8378208580147191,
      "grad_norm": 2.6338577270507812,
      "learning_rate": 0.0,
      "loss": 0.7245,
      "step": 9335
    },
    {
      "epoch": 0.8379106085083468,
      "grad_norm": 3.3214263916015625,
      "learning_rate": 0.0,
      "loss": 0.6899,
      "step": 9336
    },
    {
      "epoch": 0.8380003590019746,
      "grad_norm": 2.963829755783081,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 9337
    },
    {
      "epoch": 0.8380901094956023,
      "grad_norm": 2.5260255336761475,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 9338
    },
    {
      "epoch": 0.83817985998923,
      "grad_norm": 3.4809367656707764,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 9339
    },
    {
      "epoch": 0.8382696104828576,
      "grad_norm": 2.56687593460083,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 9340
    },
    {
      "epoch": 0.8383593609764853,
      "grad_norm": 2.5187783241271973,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 9341
    },
    {
      "epoch": 0.8384491114701131,
      "grad_norm": 2.427492618560791,
      "learning_rate": 0.0,
      "loss": 0.7526,
      "step": 9342
    },
    {
      "epoch": 0.8385388619637408,
      "grad_norm": 2.691481351852417,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 9343
    },
    {
      "epoch": 0.8386286124573685,
      "grad_norm": 2.5651755332946777,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 9344
    },
    {
      "epoch": 0.8387183629509962,
      "grad_norm": 2.898749828338623,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 9345
    },
    {
      "epoch": 0.8388081134446239,
      "grad_norm": 2.3192954063415527,
      "learning_rate": 0.0,
      "loss": 0.7688,
      "step": 9346
    },
    {
      "epoch": 0.8388978639382516,
      "grad_norm": 2.7008416652679443,
      "learning_rate": 0.0,
      "loss": 0.794,
      "step": 9347
    },
    {
      "epoch": 0.8389876144318794,
      "grad_norm": 3.428917646408081,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 9348
    },
    {
      "epoch": 0.8390773649255071,
      "grad_norm": 2.588142156600952,
      "learning_rate": 0.0,
      "loss": 0.6477,
      "step": 9349
    },
    {
      "epoch": 0.8391671154191348,
      "grad_norm": 2.99691104888916,
      "learning_rate": 0.0,
      "loss": 0.7648,
      "step": 9350
    },
    {
      "epoch": 0.8392568659127625,
      "grad_norm": 3.0366952419281006,
      "learning_rate": 0.0,
      "loss": 0.6811,
      "step": 9351
    },
    {
      "epoch": 0.8393466164063902,
      "grad_norm": 3.8653106689453125,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 9352
    },
    {
      "epoch": 0.839436366900018,
      "grad_norm": 2.325894832611084,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 9353
    },
    {
      "epoch": 0.8395261173936457,
      "grad_norm": 2.8511710166931152,
      "learning_rate": 0.0,
      "loss": 0.7285,
      "step": 9354
    },
    {
      "epoch": 0.8396158678872734,
      "grad_norm": 2.6571943759918213,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 9355
    },
    {
      "epoch": 0.8397056183809011,
      "grad_norm": 2.830000162124634,
      "learning_rate": 0.0,
      "loss": 0.6835,
      "step": 9356
    },
    {
      "epoch": 0.8397953688745288,
      "grad_norm": 2.379809856414795,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 9357
    },
    {
      "epoch": 0.8398851193681566,
      "grad_norm": 3.031477212905884,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 9358
    },
    {
      "epoch": 0.8399748698617843,
      "grad_norm": 3.2813358306884766,
      "learning_rate": 0.0,
      "loss": 0.7101,
      "step": 9359
    },
    {
      "epoch": 0.840064620355412,
      "grad_norm": 2.232210397720337,
      "learning_rate": 0.0,
      "loss": 0.7713,
      "step": 9360
    },
    {
      "epoch": 0.8401543708490397,
      "grad_norm": 3.0446033477783203,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 9361
    },
    {
      "epoch": 0.8402441213426673,
      "grad_norm": 3.1837384700775146,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 9362
    },
    {
      "epoch": 0.8403338718362952,
      "grad_norm": 3.2334861755371094,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 9363
    },
    {
      "epoch": 0.8404236223299228,
      "grad_norm": 3.3267159461975098,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 9364
    },
    {
      "epoch": 0.8405133728235505,
      "grad_norm": 3.5266213417053223,
      "learning_rate": 0.0,
      "loss": 0.6606,
      "step": 9365
    },
    {
      "epoch": 0.8406031233171782,
      "grad_norm": 2.615840435028076,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 9366
    },
    {
      "epoch": 0.8406928738108059,
      "grad_norm": 2.626511335372925,
      "learning_rate": 0.0,
      "loss": 0.6887,
      "step": 9367
    },
    {
      "epoch": 0.8407826243044336,
      "grad_norm": 3.358355760574341,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 9368
    },
    {
      "epoch": 0.8408723747980614,
      "grad_norm": 2.27482008934021,
      "learning_rate": 0.0,
      "loss": 0.7459,
      "step": 9369
    },
    {
      "epoch": 0.8409621252916891,
      "grad_norm": 2.431227922439575,
      "learning_rate": 0.0,
      "loss": 0.7623,
      "step": 9370
    },
    {
      "epoch": 0.8410518757853168,
      "grad_norm": 2.1647157669067383,
      "learning_rate": 0.0,
      "loss": 0.6598,
      "step": 9371
    },
    {
      "epoch": 0.8411416262789445,
      "grad_norm": 2.9425950050354004,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 9372
    },
    {
      "epoch": 0.8412313767725722,
      "grad_norm": 2.425677537918091,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 9373
    },
    {
      "epoch": 0.8413211272662,
      "grad_norm": 6.658446311950684,
      "learning_rate": 0.0,
      "loss": 0.6869,
      "step": 9374
    },
    {
      "epoch": 0.8414108777598277,
      "grad_norm": 2.6160809993743896,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 9375
    },
    {
      "epoch": 0.8415006282534554,
      "grad_norm": 3.033351421356201,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 9376
    },
    {
      "epoch": 0.8415903787470831,
      "grad_norm": 4.538351058959961,
      "learning_rate": 0.0,
      "loss": 0.6865,
      "step": 9377
    },
    {
      "epoch": 0.8416801292407108,
      "grad_norm": 3.402972936630249,
      "learning_rate": 0.0,
      "loss": 0.7322,
      "step": 9378
    },
    {
      "epoch": 0.8417698797343386,
      "grad_norm": 2.7848153114318848,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 9379
    },
    {
      "epoch": 0.8418596302279663,
      "grad_norm": 2.839933156967163,
      "learning_rate": 0.0,
      "loss": 0.6918,
      "step": 9380
    },
    {
      "epoch": 0.841949380721594,
      "grad_norm": 3.3325555324554443,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 9381
    },
    {
      "epoch": 0.8420391312152217,
      "grad_norm": 3.232254981994629,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 9382
    },
    {
      "epoch": 0.8421288817088494,
      "grad_norm": 3.178088665008545,
      "learning_rate": 0.0,
      "loss": 0.7381,
      "step": 9383
    },
    {
      "epoch": 0.8422186322024772,
      "grad_norm": 2.3703677654266357,
      "learning_rate": 0.0,
      "loss": 0.7748,
      "step": 9384
    },
    {
      "epoch": 0.8423083826961049,
      "grad_norm": 2.6096150875091553,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 9385
    },
    {
      "epoch": 0.8423981331897326,
      "grad_norm": 2.895691394805908,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 9386
    },
    {
      "epoch": 0.8424878836833603,
      "grad_norm": 2.776149034500122,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 9387
    },
    {
      "epoch": 0.8425776341769879,
      "grad_norm": 2.7856271266937256,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 9388
    },
    {
      "epoch": 0.8426673846706156,
      "grad_norm": 2.324193000793457,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 9389
    },
    {
      "epoch": 0.8427571351642434,
      "grad_norm": 2.332258462905884,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 9390
    },
    {
      "epoch": 0.8428468856578711,
      "grad_norm": 5.141726493835449,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 9391
    },
    {
      "epoch": 0.8429366361514988,
      "grad_norm": 3.1909289360046387,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 9392
    },
    {
      "epoch": 0.8430263866451265,
      "grad_norm": 2.3417305946350098,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 9393
    },
    {
      "epoch": 0.8431161371387542,
      "grad_norm": 3.147639274597168,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 9394
    },
    {
      "epoch": 0.843205887632382,
      "grad_norm": 3.480738878250122,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 9395
    },
    {
      "epoch": 0.8432956381260097,
      "grad_norm": 3.142143487930298,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 9396
    },
    {
      "epoch": 0.8433853886196374,
      "grad_norm": 2.7538862228393555,
      "learning_rate": 0.0,
      "loss": 0.628,
      "step": 9397
    },
    {
      "epoch": 0.8434751391132651,
      "grad_norm": 3.186856269836426,
      "learning_rate": 0.0,
      "loss": 0.6647,
      "step": 9398
    },
    {
      "epoch": 0.8435648896068928,
      "grad_norm": 2.395681381225586,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 9399
    },
    {
      "epoch": 0.8436546401005206,
      "grad_norm": 3.102872848510742,
      "learning_rate": 0.0,
      "loss": 0.67,
      "step": 9400
    },
    {
      "epoch": 0.8437443905941483,
      "grad_norm": 4.217841148376465,
      "learning_rate": 0.0,
      "loss": 0.6426,
      "step": 9401
    },
    {
      "epoch": 0.843834141087776,
      "grad_norm": 3.6352415084838867,
      "learning_rate": 0.0,
      "loss": 0.6765,
      "step": 9402
    },
    {
      "epoch": 0.8439238915814037,
      "grad_norm": 2.5448923110961914,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 9403
    },
    {
      "epoch": 0.8440136420750314,
      "grad_norm": 2.933864116668701,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 9404
    },
    {
      "epoch": 0.8441033925686591,
      "grad_norm": 1.946449875831604,
      "learning_rate": 0.0,
      "loss": 0.6728,
      "step": 9405
    },
    {
      "epoch": 0.8441931430622869,
      "grad_norm": 2.5551092624664307,
      "learning_rate": 0.0,
      "loss": 0.681,
      "step": 9406
    },
    {
      "epoch": 0.8442828935559146,
      "grad_norm": 2.6595959663391113,
      "learning_rate": 0.0,
      "loss": 0.7616,
      "step": 9407
    },
    {
      "epoch": 0.8443726440495423,
      "grad_norm": 2.610306978225708,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 9408
    },
    {
      "epoch": 0.84446239454317,
      "grad_norm": 2.7738966941833496,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 9409
    },
    {
      "epoch": 0.8445521450367977,
      "grad_norm": 2.5940682888031006,
      "learning_rate": 0.0,
      "loss": 0.7751,
      "step": 9410
    },
    {
      "epoch": 0.8446418955304255,
      "grad_norm": 2.2473976612091064,
      "learning_rate": 0.0,
      "loss": 0.6553,
      "step": 9411
    },
    {
      "epoch": 0.8447316460240532,
      "grad_norm": 2.7500479221343994,
      "learning_rate": 0.0,
      "loss": 0.7887,
      "step": 9412
    },
    {
      "epoch": 0.8448213965176808,
      "grad_norm": 2.45684814453125,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 9413
    },
    {
      "epoch": 0.8449111470113085,
      "grad_norm": 2.3408360481262207,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 9414
    },
    {
      "epoch": 0.8450008975049362,
      "grad_norm": 2.210996627807617,
      "learning_rate": 0.0,
      "loss": 0.6518,
      "step": 9415
    },
    {
      "epoch": 0.845090647998564,
      "grad_norm": 3.0969231128692627,
      "learning_rate": 0.0,
      "loss": 0.6528,
      "step": 9416
    },
    {
      "epoch": 0.8451803984921917,
      "grad_norm": 2.3137896060943604,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 9417
    },
    {
      "epoch": 0.8452701489858194,
      "grad_norm": 2.5166304111480713,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 9418
    },
    {
      "epoch": 0.8453598994794471,
      "grad_norm": 7.296230792999268,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 9419
    },
    {
      "epoch": 0.8454496499730748,
      "grad_norm": 3.7258899211883545,
      "learning_rate": 0.0,
      "loss": 0.6742,
      "step": 9420
    },
    {
      "epoch": 0.8455394004667026,
      "grad_norm": 3.4702682495117188,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 9421
    },
    {
      "epoch": 0.8456291509603303,
      "grad_norm": 2.79252552986145,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 9422
    },
    {
      "epoch": 0.845718901453958,
      "grad_norm": 2.6780827045440674,
      "learning_rate": 0.0,
      "loss": 0.8067,
      "step": 9423
    },
    {
      "epoch": 0.8458086519475857,
      "grad_norm": 3.170642375946045,
      "learning_rate": 0.0,
      "loss": 0.6834,
      "step": 9424
    },
    {
      "epoch": 0.8458984024412134,
      "grad_norm": 3.0589003562927246,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 9425
    },
    {
      "epoch": 0.8459881529348411,
      "grad_norm": 3.5440263748168945,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 9426
    },
    {
      "epoch": 0.8460779034284689,
      "grad_norm": 2.7009329795837402,
      "learning_rate": 0.0,
      "loss": 0.6769,
      "step": 9427
    },
    {
      "epoch": 0.8461676539220966,
      "grad_norm": 3.2314767837524414,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 9428
    },
    {
      "epoch": 0.8462574044157243,
      "grad_norm": 3.8788230419158936,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 9429
    },
    {
      "epoch": 0.846347154909352,
      "grad_norm": 3.9446802139282227,
      "learning_rate": 0.0,
      "loss": 0.6787,
      "step": 9430
    },
    {
      "epoch": 0.8464369054029797,
      "grad_norm": 3.3862833976745605,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 9431
    },
    {
      "epoch": 0.8465266558966075,
      "grad_norm": 2.366297483444214,
      "learning_rate": 0.0,
      "loss": 0.6104,
      "step": 9432
    },
    {
      "epoch": 0.8466164063902352,
      "grad_norm": 2.696733236312866,
      "learning_rate": 0.0,
      "loss": 0.6654,
      "step": 9433
    },
    {
      "epoch": 0.8467061568838629,
      "grad_norm": 3.726750135421753,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 9434
    },
    {
      "epoch": 0.8467959073774906,
      "grad_norm": 3.7185590267181396,
      "learning_rate": 0.0,
      "loss": 0.6775,
      "step": 9435
    },
    {
      "epoch": 0.8468856578711182,
      "grad_norm": 4.933427810668945,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 9436
    },
    {
      "epoch": 0.846975408364746,
      "grad_norm": 3.008574962615967,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 9437
    },
    {
      "epoch": 0.8470651588583737,
      "grad_norm": 2.5304830074310303,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 9438
    },
    {
      "epoch": 0.8471549093520014,
      "grad_norm": 2.9645068645477295,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 9439
    },
    {
      "epoch": 0.8472446598456291,
      "grad_norm": 10.38257122039795,
      "learning_rate": 0.0,
      "loss": 0.6788,
      "step": 9440
    },
    {
      "epoch": 0.8473344103392568,
      "grad_norm": 2.723370313644409,
      "learning_rate": 0.0,
      "loss": 0.7131,
      "step": 9441
    },
    {
      "epoch": 0.8474241608328846,
      "grad_norm": 3.3322675228118896,
      "learning_rate": 0.0,
      "loss": 0.6747,
      "step": 9442
    },
    {
      "epoch": 0.8475139113265123,
      "grad_norm": 2.424180507659912,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 9443
    },
    {
      "epoch": 0.84760366182014,
      "grad_norm": 2.37355899810791,
      "learning_rate": 0.0,
      "loss": 0.7113,
      "step": 9444
    },
    {
      "epoch": 0.8476934123137677,
      "grad_norm": 2.908698320388794,
      "learning_rate": 0.0,
      "loss": 0.6701,
      "step": 9445
    },
    {
      "epoch": 0.8477831628073954,
      "grad_norm": 2.226691246032715,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 9446
    },
    {
      "epoch": 0.8478729133010231,
      "grad_norm": 2.483973503112793,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 9447
    },
    {
      "epoch": 0.8479626637946509,
      "grad_norm": 3.4965953826904297,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 9448
    },
    {
      "epoch": 0.8480524142882786,
      "grad_norm": 2.7991230487823486,
      "learning_rate": 0.0,
      "loss": 0.6949,
      "step": 9449
    },
    {
      "epoch": 0.8481421647819063,
      "grad_norm": 1.862309455871582,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 9450
    },
    {
      "epoch": 0.848231915275534,
      "grad_norm": 2.4079883098602295,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 9451
    },
    {
      "epoch": 0.8483216657691617,
      "grad_norm": 2.794114828109741,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 9452
    },
    {
      "epoch": 0.8484114162627895,
      "grad_norm": 3.3734068870544434,
      "learning_rate": 0.0,
      "loss": 0.7841,
      "step": 9453
    },
    {
      "epoch": 0.8485011667564172,
      "grad_norm": 2.5458908081054688,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 9454
    },
    {
      "epoch": 0.8485909172500449,
      "grad_norm": 2.6477560997009277,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 9455
    },
    {
      "epoch": 0.8486806677436726,
      "grad_norm": 2.2350306510925293,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 9456
    },
    {
      "epoch": 0.8487704182373003,
      "grad_norm": 3.0588226318359375,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 9457
    },
    {
      "epoch": 0.8488601687309281,
      "grad_norm": 2.7108728885650635,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 9458
    },
    {
      "epoch": 0.8489499192245558,
      "grad_norm": 2.3727774620056152,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 9459
    },
    {
      "epoch": 0.8490396697181835,
      "grad_norm": 2.3982300758361816,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 9460
    },
    {
      "epoch": 0.8491294202118111,
      "grad_norm": 2.531244993209839,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 9461
    },
    {
      "epoch": 0.8492191707054388,
      "grad_norm": 2.5841898918151855,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 9462
    },
    {
      "epoch": 0.8493089211990666,
      "grad_norm": 2.217494010925293,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 9463
    },
    {
      "epoch": 0.8493986716926943,
      "grad_norm": 2.1798155307769775,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 9464
    },
    {
      "epoch": 0.849488422186322,
      "grad_norm": 2.378091335296631,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 9465
    },
    {
      "epoch": 0.8495781726799497,
      "grad_norm": 2.9492037296295166,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 9466
    },
    {
      "epoch": 0.8496679231735774,
      "grad_norm": 3.6150460243225098,
      "learning_rate": 0.0,
      "loss": 0.6675,
      "step": 9467
    },
    {
      "epoch": 0.8497576736672051,
      "grad_norm": 3.5524470806121826,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 9468
    },
    {
      "epoch": 0.8498474241608329,
      "grad_norm": 2.579584836959839,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 9469
    },
    {
      "epoch": 0.8499371746544606,
      "grad_norm": 2.8936820030212402,
      "learning_rate": 0.0,
      "loss": 0.8087,
      "step": 9470
    },
    {
      "epoch": 0.8500269251480883,
      "grad_norm": 2.7890498638153076,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 9471
    },
    {
      "epoch": 0.850116675641716,
      "grad_norm": 3.5676591396331787,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 9472
    },
    {
      "epoch": 0.8502064261353437,
      "grad_norm": 3.0838797092437744,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 9473
    },
    {
      "epoch": 0.8502961766289715,
      "grad_norm": 2.8250248432159424,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 9474
    },
    {
      "epoch": 0.8503859271225992,
      "grad_norm": 2.9250590801239014,
      "learning_rate": 0.0,
      "loss": 0.7235,
      "step": 9475
    },
    {
      "epoch": 0.8504756776162269,
      "grad_norm": 3.1265408992767334,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 9476
    },
    {
      "epoch": 0.8505654281098546,
      "grad_norm": 2.040381908416748,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 9477
    },
    {
      "epoch": 0.8506551786034823,
      "grad_norm": 3.377274751663208,
      "learning_rate": 0.0,
      "loss": 0.6788,
      "step": 9478
    },
    {
      "epoch": 0.8507449290971101,
      "grad_norm": 3.0809524059295654,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 9479
    },
    {
      "epoch": 0.8508346795907378,
      "grad_norm": 3.1666834354400635,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 9480
    },
    {
      "epoch": 0.8509244300843655,
      "grad_norm": 2.86509370803833,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 9481
    },
    {
      "epoch": 0.8510141805779932,
      "grad_norm": 2.8967416286468506,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 9482
    },
    {
      "epoch": 0.8511039310716209,
      "grad_norm": 2.5679867267608643,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 9483
    },
    {
      "epoch": 0.8511936815652487,
      "grad_norm": 3.133267879486084,
      "learning_rate": 0.0,
      "loss": 0.664,
      "step": 9484
    },
    {
      "epoch": 0.8512834320588764,
      "grad_norm": 2.159682273864746,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 9485
    },
    {
      "epoch": 0.851373182552504,
      "grad_norm": 3.0723114013671875,
      "learning_rate": 0.0,
      "loss": 0.6683,
      "step": 9486
    },
    {
      "epoch": 0.8514629330461317,
      "grad_norm": 2.9534740447998047,
      "learning_rate": 0.0,
      "loss": 0.6888,
      "step": 9487
    },
    {
      "epoch": 0.8515526835397594,
      "grad_norm": 2.759340286254883,
      "learning_rate": 0.0,
      "loss": 0.7313,
      "step": 9488
    },
    {
      "epoch": 0.8516424340333871,
      "grad_norm": 3.169285535812378,
      "learning_rate": 0.0,
      "loss": 0.6175,
      "step": 9489
    },
    {
      "epoch": 0.8517321845270149,
      "grad_norm": 2.622105121612549,
      "learning_rate": 0.0,
      "loss": 0.7662,
      "step": 9490
    },
    {
      "epoch": 0.8518219350206426,
      "grad_norm": 3.414710283279419,
      "learning_rate": 0.0,
      "loss": 0.7881,
      "step": 9491
    },
    {
      "epoch": 0.8519116855142703,
      "grad_norm": 2.825711250305176,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 9492
    },
    {
      "epoch": 0.852001436007898,
      "grad_norm": 2.5792925357818604,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 9493
    },
    {
      "epoch": 0.8520911865015257,
      "grad_norm": 3.2336204051971436,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 9494
    },
    {
      "epoch": 0.8521809369951535,
      "grad_norm": 2.4071829319000244,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 9495
    },
    {
      "epoch": 0.8522706874887812,
      "grad_norm": 2.941019058227539,
      "learning_rate": 0.0,
      "loss": 0.6977,
      "step": 9496
    },
    {
      "epoch": 0.8523604379824089,
      "grad_norm": 3.007566452026367,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 9497
    },
    {
      "epoch": 0.8524501884760366,
      "grad_norm": 2.37253475189209,
      "learning_rate": 0.0,
      "loss": 0.669,
      "step": 9498
    },
    {
      "epoch": 0.8525399389696643,
      "grad_norm": 3.3543968200683594,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 9499
    },
    {
      "epoch": 0.8526296894632921,
      "grad_norm": 9.064366340637207,
      "learning_rate": 0.0,
      "loss": 0.6522,
      "step": 9500
    },
    {
      "epoch": 0.8527194399569198,
      "grad_norm": 5.212370872497559,
      "learning_rate": 0.0,
      "loss": 0.6852,
      "step": 9501
    },
    {
      "epoch": 0.8528091904505475,
      "grad_norm": 3.1406991481781006,
      "learning_rate": 0.0,
      "loss": 0.679,
      "step": 9502
    },
    {
      "epoch": 0.8528989409441752,
      "grad_norm": 2.7683162689208984,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 9503
    },
    {
      "epoch": 0.8529886914378029,
      "grad_norm": 2.8621950149536133,
      "learning_rate": 0.0,
      "loss": 0.6287,
      "step": 9504
    },
    {
      "epoch": 0.8530784419314307,
      "grad_norm": 7.612265110015869,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 9505
    },
    {
      "epoch": 0.8531681924250584,
      "grad_norm": 3.3465957641601562,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 9506
    },
    {
      "epoch": 0.8532579429186861,
      "grad_norm": 2.7726175785064697,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 9507
    },
    {
      "epoch": 0.8533476934123138,
      "grad_norm": 2.610511541366577,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 9508
    },
    {
      "epoch": 0.8534374439059415,
      "grad_norm": 3.104344367980957,
      "learning_rate": 0.0,
      "loss": 0.6659,
      "step": 9509
    },
    {
      "epoch": 0.8535271943995691,
      "grad_norm": 2.7100086212158203,
      "learning_rate": 0.0,
      "loss": 0.6633,
      "step": 9510
    },
    {
      "epoch": 0.853616944893197,
      "grad_norm": 3.3385891914367676,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 9511
    },
    {
      "epoch": 0.8537066953868246,
      "grad_norm": 4.150722026824951,
      "learning_rate": 0.0,
      "loss": 0.6736,
      "step": 9512
    },
    {
      "epoch": 0.8537964458804523,
      "grad_norm": 5.9469828605651855,
      "learning_rate": 0.0,
      "loss": 0.6783,
      "step": 9513
    },
    {
      "epoch": 0.85388619637408,
      "grad_norm": 3.218552350997925,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 9514
    },
    {
      "epoch": 0.8539759468677077,
      "grad_norm": 4.446848392486572,
      "learning_rate": 0.0,
      "loss": 0.7418,
      "step": 9515
    },
    {
      "epoch": 0.8540656973613355,
      "grad_norm": 3.776156425476074,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 9516
    },
    {
      "epoch": 0.8541554478549632,
      "grad_norm": 3.105928897857666,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 9517
    },
    {
      "epoch": 0.8542451983485909,
      "grad_norm": 2.2205183506011963,
      "learning_rate": 0.0,
      "loss": 0.6527,
      "step": 9518
    },
    {
      "epoch": 0.8543349488422186,
      "grad_norm": 2.3476805686950684,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 9519
    },
    {
      "epoch": 0.8544246993358463,
      "grad_norm": 3.342224597930908,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 9520
    },
    {
      "epoch": 0.8545144498294741,
      "grad_norm": 2.5182483196258545,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 9521
    },
    {
      "epoch": 0.8546042003231018,
      "grad_norm": 2.4830076694488525,
      "learning_rate": 0.0,
      "loss": 0.7465,
      "step": 9522
    },
    {
      "epoch": 0.8546939508167295,
      "grad_norm": 3.2287981510162354,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 9523
    },
    {
      "epoch": 0.8547837013103572,
      "grad_norm": 3.906445264816284,
      "learning_rate": 0.0,
      "loss": 0.6735,
      "step": 9524
    },
    {
      "epoch": 0.8548734518039849,
      "grad_norm": 2.9147391319274902,
      "learning_rate": 0.0,
      "loss": 0.7024,
      "step": 9525
    },
    {
      "epoch": 0.8549632022976127,
      "grad_norm": 3.131052017211914,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 9526
    },
    {
      "epoch": 0.8550529527912404,
      "grad_norm": 2.475672721862793,
      "learning_rate": 0.0,
      "loss": 0.6172,
      "step": 9527
    },
    {
      "epoch": 0.8551427032848681,
      "grad_norm": 3.624664783477783,
      "learning_rate": 0.0,
      "loss": 0.7508,
      "step": 9528
    },
    {
      "epoch": 0.8552324537784958,
      "grad_norm": 3.1075830459594727,
      "learning_rate": 0.0,
      "loss": 0.8161,
      "step": 9529
    },
    {
      "epoch": 0.8553222042721235,
      "grad_norm": 2.3799502849578857,
      "learning_rate": 0.0,
      "loss": 0.7832,
      "step": 9530
    },
    {
      "epoch": 0.8554119547657512,
      "grad_norm": 2.364079475402832,
      "learning_rate": 0.0,
      "loss": 0.6933,
      "step": 9531
    },
    {
      "epoch": 0.855501705259379,
      "grad_norm": 3.636788845062256,
      "learning_rate": 0.0,
      "loss": 0.6907,
      "step": 9532
    },
    {
      "epoch": 0.8555914557530067,
      "grad_norm": 3.080605983734131,
      "learning_rate": 0.0,
      "loss": 0.6656,
      "step": 9533
    },
    {
      "epoch": 0.8556812062466344,
      "grad_norm": 3.277766704559326,
      "learning_rate": 0.0,
      "loss": 0.7044,
      "step": 9534
    },
    {
      "epoch": 0.855770956740262,
      "grad_norm": 3.866159677505493,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 9535
    },
    {
      "epoch": 0.8558607072338897,
      "grad_norm": 5.013427257537842,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 9536
    },
    {
      "epoch": 0.8559504577275175,
      "grad_norm": 3.597883701324463,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 9537
    },
    {
      "epoch": 0.8560402082211452,
      "grad_norm": 3.192354917526245,
      "learning_rate": 0.0,
      "loss": 0.6751,
      "step": 9538
    },
    {
      "epoch": 0.8561299587147729,
      "grad_norm": 2.6561124324798584,
      "learning_rate": 0.0,
      "loss": 0.6896,
      "step": 9539
    },
    {
      "epoch": 0.8562197092084006,
      "grad_norm": 2.3138580322265625,
      "learning_rate": 0.0,
      "loss": 0.6576,
      "step": 9540
    },
    {
      "epoch": 0.8563094597020283,
      "grad_norm": 3.964700222015381,
      "learning_rate": 0.0,
      "loss": 0.6618,
      "step": 9541
    },
    {
      "epoch": 0.8563992101956561,
      "grad_norm": 4.911686897277832,
      "learning_rate": 0.0,
      "loss": 0.661,
      "step": 9542
    },
    {
      "epoch": 0.8564889606892838,
      "grad_norm": 3.874199151992798,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 9543
    },
    {
      "epoch": 0.8565787111829115,
      "grad_norm": 2.2704100608825684,
      "learning_rate": 0.0,
      "loss": 0.655,
      "step": 9544
    },
    {
      "epoch": 0.8566684616765392,
      "grad_norm": 4.692509174346924,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 9545
    },
    {
      "epoch": 0.8567582121701669,
      "grad_norm": 2.895695924758911,
      "learning_rate": 0.0,
      "loss": 0.6972,
      "step": 9546
    },
    {
      "epoch": 0.8568479626637947,
      "grad_norm": 3.464892625808716,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 9547
    },
    {
      "epoch": 0.8569377131574224,
      "grad_norm": 2.8310294151306152,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 9548
    },
    {
      "epoch": 0.8570274636510501,
      "grad_norm": 4.768579959869385,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 9549
    },
    {
      "epoch": 0.8571172141446778,
      "grad_norm": 2.6976475715637207,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 9550
    },
    {
      "epoch": 0.8572069646383055,
      "grad_norm": 2.2107417583465576,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 9551
    },
    {
      "epoch": 0.8572967151319332,
      "grad_norm": 2.208052396774292,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 9552
    },
    {
      "epoch": 0.857386465625561,
      "grad_norm": 3.711430311203003,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 9553
    },
    {
      "epoch": 0.8574762161191887,
      "grad_norm": 2.4059417247772217,
      "learning_rate": 0.0,
      "loss": 0.6825,
      "step": 9554
    },
    {
      "epoch": 0.8575659666128164,
      "grad_norm": 2.1526472568511963,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 9555
    },
    {
      "epoch": 0.8576557171064441,
      "grad_norm": 3.6501991748809814,
      "learning_rate": 0.0,
      "loss": 0.6792,
      "step": 9556
    },
    {
      "epoch": 0.8577454676000718,
      "grad_norm": 2.8747761249542236,
      "learning_rate": 0.0,
      "loss": 0.652,
      "step": 9557
    },
    {
      "epoch": 0.8578352180936996,
      "grad_norm": 3.314131259918213,
      "learning_rate": 0.0,
      "loss": 0.6703,
      "step": 9558
    },
    {
      "epoch": 0.8579249685873273,
      "grad_norm": 2.9698543548583984,
      "learning_rate": 0.0,
      "loss": 0.6812,
      "step": 9559
    },
    {
      "epoch": 0.858014719080955,
      "grad_norm": 3.1344058513641357,
      "learning_rate": 0.0,
      "loss": 0.7747,
      "step": 9560
    },
    {
      "epoch": 0.8581044695745826,
      "grad_norm": 3.2361209392547607,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 9561
    },
    {
      "epoch": 0.8581942200682103,
      "grad_norm": 2.6001577377319336,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 9562
    },
    {
      "epoch": 0.8582839705618381,
      "grad_norm": 5.265576362609863,
      "learning_rate": 0.0,
      "loss": 0.7346,
      "step": 9563
    },
    {
      "epoch": 0.8583737210554658,
      "grad_norm": 3.118046998977661,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 9564
    },
    {
      "epoch": 0.8584634715490935,
      "grad_norm": 2.2247464656829834,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 9565
    },
    {
      "epoch": 0.8585532220427212,
      "grad_norm": 2.6367948055267334,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 9566
    },
    {
      "epoch": 0.8586429725363489,
      "grad_norm": 2.923389196395874,
      "learning_rate": 0.0,
      "loss": 0.6947,
      "step": 9567
    },
    {
      "epoch": 0.8587327230299767,
      "grad_norm": 3.5681002140045166,
      "learning_rate": 0.0,
      "loss": 0.7246,
      "step": 9568
    },
    {
      "epoch": 0.8588224735236044,
      "grad_norm": 2.840496778488159,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 9569
    },
    {
      "epoch": 0.8589122240172321,
      "grad_norm": 2.839163303375244,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 9570
    },
    {
      "epoch": 0.8590019745108598,
      "grad_norm": 2.6100990772247314,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 9571
    },
    {
      "epoch": 0.8590917250044875,
      "grad_norm": 3.4433321952819824,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 9572
    },
    {
      "epoch": 0.8591814754981152,
      "grad_norm": 3.099733352661133,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 9573
    },
    {
      "epoch": 0.859271225991743,
      "grad_norm": 3.3619678020477295,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 9574
    },
    {
      "epoch": 0.8593609764853707,
      "grad_norm": 2.509363889694214,
      "learning_rate": 0.0,
      "loss": 0.7887,
      "step": 9575
    },
    {
      "epoch": 0.8594507269789984,
      "grad_norm": 3.387143850326538,
      "learning_rate": 0.0,
      "loss": 0.6841,
      "step": 9576
    },
    {
      "epoch": 0.8595404774726261,
      "grad_norm": 3.4922213554382324,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 9577
    },
    {
      "epoch": 0.8596302279662538,
      "grad_norm": 3.092576503753662,
      "learning_rate": 0.0,
      "loss": 0.7086,
      "step": 9578
    },
    {
      "epoch": 0.8597199784598816,
      "grad_norm": 3.027127265930176,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 9579
    },
    {
      "epoch": 0.8598097289535093,
      "grad_norm": 3.237229824066162,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 9580
    },
    {
      "epoch": 0.859899479447137,
      "grad_norm": 3.178365468978882,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 9581
    },
    {
      "epoch": 0.8599892299407647,
      "grad_norm": 2.915666341781616,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 9582
    },
    {
      "epoch": 0.8600789804343923,
      "grad_norm": 2.4330408573150635,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 9583
    },
    {
      "epoch": 0.8601687309280202,
      "grad_norm": 2.3735625743865967,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 9584
    },
    {
      "epoch": 0.8602584814216478,
      "grad_norm": 2.3518266677856445,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 9585
    },
    {
      "epoch": 0.8603482319152755,
      "grad_norm": 3.2171919345855713,
      "learning_rate": 0.0,
      "loss": 0.6491,
      "step": 9586
    },
    {
      "epoch": 0.8604379824089032,
      "grad_norm": 2.9130618572235107,
      "learning_rate": 0.0,
      "loss": 0.6783,
      "step": 9587
    },
    {
      "epoch": 0.8605277329025309,
      "grad_norm": 2.5644373893737793,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 9588
    },
    {
      "epoch": 0.8606174833961587,
      "grad_norm": 4.129367828369141,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 9589
    },
    {
      "epoch": 0.8607072338897864,
      "grad_norm": 2.8404011726379395,
      "learning_rate": 0.0,
      "loss": 0.6836,
      "step": 9590
    },
    {
      "epoch": 0.8607969843834141,
      "grad_norm": 2.5957062244415283,
      "learning_rate": 0.0,
      "loss": 0.699,
      "step": 9591
    },
    {
      "epoch": 0.8608867348770418,
      "grad_norm": 3.288943290710449,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 9592
    },
    {
      "epoch": 0.8609764853706695,
      "grad_norm": 3.1379079818725586,
      "learning_rate": 0.0,
      "loss": 0.7106,
      "step": 9593
    },
    {
      "epoch": 0.8610662358642972,
      "grad_norm": 3.6446986198425293,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 9594
    },
    {
      "epoch": 0.861155986357925,
      "grad_norm": 2.6874642372131348,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 9595
    },
    {
      "epoch": 0.8612457368515527,
      "grad_norm": 2.7802212238311768,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 9596
    },
    {
      "epoch": 0.8613354873451804,
      "grad_norm": 3.177882671356201,
      "learning_rate": 0.0,
      "loss": 0.7723,
      "step": 9597
    },
    {
      "epoch": 0.8614252378388081,
      "grad_norm": 3.2696712017059326,
      "learning_rate": 0.0,
      "loss": 0.7175,
      "step": 9598
    },
    {
      "epoch": 0.8615149883324358,
      "grad_norm": 2.585197925567627,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 9599
    },
    {
      "epoch": 0.8616047388260636,
      "grad_norm": 2.8384907245635986,
      "learning_rate": 0.0,
      "loss": 0.8052,
      "step": 9600
    },
    {
      "epoch": 0.8616944893196913,
      "grad_norm": 2.847116231918335,
      "learning_rate": 0.0,
      "loss": 0.6652,
      "step": 9601
    },
    {
      "epoch": 0.861784239813319,
      "grad_norm": 2.751081943511963,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 9602
    },
    {
      "epoch": 0.8618739903069467,
      "grad_norm": 2.5802559852600098,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 9603
    },
    {
      "epoch": 0.8619637408005744,
      "grad_norm": 2.3225183486938477,
      "learning_rate": 0.0,
      "loss": 0.6648,
      "step": 9604
    },
    {
      "epoch": 0.8620534912942022,
      "grad_norm": 2.9852569103240967,
      "learning_rate": 0.0,
      "loss": 0.689,
      "step": 9605
    },
    {
      "epoch": 0.8621432417878299,
      "grad_norm": 2.620569944381714,
      "learning_rate": 0.0,
      "loss": 0.7494,
      "step": 9606
    },
    {
      "epoch": 0.8622329922814576,
      "grad_norm": 2.020899772644043,
      "learning_rate": 0.0,
      "loss": 0.6574,
      "step": 9607
    },
    {
      "epoch": 0.8623227427750852,
      "grad_norm": 2.8235113620758057,
      "learning_rate": 0.0,
      "loss": 0.6833,
      "step": 9608
    },
    {
      "epoch": 0.8624124932687129,
      "grad_norm": 2.745820999145508,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 9609
    },
    {
      "epoch": 0.8625022437623407,
      "grad_norm": 2.6258223056793213,
      "learning_rate": 0.0,
      "loss": 0.749,
      "step": 9610
    },
    {
      "epoch": 0.8625919942559684,
      "grad_norm": 2.5925843715667725,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 9611
    },
    {
      "epoch": 0.8626817447495961,
      "grad_norm": 3.3632020950317383,
      "learning_rate": 0.0,
      "loss": 0.7552,
      "step": 9612
    },
    {
      "epoch": 0.8627714952432238,
      "grad_norm": 3.430074691772461,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 9613
    },
    {
      "epoch": 0.8628612457368515,
      "grad_norm": 2.407486915588379,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 9614
    },
    {
      "epoch": 0.8629509962304792,
      "grad_norm": 3.10740065574646,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 9615
    },
    {
      "epoch": 0.863040746724107,
      "grad_norm": 3.4670674800872803,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 9616
    },
    {
      "epoch": 0.8631304972177347,
      "grad_norm": 2.4070582389831543,
      "learning_rate": 0.0,
      "loss": 0.6713,
      "step": 9617
    },
    {
      "epoch": 0.8632202477113624,
      "grad_norm": 3.4922146797180176,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 9618
    },
    {
      "epoch": 0.8633099982049901,
      "grad_norm": 3.9209823608398438,
      "learning_rate": 0.0,
      "loss": 0.6807,
      "step": 9619
    },
    {
      "epoch": 0.8633997486986178,
      "grad_norm": 2.7589643001556396,
      "learning_rate": 0.0,
      "loss": 0.7072,
      "step": 9620
    },
    {
      "epoch": 0.8634894991922456,
      "grad_norm": 2.7902071475982666,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 9621
    },
    {
      "epoch": 0.8635792496858733,
      "grad_norm": 3.0441360473632812,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 9622
    },
    {
      "epoch": 0.863669000179501,
      "grad_norm": 3.3407888412475586,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 9623
    },
    {
      "epoch": 0.8637587506731287,
      "grad_norm": 2.6524810791015625,
      "learning_rate": 0.0,
      "loss": 0.6911,
      "step": 9624
    },
    {
      "epoch": 0.8638485011667564,
      "grad_norm": 2.5660312175750732,
      "learning_rate": 0.0,
      "loss": 0.7387,
      "step": 9625
    },
    {
      "epoch": 0.8639382516603842,
      "grad_norm": 2.583221673965454,
      "learning_rate": 0.0,
      "loss": 0.6843,
      "step": 9626
    },
    {
      "epoch": 0.8640280021540119,
      "grad_norm": 2.8123857975006104,
      "learning_rate": 0.0,
      "loss": 0.6922,
      "step": 9627
    },
    {
      "epoch": 0.8641177526476396,
      "grad_norm": 2.290881395339966,
      "learning_rate": 0.0,
      "loss": 0.6757,
      "step": 9628
    },
    {
      "epoch": 0.8642075031412673,
      "grad_norm": 2.7960455417633057,
      "learning_rate": 0.0,
      "loss": 0.6674,
      "step": 9629
    },
    {
      "epoch": 0.864297253634895,
      "grad_norm": 2.3212661743164062,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 9630
    },
    {
      "epoch": 0.8643870041285228,
      "grad_norm": 3.0041677951812744,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 9631
    },
    {
      "epoch": 0.8644767546221505,
      "grad_norm": 3.060112714767456,
      "learning_rate": 0.0,
      "loss": 0.7398,
      "step": 9632
    },
    {
      "epoch": 0.8645665051157781,
      "grad_norm": 2.5880179405212402,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 9633
    },
    {
      "epoch": 0.8646562556094058,
      "grad_norm": 2.7021923065185547,
      "learning_rate": 0.0,
      "loss": 0.6332,
      "step": 9634
    },
    {
      "epoch": 0.8647460061030335,
      "grad_norm": 49.046634674072266,
      "learning_rate": 0.0,
      "loss": 0.7431,
      "step": 9635
    },
    {
      "epoch": 0.8648357565966612,
      "grad_norm": 3.441676616668701,
      "learning_rate": 0.0,
      "loss": 0.7207,
      "step": 9636
    },
    {
      "epoch": 0.864925507090289,
      "grad_norm": 2.423152446746826,
      "learning_rate": 0.0,
      "loss": 0.7379,
      "step": 9637
    },
    {
      "epoch": 0.8650152575839167,
      "grad_norm": 2.8338797092437744,
      "learning_rate": 0.0,
      "loss": 0.7079,
      "step": 9638
    },
    {
      "epoch": 0.8651050080775444,
      "grad_norm": 3.0253043174743652,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 9639
    },
    {
      "epoch": 0.8651947585711721,
      "grad_norm": 2.9311134815216064,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 9640
    },
    {
      "epoch": 0.8652845090647998,
      "grad_norm": 2.660963773727417,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 9641
    },
    {
      "epoch": 0.8653742595584276,
      "grad_norm": 3.064556360244751,
      "learning_rate": 0.0,
      "loss": 0.7057,
      "step": 9642
    },
    {
      "epoch": 0.8654640100520553,
      "grad_norm": 2.993267059326172,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 9643
    },
    {
      "epoch": 0.865553760545683,
      "grad_norm": 3.19938588142395,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 9644
    },
    {
      "epoch": 0.8656435110393107,
      "grad_norm": 2.5618104934692383,
      "learning_rate": 0.0,
      "loss": 0.7003,
      "step": 9645
    },
    {
      "epoch": 0.8657332615329384,
      "grad_norm": 3.7555787563323975,
      "learning_rate": 0.0,
      "loss": 0.6727,
      "step": 9646
    },
    {
      "epoch": 0.8658230120265662,
      "grad_norm": 2.9835972785949707,
      "learning_rate": 0.0,
      "loss": 0.7699,
      "step": 9647
    },
    {
      "epoch": 0.8659127625201939,
      "grad_norm": 2.3324503898620605,
      "learning_rate": 0.0,
      "loss": 0.6684,
      "step": 9648
    },
    {
      "epoch": 0.8660025130138216,
      "grad_norm": 2.7420146465301514,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 9649
    },
    {
      "epoch": 0.8660922635074493,
      "grad_norm": 3.2382545471191406,
      "learning_rate": 0.0,
      "loss": 0.7382,
      "step": 9650
    },
    {
      "epoch": 0.866182014001077,
      "grad_norm": 4.214287757873535,
      "learning_rate": 0.0,
      "loss": 0.6803,
      "step": 9651
    },
    {
      "epoch": 0.8662717644947047,
      "grad_norm": 2.8238189220428467,
      "learning_rate": 0.0,
      "loss": 0.7225,
      "step": 9652
    },
    {
      "epoch": 0.8663615149883325,
      "grad_norm": 2.3752472400665283,
      "learning_rate": 0.0,
      "loss": 0.685,
      "step": 9653
    },
    {
      "epoch": 0.8664512654819602,
      "grad_norm": 2.5678460597991943,
      "learning_rate": 0.0,
      "loss": 0.7072,
      "step": 9654
    },
    {
      "epoch": 0.8665410159755879,
      "grad_norm": 2.9167168140411377,
      "learning_rate": 0.0,
      "loss": 0.6817,
      "step": 9655
    },
    {
      "epoch": 0.8666307664692156,
      "grad_norm": 2.725775957107544,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 9656
    },
    {
      "epoch": 0.8667205169628432,
      "grad_norm": 2.3337388038635254,
      "learning_rate": 0.0,
      "loss": 0.7361,
      "step": 9657
    },
    {
      "epoch": 0.866810267456471,
      "grad_norm": 2.8758795261383057,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 9658
    },
    {
      "epoch": 0.8669000179500987,
      "grad_norm": 5.208138465881348,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 9659
    },
    {
      "epoch": 0.8669897684437264,
      "grad_norm": 2.7823398113250732,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 9660
    },
    {
      "epoch": 0.8670795189373541,
      "grad_norm": 2.575650691986084,
      "learning_rate": 0.0,
      "loss": 0.6528,
      "step": 9661
    },
    {
      "epoch": 0.8671692694309818,
      "grad_norm": 2.544926881790161,
      "learning_rate": 0.0,
      "loss": 0.6909,
      "step": 9662
    },
    {
      "epoch": 0.8672590199246096,
      "grad_norm": 2.584679126739502,
      "learning_rate": 0.0,
      "loss": 0.6972,
      "step": 9663
    },
    {
      "epoch": 0.8673487704182373,
      "grad_norm": 3.436619758605957,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 9664
    },
    {
      "epoch": 0.867438520911865,
      "grad_norm": 3.3902242183685303,
      "learning_rate": 0.0,
      "loss": 0.6538,
      "step": 9665
    },
    {
      "epoch": 0.8675282714054927,
      "grad_norm": 4.159079074859619,
      "learning_rate": 0.0,
      "loss": 0.6852,
      "step": 9666
    },
    {
      "epoch": 0.8676180218991204,
      "grad_norm": 2.952467918395996,
      "learning_rate": 0.0,
      "loss": 0.6811,
      "step": 9667
    },
    {
      "epoch": 0.8677077723927482,
      "grad_norm": 7.23113489151001,
      "learning_rate": 0.0,
      "loss": 0.7722,
      "step": 9668
    },
    {
      "epoch": 0.8677975228863759,
      "grad_norm": 3.383317708969116,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 9669
    },
    {
      "epoch": 0.8678872733800036,
      "grad_norm": 2.973670244216919,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 9670
    },
    {
      "epoch": 0.8679770238736313,
      "grad_norm": 2.42561936378479,
      "learning_rate": 0.0,
      "loss": 0.6728,
      "step": 9671
    },
    {
      "epoch": 0.868066774367259,
      "grad_norm": 3.0706357955932617,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 9672
    },
    {
      "epoch": 0.8681565248608867,
      "grad_norm": 2.6079185009002686,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 9673
    },
    {
      "epoch": 0.8682462753545145,
      "grad_norm": 2.595698595046997,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 9674
    },
    {
      "epoch": 0.8683360258481422,
      "grad_norm": 2.5355231761932373,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 9675
    },
    {
      "epoch": 0.8684257763417699,
      "grad_norm": 2.294668436050415,
      "learning_rate": 0.0,
      "loss": 0.7396,
      "step": 9676
    },
    {
      "epoch": 0.8685155268353976,
      "grad_norm": 3.3060264587402344,
      "learning_rate": 0.0,
      "loss": 0.7477,
      "step": 9677
    },
    {
      "epoch": 0.8686052773290253,
      "grad_norm": 2.7811977863311768,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 9678
    },
    {
      "epoch": 0.8686950278226531,
      "grad_norm": 2.904665946960449,
      "learning_rate": 0.0,
      "loss": 0.7613,
      "step": 9679
    },
    {
      "epoch": 0.8687847783162808,
      "grad_norm": 2.267533540725708,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 9680
    },
    {
      "epoch": 0.8688745288099085,
      "grad_norm": 2.7827975749969482,
      "learning_rate": 0.0,
      "loss": 0.722,
      "step": 9681
    },
    {
      "epoch": 0.8689642793035361,
      "grad_norm": 3.1501989364624023,
      "learning_rate": 0.0,
      "loss": 0.6611,
      "step": 9682
    },
    {
      "epoch": 0.8690540297971638,
      "grad_norm": 3.9660065174102783,
      "learning_rate": 0.0,
      "loss": 0.796,
      "step": 9683
    },
    {
      "epoch": 0.8691437802907916,
      "grad_norm": 2.742856740951538,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 9684
    },
    {
      "epoch": 0.8692335307844193,
      "grad_norm": 2.363987445831299,
      "learning_rate": 0.0,
      "loss": 0.6741,
      "step": 9685
    },
    {
      "epoch": 0.869323281278047,
      "grad_norm": 2.13496994972229,
      "learning_rate": 0.0,
      "loss": 0.6687,
      "step": 9686
    },
    {
      "epoch": 0.8694130317716747,
      "grad_norm": 2.4732425212860107,
      "learning_rate": 0.0,
      "loss": 0.6637,
      "step": 9687
    },
    {
      "epoch": 0.8695027822653024,
      "grad_norm": 2.442112445831299,
      "learning_rate": 0.0,
      "loss": 0.6677,
      "step": 9688
    },
    {
      "epoch": 0.8695925327589302,
      "grad_norm": 2.0199174880981445,
      "learning_rate": 0.0,
      "loss": 0.7006,
      "step": 9689
    },
    {
      "epoch": 0.8696822832525579,
      "grad_norm": 5.6587982177734375,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 9690
    },
    {
      "epoch": 0.8697720337461856,
      "grad_norm": 3.4566469192504883,
      "learning_rate": 0.0,
      "loss": 0.7861,
      "step": 9691
    },
    {
      "epoch": 0.8698617842398133,
      "grad_norm": 3.434884786605835,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 9692
    },
    {
      "epoch": 0.869951534733441,
      "grad_norm": 2.6152825355529785,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 9693
    },
    {
      "epoch": 0.8700412852270687,
      "grad_norm": 2.631023645401001,
      "learning_rate": 0.0,
      "loss": 0.673,
      "step": 9694
    },
    {
      "epoch": 0.8701310357206965,
      "grad_norm": 2.971123456954956,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 9695
    },
    {
      "epoch": 0.8702207862143242,
      "grad_norm": 2.5598232746124268,
      "learning_rate": 0.0,
      "loss": 0.7683,
      "step": 9696
    },
    {
      "epoch": 0.8703105367079519,
      "grad_norm": 3.075291872024536,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 9697
    },
    {
      "epoch": 0.8704002872015796,
      "grad_norm": 2.66147518157959,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 9698
    },
    {
      "epoch": 0.8704900376952073,
      "grad_norm": 2.7394347190856934,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 9699
    },
    {
      "epoch": 0.8705797881888351,
      "grad_norm": 2.168046712875366,
      "learning_rate": 0.0,
      "loss": 0.7793,
      "step": 9700
    },
    {
      "epoch": 0.8706695386824628,
      "grad_norm": 2.784761667251587,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 9701
    },
    {
      "epoch": 0.8707592891760905,
      "grad_norm": 23.696138381958008,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 9702
    },
    {
      "epoch": 0.8708490396697182,
      "grad_norm": 2.347450017929077,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 9703
    },
    {
      "epoch": 0.8709387901633459,
      "grad_norm": 3.0442965030670166,
      "learning_rate": 0.0,
      "loss": 0.7362,
      "step": 9704
    },
    {
      "epoch": 0.8710285406569737,
      "grad_norm": 3.48922061920166,
      "learning_rate": 0.0,
      "loss": 0.6724,
      "step": 9705
    },
    {
      "epoch": 0.8711182911506014,
      "grad_norm": 2.4685189723968506,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 9706
    },
    {
      "epoch": 0.871208041644229,
      "grad_norm": 2.3171374797821045,
      "learning_rate": 0.0,
      "loss": 0.7757,
      "step": 9707
    },
    {
      "epoch": 0.8712977921378567,
      "grad_norm": 2.2597262859344482,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 9708
    },
    {
      "epoch": 0.8713875426314844,
      "grad_norm": 2.3645598888397217,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 9709
    },
    {
      "epoch": 0.8714772931251122,
      "grad_norm": 2.5711047649383545,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 9710
    },
    {
      "epoch": 0.8715670436187399,
      "grad_norm": 2.6239819526672363,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 9711
    },
    {
      "epoch": 0.8716567941123676,
      "grad_norm": 2.577772617340088,
      "learning_rate": 0.0,
      "loss": 0.7629,
      "step": 9712
    },
    {
      "epoch": 0.8717465446059953,
      "grad_norm": 3.1342458724975586,
      "learning_rate": 0.0,
      "loss": 0.6803,
      "step": 9713
    },
    {
      "epoch": 0.871836295099623,
      "grad_norm": 3.001863718032837,
      "learning_rate": 0.0,
      "loss": 0.7386,
      "step": 9714
    },
    {
      "epoch": 0.8719260455932507,
      "grad_norm": 2.0637898445129395,
      "learning_rate": 0.0,
      "loss": 0.6822,
      "step": 9715
    },
    {
      "epoch": 0.8720157960868785,
      "grad_norm": 3.232445240020752,
      "learning_rate": 0.0,
      "loss": 0.6817,
      "step": 9716
    },
    {
      "epoch": 0.8721055465805062,
      "grad_norm": 3.4491655826568604,
      "learning_rate": 0.0,
      "loss": 0.7732,
      "step": 9717
    },
    {
      "epoch": 0.8721952970741339,
      "grad_norm": 2.4065768718719482,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 9718
    },
    {
      "epoch": 0.8722850475677616,
      "grad_norm": 2.789982795715332,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 9719
    },
    {
      "epoch": 0.8723747980613893,
      "grad_norm": 3.1123952865600586,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 9720
    },
    {
      "epoch": 0.8724645485550171,
      "grad_norm": 3.819262742996216,
      "learning_rate": 0.0,
      "loss": 0.7586,
      "step": 9721
    },
    {
      "epoch": 0.8725542990486448,
      "grad_norm": 2.6994073390960693,
      "learning_rate": 0.0,
      "loss": 0.6425,
      "step": 9722
    },
    {
      "epoch": 0.8726440495422725,
      "grad_norm": 3.695388078689575,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 9723
    },
    {
      "epoch": 0.8727338000359002,
      "grad_norm": 2.3554017543792725,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 9724
    },
    {
      "epoch": 0.8728235505295279,
      "grad_norm": 3.8895609378814697,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 9725
    },
    {
      "epoch": 0.8729133010231557,
      "grad_norm": 3.0095298290252686,
      "learning_rate": 0.0,
      "loss": 0.6771,
      "step": 9726
    },
    {
      "epoch": 0.8730030515167834,
      "grad_norm": 2.2162513732910156,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 9727
    },
    {
      "epoch": 0.8730928020104111,
      "grad_norm": 3.8654096126556396,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 9728
    },
    {
      "epoch": 0.8731825525040388,
      "grad_norm": 3.240111827850342,
      "learning_rate": 0.0,
      "loss": 0.6828,
      "step": 9729
    },
    {
      "epoch": 0.8732723029976664,
      "grad_norm": 2.807990312576294,
      "learning_rate": 0.0,
      "loss": 0.7506,
      "step": 9730
    },
    {
      "epoch": 0.8733620534912943,
      "grad_norm": 2.3293166160583496,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 9731
    },
    {
      "epoch": 0.873451803984922,
      "grad_norm": 2.6673080921173096,
      "learning_rate": 0.0,
      "loss": 0.705,
      "step": 9732
    },
    {
      "epoch": 0.8735415544785496,
      "grad_norm": 3.957853078842163,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 9733
    },
    {
      "epoch": 0.8736313049721773,
      "grad_norm": 3.492220401763916,
      "learning_rate": 0.0,
      "loss": 0.7544,
      "step": 9734
    },
    {
      "epoch": 0.873721055465805,
      "grad_norm": 3.018533945083618,
      "learning_rate": 0.0,
      "loss": 0.7738,
      "step": 9735
    },
    {
      "epoch": 0.8738108059594327,
      "grad_norm": 2.62717342376709,
      "learning_rate": 0.0,
      "loss": 0.6766,
      "step": 9736
    },
    {
      "epoch": 0.8739005564530605,
      "grad_norm": 5.522078514099121,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 9737
    },
    {
      "epoch": 0.8739903069466882,
      "grad_norm": 3.9980902671813965,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 9738
    },
    {
      "epoch": 0.8740800574403159,
      "grad_norm": 4.433104991912842,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 9739
    },
    {
      "epoch": 0.8741698079339436,
      "grad_norm": 3.1663131713867188,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 9740
    },
    {
      "epoch": 0.8742595584275713,
      "grad_norm": 2.681605100631714,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 9741
    },
    {
      "epoch": 0.8743493089211991,
      "grad_norm": 2.1714894771575928,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 9742
    },
    {
      "epoch": 0.8744390594148268,
      "grad_norm": 2.458643913269043,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 9743
    },
    {
      "epoch": 0.8745288099084545,
      "grad_norm": 2.7028284072875977,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 9744
    },
    {
      "epoch": 0.8746185604020822,
      "grad_norm": 2.730398654937744,
      "learning_rate": 0.0,
      "loss": 0.7065,
      "step": 9745
    },
    {
      "epoch": 0.8747083108957099,
      "grad_norm": 2.5988409519195557,
      "learning_rate": 0.0,
      "loss": 0.7485,
      "step": 9746
    },
    {
      "epoch": 0.8747980613893377,
      "grad_norm": 2.6229913234710693,
      "learning_rate": 0.0,
      "loss": 0.7711,
      "step": 9747
    },
    {
      "epoch": 0.8748878118829654,
      "grad_norm": 2.469888210296631,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 9748
    },
    {
      "epoch": 0.8749775623765931,
      "grad_norm": 3.3985705375671387,
      "learning_rate": 0.0,
      "loss": 0.7139,
      "step": 9749
    },
    {
      "epoch": 0.8750673128702208,
      "grad_norm": 4.175319671630859,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 9750
    },
    {
      "epoch": 0.8751570633638485,
      "grad_norm": 4.430607318878174,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 9751
    },
    {
      "epoch": 0.8752468138574763,
      "grad_norm": 3.231539249420166,
      "learning_rate": 0.0,
      "loss": 0.6913,
      "step": 9752
    },
    {
      "epoch": 0.875336564351104,
      "grad_norm": 2.9533398151397705,
      "learning_rate": 0.0,
      "loss": 0.7471,
      "step": 9753
    },
    {
      "epoch": 0.8754263148447317,
      "grad_norm": 2.6812334060668945,
      "learning_rate": 0.0,
      "loss": 0.7151,
      "step": 9754
    },
    {
      "epoch": 0.8755160653383594,
      "grad_norm": 3.3874237537384033,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 9755
    },
    {
      "epoch": 0.875605815831987,
      "grad_norm": 3.374267578125,
      "learning_rate": 0.0,
      "loss": 0.6678,
      "step": 9756
    },
    {
      "epoch": 0.8756955663256147,
      "grad_norm": 3.1102774143218994,
      "learning_rate": 0.0,
      "loss": 0.6768,
      "step": 9757
    },
    {
      "epoch": 0.8757853168192425,
      "grad_norm": 3.083702564239502,
      "learning_rate": 0.0,
      "loss": 0.7492,
      "step": 9758
    },
    {
      "epoch": 0.8758750673128702,
      "grad_norm": 3.257195472717285,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 9759
    },
    {
      "epoch": 0.8759648178064979,
      "grad_norm": 3.149761915206909,
      "learning_rate": 0.0,
      "loss": 0.673,
      "step": 9760
    },
    {
      "epoch": 0.8760545683001256,
      "grad_norm": 3.0009572505950928,
      "learning_rate": 0.0,
      "loss": 0.7718,
      "step": 9761
    },
    {
      "epoch": 0.8761443187937533,
      "grad_norm": 11.72973346710205,
      "learning_rate": 0.0,
      "loss": 0.7102,
      "step": 9762
    },
    {
      "epoch": 0.8762340692873811,
      "grad_norm": 3.3293538093566895,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 9763
    },
    {
      "epoch": 0.8763238197810088,
      "grad_norm": 3.7050721645355225,
      "learning_rate": 0.0,
      "loss": 0.7433,
      "step": 9764
    },
    {
      "epoch": 0.8764135702746365,
      "grad_norm": 3.3427224159240723,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 9765
    },
    {
      "epoch": 0.8765033207682642,
      "grad_norm": 3.3621723651885986,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 9766
    },
    {
      "epoch": 0.8765930712618919,
      "grad_norm": 7.6541337966918945,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 9767
    },
    {
      "epoch": 0.8766828217555197,
      "grad_norm": 4.18806266784668,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 9768
    },
    {
      "epoch": 0.8767725722491474,
      "grad_norm": 3.6141560077667236,
      "learning_rate": 0.0,
      "loss": 0.7572,
      "step": 9769
    },
    {
      "epoch": 0.8768623227427751,
      "grad_norm": 3.0868029594421387,
      "learning_rate": 0.0,
      "loss": 0.6806,
      "step": 9770
    },
    {
      "epoch": 0.8769520732364028,
      "grad_norm": 4.235915184020996,
      "learning_rate": 0.0,
      "loss": 0.6725,
      "step": 9771
    },
    {
      "epoch": 0.8770418237300305,
      "grad_norm": 3.1384196281433105,
      "learning_rate": 0.0,
      "loss": 0.6881,
      "step": 9772
    },
    {
      "epoch": 0.8771315742236583,
      "grad_norm": 4.203793525695801,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 9773
    },
    {
      "epoch": 0.877221324717286,
      "grad_norm": 2.657784938812256,
      "learning_rate": 0.0,
      "loss": 0.7484,
      "step": 9774
    },
    {
      "epoch": 0.8773110752109137,
      "grad_norm": 2.3660507202148438,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 9775
    },
    {
      "epoch": 0.8774008257045414,
      "grad_norm": 2.768126964569092,
      "learning_rate": 0.0,
      "loss": 0.7183,
      "step": 9776
    },
    {
      "epoch": 0.8774905761981691,
      "grad_norm": 3.5004777908325195,
      "learning_rate": 0.0,
      "loss": 0.6898,
      "step": 9777
    },
    {
      "epoch": 0.8775803266917968,
      "grad_norm": 3.717111587524414,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 9778
    },
    {
      "epoch": 0.8776700771854246,
      "grad_norm": 2.52194881439209,
      "learning_rate": 0.0,
      "loss": 0.7367,
      "step": 9779
    },
    {
      "epoch": 0.8777598276790523,
      "grad_norm": 2.967424154281616,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 9780
    },
    {
      "epoch": 0.87784957817268,
      "grad_norm": 2.6065475940704346,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 9781
    },
    {
      "epoch": 0.8779393286663076,
      "grad_norm": 2.320801258087158,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 9782
    },
    {
      "epoch": 0.8780290791599353,
      "grad_norm": 3.084439754486084,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 9783
    },
    {
      "epoch": 0.8781188296535631,
      "grad_norm": 3.102224826812744,
      "learning_rate": 0.0,
      "loss": 0.6741,
      "step": 9784
    },
    {
      "epoch": 0.8782085801471908,
      "grad_norm": 3.6039586067199707,
      "learning_rate": 0.0,
      "loss": 0.6652,
      "step": 9785
    },
    {
      "epoch": 0.8782983306408185,
      "grad_norm": 2.8499186038970947,
      "learning_rate": 0.0,
      "loss": 0.6445,
      "step": 9786
    },
    {
      "epoch": 0.8783880811344462,
      "grad_norm": 3.1691181659698486,
      "learning_rate": 0.0,
      "loss": 0.758,
      "step": 9787
    },
    {
      "epoch": 0.8784778316280739,
      "grad_norm": 7.851556777954102,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 9788
    },
    {
      "epoch": 0.8785675821217017,
      "grad_norm": 4.247547149658203,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 9789
    },
    {
      "epoch": 0.8786573326153294,
      "grad_norm": 4.141777515411377,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 9790
    },
    {
      "epoch": 0.8787470831089571,
      "grad_norm": 2.555539608001709,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 9791
    },
    {
      "epoch": 0.8788368336025848,
      "grad_norm": 3.445990800857544,
      "learning_rate": 0.0,
      "loss": 0.7949,
      "step": 9792
    },
    {
      "epoch": 0.8789265840962125,
      "grad_norm": 4.33269739151001,
      "learning_rate": 0.0,
      "loss": 0.6819,
      "step": 9793
    },
    {
      "epoch": 0.8790163345898403,
      "grad_norm": 2.999307155609131,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 9794
    },
    {
      "epoch": 0.879106085083468,
      "grad_norm": 3.0416555404663086,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 9795
    },
    {
      "epoch": 0.8791958355770957,
      "grad_norm": 2.6212215423583984,
      "learning_rate": 0.0,
      "loss": 0.7523,
      "step": 9796
    },
    {
      "epoch": 0.8792855860707234,
      "grad_norm": 2.6023576259613037,
      "learning_rate": 0.0,
      "loss": 0.6826,
      "step": 9797
    },
    {
      "epoch": 0.8793753365643511,
      "grad_norm": 3.8938634395599365,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 9798
    },
    {
      "epoch": 0.8794650870579788,
      "grad_norm": 3.239060640335083,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 9799
    },
    {
      "epoch": 0.8795548375516066,
      "grad_norm": 3.585836887359619,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 9800
    },
    {
      "epoch": 0.8796445880452343,
      "grad_norm": 3.4081432819366455,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 9801
    },
    {
      "epoch": 0.879734338538862,
      "grad_norm": 2.9032630920410156,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 9802
    },
    {
      "epoch": 0.8798240890324897,
      "grad_norm": 2.734276294708252,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 9803
    },
    {
      "epoch": 0.8799138395261173,
      "grad_norm": 3.2029294967651367,
      "learning_rate": 0.0,
      "loss": 0.7582,
      "step": 9804
    },
    {
      "epoch": 0.8800035900197452,
      "grad_norm": 3.229733467102051,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 9805
    },
    {
      "epoch": 0.8800933405133728,
      "grad_norm": 4.155483722686768,
      "learning_rate": 0.0,
      "loss": 0.6572,
      "step": 9806
    },
    {
      "epoch": 0.8801830910070005,
      "grad_norm": 4.23486852645874,
      "learning_rate": 0.0,
      "loss": 0.6783,
      "step": 9807
    },
    {
      "epoch": 0.8802728415006282,
      "grad_norm": 4.474471092224121,
      "learning_rate": 0.0,
      "loss": 0.6553,
      "step": 9808
    },
    {
      "epoch": 0.8803625919942559,
      "grad_norm": 3.143176794052124,
      "learning_rate": 0.0,
      "loss": 0.8246,
      "step": 9809
    },
    {
      "epoch": 0.8804523424878837,
      "grad_norm": 3.0597951412200928,
      "learning_rate": 0.0,
      "loss": 0.7654,
      "step": 9810
    },
    {
      "epoch": 0.8805420929815114,
      "grad_norm": 3.177591562271118,
      "learning_rate": 0.0,
      "loss": 0.7827,
      "step": 9811
    },
    {
      "epoch": 0.8806318434751391,
      "grad_norm": 2.426663637161255,
      "learning_rate": 0.0,
      "loss": 0.7161,
      "step": 9812
    },
    {
      "epoch": 0.8807215939687668,
      "grad_norm": 4.917906761169434,
      "learning_rate": 0.0,
      "loss": 0.6901,
      "step": 9813
    },
    {
      "epoch": 0.8808113444623945,
      "grad_norm": 3.328885793685913,
      "learning_rate": 0.0,
      "loss": 0.775,
      "step": 9814
    },
    {
      "epoch": 0.8809010949560223,
      "grad_norm": 3.717656373977661,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 9815
    },
    {
      "epoch": 0.88099084544965,
      "grad_norm": 2.793219566345215,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 9816
    },
    {
      "epoch": 0.8810805959432777,
      "grad_norm": 3.172935962677002,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 9817
    },
    {
      "epoch": 0.8811703464369054,
      "grad_norm": 4.225831508636475,
      "learning_rate": 0.0,
      "loss": 0.7555,
      "step": 9818
    },
    {
      "epoch": 0.8812600969305331,
      "grad_norm": 3.4578120708465576,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 9819
    },
    {
      "epoch": 0.8813498474241608,
      "grad_norm": 2.3619508743286133,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 9820
    },
    {
      "epoch": 0.8814395979177886,
      "grad_norm": 2.9513909816741943,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 9821
    },
    {
      "epoch": 0.8815293484114163,
      "grad_norm": 2.6166117191314697,
      "learning_rate": 0.0,
      "loss": 0.7462,
      "step": 9822
    },
    {
      "epoch": 0.881619098905044,
      "grad_norm": 3.7472469806671143,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 9823
    },
    {
      "epoch": 0.8817088493986717,
      "grad_norm": 2.9681286811828613,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 9824
    },
    {
      "epoch": 0.8817985998922994,
      "grad_norm": 2.5274906158447266,
      "learning_rate": 0.0,
      "loss": 0.6795,
      "step": 9825
    },
    {
      "epoch": 0.8818883503859272,
      "grad_norm": 2.5815999507904053,
      "learning_rate": 0.0,
      "loss": 0.763,
      "step": 9826
    },
    {
      "epoch": 0.8819781008795549,
      "grad_norm": 2.754624366760254,
      "learning_rate": 0.0,
      "loss": 0.6697,
      "step": 9827
    },
    {
      "epoch": 0.8820678513731826,
      "grad_norm": 3.109006643295288,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 9828
    },
    {
      "epoch": 0.8821576018668102,
      "grad_norm": 2.538717746734619,
      "learning_rate": 0.0,
      "loss": 0.6515,
      "step": 9829
    },
    {
      "epoch": 0.8822473523604379,
      "grad_norm": 2.192355155944824,
      "learning_rate": 0.0,
      "loss": 0.733,
      "step": 9830
    },
    {
      "epoch": 0.8823371028540657,
      "grad_norm": 2.4756104946136475,
      "learning_rate": 0.0,
      "loss": 0.6496,
      "step": 9831
    },
    {
      "epoch": 0.8824268533476934,
      "grad_norm": 2.677999258041382,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 9832
    },
    {
      "epoch": 0.8825166038413211,
      "grad_norm": 3.051478385925293,
      "learning_rate": 0.0,
      "loss": 0.7615,
      "step": 9833
    },
    {
      "epoch": 0.8826063543349488,
      "grad_norm": 3.225400924682617,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 9834
    },
    {
      "epoch": 0.8826961048285765,
      "grad_norm": 2.9072344303131104,
      "learning_rate": 0.0,
      "loss": 0.8343,
      "step": 9835
    },
    {
      "epoch": 0.8827858553222043,
      "grad_norm": 2.6907477378845215,
      "learning_rate": 0.0,
      "loss": 0.751,
      "step": 9836
    },
    {
      "epoch": 0.882875605815832,
      "grad_norm": 2.919384241104126,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 9837
    },
    {
      "epoch": 0.8829653563094597,
      "grad_norm": 2.4475502967834473,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 9838
    },
    {
      "epoch": 0.8830551068030874,
      "grad_norm": 3.5270237922668457,
      "learning_rate": 0.0,
      "loss": 0.7212,
      "step": 9839
    },
    {
      "epoch": 0.8831448572967151,
      "grad_norm": 3.2679007053375244,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 9840
    },
    {
      "epoch": 0.8832346077903428,
      "grad_norm": 3.7721407413482666,
      "learning_rate": 0.0,
      "loss": 0.6418,
      "step": 9841
    },
    {
      "epoch": 0.8833243582839706,
      "grad_norm": 3.2935938835144043,
      "learning_rate": 0.0,
      "loss": 0.63,
      "step": 9842
    },
    {
      "epoch": 0.8834141087775983,
      "grad_norm": 2.4263226985931396,
      "learning_rate": 0.0,
      "loss": 0.6874,
      "step": 9843
    },
    {
      "epoch": 0.883503859271226,
      "grad_norm": 3.1720738410949707,
      "learning_rate": 0.0,
      "loss": 0.714,
      "step": 9844
    },
    {
      "epoch": 0.8835936097648537,
      "grad_norm": 2.745215654373169,
      "learning_rate": 0.0,
      "loss": 0.7923,
      "step": 9845
    },
    {
      "epoch": 0.8836833602584814,
      "grad_norm": 6.425050735473633,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 9846
    },
    {
      "epoch": 0.8837731107521092,
      "grad_norm": 2.7784249782562256,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 9847
    },
    {
      "epoch": 0.8838628612457369,
      "grad_norm": 4.044942378997803,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 9848
    },
    {
      "epoch": 0.8839526117393646,
      "grad_norm": 8.01665210723877,
      "learning_rate": 0.0,
      "loss": 0.7721,
      "step": 9849
    },
    {
      "epoch": 0.8840423622329923,
      "grad_norm": 2.7832751274108887,
      "learning_rate": 0.0,
      "loss": 0.6715,
      "step": 9850
    },
    {
      "epoch": 0.88413211272662,
      "grad_norm": 3.9261844158172607,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 9851
    },
    {
      "epoch": 0.8842218632202478,
      "grad_norm": 5.842193126678467,
      "learning_rate": 0.0,
      "loss": 0.761,
      "step": 9852
    },
    {
      "epoch": 0.8843116137138755,
      "grad_norm": 2.7692923545837402,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 9853
    },
    {
      "epoch": 0.8844013642075031,
      "grad_norm": 2.6855733394622803,
      "learning_rate": 0.0,
      "loss": 0.7576,
      "step": 9854
    },
    {
      "epoch": 0.8844911147011308,
      "grad_norm": 3.075535535812378,
      "learning_rate": 0.0,
      "loss": 0.7806,
      "step": 9855
    },
    {
      "epoch": 0.8845808651947585,
      "grad_norm": 3.25252628326416,
      "learning_rate": 0.0,
      "loss": 0.6659,
      "step": 9856
    },
    {
      "epoch": 0.8846706156883863,
      "grad_norm": 3.185302972793579,
      "learning_rate": 0.0,
      "loss": 0.6602,
      "step": 9857
    },
    {
      "epoch": 0.884760366182014,
      "grad_norm": 3.394672393798828,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 9858
    },
    {
      "epoch": 0.8848501166756417,
      "grad_norm": 2.8118019104003906,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 9859
    },
    {
      "epoch": 0.8849398671692694,
      "grad_norm": 2.787229299545288,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 9860
    },
    {
      "epoch": 0.8850296176628971,
      "grad_norm": 4.297991752624512,
      "learning_rate": 0.0,
      "loss": 0.7188,
      "step": 9861
    },
    {
      "epoch": 0.8851193681565248,
      "grad_norm": 2.845262050628662,
      "learning_rate": 0.0,
      "loss": 0.7021,
      "step": 9862
    },
    {
      "epoch": 0.8852091186501526,
      "grad_norm": 3.6700267791748047,
      "learning_rate": 0.0,
      "loss": 0.7193,
      "step": 9863
    },
    {
      "epoch": 0.8852988691437803,
      "grad_norm": 3.444197654724121,
      "learning_rate": 0.0,
      "loss": 0.7665,
      "step": 9864
    },
    {
      "epoch": 0.885388619637408,
      "grad_norm": 2.5092105865478516,
      "learning_rate": 0.0,
      "loss": 0.7408,
      "step": 9865
    },
    {
      "epoch": 0.8854783701310357,
      "grad_norm": 2.7903761863708496,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 9866
    },
    {
      "epoch": 0.8855681206246634,
      "grad_norm": 2.7660162448883057,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 9867
    },
    {
      "epoch": 0.8856578711182912,
      "grad_norm": 2.9173684120178223,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 9868
    },
    {
      "epoch": 0.8857476216119189,
      "grad_norm": 2.7956225872039795,
      "learning_rate": 0.0,
      "loss": 0.7482,
      "step": 9869
    },
    {
      "epoch": 0.8858373721055466,
      "grad_norm": 2.818894147872925,
      "learning_rate": 0.0,
      "loss": 0.6767,
      "step": 9870
    },
    {
      "epoch": 0.8859271225991743,
      "grad_norm": 2.207939386367798,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 9871
    },
    {
      "epoch": 0.886016873092802,
      "grad_norm": 4.0423150062561035,
      "learning_rate": 0.0,
      "loss": 0.6874,
      "step": 9872
    },
    {
      "epoch": 0.8861066235864298,
      "grad_norm": 3.3174374103546143,
      "learning_rate": 0.0,
      "loss": 0.7698,
      "step": 9873
    },
    {
      "epoch": 0.8861963740800575,
      "grad_norm": 2.543457508087158,
      "learning_rate": 0.0,
      "loss": 0.7771,
      "step": 9874
    },
    {
      "epoch": 0.8862861245736852,
      "grad_norm": 2.354074716567993,
      "learning_rate": 0.0,
      "loss": 0.7125,
      "step": 9875
    },
    {
      "epoch": 0.8863758750673129,
      "grad_norm": 4.005518436431885,
      "learning_rate": 0.0,
      "loss": 0.6403,
      "step": 9876
    },
    {
      "epoch": 0.8864656255609406,
      "grad_norm": 4.056360244750977,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 9877
    },
    {
      "epoch": 0.8865553760545682,
      "grad_norm": 2.915255069732666,
      "learning_rate": 0.0,
      "loss": 0.6689,
      "step": 9878
    },
    {
      "epoch": 0.886645126548196,
      "grad_norm": 13.956435203552246,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 9879
    },
    {
      "epoch": 0.8867348770418237,
      "grad_norm": 3.770256280899048,
      "learning_rate": 0.0,
      "loss": 0.6637,
      "step": 9880
    },
    {
      "epoch": 0.8868246275354514,
      "grad_norm": 26.34550666809082,
      "learning_rate": 0.0,
      "loss": 0.7731,
      "step": 9881
    },
    {
      "epoch": 0.8869143780290791,
      "grad_norm": 2.9688332080841064,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 9882
    },
    {
      "epoch": 0.8870041285227068,
      "grad_norm": 2.465146541595459,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 9883
    },
    {
      "epoch": 0.8870938790163346,
      "grad_norm": 5.969688892364502,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 9884
    },
    {
      "epoch": 0.8871836295099623,
      "grad_norm": 4.412847518920898,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 9885
    },
    {
      "epoch": 0.88727338000359,
      "grad_norm": 2.661935329437256,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 9886
    },
    {
      "epoch": 0.8873631304972177,
      "grad_norm": 2.8213839530944824,
      "learning_rate": 0.0,
      "loss": 0.8096,
      "step": 9887
    },
    {
      "epoch": 0.8874528809908454,
      "grad_norm": 3.128570795059204,
      "learning_rate": 0.0,
      "loss": 0.6553,
      "step": 9888
    },
    {
      "epoch": 0.8875426314844732,
      "grad_norm": 4.851536273956299,
      "learning_rate": 0.0,
      "loss": 0.7735,
      "step": 9889
    },
    {
      "epoch": 0.8876323819781009,
      "grad_norm": 6.378001689910889,
      "learning_rate": 0.0,
      "loss": 0.7169,
      "step": 9890
    },
    {
      "epoch": 0.8877221324717286,
      "grad_norm": 3.9775829315185547,
      "learning_rate": 0.0,
      "loss": 0.7078,
      "step": 9891
    },
    {
      "epoch": 0.8878118829653563,
      "grad_norm": 5.55187463760376,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 9892
    },
    {
      "epoch": 0.887901633458984,
      "grad_norm": 4.474068641662598,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 9893
    },
    {
      "epoch": 0.8879913839526118,
      "grad_norm": 3.02494478225708,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 9894
    },
    {
      "epoch": 0.8880811344462395,
      "grad_norm": 4.072444915771484,
      "learning_rate": 0.0,
      "loss": 0.709,
      "step": 9895
    },
    {
      "epoch": 0.8881708849398672,
      "grad_norm": 3.5477242469787598,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 9896
    },
    {
      "epoch": 0.8882606354334949,
      "grad_norm": 3.319692850112915,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 9897
    },
    {
      "epoch": 0.8883503859271226,
      "grad_norm": 2.196538209915161,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 9898
    },
    {
      "epoch": 0.8884401364207503,
      "grad_norm": 3.0037434101104736,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 9899
    },
    {
      "epoch": 0.8885298869143781,
      "grad_norm": 2.91497802734375,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 9900
    },
    {
      "epoch": 0.8886196374080058,
      "grad_norm": 2.4670321941375732,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 9901
    },
    {
      "epoch": 0.8887093879016335,
      "grad_norm": 2.7738583087921143,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 9902
    },
    {
      "epoch": 0.8887991383952611,
      "grad_norm": 3.0299532413482666,
      "learning_rate": 0.0,
      "loss": 0.7509,
      "step": 9903
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 5.480081558227539,
      "learning_rate": 0.0,
      "loss": 0.6886,
      "step": 9904
    },
    {
      "epoch": 0.8889786393825166,
      "grad_norm": 3.233747720718384,
      "learning_rate": 0.0,
      "loss": 0.653,
      "step": 9905
    },
    {
      "epoch": 0.8890683898761443,
      "grad_norm": 2.985177993774414,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 9906
    },
    {
      "epoch": 0.889158140369772,
      "grad_norm": 3.20973801612854,
      "learning_rate": 0.0,
      "loss": 0.6469,
      "step": 9907
    },
    {
      "epoch": 0.8892478908633997,
      "grad_norm": 3.172611713409424,
      "learning_rate": 0.0,
      "loss": 0.6885,
      "step": 9908
    },
    {
      "epoch": 0.8893376413570274,
      "grad_norm": 4.232844352722168,
      "learning_rate": 0.0,
      "loss": 0.7193,
      "step": 9909
    },
    {
      "epoch": 0.8894273918506552,
      "grad_norm": 3.592292547225952,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 9910
    },
    {
      "epoch": 0.8895171423442829,
      "grad_norm": 2.8512463569641113,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 9911
    },
    {
      "epoch": 0.8896068928379106,
      "grad_norm": 3.5533056259155273,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 9912
    },
    {
      "epoch": 0.8896966433315383,
      "grad_norm": 3.721367597579956,
      "learning_rate": 0.0,
      "loss": 0.7071,
      "step": 9913
    },
    {
      "epoch": 0.889786393825166,
      "grad_norm": 4.4199628829956055,
      "learning_rate": 0.0,
      "loss": 0.6718,
      "step": 9914
    },
    {
      "epoch": 0.8898761443187938,
      "grad_norm": 3.754814863204956,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 9915
    },
    {
      "epoch": 0.8899658948124215,
      "grad_norm": 3.2090585231781006,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 9916
    },
    {
      "epoch": 0.8900556453060492,
      "grad_norm": 3.3910961151123047,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 9917
    },
    {
      "epoch": 0.8901453957996769,
      "grad_norm": 2.892509937286377,
      "learning_rate": 0.0,
      "loss": 0.7566,
      "step": 9918
    },
    {
      "epoch": 0.8902351462933046,
      "grad_norm": 5.840766906738281,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 9919
    },
    {
      "epoch": 0.8903248967869323,
      "grad_norm": 2.9928689002990723,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 9920
    },
    {
      "epoch": 0.8904146472805601,
      "grad_norm": 4.411932945251465,
      "learning_rate": 0.0,
      "loss": 0.7392,
      "step": 9921
    },
    {
      "epoch": 0.8905043977741878,
      "grad_norm": 3.135359764099121,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 9922
    },
    {
      "epoch": 0.8905941482678155,
      "grad_norm": 3.501634359359741,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 9923
    },
    {
      "epoch": 0.8906838987614432,
      "grad_norm": 2.1712417602539062,
      "learning_rate": 0.0,
      "loss": 0.7965,
      "step": 9924
    },
    {
      "epoch": 0.8907736492550709,
      "grad_norm": 2.7616591453552246,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 9925
    },
    {
      "epoch": 0.8908633997486987,
      "grad_norm": 2.633417844772339,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 9926
    },
    {
      "epoch": 0.8909531502423264,
      "grad_norm": 3.795513391494751,
      "learning_rate": 0.0,
      "loss": 0.6273,
      "step": 9927
    },
    {
      "epoch": 0.891042900735954,
      "grad_norm": 3.490133762359619,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 9928
    },
    {
      "epoch": 0.8911326512295817,
      "grad_norm": 2.935832977294922,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 9929
    },
    {
      "epoch": 0.8912224017232094,
      "grad_norm": 3.363658905029297,
      "learning_rate": 0.0,
      "loss": 0.7056,
      "step": 9930
    },
    {
      "epoch": 0.8913121522168372,
      "grad_norm": 4.2660675048828125,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 9931
    },
    {
      "epoch": 0.8914019027104649,
      "grad_norm": 3.566138744354248,
      "learning_rate": 0.0,
      "loss": 0.7375,
      "step": 9932
    },
    {
      "epoch": 0.8914916532040926,
      "grad_norm": 6.097500801086426,
      "learning_rate": 0.0,
      "loss": 0.6859,
      "step": 9933
    },
    {
      "epoch": 0.8915814036977203,
      "grad_norm": 3.9211313724517822,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 9934
    },
    {
      "epoch": 0.891671154191348,
      "grad_norm": 3.264833450317383,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 9935
    },
    {
      "epoch": 0.8917609046849758,
      "grad_norm": 2.4612576961517334,
      "learning_rate": 0.0,
      "loss": 0.7316,
      "step": 9936
    },
    {
      "epoch": 0.8918506551786035,
      "grad_norm": 3.130953550338745,
      "learning_rate": 0.0,
      "loss": 0.6872,
      "step": 9937
    },
    {
      "epoch": 0.8919404056722312,
      "grad_norm": 2.978119134902954,
      "learning_rate": 0.0,
      "loss": 0.6702,
      "step": 9938
    },
    {
      "epoch": 0.8920301561658589,
      "grad_norm": 3.0434529781341553,
      "learning_rate": 0.0,
      "loss": 0.6677,
      "step": 9939
    },
    {
      "epoch": 0.8921199066594866,
      "grad_norm": 3.2161362171173096,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 9940
    },
    {
      "epoch": 0.8922096571531143,
      "grad_norm": 3.2741336822509766,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 9941
    },
    {
      "epoch": 0.8922994076467421,
      "grad_norm": 3.2820541858673096,
      "learning_rate": 0.0,
      "loss": 0.736,
      "step": 9942
    },
    {
      "epoch": 0.8923891581403698,
      "grad_norm": 3.9391632080078125,
      "learning_rate": 0.0,
      "loss": 0.6918,
      "step": 9943
    },
    {
      "epoch": 0.8924789086339975,
      "grad_norm": 3.6852500438690186,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 9944
    },
    {
      "epoch": 0.8925686591276252,
      "grad_norm": 4.783285140991211,
      "learning_rate": 0.0,
      "loss": 0.7543,
      "step": 9945
    },
    {
      "epoch": 0.8926584096212529,
      "grad_norm": 2.805548906326294,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 9946
    },
    {
      "epoch": 0.8927481601148807,
      "grad_norm": 2.943809986114502,
      "learning_rate": 0.0,
      "loss": 0.6447,
      "step": 9947
    },
    {
      "epoch": 0.8928379106085084,
      "grad_norm": 3.4833834171295166,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 9948
    },
    {
      "epoch": 0.8929276611021361,
      "grad_norm": 4.86877965927124,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 9949
    },
    {
      "epoch": 0.8930174115957638,
      "grad_norm": 2.953629732131958,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 9950
    },
    {
      "epoch": 0.8931071620893914,
      "grad_norm": 4.0587263107299805,
      "learning_rate": 0.0,
      "loss": 0.7617,
      "step": 9951
    },
    {
      "epoch": 0.8931969125830193,
      "grad_norm": 4.007565975189209,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 9952
    },
    {
      "epoch": 0.893286663076647,
      "grad_norm": 4.284980773925781,
      "learning_rate": 0.0,
      "loss": 0.7076,
      "step": 9953
    },
    {
      "epoch": 0.8933764135702746,
      "grad_norm": 2.2217695713043213,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 9954
    },
    {
      "epoch": 0.8934661640639023,
      "grad_norm": 2.7629098892211914,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 9955
    },
    {
      "epoch": 0.89355591455753,
      "grad_norm": 3.6341280937194824,
      "learning_rate": 0.0,
      "loss": 0.6797,
      "step": 9956
    },
    {
      "epoch": 0.8936456650511578,
      "grad_norm": 3.9946768283843994,
      "learning_rate": 0.0,
      "loss": 0.7469,
      "step": 9957
    },
    {
      "epoch": 0.8937354155447855,
      "grad_norm": 3.56427264213562,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 9958
    },
    {
      "epoch": 0.8938251660384132,
      "grad_norm": 3.3693597316741943,
      "learning_rate": 0.0,
      "loss": 0.6752,
      "step": 9959
    },
    {
      "epoch": 0.8939149165320409,
      "grad_norm": 2.5850491523742676,
      "learning_rate": 0.0,
      "loss": 0.7644,
      "step": 9960
    },
    {
      "epoch": 0.8940046670256686,
      "grad_norm": 3.069213390350342,
      "learning_rate": 0.0,
      "loss": 0.7496,
      "step": 9961
    },
    {
      "epoch": 0.8940944175192963,
      "grad_norm": 2.634676694869995,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 9962
    },
    {
      "epoch": 0.8941841680129241,
      "grad_norm": 2.8495304584503174,
      "learning_rate": 0.0,
      "loss": 0.7037,
      "step": 9963
    },
    {
      "epoch": 0.8942739185065518,
      "grad_norm": 4.18410062789917,
      "learning_rate": 0.0,
      "loss": 0.7279,
      "step": 9964
    },
    {
      "epoch": 0.8943636690001795,
      "grad_norm": 4.233969211578369,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 9965
    },
    {
      "epoch": 0.8944534194938072,
      "grad_norm": 2.7793896198272705,
      "learning_rate": 0.0,
      "loss": 0.6885,
      "step": 9966
    },
    {
      "epoch": 0.8945431699874349,
      "grad_norm": 2.597170114517212,
      "learning_rate": 0.0,
      "loss": 0.6749,
      "step": 9967
    },
    {
      "epoch": 0.8946329204810627,
      "grad_norm": 2.757016658782959,
      "learning_rate": 0.0,
      "loss": 0.6978,
      "step": 9968
    },
    {
      "epoch": 0.8947226709746904,
      "grad_norm": 3.504753828048706,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 9969
    },
    {
      "epoch": 0.8948124214683181,
      "grad_norm": 3.0420944690704346,
      "learning_rate": 0.0,
      "loss": 0.6942,
      "step": 9970
    },
    {
      "epoch": 0.8949021719619458,
      "grad_norm": 2.7001469135284424,
      "learning_rate": 0.0,
      "loss": 0.7811,
      "step": 9971
    },
    {
      "epoch": 0.8949919224555735,
      "grad_norm": 4.375129699707031,
      "learning_rate": 0.0,
      "loss": 0.6883,
      "step": 9972
    },
    {
      "epoch": 0.8950816729492013,
      "grad_norm": 3.935089349746704,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 9973
    },
    {
      "epoch": 0.895171423442829,
      "grad_norm": 9.505252838134766,
      "learning_rate": 0.0,
      "loss": 0.6916,
      "step": 9974
    },
    {
      "epoch": 0.8952611739364567,
      "grad_norm": 3.2109878063201904,
      "learning_rate": 0.0,
      "loss": 0.6695,
      "step": 9975
    },
    {
      "epoch": 0.8953509244300843,
      "grad_norm": 2.5253639221191406,
      "learning_rate": 0.0,
      "loss": 0.7119,
      "step": 9976
    },
    {
      "epoch": 0.895440674923712,
      "grad_norm": 2.946929454803467,
      "learning_rate": 0.0,
      "loss": 0.6729,
      "step": 9977
    },
    {
      "epoch": 0.8955304254173398,
      "grad_norm": 2.843804359436035,
      "learning_rate": 0.0,
      "loss": 0.6719,
      "step": 9978
    },
    {
      "epoch": 0.8956201759109675,
      "grad_norm": 3.2516119480133057,
      "learning_rate": 0.0,
      "loss": 0.7081,
      "step": 9979
    },
    {
      "epoch": 0.8957099264045952,
      "grad_norm": 2.9617085456848145,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 9980
    },
    {
      "epoch": 0.8957996768982229,
      "grad_norm": 2.6102712154388428,
      "learning_rate": 0.0,
      "loss": 0.6944,
      "step": 9981
    },
    {
      "epoch": 0.8958894273918506,
      "grad_norm": 2.9267961978912354,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 9982
    },
    {
      "epoch": 0.8959791778854783,
      "grad_norm": 2.4954817295074463,
      "learning_rate": 0.0,
      "loss": 0.682,
      "step": 9983
    },
    {
      "epoch": 0.8960689283791061,
      "grad_norm": 4.8087053298950195,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 9984
    },
    {
      "epoch": 0.8961586788727338,
      "grad_norm": 2.7066457271575928,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 9985
    },
    {
      "epoch": 0.8962484293663615,
      "grad_norm": 5.871907711029053,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 9986
    },
    {
      "epoch": 0.8963381798599892,
      "grad_norm": 2.7415664196014404,
      "learning_rate": 0.0,
      "loss": 0.7672,
      "step": 9987
    },
    {
      "epoch": 0.8964279303536169,
      "grad_norm": 2.770646095275879,
      "learning_rate": 0.0,
      "loss": 0.7284,
      "step": 9988
    },
    {
      "epoch": 0.8965176808472447,
      "grad_norm": 3.7765426635742188,
      "learning_rate": 0.0,
      "loss": 0.6423,
      "step": 9989
    },
    {
      "epoch": 0.8966074313408724,
      "grad_norm": 3.347414493560791,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 9990
    },
    {
      "epoch": 0.8966971818345001,
      "grad_norm": 3.2425665855407715,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 9991
    },
    {
      "epoch": 0.8967869323281278,
      "grad_norm": 2.959550142288208,
      "learning_rate": 0.0,
      "loss": 0.7458,
      "step": 9992
    },
    {
      "epoch": 0.8968766828217555,
      "grad_norm": 3.6725399494171143,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 9993
    },
    {
      "epoch": 0.8969664333153833,
      "grad_norm": 2.411914348602295,
      "learning_rate": 0.0,
      "loss": 0.653,
      "step": 9994
    },
    {
      "epoch": 0.897056183809011,
      "grad_norm": 3.290034055709839,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 9995
    },
    {
      "epoch": 0.8971459343026387,
      "grad_norm": 2.4588801860809326,
      "learning_rate": 0.0,
      "loss": 0.651,
      "step": 9996
    },
    {
      "epoch": 0.8972356847962664,
      "grad_norm": 2.8794734477996826,
      "learning_rate": 0.0,
      "loss": 0.73,
      "step": 9997
    },
    {
      "epoch": 0.8973254352898941,
      "grad_norm": 11.227107048034668,
      "learning_rate": 0.0,
      "loss": 0.6709,
      "step": 9998
    },
    {
      "epoch": 0.8974151857835219,
      "grad_norm": 2.7004075050354004,
      "learning_rate": 0.0,
      "loss": 0.7439,
      "step": 9999
    },
    {
      "epoch": 0.8975049362771496,
      "grad_norm": 3.0866339206695557,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 10000
    },
    {
      "epoch": 0.8975946867707773,
      "grad_norm": 3.0141382217407227,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 10001
    },
    {
      "epoch": 0.8976844372644049,
      "grad_norm": 3.652933120727539,
      "learning_rate": 0.0,
      "loss": 0.6793,
      "step": 10002
    },
    {
      "epoch": 0.8977741877580326,
      "grad_norm": 3.1195757389068604,
      "learning_rate": 0.0,
      "loss": 0.679,
      "step": 10003
    },
    {
      "epoch": 0.8978639382516603,
      "grad_norm": 2.7239508628845215,
      "learning_rate": 0.0,
      "loss": 0.6625,
      "step": 10004
    },
    {
      "epoch": 0.8979536887452881,
      "grad_norm": 5.533513069152832,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 10005
    },
    {
      "epoch": 0.8980434392389158,
      "grad_norm": 2.7448534965515137,
      "learning_rate": 0.0,
      "loss": 0.7925,
      "step": 10006
    },
    {
      "epoch": 0.8981331897325435,
      "grad_norm": 3.1424288749694824,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 10007
    },
    {
      "epoch": 0.8982229402261712,
      "grad_norm": 2.293781042098999,
      "learning_rate": 0.0,
      "loss": 0.6593,
      "step": 10008
    },
    {
      "epoch": 0.8983126907197989,
      "grad_norm": 2.638373374938965,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 10009
    },
    {
      "epoch": 0.8984024412134267,
      "grad_norm": 2.9284684658050537,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 10010
    },
    {
      "epoch": 0.8984921917070544,
      "grad_norm": 2.4747109413146973,
      "learning_rate": 0.0,
      "loss": 0.7184,
      "step": 10011
    },
    {
      "epoch": 0.8985819422006821,
      "grad_norm": 2.6733381748199463,
      "learning_rate": 0.0,
      "loss": 0.7109,
      "step": 10012
    },
    {
      "epoch": 0.8986716926943098,
      "grad_norm": 2.4263622760772705,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 10013
    },
    {
      "epoch": 0.8987614431879375,
      "grad_norm": 4.178660869598389,
      "learning_rate": 0.0,
      "loss": 0.7634,
      "step": 10014
    },
    {
      "epoch": 0.8988511936815653,
      "grad_norm": 4.034685134887695,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10015
    },
    {
      "epoch": 0.898940944175193,
      "grad_norm": 7.051314830780029,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 10016
    },
    {
      "epoch": 0.8990306946688207,
      "grad_norm": 3.210188150405884,
      "learning_rate": 0.0,
      "loss": 0.6988,
      "step": 10017
    },
    {
      "epoch": 0.8991204451624484,
      "grad_norm": 2.5279035568237305,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 10018
    },
    {
      "epoch": 0.8992101956560761,
      "grad_norm": 2.8994996547698975,
      "learning_rate": 0.0,
      "loss": 0.6919,
      "step": 10019
    },
    {
      "epoch": 0.8992999461497039,
      "grad_norm": 2.8090994358062744,
      "learning_rate": 0.0,
      "loss": 0.7099,
      "step": 10020
    },
    {
      "epoch": 0.8993896966433316,
      "grad_norm": 3.7925946712493896,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 10021
    },
    {
      "epoch": 0.8994794471369593,
      "grad_norm": 3.4458749294281006,
      "learning_rate": 0.0,
      "loss": 0.7584,
      "step": 10022
    },
    {
      "epoch": 0.899569197630587,
      "grad_norm": 4.357393741607666,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 10023
    },
    {
      "epoch": 0.8996589481242147,
      "grad_norm": 2.824254035949707,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 10024
    },
    {
      "epoch": 0.8997486986178423,
      "grad_norm": 5.167358875274658,
      "learning_rate": 0.0,
      "loss": 0.7126,
      "step": 10025
    },
    {
      "epoch": 0.8998384491114702,
      "grad_norm": 3.1364047527313232,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 10026
    },
    {
      "epoch": 0.8999281996050978,
      "grad_norm": 2.719299554824829,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 10027
    },
    {
      "epoch": 0.9000179500987255,
      "grad_norm": 4.909383296966553,
      "learning_rate": 0.0,
      "loss": 0.6621,
      "step": 10028
    },
    {
      "epoch": 0.9001077005923532,
      "grad_norm": 3.928455114364624,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 10029
    },
    {
      "epoch": 0.9001974510859809,
      "grad_norm": 3.228109836578369,
      "learning_rate": 0.0,
      "loss": 0.725,
      "step": 10030
    },
    {
      "epoch": 0.9002872015796087,
      "grad_norm": 3.0860562324523926,
      "learning_rate": 0.0,
      "loss": 0.7399,
      "step": 10031
    },
    {
      "epoch": 0.9003769520732364,
      "grad_norm": 3.09167742729187,
      "learning_rate": 0.0,
      "loss": 0.7516,
      "step": 10032
    },
    {
      "epoch": 0.9004667025668641,
      "grad_norm": 2.689293384552002,
      "learning_rate": 0.0,
      "loss": 0.7581,
      "step": 10033
    },
    {
      "epoch": 0.9005564530604918,
      "grad_norm": 3.218040704727173,
      "learning_rate": 0.0,
      "loss": 0.7813,
      "step": 10034
    },
    {
      "epoch": 0.9006462035541195,
      "grad_norm": 3.2648658752441406,
      "learning_rate": 0.0,
      "loss": 0.7086,
      "step": 10035
    },
    {
      "epoch": 0.9007359540477473,
      "grad_norm": 2.7184836864471436,
      "learning_rate": 0.0,
      "loss": 0.7573,
      "step": 10036
    },
    {
      "epoch": 0.900825704541375,
      "grad_norm": 2.731344223022461,
      "learning_rate": 0.0,
      "loss": 0.6698,
      "step": 10037
    },
    {
      "epoch": 0.9009154550350027,
      "grad_norm": 2.977799415588379,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 10038
    },
    {
      "epoch": 0.9010052055286304,
      "grad_norm": 2.801001787185669,
      "learning_rate": 0.0,
      "loss": 0.6672,
      "step": 10039
    },
    {
      "epoch": 0.9010949560222581,
      "grad_norm": 3.156843423843384,
      "learning_rate": 0.0,
      "loss": 0.733,
      "step": 10040
    },
    {
      "epoch": 0.9011847065158859,
      "grad_norm": 4.134082317352295,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 10041
    },
    {
      "epoch": 0.9012744570095136,
      "grad_norm": 2.610212564468384,
      "learning_rate": 0.0,
      "loss": 0.7118,
      "step": 10042
    },
    {
      "epoch": 0.9013642075031413,
      "grad_norm": 3.2870044708251953,
      "learning_rate": 0.0,
      "loss": 0.7293,
      "step": 10043
    },
    {
      "epoch": 0.901453957996769,
      "grad_norm": 2.7645552158355713,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 10044
    },
    {
      "epoch": 0.9015437084903967,
      "grad_norm": 3.197070837020874,
      "learning_rate": 0.0,
      "loss": 0.697,
      "step": 10045
    },
    {
      "epoch": 0.9016334589840244,
      "grad_norm": 3.1058876514434814,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 10046
    },
    {
      "epoch": 0.9017232094776522,
      "grad_norm": 2.403412342071533,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 10047
    },
    {
      "epoch": 0.9018129599712799,
      "grad_norm": 5.307126522064209,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 10048
    },
    {
      "epoch": 0.9019027104649076,
      "grad_norm": 9.619297981262207,
      "learning_rate": 0.0,
      "loss": 0.6697,
      "step": 10049
    },
    {
      "epoch": 0.9019924609585352,
      "grad_norm": 11.852752685546875,
      "learning_rate": 0.0,
      "loss": 0.7684,
      "step": 10050
    },
    {
      "epoch": 0.9020822114521629,
      "grad_norm": 2.8520374298095703,
      "learning_rate": 0.0,
      "loss": 0.7756,
      "step": 10051
    },
    {
      "epoch": 0.9021719619457907,
      "grad_norm": 3.1651217937469482,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 10052
    },
    {
      "epoch": 0.9022617124394184,
      "grad_norm": 2.4597017765045166,
      "learning_rate": 0.0,
      "loss": 0.6578,
      "step": 10053
    },
    {
      "epoch": 0.9023514629330461,
      "grad_norm": 2.5241990089416504,
      "learning_rate": 0.0,
      "loss": 0.6888,
      "step": 10054
    },
    {
      "epoch": 0.9024412134266738,
      "grad_norm": 2.7831194400787354,
      "learning_rate": 0.0,
      "loss": 0.6678,
      "step": 10055
    },
    {
      "epoch": 0.9025309639203015,
      "grad_norm": 2.5779197216033936,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 10056
    },
    {
      "epoch": 0.9026207144139293,
      "grad_norm": 2.8072988986968994,
      "learning_rate": 0.0,
      "loss": 0.7412,
      "step": 10057
    },
    {
      "epoch": 0.902710464907557,
      "grad_norm": 2.6810643672943115,
      "learning_rate": 0.0,
      "loss": 0.7015,
      "step": 10058
    },
    {
      "epoch": 0.9028002154011847,
      "grad_norm": 4.601724624633789,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 10059
    },
    {
      "epoch": 0.9028899658948124,
      "grad_norm": 2.237393379211426,
      "learning_rate": 0.0,
      "loss": 0.7057,
      "step": 10060
    },
    {
      "epoch": 0.9029797163884401,
      "grad_norm": 8.031628608703613,
      "learning_rate": 0.0,
      "loss": 0.7172,
      "step": 10061
    },
    {
      "epoch": 0.9030694668820679,
      "grad_norm": 3.291378974914551,
      "learning_rate": 0.0,
      "loss": 0.6794,
      "step": 10062
    },
    {
      "epoch": 0.9031592173756956,
      "grad_norm": 3.50329852104187,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 10063
    },
    {
      "epoch": 0.9032489678693233,
      "grad_norm": 3.866290807723999,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 10064
    },
    {
      "epoch": 0.903338718362951,
      "grad_norm": 3.627969264984131,
      "learning_rate": 0.0,
      "loss": 0.6734,
      "step": 10065
    },
    {
      "epoch": 0.9034284688565787,
      "grad_norm": 3.2269325256347656,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 10066
    },
    {
      "epoch": 0.9035182193502064,
      "grad_norm": 5.336329936981201,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 10067
    },
    {
      "epoch": 0.9036079698438342,
      "grad_norm": 3.407447338104248,
      "learning_rate": 0.0,
      "loss": 0.6721,
      "step": 10068
    },
    {
      "epoch": 0.9036977203374619,
      "grad_norm": 2.9300425052642822,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10069
    },
    {
      "epoch": 0.9037874708310896,
      "grad_norm": 2.9290385246276855,
      "learning_rate": 0.0,
      "loss": 0.7124,
      "step": 10070
    },
    {
      "epoch": 0.9038772213247173,
      "grad_norm": 3.5935733318328857,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 10071
    },
    {
      "epoch": 0.903966971818345,
      "grad_norm": 3.999493360519409,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 10072
    },
    {
      "epoch": 0.9040567223119728,
      "grad_norm": 3.209001302719116,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 10073
    },
    {
      "epoch": 0.9041464728056005,
      "grad_norm": 14.616671562194824,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 10074
    },
    {
      "epoch": 0.9042362232992281,
      "grad_norm": 3.7365376949310303,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 10075
    },
    {
      "epoch": 0.9043259737928558,
      "grad_norm": 2.9464194774627686,
      "learning_rate": 0.0,
      "loss": 0.7661,
      "step": 10076
    },
    {
      "epoch": 0.9044157242864835,
      "grad_norm": 6.221938133239746,
      "learning_rate": 0.0,
      "loss": 0.7442,
      "step": 10077
    },
    {
      "epoch": 0.9045054747801113,
      "grad_norm": 3.9543349742889404,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 10078
    },
    {
      "epoch": 0.904595225273739,
      "grad_norm": 3.422245740890503,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 10079
    },
    {
      "epoch": 0.9046849757673667,
      "grad_norm": 5.820250511169434,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 10080
    },
    {
      "epoch": 0.9047747262609944,
      "grad_norm": 2.6391797065734863,
      "learning_rate": 0.0,
      "loss": 0.7057,
      "step": 10081
    },
    {
      "epoch": 0.9048644767546221,
      "grad_norm": 3.4738943576812744,
      "learning_rate": 0.0,
      "loss": 0.7452,
      "step": 10082
    },
    {
      "epoch": 0.9049542272482499,
      "grad_norm": 3.753166913986206,
      "learning_rate": 0.0,
      "loss": 0.7404,
      "step": 10083
    },
    {
      "epoch": 0.9050439777418776,
      "grad_norm": 3.652850389480591,
      "learning_rate": 0.0,
      "loss": 0.7363,
      "step": 10084
    },
    {
      "epoch": 0.9051337282355053,
      "grad_norm": 2.4947829246520996,
      "learning_rate": 0.0,
      "loss": 0.6538,
      "step": 10085
    },
    {
      "epoch": 0.905223478729133,
      "grad_norm": 2.8904478549957275,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 10086
    },
    {
      "epoch": 0.9053132292227607,
      "grad_norm": 3.047642946243286,
      "learning_rate": 0.0,
      "loss": 0.7487,
      "step": 10087
    },
    {
      "epoch": 0.9054029797163884,
      "grad_norm": 8.553348541259766,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 10088
    },
    {
      "epoch": 0.9054927302100162,
      "grad_norm": 3.729343891143799,
      "learning_rate": 0.0,
      "loss": 0.7449,
      "step": 10089
    },
    {
      "epoch": 0.9055824807036439,
      "grad_norm": 3.1410319805145264,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 10090
    },
    {
      "epoch": 0.9056722311972716,
      "grad_norm": 5.542664051055908,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 10091
    },
    {
      "epoch": 0.9057619816908993,
      "grad_norm": 4.505124092102051,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 10092
    },
    {
      "epoch": 0.905851732184527,
      "grad_norm": 4.052069187164307,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 10093
    },
    {
      "epoch": 0.9059414826781548,
      "grad_norm": 6.15771484375,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 10094
    },
    {
      "epoch": 0.9060312331717825,
      "grad_norm": 4.412780284881592,
      "learning_rate": 0.0,
      "loss": 0.7751,
      "step": 10095
    },
    {
      "epoch": 0.9061209836654102,
      "grad_norm": 3.124404191970825,
      "learning_rate": 0.0,
      "loss": 0.7324,
      "step": 10096
    },
    {
      "epoch": 0.9062107341590379,
      "grad_norm": 3.172375202178955,
      "learning_rate": 0.0,
      "loss": 0.7077,
      "step": 10097
    },
    {
      "epoch": 0.9063004846526656,
      "grad_norm": 2.8777475357055664,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 10098
    },
    {
      "epoch": 0.9063902351462934,
      "grad_norm": 3.7279152870178223,
      "learning_rate": 0.0,
      "loss": 0.6419,
      "step": 10099
    },
    {
      "epoch": 0.906479985639921,
      "grad_norm": 2.4861490726470947,
      "learning_rate": 0.0,
      "loss": 0.6866,
      "step": 10100
    },
    {
      "epoch": 0.9065697361335487,
      "grad_norm": 3.6362881660461426,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 10101
    },
    {
      "epoch": 0.9066594866271764,
      "grad_norm": 3.440657377243042,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 10102
    },
    {
      "epoch": 0.9067492371208041,
      "grad_norm": 4.0064616203308105,
      "learning_rate": 0.0,
      "loss": 0.6695,
      "step": 10103
    },
    {
      "epoch": 0.9068389876144318,
      "grad_norm": 5.378572940826416,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 10104
    },
    {
      "epoch": 0.9069287381080596,
      "grad_norm": 3.151083469390869,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 10105
    },
    {
      "epoch": 0.9070184886016873,
      "grad_norm": 3.7395269870758057,
      "learning_rate": 0.0,
      "loss": 0.6893,
      "step": 10106
    },
    {
      "epoch": 0.907108239095315,
      "grad_norm": 2.2738215923309326,
      "learning_rate": 0.0,
      "loss": 0.7685,
      "step": 10107
    },
    {
      "epoch": 0.9071979895889427,
      "grad_norm": 3.419149875640869,
      "learning_rate": 0.0,
      "loss": 0.6967,
      "step": 10108
    },
    {
      "epoch": 0.9072877400825704,
      "grad_norm": 3.2896974086761475,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 10109
    },
    {
      "epoch": 0.9073774905761982,
      "grad_norm": 3.162022352218628,
      "learning_rate": 0.0,
      "loss": 0.6914,
      "step": 10110
    },
    {
      "epoch": 0.9074672410698259,
      "grad_norm": 2.976503849029541,
      "learning_rate": 0.0,
      "loss": 0.7108,
      "step": 10111
    },
    {
      "epoch": 0.9075569915634536,
      "grad_norm": 2.4892425537109375,
      "learning_rate": 0.0,
      "loss": 0.6766,
      "step": 10112
    },
    {
      "epoch": 0.9076467420570813,
      "grad_norm": 3.0232229232788086,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 10113
    },
    {
      "epoch": 0.907736492550709,
      "grad_norm": 3.5693600177764893,
      "learning_rate": 0.0,
      "loss": 0.7401,
      "step": 10114
    },
    {
      "epoch": 0.9078262430443368,
      "grad_norm": 3.659461259841919,
      "learning_rate": 0.0,
      "loss": 0.6539,
      "step": 10115
    },
    {
      "epoch": 0.9079159935379645,
      "grad_norm": 3.0974090099334717,
      "learning_rate": 0.0,
      "loss": 0.7288,
      "step": 10116
    },
    {
      "epoch": 0.9080057440315922,
      "grad_norm": 3.453045129776001,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 10117
    },
    {
      "epoch": 0.9080954945252199,
      "grad_norm": 2.79557728767395,
      "learning_rate": 0.0,
      "loss": 0.7557,
      "step": 10118
    },
    {
      "epoch": 0.9081852450188476,
      "grad_norm": 2.7175941467285156,
      "learning_rate": 0.0,
      "loss": 0.7075,
      "step": 10119
    },
    {
      "epoch": 0.9082749955124754,
      "grad_norm": 2.0794763565063477,
      "learning_rate": 0.0,
      "loss": 0.7364,
      "step": 10120
    },
    {
      "epoch": 0.9083647460061031,
      "grad_norm": 3.2579593658447266,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 10121
    },
    {
      "epoch": 0.9084544964997308,
      "grad_norm": 2.985043525695801,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 10122
    },
    {
      "epoch": 0.9085442469933585,
      "grad_norm": 2.6576616764068604,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 10123
    },
    {
      "epoch": 0.9086339974869861,
      "grad_norm": 3.8569447994232178,
      "learning_rate": 0.0,
      "loss": 0.713,
      "step": 10124
    },
    {
      "epoch": 0.9087237479806138,
      "grad_norm": 5.180644989013672,
      "learning_rate": 0.0,
      "loss": 0.6897,
      "step": 10125
    },
    {
      "epoch": 0.9088134984742416,
      "grad_norm": 2.6534318923950195,
      "learning_rate": 0.0,
      "loss": 0.6888,
      "step": 10126
    },
    {
      "epoch": 0.9089032489678693,
      "grad_norm": 2.338066339492798,
      "learning_rate": 0.0,
      "loss": 0.7265,
      "step": 10127
    },
    {
      "epoch": 0.908992999461497,
      "grad_norm": 3.8015151023864746,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 10128
    },
    {
      "epoch": 0.9090827499551247,
      "grad_norm": 3.706685781478882,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 10129
    },
    {
      "epoch": 0.9091725004487524,
      "grad_norm": 2.9205546379089355,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 10130
    },
    {
      "epoch": 0.9092622509423802,
      "grad_norm": 2.8885014057159424,
      "learning_rate": 0.0,
      "loss": 0.6782,
      "step": 10131
    },
    {
      "epoch": 0.9093520014360079,
      "grad_norm": 4.273576736450195,
      "learning_rate": 0.0,
      "loss": 0.688,
      "step": 10132
    },
    {
      "epoch": 0.9094417519296356,
      "grad_norm": 3.504739284515381,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 10133
    },
    {
      "epoch": 0.9095315024232633,
      "grad_norm": 3.531097412109375,
      "learning_rate": 0.0,
      "loss": 0.8563,
      "step": 10134
    },
    {
      "epoch": 0.909621252916891,
      "grad_norm": 3.0608527660369873,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 10135
    },
    {
      "epoch": 0.9097110034105188,
      "grad_norm": 3.336928606033325,
      "learning_rate": 0.0,
      "loss": 0.741,
      "step": 10136
    },
    {
      "epoch": 0.9098007539041465,
      "grad_norm": 2.7274038791656494,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 10137
    },
    {
      "epoch": 0.9098905043977742,
      "grad_norm": 2.2524635791778564,
      "learning_rate": 0.0,
      "loss": 0.6825,
      "step": 10138
    },
    {
      "epoch": 0.9099802548914019,
      "grad_norm": 3.1516268253326416,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 10139
    },
    {
      "epoch": 0.9100700053850296,
      "grad_norm": 4.99823522567749,
      "learning_rate": 0.0,
      "loss": 0.6797,
      "step": 10140
    },
    {
      "epoch": 0.9101597558786574,
      "grad_norm": 2.9225711822509766,
      "learning_rate": 0.0,
      "loss": 0.8028,
      "step": 10141
    },
    {
      "epoch": 0.9102495063722851,
      "grad_norm": 3.3036959171295166,
      "learning_rate": 0.0,
      "loss": 0.7141,
      "step": 10142
    },
    {
      "epoch": 0.9103392568659128,
      "grad_norm": 3.350227117538452,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 10143
    },
    {
      "epoch": 0.9104290073595405,
      "grad_norm": 3.3346245288848877,
      "learning_rate": 0.0,
      "loss": 0.6992,
      "step": 10144
    },
    {
      "epoch": 0.9105187578531682,
      "grad_norm": 2.3591954708099365,
      "learning_rate": 0.0,
      "loss": 0.7159,
      "step": 10145
    },
    {
      "epoch": 0.9106085083467959,
      "grad_norm": 2.5409598350524902,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 10146
    },
    {
      "epoch": 0.9106982588404237,
      "grad_norm": 2.884538173675537,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 10147
    },
    {
      "epoch": 0.9107880093340514,
      "grad_norm": 4.654118537902832,
      "learning_rate": 0.0,
      "loss": 0.7366,
      "step": 10148
    },
    {
      "epoch": 0.910877759827679,
      "grad_norm": 3.196575164794922,
      "learning_rate": 0.0,
      "loss": 0.7365,
      "step": 10149
    },
    {
      "epoch": 0.9109675103213067,
      "grad_norm": 2.8969717025756836,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10150
    },
    {
      "epoch": 0.9110572608149344,
      "grad_norm": 3.006758213043213,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 10151
    },
    {
      "epoch": 0.9111470113085622,
      "grad_norm": 2.704314947128296,
      "learning_rate": 0.0,
      "loss": 0.6695,
      "step": 10152
    },
    {
      "epoch": 0.9112367618021899,
      "grad_norm": 2.767784595489502,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 10153
    },
    {
      "epoch": 0.9113265122958176,
      "grad_norm": 3.533330202102661,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 10154
    },
    {
      "epoch": 0.9114162627894453,
      "grad_norm": 2.9397499561309814,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 10155
    },
    {
      "epoch": 0.911506013283073,
      "grad_norm": 3.3707520961761475,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 10156
    },
    {
      "epoch": 0.9115957637767008,
      "grad_norm": 3.272007942199707,
      "learning_rate": 0.0,
      "loss": 0.7802,
      "step": 10157
    },
    {
      "epoch": 0.9116855142703285,
      "grad_norm": 3.1304128170013428,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 10158
    },
    {
      "epoch": 0.9117752647639562,
      "grad_norm": 2.814103364944458,
      "learning_rate": 0.0,
      "loss": 0.7218,
      "step": 10159
    },
    {
      "epoch": 0.9118650152575839,
      "grad_norm": 2.484588623046875,
      "learning_rate": 0.0,
      "loss": 0.7051,
      "step": 10160
    },
    {
      "epoch": 0.9119547657512116,
      "grad_norm": 4.484694480895996,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 10161
    },
    {
      "epoch": 0.9120445162448394,
      "grad_norm": 3.5008556842803955,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 10162
    },
    {
      "epoch": 0.9121342667384671,
      "grad_norm": 3.531963348388672,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 10163
    },
    {
      "epoch": 0.9122240172320948,
      "grad_norm": 3.1975371837615967,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 10164
    },
    {
      "epoch": 0.9123137677257225,
      "grad_norm": 3.3959856033325195,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 10165
    },
    {
      "epoch": 0.9124035182193502,
      "grad_norm": 3.286102533340454,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 10166
    },
    {
      "epoch": 0.9124932687129779,
      "grad_norm": 3.8568384647369385,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 10167
    },
    {
      "epoch": 0.9125830192066057,
      "grad_norm": 2.809938907623291,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 10168
    },
    {
      "epoch": 0.9126727697002334,
      "grad_norm": 2.30076265335083,
      "learning_rate": 0.0,
      "loss": 0.6213,
      "step": 10169
    },
    {
      "epoch": 0.9127625201938611,
      "grad_norm": 3.5074198246002197,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 10170
    },
    {
      "epoch": 0.9128522706874888,
      "grad_norm": 3.451009511947632,
      "learning_rate": 0.0,
      "loss": 0.6245,
      "step": 10171
    },
    {
      "epoch": 0.9129420211811164,
      "grad_norm": 2.895697593688965,
      "learning_rate": 0.0,
      "loss": 0.6924,
      "step": 10172
    },
    {
      "epoch": 0.9130317716747443,
      "grad_norm": 3.6157705783843994,
      "learning_rate": 0.0,
      "loss": 0.6982,
      "step": 10173
    },
    {
      "epoch": 0.913121522168372,
      "grad_norm": 3.9231297969818115,
      "learning_rate": 0.0,
      "loss": 0.7163,
      "step": 10174
    },
    {
      "epoch": 0.9132112726619996,
      "grad_norm": 3.8118655681610107,
      "learning_rate": 0.0,
      "loss": 0.7715,
      "step": 10175
    },
    {
      "epoch": 0.9133010231556273,
      "grad_norm": 3.1196413040161133,
      "learning_rate": 0.0,
      "loss": 0.7112,
      "step": 10176
    },
    {
      "epoch": 0.913390773649255,
      "grad_norm": 3.1094040870666504,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 10177
    },
    {
      "epoch": 0.9134805241428828,
      "grad_norm": 3.227642774581909,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 10178
    },
    {
      "epoch": 0.9135702746365105,
      "grad_norm": 4.405918121337891,
      "learning_rate": 0.0,
      "loss": 0.6966,
      "step": 10179
    },
    {
      "epoch": 0.9136600251301382,
      "grad_norm": 3.372072458267212,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 10180
    },
    {
      "epoch": 0.9137497756237659,
      "grad_norm": 3.8331868648529053,
      "learning_rate": 0.0,
      "loss": 0.7636,
      "step": 10181
    },
    {
      "epoch": 0.9138395261173936,
      "grad_norm": 3.902714252471924,
      "learning_rate": 0.0,
      "loss": 0.6915,
      "step": 10182
    },
    {
      "epoch": 0.9139292766110214,
      "grad_norm": 3.029968500137329,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 10183
    },
    {
      "epoch": 0.9140190271046491,
      "grad_norm": 3.2292275428771973,
      "learning_rate": 0.0,
      "loss": 0.7014,
      "step": 10184
    },
    {
      "epoch": 0.9141087775982768,
      "grad_norm": 3.238468885421753,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 10185
    },
    {
      "epoch": 0.9141985280919045,
      "grad_norm": 3.6298019886016846,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 10186
    },
    {
      "epoch": 0.9142882785855322,
      "grad_norm": 2.9321835041046143,
      "learning_rate": 0.0,
      "loss": 0.5808,
      "step": 10187
    },
    {
      "epoch": 0.9143780290791599,
      "grad_norm": 2.9686739444732666,
      "learning_rate": 0.0,
      "loss": 0.7219,
      "step": 10188
    },
    {
      "epoch": 0.9144677795727877,
      "grad_norm": 2.8831262588500977,
      "learning_rate": 0.0,
      "loss": 0.6887,
      "step": 10189
    },
    {
      "epoch": 0.9145575300664154,
      "grad_norm": 3.4498207569122314,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 10190
    },
    {
      "epoch": 0.9146472805600431,
      "grad_norm": 3.8437836170196533,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 10191
    },
    {
      "epoch": 0.9147370310536708,
      "grad_norm": 3.479013681411743,
      "learning_rate": 0.0,
      "loss": 0.7419,
      "step": 10192
    },
    {
      "epoch": 0.9148267815472985,
      "grad_norm": 2.7114789485931396,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 10193
    },
    {
      "epoch": 0.9149165320409263,
      "grad_norm": 3.508896827697754,
      "learning_rate": 0.0,
      "loss": 0.6961,
      "step": 10194
    },
    {
      "epoch": 0.915006282534554,
      "grad_norm": 4.461129665374756,
      "learning_rate": 0.0,
      "loss": 0.7465,
      "step": 10195
    },
    {
      "epoch": 0.9150960330281817,
      "grad_norm": 3.050344944000244,
      "learning_rate": 0.0,
      "loss": 0.6626,
      "step": 10196
    },
    {
      "epoch": 0.9151857835218093,
      "grad_norm": 3.3041903972625732,
      "learning_rate": 0.0,
      "loss": 0.7238,
      "step": 10197
    },
    {
      "epoch": 0.915275534015437,
      "grad_norm": 3.4630744457244873,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 10198
    },
    {
      "epoch": 0.9153652845090648,
      "grad_norm": 3.459973096847534,
      "learning_rate": 0.0,
      "loss": 0.7823,
      "step": 10199
    },
    {
      "epoch": 0.9154550350026925,
      "grad_norm": 3.667301654815674,
      "learning_rate": 0.0,
      "loss": 0.6832,
      "step": 10200
    },
    {
      "epoch": 0.9155447854963202,
      "grad_norm": 3.8406453132629395,
      "learning_rate": 0.0,
      "loss": 0.7378,
      "step": 10201
    },
    {
      "epoch": 0.9156345359899479,
      "grad_norm": 3.1502225399017334,
      "learning_rate": 0.0,
      "loss": 0.6416,
      "step": 10202
    },
    {
      "epoch": 0.9157242864835756,
      "grad_norm": 3.1877119541168213,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 10203
    },
    {
      "epoch": 0.9158140369772034,
      "grad_norm": 3.790532112121582,
      "learning_rate": 0.0,
      "loss": 0.6896,
      "step": 10204
    },
    {
      "epoch": 0.9159037874708311,
      "grad_norm": 2.6501452922821045,
      "learning_rate": 0.0,
      "loss": 0.6587,
      "step": 10205
    },
    {
      "epoch": 0.9159935379644588,
      "grad_norm": 2.4473252296447754,
      "learning_rate": 0.0,
      "loss": 0.7436,
      "step": 10206
    },
    {
      "epoch": 0.9160832884580865,
      "grad_norm": 3.011867046356201,
      "learning_rate": 0.0,
      "loss": 0.6747,
      "step": 10207
    },
    {
      "epoch": 0.9161730389517142,
      "grad_norm": 2.462745428085327,
      "learning_rate": 0.0,
      "loss": 0.7136,
      "step": 10208
    },
    {
      "epoch": 0.9162627894453419,
      "grad_norm": 2.817065954208374,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 10209
    },
    {
      "epoch": 0.9163525399389697,
      "grad_norm": 2.6640124320983887,
      "learning_rate": 0.0,
      "loss": 0.7358,
      "step": 10210
    },
    {
      "epoch": 0.9164422904325974,
      "grad_norm": 2.674318790435791,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 10211
    },
    {
      "epoch": 0.9165320409262251,
      "grad_norm": 3.6329963207244873,
      "learning_rate": 0.0,
      "loss": 0.67,
      "step": 10212
    },
    {
      "epoch": 0.9166217914198528,
      "grad_norm": 2.998854398727417,
      "learning_rate": 0.0,
      "loss": 0.6814,
      "step": 10213
    },
    {
      "epoch": 0.9167115419134805,
      "grad_norm": 2.5104973316192627,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 10214
    },
    {
      "epoch": 0.9168012924071083,
      "grad_norm": 2.8112730979919434,
      "learning_rate": 0.0,
      "loss": 0.6953,
      "step": 10215
    },
    {
      "epoch": 0.916891042900736,
      "grad_norm": 3.302190065383911,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 10216
    },
    {
      "epoch": 0.9169807933943637,
      "grad_norm": 3.271552324295044,
      "learning_rate": 0.0,
      "loss": 0.6894,
      "step": 10217
    },
    {
      "epoch": 0.9170705438879914,
      "grad_norm": 2.8768603801727295,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 10218
    },
    {
      "epoch": 0.9171602943816191,
      "grad_norm": 2.6678566932678223,
      "learning_rate": 0.0,
      "loss": 0.6733,
      "step": 10219
    },
    {
      "epoch": 0.9172500448752469,
      "grad_norm": 3.583522081375122,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 10220
    },
    {
      "epoch": 0.9173397953688746,
      "grad_norm": 2.9904966354370117,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 10221
    },
    {
      "epoch": 0.9174295458625022,
      "grad_norm": 2.182698965072632,
      "learning_rate": 0.0,
      "loss": 0.7031,
      "step": 10222
    },
    {
      "epoch": 0.9175192963561299,
      "grad_norm": 3.454010248184204,
      "learning_rate": 0.0,
      "loss": 0.693,
      "step": 10223
    },
    {
      "epoch": 0.9176090468497576,
      "grad_norm": 5.727562427520752,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 10224
    },
    {
      "epoch": 0.9176987973433854,
      "grad_norm": 4.322498798370361,
      "learning_rate": 0.0,
      "loss": 0.6882,
      "step": 10225
    },
    {
      "epoch": 0.9177885478370131,
      "grad_norm": 3.4661691188812256,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 10226
    },
    {
      "epoch": 0.9178782983306408,
      "grad_norm": 3.8146791458129883,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 10227
    },
    {
      "epoch": 0.9179680488242685,
      "grad_norm": 2.778684377670288,
      "learning_rate": 0.0,
      "loss": 0.7597,
      "step": 10228
    },
    {
      "epoch": 0.9180577993178962,
      "grad_norm": 3.350811243057251,
      "learning_rate": 0.0,
      "loss": 0.6778,
      "step": 10229
    },
    {
      "epoch": 0.9181475498115239,
      "grad_norm": 3.030493974685669,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 10230
    },
    {
      "epoch": 0.9182373003051517,
      "grad_norm": 3.1650476455688477,
      "learning_rate": 0.0,
      "loss": 0.6371,
      "step": 10231
    },
    {
      "epoch": 0.9183270507987794,
      "grad_norm": 2.71881365776062,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 10232
    },
    {
      "epoch": 0.9184168012924071,
      "grad_norm": 2.6601643562316895,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 10233
    },
    {
      "epoch": 0.9185065517860348,
      "grad_norm": 5.024848461151123,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 10234
    },
    {
      "epoch": 0.9185963022796625,
      "grad_norm": 2.858341693878174,
      "learning_rate": 0.0,
      "loss": 0.6822,
      "step": 10235
    },
    {
      "epoch": 0.9186860527732903,
      "grad_norm": 6.316837310791016,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 10236
    },
    {
      "epoch": 0.918775803266918,
      "grad_norm": 2.7088847160339355,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 10237
    },
    {
      "epoch": 0.9188655537605457,
      "grad_norm": 2.919377565383911,
      "learning_rate": 0.0,
      "loss": 0.7256,
      "step": 10238
    },
    {
      "epoch": 0.9189553042541734,
      "grad_norm": 3.475165605545044,
      "learning_rate": 0.0,
      "loss": 0.7415,
      "step": 10239
    },
    {
      "epoch": 0.9190450547478011,
      "grad_norm": 3.7804083824157715,
      "learning_rate": 0.0,
      "loss": 0.7325,
      "step": 10240
    },
    {
      "epoch": 0.9191348052414289,
      "grad_norm": 3.4272825717926025,
      "learning_rate": 0.0,
      "loss": 0.6974,
      "step": 10241
    },
    {
      "epoch": 0.9192245557350566,
      "grad_norm": 2.5131490230560303,
      "learning_rate": 0.0,
      "loss": 0.6858,
      "step": 10242
    },
    {
      "epoch": 0.9193143062286843,
      "grad_norm": 2.4699838161468506,
      "learning_rate": 0.0,
      "loss": 0.7087,
      "step": 10243
    },
    {
      "epoch": 0.919404056722312,
      "grad_norm": 2.7427656650543213,
      "learning_rate": 0.0,
      "loss": 0.6652,
      "step": 10244
    },
    {
      "epoch": 0.9194938072159397,
      "grad_norm": 3.5362539291381836,
      "learning_rate": 0.0,
      "loss": 0.6754,
      "step": 10245
    },
    {
      "epoch": 0.9195835577095675,
      "grad_norm": 3.0266847610473633,
      "learning_rate": 0.0,
      "loss": 0.6926,
      "step": 10246
    },
    {
      "epoch": 0.9196733082031952,
      "grad_norm": 2.4111499786376953,
      "learning_rate": 0.0,
      "loss": 0.7009,
      "step": 10247
    },
    {
      "epoch": 0.9197630586968228,
      "grad_norm": 2.7047531604766846,
      "learning_rate": 0.0,
      "loss": 0.6493,
      "step": 10248
    },
    {
      "epoch": 0.9198528091904505,
      "grad_norm": 2.986292600631714,
      "learning_rate": 0.0,
      "loss": 0.6393,
      "step": 10249
    },
    {
      "epoch": 0.9199425596840782,
      "grad_norm": 2.703545570373535,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 10250
    },
    {
      "epoch": 0.9200323101777059,
      "grad_norm": 2.798250198364258,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 10251
    },
    {
      "epoch": 0.9201220606713337,
      "grad_norm": 2.623756170272827,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 10252
    },
    {
      "epoch": 0.9202118111649614,
      "grad_norm": 2.169757843017578,
      "learning_rate": 0.0,
      "loss": 0.7837,
      "step": 10253
    },
    {
      "epoch": 0.9203015616585891,
      "grad_norm": 3.7386245727539062,
      "learning_rate": 0.0,
      "loss": 0.7311,
      "step": 10254
    },
    {
      "epoch": 0.9203913121522168,
      "grad_norm": 8.110010147094727,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 10255
    },
    {
      "epoch": 0.9204810626458445,
      "grad_norm": 2.83897066116333,
      "learning_rate": 0.0,
      "loss": 0.7394,
      "step": 10256
    },
    {
      "epoch": 0.9205708131394723,
      "grad_norm": 3.3265275955200195,
      "learning_rate": 0.0,
      "loss": 0.6934,
      "step": 10257
    },
    {
      "epoch": 0.9206605636331,
      "grad_norm": 3.619863271713257,
      "learning_rate": 0.0,
      "loss": 0.7042,
      "step": 10258
    },
    {
      "epoch": 0.9207503141267277,
      "grad_norm": 3.7600269317626953,
      "learning_rate": 0.0,
      "loss": 0.767,
      "step": 10259
    },
    {
      "epoch": 0.9208400646203554,
      "grad_norm": 3.3989810943603516,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 10260
    },
    {
      "epoch": 0.9209298151139831,
      "grad_norm": 3.2719056606292725,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 10261
    },
    {
      "epoch": 0.9210195656076109,
      "grad_norm": 2.5754833221435547,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 10262
    },
    {
      "epoch": 0.9211093161012386,
      "grad_norm": 8.90914535522461,
      "learning_rate": 0.0,
      "loss": 0.7237,
      "step": 10263
    },
    {
      "epoch": 0.9211990665948663,
      "grad_norm": 3.9376673698425293,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 10264
    },
    {
      "epoch": 0.921288817088494,
      "grad_norm": 2.968796491622925,
      "learning_rate": 0.0,
      "loss": 0.6722,
      "step": 10265
    },
    {
      "epoch": 0.9213785675821217,
      "grad_norm": 2.841794967651367,
      "learning_rate": 0.0,
      "loss": 0.7846,
      "step": 10266
    },
    {
      "epoch": 0.9214683180757495,
      "grad_norm": 3.576972484588623,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 10267
    },
    {
      "epoch": 0.9215580685693772,
      "grad_norm": 2.498917579650879,
      "learning_rate": 0.0,
      "loss": 0.6641,
      "step": 10268
    },
    {
      "epoch": 0.9216478190630049,
      "grad_norm": 2.4977283477783203,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 10269
    },
    {
      "epoch": 0.9217375695566326,
      "grad_norm": 3.2973148822784424,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 10270
    },
    {
      "epoch": 0.9218273200502602,
      "grad_norm": 3.40762996673584,
      "learning_rate": 0.0,
      "loss": 0.7322,
      "step": 10271
    },
    {
      "epoch": 0.9219170705438879,
      "grad_norm": 2.8004071712493896,
      "learning_rate": 0.0,
      "loss": 0.7532,
      "step": 10272
    },
    {
      "epoch": 0.9220068210375157,
      "grad_norm": 3.2091054916381836,
      "learning_rate": 0.0,
      "loss": 0.6386,
      "step": 10273
    },
    {
      "epoch": 0.9220965715311434,
      "grad_norm": 4.028809070587158,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 10274
    },
    {
      "epoch": 0.9221863220247711,
      "grad_norm": 3.6214447021484375,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 10275
    },
    {
      "epoch": 0.9222760725183988,
      "grad_norm": 2.588838815689087,
      "learning_rate": 0.0,
      "loss": 0.6758,
      "step": 10276
    },
    {
      "epoch": 0.9223658230120265,
      "grad_norm": 3.551576852798462,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 10277
    },
    {
      "epoch": 0.9224555735056543,
      "grad_norm": 2.179638147354126,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 10278
    },
    {
      "epoch": 0.922545323999282,
      "grad_norm": 3.23345947265625,
      "learning_rate": 0.0,
      "loss": 0.6771,
      "step": 10279
    },
    {
      "epoch": 0.9226350744929097,
      "grad_norm": 2.93637752532959,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 10280
    },
    {
      "epoch": 0.9227248249865374,
      "grad_norm": 5.169783115386963,
      "learning_rate": 0.0,
      "loss": 0.7522,
      "step": 10281
    },
    {
      "epoch": 0.9228145754801651,
      "grad_norm": 3.704380989074707,
      "learning_rate": 0.0,
      "loss": 0.7598,
      "step": 10282
    },
    {
      "epoch": 0.9229043259737929,
      "grad_norm": 3.9273009300231934,
      "learning_rate": 0.0,
      "loss": 0.6744,
      "step": 10283
    },
    {
      "epoch": 0.9229940764674206,
      "grad_norm": 3.4857449531555176,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 10284
    },
    {
      "epoch": 0.9230838269610483,
      "grad_norm": 15.653307914733887,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 10285
    },
    {
      "epoch": 0.923173577454676,
      "grad_norm": 3.318074941635132,
      "learning_rate": 0.0,
      "loss": 0.6609,
      "step": 10286
    },
    {
      "epoch": 0.9232633279483037,
      "grad_norm": 2.9727120399475098,
      "learning_rate": 0.0,
      "loss": 0.628,
      "step": 10287
    },
    {
      "epoch": 0.9233530784419315,
      "grad_norm": 5.140704154968262,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 10288
    },
    {
      "epoch": 0.9234428289355592,
      "grad_norm": 3.3987040519714355,
      "learning_rate": 0.0,
      "loss": 0.7156,
      "step": 10289
    },
    {
      "epoch": 0.9235325794291869,
      "grad_norm": 4.359943866729736,
      "learning_rate": 0.0,
      "loss": 0.7158,
      "step": 10290
    },
    {
      "epoch": 0.9236223299228146,
      "grad_norm": 3.2337803840637207,
      "learning_rate": 0.0,
      "loss": 0.6439,
      "step": 10291
    },
    {
      "epoch": 0.9237120804164423,
      "grad_norm": 2.8161444664001465,
      "learning_rate": 0.0,
      "loss": 0.6886,
      "step": 10292
    },
    {
      "epoch": 0.92380183091007,
      "grad_norm": 3.24418568611145,
      "learning_rate": 0.0,
      "loss": 0.7207,
      "step": 10293
    },
    {
      "epoch": 0.9238915814036978,
      "grad_norm": 3.8248188495635986,
      "learning_rate": 0.0,
      "loss": 0.6713,
      "step": 10294
    },
    {
      "epoch": 0.9239813318973255,
      "grad_norm": 2.8514962196350098,
      "learning_rate": 0.0,
      "loss": 0.6913,
      "step": 10295
    },
    {
      "epoch": 0.9240710823909531,
      "grad_norm": 3.5836548805236816,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 10296
    },
    {
      "epoch": 0.9241608328845808,
      "grad_norm": 3.2809274196624756,
      "learning_rate": 0.0,
      "loss": 0.7512,
      "step": 10297
    },
    {
      "epoch": 0.9242505833782085,
      "grad_norm": 3.0720300674438477,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 10298
    },
    {
      "epoch": 0.9243403338718363,
      "grad_norm": 3.346179246902466,
      "learning_rate": 0.0,
      "loss": 0.7672,
      "step": 10299
    },
    {
      "epoch": 0.924430084365464,
      "grad_norm": 4.272866725921631,
      "learning_rate": 0.0,
      "loss": 0.5988,
      "step": 10300
    },
    {
      "epoch": 0.9245198348590917,
      "grad_norm": 3.4980928897857666,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 10301
    },
    {
      "epoch": 0.9246095853527194,
      "grad_norm": 3.6334590911865234,
      "learning_rate": 0.0,
      "loss": 0.7409,
      "step": 10302
    },
    {
      "epoch": 0.9246993358463471,
      "grad_norm": 2.9452717304229736,
      "learning_rate": 0.0,
      "loss": 0.6664,
      "step": 10303
    },
    {
      "epoch": 0.9247890863399749,
      "grad_norm": 3.858264207839966,
      "learning_rate": 0.0,
      "loss": 0.6782,
      "step": 10304
    },
    {
      "epoch": 0.9248788368336026,
      "grad_norm": 3.0502827167510986,
      "learning_rate": 0.0,
      "loss": 0.7005,
      "step": 10305
    },
    {
      "epoch": 0.9249685873272303,
      "grad_norm": 2.643545389175415,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 10306
    },
    {
      "epoch": 0.925058337820858,
      "grad_norm": 4.113767147064209,
      "learning_rate": 0.0,
      "loss": 0.7441,
      "step": 10307
    },
    {
      "epoch": 0.9251480883144857,
      "grad_norm": 2.706782579421997,
      "learning_rate": 0.0,
      "loss": 0.6811,
      "step": 10308
    },
    {
      "epoch": 0.9252378388081135,
      "grad_norm": 3.5010266304016113,
      "learning_rate": 0.0,
      "loss": 0.7277,
      "step": 10309
    },
    {
      "epoch": 0.9253275893017412,
      "grad_norm": 2.945324420928955,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 10310
    },
    {
      "epoch": 0.9254173397953689,
      "grad_norm": 4.569239616394043,
      "learning_rate": 0.0,
      "loss": 0.7209,
      "step": 10311
    },
    {
      "epoch": 0.9255070902889966,
      "grad_norm": 3.123443126678467,
      "learning_rate": 0.0,
      "loss": 0.684,
      "step": 10312
    },
    {
      "epoch": 0.9255968407826243,
      "grad_norm": 2.4119489192962646,
      "learning_rate": 0.0,
      "loss": 0.6983,
      "step": 10313
    },
    {
      "epoch": 0.925686591276252,
      "grad_norm": 3.3145151138305664,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 10314
    },
    {
      "epoch": 0.9257763417698798,
      "grad_norm": 3.1883630752563477,
      "learning_rate": 0.0,
      "loss": 0.7054,
      "step": 10315
    },
    {
      "epoch": 0.9258660922635075,
      "grad_norm": 8.524190902709961,
      "learning_rate": 0.0,
      "loss": 0.7422,
      "step": 10316
    },
    {
      "epoch": 0.9259558427571352,
      "grad_norm": 6.090651035308838,
      "learning_rate": 0.0,
      "loss": 0.6853,
      "step": 10317
    },
    {
      "epoch": 0.9260455932507629,
      "grad_norm": 3.69857120513916,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 10318
    },
    {
      "epoch": 0.9261353437443905,
      "grad_norm": 2.538210153579712,
      "learning_rate": 0.0,
      "loss": 0.7252,
      "step": 10319
    },
    {
      "epoch": 0.9262250942380184,
      "grad_norm": 2.9818830490112305,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 10320
    },
    {
      "epoch": 0.926314844731646,
      "grad_norm": 2.9685604572296143,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 10321
    },
    {
      "epoch": 0.9264045952252737,
      "grad_norm": 2.5572352409362793,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 10322
    },
    {
      "epoch": 0.9264943457189014,
      "grad_norm": 2.5362131595611572,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 10323
    },
    {
      "epoch": 0.9265840962125291,
      "grad_norm": 2.4835240840911865,
      "learning_rate": 0.0,
      "loss": 0.7383,
      "step": 10324
    },
    {
      "epoch": 0.9266738467061569,
      "grad_norm": 2.512939453125,
      "learning_rate": 0.0,
      "loss": 0.6632,
      "step": 10325
    },
    {
      "epoch": 0.9267635971997846,
      "grad_norm": 4.078551769256592,
      "learning_rate": 0.0,
      "loss": 0.719,
      "step": 10326
    },
    {
      "epoch": 0.9268533476934123,
      "grad_norm": 12.818903923034668,
      "learning_rate": 0.0,
      "loss": 0.6712,
      "step": 10327
    },
    {
      "epoch": 0.92694309818704,
      "grad_norm": 3.254333019256592,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 10328
    },
    {
      "epoch": 0.9270328486806677,
      "grad_norm": 4.491297245025635,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 10329
    },
    {
      "epoch": 0.9271225991742954,
      "grad_norm": 3.0957424640655518,
      "learning_rate": 0.0,
      "loss": 0.6855,
      "step": 10330
    },
    {
      "epoch": 0.9272123496679232,
      "grad_norm": 2.3207767009735107,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 10331
    },
    {
      "epoch": 0.9273021001615509,
      "grad_norm": 3.1072275638580322,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 10332
    },
    {
      "epoch": 0.9273918506551786,
      "grad_norm": 5.074292182922363,
      "learning_rate": 0.0,
      "loss": 0.676,
      "step": 10333
    },
    {
      "epoch": 0.9274816011488063,
      "grad_norm": 2.959742784500122,
      "learning_rate": 0.0,
      "loss": 0.7012,
      "step": 10334
    },
    {
      "epoch": 0.927571351642434,
      "grad_norm": 108.6180648803711,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 10335
    },
    {
      "epoch": 0.9276611021360618,
      "grad_norm": 2.9784233570098877,
      "learning_rate": 0.0,
      "loss": 0.6605,
      "step": 10336
    },
    {
      "epoch": 0.9277508526296895,
      "grad_norm": 2.419617176055908,
      "learning_rate": 0.0,
      "loss": 0.6828,
      "step": 10337
    },
    {
      "epoch": 0.9278406031233172,
      "grad_norm": 3.1453375816345215,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 10338
    },
    {
      "epoch": 0.9279303536169449,
      "grad_norm": 2.742723226547241,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 10339
    },
    {
      "epoch": 0.9280201041105726,
      "grad_norm": 3.2424802780151367,
      "learning_rate": 0.0,
      "loss": 0.7529,
      "step": 10340
    },
    {
      "epoch": 0.9281098546042004,
      "grad_norm": 3.1851072311401367,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 10341
    },
    {
      "epoch": 0.9281996050978281,
      "grad_norm": 3.681300163269043,
      "learning_rate": 0.0,
      "loss": 0.7464,
      "step": 10342
    },
    {
      "epoch": 0.9282893555914558,
      "grad_norm": 5.140341758728027,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 10343
    },
    {
      "epoch": 0.9283791060850834,
      "grad_norm": 2.787576913833618,
      "learning_rate": 0.0,
      "loss": 0.7172,
      "step": 10344
    },
    {
      "epoch": 0.9284688565787111,
      "grad_norm": 4.088571548461914,
      "learning_rate": 0.0,
      "loss": 0.7423,
      "step": 10345
    },
    {
      "epoch": 0.928558607072339,
      "grad_norm": 3.3527393341064453,
      "learning_rate": 0.0,
      "loss": 0.6991,
      "step": 10346
    },
    {
      "epoch": 0.9286483575659666,
      "grad_norm": 2.3652753829956055,
      "learning_rate": 0.0,
      "loss": 0.7354,
      "step": 10347
    },
    {
      "epoch": 0.9287381080595943,
      "grad_norm": 2.6415462493896484,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 10348
    },
    {
      "epoch": 0.928827858553222,
      "grad_norm": 3.2354609966278076,
      "learning_rate": 0.0,
      "loss": 0.6854,
      "step": 10349
    },
    {
      "epoch": 0.9289176090468497,
      "grad_norm": 6.529700756072998,
      "learning_rate": 0.0,
      "loss": 0.6436,
      "step": 10350
    },
    {
      "epoch": 0.9290073595404774,
      "grad_norm": 3.361039161682129,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 10351
    },
    {
      "epoch": 0.9290971100341052,
      "grad_norm": 4.198440074920654,
      "learning_rate": 0.0,
      "loss": 0.674,
      "step": 10352
    },
    {
      "epoch": 0.9291868605277329,
      "grad_norm": 3.79638409614563,
      "learning_rate": 0.0,
      "loss": 0.6875,
      "step": 10353
    },
    {
      "epoch": 0.9292766110213606,
      "grad_norm": 2.6626100540161133,
      "learning_rate": 0.0,
      "loss": 0.6964,
      "step": 10354
    },
    {
      "epoch": 0.9293663615149883,
      "grad_norm": 3.1398162841796875,
      "learning_rate": 0.0,
      "loss": 0.6701,
      "step": 10355
    },
    {
      "epoch": 0.929456112008616,
      "grad_norm": 3.110940456390381,
      "learning_rate": 0.0,
      "loss": 0.6742,
      "step": 10356
    },
    {
      "epoch": 0.9295458625022438,
      "grad_norm": 4.993081092834473,
      "learning_rate": 0.0,
      "loss": 0.6877,
      "step": 10357
    },
    {
      "epoch": 0.9296356129958715,
      "grad_norm": 3.3289401531219482,
      "learning_rate": 0.0,
      "loss": 0.6811,
      "step": 10358
    },
    {
      "epoch": 0.9297253634894992,
      "grad_norm": 3.629962921142578,
      "learning_rate": 0.0,
      "loss": 0.655,
      "step": 10359
    },
    {
      "epoch": 0.9298151139831269,
      "grad_norm": 3.0069658756256104,
      "learning_rate": 0.0,
      "loss": 0.6735,
      "step": 10360
    },
    {
      "epoch": 0.9299048644767546,
      "grad_norm": 3.178330421447754,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 10361
    },
    {
      "epoch": 0.9299946149703824,
      "grad_norm": 8.474714279174805,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 10362
    },
    {
      "epoch": 0.9300843654640101,
      "grad_norm": 3.353675127029419,
      "learning_rate": 0.0,
      "loss": 0.7513,
      "step": 10363
    },
    {
      "epoch": 0.9301741159576378,
      "grad_norm": 2.620784282684326,
      "learning_rate": 0.0,
      "loss": 0.6649,
      "step": 10364
    },
    {
      "epoch": 0.9302638664512655,
      "grad_norm": 2.551725387573242,
      "learning_rate": 0.0,
      "loss": 0.7628,
      "step": 10365
    },
    {
      "epoch": 0.9303536169448932,
      "grad_norm": 3.044853925704956,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 10366
    },
    {
      "epoch": 0.930443367438521,
      "grad_norm": 2.956824541091919,
      "learning_rate": 0.0,
      "loss": 0.7956,
      "step": 10367
    },
    {
      "epoch": 0.9305331179321487,
      "grad_norm": 2.5079004764556885,
      "learning_rate": 0.0,
      "loss": 0.7055,
      "step": 10368
    },
    {
      "epoch": 0.9306228684257764,
      "grad_norm": 7.1651740074157715,
      "learning_rate": 0.0,
      "loss": 0.7081,
      "step": 10369
    },
    {
      "epoch": 0.930712618919404,
      "grad_norm": 3.229098320007324,
      "learning_rate": 0.0,
      "loss": 0.6887,
      "step": 10370
    },
    {
      "epoch": 0.9308023694130317,
      "grad_norm": 3.320488214492798,
      "learning_rate": 0.0,
      "loss": 0.6691,
      "step": 10371
    },
    {
      "epoch": 0.9308921199066594,
      "grad_norm": 3.342498779296875,
      "learning_rate": 0.0,
      "loss": 0.6882,
      "step": 10372
    },
    {
      "epoch": 0.9309818704002872,
      "grad_norm": 3.0611958503723145,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 10373
    },
    {
      "epoch": 0.9310716208939149,
      "grad_norm": 4.207193851470947,
      "learning_rate": 0.0,
      "loss": 0.6738,
      "step": 10374
    },
    {
      "epoch": 0.9311613713875426,
      "grad_norm": 3.1810550689697266,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 10375
    },
    {
      "epoch": 0.9312511218811703,
      "grad_norm": 3.1676387786865234,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 10376
    },
    {
      "epoch": 0.931340872374798,
      "grad_norm": 2.942355155944824,
      "learning_rate": 0.0,
      "loss": 0.7327,
      "step": 10377
    },
    {
      "epoch": 0.9314306228684258,
      "grad_norm": 3.2349109649658203,
      "learning_rate": 0.0,
      "loss": 0.7739,
      "step": 10378
    },
    {
      "epoch": 0.9315203733620535,
      "grad_norm": 2.602916955947876,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 10379
    },
    {
      "epoch": 0.9316101238556812,
      "grad_norm": 4.708733558654785,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 10380
    },
    {
      "epoch": 0.9316998743493089,
      "grad_norm": 9.002306938171387,
      "learning_rate": 0.0,
      "loss": 0.6836,
      "step": 10381
    },
    {
      "epoch": 0.9317896248429366,
      "grad_norm": 4.683501243591309,
      "learning_rate": 0.0,
      "loss": 0.7413,
      "step": 10382
    },
    {
      "epoch": 0.9318793753365644,
      "grad_norm": 3.930699348449707,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 10383
    },
    {
      "epoch": 0.9319691258301921,
      "grad_norm": 2.692206621170044,
      "learning_rate": 0.0,
      "loss": 0.7253,
      "step": 10384
    },
    {
      "epoch": 0.9320588763238198,
      "grad_norm": 2.9129979610443115,
      "learning_rate": 0.0,
      "loss": 0.678,
      "step": 10385
    },
    {
      "epoch": 0.9321486268174475,
      "grad_norm": 3.086581230163574,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 10386
    },
    {
      "epoch": 0.9322383773110752,
      "grad_norm": 2.5898499488830566,
      "learning_rate": 0.0,
      "loss": 0.6929,
      "step": 10387
    },
    {
      "epoch": 0.932328127804703,
      "grad_norm": 2.9490737915039062,
      "learning_rate": 0.0,
      "loss": 0.6652,
      "step": 10388
    },
    {
      "epoch": 0.9324178782983307,
      "grad_norm": 2.579155921936035,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 10389
    },
    {
      "epoch": 0.9325076287919584,
      "grad_norm": 3.7346677780151367,
      "learning_rate": 0.0,
      "loss": 0.6919,
      "step": 10390
    },
    {
      "epoch": 0.9325973792855861,
      "grad_norm": 2.5968520641326904,
      "learning_rate": 0.0,
      "loss": 0.7374,
      "step": 10391
    },
    {
      "epoch": 0.9326871297792138,
      "grad_norm": 2.6098766326904297,
      "learning_rate": 0.0,
      "loss": 0.6796,
      "step": 10392
    },
    {
      "epoch": 0.9327768802728414,
      "grad_norm": 3.6371328830718994,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 10393
    },
    {
      "epoch": 0.9328666307664693,
      "grad_norm": 3.6584455966949463,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 10394
    },
    {
      "epoch": 0.932956381260097,
      "grad_norm": 3.4686877727508545,
      "learning_rate": 0.0,
      "loss": 0.7481,
      "step": 10395
    },
    {
      "epoch": 0.9330461317537246,
      "grad_norm": 3.104309320449829,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 10396
    },
    {
      "epoch": 0.9331358822473523,
      "grad_norm": 2.4829766750335693,
      "learning_rate": 0.0,
      "loss": 0.7083,
      "step": 10397
    },
    {
      "epoch": 0.93322563274098,
      "grad_norm": 25.364990234375,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 10398
    },
    {
      "epoch": 0.9333153832346078,
      "grad_norm": 3.5267131328582764,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 10399
    },
    {
      "epoch": 0.9334051337282355,
      "grad_norm": 3.1418004035949707,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 10400
    },
    {
      "epoch": 0.9334948842218632,
      "grad_norm": 2.7415225505828857,
      "learning_rate": 0.0,
      "loss": 0.8059,
      "step": 10401
    },
    {
      "epoch": 0.9335846347154909,
      "grad_norm": 3.074059247970581,
      "learning_rate": 0.0,
      "loss": 0.7052,
      "step": 10402
    },
    {
      "epoch": 0.9336743852091186,
      "grad_norm": 2.9878969192504883,
      "learning_rate": 0.0,
      "loss": 0.684,
      "step": 10403
    },
    {
      "epoch": 0.9337641357027464,
      "grad_norm": 2.8395047187805176,
      "learning_rate": 0.0,
      "loss": 0.755,
      "step": 10404
    },
    {
      "epoch": 0.9338538861963741,
      "grad_norm": 3.0695266723632812,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 10405
    },
    {
      "epoch": 0.9339436366900018,
      "grad_norm": 3.391136407852173,
      "learning_rate": 0.0,
      "loss": 0.694,
      "step": 10406
    },
    {
      "epoch": 0.9340333871836295,
      "grad_norm": 3.2551653385162354,
      "learning_rate": 0.0,
      "loss": 0.6479,
      "step": 10407
    },
    {
      "epoch": 0.9341231376772572,
      "grad_norm": 3.655172824859619,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 10408
    },
    {
      "epoch": 0.934212888170885,
      "grad_norm": 3.192992925643921,
      "learning_rate": 0.0,
      "loss": 0.6777,
      "step": 10409
    },
    {
      "epoch": 0.9343026386645127,
      "grad_norm": 3.745079517364502,
      "learning_rate": 0.0,
      "loss": 0.7115,
      "step": 10410
    },
    {
      "epoch": 0.9343923891581404,
      "grad_norm": 4.754045486450195,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 10411
    },
    {
      "epoch": 0.9344821396517681,
      "grad_norm": 2.682469367980957,
      "learning_rate": 0.0,
      "loss": 0.6681,
      "step": 10412
    },
    {
      "epoch": 0.9345718901453958,
      "grad_norm": 3.0720956325531006,
      "learning_rate": 0.0,
      "loss": 0.7257,
      "step": 10413
    },
    {
      "epoch": 0.9346616406390235,
      "grad_norm": 2.9214723110198975,
      "learning_rate": 0.0,
      "loss": 0.7315,
      "step": 10414
    },
    {
      "epoch": 0.9347513911326513,
      "grad_norm": 3.2577409744262695,
      "learning_rate": 0.0,
      "loss": 0.6539,
      "step": 10415
    },
    {
      "epoch": 0.934841141626279,
      "grad_norm": 3.0660455226898193,
      "learning_rate": 0.0,
      "loss": 0.6908,
      "step": 10416
    },
    {
      "epoch": 0.9349308921199067,
      "grad_norm": 6.566094875335693,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 10417
    },
    {
      "epoch": 0.9350206426135343,
      "grad_norm": 2.424774408340454,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 10418
    },
    {
      "epoch": 0.935110393107162,
      "grad_norm": 2.8076765537261963,
      "learning_rate": 0.0,
      "loss": 0.6796,
      "step": 10419
    },
    {
      "epoch": 0.9352001436007898,
      "grad_norm": 2.725998640060425,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 10420
    },
    {
      "epoch": 0.9352898940944175,
      "grad_norm": 3.4617109298706055,
      "learning_rate": 0.0,
      "loss": 0.6836,
      "step": 10421
    },
    {
      "epoch": 0.9353796445880452,
      "grad_norm": 3.076979160308838,
      "learning_rate": 0.0,
      "loss": 0.6646,
      "step": 10422
    },
    {
      "epoch": 0.9354693950816729,
      "grad_norm": 2.157397985458374,
      "learning_rate": 0.0,
      "loss": 0.7066,
      "step": 10423
    },
    {
      "epoch": 0.9355591455753006,
      "grad_norm": 3.1913139820098877,
      "learning_rate": 0.0,
      "loss": 0.6705,
      "step": 10424
    },
    {
      "epoch": 0.9356488960689284,
      "grad_norm": 3.743461847305298,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 10425
    },
    {
      "epoch": 0.9357386465625561,
      "grad_norm": 2.8739511966705322,
      "learning_rate": 0.0,
      "loss": 0.7783,
      "step": 10426
    },
    {
      "epoch": 0.9358283970561838,
      "grad_norm": 3.0311710834503174,
      "learning_rate": 0.0,
      "loss": 0.6678,
      "step": 10427
    },
    {
      "epoch": 0.9359181475498115,
      "grad_norm": 2.989638328552246,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 10428
    },
    {
      "epoch": 0.9360078980434392,
      "grad_norm": 2.5877363681793213,
      "learning_rate": 0.0,
      "loss": 0.7407,
      "step": 10429
    },
    {
      "epoch": 0.936097648537067,
      "grad_norm": 3.5639665126800537,
      "learning_rate": 0.0,
      "loss": 0.7561,
      "step": 10430
    },
    {
      "epoch": 0.9361873990306947,
      "grad_norm": 2.886183738708496,
      "learning_rate": 0.0,
      "loss": 0.6732,
      "step": 10431
    },
    {
      "epoch": 0.9362771495243224,
      "grad_norm": 3.010542392730713,
      "learning_rate": 0.0,
      "loss": 0.6329,
      "step": 10432
    },
    {
      "epoch": 0.9363669000179501,
      "grad_norm": 5.241832256317139,
      "learning_rate": 0.0,
      "loss": 0.6856,
      "step": 10433
    },
    {
      "epoch": 0.9364566505115778,
      "grad_norm": 2.947519540786743,
      "learning_rate": 0.0,
      "loss": 0.7524,
      "step": 10434
    },
    {
      "epoch": 0.9365464010052055,
      "grad_norm": 3.0889179706573486,
      "learning_rate": 0.0,
      "loss": 0.6977,
      "step": 10435
    },
    {
      "epoch": 0.9366361514988333,
      "grad_norm": 2.733826160430908,
      "learning_rate": 0.0,
      "loss": 0.6285,
      "step": 10436
    },
    {
      "epoch": 0.936725901992461,
      "grad_norm": 2.344353675842285,
      "learning_rate": 0.0,
      "loss": 0.6653,
      "step": 10437
    },
    {
      "epoch": 0.9368156524860887,
      "grad_norm": 2.9812920093536377,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 10438
    },
    {
      "epoch": 0.9369054029797164,
      "grad_norm": 3.0529427528381348,
      "learning_rate": 0.0,
      "loss": 0.6712,
      "step": 10439
    },
    {
      "epoch": 0.9369951534733441,
      "grad_norm": 3.3323004245758057,
      "learning_rate": 0.0,
      "loss": 0.6917,
      "step": 10440
    },
    {
      "epoch": 0.9370849039669719,
      "grad_norm": 2.5576164722442627,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 10441
    },
    {
      "epoch": 0.9371746544605996,
      "grad_norm": 3.160634756088257,
      "learning_rate": 0.0,
      "loss": 0.7003,
      "step": 10442
    },
    {
      "epoch": 0.9372644049542272,
      "grad_norm": 2.241699695587158,
      "learning_rate": 0.0,
      "loss": 0.7345,
      "step": 10443
    },
    {
      "epoch": 0.9373541554478549,
      "grad_norm": 2.9353702068328857,
      "learning_rate": 0.0,
      "loss": 0.6817,
      "step": 10444
    },
    {
      "epoch": 0.9374439059414826,
      "grad_norm": 2.9774348735809326,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 10445
    },
    {
      "epoch": 0.9375336564351104,
      "grad_norm": 3.0665035247802734,
      "learning_rate": 0.0,
      "loss": 0.716,
      "step": 10446
    },
    {
      "epoch": 0.9376234069287381,
      "grad_norm": 3.4917027950286865,
      "learning_rate": 0.0,
      "loss": 0.7298,
      "step": 10447
    },
    {
      "epoch": 0.9377131574223658,
      "grad_norm": 2.545928716659546,
      "learning_rate": 0.0,
      "loss": 0.7289,
      "step": 10448
    },
    {
      "epoch": 0.9378029079159935,
      "grad_norm": 4.05872917175293,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 10449
    },
    {
      "epoch": 0.9378926584096212,
      "grad_norm": 2.9398229122161865,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 10450
    },
    {
      "epoch": 0.937982408903249,
      "grad_norm": 2.0076756477355957,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10451
    },
    {
      "epoch": 0.9380721593968767,
      "grad_norm": 2.41803240776062,
      "learning_rate": 0.0,
      "loss": 0.7528,
      "step": 10452
    },
    {
      "epoch": 0.9381619098905044,
      "grad_norm": 3.122899293899536,
      "learning_rate": 0.0,
      "loss": 0.7531,
      "step": 10453
    },
    {
      "epoch": 0.9382516603841321,
      "grad_norm": 2.7259957790374756,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 10454
    },
    {
      "epoch": 0.9383414108777598,
      "grad_norm": 2.9178926944732666,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 10455
    },
    {
      "epoch": 0.9384311613713875,
      "grad_norm": 3.087250232696533,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 10456
    },
    {
      "epoch": 0.9385209118650153,
      "grad_norm": 2.701354742050171,
      "learning_rate": 0.0,
      "loss": 0.7045,
      "step": 10457
    },
    {
      "epoch": 0.938610662358643,
      "grad_norm": 2.8794238567352295,
      "learning_rate": 0.0,
      "loss": 0.6691,
      "step": 10458
    },
    {
      "epoch": 0.9387004128522707,
      "grad_norm": 3.1836013793945312,
      "learning_rate": 0.0,
      "loss": 0.7084,
      "step": 10459
    },
    {
      "epoch": 0.9387901633458984,
      "grad_norm": 2.4708504676818848,
      "learning_rate": 0.0,
      "loss": 0.7905,
      "step": 10460
    },
    {
      "epoch": 0.9388799138395261,
      "grad_norm": 3.273526191711426,
      "learning_rate": 0.0,
      "loss": 0.6561,
      "step": 10461
    },
    {
      "epoch": 0.9389696643331539,
      "grad_norm": 3.9238200187683105,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 10462
    },
    {
      "epoch": 0.9390594148267816,
      "grad_norm": 2.4079973697662354,
      "learning_rate": 0.0,
      "loss": 0.672,
      "step": 10463
    },
    {
      "epoch": 0.9391491653204093,
      "grad_norm": 2.630265474319458,
      "learning_rate": 0.0,
      "loss": 0.6941,
      "step": 10464
    },
    {
      "epoch": 0.939238915814037,
      "grad_norm": 2.6676278114318848,
      "learning_rate": 0.0,
      "loss": 0.7173,
      "step": 10465
    },
    {
      "epoch": 0.9393286663076647,
      "grad_norm": 2.68282413482666,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 10466
    },
    {
      "epoch": 0.9394184168012925,
      "grad_norm": 2.6952006816864014,
      "learning_rate": 0.0,
      "loss": 0.7541,
      "step": 10467
    },
    {
      "epoch": 0.9395081672949201,
      "grad_norm": 3.279388666152954,
      "learning_rate": 0.0,
      "loss": 0.7143,
      "step": 10468
    },
    {
      "epoch": 0.9395979177885478,
      "grad_norm": 2.6184298992156982,
      "learning_rate": 0.0,
      "loss": 0.6866,
      "step": 10469
    },
    {
      "epoch": 0.9396876682821755,
      "grad_norm": 3.183688163757324,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 10470
    },
    {
      "epoch": 0.9397774187758032,
      "grad_norm": 3.2839248180389404,
      "learning_rate": 0.0,
      "loss": 0.686,
      "step": 10471
    },
    {
      "epoch": 0.939867169269431,
      "grad_norm": 2.6162426471710205,
      "learning_rate": 0.0,
      "loss": 0.6697,
      "step": 10472
    },
    {
      "epoch": 0.9399569197630587,
      "grad_norm": 3.9480292797088623,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 10473
    },
    {
      "epoch": 0.9400466702566864,
      "grad_norm": 4.275997161865234,
      "learning_rate": 0.0,
      "loss": 0.6856,
      "step": 10474
    },
    {
      "epoch": 0.9401364207503141,
      "grad_norm": 2.9287526607513428,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 10475
    },
    {
      "epoch": 0.9402261712439418,
      "grad_norm": 3.2424447536468506,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 10476
    },
    {
      "epoch": 0.9403159217375695,
      "grad_norm": 3.71010422706604,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 10477
    },
    {
      "epoch": 0.9404056722311973,
      "grad_norm": 3.418473482131958,
      "learning_rate": 0.0,
      "loss": 0.672,
      "step": 10478
    },
    {
      "epoch": 0.940495422724825,
      "grad_norm": 4.104066848754883,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 10479
    },
    {
      "epoch": 0.9405851732184527,
      "grad_norm": 2.641225814819336,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 10480
    },
    {
      "epoch": 0.9406749237120804,
      "grad_norm": 2.910543918609619,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 10481
    },
    {
      "epoch": 0.9407646742057081,
      "grad_norm": 3.0413076877593994,
      "learning_rate": 0.0,
      "loss": 0.681,
      "step": 10482
    },
    {
      "epoch": 0.9408544246993359,
      "grad_norm": 3.5337953567504883,
      "learning_rate": 0.0,
      "loss": 0.6684,
      "step": 10483
    },
    {
      "epoch": 0.9409441751929636,
      "grad_norm": 5.768782615661621,
      "learning_rate": 0.0,
      "loss": 0.6893,
      "step": 10484
    },
    {
      "epoch": 0.9410339256865913,
      "grad_norm": 3.238264560699463,
      "learning_rate": 0.0,
      "loss": 0.6801,
      "step": 10485
    },
    {
      "epoch": 0.941123676180219,
      "grad_norm": 2.402186870574951,
      "learning_rate": 0.0,
      "loss": 0.6723,
      "step": 10486
    },
    {
      "epoch": 0.9412134266738467,
      "grad_norm": 2.6559011936187744,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 10487
    },
    {
      "epoch": 0.9413031771674745,
      "grad_norm": 3.369415283203125,
      "learning_rate": 0.0,
      "loss": 0.6561,
      "step": 10488
    },
    {
      "epoch": 0.9413929276611022,
      "grad_norm": 3.5579895973205566,
      "learning_rate": 0.0,
      "loss": 0.6625,
      "step": 10489
    },
    {
      "epoch": 0.9414826781547299,
      "grad_norm": 2.900162935256958,
      "learning_rate": 0.0,
      "loss": 0.7016,
      "step": 10490
    },
    {
      "epoch": 0.9415724286483576,
      "grad_norm": 3.0555050373077393,
      "learning_rate": 0.0,
      "loss": 0.6756,
      "step": 10491
    },
    {
      "epoch": 0.9416621791419852,
      "grad_norm": 2.9592323303222656,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 10492
    },
    {
      "epoch": 0.941751929635613,
      "grad_norm": 2.8601512908935547,
      "learning_rate": 0.0,
      "loss": 0.7322,
      "step": 10493
    },
    {
      "epoch": 0.9418416801292407,
      "grad_norm": 4.688403129577637,
      "learning_rate": 0.0,
      "loss": 0.7589,
      "step": 10494
    },
    {
      "epoch": 0.9419314306228684,
      "grad_norm": 4.049013614654541,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 10495
    },
    {
      "epoch": 0.9420211811164961,
      "grad_norm": 2.9775474071502686,
      "learning_rate": 0.0,
      "loss": 0.7448,
      "step": 10496
    },
    {
      "epoch": 0.9421109316101238,
      "grad_norm": 3.7378287315368652,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 10497
    },
    {
      "epoch": 0.9422006821037515,
      "grad_norm": 4.0933966636657715,
      "learning_rate": 0.0,
      "loss": 0.662,
      "step": 10498
    },
    {
      "epoch": 0.9422904325973793,
      "grad_norm": 3.587615966796875,
      "learning_rate": 0.0,
      "loss": 0.75,
      "step": 10499
    },
    {
      "epoch": 0.942380183091007,
      "grad_norm": 3.1732046604156494,
      "learning_rate": 0.0,
      "loss": 0.6932,
      "step": 10500
    },
    {
      "epoch": 0.9424699335846347,
      "grad_norm": 15.090354919433594,
      "learning_rate": 0.0,
      "loss": 0.7377,
      "step": 10501
    },
    {
      "epoch": 0.9425596840782624,
      "grad_norm": 3.497990131378174,
      "learning_rate": 0.0,
      "loss": 0.6951,
      "step": 10502
    },
    {
      "epoch": 0.9426494345718901,
      "grad_norm": 3.797631025314331,
      "learning_rate": 0.0,
      "loss": 0.7369,
      "step": 10503
    },
    {
      "epoch": 0.9427391850655179,
      "grad_norm": 4.093072891235352,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 10504
    },
    {
      "epoch": 0.9428289355591456,
      "grad_norm": 2.344777822494507,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 10505
    },
    {
      "epoch": 0.9429186860527733,
      "grad_norm": 3.2933871746063232,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 10506
    },
    {
      "epoch": 0.943008436546401,
      "grad_norm": 4.315636157989502,
      "learning_rate": 0.0,
      "loss": 0.7473,
      "step": 10507
    },
    {
      "epoch": 0.9430981870400287,
      "grad_norm": 4.008358478546143,
      "learning_rate": 0.0,
      "loss": 0.6366,
      "step": 10508
    },
    {
      "epoch": 0.9431879375336565,
      "grad_norm": 2.855945110321045,
      "learning_rate": 0.0,
      "loss": 0.7049,
      "step": 10509
    },
    {
      "epoch": 0.9432776880272842,
      "grad_norm": 3.308610677719116,
      "learning_rate": 0.0,
      "loss": 0.6568,
      "step": 10510
    },
    {
      "epoch": 0.9433674385209119,
      "grad_norm": 3.9516546726226807,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 10511
    },
    {
      "epoch": 0.9434571890145396,
      "grad_norm": 4.758216857910156,
      "learning_rate": 0.0,
      "loss": 0.6704,
      "step": 10512
    },
    {
      "epoch": 0.9435469395081673,
      "grad_norm": 4.729433059692383,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 10513
    },
    {
      "epoch": 0.9436366900017951,
      "grad_norm": 4.353995323181152,
      "learning_rate": 0.0,
      "loss": 0.7299,
      "step": 10514
    },
    {
      "epoch": 0.9437264404954228,
      "grad_norm": 3.4118924140930176,
      "learning_rate": 0.0,
      "loss": 0.7467,
      "step": 10515
    },
    {
      "epoch": 0.9438161909890505,
      "grad_norm": 2.9900879859924316,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 10516
    },
    {
      "epoch": 0.9439059414826781,
      "grad_norm": 3.459533929824829,
      "learning_rate": 0.0,
      "loss": 0.6617,
      "step": 10517
    },
    {
      "epoch": 0.9439956919763058,
      "grad_norm": 3.701503038406372,
      "learning_rate": 0.0,
      "loss": 0.6507,
      "step": 10518
    },
    {
      "epoch": 0.9440854424699335,
      "grad_norm": 3.136815071105957,
      "learning_rate": 0.0,
      "loss": 0.7013,
      "step": 10519
    },
    {
      "epoch": 0.9441751929635613,
      "grad_norm": 2.6228995323181152,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 10520
    },
    {
      "epoch": 0.944264943457189,
      "grad_norm": 3.1883761882781982,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 10521
    },
    {
      "epoch": 0.9443546939508167,
      "grad_norm": 4.9245195388793945,
      "learning_rate": 0.0,
      "loss": 0.6159,
      "step": 10522
    },
    {
      "epoch": 0.9444444444444444,
      "grad_norm": 4.263134002685547,
      "learning_rate": 0.0,
      "loss": 0.7416,
      "step": 10523
    },
    {
      "epoch": 0.9445341949380721,
      "grad_norm": 3.647387981414795,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 10524
    },
    {
      "epoch": 0.9446239454316999,
      "grad_norm": 3.6113319396972656,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 10525
    },
    {
      "epoch": 0.9447136959253276,
      "grad_norm": 2.8577980995178223,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 10526
    },
    {
      "epoch": 0.9448034464189553,
      "grad_norm": 5.042186737060547,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10527
    },
    {
      "epoch": 0.944893196912583,
      "grad_norm": 3.730407953262329,
      "learning_rate": 0.0,
      "loss": 0.7333,
      "step": 10528
    },
    {
      "epoch": 0.9449829474062107,
      "grad_norm": 9.02573299407959,
      "learning_rate": 0.0,
      "loss": 0.6941,
      "step": 10529
    },
    {
      "epoch": 0.9450726978998385,
      "grad_norm": 3.2448055744171143,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 10530
    },
    {
      "epoch": 0.9451624483934662,
      "grad_norm": 5.229548931121826,
      "learning_rate": 0.0,
      "loss": 0.735,
      "step": 10531
    },
    {
      "epoch": 0.9452521988870939,
      "grad_norm": 3.012502670288086,
      "learning_rate": 0.0,
      "loss": 0.7232,
      "step": 10532
    },
    {
      "epoch": 0.9453419493807216,
      "grad_norm": 2.5205631256103516,
      "learning_rate": 0.0,
      "loss": 0.697,
      "step": 10533
    },
    {
      "epoch": 0.9454316998743493,
      "grad_norm": 3.2377116680145264,
      "learning_rate": 0.0,
      "loss": 0.7239,
      "step": 10534
    },
    {
      "epoch": 0.9455214503679771,
      "grad_norm": 3.282959461212158,
      "learning_rate": 0.0,
      "loss": 0.7227,
      "step": 10535
    },
    {
      "epoch": 0.9456112008616048,
      "grad_norm": 3.7683005332946777,
      "learning_rate": 0.0,
      "loss": 0.7549,
      "step": 10536
    },
    {
      "epoch": 0.9457009513552325,
      "grad_norm": 3.3457114696502686,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 10537
    },
    {
      "epoch": 0.9457907018488602,
      "grad_norm": 2.8418726921081543,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 10538
    },
    {
      "epoch": 0.9458804523424879,
      "grad_norm": 6.296527862548828,
      "learning_rate": 0.0,
      "loss": 0.7011,
      "step": 10539
    },
    {
      "epoch": 0.9459702028361155,
      "grad_norm": 4.837912082672119,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 10540
    },
    {
      "epoch": 0.9460599533297434,
      "grad_norm": 2.7862637042999268,
      "learning_rate": 0.0,
      "loss": 0.6788,
      "step": 10541
    },
    {
      "epoch": 0.946149703823371,
      "grad_norm": 2.814861536026001,
      "learning_rate": 0.0,
      "loss": 0.7144,
      "step": 10542
    },
    {
      "epoch": 0.9462394543169987,
      "grad_norm": 2.367278814315796,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 10543
    },
    {
      "epoch": 0.9463292048106264,
      "grad_norm": 3.144005537033081,
      "learning_rate": 0.0,
      "loss": 0.6885,
      "step": 10544
    },
    {
      "epoch": 0.9464189553042541,
      "grad_norm": 4.000855445861816,
      "learning_rate": 0.0,
      "loss": 0.745,
      "step": 10545
    },
    {
      "epoch": 0.9465087057978819,
      "grad_norm": 3.798564910888672,
      "learning_rate": 0.0,
      "loss": 0.8039,
      "step": 10546
    },
    {
      "epoch": 0.9465984562915096,
      "grad_norm": 2.2744200229644775,
      "learning_rate": 0.0,
      "loss": 0.6689,
      "step": 10547
    },
    {
      "epoch": 0.9466882067851373,
      "grad_norm": 2.9408161640167236,
      "learning_rate": 0.0,
      "loss": 0.7342,
      "step": 10548
    },
    {
      "epoch": 0.946777957278765,
      "grad_norm": 3.0989575386047363,
      "learning_rate": 0.0,
      "loss": 0.732,
      "step": 10549
    },
    {
      "epoch": 0.9468677077723927,
      "grad_norm": 2.6225903034210205,
      "learning_rate": 0.0,
      "loss": 0.6925,
      "step": 10550
    },
    {
      "epoch": 0.9469574582660205,
      "grad_norm": 2.776787757873535,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 10551
    },
    {
      "epoch": 0.9470472087596482,
      "grad_norm": 2.7233381271362305,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 10552
    },
    {
      "epoch": 0.9471369592532759,
      "grad_norm": 2.4365196228027344,
      "learning_rate": 0.0,
      "loss": 0.6938,
      "step": 10553
    },
    {
      "epoch": 0.9472267097469036,
      "grad_norm": 9.642890930175781,
      "learning_rate": 0.0,
      "loss": 0.7831,
      "step": 10554
    },
    {
      "epoch": 0.9473164602405313,
      "grad_norm": 4.07692289352417,
      "learning_rate": 0.0,
      "loss": 0.6862,
      "step": 10555
    },
    {
      "epoch": 0.9474062107341591,
      "grad_norm": 3.7154219150543213,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 10556
    },
    {
      "epoch": 0.9474959612277868,
      "grad_norm": 3.2162387371063232,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 10557
    },
    {
      "epoch": 0.9475857117214145,
      "grad_norm": 2.9441943168640137,
      "learning_rate": 0.0,
      "loss": 0.7307,
      "step": 10558
    },
    {
      "epoch": 0.9476754622150422,
      "grad_norm": 4.831309795379639,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 10559
    },
    {
      "epoch": 0.9477652127086699,
      "grad_norm": 6.211119174957275,
      "learning_rate": 0.0,
      "loss": 0.6627,
      "step": 10560
    },
    {
      "epoch": 0.9478549632022976,
      "grad_norm": 3.269068717956543,
      "learning_rate": 0.0,
      "loss": 0.6856,
      "step": 10561
    },
    {
      "epoch": 0.9479447136959254,
      "grad_norm": 3.8796801567077637,
      "learning_rate": 0.0,
      "loss": 0.6253,
      "step": 10562
    },
    {
      "epoch": 0.9480344641895531,
      "grad_norm": 3.701958417892456,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 10563
    },
    {
      "epoch": 0.9481242146831808,
      "grad_norm": 14.84333610534668,
      "learning_rate": 0.0,
      "loss": 0.6852,
      "step": 10564
    },
    {
      "epoch": 0.9482139651768084,
      "grad_norm": 3.960047483444214,
      "learning_rate": 0.0,
      "loss": 0.6786,
      "step": 10565
    },
    {
      "epoch": 0.9483037156704361,
      "grad_norm": 3.092717409133911,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 10566
    },
    {
      "epoch": 0.948393466164064,
      "grad_norm": 2.3090665340423584,
      "learning_rate": 0.0,
      "loss": 0.6893,
      "step": 10567
    },
    {
      "epoch": 0.9484832166576916,
      "grad_norm": 2.744904041290283,
      "learning_rate": 0.0,
      "loss": 0.6623,
      "step": 10568
    },
    {
      "epoch": 0.9485729671513193,
      "grad_norm": 3.113722562789917,
      "learning_rate": 0.0,
      "loss": 0.7272,
      "step": 10569
    },
    {
      "epoch": 0.948662717644947,
      "grad_norm": 5.251560688018799,
      "learning_rate": 0.0,
      "loss": 0.6416,
      "step": 10570
    },
    {
      "epoch": 0.9487524681385747,
      "grad_norm": 2.606720209121704,
      "learning_rate": 0.0,
      "loss": 0.8132,
      "step": 10571
    },
    {
      "epoch": 0.9488422186322025,
      "grad_norm": 4.716255187988281,
      "learning_rate": 0.0,
      "loss": 0.7179,
      "step": 10572
    },
    {
      "epoch": 0.9489319691258302,
      "grad_norm": 3.0201895236968994,
      "learning_rate": 0.0,
      "loss": 0.7266,
      "step": 10573
    },
    {
      "epoch": 0.9490217196194579,
      "grad_norm": 2.3563034534454346,
      "learning_rate": 0.0,
      "loss": 0.7048,
      "step": 10574
    },
    {
      "epoch": 0.9491114701130856,
      "grad_norm": 2.432203531265259,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 10575
    },
    {
      "epoch": 0.9492012206067133,
      "grad_norm": 2.7159650325775146,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 10576
    },
    {
      "epoch": 0.949290971100341,
      "grad_norm": 3.938610076904297,
      "learning_rate": 0.0,
      "loss": 0.708,
      "step": 10577
    },
    {
      "epoch": 0.9493807215939688,
      "grad_norm": 3.4758474826812744,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 10578
    },
    {
      "epoch": 0.9494704720875965,
      "grad_norm": 2.8637187480926514,
      "learning_rate": 0.0,
      "loss": 0.7181,
      "step": 10579
    },
    {
      "epoch": 0.9495602225812242,
      "grad_norm": 2.715867519378662,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 10580
    },
    {
      "epoch": 0.9496499730748519,
      "grad_norm": 3.399167060852051,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 10581
    },
    {
      "epoch": 0.9497397235684796,
      "grad_norm": 2.815141439437866,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 10582
    },
    {
      "epoch": 0.9498294740621074,
      "grad_norm": 2.5759193897247314,
      "learning_rate": 0.0,
      "loss": 0.7123,
      "step": 10583
    },
    {
      "epoch": 0.9499192245557351,
      "grad_norm": 3.4609148502349854,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 10584
    },
    {
      "epoch": 0.9500089750493628,
      "grad_norm": 2.5888750553131104,
      "learning_rate": 0.0,
      "loss": 0.7213,
      "step": 10585
    },
    {
      "epoch": 0.9500987255429905,
      "grad_norm": 2.7049787044525146,
      "learning_rate": 0.0,
      "loss": 0.6575,
      "step": 10586
    },
    {
      "epoch": 0.9501884760366182,
      "grad_norm": 2.96771502494812,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 10587
    },
    {
      "epoch": 0.950278226530246,
      "grad_norm": 6.087590217590332,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 10588
    },
    {
      "epoch": 0.9503679770238737,
      "grad_norm": 3.2515170574188232,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 10589
    },
    {
      "epoch": 0.9504577275175013,
      "grad_norm": 3.191133499145508,
      "learning_rate": 0.0,
      "loss": 0.667,
      "step": 10590
    },
    {
      "epoch": 0.950547478011129,
      "grad_norm": 3.629026412963867,
      "learning_rate": 0.0,
      "loss": 0.6485,
      "step": 10591
    },
    {
      "epoch": 0.9506372285047567,
      "grad_norm": 3.111525535583496,
      "learning_rate": 0.0,
      "loss": 0.6696,
      "step": 10592
    },
    {
      "epoch": 0.9507269789983845,
      "grad_norm": 3.2040858268737793,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 10593
    },
    {
      "epoch": 0.9508167294920122,
      "grad_norm": 2.813403367996216,
      "learning_rate": 0.0,
      "loss": 0.6975,
      "step": 10594
    },
    {
      "epoch": 0.9509064799856399,
      "grad_norm": 2.5960280895233154,
      "learning_rate": 0.0,
      "loss": 0.6673,
      "step": 10595
    },
    {
      "epoch": 0.9509962304792676,
      "grad_norm": 2.6726577281951904,
      "learning_rate": 0.0,
      "loss": 0.7798,
      "step": 10596
    },
    {
      "epoch": 0.9510859809728953,
      "grad_norm": 4.027924060821533,
      "learning_rate": 0.0,
      "loss": 0.6778,
      "step": 10597
    },
    {
      "epoch": 0.951175731466523,
      "grad_norm": 2.7363924980163574,
      "learning_rate": 0.0,
      "loss": 0.7189,
      "step": 10598
    },
    {
      "epoch": 0.9512654819601508,
      "grad_norm": 2.514404296875,
      "learning_rate": 0.0,
      "loss": 0.6815,
      "step": 10599
    },
    {
      "epoch": 0.9513552324537785,
      "grad_norm": 2.925013542175293,
      "learning_rate": 0.0,
      "loss": 0.6439,
      "step": 10600
    },
    {
      "epoch": 0.9514449829474062,
      "grad_norm": 3.660682439804077,
      "learning_rate": 0.0,
      "loss": 0.6948,
      "step": 10601
    },
    {
      "epoch": 0.9515347334410339,
      "grad_norm": 2.839688301086426,
      "learning_rate": 0.0,
      "loss": 0.7032,
      "step": 10602
    },
    {
      "epoch": 0.9516244839346616,
      "grad_norm": 2.701861619949341,
      "learning_rate": 0.0,
      "loss": 0.6724,
      "step": 10603
    },
    {
      "epoch": 0.9517142344282894,
      "grad_norm": 2.750824451446533,
      "learning_rate": 0.0,
      "loss": 0.7035,
      "step": 10604
    },
    {
      "epoch": 0.9518039849219171,
      "grad_norm": 2.8698697090148926,
      "learning_rate": 0.0,
      "loss": 0.7368,
      "step": 10605
    },
    {
      "epoch": 0.9518937354155448,
      "grad_norm": 2.803157091140747,
      "learning_rate": 0.0,
      "loss": 0.6747,
      "step": 10606
    },
    {
      "epoch": 0.9519834859091725,
      "grad_norm": 3.2817304134368896,
      "learning_rate": 0.0,
      "loss": 0.7234,
      "step": 10607
    },
    {
      "epoch": 0.9520732364028002,
      "grad_norm": 3.7792818546295166,
      "learning_rate": 0.0,
      "loss": 0.7165,
      "step": 10608
    },
    {
      "epoch": 0.952162986896428,
      "grad_norm": 3.1559321880340576,
      "learning_rate": 0.0,
      "loss": 0.7088,
      "step": 10609
    },
    {
      "epoch": 0.9522527373900557,
      "grad_norm": 2.124398946762085,
      "learning_rate": 0.0,
      "loss": 0.762,
      "step": 10610
    },
    {
      "epoch": 0.9523424878836834,
      "grad_norm": 3.164332151412964,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 10611
    },
    {
      "epoch": 0.9524322383773111,
      "grad_norm": 4.528585910797119,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 10612
    },
    {
      "epoch": 0.9525219888709388,
      "grad_norm": 3.449789524078369,
      "learning_rate": 0.0,
      "loss": 0.7536,
      "step": 10613
    },
    {
      "epoch": 0.9526117393645666,
      "grad_norm": 3.543231964111328,
      "learning_rate": 0.0,
      "loss": 0.6663,
      "step": 10614
    },
    {
      "epoch": 0.9527014898581943,
      "grad_norm": 3.006653070449829,
      "learning_rate": 0.0,
      "loss": 0.7335,
      "step": 10615
    },
    {
      "epoch": 0.9527912403518219,
      "grad_norm": 2.6241977214813232,
      "learning_rate": 0.0,
      "loss": 0.6629,
      "step": 10616
    },
    {
      "epoch": 0.9528809908454496,
      "grad_norm": 3.414494514465332,
      "learning_rate": 0.0,
      "loss": 0.7332,
      "step": 10617
    },
    {
      "epoch": 0.9529707413390773,
      "grad_norm": 3.596864700317383,
      "learning_rate": 0.0,
      "loss": 0.798,
      "step": 10618
    },
    {
      "epoch": 0.953060491832705,
      "grad_norm": 5.119151592254639,
      "learning_rate": 0.0,
      "loss": 0.7185,
      "step": 10619
    },
    {
      "epoch": 0.9531502423263328,
      "grad_norm": 2.9203200340270996,
      "learning_rate": 0.0,
      "loss": 0.7122,
      "step": 10620
    },
    {
      "epoch": 0.9532399928199605,
      "grad_norm": 3.2592358589172363,
      "learning_rate": 0.0,
      "loss": 0.7377,
      "step": 10621
    },
    {
      "epoch": 0.9533297433135882,
      "grad_norm": 4.008144378662109,
      "learning_rate": 0.0,
      "loss": 0.7309,
      "step": 10622
    },
    {
      "epoch": 0.9534194938072159,
      "grad_norm": 2.8132741451263428,
      "learning_rate": 0.0,
      "loss": 0.7204,
      "step": 10623
    },
    {
      "epoch": 0.9535092443008436,
      "grad_norm": 2.6416728496551514,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 10624
    },
    {
      "epoch": 0.9535989947944714,
      "grad_norm": 2.7990167140960693,
      "learning_rate": 0.0,
      "loss": 0.6735,
      "step": 10625
    },
    {
      "epoch": 0.9536887452880991,
      "grad_norm": 4.029527187347412,
      "learning_rate": 0.0,
      "loss": 0.7221,
      "step": 10626
    },
    {
      "epoch": 0.9537784957817268,
      "grad_norm": 3.334998607635498,
      "learning_rate": 0.0,
      "loss": 0.7069,
      "step": 10627
    },
    {
      "epoch": 0.9538682462753545,
      "grad_norm": 2.7113635540008545,
      "learning_rate": 0.0,
      "loss": 0.7146,
      "step": 10628
    },
    {
      "epoch": 0.9539579967689822,
      "grad_norm": 3.7241995334625244,
      "learning_rate": 0.0,
      "loss": 0.7147,
      "step": 10629
    },
    {
      "epoch": 0.95404774726261,
      "grad_norm": 3.566563129425049,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 10630
    },
    {
      "epoch": 0.9541374977562377,
      "grad_norm": 2.5666120052337646,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 10631
    },
    {
      "epoch": 0.9542272482498654,
      "grad_norm": 3.6434900760650635,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 10632
    },
    {
      "epoch": 0.9543169987434931,
      "grad_norm": 3.2488808631896973,
      "learning_rate": 0.0,
      "loss": 0.6829,
      "step": 10633
    },
    {
      "epoch": 0.9544067492371208,
      "grad_norm": 2.658539295196533,
      "learning_rate": 0.0,
      "loss": 0.7166,
      "step": 10634
    },
    {
      "epoch": 0.9544964997307486,
      "grad_norm": 2.1640515327453613,
      "learning_rate": 0.0,
      "loss": 0.7177,
      "step": 10635
    },
    {
      "epoch": 0.9545862502243763,
      "grad_norm": 2.740485429763794,
      "learning_rate": 0.0,
      "loss": 0.6856,
      "step": 10636
    },
    {
      "epoch": 0.954676000718004,
      "grad_norm": 2.71994948387146,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 10637
    },
    {
      "epoch": 0.9547657512116317,
      "grad_norm": 2.8743247985839844,
      "learning_rate": 0.0,
      "loss": 0.6955,
      "step": 10638
    },
    {
      "epoch": 0.9548555017052593,
      "grad_norm": 3.605332612991333,
      "learning_rate": 0.0,
      "loss": 0.7421,
      "step": 10639
    },
    {
      "epoch": 0.954945252198887,
      "grad_norm": 2.9977426528930664,
      "learning_rate": 0.0,
      "loss": 0.757,
      "step": 10640
    },
    {
      "epoch": 0.9550350026925148,
      "grad_norm": 2.4423117637634277,
      "learning_rate": 0.0,
      "loss": 0.6736,
      "step": 10641
    },
    {
      "epoch": 0.9551247531861425,
      "grad_norm": 2.5756776332855225,
      "learning_rate": 0.0,
      "loss": 0.7085,
      "step": 10642
    },
    {
      "epoch": 0.9552145036797702,
      "grad_norm": 2.9220924377441406,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 10643
    },
    {
      "epoch": 0.9553042541733979,
      "grad_norm": 4.551344871520996,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 10644
    },
    {
      "epoch": 0.9553940046670256,
      "grad_norm": 3.1808135509490967,
      "learning_rate": 0.0,
      "loss": 0.7429,
      "step": 10645
    },
    {
      "epoch": 0.9554837551606534,
      "grad_norm": 3.0208303928375244,
      "learning_rate": 0.0,
      "loss": 0.7178,
      "step": 10646
    },
    {
      "epoch": 0.9555735056542811,
      "grad_norm": 2.5976622104644775,
      "learning_rate": 0.0,
      "loss": 0.7385,
      "step": 10647
    },
    {
      "epoch": 0.9556632561479088,
      "grad_norm": 6.066844940185547,
      "learning_rate": 0.0,
      "loss": 0.6741,
      "step": 10648
    },
    {
      "epoch": 0.9557530066415365,
      "grad_norm": 3.439180612564087,
      "learning_rate": 0.0,
      "loss": 0.7098,
      "step": 10649
    },
    {
      "epoch": 0.9558427571351642,
      "grad_norm": 3.021979808807373,
      "learning_rate": 0.0,
      "loss": 0.693,
      "step": 10650
    },
    {
      "epoch": 0.955932507628792,
      "grad_norm": 2.6342105865478516,
      "learning_rate": 0.0,
      "loss": 0.6632,
      "step": 10651
    },
    {
      "epoch": 0.9560222581224197,
      "grad_norm": 2.960533618927002,
      "learning_rate": 0.0,
      "loss": 0.6936,
      "step": 10652
    },
    {
      "epoch": 0.9561120086160474,
      "grad_norm": 2.7368876934051514,
      "learning_rate": 0.0,
      "loss": 0.6725,
      "step": 10653
    },
    {
      "epoch": 0.9562017591096751,
      "grad_norm": 3.1911940574645996,
      "learning_rate": 0.0,
      "loss": 0.7352,
      "step": 10654
    },
    {
      "epoch": 0.9562915096033028,
      "grad_norm": 5.530120849609375,
      "learning_rate": 0.0,
      "loss": 0.676,
      "step": 10655
    },
    {
      "epoch": 0.9563812600969306,
      "grad_norm": 2.866729974746704,
      "learning_rate": 0.0,
      "loss": 0.688,
      "step": 10656
    },
    {
      "epoch": 0.9564710105905583,
      "grad_norm": 3.515314817428589,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 10657
    },
    {
      "epoch": 0.956560761084186,
      "grad_norm": 2.8646106719970703,
      "learning_rate": 0.0,
      "loss": 0.7264,
      "step": 10658
    },
    {
      "epoch": 0.9566505115778137,
      "grad_norm": 3.17622709274292,
      "learning_rate": 0.0,
      "loss": 0.7819,
      "step": 10659
    },
    {
      "epoch": 0.9567402620714414,
      "grad_norm": 9.159092903137207,
      "learning_rate": 0.0,
      "loss": 0.6904,
      "step": 10660
    },
    {
      "epoch": 0.956830012565069,
      "grad_norm": 3.93237042427063,
      "learning_rate": 0.0,
      "loss": 0.7172,
      "step": 10661
    },
    {
      "epoch": 0.9569197630586969,
      "grad_norm": 3.695436477661133,
      "learning_rate": 0.0,
      "loss": 0.7002,
      "step": 10662
    },
    {
      "epoch": 0.9570095135523246,
      "grad_norm": 2.9385457038879395,
      "learning_rate": 0.0,
      "loss": 0.686,
      "step": 10663
    },
    {
      "epoch": 0.9570992640459522,
      "grad_norm": 2.6767401695251465,
      "learning_rate": 0.0,
      "loss": 0.6863,
      "step": 10664
    },
    {
      "epoch": 0.9571890145395799,
      "grad_norm": 5.166617393493652,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 10665
    },
    {
      "epoch": 0.9572787650332076,
      "grad_norm": 3.8723084926605225,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 10666
    },
    {
      "epoch": 0.9573685155268354,
      "grad_norm": 6.298273086547852,
      "learning_rate": 0.0,
      "loss": 0.7578,
      "step": 10667
    },
    {
      "epoch": 0.9574582660204631,
      "grad_norm": 2.9004268646240234,
      "learning_rate": 0.0,
      "loss": 0.6583,
      "step": 10668
    },
    {
      "epoch": 0.9575480165140908,
      "grad_norm": 2.660320997238159,
      "learning_rate": 0.0,
      "loss": 0.6921,
      "step": 10669
    },
    {
      "epoch": 0.9576377670077185,
      "grad_norm": 3.0610456466674805,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 10670
    },
    {
      "epoch": 0.9577275175013462,
      "grad_norm": 4.017165660858154,
      "learning_rate": 0.0,
      "loss": 0.6886,
      "step": 10671
    },
    {
      "epoch": 0.957817267994974,
      "grad_norm": 3.074009418487549,
      "learning_rate": 0.0,
      "loss": 0.696,
      "step": 10672
    },
    {
      "epoch": 0.9579070184886017,
      "grad_norm": 2.883073568344116,
      "learning_rate": 0.0,
      "loss": 0.7058,
      "step": 10673
    },
    {
      "epoch": 0.9579967689822294,
      "grad_norm": 3.715709924697876,
      "learning_rate": 0.0,
      "loss": 0.6728,
      "step": 10674
    },
    {
      "epoch": 0.9580865194758571,
      "grad_norm": 3.4643309116363525,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 10675
    },
    {
      "epoch": 0.9581762699694848,
      "grad_norm": 2.7046244144439697,
      "learning_rate": 0.0,
      "loss": 0.6293,
      "step": 10676
    },
    {
      "epoch": 0.9582660204631126,
      "grad_norm": 4.2167582511901855,
      "learning_rate": 0.0,
      "loss": 0.6719,
      "step": 10677
    },
    {
      "epoch": 0.9583557709567403,
      "grad_norm": 2.489732265472412,
      "learning_rate": 0.0,
      "loss": 0.6444,
      "step": 10678
    },
    {
      "epoch": 0.958445521450368,
      "grad_norm": 3.945883274078369,
      "learning_rate": 0.0,
      "loss": 0.6771,
      "step": 10679
    },
    {
      "epoch": 0.9585352719439957,
      "grad_norm": 2.7131459712982178,
      "learning_rate": 0.0,
      "loss": 0.7043,
      "step": 10680
    },
    {
      "epoch": 0.9586250224376234,
      "grad_norm": 2.875601291656494,
      "learning_rate": 0.0,
      "loss": 0.6731,
      "step": 10681
    },
    {
      "epoch": 0.9587147729312511,
      "grad_norm": 3.1298139095306396,
      "learning_rate": 0.0,
      "loss": 0.689,
      "step": 10682
    },
    {
      "epoch": 0.9588045234248789,
      "grad_norm": 2.680948495864868,
      "learning_rate": 0.0,
      "loss": 0.6825,
      "step": 10683
    },
    {
      "epoch": 0.9588942739185066,
      "grad_norm": 2.6958346366882324,
      "learning_rate": 0.0,
      "loss": 0.7608,
      "step": 10684
    },
    {
      "epoch": 0.9589840244121343,
      "grad_norm": 2.615478515625,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 10685
    },
    {
      "epoch": 0.959073774905762,
      "grad_norm": 3.1698977947235107,
      "learning_rate": 0.0,
      "loss": 0.6992,
      "step": 10686
    },
    {
      "epoch": 0.9591635253993896,
      "grad_norm": 3.699932098388672,
      "learning_rate": 0.0,
      "loss": 0.6999,
      "step": 10687
    },
    {
      "epoch": 0.9592532758930175,
      "grad_norm": 3.6202845573425293,
      "learning_rate": 0.0,
      "loss": 0.7023,
      "step": 10688
    },
    {
      "epoch": 0.9593430263866451,
      "grad_norm": 3.659961700439453,
      "learning_rate": 0.0,
      "loss": 0.6662,
      "step": 10689
    },
    {
      "epoch": 0.9594327768802728,
      "grad_norm": 2.024207830429077,
      "learning_rate": 0.0,
      "loss": 0.6969,
      "step": 10690
    },
    {
      "epoch": 0.9595225273739005,
      "grad_norm": 2.8917109966278076,
      "learning_rate": 0.0,
      "loss": 0.7291,
      "step": 10691
    },
    {
      "epoch": 0.9596122778675282,
      "grad_norm": 2.9317948818206787,
      "learning_rate": 0.0,
      "loss": 0.7323,
      "step": 10692
    },
    {
      "epoch": 0.959702028361156,
      "grad_norm": 3.844506025314331,
      "learning_rate": 0.0,
      "loss": 0.701,
      "step": 10693
    },
    {
      "epoch": 0.9597917788547837,
      "grad_norm": 2.7683963775634766,
      "learning_rate": 0.0,
      "loss": 0.7451,
      "step": 10694
    },
    {
      "epoch": 0.9598815293484114,
      "grad_norm": 2.8940107822418213,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 10695
    },
    {
      "epoch": 0.9599712798420391,
      "grad_norm": 2.834806442260742,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 10696
    },
    {
      "epoch": 0.9600610303356668,
      "grad_norm": 3.1961307525634766,
      "learning_rate": 0.0,
      "loss": 0.6992,
      "step": 10697
    },
    {
      "epoch": 0.9601507808292946,
      "grad_norm": 3.38564133644104,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 10698
    },
    {
      "epoch": 0.9602405313229223,
      "grad_norm": 2.53865647315979,
      "learning_rate": 0.0,
      "loss": 0.703,
      "step": 10699
    },
    {
      "epoch": 0.96033028181655,
      "grad_norm": 3.015040636062622,
      "learning_rate": 0.0,
      "loss": 0.677,
      "step": 10700
    },
    {
      "epoch": 0.9604200323101777,
      "grad_norm": 2.4240822792053223,
      "learning_rate": 0.0,
      "loss": 0.7377,
      "step": 10701
    },
    {
      "epoch": 0.9605097828038054,
      "grad_norm": 2.8299851417541504,
      "learning_rate": 0.0,
      "loss": 0.7064,
      "step": 10702
    },
    {
      "epoch": 0.9605995332974331,
      "grad_norm": 3.2348151206970215,
      "learning_rate": 0.0,
      "loss": 0.7605,
      "step": 10703
    },
    {
      "epoch": 0.9606892837910609,
      "grad_norm": 3.374739646911621,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 10704
    },
    {
      "epoch": 0.9607790342846886,
      "grad_norm": 2.107326030731201,
      "learning_rate": 0.0,
      "loss": 0.7089,
      "step": 10705
    },
    {
      "epoch": 0.9608687847783163,
      "grad_norm": 3.8187334537506104,
      "learning_rate": 0.0,
      "loss": 0.7269,
      "step": 10706
    },
    {
      "epoch": 0.960958535271944,
      "grad_norm": 2.979088306427002,
      "learning_rate": 0.0,
      "loss": 0.7275,
      "step": 10707
    },
    {
      "epoch": 0.9610482857655717,
      "grad_norm": 4.057306289672852,
      "learning_rate": 0.0,
      "loss": 0.6638,
      "step": 10708
    },
    {
      "epoch": 0.9611380362591995,
      "grad_norm": 3.2751574516296387,
      "learning_rate": 0.0,
      "loss": 0.717,
      "step": 10709
    },
    {
      "epoch": 0.9612277867528272,
      "grad_norm": 2.5309298038482666,
      "learning_rate": 0.0,
      "loss": 0.6998,
      "step": 10710
    },
    {
      "epoch": 0.9613175372464549,
      "grad_norm": 2.861837387084961,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 10711
    },
    {
      "epoch": 0.9614072877400826,
      "grad_norm": 2.7661731243133545,
      "learning_rate": 0.0,
      "loss": 0.6903,
      "step": 10712
    },
    {
      "epoch": 0.9614970382337102,
      "grad_norm": 2.9371345043182373,
      "learning_rate": 0.0,
      "loss": 0.7595,
      "step": 10713
    },
    {
      "epoch": 0.961586788727338,
      "grad_norm": 2.588679313659668,
      "learning_rate": 0.0,
      "loss": 0.7162,
      "step": 10714
    },
    {
      "epoch": 0.9616765392209657,
      "grad_norm": 2.5800609588623047,
      "learning_rate": 0.0,
      "loss": 0.7889,
      "step": 10715
    },
    {
      "epoch": 0.9617662897145934,
      "grad_norm": 3.301069974899292,
      "learning_rate": 0.0,
      "loss": 0.7376,
      "step": 10716
    },
    {
      "epoch": 0.9618560402082211,
      "grad_norm": 3.2373507022857666,
      "learning_rate": 0.0,
      "loss": 0.6558,
      "step": 10717
    },
    {
      "epoch": 0.9619457907018488,
      "grad_norm": 3.2432615756988525,
      "learning_rate": 0.0,
      "loss": 0.7254,
      "step": 10718
    },
    {
      "epoch": 0.9620355411954766,
      "grad_norm": 3.2265069484710693,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 10719
    },
    {
      "epoch": 0.9621252916891043,
      "grad_norm": 4.240284442901611,
      "learning_rate": 0.0,
      "loss": 0.738,
      "step": 10720
    },
    {
      "epoch": 0.962215042182732,
      "grad_norm": 2.834725856781006,
      "learning_rate": 0.0,
      "loss": 0.7356,
      "step": 10721
    },
    {
      "epoch": 0.9623047926763597,
      "grad_norm": 2.0479111671447754,
      "learning_rate": 0.0,
      "loss": 0.6653,
      "step": 10722
    },
    {
      "epoch": 0.9623945431699874,
      "grad_norm": 2.8488948345184326,
      "learning_rate": 0.0,
      "loss": 0.7111,
      "step": 10723
    },
    {
      "epoch": 0.9624842936636151,
      "grad_norm": 3.491791248321533,
      "learning_rate": 0.0,
      "loss": 0.7384,
      "step": 10724
    },
    {
      "epoch": 0.9625740441572429,
      "grad_norm": 2.6450955867767334,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 10725
    },
    {
      "epoch": 0.9626637946508706,
      "grad_norm": 2.9988155364990234,
      "learning_rate": 0.0,
      "loss": 0.72,
      "step": 10726
    },
    {
      "epoch": 0.9627535451444983,
      "grad_norm": 2.7209722995758057,
      "learning_rate": 0.0,
      "loss": 0.7545,
      "step": 10727
    },
    {
      "epoch": 0.962843295638126,
      "grad_norm": 3.2483737468719482,
      "learning_rate": 0.0,
      "loss": 0.6864,
      "step": 10728
    },
    {
      "epoch": 0.9629330461317537,
      "grad_norm": 3.4804840087890625,
      "learning_rate": 0.0,
      "loss": 0.7686,
      "step": 10729
    },
    {
      "epoch": 0.9630227966253815,
      "grad_norm": 2.5244698524475098,
      "learning_rate": 0.0,
      "loss": 0.7,
      "step": 10730
    },
    {
      "epoch": 0.9631125471190092,
      "grad_norm": 2.04065203666687,
      "learning_rate": 0.0,
      "loss": 0.6508,
      "step": 10731
    },
    {
      "epoch": 0.9632022976126369,
      "grad_norm": 3.75066876411438,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 10732
    },
    {
      "epoch": 0.9632920481062646,
      "grad_norm": 2.369234561920166,
      "learning_rate": 0.0,
      "loss": 0.7029,
      "step": 10733
    },
    {
      "epoch": 0.9633817985998923,
      "grad_norm": 2.282959461212158,
      "learning_rate": 0.0,
      "loss": 0.6575,
      "step": 10734
    },
    {
      "epoch": 0.9634715490935201,
      "grad_norm": 2.2882299423217773,
      "learning_rate": 0.0,
      "loss": 0.6787,
      "step": 10735
    },
    {
      "epoch": 0.9635612995871478,
      "grad_norm": 2.53678297996521,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 10736
    },
    {
      "epoch": 0.9636510500807755,
      "grad_norm": 3.133054733276367,
      "learning_rate": 0.0,
      "loss": 0.7137,
      "step": 10737
    },
    {
      "epoch": 0.9637408005744031,
      "grad_norm": 2.5290753841400146,
      "learning_rate": 0.0,
      "loss": 0.7076,
      "step": 10738
    },
    {
      "epoch": 0.9638305510680308,
      "grad_norm": 2.660337448120117,
      "learning_rate": 0.0,
      "loss": 0.8008,
      "step": 10739
    },
    {
      "epoch": 0.9639203015616586,
      "grad_norm": 8.673332214355469,
      "learning_rate": 0.0,
      "loss": 0.742,
      "step": 10740
    },
    {
      "epoch": 0.9640100520552863,
      "grad_norm": 2.7605297565460205,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 10741
    },
    {
      "epoch": 0.964099802548914,
      "grad_norm": 3.0277745723724365,
      "learning_rate": 0.0,
      "loss": 0.6981,
      "step": 10742
    },
    {
      "epoch": 0.9641895530425417,
      "grad_norm": 2.983290433883667,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 10743
    },
    {
      "epoch": 0.9642793035361694,
      "grad_norm": 2.890322208404541,
      "learning_rate": 0.0,
      "loss": 0.7091,
      "step": 10744
    },
    {
      "epoch": 0.9643690540297971,
      "grad_norm": 3.187141180038452,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 10745
    },
    {
      "epoch": 0.9644588045234249,
      "grad_norm": 13.029450416564941,
      "learning_rate": 0.0,
      "loss": 0.7174,
      "step": 10746
    },
    {
      "epoch": 0.9645485550170526,
      "grad_norm": 2.357435464859009,
      "learning_rate": 0.0,
      "loss": 0.766,
      "step": 10747
    },
    {
      "epoch": 0.9646383055106803,
      "grad_norm": 2.7966511249542236,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 10748
    },
    {
      "epoch": 0.964728056004308,
      "grad_norm": 2.30964732170105,
      "learning_rate": 0.0,
      "loss": 0.7295,
      "step": 10749
    },
    {
      "epoch": 0.9648178064979357,
      "grad_norm": 2.429506778717041,
      "learning_rate": 0.0,
      "loss": 0.6807,
      "step": 10750
    },
    {
      "epoch": 0.9649075569915635,
      "grad_norm": 4.877260684967041,
      "learning_rate": 0.0,
      "loss": 0.6487,
      "step": 10751
    },
    {
      "epoch": 0.9649973074851912,
      "grad_norm": 3.2663626670837402,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 10752
    },
    {
      "epoch": 0.9650870579788189,
      "grad_norm": 7.291949272155762,
      "learning_rate": 0.0,
      "loss": 0.6785,
      "step": 10753
    },
    {
      "epoch": 0.9651768084724466,
      "grad_norm": 2.73357892036438,
      "learning_rate": 0.0,
      "loss": 0.7241,
      "step": 10754
    },
    {
      "epoch": 0.9652665589660743,
      "grad_norm": 4.213440895080566,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 10755
    },
    {
      "epoch": 0.9653563094597021,
      "grad_norm": 3.4927868843078613,
      "learning_rate": 0.0,
      "loss": 0.68,
      "step": 10756
    },
    {
      "epoch": 0.9654460599533298,
      "grad_norm": 2.884549856185913,
      "learning_rate": 0.0,
      "loss": 0.6341,
      "step": 10757
    },
    {
      "epoch": 0.9655358104469575,
      "grad_norm": 2.6597607135772705,
      "learning_rate": 0.0,
      "loss": 0.7403,
      "step": 10758
    },
    {
      "epoch": 0.9656255609405852,
      "grad_norm": 2.4950103759765625,
      "learning_rate": 0.0,
      "loss": 0.7072,
      "step": 10759
    },
    {
      "epoch": 0.9657153114342129,
      "grad_norm": 2.6891350746154785,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 10760
    },
    {
      "epoch": 0.9658050619278407,
      "grad_norm": 2.617981433868408,
      "learning_rate": 0.0,
      "loss": 0.7153,
      "step": 10761
    },
    {
      "epoch": 0.9658948124214684,
      "grad_norm": 3.654815912246704,
      "learning_rate": 0.0,
      "loss": 0.7841,
      "step": 10762
    },
    {
      "epoch": 0.965984562915096,
      "grad_norm": 2.42195463180542,
      "learning_rate": 0.0,
      "loss": 0.6818,
      "step": 10763
    },
    {
      "epoch": 0.9660743134087237,
      "grad_norm": 3.1172375679016113,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 10764
    },
    {
      "epoch": 0.9661640639023514,
      "grad_norm": 3.9085044860839844,
      "learning_rate": 0.0,
      "loss": 0.6823,
      "step": 10765
    },
    {
      "epoch": 0.9662538143959791,
      "grad_norm": 3.75185227394104,
      "learning_rate": 0.0,
      "loss": 0.7134,
      "step": 10766
    },
    {
      "epoch": 0.9663435648896069,
      "grad_norm": 2.9071383476257324,
      "learning_rate": 0.0,
      "loss": 0.7371,
      "step": 10767
    },
    {
      "epoch": 0.9664333153832346,
      "grad_norm": 3.8936939239501953,
      "learning_rate": 0.0,
      "loss": 0.7556,
      "step": 10768
    },
    {
      "epoch": 0.9665230658768623,
      "grad_norm": 3.689145803451538,
      "learning_rate": 0.0,
      "loss": 0.7039,
      "step": 10769
    },
    {
      "epoch": 0.96661281637049,
      "grad_norm": 3.383333206176758,
      "learning_rate": 0.0,
      "loss": 0.6957,
      "step": 10770
    },
    {
      "epoch": 0.9667025668641177,
      "grad_norm": 3.2578818798065186,
      "learning_rate": 0.0,
      "loss": 0.7562,
      "step": 10771
    },
    {
      "epoch": 0.9667923173577455,
      "grad_norm": 2.8164308071136475,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 10772
    },
    {
      "epoch": 0.9668820678513732,
      "grad_norm": 3.017946481704712,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 10773
    },
    {
      "epoch": 0.9669718183450009,
      "grad_norm": 2.4990031719207764,
      "learning_rate": 0.0,
      "loss": 0.7038,
      "step": 10774
    },
    {
      "epoch": 0.9670615688386286,
      "grad_norm": 2.8451714515686035,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 10775
    },
    {
      "epoch": 0.9671513193322563,
      "grad_norm": 2.846940040588379,
      "learning_rate": 0.0,
      "loss": 0.7132,
      "step": 10776
    },
    {
      "epoch": 0.9672410698258841,
      "grad_norm": 2.67615008354187,
      "learning_rate": 0.0,
      "loss": 0.7053,
      "step": 10777
    },
    {
      "epoch": 0.9673308203195118,
      "grad_norm": 2.9219720363616943,
      "learning_rate": 0.0,
      "loss": 0.704,
      "step": 10778
    },
    {
      "epoch": 0.9674205708131395,
      "grad_norm": 3.7786476612091064,
      "learning_rate": 0.0,
      "loss": 0.7331,
      "step": 10779
    },
    {
      "epoch": 0.9675103213067672,
      "grad_norm": 3.231462240219116,
      "learning_rate": 0.0,
      "loss": 0.6639,
      "step": 10780
    },
    {
      "epoch": 0.9676000718003949,
      "grad_norm": 2.7425079345703125,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 10781
    },
    {
      "epoch": 0.9676898222940227,
      "grad_norm": 2.5327706336975098,
      "learning_rate": 0.0,
      "loss": 0.77,
      "step": 10782
    },
    {
      "epoch": 0.9677795727876504,
      "grad_norm": 3.547264814376831,
      "learning_rate": 0.0,
      "loss": 0.6533,
      "step": 10783
    },
    {
      "epoch": 0.9678693232812781,
      "grad_norm": 2.7628097534179688,
      "learning_rate": 0.0,
      "loss": 0.7493,
      "step": 10784
    },
    {
      "epoch": 0.9679590737749058,
      "grad_norm": 2.813234329223633,
      "learning_rate": 0.0,
      "loss": 0.728,
      "step": 10785
    },
    {
      "epoch": 0.9680488242685334,
      "grad_norm": 3.102775812149048,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 10786
    },
    {
      "epoch": 0.9681385747621611,
      "grad_norm": 3.4882116317749023,
      "learning_rate": 0.0,
      "loss": 0.6744,
      "step": 10787
    },
    {
      "epoch": 0.968228325255789,
      "grad_norm": 2.6187503337860107,
      "learning_rate": 0.0,
      "loss": 0.7258,
      "step": 10788
    },
    {
      "epoch": 0.9683180757494166,
      "grad_norm": 3.540334939956665,
      "learning_rate": 0.0,
      "loss": 0.7287,
      "step": 10789
    },
    {
      "epoch": 0.9684078262430443,
      "grad_norm": 3.637263536453247,
      "learning_rate": 0.0,
      "loss": 0.6839,
      "step": 10790
    },
    {
      "epoch": 0.968497576736672,
      "grad_norm": 3.1800522804260254,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 10791
    },
    {
      "epoch": 0.9685873272302997,
      "grad_norm": 2.850780963897705,
      "learning_rate": 0.0,
      "loss": 0.6913,
      "step": 10792
    },
    {
      "epoch": 0.9686770777239275,
      "grad_norm": 2.796635866165161,
      "learning_rate": 0.0,
      "loss": 0.6845,
      "step": 10793
    },
    {
      "epoch": 0.9687668282175552,
      "grad_norm": 3.361349105834961,
      "learning_rate": 0.0,
      "loss": 0.6971,
      "step": 10794
    },
    {
      "epoch": 0.9688565787111829,
      "grad_norm": 3.0360493659973145,
      "learning_rate": 0.0,
      "loss": 0.6869,
      "step": 10795
    },
    {
      "epoch": 0.9689463292048106,
      "grad_norm": 3.166435956954956,
      "learning_rate": 0.0,
      "loss": 0.6243,
      "step": 10796
    },
    {
      "epoch": 0.9690360796984383,
      "grad_norm": 2.7807836532592773,
      "learning_rate": 0.0,
      "loss": 0.7171,
      "step": 10797
    },
    {
      "epoch": 0.9691258301920661,
      "grad_norm": 3.2118966579437256,
      "learning_rate": 0.0,
      "loss": 0.6712,
      "step": 10798
    },
    {
      "epoch": 0.9692155806856938,
      "grad_norm": 3.5864107608795166,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 10799
    },
    {
      "epoch": 0.9693053311793215,
      "grad_norm": 2.711509943008423,
      "learning_rate": 0.0,
      "loss": 0.6661,
      "step": 10800
    },
    {
      "epoch": 0.9693950816729492,
      "grad_norm": 4.52071475982666,
      "learning_rate": 0.0,
      "loss": 0.7216,
      "step": 10801
    },
    {
      "epoch": 0.9694848321665769,
      "grad_norm": 3.2763400077819824,
      "learning_rate": 0.0,
      "loss": 0.6978,
      "step": 10802
    },
    {
      "epoch": 0.9695745826602046,
      "grad_norm": 2.6885101795196533,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 10803
    },
    {
      "epoch": 0.9696643331538324,
      "grad_norm": 2.8993537425994873,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 10804
    },
    {
      "epoch": 0.9697540836474601,
      "grad_norm": 3.4356954097747803,
      "learning_rate": 0.0,
      "loss": 0.6805,
      "step": 10805
    },
    {
      "epoch": 0.9698438341410878,
      "grad_norm": 3.1653900146484375,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 10806
    },
    {
      "epoch": 0.9699335846347155,
      "grad_norm": 2.9966864585876465,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 10807
    },
    {
      "epoch": 0.9700233351283432,
      "grad_norm": 3.5526537895202637,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 10808
    },
    {
      "epoch": 0.970113085621971,
      "grad_norm": 3.751601219177246,
      "learning_rate": 0.0,
      "loss": 0.6662,
      "step": 10809
    },
    {
      "epoch": 0.9702028361155987,
      "grad_norm": 3.7395591735839844,
      "learning_rate": 0.0,
      "loss": 0.6778,
      "step": 10810
    },
    {
      "epoch": 0.9702925866092263,
      "grad_norm": 2.9456093311309814,
      "learning_rate": 0.0,
      "loss": 0.7353,
      "step": 10811
    },
    {
      "epoch": 0.970382337102854,
      "grad_norm": 2.9296107292175293,
      "learning_rate": 0.0,
      "loss": 0.7107,
      "step": 10812
    },
    {
      "epoch": 0.9704720875964817,
      "grad_norm": 2.9576520919799805,
      "learning_rate": 0.0,
      "loss": 0.6917,
      "step": 10813
    },
    {
      "epoch": 0.9705618380901095,
      "grad_norm": 3.8661000728607178,
      "learning_rate": 0.0,
      "loss": 0.6816,
      "step": 10814
    },
    {
      "epoch": 0.9706515885837372,
      "grad_norm": 2.7780258655548096,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 10815
    },
    {
      "epoch": 0.9707413390773649,
      "grad_norm": 3.0340871810913086,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 10816
    },
    {
      "epoch": 0.9708310895709926,
      "grad_norm": 2.8552026748657227,
      "learning_rate": 0.0,
      "loss": 0.68,
      "step": 10817
    },
    {
      "epoch": 0.9709208400646203,
      "grad_norm": 2.658226490020752,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 10818
    },
    {
      "epoch": 0.9710105905582481,
      "grad_norm": 10.936246871948242,
      "learning_rate": 0.0,
      "loss": 0.7228,
      "step": 10819
    },
    {
      "epoch": 0.9711003410518758,
      "grad_norm": 4.470572471618652,
      "learning_rate": 0.0,
      "loss": 0.6748,
      "step": 10820
    },
    {
      "epoch": 0.9711900915455035,
      "grad_norm": 3.6598684787750244,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 10821
    },
    {
      "epoch": 0.9712798420391312,
      "grad_norm": 3.2639787197113037,
      "learning_rate": 0.0,
      "loss": 0.7271,
      "step": 10822
    },
    {
      "epoch": 0.9713695925327589,
      "grad_norm": 3.4487855434417725,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 10823
    },
    {
      "epoch": 0.9714593430263866,
      "grad_norm": 2.805166482925415,
      "learning_rate": 0.0,
      "loss": 0.6799,
      "step": 10824
    },
    {
      "epoch": 0.9715490935200144,
      "grad_norm": 2.4415507316589355,
      "learning_rate": 0.0,
      "loss": 0.6729,
      "step": 10825
    },
    {
      "epoch": 0.9716388440136421,
      "grad_norm": 2.499377489089966,
      "learning_rate": 0.0,
      "loss": 0.6844,
      "step": 10826
    },
    {
      "epoch": 0.9717285945072698,
      "grad_norm": 2.461040496826172,
      "learning_rate": 0.0,
      "loss": 0.7116,
      "step": 10827
    },
    {
      "epoch": 0.9718183450008975,
      "grad_norm": 2.8857741355895996,
      "learning_rate": 0.0,
      "loss": 0.6949,
      "step": 10828
    },
    {
      "epoch": 0.9719080954945252,
      "grad_norm": 2.5894672870635986,
      "learning_rate": 0.0,
      "loss": 0.7577,
      "step": 10829
    },
    {
      "epoch": 0.971997845988153,
      "grad_norm": 3.9319519996643066,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 10830
    },
    {
      "epoch": 0.9720875964817807,
      "grad_norm": 3.2420012950897217,
      "learning_rate": 0.0,
      "loss": 0.7004,
      "step": 10831
    },
    {
      "epoch": 0.9721773469754084,
      "grad_norm": 3.132086992263794,
      "learning_rate": 0.0,
      "loss": 0.6703,
      "step": 10832
    },
    {
      "epoch": 0.9722670974690361,
      "grad_norm": 3.4885120391845703,
      "learning_rate": 0.0,
      "loss": 0.7959,
      "step": 10833
    },
    {
      "epoch": 0.9723568479626638,
      "grad_norm": 2.914914608001709,
      "learning_rate": 0.0,
      "loss": 0.6868,
      "step": 10834
    },
    {
      "epoch": 0.9724465984562916,
      "grad_norm": 3.0520834922790527,
      "learning_rate": 0.0,
      "loss": 0.7197,
      "step": 10835
    },
    {
      "epoch": 0.9725363489499192,
      "grad_norm": 3.1363255977630615,
      "learning_rate": 0.0,
      "loss": 0.6952,
      "step": 10836
    },
    {
      "epoch": 0.9726260994435469,
      "grad_norm": 3.395185708999634,
      "learning_rate": 0.0,
      "loss": 0.6895,
      "step": 10837
    },
    {
      "epoch": 0.9727158499371746,
      "grad_norm": 3.9895401000976562,
      "learning_rate": 0.0,
      "loss": 0.6792,
      "step": 10838
    },
    {
      "epoch": 0.9728056004308023,
      "grad_norm": 2.8839964866638184,
      "learning_rate": 0.0,
      "loss": 0.7355,
      "step": 10839
    },
    {
      "epoch": 0.9728953509244301,
      "grad_norm": 3.5034432411193848,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 10840
    },
    {
      "epoch": 0.9729851014180578,
      "grad_norm": 3.659705400466919,
      "learning_rate": 0.0,
      "loss": 0.7517,
      "step": 10841
    },
    {
      "epoch": 0.9730748519116855,
      "grad_norm": 2.7402894496917725,
      "learning_rate": 0.0,
      "loss": 0.7231,
      "step": 10842
    },
    {
      "epoch": 0.9731646024053132,
      "grad_norm": 2.684082269668579,
      "learning_rate": 0.0,
      "loss": 0.6948,
      "step": 10843
    },
    {
      "epoch": 0.9732543528989409,
      "grad_norm": 3.027674674987793,
      "learning_rate": 0.0,
      "loss": 0.6855,
      "step": 10844
    },
    {
      "epoch": 0.9733441033925686,
      "grad_norm": 3.0916619300842285,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 10845
    },
    {
      "epoch": 0.9734338538861964,
      "grad_norm": 2.4605295658111572,
      "learning_rate": 0.0,
      "loss": 0.6401,
      "step": 10846
    },
    {
      "epoch": 0.9735236043798241,
      "grad_norm": 2.938849687576294,
      "learning_rate": 0.0,
      "loss": 0.6782,
      "step": 10847
    },
    {
      "epoch": 0.9736133548734518,
      "grad_norm": 2.7583296298980713,
      "learning_rate": 0.0,
      "loss": 0.6809,
      "step": 10848
    },
    {
      "epoch": 0.9737031053670795,
      "grad_norm": 3.2305421829223633,
      "learning_rate": 0.0,
      "loss": 0.6684,
      "step": 10849
    },
    {
      "epoch": 0.9737928558607072,
      "grad_norm": 2.646923542022705,
      "learning_rate": 0.0,
      "loss": 0.6941,
      "step": 10850
    },
    {
      "epoch": 0.973882606354335,
      "grad_norm": 2.60383677482605,
      "learning_rate": 0.0,
      "loss": 0.6804,
      "step": 10851
    },
    {
      "epoch": 0.9739723568479627,
      "grad_norm": 2.700174570083618,
      "learning_rate": 0.0,
      "loss": 0.6867,
      "step": 10852
    },
    {
      "epoch": 0.9740621073415904,
      "grad_norm": 2.5451624393463135,
      "learning_rate": 0.0,
      "loss": 0.6931,
      "step": 10853
    },
    {
      "epoch": 0.9741518578352181,
      "grad_norm": 3.517202138900757,
      "learning_rate": 0.0,
      "loss": 0.691,
      "step": 10854
    },
    {
      "epoch": 0.9742416083288458,
      "grad_norm": 2.873244047164917,
      "learning_rate": 0.0,
      "loss": 0.7008,
      "step": 10855
    },
    {
      "epoch": 0.9743313588224736,
      "grad_norm": 3.3866772651672363,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 10856
    },
    {
      "epoch": 0.9744211093161013,
      "grad_norm": 2.7766952514648438,
      "learning_rate": 0.0,
      "loss": 0.759,
      "step": 10857
    },
    {
      "epoch": 0.974510859809729,
      "grad_norm": 2.9504377841949463,
      "learning_rate": 0.0,
      "loss": 0.7095,
      "step": 10858
    },
    {
      "epoch": 0.9746006103033567,
      "grad_norm": 2.933664560317993,
      "learning_rate": 0.0,
      "loss": 0.6906,
      "step": 10859
    },
    {
      "epoch": 0.9746903607969843,
      "grad_norm": 2.397027015686035,
      "learning_rate": 0.0,
      "loss": 0.7708,
      "step": 10860
    },
    {
      "epoch": 0.9747801112906122,
      "grad_norm": 5.290928363800049,
      "learning_rate": 0.0,
      "loss": 0.6995,
      "step": 10861
    },
    {
      "epoch": 0.9748698617842398,
      "grad_norm": 2.6641955375671387,
      "learning_rate": 0.0,
      "loss": 0.7001,
      "step": 10862
    },
    {
      "epoch": 0.9749596122778675,
      "grad_norm": 3.6193809509277344,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 10863
    },
    {
      "epoch": 0.9750493627714952,
      "grad_norm": 2.692394971847534,
      "learning_rate": 0.0,
      "loss": 0.6937,
      "step": 10864
    },
    {
      "epoch": 0.9751391132651229,
      "grad_norm": 2.823225498199463,
      "learning_rate": 0.0,
      "loss": 0.647,
      "step": 10865
    },
    {
      "epoch": 0.9752288637587506,
      "grad_norm": 3.3587934970855713,
      "learning_rate": 0.0,
      "loss": 0.7022,
      "step": 10866
    },
    {
      "epoch": 0.9753186142523784,
      "grad_norm": 4.316516399383545,
      "learning_rate": 0.0,
      "loss": 0.6442,
      "step": 10867
    },
    {
      "epoch": 0.9754083647460061,
      "grad_norm": 3.017401695251465,
      "learning_rate": 0.0,
      "loss": 0.7262,
      "step": 10868
    },
    {
      "epoch": 0.9754981152396338,
      "grad_norm": 2.8271028995513916,
      "learning_rate": 0.0,
      "loss": 0.7276,
      "step": 10869
    },
    {
      "epoch": 0.9755878657332615,
      "grad_norm": 2.6644630432128906,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 10870
    },
    {
      "epoch": 0.9756776162268892,
      "grad_norm": 3.3948729038238525,
      "learning_rate": 0.0,
      "loss": 0.66,
      "step": 10871
    },
    {
      "epoch": 0.975767366720517,
      "grad_norm": 3.142561674118042,
      "learning_rate": 0.0,
      "loss": 0.7537,
      "step": 10872
    },
    {
      "epoch": 0.9758571172141447,
      "grad_norm": 9.824342727661133,
      "learning_rate": 0.0,
      "loss": 0.7604,
      "step": 10873
    },
    {
      "epoch": 0.9759468677077724,
      "grad_norm": 3.5621869564056396,
      "learning_rate": 0.0,
      "loss": 0.7176,
      "step": 10874
    },
    {
      "epoch": 0.9760366182014001,
      "grad_norm": 2.826565980911255,
      "learning_rate": 0.0,
      "loss": 0.7063,
      "step": 10875
    },
    {
      "epoch": 0.9761263686950278,
      "grad_norm": 2.703594446182251,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 10876
    },
    {
      "epoch": 0.9762161191886556,
      "grad_norm": 5.39281702041626,
      "learning_rate": 0.0,
      "loss": 0.7145,
      "step": 10877
    },
    {
      "epoch": 0.9763058696822833,
      "grad_norm": 3.394995927810669,
      "learning_rate": 0.0,
      "loss": 0.7475,
      "step": 10878
    },
    {
      "epoch": 0.976395620175911,
      "grad_norm": 2.752136707305908,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 10879
    },
    {
      "epoch": 0.9764853706695387,
      "grad_norm": 2.8049542903900146,
      "learning_rate": 0.0,
      "loss": 0.7096,
      "step": 10880
    },
    {
      "epoch": 0.9765751211631664,
      "grad_norm": 3.0160930156707764,
      "learning_rate": 0.0,
      "loss": 0.6837,
      "step": 10881
    },
    {
      "epoch": 0.9766648716567942,
      "grad_norm": 2.822216510772705,
      "learning_rate": 0.0,
      "loss": 0.6434,
      "step": 10882
    },
    {
      "epoch": 0.9767546221504219,
      "grad_norm": 3.45719575881958,
      "learning_rate": 0.0,
      "loss": 0.7135,
      "step": 10883
    },
    {
      "epoch": 0.9768443726440496,
      "grad_norm": 2.6154227256774902,
      "learning_rate": 0.0,
      "loss": 0.6634,
      "step": 10884
    },
    {
      "epoch": 0.9769341231376772,
      "grad_norm": 3.5634679794311523,
      "learning_rate": 0.0,
      "loss": 0.6537,
      "step": 10885
    },
    {
      "epoch": 0.9770238736313049,
      "grad_norm": 6.053079605102539,
      "learning_rate": 0.0,
      "loss": 0.7186,
      "step": 10886
    },
    {
      "epoch": 0.9771136241249326,
      "grad_norm": 3.1085383892059326,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 10887
    },
    {
      "epoch": 0.9772033746185604,
      "grad_norm": 5.531793117523193,
      "learning_rate": 0.0,
      "loss": 0.7025,
      "step": 10888
    },
    {
      "epoch": 0.9772931251121881,
      "grad_norm": 3.146989583969116,
      "learning_rate": 0.0,
      "loss": 0.6508,
      "step": 10889
    },
    {
      "epoch": 0.9773828756058158,
      "grad_norm": 2.527311325073242,
      "learning_rate": 0.0,
      "loss": 0.7068,
      "step": 10890
    },
    {
      "epoch": 0.9774726260994435,
      "grad_norm": 3.4670681953430176,
      "learning_rate": 0.0,
      "loss": 0.7468,
      "step": 10891
    },
    {
      "epoch": 0.9775623765930712,
      "grad_norm": 3.365689516067505,
      "learning_rate": 0.0,
      "loss": 0.7832,
      "step": 10892
    },
    {
      "epoch": 0.977652127086699,
      "grad_norm": 3.0313785076141357,
      "learning_rate": 0.0,
      "loss": 0.777,
      "step": 10893
    },
    {
      "epoch": 0.9777418775803267,
      "grad_norm": 3.190437078475952,
      "learning_rate": 0.0,
      "loss": 0.7268,
      "step": 10894
    },
    {
      "epoch": 0.9778316280739544,
      "grad_norm": 3.7040953636169434,
      "learning_rate": 0.0,
      "loss": 0.7302,
      "step": 10895
    },
    {
      "epoch": 0.9779213785675821,
      "grad_norm": 3.778346300125122,
      "learning_rate": 0.0,
      "loss": 0.7223,
      "step": 10896
    },
    {
      "epoch": 0.9780111290612098,
      "grad_norm": 2.6999130249023438,
      "learning_rate": 0.0,
      "loss": 0.7243,
      "step": 10897
    },
    {
      "epoch": 0.9781008795548376,
      "grad_norm": 3.6300220489501953,
      "learning_rate": 0.0,
      "loss": 0.7294,
      "step": 10898
    },
    {
      "epoch": 0.9781906300484653,
      "grad_norm": 3.052529811859131,
      "learning_rate": 0.0,
      "loss": 0.7194,
      "step": 10899
    },
    {
      "epoch": 0.978280380542093,
      "grad_norm": 3.0962867736816406,
      "learning_rate": 0.0,
      "loss": 0.7215,
      "step": 10900
    },
    {
      "epoch": 0.9783701310357207,
      "grad_norm": 3.6713035106658936,
      "learning_rate": 0.0,
      "loss": 0.6786,
      "step": 10901
    },
    {
      "epoch": 0.9784598815293484,
      "grad_norm": 3.106011390686035,
      "learning_rate": 0.0,
      "loss": 0.748,
      "step": 10902
    },
    {
      "epoch": 0.9785496320229762,
      "grad_norm": 3.4365017414093018,
      "learning_rate": 0.0,
      "loss": 0.7027,
      "step": 10903
    },
    {
      "epoch": 0.9786393825166039,
      "grad_norm": 3.3653721809387207,
      "learning_rate": 0.0,
      "loss": 0.6831,
      "step": 10904
    },
    {
      "epoch": 0.9787291330102316,
      "grad_norm": 4.071310043334961,
      "learning_rate": 0.0,
      "loss": 0.7205,
      "step": 10905
    },
    {
      "epoch": 0.9788188835038593,
      "grad_norm": 4.4093122482299805,
      "learning_rate": 0.0,
      "loss": 0.6755,
      "step": 10906
    },
    {
      "epoch": 0.978908633997487,
      "grad_norm": 5.361592769622803,
      "learning_rate": 0.0,
      "loss": 0.7157,
      "step": 10907
    },
    {
      "epoch": 0.9789983844911146,
      "grad_norm": 2.7234113216400146,
      "learning_rate": 0.0,
      "loss": 0.6841,
      "step": 10908
    },
    {
      "epoch": 0.9790881349847425,
      "grad_norm": 2.885512351989746,
      "learning_rate": 0.0,
      "loss": 0.6873,
      "step": 10909
    },
    {
      "epoch": 0.9791778854783701,
      "grad_norm": 2.8081133365631104,
      "learning_rate": 0.0,
      "loss": 0.6819,
      "step": 10910
    },
    {
      "epoch": 0.9792676359719978,
      "grad_norm": 12.999303817749023,
      "learning_rate": 0.0,
      "loss": 0.6843,
      "step": 10911
    },
    {
      "epoch": 0.9793573864656255,
      "grad_norm": 4.213247776031494,
      "learning_rate": 0.0,
      "loss": 0.7483,
      "step": 10912
    },
    {
      "epoch": 0.9794471369592532,
      "grad_norm": 3.328439235687256,
      "learning_rate": 0.0,
      "loss": 0.6688,
      "step": 10913
    },
    {
      "epoch": 0.979536887452881,
      "grad_norm": 3.204432487487793,
      "learning_rate": 0.0,
      "loss": 0.7343,
      "step": 10914
    },
    {
      "epoch": 0.9796266379465087,
      "grad_norm": 2.55096173286438,
      "learning_rate": 0.0,
      "loss": 0.6748,
      "step": 10915
    },
    {
      "epoch": 0.9797163884401364,
      "grad_norm": 3.69429874420166,
      "learning_rate": 0.0,
      "loss": 0.6798,
      "step": 10916
    },
    {
      "epoch": 0.9798061389337641,
      "grad_norm": 4.091972827911377,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 10917
    },
    {
      "epoch": 0.9798958894273918,
      "grad_norm": 3.420600414276123,
      "learning_rate": 0.0,
      "loss": 0.7618,
      "step": 10918
    },
    {
      "epoch": 0.9799856399210196,
      "grad_norm": 2.5095784664154053,
      "learning_rate": 0.0,
      "loss": 0.6862,
      "step": 10919
    },
    {
      "epoch": 0.9800753904146473,
      "grad_norm": 2.776139259338379,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 10920
    },
    {
      "epoch": 0.980165140908275,
      "grad_norm": 3.365588903427124,
      "learning_rate": 0.0,
      "loss": 0.7103,
      "step": 10921
    },
    {
      "epoch": 0.9802548914019027,
      "grad_norm": 3.458228588104248,
      "learning_rate": 0.0,
      "loss": 0.7097,
      "step": 10922
    },
    {
      "epoch": 0.9803446418955304,
      "grad_norm": 3.7687904834747314,
      "learning_rate": 0.0,
      "loss": 0.6976,
      "step": 10923
    },
    {
      "epoch": 0.9804343923891582,
      "grad_norm": 2.7962779998779297,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 10924
    },
    {
      "epoch": 0.9805241428827859,
      "grad_norm": 7.301973342895508,
      "learning_rate": 0.0,
      "loss": 0.6554,
      "step": 10925
    },
    {
      "epoch": 0.9806138933764136,
      "grad_norm": 3.3601796627044678,
      "learning_rate": 0.0,
      "loss": 0.7372,
      "step": 10926
    },
    {
      "epoch": 0.9807036438700413,
      "grad_norm": 2.6611833572387695,
      "learning_rate": 0.0,
      "loss": 0.7472,
      "step": 10927
    },
    {
      "epoch": 0.980793394363669,
      "grad_norm": 4.279016971588135,
      "learning_rate": 0.0,
      "loss": 0.6849,
      "step": 10928
    },
    {
      "epoch": 0.9808831448572967,
      "grad_norm": 3.638197422027588,
      "learning_rate": 0.0,
      "loss": 0.7127,
      "step": 10929
    },
    {
      "epoch": 0.9809728953509245,
      "grad_norm": 3.0443532466888428,
      "learning_rate": 0.0,
      "loss": 0.737,
      "step": 10930
    },
    {
      "epoch": 0.9810626458445522,
      "grad_norm": 4.1572418212890625,
      "learning_rate": 0.0,
      "loss": 0.7565,
      "step": 10931
    },
    {
      "epoch": 0.9811523963381799,
      "grad_norm": 3.043668270111084,
      "learning_rate": 0.0,
      "loss": 0.7827,
      "step": 10932
    },
    {
      "epoch": 0.9812421468318075,
      "grad_norm": 3.1723642349243164,
      "learning_rate": 0.0,
      "loss": 0.6956,
      "step": 10933
    },
    {
      "epoch": 0.9813318973254352,
      "grad_norm": 3.1693782806396484,
      "learning_rate": 0.0,
      "loss": 0.6691,
      "step": 10934
    },
    {
      "epoch": 0.981421647819063,
      "grad_norm": 5.000544548034668,
      "learning_rate": 0.0,
      "loss": 0.7017,
      "step": 10935
    },
    {
      "epoch": 0.9815113983126907,
      "grad_norm": 3.261742115020752,
      "learning_rate": 0.0,
      "loss": 0.6533,
      "step": 10936
    },
    {
      "epoch": 0.9816011488063184,
      "grad_norm": 2.1475067138671875,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10937
    },
    {
      "epoch": 0.9816908992999461,
      "grad_norm": 2.9410743713378906,
      "learning_rate": 0.0,
      "loss": 0.7588,
      "step": 10938
    },
    {
      "epoch": 0.9817806497935738,
      "grad_norm": 2.758070945739746,
      "learning_rate": 0.0,
      "loss": 0.6775,
      "step": 10939
    },
    {
      "epoch": 0.9818704002872016,
      "grad_norm": 2.321643352508545,
      "learning_rate": 0.0,
      "loss": 0.7356,
      "step": 10940
    },
    {
      "epoch": 0.9819601507808293,
      "grad_norm": 4.171491622924805,
      "learning_rate": 0.0,
      "loss": 0.6415,
      "step": 10941
    },
    {
      "epoch": 0.982049901274457,
      "grad_norm": 2.507350206375122,
      "learning_rate": 0.0,
      "loss": 0.7041,
      "step": 10942
    },
    {
      "epoch": 0.9821396517680847,
      "grad_norm": 2.55080246925354,
      "learning_rate": 0.0,
      "loss": 0.6712,
      "step": 10943
    },
    {
      "epoch": 0.9822294022617124,
      "grad_norm": 4.896316051483154,
      "learning_rate": 0.0,
      "loss": 0.6925,
      "step": 10944
    },
    {
      "epoch": 0.9823191527553402,
      "grad_norm": 2.9400596618652344,
      "learning_rate": 0.0,
      "loss": 0.6714,
      "step": 10945
    },
    {
      "epoch": 0.9824089032489679,
      "grad_norm": 2.6802453994750977,
      "learning_rate": 0.0,
      "loss": 0.723,
      "step": 10946
    },
    {
      "epoch": 0.9824986537425956,
      "grad_norm": 3.4240527153015137,
      "learning_rate": 0.0,
      "loss": 0.6807,
      "step": 10947
    },
    {
      "epoch": 0.9825884042362233,
      "grad_norm": 2.7918436527252197,
      "learning_rate": 0.0,
      "loss": 0.6943,
      "step": 10948
    },
    {
      "epoch": 0.982678154729851,
      "grad_norm": 2.4572885036468506,
      "learning_rate": 0.0,
      "loss": 0.6706,
      "step": 10949
    },
    {
      "epoch": 0.9827679052234787,
      "grad_norm": 2.329427719116211,
      "learning_rate": 0.0,
      "loss": 0.6889,
      "step": 10950
    },
    {
      "epoch": 0.9828576557171065,
      "grad_norm": 2.6987903118133545,
      "learning_rate": 0.0,
      "loss": 0.726,
      "step": 10951
    },
    {
      "epoch": 0.9829474062107342,
      "grad_norm": 2.7853291034698486,
      "learning_rate": 0.0,
      "loss": 0.715,
      "step": 10952
    },
    {
      "epoch": 0.9830371567043619,
      "grad_norm": 2.957179546356201,
      "learning_rate": 0.0,
      "loss": 0.6987,
      "step": 10953
    },
    {
      "epoch": 0.9831269071979896,
      "grad_norm": 3.7372288703918457,
      "learning_rate": 0.0,
      "loss": 0.7461,
      "step": 10954
    },
    {
      "epoch": 0.9832166576916173,
      "grad_norm": 2.8245127201080322,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 10955
    },
    {
      "epoch": 0.9833064081852451,
      "grad_norm": 2.864703416824341,
      "learning_rate": 0.0,
      "loss": 0.6614,
      "step": 10956
    },
    {
      "epoch": 0.9833961586788728,
      "grad_norm": 2.897038459777832,
      "learning_rate": 0.0,
      "loss": 0.6548,
      "step": 10957
    },
    {
      "epoch": 0.9834859091725004,
      "grad_norm": 3.0355589389801025,
      "learning_rate": 0.0,
      "loss": 0.6452,
      "step": 10958
    },
    {
      "epoch": 0.9835756596661281,
      "grad_norm": 2.879621982574463,
      "learning_rate": 0.0,
      "loss": 0.6489,
      "step": 10959
    },
    {
      "epoch": 0.9836654101597558,
      "grad_norm": 5.337272644042969,
      "learning_rate": 0.0,
      "loss": 0.665,
      "step": 10960
    },
    {
      "epoch": 0.9837551606533836,
      "grad_norm": 3.6371076107025146,
      "learning_rate": 0.0,
      "loss": 0.6959,
      "step": 10961
    },
    {
      "epoch": 0.9838449111470113,
      "grad_norm": 2.935171127319336,
      "learning_rate": 0.0,
      "loss": 0.7466,
      "step": 10962
    },
    {
      "epoch": 0.983934661640639,
      "grad_norm": 4.30877685546875,
      "learning_rate": 0.0,
      "loss": 0.7438,
      "step": 10963
    },
    {
      "epoch": 0.9840244121342667,
      "grad_norm": 2.577770233154297,
      "learning_rate": 0.0,
      "loss": 0.6436,
      "step": 10964
    },
    {
      "epoch": 0.9841141626278944,
      "grad_norm": 3.5282535552978516,
      "learning_rate": 0.0,
      "loss": 0.7514,
      "step": 10965
    },
    {
      "epoch": 0.9842039131215222,
      "grad_norm": 3.004476308822632,
      "learning_rate": 0.0,
      "loss": 0.6738,
      "step": 10966
    },
    {
      "epoch": 0.9842936636151499,
      "grad_norm": 3.4703140258789062,
      "learning_rate": 0.0,
      "loss": 0.6566,
      "step": 10967
    },
    {
      "epoch": 0.9843834141087776,
      "grad_norm": 3.4750523567199707,
      "learning_rate": 0.0,
      "loss": 0.7304,
      "step": 10968
    },
    {
      "epoch": 0.9844731646024053,
      "grad_norm": 2.6053755283355713,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 10969
    },
    {
      "epoch": 0.984562915096033,
      "grad_norm": 5.874868869781494,
      "learning_rate": 0.0,
      "loss": 0.6614,
      "step": 10970
    },
    {
      "epoch": 0.9846526655896607,
      "grad_norm": 3.676142454147339,
      "learning_rate": 0.0,
      "loss": 0.751,
      "step": 10971
    },
    {
      "epoch": 0.9847424160832885,
      "grad_norm": 2.652125120162964,
      "learning_rate": 0.0,
      "loss": 0.7261,
      "step": 10972
    },
    {
      "epoch": 0.9848321665769162,
      "grad_norm": 3.6128106117248535,
      "learning_rate": 0.0,
      "loss": 0.6924,
      "step": 10973
    },
    {
      "epoch": 0.9849219170705439,
      "grad_norm": 2.8964765071868896,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 10974
    },
    {
      "epoch": 0.9850116675641716,
      "grad_norm": 4.153960227966309,
      "learning_rate": 0.0,
      "loss": 0.6573,
      "step": 10975
    },
    {
      "epoch": 0.9851014180577993,
      "grad_norm": 2.0625181198120117,
      "learning_rate": 0.0,
      "loss": 0.7314,
      "step": 10976
    },
    {
      "epoch": 0.9851911685514271,
      "grad_norm": 2.3174080848693848,
      "learning_rate": 0.0,
      "loss": 0.6977,
      "step": 10977
    },
    {
      "epoch": 0.9852809190450548,
      "grad_norm": 2.5482077598571777,
      "learning_rate": 0.0,
      "loss": 0.7148,
      "step": 10978
    },
    {
      "epoch": 0.9853706695386825,
      "grad_norm": 2.34934401512146,
      "learning_rate": 0.0,
      "loss": 0.6501,
      "step": 10979
    },
    {
      "epoch": 0.9854604200323102,
      "grad_norm": 2.9222915172576904,
      "learning_rate": 0.0,
      "loss": 0.672,
      "step": 10980
    },
    {
      "epoch": 0.9855501705259379,
      "grad_norm": 3.866144895553589,
      "learning_rate": 0.0,
      "loss": 0.7312,
      "step": 10981
    },
    {
      "epoch": 0.9856399210195657,
      "grad_norm": 2.922327756881714,
      "learning_rate": 0.0,
      "loss": 0.7248,
      "step": 10982
    },
    {
      "epoch": 0.9857296715131934,
      "grad_norm": 3.013906717300415,
      "learning_rate": 0.0,
      "loss": 0.8389,
      "step": 10983
    },
    {
      "epoch": 0.985819422006821,
      "grad_norm": 3.5205647945404053,
      "learning_rate": 0.0,
      "loss": 0.7046,
      "step": 10984
    },
    {
      "epoch": 0.9859091725004487,
      "grad_norm": 3.2268149852752686,
      "learning_rate": 0.0,
      "loss": 0.6785,
      "step": 10985
    },
    {
      "epoch": 0.9859989229940764,
      "grad_norm": 3.1119191646575928,
      "learning_rate": 0.0,
      "loss": 0.6874,
      "step": 10986
    },
    {
      "epoch": 0.9860886734877042,
      "grad_norm": 3.3874902725219727,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 10987
    },
    {
      "epoch": 0.9861784239813319,
      "grad_norm": 3.3660082817077637,
      "learning_rate": 0.0,
      "loss": 0.7355,
      "step": 10988
    },
    {
      "epoch": 0.9862681744749596,
      "grad_norm": 2.9799928665161133,
      "learning_rate": 0.0,
      "loss": 0.7164,
      "step": 10989
    },
    {
      "epoch": 0.9863579249685873,
      "grad_norm": 3.0417370796203613,
      "learning_rate": 0.0,
      "loss": 0.7187,
      "step": 10990
    },
    {
      "epoch": 0.986447675462215,
      "grad_norm": 4.280959606170654,
      "learning_rate": 0.0,
      "loss": 0.7455,
      "step": 10991
    },
    {
      "epoch": 0.9865374259558427,
      "grad_norm": 3.231684684753418,
      "learning_rate": 0.0,
      "loss": 0.6878,
      "step": 10992
    },
    {
      "epoch": 0.9866271764494705,
      "grad_norm": 3.010101795196533,
      "learning_rate": 0.0,
      "loss": 0.6447,
      "step": 10993
    },
    {
      "epoch": 0.9867169269430982,
      "grad_norm": 3.317952871322632,
      "learning_rate": 0.0,
      "loss": 0.7328,
      "step": 10994
    },
    {
      "epoch": 0.9868066774367259,
      "grad_norm": 2.9168479442596436,
      "learning_rate": 0.0,
      "loss": 0.6994,
      "step": 10995
    },
    {
      "epoch": 0.9868964279303536,
      "grad_norm": 2.4160914421081543,
      "learning_rate": 0.0,
      "loss": 0.7142,
      "step": 10996
    },
    {
      "epoch": 0.9869861784239813,
      "grad_norm": 3.0533201694488525,
      "learning_rate": 0.0,
      "loss": 0.705,
      "step": 10997
    },
    {
      "epoch": 0.9870759289176091,
      "grad_norm": 2.861661911010742,
      "learning_rate": 0.0,
      "loss": 0.6603,
      "step": 10998
    },
    {
      "epoch": 0.9871656794112368,
      "grad_norm": 2.762646198272705,
      "learning_rate": 0.0,
      "loss": 0.7052,
      "step": 10999
    },
    {
      "epoch": 0.9872554299048645,
      "grad_norm": 2.710179567337036,
      "learning_rate": 0.0,
      "loss": 0.7244,
      "step": 11000
    },
    {
      "epoch": 0.9873451803984922,
      "grad_norm": 3.8992841243743896,
      "learning_rate": 0.0,
      "loss": 0.6802,
      "step": 11001
    },
    {
      "epoch": 0.9874349308921199,
      "grad_norm": 3.495290517807007,
      "learning_rate": 0.0,
      "loss": 0.7348,
      "step": 11002
    },
    {
      "epoch": 0.9875246813857477,
      "grad_norm": 2.685689687728882,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 11003
    },
    {
      "epoch": 0.9876144318793754,
      "grad_norm": 2.9393086433410645,
      "learning_rate": 0.0,
      "loss": 0.6923,
      "step": 11004
    },
    {
      "epoch": 0.9877041823730031,
      "grad_norm": 4.11488151550293,
      "learning_rate": 0.0,
      "loss": 0.6986,
      "step": 11005
    },
    {
      "epoch": 0.9877939328666308,
      "grad_norm": 2.664952278137207,
      "learning_rate": 0.0,
      "loss": 0.671,
      "step": 11006
    },
    {
      "epoch": 0.9878836833602584,
      "grad_norm": 3.670818328857422,
      "learning_rate": 0.0,
      "loss": 0.6827,
      "step": 11007
    },
    {
      "epoch": 0.9879734338538863,
      "grad_norm": 3.011350393295288,
      "learning_rate": 0.0,
      "loss": 0.6905,
      "step": 11008
    },
    {
      "epoch": 0.988063184347514,
      "grad_norm": 3.3263847827911377,
      "learning_rate": 0.0,
      "loss": 0.6705,
      "step": 11009
    },
    {
      "epoch": 0.9881529348411416,
      "grad_norm": 2.501072406768799,
      "learning_rate": 0.0,
      "loss": 0.721,
      "step": 11010
    },
    {
      "epoch": 0.9882426853347693,
      "grad_norm": 2.6302881240844727,
      "learning_rate": 0.0,
      "loss": 0.7326,
      "step": 11011
    },
    {
      "epoch": 0.988332435828397,
      "grad_norm": 2.8750715255737305,
      "learning_rate": 0.0,
      "loss": 0.729,
      "step": 11012
    },
    {
      "epoch": 0.9884221863220247,
      "grad_norm": 3.599424362182617,
      "learning_rate": 0.0,
      "loss": 0.7259,
      "step": 11013
    },
    {
      "epoch": 0.9885119368156525,
      "grad_norm": 2.6016392707824707,
      "learning_rate": 0.0,
      "loss": 0.7242,
      "step": 11014
    },
    {
      "epoch": 0.9886016873092802,
      "grad_norm": 2.746126413345337,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 11015
    },
    {
      "epoch": 0.9886914378029079,
      "grad_norm": 2.6933915615081787,
      "learning_rate": 0.0,
      "loss": 0.6928,
      "step": 11016
    },
    {
      "epoch": 0.9887811882965356,
      "grad_norm": 6.201173782348633,
      "learning_rate": 0.0,
      "loss": 0.7363,
      "step": 11017
    },
    {
      "epoch": 0.9888709387901633,
      "grad_norm": 3.8715155124664307,
      "learning_rate": 0.0,
      "loss": 0.6851,
      "step": 11018
    },
    {
      "epoch": 0.9889606892837911,
      "grad_norm": 3.484553098678589,
      "learning_rate": 0.0,
      "loss": 0.692,
      "step": 11019
    },
    {
      "epoch": 0.9890504397774188,
      "grad_norm": 2.6193606853485107,
      "learning_rate": 0.0,
      "loss": 0.6966,
      "step": 11020
    },
    {
      "epoch": 0.9891401902710465,
      "grad_norm": 3.215958833694458,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 11021
    },
    {
      "epoch": 0.9892299407646742,
      "grad_norm": 5.158049583435059,
      "learning_rate": 0.0,
      "loss": 0.7033,
      "step": 11022
    },
    {
      "epoch": 0.9893196912583019,
      "grad_norm": 3.1851389408111572,
      "learning_rate": 0.0,
      "loss": 0.7019,
      "step": 11023
    },
    {
      "epoch": 0.9894094417519297,
      "grad_norm": 3.763613700866699,
      "learning_rate": 0.0,
      "loss": 0.7501,
      "step": 11024
    },
    {
      "epoch": 0.9894991922455574,
      "grad_norm": 2.8223061561584473,
      "learning_rate": 0.0,
      "loss": 0.6902,
      "step": 11025
    },
    {
      "epoch": 0.9895889427391851,
      "grad_norm": 2.814497470855713,
      "learning_rate": 0.0,
      "loss": 0.7117,
      "step": 11026
    },
    {
      "epoch": 0.9896786932328128,
      "grad_norm": 3.103095054626465,
      "learning_rate": 0.0,
      "loss": 0.6842,
      "step": 11027
    },
    {
      "epoch": 0.9897684437264405,
      "grad_norm": 3.1194558143615723,
      "learning_rate": 0.0,
      "loss": 0.6718,
      "step": 11028
    },
    {
      "epoch": 0.9898581942200682,
      "grad_norm": 2.490581512451172,
      "learning_rate": 0.0,
      "loss": 0.7434,
      "step": 11029
    },
    {
      "epoch": 0.989947944713696,
      "grad_norm": 2.752701997756958,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 11030
    },
    {
      "epoch": 0.9900376952073237,
      "grad_norm": 5.256373405456543,
      "learning_rate": 0.0,
      "loss": 0.6857,
      "step": 11031
    },
    {
      "epoch": 0.9901274457009513,
      "grad_norm": 3.098816394805908,
      "learning_rate": 0.0,
      "loss": 0.6996,
      "step": 11032
    },
    {
      "epoch": 0.990217196194579,
      "grad_norm": 2.97995924949646,
      "learning_rate": 0.0,
      "loss": 0.7292,
      "step": 11033
    },
    {
      "epoch": 0.9903069466882067,
      "grad_norm": 2.545510768890381,
      "learning_rate": 0.0,
      "loss": 0.727,
      "step": 11034
    },
    {
      "epoch": 0.9903966971818345,
      "grad_norm": 3.6872739791870117,
      "learning_rate": 0.0,
      "loss": 0.6924,
      "step": 11035
    },
    {
      "epoch": 0.9904864476754622,
      "grad_norm": 3.9641001224517822,
      "learning_rate": 0.0,
      "loss": 0.7497,
      "step": 11036
    },
    {
      "epoch": 0.9905761981690899,
      "grad_norm": 3.417851686477661,
      "learning_rate": 0.0,
      "loss": 0.6731,
      "step": 11037
    },
    {
      "epoch": 0.9906659486627176,
      "grad_norm": 2.708638906478882,
      "learning_rate": 0.0,
      "loss": 0.7427,
      "step": 11038
    },
    {
      "epoch": 0.9907556991563453,
      "grad_norm": 2.9727933406829834,
      "learning_rate": 0.0,
      "loss": 0.6742,
      "step": 11039
    },
    {
      "epoch": 0.9908454496499731,
      "grad_norm": 2.986424207687378,
      "learning_rate": 0.0,
      "loss": 0.7094,
      "step": 11040
    },
    {
      "epoch": 0.9909352001436008,
      "grad_norm": 2.836254835128784,
      "learning_rate": 0.0,
      "loss": 0.6954,
      "step": 11041
    },
    {
      "epoch": 0.9910249506372285,
      "grad_norm": 3.327712059020996,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 11042
    },
    {
      "epoch": 0.9911147011308562,
      "grad_norm": 2.6695311069488525,
      "learning_rate": 0.0,
      "loss": 0.702,
      "step": 11043
    },
    {
      "epoch": 0.9912044516244839,
      "grad_norm": 4.308676242828369,
      "learning_rate": 0.0,
      "loss": 0.724,
      "step": 11044
    },
    {
      "epoch": 0.9912942021181117,
      "grad_norm": 3.2477328777313232,
      "learning_rate": 0.0,
      "loss": 0.7043,
      "step": 11045
    },
    {
      "epoch": 0.9913839526117394,
      "grad_norm": 4.157419681549072,
      "learning_rate": 0.0,
      "loss": 0.6643,
      "step": 11046
    },
    {
      "epoch": 0.9914737031053671,
      "grad_norm": 3.362520694732666,
      "learning_rate": 0.0,
      "loss": 0.7082,
      "step": 11047
    },
    {
      "epoch": 0.9915634535989948,
      "grad_norm": 2.6246285438537598,
      "learning_rate": 0.0,
      "loss": 0.6847,
      "step": 11048
    },
    {
      "epoch": 0.9916532040926225,
      "grad_norm": 3.7384183406829834,
      "learning_rate": 0.0,
      "loss": 0.7569,
      "step": 11049
    },
    {
      "epoch": 0.9917429545862502,
      "grad_norm": 3.468219518661499,
      "learning_rate": 0.0,
      "loss": 0.6684,
      "step": 11050
    },
    {
      "epoch": 0.991832705079878,
      "grad_norm": 2.2280936241149902,
      "learning_rate": 0.0,
      "loss": 0.683,
      "step": 11051
    },
    {
      "epoch": 0.9919224555735057,
      "grad_norm": 2.69028902053833,
      "learning_rate": 0.0,
      "loss": 0.7073,
      "step": 11052
    },
    {
      "epoch": 0.9920122060671334,
      "grad_norm": 2.961538314819336,
      "learning_rate": 0.0,
      "loss": 0.6419,
      "step": 11053
    },
    {
      "epoch": 0.9921019565607611,
      "grad_norm": 2.5304315090179443,
      "learning_rate": 0.0,
      "loss": 0.731,
      "step": 11054
    },
    {
      "epoch": 0.9921917070543887,
      "grad_norm": 3.086193084716797,
      "learning_rate": 0.0,
      "loss": 0.7437,
      "step": 11055
    },
    {
      "epoch": 0.9922814575480166,
      "grad_norm": 2.9301366806030273,
      "learning_rate": 0.0,
      "loss": 0.6776,
      "step": 11056
    },
    {
      "epoch": 0.9923712080416442,
      "grad_norm": 2.544421911239624,
      "learning_rate": 0.0,
      "loss": 0.7191,
      "step": 11057
    },
    {
      "epoch": 0.9924609585352719,
      "grad_norm": 2.8917105197906494,
      "learning_rate": 0.0,
      "loss": 0.7224,
      "step": 11058
    },
    {
      "epoch": 0.9925507090288996,
      "grad_norm": 3.7494373321533203,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 11059
    },
    {
      "epoch": 0.9926404595225273,
      "grad_norm": 2.648857593536377,
      "learning_rate": 0.0,
      "loss": 0.7104,
      "step": 11060
    },
    {
      "epoch": 0.9927302100161551,
      "grad_norm": 3.2149507999420166,
      "learning_rate": 0.0,
      "loss": 0.6609,
      "step": 11061
    },
    {
      "epoch": 0.9928199605097828,
      "grad_norm": 2.933257579803467,
      "learning_rate": 0.0,
      "loss": 0.7777,
      "step": 11062
    },
    {
      "epoch": 0.9929097110034105,
      "grad_norm": 3.5056986808776855,
      "learning_rate": 0.0,
      "loss": 0.823,
      "step": 11063
    },
    {
      "epoch": 0.9929994614970382,
      "grad_norm": 2.7570548057556152,
      "learning_rate": 0.0,
      "loss": 0.7208,
      "step": 11064
    },
    {
      "epoch": 0.9930892119906659,
      "grad_norm": 5.612277984619141,
      "learning_rate": 0.0,
      "loss": 0.6809,
      "step": 11065
    },
    {
      "epoch": 0.9931789624842937,
      "grad_norm": 3.184615135192871,
      "learning_rate": 0.0,
      "loss": 0.6639,
      "step": 11066
    },
    {
      "epoch": 0.9932687129779214,
      "grad_norm": 3.829934597015381,
      "learning_rate": 0.0,
      "loss": 0.687,
      "step": 11067
    },
    {
      "epoch": 0.9933584634715491,
      "grad_norm": 3.118021011352539,
      "learning_rate": 0.0,
      "loss": 0.6899,
      "step": 11068
    },
    {
      "epoch": 0.9934482139651768,
      "grad_norm": 5.878944396972656,
      "learning_rate": 0.0,
      "loss": 0.6968,
      "step": 11069
    },
    {
      "epoch": 0.9935379644588045,
      "grad_norm": 4.355205535888672,
      "learning_rate": 0.0,
      "loss": 0.7349,
      "step": 11070
    },
    {
      "epoch": 0.9936277149524322,
      "grad_norm": 11.339991569519043,
      "learning_rate": 0.0,
      "loss": 0.7282,
      "step": 11071
    },
    {
      "epoch": 0.99371746544606,
      "grad_norm": 3.3326122760772705,
      "learning_rate": 0.0,
      "loss": 0.7314,
      "step": 11072
    },
    {
      "epoch": 0.9938072159396877,
      "grad_norm": 2.4477379322052,
      "learning_rate": 0.0,
      "loss": 0.7391,
      "step": 11073
    },
    {
      "epoch": 0.9938969664333154,
      "grad_norm": 6.2011799812316895,
      "learning_rate": 0.0,
      "loss": 0.6995,
      "step": 11074
    },
    {
      "epoch": 0.9939867169269431,
      "grad_norm": 4.189809799194336,
      "learning_rate": 0.0,
      "loss": 0.7121,
      "step": 11075
    },
    {
      "epoch": 0.9940764674205708,
      "grad_norm": 3.348905086517334,
      "learning_rate": 0.0,
      "loss": 0.756,
      "step": 11076
    },
    {
      "epoch": 0.9941662179141986,
      "grad_norm": 3.481325387954712,
      "learning_rate": 0.0,
      "loss": 0.6997,
      "step": 11077
    },
    {
      "epoch": 0.9942559684078263,
      "grad_norm": 4.534275054931641,
      "learning_rate": 0.0,
      "loss": 0.7316,
      "step": 11078
    },
    {
      "epoch": 0.994345718901454,
      "grad_norm": 3.8119029998779297,
      "learning_rate": 0.0,
      "loss": 0.6846,
      "step": 11079
    },
    {
      "epoch": 0.9944354693950817,
      "grad_norm": 2.7681610584259033,
      "learning_rate": 0.0,
      "loss": 0.6884,
      "step": 11080
    },
    {
      "epoch": 0.9945252198887093,
      "grad_norm": 2.6750423908233643,
      "learning_rate": 0.0,
      "loss": 0.7167,
      "step": 11081
    },
    {
      "epoch": 0.9946149703823371,
      "grad_norm": 3.2784016132354736,
      "learning_rate": 0.0,
      "loss": 0.6946,
      "step": 11082
    },
    {
      "epoch": 0.9947047208759648,
      "grad_norm": 3.078824281692505,
      "learning_rate": 0.0,
      "loss": 0.7217,
      "step": 11083
    },
    {
      "epoch": 0.9947944713695925,
      "grad_norm": 4.953436374664307,
      "learning_rate": 0.0,
      "loss": 0.6331,
      "step": 11084
    },
    {
      "epoch": 0.9948842218632202,
      "grad_norm": 2.836435556411743,
      "learning_rate": 0.0,
      "loss": 0.7794,
      "step": 11085
    },
    {
      "epoch": 0.9949739723568479,
      "grad_norm": 3.001528739929199,
      "learning_rate": 0.0,
      "loss": 0.6563,
      "step": 11086
    },
    {
      "epoch": 0.9950637228504757,
      "grad_norm": 4.107137203216553,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 11087
    },
    {
      "epoch": 0.9951534733441034,
      "grad_norm": 3.052675724029541,
      "learning_rate": 0.0,
      "loss": 0.7119,
      "step": 11088
    },
    {
      "epoch": 0.9952432238377311,
      "grad_norm": 3.0304722785949707,
      "learning_rate": 0.0,
      "loss": 0.7303,
      "step": 11089
    },
    {
      "epoch": 0.9953329743313588,
      "grad_norm": 2.9448747634887695,
      "learning_rate": 0.0,
      "loss": 0.707,
      "step": 11090
    },
    {
      "epoch": 0.9954227248249865,
      "grad_norm": 3.639298915863037,
      "learning_rate": 0.0,
      "loss": 0.7129,
      "step": 11091
    },
    {
      "epoch": 0.9955124753186142,
      "grad_norm": 3.3887014389038086,
      "learning_rate": 0.0,
      "loss": 0.7059,
      "step": 11092
    },
    {
      "epoch": 0.995602225812242,
      "grad_norm": 6.1867451667785645,
      "learning_rate": 0.0,
      "loss": 0.718,
      "step": 11093
    },
    {
      "epoch": 0.9956919763058697,
      "grad_norm": 2.7762603759765625,
      "learning_rate": 0.0,
      "loss": 0.7249,
      "step": 11094
    },
    {
      "epoch": 0.9957817267994974,
      "grad_norm": 11.517725944519043,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 11095
    },
    {
      "epoch": 0.9958714772931251,
      "grad_norm": 73.63030242919922,
      "learning_rate": 0.0,
      "loss": 0.6709,
      "step": 11096
    },
    {
      "epoch": 0.9959612277867528,
      "grad_norm": 5.871732711791992,
      "learning_rate": 0.0,
      "loss": 0.6861,
      "step": 11097
    },
    {
      "epoch": 0.9960509782803806,
      "grad_norm": 4.473134994506836,
      "learning_rate": 0.0,
      "loss": 0.6528,
      "step": 11098
    },
    {
      "epoch": 0.9961407287740083,
      "grad_norm": 3.302481174468994,
      "learning_rate": 0.0,
      "loss": 0.7201,
      "step": 11099
    },
    {
      "epoch": 0.996230479267636,
      "grad_norm": 3.6003122329711914,
      "learning_rate": 0.0,
      "loss": 0.6642,
      "step": 11100
    },
    {
      "epoch": 0.9963202297612637,
      "grad_norm": 3.5367043018341064,
      "learning_rate": 0.0,
      "loss": 0.7857,
      "step": 11101
    },
    {
      "epoch": 0.9964099802548914,
      "grad_norm": 4.358185291290283,
      "learning_rate": 0.0,
      "loss": 0.6958,
      "step": 11102
    },
    {
      "epoch": 0.9964997307485192,
      "grad_norm": 3.4661717414855957,
      "learning_rate": 0.0,
      "loss": 0.6698,
      "step": 11103
    },
    {
      "epoch": 0.9965894812421469,
      "grad_norm": 2.663438081741333,
      "learning_rate": 0.0,
      "loss": 0.7195,
      "step": 11104
    },
    {
      "epoch": 0.9966792317357746,
      "grad_norm": 3.7451870441436768,
      "learning_rate": 0.0,
      "loss": 0.668,
      "step": 11105
    },
    {
      "epoch": 0.9967689822294022,
      "grad_norm": 3.5944626331329346,
      "learning_rate": 0.0,
      "loss": 0.6753,
      "step": 11106
    },
    {
      "epoch": 0.9968587327230299,
      "grad_norm": 4.864634990692139,
      "learning_rate": 0.0,
      "loss": 0.7424,
      "step": 11107
    },
    {
      "epoch": 0.9969484832166577,
      "grad_norm": 3.3040175437927246,
      "learning_rate": 0.0,
      "loss": 0.698,
      "step": 11108
    },
    {
      "epoch": 0.9970382337102854,
      "grad_norm": 3.220135450363159,
      "learning_rate": 0.0,
      "loss": 0.706,
      "step": 11109
    },
    {
      "epoch": 0.9971279842039131,
      "grad_norm": 3.0044164657592773,
      "learning_rate": 0.0,
      "loss": 0.7081,
      "step": 11110
    },
    {
      "epoch": 0.9972177346975408,
      "grad_norm": 3.2474067211151123,
      "learning_rate": 0.0,
      "loss": 0.7138,
      "step": 11111
    },
    {
      "epoch": 0.9973074851911685,
      "grad_norm": 2.868734359741211,
      "learning_rate": 0.0,
      "loss": 0.7155,
      "step": 11112
    },
    {
      "epoch": 0.9973972356847962,
      "grad_norm": 2.725642442703247,
      "learning_rate": 0.0,
      "loss": 0.6985,
      "step": 11113
    },
    {
      "epoch": 0.997486986178424,
      "grad_norm": 3.7644288539886475,
      "learning_rate": 0.0,
      "loss": 0.6743,
      "step": 11114
    },
    {
      "epoch": 0.9975767366720517,
      "grad_norm": 3.936253309249878,
      "learning_rate": 0.0,
      "loss": 0.7563,
      "step": 11115
    },
    {
      "epoch": 0.9976664871656794,
      "grad_norm": 2.4458541870117188,
      "learning_rate": 0.0,
      "loss": 0.7092,
      "step": 11116
    },
    {
      "epoch": 0.9977562376593071,
      "grad_norm": 2.9910550117492676,
      "learning_rate": 0.0,
      "loss": 0.71,
      "step": 11117
    },
    {
      "epoch": 0.9978459881529348,
      "grad_norm": 3.594881057739258,
      "learning_rate": 0.0,
      "loss": 0.8323,
      "step": 11118
    },
    {
      "epoch": 0.9979357386465626,
      "grad_norm": 25.426456451416016,
      "learning_rate": 0.0,
      "loss": 0.6962,
      "step": 11119
    },
    {
      "epoch": 0.9980254891401903,
      "grad_norm": 3.508378028869629,
      "learning_rate": 0.0,
      "loss": 0.7308,
      "step": 11120
    },
    {
      "epoch": 0.998115239633818,
      "grad_norm": 3.7835628986358643,
      "learning_rate": 0.0,
      "loss": 0.6701,
      "step": 11121
    },
    {
      "epoch": 0.9982049901274457,
      "grad_norm": 2.6581099033355713,
      "learning_rate": 0.0,
      "loss": 0.6939,
      "step": 11122
    },
    {
      "epoch": 0.9982947406210734,
      "grad_norm": 4.332117080688477,
      "learning_rate": 0.0,
      "loss": 0.6604,
      "step": 11123
    },
    {
      "epoch": 0.9983844911147012,
      "grad_norm": 2.5690319538116455,
      "learning_rate": 0.0,
      "loss": 0.7198,
      "step": 11124
    },
    {
      "epoch": 0.9984742416083289,
      "grad_norm": 2.866208791732788,
      "learning_rate": 0.0,
      "loss": 0.7214,
      "step": 11125
    },
    {
      "epoch": 0.9985639921019566,
      "grad_norm": 5.840159893035889,
      "learning_rate": 0.0,
      "loss": 0.7251,
      "step": 11126
    },
    {
      "epoch": 0.9986537425955843,
      "grad_norm": 2.9880893230438232,
      "learning_rate": 0.0,
      "loss": 0.7305,
      "step": 11127
    },
    {
      "epoch": 0.998743493089212,
      "grad_norm": 2.8327102661132812,
      "learning_rate": 0.0,
      "loss": 0.6794,
      "step": 11128
    },
    {
      "epoch": 0.9988332435828398,
      "grad_norm": 3.708477735519409,
      "learning_rate": 0.0,
      "loss": 0.7338,
      "step": 11129
    },
    {
      "epoch": 0.9989229940764675,
      "grad_norm": 3.4198496341705322,
      "learning_rate": 0.0,
      "loss": 0.6454,
      "step": 11130
    },
    {
      "epoch": 0.9990127445700951,
      "grad_norm": 3.3282394409179688,
      "learning_rate": 0.0,
      "loss": 0.6935,
      "step": 11131
    },
    {
      "epoch": 0.9991024950637228,
      "grad_norm": 2.678849458694458,
      "learning_rate": 0.0,
      "loss": 0.6747,
      "step": 11132
    },
    {
      "epoch": 0.9991922455573505,
      "grad_norm": 3.1693596839904785,
      "learning_rate": 0.0,
      "loss": 0.6912,
      "step": 11133
    },
    {
      "epoch": 0.9992819960509782,
      "grad_norm": 6.701335430145264,
      "learning_rate": 0.0,
      "loss": 0.6726,
      "step": 11134
    },
    {
      "epoch": 0.999371746544606,
      "grad_norm": 3.177394151687622,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 11135
    },
    {
      "epoch": 0.9994614970382337,
      "grad_norm": 3.303953170776367,
      "learning_rate": 0.0,
      "loss": 0.7018,
      "step": 11136
    },
    {
      "epoch": 0.9995512475318614,
      "grad_norm": 3.209127426147461,
      "learning_rate": 0.0,
      "loss": 0.7318,
      "step": 11137
    },
    {
      "epoch": 0.9996409980254891,
      "grad_norm": 3.0383589267730713,
      "learning_rate": 0.0,
      "loss": 0.7028,
      "step": 11138
    },
    {
      "epoch": 0.9997307485191168,
      "grad_norm": 7.755995273590088,
      "learning_rate": 0.0,
      "loss": 0.7521,
      "step": 11139
    },
    {
      "epoch": 0.9998204990127446,
      "grad_norm": 3.575122594833374,
      "learning_rate": 0.0,
      "loss": 0.7347,
      "step": 11140
    },
    {
      "epoch": 0.9999102495063723,
      "grad_norm": 3.7643165588378906,
      "learning_rate": 0.0,
      "loss": 0.7491,
      "step": 11141
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.99886417388916,
      "learning_rate": 0.0,
      "loss": 0.7114,
      "step": 11142
    },
    {
      "epoch": 1.0,
      "step": 11142,
      "total_flos": 9.926611843342244e+19,
      "train_loss": 0.2848037878137327,
      "train_runtime": 77341.0073,
      "train_samples_per_second": 18.439,
      "train_steps_per_second": 0.144
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 11142,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 9.926611843342244e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}