{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 24770,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 4.037141703673799e-05,
      "grad_norm": 351.98150634765625,
      "learning_rate": 2.688172043010753e-08,
      "loss": 11.4271,
      "step": 1
    },
    {
      "epoch": 8.074283407347598e-05,
      "grad_norm": 351.7983093261719,
      "learning_rate": 5.376344086021506e-08,
      "loss": 12.6027,
      "step": 2
    },
    {
      "epoch": 0.00012111425111021397,
      "grad_norm": 314.4841613769531,
      "learning_rate": 8.064516129032259e-08,
      "loss": 11.6462,
      "step": 3
    },
    {
      "epoch": 0.00016148566814695197,
      "grad_norm": 265.8146667480469,
      "learning_rate": 1.0752688172043012e-07,
      "loss": 11.3881,
      "step": 4
    },
    {
      "epoch": 0.00020185708518368994,
      "grad_norm": 312.5400085449219,
      "learning_rate": 1.3440860215053764e-07,
      "loss": 11.5488,
      "step": 5
    },
    {
      "epoch": 0.00024222850222042794,
      "grad_norm": 425.1934509277344,
      "learning_rate": 1.6129032258064518e-07,
      "loss": 13.1081,
      "step": 6
    },
    {
      "epoch": 0.0002825999192571659,
      "grad_norm": 336.73419189453125,
      "learning_rate": 1.881720430107527e-07,
      "loss": 10.9735,
      "step": 7
    },
    {
      "epoch": 0.00032297133629390393,
      "grad_norm": 372.12945556640625,
      "learning_rate": 2.1505376344086024e-07,
      "loss": 11.1451,
      "step": 8
    },
    {
      "epoch": 0.0003633427533306419,
      "grad_norm": 240.36685180664062,
      "learning_rate": 2.4193548387096775e-07,
      "loss": 10.187,
      "step": 9
    },
    {
      "epoch": 0.0004037141703673799,
      "grad_norm": 270.36676025390625,
      "learning_rate": 2.688172043010753e-07,
      "loss": 10.2259,
      "step": 10
    },
    {
      "epoch": 0.0004440855874041179,
      "grad_norm": 354.73040771484375,
      "learning_rate": 2.956989247311828e-07,
      "loss": 10.719,
      "step": 11
    },
    {
      "epoch": 0.00048445700444085587,
      "grad_norm": 303.74432373046875,
      "learning_rate": 3.2258064516129035e-07,
      "loss": 10.7076,
      "step": 12
    },
    {
      "epoch": 0.0005248284214775939,
      "grad_norm": 294.79107666015625,
      "learning_rate": 3.4946236559139783e-07,
      "loss": 11.1963,
      "step": 13
    },
    {
      "epoch": 0.0005651998385143318,
      "grad_norm": 360.65716552734375,
      "learning_rate": 3.763440860215054e-07,
      "loss": 11.0795,
      "step": 14
    },
    {
      "epoch": 0.0006055712555510698,
      "grad_norm": 903.30712890625,
      "learning_rate": 4.032258064516129e-07,
      "loss": 10.5469,
      "step": 15
    },
    {
      "epoch": 0.0006459426725878079,
      "grad_norm": 500.467529296875,
      "learning_rate": 4.301075268817205e-07,
      "loss": 11.21,
      "step": 16
    },
    {
      "epoch": 0.0006863140896245458,
      "grad_norm": 221.2133331298828,
      "learning_rate": 4.56989247311828e-07,
      "loss": 10.5027,
      "step": 17
    },
    {
      "epoch": 0.0007266855066612838,
      "grad_norm": 417.05712890625,
      "learning_rate": 4.838709677419355e-07,
      "loss": 10.7038,
      "step": 18
    },
    {
      "epoch": 0.0007670569236980218,
      "grad_norm": 311.0486145019531,
      "learning_rate": 5.10752688172043e-07,
      "loss": 11.3367,
      "step": 19
    },
    {
      "epoch": 0.0008074283407347598,
      "grad_norm": 241.86407470703125,
      "learning_rate": 5.376344086021506e-07,
      "loss": 10.3079,
      "step": 20
    },
    {
      "epoch": 0.0008477997577714978,
      "grad_norm": 333.48828125,
      "learning_rate": 5.645161290322581e-07,
      "loss": 10.1543,
      "step": 21
    },
    {
      "epoch": 0.0008881711748082358,
      "grad_norm": 289.5560607910156,
      "learning_rate": 5.913978494623656e-07,
      "loss": 11.2231,
      "step": 22
    },
    {
      "epoch": 0.0009285425918449737,
      "grad_norm": 330.8139343261719,
      "learning_rate": 6.182795698924732e-07,
      "loss": 11.415,
      "step": 23
    },
    {
      "epoch": 0.0009689140088817117,
      "grad_norm": 295.14593505859375,
      "learning_rate": 6.451612903225807e-07,
      "loss": 10.8893,
      "step": 24
    },
    {
      "epoch": 0.0010092854259184498,
      "grad_norm": 285.7171936035156,
      "learning_rate": 6.720430107526882e-07,
      "loss": 10.3145,
      "step": 25
    },
    {
      "epoch": 0.0010496568429551878,
      "grad_norm": 284.4788513183594,
      "learning_rate": 6.989247311827957e-07,
      "loss": 10.2806,
      "step": 26
    },
    {
      "epoch": 0.0010900282599919258,
      "grad_norm": 295.0856018066406,
      "learning_rate": 7.258064516129033e-07,
      "loss": 10.4483,
      "step": 27
    },
    {
      "epoch": 0.0011303996770286636,
      "grad_norm": 239.06414794921875,
      "learning_rate": 7.526881720430108e-07,
      "loss": 9.4905,
      "step": 28
    },
    {
      "epoch": 0.0011707710940654017,
      "grad_norm": 268.61419677734375,
      "learning_rate": 7.795698924731184e-07,
      "loss": 10.0993,
      "step": 29
    },
    {
      "epoch": 0.0012111425111021397,
      "grad_norm": 319.35833740234375,
      "learning_rate": 8.064516129032258e-07,
      "loss": 10.8902,
      "step": 30
    },
    {
      "epoch": 0.0012515139281388777,
      "grad_norm": 245.96722412109375,
      "learning_rate": 8.333333333333333e-07,
      "loss": 9.9205,
      "step": 31
    },
    {
      "epoch": 0.0012918853451756157,
      "grad_norm": 299.4979248046875,
      "learning_rate": 8.60215053763441e-07,
      "loss": 10.0671,
      "step": 32
    },
    {
      "epoch": 0.0013322567622123538,
      "grad_norm": 249.27374267578125,
      "learning_rate": 8.870967741935485e-07,
      "loss": 9.1601,
      "step": 33
    },
    {
      "epoch": 0.0013726281792490916,
      "grad_norm": 241.42323303222656,
      "learning_rate": 9.13978494623656e-07,
      "loss": 8.6125,
      "step": 34
    },
    {
      "epoch": 0.0014129995962858296,
      "grad_norm": 251.71737670898438,
      "learning_rate": 9.408602150537635e-07,
      "loss": 9.5074,
      "step": 35
    },
    {
      "epoch": 0.0014533710133225676,
      "grad_norm": 256.9002685546875,
      "learning_rate": 9.67741935483871e-07,
      "loss": 8.9969,
      "step": 36
    },
    {
      "epoch": 0.0014937424303593056,
      "grad_norm": 276.7106628417969,
      "learning_rate": 9.946236559139785e-07,
      "loss": 9.2622,
      "step": 37
    },
    {
      "epoch": 0.0015341138473960437,
      "grad_norm": 213.2654266357422,
      "learning_rate": 1.021505376344086e-06,
      "loss": 8.1055,
      "step": 38
    },
    {
      "epoch": 0.0015744852644327817,
      "grad_norm": 165.10031127929688,
      "learning_rate": 1.0483870967741936e-06,
      "loss": 8.8639,
      "step": 39
    },
    {
      "epoch": 0.0016148566814695195,
      "grad_norm": 149.60536193847656,
      "learning_rate": 1.0752688172043011e-06,
      "loss": 8.2323,
      "step": 40
    },
    {
      "epoch": 0.0016552280985062575,
      "grad_norm": 223.98037719726562,
      "learning_rate": 1.1021505376344087e-06,
      "loss": 8.2023,
      "step": 41
    },
    {
      "epoch": 0.0016955995155429956,
      "grad_norm": 201.1020050048828,
      "learning_rate": 1.1290322580645162e-06,
      "loss": 7.2902,
      "step": 42
    },
    {
      "epoch": 0.0017359709325797336,
      "grad_norm": 156.13656616210938,
      "learning_rate": 1.1559139784946237e-06,
      "loss": 8.573,
      "step": 43
    },
    {
      "epoch": 0.0017763423496164716,
      "grad_norm": 191.32383728027344,
      "learning_rate": 1.1827956989247313e-06,
      "loss": 7.3028,
      "step": 44
    },
    {
      "epoch": 0.0018167137666532096,
      "grad_norm": 139.12530517578125,
      "learning_rate": 1.2096774193548388e-06,
      "loss": 8.3155,
      "step": 45
    },
    {
      "epoch": 0.0018570851836899474,
      "grad_norm": 170.83656311035156,
      "learning_rate": 1.2365591397849463e-06,
      "loss": 7.2813,
      "step": 46
    },
    {
      "epoch": 0.0018974566007266855,
      "grad_norm": 188.64096069335938,
      "learning_rate": 1.2634408602150539e-06,
      "loss": 8.9729,
      "step": 47
    },
    {
      "epoch": 0.0019378280177634235,
      "grad_norm": 146.08267211914062,
      "learning_rate": 1.2903225806451614e-06,
      "loss": 7.3958,
      "step": 48
    },
    {
      "epoch": 0.0019781994348001615,
      "grad_norm": 157.48838806152344,
      "learning_rate": 1.317204301075269e-06,
      "loss": 8.2855,
      "step": 49
    },
    {
      "epoch": 0.0020185708518368995,
      "grad_norm": 441.96588134765625,
      "learning_rate": 1.3440860215053765e-06,
      "loss": 7.7489,
      "step": 50
    },
    {
      "epoch": 0.0020589422688736376,
      "grad_norm": 165.79583740234375,
      "learning_rate": 1.3709677419354838e-06,
      "loss": 7.637,
      "step": 51
    },
    {
      "epoch": 0.0020993136859103756,
      "grad_norm": 185.55221557617188,
      "learning_rate": 1.3978494623655913e-06,
      "loss": 8.0051,
      "step": 52
    },
    {
      "epoch": 0.0021396851029471136,
      "grad_norm": 159.0415802001953,
      "learning_rate": 1.424731182795699e-06,
      "loss": 7.2715,
      "step": 53
    },
    {
      "epoch": 0.0021800565199838516,
      "grad_norm": 167.03736877441406,
      "learning_rate": 1.4516129032258066e-06,
      "loss": 7.5464,
      "step": 54
    },
    {
      "epoch": 0.0022204279370205892,
      "grad_norm": 173.00440979003906,
      "learning_rate": 1.4784946236559141e-06,
      "loss": 7.8403,
      "step": 55
    },
    {
      "epoch": 0.0022607993540573273,
      "grad_norm": 208.09014892578125,
      "learning_rate": 1.5053763440860217e-06,
      "loss": 7.0977,
      "step": 56
    },
    {
      "epoch": 0.0023011707710940653,
      "grad_norm": 173.131103515625,
      "learning_rate": 1.5322580645161292e-06,
      "loss": 6.8318,
      "step": 57
    },
    {
      "epoch": 0.0023415421881308033,
      "grad_norm": 616.3580322265625,
      "learning_rate": 1.5591397849462367e-06,
      "loss": 8.2545,
      "step": 58
    },
    {
      "epoch": 0.0023819136051675413,
      "grad_norm": 179.23451232910156,
      "learning_rate": 1.5860215053763443e-06,
      "loss": 7.4013,
      "step": 59
    },
    {
      "epoch": 0.0024222850222042794,
      "grad_norm": 129.75352478027344,
      "learning_rate": 1.6129032258064516e-06,
      "loss": 7.7119,
      "step": 60
    },
    {
      "epoch": 0.0024626564392410174,
      "grad_norm": 163.1778106689453,
      "learning_rate": 1.6397849462365591e-06,
      "loss": 7.2903,
      "step": 61
    },
    {
      "epoch": 0.0025030278562777554,
      "grad_norm": 183.93959045410156,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 7.1084,
      "step": 62
    },
    {
      "epoch": 0.0025433992733144934,
      "grad_norm": 302.5379638671875,
      "learning_rate": 1.6935483870967742e-06,
      "loss": 7.8183,
      "step": 63
    },
    {
      "epoch": 0.0025837706903512315,
      "grad_norm": 124.38919830322266,
      "learning_rate": 1.720430107526882e-06,
      "loss": 7.5037,
      "step": 64
    },
    {
      "epoch": 0.0026241421073879695,
      "grad_norm": 190.73336791992188,
      "learning_rate": 1.7473118279569895e-06,
      "loss": 7.3413,
      "step": 65
    },
    {
      "epoch": 0.0026645135244247075,
      "grad_norm": 197.9076385498047,
      "learning_rate": 1.774193548387097e-06,
      "loss": 6.6668,
      "step": 66
    },
    {
      "epoch": 0.002704884941461445,
      "grad_norm": 155.71371459960938,
      "learning_rate": 1.8010752688172045e-06,
      "loss": 7.0829,
      "step": 67
    },
    {
      "epoch": 0.002745256358498183,
      "grad_norm": 182.17523193359375,
      "learning_rate": 1.827956989247312e-06,
      "loss": 6.3079,
      "step": 68
    },
    {
      "epoch": 0.002785627775534921,
      "grad_norm": 300.212646484375,
      "learning_rate": 1.8548387096774196e-06,
      "loss": 5.9089,
      "step": 69
    },
    {
      "epoch": 0.002825999192571659,
      "grad_norm": 186.59921264648438,
      "learning_rate": 1.881720430107527e-06,
      "loss": 8.2888,
      "step": 70
    },
    {
      "epoch": 0.002866370609608397,
      "grad_norm": 158.3273468017578,
      "learning_rate": 1.9086021505376345e-06,
      "loss": 6.7978,
      "step": 71
    },
    {
      "epoch": 0.0029067420266451352,
      "grad_norm": 179.4152069091797,
      "learning_rate": 1.935483870967742e-06,
      "loss": 7.144,
      "step": 72
    },
    {
      "epoch": 0.0029471134436818733,
      "grad_norm": 237.13539123535156,
      "learning_rate": 1.9623655913978495e-06,
      "loss": 6.7827,
      "step": 73
    },
    {
      "epoch": 0.0029874848607186113,
      "grad_norm": 180.97222900390625,
      "learning_rate": 1.989247311827957e-06,
      "loss": 7.1172,
      "step": 74
    },
    {
      "epoch": 0.0030278562777553493,
      "grad_norm": 136.84031677246094,
      "learning_rate": 2.0161290322580646e-06,
      "loss": 6.7088,
      "step": 75
    },
    {
      "epoch": 0.0030682276947920873,
      "grad_norm": 188.46627807617188,
      "learning_rate": 2.043010752688172e-06,
      "loss": 6.2315,
      "step": 76
    },
    {
      "epoch": 0.0031085991118288254,
      "grad_norm": 185.7891845703125,
      "learning_rate": 2.0698924731182797e-06,
      "loss": 6.2583,
      "step": 77
    },
    {
      "epoch": 0.0031489705288655634,
      "grad_norm": 242.66665649414062,
      "learning_rate": 2.096774193548387e-06,
      "loss": 5.5965,
      "step": 78
    },
    {
      "epoch": 0.003189341945902301,
      "grad_norm": 148.83511352539062,
      "learning_rate": 2.1236559139784947e-06,
      "loss": 6.6791,
      "step": 79
    },
    {
      "epoch": 0.003229713362939039,
      "grad_norm": 131.77056884765625,
      "learning_rate": 2.1505376344086023e-06,
      "loss": 7.2191,
      "step": 80
    },
    {
      "epoch": 0.003270084779975777,
      "grad_norm": 243.5615234375,
      "learning_rate": 2.17741935483871e-06,
      "loss": 6.0999,
      "step": 81
    },
    {
      "epoch": 0.003310456197012515,
      "grad_norm": 330.57275390625,
      "learning_rate": 2.2043010752688173e-06,
      "loss": 5.5384,
      "step": 82
    },
    {
      "epoch": 0.003350827614049253,
      "grad_norm": 215.4854736328125,
      "learning_rate": 2.231182795698925e-06,
      "loss": 6.4691,
      "step": 83
    },
    {
      "epoch": 0.003391199031085991,
      "grad_norm": 141.75936889648438,
      "learning_rate": 2.2580645161290324e-06,
      "loss": 7.9281,
      "step": 84
    },
    {
      "epoch": 0.003431570448122729,
      "grad_norm": 299.9782409667969,
      "learning_rate": 2.28494623655914e-06,
      "loss": 6.503,
      "step": 85
    },
    {
      "epoch": 0.003471941865159467,
      "grad_norm": 171.2203369140625,
      "learning_rate": 2.3118279569892475e-06,
      "loss": 6.7308,
      "step": 86
    },
    {
      "epoch": 0.003512313282196205,
      "grad_norm": 150.8365936279297,
      "learning_rate": 2.338709677419355e-06,
      "loss": 7.2021,
      "step": 87
    },
    {
      "epoch": 0.003552684699232943,
      "grad_norm": 170.68563842773438,
      "learning_rate": 2.3655913978494625e-06,
      "loss": 6.3912,
      "step": 88
    },
    {
      "epoch": 0.0035930561162696812,
      "grad_norm": 165.5965118408203,
      "learning_rate": 2.39247311827957e-06,
      "loss": 6.4361,
      "step": 89
    },
    {
      "epoch": 0.0036334275333064193,
      "grad_norm": 205.67288208007812,
      "learning_rate": 2.4193548387096776e-06,
      "loss": 6.982,
      "step": 90
    },
    {
      "epoch": 0.003673798950343157,
      "grad_norm": 158.5758819580078,
      "learning_rate": 2.446236559139785e-06,
      "loss": 7.3175,
      "step": 91
    },
    {
      "epoch": 0.003714170367379895,
      "grad_norm": 146.76710510253906,
      "learning_rate": 2.4731182795698927e-06,
      "loss": 7.2685,
      "step": 92
    },
    {
      "epoch": 0.003754541784416633,
      "grad_norm": 284.1736145019531,
      "learning_rate": 2.5e-06,
      "loss": 7.1805,
      "step": 93
    },
    {
      "epoch": 0.003794913201453371,
      "grad_norm": 163.28565979003906,
      "learning_rate": 2.5268817204301077e-06,
      "loss": 5.9885,
      "step": 94
    },
    {
      "epoch": 0.003835284618490109,
      "grad_norm": 184.0115509033203,
      "learning_rate": 2.5537634408602153e-06,
      "loss": 6.9465,
      "step": 95
    },
    {
      "epoch": 0.003875656035526847,
      "grad_norm": 156.06378173828125,
      "learning_rate": 2.580645161290323e-06,
      "loss": 6.7838,
      "step": 96
    },
    {
      "epoch": 0.003916027452563585,
      "grad_norm": 201.0474853515625,
      "learning_rate": 2.6075268817204303e-06,
      "loss": 6.7712,
      "step": 97
    },
    {
      "epoch": 0.003956398869600323,
      "grad_norm": 178.62998962402344,
      "learning_rate": 2.634408602150538e-06,
      "loss": 6.5551,
      "step": 98
    },
    {
      "epoch": 0.003996770286637061,
      "grad_norm": 261.0061950683594,
      "learning_rate": 2.6612903225806454e-06,
      "loss": 6.0628,
      "step": 99
    },
    {
      "epoch": 0.004037141703673799,
      "grad_norm": 317.6354675292969,
      "learning_rate": 2.688172043010753e-06,
      "loss": 6.3592,
      "step": 100
    },
    {
      "epoch": 0.004077513120710537,
      "grad_norm": 309.9297180175781,
      "learning_rate": 2.71505376344086e-06,
      "loss": 5.6725,
      "step": 101
    },
    {
      "epoch": 0.004117884537747275,
      "grad_norm": 184.3683319091797,
      "learning_rate": 2.7419354838709676e-06,
      "loss": 7.7713,
      "step": 102
    },
    {
      "epoch": 0.004158255954784013,
      "grad_norm": 199.91000366210938,
      "learning_rate": 2.768817204301075e-06,
      "loss": 5.6121,
      "step": 103
    },
    {
      "epoch": 0.004198627371820751,
      "grad_norm": 166.21852111816406,
      "learning_rate": 2.7956989247311827e-06,
      "loss": 5.9422,
      "step": 104
    },
    {
      "epoch": 0.004238998788857489,
      "grad_norm": 179.0415802001953,
      "learning_rate": 2.822580645161291e-06,
      "loss": 5.4483,
      "step": 105
    },
    {
      "epoch": 0.004279370205894227,
      "grad_norm": 159.86790466308594,
      "learning_rate": 2.849462365591398e-06,
      "loss": 6.6918,
      "step": 106
    },
    {
      "epoch": 0.004319741622930965,
      "grad_norm": 213.2036590576172,
      "learning_rate": 2.8763440860215057e-06,
      "loss": 6.6955,
      "step": 107
    },
    {
      "epoch": 0.004360113039967703,
      "grad_norm": 217.697509765625,
      "learning_rate": 2.903225806451613e-06,
      "loss": 7.4238,
      "step": 108
    },
    {
      "epoch": 0.004400484457004441,
      "grad_norm": 191.98114013671875,
      "learning_rate": 2.9301075268817207e-06,
      "loss": 6.4298,
      "step": 109
    },
    {
      "epoch": 0.0044408558740411785,
      "grad_norm": 175.02099609375,
      "learning_rate": 2.9569892473118283e-06,
      "loss": 7.0272,
      "step": 110
    },
    {
      "epoch": 0.004481227291077917,
      "grad_norm": 168.18902587890625,
      "learning_rate": 2.983870967741936e-06,
      "loss": 6.6223,
      "step": 111
    },
    {
      "epoch": 0.0045215987081146545,
      "grad_norm": 167.71397399902344,
      "learning_rate": 3.0107526881720433e-06,
      "loss": 6.4587,
      "step": 112
    },
    {
      "epoch": 0.004561970125151393,
      "grad_norm": 242.6991424560547,
      "learning_rate": 3.037634408602151e-06,
      "loss": 5.3628,
      "step": 113
    },
    {
      "epoch": 0.004602341542188131,
      "grad_norm": 200.1949462890625,
      "learning_rate": 3.0645161290322584e-06,
      "loss": 5.6775,
      "step": 114
    },
    {
      "epoch": 0.004642712959224869,
      "grad_norm": 152.91062927246094,
      "learning_rate": 3.091397849462366e-06,
      "loss": 6.5483,
      "step": 115
    },
    {
      "epoch": 0.004683084376261607,
      "grad_norm": 178.0878143310547,
      "learning_rate": 3.1182795698924735e-06,
      "loss": 6.6829,
      "step": 116
    },
    {
      "epoch": 0.004723455793298345,
      "grad_norm": 161.53042602539062,
      "learning_rate": 3.145161290322581e-06,
      "loss": 7.751,
      "step": 117
    },
    {
      "epoch": 0.004763827210335083,
      "grad_norm": 205.35055541992188,
      "learning_rate": 3.1720430107526885e-06,
      "loss": 6.052,
      "step": 118
    },
    {
      "epoch": 0.004804198627371821,
      "grad_norm": 211.0558319091797,
      "learning_rate": 3.198924731182796e-06,
      "loss": 7.6129,
      "step": 119
    },
    {
      "epoch": 0.004844570044408559,
      "grad_norm": 210.41172790527344,
      "learning_rate": 3.225806451612903e-06,
      "loss": 4.5849,
      "step": 120
    },
    {
      "epoch": 0.004884941461445296,
      "grad_norm": 183.26947021484375,
      "learning_rate": 3.2526881720430107e-06,
      "loss": 5.6567,
      "step": 121
    },
    {
      "epoch": 0.004925312878482035,
      "grad_norm": 173.23068237304688,
      "learning_rate": 3.2795698924731183e-06,
      "loss": 5.4668,
      "step": 122
    },
    {
      "epoch": 0.004965684295518772,
      "grad_norm": 238.0470733642578,
      "learning_rate": 3.306451612903226e-06,
      "loss": 5.0906,
      "step": 123
    },
    {
      "epoch": 0.005006055712555511,
      "grad_norm": 209.64614868164062,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 5.0965,
      "step": 124
    },
    {
      "epoch": 0.005046427129592248,
      "grad_norm": 171.18679809570312,
      "learning_rate": 3.360215053763441e-06,
      "loss": 5.1078,
      "step": 125
    },
    {
      "epoch": 0.005086798546628987,
      "grad_norm": 176.75665283203125,
      "learning_rate": 3.3870967741935484e-06,
      "loss": 7.0827,
      "step": 126
    },
    {
      "epoch": 0.0051271699636657245,
      "grad_norm": 209.19798278808594,
      "learning_rate": 3.413978494623656e-06,
      "loss": 5.3554,
      "step": 127
    },
    {
      "epoch": 0.005167541380702463,
      "grad_norm": 237.3340301513672,
      "learning_rate": 3.440860215053764e-06,
      "loss": 7.1648,
      "step": 128
    },
    {
      "epoch": 0.0052079127977392005,
      "grad_norm": 176.25531005859375,
      "learning_rate": 3.4677419354838714e-06,
      "loss": 6.3032,
      "step": 129
    },
    {
      "epoch": 0.005248284214775939,
      "grad_norm": 153.005615234375,
      "learning_rate": 3.494623655913979e-06,
      "loss": 6.5005,
      "step": 130
    },
    {
      "epoch": 0.005288655631812677,
      "grad_norm": 174.07452392578125,
      "learning_rate": 3.5215053763440865e-06,
      "loss": 6.1746,
      "step": 131
    },
    {
      "epoch": 0.005329027048849415,
      "grad_norm": 288.2896728515625,
      "learning_rate": 3.548387096774194e-06,
      "loss": 7.0249,
      "step": 132
    },
    {
      "epoch": 0.005369398465886153,
      "grad_norm": 204.25050354003906,
      "learning_rate": 3.5752688172043015e-06,
      "loss": 7.9078,
      "step": 133
    },
    {
      "epoch": 0.00540976988292289,
      "grad_norm": 171.39553833007812,
      "learning_rate": 3.602150537634409e-06,
      "loss": 7.3114,
      "step": 134
    },
    {
      "epoch": 0.005450141299959629,
      "grad_norm": 312.931640625,
      "learning_rate": 3.6290322580645166e-06,
      "loss": 7.1282,
      "step": 135
    },
    {
      "epoch": 0.005490512716996366,
      "grad_norm": 153.33360290527344,
      "learning_rate": 3.655913978494624e-06,
      "loss": 7.5236,
      "step": 136
    },
    {
      "epoch": 0.005530884134033105,
      "grad_norm": 164.03880310058594,
      "learning_rate": 3.6827956989247317e-06,
      "loss": 6.4888,
      "step": 137
    },
    {
      "epoch": 0.005571255551069842,
      "grad_norm": 160.548583984375,
      "learning_rate": 3.7096774193548392e-06,
      "loss": 6.0387,
      "step": 138
    },
    {
      "epoch": 0.005611626968106581,
      "grad_norm": 155.964111328125,
      "learning_rate": 3.7365591397849468e-06,
      "loss": 7.2219,
      "step": 139
    },
    {
      "epoch": 0.005651998385143318,
      "grad_norm": 224.328125,
      "learning_rate": 3.763440860215054e-06,
      "loss": 7.5657,
      "step": 140
    },
    {
      "epoch": 0.005692369802180057,
      "grad_norm": 167.64039611816406,
      "learning_rate": 3.7903225806451614e-06,
      "loss": 6.3225,
      "step": 141
    },
    {
      "epoch": 0.005732741219216794,
      "grad_norm": 227.20704650878906,
      "learning_rate": 3.817204301075269e-06,
      "loss": 7.811,
      "step": 142
    },
    {
      "epoch": 0.005773112636253533,
      "grad_norm": 187.5634002685547,
      "learning_rate": 3.8440860215053765e-06,
      "loss": 6.1326,
      "step": 143
    },
    {
      "epoch": 0.0058134840532902705,
      "grad_norm": 198.0111083984375,
      "learning_rate": 3.870967741935484e-06,
      "loss": 6.2138,
      "step": 144
    },
    {
      "epoch": 0.005853855470327008,
      "grad_norm": 181.56182861328125,
      "learning_rate": 3.8978494623655915e-06,
      "loss": 6.1992,
      "step": 145
    },
    {
      "epoch": 0.0058942268873637465,
      "grad_norm": 154.05711364746094,
      "learning_rate": 3.924731182795699e-06,
      "loss": 5.6,
      "step": 146
    },
    {
      "epoch": 0.005934598304400484,
      "grad_norm": 273.8877868652344,
      "learning_rate": 3.951612903225807e-06,
      "loss": 6.8838,
      "step": 147
    },
    {
      "epoch": 0.005974969721437223,
      "grad_norm": 181.08224487304688,
      "learning_rate": 3.978494623655914e-06,
      "loss": 7.0432,
      "step": 148
    },
    {
      "epoch": 0.00601534113847396,
      "grad_norm": 196.79981994628906,
      "learning_rate": 4.005376344086022e-06,
      "loss": 7.6055,
      "step": 149
    },
    {
      "epoch": 0.006055712555510699,
      "grad_norm": 209.3241424560547,
      "learning_rate": 4.032258064516129e-06,
      "loss": 5.2674,
      "step": 150
    },
    {
      "epoch": 0.006096083972547436,
      "grad_norm": 171.3875274658203,
      "learning_rate": 4.059139784946237e-06,
      "loss": 5.9999,
      "step": 151
    },
    {
      "epoch": 0.006136455389584175,
      "grad_norm": 159.93841552734375,
      "learning_rate": 4.086021505376344e-06,
      "loss": 7.4748,
      "step": 152
    },
    {
      "epoch": 0.006176826806620912,
      "grad_norm": 203.91287231445312,
      "learning_rate": 4.112903225806452e-06,
      "loss": 6.5627,
      "step": 153
    },
    {
      "epoch": 0.006217198223657651,
      "grad_norm": 165.62144470214844,
      "learning_rate": 4.139784946236559e-06,
      "loss": 6.8352,
      "step": 154
    },
    {
      "epoch": 0.006257569640694388,
      "grad_norm": 173.3909454345703,
      "learning_rate": 4.166666666666667e-06,
      "loss": 6.98,
      "step": 155
    },
    {
      "epoch": 0.006297941057731127,
      "grad_norm": 241.4010009765625,
      "learning_rate": 4.193548387096774e-06,
      "loss": 5.5187,
      "step": 156
    },
    {
      "epoch": 0.006338312474767864,
      "grad_norm": 151.5513458251953,
      "learning_rate": 4.220430107526882e-06,
      "loss": 7.4073,
      "step": 157
    },
    {
      "epoch": 0.006378683891804602,
      "grad_norm": 222.0614776611328,
      "learning_rate": 4.2473118279569895e-06,
      "loss": 5.8106,
      "step": 158
    },
    {
      "epoch": 0.00641905530884134,
      "grad_norm": 117.0156021118164,
      "learning_rate": 4.274193548387097e-06,
      "loss": 6.6767,
      "step": 159
    },
    {
      "epoch": 0.006459426725878078,
      "grad_norm": 158.41123962402344,
      "learning_rate": 4.3010752688172045e-06,
      "loss": 7.0353,
      "step": 160
    },
    {
      "epoch": 0.0064997981429148165,
      "grad_norm": 184.31045532226562,
      "learning_rate": 4.327956989247312e-06,
      "loss": 7.857,
      "step": 161
    },
    {
      "epoch": 0.006540169559951554,
      "grad_norm": 204.30186462402344,
      "learning_rate": 4.35483870967742e-06,
      "loss": 6.9737,
      "step": 162
    },
    {
      "epoch": 0.0065805409769882925,
      "grad_norm": 323.2314758300781,
      "learning_rate": 4.381720430107527e-06,
      "loss": 5.9864,
      "step": 163
    },
    {
      "epoch": 0.00662091239402503,
      "grad_norm": 154.63963317871094,
      "learning_rate": 4.408602150537635e-06,
      "loss": 7.459,
      "step": 164
    },
    {
      "epoch": 0.006661283811061769,
      "grad_norm": 868.4306030273438,
      "learning_rate": 4.435483870967742e-06,
      "loss": 5.9287,
      "step": 165
    },
    {
      "epoch": 0.006701655228098506,
      "grad_norm": 186.2172088623047,
      "learning_rate": 4.46236559139785e-06,
      "loss": 7.3406,
      "step": 166
    },
    {
      "epoch": 0.006742026645135245,
      "grad_norm": 221.76576232910156,
      "learning_rate": 4.489247311827957e-06,
      "loss": 6.4337,
      "step": 167
    },
    {
      "epoch": 0.006782398062171982,
      "grad_norm": 9595.216796875,
      "learning_rate": 4.516129032258065e-06,
      "loss": 7.1429,
      "step": 168
    },
    {
      "epoch": 0.00682276947920872,
      "grad_norm": 178.44357299804688,
      "learning_rate": 4.543010752688172e-06,
      "loss": 6.7543,
      "step": 169
    },
    {
      "epoch": 0.006863140896245458,
      "grad_norm": 151.81483459472656,
      "learning_rate": 4.56989247311828e-06,
      "loss": 7.1929,
      "step": 170
    },
    {
      "epoch": 0.006903512313282196,
      "grad_norm": 203.39305114746094,
      "learning_rate": 4.596774193548387e-06,
      "loss": 6.3456,
      "step": 171
    },
    {
      "epoch": 0.006943883730318934,
      "grad_norm": 374.68707275390625,
      "learning_rate": 4.623655913978495e-06,
      "loss": 7.0906,
      "step": 172
    },
    {
      "epoch": 0.006984255147355672,
      "grad_norm": 2090.06787109375,
      "learning_rate": 4.6505376344086025e-06,
      "loss": 6.503,
      "step": 173
    },
    {
      "epoch": 0.00702462656439241,
      "grad_norm": 147.56536865234375,
      "learning_rate": 4.67741935483871e-06,
      "loss": 7.3752,
      "step": 174
    },
    {
      "epoch": 0.007064997981429148,
      "grad_norm": 179.8546600341797,
      "learning_rate": 4.7043010752688175e-06,
      "loss": 5.728,
      "step": 175
    },
    {
      "epoch": 0.007105369398465886,
      "grad_norm": 167.02294921875,
      "learning_rate": 4.731182795698925e-06,
      "loss": 7.2109,
      "step": 176
    },
    {
      "epoch": 0.007145740815502624,
      "grad_norm": 135.5717010498047,
      "learning_rate": 4.758064516129033e-06,
      "loss": 7.0728,
      "step": 177
    },
    {
      "epoch": 0.0071861122325393625,
      "grad_norm": 166.4923095703125,
      "learning_rate": 4.78494623655914e-06,
      "loss": 6.224,
      "step": 178
    },
    {
      "epoch": 0.0072264836495761,
      "grad_norm": 164.61575317382812,
      "learning_rate": 4.811827956989248e-06,
      "loss": 5.8806,
      "step": 179
    },
    {
      "epoch": 0.0072668550666128385,
      "grad_norm": 274.1067199707031,
      "learning_rate": 4.838709677419355e-06,
      "loss": 5.4665,
      "step": 180
    },
    {
      "epoch": 0.007307226483649576,
      "grad_norm": 137.16050720214844,
      "learning_rate": 4.865591397849463e-06,
      "loss": 6.7735,
      "step": 181
    },
    {
      "epoch": 0.007347597900686314,
      "grad_norm": 144.06166076660156,
      "learning_rate": 4.89247311827957e-06,
      "loss": 6.3312,
      "step": 182
    },
    {
      "epoch": 0.007387969317723052,
      "grad_norm": 148.45126342773438,
      "learning_rate": 4.919354838709678e-06,
      "loss": 5.2057,
      "step": 183
    },
    {
      "epoch": 0.00742834073475979,
      "grad_norm": 158.2405548095703,
      "learning_rate": 4.946236559139785e-06,
      "loss": 6.1569,
      "step": 184
    },
    {
      "epoch": 0.007468712151796528,
      "grad_norm": 244.5386962890625,
      "learning_rate": 4.973118279569893e-06,
      "loss": 6.768,
      "step": 185
    },
    {
      "epoch": 0.007509083568833266,
      "grad_norm": 229.32789611816406,
      "learning_rate": 5e-06,
      "loss": 6.5994,
      "step": 186
    },
    {
      "epoch": 0.007549454985870004,
      "grad_norm": 143.81138610839844,
      "learning_rate": 5.026881720430108e-06,
      "loss": 6.1656,
      "step": 187
    },
    {
      "epoch": 0.007589826402906742,
      "grad_norm": 121.43917846679688,
      "learning_rate": 5.0537634408602155e-06,
      "loss": 6.8324,
      "step": 188
    },
    {
      "epoch": 0.00763019781994348,
      "grad_norm": 179.71234130859375,
      "learning_rate": 5.080645161290323e-06,
      "loss": 6.9593,
      "step": 189
    },
    {
      "epoch": 0.007670569236980218,
      "grad_norm": 149.1662139892578,
      "learning_rate": 5.1075268817204305e-06,
      "loss": 6.3063,
      "step": 190
    },
    {
      "epoch": 0.007710940654016956,
      "grad_norm": 143.81448364257812,
      "learning_rate": 5.134408602150538e-06,
      "loss": 6.3106,
      "step": 191
    },
    {
      "epoch": 0.007751312071053694,
      "grad_norm": 136.3497772216797,
      "learning_rate": 5.161290322580646e-06,
      "loss": 6.0291,
      "step": 192
    },
    {
      "epoch": 0.0077916834880904316,
      "grad_norm": 144.73191833496094,
      "learning_rate": 5.188172043010753e-06,
      "loss": 7.6493,
      "step": 193
    },
    {
      "epoch": 0.00783205490512717,
      "grad_norm": 199.91461181640625,
      "learning_rate": 5.215053763440861e-06,
      "loss": 6.5372,
      "step": 194
    },
    {
      "epoch": 0.007872426322163908,
      "grad_norm": 174.95303344726562,
      "learning_rate": 5.241935483870968e-06,
      "loss": 5.3619,
      "step": 195
    },
    {
      "epoch": 0.007912797739200646,
      "grad_norm": 127.42781829833984,
      "learning_rate": 5.268817204301076e-06,
      "loss": 6.2577,
      "step": 196
    },
    {
      "epoch": 0.007953169156237384,
      "grad_norm": 184.4600372314453,
      "learning_rate": 5.295698924731183e-06,
      "loss": 8.092,
      "step": 197
    },
    {
      "epoch": 0.007993540573274121,
      "grad_norm": 148.27842712402344,
      "learning_rate": 5.322580645161291e-06,
      "loss": 5.9451,
      "step": 198
    },
    {
      "epoch": 0.00803391199031086,
      "grad_norm": 139.15646362304688,
      "learning_rate": 5.349462365591398e-06,
      "loss": 7.0292,
      "step": 199
    },
    {
      "epoch": 0.008074283407347598,
      "grad_norm": 234.504638671875,
      "learning_rate": 5.376344086021506e-06,
      "loss": 6.426,
      "step": 200
    },
    {
      "epoch": 0.008114654824384336,
      "grad_norm": 230.1488800048828,
      "learning_rate": 5.4032258064516126e-06,
      "loss": 5.7559,
      "step": 201
    },
    {
      "epoch": 0.008155026241421073,
      "grad_norm": 164.3559112548828,
      "learning_rate": 5.43010752688172e-06,
      "loss": 5.5022,
      "step": 202
    },
    {
      "epoch": 0.008195397658457813,
      "grad_norm": 276.7715759277344,
      "learning_rate": 5.456989247311828e-06,
      "loss": 6.6097,
      "step": 203
    },
    {
      "epoch": 0.00823576907549455,
      "grad_norm": 144.22964477539062,
      "learning_rate": 5.483870967741935e-06,
      "loss": 7.0859,
      "step": 204
    },
    {
      "epoch": 0.008276140492531288,
      "grad_norm": 188.50994873046875,
      "learning_rate": 5.510752688172043e-06,
      "loss": 6.2613,
      "step": 205
    },
    {
      "epoch": 0.008316511909568025,
      "grad_norm": 284.1933898925781,
      "learning_rate": 5.53763440860215e-06,
      "loss": 7.3929,
      "step": 206
    },
    {
      "epoch": 0.008356883326604763,
      "grad_norm": 153.447265625,
      "learning_rate": 5.564516129032258e-06,
      "loss": 6.2396,
      "step": 207
    },
    {
      "epoch": 0.008397254743641502,
      "grad_norm": 153.4487762451172,
      "learning_rate": 5.591397849462365e-06,
      "loss": 7.2955,
      "step": 208
    },
    {
      "epoch": 0.00843762616067824,
      "grad_norm": 204.87440490722656,
      "learning_rate": 5.618279569892473e-06,
      "loss": 6.5769,
      "step": 209
    },
    {
      "epoch": 0.008477997577714978,
      "grad_norm": 170.1885986328125,
      "learning_rate": 5.645161290322582e-06,
      "loss": 5.8274,
      "step": 210
    },
    {
      "epoch": 0.008518368994751715,
      "grad_norm": 185.06704711914062,
      "learning_rate": 5.672043010752689e-06,
      "loss": 5.3402,
      "step": 211
    },
    {
      "epoch": 0.008558740411788454,
      "grad_norm": 146.51202392578125,
      "learning_rate": 5.698924731182796e-06,
      "loss": 6.6024,
      "step": 212
    },
    {
      "epoch": 0.008599111828825192,
      "grad_norm": 195.1490478515625,
      "learning_rate": 5.725806451612904e-06,
      "loss": 6.5686,
      "step": 213
    },
    {
      "epoch": 0.00863948324586193,
      "grad_norm": 181.69944763183594,
      "learning_rate": 5.752688172043011e-06,
      "loss": 4.7039,
      "step": 214
    },
    {
      "epoch": 0.008679854662898667,
      "grad_norm": 179.2821044921875,
      "learning_rate": 5.779569892473119e-06,
      "loss": 6.7215,
      "step": 215
    },
    {
      "epoch": 0.008720226079935407,
      "grad_norm": 214.259765625,
      "learning_rate": 5.806451612903226e-06,
      "loss": 6.6369,
      "step": 216
    },
    {
      "epoch": 0.008760597496972144,
      "grad_norm": 335.80047607421875,
      "learning_rate": 5.833333333333334e-06,
      "loss": 5.3567,
      "step": 217
    },
    {
      "epoch": 0.008800968914008882,
      "grad_norm": 267.7880554199219,
      "learning_rate": 5.8602150537634415e-06,
      "loss": 6.3858,
      "step": 218
    },
    {
      "epoch": 0.00884134033104562,
      "grad_norm": 279.01092529296875,
      "learning_rate": 5.887096774193549e-06,
      "loss": 6.4347,
      "step": 219
    },
    {
      "epoch": 0.008881711748082357,
      "grad_norm": 208.0467529296875,
      "learning_rate": 5.9139784946236566e-06,
      "loss": 5.6141,
      "step": 220
    },
    {
      "epoch": 0.008922083165119096,
      "grad_norm": 162.35609436035156,
      "learning_rate": 5.940860215053764e-06,
      "loss": 6.3729,
      "step": 221
    },
    {
      "epoch": 0.008962454582155834,
      "grad_norm": 144.68397521972656,
      "learning_rate": 5.967741935483872e-06,
      "loss": 6.249,
      "step": 222
    },
    {
      "epoch": 0.009002825999192571,
      "grad_norm": 167.8122100830078,
      "learning_rate": 5.994623655913979e-06,
      "loss": 7.3278,
      "step": 223
    },
    {
      "epoch": 0.009043197416229309,
      "grad_norm": 266.6229553222656,
      "learning_rate": 6.021505376344087e-06,
      "loss": 5.9342,
      "step": 224
    },
    {
      "epoch": 0.009083568833266048,
      "grad_norm": 215.11297607421875,
      "learning_rate": 6.048387096774194e-06,
      "loss": 6.6834,
      "step": 225
    },
    {
      "epoch": 0.009123940250302786,
      "grad_norm": 139.09666442871094,
      "learning_rate": 6.075268817204302e-06,
      "loss": 6.977,
      "step": 226
    },
    {
      "epoch": 0.009164311667339524,
      "grad_norm": 151.60997009277344,
      "learning_rate": 6.102150537634409e-06,
      "loss": 6.3749,
      "step": 227
    },
    {
      "epoch": 0.009204683084376261,
      "grad_norm": 177.88340759277344,
      "learning_rate": 6.129032258064517e-06,
      "loss": 7.1203,
      "step": 228
    },
    {
      "epoch": 0.009245054501412999,
      "grad_norm": 172.05958557128906,
      "learning_rate": 6.155913978494624e-06,
      "loss": 6.4501,
      "step": 229
    },
    {
      "epoch": 0.009285425918449738,
      "grad_norm": 150.5163116455078,
      "learning_rate": 6.182795698924732e-06,
      "loss": 6.7048,
      "step": 230
    },
    {
      "epoch": 0.009325797335486476,
      "grad_norm": 161.68321228027344,
      "learning_rate": 6.209677419354839e-06,
      "loss": 5.6187,
      "step": 231
    },
    {
      "epoch": 0.009366168752523213,
      "grad_norm": 187.57286071777344,
      "learning_rate": 6.236559139784947e-06,
      "loss": 6.4803,
      "step": 232
    },
    {
      "epoch": 0.00940654016955995,
      "grad_norm": 157.749755859375,
      "learning_rate": 6.2634408602150545e-06,
      "loss": 6.6852,
      "step": 233
    },
    {
      "epoch": 0.00944691158659669,
      "grad_norm": 153.0474090576172,
      "learning_rate": 6.290322580645162e-06,
      "loss": 5.9366,
      "step": 234
    },
    {
      "epoch": 0.009487283003633428,
      "grad_norm": 168.70960998535156,
      "learning_rate": 6.3172043010752696e-06,
      "loss": 6.1006,
      "step": 235
    },
    {
      "epoch": 0.009527654420670165,
      "grad_norm": 138.58070373535156,
      "learning_rate": 6.344086021505377e-06,
      "loss": 5.2573,
      "step": 236
    },
    {
      "epoch": 0.009568025837706903,
      "grad_norm": 153.41761779785156,
      "learning_rate": 6.370967741935485e-06,
      "loss": 6.0996,
      "step": 237
    },
    {
      "epoch": 0.009608397254743642,
      "grad_norm": 211.60557556152344,
      "learning_rate": 6.397849462365592e-06,
      "loss": 5.3854,
      "step": 238
    },
    {
      "epoch": 0.00964876867178038,
      "grad_norm": 128.37132263183594,
      "learning_rate": 6.4247311827957e-06,
      "loss": 6.4036,
      "step": 239
    },
    {
      "epoch": 0.009689140088817117,
      "grad_norm": 179.22828674316406,
      "learning_rate": 6.451612903225806e-06,
      "loss": 5.4708,
      "step": 240
    },
    {
      "epoch": 0.009729511505853855,
      "grad_norm": 165.73745727539062,
      "learning_rate": 6.478494623655914e-06,
      "loss": 6.4853,
      "step": 241
    },
    {
      "epoch": 0.009769882922890593,
      "grad_norm": 112.76561737060547,
      "learning_rate": 6.5053763440860214e-06,
      "loss": 6.01,
      "step": 242
    },
    {
      "epoch": 0.009810254339927332,
      "grad_norm": 186.1405029296875,
      "learning_rate": 6.532258064516129e-06,
      "loss": 6.715,
      "step": 243
    },
    {
      "epoch": 0.00985062575696407,
      "grad_norm": 163.46954345703125,
      "learning_rate": 6.5591397849462365e-06,
      "loss": 6.6239,
      "step": 244
    },
    {
      "epoch": 0.009890997174000807,
      "grad_norm": 133.61183166503906,
      "learning_rate": 6.586021505376344e-06,
      "loss": 6.1591,
      "step": 245
    },
    {
      "epoch": 0.009931368591037545,
      "grad_norm": 172.98265075683594,
      "learning_rate": 6.612903225806452e-06,
      "loss": 6.4554,
      "step": 246
    },
    {
      "epoch": 0.009971740008074284,
      "grad_norm": 149.61134338378906,
      "learning_rate": 6.639784946236559e-06,
      "loss": 6.5007,
      "step": 247
    },
    {
      "epoch": 0.010012111425111022,
      "grad_norm": 179.8397674560547,
      "learning_rate": 6.666666666666667e-06,
      "loss": 5.458,
      "step": 248
    },
    {
      "epoch": 0.01005248284214776,
      "grad_norm": 137.73989868164062,
      "learning_rate": 6.693548387096774e-06,
      "loss": 7.0297,
      "step": 249
    },
    {
      "epoch": 0.010092854259184497,
      "grad_norm": 157.6536407470703,
      "learning_rate": 6.720430107526882e-06,
      "loss": 5.9914,
      "step": 250
    },
    {
      "epoch": 0.010133225676221236,
      "grad_norm": 151.87315368652344,
      "learning_rate": 6.747311827956989e-06,
      "loss": 7.3132,
      "step": 251
    },
    {
      "epoch": 0.010173597093257974,
      "grad_norm": 148.36648559570312,
      "learning_rate": 6.774193548387097e-06,
      "loss": 6.0062,
      "step": 252
    },
    {
      "epoch": 0.010213968510294711,
      "grad_norm": 155.5176544189453,
      "learning_rate": 6.801075268817204e-06,
      "loss": 7.7205,
      "step": 253
    },
    {
      "epoch": 0.010254339927331449,
      "grad_norm": 164.8324432373047,
      "learning_rate": 6.827956989247312e-06,
      "loss": 5.6221,
      "step": 254
    },
    {
      "epoch": 0.010294711344368187,
      "grad_norm": 184.9126739501953,
      "learning_rate": 6.854838709677419e-06,
      "loss": 5.9016,
      "step": 255
    },
    {
      "epoch": 0.010335082761404926,
      "grad_norm": 171.2563018798828,
      "learning_rate": 6.881720430107528e-06,
      "loss": 6.0983,
      "step": 256
    },
    {
      "epoch": 0.010375454178441663,
      "grad_norm": 186.4590301513672,
      "learning_rate": 6.908602150537635e-06,
      "loss": 7.8151,
      "step": 257
    },
    {
      "epoch": 0.010415825595478401,
      "grad_norm": 146.3933868408203,
      "learning_rate": 6.935483870967743e-06,
      "loss": 5.1712,
      "step": 258
    },
    {
      "epoch": 0.010456197012515139,
      "grad_norm": 188.5956573486328,
      "learning_rate": 6.96236559139785e-06,
      "loss": 5.825,
      "step": 259
    },
    {
      "epoch": 0.010496568429551878,
      "grad_norm": 183.62863159179688,
      "learning_rate": 6.989247311827958e-06,
      "loss": 6.5169,
      "step": 260
    },
    {
      "epoch": 0.010536939846588616,
      "grad_norm": 201.1549072265625,
      "learning_rate": 7.0161290322580654e-06,
      "loss": 5.7836,
      "step": 261
    },
    {
      "epoch": 0.010577311263625353,
      "grad_norm": 129.1268310546875,
      "learning_rate": 7.043010752688173e-06,
      "loss": 5.5589,
      "step": 262
    },
    {
      "epoch": 0.01061768268066209,
      "grad_norm": 151.9335479736328,
      "learning_rate": 7.0698924731182805e-06,
      "loss": 6.3797,
      "step": 263
    },
    {
      "epoch": 0.01065805409769883,
      "grad_norm": 188.2655029296875,
      "learning_rate": 7.096774193548388e-06,
      "loss": 6.7148,
      "step": 264
    },
    {
      "epoch": 0.010698425514735568,
      "grad_norm": 140.9820098876953,
      "learning_rate": 7.1236559139784956e-06,
      "loss": 5.2034,
      "step": 265
    },
    {
      "epoch": 0.010738796931772305,
      "grad_norm": 224.82342529296875,
      "learning_rate": 7.150537634408603e-06,
      "loss": 6.1126,
      "step": 266
    },
    {
      "epoch": 0.010779168348809043,
      "grad_norm": 193.43798828125,
      "learning_rate": 7.177419354838711e-06,
      "loss": 6.1006,
      "step": 267
    },
    {
      "epoch": 0.01081953976584578,
      "grad_norm": 140.7742156982422,
      "learning_rate": 7.204301075268818e-06,
      "loss": 5.5804,
      "step": 268
    },
    {
      "epoch": 0.01085991118288252,
      "grad_norm": 139.33636474609375,
      "learning_rate": 7.231182795698926e-06,
      "loss": 7.2362,
      "step": 269
    },
    {
      "epoch": 0.010900282599919257,
      "grad_norm": 235.5434112548828,
      "learning_rate": 7.258064516129033e-06,
      "loss": 6.8047,
      "step": 270
    },
    {
      "epoch": 0.010940654016955995,
      "grad_norm": 176.37010192871094,
      "learning_rate": 7.284946236559141e-06,
      "loss": 6.0901,
      "step": 271
    },
    {
      "epoch": 0.010981025433992733,
      "grad_norm": 178.64390563964844,
      "learning_rate": 7.311827956989248e-06,
      "loss": 5.0307,
      "step": 272
    },
    {
      "epoch": 0.011021396851029472,
      "grad_norm": 222.15716552734375,
      "learning_rate": 7.338709677419356e-06,
      "loss": 5.6092,
      "step": 273
    },
    {
      "epoch": 0.01106176826806621,
      "grad_norm": 199.15992736816406,
      "learning_rate": 7.365591397849463e-06,
      "loss": 6.6303,
      "step": 274
    },
    {
      "epoch": 0.011102139685102947,
      "grad_norm": 127.41642761230469,
      "learning_rate": 7.392473118279571e-06,
      "loss": 5.586,
      "step": 275
    },
    {
      "epoch": 0.011142511102139685,
      "grad_norm": 202.22654724121094,
      "learning_rate": 7.4193548387096784e-06,
      "loss": 5.9915,
      "step": 276
    },
    {
      "epoch": 0.011182882519176422,
      "grad_norm": 246.00289916992188,
      "learning_rate": 7.446236559139786e-06,
      "loss": 6.4439,
      "step": 277
    },
    {
      "epoch": 0.011223253936213162,
      "grad_norm": 123.17761993408203,
      "learning_rate": 7.4731182795698935e-06,
      "loss": 7.84,
      "step": 278
    },
    {
      "epoch": 0.0112636253532499,
      "grad_norm": 158.73880004882812,
      "learning_rate": 7.500000000000001e-06,
      "loss": 5.9798,
      "step": 279
    },
    {
      "epoch": 0.011303996770286637,
      "grad_norm": 160.1876220703125,
      "learning_rate": 7.526881720430108e-06,
      "loss": 5.3074,
      "step": 280
    },
    {
      "epoch": 0.011344368187323374,
      "grad_norm": 116.98528289794922,
      "learning_rate": 7.553763440860215e-06,
      "loss": 6.3973,
      "step": 281
    },
    {
      "epoch": 0.011384739604360114,
      "grad_norm": 204.15341186523438,
      "learning_rate": 7.580645161290323e-06,
      "loss": 6.2102,
      "step": 282
    },
    {
      "epoch": 0.011425111021396851,
      "grad_norm": 123.1771240234375,
      "learning_rate": 7.60752688172043e-06,
      "loss": 5.3524,
      "step": 283
    },
    {
      "epoch": 0.011465482438433589,
      "grad_norm": 170.76605224609375,
      "learning_rate": 7.634408602150538e-06,
      "loss": 6.3894,
      "step": 284
    },
    {
      "epoch": 0.011505853855470326,
      "grad_norm": 165.49606323242188,
      "learning_rate": 7.661290322580646e-06,
      "loss": 5.6161,
      "step": 285
    },
    {
      "epoch": 0.011546225272507066,
      "grad_norm": 167.19125366210938,
      "learning_rate": 7.688172043010753e-06,
      "loss": 5.5388,
      "step": 286
    },
    {
      "epoch": 0.011586596689543803,
      "grad_norm": 192.6871795654297,
      "learning_rate": 7.715053763440861e-06,
      "loss": 5.2253,
      "step": 287
    },
    {
      "epoch": 0.011626968106580541,
      "grad_norm": 156.7299346923828,
      "learning_rate": 7.741935483870968e-06,
      "loss": 6.6661,
      "step": 288
    },
    {
      "epoch": 0.011667339523617279,
      "grad_norm": 240.68226623535156,
      "learning_rate": 7.768817204301076e-06,
      "loss": 6.1269,
      "step": 289
    },
    {
      "epoch": 0.011707710940654016,
      "grad_norm": 140.9192352294922,
      "learning_rate": 7.795698924731183e-06,
      "loss": 5.9347,
      "step": 290
    },
    {
      "epoch": 0.011748082357690755,
      "grad_norm": 169.92520141601562,
      "learning_rate": 7.822580645161291e-06,
      "loss": 5.3168,
      "step": 291
    },
    {
      "epoch": 0.011788453774727493,
      "grad_norm": 143.77525329589844,
      "learning_rate": 7.849462365591398e-06,
      "loss": 6.1149,
      "step": 292
    },
    {
      "epoch": 0.01182882519176423,
      "grad_norm": 205.69912719726562,
      "learning_rate": 7.876344086021507e-06,
      "loss": 6.5812,
      "step": 293
    },
    {
      "epoch": 0.011869196608800968,
      "grad_norm": 125.02828216552734,
      "learning_rate": 7.903225806451613e-06,
      "loss": 5.2885,
      "step": 294
    },
    {
      "epoch": 0.011909568025837708,
      "grad_norm": 169.81434631347656,
      "learning_rate": 7.93010752688172e-06,
      "loss": 6.4672,
      "step": 295
    },
    {
      "epoch": 0.011949939442874445,
      "grad_norm": 575.8583374023438,
      "learning_rate": 7.956989247311828e-06,
      "loss": 5.4809,
      "step": 296
    },
    {
      "epoch": 0.011990310859911183,
      "grad_norm": 177.44961547851562,
      "learning_rate": 7.983870967741935e-06,
      "loss": 5.2502,
      "step": 297
    },
    {
      "epoch": 0.01203068227694792,
      "grad_norm": 135.90541076660156,
      "learning_rate": 8.010752688172043e-06,
      "loss": 5.5639,
      "step": 298
    },
    {
      "epoch": 0.01207105369398466,
      "grad_norm": 254.48260498046875,
      "learning_rate": 8.03763440860215e-06,
      "loss": 7.6216,
      "step": 299
    },
    {
      "epoch": 0.012111425111021397,
      "grad_norm": 148.18295288085938,
      "learning_rate": 8.064516129032258e-06,
      "loss": 5.2405,
      "step": 300
    },
    {
      "epoch": 0.012151796528058135,
      "grad_norm": 145.10894775390625,
      "learning_rate": 8.091397849462365e-06,
      "loss": 6.0695,
      "step": 301
    },
    {
      "epoch": 0.012192167945094872,
      "grad_norm": 482.37158203125,
      "learning_rate": 8.118279569892473e-06,
      "loss": 6.4616,
      "step": 302
    },
    {
      "epoch": 0.01223253936213161,
      "grad_norm": 173.00965881347656,
      "learning_rate": 8.145161290322582e-06,
      "loss": 5.4878,
      "step": 303
    },
    {
      "epoch": 0.01227291077916835,
      "grad_norm": 142.0355224609375,
      "learning_rate": 8.172043010752689e-06,
      "loss": 5.8783,
      "step": 304
    },
    {
      "epoch": 0.012313282196205087,
      "grad_norm": 288.27337646484375,
      "learning_rate": 8.198924731182797e-06,
      "loss": 6.5922,
      "step": 305
    },
    {
      "epoch": 0.012353653613241825,
      "grad_norm": 148.74903869628906,
      "learning_rate": 8.225806451612904e-06,
      "loss": 6.8375,
      "step": 306
    },
    {
      "epoch": 0.012394025030278562,
      "grad_norm": 189.55584716796875,
      "learning_rate": 8.252688172043012e-06,
      "loss": 7.0188,
      "step": 307
    },
    {
      "epoch": 0.012434396447315301,
      "grad_norm": 323.048828125,
      "learning_rate": 8.279569892473119e-06,
      "loss": 5.4304,
      "step": 308
    },
    {
      "epoch": 0.012474767864352039,
      "grad_norm": 290.6348571777344,
      "learning_rate": 8.306451612903227e-06,
      "loss": 7.0856,
      "step": 309
    },
    {
      "epoch": 0.012515139281388777,
      "grad_norm": 177.04356384277344,
      "learning_rate": 8.333333333333334e-06,
      "loss": 6.6141,
      "step": 310
    },
    {
      "epoch": 0.012555510698425514,
      "grad_norm": 185.6285400390625,
      "learning_rate": 8.360215053763442e-06,
      "loss": 5.87,
      "step": 311
    },
    {
      "epoch": 0.012595882115462254,
      "grad_norm": 149.58139038085938,
      "learning_rate": 8.387096774193549e-06,
      "loss": 5.1923,
      "step": 312
    },
    {
      "epoch": 0.012636253532498991,
      "grad_norm": 139.65866088867188,
      "learning_rate": 8.413978494623657e-06,
      "loss": 7.6333,
      "step": 313
    },
    {
      "epoch": 0.012676624949535729,
      "grad_norm": 168.72804260253906,
      "learning_rate": 8.440860215053764e-06,
      "loss": 4.7332,
      "step": 314
    },
    {
      "epoch": 0.012716996366572466,
      "grad_norm": 358.77899169921875,
      "learning_rate": 8.467741935483872e-06,
      "loss": 5.0748,
      "step": 315
    },
    {
      "epoch": 0.012757367783609204,
      "grad_norm": 140.43209838867188,
      "learning_rate": 8.494623655913979e-06,
      "loss": 6.6303,
      "step": 316
    },
    {
      "epoch": 0.012797739200645943,
      "grad_norm": 143.21954345703125,
      "learning_rate": 8.521505376344087e-06,
      "loss": 6.0013,
      "step": 317
    },
    {
      "epoch": 0.01283811061768268,
      "grad_norm": 202.56382751464844,
      "learning_rate": 8.548387096774194e-06,
      "loss": 7.6287,
      "step": 318
    },
    {
      "epoch": 0.012878482034719418,
      "grad_norm": 138.8590087890625,
      "learning_rate": 8.575268817204302e-06,
      "loss": 5.986,
      "step": 319
    },
    {
      "epoch": 0.012918853451756156,
      "grad_norm": 165.8949432373047,
      "learning_rate": 8.602150537634409e-06,
      "loss": 6.8424,
      "step": 320
    },
    {
      "epoch": 0.012959224868792895,
      "grad_norm": 160.17041015625,
      "learning_rate": 8.629032258064517e-06,
      "loss": 5.6253,
      "step": 321
    },
    {
      "epoch": 0.012999596285829633,
      "grad_norm": 270.1112060546875,
      "learning_rate": 8.655913978494624e-06,
      "loss": 7.0605,
      "step": 322
    },
    {
      "epoch": 0.01303996770286637,
      "grad_norm": 196.14642333984375,
      "learning_rate": 8.682795698924733e-06,
      "loss": 5.8336,
      "step": 323
    },
    {
      "epoch": 0.013080339119903108,
      "grad_norm": 209.6278076171875,
      "learning_rate": 8.70967741935484e-06,
      "loss": 4.8061,
      "step": 324
    },
    {
      "epoch": 0.013120710536939846,
      "grad_norm": 208.86334228515625,
      "learning_rate": 8.736559139784948e-06,
      "loss": 5.045,
      "step": 325
    },
    {
      "epoch": 0.013161081953976585,
      "grad_norm": 161.16233825683594,
      "learning_rate": 8.763440860215054e-06,
      "loss": 5.8991,
      "step": 326
    },
    {
      "epoch": 0.013201453371013323,
      "grad_norm": 138.47067260742188,
      "learning_rate": 8.790322580645163e-06,
      "loss": 6.3201,
      "step": 327
    },
    {
      "epoch": 0.01324182478805006,
      "grad_norm": 195.635498046875,
      "learning_rate": 8.81720430107527e-06,
      "loss": 4.9335,
      "step": 328
    },
    {
      "epoch": 0.013282196205086798,
      "grad_norm": 141.53042602539062,
      "learning_rate": 8.844086021505378e-06,
      "loss": 6.0436,
      "step": 329
    },
    {
      "epoch": 0.013322567622123537,
      "grad_norm": 160.7103729248047,
      "learning_rate": 8.870967741935484e-06,
      "loss": 6.7767,
      "step": 330
    },
    {
      "epoch": 0.013362939039160275,
      "grad_norm": 153.08580017089844,
      "learning_rate": 8.897849462365593e-06,
      "loss": 6.8035,
      "step": 331
    },
    {
      "epoch": 0.013403310456197012,
      "grad_norm": 157.11683654785156,
      "learning_rate": 8.9247311827957e-06,
      "loss": 5.6123,
      "step": 332
    },
    {
      "epoch": 0.01344368187323375,
      "grad_norm": 149.35946655273438,
      "learning_rate": 8.951612903225806e-06,
      "loss": 6.5969,
      "step": 333
    },
    {
      "epoch": 0.01348405329027049,
      "grad_norm": 135.2218780517578,
      "learning_rate": 8.978494623655915e-06,
      "loss": 5.6346,
      "step": 334
    },
    {
      "epoch": 0.013524424707307227,
      "grad_norm": 149.80130004882812,
      "learning_rate": 9.005376344086021e-06,
      "loss": 6.1249,
      "step": 335
    },
    {
      "epoch": 0.013564796124343964,
      "grad_norm": 183.02622985839844,
      "learning_rate": 9.03225806451613e-06,
      "loss": 6.1385,
      "step": 336
    },
    {
      "epoch": 0.013605167541380702,
      "grad_norm": 232.41455078125,
      "learning_rate": 9.059139784946236e-06,
      "loss": 5.5874,
      "step": 337
    },
    {
      "epoch": 0.01364553895841744,
      "grad_norm": 124.25413513183594,
      "learning_rate": 9.086021505376345e-06,
      "loss": 5.587,
      "step": 338
    },
    {
      "epoch": 0.013685910375454179,
      "grad_norm": 152.35928344726562,
      "learning_rate": 9.112903225806451e-06,
      "loss": 4.9628,
      "step": 339
    },
    {
      "epoch": 0.013726281792490917,
      "grad_norm": 148.73828125,
      "learning_rate": 9.13978494623656e-06,
      "loss": 6.4625,
      "step": 340
    },
    {
      "epoch": 0.013766653209527654,
      "grad_norm": 131.19888305664062,
      "learning_rate": 9.166666666666666e-06,
      "loss": 6.8384,
      "step": 341
    },
    {
      "epoch": 0.013807024626564392,
      "grad_norm": 241.0561981201172,
      "learning_rate": 9.193548387096775e-06,
      "loss": 6.9925,
      "step": 342
    },
    {
      "epoch": 0.013847396043601131,
      "grad_norm": 133.66436767578125,
      "learning_rate": 9.220430107526881e-06,
      "loss": 5.6858,
      "step": 343
    },
    {
      "epoch": 0.013887767460637869,
      "grad_norm": 197.45689392089844,
      "learning_rate": 9.24731182795699e-06,
      "loss": 4.3137,
      "step": 344
    },
    {
      "epoch": 0.013928138877674606,
      "grad_norm": 203.11627197265625,
      "learning_rate": 9.274193548387097e-06,
      "loss": 6.47,
      "step": 345
    },
    {
      "epoch": 0.013968510294711344,
      "grad_norm": 158.8013916015625,
      "learning_rate": 9.301075268817205e-06,
      "loss": 7.2793,
      "step": 346
    },
    {
      "epoch": 0.014008881711748083,
      "grad_norm": 523.7225341796875,
      "learning_rate": 9.327956989247312e-06,
      "loss": 4.8915,
      "step": 347
    },
    {
      "epoch": 0.01404925312878482,
      "grad_norm": 167.30880737304688,
      "learning_rate": 9.35483870967742e-06,
      "loss": 5.002,
      "step": 348
    },
    {
      "epoch": 0.014089624545821558,
      "grad_norm": 224.39634704589844,
      "learning_rate": 9.381720430107528e-06,
      "loss": 6.1912,
      "step": 349
    },
    {
      "epoch": 0.014129995962858296,
      "grad_norm": 204.5623016357422,
      "learning_rate": 9.408602150537635e-06,
      "loss": 5.7262,
      "step": 350
    },
    {
      "epoch": 0.014170367379895034,
      "grad_norm": 133.5630645751953,
      "learning_rate": 9.435483870967743e-06,
      "loss": 5.5046,
      "step": 351
    },
    {
      "epoch": 0.014210738796931773,
      "grad_norm": 161.63250732421875,
      "learning_rate": 9.46236559139785e-06,
      "loss": 5.9697,
      "step": 352
    },
    {
      "epoch": 0.01425111021396851,
      "grad_norm": 130.17437744140625,
      "learning_rate": 9.489247311827959e-06,
      "loss": 7.5775,
      "step": 353
    },
    {
      "epoch": 0.014291481631005248,
      "grad_norm": 177.56362915039062,
      "learning_rate": 9.516129032258065e-06,
      "loss": 6.6655,
      "step": 354
    },
    {
      "epoch": 0.014331853048041986,
      "grad_norm": 163.59803771972656,
      "learning_rate": 9.543010752688174e-06,
      "loss": 5.3992,
      "step": 355
    },
    {
      "epoch": 0.014372224465078725,
      "grad_norm": 158.73101806640625,
      "learning_rate": 9.56989247311828e-06,
      "loss": 5.9386,
      "step": 356
    },
    {
      "epoch": 0.014412595882115463,
      "grad_norm": 392.4779357910156,
      "learning_rate": 9.596774193548389e-06,
      "loss": 4.5445,
      "step": 357
    },
    {
      "epoch": 0.0144529672991522,
      "grad_norm": 121.7234878540039,
      "learning_rate": 9.623655913978495e-06,
      "loss": 6.5728,
      "step": 358
    },
    {
      "epoch": 0.014493338716188938,
      "grad_norm": 222.40084838867188,
      "learning_rate": 9.650537634408604e-06,
      "loss": 5.9132,
      "step": 359
    },
    {
      "epoch": 0.014533710133225677,
      "grad_norm": 239.0767059326172,
      "learning_rate": 9.67741935483871e-06,
      "loss": 5.7828,
      "step": 360
    },
    {
      "epoch": 0.014574081550262415,
      "grad_norm": 146.70181274414062,
      "learning_rate": 9.704301075268819e-06,
      "loss": 6.3687,
      "step": 361
    },
    {
      "epoch": 0.014614452967299152,
      "grad_norm": 137.30978393554688,
      "learning_rate": 9.731182795698925e-06,
      "loss": 6.5584,
      "step": 362
    },
    {
      "epoch": 0.01465482438433589,
      "grad_norm": 157.4445037841797,
      "learning_rate": 9.758064516129034e-06,
      "loss": 6.2801,
      "step": 363
    },
    {
      "epoch": 0.014695195801372627,
      "grad_norm": 232.50955200195312,
      "learning_rate": 9.78494623655914e-06,
      "loss": 7.2511,
      "step": 364
    },
    {
      "epoch": 0.014735567218409367,
      "grad_norm": 145.9236297607422,
      "learning_rate": 9.811827956989249e-06,
      "loss": 5.2777,
      "step": 365
    },
    {
      "epoch": 0.014775938635446104,
      "grad_norm": 145.07325744628906,
      "learning_rate": 9.838709677419356e-06,
      "loss": 6.6541,
      "step": 366
    },
    {
      "epoch": 0.014816310052482842,
      "grad_norm": 139.48092651367188,
      "learning_rate": 9.865591397849464e-06,
      "loss": 6.6925,
      "step": 367
    },
    {
      "epoch": 0.01485668146951958,
      "grad_norm": 170.52194213867188,
      "learning_rate": 9.89247311827957e-06,
      "loss": 7.4683,
      "step": 368
    },
    {
      "epoch": 0.014897052886556319,
      "grad_norm": 13644.37890625,
      "learning_rate": 9.919354838709679e-06,
      "loss": 6.1699,
      "step": 369
    },
    {
      "epoch": 0.014937424303593056,
      "grad_norm": 148.9525146484375,
      "learning_rate": 9.946236559139786e-06,
      "loss": 5.8603,
      "step": 370
    },
    {
      "epoch": 0.014977795720629794,
      "grad_norm": 123.65349578857422,
      "learning_rate": 9.973118279569894e-06,
      "loss": 5.9559,
      "step": 371
    },
    {
      "epoch": 0.015018167137666532,
      "grad_norm": 164.55126953125,
      "learning_rate": 1e-05,
      "loss": 5.2741,
      "step": 372
    },
    {
      "epoch": 0.015058538554703271,
      "grad_norm": 245.86264038085938,
      "learning_rate": 1.0026881720430108e-05,
      "loss": 5.7711,
      "step": 373
    },
    {
      "epoch": 0.015098909971740009,
      "grad_norm": 626.5424194335938,
      "learning_rate": 1.0053763440860216e-05,
      "loss": 4.2301,
      "step": 374
    },
    {
      "epoch": 0.015139281388776746,
      "grad_norm": 140.90670776367188,
      "learning_rate": 1.0080645161290323e-05,
      "loss": 7.2655,
      "step": 375
    },
    {
      "epoch": 0.015179652805813484,
      "grad_norm": 147.33497619628906,
      "learning_rate": 1.0107526881720431e-05,
      "loss": 6.9832,
      "step": 376
    },
    {
      "epoch": 0.015220024222850221,
      "grad_norm": 111.72758483886719,
      "learning_rate": 1.0134408602150538e-05,
      "loss": 5.1442,
      "step": 377
    },
    {
      "epoch": 0.01526039563988696,
      "grad_norm": 160.34999084472656,
      "learning_rate": 1.0161290322580646e-05,
      "loss": 6.0374,
      "step": 378
    },
    {
      "epoch": 0.015300767056923698,
      "grad_norm": 197.53237915039062,
      "learning_rate": 1.0188172043010753e-05,
      "loss": 6.3797,
      "step": 379
    },
    {
      "epoch": 0.015341138473960436,
      "grad_norm": 173.19981384277344,
      "learning_rate": 1.0215053763440861e-05,
      "loss": 4.6909,
      "step": 380
    },
    {
      "epoch": 0.015381509890997173,
      "grad_norm": 147.51914978027344,
      "learning_rate": 1.0241935483870968e-05,
      "loss": 5.9532,
      "step": 381
    },
    {
      "epoch": 0.015421881308033913,
      "grad_norm": 142.89993286132812,
      "learning_rate": 1.0268817204301076e-05,
      "loss": 5.5397,
      "step": 382
    },
    {
      "epoch": 0.01546225272507065,
      "grad_norm": 126.34602355957031,
      "learning_rate": 1.0295698924731183e-05,
      "loss": 5.5534,
      "step": 383
    },
    {
      "epoch": 0.015502624142107388,
      "grad_norm": 174.06512451171875,
      "learning_rate": 1.0322580645161291e-05,
      "loss": 5.6117,
      "step": 384
    },
    {
      "epoch": 0.015542995559144126,
      "grad_norm": 135.4725799560547,
      "learning_rate": 1.0349462365591398e-05,
      "loss": 6.4091,
      "step": 385
    },
    {
      "epoch": 0.015583366976180863,
      "grad_norm": 154.09307861328125,
      "learning_rate": 1.0376344086021506e-05,
      "loss": 6.5918,
      "step": 386
    },
    {
      "epoch": 0.015623738393217602,
      "grad_norm": 226.48394775390625,
      "learning_rate": 1.0403225806451613e-05,
      "loss": 5.4181,
      "step": 387
    },
    {
      "epoch": 0.01566410981025434,
      "grad_norm": 265.7686767578125,
      "learning_rate": 1.0430107526881721e-05,
      "loss": 4.4822,
      "step": 388
    },
    {
      "epoch": 0.015704481227291078,
      "grad_norm": 231.7173614501953,
      "learning_rate": 1.0456989247311828e-05,
      "loss": 6.3022,
      "step": 389
    },
    {
      "epoch": 0.015744852644327817,
      "grad_norm": 217.4002227783203,
      "learning_rate": 1.0483870967741936e-05,
      "loss": 7.0061,
      "step": 390
    },
    {
      "epoch": 0.015785224061364553,
      "grad_norm": 123.57698822021484,
      "learning_rate": 1.0510752688172043e-05,
      "loss": 6.5164,
      "step": 391
    },
    {
      "epoch": 0.015825595478401292,
      "grad_norm": 175.33197021484375,
      "learning_rate": 1.0537634408602151e-05,
      "loss": 6.3971,
      "step": 392
    },
    {
      "epoch": 0.01586596689543803,
      "grad_norm": 162.5196075439453,
      "learning_rate": 1.0564516129032258e-05,
      "loss": 5.9727,
      "step": 393
    },
    {
      "epoch": 0.015906338312474767,
      "grad_norm": 110.57746124267578,
      "learning_rate": 1.0591397849462367e-05,
      "loss": 5.9133,
      "step": 394
    },
    {
      "epoch": 0.015946709729511507,
      "grad_norm": 130.94720458984375,
      "learning_rate": 1.0618279569892473e-05,
      "loss": 5.3269,
      "step": 395
    },
    {
      "epoch": 0.015987081146548242,
      "grad_norm": 121.32113647460938,
      "learning_rate": 1.0645161290322582e-05,
      "loss": 7.2055,
      "step": 396
    },
    {
      "epoch": 0.016027452563584982,
      "grad_norm": 154.0959014892578,
      "learning_rate": 1.0672043010752688e-05,
      "loss": 6.6666,
      "step": 397
    },
    {
      "epoch": 0.01606782398062172,
      "grad_norm": 149.89096069335938,
      "learning_rate": 1.0698924731182797e-05,
      "loss": 6.5601,
      "step": 398
    },
    {
      "epoch": 0.016108195397658457,
      "grad_norm": 183.6191864013672,
      "learning_rate": 1.0725806451612903e-05,
      "loss": 5.1799,
      "step": 399
    },
    {
      "epoch": 0.016148566814695196,
      "grad_norm": 262.8418884277344,
      "learning_rate": 1.0752688172043012e-05,
      "loss": 5.7721,
      "step": 400
    },
    {
      "epoch": 0.016188938231731932,
      "grad_norm": 150.2283935546875,
      "learning_rate": 1.0779569892473118e-05,
      "loss": 6.1774,
      "step": 401
    },
    {
      "epoch": 0.01622930964876867,
      "grad_norm": 156.897216796875,
      "learning_rate": 1.0806451612903225e-05,
      "loss": 6.8585,
      "step": 402
    },
    {
      "epoch": 0.01626968106580541,
      "grad_norm": 175.2357940673828,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 5.2879,
      "step": 403
    },
    {
      "epoch": 0.016310052482842147,
      "grad_norm": 111.75479125976562,
      "learning_rate": 1.086021505376344e-05,
      "loss": 6.1234,
      "step": 404
    },
    {
      "epoch": 0.016350423899878886,
      "grad_norm": 132.54571533203125,
      "learning_rate": 1.0887096774193549e-05,
      "loss": 6.5281,
      "step": 405
    },
    {
      "epoch": 0.016390795316915625,
      "grad_norm": 225.41932678222656,
      "learning_rate": 1.0913978494623655e-05,
      "loss": 6.0382,
      "step": 406
    },
    {
      "epoch": 0.01643116673395236,
      "grad_norm": 133.85804748535156,
      "learning_rate": 1.0940860215053764e-05,
      "loss": 4.7663,
      "step": 407
    },
    {
      "epoch": 0.0164715381509891,
      "grad_norm": 200.77182006835938,
      "learning_rate": 1.096774193548387e-05,
      "loss": 6.2296,
      "step": 408
    },
    {
      "epoch": 0.016511909568025836,
      "grad_norm": 159.60482788085938,
      "learning_rate": 1.0994623655913979e-05,
      "loss": 5.5171,
      "step": 409
    },
    {
      "epoch": 0.016552280985062576,
      "grad_norm": 128.7032012939453,
      "learning_rate": 1.1021505376344085e-05,
      "loss": 7.7746,
      "step": 410
    },
    {
      "epoch": 0.016592652402099315,
      "grad_norm": 138.5675811767578,
      "learning_rate": 1.1048387096774194e-05,
      "loss": 6.232,
      "step": 411
    },
    {
      "epoch": 0.01663302381913605,
      "grad_norm": 133.53749084472656,
      "learning_rate": 1.10752688172043e-05,
      "loss": 5.9334,
      "step": 412
    },
    {
      "epoch": 0.01667339523617279,
      "grad_norm": 198.6153564453125,
      "learning_rate": 1.1102150537634409e-05,
      "loss": 4.504,
      "step": 413
    },
    {
      "epoch": 0.016713766653209526,
      "grad_norm": 129.776123046875,
      "learning_rate": 1.1129032258064516e-05,
      "loss": 7.0575,
      "step": 414
    },
    {
      "epoch": 0.016754138070246265,
      "grad_norm": 126.55648803710938,
      "learning_rate": 1.1155913978494624e-05,
      "loss": 7.5038,
      "step": 415
    },
    {
      "epoch": 0.016794509487283005,
      "grad_norm": 108.5566177368164,
      "learning_rate": 1.118279569892473e-05,
      "loss": 5.2583,
      "step": 416
    },
    {
      "epoch": 0.01683488090431974,
      "grad_norm": 154.91139221191406,
      "learning_rate": 1.1209677419354839e-05,
      "loss": 5.5951,
      "step": 417
    },
    {
      "epoch": 0.01687525232135648,
      "grad_norm": 193.0041046142578,
      "learning_rate": 1.1236559139784946e-05,
      "loss": 5.1785,
      "step": 418
    },
    {
      "epoch": 0.01691562373839322,
      "grad_norm": 189.43017578125,
      "learning_rate": 1.1263440860215056e-05,
      "loss": 4.9217,
      "step": 419
    },
    {
      "epoch": 0.016955995155429955,
      "grad_norm": 157.4368133544922,
      "learning_rate": 1.1290322580645164e-05,
      "loss": 6.3803,
      "step": 420
    },
    {
      "epoch": 0.016996366572466694,
      "grad_norm": 158.2810516357422,
      "learning_rate": 1.131720430107527e-05,
      "loss": 3.6114,
      "step": 421
    },
    {
      "epoch": 0.01703673798950343,
      "grad_norm": 155.56512451171875,
      "learning_rate": 1.1344086021505377e-05,
      "loss": 7.0304,
      "step": 422
    },
    {
      "epoch": 0.01707710940654017,
      "grad_norm": 179.9307098388672,
      "learning_rate": 1.1370967741935486e-05,
      "loss": 6.0151,
      "step": 423
    },
    {
      "epoch": 0.01711748082357691,
      "grad_norm": 149.67184448242188,
      "learning_rate": 1.1397849462365593e-05,
      "loss": 5.0542,
      "step": 424
    },
    {
      "epoch": 0.017157852240613645,
      "grad_norm": 145.65126037597656,
      "learning_rate": 1.1424731182795701e-05,
      "loss": 6.8834,
      "step": 425
    },
    {
      "epoch": 0.017198223657650384,
      "grad_norm": 165.5321044921875,
      "learning_rate": 1.1451612903225808e-05,
      "loss": 6.5278,
      "step": 426
    },
    {
      "epoch": 0.01723859507468712,
      "grad_norm": 138.70742797851562,
      "learning_rate": 1.1478494623655916e-05,
      "loss": 5.0732,
      "step": 427
    },
    {
      "epoch": 0.01727896649172386,
      "grad_norm": 157.5403594970703,
      "learning_rate": 1.1505376344086023e-05,
      "loss": 6.5279,
      "step": 428
    },
    {
      "epoch": 0.0173193379087606,
      "grad_norm": 205.47296142578125,
      "learning_rate": 1.1532258064516131e-05,
      "loss": 6.5384,
      "step": 429
    },
    {
      "epoch": 0.017359709325797335,
      "grad_norm": 215.49832153320312,
      "learning_rate": 1.1559139784946238e-05,
      "loss": 5.6662,
      "step": 430
    },
    {
      "epoch": 0.017400080742834074,
      "grad_norm": 216.32102966308594,
      "learning_rate": 1.1586021505376346e-05,
      "loss": 5.6082,
      "step": 431
    },
    {
      "epoch": 0.017440452159870813,
      "grad_norm": 195.87660217285156,
      "learning_rate": 1.1612903225806453e-05,
      "loss": 5.4258,
      "step": 432
    },
    {
      "epoch": 0.01748082357690755,
      "grad_norm": 202.0609588623047,
      "learning_rate": 1.1639784946236561e-05,
      "loss": 6.247,
      "step": 433
    },
    {
      "epoch": 0.01752119499394429,
      "grad_norm": 268.0738525390625,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 5.5087,
      "step": 434
    },
    {
      "epoch": 0.017561566410981024,
      "grad_norm": 127.97981262207031,
      "learning_rate": 1.1693548387096776e-05,
      "loss": 4.0958,
      "step": 435
    },
    {
      "epoch": 0.017601937828017764,
      "grad_norm": 203.92222595214844,
      "learning_rate": 1.1720430107526883e-05,
      "loss": 6.7371,
      "step": 436
    },
    {
      "epoch": 0.017642309245054503,
      "grad_norm": 176.6006317138672,
      "learning_rate": 1.1747311827956991e-05,
      "loss": 6.4163,
      "step": 437
    },
    {
      "epoch": 0.01768268066209124,
      "grad_norm": 178.9433135986328,
      "learning_rate": 1.1774193548387098e-05,
      "loss": 5.6969,
      "step": 438
    },
    {
      "epoch": 0.017723052079127978,
      "grad_norm": 165.15541076660156,
      "learning_rate": 1.1801075268817206e-05,
      "loss": 7.3538,
      "step": 439
    },
    {
      "epoch": 0.017763423496164714,
      "grad_norm": 130.01419067382812,
      "learning_rate": 1.1827956989247313e-05,
      "loss": 5.8468,
      "step": 440
    },
    {
      "epoch": 0.017803794913201453,
      "grad_norm": 140.08682250976562,
      "learning_rate": 1.1854838709677421e-05,
      "loss": 6.5653,
      "step": 441
    },
    {
      "epoch": 0.017844166330238193,
      "grad_norm": 129.1420135498047,
      "learning_rate": 1.1881720430107528e-05,
      "loss": 5.2571,
      "step": 442
    },
    {
      "epoch": 0.01788453774727493,
      "grad_norm": 121.18665313720703,
      "learning_rate": 1.1908602150537637e-05,
      "loss": 5.9724,
      "step": 443
    },
    {
      "epoch": 0.017924909164311668,
      "grad_norm": 123.23056030273438,
      "learning_rate": 1.1935483870967743e-05,
      "loss": 6.1949,
      "step": 444
    },
    {
      "epoch": 0.017965280581348407,
      "grad_norm": 233.09861755371094,
      "learning_rate": 1.1962365591397852e-05,
      "loss": 6.3752,
      "step": 445
    },
    {
      "epoch": 0.018005651998385143,
      "grad_norm": 160.4254913330078,
      "learning_rate": 1.1989247311827958e-05,
      "loss": 6.3534,
      "step": 446
    },
    {
      "epoch": 0.018046023415421882,
      "grad_norm": 154.3227081298828,
      "learning_rate": 1.2016129032258067e-05,
      "loss": 6.0704,
      "step": 447
    },
    {
      "epoch": 0.018086394832458618,
      "grad_norm": 157.4697265625,
      "learning_rate": 1.2043010752688173e-05,
      "loss": 5.7853,
      "step": 448
    },
    {
      "epoch": 0.018126766249495357,
      "grad_norm": 183.89439392089844,
      "learning_rate": 1.2069892473118282e-05,
      "loss": 4.654,
      "step": 449
    },
    {
      "epoch": 0.018167137666532097,
      "grad_norm": 136.08474731445312,
      "learning_rate": 1.2096774193548388e-05,
      "loss": 4.9536,
      "step": 450
    },
    {
      "epoch": 0.018207509083568833,
      "grad_norm": 162.51724243164062,
      "learning_rate": 1.2123655913978495e-05,
      "loss": 6.9463,
      "step": 451
    },
    {
      "epoch": 0.018247880500605572,
      "grad_norm": 140.1771240234375,
      "learning_rate": 1.2150537634408604e-05,
      "loss": 6.09,
      "step": 452
    },
    {
      "epoch": 0.018288251917642308,
      "grad_norm": 158.595947265625,
      "learning_rate": 1.217741935483871e-05,
      "loss": 5.8169,
      "step": 453
    },
    {
      "epoch": 0.018328623334679047,
      "grad_norm": 154.57568359375,
      "learning_rate": 1.2204301075268819e-05,
      "loss": 6.1423,
      "step": 454
    },
    {
      "epoch": 0.018368994751715786,
      "grad_norm": 143.8892822265625,
      "learning_rate": 1.2231182795698925e-05,
      "loss": 6.5925,
      "step": 455
    },
    {
      "epoch": 0.018409366168752522,
      "grad_norm": 132.10540771484375,
      "learning_rate": 1.2258064516129034e-05,
      "loss": 6.8232,
      "step": 456
    },
    {
      "epoch": 0.01844973758578926,
      "grad_norm": 141.7972412109375,
      "learning_rate": 1.228494623655914e-05,
      "loss": 5.5431,
      "step": 457
    },
    {
      "epoch": 0.018490109002825997,
      "grad_norm": 222.54046630859375,
      "learning_rate": 1.2311827956989249e-05,
      "loss": 5.2549,
      "step": 458
    },
    {
      "epoch": 0.018530480419862737,
      "grad_norm": 165.13946533203125,
      "learning_rate": 1.2338709677419355e-05,
      "loss": 5.2477,
      "step": 459
    },
    {
      "epoch": 0.018570851836899476,
      "grad_norm": 116.0902099609375,
      "learning_rate": 1.2365591397849464e-05,
      "loss": 7.0897,
      "step": 460
    },
    {
      "epoch": 0.018611223253936212,
      "grad_norm": 147.14744567871094,
      "learning_rate": 1.239247311827957e-05,
      "loss": 6.372,
      "step": 461
    },
    {
      "epoch": 0.01865159467097295,
      "grad_norm": 177.95851135253906,
      "learning_rate": 1.2419354838709679e-05,
      "loss": 5.5598,
      "step": 462
    },
    {
      "epoch": 0.01869196608800969,
      "grad_norm": 162.2525634765625,
      "learning_rate": 1.2446236559139786e-05,
      "loss": 6.4187,
      "step": 463
    },
    {
      "epoch": 0.018732337505046427,
      "grad_norm": 173.90838623046875,
      "learning_rate": 1.2473118279569894e-05,
      "loss": 5.659,
      "step": 464
    },
    {
      "epoch": 0.018772708922083166,
      "grad_norm": 130.22509765625,
      "learning_rate": 1.25e-05,
      "loss": 6.7628,
      "step": 465
    },
    {
      "epoch": 0.0188130803391199,
      "grad_norm": 180.35108947753906,
      "learning_rate": 1.2526881720430109e-05,
      "loss": 6.4357,
      "step": 466
    },
    {
      "epoch": 0.01885345175615664,
      "grad_norm": 145.4464111328125,
      "learning_rate": 1.2553763440860216e-05,
      "loss": 4.5287,
      "step": 467
    },
    {
      "epoch": 0.01889382317319338,
      "grad_norm": 190.9717559814453,
      "learning_rate": 1.2580645161290324e-05,
      "loss": 5.2449,
      "step": 468
    },
    {
      "epoch": 0.018934194590230116,
      "grad_norm": 340.444091796875,
      "learning_rate": 1.260752688172043e-05,
      "loss": 5.6367,
      "step": 469
    },
    {
      "epoch": 0.018974566007266856,
      "grad_norm": 134.67849731445312,
      "learning_rate": 1.2634408602150539e-05,
      "loss": 5.9239,
      "step": 470
    },
    {
      "epoch": 0.01901493742430359,
      "grad_norm": 157.6857147216797,
      "learning_rate": 1.2661290322580646e-05,
      "loss": 4.8323,
      "step": 471
    },
    {
      "epoch": 0.01905530884134033,
      "grad_norm": 336.7962951660156,
      "learning_rate": 1.2688172043010754e-05,
      "loss": 5.886,
      "step": 472
    },
    {
      "epoch": 0.01909568025837707,
      "grad_norm": 147.91932678222656,
      "learning_rate": 1.2715053763440861e-05,
      "loss": 4.1127,
      "step": 473
    },
    {
      "epoch": 0.019136051675413806,
      "grad_norm": 125.0755615234375,
      "learning_rate": 1.274193548387097e-05,
      "loss": 5.6105,
      "step": 474
    },
    {
      "epoch": 0.019176423092450545,
      "grad_norm": 154.40176391601562,
      "learning_rate": 1.2768817204301076e-05,
      "loss": 6.1048,
      "step": 475
    },
    {
      "epoch": 0.019216794509487285,
      "grad_norm": 164.03172302246094,
      "learning_rate": 1.2795698924731184e-05,
      "loss": 5.2777,
      "step": 476
    },
    {
      "epoch": 0.01925716592652402,
      "grad_norm": 167.60574340820312,
      "learning_rate": 1.2822580645161291e-05,
      "loss": 5.5388,
      "step": 477
    },
    {
      "epoch": 0.01929753734356076,
      "grad_norm": 153.51268005371094,
      "learning_rate": 1.28494623655914e-05,
      "loss": 5.8657,
      "step": 478
    },
    {
      "epoch": 0.019337908760597496,
      "grad_norm": 222.043701171875,
      "learning_rate": 1.2876344086021506e-05,
      "loss": 4.7873,
      "step": 479
    },
    {
      "epoch": 0.019378280177634235,
      "grad_norm": 160.37368774414062,
      "learning_rate": 1.2903225806451613e-05,
      "loss": 4.1442,
      "step": 480
    },
    {
      "epoch": 0.019418651594670974,
      "grad_norm": 164.6867218017578,
      "learning_rate": 1.2930107526881721e-05,
      "loss": 5.8106,
      "step": 481
    },
    {
      "epoch": 0.01945902301170771,
      "grad_norm": 173.81631469726562,
      "learning_rate": 1.2956989247311828e-05,
      "loss": 4.9017,
      "step": 482
    },
    {
      "epoch": 0.01949939442874445,
      "grad_norm": 252.2703399658203,
      "learning_rate": 1.2983870967741936e-05,
      "loss": 4.151,
      "step": 483
    },
    {
      "epoch": 0.019539765845781185,
      "grad_norm": 429.1289978027344,
      "learning_rate": 1.3010752688172043e-05,
      "loss": 6.8413,
      "step": 484
    },
    {
      "epoch": 0.019580137262817925,
      "grad_norm": 152.2969512939453,
      "learning_rate": 1.3037634408602151e-05,
      "loss": 6.5504,
      "step": 485
    },
    {
      "epoch": 0.019620508679854664,
      "grad_norm": 149.96685791015625,
      "learning_rate": 1.3064516129032258e-05,
      "loss": 6.0544,
      "step": 486
    },
    {
      "epoch": 0.0196608800968914,
      "grad_norm": 229.15603637695312,
      "learning_rate": 1.3091397849462366e-05,
      "loss": 7.7075,
      "step": 487
    },
    {
      "epoch": 0.01970125151392814,
      "grad_norm": 146.4707794189453,
      "learning_rate": 1.3118279569892473e-05,
      "loss": 4.1494,
      "step": 488
    },
    {
      "epoch": 0.01974162293096488,
      "grad_norm": 222.1854705810547,
      "learning_rate": 1.3145161290322581e-05,
      "loss": 6.6263,
      "step": 489
    },
    {
      "epoch": 0.019781994348001614,
      "grad_norm": 128.1598663330078,
      "learning_rate": 1.3172043010752688e-05,
      "loss": 6.0056,
      "step": 490
    },
    {
      "epoch": 0.019822365765038354,
      "grad_norm": 147.78683471679688,
      "learning_rate": 1.3198924731182796e-05,
      "loss": 5.6854,
      "step": 491
    },
    {
      "epoch": 0.01986273718207509,
      "grad_norm": 148.38124084472656,
      "learning_rate": 1.3225806451612903e-05,
      "loss": 5.6243,
      "step": 492
    },
    {
      "epoch": 0.01990310859911183,
      "grad_norm": 150.5673065185547,
      "learning_rate": 1.3252688172043012e-05,
      "loss": 5.4373,
      "step": 493
    },
    {
      "epoch": 0.019943480016148568,
      "grad_norm": 169.80259704589844,
      "learning_rate": 1.3279569892473118e-05,
      "loss": 6.027,
      "step": 494
    },
    {
      "epoch": 0.019983851433185304,
      "grad_norm": 181.9217987060547,
      "learning_rate": 1.3306451612903227e-05,
      "loss": 7.0961,
      "step": 495
    },
    {
      "epoch": 0.020024222850222043,
      "grad_norm": 285.10821533203125,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 5.4508,
      "step": 496
    },
    {
      "epoch": 0.02006459426725878,
      "grad_norm": 228.27134704589844,
      "learning_rate": 1.3360215053763442e-05,
      "loss": 6.6832,
      "step": 497
    },
    {
      "epoch": 0.02010496568429552,
      "grad_norm": 138.31492614746094,
      "learning_rate": 1.3387096774193548e-05,
      "loss": 5.5482,
      "step": 498
    },
    {
      "epoch": 0.020145337101332258,
      "grad_norm": 131.50665283203125,
      "learning_rate": 1.3413978494623657e-05,
      "loss": 5.7848,
      "step": 499
    },
    {
      "epoch": 0.020185708518368994,
      "grad_norm": 210.8549346923828,
      "learning_rate": 1.3440860215053763e-05,
      "loss": 4.5027,
      "step": 500
    },
    {
      "epoch": 0.020226079935405733,
      "grad_norm": 148.09466552734375,
      "learning_rate": 1.3467741935483872e-05,
      "loss": 6.4352,
      "step": 501
    },
    {
      "epoch": 0.020266451352442472,
      "grad_norm": 211.22610473632812,
      "learning_rate": 1.3494623655913978e-05,
      "loss": 6.44,
      "step": 502
    },
    {
      "epoch": 0.020306822769479208,
      "grad_norm": 170.28091430664062,
      "learning_rate": 1.3521505376344087e-05,
      "loss": 4.5521,
      "step": 503
    },
    {
      "epoch": 0.020347194186515948,
      "grad_norm": 134.76600646972656,
      "learning_rate": 1.3548387096774194e-05,
      "loss": 6.234,
      "step": 504
    },
    {
      "epoch": 0.020387565603552683,
      "grad_norm": 153.96043395996094,
      "learning_rate": 1.3575268817204302e-05,
      "loss": 5.6811,
      "step": 505
    },
    {
      "epoch": 0.020427937020589423,
      "grad_norm": 142.06321716308594,
      "learning_rate": 1.3602150537634409e-05,
      "loss": 4.7508,
      "step": 506
    },
    {
      "epoch": 0.020468308437626162,
      "grad_norm": 152.0278778076172,
      "learning_rate": 1.3629032258064517e-05,
      "loss": 5.9227,
      "step": 507
    },
    {
      "epoch": 0.020508679854662898,
      "grad_norm": 164.1175537109375,
      "learning_rate": 1.3655913978494624e-05,
      "loss": 7.4278,
      "step": 508
    },
    {
      "epoch": 0.020549051271699637,
      "grad_norm": 112.88445281982422,
      "learning_rate": 1.3682795698924732e-05,
      "loss": 7.6661,
      "step": 509
    },
    {
      "epoch": 0.020589422688736373,
      "grad_norm": 156.57691955566406,
      "learning_rate": 1.3709677419354839e-05,
      "loss": 6.7882,
      "step": 510
    },
    {
      "epoch": 0.020629794105773112,
      "grad_norm": 189.7261199951172,
      "learning_rate": 1.3736559139784945e-05,
      "loss": 4.2859,
      "step": 511
    },
    {
      "epoch": 0.02067016552280985,
      "grad_norm": 226.26979064941406,
      "learning_rate": 1.3763440860215056e-05,
      "loss": 5.9558,
      "step": 512
    },
    {
      "epoch": 0.020710536939846588,
      "grad_norm": 121.927734375,
      "learning_rate": 1.3790322580645164e-05,
      "loss": 5.7444,
      "step": 513
    },
    {
      "epoch": 0.020750908356883327,
      "grad_norm": 130.47793579101562,
      "learning_rate": 1.381720430107527e-05,
      "loss": 4.7384,
      "step": 514
    },
    {
      "epoch": 0.020791279773920066,
      "grad_norm": 113.74285888671875,
      "learning_rate": 1.3844086021505379e-05,
      "loss": 6.8356,
      "step": 515
    },
    {
      "epoch": 0.020831651190956802,
      "grad_norm": 134.41220092773438,
      "learning_rate": 1.3870967741935486e-05,
      "loss": 5.5465,
      "step": 516
    },
    {
      "epoch": 0.02087202260799354,
      "grad_norm": 139.87892150878906,
      "learning_rate": 1.3897849462365594e-05,
      "loss": 6.2675,
      "step": 517
    },
    {
      "epoch": 0.020912394025030277,
      "grad_norm": 109.64351654052734,
      "learning_rate": 1.39247311827957e-05,
      "loss": 6.7516,
      "step": 518
    },
    {
      "epoch": 0.020952765442067017,
      "grad_norm": 171.494384765625,
      "learning_rate": 1.3951612903225809e-05,
      "loss": 6.4163,
      "step": 519
    },
    {
      "epoch": 0.020993136859103756,
      "grad_norm": 117.0684585571289,
      "learning_rate": 1.3978494623655916e-05,
      "loss": 4.3163,
      "step": 520
    },
    {
      "epoch": 0.021033508276140492,
      "grad_norm": 107.31255340576172,
      "learning_rate": 1.4005376344086024e-05,
      "loss": 5.4907,
      "step": 521
    },
    {
      "epoch": 0.02107387969317723,
      "grad_norm": 172.38150024414062,
      "learning_rate": 1.4032258064516131e-05,
      "loss": 5.8591,
      "step": 522
    },
    {
      "epoch": 0.021114251110213967,
      "grad_norm": 149.27978515625,
      "learning_rate": 1.405913978494624e-05,
      "loss": 5.7178,
      "step": 523
    },
    {
      "epoch": 0.021154622527250706,
      "grad_norm": 157.1072235107422,
      "learning_rate": 1.4086021505376346e-05,
      "loss": 5.443,
      "step": 524
    },
    {
      "epoch": 0.021194993944287446,
      "grad_norm": 122.15425872802734,
      "learning_rate": 1.4112903225806454e-05,
      "loss": 6.9627,
      "step": 525
    },
    {
      "epoch": 0.02123536536132418,
      "grad_norm": 182.72010803222656,
      "learning_rate": 1.4139784946236561e-05,
      "loss": 6.5084,
      "step": 526
    },
    {
      "epoch": 0.02127573677836092,
      "grad_norm": 131.994384765625,
      "learning_rate": 1.416666666666667e-05,
      "loss": 5.9253,
      "step": 527
    },
    {
      "epoch": 0.02131610819539766,
      "grad_norm": 175.42507934570312,
      "learning_rate": 1.4193548387096776e-05,
      "loss": 5.583,
      "step": 528
    },
    {
      "epoch": 0.021356479612434396,
      "grad_norm": 147.2372589111328,
      "learning_rate": 1.4220430107526883e-05,
      "loss": 4.5025,
      "step": 529
    },
    {
      "epoch": 0.021396851029471135,
      "grad_norm": 141.43814086914062,
      "learning_rate": 1.4247311827956991e-05,
      "loss": 6.7991,
      "step": 530
    },
    {
      "epoch": 0.02143722244650787,
      "grad_norm": 161.80357360839844,
      "learning_rate": 1.4274193548387098e-05,
      "loss": 6.3557,
      "step": 531
    },
    {
      "epoch": 0.02147759386354461,
      "grad_norm": 138.95582580566406,
      "learning_rate": 1.4301075268817206e-05,
      "loss": 6.1138,
      "step": 532
    },
    {
      "epoch": 0.02151796528058135,
      "grad_norm": 145.18211364746094,
      "learning_rate": 1.4327956989247313e-05,
      "loss": 6.6276,
      "step": 533
    },
    {
      "epoch": 0.021558336697618086,
      "grad_norm": 155.25843811035156,
      "learning_rate": 1.4354838709677421e-05,
      "loss": 6.294,
      "step": 534
    },
    {
      "epoch": 0.021598708114654825,
      "grad_norm": 151.62864685058594,
      "learning_rate": 1.4381720430107528e-05,
      "loss": 6.177,
      "step": 535
    },
    {
      "epoch": 0.02163907953169156,
      "grad_norm": 137.05343627929688,
      "learning_rate": 1.4408602150537636e-05,
      "loss": 6.5441,
      "step": 536
    },
    {
      "epoch": 0.0216794509487283,
      "grad_norm": 126.61038208007812,
      "learning_rate": 1.4435483870967743e-05,
      "loss": 6.7186,
      "step": 537
    },
    {
      "epoch": 0.02171982236576504,
      "grad_norm": 200.427490234375,
      "learning_rate": 1.4462365591397851e-05,
      "loss": 5.4796,
      "step": 538
    },
    {
      "epoch": 0.021760193782801775,
      "grad_norm": 185.020263671875,
      "learning_rate": 1.4489247311827958e-05,
      "loss": 5.726,
      "step": 539
    },
    {
      "epoch": 0.021800565199838515,
      "grad_norm": 164.3673553466797,
      "learning_rate": 1.4516129032258066e-05,
      "loss": 5.3388,
      "step": 540
    },
    {
      "epoch": 0.021840936616875254,
      "grad_norm": 180.66831970214844,
      "learning_rate": 1.4543010752688173e-05,
      "loss": 6.1209,
      "step": 541
    },
    {
      "epoch": 0.02188130803391199,
      "grad_norm": 175.90594482421875,
      "learning_rate": 1.4569892473118282e-05,
      "loss": 5.4979,
      "step": 542
    },
    {
      "epoch": 0.02192167945094873,
      "grad_norm": 165.7836456298828,
      "learning_rate": 1.4596774193548388e-05,
      "loss": 5.0768,
      "step": 543
    },
    {
      "epoch": 0.021962050867985465,
      "grad_norm": 207.5694122314453,
      "learning_rate": 1.4623655913978497e-05,
      "loss": 5.6251,
      "step": 544
    },
    {
      "epoch": 0.022002422285022204,
      "grad_norm": 135.43511962890625,
      "learning_rate": 1.4650537634408603e-05,
      "loss": 6.3902,
      "step": 545
    },
    {
      "epoch": 0.022042793702058944,
      "grad_norm": 125.83143615722656,
      "learning_rate": 1.4677419354838712e-05,
      "loss": 6.404,
      "step": 546
    },
    {
      "epoch": 0.02208316511909568,
      "grad_norm": 153.85035705566406,
      "learning_rate": 1.4704301075268818e-05,
      "loss": 6.3865,
      "step": 547
    },
    {
      "epoch": 0.02212353653613242,
      "grad_norm": 211.71714782714844,
      "learning_rate": 1.4731182795698927e-05,
      "loss": 6.5082,
      "step": 548
    },
    {
      "epoch": 0.022163907953169155,
      "grad_norm": 145.03741455078125,
      "learning_rate": 1.4758064516129033e-05,
      "loss": 6.4733,
      "step": 549
    },
    {
      "epoch": 0.022204279370205894,
      "grad_norm": 124.92852020263672,
      "learning_rate": 1.4784946236559142e-05,
      "loss": 6.1846,
      "step": 550
    },
    {
      "epoch": 0.022244650787242633,
      "grad_norm": 143.94488525390625,
      "learning_rate": 1.4811827956989248e-05,
      "loss": 5.5923,
      "step": 551
    },
    {
      "epoch": 0.02228502220427937,
      "grad_norm": 154.1079559326172,
      "learning_rate": 1.4838709677419357e-05,
      "loss": 6.7173,
      "step": 552
    },
    {
      "epoch": 0.02232539362131611,
      "grad_norm": 151.94642639160156,
      "learning_rate": 1.4865591397849464e-05,
      "loss": 6.9467,
      "step": 553
    },
    {
      "epoch": 0.022365765038352844,
      "grad_norm": 137.62083435058594,
      "learning_rate": 1.4892473118279572e-05,
      "loss": 5.8918,
      "step": 554
    },
    {
      "epoch": 0.022406136455389584,
      "grad_norm": 142.54074096679688,
      "learning_rate": 1.4919354838709679e-05,
      "loss": 5.9434,
      "step": 555
    },
    {
      "epoch": 0.022446507872426323,
      "grad_norm": 183.89808654785156,
      "learning_rate": 1.4946236559139787e-05,
      "loss": 5.8082,
      "step": 556
    },
    {
      "epoch": 0.02248687928946306,
      "grad_norm": 177.196533203125,
      "learning_rate": 1.4973118279569894e-05,
      "loss": 7.2252,
      "step": 557
    },
    {
      "epoch": 0.0225272507064998,
      "grad_norm": 236.2783966064453,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 6.3243,
      "step": 558
    },
    {
      "epoch": 0.022567622123536538,
      "grad_norm": 184.65428161621094,
      "learning_rate": 1.5026881720430109e-05,
      "loss": 7.0505,
      "step": 559
    },
    {
      "epoch": 0.022607993540573273,
      "grad_norm": 185.6514129638672,
      "learning_rate": 1.5053763440860215e-05,
      "loss": 5.366,
      "step": 560
    },
    {
      "epoch": 0.022648364957610013,
      "grad_norm": 168.14295959472656,
      "learning_rate": 1.5080645161290324e-05,
      "loss": 5.3689,
      "step": 561
    },
    {
      "epoch": 0.02268873637464675,
      "grad_norm": 2504.314697265625,
      "learning_rate": 1.510752688172043e-05,
      "loss": 5.6775,
      "step": 562
    },
    {
      "epoch": 0.022729107791683488,
      "grad_norm": 119.32272338867188,
      "learning_rate": 1.5134408602150539e-05,
      "loss": 5.8151,
      "step": 563
    },
    {
      "epoch": 0.022769479208720227,
      "grad_norm": 123.15681457519531,
      "learning_rate": 1.5161290322580646e-05,
      "loss": 6.0318,
      "step": 564
    },
    {
      "epoch": 0.022809850625756963,
      "grad_norm": 123.29806518554688,
      "learning_rate": 1.5188172043010754e-05,
      "loss": 5.968,
      "step": 565
    },
    {
      "epoch": 0.022850222042793703,
      "grad_norm": 150.88287353515625,
      "learning_rate": 1.521505376344086e-05,
      "loss": 5.7258,
      "step": 566
    },
    {
      "epoch": 0.02289059345983044,
      "grad_norm": 153.10948181152344,
      "learning_rate": 1.5241935483870969e-05,
      "loss": 5.5669,
      "step": 567
    },
    {
      "epoch": 0.022930964876867178,
      "grad_norm": 156.1638641357422,
      "learning_rate": 1.5268817204301076e-05,
      "loss": 6.2768,
      "step": 568
    },
    {
      "epoch": 0.022971336293903917,
      "grad_norm": 109.63343811035156,
      "learning_rate": 1.5295698924731182e-05,
      "loss": 6.2372,
      "step": 569
    },
    {
      "epoch": 0.023011707710940653,
      "grad_norm": 263.1455383300781,
      "learning_rate": 1.5322580645161292e-05,
      "loss": 6.7824,
      "step": 570
    },
    {
      "epoch": 0.023052079127977392,
      "grad_norm": 177.3002166748047,
      "learning_rate": 1.53494623655914e-05,
      "loss": 5.6733,
      "step": 571
    },
    {
      "epoch": 0.02309245054501413,
      "grad_norm": 102.47732543945312,
      "learning_rate": 1.5376344086021506e-05,
      "loss": 5.7081,
      "step": 572
    },
    {
      "epoch": 0.023132821962050867,
      "grad_norm": 174.24903869628906,
      "learning_rate": 1.5403225806451613e-05,
      "loss": 5.6392,
      "step": 573
    },
    {
      "epoch": 0.023173193379087607,
      "grad_norm": 132.34539794921875,
      "learning_rate": 1.5430107526881723e-05,
      "loss": 6.3288,
      "step": 574
    },
    {
      "epoch": 0.023213564796124343,
      "grad_norm": 136.71127319335938,
      "learning_rate": 1.545698924731183e-05,
      "loss": 5.7994,
      "step": 575
    },
    {
      "epoch": 0.023253936213161082,
      "grad_norm": 150.71095275878906,
      "learning_rate": 1.5483870967741936e-05,
      "loss": 6.5258,
      "step": 576
    },
    {
      "epoch": 0.02329430763019782,
      "grad_norm": 162.30943298339844,
      "learning_rate": 1.5510752688172043e-05,
      "loss": 5.8379,
      "step": 577
    },
    {
      "epoch": 0.023334679047234557,
      "grad_norm": 116.65691375732422,
      "learning_rate": 1.5537634408602153e-05,
      "loss": 6.054,
      "step": 578
    },
    {
      "epoch": 0.023375050464271296,
      "grad_norm": 177.6289520263672,
      "learning_rate": 1.556451612903226e-05,
      "loss": 5.8119,
      "step": 579
    },
    {
      "epoch": 0.023415421881308032,
      "grad_norm": 131.56021118164062,
      "learning_rate": 1.5591397849462366e-05,
      "loss": 4.8685,
      "step": 580
    },
    {
      "epoch": 0.02345579329834477,
      "grad_norm": 131.4651336669922,
      "learning_rate": 1.5618279569892473e-05,
      "loss": 5.3127,
      "step": 581
    },
    {
      "epoch": 0.02349616471538151,
      "grad_norm": 244.65628051757812,
      "learning_rate": 1.5645161290322583e-05,
      "loss": 6.0902,
      "step": 582
    },
    {
      "epoch": 0.023536536132418247,
      "grad_norm": 184.0985565185547,
      "learning_rate": 1.567204301075269e-05,
      "loss": 5.9578,
      "step": 583
    },
    {
      "epoch": 0.023576907549454986,
      "grad_norm": 377.12677001953125,
      "learning_rate": 1.5698924731182796e-05,
      "loss": 6.5736,
      "step": 584
    },
    {
      "epoch": 0.023617278966491725,
      "grad_norm": 182.83245849609375,
      "learning_rate": 1.5725806451612903e-05,
      "loss": 5.8241,
      "step": 585
    },
    {
      "epoch": 0.02365765038352846,
      "grad_norm": 126.13069915771484,
      "learning_rate": 1.5752688172043013e-05,
      "loss": 6.0728,
      "step": 586
    },
    {
      "epoch": 0.0236980218005652,
      "grad_norm": 160.25808715820312,
      "learning_rate": 1.577956989247312e-05,
      "loss": 5.5867,
      "step": 587
    },
    {
      "epoch": 0.023738393217601936,
      "grad_norm": 141.198974609375,
      "learning_rate": 1.5806451612903226e-05,
      "loss": 5.5139,
      "step": 588
    },
    {
      "epoch": 0.023778764634638676,
      "grad_norm": 115.4983139038086,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 6.7832,
      "step": 589
    },
    {
      "epoch": 0.023819136051675415,
      "grad_norm": 101.8701171875,
      "learning_rate": 1.586021505376344e-05,
      "loss": 5.8427,
      "step": 590
    },
    {
      "epoch": 0.02385950746871215,
      "grad_norm": 148.01441955566406,
      "learning_rate": 1.588709677419355e-05,
      "loss": 6.6249,
      "step": 591
    },
    {
      "epoch": 0.02389987888574889,
      "grad_norm": 99.00044250488281,
      "learning_rate": 1.5913978494623657e-05,
      "loss": 6.0496,
      "step": 592
    },
    {
      "epoch": 0.023940250302785626,
      "grad_norm": 156.13381958007812,
      "learning_rate": 1.5940860215053763e-05,
      "loss": 4.8639,
      "step": 593
    },
    {
      "epoch": 0.023980621719822365,
      "grad_norm": 98.78207397460938,
      "learning_rate": 1.596774193548387e-05,
      "loss": 4.9374,
      "step": 594
    },
    {
      "epoch": 0.024020993136859105,
      "grad_norm": 142.64834594726562,
      "learning_rate": 1.599462365591398e-05,
      "loss": 4.6922,
      "step": 595
    },
    {
      "epoch": 0.02406136455389584,
      "grad_norm": 105.46001434326172,
      "learning_rate": 1.6021505376344087e-05,
      "loss": 5.764,
      "step": 596
    },
    {
      "epoch": 0.02410173597093258,
      "grad_norm": 168.6419677734375,
      "learning_rate": 1.6048387096774193e-05,
      "loss": 6.608,
      "step": 597
    },
    {
      "epoch": 0.02414210738796932,
      "grad_norm": 137.3488006591797,
      "learning_rate": 1.60752688172043e-05,
      "loss": 7.2622,
      "step": 598
    },
    {
      "epoch": 0.024182478805006055,
      "grad_norm": 227.88780212402344,
      "learning_rate": 1.610215053763441e-05,
      "loss": 6.8922,
      "step": 599
    },
    {
      "epoch": 0.024222850222042795,
      "grad_norm": 141.9279327392578,
      "learning_rate": 1.6129032258064517e-05,
      "loss": 4.0567,
      "step": 600
    },
    {
      "epoch": 0.02426322163907953,
      "grad_norm": 196.36996459960938,
      "learning_rate": 1.6155913978494623e-05,
      "loss": 6.7318,
      "step": 601
    },
    {
      "epoch": 0.02430359305611627,
      "grad_norm": 206.89303588867188,
      "learning_rate": 1.618279569892473e-05,
      "loss": 4.7781,
      "step": 602
    },
    {
      "epoch": 0.02434396447315301,
      "grad_norm": 131.1927947998047,
      "learning_rate": 1.620967741935484e-05,
      "loss": 6.7856,
      "step": 603
    },
    {
      "epoch": 0.024384335890189745,
      "grad_norm": 189.5560760498047,
      "learning_rate": 1.6236559139784947e-05,
      "loss": 4.5687,
      "step": 604
    },
    {
      "epoch": 0.024424707307226484,
      "grad_norm": 154.151611328125,
      "learning_rate": 1.6263440860215057e-05,
      "loss": 5.23,
      "step": 605
    },
    {
      "epoch": 0.02446507872426322,
      "grad_norm": 157.82769775390625,
      "learning_rate": 1.6290322580645164e-05,
      "loss": 6.1203,
      "step": 606
    },
    {
      "epoch": 0.02450545014129996,
      "grad_norm": 105.3533935546875,
      "learning_rate": 1.631720430107527e-05,
      "loss": 5.8395,
      "step": 607
    },
    {
      "epoch": 0.0245458215583367,
      "grad_norm": 155.65118408203125,
      "learning_rate": 1.6344086021505377e-05,
      "loss": 5.766,
      "step": 608
    },
    {
      "epoch": 0.024586192975373435,
      "grad_norm": 145.38436889648438,
      "learning_rate": 1.6370967741935487e-05,
      "loss": 5.6421,
      "step": 609
    },
    {
      "epoch": 0.024626564392410174,
      "grad_norm": 246.2040252685547,
      "learning_rate": 1.6397849462365594e-05,
      "loss": 4.7462,
      "step": 610
    },
    {
      "epoch": 0.024666935809446913,
      "grad_norm": 131.02540588378906,
      "learning_rate": 1.64247311827957e-05,
      "loss": 5.3234,
      "step": 611
    },
    {
      "epoch": 0.02470730722648365,
      "grad_norm": 137.05413818359375,
      "learning_rate": 1.6451612903225807e-05,
      "loss": 7.1081,
      "step": 612
    },
    {
      "epoch": 0.02474767864352039,
      "grad_norm": 122.85840606689453,
      "learning_rate": 1.6478494623655917e-05,
      "loss": 7.2581,
      "step": 613
    },
    {
      "epoch": 0.024788050060557124,
      "grad_norm": 184.2938232421875,
      "learning_rate": 1.6505376344086024e-05,
      "loss": 6.8739,
      "step": 614
    },
    {
      "epoch": 0.024828421477593864,
      "grad_norm": 109.57830047607422,
      "learning_rate": 1.653225806451613e-05,
      "loss": 6.5946,
      "step": 615
    },
    {
      "epoch": 0.024868792894630603,
      "grad_norm": 149.72373962402344,
      "learning_rate": 1.6559139784946237e-05,
      "loss": 5.6897,
      "step": 616
    },
    {
      "epoch": 0.02490916431166734,
      "grad_norm": 112.42048645019531,
      "learning_rate": 1.6586021505376347e-05,
      "loss": 6.394,
      "step": 617
    },
    {
      "epoch": 0.024949535728704078,
      "grad_norm": 115.7994384765625,
      "learning_rate": 1.6612903225806454e-05,
      "loss": 6.3147,
      "step": 618
    },
    {
      "epoch": 0.024989907145740814,
      "grad_norm": 151.27850341796875,
      "learning_rate": 1.663978494623656e-05,
      "loss": 6.0057,
      "step": 619
    },
    {
      "epoch": 0.025030278562777553,
      "grad_norm": 195.1448211669922,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 5.9696,
      "step": 620
    },
    {
      "epoch": 0.025070649979814293,
      "grad_norm": 173.48692321777344,
      "learning_rate": 1.6693548387096778e-05,
      "loss": 6.8192,
      "step": 621
    },
    {
      "epoch": 0.02511102139685103,
      "grad_norm": 168.95204162597656,
      "learning_rate": 1.6720430107526884e-05,
      "loss": 5.6877,
      "step": 622
    },
    {
      "epoch": 0.025151392813887768,
      "grad_norm": 176.62222290039062,
      "learning_rate": 1.674731182795699e-05,
      "loss": 5.9178,
      "step": 623
    },
    {
      "epoch": 0.025191764230924507,
      "grad_norm": 151.5782012939453,
      "learning_rate": 1.6774193548387098e-05,
      "loss": 5.9982,
      "step": 624
    },
    {
      "epoch": 0.025232135647961243,
      "grad_norm": 129.68963623046875,
      "learning_rate": 1.6801075268817208e-05,
      "loss": 4.7842,
      "step": 625
    },
    {
      "epoch": 0.025272507064997982,
      "grad_norm": 193.21202087402344,
      "learning_rate": 1.6827956989247314e-05,
      "loss": 7.0383,
      "step": 626
    },
    {
      "epoch": 0.025312878482034718,
      "grad_norm": 153.1450653076172,
      "learning_rate": 1.685483870967742e-05,
      "loss": 6.7622,
      "step": 627
    },
    {
      "epoch": 0.025353249899071457,
      "grad_norm": 134.05064392089844,
      "learning_rate": 1.6881720430107528e-05,
      "loss": 5.7496,
      "step": 628
    },
    {
      "epoch": 0.025393621316108197,
      "grad_norm": 160.38987731933594,
      "learning_rate": 1.6908602150537638e-05,
      "loss": 6.2858,
      "step": 629
    },
    {
      "epoch": 0.025433992733144933,
      "grad_norm": 162.4712371826172,
      "learning_rate": 1.6935483870967744e-05,
      "loss": 6.2994,
      "step": 630
    },
    {
      "epoch": 0.025474364150181672,
      "grad_norm": 148.18630981445312,
      "learning_rate": 1.696236559139785e-05,
      "loss": 7.0066,
      "step": 631
    },
    {
      "epoch": 0.025514735567218408,
      "grad_norm": 307.47900390625,
      "learning_rate": 1.6989247311827958e-05,
      "loss": 5.8822,
      "step": 632
    },
    {
      "epoch": 0.025555106984255147,
      "grad_norm": 343.90960693359375,
      "learning_rate": 1.7016129032258068e-05,
      "loss": 7.5116,
      "step": 633
    },
    {
      "epoch": 0.025595478401291887,
      "grad_norm": 162.1150665283203,
      "learning_rate": 1.7043010752688175e-05,
      "loss": 6.0708,
      "step": 634
    },
    {
      "epoch": 0.025635849818328622,
      "grad_norm": 206.94627380371094,
      "learning_rate": 1.706989247311828e-05,
      "loss": 5.1928,
      "step": 635
    },
    {
      "epoch": 0.02567622123536536,
      "grad_norm": 169.7374267578125,
      "learning_rate": 1.7096774193548388e-05,
      "loss": 6.0916,
      "step": 636
    },
    {
      "epoch": 0.0257165926524021,
      "grad_norm": 105.36103057861328,
      "learning_rate": 1.7123655913978495e-05,
      "loss": 5.7946,
      "step": 637
    },
    {
      "epoch": 0.025756964069438837,
      "grad_norm": 216.76124572753906,
      "learning_rate": 1.7150537634408605e-05,
      "loss": 5.4573,
      "step": 638
    },
    {
      "epoch": 0.025797335486475576,
      "grad_norm": 175.11703491210938,
      "learning_rate": 1.717741935483871e-05,
      "loss": 7.4343,
      "step": 639
    },
    {
      "epoch": 0.025837706903512312,
      "grad_norm": 134.11630249023438,
      "learning_rate": 1.7204301075268818e-05,
      "loss": 5.3323,
      "step": 640
    },
    {
      "epoch": 0.02587807832054905,
      "grad_norm": 124.01092529296875,
      "learning_rate": 1.7231182795698925e-05,
      "loss": 6.9354,
      "step": 641
    },
    {
      "epoch": 0.02591844973758579,
      "grad_norm": 117.30741882324219,
      "learning_rate": 1.7258064516129035e-05,
      "loss": 6.0889,
      "step": 642
    },
    {
      "epoch": 0.025958821154622527,
      "grad_norm": 113.90869903564453,
      "learning_rate": 1.728494623655914e-05,
      "loss": 7.5824,
      "step": 643
    },
    {
      "epoch": 0.025999192571659266,
      "grad_norm": 137.9143524169922,
      "learning_rate": 1.7311827956989248e-05,
      "loss": 4.9483,
      "step": 644
    },
    {
      "epoch": 0.026039563988696002,
      "grad_norm": 118.71682739257812,
      "learning_rate": 1.7338709677419355e-05,
      "loss": 6.6939,
      "step": 645
    },
    {
      "epoch": 0.02607993540573274,
      "grad_norm": 133.0789031982422,
      "learning_rate": 1.7365591397849465e-05,
      "loss": 4.9209,
      "step": 646
    },
    {
      "epoch": 0.02612030682276948,
      "grad_norm": 419.6683654785156,
      "learning_rate": 1.7392473118279572e-05,
      "loss": 5.4395,
      "step": 647
    },
    {
      "epoch": 0.026160678239806216,
      "grad_norm": 126.87855529785156,
      "learning_rate": 1.741935483870968e-05,
      "loss": 6.8796,
      "step": 648
    },
    {
      "epoch": 0.026201049656842956,
      "grad_norm": 143.43040466308594,
      "learning_rate": 1.7446236559139785e-05,
      "loss": 4.8485,
      "step": 649
    },
    {
      "epoch": 0.02624142107387969,
      "grad_norm": 144.1331787109375,
      "learning_rate": 1.7473118279569895e-05,
      "loss": 5.8846,
      "step": 650
    },
    {
      "epoch": 0.02628179249091643,
      "grad_norm": 146.27528381347656,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 5.7941,
      "step": 651
    },
    {
      "epoch": 0.02632216390795317,
      "grad_norm": 243.48419189453125,
      "learning_rate": 1.752688172043011e-05,
      "loss": 5.5227,
      "step": 652
    },
    {
      "epoch": 0.026362535324989906,
      "grad_norm": 124.34981536865234,
      "learning_rate": 1.7553763440860215e-05,
      "loss": 5.7255,
      "step": 653
    },
    {
      "epoch": 0.026402906742026645,
      "grad_norm": 157.35003662109375,
      "learning_rate": 1.7580645161290325e-05,
      "loss": 4.4257,
      "step": 654
    },
    {
      "epoch": 0.026443278159063385,
      "grad_norm": 139.78404235839844,
      "learning_rate": 1.7607526881720432e-05,
      "loss": 6.8506,
      "step": 655
    },
    {
      "epoch": 0.02648364957610012,
      "grad_norm": 144.61541748046875,
      "learning_rate": 1.763440860215054e-05,
      "loss": 6.008,
      "step": 656
    },
    {
      "epoch": 0.02652402099313686,
      "grad_norm": 139.70672607421875,
      "learning_rate": 1.7661290322580645e-05,
      "loss": 5.6927,
      "step": 657
    },
    {
      "epoch": 0.026564392410173596,
      "grad_norm": 118.31842803955078,
      "learning_rate": 1.7688172043010755e-05,
      "loss": 7.0795,
      "step": 658
    },
    {
      "epoch": 0.026604763827210335,
      "grad_norm": 128.8408966064453,
      "learning_rate": 1.7715053763440862e-05,
      "loss": 6.648,
      "step": 659
    },
    {
      "epoch": 0.026645135244247074,
      "grad_norm": 156.77145385742188,
      "learning_rate": 1.774193548387097e-05,
      "loss": 5.6915,
      "step": 660
    },
    {
      "epoch": 0.02668550666128381,
      "grad_norm": 102.55547332763672,
      "learning_rate": 1.7768817204301075e-05,
      "loss": 5.2892,
      "step": 661
    },
    {
      "epoch": 0.02672587807832055,
      "grad_norm": 105.03170013427734,
      "learning_rate": 1.7795698924731186e-05,
      "loss": 5.3158,
      "step": 662
    },
    {
      "epoch": 0.026766249495357285,
      "grad_norm": 168.19140625,
      "learning_rate": 1.7822580645161292e-05,
      "loss": 4.7904,
      "step": 663
    },
    {
      "epoch": 0.026806620912394025,
      "grad_norm": 132.76382446289062,
      "learning_rate": 1.78494623655914e-05,
      "loss": 6.0434,
      "step": 664
    },
    {
      "epoch": 0.026846992329430764,
      "grad_norm": 160.90626525878906,
      "learning_rate": 1.7876344086021506e-05,
      "loss": 5.0851,
      "step": 665
    },
    {
      "epoch": 0.0268873637464675,
      "grad_norm": 157.7742462158203,
      "learning_rate": 1.7903225806451612e-05,
      "loss": 6.65,
      "step": 666
    },
    {
      "epoch": 0.02692773516350424,
      "grad_norm": 210.41253662109375,
      "learning_rate": 1.7930107526881722e-05,
      "loss": 6.6192,
      "step": 667
    },
    {
      "epoch": 0.02696810658054098,
      "grad_norm": 203.73492431640625,
      "learning_rate": 1.795698924731183e-05,
      "loss": 3.9774,
      "step": 668
    },
    {
      "epoch": 0.027008477997577714,
      "grad_norm": 291.2826843261719,
      "learning_rate": 1.7983870967741936e-05,
      "loss": 5.2776,
      "step": 669
    },
    {
      "epoch": 0.027048849414614454,
      "grad_norm": 126.52214813232422,
      "learning_rate": 1.8010752688172042e-05,
      "loss": 7.521,
      "step": 670
    },
    {
      "epoch": 0.02708922083165119,
      "grad_norm": 180.26870727539062,
      "learning_rate": 1.8037634408602153e-05,
      "loss": 3.972,
      "step": 671
    },
    {
      "epoch": 0.02712959224868793,
      "grad_norm": 174.74668884277344,
      "learning_rate": 1.806451612903226e-05,
      "loss": 5.422,
      "step": 672
    },
    {
      "epoch": 0.027169963665724668,
      "grad_norm": 130.19967651367188,
      "learning_rate": 1.8091397849462366e-05,
      "loss": 7.6373,
      "step": 673
    },
    {
      "epoch": 0.027210335082761404,
      "grad_norm": 122.52957916259766,
      "learning_rate": 1.8118279569892473e-05,
      "loss": 5.5023,
      "step": 674
    },
    {
      "epoch": 0.027250706499798143,
      "grad_norm": 119.37210845947266,
      "learning_rate": 1.8145161290322583e-05,
      "loss": 4.2455,
      "step": 675
    },
    {
      "epoch": 0.02729107791683488,
      "grad_norm": 204.00601196289062,
      "learning_rate": 1.817204301075269e-05,
      "loss": 5.5825,
      "step": 676
    },
    {
      "epoch": 0.02733144933387162,
      "grad_norm": 150.21426391601562,
      "learning_rate": 1.8198924731182796e-05,
      "loss": 6.1065,
      "step": 677
    },
    {
      "epoch": 0.027371820750908358,
      "grad_norm": 136.64076232910156,
      "learning_rate": 1.8225806451612903e-05,
      "loss": 5.9277,
      "step": 678
    },
    {
      "epoch": 0.027412192167945094,
      "grad_norm": 181.78286743164062,
      "learning_rate": 1.8252688172043013e-05,
      "loss": 5.2932,
      "step": 679
    },
    {
      "epoch": 0.027452563584981833,
      "grad_norm": 134.0491180419922,
      "learning_rate": 1.827956989247312e-05,
      "loss": 6.9872,
      "step": 680
    },
    {
      "epoch": 0.027492935002018572,
      "grad_norm": 112.20816040039062,
      "learning_rate": 1.8306451612903226e-05,
      "loss": 5.193,
      "step": 681
    },
    {
      "epoch": 0.027533306419055308,
      "grad_norm": 180.65318298339844,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 5.0837,
      "step": 682
    },
    {
      "epoch": 0.027573677836092048,
      "grad_norm": 152.870849609375,
      "learning_rate": 1.8360215053763443e-05,
      "loss": 6.3191,
      "step": 683
    },
    {
      "epoch": 0.027614049253128783,
      "grad_norm": 95.07835388183594,
      "learning_rate": 1.838709677419355e-05,
      "loss": 6.9891,
      "step": 684
    },
    {
      "epoch": 0.027654420670165523,
      "grad_norm": 264.2321472167969,
      "learning_rate": 1.8413978494623656e-05,
      "loss": 6.0796,
      "step": 685
    },
    {
      "epoch": 0.027694792087202262,
      "grad_norm": 156.23619079589844,
      "learning_rate": 1.8440860215053763e-05,
      "loss": 6.3964,
      "step": 686
    },
    {
      "epoch": 0.027735163504238998,
      "grad_norm": 170.87814331054688,
      "learning_rate": 1.8467741935483873e-05,
      "loss": 5.8791,
      "step": 687
    },
    {
      "epoch": 0.027775534921275737,
      "grad_norm": 149.30462646484375,
      "learning_rate": 1.849462365591398e-05,
      "loss": 6.6088,
      "step": 688
    },
    {
      "epoch": 0.027815906338312473,
      "grad_norm": 148.1158447265625,
      "learning_rate": 1.8521505376344086e-05,
      "loss": 4.8396,
      "step": 689
    },
    {
      "epoch": 0.027856277755349212,
      "grad_norm": 144.70001220703125,
      "learning_rate": 1.8548387096774193e-05,
      "loss": 6.416,
      "step": 690
    },
    {
      "epoch": 0.027896649172385952,
      "grad_norm": 135.49652099609375,
      "learning_rate": 1.8575268817204303e-05,
      "loss": 5.7396,
      "step": 691
    },
    {
      "epoch": 0.027937020589422688,
      "grad_norm": 141.27232360839844,
      "learning_rate": 1.860215053763441e-05,
      "loss": 6.5657,
      "step": 692
    },
    {
      "epoch": 0.027977392006459427,
      "grad_norm": 172.46424865722656,
      "learning_rate": 1.8629032258064517e-05,
      "loss": 6.1763,
      "step": 693
    },
    {
      "epoch": 0.028017763423496166,
      "grad_norm": 99.00566101074219,
      "learning_rate": 1.8655913978494623e-05,
      "loss": 6.1374,
      "step": 694
    },
    {
      "epoch": 0.028058134840532902,
      "grad_norm": 144.7807159423828,
      "learning_rate": 1.8682795698924733e-05,
      "loss": 6.3421,
      "step": 695
    },
    {
      "epoch": 0.02809850625756964,
      "grad_norm": 95.53231811523438,
      "learning_rate": 1.870967741935484e-05,
      "loss": 4.8389,
      "step": 696
    },
    {
      "epoch": 0.028138877674606377,
      "grad_norm": 156.76622009277344,
      "learning_rate": 1.8736559139784947e-05,
      "loss": 5.7737,
      "step": 697
    },
    {
      "epoch": 0.028179249091643117,
      "grad_norm": 140.9648895263672,
      "learning_rate": 1.8763440860215057e-05,
      "loss": 4.6424,
      "step": 698
    },
    {
      "epoch": 0.028219620508679856,
      "grad_norm": 235.06712341308594,
      "learning_rate": 1.8790322580645163e-05,
      "loss": 6.6876,
      "step": 699
    },
    {
      "epoch": 0.028259991925716592,
      "grad_norm": 179.94549560546875,
      "learning_rate": 1.881720430107527e-05,
      "loss": 4.8381,
      "step": 700
    },
    {
      "epoch": 0.02830036334275333,
      "grad_norm": 221.57965087890625,
      "learning_rate": 1.884408602150538e-05,
      "loss": 5.3553,
      "step": 701
    },
    {
      "epoch": 0.028340734759790067,
      "grad_norm": 136.35244750976562,
      "learning_rate": 1.8870967741935487e-05,
      "loss": 5.5232,
      "step": 702
    },
    {
      "epoch": 0.028381106176826806,
      "grad_norm": 280.5992736816406,
      "learning_rate": 1.8897849462365594e-05,
      "loss": 7.0797,
      "step": 703
    },
    {
      "epoch": 0.028421477593863546,
      "grad_norm": 171.9462890625,
      "learning_rate": 1.89247311827957e-05,
      "loss": 4.6355,
      "step": 704
    },
    {
      "epoch": 0.02846184901090028,
      "grad_norm": 140.6548614501953,
      "learning_rate": 1.895161290322581e-05,
      "loss": 4.6841,
      "step": 705
    },
    {
      "epoch": 0.02850222042793702,
      "grad_norm": 196.39076232910156,
      "learning_rate": 1.8978494623655917e-05,
      "loss": 5.1841,
      "step": 706
    },
    {
      "epoch": 0.02854259184497376,
      "grad_norm": 127.93045043945312,
      "learning_rate": 1.9005376344086024e-05,
      "loss": 5.1453,
      "step": 707
    },
    {
      "epoch": 0.028582963262010496,
      "grad_norm": 104.76632690429688,
      "learning_rate": 1.903225806451613e-05,
      "loss": 5.5775,
      "step": 708
    },
    {
      "epoch": 0.028623334679047235,
      "grad_norm": 194.8694610595703,
      "learning_rate": 1.905913978494624e-05,
      "loss": 6.3428,
      "step": 709
    },
    {
      "epoch": 0.02866370609608397,
      "grad_norm": 150.13272094726562,
      "learning_rate": 1.9086021505376347e-05,
      "loss": 4.6135,
      "step": 710
    },
    {
      "epoch": 0.02870407751312071,
      "grad_norm": 119.21205139160156,
      "learning_rate": 1.9112903225806454e-05,
      "loss": 6.0261,
      "step": 711
    },
    {
      "epoch": 0.02874444893015745,
      "grad_norm": 99.04373931884766,
      "learning_rate": 1.913978494623656e-05,
      "loss": 7.1813,
      "step": 712
    },
    {
      "epoch": 0.028784820347194186,
      "grad_norm": 164.93875122070312,
      "learning_rate": 1.916666666666667e-05,
      "loss": 5.7909,
      "step": 713
    },
    {
      "epoch": 0.028825191764230925,
      "grad_norm": 153.8946990966797,
      "learning_rate": 1.9193548387096777e-05,
      "loss": 5.163,
      "step": 714
    },
    {
      "epoch": 0.02886556318126766,
      "grad_norm": 133.54054260253906,
      "learning_rate": 1.9220430107526884e-05,
      "loss": 6.4181,
      "step": 715
    },
    {
      "epoch": 0.0289059345983044,
      "grad_norm": 118.13850402832031,
      "learning_rate": 1.924731182795699e-05,
      "loss": 6.2947,
      "step": 716
    },
    {
      "epoch": 0.02894630601534114,
      "grad_norm": 138.39599609375,
      "learning_rate": 1.9274193548387097e-05,
      "loss": 5.7015,
      "step": 717
    },
    {
      "epoch": 0.028986677432377875,
      "grad_norm": 137.29310607910156,
      "learning_rate": 1.9301075268817207e-05,
      "loss": 6.2111,
      "step": 718
    },
    {
      "epoch": 0.029027048849414615,
      "grad_norm": 145.45257568359375,
      "learning_rate": 1.9327956989247314e-05,
      "loss": 5.6258,
      "step": 719
    },
    {
      "epoch": 0.029067420266451354,
      "grad_norm": 130.517333984375,
      "learning_rate": 1.935483870967742e-05,
      "loss": 5.1447,
      "step": 720
    },
    {
      "epoch": 0.02910779168348809,
      "grad_norm": 104.35297393798828,
      "learning_rate": 1.9381720430107528e-05,
      "loss": 6.5716,
      "step": 721
    },
    {
      "epoch": 0.02914816310052483,
      "grad_norm": 118.76133728027344,
      "learning_rate": 1.9408602150537638e-05,
      "loss": 6.3603,
      "step": 722
    },
    {
      "epoch": 0.029188534517561565,
      "grad_norm": 203.42645263671875,
      "learning_rate": 1.9435483870967744e-05,
      "loss": 5.9606,
      "step": 723
    },
    {
      "epoch": 0.029228905934598304,
      "grad_norm": 99.12989044189453,
      "learning_rate": 1.946236559139785e-05,
      "loss": 5.6205,
      "step": 724
    },
    {
      "epoch": 0.029269277351635044,
      "grad_norm": 119.08329010009766,
      "learning_rate": 1.9489247311827958e-05,
      "loss": 6.4766,
      "step": 725
    },
    {
      "epoch": 0.02930964876867178,
      "grad_norm": 173.40542602539062,
      "learning_rate": 1.9516129032258068e-05,
      "loss": 6.9503,
      "step": 726
    },
    {
      "epoch": 0.02935002018570852,
      "grad_norm": 126.43241119384766,
      "learning_rate": 1.9543010752688174e-05,
      "loss": 5.9944,
      "step": 727
    },
    {
      "epoch": 0.029390391602745255,
      "grad_norm": 180.92044067382812,
      "learning_rate": 1.956989247311828e-05,
      "loss": 6.0789,
      "step": 728
    },
    {
      "epoch": 0.029430763019781994,
      "grad_norm": 121.26048278808594,
      "learning_rate": 1.9596774193548388e-05,
      "loss": 6.0207,
      "step": 729
    },
    {
      "epoch": 0.029471134436818733,
      "grad_norm": 429.0849914550781,
      "learning_rate": 1.9623655913978498e-05,
      "loss": 5.4449,
      "step": 730
    },
    {
      "epoch": 0.02951150585385547,
      "grad_norm": 109.97157287597656,
      "learning_rate": 1.9650537634408605e-05,
      "loss": 7.1009,
      "step": 731
    },
    {
      "epoch": 0.02955187727089221,
      "grad_norm": 123.63129425048828,
      "learning_rate": 1.967741935483871e-05,
      "loss": 7.2187,
      "step": 732
    },
    {
      "epoch": 0.029592248687928948,
      "grad_norm": 104.45331573486328,
      "learning_rate": 1.9704301075268818e-05,
      "loss": 5.7947,
      "step": 733
    },
    {
      "epoch": 0.029632620104965684,
      "grad_norm": 203.1223602294922,
      "learning_rate": 1.9731182795698928e-05,
      "loss": 6.1519,
      "step": 734
    },
    {
      "epoch": 0.029672991522002423,
      "grad_norm": 108.81645202636719,
      "learning_rate": 1.9758064516129035e-05,
      "loss": 5.3871,
      "step": 735
    },
    {
      "epoch": 0.02971336293903916,
      "grad_norm": 126.79362487792969,
      "learning_rate": 1.978494623655914e-05,
      "loss": 6.4092,
      "step": 736
    },
    {
      "epoch": 0.0297537343560759,
      "grad_norm": 169.3652801513672,
      "learning_rate": 1.9811827956989248e-05,
      "loss": 4.9272,
      "step": 737
    },
    {
      "epoch": 0.029794105773112638,
      "grad_norm": 161.84059143066406,
      "learning_rate": 1.9838709677419358e-05,
      "loss": 5.1511,
      "step": 738
    },
    {
      "epoch": 0.029834477190149374,
      "grad_norm": 260.0711669921875,
      "learning_rate": 1.9865591397849465e-05,
      "loss": 5.5186,
      "step": 739
    },
    {
      "epoch": 0.029874848607186113,
      "grad_norm": 144.03738403320312,
      "learning_rate": 1.989247311827957e-05,
      "loss": 5.7457,
      "step": 740
    },
    {
      "epoch": 0.02991522002422285,
      "grad_norm": 121.43706512451172,
      "learning_rate": 1.9919354838709678e-05,
      "loss": 6.307,
      "step": 741
    },
    {
      "epoch": 0.029955591441259588,
      "grad_norm": 144.45071411132812,
      "learning_rate": 1.9946236559139788e-05,
      "loss": 5.0112,
      "step": 742
    },
    {
      "epoch": 0.029995962858296327,
      "grad_norm": 110.13965606689453,
      "learning_rate": 1.9973118279569895e-05,
      "loss": 5.692,
      "step": 743
    },
    {
      "epoch": 0.030036334275333063,
      "grad_norm": 121.21327209472656,
      "learning_rate": 2e-05,
      "loss": 5.8425,
      "step": 744
    },
    {
      "epoch": 0.030076705692369803,
      "grad_norm": 138.83041381835938,
      "learning_rate": 1.999999991451168e-05,
      "loss": 4.4493,
      "step": 745
    },
    {
      "epoch": 0.030117077109406542,
      "grad_norm": 125.09162902832031,
      "learning_rate": 1.9999999658046706e-05,
      "loss": 5.8006,
      "step": 746
    },
    {
      "epoch": 0.030157448526443278,
      "grad_norm": 118.01892852783203,
      "learning_rate": 1.9999999230605095e-05,
      "loss": 6.9616,
      "step": 747
    },
    {
      "epoch": 0.030197819943480017,
      "grad_norm": 127.29279327392578,
      "learning_rate": 1.9999998632186845e-05,
      "loss": 6.3564,
      "step": 748
    },
    {
      "epoch": 0.030238191360516753,
      "grad_norm": 164.21701049804688,
      "learning_rate": 1.999999786279197e-05,
      "loss": 5.8853,
      "step": 749
    },
    {
      "epoch": 0.030278562777553492,
      "grad_norm": 98.62155151367188,
      "learning_rate": 1.9999996922420486e-05,
      "loss": 7.1547,
      "step": 750
    },
    {
      "epoch": 0.03031893419459023,
      "grad_norm": 135.09359741210938,
      "learning_rate": 1.9999995811072405e-05,
      "loss": 5.522,
      "step": 751
    },
    {
      "epoch": 0.030359305611626967,
      "grad_norm": 132.3077392578125,
      "learning_rate": 1.9999994528747746e-05,
      "loss": 6.3703,
      "step": 752
    },
    {
      "epoch": 0.030399677028663707,
      "grad_norm": 110.67572784423828,
      "learning_rate": 1.999999307544653e-05,
      "loss": 5.0506,
      "step": 753
    },
    {
      "epoch": 0.030440048445700443,
      "grad_norm": 138.02127075195312,
      "learning_rate": 1.999999145116879e-05,
      "loss": 7.1428,
      "step": 754
    },
    {
      "epoch": 0.030480419862737182,
      "grad_norm": 117.11345672607422,
      "learning_rate": 1.9999989655914547e-05,
      "loss": 6.6004,
      "step": 755
    },
    {
      "epoch": 0.03052079127977392,
      "grad_norm": 131.84295654296875,
      "learning_rate": 1.999998768968383e-05,
      "loss": 6.5441,
      "step": 756
    },
    {
      "epoch": 0.030561162696810657,
      "grad_norm": 165.1290740966797,
      "learning_rate": 1.9999985552476678e-05,
      "loss": 6.6317,
      "step": 757
    },
    {
      "epoch": 0.030601534113847396,
      "grad_norm": 163.119873046875,
      "learning_rate": 1.9999983244293123e-05,
      "loss": 5.7838,
      "step": 758
    },
    {
      "epoch": 0.030641905530884132,
      "grad_norm": 167.27133178710938,
      "learning_rate": 1.9999980765133203e-05,
      "loss": 4.8037,
      "step": 759
    },
    {
      "epoch": 0.03068227694792087,
      "grad_norm": 124.6041488647461,
      "learning_rate": 1.999997811499697e-05,
      "loss": 7.1724,
      "step": 760
    },
    {
      "epoch": 0.03072264836495761,
      "grad_norm": 162.15769958496094,
      "learning_rate": 1.9999975293884457e-05,
      "loss": 4.1169,
      "step": 761
    },
    {
      "epoch": 0.030763019781994347,
      "grad_norm": 121.73171997070312,
      "learning_rate": 1.999997230179572e-05,
      "loss": 4.9689,
      "step": 762
    },
    {
      "epoch": 0.030803391199031086,
      "grad_norm": 193.22073364257812,
      "learning_rate": 1.9999969138730807e-05,
      "loss": 5.6079,
      "step": 763
    },
    {
      "epoch": 0.030843762616067825,
      "grad_norm": 149.10311889648438,
      "learning_rate": 1.999996580468978e-05,
      "loss": 6.0131,
      "step": 764
    },
    {
      "epoch": 0.03088413403310456,
      "grad_norm": 103.65576171875,
      "learning_rate": 1.9999962299672682e-05,
      "loss": 5.6153,
      "step": 765
    },
    {
      "epoch": 0.0309245054501413,
      "grad_norm": 128.1395263671875,
      "learning_rate": 1.999995862367958e-05,
      "loss": 5.5463,
      "step": 766
    },
    {
      "epoch": 0.030964876867178037,
      "grad_norm": 154.79025268554688,
      "learning_rate": 1.9999954776710543e-05,
      "loss": 6.3778,
      "step": 767
    },
    {
      "epoch": 0.031005248284214776,
      "grad_norm": 127.26738739013672,
      "learning_rate": 1.9999950758765627e-05,
      "loss": 6.1018,
      "step": 768
    },
    {
      "epoch": 0.031045619701251515,
      "grad_norm": 137.2698974609375,
      "learning_rate": 1.9999946569844906e-05,
      "loss": 5.8867,
      "step": 769
    },
    {
      "epoch": 0.03108599111828825,
      "grad_norm": 125.1261215209961,
      "learning_rate": 1.999994220994845e-05,
      "loss": 6.8871,
      "step": 770
    },
    {
      "epoch": 0.03112636253532499,
      "grad_norm": 132.70816040039062,
      "learning_rate": 1.9999937679076332e-05,
      "loss": 5.7515,
      "step": 771
    },
    {
      "epoch": 0.031166733952361726,
      "grad_norm": 121.54830932617188,
      "learning_rate": 1.9999932977228632e-05,
      "loss": 5.9486,
      "step": 772
    },
    {
      "epoch": 0.031207105369398466,
      "grad_norm": 120.91377258300781,
      "learning_rate": 1.999992810440543e-05,
      "loss": 5.6232,
      "step": 773
    },
    {
      "epoch": 0.031247476786435205,
      "grad_norm": 133.53346252441406,
      "learning_rate": 1.9999923060606808e-05,
      "loss": 6.4667,
      "step": 774
    },
    {
      "epoch": 0.03128784820347194,
      "grad_norm": 126.83102416992188,
      "learning_rate": 1.9999917845832855e-05,
      "loss": 7.3893,
      "step": 775
    },
    {
      "epoch": 0.03132821962050868,
      "grad_norm": 124.92578887939453,
      "learning_rate": 1.9999912460083656e-05,
      "loss": 6.6121,
      "step": 776
    },
    {
      "epoch": 0.03136859103754542,
      "grad_norm": 239.8113555908203,
      "learning_rate": 1.9999906903359308e-05,
      "loss": 5.9739,
      "step": 777
    },
    {
      "epoch": 0.031408962454582155,
      "grad_norm": 132.5224151611328,
      "learning_rate": 1.9999901175659902e-05,
      "loss": 6.1072,
      "step": 778
    },
    {
      "epoch": 0.03144933387161889,
      "grad_norm": 133.64418029785156,
      "learning_rate": 1.9999895276985538e-05,
      "loss": 6.6841,
      "step": 779
    },
    {
      "epoch": 0.031489705288655634,
      "grad_norm": 139.95802307128906,
      "learning_rate": 1.9999889207336316e-05,
      "loss": 6.5916,
      "step": 780
    },
    {
      "epoch": 0.03153007670569237,
      "grad_norm": 147.82217407226562,
      "learning_rate": 1.9999882966712342e-05,
      "loss": 6.8454,
      "step": 781
    },
    {
      "epoch": 0.031570448122729106,
      "grad_norm": 138.67330932617188,
      "learning_rate": 1.9999876555113714e-05,
      "loss": 7.1796,
      "step": 782
    },
    {
      "epoch": 0.03161081953976585,
      "grad_norm": 165.6301727294922,
      "learning_rate": 1.9999869972540553e-05,
      "loss": 6.4903,
      "step": 783
    },
    {
      "epoch": 0.031651190956802584,
      "grad_norm": 178.33740234375,
      "learning_rate": 1.999986321899297e-05,
      "loss": 3.8832,
      "step": 784
    },
    {
      "epoch": 0.03169156237383932,
      "grad_norm": 107.67662048339844,
      "learning_rate": 1.9999856294471073e-05,
      "loss": 5.8186,
      "step": 785
    },
    {
      "epoch": 0.03173193379087606,
      "grad_norm": 111.66432189941406,
      "learning_rate": 1.9999849198974987e-05,
      "loss": 6.8261,
      "step": 786
    },
    {
      "epoch": 0.0317723052079128,
      "grad_norm": 104.31240844726562,
      "learning_rate": 1.999984193250483e-05,
      "loss": 7.392,
      "step": 787
    },
    {
      "epoch": 0.031812676624949535,
      "grad_norm": 111.12522888183594,
      "learning_rate": 1.9999834495060723e-05,
      "loss": 6.8899,
      "step": 788
    },
    {
      "epoch": 0.03185304804198627,
      "grad_norm": 151.4324188232422,
      "learning_rate": 1.9999826886642802e-05,
      "loss": 5.0698,
      "step": 789
    },
    {
      "epoch": 0.03189341945902301,
      "grad_norm": 216.28988647460938,
      "learning_rate": 1.9999819107251193e-05,
      "loss": 6.3006,
      "step": 790
    },
    {
      "epoch": 0.03193379087605975,
      "grad_norm": 100.34001922607422,
      "learning_rate": 1.9999811156886028e-05,
      "loss": 6.3766,
      "step": 791
    },
    {
      "epoch": 0.031974162293096485,
      "grad_norm": 117.63668060302734,
      "learning_rate": 1.999980303554744e-05,
      "loss": 5.5044,
      "step": 792
    },
    {
      "epoch": 0.03201453371013323,
      "grad_norm": 167.00457763671875,
      "learning_rate": 1.9999794743235576e-05,
      "loss": 5.4703,
      "step": 793
    },
    {
      "epoch": 0.032054905127169964,
      "grad_norm": 157.72796630859375,
      "learning_rate": 1.9999786279950572e-05,
      "loss": 4.1719,
      "step": 794
    },
    {
      "epoch": 0.0320952765442067,
      "grad_norm": 99.31024169921875,
      "learning_rate": 1.9999777645692574e-05,
      "loss": 6.4786,
      "step": 795
    },
    {
      "epoch": 0.03213564796124344,
      "grad_norm": 178.66697692871094,
      "learning_rate": 1.9999768840461727e-05,
      "loss": 5.0713,
      "step": 796
    },
    {
      "epoch": 0.03217601937828018,
      "grad_norm": 205.4830780029297,
      "learning_rate": 1.9999759864258185e-05,
      "loss": 6.0395,
      "step": 797
    },
    {
      "epoch": 0.032216390795316914,
      "grad_norm": 405.69915771484375,
      "learning_rate": 1.99997507170821e-05,
      "loss": 4.6259,
      "step": 798
    },
    {
      "epoch": 0.03225676221235366,
      "grad_norm": 124.84907531738281,
      "learning_rate": 1.9999741398933632e-05,
      "loss": 6.1442,
      "step": 799
    },
    {
      "epoch": 0.03229713362939039,
      "grad_norm": 139.66835021972656,
      "learning_rate": 1.9999731909812935e-05,
      "loss": 3.7666,
      "step": 800
    },
    {
      "epoch": 0.03233750504642713,
      "grad_norm": 131.73159790039062,
      "learning_rate": 1.999972224972017e-05,
      "loss": 6.3307,
      "step": 801
    },
    {
      "epoch": 0.032377876463463864,
      "grad_norm": 248.66162109375,
      "learning_rate": 1.999971241865551e-05,
      "loss": 6.1519,
      "step": 802
    },
    {
      "epoch": 0.03241824788050061,
      "grad_norm": 119.54292297363281,
      "learning_rate": 1.9999702416619118e-05,
      "loss": 6.3005,
      "step": 803
    },
    {
      "epoch": 0.03245861929753734,
      "grad_norm": 111.73394012451172,
      "learning_rate": 1.9999692243611167e-05,
      "loss": 7.1818,
      "step": 804
    },
    {
      "epoch": 0.03249899071457408,
      "grad_norm": 137.66334533691406,
      "learning_rate": 1.9999681899631825e-05,
      "loss": 7.6803,
      "step": 805
    },
    {
      "epoch": 0.03253936213161082,
      "grad_norm": 152.60023498535156,
      "learning_rate": 1.9999671384681275e-05,
      "loss": 6.1806,
      "step": 806
    },
    {
      "epoch": 0.03257973354864756,
      "grad_norm": 123.67101287841797,
      "learning_rate": 1.9999660698759696e-05,
      "loss": 6.1481,
      "step": 807
    },
    {
      "epoch": 0.03262010496568429,
      "grad_norm": 98.63252258300781,
      "learning_rate": 1.999964984186727e-05,
      "loss": 6.7035,
      "step": 808
    },
    {
      "epoch": 0.032660476382721036,
      "grad_norm": 111.173583984375,
      "learning_rate": 1.9999638814004186e-05,
      "loss": 6.3828,
      "step": 809
    },
    {
      "epoch": 0.03270084779975777,
      "grad_norm": 173.13311767578125,
      "learning_rate": 1.9999627615170625e-05,
      "loss": 6.7085,
      "step": 810
    },
    {
      "epoch": 0.03274121921679451,
      "grad_norm": 97.05487060546875,
      "learning_rate": 1.9999616245366784e-05,
      "loss": 5.1004,
      "step": 811
    },
    {
      "epoch": 0.03278159063383125,
      "grad_norm": 89.97145080566406,
      "learning_rate": 1.9999604704592856e-05,
      "loss": 5.4158,
      "step": 812
    },
    {
      "epoch": 0.03282196205086799,
      "grad_norm": 134.04937744140625,
      "learning_rate": 1.9999592992849038e-05,
      "loss": 6.5322,
      "step": 813
    },
    {
      "epoch": 0.03286233346790472,
      "grad_norm": 135.50779724121094,
      "learning_rate": 1.9999581110135532e-05,
      "loss": 6.217,
      "step": 814
    },
    {
      "epoch": 0.03290270488494146,
      "grad_norm": 123.86219787597656,
      "learning_rate": 1.9999569056452536e-05,
      "loss": 4.9818,
      "step": 815
    },
    {
      "epoch": 0.0329430763019782,
      "grad_norm": 141.5067138671875,
      "learning_rate": 1.9999556831800263e-05,
      "loss": 5.1641,
      "step": 816
    },
    {
      "epoch": 0.03298344771901494,
      "grad_norm": 112.55216979980469,
      "learning_rate": 1.9999544436178916e-05,
      "loss": 5.1368,
      "step": 817
    },
    {
      "epoch": 0.03302381913605167,
      "grad_norm": 119.71576690673828,
      "learning_rate": 1.999953186958871e-05,
      "loss": 5.6168,
      "step": 818
    },
    {
      "epoch": 0.033064190553088416,
      "grad_norm": 1255.5794677734375,
      "learning_rate": 1.9999519132029864e-05,
      "loss": 6.5912,
      "step": 819
    },
    {
      "epoch": 0.03310456197012515,
      "grad_norm": 146.68553161621094,
      "learning_rate": 1.9999506223502586e-05,
      "loss": 7.9061,
      "step": 820
    },
    {
      "epoch": 0.03314493338716189,
      "grad_norm": 119.3077163696289,
      "learning_rate": 1.9999493144007105e-05,
      "loss": 5.4123,
      "step": 821
    },
    {
      "epoch": 0.03318530480419863,
      "grad_norm": 100.68647766113281,
      "learning_rate": 1.999947989354364e-05,
      "loss": 5.6525,
      "step": 822
    },
    {
      "epoch": 0.033225676221235366,
      "grad_norm": 147.1737060546875,
      "learning_rate": 1.999946647211242e-05,
      "loss": 4.9275,
      "step": 823
    },
    {
      "epoch": 0.0332660476382721,
      "grad_norm": 132.98434448242188,
      "learning_rate": 1.9999452879713676e-05,
      "loss": 6.3454,
      "step": 824
    },
    {
      "epoch": 0.033306419055308845,
      "grad_norm": 152.3587188720703,
      "learning_rate": 1.999943911634764e-05,
      "loss": 7.022,
      "step": 825
    },
    {
      "epoch": 0.03334679047234558,
      "grad_norm": 181.61106872558594,
      "learning_rate": 1.999942518201454e-05,
      "loss": 5.7071,
      "step": 826
    },
    {
      "epoch": 0.033387161889382316,
      "grad_norm": 156.72279357910156,
      "learning_rate": 1.9999411076714622e-05,
      "loss": 4.6866,
      "step": 827
    },
    {
      "epoch": 0.03342753330641905,
      "grad_norm": 162.09695434570312,
      "learning_rate": 1.9999396800448127e-05,
      "loss": 5.8258,
      "step": 828
    },
    {
      "epoch": 0.033467904723455795,
      "grad_norm": 140.93106079101562,
      "learning_rate": 1.9999382353215295e-05,
      "loss": 6.5438,
      "step": 829
    },
    {
      "epoch": 0.03350827614049253,
      "grad_norm": 117.83767700195312,
      "learning_rate": 1.999936773501637e-05,
      "loss": 6.9841,
      "step": 830
    },
    {
      "epoch": 0.03354864755752927,
      "grad_norm": 114.69205474853516,
      "learning_rate": 1.9999352945851615e-05,
      "loss": 6.3622,
      "step": 831
    },
    {
      "epoch": 0.03358901897456601,
      "grad_norm": 2965.114501953125,
      "learning_rate": 1.999933798572127e-05,
      "loss": 5.8361,
      "step": 832
    },
    {
      "epoch": 0.033629390391602745,
      "grad_norm": 102.21658325195312,
      "learning_rate": 1.99993228546256e-05,
      "loss": 5.9281,
      "step": 833
    },
    {
      "epoch": 0.03366976180863948,
      "grad_norm": 163.72113037109375,
      "learning_rate": 1.9999307552564853e-05,
      "loss": 6.1478,
      "step": 834
    },
    {
      "epoch": 0.033710133225676224,
      "grad_norm": 109.87132263183594,
      "learning_rate": 1.9999292079539302e-05,
      "loss": 6.0031,
      "step": 835
    },
    {
      "epoch": 0.03375050464271296,
      "grad_norm": 100.67647552490234,
      "learning_rate": 1.99992764355492e-05,
      "loss": 6.036,
      "step": 836
    },
    {
      "epoch": 0.033790876059749696,
      "grad_norm": 112.22637176513672,
      "learning_rate": 1.999926062059483e-05,
      "loss": 7.2883,
      "step": 837
    },
    {
      "epoch": 0.03383124747678644,
      "grad_norm": 150.297119140625,
      "learning_rate": 1.9999244634676448e-05,
      "loss": 5.5879,
      "step": 838
    },
    {
      "epoch": 0.033871618893823174,
      "grad_norm": 115.50716400146484,
      "learning_rate": 1.999922847779433e-05,
      "loss": 5.1424,
      "step": 839
    },
    {
      "epoch": 0.03391199031085991,
      "grad_norm": 107.10466766357422,
      "learning_rate": 1.999921214994876e-05,
      "loss": 6.3032,
      "step": 840
    },
    {
      "epoch": 0.033952361727896646,
      "grad_norm": 161.39691162109375,
      "learning_rate": 1.9999195651140013e-05,
      "loss": 6.2612,
      "step": 841
    },
    {
      "epoch": 0.03399273314493339,
      "grad_norm": 145.865966796875,
      "learning_rate": 1.9999178981368364e-05,
      "loss": 5.838,
      "step": 842
    },
    {
      "epoch": 0.034033104561970125,
      "grad_norm": 143.28738403320312,
      "learning_rate": 1.999916214063411e-05,
      "loss": 6.2777,
      "step": 843
    },
    {
      "epoch": 0.03407347597900686,
      "grad_norm": 118.11651611328125,
      "learning_rate": 1.9999145128937533e-05,
      "loss": 5.1015,
      "step": 844
    },
    {
      "epoch": 0.0341138473960436,
      "grad_norm": 179.6404266357422,
      "learning_rate": 1.999912794627892e-05,
      "loss": 7.3299,
      "step": 845
    },
    {
      "epoch": 0.03415421881308034,
      "grad_norm": 130.1668243408203,
      "learning_rate": 1.9999110592658572e-05,
      "loss": 7.1905,
      "step": 846
    },
    {
      "epoch": 0.034194590230117075,
      "grad_norm": 278.6486511230469,
      "learning_rate": 1.999909306807678e-05,
      "loss": 4.2691,
      "step": 847
    },
    {
      "epoch": 0.03423496164715382,
      "grad_norm": 111.74417877197266,
      "learning_rate": 1.9999075372533846e-05,
      "loss": 6.3425,
      "step": 848
    },
    {
      "epoch": 0.034275333064190554,
      "grad_norm": 105.1785659790039,
      "learning_rate": 1.9999057506030075e-05,
      "loss": 7.0636,
      "step": 849
    },
    {
      "epoch": 0.03431570448122729,
      "grad_norm": 148.92391967773438,
      "learning_rate": 1.999903946856577e-05,
      "loss": 6.4596,
      "step": 850
    },
    {
      "epoch": 0.03435607589826403,
      "grad_norm": 114.30416107177734,
      "learning_rate": 1.9999021260141237e-05,
      "loss": 5.7858,
      "step": 851
    },
    {
      "epoch": 0.03439644731530077,
      "grad_norm": 208.1607208251953,
      "learning_rate": 1.9999002880756793e-05,
      "loss": 6.1318,
      "step": 852
    },
    {
      "epoch": 0.034436818732337504,
      "grad_norm": 144.99942016601562,
      "learning_rate": 1.999898433041275e-05,
      "loss": 5.659,
      "step": 853
    },
    {
      "epoch": 0.03447719014937424,
      "grad_norm": 110.06531524658203,
      "learning_rate": 1.999896560910942e-05,
      "loss": 6.6841,
      "step": 854
    },
    {
      "epoch": 0.03451756156641098,
      "grad_norm": 90.0593490600586,
      "learning_rate": 1.9998946716847126e-05,
      "loss": 6.6324,
      "step": 855
    },
    {
      "epoch": 0.03455793298344772,
      "grad_norm": 334.1894836425781,
      "learning_rate": 1.9998927653626197e-05,
      "loss": 6.5766,
      "step": 856
    },
    {
      "epoch": 0.034598304400484454,
      "grad_norm": 158.49497985839844,
      "learning_rate": 1.9998908419446952e-05,
      "loss": 6.5332,
      "step": 857
    },
    {
      "epoch": 0.0346386758175212,
      "grad_norm": 96.22637939453125,
      "learning_rate": 1.9998889014309726e-05,
      "loss": 5.4182,
      "step": 858
    },
    {
      "epoch": 0.03467904723455793,
      "grad_norm": 112.8780746459961,
      "learning_rate": 1.9998869438214844e-05,
      "loss": 6.6751,
      "step": 859
    },
    {
      "epoch": 0.03471941865159467,
      "grad_norm": 118.67124938964844,
      "learning_rate": 1.9998849691162642e-05,
      "loss": 5.0382,
      "step": 860
    },
    {
      "epoch": 0.03475979006863141,
      "grad_norm": 120.44662475585938,
      "learning_rate": 1.999882977315346e-05,
      "loss": 6.7022,
      "step": 861
    },
    {
      "epoch": 0.03480016148566815,
      "grad_norm": 110.14851379394531,
      "learning_rate": 1.9998809684187638e-05,
      "loss": 4.8105,
      "step": 862
    },
    {
      "epoch": 0.034840532902704884,
      "grad_norm": 170.38584899902344,
      "learning_rate": 1.999878942426552e-05,
      "loss": 6.6339,
      "step": 863
    },
    {
      "epoch": 0.034880904319741626,
      "grad_norm": 167.00146484375,
      "learning_rate": 1.9998768993387455e-05,
      "loss": 4.9121,
      "step": 864
    },
    {
      "epoch": 0.03492127573677836,
      "grad_norm": 207.1678009033203,
      "learning_rate": 1.9998748391553784e-05,
      "loss": 6.4157,
      "step": 865
    },
    {
      "epoch": 0.0349616471538151,
      "grad_norm": 105.9411849975586,
      "learning_rate": 1.9998727618764865e-05,
      "loss": 4.8144,
      "step": 866
    },
    {
      "epoch": 0.035002018570851834,
      "grad_norm": 174.37327575683594,
      "learning_rate": 1.9998706675021052e-05,
      "loss": 6.2905,
      "step": 867
    },
    {
      "epoch": 0.03504238998788858,
      "grad_norm": 227.50369262695312,
      "learning_rate": 1.9998685560322707e-05,
      "loss": 5.5942,
      "step": 868
    },
    {
      "epoch": 0.03508276140492531,
      "grad_norm": 144.28021240234375,
      "learning_rate": 1.9998664274670186e-05,
      "loss": 5.6111,
      "step": 869
    },
    {
      "epoch": 0.03512313282196205,
      "grad_norm": 133.5569305419922,
      "learning_rate": 1.999864281806385e-05,
      "loss": 5.0574,
      "step": 870
    },
    {
      "epoch": 0.03516350423899879,
      "grad_norm": 187.4175262451172,
      "learning_rate": 1.999862119050408e-05,
      "loss": 6.5224,
      "step": 871
    },
    {
      "epoch": 0.03520387565603553,
      "grad_norm": 127.78020477294922,
      "learning_rate": 1.9998599391991228e-05,
      "loss": 4.2677,
      "step": 872
    },
    {
      "epoch": 0.03524424707307226,
      "grad_norm": 147.3642120361328,
      "learning_rate": 1.9998577422525674e-05,
      "loss": 7.7485,
      "step": 873
    },
    {
      "epoch": 0.035284618490109006,
      "grad_norm": 159.56805419921875,
      "learning_rate": 1.9998555282107798e-05,
      "loss": 5.4914,
      "step": 874
    },
    {
      "epoch": 0.03532498990714574,
      "grad_norm": 143.84815979003906,
      "learning_rate": 1.9998532970737975e-05,
      "loss": 6.658,
      "step": 875
    },
    {
      "epoch": 0.03536536132418248,
      "grad_norm": 151.6565399169922,
      "learning_rate": 1.9998510488416588e-05,
      "loss": 5.4857,
      "step": 876
    },
    {
      "epoch": 0.03540573274121922,
      "grad_norm": 178.61122131347656,
      "learning_rate": 1.9998487835144016e-05,
      "loss": 5.7644,
      "step": 877
    },
    {
      "epoch": 0.035446104158255956,
      "grad_norm": 153.80569458007812,
      "learning_rate": 1.9998465010920653e-05,
      "loss": 5.8891,
      "step": 878
    },
    {
      "epoch": 0.03548647557529269,
      "grad_norm": 147.05599975585938,
      "learning_rate": 1.9998442015746888e-05,
      "loss": 4.7798,
      "step": 879
    },
    {
      "epoch": 0.03552684699232943,
      "grad_norm": 183.11172485351562,
      "learning_rate": 1.9998418849623107e-05,
      "loss": 6.1611,
      "step": 880
    },
    {
      "epoch": 0.03556721840936617,
      "grad_norm": 255.2064666748047,
      "learning_rate": 1.9998395512549714e-05,
      "loss": 6.4539,
      "step": 881
    },
    {
      "epoch": 0.035607589826402906,
      "grad_norm": 103.7615966796875,
      "learning_rate": 1.9998372004527108e-05,
      "loss": 4.6862,
      "step": 882
    },
    {
      "epoch": 0.03564796124343964,
      "grad_norm": 853.4181518554688,
      "learning_rate": 1.9998348325555686e-05,
      "loss": 5.9327,
      "step": 883
    },
    {
      "epoch": 0.035688332660476385,
      "grad_norm": 111.24153137207031,
      "learning_rate": 1.9998324475635854e-05,
      "loss": 7.1855,
      "step": 884
    },
    {
      "epoch": 0.03572870407751312,
      "grad_norm": 110.569580078125,
      "learning_rate": 1.9998300454768027e-05,
      "loss": 6.7541,
      "step": 885
    },
    {
      "epoch": 0.03576907549454986,
      "grad_norm": 168.54840087890625,
      "learning_rate": 1.9998276262952606e-05,
      "loss": 6.0036,
      "step": 886
    },
    {
      "epoch": 0.0358094469115866,
      "grad_norm": 131.44943237304688,
      "learning_rate": 1.9998251900190008e-05,
      "loss": 4.2223,
      "step": 887
    },
    {
      "epoch": 0.035849818328623335,
      "grad_norm": 131.08709716796875,
      "learning_rate": 1.999822736648065e-05,
      "loss": 5.4604,
      "step": 888
    },
    {
      "epoch": 0.03589018974566007,
      "grad_norm": 98.04330444335938,
      "learning_rate": 1.9998202661824955e-05,
      "loss": 6.9914,
      "step": 889
    },
    {
      "epoch": 0.035930561162696814,
      "grad_norm": 99.44563293457031,
      "learning_rate": 1.9998177786223342e-05,
      "loss": 6.5469,
      "step": 890
    },
    {
      "epoch": 0.03597093257973355,
      "grad_norm": 6854.75927734375,
      "learning_rate": 1.999815273967623e-05,
      "loss": 6.031,
      "step": 891
    },
    {
      "epoch": 0.036011303996770286,
      "grad_norm": 134.90478515625,
      "learning_rate": 1.999812752218406e-05,
      "loss": 4.6541,
      "step": 892
    },
    {
      "epoch": 0.03605167541380702,
      "grad_norm": 154.25193786621094,
      "learning_rate": 1.9998102133747254e-05,
      "loss": 5.6037,
      "step": 893
    },
    {
      "epoch": 0.036092046830843764,
      "grad_norm": 136.3806610107422,
      "learning_rate": 1.9998076574366247e-05,
      "loss": 5.7624,
      "step": 894
    },
    {
      "epoch": 0.0361324182478805,
      "grad_norm": 164.82949829101562,
      "learning_rate": 1.9998050844041482e-05,
      "loss": 6.7703,
      "step": 895
    },
    {
      "epoch": 0.036172789664917236,
      "grad_norm": 177.2495880126953,
      "learning_rate": 1.999802494277339e-05,
      "loss": 5.0122,
      "step": 896
    },
    {
      "epoch": 0.03621316108195398,
      "grad_norm": 116.48658752441406,
      "learning_rate": 1.9997998870562423e-05,
      "loss": 4.6127,
      "step": 897
    },
    {
      "epoch": 0.036253532498990715,
      "grad_norm": 113.8570556640625,
      "learning_rate": 1.9997972627409018e-05,
      "loss": 6.0257,
      "step": 898
    },
    {
      "epoch": 0.03629390391602745,
      "grad_norm": 144.25025939941406,
      "learning_rate": 1.999794621331363e-05,
      "loss": 6.6199,
      "step": 899
    },
    {
      "epoch": 0.036334275333064193,
      "grad_norm": 188.1676025390625,
      "learning_rate": 1.999791962827671e-05,
      "loss": 5.9697,
      "step": 900
    },
    {
      "epoch": 0.03637464675010093,
      "grad_norm": 147.04656982421875,
      "learning_rate": 1.999789287229871e-05,
      "loss": 7.0125,
      "step": 901
    },
    {
      "epoch": 0.036415018167137665,
      "grad_norm": 100.23082733154297,
      "learning_rate": 1.9997865945380087e-05,
      "loss": 4.7269,
      "step": 902
    },
    {
      "epoch": 0.0364553895841744,
      "grad_norm": 132.87904357910156,
      "learning_rate": 1.9997838847521307e-05,
      "loss": 6.7407,
      "step": 903
    },
    {
      "epoch": 0.036495761001211144,
      "grad_norm": 191.15956115722656,
      "learning_rate": 1.9997811578722827e-05,
      "loss": 5.1924,
      "step": 904
    },
    {
      "epoch": 0.03653613241824788,
      "grad_norm": 144.55369567871094,
      "learning_rate": 1.9997784138985116e-05,
      "loss": 5.4168,
      "step": 905
    },
    {
      "epoch": 0.036576503835284616,
      "grad_norm": 138.95738220214844,
      "learning_rate": 1.9997756528308642e-05,
      "loss": 6.649,
      "step": 906
    },
    {
      "epoch": 0.03661687525232136,
      "grad_norm": 122.36380767822266,
      "learning_rate": 1.9997728746693878e-05,
      "loss": 6.8238,
      "step": 907
    },
    {
      "epoch": 0.036657246669358094,
      "grad_norm": 193.75323486328125,
      "learning_rate": 1.99977007941413e-05,
      "loss": 6.8931,
      "step": 908
    },
    {
      "epoch": 0.03669761808639483,
      "grad_norm": 157.1030731201172,
      "learning_rate": 1.9997672670651384e-05,
      "loss": 6.5694,
      "step": 909
    },
    {
      "epoch": 0.03673798950343157,
      "grad_norm": 164.32247924804688,
      "learning_rate": 1.999764437622461e-05,
      "loss": 7.6163,
      "step": 910
    },
    {
      "epoch": 0.03677836092046831,
      "grad_norm": 198.592529296875,
      "learning_rate": 1.9997615910861465e-05,
      "loss": 6.4092,
      "step": 911
    },
    {
      "epoch": 0.036818732337505045,
      "grad_norm": 143.24327087402344,
      "learning_rate": 1.9997587274562434e-05,
      "loss": 6.7707,
      "step": 912
    },
    {
      "epoch": 0.03685910375454179,
      "grad_norm": 97.30104064941406,
      "learning_rate": 1.9997558467328004e-05,
      "loss": 6.297,
      "step": 913
    },
    {
      "epoch": 0.03689947517157852,
      "grad_norm": 137.4681854248047,
      "learning_rate": 1.9997529489158673e-05,
      "loss": 5.6341,
      "step": 914
    },
    {
      "epoch": 0.03693984658861526,
      "grad_norm": 156.59657287597656,
      "learning_rate": 1.9997500340054936e-05,
      "loss": 4.1778,
      "step": 915
    },
    {
      "epoch": 0.036980218005651995,
      "grad_norm": 127.64611053466797,
      "learning_rate": 1.9997471020017283e-05,
      "loss": 6.7661,
      "step": 916
    },
    {
      "epoch": 0.03702058942268874,
      "grad_norm": 118.0228042602539,
      "learning_rate": 1.9997441529046226e-05,
      "loss": 4.8889,
      "step": 917
    },
    {
      "epoch": 0.037060960839725474,
      "grad_norm": 123.61502075195312,
      "learning_rate": 1.999741186714226e-05,
      "loss": 5.7811,
      "step": 918
    },
    {
      "epoch": 0.03710133225676221,
      "grad_norm": 150.52932739257812,
      "learning_rate": 1.9997382034305902e-05,
      "loss": 5.6449,
      "step": 919
    },
    {
      "epoch": 0.03714170367379895,
      "grad_norm": 127.29996490478516,
      "learning_rate": 1.9997352030537657e-05,
      "loss": 6.391,
      "step": 920
    },
    {
      "epoch": 0.03718207509083569,
      "grad_norm": 128.7767791748047,
      "learning_rate": 1.9997321855838034e-05,
      "loss": 5.6335,
      "step": 921
    },
    {
      "epoch": 0.037222446507872424,
      "grad_norm": 128.91001892089844,
      "learning_rate": 1.9997291510207556e-05,
      "loss": 6.7215,
      "step": 922
    },
    {
      "epoch": 0.03726281792490917,
      "grad_norm": 114.8734359741211,
      "learning_rate": 1.9997260993646734e-05,
      "loss": 7.087,
      "step": 923
    },
    {
      "epoch": 0.0373031893419459,
      "grad_norm": 128.84893798828125,
      "learning_rate": 1.9997230306156096e-05,
      "loss": 6.2939,
      "step": 924
    },
    {
      "epoch": 0.03734356075898264,
      "grad_norm": 100.06632995605469,
      "learning_rate": 1.9997199447736167e-05,
      "loss": 5.6185,
      "step": 925
    },
    {
      "epoch": 0.03738393217601938,
      "grad_norm": 197.33523559570312,
      "learning_rate": 1.999716841838747e-05,
      "loss": 6.6096,
      "step": 926
    },
    {
      "epoch": 0.03742430359305612,
      "grad_norm": 467.89752197265625,
      "learning_rate": 1.999713721811054e-05,
      "loss": 5.7008,
      "step": 927
    },
    {
      "epoch": 0.03746467501009285,
      "grad_norm": 133.04519653320312,
      "learning_rate": 1.9997105846905908e-05,
      "loss": 7.1016,
      "step": 928
    },
    {
      "epoch": 0.03750504642712959,
      "grad_norm": 170.48391723632812,
      "learning_rate": 1.999707430477411e-05,
      "loss": 6.604,
      "step": 929
    },
    {
      "epoch": 0.03754541784416633,
      "grad_norm": 129.48193359375,
      "learning_rate": 1.999704259171569e-05,
      "loss": 6.4091,
      "step": 930
    },
    {
      "epoch": 0.03758578926120307,
      "grad_norm": 99.29751586914062,
      "learning_rate": 1.9997010707731182e-05,
      "loss": 6.9417,
      "step": 931
    },
    {
      "epoch": 0.0376261606782398,
      "grad_norm": 309.6905517578125,
      "learning_rate": 1.999697865282114e-05,
      "loss": 6.2672,
      "step": 932
    },
    {
      "epoch": 0.037666532095276546,
      "grad_norm": 125.03262329101562,
      "learning_rate": 1.9996946426986104e-05,
      "loss": 6.0682,
      "step": 933
    },
    {
      "epoch": 0.03770690351231328,
      "grad_norm": 120.32252502441406,
      "learning_rate": 1.9996914030226628e-05,
      "loss": 6.4274,
      "step": 934
    },
    {
      "epoch": 0.03774727492935002,
      "grad_norm": 118.69168090820312,
      "learning_rate": 1.999688146254327e-05,
      "loss": 5.3834,
      "step": 935
    },
    {
      "epoch": 0.03778764634638676,
      "grad_norm": 127.2247543334961,
      "learning_rate": 1.9996848723936582e-05,
      "loss": 6.0703,
      "step": 936
    },
    {
      "epoch": 0.0378280177634235,
      "grad_norm": 104.85906982421875,
      "learning_rate": 1.9996815814407124e-05,
      "loss": 5.1061,
      "step": 937
    },
    {
      "epoch": 0.03786838918046023,
      "grad_norm": 115.37767028808594,
      "learning_rate": 1.999678273395546e-05,
      "loss": 6.6547,
      "step": 938
    },
    {
      "epoch": 0.037908760597496975,
      "grad_norm": 121.0619125366211,
      "learning_rate": 1.9996749482582156e-05,
      "loss": 5.1076,
      "step": 939
    },
    {
      "epoch": 0.03794913201453371,
      "grad_norm": 133.4525604248047,
      "learning_rate": 1.999671606028778e-05,
      "loss": 5.9587,
      "step": 940
    },
    {
      "epoch": 0.03798950343157045,
      "grad_norm": 89.16769409179688,
      "learning_rate": 1.9996682467072902e-05,
      "loss": 6.1054,
      "step": 941
    },
    {
      "epoch": 0.03802987484860718,
      "grad_norm": 147.2839813232422,
      "learning_rate": 1.99966487029381e-05,
      "loss": 5.2117,
      "step": 942
    },
    {
      "epoch": 0.038070246265643926,
      "grad_norm": 149.19776916503906,
      "learning_rate": 1.999661476788395e-05,
      "loss": 6.3481,
      "step": 943
    },
    {
      "epoch": 0.03811061768268066,
      "grad_norm": 139.96934509277344,
      "learning_rate": 1.9996580661911026e-05,
      "loss": 6.1176,
      "step": 944
    },
    {
      "epoch": 0.0381509890997174,
      "grad_norm": 139.48316955566406,
      "learning_rate": 1.999654638501992e-05,
      "loss": 5.4006,
      "step": 945
    },
    {
      "epoch": 0.03819136051675414,
      "grad_norm": 132.18397521972656,
      "learning_rate": 1.9996511937211216e-05,
      "loss": 5.6754,
      "step": 946
    },
    {
      "epoch": 0.038231731933790876,
      "grad_norm": 281.3278503417969,
      "learning_rate": 1.9996477318485497e-05,
      "loss": 6.1551,
      "step": 947
    },
    {
      "epoch": 0.03827210335082761,
      "grad_norm": 168.34584045410156,
      "learning_rate": 1.9996442528843363e-05,
      "loss": 6.3553,
      "step": 948
    },
    {
      "epoch": 0.038312474767864355,
      "grad_norm": 164.53787231445312,
      "learning_rate": 1.9996407568285402e-05,
      "loss": 5.081,
      "step": 949
    },
    {
      "epoch": 0.03835284618490109,
      "grad_norm": 128.21112060546875,
      "learning_rate": 1.9996372436812218e-05,
      "loss": 5.3499,
      "step": 950
    },
    {
      "epoch": 0.038393217601937826,
      "grad_norm": 101.1803207397461,
      "learning_rate": 1.9996337134424408e-05,
      "loss": 4.8453,
      "step": 951
    },
    {
      "epoch": 0.03843358901897457,
      "grad_norm": 138.21238708496094,
      "learning_rate": 1.9996301661122578e-05,
      "loss": 6.9759,
      "step": 952
    },
    {
      "epoch": 0.038473960436011305,
      "grad_norm": 167.57666015625,
      "learning_rate": 1.9996266016907327e-05,
      "loss": 5.7166,
      "step": 953
    },
    {
      "epoch": 0.03851433185304804,
      "grad_norm": 152.6741485595703,
      "learning_rate": 1.9996230201779273e-05,
      "loss": 5.0386,
      "step": 954
    },
    {
      "epoch": 0.03855470327008478,
      "grad_norm": 143.95736694335938,
      "learning_rate": 1.9996194215739028e-05,
      "loss": 7.0341,
      "step": 955
    },
    {
      "epoch": 0.03859507468712152,
      "grad_norm": 125.93714141845703,
      "learning_rate": 1.9996158058787202e-05,
      "loss": 6.1694,
      "step": 956
    },
    {
      "epoch": 0.038635446104158255,
      "grad_norm": 124.6629867553711,
      "learning_rate": 1.9996121730924412e-05,
      "loss": 5.3701,
      "step": 957
    },
    {
      "epoch": 0.03867581752119499,
      "grad_norm": 85.90052032470703,
      "learning_rate": 1.999608523215129e-05,
      "loss": 5.6298,
      "step": 958
    },
    {
      "epoch": 0.038716188938231734,
      "grad_norm": 110.55410766601562,
      "learning_rate": 1.9996048562468452e-05,
      "loss": 5.8688,
      "step": 959
    },
    {
      "epoch": 0.03875656035526847,
      "grad_norm": 103.70042419433594,
      "learning_rate": 1.9996011721876524e-05,
      "loss": 6.2996,
      "step": 960
    },
    {
      "epoch": 0.038796931772305206,
      "grad_norm": 134.8916015625,
      "learning_rate": 1.9995974710376136e-05,
      "loss": 5.5916,
      "step": 961
    },
    {
      "epoch": 0.03883730318934195,
      "grad_norm": 146.52554321289062,
      "learning_rate": 1.9995937527967925e-05,
      "loss": 4.897,
      "step": 962
    },
    {
      "epoch": 0.038877674606378684,
      "grad_norm": 122.42878723144531,
      "learning_rate": 1.9995900174652525e-05,
      "loss": 6.1402,
      "step": 963
    },
    {
      "epoch": 0.03891804602341542,
      "grad_norm": 88.62979888916016,
      "learning_rate": 1.999586265043057e-05,
      "loss": 4.6829,
      "step": 964
    },
    {
      "epoch": 0.03895841744045216,
      "grad_norm": 338.20709228515625,
      "learning_rate": 1.9995824955302713e-05,
      "loss": 5.4198,
      "step": 965
    },
    {
      "epoch": 0.0389987888574889,
      "grad_norm": 194.75674438476562,
      "learning_rate": 1.999578708926959e-05,
      "loss": 4.6442,
      "step": 966
    },
    {
      "epoch": 0.039039160274525635,
      "grad_norm": 91.43264770507812,
      "learning_rate": 1.9995749052331847e-05,
      "loss": 5.8003,
      "step": 967
    },
    {
      "epoch": 0.03907953169156237,
      "grad_norm": 108.68960571289062,
      "learning_rate": 1.9995710844490134e-05,
      "loss": 5.6296,
      "step": 968
    },
    {
      "epoch": 0.03911990310859911,
      "grad_norm": 157.9378662109375,
      "learning_rate": 1.999567246574511e-05,
      "loss": 5.2767,
      "step": 969
    },
    {
      "epoch": 0.03916027452563585,
      "grad_norm": 101.71186065673828,
      "learning_rate": 1.999563391609743e-05,
      "loss": 5.5163,
      "step": 970
    },
    {
      "epoch": 0.039200645942672585,
      "grad_norm": 117.37360382080078,
      "learning_rate": 1.999559519554775e-05,
      "loss": 6.3668,
      "step": 971
    },
    {
      "epoch": 0.03924101735970933,
      "grad_norm": 149.66427612304688,
      "learning_rate": 1.9995556304096736e-05,
      "loss": 3.6827,
      "step": 972
    },
    {
      "epoch": 0.039281388776746064,
      "grad_norm": 105.66918182373047,
      "learning_rate": 1.9995517241745045e-05,
      "loss": 5.0998,
      "step": 973
    },
    {
      "epoch": 0.0393217601937828,
      "grad_norm": 85.28500366210938,
      "learning_rate": 1.9995478008493354e-05,
      "loss": 5.542,
      "step": 974
    },
    {
      "epoch": 0.03936213161081954,
      "grad_norm": 113.61692810058594,
      "learning_rate": 1.999543860434233e-05,
      "loss": 6.3258,
      "step": 975
    },
    {
      "epoch": 0.03940250302785628,
      "grad_norm": 117.71403503417969,
      "learning_rate": 1.9995399029292648e-05,
      "loss": 7.1242,
      "step": 976
    },
    {
      "epoch": 0.039442874444893014,
      "grad_norm": 109.2782974243164,
      "learning_rate": 1.9995359283344982e-05,
      "loss": 6.3207,
      "step": 977
    },
    {
      "epoch": 0.03948324586192976,
      "grad_norm": 199.2484588623047,
      "learning_rate": 1.9995319366500013e-05,
      "loss": 5.3539,
      "step": 978
    },
    {
      "epoch": 0.03952361727896649,
      "grad_norm": 90.01123046875,
      "learning_rate": 1.999527927875842e-05,
      "loss": 6.6988,
      "step": 979
    },
    {
      "epoch": 0.03956398869600323,
      "grad_norm": 210.77865600585938,
      "learning_rate": 1.99952390201209e-05,
      "loss": 5.7963,
      "step": 980
    },
    {
      "epoch": 0.039604360113039964,
      "grad_norm": 288.1502685546875,
      "learning_rate": 1.999519859058813e-05,
      "loss": 6.5536,
      "step": 981
    },
    {
      "epoch": 0.03964473153007671,
      "grad_norm": 97.8677978515625,
      "learning_rate": 1.9995157990160804e-05,
      "loss": 5.6687,
      "step": 982
    },
    {
      "epoch": 0.03968510294711344,
      "grad_norm": 121.72075653076172,
      "learning_rate": 1.9995117218839616e-05,
      "loss": 6.5576,
      "step": 983
    },
    {
      "epoch": 0.03972547436415018,
      "grad_norm": 144.9723663330078,
      "learning_rate": 1.9995076276625262e-05,
      "loss": 6.3629,
      "step": 984
    },
    {
      "epoch": 0.03976584578118692,
      "grad_norm": 138.28543090820312,
      "learning_rate": 1.9995035163518447e-05,
      "loss": 5.7012,
      "step": 985
    },
    {
      "epoch": 0.03980621719822366,
      "grad_norm": 160.4176025390625,
      "learning_rate": 1.9994993879519868e-05,
      "loss": 5.0687,
      "step": 986
    },
    {
      "epoch": 0.039846588615260393,
      "grad_norm": 145.7481231689453,
      "learning_rate": 1.9994952424630237e-05,
      "loss": 5.5478,
      "step": 987
    },
    {
      "epoch": 0.039886960032297136,
      "grad_norm": 139.82151794433594,
      "learning_rate": 1.999491079885026e-05,
      "loss": 6.8898,
      "step": 988
    },
    {
      "epoch": 0.03992733144933387,
      "grad_norm": 129.06231689453125,
      "learning_rate": 1.9994869002180645e-05,
      "loss": 5.6311,
      "step": 989
    },
    {
      "epoch": 0.03996770286637061,
      "grad_norm": 174.01962280273438,
      "learning_rate": 1.999482703462211e-05,
      "loss": 6.4049,
      "step": 990
    },
    {
      "epoch": 0.04000807428340735,
      "grad_norm": 145.07606506347656,
      "learning_rate": 1.9994784896175373e-05,
      "loss": 5.3758,
      "step": 991
    },
    {
      "epoch": 0.04004844570044409,
      "grad_norm": 155.2579803466797,
      "learning_rate": 1.9994742586841155e-05,
      "loss": 6.2345,
      "step": 992
    },
    {
      "epoch": 0.04008881711748082,
      "grad_norm": 200.17706298828125,
      "learning_rate": 1.9994700106620178e-05,
      "loss": 5.3774,
      "step": 993
    },
    {
      "epoch": 0.04012918853451756,
      "grad_norm": 3837.936279296875,
      "learning_rate": 1.9994657455513166e-05,
      "loss": 6.2526,
      "step": 994
    },
    {
      "epoch": 0.0401695599515543,
      "grad_norm": 113.0466537475586,
      "learning_rate": 1.9994614633520856e-05,
      "loss": 3.3473,
      "step": 995
    },
    {
      "epoch": 0.04020993136859104,
      "grad_norm": 156.19361877441406,
      "learning_rate": 1.9994571640643968e-05,
      "loss": 6.0611,
      "step": 996
    },
    {
      "epoch": 0.04025030278562777,
      "grad_norm": 183.8043212890625,
      "learning_rate": 1.999452847688325e-05,
      "loss": 6.81,
      "step": 997
    },
    {
      "epoch": 0.040290674202664516,
      "grad_norm": 111.61248779296875,
      "learning_rate": 1.999448514223943e-05,
      "loss": 6.0979,
      "step": 998
    },
    {
      "epoch": 0.04033104561970125,
      "grad_norm": 119.92034149169922,
      "learning_rate": 1.9994441636713252e-05,
      "loss": 5.2906,
      "step": 999
    },
    {
      "epoch": 0.04037141703673799,
      "grad_norm": 137.87997436523438,
      "learning_rate": 1.9994397960305462e-05,
      "loss": 4.7445,
      "step": 1000
    },
    {
      "epoch": 0.04041178845377473,
      "grad_norm": 102.00938415527344,
      "learning_rate": 1.9994354113016808e-05,
      "loss": 5.684,
      "step": 1001
    },
    {
      "epoch": 0.040452159870811466,
      "grad_norm": 255.00404357910156,
      "learning_rate": 1.9994310094848036e-05,
      "loss": 6.2538,
      "step": 1002
    },
    {
      "epoch": 0.0404925312878482,
      "grad_norm": 117.60134887695312,
      "learning_rate": 1.9994265905799897e-05,
      "loss": 4.4035,
      "step": 1003
    },
    {
      "epoch": 0.040532902704884945,
      "grad_norm": 124.43950653076172,
      "learning_rate": 1.999422154587315e-05,
      "loss": 7.2584,
      "step": 1004
    },
    {
      "epoch": 0.04057327412192168,
      "grad_norm": 135.688232421875,
      "learning_rate": 1.9994177015068556e-05,
      "loss": 6.0761,
      "step": 1005
    },
    {
      "epoch": 0.040613645538958416,
      "grad_norm": 161.12353515625,
      "learning_rate": 1.9994132313386872e-05,
      "loss": 5.5247,
      "step": 1006
    },
    {
      "epoch": 0.04065401695599515,
      "grad_norm": 220.08143615722656,
      "learning_rate": 1.999408744082886e-05,
      "loss": 6.3625,
      "step": 1007
    },
    {
      "epoch": 0.040694388373031895,
      "grad_norm": 112.01996612548828,
      "learning_rate": 1.9994042397395296e-05,
      "loss": 6.3885,
      "step": 1008
    },
    {
      "epoch": 0.04073475979006863,
      "grad_norm": 122.0446548461914,
      "learning_rate": 1.999399718308694e-05,
      "loss": 3.8992,
      "step": 1009
    },
    {
      "epoch": 0.04077513120710537,
      "grad_norm": 198.57371520996094,
      "learning_rate": 1.9993951797904573e-05,
      "loss": 3.8662,
      "step": 1010
    },
    {
      "epoch": 0.04081550262414211,
      "grad_norm": 130.92259216308594,
      "learning_rate": 1.999390624184897e-05,
      "loss": 8.0121,
      "step": 1011
    },
    {
      "epoch": 0.040855874041178845,
      "grad_norm": 144.07186889648438,
      "learning_rate": 1.9993860514920903e-05,
      "loss": 5.5241,
      "step": 1012
    },
    {
      "epoch": 0.04089624545821558,
      "grad_norm": 107.11290740966797,
      "learning_rate": 1.9993814617121158e-05,
      "loss": 5.507,
      "step": 1013
    },
    {
      "epoch": 0.040936616875252324,
      "grad_norm": 176.821533203125,
      "learning_rate": 1.9993768548450524e-05,
      "loss": 6.052,
      "step": 1014
    },
    {
      "epoch": 0.04097698829228906,
      "grad_norm": 141.5400848388672,
      "learning_rate": 1.9993722308909783e-05,
      "loss": 6.377,
      "step": 1015
    },
    {
      "epoch": 0.041017359709325796,
      "grad_norm": 131.45376586914062,
      "learning_rate": 1.999367589849973e-05,
      "loss": 4.6594,
      "step": 1016
    },
    {
      "epoch": 0.04105773112636254,
      "grad_norm": 162.34950256347656,
      "learning_rate": 1.9993629317221154e-05,
      "loss": 6.831,
      "step": 1017
    },
    {
      "epoch": 0.041098102543399274,
      "grad_norm": 100.12068939208984,
      "learning_rate": 1.999358256507485e-05,
      "loss": 6.7989,
      "step": 1018
    },
    {
      "epoch": 0.04113847396043601,
      "grad_norm": 120.42552947998047,
      "learning_rate": 1.9993535642061626e-05,
      "loss": 6.1127,
      "step": 1019
    },
    {
      "epoch": 0.041178845377472746,
      "grad_norm": 78.30921936035156,
      "learning_rate": 1.9993488548182276e-05,
      "loss": 6.2417,
      "step": 1020
    },
    {
      "epoch": 0.04121921679450949,
      "grad_norm": 169.60047912597656,
      "learning_rate": 1.999344128343761e-05,
      "loss": 5.0476,
      "step": 1021
    },
    {
      "epoch": 0.041259588211546225,
      "grad_norm": 141.7161102294922,
      "learning_rate": 1.9993393847828435e-05,
      "loss": 5.8175,
      "step": 1022
    },
    {
      "epoch": 0.04129995962858296,
      "grad_norm": 93.63201141357422,
      "learning_rate": 1.999334624135556e-05,
      "loss": 5.7474,
      "step": 1023
    },
    {
      "epoch": 0.0413403310456197,
      "grad_norm": 126.76483154296875,
      "learning_rate": 1.9993298464019802e-05,
      "loss": 4.5523,
      "step": 1024
    },
    {
      "epoch": 0.04138070246265644,
      "grad_norm": 151.1031494140625,
      "learning_rate": 1.9993250515821974e-05,
      "loss": 4.6177,
      "step": 1025
    },
    {
      "epoch": 0.041421073879693175,
      "grad_norm": 83.42364501953125,
      "learning_rate": 1.99932023967629e-05,
      "loss": 5.4423,
      "step": 1026
    },
    {
      "epoch": 0.04146144529672992,
      "grad_norm": 335.41278076171875,
      "learning_rate": 1.99931541068434e-05,
      "loss": 5.7193,
      "step": 1027
    },
    {
      "epoch": 0.041501816713766654,
      "grad_norm": 104.93643188476562,
      "learning_rate": 1.9993105646064298e-05,
      "loss": 5.714,
      "step": 1028
    },
    {
      "epoch": 0.04154218813080339,
      "grad_norm": 143.9462127685547,
      "learning_rate": 1.999305701442643e-05,
      "loss": 6.2448,
      "step": 1029
    },
    {
      "epoch": 0.04158255954784013,
      "grad_norm": 95.14167022705078,
      "learning_rate": 1.9993008211930617e-05,
      "loss": 6.8909,
      "step": 1030
    },
    {
      "epoch": 0.04162293096487687,
      "grad_norm": 174.93711853027344,
      "learning_rate": 1.9992959238577707e-05,
      "loss": 6.6198,
      "step": 1031
    },
    {
      "epoch": 0.041663302381913604,
      "grad_norm": 128.9568328857422,
      "learning_rate": 1.999291009436852e-05,
      "loss": 6.2641,
      "step": 1032
    },
    {
      "epoch": 0.04170367379895034,
      "grad_norm": 89.43922424316406,
      "learning_rate": 1.9992860779303912e-05,
      "loss": 5.5269,
      "step": 1033
    },
    {
      "epoch": 0.04174404521598708,
      "grad_norm": 103.56703186035156,
      "learning_rate": 1.9992811293384714e-05,
      "loss": 4.853,
      "step": 1034
    },
    {
      "epoch": 0.04178441663302382,
      "grad_norm": 146.66673278808594,
      "learning_rate": 1.9992761636611782e-05,
      "loss": 6.941,
      "step": 1035
    },
    {
      "epoch": 0.041824788050060555,
      "grad_norm": 172.77175903320312,
      "learning_rate": 1.999271180898596e-05,
      "loss": 5.2455,
      "step": 1036
    },
    {
      "epoch": 0.0418651594670973,
      "grad_norm": 192.4292449951172,
      "learning_rate": 1.99926618105081e-05,
      "loss": 6.6528,
      "step": 1037
    },
    {
      "epoch": 0.04190553088413403,
      "grad_norm": 183.29483032226562,
      "learning_rate": 1.999261164117906e-05,
      "loss": 5.2244,
      "step": 1038
    },
    {
      "epoch": 0.04194590230117077,
      "grad_norm": 139.70034790039062,
      "learning_rate": 1.9992561300999692e-05,
      "loss": 6.1491,
      "step": 1039
    },
    {
      "epoch": 0.04198627371820751,
      "grad_norm": 119.01409912109375,
      "learning_rate": 1.9992510789970865e-05,
      "loss": 5.5254,
      "step": 1040
    },
    {
      "epoch": 0.04202664513524425,
      "grad_norm": 112.80599975585938,
      "learning_rate": 1.999246010809343e-05,
      "loss": 5.0487,
      "step": 1041
    },
    {
      "epoch": 0.042067016552280984,
      "grad_norm": 159.04397583007812,
      "learning_rate": 1.9992409255368267e-05,
      "loss": 5.7583,
      "step": 1042
    },
    {
      "epoch": 0.042107387969317726,
      "grad_norm": 112.71247100830078,
      "learning_rate": 1.9992358231796238e-05,
      "loss": 7.576,
      "step": 1043
    },
    {
      "epoch": 0.04214775938635446,
      "grad_norm": 115.05451965332031,
      "learning_rate": 1.9992307037378216e-05,
      "loss": 6.158,
      "step": 1044
    },
    {
      "epoch": 0.0421881308033912,
      "grad_norm": 107.19140625,
      "learning_rate": 1.9992255672115082e-05,
      "loss": 4.3525,
      "step": 1045
    },
    {
      "epoch": 0.042228502220427934,
      "grad_norm": 180.7394256591797,
      "learning_rate": 1.9992204136007707e-05,
      "loss": 7.2866,
      "step": 1046
    },
    {
      "epoch": 0.04226887363746468,
      "grad_norm": 166.24951171875,
      "learning_rate": 1.9992152429056974e-05,
      "loss": 5.9512,
      "step": 1047
    },
    {
      "epoch": 0.04230924505450141,
      "grad_norm": 315.093505859375,
      "learning_rate": 1.9992100551263768e-05,
      "loss": 7.0883,
      "step": 1048
    },
    {
      "epoch": 0.04234961647153815,
      "grad_norm": 192.5592803955078,
      "learning_rate": 1.9992048502628977e-05,
      "loss": 6.495,
      "step": 1049
    },
    {
      "epoch": 0.04238998788857489,
      "grad_norm": 178.5362091064453,
      "learning_rate": 1.999199628315349e-05,
      "loss": 5.7025,
      "step": 1050
    },
    {
      "epoch": 0.04243035930561163,
      "grad_norm": 109.99882507324219,
      "learning_rate": 1.9991943892838194e-05,
      "loss": 5.7692,
      "step": 1051
    },
    {
      "epoch": 0.04247073072264836,
      "grad_norm": 126.00344848632812,
      "learning_rate": 1.9991891331683993e-05,
      "loss": 5.2244,
      "step": 1052
    },
    {
      "epoch": 0.042511102139685106,
      "grad_norm": 182.3551025390625,
      "learning_rate": 1.9991838599691786e-05,
      "loss": 5.6908,
      "step": 1053
    },
    {
      "epoch": 0.04255147355672184,
      "grad_norm": 144.2254638671875,
      "learning_rate": 1.999178569686247e-05,
      "loss": 5.25,
      "step": 1054
    },
    {
      "epoch": 0.04259184497375858,
      "grad_norm": 159.97348022460938,
      "learning_rate": 1.999173262319695e-05,
      "loss": 6.048,
      "step": 1055
    },
    {
      "epoch": 0.04263221639079532,
      "grad_norm": 175.5627899169922,
      "learning_rate": 1.9991679378696135e-05,
      "loss": 6.2981,
      "step": 1056
    },
    {
      "epoch": 0.042672587807832056,
      "grad_norm": 107.17809295654297,
      "learning_rate": 1.9991625963360937e-05,
      "loss": 6.279,
      "step": 1057
    },
    {
      "epoch": 0.04271295922486879,
      "grad_norm": 124.8489990234375,
      "learning_rate": 1.9991572377192266e-05,
      "loss": 5.1624,
      "step": 1058
    },
    {
      "epoch": 0.04275333064190553,
      "grad_norm": 283.9469299316406,
      "learning_rate": 1.9991518620191038e-05,
      "loss": 6.1413,
      "step": 1059
    },
    {
      "epoch": 0.04279370205894227,
      "grad_norm": 94.5099105834961,
      "learning_rate": 1.9991464692358177e-05,
      "loss": 6.3815,
      "step": 1060
    },
    {
      "epoch": 0.042834073475979006,
      "grad_norm": 309.3096618652344,
      "learning_rate": 1.9991410593694595e-05,
      "loss": 6.4084,
      "step": 1061
    },
    {
      "epoch": 0.04287444489301574,
      "grad_norm": 348.6915588378906,
      "learning_rate": 1.999135632420123e-05,
      "loss": 6.4716,
      "step": 1062
    },
    {
      "epoch": 0.042914816310052485,
      "grad_norm": 111.28377532958984,
      "learning_rate": 1.9991301883878998e-05,
      "loss": 5.9851,
      "step": 1063
    },
    {
      "epoch": 0.04295518772708922,
      "grad_norm": 129.09036254882812,
      "learning_rate": 1.999124727272884e-05,
      "loss": 6.3508,
      "step": 1064
    },
    {
      "epoch": 0.04299555914412596,
      "grad_norm": 151.96238708496094,
      "learning_rate": 1.999119249075168e-05,
      "loss": 6.7268,
      "step": 1065
    },
    {
      "epoch": 0.0430359305611627,
      "grad_norm": 226.21957397460938,
      "learning_rate": 1.9991137537948463e-05,
      "loss": 5.8406,
      "step": 1066
    },
    {
      "epoch": 0.043076301978199436,
      "grad_norm": 130.66250610351562,
      "learning_rate": 1.9991082414320125e-05,
      "loss": 6.1135,
      "step": 1067
    },
    {
      "epoch": 0.04311667339523617,
      "grad_norm": 110.4659423828125,
      "learning_rate": 1.9991027119867606e-05,
      "loss": 5.52,
      "step": 1068
    },
    {
      "epoch": 0.043157044812272914,
      "grad_norm": 248.5395050048828,
      "learning_rate": 1.999097165459186e-05,
      "loss": 5.4924,
      "step": 1069
    },
    {
      "epoch": 0.04319741622930965,
      "grad_norm": 123.04351806640625,
      "learning_rate": 1.999091601849382e-05,
      "loss": 6.5307,
      "step": 1070
    },
    {
      "epoch": 0.043237787646346386,
      "grad_norm": 155.90274047851562,
      "learning_rate": 1.9990860211574454e-05,
      "loss": 5.6288,
      "step": 1071
    },
    {
      "epoch": 0.04327815906338312,
      "grad_norm": 134.09983825683594,
      "learning_rate": 1.9990804233834707e-05,
      "loss": 5.8077,
      "step": 1072
    },
    {
      "epoch": 0.043318530480419865,
      "grad_norm": 105.96923065185547,
      "learning_rate": 1.9990748085275536e-05,
      "loss": 5.5008,
      "step": 1073
    },
    {
      "epoch": 0.0433589018974566,
      "grad_norm": 108.15873718261719,
      "learning_rate": 1.9990691765897904e-05,
      "loss": 7.3744,
      "step": 1074
    },
    {
      "epoch": 0.043399273314493336,
      "grad_norm": 195.2692108154297,
      "learning_rate": 1.9990635275702774e-05,
      "loss": 5.7087,
      "step": 1075
    },
    {
      "epoch": 0.04343964473153008,
      "grad_norm": 96.03929901123047,
      "learning_rate": 1.999057861469111e-05,
      "loss": 5.8944,
      "step": 1076
    },
    {
      "epoch": 0.043480016148566815,
      "grad_norm": 143.28231811523438,
      "learning_rate": 1.9990521782863878e-05,
      "loss": 6.0511,
      "step": 1077
    },
    {
      "epoch": 0.04352038756560355,
      "grad_norm": 151.7787628173828,
      "learning_rate": 1.9990464780222057e-05,
      "loss": 5.6254,
      "step": 1078
    },
    {
      "epoch": 0.043560758982640294,
      "grad_norm": 179.39833068847656,
      "learning_rate": 1.9990407606766617e-05,
      "loss": 7.0378,
      "step": 1079
    },
    {
      "epoch": 0.04360113039967703,
      "grad_norm": 105.22634887695312,
      "learning_rate": 1.9990350262498537e-05,
      "loss": 4.819,
      "step": 1080
    },
    {
      "epoch": 0.043641501816713765,
      "grad_norm": 90.2320556640625,
      "learning_rate": 1.9990292747418795e-05,
      "loss": 7.245,
      "step": 1081
    },
    {
      "epoch": 0.04368187323375051,
      "grad_norm": 188.13011169433594,
      "learning_rate": 1.9990235061528375e-05,
      "loss": 5.8594,
      "step": 1082
    },
    {
      "epoch": 0.043722244650787244,
      "grad_norm": 155.65562438964844,
      "learning_rate": 1.9990177204828268e-05,
      "loss": 6.7277,
      "step": 1083
    },
    {
      "epoch": 0.04376261606782398,
      "grad_norm": 205.455322265625,
      "learning_rate": 1.9990119177319455e-05,
      "loss": 7.6742,
      "step": 1084
    },
    {
      "epoch": 0.043802987484860716,
      "grad_norm": 2903.569580078125,
      "learning_rate": 1.9990060979002934e-05,
      "loss": 7.1365,
      "step": 1085
    },
    {
      "epoch": 0.04384335890189746,
      "grad_norm": 124.10847473144531,
      "learning_rate": 1.9990002609879703e-05,
      "loss": 5.2152,
      "step": 1086
    },
    {
      "epoch": 0.043883730318934194,
      "grad_norm": 112.20275115966797,
      "learning_rate": 1.998994406995075e-05,
      "loss": 5.4728,
      "step": 1087
    },
    {
      "epoch": 0.04392410173597093,
      "grad_norm": 96.0655517578125,
      "learning_rate": 1.9989885359217083e-05,
      "loss": 4.8087,
      "step": 1088
    },
    {
      "epoch": 0.04396447315300767,
      "grad_norm": 149.8226318359375,
      "learning_rate": 1.9989826477679705e-05,
      "loss": 6.432,
      "step": 1089
    },
    {
      "epoch": 0.04400484457004441,
      "grad_norm": 171.94805908203125,
      "learning_rate": 1.9989767425339623e-05,
      "loss": 6.8509,
      "step": 1090
    },
    {
      "epoch": 0.044045215987081145,
      "grad_norm": 157.30709838867188,
      "learning_rate": 1.9989708202197845e-05,
      "loss": 6.622,
      "step": 1091
    },
    {
      "epoch": 0.04408558740411789,
      "grad_norm": 241.21519470214844,
      "learning_rate": 1.9989648808255383e-05,
      "loss": 5.7146,
      "step": 1092
    },
    {
      "epoch": 0.04412595882115462,
      "grad_norm": 102.68975830078125,
      "learning_rate": 1.9989589243513257e-05,
      "loss": 5.207,
      "step": 1093
    },
    {
      "epoch": 0.04416633023819136,
      "grad_norm": 101.56269073486328,
      "learning_rate": 1.9989529507972484e-05,
      "loss": 5.7021,
      "step": 1094
    },
    {
      "epoch": 0.0442067016552281,
      "grad_norm": 134.66754150390625,
      "learning_rate": 1.9989469601634076e-05,
      "loss": 5.555,
      "step": 1095
    },
    {
      "epoch": 0.04424707307226484,
      "grad_norm": 161.25537109375,
      "learning_rate": 1.9989409524499074e-05,
      "loss": 6.4348,
      "step": 1096
    },
    {
      "epoch": 0.044287444489301574,
      "grad_norm": 156.83226013183594,
      "learning_rate": 1.9989349276568488e-05,
      "loss": 5.0079,
      "step": 1097
    },
    {
      "epoch": 0.04432781590633831,
      "grad_norm": 108.50226593017578,
      "learning_rate": 1.998928885784336e-05,
      "loss": 6.5554,
      "step": 1098
    },
    {
      "epoch": 0.04436818732337505,
      "grad_norm": 158.862060546875,
      "learning_rate": 1.9989228268324718e-05,
      "loss": 5.7881,
      "step": 1099
    },
    {
      "epoch": 0.04440855874041179,
      "grad_norm": 132.00265502929688,
      "learning_rate": 1.9989167508013602e-05,
      "loss": 6.2163,
      "step": 1100
    },
    {
      "epoch": 0.044448930157448524,
      "grad_norm": 91.14561462402344,
      "learning_rate": 1.9989106576911045e-05,
      "loss": 6.7107,
      "step": 1101
    },
    {
      "epoch": 0.04448930157448527,
      "grad_norm": 181.92198181152344,
      "learning_rate": 1.9989045475018094e-05,
      "loss": 7.1868,
      "step": 1102
    },
    {
      "epoch": 0.044529672991522,
      "grad_norm": 108.36198425292969,
      "learning_rate": 1.998898420233579e-05,
      "loss": 6.6693,
      "step": 1103
    },
    {
      "epoch": 0.04457004440855874,
      "grad_norm": 144.49029541015625,
      "learning_rate": 1.998892275886518e-05,
      "loss": 5.6773,
      "step": 1104
    },
    {
      "epoch": 0.04461041582559548,
      "grad_norm": 179.79751586914062,
      "learning_rate": 1.998886114460732e-05,
      "loss": 5.5042,
      "step": 1105
    },
    {
      "epoch": 0.04465078724263222,
      "grad_norm": 158.31509399414062,
      "learning_rate": 1.998879935956326e-05,
      "loss": 5.7673,
      "step": 1106
    },
    {
      "epoch": 0.04469115865966895,
      "grad_norm": 108.93700408935547,
      "learning_rate": 1.9988737403734055e-05,
      "loss": 6.1067,
      "step": 1107
    },
    {
      "epoch": 0.04473153007670569,
      "grad_norm": 126.14005279541016,
      "learning_rate": 1.9988675277120766e-05,
      "loss": 3.825,
      "step": 1108
    },
    {
      "epoch": 0.04477190149374243,
      "grad_norm": 263.8062744140625,
      "learning_rate": 1.9988612979724456e-05,
      "loss": 6.5793,
      "step": 1109
    },
    {
      "epoch": 0.04481227291077917,
      "grad_norm": 154.64369201660156,
      "learning_rate": 1.998855051154619e-05,
      "loss": 6.9673,
      "step": 1110
    },
    {
      "epoch": 0.0448526443278159,
      "grad_norm": 479.53143310546875,
      "learning_rate": 1.998848787258703e-05,
      "loss": 5.5499,
      "step": 1111
    },
    {
      "epoch": 0.044893015744852646,
      "grad_norm": 133.3762664794922,
      "learning_rate": 1.9988425062848057e-05,
      "loss": 6.0206,
      "step": 1112
    },
    {
      "epoch": 0.04493338716188938,
      "grad_norm": 105.933837890625,
      "learning_rate": 1.9988362082330338e-05,
      "loss": 6.4531,
      "step": 1113
    },
    {
      "epoch": 0.04497375857892612,
      "grad_norm": 118.12909698486328,
      "learning_rate": 1.9988298931034954e-05,
      "loss": 6.0791,
      "step": 1114
    },
    {
      "epoch": 0.04501412999596286,
      "grad_norm": 103.60955047607422,
      "learning_rate": 1.9988235608962982e-05,
      "loss": 6.1467,
      "step": 1115
    },
    {
      "epoch": 0.0450545014129996,
      "grad_norm": 130.87034606933594,
      "learning_rate": 1.9988172116115504e-05,
      "loss": 5.7403,
      "step": 1116
    },
    {
      "epoch": 0.04509487283003633,
      "grad_norm": 116.55811309814453,
      "learning_rate": 1.9988108452493607e-05,
      "loss": 6.0257,
      "step": 1117
    },
    {
      "epoch": 0.045135244247073075,
      "grad_norm": 155.8071746826172,
      "learning_rate": 1.9988044618098383e-05,
      "loss": 4.4004,
      "step": 1118
    },
    {
      "epoch": 0.04517561566410981,
      "grad_norm": 112.07211303710938,
      "learning_rate": 1.9987980612930914e-05,
      "loss": 5.6258,
      "step": 1119
    },
    {
      "epoch": 0.04521598708114655,
      "grad_norm": 101.42943572998047,
      "learning_rate": 1.9987916436992303e-05,
      "loss": 4.4282,
      "step": 1120
    },
    {
      "epoch": 0.04525635849818328,
      "grad_norm": 96.31477355957031,
      "learning_rate": 1.9987852090283644e-05,
      "loss": 5.7906,
      "step": 1121
    },
    {
      "epoch": 0.045296729915220026,
      "grad_norm": 130.06105041503906,
      "learning_rate": 1.998778757280604e-05,
      "loss": 4.8594,
      "step": 1122
    },
    {
      "epoch": 0.04533710133225676,
      "grad_norm": 193.94432067871094,
      "learning_rate": 1.9987722884560588e-05,
      "loss": 5.3827,
      "step": 1123
    },
    {
      "epoch": 0.0453774727492935,
      "grad_norm": 98.28860473632812,
      "learning_rate": 1.9987658025548398e-05,
      "loss": 6.0421,
      "step": 1124
    },
    {
      "epoch": 0.04541784416633024,
      "grad_norm": 268.8415222167969,
      "learning_rate": 1.998759299577058e-05,
      "loss": 6.2585,
      "step": 1125
    },
    {
      "epoch": 0.045458215583366976,
      "grad_norm": 143.6586456298828,
      "learning_rate": 1.9987527795228245e-05,
      "loss": 4.8053,
      "step": 1126
    },
    {
      "epoch": 0.04549858700040371,
      "grad_norm": 146.8716583251953,
      "learning_rate": 1.9987462423922505e-05,
      "loss": 6.7209,
      "step": 1127
    },
    {
      "epoch": 0.045538958417440455,
      "grad_norm": 107.2955322265625,
      "learning_rate": 1.9987396881854485e-05,
      "loss": 5.7859,
      "step": 1128
    },
    {
      "epoch": 0.04557932983447719,
      "grad_norm": 141.61997985839844,
      "learning_rate": 1.9987331169025295e-05,
      "loss": 4.5948,
      "step": 1129
    },
    {
      "epoch": 0.045619701251513926,
      "grad_norm": 204.98492431640625,
      "learning_rate": 1.9987265285436065e-05,
      "loss": 6.7202,
      "step": 1130
    },
    {
      "epoch": 0.04566007266855067,
      "grad_norm": 158.23452758789062,
      "learning_rate": 1.9987199231087923e-05,
      "loss": 6.9996,
      "step": 1131
    },
    {
      "epoch": 0.045700444085587405,
      "grad_norm": 164.91954040527344,
      "learning_rate": 1.9987133005981995e-05,
      "loss": 5.6641,
      "step": 1132
    },
    {
      "epoch": 0.04574081550262414,
      "grad_norm": 153.7523193359375,
      "learning_rate": 1.998706661011941e-05,
      "loss": 5.621,
      "step": 1133
    },
    {
      "epoch": 0.04578118691966088,
      "grad_norm": 204.88934326171875,
      "learning_rate": 1.9987000043501315e-05,
      "loss": 5.4472,
      "step": 1134
    },
    {
      "epoch": 0.04582155833669762,
      "grad_norm": 206.99195861816406,
      "learning_rate": 1.9986933306128837e-05,
      "loss": 5.1346,
      "step": 1135
    },
    {
      "epoch": 0.045861929753734355,
      "grad_norm": 127.14704895019531,
      "learning_rate": 1.998686639800312e-05,
      "loss": 6.4709,
      "step": 1136
    },
    {
      "epoch": 0.04590230117077109,
      "grad_norm": 140.1676483154297,
      "learning_rate": 1.9986799319125308e-05,
      "loss": 6.3518,
      "step": 1137
    },
    {
      "epoch": 0.045942672587807834,
      "grad_norm": 204.47178649902344,
      "learning_rate": 1.998673206949655e-05,
      "loss": 6.0609,
      "step": 1138
    },
    {
      "epoch": 0.04598304400484457,
      "grad_norm": 175.15330505371094,
      "learning_rate": 1.9986664649117995e-05,
      "loss": 5.0675,
      "step": 1139
    },
    {
      "epoch": 0.046023415421881306,
      "grad_norm": 126.6028823852539,
      "learning_rate": 1.9986597057990794e-05,
      "loss": 5.6507,
      "step": 1140
    },
    {
      "epoch": 0.04606378683891805,
      "grad_norm": 97.00761413574219,
      "learning_rate": 1.9986529296116103e-05,
      "loss": 6.4741,
      "step": 1141
    },
    {
      "epoch": 0.046104158255954784,
      "grad_norm": 131.4276123046875,
      "learning_rate": 1.998646136349508e-05,
      "loss": 6.4419,
      "step": 1142
    },
    {
      "epoch": 0.04614452967299152,
      "grad_norm": 156.15280151367188,
      "learning_rate": 1.998639326012889e-05,
      "loss": 6.7203,
      "step": 1143
    },
    {
      "epoch": 0.04618490109002826,
      "grad_norm": 113.69229125976562,
      "learning_rate": 1.99863249860187e-05,
      "loss": 5.0483,
      "step": 1144
    },
    {
      "epoch": 0.046225272507065,
      "grad_norm": 142.95556640625,
      "learning_rate": 1.9986256541165665e-05,
      "loss": 5.8937,
      "step": 1145
    },
    {
      "epoch": 0.046265643924101735,
      "grad_norm": 744.6958618164062,
      "learning_rate": 1.9986187925570967e-05,
      "loss": 6.1853,
      "step": 1146
    },
    {
      "epoch": 0.04630601534113847,
      "grad_norm": 169.28131103515625,
      "learning_rate": 1.9986119139235767e-05,
      "loss": 5.5197,
      "step": 1147
    },
    {
      "epoch": 0.04634638675817521,
      "grad_norm": 164.5127716064453,
      "learning_rate": 1.9986050182161257e-05,
      "loss": 6.3689,
      "step": 1148
    },
    {
      "epoch": 0.04638675817521195,
      "grad_norm": 102.30970001220703,
      "learning_rate": 1.9985981054348608e-05,
      "loss": 5.3177,
      "step": 1149
    },
    {
      "epoch": 0.046427129592248685,
      "grad_norm": 499.9005126953125,
      "learning_rate": 1.9985911755799e-05,
      "loss": 5.8526,
      "step": 1150
    },
    {
      "epoch": 0.04646750100928543,
      "grad_norm": 121.54989624023438,
      "learning_rate": 1.9985842286513617e-05,
      "loss": 5.7623,
      "step": 1151
    },
    {
      "epoch": 0.046507872426322164,
      "grad_norm": 76.56183624267578,
      "learning_rate": 1.9985772646493653e-05,
      "loss": 4.4649,
      "step": 1152
    },
    {
      "epoch": 0.0465482438433589,
      "grad_norm": 86.53046417236328,
      "learning_rate": 1.9985702835740293e-05,
      "loss": 4.7338,
      "step": 1153
    },
    {
      "epoch": 0.04658861526039564,
      "grad_norm": 121.79015350341797,
      "learning_rate": 1.9985632854254735e-05,
      "loss": 5.9949,
      "step": 1154
    },
    {
      "epoch": 0.04662898667743238,
      "grad_norm": 148.50816345214844,
      "learning_rate": 1.9985562702038176e-05,
      "loss": 6.8814,
      "step": 1155
    },
    {
      "epoch": 0.046669358094469114,
      "grad_norm": 152.8815460205078,
      "learning_rate": 1.9985492379091806e-05,
      "loss": 4.3406,
      "step": 1156
    },
    {
      "epoch": 0.04670972951150586,
      "grad_norm": 129.44300842285156,
      "learning_rate": 1.998542188541684e-05,
      "loss": 5.5322,
      "step": 1157
    },
    {
      "epoch": 0.04675010092854259,
      "grad_norm": 82.39713287353516,
      "learning_rate": 1.9985351221014473e-05,
      "loss": 5.5798,
      "step": 1158
    },
    {
      "epoch": 0.04679047234557933,
      "grad_norm": 114.67699432373047,
      "learning_rate": 1.998528038588592e-05,
      "loss": 5.1974,
      "step": 1159
    },
    {
      "epoch": 0.046830843762616065,
      "grad_norm": 132.25848388671875,
      "learning_rate": 1.9985209380032388e-05,
      "loss": 4.6063,
      "step": 1160
    },
    {
      "epoch": 0.04687121517965281,
      "grad_norm": 395.37213134765625,
      "learning_rate": 1.9985138203455095e-05,
      "loss": 5.0153,
      "step": 1161
    },
    {
      "epoch": 0.04691158659668954,
      "grad_norm": 203.62266540527344,
      "learning_rate": 1.998506685615525e-05,
      "loss": 5.9101,
      "step": 1162
    },
    {
      "epoch": 0.04695195801372628,
      "grad_norm": 139.44264221191406,
      "learning_rate": 1.9984995338134084e-05,
      "loss": 4.5079,
      "step": 1163
    },
    {
      "epoch": 0.04699232943076302,
      "grad_norm": 185.20130920410156,
      "learning_rate": 1.9984923649392817e-05,
      "loss": 6.4481,
      "step": 1164
    },
    {
      "epoch": 0.04703270084779976,
      "grad_norm": 133.33383178710938,
      "learning_rate": 1.9984851789932666e-05,
      "loss": 4.3917,
      "step": 1165
    },
    {
      "epoch": 0.047073072264836494,
      "grad_norm": 126.76311492919922,
      "learning_rate": 1.998477975975487e-05,
      "loss": 7.6036,
      "step": 1166
    },
    {
      "epoch": 0.047113443681873236,
      "grad_norm": 143.07958984375,
      "learning_rate": 1.998470755886065e-05,
      "loss": 6.2858,
      "step": 1167
    },
    {
      "epoch": 0.04715381509890997,
      "grad_norm": 95.10242462158203,
      "learning_rate": 1.9984635187251254e-05,
      "loss": 6.2297,
      "step": 1168
    },
    {
      "epoch": 0.04719418651594671,
      "grad_norm": 124.74854278564453,
      "learning_rate": 1.998456264492791e-05,
      "loss": 3.1416,
      "step": 1169
    },
    {
      "epoch": 0.04723455793298345,
      "grad_norm": 199.1131134033203,
      "learning_rate": 1.998448993189186e-05,
      "loss": 6.4752,
      "step": 1170
    },
    {
      "epoch": 0.04727492935002019,
      "grad_norm": 118.0827865600586,
      "learning_rate": 1.9984417048144348e-05,
      "loss": 7.6261,
      "step": 1171
    },
    {
      "epoch": 0.04731530076705692,
      "grad_norm": 126.98328399658203,
      "learning_rate": 1.9984343993686618e-05,
      "loss": 5.7251,
      "step": 1172
    },
    {
      "epoch": 0.04735567218409366,
      "grad_norm": 183.26153564453125,
      "learning_rate": 1.998427076851992e-05,
      "loss": 6.8422,
      "step": 1173
    },
    {
      "epoch": 0.0473960436011304,
      "grad_norm": 108.1826400756836,
      "learning_rate": 1.998419737264551e-05,
      "loss": 7.0748,
      "step": 1174
    },
    {
      "epoch": 0.04743641501816714,
      "grad_norm": 103.74041748046875,
      "learning_rate": 1.998412380606464e-05,
      "loss": 5.3999,
      "step": 1175
    },
    {
      "epoch": 0.04747678643520387,
      "grad_norm": 110.30939483642578,
      "learning_rate": 1.9984050068778564e-05,
      "loss": 7.1862,
      "step": 1176
    },
    {
      "epoch": 0.047517157852240616,
      "grad_norm": 103.36114501953125,
      "learning_rate": 1.998397616078855e-05,
      "loss": 5.6443,
      "step": 1177
    },
    {
      "epoch": 0.04755752926927735,
      "grad_norm": 451.5368957519531,
      "learning_rate": 1.9983902082095855e-05,
      "loss": 4.42,
      "step": 1178
    },
    {
      "epoch": 0.04759790068631409,
      "grad_norm": 117.05953979492188,
      "learning_rate": 1.998382783270175e-05,
      "loss": 6.5285,
      "step": 1179
    },
    {
      "epoch": 0.04763827210335083,
      "grad_norm": 130.06918334960938,
      "learning_rate": 1.9983753412607504e-05,
      "loss": 5.0284,
      "step": 1180
    },
    {
      "epoch": 0.047678643520387566,
      "grad_norm": 108.07593536376953,
      "learning_rate": 1.9983678821814384e-05,
      "loss": 4.9273,
      "step": 1181
    },
    {
      "epoch": 0.0477190149374243,
      "grad_norm": 141.99127197265625,
      "learning_rate": 1.9983604060323674e-05,
      "loss": 4.8292,
      "step": 1182
    },
    {
      "epoch": 0.047759386354461045,
      "grad_norm": 110.19050598144531,
      "learning_rate": 1.9983529128136646e-05,
      "loss": 5.6606,
      "step": 1183
    },
    {
      "epoch": 0.04779975777149778,
      "grad_norm": 97.05155944824219,
      "learning_rate": 1.9983454025254582e-05,
      "loss": 6.6145,
      "step": 1184
    },
    {
      "epoch": 0.047840129188534516,
      "grad_norm": 118.29586791992188,
      "learning_rate": 1.9983378751678768e-05,
      "loss": 5.726,
      "step": 1185
    },
    {
      "epoch": 0.04788050060557125,
      "grad_norm": 112.26173400878906,
      "learning_rate": 1.9983303307410493e-05,
      "loss": 6.9124,
      "step": 1186
    },
    {
      "epoch": 0.047920872022607995,
      "grad_norm": 153.64259338378906,
      "learning_rate": 1.9983227692451043e-05,
      "loss": 6.8047,
      "step": 1187
    },
    {
      "epoch": 0.04796124343964473,
      "grad_norm": 90.42520904541016,
      "learning_rate": 1.998315190680171e-05,
      "loss": 7.2068,
      "step": 1188
    },
    {
      "epoch": 0.04800161485668147,
      "grad_norm": 133.5157470703125,
      "learning_rate": 1.998307595046379e-05,
      "loss": 5.8454,
      "step": 1189
    },
    {
      "epoch": 0.04804198627371821,
      "grad_norm": 95.96729278564453,
      "learning_rate": 1.998299982343859e-05,
      "loss": 6.1759,
      "step": 1190
    },
    {
      "epoch": 0.048082357690754945,
      "grad_norm": 120.46910095214844,
      "learning_rate": 1.99829235257274e-05,
      "loss": 6.9672,
      "step": 1191
    },
    {
      "epoch": 0.04812272910779168,
      "grad_norm": 136.16871643066406,
      "learning_rate": 1.998284705733153e-05,
      "loss": 6.7573,
      "step": 1192
    },
    {
      "epoch": 0.048163100524828424,
      "grad_norm": 229.87266540527344,
      "learning_rate": 1.9982770418252292e-05,
      "loss": 6.9194,
      "step": 1193
    },
    {
      "epoch": 0.04820347194186516,
      "grad_norm": 311.719970703125,
      "learning_rate": 1.9982693608490985e-05,
      "loss": 4.3009,
      "step": 1194
    },
    {
      "epoch": 0.048243843358901896,
      "grad_norm": 245.3404083251953,
      "learning_rate": 1.998261662804893e-05,
      "loss": 7.0043,
      "step": 1195
    },
    {
      "epoch": 0.04828421477593864,
      "grad_norm": 149.14810180664062,
      "learning_rate": 1.998253947692744e-05,
      "loss": 3.1253,
      "step": 1196
    },
    {
      "epoch": 0.048324586192975374,
      "grad_norm": 265.0834045410156,
      "learning_rate": 1.9982462155127843e-05,
      "loss": 6.269,
      "step": 1197
    },
    {
      "epoch": 0.04836495761001211,
      "grad_norm": 208.91168212890625,
      "learning_rate": 1.9982384662651445e-05,
      "loss": 6.7948,
      "step": 1198
    },
    {
      "epoch": 0.048405329027048846,
      "grad_norm": 118.25501251220703,
      "learning_rate": 1.9982306999499586e-05,
      "loss": 5.3985,
      "step": 1199
    },
    {
      "epoch": 0.04844570044408559,
      "grad_norm": 119.06272888183594,
      "learning_rate": 1.9982229165673582e-05,
      "loss": 5.714,
      "step": 1200
    },
    {
      "epoch": 0.048486071861122325,
      "grad_norm": 165.54046630859375,
      "learning_rate": 1.9982151161174777e-05,
      "loss": 5.674,
      "step": 1201
    },
    {
      "epoch": 0.04852644327815906,
      "grad_norm": 142.63671875,
      "learning_rate": 1.9982072986004494e-05,
      "loss": 5.7368,
      "step": 1202
    },
    {
      "epoch": 0.048566814695195804,
      "grad_norm": 99.59209442138672,
      "learning_rate": 1.998199464016407e-05,
      "loss": 4.8528,
      "step": 1203
    },
    {
      "epoch": 0.04860718611223254,
      "grad_norm": 95.27738189697266,
      "learning_rate": 1.998191612365485e-05,
      "loss": 6.0066,
      "step": 1204
    },
    {
      "epoch": 0.048647557529269275,
      "grad_norm": 104.5033950805664,
      "learning_rate": 1.9981837436478172e-05,
      "loss": 5.7611,
      "step": 1205
    },
    {
      "epoch": 0.04868792894630602,
      "grad_norm": 95.80033874511719,
      "learning_rate": 1.9981758578635383e-05,
      "loss": 5.5863,
      "step": 1206
    },
    {
      "epoch": 0.048728300363342754,
      "grad_norm": 155.98773193359375,
      "learning_rate": 1.9981679550127833e-05,
      "loss": 6.5733,
      "step": 1207
    },
    {
      "epoch": 0.04876867178037949,
      "grad_norm": 187.00230407714844,
      "learning_rate": 1.9981600350956872e-05,
      "loss": 5.6639,
      "step": 1208
    },
    {
      "epoch": 0.04880904319741623,
      "grad_norm": 124.65972137451172,
      "learning_rate": 1.9981520981123853e-05,
      "loss": 6.1631,
      "step": 1209
    },
    {
      "epoch": 0.04884941461445297,
      "grad_norm": 85.53620147705078,
      "learning_rate": 1.998144144063013e-05,
      "loss": 7.0589,
      "step": 1210
    },
    {
      "epoch": 0.048889786031489704,
      "grad_norm": 132.5107421875,
      "learning_rate": 1.9981361729477075e-05,
      "loss": 5.7813,
      "step": 1211
    },
    {
      "epoch": 0.04893015744852644,
      "grad_norm": 213.98443603515625,
      "learning_rate": 1.9981281847666036e-05,
      "loss": 6.7952,
      "step": 1212
    },
    {
      "epoch": 0.04897052886556318,
      "grad_norm": 121.86805725097656,
      "learning_rate": 1.9981201795198387e-05,
      "loss": 7.4908,
      "step": 1213
    },
    {
      "epoch": 0.04901090028259992,
      "grad_norm": 141.57179260253906,
      "learning_rate": 1.9981121572075497e-05,
      "loss": 6.7372,
      "step": 1214
    },
    {
      "epoch": 0.049051271699636655,
      "grad_norm": 149.16043090820312,
      "learning_rate": 1.9981041178298734e-05,
      "loss": 5.4855,
      "step": 1215
    },
    {
      "epoch": 0.0490916431166734,
      "grad_norm": 97.88445281982422,
      "learning_rate": 1.998096061386947e-05,
      "loss": 5.7917,
      "step": 1216
    },
    {
      "epoch": 0.04913201453371013,
      "grad_norm": 144.01998901367188,
      "learning_rate": 1.9980879878789093e-05,
      "loss": 5.1578,
      "step": 1217
    },
    {
      "epoch": 0.04917238595074687,
      "grad_norm": 125.4499740600586,
      "learning_rate": 1.9980798973058976e-05,
      "loss": 6.191,
      "step": 1218
    },
    {
      "epoch": 0.04921275736778361,
      "grad_norm": 90.17740631103516,
      "learning_rate": 1.9980717896680503e-05,
      "loss": 6.5544,
      "step": 1219
    },
    {
      "epoch": 0.04925312878482035,
      "grad_norm": 103.08251190185547,
      "learning_rate": 1.9980636649655058e-05,
      "loss": 6.4039,
      "step": 1220
    },
    {
      "epoch": 0.049293500201857084,
      "grad_norm": 104.00209045410156,
      "learning_rate": 1.9980555231984035e-05,
      "loss": 5.4482,
      "step": 1221
    },
    {
      "epoch": 0.049333871618893826,
      "grad_norm": 86.6725082397461,
      "learning_rate": 1.9980473643668823e-05,
      "loss": 5.1435,
      "step": 1222
    },
    {
      "epoch": 0.04937424303593056,
      "grad_norm": 96.4255599975586,
      "learning_rate": 1.998039188471082e-05,
      "loss": 4.8318,
      "step": 1223
    },
    {
      "epoch": 0.0494146144529673,
      "grad_norm": 83.86189270019531,
      "learning_rate": 1.998030995511142e-05,
      "loss": 5.998,
      "step": 1224
    },
    {
      "epoch": 0.049454985870004034,
      "grad_norm": 122.8338394165039,
      "learning_rate": 1.9980227854872027e-05,
      "loss": 5.4479,
      "step": 1225
    },
    {
      "epoch": 0.04949535728704078,
      "grad_norm": 101.50725555419922,
      "learning_rate": 1.998014558399404e-05,
      "loss": 5.2056,
      "step": 1226
    },
    {
      "epoch": 0.04953572870407751,
      "grad_norm": 90.2826919555664,
      "learning_rate": 1.998006314247887e-05,
      "loss": 6.4181,
      "step": 1227
    },
    {
      "epoch": 0.04957610012111425,
      "grad_norm": 127.66415405273438,
      "learning_rate": 1.9979980530327925e-05,
      "loss": 6.7047,
      "step": 1228
    },
    {
      "epoch": 0.04961647153815099,
      "grad_norm": 116.70962524414062,
      "learning_rate": 1.997989774754262e-05,
      "loss": 5.4378,
      "step": 1229
    },
    {
      "epoch": 0.04965684295518773,
      "grad_norm": 111.18699645996094,
      "learning_rate": 1.997981479412437e-05,
      "loss": 6.5047,
      "step": 1230
    },
    {
      "epoch": 0.04969721437222446,
      "grad_norm": 140.8634033203125,
      "learning_rate": 1.9979731670074587e-05,
      "loss": 5.7454,
      "step": 1231
    },
    {
      "epoch": 0.049737585789261206,
      "grad_norm": 113.32780456542969,
      "learning_rate": 1.9979648375394694e-05,
      "loss": 6.6421,
      "step": 1232
    },
    {
      "epoch": 0.04977795720629794,
      "grad_norm": 137.40892028808594,
      "learning_rate": 1.9979564910086123e-05,
      "loss": 6.0854,
      "step": 1233
    },
    {
      "epoch": 0.04981832862333468,
      "grad_norm": 126.26129150390625,
      "learning_rate": 1.99794812741503e-05,
      "loss": 5.9547,
      "step": 1234
    },
    {
      "epoch": 0.04985870004037142,
      "grad_norm": 113.43743896484375,
      "learning_rate": 1.9979397467588644e-05,
      "loss": 6.1343,
      "step": 1235
    },
    {
      "epoch": 0.049899071457408156,
      "grad_norm": 153.7223358154297,
      "learning_rate": 1.99793134904026e-05,
      "loss": 7.7202,
      "step": 1236
    },
    {
      "epoch": 0.04993944287444489,
      "grad_norm": 113.13749694824219,
      "learning_rate": 1.9979229342593595e-05,
      "loss": 5.5937,
      "step": 1237
    },
    {
      "epoch": 0.04997981429148163,
      "grad_norm": 243.01681518554688,
      "learning_rate": 1.997914502416307e-05,
      "loss": 6.2663,
      "step": 1238
    },
    {
      "epoch": 0.05002018570851837,
      "grad_norm": 106.47587585449219,
      "learning_rate": 1.9979060535112472e-05,
      "loss": 4.2929,
      "step": 1239
    },
    {
      "epoch": 0.05006055712555511,
      "grad_norm": 171.2425079345703,
      "learning_rate": 1.997897587544324e-05,
      "loss": 5.916,
      "step": 1240
    },
    {
      "epoch": 0.05010092854259184,
      "grad_norm": 148.08729553222656,
      "learning_rate": 1.9978891045156822e-05,
      "loss": 6.0173,
      "step": 1241
    },
    {
      "epoch": 0.050141299959628585,
      "grad_norm": 358.9314270019531,
      "learning_rate": 1.9978806044254672e-05,
      "loss": 4.385,
      "step": 1242
    },
    {
      "epoch": 0.05018167137666532,
      "grad_norm": 158.27859497070312,
      "learning_rate": 1.9978720872738236e-05,
      "loss": 5.5109,
      "step": 1243
    },
    {
      "epoch": 0.05022204279370206,
      "grad_norm": 102.37310791015625,
      "learning_rate": 1.997863553060898e-05,
      "loss": 5.7918,
      "step": 1244
    },
    {
      "epoch": 0.0502624142107388,
      "grad_norm": 123.58358001708984,
      "learning_rate": 1.9978550017868355e-05,
      "loss": 5.7804,
      "step": 1245
    },
    {
      "epoch": 0.050302785627775536,
      "grad_norm": 288.0574035644531,
      "learning_rate": 1.997846433451783e-05,
      "loss": 5.6598,
      "step": 1246
    },
    {
      "epoch": 0.05034315704481227,
      "grad_norm": 121.70074462890625,
      "learning_rate": 1.9978378480558865e-05,
      "loss": 5.6359,
      "step": 1247
    },
    {
      "epoch": 0.050383528461849014,
      "grad_norm": 122.85714721679688,
      "learning_rate": 1.997829245599293e-05,
      "loss": 5.2886,
      "step": 1248
    },
    {
      "epoch": 0.05042389987888575,
      "grad_norm": 149.9478759765625,
      "learning_rate": 1.997820626082149e-05,
      "loss": 4.8303,
      "step": 1249
    },
    {
      "epoch": 0.050464271295922486,
      "grad_norm": 103.06248474121094,
      "learning_rate": 1.9978119895046027e-05,
      "loss": 7.3112,
      "step": 1250
    },
    {
      "epoch": 0.05050464271295922,
      "grad_norm": 243.74024963378906,
      "learning_rate": 1.9978033358668015e-05,
      "loss": 5.5019,
      "step": 1251
    },
    {
      "epoch": 0.050545014129995965,
      "grad_norm": 110.84378814697266,
      "learning_rate": 1.9977946651688935e-05,
      "loss": 5.6039,
      "step": 1252
    },
    {
      "epoch": 0.0505853855470327,
      "grad_norm": 139.84494018554688,
      "learning_rate": 1.9977859774110263e-05,
      "loss": 5.1439,
      "step": 1253
    },
    {
      "epoch": 0.050625756964069436,
      "grad_norm": 281.1864013671875,
      "learning_rate": 1.997777272593349e-05,
      "loss": 5.114,
      "step": 1254
    },
    {
      "epoch": 0.05066612838110618,
      "grad_norm": 115.54508209228516,
      "learning_rate": 1.9977685507160106e-05,
      "loss": 5.5498,
      "step": 1255
    },
    {
      "epoch": 0.050706499798142915,
      "grad_norm": 129.82728576660156,
      "learning_rate": 1.9977598117791594e-05,
      "loss": 5.8948,
      "step": 1256
    },
    {
      "epoch": 0.05074687121517965,
      "grad_norm": 124.21087646484375,
      "learning_rate": 1.9977510557829462e-05,
      "loss": 5.9127,
      "step": 1257
    },
    {
      "epoch": 0.050787242632216394,
      "grad_norm": 96.66398620605469,
      "learning_rate": 1.997742282727519e-05,
      "loss": 5.892,
      "step": 1258
    },
    {
      "epoch": 0.05082761404925313,
      "grad_norm": 353.22613525390625,
      "learning_rate": 1.997733492613029e-05,
      "loss": 5.2577,
      "step": 1259
    },
    {
      "epoch": 0.050867985466289865,
      "grad_norm": 158.43710327148438,
      "learning_rate": 1.997724685439626e-05,
      "loss": 6.7272,
      "step": 1260
    },
    {
      "epoch": 0.05090835688332661,
      "grad_norm": 124.4527587890625,
      "learning_rate": 1.997715861207461e-05,
      "loss": 6.3207,
      "step": 1261
    },
    {
      "epoch": 0.050948728300363344,
      "grad_norm": 120.90972137451172,
      "learning_rate": 1.9977070199166845e-05,
      "loss": 7.0641,
      "step": 1262
    },
    {
      "epoch": 0.05098909971740008,
      "grad_norm": 170.68565368652344,
      "learning_rate": 1.9976981615674476e-05,
      "loss": 5.7514,
      "step": 1263
    },
    {
      "epoch": 0.051029471134436816,
      "grad_norm": 75.56718444824219,
      "learning_rate": 1.9976892861599024e-05,
      "loss": 6.58,
      "step": 1264
    },
    {
      "epoch": 0.05106984255147356,
      "grad_norm": 104.8825454711914,
      "learning_rate": 1.9976803936942e-05,
      "loss": 5.4687,
      "step": 1265
    },
    {
      "epoch": 0.051110213968510294,
      "grad_norm": 131.8006591796875,
      "learning_rate": 1.9976714841704923e-05,
      "loss": 5.9328,
      "step": 1266
    },
    {
      "epoch": 0.05115058538554703,
      "grad_norm": 141.2397918701172,
      "learning_rate": 1.997662557588932e-05,
      "loss": 6.4069,
      "step": 1267
    },
    {
      "epoch": 0.05119095680258377,
      "grad_norm": 97.68647003173828,
      "learning_rate": 1.997653613949672e-05,
      "loss": 6.7659,
      "step": 1268
    },
    {
      "epoch": 0.05123132821962051,
      "grad_norm": 91.86589050292969,
      "learning_rate": 1.9976446532528648e-05,
      "loss": 6.9452,
      "step": 1269
    },
    {
      "epoch": 0.051271699636657245,
      "grad_norm": 92.57233428955078,
      "learning_rate": 1.9976356754986636e-05,
      "loss": 6.055,
      "step": 1270
    },
    {
      "epoch": 0.05131207105369399,
      "grad_norm": 102.03011322021484,
      "learning_rate": 1.997626680687222e-05,
      "loss": 6.2228,
      "step": 1271
    },
    {
      "epoch": 0.05135244247073072,
      "grad_norm": 107.82308959960938,
      "learning_rate": 1.997617668818694e-05,
      "loss": 5.2351,
      "step": 1272
    },
    {
      "epoch": 0.05139281388776746,
      "grad_norm": 80.29537963867188,
      "learning_rate": 1.997608639893233e-05,
      "loss": 5.9063,
      "step": 1273
    },
    {
      "epoch": 0.0514331853048042,
      "grad_norm": 173.21469116210938,
      "learning_rate": 1.997599593910994e-05,
      "loss": 7.349,
      "step": 1274
    },
    {
      "epoch": 0.05147355672184094,
      "grad_norm": 132.235595703125,
      "learning_rate": 1.997590530872132e-05,
      "loss": 6.5699,
      "step": 1275
    },
    {
      "epoch": 0.051513928138877674,
      "grad_norm": 98.20887756347656,
      "learning_rate": 1.9975814507768006e-05,
      "loss": 6.6151,
      "step": 1276
    },
    {
      "epoch": 0.05155429955591441,
      "grad_norm": 169.0183868408203,
      "learning_rate": 1.9975723536251567e-05,
      "loss": 5.6832,
      "step": 1277
    },
    {
      "epoch": 0.05159467097295115,
      "grad_norm": 129.72747802734375,
      "learning_rate": 1.9975632394173548e-05,
      "loss": 6.3958,
      "step": 1278
    },
    {
      "epoch": 0.05163504238998789,
      "grad_norm": 172.88772583007812,
      "learning_rate": 1.9975541081535508e-05,
      "loss": 6.6314,
      "step": 1279
    },
    {
      "epoch": 0.051675413807024624,
      "grad_norm": 111.8676986694336,
      "learning_rate": 1.997544959833901e-05,
      "loss": 4.8675,
      "step": 1280
    },
    {
      "epoch": 0.05171578522406137,
      "grad_norm": 158.0574188232422,
      "learning_rate": 1.9975357944585624e-05,
      "loss": 5.5786,
      "step": 1281
    },
    {
      "epoch": 0.0517561566410981,
      "grad_norm": 227.26177978515625,
      "learning_rate": 1.9975266120276905e-05,
      "loss": 5.521,
      "step": 1282
    },
    {
      "epoch": 0.05179652805813484,
      "grad_norm": 90.82671356201172,
      "learning_rate": 1.9975174125414433e-05,
      "loss": 6.3237,
      "step": 1283
    },
    {
      "epoch": 0.05183689947517158,
      "grad_norm": 133.310546875,
      "learning_rate": 1.9975081959999772e-05,
      "loss": 4.8175,
      "step": 1284
    },
    {
      "epoch": 0.05187727089220832,
      "grad_norm": 133.6127471923828,
      "learning_rate": 1.9974989624034506e-05,
      "loss": 4.785,
      "step": 1285
    },
    {
      "epoch": 0.05191764230924505,
      "grad_norm": 147.43511962890625,
      "learning_rate": 1.997489711752021e-05,
      "loss": 5.1489,
      "step": 1286
    },
    {
      "epoch": 0.051958013726281796,
      "grad_norm": 135.45767211914062,
      "learning_rate": 1.997480444045847e-05,
      "loss": 5.0573,
      "step": 1287
    },
    {
      "epoch": 0.05199838514331853,
      "grad_norm": 161.76026916503906,
      "learning_rate": 1.9974711592850862e-05,
      "loss": 6.9806,
      "step": 1288
    },
    {
      "epoch": 0.05203875656035527,
      "grad_norm": 518.6951904296875,
      "learning_rate": 1.9974618574698983e-05,
      "loss": 5.38,
      "step": 1289
    },
    {
      "epoch": 0.052079127977392004,
      "grad_norm": 105.23979949951172,
      "learning_rate": 1.9974525386004415e-05,
      "loss": 6.0424,
      "step": 1290
    },
    {
      "epoch": 0.052119499394428746,
      "grad_norm": 143.75657653808594,
      "learning_rate": 1.9974432026768756e-05,
      "loss": 4.7859,
      "step": 1291
    },
    {
      "epoch": 0.05215987081146548,
      "grad_norm": 114.5059814453125,
      "learning_rate": 1.99743384969936e-05,
      "loss": 6.5642,
      "step": 1292
    },
    {
      "epoch": 0.05220024222850222,
      "grad_norm": 150.5408935546875,
      "learning_rate": 1.997424479668055e-05,
      "loss": 5.9597,
      "step": 1293
    },
    {
      "epoch": 0.05224061364553896,
      "grad_norm": 116.34163665771484,
      "learning_rate": 1.9974150925831205e-05,
      "loss": 5.8919,
      "step": 1294
    },
    {
      "epoch": 0.0522809850625757,
      "grad_norm": 166.8064727783203,
      "learning_rate": 1.997405688444717e-05,
      "loss": 4.5232,
      "step": 1295
    },
    {
      "epoch": 0.05232135647961243,
      "grad_norm": 108.86919403076172,
      "learning_rate": 1.9973962672530054e-05,
      "loss": 5.6078,
      "step": 1296
    },
    {
      "epoch": 0.052361727896649175,
      "grad_norm": 104.56317901611328,
      "learning_rate": 1.9973868290081467e-05,
      "loss": 5.7883,
      "step": 1297
    },
    {
      "epoch": 0.05240209931368591,
      "grad_norm": 114.63306427001953,
      "learning_rate": 1.9973773737103024e-05,
      "loss": 6.5759,
      "step": 1298
    },
    {
      "epoch": 0.05244247073072265,
      "grad_norm": 281.47845458984375,
      "learning_rate": 1.997367901359634e-05,
      "loss": 4.5928,
      "step": 1299
    },
    {
      "epoch": 0.05248284214775938,
      "grad_norm": 132.67547607421875,
      "learning_rate": 1.997358411956303e-05,
      "loss": 8.01,
      "step": 1300
    },
    {
      "epoch": 0.052523213564796126,
      "grad_norm": 108.06166076660156,
      "learning_rate": 1.9973489055004728e-05,
      "loss": 5.8204,
      "step": 1301
    },
    {
      "epoch": 0.05256358498183286,
      "grad_norm": 144.85549926757812,
      "learning_rate": 1.9973393819923053e-05,
      "loss": 5.8445,
      "step": 1302
    },
    {
      "epoch": 0.0526039563988696,
      "grad_norm": 155.4110870361328,
      "learning_rate": 1.9973298414319632e-05,
      "loss": 4.966,
      "step": 1303
    },
    {
      "epoch": 0.05264432781590634,
      "grad_norm": 87.1773452758789,
      "learning_rate": 1.9973202838196096e-05,
      "loss": 6.0278,
      "step": 1304
    },
    {
      "epoch": 0.052684699232943076,
      "grad_norm": 401.89263916015625,
      "learning_rate": 1.997310709155408e-05,
      "loss": 6.3714,
      "step": 1305
    },
    {
      "epoch": 0.05272507064997981,
      "grad_norm": 100.06290435791016,
      "learning_rate": 1.997301117439522e-05,
      "loss": 4.8536,
      "step": 1306
    },
    {
      "epoch": 0.052765442067016555,
      "grad_norm": 327.6883239746094,
      "learning_rate": 1.9972915086721162e-05,
      "loss": 6.325,
      "step": 1307
    },
    {
      "epoch": 0.05280581348405329,
      "grad_norm": 114.58361053466797,
      "learning_rate": 1.9972818828533543e-05,
      "loss": 6.4542,
      "step": 1308
    },
    {
      "epoch": 0.052846184901090026,
      "grad_norm": 130.04104614257812,
      "learning_rate": 1.997272239983401e-05,
      "loss": 5.7951,
      "step": 1309
    },
    {
      "epoch": 0.05288655631812677,
      "grad_norm": 142.943115234375,
      "learning_rate": 1.997262580062421e-05,
      "loss": 5.5936,
      "step": 1310
    },
    {
      "epoch": 0.052926927735163505,
      "grad_norm": 142.7476043701172,
      "learning_rate": 1.9972529030905798e-05,
      "loss": 6.7457,
      "step": 1311
    },
    {
      "epoch": 0.05296729915220024,
      "grad_norm": 141.2349090576172,
      "learning_rate": 1.997243209068043e-05,
      "loss": 5.7022,
      "step": 1312
    },
    {
      "epoch": 0.05300767056923698,
      "grad_norm": 147.21990966796875,
      "learning_rate": 1.9972334979949755e-05,
      "loss": 6.8218,
      "step": 1313
    },
    {
      "epoch": 0.05304804198627372,
      "grad_norm": 117.95244598388672,
      "learning_rate": 1.9972237698715445e-05,
      "loss": 4.1749,
      "step": 1314
    },
    {
      "epoch": 0.053088413403310455,
      "grad_norm": 134.78895568847656,
      "learning_rate": 1.9972140246979152e-05,
      "loss": 6.3589,
      "step": 1315
    },
    {
      "epoch": 0.05312878482034719,
      "grad_norm": 237.93527221679688,
      "learning_rate": 1.997204262474255e-05,
      "loss": 4.9685,
      "step": 1316
    },
    {
      "epoch": 0.053169156237383934,
      "grad_norm": 140.41966247558594,
      "learning_rate": 1.9971944832007306e-05,
      "loss": 5.2469,
      "step": 1317
    },
    {
      "epoch": 0.05320952765442067,
      "grad_norm": 110.95645141601562,
      "learning_rate": 1.997184686877509e-05,
      "loss": 7.3303,
      "step": 1318
    },
    {
      "epoch": 0.053249899071457406,
      "grad_norm": 116.3614501953125,
      "learning_rate": 1.9971748735047577e-05,
      "loss": 5.7821,
      "step": 1319
    },
    {
      "epoch": 0.05329027048849415,
      "grad_norm": 122.58361053466797,
      "learning_rate": 1.997165043082645e-05,
      "loss": 6.9356,
      "step": 1320
    },
    {
      "epoch": 0.053330641905530884,
      "grad_norm": 115.9233627319336,
      "learning_rate": 1.9971551956113386e-05,
      "loss": 3.0719,
      "step": 1321
    },
    {
      "epoch": 0.05337101332256762,
      "grad_norm": 126.41838073730469,
      "learning_rate": 1.9971453310910065e-05,
      "loss": 5.8304,
      "step": 1322
    },
    {
      "epoch": 0.05341138473960436,
      "grad_norm": 84.4926986694336,
      "learning_rate": 1.9971354495218183e-05,
      "loss": 7.2109,
      "step": 1323
    },
    {
      "epoch": 0.0534517561566411,
      "grad_norm": 207.32232666015625,
      "learning_rate": 1.997125550903942e-05,
      "loss": 4.9023,
      "step": 1324
    },
    {
      "epoch": 0.053492127573677835,
      "grad_norm": 132.0852508544922,
      "learning_rate": 1.9971156352375473e-05,
      "loss": 6.1959,
      "step": 1325
    },
    {
      "epoch": 0.05353249899071457,
      "grad_norm": 110.9395523071289,
      "learning_rate": 1.9971057025228034e-05,
      "loss": 6.5302,
      "step": 1326
    },
    {
      "epoch": 0.053572870407751313,
      "grad_norm": 105.74017333984375,
      "learning_rate": 1.9970957527598806e-05,
      "loss": 5.6086,
      "step": 1327
    },
    {
      "epoch": 0.05361324182478805,
      "grad_norm": 123.76982116699219,
      "learning_rate": 1.997085785948949e-05,
      "loss": 5.1862,
      "step": 1328
    },
    {
      "epoch": 0.053653613241824785,
      "grad_norm": 110.63179016113281,
      "learning_rate": 1.997075802090179e-05,
      "loss": 4.9269,
      "step": 1329
    },
    {
      "epoch": 0.05369398465886153,
      "grad_norm": 112.92047119140625,
      "learning_rate": 1.9970658011837404e-05,
      "loss": 5.3228,
      "step": 1330
    },
    {
      "epoch": 0.053734356075898264,
      "grad_norm": 109.95751953125,
      "learning_rate": 1.9970557832298058e-05,
      "loss": 5.0943,
      "step": 1331
    },
    {
      "epoch": 0.053774727492935,
      "grad_norm": 179.01670837402344,
      "learning_rate": 1.997045748228545e-05,
      "loss": 6.7075,
      "step": 1332
    },
    {
      "epoch": 0.05381509890997174,
      "grad_norm": 113.90896606445312,
      "learning_rate": 1.99703569618013e-05,
      "loss": 6.886,
      "step": 1333
    },
    {
      "epoch": 0.05385547032700848,
      "grad_norm": 134.4506378173828,
      "learning_rate": 1.9970256270847334e-05,
      "loss": 6.0885,
      "step": 1334
    },
    {
      "epoch": 0.053895841744045214,
      "grad_norm": 110.88272094726562,
      "learning_rate": 1.9970155409425268e-05,
      "loss": 3.5956,
      "step": 1335
    },
    {
      "epoch": 0.05393621316108196,
      "grad_norm": 108.34318542480469,
      "learning_rate": 1.997005437753682e-05,
      "loss": 4.9301,
      "step": 1336
    },
    {
      "epoch": 0.05397658457811869,
      "grad_norm": 171.2127685546875,
      "learning_rate": 1.996995317518373e-05,
      "loss": 6.5134,
      "step": 1337
    },
    {
      "epoch": 0.05401695599515543,
      "grad_norm": 127.78959655761719,
      "learning_rate": 1.996985180236772e-05,
      "loss": 5.5528,
      "step": 1338
    },
    {
      "epoch": 0.054057327412192165,
      "grad_norm": 120.22968292236328,
      "learning_rate": 1.9969750259090526e-05,
      "loss": 6.5759,
      "step": 1339
    },
    {
      "epoch": 0.05409769882922891,
      "grad_norm": 154.6633758544922,
      "learning_rate": 1.996964854535388e-05,
      "loss": 6.5567,
      "step": 1340
    },
    {
      "epoch": 0.05413807024626564,
      "grad_norm": 116.6657943725586,
      "learning_rate": 1.9969546661159527e-05,
      "loss": 4.835,
      "step": 1341
    },
    {
      "epoch": 0.05417844166330238,
      "grad_norm": 118.34307098388672,
      "learning_rate": 1.9969444606509213e-05,
      "loss": 8.3909,
      "step": 1342
    },
    {
      "epoch": 0.05421881308033912,
      "grad_norm": 536.1617431640625,
      "learning_rate": 1.996934238140467e-05,
      "loss": 5.6739,
      "step": 1343
    },
    {
      "epoch": 0.05425918449737586,
      "grad_norm": 249.28269958496094,
      "learning_rate": 1.9969239985847653e-05,
      "loss": 5.5741,
      "step": 1344
    },
    {
      "epoch": 0.054299555914412594,
      "grad_norm": 101.33755493164062,
      "learning_rate": 1.9969137419839913e-05,
      "loss": 6.1062,
      "step": 1345
    },
    {
      "epoch": 0.054339927331449336,
      "grad_norm": 109.74528503417969,
      "learning_rate": 1.99690346833832e-05,
      "loss": 6.5134,
      "step": 1346
    },
    {
      "epoch": 0.05438029874848607,
      "grad_norm": 149.09423828125,
      "learning_rate": 1.9968931776479276e-05,
      "loss": 5.654,
      "step": 1347
    },
    {
      "epoch": 0.05442067016552281,
      "grad_norm": 88.20384979248047,
      "learning_rate": 1.99688286991299e-05,
      "loss": 5.982,
      "step": 1348
    },
    {
      "epoch": 0.05446104158255955,
      "grad_norm": 140.06460571289062,
      "learning_rate": 1.996872545133683e-05,
      "loss": 6.4891,
      "step": 1349
    },
    {
      "epoch": 0.05450141299959629,
      "grad_norm": 98.46100616455078,
      "learning_rate": 1.996862203310183e-05,
      "loss": 5.2314,
      "step": 1350
    },
    {
      "epoch": 0.05454178441663302,
      "grad_norm": 82.77770233154297,
      "learning_rate": 1.9968518444426674e-05,
      "loss": 6.3449,
      "step": 1351
    },
    {
      "epoch": 0.05458215583366976,
      "grad_norm": 107.9808349609375,
      "learning_rate": 1.9968414685313133e-05,
      "loss": 6.2242,
      "step": 1352
    },
    {
      "epoch": 0.0546225272507065,
      "grad_norm": 93.7203598022461,
      "learning_rate": 1.9968310755762976e-05,
      "loss": 6.2333,
      "step": 1353
    },
    {
      "epoch": 0.05466289866774324,
      "grad_norm": 111.6139144897461,
      "learning_rate": 1.9968206655777983e-05,
      "loss": 6.4473,
      "step": 1354
    },
    {
      "epoch": 0.05470327008477997,
      "grad_norm": 179.6400604248047,
      "learning_rate": 1.9968102385359938e-05,
      "loss": 5.6957,
      "step": 1355
    },
    {
      "epoch": 0.054743641501816716,
      "grad_norm": 104.324951171875,
      "learning_rate": 1.9967997944510615e-05,
      "loss": 4.7908,
      "step": 1356
    },
    {
      "epoch": 0.05478401291885345,
      "grad_norm": 124.12268829345703,
      "learning_rate": 1.9967893333231804e-05,
      "loss": 4.5551,
      "step": 1357
    },
    {
      "epoch": 0.05482438433589019,
      "grad_norm": 109.22151947021484,
      "learning_rate": 1.9967788551525293e-05,
      "loss": 6.2264,
      "step": 1358
    },
    {
      "epoch": 0.05486475575292693,
      "grad_norm": 98.8077392578125,
      "learning_rate": 1.996768359939288e-05,
      "loss": 4.285,
      "step": 1359
    },
    {
      "epoch": 0.054905127169963666,
      "grad_norm": 136.1836700439453,
      "learning_rate": 1.9967578476836345e-05,
      "loss": 6.4551,
      "step": 1360
    },
    {
      "epoch": 0.0549454985870004,
      "grad_norm": 147.06640625,
      "learning_rate": 1.99674731838575e-05,
      "loss": 5.7439,
      "step": 1361
    },
    {
      "epoch": 0.054985870004037145,
      "grad_norm": 127.8466567993164,
      "learning_rate": 1.996736772045814e-05,
      "loss": 5.2484,
      "step": 1362
    },
    {
      "epoch": 0.05502624142107388,
      "grad_norm": 190.0849151611328,
      "learning_rate": 1.9967262086640063e-05,
      "loss": 5.989,
      "step": 1363
    },
    {
      "epoch": 0.055066612838110617,
      "grad_norm": 112.18577575683594,
      "learning_rate": 1.996715628240508e-05,
      "loss": 6.1048,
      "step": 1364
    },
    {
      "epoch": 0.05510698425514735,
      "grad_norm": 151.5193328857422,
      "learning_rate": 1.9967050307755005e-05,
      "loss": 4.8076,
      "step": 1365
    },
    {
      "epoch": 0.055147355672184095,
      "grad_norm": 206.59564208984375,
      "learning_rate": 1.9966944162691638e-05,
      "loss": 4.8581,
      "step": 1366
    },
    {
      "epoch": 0.05518772708922083,
      "grad_norm": 120.77896118164062,
      "learning_rate": 1.9966837847216805e-05,
      "loss": 5.4108,
      "step": 1367
    },
    {
      "epoch": 0.05522809850625757,
      "grad_norm": 123.58446502685547,
      "learning_rate": 1.996673136133232e-05,
      "loss": 6.6673,
      "step": 1368
    },
    {
      "epoch": 0.05526846992329431,
      "grad_norm": 100.06644439697266,
      "learning_rate": 1.9966624705039995e-05,
      "loss": 6.1017,
      "step": 1369
    },
    {
      "epoch": 0.055308841340331046,
      "grad_norm": 176.4990692138672,
      "learning_rate": 1.996651787834167e-05,
      "loss": 6.6409,
      "step": 1370
    },
    {
      "epoch": 0.05534921275736778,
      "grad_norm": 82.99078369140625,
      "learning_rate": 1.9966410881239156e-05,
      "loss": 5.961,
      "step": 1371
    },
    {
      "epoch": 0.055389584174404524,
      "grad_norm": 92.76287078857422,
      "learning_rate": 1.9966303713734295e-05,
      "loss": 6.4018,
      "step": 1372
    },
    {
      "epoch": 0.05542995559144126,
      "grad_norm": 98.77336883544922,
      "learning_rate": 1.996619637582891e-05,
      "loss": 5.3318,
      "step": 1373
    },
    {
      "epoch": 0.055470327008477996,
      "grad_norm": 86.7417221069336,
      "learning_rate": 1.996608886752484e-05,
      "loss": 6.3045,
      "step": 1374
    },
    {
      "epoch": 0.05551069842551474,
      "grad_norm": 124.87049102783203,
      "learning_rate": 1.9965981188823925e-05,
      "loss": 6.2822,
      "step": 1375
    },
    {
      "epoch": 0.055551069842551475,
      "grad_norm": 103.38391876220703,
      "learning_rate": 1.9965873339728002e-05,
      "loss": 6.3253,
      "step": 1376
    },
    {
      "epoch": 0.05559144125958821,
      "grad_norm": 77.05077362060547,
      "learning_rate": 1.9965765320238914e-05,
      "loss": 5.3065,
      "step": 1377
    },
    {
      "epoch": 0.055631812676624946,
      "grad_norm": 432.8467102050781,
      "learning_rate": 1.9965657130358516e-05,
      "loss": 6.0368,
      "step": 1378
    },
    {
      "epoch": 0.05567218409366169,
      "grad_norm": 199.44313049316406,
      "learning_rate": 1.996554877008865e-05,
      "loss": 5.5631,
      "step": 1379
    },
    {
      "epoch": 0.055712555510698425,
      "grad_norm": 91.54793548583984,
      "learning_rate": 1.9965440239431172e-05,
      "loss": 6.2522,
      "step": 1380
    },
    {
      "epoch": 0.05575292692773516,
      "grad_norm": 128.38710021972656,
      "learning_rate": 1.9965331538387936e-05,
      "loss": 5.2115,
      "step": 1381
    },
    {
      "epoch": 0.055793298344771904,
      "grad_norm": 192.83351135253906,
      "learning_rate": 1.99652226669608e-05,
      "loss": 7.2918,
      "step": 1382
    },
    {
      "epoch": 0.05583366976180864,
      "grad_norm": 151.4713134765625,
      "learning_rate": 1.9965113625151632e-05,
      "loss": 5.8622,
      "step": 1383
    },
    {
      "epoch": 0.055874041178845375,
      "grad_norm": 84.69669342041016,
      "learning_rate": 1.9965004412962288e-05,
      "loss": 4.2366,
      "step": 1384
    },
    {
      "epoch": 0.05591441259588212,
      "grad_norm": 100.39618682861328,
      "learning_rate": 1.9964895030394638e-05,
      "loss": 5.988,
      "step": 1385
    },
    {
      "epoch": 0.055954784012918854,
      "grad_norm": 107.23945617675781,
      "learning_rate": 1.9964785477450556e-05,
      "loss": 8.0653,
      "step": 1386
    },
    {
      "epoch": 0.05599515542995559,
      "grad_norm": 120.10631561279297,
      "learning_rate": 1.9964675754131905e-05,
      "loss": 5.8852,
      "step": 1387
    },
    {
      "epoch": 0.05603552684699233,
      "grad_norm": 155.29847717285156,
      "learning_rate": 1.9964565860440574e-05,
      "loss": 5.0119,
      "step": 1388
    },
    {
      "epoch": 0.05607589826402907,
      "grad_norm": 95.49667358398438,
      "learning_rate": 1.996445579637843e-05,
      "loss": 4.4053,
      "step": 1389
    },
    {
      "epoch": 0.056116269681065804,
      "grad_norm": 110.25637817382812,
      "learning_rate": 1.9964345561947364e-05,
      "loss": 7.5672,
      "step": 1390
    },
    {
      "epoch": 0.05615664109810254,
      "grad_norm": 124.48099517822266,
      "learning_rate": 1.996423515714926e-05,
      "loss": 4.5659,
      "step": 1391
    },
    {
      "epoch": 0.05619701251513928,
      "grad_norm": 77.3618392944336,
      "learning_rate": 1.9964124581986e-05,
      "loss": 6.2912,
      "step": 1392
    },
    {
      "epoch": 0.05623738393217602,
      "grad_norm": 108.55380249023438,
      "learning_rate": 1.996401383645948e-05,
      "loss": 6.4321,
      "step": 1393
    },
    {
      "epoch": 0.056277755349212755,
      "grad_norm": 102.7417984008789,
      "learning_rate": 1.9963902920571588e-05,
      "loss": 6.3942,
      "step": 1394
    },
    {
      "epoch": 0.0563181267662495,
      "grad_norm": 275.8184509277344,
      "learning_rate": 1.9963791834324223e-05,
      "loss": 6.3021,
      "step": 1395
    },
    {
      "epoch": 0.05635849818328623,
      "grad_norm": 137.24957275390625,
      "learning_rate": 1.9963680577719288e-05,
      "loss": 5.9949,
      "step": 1396
    },
    {
      "epoch": 0.05639886960032297,
      "grad_norm": 145.0596160888672,
      "learning_rate": 1.9963569150758676e-05,
      "loss": 4.0378,
      "step": 1397
    },
    {
      "epoch": 0.05643924101735971,
      "grad_norm": 120.41510772705078,
      "learning_rate": 1.9963457553444306e-05,
      "loss": 6.2029,
      "step": 1398
    },
    {
      "epoch": 0.05647961243439645,
      "grad_norm": 93.96385192871094,
      "learning_rate": 1.996334578577807e-05,
      "loss": 6.0202,
      "step": 1399
    },
    {
      "epoch": 0.056519983851433184,
      "grad_norm": 214.61378479003906,
      "learning_rate": 1.9963233847761896e-05,
      "loss": 6.7777,
      "step": 1400
    },
    {
      "epoch": 0.056560355268469927,
      "grad_norm": 115.35298919677734,
      "learning_rate": 1.9963121739397685e-05,
      "loss": 4.9062,
      "step": 1401
    },
    {
      "epoch": 0.05660072668550666,
      "grad_norm": 122.58586120605469,
      "learning_rate": 1.9963009460687356e-05,
      "loss": 5.0145,
      "step": 1402
    },
    {
      "epoch": 0.0566410981025434,
      "grad_norm": 197.60891723632812,
      "learning_rate": 1.9962897011632832e-05,
      "loss": 5.3184,
      "step": 1403
    },
    {
      "epoch": 0.056681469519580134,
      "grad_norm": 174.8152313232422,
      "learning_rate": 1.9962784392236036e-05,
      "loss": 7.3522,
      "step": 1404
    },
    {
      "epoch": 0.05672184093661688,
      "grad_norm": 138.59405517578125,
      "learning_rate": 1.9962671602498888e-05,
      "loss": 5.4411,
      "step": 1405
    },
    {
      "epoch": 0.05676221235365361,
      "grad_norm": 122.01170349121094,
      "learning_rate": 1.9962558642423322e-05,
      "loss": 6.8458,
      "step": 1406
    },
    {
      "epoch": 0.05680258377069035,
      "grad_norm": 224.06895446777344,
      "learning_rate": 1.9962445512011264e-05,
      "loss": 6.931,
      "step": 1407
    },
    {
      "epoch": 0.05684295518772709,
      "grad_norm": 226.90664672851562,
      "learning_rate": 1.996233221126466e-05,
      "loss": 7.0031,
      "step": 1408
    },
    {
      "epoch": 0.05688332660476383,
      "grad_norm": 190.85533142089844,
      "learning_rate": 1.9962218740185432e-05,
      "loss": 6.006,
      "step": 1409
    },
    {
      "epoch": 0.05692369802180056,
      "grad_norm": 146.76158142089844,
      "learning_rate": 1.9962105098775528e-05,
      "loss": 6.3435,
      "step": 1410
    },
    {
      "epoch": 0.056964069438837306,
      "grad_norm": 175.10829162597656,
      "learning_rate": 1.996199128703689e-05,
      "loss": 6.0849,
      "step": 1411
    },
    {
      "epoch": 0.05700444085587404,
      "grad_norm": 160.74778747558594,
      "learning_rate": 1.9961877304971468e-05,
      "loss": 6.2549,
      "step": 1412
    },
    {
      "epoch": 0.05704481227291078,
      "grad_norm": 297.36285400390625,
      "learning_rate": 1.9961763152581203e-05,
      "loss": 6.5934,
      "step": 1413
    },
    {
      "epoch": 0.05708518368994752,
      "grad_norm": 103.0043716430664,
      "learning_rate": 1.996164882986805e-05,
      "loss": 7.2749,
      "step": 1414
    },
    {
      "epoch": 0.057125555106984256,
      "grad_norm": 237.12811279296875,
      "learning_rate": 1.9961534336833966e-05,
      "loss": 4.3022,
      "step": 1415
    },
    {
      "epoch": 0.05716592652402099,
      "grad_norm": 121.41175842285156,
      "learning_rate": 1.9961419673480908e-05,
      "loss": 6.7779,
      "step": 1416
    },
    {
      "epoch": 0.05720629794105773,
      "grad_norm": 119.34642028808594,
      "learning_rate": 1.996130483981083e-05,
      "loss": 6.3724,
      "step": 1417
    },
    {
      "epoch": 0.05724666935809447,
      "grad_norm": 120.90609741210938,
      "learning_rate": 1.9961189835825706e-05,
      "loss": 7.0254,
      "step": 1418
    },
    {
      "epoch": 0.05728704077513121,
      "grad_norm": 103.82850646972656,
      "learning_rate": 1.9961074661527494e-05,
      "loss": 6.697,
      "step": 1419
    },
    {
      "epoch": 0.05732741219216794,
      "grad_norm": 113.8655776977539,
      "learning_rate": 1.996095931691817e-05,
      "loss": 5.7981,
      "step": 1420
    },
    {
      "epoch": 0.057367783609204685,
      "grad_norm": 150.4202880859375,
      "learning_rate": 1.9960843801999693e-05,
      "loss": 4.632,
      "step": 1421
    },
    {
      "epoch": 0.05740815502624142,
      "grad_norm": 110.51494598388672,
      "learning_rate": 1.9960728116774053e-05,
      "loss": 5.7342,
      "step": 1422
    },
    {
      "epoch": 0.05744852644327816,
      "grad_norm": 151.89984130859375,
      "learning_rate": 1.9960612261243222e-05,
      "loss": 4.6578,
      "step": 1423
    },
    {
      "epoch": 0.0574888978603149,
      "grad_norm": 163.43869018554688,
      "learning_rate": 1.9960496235409184e-05,
      "loss": 5.956,
      "step": 1424
    },
    {
      "epoch": 0.057529269277351636,
      "grad_norm": 132.24447631835938,
      "learning_rate": 1.9960380039273914e-05,
      "loss": 7.9225,
      "step": 1425
    },
    {
      "epoch": 0.05756964069438837,
      "grad_norm": 116.61181640625,
      "learning_rate": 1.9960263672839407e-05,
      "loss": 3.5009,
      "step": 1426
    },
    {
      "epoch": 0.057610012111425114,
      "grad_norm": 158.34695434570312,
      "learning_rate": 1.996014713610765e-05,
      "loss": 5.5728,
      "step": 1427
    },
    {
      "epoch": 0.05765038352846185,
      "grad_norm": 97.8939437866211,
      "learning_rate": 1.996003042908064e-05,
      "loss": 5.1086,
      "step": 1428
    },
    {
      "epoch": 0.057690754945498586,
      "grad_norm": 153.2526397705078,
      "learning_rate": 1.9959913551760363e-05,
      "loss": 5.3998,
      "step": 1429
    },
    {
      "epoch": 0.05773112636253532,
      "grad_norm": 95.66490936279297,
      "learning_rate": 1.9959796504148824e-05,
      "loss": 6.3313,
      "step": 1430
    },
    {
      "epoch": 0.057771497779572065,
      "grad_norm": 138.32958984375,
      "learning_rate": 1.9959679286248022e-05,
      "loss": 6.695,
      "step": 1431
    },
    {
      "epoch": 0.0578118691966088,
      "grad_norm": 87.51480102539062,
      "learning_rate": 1.995956189805996e-05,
      "loss": 6.2292,
      "step": 1432
    },
    {
      "epoch": 0.057852240613645536,
      "grad_norm": 124.3735122680664,
      "learning_rate": 1.995944433958665e-05,
      "loss": 3.7578,
      "step": 1433
    },
    {
      "epoch": 0.05789261203068228,
      "grad_norm": 113.23728942871094,
      "learning_rate": 1.9959326610830097e-05,
      "loss": 6.4644,
      "step": 1434
    },
    {
      "epoch": 0.057932983447719015,
      "grad_norm": 112.90052032470703,
      "learning_rate": 1.9959208711792318e-05,
      "loss": 7.7881,
      "step": 1435
    },
    {
      "epoch": 0.05797335486475575,
      "grad_norm": 85.04566955566406,
      "learning_rate": 1.9959090642475323e-05,
      "loss": 5.4968,
      "step": 1436
    },
    {
      "epoch": 0.058013726281792494,
      "grad_norm": 108.26045227050781,
      "learning_rate": 1.9958972402881135e-05,
      "loss": 5.3725,
      "step": 1437
    },
    {
      "epoch": 0.05805409769882923,
      "grad_norm": 105.72722625732422,
      "learning_rate": 1.9958853993011775e-05,
      "loss": 5.2276,
      "step": 1438
    },
    {
      "epoch": 0.058094469115865965,
      "grad_norm": 170.52316284179688,
      "learning_rate": 1.995873541286927e-05,
      "loss": 5.8147,
      "step": 1439
    },
    {
      "epoch": 0.05813484053290271,
      "grad_norm": 89.7197036743164,
      "learning_rate": 1.9958616662455645e-05,
      "loss": 5.3027,
      "step": 1440
    },
    {
      "epoch": 0.058175211949939444,
      "grad_norm": 106.30420684814453,
      "learning_rate": 1.9958497741772926e-05,
      "loss": 5.9191,
      "step": 1441
    },
    {
      "epoch": 0.05821558336697618,
      "grad_norm": 136.28073120117188,
      "learning_rate": 1.995837865082315e-05,
      "loss": 5.8831,
      "step": 1442
    },
    {
      "epoch": 0.058255954784012916,
      "grad_norm": 166.91238403320312,
      "learning_rate": 1.9958259389608355e-05,
      "loss": 6.4079,
      "step": 1443
    },
    {
      "epoch": 0.05829632620104966,
      "grad_norm": 114.32855224609375,
      "learning_rate": 1.995813995813058e-05,
      "loss": 7.2184,
      "step": 1444
    },
    {
      "epoch": 0.058336697618086394,
      "grad_norm": 106.43305969238281,
      "learning_rate": 1.9958020356391866e-05,
      "loss": 5.7189,
      "step": 1445
    },
    {
      "epoch": 0.05837706903512313,
      "grad_norm": 127.0048599243164,
      "learning_rate": 1.9957900584394255e-05,
      "loss": 5.5501,
      "step": 1446
    },
    {
      "epoch": 0.05841744045215987,
      "grad_norm": 234.7830352783203,
      "learning_rate": 1.99577806421398e-05,
      "loss": 4.2056,
      "step": 1447
    },
    {
      "epoch": 0.05845781186919661,
      "grad_norm": 106.35542297363281,
      "learning_rate": 1.9957660529630546e-05,
      "loss": 5.9943,
      "step": 1448
    },
    {
      "epoch": 0.058498183286233345,
      "grad_norm": 114.5039291381836,
      "learning_rate": 1.995754024686855e-05,
      "loss": 6.233,
      "step": 1449
    },
    {
      "epoch": 0.05853855470327009,
      "grad_norm": 394.344482421875,
      "learning_rate": 1.995741979385587e-05,
      "loss": 4.6356,
      "step": 1450
    },
    {
      "epoch": 0.05857892612030682,
      "grad_norm": 281.47271728515625,
      "learning_rate": 1.9957299170594567e-05,
      "loss": 6.2634,
      "step": 1451
    },
    {
      "epoch": 0.05861929753734356,
      "grad_norm": 117.24767303466797,
      "learning_rate": 1.9957178377086697e-05,
      "loss": 5.823,
      "step": 1452
    },
    {
      "epoch": 0.0586596689543803,
      "grad_norm": 122.3797378540039,
      "learning_rate": 1.9957057413334326e-05,
      "loss": 5.9227,
      "step": 1453
    },
    {
      "epoch": 0.05870004037141704,
      "grad_norm": 114.74100494384766,
      "learning_rate": 1.9956936279339527e-05,
      "loss": 5.4567,
      "step": 1454
    },
    {
      "epoch": 0.058740411788453774,
      "grad_norm": 132.6952362060547,
      "learning_rate": 1.995681497510437e-05,
      "loss": 4.924,
      "step": 1455
    },
    {
      "epoch": 0.05878078320549051,
      "grad_norm": 101.29682922363281,
      "learning_rate": 1.9956693500630928e-05,
      "loss": 6.1477,
      "step": 1456
    },
    {
      "epoch": 0.05882115462252725,
      "grad_norm": 197.72779846191406,
      "learning_rate": 1.9956571855921276e-05,
      "loss": 5.4604,
      "step": 1457
    },
    {
      "epoch": 0.05886152603956399,
      "grad_norm": 152.42259216308594,
      "learning_rate": 1.9956450040977497e-05,
      "loss": 3.787,
      "step": 1458
    },
    {
      "epoch": 0.058901897456600724,
      "grad_norm": 100.11444854736328,
      "learning_rate": 1.9956328055801667e-05,
      "loss": 5.8518,
      "step": 1459
    },
    {
      "epoch": 0.05894226887363747,
      "grad_norm": 352.9201965332031,
      "learning_rate": 1.995620590039588e-05,
      "loss": 5.5975,
      "step": 1460
    },
    {
      "epoch": 0.0589826402906742,
      "grad_norm": 274.33172607421875,
      "learning_rate": 1.9956083574762223e-05,
      "loss": 7.5685,
      "step": 1461
    },
    {
      "epoch": 0.05902301170771094,
      "grad_norm": 246.66305541992188,
      "learning_rate": 1.9955961078902786e-05,
      "loss": 5.3661,
      "step": 1462
    },
    {
      "epoch": 0.05906338312474768,
      "grad_norm": 162.47286987304688,
      "learning_rate": 1.9955838412819664e-05,
      "loss": 8.0901,
      "step": 1463
    },
    {
      "epoch": 0.05910375454178442,
      "grad_norm": 727.5283203125,
      "learning_rate": 1.9955715576514954e-05,
      "loss": 5.0392,
      "step": 1464
    },
    {
      "epoch": 0.05914412595882115,
      "grad_norm": 175.2860870361328,
      "learning_rate": 1.995559256999075e-05,
      "loss": 6.2813,
      "step": 1465
    },
    {
      "epoch": 0.059184497375857896,
      "grad_norm": 143.57073974609375,
      "learning_rate": 1.995546939324917e-05,
      "loss": 4.8084,
      "step": 1466
    },
    {
      "epoch": 0.05922486879289463,
      "grad_norm": 100.51502227783203,
      "learning_rate": 1.9955346046292307e-05,
      "loss": 5.2998,
      "step": 1467
    },
    {
      "epoch": 0.05926524020993137,
      "grad_norm": 215.86383056640625,
      "learning_rate": 1.995522252912227e-05,
      "loss": 6.4338,
      "step": 1468
    },
    {
      "epoch": 0.059305611626968104,
      "grad_norm": 116.17451477050781,
      "learning_rate": 1.995509884174118e-05,
      "loss": 5.1659,
      "step": 1469
    },
    {
      "epoch": 0.059345983044004846,
      "grad_norm": 175.90036010742188,
      "learning_rate": 1.9954974984151146e-05,
      "loss": 4.7145,
      "step": 1470
    },
    {
      "epoch": 0.05938635446104158,
      "grad_norm": 163.61471557617188,
      "learning_rate": 1.9954850956354283e-05,
      "loss": 6.458,
      "step": 1471
    },
    {
      "epoch": 0.05942672587807832,
      "grad_norm": 143.47288513183594,
      "learning_rate": 1.995472675835272e-05,
      "loss": 6.1447,
      "step": 1472
    },
    {
      "epoch": 0.05946709729511506,
      "grad_norm": 172.5647430419922,
      "learning_rate": 1.995460239014857e-05,
      "loss": 4.1914,
      "step": 1473
    },
    {
      "epoch": 0.0595074687121518,
      "grad_norm": 128.33624267578125,
      "learning_rate": 1.995447785174397e-05,
      "loss": 4.8591,
      "step": 1474
    },
    {
      "epoch": 0.05954784012918853,
      "grad_norm": 106.87279510498047,
      "learning_rate": 1.9954353143141038e-05,
      "loss": 4.9698,
      "step": 1475
    },
    {
      "epoch": 0.059588211546225275,
      "grad_norm": 100.51335906982422,
      "learning_rate": 1.9954228264341916e-05,
      "loss": 5.5765,
      "step": 1476
    },
    {
      "epoch": 0.05962858296326201,
      "grad_norm": 193.88165283203125,
      "learning_rate": 1.9954103215348736e-05,
      "loss": 5.1797,
      "step": 1477
    },
    {
      "epoch": 0.05966895438029875,
      "grad_norm": 397.9445495605469,
      "learning_rate": 1.9953977996163638e-05,
      "loss": 6.0,
      "step": 1478
    },
    {
      "epoch": 0.05970932579733549,
      "grad_norm": 167.7197265625,
      "learning_rate": 1.9953852606788756e-05,
      "loss": 6.2254,
      "step": 1479
    },
    {
      "epoch": 0.059749697214372226,
      "grad_norm": 90.08423614501953,
      "learning_rate": 1.995372704722624e-05,
      "loss": 5.882,
      "step": 1480
    },
    {
      "epoch": 0.05979006863140896,
      "grad_norm": 200.923828125,
      "learning_rate": 1.9953601317478233e-05,
      "loss": 6.7769,
      "step": 1481
    },
    {
      "epoch": 0.0598304400484457,
      "grad_norm": 148.6149139404297,
      "learning_rate": 1.995347541754689e-05,
      "loss": 5.6193,
      "step": 1482
    },
    {
      "epoch": 0.05987081146548244,
      "grad_norm": 124.97581481933594,
      "learning_rate": 1.995334934743436e-05,
      "loss": 4.8216,
      "step": 1483
    },
    {
      "epoch": 0.059911182882519176,
      "grad_norm": 409.52490234375,
      "learning_rate": 1.9953223107142797e-05,
      "loss": 6.2868,
      "step": 1484
    },
    {
      "epoch": 0.05995155429955591,
      "grad_norm": 89.46250915527344,
      "learning_rate": 1.995309669667436e-05,
      "loss": 5.8252,
      "step": 1485
    },
    {
      "epoch": 0.059991925716592655,
      "grad_norm": 128.5953369140625,
      "learning_rate": 1.9952970116031216e-05,
      "loss": 5.7413,
      "step": 1486
    },
    {
      "epoch": 0.06003229713362939,
      "grad_norm": 96.714599609375,
      "learning_rate": 1.9952843365215522e-05,
      "loss": 6.3587,
      "step": 1487
    },
    {
      "epoch": 0.060072668550666126,
      "grad_norm": 258.1334228515625,
      "learning_rate": 1.9952716444229448e-05,
      "loss": 6.4257,
      "step": 1488
    },
    {
      "epoch": 0.06011303996770287,
      "grad_norm": 104.4274673461914,
      "learning_rate": 1.9952589353075167e-05,
      "loss": 5.3646,
      "step": 1489
    },
    {
      "epoch": 0.060153411384739605,
      "grad_norm": 90.27049255371094,
      "learning_rate": 1.9952462091754846e-05,
      "loss": 6.375,
      "step": 1490
    },
    {
      "epoch": 0.06019378280177634,
      "grad_norm": 107.02213287353516,
      "learning_rate": 1.995233466027066e-05,
      "loss": 6.0451,
      "step": 1491
    },
    {
      "epoch": 0.060234154218813084,
      "grad_norm": 121.76402282714844,
      "learning_rate": 1.9952207058624798e-05,
      "loss": 6.342,
      "step": 1492
    },
    {
      "epoch": 0.06027452563584982,
      "grad_norm": 158.2082977294922,
      "learning_rate": 1.9952079286819438e-05,
      "loss": 6.1938,
      "step": 1493
    },
    {
      "epoch": 0.060314897052886556,
      "grad_norm": 184.11541748046875,
      "learning_rate": 1.9951951344856756e-05,
      "loss": 6.7473,
      "step": 1494
    },
    {
      "epoch": 0.06035526846992329,
      "grad_norm": 168.67408752441406,
      "learning_rate": 1.995182323273895e-05,
      "loss": 5.8088,
      "step": 1495
    },
    {
      "epoch": 0.060395639886960034,
      "grad_norm": 98.55520629882812,
      "learning_rate": 1.99516949504682e-05,
      "loss": 7.248,
      "step": 1496
    },
    {
      "epoch": 0.06043601130399677,
      "grad_norm": 117.94253540039062,
      "learning_rate": 1.9951566498046707e-05,
      "loss": 6.5466,
      "step": 1497
    },
    {
      "epoch": 0.060476382721033506,
      "grad_norm": 119.6575698852539,
      "learning_rate": 1.995143787547667e-05,
      "loss": 6.8005,
      "step": 1498
    },
    {
      "epoch": 0.06051675413807025,
      "grad_norm": 129.17994689941406,
      "learning_rate": 1.995130908276028e-05,
      "loss": 5.6495,
      "step": 1499
    },
    {
      "epoch": 0.060557125555106985,
      "grad_norm": 101.28902435302734,
      "learning_rate": 1.995118011989974e-05,
      "loss": 5.7256,
      "step": 1500
    },
    {
      "epoch": 0.06059749697214372,
      "grad_norm": 152.60130310058594,
      "learning_rate": 1.9951050986897263e-05,
      "loss": 6.9197,
      "step": 1501
    },
    {
      "epoch": 0.06063786838918046,
      "grad_norm": 149.74072265625,
      "learning_rate": 1.9950921683755048e-05,
      "loss": 6.6978,
      "step": 1502
    },
    {
      "epoch": 0.0606782398062172,
      "grad_norm": 196.50799560546875,
      "learning_rate": 1.995079221047531e-05,
      "loss": 6.169,
      "step": 1503
    },
    {
      "epoch": 0.060718611223253935,
      "grad_norm": 152.36219787597656,
      "learning_rate": 1.9950662567060265e-05,
      "loss": 6.3511,
      "step": 1504
    },
    {
      "epoch": 0.06075898264029067,
      "grad_norm": 147.63780212402344,
      "learning_rate": 1.9950532753512123e-05,
      "loss": 6.5609,
      "step": 1505
    },
    {
      "epoch": 0.060799354057327414,
      "grad_norm": 171.07525634765625,
      "learning_rate": 1.9950402769833108e-05,
      "loss": 5.7201,
      "step": 1506
    },
    {
      "epoch": 0.06083972547436415,
      "grad_norm": 107.9576416015625,
      "learning_rate": 1.995027261602544e-05,
      "loss": 6.1627,
      "step": 1507
    },
    {
      "epoch": 0.060880096891400885,
      "grad_norm": 102.08840942382812,
      "learning_rate": 1.995014229209135e-05,
      "loss": 5.7113,
      "step": 1508
    },
    {
      "epoch": 0.06092046830843763,
      "grad_norm": 82.65536499023438,
      "learning_rate": 1.995001179803306e-05,
      "loss": 5.8582,
      "step": 1509
    },
    {
      "epoch": 0.060960839725474364,
      "grad_norm": 155.54110717773438,
      "learning_rate": 1.9949881133852803e-05,
      "loss": 5.0617,
      "step": 1510
    },
    {
      "epoch": 0.0610012111425111,
      "grad_norm": 149.17276000976562,
      "learning_rate": 1.9949750299552812e-05,
      "loss": 4.6269,
      "step": 1511
    },
    {
      "epoch": 0.06104158255954784,
      "grad_norm": 76.03850555419922,
      "learning_rate": 1.9949619295135327e-05,
      "loss": 5.2553,
      "step": 1512
    },
    {
      "epoch": 0.06108195397658458,
      "grad_norm": 185.4452667236328,
      "learning_rate": 1.9949488120602584e-05,
      "loss": 4.8772,
      "step": 1513
    },
    {
      "epoch": 0.061122325393621314,
      "grad_norm": 155.29129028320312,
      "learning_rate": 1.9949356775956826e-05,
      "loss": 5.3937,
      "step": 1514
    },
    {
      "epoch": 0.06116269681065806,
      "grad_norm": 122.78433227539062,
      "learning_rate": 1.9949225261200306e-05,
      "loss": 5.3617,
      "step": 1515
    },
    {
      "epoch": 0.06120306822769479,
      "grad_norm": 95.36320495605469,
      "learning_rate": 1.9949093576335264e-05,
      "loss": 6.8574,
      "step": 1516
    },
    {
      "epoch": 0.06124343964473153,
      "grad_norm": 93.16735076904297,
      "learning_rate": 1.9948961721363956e-05,
      "loss": 6.2139,
      "step": 1517
    },
    {
      "epoch": 0.061283811061768265,
      "grad_norm": 109.51972198486328,
      "learning_rate": 1.9948829696288635e-05,
      "loss": 5.2139,
      "step": 1518
    },
    {
      "epoch": 0.06132418247880501,
      "grad_norm": 515.4991455078125,
      "learning_rate": 1.994869750111156e-05,
      "loss": 6.0709,
      "step": 1519
    },
    {
      "epoch": 0.06136455389584174,
      "grad_norm": 119.90030670166016,
      "learning_rate": 1.9948565135834986e-05,
      "loss": 7.4604,
      "step": 1520
    },
    {
      "epoch": 0.06140492531287848,
      "grad_norm": 117.67559814453125,
      "learning_rate": 1.994843260046118e-05,
      "loss": 6.2452,
      "step": 1521
    },
    {
      "epoch": 0.06144529672991522,
      "grad_norm": 98.96373748779297,
      "learning_rate": 1.9948299894992407e-05,
      "loss": 6.6372,
      "step": 1522
    },
    {
      "epoch": 0.06148566814695196,
      "grad_norm": 125.09652709960938,
      "learning_rate": 1.9948167019430942e-05,
      "loss": 6.6684,
      "step": 1523
    },
    {
      "epoch": 0.061526039563988694,
      "grad_norm": 181.398681640625,
      "learning_rate": 1.9948033973779047e-05,
      "loss": 5.368,
      "step": 1524
    },
    {
      "epoch": 0.061566410981025436,
      "grad_norm": 95.960205078125,
      "learning_rate": 1.9947900758039003e-05,
      "loss": 4.869,
      "step": 1525
    },
    {
      "epoch": 0.06160678239806217,
      "grad_norm": 188.16563415527344,
      "learning_rate": 1.9947767372213086e-05,
      "loss": 4.5874,
      "step": 1526
    },
    {
      "epoch": 0.06164715381509891,
      "grad_norm": 241.0452117919922,
      "learning_rate": 1.9947633816303577e-05,
      "loss": 4.8891,
      "step": 1527
    },
    {
      "epoch": 0.06168752523213565,
      "grad_norm": 122.46715545654297,
      "learning_rate": 1.994750009031276e-05,
      "loss": 4.4882,
      "step": 1528
    },
    {
      "epoch": 0.06172789664917239,
      "grad_norm": 111.7219009399414,
      "learning_rate": 1.9947366194242915e-05,
      "loss": 5.9861,
      "step": 1529
    },
    {
      "epoch": 0.06176826806620912,
      "grad_norm": 107.13919067382812,
      "learning_rate": 1.9947232128096344e-05,
      "loss": 5.6008,
      "step": 1530
    },
    {
      "epoch": 0.06180863948324586,
      "grad_norm": 88.89949798583984,
      "learning_rate": 1.9947097891875328e-05,
      "loss": 5.3433,
      "step": 1531
    },
    {
      "epoch": 0.0618490109002826,
      "grad_norm": 113.71337127685547,
      "learning_rate": 1.994696348558217e-05,
      "loss": 5.7972,
      "step": 1532
    },
    {
      "epoch": 0.06188938231731934,
      "grad_norm": 111.29682159423828,
      "learning_rate": 1.994682890921916e-05,
      "loss": 4.7471,
      "step": 1533
    },
    {
      "epoch": 0.06192975373435607,
      "grad_norm": 85.62887573242188,
      "learning_rate": 1.9946694162788606e-05,
      "loss": 5.9239,
      "step": 1534
    },
    {
      "epoch": 0.061970125151392816,
      "grad_norm": 132.90965270996094,
      "learning_rate": 1.9946559246292808e-05,
      "loss": 6.3693,
      "step": 1535
    },
    {
      "epoch": 0.06201049656842955,
      "grad_norm": 531.9140625,
      "learning_rate": 1.994642415973407e-05,
      "loss": 5.584,
      "step": 1536
    },
    {
      "epoch": 0.06205086798546629,
      "grad_norm": 132.43699645996094,
      "learning_rate": 1.994628890311471e-05,
      "loss": 5.7553,
      "step": 1537
    },
    {
      "epoch": 0.06209123940250303,
      "grad_norm": 81.63340759277344,
      "learning_rate": 1.9946153476437033e-05,
      "loss": 6.345,
      "step": 1538
    },
    {
      "epoch": 0.062131610819539766,
      "grad_norm": 118.49739837646484,
      "learning_rate": 1.9946017879703362e-05,
      "loss": 5.1963,
      "step": 1539
    },
    {
      "epoch": 0.0621719822365765,
      "grad_norm": 186.54367065429688,
      "learning_rate": 1.9945882112916006e-05,
      "loss": 5.4134,
      "step": 1540
    },
    {
      "epoch": 0.062212353653613245,
      "grad_norm": 116.50935363769531,
      "learning_rate": 1.9945746176077296e-05,
      "loss": 6.4258,
      "step": 1541
    },
    {
      "epoch": 0.06225272507064998,
      "grad_norm": 106.38423919677734,
      "learning_rate": 1.9945610069189546e-05,
      "loss": 5.9296,
      "step": 1542
    },
    {
      "epoch": 0.06229309648768672,
      "grad_norm": 131.09840393066406,
      "learning_rate": 1.9945473792255093e-05,
      "loss": 5.9549,
      "step": 1543
    },
    {
      "epoch": 0.06233346790472345,
      "grad_norm": 145.5330810546875,
      "learning_rate": 1.9945337345276262e-05,
      "loss": 5.804,
      "step": 1544
    },
    {
      "epoch": 0.062373839321760195,
      "grad_norm": 128.3846435546875,
      "learning_rate": 1.9945200728255384e-05,
      "loss": 7.5283,
      "step": 1545
    },
    {
      "epoch": 0.06241421073879693,
      "grad_norm": 100.63314819335938,
      "learning_rate": 1.9945063941194797e-05,
      "loss": 5.5348,
      "step": 1546
    },
    {
      "epoch": 0.06245458215583367,
      "grad_norm": 113.97775268554688,
      "learning_rate": 1.994492698409684e-05,
      "loss": 5.6439,
      "step": 1547
    },
    {
      "epoch": 0.06249495357287041,
      "grad_norm": 132.90866088867188,
      "learning_rate": 1.994478985696386e-05,
      "loss": 6.7947,
      "step": 1548
    },
    {
      "epoch": 0.06253532498990715,
      "grad_norm": 146.5832061767578,
      "learning_rate": 1.994465255979819e-05,
      "loss": 6.4733,
      "step": 1549
    },
    {
      "epoch": 0.06257569640694388,
      "grad_norm": 163.27786254882812,
      "learning_rate": 1.9944515092602188e-05,
      "loss": 5.1701,
      "step": 1550
    },
    {
      "epoch": 0.06261606782398062,
      "grad_norm": 126.55949401855469,
      "learning_rate": 1.9944377455378196e-05,
      "loss": 5.708,
      "step": 1551
    },
    {
      "epoch": 0.06265643924101735,
      "grad_norm": 124.85954284667969,
      "learning_rate": 1.9944239648128576e-05,
      "loss": 6.5026,
      "step": 1552
    },
    {
      "epoch": 0.0626968106580541,
      "grad_norm": 119.41648864746094,
      "learning_rate": 1.9944101670855675e-05,
      "loss": 5.6558,
      "step": 1553
    },
    {
      "epoch": 0.06273718207509084,
      "grad_norm": 63.013206481933594,
      "learning_rate": 1.994396352356186e-05,
      "loss": 5.7986,
      "step": 1554
    },
    {
      "epoch": 0.06277755349212757,
      "grad_norm": 83.4732437133789,
      "learning_rate": 1.9943825206249486e-05,
      "loss": 5.8906,
      "step": 1555
    },
    {
      "epoch": 0.06281792490916431,
      "grad_norm": 110.49275970458984,
      "learning_rate": 1.9943686718920923e-05,
      "loss": 6.6545,
      "step": 1556
    },
    {
      "epoch": 0.06285829632620105,
      "grad_norm": 118.57340240478516,
      "learning_rate": 1.994354806157854e-05,
      "loss": 6.2004,
      "step": 1557
    },
    {
      "epoch": 0.06289866774323778,
      "grad_norm": 314.4451904296875,
      "learning_rate": 1.9943409234224702e-05,
      "loss": 6.0065,
      "step": 1558
    },
    {
      "epoch": 0.06293903916027453,
      "grad_norm": 92.13491821289062,
      "learning_rate": 1.9943270236861787e-05,
      "loss": 6.869,
      "step": 1559
    },
    {
      "epoch": 0.06297941057731127,
      "grad_norm": 128.2556915283203,
      "learning_rate": 1.994313106949217e-05,
      "loss": 6.2046,
      "step": 1560
    },
    {
      "epoch": 0.063019781994348,
      "grad_norm": 107.02442169189453,
      "learning_rate": 1.9942991732118228e-05,
      "loss": 4.7865,
      "step": 1561
    },
    {
      "epoch": 0.06306015341138474,
      "grad_norm": 130.2424774169922,
      "learning_rate": 1.9942852224742348e-05,
      "loss": 6.0027,
      "step": 1562
    },
    {
      "epoch": 0.06310052482842148,
      "grad_norm": 109.2168960571289,
      "learning_rate": 1.9942712547366915e-05,
      "loss": 6.7856,
      "step": 1563
    },
    {
      "epoch": 0.06314089624545821,
      "grad_norm": 105.3587875366211,
      "learning_rate": 1.994257269999431e-05,
      "loss": 5.6973,
      "step": 1564
    },
    {
      "epoch": 0.06318126766249495,
      "grad_norm": 620.985107421875,
      "learning_rate": 1.9942432682626937e-05,
      "loss": 6.388,
      "step": 1565
    },
    {
      "epoch": 0.0632216390795317,
      "grad_norm": 162.9539031982422,
      "learning_rate": 1.9942292495267176e-05,
      "loss": 6.0714,
      "step": 1566
    },
    {
      "epoch": 0.06326201049656843,
      "grad_norm": 130.16455078125,
      "learning_rate": 1.994215213791743e-05,
      "loss": 5.5595,
      "step": 1567
    },
    {
      "epoch": 0.06330238191360517,
      "grad_norm": 287.8119201660156,
      "learning_rate": 1.9942011610580105e-05,
      "loss": 5.0441,
      "step": 1568
    },
    {
      "epoch": 0.0633427533306419,
      "grad_norm": 143.54098510742188,
      "learning_rate": 1.9941870913257592e-05,
      "loss": 5.8653,
      "step": 1569
    },
    {
      "epoch": 0.06338312474767864,
      "grad_norm": 171.5598907470703,
      "learning_rate": 1.9941730045952304e-05,
      "loss": 5.3618,
      "step": 1570
    },
    {
      "epoch": 0.06342349616471538,
      "grad_norm": 122.2398452758789,
      "learning_rate": 1.994158900866665e-05,
      "loss": 5.8967,
      "step": 1571
    },
    {
      "epoch": 0.06346386758175213,
      "grad_norm": 95.57598876953125,
      "learning_rate": 1.994144780140304e-05,
      "loss": 6.8953,
      "step": 1572
    },
    {
      "epoch": 0.06350423899878886,
      "grad_norm": 274.1950988769531,
      "learning_rate": 1.9941306424163883e-05,
      "loss": 4.6126,
      "step": 1573
    },
    {
      "epoch": 0.0635446104158256,
      "grad_norm": 117.72879028320312,
      "learning_rate": 1.9941164876951604e-05,
      "loss": 4.5961,
      "step": 1574
    },
    {
      "epoch": 0.06358498183286233,
      "grad_norm": 93.66922760009766,
      "learning_rate": 1.994102315976862e-05,
      "loss": 6.0809,
      "step": 1575
    },
    {
      "epoch": 0.06362535324989907,
      "grad_norm": 128.4038848876953,
      "learning_rate": 1.9940881272617353e-05,
      "loss": 4.7212,
      "step": 1576
    },
    {
      "epoch": 0.0636657246669358,
      "grad_norm": 116.47521209716797,
      "learning_rate": 1.994073921550023e-05,
      "loss": 5.4719,
      "step": 1577
    },
    {
      "epoch": 0.06370609608397254,
      "grad_norm": 111.32181549072266,
      "learning_rate": 1.994059698841968e-05,
      "loss": 4.975,
      "step": 1578
    },
    {
      "epoch": 0.06374646750100929,
      "grad_norm": 123.74759674072266,
      "learning_rate": 1.994045459137813e-05,
      "loss": 4.9663,
      "step": 1579
    },
    {
      "epoch": 0.06378683891804603,
      "grad_norm": 262.45635986328125,
      "learning_rate": 1.9940312024378027e-05,
      "loss": 5.7832,
      "step": 1580
    },
    {
      "epoch": 0.06382721033508276,
      "grad_norm": 112.51979064941406,
      "learning_rate": 1.9940169287421797e-05,
      "loss": 5.2741,
      "step": 1581
    },
    {
      "epoch": 0.0638675817521195,
      "grad_norm": 126.9100112915039,
      "learning_rate": 1.9940026380511882e-05,
      "loss": 5.4087,
      "step": 1582
    },
    {
      "epoch": 0.06390795316915623,
      "grad_norm": 98.50367736816406,
      "learning_rate": 1.993988330365073e-05,
      "loss": 5.6945,
      "step": 1583
    },
    {
      "epoch": 0.06394832458619297,
      "grad_norm": 108.72956085205078,
      "learning_rate": 1.9939740056840784e-05,
      "loss": 5.7665,
      "step": 1584
    },
    {
      "epoch": 0.06398869600322972,
      "grad_norm": 185.1424102783203,
      "learning_rate": 1.9939596640084493e-05,
      "loss": 5.5855,
      "step": 1585
    },
    {
      "epoch": 0.06402906742026646,
      "grad_norm": 129.7752685546875,
      "learning_rate": 1.993945305338431e-05,
      "loss": 5.6199,
      "step": 1586
    },
    {
      "epoch": 0.06406943883730319,
      "grad_norm": 116.93241882324219,
      "learning_rate": 1.9939309296742694e-05,
      "loss": 6.4298,
      "step": 1587
    },
    {
      "epoch": 0.06410981025433993,
      "grad_norm": 123.58941650390625,
      "learning_rate": 1.9939165370162093e-05,
      "loss": 6.7464,
      "step": 1588
    },
    {
      "epoch": 0.06415018167137666,
      "grad_norm": 113.59819030761719,
      "learning_rate": 1.9939021273644982e-05,
      "loss": 4.9215,
      "step": 1589
    },
    {
      "epoch": 0.0641905530884134,
      "grad_norm": 93.8616943359375,
      "learning_rate": 1.993887700719381e-05,
      "loss": 5.9149,
      "step": 1590
    },
    {
      "epoch": 0.06423092450545013,
      "grad_norm": 104.9310531616211,
      "learning_rate": 1.993873257081105e-05,
      "loss": 6.6149,
      "step": 1591
    },
    {
      "epoch": 0.06427129592248688,
      "grad_norm": 114.50337219238281,
      "learning_rate": 1.9938587964499175e-05,
      "loss": 4.7342,
      "step": 1592
    },
    {
      "epoch": 0.06431166733952362,
      "grad_norm": 148.0487060546875,
      "learning_rate": 1.993844318826065e-05,
      "loss": 4.9652,
      "step": 1593
    },
    {
      "epoch": 0.06435203875656036,
      "grad_norm": 110.51678466796875,
      "learning_rate": 1.993829824209796e-05,
      "loss": 7.0405,
      "step": 1594
    },
    {
      "epoch": 0.06439241017359709,
      "grad_norm": 178.2396697998047,
      "learning_rate": 1.9938153126013572e-05,
      "loss": 6.0002,
      "step": 1595
    },
    {
      "epoch": 0.06443278159063383,
      "grad_norm": 108.68547821044922,
      "learning_rate": 1.9938007840009977e-05,
      "loss": 4.7155,
      "step": 1596
    },
    {
      "epoch": 0.06447315300767056,
      "grad_norm": 130.43533325195312,
      "learning_rate": 1.9937862384089652e-05,
      "loss": 4.0895,
      "step": 1597
    },
    {
      "epoch": 0.06451352442470731,
      "grad_norm": 120.21714782714844,
      "learning_rate": 1.9937716758255086e-05,
      "loss": 5.5646,
      "step": 1598
    },
    {
      "epoch": 0.06455389584174405,
      "grad_norm": 690.8196411132812,
      "learning_rate": 1.9937570962508774e-05,
      "loss": 5.0615,
      "step": 1599
    },
    {
      "epoch": 0.06459426725878079,
      "grad_norm": 124.42536163330078,
      "learning_rate": 1.99374249968532e-05,
      "loss": 5.177,
      "step": 1600
    },
    {
      "epoch": 0.06463463867581752,
      "grad_norm": 150.54791259765625,
      "learning_rate": 1.993727886129087e-05,
      "loss": 5.4925,
      "step": 1601
    },
    {
      "epoch": 0.06467501009285426,
      "grad_norm": 118.17633056640625,
      "learning_rate": 1.993713255582427e-05,
      "loss": 5.5994,
      "step": 1602
    },
    {
      "epoch": 0.06471538150989099,
      "grad_norm": 98.8293685913086,
      "learning_rate": 1.993698608045591e-05,
      "loss": 6.7415,
      "step": 1603
    },
    {
      "epoch": 0.06475575292692773,
      "grad_norm": 225.73492431640625,
      "learning_rate": 1.9936839435188295e-05,
      "loss": 6.0889,
      "step": 1604
    },
    {
      "epoch": 0.06479612434396448,
      "grad_norm": 182.0449981689453,
      "learning_rate": 1.993669262002393e-05,
      "loss": 5.8162,
      "step": 1605
    },
    {
      "epoch": 0.06483649576100121,
      "grad_norm": 288.2630310058594,
      "learning_rate": 1.9936545634965324e-05,
      "loss": 4.9858,
      "step": 1606
    },
    {
      "epoch": 0.06487686717803795,
      "grad_norm": 128.86181640625,
      "learning_rate": 1.993639848001499e-05,
      "loss": 4.7435,
      "step": 1607
    },
    {
      "epoch": 0.06491723859507469,
      "grad_norm": 257.7459716796875,
      "learning_rate": 1.9936251155175447e-05,
      "loss": 6.0764,
      "step": 1608
    },
    {
      "epoch": 0.06495761001211142,
      "grad_norm": 124.08128356933594,
      "learning_rate": 1.993610366044921e-05,
      "loss": 6.1908,
      "step": 1609
    },
    {
      "epoch": 0.06499798142914816,
      "grad_norm": 89.81301879882812,
      "learning_rate": 1.99359559958388e-05,
      "loss": 6.2674,
      "step": 1610
    },
    {
      "epoch": 0.06503835284618491,
      "grad_norm": 119.206298828125,
      "learning_rate": 1.993580816134675e-05,
      "loss": 5.3859,
      "step": 1611
    },
    {
      "epoch": 0.06507872426322164,
      "grad_norm": 111.39190673828125,
      "learning_rate": 1.993566015697558e-05,
      "loss": 6.0143,
      "step": 1612
    },
    {
      "epoch": 0.06511909568025838,
      "grad_norm": 96.46068572998047,
      "learning_rate": 1.9935511982727825e-05,
      "loss": 4.9698,
      "step": 1613
    },
    {
      "epoch": 0.06515946709729512,
      "grad_norm": 162.4373016357422,
      "learning_rate": 1.9935363638606014e-05,
      "loss": 2.9742,
      "step": 1614
    },
    {
      "epoch": 0.06519983851433185,
      "grad_norm": 292.0318603515625,
      "learning_rate": 1.9935215124612685e-05,
      "loss": 5.5513,
      "step": 1615
    },
    {
      "epoch": 0.06524020993136859,
      "grad_norm": 131.0552215576172,
      "learning_rate": 1.993506644075038e-05,
      "loss": 5.4934,
      "step": 1616
    },
    {
      "epoch": 0.06528058134840532,
      "grad_norm": 117.34226989746094,
      "learning_rate": 1.993491758702164e-05,
      "loss": 6.5002,
      "step": 1617
    },
    {
      "epoch": 0.06532095276544207,
      "grad_norm": 168.38980102539062,
      "learning_rate": 1.9934768563429007e-05,
      "loss": 4.881,
      "step": 1618
    },
    {
      "epoch": 0.06536132418247881,
      "grad_norm": 172.4776153564453,
      "learning_rate": 1.993461936997503e-05,
      "loss": 5.655,
      "step": 1619
    },
    {
      "epoch": 0.06540169559951554,
      "grad_norm": 179.71788024902344,
      "learning_rate": 1.993447000666226e-05,
      "loss": 5.4504,
      "step": 1620
    },
    {
      "epoch": 0.06544206701655228,
      "grad_norm": 294.0772705078125,
      "learning_rate": 1.9934320473493255e-05,
      "loss": 6.2202,
      "step": 1621
    },
    {
      "epoch": 0.06548243843358902,
      "grad_norm": 167.14337158203125,
      "learning_rate": 1.9934170770470565e-05,
      "loss": 7.7742,
      "step": 1622
    },
    {
      "epoch": 0.06552280985062575,
      "grad_norm": 170.9562225341797,
      "learning_rate": 1.9934020897596752e-05,
      "loss": 5.5477,
      "step": 1623
    },
    {
      "epoch": 0.0655631812676625,
      "grad_norm": 87.64669799804688,
      "learning_rate": 1.9933870854874382e-05,
      "loss": 6.5362,
      "step": 1624
    },
    {
      "epoch": 0.06560355268469924,
      "grad_norm": 186.20875549316406,
      "learning_rate": 1.9933720642306016e-05,
      "loss": 4.3786,
      "step": 1625
    },
    {
      "epoch": 0.06564392410173597,
      "grad_norm": 265.85443115234375,
      "learning_rate": 1.9933570259894224e-05,
      "loss": 6.1014,
      "step": 1626
    },
    {
      "epoch": 0.06568429551877271,
      "grad_norm": 115.99775695800781,
      "learning_rate": 1.993341970764158e-05,
      "loss": 4.977,
      "step": 1627
    },
    {
      "epoch": 0.06572466693580944,
      "grad_norm": 136.0398406982422,
      "learning_rate": 1.993326898555065e-05,
      "loss": 5.5099,
      "step": 1628
    },
    {
      "epoch": 0.06576503835284618,
      "grad_norm": 108.04027557373047,
      "learning_rate": 1.993311809362402e-05,
      "loss": 6.1684,
      "step": 1629
    },
    {
      "epoch": 0.06580540976988292,
      "grad_norm": 190.7596893310547,
      "learning_rate": 1.9932967031864265e-05,
      "loss": 6.074,
      "step": 1630
    },
    {
      "epoch": 0.06584578118691967,
      "grad_norm": 102.33934783935547,
      "learning_rate": 1.993281580027397e-05,
      "loss": 5.193,
      "step": 1631
    },
    {
      "epoch": 0.0658861526039564,
      "grad_norm": 199.87950134277344,
      "learning_rate": 1.993266439885572e-05,
      "loss": 5.9402,
      "step": 1632
    },
    {
      "epoch": 0.06592652402099314,
      "grad_norm": 138.40231323242188,
      "learning_rate": 1.99325128276121e-05,
      "loss": 6.4382,
      "step": 1633
    },
    {
      "epoch": 0.06596689543802987,
      "grad_norm": 179.5316162109375,
      "learning_rate": 1.993236108654571e-05,
      "loss": 5.2246,
      "step": 1634
    },
    {
      "epoch": 0.06600726685506661,
      "grad_norm": 132.97549438476562,
      "learning_rate": 1.9932209175659133e-05,
      "loss": 6.4954,
      "step": 1635
    },
    {
      "epoch": 0.06604763827210335,
      "grad_norm": 163.92930603027344,
      "learning_rate": 1.9932057094954976e-05,
      "loss": 5.6359,
      "step": 1636
    },
    {
      "epoch": 0.0660880096891401,
      "grad_norm": 115.22415161132812,
      "learning_rate": 1.9931904844435835e-05,
      "loss": 5.0497,
      "step": 1637
    },
    {
      "epoch": 0.06612838110617683,
      "grad_norm": 442.66571044921875,
      "learning_rate": 1.9931752424104314e-05,
      "loss": 4.9795,
      "step": 1638
    },
    {
      "epoch": 0.06616875252321357,
      "grad_norm": 107.10493469238281,
      "learning_rate": 1.9931599833963017e-05,
      "loss": 5.8073,
      "step": 1639
    },
    {
      "epoch": 0.0662091239402503,
      "grad_norm": 162.8252716064453,
      "learning_rate": 1.9931447074014558e-05,
      "loss": 6.2479,
      "step": 1640
    },
    {
      "epoch": 0.06624949535728704,
      "grad_norm": 141.9288787841797,
      "learning_rate": 1.9931294144261546e-05,
      "loss": 4.8422,
      "step": 1641
    },
    {
      "epoch": 0.06628986677432377,
      "grad_norm": 179.47872924804688,
      "learning_rate": 1.9931141044706594e-05,
      "loss": 5.8228,
      "step": 1642
    },
    {
      "epoch": 0.06633023819136051,
      "grad_norm": 129.0235595703125,
      "learning_rate": 1.9930987775352317e-05,
      "loss": 6.739,
      "step": 1643
    },
    {
      "epoch": 0.06637060960839726,
      "grad_norm": 166.37457275390625,
      "learning_rate": 1.9930834336201345e-05,
      "loss": 4.9566,
      "step": 1644
    },
    {
      "epoch": 0.066410981025434,
      "grad_norm": 142.76162719726562,
      "learning_rate": 1.9930680727256292e-05,
      "loss": 7.3313,
      "step": 1645
    },
    {
      "epoch": 0.06645135244247073,
      "grad_norm": 318.1599426269531,
      "learning_rate": 1.9930526948519793e-05,
      "loss": 4.4665,
      "step": 1646
    },
    {
      "epoch": 0.06649172385950747,
      "grad_norm": 180.39825439453125,
      "learning_rate": 1.9930372999994468e-05,
      "loss": 5.9017,
      "step": 1647
    },
    {
      "epoch": 0.0665320952765442,
      "grad_norm": 141.23345947265625,
      "learning_rate": 1.9930218881682958e-05,
      "loss": 5.7238,
      "step": 1648
    },
    {
      "epoch": 0.06657246669358094,
      "grad_norm": 154.92291259765625,
      "learning_rate": 1.993006459358789e-05,
      "loss": 6.4229,
      "step": 1649
    },
    {
      "epoch": 0.06661283811061769,
      "grad_norm": 245.23304748535156,
      "learning_rate": 1.9929910135711908e-05,
      "loss": 4.8666,
      "step": 1650
    },
    {
      "epoch": 0.06665320952765443,
      "grad_norm": 106.48798370361328,
      "learning_rate": 1.992975550805765e-05,
      "loss": 5.9378,
      "step": 1651
    },
    {
      "epoch": 0.06669358094469116,
      "grad_norm": 109.08892059326172,
      "learning_rate": 1.9929600710627764e-05,
      "loss": 6.9964,
      "step": 1652
    },
    {
      "epoch": 0.0667339523617279,
      "grad_norm": 132.08181762695312,
      "learning_rate": 1.992944574342489e-05,
      "loss": 7.0324,
      "step": 1653
    },
    {
      "epoch": 0.06677432377876463,
      "grad_norm": 133.6499786376953,
      "learning_rate": 1.992929060645168e-05,
      "loss": 4.6739,
      "step": 1654
    },
    {
      "epoch": 0.06681469519580137,
      "grad_norm": 110.5031967163086,
      "learning_rate": 1.9929135299710788e-05,
      "loss": 5.3596,
      "step": 1655
    },
    {
      "epoch": 0.0668550666128381,
      "grad_norm": 81.60302734375,
      "learning_rate": 1.9928979823204865e-05,
      "loss": 5.8009,
      "step": 1656
    },
    {
      "epoch": 0.06689543802987485,
      "grad_norm": 130.33181762695312,
      "learning_rate": 1.9928824176936577e-05,
      "loss": 4.1724,
      "step": 1657
    },
    {
      "epoch": 0.06693580944691159,
      "grad_norm": 279.08905029296875,
      "learning_rate": 1.992866836090858e-05,
      "loss": 6.5257,
      "step": 1658
    },
    {
      "epoch": 0.06697618086394833,
      "grad_norm": 10124.654296875,
      "learning_rate": 1.992851237512354e-05,
      "loss": 5.8434,
      "step": 1659
    },
    {
      "epoch": 0.06701655228098506,
      "grad_norm": 147.22401428222656,
      "learning_rate": 1.9928356219584123e-05,
      "loss": 5.5122,
      "step": 1660
    },
    {
      "epoch": 0.0670569236980218,
      "grad_norm": 140.52455139160156,
      "learning_rate": 1.9928199894293e-05,
      "loss": 6.87,
      "step": 1661
    },
    {
      "epoch": 0.06709729511505853,
      "grad_norm": 150.8752899169922,
      "learning_rate": 1.992804339925284e-05,
      "loss": 6.2206,
      "step": 1662
    },
    {
      "epoch": 0.06713766653209528,
      "grad_norm": 147.79031372070312,
      "learning_rate": 1.9927886734466325e-05,
      "loss": 6.7209,
      "step": 1663
    },
    {
      "epoch": 0.06717803794913202,
      "grad_norm": 134.18760681152344,
      "learning_rate": 1.9927729899936126e-05,
      "loss": 5.2112,
      "step": 1664
    },
    {
      "epoch": 0.06721840936616875,
      "grad_norm": 222.73956298828125,
      "learning_rate": 1.992757289566493e-05,
      "loss": 5.7339,
      "step": 1665
    },
    {
      "epoch": 0.06725878078320549,
      "grad_norm": 123.46160888671875,
      "learning_rate": 1.992741572165542e-05,
      "loss": 5.987,
      "step": 1666
    },
    {
      "epoch": 0.06729915220024223,
      "grad_norm": 127.80147552490234,
      "learning_rate": 1.9927258377910285e-05,
      "loss": 4.853,
      "step": 1667
    },
    {
      "epoch": 0.06733952361727896,
      "grad_norm": 120.11738586425781,
      "learning_rate": 1.9927100864432215e-05,
      "loss": 7.2305,
      "step": 1668
    },
    {
      "epoch": 0.0673798950343157,
      "grad_norm": 470.1526794433594,
      "learning_rate": 1.9926943181223897e-05,
      "loss": 5.4459,
      "step": 1669
    },
    {
      "epoch": 0.06742026645135245,
      "grad_norm": 116.67474365234375,
      "learning_rate": 1.9926785328288035e-05,
      "loss": 5.8898,
      "step": 1670
    },
    {
      "epoch": 0.06746063786838918,
      "grad_norm": 119.53043365478516,
      "learning_rate": 1.9926627305627323e-05,
      "loss": 5.619,
      "step": 1671
    },
    {
      "epoch": 0.06750100928542592,
      "grad_norm": 148.80960083007812,
      "learning_rate": 1.9926469113244465e-05,
      "loss": 5.4636,
      "step": 1672
    },
    {
      "epoch": 0.06754138070246266,
      "grad_norm": 95.48552703857422,
      "learning_rate": 1.9926310751142167e-05,
      "loss": 5.9765,
      "step": 1673
    },
    {
      "epoch": 0.06758175211949939,
      "grad_norm": 148.12510681152344,
      "learning_rate": 1.992615221932313e-05,
      "loss": 5.4027,
      "step": 1674
    },
    {
      "epoch": 0.06762212353653613,
      "grad_norm": 118.06719207763672,
      "learning_rate": 1.992599351779007e-05,
      "loss": 4.167,
      "step": 1675
    },
    {
      "epoch": 0.06766249495357288,
      "grad_norm": 140.48939514160156,
      "learning_rate": 1.9925834646545705e-05,
      "loss": 5.4689,
      "step": 1676
    },
    {
      "epoch": 0.06770286637060961,
      "grad_norm": 156.4361572265625,
      "learning_rate": 1.992567560559274e-05,
      "loss": 5.2874,
      "step": 1677
    },
    {
      "epoch": 0.06774323778764635,
      "grad_norm": 143.88449096679688,
      "learning_rate": 1.9925516394933907e-05,
      "loss": 4.2446,
      "step": 1678
    },
    {
      "epoch": 0.06778360920468308,
      "grad_norm": 144.2994842529297,
      "learning_rate": 1.9925357014571914e-05,
      "loss": 5.6942,
      "step": 1679
    },
    {
      "epoch": 0.06782398062171982,
      "grad_norm": 129.325439453125,
      "learning_rate": 1.99251974645095e-05,
      "loss": 4.6331,
      "step": 1680
    },
    {
      "epoch": 0.06786435203875656,
      "grad_norm": 317.47064208984375,
      "learning_rate": 1.9925037744749383e-05,
      "loss": 6.2017,
      "step": 1681
    },
    {
      "epoch": 0.06790472345579329,
      "grad_norm": 554.7642822265625,
      "learning_rate": 1.9924877855294297e-05,
      "loss": 5.5785,
      "step": 1682
    },
    {
      "epoch": 0.06794509487283004,
      "grad_norm": 127.0877685546875,
      "learning_rate": 1.9924717796146974e-05,
      "loss": 6.5969,
      "step": 1683
    },
    {
      "epoch": 0.06798546628986678,
      "grad_norm": 134.00277709960938,
      "learning_rate": 1.9924557567310157e-05,
      "loss": 5.6122,
      "step": 1684
    },
    {
      "epoch": 0.06802583770690351,
      "grad_norm": 132.3541259765625,
      "learning_rate": 1.992439716878658e-05,
      "loss": 6.7327,
      "step": 1685
    },
    {
      "epoch": 0.06806620912394025,
      "grad_norm": 152.60971069335938,
      "learning_rate": 1.9924236600578984e-05,
      "loss": 6.4191,
      "step": 1686
    },
    {
      "epoch": 0.06810658054097699,
      "grad_norm": 122.47472381591797,
      "learning_rate": 1.9924075862690117e-05,
      "loss": 5.9375,
      "step": 1687
    },
    {
      "epoch": 0.06814695195801372,
      "grad_norm": 115.3364486694336,
      "learning_rate": 1.9923914955122733e-05,
      "loss": 4.3511,
      "step": 1688
    },
    {
      "epoch": 0.06818732337505047,
      "grad_norm": 146.11526489257812,
      "learning_rate": 1.9923753877879572e-05,
      "loss": 5.2265,
      "step": 1689
    },
    {
      "epoch": 0.0682276947920872,
      "grad_norm": 92.83424377441406,
      "learning_rate": 1.9923592630963392e-05,
      "loss": 4.6623,
      "step": 1690
    },
    {
      "epoch": 0.06826806620912394,
      "grad_norm": 110.59848022460938,
      "learning_rate": 1.9923431214376956e-05,
      "loss": 4.9655,
      "step": 1691
    },
    {
      "epoch": 0.06830843762616068,
      "grad_norm": 177.49314880371094,
      "learning_rate": 1.9923269628123013e-05,
      "loss": 6.3439,
      "step": 1692
    },
    {
      "epoch": 0.06834880904319741,
      "grad_norm": 778.8950805664062,
      "learning_rate": 1.9923107872204335e-05,
      "loss": 5.8106,
      "step": 1693
    },
    {
      "epoch": 0.06838918046023415,
      "grad_norm": 108.67484283447266,
      "learning_rate": 1.9922945946623685e-05,
      "loss": 6.8603,
      "step": 1694
    },
    {
      "epoch": 0.06842955187727089,
      "grad_norm": 114.06116485595703,
      "learning_rate": 1.9922783851383832e-05,
      "loss": 6.3752,
      "step": 1695
    },
    {
      "epoch": 0.06846992329430764,
      "grad_norm": 122.94540405273438,
      "learning_rate": 1.9922621586487546e-05,
      "loss": 5.7607,
      "step": 1696
    },
    {
      "epoch": 0.06851029471134437,
      "grad_norm": 172.1855010986328,
      "learning_rate": 1.99224591519376e-05,
      "loss": 5.5034,
      "step": 1697
    },
    {
      "epoch": 0.06855066612838111,
      "grad_norm": 119.59555053710938,
      "learning_rate": 1.9922296547736776e-05,
      "loss": 5.5301,
      "step": 1698
    },
    {
      "epoch": 0.06859103754541784,
      "grad_norm": 158.2825469970703,
      "learning_rate": 1.992213377388785e-05,
      "loss": 4.9623,
      "step": 1699
    },
    {
      "epoch": 0.06863140896245458,
      "grad_norm": 97.82806396484375,
      "learning_rate": 1.9921970830393606e-05,
      "loss": 5.5475,
      "step": 1700
    },
    {
      "epoch": 0.06867178037949132,
      "grad_norm": 383.0748596191406,
      "learning_rate": 1.9921807717256833e-05,
      "loss": 5.5699,
      "step": 1701
    },
    {
      "epoch": 0.06871215179652806,
      "grad_norm": 85.39448547363281,
      "learning_rate": 1.9921644434480312e-05,
      "loss": 6.3961,
      "step": 1702
    },
    {
      "epoch": 0.0687525232135648,
      "grad_norm": 104.17704010009766,
      "learning_rate": 1.9921480982066842e-05,
      "loss": 7.3318,
      "step": 1703
    },
    {
      "epoch": 0.06879289463060154,
      "grad_norm": 146.30645751953125,
      "learning_rate": 1.9921317360019216e-05,
      "loss": 7.1383,
      "step": 1704
    },
    {
      "epoch": 0.06883326604763827,
      "grad_norm": 163.63003540039062,
      "learning_rate": 1.9921153568340232e-05,
      "loss": 5.6799,
      "step": 1705
    },
    {
      "epoch": 0.06887363746467501,
      "grad_norm": 119.18814849853516,
      "learning_rate": 1.992098960703269e-05,
      "loss": 5.282,
      "step": 1706
    },
    {
      "epoch": 0.06891400888171174,
      "grad_norm": 153.18756103515625,
      "learning_rate": 1.992082547609939e-05,
      "loss": 7.0382,
      "step": 1707
    },
    {
      "epoch": 0.06895438029874848,
      "grad_norm": 161.85293579101562,
      "learning_rate": 1.9920661175543144e-05,
      "loss": 7.0459,
      "step": 1708
    },
    {
      "epoch": 0.06899475171578523,
      "grad_norm": 96.07947540283203,
      "learning_rate": 1.9920496705366754e-05,
      "loss": 5.5592,
      "step": 1709
    },
    {
      "epoch": 0.06903512313282197,
      "grad_norm": 190.99343872070312,
      "learning_rate": 1.992033206557304e-05,
      "loss": 5.9466,
      "step": 1710
    },
    {
      "epoch": 0.0690754945498587,
      "grad_norm": 136.62149047851562,
      "learning_rate": 1.9920167256164812e-05,
      "loss": 5.9766,
      "step": 1711
    },
    {
      "epoch": 0.06911586596689544,
      "grad_norm": 132.93321228027344,
      "learning_rate": 1.9920002277144892e-05,
      "loss": 4.9107,
      "step": 1712
    },
    {
      "epoch": 0.06915623738393217,
      "grad_norm": 162.0550994873047,
      "learning_rate": 1.9919837128516094e-05,
      "loss": 7.6266,
      "step": 1713
    },
    {
      "epoch": 0.06919660880096891,
      "grad_norm": 116.30625915527344,
      "learning_rate": 1.991967181028125e-05,
      "loss": 5.6091,
      "step": 1714
    },
    {
      "epoch": 0.06923698021800566,
      "grad_norm": 96.16675567626953,
      "learning_rate": 1.9919506322443177e-05,
      "loss": 6.737,
      "step": 1715
    },
    {
      "epoch": 0.0692773516350424,
      "grad_norm": 98.60612487792969,
      "learning_rate": 1.991934066500471e-05,
      "loss": 5.543,
      "step": 1716
    },
    {
      "epoch": 0.06931772305207913,
      "grad_norm": 156.21401977539062,
      "learning_rate": 1.9919174837968685e-05,
      "loss": 6.2473,
      "step": 1717
    },
    {
      "epoch": 0.06935809446911587,
      "grad_norm": 99.19168090820312,
      "learning_rate": 1.9919008841337934e-05,
      "loss": 6.1339,
      "step": 1718
    },
    {
      "epoch": 0.0693984658861526,
      "grad_norm": 113.13419342041016,
      "learning_rate": 1.991884267511529e-05,
      "loss": 5.8257,
      "step": 1719
    },
    {
      "epoch": 0.06943883730318934,
      "grad_norm": 114.2751235961914,
      "learning_rate": 1.99186763393036e-05,
      "loss": 5.4808,
      "step": 1720
    },
    {
      "epoch": 0.06947920872022607,
      "grad_norm": 117.11168670654297,
      "learning_rate": 1.9918509833905702e-05,
      "loss": 7.6185,
      "step": 1721
    },
    {
      "epoch": 0.06951958013726282,
      "grad_norm": 125.74390411376953,
      "learning_rate": 1.9918343158924455e-05,
      "loss": 5.4151,
      "step": 1722
    },
    {
      "epoch": 0.06955995155429956,
      "grad_norm": 92.68668365478516,
      "learning_rate": 1.9918176314362698e-05,
      "loss": 5.1154,
      "step": 1723
    },
    {
      "epoch": 0.0696003229713363,
      "grad_norm": 139.58291625976562,
      "learning_rate": 1.9918009300223285e-05,
      "loss": 4.7775,
      "step": 1724
    },
    {
      "epoch": 0.06964069438837303,
      "grad_norm": 97.1887435913086,
      "learning_rate": 1.9917842116509076e-05,
      "loss": 6.6637,
      "step": 1725
    },
    {
      "epoch": 0.06968106580540977,
      "grad_norm": 6192.01171875,
      "learning_rate": 1.9917674763222923e-05,
      "loss": 6.8527,
      "step": 1726
    },
    {
      "epoch": 0.0697214372224465,
      "grad_norm": 98.63325500488281,
      "learning_rate": 1.9917507240367696e-05,
      "loss": 4.9299,
      "step": 1727
    },
    {
      "epoch": 0.06976180863948325,
      "grad_norm": 98.72211456298828,
      "learning_rate": 1.9917339547946247e-05,
      "loss": 5.8346,
      "step": 1728
    },
    {
      "epoch": 0.06980218005651999,
      "grad_norm": 110.61398315429688,
      "learning_rate": 1.9917171685961457e-05,
      "loss": 5.8575,
      "step": 1729
    },
    {
      "epoch": 0.06984255147355672,
      "grad_norm": 139.39056396484375,
      "learning_rate": 1.991700365441619e-05,
      "loss": 5.7262,
      "step": 1730
    },
    {
      "epoch": 0.06988292289059346,
      "grad_norm": 161.8549346923828,
      "learning_rate": 1.9916835453313312e-05,
      "loss": 5.1792,
      "step": 1731
    },
    {
      "epoch": 0.0699232943076302,
      "grad_norm": 95.90702056884766,
      "learning_rate": 1.9916667082655706e-05,
      "loss": 5.8488,
      "step": 1732
    },
    {
      "epoch": 0.06996366572466693,
      "grad_norm": 90.40072631835938,
      "learning_rate": 1.9916498542446255e-05,
      "loss": 6.7057,
      "step": 1733
    },
    {
      "epoch": 0.07000403714170367,
      "grad_norm": 125.92350769042969,
      "learning_rate": 1.991632983268783e-05,
      "loss": 4.6561,
      "step": 1734
    },
    {
      "epoch": 0.07004440855874042,
      "grad_norm": 164.61129760742188,
      "learning_rate": 1.991616095338333e-05,
      "loss": 6.6919,
      "step": 1735
    },
    {
      "epoch": 0.07008477997577715,
      "grad_norm": 95.05935668945312,
      "learning_rate": 1.9915991904535625e-05,
      "loss": 6.8412,
      "step": 1736
    },
    {
      "epoch": 0.07012515139281389,
      "grad_norm": 102.64321899414062,
      "learning_rate": 1.9915822686147618e-05,
      "loss": 5.3519,
      "step": 1737
    },
    {
      "epoch": 0.07016552280985063,
      "grad_norm": 101.0239486694336,
      "learning_rate": 1.9915653298222196e-05,
      "loss": 5.7489,
      "step": 1738
    },
    {
      "epoch": 0.07020589422688736,
      "grad_norm": 143.63768005371094,
      "learning_rate": 1.9915483740762262e-05,
      "loss": 7.6912,
      "step": 1739
    },
    {
      "epoch": 0.0702462656439241,
      "grad_norm": 92.06588745117188,
      "learning_rate": 1.9915314013770703e-05,
      "loss": 6.1418,
      "step": 1740
    },
    {
      "epoch": 0.07028663706096085,
      "grad_norm": 92.16914367675781,
      "learning_rate": 1.991514411725043e-05,
      "loss": 5.3705,
      "step": 1741
    },
    {
      "epoch": 0.07032700847799758,
      "grad_norm": 108.20307159423828,
      "learning_rate": 1.9914974051204348e-05,
      "loss": 6.2212,
      "step": 1742
    },
    {
      "epoch": 0.07036737989503432,
      "grad_norm": 123.94193267822266,
      "learning_rate": 1.9914803815635366e-05,
      "loss": 4.2645,
      "step": 1743
    },
    {
      "epoch": 0.07040775131207105,
      "grad_norm": 136.49266052246094,
      "learning_rate": 1.9914633410546388e-05,
      "loss": 5.9133,
      "step": 1744
    },
    {
      "epoch": 0.07044812272910779,
      "grad_norm": 114.97510528564453,
      "learning_rate": 1.9914462835940332e-05,
      "loss": 4.8438,
      "step": 1745
    },
    {
      "epoch": 0.07048849414614453,
      "grad_norm": 128.865478515625,
      "learning_rate": 1.991429209182011e-05,
      "loss": 5.0687,
      "step": 1746
    },
    {
      "epoch": 0.07052886556318126,
      "grad_norm": 117.34593200683594,
      "learning_rate": 1.991412117818865e-05,
      "loss": 6.2974,
      "step": 1747
    },
    {
      "epoch": 0.07056923698021801,
      "grad_norm": 135.91990661621094,
      "learning_rate": 1.9913950095048866e-05,
      "loss": 5.6746,
      "step": 1748
    },
    {
      "epoch": 0.07060960839725475,
      "grad_norm": 94.5130844116211,
      "learning_rate": 1.9913778842403687e-05,
      "loss": 6.6926,
      "step": 1749
    },
    {
      "epoch": 0.07064997981429148,
      "grad_norm": 104.0962905883789,
      "learning_rate": 1.991360742025604e-05,
      "loss": 6.2549,
      "step": 1750
    },
    {
      "epoch": 0.07069035123132822,
      "grad_norm": 222.51287841796875,
      "learning_rate": 1.9913435828608856e-05,
      "loss": 5.1365,
      "step": 1751
    },
    {
      "epoch": 0.07073072264836495,
      "grad_norm": 145.4353790283203,
      "learning_rate": 1.991326406746507e-05,
      "loss": 4.9843,
      "step": 1752
    },
    {
      "epoch": 0.07077109406540169,
      "grad_norm": 167.77215576171875,
      "learning_rate": 1.9913092136827618e-05,
      "loss": 5.2836,
      "step": 1753
    },
    {
      "epoch": 0.07081146548243844,
      "grad_norm": 95.60668182373047,
      "learning_rate": 1.9912920036699436e-05,
      "loss": 5.481,
      "step": 1754
    },
    {
      "epoch": 0.07085183689947518,
      "grad_norm": 88.59827423095703,
      "learning_rate": 1.991274776708347e-05,
      "loss": 5.3011,
      "step": 1755
    },
    {
      "epoch": 0.07089220831651191,
      "grad_norm": 131.02020263671875,
      "learning_rate": 1.9912575327982665e-05,
      "loss": 6.0966,
      "step": 1756
    },
    {
      "epoch": 0.07093257973354865,
      "grad_norm": 125.69205474853516,
      "learning_rate": 1.991240271939997e-05,
      "loss": 6.454,
      "step": 1757
    },
    {
      "epoch": 0.07097295115058538,
      "grad_norm": 103.26402282714844,
      "learning_rate": 1.991222994133834e-05,
      "loss": 4.4783,
      "step": 1758
    },
    {
      "epoch": 0.07101332256762212,
      "grad_norm": 80.71549224853516,
      "learning_rate": 1.9912056993800723e-05,
      "loss": 6.1187,
      "step": 1759
    },
    {
      "epoch": 0.07105369398465886,
      "grad_norm": 129.2772674560547,
      "learning_rate": 1.9911883876790074e-05,
      "loss": 6.2572,
      "step": 1760
    },
    {
      "epoch": 0.0710940654016956,
      "grad_norm": 100.05765533447266,
      "learning_rate": 1.991171059030936e-05,
      "loss": 5.6628,
      "step": 1761
    },
    {
      "epoch": 0.07113443681873234,
      "grad_norm": 95.963134765625,
      "learning_rate": 1.9911537134361535e-05,
      "loss": 5.5915,
      "step": 1762
    },
    {
      "epoch": 0.07117480823576908,
      "grad_norm": 128.97193908691406,
      "learning_rate": 1.9911363508949578e-05,
      "loss": 5.2406,
      "step": 1763
    },
    {
      "epoch": 0.07121517965280581,
      "grad_norm": 115.15621948242188,
      "learning_rate": 1.9911189714076445e-05,
      "loss": 5.6652,
      "step": 1764
    },
    {
      "epoch": 0.07125555106984255,
      "grad_norm": 141.19703674316406,
      "learning_rate": 1.9911015749745112e-05,
      "loss": 4.0117,
      "step": 1765
    },
    {
      "epoch": 0.07129592248687928,
      "grad_norm": 537.0606689453125,
      "learning_rate": 1.9910841615958552e-05,
      "loss": 6.6402,
      "step": 1766
    },
    {
      "epoch": 0.07133629390391603,
      "grad_norm": 131.88836669921875,
      "learning_rate": 1.9910667312719748e-05,
      "loss": 5.1102,
      "step": 1767
    },
    {
      "epoch": 0.07137666532095277,
      "grad_norm": 133.92144775390625,
      "learning_rate": 1.9910492840031678e-05,
      "loss": 5.4005,
      "step": 1768
    },
    {
      "epoch": 0.0714170367379895,
      "grad_norm": 105.9146728515625,
      "learning_rate": 1.991031819789732e-05,
      "loss": 6.7832,
      "step": 1769
    },
    {
      "epoch": 0.07145740815502624,
      "grad_norm": 162.76087951660156,
      "learning_rate": 1.991014338631966e-05,
      "loss": 5.9911,
      "step": 1770
    },
    {
      "epoch": 0.07149777957206298,
      "grad_norm": 123.79119110107422,
      "learning_rate": 1.990996840530169e-05,
      "loss": 6.0916,
      "step": 1771
    },
    {
      "epoch": 0.07153815098909971,
      "grad_norm": 124.1023178100586,
      "learning_rate": 1.9909793254846404e-05,
      "loss": 5.0601,
      "step": 1772
    },
    {
      "epoch": 0.07157852240613645,
      "grad_norm": 178.4263458251953,
      "learning_rate": 1.9909617934956797e-05,
      "loss": 6.824,
      "step": 1773
    },
    {
      "epoch": 0.0716188938231732,
      "grad_norm": 120.62239074707031,
      "learning_rate": 1.9909442445635863e-05,
      "loss": 6.0426,
      "step": 1774
    },
    {
      "epoch": 0.07165926524020994,
      "grad_norm": 108.2620849609375,
      "learning_rate": 1.99092667868866e-05,
      "loss": 5.8122,
      "step": 1775
    },
    {
      "epoch": 0.07169963665724667,
      "grad_norm": 145.03871154785156,
      "learning_rate": 1.9909090958712017e-05,
      "loss": 6.0397,
      "step": 1776
    },
    {
      "epoch": 0.0717400080742834,
      "grad_norm": 130.08230590820312,
      "learning_rate": 1.990891496111512e-05,
      "loss": 4.9527,
      "step": 1777
    },
    {
      "epoch": 0.07178037949132014,
      "grad_norm": 439.7630615234375,
      "learning_rate": 1.990873879409891e-05,
      "loss": 6.5179,
      "step": 1778
    },
    {
      "epoch": 0.07182075090835688,
      "grad_norm": 104.50048828125,
      "learning_rate": 1.990856245766641e-05,
      "loss": 6.3818,
      "step": 1779
    },
    {
      "epoch": 0.07186112232539363,
      "grad_norm": 145.87435913085938,
      "learning_rate": 1.990838595182063e-05,
      "loss": 6.0966,
      "step": 1780
    },
    {
      "epoch": 0.07190149374243036,
      "grad_norm": 127.4691162109375,
      "learning_rate": 1.9908209276564587e-05,
      "loss": 5.0409,
      "step": 1781
    },
    {
      "epoch": 0.0719418651594671,
      "grad_norm": 117.0577621459961,
      "learning_rate": 1.99080324319013e-05,
      "loss": 4.2227,
      "step": 1782
    },
    {
      "epoch": 0.07198223657650384,
      "grad_norm": 131.85328674316406,
      "learning_rate": 1.99078554178338e-05,
      "loss": 4.4521,
      "step": 1783
    },
    {
      "epoch": 0.07202260799354057,
      "grad_norm": 102.79983520507812,
      "learning_rate": 1.9907678234365108e-05,
      "loss": 4.8153,
      "step": 1784
    },
    {
      "epoch": 0.07206297941057731,
      "grad_norm": 86.38661193847656,
      "learning_rate": 1.9907500881498252e-05,
      "loss": 5.3559,
      "step": 1785
    },
    {
      "epoch": 0.07210335082761404,
      "grad_norm": 131.2575225830078,
      "learning_rate": 1.990732335923627e-05,
      "loss": 7.1217,
      "step": 1786
    },
    {
      "epoch": 0.0721437222446508,
      "grad_norm": 162.85137939453125,
      "learning_rate": 1.990714566758219e-05,
      "loss": 6.3964,
      "step": 1787
    },
    {
      "epoch": 0.07218409366168753,
      "grad_norm": 152.83566284179688,
      "learning_rate": 1.9906967806539054e-05,
      "loss": 4.9,
      "step": 1788
    },
    {
      "epoch": 0.07222446507872426,
      "grad_norm": 141.93333435058594,
      "learning_rate": 1.9906789776109903e-05,
      "loss": 5.5896,
      "step": 1789
    },
    {
      "epoch": 0.072264836495761,
      "grad_norm": 166.11767578125,
      "learning_rate": 1.9906611576297785e-05,
      "loss": 4.8633,
      "step": 1790
    },
    {
      "epoch": 0.07230520791279774,
      "grad_norm": 127.70752716064453,
      "learning_rate": 1.9906433207105742e-05,
      "loss": 4.6873,
      "step": 1791
    },
    {
      "epoch": 0.07234557932983447,
      "grad_norm": 111.69345092773438,
      "learning_rate": 1.990625466853682e-05,
      "loss": 6.0997,
      "step": 1792
    },
    {
      "epoch": 0.07238595074687122,
      "grad_norm": 130.03439331054688,
      "learning_rate": 1.990607596059408e-05,
      "loss": 4.461,
      "step": 1793
    },
    {
      "epoch": 0.07242632216390796,
      "grad_norm": 298.9710388183594,
      "learning_rate": 1.990589708328057e-05,
      "loss": 5.0888,
      "step": 1794
    },
    {
      "epoch": 0.0724666935809447,
      "grad_norm": 140.7045135498047,
      "learning_rate": 1.9905718036599356e-05,
      "loss": 6.8343,
      "step": 1795
    },
    {
      "epoch": 0.07250706499798143,
      "grad_norm": 135.176513671875,
      "learning_rate": 1.9905538820553493e-05,
      "loss": 5.4305,
      "step": 1796
    },
    {
      "epoch": 0.07254743641501817,
      "grad_norm": 202.6995391845703,
      "learning_rate": 1.9905359435146045e-05,
      "loss": 6.2684,
      "step": 1797
    },
    {
      "epoch": 0.0725878078320549,
      "grad_norm": 140.6923370361328,
      "learning_rate": 1.9905179880380082e-05,
      "loss": 4.8413,
      "step": 1798
    },
    {
      "epoch": 0.07262817924909164,
      "grad_norm": 144.4919891357422,
      "learning_rate": 1.9905000156258676e-05,
      "loss": 4.2866,
      "step": 1799
    },
    {
      "epoch": 0.07266855066612839,
      "grad_norm": 193.60757446289062,
      "learning_rate": 1.9904820262784895e-05,
      "loss": 5.3485,
      "step": 1800
    },
    {
      "epoch": 0.07270892208316512,
      "grad_norm": 139.22561645507812,
      "learning_rate": 1.990464019996182e-05,
      "loss": 6.6256,
      "step": 1801
    },
    {
      "epoch": 0.07274929350020186,
      "grad_norm": 87.09326171875,
      "learning_rate": 1.9904459967792524e-05,
      "loss": 5.3175,
      "step": 1802
    },
    {
      "epoch": 0.0727896649172386,
      "grad_norm": 106.77767944335938,
      "learning_rate": 1.990427956628009e-05,
      "loss": 7.0266,
      "step": 1803
    },
    {
      "epoch": 0.07283003633427533,
      "grad_norm": 129.63929748535156,
      "learning_rate": 1.9904098995427606e-05,
      "loss": 4.69,
      "step": 1804
    },
    {
      "epoch": 0.07287040775131207,
      "grad_norm": 116.97301483154297,
      "learning_rate": 1.9903918255238156e-05,
      "loss": 5.3029,
      "step": 1805
    },
    {
      "epoch": 0.0729107791683488,
      "grad_norm": 250.4459686279297,
      "learning_rate": 1.990373734571483e-05,
      "loss": 7.0586,
      "step": 1806
    },
    {
      "epoch": 0.07295115058538555,
      "grad_norm": 153.44752502441406,
      "learning_rate": 1.9903556266860722e-05,
      "loss": 6.03,
      "step": 1807
    },
    {
      "epoch": 0.07299152200242229,
      "grad_norm": 169.89913940429688,
      "learning_rate": 1.990337501867893e-05,
      "loss": 5.2407,
      "step": 1808
    },
    {
      "epoch": 0.07303189341945902,
      "grad_norm": 776.032958984375,
      "learning_rate": 1.990319360117255e-05,
      "loss": 5.3935,
      "step": 1809
    },
    {
      "epoch": 0.07307226483649576,
      "grad_norm": 171.4215545654297,
      "learning_rate": 1.9903012014344688e-05,
      "loss": 6.675,
      "step": 1810
    },
    {
      "epoch": 0.0731126362535325,
      "grad_norm": 200.2425537109375,
      "learning_rate": 1.9902830258198445e-05,
      "loss": 5.9913,
      "step": 1811
    },
    {
      "epoch": 0.07315300767056923,
      "grad_norm": 168.92654418945312,
      "learning_rate": 1.9902648332736927e-05,
      "loss": 6.4698,
      "step": 1812
    },
    {
      "epoch": 0.07319337908760598,
      "grad_norm": 136.87167358398438,
      "learning_rate": 1.9902466237963246e-05,
      "loss": 5.0343,
      "step": 1813
    },
    {
      "epoch": 0.07323375050464272,
      "grad_norm": 105.36048889160156,
      "learning_rate": 1.990228397388052e-05,
      "loss": 5.0918,
      "step": 1814
    },
    {
      "epoch": 0.07327412192167945,
      "grad_norm": 157.10731506347656,
      "learning_rate": 1.990210154049186e-05,
      "loss": 5.9781,
      "step": 1815
    },
    {
      "epoch": 0.07331449333871619,
      "grad_norm": 107.04386138916016,
      "learning_rate": 1.9901918937800384e-05,
      "loss": 2.9148,
      "step": 1816
    },
    {
      "epoch": 0.07335486475575292,
      "grad_norm": 115.8561019897461,
      "learning_rate": 1.990173616580922e-05,
      "loss": 6.017,
      "step": 1817
    },
    {
      "epoch": 0.07339523617278966,
      "grad_norm": 151.2177276611328,
      "learning_rate": 1.990155322452149e-05,
      "loss": 7.2344,
      "step": 1818
    },
    {
      "epoch": 0.0734356075898264,
      "grad_norm": 101.41718292236328,
      "learning_rate": 1.9901370113940318e-05,
      "loss": 5.1604,
      "step": 1819
    },
    {
      "epoch": 0.07347597900686315,
      "grad_norm": 104.00305938720703,
      "learning_rate": 1.990118683406884e-05,
      "loss": 6.4528,
      "step": 1820
    },
    {
      "epoch": 0.07351635042389988,
      "grad_norm": 97.99864959716797,
      "learning_rate": 1.9901003384910192e-05,
      "loss": 5.7567,
      "step": 1821
    },
    {
      "epoch": 0.07355672184093662,
      "grad_norm": 114.15586853027344,
      "learning_rate": 1.9900819766467503e-05,
      "loss": 6.4243,
      "step": 1822
    },
    {
      "epoch": 0.07359709325797335,
      "grad_norm": 86.38786315917969,
      "learning_rate": 1.9900635978743917e-05,
      "loss": 5.1107,
      "step": 1823
    },
    {
      "epoch": 0.07363746467501009,
      "grad_norm": 160.09486389160156,
      "learning_rate": 1.9900452021742572e-05,
      "loss": 6.2676,
      "step": 1824
    },
    {
      "epoch": 0.07367783609204683,
      "grad_norm": 210.24557495117188,
      "learning_rate": 1.9900267895466623e-05,
      "loss": 4.7993,
      "step": 1825
    },
    {
      "epoch": 0.07371820750908357,
      "grad_norm": 221.0330047607422,
      "learning_rate": 1.990008359991921e-05,
      "loss": 5.046,
      "step": 1826
    },
    {
      "epoch": 0.07375857892612031,
      "grad_norm": 111.18528747558594,
      "learning_rate": 1.9899899135103483e-05,
      "loss": 5.8427,
      "step": 1827
    },
    {
      "epoch": 0.07379895034315705,
      "grad_norm": 327.2301940917969,
      "learning_rate": 1.98997145010226e-05,
      "loss": 5.8749,
      "step": 1828
    },
    {
      "epoch": 0.07383932176019378,
      "grad_norm": 153.92625427246094,
      "learning_rate": 1.9899529697679715e-05,
      "loss": 4.8932,
      "step": 1829
    },
    {
      "epoch": 0.07387969317723052,
      "grad_norm": 106.25981903076172,
      "learning_rate": 1.9899344725077992e-05,
      "loss": 5.8456,
      "step": 1830
    },
    {
      "epoch": 0.07392006459426725,
      "grad_norm": 386.074462890625,
      "learning_rate": 1.989915958322059e-05,
      "loss": 6.0178,
      "step": 1831
    },
    {
      "epoch": 0.07396043601130399,
      "grad_norm": 118.2993392944336,
      "learning_rate": 1.9898974272110675e-05,
      "loss": 5.7643,
      "step": 1832
    },
    {
      "epoch": 0.07400080742834074,
      "grad_norm": 119.60366821289062,
      "learning_rate": 1.9898788791751417e-05,
      "loss": 6.3869,
      "step": 1833
    },
    {
      "epoch": 0.07404117884537748,
      "grad_norm": 93.53543090820312,
      "learning_rate": 1.9898603142145984e-05,
      "loss": 4.5394,
      "step": 1834
    },
    {
      "epoch": 0.07408155026241421,
      "grad_norm": 305.4543151855469,
      "learning_rate": 1.9898417323297554e-05,
      "loss": 4.8148,
      "step": 1835
    },
    {
      "epoch": 0.07412192167945095,
      "grad_norm": 139.89517211914062,
      "learning_rate": 1.9898231335209298e-05,
      "loss": 4.2643,
      "step": 1836
    },
    {
      "epoch": 0.07416229309648768,
      "grad_norm": 109.9010238647461,
      "learning_rate": 1.9898045177884405e-05,
      "loss": 5.7873,
      "step": 1837
    },
    {
      "epoch": 0.07420266451352442,
      "grad_norm": 101.31986236572266,
      "learning_rate": 1.9897858851326052e-05,
      "loss": 5.7345,
      "step": 1838
    },
    {
      "epoch": 0.07424303593056117,
      "grad_norm": 169.4073944091797,
      "learning_rate": 1.9897672355537425e-05,
      "loss": 5.9274,
      "step": 1839
    },
    {
      "epoch": 0.0742834073475979,
      "grad_norm": 103.42391204833984,
      "learning_rate": 1.9897485690521716e-05,
      "loss": 4.8322,
      "step": 1840
    },
    {
      "epoch": 0.07432377876463464,
      "grad_norm": 94.99114990234375,
      "learning_rate": 1.989729885628211e-05,
      "loss": 4.4697,
      "step": 1841
    },
    {
      "epoch": 0.07436415018167138,
      "grad_norm": 110.43385314941406,
      "learning_rate": 1.9897111852821807e-05,
      "loss": 5.4684,
      "step": 1842
    },
    {
      "epoch": 0.07440452159870811,
      "grad_norm": 89.65741729736328,
      "learning_rate": 1.9896924680144002e-05,
      "loss": 6.6112,
      "step": 1843
    },
    {
      "epoch": 0.07444489301574485,
      "grad_norm": 103.82837677001953,
      "learning_rate": 1.9896737338251894e-05,
      "loss": 6.7801,
      "step": 1844
    },
    {
      "epoch": 0.07448526443278158,
      "grad_norm": 105.6393814086914,
      "learning_rate": 1.9896549827148692e-05,
      "loss": 6.5639,
      "step": 1845
    },
    {
      "epoch": 0.07452563584981833,
      "grad_norm": 274.3399353027344,
      "learning_rate": 1.9896362146837597e-05,
      "loss": 6.0866,
      "step": 1846
    },
    {
      "epoch": 0.07456600726685507,
      "grad_norm": 230.8987579345703,
      "learning_rate": 1.989617429732182e-05,
      "loss": 5.9735,
      "step": 1847
    },
    {
      "epoch": 0.0746063786838918,
      "grad_norm": 126.4937515258789,
      "learning_rate": 1.989598627860457e-05,
      "loss": 5.1935,
      "step": 1848
    },
    {
      "epoch": 0.07464675010092854,
      "grad_norm": 116.60391998291016,
      "learning_rate": 1.989579809068906e-05,
      "loss": 5.8787,
      "step": 1849
    },
    {
      "epoch": 0.07468712151796528,
      "grad_norm": 221.43113708496094,
      "learning_rate": 1.989560973357852e-05,
      "loss": 6.3544,
      "step": 1850
    },
    {
      "epoch": 0.07472749293500201,
      "grad_norm": 160.0850067138672,
      "learning_rate": 1.9895421207276155e-05,
      "loss": 6.9351,
      "step": 1851
    },
    {
      "epoch": 0.07476786435203876,
      "grad_norm": 167.5081024169922,
      "learning_rate": 1.9895232511785197e-05,
      "loss": 5.8142,
      "step": 1852
    },
    {
      "epoch": 0.0748082357690755,
      "grad_norm": 207.78497314453125,
      "learning_rate": 1.9895043647108868e-05,
      "loss": 6.3483,
      "step": 1853
    },
    {
      "epoch": 0.07484860718611223,
      "grad_norm": 117.57598876953125,
      "learning_rate": 1.9894854613250403e-05,
      "loss": 5.2691,
      "step": 1854
    },
    {
      "epoch": 0.07488897860314897,
      "grad_norm": 105.7937240600586,
      "learning_rate": 1.9894665410213027e-05,
      "loss": 6.6047,
      "step": 1855
    },
    {
      "epoch": 0.0749293500201857,
      "grad_norm": 111.07435607910156,
      "learning_rate": 1.989447603799998e-05,
      "loss": 5.5948,
      "step": 1856
    },
    {
      "epoch": 0.07496972143722244,
      "grad_norm": 106.8979263305664,
      "learning_rate": 1.9894286496614494e-05,
      "loss": 6.4643,
      "step": 1857
    },
    {
      "epoch": 0.07501009285425918,
      "grad_norm": 188.8857879638672,
      "learning_rate": 1.9894096786059818e-05,
      "loss": 5.3162,
      "step": 1858
    },
    {
      "epoch": 0.07505046427129593,
      "grad_norm": 92.2507095336914,
      "learning_rate": 1.9893906906339188e-05,
      "loss": 6.5472,
      "step": 1859
    },
    {
      "epoch": 0.07509083568833266,
      "grad_norm": 88.99063873291016,
      "learning_rate": 1.9893716857455856e-05,
      "loss": 6.1003,
      "step": 1860
    },
    {
      "epoch": 0.0751312071053694,
      "grad_norm": 86.45674133300781,
      "learning_rate": 1.9893526639413073e-05,
      "loss": 6.7259,
      "step": 1861
    },
    {
      "epoch": 0.07517157852240614,
      "grad_norm": 119.3402099609375,
      "learning_rate": 1.9893336252214082e-05,
      "loss": 6.0646,
      "step": 1862
    },
    {
      "epoch": 0.07521194993944287,
      "grad_norm": 114.8162612915039,
      "learning_rate": 1.9893145695862144e-05,
      "loss": 5.7796,
      "step": 1863
    },
    {
      "epoch": 0.0752523213564796,
      "grad_norm": 101.49811553955078,
      "learning_rate": 1.989295497036052e-05,
      "loss": 5.8881,
      "step": 1864
    },
    {
      "epoch": 0.07529269277351636,
      "grad_norm": 95.74071502685547,
      "learning_rate": 1.9892764075712465e-05,
      "loss": 3.9196,
      "step": 1865
    },
    {
      "epoch": 0.07533306419055309,
      "grad_norm": 117.90540313720703,
      "learning_rate": 1.9892573011921246e-05,
      "loss": 6.0476,
      "step": 1866
    },
    {
      "epoch": 0.07537343560758983,
      "grad_norm": 113.55733489990234,
      "learning_rate": 1.9892381778990127e-05,
      "loss": 6.7603,
      "step": 1867
    },
    {
      "epoch": 0.07541380702462656,
      "grad_norm": 130.92050170898438,
      "learning_rate": 1.9892190376922386e-05,
      "loss": 5.5085,
      "step": 1868
    },
    {
      "epoch": 0.0754541784416633,
      "grad_norm": 118.63275146484375,
      "learning_rate": 1.9891998805721283e-05,
      "loss": 5.7959,
      "step": 1869
    },
    {
      "epoch": 0.07549454985870004,
      "grad_norm": 191.57411193847656,
      "learning_rate": 1.9891807065390104e-05,
      "loss": 6.631,
      "step": 1870
    },
    {
      "epoch": 0.07553492127573677,
      "grad_norm": 90.36957550048828,
      "learning_rate": 1.9891615155932124e-05,
      "loss": 5.8302,
      "step": 1871
    },
    {
      "epoch": 0.07557529269277352,
      "grad_norm": 98.45429992675781,
      "learning_rate": 1.9891423077350622e-05,
      "loss": 5.4524,
      "step": 1872
    },
    {
      "epoch": 0.07561566410981026,
      "grad_norm": 112.0555191040039,
      "learning_rate": 1.9891230829648882e-05,
      "loss": 7.7251,
      "step": 1873
    },
    {
      "epoch": 0.075656035526847,
      "grad_norm": 128.86781311035156,
      "learning_rate": 1.9891038412830195e-05,
      "loss": 5.243,
      "step": 1874
    },
    {
      "epoch": 0.07569640694388373,
      "grad_norm": 142.8791961669922,
      "learning_rate": 1.9890845826897845e-05,
      "loss": 5.1852,
      "step": 1875
    },
    {
      "epoch": 0.07573677836092046,
      "grad_norm": 116.29597473144531,
      "learning_rate": 1.9890653071855134e-05,
      "loss": 6.2784,
      "step": 1876
    },
    {
      "epoch": 0.0757771497779572,
      "grad_norm": 125.3206558227539,
      "learning_rate": 1.9890460147705347e-05,
      "loss": 7.1307,
      "step": 1877
    },
    {
      "epoch": 0.07581752119499395,
      "grad_norm": 108.24259185791016,
      "learning_rate": 1.989026705445179e-05,
      "loss": 5.9076,
      "step": 1878
    },
    {
      "epoch": 0.07585789261203069,
      "grad_norm": 89.49580383300781,
      "learning_rate": 1.989007379209776e-05,
      "loss": 5.3899,
      "step": 1879
    },
    {
      "epoch": 0.07589826402906742,
      "grad_norm": 191.06031799316406,
      "learning_rate": 1.988988036064656e-05,
      "loss": 6.1744,
      "step": 1880
    },
    {
      "epoch": 0.07593863544610416,
      "grad_norm": 95.87675476074219,
      "learning_rate": 1.9889686760101504e-05,
      "loss": 6.1568,
      "step": 1881
    },
    {
      "epoch": 0.0759790068631409,
      "grad_norm": 101.82769012451172,
      "learning_rate": 1.9889492990465896e-05,
      "loss": 6.4305,
      "step": 1882
    },
    {
      "epoch": 0.07601937828017763,
      "grad_norm": 101.6128158569336,
      "learning_rate": 1.9889299051743057e-05,
      "loss": 6.3074,
      "step": 1883
    },
    {
      "epoch": 0.07605974969721437,
      "grad_norm": 131.87550354003906,
      "learning_rate": 1.988910494393629e-05,
      "loss": 5.0865,
      "step": 1884
    },
    {
      "epoch": 0.07610012111425112,
      "grad_norm": 125.86795806884766,
      "learning_rate": 1.9888910667048928e-05,
      "loss": 7.3072,
      "step": 1885
    },
    {
      "epoch": 0.07614049253128785,
      "grad_norm": 116.96745300292969,
      "learning_rate": 1.988871622108428e-05,
      "loss": 6.9535,
      "step": 1886
    },
    {
      "epoch": 0.07618086394832459,
      "grad_norm": 94.38243103027344,
      "learning_rate": 1.9888521606045682e-05,
      "loss": 6.5642,
      "step": 1887
    },
    {
      "epoch": 0.07622123536536132,
      "grad_norm": 121.09675598144531,
      "learning_rate": 1.9888326821936454e-05,
      "loss": 5.963,
      "step": 1888
    },
    {
      "epoch": 0.07626160678239806,
      "grad_norm": 125.22057342529297,
      "learning_rate": 1.9888131868759927e-05,
      "loss": 5.6777,
      "step": 1889
    },
    {
      "epoch": 0.0763019781994348,
      "grad_norm": 107.35791015625,
      "learning_rate": 1.9887936746519435e-05,
      "loss": 5.8238,
      "step": 1890
    },
    {
      "epoch": 0.07634234961647154,
      "grad_norm": 101.59339904785156,
      "learning_rate": 1.988774145521832e-05,
      "loss": 5.3321,
      "step": 1891
    },
    {
      "epoch": 0.07638272103350828,
      "grad_norm": 120.49222564697266,
      "learning_rate": 1.988754599485991e-05,
      "loss": 6.8588,
      "step": 1892
    },
    {
      "epoch": 0.07642309245054502,
      "grad_norm": 126.41382598876953,
      "learning_rate": 1.9887350365447553e-05,
      "loss": 5.9611,
      "step": 1893
    },
    {
      "epoch": 0.07646346386758175,
      "grad_norm": 124.25183868408203,
      "learning_rate": 1.9887154566984596e-05,
      "loss": 6.3966,
      "step": 1894
    },
    {
      "epoch": 0.07650383528461849,
      "grad_norm": 77.46491241455078,
      "learning_rate": 1.988695859947438e-05,
      "loss": 6.8735,
      "step": 1895
    },
    {
      "epoch": 0.07654420670165522,
      "grad_norm": 119.44178771972656,
      "learning_rate": 1.9886762462920264e-05,
      "loss": 5.1664,
      "step": 1896
    },
    {
      "epoch": 0.07658457811869196,
      "grad_norm": 114.0929946899414,
      "learning_rate": 1.9886566157325596e-05,
      "loss": 5.1917,
      "step": 1897
    },
    {
      "epoch": 0.07662494953572871,
      "grad_norm": 172.5492401123047,
      "learning_rate": 1.988636968269373e-05,
      "loss": 7.0192,
      "step": 1898
    },
    {
      "epoch": 0.07666532095276545,
      "grad_norm": 211.506103515625,
      "learning_rate": 1.988617303902803e-05,
      "loss": 5.0135,
      "step": 1899
    },
    {
      "epoch": 0.07670569236980218,
      "grad_norm": 99.43161010742188,
      "learning_rate": 1.988597622633186e-05,
      "loss": 5.9222,
      "step": 1900
    },
    {
      "epoch": 0.07674606378683892,
      "grad_norm": 91.55101013183594,
      "learning_rate": 1.988577924460858e-05,
      "loss": 5.8603,
      "step": 1901
    },
    {
      "epoch": 0.07678643520387565,
      "grad_norm": 125.7489013671875,
      "learning_rate": 1.988558209386156e-05,
      "loss": 7.2776,
      "step": 1902
    },
    {
      "epoch": 0.07682680662091239,
      "grad_norm": 94.0113525390625,
      "learning_rate": 1.988538477409417e-05,
      "loss": 6.0274,
      "step": 1903
    },
    {
      "epoch": 0.07686717803794914,
      "grad_norm": 81.71974182128906,
      "learning_rate": 1.988518728530979e-05,
      "loss": 5.5291,
      "step": 1904
    },
    {
      "epoch": 0.07690754945498587,
      "grad_norm": 179.83377075195312,
      "learning_rate": 1.9884989627511782e-05,
      "loss": 6.5961,
      "step": 1905
    },
    {
      "epoch": 0.07694792087202261,
      "grad_norm": 259.1990051269531,
      "learning_rate": 1.988479180070354e-05,
      "loss": 4.3631,
      "step": 1906
    },
    {
      "epoch": 0.07698829228905935,
      "grad_norm": 88.33531951904297,
      "learning_rate": 1.988459380488844e-05,
      "loss": 5.3303,
      "step": 1907
    },
    {
      "epoch": 0.07702866370609608,
      "grad_norm": 96.69048309326172,
      "learning_rate": 1.9884395640069868e-05,
      "loss": 5.5104,
      "step": 1908
    },
    {
      "epoch": 0.07706903512313282,
      "grad_norm": 88.22119140625,
      "learning_rate": 1.9884197306251213e-05,
      "loss": 5.4551,
      "step": 1909
    },
    {
      "epoch": 0.07710940654016955,
      "grad_norm": 223.1654510498047,
      "learning_rate": 1.9883998803435863e-05,
      "loss": 5.3865,
      "step": 1910
    },
    {
      "epoch": 0.0771497779572063,
      "grad_norm": 112.607177734375,
      "learning_rate": 1.9883800131627217e-05,
      "loss": 3.773,
      "step": 1911
    },
    {
      "epoch": 0.07719014937424304,
      "grad_norm": 116.22234344482422,
      "learning_rate": 1.9883601290828664e-05,
      "loss": 5.565,
      "step": 1912
    },
    {
      "epoch": 0.07723052079127977,
      "grad_norm": 89.01900482177734,
      "learning_rate": 1.9883402281043613e-05,
      "loss": 6.1269,
      "step": 1913
    },
    {
      "epoch": 0.07727089220831651,
      "grad_norm": 98.02239227294922,
      "learning_rate": 1.9883203102275462e-05,
      "loss": 6.8087,
      "step": 1914
    },
    {
      "epoch": 0.07731126362535325,
      "grad_norm": 117.91989135742188,
      "learning_rate": 1.9883003754527616e-05,
      "loss": 7.58,
      "step": 1915
    },
    {
      "epoch": 0.07735163504238998,
      "grad_norm": 81.195556640625,
      "learning_rate": 1.9882804237803487e-05,
      "loss": 6.3768,
      "step": 1916
    },
    {
      "epoch": 0.07739200645942673,
      "grad_norm": 102.49732971191406,
      "learning_rate": 1.9882604552106483e-05,
      "loss": 5.9766,
      "step": 1917
    },
    {
      "epoch": 0.07743237787646347,
      "grad_norm": 101.73907470703125,
      "learning_rate": 1.9882404697440016e-05,
      "loss": 5.9918,
      "step": 1918
    },
    {
      "epoch": 0.0774727492935002,
      "grad_norm": 106.3665771484375,
      "learning_rate": 1.9882204673807508e-05,
      "loss": 4.7682,
      "step": 1919
    },
    {
      "epoch": 0.07751312071053694,
      "grad_norm": 167.4385223388672,
      "learning_rate": 1.9882004481212375e-05,
      "loss": 6.4729,
      "step": 1920
    },
    {
      "epoch": 0.07755349212757368,
      "grad_norm": 207.7932891845703,
      "learning_rate": 1.988180411965804e-05,
      "loss": 4.6783,
      "step": 1921
    },
    {
      "epoch": 0.07759386354461041,
      "grad_norm": 100.74845123291016,
      "learning_rate": 1.9881603589147933e-05,
      "loss": 5.8104,
      "step": 1922
    },
    {
      "epoch": 0.07763423496164715,
      "grad_norm": 100.43582916259766,
      "learning_rate": 1.988140288968548e-05,
      "loss": 6.0131,
      "step": 1923
    },
    {
      "epoch": 0.0776746063786839,
      "grad_norm": 159.09002685546875,
      "learning_rate": 1.9881202021274114e-05,
      "loss": 3.8453,
      "step": 1924
    },
    {
      "epoch": 0.07771497779572063,
      "grad_norm": 154.7053985595703,
      "learning_rate": 1.988100098391726e-05,
      "loss": 4.9666,
      "step": 1925
    },
    {
      "epoch": 0.07775534921275737,
      "grad_norm": 100.71346282958984,
      "learning_rate": 1.9880799777618372e-05,
      "loss": 6.74,
      "step": 1926
    },
    {
      "epoch": 0.0777957206297941,
      "grad_norm": 154.07235717773438,
      "learning_rate": 1.9880598402380876e-05,
      "loss": 5.7302,
      "step": 1927
    },
    {
      "epoch": 0.07783609204683084,
      "grad_norm": 111.46439361572266,
      "learning_rate": 1.988039685820822e-05,
      "loss": 5.8456,
      "step": 1928
    },
    {
      "epoch": 0.07787646346386758,
      "grad_norm": 152.37303161621094,
      "learning_rate": 1.9880195145103855e-05,
      "loss": 6.9628,
      "step": 1929
    },
    {
      "epoch": 0.07791683488090433,
      "grad_norm": 107.06867218017578,
      "learning_rate": 1.987999326307122e-05,
      "loss": 4.9138,
      "step": 1930
    },
    {
      "epoch": 0.07795720629794106,
      "grad_norm": 120.81555938720703,
      "learning_rate": 1.987979121211377e-05,
      "loss": 5.9307,
      "step": 1931
    },
    {
      "epoch": 0.0779975777149778,
      "grad_norm": 107.52545928955078,
      "learning_rate": 1.987958899223497e-05,
      "loss": 6.7285,
      "step": 1932
    },
    {
      "epoch": 0.07803794913201453,
      "grad_norm": 118.79673767089844,
      "learning_rate": 1.9879386603438262e-05,
      "loss": 6.6767,
      "step": 1933
    },
    {
      "epoch": 0.07807832054905127,
      "grad_norm": 132.1378936767578,
      "learning_rate": 1.9879184045727115e-05,
      "loss": 5.3,
      "step": 1934
    },
    {
      "epoch": 0.078118691966088,
      "grad_norm": 89.77194213867188,
      "learning_rate": 1.9878981319104987e-05,
      "loss": 7.1584,
      "step": 1935
    },
    {
      "epoch": 0.07815906338312474,
      "grad_norm": 128.04403686523438,
      "learning_rate": 1.9878778423575353e-05,
      "loss": 5.9548,
      "step": 1936
    },
    {
      "epoch": 0.07819943480016149,
      "grad_norm": 107.3753433227539,
      "learning_rate": 1.9878575359141673e-05,
      "loss": 5.7685,
      "step": 1937
    },
    {
      "epoch": 0.07823980621719823,
      "grad_norm": 98.47303771972656,
      "learning_rate": 1.9878372125807425e-05,
      "loss": 5.7181,
      "step": 1938
    },
    {
      "epoch": 0.07828017763423496,
      "grad_norm": 123.92550659179688,
      "learning_rate": 1.9878168723576077e-05,
      "loss": 6.0711,
      "step": 1939
    },
    {
      "epoch": 0.0783205490512717,
      "grad_norm": 93.65348052978516,
      "learning_rate": 1.9877965152451117e-05,
      "loss": 6.0917,
      "step": 1940
    },
    {
      "epoch": 0.07836092046830843,
      "grad_norm": 128.97482299804688,
      "learning_rate": 1.9877761412436013e-05,
      "loss": 4.1336,
      "step": 1941
    },
    {
      "epoch": 0.07840129188534517,
      "grad_norm": 114.72053527832031,
      "learning_rate": 1.9877557503534256e-05,
      "loss": 4.499,
      "step": 1942
    },
    {
      "epoch": 0.07844166330238192,
      "grad_norm": 141.9522705078125,
      "learning_rate": 1.9877353425749336e-05,
      "loss": 5.4436,
      "step": 1943
    },
    {
      "epoch": 0.07848203471941866,
      "grad_norm": 81.52469635009766,
      "learning_rate": 1.9877149179084738e-05,
      "loss": 6.1473,
      "step": 1944
    },
    {
      "epoch": 0.07852240613645539,
      "grad_norm": 142.90896606445312,
      "learning_rate": 1.987694476354395e-05,
      "loss": 5.0773,
      "step": 1945
    },
    {
      "epoch": 0.07856277755349213,
      "grad_norm": 123.60101318359375,
      "learning_rate": 1.987674017913047e-05,
      "loss": 4.9098,
      "step": 1946
    },
    {
      "epoch": 0.07860314897052886,
      "grad_norm": 108.86145782470703,
      "learning_rate": 1.9876535425847797e-05,
      "loss": 6.666,
      "step": 1947
    },
    {
      "epoch": 0.0786435203875656,
      "grad_norm": 132.25656127929688,
      "learning_rate": 1.9876330503699438e-05,
      "loss": 5.4402,
      "step": 1948
    },
    {
      "epoch": 0.07868389180460234,
      "grad_norm": 74.38346099853516,
      "learning_rate": 1.9876125412688885e-05,
      "loss": 6.68,
      "step": 1949
    },
    {
      "epoch": 0.07872426322163908,
      "grad_norm": 89.47254180908203,
      "learning_rate": 1.987592015281965e-05,
      "loss": 4.2797,
      "step": 1950
    },
    {
      "epoch": 0.07876463463867582,
      "grad_norm": 176.31442260742188,
      "learning_rate": 1.9875714724095244e-05,
      "loss": 6.5676,
      "step": 1951
    },
    {
      "epoch": 0.07880500605571256,
      "grad_norm": 102.73755645751953,
      "learning_rate": 1.9875509126519174e-05,
      "loss": 4.3667,
      "step": 1952
    },
    {
      "epoch": 0.07884537747274929,
      "grad_norm": 89.821533203125,
      "learning_rate": 1.9875303360094964e-05,
      "loss": 6.3645,
      "step": 1953
    },
    {
      "epoch": 0.07888574888978603,
      "grad_norm": 88.2324447631836,
      "learning_rate": 1.9875097424826123e-05,
      "loss": 4.4264,
      "step": 1954
    },
    {
      "epoch": 0.07892612030682276,
      "grad_norm": 72.03790283203125,
      "learning_rate": 1.9874891320716178e-05,
      "loss": 6.4952,
      "step": 1955
    },
    {
      "epoch": 0.07896649172385951,
      "grad_norm": 224.8321533203125,
      "learning_rate": 1.987468504776865e-05,
      "loss": 7.0319,
      "step": 1956
    },
    {
      "epoch": 0.07900686314089625,
      "grad_norm": 121.46800994873047,
      "learning_rate": 1.9874478605987066e-05,
      "loss": 6.198,
      "step": 1957
    },
    {
      "epoch": 0.07904723455793299,
      "grad_norm": 89.71409606933594,
      "learning_rate": 1.987427199537496e-05,
      "loss": 5.705,
      "step": 1958
    },
    {
      "epoch": 0.07908760597496972,
      "grad_norm": 129.70582580566406,
      "learning_rate": 1.987406521593586e-05,
      "loss": 6.6201,
      "step": 1959
    },
    {
      "epoch": 0.07912797739200646,
      "grad_norm": 119.44322967529297,
      "learning_rate": 1.9873858267673302e-05,
      "loss": 5.2647,
      "step": 1960
    },
    {
      "epoch": 0.07916834880904319,
      "grad_norm": 98.0620346069336,
      "learning_rate": 1.987365115059082e-05,
      "loss": 4.5075,
      "step": 1961
    },
    {
      "epoch": 0.07920872022607993,
      "grad_norm": 284.1201171875,
      "learning_rate": 1.9873443864691968e-05,
      "loss": 5.4621,
      "step": 1962
    },
    {
      "epoch": 0.07924909164311668,
      "grad_norm": 104.43948364257812,
      "learning_rate": 1.9873236409980274e-05,
      "loss": 4.5174,
      "step": 1963
    },
    {
      "epoch": 0.07928946306015341,
      "grad_norm": 184.47039794921875,
      "learning_rate": 1.9873028786459298e-05,
      "loss": 5.2832,
      "step": 1964
    },
    {
      "epoch": 0.07932983447719015,
      "grad_norm": 106.41658020019531,
      "learning_rate": 1.9872820994132584e-05,
      "loss": 6.4186,
      "step": 1965
    },
    {
      "epoch": 0.07937020589422689,
      "grad_norm": 121.69325256347656,
      "learning_rate": 1.9872613033003686e-05,
      "loss": 5.695,
      "step": 1966
    },
    {
      "epoch": 0.07941057731126362,
      "grad_norm": 103.270751953125,
      "learning_rate": 1.9872404903076158e-05,
      "loss": 6.4007,
      "step": 1967
    },
    {
      "epoch": 0.07945094872830036,
      "grad_norm": 82.34626007080078,
      "learning_rate": 1.987219660435356e-05,
      "loss": 5.0928,
      "step": 1968
    },
    {
      "epoch": 0.07949132014533711,
      "grad_norm": 98.82298278808594,
      "learning_rate": 1.9871988136839456e-05,
      "loss": 4.8743,
      "step": 1969
    },
    {
      "epoch": 0.07953169156237384,
      "grad_norm": 110.74838256835938,
      "learning_rate": 1.9871779500537404e-05,
      "loss": 5.1219,
      "step": 1970
    },
    {
      "epoch": 0.07957206297941058,
      "grad_norm": 111.04712677001953,
      "learning_rate": 1.9871570695450976e-05,
      "loss": 4.5885,
      "step": 1971
    },
    {
      "epoch": 0.07961243439644732,
      "grad_norm": 104.10242462158203,
      "learning_rate": 1.9871361721583743e-05,
      "loss": 4.988,
      "step": 1972
    },
    {
      "epoch": 0.07965280581348405,
      "grad_norm": 89.39945220947266,
      "learning_rate": 1.9871152578939273e-05,
      "loss": 6.3698,
      "step": 1973
    },
    {
      "epoch": 0.07969317723052079,
      "grad_norm": 205.84588623046875,
      "learning_rate": 1.9870943267521144e-05,
      "loss": 5.6694,
      "step": 1974
    },
    {
      "epoch": 0.07973354864755752,
      "grad_norm": 107.09418487548828,
      "learning_rate": 1.987073378733294e-05,
      "loss": 6.8338,
      "step": 1975
    },
    {
      "epoch": 0.07977392006459427,
      "grad_norm": 80.15637969970703,
      "learning_rate": 1.9870524138378236e-05,
      "loss": 5.3055,
      "step": 1976
    },
    {
      "epoch": 0.07981429148163101,
      "grad_norm": 116.8589096069336,
      "learning_rate": 1.9870314320660616e-05,
      "loss": 6.106,
      "step": 1977
    },
    {
      "epoch": 0.07985466289866774,
      "grad_norm": 102.98173522949219,
      "learning_rate": 1.9870104334183672e-05,
      "loss": 5.8491,
      "step": 1978
    },
    {
      "epoch": 0.07989503431570448,
      "grad_norm": 153.498046875,
      "learning_rate": 1.9869894178950992e-05,
      "loss": 6.9404,
      "step": 1979
    },
    {
      "epoch": 0.07993540573274122,
      "grad_norm": 94.18878173828125,
      "learning_rate": 1.986968385496617e-05,
      "loss": 4.2865,
      "step": 1980
    },
    {
      "epoch": 0.07997577714977795,
      "grad_norm": 121.67449951171875,
      "learning_rate": 1.98694733622328e-05,
      "loss": 6.7304,
      "step": 1981
    },
    {
      "epoch": 0.0800161485668147,
      "grad_norm": 119.22216033935547,
      "learning_rate": 1.9869262700754483e-05,
      "loss": 4.8853,
      "step": 1982
    },
    {
      "epoch": 0.08005651998385144,
      "grad_norm": 130.79241943359375,
      "learning_rate": 1.986905187053482e-05,
      "loss": 5.9887,
      "step": 1983
    },
    {
      "epoch": 0.08009689140088817,
      "grad_norm": 99.89208984375,
      "learning_rate": 1.986884087157742e-05,
      "loss": 3.8899,
      "step": 1984
    },
    {
      "epoch": 0.08013726281792491,
      "grad_norm": 179.905029296875,
      "learning_rate": 1.986862970388588e-05,
      "loss": 5.2465,
      "step": 1985
    },
    {
      "epoch": 0.08017763423496164,
      "grad_norm": 109.9591293334961,
      "learning_rate": 1.986841836746382e-05,
      "loss": 4.6956,
      "step": 1986
    },
    {
      "epoch": 0.08021800565199838,
      "grad_norm": 112.28936767578125,
      "learning_rate": 1.986820686231485e-05,
      "loss": 5.2578,
      "step": 1987
    },
    {
      "epoch": 0.08025837706903512,
      "grad_norm": 123.91624450683594,
      "learning_rate": 1.9867995188442588e-05,
      "loss": 6.2109,
      "step": 1988
    },
    {
      "epoch": 0.08029874848607187,
      "grad_norm": 179.40298461914062,
      "learning_rate": 1.9867783345850647e-05,
      "loss": 6.9639,
      "step": 1989
    },
    {
      "epoch": 0.0803391199031086,
      "grad_norm": 159.2438507080078,
      "learning_rate": 1.986757133454266e-05,
      "loss": 5.0827,
      "step": 1990
    },
    {
      "epoch": 0.08037949132014534,
      "grad_norm": 182.34011840820312,
      "learning_rate": 1.9867359154522242e-05,
      "loss": 5.9323,
      "step": 1991
    },
    {
      "epoch": 0.08041986273718207,
      "grad_norm": 84.92507934570312,
      "learning_rate": 1.9867146805793025e-05,
      "loss": 4.5718,
      "step": 1992
    },
    {
      "epoch": 0.08046023415421881,
      "grad_norm": 117.67469787597656,
      "learning_rate": 1.9866934288358642e-05,
      "loss": 6.7429,
      "step": 1993
    },
    {
      "epoch": 0.08050060557125555,
      "grad_norm": 125.4376220703125,
      "learning_rate": 1.986672160222272e-05,
      "loss": 4.6525,
      "step": 1994
    },
    {
      "epoch": 0.0805409769882923,
      "grad_norm": 136.58604431152344,
      "learning_rate": 1.9866508747388898e-05,
      "loss": 6.7674,
      "step": 1995
    },
    {
      "epoch": 0.08058134840532903,
      "grad_norm": 141.8885498046875,
      "learning_rate": 1.986629572386082e-05,
      "loss": 5.3968,
      "step": 1996
    },
    {
      "epoch": 0.08062171982236577,
      "grad_norm": 184.5519561767578,
      "learning_rate": 1.9866082531642123e-05,
      "loss": 4.1276,
      "step": 1997
    },
    {
      "epoch": 0.0806620912394025,
      "grad_norm": 105.85820770263672,
      "learning_rate": 1.9865869170736452e-05,
      "loss": 6.1071,
      "step": 1998
    },
    {
      "epoch": 0.08070246265643924,
      "grad_norm": 202.98399353027344,
      "learning_rate": 1.986565564114746e-05,
      "loss": 5.7336,
      "step": 1999
    },
    {
      "epoch": 0.08074283407347597,
      "grad_norm": 106.7049560546875,
      "learning_rate": 1.986544194287879e-05,
      "loss": 5.9921,
      "step": 2000
    },
    {
      "epoch": 0.08078320549051271,
      "grad_norm": 85.287353515625,
      "learning_rate": 1.9865228075934106e-05,
      "loss": 5.1627,
      "step": 2001
    },
    {
      "epoch": 0.08082357690754946,
      "grad_norm": 104.07858276367188,
      "learning_rate": 1.9865014040317056e-05,
      "loss": 5.5211,
      "step": 2002
    },
    {
      "epoch": 0.0808639483245862,
      "grad_norm": 111.3232192993164,
      "learning_rate": 1.98647998360313e-05,
      "loss": 5.7015,
      "step": 2003
    },
    {
      "epoch": 0.08090431974162293,
      "grad_norm": 103.76757049560547,
      "learning_rate": 1.9864585463080507e-05,
      "loss": 5.2029,
      "step": 2004
    },
    {
      "epoch": 0.08094469115865967,
      "grad_norm": 105.62128448486328,
      "learning_rate": 1.9864370921468333e-05,
      "loss": 6.167,
      "step": 2005
    },
    {
      "epoch": 0.0809850625756964,
      "grad_norm": 231.81883239746094,
      "learning_rate": 1.9864156211198455e-05,
      "loss": 5.6899,
      "step": 2006
    },
    {
      "epoch": 0.08102543399273314,
      "grad_norm": 104.58550262451172,
      "learning_rate": 1.9863941332274542e-05,
      "loss": 5.3147,
      "step": 2007
    },
    {
      "epoch": 0.08106580540976989,
      "grad_norm": 100.51402282714844,
      "learning_rate": 1.986372628470026e-05,
      "loss": 5.1547,
      "step": 2008
    },
    {
      "epoch": 0.08110617682680663,
      "grad_norm": 95.66789245605469,
      "learning_rate": 1.9863511068479297e-05,
      "loss": 5.4916,
      "step": 2009
    },
    {
      "epoch": 0.08114654824384336,
      "grad_norm": 114.55594635009766,
      "learning_rate": 1.9863295683615328e-05,
      "loss": 6.5903,
      "step": 2010
    },
    {
      "epoch": 0.0811869196608801,
      "grad_norm": 233.0817108154297,
      "learning_rate": 1.9863080130112035e-05,
      "loss": 5.9088,
      "step": 2011
    },
    {
      "epoch": 0.08122729107791683,
      "grad_norm": 168.80926513671875,
      "learning_rate": 1.98628644079731e-05,
      "loss": 5.349,
      "step": 2012
    },
    {
      "epoch": 0.08126766249495357,
      "grad_norm": 96.36091613769531,
      "learning_rate": 1.9862648517202217e-05,
      "loss": 4.808,
      "step": 2013
    },
    {
      "epoch": 0.0813080339119903,
      "grad_norm": 132.49085998535156,
      "learning_rate": 1.986243245780308e-05,
      "loss": 6.0159,
      "step": 2014
    },
    {
      "epoch": 0.08134840532902705,
      "grad_norm": 113.02725219726562,
      "learning_rate": 1.986221622977937e-05,
      "loss": 6.1346,
      "step": 2015
    },
    {
      "epoch": 0.08138877674606379,
      "grad_norm": 98.63304901123047,
      "learning_rate": 1.9861999833134798e-05,
      "loss": 4.3391,
      "step": 2016
    },
    {
      "epoch": 0.08142914816310053,
      "grad_norm": 100.2426986694336,
      "learning_rate": 1.9861783267873055e-05,
      "loss": 5.8746,
      "step": 2017
    },
    {
      "epoch": 0.08146951958013726,
      "grad_norm": 136.98724365234375,
      "learning_rate": 1.986156653399785e-05,
      "loss": 5.2509,
      "step": 2018
    },
    {
      "epoch": 0.081509890997174,
      "grad_norm": 81.86554718017578,
      "learning_rate": 1.9861349631512884e-05,
      "loss": 5.0602,
      "step": 2019
    },
    {
      "epoch": 0.08155026241421073,
      "grad_norm": 92.0208740234375,
      "learning_rate": 1.9861132560421866e-05,
      "loss": 6.4424,
      "step": 2020
    },
    {
      "epoch": 0.08159063383124748,
      "grad_norm": 92.87113952636719,
      "learning_rate": 1.9860915320728513e-05,
      "loss": 5.0553,
      "step": 2021
    },
    {
      "epoch": 0.08163100524828422,
      "grad_norm": 179.59471130371094,
      "learning_rate": 1.986069791243653e-05,
      "loss": 5.7204,
      "step": 2022
    },
    {
      "epoch": 0.08167137666532095,
      "grad_norm": 107.37735748291016,
      "learning_rate": 1.9860480335549645e-05,
      "loss": 5.6311,
      "step": 2023
    },
    {
      "epoch": 0.08171174808235769,
      "grad_norm": 133.8788604736328,
      "learning_rate": 1.9860262590071567e-05,
      "loss": 5.5043,
      "step": 2024
    },
    {
      "epoch": 0.08175211949939443,
      "grad_norm": 186.01876831054688,
      "learning_rate": 1.9860044676006024e-05,
      "loss": 7.0377,
      "step": 2025
    },
    {
      "epoch": 0.08179249091643116,
      "grad_norm": 121.64598846435547,
      "learning_rate": 1.9859826593356744e-05,
      "loss": 7.0362,
      "step": 2026
    },
    {
      "epoch": 0.0818328623334679,
      "grad_norm": 148.69424438476562,
      "learning_rate": 1.9859608342127453e-05,
      "loss": 5.2567,
      "step": 2027
    },
    {
      "epoch": 0.08187323375050465,
      "grad_norm": 138.74212646484375,
      "learning_rate": 1.9859389922321882e-05,
      "loss": 5.4125,
      "step": 2028
    },
    {
      "epoch": 0.08191360516754138,
      "grad_norm": 96.28438568115234,
      "learning_rate": 1.985917133394377e-05,
      "loss": 5.2851,
      "step": 2029
    },
    {
      "epoch": 0.08195397658457812,
      "grad_norm": 160.43968200683594,
      "learning_rate": 1.9858952576996848e-05,
      "loss": 5.6914,
      "step": 2030
    },
    {
      "epoch": 0.08199434800161486,
      "grad_norm": 118.40886688232422,
      "learning_rate": 1.985873365148486e-05,
      "loss": 5.8788,
      "step": 2031
    },
    {
      "epoch": 0.08203471941865159,
      "grad_norm": 151.41526794433594,
      "learning_rate": 1.985851455741155e-05,
      "loss": 6.2456,
      "step": 2032
    },
    {
      "epoch": 0.08207509083568833,
      "grad_norm": 155.63941955566406,
      "learning_rate": 1.985829529478066e-05,
      "loss": 5.2187,
      "step": 2033
    },
    {
      "epoch": 0.08211546225272508,
      "grad_norm": 93.47346496582031,
      "learning_rate": 1.9858075863595942e-05,
      "loss": 5.5552,
      "step": 2034
    },
    {
      "epoch": 0.08215583366976181,
      "grad_norm": 143.72047424316406,
      "learning_rate": 1.9857856263861146e-05,
      "loss": 6.5489,
      "step": 2035
    },
    {
      "epoch": 0.08219620508679855,
      "grad_norm": 109.94172668457031,
      "learning_rate": 1.985763649558003e-05,
      "loss": 6.392,
      "step": 2036
    },
    {
      "epoch": 0.08223657650383528,
      "grad_norm": 128.90989685058594,
      "learning_rate": 1.9857416558756347e-05,
      "loss": 7.4584,
      "step": 2037
    },
    {
      "epoch": 0.08227694792087202,
      "grad_norm": 119.23197174072266,
      "learning_rate": 1.9857196453393863e-05,
      "loss": 6.2754,
      "step": 2038
    },
    {
      "epoch": 0.08231731933790876,
      "grad_norm": 133.70016479492188,
      "learning_rate": 1.9856976179496335e-05,
      "loss": 5.4209,
      "step": 2039
    },
    {
      "epoch": 0.08235769075494549,
      "grad_norm": 119.08110809326172,
      "learning_rate": 1.985675573706753e-05,
      "loss": 5.1113,
      "step": 2040
    },
    {
      "epoch": 0.08239806217198224,
      "grad_norm": 74.24480438232422,
      "learning_rate": 1.9856535126111223e-05,
      "loss": 5.5229,
      "step": 2041
    },
    {
      "epoch": 0.08243843358901898,
      "grad_norm": 105.50193786621094,
      "learning_rate": 1.985631434663118e-05,
      "loss": 3.852,
      "step": 2042
    },
    {
      "epoch": 0.08247880500605571,
      "grad_norm": 123.55300903320312,
      "learning_rate": 1.9856093398631176e-05,
      "loss": 4.7525,
      "step": 2043
    },
    {
      "epoch": 0.08251917642309245,
      "grad_norm": 124.4854736328125,
      "learning_rate": 1.9855872282114993e-05,
      "loss": 6.4898,
      "step": 2044
    },
    {
      "epoch": 0.08255954784012919,
      "grad_norm": 112.19326782226562,
      "learning_rate": 1.985565099708641e-05,
      "loss": 5.4671,
      "step": 2045
    },
    {
      "epoch": 0.08259991925716592,
      "grad_norm": 123.13401794433594,
      "learning_rate": 1.985542954354921e-05,
      "loss": 4.9369,
      "step": 2046
    },
    {
      "epoch": 0.08264029067420267,
      "grad_norm": 80.83904266357422,
      "learning_rate": 1.9855207921507176e-05,
      "loss": 5.4082,
      "step": 2047
    },
    {
      "epoch": 0.0826806620912394,
      "grad_norm": 89.1283187866211,
      "learning_rate": 1.98549861309641e-05,
      "loss": 6.0053,
      "step": 2048
    },
    {
      "epoch": 0.08272103350827614,
      "grad_norm": 71.70372009277344,
      "learning_rate": 1.9854764171923778e-05,
      "loss": 5.3211,
      "step": 2049
    },
    {
      "epoch": 0.08276140492531288,
      "grad_norm": 120.28062438964844,
      "learning_rate": 1.985454204439e-05,
      "loss": 6.8398,
      "step": 2050
    },
    {
      "epoch": 0.08280177634234961,
      "grad_norm": 100.44122314453125,
      "learning_rate": 1.985431974836656e-05,
      "loss": 7.5472,
      "step": 2051
    },
    {
      "epoch": 0.08284214775938635,
      "grad_norm": 88.6766128540039,
      "learning_rate": 1.985409728385727e-05,
      "loss": 5.921,
      "step": 2052
    },
    {
      "epoch": 0.08288251917642309,
      "grad_norm": 166.30892944335938,
      "learning_rate": 1.9853874650865923e-05,
      "loss": 4.7294,
      "step": 2053
    },
    {
      "epoch": 0.08292289059345984,
      "grad_norm": 150.5365753173828,
      "learning_rate": 1.9853651849396334e-05,
      "loss": 6.209,
      "step": 2054
    },
    {
      "epoch": 0.08296326201049657,
      "grad_norm": 134.60528564453125,
      "learning_rate": 1.9853428879452305e-05,
      "loss": 4.2265,
      "step": 2055
    },
    {
      "epoch": 0.08300363342753331,
      "grad_norm": 95.53876495361328,
      "learning_rate": 1.9853205741037652e-05,
      "loss": 5.7352,
      "step": 2056
    },
    {
      "epoch": 0.08304400484457004,
      "grad_norm": 126.67676544189453,
      "learning_rate": 1.9852982434156192e-05,
      "loss": 5.9881,
      "step": 2057
    },
    {
      "epoch": 0.08308437626160678,
      "grad_norm": 144.27752685546875,
      "learning_rate": 1.985275895881174e-05,
      "loss": 5.359,
      "step": 2058
    },
    {
      "epoch": 0.08312474767864352,
      "grad_norm": 123.49203491210938,
      "learning_rate": 1.9852535315008117e-05,
      "loss": 5.8412,
      "step": 2059
    },
    {
      "epoch": 0.08316511909568026,
      "grad_norm": 138.7538604736328,
      "learning_rate": 1.9852311502749145e-05,
      "loss": 5.9605,
      "step": 2060
    },
    {
      "epoch": 0.083205490512717,
      "grad_norm": 109.51343536376953,
      "learning_rate": 1.9852087522038657e-05,
      "loss": 5.3728,
      "step": 2061
    },
    {
      "epoch": 0.08324586192975374,
      "grad_norm": 196.85223388671875,
      "learning_rate": 1.985186337288048e-05,
      "loss": 4.8164,
      "step": 2062
    },
    {
      "epoch": 0.08328623334679047,
      "grad_norm": 159.60040283203125,
      "learning_rate": 1.985163905527844e-05,
      "loss": 5.3532,
      "step": 2063
    },
    {
      "epoch": 0.08332660476382721,
      "grad_norm": 93.53486633300781,
      "learning_rate": 1.985141456923638e-05,
      "loss": 5.8515,
      "step": 2064
    },
    {
      "epoch": 0.08336697618086394,
      "grad_norm": 220.28643798828125,
      "learning_rate": 1.9851189914758135e-05,
      "loss": 4.8432,
      "step": 2065
    },
    {
      "epoch": 0.08340734759790068,
      "grad_norm": 216.19830322265625,
      "learning_rate": 1.9850965091847544e-05,
      "loss": 4.2934,
      "step": 2066
    },
    {
      "epoch": 0.08344771901493743,
      "grad_norm": 218.0153045654297,
      "learning_rate": 1.9850740100508458e-05,
      "loss": 5.9715,
      "step": 2067
    },
    {
      "epoch": 0.08348809043197417,
      "grad_norm": 79.01233673095703,
      "learning_rate": 1.985051494074472e-05,
      "loss": 5.7082,
      "step": 2068
    },
    {
      "epoch": 0.0835284618490109,
      "grad_norm": 130.90380859375,
      "learning_rate": 1.9850289612560175e-05,
      "loss": 5.1979,
      "step": 2069
    },
    {
      "epoch": 0.08356883326604764,
      "grad_norm": 95.17967987060547,
      "learning_rate": 1.985006411595868e-05,
      "loss": 4.9006,
      "step": 2070
    },
    {
      "epoch": 0.08360920468308437,
      "grad_norm": 165.19520568847656,
      "learning_rate": 1.9849838450944093e-05,
      "loss": 5.3603,
      "step": 2071
    },
    {
      "epoch": 0.08364957610012111,
      "grad_norm": 120.33557891845703,
      "learning_rate": 1.984961261752027e-05,
      "loss": 5.8377,
      "step": 2072
    },
    {
      "epoch": 0.08368994751715786,
      "grad_norm": 102.62162017822266,
      "learning_rate": 1.984938661569107e-05,
      "loss": 5.9195,
      "step": 2073
    },
    {
      "epoch": 0.0837303189341946,
      "grad_norm": 113.66693878173828,
      "learning_rate": 1.9849160445460357e-05,
      "loss": 4.8677,
      "step": 2074
    },
    {
      "epoch": 0.08377069035123133,
      "grad_norm": 165.81985473632812,
      "learning_rate": 1.9848934106832002e-05,
      "loss": 6.3401,
      "step": 2075
    },
    {
      "epoch": 0.08381106176826807,
      "grad_norm": 139.712158203125,
      "learning_rate": 1.9848707599809872e-05,
      "loss": 5.0188,
      "step": 2076
    },
    {
      "epoch": 0.0838514331853048,
      "grad_norm": 127.0848388671875,
      "learning_rate": 1.9848480924397843e-05,
      "loss": 5.1864,
      "step": 2077
    },
    {
      "epoch": 0.08389180460234154,
      "grad_norm": 144.3223876953125,
      "learning_rate": 1.9848254080599785e-05,
      "loss": 5.9493,
      "step": 2078
    },
    {
      "epoch": 0.08393217601937827,
      "grad_norm": 161.8893585205078,
      "learning_rate": 1.9848027068419582e-05,
      "loss": 5.7738,
      "step": 2079
    },
    {
      "epoch": 0.08397254743641502,
      "grad_norm": 158.5736083984375,
      "learning_rate": 1.984779988786111e-05,
      "loss": 3.1467,
      "step": 2080
    },
    {
      "epoch": 0.08401291885345176,
      "grad_norm": 137.8657684326172,
      "learning_rate": 1.9847572538928256e-05,
      "loss": 7.2012,
      "step": 2081
    },
    {
      "epoch": 0.0840532902704885,
      "grad_norm": 130.83255004882812,
      "learning_rate": 1.9847345021624912e-05,
      "loss": 6.509,
      "step": 2082
    },
    {
      "epoch": 0.08409366168752523,
      "grad_norm": 99.54427337646484,
      "learning_rate": 1.984711733595496e-05,
      "loss": 5.0412,
      "step": 2083
    },
    {
      "epoch": 0.08413403310456197,
      "grad_norm": 94.5191650390625,
      "learning_rate": 1.9846889481922297e-05,
      "loss": 6.083,
      "step": 2084
    },
    {
      "epoch": 0.0841744045215987,
      "grad_norm": 139.97720336914062,
      "learning_rate": 1.9846661459530815e-05,
      "loss": 5.0276,
      "step": 2085
    },
    {
      "epoch": 0.08421477593863545,
      "grad_norm": 95.88870239257812,
      "learning_rate": 1.984643326878442e-05,
      "loss": 6.2473,
      "step": 2086
    },
    {
      "epoch": 0.08425514735567219,
      "grad_norm": 125.56643676757812,
      "learning_rate": 1.9846204909687006e-05,
      "loss": 5.9427,
      "step": 2087
    },
    {
      "epoch": 0.08429551877270892,
      "grad_norm": 86.38180541992188,
      "learning_rate": 1.9845976382242483e-05,
      "loss": 4.4485,
      "step": 2088
    },
    {
      "epoch": 0.08433589018974566,
      "grad_norm": 116.34054565429688,
      "learning_rate": 1.9845747686454752e-05,
      "loss": 6.3109,
      "step": 2089
    },
    {
      "epoch": 0.0843762616067824,
      "grad_norm": 129.42605590820312,
      "learning_rate": 1.984551882232773e-05,
      "loss": 6.8094,
      "step": 2090
    },
    {
      "epoch": 0.08441663302381913,
      "grad_norm": 138.9141845703125,
      "learning_rate": 1.984528978986533e-05,
      "loss": 5.4267,
      "step": 2091
    },
    {
      "epoch": 0.08445700444085587,
      "grad_norm": 119.94247436523438,
      "learning_rate": 1.984506058907146e-05,
      "loss": 6.0045,
      "step": 2092
    },
    {
      "epoch": 0.08449737585789262,
      "grad_norm": 153.88917541503906,
      "learning_rate": 1.9844831219950043e-05,
      "loss": 5.6564,
      "step": 2093
    },
    {
      "epoch": 0.08453774727492935,
      "grad_norm": 175.90374755859375,
      "learning_rate": 1.9844601682505006e-05,
      "loss": 5.9274,
      "step": 2094
    },
    {
      "epoch": 0.08457811869196609,
      "grad_norm": 260.2723388671875,
      "learning_rate": 1.9844371976740268e-05,
      "loss": 5.3513,
      "step": 2095
    },
    {
      "epoch": 0.08461849010900283,
      "grad_norm": 106.64266967773438,
      "learning_rate": 1.9844142102659756e-05,
      "loss": 5.6207,
      "step": 2096
    },
    {
      "epoch": 0.08465886152603956,
      "grad_norm": 218.7840576171875,
      "learning_rate": 1.98439120602674e-05,
      "loss": 6.1622,
      "step": 2097
    },
    {
      "epoch": 0.0846992329430763,
      "grad_norm": 95.66950225830078,
      "learning_rate": 1.984368184956714e-05,
      "loss": 5.1597,
      "step": 2098
    },
    {
      "epoch": 0.08473960436011305,
      "grad_norm": 138.58511352539062,
      "learning_rate": 1.9843451470562906e-05,
      "loss": 4.333,
      "step": 2099
    },
    {
      "epoch": 0.08477997577714978,
      "grad_norm": 93.72059631347656,
      "learning_rate": 1.9843220923258636e-05,
      "loss": 6.6956,
      "step": 2100
    },
    {
      "epoch": 0.08482034719418652,
      "grad_norm": 117.31832122802734,
      "learning_rate": 1.9842990207658274e-05,
      "loss": 6.8787,
      "step": 2101
    },
    {
      "epoch": 0.08486071861122325,
      "grad_norm": 106.70864868164062,
      "learning_rate": 1.9842759323765764e-05,
      "loss": 6.4906,
      "step": 2102
    },
    {
      "epoch": 0.08490109002825999,
      "grad_norm": 81.65569305419922,
      "learning_rate": 1.9842528271585057e-05,
      "loss": 4.7143,
      "step": 2103
    },
    {
      "epoch": 0.08494146144529673,
      "grad_norm": 86.97477722167969,
      "learning_rate": 1.9842297051120095e-05,
      "loss": 5.0883,
      "step": 2104
    },
    {
      "epoch": 0.08498183286233346,
      "grad_norm": 126.62176513671875,
      "learning_rate": 1.984206566237484e-05,
      "loss": 4.7835,
      "step": 2105
    },
    {
      "epoch": 0.08502220427937021,
      "grad_norm": 97.17850494384766,
      "learning_rate": 1.9841834105353246e-05,
      "loss": 5.9586,
      "step": 2106
    },
    {
      "epoch": 0.08506257569640695,
      "grad_norm": 102.35943603515625,
      "learning_rate": 1.9841602380059267e-05,
      "loss": 7.1216,
      "step": 2107
    },
    {
      "epoch": 0.08510294711344368,
      "grad_norm": 104.65343475341797,
      "learning_rate": 1.9841370486496877e-05,
      "loss": 5.5258,
      "step": 2108
    },
    {
      "epoch": 0.08514331853048042,
      "grad_norm": 67.53961944580078,
      "learning_rate": 1.9841138424670027e-05,
      "loss": 4.9543,
      "step": 2109
    },
    {
      "epoch": 0.08518368994751715,
      "grad_norm": 319.4413146972656,
      "learning_rate": 1.984090619458269e-05,
      "loss": 5.2259,
      "step": 2110
    },
    {
      "epoch": 0.08522406136455389,
      "grad_norm": 116.8298110961914,
      "learning_rate": 1.9840673796238842e-05,
      "loss": 3.1006,
      "step": 2111
    },
    {
      "epoch": 0.08526443278159064,
      "grad_norm": 132.80560302734375,
      "learning_rate": 1.9840441229642446e-05,
      "loss": 5.9383,
      "step": 2112
    },
    {
      "epoch": 0.08530480419862738,
      "grad_norm": 118.22948455810547,
      "learning_rate": 1.984020849479749e-05,
      "loss": 5.5959,
      "step": 2113
    },
    {
      "epoch": 0.08534517561566411,
      "grad_norm": 110.34417724609375,
      "learning_rate": 1.983997559170794e-05,
      "loss": 5.6315,
      "step": 2114
    },
    {
      "epoch": 0.08538554703270085,
      "grad_norm": 102.61003875732422,
      "learning_rate": 1.9839742520377792e-05,
      "loss": 5.8043,
      "step": 2115
    },
    {
      "epoch": 0.08542591844973758,
      "grad_norm": 114.74486541748047,
      "learning_rate": 1.983950928081102e-05,
      "loss": 6.5242,
      "step": 2116
    },
    {
      "epoch": 0.08546628986677432,
      "grad_norm": 418.97930908203125,
      "learning_rate": 1.983927587301162e-05,
      "loss": 5.2764,
      "step": 2117
    },
    {
      "epoch": 0.08550666128381106,
      "grad_norm": 120.55967712402344,
      "learning_rate": 1.9839042296983577e-05,
      "loss": 5.0631,
      "step": 2118
    },
    {
      "epoch": 0.0855470327008478,
      "grad_norm": 121.4141616821289,
      "learning_rate": 1.9838808552730885e-05,
      "loss": 6.8166,
      "step": 2119
    },
    {
      "epoch": 0.08558740411788454,
      "grad_norm": 133.44818115234375,
      "learning_rate": 1.9838574640257543e-05,
      "loss": 4.3235,
      "step": 2120
    },
    {
      "epoch": 0.08562777553492128,
      "grad_norm": 163.26707458496094,
      "learning_rate": 1.983834055956755e-05,
      "loss": 5.7412,
      "step": 2121
    },
    {
      "epoch": 0.08566814695195801,
      "grad_norm": 137.24407958984375,
      "learning_rate": 1.9838106310664907e-05,
      "loss": 5.1376,
      "step": 2122
    },
    {
      "epoch": 0.08570851836899475,
      "grad_norm": 244.01173400878906,
      "learning_rate": 1.9837871893553622e-05,
      "loss": 6.0811,
      "step": 2123
    },
    {
      "epoch": 0.08574888978603148,
      "grad_norm": 112.28284454345703,
      "learning_rate": 1.9837637308237698e-05,
      "loss": 6.0921,
      "step": 2124
    },
    {
      "epoch": 0.08578926120306823,
      "grad_norm": 118.16293334960938,
      "learning_rate": 1.9837402554721152e-05,
      "loss": 6.2423,
      "step": 2125
    },
    {
      "epoch": 0.08582963262010497,
      "grad_norm": 97.93961334228516,
      "learning_rate": 1.983716763300799e-05,
      "loss": 6.2425,
      "step": 2126
    },
    {
      "epoch": 0.0858700040371417,
      "grad_norm": 94.20402526855469,
      "learning_rate": 1.9836932543102236e-05,
      "loss": 4.8274,
      "step": 2127
    },
    {
      "epoch": 0.08591037545417844,
      "grad_norm": 165.13906860351562,
      "learning_rate": 1.9836697285007906e-05,
      "loss": 5.0404,
      "step": 2128
    },
    {
      "epoch": 0.08595074687121518,
      "grad_norm": 192.66090393066406,
      "learning_rate": 1.9836461858729018e-05,
      "loss": 4.0823,
      "step": 2129
    },
    {
      "epoch": 0.08599111828825191,
      "grad_norm": 129.31454467773438,
      "learning_rate": 1.9836226264269607e-05,
      "loss": 5.3783,
      "step": 2130
    },
    {
      "epoch": 0.08603148970528865,
      "grad_norm": 136.17535400390625,
      "learning_rate": 1.9835990501633695e-05,
      "loss": 6.8655,
      "step": 2131
    },
    {
      "epoch": 0.0860718611223254,
      "grad_norm": 116.67044067382812,
      "learning_rate": 1.9835754570825317e-05,
      "loss": 3.5733,
      "step": 2132
    },
    {
      "epoch": 0.08611223253936214,
      "grad_norm": 122.43444061279297,
      "learning_rate": 1.9835518471848498e-05,
      "loss": 5.5005,
      "step": 2133
    },
    {
      "epoch": 0.08615260395639887,
      "grad_norm": 126.18335723876953,
      "learning_rate": 1.9835282204707288e-05,
      "loss": 6.01,
      "step": 2134
    },
    {
      "epoch": 0.0861929753734356,
      "grad_norm": 108.89359283447266,
      "learning_rate": 1.9835045769405714e-05,
      "loss": 6.32,
      "step": 2135
    },
    {
      "epoch": 0.08623334679047234,
      "grad_norm": 94.54615783691406,
      "learning_rate": 1.9834809165947825e-05,
      "loss": 6.3455,
      "step": 2136
    },
    {
      "epoch": 0.08627371820750908,
      "grad_norm": 96.77783203125,
      "learning_rate": 1.9834572394337664e-05,
      "loss": 4.3333,
      "step": 2137
    },
    {
      "epoch": 0.08631408962454583,
      "grad_norm": 110.4371337890625,
      "learning_rate": 1.9834335454579283e-05,
      "loss": 5.8226,
      "step": 2138
    },
    {
      "epoch": 0.08635446104158256,
      "grad_norm": 127.67227935791016,
      "learning_rate": 1.983409834667673e-05,
      "loss": 4.9945,
      "step": 2139
    },
    {
      "epoch": 0.0863948324586193,
      "grad_norm": 112.25853729248047,
      "learning_rate": 1.9833861070634057e-05,
      "loss": 6.2466,
      "step": 2140
    },
    {
      "epoch": 0.08643520387565604,
      "grad_norm": 238.8031463623047,
      "learning_rate": 1.9833623626455324e-05,
      "loss": 5.3973,
      "step": 2141
    },
    {
      "epoch": 0.08647557529269277,
      "grad_norm": 223.98178100585938,
      "learning_rate": 1.983338601414459e-05,
      "loss": 6.0458,
      "step": 2142
    },
    {
      "epoch": 0.08651594670972951,
      "grad_norm": 206.78073120117188,
      "learning_rate": 1.983314823370592e-05,
      "loss": 6.3419,
      "step": 2143
    },
    {
      "epoch": 0.08655631812676624,
      "grad_norm": 98.42935180664062,
      "learning_rate": 1.9832910285143378e-05,
      "loss": 4.4793,
      "step": 2144
    },
    {
      "epoch": 0.086596689543803,
      "grad_norm": 106.09574890136719,
      "learning_rate": 1.983267216846103e-05,
      "loss": 5.0112,
      "step": 2145
    },
    {
      "epoch": 0.08663706096083973,
      "grad_norm": 169.34805297851562,
      "learning_rate": 1.983243388366295e-05,
      "loss": 6.1414,
      "step": 2146
    },
    {
      "epoch": 0.08667743237787646,
      "grad_norm": 104.78734588623047,
      "learning_rate": 1.983219543075321e-05,
      "loss": 7.174,
      "step": 2147
    },
    {
      "epoch": 0.0867178037949132,
      "grad_norm": 81.46173095703125,
      "learning_rate": 1.9831956809735887e-05,
      "loss": 5.8269,
      "step": 2148
    },
    {
      "epoch": 0.08675817521194994,
      "grad_norm": 168.17015075683594,
      "learning_rate": 1.9831718020615062e-05,
      "loss": 6.4652,
      "step": 2149
    },
    {
      "epoch": 0.08679854662898667,
      "grad_norm": 228.36129760742188,
      "learning_rate": 1.9831479063394816e-05,
      "loss": 4.8284,
      "step": 2150
    },
    {
      "epoch": 0.08683891804602342,
      "grad_norm": 135.439697265625,
      "learning_rate": 1.9831239938079242e-05,
      "loss": 6.3548,
      "step": 2151
    },
    {
      "epoch": 0.08687928946306016,
      "grad_norm": 145.0295867919922,
      "learning_rate": 1.983100064467242e-05,
      "loss": 7.1955,
      "step": 2152
    },
    {
      "epoch": 0.0869196608800969,
      "grad_norm": 115.35787963867188,
      "learning_rate": 1.983076118317844e-05,
      "loss": 5.737,
      "step": 2153
    },
    {
      "epoch": 0.08696003229713363,
      "grad_norm": 115.9648208618164,
      "learning_rate": 1.9830521553601402e-05,
      "loss": 6.6026,
      "step": 2154
    },
    {
      "epoch": 0.08700040371417037,
      "grad_norm": 97.03329467773438,
      "learning_rate": 1.9830281755945403e-05,
      "loss": 5.8601,
      "step": 2155
    },
    {
      "epoch": 0.0870407751312071,
      "grad_norm": 83.0394058227539,
      "learning_rate": 1.983004179021454e-05,
      "loss": 6.5972,
      "step": 2156
    },
    {
      "epoch": 0.08708114654824384,
      "grad_norm": 81.39657592773438,
      "learning_rate": 1.982980165641292e-05,
      "loss": 5.653,
      "step": 2157
    },
    {
      "epoch": 0.08712151796528059,
      "grad_norm": 105.61841583251953,
      "learning_rate": 1.9829561354544645e-05,
      "loss": 5.915,
      "step": 2158
    },
    {
      "epoch": 0.08716188938231732,
      "grad_norm": 85.0416030883789,
      "learning_rate": 1.9829320884613822e-05,
      "loss": 5.1936,
      "step": 2159
    },
    {
      "epoch": 0.08720226079935406,
      "grad_norm": 113.2691879272461,
      "learning_rate": 1.9829080246624564e-05,
      "loss": 6.2419,
      "step": 2160
    },
    {
      "epoch": 0.0872426322163908,
      "grad_norm": 135.74826049804688,
      "learning_rate": 1.982883944058099e-05,
      "loss": 5.7102,
      "step": 2161
    },
    {
      "epoch": 0.08728300363342753,
      "grad_norm": 97.65250396728516,
      "learning_rate": 1.9828598466487214e-05,
      "loss": 4.5731,
      "step": 2162
    },
    {
      "epoch": 0.08732337505046427,
      "grad_norm": 110.77659606933594,
      "learning_rate": 1.9828357324347354e-05,
      "loss": 6.1647,
      "step": 2163
    },
    {
      "epoch": 0.08736374646750102,
      "grad_norm": 690.9628295898438,
      "learning_rate": 1.9828116014165536e-05,
      "loss": 5.9312,
      "step": 2164
    },
    {
      "epoch": 0.08740411788453775,
      "grad_norm": 96.14271545410156,
      "learning_rate": 1.982787453594588e-05,
      "loss": 6.5285,
      "step": 2165
    },
    {
      "epoch": 0.08744448930157449,
      "grad_norm": 102.8469009399414,
      "learning_rate": 1.9827632889692522e-05,
      "loss": 7.0329,
      "step": 2166
    },
    {
      "epoch": 0.08748486071861122,
      "grad_norm": 123.62963104248047,
      "learning_rate": 1.9827391075409593e-05,
      "loss": 5.0605,
      "step": 2167
    },
    {
      "epoch": 0.08752523213564796,
      "grad_norm": 104.00119018554688,
      "learning_rate": 1.9827149093101223e-05,
      "loss": 5.6811,
      "step": 2168
    },
    {
      "epoch": 0.0875656035526847,
      "grad_norm": 98.93212127685547,
      "learning_rate": 1.982690694277155e-05,
      "loss": 6.2134,
      "step": 2169
    },
    {
      "epoch": 0.08760597496972143,
      "grad_norm": 176.06134033203125,
      "learning_rate": 1.982666462442472e-05,
      "loss": 5.9844,
      "step": 2170
    },
    {
      "epoch": 0.08764634638675818,
      "grad_norm": 127.60376739501953,
      "learning_rate": 1.9826422138064867e-05,
      "loss": 5.3665,
      "step": 2171
    },
    {
      "epoch": 0.08768671780379492,
      "grad_norm": 95.46189880371094,
      "learning_rate": 1.9826179483696144e-05,
      "loss": 5.442,
      "step": 2172
    },
    {
      "epoch": 0.08772708922083165,
      "grad_norm": 97.60729217529297,
      "learning_rate": 1.98259366613227e-05,
      "loss": 5.8151,
      "step": 2173
    },
    {
      "epoch": 0.08776746063786839,
      "grad_norm": 117.57756042480469,
      "learning_rate": 1.9825693670948682e-05,
      "loss": 7.0433,
      "step": 2174
    },
    {
      "epoch": 0.08780783205490512,
      "grad_norm": 136.76663208007812,
      "learning_rate": 1.9825450512578246e-05,
      "loss": 5.515,
      "step": 2175
    },
    {
      "epoch": 0.08784820347194186,
      "grad_norm": 109.28760528564453,
      "learning_rate": 1.9825207186215552e-05,
      "loss": 5.7862,
      "step": 2176
    },
    {
      "epoch": 0.08788857488897861,
      "grad_norm": 105.21770477294922,
      "learning_rate": 1.9824963691864758e-05,
      "loss": 5.1765,
      "step": 2177
    },
    {
      "epoch": 0.08792894630601535,
      "grad_norm": 107.5025405883789,
      "learning_rate": 1.9824720029530026e-05,
      "loss": 5.2538,
      "step": 2178
    },
    {
      "epoch": 0.08796931772305208,
      "grad_norm": 113.113037109375,
      "learning_rate": 1.9824476199215526e-05,
      "loss": 5.1642,
      "step": 2179
    },
    {
      "epoch": 0.08800968914008882,
      "grad_norm": 90.41448974609375,
      "learning_rate": 1.9824232200925424e-05,
      "loss": 4.628,
      "step": 2180
    },
    {
      "epoch": 0.08805006055712555,
      "grad_norm": 182.5604705810547,
      "learning_rate": 1.9823988034663894e-05,
      "loss": 6.6515,
      "step": 2181
    },
    {
      "epoch": 0.08809043197416229,
      "grad_norm": 140.58477783203125,
      "learning_rate": 1.9823743700435107e-05,
      "loss": 3.6787,
      "step": 2182
    },
    {
      "epoch": 0.08813080339119903,
      "grad_norm": 222.0042266845703,
      "learning_rate": 1.9823499198243245e-05,
      "loss": 4.9783,
      "step": 2183
    },
    {
      "epoch": 0.08817117480823577,
      "grad_norm": 115.29925537109375,
      "learning_rate": 1.9823254528092484e-05,
      "loss": 5.4462,
      "step": 2184
    },
    {
      "epoch": 0.08821154622527251,
      "grad_norm": 92.5673599243164,
      "learning_rate": 1.9823009689987012e-05,
      "loss": 4.4486,
      "step": 2185
    },
    {
      "epoch": 0.08825191764230925,
      "grad_norm": 125.40091705322266,
      "learning_rate": 1.982276468393101e-05,
      "loss": 6.6158,
      "step": 2186
    },
    {
      "epoch": 0.08829228905934598,
      "grad_norm": 139.9507293701172,
      "learning_rate": 1.982251950992867e-05,
      "loss": 7.8213,
      "step": 2187
    },
    {
      "epoch": 0.08833266047638272,
      "grad_norm": 142.94012451171875,
      "learning_rate": 1.9822274167984188e-05,
      "loss": 6.2365,
      "step": 2188
    },
    {
      "epoch": 0.08837303189341945,
      "grad_norm": 113.96045684814453,
      "learning_rate": 1.982202865810175e-05,
      "loss": 6.4953,
      "step": 2189
    },
    {
      "epoch": 0.0884134033104562,
      "grad_norm": 254.0894775390625,
      "learning_rate": 1.982178298028556e-05,
      "loss": 4.4431,
      "step": 2190
    },
    {
      "epoch": 0.08845377472749294,
      "grad_norm": 145.3498992919922,
      "learning_rate": 1.9821537134539818e-05,
      "loss": 5.8581,
      "step": 2191
    },
    {
      "epoch": 0.08849414614452968,
      "grad_norm": 131.11355590820312,
      "learning_rate": 1.982129112086872e-05,
      "loss": 6.5991,
      "step": 2192
    },
    {
      "epoch": 0.08853451756156641,
      "grad_norm": 116.85620880126953,
      "learning_rate": 1.9821044939276484e-05,
      "loss": 5.6086,
      "step": 2193
    },
    {
      "epoch": 0.08857488897860315,
      "grad_norm": 116.54145050048828,
      "learning_rate": 1.982079858976731e-05,
      "loss": 4.7746,
      "step": 2194
    },
    {
      "epoch": 0.08861526039563988,
      "grad_norm": 210.90591430664062,
      "learning_rate": 1.9820552072345416e-05,
      "loss": 7.0013,
      "step": 2195
    },
    {
      "epoch": 0.08865563181267662,
      "grad_norm": 213.26455688476562,
      "learning_rate": 1.9820305387015013e-05,
      "loss": 3.8589,
      "step": 2196
    },
    {
      "epoch": 0.08869600322971337,
      "grad_norm": 112.44513702392578,
      "learning_rate": 1.982005853378032e-05,
      "loss": 6.108,
      "step": 2197
    },
    {
      "epoch": 0.0887363746467501,
      "grad_norm": 100.97592163085938,
      "learning_rate": 1.9819811512645556e-05,
      "loss": 6.3751,
      "step": 2198
    },
    {
      "epoch": 0.08877674606378684,
      "grad_norm": 78.61543273925781,
      "learning_rate": 1.9819564323614945e-05,
      "loss": 5.7832,
      "step": 2199
    },
    {
      "epoch": 0.08881711748082358,
      "grad_norm": 81.277099609375,
      "learning_rate": 1.9819316966692718e-05,
      "loss": 6.2594,
      "step": 2200
    },
    {
      "epoch": 0.08885748889786031,
      "grad_norm": 75.14210510253906,
      "learning_rate": 1.98190694418831e-05,
      "loss": 7.1431,
      "step": 2201
    },
    {
      "epoch": 0.08889786031489705,
      "grad_norm": 103.13385772705078,
      "learning_rate": 1.9818821749190323e-05,
      "loss": 4.2039,
      "step": 2202
    },
    {
      "epoch": 0.08893823173193378,
      "grad_norm": 127.22088623046875,
      "learning_rate": 1.9818573888618622e-05,
      "loss": 6.7284,
      "step": 2203
    },
    {
      "epoch": 0.08897860314897053,
      "grad_norm": 125.745361328125,
      "learning_rate": 1.9818325860172236e-05,
      "loss": 5.7255,
      "step": 2204
    },
    {
      "epoch": 0.08901897456600727,
      "grad_norm": 100.33792114257812,
      "learning_rate": 1.9818077663855408e-05,
      "loss": 5.8442,
      "step": 2205
    },
    {
      "epoch": 0.089059345983044,
      "grad_norm": 139.68695068359375,
      "learning_rate": 1.9817829299672374e-05,
      "loss": 5.9261,
      "step": 2206
    },
    {
      "epoch": 0.08909971740008074,
      "grad_norm": 116.30037689208984,
      "learning_rate": 1.9817580767627386e-05,
      "loss": 5.2551,
      "step": 2207
    },
    {
      "epoch": 0.08914008881711748,
      "grad_norm": 128.7925262451172,
      "learning_rate": 1.9817332067724697e-05,
      "loss": 7.3125,
      "step": 2208
    },
    {
      "epoch": 0.08918046023415421,
      "grad_norm": 217.42247009277344,
      "learning_rate": 1.9817083199968552e-05,
      "loss": 5.4932,
      "step": 2209
    },
    {
      "epoch": 0.08922083165119096,
      "grad_norm": 103.61231231689453,
      "learning_rate": 1.981683416436321e-05,
      "loss": 5.4941,
      "step": 2210
    },
    {
      "epoch": 0.0892612030682277,
      "grad_norm": 75.55675506591797,
      "learning_rate": 1.9816584960912924e-05,
      "loss": 6.2373,
      "step": 2211
    },
    {
      "epoch": 0.08930157448526443,
      "grad_norm": 141.7594451904297,
      "learning_rate": 1.9816335589621963e-05,
      "loss": 4.6426,
      "step": 2212
    },
    {
      "epoch": 0.08934194590230117,
      "grad_norm": 99.74600219726562,
      "learning_rate": 1.981608605049458e-05,
      "loss": 5.7491,
      "step": 2213
    },
    {
      "epoch": 0.0893823173193379,
      "grad_norm": 107.24551391601562,
      "learning_rate": 1.9815836343535055e-05,
      "loss": 5.5148,
      "step": 2214
    },
    {
      "epoch": 0.08942268873637464,
      "grad_norm": 74.61307525634766,
      "learning_rate": 1.981558646874765e-05,
      "loss": 5.1297,
      "step": 2215
    },
    {
      "epoch": 0.08946306015341138,
      "grad_norm": 85.4831314086914,
      "learning_rate": 1.981533642613663e-05,
      "loss": 6.9106,
      "step": 2216
    },
    {
      "epoch": 0.08950343157044813,
      "grad_norm": 131.80587768554688,
      "learning_rate": 1.9815086215706282e-05,
      "loss": 4.9977,
      "step": 2217
    },
    {
      "epoch": 0.08954380298748486,
      "grad_norm": 107.43572998046875,
      "learning_rate": 1.981483583746088e-05,
      "loss": 4.7337,
      "step": 2218
    },
    {
      "epoch": 0.0895841744045216,
      "grad_norm": 121.53164672851562,
      "learning_rate": 1.9814585291404705e-05,
      "loss": 5.8451,
      "step": 2219
    },
    {
      "epoch": 0.08962454582155834,
      "grad_norm": 65.402587890625,
      "learning_rate": 1.981433457754204e-05,
      "loss": 6.1885,
      "step": 2220
    },
    {
      "epoch": 0.08966491723859507,
      "grad_norm": 103.09114837646484,
      "learning_rate": 1.981408369587717e-05,
      "loss": 4.5865,
      "step": 2221
    },
    {
      "epoch": 0.0897052886556318,
      "grad_norm": 117.43621063232422,
      "learning_rate": 1.9813832646414387e-05,
      "loss": 5.7383,
      "step": 2222
    },
    {
      "epoch": 0.08974566007266856,
      "grad_norm": 121.44063568115234,
      "learning_rate": 1.9813581429157985e-05,
      "loss": 5.4397,
      "step": 2223
    },
    {
      "epoch": 0.08978603148970529,
      "grad_norm": 117.38172149658203,
      "learning_rate": 1.9813330044112257e-05,
      "loss": 5.3304,
      "step": 2224
    },
    {
      "epoch": 0.08982640290674203,
      "grad_norm": 190.71731567382812,
      "learning_rate": 1.9813078491281496e-05,
      "loss": 5.9187,
      "step": 2225
    },
    {
      "epoch": 0.08986677432377876,
      "grad_norm": 93.0138931274414,
      "learning_rate": 1.9812826770670012e-05,
      "loss": 5.8654,
      "step": 2226
    },
    {
      "epoch": 0.0899071457408155,
      "grad_norm": 114.99740600585938,
      "learning_rate": 1.9812574882282104e-05,
      "loss": 4.0452,
      "step": 2227
    },
    {
      "epoch": 0.08994751715785224,
      "grad_norm": 139.27610778808594,
      "learning_rate": 1.981232282612208e-05,
      "loss": 5.6433,
      "step": 2228
    },
    {
      "epoch": 0.08998788857488897,
      "grad_norm": 115.88105010986328,
      "learning_rate": 1.9812070602194246e-05,
      "loss": 6.3219,
      "step": 2229
    },
    {
      "epoch": 0.09002825999192572,
      "grad_norm": 96.26661682128906,
      "learning_rate": 1.9811818210502922e-05,
      "loss": 5.4271,
      "step": 2230
    },
    {
      "epoch": 0.09006863140896246,
      "grad_norm": 104.62599182128906,
      "learning_rate": 1.9811565651052416e-05,
      "loss": 5.8176,
      "step": 2231
    },
    {
      "epoch": 0.0901090028259992,
      "grad_norm": 87.36883544921875,
      "learning_rate": 1.981131292384705e-05,
      "loss": 5.8695,
      "step": 2232
    },
    {
      "epoch": 0.09014937424303593,
      "grad_norm": 84.41908264160156,
      "learning_rate": 1.9811060028891143e-05,
      "loss": 5.163,
      "step": 2233
    },
    {
      "epoch": 0.09018974566007266,
      "grad_norm": 91.57876586914062,
      "learning_rate": 1.9810806966189022e-05,
      "loss": 7.0007,
      "step": 2234
    },
    {
      "epoch": 0.0902301170771094,
      "grad_norm": 110.20092010498047,
      "learning_rate": 1.9810553735745013e-05,
      "loss": 4.9794,
      "step": 2235
    },
    {
      "epoch": 0.09027048849414615,
      "grad_norm": 179.509033203125,
      "learning_rate": 1.9810300337563442e-05,
      "loss": 5.2045,
      "step": 2236
    },
    {
      "epoch": 0.09031085991118289,
      "grad_norm": 100.25807189941406,
      "learning_rate": 1.9810046771648642e-05,
      "loss": 5.2352,
      "step": 2237
    },
    {
      "epoch": 0.09035123132821962,
      "grad_norm": 258.3427429199219,
      "learning_rate": 1.980979303800495e-05,
      "loss": 5.9712,
      "step": 2238
    },
    {
      "epoch": 0.09039160274525636,
      "grad_norm": 170.7792510986328,
      "learning_rate": 1.980953913663671e-05,
      "loss": 6.1068,
      "step": 2239
    },
    {
      "epoch": 0.0904319741622931,
      "grad_norm": 100.12876892089844,
      "learning_rate": 1.9809285067548254e-05,
      "loss": 5.9651,
      "step": 2240
    },
    {
      "epoch": 0.09047234557932983,
      "grad_norm": 140.2744598388672,
      "learning_rate": 1.980903083074393e-05,
      "loss": 4.7334,
      "step": 2241
    },
    {
      "epoch": 0.09051271699636657,
      "grad_norm": 293.11669921875,
      "learning_rate": 1.9808776426228084e-05,
      "loss": 5.7218,
      "step": 2242
    },
    {
      "epoch": 0.09055308841340332,
      "grad_norm": 134.39291381835938,
      "learning_rate": 1.9808521854005068e-05,
      "loss": 7.2991,
      "step": 2243
    },
    {
      "epoch": 0.09059345983044005,
      "grad_norm": 117.8031005859375,
      "learning_rate": 1.9808267114079235e-05,
      "loss": 4.7787,
      "step": 2244
    },
    {
      "epoch": 0.09063383124747679,
      "grad_norm": 143.49578857421875,
      "learning_rate": 1.9808012206454933e-05,
      "loss": 5.4903,
      "step": 2245
    },
    {
      "epoch": 0.09067420266451352,
      "grad_norm": 121.48588562011719,
      "learning_rate": 1.9807757131136526e-05,
      "loss": 4.9836,
      "step": 2246
    },
    {
      "epoch": 0.09071457408155026,
      "grad_norm": 188.20591735839844,
      "learning_rate": 1.980750188812838e-05,
      "loss": 7.0584,
      "step": 2247
    },
    {
      "epoch": 0.090754945498587,
      "grad_norm": 100.95034790039062,
      "learning_rate": 1.980724647743485e-05,
      "loss": 4.4826,
      "step": 2248
    },
    {
      "epoch": 0.09079531691562374,
      "grad_norm": 111.08976745605469,
      "learning_rate": 1.9806990899060307e-05,
      "loss": 5.3248,
      "step": 2249
    },
    {
      "epoch": 0.09083568833266048,
      "grad_norm": 173.67108154296875,
      "learning_rate": 1.9806735153009123e-05,
      "loss": 5.3545,
      "step": 2250
    },
    {
      "epoch": 0.09087605974969722,
      "grad_norm": 136.57228088378906,
      "learning_rate": 1.9806479239285666e-05,
      "loss": 6.9854,
      "step": 2251
    },
    {
      "epoch": 0.09091643116673395,
      "grad_norm": 149.18807983398438,
      "learning_rate": 1.980622315789431e-05,
      "loss": 7.2004,
      "step": 2252
    },
    {
      "epoch": 0.09095680258377069,
      "grad_norm": 106.90599822998047,
      "learning_rate": 1.9805966908839444e-05,
      "loss": 4.6693,
      "step": 2253
    },
    {
      "epoch": 0.09099717400080742,
      "grad_norm": 104.22730255126953,
      "learning_rate": 1.9805710492125437e-05,
      "loss": 4.5897,
      "step": 2254
    },
    {
      "epoch": 0.09103754541784416,
      "grad_norm": 132.14390563964844,
      "learning_rate": 1.9805453907756682e-05,
      "loss": 5.7935,
      "step": 2255
    },
    {
      "epoch": 0.09107791683488091,
      "grad_norm": 159.92652893066406,
      "learning_rate": 1.980519715573756e-05,
      "loss": 5.6223,
      "step": 2256
    },
    {
      "epoch": 0.09111828825191765,
      "grad_norm": 122.42423248291016,
      "learning_rate": 1.9804940236072467e-05,
      "loss": 4.548,
      "step": 2257
    },
    {
      "epoch": 0.09115865966895438,
      "grad_norm": 124.39949035644531,
      "learning_rate": 1.980468314876579e-05,
      "loss": 6.1224,
      "step": 2258
    },
    {
      "epoch": 0.09119903108599112,
      "grad_norm": 185.00289916992188,
      "learning_rate": 1.9804425893821927e-05,
      "loss": 5.1866,
      "step": 2259
    },
    {
      "epoch": 0.09123940250302785,
      "grad_norm": 170.29408264160156,
      "learning_rate": 1.9804168471245278e-05,
      "loss": 6.7437,
      "step": 2260
    },
    {
      "epoch": 0.09127977392006459,
      "grad_norm": 127.19229888916016,
      "learning_rate": 1.9803910881040242e-05,
      "loss": 6.9475,
      "step": 2261
    },
    {
      "epoch": 0.09132014533710134,
      "grad_norm": 94.51419067382812,
      "learning_rate": 1.9803653123211218e-05,
      "loss": 5.9425,
      "step": 2262
    },
    {
      "epoch": 0.09136051675413807,
      "grad_norm": 111.17094421386719,
      "learning_rate": 1.9803395197762624e-05,
      "loss": 5.8096,
      "step": 2263
    },
    {
      "epoch": 0.09140088817117481,
      "grad_norm": 134.45516967773438,
      "learning_rate": 1.9803137104698868e-05,
      "loss": 5.7529,
      "step": 2264
    },
    {
      "epoch": 0.09144125958821155,
      "grad_norm": 155.88006591796875,
      "learning_rate": 1.9802878844024354e-05,
      "loss": 7.0427,
      "step": 2265
    },
    {
      "epoch": 0.09148163100524828,
      "grad_norm": 117.53276062011719,
      "learning_rate": 1.9802620415743507e-05,
      "loss": 5.4529,
      "step": 2266
    },
    {
      "epoch": 0.09152200242228502,
      "grad_norm": 146.3871612548828,
      "learning_rate": 1.9802361819860735e-05,
      "loss": 5.8987,
      "step": 2267
    },
    {
      "epoch": 0.09156237383932175,
      "grad_norm": 97.60474395751953,
      "learning_rate": 1.980210305638047e-05,
      "loss": 6.6239,
      "step": 2268
    },
    {
      "epoch": 0.0916027452563585,
      "grad_norm": 144.7122039794922,
      "learning_rate": 1.980184412530713e-05,
      "loss": 5.9369,
      "step": 2269
    },
    {
      "epoch": 0.09164311667339524,
      "grad_norm": 323.5594177246094,
      "learning_rate": 1.9801585026645152e-05,
      "loss": 6.2135,
      "step": 2270
    },
    {
      "epoch": 0.09168348809043197,
      "grad_norm": 171.6890106201172,
      "learning_rate": 1.9801325760398954e-05,
      "loss": 4.9487,
      "step": 2271
    },
    {
      "epoch": 0.09172385950746871,
      "grad_norm": 233.91119384765625,
      "learning_rate": 1.980106632657297e-05,
      "loss": 4.947,
      "step": 2272
    },
    {
      "epoch": 0.09176423092450545,
      "grad_norm": 129.61575317382812,
      "learning_rate": 1.9800806725171644e-05,
      "loss": 5.0776,
      "step": 2273
    },
    {
      "epoch": 0.09180460234154218,
      "grad_norm": 162.14263916015625,
      "learning_rate": 1.9800546956199404e-05,
      "loss": 5.2519,
      "step": 2274
    },
    {
      "epoch": 0.09184497375857893,
      "grad_norm": 132.58631896972656,
      "learning_rate": 1.98002870196607e-05,
      "loss": 5.5727,
      "step": 2275
    },
    {
      "epoch": 0.09188534517561567,
      "grad_norm": 110.25311279296875,
      "learning_rate": 1.9800026915559974e-05,
      "loss": 5.4968,
      "step": 2276
    },
    {
      "epoch": 0.0919257165926524,
      "grad_norm": 113.9887924194336,
      "learning_rate": 1.9799766643901673e-05,
      "loss": 6.1204,
      "step": 2277
    },
    {
      "epoch": 0.09196608800968914,
      "grad_norm": 95.31875610351562,
      "learning_rate": 1.9799506204690244e-05,
      "loss": 6.6676,
      "step": 2278
    },
    {
      "epoch": 0.09200645942672588,
      "grad_norm": 134.39259338378906,
      "learning_rate": 1.9799245597930144e-05,
      "loss": 4.8674,
      "step": 2279
    },
    {
      "epoch": 0.09204683084376261,
      "grad_norm": 96.806396484375,
      "learning_rate": 1.979898482362583e-05,
      "loss": 4.6168,
      "step": 2280
    },
    {
      "epoch": 0.09208720226079935,
      "grad_norm": 89.37568664550781,
      "learning_rate": 1.9798723881781753e-05,
      "loss": 5.2597,
      "step": 2281
    },
    {
      "epoch": 0.0921275736778361,
      "grad_norm": 81.1866226196289,
      "learning_rate": 1.979846277240238e-05,
      "loss": 6.2249,
      "step": 2282
    },
    {
      "epoch": 0.09216794509487283,
      "grad_norm": 140.55853271484375,
      "learning_rate": 1.9798201495492176e-05,
      "loss": 4.1801,
      "step": 2283
    },
    {
      "epoch": 0.09220831651190957,
      "grad_norm": 112.84611511230469,
      "learning_rate": 1.9797940051055607e-05,
      "loss": 4.9167,
      "step": 2284
    },
    {
      "epoch": 0.0922486879289463,
      "grad_norm": 121.91302490234375,
      "learning_rate": 1.979767843909714e-05,
      "loss": 5.1107,
      "step": 2285
    },
    {
      "epoch": 0.09228905934598304,
      "grad_norm": 95.9238052368164,
      "learning_rate": 1.9797416659621255e-05,
      "loss": 7.3513,
      "step": 2286
    },
    {
      "epoch": 0.09232943076301978,
      "grad_norm": 84.1207504272461,
      "learning_rate": 1.9797154712632422e-05,
      "loss": 5.9055,
      "step": 2287
    },
    {
      "epoch": 0.09236980218005653,
      "grad_norm": 108.49308013916016,
      "learning_rate": 1.9796892598135123e-05,
      "loss": 5.3336,
      "step": 2288
    },
    {
      "epoch": 0.09241017359709326,
      "grad_norm": 168.24794006347656,
      "learning_rate": 1.9796630316133836e-05,
      "loss": 5.908,
      "step": 2289
    },
    {
      "epoch": 0.09245054501413,
      "grad_norm": 138.62802124023438,
      "learning_rate": 1.9796367866633044e-05,
      "loss": 5.2855,
      "step": 2290
    },
    {
      "epoch": 0.09249091643116673,
      "grad_norm": 121.47095489501953,
      "learning_rate": 1.9796105249637244e-05,
      "loss": 6.0461,
      "step": 2291
    },
    {
      "epoch": 0.09253128784820347,
      "grad_norm": 112.92188262939453,
      "learning_rate": 1.9795842465150917e-05,
      "loss": 3.7215,
      "step": 2292
    },
    {
      "epoch": 0.0925716592652402,
      "grad_norm": 115.9825668334961,
      "learning_rate": 1.979557951317856e-05,
      "loss": 4.2446,
      "step": 2293
    },
    {
      "epoch": 0.09261203068227694,
      "grad_norm": 114.32170867919922,
      "learning_rate": 1.9795316393724665e-05,
      "loss": 6.1851,
      "step": 2294
    },
    {
      "epoch": 0.09265240209931369,
      "grad_norm": 115.94661712646484,
      "learning_rate": 1.9795053106793735e-05,
      "loss": 7.2886,
      "step": 2295
    },
    {
      "epoch": 0.09269277351635043,
      "grad_norm": 111.6320571899414,
      "learning_rate": 1.979478965239027e-05,
      "loss": 4.7347,
      "step": 2296
    },
    {
      "epoch": 0.09273314493338716,
      "grad_norm": 101.70513153076172,
      "learning_rate": 1.9794526030518776e-05,
      "loss": 7.7326,
      "step": 2297
    },
    {
      "epoch": 0.0927735163504239,
      "grad_norm": 120.6712417602539,
      "learning_rate": 1.9794262241183755e-05,
      "loss": 6.0408,
      "step": 2298
    },
    {
      "epoch": 0.09281388776746063,
      "grad_norm": 107.38922119140625,
      "learning_rate": 1.9793998284389725e-05,
      "loss": 4.9071,
      "step": 2299
    },
    {
      "epoch": 0.09285425918449737,
      "grad_norm": 108.10201263427734,
      "learning_rate": 1.979373416014119e-05,
      "loss": 4.7562,
      "step": 2300
    },
    {
      "epoch": 0.09289463060153412,
      "grad_norm": 119.29657745361328,
      "learning_rate": 1.9793469868442675e-05,
      "loss": 4.1135,
      "step": 2301
    },
    {
      "epoch": 0.09293500201857086,
      "grad_norm": 90.95012664794922,
      "learning_rate": 1.9793205409298696e-05,
      "loss": 5.4667,
      "step": 2302
    },
    {
      "epoch": 0.09297537343560759,
      "grad_norm": 107.23379516601562,
      "learning_rate": 1.979294078271377e-05,
      "loss": 3.8368,
      "step": 2303
    },
    {
      "epoch": 0.09301574485264433,
      "grad_norm": 90.52338409423828,
      "learning_rate": 1.9792675988692424e-05,
      "loss": 6.0207,
      "step": 2304
    },
    {
      "epoch": 0.09305611626968106,
      "grad_norm": 106.52205657958984,
      "learning_rate": 1.9792411027239188e-05,
      "loss": 5.1558,
      "step": 2305
    },
    {
      "epoch": 0.0930964876867178,
      "grad_norm": 104.22769165039062,
      "learning_rate": 1.9792145898358593e-05,
      "loss": 6.69,
      "step": 2306
    },
    {
      "epoch": 0.09313685910375454,
      "grad_norm": 104.35884857177734,
      "learning_rate": 1.9791880602055164e-05,
      "loss": 7.2014,
      "step": 2307
    },
    {
      "epoch": 0.09317723052079128,
      "grad_norm": 99.13348388671875,
      "learning_rate": 1.9791615138333443e-05,
      "loss": 6.7892,
      "step": 2308
    },
    {
      "epoch": 0.09321760193782802,
      "grad_norm": 96.70148468017578,
      "learning_rate": 1.979134950719797e-05,
      "loss": 5.4588,
      "step": 2309
    },
    {
      "epoch": 0.09325797335486476,
      "grad_norm": 96.77464294433594,
      "learning_rate": 1.9791083708653288e-05,
      "loss": 6.0113,
      "step": 2310
    },
    {
      "epoch": 0.09329834477190149,
      "grad_norm": 76.1173324584961,
      "learning_rate": 1.9790817742703937e-05,
      "loss": 5.5627,
      "step": 2311
    },
    {
      "epoch": 0.09333871618893823,
      "grad_norm": 92.93948364257812,
      "learning_rate": 1.979055160935446e-05,
      "loss": 6.3001,
      "step": 2312
    },
    {
      "epoch": 0.09337908760597496,
      "grad_norm": 92.49947357177734,
      "learning_rate": 1.979028530860942e-05,
      "loss": 5.4663,
      "step": 2313
    },
    {
      "epoch": 0.09341945902301171,
      "grad_norm": 101.43351745605469,
      "learning_rate": 1.979001884047336e-05,
      "loss": 5.1167,
      "step": 2314
    },
    {
      "epoch": 0.09345983044004845,
      "grad_norm": 101.66634368896484,
      "learning_rate": 1.978975220495084e-05,
      "loss": 6.6364,
      "step": 2315
    },
    {
      "epoch": 0.09350020185708519,
      "grad_norm": 115.13179779052734,
      "learning_rate": 1.9789485402046414e-05,
      "loss": 5.6122,
      "step": 2316
    },
    {
      "epoch": 0.09354057327412192,
      "grad_norm": 99.44349670410156,
      "learning_rate": 1.9789218431764653e-05,
      "loss": 5.484,
      "step": 2317
    },
    {
      "epoch": 0.09358094469115866,
      "grad_norm": 135.696044921875,
      "learning_rate": 1.9788951294110116e-05,
      "loss": 6.0433,
      "step": 2318
    },
    {
      "epoch": 0.0936213161081954,
      "grad_norm": 113.94100952148438,
      "learning_rate": 1.9788683989087368e-05,
      "loss": 4.9364,
      "step": 2319
    },
    {
      "epoch": 0.09366168752523213,
      "grad_norm": 208.2609405517578,
      "learning_rate": 1.9788416516700986e-05,
      "loss": 6.2237,
      "step": 2320
    },
    {
      "epoch": 0.09370205894226888,
      "grad_norm": 144.86404418945312,
      "learning_rate": 1.9788148876955534e-05,
      "loss": 5.8936,
      "step": 2321
    },
    {
      "epoch": 0.09374243035930561,
      "grad_norm": 93.72992706298828,
      "learning_rate": 1.9787881069855595e-05,
      "loss": 7.4891,
      "step": 2322
    },
    {
      "epoch": 0.09378280177634235,
      "grad_norm": 94.86470794677734,
      "learning_rate": 1.9787613095405748e-05,
      "loss": 6.0105,
      "step": 2323
    },
    {
      "epoch": 0.09382317319337909,
      "grad_norm": 147.3050079345703,
      "learning_rate": 1.9787344953610573e-05,
      "loss": 7.1521,
      "step": 2324
    },
    {
      "epoch": 0.09386354461041582,
      "grad_norm": 87.63600158691406,
      "learning_rate": 1.978707664447465e-05,
      "loss": 4.4435,
      "step": 2325
    },
    {
      "epoch": 0.09390391602745256,
      "grad_norm": 153.81283569335938,
      "learning_rate": 1.9786808168002575e-05,
      "loss": 5.1712,
      "step": 2326
    },
    {
      "epoch": 0.09394428744448931,
      "grad_norm": 84.62300872802734,
      "learning_rate": 1.9786539524198934e-05,
      "loss": 5.4592,
      "step": 2327
    },
    {
      "epoch": 0.09398465886152604,
      "grad_norm": 95.78121948242188,
      "learning_rate": 1.978627071306832e-05,
      "loss": 6.2378,
      "step": 2328
    },
    {
      "epoch": 0.09402503027856278,
      "grad_norm": 112.19807434082031,
      "learning_rate": 1.9786001734615327e-05,
      "loss": 5.4804,
      "step": 2329
    },
    {
      "epoch": 0.09406540169559952,
      "grad_norm": 93.00503540039062,
      "learning_rate": 1.978573258884456e-05,
      "loss": 4.5626,
      "step": 2330
    },
    {
      "epoch": 0.09410577311263625,
      "grad_norm": 171.38075256347656,
      "learning_rate": 1.9785463275760613e-05,
      "loss": 5.7745,
      "step": 2331
    },
    {
      "epoch": 0.09414614452967299,
      "grad_norm": 96.51007843017578,
      "learning_rate": 1.9785193795368097e-05,
      "loss": 5.3418,
      "step": 2332
    },
    {
      "epoch": 0.09418651594670972,
      "grad_norm": 79.57819366455078,
      "learning_rate": 1.978492414767162e-05,
      "loss": 5.0632,
      "step": 2333
    },
    {
      "epoch": 0.09422688736374647,
      "grad_norm": 251.2281494140625,
      "learning_rate": 1.9784654332675783e-05,
      "loss": 4.3337,
      "step": 2334
    },
    {
      "epoch": 0.09426725878078321,
      "grad_norm": 88.21846771240234,
      "learning_rate": 1.978438435038521e-05,
      "loss": 5.6304,
      "step": 2335
    },
    {
      "epoch": 0.09430763019781994,
      "grad_norm": 380.225341796875,
      "learning_rate": 1.9784114200804513e-05,
      "loss": 5.9152,
      "step": 2336
    },
    {
      "epoch": 0.09434800161485668,
      "grad_norm": 111.36785888671875,
      "learning_rate": 1.9783843883938308e-05,
      "loss": 5.751,
      "step": 2337
    },
    {
      "epoch": 0.09438837303189342,
      "grad_norm": 91.47245025634766,
      "learning_rate": 1.9783573399791224e-05,
      "loss": 5.1147,
      "step": 2338
    },
    {
      "epoch": 0.09442874444893015,
      "grad_norm": 136.1172332763672,
      "learning_rate": 1.9783302748367873e-05,
      "loss": 8.1242,
      "step": 2339
    },
    {
      "epoch": 0.0944691158659669,
      "grad_norm": 116.3007583618164,
      "learning_rate": 1.97830319296729e-05,
      "loss": 4.981,
      "step": 2340
    },
    {
      "epoch": 0.09450948728300364,
      "grad_norm": 233.16470336914062,
      "learning_rate": 1.978276094371092e-05,
      "loss": 6.5625,
      "step": 2341
    },
    {
      "epoch": 0.09454985870004037,
      "grad_norm": 135.47801208496094,
      "learning_rate": 1.9782489790486575e-05,
      "loss": 5.9647,
      "step": 2342
    },
    {
      "epoch": 0.09459023011707711,
      "grad_norm": 165.12081909179688,
      "learning_rate": 1.9782218470004496e-05,
      "loss": 6.0523,
      "step": 2343
    },
    {
      "epoch": 0.09463060153411385,
      "grad_norm": 90.49543762207031,
      "learning_rate": 1.9781946982269327e-05,
      "loss": 5.0558,
      "step": 2344
    },
    {
      "epoch": 0.09467097295115058,
      "grad_norm": 183.75450134277344,
      "learning_rate": 1.9781675327285707e-05,
      "loss": 6.4558,
      "step": 2345
    },
    {
      "epoch": 0.09471134436818732,
      "grad_norm": 90.9228744506836,
      "learning_rate": 1.9781403505058277e-05,
      "loss": 5.227,
      "step": 2346
    },
    {
      "epoch": 0.09475171578522407,
      "grad_norm": 116.44355773925781,
      "learning_rate": 1.9781131515591693e-05,
      "loss": 4.5464,
      "step": 2347
    },
    {
      "epoch": 0.0947920872022608,
      "grad_norm": 98.64592742919922,
      "learning_rate": 1.9780859358890597e-05,
      "loss": 4.0013,
      "step": 2348
    },
    {
      "epoch": 0.09483245861929754,
      "grad_norm": 112.28182220458984,
      "learning_rate": 1.9780587034959648e-05,
      "loss": 5.3745,
      "step": 2349
    },
    {
      "epoch": 0.09487283003633427,
      "grad_norm": 115.33699035644531,
      "learning_rate": 1.97803145438035e-05,
      "loss": 4.5862,
      "step": 2350
    },
    {
      "epoch": 0.09491320145337101,
      "grad_norm": 102.9142074584961,
      "learning_rate": 1.9780041885426812e-05,
      "loss": 5.6042,
      "step": 2351
    },
    {
      "epoch": 0.09495357287040775,
      "grad_norm": 85.0650634765625,
      "learning_rate": 1.9779769059834243e-05,
      "loss": 6.5557,
      "step": 2352
    },
    {
      "epoch": 0.0949939442874445,
      "grad_norm": 117.58118438720703,
      "learning_rate": 1.9779496067030462e-05,
      "loss": 6.0937,
      "step": 2353
    },
    {
      "epoch": 0.09503431570448123,
      "grad_norm": 131.29441833496094,
      "learning_rate": 1.9779222907020138e-05,
      "loss": 5.4236,
      "step": 2354
    },
    {
      "epoch": 0.09507468712151797,
      "grad_norm": 92.47189331054688,
      "learning_rate": 1.9778949579807936e-05,
      "loss": 6.6408,
      "step": 2355
    },
    {
      "epoch": 0.0951150585385547,
      "grad_norm": 82.05181121826172,
      "learning_rate": 1.977867608539853e-05,
      "loss": 5.6399,
      "step": 2356
    },
    {
      "epoch": 0.09515542995559144,
      "grad_norm": 158.9532470703125,
      "learning_rate": 1.9778402423796603e-05,
      "loss": 6.7277,
      "step": 2357
    },
    {
      "epoch": 0.09519580137262817,
      "grad_norm": 21014.9140625,
      "learning_rate": 1.9778128595006825e-05,
      "loss": 5.5439,
      "step": 2358
    },
    {
      "epoch": 0.09523617278966491,
      "grad_norm": 419.420654296875,
      "learning_rate": 1.9777854599033884e-05,
      "loss": 5.1918,
      "step": 2359
    },
    {
      "epoch": 0.09527654420670166,
      "grad_norm": 97.3004379272461,
      "learning_rate": 1.9777580435882457e-05,
      "loss": 5.3451,
      "step": 2360
    },
    {
      "epoch": 0.0953169156237384,
      "grad_norm": 120.38861083984375,
      "learning_rate": 1.977730610555724e-05,
      "loss": 6.5143,
      "step": 2361
    },
    {
      "epoch": 0.09535728704077513,
      "grad_norm": 79.10317993164062,
      "learning_rate": 1.9777031608062922e-05,
      "loss": 4.466,
      "step": 2362
    },
    {
      "epoch": 0.09539765845781187,
      "grad_norm": 90.37775421142578,
      "learning_rate": 1.9776756943404195e-05,
      "loss": 5.3652,
      "step": 2363
    },
    {
      "epoch": 0.0954380298748486,
      "grad_norm": 89.27942657470703,
      "learning_rate": 1.9776482111585752e-05,
      "loss": 5.3386,
      "step": 2364
    },
    {
      "epoch": 0.09547840129188534,
      "grad_norm": 115.68490600585938,
      "learning_rate": 1.9776207112612296e-05,
      "loss": 4.7447,
      "step": 2365
    },
    {
      "epoch": 0.09551877270892209,
      "grad_norm": 126.11420440673828,
      "learning_rate": 1.9775931946488523e-05,
      "loss": 5.5317,
      "step": 2366
    },
    {
      "epoch": 0.09555914412595883,
      "grad_norm": 136.37831115722656,
      "learning_rate": 1.977565661321915e-05,
      "loss": 5.9543,
      "step": 2367
    },
    {
      "epoch": 0.09559951554299556,
      "grad_norm": 156.45794677734375,
      "learning_rate": 1.977538111280887e-05,
      "loss": 4.5601,
      "step": 2368
    },
    {
      "epoch": 0.0956398869600323,
      "grad_norm": 117.30919647216797,
      "learning_rate": 1.9775105445262405e-05,
      "loss": 6.8923,
      "step": 2369
    },
    {
      "epoch": 0.09568025837706903,
      "grad_norm": 124.58158874511719,
      "learning_rate": 1.977482961058446e-05,
      "loss": 5.3117,
      "step": 2370
    },
    {
      "epoch": 0.09572062979410577,
      "grad_norm": 401.3665466308594,
      "learning_rate": 1.9774553608779757e-05,
      "loss": 6.4789,
      "step": 2371
    },
    {
      "epoch": 0.0957610012111425,
      "grad_norm": 110.76280212402344,
      "learning_rate": 1.977427743985301e-05,
      "loss": 6.8461,
      "step": 2372
    },
    {
      "epoch": 0.09580137262817925,
      "grad_norm": 118.93902587890625,
      "learning_rate": 1.9774001103808943e-05,
      "loss": 6.5249,
      "step": 2373
    },
    {
      "epoch": 0.09584174404521599,
      "grad_norm": 156.23147583007812,
      "learning_rate": 1.9773724600652282e-05,
      "loss": 6.3835,
      "step": 2374
    },
    {
      "epoch": 0.09588211546225273,
      "grad_norm": 131.27078247070312,
      "learning_rate": 1.9773447930387753e-05,
      "loss": 4.762,
      "step": 2375
    },
    {
      "epoch": 0.09592248687928946,
      "grad_norm": 86.8395767211914,
      "learning_rate": 1.977317109302009e-05,
      "loss": 6.3053,
      "step": 2376
    },
    {
      "epoch": 0.0959628582963262,
      "grad_norm": 136.28729248046875,
      "learning_rate": 1.9772894088554024e-05,
      "loss": 6.9511,
      "step": 2377
    },
    {
      "epoch": 0.09600322971336293,
      "grad_norm": 112.77987670898438,
      "learning_rate": 1.9772616916994286e-05,
      "loss": 5.9188,
      "step": 2378
    },
    {
      "epoch": 0.09604360113039968,
      "grad_norm": 89.4703369140625,
      "learning_rate": 1.977233957834562e-05,
      "loss": 5.6671,
      "step": 2379
    },
    {
      "epoch": 0.09608397254743642,
      "grad_norm": 169.16009521484375,
      "learning_rate": 1.977206207261277e-05,
      "loss": 4.751,
      "step": 2380
    },
    {
      "epoch": 0.09612434396447316,
      "grad_norm": 97.43561553955078,
      "learning_rate": 1.9771784399800475e-05,
      "loss": 4.4651,
      "step": 2381
    },
    {
      "epoch": 0.09616471538150989,
      "grad_norm": 103.71173095703125,
      "learning_rate": 1.977150655991349e-05,
      "loss": 5.3489,
      "step": 2382
    },
    {
      "epoch": 0.09620508679854663,
      "grad_norm": 144.3560333251953,
      "learning_rate": 1.9771228552956557e-05,
      "loss": 5.6104,
      "step": 2383
    },
    {
      "epoch": 0.09624545821558336,
      "grad_norm": 159.3746337890625,
      "learning_rate": 1.9770950378934433e-05,
      "loss": 5.8634,
      "step": 2384
    },
    {
      "epoch": 0.0962858296326201,
      "grad_norm": 106.44154357910156,
      "learning_rate": 1.9770672037851873e-05,
      "loss": 5.7318,
      "step": 2385
    },
    {
      "epoch": 0.09632620104965685,
      "grad_norm": 138.85693359375,
      "learning_rate": 1.9770393529713643e-05,
      "loss": 5.6036,
      "step": 2386
    },
    {
      "epoch": 0.09636657246669358,
      "grad_norm": 75.82081604003906,
      "learning_rate": 1.9770114854524497e-05,
      "loss": 5.4529,
      "step": 2387
    },
    {
      "epoch": 0.09640694388373032,
      "grad_norm": 156.29608154296875,
      "learning_rate": 1.9769836012289198e-05,
      "loss": 5.7007,
      "step": 2388
    },
    {
      "epoch": 0.09644731530076706,
      "grad_norm": 104.03592681884766,
      "learning_rate": 1.976955700301252e-05,
      "loss": 6.9661,
      "step": 2389
    },
    {
      "epoch": 0.09648768671780379,
      "grad_norm": 121.13468933105469,
      "learning_rate": 1.9769277826699235e-05,
      "loss": 6.7216,
      "step": 2390
    },
    {
      "epoch": 0.09652805813484053,
      "grad_norm": 195.54367065429688,
      "learning_rate": 1.9768998483354104e-05,
      "loss": 4.0883,
      "step": 2391
    },
    {
      "epoch": 0.09656842955187728,
      "grad_norm": 250.63702392578125,
      "learning_rate": 1.9768718972981917e-05,
      "loss": 6.4465,
      "step": 2392
    },
    {
      "epoch": 0.09660880096891401,
      "grad_norm": 350.23681640625,
      "learning_rate": 1.9768439295587446e-05,
      "loss": 6.243,
      "step": 2393
    },
    {
      "epoch": 0.09664917238595075,
      "grad_norm": 96.31787109375,
      "learning_rate": 1.9768159451175473e-05,
      "loss": 5.5886,
      "step": 2394
    },
    {
      "epoch": 0.09668954380298748,
      "grad_norm": 75.61261749267578,
      "learning_rate": 1.9767879439750785e-05,
      "loss": 4.9474,
      "step": 2395
    },
    {
      "epoch": 0.09672991522002422,
      "grad_norm": 104.25740814208984,
      "learning_rate": 1.9767599261318166e-05,
      "loss": 5.4049,
      "step": 2396
    },
    {
      "epoch": 0.09677028663706096,
      "grad_norm": 107.0050048828125,
      "learning_rate": 1.9767318915882413e-05,
      "loss": 5.5263,
      "step": 2397
    },
    {
      "epoch": 0.09681065805409769,
      "grad_norm": 82.25096893310547,
      "learning_rate": 1.9767038403448308e-05,
      "loss": 4.4,
      "step": 2398
    },
    {
      "epoch": 0.09685102947113444,
      "grad_norm": 78.97906494140625,
      "learning_rate": 1.976675772402066e-05,
      "loss": 4.3633,
      "step": 2399
    },
    {
      "epoch": 0.09689140088817118,
      "grad_norm": 215.43052673339844,
      "learning_rate": 1.976647687760426e-05,
      "loss": 5.5724,
      "step": 2400
    },
    {
      "epoch": 0.09693177230520791,
      "grad_norm": 80.88265991210938,
      "learning_rate": 1.976619586420391e-05,
      "loss": 6.1246,
      "step": 2401
    },
    {
      "epoch": 0.09697214372224465,
      "grad_norm": 126.0987319946289,
      "learning_rate": 1.9765914683824418e-05,
      "loss": 5.137,
      "step": 2402
    },
    {
      "epoch": 0.09701251513928139,
      "grad_norm": 113.81079864501953,
      "learning_rate": 1.976563333647059e-05,
      "loss": 5.6149,
      "step": 2403
    },
    {
      "epoch": 0.09705288655631812,
      "grad_norm": 176.6213836669922,
      "learning_rate": 1.9765351822147236e-05,
      "loss": 7.4619,
      "step": 2404
    },
    {
      "epoch": 0.09709325797335487,
      "grad_norm": 95.76631927490234,
      "learning_rate": 1.976507014085917e-05,
      "loss": 5.762,
      "step": 2405
    },
    {
      "epoch": 0.09713362939039161,
      "grad_norm": 84.6158447265625,
      "learning_rate": 1.976478829261121e-05,
      "loss": 5.0327,
      "step": 2406
    },
    {
      "epoch": 0.09717400080742834,
      "grad_norm": 89.80198669433594,
      "learning_rate": 1.976450627740817e-05,
      "loss": 6.727,
      "step": 2407
    },
    {
      "epoch": 0.09721437222446508,
      "grad_norm": 120.7254409790039,
      "learning_rate": 1.9764224095254874e-05,
      "loss": 4.9881,
      "step": 2408
    },
    {
      "epoch": 0.09725474364150181,
      "grad_norm": 99.71349334716797,
      "learning_rate": 1.9763941746156146e-05,
      "loss": 4.5719,
      "step": 2409
    },
    {
      "epoch": 0.09729511505853855,
      "grad_norm": 100.69859313964844,
      "learning_rate": 1.9763659230116815e-05,
      "loss": 4.8115,
      "step": 2410
    },
    {
      "epoch": 0.09733548647557529,
      "grad_norm": 101.13951873779297,
      "learning_rate": 1.9763376547141714e-05,
      "loss": 6.4472,
      "step": 2411
    },
    {
      "epoch": 0.09737585789261204,
      "grad_norm": 142.40687561035156,
      "learning_rate": 1.976309369723567e-05,
      "loss": 8.0693,
      "step": 2412
    },
    {
      "epoch": 0.09741622930964877,
      "grad_norm": 111.9431381225586,
      "learning_rate": 1.9762810680403522e-05,
      "loss": 6.8331,
      "step": 2413
    },
    {
      "epoch": 0.09745660072668551,
      "grad_norm": 116.91901397705078,
      "learning_rate": 1.976252749665011e-05,
      "loss": 5.2471,
      "step": 2414
    },
    {
      "epoch": 0.09749697214372224,
      "grad_norm": 104.25396728515625,
      "learning_rate": 1.976224414598028e-05,
      "loss": 5.5158,
      "step": 2415
    },
    {
      "epoch": 0.09753734356075898,
      "grad_norm": 108.5449447631836,
      "learning_rate": 1.9761960628398865e-05,
      "loss": 6.3574,
      "step": 2416
    },
    {
      "epoch": 0.09757771497779572,
      "grad_norm": 109.01921081542969,
      "learning_rate": 1.9761676943910723e-05,
      "loss": 3.7665,
      "step": 2417
    },
    {
      "epoch": 0.09761808639483247,
      "grad_norm": 121.7015380859375,
      "learning_rate": 1.97613930925207e-05,
      "loss": 5.6646,
      "step": 2418
    },
    {
      "epoch": 0.0976584578118692,
      "grad_norm": 190.6981658935547,
      "learning_rate": 1.9761109074233647e-05,
      "loss": 5.2609,
      "step": 2419
    },
    {
      "epoch": 0.09769882922890594,
      "grad_norm": 169.84527587890625,
      "learning_rate": 1.9760824889054422e-05,
      "loss": 6.4619,
      "step": 2420
    },
    {
      "epoch": 0.09773920064594267,
      "grad_norm": 101.99590301513672,
      "learning_rate": 1.9760540536987888e-05,
      "loss": 5.7347,
      "step": 2421
    },
    {
      "epoch": 0.09777957206297941,
      "grad_norm": 74.65457916259766,
      "learning_rate": 1.9760256018038902e-05,
      "loss": 5.9371,
      "step": 2422
    },
    {
      "epoch": 0.09781994348001614,
      "grad_norm": 108.81520080566406,
      "learning_rate": 1.975997133221233e-05,
      "loss": 7.0482,
      "step": 2423
    },
    {
      "epoch": 0.09786031489705288,
      "grad_norm": 153.23330688476562,
      "learning_rate": 1.9759686479513037e-05,
      "loss": 3.2813,
      "step": 2424
    },
    {
      "epoch": 0.09790068631408963,
      "grad_norm": 117.38907623291016,
      "learning_rate": 1.97594014599459e-05,
      "loss": 5.4084,
      "step": 2425
    },
    {
      "epoch": 0.09794105773112637,
      "grad_norm": 83.6557388305664,
      "learning_rate": 1.9759116273515787e-05,
      "loss": 2.5545,
      "step": 2426
    },
    {
      "epoch": 0.0979814291481631,
      "grad_norm": 108.98946380615234,
      "learning_rate": 1.975883092022757e-05,
      "loss": 5.4983,
      "step": 2427
    },
    {
      "epoch": 0.09802180056519984,
      "grad_norm": 110.91129302978516,
      "learning_rate": 1.975854540008614e-05,
      "loss": 5.0113,
      "step": 2428
    },
    {
      "epoch": 0.09806217198223657,
      "grad_norm": 142.40296936035156,
      "learning_rate": 1.975825971309637e-05,
      "loss": 6.688,
      "step": 2429
    },
    {
      "epoch": 0.09810254339927331,
      "grad_norm": 96.13964080810547,
      "learning_rate": 1.9757973859263143e-05,
      "loss": 4.8962,
      "step": 2430
    },
    {
      "epoch": 0.09814291481631006,
      "grad_norm": 92.86791229248047,
      "learning_rate": 1.975768783859135e-05,
      "loss": 5.1972,
      "step": 2431
    },
    {
      "epoch": 0.0981832862333468,
      "grad_norm": 150.2454833984375,
      "learning_rate": 1.975740165108588e-05,
      "loss": 4.3848,
      "step": 2432
    },
    {
      "epoch": 0.09822365765038353,
      "grad_norm": 97.1158676147461,
      "learning_rate": 1.975711529675163e-05,
      "loss": 6.299,
      "step": 2433
    },
    {
      "epoch": 0.09826402906742027,
      "grad_norm": 185.18238830566406,
      "learning_rate": 1.9756828775593487e-05,
      "loss": 5.858,
      "step": 2434
    },
    {
      "epoch": 0.098304400484457,
      "grad_norm": 92.04414367675781,
      "learning_rate": 1.975654208761636e-05,
      "loss": 4.5131,
      "step": 2435
    },
    {
      "epoch": 0.09834477190149374,
      "grad_norm": 173.89541625976562,
      "learning_rate": 1.975625523282515e-05,
      "loss": 5.7573,
      "step": 2436
    },
    {
      "epoch": 0.09838514331853047,
      "grad_norm": 105.96553039550781,
      "learning_rate": 1.9755968211224753e-05,
      "loss": 5.6035,
      "step": 2437
    },
    {
      "epoch": 0.09842551473556722,
      "grad_norm": 97.43311309814453,
      "learning_rate": 1.9755681022820083e-05,
      "loss": 5.9665,
      "step": 2438
    },
    {
      "epoch": 0.09846588615260396,
      "grad_norm": 90.70661926269531,
      "learning_rate": 1.9755393667616046e-05,
      "loss": 5.0422,
      "step": 2439
    },
    {
      "epoch": 0.0985062575696407,
      "grad_norm": 143.28414916992188,
      "learning_rate": 1.975510614561756e-05,
      "loss": 6.3037,
      "step": 2440
    },
    {
      "epoch": 0.09854662898667743,
      "grad_norm": 201.35520935058594,
      "learning_rate": 1.9754818456829543e-05,
      "loss": 6.4978,
      "step": 2441
    },
    {
      "epoch": 0.09858700040371417,
      "grad_norm": 100.08476257324219,
      "learning_rate": 1.9754530601256902e-05,
      "loss": 5.0142,
      "step": 2442
    },
    {
      "epoch": 0.0986273718207509,
      "grad_norm": 80.16302490234375,
      "learning_rate": 1.9754242578904573e-05,
      "loss": 6.091,
      "step": 2443
    },
    {
      "epoch": 0.09866774323778765,
      "grad_norm": 136.3591766357422,
      "learning_rate": 1.975395438977747e-05,
      "loss": 4.8208,
      "step": 2444
    },
    {
      "epoch": 0.09870811465482439,
      "grad_norm": 74.02751922607422,
      "learning_rate": 1.9753666033880522e-05,
      "loss": 6.3246,
      "step": 2445
    },
    {
      "epoch": 0.09874848607186112,
      "grad_norm": 103.34249877929688,
      "learning_rate": 1.9753377511218665e-05,
      "loss": 6.3329,
      "step": 2446
    },
    {
      "epoch": 0.09878885748889786,
      "grad_norm": 86.33425903320312,
      "learning_rate": 1.975308882179683e-05,
      "loss": 6.8257,
      "step": 2447
    },
    {
      "epoch": 0.0988292289059346,
      "grad_norm": 137.08595275878906,
      "learning_rate": 1.9752799965619947e-05,
      "loss": 4.8271,
      "step": 2448
    },
    {
      "epoch": 0.09886960032297133,
      "grad_norm": 124.26188659667969,
      "learning_rate": 1.975251094269296e-05,
      "loss": 6.8915,
      "step": 2449
    },
    {
      "epoch": 0.09890997174000807,
      "grad_norm": 109.66532135009766,
      "learning_rate": 1.975222175302081e-05,
      "loss": 6.2231,
      "step": 2450
    },
    {
      "epoch": 0.09895034315704482,
      "grad_norm": 132.03614807128906,
      "learning_rate": 1.9751932396608443e-05,
      "loss": 3.9577,
      "step": 2451
    },
    {
      "epoch": 0.09899071457408155,
      "grad_norm": 99.2720718383789,
      "learning_rate": 1.97516428734608e-05,
      "loss": 4.4895,
      "step": 2452
    },
    {
      "epoch": 0.09903108599111829,
      "grad_norm": 97.7604751586914,
      "learning_rate": 1.9751353183582838e-05,
      "loss": 5.9736,
      "step": 2453
    },
    {
      "epoch": 0.09907145740815503,
      "grad_norm": 104.16791534423828,
      "learning_rate": 1.975106332697951e-05,
      "loss": 6.5462,
      "step": 2454
    },
    {
      "epoch": 0.09911182882519176,
      "grad_norm": 128.75912475585938,
      "learning_rate": 1.9750773303655765e-05,
      "loss": 6.0572,
      "step": 2455
    },
    {
      "epoch": 0.0991522002422285,
      "grad_norm": 126.58268737792969,
      "learning_rate": 1.9750483113616568e-05,
      "loss": 4.322,
      "step": 2456
    },
    {
      "epoch": 0.09919257165926525,
      "grad_norm": 92.10576629638672,
      "learning_rate": 1.975019275686688e-05,
      "loss": 6.8905,
      "step": 2457
    },
    {
      "epoch": 0.09923294307630198,
      "grad_norm": 226.9077911376953,
      "learning_rate": 1.9749902233411663e-05,
      "loss": 5.3288,
      "step": 2458
    },
    {
      "epoch": 0.09927331449333872,
      "grad_norm": 151.6121368408203,
      "learning_rate": 1.9749611543255885e-05,
      "loss": 5.6221,
      "step": 2459
    },
    {
      "epoch": 0.09931368591037545,
      "grad_norm": 105.63053894042969,
      "learning_rate": 1.9749320686404518e-05,
      "loss": 5.1039,
      "step": 2460
    },
    {
      "epoch": 0.09935405732741219,
      "grad_norm": 129.82736206054688,
      "learning_rate": 1.9749029662862532e-05,
      "loss": 4.8188,
      "step": 2461
    },
    {
      "epoch": 0.09939442874444893,
      "grad_norm": 127.94901275634766,
      "learning_rate": 1.9748738472634908e-05,
      "loss": 5.9799,
      "step": 2462
    },
    {
      "epoch": 0.09943480016148566,
      "grad_norm": 162.7176055908203,
      "learning_rate": 1.9748447115726616e-05,
      "loss": 6.0217,
      "step": 2463
    },
    {
      "epoch": 0.09947517157852241,
      "grad_norm": 131.2802276611328,
      "learning_rate": 1.9748155592142646e-05,
      "loss": 5.6631,
      "step": 2464
    },
    {
      "epoch": 0.09951554299555915,
      "grad_norm": 109.24333953857422,
      "learning_rate": 1.9747863901887976e-05,
      "loss": 6.9856,
      "step": 2465
    },
    {
      "epoch": 0.09955591441259588,
      "grad_norm": 137.67303466796875,
      "learning_rate": 1.97475720449676e-05,
      "loss": 5.3914,
      "step": 2466
    },
    {
      "epoch": 0.09959628582963262,
      "grad_norm": 131.32705688476562,
      "learning_rate": 1.9747280021386503e-05,
      "loss": 5.7909,
      "step": 2467
    },
    {
      "epoch": 0.09963665724666936,
      "grad_norm": 108.51913452148438,
      "learning_rate": 1.9746987831149677e-05,
      "loss": 6.1216,
      "step": 2468
    },
    {
      "epoch": 0.09967702866370609,
      "grad_norm": 98.90953063964844,
      "learning_rate": 1.974669547426212e-05,
      "loss": 5.9725,
      "step": 2469
    },
    {
      "epoch": 0.09971740008074284,
      "grad_norm": 543.4216918945312,
      "learning_rate": 1.9746402950728835e-05,
      "loss": 5.3916,
      "step": 2470
    },
    {
      "epoch": 0.09975777149777958,
      "grad_norm": 132.084716796875,
      "learning_rate": 1.9746110260554813e-05,
      "loss": 6.2923,
      "step": 2471
    },
    {
      "epoch": 0.09979814291481631,
      "grad_norm": 76.45973205566406,
      "learning_rate": 1.9745817403745068e-05,
      "loss": 5.6958,
      "step": 2472
    },
    {
      "epoch": 0.09983851433185305,
      "grad_norm": 93.71354675292969,
      "learning_rate": 1.9745524380304605e-05,
      "loss": 5.3357,
      "step": 2473
    },
    {
      "epoch": 0.09987888574888978,
      "grad_norm": 87.7344741821289,
      "learning_rate": 1.974523119023843e-05,
      "loss": 7.2914,
      "step": 2474
    },
    {
      "epoch": 0.09991925716592652,
      "grad_norm": 95.80567932128906,
      "learning_rate": 1.974493783355156e-05,
      "loss": 6.5136,
      "step": 2475
    },
    {
      "epoch": 0.09995962858296326,
      "grad_norm": 157.66017150878906,
      "learning_rate": 1.9744644310249006e-05,
      "loss": 7.1039,
      "step": 2476
    },
    {
      "epoch": 0.1,
      "grad_norm": 133.5247039794922,
      "learning_rate": 1.9744350620335794e-05,
      "loss": 5.2919,
      "step": 2477
    },
    {
      "epoch": 0.10004037141703674,
      "grad_norm": 119.17767333984375,
      "learning_rate": 1.9744056763816937e-05,
      "loss": 5.5264,
      "step": 2478
    },
    {
      "epoch": 0.10008074283407348,
      "grad_norm": 84.1343994140625,
      "learning_rate": 1.9743762740697466e-05,
      "loss": 4.896,
      "step": 2479
    },
    {
      "epoch": 0.10012111425111021,
      "grad_norm": 158.8845977783203,
      "learning_rate": 1.9743468550982404e-05,
      "loss": 6.1447,
      "step": 2480
    },
    {
      "epoch": 0.10016148566814695,
      "grad_norm": 175.55606079101562,
      "learning_rate": 1.974317419467678e-05,
      "loss": 5.5828,
      "step": 2481
    },
    {
      "epoch": 0.10020185708518368,
      "grad_norm": 154.7248077392578,
      "learning_rate": 1.9742879671785635e-05,
      "loss": 6.3925,
      "step": 2482
    },
    {
      "epoch": 0.10024222850222043,
      "grad_norm": 170.72450256347656,
      "learning_rate": 1.9742584982313998e-05,
      "loss": 4.518,
      "step": 2483
    },
    {
      "epoch": 0.10028259991925717,
      "grad_norm": 114.38961029052734,
      "learning_rate": 1.97422901262669e-05,
      "loss": 6.5977,
      "step": 2484
    },
    {
      "epoch": 0.1003229713362939,
      "grad_norm": 149.61013793945312,
      "learning_rate": 1.97419951036494e-05,
      "loss": 4.4861,
      "step": 2485
    },
    {
      "epoch": 0.10036334275333064,
      "grad_norm": 115.8669204711914,
      "learning_rate": 1.9741699914466533e-05,
      "loss": 5.8773,
      "step": 2486
    },
    {
      "epoch": 0.10040371417036738,
      "grad_norm": 113.4669418334961,
      "learning_rate": 1.974140455872334e-05,
      "loss": 6.3618,
      "step": 2487
    },
    {
      "epoch": 0.10044408558740411,
      "grad_norm": 105.41571044921875,
      "learning_rate": 1.974110903642488e-05,
      "loss": 4.5093,
      "step": 2488
    },
    {
      "epoch": 0.10048445700444085,
      "grad_norm": 183.53001403808594,
      "learning_rate": 1.9740813347576204e-05,
      "loss": 5.7829,
      "step": 2489
    },
    {
      "epoch": 0.1005248284214776,
      "grad_norm": 115.10066223144531,
      "learning_rate": 1.9740517492182363e-05,
      "loss": 5.3498,
      "step": 2490
    },
    {
      "epoch": 0.10056519983851434,
      "grad_norm": 123.7801284790039,
      "learning_rate": 1.9740221470248422e-05,
      "loss": 5.6118,
      "step": 2491
    },
    {
      "epoch": 0.10060557125555107,
      "grad_norm": 107.53775787353516,
      "learning_rate": 1.9739925281779437e-05,
      "loss": 5.7416,
      "step": 2492
    },
    {
      "epoch": 0.10064594267258781,
      "grad_norm": 119.26774597167969,
      "learning_rate": 1.9739628926780476e-05,
      "loss": 6.2826,
      "step": 2493
    },
    {
      "epoch": 0.10068631408962454,
      "grad_norm": 114.21176147460938,
      "learning_rate": 1.97393324052566e-05,
      "loss": 6.6671,
      "step": 2494
    },
    {
      "epoch": 0.10072668550666128,
      "grad_norm": 98.68234252929688,
      "learning_rate": 1.9739035717212887e-05,
      "loss": 6.0648,
      "step": 2495
    },
    {
      "epoch": 0.10076705692369803,
      "grad_norm": 153.3808135986328,
      "learning_rate": 1.9738738862654402e-05,
      "loss": 5.3827,
      "step": 2496
    },
    {
      "epoch": 0.10080742834073476,
      "grad_norm": 259.301025390625,
      "learning_rate": 1.9738441841586223e-05,
      "loss": 5.09,
      "step": 2497
    },
    {
      "epoch": 0.1008477997577715,
      "grad_norm": 97.29655456542969,
      "learning_rate": 1.9738144654013434e-05,
      "loss": 6.3729,
      "step": 2498
    },
    {
      "epoch": 0.10088817117480824,
      "grad_norm": 63.57835006713867,
      "learning_rate": 1.973784729994111e-05,
      "loss": 3.8346,
      "step": 2499
    },
    {
      "epoch": 0.10092854259184497,
      "grad_norm": 97.43134307861328,
      "learning_rate": 1.9737549779374337e-05,
      "loss": 6.1104,
      "step": 2500
    },
    {
      "epoch": 0.10096891400888171,
      "grad_norm": 118.58113098144531,
      "learning_rate": 1.97372520923182e-05,
      "loss": 5.4479,
      "step": 2501
    },
    {
      "epoch": 0.10100928542591844,
      "grad_norm": 157.2510986328125,
      "learning_rate": 1.9736954238777793e-05,
      "loss": 4.9847,
      "step": 2502
    },
    {
      "epoch": 0.1010496568429552,
      "grad_norm": 87.1435546875,
      "learning_rate": 1.9736656218758202e-05,
      "loss": 5.6578,
      "step": 2503
    },
    {
      "epoch": 0.10109002825999193,
      "grad_norm": 164.38827514648438,
      "learning_rate": 1.973635803226453e-05,
      "loss": 4.5693,
      "step": 2504
    },
    {
      "epoch": 0.10113039967702867,
      "grad_norm": 152.88441467285156,
      "learning_rate": 1.973605967930187e-05,
      "loss": 7.257,
      "step": 2505
    },
    {
      "epoch": 0.1011707710940654,
      "grad_norm": 200.15550231933594,
      "learning_rate": 1.9735761159875326e-05,
      "loss": 6.8499,
      "step": 2506
    },
    {
      "epoch": 0.10121114251110214,
      "grad_norm": 136.20941162109375,
      "learning_rate": 1.973546247399e-05,
      "loss": 5.9891,
      "step": 2507
    },
    {
      "epoch": 0.10125151392813887,
      "grad_norm": 111.5837173461914,
      "learning_rate": 1.9735163621650998e-05,
      "loss": 4.1951,
      "step": 2508
    },
    {
      "epoch": 0.10129188534517562,
      "grad_norm": 152.20111083984375,
      "learning_rate": 1.9734864602863433e-05,
      "loss": 4.5588,
      "step": 2509
    },
    {
      "epoch": 0.10133225676221236,
      "grad_norm": 110.37327575683594,
      "learning_rate": 1.9734565417632417e-05,
      "loss": 5.2265,
      "step": 2510
    },
    {
      "epoch": 0.1013726281792491,
      "grad_norm": 110.14143371582031,
      "learning_rate": 1.9734266065963062e-05,
      "loss": 6.2384,
      "step": 2511
    },
    {
      "epoch": 0.10141299959628583,
      "grad_norm": 131.9818572998047,
      "learning_rate": 1.973396654786049e-05,
      "loss": 6.1516,
      "step": 2512
    },
    {
      "epoch": 0.10145337101332257,
      "grad_norm": 85.44729614257812,
      "learning_rate": 1.973366686332982e-05,
      "loss": 3.6714,
      "step": 2513
    },
    {
      "epoch": 0.1014937424303593,
      "grad_norm": 111.82623291015625,
      "learning_rate": 1.9733367012376178e-05,
      "loss": 5.8927,
      "step": 2514
    },
    {
      "epoch": 0.10153411384739604,
      "grad_norm": 137.7482452392578,
      "learning_rate": 1.9733066995004683e-05,
      "loss": 6.1159,
      "step": 2515
    },
    {
      "epoch": 0.10157448526443279,
      "grad_norm": 94.08084106445312,
      "learning_rate": 1.9732766811220478e-05,
      "loss": 5.3511,
      "step": 2516
    },
    {
      "epoch": 0.10161485668146952,
      "grad_norm": 148.0215301513672,
      "learning_rate": 1.9732466461028685e-05,
      "loss": 5.4384,
      "step": 2517
    },
    {
      "epoch": 0.10165522809850626,
      "grad_norm": 125.46881103515625,
      "learning_rate": 1.9732165944434443e-05,
      "loss": 6.2828,
      "step": 2518
    },
    {
      "epoch": 0.101695599515543,
      "grad_norm": 109.93425750732422,
      "learning_rate": 1.973186526144289e-05,
      "loss": 5.9983,
      "step": 2519
    },
    {
      "epoch": 0.10173597093257973,
      "grad_norm": 190.96397399902344,
      "learning_rate": 1.9731564412059165e-05,
      "loss": 6.4492,
      "step": 2520
    },
    {
      "epoch": 0.10177634234961647,
      "grad_norm": 136.37091064453125,
      "learning_rate": 1.9731263396288414e-05,
      "loss": 5.4929,
      "step": 2521
    },
    {
      "epoch": 0.10181671376665322,
      "grad_norm": 111.9578857421875,
      "learning_rate": 1.973096221413578e-05,
      "loss": 6.1867,
      "step": 2522
    },
    {
      "epoch": 0.10185708518368995,
      "grad_norm": 109.77230834960938,
      "learning_rate": 1.9730660865606423e-05,
      "loss": 5.2992,
      "step": 2523
    },
    {
      "epoch": 0.10189745660072669,
      "grad_norm": 129.55230712890625,
      "learning_rate": 1.9730359350705482e-05,
      "loss": 5.1258,
      "step": 2524
    },
    {
      "epoch": 0.10193782801776342,
      "grad_norm": 102.20885467529297,
      "learning_rate": 1.973005766943812e-05,
      "loss": 5.341,
      "step": 2525
    },
    {
      "epoch": 0.10197819943480016,
      "grad_norm": 109.72834014892578,
      "learning_rate": 1.9729755821809494e-05,
      "loss": 4.5812,
      "step": 2526
    },
    {
      "epoch": 0.1020185708518369,
      "grad_norm": 138.123046875,
      "learning_rate": 1.972945380782476e-05,
      "loss": 6.0073,
      "step": 2527
    },
    {
      "epoch": 0.10205894226887363,
      "grad_norm": 108.15612030029297,
      "learning_rate": 1.9729151627489092e-05,
      "loss": 6.1533,
      "step": 2528
    },
    {
      "epoch": 0.10209931368591038,
      "grad_norm": 95.0414810180664,
      "learning_rate": 1.972884928080765e-05,
      "loss": 6.8045,
      "step": 2529
    },
    {
      "epoch": 0.10213968510294712,
      "grad_norm": 86.7470932006836,
      "learning_rate": 1.97285467677856e-05,
      "loss": 4.1296,
      "step": 2530
    },
    {
      "epoch": 0.10218005651998385,
      "grad_norm": 108.9822769165039,
      "learning_rate": 1.9728244088428118e-05,
      "loss": 5.0396,
      "step": 2531
    },
    {
      "epoch": 0.10222042793702059,
      "grad_norm": 128.39035034179688,
      "learning_rate": 1.9727941242740378e-05,
      "loss": 5.7919,
      "step": 2532
    },
    {
      "epoch": 0.10226079935405732,
      "grad_norm": 127.35713195800781,
      "learning_rate": 1.9727638230727565e-05,
      "loss": 5.6114,
      "step": 2533
    },
    {
      "epoch": 0.10230117077109406,
      "grad_norm": 144.91558837890625,
      "learning_rate": 1.972733505239485e-05,
      "loss": 4.5266,
      "step": 2534
    },
    {
      "epoch": 0.10234154218813081,
      "grad_norm": 94.20507049560547,
      "learning_rate": 1.972703170774742e-05,
      "loss": 6.1131,
      "step": 2535
    },
    {
      "epoch": 0.10238191360516755,
      "grad_norm": 207.52317810058594,
      "learning_rate": 1.9726728196790462e-05,
      "loss": 5.4572,
      "step": 2536
    },
    {
      "epoch": 0.10242228502220428,
      "grad_norm": 123.73387145996094,
      "learning_rate": 1.9726424519529166e-05,
      "loss": 4.9759,
      "step": 2537
    },
    {
      "epoch": 0.10246265643924102,
      "grad_norm": 110.73001098632812,
      "learning_rate": 1.9726120675968727e-05,
      "loss": 5.1101,
      "step": 2538
    },
    {
      "epoch": 0.10250302785627775,
      "grad_norm": 116.28274536132812,
      "learning_rate": 1.9725816666114334e-05,
      "loss": 5.8225,
      "step": 2539
    },
    {
      "epoch": 0.10254339927331449,
      "grad_norm": 104.17135620117188,
      "learning_rate": 1.9725512489971184e-05,
      "loss": 5.3714,
      "step": 2540
    },
    {
      "epoch": 0.10258377069035123,
      "grad_norm": 134.6452178955078,
      "learning_rate": 1.9725208147544483e-05,
      "loss": 6.7088,
      "step": 2541
    },
    {
      "epoch": 0.10262414210738798,
      "grad_norm": 99.80682373046875,
      "learning_rate": 1.9724903638839435e-05,
      "loss": 7.5401,
      "step": 2542
    },
    {
      "epoch": 0.10266451352442471,
      "grad_norm": 140.43804931640625,
      "learning_rate": 1.9724598963861243e-05,
      "loss": 6.2188,
      "step": 2543
    },
    {
      "epoch": 0.10270488494146145,
      "grad_norm": 112.00211334228516,
      "learning_rate": 1.9724294122615118e-05,
      "loss": 6.8431,
      "step": 2544
    },
    {
      "epoch": 0.10274525635849818,
      "grad_norm": 148.23411560058594,
      "learning_rate": 1.9723989115106268e-05,
      "loss": 5.5326,
      "step": 2545
    },
    {
      "epoch": 0.10278562777553492,
      "grad_norm": 138.24122619628906,
      "learning_rate": 1.9723683941339914e-05,
      "loss": 5.7793,
      "step": 2546
    },
    {
      "epoch": 0.10282599919257165,
      "grad_norm": 101.29670715332031,
      "learning_rate": 1.972337860132127e-05,
      "loss": 7.0232,
      "step": 2547
    },
    {
      "epoch": 0.1028663706096084,
      "grad_norm": 107.5257568359375,
      "learning_rate": 1.972307309505556e-05,
      "loss": 4.5421,
      "step": 2548
    },
    {
      "epoch": 0.10290674202664514,
      "grad_norm": 110.82424926757812,
      "learning_rate": 1.9722767422548003e-05,
      "loss": 6.4327,
      "step": 2549
    },
    {
      "epoch": 0.10294711344368188,
      "grad_norm": 157.67478942871094,
      "learning_rate": 1.9722461583803828e-05,
      "loss": 4.8074,
      "step": 2550
    },
    {
      "epoch": 0.10298748486071861,
      "grad_norm": 101.05438232421875,
      "learning_rate": 1.9722155578828265e-05,
      "loss": 5.6429,
      "step": 2551
    },
    {
      "epoch": 0.10302785627775535,
      "grad_norm": 85.09236145019531,
      "learning_rate": 1.972184940762654e-05,
      "loss": 4.7735,
      "step": 2552
    },
    {
      "epoch": 0.10306822769479208,
      "grad_norm": 90.09942626953125,
      "learning_rate": 1.97215430702039e-05,
      "loss": 4.2794,
      "step": 2553
    },
    {
      "epoch": 0.10310859911182882,
      "grad_norm": 102.11453247070312,
      "learning_rate": 1.9721236566565568e-05,
      "loss": 5.6923,
      "step": 2554
    },
    {
      "epoch": 0.10314897052886557,
      "grad_norm": 125.58433532714844,
      "learning_rate": 1.9720929896716792e-05,
      "loss": 5.9487,
      "step": 2555
    },
    {
      "epoch": 0.1031893419459023,
      "grad_norm": 228.82461547851562,
      "learning_rate": 1.9720623060662814e-05,
      "loss": 5.7441,
      "step": 2556
    },
    {
      "epoch": 0.10322971336293904,
      "grad_norm": 121.3914794921875,
      "learning_rate": 1.9720316058408887e-05,
      "loss": 5.1434,
      "step": 2557
    },
    {
      "epoch": 0.10327008477997578,
      "grad_norm": 141.2384033203125,
      "learning_rate": 1.9720008889960248e-05,
      "loss": 4.7731,
      "step": 2558
    },
    {
      "epoch": 0.10331045619701251,
      "grad_norm": 96.4231185913086,
      "learning_rate": 1.9719701555322156e-05,
      "loss": 5.3595,
      "step": 2559
    },
    {
      "epoch": 0.10335082761404925,
      "grad_norm": 118.96881866455078,
      "learning_rate": 1.9719394054499865e-05,
      "loss": 4.6058,
      "step": 2560
    },
    {
      "epoch": 0.103391199031086,
      "grad_norm": 96.8377685546875,
      "learning_rate": 1.9719086387498634e-05,
      "loss": 5.0943,
      "step": 2561
    },
    {
      "epoch": 0.10343157044812273,
      "grad_norm": 104.12934112548828,
      "learning_rate": 1.9718778554323722e-05,
      "loss": 6.2366,
      "step": 2562
    },
    {
      "epoch": 0.10347194186515947,
      "grad_norm": 95.96253967285156,
      "learning_rate": 1.9718470554980387e-05,
      "loss": 6.1297,
      "step": 2563
    },
    {
      "epoch": 0.1035123132821962,
      "grad_norm": 157.25314331054688,
      "learning_rate": 1.9718162389473904e-05,
      "loss": 6.1269,
      "step": 2564
    },
    {
      "epoch": 0.10355268469923294,
      "grad_norm": 131.66363525390625,
      "learning_rate": 1.9717854057809535e-05,
      "loss": 6.2809,
      "step": 2565
    },
    {
      "epoch": 0.10359305611626968,
      "grad_norm": 134.1949920654297,
      "learning_rate": 1.971754555999256e-05,
      "loss": 6.0287,
      "step": 2566
    },
    {
      "epoch": 0.10363342753330641,
      "grad_norm": 77.41162872314453,
      "learning_rate": 1.971723689602824e-05,
      "loss": 6.2782,
      "step": 2567
    },
    {
      "epoch": 0.10367379895034316,
      "grad_norm": 108.18439483642578,
      "learning_rate": 1.9716928065921865e-05,
      "loss": 5.7477,
      "step": 2568
    },
    {
      "epoch": 0.1037141703673799,
      "grad_norm": 84.10678100585938,
      "learning_rate": 1.9716619069678708e-05,
      "loss": 4.7555,
      "step": 2569
    },
    {
      "epoch": 0.10375454178441663,
      "grad_norm": 118.70947265625,
      "learning_rate": 1.9716309907304055e-05,
      "loss": 4.9544,
      "step": 2570
    },
    {
      "epoch": 0.10379491320145337,
      "grad_norm": 143.5721893310547,
      "learning_rate": 1.9716000578803196e-05,
      "loss": 5.8085,
      "step": 2571
    },
    {
      "epoch": 0.1038352846184901,
      "grad_norm": 85.77781677246094,
      "learning_rate": 1.9715691084181414e-05,
      "loss": 7.0073,
      "step": 2572
    },
    {
      "epoch": 0.10387565603552684,
      "grad_norm": 107.39401245117188,
      "learning_rate": 1.9715381423444004e-05,
      "loss": 5.6258,
      "step": 2573
    },
    {
      "epoch": 0.10391602745256359,
      "grad_norm": 102.1375961303711,
      "learning_rate": 1.9715071596596253e-05,
      "loss": 5.666,
      "step": 2574
    },
    {
      "epoch": 0.10395639886960033,
      "grad_norm": 97.93012237548828,
      "learning_rate": 1.9714761603643467e-05,
      "loss": 7.272,
      "step": 2575
    },
    {
      "epoch": 0.10399677028663706,
      "grad_norm": 134.44204711914062,
      "learning_rate": 1.9714451444590943e-05,
      "loss": 5.4364,
      "step": 2576
    },
    {
      "epoch": 0.1040371417036738,
      "grad_norm": 96.56672668457031,
      "learning_rate": 1.9714141119443985e-05,
      "loss": 5.2736,
      "step": 2577
    },
    {
      "epoch": 0.10407751312071054,
      "grad_norm": 73.19062042236328,
      "learning_rate": 1.9713830628207897e-05,
      "loss": 3.8792,
      "step": 2578
    },
    {
      "epoch": 0.10411788453774727,
      "grad_norm": 176.65281677246094,
      "learning_rate": 1.9713519970887987e-05,
      "loss": 4.9779,
      "step": 2579
    },
    {
      "epoch": 0.10415825595478401,
      "grad_norm": 150.16563415527344,
      "learning_rate": 1.9713209147489572e-05,
      "loss": 4.294,
      "step": 2580
    },
    {
      "epoch": 0.10419862737182076,
      "grad_norm": 109.28833770751953,
      "learning_rate": 1.971289815801796e-05,
      "loss": 4.1821,
      "step": 2581
    },
    {
      "epoch": 0.10423899878885749,
      "grad_norm": 75.11372375488281,
      "learning_rate": 1.9712587002478472e-05,
      "loss": 5.2041,
      "step": 2582
    },
    {
      "epoch": 0.10427937020589423,
      "grad_norm": 132.49339294433594,
      "learning_rate": 1.9712275680876426e-05,
      "loss": 4.9164,
      "step": 2583
    },
    {
      "epoch": 0.10431974162293096,
      "grad_norm": 114.01103973388672,
      "learning_rate": 1.9711964193217148e-05,
      "loss": 4.8901,
      "step": 2584
    },
    {
      "epoch": 0.1043601130399677,
      "grad_norm": 103.84362030029297,
      "learning_rate": 1.971165253950596e-05,
      "loss": 5.604,
      "step": 2585
    },
    {
      "epoch": 0.10440048445700444,
      "grad_norm": 100.9261703491211,
      "learning_rate": 1.971134071974819e-05,
      "loss": 6.5996,
      "step": 2586
    },
    {
      "epoch": 0.10444085587404119,
      "grad_norm": 84.63832092285156,
      "learning_rate": 1.971102873394917e-05,
      "loss": 6.1442,
      "step": 2587
    },
    {
      "epoch": 0.10448122729107792,
      "grad_norm": 85.23934936523438,
      "learning_rate": 1.971071658211424e-05,
      "loss": 4.5554,
      "step": 2588
    },
    {
      "epoch": 0.10452159870811466,
      "grad_norm": 130.8696746826172,
      "learning_rate": 1.9710404264248734e-05,
      "loss": 5.6468,
      "step": 2589
    },
    {
      "epoch": 0.1045619701251514,
      "grad_norm": 126.74520874023438,
      "learning_rate": 1.9710091780357985e-05,
      "loss": 7.0587,
      "step": 2590
    },
    {
      "epoch": 0.10460234154218813,
      "grad_norm": 67.29168701171875,
      "learning_rate": 1.970977913044734e-05,
      "loss": 4.6017,
      "step": 2591
    },
    {
      "epoch": 0.10464271295922487,
      "grad_norm": 102.0086898803711,
      "learning_rate": 1.9709466314522153e-05,
      "loss": 5.9794,
      "step": 2592
    },
    {
      "epoch": 0.1046830843762616,
      "grad_norm": 110.45195770263672,
      "learning_rate": 1.970915333258776e-05,
      "loss": 5.617,
      "step": 2593
    },
    {
      "epoch": 0.10472345579329835,
      "grad_norm": 191.82870483398438,
      "learning_rate": 1.970884018464952e-05,
      "loss": 4.6854,
      "step": 2594
    },
    {
      "epoch": 0.10476382721033509,
      "grad_norm": 150.11212158203125,
      "learning_rate": 1.9708526870712784e-05,
      "loss": 4.5887,
      "step": 2595
    },
    {
      "epoch": 0.10480419862737182,
      "grad_norm": 78.15296936035156,
      "learning_rate": 1.970821339078291e-05,
      "loss": 6.3133,
      "step": 2596
    },
    {
      "epoch": 0.10484457004440856,
      "grad_norm": 119.4633560180664,
      "learning_rate": 1.9707899744865255e-05,
      "loss": 5.9729,
      "step": 2597
    },
    {
      "epoch": 0.1048849414614453,
      "grad_norm": 250.44256591796875,
      "learning_rate": 1.970758593296519e-05,
      "loss": 6.3405,
      "step": 2598
    },
    {
      "epoch": 0.10492531287848203,
      "grad_norm": 147.5523681640625,
      "learning_rate": 1.9707271955088067e-05,
      "loss": 6.2877,
      "step": 2599
    },
    {
      "epoch": 0.10496568429551877,
      "grad_norm": 188.03140258789062,
      "learning_rate": 1.9706957811239265e-05,
      "loss": 4.8079,
      "step": 2600
    },
    {
      "epoch": 0.10500605571255552,
      "grad_norm": 118.70106506347656,
      "learning_rate": 1.970664350142415e-05,
      "loss": 6.9744,
      "step": 2601
    },
    {
      "epoch": 0.10504642712959225,
      "grad_norm": 114.62787628173828,
      "learning_rate": 1.9706329025648097e-05,
      "loss": 5.9019,
      "step": 2602
    },
    {
      "epoch": 0.10508679854662899,
      "grad_norm": 98.14299011230469,
      "learning_rate": 1.9706014383916486e-05,
      "loss": 5.8604,
      "step": 2603
    },
    {
      "epoch": 0.10512716996366572,
      "grad_norm": 111.39767456054688,
      "learning_rate": 1.9705699576234695e-05,
      "loss": 4.7667,
      "step": 2604
    },
    {
      "epoch": 0.10516754138070246,
      "grad_norm": 116.04476165771484,
      "learning_rate": 1.9705384602608108e-05,
      "loss": 5.3393,
      "step": 2605
    },
    {
      "epoch": 0.1052079127977392,
      "grad_norm": 101.13925170898438,
      "learning_rate": 1.97050694630421e-05,
      "loss": 4.8698,
      "step": 2606
    },
    {
      "epoch": 0.10524828421477594,
      "grad_norm": 95.3711929321289,
      "learning_rate": 1.9704754157542073e-05,
      "loss": 5.5236,
      "step": 2607
    },
    {
      "epoch": 0.10528865563181268,
      "grad_norm": 95.15745544433594,
      "learning_rate": 1.9704438686113412e-05,
      "loss": 6.0573,
      "step": 2608
    },
    {
      "epoch": 0.10532902704884942,
      "grad_norm": 84.91365051269531,
      "learning_rate": 1.970412304876151e-05,
      "loss": 5.6221,
      "step": 2609
    },
    {
      "epoch": 0.10536939846588615,
      "grad_norm": 108.67208099365234,
      "learning_rate": 1.9703807245491763e-05,
      "loss": 5.7014,
      "step": 2610
    },
    {
      "epoch": 0.10540976988292289,
      "grad_norm": 80.4068832397461,
      "learning_rate": 1.9703491276309575e-05,
      "loss": 6.3423,
      "step": 2611
    },
    {
      "epoch": 0.10545014129995962,
      "grad_norm": 100.92513275146484,
      "learning_rate": 1.9703175141220345e-05,
      "loss": 5.7356,
      "step": 2612
    },
    {
      "epoch": 0.10549051271699636,
      "grad_norm": 117.83810424804688,
      "learning_rate": 1.9702858840229476e-05,
      "loss": 6.1097,
      "step": 2613
    },
    {
      "epoch": 0.10553088413403311,
      "grad_norm": 194.0987548828125,
      "learning_rate": 1.970254237334238e-05,
      "loss": 8.1186,
      "step": 2614
    },
    {
      "epoch": 0.10557125555106985,
      "grad_norm": 92.41117095947266,
      "learning_rate": 1.9702225740564468e-05,
      "loss": 6.1437,
      "step": 2615
    },
    {
      "epoch": 0.10561162696810658,
      "grad_norm": 87.41345977783203,
      "learning_rate": 1.9701908941901153e-05,
      "loss": 5.2244,
      "step": 2616
    },
    {
      "epoch": 0.10565199838514332,
      "grad_norm": 132.5094451904297,
      "learning_rate": 1.9701591977357846e-05,
      "loss": 5.636,
      "step": 2617
    },
    {
      "epoch": 0.10569236980218005,
      "grad_norm": 188.30368041992188,
      "learning_rate": 1.9701274846939973e-05,
      "loss": 4.5185,
      "step": 2618
    },
    {
      "epoch": 0.10573274121921679,
      "grad_norm": 139.55931091308594,
      "learning_rate": 1.9700957550652953e-05,
      "loss": 6.8268,
      "step": 2619
    },
    {
      "epoch": 0.10577311263625354,
      "grad_norm": 116.17301940917969,
      "learning_rate": 1.970064008850222e-05,
      "loss": 5.1909,
      "step": 2620
    },
    {
      "epoch": 0.10581348405329027,
      "grad_norm": 106.5714340209961,
      "learning_rate": 1.9700322460493187e-05,
      "loss": 5.3907,
      "step": 2621
    },
    {
      "epoch": 0.10585385547032701,
      "grad_norm": 100.04379272460938,
      "learning_rate": 1.9700004666631293e-05,
      "loss": 5.5587,
      "step": 2622
    },
    {
      "epoch": 0.10589422688736375,
      "grad_norm": 108.55387115478516,
      "learning_rate": 1.9699686706921967e-05,
      "loss": 6.7545,
      "step": 2623
    },
    {
      "epoch": 0.10593459830440048,
      "grad_norm": 93.54020690917969,
      "learning_rate": 1.9699368581370654e-05,
      "loss": 6.0296,
      "step": 2624
    },
    {
      "epoch": 0.10597496972143722,
      "grad_norm": 85.30261993408203,
      "learning_rate": 1.9699050289982786e-05,
      "loss": 5.7285,
      "step": 2625
    },
    {
      "epoch": 0.10601534113847395,
      "grad_norm": 220.08755493164062,
      "learning_rate": 1.9698731832763806e-05,
      "loss": 6.3717,
      "step": 2626
    },
    {
      "epoch": 0.1060557125555107,
      "grad_norm": 114.74652099609375,
      "learning_rate": 1.9698413209719164e-05,
      "loss": 5.0139,
      "step": 2627
    },
    {
      "epoch": 0.10609608397254744,
      "grad_norm": 98.0353012084961,
      "learning_rate": 1.96980944208543e-05,
      "loss": 5.1844,
      "step": 2628
    },
    {
      "epoch": 0.10613645538958418,
      "grad_norm": 129.4350128173828,
      "learning_rate": 1.9697775466174667e-05,
      "loss": 4.5826,
      "step": 2629
    },
    {
      "epoch": 0.10617682680662091,
      "grad_norm": 164.28367614746094,
      "learning_rate": 1.969745634568572e-05,
      "loss": 5.916,
      "step": 2630
    },
    {
      "epoch": 0.10621719822365765,
      "grad_norm": 166.46437072753906,
      "learning_rate": 1.9697137059392916e-05,
      "loss": 5.6743,
      "step": 2631
    },
    {
      "epoch": 0.10625756964069438,
      "grad_norm": 184.59552001953125,
      "learning_rate": 1.9696817607301712e-05,
      "loss": 5.8661,
      "step": 2632
    },
    {
      "epoch": 0.10629794105773113,
      "grad_norm": 135.20249938964844,
      "learning_rate": 1.9696497989417575e-05,
      "loss": 6.238,
      "step": 2633
    },
    {
      "epoch": 0.10633831247476787,
      "grad_norm": 71.96373748779297,
      "learning_rate": 1.969617820574596e-05,
      "loss": 5.4644,
      "step": 2634
    },
    {
      "epoch": 0.1063786838918046,
      "grad_norm": 342.41375732421875,
      "learning_rate": 1.969585825629234e-05,
      "loss": 5.9151,
      "step": 2635
    },
    {
      "epoch": 0.10641905530884134,
      "grad_norm": 119.23564910888672,
      "learning_rate": 1.9695538141062186e-05,
      "loss": 6.0335,
      "step": 2636
    },
    {
      "epoch": 0.10645942672587808,
      "grad_norm": 166.64352416992188,
      "learning_rate": 1.9695217860060972e-05,
      "loss": 4.2994,
      "step": 2637
    },
    {
      "epoch": 0.10649979814291481,
      "grad_norm": 102.10687255859375,
      "learning_rate": 1.969489741329417e-05,
      "loss": 4.5038,
      "step": 2638
    },
    {
      "epoch": 0.10654016955995155,
      "grad_norm": 154.52581787109375,
      "learning_rate": 1.9694576800767263e-05,
      "loss": 6.2312,
      "step": 2639
    },
    {
      "epoch": 0.1065805409769883,
      "grad_norm": 129.77296447753906,
      "learning_rate": 1.969425602248573e-05,
      "loss": 6.0395,
      "step": 2640
    },
    {
      "epoch": 0.10662091239402503,
      "grad_norm": 212.9022979736328,
      "learning_rate": 1.9693935078455058e-05,
      "loss": 5.4298,
      "step": 2641
    },
    {
      "epoch": 0.10666128381106177,
      "grad_norm": 124.5867919921875,
      "learning_rate": 1.9693613968680732e-05,
      "loss": 6.3686,
      "step": 2642
    },
    {
      "epoch": 0.1067016552280985,
      "grad_norm": 151.66441345214844,
      "learning_rate": 1.969329269316825e-05,
      "loss": 3.9214,
      "step": 2643
    },
    {
      "epoch": 0.10674202664513524,
      "grad_norm": 108.09437561035156,
      "learning_rate": 1.969297125192309e-05,
      "loss": 6.6491,
      "step": 2644
    },
    {
      "epoch": 0.10678239806217198,
      "grad_norm": 89.48855590820312,
      "learning_rate": 1.969264964495076e-05,
      "loss": 5.044,
      "step": 2645
    },
    {
      "epoch": 0.10682276947920873,
      "grad_norm": 133.32205200195312,
      "learning_rate": 1.9692327872256753e-05,
      "loss": 5.5978,
      "step": 2646
    },
    {
      "epoch": 0.10686314089624546,
      "grad_norm": 105.73139953613281,
      "learning_rate": 1.9692005933846575e-05,
      "loss": 4.9005,
      "step": 2647
    },
    {
      "epoch": 0.1069035123132822,
      "grad_norm": 113.79742431640625,
      "learning_rate": 1.9691683829725723e-05,
      "loss": 4.343,
      "step": 2648
    },
    {
      "epoch": 0.10694388373031893,
      "grad_norm": 236.9605255126953,
      "learning_rate": 1.9691361559899714e-05,
      "loss": 4.9184,
      "step": 2649
    },
    {
      "epoch": 0.10698425514735567,
      "grad_norm": 74.60645294189453,
      "learning_rate": 1.969103912437405e-05,
      "loss": 5.4579,
      "step": 2650
    },
    {
      "epoch": 0.1070246265643924,
      "grad_norm": 222.4517364501953,
      "learning_rate": 1.969071652315425e-05,
      "loss": 5.7077,
      "step": 2651
    },
    {
      "epoch": 0.10706499798142914,
      "grad_norm": 93.40264129638672,
      "learning_rate": 1.9690393756245826e-05,
      "loss": 5.7468,
      "step": 2652
    },
    {
      "epoch": 0.10710536939846589,
      "grad_norm": 355.71563720703125,
      "learning_rate": 1.96900708236543e-05,
      "loss": 5.6881,
      "step": 2653
    },
    {
      "epoch": 0.10714574081550263,
      "grad_norm": 107.8689956665039,
      "learning_rate": 1.968974772538519e-05,
      "loss": 6.4172,
      "step": 2654
    },
    {
      "epoch": 0.10718611223253936,
      "grad_norm": 152.89169311523438,
      "learning_rate": 1.9689424461444018e-05,
      "loss": 6.2046,
      "step": 2655
    },
    {
      "epoch": 0.1072264836495761,
      "grad_norm": 151.7646942138672,
      "learning_rate": 1.9689101031836316e-05,
      "loss": 6.5479,
      "step": 2656
    },
    {
      "epoch": 0.10726685506661283,
      "grad_norm": 112.82920837402344,
      "learning_rate": 1.9688777436567612e-05,
      "loss": 6.5639,
      "step": 2657
    },
    {
      "epoch": 0.10730722648364957,
      "grad_norm": 139.19729614257812,
      "learning_rate": 1.9688453675643438e-05,
      "loss": 5.1304,
      "step": 2658
    },
    {
      "epoch": 0.10734759790068632,
      "grad_norm": 88.89505004882812,
      "learning_rate": 1.968812974906933e-05,
      "loss": 5.9187,
      "step": 2659
    },
    {
      "epoch": 0.10738796931772306,
      "grad_norm": 94.94894409179688,
      "learning_rate": 1.9687805656850827e-05,
      "loss": 4.4399,
      "step": 2660
    },
    {
      "epoch": 0.10742834073475979,
      "grad_norm": 144.69229125976562,
      "learning_rate": 1.968748139899347e-05,
      "loss": 5.3049,
      "step": 2661
    },
    {
      "epoch": 0.10746871215179653,
      "grad_norm": 195.33795166015625,
      "learning_rate": 1.9687156975502803e-05,
      "loss": 4.7895,
      "step": 2662
    },
    {
      "epoch": 0.10750908356883326,
      "grad_norm": 197.56130981445312,
      "learning_rate": 1.9686832386384376e-05,
      "loss": 4.9181,
      "step": 2663
    },
    {
      "epoch": 0.10754945498587,
      "grad_norm": 80.0417251586914,
      "learning_rate": 1.9686507631643732e-05,
      "loss": 4.6066,
      "step": 2664
    },
    {
      "epoch": 0.10758982640290674,
      "grad_norm": 158.60853576660156,
      "learning_rate": 1.968618271128643e-05,
      "loss": 4.4774,
      "step": 2665
    },
    {
      "epoch": 0.10763019781994349,
      "grad_norm": 137.49774169921875,
      "learning_rate": 1.9685857625318017e-05,
      "loss": 5.2282,
      "step": 2666
    },
    {
      "epoch": 0.10767056923698022,
      "grad_norm": 173.4360809326172,
      "learning_rate": 1.968553237374406e-05,
      "loss": 4.4282,
      "step": 2667
    },
    {
      "epoch": 0.10771094065401696,
      "grad_norm": 153.3223419189453,
      "learning_rate": 1.968520695657012e-05,
      "loss": 6.0245,
      "step": 2668
    },
    {
      "epoch": 0.10775131207105369,
      "grad_norm": 103.83821868896484,
      "learning_rate": 1.9684881373801755e-05,
      "loss": 5.3274,
      "step": 2669
    },
    {
      "epoch": 0.10779168348809043,
      "grad_norm": 216.6870574951172,
      "learning_rate": 1.9684555625444532e-05,
      "loss": 5.8951,
      "step": 2670
    },
    {
      "epoch": 0.10783205490512716,
      "grad_norm": 92.9861068725586,
      "learning_rate": 1.9684229711504026e-05,
      "loss": 5.986,
      "step": 2671
    },
    {
      "epoch": 0.10787242632216391,
      "grad_norm": 132.7876434326172,
      "learning_rate": 1.9683903631985805e-05,
      "loss": 4.7427,
      "step": 2672
    },
    {
      "epoch": 0.10791279773920065,
      "grad_norm": 92.98108673095703,
      "learning_rate": 1.9683577386895444e-05,
      "loss": 5.807,
      "step": 2673
    },
    {
      "epoch": 0.10795316915623739,
      "grad_norm": 169.07704162597656,
      "learning_rate": 1.9683250976238525e-05,
      "loss": 6.332,
      "step": 2674
    },
    {
      "epoch": 0.10799354057327412,
      "grad_norm": 151.0231170654297,
      "learning_rate": 1.9682924400020623e-05,
      "loss": 4.4067,
      "step": 2675
    },
    {
      "epoch": 0.10803391199031086,
      "grad_norm": 129.09718322753906,
      "learning_rate": 1.968259765824733e-05,
      "loss": 4.92,
      "step": 2676
    },
    {
      "epoch": 0.1080742834073476,
      "grad_norm": 131.13685607910156,
      "learning_rate": 1.9682270750924223e-05,
      "loss": 5.505,
      "step": 2677
    },
    {
      "epoch": 0.10811465482438433,
      "grad_norm": 87.21732330322266,
      "learning_rate": 1.9681943678056898e-05,
      "loss": 5.8399,
      "step": 2678
    },
    {
      "epoch": 0.10815502624142108,
      "grad_norm": 110.43151092529297,
      "learning_rate": 1.9681616439650945e-05,
      "loss": 7.1143,
      "step": 2679
    },
    {
      "epoch": 0.10819539765845781,
      "grad_norm": 107.86040496826172,
      "learning_rate": 1.968128903571196e-05,
      "loss": 6.4973,
      "step": 2680
    },
    {
      "epoch": 0.10823576907549455,
      "grad_norm": 144.8408203125,
      "learning_rate": 1.968096146624554e-05,
      "loss": 5.446,
      "step": 2681
    },
    {
      "epoch": 0.10827614049253129,
      "grad_norm": 88.34674072265625,
      "learning_rate": 1.9680633731257282e-05,
      "loss": 4.8272,
      "step": 2682
    },
    {
      "epoch": 0.10831651190956802,
      "grad_norm": 96.61758422851562,
      "learning_rate": 1.9680305830752793e-05,
      "loss": 5.1653,
      "step": 2683
    },
    {
      "epoch": 0.10835688332660476,
      "grad_norm": 188.47877502441406,
      "learning_rate": 1.9679977764737684e-05,
      "loss": 7.1439,
      "step": 2684
    },
    {
      "epoch": 0.10839725474364151,
      "grad_norm": 201.47262573242188,
      "learning_rate": 1.967964953321756e-05,
      "loss": 6.0858,
      "step": 2685
    },
    {
      "epoch": 0.10843762616067824,
      "grad_norm": 99.53646087646484,
      "learning_rate": 1.967932113619803e-05,
      "loss": 5.5808,
      "step": 2686
    },
    {
      "epoch": 0.10847799757771498,
      "grad_norm": 82.04022979736328,
      "learning_rate": 1.967899257368471e-05,
      "loss": 5.3701,
      "step": 2687
    },
    {
      "epoch": 0.10851836899475172,
      "grad_norm": 79.04789733886719,
      "learning_rate": 1.967866384568322e-05,
      "loss": 6.5256,
      "step": 2688
    },
    {
      "epoch": 0.10855874041178845,
      "grad_norm": 111.82592010498047,
      "learning_rate": 1.967833495219918e-05,
      "loss": 3.9067,
      "step": 2689
    },
    {
      "epoch": 0.10859911182882519,
      "grad_norm": 83.71015930175781,
      "learning_rate": 1.9678005893238216e-05,
      "loss": 5.8508,
      "step": 2690
    },
    {
      "epoch": 0.10863948324586192,
      "grad_norm": 99.05358123779297,
      "learning_rate": 1.9677676668805947e-05,
      "loss": 4.768,
      "step": 2691
    },
    {
      "epoch": 0.10867985466289867,
      "grad_norm": 87.28068542480469,
      "learning_rate": 1.9677347278908004e-05,
      "loss": 5.883,
      "step": 2692
    },
    {
      "epoch": 0.10872022607993541,
      "grad_norm": 84.33301544189453,
      "learning_rate": 1.9677017723550026e-05,
      "loss": 4.5151,
      "step": 2693
    },
    {
      "epoch": 0.10876059749697214,
      "grad_norm": 132.02684020996094,
      "learning_rate": 1.967668800273764e-05,
      "loss": 5.3455,
      "step": 2694
    },
    {
      "epoch": 0.10880096891400888,
      "grad_norm": 285.9354553222656,
      "learning_rate": 1.9676358116476485e-05,
      "loss": 7.0618,
      "step": 2695
    },
    {
      "epoch": 0.10884134033104562,
      "grad_norm": 147.73843383789062,
      "learning_rate": 1.9676028064772202e-05,
      "loss": 5.5057,
      "step": 2696
    },
    {
      "epoch": 0.10888171174808235,
      "grad_norm": 122.8831558227539,
      "learning_rate": 1.9675697847630438e-05,
      "loss": 5.1902,
      "step": 2697
    },
    {
      "epoch": 0.1089220831651191,
      "grad_norm": 101.16261291503906,
      "learning_rate": 1.967536746505683e-05,
      "loss": 5.302,
      "step": 2698
    },
    {
      "epoch": 0.10896245458215584,
      "grad_norm": 78.31468200683594,
      "learning_rate": 1.9675036917057035e-05,
      "loss": 5.453,
      "step": 2699
    },
    {
      "epoch": 0.10900282599919257,
      "grad_norm": 152.2611846923828,
      "learning_rate": 1.96747062036367e-05,
      "loss": 6.0572,
      "step": 2700
    },
    {
      "epoch": 0.10904319741622931,
      "grad_norm": 133.39169311523438,
      "learning_rate": 1.9674375324801486e-05,
      "loss": 6.0495,
      "step": 2701
    },
    {
      "epoch": 0.10908356883326605,
      "grad_norm": 180.81138610839844,
      "learning_rate": 1.967404428055704e-05,
      "loss": 4.8888,
      "step": 2702
    },
    {
      "epoch": 0.10912394025030278,
      "grad_norm": 205.75074768066406,
      "learning_rate": 1.9673713070909028e-05,
      "loss": 4.2373,
      "step": 2703
    },
    {
      "epoch": 0.10916431166733952,
      "grad_norm": 136.2828826904297,
      "learning_rate": 1.967338169586311e-05,
      "loss": 7.0191,
      "step": 2704
    },
    {
      "epoch": 0.10920468308437627,
      "grad_norm": 153.48793029785156,
      "learning_rate": 1.967305015542496e-05,
      "loss": 6.4802,
      "step": 2705
    },
    {
      "epoch": 0.109245054501413,
      "grad_norm": 77.16936492919922,
      "learning_rate": 1.9672718449600238e-05,
      "loss": 4.667,
      "step": 2706
    },
    {
      "epoch": 0.10928542591844974,
      "grad_norm": 132.62249755859375,
      "learning_rate": 1.9672386578394615e-05,
      "loss": 6.2785,
      "step": 2707
    },
    {
      "epoch": 0.10932579733548647,
      "grad_norm": 102.31893920898438,
      "learning_rate": 1.967205454181377e-05,
      "loss": 6.309,
      "step": 2708
    },
    {
      "epoch": 0.10936616875252321,
      "grad_norm": 206.86143493652344,
      "learning_rate": 1.9671722339863378e-05,
      "loss": 3.8764,
      "step": 2709
    },
    {
      "epoch": 0.10940654016955995,
      "grad_norm": 171.95347595214844,
      "learning_rate": 1.967138997254912e-05,
      "loss": 4.6988,
      "step": 2710
    },
    {
      "epoch": 0.1094469115865967,
      "grad_norm": 283.2774353027344,
      "learning_rate": 1.967105743987668e-05,
      "loss": 5.4987,
      "step": 2711
    },
    {
      "epoch": 0.10948728300363343,
      "grad_norm": 128.2808380126953,
      "learning_rate": 1.9670724741851737e-05,
      "loss": 6.4836,
      "step": 2712
    },
    {
      "epoch": 0.10952765442067017,
      "grad_norm": 118.20075225830078,
      "learning_rate": 1.9670391878479983e-05,
      "loss": 4.838,
      "step": 2713
    },
    {
      "epoch": 0.1095680258377069,
      "grad_norm": 117.70574951171875,
      "learning_rate": 1.9670058849767114e-05,
      "loss": 4.9787,
      "step": 2714
    },
    {
      "epoch": 0.10960839725474364,
      "grad_norm": 114.47407531738281,
      "learning_rate": 1.966972565571882e-05,
      "loss": 5.5536,
      "step": 2715
    },
    {
      "epoch": 0.10964876867178038,
      "grad_norm": 207.23367309570312,
      "learning_rate": 1.9669392296340795e-05,
      "loss": 5.5549,
      "step": 2716
    },
    {
      "epoch": 0.10968914008881711,
      "grad_norm": 119.44831085205078,
      "learning_rate": 1.9669058771638744e-05,
      "loss": 6.0104,
      "step": 2717
    },
    {
      "epoch": 0.10972951150585386,
      "grad_norm": 164.21868896484375,
      "learning_rate": 1.9668725081618362e-05,
      "loss": 4.6507,
      "step": 2718
    },
    {
      "epoch": 0.1097698829228906,
      "grad_norm": 82.98907470703125,
      "learning_rate": 1.9668391226285365e-05,
      "loss": 5.4563,
      "step": 2719
    },
    {
      "epoch": 0.10981025433992733,
      "grad_norm": 208.5867462158203,
      "learning_rate": 1.966805720564545e-05,
      "loss": 6.5007,
      "step": 2720
    },
    {
      "epoch": 0.10985062575696407,
      "grad_norm": 92.16706085205078,
      "learning_rate": 1.9667723019704335e-05,
      "loss": 5.9099,
      "step": 2721
    },
    {
      "epoch": 0.1098909971740008,
      "grad_norm": 87.0771255493164,
      "learning_rate": 1.9667388668467734e-05,
      "loss": 6.9185,
      "step": 2722
    },
    {
      "epoch": 0.10993136859103754,
      "grad_norm": 129.11923217773438,
      "learning_rate": 1.9667054151941364e-05,
      "loss": 6.1049,
      "step": 2723
    },
    {
      "epoch": 0.10997174000807429,
      "grad_norm": 114.85309600830078,
      "learning_rate": 1.9666719470130938e-05,
      "loss": 6.095,
      "step": 2724
    },
    {
      "epoch": 0.11001211142511103,
      "grad_norm": 131.4066619873047,
      "learning_rate": 1.9666384623042184e-05,
      "loss": 6.5733,
      "step": 2725
    },
    {
      "epoch": 0.11005248284214776,
      "grad_norm": 177.86988830566406,
      "learning_rate": 1.9666049610680823e-05,
      "loss": 5.4493,
      "step": 2726
    },
    {
      "epoch": 0.1100928542591845,
      "grad_norm": 87.15799713134766,
      "learning_rate": 1.966571443305259e-05,
      "loss": 5.6508,
      "step": 2727
    },
    {
      "epoch": 0.11013322567622123,
      "grad_norm": 92.76204681396484,
      "learning_rate": 1.9665379090163207e-05,
      "loss": 6.1289,
      "step": 2728
    },
    {
      "epoch": 0.11017359709325797,
      "grad_norm": 71.47294616699219,
      "learning_rate": 1.966504358201841e-05,
      "loss": 6.5166,
      "step": 2729
    },
    {
      "epoch": 0.1102139685102947,
      "grad_norm": 121.33773040771484,
      "learning_rate": 1.9664707908623946e-05,
      "loss": 6.1371,
      "step": 2730
    },
    {
      "epoch": 0.11025433992733145,
      "grad_norm": 135.74302673339844,
      "learning_rate": 1.966437206998554e-05,
      "loss": 5.3622,
      "step": 2731
    },
    {
      "epoch": 0.11029471134436819,
      "grad_norm": 98.13872528076172,
      "learning_rate": 1.966403606610894e-05,
      "loss": 5.3332,
      "step": 2732
    },
    {
      "epoch": 0.11033508276140493,
      "grad_norm": 80.51935577392578,
      "learning_rate": 1.9663699896999892e-05,
      "loss": 5.1662,
      "step": 2733
    },
    {
      "epoch": 0.11037545417844166,
      "grad_norm": 92.3671646118164,
      "learning_rate": 1.966336356266414e-05,
      "loss": 3.5104,
      "step": 2734
    },
    {
      "epoch": 0.1104158255954784,
      "grad_norm": 88.51487731933594,
      "learning_rate": 1.9663027063107436e-05,
      "loss": 5.9882,
      "step": 2735
    },
    {
      "epoch": 0.11045619701251513,
      "grad_norm": 67.61792755126953,
      "learning_rate": 1.9662690398335536e-05,
      "loss": 6.1057,
      "step": 2736
    },
    {
      "epoch": 0.11049656842955188,
      "grad_norm": 171.9068603515625,
      "learning_rate": 1.9662353568354193e-05,
      "loss": 5.8131,
      "step": 2737
    },
    {
      "epoch": 0.11053693984658862,
      "grad_norm": 170.0095672607422,
      "learning_rate": 1.9662016573169167e-05,
      "loss": 6.7173,
      "step": 2738
    },
    {
      "epoch": 0.11057731126362536,
      "grad_norm": 86.67150115966797,
      "learning_rate": 1.966167941278622e-05,
      "loss": 5.9069,
      "step": 2739
    },
    {
      "epoch": 0.11061768268066209,
      "grad_norm": 76.1494369506836,
      "learning_rate": 1.9661342087211118e-05,
      "loss": 5.2241,
      "step": 2740
    },
    {
      "epoch": 0.11065805409769883,
      "grad_norm": 73.80950164794922,
      "learning_rate": 1.9661004596449625e-05,
      "loss": 4.2618,
      "step": 2741
    },
    {
      "epoch": 0.11069842551473556,
      "grad_norm": 91.28389739990234,
      "learning_rate": 1.966066694050752e-05,
      "loss": 6.7515,
      "step": 2742
    },
    {
      "epoch": 0.1107387969317723,
      "grad_norm": 149.15283203125,
      "learning_rate": 1.9660329119390563e-05,
      "loss": 6.5299,
      "step": 2743
    },
    {
      "epoch": 0.11077916834880905,
      "grad_norm": 90.18011474609375,
      "learning_rate": 1.9659991133104536e-05,
      "loss": 4.7848,
      "step": 2744
    },
    {
      "epoch": 0.11081953976584578,
      "grad_norm": 102.95455932617188,
      "learning_rate": 1.965965298165522e-05,
      "loss": 5.9288,
      "step": 2745
    },
    {
      "epoch": 0.11085991118288252,
      "grad_norm": 94.59963989257812,
      "learning_rate": 1.9659314665048398e-05,
      "loss": 7.498,
      "step": 2746
    },
    {
      "epoch": 0.11090028259991926,
      "grad_norm": 77.82449340820312,
      "learning_rate": 1.9658976183289846e-05,
      "loss": 6.4859,
      "step": 2747
    },
    {
      "epoch": 0.11094065401695599,
      "grad_norm": 101.38864135742188,
      "learning_rate": 1.965863753638536e-05,
      "loss": 3.9204,
      "step": 2748
    },
    {
      "epoch": 0.11098102543399273,
      "grad_norm": 200.99313354492188,
      "learning_rate": 1.9658298724340725e-05,
      "loss": 5.4064,
      "step": 2749
    },
    {
      "epoch": 0.11102139685102948,
      "grad_norm": 699.804931640625,
      "learning_rate": 1.9657959747161737e-05,
      "loss": 5.9012,
      "step": 2750
    },
    {
      "epoch": 0.11106176826806621,
      "grad_norm": 86.67945861816406,
      "learning_rate": 1.965762060485419e-05,
      "loss": 5.3671,
      "step": 2751
    },
    {
      "epoch": 0.11110213968510295,
      "grad_norm": 110.31659698486328,
      "learning_rate": 1.9657281297423883e-05,
      "loss": 4.6088,
      "step": 2752
    },
    {
      "epoch": 0.11114251110213968,
      "grad_norm": 208.73526000976562,
      "learning_rate": 1.9656941824876615e-05,
      "loss": 5.5762,
      "step": 2753
    },
    {
      "epoch": 0.11118288251917642,
      "grad_norm": 89.10165405273438,
      "learning_rate": 1.9656602187218198e-05,
      "loss": 6.614,
      "step": 2754
    },
    {
      "epoch": 0.11122325393621316,
      "grad_norm": 68.56206512451172,
      "learning_rate": 1.9656262384454427e-05,
      "loss": 5.4182,
      "step": 2755
    },
    {
      "epoch": 0.11126362535324989,
      "grad_norm": 159.51451110839844,
      "learning_rate": 1.965592241659112e-05,
      "loss": 6.1137,
      "step": 2756
    },
    {
      "epoch": 0.11130399677028664,
      "grad_norm": 111.06620025634766,
      "learning_rate": 1.9655582283634088e-05,
      "loss": 4.4683,
      "step": 2757
    },
    {
      "epoch": 0.11134436818732338,
      "grad_norm": 179.10015869140625,
      "learning_rate": 1.965524198558915e-05,
      "loss": 6.752,
      "step": 2758
    },
    {
      "epoch": 0.11138473960436011,
      "grad_norm": 77.77751159667969,
      "learning_rate": 1.965490152246212e-05,
      "loss": 5.0539,
      "step": 2759
    },
    {
      "epoch": 0.11142511102139685,
      "grad_norm": 173.34475708007812,
      "learning_rate": 1.9654560894258816e-05,
      "loss": 5.4098,
      "step": 2760
    },
    {
      "epoch": 0.11146548243843359,
      "grad_norm": 105.49647521972656,
      "learning_rate": 1.9654220100985064e-05,
      "loss": 5.6273,
      "step": 2761
    },
    {
      "epoch": 0.11150585385547032,
      "grad_norm": 177.3513641357422,
      "learning_rate": 1.9653879142646695e-05,
      "loss": 4.8949,
      "step": 2762
    },
    {
      "epoch": 0.11154622527250707,
      "grad_norm": 173.9185791015625,
      "learning_rate": 1.965353801924954e-05,
      "loss": 5.2876,
      "step": 2763
    },
    {
      "epoch": 0.11158659668954381,
      "grad_norm": 214.3805694580078,
      "learning_rate": 1.9653196730799423e-05,
      "loss": 7.2642,
      "step": 2764
    },
    {
      "epoch": 0.11162696810658054,
      "grad_norm": 142.41195678710938,
      "learning_rate": 1.9652855277302185e-05,
      "loss": 6.5804,
      "step": 2765
    },
    {
      "epoch": 0.11166733952361728,
      "grad_norm": 103.77876281738281,
      "learning_rate": 1.9652513658763663e-05,
      "loss": 5.6285,
      "step": 2766
    },
    {
      "epoch": 0.11170771094065401,
      "grad_norm": 136.21725463867188,
      "learning_rate": 1.96521718751897e-05,
      "loss": 5.6197,
      "step": 2767
    },
    {
      "epoch": 0.11174808235769075,
      "grad_norm": 96.31502532958984,
      "learning_rate": 1.9651829926586132e-05,
      "loss": 5.3485,
      "step": 2768
    },
    {
      "epoch": 0.11178845377472749,
      "grad_norm": 146.92745971679688,
      "learning_rate": 1.9651487812958815e-05,
      "loss": 5.5799,
      "step": 2769
    },
    {
      "epoch": 0.11182882519176424,
      "grad_norm": 65.36015319824219,
      "learning_rate": 1.9651145534313593e-05,
      "loss": 6.408,
      "step": 2770
    },
    {
      "epoch": 0.11186919660880097,
      "grad_norm": 146.04733276367188,
      "learning_rate": 1.965080309065632e-05,
      "loss": 5.2792,
      "step": 2771
    },
    {
      "epoch": 0.11190956802583771,
      "grad_norm": 78.67916870117188,
      "learning_rate": 1.9650460481992848e-05,
      "loss": 5.5544,
      "step": 2772
    },
    {
      "epoch": 0.11194993944287444,
      "grad_norm": 114.4932632446289,
      "learning_rate": 1.965011770832904e-05,
      "loss": 6.2513,
      "step": 2773
    },
    {
      "epoch": 0.11199031085991118,
      "grad_norm": 111.65922546386719,
      "learning_rate": 1.964977476967075e-05,
      "loss": 6.2636,
      "step": 2774
    },
    {
      "epoch": 0.11203068227694792,
      "grad_norm": 90.27752685546875,
      "learning_rate": 1.964943166602385e-05,
      "loss": 5.8013,
      "step": 2775
    },
    {
      "epoch": 0.11207105369398467,
      "grad_norm": 90.30661010742188,
      "learning_rate": 1.96490883973942e-05,
      "loss": 4.5025,
      "step": 2776
    },
    {
      "epoch": 0.1121114251110214,
      "grad_norm": 87.75947570800781,
      "learning_rate": 1.964874496378767e-05,
      "loss": 4.9179,
      "step": 2777
    },
    {
      "epoch": 0.11215179652805814,
      "grad_norm": 109.86922454833984,
      "learning_rate": 1.9648401365210133e-05,
      "loss": 4.7651,
      "step": 2778
    },
    {
      "epoch": 0.11219216794509487,
      "grad_norm": 85.62980651855469,
      "learning_rate": 1.9648057601667463e-05,
      "loss": 4.0256,
      "step": 2779
    },
    {
      "epoch": 0.11223253936213161,
      "grad_norm": 120.26880645751953,
      "learning_rate": 1.9647713673165537e-05,
      "loss": 5.6492,
      "step": 2780
    },
    {
      "epoch": 0.11227291077916834,
      "grad_norm": 173.71974182128906,
      "learning_rate": 1.964736957971024e-05,
      "loss": 6.8869,
      "step": 2781
    },
    {
      "epoch": 0.11231328219620508,
      "grad_norm": 127.70679473876953,
      "learning_rate": 1.9647025321307446e-05,
      "loss": 5.3732,
      "step": 2782
    },
    {
      "epoch": 0.11235365361324183,
      "grad_norm": 135.20970153808594,
      "learning_rate": 1.964668089796305e-05,
      "loss": 5.0591,
      "step": 2783
    },
    {
      "epoch": 0.11239402503027857,
      "grad_norm": 70.73987579345703,
      "learning_rate": 1.9646336309682937e-05,
      "loss": 4.4784,
      "step": 2784
    },
    {
      "epoch": 0.1124343964473153,
      "grad_norm": 111.53882598876953,
      "learning_rate": 1.9645991556473e-05,
      "loss": 5.5901,
      "step": 2785
    },
    {
      "epoch": 0.11247476786435204,
      "grad_norm": 153.1847381591797,
      "learning_rate": 1.964564663833913e-05,
      "loss": 5.2048,
      "step": 2786
    },
    {
      "epoch": 0.11251513928138877,
      "grad_norm": 81.9525375366211,
      "learning_rate": 1.9645301555287227e-05,
      "loss": 7.1461,
      "step": 2787
    },
    {
      "epoch": 0.11255551069842551,
      "grad_norm": 120.13416290283203,
      "learning_rate": 1.964495630732319e-05,
      "loss": 6.8425,
      "step": 2788
    },
    {
      "epoch": 0.11259588211546226,
      "grad_norm": 179.29165649414062,
      "learning_rate": 1.9644610894452924e-05,
      "loss": 5.0358,
      "step": 2789
    },
    {
      "epoch": 0.112636253532499,
      "grad_norm": 88.42350006103516,
      "learning_rate": 1.9644265316682334e-05,
      "loss": 5.8868,
      "step": 2790
    },
    {
      "epoch": 0.11267662494953573,
      "grad_norm": 159.350341796875,
      "learning_rate": 1.9643919574017327e-05,
      "loss": 5.5367,
      "step": 2791
    },
    {
      "epoch": 0.11271699636657247,
      "grad_norm": 119.60884094238281,
      "learning_rate": 1.9643573666463814e-05,
      "loss": 6.5571,
      "step": 2792
    },
    {
      "epoch": 0.1127573677836092,
      "grad_norm": 79.23350524902344,
      "learning_rate": 1.9643227594027715e-05,
      "loss": 5.4646,
      "step": 2793
    },
    {
      "epoch": 0.11279773920064594,
      "grad_norm": 128.7123260498047,
      "learning_rate": 1.964288135671494e-05,
      "loss": 4.8285,
      "step": 2794
    },
    {
      "epoch": 0.11283811061768267,
      "grad_norm": 153.421875,
      "learning_rate": 1.964253495453141e-05,
      "loss": 4.8945,
      "step": 2795
    },
    {
      "epoch": 0.11287848203471942,
      "grad_norm": 89.73920440673828,
      "learning_rate": 1.9642188387483052e-05,
      "loss": 5.5002,
      "step": 2796
    },
    {
      "epoch": 0.11291885345175616,
      "grad_norm": 86.42562103271484,
      "learning_rate": 1.9641841655575787e-05,
      "loss": 5.2801,
      "step": 2797
    },
    {
      "epoch": 0.1129592248687929,
      "grad_norm": 140.92825317382812,
      "learning_rate": 1.964149475881555e-05,
      "loss": 7.4069,
      "step": 2798
    },
    {
      "epoch": 0.11299959628582963,
      "grad_norm": 92.03694152832031,
      "learning_rate": 1.9641147697208258e-05,
      "loss": 4.2304,
      "step": 2799
    },
    {
      "epoch": 0.11303996770286637,
      "grad_norm": 546.4337768554688,
      "learning_rate": 1.9640800470759858e-05,
      "loss": 4.072,
      "step": 2800
    },
    {
      "epoch": 0.1130803391199031,
      "grad_norm": 94.56942749023438,
      "learning_rate": 1.9640453079476286e-05,
      "loss": 5.1217,
      "step": 2801
    },
    {
      "epoch": 0.11312071053693985,
      "grad_norm": 71.20248413085938,
      "learning_rate": 1.9640105523363476e-05,
      "loss": 6.3007,
      "step": 2802
    },
    {
      "epoch": 0.11316108195397659,
      "grad_norm": 115.83833312988281,
      "learning_rate": 1.9639757802427373e-05,
      "loss": 6.0849,
      "step": 2803
    },
    {
      "epoch": 0.11320145337101332,
      "grad_norm": 88.13482666015625,
      "learning_rate": 1.9639409916673924e-05,
      "loss": 6.7807,
      "step": 2804
    },
    {
      "epoch": 0.11324182478805006,
      "grad_norm": 116.17980194091797,
      "learning_rate": 1.963906186610907e-05,
      "loss": 5.4092,
      "step": 2805
    },
    {
      "epoch": 0.1132821962050868,
      "grad_norm": 101.0732650756836,
      "learning_rate": 1.9638713650738772e-05,
      "loss": 5.271,
      "step": 2806
    },
    {
      "epoch": 0.11332256762212353,
      "grad_norm": 115.60160064697266,
      "learning_rate": 1.9638365270568977e-05,
      "loss": 5.1927,
      "step": 2807
    },
    {
      "epoch": 0.11336293903916027,
      "grad_norm": 124.50871276855469,
      "learning_rate": 1.9638016725605642e-05,
      "loss": 5.769,
      "step": 2808
    },
    {
      "epoch": 0.11340331045619702,
      "grad_norm": 115.40962219238281,
      "learning_rate": 1.9637668015854728e-05,
      "loss": 4.2861,
      "step": 2809
    },
    {
      "epoch": 0.11344368187323375,
      "grad_norm": 148.73101806640625,
      "learning_rate": 1.9637319141322197e-05,
      "loss": 5.3524,
      "step": 2810
    },
    {
      "epoch": 0.11348405329027049,
      "grad_norm": 120.39436340332031,
      "learning_rate": 1.9636970102014015e-05,
      "loss": 5.0248,
      "step": 2811
    },
    {
      "epoch": 0.11352442470730723,
      "grad_norm": 67.48529815673828,
      "learning_rate": 1.9636620897936143e-05,
      "loss": 4.8558,
      "step": 2812
    },
    {
      "epoch": 0.11356479612434396,
      "grad_norm": 123.4802474975586,
      "learning_rate": 1.963627152909456e-05,
      "loss": 4.7704,
      "step": 2813
    },
    {
      "epoch": 0.1136051675413807,
      "grad_norm": 98.13111877441406,
      "learning_rate": 1.9635921995495238e-05,
      "loss": 4.9079,
      "step": 2814
    },
    {
      "epoch": 0.11364553895841745,
      "grad_norm": 199.27493286132812,
      "learning_rate": 1.963557229714415e-05,
      "loss": 6.4064,
      "step": 2815
    },
    {
      "epoch": 0.11368591037545418,
      "grad_norm": 101.43302154541016,
      "learning_rate": 1.9635222434047273e-05,
      "loss": 6.2528,
      "step": 2816
    },
    {
      "epoch": 0.11372628179249092,
      "grad_norm": 110.05504608154297,
      "learning_rate": 1.9634872406210595e-05,
      "loss": 5.0841,
      "step": 2817
    },
    {
      "epoch": 0.11376665320952765,
      "grad_norm": 109.40988159179688,
      "learning_rate": 1.96345222136401e-05,
      "loss": 6.1433,
      "step": 2818
    },
    {
      "epoch": 0.11380702462656439,
      "grad_norm": 132.29324340820312,
      "learning_rate": 1.963417185634177e-05,
      "loss": 6.5193,
      "step": 2819
    },
    {
      "epoch": 0.11384739604360113,
      "grad_norm": 103.4073486328125,
      "learning_rate": 1.9633821334321603e-05,
      "loss": 4.7657,
      "step": 2820
    },
    {
      "epoch": 0.11388776746063786,
      "grad_norm": 98.85749053955078,
      "learning_rate": 1.9633470647585584e-05,
      "loss": 6.0999,
      "step": 2821
    },
    {
      "epoch": 0.11392813887767461,
      "grad_norm": 92.59516143798828,
      "learning_rate": 1.9633119796139713e-05,
      "loss": 5.4791,
      "step": 2822
    },
    {
      "epoch": 0.11396851029471135,
      "grad_norm": 88.88692474365234,
      "learning_rate": 1.963276877998999e-05,
      "loss": 6.5281,
      "step": 2823
    },
    {
      "epoch": 0.11400888171174808,
      "grad_norm": 147.3623504638672,
      "learning_rate": 1.9632417599142414e-05,
      "loss": 8.1979,
      "step": 2824
    },
    {
      "epoch": 0.11404925312878482,
      "grad_norm": 103.4726333618164,
      "learning_rate": 1.9632066253602987e-05,
      "loss": 5.2926,
      "step": 2825
    },
    {
      "epoch": 0.11408962454582156,
      "grad_norm": 81.62506866455078,
      "learning_rate": 1.9631714743377726e-05,
      "loss": 4.0655,
      "step": 2826
    },
    {
      "epoch": 0.11412999596285829,
      "grad_norm": 111.88246154785156,
      "learning_rate": 1.963136306847263e-05,
      "loss": 5.1765,
      "step": 2827
    },
    {
      "epoch": 0.11417036737989504,
      "grad_norm": 77.2559814453125,
      "learning_rate": 1.963101122889372e-05,
      "loss": 4.3034,
      "step": 2828
    },
    {
      "epoch": 0.11421073879693178,
      "grad_norm": 99.50736999511719,
      "learning_rate": 1.9630659224647005e-05,
      "loss": 5.8773,
      "step": 2829
    },
    {
      "epoch": 0.11425111021396851,
      "grad_norm": 115.68878173828125,
      "learning_rate": 1.963030705573851e-05,
      "loss": 4.7484,
      "step": 2830
    },
    {
      "epoch": 0.11429148163100525,
      "grad_norm": 93.8062973022461,
      "learning_rate": 1.9629954722174247e-05,
      "loss": 6.1598,
      "step": 2831
    },
    {
      "epoch": 0.11433185304804198,
      "grad_norm": 155.60586547851562,
      "learning_rate": 1.962960222396025e-05,
      "loss": 5.0417,
      "step": 2832
    },
    {
      "epoch": 0.11437222446507872,
      "grad_norm": 91.51184844970703,
      "learning_rate": 1.9629249561102538e-05,
      "loss": 5.0009,
      "step": 2833
    },
    {
      "epoch": 0.11441259588211546,
      "grad_norm": 106.79249572753906,
      "learning_rate": 1.962889673360715e-05,
      "loss": 5.1362,
      "step": 2834
    },
    {
      "epoch": 0.1144529672991522,
      "grad_norm": 152.72242736816406,
      "learning_rate": 1.9628543741480107e-05,
      "loss": 4.5382,
      "step": 2835
    },
    {
      "epoch": 0.11449333871618894,
      "grad_norm": 81.97557830810547,
      "learning_rate": 1.9628190584727456e-05,
      "loss": 5.2015,
      "step": 2836
    },
    {
      "epoch": 0.11453371013322568,
      "grad_norm": 299.4334411621094,
      "learning_rate": 1.9627837263355225e-05,
      "loss": 5.2198,
      "step": 2837
    },
    {
      "epoch": 0.11457408155026241,
      "grad_norm": 105.39899444580078,
      "learning_rate": 1.9627483777369463e-05,
      "loss": 3.4496,
      "step": 2838
    },
    {
      "epoch": 0.11461445296729915,
      "grad_norm": 201.14503479003906,
      "learning_rate": 1.962713012677621e-05,
      "loss": 4.2436,
      "step": 2839
    },
    {
      "epoch": 0.11465482438433588,
      "grad_norm": 85.05013275146484,
      "learning_rate": 1.9626776311581508e-05,
      "loss": 6.7527,
      "step": 2840
    },
    {
      "epoch": 0.11469519580137263,
      "grad_norm": 112.35802459716797,
      "learning_rate": 1.9626422331791414e-05,
      "loss": 5.3063,
      "step": 2841
    },
    {
      "epoch": 0.11473556721840937,
      "grad_norm": 89.08196258544922,
      "learning_rate": 1.9626068187411976e-05,
      "loss": 5.3895,
      "step": 2842
    },
    {
      "epoch": 0.1147759386354461,
      "grad_norm": 129.64697265625,
      "learning_rate": 1.9625713878449257e-05,
      "loss": 6.1539,
      "step": 2843
    },
    {
      "epoch": 0.11481631005248284,
      "grad_norm": 141.96945190429688,
      "learning_rate": 1.9625359404909302e-05,
      "loss": 6.0302,
      "step": 2844
    },
    {
      "epoch": 0.11485668146951958,
      "grad_norm": 116.68212127685547,
      "learning_rate": 1.962500476679818e-05,
      "loss": 5.9295,
      "step": 2845
    },
    {
      "epoch": 0.11489705288655631,
      "grad_norm": 86.8970718383789,
      "learning_rate": 1.962464996412195e-05,
      "loss": 3.9244,
      "step": 2846
    },
    {
      "epoch": 0.11493742430359305,
      "grad_norm": 766.6416015625,
      "learning_rate": 1.9624294996886684e-05,
      "loss": 4.806,
      "step": 2847
    },
    {
      "epoch": 0.1149777957206298,
      "grad_norm": 184.16143798828125,
      "learning_rate": 1.9623939865098446e-05,
      "loss": 4.5023,
      "step": 2848
    },
    {
      "epoch": 0.11501816713766654,
      "grad_norm": 168.40994262695312,
      "learning_rate": 1.962358456876331e-05,
      "loss": 5.3968,
      "step": 2849
    },
    {
      "epoch": 0.11505853855470327,
      "grad_norm": 141.6181182861328,
      "learning_rate": 1.962322910788735e-05,
      "loss": 6.7644,
      "step": 2850
    },
    {
      "epoch": 0.11509890997174001,
      "grad_norm": 105.4617691040039,
      "learning_rate": 1.962287348247665e-05,
      "loss": 6.2104,
      "step": 2851
    },
    {
      "epoch": 0.11513928138877674,
      "grad_norm": 110.53770446777344,
      "learning_rate": 1.962251769253728e-05,
      "loss": 5.3787,
      "step": 2852
    },
    {
      "epoch": 0.11517965280581348,
      "grad_norm": 129.81622314453125,
      "learning_rate": 1.9622161738075325e-05,
      "loss": 7.238,
      "step": 2853
    },
    {
      "epoch": 0.11522002422285023,
      "grad_norm": 131.30966186523438,
      "learning_rate": 1.9621805619096875e-05,
      "loss": 4.7889,
      "step": 2854
    },
    {
      "epoch": 0.11526039563988696,
      "grad_norm": 126.84219360351562,
      "learning_rate": 1.9621449335608022e-05,
      "loss": 5.7243,
      "step": 2855
    },
    {
      "epoch": 0.1153007670569237,
      "grad_norm": 166.01499938964844,
      "learning_rate": 1.9621092887614847e-05,
      "loss": 4.944,
      "step": 2856
    },
    {
      "epoch": 0.11534113847396044,
      "grad_norm": 101.75765991210938,
      "learning_rate": 1.9620736275123452e-05,
      "loss": 5.3089,
      "step": 2857
    },
    {
      "epoch": 0.11538150989099717,
      "grad_norm": 104.65613555908203,
      "learning_rate": 1.9620379498139933e-05,
      "loss": 4.5371,
      "step": 2858
    },
    {
      "epoch": 0.11542188130803391,
      "grad_norm": 81.9441909790039,
      "learning_rate": 1.9620022556670394e-05,
      "loss": 4.9534,
      "step": 2859
    },
    {
      "epoch": 0.11546225272507064,
      "grad_norm": 131.56619262695312,
      "learning_rate": 1.9619665450720928e-05,
      "loss": 4.2583,
      "step": 2860
    },
    {
      "epoch": 0.1155026241421074,
      "grad_norm": 350.3829345703125,
      "learning_rate": 1.9619308180297646e-05,
      "loss": 5.2996,
      "step": 2861
    },
    {
      "epoch": 0.11554299555914413,
      "grad_norm": 80.65410614013672,
      "learning_rate": 1.9618950745406662e-05,
      "loss": 4.6947,
      "step": 2862
    },
    {
      "epoch": 0.11558336697618087,
      "grad_norm": 92.345703125,
      "learning_rate": 1.961859314605408e-05,
      "loss": 6.2504,
      "step": 2863
    },
    {
      "epoch": 0.1156237383932176,
      "grad_norm": 99.54293823242188,
      "learning_rate": 1.9618235382246013e-05,
      "loss": 5.215,
      "step": 2864
    },
    {
      "epoch": 0.11566410981025434,
      "grad_norm": 115.34518432617188,
      "learning_rate": 1.9617877453988587e-05,
      "loss": 4.0748,
      "step": 2865
    },
    {
      "epoch": 0.11570448122729107,
      "grad_norm": 104.05806732177734,
      "learning_rate": 1.9617519361287912e-05,
      "loss": 7.8335,
      "step": 2866
    },
    {
      "epoch": 0.11574485264432782,
      "grad_norm": 96.41213989257812,
      "learning_rate": 1.9617161104150117e-05,
      "loss": 4.4058,
      "step": 2867
    },
    {
      "epoch": 0.11578522406136456,
      "grad_norm": 146.3758087158203,
      "learning_rate": 1.961680268258132e-05,
      "loss": 5.2473,
      "step": 2868
    },
    {
      "epoch": 0.1158255954784013,
      "grad_norm": 152.62484741210938,
      "learning_rate": 1.961644409658766e-05,
      "loss": 5.7922,
      "step": 2869
    },
    {
      "epoch": 0.11586596689543803,
      "grad_norm": 144.20220947265625,
      "learning_rate": 1.9616085346175256e-05,
      "loss": 6.0236,
      "step": 2870
    },
    {
      "epoch": 0.11590633831247477,
      "grad_norm": 101.10020446777344,
      "learning_rate": 1.961572643135025e-05,
      "loss": 5.088,
      "step": 2871
    },
    {
      "epoch": 0.1159467097295115,
      "grad_norm": 303.2568664550781,
      "learning_rate": 1.961536735211878e-05,
      "loss": 7.4898,
      "step": 2872
    },
    {
      "epoch": 0.11598708114654824,
      "grad_norm": 228.3203582763672,
      "learning_rate": 1.9615008108486977e-05,
      "loss": 5.2487,
      "step": 2873
    },
    {
      "epoch": 0.11602745256358499,
      "grad_norm": 99.53047943115234,
      "learning_rate": 1.961464870046099e-05,
      "loss": 5.1911,
      "step": 2874
    },
    {
      "epoch": 0.11606782398062172,
      "grad_norm": 114.30912017822266,
      "learning_rate": 1.9614289128046963e-05,
      "loss": 5.4509,
      "step": 2875
    },
    {
      "epoch": 0.11610819539765846,
      "grad_norm": 141.8883819580078,
      "learning_rate": 1.9613929391251042e-05,
      "loss": 6.4241,
      "step": 2876
    },
    {
      "epoch": 0.1161485668146952,
      "grad_norm": 119.4512710571289,
      "learning_rate": 1.9613569490079375e-05,
      "loss": 5.3816,
      "step": 2877
    },
    {
      "epoch": 0.11618893823173193,
      "grad_norm": 90.92249298095703,
      "learning_rate": 1.9613209424538126e-05,
      "loss": 6.9495,
      "step": 2878
    },
    {
      "epoch": 0.11622930964876867,
      "grad_norm": 125.96381378173828,
      "learning_rate": 1.961284919463344e-05,
      "loss": 6.0646,
      "step": 2879
    },
    {
      "epoch": 0.11626968106580542,
      "grad_norm": 89.2419662475586,
      "learning_rate": 1.961248880037148e-05,
      "loss": 6.1195,
      "step": 2880
    },
    {
      "epoch": 0.11631005248284215,
      "grad_norm": 124.45516204833984,
      "learning_rate": 1.9612128241758407e-05,
      "loss": 5.8046,
      "step": 2881
    },
    {
      "epoch": 0.11635042389987889,
      "grad_norm": 214.4410400390625,
      "learning_rate": 1.961176751880039e-05,
      "loss": 5.2036,
      "step": 2882
    },
    {
      "epoch": 0.11639079531691562,
      "grad_norm": 144.55215454101562,
      "learning_rate": 1.9611406631503594e-05,
      "loss": 6.7454,
      "step": 2883
    },
    {
      "epoch": 0.11643116673395236,
      "grad_norm": 197.62332153320312,
      "learning_rate": 1.9611045579874187e-05,
      "loss": 4.0662,
      "step": 2884
    },
    {
      "epoch": 0.1164715381509891,
      "grad_norm": 136.26785278320312,
      "learning_rate": 1.9610684363918346e-05,
      "loss": 3.9694,
      "step": 2885
    },
    {
      "epoch": 0.11651190956802583,
      "grad_norm": 133.50247192382812,
      "learning_rate": 1.9610322983642243e-05,
      "loss": 4.9842,
      "step": 2886
    },
    {
      "epoch": 0.11655228098506258,
      "grad_norm": 95.04391479492188,
      "learning_rate": 1.9609961439052058e-05,
      "loss": 7.19,
      "step": 2887
    },
    {
      "epoch": 0.11659265240209932,
      "grad_norm": 179.08856201171875,
      "learning_rate": 1.9609599730153978e-05,
      "loss": 4.9895,
      "step": 2888
    },
    {
      "epoch": 0.11663302381913605,
      "grad_norm": 191.7165985107422,
      "learning_rate": 1.9609237856954176e-05,
      "loss": 5.3786,
      "step": 2889
    },
    {
      "epoch": 0.11667339523617279,
      "grad_norm": 113.36298370361328,
      "learning_rate": 1.960887581945885e-05,
      "loss": 6.2747,
      "step": 2890
    },
    {
      "epoch": 0.11671376665320952,
      "grad_norm": 142.67823791503906,
      "learning_rate": 1.960851361767418e-05,
      "loss": 5.7507,
      "step": 2891
    },
    {
      "epoch": 0.11675413807024626,
      "grad_norm": 96.20723724365234,
      "learning_rate": 1.9608151251606372e-05,
      "loss": 6.3505,
      "step": 2892
    },
    {
      "epoch": 0.11679450948728301,
      "grad_norm": 139.81643676757812,
      "learning_rate": 1.9607788721261615e-05,
      "loss": 6.5428,
      "step": 2893
    },
    {
      "epoch": 0.11683488090431975,
      "grad_norm": 111.38191223144531,
      "learning_rate": 1.96074260266461e-05,
      "loss": 5.8122,
      "step": 2894
    },
    {
      "epoch": 0.11687525232135648,
      "grad_norm": 121.13592529296875,
      "learning_rate": 1.9607063167766042e-05,
      "loss": 5.5175,
      "step": 2895
    },
    {
      "epoch": 0.11691562373839322,
      "grad_norm": 93.2664794921875,
      "learning_rate": 1.9606700144627633e-05,
      "loss": 6.7932,
      "step": 2896
    },
    {
      "epoch": 0.11695599515542995,
      "grad_norm": 182.9295196533203,
      "learning_rate": 1.9606336957237088e-05,
      "loss": 5.8873,
      "step": 2897
    },
    {
      "epoch": 0.11699636657246669,
      "grad_norm": 83.4369888305664,
      "learning_rate": 1.960597360560061e-05,
      "loss": 4.6321,
      "step": 2898
    },
    {
      "epoch": 0.11703673798950343,
      "grad_norm": 139.7476043701172,
      "learning_rate": 1.9605610089724413e-05,
      "loss": 5.5953,
      "step": 2899
    },
    {
      "epoch": 0.11707710940654018,
      "grad_norm": 102.17660522460938,
      "learning_rate": 1.960524640961472e-05,
      "loss": 5.0627,
      "step": 2900
    },
    {
      "epoch": 0.11711748082357691,
      "grad_norm": 187.48757934570312,
      "learning_rate": 1.960488256527774e-05,
      "loss": 4.4524,
      "step": 2901
    },
    {
      "epoch": 0.11715785224061365,
      "grad_norm": 99.74190521240234,
      "learning_rate": 1.9604518556719697e-05,
      "loss": 4.405,
      "step": 2902
    },
    {
      "epoch": 0.11719822365765038,
      "grad_norm": 121.74100494384766,
      "learning_rate": 1.960415438394682e-05,
      "loss": 5.5186,
      "step": 2903
    },
    {
      "epoch": 0.11723859507468712,
      "grad_norm": 80.46865844726562,
      "learning_rate": 1.9603790046965324e-05,
      "loss": 4.5028,
      "step": 2904
    },
    {
      "epoch": 0.11727896649172385,
      "grad_norm": 173.13572692871094,
      "learning_rate": 1.9603425545781446e-05,
      "loss": 5.1696,
      "step": 2905
    },
    {
      "epoch": 0.1173193379087606,
      "grad_norm": 149.07937622070312,
      "learning_rate": 1.9603060880401416e-05,
      "loss": 4.4672,
      "step": 2906
    },
    {
      "epoch": 0.11735970932579734,
      "grad_norm": 124.09632873535156,
      "learning_rate": 1.960269605083147e-05,
      "loss": 4.9772,
      "step": 2907
    },
    {
      "epoch": 0.11740008074283408,
      "grad_norm": 78.31407928466797,
      "learning_rate": 1.960233105707785e-05,
      "loss": 4.7694,
      "step": 2908
    },
    {
      "epoch": 0.11744045215987081,
      "grad_norm": 111.23736572265625,
      "learning_rate": 1.960196589914679e-05,
      "loss": 6.2665,
      "step": 2909
    },
    {
      "epoch": 0.11748082357690755,
      "grad_norm": 107.29472351074219,
      "learning_rate": 1.960160057704453e-05,
      "loss": 5.496,
      "step": 2910
    },
    {
      "epoch": 0.11752119499394428,
      "grad_norm": 148.81829833984375,
      "learning_rate": 1.9601235090777328e-05,
      "loss": 4.8982,
      "step": 2911
    },
    {
      "epoch": 0.11756156641098102,
      "grad_norm": 147.24754333496094,
      "learning_rate": 1.9600869440351422e-05,
      "loss": 6.8848,
      "step": 2912
    },
    {
      "epoch": 0.11760193782801777,
      "grad_norm": 151.43983459472656,
      "learning_rate": 1.9600503625773072e-05,
      "loss": 5.4498,
      "step": 2913
    },
    {
      "epoch": 0.1176423092450545,
      "grad_norm": 93.7928237915039,
      "learning_rate": 1.9600137647048526e-05,
      "loss": 4.8497,
      "step": 2914
    },
    {
      "epoch": 0.11768268066209124,
      "grad_norm": 134.52220153808594,
      "learning_rate": 1.9599771504184045e-05,
      "loss": 4.9012,
      "step": 2915
    },
    {
      "epoch": 0.11772305207912798,
      "grad_norm": 171.19638061523438,
      "learning_rate": 1.9599405197185886e-05,
      "loss": 6.6919,
      "step": 2916
    },
    {
      "epoch": 0.11776342349616471,
      "grad_norm": 120.67648315429688,
      "learning_rate": 1.959903872606032e-05,
      "loss": 4.9511,
      "step": 2917
    },
    {
      "epoch": 0.11780379491320145,
      "grad_norm": 123.06472778320312,
      "learning_rate": 1.9598672090813603e-05,
      "loss": 5.9724,
      "step": 2918
    },
    {
      "epoch": 0.1178441663302382,
      "grad_norm": 135.66127014160156,
      "learning_rate": 1.9598305291452008e-05,
      "loss": 6.5138,
      "step": 2919
    },
    {
      "epoch": 0.11788453774727493,
      "grad_norm": 132.5623321533203,
      "learning_rate": 1.9597938327981808e-05,
      "loss": 6.6635,
      "step": 2920
    },
    {
      "epoch": 0.11792490916431167,
      "grad_norm": 159.81800842285156,
      "learning_rate": 1.9597571200409274e-05,
      "loss": 5.5121,
      "step": 2921
    },
    {
      "epoch": 0.1179652805813484,
      "grad_norm": 150.41600036621094,
      "learning_rate": 1.9597203908740684e-05,
      "loss": 5.0995,
      "step": 2922
    },
    {
      "epoch": 0.11800565199838514,
      "grad_norm": 99.0184097290039,
      "learning_rate": 1.959683645298232e-05,
      "loss": 6.6382,
      "step": 2923
    },
    {
      "epoch": 0.11804602341542188,
      "grad_norm": 141.140869140625,
      "learning_rate": 1.9596468833140462e-05,
      "loss": 5.1436,
      "step": 2924
    },
    {
      "epoch": 0.11808639483245861,
      "grad_norm": 94.65485382080078,
      "learning_rate": 1.9596101049221396e-05,
      "loss": 5.2972,
      "step": 2925
    },
    {
      "epoch": 0.11812676624949536,
      "grad_norm": 169.31959533691406,
      "learning_rate": 1.959573310123141e-05,
      "loss": 5.3911,
      "step": 2926
    },
    {
      "epoch": 0.1181671376665321,
      "grad_norm": 174.7076416015625,
      "learning_rate": 1.9595364989176796e-05,
      "loss": 4.6446,
      "step": 2927
    },
    {
      "epoch": 0.11820750908356883,
      "grad_norm": 136.83053588867188,
      "learning_rate": 1.959499671306385e-05,
      "loss": 5.4487,
      "step": 2928
    },
    {
      "epoch": 0.11824788050060557,
      "grad_norm": 108.09437561035156,
      "learning_rate": 1.9594628272898864e-05,
      "loss": 6.0135,
      "step": 2929
    },
    {
      "epoch": 0.1182882519176423,
      "grad_norm": 92.3945541381836,
      "learning_rate": 1.959425966868814e-05,
      "loss": 3.7237,
      "step": 2930
    },
    {
      "epoch": 0.11832862333467904,
      "grad_norm": 122.04711151123047,
      "learning_rate": 1.9593890900437976e-05,
      "loss": 6.0363,
      "step": 2931
    },
    {
      "epoch": 0.11836899475171579,
      "grad_norm": 152.0885009765625,
      "learning_rate": 1.9593521968154687e-05,
      "loss": 5.6422,
      "step": 2932
    },
    {
      "epoch": 0.11840936616875253,
      "grad_norm": 115.66700744628906,
      "learning_rate": 1.959315287184457e-05,
      "loss": 5.1413,
      "step": 2933
    },
    {
      "epoch": 0.11844973758578926,
      "grad_norm": 130.65443420410156,
      "learning_rate": 1.9592783611513942e-05,
      "loss": 5.0565,
      "step": 2934
    },
    {
      "epoch": 0.118490109002826,
      "grad_norm": 113.89570617675781,
      "learning_rate": 1.959241418716912e-05,
      "loss": 6.0548,
      "step": 2935
    },
    {
      "epoch": 0.11853048041986274,
      "grad_norm": 109.5886001586914,
      "learning_rate": 1.959204459881641e-05,
      "loss": 5.8886,
      "step": 2936
    },
    {
      "epoch": 0.11857085183689947,
      "grad_norm": 98.13663482666016,
      "learning_rate": 1.9591674846462135e-05,
      "loss": 6.1392,
      "step": 2937
    },
    {
      "epoch": 0.11861122325393621,
      "grad_norm": 139.28468322753906,
      "learning_rate": 1.9591304930112623e-05,
      "loss": 4.6899,
      "step": 2938
    },
    {
      "epoch": 0.11865159467097296,
      "grad_norm": 116.06629180908203,
      "learning_rate": 1.9590934849774192e-05,
      "loss": 4.5691,
      "step": 2939
    },
    {
      "epoch": 0.11869196608800969,
      "grad_norm": 118.36213684082031,
      "learning_rate": 1.959056460545317e-05,
      "loss": 5.2656,
      "step": 2940
    },
    {
      "epoch": 0.11873233750504643,
      "grad_norm": 130.56935119628906,
      "learning_rate": 1.9590194197155888e-05,
      "loss": 5.3649,
      "step": 2941
    },
    {
      "epoch": 0.11877270892208316,
      "grad_norm": 78.23484802246094,
      "learning_rate": 1.9589823624888683e-05,
      "loss": 5.7114,
      "step": 2942
    },
    {
      "epoch": 0.1188130803391199,
      "grad_norm": 137.72607421875,
      "learning_rate": 1.9589452888657886e-05,
      "loss": 5.7488,
      "step": 2943
    },
    {
      "epoch": 0.11885345175615664,
      "grad_norm": 76.31697082519531,
      "learning_rate": 1.958908198846984e-05,
      "loss": 6.8044,
      "step": 2944
    },
    {
      "epoch": 0.11889382317319339,
      "grad_norm": 117.3508071899414,
      "learning_rate": 1.9588710924330882e-05,
      "loss": 5.9808,
      "step": 2945
    },
    {
      "epoch": 0.11893419459023012,
      "grad_norm": 91.94345092773438,
      "learning_rate": 1.958833969624736e-05,
      "loss": 5.8062,
      "step": 2946
    },
    {
      "epoch": 0.11897456600726686,
      "grad_norm": 136.0702667236328,
      "learning_rate": 1.9587968304225617e-05,
      "loss": 6.501,
      "step": 2947
    },
    {
      "epoch": 0.1190149374243036,
      "grad_norm": 139.49472045898438,
      "learning_rate": 1.9587596748272007e-05,
      "loss": 4.8585,
      "step": 2948
    },
    {
      "epoch": 0.11905530884134033,
      "grad_norm": 118.65191650390625,
      "learning_rate": 1.9587225028392883e-05,
      "loss": 5.8989,
      "step": 2949
    },
    {
      "epoch": 0.11909568025837707,
      "grad_norm": 324.77294921875,
      "learning_rate": 1.9586853144594594e-05,
      "loss": 5.4628,
      "step": 2950
    },
    {
      "epoch": 0.1191360516754138,
      "grad_norm": 115.7116470336914,
      "learning_rate": 1.9586481096883507e-05,
      "loss": 5.3443,
      "step": 2951
    },
    {
      "epoch": 0.11917642309245055,
      "grad_norm": 89.85337829589844,
      "learning_rate": 1.958610888526598e-05,
      "loss": 6.5504,
      "step": 2952
    },
    {
      "epoch": 0.11921679450948729,
      "grad_norm": 103.36700439453125,
      "learning_rate": 1.9585736509748374e-05,
      "loss": 6.5408,
      "step": 2953
    },
    {
      "epoch": 0.11925716592652402,
      "grad_norm": 143.60659790039062,
      "learning_rate": 1.9585363970337057e-05,
      "loss": 4.0355,
      "step": 2954
    },
    {
      "epoch": 0.11929753734356076,
      "grad_norm": 119.92741394042969,
      "learning_rate": 1.9584991267038404e-05,
      "loss": 3.628,
      "step": 2955
    },
    {
      "epoch": 0.1193379087605975,
      "grad_norm": 175.2223358154297,
      "learning_rate": 1.958461839985878e-05,
      "loss": 7.1497,
      "step": 2956
    },
    {
      "epoch": 0.11937828017763423,
      "grad_norm": 92.30524444580078,
      "learning_rate": 1.9584245368804563e-05,
      "loss": 4.9126,
      "step": 2957
    },
    {
      "epoch": 0.11941865159467098,
      "grad_norm": 168.05770874023438,
      "learning_rate": 1.958387217388213e-05,
      "loss": 5.7202,
      "step": 2958
    },
    {
      "epoch": 0.11945902301170772,
      "grad_norm": 109.39268493652344,
      "learning_rate": 1.9583498815097865e-05,
      "loss": 6.3297,
      "step": 2959
    },
    {
      "epoch": 0.11949939442874445,
      "grad_norm": 120.05841064453125,
      "learning_rate": 1.9583125292458146e-05,
      "loss": 5.604,
      "step": 2960
    },
    {
      "epoch": 0.11953976584578119,
      "grad_norm": 109.77481079101562,
      "learning_rate": 1.9582751605969367e-05,
      "loss": 6.2526,
      "step": 2961
    },
    {
      "epoch": 0.11958013726281792,
      "grad_norm": 80.87390899658203,
      "learning_rate": 1.958237775563791e-05,
      "loss": 7.3204,
      "step": 2962
    },
    {
      "epoch": 0.11962050867985466,
      "grad_norm": 128.0760040283203,
      "learning_rate": 1.9582003741470173e-05,
      "loss": 4.9704,
      "step": 2963
    },
    {
      "epoch": 0.1196608800968914,
      "grad_norm": 85.79962921142578,
      "learning_rate": 1.9581629563472546e-05,
      "loss": 5.1826,
      "step": 2964
    },
    {
      "epoch": 0.11970125151392814,
      "grad_norm": 145.7322235107422,
      "learning_rate": 1.9581255221651428e-05,
      "loss": 4.3559,
      "step": 2965
    },
    {
      "epoch": 0.11974162293096488,
      "grad_norm": 162.54034423828125,
      "learning_rate": 1.9580880716013216e-05,
      "loss": 5.3754,
      "step": 2966
    },
    {
      "epoch": 0.11978199434800162,
      "grad_norm": 143.62171936035156,
      "learning_rate": 1.958050604656432e-05,
      "loss": 6.867,
      "step": 2967
    },
    {
      "epoch": 0.11982236576503835,
      "grad_norm": 126.75370788574219,
      "learning_rate": 1.9580131213311144e-05,
      "loss": 6.551,
      "step": 2968
    },
    {
      "epoch": 0.11986273718207509,
      "grad_norm": 201.0367431640625,
      "learning_rate": 1.9579756216260093e-05,
      "loss": 6.5312,
      "step": 2969
    },
    {
      "epoch": 0.11990310859911182,
      "grad_norm": 84.67472839355469,
      "learning_rate": 1.9579381055417584e-05,
      "loss": 6.3217,
      "step": 2970
    },
    {
      "epoch": 0.11994348001614857,
      "grad_norm": 124.68359375,
      "learning_rate": 1.9579005730790025e-05,
      "loss": 6.0689,
      "step": 2971
    },
    {
      "epoch": 0.11998385143318531,
      "grad_norm": 78.9422607421875,
      "learning_rate": 1.957863024238384e-05,
      "loss": 5.2486,
      "step": 2972
    },
    {
      "epoch": 0.12002422285022205,
      "grad_norm": 89.27397155761719,
      "learning_rate": 1.9578254590205442e-05,
      "loss": 6.2549,
      "step": 2973
    },
    {
      "epoch": 0.12006459426725878,
      "grad_norm": 82.27686309814453,
      "learning_rate": 1.9577878774261258e-05,
      "loss": 6.5253,
      "step": 2974
    },
    {
      "epoch": 0.12010496568429552,
      "grad_norm": 116.25216674804688,
      "learning_rate": 1.9577502794557714e-05,
      "loss": 7.0833,
      "step": 2975
    },
    {
      "epoch": 0.12014533710133225,
      "grad_norm": 151.03070068359375,
      "learning_rate": 1.957712665110124e-05,
      "loss": 5.6706,
      "step": 2976
    },
    {
      "epoch": 0.12018570851836899,
      "grad_norm": 65.24050903320312,
      "learning_rate": 1.957675034389826e-05,
      "loss": 5.7581,
      "step": 2977
    },
    {
      "epoch": 0.12022607993540574,
      "grad_norm": 115.97901153564453,
      "learning_rate": 1.9576373872955212e-05,
      "loss": 5.136,
      "step": 2978
    },
    {
      "epoch": 0.12026645135244247,
      "grad_norm": 90.95303344726562,
      "learning_rate": 1.9575997238278536e-05,
      "loss": 3.8348,
      "step": 2979
    },
    {
      "epoch": 0.12030682276947921,
      "grad_norm": 127.73001098632812,
      "learning_rate": 1.957562043987467e-05,
      "loss": 4.2081,
      "step": 2980
    },
    {
      "epoch": 0.12034719418651595,
      "grad_norm": 144.21141052246094,
      "learning_rate": 1.957524347775005e-05,
      "loss": 6.8138,
      "step": 2981
    },
    {
      "epoch": 0.12038756560355268,
      "grad_norm": 100.97782135009766,
      "learning_rate": 1.957486635191113e-05,
      "loss": 5.3042,
      "step": 2982
    },
    {
      "epoch": 0.12042793702058942,
      "grad_norm": 276.9176940917969,
      "learning_rate": 1.9574489062364356e-05,
      "loss": 5.2111,
      "step": 2983
    },
    {
      "epoch": 0.12046830843762617,
      "grad_norm": 87.55242156982422,
      "learning_rate": 1.9574111609116174e-05,
      "loss": 6.1416,
      "step": 2984
    },
    {
      "epoch": 0.1205086798546629,
      "grad_norm": 96.09854125976562,
      "learning_rate": 1.9573733992173045e-05,
      "loss": 6.1779,
      "step": 2985
    },
    {
      "epoch": 0.12054905127169964,
      "grad_norm": 145.1275634765625,
      "learning_rate": 1.9573356211541414e-05,
      "loss": 5.929,
      "step": 2986
    },
    {
      "epoch": 0.12058942268873638,
      "grad_norm": 96.9312515258789,
      "learning_rate": 1.957297826722775e-05,
      "loss": 5.788,
      "step": 2987
    },
    {
      "epoch": 0.12062979410577311,
      "grad_norm": 102.59430694580078,
      "learning_rate": 1.9572600159238514e-05,
      "loss": 4.8528,
      "step": 2988
    },
    {
      "epoch": 0.12067016552280985,
      "grad_norm": 90.50984191894531,
      "learning_rate": 1.9572221887580167e-05,
      "loss": 6.8133,
      "step": 2989
    },
    {
      "epoch": 0.12071053693984658,
      "grad_norm": 129.3245849609375,
      "learning_rate": 1.957184345225918e-05,
      "loss": 5.221,
      "step": 2990
    },
    {
      "epoch": 0.12075090835688333,
      "grad_norm": 108.10220336914062,
      "learning_rate": 1.957146485328202e-05,
      "loss": 5.9005,
      "step": 2991
    },
    {
      "epoch": 0.12079127977392007,
      "grad_norm": 79.42806243896484,
      "learning_rate": 1.9571086090655166e-05,
      "loss": 6.4927,
      "step": 2992
    },
    {
      "epoch": 0.1208316511909568,
      "grad_norm": 128.95445251464844,
      "learning_rate": 1.9570707164385085e-05,
      "loss": 5.6452,
      "step": 2993
    },
    {
      "epoch": 0.12087202260799354,
      "grad_norm": 87.2731704711914,
      "learning_rate": 1.9570328074478263e-05,
      "loss": 5.2886,
      "step": 2994
    },
    {
      "epoch": 0.12091239402503028,
      "grad_norm": 138.598876953125,
      "learning_rate": 1.9569948820941177e-05,
      "loss": 5.9654,
      "step": 2995
    },
    {
      "epoch": 0.12095276544206701,
      "grad_norm": 95.38597869873047,
      "learning_rate": 1.956956940378032e-05,
      "loss": 4.7492,
      "step": 2996
    },
    {
      "epoch": 0.12099313685910375,
      "grad_norm": 106.25130462646484,
      "learning_rate": 1.9569189823002167e-05,
      "loss": 6.6264,
      "step": 2997
    },
    {
      "epoch": 0.1210335082761405,
      "grad_norm": 110.90544891357422,
      "learning_rate": 1.9568810078613215e-05,
      "loss": 6.403,
      "step": 2998
    },
    {
      "epoch": 0.12107387969317723,
      "grad_norm": 80.05277252197266,
      "learning_rate": 1.9568430170619953e-05,
      "loss": 7.0713,
      "step": 2999
    },
    {
      "epoch": 0.12111425111021397,
      "grad_norm": 130.12994384765625,
      "learning_rate": 1.9568050099028885e-05,
      "loss": 6.6492,
      "step": 3000
    },
    {
      "epoch": 0.1211546225272507,
      "grad_norm": 114.90186309814453,
      "learning_rate": 1.9567669863846498e-05,
      "loss": 2.923,
      "step": 3001
    },
    {
      "epoch": 0.12119499394428744,
      "grad_norm": 102.87138366699219,
      "learning_rate": 1.95672894650793e-05,
      "loss": 5.6483,
      "step": 3002
    },
    {
      "epoch": 0.12123536536132418,
      "grad_norm": 136.70274353027344,
      "learning_rate": 1.9566908902733792e-05,
      "loss": 5.2523,
      "step": 3003
    },
    {
      "epoch": 0.12127573677836093,
      "grad_norm": 107.27422332763672,
      "learning_rate": 1.9566528176816484e-05,
      "loss": 5.9297,
      "step": 3004
    },
    {
      "epoch": 0.12131610819539766,
      "grad_norm": 158.54165649414062,
      "learning_rate": 1.9566147287333885e-05,
      "loss": 5.5433,
      "step": 3005
    },
    {
      "epoch": 0.1213564796124344,
      "grad_norm": 185.38526916503906,
      "learning_rate": 1.9565766234292503e-05,
      "loss": 4.9052,
      "step": 3006
    },
    {
      "epoch": 0.12139685102947113,
      "grad_norm": 101.69034576416016,
      "learning_rate": 1.9565385017698858e-05,
      "loss": 5.496,
      "step": 3007
    },
    {
      "epoch": 0.12143722244650787,
      "grad_norm": 99.71556091308594,
      "learning_rate": 1.9565003637559465e-05,
      "loss": 5.7557,
      "step": 3008
    },
    {
      "epoch": 0.1214775938635446,
      "grad_norm": 158.87655639648438,
      "learning_rate": 1.9564622093880845e-05,
      "loss": 6.4465,
      "step": 3009
    },
    {
      "epoch": 0.12151796528058134,
      "grad_norm": 214.7599334716797,
      "learning_rate": 1.9564240386669525e-05,
      "loss": 7.1374,
      "step": 3010
    },
    {
      "epoch": 0.12155833669761809,
      "grad_norm": 88.59691619873047,
      "learning_rate": 1.9563858515932026e-05,
      "loss": 5.6715,
      "step": 3011
    },
    {
      "epoch": 0.12159870811465483,
      "grad_norm": 197.96749877929688,
      "learning_rate": 1.9563476481674878e-05,
      "loss": 6.1516,
      "step": 3012
    },
    {
      "epoch": 0.12163907953169156,
      "grad_norm": 141.39268493652344,
      "learning_rate": 1.956309428390462e-05,
      "loss": 4.1712,
      "step": 3013
    },
    {
      "epoch": 0.1216794509487283,
      "grad_norm": 111.05030059814453,
      "learning_rate": 1.9562711922627777e-05,
      "loss": 5.4761,
      "step": 3014
    },
    {
      "epoch": 0.12171982236576503,
      "grad_norm": 97.3577651977539,
      "learning_rate": 1.956232939785089e-05,
      "loss": 6.8647,
      "step": 3015
    },
    {
      "epoch": 0.12176019378280177,
      "grad_norm": 106.87751007080078,
      "learning_rate": 1.9561946709580502e-05,
      "loss": 5.6824,
      "step": 3016
    },
    {
      "epoch": 0.12180056519983852,
      "grad_norm": 111.0870361328125,
      "learning_rate": 1.9561563857823156e-05,
      "loss": 6.5317,
      "step": 3017
    },
    {
      "epoch": 0.12184093661687526,
      "grad_norm": 79.923583984375,
      "learning_rate": 1.956118084258539e-05,
      "loss": 5.1934,
      "step": 3018
    },
    {
      "epoch": 0.12188130803391199,
      "grad_norm": 275.3927307128906,
      "learning_rate": 1.9560797663873764e-05,
      "loss": 3.9576,
      "step": 3019
    },
    {
      "epoch": 0.12192167945094873,
      "grad_norm": 133.6530303955078,
      "learning_rate": 1.956041432169482e-05,
      "loss": 4.8742,
      "step": 3020
    },
    {
      "epoch": 0.12196205086798546,
      "grad_norm": 97.6750259399414,
      "learning_rate": 1.956003081605512e-05,
      "loss": 4.5413,
      "step": 3021
    },
    {
      "epoch": 0.1220024222850222,
      "grad_norm": 87.14346313476562,
      "learning_rate": 1.9559647146961216e-05,
      "loss": 5.4227,
      "step": 3022
    },
    {
      "epoch": 0.12204279370205894,
      "grad_norm": 237.3148956298828,
      "learning_rate": 1.9559263314419666e-05,
      "loss": 6.2091,
      "step": 3023
    },
    {
      "epoch": 0.12208316511909569,
      "grad_norm": 85.06988525390625,
      "learning_rate": 1.955887931843704e-05,
      "loss": 5.7319,
      "step": 3024
    },
    {
      "epoch": 0.12212353653613242,
      "grad_norm": 114.53141021728516,
      "learning_rate": 1.9558495159019898e-05,
      "loss": 6.0929,
      "step": 3025
    },
    {
      "epoch": 0.12216390795316916,
      "grad_norm": 144.83270263671875,
      "learning_rate": 1.9558110836174807e-05,
      "loss": 5.2912,
      "step": 3026
    },
    {
      "epoch": 0.12220427937020589,
      "grad_norm": 168.52305603027344,
      "learning_rate": 1.9557726349908347e-05,
      "loss": 4.3041,
      "step": 3027
    },
    {
      "epoch": 0.12224465078724263,
      "grad_norm": 133.90745544433594,
      "learning_rate": 1.955734170022708e-05,
      "loss": 6.0727,
      "step": 3028
    },
    {
      "epoch": 0.12228502220427936,
      "grad_norm": 134.48501586914062,
      "learning_rate": 1.9556956887137588e-05,
      "loss": 6.4291,
      "step": 3029
    },
    {
      "epoch": 0.12232539362131611,
      "grad_norm": 106.09542083740234,
      "learning_rate": 1.9556571910646452e-05,
      "loss": 6.3703,
      "step": 3030
    },
    {
      "epoch": 0.12236576503835285,
      "grad_norm": 181.73464965820312,
      "learning_rate": 1.9556186770760252e-05,
      "loss": 4.5149,
      "step": 3031
    },
    {
      "epoch": 0.12240613645538959,
      "grad_norm": 86.91201782226562,
      "learning_rate": 1.9555801467485574e-05,
      "loss": 5.1394,
      "step": 3032
    },
    {
      "epoch": 0.12244650787242632,
      "grad_norm": 124.71038818359375,
      "learning_rate": 1.9555416000829004e-05,
      "loss": 6.3472,
      "step": 3033
    },
    {
      "epoch": 0.12248687928946306,
      "grad_norm": 98.9211196899414,
      "learning_rate": 1.9555030370797133e-05,
      "loss": 5.8807,
      "step": 3034
    },
    {
      "epoch": 0.1225272507064998,
      "grad_norm": 117.03617858886719,
      "learning_rate": 1.9554644577396557e-05,
      "loss": 5.0495,
      "step": 3035
    },
    {
      "epoch": 0.12256762212353653,
      "grad_norm": 134.26441955566406,
      "learning_rate": 1.955425862063387e-05,
      "loss": 6.0457,
      "step": 3036
    },
    {
      "epoch": 0.12260799354057328,
      "grad_norm": 97.81014251708984,
      "learning_rate": 1.955387250051567e-05,
      "loss": 5.798,
      "step": 3037
    },
    {
      "epoch": 0.12264836495761001,
      "grad_norm": 94.00525665283203,
      "learning_rate": 1.9553486217048564e-05,
      "loss": 6.821,
      "step": 3038
    },
    {
      "epoch": 0.12268873637464675,
      "grad_norm": 92.54986572265625,
      "learning_rate": 1.955309977023915e-05,
      "loss": 6.0429,
      "step": 3039
    },
    {
      "epoch": 0.12272910779168349,
      "grad_norm": 119.48859405517578,
      "learning_rate": 1.955271316009404e-05,
      "loss": 5.0727,
      "step": 3040
    },
    {
      "epoch": 0.12276947920872022,
      "grad_norm": 103.13643646240234,
      "learning_rate": 1.955232638661984e-05,
      "loss": 5.4065,
      "step": 3041
    },
    {
      "epoch": 0.12280985062575696,
      "grad_norm": 194.2552947998047,
      "learning_rate": 1.9551939449823164e-05,
      "loss": 4.7399,
      "step": 3042
    },
    {
      "epoch": 0.12285022204279371,
      "grad_norm": 121.2883071899414,
      "learning_rate": 1.9551552349710634e-05,
      "loss": 6.3073,
      "step": 3043
    },
    {
      "epoch": 0.12289059345983044,
      "grad_norm": 170.08029174804688,
      "learning_rate": 1.9551165086288857e-05,
      "loss": 5.938,
      "step": 3044
    },
    {
      "epoch": 0.12293096487686718,
      "grad_norm": 106.63807678222656,
      "learning_rate": 1.9550777659564467e-05,
      "loss": 6.4954,
      "step": 3045
    },
    {
      "epoch": 0.12297133629390392,
      "grad_norm": 101.16615295410156,
      "learning_rate": 1.9550390069544076e-05,
      "loss": 4.1097,
      "step": 3046
    },
    {
      "epoch": 0.12301170771094065,
      "grad_norm": 90.30718994140625,
      "learning_rate": 1.955000231623432e-05,
      "loss": 6.1523,
      "step": 3047
    },
    {
      "epoch": 0.12305207912797739,
      "grad_norm": 108.41767120361328,
      "learning_rate": 1.954961439964182e-05,
      "loss": 6.6655,
      "step": 3048
    },
    {
      "epoch": 0.12309245054501412,
      "grad_norm": 70.23261260986328,
      "learning_rate": 1.954922631977322e-05,
      "loss": 6.6084,
      "step": 3049
    },
    {
      "epoch": 0.12313282196205087,
      "grad_norm": 99.912353515625,
      "learning_rate": 1.9548838076635146e-05,
      "loss": 5.3154,
      "step": 3050
    },
    {
      "epoch": 0.12317319337908761,
      "grad_norm": 125.38279724121094,
      "learning_rate": 1.954844967023424e-05,
      "loss": 6.5651,
      "step": 3051
    },
    {
      "epoch": 0.12321356479612434,
      "grad_norm": 80.5986328125,
      "learning_rate": 1.9548061100577143e-05,
      "loss": 5.4668,
      "step": 3052
    },
    {
      "epoch": 0.12325393621316108,
      "grad_norm": 61.0533447265625,
      "learning_rate": 1.9547672367670496e-05,
      "loss": 4.6164,
      "step": 3053
    },
    {
      "epoch": 0.12329430763019782,
      "grad_norm": 73.76923370361328,
      "learning_rate": 1.9547283471520946e-05,
      "loss": 6.635,
      "step": 3054
    },
    {
      "epoch": 0.12333467904723455,
      "grad_norm": 77.86808013916016,
      "learning_rate": 1.9546894412135144e-05,
      "loss": 3.819,
      "step": 3055
    },
    {
      "epoch": 0.1233750504642713,
      "grad_norm": 84.33065032958984,
      "learning_rate": 1.954650518951974e-05,
      "loss": 5.5865,
      "step": 3056
    },
    {
      "epoch": 0.12341542188130804,
      "grad_norm": 113.69490814208984,
      "learning_rate": 1.9546115803681395e-05,
      "loss": 5.3267,
      "step": 3057
    },
    {
      "epoch": 0.12345579329834477,
      "grad_norm": 85.44307708740234,
      "learning_rate": 1.9545726254626753e-05,
      "loss": 6.4398,
      "step": 3058
    },
    {
      "epoch": 0.12349616471538151,
      "grad_norm": 95.12298583984375,
      "learning_rate": 1.954533654236249e-05,
      "loss": 5.0643,
      "step": 3059
    },
    {
      "epoch": 0.12353653613241825,
      "grad_norm": 101.4286880493164,
      "learning_rate": 1.954494666689526e-05,
      "loss": 5.5623,
      "step": 3060
    },
    {
      "epoch": 0.12357690754945498,
      "grad_norm": 101.72466278076172,
      "learning_rate": 1.954455662823173e-05,
      "loss": 4.761,
      "step": 3061
    },
    {
      "epoch": 0.12361727896649172,
      "grad_norm": 149.30429077148438,
      "learning_rate": 1.954416642637857e-05,
      "loss": 4.6264,
      "step": 3062
    },
    {
      "epoch": 0.12365765038352847,
      "grad_norm": 100.23894500732422,
      "learning_rate": 1.9543776061342453e-05,
      "loss": 5.6284,
      "step": 3063
    },
    {
      "epoch": 0.1236980218005652,
      "grad_norm": 86.8613510131836,
      "learning_rate": 1.954338553313005e-05,
      "loss": 7.0004,
      "step": 3064
    },
    {
      "epoch": 0.12373839321760194,
      "grad_norm": 99.33417510986328,
      "learning_rate": 1.954299484174804e-05,
      "loss": 5.5004,
      "step": 3065
    },
    {
      "epoch": 0.12377876463463867,
      "grad_norm": 87.9779281616211,
      "learning_rate": 1.9542603987203103e-05,
      "loss": 6.307,
      "step": 3066
    },
    {
      "epoch": 0.12381913605167541,
      "grad_norm": 116.65901184082031,
      "learning_rate": 1.954221296950192e-05,
      "loss": 6.4989,
      "step": 3067
    },
    {
      "epoch": 0.12385950746871215,
      "grad_norm": 153.35182189941406,
      "learning_rate": 1.954182178865118e-05,
      "loss": 5.6255,
      "step": 3068
    },
    {
      "epoch": 0.1238998788857489,
      "grad_norm": 107.4705581665039,
      "learning_rate": 1.954143044465757e-05,
      "loss": 6.6872,
      "step": 3069
    },
    {
      "epoch": 0.12394025030278563,
      "grad_norm": 140.2205810546875,
      "learning_rate": 1.9541038937527778e-05,
      "loss": 6.5575,
      "step": 3070
    },
    {
      "epoch": 0.12398062171982237,
      "grad_norm": 234.05703735351562,
      "learning_rate": 1.9540647267268503e-05,
      "loss": 5.8685,
      "step": 3071
    },
    {
      "epoch": 0.1240209931368591,
      "grad_norm": 121.8825912475586,
      "learning_rate": 1.9540255433886437e-05,
      "loss": 4.7414,
      "step": 3072
    },
    {
      "epoch": 0.12406136455389584,
      "grad_norm": 154.19920349121094,
      "learning_rate": 1.953986343738828e-05,
      "loss": 5.9981,
      "step": 3073
    },
    {
      "epoch": 0.12410173597093258,
      "grad_norm": 269.42071533203125,
      "learning_rate": 1.9539471277780734e-05,
      "loss": 5.8749,
      "step": 3074
    },
    {
      "epoch": 0.12414210738796931,
      "grad_norm": 78.53853607177734,
      "learning_rate": 1.953907895507051e-05,
      "loss": 4.967,
      "step": 3075
    },
    {
      "epoch": 0.12418247880500606,
      "grad_norm": 77.57672882080078,
      "learning_rate": 1.9538686469264308e-05,
      "loss": 5.215,
      "step": 3076
    },
    {
      "epoch": 0.1242228502220428,
      "grad_norm": 131.1054229736328,
      "learning_rate": 1.9538293820368845e-05,
      "loss": 5.2825,
      "step": 3077
    },
    {
      "epoch": 0.12426322163907953,
      "grad_norm": 206.3662872314453,
      "learning_rate": 1.9537901008390828e-05,
      "loss": 5.4532,
      "step": 3078
    },
    {
      "epoch": 0.12430359305611627,
      "grad_norm": 101.35773468017578,
      "learning_rate": 1.9537508033336976e-05,
      "loss": 5.1707,
      "step": 3079
    },
    {
      "epoch": 0.124343964473153,
      "grad_norm": 190.18592834472656,
      "learning_rate": 1.953711489521401e-05,
      "loss": 5.196,
      "step": 3080
    },
    {
      "epoch": 0.12438433589018974,
      "grad_norm": 144.19247436523438,
      "learning_rate": 1.953672159402865e-05,
      "loss": 4.3961,
      "step": 3081
    },
    {
      "epoch": 0.12442470730722649,
      "grad_norm": 130.47445678710938,
      "learning_rate": 1.9536328129787617e-05,
      "loss": 6.3564,
      "step": 3082
    },
    {
      "epoch": 0.12446507872426323,
      "grad_norm": 176.82093811035156,
      "learning_rate": 1.9535934502497648e-05,
      "loss": 5.2432,
      "step": 3083
    },
    {
      "epoch": 0.12450545014129996,
      "grad_norm": 135.08218383789062,
      "learning_rate": 1.953554071216546e-05,
      "loss": 4.7176,
      "step": 3084
    },
    {
      "epoch": 0.1245458215583367,
      "grad_norm": 67.4820785522461,
      "learning_rate": 1.95351467587978e-05,
      "loss": 5.4931,
      "step": 3085
    },
    {
      "epoch": 0.12458619297537343,
      "grad_norm": 167.80418395996094,
      "learning_rate": 1.9534752642401395e-05,
      "loss": 5.1622,
      "step": 3086
    },
    {
      "epoch": 0.12462656439241017,
      "grad_norm": 133.0956268310547,
      "learning_rate": 1.9534358362982982e-05,
      "loss": 5.0061,
      "step": 3087
    },
    {
      "epoch": 0.1246669358094469,
      "grad_norm": 123.34699249267578,
      "learning_rate": 1.9533963920549307e-05,
      "loss": 4.1938,
      "step": 3088
    },
    {
      "epoch": 0.12470730722648365,
      "grad_norm": 191.40663146972656,
      "learning_rate": 1.9533569315107113e-05,
      "loss": 5.1814,
      "step": 3089
    },
    {
      "epoch": 0.12474767864352039,
      "grad_norm": 104.7121810913086,
      "learning_rate": 1.9533174546663145e-05,
      "loss": 5.1499,
      "step": 3090
    },
    {
      "epoch": 0.12478805006055713,
      "grad_norm": 117.82018280029297,
      "learning_rate": 1.9532779615224156e-05,
      "loss": 5.6493,
      "step": 3091
    },
    {
      "epoch": 0.12482842147759386,
      "grad_norm": 109.4085464477539,
      "learning_rate": 1.9532384520796892e-05,
      "loss": 6.3299,
      "step": 3092
    },
    {
      "epoch": 0.1248687928946306,
      "grad_norm": 156.14854431152344,
      "learning_rate": 1.9531989263388116e-05,
      "loss": 5.2113,
      "step": 3093
    },
    {
      "epoch": 0.12490916431166733,
      "grad_norm": 156.88377380371094,
      "learning_rate": 1.9531593843004583e-05,
      "loss": 6.3261,
      "step": 3094
    },
    {
      "epoch": 0.12494953572870408,
      "grad_norm": 123.63557434082031,
      "learning_rate": 1.953119825965305e-05,
      "loss": 6.6006,
      "step": 3095
    },
    {
      "epoch": 0.12498990714574082,
      "grad_norm": 100.6690444946289,
      "learning_rate": 1.9530802513340284e-05,
      "loss": 4.2996,
      "step": 3096
    },
    {
      "epoch": 0.12503027856277754,
      "grad_norm": 172.97926330566406,
      "learning_rate": 1.9530406604073054e-05,
      "loss": 6.198,
      "step": 3097
    },
    {
      "epoch": 0.1250706499798143,
      "grad_norm": 161.01112365722656,
      "learning_rate": 1.9530010531858122e-05,
      "loss": 5.8132,
      "step": 3098
    },
    {
      "epoch": 0.12511102139685104,
      "grad_norm": 103.4471664428711,
      "learning_rate": 1.9529614296702267e-05,
      "loss": 5.535,
      "step": 3099
    },
    {
      "epoch": 0.12515139281388776,
      "grad_norm": 133.88436889648438,
      "learning_rate": 1.9529217898612262e-05,
      "loss": 4.138,
      "step": 3100
    },
    {
      "epoch": 0.1251917642309245,
      "grad_norm": 141.58786010742188,
      "learning_rate": 1.952882133759488e-05,
      "loss": 6.1002,
      "step": 3101
    },
    {
      "epoch": 0.12523213564796123,
      "grad_norm": 141.3439483642578,
      "learning_rate": 1.9528424613656908e-05,
      "loss": 5.5198,
      "step": 3102
    },
    {
      "epoch": 0.12527250706499798,
      "grad_norm": 181.4371795654297,
      "learning_rate": 1.9528027726805122e-05,
      "loss": 4.6176,
      "step": 3103
    },
    {
      "epoch": 0.1253128784820347,
      "grad_norm": 127.01652526855469,
      "learning_rate": 1.9527630677046314e-05,
      "loss": 6.4827,
      "step": 3104
    },
    {
      "epoch": 0.12535324989907146,
      "grad_norm": 114.85302734375,
      "learning_rate": 1.952723346438727e-05,
      "loss": 6.4399,
      "step": 3105
    },
    {
      "epoch": 0.1253936213161082,
      "grad_norm": 96.41173553466797,
      "learning_rate": 1.952683608883478e-05,
      "loss": 5.2916,
      "step": 3106
    },
    {
      "epoch": 0.12543399273314493,
      "grad_norm": 94.17321014404297,
      "learning_rate": 1.952643855039564e-05,
      "loss": 5.5977,
      "step": 3107
    },
    {
      "epoch": 0.12547436415018168,
      "grad_norm": 115.5965576171875,
      "learning_rate": 1.9526040849076646e-05,
      "loss": 5.9967,
      "step": 3108
    },
    {
      "epoch": 0.1255147355672184,
      "grad_norm": 145.9651336669922,
      "learning_rate": 1.95256429848846e-05,
      "loss": 6.3116,
      "step": 3109
    },
    {
      "epoch": 0.12555510698425515,
      "grad_norm": 177.015380859375,
      "learning_rate": 1.95252449578263e-05,
      "loss": 5.1274,
      "step": 3110
    },
    {
      "epoch": 0.1255954784012919,
      "grad_norm": 210.09246826171875,
      "learning_rate": 1.952484676790856e-05,
      "loss": 6.6851,
      "step": 3111
    },
    {
      "epoch": 0.12563584981832862,
      "grad_norm": 155.3188934326172,
      "learning_rate": 1.9524448415138176e-05,
      "loss": 4.9588,
      "step": 3112
    },
    {
      "epoch": 0.12567622123536537,
      "grad_norm": 78.56980895996094,
      "learning_rate": 1.952404989952197e-05,
      "loss": 5.8937,
      "step": 3113
    },
    {
      "epoch": 0.1257165926524021,
      "grad_norm": 94.26100158691406,
      "learning_rate": 1.952365122106675e-05,
      "loss": 6.6718,
      "step": 3114
    },
    {
      "epoch": 0.12575696406943884,
      "grad_norm": 174.9249267578125,
      "learning_rate": 1.9523252379779333e-05,
      "loss": 6.867,
      "step": 3115
    },
    {
      "epoch": 0.12579733548647556,
      "grad_norm": 90.84963989257812,
      "learning_rate": 1.9522853375666536e-05,
      "loss": 5.2177,
      "step": 3116
    },
    {
      "epoch": 0.12583770690351231,
      "grad_norm": 183.90719604492188,
      "learning_rate": 1.9522454208735185e-05,
      "loss": 6.1207,
      "step": 3117
    },
    {
      "epoch": 0.12587807832054906,
      "grad_norm": 107.42868041992188,
      "learning_rate": 1.952205487899211e-05,
      "loss": 5.577,
      "step": 3118
    },
    {
      "epoch": 0.12591844973758579,
      "grad_norm": 91.70764923095703,
      "learning_rate": 1.9521655386444122e-05,
      "loss": 5.6016,
      "step": 3119
    },
    {
      "epoch": 0.12595882115462254,
      "grad_norm": 123.33866882324219,
      "learning_rate": 1.9521255731098066e-05,
      "loss": 5.6921,
      "step": 3120
    },
    {
      "epoch": 0.12599919257165926,
      "grad_norm": 134.3070831298828,
      "learning_rate": 1.952085591296077e-05,
      "loss": 5.6766,
      "step": 3121
    },
    {
      "epoch": 0.126039563988696,
      "grad_norm": 71.31695556640625,
      "learning_rate": 1.9520455932039074e-05,
      "loss": 6.4608,
      "step": 3122
    },
    {
      "epoch": 0.12607993540573273,
      "grad_norm": 93.03102111816406,
      "learning_rate": 1.952005578833981e-05,
      "loss": 5.3501,
      "step": 3123
    },
    {
      "epoch": 0.12612030682276948,
      "grad_norm": 81.07150268554688,
      "learning_rate": 1.9519655481869826e-05,
      "loss": 6.9141,
      "step": 3124
    },
    {
      "epoch": 0.12616067823980623,
      "grad_norm": 104.09778594970703,
      "learning_rate": 1.9519255012635958e-05,
      "loss": 4.8043,
      "step": 3125
    },
    {
      "epoch": 0.12620104965684295,
      "grad_norm": 172.8452606201172,
      "learning_rate": 1.951885438064506e-05,
      "loss": 5.5278,
      "step": 3126
    },
    {
      "epoch": 0.1262414210738797,
      "grad_norm": 117.44365692138672,
      "learning_rate": 1.9518453585903983e-05,
      "loss": 5.9439,
      "step": 3127
    },
    {
      "epoch": 0.12628179249091642,
      "grad_norm": 1460.0037841796875,
      "learning_rate": 1.951805262841958e-05,
      "loss": 4.5348,
      "step": 3128
    },
    {
      "epoch": 0.12632216390795317,
      "grad_norm": 100.92747497558594,
      "learning_rate": 1.9517651508198695e-05,
      "loss": 4.8644,
      "step": 3129
    },
    {
      "epoch": 0.1263625353249899,
      "grad_norm": 100.85173034667969,
      "learning_rate": 1.9517250225248202e-05,
      "loss": 6.8637,
      "step": 3130
    },
    {
      "epoch": 0.12640290674202664,
      "grad_norm": 140.47178649902344,
      "learning_rate": 1.951684877957495e-05,
      "loss": 3.9466,
      "step": 3131
    },
    {
      "epoch": 0.1264432781590634,
      "grad_norm": 89.65636444091797,
      "learning_rate": 1.9516447171185806e-05,
      "loss": 4.6231,
      "step": 3132
    },
    {
      "epoch": 0.12648364957610012,
      "grad_norm": 71.27005767822266,
      "learning_rate": 1.951604540008764e-05,
      "loss": 5.5932,
      "step": 3133
    },
    {
      "epoch": 0.12652402099313687,
      "grad_norm": 76.7610855102539,
      "learning_rate": 1.9515643466287322e-05,
      "loss": 6.3197,
      "step": 3134
    },
    {
      "epoch": 0.1265643924101736,
      "grad_norm": 108.06694030761719,
      "learning_rate": 1.9515241369791716e-05,
      "loss": 5.5892,
      "step": 3135
    },
    {
      "epoch": 0.12660476382721034,
      "grad_norm": 93.9175033569336,
      "learning_rate": 1.9514839110607705e-05,
      "loss": 7.292,
      "step": 3136
    },
    {
      "epoch": 0.1266451352442471,
      "grad_norm": 167.29388427734375,
      "learning_rate": 1.9514436688742163e-05,
      "loss": 5.4267,
      "step": 3137
    },
    {
      "epoch": 0.1266855066612838,
      "grad_norm": 81.61820983886719,
      "learning_rate": 1.9514034104201976e-05,
      "loss": 3.9218,
      "step": 3138
    },
    {
      "epoch": 0.12672587807832056,
      "grad_norm": 90.28086853027344,
      "learning_rate": 1.951363135699402e-05,
      "loss": 5.7686,
      "step": 3139
    },
    {
      "epoch": 0.12676624949535728,
      "grad_norm": 962.6541137695312,
      "learning_rate": 1.951322844712518e-05,
      "loss": 5.3711,
      "step": 3140
    },
    {
      "epoch": 0.12680662091239403,
      "grad_norm": 152.41455078125,
      "learning_rate": 1.9512825374602353e-05,
      "loss": 5.1325,
      "step": 3141
    },
    {
      "epoch": 0.12684699232943075,
      "grad_norm": 131.80691528320312,
      "learning_rate": 1.9512422139432424e-05,
      "loss": 5.6985,
      "step": 3142
    },
    {
      "epoch": 0.1268873637464675,
      "grad_norm": 132.0167999267578,
      "learning_rate": 1.951201874162229e-05,
      "loss": 6.1398,
      "step": 3143
    },
    {
      "epoch": 0.12692773516350425,
      "grad_norm": 93.77043151855469,
      "learning_rate": 1.9511615181178846e-05,
      "loss": 6.3722,
      "step": 3144
    },
    {
      "epoch": 0.12696810658054097,
      "grad_norm": 86.09420013427734,
      "learning_rate": 1.9511211458108995e-05,
      "loss": 5.1158,
      "step": 3145
    },
    {
      "epoch": 0.12700847799757772,
      "grad_norm": 198.58407592773438,
      "learning_rate": 1.951080757241964e-05,
      "loss": 4.8486,
      "step": 3146
    },
    {
      "epoch": 0.12704884941461445,
      "grad_norm": 143.14112854003906,
      "learning_rate": 1.9510403524117686e-05,
      "loss": 6.7519,
      "step": 3147
    },
    {
      "epoch": 0.1270892208316512,
      "grad_norm": 119.87458801269531,
      "learning_rate": 1.9509999313210034e-05,
      "loss": 4.5264,
      "step": 3148
    },
    {
      "epoch": 0.12712959224868792,
      "grad_norm": 125.62178802490234,
      "learning_rate": 1.9509594939703607e-05,
      "loss": 7.9084,
      "step": 3149
    },
    {
      "epoch": 0.12716996366572467,
      "grad_norm": 116.0595474243164,
      "learning_rate": 1.9509190403605313e-05,
      "loss": 5.8319,
      "step": 3150
    },
    {
      "epoch": 0.12721033508276142,
      "grad_norm": 101.18958282470703,
      "learning_rate": 1.9508785704922063e-05,
      "loss": 5.4206,
      "step": 3151
    },
    {
      "epoch": 0.12725070649979814,
      "grad_norm": 88.44606018066406,
      "learning_rate": 1.9508380843660787e-05,
      "loss": 6.1501,
      "step": 3152
    },
    {
      "epoch": 0.1272910779168349,
      "grad_norm": 102.7656021118164,
      "learning_rate": 1.95079758198284e-05,
      "loss": 6.341,
      "step": 3153
    },
    {
      "epoch": 0.1273314493338716,
      "grad_norm": 124.39763641357422,
      "learning_rate": 1.950757063343183e-05,
      "loss": 5.8145,
      "step": 3154
    },
    {
      "epoch": 0.12737182075090836,
      "grad_norm": 121.25338745117188,
      "learning_rate": 1.9507165284478007e-05,
      "loss": 5.8822,
      "step": 3155
    },
    {
      "epoch": 0.12741219216794508,
      "grad_norm": 111.05423736572266,
      "learning_rate": 1.9506759772973854e-05,
      "loss": 5.9841,
      "step": 3156
    },
    {
      "epoch": 0.12745256358498183,
      "grad_norm": 127.81483459472656,
      "learning_rate": 1.950635409892631e-05,
      "loss": 5.1001,
      "step": 3157
    },
    {
      "epoch": 0.12749293500201858,
      "grad_norm": 104.02941131591797,
      "learning_rate": 1.9505948262342313e-05,
      "loss": 5.4153,
      "step": 3158
    },
    {
      "epoch": 0.1275333064190553,
      "grad_norm": 457.1528625488281,
      "learning_rate": 1.9505542263228793e-05,
      "loss": 5.8576,
      "step": 3159
    },
    {
      "epoch": 0.12757367783609205,
      "grad_norm": 364.9958801269531,
      "learning_rate": 1.9505136101592697e-05,
      "loss": 6.0986,
      "step": 3160
    },
    {
      "epoch": 0.12761404925312878,
      "grad_norm": 106.61859893798828,
      "learning_rate": 1.9504729777440977e-05,
      "loss": 6.6491,
      "step": 3161
    },
    {
      "epoch": 0.12765442067016552,
      "grad_norm": 95.226806640625,
      "learning_rate": 1.9504323290780568e-05,
      "loss": 6.5037,
      "step": 3162
    },
    {
      "epoch": 0.12769479208720227,
      "grad_norm": 70.50912475585938,
      "learning_rate": 1.9503916641618427e-05,
      "loss": 4.7319,
      "step": 3163
    },
    {
      "epoch": 0.127735163504239,
      "grad_norm": 96.4969711303711,
      "learning_rate": 1.95035098299615e-05,
      "loss": 4.4651,
      "step": 3164
    },
    {
      "epoch": 0.12777553492127575,
      "grad_norm": 408.3289794921875,
      "learning_rate": 1.9503102855816752e-05,
      "loss": 5.5106,
      "step": 3165
    },
    {
      "epoch": 0.12781590633831247,
      "grad_norm": 82.41770935058594,
      "learning_rate": 1.9502695719191137e-05,
      "loss": 4.4673,
      "step": 3166
    },
    {
      "epoch": 0.12785627775534922,
      "grad_norm": 119.48365020751953,
      "learning_rate": 1.9502288420091613e-05,
      "loss": 5.1843,
      "step": 3167
    },
    {
      "epoch": 0.12789664917238594,
      "grad_norm": 96.78902435302734,
      "learning_rate": 1.9501880958525145e-05,
      "loss": 6.116,
      "step": 3168
    },
    {
      "epoch": 0.1279370205894227,
      "grad_norm": 181.0403289794922,
      "learning_rate": 1.9501473334498704e-05,
      "loss": 5.8933,
      "step": 3169
    },
    {
      "epoch": 0.12797739200645944,
      "grad_norm": 98.83299255371094,
      "learning_rate": 1.9501065548019254e-05,
      "loss": 5.7277,
      "step": 3170
    },
    {
      "epoch": 0.12801776342349616,
      "grad_norm": 115.35496520996094,
      "learning_rate": 1.9500657599093773e-05,
      "loss": 6.3701,
      "step": 3171
    },
    {
      "epoch": 0.1280581348405329,
      "grad_norm": 113.84945678710938,
      "learning_rate": 1.950024948772923e-05,
      "loss": 5.1417,
      "step": 3172
    },
    {
      "epoch": 0.12809850625756963,
      "grad_norm": 78.15037536621094,
      "learning_rate": 1.9499841213932606e-05,
      "loss": 7.956,
      "step": 3173
    },
    {
      "epoch": 0.12813887767460638,
      "grad_norm": 86.02912139892578,
      "learning_rate": 1.9499432777710882e-05,
      "loss": 5.6442,
      "step": 3174
    },
    {
      "epoch": 0.1281792490916431,
      "grad_norm": 83.85205841064453,
      "learning_rate": 1.949902417907104e-05,
      "loss": 3.9904,
      "step": 3175
    },
    {
      "epoch": 0.12821962050867985,
      "grad_norm": 180.8473663330078,
      "learning_rate": 1.9498615418020068e-05,
      "loss": 4.8135,
      "step": 3176
    },
    {
      "epoch": 0.1282599919257166,
      "grad_norm": 139.8681640625,
      "learning_rate": 1.9498206494564952e-05,
      "loss": 7.2173,
      "step": 3177
    },
    {
      "epoch": 0.12830036334275333,
      "grad_norm": 146.27513122558594,
      "learning_rate": 1.9497797408712686e-05,
      "loss": 4.9665,
      "step": 3178
    },
    {
      "epoch": 0.12834073475979008,
      "grad_norm": 254.8520965576172,
      "learning_rate": 1.9497388160470258e-05,
      "loss": 6.0632,
      "step": 3179
    },
    {
      "epoch": 0.1283811061768268,
      "grad_norm": 118.97098541259766,
      "learning_rate": 1.9496978749844675e-05,
      "loss": 4.7311,
      "step": 3180
    },
    {
      "epoch": 0.12842147759386355,
      "grad_norm": 118.8161392211914,
      "learning_rate": 1.9496569176842933e-05,
      "loss": 5.382,
      "step": 3181
    },
    {
      "epoch": 0.12846184901090027,
      "grad_norm": 141.4058380126953,
      "learning_rate": 1.949615944147203e-05,
      "loss": 6.5726,
      "step": 3182
    },
    {
      "epoch": 0.12850222042793702,
      "grad_norm": 126.85946655273438,
      "learning_rate": 1.949574954373898e-05,
      "loss": 4.5996,
      "step": 3183
    },
    {
      "epoch": 0.12854259184497377,
      "grad_norm": 100.27180480957031,
      "learning_rate": 1.9495339483650784e-05,
      "loss": 5.5318,
      "step": 3184
    },
    {
      "epoch": 0.1285829632620105,
      "grad_norm": 152.4421844482422,
      "learning_rate": 1.9494929261214456e-05,
      "loss": 5.3274,
      "step": 3185
    },
    {
      "epoch": 0.12862333467904724,
      "grad_norm": 68.88300323486328,
      "learning_rate": 1.9494518876437012e-05,
      "loss": 5.3742,
      "step": 3186
    },
    {
      "epoch": 0.12866370609608396,
      "grad_norm": 87.66486358642578,
      "learning_rate": 1.9494108329325467e-05,
      "loss": 4.6029,
      "step": 3187
    },
    {
      "epoch": 0.1287040775131207,
      "grad_norm": 67.03292083740234,
      "learning_rate": 1.9493697619886837e-05,
      "loss": 4.0332,
      "step": 3188
    },
    {
      "epoch": 0.12874444893015746,
      "grad_norm": 130.64210510253906,
      "learning_rate": 1.949328674812815e-05,
      "loss": 4.937,
      "step": 3189
    },
    {
      "epoch": 0.12878482034719418,
      "grad_norm": 93.13233184814453,
      "learning_rate": 1.9492875714056424e-05,
      "loss": 5.5231,
      "step": 3190
    },
    {
      "epoch": 0.12882519176423093,
      "grad_norm": 97.99835205078125,
      "learning_rate": 1.9492464517678695e-05,
      "loss": 7.082,
      "step": 3191
    },
    {
      "epoch": 0.12886556318126766,
      "grad_norm": 100.63081359863281,
      "learning_rate": 1.9492053159001986e-05,
      "loss": 5.9885,
      "step": 3192
    },
    {
      "epoch": 0.1289059345983044,
      "grad_norm": 83.83587646484375,
      "learning_rate": 1.9491641638033338e-05,
      "loss": 6.0392,
      "step": 3193
    },
    {
      "epoch": 0.12894630601534113,
      "grad_norm": 141.4703369140625,
      "learning_rate": 1.9491229954779777e-05,
      "loss": 3.4573,
      "step": 3194
    },
    {
      "epoch": 0.12898667743237788,
      "grad_norm": 92.18822479248047,
      "learning_rate": 1.949081810924835e-05,
      "loss": 5.8064,
      "step": 3195
    },
    {
      "epoch": 0.12902704884941463,
      "grad_norm": 122.73970031738281,
      "learning_rate": 1.94904061014461e-05,
      "loss": 6.5058,
      "step": 3196
    },
    {
      "epoch": 0.12906742026645135,
      "grad_norm": 152.00083923339844,
      "learning_rate": 1.948999393138006e-05,
      "loss": 4.6347,
      "step": 3197
    },
    {
      "epoch": 0.1291077916834881,
      "grad_norm": 109.86811828613281,
      "learning_rate": 1.948958159905729e-05,
      "loss": 3.5609,
      "step": 3198
    },
    {
      "epoch": 0.12914816310052482,
      "grad_norm": 206.6925048828125,
      "learning_rate": 1.948916910448483e-05,
      "loss": 6.1567,
      "step": 3199
    },
    {
      "epoch": 0.12918853451756157,
      "grad_norm": 169.38816833496094,
      "learning_rate": 1.948875644766974e-05,
      "loss": 5.7446,
      "step": 3200
    },
    {
      "epoch": 0.1292289059345983,
      "grad_norm": 105.80956268310547,
      "learning_rate": 1.948834362861907e-05,
      "loss": 5.1136,
      "step": 3201
    },
    {
      "epoch": 0.12926927735163504,
      "grad_norm": 179.93685913085938,
      "learning_rate": 1.9487930647339884e-05,
      "loss": 4.7407,
      "step": 3202
    },
    {
      "epoch": 0.1293096487686718,
      "grad_norm": 127.34744262695312,
      "learning_rate": 1.948751750383924e-05,
      "loss": 7.0137,
      "step": 3203
    },
    {
      "epoch": 0.12935002018570851,
      "grad_norm": 93.1299057006836,
      "learning_rate": 1.94871041981242e-05,
      "loss": 5.6658,
      "step": 3204
    },
    {
      "epoch": 0.12939039160274526,
      "grad_norm": 98.42888641357422,
      "learning_rate": 1.9486690730201832e-05,
      "loss": 4.431,
      "step": 3205
    },
    {
      "epoch": 0.12943076301978199,
      "grad_norm": 275.58197021484375,
      "learning_rate": 1.9486277100079207e-05,
      "loss": 4.6908,
      "step": 3206
    },
    {
      "epoch": 0.12947113443681874,
      "grad_norm": 118.04266357421875,
      "learning_rate": 1.9485863307763395e-05,
      "loss": 5.9941,
      "step": 3207
    },
    {
      "epoch": 0.12951150585385546,
      "grad_norm": 126.00170135498047,
      "learning_rate": 1.948544935326147e-05,
      "loss": 6.1959,
      "step": 3208
    },
    {
      "epoch": 0.1295518772708922,
      "grad_norm": 196.1224365234375,
      "learning_rate": 1.9485035236580513e-05,
      "loss": 5.593,
      "step": 3209
    },
    {
      "epoch": 0.12959224868792896,
      "grad_norm": 304.8804626464844,
      "learning_rate": 1.94846209577276e-05,
      "loss": 4.6819,
      "step": 3210
    },
    {
      "epoch": 0.12963262010496568,
      "grad_norm": 87.31375885009766,
      "learning_rate": 1.9484206516709822e-05,
      "loss": 4.7696,
      "step": 3211
    },
    {
      "epoch": 0.12967299152200243,
      "grad_norm": 116.06403350830078,
      "learning_rate": 1.9483791913534256e-05,
      "loss": 5.0828,
      "step": 3212
    },
    {
      "epoch": 0.12971336293903915,
      "grad_norm": 78.00618743896484,
      "learning_rate": 1.9483377148207995e-05,
      "loss": 4.6001,
      "step": 3213
    },
    {
      "epoch": 0.1297537343560759,
      "grad_norm": 253.82589721679688,
      "learning_rate": 1.948296222073813e-05,
      "loss": 5.2979,
      "step": 3214
    },
    {
      "epoch": 0.12979410577311262,
      "grad_norm": 165.45179748535156,
      "learning_rate": 1.9482547131131756e-05,
      "loss": 5.635,
      "step": 3215
    },
    {
      "epoch": 0.12983447719014937,
      "grad_norm": 158.18174743652344,
      "learning_rate": 1.9482131879395968e-05,
      "loss": 4.6624,
      "step": 3216
    },
    {
      "epoch": 0.12987484860718612,
      "grad_norm": 164.29209899902344,
      "learning_rate": 1.948171646553787e-05,
      "loss": 5.7497,
      "step": 3217
    },
    {
      "epoch": 0.12991522002422284,
      "grad_norm": 73.23007202148438,
      "learning_rate": 1.9481300889564558e-05,
      "loss": 4.7969,
      "step": 3218
    },
    {
      "epoch": 0.1299555914412596,
      "grad_norm": 81.46464538574219,
      "learning_rate": 1.9480885151483143e-05,
      "loss": 5.6643,
      "step": 3219
    },
    {
      "epoch": 0.12999596285829632,
      "grad_norm": 75.98135375976562,
      "learning_rate": 1.9480469251300733e-05,
      "loss": 5.8436,
      "step": 3220
    },
    {
      "epoch": 0.13003633427533307,
      "grad_norm": 126.89350128173828,
      "learning_rate": 1.9480053189024436e-05,
      "loss": 5.1863,
      "step": 3221
    },
    {
      "epoch": 0.13007670569236982,
      "grad_norm": 110.51506805419922,
      "learning_rate": 1.9479636964661366e-05,
      "loss": 6.6147,
      "step": 3222
    },
    {
      "epoch": 0.13011707710940654,
      "grad_norm": 199.8989715576172,
      "learning_rate": 1.9479220578218642e-05,
      "loss": 5.0177,
      "step": 3223
    },
    {
      "epoch": 0.1301574485264433,
      "grad_norm": 119.01211547851562,
      "learning_rate": 1.947880402970338e-05,
      "loss": 5.9543,
      "step": 3224
    },
    {
      "epoch": 0.13019781994348,
      "grad_norm": 209.08937072753906,
      "learning_rate": 1.9478387319122706e-05,
      "loss": 5.5865,
      "step": 3225
    },
    {
      "epoch": 0.13023819136051676,
      "grad_norm": 91.69136810302734,
      "learning_rate": 1.947797044648374e-05,
      "loss": 6.1968,
      "step": 3226
    },
    {
      "epoch": 0.13027856277755348,
      "grad_norm": 168.4530487060547,
      "learning_rate": 1.947755341179361e-05,
      "loss": 4.6273,
      "step": 3227
    },
    {
      "epoch": 0.13031893419459023,
      "grad_norm": 131.59552001953125,
      "learning_rate": 1.9477136215059455e-05,
      "loss": 3.0049,
      "step": 3228
    },
    {
      "epoch": 0.13035930561162698,
      "grad_norm": 100.44084930419922,
      "learning_rate": 1.9476718856288397e-05,
      "loss": 4.2719,
      "step": 3229
    },
    {
      "epoch": 0.1303996770286637,
      "grad_norm": 125.27420043945312,
      "learning_rate": 1.9476301335487576e-05,
      "loss": 4.7105,
      "step": 3230
    },
    {
      "epoch": 0.13044004844570045,
      "grad_norm": 85.9947280883789,
      "learning_rate": 1.9475883652664135e-05,
      "loss": 6.9898,
      "step": 3231
    },
    {
      "epoch": 0.13048041986273717,
      "grad_norm": 161.12876892089844,
      "learning_rate": 1.9475465807825206e-05,
      "loss": 5.1132,
      "step": 3232
    },
    {
      "epoch": 0.13052079127977392,
      "grad_norm": 87.92362976074219,
      "learning_rate": 1.947504780097794e-05,
      "loss": 3.893,
      "step": 3233
    },
    {
      "epoch": 0.13056116269681065,
      "grad_norm": 121.12775421142578,
      "learning_rate": 1.9474629632129485e-05,
      "loss": 5.0765,
      "step": 3234
    },
    {
      "epoch": 0.1306015341138474,
      "grad_norm": 138.9753875732422,
      "learning_rate": 1.9474211301286988e-05,
      "loss": 5.1238,
      "step": 3235
    },
    {
      "epoch": 0.13064190553088414,
      "grad_norm": 153.02346801757812,
      "learning_rate": 1.9473792808457598e-05,
      "loss": 5.6296,
      "step": 3236
    },
    {
      "epoch": 0.13068227694792087,
      "grad_norm": 248.88995361328125,
      "learning_rate": 1.9473374153648478e-05,
      "loss": 5.144,
      "step": 3237
    },
    {
      "epoch": 0.13072264836495762,
      "grad_norm": 110.5552978515625,
      "learning_rate": 1.947295533686678e-05,
      "loss": 6.0593,
      "step": 3238
    },
    {
      "epoch": 0.13076301978199434,
      "grad_norm": 207.03245544433594,
      "learning_rate": 1.9472536358119664e-05,
      "loss": 3.7894,
      "step": 3239
    },
    {
      "epoch": 0.1308033911990311,
      "grad_norm": 82.10330200195312,
      "learning_rate": 1.9472117217414302e-05,
      "loss": 4.9759,
      "step": 3240
    },
    {
      "epoch": 0.1308437626160678,
      "grad_norm": 85.54402160644531,
      "learning_rate": 1.9471697914757846e-05,
      "loss": 5.9264,
      "step": 3241
    },
    {
      "epoch": 0.13088413403310456,
      "grad_norm": 95.7772445678711,
      "learning_rate": 1.9471278450157484e-05,
      "loss": 6.4472,
      "step": 3242
    },
    {
      "epoch": 0.1309245054501413,
      "grad_norm": 159.6626739501953,
      "learning_rate": 1.9470858823620372e-05,
      "loss": 4.3685,
      "step": 3243
    },
    {
      "epoch": 0.13096487686717803,
      "grad_norm": 89.05668640136719,
      "learning_rate": 1.947043903515369e-05,
      "loss": 6.2884,
      "step": 3244
    },
    {
      "epoch": 0.13100524828421478,
      "grad_norm": 101.41351318359375,
      "learning_rate": 1.9470019084764617e-05,
      "loss": 5.2207,
      "step": 3245
    },
    {
      "epoch": 0.1310456197012515,
      "grad_norm": 154.85411071777344,
      "learning_rate": 1.9469598972460328e-05,
      "loss": 5.3538,
      "step": 3246
    },
    {
      "epoch": 0.13108599111828825,
      "grad_norm": 89.6535415649414,
      "learning_rate": 1.9469178698248015e-05,
      "loss": 6.366,
      "step": 3247
    },
    {
      "epoch": 0.131126362535325,
      "grad_norm": 82.96221160888672,
      "learning_rate": 1.9468758262134858e-05,
      "loss": 4.2658,
      "step": 3248
    },
    {
      "epoch": 0.13116673395236172,
      "grad_norm": 87.32096099853516,
      "learning_rate": 1.9468337664128043e-05,
      "loss": 6.0864,
      "step": 3249
    },
    {
      "epoch": 0.13120710536939847,
      "grad_norm": 73.81517791748047,
      "learning_rate": 1.9467916904234764e-05,
      "loss": 6.2588,
      "step": 3250
    },
    {
      "epoch": 0.1312474767864352,
      "grad_norm": 104.30469512939453,
      "learning_rate": 1.9467495982462215e-05,
      "loss": 5.5714,
      "step": 3251
    },
    {
      "epoch": 0.13128784820347195,
      "grad_norm": 132.2548065185547,
      "learning_rate": 1.9467074898817595e-05,
      "loss": 6.0432,
      "step": 3252
    },
    {
      "epoch": 0.13132821962050867,
      "grad_norm": 108.83547973632812,
      "learning_rate": 1.9466653653308102e-05,
      "loss": 3.107,
      "step": 3253
    },
    {
      "epoch": 0.13136859103754542,
      "grad_norm": 106.51299285888672,
      "learning_rate": 1.9466232245940933e-05,
      "loss": 5.9568,
      "step": 3254
    },
    {
      "epoch": 0.13140896245458217,
      "grad_norm": 84.51371765136719,
      "learning_rate": 1.9465810676723302e-05,
      "loss": 6.0527,
      "step": 3255
    },
    {
      "epoch": 0.1314493338716189,
      "grad_norm": 195.7965850830078,
      "learning_rate": 1.946538894566241e-05,
      "loss": 5.4899,
      "step": 3256
    },
    {
      "epoch": 0.13148970528865564,
      "grad_norm": 213.32115173339844,
      "learning_rate": 1.9464967052765474e-05,
      "loss": 5.088,
      "step": 3257
    },
    {
      "epoch": 0.13153007670569236,
      "grad_norm": 133.66290283203125,
      "learning_rate": 1.94645449980397e-05,
      "loss": 5.707,
      "step": 3258
    },
    {
      "epoch": 0.1315704481227291,
      "grad_norm": 89.26368713378906,
      "learning_rate": 1.9464122781492312e-05,
      "loss": 5.8986,
      "step": 3259
    },
    {
      "epoch": 0.13161081953976583,
      "grad_norm": 168.44091796875,
      "learning_rate": 1.946370040313052e-05,
      "loss": 5.3337,
      "step": 3260
    },
    {
      "epoch": 0.13165119095680258,
      "grad_norm": 91.1526870727539,
      "learning_rate": 1.9463277862961555e-05,
      "loss": 5.5012,
      "step": 3261
    },
    {
      "epoch": 0.13169156237383933,
      "grad_norm": 100.60990905761719,
      "learning_rate": 1.9462855160992635e-05,
      "loss": 4.9795,
      "step": 3262
    },
    {
      "epoch": 0.13173193379087605,
      "grad_norm": 141.54388427734375,
      "learning_rate": 1.946243229723099e-05,
      "loss": 6.3939,
      "step": 3263
    },
    {
      "epoch": 0.1317723052079128,
      "grad_norm": 250.77392578125,
      "learning_rate": 1.9462009271683845e-05,
      "loss": 5.5709,
      "step": 3264
    },
    {
      "epoch": 0.13181267662494953,
      "grad_norm": 95.60781860351562,
      "learning_rate": 1.9461586084358442e-05,
      "loss": 6.2293,
      "step": 3265
    },
    {
      "epoch": 0.13185304804198628,
      "grad_norm": 118.42851257324219,
      "learning_rate": 1.946116273526201e-05,
      "loss": 6.1646,
      "step": 3266
    },
    {
      "epoch": 0.131893419459023,
      "grad_norm": 89.53085327148438,
      "learning_rate": 1.9460739224401784e-05,
      "loss": 5.0418,
      "step": 3267
    },
    {
      "epoch": 0.13193379087605975,
      "grad_norm": 193.77220153808594,
      "learning_rate": 1.9460315551785012e-05,
      "loss": 4.9867,
      "step": 3268
    },
    {
      "epoch": 0.1319741622930965,
      "grad_norm": 109.15733337402344,
      "learning_rate": 1.9459891717418938e-05,
      "loss": 6.3944,
      "step": 3269
    },
    {
      "epoch": 0.13201453371013322,
      "grad_norm": 106.28387451171875,
      "learning_rate": 1.9459467721310807e-05,
      "loss": 6.138,
      "step": 3270
    },
    {
      "epoch": 0.13205490512716997,
      "grad_norm": 243.5049285888672,
      "learning_rate": 1.9459043563467864e-05,
      "loss": 6.2621,
      "step": 3271
    },
    {
      "epoch": 0.1320952765442067,
      "grad_norm": 109.0179214477539,
      "learning_rate": 1.9458619243897366e-05,
      "loss": 5.0656,
      "step": 3272
    },
    {
      "epoch": 0.13213564796124344,
      "grad_norm": 95.69435119628906,
      "learning_rate": 1.9458194762606564e-05,
      "loss": 5.5597,
      "step": 3273
    },
    {
      "epoch": 0.1321760193782802,
      "grad_norm": 106.61656188964844,
      "learning_rate": 1.9457770119602722e-05,
      "loss": 3.46,
      "step": 3274
    },
    {
      "epoch": 0.1322163907953169,
      "grad_norm": 144.73382568359375,
      "learning_rate": 1.9457345314893094e-05,
      "loss": 6.9422,
      "step": 3275
    },
    {
      "epoch": 0.13225676221235366,
      "grad_norm": 108.08028411865234,
      "learning_rate": 1.9456920348484948e-05,
      "loss": 5.8713,
      "step": 3276
    },
    {
      "epoch": 0.13229713362939038,
      "grad_norm": 90.83871459960938,
      "learning_rate": 1.9456495220385544e-05,
      "loss": 6.0125,
      "step": 3277
    },
    {
      "epoch": 0.13233750504642713,
      "grad_norm": 63.7999382019043,
      "learning_rate": 1.9456069930602156e-05,
      "loss": 5.4809,
      "step": 3278
    },
    {
      "epoch": 0.13237787646346386,
      "grad_norm": 91.9803466796875,
      "learning_rate": 1.9455644479142057e-05,
      "loss": 5.5495,
      "step": 3279
    },
    {
      "epoch": 0.1324182478805006,
      "grad_norm": 94.42926025390625,
      "learning_rate": 1.945521886601251e-05,
      "loss": 7.7196,
      "step": 3280
    },
    {
      "epoch": 0.13245861929753736,
      "grad_norm": 96.47698211669922,
      "learning_rate": 1.9454793091220806e-05,
      "loss": 4.0794,
      "step": 3281
    },
    {
      "epoch": 0.13249899071457408,
      "grad_norm": 128.5245819091797,
      "learning_rate": 1.9454367154774218e-05,
      "loss": 4.5893,
      "step": 3282
    },
    {
      "epoch": 0.13253936213161083,
      "grad_norm": 106.32999420166016,
      "learning_rate": 1.945394105668003e-05,
      "loss": 4.3844,
      "step": 3283
    },
    {
      "epoch": 0.13257973354864755,
      "grad_norm": 144.27597045898438,
      "learning_rate": 1.9453514796945522e-05,
      "loss": 5.9766,
      "step": 3284
    },
    {
      "epoch": 0.1326201049656843,
      "grad_norm": 88.07316589355469,
      "learning_rate": 1.9453088375577992e-05,
      "loss": 5.5383,
      "step": 3285
    },
    {
      "epoch": 0.13266047638272102,
      "grad_norm": 112.63182067871094,
      "learning_rate": 1.945266179258472e-05,
      "loss": 6.0915,
      "step": 3286
    },
    {
      "epoch": 0.13270084779975777,
      "grad_norm": 146.08230590820312,
      "learning_rate": 1.945223504797301e-05,
      "loss": 6.8816,
      "step": 3287
    },
    {
      "epoch": 0.13274121921679452,
      "grad_norm": 368.0826416015625,
      "learning_rate": 1.945180814175015e-05,
      "loss": 6.0238,
      "step": 3288
    },
    {
      "epoch": 0.13278159063383124,
      "grad_norm": 106.12474060058594,
      "learning_rate": 1.9451381073923445e-05,
      "loss": 4.7713,
      "step": 3289
    },
    {
      "epoch": 0.132821962050868,
      "grad_norm": 197.60906982421875,
      "learning_rate": 1.9450953844500192e-05,
      "loss": 6.1192,
      "step": 3290
    },
    {
      "epoch": 0.13286233346790471,
      "grad_norm": 111.8470458984375,
      "learning_rate": 1.94505264534877e-05,
      "loss": 6.9531,
      "step": 3291
    },
    {
      "epoch": 0.13290270488494146,
      "grad_norm": 108.17645263671875,
      "learning_rate": 1.9450098900893272e-05,
      "loss": 5.624,
      "step": 3292
    },
    {
      "epoch": 0.13294307630197819,
      "grad_norm": 97.54194641113281,
      "learning_rate": 1.944967118672422e-05,
      "loss": 5.9628,
      "step": 3293
    },
    {
      "epoch": 0.13298344771901494,
      "grad_norm": 194.01182556152344,
      "learning_rate": 1.9449243310987856e-05,
      "loss": 3.8082,
      "step": 3294
    },
    {
      "epoch": 0.13302381913605169,
      "grad_norm": 99.01774597167969,
      "learning_rate": 1.94488152736915e-05,
      "loss": 6.5466,
      "step": 3295
    },
    {
      "epoch": 0.1330641905530884,
      "grad_norm": 76.71434783935547,
      "learning_rate": 1.944838707484247e-05,
      "loss": 3.4665,
      "step": 3296
    },
    {
      "epoch": 0.13310456197012516,
      "grad_norm": 81.6198959350586,
      "learning_rate": 1.9447958714448075e-05,
      "loss": 6.5242,
      "step": 3297
    },
    {
      "epoch": 0.13314493338716188,
      "grad_norm": 109.91975402832031,
      "learning_rate": 1.944753019251566e-05,
      "loss": 4.8455,
      "step": 3298
    },
    {
      "epoch": 0.13318530480419863,
      "grad_norm": 82.9660415649414,
      "learning_rate": 1.9447101509052532e-05,
      "loss": 7.831,
      "step": 3299
    },
    {
      "epoch": 0.13322567622123538,
      "grad_norm": 103.75143432617188,
      "learning_rate": 1.944667266406603e-05,
      "loss": 6.4679,
      "step": 3300
    },
    {
      "epoch": 0.1332660476382721,
      "grad_norm": 154.2065887451172,
      "learning_rate": 1.9446243657563486e-05,
      "loss": 4.9364,
      "step": 3301
    },
    {
      "epoch": 0.13330641905530885,
      "grad_norm": 167.5544891357422,
      "learning_rate": 1.9445814489552237e-05,
      "loss": 6.9331,
      "step": 3302
    },
    {
      "epoch": 0.13334679047234557,
      "grad_norm": 88.5614013671875,
      "learning_rate": 1.9445385160039615e-05,
      "loss": 6.5092,
      "step": 3303
    },
    {
      "epoch": 0.13338716188938232,
      "grad_norm": 139.28005981445312,
      "learning_rate": 1.9444955669032962e-05,
      "loss": 5.5774,
      "step": 3304
    },
    {
      "epoch": 0.13342753330641904,
      "grad_norm": 118.83869934082031,
      "learning_rate": 1.9444526016539624e-05,
      "loss": 5.198,
      "step": 3305
    },
    {
      "epoch": 0.1334679047234558,
      "grad_norm": 89.3421859741211,
      "learning_rate": 1.9444096202566948e-05,
      "loss": 4.2894,
      "step": 3306
    },
    {
      "epoch": 0.13350827614049254,
      "grad_norm": 129.38438415527344,
      "learning_rate": 1.9443666227122278e-05,
      "loss": 6.8367,
      "step": 3307
    },
    {
      "epoch": 0.13354864755752927,
      "grad_norm": 141.30955505371094,
      "learning_rate": 1.944323609021297e-05,
      "loss": 5.3659,
      "step": 3308
    },
    {
      "epoch": 0.13358901897456602,
      "grad_norm": 108.14979553222656,
      "learning_rate": 1.9442805791846372e-05,
      "loss": 4.1858,
      "step": 3309
    },
    {
      "epoch": 0.13362939039160274,
      "grad_norm": 88.82259368896484,
      "learning_rate": 1.944237533202985e-05,
      "loss": 6.5486,
      "step": 3310
    },
    {
      "epoch": 0.1336697618086395,
      "grad_norm": 230.26577758789062,
      "learning_rate": 1.944194471077076e-05,
      "loss": 6.1125,
      "step": 3311
    },
    {
      "epoch": 0.1337101332256762,
      "grad_norm": 192.3295135498047,
      "learning_rate": 1.944151392807646e-05,
      "loss": 5.8023,
      "step": 3312
    },
    {
      "epoch": 0.13375050464271296,
      "grad_norm": 145.8681640625,
      "learning_rate": 1.944108298395432e-05,
      "loss": 6.8911,
      "step": 3313
    },
    {
      "epoch": 0.1337908760597497,
      "grad_norm": 165.04254150390625,
      "learning_rate": 1.944065187841171e-05,
      "loss": 5.8427,
      "step": 3314
    },
    {
      "epoch": 0.13383124747678643,
      "grad_norm": 104.4214096069336,
      "learning_rate": 1.9440220611455997e-05,
      "loss": 7.078,
      "step": 3315
    },
    {
      "epoch": 0.13387161889382318,
      "grad_norm": 112.21549224853516,
      "learning_rate": 1.9439789183094557e-05,
      "loss": 4.661,
      "step": 3316
    },
    {
      "epoch": 0.1339119903108599,
      "grad_norm": 146.01173400878906,
      "learning_rate": 1.9439357593334763e-05,
      "loss": 4.9784,
      "step": 3317
    },
    {
      "epoch": 0.13395236172789665,
      "grad_norm": 68.00843811035156,
      "learning_rate": 1.9438925842183998e-05,
      "loss": 6.1325,
      "step": 3318
    },
    {
      "epoch": 0.13399273314493337,
      "grad_norm": 129.29934692382812,
      "learning_rate": 1.9438493929649645e-05,
      "loss": 5.5168,
      "step": 3319
    },
    {
      "epoch": 0.13403310456197012,
      "grad_norm": 83.48184204101562,
      "learning_rate": 1.9438061855739083e-05,
      "loss": 6.1087,
      "step": 3320
    },
    {
      "epoch": 0.13407347597900687,
      "grad_norm": 85.13540649414062,
      "learning_rate": 1.9437629620459702e-05,
      "loss": 5.0517,
      "step": 3321
    },
    {
      "epoch": 0.1341138473960436,
      "grad_norm": 101.96695709228516,
      "learning_rate": 1.9437197223818894e-05,
      "loss": 4.8153,
      "step": 3322
    },
    {
      "epoch": 0.13415421881308034,
      "grad_norm": 107.2568359375,
      "learning_rate": 1.9436764665824053e-05,
      "loss": 5.3692,
      "step": 3323
    },
    {
      "epoch": 0.13419459023011707,
      "grad_norm": 89.34864044189453,
      "learning_rate": 1.943633194648257e-05,
      "loss": 5.4074,
      "step": 3324
    },
    {
      "epoch": 0.13423496164715382,
      "grad_norm": 95.2779312133789,
      "learning_rate": 1.9435899065801848e-05,
      "loss": 5.1342,
      "step": 3325
    },
    {
      "epoch": 0.13427533306419057,
      "grad_norm": 112.75835418701172,
      "learning_rate": 1.9435466023789287e-05,
      "loss": 5.5968,
      "step": 3326
    },
    {
      "epoch": 0.1343157044812273,
      "grad_norm": 130.62464904785156,
      "learning_rate": 1.9435032820452288e-05,
      "loss": 6.6243,
      "step": 3327
    },
    {
      "epoch": 0.13435607589826404,
      "grad_norm": 182.60084533691406,
      "learning_rate": 1.943459945579826e-05,
      "loss": 6.2529,
      "step": 3328
    },
    {
      "epoch": 0.13439644731530076,
      "grad_norm": 111.53166198730469,
      "learning_rate": 1.9434165929834618e-05,
      "loss": 5.4003,
      "step": 3329
    },
    {
      "epoch": 0.1344368187323375,
      "grad_norm": 116.57920837402344,
      "learning_rate": 1.9433732242568763e-05,
      "loss": 5.7743,
      "step": 3330
    },
    {
      "epoch": 0.13447719014937423,
      "grad_norm": 249.3780517578125,
      "learning_rate": 1.9433298394008116e-05,
      "loss": 5.4822,
      "step": 3331
    },
    {
      "epoch": 0.13451756156641098,
      "grad_norm": 112.14811706542969,
      "learning_rate": 1.94328643841601e-05,
      "loss": 6.2463,
      "step": 3332
    },
    {
      "epoch": 0.13455793298344773,
      "grad_norm": 135.988037109375,
      "learning_rate": 1.9432430213032126e-05,
      "loss": 6.4099,
      "step": 3333
    },
    {
      "epoch": 0.13459830440048445,
      "grad_norm": 101.42012786865234,
      "learning_rate": 1.9431995880631623e-05,
      "loss": 5.1661,
      "step": 3334
    },
    {
      "epoch": 0.1346386758175212,
      "grad_norm": 111.67835998535156,
      "learning_rate": 1.9431561386966015e-05,
      "loss": 6.4289,
      "step": 3335
    },
    {
      "epoch": 0.13467904723455792,
      "grad_norm": 91.31681060791016,
      "learning_rate": 1.943112673204273e-05,
      "loss": 4.9303,
      "step": 3336
    },
    {
      "epoch": 0.13471941865159467,
      "grad_norm": 148.42626953125,
      "learning_rate": 1.9430691915869206e-05,
      "loss": 5.315,
      "step": 3337
    },
    {
      "epoch": 0.1347597900686314,
      "grad_norm": 154.7969207763672,
      "learning_rate": 1.943025693845287e-05,
      "loss": 5.672,
      "step": 3338
    },
    {
      "epoch": 0.13480016148566815,
      "grad_norm": 160.3094940185547,
      "learning_rate": 1.942982179980116e-05,
      "loss": 6.9855,
      "step": 3339
    },
    {
      "epoch": 0.1348405329027049,
      "grad_norm": 128.33444213867188,
      "learning_rate": 1.942938649992152e-05,
      "loss": 5.742,
      "step": 3340
    },
    {
      "epoch": 0.13488090431974162,
      "grad_norm": 142.6033172607422,
      "learning_rate": 1.9428951038821386e-05,
      "loss": 5.5172,
      "step": 3341
    },
    {
      "epoch": 0.13492127573677837,
      "grad_norm": 84.66514587402344,
      "learning_rate": 1.9428515416508214e-05,
      "loss": 6.1173,
      "step": 3342
    },
    {
      "epoch": 0.1349616471538151,
      "grad_norm": 211.94583129882812,
      "learning_rate": 1.942807963298944e-05,
      "loss": 6.0091,
      "step": 3343
    },
    {
      "epoch": 0.13500201857085184,
      "grad_norm": 136.3842010498047,
      "learning_rate": 1.9427643688272523e-05,
      "loss": 6.4946,
      "step": 3344
    },
    {
      "epoch": 0.13504238998788856,
      "grad_norm": 116.39862060546875,
      "learning_rate": 1.942720758236491e-05,
      "loss": 5.9893,
      "step": 3345
    },
    {
      "epoch": 0.1350827614049253,
      "grad_norm": 112.91094970703125,
      "learning_rate": 1.9426771315274068e-05,
      "loss": 5.6848,
      "step": 3346
    },
    {
      "epoch": 0.13512313282196206,
      "grad_norm": 184.3402862548828,
      "learning_rate": 1.9426334887007444e-05,
      "loss": 5.5758,
      "step": 3347
    },
    {
      "epoch": 0.13516350423899878,
      "grad_norm": 240.90696716308594,
      "learning_rate": 1.942589829757251e-05,
      "loss": 6.6528,
      "step": 3348
    },
    {
      "epoch": 0.13520387565603553,
      "grad_norm": 140.7642364501953,
      "learning_rate": 1.9425461546976722e-05,
      "loss": 4.902,
      "step": 3349
    },
    {
      "epoch": 0.13524424707307225,
      "grad_norm": 111.94091796875,
      "learning_rate": 1.9425024635227555e-05,
      "loss": 5.7496,
      "step": 3350
    },
    {
      "epoch": 0.135284618490109,
      "grad_norm": 103.6980972290039,
      "learning_rate": 1.9424587562332476e-05,
      "loss": 5.9216,
      "step": 3351
    },
    {
      "epoch": 0.13532498990714575,
      "grad_norm": 151.18527221679688,
      "learning_rate": 1.9424150328298955e-05,
      "loss": 5.8425,
      "step": 3352
    },
    {
      "epoch": 0.13536536132418248,
      "grad_norm": 103.53816986083984,
      "learning_rate": 1.942371293313447e-05,
      "loss": 6.149,
      "step": 3353
    },
    {
      "epoch": 0.13540573274121923,
      "grad_norm": 98.70388793945312,
      "learning_rate": 1.94232753768465e-05,
      "loss": 5.3575,
      "step": 3354
    },
    {
      "epoch": 0.13544610415825595,
      "grad_norm": 80.51641845703125,
      "learning_rate": 1.942283765944253e-05,
      "loss": 6.1759,
      "step": 3355
    },
    {
      "epoch": 0.1354864755752927,
      "grad_norm": 83.75017547607422,
      "learning_rate": 1.9422399780930035e-05,
      "loss": 5.8526,
      "step": 3356
    },
    {
      "epoch": 0.13552684699232942,
      "grad_norm": 120.78399658203125,
      "learning_rate": 1.942196174131651e-05,
      "loss": 5.4826,
      "step": 3357
    },
    {
      "epoch": 0.13556721840936617,
      "grad_norm": 96.30940246582031,
      "learning_rate": 1.942152354060944e-05,
      "loss": 7.3024,
      "step": 3358
    },
    {
      "epoch": 0.13560758982640292,
      "grad_norm": 125.32355499267578,
      "learning_rate": 1.9421085178816318e-05,
      "loss": 5.4453,
      "step": 3359
    },
    {
      "epoch": 0.13564796124343964,
      "grad_norm": 128.83120727539062,
      "learning_rate": 1.942064665594464e-05,
      "loss": 4.6941,
      "step": 3360
    },
    {
      "epoch": 0.1356883326604764,
      "grad_norm": 131.9399871826172,
      "learning_rate": 1.94202079720019e-05,
      "loss": 5.2921,
      "step": 3361
    },
    {
      "epoch": 0.1357287040775131,
      "grad_norm": 102.6584701538086,
      "learning_rate": 1.9419769126995604e-05,
      "loss": 5.9052,
      "step": 3362
    },
    {
      "epoch": 0.13576907549454986,
      "grad_norm": 183.1845703125,
      "learning_rate": 1.9419330120933253e-05,
      "loss": 5.4111,
      "step": 3363
    },
    {
      "epoch": 0.13580944691158658,
      "grad_norm": 110.43077087402344,
      "learning_rate": 1.9418890953822352e-05,
      "loss": 5.6846,
      "step": 3364
    },
    {
      "epoch": 0.13584981832862333,
      "grad_norm": 119.34333801269531,
      "learning_rate": 1.941845162567041e-05,
      "loss": 6.1384,
      "step": 3365
    },
    {
      "epoch": 0.13589018974566008,
      "grad_norm": 143.51966857910156,
      "learning_rate": 1.9418012136484942e-05,
      "loss": 4.3893,
      "step": 3366
    },
    {
      "epoch": 0.1359305611626968,
      "grad_norm": 126.89920043945312,
      "learning_rate": 1.9417572486273455e-05,
      "loss": 5.5538,
      "step": 3367
    },
    {
      "epoch": 0.13597093257973356,
      "grad_norm": 112.1628646850586,
      "learning_rate": 1.941713267504347e-05,
      "loss": 6.1823,
      "step": 3368
    },
    {
      "epoch": 0.13601130399677028,
      "grad_norm": 92.99742889404297,
      "learning_rate": 1.941669270280251e-05,
      "loss": 5.0125,
      "step": 3369
    },
    {
      "epoch": 0.13605167541380703,
      "grad_norm": 99.05742645263672,
      "learning_rate": 1.941625256955809e-05,
      "loss": 5.5084,
      "step": 3370
    },
    {
      "epoch": 0.13609204683084375,
      "grad_norm": 105.65899658203125,
      "learning_rate": 1.9415812275317743e-05,
      "loss": 4.8656,
      "step": 3371
    },
    {
      "epoch": 0.1361324182478805,
      "grad_norm": 96.91753387451172,
      "learning_rate": 1.9415371820088996e-05,
      "loss": 7.0081,
      "step": 3372
    },
    {
      "epoch": 0.13617278966491725,
      "grad_norm": 100.9683609008789,
      "learning_rate": 1.941493120387937e-05,
      "loss": 5.3268,
      "step": 3373
    },
    {
      "epoch": 0.13621316108195397,
      "grad_norm": 102.24005126953125,
      "learning_rate": 1.9414490426696413e-05,
      "loss": 6.4213,
      "step": 3374
    },
    {
      "epoch": 0.13625353249899072,
      "grad_norm": 90.09848022460938,
      "learning_rate": 1.941404948854765e-05,
      "loss": 5.9601,
      "step": 3375
    },
    {
      "epoch": 0.13629390391602744,
      "grad_norm": 463.209228515625,
      "learning_rate": 1.9413608389440625e-05,
      "loss": 5.7331,
      "step": 3376
    },
    {
      "epoch": 0.1363342753330642,
      "grad_norm": 208.20904541015625,
      "learning_rate": 1.941316712938288e-05,
      "loss": 5.754,
      "step": 3377
    },
    {
      "epoch": 0.13637464675010094,
      "grad_norm": 95.92414093017578,
      "learning_rate": 1.9412725708381958e-05,
      "loss": 7.3838,
      "step": 3378
    },
    {
      "epoch": 0.13641501816713766,
      "grad_norm": 84.8706283569336,
      "learning_rate": 1.9412284126445407e-05,
      "loss": 6.312,
      "step": 3379
    },
    {
      "epoch": 0.1364553895841744,
      "grad_norm": 269.07281494140625,
      "learning_rate": 1.9411842383580775e-05,
      "loss": 4.7389,
      "step": 3380
    },
    {
      "epoch": 0.13649576100121114,
      "grad_norm": 137.5897674560547,
      "learning_rate": 1.9411400479795618e-05,
      "loss": 5.8353,
      "step": 3381
    },
    {
      "epoch": 0.13653613241824789,
      "grad_norm": 100.27400207519531,
      "learning_rate": 1.941095841509749e-05,
      "loss": 5.2366,
      "step": 3382
    },
    {
      "epoch": 0.1365765038352846,
      "grad_norm": 129.56777954101562,
      "learning_rate": 1.941051618949395e-05,
      "loss": 3.3494,
      "step": 3383
    },
    {
      "epoch": 0.13661687525232136,
      "grad_norm": 138.2855682373047,
      "learning_rate": 1.9410073802992555e-05,
      "loss": 6.6008,
      "step": 3384
    },
    {
      "epoch": 0.1366572466693581,
      "grad_norm": 117.17613983154297,
      "learning_rate": 1.9409631255600875e-05,
      "loss": 3.4698,
      "step": 3385
    },
    {
      "epoch": 0.13669761808639483,
      "grad_norm": 69.67994689941406,
      "learning_rate": 1.940918854732647e-05,
      "loss": 5.0542,
      "step": 3386
    },
    {
      "epoch": 0.13673798950343158,
      "grad_norm": 191.21353149414062,
      "learning_rate": 1.9408745678176918e-05,
      "loss": 5.9425,
      "step": 3387
    },
    {
      "epoch": 0.1367783609204683,
      "grad_norm": 175.67758178710938,
      "learning_rate": 1.9408302648159783e-05,
      "loss": 6.3047,
      "step": 3388
    },
    {
      "epoch": 0.13681873233750505,
      "grad_norm": 110.79118347167969,
      "learning_rate": 1.940785945728264e-05,
      "loss": 6.5838,
      "step": 3389
    },
    {
      "epoch": 0.13685910375454177,
      "grad_norm": 82.39977264404297,
      "learning_rate": 1.9407416105553075e-05,
      "loss": 5.6028,
      "step": 3390
    },
    {
      "epoch": 0.13689947517157852,
      "grad_norm": 133.2849884033203,
      "learning_rate": 1.940697259297866e-05,
      "loss": 6.2807,
      "step": 3391
    },
    {
      "epoch": 0.13693984658861527,
      "grad_norm": 125.89656829833984,
      "learning_rate": 1.9406528919566982e-05,
      "loss": 6.2687,
      "step": 3392
    },
    {
      "epoch": 0.136980218005652,
      "grad_norm": 96.854736328125,
      "learning_rate": 1.9406085085325623e-05,
      "loss": 4.8216,
      "step": 3393
    },
    {
      "epoch": 0.13702058942268874,
      "grad_norm": 157.70254516601562,
      "learning_rate": 1.9405641090262174e-05,
      "loss": 5.463,
      "step": 3394
    },
    {
      "epoch": 0.13706096083972547,
      "grad_norm": 118.19808197021484,
      "learning_rate": 1.9405196934384224e-05,
      "loss": 4.0495,
      "step": 3395
    },
    {
      "epoch": 0.13710133225676222,
      "grad_norm": 105.85313415527344,
      "learning_rate": 1.9404752617699373e-05,
      "loss": 6.0704,
      "step": 3396
    },
    {
      "epoch": 0.13714170367379894,
      "grad_norm": 127.5667495727539,
      "learning_rate": 1.9404308140215217e-05,
      "loss": 5.7017,
      "step": 3397
    },
    {
      "epoch": 0.1371820750908357,
      "grad_norm": 143.86773681640625,
      "learning_rate": 1.9403863501939345e-05,
      "loss": 4.2753,
      "step": 3398
    },
    {
      "epoch": 0.13722244650787244,
      "grad_norm": 85.23965454101562,
      "learning_rate": 1.9403418702879372e-05,
      "loss": 5.5905,
      "step": 3399
    },
    {
      "epoch": 0.13726281792490916,
      "grad_norm": 290.0005187988281,
      "learning_rate": 1.9402973743042892e-05,
      "loss": 7.4643,
      "step": 3400
    },
    {
      "epoch": 0.1373031893419459,
      "grad_norm": 71.74673461914062,
      "learning_rate": 1.9402528622437526e-05,
      "loss": 5.8323,
      "step": 3401
    },
    {
      "epoch": 0.13734356075898263,
      "grad_norm": 107.08522033691406,
      "learning_rate": 1.9402083341070873e-05,
      "loss": 5.5889,
      "step": 3402
    },
    {
      "epoch": 0.13738393217601938,
      "grad_norm": 103.03728485107422,
      "learning_rate": 1.940163789895055e-05,
      "loss": 6.0339,
      "step": 3403
    },
    {
      "epoch": 0.13742430359305613,
      "grad_norm": 119.68949127197266,
      "learning_rate": 1.9401192296084173e-05,
      "loss": 6.1457,
      "step": 3404
    },
    {
      "epoch": 0.13746467501009285,
      "grad_norm": 86.30003356933594,
      "learning_rate": 1.940074653247936e-05,
      "loss": 5.1416,
      "step": 3405
    },
    {
      "epoch": 0.1375050464271296,
      "grad_norm": 83.71137237548828,
      "learning_rate": 1.9400300608143735e-05,
      "loss": 5.822,
      "step": 3406
    },
    {
      "epoch": 0.13754541784416632,
      "grad_norm": 78.01338195800781,
      "learning_rate": 1.9399854523084923e-05,
      "loss": 5.8128,
      "step": 3407
    },
    {
      "epoch": 0.13758578926120307,
      "grad_norm": 146.22154235839844,
      "learning_rate": 1.9399408277310548e-05,
      "loss": 5.7775,
      "step": 3408
    },
    {
      "epoch": 0.1376261606782398,
      "grad_norm": 110.7657470703125,
      "learning_rate": 1.939896187082824e-05,
      "loss": 6.296,
      "step": 3409
    },
    {
      "epoch": 0.13766653209527654,
      "grad_norm": 158.32252502441406,
      "learning_rate": 1.939851530364563e-05,
      "loss": 5.1153,
      "step": 3410
    },
    {
      "epoch": 0.1377069035123133,
      "grad_norm": 79.98668670654297,
      "learning_rate": 1.939806857577036e-05,
      "loss": 5.4431,
      "step": 3411
    },
    {
      "epoch": 0.13774727492935002,
      "grad_norm": 159.1708221435547,
      "learning_rate": 1.939762168721006e-05,
      "loss": 5.6677,
      "step": 3412
    },
    {
      "epoch": 0.13778764634638677,
      "grad_norm": 188.1392364501953,
      "learning_rate": 1.9397174637972375e-05,
      "loss": 5.5917,
      "step": 3413
    },
    {
      "epoch": 0.1378280177634235,
      "grad_norm": 124.83942413330078,
      "learning_rate": 1.9396727428064946e-05,
      "loss": 7.1905,
      "step": 3414
    },
    {
      "epoch": 0.13786838918046024,
      "grad_norm": 93.80907440185547,
      "learning_rate": 1.9396280057495425e-05,
      "loss": 6.8121,
      "step": 3415
    },
    {
      "epoch": 0.13790876059749696,
      "grad_norm": 107.45306396484375,
      "learning_rate": 1.9395832526271456e-05,
      "loss": 6.1932,
      "step": 3416
    },
    {
      "epoch": 0.1379491320145337,
      "grad_norm": 101.43861389160156,
      "learning_rate": 1.9395384834400688e-05,
      "loss": 4.8089,
      "step": 3417
    },
    {
      "epoch": 0.13798950343157046,
      "grad_norm": 80.82820129394531,
      "learning_rate": 1.9394936981890782e-05,
      "loss": 6.1251,
      "step": 3418
    },
    {
      "epoch": 0.13802987484860718,
      "grad_norm": 120.01830291748047,
      "learning_rate": 1.9394488968749392e-05,
      "loss": 5.7772,
      "step": 3419
    },
    {
      "epoch": 0.13807024626564393,
      "grad_norm": 73.3323745727539,
      "learning_rate": 1.939404079498418e-05,
      "loss": 5.8862,
      "step": 3420
    },
    {
      "epoch": 0.13811061768268065,
      "grad_norm": 120.79756927490234,
      "learning_rate": 1.9393592460602803e-05,
      "loss": 2.6042,
      "step": 3421
    },
    {
      "epoch": 0.1381509890997174,
      "grad_norm": 109.55097198486328,
      "learning_rate": 1.939314396561293e-05,
      "loss": 5.0073,
      "step": 3422
    },
    {
      "epoch": 0.13819136051675412,
      "grad_norm": 101.20858764648438,
      "learning_rate": 1.9392695310022235e-05,
      "loss": 5.9709,
      "step": 3423
    },
    {
      "epoch": 0.13823173193379087,
      "grad_norm": 89.82926940917969,
      "learning_rate": 1.9392246493838382e-05,
      "loss": 6.3348,
      "step": 3424
    },
    {
      "epoch": 0.13827210335082762,
      "grad_norm": 92.95947265625,
      "learning_rate": 1.9391797517069045e-05,
      "loss": 5.2315,
      "step": 3425
    },
    {
      "epoch": 0.13831247476786435,
      "grad_norm": 106.9877700805664,
      "learning_rate": 1.9391348379721902e-05,
      "loss": 5.8508,
      "step": 3426
    },
    {
      "epoch": 0.1383528461849011,
      "grad_norm": 95.5060043334961,
      "learning_rate": 1.9390899081804634e-05,
      "loss": 6.1176,
      "step": 3427
    },
    {
      "epoch": 0.13839321760193782,
      "grad_norm": 572.8359985351562,
      "learning_rate": 1.939044962332492e-05,
      "loss": 6.07,
      "step": 3428
    },
    {
      "epoch": 0.13843358901897457,
      "grad_norm": 135.0657501220703,
      "learning_rate": 1.9390000004290445e-05,
      "loss": 5.0093,
      "step": 3429
    },
    {
      "epoch": 0.13847396043601132,
      "grad_norm": 90.7294692993164,
      "learning_rate": 1.9389550224708898e-05,
      "loss": 4.8581,
      "step": 3430
    },
    {
      "epoch": 0.13851433185304804,
      "grad_norm": 147.1729278564453,
      "learning_rate": 1.9389100284587966e-05,
      "loss": 5.6834,
      "step": 3431
    },
    {
      "epoch": 0.1385547032700848,
      "grad_norm": 127.37533569335938,
      "learning_rate": 1.938865018393535e-05,
      "loss": 5.2041,
      "step": 3432
    },
    {
      "epoch": 0.1385950746871215,
      "grad_norm": 132.1737823486328,
      "learning_rate": 1.9388199922758732e-05,
      "loss": 5.3039,
      "step": 3433
    },
    {
      "epoch": 0.13863544610415826,
      "grad_norm": 89.04087829589844,
      "learning_rate": 1.938774950106582e-05,
      "loss": 5.9975,
      "step": 3434
    },
    {
      "epoch": 0.13867581752119498,
      "grad_norm": 152.69482421875,
      "learning_rate": 1.938729891886432e-05,
      "loss": 4.57,
      "step": 3435
    },
    {
      "epoch": 0.13871618893823173,
      "grad_norm": 149.942138671875,
      "learning_rate": 1.9386848176161925e-05,
      "loss": 4.3957,
      "step": 3436
    },
    {
      "epoch": 0.13875656035526848,
      "grad_norm": 143.32833862304688,
      "learning_rate": 1.9386397272966345e-05,
      "loss": 5.6777,
      "step": 3437
    },
    {
      "epoch": 0.1387969317723052,
      "grad_norm": 98.95429992675781,
      "learning_rate": 1.938594620928529e-05,
      "loss": 5.2818,
      "step": 3438
    },
    {
      "epoch": 0.13883730318934195,
      "grad_norm": 103.62015533447266,
      "learning_rate": 1.9385494985126476e-05,
      "loss": 4.2726,
      "step": 3439
    },
    {
      "epoch": 0.13887767460637868,
      "grad_norm": 232.03982543945312,
      "learning_rate": 1.9385043600497614e-05,
      "loss": 3.7144,
      "step": 3440
    },
    {
      "epoch": 0.13891804602341543,
      "grad_norm": 97.54658508300781,
      "learning_rate": 1.938459205540642e-05,
      "loss": 3.7564,
      "step": 3441
    },
    {
      "epoch": 0.13895841744045215,
      "grad_norm": 150.98878479003906,
      "learning_rate": 1.9384140349860615e-05,
      "loss": 5.249,
      "step": 3442
    },
    {
      "epoch": 0.1389987888574889,
      "grad_norm": 111.62024688720703,
      "learning_rate": 1.9383688483867926e-05,
      "loss": 7.1765,
      "step": 3443
    },
    {
      "epoch": 0.13903916027452565,
      "grad_norm": 153.860107421875,
      "learning_rate": 1.9383236457436077e-05,
      "loss": 5.1704,
      "step": 3444
    },
    {
      "epoch": 0.13907953169156237,
      "grad_norm": 133.12637329101562,
      "learning_rate": 1.9382784270572796e-05,
      "loss": 5.4835,
      "step": 3445
    },
    {
      "epoch": 0.13911990310859912,
      "grad_norm": 665.4515991210938,
      "learning_rate": 1.9382331923285815e-05,
      "loss": 6.1667,
      "step": 3446
    },
    {
      "epoch": 0.13916027452563584,
      "grad_norm": 112.1897964477539,
      "learning_rate": 1.9381879415582866e-05,
      "loss": 5.587,
      "step": 3447
    },
    {
      "epoch": 0.1392006459426726,
      "grad_norm": 114.30377960205078,
      "learning_rate": 1.9381426747471687e-05,
      "loss": 5.5438,
      "step": 3448
    },
    {
      "epoch": 0.1392410173597093,
      "grad_norm": 109.46294403076172,
      "learning_rate": 1.9380973918960016e-05,
      "loss": 5.9005,
      "step": 3449
    },
    {
      "epoch": 0.13928138877674606,
      "grad_norm": 106.91893768310547,
      "learning_rate": 1.9380520930055603e-05,
      "loss": 4.8762,
      "step": 3450
    },
    {
      "epoch": 0.1393217601937828,
      "grad_norm": 87.37007904052734,
      "learning_rate": 1.938006778076618e-05,
      "loss": 4.701,
      "step": 3451
    },
    {
      "epoch": 0.13936213161081953,
      "grad_norm": 131.04763793945312,
      "learning_rate": 1.9379614471099506e-05,
      "loss": 5.3241,
      "step": 3452
    },
    {
      "epoch": 0.13940250302785628,
      "grad_norm": 192.5240936279297,
      "learning_rate": 1.9379161001063327e-05,
      "loss": 4.4516,
      "step": 3453
    },
    {
      "epoch": 0.139442874444893,
      "grad_norm": 109.56015014648438,
      "learning_rate": 1.9378707370665396e-05,
      "loss": 5.8176,
      "step": 3454
    },
    {
      "epoch": 0.13948324586192976,
      "grad_norm": 97.68437194824219,
      "learning_rate": 1.9378253579913472e-05,
      "loss": 6.3807,
      "step": 3455
    },
    {
      "epoch": 0.1395236172789665,
      "grad_norm": 209.12344360351562,
      "learning_rate": 1.937779962881531e-05,
      "loss": 5.6843,
      "step": 3456
    },
    {
      "epoch": 0.13956398869600323,
      "grad_norm": 229.84671020507812,
      "learning_rate": 1.9377345517378673e-05,
      "loss": 4.699,
      "step": 3457
    },
    {
      "epoch": 0.13960436011303998,
      "grad_norm": 112.0840072631836,
      "learning_rate": 1.9376891245611328e-05,
      "loss": 4.8258,
      "step": 3458
    },
    {
      "epoch": 0.1396447315300767,
      "grad_norm": 209.38330078125,
      "learning_rate": 1.9376436813521036e-05,
      "loss": 6.237,
      "step": 3459
    },
    {
      "epoch": 0.13968510294711345,
      "grad_norm": 148.67926025390625,
      "learning_rate": 1.937598222111557e-05,
      "loss": 5.4114,
      "step": 3460
    },
    {
      "epoch": 0.13972547436415017,
      "grad_norm": 116.4202880859375,
      "learning_rate": 1.9375527468402703e-05,
      "loss": 5.1466,
      "step": 3461
    },
    {
      "epoch": 0.13976584578118692,
      "grad_norm": 253.19827270507812,
      "learning_rate": 1.937507255539021e-05,
      "loss": 7.0873,
      "step": 3462
    },
    {
      "epoch": 0.13980621719822367,
      "grad_norm": 143.52891540527344,
      "learning_rate": 1.937461748208587e-05,
      "loss": 6.2238,
      "step": 3463
    },
    {
      "epoch": 0.1398465886152604,
      "grad_norm": 158.23468017578125,
      "learning_rate": 1.9374162248497462e-05,
      "loss": 6.7467,
      "step": 3464
    },
    {
      "epoch": 0.13988696003229714,
      "grad_norm": 109.09382629394531,
      "learning_rate": 1.9373706854632768e-05,
      "loss": 5.5147,
      "step": 3465
    },
    {
      "epoch": 0.13992733144933386,
      "grad_norm": 145.29591369628906,
      "learning_rate": 1.9373251300499577e-05,
      "loss": 6.1363,
      "step": 3466
    },
    {
      "epoch": 0.1399677028663706,
      "grad_norm": 124.24031066894531,
      "learning_rate": 1.9372795586105676e-05,
      "loss": 6.5833,
      "step": 3467
    },
    {
      "epoch": 0.14000807428340734,
      "grad_norm": 74.6865463256836,
      "learning_rate": 1.937233971145886e-05,
      "loss": 6.3999,
      "step": 3468
    },
    {
      "epoch": 0.14004844570044409,
      "grad_norm": 83.77539825439453,
      "learning_rate": 1.937188367656692e-05,
      "loss": 6.8851,
      "step": 3469
    },
    {
      "epoch": 0.14008881711748083,
      "grad_norm": 88.30918884277344,
      "learning_rate": 1.9371427481437653e-05,
      "loss": 5.987,
      "step": 3470
    },
    {
      "epoch": 0.14012918853451756,
      "grad_norm": 110.93408203125,
      "learning_rate": 1.937097112607886e-05,
      "loss": 5.0122,
      "step": 3471
    },
    {
      "epoch": 0.1401695599515543,
      "grad_norm": 85.93838500976562,
      "learning_rate": 1.9370514610498345e-05,
      "loss": 6.1008,
      "step": 3472
    },
    {
      "epoch": 0.14020993136859103,
      "grad_norm": 121.27180480957031,
      "learning_rate": 1.937005793470391e-05,
      "loss": 4.9272,
      "step": 3473
    },
    {
      "epoch": 0.14025030278562778,
      "grad_norm": 106.64958190917969,
      "learning_rate": 1.9369601098703365e-05,
      "loss": 5.7202,
      "step": 3474
    },
    {
      "epoch": 0.1402906742026645,
      "grad_norm": 88.40930938720703,
      "learning_rate": 1.9369144102504523e-05,
      "loss": 6.5601,
      "step": 3475
    },
    {
      "epoch": 0.14033104561970125,
      "grad_norm": 366.6150817871094,
      "learning_rate": 1.936868694611519e-05,
      "loss": 5.9418,
      "step": 3476
    },
    {
      "epoch": 0.140371417036738,
      "grad_norm": 78.47014617919922,
      "learning_rate": 1.9368229629543192e-05,
      "loss": 4.9177,
      "step": 3477
    },
    {
      "epoch": 0.14041178845377472,
      "grad_norm": 120.26966857910156,
      "learning_rate": 1.9367772152796346e-05,
      "loss": 5.3587,
      "step": 3478
    },
    {
      "epoch": 0.14045215987081147,
      "grad_norm": 92.56312561035156,
      "learning_rate": 1.9367314515882467e-05,
      "loss": 4.3132,
      "step": 3479
    },
    {
      "epoch": 0.1404925312878482,
      "grad_norm": 155.17434692382812,
      "learning_rate": 1.9366856718809386e-05,
      "loss": 3.9448,
      "step": 3480
    },
    {
      "epoch": 0.14053290270488494,
      "grad_norm": 102.18831634521484,
      "learning_rate": 1.9366398761584927e-05,
      "loss": 6.0245,
      "step": 3481
    },
    {
      "epoch": 0.1405732741219217,
      "grad_norm": 94.62721252441406,
      "learning_rate": 1.9365940644216925e-05,
      "loss": 4.7138,
      "step": 3482
    },
    {
      "epoch": 0.14061364553895842,
      "grad_norm": 135.030029296875,
      "learning_rate": 1.9365482366713208e-05,
      "loss": 5.6126,
      "step": 3483
    },
    {
      "epoch": 0.14065401695599516,
      "grad_norm": 97.6307144165039,
      "learning_rate": 1.9365023929081608e-05,
      "loss": 6.5592,
      "step": 3484
    },
    {
      "epoch": 0.1406943883730319,
      "grad_norm": 152.89675903320312,
      "learning_rate": 1.9364565331329973e-05,
      "loss": 5.5239,
      "step": 3485
    },
    {
      "epoch": 0.14073475979006864,
      "grad_norm": 119.86103057861328,
      "learning_rate": 1.9364106573466133e-05,
      "loss": 3.8081,
      "step": 3486
    },
    {
      "epoch": 0.14077513120710536,
      "grad_norm": 100.88007354736328,
      "learning_rate": 1.936364765549794e-05,
      "loss": 7.0173,
      "step": 3487
    },
    {
      "epoch": 0.1408155026241421,
      "grad_norm": 97.09559631347656,
      "learning_rate": 1.936318857743324e-05,
      "loss": 5.8488,
      "step": 3488
    },
    {
      "epoch": 0.14085587404117886,
      "grad_norm": 123.0761947631836,
      "learning_rate": 1.9362729339279877e-05,
      "loss": 5.3016,
      "step": 3489
    },
    {
      "epoch": 0.14089624545821558,
      "grad_norm": 131.9690399169922,
      "learning_rate": 1.9362269941045708e-05,
      "loss": 5.1574,
      "step": 3490
    },
    {
      "epoch": 0.14093661687525233,
      "grad_norm": 94.21881866455078,
      "learning_rate": 1.9361810382738585e-05,
      "loss": 6.7442,
      "step": 3491
    },
    {
      "epoch": 0.14097698829228905,
      "grad_norm": 122.9939193725586,
      "learning_rate": 1.9361350664366366e-05,
      "loss": 6.9363,
      "step": 3492
    },
    {
      "epoch": 0.1410173597093258,
      "grad_norm": 161.874755859375,
      "learning_rate": 1.9360890785936914e-05,
      "loss": 4.656,
      "step": 3493
    },
    {
      "epoch": 0.14105773112636252,
      "grad_norm": 95.06200408935547,
      "learning_rate": 1.9360430747458082e-05,
      "loss": 6.7382,
      "step": 3494
    },
    {
      "epoch": 0.14109810254339927,
      "grad_norm": 178.8581085205078,
      "learning_rate": 1.9359970548937746e-05,
      "loss": 5.164,
      "step": 3495
    },
    {
      "epoch": 0.14113847396043602,
      "grad_norm": 138.6744842529297,
      "learning_rate": 1.935951019038377e-05,
      "loss": 6.306,
      "step": 3496
    },
    {
      "epoch": 0.14117884537747274,
      "grad_norm": 106.2161865234375,
      "learning_rate": 1.9359049671804026e-05,
      "loss": 6.1815,
      "step": 3497
    },
    {
      "epoch": 0.1412192167945095,
      "grad_norm": 99.68490600585938,
      "learning_rate": 1.9358588993206387e-05,
      "loss": 4.7051,
      "step": 3498
    },
    {
      "epoch": 0.14125958821154622,
      "grad_norm": 119.52898406982422,
      "learning_rate": 1.9358128154598732e-05,
      "loss": 4.7585,
      "step": 3499
    },
    {
      "epoch": 0.14129995962858297,
      "grad_norm": 127.5406494140625,
      "learning_rate": 1.9357667155988936e-05,
      "loss": 7.4328,
      "step": 3500
    },
    {
      "epoch": 0.1413403310456197,
      "grad_norm": 125.97950744628906,
      "learning_rate": 1.9357205997384885e-05,
      "loss": 7.4854,
      "step": 3501
    },
    {
      "epoch": 0.14138070246265644,
      "grad_norm": 217.1858673095703,
      "learning_rate": 1.935674467879446e-05,
      "loss": 4.8531,
      "step": 3502
    },
    {
      "epoch": 0.1414210738796932,
      "grad_norm": 160.41448974609375,
      "learning_rate": 1.9356283200225558e-05,
      "loss": 6.7179,
      "step": 3503
    },
    {
      "epoch": 0.1414614452967299,
      "grad_norm": 97.78870391845703,
      "learning_rate": 1.9355821561686053e-05,
      "loss": 5.7072,
      "step": 3504
    },
    {
      "epoch": 0.14150181671376666,
      "grad_norm": 132.96397399902344,
      "learning_rate": 1.935535976318385e-05,
      "loss": 5.3878,
      "step": 3505
    },
    {
      "epoch": 0.14154218813080338,
      "grad_norm": 149.9427947998047,
      "learning_rate": 1.935489780472684e-05,
      "loss": 5.7942,
      "step": 3506
    },
    {
      "epoch": 0.14158255954784013,
      "grad_norm": 125.00550842285156,
      "learning_rate": 1.9354435686322924e-05,
      "loss": 5.9725,
      "step": 3507
    },
    {
      "epoch": 0.14162293096487688,
      "grad_norm": 82.47331237792969,
      "learning_rate": 1.9353973407980003e-05,
      "loss": 4.6103,
      "step": 3508
    },
    {
      "epoch": 0.1416633023819136,
      "grad_norm": 91.37753295898438,
      "learning_rate": 1.935351096970598e-05,
      "loss": 6.1846,
      "step": 3509
    },
    {
      "epoch": 0.14170367379895035,
      "grad_norm": 156.91920471191406,
      "learning_rate": 1.935304837150876e-05,
      "loss": 5.3801,
      "step": 3510
    },
    {
      "epoch": 0.14174404521598707,
      "grad_norm": 127.04059600830078,
      "learning_rate": 1.9352585613396253e-05,
      "loss": 6.3607,
      "step": 3511
    },
    {
      "epoch": 0.14178441663302382,
      "grad_norm": 823.8863525390625,
      "learning_rate": 1.935212269537637e-05,
      "loss": 6.3101,
      "step": 3512
    },
    {
      "epoch": 0.14182478805006055,
      "grad_norm": 155.34628295898438,
      "learning_rate": 1.935165961745703e-05,
      "loss": 5.2789,
      "step": 3513
    },
    {
      "epoch": 0.1418651594670973,
      "grad_norm": 250.83714294433594,
      "learning_rate": 1.935119637964615e-05,
      "loss": 5.9178,
      "step": 3514
    },
    {
      "epoch": 0.14190553088413405,
      "grad_norm": 90.9480972290039,
      "learning_rate": 1.9350732981951647e-05,
      "loss": 5.2529,
      "step": 3515
    },
    {
      "epoch": 0.14194590230117077,
      "grad_norm": 71.97303009033203,
      "learning_rate": 1.935026942438145e-05,
      "loss": 6.2269,
      "step": 3516
    },
    {
      "epoch": 0.14198627371820752,
      "grad_norm": 131.41671752929688,
      "learning_rate": 1.9349805706943475e-05,
      "loss": 5.8507,
      "step": 3517
    },
    {
      "epoch": 0.14202664513524424,
      "grad_norm": 123.97774505615234,
      "learning_rate": 1.9349341829645655e-05,
      "loss": 4.158,
      "step": 3518
    },
    {
      "epoch": 0.142067016552281,
      "grad_norm": 112.79246520996094,
      "learning_rate": 1.9348877792495925e-05,
      "loss": 5.7832,
      "step": 3519
    },
    {
      "epoch": 0.1421073879693177,
      "grad_norm": 216.2462921142578,
      "learning_rate": 1.9348413595502212e-05,
      "loss": 4.5783,
      "step": 3520
    },
    {
      "epoch": 0.14214775938635446,
      "grad_norm": 124.58135986328125,
      "learning_rate": 1.934794923867246e-05,
      "loss": 4.1927,
      "step": 3521
    },
    {
      "epoch": 0.1421881308033912,
      "grad_norm": 87.14722442626953,
      "learning_rate": 1.9347484722014605e-05,
      "loss": 4.7256,
      "step": 3522
    },
    {
      "epoch": 0.14222850222042793,
      "grad_norm": 129.69345092773438,
      "learning_rate": 1.9347020045536587e-05,
      "loss": 6.5426,
      "step": 3523
    },
    {
      "epoch": 0.14226887363746468,
      "grad_norm": 84.46060180664062,
      "learning_rate": 1.9346555209246356e-05,
      "loss": 5.7375,
      "step": 3524
    },
    {
      "epoch": 0.1423092450545014,
      "grad_norm": 125.60585021972656,
      "learning_rate": 1.9346090213151857e-05,
      "loss": 5.184,
      "step": 3525
    },
    {
      "epoch": 0.14234961647153815,
      "grad_norm": 145.08677673339844,
      "learning_rate": 1.9345625057261038e-05,
      "loss": 3.392,
      "step": 3526
    },
    {
      "epoch": 0.14238998788857488,
      "grad_norm": 81.25349426269531,
      "learning_rate": 1.9345159741581855e-05,
      "loss": 4.5947,
      "step": 3527
    },
    {
      "epoch": 0.14243035930561163,
      "grad_norm": 73.85813903808594,
      "learning_rate": 1.9344694266122263e-05,
      "loss": 5.2127,
      "step": 3528
    },
    {
      "epoch": 0.14247073072264838,
      "grad_norm": 130.00575256347656,
      "learning_rate": 1.9344228630890217e-05,
      "loss": 5.6363,
      "step": 3529
    },
    {
      "epoch": 0.1425111021396851,
      "grad_norm": 120.9537353515625,
      "learning_rate": 1.9343762835893683e-05,
      "loss": 5.5284,
      "step": 3530
    },
    {
      "epoch": 0.14255147355672185,
      "grad_norm": 103.31277465820312,
      "learning_rate": 1.9343296881140627e-05,
      "loss": 5.117,
      "step": 3531
    },
    {
      "epoch": 0.14259184497375857,
      "grad_norm": 102.68961334228516,
      "learning_rate": 1.9342830766639013e-05,
      "loss": 5.6086,
      "step": 3532
    },
    {
      "epoch": 0.14263221639079532,
      "grad_norm": 118.49593353271484,
      "learning_rate": 1.9342364492396808e-05,
      "loss": 7.0909,
      "step": 3533
    },
    {
      "epoch": 0.14267258780783207,
      "grad_norm": 107.71573638916016,
      "learning_rate": 1.9341898058421987e-05,
      "loss": 6.4013,
      "step": 3534
    },
    {
      "epoch": 0.1427129592248688,
      "grad_norm": 176.8319549560547,
      "learning_rate": 1.934143146472252e-05,
      "loss": 6.0523,
      "step": 3535
    },
    {
      "epoch": 0.14275333064190554,
      "grad_norm": 95.93356323242188,
      "learning_rate": 1.9340964711306395e-05,
      "loss": 6.0941,
      "step": 3536
    },
    {
      "epoch": 0.14279370205894226,
      "grad_norm": 238.49517822265625,
      "learning_rate": 1.934049779818158e-05,
      "loss": 5.785,
      "step": 3537
    },
    {
      "epoch": 0.142834073475979,
      "grad_norm": 111.50775146484375,
      "learning_rate": 1.9340030725356068e-05,
      "loss": 6.091,
      "step": 3538
    },
    {
      "epoch": 0.14287444489301573,
      "grad_norm": 98.77796936035156,
      "learning_rate": 1.9339563492837842e-05,
      "loss": 5.8218,
      "step": 3539
    },
    {
      "epoch": 0.14291481631005248,
      "grad_norm": 118.71866607666016,
      "learning_rate": 1.933909610063489e-05,
      "loss": 6.0254,
      "step": 3540
    },
    {
      "epoch": 0.14295518772708923,
      "grad_norm": 84.07167053222656,
      "learning_rate": 1.93386285487552e-05,
      "loss": 7.6336,
      "step": 3541
    },
    {
      "epoch": 0.14299555914412596,
      "grad_norm": 143.7473907470703,
      "learning_rate": 1.933816083720677e-05,
      "loss": 5.9178,
      "step": 3542
    },
    {
      "epoch": 0.1430359305611627,
      "grad_norm": 196.19093322753906,
      "learning_rate": 1.9337692965997597e-05,
      "loss": 5.9608,
      "step": 3543
    },
    {
      "epoch": 0.14307630197819943,
      "grad_norm": 171.09109497070312,
      "learning_rate": 1.9337224935135677e-05,
      "loss": 5.4392,
      "step": 3544
    },
    {
      "epoch": 0.14311667339523618,
      "grad_norm": 78.10480499267578,
      "learning_rate": 1.9336756744629016e-05,
      "loss": 5.8525,
      "step": 3545
    },
    {
      "epoch": 0.1431570448122729,
      "grad_norm": 112.6042709350586,
      "learning_rate": 1.933628839448562e-05,
      "loss": 7.4488,
      "step": 3546
    },
    {
      "epoch": 0.14319741622930965,
      "grad_norm": 223.6512908935547,
      "learning_rate": 1.933581988471349e-05,
      "loss": 6.2438,
      "step": 3547
    },
    {
      "epoch": 0.1432377876463464,
      "grad_norm": 157.65109252929688,
      "learning_rate": 1.9335351215320644e-05,
      "loss": 6.9975,
      "step": 3548
    },
    {
      "epoch": 0.14327815906338312,
      "grad_norm": 87.92195892333984,
      "learning_rate": 1.933488238631509e-05,
      "loss": 6.5831,
      "step": 3549
    },
    {
      "epoch": 0.14331853048041987,
      "grad_norm": 81.61279296875,
      "learning_rate": 1.9334413397704847e-05,
      "loss": 5.7663,
      "step": 3550
    },
    {
      "epoch": 0.1433589018974566,
      "grad_norm": 424.72833251953125,
      "learning_rate": 1.933394424949793e-05,
      "loss": 3.5799,
      "step": 3551
    },
    {
      "epoch": 0.14339927331449334,
      "grad_norm": 86.70234680175781,
      "learning_rate": 1.9333474941702367e-05,
      "loss": 6.0296,
      "step": 3552
    },
    {
      "epoch": 0.14343964473153006,
      "grad_norm": 188.577880859375,
      "learning_rate": 1.9333005474326175e-05,
      "loss": 4.3097,
      "step": 3553
    },
    {
      "epoch": 0.1434800161485668,
      "grad_norm": 148.4987335205078,
      "learning_rate": 1.9332535847377382e-05,
      "loss": 4.6588,
      "step": 3554
    },
    {
      "epoch": 0.14352038756560356,
      "grad_norm": 107.51017761230469,
      "learning_rate": 1.933206606086402e-05,
      "loss": 5.8206,
      "step": 3555
    },
    {
      "epoch": 0.14356075898264029,
      "grad_norm": 97.16667938232422,
      "learning_rate": 1.9331596114794122e-05,
      "loss": 6.5005,
      "step": 3556
    },
    {
      "epoch": 0.14360113039967703,
      "grad_norm": 171.72683715820312,
      "learning_rate": 1.933112600917572e-05,
      "loss": 4.8977,
      "step": 3557
    },
    {
      "epoch": 0.14364150181671376,
      "grad_norm": 137.19363403320312,
      "learning_rate": 1.9330655744016852e-05,
      "loss": 6.2419,
      "step": 3558
    },
    {
      "epoch": 0.1436818732337505,
      "grad_norm": 224.09214782714844,
      "learning_rate": 1.9330185319325562e-05,
      "loss": 5.7997,
      "step": 3559
    },
    {
      "epoch": 0.14372224465078726,
      "grad_norm": 77.97261047363281,
      "learning_rate": 1.932971473510989e-05,
      "loss": 3.7523,
      "step": 3560
    },
    {
      "epoch": 0.14376261606782398,
      "grad_norm": 95.38113403320312,
      "learning_rate": 1.932924399137788e-05,
      "loss": 5.3518,
      "step": 3561
    },
    {
      "epoch": 0.14380298748486073,
      "grad_norm": 85.13938903808594,
      "learning_rate": 1.9328773088137586e-05,
      "loss": 5.8082,
      "step": 3562
    },
    {
      "epoch": 0.14384335890189745,
      "grad_norm": 226.56588745117188,
      "learning_rate": 1.9328302025397054e-05,
      "loss": 5.6897,
      "step": 3563
    },
    {
      "epoch": 0.1438837303189342,
      "grad_norm": 79.9389419555664,
      "learning_rate": 1.9327830803164342e-05,
      "loss": 7.4341,
      "step": 3564
    },
    {
      "epoch": 0.14392410173597092,
      "grad_norm": 90.983642578125,
      "learning_rate": 1.9327359421447506e-05,
      "loss": 6.555,
      "step": 3565
    },
    {
      "epoch": 0.14396447315300767,
      "grad_norm": 114.46060180664062,
      "learning_rate": 1.9326887880254603e-05,
      "loss": 5.824,
      "step": 3566
    },
    {
      "epoch": 0.14400484457004442,
      "grad_norm": 140.4686279296875,
      "learning_rate": 1.9326416179593698e-05,
      "loss": 4.9636,
      "step": 3567
    },
    {
      "epoch": 0.14404521598708114,
      "grad_norm": 85.91769409179688,
      "learning_rate": 1.9325944319472856e-05,
      "loss": 6.1986,
      "step": 3568
    },
    {
      "epoch": 0.1440855874041179,
      "grad_norm": 109.5744400024414,
      "learning_rate": 1.9325472299900143e-05,
      "loss": 4.1003,
      "step": 3569
    },
    {
      "epoch": 0.14412595882115461,
      "grad_norm": 94.48709106445312,
      "learning_rate": 1.932500012088363e-05,
      "loss": 4.3342,
      "step": 3570
    },
    {
      "epoch": 0.14416633023819136,
      "grad_norm": 93.00914764404297,
      "learning_rate": 1.932452778243139e-05,
      "loss": 3.7278,
      "step": 3571
    },
    {
      "epoch": 0.1442067016552281,
      "grad_norm": 115.4797134399414,
      "learning_rate": 1.9324055284551497e-05,
      "loss": 5.3823,
      "step": 3572
    },
    {
      "epoch": 0.14424707307226484,
      "grad_norm": 154.8107452392578,
      "learning_rate": 1.9323582627252038e-05,
      "loss": 3.7174,
      "step": 3573
    },
    {
      "epoch": 0.1442874444893016,
      "grad_norm": 89.90785217285156,
      "learning_rate": 1.9323109810541083e-05,
      "loss": 4.7014,
      "step": 3574
    },
    {
      "epoch": 0.1443278159063383,
      "grad_norm": 113.37715911865234,
      "learning_rate": 1.932263683442672e-05,
      "loss": 5.0965,
      "step": 3575
    },
    {
      "epoch": 0.14436818732337506,
      "grad_norm": 104.2426986694336,
      "learning_rate": 1.9322163698917044e-05,
      "loss": 5.8957,
      "step": 3576
    },
    {
      "epoch": 0.14440855874041178,
      "grad_norm": 98.7147216796875,
      "learning_rate": 1.932169040402013e-05,
      "loss": 6.6614,
      "step": 3577
    },
    {
      "epoch": 0.14444893015744853,
      "grad_norm": 93.59056091308594,
      "learning_rate": 1.9321216949744082e-05,
      "loss": 6.3577,
      "step": 3578
    },
    {
      "epoch": 0.14448930157448525,
      "grad_norm": 108.4383316040039,
      "learning_rate": 1.9320743336096992e-05,
      "loss": 5.5439,
      "step": 3579
    },
    {
      "epoch": 0.144529672991522,
      "grad_norm": 155.3631134033203,
      "learning_rate": 1.9320269563086955e-05,
      "loss": 6.0584,
      "step": 3580
    },
    {
      "epoch": 0.14457004440855875,
      "grad_norm": 80.2646255493164,
      "learning_rate": 1.9319795630722075e-05,
      "loss": 4.4323,
      "step": 3581
    },
    {
      "epoch": 0.14461041582559547,
      "grad_norm": 80.0129165649414,
      "learning_rate": 1.9319321539010447e-05,
      "loss": 5.9557,
      "step": 3582
    },
    {
      "epoch": 0.14465078724263222,
      "grad_norm": 114.2190170288086,
      "learning_rate": 1.9318847287960188e-05,
      "loss": 3.9278,
      "step": 3583
    },
    {
      "epoch": 0.14469115865966894,
      "grad_norm": 260.38067626953125,
      "learning_rate": 1.93183728775794e-05,
      "loss": 5.9845,
      "step": 3584
    },
    {
      "epoch": 0.1447315300767057,
      "grad_norm": 292.52850341796875,
      "learning_rate": 1.9317898307876196e-05,
      "loss": 5.3321,
      "step": 3585
    },
    {
      "epoch": 0.14477190149374244,
      "grad_norm": 112.51070404052734,
      "learning_rate": 1.931742357885869e-05,
      "loss": 3.7735,
      "step": 3586
    },
    {
      "epoch": 0.14481227291077917,
      "grad_norm": 100.77621459960938,
      "learning_rate": 1.9316948690534997e-05,
      "loss": 4.9664,
      "step": 3587
    },
    {
      "epoch": 0.14485264432781592,
      "grad_norm": 141.72817993164062,
      "learning_rate": 1.9316473642913237e-05,
      "loss": 5.525,
      "step": 3588
    },
    {
      "epoch": 0.14489301574485264,
      "grad_norm": 93.79303741455078,
      "learning_rate": 1.9315998436001537e-05,
      "loss": 5.8141,
      "step": 3589
    },
    {
      "epoch": 0.1449333871618894,
      "grad_norm": 146.48472595214844,
      "learning_rate": 1.9315523069808015e-05,
      "loss": 5.1938,
      "step": 3590
    },
    {
      "epoch": 0.1449737585789261,
      "grad_norm": 91.95085144042969,
      "learning_rate": 1.9315047544340803e-05,
      "loss": 4.403,
      "step": 3591
    },
    {
      "epoch": 0.14501412999596286,
      "grad_norm": 86.68586730957031,
      "learning_rate": 1.9314571859608033e-05,
      "loss": 6.5274,
      "step": 3592
    },
    {
      "epoch": 0.1450545014129996,
      "grad_norm": 72.5445785522461,
      "learning_rate": 1.931409601561783e-05,
      "loss": 5.2942,
      "step": 3593
    },
    {
      "epoch": 0.14509487283003633,
      "grad_norm": 104.90951538085938,
      "learning_rate": 1.931362001237834e-05,
      "loss": 6.09,
      "step": 3594
    },
    {
      "epoch": 0.14513524424707308,
      "grad_norm": 98.39884948730469,
      "learning_rate": 1.9313143849897695e-05,
      "loss": 5.464,
      "step": 3595
    },
    {
      "epoch": 0.1451756156641098,
      "grad_norm": 203.84300231933594,
      "learning_rate": 1.9312667528184035e-05,
      "loss": 5.2489,
      "step": 3596
    },
    {
      "epoch": 0.14521598708114655,
      "grad_norm": 115.5306396484375,
      "learning_rate": 1.931219104724551e-05,
      "loss": 5.6259,
      "step": 3597
    },
    {
      "epoch": 0.14525635849818327,
      "grad_norm": 86.88900756835938,
      "learning_rate": 1.9311714407090264e-05,
      "loss": 4.5644,
      "step": 3598
    },
    {
      "epoch": 0.14529672991522002,
      "grad_norm": 101.34805297851562,
      "learning_rate": 1.931123760772644e-05,
      "loss": 6.3142,
      "step": 3599
    },
    {
      "epoch": 0.14533710133225677,
      "grad_norm": 94.7823715209961,
      "learning_rate": 1.9310760649162207e-05,
      "loss": 6.8922,
      "step": 3600
    },
    {
      "epoch": 0.1453774727492935,
      "grad_norm": 90.02969360351562,
      "learning_rate": 1.93102835314057e-05,
      "loss": 7.4287,
      "step": 3601
    },
    {
      "epoch": 0.14541784416633025,
      "grad_norm": 92.24016571044922,
      "learning_rate": 1.9309806254465088e-05,
      "loss": 4.4378,
      "step": 3602
    },
    {
      "epoch": 0.14545821558336697,
      "grad_norm": 110.00205993652344,
      "learning_rate": 1.930932881834853e-05,
      "loss": 6.2209,
      "step": 3603
    },
    {
      "epoch": 0.14549858700040372,
      "grad_norm": 125.7967300415039,
      "learning_rate": 1.930885122306419e-05,
      "loss": 5.6283,
      "step": 3604
    },
    {
      "epoch": 0.14553895841744044,
      "grad_norm": 104.49606323242188,
      "learning_rate": 1.930837346862023e-05,
      "loss": 6.0005,
      "step": 3605
    },
    {
      "epoch": 0.1455793298344772,
      "grad_norm": 133.75888061523438,
      "learning_rate": 1.930789555502482e-05,
      "loss": 5.8906,
      "step": 3606
    },
    {
      "epoch": 0.14561970125151394,
      "grad_norm": 112.8327865600586,
      "learning_rate": 1.930741748228613e-05,
      "loss": 5.3415,
      "step": 3607
    },
    {
      "epoch": 0.14566007266855066,
      "grad_norm": 128.73715209960938,
      "learning_rate": 1.9306939250412338e-05,
      "loss": 5.8899,
      "step": 3608
    },
    {
      "epoch": 0.1457004440855874,
      "grad_norm": 102.1942367553711,
      "learning_rate": 1.930646085941162e-05,
      "loss": 4.7118,
      "step": 3609
    },
    {
      "epoch": 0.14574081550262413,
      "grad_norm": 150.71868896484375,
      "learning_rate": 1.9305982309292148e-05,
      "loss": 5.2409,
      "step": 3610
    },
    {
      "epoch": 0.14578118691966088,
      "grad_norm": 126.26303100585938,
      "learning_rate": 1.930550360006211e-05,
      "loss": 4.1972,
      "step": 3611
    },
    {
      "epoch": 0.1458215583366976,
      "grad_norm": 78.87185668945312,
      "learning_rate": 1.9305024731729694e-05,
      "loss": 5.4628,
      "step": 3612
    },
    {
      "epoch": 0.14586192975373435,
      "grad_norm": 110.90478515625,
      "learning_rate": 1.9304545704303082e-05,
      "loss": 5.9899,
      "step": 3613
    },
    {
      "epoch": 0.1459023011707711,
      "grad_norm": 368.8982238769531,
      "learning_rate": 1.9304066517790465e-05,
      "loss": 6.3893,
      "step": 3614
    },
    {
      "epoch": 0.14594267258780783,
      "grad_norm": 109.07760620117188,
      "learning_rate": 1.930358717220004e-05,
      "loss": 6.9268,
      "step": 3615
    },
    {
      "epoch": 0.14598304400484458,
      "grad_norm": 100.78929901123047,
      "learning_rate": 1.930310766754e-05,
      "loss": 5.4515,
      "step": 3616
    },
    {
      "epoch": 0.1460234154218813,
      "grad_norm": 113.44183349609375,
      "learning_rate": 1.930262800381854e-05,
      "loss": 6.1895,
      "step": 3617
    },
    {
      "epoch": 0.14606378683891805,
      "grad_norm": 109.19171142578125,
      "learning_rate": 1.9302148181043864e-05,
      "loss": 5.9195,
      "step": 3618
    },
    {
      "epoch": 0.1461041582559548,
      "grad_norm": 118.35230255126953,
      "learning_rate": 1.930166819922418e-05,
      "loss": 6.4175,
      "step": 3619
    },
    {
      "epoch": 0.14614452967299152,
      "grad_norm": 83.70162963867188,
      "learning_rate": 1.9301188058367686e-05,
      "loss": 4.7199,
      "step": 3620
    },
    {
      "epoch": 0.14618490109002827,
      "grad_norm": 126.99000549316406,
      "learning_rate": 1.93007077584826e-05,
      "loss": 5.1711,
      "step": 3621
    },
    {
      "epoch": 0.146225272507065,
      "grad_norm": 132.8632049560547,
      "learning_rate": 1.930022729957713e-05,
      "loss": 5.2762,
      "step": 3622
    },
    {
      "epoch": 0.14626564392410174,
      "grad_norm": 277.0420227050781,
      "learning_rate": 1.929974668165949e-05,
      "loss": 5.6411,
      "step": 3623
    },
    {
      "epoch": 0.14630601534113846,
      "grad_norm": 230.2034149169922,
      "learning_rate": 1.9299265904737898e-05,
      "loss": 5.3893,
      "step": 3624
    },
    {
      "epoch": 0.1463463867581752,
      "grad_norm": 239.4950714111328,
      "learning_rate": 1.9298784968820573e-05,
      "loss": 6.0285,
      "step": 3625
    },
    {
      "epoch": 0.14638675817521196,
      "grad_norm": 82.32695007324219,
      "learning_rate": 1.9298303873915743e-05,
      "loss": 4.8171,
      "step": 3626
    },
    {
      "epoch": 0.14642712959224868,
      "grad_norm": 132.1372528076172,
      "learning_rate": 1.9297822620031628e-05,
      "loss": 5.8493,
      "step": 3627
    },
    {
      "epoch": 0.14646750100928543,
      "grad_norm": 159.8011016845703,
      "learning_rate": 1.9297341207176457e-05,
      "loss": 5.2752,
      "step": 3628
    },
    {
      "epoch": 0.14650787242632216,
      "grad_norm": 101.35397338867188,
      "learning_rate": 1.929685963535846e-05,
      "loss": 6.1313,
      "step": 3629
    },
    {
      "epoch": 0.1465482438433589,
      "grad_norm": 127.68951416015625,
      "learning_rate": 1.9296377904585876e-05,
      "loss": 6.4284,
      "step": 3630
    },
    {
      "epoch": 0.14658861526039563,
      "grad_norm": 170.88858032226562,
      "learning_rate": 1.929589601486694e-05,
      "loss": 4.5543,
      "step": 3631
    },
    {
      "epoch": 0.14662898667743238,
      "grad_norm": 118.27503967285156,
      "learning_rate": 1.9295413966209887e-05,
      "loss": 5.851,
      "step": 3632
    },
    {
      "epoch": 0.14666935809446913,
      "grad_norm": 174.62969970703125,
      "learning_rate": 1.9294931758622964e-05,
      "loss": 6.7702,
      "step": 3633
    },
    {
      "epoch": 0.14670972951150585,
      "grad_norm": 78.96407318115234,
      "learning_rate": 1.9294449392114408e-05,
      "loss": 6.241,
      "step": 3634
    },
    {
      "epoch": 0.1467501009285426,
      "grad_norm": 138.1212615966797,
      "learning_rate": 1.9293966866692476e-05,
      "loss": 4.6455,
      "step": 3635
    },
    {
      "epoch": 0.14679047234557932,
      "grad_norm": 86.24057006835938,
      "learning_rate": 1.9293484182365415e-05,
      "loss": 5.4466,
      "step": 3636
    },
    {
      "epoch": 0.14683084376261607,
      "grad_norm": 140.61509704589844,
      "learning_rate": 1.929300133914147e-05,
      "loss": 6.3395,
      "step": 3637
    },
    {
      "epoch": 0.1468712151796528,
      "grad_norm": 121.98310089111328,
      "learning_rate": 1.929251833702891e-05,
      "loss": 5.6721,
      "step": 3638
    },
    {
      "epoch": 0.14691158659668954,
      "grad_norm": 87.39095306396484,
      "learning_rate": 1.929203517603598e-05,
      "loss": 5.6948,
      "step": 3639
    },
    {
      "epoch": 0.1469519580137263,
      "grad_norm": 157.3022918701172,
      "learning_rate": 1.9291551856170955e-05,
      "loss": 4.6619,
      "step": 3640
    },
    {
      "epoch": 0.146992329430763,
      "grad_norm": 87.52153015136719,
      "learning_rate": 1.9291068377442084e-05,
      "loss": 4.5763,
      "step": 3641
    },
    {
      "epoch": 0.14703270084779976,
      "grad_norm": 86.02887725830078,
      "learning_rate": 1.9290584739857642e-05,
      "loss": 5.7647,
      "step": 3642
    },
    {
      "epoch": 0.14707307226483649,
      "grad_norm": 86.03722381591797,
      "learning_rate": 1.9290100943425893e-05,
      "loss": 6.6965,
      "step": 3643
    },
    {
      "epoch": 0.14711344368187323,
      "grad_norm": 94.45664978027344,
      "learning_rate": 1.9289616988155114e-05,
      "loss": 5.2545,
      "step": 3644
    },
    {
      "epoch": 0.14715381509890998,
      "grad_norm": 84.04740142822266,
      "learning_rate": 1.9289132874053575e-05,
      "loss": 7.5814,
      "step": 3645
    },
    {
      "epoch": 0.1471941865159467,
      "grad_norm": 91.56331634521484,
      "learning_rate": 1.928864860112956e-05,
      "loss": 6.8265,
      "step": 3646
    },
    {
      "epoch": 0.14723455793298346,
      "grad_norm": 82.84741973876953,
      "learning_rate": 1.9288164169391344e-05,
      "loss": 5.8728,
      "step": 3647
    },
    {
      "epoch": 0.14727492935002018,
      "grad_norm": 169.87759399414062,
      "learning_rate": 1.9287679578847206e-05,
      "loss": 4.9576,
      "step": 3648
    },
    {
      "epoch": 0.14731530076705693,
      "grad_norm": 107.775390625,
      "learning_rate": 1.9287194829505438e-05,
      "loss": 5.9079,
      "step": 3649
    },
    {
      "epoch": 0.14735567218409365,
      "grad_norm": 100.05482482910156,
      "learning_rate": 1.9286709921374326e-05,
      "loss": 5.6946,
      "step": 3650
    },
    {
      "epoch": 0.1473960436011304,
      "grad_norm": 101.66407775878906,
      "learning_rate": 1.9286224854462162e-05,
      "loss": 6.401,
      "step": 3651
    },
    {
      "epoch": 0.14743641501816715,
      "grad_norm": 95.38565063476562,
      "learning_rate": 1.9285739628777233e-05,
      "loss": 4.584,
      "step": 3652
    },
    {
      "epoch": 0.14747678643520387,
      "grad_norm": 106.8643569946289,
      "learning_rate": 1.9285254244327844e-05,
      "loss": 5.8503,
      "step": 3653
    },
    {
      "epoch": 0.14751715785224062,
      "grad_norm": 110.3509292602539,
      "learning_rate": 1.928476870112229e-05,
      "loss": 5.6935,
      "step": 3654
    },
    {
      "epoch": 0.14755752926927734,
      "grad_norm": 72.33702087402344,
      "learning_rate": 1.9284282999168872e-05,
      "loss": 6.1787,
      "step": 3655
    },
    {
      "epoch": 0.1475979006863141,
      "grad_norm": 100.62427520751953,
      "learning_rate": 1.9283797138475892e-05,
      "loss": 4.8652,
      "step": 3656
    },
    {
      "epoch": 0.14763827210335081,
      "grad_norm": 124.80619812011719,
      "learning_rate": 1.9283311119051665e-05,
      "loss": 3.5669,
      "step": 3657
    },
    {
      "epoch": 0.14767864352038756,
      "grad_norm": 109.9373550415039,
      "learning_rate": 1.9282824940904493e-05,
      "loss": 6.4184,
      "step": 3658
    },
    {
      "epoch": 0.14771901493742431,
      "grad_norm": 107.4551010131836,
      "learning_rate": 1.9282338604042695e-05,
      "loss": 6.4444,
      "step": 3659
    },
    {
      "epoch": 0.14775938635446104,
      "grad_norm": 98.18954467773438,
      "learning_rate": 1.9281852108474578e-05,
      "loss": 5.5189,
      "step": 3660
    },
    {
      "epoch": 0.1477997577714978,
      "grad_norm": 123.82550048828125,
      "learning_rate": 1.9281365454208468e-05,
      "loss": 7.9856,
      "step": 3661
    },
    {
      "epoch": 0.1478401291885345,
      "grad_norm": 116.35289001464844,
      "learning_rate": 1.928087864125268e-05,
      "loss": 5.5533,
      "step": 3662
    },
    {
      "epoch": 0.14788050060557126,
      "grad_norm": 105.60775756835938,
      "learning_rate": 1.9280391669615538e-05,
      "loss": 4.3919,
      "step": 3663
    },
    {
      "epoch": 0.14792087202260798,
      "grad_norm": 104.371337890625,
      "learning_rate": 1.9279904539305375e-05,
      "loss": 4.9034,
      "step": 3664
    },
    {
      "epoch": 0.14796124343964473,
      "grad_norm": 89.67601013183594,
      "learning_rate": 1.927941725033051e-05,
      "loss": 6.6822,
      "step": 3665
    },
    {
      "epoch": 0.14800161485668148,
      "grad_norm": 77.57770538330078,
      "learning_rate": 1.927892980269928e-05,
      "loss": 5.3078,
      "step": 3666
    },
    {
      "epoch": 0.1480419862737182,
      "grad_norm": 1262.39599609375,
      "learning_rate": 1.927844219642002e-05,
      "loss": 5.5793,
      "step": 3667
    },
    {
      "epoch": 0.14808235769075495,
      "grad_norm": 79.45745849609375,
      "learning_rate": 1.9277954431501065e-05,
      "loss": 6.2982,
      "step": 3668
    },
    {
      "epoch": 0.14812272910779167,
      "grad_norm": 210.9683380126953,
      "learning_rate": 1.9277466507950754e-05,
      "loss": 4.9917,
      "step": 3669
    },
    {
      "epoch": 0.14816310052482842,
      "grad_norm": 96.81055450439453,
      "learning_rate": 1.927697842577743e-05,
      "loss": 7.2746,
      "step": 3670
    },
    {
      "epoch": 0.14820347194186517,
      "grad_norm": 101.4755859375,
      "learning_rate": 1.9276490184989438e-05,
      "loss": 4.5547,
      "step": 3671
    },
    {
      "epoch": 0.1482438433589019,
      "grad_norm": 114.89277648925781,
      "learning_rate": 1.9276001785595126e-05,
      "loss": 5.1964,
      "step": 3672
    },
    {
      "epoch": 0.14828421477593864,
      "grad_norm": 138.10777282714844,
      "learning_rate": 1.9275513227602846e-05,
      "loss": 6.7894,
      "step": 3673
    },
    {
      "epoch": 0.14832458619297537,
      "grad_norm": 102.05095672607422,
      "learning_rate": 1.927502451102095e-05,
      "loss": 7.2602,
      "step": 3674
    },
    {
      "epoch": 0.14836495761001212,
      "grad_norm": 189.55531311035156,
      "learning_rate": 1.927453563585779e-05,
      "loss": 5.6902,
      "step": 3675
    },
    {
      "epoch": 0.14840532902704884,
      "grad_norm": 121.90968322753906,
      "learning_rate": 1.927404660212173e-05,
      "loss": 5.8166,
      "step": 3676
    },
    {
      "epoch": 0.1484457004440856,
      "grad_norm": 108.98853302001953,
      "learning_rate": 1.927355740982113e-05,
      "loss": 5.1709,
      "step": 3677
    },
    {
      "epoch": 0.14848607186112234,
      "grad_norm": 91.56439971923828,
      "learning_rate": 1.9273068058964355e-05,
      "loss": 5.2805,
      "step": 3678
    },
    {
      "epoch": 0.14852644327815906,
      "grad_norm": 144.70236206054688,
      "learning_rate": 1.927257854955977e-05,
      "loss": 5.7455,
      "step": 3679
    },
    {
      "epoch": 0.1485668146951958,
      "grad_norm": 98.75641632080078,
      "learning_rate": 1.9272088881615742e-05,
      "loss": 5.4915,
      "step": 3680
    },
    {
      "epoch": 0.14860718611223253,
      "grad_norm": 92.40211486816406,
      "learning_rate": 1.927159905514065e-05,
      "loss": 6.1852,
      "step": 3681
    },
    {
      "epoch": 0.14864755752926928,
      "grad_norm": 104.24855041503906,
      "learning_rate": 1.927110907014286e-05,
      "loss": 6.7234,
      "step": 3682
    },
    {
      "epoch": 0.148687928946306,
      "grad_norm": 154.59536743164062,
      "learning_rate": 1.927061892663076e-05,
      "loss": 6.0282,
      "step": 3683
    },
    {
      "epoch": 0.14872830036334275,
      "grad_norm": 98.92107391357422,
      "learning_rate": 1.9270128624612717e-05,
      "loss": 6.3162,
      "step": 3684
    },
    {
      "epoch": 0.1487686717803795,
      "grad_norm": 111.96880340576172,
      "learning_rate": 1.926963816409713e-05,
      "loss": 5.1473,
      "step": 3685
    },
    {
      "epoch": 0.14880904319741622,
      "grad_norm": 117.66322326660156,
      "learning_rate": 1.9269147545092373e-05,
      "loss": 4.8951,
      "step": 3686
    },
    {
      "epoch": 0.14884941461445297,
      "grad_norm": 70.74616241455078,
      "learning_rate": 1.926865676760684e-05,
      "loss": 4.3459,
      "step": 3687
    },
    {
      "epoch": 0.1488897860314897,
      "grad_norm": 112.54205322265625,
      "learning_rate": 1.926816583164892e-05,
      "loss": 5.6428,
      "step": 3688
    },
    {
      "epoch": 0.14893015744852645,
      "grad_norm": 97.68804931640625,
      "learning_rate": 1.9267674737227002e-05,
      "loss": 5.5514,
      "step": 3689
    },
    {
      "epoch": 0.14897052886556317,
      "grad_norm": 90.52777099609375,
      "learning_rate": 1.9267183484349493e-05,
      "loss": 5.6062,
      "step": 3690
    },
    {
      "epoch": 0.14901090028259992,
      "grad_norm": 172.7816925048828,
      "learning_rate": 1.9266692073024785e-05,
      "loss": 5.8605,
      "step": 3691
    },
    {
      "epoch": 0.14905127169963667,
      "grad_norm": 111.94236755371094,
      "learning_rate": 1.926620050326128e-05,
      "loss": 6.5909,
      "step": 3692
    },
    {
      "epoch": 0.1490916431166734,
      "grad_norm": 110.49031066894531,
      "learning_rate": 1.9265708775067388e-05,
      "loss": 5.5533,
      "step": 3693
    },
    {
      "epoch": 0.14913201453371014,
      "grad_norm": 129.4652099609375,
      "learning_rate": 1.926521688845151e-05,
      "loss": 5.9881,
      "step": 3694
    },
    {
      "epoch": 0.14917238595074686,
      "grad_norm": 73.16157531738281,
      "learning_rate": 1.926472484342206e-05,
      "loss": 6.8215,
      "step": 3695
    },
    {
      "epoch": 0.1492127573677836,
      "grad_norm": 142.85174560546875,
      "learning_rate": 1.926423263998745e-05,
      "loss": 5.3166,
      "step": 3696
    },
    {
      "epoch": 0.14925312878482036,
      "grad_norm": 105.38622283935547,
      "learning_rate": 1.9263740278156095e-05,
      "loss": 5.3809,
      "step": 3697
    },
    {
      "epoch": 0.14929350020185708,
      "grad_norm": 81.71334075927734,
      "learning_rate": 1.9263247757936415e-05,
      "loss": 4.7848,
      "step": 3698
    },
    {
      "epoch": 0.14933387161889383,
      "grad_norm": 84.7408676147461,
      "learning_rate": 1.9262755079336828e-05,
      "loss": 3.9607,
      "step": 3699
    },
    {
      "epoch": 0.14937424303593055,
      "grad_norm": 97.11955261230469,
      "learning_rate": 1.926226224236576e-05,
      "loss": 6.2519,
      "step": 3700
    },
    {
      "epoch": 0.1494146144529673,
      "grad_norm": 270.22174072265625,
      "learning_rate": 1.9261769247031637e-05,
      "loss": 4.1121,
      "step": 3701
    },
    {
      "epoch": 0.14945498587000403,
      "grad_norm": 145.2192840576172,
      "learning_rate": 1.9261276093342888e-05,
      "loss": 4.011,
      "step": 3702
    },
    {
      "epoch": 0.14949535728704078,
      "grad_norm": 124.568115234375,
      "learning_rate": 1.9260782781307942e-05,
      "loss": 5.9783,
      "step": 3703
    },
    {
      "epoch": 0.14953572870407753,
      "grad_norm": 143.80374145507812,
      "learning_rate": 1.926028931093524e-05,
      "loss": 4.7034,
      "step": 3704
    },
    {
      "epoch": 0.14957610012111425,
      "grad_norm": 96.74362182617188,
      "learning_rate": 1.925979568223321e-05,
      "loss": 6.1873,
      "step": 3705
    },
    {
      "epoch": 0.149616471538151,
      "grad_norm": 100.5030288696289,
      "learning_rate": 1.92593018952103e-05,
      "loss": 5.66,
      "step": 3706
    },
    {
      "epoch": 0.14965684295518772,
      "grad_norm": 141.68875122070312,
      "learning_rate": 1.925880794987495e-05,
      "loss": 5.6959,
      "step": 3707
    },
    {
      "epoch": 0.14969721437222447,
      "grad_norm": 86.6017074584961,
      "learning_rate": 1.9258313846235602e-05,
      "loss": 5.4392,
      "step": 3708
    },
    {
      "epoch": 0.1497375857892612,
      "grad_norm": 136.210693359375,
      "learning_rate": 1.925781958430071e-05,
      "loss": 6.2325,
      "step": 3709
    },
    {
      "epoch": 0.14977795720629794,
      "grad_norm": 138.23419189453125,
      "learning_rate": 1.9257325164078723e-05,
      "loss": 6.5412,
      "step": 3710
    },
    {
      "epoch": 0.1498183286233347,
      "grad_norm": 146.1085662841797,
      "learning_rate": 1.9256830585578092e-05,
      "loss": 5.5179,
      "step": 3711
    },
    {
      "epoch": 0.1498587000403714,
      "grad_norm": 150.65057373046875,
      "learning_rate": 1.925633584880727e-05,
      "loss": 4.5317,
      "step": 3712
    },
    {
      "epoch": 0.14989907145740816,
      "grad_norm": 87.79476928710938,
      "learning_rate": 1.9255840953774724e-05,
      "loss": 6.2142,
      "step": 3713
    },
    {
      "epoch": 0.14993944287444488,
      "grad_norm": 165.9092559814453,
      "learning_rate": 1.925534590048891e-05,
      "loss": 6.3154,
      "step": 3714
    },
    {
      "epoch": 0.14997981429148163,
      "grad_norm": 108.52133178710938,
      "learning_rate": 1.9254850688958294e-05,
      "loss": 5.0092,
      "step": 3715
    },
    {
      "epoch": 0.15002018570851836,
      "grad_norm": 150.67459106445312,
      "learning_rate": 1.9254355319191346e-05,
      "loss": 5.695,
      "step": 3716
    },
    {
      "epoch": 0.1500605571255551,
      "grad_norm": 143.84242248535156,
      "learning_rate": 1.925385979119653e-05,
      "loss": 5.8326,
      "step": 3717
    },
    {
      "epoch": 0.15010092854259185,
      "grad_norm": 206.73550415039062,
      "learning_rate": 1.9253364104982323e-05,
      "loss": 5.7848,
      "step": 3718
    },
    {
      "epoch": 0.15014129995962858,
      "grad_norm": 483.6385498046875,
      "learning_rate": 1.9252868260557193e-05,
      "loss": 2.6412,
      "step": 3719
    },
    {
      "epoch": 0.15018167137666533,
      "grad_norm": 101.43060302734375,
      "learning_rate": 1.925237225792963e-05,
      "loss": 5.6491,
      "step": 3720
    },
    {
      "epoch": 0.15022204279370205,
      "grad_norm": 109.68919372558594,
      "learning_rate": 1.92518760971081e-05,
      "loss": 6.4198,
      "step": 3721
    },
    {
      "epoch": 0.1502624142107388,
      "grad_norm": 103.93772888183594,
      "learning_rate": 1.9251379778101097e-05,
      "loss": 5.1935,
      "step": 3722
    },
    {
      "epoch": 0.15030278562777555,
      "grad_norm": 77.80084991455078,
      "learning_rate": 1.9250883300917104e-05,
      "loss": 6.0401,
      "step": 3723
    },
    {
      "epoch": 0.15034315704481227,
      "grad_norm": 178.96383666992188,
      "learning_rate": 1.9250386665564608e-05,
      "loss": 5.0654,
      "step": 3724
    },
    {
      "epoch": 0.15038352846184902,
      "grad_norm": 80.65153503417969,
      "learning_rate": 1.9249889872052098e-05,
      "loss": 5.6649,
      "step": 3725
    },
    {
      "epoch": 0.15042389987888574,
      "grad_norm": 69.43978881835938,
      "learning_rate": 1.9249392920388075e-05,
      "loss": 5.9882,
      "step": 3726
    },
    {
      "epoch": 0.1504642712959225,
      "grad_norm": 100.36404418945312,
      "learning_rate": 1.924889581058103e-05,
      "loss": 5.3094,
      "step": 3727
    },
    {
      "epoch": 0.1505046427129592,
      "grad_norm": 85.74613189697266,
      "learning_rate": 1.9248398542639465e-05,
      "loss": 6.0829,
      "step": 3728
    },
    {
      "epoch": 0.15054501412999596,
      "grad_norm": 68.46339416503906,
      "learning_rate": 1.9247901116571885e-05,
      "loss": 6.6913,
      "step": 3729
    },
    {
      "epoch": 0.1505853855470327,
      "grad_norm": 62.677616119384766,
      "learning_rate": 1.9247403532386787e-05,
      "loss": 6.1877,
      "step": 3730
    },
    {
      "epoch": 0.15062575696406943,
      "grad_norm": 115.47235107421875,
      "learning_rate": 1.9246905790092683e-05,
      "loss": 6.0418,
      "step": 3731
    },
    {
      "epoch": 0.15066612838110618,
      "grad_norm": 90.11369323730469,
      "learning_rate": 1.9246407889698083e-05,
      "loss": 5.7824,
      "step": 3732
    },
    {
      "epoch": 0.1507064997981429,
      "grad_norm": 151.76832580566406,
      "learning_rate": 1.9245909831211503e-05,
      "loss": 3.7034,
      "step": 3733
    },
    {
      "epoch": 0.15074687121517966,
      "grad_norm": 135.9447784423828,
      "learning_rate": 1.9245411614641453e-05,
      "loss": 4.9412,
      "step": 3734
    },
    {
      "epoch": 0.15078724263221638,
      "grad_norm": 142.3673095703125,
      "learning_rate": 1.9244913239996453e-05,
      "loss": 5.3683,
      "step": 3735
    },
    {
      "epoch": 0.15082761404925313,
      "grad_norm": 118.97039031982422,
      "learning_rate": 1.924441470728503e-05,
      "loss": 5.6516,
      "step": 3736
    },
    {
      "epoch": 0.15086798546628988,
      "grad_norm": 117.20911407470703,
      "learning_rate": 1.9243916016515697e-05,
      "loss": 5.4981,
      "step": 3737
    },
    {
      "epoch": 0.1509083568833266,
      "grad_norm": 170.0854034423828,
      "learning_rate": 1.9243417167696988e-05,
      "loss": 6.6401,
      "step": 3738
    },
    {
      "epoch": 0.15094872830036335,
      "grad_norm": 105.8757095336914,
      "learning_rate": 1.9242918160837433e-05,
      "loss": 4.6142,
      "step": 3739
    },
    {
      "epoch": 0.15098909971740007,
      "grad_norm": 10854.400390625,
      "learning_rate": 1.9242418995945557e-05,
      "loss": 6.3919,
      "step": 3740
    },
    {
      "epoch": 0.15102947113443682,
      "grad_norm": 148.17578125,
      "learning_rate": 1.9241919673029905e-05,
      "loss": 6.0757,
      "step": 3741
    },
    {
      "epoch": 0.15106984255147354,
      "grad_norm": 147.6841583251953,
      "learning_rate": 1.9241420192099002e-05,
      "loss": 5.2071,
      "step": 3742
    },
    {
      "epoch": 0.1511102139685103,
      "grad_norm": 111.83880615234375,
      "learning_rate": 1.9240920553161397e-05,
      "loss": 5.9862,
      "step": 3743
    },
    {
      "epoch": 0.15115058538554704,
      "grad_norm": 85.80765533447266,
      "learning_rate": 1.9240420756225626e-05,
      "loss": 4.2715,
      "step": 3744
    },
    {
      "epoch": 0.15119095680258376,
      "grad_norm": 182.4120635986328,
      "learning_rate": 1.9239920801300242e-05,
      "loss": 4.9738,
      "step": 3745
    },
    {
      "epoch": 0.15123132821962051,
      "grad_norm": 63.498046875,
      "learning_rate": 1.923942068839379e-05,
      "loss": 6.8359,
      "step": 3746
    },
    {
      "epoch": 0.15127169963665724,
      "grad_norm": 118.58452606201172,
      "learning_rate": 1.9238920417514816e-05,
      "loss": 5.883,
      "step": 3747
    },
    {
      "epoch": 0.151312071053694,
      "grad_norm": 108.62731170654297,
      "learning_rate": 1.923841998867188e-05,
      "loss": 3.8612,
      "step": 3748
    },
    {
      "epoch": 0.15135244247073074,
      "grad_norm": 154.09796142578125,
      "learning_rate": 1.9237919401873535e-05,
      "loss": 4.6299,
      "step": 3749
    },
    {
      "epoch": 0.15139281388776746,
      "grad_norm": 123.81061553955078,
      "learning_rate": 1.9237418657128336e-05,
      "loss": 5.9077,
      "step": 3750
    },
    {
      "epoch": 0.1514331853048042,
      "grad_norm": 269.90985107421875,
      "learning_rate": 1.9236917754444855e-05,
      "loss": 5.5692,
      "step": 3751
    },
    {
      "epoch": 0.15147355672184093,
      "grad_norm": 87.42609405517578,
      "learning_rate": 1.9236416693831648e-05,
      "loss": 5.2977,
      "step": 3752
    },
    {
      "epoch": 0.15151392813887768,
      "grad_norm": 137.17813110351562,
      "learning_rate": 1.9235915475297286e-05,
      "loss": 6.5886,
      "step": 3753
    },
    {
      "epoch": 0.1515542995559144,
      "grad_norm": 131.7044677734375,
      "learning_rate": 1.923541409885033e-05,
      "loss": 7.0795,
      "step": 3754
    },
    {
      "epoch": 0.15159467097295115,
      "grad_norm": 139.23263549804688,
      "learning_rate": 1.9234912564499365e-05,
      "loss": 6.0211,
      "step": 3755
    },
    {
      "epoch": 0.1516350423899879,
      "grad_norm": 226.3516387939453,
      "learning_rate": 1.923441087225296e-05,
      "loss": 4.7919,
      "step": 3756
    },
    {
      "epoch": 0.15167541380702462,
      "grad_norm": 177.46734619140625,
      "learning_rate": 1.9233909022119693e-05,
      "loss": 6.6227,
      "step": 3757
    },
    {
      "epoch": 0.15171578522406137,
      "grad_norm": 157.786376953125,
      "learning_rate": 1.923340701410814e-05,
      "loss": 5.9612,
      "step": 3758
    },
    {
      "epoch": 0.1517561566410981,
      "grad_norm": 98.35379028320312,
      "learning_rate": 1.9232904848226895e-05,
      "loss": 5.7497,
      "step": 3759
    },
    {
      "epoch": 0.15179652805813484,
      "grad_norm": 138.4340362548828,
      "learning_rate": 1.9232402524484533e-05,
      "loss": 4.8604,
      "step": 3760
    },
    {
      "epoch": 0.15183689947517157,
      "grad_norm": 114.63153076171875,
      "learning_rate": 1.923190004288965e-05,
      "loss": 5.0329,
      "step": 3761
    },
    {
      "epoch": 0.15187727089220832,
      "grad_norm": 87.94915771484375,
      "learning_rate": 1.923139740345083e-05,
      "loss": 4.9879,
      "step": 3762
    },
    {
      "epoch": 0.15191764230924507,
      "grad_norm": 140.82801818847656,
      "learning_rate": 1.923089460617668e-05,
      "loss": 5.7113,
      "step": 3763
    },
    {
      "epoch": 0.1519580137262818,
      "grad_norm": 92.23088073730469,
      "learning_rate": 1.9230391651075778e-05,
      "loss": 6.4777,
      "step": 3764
    },
    {
      "epoch": 0.15199838514331854,
      "grad_norm": 122.27082061767578,
      "learning_rate": 1.922988853815674e-05,
      "loss": 4.7166,
      "step": 3765
    },
    {
      "epoch": 0.15203875656035526,
      "grad_norm": 120.27598571777344,
      "learning_rate": 1.9229385267428154e-05,
      "loss": 5.3288,
      "step": 3766
    },
    {
      "epoch": 0.152079127977392,
      "grad_norm": 104.30023193359375,
      "learning_rate": 1.922888183889864e-05,
      "loss": 4.9039,
      "step": 3767
    },
    {
      "epoch": 0.15211949939442873,
      "grad_norm": 115.22828674316406,
      "learning_rate": 1.9228378252576792e-05,
      "loss": 4.4528,
      "step": 3768
    },
    {
      "epoch": 0.15215987081146548,
      "grad_norm": 153.5281219482422,
      "learning_rate": 1.922787450847123e-05,
      "loss": 5.0452,
      "step": 3769
    },
    {
      "epoch": 0.15220024222850223,
      "grad_norm": 77.29727935791016,
      "learning_rate": 1.922737060659056e-05,
      "loss": 6.1528,
      "step": 3770
    },
    {
      "epoch": 0.15224061364553895,
      "grad_norm": 135.4048614501953,
      "learning_rate": 1.9226866546943398e-05,
      "loss": 7.729,
      "step": 3771
    },
    {
      "epoch": 0.1522809850625757,
      "grad_norm": 128.64239501953125,
      "learning_rate": 1.922636232953837e-05,
      "loss": 6.8452,
      "step": 3772
    },
    {
      "epoch": 0.15232135647961242,
      "grad_norm": 129.7314910888672,
      "learning_rate": 1.9225857954384088e-05,
      "loss": 5.8136,
      "step": 3773
    },
    {
      "epoch": 0.15236172789664917,
      "grad_norm": 96.27642059326172,
      "learning_rate": 1.922535342148918e-05,
      "loss": 5.0984,
      "step": 3774
    },
    {
      "epoch": 0.15240209931368592,
      "grad_norm": 202.80789184570312,
      "learning_rate": 1.9224848730862267e-05,
      "loss": 6.5672,
      "step": 3775
    },
    {
      "epoch": 0.15244247073072265,
      "grad_norm": 118.69708251953125,
      "learning_rate": 1.922434388251199e-05,
      "loss": 5.8977,
      "step": 3776
    },
    {
      "epoch": 0.1524828421477594,
      "grad_norm": 140.6011962890625,
      "learning_rate": 1.9223838876446964e-05,
      "loss": 4.0179,
      "step": 3777
    },
    {
      "epoch": 0.15252321356479612,
      "grad_norm": 155.58407592773438,
      "learning_rate": 1.922333371267584e-05,
      "loss": 4.8544,
      "step": 3778
    },
    {
      "epoch": 0.15256358498183287,
      "grad_norm": 122.5542221069336,
      "learning_rate": 1.9222828391207244e-05,
      "loss": 4.4551,
      "step": 3779
    },
    {
      "epoch": 0.1526039563988696,
      "grad_norm": 107.63692474365234,
      "learning_rate": 1.922232291204982e-05,
      "loss": 6.2912,
      "step": 3780
    },
    {
      "epoch": 0.15264432781590634,
      "grad_norm": 96.39556884765625,
      "learning_rate": 1.922181727521221e-05,
      "loss": 4.3602,
      "step": 3781
    },
    {
      "epoch": 0.1526846992329431,
      "grad_norm": 141.50509643554688,
      "learning_rate": 1.9221311480703058e-05,
      "loss": 4.6797,
      "step": 3782
    },
    {
      "epoch": 0.1527250706499798,
      "grad_norm": 104.08492279052734,
      "learning_rate": 1.9220805528531013e-05,
      "loss": 5.6863,
      "step": 3783
    },
    {
      "epoch": 0.15276544206701656,
      "grad_norm": 98.154541015625,
      "learning_rate": 1.922029941870473e-05,
      "loss": 4.7637,
      "step": 3784
    },
    {
      "epoch": 0.15280581348405328,
      "grad_norm": 108.73468780517578,
      "learning_rate": 1.9219793151232853e-05,
      "loss": 5.735,
      "step": 3785
    },
    {
      "epoch": 0.15284618490109003,
      "grad_norm": 161.84490966796875,
      "learning_rate": 1.9219286726124045e-05,
      "loss": 5.5908,
      "step": 3786
    },
    {
      "epoch": 0.15288655631812675,
      "grad_norm": 87.8739242553711,
      "learning_rate": 1.9218780143386962e-05,
      "loss": 5.2857,
      "step": 3787
    },
    {
      "epoch": 0.1529269277351635,
      "grad_norm": 109.95288848876953,
      "learning_rate": 1.9218273403030265e-05,
      "loss": 5.9346,
      "step": 3788
    },
    {
      "epoch": 0.15296729915220025,
      "grad_norm": 278.6561584472656,
      "learning_rate": 1.9217766505062617e-05,
      "loss": 5.0095,
      "step": 3789
    },
    {
      "epoch": 0.15300767056923698,
      "grad_norm": 87.41838836669922,
      "learning_rate": 1.921725944949269e-05,
      "loss": 6.2738,
      "step": 3790
    },
    {
      "epoch": 0.15304804198627373,
      "grad_norm": 104.70514678955078,
      "learning_rate": 1.9216752236329152e-05,
      "loss": 7.6218,
      "step": 3791
    },
    {
      "epoch": 0.15308841340331045,
      "grad_norm": 159.63497924804688,
      "learning_rate": 1.9216244865580673e-05,
      "loss": 6.5278,
      "step": 3792
    },
    {
      "epoch": 0.1531287848203472,
      "grad_norm": 107.8621597290039,
      "learning_rate": 1.9215737337255923e-05,
      "loss": 5.999,
      "step": 3793
    },
    {
      "epoch": 0.15316915623738392,
      "grad_norm": 123.8042984008789,
      "learning_rate": 1.921522965136359e-05,
      "loss": 4.9842,
      "step": 3794
    },
    {
      "epoch": 0.15320952765442067,
      "grad_norm": 73.23761749267578,
      "learning_rate": 1.9214721807912346e-05,
      "loss": 3.0591,
      "step": 3795
    },
    {
      "epoch": 0.15324989907145742,
      "grad_norm": 107.66911315917969,
      "learning_rate": 1.9214213806910882e-05,
      "loss": 6.3446,
      "step": 3796
    },
    {
      "epoch": 0.15329027048849414,
      "grad_norm": 79.80415344238281,
      "learning_rate": 1.9213705648367874e-05,
      "loss": 5.686,
      "step": 3797
    },
    {
      "epoch": 0.1533306419055309,
      "grad_norm": 96.27152252197266,
      "learning_rate": 1.9213197332292014e-05,
      "loss": 4.3635,
      "step": 3798
    },
    {
      "epoch": 0.1533710133225676,
      "grad_norm": 135.2897491455078,
      "learning_rate": 1.9212688858692e-05,
      "loss": 5.6637,
      "step": 3799
    },
    {
      "epoch": 0.15341138473960436,
      "grad_norm": 87.25790405273438,
      "learning_rate": 1.9212180227576513e-05,
      "loss": 6.3889,
      "step": 3800
    },
    {
      "epoch": 0.1534517561566411,
      "grad_norm": 92.07347106933594,
      "learning_rate": 1.921167143895426e-05,
      "loss": 5.3751,
      "step": 3801
    },
    {
      "epoch": 0.15349212757367783,
      "grad_norm": 145.76377868652344,
      "learning_rate": 1.9211162492833936e-05,
      "loss": 5.7303,
      "step": 3802
    },
    {
      "epoch": 0.15353249899071458,
      "grad_norm": 117.75994873046875,
      "learning_rate": 1.921065338922424e-05,
      "loss": 5.8076,
      "step": 3803
    },
    {
      "epoch": 0.1535728704077513,
      "grad_norm": 111.5917739868164,
      "learning_rate": 1.921014412813388e-05,
      "loss": 5.2596,
      "step": 3804
    },
    {
      "epoch": 0.15361324182478805,
      "grad_norm": 111.27794647216797,
      "learning_rate": 1.9209634709571565e-05,
      "loss": 4.1246,
      "step": 3805
    },
    {
      "epoch": 0.15365361324182478,
      "grad_norm": 88.95201110839844,
      "learning_rate": 1.9209125133546003e-05,
      "loss": 4.7957,
      "step": 3806
    },
    {
      "epoch": 0.15369398465886153,
      "grad_norm": 157.19810485839844,
      "learning_rate": 1.9208615400065908e-05,
      "loss": 5.829,
      "step": 3807
    },
    {
      "epoch": 0.15373435607589828,
      "grad_norm": 114.81649017333984,
      "learning_rate": 1.920810550913999e-05,
      "loss": 5.4688,
      "step": 3808
    },
    {
      "epoch": 0.153774727492935,
      "grad_norm": 88.26560974121094,
      "learning_rate": 1.920759546077697e-05,
      "loss": 5.1328,
      "step": 3809
    },
    {
      "epoch": 0.15381509890997175,
      "grad_norm": 161.12063598632812,
      "learning_rate": 1.9207085254985567e-05,
      "loss": 4.6499,
      "step": 3810
    },
    {
      "epoch": 0.15385547032700847,
      "grad_norm": 100.03279876708984,
      "learning_rate": 1.9206574891774507e-05,
      "loss": 6.779,
      "step": 3811
    },
    {
      "epoch": 0.15389584174404522,
      "grad_norm": 125.41998291015625,
      "learning_rate": 1.9206064371152518e-05,
      "loss": 5.1133,
      "step": 3812
    },
    {
      "epoch": 0.15393621316108194,
      "grad_norm": 123.18319702148438,
      "learning_rate": 1.9205553693128324e-05,
      "loss": 4.4742,
      "step": 3813
    },
    {
      "epoch": 0.1539765845781187,
      "grad_norm": 78.47528076171875,
      "learning_rate": 1.920504285771066e-05,
      "loss": 6.1327,
      "step": 3814
    },
    {
      "epoch": 0.15401695599515544,
      "grad_norm": 110.43328857421875,
      "learning_rate": 1.9204531864908257e-05,
      "loss": 4.3935,
      "step": 3815
    },
    {
      "epoch": 0.15405732741219216,
      "grad_norm": 130.26243591308594,
      "learning_rate": 1.9204020714729853e-05,
      "loss": 5.9843,
      "step": 3816
    },
    {
      "epoch": 0.1540976988292289,
      "grad_norm": 103.24874877929688,
      "learning_rate": 1.920350940718419e-05,
      "loss": 6.8862,
      "step": 3817
    },
    {
      "epoch": 0.15413807024626563,
      "grad_norm": 92.59767150878906,
      "learning_rate": 1.9202997942280004e-05,
      "loss": 3.5016,
      "step": 3818
    },
    {
      "epoch": 0.15417844166330238,
      "grad_norm": 96.37419128417969,
      "learning_rate": 1.9202486320026045e-05,
      "loss": 6.542,
      "step": 3819
    },
    {
      "epoch": 0.1542188130803391,
      "grad_norm": 185.49075317382812,
      "learning_rate": 1.920197454043106e-05,
      "loss": 4.3665,
      "step": 3820
    },
    {
      "epoch": 0.15425918449737586,
      "grad_norm": 144.15650939941406,
      "learning_rate": 1.92014626035038e-05,
      "loss": 5.5602,
      "step": 3821
    },
    {
      "epoch": 0.1542995559144126,
      "grad_norm": 100.53197479248047,
      "learning_rate": 1.9200950509253018e-05,
      "loss": 5.4586,
      "step": 3822
    },
    {
      "epoch": 0.15433992733144933,
      "grad_norm": 105.23015594482422,
      "learning_rate": 1.9200438257687464e-05,
      "loss": 5.8772,
      "step": 3823
    },
    {
      "epoch": 0.15438029874848608,
      "grad_norm": 112.77274322509766,
      "learning_rate": 1.91999258488159e-05,
      "loss": 6.798,
      "step": 3824
    },
    {
      "epoch": 0.1544206701655228,
      "grad_norm": 113.65153503417969,
      "learning_rate": 1.9199413282647092e-05,
      "loss": 4.6407,
      "step": 3825
    },
    {
      "epoch": 0.15446104158255955,
      "grad_norm": 285.8592224121094,
      "learning_rate": 1.9198900559189793e-05,
      "loss": 5.9595,
      "step": 3826
    },
    {
      "epoch": 0.1545014129995963,
      "grad_norm": 130.150146484375,
      "learning_rate": 1.919838767845278e-05,
      "loss": 4.3556,
      "step": 3827
    },
    {
      "epoch": 0.15454178441663302,
      "grad_norm": 106.91216278076172,
      "learning_rate": 1.9197874640444817e-05,
      "loss": 5.7897,
      "step": 3828
    },
    {
      "epoch": 0.15458215583366977,
      "grad_norm": 104.07172393798828,
      "learning_rate": 1.9197361445174673e-05,
      "loss": 3.9998,
      "step": 3829
    },
    {
      "epoch": 0.1546225272507065,
      "grad_norm": 211.45394897460938,
      "learning_rate": 1.919684809265113e-05,
      "loss": 5.0099,
      "step": 3830
    },
    {
      "epoch": 0.15466289866774324,
      "grad_norm": 100.65733337402344,
      "learning_rate": 1.919633458288296e-05,
      "loss": 6.3622,
      "step": 3831
    },
    {
      "epoch": 0.15470327008477996,
      "grad_norm": 97.56356811523438,
      "learning_rate": 1.919582091587894e-05,
      "loss": 5.9596,
      "step": 3832
    },
    {
      "epoch": 0.15474364150181671,
      "grad_norm": 92.90740203857422,
      "learning_rate": 1.919530709164786e-05,
      "loss": 5.4931,
      "step": 3833
    },
    {
      "epoch": 0.15478401291885346,
      "grad_norm": 129.1946563720703,
      "learning_rate": 1.91947931101985e-05,
      "loss": 4.703,
      "step": 3834
    },
    {
      "epoch": 0.15482438433589019,
      "grad_norm": 89.61626434326172,
      "learning_rate": 1.919427897153965e-05,
      "loss": 6.6362,
      "step": 3835
    },
    {
      "epoch": 0.15486475575292694,
      "grad_norm": 85.99122619628906,
      "learning_rate": 1.9193764675680097e-05,
      "loss": 6.4848,
      "step": 3836
    },
    {
      "epoch": 0.15490512716996366,
      "grad_norm": 95.73828887939453,
      "learning_rate": 1.919325022262864e-05,
      "loss": 4.1881,
      "step": 3837
    },
    {
      "epoch": 0.1549454985870004,
      "grad_norm": 92.7774658203125,
      "learning_rate": 1.919273561239407e-05,
      "loss": 6.1745,
      "step": 3838
    },
    {
      "epoch": 0.15498587000403713,
      "grad_norm": 84.06645202636719,
      "learning_rate": 1.9192220844985184e-05,
      "loss": 4.7018,
      "step": 3839
    },
    {
      "epoch": 0.15502624142107388,
      "grad_norm": 106.32046508789062,
      "learning_rate": 1.9191705920410793e-05,
      "loss": 5.4072,
      "step": 3840
    },
    {
      "epoch": 0.15506661283811063,
      "grad_norm": 136.3414764404297,
      "learning_rate": 1.9191190838679694e-05,
      "loss": 4.727,
      "step": 3841
    },
    {
      "epoch": 0.15510698425514735,
      "grad_norm": 102.09601593017578,
      "learning_rate": 1.9190675599800693e-05,
      "loss": 4.3206,
      "step": 3842
    },
    {
      "epoch": 0.1551473556721841,
      "grad_norm": 104.02631378173828,
      "learning_rate": 1.91901602037826e-05,
      "loss": 6.4174,
      "step": 3843
    },
    {
      "epoch": 0.15518772708922082,
      "grad_norm": 143.0424346923828,
      "learning_rate": 1.9189644650634234e-05,
      "loss": 6.4546,
      "step": 3844
    },
    {
      "epoch": 0.15522809850625757,
      "grad_norm": 423.7908020019531,
      "learning_rate": 1.91891289403644e-05,
      "loss": 4.8815,
      "step": 3845
    },
    {
      "epoch": 0.1552684699232943,
      "grad_norm": 146.44236755371094,
      "learning_rate": 1.9188613072981917e-05,
      "loss": 6.9226,
      "step": 3846
    },
    {
      "epoch": 0.15530884134033104,
      "grad_norm": 103.1520767211914,
      "learning_rate": 1.918809704849561e-05,
      "loss": 4.9769,
      "step": 3847
    },
    {
      "epoch": 0.1553492127573678,
      "grad_norm": 90.11070251464844,
      "learning_rate": 1.9187580866914296e-05,
      "loss": 4.6562,
      "step": 3848
    },
    {
      "epoch": 0.15538958417440452,
      "grad_norm": 102.50076293945312,
      "learning_rate": 1.9187064528246808e-05,
      "loss": 3.7752,
      "step": 3849
    },
    {
      "epoch": 0.15542995559144127,
      "grad_norm": 107.78345489501953,
      "learning_rate": 1.9186548032501967e-05,
      "loss": 4.4501,
      "step": 3850
    },
    {
      "epoch": 0.155470327008478,
      "grad_norm": 179.51930236816406,
      "learning_rate": 1.918603137968861e-05,
      "loss": 5.2928,
      "step": 3851
    },
    {
      "epoch": 0.15551069842551474,
      "grad_norm": 113.52947998046875,
      "learning_rate": 1.9185514569815567e-05,
      "loss": 5.4808,
      "step": 3852
    },
    {
      "epoch": 0.1555510698425515,
      "grad_norm": 144.39065551757812,
      "learning_rate": 1.918499760289167e-05,
      "loss": 5.3299,
      "step": 3853
    },
    {
      "epoch": 0.1555914412595882,
      "grad_norm": 104.77790832519531,
      "learning_rate": 1.9184480478925765e-05,
      "loss": 5.995,
      "step": 3854
    },
    {
      "epoch": 0.15563181267662496,
      "grad_norm": 232.04954528808594,
      "learning_rate": 1.9183963197926694e-05,
      "loss": 3.9691,
      "step": 3855
    },
    {
      "epoch": 0.15567218409366168,
      "grad_norm": 146.5750274658203,
      "learning_rate": 1.9183445759903296e-05,
      "loss": 5.7228,
      "step": 3856
    },
    {
      "epoch": 0.15571255551069843,
      "grad_norm": 89.9005126953125,
      "learning_rate": 1.918292816486442e-05,
      "loss": 5.0977,
      "step": 3857
    },
    {
      "epoch": 0.15575292692773515,
      "grad_norm": 81.31658172607422,
      "learning_rate": 1.9182410412818917e-05,
      "loss": 5.7811,
      "step": 3858
    },
    {
      "epoch": 0.1557932983447719,
      "grad_norm": 82.50636291503906,
      "learning_rate": 1.9181892503775635e-05,
      "loss": 6.9076,
      "step": 3859
    },
    {
      "epoch": 0.15583366976180865,
      "grad_norm": 320.40228271484375,
      "learning_rate": 1.918137443774344e-05,
      "loss": 5.6191,
      "step": 3860
    },
    {
      "epoch": 0.15587404117884537,
      "grad_norm": 166.17507934570312,
      "learning_rate": 1.9180856214731178e-05,
      "loss": 5.5467,
      "step": 3861
    },
    {
      "epoch": 0.15591441259588212,
      "grad_norm": 100.89933776855469,
      "learning_rate": 1.9180337834747708e-05,
      "loss": 5.3441,
      "step": 3862
    },
    {
      "epoch": 0.15595478401291885,
      "grad_norm": 130.15736389160156,
      "learning_rate": 1.9179819297801904e-05,
      "loss": 4.7539,
      "step": 3863
    },
    {
      "epoch": 0.1559951554299556,
      "grad_norm": 106.48886108398438,
      "learning_rate": 1.9179300603902627e-05,
      "loss": 5.044,
      "step": 3864
    },
    {
      "epoch": 0.15603552684699232,
      "grad_norm": 219.07896423339844,
      "learning_rate": 1.9178781753058743e-05,
      "loss": 5.1911,
      "step": 3865
    },
    {
      "epoch": 0.15607589826402907,
      "grad_norm": 94.42346954345703,
      "learning_rate": 1.917826274527913e-05,
      "loss": 5.7922,
      "step": 3866
    },
    {
      "epoch": 0.15611626968106582,
      "grad_norm": 126.49922943115234,
      "learning_rate": 1.917774358057265e-05,
      "loss": 3.8203,
      "step": 3867
    },
    {
      "epoch": 0.15615664109810254,
      "grad_norm": 105.68155670166016,
      "learning_rate": 1.917722425894819e-05,
      "loss": 4.9529,
      "step": 3868
    },
    {
      "epoch": 0.1561970125151393,
      "grad_norm": 100.96878051757812,
      "learning_rate": 1.9176704780414622e-05,
      "loss": 5.9561,
      "step": 3869
    },
    {
      "epoch": 0.156237383932176,
      "grad_norm": 90.3969497680664,
      "learning_rate": 1.9176185144980836e-05,
      "loss": 6.6529,
      "step": 3870
    },
    {
      "epoch": 0.15627775534921276,
      "grad_norm": 79.8398666381836,
      "learning_rate": 1.917566535265571e-05,
      "loss": 4.483,
      "step": 3871
    },
    {
      "epoch": 0.15631812676624948,
      "grad_norm": 84.0641098022461,
      "learning_rate": 1.9175145403448134e-05,
      "loss": 5.2695,
      "step": 3872
    },
    {
      "epoch": 0.15635849818328623,
      "grad_norm": 166.74095153808594,
      "learning_rate": 1.9174625297366996e-05,
      "loss": 5.8682,
      "step": 3873
    },
    {
      "epoch": 0.15639886960032298,
      "grad_norm": 92.5826187133789,
      "learning_rate": 1.917410503442119e-05,
      "loss": 5.6523,
      "step": 3874
    },
    {
      "epoch": 0.1564392410173597,
      "grad_norm": 124.00626373291016,
      "learning_rate": 1.917358461461961e-05,
      "loss": 6.0841,
      "step": 3875
    },
    {
      "epoch": 0.15647961243439645,
      "grad_norm": 111.89415740966797,
      "learning_rate": 1.9173064037971158e-05,
      "loss": 5.0409,
      "step": 3876
    },
    {
      "epoch": 0.15651998385143318,
      "grad_norm": 186.23828125,
      "learning_rate": 1.917254330448473e-05,
      "loss": 3.9313,
      "step": 3877
    },
    {
      "epoch": 0.15656035526846993,
      "grad_norm": 123.09004974365234,
      "learning_rate": 1.917202241416923e-05,
      "loss": 5.8214,
      "step": 3878
    },
    {
      "epoch": 0.15660072668550667,
      "grad_norm": 103.0923080444336,
      "learning_rate": 1.9171501367033566e-05,
      "loss": 4.8682,
      "step": 3879
    },
    {
      "epoch": 0.1566410981025434,
      "grad_norm": 101.62142944335938,
      "learning_rate": 1.9170980163086645e-05,
      "loss": 6.6434,
      "step": 3880
    },
    {
      "epoch": 0.15668146951958015,
      "grad_norm": 85.72885131835938,
      "learning_rate": 1.917045880233738e-05,
      "loss": 5.6331,
      "step": 3881
    },
    {
      "epoch": 0.15672184093661687,
      "grad_norm": 101.6050796508789,
      "learning_rate": 1.916993728479468e-05,
      "loss": 4.2127,
      "step": 3882
    },
    {
      "epoch": 0.15676221235365362,
      "grad_norm": 90.916748046875,
      "learning_rate": 1.916941561046747e-05,
      "loss": 6.4963,
      "step": 3883
    },
    {
      "epoch": 0.15680258377069034,
      "grad_norm": 66.53782653808594,
      "learning_rate": 1.9168893779364667e-05,
      "loss": 4.2458,
      "step": 3884
    },
    {
      "epoch": 0.1568429551877271,
      "grad_norm": 98.9454345703125,
      "learning_rate": 1.9168371791495187e-05,
      "loss": 5.5303,
      "step": 3885
    },
    {
      "epoch": 0.15688332660476384,
      "grad_norm": 172.99000549316406,
      "learning_rate": 1.916784964686796e-05,
      "loss": 5.0835,
      "step": 3886
    },
    {
      "epoch": 0.15692369802180056,
      "grad_norm": 79.3277359008789,
      "learning_rate": 1.916732734549191e-05,
      "loss": 5.6225,
      "step": 3887
    },
    {
      "epoch": 0.1569640694388373,
      "grad_norm": 124.99447631835938,
      "learning_rate": 1.9166804887375976e-05,
      "loss": 5.2979,
      "step": 3888
    },
    {
      "epoch": 0.15700444085587403,
      "grad_norm": 85.89258575439453,
      "learning_rate": 1.916628227252908e-05,
      "loss": 7.0683,
      "step": 3889
    },
    {
      "epoch": 0.15704481227291078,
      "grad_norm": 81.53611755371094,
      "learning_rate": 1.916575950096016e-05,
      "loss": 6.2826,
      "step": 3890
    },
    {
      "epoch": 0.1570851836899475,
      "grad_norm": 108.62158966064453,
      "learning_rate": 1.916523657267816e-05,
      "loss": 6.8064,
      "step": 3891
    },
    {
      "epoch": 0.15712555510698425,
      "grad_norm": 531.5919799804688,
      "learning_rate": 1.9164713487692018e-05,
      "loss": 5.3882,
      "step": 3892
    },
    {
      "epoch": 0.157165926524021,
      "grad_norm": 94.5216293334961,
      "learning_rate": 1.9164190246010675e-05,
      "loss": 6.0959,
      "step": 3893
    },
    {
      "epoch": 0.15720629794105773,
      "grad_norm": 69.82068634033203,
      "learning_rate": 1.9163666847643076e-05,
      "loss": 5.404,
      "step": 3894
    },
    {
      "epoch": 0.15724666935809448,
      "grad_norm": 134.36293029785156,
      "learning_rate": 1.9163143292598178e-05,
      "loss": 5.0401,
      "step": 3895
    },
    {
      "epoch": 0.1572870407751312,
      "grad_norm": 163.00587463378906,
      "learning_rate": 1.9162619580884923e-05,
      "loss": 4.7043,
      "step": 3896
    },
    {
      "epoch": 0.15732741219216795,
      "grad_norm": 109.36251831054688,
      "learning_rate": 1.916209571251227e-05,
      "loss": 6.1894,
      "step": 3897
    },
    {
      "epoch": 0.15736778360920467,
      "grad_norm": 84.64905548095703,
      "learning_rate": 1.916157168748918e-05,
      "loss": 6.0063,
      "step": 3898
    },
    {
      "epoch": 0.15740815502624142,
      "grad_norm": 108.7936782836914,
      "learning_rate": 1.91610475058246e-05,
      "loss": 5.7981,
      "step": 3899
    },
    {
      "epoch": 0.15744852644327817,
      "grad_norm": 98.02617645263672,
      "learning_rate": 1.9160523167527505e-05,
      "loss": 6.9376,
      "step": 3900
    },
    {
      "epoch": 0.1574888978603149,
      "grad_norm": 126.2669906616211,
      "learning_rate": 1.9159998672606855e-05,
      "loss": 3.1107,
      "step": 3901
    },
    {
      "epoch": 0.15752926927735164,
      "grad_norm": 116.79812622070312,
      "learning_rate": 1.9159474021071616e-05,
      "loss": 6.5994,
      "step": 3902
    },
    {
      "epoch": 0.15756964069438836,
      "grad_norm": 111.45769500732422,
      "learning_rate": 1.915894921293076e-05,
      "loss": 5.8057,
      "step": 3903
    },
    {
      "epoch": 0.1576100121114251,
      "grad_norm": 85.26962280273438,
      "learning_rate": 1.9158424248193263e-05,
      "loss": 6.0736,
      "step": 3904
    },
    {
      "epoch": 0.15765038352846186,
      "grad_norm": 146.6771697998047,
      "learning_rate": 1.9157899126868096e-05,
      "loss": 5.6944,
      "step": 3905
    },
    {
      "epoch": 0.15769075494549858,
      "grad_norm": 75.33729553222656,
      "learning_rate": 1.9157373848964237e-05,
      "loss": 4.6871,
      "step": 3906
    },
    {
      "epoch": 0.15773112636253533,
      "grad_norm": 171.7278594970703,
      "learning_rate": 1.915684841449067e-05,
      "loss": 6.8044,
      "step": 3907
    },
    {
      "epoch": 0.15777149777957206,
      "grad_norm": 182.3719024658203,
      "learning_rate": 1.915632282345638e-05,
      "loss": 5.5221,
      "step": 3908
    },
    {
      "epoch": 0.1578118691966088,
      "grad_norm": 173.2001190185547,
      "learning_rate": 1.9155797075870353e-05,
      "loss": 5.0898,
      "step": 3909
    },
    {
      "epoch": 0.15785224061364553,
      "grad_norm": 115.56017303466797,
      "learning_rate": 1.9155271171741575e-05,
      "loss": 6.0506,
      "step": 3910
    },
    {
      "epoch": 0.15789261203068228,
      "grad_norm": 99.82826232910156,
      "learning_rate": 1.9154745111079038e-05,
      "loss": 5.6836,
      "step": 3911
    },
    {
      "epoch": 0.15793298344771903,
      "grad_norm": 91.11571502685547,
      "learning_rate": 1.9154218893891738e-05,
      "loss": 6.2547,
      "step": 3912
    },
    {
      "epoch": 0.15797335486475575,
      "grad_norm": 102.98213958740234,
      "learning_rate": 1.915369252018867e-05,
      "loss": 6.1502,
      "step": 3913
    },
    {
      "epoch": 0.1580137262817925,
      "grad_norm": 105.15494537353516,
      "learning_rate": 1.9153165989978836e-05,
      "loss": 5.8476,
      "step": 3914
    },
    {
      "epoch": 0.15805409769882922,
      "grad_norm": 123.3259506225586,
      "learning_rate": 1.915263930327124e-05,
      "loss": 5.8095,
      "step": 3915
    },
    {
      "epoch": 0.15809446911586597,
      "grad_norm": 87.7287368774414,
      "learning_rate": 1.9152112460074882e-05,
      "loss": 5.7705,
      "step": 3916
    },
    {
      "epoch": 0.1581348405329027,
      "grad_norm": 148.3131103515625,
      "learning_rate": 1.9151585460398775e-05,
      "loss": 5.4359,
      "step": 3917
    },
    {
      "epoch": 0.15817521194993944,
      "grad_norm": 91.71047973632812,
      "learning_rate": 1.9151058304251923e-05,
      "loss": 6.4413,
      "step": 3918
    },
    {
      "epoch": 0.1582155833669762,
      "grad_norm": 137.56906127929688,
      "learning_rate": 1.9150530991643346e-05,
      "loss": 6.1869,
      "step": 3919
    },
    {
      "epoch": 0.15825595478401291,
      "grad_norm": 146.1921844482422,
      "learning_rate": 1.915000352258206e-05,
      "loss": 6.3253,
      "step": 3920
    },
    {
      "epoch": 0.15829632620104966,
      "grad_norm": 102.56863403320312,
      "learning_rate": 1.9149475897077075e-05,
      "loss": 6.0892,
      "step": 3921
    },
    {
      "epoch": 0.15833669761808639,
      "grad_norm": 132.3795928955078,
      "learning_rate": 1.914894811513742e-05,
      "loss": 4.5045,
      "step": 3922
    },
    {
      "epoch": 0.15837706903512314,
      "grad_norm": 126.80741119384766,
      "learning_rate": 1.914842017677212e-05,
      "loss": 5.2738,
      "step": 3923
    },
    {
      "epoch": 0.15841744045215986,
      "grad_norm": 100.40518951416016,
      "learning_rate": 1.9147892081990192e-05,
      "loss": 4.8895,
      "step": 3924
    },
    {
      "epoch": 0.1584578118691966,
      "grad_norm": 105.7684097290039,
      "learning_rate": 1.9147363830800673e-05,
      "loss": 4.0666,
      "step": 3925
    },
    {
      "epoch": 0.15849818328623336,
      "grad_norm": 388.7333679199219,
      "learning_rate": 1.91468354232126e-05,
      "loss": 4.9522,
      "step": 3926
    },
    {
      "epoch": 0.15853855470327008,
      "grad_norm": 95.99212646484375,
      "learning_rate": 1.9146306859234992e-05,
      "loss": 5.6164,
      "step": 3927
    },
    {
      "epoch": 0.15857892612030683,
      "grad_norm": 94.7569351196289,
      "learning_rate": 1.9145778138876898e-05,
      "loss": 5.8863,
      "step": 3928
    },
    {
      "epoch": 0.15861929753734355,
      "grad_norm": 100.23162078857422,
      "learning_rate": 1.9145249262147354e-05,
      "loss": 5.17,
      "step": 3929
    },
    {
      "epoch": 0.1586596689543803,
      "grad_norm": 101.74520874023438,
      "learning_rate": 1.9144720229055403e-05,
      "loss": 5.734,
      "step": 3930
    },
    {
      "epoch": 0.15870004037141705,
      "grad_norm": 104.2304916381836,
      "learning_rate": 1.914419103961009e-05,
      "loss": 5.7291,
      "step": 3931
    },
    {
      "epoch": 0.15874041178845377,
      "grad_norm": 106.92723083496094,
      "learning_rate": 1.9143661693820467e-05,
      "loss": 5.5143,
      "step": 3932
    },
    {
      "epoch": 0.15878078320549052,
      "grad_norm": 108.63618469238281,
      "learning_rate": 1.9143132191695578e-05,
      "loss": 5.8028,
      "step": 3933
    },
    {
      "epoch": 0.15882115462252724,
      "grad_norm": 130.43641662597656,
      "learning_rate": 1.914260253324448e-05,
      "loss": 6.7406,
      "step": 3934
    },
    {
      "epoch": 0.158861526039564,
      "grad_norm": 113.4793472290039,
      "learning_rate": 1.914207271847623e-05,
      "loss": 5.2397,
      "step": 3935
    },
    {
      "epoch": 0.15890189745660072,
      "grad_norm": 110.32916259765625,
      "learning_rate": 1.914154274739988e-05,
      "loss": 6.1643,
      "step": 3936
    },
    {
      "epoch": 0.15894226887363747,
      "grad_norm": 102.65760040283203,
      "learning_rate": 1.91410126200245e-05,
      "loss": 4.646,
      "step": 3937
    },
    {
      "epoch": 0.15898264029067422,
      "grad_norm": 154.8840789794922,
      "learning_rate": 1.9140482336359153e-05,
      "loss": 5.1984,
      "step": 3938
    },
    {
      "epoch": 0.15902301170771094,
      "grad_norm": 94.89183807373047,
      "learning_rate": 1.9139951896412897e-05,
      "loss": 5.2951,
      "step": 3939
    },
    {
      "epoch": 0.1590633831247477,
      "grad_norm": 111.39331817626953,
      "learning_rate": 1.9139421300194808e-05,
      "loss": 5.3678,
      "step": 3940
    },
    {
      "epoch": 0.1591037545417844,
      "grad_norm": 155.2172393798828,
      "learning_rate": 1.9138890547713957e-05,
      "loss": 5.4804,
      "step": 3941
    },
    {
      "epoch": 0.15914412595882116,
      "grad_norm": 137.45166015625,
      "learning_rate": 1.9138359638979423e-05,
      "loss": 6.4363,
      "step": 3942
    },
    {
      "epoch": 0.15918449737585788,
      "grad_norm": 122.55270385742188,
      "learning_rate": 1.9137828574000274e-05,
      "loss": 5.7756,
      "step": 3943
    },
    {
      "epoch": 0.15922486879289463,
      "grad_norm": 100.35802459716797,
      "learning_rate": 1.9137297352785596e-05,
      "loss": 5.6742,
      "step": 3944
    },
    {
      "epoch": 0.15926524020993138,
      "grad_norm": 85.63905334472656,
      "learning_rate": 1.913676597534447e-05,
      "loss": 6.197,
      "step": 3945
    },
    {
      "epoch": 0.1593056116269681,
      "grad_norm": 135.42236328125,
      "learning_rate": 1.913623444168598e-05,
      "loss": 5.7127,
      "step": 3946
    },
    {
      "epoch": 0.15934598304400485,
      "grad_norm": 99.94322967529297,
      "learning_rate": 1.913570275181922e-05,
      "loss": 5.7077,
      "step": 3947
    },
    {
      "epoch": 0.15938635446104157,
      "grad_norm": 100.1816635131836,
      "learning_rate": 1.913517090575327e-05,
      "loss": 5.8169,
      "step": 3948
    },
    {
      "epoch": 0.15942672587807832,
      "grad_norm": 84.45658111572266,
      "learning_rate": 1.9134638903497236e-05,
      "loss": 4.9217,
      "step": 3949
    },
    {
      "epoch": 0.15946709729511505,
      "grad_norm": 105.34477233886719,
      "learning_rate": 1.9134106745060203e-05,
      "loss": 4.2686,
      "step": 3950
    },
    {
      "epoch": 0.1595074687121518,
      "grad_norm": 165.08877563476562,
      "learning_rate": 1.913357443045128e-05,
      "loss": 5.1461,
      "step": 3951
    },
    {
      "epoch": 0.15954784012918855,
      "grad_norm": 125.07987976074219,
      "learning_rate": 1.9133041959679557e-05,
      "loss": 6.4693,
      "step": 3952
    },
    {
      "epoch": 0.15958821154622527,
      "grad_norm": 158.3524932861328,
      "learning_rate": 1.9132509332754147e-05,
      "loss": 5.7928,
      "step": 3953
    },
    {
      "epoch": 0.15962858296326202,
      "grad_norm": 123.15573120117188,
      "learning_rate": 1.913197654968415e-05,
      "loss": 5.0837,
      "step": 3954
    },
    {
      "epoch": 0.15966895438029874,
      "grad_norm": 124.6296615600586,
      "learning_rate": 1.913144361047868e-05,
      "loss": 5.9476,
      "step": 3955
    },
    {
      "epoch": 0.1597093257973355,
      "grad_norm": 136.7162322998047,
      "learning_rate": 1.913091051514685e-05,
      "loss": 4.7164,
      "step": 3956
    },
    {
      "epoch": 0.15974969721437224,
      "grad_norm": 90.02531433105469,
      "learning_rate": 1.913037726369777e-05,
      "loss": 6.0712,
      "step": 3957
    },
    {
      "epoch": 0.15979006863140896,
      "grad_norm": 97.0315933227539,
      "learning_rate": 1.9129843856140563e-05,
      "loss": 4.2391,
      "step": 3958
    },
    {
      "epoch": 0.1598304400484457,
      "grad_norm": 217.79039001464844,
      "learning_rate": 1.9129310292484343e-05,
      "loss": 6.2535,
      "step": 3959
    },
    {
      "epoch": 0.15987081146548243,
      "grad_norm": 76.10218811035156,
      "learning_rate": 1.9128776572738234e-05,
      "loss": 4.7843,
      "step": 3960
    },
    {
      "epoch": 0.15991118288251918,
      "grad_norm": 153.13623046875,
      "learning_rate": 1.9128242696911366e-05,
      "loss": 5.9622,
      "step": 3961
    },
    {
      "epoch": 0.1599515542995559,
      "grad_norm": 82.0076904296875,
      "learning_rate": 1.9127708665012863e-05,
      "loss": 6.2114,
      "step": 3962
    },
    {
      "epoch": 0.15999192571659265,
      "grad_norm": 98.4731674194336,
      "learning_rate": 1.9127174477051854e-05,
      "loss": 4.0699,
      "step": 3963
    },
    {
      "epoch": 0.1600322971336294,
      "grad_norm": 96.61719512939453,
      "learning_rate": 1.912664013303748e-05,
      "loss": 6.4087,
      "step": 3964
    },
    {
      "epoch": 0.16007266855066613,
      "grad_norm": 93.8865737915039,
      "learning_rate": 1.912610563297887e-05,
      "loss": 5.9403,
      "step": 3965
    },
    {
      "epoch": 0.16011303996770287,
      "grad_norm": 126.47637176513672,
      "learning_rate": 1.912557097688516e-05,
      "loss": 5.3573,
      "step": 3966
    },
    {
      "epoch": 0.1601534113847396,
      "grad_norm": 107.08084106445312,
      "learning_rate": 1.9125036164765502e-05,
      "loss": 6.3417,
      "step": 3967
    },
    {
      "epoch": 0.16019378280177635,
      "grad_norm": 106.0974349975586,
      "learning_rate": 1.912450119662903e-05,
      "loss": 6.8122,
      "step": 3968
    },
    {
      "epoch": 0.16023415421881307,
      "grad_norm": 194.20257568359375,
      "learning_rate": 1.9123966072484898e-05,
      "loss": 6.0706,
      "step": 3969
    },
    {
      "epoch": 0.16027452563584982,
      "grad_norm": 78.22774505615234,
      "learning_rate": 1.9123430792342252e-05,
      "loss": 5.3145,
      "step": 3970
    },
    {
      "epoch": 0.16031489705288657,
      "grad_norm": 141.30972290039062,
      "learning_rate": 1.9122895356210242e-05,
      "loss": 4.8376,
      "step": 3971
    },
    {
      "epoch": 0.1603552684699233,
      "grad_norm": 101.20545196533203,
      "learning_rate": 1.9122359764098027e-05,
      "loss": 6.4147,
      "step": 3972
    },
    {
      "epoch": 0.16039563988696004,
      "grad_norm": 118.63796997070312,
      "learning_rate": 1.912182401601476e-05,
      "loss": 4.885,
      "step": 3973
    },
    {
      "epoch": 0.16043601130399676,
      "grad_norm": 140.35304260253906,
      "learning_rate": 1.9121288111969604e-05,
      "loss": 6.7794,
      "step": 3974
    },
    {
      "epoch": 0.1604763827210335,
      "grad_norm": 110.82140350341797,
      "learning_rate": 1.912075205197172e-05,
      "loss": 4.2794,
      "step": 3975
    },
    {
      "epoch": 0.16051675413807023,
      "grad_norm": 90.99923706054688,
      "learning_rate": 1.912021583603028e-05,
      "loss": 6.3708,
      "step": 3976
    },
    {
      "epoch": 0.16055712555510698,
      "grad_norm": 198.80886840820312,
      "learning_rate": 1.9119679464154443e-05,
      "loss": 4.9943,
      "step": 3977
    },
    {
      "epoch": 0.16059749697214373,
      "grad_norm": 99.243408203125,
      "learning_rate": 1.9119142936353377e-05,
      "loss": 5.6471,
      "step": 3978
    },
    {
      "epoch": 0.16063786838918045,
      "grad_norm": 79.23453521728516,
      "learning_rate": 1.911860625263627e-05,
      "loss": 4.0337,
      "step": 3979
    },
    {
      "epoch": 0.1606782398062172,
      "grad_norm": 117.66461944580078,
      "learning_rate": 1.9118069413012287e-05,
      "loss": 4.3912,
      "step": 3980
    },
    {
      "epoch": 0.16071861122325393,
      "grad_norm": 98.50515747070312,
      "learning_rate": 1.911753241749061e-05,
      "loss": 6.7261,
      "step": 3981
    },
    {
      "epoch": 0.16075898264029068,
      "grad_norm": 90.6673812866211,
      "learning_rate": 1.9116995266080418e-05,
      "loss": 5.3381,
      "step": 3982
    },
    {
      "epoch": 0.16079935405732743,
      "grad_norm": 133.07720947265625,
      "learning_rate": 1.9116457958790896e-05,
      "loss": 4.8208,
      "step": 3983
    },
    {
      "epoch": 0.16083972547436415,
      "grad_norm": 108.68301391601562,
      "learning_rate": 1.911592049563123e-05,
      "loss": 5.0897,
      "step": 3984
    },
    {
      "epoch": 0.1608800968914009,
      "grad_norm": 135.99127197265625,
      "learning_rate": 1.9115382876610617e-05,
      "loss": 5.3777,
      "step": 3985
    },
    {
      "epoch": 0.16092046830843762,
      "grad_norm": 147.8561248779297,
      "learning_rate": 1.9114845101738237e-05,
      "loss": 6.2688,
      "step": 3986
    },
    {
      "epoch": 0.16096083972547437,
      "grad_norm": 95.6446304321289,
      "learning_rate": 1.9114307171023294e-05,
      "loss": 5.1587,
      "step": 3987
    },
    {
      "epoch": 0.1610012111425111,
      "grad_norm": 95.25577545166016,
      "learning_rate": 1.911376908447498e-05,
      "loss": 4.9634,
      "step": 3988
    },
    {
      "epoch": 0.16104158255954784,
      "grad_norm": 113.18011474609375,
      "learning_rate": 1.9113230842102498e-05,
      "loss": 5.8379,
      "step": 3989
    },
    {
      "epoch": 0.1610819539765846,
      "grad_norm": 104.29454803466797,
      "learning_rate": 1.911269244391505e-05,
      "loss": 4.9243,
      "step": 3990
    },
    {
      "epoch": 0.1611223253936213,
      "grad_norm": 113.06546020507812,
      "learning_rate": 1.9112153889921835e-05,
      "loss": 4.4285,
      "step": 3991
    },
    {
      "epoch": 0.16116269681065806,
      "grad_norm": 106.59069061279297,
      "learning_rate": 1.9111615180132072e-05,
      "loss": 5.8801,
      "step": 3992
    },
    {
      "epoch": 0.16120306822769478,
      "grad_norm": 143.04190063476562,
      "learning_rate": 1.9111076314554967e-05,
      "loss": 6.3994,
      "step": 3993
    },
    {
      "epoch": 0.16124343964473153,
      "grad_norm": 142.7012939453125,
      "learning_rate": 1.911053729319973e-05,
      "loss": 5.4754,
      "step": 3994
    },
    {
      "epoch": 0.16128381106176826,
      "grad_norm": 74.2397232055664,
      "learning_rate": 1.9109998116075582e-05,
      "loss": 6.0676,
      "step": 3995
    },
    {
      "epoch": 0.161324182478805,
      "grad_norm": 111.53477478027344,
      "learning_rate": 1.910945878319174e-05,
      "loss": 6.6508,
      "step": 3996
    },
    {
      "epoch": 0.16136455389584176,
      "grad_norm": 174.60548400878906,
      "learning_rate": 1.9108919294557426e-05,
      "loss": 5.8662,
      "step": 3997
    },
    {
      "epoch": 0.16140492531287848,
      "grad_norm": 106.21039581298828,
      "learning_rate": 1.910837965018186e-05,
      "loss": 4.1909,
      "step": 3998
    },
    {
      "epoch": 0.16144529672991523,
      "grad_norm": 114.99520874023438,
      "learning_rate": 1.9107839850074274e-05,
      "loss": 5.1887,
      "step": 3999
    },
    {
      "epoch": 0.16148566814695195,
      "grad_norm": 109.22476959228516,
      "learning_rate": 1.910729989424389e-05,
      "loss": 5.8319,
      "step": 4000
    },
    {
      "epoch": 0.1615260395639887,
      "grad_norm": 96.91830444335938,
      "learning_rate": 1.9106759782699954e-05,
      "loss": 6.8157,
      "step": 4001
    },
    {
      "epoch": 0.16156641098102542,
      "grad_norm": 383.90252685546875,
      "learning_rate": 1.9106219515451684e-05,
      "loss": 6.5783,
      "step": 4002
    },
    {
      "epoch": 0.16160678239806217,
      "grad_norm": 79.24076080322266,
      "learning_rate": 1.9105679092508325e-05,
      "loss": 3.4462,
      "step": 4003
    },
    {
      "epoch": 0.16164715381509892,
      "grad_norm": 96.4423599243164,
      "learning_rate": 1.910513851387912e-05,
      "loss": 5.7474,
      "step": 4004
    },
    {
      "epoch": 0.16168752523213564,
      "grad_norm": 209.10325622558594,
      "learning_rate": 1.9104597779573307e-05,
      "loss": 4.9281,
      "step": 4005
    },
    {
      "epoch": 0.1617278966491724,
      "grad_norm": 141.02557373046875,
      "learning_rate": 1.9104056889600133e-05,
      "loss": 5.7952,
      "step": 4006
    },
    {
      "epoch": 0.16176826806620911,
      "grad_norm": 75.89485168457031,
      "learning_rate": 1.9103515843968848e-05,
      "loss": 6.2721,
      "step": 4007
    },
    {
      "epoch": 0.16180863948324586,
      "grad_norm": 180.1595916748047,
      "learning_rate": 1.9102974642688697e-05,
      "loss": 4.265,
      "step": 4008
    },
    {
      "epoch": 0.16184901090028259,
      "grad_norm": 105.26322937011719,
      "learning_rate": 1.910243328576894e-05,
      "loss": 6.8938,
      "step": 4009
    },
    {
      "epoch": 0.16188938231731934,
      "grad_norm": 103.9598159790039,
      "learning_rate": 1.9101891773218823e-05,
      "loss": 5.5644,
      "step": 4010
    },
    {
      "epoch": 0.16192975373435609,
      "grad_norm": 198.06109619140625,
      "learning_rate": 1.9101350105047616e-05,
      "loss": 4.7664,
      "step": 4011
    },
    {
      "epoch": 0.1619701251513928,
      "grad_norm": 90.69308471679688,
      "learning_rate": 1.9100808281264575e-05,
      "loss": 5.6809,
      "step": 4012
    },
    {
      "epoch": 0.16201049656842956,
      "grad_norm": 135.27056884765625,
      "learning_rate": 1.9100266301878962e-05,
      "loss": 4.8187,
      "step": 4013
    },
    {
      "epoch": 0.16205086798546628,
      "grad_norm": 111.59947967529297,
      "learning_rate": 1.9099724166900048e-05,
      "loss": 5.1737,
      "step": 4014
    },
    {
      "epoch": 0.16209123940250303,
      "grad_norm": 107.85392761230469,
      "learning_rate": 1.9099181876337102e-05,
      "loss": 5.9001,
      "step": 4015
    },
    {
      "epoch": 0.16213161081953978,
      "grad_norm": 97.7159652709961,
      "learning_rate": 1.909863943019939e-05,
      "loss": 6.5663,
      "step": 4016
    },
    {
      "epoch": 0.1621719822365765,
      "grad_norm": 135.15415954589844,
      "learning_rate": 1.9098096828496192e-05,
      "loss": 4.6506,
      "step": 4017
    },
    {
      "epoch": 0.16221235365361325,
      "grad_norm": 119.85482788085938,
      "learning_rate": 1.9097554071236785e-05,
      "loss": 5.2539,
      "step": 4018
    },
    {
      "epoch": 0.16225272507064997,
      "grad_norm": 155.12481689453125,
      "learning_rate": 1.909701115843045e-05,
      "loss": 5.0054,
      "step": 4019
    },
    {
      "epoch": 0.16229309648768672,
      "grad_norm": 345.74774169921875,
      "learning_rate": 1.9096468090086462e-05,
      "loss": 5.1443,
      "step": 4020
    },
    {
      "epoch": 0.16233346790472344,
      "grad_norm": 76.72197723388672,
      "learning_rate": 1.9095924866214114e-05,
      "loss": 5.7788,
      "step": 4021
    },
    {
      "epoch": 0.1623738393217602,
      "grad_norm": 140.22189331054688,
      "learning_rate": 1.909538148682269e-05,
      "loss": 6.7518,
      "step": 4022
    },
    {
      "epoch": 0.16241421073879694,
      "grad_norm": 105.91313934326172,
      "learning_rate": 1.9094837951921484e-05,
      "loss": 6.2089,
      "step": 4023
    },
    {
      "epoch": 0.16245458215583367,
      "grad_norm": 100.02558898925781,
      "learning_rate": 1.9094294261519787e-05,
      "loss": 4.8923,
      "step": 4024
    },
    {
      "epoch": 0.16249495357287042,
      "grad_norm": 132.2882080078125,
      "learning_rate": 1.9093750415626894e-05,
      "loss": 6.6027,
      "step": 4025
    },
    {
      "epoch": 0.16253532498990714,
      "grad_norm": 147.0712890625,
      "learning_rate": 1.90932064142521e-05,
      "loss": 5.9526,
      "step": 4026
    },
    {
      "epoch": 0.1625756964069439,
      "grad_norm": 164.9053497314453,
      "learning_rate": 1.9092662257404714e-05,
      "loss": 6.2903,
      "step": 4027
    },
    {
      "epoch": 0.1626160678239806,
      "grad_norm": 90.6007080078125,
      "learning_rate": 1.9092117945094036e-05,
      "loss": 7.0979,
      "step": 4028
    },
    {
      "epoch": 0.16265643924101736,
      "grad_norm": 84.02472686767578,
      "learning_rate": 1.9091573477329376e-05,
      "loss": 4.4982,
      "step": 4029
    },
    {
      "epoch": 0.1626968106580541,
      "grad_norm": 74.20033264160156,
      "learning_rate": 1.909102885412004e-05,
      "loss": 6.9015,
      "step": 4030
    },
    {
      "epoch": 0.16273718207509083,
      "grad_norm": 113.71038818359375,
      "learning_rate": 1.9090484075475334e-05,
      "loss": 5.8862,
      "step": 4031
    },
    {
      "epoch": 0.16277755349212758,
      "grad_norm": 62.95146560668945,
      "learning_rate": 1.908993914140458e-05,
      "loss": 5.7411,
      "step": 4032
    },
    {
      "epoch": 0.1628179249091643,
      "grad_norm": 132.07974243164062,
      "learning_rate": 1.9089394051917096e-05,
      "loss": 5.3276,
      "step": 4033
    },
    {
      "epoch": 0.16285829632620105,
      "grad_norm": 115.04995727539062,
      "learning_rate": 1.9088848807022198e-05,
      "loss": 5.7735,
      "step": 4034
    },
    {
      "epoch": 0.16289866774323777,
      "grad_norm": 74.44882202148438,
      "learning_rate": 1.908830340672921e-05,
      "loss": 5.0085,
      "step": 4035
    },
    {
      "epoch": 0.16293903916027452,
      "grad_norm": 99.58770751953125,
      "learning_rate": 1.9087757851047453e-05,
      "loss": 4.2359,
      "step": 4036
    },
    {
      "epoch": 0.16297941057731127,
      "grad_norm": 132.5151824951172,
      "learning_rate": 1.908721213998626e-05,
      "loss": 6.156,
      "step": 4037
    },
    {
      "epoch": 0.163019781994348,
      "grad_norm": 97.74227142333984,
      "learning_rate": 1.908666627355496e-05,
      "loss": 5.5453,
      "step": 4038
    },
    {
      "epoch": 0.16306015341138475,
      "grad_norm": 147.73458862304688,
      "learning_rate": 1.9086120251762886e-05,
      "loss": 3.774,
      "step": 4039
    },
    {
      "epoch": 0.16310052482842147,
      "grad_norm": 111.33139038085938,
      "learning_rate": 1.9085574074619372e-05,
      "loss": 5.318,
      "step": 4040
    },
    {
      "epoch": 0.16314089624545822,
      "grad_norm": 104.3404541015625,
      "learning_rate": 1.908502774213376e-05,
      "loss": 6.2353,
      "step": 4041
    },
    {
      "epoch": 0.16318126766249497,
      "grad_norm": 120.53575134277344,
      "learning_rate": 1.9084481254315386e-05,
      "loss": 5.5745,
      "step": 4042
    },
    {
      "epoch": 0.1632216390795317,
      "grad_norm": 105.86483764648438,
      "learning_rate": 1.90839346111736e-05,
      "loss": 6.6611,
      "step": 4043
    },
    {
      "epoch": 0.16326201049656844,
      "grad_norm": 194.79164123535156,
      "learning_rate": 1.908338781271774e-05,
      "loss": 6.2374,
      "step": 4044
    },
    {
      "epoch": 0.16330238191360516,
      "grad_norm": 94.53579711914062,
      "learning_rate": 1.908284085895716e-05,
      "loss": 5.7624,
      "step": 4045
    },
    {
      "epoch": 0.1633427533306419,
      "grad_norm": 88.6360092163086,
      "learning_rate": 1.9082293749901215e-05,
      "loss": 6.5686,
      "step": 4046
    },
    {
      "epoch": 0.16338312474767863,
      "grad_norm": 101.64668273925781,
      "learning_rate": 1.9081746485559255e-05,
      "loss": 5.9631,
      "step": 4047
    },
    {
      "epoch": 0.16342349616471538,
      "grad_norm": 82.78384399414062,
      "learning_rate": 1.9081199065940638e-05,
      "loss": 5.4912,
      "step": 4048
    },
    {
      "epoch": 0.16346386758175213,
      "grad_norm": 76.64469146728516,
      "learning_rate": 1.9080651491054718e-05,
      "loss": 5.5533,
      "step": 4049
    },
    {
      "epoch": 0.16350423899878885,
      "grad_norm": 83.40351104736328,
      "learning_rate": 1.9080103760910868e-05,
      "loss": 5.1014,
      "step": 4050
    },
    {
      "epoch": 0.1635446104158256,
      "grad_norm": 128.70823669433594,
      "learning_rate": 1.907955587551844e-05,
      "loss": 5.5072,
      "step": 4051
    },
    {
      "epoch": 0.16358498183286233,
      "grad_norm": 100.43354797363281,
      "learning_rate": 1.9079007834886817e-05,
      "loss": 5.5931,
      "step": 4052
    },
    {
      "epoch": 0.16362535324989907,
      "grad_norm": 82.65833282470703,
      "learning_rate": 1.9078459639025357e-05,
      "loss": 6.184,
      "step": 4053
    },
    {
      "epoch": 0.1636657246669358,
      "grad_norm": 99.08235931396484,
      "learning_rate": 1.907791128794344e-05,
      "loss": 5.7037,
      "step": 4054
    },
    {
      "epoch": 0.16370609608397255,
      "grad_norm": 86.3868637084961,
      "learning_rate": 1.9077362781650434e-05,
      "loss": 6.5099,
      "step": 4055
    },
    {
      "epoch": 0.1637464675010093,
      "grad_norm": 78.24600219726562,
      "learning_rate": 1.907681412015572e-05,
      "loss": 5.4026,
      "step": 4056
    },
    {
      "epoch": 0.16378683891804602,
      "grad_norm": 115.11271667480469,
      "learning_rate": 1.907626530346869e-05,
      "loss": 6.8735,
      "step": 4057
    },
    {
      "epoch": 0.16382721033508277,
      "grad_norm": 105.859130859375,
      "learning_rate": 1.907571633159871e-05,
      "loss": 5.3904,
      "step": 4058
    },
    {
      "epoch": 0.1638675817521195,
      "grad_norm": 125.93933868408203,
      "learning_rate": 1.9075167204555177e-05,
      "loss": 6.7261,
      "step": 4059
    },
    {
      "epoch": 0.16390795316915624,
      "grad_norm": 123.56517791748047,
      "learning_rate": 1.9074617922347473e-05,
      "loss": 6.5817,
      "step": 4060
    },
    {
      "epoch": 0.16394832458619296,
      "grad_norm": 100.94164276123047,
      "learning_rate": 1.9074068484984997e-05,
      "loss": 5.8788,
      "step": 4061
    },
    {
      "epoch": 0.1639886960032297,
      "grad_norm": 145.5796661376953,
      "learning_rate": 1.907351889247714e-05,
      "loss": 4.5103,
      "step": 4062
    },
    {
      "epoch": 0.16402906742026646,
      "grad_norm": 402.3216247558594,
      "learning_rate": 1.9072969144833297e-05,
      "loss": 6.3706,
      "step": 4063
    },
    {
      "epoch": 0.16406943883730318,
      "grad_norm": 118.09580993652344,
      "learning_rate": 1.9072419242062866e-05,
      "loss": 7.1538,
      "step": 4064
    },
    {
      "epoch": 0.16410981025433993,
      "grad_norm": 93.23589324951172,
      "learning_rate": 1.9071869184175254e-05,
      "loss": 4.9067,
      "step": 4065
    },
    {
      "epoch": 0.16415018167137665,
      "grad_norm": 96.17608642578125,
      "learning_rate": 1.9071318971179863e-05,
      "loss": 4.4204,
      "step": 4066
    },
    {
      "epoch": 0.1641905530884134,
      "grad_norm": 116.03299713134766,
      "learning_rate": 1.9070768603086103e-05,
      "loss": 4.3518,
      "step": 4067
    },
    {
      "epoch": 0.16423092450545015,
      "grad_norm": 6645.49462890625,
      "learning_rate": 1.9070218079903378e-05,
      "loss": 4.5217,
      "step": 4068
    },
    {
      "epoch": 0.16427129592248688,
      "grad_norm": 96.7618637084961,
      "learning_rate": 1.9069667401641107e-05,
      "loss": 4.2972,
      "step": 4069
    },
    {
      "epoch": 0.16431166733952363,
      "grad_norm": 125.47119903564453,
      "learning_rate": 1.90691165683087e-05,
      "loss": 4.4301,
      "step": 4070
    },
    {
      "epoch": 0.16435203875656035,
      "grad_norm": 99.1823501586914,
      "learning_rate": 1.9068565579915578e-05,
      "loss": 6.945,
      "step": 4071
    },
    {
      "epoch": 0.1643924101735971,
      "grad_norm": 121.04417419433594,
      "learning_rate": 1.9068014436471164e-05,
      "loss": 4.3921,
      "step": 4072
    },
    {
      "epoch": 0.16443278159063382,
      "grad_norm": 162.53204345703125,
      "learning_rate": 1.9067463137984878e-05,
      "loss": 3.9693,
      "step": 4073
    },
    {
      "epoch": 0.16447315300767057,
      "grad_norm": 102.80314636230469,
      "learning_rate": 1.9066911684466144e-05,
      "loss": 6.6074,
      "step": 4074
    },
    {
      "epoch": 0.16451352442470732,
      "grad_norm": 123.56639862060547,
      "learning_rate": 1.9066360075924393e-05,
      "loss": 6.0853,
      "step": 4075
    },
    {
      "epoch": 0.16455389584174404,
      "grad_norm": 112.98027801513672,
      "learning_rate": 1.9065808312369056e-05,
      "loss": 6.8677,
      "step": 4076
    },
    {
      "epoch": 0.1645942672587808,
      "grad_norm": 78.18863677978516,
      "learning_rate": 1.906525639380957e-05,
      "loss": 5.39,
      "step": 4077
    },
    {
      "epoch": 0.1646346386758175,
      "grad_norm": 144.83135986328125,
      "learning_rate": 1.9064704320255366e-05,
      "loss": 5.1476,
      "step": 4078
    },
    {
      "epoch": 0.16467501009285426,
      "grad_norm": 92.30769348144531,
      "learning_rate": 1.9064152091715885e-05,
      "loss": 4.5332,
      "step": 4079
    },
    {
      "epoch": 0.16471538150989098,
      "grad_norm": 102.99739074707031,
      "learning_rate": 1.9063599708200573e-05,
      "loss": 5.5085,
      "step": 4080
    },
    {
      "epoch": 0.16475575292692773,
      "grad_norm": 106.56449890136719,
      "learning_rate": 1.906304716971887e-05,
      "loss": 3.8602,
      "step": 4081
    },
    {
      "epoch": 0.16479612434396448,
      "grad_norm": 139.52052307128906,
      "learning_rate": 1.9062494476280222e-05,
      "loss": 6.7622,
      "step": 4082
    },
    {
      "epoch": 0.1648364957610012,
      "grad_norm": 69.17138671875,
      "learning_rate": 1.9061941627894082e-05,
      "loss": 5.237,
      "step": 4083
    },
    {
      "epoch": 0.16487686717803796,
      "grad_norm": 190.20262145996094,
      "learning_rate": 1.9061388624569902e-05,
      "loss": 5.1789,
      "step": 4084
    },
    {
      "epoch": 0.16491723859507468,
      "grad_norm": 137.7689971923828,
      "learning_rate": 1.9060835466317134e-05,
      "loss": 7.6556,
      "step": 4085
    },
    {
      "epoch": 0.16495761001211143,
      "grad_norm": 129.9397735595703,
      "learning_rate": 1.9060282153145243e-05,
      "loss": 5.2559,
      "step": 4086
    },
    {
      "epoch": 0.16499798142914815,
      "grad_norm": 103.86064910888672,
      "learning_rate": 1.905972868506368e-05,
      "loss": 5.469,
      "step": 4087
    },
    {
      "epoch": 0.1650383528461849,
      "grad_norm": 104.40564727783203,
      "learning_rate": 1.905917506208191e-05,
      "loss": 7.133,
      "step": 4088
    },
    {
      "epoch": 0.16507872426322165,
      "grad_norm": 95.8954086303711,
      "learning_rate": 1.9058621284209405e-05,
      "loss": 7.3419,
      "step": 4089
    },
    {
      "epoch": 0.16511909568025837,
      "grad_norm": 151.38523864746094,
      "learning_rate": 1.9058067351455633e-05,
      "loss": 7.0656,
      "step": 4090
    },
    {
      "epoch": 0.16515946709729512,
      "grad_norm": 130.619873046875,
      "learning_rate": 1.9057513263830055e-05,
      "loss": 5.6763,
      "step": 4091
    },
    {
      "epoch": 0.16519983851433184,
      "grad_norm": 134.06884765625,
      "learning_rate": 1.9056959021342156e-05,
      "loss": 7.241,
      "step": 4092
    },
    {
      "epoch": 0.1652402099313686,
      "grad_norm": 141.93728637695312,
      "learning_rate": 1.9056404624001407e-05,
      "loss": 5.7218,
      "step": 4093
    },
    {
      "epoch": 0.16528058134840534,
      "grad_norm": 157.52896118164062,
      "learning_rate": 1.9055850071817282e-05,
      "loss": 6.4252,
      "step": 4094
    },
    {
      "epoch": 0.16532095276544206,
      "grad_norm": 107.95218658447266,
      "learning_rate": 1.9055295364799272e-05,
      "loss": 6.0898,
      "step": 4095
    },
    {
      "epoch": 0.1653613241824788,
      "grad_norm": 232.1695098876953,
      "learning_rate": 1.9054740502956853e-05,
      "loss": 7.0902,
      "step": 4096
    },
    {
      "epoch": 0.16540169559951554,
      "grad_norm": 103.5868911743164,
      "learning_rate": 1.905418548629952e-05,
      "loss": 6.483,
      "step": 4097
    },
    {
      "epoch": 0.16544206701655229,
      "grad_norm": 86.10774993896484,
      "learning_rate": 1.9053630314836758e-05,
      "loss": 6.9961,
      "step": 4098
    },
    {
      "epoch": 0.165482438433589,
      "grad_norm": 112.5691146850586,
      "learning_rate": 1.905307498857806e-05,
      "loss": 5.9893,
      "step": 4099
    },
    {
      "epoch": 0.16552280985062576,
      "grad_norm": 90.68333435058594,
      "learning_rate": 1.9052519507532916e-05,
      "loss": 6.9197,
      "step": 4100
    },
    {
      "epoch": 0.1655631812676625,
      "grad_norm": 182.44346618652344,
      "learning_rate": 1.905196387171083e-05,
      "loss": 7.5149,
      "step": 4101
    },
    {
      "epoch": 0.16560355268469923,
      "grad_norm": 63.17719650268555,
      "learning_rate": 1.90514080811213e-05,
      "loss": 5.9489,
      "step": 4102
    },
    {
      "epoch": 0.16564392410173598,
      "grad_norm": 106.92493438720703,
      "learning_rate": 1.9050852135773828e-05,
      "loss": 5.9056,
      "step": 4103
    },
    {
      "epoch": 0.1656842955187727,
      "grad_norm": 94.76206970214844,
      "learning_rate": 1.9050296035677917e-05,
      "loss": 5.1487,
      "step": 4104
    },
    {
      "epoch": 0.16572466693580945,
      "grad_norm": 106.99100494384766,
      "learning_rate": 1.904973978084308e-05,
      "loss": 6.2031,
      "step": 4105
    },
    {
      "epoch": 0.16576503835284617,
      "grad_norm": 133.64625549316406,
      "learning_rate": 1.9049183371278828e-05,
      "loss": 5.9967,
      "step": 4106
    },
    {
      "epoch": 0.16580540976988292,
      "grad_norm": 74.966064453125,
      "learning_rate": 1.904862680699467e-05,
      "loss": 6.1692,
      "step": 4107
    },
    {
      "epoch": 0.16584578118691967,
      "grad_norm": 79.02952575683594,
      "learning_rate": 1.9048070088000123e-05,
      "loss": 5.4911,
      "step": 4108
    },
    {
      "epoch": 0.1658861526039564,
      "grad_norm": 87.399169921875,
      "learning_rate": 1.9047513214304705e-05,
      "loss": 5.2197,
      "step": 4109
    },
    {
      "epoch": 0.16592652402099314,
      "grad_norm": 71.46336364746094,
      "learning_rate": 1.904695618591794e-05,
      "loss": 7.3174,
      "step": 4110
    },
    {
      "epoch": 0.16596689543802987,
      "grad_norm": 95.15709686279297,
      "learning_rate": 1.9046399002849353e-05,
      "loss": 5.7489,
      "step": 4111
    },
    {
      "epoch": 0.16600726685506662,
      "grad_norm": 131.6155548095703,
      "learning_rate": 1.9045841665108463e-05,
      "loss": 6.5253,
      "step": 4112
    },
    {
      "epoch": 0.16604763827210334,
      "grad_norm": 101.35112762451172,
      "learning_rate": 1.9045284172704805e-05,
      "loss": 5.8085,
      "step": 4113
    },
    {
      "epoch": 0.1660880096891401,
      "grad_norm": 169.57196044921875,
      "learning_rate": 1.9044726525647914e-05,
      "loss": 5.9685,
      "step": 4114
    },
    {
      "epoch": 0.16612838110617684,
      "grad_norm": 193.1352996826172,
      "learning_rate": 1.9044168723947316e-05,
      "loss": 4.9276,
      "step": 4115
    },
    {
      "epoch": 0.16616875252321356,
      "grad_norm": 262.29437255859375,
      "learning_rate": 1.9043610767612555e-05,
      "loss": 4.5376,
      "step": 4116
    },
    {
      "epoch": 0.1662091239402503,
      "grad_norm": 79.70748901367188,
      "learning_rate": 1.9043052656653167e-05,
      "loss": 5.2775,
      "step": 4117
    },
    {
      "epoch": 0.16624949535728703,
      "grad_norm": 92.41650390625,
      "learning_rate": 1.9042494391078693e-05,
      "loss": 5.616,
      "step": 4118
    },
    {
      "epoch": 0.16628986677432378,
      "grad_norm": 98.44725799560547,
      "learning_rate": 1.9041935970898685e-05,
      "loss": 3.7636,
      "step": 4119
    },
    {
      "epoch": 0.16633023819136053,
      "grad_norm": 184.486572265625,
      "learning_rate": 1.9041377396122683e-05,
      "loss": 4.9798,
      "step": 4120
    },
    {
      "epoch": 0.16637060960839725,
      "grad_norm": 137.54624938964844,
      "learning_rate": 1.9040818666760245e-05,
      "loss": 5.1334,
      "step": 4121
    },
    {
      "epoch": 0.166410981025434,
      "grad_norm": 90.06401824951172,
      "learning_rate": 1.9040259782820917e-05,
      "loss": 4.5662,
      "step": 4122
    },
    {
      "epoch": 0.16645135244247072,
      "grad_norm": 108.8695297241211,
      "learning_rate": 1.9039700744314258e-05,
      "loss": 5.9155,
      "step": 4123
    },
    {
      "epoch": 0.16649172385950747,
      "grad_norm": 188.4958953857422,
      "learning_rate": 1.9039141551249826e-05,
      "loss": 5.9069,
      "step": 4124
    },
    {
      "epoch": 0.1665320952765442,
      "grad_norm": 110.21473693847656,
      "learning_rate": 1.9038582203637178e-05,
      "loss": 5.7006,
      "step": 4125
    },
    {
      "epoch": 0.16657246669358095,
      "grad_norm": 96.16175842285156,
      "learning_rate": 1.9038022701485883e-05,
      "loss": 6.1897,
      "step": 4126
    },
    {
      "epoch": 0.1666128381106177,
      "grad_norm": 98.81534576416016,
      "learning_rate": 1.9037463044805506e-05,
      "loss": 5.3431,
      "step": 4127
    },
    {
      "epoch": 0.16665320952765442,
      "grad_norm": 100.84362030029297,
      "learning_rate": 1.9036903233605614e-05,
      "loss": 5.1083,
      "step": 4128
    },
    {
      "epoch": 0.16669358094469117,
      "grad_norm": 177.21241760253906,
      "learning_rate": 1.903634326789578e-05,
      "loss": 5.5142,
      "step": 4129
    },
    {
      "epoch": 0.1667339523617279,
      "grad_norm": 152.8732452392578,
      "learning_rate": 1.9035783147685576e-05,
      "loss": 5.9253,
      "step": 4130
    },
    {
      "epoch": 0.16677432377876464,
      "grad_norm": 97.14341735839844,
      "learning_rate": 1.9035222872984584e-05,
      "loss": 6.4941,
      "step": 4131
    },
    {
      "epoch": 0.16681469519580136,
      "grad_norm": 102.87915802001953,
      "learning_rate": 1.9034662443802373e-05,
      "loss": 4.8704,
      "step": 4132
    },
    {
      "epoch": 0.1668550666128381,
      "grad_norm": 85.7868423461914,
      "learning_rate": 1.9034101860148538e-05,
      "loss": 6.0307,
      "step": 4133
    },
    {
      "epoch": 0.16689543802987486,
      "grad_norm": 89.29340362548828,
      "learning_rate": 1.9033541122032656e-05,
      "loss": 4.9525,
      "step": 4134
    },
    {
      "epoch": 0.16693580944691158,
      "grad_norm": 103.85648345947266,
      "learning_rate": 1.9032980229464313e-05,
      "loss": 5.3088,
      "step": 4135
    },
    {
      "epoch": 0.16697618086394833,
      "grad_norm": 94.57615661621094,
      "learning_rate": 1.9032419182453102e-05,
      "loss": 3.9293,
      "step": 4136
    },
    {
      "epoch": 0.16701655228098505,
      "grad_norm": 90.00204467773438,
      "learning_rate": 1.903185798100862e-05,
      "loss": 5.8175,
      "step": 4137
    },
    {
      "epoch": 0.1670569236980218,
      "grad_norm": 107.16587829589844,
      "learning_rate": 1.9031296625140453e-05,
      "loss": 7.0994,
      "step": 4138
    },
    {
      "epoch": 0.16709729511505853,
      "grad_norm": 151.1944580078125,
      "learning_rate": 1.9030735114858203e-05,
      "loss": 5.7434,
      "step": 4139
    },
    {
      "epoch": 0.16713766653209527,
      "grad_norm": 96.09297943115234,
      "learning_rate": 1.903017345017147e-05,
      "loss": 6.0184,
      "step": 4140
    },
    {
      "epoch": 0.16717803794913202,
      "grad_norm": 100.47521209716797,
      "learning_rate": 1.902961163108986e-05,
      "loss": 6.2148,
      "step": 4141
    },
    {
      "epoch": 0.16721840936616875,
      "grad_norm": 90.32920837402344,
      "learning_rate": 1.9029049657622978e-05,
      "loss": 6.5791,
      "step": 4142
    },
    {
      "epoch": 0.1672587807832055,
      "grad_norm": 89.8511734008789,
      "learning_rate": 1.902848752978043e-05,
      "loss": 5.6532,
      "step": 4143
    },
    {
      "epoch": 0.16729915220024222,
      "grad_norm": 94.14183807373047,
      "learning_rate": 1.902792524757183e-05,
      "loss": 4.9935,
      "step": 4144
    },
    {
      "epoch": 0.16733952361727897,
      "grad_norm": 88.61791229248047,
      "learning_rate": 1.9027362811006786e-05,
      "loss": 5.1199,
      "step": 4145
    },
    {
      "epoch": 0.16737989503431572,
      "grad_norm": 117.57915496826172,
      "learning_rate": 1.902680022009492e-05,
      "loss": 4.2177,
      "step": 4146
    },
    {
      "epoch": 0.16742026645135244,
      "grad_norm": 95.68894958496094,
      "learning_rate": 1.9026237474845852e-05,
      "loss": 4.3176,
      "step": 4147
    },
    {
      "epoch": 0.1674606378683892,
      "grad_norm": 153.32875061035156,
      "learning_rate": 1.90256745752692e-05,
      "loss": 4.0271,
      "step": 4148
    },
    {
      "epoch": 0.1675010092854259,
      "grad_norm": 186.81011962890625,
      "learning_rate": 1.9025111521374585e-05,
      "loss": 3.6058,
      "step": 4149
    },
    {
      "epoch": 0.16754138070246266,
      "grad_norm": 899.31884765625,
      "learning_rate": 1.9024548313171644e-05,
      "loss": 4.3023,
      "step": 4150
    },
    {
      "epoch": 0.16758175211949938,
      "grad_norm": 138.85247802734375,
      "learning_rate": 1.9023984950669996e-05,
      "loss": 4.3778,
      "step": 4151
    },
    {
      "epoch": 0.16762212353653613,
      "grad_norm": 89.70076751708984,
      "learning_rate": 1.902342143387928e-05,
      "loss": 4.8617,
      "step": 4152
    },
    {
      "epoch": 0.16766249495357288,
      "grad_norm": 90.00421905517578,
      "learning_rate": 1.9022857762809127e-05,
      "loss": 6.6829,
      "step": 4153
    },
    {
      "epoch": 0.1677028663706096,
      "grad_norm": 101.04619598388672,
      "learning_rate": 1.9022293937469178e-05,
      "loss": 3.5474,
      "step": 4154
    },
    {
      "epoch": 0.16774323778764635,
      "grad_norm": 84.4128646850586,
      "learning_rate": 1.902172995786907e-05,
      "loss": 4.1705,
      "step": 4155
    },
    {
      "epoch": 0.16778360920468308,
      "grad_norm": 116.91746520996094,
      "learning_rate": 1.902116582401845e-05,
      "loss": 5.9137,
      "step": 4156
    },
    {
      "epoch": 0.16782398062171983,
      "grad_norm": 102.9795913696289,
      "learning_rate": 1.9020601535926956e-05,
      "loss": 3.8839,
      "step": 4157
    },
    {
      "epoch": 0.16786435203875655,
      "grad_norm": 157.0396728515625,
      "learning_rate": 1.9020037093604244e-05,
      "loss": 5.2967,
      "step": 4158
    },
    {
      "epoch": 0.1679047234557933,
      "grad_norm": 85.284912109375,
      "learning_rate": 1.901947249705996e-05,
      "loss": 6.746,
      "step": 4159
    },
    {
      "epoch": 0.16794509487283005,
      "grad_norm": 183.19537353515625,
      "learning_rate": 1.9018907746303756e-05,
      "loss": 6.013,
      "step": 4160
    },
    {
      "epoch": 0.16798546628986677,
      "grad_norm": 120.3584213256836,
      "learning_rate": 1.9018342841345293e-05,
      "loss": 5.7261,
      "step": 4161
    },
    {
      "epoch": 0.16802583770690352,
      "grad_norm": 175.1416778564453,
      "learning_rate": 1.9017777782194225e-05,
      "loss": 6.3685,
      "step": 4162
    },
    {
      "epoch": 0.16806620912394024,
      "grad_norm": 1254.9954833984375,
      "learning_rate": 1.9017212568860217e-05,
      "loss": 5.1838,
      "step": 4163
    },
    {
      "epoch": 0.168106580540977,
      "grad_norm": 102.29475402832031,
      "learning_rate": 1.9016647201352927e-05,
      "loss": 5.8194,
      "step": 4164
    },
    {
      "epoch": 0.1681469519580137,
      "grad_norm": 107.27105712890625,
      "learning_rate": 1.901608167968203e-05,
      "loss": 5.7277,
      "step": 4165
    },
    {
      "epoch": 0.16818732337505046,
      "grad_norm": 109.39690399169922,
      "learning_rate": 1.9015516003857186e-05,
      "loss": 5.9511,
      "step": 4166
    },
    {
      "epoch": 0.1682276947920872,
      "grad_norm": 111.22325897216797,
      "learning_rate": 1.9014950173888075e-05,
      "loss": 6.9531,
      "step": 4167
    },
    {
      "epoch": 0.16826806620912393,
      "grad_norm": 119.74193572998047,
      "learning_rate": 1.9014384189784364e-05,
      "loss": 6.2587,
      "step": 4168
    },
    {
      "epoch": 0.16830843762616068,
      "grad_norm": 119.28367614746094,
      "learning_rate": 1.9013818051555738e-05,
      "loss": 6.5552,
      "step": 4169
    },
    {
      "epoch": 0.1683488090431974,
      "grad_norm": 133.93670654296875,
      "learning_rate": 1.9013251759211866e-05,
      "loss": 3.7263,
      "step": 4170
    },
    {
      "epoch": 0.16838918046023416,
      "grad_norm": 73.2082748413086,
      "learning_rate": 1.901268531276244e-05,
      "loss": 5.8792,
      "step": 4171
    },
    {
      "epoch": 0.1684295518772709,
      "grad_norm": 123.25196838378906,
      "learning_rate": 1.901211871221714e-05,
      "loss": 5.5786,
      "step": 4172
    },
    {
      "epoch": 0.16846992329430763,
      "grad_norm": 90.04666900634766,
      "learning_rate": 1.9011551957585657e-05,
      "loss": 6.0588,
      "step": 4173
    },
    {
      "epoch": 0.16851029471134438,
      "grad_norm": 151.9364013671875,
      "learning_rate": 1.9010985048877677e-05,
      "loss": 6.5058,
      "step": 4174
    },
    {
      "epoch": 0.1685506661283811,
      "grad_norm": 211.07412719726562,
      "learning_rate": 1.9010417986102896e-05,
      "loss": 5.9221,
      "step": 4175
    },
    {
      "epoch": 0.16859103754541785,
      "grad_norm": 120.45613861083984,
      "learning_rate": 1.900985076927101e-05,
      "loss": 4.5502,
      "step": 4176
    },
    {
      "epoch": 0.16863140896245457,
      "grad_norm": 80.73838806152344,
      "learning_rate": 1.9009283398391712e-05,
      "loss": 5.4156,
      "step": 4177
    },
    {
      "epoch": 0.16867178037949132,
      "grad_norm": 97.24917602539062,
      "learning_rate": 1.9008715873474704e-05,
      "loss": 4.545,
      "step": 4178
    },
    {
      "epoch": 0.16871215179652807,
      "grad_norm": 84.22090148925781,
      "learning_rate": 1.9008148194529695e-05,
      "loss": 6.3703,
      "step": 4179
    },
    {
      "epoch": 0.1687525232135648,
      "grad_norm": 117.14862060546875,
      "learning_rate": 1.9007580361566387e-05,
      "loss": 5.9078,
      "step": 4180
    },
    {
      "epoch": 0.16879289463060154,
      "grad_norm": 144.3307342529297,
      "learning_rate": 1.900701237459449e-05,
      "loss": 5.7247,
      "step": 4181
    },
    {
      "epoch": 0.16883326604763826,
      "grad_norm": 134.5896453857422,
      "learning_rate": 1.9006444233623713e-05,
      "loss": 4.1185,
      "step": 4182
    },
    {
      "epoch": 0.168873637464675,
      "grad_norm": 184.36282348632812,
      "learning_rate": 1.900587593866377e-05,
      "loss": 5.842,
      "step": 4183
    },
    {
      "epoch": 0.16891400888171174,
      "grad_norm": 94.98867797851562,
      "learning_rate": 1.9005307489724382e-05,
      "loss": 5.5374,
      "step": 4184
    },
    {
      "epoch": 0.16895438029874849,
      "grad_norm": 94.30541229248047,
      "learning_rate": 1.900473888681526e-05,
      "loss": 5.8547,
      "step": 4185
    },
    {
      "epoch": 0.16899475171578524,
      "grad_norm": 91.1836166381836,
      "learning_rate": 1.9004170129946135e-05,
      "loss": 5.7664,
      "step": 4186
    },
    {
      "epoch": 0.16903512313282196,
      "grad_norm": 90.03325653076172,
      "learning_rate": 1.9003601219126727e-05,
      "loss": 5.9003,
      "step": 4187
    },
    {
      "epoch": 0.1690754945498587,
      "grad_norm": 214.6151885986328,
      "learning_rate": 1.9003032154366762e-05,
      "loss": 5.3339,
      "step": 4188
    },
    {
      "epoch": 0.16911586596689543,
      "grad_norm": 69.40335845947266,
      "learning_rate": 1.900246293567597e-05,
      "loss": 6.7055,
      "step": 4189
    },
    {
      "epoch": 0.16915623738393218,
      "grad_norm": 5054.09716796875,
      "learning_rate": 1.9001893563064084e-05,
      "loss": 5.4764,
      "step": 4190
    },
    {
      "epoch": 0.1691966088009689,
      "grad_norm": 111.19458770751953,
      "learning_rate": 1.9001324036540838e-05,
      "loss": 5.2081,
      "step": 4191
    },
    {
      "epoch": 0.16923698021800565,
      "grad_norm": 132.93966674804688,
      "learning_rate": 1.9000754356115975e-05,
      "loss": 6.2247,
      "step": 4192
    },
    {
      "epoch": 0.1692773516350424,
      "grad_norm": 141.35093688964844,
      "learning_rate": 1.900018452179923e-05,
      "loss": 4.1541,
      "step": 4193
    },
    {
      "epoch": 0.16931772305207912,
      "grad_norm": 293.621337890625,
      "learning_rate": 1.899961453360034e-05,
      "loss": 6.2531,
      "step": 4194
    },
    {
      "epoch": 0.16935809446911587,
      "grad_norm": 181.09376525878906,
      "learning_rate": 1.8999044391529063e-05,
      "loss": 5.3843,
      "step": 4195
    },
    {
      "epoch": 0.1693984658861526,
      "grad_norm": 104.20925903320312,
      "learning_rate": 1.899847409559514e-05,
      "loss": 5.5432,
      "step": 4196
    },
    {
      "epoch": 0.16943883730318934,
      "grad_norm": 287.8548583984375,
      "learning_rate": 1.8997903645808322e-05,
      "loss": 4.5229,
      "step": 4197
    },
    {
      "epoch": 0.1694792087202261,
      "grad_norm": 113.3355712890625,
      "learning_rate": 1.8997333042178364e-05,
      "loss": 6.1457,
      "step": 4198
    },
    {
      "epoch": 0.16951958013726282,
      "grad_norm": 154.2113037109375,
      "learning_rate": 1.899676228471502e-05,
      "loss": 5.1761,
      "step": 4199
    },
    {
      "epoch": 0.16955995155429957,
      "grad_norm": 91.9105453491211,
      "learning_rate": 1.8996191373428053e-05,
      "loss": 3.3534,
      "step": 4200
    },
    {
      "epoch": 0.1696003229713363,
      "grad_norm": 128.34974670410156,
      "learning_rate": 1.899562030832722e-05,
      "loss": 4.2644,
      "step": 4201
    },
    {
      "epoch": 0.16964069438837304,
      "grad_norm": 98.91584014892578,
      "learning_rate": 1.8995049089422282e-05,
      "loss": 5.2877,
      "step": 4202
    },
    {
      "epoch": 0.16968106580540976,
      "grad_norm": 102.93064880371094,
      "learning_rate": 1.8994477716723013e-05,
      "loss": 4.8545,
      "step": 4203
    },
    {
      "epoch": 0.1697214372224465,
      "grad_norm": 107.35980224609375,
      "learning_rate": 1.899390619023918e-05,
      "loss": 5.3798,
      "step": 4204
    },
    {
      "epoch": 0.16976180863948326,
      "grad_norm": 154.0720672607422,
      "learning_rate": 1.899333450998055e-05,
      "loss": 5.4007,
      "step": 4205
    },
    {
      "epoch": 0.16980218005651998,
      "grad_norm": 126.16107940673828,
      "learning_rate": 1.8992762675956905e-05,
      "loss": 4.8053,
      "step": 4206
    },
    {
      "epoch": 0.16984255147355673,
      "grad_norm": 86.44194030761719,
      "learning_rate": 1.8992190688178013e-05,
      "loss": 5.2059,
      "step": 4207
    },
    {
      "epoch": 0.16988292289059345,
      "grad_norm": 96.4544677734375,
      "learning_rate": 1.8991618546653664e-05,
      "loss": 4.2189,
      "step": 4208
    },
    {
      "epoch": 0.1699232943076302,
      "grad_norm": 88.41159057617188,
      "learning_rate": 1.899104625139363e-05,
      "loss": 5.0426,
      "step": 4209
    },
    {
      "epoch": 0.16996366572466692,
      "grad_norm": 138.53504943847656,
      "learning_rate": 1.8990473802407704e-05,
      "loss": 4.7355,
      "step": 4210
    },
    {
      "epoch": 0.17000403714170367,
      "grad_norm": 119.17308044433594,
      "learning_rate": 1.8989901199705668e-05,
      "loss": 5.1554,
      "step": 4211
    },
    {
      "epoch": 0.17004440855874042,
      "grad_norm": 102.62602996826172,
      "learning_rate": 1.8989328443297313e-05,
      "loss": 6.72,
      "step": 4212
    },
    {
      "epoch": 0.17008477997577715,
      "grad_norm": 1324.5916748046875,
      "learning_rate": 1.8988755533192438e-05,
      "loss": 6.8357,
      "step": 4213
    },
    {
      "epoch": 0.1701251513928139,
      "grad_norm": 105.89481353759766,
      "learning_rate": 1.898818246940083e-05,
      "loss": 5.5262,
      "step": 4214
    },
    {
      "epoch": 0.17016552280985062,
      "grad_norm": 669.1353759765625,
      "learning_rate": 1.898760925193229e-05,
      "loss": 6.4031,
      "step": 4215
    },
    {
      "epoch": 0.17020589422688737,
      "grad_norm": 122.48753356933594,
      "learning_rate": 1.898703588079662e-05,
      "loss": 5.6789,
      "step": 4216
    },
    {
      "epoch": 0.1702462656439241,
      "grad_norm": 154.03829956054688,
      "learning_rate": 1.898646235600363e-05,
      "loss": 4.6253,
      "step": 4217
    },
    {
      "epoch": 0.17028663706096084,
      "grad_norm": 100.88961029052734,
      "learning_rate": 1.898588867756311e-05,
      "loss": 6.7036,
      "step": 4218
    },
    {
      "epoch": 0.1703270084779976,
      "grad_norm": 114.43960571289062,
      "learning_rate": 1.898531484548488e-05,
      "loss": 5.8458,
      "step": 4219
    },
    {
      "epoch": 0.1703673798950343,
      "grad_norm": 107.02978515625,
      "learning_rate": 1.8984740859778744e-05,
      "loss": 7.3679,
      "step": 4220
    },
    {
      "epoch": 0.17040775131207106,
      "grad_norm": 115.21469116210938,
      "learning_rate": 1.8984166720454526e-05,
      "loss": 5.9607,
      "step": 4221
    },
    {
      "epoch": 0.17044812272910778,
      "grad_norm": 102.32557678222656,
      "learning_rate": 1.8983592427522035e-05,
      "loss": 5.9562,
      "step": 4222
    },
    {
      "epoch": 0.17048849414614453,
      "grad_norm": 116.14411163330078,
      "learning_rate": 1.8983017980991094e-05,
      "loss": 6.7058,
      "step": 4223
    },
    {
      "epoch": 0.17052886556318128,
      "grad_norm": 111.8249740600586,
      "learning_rate": 1.8982443380871516e-05,
      "loss": 6.5809,
      "step": 4224
    },
    {
      "epoch": 0.170569236980218,
      "grad_norm": 98.55652618408203,
      "learning_rate": 1.8981868627173136e-05,
      "loss": 5.0377,
      "step": 4225
    },
    {
      "epoch": 0.17060960839725475,
      "grad_norm": 96.88633728027344,
      "learning_rate": 1.8981293719905775e-05,
      "loss": 6.1959,
      "step": 4226
    },
    {
      "epoch": 0.17064997981429147,
      "grad_norm": 128.22996520996094,
      "learning_rate": 1.8980718659079266e-05,
      "loss": 5.4922,
      "step": 4227
    },
    {
      "epoch": 0.17069035123132822,
      "grad_norm": 99.76618194580078,
      "learning_rate": 1.898014344470344e-05,
      "loss": 6.6138,
      "step": 4228
    },
    {
      "epoch": 0.17073072264836495,
      "grad_norm": 110.67215728759766,
      "learning_rate": 1.897956807678813e-05,
      "loss": 5.1532,
      "step": 4229
    },
    {
      "epoch": 0.1707710940654017,
      "grad_norm": 163.2844696044922,
      "learning_rate": 1.8978992555343173e-05,
      "loss": 3.5371,
      "step": 4230
    },
    {
      "epoch": 0.17081146548243845,
      "grad_norm": 104.3866195678711,
      "learning_rate": 1.8978416880378413e-05,
      "loss": 5.5749,
      "step": 4231
    },
    {
      "epoch": 0.17085183689947517,
      "grad_norm": 141.9801483154297,
      "learning_rate": 1.897784105190369e-05,
      "loss": 5.6103,
      "step": 4232
    },
    {
      "epoch": 0.17089220831651192,
      "grad_norm": 68.60130310058594,
      "learning_rate": 1.897726506992885e-05,
      "loss": 4.7231,
      "step": 4233
    },
    {
      "epoch": 0.17093257973354864,
      "grad_norm": 100.0654296875,
      "learning_rate": 1.897668893446374e-05,
      "loss": 5.4935,
      "step": 4234
    },
    {
      "epoch": 0.1709729511505854,
      "grad_norm": 118.10639190673828,
      "learning_rate": 1.897611264551821e-05,
      "loss": 5.3369,
      "step": 4235
    },
    {
      "epoch": 0.1710133225676221,
      "grad_norm": 120.39393615722656,
      "learning_rate": 1.897553620310212e-05,
      "loss": 5.1121,
      "step": 4236
    },
    {
      "epoch": 0.17105369398465886,
      "grad_norm": 79.29304504394531,
      "learning_rate": 1.8974959607225318e-05,
      "loss": 5.9687,
      "step": 4237
    },
    {
      "epoch": 0.1710940654016956,
      "grad_norm": 124.65324401855469,
      "learning_rate": 1.897438285789766e-05,
      "loss": 4.7266,
      "step": 4238
    },
    {
      "epoch": 0.17113443681873233,
      "grad_norm": 173.00332641601562,
      "learning_rate": 1.8973805955129013e-05,
      "loss": 6.3174,
      "step": 4239
    },
    {
      "epoch": 0.17117480823576908,
      "grad_norm": 122.26052856445312,
      "learning_rate": 1.8973228898929243e-05,
      "loss": 5.3441,
      "step": 4240
    },
    {
      "epoch": 0.1712151796528058,
      "grad_norm": 153.40626525878906,
      "learning_rate": 1.897265168930821e-05,
      "loss": 5.5468,
      "step": 4241
    },
    {
      "epoch": 0.17125555106984255,
      "grad_norm": 112.3570556640625,
      "learning_rate": 1.897207432627579e-05,
      "loss": 6.1082,
      "step": 4242
    },
    {
      "epoch": 0.17129592248687928,
      "grad_norm": 148.0279083251953,
      "learning_rate": 1.8971496809841847e-05,
      "loss": 5.4623,
      "step": 4243
    },
    {
      "epoch": 0.17133629390391603,
      "grad_norm": 86.75447082519531,
      "learning_rate": 1.8970919140016257e-05,
      "loss": 4.9301,
      "step": 4244
    },
    {
      "epoch": 0.17137666532095278,
      "grad_norm": 78.48841857910156,
      "learning_rate": 1.89703413168089e-05,
      "loss": 5.742,
      "step": 4245
    },
    {
      "epoch": 0.1714170367379895,
      "grad_norm": 166.4467010498047,
      "learning_rate": 1.8969763340229653e-05,
      "loss": 4.0886,
      "step": 4246
    },
    {
      "epoch": 0.17145740815502625,
      "grad_norm": 82.85955810546875,
      "learning_rate": 1.89691852102884e-05,
      "loss": 5.1158,
      "step": 4247
    },
    {
      "epoch": 0.17149777957206297,
      "grad_norm": 102.6223373413086,
      "learning_rate": 1.8968606926995026e-05,
      "loss": 6.7551,
      "step": 4248
    },
    {
      "epoch": 0.17153815098909972,
      "grad_norm": 137.2556915283203,
      "learning_rate": 1.8968028490359417e-05,
      "loss": 5.8831,
      "step": 4249
    },
    {
      "epoch": 0.17157852240613647,
      "grad_norm": 110.02703094482422,
      "learning_rate": 1.896744990039146e-05,
      "loss": 6.7385,
      "step": 4250
    },
    {
      "epoch": 0.1716188938231732,
      "grad_norm": 172.33905029296875,
      "learning_rate": 1.896687115710105e-05,
      "loss": 6.3463,
      "step": 4251
    },
    {
      "epoch": 0.17165926524020994,
      "grad_norm": 116.70018768310547,
      "learning_rate": 1.896629226049808e-05,
      "loss": 5.8669,
      "step": 4252
    },
    {
      "epoch": 0.17169963665724666,
      "grad_norm": 110.16179656982422,
      "learning_rate": 1.8965713210592458e-05,
      "loss": 5.1504,
      "step": 4253
    },
    {
      "epoch": 0.1717400080742834,
      "grad_norm": 187.35208129882812,
      "learning_rate": 1.896513400739407e-05,
      "loss": 5.365,
      "step": 4254
    },
    {
      "epoch": 0.17178037949132013,
      "grad_norm": 226.13157653808594,
      "learning_rate": 1.8964554650912828e-05,
      "loss": 6.3289,
      "step": 4255
    },
    {
      "epoch": 0.17182075090835688,
      "grad_norm": 105.42972564697266,
      "learning_rate": 1.8963975141158635e-05,
      "loss": 5.4721,
      "step": 4256
    },
    {
      "epoch": 0.17186112232539363,
      "grad_norm": 129.8288116455078,
      "learning_rate": 1.8963395478141396e-05,
      "loss": 6.2259,
      "step": 4257
    },
    {
      "epoch": 0.17190149374243036,
      "grad_norm": 105.34442901611328,
      "learning_rate": 1.8962815661871032e-05,
      "loss": 4.8344,
      "step": 4258
    },
    {
      "epoch": 0.1719418651594671,
      "grad_norm": 263.18896484375,
      "learning_rate": 1.8962235692357445e-05,
      "loss": 7.2312,
      "step": 4259
    },
    {
      "epoch": 0.17198223657650383,
      "grad_norm": 83.94623565673828,
      "learning_rate": 1.8961655569610557e-05,
      "loss": 5.7385,
      "step": 4260
    },
    {
      "epoch": 0.17202260799354058,
      "grad_norm": 275.3376770019531,
      "learning_rate": 1.8961075293640287e-05,
      "loss": 6.179,
      "step": 4261
    },
    {
      "epoch": 0.1720629794105773,
      "grad_norm": 165.7292022705078,
      "learning_rate": 1.8960494864456557e-05,
      "loss": 6.4699,
      "step": 4262
    },
    {
      "epoch": 0.17210335082761405,
      "grad_norm": 114.7677230834961,
      "learning_rate": 1.8959914282069283e-05,
      "loss": 5.2047,
      "step": 4263
    },
    {
      "epoch": 0.1721437222446508,
      "grad_norm": 88.88772583007812,
      "learning_rate": 1.8959333546488402e-05,
      "loss": 6.327,
      "step": 4264
    },
    {
      "epoch": 0.17218409366168752,
      "grad_norm": 87.80389404296875,
      "learning_rate": 1.8958752657723836e-05,
      "loss": 6.3076,
      "step": 4265
    },
    {
      "epoch": 0.17222446507872427,
      "grad_norm": 157.89207458496094,
      "learning_rate": 1.8958171615785524e-05,
      "loss": 6.4497,
      "step": 4266
    },
    {
      "epoch": 0.172264836495761,
      "grad_norm": 82.8712158203125,
      "learning_rate": 1.895759042068339e-05,
      "loss": 6.2759,
      "step": 4267
    },
    {
      "epoch": 0.17230520791279774,
      "grad_norm": 106.5311279296875,
      "learning_rate": 1.8957009072427385e-05,
      "loss": 6.9463,
      "step": 4268
    },
    {
      "epoch": 0.17234557932983446,
      "grad_norm": 199.5565643310547,
      "learning_rate": 1.8956427571027435e-05,
      "loss": 4.4828,
      "step": 4269
    },
    {
      "epoch": 0.1723859507468712,
      "grad_norm": 104.95059967041016,
      "learning_rate": 1.8955845916493493e-05,
      "loss": 5.0501,
      "step": 4270
    },
    {
      "epoch": 0.17242632216390796,
      "grad_norm": 107.31655883789062,
      "learning_rate": 1.8955264108835493e-05,
      "loss": 5.2952,
      "step": 4271
    },
    {
      "epoch": 0.17246669358094469,
      "grad_norm": 105.42154693603516,
      "learning_rate": 1.8954682148063393e-05,
      "loss": 5.4696,
      "step": 4272
    },
    {
      "epoch": 0.17250706499798144,
      "grad_norm": 216.18153381347656,
      "learning_rate": 1.895410003418714e-05,
      "loss": 5.7277,
      "step": 4273
    },
    {
      "epoch": 0.17254743641501816,
      "grad_norm": 125.58244323730469,
      "learning_rate": 1.8953517767216684e-05,
      "loss": 5.6284,
      "step": 4274
    },
    {
      "epoch": 0.1725878078320549,
      "grad_norm": 94.55319213867188,
      "learning_rate": 1.895293534716198e-05,
      "loss": 6.361,
      "step": 4275
    },
    {
      "epoch": 0.17262817924909166,
      "grad_norm": 162.66690063476562,
      "learning_rate": 1.8952352774032992e-05,
      "loss": 6.0862,
      "step": 4276
    },
    {
      "epoch": 0.17266855066612838,
      "grad_norm": 183.30479431152344,
      "learning_rate": 1.8951770047839674e-05,
      "loss": 4.4943,
      "step": 4277
    },
    {
      "epoch": 0.17270892208316513,
      "grad_norm": 128.63565063476562,
      "learning_rate": 1.8951187168591995e-05,
      "loss": 5.0681,
      "step": 4278
    },
    {
      "epoch": 0.17274929350020185,
      "grad_norm": 108.14425659179688,
      "learning_rate": 1.8950604136299916e-05,
      "loss": 4.3277,
      "step": 4279
    },
    {
      "epoch": 0.1727896649172386,
      "grad_norm": 159.31463623046875,
      "learning_rate": 1.895002095097341e-05,
      "loss": 6.5945,
      "step": 4280
    },
    {
      "epoch": 0.17283003633427532,
      "grad_norm": 85.77081298828125,
      "learning_rate": 1.894943761262244e-05,
      "loss": 4.0403,
      "step": 4281
    },
    {
      "epoch": 0.17287040775131207,
      "grad_norm": 104.10196685791016,
      "learning_rate": 1.894885412125699e-05,
      "loss": 5.0386,
      "step": 4282
    },
    {
      "epoch": 0.17291077916834882,
      "grad_norm": 104.24125671386719,
      "learning_rate": 1.8948270476887028e-05,
      "loss": 6.5472,
      "step": 4283
    },
    {
      "epoch": 0.17295115058538554,
      "grad_norm": 107.15171813964844,
      "learning_rate": 1.894768667952254e-05,
      "loss": 7.0814,
      "step": 4284
    },
    {
      "epoch": 0.1729915220024223,
      "grad_norm": 114.07896423339844,
      "learning_rate": 1.8947102729173502e-05,
      "loss": 6.0692,
      "step": 4285
    },
    {
      "epoch": 0.17303189341945902,
      "grad_norm": 117.51366424560547,
      "learning_rate": 1.8946518625849902e-05,
      "loss": 5.7548,
      "step": 4286
    },
    {
      "epoch": 0.17307226483649576,
      "grad_norm": 87.33261108398438,
      "learning_rate": 1.8945934369561722e-05,
      "loss": 5.3405,
      "step": 4287
    },
    {
      "epoch": 0.1731126362535325,
      "grad_norm": 111.5725326538086,
      "learning_rate": 1.894534996031896e-05,
      "loss": 5.0979,
      "step": 4288
    },
    {
      "epoch": 0.17315300767056924,
      "grad_norm": 80.76530456542969,
      "learning_rate": 1.8944765398131597e-05,
      "loss": 6.731,
      "step": 4289
    },
    {
      "epoch": 0.173193379087606,
      "grad_norm": 99.94351196289062,
      "learning_rate": 1.8944180683009638e-05,
      "loss": 5.477,
      "step": 4290
    },
    {
      "epoch": 0.1732337505046427,
      "grad_norm": 88.23161315917969,
      "learning_rate": 1.8943595814963073e-05,
      "loss": 5.2259,
      "step": 4291
    },
    {
      "epoch": 0.17327412192167946,
      "grad_norm": 114.78129577636719,
      "learning_rate": 1.89430107940019e-05,
      "loss": 5.9635,
      "step": 4292
    },
    {
      "epoch": 0.17331449333871618,
      "grad_norm": 116.90109252929688,
      "learning_rate": 1.8942425620136133e-05,
      "loss": 6.0212,
      "step": 4293
    },
    {
      "epoch": 0.17335486475575293,
      "grad_norm": 174.61082458496094,
      "learning_rate": 1.8941840293375765e-05,
      "loss": 5.7156,
      "step": 4294
    },
    {
      "epoch": 0.17339523617278965,
      "grad_norm": 105.21041107177734,
      "learning_rate": 1.8941254813730815e-05,
      "loss": 6.0219,
      "step": 4295
    },
    {
      "epoch": 0.1734356075898264,
      "grad_norm": 86.59598541259766,
      "learning_rate": 1.8940669181211283e-05,
      "loss": 3.9514,
      "step": 4296
    },
    {
      "epoch": 0.17347597900686315,
      "grad_norm": 75.17659759521484,
      "learning_rate": 1.8940083395827185e-05,
      "loss": 5.5525,
      "step": 4297
    },
    {
      "epoch": 0.17351635042389987,
      "grad_norm": 119.03913879394531,
      "learning_rate": 1.8939497457588536e-05,
      "loss": 5.0654,
      "step": 4298
    },
    {
      "epoch": 0.17355672184093662,
      "grad_norm": 76.01866912841797,
      "learning_rate": 1.893891136650536e-05,
      "loss": 6.9018,
      "step": 4299
    },
    {
      "epoch": 0.17359709325797335,
      "grad_norm": 85.0959701538086,
      "learning_rate": 1.893832512258767e-05,
      "loss": 5.6157,
      "step": 4300
    },
    {
      "epoch": 0.1736374646750101,
      "grad_norm": 123.35259246826172,
      "learning_rate": 1.8937738725845494e-05,
      "loss": 5.0928,
      "step": 4301
    },
    {
      "epoch": 0.17367783609204684,
      "grad_norm": 82.12181854248047,
      "learning_rate": 1.893715217628886e-05,
      "loss": 6.5979,
      "step": 4302
    },
    {
      "epoch": 0.17371820750908357,
      "grad_norm": 107.11579132080078,
      "learning_rate": 1.8936565473927787e-05,
      "loss": 7.006,
      "step": 4303
    },
    {
      "epoch": 0.17375857892612032,
      "grad_norm": 116.5339584350586,
      "learning_rate": 1.8935978618772316e-05,
      "loss": 5.3016,
      "step": 4304
    },
    {
      "epoch": 0.17379895034315704,
      "grad_norm": 92.93474578857422,
      "learning_rate": 1.893539161083248e-05,
      "loss": 6.8924,
      "step": 4305
    },
    {
      "epoch": 0.1738393217601938,
      "grad_norm": 156.7423858642578,
      "learning_rate": 1.8934804450118314e-05,
      "loss": 5.684,
      "step": 4306
    },
    {
      "epoch": 0.1738796931772305,
      "grad_norm": 97.36425018310547,
      "learning_rate": 1.893421713663985e-05,
      "loss": 5.6859,
      "step": 4307
    },
    {
      "epoch": 0.17392006459426726,
      "grad_norm": 80.54924011230469,
      "learning_rate": 1.893362967040714e-05,
      "loss": 5.4963,
      "step": 4308
    },
    {
      "epoch": 0.173960436011304,
      "grad_norm": 127.78780364990234,
      "learning_rate": 1.8933042051430222e-05,
      "loss": 3.9906,
      "step": 4309
    },
    {
      "epoch": 0.17400080742834073,
      "grad_norm": 83.98343658447266,
      "learning_rate": 1.893245427971915e-05,
      "loss": 4.1255,
      "step": 4310
    },
    {
      "epoch": 0.17404117884537748,
      "grad_norm": 179.68955993652344,
      "learning_rate": 1.8931866355283963e-05,
      "loss": 6.3659,
      "step": 4311
    },
    {
      "epoch": 0.1740815502624142,
      "grad_norm": 103.07471466064453,
      "learning_rate": 1.8931278278134723e-05,
      "loss": 5.3209,
      "step": 4312
    },
    {
      "epoch": 0.17412192167945095,
      "grad_norm": 238.84881591796875,
      "learning_rate": 1.893069004828148e-05,
      "loss": 6.1459,
      "step": 4313
    },
    {
      "epoch": 0.17416229309648767,
      "grad_norm": 84.2608642578125,
      "learning_rate": 1.893010166573429e-05,
      "loss": 6.7804,
      "step": 4314
    },
    {
      "epoch": 0.17420266451352442,
      "grad_norm": 194.1399688720703,
      "learning_rate": 1.8929513130503214e-05,
      "loss": 5.7145,
      "step": 4315
    },
    {
      "epoch": 0.17424303593056117,
      "grad_norm": 136.68179321289062,
      "learning_rate": 1.8928924442598317e-05,
      "loss": 5.6546,
      "step": 4316
    },
    {
      "epoch": 0.1742834073475979,
      "grad_norm": 137.171630859375,
      "learning_rate": 1.8928335602029662e-05,
      "loss": 5.2695,
      "step": 4317
    },
    {
      "epoch": 0.17432377876463465,
      "grad_norm": 93.1363525390625,
      "learning_rate": 1.8927746608807317e-05,
      "loss": 6.7131,
      "step": 4318
    },
    {
      "epoch": 0.17436415018167137,
      "grad_norm": 99.93563842773438,
      "learning_rate": 1.892715746294135e-05,
      "loss": 6.3893,
      "step": 4319
    },
    {
      "epoch": 0.17440452159870812,
      "grad_norm": 112.11433410644531,
      "learning_rate": 1.8926568164441843e-05,
      "loss": 5.9022,
      "step": 4320
    },
    {
      "epoch": 0.17444489301574484,
      "grad_norm": 209.1799774169922,
      "learning_rate": 1.8925978713318864e-05,
      "loss": 4.7022,
      "step": 4321
    },
    {
      "epoch": 0.1744852644327816,
      "grad_norm": 126.39643096923828,
      "learning_rate": 1.892538910958249e-05,
      "loss": 6.0878,
      "step": 4322
    },
    {
      "epoch": 0.17452563584981834,
      "grad_norm": 151.9305419921875,
      "learning_rate": 1.8924799353242808e-05,
      "loss": 4.8978,
      "step": 4323
    },
    {
      "epoch": 0.17456600726685506,
      "grad_norm": 73.51664733886719,
      "learning_rate": 1.892420944430989e-05,
      "loss": 5.7671,
      "step": 4324
    },
    {
      "epoch": 0.1746063786838918,
      "grad_norm": 121.8717269897461,
      "learning_rate": 1.8923619382793838e-05,
      "loss": 6.6212,
      "step": 4325
    },
    {
      "epoch": 0.17464675010092853,
      "grad_norm": 77.9253921508789,
      "learning_rate": 1.892302916870473e-05,
      "loss": 5.2026,
      "step": 4326
    },
    {
      "epoch": 0.17468712151796528,
      "grad_norm": 141.2133331298828,
      "learning_rate": 1.892243880205266e-05,
      "loss": 6.3587,
      "step": 4327
    },
    {
      "epoch": 0.17472749293500203,
      "grad_norm": 225.69204711914062,
      "learning_rate": 1.8921848282847724e-05,
      "loss": 5.1854,
      "step": 4328
    },
    {
      "epoch": 0.17476786435203875,
      "grad_norm": 82.07688903808594,
      "learning_rate": 1.8921257611100014e-05,
      "loss": 6.9705,
      "step": 4329
    },
    {
      "epoch": 0.1748082357690755,
      "grad_norm": 170.2679443359375,
      "learning_rate": 1.8920666786819632e-05,
      "loss": 6.7027,
      "step": 4330
    },
    {
      "epoch": 0.17484860718611223,
      "grad_norm": 127.25004577636719,
      "learning_rate": 1.892007581001668e-05,
      "loss": 5.3603,
      "step": 4331
    },
    {
      "epoch": 0.17488897860314898,
      "grad_norm": 115.24639129638672,
      "learning_rate": 1.891948468070126e-05,
      "loss": 5.2388,
      "step": 4332
    },
    {
      "epoch": 0.1749293500201857,
      "grad_norm": 91.51602172851562,
      "learning_rate": 1.891889339888348e-05,
      "loss": 6.2433,
      "step": 4333
    },
    {
      "epoch": 0.17496972143722245,
      "grad_norm": 120.0892333984375,
      "learning_rate": 1.891830196457345e-05,
      "loss": 5.8635,
      "step": 4334
    },
    {
      "epoch": 0.1750100928542592,
      "grad_norm": 72.91862487792969,
      "learning_rate": 1.8917710377781286e-05,
      "loss": 6.8291,
      "step": 4335
    },
    {
      "epoch": 0.17505046427129592,
      "grad_norm": 93.6216049194336,
      "learning_rate": 1.8917118638517093e-05,
      "loss": 6.1821,
      "step": 4336
    },
    {
      "epoch": 0.17509083568833267,
      "grad_norm": 128.1939239501953,
      "learning_rate": 1.8916526746791e-05,
      "loss": 6.3393,
      "step": 4337
    },
    {
      "epoch": 0.1751312071053694,
      "grad_norm": 103.54537200927734,
      "learning_rate": 1.8915934702613118e-05,
      "loss": 7.0722,
      "step": 4338
    },
    {
      "epoch": 0.17517157852240614,
      "grad_norm": 113.19868469238281,
      "learning_rate": 1.8915342505993576e-05,
      "loss": 6.3048,
      "step": 4339
    },
    {
      "epoch": 0.17521194993944286,
      "grad_norm": 101.2662353515625,
      "learning_rate": 1.891475015694249e-05,
      "loss": 4.6908,
      "step": 4340
    },
    {
      "epoch": 0.1752523213564796,
      "grad_norm": 75.81938171386719,
      "learning_rate": 1.8914157655469997e-05,
      "loss": 5.196,
      "step": 4341
    },
    {
      "epoch": 0.17529269277351636,
      "grad_norm": 117.41372680664062,
      "learning_rate": 1.8913565001586227e-05,
      "loss": 6.4389,
      "step": 4342
    },
    {
      "epoch": 0.17533306419055308,
      "grad_norm": 111.79692840576172,
      "learning_rate": 1.8912972195301305e-05,
      "loss": 5.013,
      "step": 4343
    },
    {
      "epoch": 0.17537343560758983,
      "grad_norm": 97.43540954589844,
      "learning_rate": 1.8912379236625377e-05,
      "loss": 5.9071,
      "step": 4344
    },
    {
      "epoch": 0.17541380702462656,
      "grad_norm": 76.14598846435547,
      "learning_rate": 1.891178612556857e-05,
      "loss": 5.254,
      "step": 4345
    },
    {
      "epoch": 0.1754541784416633,
      "grad_norm": 112.52153778076172,
      "learning_rate": 1.8911192862141034e-05,
      "loss": 5.6492,
      "step": 4346
    },
    {
      "epoch": 0.17549454985870003,
      "grad_norm": 128.30140686035156,
      "learning_rate": 1.891059944635291e-05,
      "loss": 5.0044,
      "step": 4347
    },
    {
      "epoch": 0.17553492127573678,
      "grad_norm": 90.2647705078125,
      "learning_rate": 1.8910005878214345e-05,
      "loss": 5.7369,
      "step": 4348
    },
    {
      "epoch": 0.17557529269277353,
      "grad_norm": 85.2386474609375,
      "learning_rate": 1.890941215773548e-05,
      "loss": 5.9377,
      "step": 4349
    },
    {
      "epoch": 0.17561566410981025,
      "grad_norm": 95.71202087402344,
      "learning_rate": 1.8908818284926474e-05,
      "loss": 4.6694,
      "step": 4350
    },
    {
      "epoch": 0.175656035526847,
      "grad_norm": 136.58367919921875,
      "learning_rate": 1.8908224259797482e-05,
      "loss": 5.6306,
      "step": 4351
    },
    {
      "epoch": 0.17569640694388372,
      "grad_norm": 98.28880310058594,
      "learning_rate": 1.8907630082358657e-05,
      "loss": 5.5689,
      "step": 4352
    },
    {
      "epoch": 0.17573677836092047,
      "grad_norm": 132.231689453125,
      "learning_rate": 1.8907035752620154e-05,
      "loss": 5.3446,
      "step": 4353
    },
    {
      "epoch": 0.17577714977795722,
      "grad_norm": 109.06616973876953,
      "learning_rate": 1.8906441270592144e-05,
      "loss": 6.6638,
      "step": 4354
    },
    {
      "epoch": 0.17581752119499394,
      "grad_norm": 108.9760971069336,
      "learning_rate": 1.8905846636284784e-05,
      "loss": 3.6037,
      "step": 4355
    },
    {
      "epoch": 0.1758578926120307,
      "grad_norm": 140.22897338867188,
      "learning_rate": 1.8905251849708243e-05,
      "loss": 5.4935,
      "step": 4356
    },
    {
      "epoch": 0.1758982640290674,
      "grad_norm": 130.37283325195312,
      "learning_rate": 1.890465691087269e-05,
      "loss": 5.4631,
      "step": 4357
    },
    {
      "epoch": 0.17593863544610416,
      "grad_norm": 150.518310546875,
      "learning_rate": 1.89040618197883e-05,
      "loss": 6.8956,
      "step": 4358
    },
    {
      "epoch": 0.17597900686314089,
      "grad_norm": 131.55589294433594,
      "learning_rate": 1.8903466576465247e-05,
      "loss": 3.6356,
      "step": 4359
    },
    {
      "epoch": 0.17601937828017764,
      "grad_norm": 73.15027618408203,
      "learning_rate": 1.89028711809137e-05,
      "loss": 5.6489,
      "step": 4360
    },
    {
      "epoch": 0.17605974969721438,
      "grad_norm": 107.19941711425781,
      "learning_rate": 1.890227563314385e-05,
      "loss": 6.6693,
      "step": 4361
    },
    {
      "epoch": 0.1761001211142511,
      "grad_norm": 129.35699462890625,
      "learning_rate": 1.8901679933165876e-05,
      "loss": 6.4859,
      "step": 4362
    },
    {
      "epoch": 0.17614049253128786,
      "grad_norm": 90.0306167602539,
      "learning_rate": 1.890108408098996e-05,
      "loss": 5.6601,
      "step": 4363
    },
    {
      "epoch": 0.17618086394832458,
      "grad_norm": 111.11522674560547,
      "learning_rate": 1.890048807662629e-05,
      "loss": 4.3342,
      "step": 4364
    },
    {
      "epoch": 0.17622123536536133,
      "grad_norm": 89.20857238769531,
      "learning_rate": 1.889989192008506e-05,
      "loss": 4.252,
      "step": 4365
    },
    {
      "epoch": 0.17626160678239805,
      "grad_norm": 112.10601806640625,
      "learning_rate": 1.8899295611376462e-05,
      "loss": 5.2879,
      "step": 4366
    },
    {
      "epoch": 0.1763019781994348,
      "grad_norm": 116.02952575683594,
      "learning_rate": 1.889869915051069e-05,
      "loss": 5.42,
      "step": 4367
    },
    {
      "epoch": 0.17634234961647155,
      "grad_norm": 160.78173828125,
      "learning_rate": 1.8898102537497943e-05,
      "loss": 4.0261,
      "step": 4368
    },
    {
      "epoch": 0.17638272103350827,
      "grad_norm": 151.0635223388672,
      "learning_rate": 1.889750577234842e-05,
      "loss": 5.2056,
      "step": 4369
    },
    {
      "epoch": 0.17642309245054502,
      "grad_norm": 89.10269165039062,
      "learning_rate": 1.8896908855072324e-05,
      "loss": 4.8354,
      "step": 4370
    },
    {
      "epoch": 0.17646346386758174,
      "grad_norm": 96.8393783569336,
      "learning_rate": 1.8896311785679865e-05,
      "loss": 5.5537,
      "step": 4371
    },
    {
      "epoch": 0.1765038352846185,
      "grad_norm": 85.42033386230469,
      "learning_rate": 1.8895714564181247e-05,
      "loss": 6.4974,
      "step": 4372
    },
    {
      "epoch": 0.17654420670165522,
      "grad_norm": 173.73458862304688,
      "learning_rate": 1.8895117190586683e-05,
      "loss": 5.8037,
      "step": 4373
    },
    {
      "epoch": 0.17658457811869196,
      "grad_norm": 82.53240966796875,
      "learning_rate": 1.8894519664906388e-05,
      "loss": 6.9345,
      "step": 4374
    },
    {
      "epoch": 0.17662494953572871,
      "grad_norm": 72.53954315185547,
      "learning_rate": 1.8893921987150577e-05,
      "loss": 4.2685,
      "step": 4375
    },
    {
      "epoch": 0.17666532095276544,
      "grad_norm": 71.6419448852539,
      "learning_rate": 1.8893324157329468e-05,
      "loss": 5.8867,
      "step": 4376
    },
    {
      "epoch": 0.1767056923698022,
      "grad_norm": 100.55894470214844,
      "learning_rate": 1.889272617545328e-05,
      "loss": 6.3811,
      "step": 4377
    },
    {
      "epoch": 0.1767460637868389,
      "grad_norm": 89.49108123779297,
      "learning_rate": 1.8892128041532246e-05,
      "loss": 6.0303,
      "step": 4378
    },
    {
      "epoch": 0.17678643520387566,
      "grad_norm": 91.38414001464844,
      "learning_rate": 1.8891529755576584e-05,
      "loss": 4.5689,
      "step": 4379
    },
    {
      "epoch": 0.1768268066209124,
      "grad_norm": 100.45668029785156,
      "learning_rate": 1.889093131759653e-05,
      "loss": 5.9183,
      "step": 4380
    },
    {
      "epoch": 0.17686717803794913,
      "grad_norm": 85.58827209472656,
      "learning_rate": 1.8890332727602307e-05,
      "loss": 4.9177,
      "step": 4381
    },
    {
      "epoch": 0.17690754945498588,
      "grad_norm": 96.21332550048828,
      "learning_rate": 1.888973398560416e-05,
      "loss": 5.5704,
      "step": 4382
    },
    {
      "epoch": 0.1769479208720226,
      "grad_norm": 106.64968872070312,
      "learning_rate": 1.8889135091612316e-05,
      "loss": 5.8376,
      "step": 4383
    },
    {
      "epoch": 0.17698829228905935,
      "grad_norm": 95.88514709472656,
      "learning_rate": 1.8888536045637023e-05,
      "loss": 5.2298,
      "step": 4384
    },
    {
      "epoch": 0.17702866370609607,
      "grad_norm": 388.1346740722656,
      "learning_rate": 1.8887936847688518e-05,
      "loss": 5.4886,
      "step": 4385
    },
    {
      "epoch": 0.17706903512313282,
      "grad_norm": 234.27561950683594,
      "learning_rate": 1.8887337497777047e-05,
      "loss": 7.1703,
      "step": 4386
    },
    {
      "epoch": 0.17710940654016957,
      "grad_norm": 118.42046356201172,
      "learning_rate": 1.888673799591286e-05,
      "loss": 6.0839,
      "step": 4387
    },
    {
      "epoch": 0.1771497779572063,
      "grad_norm": 148.50108337402344,
      "learning_rate": 1.8886138342106205e-05,
      "loss": 5.5126,
      "step": 4388
    },
    {
      "epoch": 0.17719014937424304,
      "grad_norm": 122.16866302490234,
      "learning_rate": 1.8885538536367335e-05,
      "loss": 5.9019,
      "step": 4389
    },
    {
      "epoch": 0.17723052079127977,
      "grad_norm": 153.51803588867188,
      "learning_rate": 1.8884938578706506e-05,
      "loss": 5.4978,
      "step": 4390
    },
    {
      "epoch": 0.17727089220831652,
      "grad_norm": 189.53414916992188,
      "learning_rate": 1.8884338469133974e-05,
      "loss": 4.9355,
      "step": 4391
    },
    {
      "epoch": 0.17731126362535324,
      "grad_norm": 144.73568725585938,
      "learning_rate": 1.888373820766e-05,
      "loss": 5.838,
      "step": 4392
    },
    {
      "epoch": 0.17735163504239,
      "grad_norm": 114.00287628173828,
      "learning_rate": 1.8883137794294846e-05,
      "loss": 4.7502,
      "step": 4393
    },
    {
      "epoch": 0.17739200645942674,
      "grad_norm": 74.44725036621094,
      "learning_rate": 1.8882537229048783e-05,
      "loss": 6.5599,
      "step": 4394
    },
    {
      "epoch": 0.17743237787646346,
      "grad_norm": 138.12924194335938,
      "learning_rate": 1.8881936511932078e-05,
      "loss": 5.8247,
      "step": 4395
    },
    {
      "epoch": 0.1774727492935002,
      "grad_norm": 82.49182891845703,
      "learning_rate": 1.8881335642954993e-05,
      "loss": 5.6045,
      "step": 4396
    },
    {
      "epoch": 0.17751312071053693,
      "grad_norm": 137.55299377441406,
      "learning_rate": 1.888073462212781e-05,
      "loss": 3.3912,
      "step": 4397
    },
    {
      "epoch": 0.17755349212757368,
      "grad_norm": 141.84124755859375,
      "learning_rate": 1.8880133449460804e-05,
      "loss": 6.1437,
      "step": 4398
    },
    {
      "epoch": 0.1775938635446104,
      "grad_norm": 81.22554016113281,
      "learning_rate": 1.8879532124964253e-05,
      "loss": 4.5372,
      "step": 4399
    },
    {
      "epoch": 0.17763423496164715,
      "grad_norm": 68.5572509765625,
      "learning_rate": 1.8878930648648438e-05,
      "loss": 6.0312,
      "step": 4400
    },
    {
      "epoch": 0.1776746063786839,
      "grad_norm": 93.19141387939453,
      "learning_rate": 1.887832902052364e-05,
      "loss": 4.8251,
      "step": 4401
    },
    {
      "epoch": 0.17771497779572062,
      "grad_norm": 118.00349426269531,
      "learning_rate": 1.8877727240600154e-05,
      "loss": 6.3043,
      "step": 4402
    },
    {
      "epoch": 0.17775534921275737,
      "grad_norm": 120.33145141601562,
      "learning_rate": 1.8877125308888255e-05,
      "loss": 5.2943,
      "step": 4403
    },
    {
      "epoch": 0.1777957206297941,
      "grad_norm": 84.44216918945312,
      "learning_rate": 1.887652322539825e-05,
      "loss": 5.1065,
      "step": 4404
    },
    {
      "epoch": 0.17783609204683085,
      "grad_norm": 88.32903289794922,
      "learning_rate": 1.8875920990140422e-05,
      "loss": 7.081,
      "step": 4405
    },
    {
      "epoch": 0.17787646346386757,
      "grad_norm": 91.67086029052734,
      "learning_rate": 1.887531860312507e-05,
      "loss": 5.1936,
      "step": 4406
    },
    {
      "epoch": 0.17791683488090432,
      "grad_norm": 94.27228546142578,
      "learning_rate": 1.88747160643625e-05,
      "loss": 5.0653,
      "step": 4407
    },
    {
      "epoch": 0.17795720629794107,
      "grad_norm": 136.81544494628906,
      "learning_rate": 1.8874113373863008e-05,
      "loss": 3.8883,
      "step": 4408
    },
    {
      "epoch": 0.1779975777149778,
      "grad_norm": 66.86424255371094,
      "learning_rate": 1.8873510531636898e-05,
      "loss": 6.1704,
      "step": 4409
    },
    {
      "epoch": 0.17803794913201454,
      "grad_norm": 83.96041107177734,
      "learning_rate": 1.887290753769448e-05,
      "loss": 5.7978,
      "step": 4410
    },
    {
      "epoch": 0.17807832054905126,
      "grad_norm": 122.00092315673828,
      "learning_rate": 1.8872304392046062e-05,
      "loss": 5.9384,
      "step": 4411
    },
    {
      "epoch": 0.178118691966088,
      "grad_norm": 86.45323944091797,
      "learning_rate": 1.887170109470196e-05,
      "loss": 5.6262,
      "step": 4412
    },
    {
      "epoch": 0.17815906338312476,
      "grad_norm": 121.22579193115234,
      "learning_rate": 1.8871097645672485e-05,
      "loss": 4.9013,
      "step": 4413
    },
    {
      "epoch": 0.17819943480016148,
      "grad_norm": 246.10714721679688,
      "learning_rate": 1.8870494044967955e-05,
      "loss": 7.21,
      "step": 4414
    },
    {
      "epoch": 0.17823980621719823,
      "grad_norm": 107.20317077636719,
      "learning_rate": 1.886989029259869e-05,
      "loss": 4.3196,
      "step": 4415
    },
    {
      "epoch": 0.17828017763423495,
      "grad_norm": 107.11157989501953,
      "learning_rate": 1.8869286388575017e-05,
      "loss": 4.972,
      "step": 4416
    },
    {
      "epoch": 0.1783205490512717,
      "grad_norm": 74.50264739990234,
      "learning_rate": 1.8868682332907257e-05,
      "loss": 4.034,
      "step": 4417
    },
    {
      "epoch": 0.17836092046830843,
      "grad_norm": 126.93748474121094,
      "learning_rate": 1.8868078125605736e-05,
      "loss": 5.5581,
      "step": 4418
    },
    {
      "epoch": 0.17840129188534518,
      "grad_norm": 108.2729263305664,
      "learning_rate": 1.886747376668079e-05,
      "loss": 5.4997,
      "step": 4419
    },
    {
      "epoch": 0.17844166330238193,
      "grad_norm": 159.34002685546875,
      "learning_rate": 1.886686925614275e-05,
      "loss": 5.1334,
      "step": 4420
    },
    {
      "epoch": 0.17848203471941865,
      "grad_norm": 100.7432861328125,
      "learning_rate": 1.886626459400195e-05,
      "loss": 4.2227,
      "step": 4421
    },
    {
      "epoch": 0.1785224061364554,
      "grad_norm": 141.25543212890625,
      "learning_rate": 1.886565978026873e-05,
      "loss": 5.91,
      "step": 4422
    },
    {
      "epoch": 0.17856277755349212,
      "grad_norm": 105.873779296875,
      "learning_rate": 1.886505481495343e-05,
      "loss": 3.9463,
      "step": 4423
    },
    {
      "epoch": 0.17860314897052887,
      "grad_norm": 160.2205352783203,
      "learning_rate": 1.8864449698066397e-05,
      "loss": 4.6167,
      "step": 4424
    },
    {
      "epoch": 0.1786435203875656,
      "grad_norm": 148.60630798339844,
      "learning_rate": 1.8863844429617968e-05,
      "loss": 5.7618,
      "step": 4425
    },
    {
      "epoch": 0.17868389180460234,
      "grad_norm": 95.11199951171875,
      "learning_rate": 1.8863239009618502e-05,
      "loss": 6.0312,
      "step": 4426
    },
    {
      "epoch": 0.1787242632216391,
      "grad_norm": 114.12809753417969,
      "learning_rate": 1.8862633438078347e-05,
      "loss": 6.3515,
      "step": 4427
    },
    {
      "epoch": 0.1787646346386758,
      "grad_norm": 97.59008026123047,
      "learning_rate": 1.8862027715007857e-05,
      "loss": 5.7353,
      "step": 4428
    },
    {
      "epoch": 0.17880500605571256,
      "grad_norm": 159.8184814453125,
      "learning_rate": 1.8861421840417382e-05,
      "loss": 5.7326,
      "step": 4429
    },
    {
      "epoch": 0.17884537747274928,
      "grad_norm": 113.74858093261719,
      "learning_rate": 1.886081581431729e-05,
      "loss": 5.8467,
      "step": 4430
    },
    {
      "epoch": 0.17888574888978603,
      "grad_norm": 99.7320327758789,
      "learning_rate": 1.886020963671794e-05,
      "loss": 4.8283,
      "step": 4431
    },
    {
      "epoch": 0.17892612030682276,
      "grad_norm": 65.5793228149414,
      "learning_rate": 1.885960330762969e-05,
      "loss": 5.2951,
      "step": 4432
    },
    {
      "epoch": 0.1789664917238595,
      "grad_norm": 153.36317443847656,
      "learning_rate": 1.8858996827062918e-05,
      "loss": 4.0483,
      "step": 4433
    },
    {
      "epoch": 0.17900686314089626,
      "grad_norm": 80.16141510009766,
      "learning_rate": 1.8858390195027986e-05,
      "loss": 5.1683,
      "step": 4434
    },
    {
      "epoch": 0.17904723455793298,
      "grad_norm": 72.32886505126953,
      "learning_rate": 1.8857783411535265e-05,
      "loss": 3.9403,
      "step": 4435
    },
    {
      "epoch": 0.17908760597496973,
      "grad_norm": 118.43769836425781,
      "learning_rate": 1.8857176476595133e-05,
      "loss": 4.8524,
      "step": 4436
    },
    {
      "epoch": 0.17912797739200645,
      "grad_norm": 89.24971008300781,
      "learning_rate": 1.885656939021797e-05,
      "loss": 5.0582,
      "step": 4437
    },
    {
      "epoch": 0.1791683488090432,
      "grad_norm": 228.106689453125,
      "learning_rate": 1.885596215241415e-05,
      "loss": 5.86,
      "step": 4438
    },
    {
      "epoch": 0.17920872022607995,
      "grad_norm": 81.9186019897461,
      "learning_rate": 1.8855354763194053e-05,
      "loss": 5.538,
      "step": 4439
    },
    {
      "epoch": 0.17924909164311667,
      "grad_norm": 81.92935943603516,
      "learning_rate": 1.8854747222568074e-05,
      "loss": 4.0843,
      "step": 4440
    },
    {
      "epoch": 0.17928946306015342,
      "grad_norm": 110.66150665283203,
      "learning_rate": 1.885413953054659e-05,
      "loss": 5.6086,
      "step": 4441
    },
    {
      "epoch": 0.17932983447719014,
      "grad_norm": 100.43522644042969,
      "learning_rate": 1.885353168714e-05,
      "loss": 5.1023,
      "step": 4442
    },
    {
      "epoch": 0.1793702058942269,
      "grad_norm": 140.85289001464844,
      "learning_rate": 1.8852923692358693e-05,
      "loss": 7.257,
      "step": 4443
    },
    {
      "epoch": 0.1794105773112636,
      "grad_norm": 89.76032257080078,
      "learning_rate": 1.885231554621306e-05,
      "loss": 6.0,
      "step": 4444
    },
    {
      "epoch": 0.17945094872830036,
      "grad_norm": 115.17117309570312,
      "learning_rate": 1.8851707248713505e-05,
      "loss": 5.903,
      "step": 4445
    },
    {
      "epoch": 0.1794913201453371,
      "grad_norm": 76.9775390625,
      "learning_rate": 1.8851098799870428e-05,
      "loss": 4.7427,
      "step": 4446
    },
    {
      "epoch": 0.17953169156237384,
      "grad_norm": 73.22766876220703,
      "learning_rate": 1.8850490199694226e-05,
      "loss": 5.6298,
      "step": 4447
    },
    {
      "epoch": 0.17957206297941058,
      "grad_norm": 153.73846435546875,
      "learning_rate": 1.8849881448195316e-05,
      "loss": 7.8195,
      "step": 4448
    },
    {
      "epoch": 0.1796124343964473,
      "grad_norm": 152.6529083251953,
      "learning_rate": 1.8849272545384094e-05,
      "loss": 4.8347,
      "step": 4449
    },
    {
      "epoch": 0.17965280581348406,
      "grad_norm": 124.18060302734375,
      "learning_rate": 1.8848663491270975e-05,
      "loss": 6.1975,
      "step": 4450
    },
    {
      "epoch": 0.17969317723052078,
      "grad_norm": 108.05548858642578,
      "learning_rate": 1.8848054285866377e-05,
      "loss": 5.2417,
      "step": 4451
    },
    {
      "epoch": 0.17973354864755753,
      "grad_norm": 80.8581314086914,
      "learning_rate": 1.884744492918071e-05,
      "loss": 6.2968,
      "step": 4452
    },
    {
      "epoch": 0.17977392006459428,
      "grad_norm": 93.52590942382812,
      "learning_rate": 1.8846835421224395e-05,
      "loss": 4.9711,
      "step": 4453
    },
    {
      "epoch": 0.179814291481631,
      "grad_norm": 113.71490478515625,
      "learning_rate": 1.8846225762007852e-05,
      "loss": 4.5582,
      "step": 4454
    },
    {
      "epoch": 0.17985466289866775,
      "grad_norm": 74.62390899658203,
      "learning_rate": 1.8845615951541508e-05,
      "loss": 5.529,
      "step": 4455
    },
    {
      "epoch": 0.17989503431570447,
      "grad_norm": 74.16822052001953,
      "learning_rate": 1.8845005989835787e-05,
      "loss": 4.3613,
      "step": 4456
    },
    {
      "epoch": 0.17993540573274122,
      "grad_norm": 88.5512924194336,
      "learning_rate": 1.884439587690112e-05,
      "loss": 5.4144,
      "step": 4457
    },
    {
      "epoch": 0.17997577714977794,
      "grad_norm": 78.23040771484375,
      "learning_rate": 1.8843785612747933e-05,
      "loss": 5.3729,
      "step": 4458
    },
    {
      "epoch": 0.1800161485668147,
      "grad_norm": 131.5013427734375,
      "learning_rate": 1.8843175197386667e-05,
      "loss": 3.768,
      "step": 4459
    },
    {
      "epoch": 0.18005651998385144,
      "grad_norm": 211.2291717529297,
      "learning_rate": 1.884256463082775e-05,
      "loss": 6.5182,
      "step": 4460
    },
    {
      "epoch": 0.18009689140088816,
      "grad_norm": 122.73101043701172,
      "learning_rate": 1.8841953913081634e-05,
      "loss": 3.0766,
      "step": 4461
    },
    {
      "epoch": 0.18013726281792491,
      "grad_norm": 160.0671844482422,
      "learning_rate": 1.8841343044158748e-05,
      "loss": 5.0735,
      "step": 4462
    },
    {
      "epoch": 0.18017763423496164,
      "grad_norm": 146.03492736816406,
      "learning_rate": 1.8840732024069544e-05,
      "loss": 5.5035,
      "step": 4463
    },
    {
      "epoch": 0.1802180056519984,
      "grad_norm": 90.28909301757812,
      "learning_rate": 1.8840120852824468e-05,
      "loss": 4.6891,
      "step": 4464
    },
    {
      "epoch": 0.18025837706903514,
      "grad_norm": 89.01673126220703,
      "learning_rate": 1.883950953043397e-05,
      "loss": 4.7759,
      "step": 4465
    },
    {
      "epoch": 0.18029874848607186,
      "grad_norm": 123.09638977050781,
      "learning_rate": 1.8838898056908496e-05,
      "loss": 6.132,
      "step": 4466
    },
    {
      "epoch": 0.1803391199031086,
      "grad_norm": 139.55429077148438,
      "learning_rate": 1.8838286432258508e-05,
      "loss": 5.1549,
      "step": 4467
    },
    {
      "epoch": 0.18037949132014533,
      "grad_norm": 100.36237335205078,
      "learning_rate": 1.883767465649446e-05,
      "loss": 5.7518,
      "step": 4468
    },
    {
      "epoch": 0.18041986273718208,
      "grad_norm": 333.47674560546875,
      "learning_rate": 1.8837062729626814e-05,
      "loss": 6.9936,
      "step": 4469
    },
    {
      "epoch": 0.1804602341542188,
      "grad_norm": 121.43849182128906,
      "learning_rate": 1.883645065166603e-05,
      "loss": 7.016,
      "step": 4470
    },
    {
      "epoch": 0.18050060557125555,
      "grad_norm": 121.38639831542969,
      "learning_rate": 1.8835838422622577e-05,
      "loss": 6.4839,
      "step": 4471
    },
    {
      "epoch": 0.1805409769882923,
      "grad_norm": 101.2248306274414,
      "learning_rate": 1.883522604250692e-05,
      "loss": 5.1063,
      "step": 4472
    },
    {
      "epoch": 0.18058134840532902,
      "grad_norm": 93.49028778076172,
      "learning_rate": 1.8834613511329524e-05,
      "loss": 6.1625,
      "step": 4473
    },
    {
      "epoch": 0.18062171982236577,
      "grad_norm": 79.80825805664062,
      "learning_rate": 1.8834000829100873e-05,
      "loss": 6.0757,
      "step": 4474
    },
    {
      "epoch": 0.1806620912394025,
      "grad_norm": 80.66144561767578,
      "learning_rate": 1.8833387995831435e-05,
      "loss": 6.5418,
      "step": 4475
    },
    {
      "epoch": 0.18070246265643924,
      "grad_norm": 95.54918670654297,
      "learning_rate": 1.883277501153169e-05,
      "loss": 7.2664,
      "step": 4476
    },
    {
      "epoch": 0.18074283407347597,
      "grad_norm": 89.80803680419922,
      "learning_rate": 1.8832161876212117e-05,
      "loss": 4.8657,
      "step": 4477
    },
    {
      "epoch": 0.18078320549051272,
      "grad_norm": 83.63655090332031,
      "learning_rate": 1.88315485898832e-05,
      "loss": 7.0782,
      "step": 4478
    },
    {
      "epoch": 0.18082357690754947,
      "grad_norm": 91.28623962402344,
      "learning_rate": 1.8830935152555427e-05,
      "loss": 5.5776,
      "step": 4479
    },
    {
      "epoch": 0.1808639483245862,
      "grad_norm": 310.4918212890625,
      "learning_rate": 1.8830321564239284e-05,
      "loss": 5.7844,
      "step": 4480
    },
    {
      "epoch": 0.18090431974162294,
      "grad_norm": 104.84236907958984,
      "learning_rate": 1.8829707824945265e-05,
      "loss": 4.1929,
      "step": 4481
    },
    {
      "epoch": 0.18094469115865966,
      "grad_norm": 123.40275573730469,
      "learning_rate": 1.882909393468386e-05,
      "loss": 5.8209,
      "step": 4482
    },
    {
      "epoch": 0.1809850625756964,
      "grad_norm": 93.07273864746094,
      "learning_rate": 1.8828479893465562e-05,
      "loss": 7.0469,
      "step": 4483
    },
    {
      "epoch": 0.18102543399273313,
      "grad_norm": 104.39033508300781,
      "learning_rate": 1.8827865701300875e-05,
      "loss": 3.6959,
      "step": 4484
    },
    {
      "epoch": 0.18106580540976988,
      "grad_norm": 74.88325500488281,
      "learning_rate": 1.8827251358200304e-05,
      "loss": 5.4995,
      "step": 4485
    },
    {
      "epoch": 0.18110617682680663,
      "grad_norm": 134.93618774414062,
      "learning_rate": 1.8826636864174345e-05,
      "loss": 7.3921,
      "step": 4486
    },
    {
      "epoch": 0.18114654824384335,
      "grad_norm": 143.16781616210938,
      "learning_rate": 1.8826022219233505e-05,
      "loss": 4.4337,
      "step": 4487
    },
    {
      "epoch": 0.1811869196608801,
      "grad_norm": 139.50418090820312,
      "learning_rate": 1.8825407423388297e-05,
      "loss": 5.8288,
      "step": 4488
    },
    {
      "epoch": 0.18122729107791682,
      "grad_norm": 83.9513931274414,
      "learning_rate": 1.8824792476649232e-05,
      "loss": 6.9773,
      "step": 4489
    },
    {
      "epoch": 0.18126766249495357,
      "grad_norm": 97.55541229248047,
      "learning_rate": 1.8824177379026822e-05,
      "loss": 5.1193,
      "step": 4490
    },
    {
      "epoch": 0.18130803391199032,
      "grad_norm": 82.30827331542969,
      "learning_rate": 1.8823562130531585e-05,
      "loss": 5.7657,
      "step": 4491
    },
    {
      "epoch": 0.18134840532902705,
      "grad_norm": 76.009521484375,
      "learning_rate": 1.882294673117404e-05,
      "loss": 6.414,
      "step": 4492
    },
    {
      "epoch": 0.1813887767460638,
      "grad_norm": 111.17132568359375,
      "learning_rate": 1.882233118096471e-05,
      "loss": 6.1006,
      "step": 4493
    },
    {
      "epoch": 0.18142914816310052,
      "grad_norm": 80.2989501953125,
      "learning_rate": 1.8821715479914117e-05,
      "loss": 6.3655,
      "step": 4494
    },
    {
      "epoch": 0.18146951958013727,
      "grad_norm": 124.02296447753906,
      "learning_rate": 1.8821099628032788e-05,
      "loss": 4.7848,
      "step": 4495
    },
    {
      "epoch": 0.181509890997174,
      "grad_norm": 90.29515075683594,
      "learning_rate": 1.8820483625331256e-05,
      "loss": 6.1199,
      "step": 4496
    },
    {
      "epoch": 0.18155026241421074,
      "grad_norm": 93.01142120361328,
      "learning_rate": 1.881986747182005e-05,
      "loss": 5.6194,
      "step": 4497
    },
    {
      "epoch": 0.1815906338312475,
      "grad_norm": 114.77101135253906,
      "learning_rate": 1.881925116750971e-05,
      "loss": 5.5506,
      "step": 4498
    },
    {
      "epoch": 0.1816310052482842,
      "grad_norm": 91.68521881103516,
      "learning_rate": 1.8818634712410765e-05,
      "loss": 5.4082,
      "step": 4499
    },
    {
      "epoch": 0.18167137666532096,
      "grad_norm": 105.91889953613281,
      "learning_rate": 1.8818018106533764e-05,
      "loss": 4.5541,
      "step": 4500
    },
    {
      "epoch": 0.18171174808235768,
      "grad_norm": 59.757076263427734,
      "learning_rate": 1.8817401349889242e-05,
      "loss": 3.8418,
      "step": 4501
    },
    {
      "epoch": 0.18175211949939443,
      "grad_norm": 117.97687530517578,
      "learning_rate": 1.8816784442487747e-05,
      "loss": 5.4241,
      "step": 4502
    },
    {
      "epoch": 0.18179249091643115,
      "grad_norm": 92.6259994506836,
      "learning_rate": 1.881616738433983e-05,
      "loss": 5.0647,
      "step": 4503
    },
    {
      "epoch": 0.1818328623334679,
      "grad_norm": 118.64775085449219,
      "learning_rate": 1.881555017545603e-05,
      "loss": 5.7001,
      "step": 4504
    },
    {
      "epoch": 0.18187323375050465,
      "grad_norm": 113.04720306396484,
      "learning_rate": 1.8814932815846915e-05,
      "loss": 3.6668,
      "step": 4505
    },
    {
      "epoch": 0.18191360516754138,
      "grad_norm": 93.13751220703125,
      "learning_rate": 1.881431530552303e-05,
      "loss": 5.0572,
      "step": 4506
    },
    {
      "epoch": 0.18195397658457813,
      "grad_norm": 76.44294738769531,
      "learning_rate": 1.8813697644494936e-05,
      "loss": 4.9439,
      "step": 4507
    },
    {
      "epoch": 0.18199434800161485,
      "grad_norm": 77.31818389892578,
      "learning_rate": 1.88130798327732e-05,
      "loss": 8.0451,
      "step": 4508
    },
    {
      "epoch": 0.1820347194186516,
      "grad_norm": 139.33328247070312,
      "learning_rate": 1.881246187036837e-05,
      "loss": 5.0412,
      "step": 4509
    },
    {
      "epoch": 0.18207509083568832,
      "grad_norm": 124.3337173461914,
      "learning_rate": 1.8811843757291026e-05,
      "loss": 6.1502,
      "step": 4510
    },
    {
      "epoch": 0.18211546225272507,
      "grad_norm": 147.2711944580078,
      "learning_rate": 1.8811225493551728e-05,
      "loss": 5.0051,
      "step": 4511
    },
    {
      "epoch": 0.18215583366976182,
      "grad_norm": 97.00089263916016,
      "learning_rate": 1.881060707916105e-05,
      "loss": 4.5399,
      "step": 4512
    },
    {
      "epoch": 0.18219620508679854,
      "grad_norm": 96.81366729736328,
      "learning_rate": 1.8809988514129566e-05,
      "loss": 5.6032,
      "step": 4513
    },
    {
      "epoch": 0.1822365765038353,
      "grad_norm": 99.57076263427734,
      "learning_rate": 1.8809369798467854e-05,
      "loss": 5.4562,
      "step": 4514
    },
    {
      "epoch": 0.182276947920872,
      "grad_norm": 189.3133544921875,
      "learning_rate": 1.8808750932186486e-05,
      "loss": 4.5436,
      "step": 4515
    },
    {
      "epoch": 0.18231731933790876,
      "grad_norm": 202.04534912109375,
      "learning_rate": 1.8808131915296046e-05,
      "loss": 5.9051,
      "step": 4516
    },
    {
      "epoch": 0.1823576907549455,
      "grad_norm": 121.9220199584961,
      "learning_rate": 1.8807512747807122e-05,
      "loss": 6.0319,
      "step": 4517
    },
    {
      "epoch": 0.18239806217198223,
      "grad_norm": 142.06507873535156,
      "learning_rate": 1.8806893429730296e-05,
      "loss": 5.1265,
      "step": 4518
    },
    {
      "epoch": 0.18243843358901898,
      "grad_norm": 124.68856811523438,
      "learning_rate": 1.8806273961076158e-05,
      "loss": 4.0456,
      "step": 4519
    },
    {
      "epoch": 0.1824788050060557,
      "grad_norm": 180.44805908203125,
      "learning_rate": 1.88056543418553e-05,
      "loss": 4.4207,
      "step": 4520
    },
    {
      "epoch": 0.18251917642309246,
      "grad_norm": 98.07128143310547,
      "learning_rate": 1.8805034572078314e-05,
      "loss": 4.4753,
      "step": 4521
    },
    {
      "epoch": 0.18255954784012918,
      "grad_norm": 311.02545166015625,
      "learning_rate": 1.8804414651755797e-05,
      "loss": 5.8942,
      "step": 4522
    },
    {
      "epoch": 0.18259991925716593,
      "grad_norm": 112.94286346435547,
      "learning_rate": 1.880379458089835e-05,
      "loss": 5.8283,
      "step": 4523
    },
    {
      "epoch": 0.18264029067420268,
      "grad_norm": 143.62294006347656,
      "learning_rate": 1.8803174359516577e-05,
      "loss": 5.4056,
      "step": 4524
    },
    {
      "epoch": 0.1826806620912394,
      "grad_norm": 115.24046325683594,
      "learning_rate": 1.8802553987621077e-05,
      "loss": 4.9882,
      "step": 4525
    },
    {
      "epoch": 0.18272103350827615,
      "grad_norm": 234.4067840576172,
      "learning_rate": 1.880193346522246e-05,
      "loss": 6.1693,
      "step": 4526
    },
    {
      "epoch": 0.18276140492531287,
      "grad_norm": 87.65225219726562,
      "learning_rate": 1.8801312792331334e-05,
      "loss": 4.7085,
      "step": 4527
    },
    {
      "epoch": 0.18280177634234962,
      "grad_norm": 96.2287368774414,
      "learning_rate": 1.8800691968958312e-05,
      "loss": 6.4881,
      "step": 4528
    },
    {
      "epoch": 0.18284214775938634,
      "grad_norm": 110.61905670166016,
      "learning_rate": 1.880007099511401e-05,
      "loss": 6.4581,
      "step": 4529
    },
    {
      "epoch": 0.1828825191764231,
      "grad_norm": 73.80612182617188,
      "learning_rate": 1.8799449870809042e-05,
      "loss": 5.2856,
      "step": 4530
    },
    {
      "epoch": 0.18292289059345984,
      "grad_norm": 79.30680847167969,
      "learning_rate": 1.879882859605403e-05,
      "loss": 5.0945,
      "step": 4531
    },
    {
      "epoch": 0.18296326201049656,
      "grad_norm": 81.67464447021484,
      "learning_rate": 1.879820717085959e-05,
      "loss": 5.4015,
      "step": 4532
    },
    {
      "epoch": 0.1830036334275333,
      "grad_norm": 93.5469970703125,
      "learning_rate": 1.8797585595236364e-05,
      "loss": 6.3427,
      "step": 4533
    },
    {
      "epoch": 0.18304400484457004,
      "grad_norm": 89.47593688964844,
      "learning_rate": 1.879696386919496e-05,
      "loss": 6.1791,
      "step": 4534
    },
    {
      "epoch": 0.18308437626160678,
      "grad_norm": 118.42133331298828,
      "learning_rate": 1.8796341992746016e-05,
      "loss": 6.8079,
      "step": 4535
    },
    {
      "epoch": 0.1831247476786435,
      "grad_norm": 105.60814666748047,
      "learning_rate": 1.8795719965900166e-05,
      "loss": 7.0384,
      "step": 4536
    },
    {
      "epoch": 0.18316511909568026,
      "grad_norm": 126.54280853271484,
      "learning_rate": 1.8795097788668048e-05,
      "loss": 5.543,
      "step": 4537
    },
    {
      "epoch": 0.183205490512717,
      "grad_norm": 175.79116821289062,
      "learning_rate": 1.8794475461060294e-05,
      "loss": 4.6904,
      "step": 4538
    },
    {
      "epoch": 0.18324586192975373,
      "grad_norm": 135.79490661621094,
      "learning_rate": 1.8793852983087545e-05,
      "loss": 5.3265,
      "step": 4539
    },
    {
      "epoch": 0.18328623334679048,
      "grad_norm": 101.34138488769531,
      "learning_rate": 1.8793230354760445e-05,
      "loss": 6.6521,
      "step": 4540
    },
    {
      "epoch": 0.1833266047638272,
      "grad_norm": 122.11870574951172,
      "learning_rate": 1.879260757608964e-05,
      "loss": 5.7105,
      "step": 4541
    },
    {
      "epoch": 0.18336697618086395,
      "grad_norm": 115.12596893310547,
      "learning_rate": 1.8791984647085777e-05,
      "loss": 4.7471,
      "step": 4542
    },
    {
      "epoch": 0.1834073475979007,
      "grad_norm": 83.57764434814453,
      "learning_rate": 1.879136156775951e-05,
      "loss": 7.1048,
      "step": 4543
    },
    {
      "epoch": 0.18344771901493742,
      "grad_norm": 88.11918640136719,
      "learning_rate": 1.8790738338121485e-05,
      "loss": 5.8927,
      "step": 4544
    },
    {
      "epoch": 0.18348809043197417,
      "grad_norm": 89.87812042236328,
      "learning_rate": 1.8790114958182365e-05,
      "loss": 4.3498,
      "step": 4545
    },
    {
      "epoch": 0.1835284618490109,
      "grad_norm": 106.62186431884766,
      "learning_rate": 1.8789491427952808e-05,
      "loss": 6.1206,
      "step": 4546
    },
    {
      "epoch": 0.18356883326604764,
      "grad_norm": 144.27914428710938,
      "learning_rate": 1.8788867747443467e-05,
      "loss": 5.073,
      "step": 4547
    },
    {
      "epoch": 0.18360920468308436,
      "grad_norm": 77.76458740234375,
      "learning_rate": 1.8788243916665014e-05,
      "loss": 5.4746,
      "step": 4548
    },
    {
      "epoch": 0.18364957610012111,
      "grad_norm": 106.73190307617188,
      "learning_rate": 1.8787619935628113e-05,
      "loss": 5.339,
      "step": 4549
    },
    {
      "epoch": 0.18368994751715786,
      "grad_norm": 68.28231811523438,
      "learning_rate": 1.878699580434343e-05,
      "loss": 6.056,
      "step": 4550
    },
    {
      "epoch": 0.1837303189341946,
      "grad_norm": 108.1685791015625,
      "learning_rate": 1.8786371522821638e-05,
      "loss": 4.276,
      "step": 4551
    },
    {
      "epoch": 0.18377069035123134,
      "grad_norm": 108.0763168334961,
      "learning_rate": 1.8785747091073413e-05,
      "loss": 4.7265,
      "step": 4552
    },
    {
      "epoch": 0.18381106176826806,
      "grad_norm": 97.5424575805664,
      "learning_rate": 1.8785122509109425e-05,
      "loss": 4.0697,
      "step": 4553
    },
    {
      "epoch": 0.1838514331853048,
      "grad_norm": 62.86819076538086,
      "learning_rate": 1.878449777694036e-05,
      "loss": 5.9663,
      "step": 4554
    },
    {
      "epoch": 0.18389180460234153,
      "grad_norm": 128.21316528320312,
      "learning_rate": 1.8783872894576895e-05,
      "loss": 5.5223,
      "step": 4555
    },
    {
      "epoch": 0.18393217601937828,
      "grad_norm": 165.86068725585938,
      "learning_rate": 1.8783247862029715e-05,
      "loss": 7.176,
      "step": 4556
    },
    {
      "epoch": 0.18397254743641503,
      "grad_norm": 74.89310455322266,
      "learning_rate": 1.878262267930951e-05,
      "loss": 3.9437,
      "step": 4557
    },
    {
      "epoch": 0.18401291885345175,
      "grad_norm": 94.79899597167969,
      "learning_rate": 1.878199734642696e-05,
      "loss": 5.0494,
      "step": 4558
    },
    {
      "epoch": 0.1840532902704885,
      "grad_norm": 72.45326232910156,
      "learning_rate": 1.878137186339277e-05,
      "loss": 5.4964,
      "step": 4559
    },
    {
      "epoch": 0.18409366168752522,
      "grad_norm": 84.77030181884766,
      "learning_rate": 1.8780746230217623e-05,
      "loss": 4.1005,
      "step": 4560
    },
    {
      "epoch": 0.18413403310456197,
      "grad_norm": 84.17103576660156,
      "learning_rate": 1.8780120446912216e-05,
      "loss": 6.3096,
      "step": 4561
    },
    {
      "epoch": 0.1841744045215987,
      "grad_norm": 90.32174682617188,
      "learning_rate": 1.877949451348726e-05,
      "loss": 5.3859,
      "step": 4562
    },
    {
      "epoch": 0.18421477593863544,
      "grad_norm": 148.79637145996094,
      "learning_rate": 1.8778868429953446e-05,
      "loss": 4.5847,
      "step": 4563
    },
    {
      "epoch": 0.1842551473556722,
      "grad_norm": 160.26889038085938,
      "learning_rate": 1.8778242196321482e-05,
      "loss": 8.1913,
      "step": 4564
    },
    {
      "epoch": 0.18429551877270892,
      "grad_norm": 93.62619018554688,
      "learning_rate": 1.8777615812602073e-05,
      "loss": 5.5482,
      "step": 4565
    },
    {
      "epoch": 0.18433589018974567,
      "grad_norm": 118.01126861572266,
      "learning_rate": 1.8776989278805936e-05,
      "loss": 6.0355,
      "step": 4566
    },
    {
      "epoch": 0.1843762616067824,
      "grad_norm": 134.16441345214844,
      "learning_rate": 1.877636259494377e-05,
      "loss": 4.6085,
      "step": 4567
    },
    {
      "epoch": 0.18441663302381914,
      "grad_norm": 82.48136901855469,
      "learning_rate": 1.8775735761026306e-05,
      "loss": 5.1397,
      "step": 4568
    },
    {
      "epoch": 0.1844570044408559,
      "grad_norm": 85.60054779052734,
      "learning_rate": 1.877510877706425e-05,
      "loss": 6.6248,
      "step": 4569
    },
    {
      "epoch": 0.1844973758578926,
      "grad_norm": 88.70692443847656,
      "learning_rate": 1.8774481643068324e-05,
      "loss": 5.2367,
      "step": 4570
    },
    {
      "epoch": 0.18453774727492936,
      "grad_norm": 85.86957550048828,
      "learning_rate": 1.8773854359049254e-05,
      "loss": 4.9919,
      "step": 4571
    },
    {
      "epoch": 0.18457811869196608,
      "grad_norm": 109.31304168701172,
      "learning_rate": 1.8773226925017766e-05,
      "loss": 6.9182,
      "step": 4572
    },
    {
      "epoch": 0.18461849010900283,
      "grad_norm": 76.15506744384766,
      "learning_rate": 1.877259934098458e-05,
      "loss": 5.3503,
      "step": 4573
    },
    {
      "epoch": 0.18465886152603955,
      "grad_norm": 124.04484558105469,
      "learning_rate": 1.8771971606960432e-05,
      "loss": 3.8786,
      "step": 4574
    },
    {
      "epoch": 0.1846992329430763,
      "grad_norm": 81.67253112792969,
      "learning_rate": 1.877134372295605e-05,
      "loss": 5.3611,
      "step": 4575
    },
    {
      "epoch": 0.18473960436011305,
      "grad_norm": 119.49285125732422,
      "learning_rate": 1.8770715688982174e-05,
      "loss": 5.0978,
      "step": 4576
    },
    {
      "epoch": 0.18477997577714977,
      "grad_norm": 264.64727783203125,
      "learning_rate": 1.8770087505049547e-05,
      "loss": 5.7146,
      "step": 4577
    },
    {
      "epoch": 0.18482034719418652,
      "grad_norm": 101.76364135742188,
      "learning_rate": 1.8769459171168897e-05,
      "loss": 5.1583,
      "step": 4578
    },
    {
      "epoch": 0.18486071861122325,
      "grad_norm": 180.89923095703125,
      "learning_rate": 1.8768830687350975e-05,
      "loss": 4.3609,
      "step": 4579
    },
    {
      "epoch": 0.18490109002826,
      "grad_norm": 394.7458190917969,
      "learning_rate": 1.8768202053606526e-05,
      "loss": 5.5377,
      "step": 4580
    },
    {
      "epoch": 0.18494146144529672,
      "grad_norm": 111.14640808105469,
      "learning_rate": 1.8767573269946296e-05,
      "loss": 5.0642,
      "step": 4581
    },
    {
      "epoch": 0.18498183286233347,
      "grad_norm": 84.74592590332031,
      "learning_rate": 1.8766944336381038e-05,
      "loss": 6.4584,
      "step": 4582
    },
    {
      "epoch": 0.18502220427937022,
      "grad_norm": 291.5816955566406,
      "learning_rate": 1.87663152529215e-05,
      "loss": 5.4919,
      "step": 4583
    },
    {
      "epoch": 0.18506257569640694,
      "grad_norm": 102.78178405761719,
      "learning_rate": 1.876568601957845e-05,
      "loss": 4.8977,
      "step": 4584
    },
    {
      "epoch": 0.1851029471134437,
      "grad_norm": 109.40532684326172,
      "learning_rate": 1.8765056636362633e-05,
      "loss": 5.4205,
      "step": 4585
    },
    {
      "epoch": 0.1851433185304804,
      "grad_norm": 110.64908599853516,
      "learning_rate": 1.8764427103284816e-05,
      "loss": 6.5248,
      "step": 4586
    },
    {
      "epoch": 0.18518368994751716,
      "grad_norm": 93.3238525390625,
      "learning_rate": 1.8763797420355763e-05,
      "loss": 4.8328,
      "step": 4587
    },
    {
      "epoch": 0.18522406136455388,
      "grad_norm": 136.72874450683594,
      "learning_rate": 1.8763167587586237e-05,
      "loss": 4.6043,
      "step": 4588
    },
    {
      "epoch": 0.18526443278159063,
      "grad_norm": 80.10401916503906,
      "learning_rate": 1.876253760498701e-05,
      "loss": 5.354,
      "step": 4589
    },
    {
      "epoch": 0.18530480419862738,
      "grad_norm": 146.45364379882812,
      "learning_rate": 1.8761907472568853e-05,
      "loss": 4.0519,
      "step": 4590
    },
    {
      "epoch": 0.1853451756156641,
      "grad_norm": 114.4381103515625,
      "learning_rate": 1.8761277190342537e-05,
      "loss": 4.4613,
      "step": 4591
    },
    {
      "epoch": 0.18538554703270085,
      "grad_norm": 208.7165985107422,
      "learning_rate": 1.876064675831884e-05,
      "loss": 7.2959,
      "step": 4592
    },
    {
      "epoch": 0.18542591844973758,
      "grad_norm": 197.22837829589844,
      "learning_rate": 1.876001617650854e-05,
      "loss": 6.6824,
      "step": 4593
    },
    {
      "epoch": 0.18546628986677433,
      "grad_norm": 99.55792999267578,
      "learning_rate": 1.8759385444922424e-05,
      "loss": 4.8318,
      "step": 4594
    },
    {
      "epoch": 0.18550666128381108,
      "grad_norm": 104.09737396240234,
      "learning_rate": 1.875875456357127e-05,
      "loss": 4.4066,
      "step": 4595
    },
    {
      "epoch": 0.1855470327008478,
      "grad_norm": 104.06756591796875,
      "learning_rate": 1.8758123532465865e-05,
      "loss": 6.0926,
      "step": 4596
    },
    {
      "epoch": 0.18558740411788455,
      "grad_norm": 151.07562255859375,
      "learning_rate": 1.8757492351617e-05,
      "loss": 5.8571,
      "step": 4597
    },
    {
      "epoch": 0.18562777553492127,
      "grad_norm": 211.13758850097656,
      "learning_rate": 1.8756861021035463e-05,
      "loss": 5.3625,
      "step": 4598
    },
    {
      "epoch": 0.18566814695195802,
      "grad_norm": 183.5526580810547,
      "learning_rate": 1.8756229540732055e-05,
      "loss": 3.7272,
      "step": 4599
    },
    {
      "epoch": 0.18570851836899474,
      "grad_norm": 174.64248657226562,
      "learning_rate": 1.8755597910717566e-05,
      "loss": 5.5487,
      "step": 4600
    },
    {
      "epoch": 0.1857488897860315,
      "grad_norm": 1140.3538818359375,
      "learning_rate": 1.87549661310028e-05,
      "loss": 5.2326,
      "step": 4601
    },
    {
      "epoch": 0.18578926120306824,
      "grad_norm": 160.7980499267578,
      "learning_rate": 1.875433420159856e-05,
      "loss": 4.9857,
      "step": 4602
    },
    {
      "epoch": 0.18582963262010496,
      "grad_norm": 133.35687255859375,
      "learning_rate": 1.8753702122515644e-05,
      "loss": 7.4035,
      "step": 4603
    },
    {
      "epoch": 0.1858700040371417,
      "grad_norm": 112.6512680053711,
      "learning_rate": 1.8753069893764863e-05,
      "loss": 7.0837,
      "step": 4604
    },
    {
      "epoch": 0.18591037545417843,
      "grad_norm": 97.68309783935547,
      "learning_rate": 1.875243751535703e-05,
      "loss": 6.8406,
      "step": 4605
    },
    {
      "epoch": 0.18595074687121518,
      "grad_norm": 127.88373565673828,
      "learning_rate": 1.875180498730295e-05,
      "loss": 5.3508,
      "step": 4606
    },
    {
      "epoch": 0.1859911182882519,
      "grad_norm": 108.87846374511719,
      "learning_rate": 1.8751172309613447e-05,
      "loss": 3.6682,
      "step": 4607
    },
    {
      "epoch": 0.18603148970528866,
      "grad_norm": 377.88079833984375,
      "learning_rate": 1.875053948229933e-05,
      "loss": 6.1314,
      "step": 4608
    },
    {
      "epoch": 0.1860718611223254,
      "grad_norm": 88.97782897949219,
      "learning_rate": 1.874990650537142e-05,
      "loss": 5.1773,
      "step": 4609
    },
    {
      "epoch": 0.18611223253936213,
      "grad_norm": 126.72435760498047,
      "learning_rate": 1.8749273378840544e-05,
      "loss": 6.0203,
      "step": 4610
    },
    {
      "epoch": 0.18615260395639888,
      "grad_norm": 85.60004425048828,
      "learning_rate": 1.8748640102717522e-05,
      "loss": 5.5155,
      "step": 4611
    },
    {
      "epoch": 0.1861929753734356,
      "grad_norm": 92.2518310546875,
      "learning_rate": 1.8748006677013184e-05,
      "loss": 5.6641,
      "step": 4612
    },
    {
      "epoch": 0.18623334679047235,
      "grad_norm": 119.16898345947266,
      "learning_rate": 1.8747373101738363e-05,
      "loss": 5.2121,
      "step": 4613
    },
    {
      "epoch": 0.18627371820750907,
      "grad_norm": 174.179443359375,
      "learning_rate": 1.8746739376903886e-05,
      "loss": 6.5645,
      "step": 4614
    },
    {
      "epoch": 0.18631408962454582,
      "grad_norm": 104.62369537353516,
      "learning_rate": 1.8746105502520592e-05,
      "loss": 6.2914,
      "step": 4615
    },
    {
      "epoch": 0.18635446104158257,
      "grad_norm": 100.8128662109375,
      "learning_rate": 1.8745471478599317e-05,
      "loss": 5.032,
      "step": 4616
    },
    {
      "epoch": 0.1863948324586193,
      "grad_norm": 114.3680648803711,
      "learning_rate": 1.87448373051509e-05,
      "loss": 5.2103,
      "step": 4617
    },
    {
      "epoch": 0.18643520387565604,
      "grad_norm": 125.16712188720703,
      "learning_rate": 1.874420298218619e-05,
      "loss": 4.5779,
      "step": 4618
    },
    {
      "epoch": 0.18647557529269276,
      "grad_norm": 89.89887237548828,
      "learning_rate": 1.8743568509716022e-05,
      "loss": 5.6824,
      "step": 4619
    },
    {
      "epoch": 0.1865159467097295,
      "grad_norm": 90.58696746826172,
      "learning_rate": 1.8742933887751254e-05,
      "loss": 5.5574,
      "step": 4620
    },
    {
      "epoch": 0.18655631812676626,
      "grad_norm": 96.68438720703125,
      "learning_rate": 1.8742299116302732e-05,
      "loss": 6.5173,
      "step": 4621
    },
    {
      "epoch": 0.18659668954380298,
      "grad_norm": 103.42584991455078,
      "learning_rate": 1.8741664195381314e-05,
      "loss": 6.4233,
      "step": 4622
    },
    {
      "epoch": 0.18663706096083973,
      "grad_norm": 119.9439468383789,
      "learning_rate": 1.8741029124997846e-05,
      "loss": 5.5412,
      "step": 4623
    },
    {
      "epoch": 0.18667743237787646,
      "grad_norm": 80.63178253173828,
      "learning_rate": 1.874039390516319e-05,
      "loss": 5.0771,
      "step": 4624
    },
    {
      "epoch": 0.1867178037949132,
      "grad_norm": 159.6189422607422,
      "learning_rate": 1.8739758535888216e-05,
      "loss": 4.4143,
      "step": 4625
    },
    {
      "epoch": 0.18675817521194993,
      "grad_norm": 114.33844757080078,
      "learning_rate": 1.8739123017183772e-05,
      "loss": 4.403,
      "step": 4626
    },
    {
      "epoch": 0.18679854662898668,
      "grad_norm": 113.18330383300781,
      "learning_rate": 1.8738487349060737e-05,
      "loss": 5.7331,
      "step": 4627
    },
    {
      "epoch": 0.18683891804602343,
      "grad_norm": 98.85076904296875,
      "learning_rate": 1.8737851531529974e-05,
      "loss": 6.4727,
      "step": 4628
    },
    {
      "epoch": 0.18687928946306015,
      "grad_norm": 156.97506713867188,
      "learning_rate": 1.8737215564602352e-05,
      "loss": 5.1976,
      "step": 4629
    },
    {
      "epoch": 0.1869196608800969,
      "grad_norm": 89.0325927734375,
      "learning_rate": 1.873657944828875e-05,
      "loss": 5.8316,
      "step": 4630
    },
    {
      "epoch": 0.18696003229713362,
      "grad_norm": 77.89742279052734,
      "learning_rate": 1.8735943182600035e-05,
      "loss": 5.6329,
      "step": 4631
    },
    {
      "epoch": 0.18700040371417037,
      "grad_norm": 138.6244659423828,
      "learning_rate": 1.8735306767547097e-05,
      "loss": 3.0575,
      "step": 4632
    },
    {
      "epoch": 0.1870407751312071,
      "grad_norm": 107.50379943847656,
      "learning_rate": 1.873467020314081e-05,
      "loss": 5.1456,
      "step": 4633
    },
    {
      "epoch": 0.18708114654824384,
      "grad_norm": 88.44583892822266,
      "learning_rate": 1.873403348939206e-05,
      "loss": 6.1464,
      "step": 4634
    },
    {
      "epoch": 0.1871215179652806,
      "grad_norm": 117.69268035888672,
      "learning_rate": 1.873339662631173e-05,
      "loss": 6.4857,
      "step": 4635
    },
    {
      "epoch": 0.18716188938231731,
      "grad_norm": 78.84596252441406,
      "learning_rate": 1.8732759613910712e-05,
      "loss": 5.2976,
      "step": 4636
    },
    {
      "epoch": 0.18720226079935406,
      "grad_norm": 133.46304321289062,
      "learning_rate": 1.8732122452199903e-05,
      "loss": 7.0392,
      "step": 4637
    },
    {
      "epoch": 0.1872426322163908,
      "grad_norm": 101.65946960449219,
      "learning_rate": 1.8731485141190185e-05,
      "loss": 5.615,
      "step": 4638
    },
    {
      "epoch": 0.18728300363342754,
      "grad_norm": 117.65607452392578,
      "learning_rate": 1.873084768089246e-05,
      "loss": 5.5191,
      "step": 4639
    },
    {
      "epoch": 0.18732337505046426,
      "grad_norm": 168.76449584960938,
      "learning_rate": 1.8730210071317633e-05,
      "loss": 4.1252,
      "step": 4640
    },
    {
      "epoch": 0.187363746467501,
      "grad_norm": 191.9886016845703,
      "learning_rate": 1.8729572312476595e-05,
      "loss": 3.7075,
      "step": 4641
    },
    {
      "epoch": 0.18740411788453776,
      "grad_norm": 169.9391326904297,
      "learning_rate": 1.8728934404380257e-05,
      "loss": 6.2555,
      "step": 4642
    },
    {
      "epoch": 0.18744448930157448,
      "grad_norm": 112.89250183105469,
      "learning_rate": 1.8728296347039526e-05,
      "loss": 5.675,
      "step": 4643
    },
    {
      "epoch": 0.18748486071861123,
      "grad_norm": 69.27188873291016,
      "learning_rate": 1.872765814046531e-05,
      "loss": 5.9559,
      "step": 4644
    },
    {
      "epoch": 0.18752523213564795,
      "grad_norm": 124.44142150878906,
      "learning_rate": 1.8727019784668518e-05,
      "loss": 5.1695,
      "step": 4645
    },
    {
      "epoch": 0.1875656035526847,
      "grad_norm": 102.68328857421875,
      "learning_rate": 1.8726381279660066e-05,
      "loss": 5.1968,
      "step": 4646
    },
    {
      "epoch": 0.18760597496972145,
      "grad_norm": 130.45730590820312,
      "learning_rate": 1.8725742625450874e-05,
      "loss": 5.5013,
      "step": 4647
    },
    {
      "epoch": 0.18764634638675817,
      "grad_norm": 105.45684814453125,
      "learning_rate": 1.8725103822051858e-05,
      "loss": 6.03,
      "step": 4648
    },
    {
      "epoch": 0.18768671780379492,
      "grad_norm": 117.01299285888672,
      "learning_rate": 1.872446486947394e-05,
      "loss": 5.8874,
      "step": 4649
    },
    {
      "epoch": 0.18772708922083164,
      "grad_norm": 184.73756408691406,
      "learning_rate": 1.8723825767728048e-05,
      "loss": 3.9098,
      "step": 4650
    },
    {
      "epoch": 0.1877674606378684,
      "grad_norm": 195.44070434570312,
      "learning_rate": 1.8723186516825107e-05,
      "loss": 3.3912,
      "step": 4651
    },
    {
      "epoch": 0.18780783205490512,
      "grad_norm": 116.27981567382812,
      "learning_rate": 1.8722547116776046e-05,
      "loss": 6.8502,
      "step": 4652
    },
    {
      "epoch": 0.18784820347194187,
      "grad_norm": 113.01802062988281,
      "learning_rate": 1.8721907567591798e-05,
      "loss": 5.2091,
      "step": 4653
    },
    {
      "epoch": 0.18788857488897862,
      "grad_norm": 81.21910858154297,
      "learning_rate": 1.8721267869283296e-05,
      "loss": 5.7085,
      "step": 4654
    },
    {
      "epoch": 0.18792894630601534,
      "grad_norm": 79.19062042236328,
      "learning_rate": 1.872062802186148e-05,
      "loss": 4.3639,
      "step": 4655
    },
    {
      "epoch": 0.1879693177230521,
      "grad_norm": 120.57144165039062,
      "learning_rate": 1.871998802533729e-05,
      "loss": 6.4804,
      "step": 4656
    },
    {
      "epoch": 0.1880096891400888,
      "grad_norm": 158.0596160888672,
      "learning_rate": 1.8719347879721667e-05,
      "loss": 3.3326,
      "step": 4657
    },
    {
      "epoch": 0.18805006055712556,
      "grad_norm": 101.56995391845703,
      "learning_rate": 1.8718707585025558e-05,
      "loss": 5.7061,
      "step": 4658
    },
    {
      "epoch": 0.18809043197416228,
      "grad_norm": 77.73393249511719,
      "learning_rate": 1.8718067141259904e-05,
      "loss": 5.4708,
      "step": 4659
    },
    {
      "epoch": 0.18813080339119903,
      "grad_norm": 96.7105484008789,
      "learning_rate": 1.8717426548435665e-05,
      "loss": 5.9939,
      "step": 4660
    },
    {
      "epoch": 0.18817117480823578,
      "grad_norm": 118.5682601928711,
      "learning_rate": 1.871678580656379e-05,
      "loss": 4.3706,
      "step": 4661
    },
    {
      "epoch": 0.1882115462252725,
      "grad_norm": 106.67512512207031,
      "learning_rate": 1.871614491565523e-05,
      "loss": 6.2372,
      "step": 4662
    },
    {
      "epoch": 0.18825191764230925,
      "grad_norm": 147.2771759033203,
      "learning_rate": 1.8715503875720942e-05,
      "loss": 4.8448,
      "step": 4663
    },
    {
      "epoch": 0.18829228905934597,
      "grad_norm": 85.09869384765625,
      "learning_rate": 1.8714862686771897e-05,
      "loss": 4.249,
      "step": 4664
    },
    {
      "epoch": 0.18833266047638272,
      "grad_norm": 89.09709167480469,
      "learning_rate": 1.8714221348819042e-05,
      "loss": 6.0057,
      "step": 4665
    },
    {
      "epoch": 0.18837303189341945,
      "grad_norm": 161.03196716308594,
      "learning_rate": 1.871357986187336e-05,
      "loss": 3.6161,
      "step": 4666
    },
    {
      "epoch": 0.1884134033104562,
      "grad_norm": 110.88042449951172,
      "learning_rate": 1.8712938225945804e-05,
      "loss": 6.7575,
      "step": 4667
    },
    {
      "epoch": 0.18845377472749295,
      "grad_norm": 91.88871765136719,
      "learning_rate": 1.8712296441047355e-05,
      "loss": 6.5584,
      "step": 4668
    },
    {
      "epoch": 0.18849414614452967,
      "grad_norm": 86.89176177978516,
      "learning_rate": 1.871165450718898e-05,
      "loss": 5.4329,
      "step": 4669
    },
    {
      "epoch": 0.18853451756156642,
      "grad_norm": 90.94937133789062,
      "learning_rate": 1.871101242438165e-05,
      "loss": 4.865,
      "step": 4670
    },
    {
      "epoch": 0.18857488897860314,
      "grad_norm": 131.33595275878906,
      "learning_rate": 1.8710370192636357e-05,
      "loss": 7.2049,
      "step": 4671
    },
    {
      "epoch": 0.1886152603956399,
      "grad_norm": 103.18208312988281,
      "learning_rate": 1.8709727811964074e-05,
      "loss": 6.4453,
      "step": 4672
    },
    {
      "epoch": 0.18865563181267664,
      "grad_norm": 106.81297302246094,
      "learning_rate": 1.870908528237578e-05,
      "loss": 5.8214,
      "step": 4673
    },
    {
      "epoch": 0.18869600322971336,
      "grad_norm": 205.4401397705078,
      "learning_rate": 1.8708442603882467e-05,
      "loss": 4.1165,
      "step": 4674
    },
    {
      "epoch": 0.1887363746467501,
      "grad_norm": 94.43231964111328,
      "learning_rate": 1.8707799776495118e-05,
      "loss": 4.647,
      "step": 4675
    },
    {
      "epoch": 0.18877674606378683,
      "grad_norm": 68.71650695800781,
      "learning_rate": 1.8707156800224733e-05,
      "loss": 6.3303,
      "step": 4676
    },
    {
      "epoch": 0.18881711748082358,
      "grad_norm": 100.3755111694336,
      "learning_rate": 1.8706513675082294e-05,
      "loss": 6.2467,
      "step": 4677
    },
    {
      "epoch": 0.1888574888978603,
      "grad_norm": 104.98118591308594,
      "learning_rate": 1.8705870401078806e-05,
      "loss": 5.671,
      "step": 4678
    },
    {
      "epoch": 0.18889786031489705,
      "grad_norm": 75.91372680664062,
      "learning_rate": 1.870522697822526e-05,
      "loss": 5.4935,
      "step": 4679
    },
    {
      "epoch": 0.1889382317319338,
      "grad_norm": 115.4427490234375,
      "learning_rate": 1.8704583406532664e-05,
      "loss": 5.7735,
      "step": 4680
    },
    {
      "epoch": 0.18897860314897053,
      "grad_norm": 81.36875915527344,
      "learning_rate": 1.8703939686012017e-05,
      "loss": 5.5721,
      "step": 4681
    },
    {
      "epoch": 0.18901897456600728,
      "grad_norm": 85.480224609375,
      "learning_rate": 1.8703295816674328e-05,
      "loss": 4.217,
      "step": 4682
    },
    {
      "epoch": 0.189059345983044,
      "grad_norm": 90.6861572265625,
      "learning_rate": 1.8702651798530603e-05,
      "loss": 5.9299,
      "step": 4683
    },
    {
      "epoch": 0.18909971740008075,
      "grad_norm": 149.24301147460938,
      "learning_rate": 1.8702007631591853e-05,
      "loss": 5.1873,
      "step": 4684
    },
    {
      "epoch": 0.18914008881711747,
      "grad_norm": 116.06559753417969,
      "learning_rate": 1.8701363315869095e-05,
      "loss": 4.1417,
      "step": 4685
    },
    {
      "epoch": 0.18918046023415422,
      "grad_norm": 92.03489685058594,
      "learning_rate": 1.8700718851373342e-05,
      "loss": 5.8553,
      "step": 4686
    },
    {
      "epoch": 0.18922083165119097,
      "grad_norm": 90.68152618408203,
      "learning_rate": 1.8700074238115618e-05,
      "loss": 5.2933,
      "step": 4687
    },
    {
      "epoch": 0.1892612030682277,
      "grad_norm": 270.50433349609375,
      "learning_rate": 1.8699429476106935e-05,
      "loss": 6.0374,
      "step": 4688
    },
    {
      "epoch": 0.18930157448526444,
      "grad_norm": 92.64954376220703,
      "learning_rate": 1.8698784565358328e-05,
      "loss": 4.3545,
      "step": 4689
    },
    {
      "epoch": 0.18934194590230116,
      "grad_norm": 136.528076171875,
      "learning_rate": 1.8698139505880815e-05,
      "loss": 6.0769,
      "step": 4690
    },
    {
      "epoch": 0.1893823173193379,
      "grad_norm": 118.97066497802734,
      "learning_rate": 1.869749429768543e-05,
      "loss": 5.9251,
      "step": 4691
    },
    {
      "epoch": 0.18942268873637463,
      "grad_norm": 67.71609497070312,
      "learning_rate": 1.86968489407832e-05,
      "loss": 5.4032,
      "step": 4692
    },
    {
      "epoch": 0.18946306015341138,
      "grad_norm": 106.60440826416016,
      "learning_rate": 1.8696203435185163e-05,
      "loss": 4.1165,
      "step": 4693
    },
    {
      "epoch": 0.18950343157044813,
      "grad_norm": 84.49247741699219,
      "learning_rate": 1.8695557780902354e-05,
      "loss": 5.5965,
      "step": 4694
    },
    {
      "epoch": 0.18954380298748486,
      "grad_norm": 138.6885223388672,
      "learning_rate": 1.8694911977945812e-05,
      "loss": 5.3512,
      "step": 4695
    },
    {
      "epoch": 0.1895841744045216,
      "grad_norm": 132.510986328125,
      "learning_rate": 1.869426602632658e-05,
      "loss": 5.8442,
      "step": 4696
    },
    {
      "epoch": 0.18962454582155833,
      "grad_norm": 75.69818878173828,
      "learning_rate": 1.8693619926055702e-05,
      "loss": 6.3609,
      "step": 4697
    },
    {
      "epoch": 0.18966491723859508,
      "grad_norm": 95.26174926757812,
      "learning_rate": 1.8692973677144225e-05,
      "loss": 4.3642,
      "step": 4698
    },
    {
      "epoch": 0.18970528865563183,
      "grad_norm": 118.26614379882812,
      "learning_rate": 1.8692327279603195e-05,
      "loss": 6.9534,
      "step": 4699
    },
    {
      "epoch": 0.18974566007266855,
      "grad_norm": 109.79351043701172,
      "learning_rate": 1.869168073344367e-05,
      "loss": 6.0302,
      "step": 4700
    },
    {
      "epoch": 0.1897860314897053,
      "grad_norm": 103.05778503417969,
      "learning_rate": 1.8691034038676697e-05,
      "loss": 6.2306,
      "step": 4701
    },
    {
      "epoch": 0.18982640290674202,
      "grad_norm": 128.18421936035156,
      "learning_rate": 1.869038719531334e-05,
      "loss": 5.795,
      "step": 4702
    },
    {
      "epoch": 0.18986677432377877,
      "grad_norm": 113.27005767822266,
      "learning_rate": 1.8689740203364656e-05,
      "loss": 5.0523,
      "step": 4703
    },
    {
      "epoch": 0.1899071457408155,
      "grad_norm": 94.1351089477539,
      "learning_rate": 1.8689093062841704e-05,
      "loss": 4.9823,
      "step": 4704
    },
    {
      "epoch": 0.18994751715785224,
      "grad_norm": 133.09014892578125,
      "learning_rate": 1.8688445773755553e-05,
      "loss": 6.7669,
      "step": 4705
    },
    {
      "epoch": 0.189987888574889,
      "grad_norm": 198.68260192871094,
      "learning_rate": 1.8687798336117268e-05,
      "loss": 5.0463,
      "step": 4706
    },
    {
      "epoch": 0.1900282599919257,
      "grad_norm": 123.72967529296875,
      "learning_rate": 1.868715074993792e-05,
      "loss": 6.0311,
      "step": 4707
    },
    {
      "epoch": 0.19006863140896246,
      "grad_norm": 79.53145599365234,
      "learning_rate": 1.8686503015228578e-05,
      "loss": 6.3756,
      "step": 4708
    },
    {
      "epoch": 0.19010900282599918,
      "grad_norm": 85.97773742675781,
      "learning_rate": 1.868585513200032e-05,
      "loss": 5.1376,
      "step": 4709
    },
    {
      "epoch": 0.19014937424303593,
      "grad_norm": 63.505828857421875,
      "learning_rate": 1.868520710026422e-05,
      "loss": 3.7799,
      "step": 4710
    },
    {
      "epoch": 0.19018974566007266,
      "grad_norm": 99.81352233886719,
      "learning_rate": 1.868455892003136e-05,
      "loss": 7.4224,
      "step": 4711
    },
    {
      "epoch": 0.1902301170771094,
      "grad_norm": 75.07951354980469,
      "learning_rate": 1.8683910591312826e-05,
      "loss": 5.7708,
      "step": 4712
    },
    {
      "epoch": 0.19027048849414616,
      "grad_norm": 104.00457000732422,
      "learning_rate": 1.8683262114119696e-05,
      "loss": 6.8951,
      "step": 4713
    },
    {
      "epoch": 0.19031085991118288,
      "grad_norm": 119.66724395751953,
      "learning_rate": 1.8682613488463062e-05,
      "loss": 5.8834,
      "step": 4714
    },
    {
      "epoch": 0.19035123132821963,
      "grad_norm": 108.14998626708984,
      "learning_rate": 1.868196471435401e-05,
      "loss": 6.8662,
      "step": 4715
    },
    {
      "epoch": 0.19039160274525635,
      "grad_norm": 81.86827850341797,
      "learning_rate": 1.8681315791803642e-05,
      "loss": 5.471,
      "step": 4716
    },
    {
      "epoch": 0.1904319741622931,
      "grad_norm": 96.08167266845703,
      "learning_rate": 1.8680666720823045e-05,
      "loss": 4.8227,
      "step": 4717
    },
    {
      "epoch": 0.19047234557932982,
      "grad_norm": 101.87183380126953,
      "learning_rate": 1.8680017501423312e-05,
      "loss": 4.5756,
      "step": 4718
    },
    {
      "epoch": 0.19051271699636657,
      "grad_norm": 75.06839752197266,
      "learning_rate": 1.8679368133615553e-05,
      "loss": 2.2549,
      "step": 4719
    },
    {
      "epoch": 0.19055308841340332,
      "grad_norm": 121.73297882080078,
      "learning_rate": 1.8678718617410867e-05,
      "loss": 3.8696,
      "step": 4720
    },
    {
      "epoch": 0.19059345983044004,
      "grad_norm": 79.00727844238281,
      "learning_rate": 1.8678068952820355e-05,
      "loss": 5.355,
      "step": 4721
    },
    {
      "epoch": 0.1906338312474768,
      "grad_norm": 111.98582458496094,
      "learning_rate": 1.867741913985513e-05,
      "loss": 4.5648,
      "step": 4722
    },
    {
      "epoch": 0.19067420266451351,
      "grad_norm": 61.40840530395508,
      "learning_rate": 1.8676769178526304e-05,
      "loss": 5.9198,
      "step": 4723
    },
    {
      "epoch": 0.19071457408155026,
      "grad_norm": 98.4198989868164,
      "learning_rate": 1.8676119068844986e-05,
      "loss": 5.5417,
      "step": 4724
    },
    {
      "epoch": 0.19075494549858701,
      "grad_norm": 100.87016296386719,
      "learning_rate": 1.867546881082229e-05,
      "loss": 5.6573,
      "step": 4725
    },
    {
      "epoch": 0.19079531691562374,
      "grad_norm": 99.92564392089844,
      "learning_rate": 1.867481840446934e-05,
      "loss": 3.8937,
      "step": 4726
    },
    {
      "epoch": 0.19083568833266049,
      "grad_norm": 116.37054443359375,
      "learning_rate": 1.8674167849797247e-05,
      "loss": 5.755,
      "step": 4727
    },
    {
      "epoch": 0.1908760597496972,
      "grad_norm": 155.58255004882812,
      "learning_rate": 1.8673517146817146e-05,
      "loss": 5.476,
      "step": 4728
    },
    {
      "epoch": 0.19091643116673396,
      "grad_norm": 312.6177673339844,
      "learning_rate": 1.867286629554015e-05,
      "loss": 7.5141,
      "step": 4729
    },
    {
      "epoch": 0.19095680258377068,
      "grad_norm": 118.71395874023438,
      "learning_rate": 1.8672215295977393e-05,
      "loss": 5.9455,
      "step": 4730
    },
    {
      "epoch": 0.19099717400080743,
      "grad_norm": 70.23538208007812,
      "learning_rate": 1.867156414814001e-05,
      "loss": 6.6839,
      "step": 4731
    },
    {
      "epoch": 0.19103754541784418,
      "grad_norm": 118.41507720947266,
      "learning_rate": 1.8670912852039126e-05,
      "loss": 5.5495,
      "step": 4732
    },
    {
      "epoch": 0.1910779168348809,
      "grad_norm": 118.27098083496094,
      "learning_rate": 1.867026140768588e-05,
      "loss": 5.6354,
      "step": 4733
    },
    {
      "epoch": 0.19111828825191765,
      "grad_norm": 109.0478286743164,
      "learning_rate": 1.8669609815091416e-05,
      "loss": 4.2568,
      "step": 4734
    },
    {
      "epoch": 0.19115865966895437,
      "grad_norm": 93.25469207763672,
      "learning_rate": 1.866895807426686e-05,
      "loss": 5.6239,
      "step": 4735
    },
    {
      "epoch": 0.19119903108599112,
      "grad_norm": 108.06110382080078,
      "learning_rate": 1.8668306185223373e-05,
      "loss": 5.3415,
      "step": 4736
    },
    {
      "epoch": 0.19123940250302784,
      "grad_norm": 94.33202362060547,
      "learning_rate": 1.8667654147972087e-05,
      "loss": 7.4492,
      "step": 4737
    },
    {
      "epoch": 0.1912797739200646,
      "grad_norm": 153.3216552734375,
      "learning_rate": 1.8667001962524157e-05,
      "loss": 5.5412,
      "step": 4738
    },
    {
      "epoch": 0.19132014533710134,
      "grad_norm": 107.41127014160156,
      "learning_rate": 1.8666349628890733e-05,
      "loss": 7.2681,
      "step": 4739
    },
    {
      "epoch": 0.19136051675413807,
      "grad_norm": 111.03934478759766,
      "learning_rate": 1.866569714708297e-05,
      "loss": 5.2059,
      "step": 4740
    },
    {
      "epoch": 0.19140088817117482,
      "grad_norm": 114.05287170410156,
      "learning_rate": 1.8665044517112017e-05,
      "loss": 4.4378,
      "step": 4741
    },
    {
      "epoch": 0.19144125958821154,
      "grad_norm": 77.04254913330078,
      "learning_rate": 1.866439173898904e-05,
      "loss": 4.985,
      "step": 4742
    },
    {
      "epoch": 0.1914816310052483,
      "grad_norm": 98.36638641357422,
      "learning_rate": 1.8663738812725195e-05,
      "loss": 5.6701,
      "step": 4743
    },
    {
      "epoch": 0.191522002422285,
      "grad_norm": 88.83175659179688,
      "learning_rate": 1.8663085738331647e-05,
      "loss": 5.5172,
      "step": 4744
    },
    {
      "epoch": 0.19156237383932176,
      "grad_norm": 87.14396667480469,
      "learning_rate": 1.8662432515819565e-05,
      "loss": 6.2763,
      "step": 4745
    },
    {
      "epoch": 0.1916027452563585,
      "grad_norm": 90.65481567382812,
      "learning_rate": 1.8661779145200113e-05,
      "loss": 5.5795,
      "step": 4746
    },
    {
      "epoch": 0.19164311667339523,
      "grad_norm": 120.23503112792969,
      "learning_rate": 1.8661125626484465e-05,
      "loss": 5.9594,
      "step": 4747
    },
    {
      "epoch": 0.19168348809043198,
      "grad_norm": 87.49078369140625,
      "learning_rate": 1.8660471959683795e-05,
      "loss": 4.1925,
      "step": 4748
    },
    {
      "epoch": 0.1917238595074687,
      "grad_norm": 130.3988494873047,
      "learning_rate": 1.8659818144809278e-05,
      "loss": 6.0315,
      "step": 4749
    },
    {
      "epoch": 0.19176423092450545,
      "grad_norm": 128.6296844482422,
      "learning_rate": 1.8659164181872093e-05,
      "loss": 5.3371,
      "step": 4750
    },
    {
      "epoch": 0.1918046023415422,
      "grad_norm": 77.69158172607422,
      "learning_rate": 1.8658510070883422e-05,
      "loss": 4.7941,
      "step": 4751
    },
    {
      "epoch": 0.19184497375857892,
      "grad_norm": 136.35565185546875,
      "learning_rate": 1.8657855811854446e-05,
      "loss": 7.5555,
      "step": 4752
    },
    {
      "epoch": 0.19188534517561567,
      "grad_norm": 90.11695861816406,
      "learning_rate": 1.8657201404796352e-05,
      "loss": 6.1961,
      "step": 4753
    },
    {
      "epoch": 0.1919257165926524,
      "grad_norm": 95.17582702636719,
      "learning_rate": 1.8656546849720335e-05,
      "loss": 5.9018,
      "step": 4754
    },
    {
      "epoch": 0.19196608800968915,
      "grad_norm": 76.76593780517578,
      "learning_rate": 1.8655892146637578e-05,
      "loss": 5.014,
      "step": 4755
    },
    {
      "epoch": 0.19200645942672587,
      "grad_norm": 66.61088562011719,
      "learning_rate": 1.8655237295559277e-05,
      "loss": 6.503,
      "step": 4756
    },
    {
      "epoch": 0.19204683084376262,
      "grad_norm": 93.97103118896484,
      "learning_rate": 1.8654582296496633e-05,
      "loss": 5.547,
      "step": 4757
    },
    {
      "epoch": 0.19208720226079937,
      "grad_norm": 88.95523071289062,
      "learning_rate": 1.865392714946084e-05,
      "loss": 5.6641,
      "step": 4758
    },
    {
      "epoch": 0.1921275736778361,
      "grad_norm": 119.11508178710938,
      "learning_rate": 1.8653271854463097e-05,
      "loss": 5.6125,
      "step": 4759
    },
    {
      "epoch": 0.19216794509487284,
      "grad_norm": 89.55319213867188,
      "learning_rate": 1.865261641151462e-05,
      "loss": 5.4054,
      "step": 4760
    },
    {
      "epoch": 0.19220831651190956,
      "grad_norm": 113.63619995117188,
      "learning_rate": 1.86519608206266e-05,
      "loss": 5.2799,
      "step": 4761
    },
    {
      "epoch": 0.1922486879289463,
      "grad_norm": 138.10984802246094,
      "learning_rate": 1.865130508181026e-05,
      "loss": 5.592,
      "step": 4762
    },
    {
      "epoch": 0.19228905934598303,
      "grad_norm": 170.0149383544922,
      "learning_rate": 1.8650649195076802e-05,
      "loss": 6.8499,
      "step": 4763
    },
    {
      "epoch": 0.19232943076301978,
      "grad_norm": 89.59378814697266,
      "learning_rate": 1.8649993160437442e-05,
      "loss": 5.5001,
      "step": 4764
    },
    {
      "epoch": 0.19236980218005653,
      "grad_norm": 80.35101318359375,
      "learning_rate": 1.8649336977903403e-05,
      "loss": 4.5487,
      "step": 4765
    },
    {
      "epoch": 0.19241017359709325,
      "grad_norm": 86.93383026123047,
      "learning_rate": 1.86486806474859e-05,
      "loss": 6.2883,
      "step": 4766
    },
    {
      "epoch": 0.19245054501413,
      "grad_norm": 97.7171630859375,
      "learning_rate": 1.8648024169196146e-05,
      "loss": 4.4607,
      "step": 4767
    },
    {
      "epoch": 0.19249091643116673,
      "grad_norm": 139.7764129638672,
      "learning_rate": 1.864736754304538e-05,
      "loss": 6.3517,
      "step": 4768
    },
    {
      "epoch": 0.19253128784820348,
      "grad_norm": 105.7029037475586,
      "learning_rate": 1.864671076904482e-05,
      "loss": 6.4318,
      "step": 4769
    },
    {
      "epoch": 0.1925716592652402,
      "grad_norm": 97.43148040771484,
      "learning_rate": 1.8646053847205697e-05,
      "loss": 5.7567,
      "step": 4770
    },
    {
      "epoch": 0.19261203068227695,
      "grad_norm": 94.2164306640625,
      "learning_rate": 1.8645396777539246e-05,
      "loss": 4.0574,
      "step": 4771
    },
    {
      "epoch": 0.1926524020993137,
      "grad_norm": 108.70549774169922,
      "learning_rate": 1.8644739560056692e-05,
      "loss": 5.3346,
      "step": 4772
    },
    {
      "epoch": 0.19269277351635042,
      "grad_norm": 95.59941101074219,
      "learning_rate": 1.8644082194769284e-05,
      "loss": 6.4483,
      "step": 4773
    },
    {
      "epoch": 0.19273314493338717,
      "grad_norm": 113.7093734741211,
      "learning_rate": 1.8643424681688254e-05,
      "loss": 4.173,
      "step": 4774
    },
    {
      "epoch": 0.1927735163504239,
      "grad_norm": 86.1368179321289,
      "learning_rate": 1.8642767020824847e-05,
      "loss": 4.9902,
      "step": 4775
    },
    {
      "epoch": 0.19281388776746064,
      "grad_norm": 123.92668151855469,
      "learning_rate": 1.8642109212190305e-05,
      "loss": 4.1664,
      "step": 4776
    },
    {
      "epoch": 0.1928542591844974,
      "grad_norm": 148.00758361816406,
      "learning_rate": 1.8641451255795876e-05,
      "loss": 4.128,
      "step": 4777
    },
    {
      "epoch": 0.1928946306015341,
      "grad_norm": 182.20521545410156,
      "learning_rate": 1.8640793151652813e-05,
      "loss": 3.5608,
      "step": 4778
    },
    {
      "epoch": 0.19293500201857086,
      "grad_norm": 119.55738067626953,
      "learning_rate": 1.864013489977236e-05,
      "loss": 5.1038,
      "step": 4779
    },
    {
      "epoch": 0.19297537343560758,
      "grad_norm": 112.50438690185547,
      "learning_rate": 1.863947650016578e-05,
      "loss": 6.2284,
      "step": 4780
    },
    {
      "epoch": 0.19301574485264433,
      "grad_norm": 107.50910949707031,
      "learning_rate": 1.8638817952844325e-05,
      "loss": 5.9246,
      "step": 4781
    },
    {
      "epoch": 0.19305611626968106,
      "grad_norm": 184.98397827148438,
      "learning_rate": 1.8638159257819257e-05,
      "loss": 3.5058,
      "step": 4782
    },
    {
      "epoch": 0.1930964876867178,
      "grad_norm": 112.08236694335938,
      "learning_rate": 1.8637500415101834e-05,
      "loss": 6.8813,
      "step": 4783
    },
    {
      "epoch": 0.19313685910375455,
      "grad_norm": 118.9689712524414,
      "learning_rate": 1.863684142470333e-05,
      "loss": 6.7274,
      "step": 4784
    },
    {
      "epoch": 0.19317723052079128,
      "grad_norm": 97.1254653930664,
      "learning_rate": 1.8636182286635e-05,
      "loss": 4.8726,
      "step": 4785
    },
    {
      "epoch": 0.19321760193782803,
      "grad_norm": 147.84568786621094,
      "learning_rate": 1.8635523000908125e-05,
      "loss": 6.0432,
      "step": 4786
    },
    {
      "epoch": 0.19325797335486475,
      "grad_norm": 192.202392578125,
      "learning_rate": 1.8634863567533968e-05,
      "loss": 5.5632,
      "step": 4787
    },
    {
      "epoch": 0.1932983447719015,
      "grad_norm": 149.2974395751953,
      "learning_rate": 1.863420398652381e-05,
      "loss": 6.4126,
      "step": 4788
    },
    {
      "epoch": 0.19333871618893822,
      "grad_norm": 100.84990692138672,
      "learning_rate": 1.8633544257888926e-05,
      "loss": 6.2878,
      "step": 4789
    },
    {
      "epoch": 0.19337908760597497,
      "grad_norm": 107.88312530517578,
      "learning_rate": 1.8632884381640593e-05,
      "loss": 5.8616,
      "step": 4790
    },
    {
      "epoch": 0.19341945902301172,
      "grad_norm": 86.75746154785156,
      "learning_rate": 1.86322243577901e-05,
      "loss": 4.0253,
      "step": 4791
    },
    {
      "epoch": 0.19345983044004844,
      "grad_norm": 123.92704010009766,
      "learning_rate": 1.8631564186348727e-05,
      "loss": 5.1451,
      "step": 4792
    },
    {
      "epoch": 0.1935002018570852,
      "grad_norm": 110.49852752685547,
      "learning_rate": 1.863090386732776e-05,
      "loss": 5.6707,
      "step": 4793
    },
    {
      "epoch": 0.1935405732741219,
      "grad_norm": 217.33116149902344,
      "learning_rate": 1.8630243400738498e-05,
      "loss": 6.3241,
      "step": 4794
    },
    {
      "epoch": 0.19358094469115866,
      "grad_norm": 163.71144104003906,
      "learning_rate": 1.8629582786592223e-05,
      "loss": 4.5579,
      "step": 4795
    },
    {
      "epoch": 0.19362131610819538,
      "grad_norm": 73.92425537109375,
      "learning_rate": 1.8628922024900235e-05,
      "loss": 4.1355,
      "step": 4796
    },
    {
      "epoch": 0.19366168752523213,
      "grad_norm": 204.763671875,
      "learning_rate": 1.8628261115673827e-05,
      "loss": 5.3029,
      "step": 4797
    },
    {
      "epoch": 0.19370205894226888,
      "grad_norm": 123.63407897949219,
      "learning_rate": 1.8627600058924305e-05,
      "loss": 6.4943,
      "step": 4798
    },
    {
      "epoch": 0.1937424303593056,
      "grad_norm": 251.43995666503906,
      "learning_rate": 1.862693885466297e-05,
      "loss": 4.9278,
      "step": 4799
    },
    {
      "epoch": 0.19378280177634236,
      "grad_norm": 65.36033630371094,
      "learning_rate": 1.8626277502901122e-05,
      "loss": 7.036,
      "step": 4800
    },
    {
      "epoch": 0.19382317319337908,
      "grad_norm": 118.01177978515625,
      "learning_rate": 1.8625616003650075e-05,
      "loss": 6.0799,
      "step": 4801
    },
    {
      "epoch": 0.19386354461041583,
      "grad_norm": 161.36932373046875,
      "learning_rate": 1.8624954356921137e-05,
      "loss": 5.2954,
      "step": 4802
    },
    {
      "epoch": 0.19390391602745255,
      "grad_norm": 109.3651351928711,
      "learning_rate": 1.8624292562725622e-05,
      "loss": 4.4405,
      "step": 4803
    },
    {
      "epoch": 0.1939442874444893,
      "grad_norm": 111.38175201416016,
      "learning_rate": 1.862363062107484e-05,
      "loss": 6.6711,
      "step": 4804
    },
    {
      "epoch": 0.19398465886152605,
      "grad_norm": 85.81491088867188,
      "learning_rate": 1.8622968531980113e-05,
      "loss": 4.4597,
      "step": 4805
    },
    {
      "epoch": 0.19402503027856277,
      "grad_norm": 77.40961456298828,
      "learning_rate": 1.862230629545276e-05,
      "loss": 5.3881,
      "step": 4806
    },
    {
      "epoch": 0.19406540169559952,
      "grad_norm": 117.89636993408203,
      "learning_rate": 1.8621643911504103e-05,
      "loss": 5.5418,
      "step": 4807
    },
    {
      "epoch": 0.19410577311263624,
      "grad_norm": 151.41998291015625,
      "learning_rate": 1.8620981380145468e-05,
      "loss": 3.4444,
      "step": 4808
    },
    {
      "epoch": 0.194146144529673,
      "grad_norm": 84.79239654541016,
      "learning_rate": 1.8620318701388184e-05,
      "loss": 6.2378,
      "step": 4809
    },
    {
      "epoch": 0.19418651594670974,
      "grad_norm": 90.66378784179688,
      "learning_rate": 1.8619655875243584e-05,
      "loss": 5.4793,
      "step": 4810
    },
    {
      "epoch": 0.19422688736374646,
      "grad_norm": 139.84805297851562,
      "learning_rate": 1.861899290172299e-05,
      "loss": 5.9123,
      "step": 4811
    },
    {
      "epoch": 0.19426725878078321,
      "grad_norm": 120.90171813964844,
      "learning_rate": 1.8618329780837746e-05,
      "loss": 6.4242,
      "step": 4812
    },
    {
      "epoch": 0.19430763019781994,
      "grad_norm": 92.70111846923828,
      "learning_rate": 1.861766651259919e-05,
      "loss": 5.0835,
      "step": 4813
    },
    {
      "epoch": 0.19434800161485669,
      "grad_norm": 97.79780578613281,
      "learning_rate": 1.861700309701866e-05,
      "loss": 4.0319,
      "step": 4814
    },
    {
      "epoch": 0.1943883730318934,
      "grad_norm": 91.77044677734375,
      "learning_rate": 1.86163395341075e-05,
      "loss": 4.613,
      "step": 4815
    },
    {
      "epoch": 0.19442874444893016,
      "grad_norm": 143.71621704101562,
      "learning_rate": 1.8615675823877052e-05,
      "loss": 3.4671,
      "step": 4816
    },
    {
      "epoch": 0.1944691158659669,
      "grad_norm": 138.08126831054688,
      "learning_rate": 1.8615011966338667e-05,
      "loss": 4.2247,
      "step": 4817
    },
    {
      "epoch": 0.19450948728300363,
      "grad_norm": 85.08323669433594,
      "learning_rate": 1.8614347961503695e-05,
      "loss": 6.9165,
      "step": 4818
    },
    {
      "epoch": 0.19454985870004038,
      "grad_norm": 76.54813385009766,
      "learning_rate": 1.8613683809383493e-05,
      "loss": 6.2504,
      "step": 4819
    },
    {
      "epoch": 0.1945902301170771,
      "grad_norm": 102.93978118896484,
      "learning_rate": 1.8613019509989408e-05,
      "loss": 6.2611,
      "step": 4820
    },
    {
      "epoch": 0.19463060153411385,
      "grad_norm": 149.5154571533203,
      "learning_rate": 1.86123550633328e-05,
      "loss": 4.0092,
      "step": 4821
    },
    {
      "epoch": 0.19467097295115057,
      "grad_norm": 145.53927612304688,
      "learning_rate": 1.8611690469425035e-05,
      "loss": 6.9646,
      "step": 4822
    },
    {
      "epoch": 0.19471134436818732,
      "grad_norm": 168.66494750976562,
      "learning_rate": 1.8611025728277475e-05,
      "loss": 4.0747,
      "step": 4823
    },
    {
      "epoch": 0.19475171578522407,
      "grad_norm": 89.15113830566406,
      "learning_rate": 1.8610360839901476e-05,
      "loss": 6.2204,
      "step": 4824
    },
    {
      "epoch": 0.1947920872022608,
      "grad_norm": 91.0371322631836,
      "learning_rate": 1.860969580430842e-05,
      "loss": 5.1379,
      "step": 4825
    },
    {
      "epoch": 0.19483245861929754,
      "grad_norm": 104.37476348876953,
      "learning_rate": 1.860903062150967e-05,
      "loss": 5.1733,
      "step": 4826
    },
    {
      "epoch": 0.19487283003633427,
      "grad_norm": 131.3164825439453,
      "learning_rate": 1.8608365291516598e-05,
      "loss": 7.1906,
      "step": 4827
    },
    {
      "epoch": 0.19491320145337102,
      "grad_norm": 134.3671875,
      "learning_rate": 1.860769981434058e-05,
      "loss": 5.9872,
      "step": 4828
    },
    {
      "epoch": 0.19495357287040774,
      "grad_norm": 97.8207778930664,
      "learning_rate": 1.8607034189993e-05,
      "loss": 5.5709,
      "step": 4829
    },
    {
      "epoch": 0.1949939442874445,
      "grad_norm": 90.6044692993164,
      "learning_rate": 1.860636841848523e-05,
      "loss": 4.7356,
      "step": 4830
    },
    {
      "epoch": 0.19503431570448124,
      "grad_norm": 123.28828430175781,
      "learning_rate": 1.8605702499828663e-05,
      "loss": 5.6961,
      "step": 4831
    },
    {
      "epoch": 0.19507468712151796,
      "grad_norm": 119.25023651123047,
      "learning_rate": 1.8605036434034672e-05,
      "loss": 6.7978,
      "step": 4832
    },
    {
      "epoch": 0.1951150585385547,
      "grad_norm": 115.59984588623047,
      "learning_rate": 1.8604370221114656e-05,
      "loss": 4.9137,
      "step": 4833
    },
    {
      "epoch": 0.19515542995559143,
      "grad_norm": 315.84136962890625,
      "learning_rate": 1.860370386108e-05,
      "loss": 6.5034,
      "step": 4834
    },
    {
      "epoch": 0.19519580137262818,
      "grad_norm": 90.6501693725586,
      "learning_rate": 1.86030373539421e-05,
      "loss": 3.8689,
      "step": 4835
    },
    {
      "epoch": 0.19523617278966493,
      "grad_norm": 76.4385757446289,
      "learning_rate": 1.8602370699712354e-05,
      "loss": 4.6205,
      "step": 4836
    },
    {
      "epoch": 0.19527654420670165,
      "grad_norm": 130.45411682128906,
      "learning_rate": 1.860170389840215e-05,
      "loss": 6.8073,
      "step": 4837
    },
    {
      "epoch": 0.1953169156237384,
      "grad_norm": 99.59048461914062,
      "learning_rate": 1.8601036950022903e-05,
      "loss": 6.0955,
      "step": 4838
    },
    {
      "epoch": 0.19535728704077512,
      "grad_norm": 131.6742706298828,
      "learning_rate": 1.8600369854586005e-05,
      "loss": 5.8296,
      "step": 4839
    },
    {
      "epoch": 0.19539765845781187,
      "grad_norm": 144.8249969482422,
      "learning_rate": 1.8599702612102867e-05,
      "loss": 5.0814,
      "step": 4840
    },
    {
      "epoch": 0.1954380298748486,
      "grad_norm": 102.74249267578125,
      "learning_rate": 1.8599035222584895e-05,
      "loss": 6.0061,
      "step": 4841
    },
    {
      "epoch": 0.19547840129188535,
      "grad_norm": 107.20014190673828,
      "learning_rate": 1.8598367686043505e-05,
      "loss": 5.4693,
      "step": 4842
    },
    {
      "epoch": 0.1955187727089221,
      "grad_norm": 85.35926818847656,
      "learning_rate": 1.85977000024901e-05,
      "loss": 5.0402,
      "step": 4843
    },
    {
      "epoch": 0.19555914412595882,
      "grad_norm": 127.25595092773438,
      "learning_rate": 1.8597032171936103e-05,
      "loss": 5.1303,
      "step": 4844
    },
    {
      "epoch": 0.19559951554299557,
      "grad_norm": 124.31324005126953,
      "learning_rate": 1.859636419439294e-05,
      "loss": 6.5898,
      "step": 4845
    },
    {
      "epoch": 0.1956398869600323,
      "grad_norm": 106.85125732421875,
      "learning_rate": 1.8595696069872013e-05,
      "loss": 5.932,
      "step": 4846
    },
    {
      "epoch": 0.19568025837706904,
      "grad_norm": 147.33921813964844,
      "learning_rate": 1.859502779838476e-05,
      "loss": 4.7711,
      "step": 4847
    },
    {
      "epoch": 0.19572062979410576,
      "grad_norm": 96.01114654541016,
      "learning_rate": 1.8594359379942602e-05,
      "loss": 6.0941,
      "step": 4848
    },
    {
      "epoch": 0.1957610012111425,
      "grad_norm": 103.79877471923828,
      "learning_rate": 1.8593690814556966e-05,
      "loss": 6.3345,
      "step": 4849
    },
    {
      "epoch": 0.19580137262817926,
      "grad_norm": 102.62596893310547,
      "learning_rate": 1.8593022102239286e-05,
      "loss": 6.6928,
      "step": 4850
    },
    {
      "epoch": 0.19584174404521598,
      "grad_norm": 136.83547973632812,
      "learning_rate": 1.8592353243000996e-05,
      "loss": 4.8405,
      "step": 4851
    },
    {
      "epoch": 0.19588211546225273,
      "grad_norm": 88.58300018310547,
      "learning_rate": 1.8591684236853526e-05,
      "loss": 5.4382,
      "step": 4852
    },
    {
      "epoch": 0.19592248687928945,
      "grad_norm": 78.21292877197266,
      "learning_rate": 1.8591015083808323e-05,
      "loss": 6.2926,
      "step": 4853
    },
    {
      "epoch": 0.1959628582963262,
      "grad_norm": 89.00082397460938,
      "learning_rate": 1.8590345783876823e-05,
      "loss": 6.6365,
      "step": 4854
    },
    {
      "epoch": 0.19600322971336293,
      "grad_norm": 214.3539276123047,
      "learning_rate": 1.8589676337070466e-05,
      "loss": 4.5951,
      "step": 4855
    },
    {
      "epoch": 0.19604360113039968,
      "grad_norm": 103.34227752685547,
      "learning_rate": 1.8589006743400707e-05,
      "loss": 6.3539,
      "step": 4856
    },
    {
      "epoch": 0.19608397254743642,
      "grad_norm": 93.33203887939453,
      "learning_rate": 1.8588337002878988e-05,
      "loss": 6.6002,
      "step": 4857
    },
    {
      "epoch": 0.19612434396447315,
      "grad_norm": 68.30083465576172,
      "learning_rate": 1.858766711551676e-05,
      "loss": 6.7598,
      "step": 4858
    },
    {
      "epoch": 0.1961647153815099,
      "grad_norm": 109.84249877929688,
      "learning_rate": 1.858699708132548e-05,
      "loss": 5.0056,
      "step": 4859
    },
    {
      "epoch": 0.19620508679854662,
      "grad_norm": 112.03909301757812,
      "learning_rate": 1.8586326900316598e-05,
      "loss": 6.1684,
      "step": 4860
    },
    {
      "epoch": 0.19624545821558337,
      "grad_norm": 217.8072052001953,
      "learning_rate": 1.858565657250158e-05,
      "loss": 5.2455,
      "step": 4861
    },
    {
      "epoch": 0.19628582963262012,
      "grad_norm": 264.54534912109375,
      "learning_rate": 1.8584986097891883e-05,
      "loss": 4.5285,
      "step": 4862
    },
    {
      "epoch": 0.19632620104965684,
      "grad_norm": 66.34380340576172,
      "learning_rate": 1.858431547649897e-05,
      "loss": 5.0854,
      "step": 4863
    },
    {
      "epoch": 0.1963665724666936,
      "grad_norm": 85.09481811523438,
      "learning_rate": 1.858364470833431e-05,
      "loss": 6.3892,
      "step": 4864
    },
    {
      "epoch": 0.1964069438837303,
      "grad_norm": 95.25362396240234,
      "learning_rate": 1.8582973793409366e-05,
      "loss": 4.208,
      "step": 4865
    },
    {
      "epoch": 0.19644731530076706,
      "grad_norm": 101.05084228515625,
      "learning_rate": 1.8582302731735614e-05,
      "loss": 5.266,
      "step": 4866
    },
    {
      "epoch": 0.19648768671780378,
      "grad_norm": 165.27308654785156,
      "learning_rate": 1.858163152332453e-05,
      "loss": 4.1829,
      "step": 4867
    },
    {
      "epoch": 0.19652805813484053,
      "grad_norm": 105.19932556152344,
      "learning_rate": 1.8580960168187585e-05,
      "loss": 5.4586,
      "step": 4868
    },
    {
      "epoch": 0.19656842955187728,
      "grad_norm": 102.7529525756836,
      "learning_rate": 1.858028866633626e-05,
      "loss": 4.5976,
      "step": 4869
    },
    {
      "epoch": 0.196608800968914,
      "grad_norm": 106.1096420288086,
      "learning_rate": 1.857961701778203e-05,
      "loss": 5.8896,
      "step": 4870
    },
    {
      "epoch": 0.19664917238595075,
      "grad_norm": 93.889892578125,
      "learning_rate": 1.857894522253639e-05,
      "loss": 5.3455,
      "step": 4871
    },
    {
      "epoch": 0.19668954380298748,
      "grad_norm": 87.29039001464844,
      "learning_rate": 1.857827328061082e-05,
      "loss": 5.4903,
      "step": 4872
    },
    {
      "epoch": 0.19672991522002423,
      "grad_norm": 93.6878433227539,
      "learning_rate": 1.85776011920168e-05,
      "loss": 4.6226,
      "step": 4873
    },
    {
      "epoch": 0.19677028663706095,
      "grad_norm": 131.0951385498047,
      "learning_rate": 1.8576928956765837e-05,
      "loss": 5.5362,
      "step": 4874
    },
    {
      "epoch": 0.1968106580540977,
      "grad_norm": 110.63201904296875,
      "learning_rate": 1.857625657486942e-05,
      "loss": 5.7222,
      "step": 4875
    },
    {
      "epoch": 0.19685102947113445,
      "grad_norm": 109.08722686767578,
      "learning_rate": 1.8575584046339036e-05,
      "loss": 5.1905,
      "step": 4876
    },
    {
      "epoch": 0.19689140088817117,
      "grad_norm": 89.2757568359375,
      "learning_rate": 1.8574911371186193e-05,
      "loss": 5.078,
      "step": 4877
    },
    {
      "epoch": 0.19693177230520792,
      "grad_norm": 115.27236938476562,
      "learning_rate": 1.857423854942239e-05,
      "loss": 5.3533,
      "step": 4878
    },
    {
      "epoch": 0.19697214372224464,
      "grad_norm": 127.20756530761719,
      "learning_rate": 1.857356558105913e-05,
      "loss": 4.2185,
      "step": 4879
    },
    {
      "epoch": 0.1970125151392814,
      "grad_norm": 117.1166763305664,
      "learning_rate": 1.857289246610792e-05,
      "loss": 4.5098,
      "step": 4880
    },
    {
      "epoch": 0.1970528865563181,
      "grad_norm": 95.00489044189453,
      "learning_rate": 1.8572219204580266e-05,
      "loss": 5.1725,
      "step": 4881
    },
    {
      "epoch": 0.19709325797335486,
      "grad_norm": 121.29075622558594,
      "learning_rate": 1.8571545796487682e-05,
      "loss": 5.3183,
      "step": 4882
    },
    {
      "epoch": 0.1971336293903916,
      "grad_norm": 125.04036712646484,
      "learning_rate": 1.857087224184168e-05,
      "loss": 6.1597,
      "step": 4883
    },
    {
      "epoch": 0.19717400080742833,
      "grad_norm": 107.12779235839844,
      "learning_rate": 1.857019854065378e-05,
      "loss": 6.8605,
      "step": 4884
    },
    {
      "epoch": 0.19721437222446508,
      "grad_norm": 101.03242492675781,
      "learning_rate": 1.8569524692935497e-05,
      "loss": 5.2071,
      "step": 4885
    },
    {
      "epoch": 0.1972547436415018,
      "grad_norm": 109.64463806152344,
      "learning_rate": 1.856885069869835e-05,
      "loss": 6.3251,
      "step": 4886
    },
    {
      "epoch": 0.19729511505853856,
      "grad_norm": 109.06224822998047,
      "learning_rate": 1.8568176557953867e-05,
      "loss": 5.6485,
      "step": 4887
    },
    {
      "epoch": 0.1973354864755753,
      "grad_norm": 281.6580505371094,
      "learning_rate": 1.8567502270713573e-05,
      "loss": 4.3484,
      "step": 4888
    },
    {
      "epoch": 0.19737585789261203,
      "grad_norm": 131.00616455078125,
      "learning_rate": 1.8566827836989e-05,
      "loss": 6.6216,
      "step": 4889
    },
    {
      "epoch": 0.19741622930964878,
      "grad_norm": 121.8570327758789,
      "learning_rate": 1.8566153256791672e-05,
      "loss": 4.7406,
      "step": 4890
    },
    {
      "epoch": 0.1974566007266855,
      "grad_norm": 143.43882751464844,
      "learning_rate": 1.8565478530133128e-05,
      "loss": 5.7331,
      "step": 4891
    },
    {
      "epoch": 0.19749697214372225,
      "grad_norm": 131.1141357421875,
      "learning_rate": 1.85648036570249e-05,
      "loss": 5.1621,
      "step": 4892
    },
    {
      "epoch": 0.19753734356075897,
      "grad_norm": 99.57818603515625,
      "learning_rate": 1.8564128637478535e-05,
      "loss": 4.3198,
      "step": 4893
    },
    {
      "epoch": 0.19757771497779572,
      "grad_norm": 114.64282989501953,
      "learning_rate": 1.8563453471505567e-05,
      "loss": 5.6482,
      "step": 4894
    },
    {
      "epoch": 0.19761808639483247,
      "grad_norm": 99.34489440917969,
      "learning_rate": 1.8562778159117542e-05,
      "loss": 5.6012,
      "step": 4895
    },
    {
      "epoch": 0.1976584578118692,
      "grad_norm": 128.09109497070312,
      "learning_rate": 1.8562102700326002e-05,
      "loss": 5.9626,
      "step": 4896
    },
    {
      "epoch": 0.19769882922890594,
      "grad_norm": 192.43115234375,
      "learning_rate": 1.8561427095142507e-05,
      "loss": 5.8135,
      "step": 4897
    },
    {
      "epoch": 0.19773920064594266,
      "grad_norm": 95.13568878173828,
      "learning_rate": 1.8560751343578596e-05,
      "loss": 6.897,
      "step": 4898
    },
    {
      "epoch": 0.19777957206297941,
      "grad_norm": 94.951416015625,
      "learning_rate": 1.856007544564583e-05,
      "loss": 6.043,
      "step": 4899
    },
    {
      "epoch": 0.19781994348001614,
      "grad_norm": 93.05984497070312,
      "learning_rate": 1.8559399401355762e-05,
      "loss": 4.8024,
      "step": 4900
    },
    {
      "epoch": 0.19786031489705289,
      "grad_norm": 93.48007202148438,
      "learning_rate": 1.8558723210719953e-05,
      "loss": 4.3908,
      "step": 4901
    },
    {
      "epoch": 0.19790068631408964,
      "grad_norm": 93.48780059814453,
      "learning_rate": 1.8558046873749962e-05,
      "loss": 5.4533,
      "step": 4902
    },
    {
      "epoch": 0.19794105773112636,
      "grad_norm": 104.57998657226562,
      "learning_rate": 1.855737039045735e-05,
      "loss": 5.2692,
      "step": 4903
    },
    {
      "epoch": 0.1979814291481631,
      "grad_norm": 163.77503967285156,
      "learning_rate": 1.8556693760853693e-05,
      "loss": 4.808,
      "step": 4904
    },
    {
      "epoch": 0.19802180056519983,
      "grad_norm": 164.2623291015625,
      "learning_rate": 1.8556016984950552e-05,
      "loss": 6.6342,
      "step": 4905
    },
    {
      "epoch": 0.19806217198223658,
      "grad_norm": 81.42506408691406,
      "learning_rate": 1.85553400627595e-05,
      "loss": 6.6835,
      "step": 4906
    },
    {
      "epoch": 0.1981025433992733,
      "grad_norm": 95.14846801757812,
      "learning_rate": 1.8554662994292112e-05,
      "loss": 5.0887,
      "step": 4907
    },
    {
      "epoch": 0.19814291481631005,
      "grad_norm": 120.41236114501953,
      "learning_rate": 1.8553985779559964e-05,
      "loss": 5.5494,
      "step": 4908
    },
    {
      "epoch": 0.1981832862333468,
      "grad_norm": 116.59016418457031,
      "learning_rate": 1.8553308418574635e-05,
      "loss": 6.3448,
      "step": 4909
    },
    {
      "epoch": 0.19822365765038352,
      "grad_norm": 104.17057037353516,
      "learning_rate": 1.8552630911347703e-05,
      "loss": 5.8219,
      "step": 4910
    },
    {
      "epoch": 0.19826402906742027,
      "grad_norm": 155.28140258789062,
      "learning_rate": 1.8551953257890757e-05,
      "loss": 5.3571,
      "step": 4911
    },
    {
      "epoch": 0.198304400484457,
      "grad_norm": 120.52297973632812,
      "learning_rate": 1.8551275458215373e-05,
      "loss": 4.6345,
      "step": 4912
    },
    {
      "epoch": 0.19834477190149374,
      "grad_norm": 72.71697998046875,
      "learning_rate": 1.8550597512333153e-05,
      "loss": 5.4637,
      "step": 4913
    },
    {
      "epoch": 0.1983851433185305,
      "grad_norm": 88.30501556396484,
      "learning_rate": 1.8549919420255683e-05,
      "loss": 4.8317,
      "step": 4914
    },
    {
      "epoch": 0.19842551473556722,
      "grad_norm": 96.79143524169922,
      "learning_rate": 1.8549241181994557e-05,
      "loss": 5.458,
      "step": 4915
    },
    {
      "epoch": 0.19846588615260397,
      "grad_norm": 73.23812103271484,
      "learning_rate": 1.8548562797561368e-05,
      "loss": 5.8353,
      "step": 4916
    },
    {
      "epoch": 0.1985062575696407,
      "grad_norm": 199.05007934570312,
      "learning_rate": 1.854788426696772e-05,
      "loss": 6.886,
      "step": 4917
    },
    {
      "epoch": 0.19854662898667744,
      "grad_norm": 106.04366302490234,
      "learning_rate": 1.8547205590225214e-05,
      "loss": 5.2212,
      "step": 4918
    },
    {
      "epoch": 0.19858700040371416,
      "grad_norm": 75.47307586669922,
      "learning_rate": 1.8546526767345447e-05,
      "loss": 6.0323,
      "step": 4919
    },
    {
      "epoch": 0.1986273718207509,
      "grad_norm": 126.6998519897461,
      "learning_rate": 1.854584779834003e-05,
      "loss": 5.9622,
      "step": 4920
    },
    {
      "epoch": 0.19866774323778766,
      "grad_norm": 95.21817016601562,
      "learning_rate": 1.8545168683220572e-05,
      "loss": 5.8852,
      "step": 4921
    },
    {
      "epoch": 0.19870811465482438,
      "grad_norm": 80.60272979736328,
      "learning_rate": 1.8544489421998687e-05,
      "loss": 5.0958,
      "step": 4922
    },
    {
      "epoch": 0.19874848607186113,
      "grad_norm": 98.78398132324219,
      "learning_rate": 1.8543810014685984e-05,
      "loss": 5.3289,
      "step": 4923
    },
    {
      "epoch": 0.19878885748889785,
      "grad_norm": 133.17881774902344,
      "learning_rate": 1.854313046129408e-05,
      "loss": 6.8013,
      "step": 4924
    },
    {
      "epoch": 0.1988292289059346,
      "grad_norm": 107.18929290771484,
      "learning_rate": 1.8542450761834593e-05,
      "loss": 4.9786,
      "step": 4925
    },
    {
      "epoch": 0.19886960032297132,
      "grad_norm": 125.11835479736328,
      "learning_rate": 1.8541770916319152e-05,
      "loss": 4.4164,
      "step": 4926
    },
    {
      "epoch": 0.19890997174000807,
      "grad_norm": 92.29890441894531,
      "learning_rate": 1.8541090924759368e-05,
      "loss": 5.0853,
      "step": 4927
    },
    {
      "epoch": 0.19895034315704482,
      "grad_norm": 97.93994903564453,
      "learning_rate": 1.8540410787166878e-05,
      "loss": 5.8646,
      "step": 4928
    },
    {
      "epoch": 0.19899071457408155,
      "grad_norm": 230.45407104492188,
      "learning_rate": 1.8539730503553305e-05,
      "loss": 4.002,
      "step": 4929
    },
    {
      "epoch": 0.1990310859911183,
      "grad_norm": 100.95801544189453,
      "learning_rate": 1.8539050073930282e-05,
      "loss": 4.6876,
      "step": 4930
    },
    {
      "epoch": 0.19907145740815502,
      "grad_norm": 86.30387878417969,
      "learning_rate": 1.8538369498309446e-05,
      "loss": 5.3025,
      "step": 4931
    },
    {
      "epoch": 0.19911182882519177,
      "grad_norm": 153.69625854492188,
      "learning_rate": 1.8537688776702427e-05,
      "loss": 5.7149,
      "step": 4932
    },
    {
      "epoch": 0.1991522002422285,
      "grad_norm": 117.7066879272461,
      "learning_rate": 1.853700790912087e-05,
      "loss": 7.0081,
      "step": 4933
    },
    {
      "epoch": 0.19919257165926524,
      "grad_norm": 105.77389526367188,
      "learning_rate": 1.853632689557641e-05,
      "loss": 5.9021,
      "step": 4934
    },
    {
      "epoch": 0.199232943076302,
      "grad_norm": 140.06494140625,
      "learning_rate": 1.8535645736080693e-05,
      "loss": 6.0098,
      "step": 4935
    },
    {
      "epoch": 0.1992733144933387,
      "grad_norm": 125.19619750976562,
      "learning_rate": 1.853496443064537e-05,
      "loss": 7.0043,
      "step": 4936
    },
    {
      "epoch": 0.19931368591037546,
      "grad_norm": 255.2788543701172,
      "learning_rate": 1.8534282979282082e-05,
      "loss": 5.4391,
      "step": 4937
    },
    {
      "epoch": 0.19935405732741218,
      "grad_norm": 215.369384765625,
      "learning_rate": 1.8533601382002485e-05,
      "loss": 6.1485,
      "step": 4938
    },
    {
      "epoch": 0.19939442874444893,
      "grad_norm": 112.17334747314453,
      "learning_rate": 1.8532919638818234e-05,
      "loss": 4.9055,
      "step": 4939
    },
    {
      "epoch": 0.19943480016148568,
      "grad_norm": 84.7526626586914,
      "learning_rate": 1.853223774974098e-05,
      "loss": 5.8251,
      "step": 4940
    },
    {
      "epoch": 0.1994751715785224,
      "grad_norm": 106.96025848388672,
      "learning_rate": 1.8531555714782387e-05,
      "loss": 4.5875,
      "step": 4941
    },
    {
      "epoch": 0.19951554299555915,
      "grad_norm": 148.27439880371094,
      "learning_rate": 1.8530873533954112e-05,
      "loss": 4.0548,
      "step": 4942
    },
    {
      "epoch": 0.19955591441259588,
      "grad_norm": 220.9293212890625,
      "learning_rate": 1.8530191207267823e-05,
      "loss": 4.1998,
      "step": 4943
    },
    {
      "epoch": 0.19959628582963262,
      "grad_norm": 138.3700408935547,
      "learning_rate": 1.8529508734735184e-05,
      "loss": 4.9117,
      "step": 4944
    },
    {
      "epoch": 0.19963665724666935,
      "grad_norm": 98.98326110839844,
      "learning_rate": 1.852882611636786e-05,
      "loss": 5.6253,
      "step": 4945
    },
    {
      "epoch": 0.1996770286637061,
      "grad_norm": 91.97439575195312,
      "learning_rate": 1.8528143352177526e-05,
      "loss": 6.5688,
      "step": 4946
    },
    {
      "epoch": 0.19971740008074285,
      "grad_norm": 87.86157989501953,
      "learning_rate": 1.852746044217586e-05,
      "loss": 4.0512,
      "step": 4947
    },
    {
      "epoch": 0.19975777149777957,
      "grad_norm": 85.9807357788086,
      "learning_rate": 1.852677738637453e-05,
      "loss": 3.4069,
      "step": 4948
    },
    {
      "epoch": 0.19979814291481632,
      "grad_norm": 99.21221923828125,
      "learning_rate": 1.852609418478522e-05,
      "loss": 5.1255,
      "step": 4949
    },
    {
      "epoch": 0.19983851433185304,
      "grad_norm": 73.14861297607422,
      "learning_rate": 1.8525410837419606e-05,
      "loss": 5.4844,
      "step": 4950
    },
    {
      "epoch": 0.1998788857488898,
      "grad_norm": 67.09971618652344,
      "learning_rate": 1.8524727344289375e-05,
      "loss": 6.5754,
      "step": 4951
    },
    {
      "epoch": 0.1999192571659265,
      "grad_norm": 87.83251953125,
      "learning_rate": 1.8524043705406215e-05,
      "loss": 5.1838,
      "step": 4952
    },
    {
      "epoch": 0.19995962858296326,
      "grad_norm": 99.32026672363281,
      "learning_rate": 1.8523359920781813e-05,
      "loss": 5.3016,
      "step": 4953
    },
    {
      "epoch": 0.2,
      "grad_norm": 77.6196517944336,
      "learning_rate": 1.852267599042786e-05,
      "loss": 5.7782,
      "step": 4954
    },
    {
      "epoch": 0.20004037141703673,
      "grad_norm": 92.8257064819336,
      "learning_rate": 1.852199191435605e-05,
      "loss": 3.9632,
      "step": 4955
    },
    {
      "epoch": 0.20008074283407348,
      "grad_norm": 107.4832992553711,
      "learning_rate": 1.8521307692578076e-05,
      "loss": 5.7624,
      "step": 4956
    },
    {
      "epoch": 0.2001211142511102,
      "grad_norm": 225.6306915283203,
      "learning_rate": 1.852062332510564e-05,
      "loss": 4.6567,
      "step": 4957
    },
    {
      "epoch": 0.20016148566814695,
      "grad_norm": 126.31206512451172,
      "learning_rate": 1.8519938811950444e-05,
      "loss": 4.9306,
      "step": 4958
    },
    {
      "epoch": 0.20020185708518368,
      "grad_norm": 81.8460922241211,
      "learning_rate": 1.851925415312419e-05,
      "loss": 4.0029,
      "step": 4959
    },
    {
      "epoch": 0.20024222850222043,
      "grad_norm": 106.06522369384766,
      "learning_rate": 1.851856934863858e-05,
      "loss": 6.157,
      "step": 4960
    },
    {
      "epoch": 0.20028259991925718,
      "grad_norm": 125.34764862060547,
      "learning_rate": 1.851788439850533e-05,
      "loss": 5.5344,
      "step": 4961
    },
    {
      "epoch": 0.2003229713362939,
      "grad_norm": 190.96994018554688,
      "learning_rate": 1.8517199302736145e-05,
      "loss": 4.7176,
      "step": 4962
    },
    {
      "epoch": 0.20036334275333065,
      "grad_norm": 81.70188903808594,
      "learning_rate": 1.8516514061342742e-05,
      "loss": 5.1931,
      "step": 4963
    },
    {
      "epoch": 0.20040371417036737,
      "grad_norm": 238.18475341796875,
      "learning_rate": 1.8515828674336837e-05,
      "loss": 6.0363,
      "step": 4964
    },
    {
      "epoch": 0.20044408558740412,
      "grad_norm": 134.2950897216797,
      "learning_rate": 1.8515143141730146e-05,
      "loss": 6.0611,
      "step": 4965
    },
    {
      "epoch": 0.20048445700444087,
      "grad_norm": 157.68072509765625,
      "learning_rate": 1.8514457463534392e-05,
      "loss": 5.1347,
      "step": 4966
    },
    {
      "epoch": 0.2005248284214776,
      "grad_norm": 117.1812744140625,
      "learning_rate": 1.85137716397613e-05,
      "loss": 6.5047,
      "step": 4967
    },
    {
      "epoch": 0.20056519983851434,
      "grad_norm": 115.18313598632812,
      "learning_rate": 1.8513085670422593e-05,
      "loss": 5.3937,
      "step": 4968
    },
    {
      "epoch": 0.20060557125555106,
      "grad_norm": 172.46092224121094,
      "learning_rate": 1.851239955553e-05,
      "loss": 6.0024,
      "step": 4969
    },
    {
      "epoch": 0.2006459426725878,
      "grad_norm": 88.94463348388672,
      "learning_rate": 1.851171329509525e-05,
      "loss": 4.9316,
      "step": 4970
    },
    {
      "epoch": 0.20068631408962453,
      "grad_norm": 111.26085662841797,
      "learning_rate": 1.8511026889130082e-05,
      "loss": 5.0542,
      "step": 4971
    },
    {
      "epoch": 0.20072668550666128,
      "grad_norm": 82.51070404052734,
      "learning_rate": 1.8510340337646232e-05,
      "loss": 4.8279,
      "step": 4972
    },
    {
      "epoch": 0.20076705692369803,
      "grad_norm": 106.73870849609375,
      "learning_rate": 1.850965364065543e-05,
      "loss": 3.8367,
      "step": 4973
    },
    {
      "epoch": 0.20080742834073476,
      "grad_norm": 151.7615509033203,
      "learning_rate": 1.8508966798169425e-05,
      "loss": 4.0257,
      "step": 4974
    },
    {
      "epoch": 0.2008477997577715,
      "grad_norm": 104.86776733398438,
      "learning_rate": 1.8508279810199953e-05,
      "loss": 5.7915,
      "step": 4975
    },
    {
      "epoch": 0.20088817117480823,
      "grad_norm": 80.43273162841797,
      "learning_rate": 1.850759267675877e-05,
      "loss": 5.745,
      "step": 4976
    },
    {
      "epoch": 0.20092854259184498,
      "grad_norm": 132.6168670654297,
      "learning_rate": 1.8506905397857617e-05,
      "loss": 6.2945,
      "step": 4977
    },
    {
      "epoch": 0.2009689140088817,
      "grad_norm": 104.23455810546875,
      "learning_rate": 1.8506217973508247e-05,
      "loss": 6.2846,
      "step": 4978
    },
    {
      "epoch": 0.20100928542591845,
      "grad_norm": 105.3067626953125,
      "learning_rate": 1.8505530403722414e-05,
      "loss": 4.8199,
      "step": 4979
    },
    {
      "epoch": 0.2010496568429552,
      "grad_norm": 346.0680847167969,
      "learning_rate": 1.850484268851187e-05,
      "loss": 4.4398,
      "step": 4980
    },
    {
      "epoch": 0.20109002825999192,
      "grad_norm": 122.01172637939453,
      "learning_rate": 1.850415482788838e-05,
      "loss": 6.8835,
      "step": 4981
    },
    {
      "epoch": 0.20113039967702867,
      "grad_norm": 87.10564422607422,
      "learning_rate": 1.85034668218637e-05,
      "loss": 4.8814,
      "step": 4982
    },
    {
      "epoch": 0.2011707710940654,
      "grad_norm": 87.20597839355469,
      "learning_rate": 1.850277867044959e-05,
      "loss": 5.5494,
      "step": 4983
    },
    {
      "epoch": 0.20121114251110214,
      "grad_norm": 96.57704162597656,
      "learning_rate": 1.8502090373657824e-05,
      "loss": 5.8068,
      "step": 4984
    },
    {
      "epoch": 0.20125151392813886,
      "grad_norm": 224.09494018554688,
      "learning_rate": 1.8501401931500165e-05,
      "loss": 5.5208,
      "step": 4985
    },
    {
      "epoch": 0.20129188534517561,
      "grad_norm": 143.15467834472656,
      "learning_rate": 1.8500713343988387e-05,
      "loss": 4.9675,
      "step": 4986
    },
    {
      "epoch": 0.20133225676221236,
      "grad_norm": 78.62347412109375,
      "learning_rate": 1.8500024611134263e-05,
      "loss": 5.0604,
      "step": 4987
    },
    {
      "epoch": 0.20137262817924909,
      "grad_norm": 110.84603118896484,
      "learning_rate": 1.8499335732949563e-05,
      "loss": 5.7895,
      "step": 4988
    },
    {
      "epoch": 0.20141299959628584,
      "grad_norm": 166.09017944335938,
      "learning_rate": 1.8498646709446067e-05,
      "loss": 5.8708,
      "step": 4989
    },
    {
      "epoch": 0.20145337101332256,
      "grad_norm": 136.34690856933594,
      "learning_rate": 1.8497957540635564e-05,
      "loss": 5.9919,
      "step": 4990
    },
    {
      "epoch": 0.2014937424303593,
      "grad_norm": 134.1013946533203,
      "learning_rate": 1.8497268226529833e-05,
      "loss": 6.496,
      "step": 4991
    },
    {
      "epoch": 0.20153411384739606,
      "grad_norm": 97.89450073242188,
      "learning_rate": 1.8496578767140652e-05,
      "loss": 5.0461,
      "step": 4992
    },
    {
      "epoch": 0.20157448526443278,
      "grad_norm": 120.35967254638672,
      "learning_rate": 1.849588916247982e-05,
      "loss": 4.4852,
      "step": 4993
    },
    {
      "epoch": 0.20161485668146953,
      "grad_norm": 91.1749038696289,
      "learning_rate": 1.8495199412559116e-05,
      "loss": 6.8275,
      "step": 4994
    },
    {
      "epoch": 0.20165522809850625,
      "grad_norm": 258.68505859375,
      "learning_rate": 1.8494509517390345e-05,
      "loss": 6.1002,
      "step": 4995
    },
    {
      "epoch": 0.201695599515543,
      "grad_norm": 123.55696105957031,
      "learning_rate": 1.8493819476985294e-05,
      "loss": 5.6169,
      "step": 4996
    },
    {
      "epoch": 0.20173597093257972,
      "grad_norm": 103.75286865234375,
      "learning_rate": 1.849312929135577e-05,
      "loss": 4.3667,
      "step": 4997
    },
    {
      "epoch": 0.20177634234961647,
      "grad_norm": 246.10791015625,
      "learning_rate": 1.8492438960513563e-05,
      "loss": 5.7013,
      "step": 4998
    },
    {
      "epoch": 0.20181671376665322,
      "grad_norm": 169.66256713867188,
      "learning_rate": 1.8491748484470478e-05,
      "loss": 5.5176,
      "step": 4999
    },
    {
      "epoch": 0.20185708518368994,
      "grad_norm": 83.66153717041016,
      "learning_rate": 1.849105786323833e-05,
      "loss": 7.4727,
      "step": 5000
    },
    {
      "epoch": 0.2018974566007267,
      "grad_norm": 86.1142349243164,
      "learning_rate": 1.8490367096828922e-05,
      "loss": 5.5252,
      "step": 5001
    },
    {
      "epoch": 0.20193782801776342,
      "grad_norm": 116.21569061279297,
      "learning_rate": 1.848967618525406e-05,
      "loss": 5.1324,
      "step": 5002
    },
    {
      "epoch": 0.20197819943480017,
      "grad_norm": 116.5449447631836,
      "learning_rate": 1.8488985128525558e-05,
      "loss": 5.1802,
      "step": 5003
    },
    {
      "epoch": 0.2020185708518369,
      "grad_norm": 95.64234161376953,
      "learning_rate": 1.8488293926655235e-05,
      "loss": 6.8439,
      "step": 5004
    },
    {
      "epoch": 0.20205894226887364,
      "grad_norm": 79.69072723388672,
      "learning_rate": 1.848760257965491e-05,
      "loss": 5.8568,
      "step": 5005
    },
    {
      "epoch": 0.2020993136859104,
      "grad_norm": 86.83350372314453,
      "learning_rate": 1.8486911087536397e-05,
      "loss": 5.3073,
      "step": 5006
    },
    {
      "epoch": 0.2021396851029471,
      "grad_norm": 105.94661712646484,
      "learning_rate": 1.8486219450311522e-05,
      "loss": 5.0847,
      "step": 5007
    },
    {
      "epoch": 0.20218005651998386,
      "grad_norm": 90.13518524169922,
      "learning_rate": 1.8485527667992115e-05,
      "loss": 5.3541,
      "step": 5008
    },
    {
      "epoch": 0.20222042793702058,
      "grad_norm": 106.85763549804688,
      "learning_rate": 1.848483574059e-05,
      "loss": 5.2405,
      "step": 5009
    },
    {
      "epoch": 0.20226079935405733,
      "grad_norm": 207.08358764648438,
      "learning_rate": 1.8484143668117003e-05,
      "loss": 4.3225,
      "step": 5010
    },
    {
      "epoch": 0.20230117077109405,
      "grad_norm": 95.6512451171875,
      "learning_rate": 1.8483451450584966e-05,
      "loss": 4.9467,
      "step": 5011
    },
    {
      "epoch": 0.2023415421881308,
      "grad_norm": 99.3128662109375,
      "learning_rate": 1.8482759088005713e-05,
      "loss": 3.3799,
      "step": 5012
    },
    {
      "epoch": 0.20238191360516755,
      "grad_norm": 129.804443359375,
      "learning_rate": 1.8482066580391096e-05,
      "loss": 5.2899,
      "step": 5013
    },
    {
      "epoch": 0.20242228502220427,
      "grad_norm": 75.7537841796875,
      "learning_rate": 1.8481373927752946e-05,
      "loss": 4.6528,
      "step": 5014
    },
    {
      "epoch": 0.20246265643924102,
      "grad_norm": 78.31391143798828,
      "learning_rate": 1.8480681130103104e-05,
      "loss": 5.796,
      "step": 5015
    },
    {
      "epoch": 0.20250302785627775,
      "grad_norm": 93.2804183959961,
      "learning_rate": 1.847998818745342e-05,
      "loss": 5.1603,
      "step": 5016
    },
    {
      "epoch": 0.2025433992733145,
      "grad_norm": 150.68775939941406,
      "learning_rate": 1.8479295099815742e-05,
      "loss": 3.3048,
      "step": 5017
    },
    {
      "epoch": 0.20258377069035124,
      "grad_norm": 210.68177795410156,
      "learning_rate": 1.8478601867201912e-05,
      "loss": 6.427,
      "step": 5018
    },
    {
      "epoch": 0.20262414210738797,
      "grad_norm": 122.24403381347656,
      "learning_rate": 1.8477908489623794e-05,
      "loss": 5.0742,
      "step": 5019
    },
    {
      "epoch": 0.20266451352442472,
      "grad_norm": 82.13291931152344,
      "learning_rate": 1.847721496709324e-05,
      "loss": 5.7796,
      "step": 5020
    },
    {
      "epoch": 0.20270488494146144,
      "grad_norm": 151.28680419921875,
      "learning_rate": 1.8476521299622102e-05,
      "loss": 5.5611,
      "step": 5021
    },
    {
      "epoch": 0.2027452563584982,
      "grad_norm": 93.18706512451172,
      "learning_rate": 1.847582748722225e-05,
      "loss": 5.6744,
      "step": 5022
    },
    {
      "epoch": 0.2027856277755349,
      "grad_norm": 114.61177825927734,
      "learning_rate": 1.8475133529905532e-05,
      "loss": 6.5118,
      "step": 5023
    },
    {
      "epoch": 0.20282599919257166,
      "grad_norm": 86.34323120117188,
      "learning_rate": 1.8474439427683825e-05,
      "loss": 4.0628,
      "step": 5024
    },
    {
      "epoch": 0.2028663706096084,
      "grad_norm": 118.09308624267578,
      "learning_rate": 1.8473745180568995e-05,
      "loss": 4.6408,
      "step": 5025
    },
    {
      "epoch": 0.20290674202664513,
      "grad_norm": 174.7678985595703,
      "learning_rate": 1.847305078857291e-05,
      "loss": 4.7371,
      "step": 5026
    },
    {
      "epoch": 0.20294711344368188,
      "grad_norm": 212.07220458984375,
      "learning_rate": 1.847235625170744e-05,
      "loss": 5.1118,
      "step": 5027
    },
    {
      "epoch": 0.2029874848607186,
      "grad_norm": 89.12480163574219,
      "learning_rate": 1.8471661569984464e-05,
      "loss": 4.8205,
      "step": 5028
    },
    {
      "epoch": 0.20302785627775535,
      "grad_norm": 127.58792877197266,
      "learning_rate": 1.8470966743415854e-05,
      "loss": 7.0757,
      "step": 5029
    },
    {
      "epoch": 0.20306822769479208,
      "grad_norm": 113.24164581298828,
      "learning_rate": 1.8470271772013498e-05,
      "loss": 3.0101,
      "step": 5030
    },
    {
      "epoch": 0.20310859911182882,
      "grad_norm": 91.78125762939453,
      "learning_rate": 1.8469576655789274e-05,
      "loss": 6.3874,
      "step": 5031
    },
    {
      "epoch": 0.20314897052886557,
      "grad_norm": 107.8099594116211,
      "learning_rate": 1.8468881394755067e-05,
      "loss": 3.7498,
      "step": 5032
    },
    {
      "epoch": 0.2031893419459023,
      "grad_norm": 200.18968200683594,
      "learning_rate": 1.8468185988922763e-05,
      "loss": 5.5253,
      "step": 5033
    },
    {
      "epoch": 0.20322971336293905,
      "grad_norm": 219.14984130859375,
      "learning_rate": 1.8467490438304254e-05,
      "loss": 4.3831,
      "step": 5034
    },
    {
      "epoch": 0.20327008477997577,
      "grad_norm": 114.40474700927734,
      "learning_rate": 1.8466794742911433e-05,
      "loss": 6.3612,
      "step": 5035
    },
    {
      "epoch": 0.20331045619701252,
      "grad_norm": 126.12416076660156,
      "learning_rate": 1.8466098902756193e-05,
      "loss": 5.0612,
      "step": 5036
    },
    {
      "epoch": 0.20335082761404924,
      "grad_norm": 104.24267578125,
      "learning_rate": 1.8465402917850428e-05,
      "loss": 5.6295,
      "step": 5037
    },
    {
      "epoch": 0.203391199031086,
      "grad_norm": 130.889404296875,
      "learning_rate": 1.8464706788206044e-05,
      "loss": 4.876,
      "step": 5038
    },
    {
      "epoch": 0.20343157044812274,
      "grad_norm": 91.57534790039062,
      "learning_rate": 1.846401051383494e-05,
      "loss": 4.645,
      "step": 5039
    },
    {
      "epoch": 0.20347194186515946,
      "grad_norm": 73.70066833496094,
      "learning_rate": 1.8463314094749022e-05,
      "loss": 6.4828,
      "step": 5040
    },
    {
      "epoch": 0.2035123132821962,
      "grad_norm": 267.2913818359375,
      "learning_rate": 1.8462617530960196e-05,
      "loss": 6.0808,
      "step": 5041
    },
    {
      "epoch": 0.20355268469923293,
      "grad_norm": 90.06062316894531,
      "learning_rate": 1.8461920822480367e-05,
      "loss": 5.6889,
      "step": 5042
    },
    {
      "epoch": 0.20359305611626968,
      "grad_norm": 232.57879638671875,
      "learning_rate": 1.8461223969321457e-05,
      "loss": 5.9531,
      "step": 5043
    },
    {
      "epoch": 0.20363342753330643,
      "grad_norm": 91.65325927734375,
      "learning_rate": 1.8460526971495377e-05,
      "loss": 5.1754,
      "step": 5044
    },
    {
      "epoch": 0.20367379895034315,
      "grad_norm": 116.02761840820312,
      "learning_rate": 1.845982982901404e-05,
      "loss": 6.6467,
      "step": 5045
    },
    {
      "epoch": 0.2037141703673799,
      "grad_norm": 69.4737548828125,
      "learning_rate": 1.8459132541889367e-05,
      "loss": 4.8284,
      "step": 5046
    },
    {
      "epoch": 0.20375454178441663,
      "grad_norm": 89.45829010009766,
      "learning_rate": 1.8458435110133282e-05,
      "loss": 4.4244,
      "step": 5047
    },
    {
      "epoch": 0.20379491320145338,
      "grad_norm": 88.78694152832031,
      "learning_rate": 1.845773753375771e-05,
      "loss": 6.2614,
      "step": 5048
    },
    {
      "epoch": 0.2038352846184901,
      "grad_norm": 84.72100067138672,
      "learning_rate": 1.8457039812774574e-05,
      "loss": 4.145,
      "step": 5049
    },
    {
      "epoch": 0.20387565603552685,
      "grad_norm": 68.18778991699219,
      "learning_rate": 1.845634194719581e-05,
      "loss": 3.7965,
      "step": 5050
    },
    {
      "epoch": 0.2039160274525636,
      "grad_norm": 137.79786682128906,
      "learning_rate": 1.845564393703334e-05,
      "loss": 3.9478,
      "step": 5051
    },
    {
      "epoch": 0.20395639886960032,
      "grad_norm": 69.57941436767578,
      "learning_rate": 1.8454945782299106e-05,
      "loss": 4.8808,
      "step": 5052
    },
    {
      "epoch": 0.20399677028663707,
      "grad_norm": 126.3305435180664,
      "learning_rate": 1.8454247483005043e-05,
      "loss": 3.3543,
      "step": 5053
    },
    {
      "epoch": 0.2040371417036738,
      "grad_norm": 97.44004821777344,
      "learning_rate": 1.8453549039163093e-05,
      "loss": 6.3033,
      "step": 5054
    },
    {
      "epoch": 0.20407751312071054,
      "grad_norm": 114.55326080322266,
      "learning_rate": 1.8452850450785192e-05,
      "loss": 4.2956,
      "step": 5055
    },
    {
      "epoch": 0.20411788453774726,
      "grad_norm": 160.48565673828125,
      "learning_rate": 1.8452151717883288e-05,
      "loss": 4.8601,
      "step": 5056
    },
    {
      "epoch": 0.204158255954784,
      "grad_norm": 114.47620391845703,
      "learning_rate": 1.8451452840469326e-05,
      "loss": 5.0405,
      "step": 5057
    },
    {
      "epoch": 0.20419862737182076,
      "grad_norm": 95.36593627929688,
      "learning_rate": 1.8450753818555257e-05,
      "loss": 5.6766,
      "step": 5058
    },
    {
      "epoch": 0.20423899878885748,
      "grad_norm": 155.19500732421875,
      "learning_rate": 1.845005465215303e-05,
      "loss": 5.9068,
      "step": 5059
    },
    {
      "epoch": 0.20427937020589423,
      "grad_norm": 122.14095306396484,
      "learning_rate": 1.8449355341274603e-05,
      "loss": 6.2661,
      "step": 5060
    },
    {
      "epoch": 0.20431974162293096,
      "grad_norm": 129.3920135498047,
      "learning_rate": 1.844865588593193e-05,
      "loss": 5.0553,
      "step": 5061
    },
    {
      "epoch": 0.2043601130399677,
      "grad_norm": 76.10726928710938,
      "learning_rate": 1.844795628613697e-05,
      "loss": 3.4579,
      "step": 5062
    },
    {
      "epoch": 0.20440048445700443,
      "grad_norm": 109.53791809082031,
      "learning_rate": 1.8447256541901683e-05,
      "loss": 4.9918,
      "step": 5063
    },
    {
      "epoch": 0.20444085587404118,
      "grad_norm": 139.8856201171875,
      "learning_rate": 1.8446556653238036e-05,
      "loss": 5.4274,
      "step": 5064
    },
    {
      "epoch": 0.20448122729107793,
      "grad_norm": 107.19644165039062,
      "learning_rate": 1.8445856620157997e-05,
      "loss": 4.4988,
      "step": 5065
    },
    {
      "epoch": 0.20452159870811465,
      "grad_norm": 114.94694519042969,
      "learning_rate": 1.8445156442673527e-05,
      "loss": 5.0545,
      "step": 5066
    },
    {
      "epoch": 0.2045619701251514,
      "grad_norm": 153.0474395751953,
      "learning_rate": 1.8444456120796608e-05,
      "loss": 4.0825,
      "step": 5067
    },
    {
      "epoch": 0.20460234154218812,
      "grad_norm": 91.46851348876953,
      "learning_rate": 1.8443755654539206e-05,
      "loss": 5.6042,
      "step": 5068
    },
    {
      "epoch": 0.20464271295922487,
      "grad_norm": 104.62637329101562,
      "learning_rate": 1.84430550439133e-05,
      "loss": 6.8523,
      "step": 5069
    },
    {
      "epoch": 0.20468308437626162,
      "grad_norm": 119.94732666015625,
      "learning_rate": 1.844235428893087e-05,
      "loss": 5.8173,
      "step": 5070
    },
    {
      "epoch": 0.20472345579329834,
      "grad_norm": 94.8011703491211,
      "learning_rate": 1.844165338960389e-05,
      "loss": 6.5277,
      "step": 5071
    },
    {
      "epoch": 0.2047638272103351,
      "grad_norm": 83.98675537109375,
      "learning_rate": 1.844095234594436e-05,
      "loss": 5.1089,
      "step": 5072
    },
    {
      "epoch": 0.20480419862737181,
      "grad_norm": 86.16030883789062,
      "learning_rate": 1.8440251157964248e-05,
      "loss": 4.5808,
      "step": 5073
    },
    {
      "epoch": 0.20484457004440856,
      "grad_norm": 148.60800170898438,
      "learning_rate": 1.8439549825675553e-05,
      "loss": 5.0984,
      "step": 5074
    },
    {
      "epoch": 0.20488494146144529,
      "grad_norm": 134.3030242919922,
      "learning_rate": 1.8438848349090265e-05,
      "loss": 5.165,
      "step": 5075
    },
    {
      "epoch": 0.20492531287848204,
      "grad_norm": 161.31932067871094,
      "learning_rate": 1.8438146728220376e-05,
      "loss": 4.5183,
      "step": 5076
    },
    {
      "epoch": 0.20496568429551879,
      "grad_norm": 134.71458435058594,
      "learning_rate": 1.843744496307788e-05,
      "loss": 5.9051,
      "step": 5077
    },
    {
      "epoch": 0.2050060557125555,
      "grad_norm": 159.5782012939453,
      "learning_rate": 1.843674305367478e-05,
      "loss": 5.8476,
      "step": 5078
    },
    {
      "epoch": 0.20504642712959226,
      "grad_norm": 121.14154815673828,
      "learning_rate": 1.8436041000023076e-05,
      "loss": 4.5283,
      "step": 5079
    },
    {
      "epoch": 0.20508679854662898,
      "grad_norm": 96.7656021118164,
      "learning_rate": 1.8435338802134767e-05,
      "loss": 4.4541,
      "step": 5080
    },
    {
      "epoch": 0.20512716996366573,
      "grad_norm": 98.44661712646484,
      "learning_rate": 1.8434636460021866e-05,
      "loss": 5.9993,
      "step": 5081
    },
    {
      "epoch": 0.20516754138070245,
      "grad_norm": 81.9199447631836,
      "learning_rate": 1.8433933973696374e-05,
      "loss": 5.4397,
      "step": 5082
    },
    {
      "epoch": 0.2052079127977392,
      "grad_norm": 144.7051544189453,
      "learning_rate": 1.843323134317031e-05,
      "loss": 7.5745,
      "step": 5083
    },
    {
      "epoch": 0.20524828421477595,
      "grad_norm": 111.93897247314453,
      "learning_rate": 1.843252856845568e-05,
      "loss": 6.2197,
      "step": 5084
    },
    {
      "epoch": 0.20528865563181267,
      "grad_norm": 90.19837951660156,
      "learning_rate": 1.8431825649564502e-05,
      "loss": 5.9821,
      "step": 5085
    },
    {
      "epoch": 0.20532902704884942,
      "grad_norm": 164.12367248535156,
      "learning_rate": 1.8431122586508798e-05,
      "loss": 4.9369,
      "step": 5086
    },
    {
      "epoch": 0.20536939846588614,
      "grad_norm": 90.80906677246094,
      "learning_rate": 1.8430419379300586e-05,
      "loss": 4.9013,
      "step": 5087
    },
    {
      "epoch": 0.2054097698829229,
      "grad_norm": 108.17987060546875,
      "learning_rate": 1.842971602795189e-05,
      "loss": 5.9081,
      "step": 5088
    },
    {
      "epoch": 0.20545014129995962,
      "grad_norm": 222.254638671875,
      "learning_rate": 1.8429012532474728e-05,
      "loss": 3.6518,
      "step": 5089
    },
    {
      "epoch": 0.20549051271699637,
      "grad_norm": 111.11985778808594,
      "learning_rate": 1.8428308892881143e-05,
      "loss": 6.1633,
      "step": 5090
    },
    {
      "epoch": 0.20553088413403311,
      "grad_norm": 147.81304931640625,
      "learning_rate": 1.842760510918315e-05,
      "loss": 4.6019,
      "step": 5091
    },
    {
      "epoch": 0.20557125555106984,
      "grad_norm": 83.19377136230469,
      "learning_rate": 1.8426901181392794e-05,
      "loss": 5.5501,
      "step": 5092
    },
    {
      "epoch": 0.2056116269681066,
      "grad_norm": 122.54945373535156,
      "learning_rate": 1.8426197109522107e-05,
      "loss": 6.2668,
      "step": 5093
    },
    {
      "epoch": 0.2056519983851433,
      "grad_norm": 105.2330551147461,
      "learning_rate": 1.8425492893583127e-05,
      "loss": 5.7892,
      "step": 5094
    },
    {
      "epoch": 0.20569236980218006,
      "grad_norm": 156.41314697265625,
      "learning_rate": 1.842478853358789e-05,
      "loss": 4.8319,
      "step": 5095
    },
    {
      "epoch": 0.2057327412192168,
      "grad_norm": 140.17384338378906,
      "learning_rate": 1.842408402954844e-05,
      "loss": 5.5439,
      "step": 5096
    },
    {
      "epoch": 0.20577311263625353,
      "grad_norm": 242.1778106689453,
      "learning_rate": 1.842337938147683e-05,
      "loss": 7.2339,
      "step": 5097
    },
    {
      "epoch": 0.20581348405329028,
      "grad_norm": 133.16770935058594,
      "learning_rate": 1.8422674589385103e-05,
      "loss": 6.0661,
      "step": 5098
    },
    {
      "epoch": 0.205853855470327,
      "grad_norm": 90.73487091064453,
      "learning_rate": 1.8421969653285307e-05,
      "loss": 3.2064,
      "step": 5099
    },
    {
      "epoch": 0.20589422688736375,
      "grad_norm": 81.70140838623047,
      "learning_rate": 1.8421264573189493e-05,
      "loss": 5.5523,
      "step": 5100
    },
    {
      "epoch": 0.20593459830440047,
      "grad_norm": 114.99794006347656,
      "learning_rate": 1.8420559349109724e-05,
      "loss": 4.3538,
      "step": 5101
    },
    {
      "epoch": 0.20597496972143722,
      "grad_norm": 123.52117156982422,
      "learning_rate": 1.841985398105805e-05,
      "loss": 6.3564,
      "step": 5102
    },
    {
      "epoch": 0.20601534113847397,
      "grad_norm": 117.29888153076172,
      "learning_rate": 1.841914846904654e-05,
      "loss": 5.036,
      "step": 5103
    },
    {
      "epoch": 0.2060557125555107,
      "grad_norm": 103.85919189453125,
      "learning_rate": 1.8418442813087246e-05,
      "loss": 6.3403,
      "step": 5104
    },
    {
      "epoch": 0.20609608397254744,
      "grad_norm": 105.91909790039062,
      "learning_rate": 1.8417737013192243e-05,
      "loss": 4.3974,
      "step": 5105
    },
    {
      "epoch": 0.20613645538958417,
      "grad_norm": 162.64527893066406,
      "learning_rate": 1.8417031069373592e-05,
      "loss": 7.0698,
      "step": 5106
    },
    {
      "epoch": 0.20617682680662092,
      "grad_norm": 74.01446533203125,
      "learning_rate": 1.8416324981643363e-05,
      "loss": 5.2134,
      "step": 5107
    },
    {
      "epoch": 0.20621719822365764,
      "grad_norm": 146.7371826171875,
      "learning_rate": 1.8415618750013634e-05,
      "loss": 5.0079,
      "step": 5108
    },
    {
      "epoch": 0.2062575696406944,
      "grad_norm": 90.10218048095703,
      "learning_rate": 1.8414912374496473e-05,
      "loss": 5.2444,
      "step": 5109
    },
    {
      "epoch": 0.20629794105773114,
      "grad_norm": 101.58429718017578,
      "learning_rate": 1.8414205855103963e-05,
      "loss": 4.9989,
      "step": 5110
    },
    {
      "epoch": 0.20633831247476786,
      "grad_norm": 120.81748962402344,
      "learning_rate": 1.841349919184818e-05,
      "loss": 4.8725,
      "step": 5111
    },
    {
      "epoch": 0.2063786838918046,
      "grad_norm": 81.07373046875,
      "learning_rate": 1.841279238474121e-05,
      "loss": 6.4893,
      "step": 5112
    },
    {
      "epoch": 0.20641905530884133,
      "grad_norm": 84.76177978515625,
      "learning_rate": 1.8412085433795132e-05,
      "loss": 5.2274,
      "step": 5113
    },
    {
      "epoch": 0.20645942672587808,
      "grad_norm": 130.214111328125,
      "learning_rate": 1.8411378339022038e-05,
      "loss": 5.6967,
      "step": 5114
    },
    {
      "epoch": 0.2064997981429148,
      "grad_norm": 121.05789947509766,
      "learning_rate": 1.841067110043402e-05,
      "loss": 4.6629,
      "step": 5115
    },
    {
      "epoch": 0.20654016955995155,
      "grad_norm": 77.53231811523438,
      "learning_rate": 1.840996371804316e-05,
      "loss": 4.2998,
      "step": 5116
    },
    {
      "epoch": 0.2065805409769883,
      "grad_norm": 87.83821868896484,
      "learning_rate": 1.840925619186156e-05,
      "loss": 5.3762,
      "step": 5117
    },
    {
      "epoch": 0.20662091239402502,
      "grad_norm": 77.48865509033203,
      "learning_rate": 1.8408548521901322e-05,
      "loss": 6.2951,
      "step": 5118
    },
    {
      "epoch": 0.20666128381106177,
      "grad_norm": 92.0673599243164,
      "learning_rate": 1.8407840708174536e-05,
      "loss": 5.3139,
      "step": 5119
    },
    {
      "epoch": 0.2067016552280985,
      "grad_norm": 87.56224822998047,
      "learning_rate": 1.840713275069331e-05,
      "loss": 5.7295,
      "step": 5120
    },
    {
      "epoch": 0.20674202664513525,
      "grad_norm": 100.58861541748047,
      "learning_rate": 1.8406424649469742e-05,
      "loss": 6.0163,
      "step": 5121
    },
    {
      "epoch": 0.206782398062172,
      "grad_norm": 128.2412872314453,
      "learning_rate": 1.8405716404515942e-05,
      "loss": 5.7354,
      "step": 5122
    },
    {
      "epoch": 0.20682276947920872,
      "grad_norm": 95.44880676269531,
      "learning_rate": 1.8405008015844025e-05,
      "loss": 6.3392,
      "step": 5123
    },
    {
      "epoch": 0.20686314089624547,
      "grad_norm": 81.12798309326172,
      "learning_rate": 1.8404299483466098e-05,
      "loss": 4.3718,
      "step": 5124
    },
    {
      "epoch": 0.2069035123132822,
      "grad_norm": 100.49008178710938,
      "learning_rate": 1.8403590807394272e-05,
      "loss": 4.2538,
      "step": 5125
    },
    {
      "epoch": 0.20694388373031894,
      "grad_norm": 95.39754486083984,
      "learning_rate": 1.8402881987640668e-05,
      "loss": 4.9519,
      "step": 5126
    },
    {
      "epoch": 0.20698425514735566,
      "grad_norm": 90.97562408447266,
      "learning_rate": 1.8402173024217405e-05,
      "loss": 3.6708,
      "step": 5127
    },
    {
      "epoch": 0.2070246265643924,
      "grad_norm": 109.26970672607422,
      "learning_rate": 1.84014639171366e-05,
      "loss": 6.0669,
      "step": 5128
    },
    {
      "epoch": 0.20706499798142916,
      "grad_norm": 75.22322845458984,
      "learning_rate": 1.8400754666410384e-05,
      "loss": 5.4894,
      "step": 5129
    },
    {
      "epoch": 0.20710536939846588,
      "grad_norm": 94.02415466308594,
      "learning_rate": 1.8400045272050885e-05,
      "loss": 4.2207,
      "step": 5130
    },
    {
      "epoch": 0.20714574081550263,
      "grad_norm": 124.73999786376953,
      "learning_rate": 1.8399335734070222e-05,
      "loss": 3.8246,
      "step": 5131
    },
    {
      "epoch": 0.20718611223253935,
      "grad_norm": 142.88174438476562,
      "learning_rate": 1.8398626052480533e-05,
      "loss": 4.123,
      "step": 5132
    },
    {
      "epoch": 0.2072264836495761,
      "grad_norm": 228.8887939453125,
      "learning_rate": 1.839791622729395e-05,
      "loss": 5.5922,
      "step": 5133
    },
    {
      "epoch": 0.20726685506661283,
      "grad_norm": 83.36515808105469,
      "learning_rate": 1.839720625852261e-05,
      "loss": 4.631,
      "step": 5134
    },
    {
      "epoch": 0.20730722648364958,
      "grad_norm": 116.33688354492188,
      "learning_rate": 1.8396496146178654e-05,
      "loss": 4.3609,
      "step": 5135
    },
    {
      "epoch": 0.20734759790068633,
      "grad_norm": 95.38289642333984,
      "learning_rate": 1.8395785890274216e-05,
      "loss": 3.394,
      "step": 5136
    },
    {
      "epoch": 0.20738796931772305,
      "grad_norm": 136.5562286376953,
      "learning_rate": 1.8395075490821452e-05,
      "loss": 5.7875,
      "step": 5137
    },
    {
      "epoch": 0.2074283407347598,
      "grad_norm": 130.8564453125,
      "learning_rate": 1.8394364947832493e-05,
      "loss": 7.3221,
      "step": 5138
    },
    {
      "epoch": 0.20746871215179652,
      "grad_norm": 188.93006896972656,
      "learning_rate": 1.8393654261319504e-05,
      "loss": 4.5171,
      "step": 5139
    },
    {
      "epoch": 0.20750908356883327,
      "grad_norm": 122.69171142578125,
      "learning_rate": 1.839294343129462e-05,
      "loss": 5.542,
      "step": 5140
    },
    {
      "epoch": 0.20754945498587,
      "grad_norm": 103.13880157470703,
      "learning_rate": 1.8392232457770006e-05,
      "loss": 4.6593,
      "step": 5141
    },
    {
      "epoch": 0.20758982640290674,
      "grad_norm": 180.08399963378906,
      "learning_rate": 1.8391521340757817e-05,
      "loss": 4.4713,
      "step": 5142
    },
    {
      "epoch": 0.2076301978199435,
      "grad_norm": 126.01576232910156,
      "learning_rate": 1.8390810080270203e-05,
      "loss": 6.0341,
      "step": 5143
    },
    {
      "epoch": 0.2076705692369802,
      "grad_norm": 118.01785278320312,
      "learning_rate": 1.8390098676319335e-05,
      "loss": 5.5027,
      "step": 5144
    },
    {
      "epoch": 0.20771094065401696,
      "grad_norm": 96.8524398803711,
      "learning_rate": 1.838938712891737e-05,
      "loss": 4.8162,
      "step": 5145
    },
    {
      "epoch": 0.20775131207105368,
      "grad_norm": 86.77902221679688,
      "learning_rate": 1.8388675438076476e-05,
      "loss": 6.0272,
      "step": 5146
    },
    {
      "epoch": 0.20779168348809043,
      "grad_norm": 98.99378967285156,
      "learning_rate": 1.8387963603808822e-05,
      "loss": 6.8788,
      "step": 5147
    },
    {
      "epoch": 0.20783205490512718,
      "grad_norm": 76.0437240600586,
      "learning_rate": 1.8387251626126574e-05,
      "loss": 5.7854,
      "step": 5148
    },
    {
      "epoch": 0.2078724263221639,
      "grad_norm": 85.7993392944336,
      "learning_rate": 1.8386539505041915e-05,
      "loss": 6.1408,
      "step": 5149
    },
    {
      "epoch": 0.20791279773920066,
      "grad_norm": 134.4572296142578,
      "learning_rate": 1.838582724056701e-05,
      "loss": 5.2945,
      "step": 5150
    },
    {
      "epoch": 0.20795316915623738,
      "grad_norm": 118.93092346191406,
      "learning_rate": 1.8385114832714038e-05,
      "loss": 2.6872,
      "step": 5151
    },
    {
      "epoch": 0.20799354057327413,
      "grad_norm": 84.37199401855469,
      "learning_rate": 1.838440228149519e-05,
      "loss": 6.2064,
      "step": 5152
    },
    {
      "epoch": 0.20803391199031085,
      "grad_norm": 98.21096801757812,
      "learning_rate": 1.838368958692264e-05,
      "loss": 4.7306,
      "step": 5153
    },
    {
      "epoch": 0.2080742834073476,
      "grad_norm": 129.60072326660156,
      "learning_rate": 1.8382976749008573e-05,
      "loss": 5.9813,
      "step": 5154
    },
    {
      "epoch": 0.20811465482438435,
      "grad_norm": 78.70340728759766,
      "learning_rate": 1.8382263767765182e-05,
      "loss": 6.8702,
      "step": 5155
    },
    {
      "epoch": 0.20815502624142107,
      "grad_norm": 155.06752014160156,
      "learning_rate": 1.8381550643204657e-05,
      "loss": 5.0269,
      "step": 5156
    },
    {
      "epoch": 0.20819539765845782,
      "grad_norm": 128.52102661132812,
      "learning_rate": 1.838083737533918e-05,
      "loss": 6.3562,
      "step": 5157
    },
    {
      "epoch": 0.20823576907549454,
      "grad_norm": 90.19061279296875,
      "learning_rate": 1.838012396418096e-05,
      "loss": 4.7793,
      "step": 5158
    },
    {
      "epoch": 0.2082761404925313,
      "grad_norm": 91.23539733886719,
      "learning_rate": 1.837941040974219e-05,
      "loss": 5.5172,
      "step": 5159
    },
    {
      "epoch": 0.20831651190956801,
      "grad_norm": 231.5214080810547,
      "learning_rate": 1.837869671203507e-05,
      "loss": 4.9811,
      "step": 5160
    },
    {
      "epoch": 0.20835688332660476,
      "grad_norm": 102.54190063476562,
      "learning_rate": 1.8377982871071796e-05,
      "loss": 4.5044,
      "step": 5161
    },
    {
      "epoch": 0.2083972547436415,
      "grad_norm": 80.29256439208984,
      "learning_rate": 1.8377268886864587e-05,
      "loss": 5.7121,
      "step": 5162
    },
    {
      "epoch": 0.20843762616067824,
      "grad_norm": 170.9859161376953,
      "learning_rate": 1.8376554759425634e-05,
      "loss": 6.3472,
      "step": 5163
    },
    {
      "epoch": 0.20847799757771499,
      "grad_norm": 109.94129943847656,
      "learning_rate": 1.8375840488767163e-05,
      "loss": 5.7699,
      "step": 5164
    },
    {
      "epoch": 0.2085183689947517,
      "grad_norm": 130.2085418701172,
      "learning_rate": 1.8375126074901373e-05,
      "loss": 6.0148,
      "step": 5165
    },
    {
      "epoch": 0.20855874041178846,
      "grad_norm": 144.13101196289062,
      "learning_rate": 1.8374411517840485e-05,
      "loss": 5.258,
      "step": 5166
    },
    {
      "epoch": 0.20859911182882518,
      "grad_norm": 135.5904998779297,
      "learning_rate": 1.8373696817596722e-05,
      "loss": 4.0599,
      "step": 5167
    },
    {
      "epoch": 0.20863948324586193,
      "grad_norm": 66.3238525390625,
      "learning_rate": 1.8372981974182293e-05,
      "loss": 3.7381,
      "step": 5168
    },
    {
      "epoch": 0.20867985466289868,
      "grad_norm": 99.14871215820312,
      "learning_rate": 1.8372266987609422e-05,
      "loss": 5.2258,
      "step": 5169
    },
    {
      "epoch": 0.2087202260799354,
      "grad_norm": 87.63042449951172,
      "learning_rate": 1.837155185789034e-05,
      "loss": 6.0517,
      "step": 5170
    },
    {
      "epoch": 0.20876059749697215,
      "grad_norm": 102.2785415649414,
      "learning_rate": 1.8370836585037266e-05,
      "loss": 5.057,
      "step": 5171
    },
    {
      "epoch": 0.20880096891400887,
      "grad_norm": 98.83769989013672,
      "learning_rate": 1.8370121169062436e-05,
      "loss": 5.6146,
      "step": 5172
    },
    {
      "epoch": 0.20884134033104562,
      "grad_norm": 150.6449737548828,
      "learning_rate": 1.8369405609978083e-05,
      "loss": 4.9374,
      "step": 5173
    },
    {
      "epoch": 0.20888171174808237,
      "grad_norm": 138.15171813964844,
      "learning_rate": 1.8368689907796434e-05,
      "loss": 5.0506,
      "step": 5174
    },
    {
      "epoch": 0.2089220831651191,
      "grad_norm": 128.2700958251953,
      "learning_rate": 1.836797406252973e-05,
      "loss": 4.618,
      "step": 5175
    },
    {
      "epoch": 0.20896245458215584,
      "grad_norm": 129.57846069335938,
      "learning_rate": 1.836725807419021e-05,
      "loss": 4.1773,
      "step": 5176
    },
    {
      "epoch": 0.20900282599919257,
      "grad_norm": 96.59933471679688,
      "learning_rate": 1.8366541942790117e-05,
      "loss": 5.4834,
      "step": 5177
    },
    {
      "epoch": 0.20904319741622931,
      "grad_norm": 103.94254302978516,
      "learning_rate": 1.8365825668341696e-05,
      "loss": 5.1561,
      "step": 5178
    },
    {
      "epoch": 0.20908356883326604,
      "grad_norm": 72.5163345336914,
      "learning_rate": 1.836510925085719e-05,
      "loss": 4.2094,
      "step": 5179
    },
    {
      "epoch": 0.2091239402503028,
      "grad_norm": 193.1908416748047,
      "learning_rate": 1.8364392690348848e-05,
      "loss": 5.6856,
      "step": 5180
    },
    {
      "epoch": 0.20916431166733954,
      "grad_norm": 82.77017211914062,
      "learning_rate": 1.8363675986828925e-05,
      "loss": 5.3293,
      "step": 5181
    },
    {
      "epoch": 0.20920468308437626,
      "grad_norm": 204.53054809570312,
      "learning_rate": 1.8362959140309674e-05,
      "loss": 5.3033,
      "step": 5182
    },
    {
      "epoch": 0.209245054501413,
      "grad_norm": 122.88172912597656,
      "learning_rate": 1.8362242150803347e-05,
      "loss": 5.6675,
      "step": 5183
    },
    {
      "epoch": 0.20928542591844973,
      "grad_norm": 85.31038665771484,
      "learning_rate": 1.836152501832221e-05,
      "loss": 4.5254,
      "step": 5184
    },
    {
      "epoch": 0.20932579733548648,
      "grad_norm": 93.06494140625,
      "learning_rate": 1.8360807742878518e-05,
      "loss": 5.472,
      "step": 5185
    },
    {
      "epoch": 0.2093661687525232,
      "grad_norm": 163.84469604492188,
      "learning_rate": 1.836009032448454e-05,
      "loss": 6.4439,
      "step": 5186
    },
    {
      "epoch": 0.20940654016955995,
      "grad_norm": 130.25003051757812,
      "learning_rate": 1.8359372763152537e-05,
      "loss": 5.5095,
      "step": 5187
    },
    {
      "epoch": 0.2094469115865967,
      "grad_norm": 86.89057922363281,
      "learning_rate": 1.8358655058894785e-05,
      "loss": 6.6241,
      "step": 5188
    },
    {
      "epoch": 0.20948728300363342,
      "grad_norm": 116.39128112792969,
      "learning_rate": 1.8357937211723546e-05,
      "loss": 4.9757,
      "step": 5189
    },
    {
      "epoch": 0.20952765442067017,
      "grad_norm": 104.23992919921875,
      "learning_rate": 1.83572192216511e-05,
      "loss": 5.4369,
      "step": 5190
    },
    {
      "epoch": 0.2095680258377069,
      "grad_norm": 84.23524475097656,
      "learning_rate": 1.835650108868972e-05,
      "loss": 5.9727,
      "step": 5191
    },
    {
      "epoch": 0.20960839725474364,
      "grad_norm": 163.86363220214844,
      "learning_rate": 1.8355782812851687e-05,
      "loss": 6.5331,
      "step": 5192
    },
    {
      "epoch": 0.20964876867178037,
      "grad_norm": 87.91470336914062,
      "learning_rate": 1.8355064394149277e-05,
      "loss": 5.7798,
      "step": 5193
    },
    {
      "epoch": 0.20968914008881712,
      "grad_norm": 111.98867797851562,
      "learning_rate": 1.8354345832594777e-05,
      "loss": 5.3948,
      "step": 5194
    },
    {
      "epoch": 0.20972951150585387,
      "grad_norm": 170.1331329345703,
      "learning_rate": 1.8353627128200473e-05,
      "loss": 5.4857,
      "step": 5195
    },
    {
      "epoch": 0.2097698829228906,
      "grad_norm": 139.9152069091797,
      "learning_rate": 1.835290828097865e-05,
      "loss": 6.2721,
      "step": 5196
    },
    {
      "epoch": 0.20981025433992734,
      "grad_norm": 127.45706176757812,
      "learning_rate": 1.8352189290941608e-05,
      "loss": 6.9603,
      "step": 5197
    },
    {
      "epoch": 0.20985062575696406,
      "grad_norm": 81.39462280273438,
      "learning_rate": 1.8351470158101624e-05,
      "loss": 5.609,
      "step": 5198
    },
    {
      "epoch": 0.2098909971740008,
      "grad_norm": 90.36161041259766,
      "learning_rate": 1.8350750882471007e-05,
      "loss": 5.7988,
      "step": 5199
    },
    {
      "epoch": 0.20993136859103753,
      "grad_norm": 128.03213500976562,
      "learning_rate": 1.835003146406205e-05,
      "loss": 4.956,
      "step": 5200
    },
    {
      "epoch": 0.20997174000807428,
      "grad_norm": 142.46775817871094,
      "learning_rate": 1.8349311902887053e-05,
      "loss": 6.0314,
      "step": 5201
    },
    {
      "epoch": 0.21001211142511103,
      "grad_norm": 84.01886749267578,
      "learning_rate": 1.834859219895832e-05,
      "loss": 7.0243,
      "step": 5202
    },
    {
      "epoch": 0.21005248284214775,
      "grad_norm": 87.82781982421875,
      "learning_rate": 1.8347872352288156e-05,
      "loss": 5.9132,
      "step": 5203
    },
    {
      "epoch": 0.2100928542591845,
      "grad_norm": 84.65574645996094,
      "learning_rate": 1.8347152362888865e-05,
      "loss": 5.2404,
      "step": 5204
    },
    {
      "epoch": 0.21013322567622122,
      "grad_norm": 74.76691436767578,
      "learning_rate": 1.8346432230772766e-05,
      "loss": 5.6432,
      "step": 5205
    },
    {
      "epoch": 0.21017359709325797,
      "grad_norm": 78.54199981689453,
      "learning_rate": 1.834571195595216e-05,
      "loss": 6.1106,
      "step": 5206
    },
    {
      "epoch": 0.21021396851029472,
      "grad_norm": 133.68936157226562,
      "learning_rate": 1.8344991538439373e-05,
      "loss": 5.1621,
      "step": 5207
    },
    {
      "epoch": 0.21025433992733145,
      "grad_norm": 226.95217895507812,
      "learning_rate": 1.8344270978246717e-05,
      "loss": 6.3922,
      "step": 5208
    },
    {
      "epoch": 0.2102947113443682,
      "grad_norm": 80.27652740478516,
      "learning_rate": 1.8343550275386513e-05,
      "loss": 5.9859,
      "step": 5209
    },
    {
      "epoch": 0.21033508276140492,
      "grad_norm": 133.48104858398438,
      "learning_rate": 1.834282942987108e-05,
      "loss": 5.1671,
      "step": 5210
    },
    {
      "epoch": 0.21037545417844167,
      "grad_norm": 99.45147705078125,
      "learning_rate": 1.834210844171275e-05,
      "loss": 5.1039,
      "step": 5211
    },
    {
      "epoch": 0.2104158255954784,
      "grad_norm": 93.32068634033203,
      "learning_rate": 1.8341387310923845e-05,
      "loss": 6.0387,
      "step": 5212
    },
    {
      "epoch": 0.21045619701251514,
      "grad_norm": 127.75816345214844,
      "learning_rate": 1.834066603751669e-05,
      "loss": 4.4691,
      "step": 5213
    },
    {
      "epoch": 0.2104965684295519,
      "grad_norm": 140.08407592773438,
      "learning_rate": 1.833994462150363e-05,
      "loss": 5.5745,
      "step": 5214
    },
    {
      "epoch": 0.2105369398465886,
      "grad_norm": 71.17708587646484,
      "learning_rate": 1.8339223062896986e-05,
      "loss": 4.2009,
      "step": 5215
    },
    {
      "epoch": 0.21057731126362536,
      "grad_norm": 185.18209838867188,
      "learning_rate": 1.8338501361709106e-05,
      "loss": 6.1889,
      "step": 5216
    },
    {
      "epoch": 0.21061768268066208,
      "grad_norm": 84.9694595336914,
      "learning_rate": 1.833777951795232e-05,
      "loss": 4.4548,
      "step": 5217
    },
    {
      "epoch": 0.21065805409769883,
      "grad_norm": 89.83562469482422,
      "learning_rate": 1.833705753163898e-05,
      "loss": 5.8524,
      "step": 5218
    },
    {
      "epoch": 0.21069842551473555,
      "grad_norm": 90.77855682373047,
      "learning_rate": 1.833633540278142e-05,
      "loss": 5.0634,
      "step": 5219
    },
    {
      "epoch": 0.2107387969317723,
      "grad_norm": 112.57235717773438,
      "learning_rate": 1.833561313139199e-05,
      "loss": 6.4773,
      "step": 5220
    },
    {
      "epoch": 0.21077916834880905,
      "grad_norm": 93.22792053222656,
      "learning_rate": 1.8334890717483045e-05,
      "loss": 5.1506,
      "step": 5221
    },
    {
      "epoch": 0.21081953976584578,
      "grad_norm": 100.708251953125,
      "learning_rate": 1.833416816106693e-05,
      "loss": 4.8002,
      "step": 5222
    },
    {
      "epoch": 0.21085991118288253,
      "grad_norm": 93.2436752319336,
      "learning_rate": 1.8333445462156002e-05,
      "loss": 6.65,
      "step": 5223
    },
    {
      "epoch": 0.21090028259991925,
      "grad_norm": 115.53718566894531,
      "learning_rate": 1.833272262076262e-05,
      "loss": 5.5578,
      "step": 5224
    },
    {
      "epoch": 0.210940654016956,
      "grad_norm": 94.31204223632812,
      "learning_rate": 1.833199963689913e-05,
      "loss": 7.0445,
      "step": 5225
    },
    {
      "epoch": 0.21098102543399272,
      "grad_norm": 94.72102355957031,
      "learning_rate": 1.833127651057791e-05,
      "loss": 5.9271,
      "step": 5226
    },
    {
      "epoch": 0.21102139685102947,
      "grad_norm": 133.11788940429688,
      "learning_rate": 1.8330553241811317e-05,
      "loss": 5.3232,
      "step": 5227
    },
    {
      "epoch": 0.21106176826806622,
      "grad_norm": 161.12428283691406,
      "learning_rate": 1.8329829830611714e-05,
      "loss": 5.5525,
      "step": 5228
    },
    {
      "epoch": 0.21110213968510294,
      "grad_norm": 113.56775665283203,
      "learning_rate": 1.832910627699147e-05,
      "loss": 5.9204,
      "step": 5229
    },
    {
      "epoch": 0.2111425111021397,
      "grad_norm": 127.01058197021484,
      "learning_rate": 1.8328382580962964e-05,
      "loss": 6.1186,
      "step": 5230
    },
    {
      "epoch": 0.2111828825191764,
      "grad_norm": 67.21422576904297,
      "learning_rate": 1.8327658742538564e-05,
      "loss": 4.7344,
      "step": 5231
    },
    {
      "epoch": 0.21122325393621316,
      "grad_norm": 156.096435546875,
      "learning_rate": 1.8326934761730643e-05,
      "loss": 4.9585,
      "step": 5232
    },
    {
      "epoch": 0.2112636253532499,
      "grad_norm": 83.05474090576172,
      "learning_rate": 1.832621063855158e-05,
      "loss": 5.2345,
      "step": 5233
    },
    {
      "epoch": 0.21130399677028663,
      "grad_norm": 96.15074920654297,
      "learning_rate": 1.8325486373013765e-05,
      "loss": 6.1931,
      "step": 5234
    },
    {
      "epoch": 0.21134436818732338,
      "grad_norm": 97.24642944335938,
      "learning_rate": 1.832476196512957e-05,
      "loss": 6.4155,
      "step": 5235
    },
    {
      "epoch": 0.2113847396043601,
      "grad_norm": 187.5957489013672,
      "learning_rate": 1.8324037414911385e-05,
      "loss": 5.7424,
      "step": 5236
    },
    {
      "epoch": 0.21142511102139686,
      "grad_norm": 96.17208862304688,
      "learning_rate": 1.83233127223716e-05,
      "loss": 5.2104,
      "step": 5237
    },
    {
      "epoch": 0.21146548243843358,
      "grad_norm": 91.79185485839844,
      "learning_rate": 1.83225878875226e-05,
      "loss": 6.329,
      "step": 5238
    },
    {
      "epoch": 0.21150585385547033,
      "grad_norm": 112.1553726196289,
      "learning_rate": 1.8321862910376786e-05,
      "loss": 5.578,
      "step": 5239
    },
    {
      "epoch": 0.21154622527250708,
      "grad_norm": 108.32706451416016,
      "learning_rate": 1.832113779094655e-05,
      "loss": 5.3265,
      "step": 5240
    },
    {
      "epoch": 0.2115865966895438,
      "grad_norm": 107.22596740722656,
      "learning_rate": 1.8320412529244286e-05,
      "loss": 5.0198,
      "step": 5241
    },
    {
      "epoch": 0.21162696810658055,
      "grad_norm": 79.22071075439453,
      "learning_rate": 1.8319687125282398e-05,
      "loss": 4.6951,
      "step": 5242
    },
    {
      "epoch": 0.21166733952361727,
      "grad_norm": 97.1541976928711,
      "learning_rate": 1.8318961579073287e-05,
      "loss": 5.4673,
      "step": 5243
    },
    {
      "epoch": 0.21170771094065402,
      "grad_norm": 97.16192626953125,
      "learning_rate": 1.831823589062936e-05,
      "loss": 5.6306,
      "step": 5244
    },
    {
      "epoch": 0.21174808235769074,
      "grad_norm": 115.57694244384766,
      "learning_rate": 1.8317510059963024e-05,
      "loss": 6.1732,
      "step": 5245
    },
    {
      "epoch": 0.2117884537747275,
      "grad_norm": 87.65924072265625,
      "learning_rate": 1.831678408708669e-05,
      "loss": 4.7181,
      "step": 5246
    },
    {
      "epoch": 0.21182882519176424,
      "grad_norm": 165.03054809570312,
      "learning_rate": 1.831605797201277e-05,
      "loss": 4.0747,
      "step": 5247
    },
    {
      "epoch": 0.21186919660880096,
      "grad_norm": 82.73961639404297,
      "learning_rate": 1.8315331714753674e-05,
      "loss": 6.0405,
      "step": 5248
    },
    {
      "epoch": 0.2119095680258377,
      "grad_norm": 116.79312896728516,
      "learning_rate": 1.8314605315321824e-05,
      "loss": 5.511,
      "step": 5249
    },
    {
      "epoch": 0.21194993944287444,
      "grad_norm": 147.4176025390625,
      "learning_rate": 1.831387877372964e-05,
      "loss": 4.631,
      "step": 5250
    },
    {
      "epoch": 0.21199031085991119,
      "grad_norm": 186.2315216064453,
      "learning_rate": 1.8313152089989544e-05,
      "loss": 6.434,
      "step": 5251
    },
    {
      "epoch": 0.2120306822769479,
      "grad_norm": 140.97996520996094,
      "learning_rate": 1.831242526411396e-05,
      "loss": 5.4538,
      "step": 5252
    },
    {
      "epoch": 0.21207105369398466,
      "grad_norm": 84.86109161376953,
      "learning_rate": 1.8311698296115316e-05,
      "loss": 6.634,
      "step": 5253
    },
    {
      "epoch": 0.2121114251110214,
      "grad_norm": 111.1642837524414,
      "learning_rate": 1.831097118600604e-05,
      "loss": 6.0621,
      "step": 5254
    },
    {
      "epoch": 0.21215179652805813,
      "grad_norm": 127.34708404541016,
      "learning_rate": 1.831024393379856e-05,
      "loss": 4.057,
      "step": 5255
    },
    {
      "epoch": 0.21219216794509488,
      "grad_norm": 130.35362243652344,
      "learning_rate": 1.830951653950532e-05,
      "loss": 5.2785,
      "step": 5256
    },
    {
      "epoch": 0.2122325393621316,
      "grad_norm": 100.52658081054688,
      "learning_rate": 1.8308789003138748e-05,
      "loss": 6.1689,
      "step": 5257
    },
    {
      "epoch": 0.21227291077916835,
      "grad_norm": 132.9512481689453,
      "learning_rate": 1.8308061324711288e-05,
      "loss": 6.0869,
      "step": 5258
    },
    {
      "epoch": 0.2123132821962051,
      "grad_norm": 124.4642105102539,
      "learning_rate": 1.830733350423538e-05,
      "loss": 5.0672,
      "step": 5259
    },
    {
      "epoch": 0.21235365361324182,
      "grad_norm": 124.4874496459961,
      "learning_rate": 1.8306605541723468e-05,
      "loss": 4.9989,
      "step": 5260
    },
    {
      "epoch": 0.21239402503027857,
      "grad_norm": 97.35135650634766,
      "learning_rate": 1.8305877437187996e-05,
      "loss": 3.9801,
      "step": 5261
    },
    {
      "epoch": 0.2124343964473153,
      "grad_norm": 88.20917510986328,
      "learning_rate": 1.8305149190641417e-05,
      "loss": 6.5645,
      "step": 5262
    },
    {
      "epoch": 0.21247476786435204,
      "grad_norm": 78.09874725341797,
      "learning_rate": 1.8304420802096184e-05,
      "loss": 5.4048,
      "step": 5263
    },
    {
      "epoch": 0.21251513928138877,
      "grad_norm": 161.88316345214844,
      "learning_rate": 1.8303692271564743e-05,
      "loss": 6.3221,
      "step": 5264
    },
    {
      "epoch": 0.21255551069842551,
      "grad_norm": 76.4183578491211,
      "learning_rate": 1.830296359905956e-05,
      "loss": 6.7249,
      "step": 5265
    },
    {
      "epoch": 0.21259588211546226,
      "grad_norm": 100.32877349853516,
      "learning_rate": 1.8302234784593086e-05,
      "loss": 4.8853,
      "step": 5266
    },
    {
      "epoch": 0.212636253532499,
      "grad_norm": 131.69126892089844,
      "learning_rate": 1.8301505828177778e-05,
      "loss": 6.3891,
      "step": 5267
    },
    {
      "epoch": 0.21267662494953574,
      "grad_norm": 187.48165893554688,
      "learning_rate": 1.8300776729826112e-05,
      "loss": 6.4649,
      "step": 5268
    },
    {
      "epoch": 0.21271699636657246,
      "grad_norm": 76.72167205810547,
      "learning_rate": 1.830004748955055e-05,
      "loss": 4.4923,
      "step": 5269
    },
    {
      "epoch": 0.2127573677836092,
      "grad_norm": 102.88972473144531,
      "learning_rate": 1.8299318107363552e-05,
      "loss": 5.0128,
      "step": 5270
    },
    {
      "epoch": 0.21279773920064593,
      "grad_norm": 91.66423797607422,
      "learning_rate": 1.82985885832776e-05,
      "loss": 4.9049,
      "step": 5271
    },
    {
      "epoch": 0.21283811061768268,
      "grad_norm": 75.51097106933594,
      "learning_rate": 1.8297858917305157e-05,
      "loss": 6.1305,
      "step": 5272
    },
    {
      "epoch": 0.21287848203471943,
      "grad_norm": 82.8017349243164,
      "learning_rate": 1.8297129109458704e-05,
      "loss": 5.7183,
      "step": 5273
    },
    {
      "epoch": 0.21291885345175615,
      "grad_norm": 91.32648468017578,
      "learning_rate": 1.8296399159750723e-05,
      "loss": 4.5917,
      "step": 5274
    },
    {
      "epoch": 0.2129592248687929,
      "grad_norm": 109.98236846923828,
      "learning_rate": 1.8295669068193684e-05,
      "loss": 5.4857,
      "step": 5275
    },
    {
      "epoch": 0.21299959628582962,
      "grad_norm": 94.56297302246094,
      "learning_rate": 1.8294938834800078e-05,
      "loss": 5.3726,
      "step": 5276
    },
    {
      "epoch": 0.21303996770286637,
      "grad_norm": 135.5232696533203,
      "learning_rate": 1.829420845958239e-05,
      "loss": 5.9332,
      "step": 5277
    },
    {
      "epoch": 0.2130803391199031,
      "grad_norm": 106.57242584228516,
      "learning_rate": 1.8293477942553104e-05,
      "loss": 5.1361,
      "step": 5278
    },
    {
      "epoch": 0.21312071053693984,
      "grad_norm": 111.76659393310547,
      "learning_rate": 1.829274728372471e-05,
      "loss": 5.0809,
      "step": 5279
    },
    {
      "epoch": 0.2131610819539766,
      "grad_norm": 90.62386322021484,
      "learning_rate": 1.8292016483109704e-05,
      "loss": 6.6386,
      "step": 5280
    },
    {
      "epoch": 0.21320145337101332,
      "grad_norm": 98.25637817382812,
      "learning_rate": 1.829128554072058e-05,
      "loss": 5.3239,
      "step": 5281
    },
    {
      "epoch": 0.21324182478805007,
      "grad_norm": 85.43645477294922,
      "learning_rate": 1.8290554456569835e-05,
      "loss": 5.4743,
      "step": 5282
    },
    {
      "epoch": 0.2132821962050868,
      "grad_norm": 89.47917938232422,
      "learning_rate": 1.8289823230669967e-05,
      "loss": 5.2015,
      "step": 5283
    },
    {
      "epoch": 0.21332256762212354,
      "grad_norm": 119.44398498535156,
      "learning_rate": 1.8289091863033484e-05,
      "loss": 6.0854,
      "step": 5284
    },
    {
      "epoch": 0.2133629390391603,
      "grad_norm": 178.77442932128906,
      "learning_rate": 1.828836035367288e-05,
      "loss": 4.9604,
      "step": 5285
    },
    {
      "epoch": 0.213403310456197,
      "grad_norm": 110.9407730102539,
      "learning_rate": 1.8287628702600674e-05,
      "loss": 4.8926,
      "step": 5286
    },
    {
      "epoch": 0.21344368187323376,
      "grad_norm": 88.48416900634766,
      "learning_rate": 1.828689690982937e-05,
      "loss": 5.8028,
      "step": 5287
    },
    {
      "epoch": 0.21348405329027048,
      "grad_norm": 90.02835083007812,
      "learning_rate": 1.8286164975371476e-05,
      "loss": 5.2861,
      "step": 5288
    },
    {
      "epoch": 0.21352442470730723,
      "grad_norm": 101.82744598388672,
      "learning_rate": 1.8285432899239515e-05,
      "loss": 5.8358,
      "step": 5289
    },
    {
      "epoch": 0.21356479612434395,
      "grad_norm": 88.01942443847656,
      "learning_rate": 1.8284700681445997e-05,
      "loss": 6.3867,
      "step": 5290
    },
    {
      "epoch": 0.2136051675413807,
      "grad_norm": 108.52264404296875,
      "learning_rate": 1.8283968322003446e-05,
      "loss": 4.6197,
      "step": 5291
    },
    {
      "epoch": 0.21364553895841745,
      "grad_norm": 86.41716003417969,
      "learning_rate": 1.8283235820924377e-05,
      "loss": 6.2928,
      "step": 5292
    },
    {
      "epoch": 0.21368591037545417,
      "grad_norm": 288.8370056152344,
      "learning_rate": 1.828250317822132e-05,
      "loss": 3.9485,
      "step": 5293
    },
    {
      "epoch": 0.21372628179249092,
      "grad_norm": 78.10604095458984,
      "learning_rate": 1.82817703939068e-05,
      "loss": 3.8232,
      "step": 5294
    },
    {
      "epoch": 0.21376665320952765,
      "grad_norm": 89.12541198730469,
      "learning_rate": 1.8281037467993345e-05,
      "loss": 4.6147,
      "step": 5295
    },
    {
      "epoch": 0.2138070246265644,
      "grad_norm": 132.8334197998047,
      "learning_rate": 1.8280304400493487e-05,
      "loss": 6.1181,
      "step": 5296
    },
    {
      "epoch": 0.21384739604360112,
      "grad_norm": 111.8790054321289,
      "learning_rate": 1.827957119141976e-05,
      "loss": 4.926,
      "step": 5297
    },
    {
      "epoch": 0.21388776746063787,
      "grad_norm": 144.2393341064453,
      "learning_rate": 1.82788378407847e-05,
      "loss": 4.5703,
      "step": 5298
    },
    {
      "epoch": 0.21392813887767462,
      "grad_norm": 199.41790771484375,
      "learning_rate": 1.8278104348600846e-05,
      "loss": 5.9999,
      "step": 5299
    },
    {
      "epoch": 0.21396851029471134,
      "grad_norm": 81.17115020751953,
      "learning_rate": 1.8277370714880737e-05,
      "loss": 6.1025,
      "step": 5300
    },
    {
      "epoch": 0.2140088817117481,
      "grad_norm": 101.17658233642578,
      "learning_rate": 1.8276636939636917e-05,
      "loss": 6.3426,
      "step": 5301
    },
    {
      "epoch": 0.2140492531287848,
      "grad_norm": 74.53887176513672,
      "learning_rate": 1.8275903022881935e-05,
      "loss": 6.1292,
      "step": 5302
    },
    {
      "epoch": 0.21408962454582156,
      "grad_norm": 142.87867736816406,
      "learning_rate": 1.8275168964628337e-05,
      "loss": 4.9779,
      "step": 5303
    },
    {
      "epoch": 0.21412999596285828,
      "grad_norm": 173.64068603515625,
      "learning_rate": 1.8274434764888673e-05,
      "loss": 5.0938,
      "step": 5304
    },
    {
      "epoch": 0.21417036737989503,
      "grad_norm": 111.71190643310547,
      "learning_rate": 1.8273700423675496e-05,
      "loss": 4.2868,
      "step": 5305
    },
    {
      "epoch": 0.21421073879693178,
      "grad_norm": 127.30928039550781,
      "learning_rate": 1.8272965941001363e-05,
      "loss": 6.7033,
      "step": 5306
    },
    {
      "epoch": 0.2142511102139685,
      "grad_norm": 142.23110961914062,
      "learning_rate": 1.827223131687883e-05,
      "loss": 6.312,
      "step": 5307
    },
    {
      "epoch": 0.21429148163100525,
      "grad_norm": 108.53653717041016,
      "learning_rate": 1.827149655132046e-05,
      "loss": 6.1464,
      "step": 5308
    },
    {
      "epoch": 0.21433185304804198,
      "grad_norm": 153.8266143798828,
      "learning_rate": 1.827076164433881e-05,
      "loss": 4.885,
      "step": 5309
    },
    {
      "epoch": 0.21437222446507873,
      "grad_norm": 114.33370208740234,
      "learning_rate": 1.8270026595946456e-05,
      "loss": 4.4701,
      "step": 5310
    },
    {
      "epoch": 0.21441259588211548,
      "grad_norm": 85.39079284667969,
      "learning_rate": 1.8269291406155954e-05,
      "loss": 6.4532,
      "step": 5311
    },
    {
      "epoch": 0.2144529672991522,
      "grad_norm": 70.0855712890625,
      "learning_rate": 1.8268556074979888e-05,
      "loss": 5.7852,
      "step": 5312
    },
    {
      "epoch": 0.21449333871618895,
      "grad_norm": 148.76174926757812,
      "learning_rate": 1.826782060243081e-05,
      "loss": 6.8364,
      "step": 5313
    },
    {
      "epoch": 0.21453371013322567,
      "grad_norm": 127.32006072998047,
      "learning_rate": 1.8267084988521313e-05,
      "loss": 5.0503,
      "step": 5314
    },
    {
      "epoch": 0.21457408155026242,
      "grad_norm": 124.43413543701172,
      "learning_rate": 1.826634923326397e-05,
      "loss": 6.4045,
      "step": 5315
    },
    {
      "epoch": 0.21461445296729914,
      "grad_norm": 181.00949096679688,
      "learning_rate": 1.826561333667135e-05,
      "loss": 6.185,
      "step": 5316
    },
    {
      "epoch": 0.2146548243843359,
      "grad_norm": 108.70549774169922,
      "learning_rate": 1.826487729875605e-05,
      "loss": 5.588,
      "step": 5317
    },
    {
      "epoch": 0.21469519580137264,
      "grad_norm": 87.47926330566406,
      "learning_rate": 1.8264141119530646e-05,
      "loss": 8.1122,
      "step": 5318
    },
    {
      "epoch": 0.21473556721840936,
      "grad_norm": 101.96343994140625,
      "learning_rate": 1.8263404799007727e-05,
      "loss": 5.7398,
      "step": 5319
    },
    {
      "epoch": 0.2147759386354461,
      "grad_norm": 128.36334228515625,
      "learning_rate": 1.8262668337199882e-05,
      "loss": 4.9399,
      "step": 5320
    },
    {
      "epoch": 0.21481631005248283,
      "grad_norm": 93.1135025024414,
      "learning_rate": 1.8261931734119705e-05,
      "loss": 6.249,
      "step": 5321
    },
    {
      "epoch": 0.21485668146951958,
      "grad_norm": 128.29806518554688,
      "learning_rate": 1.8261194989779785e-05,
      "loss": 6.8053,
      "step": 5322
    },
    {
      "epoch": 0.2148970528865563,
      "grad_norm": 87.93934631347656,
      "learning_rate": 1.8260458104192724e-05,
      "loss": 5.3957,
      "step": 5323
    },
    {
      "epoch": 0.21493742430359306,
      "grad_norm": 91.28656005859375,
      "learning_rate": 1.8259721077371118e-05,
      "loss": 5.1705,
      "step": 5324
    },
    {
      "epoch": 0.2149777957206298,
      "grad_norm": 114.17808532714844,
      "learning_rate": 1.825898390932757e-05,
      "loss": 6.5127,
      "step": 5325
    },
    {
      "epoch": 0.21501816713766653,
      "grad_norm": 111.76937866210938,
      "learning_rate": 1.8258246600074688e-05,
      "loss": 4.6857,
      "step": 5326
    },
    {
      "epoch": 0.21505853855470328,
      "grad_norm": 91.64264678955078,
      "learning_rate": 1.8257509149625067e-05,
      "loss": 3.1023,
      "step": 5327
    },
    {
      "epoch": 0.21509890997174,
      "grad_norm": 109.34944152832031,
      "learning_rate": 1.8256771557991328e-05,
      "loss": 5.2678,
      "step": 5328
    },
    {
      "epoch": 0.21513928138877675,
      "grad_norm": 114.74817657470703,
      "learning_rate": 1.8256033825186066e-05,
      "loss": 5.4446,
      "step": 5329
    },
    {
      "epoch": 0.21517965280581347,
      "grad_norm": 109.86454010009766,
      "learning_rate": 1.8255295951221917e-05,
      "loss": 6.298,
      "step": 5330
    },
    {
      "epoch": 0.21522002422285022,
      "grad_norm": 75.55792236328125,
      "learning_rate": 1.8254557936111478e-05,
      "loss": 5.8245,
      "step": 5331
    },
    {
      "epoch": 0.21526039563988697,
      "grad_norm": 60.574485778808594,
      "learning_rate": 1.8253819779867378e-05,
      "loss": 5.371,
      "step": 5332
    },
    {
      "epoch": 0.2153007670569237,
      "grad_norm": 175.02740478515625,
      "learning_rate": 1.825308148250223e-05,
      "loss": 4.4014,
      "step": 5333
    },
    {
      "epoch": 0.21534113847396044,
      "grad_norm": 149.4921417236328,
      "learning_rate": 1.825234304402866e-05,
      "loss": 5.8153,
      "step": 5334
    },
    {
      "epoch": 0.21538150989099716,
      "grad_norm": 81.73567199707031,
      "learning_rate": 1.8251604464459296e-05,
      "loss": 6.8477,
      "step": 5335
    },
    {
      "epoch": 0.2154218813080339,
      "grad_norm": 105.35745239257812,
      "learning_rate": 1.8250865743806766e-05,
      "loss": 4.0533,
      "step": 5336
    },
    {
      "epoch": 0.21546225272507066,
      "grad_norm": 109.8779525756836,
      "learning_rate": 1.8250126882083697e-05,
      "loss": 6.7391,
      "step": 5337
    },
    {
      "epoch": 0.21550262414210739,
      "grad_norm": 79.0563735961914,
      "learning_rate": 1.8249387879302727e-05,
      "loss": 5.8386,
      "step": 5338
    },
    {
      "epoch": 0.21554299555914413,
      "grad_norm": 128.17474365234375,
      "learning_rate": 1.8248648735476483e-05,
      "loss": 4.254,
      "step": 5339
    },
    {
      "epoch": 0.21558336697618086,
      "grad_norm": 101.17436218261719,
      "learning_rate": 1.8247909450617614e-05,
      "loss": 6.0559,
      "step": 5340
    },
    {
      "epoch": 0.2156237383932176,
      "grad_norm": 101.41075134277344,
      "learning_rate": 1.824717002473875e-05,
      "loss": 6.3101,
      "step": 5341
    },
    {
      "epoch": 0.21566410981025433,
      "grad_norm": 127.82572937011719,
      "learning_rate": 1.8246430457852534e-05,
      "loss": 5.1584,
      "step": 5342
    },
    {
      "epoch": 0.21570448122729108,
      "grad_norm": 87.25179290771484,
      "learning_rate": 1.8245690749971618e-05,
      "loss": 5.6786,
      "step": 5343
    },
    {
      "epoch": 0.21574485264432783,
      "grad_norm": 127.26126861572266,
      "learning_rate": 1.8244950901108648e-05,
      "loss": 5.0508,
      "step": 5344
    },
    {
      "epoch": 0.21578522406136455,
      "grad_norm": 87.97650909423828,
      "learning_rate": 1.8244210911276264e-05,
      "loss": 6.7527,
      "step": 5345
    },
    {
      "epoch": 0.2158255954784013,
      "grad_norm": 113.64728546142578,
      "learning_rate": 1.824347078048713e-05,
      "loss": 5.1373,
      "step": 5346
    },
    {
      "epoch": 0.21586596689543802,
      "grad_norm": 135.55772399902344,
      "learning_rate": 1.8242730508753893e-05,
      "loss": 5.2026,
      "step": 5347
    },
    {
      "epoch": 0.21590633831247477,
      "grad_norm": 126.55379486083984,
      "learning_rate": 1.8241990096089215e-05,
      "loss": 5.1026,
      "step": 5348
    },
    {
      "epoch": 0.2159467097295115,
      "grad_norm": 93.95369720458984,
      "learning_rate": 1.824124954250575e-05,
      "loss": 5.7701,
      "step": 5349
    },
    {
      "epoch": 0.21598708114654824,
      "grad_norm": 138.56787109375,
      "learning_rate": 1.8240508848016164e-05,
      "loss": 4.8219,
      "step": 5350
    },
    {
      "epoch": 0.216027452563585,
      "grad_norm": 149.61370849609375,
      "learning_rate": 1.823976801263312e-05,
      "loss": 5.263,
      "step": 5351
    },
    {
      "epoch": 0.21606782398062171,
      "grad_norm": 85.5859603881836,
      "learning_rate": 1.8239027036369284e-05,
      "loss": 4.1871,
      "step": 5352
    },
    {
      "epoch": 0.21610819539765846,
      "grad_norm": 141.74317932128906,
      "learning_rate": 1.8238285919237326e-05,
      "loss": 5.5367,
      "step": 5353
    },
    {
      "epoch": 0.2161485668146952,
      "grad_norm": 99.98514556884766,
      "learning_rate": 1.8237544661249914e-05,
      "loss": 7.3249,
      "step": 5354
    },
    {
      "epoch": 0.21618893823173194,
      "grad_norm": 97.35282897949219,
      "learning_rate": 1.8236803262419727e-05,
      "loss": 5.4371,
      "step": 5355
    },
    {
      "epoch": 0.21622930964876866,
      "grad_norm": 95.41091918945312,
      "learning_rate": 1.8236061722759438e-05,
      "loss": 6.4576,
      "step": 5356
    },
    {
      "epoch": 0.2162696810658054,
      "grad_norm": 100.06019592285156,
      "learning_rate": 1.8235320042281723e-05,
      "loss": 4.1303,
      "step": 5357
    },
    {
      "epoch": 0.21631005248284216,
      "grad_norm": 96.01253509521484,
      "learning_rate": 1.8234578220999268e-05,
      "loss": 5.5932,
      "step": 5358
    },
    {
      "epoch": 0.21635042389987888,
      "grad_norm": 101.2815170288086,
      "learning_rate": 1.8233836258924755e-05,
      "loss": 5.6228,
      "step": 5359
    },
    {
      "epoch": 0.21639079531691563,
      "grad_norm": 111.14167785644531,
      "learning_rate": 1.8233094156070872e-05,
      "loss": 5.9488,
      "step": 5360
    },
    {
      "epoch": 0.21643116673395235,
      "grad_norm": 77.86092376708984,
      "learning_rate": 1.8232351912450302e-05,
      "loss": 3.0972,
      "step": 5361
    },
    {
      "epoch": 0.2164715381509891,
      "grad_norm": 113.79451751708984,
      "learning_rate": 1.8231609528075737e-05,
      "loss": 5.3041,
      "step": 5362
    },
    {
      "epoch": 0.21651190956802585,
      "grad_norm": 112.91866302490234,
      "learning_rate": 1.823086700295987e-05,
      "loss": 5.2535,
      "step": 5363
    },
    {
      "epoch": 0.21655228098506257,
      "grad_norm": 84.43408203125,
      "learning_rate": 1.8230124337115406e-05,
      "loss": 5.7811,
      "step": 5364
    },
    {
      "epoch": 0.21659265240209932,
      "grad_norm": 113.53007507324219,
      "learning_rate": 1.8229381530555024e-05,
      "loss": 6.1346,
      "step": 5365
    },
    {
      "epoch": 0.21663302381913604,
      "grad_norm": 85.41302490234375,
      "learning_rate": 1.8228638583291442e-05,
      "loss": 6.451,
      "step": 5366
    },
    {
      "epoch": 0.2166733952361728,
      "grad_norm": 89.40203094482422,
      "learning_rate": 1.8227895495337354e-05,
      "loss": 6.005,
      "step": 5367
    },
    {
      "epoch": 0.21671376665320952,
      "grad_norm": 164.45199584960938,
      "learning_rate": 1.8227152266705467e-05,
      "loss": 5.5429,
      "step": 5368
    },
    {
      "epoch": 0.21675413807024627,
      "grad_norm": 243.4508514404297,
      "learning_rate": 1.8226408897408484e-05,
      "loss": 5.867,
      "step": 5369
    },
    {
      "epoch": 0.21679450948728302,
      "grad_norm": 94.8363037109375,
      "learning_rate": 1.8225665387459123e-05,
      "loss": 5.7204,
      "step": 5370
    },
    {
      "epoch": 0.21683488090431974,
      "grad_norm": 101.0843276977539,
      "learning_rate": 1.8224921736870088e-05,
      "loss": 6.6813,
      "step": 5371
    },
    {
      "epoch": 0.2168752523213565,
      "grad_norm": 371.1473693847656,
      "learning_rate": 1.8224177945654102e-05,
      "loss": 5.1869,
      "step": 5372
    },
    {
      "epoch": 0.2169156237383932,
      "grad_norm": 92.75496673583984,
      "learning_rate": 1.8223434013823875e-05,
      "loss": 5.8171,
      "step": 5373
    },
    {
      "epoch": 0.21695599515542996,
      "grad_norm": 75.66838073730469,
      "learning_rate": 1.822268994139213e-05,
      "loss": 5.93,
      "step": 5374
    },
    {
      "epoch": 0.21699636657246668,
      "grad_norm": 60.56167984008789,
      "learning_rate": 1.8221945728371592e-05,
      "loss": 7.0012,
      "step": 5375
    },
    {
      "epoch": 0.21703673798950343,
      "grad_norm": 126.25244903564453,
      "learning_rate": 1.822120137477498e-05,
      "loss": 5.3071,
      "step": 5376
    },
    {
      "epoch": 0.21707710940654018,
      "grad_norm": 103.90780639648438,
      "learning_rate": 1.822045688061502e-05,
      "loss": 5.7626,
      "step": 5377
    },
    {
      "epoch": 0.2171174808235769,
      "grad_norm": 92.8706283569336,
      "learning_rate": 1.8219712245904446e-05,
      "loss": 4.6402,
      "step": 5378
    },
    {
      "epoch": 0.21715785224061365,
      "grad_norm": 84.40364837646484,
      "learning_rate": 1.8218967470655988e-05,
      "loss": 4.381,
      "step": 5379
    },
    {
      "epoch": 0.21719822365765037,
      "grad_norm": 112.45024108886719,
      "learning_rate": 1.8218222554882374e-05,
      "loss": 6.0935,
      "step": 5380
    },
    {
      "epoch": 0.21723859507468712,
      "grad_norm": 74.727783203125,
      "learning_rate": 1.821747749859635e-05,
      "loss": 5.404,
      "step": 5381
    },
    {
      "epoch": 0.21727896649172385,
      "grad_norm": 95.77996826171875,
      "learning_rate": 1.821673230181065e-05,
      "loss": 4.6158,
      "step": 5382
    },
    {
      "epoch": 0.2173193379087606,
      "grad_norm": 75.59547424316406,
      "learning_rate": 1.8215986964538008e-05,
      "loss": 5.5111,
      "step": 5383
    },
    {
      "epoch": 0.21735970932579735,
      "grad_norm": 89.24848175048828,
      "learning_rate": 1.8215241486791183e-05,
      "loss": 5.4871,
      "step": 5384
    },
    {
      "epoch": 0.21740008074283407,
      "grad_norm": 143.99221801757812,
      "learning_rate": 1.8214495868582908e-05,
      "loss": 5.1105,
      "step": 5385
    },
    {
      "epoch": 0.21744045215987082,
      "grad_norm": 75.06140899658203,
      "learning_rate": 1.821375010992594e-05,
      "loss": 5.3706,
      "step": 5386
    },
    {
      "epoch": 0.21748082357690754,
      "grad_norm": 125.39849853515625,
      "learning_rate": 1.821300421083302e-05,
      "loss": 6.5476,
      "step": 5387
    },
    {
      "epoch": 0.2175211949939443,
      "grad_norm": 97.41854095458984,
      "learning_rate": 1.821225817131691e-05,
      "loss": 5.5949,
      "step": 5388
    },
    {
      "epoch": 0.21756156641098104,
      "grad_norm": 75.30164337158203,
      "learning_rate": 1.821151199139036e-05,
      "loss": 6.7796,
      "step": 5389
    },
    {
      "epoch": 0.21760193782801776,
      "grad_norm": 89.52284240722656,
      "learning_rate": 1.8210765671066135e-05,
      "loss": 5.7139,
      "step": 5390
    },
    {
      "epoch": 0.2176423092450545,
      "grad_norm": 121.22701263427734,
      "learning_rate": 1.8210019210356986e-05,
      "loss": 5.8644,
      "step": 5391
    },
    {
      "epoch": 0.21768268066209123,
      "grad_norm": 99.9977798461914,
      "learning_rate": 1.820927260927568e-05,
      "loss": 5.9148,
      "step": 5392
    },
    {
      "epoch": 0.21772305207912798,
      "grad_norm": 78.09201049804688,
      "learning_rate": 1.8208525867834987e-05,
      "loss": 5.8915,
      "step": 5393
    },
    {
      "epoch": 0.2177634234961647,
      "grad_norm": 95.56829071044922,
      "learning_rate": 1.8207778986047668e-05,
      "loss": 4.929,
      "step": 5394
    },
    {
      "epoch": 0.21780379491320145,
      "grad_norm": 81.27755737304688,
      "learning_rate": 1.8207031963926494e-05,
      "loss": 5.631,
      "step": 5395
    },
    {
      "epoch": 0.2178441663302382,
      "grad_norm": 92.38385772705078,
      "learning_rate": 1.8206284801484237e-05,
      "loss": 4.3994,
      "step": 5396
    },
    {
      "epoch": 0.21788453774727493,
      "grad_norm": 108.43315887451172,
      "learning_rate": 1.8205537498733675e-05,
      "loss": 4.2358,
      "step": 5397
    },
    {
      "epoch": 0.21792490916431168,
      "grad_norm": 65.001953125,
      "learning_rate": 1.8204790055687583e-05,
      "loss": 5.6244,
      "step": 5398
    },
    {
      "epoch": 0.2179652805813484,
      "grad_norm": 131.18597412109375,
      "learning_rate": 1.8204042472358737e-05,
      "loss": 7.1512,
      "step": 5399
    },
    {
      "epoch": 0.21800565199838515,
      "grad_norm": 145.33596801757812,
      "learning_rate": 1.8203294748759926e-05,
      "loss": 6.9965,
      "step": 5400
    },
    {
      "epoch": 0.21804602341542187,
      "grad_norm": 87.66667175292969,
      "learning_rate": 1.8202546884903932e-05,
      "loss": 5.4672,
      "step": 5401
    },
    {
      "epoch": 0.21808639483245862,
      "grad_norm": 96.73646545410156,
      "learning_rate": 1.8201798880803538e-05,
      "loss": 5.204,
      "step": 5402
    },
    {
      "epoch": 0.21812676624949537,
      "grad_norm": 91.9936752319336,
      "learning_rate": 1.8201050736471537e-05,
      "loss": 5.4392,
      "step": 5403
    },
    {
      "epoch": 0.2181671376665321,
      "grad_norm": 80.16327667236328,
      "learning_rate": 1.820030245192072e-05,
      "loss": 4.7084,
      "step": 5404
    },
    {
      "epoch": 0.21820750908356884,
      "grad_norm": 121.86344909667969,
      "learning_rate": 1.8199554027163878e-05,
      "loss": 4.4552,
      "step": 5405
    },
    {
      "epoch": 0.21824788050060556,
      "grad_norm": 110.01303100585938,
      "learning_rate": 1.819880546221381e-05,
      "loss": 5.9915,
      "step": 5406
    },
    {
      "epoch": 0.2182882519176423,
      "grad_norm": 172.9178466796875,
      "learning_rate": 1.8198056757083314e-05,
      "loss": 4.8467,
      "step": 5407
    },
    {
      "epoch": 0.21832862333467903,
      "grad_norm": 100.20968627929688,
      "learning_rate": 1.8197307911785193e-05,
      "loss": 5.261,
      "step": 5408
    },
    {
      "epoch": 0.21836899475171578,
      "grad_norm": 97.13964080810547,
      "learning_rate": 1.819655892633225e-05,
      "loss": 7.544,
      "step": 5409
    },
    {
      "epoch": 0.21840936616875253,
      "grad_norm": 81.66860961914062,
      "learning_rate": 1.8195809800737285e-05,
      "loss": 6.5713,
      "step": 5410
    },
    {
      "epoch": 0.21844973758578926,
      "grad_norm": 111.65184783935547,
      "learning_rate": 1.8195060535013113e-05,
      "loss": 5.1778,
      "step": 5411
    },
    {
      "epoch": 0.218490109002826,
      "grad_norm": 71.44951629638672,
      "learning_rate": 1.8194311129172544e-05,
      "loss": 6.2308,
      "step": 5412
    },
    {
      "epoch": 0.21853048041986273,
      "grad_norm": 115.83464813232422,
      "learning_rate": 1.819356158322839e-05,
      "loss": 5.8375,
      "step": 5413
    },
    {
      "epoch": 0.21857085183689948,
      "grad_norm": 131.20556640625,
      "learning_rate": 1.8192811897193462e-05,
      "loss": 5.159,
      "step": 5414
    },
    {
      "epoch": 0.21861122325393623,
      "grad_norm": 116.12657165527344,
      "learning_rate": 1.8192062071080587e-05,
      "loss": 7.6297,
      "step": 5415
    },
    {
      "epoch": 0.21865159467097295,
      "grad_norm": 71.2251205444336,
      "learning_rate": 1.819131210490258e-05,
      "loss": 5.4591,
      "step": 5416
    },
    {
      "epoch": 0.2186919660880097,
      "grad_norm": 120.88654327392578,
      "learning_rate": 1.8190561998672263e-05,
      "loss": 4.4341,
      "step": 5417
    },
    {
      "epoch": 0.21873233750504642,
      "grad_norm": 78.89833068847656,
      "learning_rate": 1.818981175240246e-05,
      "loss": 4.787,
      "step": 5418
    },
    {
      "epoch": 0.21877270892208317,
      "grad_norm": 80.40131378173828,
      "learning_rate": 1.8189061366106005e-05,
      "loss": 5.8005,
      "step": 5419
    },
    {
      "epoch": 0.2188130803391199,
      "grad_norm": 162.01820373535156,
      "learning_rate": 1.818831083979572e-05,
      "loss": 4.3354,
      "step": 5420
    },
    {
      "epoch": 0.21885345175615664,
      "grad_norm": 75.10767364501953,
      "learning_rate": 1.8187560173484442e-05,
      "loss": 4.4791,
      "step": 5421
    },
    {
      "epoch": 0.2188938231731934,
      "grad_norm": 165.4917755126953,
      "learning_rate": 1.8186809367185003e-05,
      "loss": 7.0179,
      "step": 5422
    },
    {
      "epoch": 0.2189341945902301,
      "grad_norm": 147.8934783935547,
      "learning_rate": 1.8186058420910245e-05,
      "loss": 3.2584,
      "step": 5423
    },
    {
      "epoch": 0.21897456600726686,
      "grad_norm": 95.11578369140625,
      "learning_rate": 1.8185307334673e-05,
      "loss": 5.4508,
      "step": 5424
    },
    {
      "epoch": 0.21901493742430359,
      "grad_norm": 85.7662582397461,
      "learning_rate": 1.8184556108486115e-05,
      "loss": 5.1045,
      "step": 5425
    },
    {
      "epoch": 0.21905530884134033,
      "grad_norm": 116.12051391601562,
      "learning_rate": 1.818380474236243e-05,
      "loss": 6.5222,
      "step": 5426
    },
    {
      "epoch": 0.21909568025837706,
      "grad_norm": 94.99061584472656,
      "learning_rate": 1.81830532363148e-05,
      "loss": 6.2366,
      "step": 5427
    },
    {
      "epoch": 0.2191360516754138,
      "grad_norm": 111.49121856689453,
      "learning_rate": 1.8182301590356065e-05,
      "loss": 5.7528,
      "step": 5428
    },
    {
      "epoch": 0.21917642309245056,
      "grad_norm": 156.98928833007812,
      "learning_rate": 1.8181549804499082e-05,
      "loss": 6.4348,
      "step": 5429
    },
    {
      "epoch": 0.21921679450948728,
      "grad_norm": 188.8536376953125,
      "learning_rate": 1.8180797878756702e-05,
      "loss": 5.0314,
      "step": 5430
    },
    {
      "epoch": 0.21925716592652403,
      "grad_norm": 96.6431884765625,
      "learning_rate": 1.818004581314178e-05,
      "loss": 5.164,
      "step": 5431
    },
    {
      "epoch": 0.21929753734356075,
      "grad_norm": 160.5419464111328,
      "learning_rate": 1.8179293607667177e-05,
      "loss": 5.5584,
      "step": 5432
    },
    {
      "epoch": 0.2193379087605975,
      "grad_norm": 75.20992279052734,
      "learning_rate": 1.8178541262345756e-05,
      "loss": 5.4325,
      "step": 5433
    },
    {
      "epoch": 0.21937828017763422,
      "grad_norm": 91.26219940185547,
      "learning_rate": 1.8177788777190376e-05,
      "loss": 6.0488,
      "step": 5434
    },
    {
      "epoch": 0.21941865159467097,
      "grad_norm": 123.61161804199219,
      "learning_rate": 1.8177036152213902e-05,
      "loss": 4.0192,
      "step": 5435
    },
    {
      "epoch": 0.21945902301170772,
      "grad_norm": 96.13117980957031,
      "learning_rate": 1.8176283387429205e-05,
      "loss": 5.6493,
      "step": 5436
    },
    {
      "epoch": 0.21949939442874444,
      "grad_norm": 117.7236557006836,
      "learning_rate": 1.817553048284916e-05,
      "loss": 5.5795,
      "step": 5437
    },
    {
      "epoch": 0.2195397658457812,
      "grad_norm": 83.93763732910156,
      "learning_rate": 1.817477743848663e-05,
      "loss": 5.8843,
      "step": 5438
    },
    {
      "epoch": 0.21958013726281791,
      "grad_norm": 83.81011199951172,
      "learning_rate": 1.8174024254354495e-05,
      "loss": 4.7547,
      "step": 5439
    },
    {
      "epoch": 0.21962050867985466,
      "grad_norm": 112.24099731445312,
      "learning_rate": 1.8173270930465637e-05,
      "loss": 6.033,
      "step": 5440
    },
    {
      "epoch": 0.21966088009689141,
      "grad_norm": 157.25802612304688,
      "learning_rate": 1.8172517466832928e-05,
      "loss": 5.8741,
      "step": 5441
    },
    {
      "epoch": 0.21970125151392814,
      "grad_norm": 80.93051147460938,
      "learning_rate": 1.817176386346926e-05,
      "loss": 3.9399,
      "step": 5442
    },
    {
      "epoch": 0.21974162293096489,
      "grad_norm": 100.53251647949219,
      "learning_rate": 1.8171010120387505e-05,
      "loss": 5.8444,
      "step": 5443
    },
    {
      "epoch": 0.2197819943480016,
      "grad_norm": 94.42731475830078,
      "learning_rate": 1.8170256237600562e-05,
      "loss": 5.4052,
      "step": 5444
    },
    {
      "epoch": 0.21982236576503836,
      "grad_norm": 131.13461303710938,
      "learning_rate": 1.816950221512132e-05,
      "loss": 4.8183,
      "step": 5445
    },
    {
      "epoch": 0.21986273718207508,
      "grad_norm": 91.0704116821289,
      "learning_rate": 1.8168748052962662e-05,
      "loss": 5.8167,
      "step": 5446
    },
    {
      "epoch": 0.21990310859911183,
      "grad_norm": 306.8412780761719,
      "learning_rate": 1.8167993751137488e-05,
      "loss": 5.629,
      "step": 5447
    },
    {
      "epoch": 0.21994348001614858,
      "grad_norm": 76.59649658203125,
      "learning_rate": 1.8167239309658696e-05,
      "loss": 4.3016,
      "step": 5448
    },
    {
      "epoch": 0.2199838514331853,
      "grad_norm": 104.41143798828125,
      "learning_rate": 1.8166484728539183e-05,
      "loss": 5.4637,
      "step": 5449
    },
    {
      "epoch": 0.22002422285022205,
      "grad_norm": 135.8785400390625,
      "learning_rate": 1.816573000779185e-05,
      "loss": 5.649,
      "step": 5450
    },
    {
      "epoch": 0.22006459426725877,
      "grad_norm": 94.92631530761719,
      "learning_rate": 1.8164975147429603e-05,
      "loss": 5.3997,
      "step": 5451
    },
    {
      "epoch": 0.22010496568429552,
      "grad_norm": 90.49381256103516,
      "learning_rate": 1.8164220147465346e-05,
      "loss": 5.017,
      "step": 5452
    },
    {
      "epoch": 0.22014533710133224,
      "grad_norm": 78.56593322753906,
      "learning_rate": 1.816346500791199e-05,
      "loss": 3.6875,
      "step": 5453
    },
    {
      "epoch": 0.220185708518369,
      "grad_norm": 113.817626953125,
      "learning_rate": 1.8162709728782448e-05,
      "loss": 6.1252,
      "step": 5454
    },
    {
      "epoch": 0.22022607993540574,
      "grad_norm": 106.97196197509766,
      "learning_rate": 1.816195431008963e-05,
      "loss": 5.4938,
      "step": 5455
    },
    {
      "epoch": 0.22026645135244247,
      "grad_norm": 135.59646606445312,
      "learning_rate": 1.816119875184645e-05,
      "loss": 5.9982,
      "step": 5456
    },
    {
      "epoch": 0.22030682276947922,
      "grad_norm": 152.4300079345703,
      "learning_rate": 1.8160443054065836e-05,
      "loss": 4.9644,
      "step": 5457
    },
    {
      "epoch": 0.22034719418651594,
      "grad_norm": 153.7608184814453,
      "learning_rate": 1.8159687216760694e-05,
      "loss": 5.6802,
      "step": 5458
    },
    {
      "epoch": 0.2203875656035527,
      "grad_norm": 358.0734558105469,
      "learning_rate": 1.8158931239943957e-05,
      "loss": 4.4976,
      "step": 5459
    },
    {
      "epoch": 0.2204279370205894,
      "grad_norm": 144.22491455078125,
      "learning_rate": 1.815817512362855e-05,
      "loss": 4.1694,
      "step": 5460
    },
    {
      "epoch": 0.22046830843762616,
      "grad_norm": 115.63996887207031,
      "learning_rate": 1.81574188678274e-05,
      "loss": 5.3881,
      "step": 5461
    },
    {
      "epoch": 0.2205086798546629,
      "grad_norm": 119.21199798583984,
      "learning_rate": 1.8156662472553435e-05,
      "loss": 5.5492,
      "step": 5462
    },
    {
      "epoch": 0.22054905127169963,
      "grad_norm": 129.82530212402344,
      "learning_rate": 1.8155905937819586e-05,
      "loss": 6.215,
      "step": 5463
    },
    {
      "epoch": 0.22058942268873638,
      "grad_norm": 73.7117919921875,
      "learning_rate": 1.8155149263638797e-05,
      "loss": 4.0463,
      "step": 5464
    },
    {
      "epoch": 0.2206297941057731,
      "grad_norm": 102.76744079589844,
      "learning_rate": 1.8154392450023995e-05,
      "loss": 4.4531,
      "step": 5465
    },
    {
      "epoch": 0.22067016552280985,
      "grad_norm": 107.4192886352539,
      "learning_rate": 1.8153635496988126e-05,
      "loss": 7.5507,
      "step": 5466
    },
    {
      "epoch": 0.2207105369398466,
      "grad_norm": 91.55069732666016,
      "learning_rate": 1.8152878404544125e-05,
      "loss": 4.8233,
      "step": 5467
    },
    {
      "epoch": 0.22075090835688332,
      "grad_norm": 73.39828491210938,
      "learning_rate": 1.8152121172704945e-05,
      "loss": 6.1795,
      "step": 5468
    },
    {
      "epoch": 0.22079127977392007,
      "grad_norm": 103.61373138427734,
      "learning_rate": 1.8151363801483533e-05,
      "loss": 5.4299,
      "step": 5469
    },
    {
      "epoch": 0.2208316511909568,
      "grad_norm": 92.15950775146484,
      "learning_rate": 1.815060629089283e-05,
      "loss": 5.5931,
      "step": 5470
    },
    {
      "epoch": 0.22087202260799355,
      "grad_norm": 105.43399810791016,
      "learning_rate": 1.8149848640945794e-05,
      "loss": 4.8874,
      "step": 5471
    },
    {
      "epoch": 0.22091239402503027,
      "grad_norm": 139.50709533691406,
      "learning_rate": 1.814909085165538e-05,
      "loss": 5.6326,
      "step": 5472
    },
    {
      "epoch": 0.22095276544206702,
      "grad_norm": 243.6942901611328,
      "learning_rate": 1.8148332923034538e-05,
      "loss": 5.5356,
      "step": 5473
    },
    {
      "epoch": 0.22099313685910377,
      "grad_norm": 85.68035888671875,
      "learning_rate": 1.8147574855096234e-05,
      "loss": 5.0685,
      "step": 5474
    },
    {
      "epoch": 0.2210335082761405,
      "grad_norm": 116.79997253417969,
      "learning_rate": 1.814681664785342e-05,
      "loss": 5.6814,
      "step": 5475
    },
    {
      "epoch": 0.22107387969317724,
      "grad_norm": 81.93087005615234,
      "learning_rate": 1.8146058301319073e-05,
      "loss": 4.6429,
      "step": 5476
    },
    {
      "epoch": 0.22111425111021396,
      "grad_norm": 89.57183837890625,
      "learning_rate": 1.814529981550615e-05,
      "loss": 5.237,
      "step": 5477
    },
    {
      "epoch": 0.2211546225272507,
      "grad_norm": 75.65913391113281,
      "learning_rate": 1.8144541190427617e-05,
      "loss": 6.2842,
      "step": 5478
    },
    {
      "epoch": 0.22119499394428743,
      "grad_norm": 111.90998077392578,
      "learning_rate": 1.814378242609645e-05,
      "loss": 5.8315,
      "step": 5479
    },
    {
      "epoch": 0.22123536536132418,
      "grad_norm": 80.30455017089844,
      "learning_rate": 1.814302352252562e-05,
      "loss": 5.9436,
      "step": 5480
    },
    {
      "epoch": 0.22127573677836093,
      "grad_norm": 400.4461975097656,
      "learning_rate": 1.8142264479728106e-05,
      "loss": 5.1224,
      "step": 5481
    },
    {
      "epoch": 0.22131610819539765,
      "grad_norm": 151.66517639160156,
      "learning_rate": 1.814150529771688e-05,
      "loss": 6.1565,
      "step": 5482
    },
    {
      "epoch": 0.2213564796124344,
      "grad_norm": 72.0406265258789,
      "learning_rate": 1.814074597650493e-05,
      "loss": 6.3657,
      "step": 5483
    },
    {
      "epoch": 0.22139685102947113,
      "grad_norm": 73.93132019042969,
      "learning_rate": 1.813998651610523e-05,
      "loss": 6.5232,
      "step": 5484
    },
    {
      "epoch": 0.22143722244650788,
      "grad_norm": 97.68586730957031,
      "learning_rate": 1.813922691653077e-05,
      "loss": 5.8,
      "step": 5485
    },
    {
      "epoch": 0.2214775938635446,
      "grad_norm": 120.62646484375,
      "learning_rate": 1.8138467177794532e-05,
      "loss": 6.3359,
      "step": 5486
    },
    {
      "epoch": 0.22151796528058135,
      "grad_norm": 124.9734115600586,
      "learning_rate": 1.813770729990951e-05,
      "loss": 5.7792,
      "step": 5487
    },
    {
      "epoch": 0.2215583366976181,
      "grad_norm": 79.4189682006836,
      "learning_rate": 1.8136947282888704e-05,
      "loss": 4.6531,
      "step": 5488
    },
    {
      "epoch": 0.22159870811465482,
      "grad_norm": 90.59032440185547,
      "learning_rate": 1.8136187126745096e-05,
      "loss": 3.667,
      "step": 5489
    },
    {
      "epoch": 0.22163907953169157,
      "grad_norm": 151.8572998046875,
      "learning_rate": 1.813542683149169e-05,
      "loss": 4.9824,
      "step": 5490
    },
    {
      "epoch": 0.2216794509487283,
      "grad_norm": 94.71228790283203,
      "learning_rate": 1.8134666397141475e-05,
      "loss": 4.1354,
      "step": 5491
    },
    {
      "epoch": 0.22171982236576504,
      "grad_norm": 100.46756744384766,
      "learning_rate": 1.813390582370747e-05,
      "loss": 5.0164,
      "step": 5492
    },
    {
      "epoch": 0.2217601937828018,
      "grad_norm": 93.71784210205078,
      "learning_rate": 1.8133145111202664e-05,
      "loss": 5.6407,
      "step": 5493
    },
    {
      "epoch": 0.2218005651998385,
      "grad_norm": 98.87178039550781,
      "learning_rate": 1.813238425964007e-05,
      "loss": 3.7689,
      "step": 5494
    },
    {
      "epoch": 0.22184093661687526,
      "grad_norm": 90.83344268798828,
      "learning_rate": 1.8131623269032694e-05,
      "loss": 6.0592,
      "step": 5495
    },
    {
      "epoch": 0.22188130803391198,
      "grad_norm": 91.5707015991211,
      "learning_rate": 1.813086213939355e-05,
      "loss": 5.1985,
      "step": 5496
    },
    {
      "epoch": 0.22192167945094873,
      "grad_norm": 104.23589324951172,
      "learning_rate": 1.813010087073565e-05,
      "loss": 6.5357,
      "step": 5497
    },
    {
      "epoch": 0.22196205086798546,
      "grad_norm": 108.50618743896484,
      "learning_rate": 1.8129339463072014e-05,
      "loss": 7.2934,
      "step": 5498
    },
    {
      "epoch": 0.2220024222850222,
      "grad_norm": 85.52540588378906,
      "learning_rate": 1.812857791641565e-05,
      "loss": 5.4005,
      "step": 5499
    },
    {
      "epoch": 0.22204279370205895,
      "grad_norm": 139.1644287109375,
      "learning_rate": 1.8127816230779588e-05,
      "loss": 6.447,
      "step": 5500
    },
    {
      "epoch": 0.22208316511909568,
      "grad_norm": 94.95746612548828,
      "learning_rate": 1.812705440617685e-05,
      "loss": 6.0379,
      "step": 5501
    },
    {
      "epoch": 0.22212353653613243,
      "grad_norm": 84.16495513916016,
      "learning_rate": 1.8126292442620463e-05,
      "loss": 6.5533,
      "step": 5502
    },
    {
      "epoch": 0.22216390795316915,
      "grad_norm": 127.25960540771484,
      "learning_rate": 1.8125530340123444e-05,
      "loss": 5.9736,
      "step": 5503
    },
    {
      "epoch": 0.2222042793702059,
      "grad_norm": 198.6323699951172,
      "learning_rate": 1.8124768098698836e-05,
      "loss": 6.2136,
      "step": 5504
    },
    {
      "epoch": 0.22224465078724262,
      "grad_norm": 89.81734466552734,
      "learning_rate": 1.8124005718359662e-05,
      "loss": 5.3479,
      "step": 5505
    },
    {
      "epoch": 0.22228502220427937,
      "grad_norm": 117.7151107788086,
      "learning_rate": 1.8123243199118964e-05,
      "loss": 4.5513,
      "step": 5506
    },
    {
      "epoch": 0.22232539362131612,
      "grad_norm": 93.87519073486328,
      "learning_rate": 1.8122480540989776e-05,
      "loss": 5.6194,
      "step": 5507
    },
    {
      "epoch": 0.22236576503835284,
      "grad_norm": 83.3584213256836,
      "learning_rate": 1.8121717743985138e-05,
      "loss": 5.1756,
      "step": 5508
    },
    {
      "epoch": 0.2224061364553896,
      "grad_norm": 97.22968292236328,
      "learning_rate": 1.812095480811809e-05,
      "loss": 5.0194,
      "step": 5509
    },
    {
      "epoch": 0.2224465078724263,
      "grad_norm": 125.25020599365234,
      "learning_rate": 1.812019173340168e-05,
      "loss": 5.9609,
      "step": 5510
    },
    {
      "epoch": 0.22248687928946306,
      "grad_norm": 162.92405700683594,
      "learning_rate": 1.8119428519848953e-05,
      "loss": 5.6645,
      "step": 5511
    },
    {
      "epoch": 0.22252725070649979,
      "grad_norm": 130.5648193359375,
      "learning_rate": 1.811866516747296e-05,
      "loss": 4.5957,
      "step": 5512
    },
    {
      "epoch": 0.22256762212353653,
      "grad_norm": 73.32884216308594,
      "learning_rate": 1.8117901676286752e-05,
      "loss": 4.3651,
      "step": 5513
    },
    {
      "epoch": 0.22260799354057328,
      "grad_norm": 223.1475830078125,
      "learning_rate": 1.8117138046303378e-05,
      "loss": 5.0276,
      "step": 5514
    },
    {
      "epoch": 0.22264836495761,
      "grad_norm": 320.97979736328125,
      "learning_rate": 1.8116374277535902e-05,
      "loss": 4.8269,
      "step": 5515
    },
    {
      "epoch": 0.22268873637464676,
      "grad_norm": 71.57271575927734,
      "learning_rate": 1.811561036999738e-05,
      "loss": 5.1305,
      "step": 5516
    },
    {
      "epoch": 0.22272910779168348,
      "grad_norm": 156.94200134277344,
      "learning_rate": 1.8114846323700868e-05,
      "loss": 5.2022,
      "step": 5517
    },
    {
      "epoch": 0.22276947920872023,
      "grad_norm": 118.09589385986328,
      "learning_rate": 1.8114082138659433e-05,
      "loss": 4.5944,
      "step": 5518
    },
    {
      "epoch": 0.22280985062575698,
      "grad_norm": 118.62754821777344,
      "learning_rate": 1.8113317814886144e-05,
      "loss": 3.8619,
      "step": 5519
    },
    {
      "epoch": 0.2228502220427937,
      "grad_norm": 94.25054931640625,
      "learning_rate": 1.8112553352394066e-05,
      "loss": 5.8226,
      "step": 5520
    },
    {
      "epoch": 0.22289059345983045,
      "grad_norm": 168.29470825195312,
      "learning_rate": 1.811178875119627e-05,
      "loss": 6.3357,
      "step": 5521
    },
    {
      "epoch": 0.22293096487686717,
      "grad_norm": 120.21690368652344,
      "learning_rate": 1.811102401130583e-05,
      "loss": 4.4241,
      "step": 5522
    },
    {
      "epoch": 0.22297133629390392,
      "grad_norm": 111.1507568359375,
      "learning_rate": 1.8110259132735815e-05,
      "loss": 5.3336,
      "step": 5523
    },
    {
      "epoch": 0.22301170771094064,
      "grad_norm": 81.91039276123047,
      "learning_rate": 1.8109494115499312e-05,
      "loss": 5.3493,
      "step": 5524
    },
    {
      "epoch": 0.2230520791279774,
      "grad_norm": 116.09185791015625,
      "learning_rate": 1.8108728959609397e-05,
      "loss": 6.0587,
      "step": 5525
    },
    {
      "epoch": 0.22309245054501414,
      "grad_norm": 90.28248596191406,
      "learning_rate": 1.810796366507915e-05,
      "loss": 4.5993,
      "step": 5526
    },
    {
      "epoch": 0.22313282196205086,
      "grad_norm": 101.12045288085938,
      "learning_rate": 1.810719823192166e-05,
      "loss": 5.868,
      "step": 5527
    },
    {
      "epoch": 0.22317319337908761,
      "grad_norm": 114.48418426513672,
      "learning_rate": 1.8106432660150006e-05,
      "loss": 5.1982,
      "step": 5528
    },
    {
      "epoch": 0.22321356479612434,
      "grad_norm": 120.97045135498047,
      "learning_rate": 1.810566694977729e-05,
      "loss": 6.9809,
      "step": 5529
    },
    {
      "epoch": 0.22325393621316109,
      "grad_norm": 87.3619384765625,
      "learning_rate": 1.8104901100816595e-05,
      "loss": 5.2944,
      "step": 5530
    },
    {
      "epoch": 0.2232943076301978,
      "grad_norm": 97.82561492919922,
      "learning_rate": 1.8104135113281015e-05,
      "loss": 5.7894,
      "step": 5531
    },
    {
      "epoch": 0.22333467904723456,
      "grad_norm": 102.12408447265625,
      "learning_rate": 1.810336898718365e-05,
      "loss": 7.1812,
      "step": 5532
    },
    {
      "epoch": 0.2233750504642713,
      "grad_norm": 80.06468200683594,
      "learning_rate": 1.81026027225376e-05,
      "loss": 4.1149,
      "step": 5533
    },
    {
      "epoch": 0.22341542188130803,
      "grad_norm": 115.24301147460938,
      "learning_rate": 1.8101836319355967e-05,
      "loss": 4.5483,
      "step": 5534
    },
    {
      "epoch": 0.22345579329834478,
      "grad_norm": 129.00697326660156,
      "learning_rate": 1.810106977765185e-05,
      "loss": 6.0251,
      "step": 5535
    },
    {
      "epoch": 0.2234961647153815,
      "grad_norm": 87.55050659179688,
      "learning_rate": 1.8100303097438355e-05,
      "loss": 5.2615,
      "step": 5536
    },
    {
      "epoch": 0.22353653613241825,
      "grad_norm": 86.26715850830078,
      "learning_rate": 1.8099536278728592e-05,
      "loss": 4.0467,
      "step": 5537
    },
    {
      "epoch": 0.22357690754945497,
      "grad_norm": 165.0093231201172,
      "learning_rate": 1.8098769321535674e-05,
      "loss": 4.103,
      "step": 5538
    },
    {
      "epoch": 0.22361727896649172,
      "grad_norm": 96.84644317626953,
      "learning_rate": 1.8098002225872713e-05,
      "loss": 4.6876,
      "step": 5539
    },
    {
      "epoch": 0.22365765038352847,
      "grad_norm": 80.50726318359375,
      "learning_rate": 1.8097234991752822e-05,
      "loss": 4.9162,
      "step": 5540
    },
    {
      "epoch": 0.2236980218005652,
      "grad_norm": 109.84198760986328,
      "learning_rate": 1.8096467619189123e-05,
      "loss": 7.129,
      "step": 5541
    },
    {
      "epoch": 0.22373839321760194,
      "grad_norm": 141.6226806640625,
      "learning_rate": 1.8095700108194737e-05,
      "loss": 5.942,
      "step": 5542
    },
    {
      "epoch": 0.22377876463463867,
      "grad_norm": 121.44781494140625,
      "learning_rate": 1.809493245878278e-05,
      "loss": 4.9884,
      "step": 5543
    },
    {
      "epoch": 0.22381913605167542,
      "grad_norm": 121.4617691040039,
      "learning_rate": 1.809416467096638e-05,
      "loss": 5.4449,
      "step": 5544
    },
    {
      "epoch": 0.22385950746871217,
      "grad_norm": 118.9612808227539,
      "learning_rate": 1.809339674475867e-05,
      "loss": 5.925,
      "step": 5545
    },
    {
      "epoch": 0.2238998788857489,
      "grad_norm": 94.23463439941406,
      "learning_rate": 1.8092628680172773e-05,
      "loss": 4.9009,
      "step": 5546
    },
    {
      "epoch": 0.22394025030278564,
      "grad_norm": 166.32598876953125,
      "learning_rate": 1.8091860477221824e-05,
      "loss": 6.8029,
      "step": 5547
    },
    {
      "epoch": 0.22398062171982236,
      "grad_norm": 118.76512145996094,
      "learning_rate": 1.8091092135918954e-05,
      "loss": 5.2286,
      "step": 5548
    },
    {
      "epoch": 0.2240209931368591,
      "grad_norm": 102.49996948242188,
      "learning_rate": 1.80903236562773e-05,
      "loss": 6.0778,
      "step": 5549
    },
    {
      "epoch": 0.22406136455389583,
      "grad_norm": 130.42904663085938,
      "learning_rate": 1.808955503831001e-05,
      "loss": 5.8257,
      "step": 5550
    },
    {
      "epoch": 0.22410173597093258,
      "grad_norm": 112.34172821044922,
      "learning_rate": 1.8088786282030218e-05,
      "loss": 6.4729,
      "step": 5551
    },
    {
      "epoch": 0.22414210738796933,
      "grad_norm": 192.08331298828125,
      "learning_rate": 1.8088017387451068e-05,
      "loss": 5.0906,
      "step": 5552
    },
    {
      "epoch": 0.22418247880500605,
      "grad_norm": 101.43475341796875,
      "learning_rate": 1.8087248354585708e-05,
      "loss": 6.0715,
      "step": 5553
    },
    {
      "epoch": 0.2242228502220428,
      "grad_norm": 138.08616638183594,
      "learning_rate": 1.8086479183447285e-05,
      "loss": 6.1536,
      "step": 5554
    },
    {
      "epoch": 0.22426322163907952,
      "grad_norm": 165.26910400390625,
      "learning_rate": 1.8085709874048955e-05,
      "loss": 6.0553,
      "step": 5555
    },
    {
      "epoch": 0.22430359305611627,
      "grad_norm": 90.29935455322266,
      "learning_rate": 1.8084940426403866e-05,
      "loss": 4.7041,
      "step": 5556
    },
    {
      "epoch": 0.224343964473153,
      "grad_norm": 104.26594543457031,
      "learning_rate": 1.8084170840525173e-05,
      "loss": 6.6144,
      "step": 5557
    },
    {
      "epoch": 0.22438433589018975,
      "grad_norm": 122.10856628417969,
      "learning_rate": 1.8083401116426037e-05,
      "loss": 5.5766,
      "step": 5558
    },
    {
      "epoch": 0.2244247073072265,
      "grad_norm": 120.77625274658203,
      "learning_rate": 1.8082631254119622e-05,
      "loss": 6.842,
      "step": 5559
    },
    {
      "epoch": 0.22446507872426322,
      "grad_norm": 142.00509643554688,
      "learning_rate": 1.8081861253619083e-05,
      "loss": 3.2604,
      "step": 5560
    },
    {
      "epoch": 0.22450545014129997,
      "grad_norm": 140.1276092529297,
      "learning_rate": 1.8081091114937588e-05,
      "loss": 5.9168,
      "step": 5561
    },
    {
      "epoch": 0.2245458215583367,
      "grad_norm": 90.4361572265625,
      "learning_rate": 1.8080320838088308e-05,
      "loss": 5.1541,
      "step": 5562
    },
    {
      "epoch": 0.22458619297537344,
      "grad_norm": 110.9000015258789,
      "learning_rate": 1.807955042308441e-05,
      "loss": 3.9275,
      "step": 5563
    },
    {
      "epoch": 0.22462656439241016,
      "grad_norm": 86.75226593017578,
      "learning_rate": 1.8078779869939066e-05,
      "loss": 5.337,
      "step": 5564
    },
    {
      "epoch": 0.2246669358094469,
      "grad_norm": 121.09130096435547,
      "learning_rate": 1.8078009178665452e-05,
      "loss": 7.0663,
      "step": 5565
    },
    {
      "epoch": 0.22470730722648366,
      "grad_norm": 106.36873626708984,
      "learning_rate": 1.8077238349276744e-05,
      "loss": 4.4787,
      "step": 5566
    },
    {
      "epoch": 0.22474767864352038,
      "grad_norm": 149.50205993652344,
      "learning_rate": 1.8076467381786123e-05,
      "loss": 5.4717,
      "step": 5567
    },
    {
      "epoch": 0.22478805006055713,
      "grad_norm": 108.14152526855469,
      "learning_rate": 1.807569627620677e-05,
      "loss": 3.8946,
      "step": 5568
    },
    {
      "epoch": 0.22482842147759385,
      "grad_norm": 83.14802551269531,
      "learning_rate": 1.807492503255187e-05,
      "loss": 4.4028,
      "step": 5569
    },
    {
      "epoch": 0.2248687928946306,
      "grad_norm": 92.63927459716797,
      "learning_rate": 1.8074153650834606e-05,
      "loss": 6.4381,
      "step": 5570
    },
    {
      "epoch": 0.22490916431166735,
      "grad_norm": 98.04116821289062,
      "learning_rate": 1.807338213106817e-05,
      "loss": 6.37,
      "step": 5571
    },
    {
      "epoch": 0.22494953572870408,
      "grad_norm": 172.94984436035156,
      "learning_rate": 1.8072610473265752e-05,
      "loss": 5.5393,
      "step": 5572
    },
    {
      "epoch": 0.22498990714574083,
      "grad_norm": 221.4795379638672,
      "learning_rate": 1.807183867744055e-05,
      "loss": 6.8032,
      "step": 5573
    },
    {
      "epoch": 0.22503027856277755,
      "grad_norm": 133.29690551757812,
      "learning_rate": 1.8071066743605746e-05,
      "loss": 5.0895,
      "step": 5574
    },
    {
      "epoch": 0.2250706499798143,
      "grad_norm": 98.94744110107422,
      "learning_rate": 1.8070294671774555e-05,
      "loss": 4.2257,
      "step": 5575
    },
    {
      "epoch": 0.22511102139685102,
      "grad_norm": 114.63619232177734,
      "learning_rate": 1.806952246196017e-05,
      "loss": 5.1254,
      "step": 5576
    },
    {
      "epoch": 0.22515139281388777,
      "grad_norm": 110.59832000732422,
      "learning_rate": 1.806875011417579e-05,
      "loss": 5.5988,
      "step": 5577
    },
    {
      "epoch": 0.22519176423092452,
      "grad_norm": 148.47479248046875,
      "learning_rate": 1.8067977628434632e-05,
      "loss": 5.938,
      "step": 5578
    },
    {
      "epoch": 0.22523213564796124,
      "grad_norm": 137.329345703125,
      "learning_rate": 1.8067205004749892e-05,
      "loss": 5.1322,
      "step": 5579
    },
    {
      "epoch": 0.225272507064998,
      "grad_norm": 125.2515869140625,
      "learning_rate": 1.8066432243134785e-05,
      "loss": 5.108,
      "step": 5580
    },
    {
      "epoch": 0.2253128784820347,
      "grad_norm": 125.37507629394531,
      "learning_rate": 1.8065659343602523e-05,
      "loss": 6.9518,
      "step": 5581
    },
    {
      "epoch": 0.22535324989907146,
      "grad_norm": 120.81842041015625,
      "learning_rate": 1.8064886306166326e-05,
      "loss": 5.5872,
      "step": 5582
    },
    {
      "epoch": 0.22539362131610818,
      "grad_norm": 140.11228942871094,
      "learning_rate": 1.80641131308394e-05,
      "loss": 6.8269,
      "step": 5583
    },
    {
      "epoch": 0.22543399273314493,
      "grad_norm": 103.69916534423828,
      "learning_rate": 1.8063339817634972e-05,
      "loss": 6.4876,
      "step": 5584
    },
    {
      "epoch": 0.22547436415018168,
      "grad_norm": 88.82202911376953,
      "learning_rate": 1.8062566366566263e-05,
      "loss": 4.8068,
      "step": 5585
    },
    {
      "epoch": 0.2255147355672184,
      "grad_norm": 66.32856750488281,
      "learning_rate": 1.8061792777646497e-05,
      "loss": 5.2788,
      "step": 5586
    },
    {
      "epoch": 0.22555510698425515,
      "grad_norm": 108.1882553100586,
      "learning_rate": 1.80610190508889e-05,
      "loss": 5.6819,
      "step": 5587
    },
    {
      "epoch": 0.22559547840129188,
      "grad_norm": 104.65766143798828,
      "learning_rate": 1.80602451863067e-05,
      "loss": 6.4634,
      "step": 5588
    },
    {
      "epoch": 0.22563584981832863,
      "grad_norm": 94.74604034423828,
      "learning_rate": 1.805947118391313e-05,
      "loss": 6.6303,
      "step": 5589
    },
    {
      "epoch": 0.22567622123536535,
      "grad_norm": 115.4878158569336,
      "learning_rate": 1.8058697043721424e-05,
      "loss": 5.8565,
      "step": 5590
    },
    {
      "epoch": 0.2257165926524021,
      "grad_norm": 84.78722381591797,
      "learning_rate": 1.8057922765744814e-05,
      "loss": 6.0246,
      "step": 5591
    },
    {
      "epoch": 0.22575696406943885,
      "grad_norm": 118.30561828613281,
      "learning_rate": 1.8057148349996546e-05,
      "loss": 6.549,
      "step": 5592
    },
    {
      "epoch": 0.22579733548647557,
      "grad_norm": 85.86272430419922,
      "learning_rate": 1.8056373796489853e-05,
      "loss": 5.4358,
      "step": 5593
    },
    {
      "epoch": 0.22583770690351232,
      "grad_norm": 86.72789001464844,
      "learning_rate": 1.805559910523798e-05,
      "loss": 6.4137,
      "step": 5594
    },
    {
      "epoch": 0.22587807832054904,
      "grad_norm": 94.64897918701172,
      "learning_rate": 1.8054824276254176e-05,
      "loss": 6.2775,
      "step": 5595
    },
    {
      "epoch": 0.2259184497375858,
      "grad_norm": 123.46101379394531,
      "learning_rate": 1.8054049309551685e-05,
      "loss": 5.1156,
      "step": 5596
    },
    {
      "epoch": 0.2259588211546225,
      "grad_norm": 106.2701187133789,
      "learning_rate": 1.8053274205143758e-05,
      "loss": 5.8844,
      "step": 5597
    },
    {
      "epoch": 0.22599919257165926,
      "grad_norm": 98.38101196289062,
      "learning_rate": 1.805249896304365e-05,
      "loss": 5.6828,
      "step": 5598
    },
    {
      "epoch": 0.226039563988696,
      "grad_norm": 233.81085205078125,
      "learning_rate": 1.805172358326461e-05,
      "loss": 6.0786,
      "step": 5599
    },
    {
      "epoch": 0.22607993540573273,
      "grad_norm": 75.73099517822266,
      "learning_rate": 1.80509480658199e-05,
      "loss": 7.5564,
      "step": 5600
    },
    {
      "epoch": 0.22612030682276948,
      "grad_norm": 170.10655212402344,
      "learning_rate": 1.805017241072278e-05,
      "loss": 5.9271,
      "step": 5601
    },
    {
      "epoch": 0.2261606782398062,
      "grad_norm": 93.39633178710938,
      "learning_rate": 1.8049396617986506e-05,
      "loss": 5.1634,
      "step": 5602
    },
    {
      "epoch": 0.22620104965684296,
      "grad_norm": 85.21745300292969,
      "learning_rate": 1.8048620687624352e-05,
      "loss": 5.4465,
      "step": 5603
    },
    {
      "epoch": 0.2262414210738797,
      "grad_norm": 103.02625274658203,
      "learning_rate": 1.8047844619649576e-05,
      "loss": 4.1954,
      "step": 5604
    },
    {
      "epoch": 0.22628179249091643,
      "grad_norm": 67.85591125488281,
      "learning_rate": 1.8047068414075452e-05,
      "loss": 6.4935,
      "step": 5605
    },
    {
      "epoch": 0.22632216390795318,
      "grad_norm": 112.40576934814453,
      "learning_rate": 1.804629207091525e-05,
      "loss": 5.078,
      "step": 5606
    },
    {
      "epoch": 0.2263625353249899,
      "grad_norm": 73.60478973388672,
      "learning_rate": 1.804551559018224e-05,
      "loss": 6.3799,
      "step": 5607
    },
    {
      "epoch": 0.22640290674202665,
      "grad_norm": 74.42050170898438,
      "learning_rate": 1.8044738971889704e-05,
      "loss": 6.1408,
      "step": 5608
    },
    {
      "epoch": 0.22644327815906337,
      "grad_norm": 68.4345932006836,
      "learning_rate": 1.8043962216050913e-05,
      "loss": 5.8425,
      "step": 5609
    },
    {
      "epoch": 0.22648364957610012,
      "grad_norm": 140.0161895751953,
      "learning_rate": 1.804318532267916e-05,
      "loss": 5.5808,
      "step": 5610
    },
    {
      "epoch": 0.22652402099313687,
      "grad_norm": 72.13984680175781,
      "learning_rate": 1.8042408291787713e-05,
      "loss": 5.0164,
      "step": 5611
    },
    {
      "epoch": 0.2265643924101736,
      "grad_norm": 90.00132751464844,
      "learning_rate": 1.8041631123389866e-05,
      "loss": 5.3793,
      "step": 5612
    },
    {
      "epoch": 0.22660476382721034,
      "grad_norm": 103.62703704833984,
      "learning_rate": 1.8040853817498907e-05,
      "loss": 5.936,
      "step": 5613
    },
    {
      "epoch": 0.22664513524424706,
      "grad_norm": 90.7685317993164,
      "learning_rate": 1.8040076374128125e-05,
      "loss": 5.2384,
      "step": 5614
    },
    {
      "epoch": 0.22668550666128381,
      "grad_norm": 110.20957946777344,
      "learning_rate": 1.803929879329081e-05,
      "loss": 4.8716,
      "step": 5615
    },
    {
      "epoch": 0.22672587807832054,
      "grad_norm": 70.57220458984375,
      "learning_rate": 1.803852107500026e-05,
      "loss": 5.0954,
      "step": 5616
    },
    {
      "epoch": 0.22676624949535729,
      "grad_norm": 98.62031555175781,
      "learning_rate": 1.803774321926977e-05,
      "loss": 4.6245,
      "step": 5617
    },
    {
      "epoch": 0.22680662091239404,
      "grad_norm": 134.47972106933594,
      "learning_rate": 1.8036965226112638e-05,
      "loss": 6.5529,
      "step": 5618
    },
    {
      "epoch": 0.22684699232943076,
      "grad_norm": 88.08474731445312,
      "learning_rate": 1.8036187095542172e-05,
      "loss": 6.0033,
      "step": 5619
    },
    {
      "epoch": 0.2268873637464675,
      "grad_norm": 122.486328125,
      "learning_rate": 1.8035408827571673e-05,
      "loss": 6.4632,
      "step": 5620
    },
    {
      "epoch": 0.22692773516350423,
      "grad_norm": 83.71237182617188,
      "learning_rate": 1.8034630422214446e-05,
      "loss": 5.2218,
      "step": 5621
    },
    {
      "epoch": 0.22696810658054098,
      "grad_norm": 115.9239501953125,
      "learning_rate": 1.80338518794838e-05,
      "loss": 6.3197,
      "step": 5622
    },
    {
      "epoch": 0.2270084779975777,
      "grad_norm": 97.0765609741211,
      "learning_rate": 1.8033073199393048e-05,
      "loss": 6.4401,
      "step": 5623
    },
    {
      "epoch": 0.22704884941461445,
      "grad_norm": 122.8407974243164,
      "learning_rate": 1.80322943819555e-05,
      "loss": 4.7982,
      "step": 5624
    },
    {
      "epoch": 0.2270892208316512,
      "grad_norm": 97.75566864013672,
      "learning_rate": 1.8031515427184478e-05,
      "loss": 6.7949,
      "step": 5625
    },
    {
      "epoch": 0.22712959224868792,
      "grad_norm": 69.27980041503906,
      "learning_rate": 1.80307363350933e-05,
      "loss": 3.5049,
      "step": 5626
    },
    {
      "epoch": 0.22716996366572467,
      "grad_norm": 85.1413345336914,
      "learning_rate": 1.802995710569528e-05,
      "loss": 5.6261,
      "step": 5627
    },
    {
      "epoch": 0.2272103350827614,
      "grad_norm": 71.49475860595703,
      "learning_rate": 1.8029177739003745e-05,
      "loss": 6.1974,
      "step": 5628
    },
    {
      "epoch": 0.22725070649979814,
      "grad_norm": 85.12317657470703,
      "learning_rate": 1.802839823503202e-05,
      "loss": 7.4836,
      "step": 5629
    },
    {
      "epoch": 0.2272910779168349,
      "grad_norm": 103.32051849365234,
      "learning_rate": 1.8027618593793433e-05,
      "loss": 4.7518,
      "step": 5630
    },
    {
      "epoch": 0.22733144933387162,
      "grad_norm": 110.83341217041016,
      "learning_rate": 1.8026838815301313e-05,
      "loss": 5.0794,
      "step": 5631
    },
    {
      "epoch": 0.22737182075090837,
      "grad_norm": 128.82516479492188,
      "learning_rate": 1.8026058899568995e-05,
      "loss": 5.8097,
      "step": 5632
    },
    {
      "epoch": 0.2274121921679451,
      "grad_norm": 89.64065551757812,
      "learning_rate": 1.8025278846609812e-05,
      "loss": 4.5469,
      "step": 5633
    },
    {
      "epoch": 0.22745256358498184,
      "grad_norm": 116.33142852783203,
      "learning_rate": 1.8024498656437098e-05,
      "loss": 5.3146,
      "step": 5634
    },
    {
      "epoch": 0.22749293500201856,
      "grad_norm": 111.53533172607422,
      "learning_rate": 1.80237183290642e-05,
      "loss": 5.6588,
      "step": 5635
    },
    {
      "epoch": 0.2275333064190553,
      "grad_norm": 89.60541534423828,
      "learning_rate": 1.8022937864504453e-05,
      "loss": 6.1813,
      "step": 5636
    },
    {
      "epoch": 0.22757367783609206,
      "grad_norm": 125.35485076904297,
      "learning_rate": 1.8022157262771203e-05,
      "loss": 6.074,
      "step": 5637
    },
    {
      "epoch": 0.22761404925312878,
      "grad_norm": 125.15757751464844,
      "learning_rate": 1.8021376523877797e-05,
      "loss": 5.3296,
      "step": 5638
    },
    {
      "epoch": 0.22765442067016553,
      "grad_norm": 92.1509017944336,
      "learning_rate": 1.8020595647837586e-05,
      "loss": 4.0889,
      "step": 5639
    },
    {
      "epoch": 0.22769479208720225,
      "grad_norm": 107.32296752929688,
      "learning_rate": 1.8019814634663916e-05,
      "loss": 5.5437,
      "step": 5640
    },
    {
      "epoch": 0.227735163504239,
      "grad_norm": 99.15511322021484,
      "learning_rate": 1.8019033484370145e-05,
      "loss": 6.2287,
      "step": 5641
    },
    {
      "epoch": 0.22777553492127572,
      "grad_norm": 121.33049774169922,
      "learning_rate": 1.8018252196969626e-05,
      "loss": 5.2499,
      "step": 5642
    },
    {
      "epoch": 0.22781590633831247,
      "grad_norm": 367.0111083984375,
      "learning_rate": 1.801747077247572e-05,
      "loss": 5.5304,
      "step": 5643
    },
    {
      "epoch": 0.22785627775534922,
      "grad_norm": 86.4073486328125,
      "learning_rate": 1.8016689210901785e-05,
      "loss": 5.779,
      "step": 5644
    },
    {
      "epoch": 0.22789664917238595,
      "grad_norm": 90.23954010009766,
      "learning_rate": 1.8015907512261183e-05,
      "loss": 6.493,
      "step": 5645
    },
    {
      "epoch": 0.2279370205894227,
      "grad_norm": 85.21354675292969,
      "learning_rate": 1.8015125676567287e-05,
      "loss": 4.5957,
      "step": 5646
    },
    {
      "epoch": 0.22797739200645942,
      "grad_norm": 124.09247589111328,
      "learning_rate": 1.801434370383345e-05,
      "loss": 4.7681,
      "step": 5647
    },
    {
      "epoch": 0.22801776342349617,
      "grad_norm": 124.06141662597656,
      "learning_rate": 1.8013561594073056e-05,
      "loss": 5.87,
      "step": 5648
    },
    {
      "epoch": 0.2280581348405329,
      "grad_norm": 119.10993957519531,
      "learning_rate": 1.801277934729947e-05,
      "loss": 6.2029,
      "step": 5649
    },
    {
      "epoch": 0.22809850625756964,
      "grad_norm": 102.44300842285156,
      "learning_rate": 1.801199696352607e-05,
      "loss": 6.6392,
      "step": 5650
    },
    {
      "epoch": 0.2281388776746064,
      "grad_norm": 79.08507537841797,
      "learning_rate": 1.8011214442766232e-05,
      "loss": 5.239,
      "step": 5651
    },
    {
      "epoch": 0.2281792490916431,
      "grad_norm": 131.56178283691406,
      "learning_rate": 1.801043178503333e-05,
      "loss": 6.4221,
      "step": 5652
    },
    {
      "epoch": 0.22821962050867986,
      "grad_norm": 115.04396057128906,
      "learning_rate": 1.8009648990340757e-05,
      "loss": 4.5138,
      "step": 5653
    },
    {
      "epoch": 0.22825999192571658,
      "grad_norm": 100.97252655029297,
      "learning_rate": 1.8008866058701887e-05,
      "loss": 4.9367,
      "step": 5654
    },
    {
      "epoch": 0.22830036334275333,
      "grad_norm": 115.81399536132812,
      "learning_rate": 1.8008082990130106e-05,
      "loss": 5.3578,
      "step": 5655
    },
    {
      "epoch": 0.22834073475979008,
      "grad_norm": 72.20263671875,
      "learning_rate": 1.800729978463881e-05,
      "loss": 6.5249,
      "step": 5656
    },
    {
      "epoch": 0.2283811061768268,
      "grad_norm": 156.3076629638672,
      "learning_rate": 1.8006516442241385e-05,
      "loss": 4.0139,
      "step": 5657
    },
    {
      "epoch": 0.22842147759386355,
      "grad_norm": 128.90293884277344,
      "learning_rate": 1.8005732962951222e-05,
      "loss": 7.6715,
      "step": 5658
    },
    {
      "epoch": 0.22846184901090028,
      "grad_norm": 107.475830078125,
      "learning_rate": 1.8004949346781724e-05,
      "loss": 6.2078,
      "step": 5659
    },
    {
      "epoch": 0.22850222042793703,
      "grad_norm": 77.01811981201172,
      "learning_rate": 1.8004165593746288e-05,
      "loss": 6.158,
      "step": 5660
    },
    {
      "epoch": 0.22854259184497375,
      "grad_norm": 129.9298858642578,
      "learning_rate": 1.8003381703858307e-05,
      "loss": 7.5612,
      "step": 5661
    },
    {
      "epoch": 0.2285829632620105,
      "grad_norm": 130.72271728515625,
      "learning_rate": 1.8002597677131187e-05,
      "loss": 5.3782,
      "step": 5662
    },
    {
      "epoch": 0.22862333467904725,
      "grad_norm": 85.90105438232422,
      "learning_rate": 1.8001813513578336e-05,
      "loss": 5.4486,
      "step": 5663
    },
    {
      "epoch": 0.22866370609608397,
      "grad_norm": 82.72710418701172,
      "learning_rate": 1.800102921321316e-05,
      "loss": 3.7425,
      "step": 5664
    },
    {
      "epoch": 0.22870407751312072,
      "grad_norm": 121.020751953125,
      "learning_rate": 1.800024477604907e-05,
      "loss": 5.5795,
      "step": 5665
    },
    {
      "epoch": 0.22874444893015744,
      "grad_norm": 131.79933166503906,
      "learning_rate": 1.7999460202099473e-05,
      "loss": 5.5336,
      "step": 5666
    },
    {
      "epoch": 0.2287848203471942,
      "grad_norm": 75.14277648925781,
      "learning_rate": 1.7998675491377788e-05,
      "loss": 5.8638,
      "step": 5667
    },
    {
      "epoch": 0.2288251917642309,
      "grad_norm": 96.02619171142578,
      "learning_rate": 1.799789064389743e-05,
      "loss": 4.6361,
      "step": 5668
    },
    {
      "epoch": 0.22886556318126766,
      "grad_norm": 168.78460693359375,
      "learning_rate": 1.799710565967182e-05,
      "loss": 6.0836,
      "step": 5669
    },
    {
      "epoch": 0.2289059345983044,
      "grad_norm": 98.9916763305664,
      "learning_rate": 1.7996320538714375e-05,
      "loss": 5.1067,
      "step": 5670
    },
    {
      "epoch": 0.22894630601534113,
      "grad_norm": 90.70431518554688,
      "learning_rate": 1.799553528103852e-05,
      "loss": 4.8461,
      "step": 5671
    },
    {
      "epoch": 0.22898667743237788,
      "grad_norm": 166.00146484375,
      "learning_rate": 1.7994749886657687e-05,
      "loss": 5.3025,
      "step": 5672
    },
    {
      "epoch": 0.2290270488494146,
      "grad_norm": 79.59408569335938,
      "learning_rate": 1.7993964355585298e-05,
      "loss": 6.188,
      "step": 5673
    },
    {
      "epoch": 0.22906742026645135,
      "grad_norm": 94.40361022949219,
      "learning_rate": 1.7993178687834784e-05,
      "loss": 4.8024,
      "step": 5674
    },
    {
      "epoch": 0.22910779168348808,
      "grad_norm": 146.80165100097656,
      "learning_rate": 1.7992392883419584e-05,
      "loss": 3.1331,
      "step": 5675
    },
    {
      "epoch": 0.22914816310052483,
      "grad_norm": 88.24270629882812,
      "learning_rate": 1.7991606942353128e-05,
      "loss": 4.8146,
      "step": 5676
    },
    {
      "epoch": 0.22918853451756158,
      "grad_norm": 99.54888153076172,
      "learning_rate": 1.7990820864648856e-05,
      "loss": 5.7565,
      "step": 5677
    },
    {
      "epoch": 0.2292289059345983,
      "grad_norm": 73.81959533691406,
      "learning_rate": 1.7990034650320203e-05,
      "loss": 5.1753,
      "step": 5678
    },
    {
      "epoch": 0.22926927735163505,
      "grad_norm": 114.28600311279297,
      "learning_rate": 1.798924829938062e-05,
      "loss": 3.9807,
      "step": 5679
    },
    {
      "epoch": 0.22930964876867177,
      "grad_norm": 97.06834411621094,
      "learning_rate": 1.7988461811843545e-05,
      "loss": 6.2349,
      "step": 5680
    },
    {
      "epoch": 0.22935002018570852,
      "grad_norm": 115.40591430664062,
      "learning_rate": 1.7987675187722426e-05,
      "loss": 5.6623,
      "step": 5681
    },
    {
      "epoch": 0.22939039160274527,
      "grad_norm": 114.65235900878906,
      "learning_rate": 1.7986888427030714e-05,
      "loss": 5.8201,
      "step": 5682
    },
    {
      "epoch": 0.229430763019782,
      "grad_norm": 84.50605010986328,
      "learning_rate": 1.7986101529781863e-05,
      "loss": 4.3141,
      "step": 5683
    },
    {
      "epoch": 0.22947113443681874,
      "grad_norm": 147.5388641357422,
      "learning_rate": 1.7985314495989324e-05,
      "loss": 4.8438,
      "step": 5684
    },
    {
      "epoch": 0.22951150585385546,
      "grad_norm": 86.02283477783203,
      "learning_rate": 1.7984527325666554e-05,
      "loss": 5.4429,
      "step": 5685
    },
    {
      "epoch": 0.2295518772708922,
      "grad_norm": 123.45439147949219,
      "learning_rate": 1.7983740018827013e-05,
      "loss": 3.3487,
      "step": 5686
    },
    {
      "epoch": 0.22959224868792893,
      "grad_norm": 168.37815856933594,
      "learning_rate": 1.798295257548416e-05,
      "loss": 3.849,
      "step": 5687
    },
    {
      "epoch": 0.22963262010496568,
      "grad_norm": 152.39547729492188,
      "learning_rate": 1.7982164995651458e-05,
      "loss": 5.4495,
      "step": 5688
    },
    {
      "epoch": 0.22967299152200243,
      "grad_norm": 128.34915161132812,
      "learning_rate": 1.7981377279342376e-05,
      "loss": 5.5661,
      "step": 5689
    },
    {
      "epoch": 0.22971336293903916,
      "grad_norm": 101.2533950805664,
      "learning_rate": 1.798058942657038e-05,
      "loss": 4.8306,
      "step": 5690
    },
    {
      "epoch": 0.2297537343560759,
      "grad_norm": 116.19120025634766,
      "learning_rate": 1.797980143734894e-05,
      "loss": 5.2859,
      "step": 5691
    },
    {
      "epoch": 0.22979410577311263,
      "grad_norm": 131.70437622070312,
      "learning_rate": 1.797901331169153e-05,
      "loss": 6.1324,
      "step": 5692
    },
    {
      "epoch": 0.22983447719014938,
      "grad_norm": 78.8419418334961,
      "learning_rate": 1.797822504961163e-05,
      "loss": 5.5199,
      "step": 5693
    },
    {
      "epoch": 0.2298748486071861,
      "grad_norm": 111.30510711669922,
      "learning_rate": 1.7977436651122707e-05,
      "loss": 6.0003,
      "step": 5694
    },
    {
      "epoch": 0.22991522002422285,
      "grad_norm": 113.63895416259766,
      "learning_rate": 1.7976648116238245e-05,
      "loss": 4.478,
      "step": 5695
    },
    {
      "epoch": 0.2299555914412596,
      "grad_norm": 167.4565887451172,
      "learning_rate": 1.797585944497173e-05,
      "loss": 6.0949,
      "step": 5696
    },
    {
      "epoch": 0.22999596285829632,
      "grad_norm": 178.18402099609375,
      "learning_rate": 1.797507063733664e-05,
      "loss": 4.0713,
      "step": 5697
    },
    {
      "epoch": 0.23003633427533307,
      "grad_norm": 188.19873046875,
      "learning_rate": 1.7974281693346466e-05,
      "loss": 3.8817,
      "step": 5698
    },
    {
      "epoch": 0.2300767056923698,
      "grad_norm": 98.34178161621094,
      "learning_rate": 1.79734926130147e-05,
      "loss": 6.0745,
      "step": 5699
    },
    {
      "epoch": 0.23011707710940654,
      "grad_norm": 109.92655181884766,
      "learning_rate": 1.7972703396354827e-05,
      "loss": 6.0627,
      "step": 5700
    },
    {
      "epoch": 0.23015744852644326,
      "grad_norm": 90.97785186767578,
      "learning_rate": 1.797191404338034e-05,
      "loss": 4.9979,
      "step": 5701
    },
    {
      "epoch": 0.23019781994348001,
      "grad_norm": 85.80762481689453,
      "learning_rate": 1.7971124554104746e-05,
      "loss": 4.8416,
      "step": 5702
    },
    {
      "epoch": 0.23023819136051676,
      "grad_norm": 85.75232696533203,
      "learning_rate": 1.797033492854153e-05,
      "loss": 6.6715,
      "step": 5703
    },
    {
      "epoch": 0.23027856277755349,
      "grad_norm": 117.2813720703125,
      "learning_rate": 1.7969545166704203e-05,
      "loss": 4.5192,
      "step": 5704
    },
    {
      "epoch": 0.23031893419459024,
      "grad_norm": 139.52857971191406,
      "learning_rate": 1.796875526860626e-05,
      "loss": 6.5465,
      "step": 5705
    },
    {
      "epoch": 0.23035930561162696,
      "grad_norm": 101.03703308105469,
      "learning_rate": 1.7967965234261214e-05,
      "loss": 5.113,
      "step": 5706
    },
    {
      "epoch": 0.2303996770286637,
      "grad_norm": 89.0457992553711,
      "learning_rate": 1.7967175063682567e-05,
      "loss": 4.6352,
      "step": 5707
    },
    {
      "epoch": 0.23044004844570046,
      "grad_norm": 208.69808959960938,
      "learning_rate": 1.796638475688383e-05,
      "loss": 5.2914,
      "step": 5708
    },
    {
      "epoch": 0.23048041986273718,
      "grad_norm": 98.917236328125,
      "learning_rate": 1.796559431387852e-05,
      "loss": 5.002,
      "step": 5709
    },
    {
      "epoch": 0.23052079127977393,
      "grad_norm": 81.16630554199219,
      "learning_rate": 1.796480373468015e-05,
      "loss": 4.8107,
      "step": 5710
    },
    {
      "epoch": 0.23056116269681065,
      "grad_norm": 86.39619445800781,
      "learning_rate": 1.7964013019302226e-05,
      "loss": 4.9586,
      "step": 5711
    },
    {
      "epoch": 0.2306015341138474,
      "grad_norm": 171.47592163085938,
      "learning_rate": 1.796322216775828e-05,
      "loss": 5.3183,
      "step": 5712
    },
    {
      "epoch": 0.23064190553088412,
      "grad_norm": 169.7366943359375,
      "learning_rate": 1.7962431180061838e-05,
      "loss": 5.0144,
      "step": 5713
    },
    {
      "epoch": 0.23068227694792087,
      "grad_norm": 82.46159362792969,
      "learning_rate": 1.7961640056226407e-05,
      "loss": 4.553,
      "step": 5714
    },
    {
      "epoch": 0.23072264836495762,
      "grad_norm": 187.6264190673828,
      "learning_rate": 1.796084879626553e-05,
      "loss": 6.0051,
      "step": 5715
    },
    {
      "epoch": 0.23076301978199434,
      "grad_norm": 84.831298828125,
      "learning_rate": 1.7960057400192724e-05,
      "loss": 5.8044,
      "step": 5716
    },
    {
      "epoch": 0.2308033911990311,
      "grad_norm": 124.86506652832031,
      "learning_rate": 1.7959265868021526e-05,
      "loss": 3.9908,
      "step": 5717
    },
    {
      "epoch": 0.23084376261606782,
      "grad_norm": 84.85760498046875,
      "learning_rate": 1.7958474199765466e-05,
      "loss": 2.577,
      "step": 5718
    },
    {
      "epoch": 0.23088413403310457,
      "grad_norm": 156.3638153076172,
      "learning_rate": 1.7957682395438084e-05,
      "loss": 5.885,
      "step": 5719
    },
    {
      "epoch": 0.2309245054501413,
      "grad_norm": 101.59007263183594,
      "learning_rate": 1.7956890455052916e-05,
      "loss": 5.4157,
      "step": 5720
    },
    {
      "epoch": 0.23096487686717804,
      "grad_norm": 99.70230865478516,
      "learning_rate": 1.79560983786235e-05,
      "loss": 5.9894,
      "step": 5721
    },
    {
      "epoch": 0.2310052482842148,
      "grad_norm": 90.23641967773438,
      "learning_rate": 1.7955306166163382e-05,
      "loss": 5.2299,
      "step": 5722
    },
    {
      "epoch": 0.2310456197012515,
      "grad_norm": 114.52140045166016,
      "learning_rate": 1.7954513817686106e-05,
      "loss": 5.831,
      "step": 5723
    },
    {
      "epoch": 0.23108599111828826,
      "grad_norm": 91.44083404541016,
      "learning_rate": 1.7953721333205216e-05,
      "loss": 6.7171,
      "step": 5724
    },
    {
      "epoch": 0.23112636253532498,
      "grad_norm": 81.962646484375,
      "learning_rate": 1.7952928712734266e-05,
      "loss": 4.7363,
      "step": 5725
    },
    {
      "epoch": 0.23116673395236173,
      "grad_norm": 134.32369995117188,
      "learning_rate": 1.7952135956286807e-05,
      "loss": 5.0189,
      "step": 5726
    },
    {
      "epoch": 0.23120710536939845,
      "grad_norm": 111.71481323242188,
      "learning_rate": 1.795134306387639e-05,
      "loss": 5.0806,
      "step": 5727
    },
    {
      "epoch": 0.2312474767864352,
      "grad_norm": 121.80787658691406,
      "learning_rate": 1.795055003551658e-05,
      "loss": 5.7829,
      "step": 5728
    },
    {
      "epoch": 0.23128784820347195,
      "grad_norm": 124.41211700439453,
      "learning_rate": 1.7949756871220927e-05,
      "loss": 4.7264,
      "step": 5729
    },
    {
      "epoch": 0.23132821962050867,
      "grad_norm": 152.04710388183594,
      "learning_rate": 1.794896357100299e-05,
      "loss": 6.5121,
      "step": 5730
    },
    {
      "epoch": 0.23136859103754542,
      "grad_norm": 91.28468322753906,
      "learning_rate": 1.794817013487635e-05,
      "loss": 5.9255,
      "step": 5731
    },
    {
      "epoch": 0.23140896245458215,
      "grad_norm": 113.23722839355469,
      "learning_rate": 1.794737656285455e-05,
      "loss": 4.5155,
      "step": 5732
    },
    {
      "epoch": 0.2314493338716189,
      "grad_norm": 119.6435775756836,
      "learning_rate": 1.7946582854951177e-05,
      "loss": 5.4605,
      "step": 5733
    },
    {
      "epoch": 0.23148970528865564,
      "grad_norm": 87.14838409423828,
      "learning_rate": 1.794578901117979e-05,
      "loss": 6.0315,
      "step": 5734
    },
    {
      "epoch": 0.23153007670569237,
      "grad_norm": 136.86712646484375,
      "learning_rate": 1.7944995031553968e-05,
      "loss": 4.8116,
      "step": 5735
    },
    {
      "epoch": 0.23157044812272912,
      "grad_norm": 93.29798889160156,
      "learning_rate": 1.794420091608728e-05,
      "loss": 5.9512,
      "step": 5736
    },
    {
      "epoch": 0.23161081953976584,
      "grad_norm": 93.06721496582031,
      "learning_rate": 1.794340666479331e-05,
      "loss": 4.9318,
      "step": 5737
    },
    {
      "epoch": 0.2316511909568026,
      "grad_norm": 95.53129577636719,
      "learning_rate": 1.7942612277685633e-05,
      "loss": 6.3528,
      "step": 5738
    },
    {
      "epoch": 0.2316915623738393,
      "grad_norm": 74.51985931396484,
      "learning_rate": 1.7941817754777833e-05,
      "loss": 5.734,
      "step": 5739
    },
    {
      "epoch": 0.23173193379087606,
      "grad_norm": 180.6527557373047,
      "learning_rate": 1.7941023096083497e-05,
      "loss": 6.2164,
      "step": 5740
    },
    {
      "epoch": 0.2317723052079128,
      "grad_norm": 71.1929702758789,
      "learning_rate": 1.7940228301616213e-05,
      "loss": 4.5887,
      "step": 5741
    },
    {
      "epoch": 0.23181267662494953,
      "grad_norm": 96.56070709228516,
      "learning_rate": 1.793943337138956e-05,
      "loss": 5.0766,
      "step": 5742
    },
    {
      "epoch": 0.23185304804198628,
      "grad_norm": 124.26860046386719,
      "learning_rate": 1.7938638305417142e-05,
      "loss": 4.981,
      "step": 5743
    },
    {
      "epoch": 0.231893419459023,
      "grad_norm": 132.4147491455078,
      "learning_rate": 1.7937843103712544e-05,
      "loss": 7.2684,
      "step": 5744
    },
    {
      "epoch": 0.23193379087605975,
      "grad_norm": 113.25308227539062,
      "learning_rate": 1.7937047766289365e-05,
      "loss": 6.1555,
      "step": 5745
    },
    {
      "epoch": 0.23197416229309648,
      "grad_norm": 134.3466339111328,
      "learning_rate": 1.7936252293161205e-05,
      "loss": 5.9608,
      "step": 5746
    },
    {
      "epoch": 0.23201453371013323,
      "grad_norm": 170.2557830810547,
      "learning_rate": 1.7935456684341663e-05,
      "loss": 4.5836,
      "step": 5747
    },
    {
      "epoch": 0.23205490512716997,
      "grad_norm": 122.7575454711914,
      "learning_rate": 1.7934660939844343e-05,
      "loss": 5.8949,
      "step": 5748
    },
    {
      "epoch": 0.2320952765442067,
      "grad_norm": 63.255523681640625,
      "learning_rate": 1.7933865059682848e-05,
      "loss": 5.3264,
      "step": 5749
    },
    {
      "epoch": 0.23213564796124345,
      "grad_norm": 195.385009765625,
      "learning_rate": 1.793306904387079e-05,
      "loss": 5.3977,
      "step": 5750
    },
    {
      "epoch": 0.23217601937828017,
      "grad_norm": 728.7676391601562,
      "learning_rate": 1.793227289242177e-05,
      "loss": 5.7531,
      "step": 5751
    },
    {
      "epoch": 0.23221639079531692,
      "grad_norm": 139.10629272460938,
      "learning_rate": 1.7931476605349413e-05,
      "loss": 5.2762,
      "step": 5752
    },
    {
      "epoch": 0.23225676221235364,
      "grad_norm": 84.63693237304688,
      "learning_rate": 1.7930680182667327e-05,
      "loss": 5.5951,
      "step": 5753
    },
    {
      "epoch": 0.2322971336293904,
      "grad_norm": 96.60188293457031,
      "learning_rate": 1.7929883624389127e-05,
      "loss": 5.7744,
      "step": 5754
    },
    {
      "epoch": 0.23233750504642714,
      "grad_norm": 121.92710876464844,
      "learning_rate": 1.7929086930528433e-05,
      "loss": 4.5293,
      "step": 5755
    },
    {
      "epoch": 0.23237787646346386,
      "grad_norm": 96.10737609863281,
      "learning_rate": 1.7928290101098873e-05,
      "loss": 4.6825,
      "step": 5756
    },
    {
      "epoch": 0.2324182478805006,
      "grad_norm": 203.21861267089844,
      "learning_rate": 1.7927493136114066e-05,
      "loss": 5.1955,
      "step": 5757
    },
    {
      "epoch": 0.23245861929753733,
      "grad_norm": 117.3794174194336,
      "learning_rate": 1.7926696035587633e-05,
      "loss": 6.8522,
      "step": 5758
    },
    {
      "epoch": 0.23249899071457408,
      "grad_norm": 119.73981475830078,
      "learning_rate": 1.792589879953321e-05,
      "loss": 7.8802,
      "step": 5759
    },
    {
      "epoch": 0.23253936213161083,
      "grad_norm": 109.20823669433594,
      "learning_rate": 1.7925101427964428e-05,
      "loss": 5.9753,
      "step": 5760
    },
    {
      "epoch": 0.23257973354864755,
      "grad_norm": 101.34404754638672,
      "learning_rate": 1.7924303920894917e-05,
      "loss": 3.9147,
      "step": 5761
    },
    {
      "epoch": 0.2326201049656843,
      "grad_norm": 99.88555145263672,
      "learning_rate": 1.7923506278338315e-05,
      "loss": 5.6416,
      "step": 5762
    },
    {
      "epoch": 0.23266047638272103,
      "grad_norm": 109.09590911865234,
      "learning_rate": 1.7922708500308256e-05,
      "loss": 5.8442,
      "step": 5763
    },
    {
      "epoch": 0.23270084779975778,
      "grad_norm": 78.8676986694336,
      "learning_rate": 1.7921910586818384e-05,
      "loss": 6.5903,
      "step": 5764
    },
    {
      "epoch": 0.2327412192167945,
      "grad_norm": 107.61288452148438,
      "learning_rate": 1.792111253788234e-05,
      "loss": 4.96,
      "step": 5765
    },
    {
      "epoch": 0.23278159063383125,
      "grad_norm": 81.23473358154297,
      "learning_rate": 1.7920314353513768e-05,
      "loss": 5.5952,
      "step": 5766
    },
    {
      "epoch": 0.232821962050868,
      "grad_norm": 148.09335327148438,
      "learning_rate": 1.7919516033726315e-05,
      "loss": 5.5069,
      "step": 5767
    },
    {
      "epoch": 0.23286233346790472,
      "grad_norm": 72.42546081542969,
      "learning_rate": 1.7918717578533632e-05,
      "loss": 5.1599,
      "step": 5768
    },
    {
      "epoch": 0.23290270488494147,
      "grad_norm": 85.85834503173828,
      "learning_rate": 1.7917918987949373e-05,
      "loss": 5.7138,
      "step": 5769
    },
    {
      "epoch": 0.2329430763019782,
      "grad_norm": 76.20933532714844,
      "learning_rate": 1.7917120261987185e-05,
      "loss": 5.647,
      "step": 5770
    },
    {
      "epoch": 0.23298344771901494,
      "grad_norm": 95.72528839111328,
      "learning_rate": 1.791632140066073e-05,
      "loss": 4.5153,
      "step": 5771
    },
    {
      "epoch": 0.23302381913605166,
      "grad_norm": 72.08460235595703,
      "learning_rate": 1.7915522403983662e-05,
      "loss": 5.3546,
      "step": 5772
    },
    {
      "epoch": 0.2330641905530884,
      "grad_norm": 84.37144470214844,
      "learning_rate": 1.791472327196965e-05,
      "loss": 5.7143,
      "step": 5773
    },
    {
      "epoch": 0.23310456197012516,
      "grad_norm": 112.14899444580078,
      "learning_rate": 1.791392400463235e-05,
      "loss": 5.1169,
      "step": 5774
    },
    {
      "epoch": 0.23314493338716188,
      "grad_norm": 82.67969512939453,
      "learning_rate": 1.7913124601985427e-05,
      "loss": 5.1399,
      "step": 5775
    },
    {
      "epoch": 0.23318530480419863,
      "grad_norm": 79.13899230957031,
      "learning_rate": 1.7912325064042555e-05,
      "loss": 4.0886,
      "step": 5776
    },
    {
      "epoch": 0.23322567622123536,
      "grad_norm": 133.96240234375,
      "learning_rate": 1.79115253908174e-05,
      "loss": 5.5804,
      "step": 5777
    },
    {
      "epoch": 0.2332660476382721,
      "grad_norm": 82.00394439697266,
      "learning_rate": 1.7910725582323634e-05,
      "loss": 3.7937,
      "step": 5778
    },
    {
      "epoch": 0.23330641905530883,
      "grad_norm": 77.23129272460938,
      "learning_rate": 1.7909925638574938e-05,
      "loss": 4.5138,
      "step": 5779
    },
    {
      "epoch": 0.23334679047234558,
      "grad_norm": 91.87251281738281,
      "learning_rate": 1.790912555958498e-05,
      "loss": 6.9645,
      "step": 5780
    },
    {
      "epoch": 0.23338716188938233,
      "grad_norm": 90.73468780517578,
      "learning_rate": 1.7908325345367446e-05,
      "loss": 3.2569,
      "step": 5781
    },
    {
      "epoch": 0.23342753330641905,
      "grad_norm": 157.10887145996094,
      "learning_rate": 1.7907524995936017e-05,
      "loss": 6.3767,
      "step": 5782
    },
    {
      "epoch": 0.2334679047234558,
      "grad_norm": 105.22591400146484,
      "learning_rate": 1.7906724511304372e-05,
      "loss": 4.1775,
      "step": 5783
    },
    {
      "epoch": 0.23350827614049252,
      "grad_norm": 86.7423324584961,
      "learning_rate": 1.790592389148621e-05,
      "loss": 6.3374,
      "step": 5784
    },
    {
      "epoch": 0.23354864755752927,
      "grad_norm": 152.9863739013672,
      "learning_rate": 1.79051231364952e-05,
      "loss": 5.6183,
      "step": 5785
    },
    {
      "epoch": 0.23358901897456602,
      "grad_norm": 135.32095336914062,
      "learning_rate": 1.7904322246345053e-05,
      "loss": 3.7699,
      "step": 5786
    },
    {
      "epoch": 0.23362939039160274,
      "grad_norm": 164.4574737548828,
      "learning_rate": 1.790352122104945e-05,
      "loss": 5.1355,
      "step": 5787
    },
    {
      "epoch": 0.2336697618086395,
      "grad_norm": 82.06427001953125,
      "learning_rate": 1.790272006062209e-05,
      "loss": 3.8314,
      "step": 5788
    },
    {
      "epoch": 0.23371013322567621,
      "grad_norm": 104.91508483886719,
      "learning_rate": 1.790191876507667e-05,
      "loss": 4.6196,
      "step": 5789
    },
    {
      "epoch": 0.23375050464271296,
      "grad_norm": 305.035888671875,
      "learning_rate": 1.7901117334426894e-05,
      "loss": 5.673,
      "step": 5790
    },
    {
      "epoch": 0.23379087605974969,
      "grad_norm": 123.82626342773438,
      "learning_rate": 1.790031576868646e-05,
      "loss": 5.7111,
      "step": 5791
    },
    {
      "epoch": 0.23383124747678644,
      "grad_norm": 149.55201721191406,
      "learning_rate": 1.7899514067869073e-05,
      "loss": 5.0413,
      "step": 5792
    },
    {
      "epoch": 0.23387161889382319,
      "grad_norm": 123.25154876708984,
      "learning_rate": 1.7898712231988448e-05,
      "loss": 6.3553,
      "step": 5793
    },
    {
      "epoch": 0.2339119903108599,
      "grad_norm": 187.88104248046875,
      "learning_rate": 1.7897910261058286e-05,
      "loss": 5.4181,
      "step": 5794
    },
    {
      "epoch": 0.23395236172789666,
      "grad_norm": 85.20164489746094,
      "learning_rate": 1.78971081550923e-05,
      "loss": 5.0619,
      "step": 5795
    },
    {
      "epoch": 0.23399273314493338,
      "grad_norm": 129.51077270507812,
      "learning_rate": 1.7896305914104205e-05,
      "loss": 5.542,
      "step": 5796
    },
    {
      "epoch": 0.23403310456197013,
      "grad_norm": 149.200927734375,
      "learning_rate": 1.789550353810772e-05,
      "loss": 2.9388,
      "step": 5797
    },
    {
      "epoch": 0.23407347597900685,
      "grad_norm": 143.4280548095703,
      "learning_rate": 1.789470102711656e-05,
      "loss": 4.6221,
      "step": 5798
    },
    {
      "epoch": 0.2341138473960436,
      "grad_norm": 99.08723449707031,
      "learning_rate": 1.789389838114445e-05,
      "loss": 5.8216,
      "step": 5799
    },
    {
      "epoch": 0.23415421881308035,
      "grad_norm": 93.5471420288086,
      "learning_rate": 1.789309560020511e-05,
      "loss": 5.2609,
      "step": 5800
    },
    {
      "epoch": 0.23419459023011707,
      "grad_norm": 108.86380004882812,
      "learning_rate": 1.789229268431227e-05,
      "loss": 5.6282,
      "step": 5801
    },
    {
      "epoch": 0.23423496164715382,
      "grad_norm": 150.30540466308594,
      "learning_rate": 1.789148963347965e-05,
      "loss": 5.5328,
      "step": 5802
    },
    {
      "epoch": 0.23427533306419054,
      "grad_norm": 92.43550109863281,
      "learning_rate": 1.789068644772099e-05,
      "loss": 5.1588,
      "step": 5803
    },
    {
      "epoch": 0.2343157044812273,
      "grad_norm": 137.36349487304688,
      "learning_rate": 1.7889883127050016e-05,
      "loss": 4.8865,
      "step": 5804
    },
    {
      "epoch": 0.23435607589826402,
      "grad_norm": 125.85485076904297,
      "learning_rate": 1.7889079671480463e-05,
      "loss": 7.0179,
      "step": 5805
    },
    {
      "epoch": 0.23439644731530077,
      "grad_norm": 86.0741958618164,
      "learning_rate": 1.788827608102607e-05,
      "loss": 6.8455,
      "step": 5806
    },
    {
      "epoch": 0.23443681873233752,
      "grad_norm": 98.4103012084961,
      "learning_rate": 1.7887472355700578e-05,
      "loss": 5.647,
      "step": 5807
    },
    {
      "epoch": 0.23447719014937424,
      "grad_norm": 163.80210876464844,
      "learning_rate": 1.788666849551773e-05,
      "loss": 5.5556,
      "step": 5808
    },
    {
      "epoch": 0.234517561566411,
      "grad_norm": 111.62737274169922,
      "learning_rate": 1.7885864500491263e-05,
      "loss": 6.0621,
      "step": 5809
    },
    {
      "epoch": 0.2345579329834477,
      "grad_norm": 96.71305847167969,
      "learning_rate": 1.788506037063493e-05,
      "loss": 6.2561,
      "step": 5810
    },
    {
      "epoch": 0.23459830440048446,
      "grad_norm": 112.26875305175781,
      "learning_rate": 1.788425610596247e-05,
      "loss": 4.6658,
      "step": 5811
    },
    {
      "epoch": 0.2346386758175212,
      "grad_norm": 65.82135009765625,
      "learning_rate": 1.7883451706487652e-05,
      "loss": 3.8113,
      "step": 5812
    },
    {
      "epoch": 0.23467904723455793,
      "grad_norm": 97.47264099121094,
      "learning_rate": 1.7882647172224213e-05,
      "loss": 5.8503,
      "step": 5813
    },
    {
      "epoch": 0.23471941865159468,
      "grad_norm": 84.87168884277344,
      "learning_rate": 1.788184250318592e-05,
      "loss": 7.102,
      "step": 5814
    },
    {
      "epoch": 0.2347597900686314,
      "grad_norm": 114.21788024902344,
      "learning_rate": 1.788103769938652e-05,
      "loss": 6.1277,
      "step": 5815
    },
    {
      "epoch": 0.23480016148566815,
      "grad_norm": 115.73108673095703,
      "learning_rate": 1.7880232760839785e-05,
      "loss": 5.0279,
      "step": 5816
    },
    {
      "epoch": 0.23484053290270487,
      "grad_norm": 66.77824401855469,
      "learning_rate": 1.7879427687559467e-05,
      "loss": 5.6764,
      "step": 5817
    },
    {
      "epoch": 0.23488090431974162,
      "grad_norm": 142.63014221191406,
      "learning_rate": 1.7878622479559337e-05,
      "loss": 4.9023,
      "step": 5818
    },
    {
      "epoch": 0.23492127573677837,
      "grad_norm": 109.60025024414062,
      "learning_rate": 1.787781713685316e-05,
      "loss": 5.9718,
      "step": 5819
    },
    {
      "epoch": 0.2349616471538151,
      "grad_norm": 128.5666961669922,
      "learning_rate": 1.7877011659454704e-05,
      "loss": 5.002,
      "step": 5820
    },
    {
      "epoch": 0.23500201857085184,
      "grad_norm": 68.99897766113281,
      "learning_rate": 1.7876206047377748e-05,
      "loss": 5.5248,
      "step": 5821
    },
    {
      "epoch": 0.23504238998788857,
      "grad_norm": 171.58282470703125,
      "learning_rate": 1.7875400300636054e-05,
      "loss": 5.1909,
      "step": 5822
    },
    {
      "epoch": 0.23508276140492532,
      "grad_norm": 201.83969116210938,
      "learning_rate": 1.787459441924341e-05,
      "loss": 4.8011,
      "step": 5823
    },
    {
      "epoch": 0.23512313282196204,
      "grad_norm": 78.0776596069336,
      "learning_rate": 1.7873788403213587e-05,
      "loss": 5.182,
      "step": 5824
    },
    {
      "epoch": 0.2351635042389988,
      "grad_norm": 83.08199310302734,
      "learning_rate": 1.787298225256037e-05,
      "loss": 6.0678,
      "step": 5825
    },
    {
      "epoch": 0.23520387565603554,
      "grad_norm": 74.15966033935547,
      "learning_rate": 1.787217596729754e-05,
      "loss": 3.7588,
      "step": 5826
    },
    {
      "epoch": 0.23524424707307226,
      "grad_norm": 145.77809143066406,
      "learning_rate": 1.787136954743889e-05,
      "loss": 6.2107,
      "step": 5827
    },
    {
      "epoch": 0.235284618490109,
      "grad_norm": 98.49092102050781,
      "learning_rate": 1.7870562992998195e-05,
      "loss": 6.0045,
      "step": 5828
    },
    {
      "epoch": 0.23532498990714573,
      "grad_norm": 122.70569610595703,
      "learning_rate": 1.7869756303989255e-05,
      "loss": 5.7516,
      "step": 5829
    },
    {
      "epoch": 0.23536536132418248,
      "grad_norm": 111.87818908691406,
      "learning_rate": 1.7868949480425858e-05,
      "loss": 3.7578,
      "step": 5830
    },
    {
      "epoch": 0.2354057327412192,
      "grad_norm": 85.7345962524414,
      "learning_rate": 1.7868142522321804e-05,
      "loss": 5.4844,
      "step": 5831
    },
    {
      "epoch": 0.23544610415825595,
      "grad_norm": 179.86752319335938,
      "learning_rate": 1.786733542969088e-05,
      "loss": 6.0694,
      "step": 5832
    },
    {
      "epoch": 0.2354864755752927,
      "grad_norm": 144.1345977783203,
      "learning_rate": 1.7866528202546895e-05,
      "loss": 5.084,
      "step": 5833
    },
    {
      "epoch": 0.23552684699232943,
      "grad_norm": 105.15867614746094,
      "learning_rate": 1.786572084090365e-05,
      "loss": 4.8855,
      "step": 5834
    },
    {
      "epoch": 0.23556721840936617,
      "grad_norm": 68.01566314697266,
      "learning_rate": 1.786491334477494e-05,
      "loss": 4.7515,
      "step": 5835
    },
    {
      "epoch": 0.2356075898264029,
      "grad_norm": 76.92391967773438,
      "learning_rate": 1.7864105714174584e-05,
      "loss": 5.2795,
      "step": 5836
    },
    {
      "epoch": 0.23564796124343965,
      "grad_norm": 92.93621826171875,
      "learning_rate": 1.786329794911638e-05,
      "loss": 3.7379,
      "step": 5837
    },
    {
      "epoch": 0.2356883326604764,
      "grad_norm": 102.5843505859375,
      "learning_rate": 1.7862490049614145e-05,
      "loss": 5.1124,
      "step": 5838
    },
    {
      "epoch": 0.23572870407751312,
      "grad_norm": 116.29241943359375,
      "learning_rate": 1.786168201568169e-05,
      "loss": 5.9836,
      "step": 5839
    },
    {
      "epoch": 0.23576907549454987,
      "grad_norm": 110.69050598144531,
      "learning_rate": 1.786087384733283e-05,
      "loss": 5.4767,
      "step": 5840
    },
    {
      "epoch": 0.2358094469115866,
      "grad_norm": 99.27491760253906,
      "learning_rate": 1.7860065544581388e-05,
      "loss": 3.7569,
      "step": 5841
    },
    {
      "epoch": 0.23584981832862334,
      "grad_norm": 86.74093627929688,
      "learning_rate": 1.7859257107441175e-05,
      "loss": 4.1242,
      "step": 5842
    },
    {
      "epoch": 0.23589018974566006,
      "grad_norm": 143.95132446289062,
      "learning_rate": 1.785844853592602e-05,
      "loss": 5.3323,
      "step": 5843
    },
    {
      "epoch": 0.2359305611626968,
      "grad_norm": 90.58242797851562,
      "learning_rate": 1.7857639830049747e-05,
      "loss": 7.1946,
      "step": 5844
    },
    {
      "epoch": 0.23597093257973356,
      "grad_norm": 99.04721069335938,
      "learning_rate": 1.785683098982618e-05,
      "loss": 6.3304,
      "step": 5845
    },
    {
      "epoch": 0.23601130399677028,
      "grad_norm": 116.48577117919922,
      "learning_rate": 1.7856022015269148e-05,
      "loss": 4.696,
      "step": 5846
    },
    {
      "epoch": 0.23605167541380703,
      "grad_norm": 71.56526947021484,
      "learning_rate": 1.785521290639249e-05,
      "loss": 6.2571,
      "step": 5847
    },
    {
      "epoch": 0.23609204683084375,
      "grad_norm": 168.0645751953125,
      "learning_rate": 1.7854403663210028e-05,
      "loss": 5.0973,
      "step": 5848
    },
    {
      "epoch": 0.2361324182478805,
      "grad_norm": 168.20346069335938,
      "learning_rate": 1.785359428573561e-05,
      "loss": 6.7968,
      "step": 5849
    },
    {
      "epoch": 0.23617278966491723,
      "grad_norm": 176.41961669921875,
      "learning_rate": 1.7852784773983067e-05,
      "loss": 6.593,
      "step": 5850
    },
    {
      "epoch": 0.23621316108195398,
      "grad_norm": 117.98722839355469,
      "learning_rate": 1.785197512796624e-05,
      "loss": 5.6032,
      "step": 5851
    },
    {
      "epoch": 0.23625353249899073,
      "grad_norm": 77.25040435791016,
      "learning_rate": 1.785116534769898e-05,
      "loss": 5.2123,
      "step": 5852
    },
    {
      "epoch": 0.23629390391602745,
      "grad_norm": 81.09008026123047,
      "learning_rate": 1.785035543319512e-05,
      "loss": 6.4921,
      "step": 5853
    },
    {
      "epoch": 0.2363342753330642,
      "grad_norm": 157.25192260742188,
      "learning_rate": 1.7849545384468518e-05,
      "loss": 5.6837,
      "step": 5854
    },
    {
      "epoch": 0.23637464675010092,
      "grad_norm": 115.4410629272461,
      "learning_rate": 1.7848735201533018e-05,
      "loss": 4.3866,
      "step": 5855
    },
    {
      "epoch": 0.23641501816713767,
      "grad_norm": 149.16586303710938,
      "learning_rate": 1.7847924884402474e-05,
      "loss": 5.1409,
      "step": 5856
    },
    {
      "epoch": 0.2364553895841744,
      "grad_norm": 120.7906265258789,
      "learning_rate": 1.7847114433090743e-05,
      "loss": 6.6503,
      "step": 5857
    },
    {
      "epoch": 0.23649576100121114,
      "grad_norm": 111.84234619140625,
      "learning_rate": 1.7846303847611678e-05,
      "loss": 5.6107,
      "step": 5858
    },
    {
      "epoch": 0.2365361324182479,
      "grad_norm": 107.4189453125,
      "learning_rate": 1.784549312797914e-05,
      "loss": 4.7508,
      "step": 5859
    },
    {
      "epoch": 0.2365765038352846,
      "grad_norm": 112.1483154296875,
      "learning_rate": 1.7844682274206987e-05,
      "loss": 5.3198,
      "step": 5860
    },
    {
      "epoch": 0.23661687525232136,
      "grad_norm": 156.68429565429688,
      "learning_rate": 1.784387128630909e-05,
      "loss": 5.0707,
      "step": 5861
    },
    {
      "epoch": 0.23665724666935808,
      "grad_norm": 102.7824478149414,
      "learning_rate": 1.7843060164299312e-05,
      "loss": 6.8526,
      "step": 5862
    },
    {
      "epoch": 0.23669761808639483,
      "grad_norm": 64.65583038330078,
      "learning_rate": 1.7842248908191516e-05,
      "loss": 6.1126,
      "step": 5863
    },
    {
      "epoch": 0.23673798950343158,
      "grad_norm": 96.18231201171875,
      "learning_rate": 1.7841437517999576e-05,
      "loss": 7.4483,
      "step": 5864
    },
    {
      "epoch": 0.2367783609204683,
      "grad_norm": 80.1888198852539,
      "learning_rate": 1.784062599373737e-05,
      "loss": 5.2553,
      "step": 5865
    },
    {
      "epoch": 0.23681873233750506,
      "grad_norm": 174.9115447998047,
      "learning_rate": 1.7839814335418765e-05,
      "loss": 4.6431,
      "step": 5866
    },
    {
      "epoch": 0.23685910375454178,
      "grad_norm": 82.74188995361328,
      "learning_rate": 1.7839002543057643e-05,
      "loss": 3.6718,
      "step": 5867
    },
    {
      "epoch": 0.23689947517157853,
      "grad_norm": 117.0372543334961,
      "learning_rate": 1.7838190616667884e-05,
      "loss": 5.6322,
      "step": 5868
    },
    {
      "epoch": 0.23693984658861525,
      "grad_norm": 97.1174545288086,
      "learning_rate": 1.7837378556263368e-05,
      "loss": 5.7676,
      "step": 5869
    },
    {
      "epoch": 0.236980218005652,
      "grad_norm": 104.53492736816406,
      "learning_rate": 1.7836566361857982e-05,
      "loss": 3.8681,
      "step": 5870
    },
    {
      "epoch": 0.23702058942268875,
      "grad_norm": 121.4781265258789,
      "learning_rate": 1.7835754033465608e-05,
      "loss": 5.1636,
      "step": 5871
    },
    {
      "epoch": 0.23706096083972547,
      "grad_norm": 85.83499145507812,
      "learning_rate": 1.783494157110014e-05,
      "loss": 6.1727,
      "step": 5872
    },
    {
      "epoch": 0.23710133225676222,
      "grad_norm": 150.25033569335938,
      "learning_rate": 1.7834128974775466e-05,
      "loss": 5.2736,
      "step": 5873
    },
    {
      "epoch": 0.23714170367379894,
      "grad_norm": 130.07302856445312,
      "learning_rate": 1.7833316244505477e-05,
      "loss": 4.3428,
      "step": 5874
    },
    {
      "epoch": 0.2371820750908357,
      "grad_norm": 103.29615783691406,
      "learning_rate": 1.783250338030408e-05,
      "loss": 5.2313,
      "step": 5875
    },
    {
      "epoch": 0.23722244650787241,
      "grad_norm": 100.09869384765625,
      "learning_rate": 1.7831690382185162e-05,
      "loss": 4.7013,
      "step": 5876
    },
    {
      "epoch": 0.23726281792490916,
      "grad_norm": 94.03850555419922,
      "learning_rate": 1.7830877250162625e-05,
      "loss": 4.4804,
      "step": 5877
    },
    {
      "epoch": 0.2373031893419459,
      "grad_norm": 101.8338851928711,
      "learning_rate": 1.7830063984250373e-05,
      "loss": 6.924,
      "step": 5878
    },
    {
      "epoch": 0.23734356075898264,
      "grad_norm": 94.169677734375,
      "learning_rate": 1.7829250584462314e-05,
      "loss": 5.3026,
      "step": 5879
    },
    {
      "epoch": 0.23738393217601939,
      "grad_norm": 81.42330932617188,
      "learning_rate": 1.7828437050812353e-05,
      "loss": 6.951,
      "step": 5880
    },
    {
      "epoch": 0.2374243035930561,
      "grad_norm": 116.28739929199219,
      "learning_rate": 1.7827623383314397e-05,
      "loss": 7.3545,
      "step": 5881
    },
    {
      "epoch": 0.23746467501009286,
      "grad_norm": 100.78177642822266,
      "learning_rate": 1.782680958198236e-05,
      "loss": 6.5361,
      "step": 5882
    },
    {
      "epoch": 0.23750504642712958,
      "grad_norm": 114.395751953125,
      "learning_rate": 1.782599564683016e-05,
      "loss": 4.3634,
      "step": 5883
    },
    {
      "epoch": 0.23754541784416633,
      "grad_norm": 101.20281219482422,
      "learning_rate": 1.7825181577871706e-05,
      "loss": 4.1055,
      "step": 5884
    },
    {
      "epoch": 0.23758578926120308,
      "grad_norm": 203.2952117919922,
      "learning_rate": 1.782436737512092e-05,
      "loss": 6.4208,
      "step": 5885
    },
    {
      "epoch": 0.2376261606782398,
      "grad_norm": 166.74359130859375,
      "learning_rate": 1.782355303859172e-05,
      "loss": 6.3845,
      "step": 5886
    },
    {
      "epoch": 0.23766653209527655,
      "grad_norm": 106.058837890625,
      "learning_rate": 1.782273856829804e-05,
      "loss": 5.2192,
      "step": 5887
    },
    {
      "epoch": 0.23770690351231327,
      "grad_norm": 130.6007080078125,
      "learning_rate": 1.7821923964253792e-05,
      "loss": 4.944,
      "step": 5888
    },
    {
      "epoch": 0.23774727492935002,
      "grad_norm": 105.00886535644531,
      "learning_rate": 1.782110922647291e-05,
      "loss": 6.957,
      "step": 5889
    },
    {
      "epoch": 0.23778764634638677,
      "grad_norm": 69.10576629638672,
      "learning_rate": 1.7820294354969326e-05,
      "loss": 6.1491,
      "step": 5890
    },
    {
      "epoch": 0.2378280177634235,
      "grad_norm": 99.26202392578125,
      "learning_rate": 1.781947934975697e-05,
      "loss": 4.3133,
      "step": 5891
    },
    {
      "epoch": 0.23786838918046024,
      "grad_norm": 86.78356170654297,
      "learning_rate": 1.7818664210849778e-05,
      "loss": 4.9723,
      "step": 5892
    },
    {
      "epoch": 0.23790876059749697,
      "grad_norm": 104.09534454345703,
      "learning_rate": 1.7817848938261684e-05,
      "loss": 6.6003,
      "step": 5893
    },
    {
      "epoch": 0.23794913201453372,
      "grad_norm": 71.85230255126953,
      "learning_rate": 1.7817033532006628e-05,
      "loss": 5.1486,
      "step": 5894
    },
    {
      "epoch": 0.23798950343157044,
      "grad_norm": 63.6766242980957,
      "learning_rate": 1.7816217992098555e-05,
      "loss": 5.0981,
      "step": 5895
    },
    {
      "epoch": 0.2380298748486072,
      "grad_norm": 221.47364807128906,
      "learning_rate": 1.7815402318551406e-05,
      "loss": 3.6867,
      "step": 5896
    },
    {
      "epoch": 0.23807024626564394,
      "grad_norm": 85.10433959960938,
      "learning_rate": 1.781458651137913e-05,
      "loss": 4.3546,
      "step": 5897
    },
    {
      "epoch": 0.23811061768268066,
      "grad_norm": 96.60132598876953,
      "learning_rate": 1.7813770570595668e-05,
      "loss": 4.9112,
      "step": 5898
    },
    {
      "epoch": 0.2381509890997174,
      "grad_norm": 154.9127197265625,
      "learning_rate": 1.7812954496214975e-05,
      "loss": 5.6668,
      "step": 5899
    },
    {
      "epoch": 0.23819136051675413,
      "grad_norm": 84.89239501953125,
      "learning_rate": 1.781213828825101e-05,
      "loss": 5.4716,
      "step": 5900
    },
    {
      "epoch": 0.23823173193379088,
      "grad_norm": 115.1965103149414,
      "learning_rate": 1.781132194671772e-05,
      "loss": 6.5985,
      "step": 5901
    },
    {
      "epoch": 0.2382721033508276,
      "grad_norm": 109.35415649414062,
      "learning_rate": 1.781050547162907e-05,
      "loss": 5.4191,
      "step": 5902
    },
    {
      "epoch": 0.23831247476786435,
      "grad_norm": 90.14212036132812,
      "learning_rate": 1.7809688862999007e-05,
      "loss": 7.2245,
      "step": 5903
    },
    {
      "epoch": 0.2383528461849011,
      "grad_norm": 100.14058685302734,
      "learning_rate": 1.7808872120841506e-05,
      "loss": 5.9217,
      "step": 5904
    },
    {
      "epoch": 0.23839321760193782,
      "grad_norm": 76.90837097167969,
      "learning_rate": 1.780805524517053e-05,
      "loss": 5.8352,
      "step": 5905
    },
    {
      "epoch": 0.23843358901897457,
      "grad_norm": 104.5289077758789,
      "learning_rate": 1.7807238236000037e-05,
      "loss": 4.9272,
      "step": 5906
    },
    {
      "epoch": 0.2384739604360113,
      "grad_norm": 65.91993713378906,
      "learning_rate": 1.7806421093344003e-05,
      "loss": 5.8089,
      "step": 5907
    },
    {
      "epoch": 0.23851433185304804,
      "grad_norm": 139.99801635742188,
      "learning_rate": 1.7805603817216397e-05,
      "loss": 5.865,
      "step": 5908
    },
    {
      "epoch": 0.23855470327008477,
      "grad_norm": 90.89962768554688,
      "learning_rate": 1.7804786407631195e-05,
      "loss": 5.7954,
      "step": 5909
    },
    {
      "epoch": 0.23859507468712152,
      "grad_norm": 79.38571166992188,
      "learning_rate": 1.780396886460237e-05,
      "loss": 5.4301,
      "step": 5910
    },
    {
      "epoch": 0.23863544610415827,
      "grad_norm": 91.54914855957031,
      "learning_rate": 1.7803151188143902e-05,
      "loss": 4.7111,
      "step": 5911
    },
    {
      "epoch": 0.238675817521195,
      "grad_norm": 105.78622436523438,
      "learning_rate": 1.780233337826977e-05,
      "loss": 5.3099,
      "step": 5912
    },
    {
      "epoch": 0.23871618893823174,
      "grad_norm": 122.58827209472656,
      "learning_rate": 1.7801515434993953e-05,
      "loss": 4.2978,
      "step": 5913
    },
    {
      "epoch": 0.23875656035526846,
      "grad_norm": 85.86949920654297,
      "learning_rate": 1.7800697358330447e-05,
      "loss": 5.8908,
      "step": 5914
    },
    {
      "epoch": 0.2387969317723052,
      "grad_norm": 99.67823028564453,
      "learning_rate": 1.7799879148293227e-05,
      "loss": 5.573,
      "step": 5915
    },
    {
      "epoch": 0.23883730318934196,
      "grad_norm": 105.11700439453125,
      "learning_rate": 1.7799060804896288e-05,
      "loss": 6.7781,
      "step": 5916
    },
    {
      "epoch": 0.23887767460637868,
      "grad_norm": 165.7634735107422,
      "learning_rate": 1.779824232815362e-05,
      "loss": 5.384,
      "step": 5917
    },
    {
      "epoch": 0.23891804602341543,
      "grad_norm": 95.22285461425781,
      "learning_rate": 1.7797423718079225e-05,
      "loss": 6.3039,
      "step": 5918
    },
    {
      "epoch": 0.23895841744045215,
      "grad_norm": 80.29763793945312,
      "learning_rate": 1.779660497468709e-05,
      "loss": 5.8587,
      "step": 5919
    },
    {
      "epoch": 0.2389987888574889,
      "grad_norm": 128.9203643798828,
      "learning_rate": 1.7795786097991214e-05,
      "loss": 5.6959,
      "step": 5920
    },
    {
      "epoch": 0.23903916027452562,
      "grad_norm": 107.01722717285156,
      "learning_rate": 1.77949670880056e-05,
      "loss": 3.7566,
      "step": 5921
    },
    {
      "epoch": 0.23907953169156237,
      "grad_norm": 77.2967758178711,
      "learning_rate": 1.779414794474425e-05,
      "loss": 5.8576,
      "step": 5922
    },
    {
      "epoch": 0.23911990310859912,
      "grad_norm": 88.23291015625,
      "learning_rate": 1.7793328668221176e-05,
      "loss": 5.6564,
      "step": 5923
    },
    {
      "epoch": 0.23916027452563585,
      "grad_norm": 145.1075897216797,
      "learning_rate": 1.7792509258450373e-05,
      "loss": 5.6735,
      "step": 5924
    },
    {
      "epoch": 0.2392006459426726,
      "grad_norm": 129.75389099121094,
      "learning_rate": 1.7791689715445865e-05,
      "loss": 4.7995,
      "step": 5925
    },
    {
      "epoch": 0.23924101735970932,
      "grad_norm": 113.92557525634766,
      "learning_rate": 1.7790870039221658e-05,
      "loss": 4.9111,
      "step": 5926
    },
    {
      "epoch": 0.23928138877674607,
      "grad_norm": 128.85130310058594,
      "learning_rate": 1.7790050229791762e-05,
      "loss": 6.3315,
      "step": 5927
    },
    {
      "epoch": 0.2393217601937828,
      "grad_norm": 130.8972930908203,
      "learning_rate": 1.77892302871702e-05,
      "loss": 6.7297,
      "step": 5928
    },
    {
      "epoch": 0.23936213161081954,
      "grad_norm": 104.3133544921875,
      "learning_rate": 1.778841021137099e-05,
      "loss": 5.3677,
      "step": 5929
    },
    {
      "epoch": 0.2394025030278563,
      "grad_norm": 123.74713897705078,
      "learning_rate": 1.7787590002408154e-05,
      "loss": 5.2215,
      "step": 5930
    },
    {
      "epoch": 0.239442874444893,
      "grad_norm": 97.66261291503906,
      "learning_rate": 1.778676966029571e-05,
      "loss": 5.3826,
      "step": 5931
    },
    {
      "epoch": 0.23948324586192976,
      "grad_norm": 146.88983154296875,
      "learning_rate": 1.7785949185047692e-05,
      "loss": 4.2508,
      "step": 5932
    },
    {
      "epoch": 0.23952361727896648,
      "grad_norm": 77.89947509765625,
      "learning_rate": 1.7785128576678123e-05,
      "loss": 3.9961,
      "step": 5933
    },
    {
      "epoch": 0.23956398869600323,
      "grad_norm": 130.0517578125,
      "learning_rate": 1.7784307835201036e-05,
      "loss": 4.8011,
      "step": 5934
    },
    {
      "epoch": 0.23960436011303995,
      "grad_norm": 139.40383911132812,
      "learning_rate": 1.7783486960630462e-05,
      "loss": 4.0446,
      "step": 5935
    },
    {
      "epoch": 0.2396447315300767,
      "grad_norm": 121.48936462402344,
      "learning_rate": 1.7782665952980437e-05,
      "loss": 5.8213,
      "step": 5936
    },
    {
      "epoch": 0.23968510294711345,
      "grad_norm": 189.64434814453125,
      "learning_rate": 1.7781844812264996e-05,
      "loss": 6.0828,
      "step": 5937
    },
    {
      "epoch": 0.23972547436415018,
      "grad_norm": 78.82970428466797,
      "learning_rate": 1.7781023538498182e-05,
      "loss": 5.5643,
      "step": 5938
    },
    {
      "epoch": 0.23976584578118693,
      "grad_norm": 88.9985122680664,
      "learning_rate": 1.7780202131694035e-05,
      "loss": 5.2801,
      "step": 5939
    },
    {
      "epoch": 0.23980621719822365,
      "grad_norm": 97.86265563964844,
      "learning_rate": 1.77793805918666e-05,
      "loss": 5.5748,
      "step": 5940
    },
    {
      "epoch": 0.2398465886152604,
      "grad_norm": 140.58607482910156,
      "learning_rate": 1.7778558919029926e-05,
      "loss": 5.15,
      "step": 5941
    },
    {
      "epoch": 0.23988696003229715,
      "grad_norm": 155.29595947265625,
      "learning_rate": 1.7777737113198054e-05,
      "loss": 4.2618,
      "step": 5942
    },
    {
      "epoch": 0.23992733144933387,
      "grad_norm": 121.82146453857422,
      "learning_rate": 1.777691517438504e-05,
      "loss": 5.404,
      "step": 5943
    },
    {
      "epoch": 0.23996770286637062,
      "grad_norm": 84.41773986816406,
      "learning_rate": 1.777609310260494e-05,
      "loss": 5.6081,
      "step": 5944
    },
    {
      "epoch": 0.24000807428340734,
      "grad_norm": 102.61873626708984,
      "learning_rate": 1.7775270897871807e-05,
      "loss": 4.7967,
      "step": 5945
    },
    {
      "epoch": 0.2400484457004441,
      "grad_norm": 203.48692321777344,
      "learning_rate": 1.7774448560199695e-05,
      "loss": 5.6524,
      "step": 5946
    },
    {
      "epoch": 0.2400888171174808,
      "grad_norm": 89.69503784179688,
      "learning_rate": 1.777362608960267e-05,
      "loss": 4.2579,
      "step": 5947
    },
    {
      "epoch": 0.24012918853451756,
      "grad_norm": 113.927001953125,
      "learning_rate": 1.777280348609479e-05,
      "loss": 6.4462,
      "step": 5948
    },
    {
      "epoch": 0.2401695599515543,
      "grad_norm": 265.9789733886719,
      "learning_rate": 1.777198074969012e-05,
      "loss": 5.8506,
      "step": 5949
    },
    {
      "epoch": 0.24020993136859103,
      "grad_norm": 82.45149230957031,
      "learning_rate": 1.777115788040273e-05,
      "loss": 5.2867,
      "step": 5950
    },
    {
      "epoch": 0.24025030278562778,
      "grad_norm": 122.41938018798828,
      "learning_rate": 1.7770334878246685e-05,
      "loss": 5.6466,
      "step": 5951
    },
    {
      "epoch": 0.2402906742026645,
      "grad_norm": 74.53356170654297,
      "learning_rate": 1.776951174323606e-05,
      "loss": 6.6565,
      "step": 5952
    },
    {
      "epoch": 0.24033104561970126,
      "grad_norm": 77.934326171875,
      "learning_rate": 1.7768688475384928e-05,
      "loss": 5.6138,
      "step": 5953
    },
    {
      "epoch": 0.24037141703673798,
      "grad_norm": 102.03507995605469,
      "learning_rate": 1.7767865074707365e-05,
      "loss": 6.1537,
      "step": 5954
    },
    {
      "epoch": 0.24041178845377473,
      "grad_norm": 80.7371597290039,
      "learning_rate": 1.7767041541217448e-05,
      "loss": 5.8415,
      "step": 5955
    },
    {
      "epoch": 0.24045215987081148,
      "grad_norm": 226.57901000976562,
      "learning_rate": 1.776621787492926e-05,
      "loss": 6.1047,
      "step": 5956
    },
    {
      "epoch": 0.2404925312878482,
      "grad_norm": 109.54818725585938,
      "learning_rate": 1.7765394075856874e-05,
      "loss": 6.5037,
      "step": 5957
    },
    {
      "epoch": 0.24053290270488495,
      "grad_norm": 87.7962875366211,
      "learning_rate": 1.776457014401439e-05,
      "loss": 4.9404,
      "step": 5958
    },
    {
      "epoch": 0.24057327412192167,
      "grad_norm": 123.05998229980469,
      "learning_rate": 1.7763746079415885e-05,
      "loss": 4.9226,
      "step": 5959
    },
    {
      "epoch": 0.24061364553895842,
      "grad_norm": 75.546630859375,
      "learning_rate": 1.7762921882075452e-05,
      "loss": 5.7193,
      "step": 5960
    },
    {
      "epoch": 0.24065401695599514,
      "grad_norm": 119.28202819824219,
      "learning_rate": 1.7762097552007183e-05,
      "loss": 5.8983,
      "step": 5961
    },
    {
      "epoch": 0.2406943883730319,
      "grad_norm": 116.26841735839844,
      "learning_rate": 1.776127308922517e-05,
      "loss": 5.1842,
      "step": 5962
    },
    {
      "epoch": 0.24073475979006864,
      "grad_norm": 223.44894409179688,
      "learning_rate": 1.7760448493743516e-05,
      "loss": 5.8287,
      "step": 5963
    },
    {
      "epoch": 0.24077513120710536,
      "grad_norm": 154.79620361328125,
      "learning_rate": 1.775962376557631e-05,
      "loss": 6.5757,
      "step": 5964
    },
    {
      "epoch": 0.2408155026241421,
      "grad_norm": 109.25288391113281,
      "learning_rate": 1.775879890473766e-05,
      "loss": 5.06,
      "step": 5965
    },
    {
      "epoch": 0.24085587404117884,
      "grad_norm": 106.20862579345703,
      "learning_rate": 1.7757973911241664e-05,
      "loss": 5.6258,
      "step": 5966
    },
    {
      "epoch": 0.24089624545821559,
      "grad_norm": 111.44542694091797,
      "learning_rate": 1.7757148785102433e-05,
      "loss": 7.0245,
      "step": 5967
    },
    {
      "epoch": 0.24093661687525234,
      "grad_norm": 101.22270965576172,
      "learning_rate": 1.775632352633407e-05,
      "loss": 5.108,
      "step": 5968
    },
    {
      "epoch": 0.24097698829228906,
      "grad_norm": 120.90257263183594,
      "learning_rate": 1.775549813495069e-05,
      "loss": 5.7393,
      "step": 5969
    },
    {
      "epoch": 0.2410173597093258,
      "grad_norm": 77.40303802490234,
      "learning_rate": 1.77546726109664e-05,
      "loss": 6.4803,
      "step": 5970
    },
    {
      "epoch": 0.24105773112636253,
      "grad_norm": 72.21610260009766,
      "learning_rate": 1.7753846954395317e-05,
      "loss": 6.5275,
      "step": 5971
    },
    {
      "epoch": 0.24109810254339928,
      "grad_norm": 80.15948486328125,
      "learning_rate": 1.775302116525156e-05,
      "loss": 5.7303,
      "step": 5972
    },
    {
      "epoch": 0.241138473960436,
      "grad_norm": 103.86457061767578,
      "learning_rate": 1.7752195243549242e-05,
      "loss": 4.996,
      "step": 5973
    },
    {
      "epoch": 0.24117884537747275,
      "grad_norm": 83.04635620117188,
      "learning_rate": 1.775136918930249e-05,
      "loss": 6.0193,
      "step": 5974
    },
    {
      "epoch": 0.2412192167945095,
      "grad_norm": 70.82820892333984,
      "learning_rate": 1.7750543002525425e-05,
      "loss": 5.1356,
      "step": 5975
    },
    {
      "epoch": 0.24125958821154622,
      "grad_norm": 87.19628143310547,
      "learning_rate": 1.7749716683232175e-05,
      "loss": 4.7669,
      "step": 5976
    },
    {
      "epoch": 0.24129995962858297,
      "grad_norm": 67.24391174316406,
      "learning_rate": 1.7748890231436867e-05,
      "loss": 5.8666,
      "step": 5977
    },
    {
      "epoch": 0.2413403310456197,
      "grad_norm": 89.5867691040039,
      "learning_rate": 1.774806364715363e-05,
      "loss": 4.1875,
      "step": 5978
    },
    {
      "epoch": 0.24138070246265644,
      "grad_norm": 138.05809020996094,
      "learning_rate": 1.7747236930396595e-05,
      "loss": 5.6728,
      "step": 5979
    },
    {
      "epoch": 0.24142107387969317,
      "grad_norm": 95.1600570678711,
      "learning_rate": 1.7746410081179903e-05,
      "loss": 4.9268,
      "step": 5980
    },
    {
      "epoch": 0.24146144529672992,
      "grad_norm": 99.0817642211914,
      "learning_rate": 1.7745583099517686e-05,
      "loss": 6.5327,
      "step": 5981
    },
    {
      "epoch": 0.24150181671376666,
      "grad_norm": 85.55270385742188,
      "learning_rate": 1.7744755985424084e-05,
      "loss": 7.2118,
      "step": 5982
    },
    {
      "epoch": 0.2415421881308034,
      "grad_norm": 73.35932159423828,
      "learning_rate": 1.774392873891324e-05,
      "loss": 5.9037,
      "step": 5983
    },
    {
      "epoch": 0.24158255954784014,
      "grad_norm": 86.7330093383789,
      "learning_rate": 1.77431013599993e-05,
      "loss": 5.7086,
      "step": 5984
    },
    {
      "epoch": 0.24162293096487686,
      "grad_norm": 87.96886444091797,
      "learning_rate": 1.774227384869641e-05,
      "loss": 5.1586,
      "step": 5985
    },
    {
      "epoch": 0.2416633023819136,
      "grad_norm": 107.66802215576172,
      "learning_rate": 1.7741446205018716e-05,
      "loss": 6.9154,
      "step": 5986
    },
    {
      "epoch": 0.24170367379895033,
      "grad_norm": 102.99546813964844,
      "learning_rate": 1.7740618428980365e-05,
      "loss": 6.2465,
      "step": 5987
    },
    {
      "epoch": 0.24174404521598708,
      "grad_norm": 98.30487823486328,
      "learning_rate": 1.7739790520595518e-05,
      "loss": 3.8461,
      "step": 5988
    },
    {
      "epoch": 0.24178441663302383,
      "grad_norm": 82.52546691894531,
      "learning_rate": 1.7738962479878327e-05,
      "loss": 5.5378,
      "step": 5989
    },
    {
      "epoch": 0.24182478805006055,
      "grad_norm": 81.04200744628906,
      "learning_rate": 1.773813430684295e-05,
      "loss": 5.7468,
      "step": 5990
    },
    {
      "epoch": 0.2418651594670973,
      "grad_norm": 113.02803802490234,
      "learning_rate": 1.7737306001503545e-05,
      "loss": 4.2761,
      "step": 5991
    },
    {
      "epoch": 0.24190553088413402,
      "grad_norm": 69.661865234375,
      "learning_rate": 1.7736477563874274e-05,
      "loss": 4.9663,
      "step": 5992
    },
    {
      "epoch": 0.24194590230117077,
      "grad_norm": 62.90067672729492,
      "learning_rate": 1.7735648993969302e-05,
      "loss": 5.1305,
      "step": 5993
    },
    {
      "epoch": 0.2419862737182075,
      "grad_norm": 81.64987182617188,
      "learning_rate": 1.7734820291802803e-05,
      "loss": 5.5736,
      "step": 5994
    },
    {
      "epoch": 0.24202664513524424,
      "grad_norm": 97.1672592163086,
      "learning_rate": 1.773399145738893e-05,
      "loss": 6.0707,
      "step": 5995
    },
    {
      "epoch": 0.242067016552281,
      "grad_norm": 139.3844451904297,
      "learning_rate": 1.7733162490741868e-05,
      "loss": 4.8614,
      "step": 5996
    },
    {
      "epoch": 0.24210738796931772,
      "grad_norm": 87.07423400878906,
      "learning_rate": 1.7732333391875785e-05,
      "loss": 6.2548,
      "step": 5997
    },
    {
      "epoch": 0.24214775938635447,
      "grad_norm": 73.55404663085938,
      "learning_rate": 1.7731504160804857e-05,
      "loss": 5.637,
      "step": 5998
    },
    {
      "epoch": 0.2421881308033912,
      "grad_norm": 131.72091674804688,
      "learning_rate": 1.7730674797543264e-05,
      "loss": 4.9836,
      "step": 5999
    },
    {
      "epoch": 0.24222850222042794,
      "grad_norm": 94.4135513305664,
      "learning_rate": 1.772984530210518e-05,
      "loss": 5.72,
      "step": 6000
    },
    {
      "epoch": 0.2422688736374647,
      "grad_norm": 120.27977752685547,
      "learning_rate": 1.7729015674504798e-05,
      "loss": 6.3021,
      "step": 6001
    },
    {
      "epoch": 0.2423092450545014,
      "grad_norm": 96.89191436767578,
      "learning_rate": 1.772818591475629e-05,
      "loss": 5.2298,
      "step": 6002
    },
    {
      "epoch": 0.24234961647153816,
      "grad_norm": 107.34725952148438,
      "learning_rate": 1.7727356022873852e-05,
      "loss": 4.9705,
      "step": 6003
    },
    {
      "epoch": 0.24238998788857488,
      "grad_norm": 112.86279296875,
      "learning_rate": 1.7726525998871676e-05,
      "loss": 5.0596,
      "step": 6004
    },
    {
      "epoch": 0.24243035930561163,
      "grad_norm": 85.18235778808594,
      "learning_rate": 1.772569584276394e-05,
      "loss": 6.2959,
      "step": 6005
    },
    {
      "epoch": 0.24247073072264835,
      "grad_norm": 114.42411804199219,
      "learning_rate": 1.7724865554564852e-05,
      "loss": 5.2135,
      "step": 6006
    },
    {
      "epoch": 0.2425111021396851,
      "grad_norm": 99.69255065917969,
      "learning_rate": 1.7724035134288597e-05,
      "loss": 5.1739,
      "step": 6007
    },
    {
      "epoch": 0.24255147355672185,
      "grad_norm": 90.63612365722656,
      "learning_rate": 1.7723204581949383e-05,
      "loss": 6.1412,
      "step": 6008
    },
    {
      "epoch": 0.24259184497375857,
      "grad_norm": 103.37932586669922,
      "learning_rate": 1.77223738975614e-05,
      "loss": 5.1453,
      "step": 6009
    },
    {
      "epoch": 0.24263221639079532,
      "grad_norm": 88.92766571044922,
      "learning_rate": 1.7721543081138862e-05,
      "loss": 6.2599,
      "step": 6010
    },
    {
      "epoch": 0.24267258780783205,
      "grad_norm": 126.55681610107422,
      "learning_rate": 1.7720712132695962e-05,
      "loss": 5.6616,
      "step": 6011
    },
    {
      "epoch": 0.2427129592248688,
      "grad_norm": 116.23165130615234,
      "learning_rate": 1.771988105224692e-05,
      "loss": 4.1736,
      "step": 6012
    },
    {
      "epoch": 0.24275333064190552,
      "grad_norm": 84.94568634033203,
      "learning_rate": 1.771904983980594e-05,
      "loss": 4.6378,
      "step": 6013
    },
    {
      "epoch": 0.24279370205894227,
      "grad_norm": 129.2796630859375,
      "learning_rate": 1.7718218495387228e-05,
      "loss": 3.7428,
      "step": 6014
    },
    {
      "epoch": 0.24283407347597902,
      "grad_norm": 249.04417419433594,
      "learning_rate": 1.7717387019005002e-05,
      "loss": 5.5604,
      "step": 6015
    },
    {
      "epoch": 0.24287444489301574,
      "grad_norm": 137.66188049316406,
      "learning_rate": 1.7716555410673485e-05,
      "loss": 5.7551,
      "step": 6016
    },
    {
      "epoch": 0.2429148163100525,
      "grad_norm": 116.61540985107422,
      "learning_rate": 1.7715723670406886e-05,
      "loss": 3.8884,
      "step": 6017
    },
    {
      "epoch": 0.2429551877270892,
      "grad_norm": 131.4147491455078,
      "learning_rate": 1.7714891798219432e-05,
      "loss": 6.1351,
      "step": 6018
    },
    {
      "epoch": 0.24299555914412596,
      "grad_norm": 72.62923431396484,
      "learning_rate": 1.771405979412534e-05,
      "loss": 5.6913,
      "step": 6019
    },
    {
      "epoch": 0.24303593056116268,
      "grad_norm": 151.78988647460938,
      "learning_rate": 1.7713227658138846e-05,
      "loss": 6.6365,
      "step": 6020
    },
    {
      "epoch": 0.24307630197819943,
      "grad_norm": 94.72834777832031,
      "learning_rate": 1.7712395390274164e-05,
      "loss": 5.7779,
      "step": 6021
    },
    {
      "epoch": 0.24311667339523618,
      "grad_norm": 142.46826171875,
      "learning_rate": 1.7711562990545534e-05,
      "loss": 5.1606,
      "step": 6022
    },
    {
      "epoch": 0.2431570448122729,
      "grad_norm": 78.40290832519531,
      "learning_rate": 1.7710730458967183e-05,
      "loss": 6.8003,
      "step": 6023
    },
    {
      "epoch": 0.24319741622930965,
      "grad_norm": 74.69883728027344,
      "learning_rate": 1.7709897795553347e-05,
      "loss": 5.1733,
      "step": 6024
    },
    {
      "epoch": 0.24323778764634638,
      "grad_norm": 96.1148681640625,
      "learning_rate": 1.7709065000318265e-05,
      "loss": 4.6638,
      "step": 6025
    },
    {
      "epoch": 0.24327815906338313,
      "grad_norm": 110.64897918701172,
      "learning_rate": 1.770823207327617e-05,
      "loss": 4.5548,
      "step": 6026
    },
    {
      "epoch": 0.24331853048041988,
      "grad_norm": 103.06077575683594,
      "learning_rate": 1.770739901444131e-05,
      "loss": 5.0149,
      "step": 6027
    },
    {
      "epoch": 0.2433589018974566,
      "grad_norm": 82.20011138916016,
      "learning_rate": 1.7706565823827922e-05,
      "loss": 6.7715,
      "step": 6028
    },
    {
      "epoch": 0.24339927331449335,
      "grad_norm": 154.46163940429688,
      "learning_rate": 1.7705732501450257e-05,
      "loss": 5.6371,
      "step": 6029
    },
    {
      "epoch": 0.24343964473153007,
      "grad_norm": 102.60125732421875,
      "learning_rate": 1.770489904732256e-05,
      "loss": 5.5695,
      "step": 6030
    },
    {
      "epoch": 0.24348001614856682,
      "grad_norm": 74.56710815429688,
      "learning_rate": 1.770406546145908e-05,
      "loss": 5.2888,
      "step": 6031
    },
    {
      "epoch": 0.24352038756560354,
      "grad_norm": 97.52302551269531,
      "learning_rate": 1.7703231743874073e-05,
      "loss": 3.7177,
      "step": 6032
    },
    {
      "epoch": 0.2435607589826403,
      "grad_norm": 81.92987823486328,
      "learning_rate": 1.770239789458179e-05,
      "loss": 5.7345,
      "step": 6033
    },
    {
      "epoch": 0.24360113039967704,
      "grad_norm": 94.10509490966797,
      "learning_rate": 1.7701563913596488e-05,
      "loss": 6.6118,
      "step": 6034
    },
    {
      "epoch": 0.24364150181671376,
      "grad_norm": 90.84699249267578,
      "learning_rate": 1.770072980093243e-05,
      "loss": 6.8364,
      "step": 6035
    },
    {
      "epoch": 0.2436818732337505,
      "grad_norm": 92.27839660644531,
      "learning_rate": 1.7699895556603875e-05,
      "loss": 5.2296,
      "step": 6036
    },
    {
      "epoch": 0.24372224465078723,
      "grad_norm": 130.2853240966797,
      "learning_rate": 1.7699061180625085e-05,
      "loss": 5.1535,
      "step": 6037
    },
    {
      "epoch": 0.24376261606782398,
      "grad_norm": 129.22256469726562,
      "learning_rate": 1.7698226673010327e-05,
      "loss": 6.9742,
      "step": 6038
    },
    {
      "epoch": 0.2438029874848607,
      "grad_norm": 82.5177230834961,
      "learning_rate": 1.769739203377387e-05,
      "loss": 5.4042,
      "step": 6039
    },
    {
      "epoch": 0.24384335890189746,
      "grad_norm": 80.31067657470703,
      "learning_rate": 1.7696557262929986e-05,
      "loss": 5.7001,
      "step": 6040
    },
    {
      "epoch": 0.2438837303189342,
      "grad_norm": 81.27384948730469,
      "learning_rate": 1.769572236049294e-05,
      "loss": 5.5448,
      "step": 6041
    },
    {
      "epoch": 0.24392410173597093,
      "grad_norm": 86.07035827636719,
      "learning_rate": 1.7694887326477018e-05,
      "loss": 5.5042,
      "step": 6042
    },
    {
      "epoch": 0.24396447315300768,
      "grad_norm": 74.6534652709961,
      "learning_rate": 1.7694052160896493e-05,
      "loss": 5.2067,
      "step": 6043
    },
    {
      "epoch": 0.2440048445700444,
      "grad_norm": 121.495849609375,
      "learning_rate": 1.769321686376564e-05,
      "loss": 4.6112,
      "step": 6044
    },
    {
      "epoch": 0.24404521598708115,
      "grad_norm": 84.95056915283203,
      "learning_rate": 1.7692381435098743e-05,
      "loss": 5.3427,
      "step": 6045
    },
    {
      "epoch": 0.24408558740411787,
      "grad_norm": 75.89482879638672,
      "learning_rate": 1.7691545874910087e-05,
      "loss": 5.4676,
      "step": 6046
    },
    {
      "epoch": 0.24412595882115462,
      "grad_norm": 101.80958557128906,
      "learning_rate": 1.7690710183213956e-05,
      "loss": 4.6998,
      "step": 6047
    },
    {
      "epoch": 0.24416633023819137,
      "grad_norm": 74.06217956542969,
      "learning_rate": 1.7689874360024643e-05,
      "loss": 3.0662,
      "step": 6048
    },
    {
      "epoch": 0.2442067016552281,
      "grad_norm": 69.0732421875,
      "learning_rate": 1.7689038405356433e-05,
      "loss": 5.7977,
      "step": 6049
    },
    {
      "epoch": 0.24424707307226484,
      "grad_norm": 165.50433349609375,
      "learning_rate": 1.768820231922362e-05,
      "loss": 6.6,
      "step": 6050
    },
    {
      "epoch": 0.24428744448930156,
      "grad_norm": 99.84991455078125,
      "learning_rate": 1.7687366101640508e-05,
      "loss": 6.2236,
      "step": 6051
    },
    {
      "epoch": 0.2443278159063383,
      "grad_norm": 86.41944885253906,
      "learning_rate": 1.768652975262138e-05,
      "loss": 4.4032,
      "step": 6052
    },
    {
      "epoch": 0.24436818732337506,
      "grad_norm": 110.58154296875,
      "learning_rate": 1.7685693272180543e-05,
      "loss": 4.2381,
      "step": 6053
    },
    {
      "epoch": 0.24440855874041179,
      "grad_norm": 86.77497863769531,
      "learning_rate": 1.76848566603323e-05,
      "loss": 6.4966,
      "step": 6054
    },
    {
      "epoch": 0.24444893015744854,
      "grad_norm": 107.31599426269531,
      "learning_rate": 1.7684019917090955e-05,
      "loss": 3.989,
      "step": 6055
    },
    {
      "epoch": 0.24448930157448526,
      "grad_norm": 107.03816986083984,
      "learning_rate": 1.7683183042470812e-05,
      "loss": 6.341,
      "step": 6056
    },
    {
      "epoch": 0.244529672991522,
      "grad_norm": 147.18862915039062,
      "learning_rate": 1.768234603648618e-05,
      "loss": 5.9328,
      "step": 6057
    },
    {
      "epoch": 0.24457004440855873,
      "grad_norm": 118.75830841064453,
      "learning_rate": 1.768150889915137e-05,
      "loss": 5.7945,
      "step": 6058
    },
    {
      "epoch": 0.24461041582559548,
      "grad_norm": 123.07266998291016,
      "learning_rate": 1.7680671630480696e-05,
      "loss": 5.7966,
      "step": 6059
    },
    {
      "epoch": 0.24465078724263223,
      "grad_norm": 106.30809783935547,
      "learning_rate": 1.7679834230488472e-05,
      "loss": 5.0478,
      "step": 6060
    },
    {
      "epoch": 0.24469115865966895,
      "grad_norm": 118.6714859008789,
      "learning_rate": 1.767899669918902e-05,
      "loss": 6.2184,
      "step": 6061
    },
    {
      "epoch": 0.2447315300767057,
      "grad_norm": 97.3495864868164,
      "learning_rate": 1.7678159036596652e-05,
      "loss": 5.7919,
      "step": 6062
    },
    {
      "epoch": 0.24477190149374242,
      "grad_norm": 100.68453979492188,
      "learning_rate": 1.7677321242725695e-05,
      "loss": 4.2524,
      "step": 6063
    },
    {
      "epoch": 0.24481227291077917,
      "grad_norm": 113.27177429199219,
      "learning_rate": 1.7676483317590473e-05,
      "loss": 4.8409,
      "step": 6064
    },
    {
      "epoch": 0.2448526443278159,
      "grad_norm": 89.00183868408203,
      "learning_rate": 1.7675645261205314e-05,
      "loss": 4.8059,
      "step": 6065
    },
    {
      "epoch": 0.24489301574485264,
      "grad_norm": 109.70138549804688,
      "learning_rate": 1.7674807073584542e-05,
      "loss": 4.9354,
      "step": 6066
    },
    {
      "epoch": 0.2449333871618894,
      "grad_norm": 139.08168029785156,
      "learning_rate": 1.7673968754742493e-05,
      "loss": 5.4364,
      "step": 6067
    },
    {
      "epoch": 0.24497375857892612,
      "grad_norm": 143.52606201171875,
      "learning_rate": 1.7673130304693496e-05,
      "loss": 4.017,
      "step": 6068
    },
    {
      "epoch": 0.24501412999596286,
      "grad_norm": 79.96031951904297,
      "learning_rate": 1.767229172345189e-05,
      "loss": 4.8106,
      "step": 6069
    },
    {
      "epoch": 0.2450545014129996,
      "grad_norm": 108.87830352783203,
      "learning_rate": 1.7671453011032014e-05,
      "loss": 6.7204,
      "step": 6070
    },
    {
      "epoch": 0.24509487283003634,
      "grad_norm": 91.47180938720703,
      "learning_rate": 1.76706141674482e-05,
      "loss": 5.8246,
      "step": 6071
    },
    {
      "epoch": 0.24513524424707306,
      "grad_norm": 131.67672729492188,
      "learning_rate": 1.76697751927148e-05,
      "loss": 6.3015,
      "step": 6072
    },
    {
      "epoch": 0.2451756156641098,
      "grad_norm": 116.12157440185547,
      "learning_rate": 1.7668936086846157e-05,
      "loss": 5.3564,
      "step": 6073
    },
    {
      "epoch": 0.24521598708114656,
      "grad_norm": 96.30015563964844,
      "learning_rate": 1.766809684985661e-05,
      "loss": 4.4719,
      "step": 6074
    },
    {
      "epoch": 0.24525635849818328,
      "grad_norm": 116.22333526611328,
      "learning_rate": 1.7667257481760517e-05,
      "loss": 3.6218,
      "step": 6075
    },
    {
      "epoch": 0.24529672991522003,
      "grad_norm": 110.10753631591797,
      "learning_rate": 1.7666417982572223e-05,
      "loss": 4.5694,
      "step": 6076
    },
    {
      "epoch": 0.24533710133225675,
      "grad_norm": 73.16168975830078,
      "learning_rate": 1.7665578352306084e-05,
      "loss": 5.8268,
      "step": 6077
    },
    {
      "epoch": 0.2453774727492935,
      "grad_norm": 112.27823638916016,
      "learning_rate": 1.7664738590976457e-05,
      "loss": 4.8347,
      "step": 6078
    },
    {
      "epoch": 0.24541784416633025,
      "grad_norm": 77.43183898925781,
      "learning_rate": 1.7663898698597697e-05,
      "loss": 5.2261,
      "step": 6079
    },
    {
      "epoch": 0.24545821558336697,
      "grad_norm": 78.77147674560547,
      "learning_rate": 1.7663058675184164e-05,
      "loss": 5.0591,
      "step": 6080
    },
    {
      "epoch": 0.24549858700040372,
      "grad_norm": 82.15838623046875,
      "learning_rate": 1.7662218520750226e-05,
      "loss": 4.9217,
      "step": 6081
    },
    {
      "epoch": 0.24553895841744044,
      "grad_norm": 328.015625,
      "learning_rate": 1.7661378235310242e-05,
      "loss": 6.144,
      "step": 6082
    },
    {
      "epoch": 0.2455793298344772,
      "grad_norm": 95.46847534179688,
      "learning_rate": 1.766053781887858e-05,
      "loss": 5.5903,
      "step": 6083
    },
    {
      "epoch": 0.24561970125151392,
      "grad_norm": 88.73502349853516,
      "learning_rate": 1.7659697271469612e-05,
      "loss": 5.9261,
      "step": 6084
    },
    {
      "epoch": 0.24566007266855067,
      "grad_norm": 132.09010314941406,
      "learning_rate": 1.765885659309771e-05,
      "loss": 5.2569,
      "step": 6085
    },
    {
      "epoch": 0.24570044408558742,
      "grad_norm": 101.49199676513672,
      "learning_rate": 1.765801578377724e-05,
      "loss": 4.677,
      "step": 6086
    },
    {
      "epoch": 0.24574081550262414,
      "grad_norm": 101.85501861572266,
      "learning_rate": 1.7657174843522585e-05,
      "loss": 5.4574,
      "step": 6087
    },
    {
      "epoch": 0.2457811869196609,
      "grad_norm": 77.16210174560547,
      "learning_rate": 1.7656333772348117e-05,
      "loss": 4.8454,
      "step": 6088
    },
    {
      "epoch": 0.2458215583366976,
      "grad_norm": 105.47467041015625,
      "learning_rate": 1.7655492570268227e-05,
      "loss": 5.1233,
      "step": 6089
    },
    {
      "epoch": 0.24586192975373436,
      "grad_norm": 103.79716491699219,
      "learning_rate": 1.7654651237297288e-05,
      "loss": 3.8593,
      "step": 6090
    },
    {
      "epoch": 0.24590230117077108,
      "grad_norm": 73.02237701416016,
      "learning_rate": 1.7653809773449682e-05,
      "loss": 6.3013,
      "step": 6091
    },
    {
      "epoch": 0.24594267258780783,
      "grad_norm": 71.37913513183594,
      "learning_rate": 1.765296817873981e-05,
      "loss": 4.116,
      "step": 6092
    },
    {
      "epoch": 0.24598304400484458,
      "grad_norm": 92.89425659179688,
      "learning_rate": 1.765212645318205e-05,
      "loss": 5.9225,
      "step": 6093
    },
    {
      "epoch": 0.2460234154218813,
      "grad_norm": 99.18700408935547,
      "learning_rate": 1.7651284596790796e-05,
      "loss": 5.0996,
      "step": 6094
    },
    {
      "epoch": 0.24606378683891805,
      "grad_norm": 90.47806549072266,
      "learning_rate": 1.765044260958044e-05,
      "loss": 5.3814,
      "step": 6095
    },
    {
      "epoch": 0.24610415825595477,
      "grad_norm": 86.29560852050781,
      "learning_rate": 1.7649600491565387e-05,
      "loss": 5.7515,
      "step": 6096
    },
    {
      "epoch": 0.24614452967299152,
      "grad_norm": 118.67134857177734,
      "learning_rate": 1.7648758242760027e-05,
      "loss": 5.5733,
      "step": 6097
    },
    {
      "epoch": 0.24618490109002825,
      "grad_norm": 84.59148406982422,
      "learning_rate": 1.764791586317876e-05,
      "loss": 5.1511,
      "step": 6098
    },
    {
      "epoch": 0.246225272507065,
      "grad_norm": 110.51509094238281,
      "learning_rate": 1.764707335283599e-05,
      "loss": 7.8693,
      "step": 6099
    },
    {
      "epoch": 0.24626564392410175,
      "grad_norm": 146.7998046875,
      "learning_rate": 1.7646230711746125e-05,
      "loss": 4.9595,
      "step": 6100
    },
    {
      "epoch": 0.24630601534113847,
      "grad_norm": 125.72041320800781,
      "learning_rate": 1.7645387939923568e-05,
      "loss": 4.5315,
      "step": 6101
    },
    {
      "epoch": 0.24634638675817522,
      "grad_norm": 74.66952514648438,
      "learning_rate": 1.7644545037382732e-05,
      "loss": 6.3275,
      "step": 6102
    },
    {
      "epoch": 0.24638675817521194,
      "grad_norm": 152.0884552001953,
      "learning_rate": 1.7643702004138028e-05,
      "loss": 6.6144,
      "step": 6103
    },
    {
      "epoch": 0.2464271295922487,
      "grad_norm": 94.10076904296875,
      "learning_rate": 1.7642858840203868e-05,
      "loss": 5.8623,
      "step": 6104
    },
    {
      "epoch": 0.24646750100928544,
      "grad_norm": 168.7364501953125,
      "learning_rate": 1.764201554559467e-05,
      "loss": 5.2295,
      "step": 6105
    },
    {
      "epoch": 0.24650787242632216,
      "grad_norm": 144.9301300048828,
      "learning_rate": 1.764117212032485e-05,
      "loss": 6.9982,
      "step": 6106
    },
    {
      "epoch": 0.2465482438433589,
      "grad_norm": 141.78639221191406,
      "learning_rate": 1.764032856440883e-05,
      "loss": 3.5671,
      "step": 6107
    },
    {
      "epoch": 0.24658861526039563,
      "grad_norm": 88.1741943359375,
      "learning_rate": 1.7639484877861036e-05,
      "loss": 4.3866,
      "step": 6108
    },
    {
      "epoch": 0.24662898667743238,
      "grad_norm": 98.63414764404297,
      "learning_rate": 1.7638641060695887e-05,
      "loss": 4.7403,
      "step": 6109
    },
    {
      "epoch": 0.2466693580944691,
      "grad_norm": 112.9139404296875,
      "learning_rate": 1.7637797112927816e-05,
      "loss": 5.8775,
      "step": 6110
    },
    {
      "epoch": 0.24670972951150585,
      "grad_norm": 89.87181854248047,
      "learning_rate": 1.7636953034571245e-05,
      "loss": 6.5028,
      "step": 6111
    },
    {
      "epoch": 0.2467501009285426,
      "grad_norm": 132.26611328125,
      "learning_rate": 1.7636108825640616e-05,
      "loss": 5.8619,
      "step": 6112
    },
    {
      "epoch": 0.24679047234557933,
      "grad_norm": 90.37429809570312,
      "learning_rate": 1.7635264486150357e-05,
      "loss": 6.175,
      "step": 6113
    },
    {
      "epoch": 0.24683084376261608,
      "grad_norm": 89.14541625976562,
      "learning_rate": 1.76344200161149e-05,
      "loss": 6.1399,
      "step": 6114
    },
    {
      "epoch": 0.2468712151796528,
      "grad_norm": 80.80571746826172,
      "learning_rate": 1.7633575415548694e-05,
      "loss": 6.1435,
      "step": 6115
    },
    {
      "epoch": 0.24691158659668955,
      "grad_norm": 78.69603729248047,
      "learning_rate": 1.7632730684466173e-05,
      "loss": 6.1004,
      "step": 6116
    },
    {
      "epoch": 0.24695195801372627,
      "grad_norm": 113.71155548095703,
      "learning_rate": 1.7631885822881777e-05,
      "loss": 5.4684,
      "step": 6117
    },
    {
      "epoch": 0.24699232943076302,
      "grad_norm": 64.37646484375,
      "learning_rate": 1.763104083080996e-05,
      "loss": 2.735,
      "step": 6118
    },
    {
      "epoch": 0.24703270084779977,
      "grad_norm": 81.4853744506836,
      "learning_rate": 1.7630195708265165e-05,
      "loss": 5.398,
      "step": 6119
    },
    {
      "epoch": 0.2470730722648365,
      "grad_norm": 78.78315734863281,
      "learning_rate": 1.7629350455261836e-05,
      "loss": 6.0654,
      "step": 6120
    },
    {
      "epoch": 0.24711344368187324,
      "grad_norm": 97.87934112548828,
      "learning_rate": 1.7628505071814434e-05,
      "loss": 5.0337,
      "step": 6121
    },
    {
      "epoch": 0.24715381509890996,
      "grad_norm": 94.76945495605469,
      "learning_rate": 1.7627659557937406e-05,
      "loss": 6.0126,
      "step": 6122
    },
    {
      "epoch": 0.2471941865159467,
      "grad_norm": 79.9699935913086,
      "learning_rate": 1.7626813913645215e-05,
      "loss": 3.9845,
      "step": 6123
    },
    {
      "epoch": 0.24723455793298343,
      "grad_norm": 87.49744415283203,
      "learning_rate": 1.7625968138952315e-05,
      "loss": 4.457,
      "step": 6124
    },
    {
      "epoch": 0.24727492935002018,
      "grad_norm": 154.5457763671875,
      "learning_rate": 1.7625122233873168e-05,
      "loss": 7.6164,
      "step": 6125
    },
    {
      "epoch": 0.24731530076705693,
      "grad_norm": 106.08487701416016,
      "learning_rate": 1.7624276198422236e-05,
      "loss": 4.1024,
      "step": 6126
    },
    {
      "epoch": 0.24735567218409366,
      "grad_norm": 157.2947540283203,
      "learning_rate": 1.7623430032613984e-05,
      "loss": 4.315,
      "step": 6127
    },
    {
      "epoch": 0.2473960436011304,
      "grad_norm": 69.86736297607422,
      "learning_rate": 1.7622583736462882e-05,
      "loss": 6.2105,
      "step": 6128
    },
    {
      "epoch": 0.24743641501816713,
      "grad_norm": 100.19100952148438,
      "learning_rate": 1.7621737309983395e-05,
      "loss": 5.7441,
      "step": 6129
    },
    {
      "epoch": 0.24747678643520388,
      "grad_norm": 118.57280731201172,
      "learning_rate": 1.762089075319e-05,
      "loss": 5.3168,
      "step": 6130
    },
    {
      "epoch": 0.24751715785224063,
      "grad_norm": 96.9669189453125,
      "learning_rate": 1.7620044066097174e-05,
      "loss": 6.2141,
      "step": 6131
    },
    {
      "epoch": 0.24755752926927735,
      "grad_norm": 97.49742889404297,
      "learning_rate": 1.7619197248719384e-05,
      "loss": 7.6759,
      "step": 6132
    },
    {
      "epoch": 0.2475979006863141,
      "grad_norm": 156.0751190185547,
      "learning_rate": 1.7618350301071112e-05,
      "loss": 6.5789,
      "step": 6133
    },
    {
      "epoch": 0.24763827210335082,
      "grad_norm": 69.15914916992188,
      "learning_rate": 1.7617503223166842e-05,
      "loss": 4.1894,
      "step": 6134
    },
    {
      "epoch": 0.24767864352038757,
      "grad_norm": 102.12586975097656,
      "learning_rate": 1.7616656015021055e-05,
      "loss": 5.5927,
      "step": 6135
    },
    {
      "epoch": 0.2477190149374243,
      "grad_norm": 86.83513641357422,
      "learning_rate": 1.7615808676648235e-05,
      "loss": 3.7166,
      "step": 6136
    },
    {
      "epoch": 0.24775938635446104,
      "grad_norm": 152.3862762451172,
      "learning_rate": 1.761496120806287e-05,
      "loss": 4.3709,
      "step": 6137
    },
    {
      "epoch": 0.2477997577714978,
      "grad_norm": 70.0562515258789,
      "learning_rate": 1.7614113609279452e-05,
      "loss": 5.8269,
      "step": 6138
    },
    {
      "epoch": 0.2478401291885345,
      "grad_norm": 160.02371215820312,
      "learning_rate": 1.7613265880312472e-05,
      "loss": 5.2421,
      "step": 6139
    },
    {
      "epoch": 0.24788050060557126,
      "grad_norm": 67.5106430053711,
      "learning_rate": 1.7612418021176424e-05,
      "loss": 5.2924,
      "step": 6140
    },
    {
      "epoch": 0.24792087202260799,
      "grad_norm": 68.50265502929688,
      "learning_rate": 1.7611570031885804e-05,
      "loss": 3.5603,
      "step": 6141
    },
    {
      "epoch": 0.24796124343964474,
      "grad_norm": 108.74456024169922,
      "learning_rate": 1.761072191245511e-05,
      "loss": 5.8484,
      "step": 6142
    },
    {
      "epoch": 0.24800161485668146,
      "grad_norm": 81.83265686035156,
      "learning_rate": 1.7609873662898845e-05,
      "loss": 5.2162,
      "step": 6143
    },
    {
      "epoch": 0.2480419862737182,
      "grad_norm": 89.62882995605469,
      "learning_rate": 1.760902528323151e-05,
      "loss": 4.7464,
      "step": 6144
    },
    {
      "epoch": 0.24808235769075496,
      "grad_norm": 437.4469299316406,
      "learning_rate": 1.760817677346761e-05,
      "loss": 6.5264,
      "step": 6145
    },
    {
      "epoch": 0.24812272910779168,
      "grad_norm": 79.71366882324219,
      "learning_rate": 1.7607328133621653e-05,
      "loss": 6.0906,
      "step": 6146
    },
    {
      "epoch": 0.24816310052482843,
      "grad_norm": 84.76907348632812,
      "learning_rate": 1.760647936370815e-05,
      "loss": 5.3397,
      "step": 6147
    },
    {
      "epoch": 0.24820347194186515,
      "grad_norm": 74.17172241210938,
      "learning_rate": 1.7605630463741613e-05,
      "loss": 4.8335,
      "step": 6148
    },
    {
      "epoch": 0.2482438433589019,
      "grad_norm": 70.44876098632812,
      "learning_rate": 1.7604781433736552e-05,
      "loss": 4.9857,
      "step": 6149
    },
    {
      "epoch": 0.24828421477593862,
      "grad_norm": 91.04230499267578,
      "learning_rate": 1.7603932273707493e-05,
      "loss": 5.7558,
      "step": 6150
    },
    {
      "epoch": 0.24832458619297537,
      "grad_norm": 85.33031463623047,
      "learning_rate": 1.7603082983668945e-05,
      "loss": 6.8825,
      "step": 6151
    },
    {
      "epoch": 0.24836495761001212,
      "grad_norm": 212.30548095703125,
      "learning_rate": 1.7602233563635434e-05,
      "loss": 5.7982,
      "step": 6152
    },
    {
      "epoch": 0.24840532902704884,
      "grad_norm": 96.59770965576172,
      "learning_rate": 1.7601384013621483e-05,
      "loss": 5.9575,
      "step": 6153
    },
    {
      "epoch": 0.2484457004440856,
      "grad_norm": 63.204978942871094,
      "learning_rate": 1.760053433364161e-05,
      "loss": 4.1513,
      "step": 6154
    },
    {
      "epoch": 0.24848607186112232,
      "grad_norm": 103.4726791381836,
      "learning_rate": 1.759968452371035e-05,
      "loss": 5.8519,
      "step": 6155
    },
    {
      "epoch": 0.24852644327815906,
      "grad_norm": 85.95557403564453,
      "learning_rate": 1.7598834583842234e-05,
      "loss": 5.7179,
      "step": 6156
    },
    {
      "epoch": 0.24856681469519581,
      "grad_norm": 186.85983276367188,
      "learning_rate": 1.7597984514051793e-05,
      "loss": 6.4891,
      "step": 6157
    },
    {
      "epoch": 0.24860718611223254,
      "grad_norm": 89.6390151977539,
      "learning_rate": 1.7597134314353555e-05,
      "loss": 4.5572,
      "step": 6158
    },
    {
      "epoch": 0.2486475575292693,
      "grad_norm": 85.59256744384766,
      "learning_rate": 1.7596283984762063e-05,
      "loss": 7.1571,
      "step": 6159
    },
    {
      "epoch": 0.248687928946306,
      "grad_norm": 97.58651733398438,
      "learning_rate": 1.7595433525291854e-05,
      "loss": 5.3949,
      "step": 6160
    },
    {
      "epoch": 0.24872830036334276,
      "grad_norm": 81.62389373779297,
      "learning_rate": 1.7594582935957465e-05,
      "loss": 5.236,
      "step": 6161
    },
    {
      "epoch": 0.24876867178037948,
      "grad_norm": 84.13124084472656,
      "learning_rate": 1.7593732216773446e-05,
      "loss": 4.5777,
      "step": 6162
    },
    {
      "epoch": 0.24880904319741623,
      "grad_norm": 145.64071655273438,
      "learning_rate": 1.7592881367754337e-05,
      "loss": 4.701,
      "step": 6163
    },
    {
      "epoch": 0.24884941461445298,
      "grad_norm": 75.11285400390625,
      "learning_rate": 1.7592030388914685e-05,
      "loss": 4.0767,
      "step": 6164
    },
    {
      "epoch": 0.2488897860314897,
      "grad_norm": 67.04360961914062,
      "learning_rate": 1.7591179280269048e-05,
      "loss": 3.6344,
      "step": 6165
    },
    {
      "epoch": 0.24893015744852645,
      "grad_norm": 93.28397369384766,
      "learning_rate": 1.759032804183197e-05,
      "loss": 3.7089,
      "step": 6166
    },
    {
      "epoch": 0.24897052886556317,
      "grad_norm": 122.21196746826172,
      "learning_rate": 1.7589476673618e-05,
      "loss": 5.4149,
      "step": 6167
    },
    {
      "epoch": 0.24901090028259992,
      "grad_norm": 72.30835723876953,
      "learning_rate": 1.758862517564171e-05,
      "loss": 4.1387,
      "step": 6168
    },
    {
      "epoch": 0.24905127169963664,
      "grad_norm": 111.94731140136719,
      "learning_rate": 1.758777354791765e-05,
      "loss": 7.5308,
      "step": 6169
    },
    {
      "epoch": 0.2490916431166734,
      "grad_norm": 85.4908676147461,
      "learning_rate": 1.7586921790460376e-05,
      "loss": 5.2088,
      "step": 6170
    },
    {
      "epoch": 0.24913201453371014,
      "grad_norm": 172.31910705566406,
      "learning_rate": 1.758606990328446e-05,
      "loss": 5.926,
      "step": 6171
    },
    {
      "epoch": 0.24917238595074687,
      "grad_norm": 104.54186248779297,
      "learning_rate": 1.7585217886404464e-05,
      "loss": 3.7288,
      "step": 6172
    },
    {
      "epoch": 0.24921275736778362,
      "grad_norm": 144.40444946289062,
      "learning_rate": 1.7584365739834953e-05,
      "loss": 5.3484,
      "step": 6173
    },
    {
      "epoch": 0.24925312878482034,
      "grad_norm": 112.33906555175781,
      "learning_rate": 1.7583513463590498e-05,
      "loss": 5.8158,
      "step": 6174
    },
    {
      "epoch": 0.2492935002018571,
      "grad_norm": 98.13909912109375,
      "learning_rate": 1.758266105768567e-05,
      "loss": 5.8801,
      "step": 6175
    },
    {
      "epoch": 0.2493338716188938,
      "grad_norm": 93.43961334228516,
      "learning_rate": 1.758180852213505e-05,
      "loss": 4.323,
      "step": 6176
    },
    {
      "epoch": 0.24937424303593056,
      "grad_norm": 70.71387481689453,
      "learning_rate": 1.7580955856953204e-05,
      "loss": 5.9202,
      "step": 6177
    },
    {
      "epoch": 0.2494146144529673,
      "grad_norm": 149.18557739257812,
      "learning_rate": 1.7580103062154718e-05,
      "loss": 4.6175,
      "step": 6178
    },
    {
      "epoch": 0.24945498587000403,
      "grad_norm": 84.14598846435547,
      "learning_rate": 1.757925013775417e-05,
      "loss": 6.3908,
      "step": 6179
    },
    {
      "epoch": 0.24949535728704078,
      "grad_norm": 91.55603790283203,
      "learning_rate": 1.7578397083766147e-05,
      "loss": 7.2505,
      "step": 6180
    },
    {
      "epoch": 0.2495357287040775,
      "grad_norm": 84.24451446533203,
      "learning_rate": 1.7577543900205224e-05,
      "loss": 6.4128,
      "step": 6181
    },
    {
      "epoch": 0.24957610012111425,
      "grad_norm": 88.58628845214844,
      "learning_rate": 1.7576690587086e-05,
      "loss": 6.0695,
      "step": 6182
    },
    {
      "epoch": 0.249616471538151,
      "grad_norm": 114.7388687133789,
      "learning_rate": 1.7575837144423058e-05,
      "loss": 5.001,
      "step": 6183
    },
    {
      "epoch": 0.24965684295518772,
      "grad_norm": 88.84111785888672,
      "learning_rate": 1.7574983572230994e-05,
      "loss": 6.2022,
      "step": 6184
    },
    {
      "epoch": 0.24969721437222447,
      "grad_norm": 86.6026611328125,
      "learning_rate": 1.7574129870524394e-05,
      "loss": 5.9355,
      "step": 6185
    },
    {
      "epoch": 0.2497375857892612,
      "grad_norm": 110.44198608398438,
      "learning_rate": 1.7573276039317864e-05,
      "loss": 5.6928,
      "step": 6186
    },
    {
      "epoch": 0.24977795720629795,
      "grad_norm": 85.6563491821289,
      "learning_rate": 1.7572422078625998e-05,
      "loss": 6.1385,
      "step": 6187
    },
    {
      "epoch": 0.24981832862333467,
      "grad_norm": 125.09893035888672,
      "learning_rate": 1.75715679884634e-05,
      "loss": 5.3785,
      "step": 6188
    },
    {
      "epoch": 0.24985870004037142,
      "grad_norm": 118.06724548339844,
      "learning_rate": 1.7570713768844667e-05,
      "loss": 5.1215,
      "step": 6189
    },
    {
      "epoch": 0.24989907145740817,
      "grad_norm": 142.54342651367188,
      "learning_rate": 1.756985941978441e-05,
      "loss": 6.728,
      "step": 6190
    },
    {
      "epoch": 0.2499394428744449,
      "grad_norm": 91.7153091430664,
      "learning_rate": 1.756900494129723e-05,
      "loss": 6.542,
      "step": 6191
    },
    {
      "epoch": 0.24997981429148164,
      "grad_norm": 110.15680694580078,
      "learning_rate": 1.7568150333397744e-05,
      "loss": 5.8676,
      "step": 6192
    },
    {
      "epoch": 0.25002018570851836,
      "grad_norm": 121.96024322509766,
      "learning_rate": 1.7567295596100558e-05,
      "loss": 5.9351,
      "step": 6193
    },
    {
      "epoch": 0.2500605571255551,
      "grad_norm": 80.46528625488281,
      "learning_rate": 1.756644072942029e-05,
      "loss": 4.8453,
      "step": 6194
    },
    {
      "epoch": 0.25010092854259186,
      "grad_norm": 115.87245178222656,
      "learning_rate": 1.7565585733371553e-05,
      "loss": 6.6531,
      "step": 6195
    },
    {
      "epoch": 0.2501412999596286,
      "grad_norm": 173.0806427001953,
      "learning_rate": 1.7564730607968967e-05,
      "loss": 5.2667,
      "step": 6196
    },
    {
      "epoch": 0.2501816713766653,
      "grad_norm": 82.59706115722656,
      "learning_rate": 1.7563875353227153e-05,
      "loss": 3.9195,
      "step": 6197
    },
    {
      "epoch": 0.2502220427937021,
      "grad_norm": 131.72958374023438,
      "learning_rate": 1.7563019969160732e-05,
      "loss": 5.1299,
      "step": 6198
    },
    {
      "epoch": 0.2502624142107388,
      "grad_norm": 187.73776245117188,
      "learning_rate": 1.756216445578433e-05,
      "loss": 4.4497,
      "step": 6199
    },
    {
      "epoch": 0.2503027856277755,
      "grad_norm": 97.48921966552734,
      "learning_rate": 1.756130881311258e-05,
      "loss": 6.0874,
      "step": 6200
    },
    {
      "epoch": 0.25034315704481225,
      "grad_norm": 96.76197052001953,
      "learning_rate": 1.7560453041160103e-05,
      "loss": 5.1443,
      "step": 6201
    },
    {
      "epoch": 0.250383528461849,
      "grad_norm": 103.25464630126953,
      "learning_rate": 1.7559597139941534e-05,
      "loss": 4.7748,
      "step": 6202
    },
    {
      "epoch": 0.25042389987888575,
      "grad_norm": 126.92750549316406,
      "learning_rate": 1.7558741109471504e-05,
      "loss": 5.8591,
      "step": 6203
    },
    {
      "epoch": 0.25046427129592247,
      "grad_norm": 134.45956420898438,
      "learning_rate": 1.7557884949764657e-05,
      "loss": 4.134,
      "step": 6204
    },
    {
      "epoch": 0.25050464271295925,
      "grad_norm": 98.98147583007812,
      "learning_rate": 1.7557028660835622e-05,
      "loss": 4.8024,
      "step": 6205
    },
    {
      "epoch": 0.25054501412999597,
      "grad_norm": 154.3301239013672,
      "learning_rate": 1.7556172242699046e-05,
      "loss": 4.5459,
      "step": 6206
    },
    {
      "epoch": 0.2505853855470327,
      "grad_norm": 161.1691131591797,
      "learning_rate": 1.755531569536957e-05,
      "loss": 3.5216,
      "step": 6207
    },
    {
      "epoch": 0.2506257569640694,
      "grad_norm": 109.20459747314453,
      "learning_rate": 1.7554459018861835e-05,
      "loss": 5.1131,
      "step": 6208
    },
    {
      "epoch": 0.2506661283811062,
      "grad_norm": 114.33198547363281,
      "learning_rate": 1.7553602213190495e-05,
      "loss": 6.0381,
      "step": 6209
    },
    {
      "epoch": 0.2507064997981429,
      "grad_norm": 111.31488037109375,
      "learning_rate": 1.7552745278370197e-05,
      "loss": 5.3981,
      "step": 6210
    },
    {
      "epoch": 0.25074687121517963,
      "grad_norm": 126.32987213134766,
      "learning_rate": 1.755188821441559e-05,
      "loss": 4.89,
      "step": 6211
    },
    {
      "epoch": 0.2507872426322164,
      "grad_norm": 161.12237548828125,
      "learning_rate": 1.755103102134133e-05,
      "loss": 6.9404,
      "step": 6212
    },
    {
      "epoch": 0.25082761404925313,
      "grad_norm": 91.447021484375,
      "learning_rate": 1.7550173699162072e-05,
      "loss": 5.7067,
      "step": 6213
    },
    {
      "epoch": 0.25086798546628986,
      "grad_norm": 101.62844848632812,
      "learning_rate": 1.754931624789247e-05,
      "loss": 4.6187,
      "step": 6214
    },
    {
      "epoch": 0.2509083568833266,
      "grad_norm": 124.40998840332031,
      "learning_rate": 1.7548458667547196e-05,
      "loss": 4.6906,
      "step": 6215
    },
    {
      "epoch": 0.25094872830036336,
      "grad_norm": 241.7193145751953,
      "learning_rate": 1.7547600958140906e-05,
      "loss": 4.0161,
      "step": 6216
    },
    {
      "epoch": 0.2509890997174001,
      "grad_norm": 105.33757019042969,
      "learning_rate": 1.754674311968826e-05,
      "loss": 5.8267,
      "step": 6217
    },
    {
      "epoch": 0.2510294711344368,
      "grad_norm": 115.40699768066406,
      "learning_rate": 1.7545885152203933e-05,
      "loss": 6.656,
      "step": 6218
    },
    {
      "epoch": 0.2510698425514736,
      "grad_norm": 1226.1015625,
      "learning_rate": 1.7545027055702587e-05,
      "loss": 5.166,
      "step": 6219
    },
    {
      "epoch": 0.2511102139685103,
      "grad_norm": 107.03456115722656,
      "learning_rate": 1.7544168830198902e-05,
      "loss": 6.1734,
      "step": 6220
    },
    {
      "epoch": 0.251150585385547,
      "grad_norm": 96.11265563964844,
      "learning_rate": 1.7543310475707543e-05,
      "loss": 5.5733,
      "step": 6221
    },
    {
      "epoch": 0.2511909568025838,
      "grad_norm": 84.55817413330078,
      "learning_rate": 1.754245199224319e-05,
      "loss": 4.1727,
      "step": 6222
    },
    {
      "epoch": 0.2512313282196205,
      "grad_norm": 208.14181518554688,
      "learning_rate": 1.7541593379820523e-05,
      "loss": 5.8632,
      "step": 6223
    },
    {
      "epoch": 0.25127169963665724,
      "grad_norm": 102.23393249511719,
      "learning_rate": 1.754073463845422e-05,
      "loss": 5.6258,
      "step": 6224
    },
    {
      "epoch": 0.25131207105369396,
      "grad_norm": 106.7198257446289,
      "learning_rate": 1.7539875768158963e-05,
      "loss": 6.5577,
      "step": 6225
    },
    {
      "epoch": 0.25135244247073074,
      "grad_norm": 115.28688049316406,
      "learning_rate": 1.753901676894944e-05,
      "loss": 7.2929,
      "step": 6226
    },
    {
      "epoch": 0.25139281388776746,
      "grad_norm": 95.27510833740234,
      "learning_rate": 1.753815764084033e-05,
      "loss": 7.0083,
      "step": 6227
    },
    {
      "epoch": 0.2514331853048042,
      "grad_norm": 89.53400421142578,
      "learning_rate": 1.753729838384633e-05,
      "loss": 4.4764,
      "step": 6228
    },
    {
      "epoch": 0.25147355672184096,
      "grad_norm": 183.0060272216797,
      "learning_rate": 1.7536438997982127e-05,
      "loss": 5.4622,
      "step": 6229
    },
    {
      "epoch": 0.2515139281388777,
      "grad_norm": 122.78810119628906,
      "learning_rate": 1.7535579483262415e-05,
      "loss": 4.9507,
      "step": 6230
    },
    {
      "epoch": 0.2515542995559144,
      "grad_norm": 83.96964263916016,
      "learning_rate": 1.7534719839701893e-05,
      "loss": 4.2164,
      "step": 6231
    },
    {
      "epoch": 0.25159467097295113,
      "grad_norm": 194.5238800048828,
      "learning_rate": 1.7533860067315256e-05,
      "loss": 5.4619,
      "step": 6232
    },
    {
      "epoch": 0.2516350423899879,
      "grad_norm": 170.473388671875,
      "learning_rate": 1.7533000166117204e-05,
      "loss": 4.615,
      "step": 6233
    },
    {
      "epoch": 0.25167541380702463,
      "grad_norm": 86.97417449951172,
      "learning_rate": 1.753214013612244e-05,
      "loss": 4.3194,
      "step": 6234
    },
    {
      "epoch": 0.25171578522406135,
      "grad_norm": 110.22871398925781,
      "learning_rate": 1.753127997734567e-05,
      "loss": 4.5581,
      "step": 6235
    },
    {
      "epoch": 0.2517561566410981,
      "grad_norm": 185.50430297851562,
      "learning_rate": 1.75304196898016e-05,
      "loss": 4.7949,
      "step": 6236
    },
    {
      "epoch": 0.25179652805813485,
      "grad_norm": 74.81050109863281,
      "learning_rate": 1.7529559273504933e-05,
      "loss": 6.8565,
      "step": 6237
    },
    {
      "epoch": 0.25183689947517157,
      "grad_norm": 94.2925033569336,
      "learning_rate": 1.7528698728470393e-05,
      "loss": 4.5759,
      "step": 6238
    },
    {
      "epoch": 0.2518772708922083,
      "grad_norm": 90.5698013305664,
      "learning_rate": 1.752783805471268e-05,
      "loss": 4.7516,
      "step": 6239
    },
    {
      "epoch": 0.25191764230924507,
      "grad_norm": 158.31991577148438,
      "learning_rate": 1.7526977252246514e-05,
      "loss": 4.2162,
      "step": 6240
    },
    {
      "epoch": 0.2519580137262818,
      "grad_norm": 135.47804260253906,
      "learning_rate": 1.7526116321086618e-05,
      "loss": 3.5299,
      "step": 6241
    },
    {
      "epoch": 0.2519983851433185,
      "grad_norm": 92.74293518066406,
      "learning_rate": 1.7525255261247705e-05,
      "loss": 6.7133,
      "step": 6242
    },
    {
      "epoch": 0.2520387565603553,
      "grad_norm": 114.84576416015625,
      "learning_rate": 1.7524394072744502e-05,
      "loss": 6.0817,
      "step": 6243
    },
    {
      "epoch": 0.252079127977392,
      "grad_norm": 75.35694885253906,
      "learning_rate": 1.752353275559173e-05,
      "loss": 5.7976,
      "step": 6244
    },
    {
      "epoch": 0.25211949939442874,
      "grad_norm": 90.60497283935547,
      "learning_rate": 1.7522671309804115e-05,
      "loss": 4.9775,
      "step": 6245
    },
    {
      "epoch": 0.25215987081146546,
      "grad_norm": 95.58950805664062,
      "learning_rate": 1.752180973539639e-05,
      "loss": 6.8952,
      "step": 6246
    },
    {
      "epoch": 0.25220024222850224,
      "grad_norm": 131.49925231933594,
      "learning_rate": 1.752094803238328e-05,
      "loss": 6.3922,
      "step": 6247
    },
    {
      "epoch": 0.25224061364553896,
      "grad_norm": 150.3668975830078,
      "learning_rate": 1.752008620077952e-05,
      "loss": 5.2366,
      "step": 6248
    },
    {
      "epoch": 0.2522809850625757,
      "grad_norm": 81.45880126953125,
      "learning_rate": 1.7519224240599853e-05,
      "loss": 3.3788,
      "step": 6249
    },
    {
      "epoch": 0.25232135647961246,
      "grad_norm": 72.12129211425781,
      "learning_rate": 1.751836215185901e-05,
      "loss": 4.6007,
      "step": 6250
    },
    {
      "epoch": 0.2523617278966492,
      "grad_norm": 115.51940155029297,
      "learning_rate": 1.7517499934571726e-05,
      "loss": 5.7957,
      "step": 6251
    },
    {
      "epoch": 0.2524020993136859,
      "grad_norm": 139.35052490234375,
      "learning_rate": 1.7516637588752746e-05,
      "loss": 6.0107,
      "step": 6252
    },
    {
      "epoch": 0.2524424707307226,
      "grad_norm": 160.02638244628906,
      "learning_rate": 1.751577511441682e-05,
      "loss": 5.7681,
      "step": 6253
    },
    {
      "epoch": 0.2524828421477594,
      "grad_norm": 91.62287139892578,
      "learning_rate": 1.751491251157869e-05,
      "loss": 5.7051,
      "step": 6254
    },
    {
      "epoch": 0.2525232135647961,
      "grad_norm": 128.94981384277344,
      "learning_rate": 1.7514049780253103e-05,
      "loss": 5.6263,
      "step": 6255
    },
    {
      "epoch": 0.25256358498183284,
      "grad_norm": 114.20903015136719,
      "learning_rate": 1.7513186920454815e-05,
      "loss": 7.7824,
      "step": 6256
    },
    {
      "epoch": 0.2526039563988696,
      "grad_norm": 86.91571807861328,
      "learning_rate": 1.751232393219857e-05,
      "loss": 5.8799,
      "step": 6257
    },
    {
      "epoch": 0.25264432781590634,
      "grad_norm": 181.08912658691406,
      "learning_rate": 1.751146081549913e-05,
      "loss": 4.0497,
      "step": 6258
    },
    {
      "epoch": 0.25268469923294307,
      "grad_norm": 123.8399887084961,
      "learning_rate": 1.751059757037125e-05,
      "loss": 5.2922,
      "step": 6259
    },
    {
      "epoch": 0.2527250706499798,
      "grad_norm": 112.92724609375,
      "learning_rate": 1.7509734196829694e-05,
      "loss": 4.353,
      "step": 6260
    },
    {
      "epoch": 0.25276544206701657,
      "grad_norm": 84.37520599365234,
      "learning_rate": 1.7508870694889213e-05,
      "loss": 5.232,
      "step": 6261
    },
    {
      "epoch": 0.2528058134840533,
      "grad_norm": 141.375244140625,
      "learning_rate": 1.7508007064564587e-05,
      "loss": 4.967,
      "step": 6262
    },
    {
      "epoch": 0.25284618490109,
      "grad_norm": 166.15736389160156,
      "learning_rate": 1.7507143305870565e-05,
      "loss": 4.902,
      "step": 6263
    },
    {
      "epoch": 0.2528865563181268,
      "grad_norm": 133.07864379882812,
      "learning_rate": 1.7506279418821927e-05,
      "loss": 5.1468,
      "step": 6264
    },
    {
      "epoch": 0.2529269277351635,
      "grad_norm": 80.92617797851562,
      "learning_rate": 1.7505415403433435e-05,
      "loss": 5.836,
      "step": 6265
    },
    {
      "epoch": 0.25296729915220023,
      "grad_norm": 119.78252410888672,
      "learning_rate": 1.750455125971987e-05,
      "loss": 4.6594,
      "step": 6266
    },
    {
      "epoch": 0.25300767056923695,
      "grad_norm": 112.9336929321289,
      "learning_rate": 1.7503686987696e-05,
      "loss": 4.4723,
      "step": 6267
    },
    {
      "epoch": 0.25304804198627373,
      "grad_norm": 105.30375671386719,
      "learning_rate": 1.7502822587376608e-05,
      "loss": 5.0268,
      "step": 6268
    },
    {
      "epoch": 0.25308841340331045,
      "grad_norm": 144.67294311523438,
      "learning_rate": 1.7501958058776468e-05,
      "loss": 6.709,
      "step": 6269
    },
    {
      "epoch": 0.2531287848203472,
      "grad_norm": 124.16264343261719,
      "learning_rate": 1.7501093401910366e-05,
      "loss": 6.6009,
      "step": 6270
    },
    {
      "epoch": 0.25316915623738395,
      "grad_norm": 132.23196411132812,
      "learning_rate": 1.750022861679308e-05,
      "loss": 6.8395,
      "step": 6271
    },
    {
      "epoch": 0.2532095276544207,
      "grad_norm": 129.5429229736328,
      "learning_rate": 1.7499363703439398e-05,
      "loss": 5.3339,
      "step": 6272
    },
    {
      "epoch": 0.2532498990714574,
      "grad_norm": 127.76680755615234,
      "learning_rate": 1.749849866186411e-05,
      "loss": 6.0568,
      "step": 6273
    },
    {
      "epoch": 0.2532902704884942,
      "grad_norm": 110.38898468017578,
      "learning_rate": 1.749763349208201e-05,
      "loss": 5.1859,
      "step": 6274
    },
    {
      "epoch": 0.2533306419055309,
      "grad_norm": 103.74479675292969,
      "learning_rate": 1.749676819410788e-05,
      "loss": 5.1959,
      "step": 6275
    },
    {
      "epoch": 0.2533710133225676,
      "grad_norm": 92.05522918701172,
      "learning_rate": 1.749590276795652e-05,
      "loss": 6.6048,
      "step": 6276
    },
    {
      "epoch": 0.25341138473960434,
      "grad_norm": 120.73477172851562,
      "learning_rate": 1.7495037213642728e-05,
      "loss": 3.6454,
      "step": 6277
    },
    {
      "epoch": 0.2534517561566411,
      "grad_norm": 72.93058776855469,
      "learning_rate": 1.7494171531181307e-05,
      "loss": 4.2538,
      "step": 6278
    },
    {
      "epoch": 0.25349212757367784,
      "grad_norm": 104.45809936523438,
      "learning_rate": 1.7493305720587047e-05,
      "loss": 5.3649,
      "step": 6279
    },
    {
      "epoch": 0.25353249899071456,
      "grad_norm": 77.50835418701172,
      "learning_rate": 1.749243978187476e-05,
      "loss": 3.519,
      "step": 6280
    },
    {
      "epoch": 0.25357287040775134,
      "grad_norm": 107.38404083251953,
      "learning_rate": 1.7491573715059247e-05,
      "loss": 6.3748,
      "step": 6281
    },
    {
      "epoch": 0.25361324182478806,
      "grad_norm": 137.98316955566406,
      "learning_rate": 1.7490707520155314e-05,
      "loss": 3.5222,
      "step": 6282
    },
    {
      "epoch": 0.2536536132418248,
      "grad_norm": 98.86771392822266,
      "learning_rate": 1.7489841197177783e-05,
      "loss": 5.1532,
      "step": 6283
    },
    {
      "epoch": 0.2536939846588615,
      "grad_norm": 90.6309814453125,
      "learning_rate": 1.748897474614145e-05,
      "loss": 5.3737,
      "step": 6284
    },
    {
      "epoch": 0.2537343560758983,
      "grad_norm": 118.05745697021484,
      "learning_rate": 1.7488108167061143e-05,
      "loss": 5.4249,
      "step": 6285
    },
    {
      "epoch": 0.253774727492935,
      "grad_norm": 122.74826049804688,
      "learning_rate": 1.7487241459951667e-05,
      "loss": 5.4277,
      "step": 6286
    },
    {
      "epoch": 0.2538150989099717,
      "grad_norm": 89.98331451416016,
      "learning_rate": 1.7486374624827848e-05,
      "loss": 5.5208,
      "step": 6287
    },
    {
      "epoch": 0.2538554703270085,
      "grad_norm": 99.07605743408203,
      "learning_rate": 1.7485507661704506e-05,
      "loss": 5.8392,
      "step": 6288
    },
    {
      "epoch": 0.2538958417440452,
      "grad_norm": 77.55787658691406,
      "learning_rate": 1.7484640570596462e-05,
      "loss": 4.6819,
      "step": 6289
    },
    {
      "epoch": 0.25393621316108195,
      "grad_norm": 111.29324340820312,
      "learning_rate": 1.7483773351518544e-05,
      "loss": 5.1715,
      "step": 6290
    },
    {
      "epoch": 0.25397658457811867,
      "grad_norm": 86.63137817382812,
      "learning_rate": 1.7482906004485573e-05,
      "loss": 5.7305,
      "step": 6291
    },
    {
      "epoch": 0.25401695599515545,
      "grad_norm": 145.24636840820312,
      "learning_rate": 1.7482038529512385e-05,
      "loss": 5.7225,
      "step": 6292
    },
    {
      "epoch": 0.25405732741219217,
      "grad_norm": 360.0796203613281,
      "learning_rate": 1.748117092661381e-05,
      "loss": 4.7689,
      "step": 6293
    },
    {
      "epoch": 0.2540976988292289,
      "grad_norm": 111.83570098876953,
      "learning_rate": 1.7480303195804684e-05,
      "loss": 4.9627,
      "step": 6294
    },
    {
      "epoch": 0.25413807024626567,
      "grad_norm": 124.73916625976562,
      "learning_rate": 1.747943533709984e-05,
      "loss": 6.1346,
      "step": 6295
    },
    {
      "epoch": 0.2541784416633024,
      "grad_norm": 100.7200927734375,
      "learning_rate": 1.747856735051412e-05,
      "loss": 5.6631,
      "step": 6296
    },
    {
      "epoch": 0.2542188130803391,
      "grad_norm": 96.4183120727539,
      "learning_rate": 1.7477699236062357e-05,
      "loss": 4.0339,
      "step": 6297
    },
    {
      "epoch": 0.25425918449737583,
      "grad_norm": 111.28185272216797,
      "learning_rate": 1.74768309937594e-05,
      "loss": 5.7538,
      "step": 6298
    },
    {
      "epoch": 0.2542995559144126,
      "grad_norm": 101.94537353515625,
      "learning_rate": 1.74759626236201e-05,
      "loss": 5.3143,
      "step": 6299
    },
    {
      "epoch": 0.25433992733144933,
      "grad_norm": 78.84424591064453,
      "learning_rate": 1.7475094125659288e-05,
      "loss": 5.8064,
      "step": 6300
    },
    {
      "epoch": 0.25438029874848606,
      "grad_norm": 120.96675872802734,
      "learning_rate": 1.7474225499891823e-05,
      "loss": 5.9637,
      "step": 6301
    },
    {
      "epoch": 0.25442067016552283,
      "grad_norm": 97.06033325195312,
      "learning_rate": 1.747335674633256e-05,
      "loss": 5.781,
      "step": 6302
    },
    {
      "epoch": 0.25446104158255956,
      "grad_norm": 76.77945709228516,
      "learning_rate": 1.7472487864996345e-05,
      "loss": 4.5461,
      "step": 6303
    },
    {
      "epoch": 0.2545014129995963,
      "grad_norm": 92.39286041259766,
      "learning_rate": 1.7471618855898035e-05,
      "loss": 6.749,
      "step": 6304
    },
    {
      "epoch": 0.254541784416633,
      "grad_norm": 76.98026275634766,
      "learning_rate": 1.7470749719052492e-05,
      "loss": 5.4574,
      "step": 6305
    },
    {
      "epoch": 0.2545821558336698,
      "grad_norm": 119.32894134521484,
      "learning_rate": 1.7469880454474575e-05,
      "loss": 6.7768,
      "step": 6306
    },
    {
      "epoch": 0.2546225272507065,
      "grad_norm": 75.5621109008789,
      "learning_rate": 1.7469011062179145e-05,
      "loss": 5.0836,
      "step": 6307
    },
    {
      "epoch": 0.2546628986677432,
      "grad_norm": 83.91072845458984,
      "learning_rate": 1.746814154218107e-05,
      "loss": 5.4244,
      "step": 6308
    },
    {
      "epoch": 0.25470327008478,
      "grad_norm": 100.27023315429688,
      "learning_rate": 1.746727189449521e-05,
      "loss": 5.3359,
      "step": 6309
    },
    {
      "epoch": 0.2547436415018167,
      "grad_norm": 86.35749053955078,
      "learning_rate": 1.746640211913644e-05,
      "loss": 5.375,
      "step": 6310
    },
    {
      "epoch": 0.25478401291885344,
      "grad_norm": 100.63524627685547,
      "learning_rate": 1.7465532216119628e-05,
      "loss": 6.2554,
      "step": 6311
    },
    {
      "epoch": 0.25482438433589016,
      "grad_norm": 61.80023956298828,
      "learning_rate": 1.7464662185459648e-05,
      "loss": 6.2274,
      "step": 6312
    },
    {
      "epoch": 0.25486475575292694,
      "grad_norm": 98.67611694335938,
      "learning_rate": 1.746379202717138e-05,
      "loss": 4.5992,
      "step": 6313
    },
    {
      "epoch": 0.25490512716996366,
      "grad_norm": 83.3555679321289,
      "learning_rate": 1.7462921741269692e-05,
      "loss": 4.7495,
      "step": 6314
    },
    {
      "epoch": 0.2549454985870004,
      "grad_norm": 95.47452545166016,
      "learning_rate": 1.746205132776947e-05,
      "loss": 4.6093,
      "step": 6315
    },
    {
      "epoch": 0.25498587000403716,
      "grad_norm": 126.46708679199219,
      "learning_rate": 1.7461180786685597e-05,
      "loss": 5.6093,
      "step": 6316
    },
    {
      "epoch": 0.2550262414210739,
      "grad_norm": 105.2287368774414,
      "learning_rate": 1.7460310118032957e-05,
      "loss": 5.0758,
      "step": 6317
    },
    {
      "epoch": 0.2550666128381106,
      "grad_norm": 92.46643829345703,
      "learning_rate": 1.7459439321826436e-05,
      "loss": 6.8501,
      "step": 6318
    },
    {
      "epoch": 0.25510698425514733,
      "grad_norm": 110.91433715820312,
      "learning_rate": 1.745856839808092e-05,
      "loss": 3.6403,
      "step": 6319
    },
    {
      "epoch": 0.2551473556721841,
      "grad_norm": 88.14583587646484,
      "learning_rate": 1.74576973468113e-05,
      "loss": 5.9619,
      "step": 6320
    },
    {
      "epoch": 0.25518772708922083,
      "grad_norm": 141.57667541503906,
      "learning_rate": 1.745682616803247e-05,
      "loss": 5.4306,
      "step": 6321
    },
    {
      "epoch": 0.25522809850625755,
      "grad_norm": 101.95860290527344,
      "learning_rate": 1.7455954861759328e-05,
      "loss": 5.6014,
      "step": 6322
    },
    {
      "epoch": 0.2552684699232943,
      "grad_norm": 83.26870727539062,
      "learning_rate": 1.7455083428006767e-05,
      "loss": 2.8598,
      "step": 6323
    },
    {
      "epoch": 0.25530884134033105,
      "grad_norm": 126.67935943603516,
      "learning_rate": 1.745421186678969e-05,
      "loss": 5.3562,
      "step": 6324
    },
    {
      "epoch": 0.25534921275736777,
      "grad_norm": 96.77794647216797,
      "learning_rate": 1.7453340178122996e-05,
      "loss": 6.6083,
      "step": 6325
    },
    {
      "epoch": 0.25538958417440455,
      "grad_norm": 110.10690307617188,
      "learning_rate": 1.7452468362021585e-05,
      "loss": 5.5261,
      "step": 6326
    },
    {
      "epoch": 0.25542995559144127,
      "grad_norm": 111.13928985595703,
      "learning_rate": 1.7451596418500375e-05,
      "loss": 5.234,
      "step": 6327
    },
    {
      "epoch": 0.255470327008478,
      "grad_norm": 89.25446319580078,
      "learning_rate": 1.7450724347574265e-05,
      "loss": 5.004,
      "step": 6328
    },
    {
      "epoch": 0.2555106984255147,
      "grad_norm": 87.22100830078125,
      "learning_rate": 1.7449852149258165e-05,
      "loss": 5.9016,
      "step": 6329
    },
    {
      "epoch": 0.2555510698425515,
      "grad_norm": 91.14813995361328,
      "learning_rate": 1.744897982356699e-05,
      "loss": 5.9671,
      "step": 6330
    },
    {
      "epoch": 0.2555914412595882,
      "grad_norm": 83.15841674804688,
      "learning_rate": 1.744810737051566e-05,
      "loss": 2.8528,
      "step": 6331
    },
    {
      "epoch": 0.25563181267662494,
      "grad_norm": 83.12249755859375,
      "learning_rate": 1.744723479011908e-05,
      "loss": 6.3958,
      "step": 6332
    },
    {
      "epoch": 0.2556721840936617,
      "grad_norm": 110.26986694335938,
      "learning_rate": 1.7446362082392178e-05,
      "loss": 6.2563,
      "step": 6333
    },
    {
      "epoch": 0.25571255551069844,
      "grad_norm": 81.53353118896484,
      "learning_rate": 1.7445489247349874e-05,
      "loss": 5.6279,
      "step": 6334
    },
    {
      "epoch": 0.25575292692773516,
      "grad_norm": 115.56771850585938,
      "learning_rate": 1.744461628500709e-05,
      "loss": 4.0971,
      "step": 6335
    },
    {
      "epoch": 0.2557932983447719,
      "grad_norm": 336.3779602050781,
      "learning_rate": 1.7443743195378752e-05,
      "loss": 6.4127,
      "step": 6336
    },
    {
      "epoch": 0.25583366976180866,
      "grad_norm": 112.94284057617188,
      "learning_rate": 1.7442869978479786e-05,
      "loss": 4.3453,
      "step": 6337
    },
    {
      "epoch": 0.2558740411788454,
      "grad_norm": 98.74012756347656,
      "learning_rate": 1.7441996634325128e-05,
      "loss": 5.971,
      "step": 6338
    },
    {
      "epoch": 0.2559144125958821,
      "grad_norm": 80.39956665039062,
      "learning_rate": 1.7441123162929702e-05,
      "loss": 4.5162,
      "step": 6339
    },
    {
      "epoch": 0.2559547840129189,
      "grad_norm": 118.728515625,
      "learning_rate": 1.7440249564308444e-05,
      "loss": 6.7702,
      "step": 6340
    },
    {
      "epoch": 0.2559951554299556,
      "grad_norm": 86.6015853881836,
      "learning_rate": 1.7439375838476298e-05,
      "loss": 4.7138,
      "step": 6341
    },
    {
      "epoch": 0.2560355268469923,
      "grad_norm": 192.9158935546875,
      "learning_rate": 1.7438501985448196e-05,
      "loss": 4.9781,
      "step": 6342
    },
    {
      "epoch": 0.25607589826402904,
      "grad_norm": 143.8426055908203,
      "learning_rate": 1.7437628005239075e-05,
      "loss": 6.0155,
      "step": 6343
    },
    {
      "epoch": 0.2561162696810658,
      "grad_norm": 82.00563049316406,
      "learning_rate": 1.743675389786389e-05,
      "loss": 5.3631,
      "step": 6344
    },
    {
      "epoch": 0.25615664109810254,
      "grad_norm": 97.64118194580078,
      "learning_rate": 1.743587966333758e-05,
      "loss": 4.9296,
      "step": 6345
    },
    {
      "epoch": 0.25619701251513927,
      "grad_norm": 86.60591888427734,
      "learning_rate": 1.7435005301675082e-05,
      "loss": 6.7394,
      "step": 6346
    },
    {
      "epoch": 0.25623738393217604,
      "grad_norm": 104.009765625,
      "learning_rate": 1.7434130812891365e-05,
      "loss": 6.0834,
      "step": 6347
    },
    {
      "epoch": 0.25627775534921277,
      "grad_norm": 87.57487487792969,
      "learning_rate": 1.743325619700137e-05,
      "loss": 5.5825,
      "step": 6348
    },
    {
      "epoch": 0.2563181267662495,
      "grad_norm": 90.99625396728516,
      "learning_rate": 1.7432381454020045e-05,
      "loss": 3.8232,
      "step": 6349
    },
    {
      "epoch": 0.2563584981832862,
      "grad_norm": 81.04083251953125,
      "learning_rate": 1.743150658396236e-05,
      "loss": 6.1561,
      "step": 6350
    },
    {
      "epoch": 0.256398869600323,
      "grad_norm": 115.5518569946289,
      "learning_rate": 1.7430631586843264e-05,
      "loss": 6.3176,
      "step": 6351
    },
    {
      "epoch": 0.2564392410173597,
      "grad_norm": 139.82582092285156,
      "learning_rate": 1.742975646267772e-05,
      "loss": 6.6821,
      "step": 6352
    },
    {
      "epoch": 0.25647961243439643,
      "grad_norm": 86.15995788574219,
      "learning_rate": 1.742888121148069e-05,
      "loss": 4.2274,
      "step": 6353
    },
    {
      "epoch": 0.2565199838514332,
      "grad_norm": 97.82230377197266,
      "learning_rate": 1.742800583326714e-05,
      "loss": 4.9644,
      "step": 6354
    },
    {
      "epoch": 0.25656035526846993,
      "grad_norm": 104.88460540771484,
      "learning_rate": 1.7427130328052032e-05,
      "loss": 4.4826,
      "step": 6355
    },
    {
      "epoch": 0.25660072668550665,
      "grad_norm": 71.73296356201172,
      "learning_rate": 1.7426254695850343e-05,
      "loss": 5.4843,
      "step": 6356
    },
    {
      "epoch": 0.2566410981025434,
      "grad_norm": 80.9051284790039,
      "learning_rate": 1.742537893667704e-05,
      "loss": 5.4991,
      "step": 6357
    },
    {
      "epoch": 0.25668146951958015,
      "grad_norm": 189.16856384277344,
      "learning_rate": 1.7424503050547095e-05,
      "loss": 3.3675,
      "step": 6358
    },
    {
      "epoch": 0.2567218409366169,
      "grad_norm": 128.1820068359375,
      "learning_rate": 1.7423627037475486e-05,
      "loss": 5.6009,
      "step": 6359
    },
    {
      "epoch": 0.2567622123536536,
      "grad_norm": 122.50395202636719,
      "learning_rate": 1.7422750897477193e-05,
      "loss": 6.7957,
      "step": 6360
    },
    {
      "epoch": 0.2568025837706904,
      "grad_norm": 100.82891082763672,
      "learning_rate": 1.7421874630567188e-05,
      "loss": 6.0727,
      "step": 6361
    },
    {
      "epoch": 0.2568429551877271,
      "grad_norm": 98.23139190673828,
      "learning_rate": 1.7420998236760463e-05,
      "loss": 5.3927,
      "step": 6362
    },
    {
      "epoch": 0.2568833266047638,
      "grad_norm": 100.3313980102539,
      "learning_rate": 1.742012171607199e-05,
      "loss": 5.7873,
      "step": 6363
    },
    {
      "epoch": 0.25692369802180054,
      "grad_norm": 95.98206329345703,
      "learning_rate": 1.741924506851677e-05,
      "loss": 4.7794,
      "step": 6364
    },
    {
      "epoch": 0.2569640694388373,
      "grad_norm": 104.8271713256836,
      "learning_rate": 1.7418368294109783e-05,
      "loss": 4.7561,
      "step": 6365
    },
    {
      "epoch": 0.25700444085587404,
      "grad_norm": 102.83257293701172,
      "learning_rate": 1.741749139286602e-05,
      "loss": 4.8918,
      "step": 6366
    },
    {
      "epoch": 0.25704481227291076,
      "grad_norm": 95.45784759521484,
      "learning_rate": 1.7416614364800475e-05,
      "loss": 4.1667,
      "step": 6367
    },
    {
      "epoch": 0.25708518368994754,
      "grad_norm": 138.2539520263672,
      "learning_rate": 1.7415737209928145e-05,
      "loss": 4.1665,
      "step": 6368
    },
    {
      "epoch": 0.25712555510698426,
      "grad_norm": 128.69287109375,
      "learning_rate": 1.7414859928264026e-05,
      "loss": 4.1661,
      "step": 6369
    },
    {
      "epoch": 0.257165926524021,
      "grad_norm": 99.81011199951172,
      "learning_rate": 1.7413982519823114e-05,
      "loss": 6.3788,
      "step": 6370
    },
    {
      "epoch": 0.2572062979410577,
      "grad_norm": 86.64713287353516,
      "learning_rate": 1.7413104984620414e-05,
      "loss": 3.9602,
      "step": 6371
    },
    {
      "epoch": 0.2572466693580945,
      "grad_norm": 177.54579162597656,
      "learning_rate": 1.7412227322670933e-05,
      "loss": 3.986,
      "step": 6372
    },
    {
      "epoch": 0.2572870407751312,
      "grad_norm": 93.10669708251953,
      "learning_rate": 1.7411349533989673e-05,
      "loss": 3.597,
      "step": 6373
    },
    {
      "epoch": 0.2573274121921679,
      "grad_norm": 132.81747436523438,
      "learning_rate": 1.7410471618591644e-05,
      "loss": 4.8813,
      "step": 6374
    },
    {
      "epoch": 0.2573677836092047,
      "grad_norm": 122.04887390136719,
      "learning_rate": 1.740959357649185e-05,
      "loss": 5.95,
      "step": 6375
    },
    {
      "epoch": 0.2574081550262414,
      "grad_norm": 95.97486877441406,
      "learning_rate": 1.740871540770531e-05,
      "loss": 5.6109,
      "step": 6376
    },
    {
      "epoch": 0.25744852644327815,
      "grad_norm": 162.91285705566406,
      "learning_rate": 1.7407837112247042e-05,
      "loss": 4.9485,
      "step": 6377
    },
    {
      "epoch": 0.2574888978603149,
      "grad_norm": 100.4346694946289,
      "learning_rate": 1.7406958690132054e-05,
      "loss": 4.2853,
      "step": 6378
    },
    {
      "epoch": 0.25752926927735165,
      "grad_norm": 128.15538024902344,
      "learning_rate": 1.7406080141375367e-05,
      "loss": 3.2215,
      "step": 6379
    },
    {
      "epoch": 0.25756964069438837,
      "grad_norm": 116.54838562011719,
      "learning_rate": 1.7405201465992005e-05,
      "loss": 5.3781,
      "step": 6380
    },
    {
      "epoch": 0.2576100121114251,
      "grad_norm": 79.11579895019531,
      "learning_rate": 1.7404322663996993e-05,
      "loss": 4.9264,
      "step": 6381
    },
    {
      "epoch": 0.25765038352846187,
      "grad_norm": 159.3199920654297,
      "learning_rate": 1.7403443735405353e-05,
      "loss": 5.7675,
      "step": 6382
    },
    {
      "epoch": 0.2576907549454986,
      "grad_norm": 146.7336883544922,
      "learning_rate": 1.7402564680232112e-05,
      "loss": 5.9655,
      "step": 6383
    },
    {
      "epoch": 0.2577311263625353,
      "grad_norm": 83.13177490234375,
      "learning_rate": 1.7401685498492302e-05,
      "loss": 4.6696,
      "step": 6384
    },
    {
      "epoch": 0.2577714977795721,
      "grad_norm": 78.89291381835938,
      "learning_rate": 1.7400806190200954e-05,
      "loss": 4.882,
      "step": 6385
    },
    {
      "epoch": 0.2578118691966088,
      "grad_norm": 124.09247589111328,
      "learning_rate": 1.7399926755373103e-05,
      "loss": 5.8899,
      "step": 6386
    },
    {
      "epoch": 0.25785224061364553,
      "grad_norm": 97.16089630126953,
      "learning_rate": 1.7399047194023785e-05,
      "loss": 4.7547,
      "step": 6387
    },
    {
      "epoch": 0.25789261203068226,
      "grad_norm": 114.3879623413086,
      "learning_rate": 1.7398167506168036e-05,
      "loss": 6.2326,
      "step": 6388
    },
    {
      "epoch": 0.25793298344771903,
      "grad_norm": 102.28378295898438,
      "learning_rate": 1.73972876918209e-05,
      "loss": 4.7078,
      "step": 6389
    },
    {
      "epoch": 0.25797335486475576,
      "grad_norm": 120.67585754394531,
      "learning_rate": 1.7396407750997416e-05,
      "loss": 4.5261,
      "step": 6390
    },
    {
      "epoch": 0.2580137262817925,
      "grad_norm": 115.43397521972656,
      "learning_rate": 1.7395527683712635e-05,
      "loss": 6.1974,
      "step": 6391
    },
    {
      "epoch": 0.25805409769882925,
      "grad_norm": 137.8250274658203,
      "learning_rate": 1.7394647489981597e-05,
      "loss": 5.9897,
      "step": 6392
    },
    {
      "epoch": 0.258094469115866,
      "grad_norm": 88.85933685302734,
      "learning_rate": 1.7393767169819358e-05,
      "loss": 4.5618,
      "step": 6393
    },
    {
      "epoch": 0.2581348405329027,
      "grad_norm": 95.21077728271484,
      "learning_rate": 1.739288672324096e-05,
      "loss": 5.0496,
      "step": 6394
    },
    {
      "epoch": 0.2581752119499394,
      "grad_norm": 128.85992431640625,
      "learning_rate": 1.7392006150261468e-05,
      "loss": 5.8913,
      "step": 6395
    },
    {
      "epoch": 0.2582155833669762,
      "grad_norm": 140.8119659423828,
      "learning_rate": 1.739112545089593e-05,
      "loss": 4.8546,
      "step": 6396
    },
    {
      "epoch": 0.2582559547840129,
      "grad_norm": 103.77928161621094,
      "learning_rate": 1.7390244625159408e-05,
      "loss": 5.5783,
      "step": 6397
    },
    {
      "epoch": 0.25829632620104964,
      "grad_norm": 116.68023681640625,
      "learning_rate": 1.7389363673066964e-05,
      "loss": 6.834,
      "step": 6398
    },
    {
      "epoch": 0.2583366976180864,
      "grad_norm": 110.11788940429688,
      "learning_rate": 1.738848259463365e-05,
      "loss": 4.4598,
      "step": 6399
    },
    {
      "epoch": 0.25837706903512314,
      "grad_norm": 71.26177215576172,
      "learning_rate": 1.7387601389874543e-05,
      "loss": 5.9944,
      "step": 6400
    },
    {
      "epoch": 0.25841744045215986,
      "grad_norm": 141.6345977783203,
      "learning_rate": 1.73867200588047e-05,
      "loss": 6.3468,
      "step": 6401
    },
    {
      "epoch": 0.2584578118691966,
      "grad_norm": 87.94096374511719,
      "learning_rate": 1.7385838601439193e-05,
      "loss": 3.3396,
      "step": 6402
    },
    {
      "epoch": 0.25849818328623336,
      "grad_norm": 115.21498107910156,
      "learning_rate": 1.7384957017793095e-05,
      "loss": 4.9668,
      "step": 6403
    },
    {
      "epoch": 0.2585385547032701,
      "grad_norm": 133.4326934814453,
      "learning_rate": 1.7384075307881474e-05,
      "loss": 5.2794,
      "step": 6404
    },
    {
      "epoch": 0.2585789261203068,
      "grad_norm": 132.0460968017578,
      "learning_rate": 1.738319347171941e-05,
      "loss": 6.3963,
      "step": 6405
    },
    {
      "epoch": 0.2586192975373436,
      "grad_norm": 97.84303283691406,
      "learning_rate": 1.738231150932198e-05,
      "loss": 5.2831,
      "step": 6406
    },
    {
      "epoch": 0.2586596689543803,
      "grad_norm": 122.96819305419922,
      "learning_rate": 1.7381429420704255e-05,
      "loss": 4.7286,
      "step": 6407
    },
    {
      "epoch": 0.25870004037141703,
      "grad_norm": 86.68722534179688,
      "learning_rate": 1.7380547205881332e-05,
      "loss": 6.5147,
      "step": 6408
    },
    {
      "epoch": 0.25874041178845375,
      "grad_norm": 97.358642578125,
      "learning_rate": 1.7379664864868284e-05,
      "loss": 5.0801,
      "step": 6409
    },
    {
      "epoch": 0.2587807832054905,
      "grad_norm": 199.8591766357422,
      "learning_rate": 1.73787823976802e-05,
      "loss": 4.7051,
      "step": 6410
    },
    {
      "epoch": 0.25882115462252725,
      "grad_norm": 82.265380859375,
      "learning_rate": 1.737789980433217e-05,
      "loss": 4.6765,
      "step": 6411
    },
    {
      "epoch": 0.25886152603956397,
      "grad_norm": 83.60492706298828,
      "learning_rate": 1.7377017084839277e-05,
      "loss": 4.929,
      "step": 6412
    },
    {
      "epoch": 0.25890189745660075,
      "grad_norm": 99.62299346923828,
      "learning_rate": 1.737613423921662e-05,
      "loss": 5.6819,
      "step": 6413
    },
    {
      "epoch": 0.25894226887363747,
      "grad_norm": 71.1746826171875,
      "learning_rate": 1.7375251267479295e-05,
      "loss": 5.8802,
      "step": 6414
    },
    {
      "epoch": 0.2589826402906742,
      "grad_norm": 61.79850387573242,
      "learning_rate": 1.7374368169642393e-05,
      "loss": 5.1839,
      "step": 6415
    },
    {
      "epoch": 0.2590230117077109,
      "grad_norm": 70.75880432128906,
      "learning_rate": 1.7373484945721015e-05,
      "loss": 4.485,
      "step": 6416
    },
    {
      "epoch": 0.2590633831247477,
      "grad_norm": 89.91203308105469,
      "learning_rate": 1.7372601595730262e-05,
      "loss": 5.3344,
      "step": 6417
    },
    {
      "epoch": 0.2591037545417844,
      "grad_norm": 108.64740753173828,
      "learning_rate": 1.737171811968524e-05,
      "loss": 5.8243,
      "step": 6418
    },
    {
      "epoch": 0.25914412595882114,
      "grad_norm": 90.0994644165039,
      "learning_rate": 1.737083451760105e-05,
      "loss": 5.7103,
      "step": 6419
    },
    {
      "epoch": 0.2591844973758579,
      "grad_norm": 253.06961059570312,
      "learning_rate": 1.7369950789492808e-05,
      "loss": 4.4901,
      "step": 6420
    },
    {
      "epoch": 0.25922486879289464,
      "grad_norm": 127.89641571044922,
      "learning_rate": 1.7369066935375614e-05,
      "loss": 6.652,
      "step": 6421
    },
    {
      "epoch": 0.25926524020993136,
      "grad_norm": 94.62724304199219,
      "learning_rate": 1.7368182955264583e-05,
      "loss": 6.9952,
      "step": 6422
    },
    {
      "epoch": 0.2593056116269681,
      "grad_norm": 129.3190460205078,
      "learning_rate": 1.736729884917483e-05,
      "loss": 5.5523,
      "step": 6423
    },
    {
      "epoch": 0.25934598304400486,
      "grad_norm": 102.39450073242188,
      "learning_rate": 1.7366414617121468e-05,
      "loss": 4.9199,
      "step": 6424
    },
    {
      "epoch": 0.2593863544610416,
      "grad_norm": 156.5802459716797,
      "learning_rate": 1.736553025911962e-05,
      "loss": 5.7958,
      "step": 6425
    },
    {
      "epoch": 0.2594267258780783,
      "grad_norm": 77.25160217285156,
      "learning_rate": 1.7364645775184407e-05,
      "loss": 5.0586,
      "step": 6426
    },
    {
      "epoch": 0.2594670972951151,
      "grad_norm": 87.84078216552734,
      "learning_rate": 1.7363761165330943e-05,
      "loss": 5.4687,
      "step": 6427
    },
    {
      "epoch": 0.2595074687121518,
      "grad_norm": 103.31646728515625,
      "learning_rate": 1.7362876429574362e-05,
      "loss": 5.6897,
      "step": 6428
    },
    {
      "epoch": 0.2595478401291885,
      "grad_norm": 114.00814819335938,
      "learning_rate": 1.7361991567929787e-05,
      "loss": 6.7003,
      "step": 6429
    },
    {
      "epoch": 0.25958821154622524,
      "grad_norm": 107.8480453491211,
      "learning_rate": 1.736110658041235e-05,
      "loss": 5.6317,
      "step": 6430
    },
    {
      "epoch": 0.259628582963262,
      "grad_norm": 90.03013610839844,
      "learning_rate": 1.736022146703718e-05,
      "loss": 5.4505,
      "step": 6431
    },
    {
      "epoch": 0.25966895438029874,
      "grad_norm": 87.48660278320312,
      "learning_rate": 1.7359336227819405e-05,
      "loss": 5.0233,
      "step": 6432
    },
    {
      "epoch": 0.25970932579733547,
      "grad_norm": 118.88618469238281,
      "learning_rate": 1.7358450862774173e-05,
      "loss": 4.4937,
      "step": 6433
    },
    {
      "epoch": 0.25974969721437224,
      "grad_norm": 97.34734344482422,
      "learning_rate": 1.7357565371916608e-05,
      "loss": 3.3607,
      "step": 6434
    },
    {
      "epoch": 0.25979006863140897,
      "grad_norm": 68.2127685546875,
      "learning_rate": 1.735667975526186e-05,
      "loss": 5.6482,
      "step": 6435
    },
    {
      "epoch": 0.2598304400484457,
      "grad_norm": 78.80931854248047,
      "learning_rate": 1.7355794012825066e-05,
      "loss": 5.3795,
      "step": 6436
    },
    {
      "epoch": 0.25987081146548247,
      "grad_norm": 91.45195770263672,
      "learning_rate": 1.735490814462137e-05,
      "loss": 4.6246,
      "step": 6437
    },
    {
      "epoch": 0.2599111828825192,
      "grad_norm": 79.68939971923828,
      "learning_rate": 1.7354022150665925e-05,
      "loss": 5.9785,
      "step": 6438
    },
    {
      "epoch": 0.2599515542995559,
      "grad_norm": 113.82240295410156,
      "learning_rate": 1.7353136030973867e-05,
      "loss": 5.3061,
      "step": 6439
    },
    {
      "epoch": 0.25999192571659263,
      "grad_norm": 166.7451171875,
      "learning_rate": 1.7352249785560355e-05,
      "loss": 5.0633,
      "step": 6440
    },
    {
      "epoch": 0.2600322971336294,
      "grad_norm": 122.62030792236328,
      "learning_rate": 1.7351363414440546e-05,
      "loss": 6.5894,
      "step": 6441
    },
    {
      "epoch": 0.26007266855066613,
      "grad_norm": 135.0858917236328,
      "learning_rate": 1.7350476917629583e-05,
      "loss": 3.927,
      "step": 6442
    },
    {
      "epoch": 0.26011303996770285,
      "grad_norm": 84.16254425048828,
      "learning_rate": 1.7349590295142637e-05,
      "loss": 5.6114,
      "step": 6443
    },
    {
      "epoch": 0.26015341138473963,
      "grad_norm": 102.7471923828125,
      "learning_rate": 1.734870354699485e-05,
      "loss": 4.9813,
      "step": 6444
    },
    {
      "epoch": 0.26019378280177635,
      "grad_norm": 97.87606811523438,
      "learning_rate": 1.7347816673201397e-05,
      "loss": 5.4403,
      "step": 6445
    },
    {
      "epoch": 0.2602341542188131,
      "grad_norm": 91.11294555664062,
      "learning_rate": 1.7346929673777438e-05,
      "loss": 4.096,
      "step": 6446
    },
    {
      "epoch": 0.2602745256358498,
      "grad_norm": 202.56375122070312,
      "learning_rate": 1.734604254873814e-05,
      "loss": 4.7384,
      "step": 6447
    },
    {
      "epoch": 0.2603148970528866,
      "grad_norm": 93.98361206054688,
      "learning_rate": 1.7345155298098665e-05,
      "loss": 4.8276,
      "step": 6448
    },
    {
      "epoch": 0.2603552684699233,
      "grad_norm": 95.86788177490234,
      "learning_rate": 1.7344267921874186e-05,
      "loss": 4.6155,
      "step": 6449
    },
    {
      "epoch": 0.26039563988696,
      "grad_norm": 91.29017639160156,
      "learning_rate": 1.7343380420079877e-05,
      "loss": 5.3539,
      "step": 6450
    },
    {
      "epoch": 0.2604360113039968,
      "grad_norm": 65.64302062988281,
      "learning_rate": 1.7342492792730912e-05,
      "loss": 4.0409,
      "step": 6451
    },
    {
      "epoch": 0.2604763827210335,
      "grad_norm": 100.54818725585938,
      "learning_rate": 1.7341605039842465e-05,
      "loss": 5.3633,
      "step": 6452
    },
    {
      "epoch": 0.26051675413807024,
      "grad_norm": 106.15442657470703,
      "learning_rate": 1.7340717161429716e-05,
      "loss": 4.9281,
      "step": 6453
    },
    {
      "epoch": 0.26055712555510696,
      "grad_norm": 99.32792663574219,
      "learning_rate": 1.7339829157507846e-05,
      "loss": 3.9634,
      "step": 6454
    },
    {
      "epoch": 0.26059749697214374,
      "grad_norm": 111.11714172363281,
      "learning_rate": 1.7338941028092037e-05,
      "loss": 5.9628,
      "step": 6455
    },
    {
      "epoch": 0.26063786838918046,
      "grad_norm": 100.44255065917969,
      "learning_rate": 1.7338052773197473e-05,
      "loss": 6.1709,
      "step": 6456
    },
    {
      "epoch": 0.2606782398062172,
      "grad_norm": 99.33821868896484,
      "learning_rate": 1.733716439283934e-05,
      "loss": 5.3246,
      "step": 6457
    },
    {
      "epoch": 0.26071861122325396,
      "grad_norm": 90.05223846435547,
      "learning_rate": 1.7336275887032835e-05,
      "loss": 4.6001,
      "step": 6458
    },
    {
      "epoch": 0.2607589826402907,
      "grad_norm": 113.90547943115234,
      "learning_rate": 1.733538725579314e-05,
      "loss": 6.0143,
      "step": 6459
    },
    {
      "epoch": 0.2607993540573274,
      "grad_norm": 87.71128845214844,
      "learning_rate": 1.733449849913545e-05,
      "loss": 5.193,
      "step": 6460
    },
    {
      "epoch": 0.2608397254743641,
      "grad_norm": 110.80097961425781,
      "learning_rate": 1.7333609617074972e-05,
      "loss": 6.1801,
      "step": 6461
    },
    {
      "epoch": 0.2608800968914009,
      "grad_norm": 81.83992004394531,
      "learning_rate": 1.7332720609626887e-05,
      "loss": 5.4982,
      "step": 6462
    },
    {
      "epoch": 0.2609204683084376,
      "grad_norm": 131.26402282714844,
      "learning_rate": 1.7331831476806407e-05,
      "loss": 5.2107,
      "step": 6463
    },
    {
      "epoch": 0.26096083972547435,
      "grad_norm": 112.12376403808594,
      "learning_rate": 1.7330942218628726e-05,
      "loss": 5.7546,
      "step": 6464
    },
    {
      "epoch": 0.2610012111425111,
      "grad_norm": 172.5349884033203,
      "learning_rate": 1.733005283510905e-05,
      "loss": 4.524,
      "step": 6465
    },
    {
      "epoch": 0.26104158255954785,
      "grad_norm": 150.8626251220703,
      "learning_rate": 1.7329163326262596e-05,
      "loss": 5.4043,
      "step": 6466
    },
    {
      "epoch": 0.26108195397658457,
      "grad_norm": 161.6379852294922,
      "learning_rate": 1.7328273692104557e-05,
      "loss": 6.709,
      "step": 6467
    },
    {
      "epoch": 0.2611223253936213,
      "grad_norm": 83.46670532226562,
      "learning_rate": 1.7327383932650153e-05,
      "loss": 6.3074,
      "step": 6468
    },
    {
      "epoch": 0.26116269681065807,
      "grad_norm": 92.94638061523438,
      "learning_rate": 1.732649404791459e-05,
      "loss": 3.9983,
      "step": 6469
    },
    {
      "epoch": 0.2612030682276948,
      "grad_norm": 76.992919921875,
      "learning_rate": 1.7325604037913093e-05,
      "loss": 5.7952,
      "step": 6470
    },
    {
      "epoch": 0.2612434396447315,
      "grad_norm": 75.04573822021484,
      "learning_rate": 1.7324713902660873e-05,
      "loss": 6.6179,
      "step": 6471
    },
    {
      "epoch": 0.2612838110617683,
      "grad_norm": 123.21736145019531,
      "learning_rate": 1.732382364217315e-05,
      "loss": 7.324,
      "step": 6472
    },
    {
      "epoch": 0.261324182478805,
      "grad_norm": 124.05339050292969,
      "learning_rate": 1.732293325646514e-05,
      "loss": 4.8369,
      "step": 6473
    },
    {
      "epoch": 0.26136455389584173,
      "grad_norm": 65.03087615966797,
      "learning_rate": 1.7322042745552077e-05,
      "loss": 4.3071,
      "step": 6474
    },
    {
      "epoch": 0.26140492531287846,
      "grad_norm": 99.01994323730469,
      "learning_rate": 1.732115210944918e-05,
      "loss": 4.7197,
      "step": 6475
    },
    {
      "epoch": 0.26144529672991523,
      "grad_norm": 63.92416763305664,
      "learning_rate": 1.7320261348171677e-05,
      "loss": 5.6902,
      "step": 6476
    },
    {
      "epoch": 0.26148566814695196,
      "grad_norm": 103.4669418334961,
      "learning_rate": 1.73193704617348e-05,
      "loss": 4.8599,
      "step": 6477
    },
    {
      "epoch": 0.2615260395639887,
      "grad_norm": 65.54827880859375,
      "learning_rate": 1.731847945015378e-05,
      "loss": 5.1484,
      "step": 6478
    },
    {
      "epoch": 0.26156641098102545,
      "grad_norm": 91.8048095703125,
      "learning_rate": 1.731758831344385e-05,
      "loss": 4.879,
      "step": 6479
    },
    {
      "epoch": 0.2616067823980622,
      "grad_norm": 70.17029571533203,
      "learning_rate": 1.731669705162025e-05,
      "loss": 3.4871,
      "step": 6480
    },
    {
      "epoch": 0.2616471538150989,
      "grad_norm": 107.4705810546875,
      "learning_rate": 1.731580566469821e-05,
      "loss": 4.9313,
      "step": 6481
    },
    {
      "epoch": 0.2616875252321356,
      "grad_norm": 101.49443817138672,
      "learning_rate": 1.7314914152692984e-05,
      "loss": 3.7084,
      "step": 6482
    },
    {
      "epoch": 0.2617278966491724,
      "grad_norm": 100.78109741210938,
      "learning_rate": 1.7314022515619805e-05,
      "loss": 4.4175,
      "step": 6483
    },
    {
      "epoch": 0.2617682680662091,
      "grad_norm": 181.7901153564453,
      "learning_rate": 1.7313130753493917e-05,
      "loss": 4.7455,
      "step": 6484
    },
    {
      "epoch": 0.26180863948324584,
      "grad_norm": 195.64602661132812,
      "learning_rate": 1.7312238866330578e-05,
      "loss": 3.84,
      "step": 6485
    },
    {
      "epoch": 0.2618490109002826,
      "grad_norm": 116.86393737792969,
      "learning_rate": 1.7311346854145023e-05,
      "loss": 5.3955,
      "step": 6486
    },
    {
      "epoch": 0.26188938231731934,
      "grad_norm": 69.9563980102539,
      "learning_rate": 1.7310454716952512e-05,
      "loss": 4.1689,
      "step": 6487
    },
    {
      "epoch": 0.26192975373435606,
      "grad_norm": 84.59313201904297,
      "learning_rate": 1.7309562454768297e-05,
      "loss": 4.0819,
      "step": 6488
    },
    {
      "epoch": 0.26197012515139284,
      "grad_norm": 96.73358917236328,
      "learning_rate": 1.7308670067607636e-05,
      "loss": 5.1552,
      "step": 6489
    },
    {
      "epoch": 0.26201049656842956,
      "grad_norm": 176.478759765625,
      "learning_rate": 1.730777755548578e-05,
      "loss": 4.2956,
      "step": 6490
    },
    {
      "epoch": 0.2620508679854663,
      "grad_norm": 90.27690124511719,
      "learning_rate": 1.7306884918417992e-05,
      "loss": 4.5674,
      "step": 6491
    },
    {
      "epoch": 0.262091239402503,
      "grad_norm": 74.45086669921875,
      "learning_rate": 1.7305992156419536e-05,
      "loss": 5.5362,
      "step": 6492
    },
    {
      "epoch": 0.2621316108195398,
      "grad_norm": 94.7879867553711,
      "learning_rate": 1.7305099269505676e-05,
      "loss": 5.8255,
      "step": 6493
    },
    {
      "epoch": 0.2621719822365765,
      "grad_norm": 67.22917938232422,
      "learning_rate": 1.7304206257691677e-05,
      "loss": 4.8352,
      "step": 6494
    },
    {
      "epoch": 0.26221235365361323,
      "grad_norm": 98.34140014648438,
      "learning_rate": 1.7303313120992805e-05,
      "loss": 5.385,
      "step": 6495
    },
    {
      "epoch": 0.26225272507065,
      "grad_norm": 92.91995239257812,
      "learning_rate": 1.7302419859424335e-05,
      "loss": 6.32,
      "step": 6496
    },
    {
      "epoch": 0.2622930964876867,
      "grad_norm": 101.4473876953125,
      "learning_rate": 1.730152647300154e-05,
      "loss": 6.0745,
      "step": 6497
    },
    {
      "epoch": 0.26233346790472345,
      "grad_norm": 109.65857696533203,
      "learning_rate": 1.730063296173969e-05,
      "loss": 4.5463,
      "step": 6498
    },
    {
      "epoch": 0.26237383932176017,
      "grad_norm": 128.3776397705078,
      "learning_rate": 1.7299739325654064e-05,
      "loss": 5.3204,
      "step": 6499
    },
    {
      "epoch": 0.26241421073879695,
      "grad_norm": 117.93672943115234,
      "learning_rate": 1.729884556475994e-05,
      "loss": 6.0544,
      "step": 6500
    },
    {
      "epoch": 0.26245458215583367,
      "grad_norm": 117.28588104248047,
      "learning_rate": 1.7297951679072604e-05,
      "loss": 4.9561,
      "step": 6501
    },
    {
      "epoch": 0.2624949535728704,
      "grad_norm": 198.29917907714844,
      "learning_rate": 1.7297057668607337e-05,
      "loss": 3.2637,
      "step": 6502
    },
    {
      "epoch": 0.26253532498990717,
      "grad_norm": 120.10796356201172,
      "learning_rate": 1.7296163533379423e-05,
      "loss": 7.0465,
      "step": 6503
    },
    {
      "epoch": 0.2625756964069439,
      "grad_norm": 119.55882263183594,
      "learning_rate": 1.7295269273404148e-05,
      "loss": 4.9333,
      "step": 6504
    },
    {
      "epoch": 0.2626160678239806,
      "grad_norm": 140.17152404785156,
      "learning_rate": 1.7294374888696808e-05,
      "loss": 6.1337,
      "step": 6505
    },
    {
      "epoch": 0.26265643924101734,
      "grad_norm": 100.84590148925781,
      "learning_rate": 1.729348037927269e-05,
      "loss": 4.9381,
      "step": 6506
    },
    {
      "epoch": 0.2626968106580541,
      "grad_norm": 81.5610580444336,
      "learning_rate": 1.7292585745147084e-05,
      "loss": 4.1757,
      "step": 6507
    },
    {
      "epoch": 0.26273718207509084,
      "grad_norm": 109.10648345947266,
      "learning_rate": 1.7291690986335296e-05,
      "loss": 4.3813,
      "step": 6508
    },
    {
      "epoch": 0.26277755349212756,
      "grad_norm": 128.10386657714844,
      "learning_rate": 1.7290796102852615e-05,
      "loss": 5.8385,
      "step": 6509
    },
    {
      "epoch": 0.26281792490916434,
      "grad_norm": 99.71379089355469,
      "learning_rate": 1.728990109471435e-05,
      "loss": 6.3894,
      "step": 6510
    },
    {
      "epoch": 0.26285829632620106,
      "grad_norm": 80.94365692138672,
      "learning_rate": 1.72890059619358e-05,
      "loss": 5.9361,
      "step": 6511
    },
    {
      "epoch": 0.2628986677432378,
      "grad_norm": 92.27545928955078,
      "learning_rate": 1.7288110704532267e-05,
      "loss": 5.8881,
      "step": 6512
    },
    {
      "epoch": 0.2629390391602745,
      "grad_norm": 81.315673828125,
      "learning_rate": 1.728721532251906e-05,
      "loss": 4.7462,
      "step": 6513
    },
    {
      "epoch": 0.2629794105773113,
      "grad_norm": 80.4803237915039,
      "learning_rate": 1.728631981591149e-05,
      "loss": 5.1255,
      "step": 6514
    },
    {
      "epoch": 0.263019781994348,
      "grad_norm": 88.64851379394531,
      "learning_rate": 1.7285424184724866e-05,
      "loss": 4.5747,
      "step": 6515
    },
    {
      "epoch": 0.2630601534113847,
      "grad_norm": 112.28235626220703,
      "learning_rate": 1.72845284289745e-05,
      "loss": 4.2937,
      "step": 6516
    },
    {
      "epoch": 0.2631005248284215,
      "grad_norm": 72.07464599609375,
      "learning_rate": 1.7283632548675708e-05,
      "loss": 6.6414,
      "step": 6517
    },
    {
      "epoch": 0.2631408962454582,
      "grad_norm": 65.63872528076172,
      "learning_rate": 1.7282736543843807e-05,
      "loss": 6.0978,
      "step": 6518
    },
    {
      "epoch": 0.26318126766249494,
      "grad_norm": 137.44888305664062,
      "learning_rate": 1.7281840414494116e-05,
      "loss": 5.6145,
      "step": 6519
    },
    {
      "epoch": 0.26322163907953167,
      "grad_norm": 218.57933044433594,
      "learning_rate": 1.7280944160641963e-05,
      "loss": 5.0681,
      "step": 6520
    },
    {
      "epoch": 0.26326201049656844,
      "grad_norm": 121.52847290039062,
      "learning_rate": 1.7280047782302665e-05,
      "loss": 4.7881,
      "step": 6521
    },
    {
      "epoch": 0.26330238191360517,
      "grad_norm": 94.34153747558594,
      "learning_rate": 1.727915127949155e-05,
      "loss": 6.1395,
      "step": 6522
    },
    {
      "epoch": 0.2633427533306419,
      "grad_norm": 80.98170471191406,
      "learning_rate": 1.7278254652223946e-05,
      "loss": 5.8922,
      "step": 6523
    },
    {
      "epoch": 0.26338312474767867,
      "grad_norm": 85.73329162597656,
      "learning_rate": 1.727735790051518e-05,
      "loss": 6.0764,
      "step": 6524
    },
    {
      "epoch": 0.2634234961647154,
      "grad_norm": 114.87275695800781,
      "learning_rate": 1.7276461024380596e-05,
      "loss": 6.4107,
      "step": 6525
    },
    {
      "epoch": 0.2634638675817521,
      "grad_norm": 411.9471740722656,
      "learning_rate": 1.7275564023835513e-05,
      "loss": 5.0588,
      "step": 6526
    },
    {
      "epoch": 0.26350423899878883,
      "grad_norm": 86.50202178955078,
      "learning_rate": 1.7274666898895277e-05,
      "loss": 6.3128,
      "step": 6527
    },
    {
      "epoch": 0.2635446104158256,
      "grad_norm": 98.5763931274414,
      "learning_rate": 1.7273769649575224e-05,
      "loss": 6.8604,
      "step": 6528
    },
    {
      "epoch": 0.26358498183286233,
      "grad_norm": 97.32089233398438,
      "learning_rate": 1.7272872275890698e-05,
      "loss": 5.0817,
      "step": 6529
    },
    {
      "epoch": 0.26362535324989905,
      "grad_norm": 107.06633758544922,
      "learning_rate": 1.727197477785704e-05,
      "loss": 5.7083,
      "step": 6530
    },
    {
      "epoch": 0.26366572466693583,
      "grad_norm": 91.51175689697266,
      "learning_rate": 1.727107715548959e-05,
      "loss": 5.0567,
      "step": 6531
    },
    {
      "epoch": 0.26370609608397255,
      "grad_norm": 146.5870361328125,
      "learning_rate": 1.7270179408803703e-05,
      "loss": 3.9876,
      "step": 6532
    },
    {
      "epoch": 0.2637464675010093,
      "grad_norm": 144.1085662841797,
      "learning_rate": 1.7269281537814722e-05,
      "loss": 6.7892,
      "step": 6533
    },
    {
      "epoch": 0.263786838918046,
      "grad_norm": 82.8897476196289,
      "learning_rate": 1.7268383542538005e-05,
      "loss": 4.8661,
      "step": 6534
    },
    {
      "epoch": 0.2638272103350828,
      "grad_norm": 180.43191528320312,
      "learning_rate": 1.72674854229889e-05,
      "loss": 4.8528,
      "step": 6535
    },
    {
      "epoch": 0.2638675817521195,
      "grad_norm": 93.56403350830078,
      "learning_rate": 1.726658717918277e-05,
      "loss": 6.4003,
      "step": 6536
    },
    {
      "epoch": 0.2639079531691562,
      "grad_norm": 64.31196594238281,
      "learning_rate": 1.7265688811134963e-05,
      "loss": 4.3815,
      "step": 6537
    },
    {
      "epoch": 0.263948324586193,
      "grad_norm": 99.08624267578125,
      "learning_rate": 1.7264790318860844e-05,
      "loss": 6.1013,
      "step": 6538
    },
    {
      "epoch": 0.2639886960032297,
      "grad_norm": 86.47508239746094,
      "learning_rate": 1.7263891702375777e-05,
      "loss": 3.0086,
      "step": 6539
    },
    {
      "epoch": 0.26402906742026644,
      "grad_norm": 72.9893569946289,
      "learning_rate": 1.7262992961695125e-05,
      "loss": 6.2926,
      "step": 6540
    },
    {
      "epoch": 0.2640694388373032,
      "grad_norm": 70.19054412841797,
      "learning_rate": 1.726209409683425e-05,
      "loss": 6.5492,
      "step": 6541
    },
    {
      "epoch": 0.26410981025433994,
      "grad_norm": 133.58119201660156,
      "learning_rate": 1.7261195107808527e-05,
      "loss": 5.6397,
      "step": 6542
    },
    {
      "epoch": 0.26415018167137666,
      "grad_norm": 128.60044860839844,
      "learning_rate": 1.7260295994633325e-05,
      "loss": 5.1695,
      "step": 6543
    },
    {
      "epoch": 0.2641905530884134,
      "grad_norm": 112.22277069091797,
      "learning_rate": 1.725939675732401e-05,
      "loss": 4.2341,
      "step": 6544
    },
    {
      "epoch": 0.26423092450545016,
      "grad_norm": 89.08710479736328,
      "learning_rate": 1.725849739589597e-05,
      "loss": 4.7744,
      "step": 6545
    },
    {
      "epoch": 0.2642712959224869,
      "grad_norm": 84.37786865234375,
      "learning_rate": 1.7257597910364568e-05,
      "loss": 4.8432,
      "step": 6546
    },
    {
      "epoch": 0.2643116673395236,
      "grad_norm": 108.58320617675781,
      "learning_rate": 1.7256698300745192e-05,
      "loss": 3.9736,
      "step": 6547
    },
    {
      "epoch": 0.2643520387565604,
      "grad_norm": 115.86542510986328,
      "learning_rate": 1.7255798567053223e-05,
      "loss": 5.7582,
      "step": 6548
    },
    {
      "epoch": 0.2643924101735971,
      "grad_norm": 426.8697204589844,
      "learning_rate": 1.725489870930404e-05,
      "loss": 4.653,
      "step": 6549
    },
    {
      "epoch": 0.2644327815906338,
      "grad_norm": 130.5552978515625,
      "learning_rate": 1.725399872751303e-05,
      "loss": 5.0517,
      "step": 6550
    },
    {
      "epoch": 0.26447315300767055,
      "grad_norm": 89.51067352294922,
      "learning_rate": 1.725309862169558e-05,
      "loss": 7.267,
      "step": 6551
    },
    {
      "epoch": 0.2645135244247073,
      "grad_norm": 131.14996337890625,
      "learning_rate": 1.7252198391867083e-05,
      "loss": 6.3835,
      "step": 6552
    },
    {
      "epoch": 0.26455389584174405,
      "grad_norm": 187.4968719482422,
      "learning_rate": 1.7251298038042926e-05,
      "loss": 5.683,
      "step": 6553
    },
    {
      "epoch": 0.26459426725878077,
      "grad_norm": 207.79200744628906,
      "learning_rate": 1.7250397560238506e-05,
      "loss": 6.4113,
      "step": 6554
    },
    {
      "epoch": 0.26463463867581755,
      "grad_norm": 88.46598815917969,
      "learning_rate": 1.7249496958469222e-05,
      "loss": 5.3989,
      "step": 6555
    },
    {
      "epoch": 0.26467501009285427,
      "grad_norm": 67.58189392089844,
      "learning_rate": 1.7248596232750466e-05,
      "loss": 5.3697,
      "step": 6556
    },
    {
      "epoch": 0.264715381509891,
      "grad_norm": 73.73224639892578,
      "learning_rate": 1.724769538309764e-05,
      "loss": 6.0385,
      "step": 6557
    },
    {
      "epoch": 0.2647557529269277,
      "grad_norm": 142.81451416015625,
      "learning_rate": 1.724679440952615e-05,
      "loss": 6.2703,
      "step": 6558
    },
    {
      "epoch": 0.2647961243439645,
      "grad_norm": 108.37516784667969,
      "learning_rate": 1.7245893312051398e-05,
      "loss": 5.2796,
      "step": 6559
    },
    {
      "epoch": 0.2648364957610012,
      "grad_norm": 121.24029541015625,
      "learning_rate": 1.7244992090688792e-05,
      "loss": 5.2563,
      "step": 6560
    },
    {
      "epoch": 0.26487686717803793,
      "grad_norm": 99.15504455566406,
      "learning_rate": 1.7244090745453733e-05,
      "loss": 6.0249,
      "step": 6561
    },
    {
      "epoch": 0.2649172385950747,
      "grad_norm": 100.36276245117188,
      "learning_rate": 1.7243189276361644e-05,
      "loss": 6.0766,
      "step": 6562
    },
    {
      "epoch": 0.26495761001211143,
      "grad_norm": 108.97128295898438,
      "learning_rate": 1.724228768342793e-05,
      "loss": 6.1554,
      "step": 6563
    },
    {
      "epoch": 0.26499798142914816,
      "grad_norm": 145.5115966796875,
      "learning_rate": 1.724138596666801e-05,
      "loss": 4.364,
      "step": 6564
    },
    {
      "epoch": 0.2650383528461849,
      "grad_norm": 96.40485382080078,
      "learning_rate": 1.7240484126097302e-05,
      "loss": 6.0039,
      "step": 6565
    },
    {
      "epoch": 0.26507872426322165,
      "grad_norm": 81.06026458740234,
      "learning_rate": 1.7239582161731218e-05,
      "loss": 4.1301,
      "step": 6566
    },
    {
      "epoch": 0.2651190956802584,
      "grad_norm": 143.96905517578125,
      "learning_rate": 1.7238680073585188e-05,
      "loss": 5.7246,
      "step": 6567
    },
    {
      "epoch": 0.2651594670972951,
      "grad_norm": 86.83251190185547,
      "learning_rate": 1.7237777861674633e-05,
      "loss": 6.9045,
      "step": 6568
    },
    {
      "epoch": 0.2651998385143319,
      "grad_norm": 90.86653137207031,
      "learning_rate": 1.7236875526014974e-05,
      "loss": 4.0287,
      "step": 6569
    },
    {
      "epoch": 0.2652402099313686,
      "grad_norm": 93.13839721679688,
      "learning_rate": 1.723597306662165e-05,
      "loss": 5.5846,
      "step": 6570
    },
    {
      "epoch": 0.2652805813484053,
      "grad_norm": 133.9046630859375,
      "learning_rate": 1.7235070483510077e-05,
      "loss": 5.4965,
      "step": 6571
    },
    {
      "epoch": 0.26532095276544204,
      "grad_norm": 129.64401245117188,
      "learning_rate": 1.7234167776695695e-05,
      "loss": 5.6948,
      "step": 6572
    },
    {
      "epoch": 0.2653613241824788,
      "grad_norm": 143.88259887695312,
      "learning_rate": 1.723326494619394e-05,
      "loss": 4.3628,
      "step": 6573
    },
    {
      "epoch": 0.26540169559951554,
      "grad_norm": 108.58419799804688,
      "learning_rate": 1.7232361992020242e-05,
      "loss": 4.852,
      "step": 6574
    },
    {
      "epoch": 0.26544206701655226,
      "grad_norm": 92.37435150146484,
      "learning_rate": 1.7231458914190047e-05,
      "loss": 4.0584,
      "step": 6575
    },
    {
      "epoch": 0.26548243843358904,
      "grad_norm": 73.34140014648438,
      "learning_rate": 1.7230555712718788e-05,
      "loss": 5.2659,
      "step": 6576
    },
    {
      "epoch": 0.26552280985062576,
      "grad_norm": 76.83888244628906,
      "learning_rate": 1.722965238762191e-05,
      "loss": 5.4749,
      "step": 6577
    },
    {
      "epoch": 0.2655631812676625,
      "grad_norm": 81.85470581054688,
      "learning_rate": 1.722874893891486e-05,
      "loss": 5.2112,
      "step": 6578
    },
    {
      "epoch": 0.2656035526846992,
      "grad_norm": 102.47541809082031,
      "learning_rate": 1.7227845366613086e-05,
      "loss": 5.2075,
      "step": 6579
    },
    {
      "epoch": 0.265643924101736,
      "grad_norm": 105.2646255493164,
      "learning_rate": 1.7226941670732033e-05,
      "loss": 5.0269,
      "step": 6580
    },
    {
      "epoch": 0.2656842955187727,
      "grad_norm": 108.20997619628906,
      "learning_rate": 1.722603785128715e-05,
      "loss": 5.3294,
      "step": 6581
    },
    {
      "epoch": 0.26572466693580943,
      "grad_norm": 186.04037475585938,
      "learning_rate": 1.7225133908293896e-05,
      "loss": 6.7455,
      "step": 6582
    },
    {
      "epoch": 0.2657650383528462,
      "grad_norm": 122.20954132080078,
      "learning_rate": 1.7224229841767726e-05,
      "loss": 4.4681,
      "step": 6583
    },
    {
      "epoch": 0.2658054097698829,
      "grad_norm": 89.92524719238281,
      "learning_rate": 1.7223325651724094e-05,
      "loss": 4.6579,
      "step": 6584
    },
    {
      "epoch": 0.26584578118691965,
      "grad_norm": 113.41470336914062,
      "learning_rate": 1.722242133817846e-05,
      "loss": 5.9053,
      "step": 6585
    },
    {
      "epoch": 0.26588615260395637,
      "grad_norm": 116.1080093383789,
      "learning_rate": 1.7221516901146286e-05,
      "loss": 4.6984,
      "step": 6586
    },
    {
      "epoch": 0.26592652402099315,
      "grad_norm": 87.26712799072266,
      "learning_rate": 1.722061234064304e-05,
      "loss": 4.9207,
      "step": 6587
    },
    {
      "epoch": 0.26596689543802987,
      "grad_norm": 102.5451431274414,
      "learning_rate": 1.7219707656684183e-05,
      "loss": 5.4983,
      "step": 6588
    },
    {
      "epoch": 0.2660072668550666,
      "grad_norm": 79.96680450439453,
      "learning_rate": 1.7218802849285185e-05,
      "loss": 6.3943,
      "step": 6589
    },
    {
      "epoch": 0.26604763827210337,
      "grad_norm": 88.04561614990234,
      "learning_rate": 1.7217897918461513e-05,
      "loss": 5.1407,
      "step": 6590
    },
    {
      "epoch": 0.2660880096891401,
      "grad_norm": 117.91165161132812,
      "learning_rate": 1.721699286422864e-05,
      "loss": 5.4157,
      "step": 6591
    },
    {
      "epoch": 0.2661283811061768,
      "grad_norm": 90.05259704589844,
      "learning_rate": 1.7216087686602045e-05,
      "loss": 4.6285,
      "step": 6592
    },
    {
      "epoch": 0.2661687525232136,
      "grad_norm": 146.1097412109375,
      "learning_rate": 1.7215182385597202e-05,
      "loss": 4.4292,
      "step": 6593
    },
    {
      "epoch": 0.2662091239402503,
      "grad_norm": 92.96125030517578,
      "learning_rate": 1.7214276961229587e-05,
      "loss": 6.6645,
      "step": 6594
    },
    {
      "epoch": 0.26624949535728704,
      "grad_norm": 102.62379455566406,
      "learning_rate": 1.7213371413514682e-05,
      "loss": 5.064,
      "step": 6595
    },
    {
      "epoch": 0.26628986677432376,
      "grad_norm": 86.68009948730469,
      "learning_rate": 1.7212465742467973e-05,
      "loss": 5.9458,
      "step": 6596
    },
    {
      "epoch": 0.26633023819136054,
      "grad_norm": 72.58829498291016,
      "learning_rate": 1.721155994810494e-05,
      "loss": 6.3812,
      "step": 6597
    },
    {
      "epoch": 0.26637060960839726,
      "grad_norm": 88.87329864501953,
      "learning_rate": 1.7210654030441075e-05,
      "loss": 5.8914,
      "step": 6598
    },
    {
      "epoch": 0.266410981025434,
      "grad_norm": 92.94698333740234,
      "learning_rate": 1.7209747989491862e-05,
      "loss": 5.1817,
      "step": 6599
    },
    {
      "epoch": 0.26645135244247076,
      "grad_norm": 118.68573760986328,
      "learning_rate": 1.7208841825272792e-05,
      "loss": 7.2545,
      "step": 6600
    },
    {
      "epoch": 0.2664917238595075,
      "grad_norm": 93.45089721679688,
      "learning_rate": 1.7207935537799362e-05,
      "loss": 3.2135,
      "step": 6601
    },
    {
      "epoch": 0.2665320952765442,
      "grad_norm": 102.0355224609375,
      "learning_rate": 1.7207029127087067e-05,
      "loss": 6.3463,
      "step": 6602
    },
    {
      "epoch": 0.2665724666935809,
      "grad_norm": 93.50227355957031,
      "learning_rate": 1.7206122593151404e-05,
      "loss": 4.4182,
      "step": 6603
    },
    {
      "epoch": 0.2666128381106177,
      "grad_norm": 142.44593811035156,
      "learning_rate": 1.720521593600787e-05,
      "loss": 4.6098,
      "step": 6604
    },
    {
      "epoch": 0.2666532095276544,
      "grad_norm": 164.8262176513672,
      "learning_rate": 1.7204309155671972e-05,
      "loss": 4.4359,
      "step": 6605
    },
    {
      "epoch": 0.26669358094469114,
      "grad_norm": 100.68806457519531,
      "learning_rate": 1.720340225215921e-05,
      "loss": 5.401,
      "step": 6606
    },
    {
      "epoch": 0.2667339523617279,
      "grad_norm": 117.5062484741211,
      "learning_rate": 1.720249522548509e-05,
      "loss": 5.4749,
      "step": 6607
    },
    {
      "epoch": 0.26677432377876464,
      "grad_norm": 80.3673095703125,
      "learning_rate": 1.720158807566512e-05,
      "loss": 6.2568,
      "step": 6608
    },
    {
      "epoch": 0.26681469519580137,
      "grad_norm": 90.88360595703125,
      "learning_rate": 1.720068080271481e-05,
      "loss": 6.2238,
      "step": 6609
    },
    {
      "epoch": 0.2668550666128381,
      "grad_norm": 237.7898712158203,
      "learning_rate": 1.7199773406649676e-05,
      "loss": 5.0178,
      "step": 6610
    },
    {
      "epoch": 0.26689543802987487,
      "grad_norm": 122.8186264038086,
      "learning_rate": 1.7198865887485228e-05,
      "loss": 5.8856,
      "step": 6611
    },
    {
      "epoch": 0.2669358094469116,
      "grad_norm": 123.58280944824219,
      "learning_rate": 1.7197958245236982e-05,
      "loss": 4.9515,
      "step": 6612
    },
    {
      "epoch": 0.2669761808639483,
      "grad_norm": 81.16875457763672,
      "learning_rate": 1.7197050479920464e-05,
      "loss": 4.2946,
      "step": 6613
    },
    {
      "epoch": 0.2670165522809851,
      "grad_norm": 156.4597625732422,
      "learning_rate": 1.7196142591551183e-05,
      "loss": 4.5941,
      "step": 6614
    },
    {
      "epoch": 0.2670569236980218,
      "grad_norm": 86.15025329589844,
      "learning_rate": 1.719523458014467e-05,
      "loss": 6.2106,
      "step": 6615
    },
    {
      "epoch": 0.26709729511505853,
      "grad_norm": 96.57862091064453,
      "learning_rate": 1.719432644571645e-05,
      "loss": 5.7894,
      "step": 6616
    },
    {
      "epoch": 0.26713766653209525,
      "grad_norm": 156.4712371826172,
      "learning_rate": 1.7193418188282044e-05,
      "loss": 5.6377,
      "step": 6617
    },
    {
      "epoch": 0.26717803794913203,
      "grad_norm": 120.93501281738281,
      "learning_rate": 1.719250980785699e-05,
      "loss": 6.359,
      "step": 6618
    },
    {
      "epoch": 0.26721840936616875,
      "grad_norm": 94.79085540771484,
      "learning_rate": 1.7191601304456807e-05,
      "loss": 5.8959,
      "step": 6619
    },
    {
      "epoch": 0.2672587807832055,
      "grad_norm": 89.23776245117188,
      "learning_rate": 1.719069267809704e-05,
      "loss": 4.7124,
      "step": 6620
    },
    {
      "epoch": 0.26729915220024225,
      "grad_norm": 111.75122833251953,
      "learning_rate": 1.7189783928793215e-05,
      "loss": 5.7291,
      "step": 6621
    },
    {
      "epoch": 0.267339523617279,
      "grad_norm": 151.85226440429688,
      "learning_rate": 1.7188875056560878e-05,
      "loss": 4.2729,
      "step": 6622
    },
    {
      "epoch": 0.2673798950343157,
      "grad_norm": 78.71054077148438,
      "learning_rate": 1.7187966061415566e-05,
      "loss": 7.4538,
      "step": 6623
    },
    {
      "epoch": 0.2674202664513524,
      "grad_norm": 96.5770034790039,
      "learning_rate": 1.718705694337282e-05,
      "loss": 4.8559,
      "step": 6624
    },
    {
      "epoch": 0.2674606378683892,
      "grad_norm": 123.04589080810547,
      "learning_rate": 1.718614770244818e-05,
      "loss": 5.5155,
      "step": 6625
    },
    {
      "epoch": 0.2675010092854259,
      "grad_norm": 125.66568756103516,
      "learning_rate": 1.7185238338657196e-05,
      "loss": 6.0466,
      "step": 6626
    },
    {
      "epoch": 0.26754138070246264,
      "grad_norm": 97.4511489868164,
      "learning_rate": 1.7184328852015416e-05,
      "loss": 5.4946,
      "step": 6627
    },
    {
      "epoch": 0.2675817521194994,
      "grad_norm": 152.4881134033203,
      "learning_rate": 1.718341924253839e-05,
      "loss": 4.0938,
      "step": 6628
    },
    {
      "epoch": 0.26762212353653614,
      "grad_norm": 106.85590362548828,
      "learning_rate": 1.7182509510241665e-05,
      "loss": 5.8803,
      "step": 6629
    },
    {
      "epoch": 0.26766249495357286,
      "grad_norm": 94.64881896972656,
      "learning_rate": 1.7181599655140803e-05,
      "loss": 4.6738,
      "step": 6630
    },
    {
      "epoch": 0.2677028663706096,
      "grad_norm": 119.96232604980469,
      "learning_rate": 1.718068967725136e-05,
      "loss": 4.0589,
      "step": 6631
    },
    {
      "epoch": 0.26774323778764636,
      "grad_norm": 145.16598510742188,
      "learning_rate": 1.7179779576588887e-05,
      "loss": 5.2895,
      "step": 6632
    },
    {
      "epoch": 0.2677836092046831,
      "grad_norm": 87.23330688476562,
      "learning_rate": 1.717886935316895e-05,
      "loss": 5.466,
      "step": 6633
    },
    {
      "epoch": 0.2678239806217198,
      "grad_norm": 76.01519775390625,
      "learning_rate": 1.7177959007007112e-05,
      "loss": 3.7512,
      "step": 6634
    },
    {
      "epoch": 0.2678643520387566,
      "grad_norm": 99.74445343017578,
      "learning_rate": 1.7177048538118937e-05,
      "loss": 3.8613,
      "step": 6635
    },
    {
      "epoch": 0.2679047234557933,
      "grad_norm": 132.28765869140625,
      "learning_rate": 1.717613794651999e-05,
      "loss": 7.1845,
      "step": 6636
    },
    {
      "epoch": 0.26794509487283,
      "grad_norm": 117.61344146728516,
      "learning_rate": 1.7175227232225844e-05,
      "loss": 5.6318,
      "step": 6637
    },
    {
      "epoch": 0.26798546628986675,
      "grad_norm": 139.38291931152344,
      "learning_rate": 1.7174316395252068e-05,
      "loss": 3.14,
      "step": 6638
    },
    {
      "epoch": 0.2680258377069035,
      "grad_norm": 127.41146087646484,
      "learning_rate": 1.7173405435614233e-05,
      "loss": 5.6917,
      "step": 6639
    },
    {
      "epoch": 0.26806620912394025,
      "grad_norm": 102.65313720703125,
      "learning_rate": 1.7172494353327914e-05,
      "loss": 5.2672,
      "step": 6640
    },
    {
      "epoch": 0.26810658054097697,
      "grad_norm": 93.96678161621094,
      "learning_rate": 1.7171583148408694e-05,
      "loss": 5.5611,
      "step": 6641
    },
    {
      "epoch": 0.26814695195801375,
      "grad_norm": 101.18846130371094,
      "learning_rate": 1.717067182087215e-05,
      "loss": 5.4156,
      "step": 6642
    },
    {
      "epoch": 0.26818732337505047,
      "grad_norm": 94.42167663574219,
      "learning_rate": 1.7169760370733856e-05,
      "loss": 6.3345,
      "step": 6643
    },
    {
      "epoch": 0.2682276947920872,
      "grad_norm": 133.2460174560547,
      "learning_rate": 1.716884879800941e-05,
      "loss": 6.4109,
      "step": 6644
    },
    {
      "epoch": 0.26826806620912397,
      "grad_norm": 155.91969299316406,
      "learning_rate": 1.7167937102714386e-05,
      "loss": 4.251,
      "step": 6645
    },
    {
      "epoch": 0.2683084376261607,
      "grad_norm": 130.11080932617188,
      "learning_rate": 1.7167025284864375e-05,
      "loss": 7.4536,
      "step": 6646
    },
    {
      "epoch": 0.2683488090431974,
      "grad_norm": 87.18595123291016,
      "learning_rate": 1.716611334447497e-05,
      "loss": 4.8975,
      "step": 6647
    },
    {
      "epoch": 0.26838918046023413,
      "grad_norm": 104.96135711669922,
      "learning_rate": 1.716520128156176e-05,
      "loss": 5.1097,
      "step": 6648
    },
    {
      "epoch": 0.2684295518772709,
      "grad_norm": 88.2764663696289,
      "learning_rate": 1.716428909614034e-05,
      "loss": 6.606,
      "step": 6649
    },
    {
      "epoch": 0.26846992329430763,
      "grad_norm": 122.88414001464844,
      "learning_rate": 1.716337678822631e-05,
      "loss": 7.4385,
      "step": 6650
    },
    {
      "epoch": 0.26851029471134436,
      "grad_norm": 131.61961364746094,
      "learning_rate": 1.716246435783526e-05,
      "loss": 4.3459,
      "step": 6651
    },
    {
      "epoch": 0.26855066612838113,
      "grad_norm": 137.95217895507812,
      "learning_rate": 1.7161551804982797e-05,
      "loss": 4.4021,
      "step": 6652
    },
    {
      "epoch": 0.26859103754541785,
      "grad_norm": 93.54940032958984,
      "learning_rate": 1.716063912968452e-05,
      "loss": 4.2735,
      "step": 6653
    },
    {
      "epoch": 0.2686314089624546,
      "grad_norm": 82.86613464355469,
      "learning_rate": 1.7159726331956036e-05,
      "loss": 4.2794,
      "step": 6654
    },
    {
      "epoch": 0.2686717803794913,
      "grad_norm": 103.31629180908203,
      "learning_rate": 1.715881341181295e-05,
      "loss": 6.4652,
      "step": 6655
    },
    {
      "epoch": 0.2687121517965281,
      "grad_norm": 130.6968231201172,
      "learning_rate": 1.7157900369270875e-05,
      "loss": 5.972,
      "step": 6656
    },
    {
      "epoch": 0.2687525232135648,
      "grad_norm": 138.47840881347656,
      "learning_rate": 1.7156987204345417e-05,
      "loss": 6.2564,
      "step": 6657
    },
    {
      "epoch": 0.2687928946306015,
      "grad_norm": 80.5005111694336,
      "learning_rate": 1.715607391705219e-05,
      "loss": 5.3809,
      "step": 6658
    },
    {
      "epoch": 0.2688332660476383,
      "grad_norm": 84.7727279663086,
      "learning_rate": 1.715516050740681e-05,
      "loss": 4.9796,
      "step": 6659
    },
    {
      "epoch": 0.268873637464675,
      "grad_norm": 93.37678527832031,
      "learning_rate": 1.7154246975424892e-05,
      "loss": 4.2793,
      "step": 6660
    },
    {
      "epoch": 0.26891400888171174,
      "grad_norm": 227.65667724609375,
      "learning_rate": 1.7153333321122056e-05,
      "loss": 5.1684,
      "step": 6661
    },
    {
      "epoch": 0.26895438029874846,
      "grad_norm": 91.35773468017578,
      "learning_rate": 1.7152419544513928e-05,
      "loss": 5.5144,
      "step": 6662
    },
    {
      "epoch": 0.26899475171578524,
      "grad_norm": 103.8681869506836,
      "learning_rate": 1.7151505645616126e-05,
      "loss": 6.6544,
      "step": 6663
    },
    {
      "epoch": 0.26903512313282196,
      "grad_norm": 99.90460968017578,
      "learning_rate": 1.715059162444428e-05,
      "loss": 5.9156,
      "step": 6664
    },
    {
      "epoch": 0.2690754945498587,
      "grad_norm": 66.02108001708984,
      "learning_rate": 1.7149677481014012e-05,
      "loss": 6.3064,
      "step": 6665
    },
    {
      "epoch": 0.26911586596689546,
      "grad_norm": 100.81643676757812,
      "learning_rate": 1.714876321534096e-05,
      "loss": 5.4542,
      "step": 6666
    },
    {
      "epoch": 0.2691562373839322,
      "grad_norm": 119.1552505493164,
      "learning_rate": 1.7147848827440743e-05,
      "loss": 4.9399,
      "step": 6667
    },
    {
      "epoch": 0.2691966088009689,
      "grad_norm": 71.12548065185547,
      "learning_rate": 1.7146934317329008e-05,
      "loss": 4.2293,
      "step": 6668
    },
    {
      "epoch": 0.26923698021800563,
      "grad_norm": 197.54898071289062,
      "learning_rate": 1.7146019685021382e-05,
      "loss": 5.3839,
      "step": 6669
    },
    {
      "epoch": 0.2692773516350424,
      "grad_norm": 133.7823028564453,
      "learning_rate": 1.714510493053351e-05,
      "loss": 3.8117,
      "step": 6670
    },
    {
      "epoch": 0.2693177230520791,
      "grad_norm": 118.42481994628906,
      "learning_rate": 1.7144190053881023e-05,
      "loss": 5.2528,
      "step": 6671
    },
    {
      "epoch": 0.26935809446911585,
      "grad_norm": 176.80770874023438,
      "learning_rate": 1.7143275055079575e-05,
      "loss": 4.8643,
      "step": 6672
    },
    {
      "epoch": 0.2693984658861526,
      "grad_norm": 149.00636291503906,
      "learning_rate": 1.7142359934144802e-05,
      "loss": 5.4277,
      "step": 6673
    },
    {
      "epoch": 0.26943883730318935,
      "grad_norm": 96.44613647460938,
      "learning_rate": 1.7141444691092354e-05,
      "loss": 6.1447,
      "step": 6674
    },
    {
      "epoch": 0.26947920872022607,
      "grad_norm": 154.36465454101562,
      "learning_rate": 1.7140529325937877e-05,
      "loss": 4.9954,
      "step": 6675
    },
    {
      "epoch": 0.2695195801372628,
      "grad_norm": 138.6244659423828,
      "learning_rate": 1.713961383869702e-05,
      "loss": 4.6184,
      "step": 6676
    },
    {
      "epoch": 0.26955995155429957,
      "grad_norm": 143.53834533691406,
      "learning_rate": 1.7138698229385443e-05,
      "loss": 5.7838,
      "step": 6677
    },
    {
      "epoch": 0.2696003229713363,
      "grad_norm": 72.75285339355469,
      "learning_rate": 1.7137782498018796e-05,
      "loss": 5.8784,
      "step": 6678
    },
    {
      "epoch": 0.269640694388373,
      "grad_norm": 70.09127044677734,
      "learning_rate": 1.7136866644612733e-05,
      "loss": 4.8349,
      "step": 6679
    },
    {
      "epoch": 0.2696810658054098,
      "grad_norm": 139.04029846191406,
      "learning_rate": 1.713595066918292e-05,
      "loss": 5.5153,
      "step": 6680
    },
    {
      "epoch": 0.2697214372224465,
      "grad_norm": 89.08656311035156,
      "learning_rate": 1.7135034571745014e-05,
      "loss": 5.7049,
      "step": 6681
    },
    {
      "epoch": 0.26976180863948324,
      "grad_norm": 92.26889038085938,
      "learning_rate": 1.7134118352314676e-05,
      "loss": 3.4514,
      "step": 6682
    },
    {
      "epoch": 0.26980218005651996,
      "grad_norm": 164.52340698242188,
      "learning_rate": 1.713320201090757e-05,
      "loss": 4.9321,
      "step": 6683
    },
    {
      "epoch": 0.26984255147355674,
      "grad_norm": 84.6821517944336,
      "learning_rate": 1.7132285547539374e-05,
      "loss": 4.3833,
      "step": 6684
    },
    {
      "epoch": 0.26988292289059346,
      "grad_norm": 81.02751922607422,
      "learning_rate": 1.7131368962225748e-05,
      "loss": 6.344,
      "step": 6685
    },
    {
      "epoch": 0.2699232943076302,
      "grad_norm": 193.52035522460938,
      "learning_rate": 1.7130452254982362e-05,
      "loss": 4.8287,
      "step": 6686
    },
    {
      "epoch": 0.26996366572466696,
      "grad_norm": 118.11158752441406,
      "learning_rate": 1.7129535425824898e-05,
      "loss": 5.6141,
      "step": 6687
    },
    {
      "epoch": 0.2700040371417037,
      "grad_norm": 153.21656799316406,
      "learning_rate": 1.7128618474769024e-05,
      "loss": 4.7596,
      "step": 6688
    },
    {
      "epoch": 0.2700444085587404,
      "grad_norm": 79.35603332519531,
      "learning_rate": 1.7127701401830422e-05,
      "loss": 5.6455,
      "step": 6689
    },
    {
      "epoch": 0.2700847799757771,
      "grad_norm": 156.10752868652344,
      "learning_rate": 1.7126784207024772e-05,
      "loss": 7.2151,
      "step": 6690
    },
    {
      "epoch": 0.2701251513928139,
      "grad_norm": 91.74716186523438,
      "learning_rate": 1.7125866890367754e-05,
      "loss": 6.0179,
      "step": 6691
    },
    {
      "epoch": 0.2701655228098506,
      "grad_norm": 71.57186126708984,
      "learning_rate": 1.712494945187505e-05,
      "loss": 6.682,
      "step": 6692
    },
    {
      "epoch": 0.27020589422688734,
      "grad_norm": 138.6996612548828,
      "learning_rate": 1.712403189156235e-05,
      "loss": 5.9157,
      "step": 6693
    },
    {
      "epoch": 0.2702462656439241,
      "grad_norm": 87.12960815429688,
      "learning_rate": 1.7123114209445336e-05,
      "loss": 5.4368,
      "step": 6694
    },
    {
      "epoch": 0.27028663706096084,
      "grad_norm": 149.0299072265625,
      "learning_rate": 1.7122196405539706e-05,
      "loss": 6.0958,
      "step": 6695
    },
    {
      "epoch": 0.27032700847799757,
      "grad_norm": 125.85977172851562,
      "learning_rate": 1.712127847986115e-05,
      "loss": 4.816,
      "step": 6696
    },
    {
      "epoch": 0.27036737989503434,
      "grad_norm": 117.78314971923828,
      "learning_rate": 1.7120360432425363e-05,
      "loss": 5.3037,
      "step": 6697
    },
    {
      "epoch": 0.27040775131207107,
      "grad_norm": 108.23990631103516,
      "learning_rate": 1.7119442263248034e-05,
      "loss": 7.7867,
      "step": 6698
    },
    {
      "epoch": 0.2704481227291078,
      "grad_norm": 83.2197494506836,
      "learning_rate": 1.711852397234487e-05,
      "loss": 5.814,
      "step": 6699
    },
    {
      "epoch": 0.2704884941461445,
      "grad_norm": 89.37845611572266,
      "learning_rate": 1.7117605559731573e-05,
      "loss": 6.8485,
      "step": 6700
    },
    {
      "epoch": 0.2705288655631813,
      "grad_norm": 71.2811508178711,
      "learning_rate": 1.7116687025423836e-05,
      "loss": 3.5342,
      "step": 6701
    },
    {
      "epoch": 0.270569236980218,
      "grad_norm": 93.48041534423828,
      "learning_rate": 1.7115768369437373e-05,
      "loss": 4.7751,
      "step": 6702
    },
    {
      "epoch": 0.27060960839725473,
      "grad_norm": 108.11287689208984,
      "learning_rate": 1.7114849591787888e-05,
      "loss": 5.8323,
      "step": 6703
    },
    {
      "epoch": 0.2706499798142915,
      "grad_norm": 121.08121490478516,
      "learning_rate": 1.7113930692491086e-05,
      "loss": 4.0837,
      "step": 6704
    },
    {
      "epoch": 0.27069035123132823,
      "grad_norm": 119.13776397705078,
      "learning_rate": 1.7113011671562683e-05,
      "loss": 5.1219,
      "step": 6705
    },
    {
      "epoch": 0.27073072264836495,
      "grad_norm": 101.87144470214844,
      "learning_rate": 1.711209252901839e-05,
      "loss": 5.1115,
      "step": 6706
    },
    {
      "epoch": 0.2707710940654017,
      "grad_norm": 113.35063171386719,
      "learning_rate": 1.7111173264873927e-05,
      "loss": 4.4045,
      "step": 6707
    },
    {
      "epoch": 0.27081146548243845,
      "grad_norm": 71.4780044555664,
      "learning_rate": 1.7110253879145e-05,
      "loss": 5.967,
      "step": 6708
    },
    {
      "epoch": 0.2708518368994752,
      "grad_norm": 105.0081558227539,
      "learning_rate": 1.710933437184734e-05,
      "loss": 6.176,
      "step": 6709
    },
    {
      "epoch": 0.2708922083165119,
      "grad_norm": 97.69917297363281,
      "learning_rate": 1.7108414742996663e-05,
      "loss": 5.3081,
      "step": 6710
    },
    {
      "epoch": 0.2709325797335487,
      "grad_norm": 93.28056335449219,
      "learning_rate": 1.710749499260869e-05,
      "loss": 7.0262,
      "step": 6711
    },
    {
      "epoch": 0.2709729511505854,
      "grad_norm": 81.34341430664062,
      "learning_rate": 1.7106575120699154e-05,
      "loss": 5.9292,
      "step": 6712
    },
    {
      "epoch": 0.2710133225676221,
      "grad_norm": 83.85417938232422,
      "learning_rate": 1.7105655127283775e-05,
      "loss": 5.7967,
      "step": 6713
    },
    {
      "epoch": 0.27105369398465884,
      "grad_norm": 70.42186737060547,
      "learning_rate": 1.710473501237829e-05,
      "loss": 6.0487,
      "step": 6714
    },
    {
      "epoch": 0.2710940654016956,
      "grad_norm": 78.20272064208984,
      "learning_rate": 1.7103814775998425e-05,
      "loss": 5.9306,
      "step": 6715
    },
    {
      "epoch": 0.27113443681873234,
      "grad_norm": 95.75785064697266,
      "learning_rate": 1.7102894418159917e-05,
      "loss": 5.057,
      "step": 6716
    },
    {
      "epoch": 0.27117480823576906,
      "grad_norm": 140.29876708984375,
      "learning_rate": 1.71019739388785e-05,
      "loss": 6.0023,
      "step": 6717
    },
    {
      "epoch": 0.27121517965280584,
      "grad_norm": 61.107173919677734,
      "learning_rate": 1.7101053338169914e-05,
      "loss": 3.959,
      "step": 6718
    },
    {
      "epoch": 0.27125555106984256,
      "grad_norm": 100.26707458496094,
      "learning_rate": 1.7100132616049895e-05,
      "loss": 5.9631,
      "step": 6719
    },
    {
      "epoch": 0.2712959224868793,
      "grad_norm": 83.2303466796875,
      "learning_rate": 1.709921177253419e-05,
      "loss": 5.3553,
      "step": 6720
    },
    {
      "epoch": 0.271336293903916,
      "grad_norm": 128.73370361328125,
      "learning_rate": 1.7098290807638544e-05,
      "loss": 5.1422,
      "step": 6721
    },
    {
      "epoch": 0.2713766653209528,
      "grad_norm": 217.14410400390625,
      "learning_rate": 1.7097369721378698e-05,
      "loss": 4.2952,
      "step": 6722
    },
    {
      "epoch": 0.2714170367379895,
      "grad_norm": 98.2079849243164,
      "learning_rate": 1.70964485137704e-05,
      "loss": 5.8912,
      "step": 6723
    },
    {
      "epoch": 0.2714574081550262,
      "grad_norm": 79.72669982910156,
      "learning_rate": 1.709552718482941e-05,
      "loss": 5.4499,
      "step": 6724
    },
    {
      "epoch": 0.271497779572063,
      "grad_norm": 83.28626251220703,
      "learning_rate": 1.7094605734571472e-05,
      "loss": 4.9205,
      "step": 6725
    },
    {
      "epoch": 0.2715381509890997,
      "grad_norm": 87.71932220458984,
      "learning_rate": 1.7093684163012346e-05,
      "loss": 5.6217,
      "step": 6726
    },
    {
      "epoch": 0.27157852240613645,
      "grad_norm": 95.55823516845703,
      "learning_rate": 1.709276247016778e-05,
      "loss": 6.7796,
      "step": 6727
    },
    {
      "epoch": 0.27161889382317317,
      "grad_norm": 125.14491271972656,
      "learning_rate": 1.709184065605354e-05,
      "loss": 4.6747,
      "step": 6728
    },
    {
      "epoch": 0.27165926524020995,
      "grad_norm": 98.15953063964844,
      "learning_rate": 1.7090918720685388e-05,
      "loss": 6.1835,
      "step": 6729
    },
    {
      "epoch": 0.27169963665724667,
      "grad_norm": 90.0787124633789,
      "learning_rate": 1.7089996664079085e-05,
      "loss": 4.191,
      "step": 6730
    },
    {
      "epoch": 0.2717400080742834,
      "grad_norm": 75.77354431152344,
      "learning_rate": 1.7089074486250394e-05,
      "loss": 6.4446,
      "step": 6731
    },
    {
      "epoch": 0.27178037949132017,
      "grad_norm": 126.89590454101562,
      "learning_rate": 1.708815218721508e-05,
      "loss": 6.4844,
      "step": 6732
    },
    {
      "epoch": 0.2718207509083569,
      "grad_norm": 79.17387390136719,
      "learning_rate": 1.708722976698892e-05,
      "loss": 4.8175,
      "step": 6733
    },
    {
      "epoch": 0.2718611223253936,
      "grad_norm": 102.62660217285156,
      "learning_rate": 1.7086307225587678e-05,
      "loss": 5.5661,
      "step": 6734
    },
    {
      "epoch": 0.27190149374243033,
      "grad_norm": 110.63492584228516,
      "learning_rate": 1.7085384563027134e-05,
      "loss": 5.2073,
      "step": 6735
    },
    {
      "epoch": 0.2719418651594671,
      "grad_norm": 88.72075653076172,
      "learning_rate": 1.7084461779323057e-05,
      "loss": 4.9761,
      "step": 6736
    },
    {
      "epoch": 0.27198223657650383,
      "grad_norm": 84.90026092529297,
      "learning_rate": 1.708353887449123e-05,
      "loss": 5.7904,
      "step": 6737
    },
    {
      "epoch": 0.27202260799354055,
      "grad_norm": 99.81644439697266,
      "learning_rate": 1.7082615848547424e-05,
      "loss": 5.7109,
      "step": 6738
    },
    {
      "epoch": 0.27206297941057733,
      "grad_norm": 74.01146697998047,
      "learning_rate": 1.708169270150743e-05,
      "loss": 5.6561,
      "step": 6739
    },
    {
      "epoch": 0.27210335082761405,
      "grad_norm": 131.32080078125,
      "learning_rate": 1.7080769433387026e-05,
      "loss": 7.1625,
      "step": 6740
    },
    {
      "epoch": 0.2721437222446508,
      "grad_norm": 95.32120513916016,
      "learning_rate": 1.7079846044202e-05,
      "loss": 5.2449,
      "step": 6741
    },
    {
      "epoch": 0.2721840936616875,
      "grad_norm": 123.02565002441406,
      "learning_rate": 1.707892253396814e-05,
      "loss": 5.8318,
      "step": 6742
    },
    {
      "epoch": 0.2722244650787243,
      "grad_norm": 109.7567367553711,
      "learning_rate": 1.7077998902701236e-05,
      "loss": 5.9062,
      "step": 6743
    },
    {
      "epoch": 0.272264836495761,
      "grad_norm": 84.64254760742188,
      "learning_rate": 1.7077075150417078e-05,
      "loss": 5.8079,
      "step": 6744
    },
    {
      "epoch": 0.2723052079127977,
      "grad_norm": 106.3527603149414,
      "learning_rate": 1.707615127713146e-05,
      "loss": 5.1136,
      "step": 6745
    },
    {
      "epoch": 0.2723455793298345,
      "grad_norm": 94.04998779296875,
      "learning_rate": 1.7075227282860177e-05,
      "loss": 5.9613,
      "step": 6746
    },
    {
      "epoch": 0.2723859507468712,
      "grad_norm": 101.80010986328125,
      "learning_rate": 1.7074303167619037e-05,
      "loss": 4.5893,
      "step": 6747
    },
    {
      "epoch": 0.27242632216390794,
      "grad_norm": 170.967041015625,
      "learning_rate": 1.7073378931423824e-05,
      "loss": 5.5377,
      "step": 6748
    },
    {
      "epoch": 0.2724666935809447,
      "grad_norm": 89.43157958984375,
      "learning_rate": 1.7072454574290355e-05,
      "loss": 6.6396,
      "step": 6749
    },
    {
      "epoch": 0.27250706499798144,
      "grad_norm": 71.71910858154297,
      "learning_rate": 1.7071530096234425e-05,
      "loss": 5.3971,
      "step": 6750
    },
    {
      "epoch": 0.27254743641501816,
      "grad_norm": 114.3143310546875,
      "learning_rate": 1.7070605497271843e-05,
      "loss": 5.1539,
      "step": 6751
    },
    {
      "epoch": 0.2725878078320549,
      "grad_norm": 81.11254119873047,
      "learning_rate": 1.7069680777418422e-05,
      "loss": 5.1032,
      "step": 6752
    },
    {
      "epoch": 0.27262817924909166,
      "grad_norm": 76.52278137207031,
      "learning_rate": 1.7068755936689964e-05,
      "loss": 5.8789,
      "step": 6753
    },
    {
      "epoch": 0.2726685506661284,
      "grad_norm": 108.78157043457031,
      "learning_rate": 1.7067830975102287e-05,
      "loss": 5.4006,
      "step": 6754
    },
    {
      "epoch": 0.2727089220831651,
      "grad_norm": 67.92649841308594,
      "learning_rate": 1.7066905892671206e-05,
      "loss": 4.6529,
      "step": 6755
    },
    {
      "epoch": 0.2727492935002019,
      "grad_norm": 213.6684112548828,
      "learning_rate": 1.7065980689412538e-05,
      "loss": 5.3534,
      "step": 6756
    },
    {
      "epoch": 0.2727896649172386,
      "grad_norm": 106.94843292236328,
      "learning_rate": 1.70650553653421e-05,
      "loss": 5.345,
      "step": 6757
    },
    {
      "epoch": 0.2728300363342753,
      "grad_norm": 200.22955322265625,
      "learning_rate": 1.7064129920475712e-05,
      "loss": 4.9986,
      "step": 6758
    },
    {
      "epoch": 0.27287040775131205,
      "grad_norm": 112.97657012939453,
      "learning_rate": 1.7063204354829193e-05,
      "loss": 4.8962,
      "step": 6759
    },
    {
      "epoch": 0.2729107791683488,
      "grad_norm": 103.89055633544922,
      "learning_rate": 1.7062278668418382e-05,
      "loss": 4.225,
      "step": 6760
    },
    {
      "epoch": 0.27295115058538555,
      "grad_norm": 132.59495544433594,
      "learning_rate": 1.7061352861259094e-05,
      "loss": 4.323,
      "step": 6761
    },
    {
      "epoch": 0.27299152200242227,
      "grad_norm": 76.79475402832031,
      "learning_rate": 1.706042693336716e-05,
      "loss": 6.347,
      "step": 6762
    },
    {
      "epoch": 0.27303189341945905,
      "grad_norm": 138.9867706298828,
      "learning_rate": 1.7059500884758417e-05,
      "loss": 4.9303,
      "step": 6763
    },
    {
      "epoch": 0.27307226483649577,
      "grad_norm": 89.55731201171875,
      "learning_rate": 1.705857471544869e-05,
      "loss": 5.7098,
      "step": 6764
    },
    {
      "epoch": 0.2731126362535325,
      "grad_norm": 118.77864074707031,
      "learning_rate": 1.7057648425453823e-05,
      "loss": 5.6769,
      "step": 6765
    },
    {
      "epoch": 0.2731530076705692,
      "grad_norm": 94.8846206665039,
      "learning_rate": 1.7056722014789645e-05,
      "loss": 5.0746,
      "step": 6766
    },
    {
      "epoch": 0.273193379087606,
      "grad_norm": 130.00267028808594,
      "learning_rate": 1.7055795483472e-05,
      "loss": 4.8597,
      "step": 6767
    },
    {
      "epoch": 0.2732337505046427,
      "grad_norm": 76.04297637939453,
      "learning_rate": 1.7054868831516727e-05,
      "loss": 4.5974,
      "step": 6768
    },
    {
      "epoch": 0.27327412192167944,
      "grad_norm": 62.30728530883789,
      "learning_rate": 1.7053942058939674e-05,
      "loss": 3.9561,
      "step": 6769
    },
    {
      "epoch": 0.2733144933387162,
      "grad_norm": 73.47811126708984,
      "learning_rate": 1.7053015165756682e-05,
      "loss": 5.0463,
      "step": 6770
    },
    {
      "epoch": 0.27335486475575294,
      "grad_norm": 88.8709945678711,
      "learning_rate": 1.7052088151983602e-05,
      "loss": 5.3783,
      "step": 6771
    },
    {
      "epoch": 0.27339523617278966,
      "grad_norm": 124.31585693359375,
      "learning_rate": 1.7051161017636283e-05,
      "loss": 7.099,
      "step": 6772
    },
    {
      "epoch": 0.2734356075898264,
      "grad_norm": 115.69828033447266,
      "learning_rate": 1.7050233762730575e-05,
      "loss": 4.7256,
      "step": 6773
    },
    {
      "epoch": 0.27347597900686316,
      "grad_norm": 111.81758117675781,
      "learning_rate": 1.704930638728233e-05,
      "loss": 5.5334,
      "step": 6774
    },
    {
      "epoch": 0.2735163504238999,
      "grad_norm": 74.57620239257812,
      "learning_rate": 1.704837889130741e-05,
      "loss": 5.8287,
      "step": 6775
    },
    {
      "epoch": 0.2735567218409366,
      "grad_norm": 86.2268295288086,
      "learning_rate": 1.7047451274821674e-05,
      "loss": 5.2321,
      "step": 6776
    },
    {
      "epoch": 0.2735970932579734,
      "grad_norm": 322.3421630859375,
      "learning_rate": 1.7046523537840974e-05,
      "loss": 6.9712,
      "step": 6777
    },
    {
      "epoch": 0.2736374646750101,
      "grad_norm": 153.71157836914062,
      "learning_rate": 1.7045595680381178e-05,
      "loss": 4.707,
      "step": 6778
    },
    {
      "epoch": 0.2736778360920468,
      "grad_norm": 101.691162109375,
      "learning_rate": 1.7044667702458144e-05,
      "loss": 5.0901,
      "step": 6779
    },
    {
      "epoch": 0.27371820750908354,
      "grad_norm": 83.26856994628906,
      "learning_rate": 1.7043739604087748e-05,
      "loss": 6.8466,
      "step": 6780
    },
    {
      "epoch": 0.2737585789261203,
      "grad_norm": 96.30915069580078,
      "learning_rate": 1.704281138528585e-05,
      "loss": 6.0356,
      "step": 6781
    },
    {
      "epoch": 0.27379895034315704,
      "grad_norm": 124.15725708007812,
      "learning_rate": 1.7041883046068327e-05,
      "loss": 7.1069,
      "step": 6782
    },
    {
      "epoch": 0.27383932176019377,
      "grad_norm": 216.10646057128906,
      "learning_rate": 1.7040954586451043e-05,
      "loss": 5.8339,
      "step": 6783
    },
    {
      "epoch": 0.27387969317723054,
      "grad_norm": 88.46839904785156,
      "learning_rate": 1.7040026006449882e-05,
      "loss": 5.9475,
      "step": 6784
    },
    {
      "epoch": 0.27392006459426727,
      "grad_norm": 95.50755310058594,
      "learning_rate": 1.7039097306080715e-05,
      "loss": 3.9687,
      "step": 6785
    },
    {
      "epoch": 0.273960436011304,
      "grad_norm": 151.54067993164062,
      "learning_rate": 1.7038168485359418e-05,
      "loss": 5.3663,
      "step": 6786
    },
    {
      "epoch": 0.2740008074283407,
      "grad_norm": 105.49918365478516,
      "learning_rate": 1.7037239544301878e-05,
      "loss": 3.3899,
      "step": 6787
    },
    {
      "epoch": 0.2740411788453775,
      "grad_norm": 91.3900375366211,
      "learning_rate": 1.7036310482923975e-05,
      "loss": 5.1251,
      "step": 6788
    },
    {
      "epoch": 0.2740815502624142,
      "grad_norm": 96.4812240600586,
      "learning_rate": 1.703538130124159e-05,
      "loss": 6.1315,
      "step": 6789
    },
    {
      "epoch": 0.27412192167945093,
      "grad_norm": 96.03691101074219,
      "learning_rate": 1.703445199927062e-05,
      "loss": 5.1335,
      "step": 6790
    },
    {
      "epoch": 0.2741622930964877,
      "grad_norm": 96.86825561523438,
      "learning_rate": 1.7033522577026944e-05,
      "loss": 5.0337,
      "step": 6791
    },
    {
      "epoch": 0.27420266451352443,
      "grad_norm": 109.15802764892578,
      "learning_rate": 1.7032593034526457e-05,
      "loss": 6.1495,
      "step": 6792
    },
    {
      "epoch": 0.27424303593056115,
      "grad_norm": 100.82378387451172,
      "learning_rate": 1.703166337178505e-05,
      "loss": 5.4999,
      "step": 6793
    },
    {
      "epoch": 0.2742834073475979,
      "grad_norm": 76.09877014160156,
      "learning_rate": 1.7030733588818622e-05,
      "loss": 6.0267,
      "step": 6794
    },
    {
      "epoch": 0.27432377876463465,
      "grad_norm": 86.71719360351562,
      "learning_rate": 1.7029803685643066e-05,
      "loss": 4.8527,
      "step": 6795
    },
    {
      "epoch": 0.2743641501816714,
      "grad_norm": 1575.1751708984375,
      "learning_rate": 1.7028873662274284e-05,
      "loss": 5.4331,
      "step": 6796
    },
    {
      "epoch": 0.2744045215987081,
      "grad_norm": 123.88077545166016,
      "learning_rate": 1.7027943518728175e-05,
      "loss": 4.6976,
      "step": 6797
    },
    {
      "epoch": 0.2744448930157449,
      "grad_norm": 89.82596588134766,
      "learning_rate": 1.702701325502064e-05,
      "loss": 4.8583,
      "step": 6798
    },
    {
      "epoch": 0.2744852644327816,
      "grad_norm": 113.63148498535156,
      "learning_rate": 1.702608287116759e-05,
      "loss": 4.4351,
      "step": 6799
    },
    {
      "epoch": 0.2745256358498183,
      "grad_norm": 163.69244384765625,
      "learning_rate": 1.7025152367184933e-05,
      "loss": 4.2472,
      "step": 6800
    },
    {
      "epoch": 0.27456600726685504,
      "grad_norm": 84.15453338623047,
      "learning_rate": 1.702422174308857e-05,
      "loss": 4.8133,
      "step": 6801
    },
    {
      "epoch": 0.2746063786838918,
      "grad_norm": 113.04182434082031,
      "learning_rate": 1.7023290998894424e-05,
      "loss": 6.2628,
      "step": 6802
    },
    {
      "epoch": 0.27464675010092854,
      "grad_norm": 117.00343322753906,
      "learning_rate": 1.70223601346184e-05,
      "loss": 4.4277,
      "step": 6803
    },
    {
      "epoch": 0.27468712151796526,
      "grad_norm": 79.5206298828125,
      "learning_rate": 1.7021429150276417e-05,
      "loss": 5.1653,
      "step": 6804
    },
    {
      "epoch": 0.27472749293500204,
      "grad_norm": 83.51285552978516,
      "learning_rate": 1.7020498045884387e-05,
      "loss": 7.2793,
      "step": 6805
    },
    {
      "epoch": 0.27476786435203876,
      "grad_norm": 158.48916625976562,
      "learning_rate": 1.701956682145824e-05,
      "loss": 6.5566,
      "step": 6806
    },
    {
      "epoch": 0.2748082357690755,
      "grad_norm": 118.67474365234375,
      "learning_rate": 1.701863547701389e-05,
      "loss": 6.2245,
      "step": 6807
    },
    {
      "epoch": 0.27484860718611226,
      "grad_norm": 86.40516662597656,
      "learning_rate": 1.701770401256726e-05,
      "loss": 5.9053,
      "step": 6808
    },
    {
      "epoch": 0.274888978603149,
      "grad_norm": 131.21575927734375,
      "learning_rate": 1.7016772428134283e-05,
      "loss": 5.2753,
      "step": 6809
    },
    {
      "epoch": 0.2749293500201857,
      "grad_norm": 172.9568634033203,
      "learning_rate": 1.7015840723730883e-05,
      "loss": 5.2505,
      "step": 6810
    },
    {
      "epoch": 0.2749697214372224,
      "grad_norm": 114.7260513305664,
      "learning_rate": 1.7014908899372987e-05,
      "loss": 5.5953,
      "step": 6811
    },
    {
      "epoch": 0.2750100928542592,
      "grad_norm": 118.4347152709961,
      "learning_rate": 1.7013976955076534e-05,
      "loss": 3.3595,
      "step": 6812
    },
    {
      "epoch": 0.2750504642712959,
      "grad_norm": 118.0241928100586,
      "learning_rate": 1.7013044890857448e-05,
      "loss": 4.6838,
      "step": 6813
    },
    {
      "epoch": 0.27509083568833265,
      "grad_norm": 98.46552276611328,
      "learning_rate": 1.701211270673168e-05,
      "loss": 4.5012,
      "step": 6814
    },
    {
      "epoch": 0.2751312071053694,
      "grad_norm": 85.13011932373047,
      "learning_rate": 1.701118040271515e-05,
      "loss": 5.5154,
      "step": 6815
    },
    {
      "epoch": 0.27517157852240615,
      "grad_norm": 111.30109405517578,
      "learning_rate": 1.701024797882381e-05,
      "loss": 6.4527,
      "step": 6816
    },
    {
      "epoch": 0.27521194993944287,
      "grad_norm": 119.20781707763672,
      "learning_rate": 1.7009315435073602e-05,
      "loss": 4.7049,
      "step": 6817
    },
    {
      "epoch": 0.2752523213564796,
      "grad_norm": 95.85858154296875,
      "learning_rate": 1.7008382771480467e-05,
      "loss": 5.053,
      "step": 6818
    },
    {
      "epoch": 0.27529269277351637,
      "grad_norm": 128.31402587890625,
      "learning_rate": 1.7007449988060348e-05,
      "loss": 6.4747,
      "step": 6819
    },
    {
      "epoch": 0.2753330641905531,
      "grad_norm": 142.25140380859375,
      "learning_rate": 1.7006517084829205e-05,
      "loss": 5.918,
      "step": 6820
    },
    {
      "epoch": 0.2753734356075898,
      "grad_norm": 96.05819702148438,
      "learning_rate": 1.7005584061802974e-05,
      "loss": 4.3374,
      "step": 6821
    },
    {
      "epoch": 0.2754138070246266,
      "grad_norm": 129.4245147705078,
      "learning_rate": 1.7004650918997622e-05,
      "loss": 7.057,
      "step": 6822
    },
    {
      "epoch": 0.2754541784416633,
      "grad_norm": 98.65269470214844,
      "learning_rate": 1.700371765642909e-05,
      "loss": 5.0524,
      "step": 6823
    },
    {
      "epoch": 0.27549454985870003,
      "grad_norm": 85.13566589355469,
      "learning_rate": 1.7002784274113345e-05,
      "loss": 6.203,
      "step": 6824
    },
    {
      "epoch": 0.27553492127573675,
      "grad_norm": 111.91275024414062,
      "learning_rate": 1.700185077206634e-05,
      "loss": 4.2131,
      "step": 6825
    },
    {
      "epoch": 0.27557529269277353,
      "grad_norm": 91.78761291503906,
      "learning_rate": 1.700091715030404e-05,
      "loss": 5.4671,
      "step": 6826
    },
    {
      "epoch": 0.27561566410981025,
      "grad_norm": 177.3418731689453,
      "learning_rate": 1.6999983408842404e-05,
      "loss": 3.7099,
      "step": 6827
    },
    {
      "epoch": 0.275656035526847,
      "grad_norm": 126.05192565917969,
      "learning_rate": 1.6999049547697394e-05,
      "loss": 4.1894,
      "step": 6828
    },
    {
      "epoch": 0.27569640694388375,
      "grad_norm": 100.8119125366211,
      "learning_rate": 1.6998115566884986e-05,
      "loss": 4.7662,
      "step": 6829
    },
    {
      "epoch": 0.2757367783609205,
      "grad_norm": 109.28681945800781,
      "learning_rate": 1.699718146642114e-05,
      "loss": 4.9402,
      "step": 6830
    },
    {
      "epoch": 0.2757771497779572,
      "grad_norm": 74.2500991821289,
      "learning_rate": 1.699624724632183e-05,
      "loss": 5.2456,
      "step": 6831
    },
    {
      "epoch": 0.2758175211949939,
      "grad_norm": 213.32119750976562,
      "learning_rate": 1.6995312906603032e-05,
      "loss": 4.5897,
      "step": 6832
    },
    {
      "epoch": 0.2758578926120307,
      "grad_norm": 100.10065460205078,
      "learning_rate": 1.6994378447280716e-05,
      "loss": 5.586,
      "step": 6833
    },
    {
      "epoch": 0.2758982640290674,
      "grad_norm": 100.5224609375,
      "learning_rate": 1.6993443868370866e-05,
      "loss": 4.3428,
      "step": 6834
    },
    {
      "epoch": 0.27593863544610414,
      "grad_norm": 80.15782165527344,
      "learning_rate": 1.699250916988945e-05,
      "loss": 6.2489,
      "step": 6835
    },
    {
      "epoch": 0.2759790068631409,
      "grad_norm": 159.45697021484375,
      "learning_rate": 1.6991574351852462e-05,
      "loss": 5.0898,
      "step": 6836
    },
    {
      "epoch": 0.27601937828017764,
      "grad_norm": 81.62725830078125,
      "learning_rate": 1.6990639414275876e-05,
      "loss": 6.1584,
      "step": 6837
    },
    {
      "epoch": 0.27605974969721436,
      "grad_norm": 139.710693359375,
      "learning_rate": 1.698970435717568e-05,
      "loss": 3.7964,
      "step": 6838
    },
    {
      "epoch": 0.2761001211142511,
      "grad_norm": 87.80329895019531,
      "learning_rate": 1.6988769180567862e-05,
      "loss": 5.9032,
      "step": 6839
    },
    {
      "epoch": 0.27614049253128786,
      "grad_norm": 128.17181396484375,
      "learning_rate": 1.6987833884468416e-05,
      "loss": 6.1841,
      "step": 6840
    },
    {
      "epoch": 0.2761808639483246,
      "grad_norm": 86.04155731201172,
      "learning_rate": 1.698689846889332e-05,
      "loss": 5.8604,
      "step": 6841
    },
    {
      "epoch": 0.2762212353653613,
      "grad_norm": 90.97322082519531,
      "learning_rate": 1.698596293385858e-05,
      "loss": 6.1482,
      "step": 6842
    },
    {
      "epoch": 0.2762616067823981,
      "grad_norm": 117.85246276855469,
      "learning_rate": 1.698502727938019e-05,
      "loss": 5.7671,
      "step": 6843
    },
    {
      "epoch": 0.2763019781994348,
      "grad_norm": 86.78941345214844,
      "learning_rate": 1.698409150547414e-05,
      "loss": 3.8508,
      "step": 6844
    },
    {
      "epoch": 0.2763423496164715,
      "grad_norm": 88.7052001953125,
      "learning_rate": 1.6983155612156434e-05,
      "loss": 5.341,
      "step": 6845
    },
    {
      "epoch": 0.27638272103350825,
      "grad_norm": 97.7032241821289,
      "learning_rate": 1.698221959944308e-05,
      "loss": 6.2453,
      "step": 6846
    },
    {
      "epoch": 0.276423092450545,
      "grad_norm": 156.91995239257812,
      "learning_rate": 1.6981283467350066e-05,
      "loss": 6.2435,
      "step": 6847
    },
    {
      "epoch": 0.27646346386758175,
      "grad_norm": 86.88849639892578,
      "learning_rate": 1.6980347215893417e-05,
      "loss": 6.3871,
      "step": 6848
    },
    {
      "epoch": 0.27650383528461847,
      "grad_norm": 197.14578247070312,
      "learning_rate": 1.6979410845089125e-05,
      "loss": 5.5693,
      "step": 6849
    },
    {
      "epoch": 0.27654420670165525,
      "grad_norm": 76.18880462646484,
      "learning_rate": 1.6978474354953208e-05,
      "loss": 6.6788,
      "step": 6850
    },
    {
      "epoch": 0.27658457811869197,
      "grad_norm": 78.44642639160156,
      "learning_rate": 1.6977537745501672e-05,
      "loss": 4.9883,
      "step": 6851
    },
    {
      "epoch": 0.2766249495357287,
      "grad_norm": 85.55854797363281,
      "learning_rate": 1.697660101675054e-05,
      "loss": 5.7077,
      "step": 6852
    },
    {
      "epoch": 0.2766653209527654,
      "grad_norm": 91.14894104003906,
      "learning_rate": 1.6975664168715816e-05,
      "loss": 5.5788,
      "step": 6853
    },
    {
      "epoch": 0.2767056923698022,
      "grad_norm": 81.55644989013672,
      "learning_rate": 1.6974727201413528e-05,
      "loss": 3.0865,
      "step": 6854
    },
    {
      "epoch": 0.2767460637868389,
      "grad_norm": 65.16583251953125,
      "learning_rate": 1.6973790114859693e-05,
      "loss": 4.9663,
      "step": 6855
    },
    {
      "epoch": 0.27678643520387564,
      "grad_norm": 78.98159790039062,
      "learning_rate": 1.6972852909070328e-05,
      "loss": 6.248,
      "step": 6856
    },
    {
      "epoch": 0.2768268066209124,
      "grad_norm": 166.5556182861328,
      "learning_rate": 1.6971915584061466e-05,
      "loss": 5.485,
      "step": 6857
    },
    {
      "epoch": 0.27686717803794914,
      "grad_norm": 187.47474670410156,
      "learning_rate": 1.6970978139849125e-05,
      "loss": 6.2439,
      "step": 6858
    },
    {
      "epoch": 0.27690754945498586,
      "grad_norm": 88.10169982910156,
      "learning_rate": 1.6970040576449334e-05,
      "loss": 6.238,
      "step": 6859
    },
    {
      "epoch": 0.27694792087202263,
      "grad_norm": 77.0615005493164,
      "learning_rate": 1.696910289387813e-05,
      "loss": 6.7566,
      "step": 6860
    },
    {
      "epoch": 0.27698829228905936,
      "grad_norm": 113.13947296142578,
      "learning_rate": 1.6968165092151535e-05,
      "loss": 4.929,
      "step": 6861
    },
    {
      "epoch": 0.2770286637060961,
      "grad_norm": 86.61907958984375,
      "learning_rate": 1.6967227171285593e-05,
      "loss": 7.5231,
      "step": 6862
    },
    {
      "epoch": 0.2770690351231328,
      "grad_norm": 104.00553131103516,
      "learning_rate": 1.6966289131296335e-05,
      "loss": 5.0307,
      "step": 6863
    },
    {
      "epoch": 0.2771094065401696,
      "grad_norm": 84.1974105834961,
      "learning_rate": 1.69653509721998e-05,
      "loss": 6.3039,
      "step": 6864
    },
    {
      "epoch": 0.2771497779572063,
      "grad_norm": 92.65310668945312,
      "learning_rate": 1.696441269401203e-05,
      "loss": 7.0983,
      "step": 6865
    },
    {
      "epoch": 0.277190149374243,
      "grad_norm": 101.0923843383789,
      "learning_rate": 1.6963474296749065e-05,
      "loss": 6.2237,
      "step": 6866
    },
    {
      "epoch": 0.2772305207912798,
      "grad_norm": 81.23795318603516,
      "learning_rate": 1.696253578042695e-05,
      "loss": 5.4327,
      "step": 6867
    },
    {
      "epoch": 0.2772708922083165,
      "grad_norm": 96.87347412109375,
      "learning_rate": 1.6961597145061733e-05,
      "loss": 6.55,
      "step": 6868
    },
    {
      "epoch": 0.27731126362535324,
      "grad_norm": 116.43115234375,
      "learning_rate": 1.696065839066946e-05,
      "loss": 5.2714,
      "step": 6869
    },
    {
      "epoch": 0.27735163504238997,
      "grad_norm": 93.24645233154297,
      "learning_rate": 1.6959719517266183e-05,
      "loss": 5.4343,
      "step": 6870
    },
    {
      "epoch": 0.27739200645942674,
      "grad_norm": 92.23976135253906,
      "learning_rate": 1.6958780524867953e-05,
      "loss": 3.0469,
      "step": 6871
    },
    {
      "epoch": 0.27743237787646347,
      "grad_norm": 110.95112609863281,
      "learning_rate": 1.695784141349083e-05,
      "loss": 6.145,
      "step": 6872
    },
    {
      "epoch": 0.2774727492935002,
      "grad_norm": 97.68108367919922,
      "learning_rate": 1.6956902183150864e-05,
      "loss": 6.9421,
      "step": 6873
    },
    {
      "epoch": 0.27751312071053696,
      "grad_norm": 67.60233306884766,
      "learning_rate": 1.695596283386412e-05,
      "loss": 4.6875,
      "step": 6874
    },
    {
      "epoch": 0.2775534921275737,
      "grad_norm": 90.38546752929688,
      "learning_rate": 1.6955023365646646e-05,
      "loss": 5.0875,
      "step": 6875
    },
    {
      "epoch": 0.2775938635446104,
      "grad_norm": 116.4218521118164,
      "learning_rate": 1.6954083778514522e-05,
      "loss": 5.0701,
      "step": 6876
    },
    {
      "epoch": 0.27763423496164713,
      "grad_norm": 115.81974029541016,
      "learning_rate": 1.69531440724838e-05,
      "loss": 5.2565,
      "step": 6877
    },
    {
      "epoch": 0.2776746063786839,
      "grad_norm": 167.61126708984375,
      "learning_rate": 1.6952204247570554e-05,
      "loss": 6.8242,
      "step": 6878
    },
    {
      "epoch": 0.27771497779572063,
      "grad_norm": 122.87470245361328,
      "learning_rate": 1.6951264303790846e-05,
      "loss": 4.3378,
      "step": 6879
    },
    {
      "epoch": 0.27775534921275735,
      "grad_norm": 84.29102325439453,
      "learning_rate": 1.6950324241160754e-05,
      "loss": 4.8108,
      "step": 6880
    },
    {
      "epoch": 0.27779572062979413,
      "grad_norm": 71.39672088623047,
      "learning_rate": 1.6949384059696345e-05,
      "loss": 4.9551,
      "step": 6881
    },
    {
      "epoch": 0.27783609204683085,
      "grad_norm": 79.1923599243164,
      "learning_rate": 1.69484437594137e-05,
      "loss": 5.9792,
      "step": 6882
    },
    {
      "epoch": 0.2778764634638676,
      "grad_norm": 59.91400909423828,
      "learning_rate": 1.694750334032889e-05,
      "loss": 6.0587,
      "step": 6883
    },
    {
      "epoch": 0.2779168348809043,
      "grad_norm": 96.5875473022461,
      "learning_rate": 1.6946562802458e-05,
      "loss": 5.8206,
      "step": 6884
    },
    {
      "epoch": 0.2779572062979411,
      "grad_norm": 112.7123031616211,
      "learning_rate": 1.69456221458171e-05,
      "loss": 6.9244,
      "step": 6885
    },
    {
      "epoch": 0.2779975777149778,
      "grad_norm": 166.78985595703125,
      "learning_rate": 1.694468137042229e-05,
      "loss": 3.423,
      "step": 6886
    },
    {
      "epoch": 0.2780379491320145,
      "grad_norm": 145.48843383789062,
      "learning_rate": 1.694374047628964e-05,
      "loss": 6.1347,
      "step": 6887
    },
    {
      "epoch": 0.2780783205490513,
      "grad_norm": 61.83201599121094,
      "learning_rate": 1.6942799463435243e-05,
      "loss": 6.5347,
      "step": 6888
    },
    {
      "epoch": 0.278118691966088,
      "grad_norm": 82.84354400634766,
      "learning_rate": 1.6941858331875186e-05,
      "loss": 5.9529,
      "step": 6889
    },
    {
      "epoch": 0.27815906338312474,
      "grad_norm": 118.0797119140625,
      "learning_rate": 1.6940917081625565e-05,
      "loss": 4.6131,
      "step": 6890
    },
    {
      "epoch": 0.27819943480016146,
      "grad_norm": 78.01482391357422,
      "learning_rate": 1.693997571270247e-05,
      "loss": 5.7917,
      "step": 6891
    },
    {
      "epoch": 0.27823980621719824,
      "grad_norm": 113.27249908447266,
      "learning_rate": 1.6939034225122e-05,
      "loss": 6.4774,
      "step": 6892
    },
    {
      "epoch": 0.27828017763423496,
      "grad_norm": 93.48929595947266,
      "learning_rate": 1.693809261890024e-05,
      "loss": 4.5606,
      "step": 6893
    },
    {
      "epoch": 0.2783205490512717,
      "grad_norm": 84.82164001464844,
      "learning_rate": 1.6937150894053306e-05,
      "loss": 4.8763,
      "step": 6894
    },
    {
      "epoch": 0.27836092046830846,
      "grad_norm": 92.06336212158203,
      "learning_rate": 1.6936209050597283e-05,
      "loss": 5.0371,
      "step": 6895
    },
    {
      "epoch": 0.2784012918853452,
      "grad_norm": 102.8565444946289,
      "learning_rate": 1.693526708854829e-05,
      "loss": 5.2804,
      "step": 6896
    },
    {
      "epoch": 0.2784416633023819,
      "grad_norm": 101.42584228515625,
      "learning_rate": 1.6934325007922418e-05,
      "loss": 4.3115,
      "step": 6897
    },
    {
      "epoch": 0.2784820347194186,
      "grad_norm": 99.83280181884766,
      "learning_rate": 1.6933382808735786e-05,
      "loss": 4.3461,
      "step": 6898
    },
    {
      "epoch": 0.2785224061364554,
      "grad_norm": 90.30842590332031,
      "learning_rate": 1.6932440491004495e-05,
      "loss": 4.8627,
      "step": 6899
    },
    {
      "epoch": 0.2785627775534921,
      "grad_norm": 145.2998809814453,
      "learning_rate": 1.6931498054744664e-05,
      "loss": 5.228,
      "step": 6900
    },
    {
      "epoch": 0.27860314897052885,
      "grad_norm": 104.25545501708984,
      "learning_rate": 1.6930555499972402e-05,
      "loss": 5.9224,
      "step": 6901
    },
    {
      "epoch": 0.2786435203875656,
      "grad_norm": 78.08849334716797,
      "learning_rate": 1.6929612826703822e-05,
      "loss": 5.4011,
      "step": 6902
    },
    {
      "epoch": 0.27868389180460235,
      "grad_norm": 87.45816802978516,
      "learning_rate": 1.6928670034955045e-05,
      "loss": 5.9102,
      "step": 6903
    },
    {
      "epoch": 0.27872426322163907,
      "grad_norm": 103.39178466796875,
      "learning_rate": 1.6927727124742194e-05,
      "loss": 6.4881,
      "step": 6904
    },
    {
      "epoch": 0.2787646346386758,
      "grad_norm": 133.6716766357422,
      "learning_rate": 1.692678409608138e-05,
      "loss": 6.0086,
      "step": 6905
    },
    {
      "epoch": 0.27880500605571257,
      "grad_norm": 127.81536865234375,
      "learning_rate": 1.692584094898874e-05,
      "loss": 6.4787,
      "step": 6906
    },
    {
      "epoch": 0.2788453774727493,
      "grad_norm": 77.13336181640625,
      "learning_rate": 1.692489768348039e-05,
      "loss": 6.4383,
      "step": 6907
    },
    {
      "epoch": 0.278885748889786,
      "grad_norm": 155.9879913330078,
      "learning_rate": 1.6923954299572457e-05,
      "loss": 5.0673,
      "step": 6908
    },
    {
      "epoch": 0.2789261203068228,
      "grad_norm": 73.27352142333984,
      "learning_rate": 1.692301079728108e-05,
      "loss": 3.5226,
      "step": 6909
    },
    {
      "epoch": 0.2789664917238595,
      "grad_norm": 125.62061309814453,
      "learning_rate": 1.692206717662238e-05,
      "loss": 4.6496,
      "step": 6910
    },
    {
      "epoch": 0.27900686314089623,
      "grad_norm": 95.91007995605469,
      "learning_rate": 1.6921123437612497e-05,
      "loss": 3.9624,
      "step": 6911
    },
    {
      "epoch": 0.279047234557933,
      "grad_norm": 93.75267791748047,
      "learning_rate": 1.6920179580267568e-05,
      "loss": 5.0754,
      "step": 6912
    },
    {
      "epoch": 0.27908760597496973,
      "grad_norm": 223.56741333007812,
      "learning_rate": 1.6919235604603723e-05,
      "loss": 5.8278,
      "step": 6913
    },
    {
      "epoch": 0.27912797739200645,
      "grad_norm": 172.69985961914062,
      "learning_rate": 1.691829151063711e-05,
      "loss": 4.2477,
      "step": 6914
    },
    {
      "epoch": 0.2791683488090432,
      "grad_norm": 134.0113067626953,
      "learning_rate": 1.6917347298383866e-05,
      "loss": 4.2364,
      "step": 6915
    },
    {
      "epoch": 0.27920872022607995,
      "grad_norm": 87.38469696044922,
      "learning_rate": 1.6916402967860135e-05,
      "loss": 4.7066,
      "step": 6916
    },
    {
      "epoch": 0.2792490916431167,
      "grad_norm": 87.33341979980469,
      "learning_rate": 1.6915458519082067e-05,
      "loss": 5.5026,
      "step": 6917
    },
    {
      "epoch": 0.2792894630601534,
      "grad_norm": 80.30044555664062,
      "learning_rate": 1.6914513952065806e-05,
      "loss": 5.3972,
      "step": 6918
    },
    {
      "epoch": 0.2793298344771902,
      "grad_norm": 86.0880126953125,
      "learning_rate": 1.6913569266827507e-05,
      "loss": 3.9074,
      "step": 6919
    },
    {
      "epoch": 0.2793702058942269,
      "grad_norm": 79.6326675415039,
      "learning_rate": 1.6912624463383313e-05,
      "loss": 5.4656,
      "step": 6920
    },
    {
      "epoch": 0.2794105773112636,
      "grad_norm": 70.07650756835938,
      "learning_rate": 1.6911679541749383e-05,
      "loss": 6.0975,
      "step": 6921
    },
    {
      "epoch": 0.27945094872830034,
      "grad_norm": 82.38229370117188,
      "learning_rate": 1.6910734501941875e-05,
      "loss": 4.5823,
      "step": 6922
    },
    {
      "epoch": 0.2794913201453371,
      "grad_norm": 117.87053680419922,
      "learning_rate": 1.6909789343976944e-05,
      "loss": 5.446,
      "step": 6923
    },
    {
      "epoch": 0.27953169156237384,
      "grad_norm": 96.91514587402344,
      "learning_rate": 1.690884406787075e-05,
      "loss": 5.4416,
      "step": 6924
    },
    {
      "epoch": 0.27957206297941056,
      "grad_norm": 106.8963623046875,
      "learning_rate": 1.6907898673639454e-05,
      "loss": 5.3003,
      "step": 6925
    },
    {
      "epoch": 0.27961243439644734,
      "grad_norm": 75.13604736328125,
      "learning_rate": 1.6906953161299228e-05,
      "loss": 6.7984,
      "step": 6926
    },
    {
      "epoch": 0.27965280581348406,
      "grad_norm": 89.5090103149414,
      "learning_rate": 1.6906007530866224e-05,
      "loss": 3.5489,
      "step": 6927
    },
    {
      "epoch": 0.2796931772305208,
      "grad_norm": 89.76486206054688,
      "learning_rate": 1.6905061782356624e-05,
      "loss": 5.9228,
      "step": 6928
    },
    {
      "epoch": 0.2797335486475575,
      "grad_norm": 102.18414306640625,
      "learning_rate": 1.690411591578659e-05,
      "loss": 8.0988,
      "step": 6929
    },
    {
      "epoch": 0.2797739200645943,
      "grad_norm": 111.33623504638672,
      "learning_rate": 1.6903169931172294e-05,
      "loss": 4.3988,
      "step": 6930
    },
    {
      "epoch": 0.279814291481631,
      "grad_norm": 105.77313995361328,
      "learning_rate": 1.6902223828529914e-05,
      "loss": 6.0138,
      "step": 6931
    },
    {
      "epoch": 0.2798546628986677,
      "grad_norm": 81.35370635986328,
      "learning_rate": 1.6901277607875625e-05,
      "loss": 4.39,
      "step": 6932
    },
    {
      "epoch": 0.2798950343157045,
      "grad_norm": 129.7156524658203,
      "learning_rate": 1.6900331269225603e-05,
      "loss": 5.3677,
      "step": 6933
    },
    {
      "epoch": 0.2799354057327412,
      "grad_norm": 268.619873046875,
      "learning_rate": 1.6899384812596032e-05,
      "loss": 4.395,
      "step": 6934
    },
    {
      "epoch": 0.27997577714977795,
      "grad_norm": 100.1133804321289,
      "learning_rate": 1.6898438238003086e-05,
      "loss": 5.5482,
      "step": 6935
    },
    {
      "epoch": 0.28001614856681467,
      "grad_norm": 142.09336853027344,
      "learning_rate": 1.6897491545462964e-05,
      "loss": 5.5954,
      "step": 6936
    },
    {
      "epoch": 0.28005651998385145,
      "grad_norm": 115.44325256347656,
      "learning_rate": 1.6896544734991837e-05,
      "loss": 5.1843,
      "step": 6937
    },
    {
      "epoch": 0.28009689140088817,
      "grad_norm": 106.04704284667969,
      "learning_rate": 1.6895597806605904e-05,
      "loss": 6.0898,
      "step": 6938
    },
    {
      "epoch": 0.2801372628179249,
      "grad_norm": 85.98809051513672,
      "learning_rate": 1.6894650760321347e-05,
      "loss": 4.0259,
      "step": 6939
    },
    {
      "epoch": 0.28017763423496167,
      "grad_norm": 76.7108154296875,
      "learning_rate": 1.6893703596154364e-05,
      "loss": 4.1351,
      "step": 6940
    },
    {
      "epoch": 0.2802180056519984,
      "grad_norm": 118.43062591552734,
      "learning_rate": 1.6892756314121147e-05,
      "loss": 4.8113,
      "step": 6941
    },
    {
      "epoch": 0.2802583770690351,
      "grad_norm": 110.89665985107422,
      "learning_rate": 1.6891808914237895e-05,
      "loss": 5.8597,
      "step": 6942
    },
    {
      "epoch": 0.28029874848607184,
      "grad_norm": 93.68592834472656,
      "learning_rate": 1.68908613965208e-05,
      "loss": 5.8051,
      "step": 6943
    },
    {
      "epoch": 0.2803391199031086,
      "grad_norm": 105.01364135742188,
      "learning_rate": 1.6889913760986066e-05,
      "loss": 4.3558,
      "step": 6944
    },
    {
      "epoch": 0.28037949132014534,
      "grad_norm": 111.5158920288086,
      "learning_rate": 1.68889660076499e-05,
      "loss": 3.8262,
      "step": 6945
    },
    {
      "epoch": 0.28041986273718206,
      "grad_norm": 104.57118225097656,
      "learning_rate": 1.6888018136528505e-05,
      "loss": 4.5435,
      "step": 6946
    },
    {
      "epoch": 0.28046023415421883,
      "grad_norm": 90.77667999267578,
      "learning_rate": 1.6887070147638077e-05,
      "loss": 6.0084,
      "step": 6947
    },
    {
      "epoch": 0.28050060557125556,
      "grad_norm": 154.10765075683594,
      "learning_rate": 1.6886122040994837e-05,
      "loss": 4.2129,
      "step": 6948
    },
    {
      "epoch": 0.2805409769882923,
      "grad_norm": 123.21892547607422,
      "learning_rate": 1.6885173816614986e-05,
      "loss": 5.2234,
      "step": 6949
    },
    {
      "epoch": 0.280581348405329,
      "grad_norm": 88.6094970703125,
      "learning_rate": 1.688422547451474e-05,
      "loss": 4.8878,
      "step": 6950
    },
    {
      "epoch": 0.2806217198223658,
      "grad_norm": 98.69248962402344,
      "learning_rate": 1.688327701471032e-05,
      "loss": 6.4263,
      "step": 6951
    },
    {
      "epoch": 0.2806620912394025,
      "grad_norm": 142.51797485351562,
      "learning_rate": 1.6882328437217935e-05,
      "loss": 5.9829,
      "step": 6952
    },
    {
      "epoch": 0.2807024626564392,
      "grad_norm": 82.6432113647461,
      "learning_rate": 1.68813797420538e-05,
      "loss": 4.0877,
      "step": 6953
    },
    {
      "epoch": 0.280742834073476,
      "grad_norm": 142.3252410888672,
      "learning_rate": 1.6880430929234148e-05,
      "loss": 3.9784,
      "step": 6954
    },
    {
      "epoch": 0.2807832054905127,
      "grad_norm": 92.1093521118164,
      "learning_rate": 1.687948199877519e-05,
      "loss": 5.889,
      "step": 6955
    },
    {
      "epoch": 0.28082357690754944,
      "grad_norm": 191.6770782470703,
      "learning_rate": 1.6878532950693156e-05,
      "loss": 5.3803,
      "step": 6956
    },
    {
      "epoch": 0.28086394832458617,
      "grad_norm": 90.8195571899414,
      "learning_rate": 1.687758378500427e-05,
      "loss": 5.4128,
      "step": 6957
    },
    {
      "epoch": 0.28090431974162294,
      "grad_norm": 80.10398864746094,
      "learning_rate": 1.6876634501724764e-05,
      "loss": 6.0138,
      "step": 6958
    },
    {
      "epoch": 0.28094469115865967,
      "grad_norm": 117.2073745727539,
      "learning_rate": 1.687568510087086e-05,
      "loss": 5.6744,
      "step": 6959
    },
    {
      "epoch": 0.2809850625756964,
      "grad_norm": 71.37039947509766,
      "learning_rate": 1.6874735582458803e-05,
      "loss": 5.7354,
      "step": 6960
    },
    {
      "epoch": 0.28102543399273316,
      "grad_norm": 88.49369049072266,
      "learning_rate": 1.6873785946504817e-05,
      "loss": 5.0247,
      "step": 6961
    },
    {
      "epoch": 0.2810658054097699,
      "grad_norm": 104.90386962890625,
      "learning_rate": 1.6872836193025145e-05,
      "loss": 6.864,
      "step": 6962
    },
    {
      "epoch": 0.2811061768268066,
      "grad_norm": 123.45093536376953,
      "learning_rate": 1.6871886322036023e-05,
      "loss": 1.7587,
      "step": 6963
    },
    {
      "epoch": 0.2811465482438434,
      "grad_norm": 145.75674438476562,
      "learning_rate": 1.6870936333553694e-05,
      "loss": 5.2926,
      "step": 6964
    },
    {
      "epoch": 0.2811869196608801,
      "grad_norm": 255.54049682617188,
      "learning_rate": 1.6869986227594396e-05,
      "loss": 4.2742,
      "step": 6965
    },
    {
      "epoch": 0.28122729107791683,
      "grad_norm": 123.5911636352539,
      "learning_rate": 1.6869036004174377e-05,
      "loss": 5.6117,
      "step": 6966
    },
    {
      "epoch": 0.28126766249495355,
      "grad_norm": 112.665771484375,
      "learning_rate": 1.6868085663309882e-05,
      "loss": 5.6497,
      "step": 6967
    },
    {
      "epoch": 0.28130803391199033,
      "grad_norm": 91.86866760253906,
      "learning_rate": 1.686713520501716e-05,
      "loss": 5.9532,
      "step": 6968
    },
    {
      "epoch": 0.28134840532902705,
      "grad_norm": 136.60401916503906,
      "learning_rate": 1.686618462931246e-05,
      "loss": 5.7492,
      "step": 6969
    },
    {
      "epoch": 0.2813887767460638,
      "grad_norm": 67.20311737060547,
      "learning_rate": 1.686523393621204e-05,
      "loss": 6.1941,
      "step": 6970
    },
    {
      "epoch": 0.28142914816310055,
      "grad_norm": 86.65936279296875,
      "learning_rate": 1.6864283125732152e-05,
      "loss": 5.0465,
      "step": 6971
    },
    {
      "epoch": 0.2814695195801373,
      "grad_norm": 141.0764617919922,
      "learning_rate": 1.6863332197889047e-05,
      "loss": 4.1224,
      "step": 6972
    },
    {
      "epoch": 0.281509890997174,
      "grad_norm": 163.3915557861328,
      "learning_rate": 1.686238115269899e-05,
      "loss": 4.7619,
      "step": 6973
    },
    {
      "epoch": 0.2815502624142107,
      "grad_norm": 81.93791961669922,
      "learning_rate": 1.6861429990178243e-05,
      "loss": 6.2563,
      "step": 6974
    },
    {
      "epoch": 0.2815906338312475,
      "grad_norm": 213.77474975585938,
      "learning_rate": 1.6860478710343065e-05,
      "loss": 5.5695,
      "step": 6975
    },
    {
      "epoch": 0.2816310052482842,
      "grad_norm": 98.25362396240234,
      "learning_rate": 1.685952731320972e-05,
      "loss": 4.4714,
      "step": 6976
    },
    {
      "epoch": 0.28167137666532094,
      "grad_norm": 199.99176025390625,
      "learning_rate": 1.6858575798794475e-05,
      "loss": 4.8083,
      "step": 6977
    },
    {
      "epoch": 0.2817117480823577,
      "grad_norm": 84.30486297607422,
      "learning_rate": 1.6857624167113605e-05,
      "loss": 5.0122,
      "step": 6978
    },
    {
      "epoch": 0.28175211949939444,
      "grad_norm": 104.88996124267578,
      "learning_rate": 1.6856672418183372e-05,
      "loss": 5.3227,
      "step": 6979
    },
    {
      "epoch": 0.28179249091643116,
      "grad_norm": 91.39482116699219,
      "learning_rate": 1.685572055202005e-05,
      "loss": 5.9461,
      "step": 6980
    },
    {
      "epoch": 0.2818328623334679,
      "grad_norm": 102.620361328125,
      "learning_rate": 1.685476856863992e-05,
      "loss": 6.5235,
      "step": 6981
    },
    {
      "epoch": 0.28187323375050466,
      "grad_norm": 88.89498138427734,
      "learning_rate": 1.685381646805925e-05,
      "loss": 5.5811,
      "step": 6982
    },
    {
      "epoch": 0.2819136051675414,
      "grad_norm": 102.10384368896484,
      "learning_rate": 1.6852864250294327e-05,
      "loss": 3.6225,
      "step": 6983
    },
    {
      "epoch": 0.2819539765845781,
      "grad_norm": 109.74011993408203,
      "learning_rate": 1.6851911915361425e-05,
      "loss": 5.599,
      "step": 6984
    },
    {
      "epoch": 0.2819943480016149,
      "grad_norm": 77.00582885742188,
      "learning_rate": 1.685095946327683e-05,
      "loss": 5.7782,
      "step": 6985
    },
    {
      "epoch": 0.2820347194186516,
      "grad_norm": 82.98307800292969,
      "learning_rate": 1.685000689405683e-05,
      "loss": 3.4476,
      "step": 6986
    },
    {
      "epoch": 0.2820750908356883,
      "grad_norm": 201.38475036621094,
      "learning_rate": 1.6849054207717705e-05,
      "loss": 4.656,
      "step": 6987
    },
    {
      "epoch": 0.28211546225272505,
      "grad_norm": 283.56048583984375,
      "learning_rate": 1.6848101404275746e-05,
      "loss": 4.4257,
      "step": 6988
    },
    {
      "epoch": 0.2821558336697618,
      "grad_norm": 105.81951141357422,
      "learning_rate": 1.6847148483747243e-05,
      "loss": 5.3931,
      "step": 6989
    },
    {
      "epoch": 0.28219620508679855,
      "grad_norm": 87.74962615966797,
      "learning_rate": 1.6846195446148497e-05,
      "loss": 4.8774,
      "step": 6990
    },
    {
      "epoch": 0.28223657650383527,
      "grad_norm": 118.7925033569336,
      "learning_rate": 1.684524229149579e-05,
      "loss": 6.1362,
      "step": 6991
    },
    {
      "epoch": 0.28227694792087205,
      "grad_norm": 90.42003631591797,
      "learning_rate": 1.6844289019805423e-05,
      "loss": 5.5836,
      "step": 6992
    },
    {
      "epoch": 0.28231731933790877,
      "grad_norm": 72.45135498046875,
      "learning_rate": 1.6843335631093706e-05,
      "loss": 5.7559,
      "step": 6993
    },
    {
      "epoch": 0.2823576907549455,
      "grad_norm": 89.1651611328125,
      "learning_rate": 1.6842382125376923e-05,
      "loss": 6.5992,
      "step": 6994
    },
    {
      "epoch": 0.2823980621719822,
      "grad_norm": 99.82646942138672,
      "learning_rate": 1.684142850267138e-05,
      "loss": 5.6914,
      "step": 6995
    },
    {
      "epoch": 0.282438433589019,
      "grad_norm": 100.45820617675781,
      "learning_rate": 1.6840474762993392e-05,
      "loss": 5.834,
      "step": 6996
    },
    {
      "epoch": 0.2824788050060557,
      "grad_norm": 92.0634536743164,
      "learning_rate": 1.6839520906359255e-05,
      "loss": 3.5762,
      "step": 6997
    },
    {
      "epoch": 0.28251917642309243,
      "grad_norm": 72.74417877197266,
      "learning_rate": 1.6838566932785284e-05,
      "loss": 4.9771,
      "step": 6998
    },
    {
      "epoch": 0.2825595478401292,
      "grad_norm": 226.05422973632812,
      "learning_rate": 1.683761284228779e-05,
      "loss": 6.0649,
      "step": 6999
    },
    {
      "epoch": 0.28259991925716593,
      "grad_norm": 166.88308715820312,
      "learning_rate": 1.683665863488308e-05,
      "loss": 6.8027,
      "step": 7000
    },
    {
      "epoch": 0.28264029067420265,
      "grad_norm": 95.8824234008789,
      "learning_rate": 1.6835704310587476e-05,
      "loss": 3.3416,
      "step": 7001
    },
    {
      "epoch": 0.2826806620912394,
      "grad_norm": 117.96073150634766,
      "learning_rate": 1.6834749869417284e-05,
      "loss": 7.4344,
      "step": 7002
    },
    {
      "epoch": 0.28272103350827615,
      "grad_norm": 77.55817413330078,
      "learning_rate": 1.6833795311388835e-05,
      "loss": 5.8996,
      "step": 7003
    },
    {
      "epoch": 0.2827614049253129,
      "grad_norm": 139.91317749023438,
      "learning_rate": 1.6832840636518445e-05,
      "loss": 4.9146,
      "step": 7004
    },
    {
      "epoch": 0.2828017763423496,
      "grad_norm": 83.5267105102539,
      "learning_rate": 1.683188584482243e-05,
      "loss": 5.8208,
      "step": 7005
    },
    {
      "epoch": 0.2828421477593864,
      "grad_norm": 87.69722747802734,
      "learning_rate": 1.6830930936317125e-05,
      "loss": 6.5394,
      "step": 7006
    },
    {
      "epoch": 0.2828825191764231,
      "grad_norm": 172.0089874267578,
      "learning_rate": 1.682997591101885e-05,
      "loss": 4.6475,
      "step": 7007
    },
    {
      "epoch": 0.2829228905934598,
      "grad_norm": 115.60958862304688,
      "learning_rate": 1.682902076894394e-05,
      "loss": 6.4503,
      "step": 7008
    },
    {
      "epoch": 0.28296326201049654,
      "grad_norm": 68.2867660522461,
      "learning_rate": 1.682806551010872e-05,
      "loss": 5.1212,
      "step": 7009
    },
    {
      "epoch": 0.2830036334275333,
      "grad_norm": 98.97969055175781,
      "learning_rate": 1.6827110134529524e-05,
      "loss": 4.5934,
      "step": 7010
    },
    {
      "epoch": 0.28304400484457004,
      "grad_norm": 100.16387939453125,
      "learning_rate": 1.6826154642222684e-05,
      "loss": 6.6034,
      "step": 7011
    },
    {
      "epoch": 0.28308437626160676,
      "grad_norm": 73.51154327392578,
      "learning_rate": 1.6825199033204545e-05,
      "loss": 5.6254,
      "step": 7012
    },
    {
      "epoch": 0.28312474767864354,
      "grad_norm": 93.1387710571289,
      "learning_rate": 1.682424330749144e-05,
      "loss": 5.9396,
      "step": 7013
    },
    {
      "epoch": 0.28316511909568026,
      "grad_norm": 94.92105865478516,
      "learning_rate": 1.6823287465099706e-05,
      "loss": 6.214,
      "step": 7014
    },
    {
      "epoch": 0.283205490512717,
      "grad_norm": 78.5295639038086,
      "learning_rate": 1.6822331506045694e-05,
      "loss": 5.883,
      "step": 7015
    },
    {
      "epoch": 0.28324586192975376,
      "grad_norm": 108.79058837890625,
      "learning_rate": 1.6821375430345744e-05,
      "loss": 5.1747,
      "step": 7016
    },
    {
      "epoch": 0.2832862333467905,
      "grad_norm": 129.03817749023438,
      "learning_rate": 1.68204192380162e-05,
      "loss": 4.8675,
      "step": 7017
    },
    {
      "epoch": 0.2833266047638272,
      "grad_norm": 101.36924743652344,
      "learning_rate": 1.6819462929073414e-05,
      "loss": 5.7632,
      "step": 7018
    },
    {
      "epoch": 0.2833669761808639,
      "grad_norm": 122.05785369873047,
      "learning_rate": 1.6818506503533743e-05,
      "loss": 5.7908,
      "step": 7019
    },
    {
      "epoch": 0.2834073475979007,
      "grad_norm": 117.40648651123047,
      "learning_rate": 1.681754996141353e-05,
      "loss": 4.473,
      "step": 7020
    },
    {
      "epoch": 0.2834477190149374,
      "grad_norm": 75.12145233154297,
      "learning_rate": 1.681659330272913e-05,
      "loss": 5.9889,
      "step": 7021
    },
    {
      "epoch": 0.28348809043197415,
      "grad_norm": 171.64141845703125,
      "learning_rate": 1.68156365274969e-05,
      "loss": 4.322,
      "step": 7022
    },
    {
      "epoch": 0.2835284618490109,
      "grad_norm": 76.0806655883789,
      "learning_rate": 1.681467963573321e-05,
      "loss": 5.9606,
      "step": 7023
    },
    {
      "epoch": 0.28356883326604765,
      "grad_norm": 83.04637908935547,
      "learning_rate": 1.6813722627454402e-05,
      "loss": 6.5752,
      "step": 7024
    },
    {
      "epoch": 0.28360920468308437,
      "grad_norm": 82.57109069824219,
      "learning_rate": 1.6812765502676856e-05,
      "loss": 5.5158,
      "step": 7025
    },
    {
      "epoch": 0.2836495761001211,
      "grad_norm": 82.07806396484375,
      "learning_rate": 1.6811808261416925e-05,
      "loss": 6.6817,
      "step": 7026
    },
    {
      "epoch": 0.28368994751715787,
      "grad_norm": 245.60482788085938,
      "learning_rate": 1.681085090369098e-05,
      "loss": 4.2346,
      "step": 7027
    },
    {
      "epoch": 0.2837303189341946,
      "grad_norm": 155.5291748046875,
      "learning_rate": 1.680989342951539e-05,
      "loss": 5.413,
      "step": 7028
    },
    {
      "epoch": 0.2837706903512313,
      "grad_norm": 73.75446319580078,
      "learning_rate": 1.6808935838906523e-05,
      "loss": 4.8581,
      "step": 7029
    },
    {
      "epoch": 0.2838110617682681,
      "grad_norm": 90.5551528930664,
      "learning_rate": 1.6807978131880752e-05,
      "loss": 5.3284,
      "step": 7030
    },
    {
      "epoch": 0.2838514331853048,
      "grad_norm": 109.1418685913086,
      "learning_rate": 1.6807020308454456e-05,
      "loss": 4.9447,
      "step": 7031
    },
    {
      "epoch": 0.28389180460234154,
      "grad_norm": 117.90022277832031,
      "learning_rate": 1.6806062368644007e-05,
      "loss": 3.4356,
      "step": 7032
    },
    {
      "epoch": 0.28393217601937826,
      "grad_norm": 86.9512939453125,
      "learning_rate": 1.6805104312465786e-05,
      "loss": 5.5517,
      "step": 7033
    },
    {
      "epoch": 0.28397254743641503,
      "grad_norm": 173.89154052734375,
      "learning_rate": 1.680414613993617e-05,
      "loss": 5.0163,
      "step": 7034
    },
    {
      "epoch": 0.28401291885345176,
      "grad_norm": 74.84910583496094,
      "learning_rate": 1.6803187851071546e-05,
      "loss": 4.8815,
      "step": 7035
    },
    {
      "epoch": 0.2840532902704885,
      "grad_norm": 87.99168395996094,
      "learning_rate": 1.6802229445888293e-05,
      "loss": 6.4282,
      "step": 7036
    },
    {
      "epoch": 0.28409366168752526,
      "grad_norm": 109.26384735107422,
      "learning_rate": 1.6801270924402806e-05,
      "loss": 5.7194,
      "step": 7037
    },
    {
      "epoch": 0.284134033104562,
      "grad_norm": 105.27792358398438,
      "learning_rate": 1.6800312286631465e-05,
      "loss": 6.3967,
      "step": 7038
    },
    {
      "epoch": 0.2841744045215987,
      "grad_norm": 69.54723358154297,
      "learning_rate": 1.6799353532590663e-05,
      "loss": 5.7829,
      "step": 7039
    },
    {
      "epoch": 0.2842147759386354,
      "grad_norm": 80.58367156982422,
      "learning_rate": 1.6798394662296795e-05,
      "loss": 4.8221,
      "step": 7040
    },
    {
      "epoch": 0.2842551473556722,
      "grad_norm": 108.89454650878906,
      "learning_rate": 1.679743567576625e-05,
      "loss": 5.3813,
      "step": 7041
    },
    {
      "epoch": 0.2842955187727089,
      "grad_norm": 91.17406463623047,
      "learning_rate": 1.6796476573015433e-05,
      "loss": 5.8897,
      "step": 7042
    },
    {
      "epoch": 0.28433589018974564,
      "grad_norm": 102.93927764892578,
      "learning_rate": 1.6795517354060736e-05,
      "loss": 6.9656,
      "step": 7043
    },
    {
      "epoch": 0.2843762616067824,
      "grad_norm": 96.72370147705078,
      "learning_rate": 1.679455801891856e-05,
      "loss": 4.5581,
      "step": 7044
    },
    {
      "epoch": 0.28441663302381914,
      "grad_norm": 82.61998748779297,
      "learning_rate": 1.679359856760531e-05,
      "loss": 4.6156,
      "step": 7045
    },
    {
      "epoch": 0.28445700444085587,
      "grad_norm": 95.4020767211914,
      "learning_rate": 1.6792639000137384e-05,
      "loss": 5.1698,
      "step": 7046
    },
    {
      "epoch": 0.2844973758578926,
      "grad_norm": 92.12443542480469,
      "learning_rate": 1.67916793165312e-05,
      "loss": 6.7665,
      "step": 7047
    },
    {
      "epoch": 0.28453774727492936,
      "grad_norm": 106.27957153320312,
      "learning_rate": 1.6790719516803157e-05,
      "loss": 4.3057,
      "step": 7048
    },
    {
      "epoch": 0.2845781186919661,
      "grad_norm": 535.7071533203125,
      "learning_rate": 1.6789759600969664e-05,
      "loss": 6.6028,
      "step": 7049
    },
    {
      "epoch": 0.2846184901090028,
      "grad_norm": 171.69073486328125,
      "learning_rate": 1.678879956904714e-05,
      "loss": 5.09,
      "step": 7050
    },
    {
      "epoch": 0.2846588615260396,
      "grad_norm": 94.07962036132812,
      "learning_rate": 1.6787839421052e-05,
      "loss": 6.0618,
      "step": 7051
    },
    {
      "epoch": 0.2846992329430763,
      "grad_norm": 96.47969055175781,
      "learning_rate": 1.678687915700065e-05,
      "loss": 6.0093,
      "step": 7052
    },
    {
      "epoch": 0.28473960436011303,
      "grad_norm": 101.48588562011719,
      "learning_rate": 1.678591877690952e-05,
      "loss": 5.4179,
      "step": 7053
    },
    {
      "epoch": 0.28477997577714975,
      "grad_norm": 108.37828826904297,
      "learning_rate": 1.6784958280795018e-05,
      "loss": 6.9691,
      "step": 7054
    },
    {
      "epoch": 0.28482034719418653,
      "grad_norm": 131.81324768066406,
      "learning_rate": 1.678399766867358e-05,
      "loss": 6.0904,
      "step": 7055
    },
    {
      "epoch": 0.28486071861122325,
      "grad_norm": 95.64656829833984,
      "learning_rate": 1.678303694056162e-05,
      "loss": 4.8448,
      "step": 7056
    },
    {
      "epoch": 0.28490109002826,
      "grad_norm": 185.91836547851562,
      "learning_rate": 1.6782076096475576e-05,
      "loss": 6.153,
      "step": 7057
    },
    {
      "epoch": 0.28494146144529675,
      "grad_norm": 83.58366394042969,
      "learning_rate": 1.678111513643186e-05,
      "loss": 5.933,
      "step": 7058
    },
    {
      "epoch": 0.2849818328623335,
      "grad_norm": 96.8852767944336,
      "learning_rate": 1.678015406044691e-05,
      "loss": 4.8421,
      "step": 7059
    },
    {
      "epoch": 0.2850222042793702,
      "grad_norm": 106.58528900146484,
      "learning_rate": 1.677919286853716e-05,
      "loss": 6.5575,
      "step": 7060
    },
    {
      "epoch": 0.2850625756964069,
      "grad_norm": 90.743408203125,
      "learning_rate": 1.6778231560719046e-05,
      "loss": 5.7133,
      "step": 7061
    },
    {
      "epoch": 0.2851029471134437,
      "grad_norm": 86.81787872314453,
      "learning_rate": 1.6777270137008996e-05,
      "loss": 5.311,
      "step": 7062
    },
    {
      "epoch": 0.2851433185304804,
      "grad_norm": 125.71150970458984,
      "learning_rate": 1.6776308597423453e-05,
      "loss": 6.0698,
      "step": 7063
    },
    {
      "epoch": 0.28518368994751714,
      "grad_norm": 118.52833557128906,
      "learning_rate": 1.6775346941978857e-05,
      "loss": 5.6336,
      "step": 7064
    },
    {
      "epoch": 0.2852240613645539,
      "grad_norm": 155.73831176757812,
      "learning_rate": 1.6774385170691652e-05,
      "loss": 2.9055,
      "step": 7065
    },
    {
      "epoch": 0.28526443278159064,
      "grad_norm": 89.38378143310547,
      "learning_rate": 1.677342328357828e-05,
      "loss": 4.7794,
      "step": 7066
    },
    {
      "epoch": 0.28530480419862736,
      "grad_norm": 109.12374114990234,
      "learning_rate": 1.6772461280655184e-05,
      "loss": 4.7506,
      "step": 7067
    },
    {
      "epoch": 0.28534517561566414,
      "grad_norm": 95.76189422607422,
      "learning_rate": 1.677149916193882e-05,
      "loss": 5.8403,
      "step": 7068
    },
    {
      "epoch": 0.28538554703270086,
      "grad_norm": 81.8535385131836,
      "learning_rate": 1.6770536927445627e-05,
      "loss": 5.1392,
      "step": 7069
    },
    {
      "epoch": 0.2854259184497376,
      "grad_norm": 83.1111831665039,
      "learning_rate": 1.6769574577192064e-05,
      "loss": 3.7594,
      "step": 7070
    },
    {
      "epoch": 0.2854662898667743,
      "grad_norm": 122.64447021484375,
      "learning_rate": 1.6768612111194586e-05,
      "loss": 4.5254,
      "step": 7071
    },
    {
      "epoch": 0.2855066612838111,
      "grad_norm": 109.85399627685547,
      "learning_rate": 1.6767649529469646e-05,
      "loss": 4.2621,
      "step": 7072
    },
    {
      "epoch": 0.2855470327008478,
      "grad_norm": 213.5938262939453,
      "learning_rate": 1.6766686832033703e-05,
      "loss": 4.7947,
      "step": 7073
    },
    {
      "epoch": 0.2855874041178845,
      "grad_norm": 116.460205078125,
      "learning_rate": 1.6765724018903214e-05,
      "loss": 5.5872,
      "step": 7074
    },
    {
      "epoch": 0.2856277755349213,
      "grad_norm": 121.55392456054688,
      "learning_rate": 1.676476109009465e-05,
      "loss": 5.1976,
      "step": 7075
    },
    {
      "epoch": 0.285668146951958,
      "grad_norm": 74.95001220703125,
      "learning_rate": 1.6763798045624462e-05,
      "loss": 6.4459,
      "step": 7076
    },
    {
      "epoch": 0.28570851836899475,
      "grad_norm": 118.12166595458984,
      "learning_rate": 1.6762834885509125e-05,
      "loss": 5.19,
      "step": 7077
    },
    {
      "epoch": 0.28574888978603147,
      "grad_norm": 110.8545913696289,
      "learning_rate": 1.67618716097651e-05,
      "loss": 4.8389,
      "step": 7078
    },
    {
      "epoch": 0.28578926120306825,
      "grad_norm": 77.11988067626953,
      "learning_rate": 1.6760908218408864e-05,
      "loss": 6.2691,
      "step": 7079
    },
    {
      "epoch": 0.28582963262010497,
      "grad_norm": 98.62699890136719,
      "learning_rate": 1.6759944711456887e-05,
      "loss": 5.7848,
      "step": 7080
    },
    {
      "epoch": 0.2858700040371417,
      "grad_norm": 118.81278991699219,
      "learning_rate": 1.6758981088925637e-05,
      "loss": 4.8571,
      "step": 7081
    },
    {
      "epoch": 0.28591037545417847,
      "grad_norm": 93.69954681396484,
      "learning_rate": 1.6758017350831595e-05,
      "loss": 5.0816,
      "step": 7082
    },
    {
      "epoch": 0.2859507468712152,
      "grad_norm": 87.20641326904297,
      "learning_rate": 1.6757053497191237e-05,
      "loss": 6.3406,
      "step": 7083
    },
    {
      "epoch": 0.2859911182882519,
      "grad_norm": 94.06993103027344,
      "learning_rate": 1.6756089528021044e-05,
      "loss": 6.6741,
      "step": 7084
    },
    {
      "epoch": 0.28603148970528863,
      "grad_norm": 93.17546844482422,
      "learning_rate": 1.6755125443337495e-05,
      "loss": 6.0367,
      "step": 7085
    },
    {
      "epoch": 0.2860718611223254,
      "grad_norm": 186.0434112548828,
      "learning_rate": 1.675416124315708e-05,
      "loss": 6.2199,
      "step": 7086
    },
    {
      "epoch": 0.28611223253936213,
      "grad_norm": 129.6484375,
      "learning_rate": 1.6753196927496276e-05,
      "loss": 6.9313,
      "step": 7087
    },
    {
      "epoch": 0.28615260395639885,
      "grad_norm": 152.83926391601562,
      "learning_rate": 1.6752232496371576e-05,
      "loss": 6.9566,
      "step": 7088
    },
    {
      "epoch": 0.28619297537343563,
      "grad_norm": 112.21560668945312,
      "learning_rate": 1.6751267949799467e-05,
      "loss": 3.9071,
      "step": 7089
    },
    {
      "epoch": 0.28623334679047235,
      "grad_norm": 101.95863342285156,
      "learning_rate": 1.6750303287796438e-05,
      "loss": 5.7368,
      "step": 7090
    },
    {
      "epoch": 0.2862737182075091,
      "grad_norm": 102.4418716430664,
      "learning_rate": 1.674933851037899e-05,
      "loss": 4.1036,
      "step": 7091
    },
    {
      "epoch": 0.2863140896245458,
      "grad_norm": 288.62213134765625,
      "learning_rate": 1.6748373617563614e-05,
      "loss": 5.4881,
      "step": 7092
    },
    {
      "epoch": 0.2863544610415826,
      "grad_norm": 95.46280670166016,
      "learning_rate": 1.6747408609366805e-05,
      "loss": 6.7888,
      "step": 7093
    },
    {
      "epoch": 0.2863948324586193,
      "grad_norm": 127.354736328125,
      "learning_rate": 1.6746443485805068e-05,
      "loss": 5.8046,
      "step": 7094
    },
    {
      "epoch": 0.286435203875656,
      "grad_norm": 106.51057434082031,
      "learning_rate": 1.6745478246894902e-05,
      "loss": 4.9875,
      "step": 7095
    },
    {
      "epoch": 0.2864755752926928,
      "grad_norm": 99.79777526855469,
      "learning_rate": 1.6744512892652805e-05,
      "loss": 5.2395,
      "step": 7096
    },
    {
      "epoch": 0.2865159467097295,
      "grad_norm": 80.63549041748047,
      "learning_rate": 1.6743547423095293e-05,
      "loss": 4.5469,
      "step": 7097
    },
    {
      "epoch": 0.28655631812676624,
      "grad_norm": 110.95275115966797,
      "learning_rate": 1.6742581838238864e-05,
      "loss": 6.2286,
      "step": 7098
    },
    {
      "epoch": 0.28659668954380296,
      "grad_norm": 86.77100372314453,
      "learning_rate": 1.674161613810003e-05,
      "loss": 5.656,
      "step": 7099
    },
    {
      "epoch": 0.28663706096083974,
      "grad_norm": 102.98941040039062,
      "learning_rate": 1.6740650322695308e-05,
      "loss": 5.4252,
      "step": 7100
    },
    {
      "epoch": 0.28667743237787646,
      "grad_norm": 91.31305694580078,
      "learning_rate": 1.67396843920412e-05,
      "loss": 6.0522,
      "step": 7101
    },
    {
      "epoch": 0.2867178037949132,
      "grad_norm": 77.24491119384766,
      "learning_rate": 1.6738718346154233e-05,
      "loss": 5.1264,
      "step": 7102
    },
    {
      "epoch": 0.28675817521194996,
      "grad_norm": 265.7178955078125,
      "learning_rate": 1.6737752185050915e-05,
      "loss": 5.4504,
      "step": 7103
    },
    {
      "epoch": 0.2867985466289867,
      "grad_norm": 81.3823471069336,
      "learning_rate": 1.6736785908747765e-05,
      "loss": 5.5964,
      "step": 7104
    },
    {
      "epoch": 0.2868389180460234,
      "grad_norm": 145.55628967285156,
      "learning_rate": 1.6735819517261315e-05,
      "loss": 4.411,
      "step": 7105
    },
    {
      "epoch": 0.2868792894630601,
      "grad_norm": 94.68301391601562,
      "learning_rate": 1.6734853010608074e-05,
      "loss": 6.0846,
      "step": 7106
    },
    {
      "epoch": 0.2869196608800969,
      "grad_norm": 96.76715087890625,
      "learning_rate": 1.6733886388804576e-05,
      "loss": 5.2269,
      "step": 7107
    },
    {
      "epoch": 0.2869600322971336,
      "grad_norm": 95.2863998413086,
      "learning_rate": 1.6732919651867346e-05,
      "loss": 4.5979,
      "step": 7108
    },
    {
      "epoch": 0.28700040371417035,
      "grad_norm": 125.13866424560547,
      "learning_rate": 1.6731952799812914e-05,
      "loss": 7.2725,
      "step": 7109
    },
    {
      "epoch": 0.2870407751312071,
      "grad_norm": 181.70498657226562,
      "learning_rate": 1.6730985832657807e-05,
      "loss": 6.3016,
      "step": 7110
    },
    {
      "epoch": 0.28708114654824385,
      "grad_norm": 138.31231689453125,
      "learning_rate": 1.673001875041856e-05,
      "loss": 6.2114,
      "step": 7111
    },
    {
      "epoch": 0.28712151796528057,
      "grad_norm": 77.97256469726562,
      "learning_rate": 1.672905155311171e-05,
      "loss": 6.4589,
      "step": 7112
    },
    {
      "epoch": 0.2871618893823173,
      "grad_norm": 82.20867156982422,
      "learning_rate": 1.6728084240753787e-05,
      "loss": 4.8504,
      "step": 7113
    },
    {
      "epoch": 0.28720226079935407,
      "grad_norm": 74.82014465332031,
      "learning_rate": 1.672711681336134e-05,
      "loss": 5.5853,
      "step": 7114
    },
    {
      "epoch": 0.2872426322163908,
      "grad_norm": 82.72476959228516,
      "learning_rate": 1.67261492709509e-05,
      "loss": 4.8117,
      "step": 7115
    },
    {
      "epoch": 0.2872830036334275,
      "grad_norm": 99.5813980102539,
      "learning_rate": 1.672518161353902e-05,
      "loss": 4.1352,
      "step": 7116
    },
    {
      "epoch": 0.2873233750504643,
      "grad_norm": 141.2634735107422,
      "learning_rate": 1.6724213841142237e-05,
      "loss": 5.2066,
      "step": 7117
    },
    {
      "epoch": 0.287363746467501,
      "grad_norm": 189.24021911621094,
      "learning_rate": 1.6723245953777096e-05,
      "loss": 4.2475,
      "step": 7118
    },
    {
      "epoch": 0.28740411788453774,
      "grad_norm": 94.59339904785156,
      "learning_rate": 1.672227795146015e-05,
      "loss": 5.5553,
      "step": 7119
    },
    {
      "epoch": 0.2874444893015745,
      "grad_norm": 110.05186462402344,
      "learning_rate": 1.672130983420795e-05,
      "loss": 3.3349,
      "step": 7120
    },
    {
      "epoch": 0.28748486071861123,
      "grad_norm": 132.23167419433594,
      "learning_rate": 1.6720341602037048e-05,
      "loss": 5.2432,
      "step": 7121
    },
    {
      "epoch": 0.28752523213564796,
      "grad_norm": 77.85865783691406,
      "learning_rate": 1.6719373254963996e-05,
      "loss": 5.0934,
      "step": 7122
    },
    {
      "epoch": 0.2875656035526847,
      "grad_norm": 119.16925048828125,
      "learning_rate": 1.6718404793005354e-05,
      "loss": 6.0196,
      "step": 7123
    },
    {
      "epoch": 0.28760597496972146,
      "grad_norm": 74.21521759033203,
      "learning_rate": 1.6717436216177677e-05,
      "loss": 4.7956,
      "step": 7124
    },
    {
      "epoch": 0.2876463463867582,
      "grad_norm": 98.11873626708984,
      "learning_rate": 1.6716467524497527e-05,
      "loss": 4.6238,
      "step": 7125
    },
    {
      "epoch": 0.2876867178037949,
      "grad_norm": 102.71147918701172,
      "learning_rate": 1.6715498717981468e-05,
      "loss": 5.616,
      "step": 7126
    },
    {
      "epoch": 0.2877270892208317,
      "grad_norm": 224.88307189941406,
      "learning_rate": 1.6714529796646064e-05,
      "loss": 6.2138,
      "step": 7127
    },
    {
      "epoch": 0.2877674606378684,
      "grad_norm": 70.25899505615234,
      "learning_rate": 1.6713560760507877e-05,
      "loss": 4.4742,
      "step": 7128
    },
    {
      "epoch": 0.2878078320549051,
      "grad_norm": 164.81118774414062,
      "learning_rate": 1.6712591609583478e-05,
      "loss": 4.2394,
      "step": 7129
    },
    {
      "epoch": 0.28784820347194184,
      "grad_norm": 144.51414489746094,
      "learning_rate": 1.671162234388944e-05,
      "loss": 4.2492,
      "step": 7130
    },
    {
      "epoch": 0.2878885748889786,
      "grad_norm": 82.10653686523438,
      "learning_rate": 1.6710652963442333e-05,
      "loss": 6.817,
      "step": 7131
    },
    {
      "epoch": 0.28792894630601534,
      "grad_norm": 132.25352478027344,
      "learning_rate": 1.670968346825873e-05,
      "loss": 4.6004,
      "step": 7132
    },
    {
      "epoch": 0.28796931772305207,
      "grad_norm": 97.77762603759766,
      "learning_rate": 1.6708713858355207e-05,
      "loss": 5.3937,
      "step": 7133
    },
    {
      "epoch": 0.28800968914008884,
      "grad_norm": 119.95800018310547,
      "learning_rate": 1.6707744133748342e-05,
      "loss": 6.1932,
      "step": 7134
    },
    {
      "epoch": 0.28805006055712556,
      "grad_norm": 75.9188461303711,
      "learning_rate": 1.6706774294454717e-05,
      "loss": 5.269,
      "step": 7135
    },
    {
      "epoch": 0.2880904319741623,
      "grad_norm": 84.26110076904297,
      "learning_rate": 1.6705804340490913e-05,
      "loss": 5.9525,
      "step": 7136
    },
    {
      "epoch": 0.288130803391199,
      "grad_norm": 89.69024658203125,
      "learning_rate": 1.6704834271873513e-05,
      "loss": 5.6755,
      "step": 7137
    },
    {
      "epoch": 0.2881711748082358,
      "grad_norm": 118.58125305175781,
      "learning_rate": 1.6703864088619104e-05,
      "loss": 6.3248,
      "step": 7138
    },
    {
      "epoch": 0.2882115462252725,
      "grad_norm": 81.4039306640625,
      "learning_rate": 1.6702893790744277e-05,
      "loss": 5.3954,
      "step": 7139
    },
    {
      "epoch": 0.28825191764230923,
      "grad_norm": 157.19606018066406,
      "learning_rate": 1.6701923378265615e-05,
      "loss": 5.151,
      "step": 7140
    },
    {
      "epoch": 0.288292289059346,
      "grad_norm": 126.53094482421875,
      "learning_rate": 1.6700952851199716e-05,
      "loss": 6.4635,
      "step": 7141
    },
    {
      "epoch": 0.28833266047638273,
      "grad_norm": 152.29745483398438,
      "learning_rate": 1.6699982209563168e-05,
      "loss": 4.6671,
      "step": 7142
    },
    {
      "epoch": 0.28837303189341945,
      "grad_norm": 132.26283264160156,
      "learning_rate": 1.6699011453372573e-05,
      "loss": 5.6093,
      "step": 7143
    },
    {
      "epoch": 0.2884134033104562,
      "grad_norm": 131.31886291503906,
      "learning_rate": 1.6698040582644523e-05,
      "loss": 6.1464,
      "step": 7144
    },
    {
      "epoch": 0.28845377472749295,
      "grad_norm": 102.54837799072266,
      "learning_rate": 1.6697069597395622e-05,
      "loss": 4.5533,
      "step": 7145
    },
    {
      "epoch": 0.2884941461445297,
      "grad_norm": 1813.66552734375,
      "learning_rate": 1.6696098497642467e-05,
      "loss": 6.648,
      "step": 7146
    },
    {
      "epoch": 0.2885345175615664,
      "grad_norm": 86.14134216308594,
      "learning_rate": 1.6695127283401665e-05,
      "loss": 7.4958,
      "step": 7147
    },
    {
      "epoch": 0.2885748889786032,
      "grad_norm": 105.59870147705078,
      "learning_rate": 1.6694155954689822e-05,
      "loss": 5.095,
      "step": 7148
    },
    {
      "epoch": 0.2886152603956399,
      "grad_norm": 189.60357666015625,
      "learning_rate": 1.6693184511523547e-05,
      "loss": 5.1856,
      "step": 7149
    },
    {
      "epoch": 0.2886556318126766,
      "grad_norm": 92.60057067871094,
      "learning_rate": 1.669221295391944e-05,
      "loss": 5.548,
      "step": 7150
    },
    {
      "epoch": 0.28869600322971334,
      "grad_norm": 129.0547332763672,
      "learning_rate": 1.6691241281894123e-05,
      "loss": 5.7768,
      "step": 7151
    },
    {
      "epoch": 0.2887363746467501,
      "grad_norm": 179.57044982910156,
      "learning_rate": 1.6690269495464205e-05,
      "loss": 3.9428,
      "step": 7152
    },
    {
      "epoch": 0.28877674606378684,
      "grad_norm": 156.0667724609375,
      "learning_rate": 1.6689297594646302e-05,
      "loss": 6.1544,
      "step": 7153
    },
    {
      "epoch": 0.28881711748082356,
      "grad_norm": 82.81535339355469,
      "learning_rate": 1.668832557945703e-05,
      "loss": 5.6014,
      "step": 7154
    },
    {
      "epoch": 0.28885748889786034,
      "grad_norm": 135.36000061035156,
      "learning_rate": 1.668735344991301e-05,
      "loss": 4.5239,
      "step": 7155
    },
    {
      "epoch": 0.28889786031489706,
      "grad_norm": 118.89881896972656,
      "learning_rate": 1.668638120603086e-05,
      "loss": 4.2335,
      "step": 7156
    },
    {
      "epoch": 0.2889382317319338,
      "grad_norm": 68.75727081298828,
      "learning_rate": 1.6685408847827206e-05,
      "loss": 5.0877,
      "step": 7157
    },
    {
      "epoch": 0.2889786031489705,
      "grad_norm": 72.95580291748047,
      "learning_rate": 1.6684436375318673e-05,
      "loss": 4.3168,
      "step": 7158
    },
    {
      "epoch": 0.2890189745660073,
      "grad_norm": 111.45895385742188,
      "learning_rate": 1.6683463788521886e-05,
      "loss": 6.0619,
      "step": 7159
    },
    {
      "epoch": 0.289059345983044,
      "grad_norm": 81.54622650146484,
      "learning_rate": 1.6682491087453476e-05,
      "loss": 5.2897,
      "step": 7160
    },
    {
      "epoch": 0.2890997174000807,
      "grad_norm": 130.02122497558594,
      "learning_rate": 1.6681518272130072e-05,
      "loss": 6.1643,
      "step": 7161
    },
    {
      "epoch": 0.2891400888171175,
      "grad_norm": 113.75897216796875,
      "learning_rate": 1.668054534256831e-05,
      "loss": 6.5758,
      "step": 7162
    },
    {
      "epoch": 0.2891804602341542,
      "grad_norm": 134.35704040527344,
      "learning_rate": 1.667957229878482e-05,
      "loss": 6.0628,
      "step": 7163
    },
    {
      "epoch": 0.28922083165119095,
      "grad_norm": 134.98338317871094,
      "learning_rate": 1.6678599140796246e-05,
      "loss": 5.6623,
      "step": 7164
    },
    {
      "epoch": 0.28926120306822767,
      "grad_norm": 95.61381530761719,
      "learning_rate": 1.6677625868619222e-05,
      "loss": 4.2725,
      "step": 7165
    },
    {
      "epoch": 0.28930157448526445,
      "grad_norm": 108.91419982910156,
      "learning_rate": 1.6676652482270383e-05,
      "loss": 2.9014,
      "step": 7166
    },
    {
      "epoch": 0.28934194590230117,
      "grad_norm": 84.08922576904297,
      "learning_rate": 1.6675678981766382e-05,
      "loss": 5.7332,
      "step": 7167
    },
    {
      "epoch": 0.2893823173193379,
      "grad_norm": 575.5741577148438,
      "learning_rate": 1.667470536712386e-05,
      "loss": 6.7097,
      "step": 7168
    },
    {
      "epoch": 0.28942268873637467,
      "grad_norm": 87.91351318359375,
      "learning_rate": 1.6673731638359463e-05,
      "loss": 3.5509,
      "step": 7169
    },
    {
      "epoch": 0.2894630601534114,
      "grad_norm": 108.1483383178711,
      "learning_rate": 1.6672757795489837e-05,
      "loss": 5.8766,
      "step": 7170
    },
    {
      "epoch": 0.2895034315704481,
      "grad_norm": 133.95228576660156,
      "learning_rate": 1.6671783838531635e-05,
      "loss": 5.1356,
      "step": 7171
    },
    {
      "epoch": 0.2895438029874849,
      "grad_norm": 87.82634735107422,
      "learning_rate": 1.6670809767501508e-05,
      "loss": 4.3057,
      "step": 7172
    },
    {
      "epoch": 0.2895841744045216,
      "grad_norm": 115.47115325927734,
      "learning_rate": 1.6669835582416114e-05,
      "loss": 6.0653,
      "step": 7173
    },
    {
      "epoch": 0.28962454582155833,
      "grad_norm": 120.65579986572266,
      "learning_rate": 1.6668861283292103e-05,
      "loss": 5.1546,
      "step": 7174
    },
    {
      "epoch": 0.28966491723859505,
      "grad_norm": 105.97090148925781,
      "learning_rate": 1.666788687014614e-05,
      "loss": 5.8286,
      "step": 7175
    },
    {
      "epoch": 0.28970528865563183,
      "grad_norm": 72.076416015625,
      "learning_rate": 1.6666912342994886e-05,
      "loss": 4.183,
      "step": 7176
    },
    {
      "epoch": 0.28974566007266855,
      "grad_norm": 94.92854309082031,
      "learning_rate": 1.6665937701854994e-05,
      "loss": 5.7829,
      "step": 7177
    },
    {
      "epoch": 0.2897860314897053,
      "grad_norm": 107.3591079711914,
      "learning_rate": 1.666496294674313e-05,
      "loss": 3.4639,
      "step": 7178
    },
    {
      "epoch": 0.28982640290674205,
      "grad_norm": 978.8682250976562,
      "learning_rate": 1.6663988077675973e-05,
      "loss": 4.668,
      "step": 7179
    },
    {
      "epoch": 0.2898667743237788,
      "grad_norm": 94.81346130371094,
      "learning_rate": 1.6663013094670174e-05,
      "loss": 6.0262,
      "step": 7180
    },
    {
      "epoch": 0.2899071457408155,
      "grad_norm": 96.1770248413086,
      "learning_rate": 1.6662037997742415e-05,
      "loss": 5.4859,
      "step": 7181
    },
    {
      "epoch": 0.2899475171578522,
      "grad_norm": 155.1367950439453,
      "learning_rate": 1.6661062786909358e-05,
      "loss": 4.5494,
      "step": 7182
    },
    {
      "epoch": 0.289987888574889,
      "grad_norm": 93.91181182861328,
      "learning_rate": 1.666008746218769e-05,
      "loss": 6.0238,
      "step": 7183
    },
    {
      "epoch": 0.2900282599919257,
      "grad_norm": 81.36820983886719,
      "learning_rate": 1.665911202359407e-05,
      "loss": 3.2565,
      "step": 7184
    },
    {
      "epoch": 0.29006863140896244,
      "grad_norm": 88.00448608398438,
      "learning_rate": 1.665813647114519e-05,
      "loss": 5.1937,
      "step": 7185
    },
    {
      "epoch": 0.2901090028259992,
      "grad_norm": 200.5377655029297,
      "learning_rate": 1.665716080485772e-05,
      "loss": 4.9807,
      "step": 7186
    },
    {
      "epoch": 0.29014937424303594,
      "grad_norm": 197.03952026367188,
      "learning_rate": 1.665618502474835e-05,
      "loss": 6.2446,
      "step": 7187
    },
    {
      "epoch": 0.29018974566007266,
      "grad_norm": 136.70523071289062,
      "learning_rate": 1.6655209130833757e-05,
      "loss": 2.7713,
      "step": 7188
    },
    {
      "epoch": 0.2902301170771094,
      "grad_norm": 138.28970336914062,
      "learning_rate": 1.6654233123130633e-05,
      "loss": 6.1203,
      "step": 7189
    },
    {
      "epoch": 0.29027048849414616,
      "grad_norm": 96.54086303710938,
      "learning_rate": 1.6653257001655652e-05,
      "loss": 5.1423,
      "step": 7190
    },
    {
      "epoch": 0.2903108599111829,
      "grad_norm": 160.9849395751953,
      "learning_rate": 1.6652280766425523e-05,
      "loss": 5.3407,
      "step": 7191
    },
    {
      "epoch": 0.2903512313282196,
      "grad_norm": 97.5699234008789,
      "learning_rate": 1.665130441745692e-05,
      "loss": 5.8768,
      "step": 7192
    },
    {
      "epoch": 0.2903916027452564,
      "grad_norm": 160.72743225097656,
      "learning_rate": 1.665032795476655e-05,
      "loss": 5.2355,
      "step": 7193
    },
    {
      "epoch": 0.2904319741622931,
      "grad_norm": 126.31873321533203,
      "learning_rate": 1.6649351378371098e-05,
      "loss": 5.3987,
      "step": 7194
    },
    {
      "epoch": 0.2904723455793298,
      "grad_norm": 131.03578186035156,
      "learning_rate": 1.664837468828726e-05,
      "loss": 6.0399,
      "step": 7195
    },
    {
      "epoch": 0.29051271699636655,
      "grad_norm": 222.56602478027344,
      "learning_rate": 1.6647397884531747e-05,
      "loss": 5.1137,
      "step": 7196
    },
    {
      "epoch": 0.2905530884134033,
      "grad_norm": 113.6751708984375,
      "learning_rate": 1.6646420967121252e-05,
      "loss": 7.1748,
      "step": 7197
    },
    {
      "epoch": 0.29059345983044005,
      "grad_norm": 105.5047378540039,
      "learning_rate": 1.6645443936072477e-05,
      "loss": 5.9451,
      "step": 7198
    },
    {
      "epoch": 0.29063383124747677,
      "grad_norm": 81.40450286865234,
      "learning_rate": 1.664446679140213e-05,
      "loss": 6.7144,
      "step": 7199
    },
    {
      "epoch": 0.29067420266451355,
      "grad_norm": 70.62767028808594,
      "learning_rate": 1.6643489533126915e-05,
      "loss": 6.1664,
      "step": 7200
    },
    {
      "epoch": 0.29071457408155027,
      "grad_norm": 138.175048828125,
      "learning_rate": 1.6642512161263545e-05,
      "loss": 4.6398,
      "step": 7201
    },
    {
      "epoch": 0.290754945498587,
      "grad_norm": 121.69737243652344,
      "learning_rate": 1.6641534675828723e-05,
      "loss": 5.6761,
      "step": 7202
    },
    {
      "epoch": 0.2907953169156237,
      "grad_norm": 121.85870361328125,
      "learning_rate": 1.6640557076839173e-05,
      "loss": 5.899,
      "step": 7203
    },
    {
      "epoch": 0.2908356883326605,
      "grad_norm": 164.81613159179688,
      "learning_rate": 1.6639579364311603e-05,
      "loss": 5.2244,
      "step": 7204
    },
    {
      "epoch": 0.2908760597496972,
      "grad_norm": 155.92501831054688,
      "learning_rate": 1.663860153826273e-05,
      "loss": 6.0751,
      "step": 7205
    },
    {
      "epoch": 0.29091643116673394,
      "grad_norm": 82.50077056884766,
      "learning_rate": 1.663762359870927e-05,
      "loss": 4.9918,
      "step": 7206
    },
    {
      "epoch": 0.2909568025837707,
      "grad_norm": 95.03050231933594,
      "learning_rate": 1.663664554566795e-05,
      "loss": 5.8054,
      "step": 7207
    },
    {
      "epoch": 0.29099717400080743,
      "grad_norm": 111.30221557617188,
      "learning_rate": 1.663566737915549e-05,
      "loss": 4.7231,
      "step": 7208
    },
    {
      "epoch": 0.29103754541784416,
      "grad_norm": 71.26830291748047,
      "learning_rate": 1.663468909918861e-05,
      "loss": 5.0989,
      "step": 7209
    },
    {
      "epoch": 0.2910779168348809,
      "grad_norm": 92.6832046508789,
      "learning_rate": 1.6633710705784037e-05,
      "loss": 5.7607,
      "step": 7210
    },
    {
      "epoch": 0.29111828825191766,
      "grad_norm": 285.4213562011719,
      "learning_rate": 1.6632732198958507e-05,
      "loss": 4.0979,
      "step": 7211
    },
    {
      "epoch": 0.2911586596689544,
      "grad_norm": 88.06118774414062,
      "learning_rate": 1.6631753578728743e-05,
      "loss": 5.9496,
      "step": 7212
    },
    {
      "epoch": 0.2911990310859911,
      "grad_norm": 117.70036315917969,
      "learning_rate": 1.6630774845111482e-05,
      "loss": 5.0998,
      "step": 7213
    },
    {
      "epoch": 0.2912394025030279,
      "grad_norm": 176.11431884765625,
      "learning_rate": 1.6629795998123453e-05,
      "loss": 5.3868,
      "step": 7214
    },
    {
      "epoch": 0.2912797739200646,
      "grad_norm": 184.06930541992188,
      "learning_rate": 1.6628817037781392e-05,
      "loss": 5.6436,
      "step": 7215
    },
    {
      "epoch": 0.2913201453371013,
      "grad_norm": 99.10858917236328,
      "learning_rate": 1.662783796410204e-05,
      "loss": 6.3236,
      "step": 7216
    },
    {
      "epoch": 0.29136051675413804,
      "grad_norm": 93.4450454711914,
      "learning_rate": 1.6626858777102138e-05,
      "loss": 5.3543,
      "step": 7217
    },
    {
      "epoch": 0.2914008881711748,
      "grad_norm": 143.3512420654297,
      "learning_rate": 1.6625879476798423e-05,
      "loss": 4.8561,
      "step": 7218
    },
    {
      "epoch": 0.29144125958821154,
      "grad_norm": 96.51358032226562,
      "learning_rate": 1.6624900063207646e-05,
      "loss": 4.7813,
      "step": 7219
    },
    {
      "epoch": 0.29148163100524827,
      "grad_norm": 83.67677307128906,
      "learning_rate": 1.6623920536346544e-05,
      "loss": 6.3298,
      "step": 7220
    },
    {
      "epoch": 0.29152200242228504,
      "grad_norm": 103.04386138916016,
      "learning_rate": 1.662294089623187e-05,
      "loss": 4.5662,
      "step": 7221
    },
    {
      "epoch": 0.29156237383932176,
      "grad_norm": 130.1688690185547,
      "learning_rate": 1.6621961142880368e-05,
      "loss": 5.4354,
      "step": 7222
    },
    {
      "epoch": 0.2916027452563585,
      "grad_norm": 117.05049133300781,
      "learning_rate": 1.6620981276308802e-05,
      "loss": 5.9107,
      "step": 7223
    },
    {
      "epoch": 0.2916431166733952,
      "grad_norm": 102.06326293945312,
      "learning_rate": 1.6620001296533912e-05,
      "loss": 4.1141,
      "step": 7224
    },
    {
      "epoch": 0.291683488090432,
      "grad_norm": 107.24504852294922,
      "learning_rate": 1.6619021203572458e-05,
      "loss": 4.386,
      "step": 7225
    },
    {
      "epoch": 0.2917238595074687,
      "grad_norm": 100.85893249511719,
      "learning_rate": 1.66180409974412e-05,
      "loss": 7.1329,
      "step": 7226
    },
    {
      "epoch": 0.29176423092450543,
      "grad_norm": 83.46745300292969,
      "learning_rate": 1.6617060678156896e-05,
      "loss": 5.0668,
      "step": 7227
    },
    {
      "epoch": 0.2918046023415422,
      "grad_norm": 92.02146911621094,
      "learning_rate": 1.6616080245736307e-05,
      "loss": 4.2135,
      "step": 7228
    },
    {
      "epoch": 0.29184497375857893,
      "grad_norm": 121.90748596191406,
      "learning_rate": 1.661509970019619e-05,
      "loss": 5.0432,
      "step": 7229
    },
    {
      "epoch": 0.29188534517561565,
      "grad_norm": 227.16233825683594,
      "learning_rate": 1.661411904155332e-05,
      "loss": 4.0718,
      "step": 7230
    },
    {
      "epoch": 0.29192571659265243,
      "grad_norm": 91.92820739746094,
      "learning_rate": 1.661313826982446e-05,
      "loss": 4.054,
      "step": 7231
    },
    {
      "epoch": 0.29196608800968915,
      "grad_norm": 101.4403305053711,
      "learning_rate": 1.6612157385026377e-05,
      "loss": 4.3102,
      "step": 7232
    },
    {
      "epoch": 0.2920064594267259,
      "grad_norm": 94.02326965332031,
      "learning_rate": 1.6611176387175842e-05,
      "loss": 5.9412,
      "step": 7233
    },
    {
      "epoch": 0.2920468308437626,
      "grad_norm": 144.14256286621094,
      "learning_rate": 1.661019527628963e-05,
      "loss": 5.624,
      "step": 7234
    },
    {
      "epoch": 0.2920872022607994,
      "grad_norm": 90.20195007324219,
      "learning_rate": 1.6609214052384514e-05,
      "loss": 5.6252,
      "step": 7235
    },
    {
      "epoch": 0.2921275736778361,
      "grad_norm": 161.83615112304688,
      "learning_rate": 1.6608232715477274e-05,
      "loss": 5.7219,
      "step": 7236
    },
    {
      "epoch": 0.2921679450948728,
      "grad_norm": 93.9181137084961,
      "learning_rate": 1.6607251265584686e-05,
      "loss": 5.9076,
      "step": 7237
    },
    {
      "epoch": 0.2922083165119096,
      "grad_norm": 118.0978775024414,
      "learning_rate": 1.6606269702723526e-05,
      "loss": 5.5942,
      "step": 7238
    },
    {
      "epoch": 0.2922486879289463,
      "grad_norm": 108.52139282226562,
      "learning_rate": 1.6605288026910585e-05,
      "loss": 4.1933,
      "step": 7239
    },
    {
      "epoch": 0.29228905934598304,
      "grad_norm": 93.4549331665039,
      "learning_rate": 1.6604306238162642e-05,
      "loss": 5.4924,
      "step": 7240
    },
    {
      "epoch": 0.29232943076301976,
      "grad_norm": 151.3306427001953,
      "learning_rate": 1.6603324336496483e-05,
      "loss": 6.3375,
      "step": 7241
    },
    {
      "epoch": 0.29236980218005654,
      "grad_norm": 111.5956802368164,
      "learning_rate": 1.66023423219289e-05,
      "loss": 5.2388,
      "step": 7242
    },
    {
      "epoch": 0.29241017359709326,
      "grad_norm": 131.69422912597656,
      "learning_rate": 1.660136019447668e-05,
      "loss": 5.489,
      "step": 7243
    },
    {
      "epoch": 0.29245054501413,
      "grad_norm": 93.70466613769531,
      "learning_rate": 1.6600377954156616e-05,
      "loss": 4.8569,
      "step": 7244
    },
    {
      "epoch": 0.29249091643116676,
      "grad_norm": 103.133056640625,
      "learning_rate": 1.65993956009855e-05,
      "loss": 6.5178,
      "step": 7245
    },
    {
      "epoch": 0.2925312878482035,
      "grad_norm": 75.35203552246094,
      "learning_rate": 1.659841313498013e-05,
      "loss": 6.2239,
      "step": 7246
    },
    {
      "epoch": 0.2925716592652402,
      "grad_norm": 105.10456085205078,
      "learning_rate": 1.6597430556157307e-05,
      "loss": 6.1383,
      "step": 7247
    },
    {
      "epoch": 0.2926120306822769,
      "grad_norm": 69.5913314819336,
      "learning_rate": 1.6596447864533828e-05,
      "loss": 4.7819,
      "step": 7248
    },
    {
      "epoch": 0.2926524020993137,
      "grad_norm": 79.07725524902344,
      "learning_rate": 1.659546506012649e-05,
      "loss": 5.2572,
      "step": 7249
    },
    {
      "epoch": 0.2926927735163504,
      "grad_norm": 116.84628295898438,
      "learning_rate": 1.6594482142952103e-05,
      "loss": 5.4398,
      "step": 7250
    },
    {
      "epoch": 0.29273314493338715,
      "grad_norm": 91.91943359375,
      "learning_rate": 1.6593499113027474e-05,
      "loss": 4.5729,
      "step": 7251
    },
    {
      "epoch": 0.2927735163504239,
      "grad_norm": 85.0927505493164,
      "learning_rate": 1.6592515970369403e-05,
      "loss": 4.7827,
      "step": 7252
    },
    {
      "epoch": 0.29281388776746065,
      "grad_norm": 86.68842315673828,
      "learning_rate": 1.6591532714994703e-05,
      "loss": 6.3339,
      "step": 7253
    },
    {
      "epoch": 0.29285425918449737,
      "grad_norm": 153.12109375,
      "learning_rate": 1.659054934692019e-05,
      "loss": 5.462,
      "step": 7254
    },
    {
      "epoch": 0.2928946306015341,
      "grad_norm": 67.44571685791016,
      "learning_rate": 1.658956586616267e-05,
      "loss": 5.8781,
      "step": 7255
    },
    {
      "epoch": 0.29293500201857087,
      "grad_norm": 91.10954284667969,
      "learning_rate": 1.658858227273896e-05,
      "loss": 5.3614,
      "step": 7256
    },
    {
      "epoch": 0.2929753734356076,
      "grad_norm": 81.49465942382812,
      "learning_rate": 1.658759856666588e-05,
      "loss": 6.8399,
      "step": 7257
    },
    {
      "epoch": 0.2930157448526443,
      "grad_norm": 113.38937377929688,
      "learning_rate": 1.6586614747960246e-05,
      "loss": 5.1442,
      "step": 7258
    },
    {
      "epoch": 0.2930561162696811,
      "grad_norm": 91.56624603271484,
      "learning_rate": 1.6585630816638886e-05,
      "loss": 6.8917,
      "step": 7259
    },
    {
      "epoch": 0.2930964876867178,
      "grad_norm": 80.41079711914062,
      "learning_rate": 1.6584646772718613e-05,
      "loss": 4.9971,
      "step": 7260
    },
    {
      "epoch": 0.29313685910375453,
      "grad_norm": 98.12834167480469,
      "learning_rate": 1.658366261621626e-05,
      "loss": 7.1272,
      "step": 7261
    },
    {
      "epoch": 0.29317723052079125,
      "grad_norm": 127.20003509521484,
      "learning_rate": 1.658267834714865e-05,
      "loss": 3.9559,
      "step": 7262
    },
    {
      "epoch": 0.29321760193782803,
      "grad_norm": 117.91317749023438,
      "learning_rate": 1.658169396553261e-05,
      "loss": 4.3762,
      "step": 7263
    },
    {
      "epoch": 0.29325797335486475,
      "grad_norm": 82.35855865478516,
      "learning_rate": 1.6580709471384972e-05,
      "loss": 6.1176,
      "step": 7264
    },
    {
      "epoch": 0.2932983447719015,
      "grad_norm": 90.36881256103516,
      "learning_rate": 1.657972486472257e-05,
      "loss": 5.5282,
      "step": 7265
    },
    {
      "epoch": 0.29333871618893825,
      "grad_norm": 70.00463104248047,
      "learning_rate": 1.6578740145562236e-05,
      "loss": 5.5998,
      "step": 7266
    },
    {
      "epoch": 0.293379087605975,
      "grad_norm": 153.76939392089844,
      "learning_rate": 1.657775531392081e-05,
      "loss": 5.4476,
      "step": 7267
    },
    {
      "epoch": 0.2934194590230117,
      "grad_norm": 69.47077178955078,
      "learning_rate": 1.657677036981513e-05,
      "loss": 5.2028,
      "step": 7268
    },
    {
      "epoch": 0.2934598304400484,
      "grad_norm": 90.1418228149414,
      "learning_rate": 1.657578531326203e-05,
      "loss": 5.6114,
      "step": 7269
    },
    {
      "epoch": 0.2935002018570852,
      "grad_norm": 67.09906005859375,
      "learning_rate": 1.6574800144278357e-05,
      "loss": 5.3047,
      "step": 7270
    },
    {
      "epoch": 0.2935405732741219,
      "grad_norm": 91.5662612915039,
      "learning_rate": 1.6573814862880956e-05,
      "loss": 4.7874,
      "step": 7271
    },
    {
      "epoch": 0.29358094469115864,
      "grad_norm": 104.05779266357422,
      "learning_rate": 1.6572829469086674e-05,
      "loss": 5.5801,
      "step": 7272
    },
    {
      "epoch": 0.2936213161081954,
      "grad_norm": 211.07386779785156,
      "learning_rate": 1.6571843962912354e-05,
      "loss": 5.363,
      "step": 7273
    },
    {
      "epoch": 0.29366168752523214,
      "grad_norm": 162.2935028076172,
      "learning_rate": 1.657085834437485e-05,
      "loss": 5.8977,
      "step": 7274
    },
    {
      "epoch": 0.29370205894226886,
      "grad_norm": 150.72242736816406,
      "learning_rate": 1.6569872613491014e-05,
      "loss": 5.052,
      "step": 7275
    },
    {
      "epoch": 0.2937424303593056,
      "grad_norm": 93.80782318115234,
      "learning_rate": 1.6568886770277693e-05,
      "loss": 4.2754,
      "step": 7276
    },
    {
      "epoch": 0.29378280177634236,
      "grad_norm": 88.44038391113281,
      "learning_rate": 1.6567900814751754e-05,
      "loss": 5.5769,
      "step": 7277
    },
    {
      "epoch": 0.2938231731933791,
      "grad_norm": 592.2020874023438,
      "learning_rate": 1.6566914746930045e-05,
      "loss": 4.7596,
      "step": 7278
    },
    {
      "epoch": 0.2938635446104158,
      "grad_norm": 81.12718963623047,
      "learning_rate": 1.6565928566829427e-05,
      "loss": 6.068,
      "step": 7279
    },
    {
      "epoch": 0.2939039160274526,
      "grad_norm": 83.90303802490234,
      "learning_rate": 1.6564942274466765e-05,
      "loss": 6.0356,
      "step": 7280
    },
    {
      "epoch": 0.2939442874444893,
      "grad_norm": 78.378173828125,
      "learning_rate": 1.6563955869858923e-05,
      "loss": 5.4825,
      "step": 7281
    },
    {
      "epoch": 0.293984658861526,
      "grad_norm": 151.5792999267578,
      "learning_rate": 1.656296935302276e-05,
      "loss": 5.0759,
      "step": 7282
    },
    {
      "epoch": 0.2940250302785628,
      "grad_norm": 109.56546020507812,
      "learning_rate": 1.6561982723975146e-05,
      "loss": 5.4061,
      "step": 7283
    },
    {
      "epoch": 0.2940654016955995,
      "grad_norm": 85.60417938232422,
      "learning_rate": 1.6560995982732954e-05,
      "loss": 3.6787,
      "step": 7284
    },
    {
      "epoch": 0.29410577311263625,
      "grad_norm": 111.82747650146484,
      "learning_rate": 1.656000912931305e-05,
      "loss": 3.6734,
      "step": 7285
    },
    {
      "epoch": 0.29414614452967297,
      "grad_norm": 131.49969482421875,
      "learning_rate": 1.6559022163732308e-05,
      "loss": 4.3951,
      "step": 7286
    },
    {
      "epoch": 0.29418651594670975,
      "grad_norm": 87.3468246459961,
      "learning_rate": 1.6558035086007606e-05,
      "loss": 5.5734,
      "step": 7287
    },
    {
      "epoch": 0.29422688736374647,
      "grad_norm": 81.851318359375,
      "learning_rate": 1.6557047896155813e-05,
      "loss": 6.7634,
      "step": 7288
    },
    {
      "epoch": 0.2942672587807832,
      "grad_norm": 101.86026763916016,
      "learning_rate": 1.6556060594193815e-05,
      "loss": 6.1056,
      "step": 7289
    },
    {
      "epoch": 0.29430763019781997,
      "grad_norm": 165.9380340576172,
      "learning_rate": 1.6555073180138493e-05,
      "loss": 5.3518,
      "step": 7290
    },
    {
      "epoch": 0.2943480016148567,
      "grad_norm": 102.42095947265625,
      "learning_rate": 1.6554085654006726e-05,
      "loss": 3.9043,
      "step": 7291
    },
    {
      "epoch": 0.2943883730318934,
      "grad_norm": 196.27304077148438,
      "learning_rate": 1.65530980158154e-05,
      "loss": 5.1009,
      "step": 7292
    },
    {
      "epoch": 0.29442874444893014,
      "grad_norm": 98.30440521240234,
      "learning_rate": 1.6552110265581395e-05,
      "loss": 4.4055,
      "step": 7293
    },
    {
      "epoch": 0.2944691158659669,
      "grad_norm": 134.01480102539062,
      "learning_rate": 1.6551122403321607e-05,
      "loss": 5.1009,
      "step": 7294
    },
    {
      "epoch": 0.29450948728300363,
      "grad_norm": 78.07803344726562,
      "learning_rate": 1.6550134429052927e-05,
      "loss": 4.2122,
      "step": 7295
    },
    {
      "epoch": 0.29454985870004036,
      "grad_norm": 101.67192077636719,
      "learning_rate": 1.654914634279224e-05,
      "loss": 5.8843,
      "step": 7296
    },
    {
      "epoch": 0.29459023011707713,
      "grad_norm": 111.17160034179688,
      "learning_rate": 1.654815814455645e-05,
      "loss": 6.5927,
      "step": 7297
    },
    {
      "epoch": 0.29463060153411386,
      "grad_norm": 108.37767791748047,
      "learning_rate": 1.654716983436244e-05,
      "loss": 5.4952,
      "step": 7298
    },
    {
      "epoch": 0.2946709729511506,
      "grad_norm": 86.06785583496094,
      "learning_rate": 1.654618141222712e-05,
      "loss": 5.7642,
      "step": 7299
    },
    {
      "epoch": 0.2947113443681873,
      "grad_norm": 305.4000244140625,
      "learning_rate": 1.654519287816738e-05,
      "loss": 5.1068,
      "step": 7300
    },
    {
      "epoch": 0.2947517157852241,
      "grad_norm": 77.90813446044922,
      "learning_rate": 1.654420423220013e-05,
      "loss": 5.8914,
      "step": 7301
    },
    {
      "epoch": 0.2947920872022608,
      "grad_norm": 94.16065979003906,
      "learning_rate": 1.6543215474342268e-05,
      "loss": 7.0715,
      "step": 7302
    },
    {
      "epoch": 0.2948324586192975,
      "grad_norm": 109.7548599243164,
      "learning_rate": 1.6542226604610698e-05,
      "loss": 4.1194,
      "step": 7303
    },
    {
      "epoch": 0.2948728300363343,
      "grad_norm": 98.64430236816406,
      "learning_rate": 1.6541237623022333e-05,
      "loss": 4.9803,
      "step": 7304
    },
    {
      "epoch": 0.294913201453371,
      "grad_norm": 81.32831573486328,
      "learning_rate": 1.6540248529594078e-05,
      "loss": 4.3738,
      "step": 7305
    },
    {
      "epoch": 0.29495357287040774,
      "grad_norm": 117.90347290039062,
      "learning_rate": 1.6539259324342848e-05,
      "loss": 4.7399,
      "step": 7306
    },
    {
      "epoch": 0.29499394428744447,
      "grad_norm": 106.57283020019531,
      "learning_rate": 1.6538270007285552e-05,
      "loss": 4.8677,
      "step": 7307
    },
    {
      "epoch": 0.29503431570448124,
      "grad_norm": 101.47188568115234,
      "learning_rate": 1.653728057843911e-05,
      "loss": 4.4686,
      "step": 7308
    },
    {
      "epoch": 0.29507468712151796,
      "grad_norm": 155.9355010986328,
      "learning_rate": 1.6536291037820433e-05,
      "loss": 4.542,
      "step": 7309
    },
    {
      "epoch": 0.2951150585385547,
      "grad_norm": 160.8304443359375,
      "learning_rate": 1.653530138544644e-05,
      "loss": 5.4143,
      "step": 7310
    },
    {
      "epoch": 0.29515542995559146,
      "grad_norm": 99.35700225830078,
      "learning_rate": 1.6534311621334058e-05,
      "loss": 6.4168,
      "step": 7311
    },
    {
      "epoch": 0.2951958013726282,
      "grad_norm": 78.380615234375,
      "learning_rate": 1.6533321745500203e-05,
      "loss": 3.9986,
      "step": 7312
    },
    {
      "epoch": 0.2952361727896649,
      "grad_norm": 122.06571960449219,
      "learning_rate": 1.6532331757961804e-05,
      "loss": 4.9621,
      "step": 7313
    },
    {
      "epoch": 0.29527654420670163,
      "grad_norm": 71.7472152709961,
      "learning_rate": 1.6531341658735786e-05,
      "loss": 4.5701,
      "step": 7314
    },
    {
      "epoch": 0.2953169156237384,
      "grad_norm": 74.67803192138672,
      "learning_rate": 1.6530351447839082e-05,
      "loss": 5.6487,
      "step": 7315
    },
    {
      "epoch": 0.29535728704077513,
      "grad_norm": 94.02667236328125,
      "learning_rate": 1.652936112528861e-05,
      "loss": 4.7585,
      "step": 7316
    },
    {
      "epoch": 0.29539765845781185,
      "grad_norm": 77.68048858642578,
      "learning_rate": 1.6528370691101318e-05,
      "loss": 5.9869,
      "step": 7317
    },
    {
      "epoch": 0.29543802987484863,
      "grad_norm": 114.47071075439453,
      "learning_rate": 1.6527380145294125e-05,
      "loss": 5.8474,
      "step": 7318
    },
    {
      "epoch": 0.29547840129188535,
      "grad_norm": 70.53148651123047,
      "learning_rate": 1.652638948788398e-05,
      "loss": 5.0997,
      "step": 7319
    },
    {
      "epoch": 0.2955187727089221,
      "grad_norm": 92.12418365478516,
      "learning_rate": 1.652539871888781e-05,
      "loss": 5.2922,
      "step": 7320
    },
    {
      "epoch": 0.2955591441259588,
      "grad_norm": 78.77169799804688,
      "learning_rate": 1.652440783832256e-05,
      "loss": 4.9155,
      "step": 7321
    },
    {
      "epoch": 0.2955995155429956,
      "grad_norm": 99.21060943603516,
      "learning_rate": 1.652341684620517e-05,
      "loss": 4.6492,
      "step": 7322
    },
    {
      "epoch": 0.2956398869600323,
      "grad_norm": 217.71832275390625,
      "learning_rate": 1.652242574255259e-05,
      "loss": 6.7226,
      "step": 7323
    },
    {
      "epoch": 0.295680258377069,
      "grad_norm": 208.06686401367188,
      "learning_rate": 1.6521434527381758e-05,
      "loss": 3.793,
      "step": 7324
    },
    {
      "epoch": 0.2957206297941058,
      "grad_norm": 131.67054748535156,
      "learning_rate": 1.6520443200709627e-05,
      "loss": 4.7958,
      "step": 7325
    },
    {
      "epoch": 0.2957610012111425,
      "grad_norm": 93.4510726928711,
      "learning_rate": 1.651945176255314e-05,
      "loss": 4.9705,
      "step": 7326
    },
    {
      "epoch": 0.29580137262817924,
      "grad_norm": 100.43107604980469,
      "learning_rate": 1.6518460212929258e-05,
      "loss": 4.124,
      "step": 7327
    },
    {
      "epoch": 0.29584174404521596,
      "grad_norm": 198.14358520507812,
      "learning_rate": 1.651746855185492e-05,
      "loss": 5.3878,
      "step": 7328
    },
    {
      "epoch": 0.29588211546225274,
      "grad_norm": 94.92047119140625,
      "learning_rate": 1.6516476779347095e-05,
      "loss": 6.144,
      "step": 7329
    },
    {
      "epoch": 0.29592248687928946,
      "grad_norm": 60.070621490478516,
      "learning_rate": 1.651548489542273e-05,
      "loss": 6.8641,
      "step": 7330
    },
    {
      "epoch": 0.2959628582963262,
      "grad_norm": 82.99150085449219,
      "learning_rate": 1.651449290009879e-05,
      "loss": 6.0173,
      "step": 7331
    },
    {
      "epoch": 0.29600322971336296,
      "grad_norm": 83.23502349853516,
      "learning_rate": 1.6513500793392235e-05,
      "loss": 4.6227,
      "step": 7332
    },
    {
      "epoch": 0.2960436011303997,
      "grad_norm": 300.1085510253906,
      "learning_rate": 1.6512508575320024e-05,
      "loss": 4.6852,
      "step": 7333
    },
    {
      "epoch": 0.2960839725474364,
      "grad_norm": 398.2463073730469,
      "learning_rate": 1.6511516245899126e-05,
      "loss": 3.95,
      "step": 7334
    },
    {
      "epoch": 0.2961243439644732,
      "grad_norm": 90.3869400024414,
      "learning_rate": 1.6510523805146504e-05,
      "loss": 6.3492,
      "step": 7335
    },
    {
      "epoch": 0.2961647153815099,
      "grad_norm": 86.74116516113281,
      "learning_rate": 1.650953125307913e-05,
      "loss": 6.4862,
      "step": 7336
    },
    {
      "epoch": 0.2962050867985466,
      "grad_norm": 189.46563720703125,
      "learning_rate": 1.6508538589713974e-05,
      "loss": 4.6103,
      "step": 7337
    },
    {
      "epoch": 0.29624545821558335,
      "grad_norm": 150.71412658691406,
      "learning_rate": 1.6507545815068e-05,
      "loss": 3.6418,
      "step": 7338
    },
    {
      "epoch": 0.2962858296326201,
      "grad_norm": 92.83818054199219,
      "learning_rate": 1.6506552929158196e-05,
      "loss": 5.0265,
      "step": 7339
    },
    {
      "epoch": 0.29632620104965685,
      "grad_norm": 114.89167785644531,
      "learning_rate": 1.6505559932001527e-05,
      "loss": 5.1151,
      "step": 7340
    },
    {
      "epoch": 0.29636657246669357,
      "grad_norm": 146.90245056152344,
      "learning_rate": 1.6504566823614978e-05,
      "loss": 6.1313,
      "step": 7341
    },
    {
      "epoch": 0.29640694388373034,
      "grad_norm": 79.6856460571289,
      "learning_rate": 1.6503573604015524e-05,
      "loss": 5.0279,
      "step": 7342
    },
    {
      "epoch": 0.29644731530076707,
      "grad_norm": 127.18577575683594,
      "learning_rate": 1.6502580273220146e-05,
      "loss": 5.5444,
      "step": 7343
    },
    {
      "epoch": 0.2964876867178038,
      "grad_norm": 98.1731948852539,
      "learning_rate": 1.6501586831245832e-05,
      "loss": 3.7905,
      "step": 7344
    },
    {
      "epoch": 0.2965280581348405,
      "grad_norm": 98.19215393066406,
      "learning_rate": 1.6500593278109563e-05,
      "loss": 6.1175,
      "step": 7345
    },
    {
      "epoch": 0.2965684295518773,
      "grad_norm": 244.67935180664062,
      "learning_rate": 1.6499599613828332e-05,
      "loss": 6.4503,
      "step": 7346
    },
    {
      "epoch": 0.296608800968914,
      "grad_norm": 88.68206787109375,
      "learning_rate": 1.6498605838419126e-05,
      "loss": 3.6692,
      "step": 7347
    },
    {
      "epoch": 0.29664917238595073,
      "grad_norm": 83.55973052978516,
      "learning_rate": 1.6497611951898932e-05,
      "loss": 5.0075,
      "step": 7348
    },
    {
      "epoch": 0.2966895438029875,
      "grad_norm": 91.6436996459961,
      "learning_rate": 1.6496617954284747e-05,
      "loss": 5.6362,
      "step": 7349
    },
    {
      "epoch": 0.29672991522002423,
      "grad_norm": 148.96087646484375,
      "learning_rate": 1.649562384559357e-05,
      "loss": 4.7394,
      "step": 7350
    },
    {
      "epoch": 0.29677028663706095,
      "grad_norm": 124.35485076904297,
      "learning_rate": 1.6494629625842388e-05,
      "loss": 2.5885,
      "step": 7351
    },
    {
      "epoch": 0.2968106580540977,
      "grad_norm": 147.07505798339844,
      "learning_rate": 1.649363529504821e-05,
      "loss": 5.9852,
      "step": 7352
    },
    {
      "epoch": 0.29685102947113445,
      "grad_norm": 133.9464111328125,
      "learning_rate": 1.649264085322803e-05,
      "loss": 5.466,
      "step": 7353
    },
    {
      "epoch": 0.2968914008881712,
      "grad_norm": 101.71126556396484,
      "learning_rate": 1.649164630039885e-05,
      "loss": 6.0374,
      "step": 7354
    },
    {
      "epoch": 0.2969317723052079,
      "grad_norm": 87.26793670654297,
      "learning_rate": 1.6490651636577685e-05,
      "loss": 4.8276,
      "step": 7355
    },
    {
      "epoch": 0.2969721437222447,
      "grad_norm": 87.34381103515625,
      "learning_rate": 1.6489656861781528e-05,
      "loss": 4.2141,
      "step": 7356
    },
    {
      "epoch": 0.2970125151392814,
      "grad_norm": 100.71920776367188,
      "learning_rate": 1.6488661976027395e-05,
      "loss": 5.425,
      "step": 7357
    },
    {
      "epoch": 0.2970528865563181,
      "grad_norm": 103.23836517333984,
      "learning_rate": 1.6487666979332292e-05,
      "loss": 6.1334,
      "step": 7358
    },
    {
      "epoch": 0.29709325797335484,
      "grad_norm": 79.87155151367188,
      "learning_rate": 1.6486671871713234e-05,
      "loss": 6.4365,
      "step": 7359
    },
    {
      "epoch": 0.2971336293903916,
      "grad_norm": 196.20726013183594,
      "learning_rate": 1.6485676653187236e-05,
      "loss": 4.9607,
      "step": 7360
    },
    {
      "epoch": 0.29717400080742834,
      "grad_norm": 94.07072448730469,
      "learning_rate": 1.6484681323771313e-05,
      "loss": 4.825,
      "step": 7361
    },
    {
      "epoch": 0.29721437222446506,
      "grad_norm": 92.07304382324219,
      "learning_rate": 1.648368588348248e-05,
      "loss": 4.3785,
      "step": 7362
    },
    {
      "epoch": 0.29725474364150184,
      "grad_norm": 123.69258117675781,
      "learning_rate": 1.6482690332337764e-05,
      "loss": 5.1509,
      "step": 7363
    },
    {
      "epoch": 0.29729511505853856,
      "grad_norm": 92.61680603027344,
      "learning_rate": 1.6481694670354172e-05,
      "loss": 6.0284,
      "step": 7364
    },
    {
      "epoch": 0.2973354864755753,
      "grad_norm": 140.86907958984375,
      "learning_rate": 1.6480698897548746e-05,
      "loss": 5.8476,
      "step": 7365
    },
    {
      "epoch": 0.297375857892612,
      "grad_norm": 83.59830474853516,
      "learning_rate": 1.64797030139385e-05,
      "loss": 4.3476,
      "step": 7366
    },
    {
      "epoch": 0.2974162293096488,
      "grad_norm": 114.5889663696289,
      "learning_rate": 1.647870701954046e-05,
      "loss": 6.0813,
      "step": 7367
    },
    {
      "epoch": 0.2974566007266855,
      "grad_norm": 111.47695922851562,
      "learning_rate": 1.647771091437166e-05,
      "loss": 6.7025,
      "step": 7368
    },
    {
      "epoch": 0.2974969721437222,
      "grad_norm": 147.89097595214844,
      "learning_rate": 1.647671469844913e-05,
      "loss": 5.7554,
      "step": 7369
    },
    {
      "epoch": 0.297537343560759,
      "grad_norm": 120.29578399658203,
      "learning_rate": 1.6475718371789908e-05,
      "loss": 6.3477,
      "step": 7370
    },
    {
      "epoch": 0.2975777149777957,
      "grad_norm": 97.24530792236328,
      "learning_rate": 1.647472193441102e-05,
      "loss": 4.9786,
      "step": 7371
    },
    {
      "epoch": 0.29761808639483245,
      "grad_norm": 71.68708801269531,
      "learning_rate": 1.6473725386329508e-05,
      "loss": 7.255,
      "step": 7372
    },
    {
      "epoch": 0.29765845781186917,
      "grad_norm": 129.4801483154297,
      "learning_rate": 1.6472728727562405e-05,
      "loss": 4.6386,
      "step": 7373
    },
    {
      "epoch": 0.29769882922890595,
      "grad_norm": 86.58757781982422,
      "learning_rate": 1.6471731958126758e-05,
      "loss": 6.2615,
      "step": 7374
    },
    {
      "epoch": 0.29773920064594267,
      "grad_norm": 149.88247680664062,
      "learning_rate": 1.647073507803961e-05,
      "loss": 4.0003,
      "step": 7375
    },
    {
      "epoch": 0.2977795720629794,
      "grad_norm": 100.01637268066406,
      "learning_rate": 1.6469738087317997e-05,
      "loss": 4.7621,
      "step": 7376
    },
    {
      "epoch": 0.29781994348001617,
      "grad_norm": 100.63935852050781,
      "learning_rate": 1.646874098597897e-05,
      "loss": 4.6894,
      "step": 7377
    },
    {
      "epoch": 0.2978603148970529,
      "grad_norm": 92.35025787353516,
      "learning_rate": 1.646774377403958e-05,
      "loss": 7.4218,
      "step": 7378
    },
    {
      "epoch": 0.2979006863140896,
      "grad_norm": 123.66570281982422,
      "learning_rate": 1.646674645151688e-05,
      "loss": 6.2626,
      "step": 7379
    },
    {
      "epoch": 0.29794105773112634,
      "grad_norm": 105.70913696289062,
      "learning_rate": 1.646574901842791e-05,
      "loss": 3.5866,
      "step": 7380
    },
    {
      "epoch": 0.2979814291481631,
      "grad_norm": 133.49378967285156,
      "learning_rate": 1.6464751474789734e-05,
      "loss": 5.9461,
      "step": 7381
    },
    {
      "epoch": 0.29802180056519983,
      "grad_norm": 95.30084991455078,
      "learning_rate": 1.6463753820619405e-05,
      "loss": 5.7043,
      "step": 7382
    },
    {
      "epoch": 0.29806217198223656,
      "grad_norm": 72.0408706665039,
      "learning_rate": 1.6462756055933974e-05,
      "loss": 5.7908,
      "step": 7383
    },
    {
      "epoch": 0.29810254339927333,
      "grad_norm": 86.13438415527344,
      "learning_rate": 1.6461758180750513e-05,
      "loss": 6.1722,
      "step": 7384
    },
    {
      "epoch": 0.29814291481631006,
      "grad_norm": 109.46932220458984,
      "learning_rate": 1.646076019508607e-05,
      "loss": 5.4595,
      "step": 7385
    },
    {
      "epoch": 0.2981832862333468,
      "grad_norm": 95.2256851196289,
      "learning_rate": 1.6459762098957718e-05,
      "loss": 5.0836,
      "step": 7386
    },
    {
      "epoch": 0.29822365765038356,
      "grad_norm": 129.74771118164062,
      "learning_rate": 1.645876389238252e-05,
      "loss": 5.5296,
      "step": 7387
    },
    {
      "epoch": 0.2982640290674203,
      "grad_norm": 81.63229370117188,
      "learning_rate": 1.645776557537754e-05,
      "loss": 4.7985,
      "step": 7388
    },
    {
      "epoch": 0.298304400484457,
      "grad_norm": 89.42779541015625,
      "learning_rate": 1.6456767147959845e-05,
      "loss": 4.0877,
      "step": 7389
    },
    {
      "epoch": 0.2983447719014937,
      "grad_norm": 128.28822326660156,
      "learning_rate": 1.6455768610146514e-05,
      "loss": 6.1195,
      "step": 7390
    },
    {
      "epoch": 0.2983851433185305,
      "grad_norm": 115.11072540283203,
      "learning_rate": 1.645476996195462e-05,
      "loss": 4.7403,
      "step": 7391
    },
    {
      "epoch": 0.2984255147355672,
      "grad_norm": 112.14423370361328,
      "learning_rate": 1.6453771203401222e-05,
      "loss": 6.3218,
      "step": 7392
    },
    {
      "epoch": 0.29846588615260394,
      "grad_norm": 91.1868896484375,
      "learning_rate": 1.6452772334503414e-05,
      "loss": 4.1359,
      "step": 7393
    },
    {
      "epoch": 0.2985062575696407,
      "grad_norm": 84.99030303955078,
      "learning_rate": 1.6451773355278263e-05,
      "loss": 5.2579,
      "step": 7394
    },
    {
      "epoch": 0.29854662898667744,
      "grad_norm": 140.30050659179688,
      "learning_rate": 1.6450774265742856e-05,
      "loss": 5.6446,
      "step": 7395
    },
    {
      "epoch": 0.29858700040371416,
      "grad_norm": 105.44902038574219,
      "learning_rate": 1.6449775065914272e-05,
      "loss": 5.4909,
      "step": 7396
    },
    {
      "epoch": 0.2986273718207509,
      "grad_norm": 94.95169067382812,
      "learning_rate": 1.6448775755809595e-05,
      "loss": 6.6853,
      "step": 7397
    },
    {
      "epoch": 0.29866774323778766,
      "grad_norm": 74.57848358154297,
      "learning_rate": 1.6447776335445914e-05,
      "loss": 5.8855,
      "step": 7398
    },
    {
      "epoch": 0.2987081146548244,
      "grad_norm": 105.4181900024414,
      "learning_rate": 1.6446776804840316e-05,
      "loss": 7.4825,
      "step": 7399
    },
    {
      "epoch": 0.2987484860718611,
      "grad_norm": 59.21658706665039,
      "learning_rate": 1.6445777164009883e-05,
      "loss": 6.0544,
      "step": 7400
    },
    {
      "epoch": 0.2987888574888979,
      "grad_norm": 150.38290405273438,
      "learning_rate": 1.6444777412971716e-05,
      "loss": 5.1645,
      "step": 7401
    },
    {
      "epoch": 0.2988292289059346,
      "grad_norm": 140.06588745117188,
      "learning_rate": 1.6443777551742903e-05,
      "loss": 7.1237,
      "step": 7402
    },
    {
      "epoch": 0.29886960032297133,
      "grad_norm": 61.685951232910156,
      "learning_rate": 1.6442777580340543e-05,
      "loss": 5.9482,
      "step": 7403
    },
    {
      "epoch": 0.29890997174000805,
      "grad_norm": 103.20390319824219,
      "learning_rate": 1.644177749878173e-05,
      "loss": 4.5603,
      "step": 7404
    },
    {
      "epoch": 0.29895034315704483,
      "grad_norm": 102.26538848876953,
      "learning_rate": 1.644077730708357e-05,
      "loss": 4.6448,
      "step": 7405
    },
    {
      "epoch": 0.29899071457408155,
      "grad_norm": 142.1682891845703,
      "learning_rate": 1.643977700526315e-05,
      "loss": 5.2611,
      "step": 7406
    },
    {
      "epoch": 0.2990310859911183,
      "grad_norm": 71.87789154052734,
      "learning_rate": 1.6438776593337584e-05,
      "loss": 4.0346,
      "step": 7407
    },
    {
      "epoch": 0.29907145740815505,
      "grad_norm": 82.28510284423828,
      "learning_rate": 1.6437776071323974e-05,
      "loss": 4.826,
      "step": 7408
    },
    {
      "epoch": 0.2991118288251918,
      "grad_norm": 134.05398559570312,
      "learning_rate": 1.6436775439239428e-05,
      "loss": 5.9297,
      "step": 7409
    },
    {
      "epoch": 0.2991522002422285,
      "grad_norm": 289.9971618652344,
      "learning_rate": 1.6435774697101053e-05,
      "loss": 5.9577,
      "step": 7410
    },
    {
      "epoch": 0.2991925716592652,
      "grad_norm": 160.5531463623047,
      "learning_rate": 1.6434773844925956e-05,
      "loss": 4.0566,
      "step": 7411
    },
    {
      "epoch": 0.299232943076302,
      "grad_norm": 119.5596694946289,
      "learning_rate": 1.6433772882731254e-05,
      "loss": 6.3147,
      "step": 7412
    },
    {
      "epoch": 0.2992733144933387,
      "grad_norm": 98.60066223144531,
      "learning_rate": 1.6432771810534064e-05,
      "loss": 5.8292,
      "step": 7413
    },
    {
      "epoch": 0.29931368591037544,
      "grad_norm": 121.75408172607422,
      "learning_rate": 1.6431770628351492e-05,
      "loss": 6.2459,
      "step": 7414
    },
    {
      "epoch": 0.2993540573274122,
      "grad_norm": 144.2238006591797,
      "learning_rate": 1.6430769336200666e-05,
      "loss": 3.7489,
      "step": 7415
    },
    {
      "epoch": 0.29939442874444894,
      "grad_norm": 107.35824584960938,
      "learning_rate": 1.6429767934098698e-05,
      "loss": 5.0251,
      "step": 7416
    },
    {
      "epoch": 0.29943480016148566,
      "grad_norm": 184.15879821777344,
      "learning_rate": 1.6428766422062712e-05,
      "loss": 5.5271,
      "step": 7417
    },
    {
      "epoch": 0.2994751715785224,
      "grad_norm": 69.28072357177734,
      "learning_rate": 1.6427764800109836e-05,
      "loss": 5.5789,
      "step": 7418
    },
    {
      "epoch": 0.29951554299555916,
      "grad_norm": 101.97411346435547,
      "learning_rate": 1.642676306825719e-05,
      "loss": 5.8026,
      "step": 7419
    },
    {
      "epoch": 0.2995559144125959,
      "grad_norm": 96.647216796875,
      "learning_rate": 1.6425761226521904e-05,
      "loss": 4.4276,
      "step": 7420
    },
    {
      "epoch": 0.2995962858296326,
      "grad_norm": 101.05750274658203,
      "learning_rate": 1.6424759274921104e-05,
      "loss": 6.9687,
      "step": 7421
    },
    {
      "epoch": 0.2996366572466694,
      "grad_norm": 108.61560821533203,
      "learning_rate": 1.6423757213471923e-05,
      "loss": 4.2253,
      "step": 7422
    },
    {
      "epoch": 0.2996770286637061,
      "grad_norm": 89.5646743774414,
      "learning_rate": 1.64227550421915e-05,
      "loss": 6.066,
      "step": 7423
    },
    {
      "epoch": 0.2997174000807428,
      "grad_norm": 134.42266845703125,
      "learning_rate": 1.6421752761096957e-05,
      "loss": 2.8517,
      "step": 7424
    },
    {
      "epoch": 0.29975777149777955,
      "grad_norm": 92.86759948730469,
      "learning_rate": 1.6420750370205443e-05,
      "loss": 5.7441,
      "step": 7425
    },
    {
      "epoch": 0.2997981429148163,
      "grad_norm": 88.04438018798828,
      "learning_rate": 1.6419747869534086e-05,
      "loss": 6.0913,
      "step": 7426
    },
    {
      "epoch": 0.29983851433185305,
      "grad_norm": 119.01581573486328,
      "learning_rate": 1.6418745259100035e-05,
      "loss": 6.3656,
      "step": 7427
    },
    {
      "epoch": 0.29987888574888977,
      "grad_norm": 91.49837493896484,
      "learning_rate": 1.641774253892043e-05,
      "loss": 4.9146,
      "step": 7428
    },
    {
      "epoch": 0.29991925716592654,
      "grad_norm": 160.96945190429688,
      "learning_rate": 1.6416739709012413e-05,
      "loss": 6.568,
      "step": 7429
    },
    {
      "epoch": 0.29995962858296327,
      "grad_norm": 153.08984375,
      "learning_rate": 1.641573676939313e-05,
      "loss": 4.7997,
      "step": 7430
    },
    {
      "epoch": 0.3,
      "grad_norm": 80.49485778808594,
      "learning_rate": 1.641473372007973e-05,
      "loss": 5.6974,
      "step": 7431
    },
    {
      "epoch": 0.3000403714170367,
      "grad_norm": 91.77813720703125,
      "learning_rate": 1.641373056108936e-05,
      "loss": 2.4724,
      "step": 7432
    },
    {
      "epoch": 0.3000807428340735,
      "grad_norm": 147.56094360351562,
      "learning_rate": 1.6412727292439182e-05,
      "loss": 5.5193,
      "step": 7433
    },
    {
      "epoch": 0.3001211142511102,
      "grad_norm": 80.6565170288086,
      "learning_rate": 1.6411723914146336e-05,
      "loss": 3.3594,
      "step": 7434
    },
    {
      "epoch": 0.30016148566814693,
      "grad_norm": 209.06251525878906,
      "learning_rate": 1.6410720426227985e-05,
      "loss": 5.4855,
      "step": 7435
    },
    {
      "epoch": 0.3002018570851837,
      "grad_norm": 70.07958984375,
      "learning_rate": 1.6409716828701284e-05,
      "loss": 5.7668,
      "step": 7436
    },
    {
      "epoch": 0.30024222850222043,
      "grad_norm": 151.43666076660156,
      "learning_rate": 1.6408713121583397e-05,
      "loss": 5.6173,
      "step": 7437
    },
    {
      "epoch": 0.30028259991925715,
      "grad_norm": 138.11354064941406,
      "learning_rate": 1.6407709304891478e-05,
      "loss": 5.0665,
      "step": 7438
    },
    {
      "epoch": 0.30032297133629393,
      "grad_norm": 76.85538482666016,
      "learning_rate": 1.6406705378642695e-05,
      "loss": 5.8037,
      "step": 7439
    },
    {
      "epoch": 0.30036334275333065,
      "grad_norm": 122.46699523925781,
      "learning_rate": 1.6405701342854208e-05,
      "loss": 4.9858,
      "step": 7440
    },
    {
      "epoch": 0.3004037141703674,
      "grad_norm": 148.69537353515625,
      "learning_rate": 1.640469719754319e-05,
      "loss": 4.1503,
      "step": 7441
    },
    {
      "epoch": 0.3004440855874041,
      "grad_norm": 163.06040954589844,
      "learning_rate": 1.64036929427268e-05,
      "loss": 5.4677,
      "step": 7442
    },
    {
      "epoch": 0.3004844570044409,
      "grad_norm": 114.53446960449219,
      "learning_rate": 1.640268857842222e-05,
      "loss": 4.9313,
      "step": 7443
    },
    {
      "epoch": 0.3005248284214776,
      "grad_norm": 94.7508773803711,
      "learning_rate": 1.6401684104646613e-05,
      "loss": 5.3233,
      "step": 7444
    },
    {
      "epoch": 0.3005651998385143,
      "grad_norm": 194.34213256835938,
      "learning_rate": 1.640067952141716e-05,
      "loss": 3.2477,
      "step": 7445
    },
    {
      "epoch": 0.3006055712555511,
      "grad_norm": 97.69505310058594,
      "learning_rate": 1.6399674828751032e-05,
      "loss": 4.9524,
      "step": 7446
    },
    {
      "epoch": 0.3006459426725878,
      "grad_norm": 111.10091400146484,
      "learning_rate": 1.6398670026665408e-05,
      "loss": 7.8167,
      "step": 7447
    },
    {
      "epoch": 0.30068631408962454,
      "grad_norm": 71.85933685302734,
      "learning_rate": 1.639766511517747e-05,
      "loss": 5.1001,
      "step": 7448
    },
    {
      "epoch": 0.30072668550666126,
      "grad_norm": 93.1054916381836,
      "learning_rate": 1.63966600943044e-05,
      "loss": 5.5292,
      "step": 7449
    },
    {
      "epoch": 0.30076705692369804,
      "grad_norm": 93.54082489013672,
      "learning_rate": 1.6395654964063377e-05,
      "loss": 6.7803,
      "step": 7450
    },
    {
      "epoch": 0.30080742834073476,
      "grad_norm": 150.19432067871094,
      "learning_rate": 1.6394649724471587e-05,
      "loss": 4.9191,
      "step": 7451
    },
    {
      "epoch": 0.3008477997577715,
      "grad_norm": 91.11298370361328,
      "learning_rate": 1.639364437554622e-05,
      "loss": 5.0588,
      "step": 7452
    },
    {
      "epoch": 0.30088817117480826,
      "grad_norm": 148.30653381347656,
      "learning_rate": 1.6392638917304468e-05,
      "loss": 5.5868,
      "step": 7453
    },
    {
      "epoch": 0.300928542591845,
      "grad_norm": 109.98098754882812,
      "learning_rate": 1.6391633349763517e-05,
      "loss": 5.1504,
      "step": 7454
    },
    {
      "epoch": 0.3009689140088817,
      "grad_norm": 78.31835174560547,
      "learning_rate": 1.639062767294056e-05,
      "loss": 6.064,
      "step": 7455
    },
    {
      "epoch": 0.3010092854259184,
      "grad_norm": 116.11622619628906,
      "learning_rate": 1.6389621886852793e-05,
      "loss": 6.1526,
      "step": 7456
    },
    {
      "epoch": 0.3010496568429552,
      "grad_norm": 123.42716217041016,
      "learning_rate": 1.6388615991517417e-05,
      "loss": 7.2523,
      "step": 7457
    },
    {
      "epoch": 0.3010900282599919,
      "grad_norm": 86.3685073852539,
      "learning_rate": 1.638760998695162e-05,
      "loss": 5.6329,
      "step": 7458
    },
    {
      "epoch": 0.30113039967702865,
      "grad_norm": 78.2977066040039,
      "learning_rate": 1.6386603873172613e-05,
      "loss": 5.9957,
      "step": 7459
    },
    {
      "epoch": 0.3011707710940654,
      "grad_norm": 93.04610443115234,
      "learning_rate": 1.6385597650197587e-05,
      "loss": 4.5929,
      "step": 7460
    },
    {
      "epoch": 0.30121114251110215,
      "grad_norm": 94.50640106201172,
      "learning_rate": 1.6384591318043763e-05,
      "loss": 6.501,
      "step": 7461
    },
    {
      "epoch": 0.30125151392813887,
      "grad_norm": 81.31101989746094,
      "learning_rate": 1.6383584876728326e-05,
      "loss": 5.4117,
      "step": 7462
    },
    {
      "epoch": 0.3012918853451756,
      "grad_norm": 76.81295776367188,
      "learning_rate": 1.63825783262685e-05,
      "loss": 4.8899,
      "step": 7463
    },
    {
      "epoch": 0.30133225676221237,
      "grad_norm": 67.47710418701172,
      "learning_rate": 1.638157166668149e-05,
      "loss": 5.7964,
      "step": 7464
    },
    {
      "epoch": 0.3013726281792491,
      "grad_norm": 109.7152328491211,
      "learning_rate": 1.63805648979845e-05,
      "loss": 6.0029,
      "step": 7465
    },
    {
      "epoch": 0.3014129995962858,
      "grad_norm": 111.3666000366211,
      "learning_rate": 1.6379558020194756e-05,
      "loss": 6.2076,
      "step": 7466
    },
    {
      "epoch": 0.3014533710133226,
      "grad_norm": 72.46192169189453,
      "learning_rate": 1.6378551033329468e-05,
      "loss": 5.6294,
      "step": 7467
    },
    {
      "epoch": 0.3014937424303593,
      "grad_norm": 150.18234252929688,
      "learning_rate": 1.6377543937405848e-05,
      "loss": 5.9033,
      "step": 7468
    },
    {
      "epoch": 0.30153411384739603,
      "grad_norm": 118.85890197753906,
      "learning_rate": 1.6376536732441123e-05,
      "loss": 4.486,
      "step": 7469
    },
    {
      "epoch": 0.30157448526443276,
      "grad_norm": 99.81301879882812,
      "learning_rate": 1.6375529418452506e-05,
      "loss": 6.2878,
      "step": 7470
    },
    {
      "epoch": 0.30161485668146953,
      "grad_norm": 100.81294250488281,
      "learning_rate": 1.6374521995457227e-05,
      "loss": 6.9018,
      "step": 7471
    },
    {
      "epoch": 0.30165522809850626,
      "grad_norm": 86.0404281616211,
      "learning_rate": 1.6373514463472505e-05,
      "loss": 5.2415,
      "step": 7472
    },
    {
      "epoch": 0.301695599515543,
      "grad_norm": 77.68433380126953,
      "learning_rate": 1.637250682251557e-05,
      "loss": 4.8763,
      "step": 7473
    },
    {
      "epoch": 0.30173597093257976,
      "grad_norm": 80.02837371826172,
      "learning_rate": 1.637149907260365e-05,
      "loss": 4.0246,
      "step": 7474
    },
    {
      "epoch": 0.3017763423496165,
      "grad_norm": 119.59439849853516,
      "learning_rate": 1.6370491213753972e-05,
      "loss": 6.1395,
      "step": 7475
    },
    {
      "epoch": 0.3018167137666532,
      "grad_norm": 98.13127136230469,
      "learning_rate": 1.6369483245983775e-05,
      "loss": 5.438,
      "step": 7476
    },
    {
      "epoch": 0.3018570851836899,
      "grad_norm": 144.42739868164062,
      "learning_rate": 1.6368475169310284e-05,
      "loss": 5.2628,
      "step": 7477
    },
    {
      "epoch": 0.3018974566007267,
      "grad_norm": 118.9273452758789,
      "learning_rate": 1.6367466983750745e-05,
      "loss": 3.6057,
      "step": 7478
    },
    {
      "epoch": 0.3019378280177634,
      "grad_norm": 95.6353988647461,
      "learning_rate": 1.6366458689322384e-05,
      "loss": 6.8939,
      "step": 7479
    },
    {
      "epoch": 0.30197819943480014,
      "grad_norm": 85.68243408203125,
      "learning_rate": 1.6365450286042448e-05,
      "loss": 4.7444,
      "step": 7480
    },
    {
      "epoch": 0.3020185708518369,
      "grad_norm": 60.36145782470703,
      "learning_rate": 1.6364441773928174e-05,
      "loss": 6.2848,
      "step": 7481
    },
    {
      "epoch": 0.30205894226887364,
      "grad_norm": 109.01210021972656,
      "learning_rate": 1.6363433152996814e-05,
      "loss": 4.8884,
      "step": 7482
    },
    {
      "epoch": 0.30209931368591036,
      "grad_norm": 108.11356353759766,
      "learning_rate": 1.63624244232656e-05,
      "loss": 6.1792,
      "step": 7483
    },
    {
      "epoch": 0.3021396851029471,
      "grad_norm": 95.67403411865234,
      "learning_rate": 1.636141558475179e-05,
      "loss": 6.0394,
      "step": 7484
    },
    {
      "epoch": 0.30218005651998386,
      "grad_norm": 73.09271240234375,
      "learning_rate": 1.6360406637472625e-05,
      "loss": 6.6116,
      "step": 7485
    },
    {
      "epoch": 0.3022204279370206,
      "grad_norm": 78.3336181640625,
      "learning_rate": 1.6359397581445363e-05,
      "loss": 6.1321,
      "step": 7486
    },
    {
      "epoch": 0.3022607993540573,
      "grad_norm": 113.80927276611328,
      "learning_rate": 1.6358388416687246e-05,
      "loss": 5.1527,
      "step": 7487
    },
    {
      "epoch": 0.3023011707710941,
      "grad_norm": 105.12442779541016,
      "learning_rate": 1.6357379143215544e-05,
      "loss": 5.5676,
      "step": 7488
    },
    {
      "epoch": 0.3023415421881308,
      "grad_norm": 139.0950469970703,
      "learning_rate": 1.63563697610475e-05,
      "loss": 6.2593,
      "step": 7489
    },
    {
      "epoch": 0.30238191360516753,
      "grad_norm": 118.98853302001953,
      "learning_rate": 1.6355360270200376e-05,
      "loss": 5.7638,
      "step": 7490
    },
    {
      "epoch": 0.3024222850222043,
      "grad_norm": 96.2948226928711,
      "learning_rate": 1.6354350670691432e-05,
      "loss": 6.2247,
      "step": 7491
    },
    {
      "epoch": 0.30246265643924103,
      "grad_norm": 94.74166870117188,
      "learning_rate": 1.635334096253793e-05,
      "loss": 4.3353,
      "step": 7492
    },
    {
      "epoch": 0.30250302785627775,
      "grad_norm": 193.2660675048828,
      "learning_rate": 1.6352331145757132e-05,
      "loss": 5.2929,
      "step": 7493
    },
    {
      "epoch": 0.3025433992733145,
      "grad_norm": 120.67294311523438,
      "learning_rate": 1.635132122036631e-05,
      "loss": 4.0083,
      "step": 7494
    },
    {
      "epoch": 0.30258377069035125,
      "grad_norm": 141.52198791503906,
      "learning_rate": 1.635031118638272e-05,
      "loss": 4.0568,
      "step": 7495
    },
    {
      "epoch": 0.302624142107388,
      "grad_norm": 70.26030731201172,
      "learning_rate": 1.6349301043823644e-05,
      "loss": 6.9858,
      "step": 7496
    },
    {
      "epoch": 0.3026645135244247,
      "grad_norm": 127.94802856445312,
      "learning_rate": 1.634829079270634e-05,
      "loss": 5.628,
      "step": 7497
    },
    {
      "epoch": 0.30270488494146147,
      "grad_norm": 84.4966812133789,
      "learning_rate": 1.6347280433048095e-05,
      "loss": 6.2265,
      "step": 7498
    },
    {
      "epoch": 0.3027452563584982,
      "grad_norm": 114.0522232055664,
      "learning_rate": 1.6346269964866172e-05,
      "loss": 5.9715,
      "step": 7499
    },
    {
      "epoch": 0.3027856277755349,
      "grad_norm": 172.15426635742188,
      "learning_rate": 1.6345259388177855e-05,
      "loss": 4.4397,
      "step": 7500
    },
    {
      "epoch": 0.30282599919257164,
      "grad_norm": 105.71906280517578,
      "learning_rate": 1.634424870300042e-05,
      "loss": 5.394,
      "step": 7501
    },
    {
      "epoch": 0.3028663706096084,
      "grad_norm": 145.44216918945312,
      "learning_rate": 1.6343237909351144e-05,
      "loss": 3.7862,
      "step": 7502
    },
    {
      "epoch": 0.30290674202664514,
      "grad_norm": 103.76620483398438,
      "learning_rate": 1.6342227007247314e-05,
      "loss": 4.3519,
      "step": 7503
    },
    {
      "epoch": 0.30294711344368186,
      "grad_norm": 87.71084594726562,
      "learning_rate": 1.6341215996706213e-05,
      "loss": 5.1226,
      "step": 7504
    },
    {
      "epoch": 0.30298748486071864,
      "grad_norm": 151.5065460205078,
      "learning_rate": 1.6340204877745126e-05,
      "loss": 4.4476,
      "step": 7505
    },
    {
      "epoch": 0.30302785627775536,
      "grad_norm": 104.3457260131836,
      "learning_rate": 1.6339193650381338e-05,
      "loss": 4.6274,
      "step": 7506
    },
    {
      "epoch": 0.3030682276947921,
      "grad_norm": 71.34832000732422,
      "learning_rate": 1.6338182314632143e-05,
      "loss": 5.2507,
      "step": 7507
    },
    {
      "epoch": 0.3031085991118288,
      "grad_norm": 93.67102813720703,
      "learning_rate": 1.6337170870514833e-05,
      "loss": 5.3779,
      "step": 7508
    },
    {
      "epoch": 0.3031489705288656,
      "grad_norm": 134.8176727294922,
      "learning_rate": 1.6336159318046698e-05,
      "loss": 5.4348,
      "step": 7509
    },
    {
      "epoch": 0.3031893419459023,
      "grad_norm": 97.72801971435547,
      "learning_rate": 1.6335147657245038e-05,
      "loss": 5.4747,
      "step": 7510
    },
    {
      "epoch": 0.303229713362939,
      "grad_norm": 125.03084564208984,
      "learning_rate": 1.6334135888127143e-05,
      "loss": 5.7421,
      "step": 7511
    },
    {
      "epoch": 0.3032700847799758,
      "grad_norm": 94.94905853271484,
      "learning_rate": 1.633312401071032e-05,
      "loss": 6.3845,
      "step": 7512
    },
    {
      "epoch": 0.3033104561970125,
      "grad_norm": 212.87261962890625,
      "learning_rate": 1.6332112025011858e-05,
      "loss": 4.9794,
      "step": 7513
    },
    {
      "epoch": 0.30335082761404925,
      "grad_norm": 110.31099700927734,
      "learning_rate": 1.6331099931049072e-05,
      "loss": 4.915,
      "step": 7514
    },
    {
      "epoch": 0.30339119903108597,
      "grad_norm": 107.5264892578125,
      "learning_rate": 1.633008772883926e-05,
      "loss": 3.5251,
      "step": 7515
    },
    {
      "epoch": 0.30343157044812274,
      "grad_norm": 88.71165466308594,
      "learning_rate": 1.6329075418399728e-05,
      "loss": 6.3352,
      "step": 7516
    },
    {
      "epoch": 0.30347194186515947,
      "grad_norm": 111.61103820800781,
      "learning_rate": 1.632806299974779e-05,
      "loss": 6.4141,
      "step": 7517
    },
    {
      "epoch": 0.3035123132821962,
      "grad_norm": 258.53790283203125,
      "learning_rate": 1.632705047290075e-05,
      "loss": 4.3971,
      "step": 7518
    },
    {
      "epoch": 0.30355268469923297,
      "grad_norm": 100.62364196777344,
      "learning_rate": 1.632603783787592e-05,
      "loss": 5.4142,
      "step": 7519
    },
    {
      "epoch": 0.3035930561162697,
      "grad_norm": 99.11235046386719,
      "learning_rate": 1.6325025094690615e-05,
      "loss": 5.1741,
      "step": 7520
    },
    {
      "epoch": 0.3036334275333064,
      "grad_norm": 98.2437973022461,
      "learning_rate": 1.6324012243362152e-05,
      "loss": 5.7469,
      "step": 7521
    },
    {
      "epoch": 0.30367379895034313,
      "grad_norm": 67.98306274414062,
      "learning_rate": 1.632299928390785e-05,
      "loss": 7.0593,
      "step": 7522
    },
    {
      "epoch": 0.3037141703673799,
      "grad_norm": 135.50209045410156,
      "learning_rate": 1.6321986216345024e-05,
      "loss": 4.7323,
      "step": 7523
    },
    {
      "epoch": 0.30375454178441663,
      "grad_norm": 145.43495178222656,
      "learning_rate": 1.6320973040690996e-05,
      "loss": 5.3693,
      "step": 7524
    },
    {
      "epoch": 0.30379491320145335,
      "grad_norm": 113.16134643554688,
      "learning_rate": 1.631995975696309e-05,
      "loss": 5.5842,
      "step": 7525
    },
    {
      "epoch": 0.30383528461849013,
      "grad_norm": 120.9155502319336,
      "learning_rate": 1.6318946365178627e-05,
      "loss": 4.8267,
      "step": 7526
    },
    {
      "epoch": 0.30387565603552685,
      "grad_norm": 98.59215545654297,
      "learning_rate": 1.631793286535494e-05,
      "loss": 6.9987,
      "step": 7527
    },
    {
      "epoch": 0.3039160274525636,
      "grad_norm": 86.31942749023438,
      "learning_rate": 1.6316919257509358e-05,
      "loss": 5.435,
      "step": 7528
    },
    {
      "epoch": 0.3039563988696003,
      "grad_norm": 85.65367126464844,
      "learning_rate": 1.63159055416592e-05,
      "loss": 5.0964,
      "step": 7529
    },
    {
      "epoch": 0.3039967702866371,
      "grad_norm": 90.46766662597656,
      "learning_rate": 1.631489171782181e-05,
      "loss": 5.6095,
      "step": 7530
    },
    {
      "epoch": 0.3040371417036738,
      "grad_norm": 115.39685821533203,
      "learning_rate": 1.6313877786014517e-05,
      "loss": 4.7283,
      "step": 7531
    },
    {
      "epoch": 0.3040775131207105,
      "grad_norm": 105.03810119628906,
      "learning_rate": 1.6312863746254658e-05,
      "loss": 4.1039,
      "step": 7532
    },
    {
      "epoch": 0.3041178845377473,
      "grad_norm": 86.06854248046875,
      "learning_rate": 1.6311849598559572e-05,
      "loss": 5.4036,
      "step": 7533
    },
    {
      "epoch": 0.304158255954784,
      "grad_norm": 165.1433868408203,
      "learning_rate": 1.6310835342946597e-05,
      "loss": 5.2329,
      "step": 7534
    },
    {
      "epoch": 0.30419862737182074,
      "grad_norm": 112.48160552978516,
      "learning_rate": 1.6309820979433073e-05,
      "loss": 5.1168,
      "step": 7535
    },
    {
      "epoch": 0.30423899878885746,
      "grad_norm": 67.8758544921875,
      "learning_rate": 1.6308806508036347e-05,
      "loss": 5.9348,
      "step": 7536
    },
    {
      "epoch": 0.30427937020589424,
      "grad_norm": 98.43855285644531,
      "learning_rate": 1.6307791928773756e-05,
      "loss": 5.1697,
      "step": 7537
    },
    {
      "epoch": 0.30431974162293096,
      "grad_norm": 104.2973861694336,
      "learning_rate": 1.630677724166266e-05,
      "loss": 6.1525,
      "step": 7538
    },
    {
      "epoch": 0.3043601130399677,
      "grad_norm": 155.6044158935547,
      "learning_rate": 1.6305762446720397e-05,
      "loss": 4.5703,
      "step": 7539
    },
    {
      "epoch": 0.30440048445700446,
      "grad_norm": 130.35140991210938,
      "learning_rate": 1.630474754396432e-05,
      "loss": 5.8151,
      "step": 7540
    },
    {
      "epoch": 0.3044408558740412,
      "grad_norm": 119.53195190429688,
      "learning_rate": 1.6303732533411784e-05,
      "loss": 4.1088,
      "step": 7541
    },
    {
      "epoch": 0.3044812272910779,
      "grad_norm": 69.62908172607422,
      "learning_rate": 1.6302717415080143e-05,
      "loss": 5.2324,
      "step": 7542
    },
    {
      "epoch": 0.3045215987081147,
      "grad_norm": 84.2533187866211,
      "learning_rate": 1.630170218898675e-05,
      "loss": 6.3547,
      "step": 7543
    },
    {
      "epoch": 0.3045619701251514,
      "grad_norm": 96.83975219726562,
      "learning_rate": 1.630068685514897e-05,
      "loss": 5.8652,
      "step": 7544
    },
    {
      "epoch": 0.3046023415421881,
      "grad_norm": 223.7336883544922,
      "learning_rate": 1.629967141358415e-05,
      "loss": 4.8446,
      "step": 7545
    },
    {
      "epoch": 0.30464271295922485,
      "grad_norm": 102.29183959960938,
      "learning_rate": 1.6298655864309665e-05,
      "loss": 5.8862,
      "step": 7546
    },
    {
      "epoch": 0.3046830843762616,
      "grad_norm": 181.0918731689453,
      "learning_rate": 1.629764020734287e-05,
      "loss": 4.8984,
      "step": 7547
    },
    {
      "epoch": 0.30472345579329835,
      "grad_norm": 108.50386047363281,
      "learning_rate": 1.6296624442701138e-05,
      "loss": 4.9886,
      "step": 7548
    },
    {
      "epoch": 0.30476382721033507,
      "grad_norm": 121.81070709228516,
      "learning_rate": 1.629560857040183e-05,
      "loss": 5.4662,
      "step": 7549
    },
    {
      "epoch": 0.30480419862737185,
      "grad_norm": 104.35112762451172,
      "learning_rate": 1.6294592590462317e-05,
      "loss": 5.1479,
      "step": 7550
    },
    {
      "epoch": 0.30484457004440857,
      "grad_norm": 90.77881622314453,
      "learning_rate": 1.6293576502899968e-05,
      "loss": 5.5684,
      "step": 7551
    },
    {
      "epoch": 0.3048849414614453,
      "grad_norm": 110.94636535644531,
      "learning_rate": 1.629256030773216e-05,
      "loss": 3.6139,
      "step": 7552
    },
    {
      "epoch": 0.304925312878482,
      "grad_norm": 70.87190246582031,
      "learning_rate": 1.629154400497626e-05,
      "loss": 6.116,
      "step": 7553
    },
    {
      "epoch": 0.3049656842955188,
      "grad_norm": 106.41337585449219,
      "learning_rate": 1.6290527594649655e-05,
      "loss": 5.1564,
      "step": 7554
    },
    {
      "epoch": 0.3050060557125555,
      "grad_norm": 92.29442596435547,
      "learning_rate": 1.6289511076769714e-05,
      "loss": 6.9186,
      "step": 7555
    },
    {
      "epoch": 0.30504642712959223,
      "grad_norm": 74.5460433959961,
      "learning_rate": 1.6288494451353824e-05,
      "loss": 6.3716,
      "step": 7556
    },
    {
      "epoch": 0.305086798546629,
      "grad_norm": 112.37510681152344,
      "learning_rate": 1.628747771841936e-05,
      "loss": 4.803,
      "step": 7557
    },
    {
      "epoch": 0.30512716996366573,
      "grad_norm": 94.90031433105469,
      "learning_rate": 1.628646087798371e-05,
      "loss": 5.4335,
      "step": 7558
    },
    {
      "epoch": 0.30516754138070246,
      "grad_norm": 76.2055892944336,
      "learning_rate": 1.628544393006426e-05,
      "loss": 5.1375,
      "step": 7559
    },
    {
      "epoch": 0.3052079127977392,
      "grad_norm": 129.70970153808594,
      "learning_rate": 1.6284426874678396e-05,
      "loss": 5.378,
      "step": 7560
    },
    {
      "epoch": 0.30524828421477596,
      "grad_norm": 106.47102355957031,
      "learning_rate": 1.6283409711843506e-05,
      "loss": 3.8904,
      "step": 7561
    },
    {
      "epoch": 0.3052886556318127,
      "grad_norm": 73.99110412597656,
      "learning_rate": 1.6282392441576985e-05,
      "loss": 4.1458,
      "step": 7562
    },
    {
      "epoch": 0.3053290270488494,
      "grad_norm": 102.9220199584961,
      "learning_rate": 1.628137506389622e-05,
      "loss": 4.3067,
      "step": 7563
    },
    {
      "epoch": 0.3053693984658862,
      "grad_norm": 112.48258209228516,
      "learning_rate": 1.6280357578818616e-05,
      "loss": 4.3785,
      "step": 7564
    },
    {
      "epoch": 0.3054097698829229,
      "grad_norm": 97.90451049804688,
      "learning_rate": 1.6279339986361558e-05,
      "loss": 7.0874,
      "step": 7565
    },
    {
      "epoch": 0.3054501412999596,
      "grad_norm": 107.09695434570312,
      "learning_rate": 1.627832228654245e-05,
      "loss": 6.0751,
      "step": 7566
    },
    {
      "epoch": 0.30549051271699634,
      "grad_norm": 101.89546966552734,
      "learning_rate": 1.627730447937869e-05,
      "loss": 4.1493,
      "step": 7567
    },
    {
      "epoch": 0.3055308841340331,
      "grad_norm": 79.65323638916016,
      "learning_rate": 1.627628656488769e-05,
      "loss": 5.328,
      "step": 7568
    },
    {
      "epoch": 0.30557125555106984,
      "grad_norm": 87.949462890625,
      "learning_rate": 1.627526854308684e-05,
      "loss": 5.2566,
      "step": 7569
    },
    {
      "epoch": 0.30561162696810656,
      "grad_norm": 88.39444732666016,
      "learning_rate": 1.6274250413993554e-05,
      "loss": 5.1231,
      "step": 7570
    },
    {
      "epoch": 0.30565199838514334,
      "grad_norm": 85.77349090576172,
      "learning_rate": 1.6273232177625234e-05,
      "loss": 4.2596,
      "step": 7571
    },
    {
      "epoch": 0.30569236980218006,
      "grad_norm": 167.4271240234375,
      "learning_rate": 1.6272213833999294e-05,
      "loss": 6.3497,
      "step": 7572
    },
    {
      "epoch": 0.3057327412192168,
      "grad_norm": 92.48560333251953,
      "learning_rate": 1.6271195383133145e-05,
      "loss": 5.2372,
      "step": 7573
    },
    {
      "epoch": 0.3057731126362535,
      "grad_norm": 115.70477294921875,
      "learning_rate": 1.62701768250442e-05,
      "loss": 4.5364,
      "step": 7574
    },
    {
      "epoch": 0.3058134840532903,
      "grad_norm": 116.05189514160156,
      "learning_rate": 1.626915815974987e-05,
      "loss": 5.8097,
      "step": 7575
    },
    {
      "epoch": 0.305853855470327,
      "grad_norm": 138.5540771484375,
      "learning_rate": 1.6268139387267578e-05,
      "loss": 5.1317,
      "step": 7576
    },
    {
      "epoch": 0.30589422688736373,
      "grad_norm": 137.4619903564453,
      "learning_rate": 1.6267120507614736e-05,
      "loss": 5.6025,
      "step": 7577
    },
    {
      "epoch": 0.3059345983044005,
      "grad_norm": 111.92935180664062,
      "learning_rate": 1.626610152080877e-05,
      "loss": 5.4502,
      "step": 7578
    },
    {
      "epoch": 0.30597496972143723,
      "grad_norm": 119.9725112915039,
      "learning_rate": 1.6265082426867104e-05,
      "loss": 6.2426,
      "step": 7579
    },
    {
      "epoch": 0.30601534113847395,
      "grad_norm": 168.3830108642578,
      "learning_rate": 1.6264063225807157e-05,
      "loss": 4.1338,
      "step": 7580
    },
    {
      "epoch": 0.3060557125555107,
      "grad_norm": 85.67469787597656,
      "learning_rate": 1.626304391764635e-05,
      "loss": 4.7122,
      "step": 7581
    },
    {
      "epoch": 0.30609608397254745,
      "grad_norm": 100.25732421875,
      "learning_rate": 1.6262024502402126e-05,
      "loss": 5.3343,
      "step": 7582
    },
    {
      "epoch": 0.3061364553895842,
      "grad_norm": 147.8374481201172,
      "learning_rate": 1.6261004980091903e-05,
      "loss": 4.2102,
      "step": 7583
    },
    {
      "epoch": 0.3061768268066209,
      "grad_norm": 118.34778594970703,
      "learning_rate": 1.6259985350733117e-05,
      "loss": 5.6689,
      "step": 7584
    },
    {
      "epoch": 0.30621719822365767,
      "grad_norm": 84.73841094970703,
      "learning_rate": 1.6258965614343197e-05,
      "loss": 5.6373,
      "step": 7585
    },
    {
      "epoch": 0.3062575696406944,
      "grad_norm": 79.9382095336914,
      "learning_rate": 1.6257945770939582e-05,
      "loss": 4.9158,
      "step": 7586
    },
    {
      "epoch": 0.3062979410577311,
      "grad_norm": 78.21572875976562,
      "learning_rate": 1.6256925820539703e-05,
      "loss": 6.5312,
      "step": 7587
    },
    {
      "epoch": 0.30633831247476784,
      "grad_norm": 112.82093811035156,
      "learning_rate": 1.625590576316101e-05,
      "loss": 4.52,
      "step": 7588
    },
    {
      "epoch": 0.3063786838918046,
      "grad_norm": 97.1787338256836,
      "learning_rate": 1.6254885598820935e-05,
      "loss": 4.2698,
      "step": 7589
    },
    {
      "epoch": 0.30641905530884134,
      "grad_norm": 184.3574981689453,
      "learning_rate": 1.625386532753692e-05,
      "loss": 6.2381,
      "step": 7590
    },
    {
      "epoch": 0.30645942672587806,
      "grad_norm": 105.17315673828125,
      "learning_rate": 1.625284494932641e-05,
      "loss": 5.9108,
      "step": 7591
    },
    {
      "epoch": 0.30649979814291484,
      "grad_norm": 113.68411254882812,
      "learning_rate": 1.625182446420686e-05,
      "loss": 5.3663,
      "step": 7592
    },
    {
      "epoch": 0.30654016955995156,
      "grad_norm": 165.33998107910156,
      "learning_rate": 1.6250803872195706e-05,
      "loss": 4.661,
      "step": 7593
    },
    {
      "epoch": 0.3065805409769883,
      "grad_norm": 107.42577362060547,
      "learning_rate": 1.6249783173310403e-05,
      "loss": 5.0892,
      "step": 7594
    },
    {
      "epoch": 0.306620912394025,
      "grad_norm": 134.5416717529297,
      "learning_rate": 1.62487623675684e-05,
      "loss": 4.9866,
      "step": 7595
    },
    {
      "epoch": 0.3066612838110618,
      "grad_norm": 69.3499984741211,
      "learning_rate": 1.624774145498716e-05,
      "loss": 4.6366,
      "step": 7596
    },
    {
      "epoch": 0.3067016552280985,
      "grad_norm": 83.708251953125,
      "learning_rate": 1.6246720435584124e-05,
      "loss": 6.1869,
      "step": 7597
    },
    {
      "epoch": 0.3067420266451352,
      "grad_norm": 78.09471130371094,
      "learning_rate": 1.6245699309376758e-05,
      "loss": 5.0622,
      "step": 7598
    },
    {
      "epoch": 0.306782398062172,
      "grad_norm": 78.0210952758789,
      "learning_rate": 1.6244678076382518e-05,
      "loss": 6.6532,
      "step": 7599
    },
    {
      "epoch": 0.3068227694792087,
      "grad_norm": 134.47447204589844,
      "learning_rate": 1.6243656736618863e-05,
      "loss": 7.1809,
      "step": 7600
    },
    {
      "epoch": 0.30686314089624545,
      "grad_norm": 147.9116668701172,
      "learning_rate": 1.624263529010326e-05,
      "loss": 5.4436,
      "step": 7601
    },
    {
      "epoch": 0.3069035123132822,
      "grad_norm": 156.8108367919922,
      "learning_rate": 1.624161373685317e-05,
      "loss": 4.7993,
      "step": 7602
    },
    {
      "epoch": 0.30694388373031894,
      "grad_norm": 101.67945098876953,
      "learning_rate": 1.6240592076886065e-05,
      "loss": 5.9857,
      "step": 7603
    },
    {
      "epoch": 0.30698425514735567,
      "grad_norm": 125.0936050415039,
      "learning_rate": 1.62395703102194e-05,
      "loss": 5.1581,
      "step": 7604
    },
    {
      "epoch": 0.3070246265643924,
      "grad_norm": 222.79762268066406,
      "learning_rate": 1.6238548436870658e-05,
      "loss": 4.9797,
      "step": 7605
    },
    {
      "epoch": 0.30706499798142917,
      "grad_norm": 94.46681213378906,
      "learning_rate": 1.623752645685731e-05,
      "loss": 5.9084,
      "step": 7606
    },
    {
      "epoch": 0.3071053693984659,
      "grad_norm": 133.17832946777344,
      "learning_rate": 1.623650437019682e-05,
      "loss": 5.17,
      "step": 7607
    },
    {
      "epoch": 0.3071457408155026,
      "grad_norm": 84.02716827392578,
      "learning_rate": 1.6235482176906668e-05,
      "loss": 5.6404,
      "step": 7608
    },
    {
      "epoch": 0.3071861122325394,
      "grad_norm": 92.60934448242188,
      "learning_rate": 1.623445987700433e-05,
      "loss": 6.4172,
      "step": 7609
    },
    {
      "epoch": 0.3072264836495761,
      "grad_norm": 85.36443328857422,
      "learning_rate": 1.623343747050729e-05,
      "loss": 5.4932,
      "step": 7610
    },
    {
      "epoch": 0.30726685506661283,
      "grad_norm": 109.06576538085938,
      "learning_rate": 1.623241495743303e-05,
      "loss": 5.3512,
      "step": 7611
    },
    {
      "epoch": 0.30730722648364955,
      "grad_norm": 76.47566223144531,
      "learning_rate": 1.6231392337799023e-05,
      "loss": 5.2912,
      "step": 7612
    },
    {
      "epoch": 0.30734759790068633,
      "grad_norm": 278.4664001464844,
      "learning_rate": 1.6230369611622757e-05,
      "loss": 5.309,
      "step": 7613
    },
    {
      "epoch": 0.30738796931772305,
      "grad_norm": 95.58027648925781,
      "learning_rate": 1.622934677892172e-05,
      "loss": 5.878,
      "step": 7614
    },
    {
      "epoch": 0.3074283407347598,
      "grad_norm": 90.37091064453125,
      "learning_rate": 1.62283238397134e-05,
      "loss": 6.1954,
      "step": 7615
    },
    {
      "epoch": 0.30746871215179655,
      "grad_norm": 133.42059326171875,
      "learning_rate": 1.6227300794015288e-05,
      "loss": 6.347,
      "step": 7616
    },
    {
      "epoch": 0.3075090835688333,
      "grad_norm": 75.6248550415039,
      "learning_rate": 1.6226277641844872e-05,
      "loss": 3.5818,
      "step": 7617
    },
    {
      "epoch": 0.30754945498587,
      "grad_norm": 87.49890899658203,
      "learning_rate": 1.622525438321965e-05,
      "loss": 5.0574,
      "step": 7618
    },
    {
      "epoch": 0.3075898264029067,
      "grad_norm": 84.61186218261719,
      "learning_rate": 1.622423101815711e-05,
      "loss": 6.1445,
      "step": 7619
    },
    {
      "epoch": 0.3076301978199435,
      "grad_norm": 83.2783432006836,
      "learning_rate": 1.6223207546674763e-05,
      "loss": 6.3844,
      "step": 7620
    },
    {
      "epoch": 0.3076705692369802,
      "grad_norm": 181.19085693359375,
      "learning_rate": 1.6222183968790093e-05,
      "loss": 5.1097,
      "step": 7621
    },
    {
      "epoch": 0.30771094065401694,
      "grad_norm": 89.96211242675781,
      "learning_rate": 1.622116028452061e-05,
      "loss": 4.6909,
      "step": 7622
    },
    {
      "epoch": 0.3077513120710537,
      "grad_norm": 80.41666412353516,
      "learning_rate": 1.6220136493883813e-05,
      "loss": 5.11,
      "step": 7623
    },
    {
      "epoch": 0.30779168348809044,
      "grad_norm": 140.24322509765625,
      "learning_rate": 1.6219112596897204e-05,
      "loss": 4.9489,
      "step": 7624
    },
    {
      "epoch": 0.30783205490512716,
      "grad_norm": 116.81394958496094,
      "learning_rate": 1.6218088593578294e-05,
      "loss": 3.2998,
      "step": 7625
    },
    {
      "epoch": 0.3078724263221639,
      "grad_norm": 95.44110870361328,
      "learning_rate": 1.6217064483944593e-05,
      "loss": 4.8129,
      "step": 7626
    },
    {
      "epoch": 0.30791279773920066,
      "grad_norm": 82.69427490234375,
      "learning_rate": 1.62160402680136e-05,
      "loss": 6.0123,
      "step": 7627
    },
    {
      "epoch": 0.3079531691562374,
      "grad_norm": 87.79987335205078,
      "learning_rate": 1.6215015945802838e-05,
      "loss": 6.2269,
      "step": 7628
    },
    {
      "epoch": 0.3079935405732741,
      "grad_norm": 74.06529235839844,
      "learning_rate": 1.6213991517329817e-05,
      "loss": 5.0326,
      "step": 7629
    },
    {
      "epoch": 0.3080339119903109,
      "grad_norm": 149.5891571044922,
      "learning_rate": 1.6212966982612052e-05,
      "loss": 5.7029,
      "step": 7630
    },
    {
      "epoch": 0.3080742834073476,
      "grad_norm": 241.68252563476562,
      "learning_rate": 1.6211942341667058e-05,
      "loss": 6.1613,
      "step": 7631
    },
    {
      "epoch": 0.3081146548243843,
      "grad_norm": 77.13164520263672,
      "learning_rate": 1.6210917594512355e-05,
      "loss": 4.5038,
      "step": 7632
    },
    {
      "epoch": 0.30815502624142105,
      "grad_norm": 122.88862609863281,
      "learning_rate": 1.6209892741165467e-05,
      "loss": 6.1371,
      "step": 7633
    },
    {
      "epoch": 0.3081953976584578,
      "grad_norm": 88.55518341064453,
      "learning_rate": 1.6208867781643914e-05,
      "loss": 5.5136,
      "step": 7634
    },
    {
      "epoch": 0.30823576907549455,
      "grad_norm": 117.6596450805664,
      "learning_rate": 1.6207842715965223e-05,
      "loss": 5.871,
      "step": 7635
    },
    {
      "epoch": 0.30827614049253127,
      "grad_norm": 90.33196258544922,
      "learning_rate": 1.6206817544146917e-05,
      "loss": 4.576,
      "step": 7636
    },
    {
      "epoch": 0.30831651190956805,
      "grad_norm": 182.34181213378906,
      "learning_rate": 1.620579226620652e-05,
      "loss": 4.7203,
      "step": 7637
    },
    {
      "epoch": 0.30835688332660477,
      "grad_norm": 96.1950912475586,
      "learning_rate": 1.620476688216157e-05,
      "loss": 6.0535,
      "step": 7638
    },
    {
      "epoch": 0.3083972547436415,
      "grad_norm": 75.50479888916016,
      "learning_rate": 1.6203741392029593e-05,
      "loss": 3.9925,
      "step": 7639
    },
    {
      "epoch": 0.3084376261606782,
      "grad_norm": 95.36465454101562,
      "learning_rate": 1.6202715795828133e-05,
      "loss": 5.5053,
      "step": 7640
    },
    {
      "epoch": 0.308477997577715,
      "grad_norm": 104.79139709472656,
      "learning_rate": 1.620169009357471e-05,
      "loss": 5.5698,
      "step": 7641
    },
    {
      "epoch": 0.3085183689947517,
      "grad_norm": 96.10147857666016,
      "learning_rate": 1.620066428528687e-05,
      "loss": 8.0721,
      "step": 7642
    },
    {
      "epoch": 0.30855874041178843,
      "grad_norm": 209.25209045410156,
      "learning_rate": 1.6199638370982152e-05,
      "loss": 4.7656,
      "step": 7643
    },
    {
      "epoch": 0.3085991118288252,
      "grad_norm": 109.20834350585938,
      "learning_rate": 1.6198612350678092e-05,
      "loss": 5.7072,
      "step": 7644
    },
    {
      "epoch": 0.30863948324586193,
      "grad_norm": 94.01226806640625,
      "learning_rate": 1.619758622439224e-05,
      "loss": 5.4121,
      "step": 7645
    },
    {
      "epoch": 0.30867985466289866,
      "grad_norm": 117.97901916503906,
      "learning_rate": 1.619655999214213e-05,
      "loss": 6.7984,
      "step": 7646
    },
    {
      "epoch": 0.3087202260799354,
      "grad_norm": 99.0185317993164,
      "learning_rate": 1.619553365394532e-05,
      "loss": 5.7814,
      "step": 7647
    },
    {
      "epoch": 0.30876059749697216,
      "grad_norm": 87.52592468261719,
      "learning_rate": 1.619450720981935e-05,
      "loss": 4.9834,
      "step": 7648
    },
    {
      "epoch": 0.3088009689140089,
      "grad_norm": 90.74344635009766,
      "learning_rate": 1.6193480659781768e-05,
      "loss": 5.4497,
      "step": 7649
    },
    {
      "epoch": 0.3088413403310456,
      "grad_norm": 123.290283203125,
      "learning_rate": 1.6192454003850136e-05,
      "loss": 6.2838,
      "step": 7650
    },
    {
      "epoch": 0.3088817117480824,
      "grad_norm": 229.6923828125,
      "learning_rate": 1.6191427242041995e-05,
      "loss": 4.6707,
      "step": 7651
    },
    {
      "epoch": 0.3089220831651191,
      "grad_norm": 246.45887756347656,
      "learning_rate": 1.6190400374374912e-05,
      "loss": 5.4562,
      "step": 7652
    },
    {
      "epoch": 0.3089624545821558,
      "grad_norm": 156.38592529296875,
      "learning_rate": 1.6189373400866434e-05,
      "loss": 6.4612,
      "step": 7653
    },
    {
      "epoch": 0.3090028259991926,
      "grad_norm": 94.43264770507812,
      "learning_rate": 1.6188346321534125e-05,
      "loss": 5.5159,
      "step": 7654
    },
    {
      "epoch": 0.3090431974162293,
      "grad_norm": 166.73182678222656,
      "learning_rate": 1.6187319136395542e-05,
      "loss": 4.5747,
      "step": 7655
    },
    {
      "epoch": 0.30908356883326604,
      "grad_norm": 76.16055297851562,
      "learning_rate": 1.6186291845468257e-05,
      "loss": 5.7592,
      "step": 7656
    },
    {
      "epoch": 0.30912394025030276,
      "grad_norm": 164.91566467285156,
      "learning_rate": 1.6185264448769823e-05,
      "loss": 5.3321,
      "step": 7657
    },
    {
      "epoch": 0.30916431166733954,
      "grad_norm": 81.40847778320312,
      "learning_rate": 1.618423694631781e-05,
      "loss": 2.8024,
      "step": 7658
    },
    {
      "epoch": 0.30920468308437626,
      "grad_norm": 127.86139678955078,
      "learning_rate": 1.618320933812979e-05,
      "loss": 5.2706,
      "step": 7659
    },
    {
      "epoch": 0.309245054501413,
      "grad_norm": 146.29782104492188,
      "learning_rate": 1.6182181624223326e-05,
      "loss": 5.7756,
      "step": 7660
    },
    {
      "epoch": 0.30928542591844976,
      "grad_norm": 103.14571380615234,
      "learning_rate": 1.6181153804615992e-05,
      "loss": 5.4334,
      "step": 7661
    },
    {
      "epoch": 0.3093257973354865,
      "grad_norm": 84.84107208251953,
      "learning_rate": 1.6180125879325363e-05,
      "loss": 5.0182,
      "step": 7662
    },
    {
      "epoch": 0.3093661687525232,
      "grad_norm": 91.34931182861328,
      "learning_rate": 1.6179097848369014e-05,
      "loss": 2.9492,
      "step": 7663
    },
    {
      "epoch": 0.30940654016955993,
      "grad_norm": 139.25193786621094,
      "learning_rate": 1.6178069711764522e-05,
      "loss": 5.7684,
      "step": 7664
    },
    {
      "epoch": 0.3094469115865967,
      "grad_norm": 94.29400634765625,
      "learning_rate": 1.617704146952946e-05,
      "loss": 5.8035,
      "step": 7665
    },
    {
      "epoch": 0.30948728300363343,
      "grad_norm": 97.57254791259766,
      "learning_rate": 1.617601312168142e-05,
      "loss": 6.1527,
      "step": 7666
    },
    {
      "epoch": 0.30952765442067015,
      "grad_norm": 96.34278869628906,
      "learning_rate": 1.6174984668237973e-05,
      "loss": 4.2291,
      "step": 7667
    },
    {
      "epoch": 0.30956802583770693,
      "grad_norm": 162.38653564453125,
      "learning_rate": 1.617395610921671e-05,
      "loss": 5.5969,
      "step": 7668
    },
    {
      "epoch": 0.30960839725474365,
      "grad_norm": 103.66012573242188,
      "learning_rate": 1.6172927444635217e-05,
      "loss": 5.0122,
      "step": 7669
    },
    {
      "epoch": 0.30964876867178037,
      "grad_norm": 73.32553100585938,
      "learning_rate": 1.6171898674511074e-05,
      "loss": 4.8768,
      "step": 7670
    },
    {
      "epoch": 0.3096891400888171,
      "grad_norm": 93.23023986816406,
      "learning_rate": 1.617086979886188e-05,
      "loss": 5.4728,
      "step": 7671
    },
    {
      "epoch": 0.30972951150585387,
      "grad_norm": 105.73017883300781,
      "learning_rate": 1.616984081770522e-05,
      "loss": 4.9783,
      "step": 7672
    },
    {
      "epoch": 0.3097698829228906,
      "grad_norm": 134.16824340820312,
      "learning_rate": 1.616881173105869e-05,
      "loss": 3.589,
      "step": 7673
    },
    {
      "epoch": 0.3098102543399273,
      "grad_norm": 109.34120178222656,
      "learning_rate": 1.6167782538939887e-05,
      "loss": 7.7409,
      "step": 7674
    },
    {
      "epoch": 0.3098506257569641,
      "grad_norm": 101.5480728149414,
      "learning_rate": 1.6166753241366403e-05,
      "loss": 4.3248,
      "step": 7675
    },
    {
      "epoch": 0.3098909971740008,
      "grad_norm": 145.07208251953125,
      "learning_rate": 1.616572383835584e-05,
      "loss": 4.6933,
      "step": 7676
    },
    {
      "epoch": 0.30993136859103754,
      "grad_norm": 116.51136016845703,
      "learning_rate": 1.6164694329925796e-05,
      "loss": 6.3348,
      "step": 7677
    },
    {
      "epoch": 0.30997174000807426,
      "grad_norm": 96.87189483642578,
      "learning_rate": 1.6163664716093875e-05,
      "loss": 6.4332,
      "step": 7678
    },
    {
      "epoch": 0.31001211142511104,
      "grad_norm": 117.38065338134766,
      "learning_rate": 1.6162634996877683e-05,
      "loss": 4.9307,
      "step": 7679
    },
    {
      "epoch": 0.31005248284214776,
      "grad_norm": 103.98995208740234,
      "learning_rate": 1.616160517229482e-05,
      "loss": 6.0432,
      "step": 7680
    },
    {
      "epoch": 0.3100928542591845,
      "grad_norm": 103.32666015625,
      "learning_rate": 1.6160575242362897e-05,
      "loss": 4.0615,
      "step": 7681
    },
    {
      "epoch": 0.31013322567622126,
      "grad_norm": 95.11256408691406,
      "learning_rate": 1.6159545207099525e-05,
      "loss": 4.8415,
      "step": 7682
    },
    {
      "epoch": 0.310173597093258,
      "grad_norm": 94.90554809570312,
      "learning_rate": 1.6158515066522314e-05,
      "loss": 3.9659,
      "step": 7683
    },
    {
      "epoch": 0.3102139685102947,
      "grad_norm": 104.55476379394531,
      "learning_rate": 1.6157484820648876e-05,
      "loss": 5.3437,
      "step": 7684
    },
    {
      "epoch": 0.3102543399273314,
      "grad_norm": 107.6470718383789,
      "learning_rate": 1.6156454469496825e-05,
      "loss": 5.1972,
      "step": 7685
    },
    {
      "epoch": 0.3102947113443682,
      "grad_norm": 95.35597229003906,
      "learning_rate": 1.6155424013083783e-05,
      "loss": 4.6859,
      "step": 7686
    },
    {
      "epoch": 0.3103350827614049,
      "grad_norm": 87.88623809814453,
      "learning_rate": 1.6154393451427358e-05,
      "loss": 5.3088,
      "step": 7687
    },
    {
      "epoch": 0.31037545417844165,
      "grad_norm": 182.10809326171875,
      "learning_rate": 1.615336278454518e-05,
      "loss": 5.9734,
      "step": 7688
    },
    {
      "epoch": 0.3104158255954784,
      "grad_norm": 136.92193603515625,
      "learning_rate": 1.615233201245487e-05,
      "loss": 6.1632,
      "step": 7689
    },
    {
      "epoch": 0.31045619701251514,
      "grad_norm": 89.55291748046875,
      "learning_rate": 1.6151301135174046e-05,
      "loss": 4.1707,
      "step": 7690
    },
    {
      "epoch": 0.31049656842955187,
      "grad_norm": 106.29689025878906,
      "learning_rate": 1.6150270152720337e-05,
      "loss": 5.4142,
      "step": 7691
    },
    {
      "epoch": 0.3105369398465886,
      "grad_norm": 91.08301544189453,
      "learning_rate": 1.614923906511137e-05,
      "loss": 4.4063,
      "step": 7692
    },
    {
      "epoch": 0.31057731126362537,
      "grad_norm": 136.2309112548828,
      "learning_rate": 1.6148207872364773e-05,
      "loss": 4.4616,
      "step": 7693
    },
    {
      "epoch": 0.3106176826806621,
      "grad_norm": 126.43840789794922,
      "learning_rate": 1.6147176574498183e-05,
      "loss": 5.5079,
      "step": 7694
    },
    {
      "epoch": 0.3106580540976988,
      "grad_norm": 132.2465362548828,
      "learning_rate": 1.6146145171529228e-05,
      "loss": 5.9348,
      "step": 7695
    },
    {
      "epoch": 0.3106984255147356,
      "grad_norm": 71.8799819946289,
      "learning_rate": 1.614511366347554e-05,
      "loss": 6.1304,
      "step": 7696
    },
    {
      "epoch": 0.3107387969317723,
      "grad_norm": 119.12367248535156,
      "learning_rate": 1.6144082050354758e-05,
      "loss": 5.5822,
      "step": 7697
    },
    {
      "epoch": 0.31077916834880903,
      "grad_norm": 208.91387939453125,
      "learning_rate": 1.6143050332184524e-05,
      "loss": 6.2539,
      "step": 7698
    },
    {
      "epoch": 0.31081953976584575,
      "grad_norm": 3814.50537109375,
      "learning_rate": 1.6142018508982472e-05,
      "loss": 3.9527,
      "step": 7699
    },
    {
      "epoch": 0.31085991118288253,
      "grad_norm": 99.55293273925781,
      "learning_rate": 1.6140986580766248e-05,
      "loss": 4.7114,
      "step": 7700
    },
    {
      "epoch": 0.31090028259991925,
      "grad_norm": 112.38621520996094,
      "learning_rate": 1.613995454755349e-05,
      "loss": 4.5043,
      "step": 7701
    },
    {
      "epoch": 0.310940654016956,
      "grad_norm": 85.4827880859375,
      "learning_rate": 1.6138922409361847e-05,
      "loss": 5.3973,
      "step": 7702
    },
    {
      "epoch": 0.31098102543399275,
      "grad_norm": 90.2845687866211,
      "learning_rate": 1.613789016620897e-05,
      "loss": 6.2184,
      "step": 7703
    },
    {
      "epoch": 0.3110213968510295,
      "grad_norm": 115.5992660522461,
      "learning_rate": 1.6136857818112502e-05,
      "loss": 6.5128,
      "step": 7704
    },
    {
      "epoch": 0.3110617682680662,
      "grad_norm": 145.56292724609375,
      "learning_rate": 1.6135825365090093e-05,
      "loss": 3.7804,
      "step": 7705
    },
    {
      "epoch": 0.311102139685103,
      "grad_norm": 107.14926147460938,
      "learning_rate": 1.61347928071594e-05,
      "loss": 6.8485,
      "step": 7706
    },
    {
      "epoch": 0.3111425111021397,
      "grad_norm": 75.59481048583984,
      "learning_rate": 1.6133760144338082e-05,
      "loss": 6.3744,
      "step": 7707
    },
    {
      "epoch": 0.3111828825191764,
      "grad_norm": 84.02385711669922,
      "learning_rate": 1.6132727376643784e-05,
      "loss": 6.2814,
      "step": 7708
    },
    {
      "epoch": 0.31122325393621314,
      "grad_norm": 77.6731185913086,
      "learning_rate": 1.6131694504094166e-05,
      "loss": 3.6117,
      "step": 7709
    },
    {
      "epoch": 0.3112636253532499,
      "grad_norm": 84.41787719726562,
      "learning_rate": 1.6130661526706897e-05,
      "loss": 4.9247,
      "step": 7710
    },
    {
      "epoch": 0.31130399677028664,
      "grad_norm": 115.4563217163086,
      "learning_rate": 1.6129628444499627e-05,
      "loss": 5.7399,
      "step": 7711
    },
    {
      "epoch": 0.31134436818732336,
      "grad_norm": 111.26152801513672,
      "learning_rate": 1.6128595257490026e-05,
      "loss": 5.3484,
      "step": 7712
    },
    {
      "epoch": 0.31138473960436014,
      "grad_norm": 79.3482437133789,
      "learning_rate": 1.6127561965695756e-05,
      "loss": 5.2204,
      "step": 7713
    },
    {
      "epoch": 0.31142511102139686,
      "grad_norm": 111.97703552246094,
      "learning_rate": 1.612652856913449e-05,
      "loss": 5.0657,
      "step": 7714
    },
    {
      "epoch": 0.3114654824384336,
      "grad_norm": 110.401611328125,
      "learning_rate": 1.6125495067823886e-05,
      "loss": 3.5502,
      "step": 7715
    },
    {
      "epoch": 0.3115058538554703,
      "grad_norm": 74.73388671875,
      "learning_rate": 1.6124461461781625e-05,
      "loss": 5.6078,
      "step": 7716
    },
    {
      "epoch": 0.3115462252725071,
      "grad_norm": 99.39370727539062,
      "learning_rate": 1.612342775102537e-05,
      "loss": 7.204,
      "step": 7717
    },
    {
      "epoch": 0.3115865966895438,
      "grad_norm": 101.35757446289062,
      "learning_rate": 1.6122393935572804e-05,
      "loss": 5.5571,
      "step": 7718
    },
    {
      "epoch": 0.3116269681065805,
      "grad_norm": 98.33853912353516,
      "learning_rate": 1.6121360015441595e-05,
      "loss": 6.0443,
      "step": 7719
    },
    {
      "epoch": 0.3116673395236173,
      "grad_norm": 77.03980255126953,
      "learning_rate": 1.6120325990649427e-05,
      "loss": 5.9237,
      "step": 7720
    },
    {
      "epoch": 0.311707710940654,
      "grad_norm": 81.47386169433594,
      "learning_rate": 1.6119291861213975e-05,
      "loss": 5.6714,
      "step": 7721
    },
    {
      "epoch": 0.31174808235769075,
      "grad_norm": 165.4454803466797,
      "learning_rate": 1.6118257627152925e-05,
      "loss": 6.9357,
      "step": 7722
    },
    {
      "epoch": 0.31178845377472747,
      "grad_norm": 120.24334716796875,
      "learning_rate": 1.6117223288483953e-05,
      "loss": 4.499,
      "step": 7723
    },
    {
      "epoch": 0.31182882519176425,
      "grad_norm": 78.8615493774414,
      "learning_rate": 1.611618884522475e-05,
      "loss": 4.3568,
      "step": 7724
    },
    {
      "epoch": 0.31186919660880097,
      "grad_norm": 86.28028106689453,
      "learning_rate": 1.6115154297392995e-05,
      "loss": 5.9167,
      "step": 7725
    },
    {
      "epoch": 0.3119095680258377,
      "grad_norm": 124.70448303222656,
      "learning_rate": 1.6114119645006385e-05,
      "loss": 4.6482,
      "step": 7726
    },
    {
      "epoch": 0.31194993944287447,
      "grad_norm": 66.73204803466797,
      "learning_rate": 1.6113084888082606e-05,
      "loss": 5.949,
      "step": 7727
    },
    {
      "epoch": 0.3119903108599112,
      "grad_norm": 98.5573959350586,
      "learning_rate": 1.6112050026639353e-05,
      "loss": 4.8742,
      "step": 7728
    },
    {
      "epoch": 0.3120306822769479,
      "grad_norm": 81.86151123046875,
      "learning_rate": 1.6111015060694315e-05,
      "loss": 6.131,
      "step": 7729
    },
    {
      "epoch": 0.31207105369398463,
      "grad_norm": 81.41410827636719,
      "learning_rate": 1.610997999026519e-05,
      "loss": 5.4033,
      "step": 7730
    },
    {
      "epoch": 0.3121114251110214,
      "grad_norm": 101.2059097290039,
      "learning_rate": 1.6108944815369674e-05,
      "loss": 3.3866,
      "step": 7731
    },
    {
      "epoch": 0.31215179652805813,
      "grad_norm": 225.0520782470703,
      "learning_rate": 1.610790953602547e-05,
      "loss": 5.7502,
      "step": 7732
    },
    {
      "epoch": 0.31219216794509486,
      "grad_norm": 79.65052795410156,
      "learning_rate": 1.6106874152250275e-05,
      "loss": 5.9168,
      "step": 7733
    },
    {
      "epoch": 0.31223253936213163,
      "grad_norm": 130.6546630859375,
      "learning_rate": 1.610583866406179e-05,
      "loss": 6.97,
      "step": 7734
    },
    {
      "epoch": 0.31227291077916836,
      "grad_norm": 74.65611267089844,
      "learning_rate": 1.6104803071477723e-05,
      "loss": 4.6202,
      "step": 7735
    },
    {
      "epoch": 0.3123132821962051,
      "grad_norm": 71.89141845703125,
      "learning_rate": 1.610376737451578e-05,
      "loss": 4.8091,
      "step": 7736
    },
    {
      "epoch": 0.3123536536132418,
      "grad_norm": 129.86965942382812,
      "learning_rate": 1.6102731573193666e-05,
      "loss": 6.3827,
      "step": 7737
    },
    {
      "epoch": 0.3123940250302786,
      "grad_norm": 136.85281372070312,
      "learning_rate": 1.6101695667529097e-05,
      "loss": 6.6847,
      "step": 7738
    },
    {
      "epoch": 0.3124343964473153,
      "grad_norm": 98.6790771484375,
      "learning_rate": 1.610065965753978e-05,
      "loss": 6.5745,
      "step": 7739
    },
    {
      "epoch": 0.312474767864352,
      "grad_norm": 66.97420501708984,
      "learning_rate": 1.6099623543243426e-05,
      "loss": 6.655,
      "step": 7740
    },
    {
      "epoch": 0.3125151392813888,
      "grad_norm": 94.32658386230469,
      "learning_rate": 1.6098587324657754e-05,
      "loss": 5.8504,
      "step": 7741
    },
    {
      "epoch": 0.3125555106984255,
      "grad_norm": 147.6631317138672,
      "learning_rate": 1.609755100180048e-05,
      "loss": 5.7065,
      "step": 7742
    },
    {
      "epoch": 0.31259588211546224,
      "grad_norm": 79.76377868652344,
      "learning_rate": 1.6096514574689323e-05,
      "loss": 6.7161,
      "step": 7743
    },
    {
      "epoch": 0.31263625353249896,
      "grad_norm": 92.92597961425781,
      "learning_rate": 1.6095478043342003e-05,
      "loss": 5.5785,
      "step": 7744
    },
    {
      "epoch": 0.31267662494953574,
      "grad_norm": 76.46699523925781,
      "learning_rate": 1.6094441407776242e-05,
      "loss": 5.835,
      "step": 7745
    },
    {
      "epoch": 0.31271699636657246,
      "grad_norm": 86.94803619384766,
      "learning_rate": 1.6093404668009767e-05,
      "loss": 5.5201,
      "step": 7746
    },
    {
      "epoch": 0.3127573677836092,
      "grad_norm": 107.35388946533203,
      "learning_rate": 1.6092367824060297e-05,
      "loss": 5.3565,
      "step": 7747
    },
    {
      "epoch": 0.31279773920064596,
      "grad_norm": 78.11737060546875,
      "learning_rate": 1.609133087594557e-05,
      "loss": 5.659,
      "step": 7748
    },
    {
      "epoch": 0.3128381106176827,
      "grad_norm": 143.5332489013672,
      "learning_rate": 1.6090293823683307e-05,
      "loss": 5.1104,
      "step": 7749
    },
    {
      "epoch": 0.3128784820347194,
      "grad_norm": 139.6876678466797,
      "learning_rate": 1.6089256667291236e-05,
      "loss": 5.7552,
      "step": 7750
    },
    {
      "epoch": 0.31291885345175613,
      "grad_norm": 71.0224609375,
      "learning_rate": 1.60882194067871e-05,
      "loss": 5.0027,
      "step": 7751
    },
    {
      "epoch": 0.3129592248687929,
      "grad_norm": 102.94406127929688,
      "learning_rate": 1.608718204218863e-05,
      "loss": 4.2589,
      "step": 7752
    },
    {
      "epoch": 0.31299959628582963,
      "grad_norm": 91.41764068603516,
      "learning_rate": 1.608614457351356e-05,
      "loss": 4.5711,
      "step": 7753
    },
    {
      "epoch": 0.31303996770286635,
      "grad_norm": 129.69143676757812,
      "learning_rate": 1.608510700077963e-05,
      "loss": 4.8951,
      "step": 7754
    },
    {
      "epoch": 0.31308033911990313,
      "grad_norm": 73.25420379638672,
      "learning_rate": 1.6084069324004582e-05,
      "loss": 6.4943,
      "step": 7755
    },
    {
      "epoch": 0.31312071053693985,
      "grad_norm": 99.51688385009766,
      "learning_rate": 1.6083031543206152e-05,
      "loss": 5.1125,
      "step": 7756
    },
    {
      "epoch": 0.31316108195397657,
      "grad_norm": 144.65704345703125,
      "learning_rate": 1.6081993658402088e-05,
      "loss": 5.6245,
      "step": 7757
    },
    {
      "epoch": 0.31320145337101335,
      "grad_norm": 122.58684539794922,
      "learning_rate": 1.6080955669610137e-05,
      "loss": 5.6293,
      "step": 7758
    },
    {
      "epoch": 0.31324182478805007,
      "grad_norm": 117.15467834472656,
      "learning_rate": 1.6079917576848046e-05,
      "loss": 5.4039,
      "step": 7759
    },
    {
      "epoch": 0.3132821962050868,
      "grad_norm": 119.48371887207031,
      "learning_rate": 1.607887938013356e-05,
      "loss": 3.4844,
      "step": 7760
    },
    {
      "epoch": 0.3133225676221235,
      "grad_norm": 104.66513061523438,
      "learning_rate": 1.6077841079484433e-05,
      "loss": 6.0883,
      "step": 7761
    },
    {
      "epoch": 0.3133629390391603,
      "grad_norm": 98.15169525146484,
      "learning_rate": 1.6076802674918414e-05,
      "loss": 3.3818,
      "step": 7762
    },
    {
      "epoch": 0.313403310456197,
      "grad_norm": 91.14521026611328,
      "learning_rate": 1.607576416645326e-05,
      "loss": 5.836,
      "step": 7763
    },
    {
      "epoch": 0.31344368187323374,
      "grad_norm": 123.10379791259766,
      "learning_rate": 1.6074725554106727e-05,
      "loss": 5.719,
      "step": 7764
    },
    {
      "epoch": 0.3134840532902705,
      "grad_norm": 91.18404388427734,
      "learning_rate": 1.6073686837896575e-05,
      "loss": 4.1909,
      "step": 7765
    },
    {
      "epoch": 0.31352442470730724,
      "grad_norm": 124.75432586669922,
      "learning_rate": 1.6072648017840557e-05,
      "loss": 6.1442,
      "step": 7766
    },
    {
      "epoch": 0.31356479612434396,
      "grad_norm": 117.93550109863281,
      "learning_rate": 1.607160909395644e-05,
      "loss": 5.1235,
      "step": 7767
    },
    {
      "epoch": 0.3136051675413807,
      "grad_norm": 100.01268768310547,
      "learning_rate": 1.6070570066261988e-05,
      "loss": 4.3669,
      "step": 7768
    },
    {
      "epoch": 0.31364553895841746,
      "grad_norm": 146.58985900878906,
      "learning_rate": 1.606953093477496e-05,
      "loss": 5.0546,
      "step": 7769
    },
    {
      "epoch": 0.3136859103754542,
      "grad_norm": 83.66029357910156,
      "learning_rate": 1.6068491699513133e-05,
      "loss": 5.4365,
      "step": 7770
    },
    {
      "epoch": 0.3137262817924909,
      "grad_norm": 117.13542938232422,
      "learning_rate": 1.606745236049426e-05,
      "loss": 4.6871,
      "step": 7771
    },
    {
      "epoch": 0.3137666532095277,
      "grad_norm": 134.2798309326172,
      "learning_rate": 1.6066412917736128e-05,
      "loss": 4.325,
      "step": 7772
    },
    {
      "epoch": 0.3138070246265644,
      "grad_norm": 85.034912109375,
      "learning_rate": 1.6065373371256496e-05,
      "loss": 7.1395,
      "step": 7773
    },
    {
      "epoch": 0.3138473960436011,
      "grad_norm": 82.31013488769531,
      "learning_rate": 1.6064333721073146e-05,
      "loss": 4.7073,
      "step": 7774
    },
    {
      "epoch": 0.31388776746063785,
      "grad_norm": 161.088134765625,
      "learning_rate": 1.606329396720385e-05,
      "loss": 6.2138,
      "step": 7775
    },
    {
      "epoch": 0.3139281388776746,
      "grad_norm": 119.26899719238281,
      "learning_rate": 1.6062254109666383e-05,
      "loss": 4.9963,
      "step": 7776
    },
    {
      "epoch": 0.31396851029471134,
      "grad_norm": 111.47457122802734,
      "learning_rate": 1.606121414847853e-05,
      "loss": 4.6661,
      "step": 7777
    },
    {
      "epoch": 0.31400888171174807,
      "grad_norm": 124.03829956054688,
      "learning_rate": 1.6060174083658068e-05,
      "loss": 5.0691,
      "step": 7778
    },
    {
      "epoch": 0.31404925312878484,
      "grad_norm": 107.07508087158203,
      "learning_rate": 1.605913391522278e-05,
      "loss": 4.9796,
      "step": 7779
    },
    {
      "epoch": 0.31408962454582157,
      "grad_norm": 157.21725463867188,
      "learning_rate": 1.6058093643190452e-05,
      "loss": 4.3138,
      "step": 7780
    },
    {
      "epoch": 0.3141299959628583,
      "grad_norm": 88.0505142211914,
      "learning_rate": 1.6057053267578867e-05,
      "loss": 3.9119,
      "step": 7781
    },
    {
      "epoch": 0.314170367379895,
      "grad_norm": 82.70779418945312,
      "learning_rate": 1.605601278840582e-05,
      "loss": 4.8934,
      "step": 7782
    },
    {
      "epoch": 0.3142107387969318,
      "grad_norm": 79.5171127319336,
      "learning_rate": 1.605497220568909e-05,
      "loss": 5.2714,
      "step": 7783
    },
    {
      "epoch": 0.3142511102139685,
      "grad_norm": 122.5922622680664,
      "learning_rate": 1.6053931519446482e-05,
      "loss": 4.9594,
      "step": 7784
    },
    {
      "epoch": 0.31429148163100523,
      "grad_norm": 93.8082275390625,
      "learning_rate": 1.605289072969578e-05,
      "loss": 5.9716,
      "step": 7785
    },
    {
      "epoch": 0.314331853048042,
      "grad_norm": 213.34420776367188,
      "learning_rate": 1.605184983645478e-05,
      "loss": 5.0903,
      "step": 7786
    },
    {
      "epoch": 0.31437222446507873,
      "grad_norm": 96.85710144042969,
      "learning_rate": 1.605080883974128e-05,
      "loss": 6.0266,
      "step": 7787
    },
    {
      "epoch": 0.31441259588211545,
      "grad_norm": 84.71764373779297,
      "learning_rate": 1.6049767739573075e-05,
      "loss": 6.1031,
      "step": 7788
    },
    {
      "epoch": 0.3144529672991522,
      "grad_norm": 66.25775146484375,
      "learning_rate": 1.6048726535967974e-05,
      "loss": 3.5743,
      "step": 7789
    },
    {
      "epoch": 0.31449333871618895,
      "grad_norm": 238.9482879638672,
      "learning_rate": 1.604768522894377e-05,
      "loss": 5.4113,
      "step": 7790
    },
    {
      "epoch": 0.3145337101332257,
      "grad_norm": 73.69754791259766,
      "learning_rate": 1.604664381851827e-05,
      "loss": 4.4892,
      "step": 7791
    },
    {
      "epoch": 0.3145740815502624,
      "grad_norm": 97.68855285644531,
      "learning_rate": 1.6045602304709288e-05,
      "loss": 5.4624,
      "step": 7792
    },
    {
      "epoch": 0.3146144529672992,
      "grad_norm": 96.74098205566406,
      "learning_rate": 1.6044560687534617e-05,
      "loss": 4.7148,
      "step": 7793
    },
    {
      "epoch": 0.3146548243843359,
      "grad_norm": 101.65998077392578,
      "learning_rate": 1.604351896701208e-05,
      "loss": 5.794,
      "step": 7794
    },
    {
      "epoch": 0.3146951958013726,
      "grad_norm": 117.10785675048828,
      "learning_rate": 1.6042477143159473e-05,
      "loss": 6.3927,
      "step": 7795
    },
    {
      "epoch": 0.31473556721840934,
      "grad_norm": 313.8560791015625,
      "learning_rate": 1.6041435215994622e-05,
      "loss": 5.1159,
      "step": 7796
    },
    {
      "epoch": 0.3147759386354461,
      "grad_norm": 100.08763122558594,
      "learning_rate": 1.6040393185535337e-05,
      "loss": 4.3841,
      "step": 7797
    },
    {
      "epoch": 0.31481631005248284,
      "grad_norm": 121.203857421875,
      "learning_rate": 1.6039351051799433e-05,
      "loss": 5.4087,
      "step": 7798
    },
    {
      "epoch": 0.31485668146951956,
      "grad_norm": 94.55425262451172,
      "learning_rate": 1.603830881480473e-05,
      "loss": 5.8498,
      "step": 7799
    },
    {
      "epoch": 0.31489705288655634,
      "grad_norm": 107.24827575683594,
      "learning_rate": 1.6037266474569043e-05,
      "loss": 4.0315,
      "step": 7800
    },
    {
      "epoch": 0.31493742430359306,
      "grad_norm": 94.83979034423828,
      "learning_rate": 1.60362240311102e-05,
      "loss": 3.6461,
      "step": 7801
    },
    {
      "epoch": 0.3149777957206298,
      "grad_norm": 72.162109375,
      "learning_rate": 1.6035181484446023e-05,
      "loss": 6.4748,
      "step": 7802
    },
    {
      "epoch": 0.3150181671376665,
      "grad_norm": 98.2156753540039,
      "learning_rate": 1.603413883459433e-05,
      "loss": 7.0803,
      "step": 7803
    },
    {
      "epoch": 0.3150585385547033,
      "grad_norm": 135.3854522705078,
      "learning_rate": 1.6033096081572962e-05,
      "loss": 5.3902,
      "step": 7804
    },
    {
      "epoch": 0.31509890997174,
      "grad_norm": 71.39429473876953,
      "learning_rate": 1.6032053225399733e-05,
      "loss": 6.039,
      "step": 7805
    },
    {
      "epoch": 0.3151392813887767,
      "grad_norm": 93.83122253417969,
      "learning_rate": 1.603101026609248e-05,
      "loss": 4.67,
      "step": 7806
    },
    {
      "epoch": 0.3151796528058135,
      "grad_norm": 106.33531951904297,
      "learning_rate": 1.6029967203669035e-05,
      "loss": 5.3981,
      "step": 7807
    },
    {
      "epoch": 0.3152200242228502,
      "grad_norm": 65.70875549316406,
      "learning_rate": 1.6028924038147237e-05,
      "loss": 6.5318,
      "step": 7808
    },
    {
      "epoch": 0.31526039563988695,
      "grad_norm": 126.33000946044922,
      "learning_rate": 1.602788076954491e-05,
      "loss": 5.3841,
      "step": 7809
    },
    {
      "epoch": 0.3153007670569237,
      "grad_norm": 182.02017211914062,
      "learning_rate": 1.6026837397879905e-05,
      "loss": 4.9369,
      "step": 7810
    },
    {
      "epoch": 0.31534113847396045,
      "grad_norm": 77.74185180664062,
      "learning_rate": 1.6025793923170048e-05,
      "loss": 5.8962,
      "step": 7811
    },
    {
      "epoch": 0.31538150989099717,
      "grad_norm": 93.11988067626953,
      "learning_rate": 1.6024750345433188e-05,
      "loss": 6.1269,
      "step": 7812
    },
    {
      "epoch": 0.3154218813080339,
      "grad_norm": 179.87013244628906,
      "learning_rate": 1.6023706664687167e-05,
      "loss": 4.8256,
      "step": 7813
    },
    {
      "epoch": 0.31546225272507067,
      "grad_norm": 91.04800415039062,
      "learning_rate": 1.6022662880949828e-05,
      "loss": 5.7948,
      "step": 7814
    },
    {
      "epoch": 0.3155026241421074,
      "grad_norm": 190.93739318847656,
      "learning_rate": 1.6021618994239015e-05,
      "loss": 4.9567,
      "step": 7815
    },
    {
      "epoch": 0.3155429955591441,
      "grad_norm": 99.28095245361328,
      "learning_rate": 1.6020575004572584e-05,
      "loss": 4.1479,
      "step": 7816
    },
    {
      "epoch": 0.3155833669761809,
      "grad_norm": 88.69280242919922,
      "learning_rate": 1.6019530911968373e-05,
      "loss": 5.3607,
      "step": 7817
    },
    {
      "epoch": 0.3156237383932176,
      "grad_norm": 108.35737609863281,
      "learning_rate": 1.6018486716444246e-05,
      "loss": 5.3131,
      "step": 7818
    },
    {
      "epoch": 0.31566410981025433,
      "grad_norm": 117.06791687011719,
      "learning_rate": 1.6017442418018046e-05,
      "loss": 4.8156,
      "step": 7819
    },
    {
      "epoch": 0.31570448122729106,
      "grad_norm": 91.69773864746094,
      "learning_rate": 1.6016398016707636e-05,
      "loss": 6.0054,
      "step": 7820
    },
    {
      "epoch": 0.31574485264432783,
      "grad_norm": 91.48892974853516,
      "learning_rate": 1.6015353512530865e-05,
      "loss": 5.4842,
      "step": 7821
    },
    {
      "epoch": 0.31578522406136456,
      "grad_norm": 127.64018249511719,
      "learning_rate": 1.60143089055056e-05,
      "loss": 6.7623,
      "step": 7822
    },
    {
      "epoch": 0.3158255954784013,
      "grad_norm": 82.30572509765625,
      "learning_rate": 1.6013264195649692e-05,
      "loss": 7.1203,
      "step": 7823
    },
    {
      "epoch": 0.31586596689543806,
      "grad_norm": 69.43525695800781,
      "learning_rate": 1.6012219382981015e-05,
      "loss": 4.9388,
      "step": 7824
    },
    {
      "epoch": 0.3159063383124748,
      "grad_norm": 77.36106872558594,
      "learning_rate": 1.601117446751742e-05,
      "loss": 5.1921,
      "step": 7825
    },
    {
      "epoch": 0.3159467097295115,
      "grad_norm": 122.53765869140625,
      "learning_rate": 1.6010129449276782e-05,
      "loss": 6.1088,
      "step": 7826
    },
    {
      "epoch": 0.3159870811465482,
      "grad_norm": 74.99198150634766,
      "learning_rate": 1.6009084328276965e-05,
      "loss": 6.4026,
      "step": 7827
    },
    {
      "epoch": 0.316027452563585,
      "grad_norm": 90.84821319580078,
      "learning_rate": 1.6008039104535836e-05,
      "loss": 6.1578,
      "step": 7828
    },
    {
      "epoch": 0.3160678239806217,
      "grad_norm": 111.06403350830078,
      "learning_rate": 1.600699377807127e-05,
      "loss": 6.3293,
      "step": 7829
    },
    {
      "epoch": 0.31610819539765844,
      "grad_norm": 84.87445068359375,
      "learning_rate": 1.6005948348901138e-05,
      "loss": 5.6105,
      "step": 7830
    },
    {
      "epoch": 0.3161485668146952,
      "grad_norm": 158.31272888183594,
      "learning_rate": 1.6004902817043314e-05,
      "loss": 5.3291,
      "step": 7831
    },
    {
      "epoch": 0.31618893823173194,
      "grad_norm": 133.08262634277344,
      "learning_rate": 1.6003857182515675e-05,
      "loss": 5.7987,
      "step": 7832
    },
    {
      "epoch": 0.31622930964876866,
      "grad_norm": 120.4376220703125,
      "learning_rate": 1.6002811445336095e-05,
      "loss": 5.4464,
      "step": 7833
    },
    {
      "epoch": 0.3162696810658054,
      "grad_norm": 157.9675750732422,
      "learning_rate": 1.600176560552246e-05,
      "loss": 5.2928,
      "step": 7834
    },
    {
      "epoch": 0.31631005248284216,
      "grad_norm": 73.41561889648438,
      "learning_rate": 1.6000719663092644e-05,
      "loss": 5.3176,
      "step": 7835
    },
    {
      "epoch": 0.3163504238998789,
      "grad_norm": 154.31607055664062,
      "learning_rate": 1.599967361806454e-05,
      "loss": 6.8257,
      "step": 7836
    },
    {
      "epoch": 0.3163907953169156,
      "grad_norm": 86.13622283935547,
      "learning_rate": 1.5998627470456024e-05,
      "loss": 4.975,
      "step": 7837
    },
    {
      "epoch": 0.3164311667339524,
      "grad_norm": 88.1004638671875,
      "learning_rate": 1.5997581220284988e-05,
      "loss": 3.5612,
      "step": 7838
    },
    {
      "epoch": 0.3164715381509891,
      "grad_norm": 73.73098754882812,
      "learning_rate": 1.5996534867569317e-05,
      "loss": 5.7176,
      "step": 7839
    },
    {
      "epoch": 0.31651190956802583,
      "grad_norm": 125.89739227294922,
      "learning_rate": 1.5995488412326904e-05,
      "loss": 5.5725,
      "step": 7840
    },
    {
      "epoch": 0.31655228098506255,
      "grad_norm": 79.36572265625,
      "learning_rate": 1.5994441854575636e-05,
      "loss": 4.4676,
      "step": 7841
    },
    {
      "epoch": 0.31659265240209933,
      "grad_norm": 75.07476806640625,
      "learning_rate": 1.5993395194333415e-05,
      "loss": 5.0507,
      "step": 7842
    },
    {
      "epoch": 0.31663302381913605,
      "grad_norm": 78.3328628540039,
      "learning_rate": 1.599234843161813e-05,
      "loss": 4.5384,
      "step": 7843
    },
    {
      "epoch": 0.31667339523617277,
      "grad_norm": 105.29019927978516,
      "learning_rate": 1.599130156644768e-05,
      "loss": 5.6304,
      "step": 7844
    },
    {
      "epoch": 0.31671376665320955,
      "grad_norm": 125.60363006591797,
      "learning_rate": 1.5990254598839963e-05,
      "loss": 4.641,
      "step": 7845
    },
    {
      "epoch": 0.31675413807024627,
      "grad_norm": 120.21719360351562,
      "learning_rate": 1.5989207528812883e-05,
      "loss": 5.344,
      "step": 7846
    },
    {
      "epoch": 0.316794509487283,
      "grad_norm": 93.9229965209961,
      "learning_rate": 1.5988160356384333e-05,
      "loss": 6.4207,
      "step": 7847
    },
    {
      "epoch": 0.3168348809043197,
      "grad_norm": 226.57640075683594,
      "learning_rate": 1.5987113081572234e-05,
      "loss": 4.4105,
      "step": 7848
    },
    {
      "epoch": 0.3168752523213565,
      "grad_norm": 121.94873046875,
      "learning_rate": 1.5986065704394476e-05,
      "loss": 4.7292,
      "step": 7849
    },
    {
      "epoch": 0.3169156237383932,
      "grad_norm": 132.71975708007812,
      "learning_rate": 1.5985018224868975e-05,
      "loss": 5.7043,
      "step": 7850
    },
    {
      "epoch": 0.31695599515542994,
      "grad_norm": 105.41967010498047,
      "learning_rate": 1.5983970643013636e-05,
      "loss": 5.2049,
      "step": 7851
    },
    {
      "epoch": 0.3169963665724667,
      "grad_norm": 81.616455078125,
      "learning_rate": 1.5982922958846378e-05,
      "loss": 6.1961,
      "step": 7852
    },
    {
      "epoch": 0.31703673798950344,
      "grad_norm": 90.82918548583984,
      "learning_rate": 1.5981875172385105e-05,
      "loss": 4.3201,
      "step": 7853
    },
    {
      "epoch": 0.31707710940654016,
      "grad_norm": 99.55569458007812,
      "learning_rate": 1.5980827283647734e-05,
      "loss": 4.9309,
      "step": 7854
    },
    {
      "epoch": 0.3171174808235769,
      "grad_norm": 83.19175720214844,
      "learning_rate": 1.5979779292652185e-05,
      "loss": 5.6118,
      "step": 7855
    },
    {
      "epoch": 0.31715785224061366,
      "grad_norm": 89.62377166748047,
      "learning_rate": 1.5978731199416375e-05,
      "loss": 5.1965,
      "step": 7856
    },
    {
      "epoch": 0.3171982236576504,
      "grad_norm": 117.93968200683594,
      "learning_rate": 1.597768300395822e-05,
      "loss": 5.3933,
      "step": 7857
    },
    {
      "epoch": 0.3172385950746871,
      "grad_norm": 237.54685974121094,
      "learning_rate": 1.5976634706295647e-05,
      "loss": 4.157,
      "step": 7858
    },
    {
      "epoch": 0.3172789664917239,
      "grad_norm": 111.29057312011719,
      "learning_rate": 1.5975586306446574e-05,
      "loss": 5.088,
      "step": 7859
    },
    {
      "epoch": 0.3173193379087606,
      "grad_norm": 86.82386016845703,
      "learning_rate": 1.597453780442893e-05,
      "loss": 5.7698,
      "step": 7860
    },
    {
      "epoch": 0.3173597093257973,
      "grad_norm": 107.61360931396484,
      "learning_rate": 1.597348920026064e-05,
      "loss": 3.8881,
      "step": 7861
    },
    {
      "epoch": 0.3174000807428341,
      "grad_norm": 67.12628173828125,
      "learning_rate": 1.597244049395964e-05,
      "loss": 4.696,
      "step": 7862
    },
    {
      "epoch": 0.3174404521598708,
      "grad_norm": 527.9368286132812,
      "learning_rate": 1.5971391685543845e-05,
      "loss": 5.6121,
      "step": 7863
    },
    {
      "epoch": 0.31748082357690754,
      "grad_norm": 78.57280731201172,
      "learning_rate": 1.5970342775031203e-05,
      "loss": 3.4467,
      "step": 7864
    },
    {
      "epoch": 0.31752119499394427,
      "grad_norm": 89.29187774658203,
      "learning_rate": 1.596929376243964e-05,
      "loss": 4.7372,
      "step": 7865
    },
    {
      "epoch": 0.31756156641098104,
      "grad_norm": 162.37855529785156,
      "learning_rate": 1.5968244647787092e-05,
      "loss": 4.3262,
      "step": 7866
    },
    {
      "epoch": 0.31760193782801777,
      "grad_norm": 82.20704650878906,
      "learning_rate": 1.5967195431091494e-05,
      "loss": 5.083,
      "step": 7867
    },
    {
      "epoch": 0.3176423092450545,
      "grad_norm": 212.9345245361328,
      "learning_rate": 1.5966146112370794e-05,
      "loss": 4.0988,
      "step": 7868
    },
    {
      "epoch": 0.31768268066209127,
      "grad_norm": 94.03981018066406,
      "learning_rate": 1.5965096691642927e-05,
      "loss": 6.3135,
      "step": 7869
    },
    {
      "epoch": 0.317723052079128,
      "grad_norm": 165.0602264404297,
      "learning_rate": 1.5964047168925834e-05,
      "loss": 3.9185,
      "step": 7870
    },
    {
      "epoch": 0.3177634234961647,
      "grad_norm": 80.1162338256836,
      "learning_rate": 1.596299754423746e-05,
      "loss": 3.5515,
      "step": 7871
    },
    {
      "epoch": 0.31780379491320143,
      "grad_norm": 109.46804809570312,
      "learning_rate": 1.5961947817595756e-05,
      "loss": 6.5166,
      "step": 7872
    },
    {
      "epoch": 0.3178441663302382,
      "grad_norm": 74.33888244628906,
      "learning_rate": 1.5960897989018668e-05,
      "loss": 5.144,
      "step": 7873
    },
    {
      "epoch": 0.31788453774727493,
      "grad_norm": 75.08055877685547,
      "learning_rate": 1.595984805852414e-05,
      "loss": 5.746,
      "step": 7874
    },
    {
      "epoch": 0.31792490916431165,
      "grad_norm": 107.36483764648438,
      "learning_rate": 1.5958798026130128e-05,
      "loss": 6.7306,
      "step": 7875
    },
    {
      "epoch": 0.31796528058134843,
      "grad_norm": 91.80753326416016,
      "learning_rate": 1.5957747891854583e-05,
      "loss": 5.4002,
      "step": 7876
    },
    {
      "epoch": 0.31800565199838515,
      "grad_norm": 141.87234497070312,
      "learning_rate": 1.5956697655715466e-05,
      "loss": 4.6998,
      "step": 7877
    },
    {
      "epoch": 0.3180460234154219,
      "grad_norm": 104.05286407470703,
      "learning_rate": 1.5955647317730727e-05,
      "loss": 3.8901,
      "step": 7878
    },
    {
      "epoch": 0.3180863948324586,
      "grad_norm": 142.64479064941406,
      "learning_rate": 1.5954596877918324e-05,
      "loss": 5.2209,
      "step": 7879
    },
    {
      "epoch": 0.3181267662494954,
      "grad_norm": 82.4615249633789,
      "learning_rate": 1.5953546336296225e-05,
      "loss": 3.9768,
      "step": 7880
    },
    {
      "epoch": 0.3181671376665321,
      "grad_norm": 123.35604858398438,
      "learning_rate": 1.595249569288238e-05,
      "loss": 4.5437,
      "step": 7881
    },
    {
      "epoch": 0.3182075090835688,
      "grad_norm": 99.86046600341797,
      "learning_rate": 1.5951444947694764e-05,
      "loss": 4.3349,
      "step": 7882
    },
    {
      "epoch": 0.3182478805006056,
      "grad_norm": 213.1149139404297,
      "learning_rate": 1.5950394100751333e-05,
      "loss": 6.3806,
      "step": 7883
    },
    {
      "epoch": 0.3182882519176423,
      "grad_norm": 191.18133544921875,
      "learning_rate": 1.594934315207006e-05,
      "loss": 5.9509,
      "step": 7884
    },
    {
      "epoch": 0.31832862333467904,
      "grad_norm": 91.93168640136719,
      "learning_rate": 1.594829210166891e-05,
      "loss": 4.9124,
      "step": 7885
    },
    {
      "epoch": 0.31836899475171576,
      "grad_norm": 88.55303955078125,
      "learning_rate": 1.5947240949565857e-05,
      "loss": 5.5759,
      "step": 7886
    },
    {
      "epoch": 0.31840936616875254,
      "grad_norm": 116.46065521240234,
      "learning_rate": 1.594618969577887e-05,
      "loss": 3.9885,
      "step": 7887
    },
    {
      "epoch": 0.31844973758578926,
      "grad_norm": 97.10783386230469,
      "learning_rate": 1.5945138340325924e-05,
      "loss": 5.0348,
      "step": 7888
    },
    {
      "epoch": 0.318490109002826,
      "grad_norm": 92.77388763427734,
      "learning_rate": 1.5944086883224995e-05,
      "loss": 3.3613,
      "step": 7889
    },
    {
      "epoch": 0.31853048041986276,
      "grad_norm": 93.28356170654297,
      "learning_rate": 1.5943035324494065e-05,
      "loss": 6.2324,
      "step": 7890
    },
    {
      "epoch": 0.3185708518368995,
      "grad_norm": 103.19768524169922,
      "learning_rate": 1.5941983664151108e-05,
      "loss": 4.8056,
      "step": 7891
    },
    {
      "epoch": 0.3186112232539362,
      "grad_norm": 75.24077606201172,
      "learning_rate": 1.5940931902214105e-05,
      "loss": 4.523,
      "step": 7892
    },
    {
      "epoch": 0.3186515946709729,
      "grad_norm": 108.9754638671875,
      "learning_rate": 1.593988003870104e-05,
      "loss": 7.2323,
      "step": 7893
    },
    {
      "epoch": 0.3186919660880097,
      "grad_norm": 97.40734100341797,
      "learning_rate": 1.593882807362989e-05,
      "loss": 5.0791,
      "step": 7894
    },
    {
      "epoch": 0.3187323375050464,
      "grad_norm": 66.80554962158203,
      "learning_rate": 1.5937776007018657e-05,
      "loss": 5.4108,
      "step": 7895
    },
    {
      "epoch": 0.31877270892208315,
      "grad_norm": 138.0582733154297,
      "learning_rate": 1.593672383888532e-05,
      "loss": 8.1947,
      "step": 7896
    },
    {
      "epoch": 0.3188130803391199,
      "grad_norm": 116.61229705810547,
      "learning_rate": 1.5935671569247863e-05,
      "loss": 1.9171,
      "step": 7897
    },
    {
      "epoch": 0.31885345175615665,
      "grad_norm": 103.1606216430664,
      "learning_rate": 1.5934619198124286e-05,
      "loss": 3.3182,
      "step": 7898
    },
    {
      "epoch": 0.31889382317319337,
      "grad_norm": 89.14549255371094,
      "learning_rate": 1.5933566725532576e-05,
      "loss": 6.4742,
      "step": 7899
    },
    {
      "epoch": 0.3189341945902301,
      "grad_norm": 117.43355560302734,
      "learning_rate": 1.5932514151490737e-05,
      "loss": 4.1532,
      "step": 7900
    },
    {
      "epoch": 0.31897456600726687,
      "grad_norm": 88.72991180419922,
      "learning_rate": 1.5931461476016753e-05,
      "loss": 5.406,
      "step": 7901
    },
    {
      "epoch": 0.3190149374243036,
      "grad_norm": 120.02606201171875,
      "learning_rate": 1.5930408699128636e-05,
      "loss": 5.9298,
      "step": 7902
    },
    {
      "epoch": 0.3190553088413403,
      "grad_norm": 73.62776947021484,
      "learning_rate": 1.592935582084437e-05,
      "loss": 4.8778,
      "step": 7903
    },
    {
      "epoch": 0.3190956802583771,
      "grad_norm": 95.46664428710938,
      "learning_rate": 1.5928302841181972e-05,
      "loss": 4.8712,
      "step": 7904
    },
    {
      "epoch": 0.3191360516754138,
      "grad_norm": 112.85442352294922,
      "learning_rate": 1.5927249760159435e-05,
      "loss": 5.3641,
      "step": 7905
    },
    {
      "epoch": 0.31917642309245053,
      "grad_norm": 99.10736083984375,
      "learning_rate": 1.5926196577794773e-05,
      "loss": 6.8332,
      "step": 7906
    },
    {
      "epoch": 0.31921679450948726,
      "grad_norm": 72.50416564941406,
      "learning_rate": 1.5925143294105986e-05,
      "loss": 5.4862,
      "step": 7907
    },
    {
      "epoch": 0.31925716592652403,
      "grad_norm": 68.58189392089844,
      "learning_rate": 1.5924089909111083e-05,
      "loss": 4.7951,
      "step": 7908
    },
    {
      "epoch": 0.31929753734356076,
      "grad_norm": 94.11516571044922,
      "learning_rate": 1.5923036422828075e-05,
      "loss": 5.8993,
      "step": 7909
    },
    {
      "epoch": 0.3193379087605975,
      "grad_norm": 76.90119934082031,
      "learning_rate": 1.592198283527498e-05,
      "loss": 4.9848,
      "step": 7910
    },
    {
      "epoch": 0.31937828017763426,
      "grad_norm": 71.71918487548828,
      "learning_rate": 1.5920929146469803e-05,
      "loss": 6.2598,
      "step": 7911
    },
    {
      "epoch": 0.319418651594671,
      "grad_norm": 124.47390747070312,
      "learning_rate": 1.5919875356430568e-05,
      "loss": 6.5023,
      "step": 7912
    },
    {
      "epoch": 0.3194590230117077,
      "grad_norm": 141.17007446289062,
      "learning_rate": 1.5918821465175284e-05,
      "loss": 6.1634,
      "step": 7913
    },
    {
      "epoch": 0.3194993944287445,
      "grad_norm": 113.63231658935547,
      "learning_rate": 1.5917767472721978e-05,
      "loss": 4.2971,
      "step": 7914
    },
    {
      "epoch": 0.3195397658457812,
      "grad_norm": 122.45350646972656,
      "learning_rate": 1.5916713379088666e-05,
      "loss": 4.3058,
      "step": 7915
    },
    {
      "epoch": 0.3195801372628179,
      "grad_norm": 137.6570587158203,
      "learning_rate": 1.591565918429337e-05,
      "loss": 5.7701,
      "step": 7916
    },
    {
      "epoch": 0.31962050867985464,
      "grad_norm": 93.71582794189453,
      "learning_rate": 1.591460488835412e-05,
      "loss": 5.1362,
      "step": 7917
    },
    {
      "epoch": 0.3196608800968914,
      "grad_norm": 91.54806518554688,
      "learning_rate": 1.5913550491288934e-05,
      "loss": 4.6073,
      "step": 7918
    },
    {
      "epoch": 0.31970125151392814,
      "grad_norm": 116.73207092285156,
      "learning_rate": 1.5912495993115848e-05,
      "loss": 4.4871,
      "step": 7919
    },
    {
      "epoch": 0.31974162293096486,
      "grad_norm": 111.69827270507812,
      "learning_rate": 1.5911441393852885e-05,
      "loss": 6.0137,
      "step": 7920
    },
    {
      "epoch": 0.31978199434800164,
      "grad_norm": 79.43096923828125,
      "learning_rate": 1.5910386693518076e-05,
      "loss": 6.0577,
      "step": 7921
    },
    {
      "epoch": 0.31982236576503836,
      "grad_norm": 93.44278717041016,
      "learning_rate": 1.590933189212946e-05,
      "loss": 6.1371,
      "step": 7922
    },
    {
      "epoch": 0.3198627371820751,
      "grad_norm": 72.54145050048828,
      "learning_rate": 1.5908276989705065e-05,
      "loss": 4.6336,
      "step": 7923
    },
    {
      "epoch": 0.3199031085991118,
      "grad_norm": 74.95919799804688,
      "learning_rate": 1.5907221986262934e-05,
      "loss": 5.2663,
      "step": 7924
    },
    {
      "epoch": 0.3199434800161486,
      "grad_norm": 87.44645690917969,
      "learning_rate": 1.5906166881821096e-05,
      "loss": 5.066,
      "step": 7925
    },
    {
      "epoch": 0.3199838514331853,
      "grad_norm": 135.95999145507812,
      "learning_rate": 1.5905111676397602e-05,
      "loss": 6.5103,
      "step": 7926
    },
    {
      "epoch": 0.32002422285022203,
      "grad_norm": 89.18450927734375,
      "learning_rate": 1.5904056370010482e-05,
      "loss": 6.0087,
      "step": 7927
    },
    {
      "epoch": 0.3200645942672588,
      "grad_norm": 60.21790313720703,
      "learning_rate": 1.590300096267779e-05,
      "loss": 5.1582,
      "step": 7928
    },
    {
      "epoch": 0.32010496568429553,
      "grad_norm": 76.87177276611328,
      "learning_rate": 1.590194545441756e-05,
      "loss": 4.0376,
      "step": 7929
    },
    {
      "epoch": 0.32014533710133225,
      "grad_norm": 93.31320190429688,
      "learning_rate": 1.5900889845247855e-05,
      "loss": 5.1046,
      "step": 7930
    },
    {
      "epoch": 0.32018570851836897,
      "grad_norm": 109.14603424072266,
      "learning_rate": 1.5899834135186705e-05,
      "loss": 6.4677,
      "step": 7931
    },
    {
      "epoch": 0.32022607993540575,
      "grad_norm": 91.30284118652344,
      "learning_rate": 1.589877832425217e-05,
      "loss": 4.2945,
      "step": 7932
    },
    {
      "epoch": 0.32026645135244247,
      "grad_norm": 161.568603515625,
      "learning_rate": 1.58977224124623e-05,
      "loss": 6.1707,
      "step": 7933
    },
    {
      "epoch": 0.3203068227694792,
      "grad_norm": 182.76913452148438,
      "learning_rate": 1.589666639983515e-05,
      "loss": 5.1939,
      "step": 7934
    },
    {
      "epoch": 0.32034719418651597,
      "grad_norm": 115.13275909423828,
      "learning_rate": 1.5895610286388773e-05,
      "loss": 5.8797,
      "step": 7935
    },
    {
      "epoch": 0.3203875656035527,
      "grad_norm": 158.3239288330078,
      "learning_rate": 1.589455407214123e-05,
      "loss": 3.1563,
      "step": 7936
    },
    {
      "epoch": 0.3204279370205894,
      "grad_norm": 97.17362213134766,
      "learning_rate": 1.5893497757110576e-05,
      "loss": 5.3952,
      "step": 7937
    },
    {
      "epoch": 0.32046830843762614,
      "grad_norm": 67.76506805419922,
      "learning_rate": 1.5892441341314873e-05,
      "loss": 6.1578,
      "step": 7938
    },
    {
      "epoch": 0.3205086798546629,
      "grad_norm": 88.10018920898438,
      "learning_rate": 1.589138482477218e-05,
      "loss": 4.6737,
      "step": 7939
    },
    {
      "epoch": 0.32054905127169964,
      "grad_norm": 69.25482177734375,
      "learning_rate": 1.5890328207500564e-05,
      "loss": 5.7687,
      "step": 7940
    },
    {
      "epoch": 0.32058942268873636,
      "grad_norm": 169.03997802734375,
      "learning_rate": 1.5889271489518092e-05,
      "loss": 4.7858,
      "step": 7941
    },
    {
      "epoch": 0.32062979410577314,
      "grad_norm": 125.11151123046875,
      "learning_rate": 1.5888214670842833e-05,
      "loss": 5.7236,
      "step": 7942
    },
    {
      "epoch": 0.32067016552280986,
      "grad_norm": 146.72482299804688,
      "learning_rate": 1.588715775149285e-05,
      "loss": 5.3821,
      "step": 7943
    },
    {
      "epoch": 0.3207105369398466,
      "grad_norm": 96.56403350830078,
      "learning_rate": 1.5886100731486215e-05,
      "loss": 4.3028,
      "step": 7944
    },
    {
      "epoch": 0.3207509083568833,
      "grad_norm": 99.30255889892578,
      "learning_rate": 1.5885043610841007e-05,
      "loss": 3.7192,
      "step": 7945
    },
    {
      "epoch": 0.3207912797739201,
      "grad_norm": 196.23001098632812,
      "learning_rate": 1.5883986389575295e-05,
      "loss": 6.0846,
      "step": 7946
    },
    {
      "epoch": 0.3208316511909568,
      "grad_norm": 140.9591522216797,
      "learning_rate": 1.5882929067707154e-05,
      "loss": 6.4734,
      "step": 7947
    },
    {
      "epoch": 0.3208720226079935,
      "grad_norm": 126.58739471435547,
      "learning_rate": 1.588187164525466e-05,
      "loss": 5.4668,
      "step": 7948
    },
    {
      "epoch": 0.3209123940250303,
      "grad_norm": 126.69715881347656,
      "learning_rate": 1.5880814122235904e-05,
      "loss": 7.5411,
      "step": 7949
    },
    {
      "epoch": 0.320952765442067,
      "grad_norm": 73.79875183105469,
      "learning_rate": 1.5879756498668956e-05,
      "loss": 3.8119,
      "step": 7950
    },
    {
      "epoch": 0.32099313685910374,
      "grad_norm": 100.86895751953125,
      "learning_rate": 1.58786987745719e-05,
      "loss": 5.6291,
      "step": 7951
    },
    {
      "epoch": 0.32103350827614047,
      "grad_norm": 145.56427001953125,
      "learning_rate": 1.5877640949962824e-05,
      "loss": 5.5335,
      "step": 7952
    },
    {
      "epoch": 0.32107387969317724,
      "grad_norm": 89.15443420410156,
      "learning_rate": 1.5876583024859814e-05,
      "loss": 6.3153,
      "step": 7953
    },
    {
      "epoch": 0.32111425111021397,
      "grad_norm": 297.700439453125,
      "learning_rate": 1.5875524999280956e-05,
      "loss": 8.1381,
      "step": 7954
    },
    {
      "epoch": 0.3211546225272507,
      "grad_norm": 161.22055053710938,
      "learning_rate": 1.587446687324434e-05,
      "loss": 4.1461,
      "step": 7955
    },
    {
      "epoch": 0.32119499394428747,
      "grad_norm": 118.23318481445312,
      "learning_rate": 1.587340864676806e-05,
      "loss": 4.5063,
      "step": 7956
    },
    {
      "epoch": 0.3212353653613242,
      "grad_norm": 176.88551330566406,
      "learning_rate": 1.587235031987021e-05,
      "loss": 6.6737,
      "step": 7957
    },
    {
      "epoch": 0.3212757367783609,
      "grad_norm": 75.00594329833984,
      "learning_rate": 1.5871291892568876e-05,
      "loss": 5.7099,
      "step": 7958
    },
    {
      "epoch": 0.32131610819539763,
      "grad_norm": 326.1627502441406,
      "learning_rate": 1.5870233364882162e-05,
      "loss": 4.2626,
      "step": 7959
    },
    {
      "epoch": 0.3213564796124344,
      "grad_norm": 115.15391540527344,
      "learning_rate": 1.586917473682817e-05,
      "loss": 4.2275,
      "step": 7960
    },
    {
      "epoch": 0.32139685102947113,
      "grad_norm": 93.78427124023438,
      "learning_rate": 1.5868116008424995e-05,
      "loss": 5.6408,
      "step": 7961
    },
    {
      "epoch": 0.32143722244650785,
      "grad_norm": 72.403076171875,
      "learning_rate": 1.5867057179690737e-05,
      "loss": 6.3685,
      "step": 7962
    },
    {
      "epoch": 0.32147759386354463,
      "grad_norm": 77.35746765136719,
      "learning_rate": 1.5865998250643505e-05,
      "loss": 6.2038,
      "step": 7963
    },
    {
      "epoch": 0.32151796528058135,
      "grad_norm": 136.92367553710938,
      "learning_rate": 1.5864939221301396e-05,
      "loss": 5.7598,
      "step": 7964
    },
    {
      "epoch": 0.3215583366976181,
      "grad_norm": 99.41307067871094,
      "learning_rate": 1.5863880091682525e-05,
      "loss": 6.2149,
      "step": 7965
    },
    {
      "epoch": 0.32159870811465485,
      "grad_norm": 86.06446075439453,
      "learning_rate": 1.5862820861804997e-05,
      "loss": 5.5006,
      "step": 7966
    },
    {
      "epoch": 0.3216390795316916,
      "grad_norm": 100.39088439941406,
      "learning_rate": 1.5861761531686922e-05,
      "loss": 5.325,
      "step": 7967
    },
    {
      "epoch": 0.3216794509487283,
      "grad_norm": 147.24917602539062,
      "learning_rate": 1.5860702101346416e-05,
      "loss": 6.1893,
      "step": 7968
    },
    {
      "epoch": 0.321719822365765,
      "grad_norm": 104.6547622680664,
      "learning_rate": 1.585964257080159e-05,
      "loss": 4.8448,
      "step": 7969
    },
    {
      "epoch": 0.3217601937828018,
      "grad_norm": 110.40914916992188,
      "learning_rate": 1.5858582940070553e-05,
      "loss": 5.3612,
      "step": 7970
    },
    {
      "epoch": 0.3218005651998385,
      "grad_norm": 102.47103881835938,
      "learning_rate": 1.5857523209171434e-05,
      "loss": 5.1933,
      "step": 7971
    },
    {
      "epoch": 0.32184093661687524,
      "grad_norm": 68.82640838623047,
      "learning_rate": 1.5856463378122343e-05,
      "loss": 3.7464,
      "step": 7972
    },
    {
      "epoch": 0.321881308033912,
      "grad_norm": 73.44393920898438,
      "learning_rate": 1.585540344694141e-05,
      "loss": 6.0975,
      "step": 7973
    },
    {
      "epoch": 0.32192167945094874,
      "grad_norm": 88.93762969970703,
      "learning_rate": 1.5854343415646744e-05,
      "loss": 4.412,
      "step": 7974
    },
    {
      "epoch": 0.32196205086798546,
      "grad_norm": 207.19349670410156,
      "learning_rate": 1.585328328425648e-05,
      "loss": 5.327,
      "step": 7975
    },
    {
      "epoch": 0.3220024222850222,
      "grad_norm": 91.3746337890625,
      "learning_rate": 1.585222305278874e-05,
      "loss": 5.0015,
      "step": 7976
    },
    {
      "epoch": 0.32204279370205896,
      "grad_norm": 128.23936462402344,
      "learning_rate": 1.5851162721261654e-05,
      "loss": 6.9047,
      "step": 7977
    },
    {
      "epoch": 0.3220831651190957,
      "grad_norm": 85.63932800292969,
      "learning_rate": 1.5850102289693343e-05,
      "loss": 5.9629,
      "step": 7978
    },
    {
      "epoch": 0.3221235365361324,
      "grad_norm": 74.53614807128906,
      "learning_rate": 1.584904175810195e-05,
      "loss": 5.5533,
      "step": 7979
    },
    {
      "epoch": 0.3221639079531692,
      "grad_norm": 65.41716766357422,
      "learning_rate": 1.58479811265056e-05,
      "loss": 5.6644,
      "step": 7980
    },
    {
      "epoch": 0.3222042793702059,
      "grad_norm": 112.4383316040039,
      "learning_rate": 1.584692039492243e-05,
      "loss": 4.0659,
      "step": 7981
    },
    {
      "epoch": 0.3222446507872426,
      "grad_norm": 113.61892700195312,
      "learning_rate": 1.584585956337057e-05,
      "loss": 5.5863,
      "step": 7982
    },
    {
      "epoch": 0.32228502220427935,
      "grad_norm": 79.5447006225586,
      "learning_rate": 1.5844798631868163e-05,
      "loss": 4.532,
      "step": 7983
    },
    {
      "epoch": 0.3223253936213161,
      "grad_norm": 71.16300201416016,
      "learning_rate": 1.5843737600433352e-05,
      "loss": 5.6997,
      "step": 7984
    },
    {
      "epoch": 0.32236576503835285,
      "grad_norm": 82.69476318359375,
      "learning_rate": 1.5842676469084272e-05,
      "loss": 5.9698,
      "step": 7985
    },
    {
      "epoch": 0.32240613645538957,
      "grad_norm": 102.0179672241211,
      "learning_rate": 1.584161523783907e-05,
      "loss": 4.8696,
      "step": 7986
    },
    {
      "epoch": 0.32244650787242635,
      "grad_norm": 332.8409729003906,
      "learning_rate": 1.5840553906715885e-05,
      "loss": 5.171,
      "step": 7987
    },
    {
      "epoch": 0.32248687928946307,
      "grad_norm": 131.57212829589844,
      "learning_rate": 1.5839492475732868e-05,
      "loss": 4.9186,
      "step": 7988
    },
    {
      "epoch": 0.3225272507064998,
      "grad_norm": 80.60038757324219,
      "learning_rate": 1.5838430944908166e-05,
      "loss": 4.4001,
      "step": 7989
    },
    {
      "epoch": 0.3225676221235365,
      "grad_norm": 108.3232192993164,
      "learning_rate": 1.583736931425993e-05,
      "loss": 4.3244,
      "step": 7990
    },
    {
      "epoch": 0.3226079935405733,
      "grad_norm": 96.45861053466797,
      "learning_rate": 1.583630758380631e-05,
      "loss": 5.4044,
      "step": 7991
    },
    {
      "epoch": 0.32264836495761,
      "grad_norm": 75.4764404296875,
      "learning_rate": 1.5835245753565457e-05,
      "loss": 6.1608,
      "step": 7992
    },
    {
      "epoch": 0.32268873637464673,
      "grad_norm": 73.6602554321289,
      "learning_rate": 1.583418382355553e-05,
      "loss": 4.6127,
      "step": 7993
    },
    {
      "epoch": 0.3227291077916835,
      "grad_norm": 80.09681701660156,
      "learning_rate": 1.583312179379468e-05,
      "loss": 2.7409,
      "step": 7994
    },
    {
      "epoch": 0.32276947920872023,
      "grad_norm": 135.07533264160156,
      "learning_rate": 1.5832059664301074e-05,
      "loss": 5.1942,
      "step": 7995
    },
    {
      "epoch": 0.32280985062575696,
      "grad_norm": 114.10836029052734,
      "learning_rate": 1.5830997435092863e-05,
      "loss": 4.2449,
      "step": 7996
    },
    {
      "epoch": 0.3228502220427937,
      "grad_norm": 252.41921997070312,
      "learning_rate": 1.5829935106188212e-05,
      "loss": 5.1457,
      "step": 7997
    },
    {
      "epoch": 0.32289059345983046,
      "grad_norm": 121.89990997314453,
      "learning_rate": 1.5828872677605284e-05,
      "loss": 5.3903,
      "step": 7998
    },
    {
      "epoch": 0.3229309648768672,
      "grad_norm": 81.73611450195312,
      "learning_rate": 1.582781014936225e-05,
      "loss": 4.9188,
      "step": 7999
    },
    {
      "epoch": 0.3229713362939039,
      "grad_norm": 164.46249389648438,
      "learning_rate": 1.5826747521477264e-05,
      "loss": 5.0586,
      "step": 8000
    },
    {
      "epoch": 0.3230117077109407,
      "grad_norm": 118.95246887207031,
      "learning_rate": 1.582568479396851e-05,
      "loss": 5.6764,
      "step": 8001
    },
    {
      "epoch": 0.3230520791279774,
      "grad_norm": 185.33934020996094,
      "learning_rate": 1.5824621966854145e-05,
      "loss": 3.5832,
      "step": 8002
    },
    {
      "epoch": 0.3230924505450141,
      "grad_norm": 139.81103515625,
      "learning_rate": 1.5823559040152347e-05,
      "loss": 5.5817,
      "step": 8003
    },
    {
      "epoch": 0.32313282196205084,
      "grad_norm": 155.98736572265625,
      "learning_rate": 1.5822496013881283e-05,
      "loss": 3.5874,
      "step": 8004
    },
    {
      "epoch": 0.3231731933790876,
      "grad_norm": 91.43597412109375,
      "learning_rate": 1.5821432888059142e-05,
      "loss": 6.0995,
      "step": 8005
    },
    {
      "epoch": 0.32321356479612434,
      "grad_norm": 115.7002944946289,
      "learning_rate": 1.5820369662704087e-05,
      "loss": 7.0973,
      "step": 8006
    },
    {
      "epoch": 0.32325393621316106,
      "grad_norm": 84.1581802368164,
      "learning_rate": 1.5819306337834303e-05,
      "loss": 6.2059,
      "step": 8007
    },
    {
      "epoch": 0.32329430763019784,
      "grad_norm": 118.30677032470703,
      "learning_rate": 1.5818242913467973e-05,
      "loss": 5.4957,
      "step": 8008
    },
    {
      "epoch": 0.32333467904723456,
      "grad_norm": 186.73190307617188,
      "learning_rate": 1.581717938962327e-05,
      "loss": 5.0241,
      "step": 8009
    },
    {
      "epoch": 0.3233750504642713,
      "grad_norm": 85.19791412353516,
      "learning_rate": 1.581611576631839e-05,
      "loss": 4.0437,
      "step": 8010
    },
    {
      "epoch": 0.323415421881308,
      "grad_norm": 135.94094848632812,
      "learning_rate": 1.581505204357151e-05,
      "loss": 5.4631,
      "step": 8011
    },
    {
      "epoch": 0.3234557932983448,
      "grad_norm": 88.80587005615234,
      "learning_rate": 1.5813988221400814e-05,
      "loss": 4.9556,
      "step": 8012
    },
    {
      "epoch": 0.3234961647153815,
      "grad_norm": 86.73670196533203,
      "learning_rate": 1.58129242998245e-05,
      "loss": 6.0163,
      "step": 8013
    },
    {
      "epoch": 0.32353653613241823,
      "grad_norm": 107.5760498046875,
      "learning_rate": 1.5811860278860757e-05,
      "loss": 6.4309,
      "step": 8014
    },
    {
      "epoch": 0.323576907549455,
      "grad_norm": 113.76426696777344,
      "learning_rate": 1.5810796158527773e-05,
      "loss": 6.5845,
      "step": 8015
    },
    {
      "epoch": 0.32361727896649173,
      "grad_norm": 145.93994140625,
      "learning_rate": 1.5809731938843743e-05,
      "loss": 4.2714,
      "step": 8016
    },
    {
      "epoch": 0.32365765038352845,
      "grad_norm": 104.80638885498047,
      "learning_rate": 1.5808667619826865e-05,
      "loss": 5.8018,
      "step": 8017
    },
    {
      "epoch": 0.32369802180056517,
      "grad_norm": 88.70365142822266,
      "learning_rate": 1.5807603201495334e-05,
      "loss": 5.3476,
      "step": 8018
    },
    {
      "epoch": 0.32373839321760195,
      "grad_norm": 193.86814880371094,
      "learning_rate": 1.580653868386735e-05,
      "loss": 4.874,
      "step": 8019
    },
    {
      "epoch": 0.32377876463463867,
      "grad_norm": 96.72603607177734,
      "learning_rate": 1.5805474066961114e-05,
      "loss": 5.6803,
      "step": 8020
    },
    {
      "epoch": 0.3238191360516754,
      "grad_norm": 155.5515594482422,
      "learning_rate": 1.5804409350794828e-05,
      "loss": 4.512,
      "step": 8021
    },
    {
      "epoch": 0.32385950746871217,
      "grad_norm": 89.625,
      "learning_rate": 1.58033445353867e-05,
      "loss": 5.4489,
      "step": 8022
    },
    {
      "epoch": 0.3238998788857489,
      "grad_norm": 77.85199737548828,
      "learning_rate": 1.5802279620754925e-05,
      "loss": 6.5447,
      "step": 8023
    },
    {
      "epoch": 0.3239402503027856,
      "grad_norm": 206.61138916015625,
      "learning_rate": 1.580121460691772e-05,
      "loss": 5.6156,
      "step": 8024
    },
    {
      "epoch": 0.3239806217198224,
      "grad_norm": 90.56869506835938,
      "learning_rate": 1.58001494938933e-05,
      "loss": 5.795,
      "step": 8025
    },
    {
      "epoch": 0.3240209931368591,
      "grad_norm": 79.69165802001953,
      "learning_rate": 1.5799084281699857e-05,
      "loss": 5.3436,
      "step": 8026
    },
    {
      "epoch": 0.32406136455389584,
      "grad_norm": 79.53755187988281,
      "learning_rate": 1.579801897035562e-05,
      "loss": 5.7133,
      "step": 8027
    },
    {
      "epoch": 0.32410173597093256,
      "grad_norm": 158.8041229248047,
      "learning_rate": 1.5796953559878796e-05,
      "loss": 4.9083,
      "step": 8028
    },
    {
      "epoch": 0.32414210738796934,
      "grad_norm": 118.55484008789062,
      "learning_rate": 1.5795888050287604e-05,
      "loss": 6.3054,
      "step": 8029
    },
    {
      "epoch": 0.32418247880500606,
      "grad_norm": 79.49674224853516,
      "learning_rate": 1.579482244160026e-05,
      "loss": 4.4608,
      "step": 8030
    },
    {
      "epoch": 0.3242228502220428,
      "grad_norm": 89.83697509765625,
      "learning_rate": 1.5793756733834987e-05,
      "loss": 5.0808,
      "step": 8031
    },
    {
      "epoch": 0.32426322163907956,
      "grad_norm": 113.99422454833984,
      "learning_rate": 1.579269092701e-05,
      "loss": 4.422,
      "step": 8032
    },
    {
      "epoch": 0.3243035930561163,
      "grad_norm": 93.7699966430664,
      "learning_rate": 1.5791625021143526e-05,
      "loss": 6.6158,
      "step": 8033
    },
    {
      "epoch": 0.324343964473153,
      "grad_norm": 80.28876495361328,
      "learning_rate": 1.579055901625379e-05,
      "loss": 6.8725,
      "step": 8034
    },
    {
      "epoch": 0.3243843358901897,
      "grad_norm": 104.34326171875,
      "learning_rate": 1.5789492912359015e-05,
      "loss": 6.0024,
      "step": 8035
    },
    {
      "epoch": 0.3244247073072265,
      "grad_norm": 95.80439758300781,
      "learning_rate": 1.578842670947743e-05,
      "loss": 4.8714,
      "step": 8036
    },
    {
      "epoch": 0.3244650787242632,
      "grad_norm": 93.51319885253906,
      "learning_rate": 1.578736040762727e-05,
      "loss": 4.5485,
      "step": 8037
    },
    {
      "epoch": 0.32450545014129994,
      "grad_norm": 67.7225112915039,
      "learning_rate": 1.5786294006826755e-05,
      "loss": 5.9856,
      "step": 8038
    },
    {
      "epoch": 0.3245458215583367,
      "grad_norm": 143.10342407226562,
      "learning_rate": 1.578522750709413e-05,
      "loss": 5.9868,
      "step": 8039
    },
    {
      "epoch": 0.32458619297537344,
      "grad_norm": 100.46102905273438,
      "learning_rate": 1.5784160908447618e-05,
      "loss": 3.9572,
      "step": 8040
    },
    {
      "epoch": 0.32462656439241017,
      "grad_norm": 79.7616195678711,
      "learning_rate": 1.5783094210905466e-05,
      "loss": 3.4299,
      "step": 8041
    },
    {
      "epoch": 0.3246669358094469,
      "grad_norm": 100.82470703125,
      "learning_rate": 1.5782027414485906e-05,
      "loss": 5.2069,
      "step": 8042
    },
    {
      "epoch": 0.32470730722648367,
      "grad_norm": 95.40550994873047,
      "learning_rate": 1.578096051920718e-05,
      "loss": 5.5394,
      "step": 8043
    },
    {
      "epoch": 0.3247476786435204,
      "grad_norm": 73.9083480834961,
      "learning_rate": 1.5779893525087527e-05,
      "loss": 5.2898,
      "step": 8044
    },
    {
      "epoch": 0.3247880500605571,
      "grad_norm": 99.07978820800781,
      "learning_rate": 1.5778826432145195e-05,
      "loss": 5.1372,
      "step": 8045
    },
    {
      "epoch": 0.3248284214775939,
      "grad_norm": 163.59841918945312,
      "learning_rate": 1.577775924039842e-05,
      "loss": 4.0276,
      "step": 8046
    },
    {
      "epoch": 0.3248687928946306,
      "grad_norm": 130.63531494140625,
      "learning_rate": 1.577669194986546e-05,
      "loss": 5.0977,
      "step": 8047
    },
    {
      "epoch": 0.32490916431166733,
      "grad_norm": 63.5425910949707,
      "learning_rate": 1.5775624560564552e-05,
      "loss": 6.5102,
      "step": 8048
    },
    {
      "epoch": 0.32494953572870405,
      "grad_norm": 113.94834899902344,
      "learning_rate": 1.5774557072513957e-05,
      "loss": 5.5383,
      "step": 8049
    },
    {
      "epoch": 0.32498990714574083,
      "grad_norm": 120.0945053100586,
      "learning_rate": 1.5773489485731915e-05,
      "loss": 6.0904,
      "step": 8050
    },
    {
      "epoch": 0.32503027856277755,
      "grad_norm": 77.96324920654297,
      "learning_rate": 1.577242180023669e-05,
      "loss": 5.7617,
      "step": 8051
    },
    {
      "epoch": 0.3250706499798143,
      "grad_norm": 158.42539978027344,
      "learning_rate": 1.5771354016046528e-05,
      "loss": 4.9232,
      "step": 8052
    },
    {
      "epoch": 0.32511102139685105,
      "grad_norm": 138.57672119140625,
      "learning_rate": 1.577028613317969e-05,
      "loss": 4.6385,
      "step": 8053
    },
    {
      "epoch": 0.3251513928138878,
      "grad_norm": 583.8086547851562,
      "learning_rate": 1.5769218151654433e-05,
      "loss": 6.2872,
      "step": 8054
    },
    {
      "epoch": 0.3251917642309245,
      "grad_norm": 118.23808288574219,
      "learning_rate": 1.576815007148902e-05,
      "loss": 5.5214,
      "step": 8055
    },
    {
      "epoch": 0.3252321356479612,
      "grad_norm": 81.53187561035156,
      "learning_rate": 1.576708189270171e-05,
      "loss": 4.7891,
      "step": 8056
    },
    {
      "epoch": 0.325272507064998,
      "grad_norm": 191.8927459716797,
      "learning_rate": 1.5766013615310764e-05,
      "loss": 4.801,
      "step": 8057
    },
    {
      "epoch": 0.3253128784820347,
      "grad_norm": 83.88213348388672,
      "learning_rate": 1.5764945239334453e-05,
      "loss": 4.4913,
      "step": 8058
    },
    {
      "epoch": 0.32535324989907144,
      "grad_norm": 94.63961029052734,
      "learning_rate": 1.576387676479104e-05,
      "loss": 4.4018,
      "step": 8059
    },
    {
      "epoch": 0.3253936213161082,
      "grad_norm": 129.19265747070312,
      "learning_rate": 1.5762808191698792e-05,
      "loss": 5.5021,
      "step": 8060
    },
    {
      "epoch": 0.32543399273314494,
      "grad_norm": 91.14935302734375,
      "learning_rate": 1.5761739520075985e-05,
      "loss": 6.7605,
      "step": 8061
    },
    {
      "epoch": 0.32547436415018166,
      "grad_norm": 71.98907470703125,
      "learning_rate": 1.5760670749940882e-05,
      "loss": 5.3809,
      "step": 8062
    },
    {
      "epoch": 0.3255147355672184,
      "grad_norm": 78.804443359375,
      "learning_rate": 1.5759601881311764e-05,
      "loss": 4.0722,
      "step": 8063
    },
    {
      "epoch": 0.32555510698425516,
      "grad_norm": 145.67214965820312,
      "learning_rate": 1.5758532914206904e-05,
      "loss": 5.6317,
      "step": 8064
    },
    {
      "epoch": 0.3255954784012919,
      "grad_norm": 101.55232238769531,
      "learning_rate": 1.5757463848644575e-05,
      "loss": 6.1705,
      "step": 8065
    },
    {
      "epoch": 0.3256358498183286,
      "grad_norm": 105.3482666015625,
      "learning_rate": 1.575639468464306e-05,
      "loss": 3.8654,
      "step": 8066
    },
    {
      "epoch": 0.3256762212353654,
      "grad_norm": 89.98358154296875,
      "learning_rate": 1.5755325422220642e-05,
      "loss": 5.8242,
      "step": 8067
    },
    {
      "epoch": 0.3257165926524021,
      "grad_norm": 133.9900665283203,
      "learning_rate": 1.5754256061395593e-05,
      "loss": 4.4381,
      "step": 8068
    },
    {
      "epoch": 0.3257569640694388,
      "grad_norm": 80.36224365234375,
      "learning_rate": 1.5753186602186207e-05,
      "loss": 5.7033,
      "step": 8069
    },
    {
      "epoch": 0.32579733548647555,
      "grad_norm": 99.82060241699219,
      "learning_rate": 1.5752117044610763e-05,
      "loss": 5.1544,
      "step": 8070
    },
    {
      "epoch": 0.3258377069035123,
      "grad_norm": 273.2958984375,
      "learning_rate": 1.575104738868755e-05,
      "loss": 3.1742,
      "step": 8071
    },
    {
      "epoch": 0.32587807832054905,
      "grad_norm": 104.71646881103516,
      "learning_rate": 1.5749977634434857e-05,
      "loss": 7.3992,
      "step": 8072
    },
    {
      "epoch": 0.32591844973758577,
      "grad_norm": 214.7588348388672,
      "learning_rate": 1.5748907781870974e-05,
      "loss": 6.4025,
      "step": 8073
    },
    {
      "epoch": 0.32595882115462255,
      "grad_norm": 105.11615753173828,
      "learning_rate": 1.574783783101419e-05,
      "loss": 4.3474,
      "step": 8074
    },
    {
      "epoch": 0.32599919257165927,
      "grad_norm": 88.67265319824219,
      "learning_rate": 1.5746767781882804e-05,
      "loss": 4.9885,
      "step": 8075
    },
    {
      "epoch": 0.326039563988696,
      "grad_norm": 120.17288208007812,
      "learning_rate": 1.5745697634495104e-05,
      "loss": 3.9413,
      "step": 8076
    },
    {
      "epoch": 0.32607993540573277,
      "grad_norm": 115.8528060913086,
      "learning_rate": 1.57446273888694e-05,
      "loss": 5.8208,
      "step": 8077
    },
    {
      "epoch": 0.3261203068227695,
      "grad_norm": 78.85003662109375,
      "learning_rate": 1.5743557045023974e-05,
      "loss": 6.2163,
      "step": 8078
    },
    {
      "epoch": 0.3261606782398062,
      "grad_norm": 85.97803497314453,
      "learning_rate": 1.574248660297714e-05,
      "loss": 7.1187,
      "step": 8079
    },
    {
      "epoch": 0.32620104965684293,
      "grad_norm": 112.86637115478516,
      "learning_rate": 1.574141606274719e-05,
      "loss": 5.1487,
      "step": 8080
    },
    {
      "epoch": 0.3262414210738797,
      "grad_norm": 85.14006042480469,
      "learning_rate": 1.5740345424352436e-05,
      "loss": 6.428,
      "step": 8081
    },
    {
      "epoch": 0.32628179249091643,
      "grad_norm": 167.9874725341797,
      "learning_rate": 1.573927468781118e-05,
      "loss": 4.6539,
      "step": 8082
    },
    {
      "epoch": 0.32632216390795316,
      "grad_norm": 74.86554718017578,
      "learning_rate": 1.5738203853141728e-05,
      "loss": 5.6685,
      "step": 8083
    },
    {
      "epoch": 0.32636253532498993,
      "grad_norm": 74.0051040649414,
      "learning_rate": 1.573713292036239e-05,
      "loss": 4.6892,
      "step": 8084
    },
    {
      "epoch": 0.32640290674202665,
      "grad_norm": 68.35941314697266,
      "learning_rate": 1.5736061889491477e-05,
      "loss": 5.8111,
      "step": 8085
    },
    {
      "epoch": 0.3264432781590634,
      "grad_norm": 84.4739761352539,
      "learning_rate": 1.57349907605473e-05,
      "loss": 5.0628,
      "step": 8086
    },
    {
      "epoch": 0.3264836495761001,
      "grad_norm": 104.9315414428711,
      "learning_rate": 1.573391953354817e-05,
      "loss": 4.0507,
      "step": 8087
    },
    {
      "epoch": 0.3265240209931369,
      "grad_norm": 110.1638412475586,
      "learning_rate": 1.573284820851241e-05,
      "loss": 5.5254,
      "step": 8088
    },
    {
      "epoch": 0.3265643924101736,
      "grad_norm": 100.95222473144531,
      "learning_rate": 1.5731776785458333e-05,
      "loss": 5.0292,
      "step": 8089
    },
    {
      "epoch": 0.3266047638272103,
      "grad_norm": 95.79684448242188,
      "learning_rate": 1.5730705264404254e-05,
      "loss": 4.2447,
      "step": 8090
    },
    {
      "epoch": 0.3266451352442471,
      "grad_norm": 121.1669921875,
      "learning_rate": 1.5729633645368497e-05,
      "loss": 4.9615,
      "step": 8091
    },
    {
      "epoch": 0.3266855066612838,
      "grad_norm": 88.26468658447266,
      "learning_rate": 1.5728561928369388e-05,
      "loss": 5.6985,
      "step": 8092
    },
    {
      "epoch": 0.32672587807832054,
      "grad_norm": 115.2071533203125,
      "learning_rate": 1.5727490113425244e-05,
      "loss": 5.6688,
      "step": 8093
    },
    {
      "epoch": 0.32676624949535726,
      "grad_norm": 102.09516906738281,
      "learning_rate": 1.5726418200554395e-05,
      "loss": 4.7312,
      "step": 8094
    },
    {
      "epoch": 0.32680662091239404,
      "grad_norm": 110.27166748046875,
      "learning_rate": 1.572534618977517e-05,
      "loss": 5.0028,
      "step": 8095
    },
    {
      "epoch": 0.32684699232943076,
      "grad_norm": 126.68907928466797,
      "learning_rate": 1.5724274081105888e-05,
      "loss": 4.8603,
      "step": 8096
    },
    {
      "epoch": 0.3268873637464675,
      "grad_norm": 129.7102813720703,
      "learning_rate": 1.5723201874564894e-05,
      "loss": 3.6939,
      "step": 8097
    },
    {
      "epoch": 0.32692773516350426,
      "grad_norm": 85.66194152832031,
      "learning_rate": 1.5722129570170507e-05,
      "loss": 6.546,
      "step": 8098
    },
    {
      "epoch": 0.326968106580541,
      "grad_norm": 76.91282653808594,
      "learning_rate": 1.572105716794107e-05,
      "loss": 4.9146,
      "step": 8099
    },
    {
      "epoch": 0.3270084779975777,
      "grad_norm": 64.86021423339844,
      "learning_rate": 1.571998466789491e-05,
      "loss": 5.9086,
      "step": 8100
    },
    {
      "epoch": 0.32704884941461443,
      "grad_norm": 161.16665649414062,
      "learning_rate": 1.5718912070050373e-05,
      "loss": 7.4129,
      "step": 8101
    },
    {
      "epoch": 0.3270892208316512,
      "grad_norm": 162.424560546875,
      "learning_rate": 1.5717839374425795e-05,
      "loss": 3.9359,
      "step": 8102
    },
    {
      "epoch": 0.32712959224868793,
      "grad_norm": 134.9999542236328,
      "learning_rate": 1.5716766581039515e-05,
      "loss": 5.9871,
      "step": 8103
    },
    {
      "epoch": 0.32716996366572465,
      "grad_norm": 128.8256378173828,
      "learning_rate": 1.5715693689909874e-05,
      "loss": 7.0759,
      "step": 8104
    },
    {
      "epoch": 0.3272103350827614,
      "grad_norm": 74.94342803955078,
      "learning_rate": 1.571462070105522e-05,
      "loss": 5.9974,
      "step": 8105
    },
    {
      "epoch": 0.32725070649979815,
      "grad_norm": 79.00865173339844,
      "learning_rate": 1.57135476144939e-05,
      "loss": 5.6726,
      "step": 8106
    },
    {
      "epoch": 0.32729107791683487,
      "grad_norm": 137.18905639648438,
      "learning_rate": 1.5712474430244254e-05,
      "loss": 6.5248,
      "step": 8107
    },
    {
      "epoch": 0.3273314493338716,
      "grad_norm": 106.45814514160156,
      "learning_rate": 1.5711401148324633e-05,
      "loss": 6.3066,
      "step": 8108
    },
    {
      "epoch": 0.32737182075090837,
      "grad_norm": 121.1650390625,
      "learning_rate": 1.571032776875339e-05,
      "loss": 6.5387,
      "step": 8109
    },
    {
      "epoch": 0.3274121921679451,
      "grad_norm": 84.13323974609375,
      "learning_rate": 1.570925429154888e-05,
      "loss": 5.4307,
      "step": 8110
    },
    {
      "epoch": 0.3274525635849818,
      "grad_norm": 194.18069458007812,
      "learning_rate": 1.570818071672945e-05,
      "loss": 5.6283,
      "step": 8111
    },
    {
      "epoch": 0.3274929350020186,
      "grad_norm": 77.37173461914062,
      "learning_rate": 1.5707107044313464e-05,
      "loss": 7.2246,
      "step": 8112
    },
    {
      "epoch": 0.3275333064190553,
      "grad_norm": 72.53289794921875,
      "learning_rate": 1.5706033274319272e-05,
      "loss": 6.2577,
      "step": 8113
    },
    {
      "epoch": 0.32757367783609204,
      "grad_norm": 106.9954833984375,
      "learning_rate": 1.5704959406765232e-05,
      "loss": 6.227,
      "step": 8114
    },
    {
      "epoch": 0.32761404925312876,
      "grad_norm": 71.58648681640625,
      "learning_rate": 1.5703885441669714e-05,
      "loss": 5.3006,
      "step": 8115
    },
    {
      "epoch": 0.32765442067016554,
      "grad_norm": 95.06490325927734,
      "learning_rate": 1.570281137905107e-05,
      "loss": 6.6933,
      "step": 8116
    },
    {
      "epoch": 0.32769479208720226,
      "grad_norm": 85.42337036132812,
      "learning_rate": 1.570173721892767e-05,
      "loss": 6.5161,
      "step": 8117
    },
    {
      "epoch": 0.327735163504239,
      "grad_norm": 102.05963897705078,
      "learning_rate": 1.5700662961317878e-05,
      "loss": 6.1003,
      "step": 8118
    },
    {
      "epoch": 0.32777553492127576,
      "grad_norm": 106.30870819091797,
      "learning_rate": 1.5699588606240066e-05,
      "loss": 5.401,
      "step": 8119
    },
    {
      "epoch": 0.3278159063383125,
      "grad_norm": 109.49081420898438,
      "learning_rate": 1.569851415371259e-05,
      "loss": 5.4173,
      "step": 8120
    },
    {
      "epoch": 0.3278562777553492,
      "grad_norm": 117.68236541748047,
      "learning_rate": 1.5697439603753836e-05,
      "loss": 4.9618,
      "step": 8121
    },
    {
      "epoch": 0.3278966491723859,
      "grad_norm": 61.22718048095703,
      "learning_rate": 1.5696364956382166e-05,
      "loss": 6.1399,
      "step": 8122
    },
    {
      "epoch": 0.3279370205894227,
      "grad_norm": 157.79653930664062,
      "learning_rate": 1.569529021161596e-05,
      "loss": 5.6635,
      "step": 8123
    },
    {
      "epoch": 0.3279773920064594,
      "grad_norm": 117.3892593383789,
      "learning_rate": 1.5694215369473584e-05,
      "loss": 5.0344,
      "step": 8124
    },
    {
      "epoch": 0.32801776342349614,
      "grad_norm": 121.85321044921875,
      "learning_rate": 1.569314042997343e-05,
      "loss": 4.6485,
      "step": 8125
    },
    {
      "epoch": 0.3280581348405329,
      "grad_norm": 100.56546020507812,
      "learning_rate": 1.5692065393133865e-05,
      "loss": 5.7041,
      "step": 8126
    },
    {
      "epoch": 0.32809850625756964,
      "grad_norm": 114.9592056274414,
      "learning_rate": 1.5690990258973275e-05,
      "loss": 6.4892,
      "step": 8127
    },
    {
      "epoch": 0.32813887767460637,
      "grad_norm": 102.46018981933594,
      "learning_rate": 1.568991502751004e-05,
      "loss": 3.8595,
      "step": 8128
    },
    {
      "epoch": 0.32817924909164314,
      "grad_norm": 78.45951080322266,
      "learning_rate": 1.5688839698762545e-05,
      "loss": 6.4996,
      "step": 8129
    },
    {
      "epoch": 0.32821962050867987,
      "grad_norm": 73.94793701171875,
      "learning_rate": 1.5687764272749177e-05,
      "loss": 5.4379,
      "step": 8130
    },
    {
      "epoch": 0.3282599919257166,
      "grad_norm": 73.11529541015625,
      "learning_rate": 1.568668874948832e-05,
      "loss": 5.187,
      "step": 8131
    },
    {
      "epoch": 0.3283003633427533,
      "grad_norm": 90.47161102294922,
      "learning_rate": 1.5685613128998364e-05,
      "loss": 6.1944,
      "step": 8132
    },
    {
      "epoch": 0.3283407347597901,
      "grad_norm": 91.06324005126953,
      "learning_rate": 1.56845374112977e-05,
      "loss": 5.9902,
      "step": 8133
    },
    {
      "epoch": 0.3283811061768268,
      "grad_norm": 86.28483581542969,
      "learning_rate": 1.5683461596404724e-05,
      "loss": 5.1901,
      "step": 8134
    },
    {
      "epoch": 0.32842147759386353,
      "grad_norm": 94.22975158691406,
      "learning_rate": 1.5682385684337823e-05,
      "loss": 6.2695,
      "step": 8135
    },
    {
      "epoch": 0.3284618490109003,
      "grad_norm": 83.98409271240234,
      "learning_rate": 1.56813096751154e-05,
      "loss": 5.8384,
      "step": 8136
    },
    {
      "epoch": 0.32850222042793703,
      "grad_norm": 96.36368560791016,
      "learning_rate": 1.5680233568755847e-05,
      "loss": 6.0071,
      "step": 8137
    },
    {
      "epoch": 0.32854259184497375,
      "grad_norm": 92.10031127929688,
      "learning_rate": 1.5679157365277564e-05,
      "loss": 5.3055,
      "step": 8138
    },
    {
      "epoch": 0.3285829632620105,
      "grad_norm": 92.338623046875,
      "learning_rate": 1.567808106469895e-05,
      "loss": 4.6857,
      "step": 8139
    },
    {
      "epoch": 0.32862333467904725,
      "grad_norm": 81.28319549560547,
      "learning_rate": 1.5677004667038412e-05,
      "loss": 6.4054,
      "step": 8140
    },
    {
      "epoch": 0.328663706096084,
      "grad_norm": 101.72694396972656,
      "learning_rate": 1.567592817231435e-05,
      "loss": 5.0807,
      "step": 8141
    },
    {
      "epoch": 0.3287040775131207,
      "grad_norm": 81.56243133544922,
      "learning_rate": 1.5674851580545168e-05,
      "loss": 7.394,
      "step": 8142
    },
    {
      "epoch": 0.3287444489301575,
      "grad_norm": 71.0301742553711,
      "learning_rate": 1.567377489174928e-05,
      "loss": 6.8482,
      "step": 8143
    },
    {
      "epoch": 0.3287848203471942,
      "grad_norm": 95.3550796508789,
      "learning_rate": 1.5672698105945086e-05,
      "loss": 4.871,
      "step": 8144
    },
    {
      "epoch": 0.3288251917642309,
      "grad_norm": 90.90491485595703,
      "learning_rate": 1.5671621223151004e-05,
      "loss": 4.7098,
      "step": 8145
    },
    {
      "epoch": 0.32886556318126764,
      "grad_norm": 80.53409576416016,
      "learning_rate": 1.5670544243385443e-05,
      "loss": 5.7214,
      "step": 8146
    },
    {
      "epoch": 0.3289059345983044,
      "grad_norm": 80.86603546142578,
      "learning_rate": 1.566946716666682e-05,
      "loss": 4.2826,
      "step": 8147
    },
    {
      "epoch": 0.32894630601534114,
      "grad_norm": 128.16233825683594,
      "learning_rate": 1.5668389993013543e-05,
      "loss": 6.2741,
      "step": 8148
    },
    {
      "epoch": 0.32898667743237786,
      "grad_norm": 140.18533325195312,
      "learning_rate": 1.5667312722444034e-05,
      "loss": 4.4432,
      "step": 8149
    },
    {
      "epoch": 0.32902704884941464,
      "grad_norm": 111.4932632446289,
      "learning_rate": 1.566623535497671e-05,
      "loss": 6.2287,
      "step": 8150
    },
    {
      "epoch": 0.32906742026645136,
      "grad_norm": 117.9592514038086,
      "learning_rate": 1.5665157890629997e-05,
      "loss": 6.0786,
      "step": 8151
    },
    {
      "epoch": 0.3291077916834881,
      "grad_norm": 75.93991088867188,
      "learning_rate": 1.5664080329422314e-05,
      "loss": 5.8819,
      "step": 8152
    },
    {
      "epoch": 0.3291481631005248,
      "grad_norm": 230.1680450439453,
      "learning_rate": 1.5663002671372084e-05,
      "loss": 6.0512,
      "step": 8153
    },
    {
      "epoch": 0.3291885345175616,
      "grad_norm": 87.67992401123047,
      "learning_rate": 1.566192491649773e-05,
      "loss": 5.4591,
      "step": 8154
    },
    {
      "epoch": 0.3292289059345983,
      "grad_norm": 107.20445251464844,
      "learning_rate": 1.5660847064817678e-05,
      "loss": 3.3942,
      "step": 8155
    },
    {
      "epoch": 0.329269277351635,
      "grad_norm": 96.2350082397461,
      "learning_rate": 1.5659769116350364e-05,
      "loss": 4.7769,
      "step": 8156
    },
    {
      "epoch": 0.3293096487686718,
      "grad_norm": 133.6276397705078,
      "learning_rate": 1.5658691071114218e-05,
      "loss": 5.407,
      "step": 8157
    },
    {
      "epoch": 0.3293500201857085,
      "grad_norm": 124.4494857788086,
      "learning_rate": 1.5657612929127662e-05,
      "loss": 4.4916,
      "step": 8158
    },
    {
      "epoch": 0.32939039160274525,
      "grad_norm": 106.08732604980469,
      "learning_rate": 1.5656534690409143e-05,
      "loss": 4.721,
      "step": 8159
    },
    {
      "epoch": 0.32943076301978197,
      "grad_norm": 80.7750015258789,
      "learning_rate": 1.5655456354977084e-05,
      "loss": 5.9922,
      "step": 8160
    },
    {
      "epoch": 0.32947113443681875,
      "grad_norm": 113.91527557373047,
      "learning_rate": 1.5654377922849932e-05,
      "loss": 5.2206,
      "step": 8161
    },
    {
      "epoch": 0.32951150585385547,
      "grad_norm": 66.9751968383789,
      "learning_rate": 1.565329939404612e-05,
      "loss": 6.1071,
      "step": 8162
    },
    {
      "epoch": 0.3295518772708922,
      "grad_norm": 195.2349853515625,
      "learning_rate": 1.565222076858409e-05,
      "loss": 4.3851,
      "step": 8163
    },
    {
      "epoch": 0.32959224868792897,
      "grad_norm": 145.05145263671875,
      "learning_rate": 1.5651142046482283e-05,
      "loss": 3.6641,
      "step": 8164
    },
    {
      "epoch": 0.3296326201049657,
      "grad_norm": 171.58892822265625,
      "learning_rate": 1.5650063227759143e-05,
      "loss": 4.5204,
      "step": 8165
    },
    {
      "epoch": 0.3296729915220024,
      "grad_norm": 167.1291046142578,
      "learning_rate": 1.5648984312433112e-05,
      "loss": 5.5058,
      "step": 8166
    },
    {
      "epoch": 0.32971336293903913,
      "grad_norm": 80.43531799316406,
      "learning_rate": 1.5647905300522648e-05,
      "loss": 6.0281,
      "step": 8167
    },
    {
      "epoch": 0.3297537343560759,
      "grad_norm": 128.13919067382812,
      "learning_rate": 1.5646826192046186e-05,
      "loss": 6.8573,
      "step": 8168
    },
    {
      "epoch": 0.32979410577311263,
      "grad_norm": 87.16279602050781,
      "learning_rate": 1.5645746987022188e-05,
      "loss": 5.7729,
      "step": 8169
    },
    {
      "epoch": 0.32983447719014936,
      "grad_norm": 165.7659912109375,
      "learning_rate": 1.5644667685469096e-05,
      "loss": 5.9107,
      "step": 8170
    },
    {
      "epoch": 0.32987484860718613,
      "grad_norm": 94.32001495361328,
      "learning_rate": 1.564358828740537e-05,
      "loss": 4.9268,
      "step": 8171
    },
    {
      "epoch": 0.32991522002422285,
      "grad_norm": 252.07266235351562,
      "learning_rate": 1.5642508792849463e-05,
      "loss": 3.9986,
      "step": 8172
    },
    {
      "epoch": 0.3299555914412596,
      "grad_norm": 88.49258422851562,
      "learning_rate": 1.564142920181983e-05,
      "loss": 4.7159,
      "step": 8173
    },
    {
      "epoch": 0.3299959628582963,
      "grad_norm": 66.87581634521484,
      "learning_rate": 1.5640349514334934e-05,
      "loss": 5.6614,
      "step": 8174
    },
    {
      "epoch": 0.3300363342753331,
      "grad_norm": 90.15835571289062,
      "learning_rate": 1.563926973041323e-05,
      "loss": 5.1831,
      "step": 8175
    },
    {
      "epoch": 0.3300767056923698,
      "grad_norm": 87.19505310058594,
      "learning_rate": 1.5638189850073184e-05,
      "loss": 7.0625,
      "step": 8176
    },
    {
      "epoch": 0.3301170771094065,
      "grad_norm": 84.85765075683594,
      "learning_rate": 1.5637109873333257e-05,
      "loss": 5.7335,
      "step": 8177
    },
    {
      "epoch": 0.3301574485264433,
      "grad_norm": 110.8461685180664,
      "learning_rate": 1.5636029800211917e-05,
      "loss": 4.9236,
      "step": 8178
    },
    {
      "epoch": 0.33019781994348,
      "grad_norm": 148.66366577148438,
      "learning_rate": 1.5634949630727627e-05,
      "loss": 4.2281,
      "step": 8179
    },
    {
      "epoch": 0.33023819136051674,
      "grad_norm": 78.84542846679688,
      "learning_rate": 1.5633869364898854e-05,
      "loss": 4.6633,
      "step": 8180
    },
    {
      "epoch": 0.3302785627775535,
      "grad_norm": 100.61678314208984,
      "learning_rate": 1.5632789002744075e-05,
      "loss": 6.6737,
      "step": 8181
    },
    {
      "epoch": 0.33031893419459024,
      "grad_norm": 91.128173828125,
      "learning_rate": 1.5631708544281756e-05,
      "loss": 5.1072,
      "step": 8182
    },
    {
      "epoch": 0.33035930561162696,
      "grad_norm": 74.9653091430664,
      "learning_rate": 1.563062798953037e-05,
      "loss": 6.1734,
      "step": 8183
    },
    {
      "epoch": 0.3303996770286637,
      "grad_norm": 99.88888549804688,
      "learning_rate": 1.562954733850839e-05,
      "loss": 6.1664,
      "step": 8184
    },
    {
      "epoch": 0.33044004844570046,
      "grad_norm": 112.79631042480469,
      "learning_rate": 1.5628466591234305e-05,
      "loss": 5.6179,
      "step": 8185
    },
    {
      "epoch": 0.3304804198627372,
      "grad_norm": 158.0160369873047,
      "learning_rate": 1.5627385747726585e-05,
      "loss": 6.1454,
      "step": 8186
    },
    {
      "epoch": 0.3305207912797739,
      "grad_norm": 124.05764770507812,
      "learning_rate": 1.5626304808003706e-05,
      "loss": 5.261,
      "step": 8187
    },
    {
      "epoch": 0.3305611626968107,
      "grad_norm": 129.6709747314453,
      "learning_rate": 1.5625223772084152e-05,
      "loss": 6.3243,
      "step": 8188
    },
    {
      "epoch": 0.3306015341138474,
      "grad_norm": 102.53315734863281,
      "learning_rate": 1.562414263998641e-05,
      "loss": 6.7038,
      "step": 8189
    },
    {
      "epoch": 0.33064190553088413,
      "grad_norm": 124.81716918945312,
      "learning_rate": 1.5623061411728956e-05,
      "loss": 5.0012,
      "step": 8190
    },
    {
      "epoch": 0.33068227694792085,
      "grad_norm": 96.75936126708984,
      "learning_rate": 1.5621980087330292e-05,
      "loss": 4.8164,
      "step": 8191
    },
    {
      "epoch": 0.3307226483649576,
      "grad_norm": 194.26678466796875,
      "learning_rate": 1.562089866680889e-05,
      "loss": 2.1639,
      "step": 8192
    },
    {
      "epoch": 0.33076301978199435,
      "grad_norm": 165.24343872070312,
      "learning_rate": 1.561981715018325e-05,
      "loss": 5.1889,
      "step": 8193
    },
    {
      "epoch": 0.33080339119903107,
      "grad_norm": 71.24667358398438,
      "learning_rate": 1.5618735537471857e-05,
      "loss": 5.948,
      "step": 8194
    },
    {
      "epoch": 0.33084376261606785,
      "grad_norm": 340.6913146972656,
      "learning_rate": 1.561765382869321e-05,
      "loss": 7.3533,
      "step": 8195
    },
    {
      "epoch": 0.33088413403310457,
      "grad_norm": 143.72251892089844,
      "learning_rate": 1.56165720238658e-05,
      "loss": 5.594,
      "step": 8196
    },
    {
      "epoch": 0.3309245054501413,
      "grad_norm": 82.3899154663086,
      "learning_rate": 1.5615490123008123e-05,
      "loss": 4.7207,
      "step": 8197
    },
    {
      "epoch": 0.330964876867178,
      "grad_norm": 110.33866882324219,
      "learning_rate": 1.561440812613868e-05,
      "loss": 5.4677,
      "step": 8198
    },
    {
      "epoch": 0.3310052482842148,
      "grad_norm": 111.70823669433594,
      "learning_rate": 1.5613326033275967e-05,
      "loss": 6.777,
      "step": 8199
    },
    {
      "epoch": 0.3310456197012515,
      "grad_norm": 85.02448272705078,
      "learning_rate": 1.5612243844438482e-05,
      "loss": 5.0645,
      "step": 8200
    },
    {
      "epoch": 0.33108599111828824,
      "grad_norm": 71.72334289550781,
      "learning_rate": 1.5611161559644742e-05,
      "loss": 4.6962,
      "step": 8201
    },
    {
      "epoch": 0.331126362535325,
      "grad_norm": 120.7423095703125,
      "learning_rate": 1.5610079178913234e-05,
      "loss": 7.3118,
      "step": 8202
    },
    {
      "epoch": 0.33116673395236174,
      "grad_norm": 152.0901336669922,
      "learning_rate": 1.560899670226248e-05,
      "loss": 5.398,
      "step": 8203
    },
    {
      "epoch": 0.33120710536939846,
      "grad_norm": 90.50787353515625,
      "learning_rate": 1.5607914129710975e-05,
      "loss": 5.9619,
      "step": 8204
    },
    {
      "epoch": 0.3312474767864352,
      "grad_norm": 82.1998291015625,
      "learning_rate": 1.560683146127724e-05,
      "loss": 5.6033,
      "step": 8205
    },
    {
      "epoch": 0.33128784820347196,
      "grad_norm": 162.19937133789062,
      "learning_rate": 1.5605748696979773e-05,
      "loss": 3.0779,
      "step": 8206
    },
    {
      "epoch": 0.3313282196205087,
      "grad_norm": 78.5672607421875,
      "learning_rate": 1.5604665836837097e-05,
      "loss": 4.0385,
      "step": 8207
    },
    {
      "epoch": 0.3313685910375454,
      "grad_norm": 80.58292388916016,
      "learning_rate": 1.5603582880867724e-05,
      "loss": 4.2111,
      "step": 8208
    },
    {
      "epoch": 0.3314089624545822,
      "grad_norm": 156.4082489013672,
      "learning_rate": 1.5602499829090163e-05,
      "loss": 4.816,
      "step": 8209
    },
    {
      "epoch": 0.3314493338716189,
      "grad_norm": 105.98148345947266,
      "learning_rate": 1.5601416681522943e-05,
      "loss": 5.1182,
      "step": 8210
    },
    {
      "epoch": 0.3314897052886556,
      "grad_norm": 155.648193359375,
      "learning_rate": 1.5600333438184578e-05,
      "loss": 4.1884,
      "step": 8211
    },
    {
      "epoch": 0.33153007670569234,
      "grad_norm": 69.74260711669922,
      "learning_rate": 1.5599250099093586e-05,
      "loss": 6.1751,
      "step": 8212
    },
    {
      "epoch": 0.3315704481227291,
      "grad_norm": 107.55493927001953,
      "learning_rate": 1.5598166664268495e-05,
      "loss": 4.9092,
      "step": 8213
    },
    {
      "epoch": 0.33161081953976584,
      "grad_norm": 135.59674072265625,
      "learning_rate": 1.5597083133727827e-05,
      "loss": 4.6947,
      "step": 8214
    },
    {
      "epoch": 0.33165119095680257,
      "grad_norm": 94.99974822998047,
      "learning_rate": 1.559599950749011e-05,
      "loss": 5.013,
      "step": 8215
    },
    {
      "epoch": 0.33169156237383934,
      "grad_norm": 106.93074035644531,
      "learning_rate": 1.559491578557386e-05,
      "loss": 5.559,
      "step": 8216
    },
    {
      "epoch": 0.33173193379087607,
      "grad_norm": 93.6051254272461,
      "learning_rate": 1.5593831967997622e-05,
      "loss": 4.4635,
      "step": 8217
    },
    {
      "epoch": 0.3317723052079128,
      "grad_norm": 136.46298217773438,
      "learning_rate": 1.559274805477992e-05,
      "loss": 4.4831,
      "step": 8218
    },
    {
      "epoch": 0.3318126766249495,
      "grad_norm": 90.17405700683594,
      "learning_rate": 1.5591664045939286e-05,
      "loss": 4.3834,
      "step": 8219
    },
    {
      "epoch": 0.3318530480419863,
      "grad_norm": 96.50128173828125,
      "learning_rate": 1.5590579941494253e-05,
      "loss": 4.3101,
      "step": 8220
    },
    {
      "epoch": 0.331893419459023,
      "grad_norm": 171.06411743164062,
      "learning_rate": 1.558949574146336e-05,
      "loss": 5.8549,
      "step": 8221
    },
    {
      "epoch": 0.33193379087605973,
      "grad_norm": 89.1474380493164,
      "learning_rate": 1.5588411445865138e-05,
      "loss": 5.3008,
      "step": 8222
    },
    {
      "epoch": 0.3319741622930965,
      "grad_norm": 82.20162200927734,
      "learning_rate": 1.5587327054718135e-05,
      "loss": 5.6323,
      "step": 8223
    },
    {
      "epoch": 0.33201453371013323,
      "grad_norm": 112.18692779541016,
      "learning_rate": 1.5586242568040882e-05,
      "loss": 5.6065,
      "step": 8224
    },
    {
      "epoch": 0.33205490512716995,
      "grad_norm": 95.84130859375,
      "learning_rate": 1.558515798585193e-05,
      "loss": 4.5544,
      "step": 8225
    },
    {
      "epoch": 0.3320952765442067,
      "grad_norm": 79.75045013427734,
      "learning_rate": 1.5584073308169812e-05,
      "loss": 4.9722,
      "step": 8226
    },
    {
      "epoch": 0.33213564796124345,
      "grad_norm": 148.78460693359375,
      "learning_rate": 1.5582988535013088e-05,
      "loss": 5.0917,
      "step": 8227
    },
    {
      "epoch": 0.3321760193782802,
      "grad_norm": 76.75033569335938,
      "learning_rate": 1.5581903666400294e-05,
      "loss": 5.2482,
      "step": 8228
    },
    {
      "epoch": 0.3322163907953169,
      "grad_norm": 90.77863311767578,
      "learning_rate": 1.5580818702349983e-05,
      "loss": 5.1694,
      "step": 8229
    },
    {
      "epoch": 0.3322567622123537,
      "grad_norm": 108.75428009033203,
      "learning_rate": 1.5579733642880703e-05,
      "loss": 5.5719,
      "step": 8230
    },
    {
      "epoch": 0.3322971336293904,
      "grad_norm": 124.28644561767578,
      "learning_rate": 1.557864848801101e-05,
      "loss": 5.3172,
      "step": 8231
    },
    {
      "epoch": 0.3323375050464271,
      "grad_norm": 96.73950958251953,
      "learning_rate": 1.557756323775945e-05,
      "loss": 5.413,
      "step": 8232
    },
    {
      "epoch": 0.3323778764634639,
      "grad_norm": 67.34583282470703,
      "learning_rate": 1.5576477892144587e-05,
      "loss": 5.386,
      "step": 8233
    },
    {
      "epoch": 0.3324182478805006,
      "grad_norm": 83.72362518310547,
      "learning_rate": 1.5575392451184976e-05,
      "loss": 6.5462,
      "step": 8234
    },
    {
      "epoch": 0.33245861929753734,
      "grad_norm": 81.21842956542969,
      "learning_rate": 1.5574306914899172e-05,
      "loss": 4.198,
      "step": 8235
    },
    {
      "epoch": 0.33249899071457406,
      "grad_norm": 102.01158142089844,
      "learning_rate": 1.557322128330573e-05,
      "loss": 5.0492,
      "step": 8236
    },
    {
      "epoch": 0.33253936213161084,
      "grad_norm": 84.86693572998047,
      "learning_rate": 1.5572135556423226e-05,
      "loss": 5.0611,
      "step": 8237
    },
    {
      "epoch": 0.33257973354864756,
      "grad_norm": 106.96942901611328,
      "learning_rate": 1.5571049734270214e-05,
      "loss": 4.7493,
      "step": 8238
    },
    {
      "epoch": 0.3326201049656843,
      "grad_norm": 88.31291198730469,
      "learning_rate": 1.5569963816865263e-05,
      "loss": 6.0332,
      "step": 8239
    },
    {
      "epoch": 0.33266047638272106,
      "grad_norm": 208.31878662109375,
      "learning_rate": 1.5568877804226934e-05,
      "loss": 5.459,
      "step": 8240
    },
    {
      "epoch": 0.3327008477997578,
      "grad_norm": 74.23319244384766,
      "learning_rate": 1.5567791696373798e-05,
      "loss": 4.6251,
      "step": 8241
    },
    {
      "epoch": 0.3327412192167945,
      "grad_norm": 79.86212158203125,
      "learning_rate": 1.5566705493324428e-05,
      "loss": 5.1414,
      "step": 8242
    },
    {
      "epoch": 0.3327815906338312,
      "grad_norm": 122.39176177978516,
      "learning_rate": 1.5565619195097394e-05,
      "loss": 4.2525,
      "step": 8243
    },
    {
      "epoch": 0.332821962050868,
      "grad_norm": 81.08171081542969,
      "learning_rate": 1.5564532801711268e-05,
      "loss": 5.4791,
      "step": 8244
    },
    {
      "epoch": 0.3328623334679047,
      "grad_norm": 121.55670928955078,
      "learning_rate": 1.5563446313184625e-05,
      "loss": 4.0273,
      "step": 8245
    },
    {
      "epoch": 0.33290270488494145,
      "grad_norm": 85.1768569946289,
      "learning_rate": 1.556235972953604e-05,
      "loss": 5.7614,
      "step": 8246
    },
    {
      "epoch": 0.3329430763019782,
      "grad_norm": 126.7115478515625,
      "learning_rate": 1.5561273050784093e-05,
      "loss": 4.5451,
      "step": 8247
    },
    {
      "epoch": 0.33298344771901495,
      "grad_norm": 97.07068634033203,
      "learning_rate": 1.5560186276947365e-05,
      "loss": 6.0687,
      "step": 8248
    },
    {
      "epoch": 0.33302381913605167,
      "grad_norm": 108.15416717529297,
      "learning_rate": 1.5559099408044436e-05,
      "loss": 4.7504,
      "step": 8249
    },
    {
      "epoch": 0.3330641905530884,
      "grad_norm": 82.71039581298828,
      "learning_rate": 1.5558012444093887e-05,
      "loss": 4.6102,
      "step": 8250
    },
    {
      "epoch": 0.33310456197012517,
      "grad_norm": 104.10630798339844,
      "learning_rate": 1.5556925385114306e-05,
      "loss": 4.7331,
      "step": 8251
    },
    {
      "epoch": 0.3331449333871619,
      "grad_norm": 207.0696563720703,
      "learning_rate": 1.5555838231124275e-05,
      "loss": 6.028,
      "step": 8252
    },
    {
      "epoch": 0.3331853048041986,
      "grad_norm": 130.3153076171875,
      "learning_rate": 1.5554750982142382e-05,
      "loss": 6.0398,
      "step": 8253
    },
    {
      "epoch": 0.3332256762212354,
      "grad_norm": 178.9852752685547,
      "learning_rate": 1.555366363818722e-05,
      "loss": 4.99,
      "step": 8254
    },
    {
      "epoch": 0.3332660476382721,
      "grad_norm": 114.84773254394531,
      "learning_rate": 1.5552576199277384e-05,
      "loss": 5.7845,
      "step": 8255
    },
    {
      "epoch": 0.33330641905530883,
      "grad_norm": 89.27747344970703,
      "learning_rate": 1.5551488665431454e-05,
      "loss": 4.8993,
      "step": 8256
    },
    {
      "epoch": 0.33334679047234556,
      "grad_norm": 109.8609848022461,
      "learning_rate": 1.5550401036668038e-05,
      "loss": 4.4442,
      "step": 8257
    },
    {
      "epoch": 0.33338716188938233,
      "grad_norm": 117.12364959716797,
      "learning_rate": 1.554931331300572e-05,
      "loss": 3.8945,
      "step": 8258
    },
    {
      "epoch": 0.33342753330641905,
      "grad_norm": 247.7454376220703,
      "learning_rate": 1.5548225494463106e-05,
      "loss": 5.5551,
      "step": 8259
    },
    {
      "epoch": 0.3334679047234558,
      "grad_norm": 109.82997131347656,
      "learning_rate": 1.554713758105879e-05,
      "loss": 4.8414,
      "step": 8260
    },
    {
      "epoch": 0.33350827614049255,
      "grad_norm": 403.7210998535156,
      "learning_rate": 1.5546049572811378e-05,
      "loss": 5.9316,
      "step": 8261
    },
    {
      "epoch": 0.3335486475575293,
      "grad_norm": 145.82620239257812,
      "learning_rate": 1.5544961469739466e-05,
      "loss": 5.0175,
      "step": 8262
    },
    {
      "epoch": 0.333589018974566,
      "grad_norm": 92.7822036743164,
      "learning_rate": 1.5543873271861663e-05,
      "loss": 6.5429,
      "step": 8263
    },
    {
      "epoch": 0.3336293903916027,
      "grad_norm": 119.39614868164062,
      "learning_rate": 1.554278497919657e-05,
      "loss": 4.7312,
      "step": 8264
    },
    {
      "epoch": 0.3336697618086395,
      "grad_norm": 79.39445495605469,
      "learning_rate": 1.5541696591762804e-05,
      "loss": 4.7685,
      "step": 8265
    },
    {
      "epoch": 0.3337101332256762,
      "grad_norm": 102.00466918945312,
      "learning_rate": 1.554060810957896e-05,
      "loss": 4.9313,
      "step": 8266
    },
    {
      "epoch": 0.33375050464271294,
      "grad_norm": 188.00198364257812,
      "learning_rate": 1.5539519532663656e-05,
      "loss": 3.8963,
      "step": 8267
    },
    {
      "epoch": 0.3337908760597497,
      "grad_norm": 148.12408447265625,
      "learning_rate": 1.553843086103551e-05,
      "loss": 4.7883,
      "step": 8268
    },
    {
      "epoch": 0.33383124747678644,
      "grad_norm": 95.91405487060547,
      "learning_rate": 1.5537342094713123e-05,
      "loss": 5.1665,
      "step": 8269
    },
    {
      "epoch": 0.33387161889382316,
      "grad_norm": 108.01093292236328,
      "learning_rate": 1.5536253233715117e-05,
      "loss": 3.6487,
      "step": 8270
    },
    {
      "epoch": 0.3339119903108599,
      "grad_norm": 394.6930847167969,
      "learning_rate": 1.553516427806011e-05,
      "loss": 5.2026,
      "step": 8271
    },
    {
      "epoch": 0.33395236172789666,
      "grad_norm": 94.71161651611328,
      "learning_rate": 1.553407522776672e-05,
      "loss": 3.5437,
      "step": 8272
    },
    {
      "epoch": 0.3339927331449334,
      "grad_norm": 90.521484375,
      "learning_rate": 1.5532986082853568e-05,
      "loss": 5.7218,
      "step": 8273
    },
    {
      "epoch": 0.3340331045619701,
      "grad_norm": 151.0395050048828,
      "learning_rate": 1.5531896843339272e-05,
      "loss": 5.3538,
      "step": 8274
    },
    {
      "epoch": 0.3340734759790069,
      "grad_norm": 106.97908020019531,
      "learning_rate": 1.5530807509242462e-05,
      "loss": 5.6644,
      "step": 8275
    },
    {
      "epoch": 0.3341138473960436,
      "grad_norm": 98.94774627685547,
      "learning_rate": 1.5529718080581755e-05,
      "loss": 5.795,
      "step": 8276
    },
    {
      "epoch": 0.33415421881308033,
      "grad_norm": 102.8499526977539,
      "learning_rate": 1.5528628557375783e-05,
      "loss": 6.0609,
      "step": 8277
    },
    {
      "epoch": 0.33419459023011705,
      "grad_norm": 99.2379150390625,
      "learning_rate": 1.5527538939643173e-05,
      "loss": 4.6814,
      "step": 8278
    },
    {
      "epoch": 0.3342349616471538,
      "grad_norm": 112.53716278076172,
      "learning_rate": 1.5526449227402554e-05,
      "loss": 3.251,
      "step": 8279
    },
    {
      "epoch": 0.33427533306419055,
      "grad_norm": 109.02169036865234,
      "learning_rate": 1.552535942067256e-05,
      "loss": 5.8407,
      "step": 8280
    },
    {
      "epoch": 0.33431570448122727,
      "grad_norm": 109.33033752441406,
      "learning_rate": 1.5524269519471823e-05,
      "loss": 5.6225,
      "step": 8281
    },
    {
      "epoch": 0.33435607589826405,
      "grad_norm": 86.4358139038086,
      "learning_rate": 1.5523179523818974e-05,
      "loss": 6.7598,
      "step": 8282
    },
    {
      "epoch": 0.33439644731530077,
      "grad_norm": 113.92192077636719,
      "learning_rate": 1.5522089433732658e-05,
      "loss": 4.9325,
      "step": 8283
    },
    {
      "epoch": 0.3344368187323375,
      "grad_norm": 81.65184020996094,
      "learning_rate": 1.55209992492315e-05,
      "loss": 5.8996,
      "step": 8284
    },
    {
      "epoch": 0.33447719014937427,
      "grad_norm": 97.87455749511719,
      "learning_rate": 1.5519908970334153e-05,
      "loss": 5.1596,
      "step": 8285
    },
    {
      "epoch": 0.334517561566411,
      "grad_norm": 205.79901123046875,
      "learning_rate": 1.551881859705925e-05,
      "loss": 4.4895,
      "step": 8286
    },
    {
      "epoch": 0.3345579329834477,
      "grad_norm": 94.27427673339844,
      "learning_rate": 1.5517728129425437e-05,
      "loss": 6.1069,
      "step": 8287
    },
    {
      "epoch": 0.33459830440048444,
      "grad_norm": 162.2668914794922,
      "learning_rate": 1.5516637567451357e-05,
      "loss": 5.4832,
      "step": 8288
    },
    {
      "epoch": 0.3346386758175212,
      "grad_norm": 92.03559875488281,
      "learning_rate": 1.5515546911155662e-05,
      "loss": 5.739,
      "step": 8289
    },
    {
      "epoch": 0.33467904723455794,
      "grad_norm": 78.71733856201172,
      "learning_rate": 1.5514456160556987e-05,
      "loss": 4.6029,
      "step": 8290
    },
    {
      "epoch": 0.33471941865159466,
      "grad_norm": 89.88956451416016,
      "learning_rate": 1.5513365315673994e-05,
      "loss": 5.0912,
      "step": 8291
    },
    {
      "epoch": 0.33475979006863144,
      "grad_norm": 106.92145538330078,
      "learning_rate": 1.5512274376525328e-05,
      "loss": 5.0027,
      "step": 8292
    },
    {
      "epoch": 0.33480016148566816,
      "grad_norm": 273.64837646484375,
      "learning_rate": 1.551118334312964e-05,
      "loss": 5.8679,
      "step": 8293
    },
    {
      "epoch": 0.3348405329027049,
      "grad_norm": 115.79048156738281,
      "learning_rate": 1.551009221550559e-05,
      "loss": 3.8595,
      "step": 8294
    },
    {
      "epoch": 0.3348809043197416,
      "grad_norm": 119.75697326660156,
      "learning_rate": 1.5509000993671826e-05,
      "loss": 4.6439,
      "step": 8295
    },
    {
      "epoch": 0.3349212757367784,
      "grad_norm": 84.465087890625,
      "learning_rate": 1.550790967764701e-05,
      "loss": 5.4905,
      "step": 8296
    },
    {
      "epoch": 0.3349616471538151,
      "grad_norm": 78.61588287353516,
      "learning_rate": 1.5506818267449804e-05,
      "loss": 7.1625,
      "step": 8297
    },
    {
      "epoch": 0.3350020185708518,
      "grad_norm": 107.1236343383789,
      "learning_rate": 1.5505726763098864e-05,
      "loss": 5.5858,
      "step": 8298
    },
    {
      "epoch": 0.3350423899878886,
      "grad_norm": 100.28002166748047,
      "learning_rate": 1.550463516461285e-05,
      "loss": 6.4601,
      "step": 8299
    },
    {
      "epoch": 0.3350827614049253,
      "grad_norm": 96.6833267211914,
      "learning_rate": 1.550354347201043e-05,
      "loss": 4.4388,
      "step": 8300
    },
    {
      "epoch": 0.33512313282196204,
      "grad_norm": 102.83013916015625,
      "learning_rate": 1.550245168531027e-05,
      "loss": 2.3274,
      "step": 8301
    },
    {
      "epoch": 0.33516350423899877,
      "grad_norm": 105.90660858154297,
      "learning_rate": 1.5501359804531034e-05,
      "loss": 5.5847,
      "step": 8302
    },
    {
      "epoch": 0.33520387565603554,
      "grad_norm": 186.52001953125,
      "learning_rate": 1.5500267829691392e-05,
      "loss": 4.8268,
      "step": 8303
    },
    {
      "epoch": 0.33524424707307227,
      "grad_norm": 164.25599670410156,
      "learning_rate": 1.5499175760810015e-05,
      "loss": 5.0494,
      "step": 8304
    },
    {
      "epoch": 0.335284618490109,
      "grad_norm": 186.2911376953125,
      "learning_rate": 1.5498083597905575e-05,
      "loss": 3.599,
      "step": 8305
    },
    {
      "epoch": 0.33532498990714577,
      "grad_norm": 100.6375503540039,
      "learning_rate": 1.549699134099674e-05,
      "loss": 6.4371,
      "step": 8306
    },
    {
      "epoch": 0.3353653613241825,
      "grad_norm": 104.82095336914062,
      "learning_rate": 1.5495898990102195e-05,
      "loss": 5.3996,
      "step": 8307
    },
    {
      "epoch": 0.3354057327412192,
      "grad_norm": 109.19189453125,
      "learning_rate": 1.549480654524061e-05,
      "loss": 4.7925,
      "step": 8308
    },
    {
      "epoch": 0.33544610415825593,
      "grad_norm": 85.63694763183594,
      "learning_rate": 1.5493714006430663e-05,
      "loss": 5.8249,
      "step": 8309
    },
    {
      "epoch": 0.3354864755752927,
      "grad_norm": 83.8934555053711,
      "learning_rate": 1.5492621373691033e-05,
      "loss": 5.7591,
      "step": 8310
    },
    {
      "epoch": 0.33552684699232943,
      "grad_norm": 73.17987060546875,
      "learning_rate": 1.5491528647040407e-05,
      "loss": 4.8843,
      "step": 8311
    },
    {
      "epoch": 0.33556721840936615,
      "grad_norm": 150.57254028320312,
      "learning_rate": 1.549043582649746e-05,
      "loss": 4.6667,
      "step": 8312
    },
    {
      "epoch": 0.33560758982640293,
      "grad_norm": 76.70359802246094,
      "learning_rate": 1.548934291208089e-05,
      "loss": 5.551,
      "step": 8313
    },
    {
      "epoch": 0.33564796124343965,
      "grad_norm": 118.37234497070312,
      "learning_rate": 1.548824990380937e-05,
      "loss": 6.0988,
      "step": 8314
    },
    {
      "epoch": 0.3356883326604764,
      "grad_norm": 154.0071258544922,
      "learning_rate": 1.548715680170159e-05,
      "loss": 8.6362,
      "step": 8315
    },
    {
      "epoch": 0.3357287040775131,
      "grad_norm": 105.41019439697266,
      "learning_rate": 1.5486063605776245e-05,
      "loss": 6.2362,
      "step": 8316
    },
    {
      "epoch": 0.3357690754945499,
      "grad_norm": 109.94169616699219,
      "learning_rate": 1.5484970316052024e-05,
      "loss": 5.2146,
      "step": 8317
    },
    {
      "epoch": 0.3358094469115866,
      "grad_norm": 102.33905029296875,
      "learning_rate": 1.5483876932547615e-05,
      "loss": 5.8593,
      "step": 8318
    },
    {
      "epoch": 0.3358498183286233,
      "grad_norm": 116.9135971069336,
      "learning_rate": 1.5482783455281724e-05,
      "loss": 6.3496,
      "step": 8319
    },
    {
      "epoch": 0.3358901897456601,
      "grad_norm": 98.89482116699219,
      "learning_rate": 1.548168988427303e-05,
      "loss": 2.8902,
      "step": 8320
    },
    {
      "epoch": 0.3359305611626968,
      "grad_norm": 105.34625244140625,
      "learning_rate": 1.5480596219540245e-05,
      "loss": 5.4778,
      "step": 8321
    },
    {
      "epoch": 0.33597093257973354,
      "grad_norm": 135.5098876953125,
      "learning_rate": 1.547950246110206e-05,
      "loss": 4.8316,
      "step": 8322
    },
    {
      "epoch": 0.33601130399677026,
      "grad_norm": 90.34600830078125,
      "learning_rate": 1.547840860897718e-05,
      "loss": 6.3261,
      "step": 8323
    },
    {
      "epoch": 0.33605167541380704,
      "grad_norm": 143.9420928955078,
      "learning_rate": 1.5477314663184307e-05,
      "loss": 4.2311,
      "step": 8324
    },
    {
      "epoch": 0.33609204683084376,
      "grad_norm": 99.4668960571289,
      "learning_rate": 1.5476220623742144e-05,
      "loss": 4.7258,
      "step": 8325
    },
    {
      "epoch": 0.3361324182478805,
      "grad_norm": 133.0107421875,
      "learning_rate": 1.5475126490669393e-05,
      "loss": 4.6255,
      "step": 8326
    },
    {
      "epoch": 0.33617278966491726,
      "grad_norm": 79.83357238769531,
      "learning_rate": 1.5474032263984766e-05,
      "loss": 4.9883,
      "step": 8327
    },
    {
      "epoch": 0.336213161081954,
      "grad_norm": 97.75086975097656,
      "learning_rate": 1.547293794370697e-05,
      "loss": 4.3101,
      "step": 8328
    },
    {
      "epoch": 0.3362535324989907,
      "grad_norm": 88.28976440429688,
      "learning_rate": 1.547184352985472e-05,
      "loss": 4.5702,
      "step": 8329
    },
    {
      "epoch": 0.3362939039160274,
      "grad_norm": 111.4886245727539,
      "learning_rate": 1.5470749022446713e-05,
      "loss": 4.6258,
      "step": 8330
    },
    {
      "epoch": 0.3363342753330642,
      "grad_norm": 128.4032440185547,
      "learning_rate": 1.546965442150168e-05,
      "loss": 4.9097,
      "step": 8331
    },
    {
      "epoch": 0.3363746467501009,
      "grad_norm": 75.24937438964844,
      "learning_rate": 1.5468559727038327e-05,
      "loss": 4.7727,
      "step": 8332
    },
    {
      "epoch": 0.33641501816713765,
      "grad_norm": 114.15418243408203,
      "learning_rate": 1.5467464939075375e-05,
      "loss": 6.2498,
      "step": 8333
    },
    {
      "epoch": 0.3364553895841744,
      "grad_norm": 145.7285614013672,
      "learning_rate": 1.546637005763154e-05,
      "loss": 5.1855,
      "step": 8334
    },
    {
      "epoch": 0.33649576100121115,
      "grad_norm": 149.27587890625,
      "learning_rate": 1.5465275082725538e-05,
      "loss": 4.3957,
      "step": 8335
    },
    {
      "epoch": 0.33653613241824787,
      "grad_norm": 113.38957977294922,
      "learning_rate": 1.5464180014376096e-05,
      "loss": 5.8126,
      "step": 8336
    },
    {
      "epoch": 0.33657650383528465,
      "grad_norm": 540.5101318359375,
      "learning_rate": 1.546308485260194e-05,
      "loss": 5.2104,
      "step": 8337
    },
    {
      "epoch": 0.33661687525232137,
      "grad_norm": 87.40089416503906,
      "learning_rate": 1.5461989597421782e-05,
      "loss": 5.1818,
      "step": 8338
    },
    {
      "epoch": 0.3366572466693581,
      "grad_norm": 123.75067901611328,
      "learning_rate": 1.546089424885436e-05,
      "loss": 4.95,
      "step": 8339
    },
    {
      "epoch": 0.3366976180863948,
      "grad_norm": 141.56797790527344,
      "learning_rate": 1.54597988069184e-05,
      "loss": 3.3818,
      "step": 8340
    },
    {
      "epoch": 0.3367379895034316,
      "grad_norm": 95.06205749511719,
      "learning_rate": 1.545870327163263e-05,
      "loss": 6.3129,
      "step": 8341
    },
    {
      "epoch": 0.3367783609204683,
      "grad_norm": 81.82474517822266,
      "learning_rate": 1.545760764301578e-05,
      "loss": 4.7414,
      "step": 8342
    },
    {
      "epoch": 0.33681873233750503,
      "grad_norm": 117.81592559814453,
      "learning_rate": 1.5456511921086585e-05,
      "loss": 4.4197,
      "step": 8343
    },
    {
      "epoch": 0.3368591037545418,
      "grad_norm": 73.0197525024414,
      "learning_rate": 1.5455416105863775e-05,
      "loss": 5.6616,
      "step": 8344
    },
    {
      "epoch": 0.33689947517157853,
      "grad_norm": 158.64573669433594,
      "learning_rate": 1.5454320197366093e-05,
      "loss": 4.9789,
      "step": 8345
    },
    {
      "epoch": 0.33693984658861525,
      "grad_norm": 183.9267578125,
      "learning_rate": 1.545322419561227e-05,
      "loss": 5.0139,
      "step": 8346
    },
    {
      "epoch": 0.336980218005652,
      "grad_norm": 136.5833740234375,
      "learning_rate": 1.545212810062105e-05,
      "loss": 5.7887,
      "step": 8347
    },
    {
      "epoch": 0.33702058942268875,
      "grad_norm": 105.86711120605469,
      "learning_rate": 1.5451031912411165e-05,
      "loss": 6.4232,
      "step": 8348
    },
    {
      "epoch": 0.3370609608397255,
      "grad_norm": 108.06082916259766,
      "learning_rate": 1.544993563100137e-05,
      "loss": 5.0923,
      "step": 8349
    },
    {
      "epoch": 0.3371013322567622,
      "grad_norm": 167.8319091796875,
      "learning_rate": 1.54488392564104e-05,
      "loss": 4.8739,
      "step": 8350
    },
    {
      "epoch": 0.337141703673799,
      "grad_norm": 77.06885528564453,
      "learning_rate": 1.5447742788657005e-05,
      "loss": 4.9156,
      "step": 8351
    },
    {
      "epoch": 0.3371820750908357,
      "grad_norm": 94.63150024414062,
      "learning_rate": 1.5446646227759924e-05,
      "loss": 5.5733,
      "step": 8352
    },
    {
      "epoch": 0.3372224465078724,
      "grad_norm": 113.03749084472656,
      "learning_rate": 1.5445549573737913e-05,
      "loss": 4.4044,
      "step": 8353
    },
    {
      "epoch": 0.33726281792490914,
      "grad_norm": 83.46092987060547,
      "learning_rate": 1.5444452826609722e-05,
      "loss": 5.2472,
      "step": 8354
    },
    {
      "epoch": 0.3373031893419459,
      "grad_norm": 141.46914672851562,
      "learning_rate": 1.54433559863941e-05,
      "loss": 5.6311,
      "step": 8355
    },
    {
      "epoch": 0.33734356075898264,
      "grad_norm": 94.70655822753906,
      "learning_rate": 1.5442259053109803e-05,
      "loss": 5.3745,
      "step": 8356
    },
    {
      "epoch": 0.33738393217601936,
      "grad_norm": 83.47167205810547,
      "learning_rate": 1.5441162026775585e-05,
      "loss": 3.3443,
      "step": 8357
    },
    {
      "epoch": 0.33742430359305614,
      "grad_norm": 139.204345703125,
      "learning_rate": 1.5440064907410204e-05,
      "loss": 5.1983,
      "step": 8358
    },
    {
      "epoch": 0.33746467501009286,
      "grad_norm": 77.22799682617188,
      "learning_rate": 1.5438967695032413e-05,
      "loss": 4.4854,
      "step": 8359
    },
    {
      "epoch": 0.3375050464271296,
      "grad_norm": 106.01973724365234,
      "learning_rate": 1.5437870389660978e-05,
      "loss": 7.1869,
      "step": 8360
    },
    {
      "epoch": 0.3375454178441663,
      "grad_norm": 119.18134307861328,
      "learning_rate": 1.5436772991314657e-05,
      "loss": 5.7584,
      "step": 8361
    },
    {
      "epoch": 0.3375857892612031,
      "grad_norm": 136.43145751953125,
      "learning_rate": 1.5435675500012212e-05,
      "loss": 5.7649,
      "step": 8362
    },
    {
      "epoch": 0.3376261606782398,
      "grad_norm": 102.92111206054688,
      "learning_rate": 1.5434577915772412e-05,
      "loss": 4.4231,
      "step": 8363
    },
    {
      "epoch": 0.33766653209527653,
      "grad_norm": 118.57856750488281,
      "learning_rate": 1.543348023861402e-05,
      "loss": 5.675,
      "step": 8364
    },
    {
      "epoch": 0.3377069035123133,
      "grad_norm": 190.92758178710938,
      "learning_rate": 1.5432382468555803e-05,
      "loss": 3.8317,
      "step": 8365
    },
    {
      "epoch": 0.33774727492935,
      "grad_norm": 134.7378387451172,
      "learning_rate": 1.5431284605616528e-05,
      "loss": 3.2942,
      "step": 8366
    },
    {
      "epoch": 0.33778764634638675,
      "grad_norm": 83.13494873046875,
      "learning_rate": 1.5430186649814975e-05,
      "loss": 5.4957,
      "step": 8367
    },
    {
      "epoch": 0.33782801776342347,
      "grad_norm": 148.5968475341797,
      "learning_rate": 1.542908860116991e-05,
      "loss": 6.1076,
      "step": 8368
    },
    {
      "epoch": 0.33786838918046025,
      "grad_norm": 178.59706115722656,
      "learning_rate": 1.5427990459700103e-05,
      "loss": 6.5378,
      "step": 8369
    },
    {
      "epoch": 0.33790876059749697,
      "grad_norm": 87.3341064453125,
      "learning_rate": 1.542689222542434e-05,
      "loss": 6.5263,
      "step": 8370
    },
    {
      "epoch": 0.3379491320145337,
      "grad_norm": 94.75331115722656,
      "learning_rate": 1.5425793898361387e-05,
      "loss": 4.9529,
      "step": 8371
    },
    {
      "epoch": 0.33798950343157047,
      "grad_norm": 102.02770233154297,
      "learning_rate": 1.542469547853003e-05,
      "loss": 5.0232,
      "step": 8372
    },
    {
      "epoch": 0.3380298748486072,
      "grad_norm": 145.793212890625,
      "learning_rate": 1.542359696594905e-05,
      "loss": 6.7423,
      "step": 8373
    },
    {
      "epoch": 0.3380702462656439,
      "grad_norm": 141.7210693359375,
      "learning_rate": 1.5422498360637225e-05,
      "loss": 5.5996,
      "step": 8374
    },
    {
      "epoch": 0.33811061768268064,
      "grad_norm": 149.1375732421875,
      "learning_rate": 1.5421399662613344e-05,
      "loss": 4.3251,
      "step": 8375
    },
    {
      "epoch": 0.3381509890997174,
      "grad_norm": 89.13043212890625,
      "learning_rate": 1.5420300871896182e-05,
      "loss": 5.3908,
      "step": 8376
    },
    {
      "epoch": 0.33819136051675414,
      "grad_norm": 67.66474151611328,
      "learning_rate": 1.5419201988504538e-05,
      "loss": 2.771,
      "step": 8377
    },
    {
      "epoch": 0.33823173193379086,
      "grad_norm": 104.09002685546875,
      "learning_rate": 1.5418103012457194e-05,
      "loss": 4.7251,
      "step": 8378
    },
    {
      "epoch": 0.33827210335082764,
      "grad_norm": 104.79765319824219,
      "learning_rate": 1.5417003943772943e-05,
      "loss": 6.3668,
      "step": 8379
    },
    {
      "epoch": 0.33831247476786436,
      "grad_norm": 97.11398315429688,
      "learning_rate": 1.541590478247057e-05,
      "loss": 4.2979,
      "step": 8380
    },
    {
      "epoch": 0.3383528461849011,
      "grad_norm": 141.99632263183594,
      "learning_rate": 1.541480552856887e-05,
      "loss": 5.0574,
      "step": 8381
    },
    {
      "epoch": 0.3383932176019378,
      "grad_norm": 106.08683013916016,
      "learning_rate": 1.5413706182086646e-05,
      "loss": 3.2818,
      "step": 8382
    },
    {
      "epoch": 0.3384335890189746,
      "grad_norm": 109.64909362792969,
      "learning_rate": 1.5412606743042687e-05,
      "loss": 3.7234,
      "step": 8383
    },
    {
      "epoch": 0.3384739604360113,
      "grad_norm": 93.56123352050781,
      "learning_rate": 1.541150721145579e-05,
      "loss": 4.4394,
      "step": 8384
    },
    {
      "epoch": 0.338514331853048,
      "grad_norm": 66.40922546386719,
      "learning_rate": 1.5410407587344755e-05,
      "loss": 5.3621,
      "step": 8385
    },
    {
      "epoch": 0.3385547032700848,
      "grad_norm": 81.66939544677734,
      "learning_rate": 1.5409307870728386e-05,
      "loss": 6.1732,
      "step": 8386
    },
    {
      "epoch": 0.3385950746871215,
      "grad_norm": 85.33052062988281,
      "learning_rate": 1.5408208061625487e-05,
      "loss": 5.5737,
      "step": 8387
    },
    {
      "epoch": 0.33863544610415824,
      "grad_norm": 89.79383850097656,
      "learning_rate": 1.5407108160054857e-05,
      "loss": 2.9499,
      "step": 8388
    },
    {
      "epoch": 0.338675817521195,
      "grad_norm": 63.615440368652344,
      "learning_rate": 1.5406008166035303e-05,
      "loss": 4.6557,
      "step": 8389
    },
    {
      "epoch": 0.33871618893823174,
      "grad_norm": 99.60513305664062,
      "learning_rate": 1.5404908079585634e-05,
      "loss": 4.48,
      "step": 8390
    },
    {
      "epoch": 0.33875656035526847,
      "grad_norm": 169.69668579101562,
      "learning_rate": 1.540380790072466e-05,
      "loss": 6.2208,
      "step": 8391
    },
    {
      "epoch": 0.3387969317723052,
      "grad_norm": 91.86029052734375,
      "learning_rate": 1.5402707629471186e-05,
      "loss": 6.1394,
      "step": 8392
    },
    {
      "epoch": 0.33883730318934197,
      "grad_norm": 204.53262329101562,
      "learning_rate": 1.5401607265844032e-05,
      "loss": 4.5626,
      "step": 8393
    },
    {
      "epoch": 0.3388776746063787,
      "grad_norm": 110.8997573852539,
      "learning_rate": 1.5400506809862003e-05,
      "loss": 4.9176,
      "step": 8394
    },
    {
      "epoch": 0.3389180460234154,
      "grad_norm": 144.338623046875,
      "learning_rate": 1.539940626154392e-05,
      "loss": 5.5661,
      "step": 8395
    },
    {
      "epoch": 0.3389584174404522,
      "grad_norm": 99.86650848388672,
      "learning_rate": 1.53983056209086e-05,
      "loss": 4.396,
      "step": 8396
    },
    {
      "epoch": 0.3389987888574889,
      "grad_norm": 325.6405029296875,
      "learning_rate": 1.539720488797486e-05,
      "loss": 7.6304,
      "step": 8397
    },
    {
      "epoch": 0.33903916027452563,
      "grad_norm": 96.28173065185547,
      "learning_rate": 1.5396104062761517e-05,
      "loss": 5.1091,
      "step": 8398
    },
    {
      "epoch": 0.33907953169156235,
      "grad_norm": 102.59526062011719,
      "learning_rate": 1.53950031452874e-05,
      "loss": 5.8885,
      "step": 8399
    },
    {
      "epoch": 0.33911990310859913,
      "grad_norm": 81.81493377685547,
      "learning_rate": 1.539390213557132e-05,
      "loss": 5.583,
      "step": 8400
    },
    {
      "epoch": 0.33916027452563585,
      "grad_norm": 123.72279357910156,
      "learning_rate": 1.5392801033632116e-05,
      "loss": 5.6377,
      "step": 8401
    },
    {
      "epoch": 0.3392006459426726,
      "grad_norm": 107.6915054321289,
      "learning_rate": 1.5391699839488605e-05,
      "loss": 5.6331,
      "step": 8402
    },
    {
      "epoch": 0.33924101735970935,
      "grad_norm": 122.13525390625,
      "learning_rate": 1.539059855315962e-05,
      "loss": 6.0301,
      "step": 8403
    },
    {
      "epoch": 0.3392813887767461,
      "grad_norm": 107.92782592773438,
      "learning_rate": 1.5389497174663985e-05,
      "loss": 6.3842,
      "step": 8404
    },
    {
      "epoch": 0.3393217601937828,
      "grad_norm": 195.4067840576172,
      "learning_rate": 1.5388395704020534e-05,
      "loss": 3.7931,
      "step": 8405
    },
    {
      "epoch": 0.3393621316108195,
      "grad_norm": 236.96636962890625,
      "learning_rate": 1.53872941412481e-05,
      "loss": 3.5983,
      "step": 8406
    },
    {
      "epoch": 0.3394025030278563,
      "grad_norm": 130.7678680419922,
      "learning_rate": 1.5386192486365512e-05,
      "loss": 4.7002,
      "step": 8407
    },
    {
      "epoch": 0.339442874444893,
      "grad_norm": 112.2562484741211,
      "learning_rate": 1.5385090739391616e-05,
      "loss": 5.3262,
      "step": 8408
    },
    {
      "epoch": 0.33948324586192974,
      "grad_norm": 135.48484802246094,
      "learning_rate": 1.538398890034524e-05,
      "loss": 4.8184,
      "step": 8409
    },
    {
      "epoch": 0.3395236172789665,
      "grad_norm": 114.56101989746094,
      "learning_rate": 1.538288696924523e-05,
      "loss": 5.0357,
      "step": 8410
    },
    {
      "epoch": 0.33956398869600324,
      "grad_norm": 99.2569808959961,
      "learning_rate": 1.538178494611042e-05,
      "loss": 5.9224,
      "step": 8411
    },
    {
      "epoch": 0.33960436011303996,
      "grad_norm": 75.67974853515625,
      "learning_rate": 1.5380682830959654e-05,
      "loss": 5.9502,
      "step": 8412
    },
    {
      "epoch": 0.3396447315300767,
      "grad_norm": 105.11449432373047,
      "learning_rate": 1.537958062381178e-05,
      "loss": 4.5846,
      "step": 8413
    },
    {
      "epoch": 0.33968510294711346,
      "grad_norm": 160.86244201660156,
      "learning_rate": 1.5378478324685633e-05,
      "loss": 4.4559,
      "step": 8414
    },
    {
      "epoch": 0.3397254743641502,
      "grad_norm": 102.03500366210938,
      "learning_rate": 1.5377375933600075e-05,
      "loss": 5.3069,
      "step": 8415
    },
    {
      "epoch": 0.3397658457811869,
      "grad_norm": 112.48688507080078,
      "learning_rate": 1.5376273450573938e-05,
      "loss": 4.6003,
      "step": 8416
    },
    {
      "epoch": 0.3398062171982237,
      "grad_norm": 101.29886627197266,
      "learning_rate": 1.5375170875626082e-05,
      "loss": 4.6328,
      "step": 8417
    },
    {
      "epoch": 0.3398465886152604,
      "grad_norm": 80.30207061767578,
      "learning_rate": 1.5374068208775358e-05,
      "loss": 3.5854,
      "step": 8418
    },
    {
      "epoch": 0.3398869600322971,
      "grad_norm": 90.6688003540039,
      "learning_rate": 1.5372965450040613e-05,
      "loss": 5.817,
      "step": 8419
    },
    {
      "epoch": 0.33992733144933385,
      "grad_norm": 104.3021011352539,
      "learning_rate": 1.5371862599440708e-05,
      "loss": 6.4215,
      "step": 8420
    },
    {
      "epoch": 0.3399677028663706,
      "grad_norm": 109.5409927368164,
      "learning_rate": 1.5370759656994495e-05,
      "loss": 4.8711,
      "step": 8421
    },
    {
      "epoch": 0.34000807428340735,
      "grad_norm": 125.64117431640625,
      "learning_rate": 1.5369656622720838e-05,
      "loss": 6.3543,
      "step": 8422
    },
    {
      "epoch": 0.34004844570044407,
      "grad_norm": 144.00379943847656,
      "learning_rate": 1.536855349663859e-05,
      "loss": 5.574,
      "step": 8423
    },
    {
      "epoch": 0.34008881711748085,
      "grad_norm": 108.82611083984375,
      "learning_rate": 1.536745027876661e-05,
      "loss": 4.5765,
      "step": 8424
    },
    {
      "epoch": 0.34012918853451757,
      "grad_norm": 138.37103271484375,
      "learning_rate": 1.536634696912377e-05,
      "loss": 6.2113,
      "step": 8425
    },
    {
      "epoch": 0.3401695599515543,
      "grad_norm": 111.5543441772461,
      "learning_rate": 1.5365243567728923e-05,
      "loss": 5.3413,
      "step": 8426
    },
    {
      "epoch": 0.340209931368591,
      "grad_norm": 110.70384979248047,
      "learning_rate": 1.5364140074600943e-05,
      "loss": 3.3214,
      "step": 8427
    },
    {
      "epoch": 0.3402503027856278,
      "grad_norm": 106.12235260009766,
      "learning_rate": 1.536303648975869e-05,
      "loss": 3.8683,
      "step": 8428
    },
    {
      "epoch": 0.3402906742026645,
      "grad_norm": 121.90370178222656,
      "learning_rate": 1.536193281322104e-05,
      "loss": 5.6173,
      "step": 8429
    },
    {
      "epoch": 0.34033104561970123,
      "grad_norm": 108.2942123413086,
      "learning_rate": 1.5360829045006858e-05,
      "loss": 4.9751,
      "step": 8430
    },
    {
      "epoch": 0.340371417036738,
      "grad_norm": 75.08218383789062,
      "learning_rate": 1.5359725185135022e-05,
      "loss": 5.7603,
      "step": 8431
    },
    {
      "epoch": 0.34041178845377473,
      "grad_norm": 120.67926025390625,
      "learning_rate": 1.5358621233624395e-05,
      "loss": 5.4275,
      "step": 8432
    },
    {
      "epoch": 0.34045215987081145,
      "grad_norm": 146.2056884765625,
      "learning_rate": 1.5357517190493864e-05,
      "loss": 5.3671,
      "step": 8433
    },
    {
      "epoch": 0.3404925312878482,
      "grad_norm": 97.29022216796875,
      "learning_rate": 1.5356413055762296e-05,
      "loss": 5.7677,
      "step": 8434
    },
    {
      "epoch": 0.34053290270488495,
      "grad_norm": 97.24764251708984,
      "learning_rate": 1.5355308829448574e-05,
      "loss": 5.2786,
      "step": 8435
    },
    {
      "epoch": 0.3405732741219217,
      "grad_norm": 91.1507797241211,
      "learning_rate": 1.5354204511571576e-05,
      "loss": 6.7185,
      "step": 8436
    },
    {
      "epoch": 0.3406136455389584,
      "grad_norm": 88.2528076171875,
      "learning_rate": 1.535310010215018e-05,
      "loss": 4.2131,
      "step": 8437
    },
    {
      "epoch": 0.3406540169559952,
      "grad_norm": 103.15640258789062,
      "learning_rate": 1.535199560120328e-05,
      "loss": 5.3938,
      "step": 8438
    },
    {
      "epoch": 0.3406943883730319,
      "grad_norm": 121.73226928710938,
      "learning_rate": 1.535089100874975e-05,
      "loss": 4.7636,
      "step": 8439
    },
    {
      "epoch": 0.3407347597900686,
      "grad_norm": 112.06773376464844,
      "learning_rate": 1.534978632480848e-05,
      "loss": 6.4229,
      "step": 8440
    },
    {
      "epoch": 0.34077513120710534,
      "grad_norm": 116.71560668945312,
      "learning_rate": 1.5348681549398355e-05,
      "loss": 5.1387,
      "step": 8441
    },
    {
      "epoch": 0.3408155026241421,
      "grad_norm": 639.7734375,
      "learning_rate": 1.5347576682538264e-05,
      "loss": 6.7901,
      "step": 8442
    },
    {
      "epoch": 0.34085587404117884,
      "grad_norm": 95.99889373779297,
      "learning_rate": 1.5346471724247102e-05,
      "loss": 5.0795,
      "step": 8443
    },
    {
      "epoch": 0.34089624545821556,
      "grad_norm": 134.5214080810547,
      "learning_rate": 1.5345366674543762e-05,
      "loss": 4.9438,
      "step": 8444
    },
    {
      "epoch": 0.34093661687525234,
      "grad_norm": 66.36017608642578,
      "learning_rate": 1.5344261533447125e-05,
      "loss": 5.6785,
      "step": 8445
    },
    {
      "epoch": 0.34097698829228906,
      "grad_norm": 92.76655578613281,
      "learning_rate": 1.5343156300976104e-05,
      "loss": 5.8061,
      "step": 8446
    },
    {
      "epoch": 0.3410173597093258,
      "grad_norm": 123.59314727783203,
      "learning_rate": 1.534205097714958e-05,
      "loss": 4.7426,
      "step": 8447
    },
    {
      "epoch": 0.34105773112636256,
      "grad_norm": 80.937744140625,
      "learning_rate": 1.5340945561986466e-05,
      "loss": 6.1784,
      "step": 8448
    },
    {
      "epoch": 0.3410981025433993,
      "grad_norm": 147.61004638671875,
      "learning_rate": 1.5339840055505647e-05,
      "loss": 4.2665,
      "step": 8449
    },
    {
      "epoch": 0.341138473960436,
      "grad_norm": 102.36641693115234,
      "learning_rate": 1.533873445772604e-05,
      "loss": 3.7139,
      "step": 8450
    },
    {
      "epoch": 0.34117884537747273,
      "grad_norm": 128.2873992919922,
      "learning_rate": 1.5337628768666533e-05,
      "loss": 7.3542,
      "step": 8451
    },
    {
      "epoch": 0.3412192167945095,
      "grad_norm": 89.62632751464844,
      "learning_rate": 1.5336522988346047e-05,
      "loss": 5.2972,
      "step": 8452
    },
    {
      "epoch": 0.3412595882115462,
      "grad_norm": 116.67137908935547,
      "learning_rate": 1.5335417116783474e-05,
      "loss": 7.5448,
      "step": 8453
    },
    {
      "epoch": 0.34129995962858295,
      "grad_norm": 100.31680297851562,
      "learning_rate": 1.5334311153997732e-05,
      "loss": 5.5131,
      "step": 8454
    },
    {
      "epoch": 0.3413403310456197,
      "grad_norm": 195.64398193359375,
      "learning_rate": 1.533320510000772e-05,
      "loss": 6.8668,
      "step": 8455
    },
    {
      "epoch": 0.34138070246265645,
      "grad_norm": 110.01821899414062,
      "learning_rate": 1.533209895483236e-05,
      "loss": 6.6168,
      "step": 8456
    },
    {
      "epoch": 0.34142107387969317,
      "grad_norm": 141.5291748046875,
      "learning_rate": 1.5330992718490555e-05,
      "loss": 4.3132,
      "step": 8457
    },
    {
      "epoch": 0.3414614452967299,
      "grad_norm": 99.83649444580078,
      "learning_rate": 1.5329886391001222e-05,
      "loss": 6.0114,
      "step": 8458
    },
    {
      "epoch": 0.34150181671376667,
      "grad_norm": 68.59078979492188,
      "learning_rate": 1.532877997238328e-05,
      "loss": 3.8946,
      "step": 8459
    },
    {
      "epoch": 0.3415421881308034,
      "grad_norm": 81.84535217285156,
      "learning_rate": 1.532767346265565e-05,
      "loss": 6.8548,
      "step": 8460
    },
    {
      "epoch": 0.3415825595478401,
      "grad_norm": 107.4296646118164,
      "learning_rate": 1.5326566861837237e-05,
      "loss": 4.3178,
      "step": 8461
    },
    {
      "epoch": 0.3416229309648769,
      "grad_norm": 106.14300537109375,
      "learning_rate": 1.5325460169946974e-05,
      "loss": 5.3453,
      "step": 8462
    },
    {
      "epoch": 0.3416633023819136,
      "grad_norm": 97.16987609863281,
      "learning_rate": 1.5324353387003773e-05,
      "loss": 4.9301,
      "step": 8463
    },
    {
      "epoch": 0.34170367379895034,
      "grad_norm": 80.60903930664062,
      "learning_rate": 1.5323246513026567e-05,
      "loss": 4.5618,
      "step": 8464
    },
    {
      "epoch": 0.34174404521598706,
      "grad_norm": 128.15728759765625,
      "learning_rate": 1.5322139548034274e-05,
      "loss": 4.7863,
      "step": 8465
    },
    {
      "epoch": 0.34178441663302384,
      "grad_norm": 80.01805114746094,
      "learning_rate": 1.5321032492045822e-05,
      "loss": 5.1794,
      "step": 8466
    },
    {
      "epoch": 0.34182478805006056,
      "grad_norm": 105.70799255371094,
      "learning_rate": 1.5319925345080142e-05,
      "loss": 5.1757,
      "step": 8467
    },
    {
      "epoch": 0.3418651594670973,
      "grad_norm": 100.47891998291016,
      "learning_rate": 1.531881810715616e-05,
      "loss": 4.7343,
      "step": 8468
    },
    {
      "epoch": 0.34190553088413406,
      "grad_norm": 92.87352752685547,
      "learning_rate": 1.531771077829281e-05,
      "loss": 6.311,
      "step": 8469
    },
    {
      "epoch": 0.3419459023011708,
      "grad_norm": 72.80714416503906,
      "learning_rate": 1.5316603358509022e-05,
      "loss": 4.5198,
      "step": 8470
    },
    {
      "epoch": 0.3419862737182075,
      "grad_norm": 86.49907684326172,
      "learning_rate": 1.5315495847823733e-05,
      "loss": 4.4417,
      "step": 8471
    },
    {
      "epoch": 0.3420266451352442,
      "grad_norm": 122.05213165283203,
      "learning_rate": 1.5314388246255877e-05,
      "loss": 3.9247,
      "step": 8472
    },
    {
      "epoch": 0.342067016552281,
      "grad_norm": 91.27777862548828,
      "learning_rate": 1.531328055382439e-05,
      "loss": 5.2572,
      "step": 8473
    },
    {
      "epoch": 0.3421073879693177,
      "grad_norm": 81.48680877685547,
      "learning_rate": 1.5312172770548216e-05,
      "loss": 6.1563,
      "step": 8474
    },
    {
      "epoch": 0.34214775938635444,
      "grad_norm": 103.6043701171875,
      "learning_rate": 1.5311064896446293e-05,
      "loss": 6.3175,
      "step": 8475
    },
    {
      "epoch": 0.3421881308033912,
      "grad_norm": 122.01641082763672,
      "learning_rate": 1.530995693153756e-05,
      "loss": 4.3767,
      "step": 8476
    },
    {
      "epoch": 0.34222850222042794,
      "grad_norm": 141.51400756835938,
      "learning_rate": 1.5308848875840964e-05,
      "loss": 4.6313,
      "step": 8477
    },
    {
      "epoch": 0.34226887363746467,
      "grad_norm": 212.3189697265625,
      "learning_rate": 1.5307740729375448e-05,
      "loss": 3.7785,
      "step": 8478
    },
    {
      "epoch": 0.3423092450545014,
      "grad_norm": 131.77195739746094,
      "learning_rate": 1.530663249215996e-05,
      "loss": 4.2265,
      "step": 8479
    },
    {
      "epoch": 0.34234961647153817,
      "grad_norm": 83.57185363769531,
      "learning_rate": 1.530552416421345e-05,
      "loss": 3.4274,
      "step": 8480
    },
    {
      "epoch": 0.3423899878885749,
      "grad_norm": 90.39282989501953,
      "learning_rate": 1.5304415745554866e-05,
      "loss": 6.3464,
      "step": 8481
    },
    {
      "epoch": 0.3424303593056116,
      "grad_norm": 128.1884002685547,
      "learning_rate": 1.530330723620316e-05,
      "loss": 5.605,
      "step": 8482
    },
    {
      "epoch": 0.3424707307226484,
      "grad_norm": 109.96416473388672,
      "learning_rate": 1.5302198636177283e-05,
      "loss": 5.4249,
      "step": 8483
    },
    {
      "epoch": 0.3425111021396851,
      "grad_norm": 197.17088317871094,
      "learning_rate": 1.5301089945496192e-05,
      "loss": 5.4678,
      "step": 8484
    },
    {
      "epoch": 0.34255147355672183,
      "grad_norm": 88.31071472167969,
      "learning_rate": 1.529998116417884e-05,
      "loss": 4.3647,
      "step": 8485
    },
    {
      "epoch": 0.34259184497375855,
      "grad_norm": 251.94903564453125,
      "learning_rate": 1.5298872292244185e-05,
      "loss": 6.7834,
      "step": 8486
    },
    {
      "epoch": 0.34263221639079533,
      "grad_norm": 108.86219787597656,
      "learning_rate": 1.5297763329711193e-05,
      "loss": 6.0862,
      "step": 8487
    },
    {
      "epoch": 0.34267258780783205,
      "grad_norm": 82.0390396118164,
      "learning_rate": 1.5296654276598815e-05,
      "loss": 4.899,
      "step": 8488
    },
    {
      "epoch": 0.3427129592248688,
      "grad_norm": 115.67002868652344,
      "learning_rate": 1.5295545132926018e-05,
      "loss": 6.1817,
      "step": 8489
    },
    {
      "epoch": 0.34275333064190555,
      "grad_norm": 77.2864990234375,
      "learning_rate": 1.529443589871177e-05,
      "loss": 4.4039,
      "step": 8490
    },
    {
      "epoch": 0.3427937020589423,
      "grad_norm": 114.14603424072266,
      "learning_rate": 1.5293326573975023e-05,
      "loss": 5.0114,
      "step": 8491
    },
    {
      "epoch": 0.342834073475979,
      "grad_norm": 111.29442596435547,
      "learning_rate": 1.5292217158734763e-05,
      "loss": 5.0155,
      "step": 8492
    },
    {
      "epoch": 0.3428744448930157,
      "grad_norm": 131.95709228515625,
      "learning_rate": 1.529110765300994e-05,
      "loss": 4.8557,
      "step": 8493
    },
    {
      "epoch": 0.3429148163100525,
      "grad_norm": 88.64203643798828,
      "learning_rate": 1.5289998056819535e-05,
      "loss": 4.106,
      "step": 8494
    },
    {
      "epoch": 0.3429551877270892,
      "grad_norm": 97.56938171386719,
      "learning_rate": 1.528888837018251e-05,
      "loss": 6.8724,
      "step": 8495
    },
    {
      "epoch": 0.34299555914412594,
      "grad_norm": 83.30854034423828,
      "learning_rate": 1.528777859311785e-05,
      "loss": 4.5637,
      "step": 8496
    },
    {
      "epoch": 0.3430359305611627,
      "grad_norm": 81.34432220458984,
      "learning_rate": 1.5286668725644523e-05,
      "loss": 6.5644,
      "step": 8497
    },
    {
      "epoch": 0.34307630197819944,
      "grad_norm": 117.0193862915039,
      "learning_rate": 1.5285558767781505e-05,
      "loss": 5.8101,
      "step": 8498
    },
    {
      "epoch": 0.34311667339523616,
      "grad_norm": 201.88592529296875,
      "learning_rate": 1.5284448719547777e-05,
      "loss": 4.7613,
      "step": 8499
    },
    {
      "epoch": 0.34315704481227294,
      "grad_norm": 172.92294311523438,
      "learning_rate": 1.5283338580962315e-05,
      "loss": 5.1808,
      "step": 8500
    },
    {
      "epoch": 0.34319741622930966,
      "grad_norm": 77.62358093261719,
      "learning_rate": 1.5282228352044102e-05,
      "loss": 6.4578,
      "step": 8501
    },
    {
      "epoch": 0.3432377876463464,
      "grad_norm": 100.6202163696289,
      "learning_rate": 1.5281118032812116e-05,
      "loss": 4.9833,
      "step": 8502
    },
    {
      "epoch": 0.3432781590633831,
      "grad_norm": 111.61825561523438,
      "learning_rate": 1.5280007623285347e-05,
      "loss": 6.2353,
      "step": 8503
    },
    {
      "epoch": 0.3433185304804199,
      "grad_norm": 104.92697143554688,
      "learning_rate": 1.5278897123482776e-05,
      "loss": 7.2896,
      "step": 8504
    },
    {
      "epoch": 0.3433589018974566,
      "grad_norm": 154.3993682861328,
      "learning_rate": 1.5277786533423393e-05,
      "loss": 4.8247,
      "step": 8505
    },
    {
      "epoch": 0.3433992733144933,
      "grad_norm": 101.87299346923828,
      "learning_rate": 1.527667585312618e-05,
      "loss": 5.1432,
      "step": 8506
    },
    {
      "epoch": 0.3434396447315301,
      "grad_norm": 157.71531677246094,
      "learning_rate": 1.527556508261013e-05,
      "loss": 6.7233,
      "step": 8507
    },
    {
      "epoch": 0.3434800161485668,
      "grad_norm": 99.9674301147461,
      "learning_rate": 1.5274454221894245e-05,
      "loss": 4.9781,
      "step": 8508
    },
    {
      "epoch": 0.34352038756560355,
      "grad_norm": 76.48567199707031,
      "learning_rate": 1.5273343270997508e-05,
      "loss": 6.0426,
      "step": 8509
    },
    {
      "epoch": 0.34356075898264027,
      "grad_norm": 193.07595825195312,
      "learning_rate": 1.527223222993891e-05,
      "loss": 5.172,
      "step": 8510
    },
    {
      "epoch": 0.34360113039967705,
      "grad_norm": 93.5719985961914,
      "learning_rate": 1.5271121098737456e-05,
      "loss": 5.6515,
      "step": 8511
    },
    {
      "epoch": 0.34364150181671377,
      "grad_norm": 60.6175422668457,
      "learning_rate": 1.527000987741214e-05,
      "loss": 4.934,
      "step": 8512
    },
    {
      "epoch": 0.3436818732337505,
      "grad_norm": 121.59712219238281,
      "learning_rate": 1.5268898565981964e-05,
      "loss": 5.1584,
      "step": 8513
    },
    {
      "epoch": 0.34372224465078727,
      "grad_norm": 138.249267578125,
      "learning_rate": 1.526778716446592e-05,
      "loss": 4.6302,
      "step": 8514
    },
    {
      "epoch": 0.343762616067824,
      "grad_norm": 83.85519409179688,
      "learning_rate": 1.526667567288302e-05,
      "loss": 6.5044,
      "step": 8515
    },
    {
      "epoch": 0.3438029874848607,
      "grad_norm": 126.48111724853516,
      "learning_rate": 1.5265564091252265e-05,
      "loss": 4.0881,
      "step": 8516
    },
    {
      "epoch": 0.34384335890189743,
      "grad_norm": 130.94102478027344,
      "learning_rate": 1.526445241959266e-05,
      "loss": 5.4833,
      "step": 8517
    },
    {
      "epoch": 0.3438837303189342,
      "grad_norm": 89.35798645019531,
      "learning_rate": 1.526334065792321e-05,
      "loss": 5.4523,
      "step": 8518
    },
    {
      "epoch": 0.34392410173597093,
      "grad_norm": 105.4628677368164,
      "learning_rate": 1.5262228806262925e-05,
      "loss": 7.191,
      "step": 8519
    },
    {
      "epoch": 0.34396447315300765,
      "grad_norm": 144.80906677246094,
      "learning_rate": 1.526111686463082e-05,
      "loss": 4.5851,
      "step": 8520
    },
    {
      "epoch": 0.34400484457004443,
      "grad_norm": 78.63787841796875,
      "learning_rate": 1.52600048330459e-05,
      "loss": 3.3285,
      "step": 8521
    },
    {
      "epoch": 0.34404521598708115,
      "grad_norm": 79.8327407836914,
      "learning_rate": 1.525889271152718e-05,
      "loss": 4.9074,
      "step": 8522
    },
    {
      "epoch": 0.3440855874041179,
      "grad_norm": 114.35507202148438,
      "learning_rate": 1.5257780500093675e-05,
      "loss": 6.1063,
      "step": 8523
    },
    {
      "epoch": 0.3441259588211546,
      "grad_norm": 140.8229217529297,
      "learning_rate": 1.5256668198764404e-05,
      "loss": 5.8001,
      "step": 8524
    },
    {
      "epoch": 0.3441663302381914,
      "grad_norm": 132.18350219726562,
      "learning_rate": 1.5255555807558378e-05,
      "loss": 5.8664,
      "step": 8525
    },
    {
      "epoch": 0.3442067016552281,
      "grad_norm": 146.91921997070312,
      "learning_rate": 1.5254443326494621e-05,
      "loss": 5.3952,
      "step": 8526
    },
    {
      "epoch": 0.3442470730722648,
      "grad_norm": 150.01016235351562,
      "learning_rate": 1.5253330755592153e-05,
      "loss": 4.2642,
      "step": 8527
    },
    {
      "epoch": 0.3442874444893016,
      "grad_norm": 60.54850387573242,
      "learning_rate": 1.525221809487e-05,
      "loss": 5.0533,
      "step": 8528
    },
    {
      "epoch": 0.3443278159063383,
      "grad_norm": 165.54425048828125,
      "learning_rate": 1.5251105344347176e-05,
      "loss": 5.9931,
      "step": 8529
    },
    {
      "epoch": 0.34436818732337504,
      "grad_norm": 95.23164367675781,
      "learning_rate": 1.5249992504042718e-05,
      "loss": 4.9381,
      "step": 8530
    },
    {
      "epoch": 0.34440855874041176,
      "grad_norm": 106.70299530029297,
      "learning_rate": 1.5248879573975646e-05,
      "loss": 6.0047,
      "step": 8531
    },
    {
      "epoch": 0.34444893015744854,
      "grad_norm": 73.23660278320312,
      "learning_rate": 1.524776655416499e-05,
      "loss": 6.1141,
      "step": 8532
    },
    {
      "epoch": 0.34448930157448526,
      "grad_norm": 103.57534790039062,
      "learning_rate": 1.5246653444629782e-05,
      "loss": 6.2529,
      "step": 8533
    },
    {
      "epoch": 0.344529672991522,
      "grad_norm": 126.04393005371094,
      "learning_rate": 1.5245540245389053e-05,
      "loss": 5.7601,
      "step": 8534
    },
    {
      "epoch": 0.34457004440855876,
      "grad_norm": 72.51183319091797,
      "learning_rate": 1.524442695646183e-05,
      "loss": 4.1677,
      "step": 8535
    },
    {
      "epoch": 0.3446104158255955,
      "grad_norm": 78.23218536376953,
      "learning_rate": 1.5243313577867157e-05,
      "loss": 5.2765,
      "step": 8536
    },
    {
      "epoch": 0.3446507872426322,
      "grad_norm": 94.37056732177734,
      "learning_rate": 1.5242200109624064e-05,
      "loss": 6.8816,
      "step": 8537
    },
    {
      "epoch": 0.34469115865966893,
      "grad_norm": 88.23895263671875,
      "learning_rate": 1.5241086551751592e-05,
      "loss": 6.2183,
      "step": 8538
    },
    {
      "epoch": 0.3447315300767057,
      "grad_norm": 122.60082244873047,
      "learning_rate": 1.5239972904268776e-05,
      "loss": 4.8358,
      "step": 8539
    },
    {
      "epoch": 0.3447719014937424,
      "grad_norm": 89.71009826660156,
      "learning_rate": 1.5238859167194661e-05,
      "loss": 6.123,
      "step": 8540
    },
    {
      "epoch": 0.34481227291077915,
      "grad_norm": 151.1520538330078,
      "learning_rate": 1.5237745340548288e-05,
      "loss": 5.3541,
      "step": 8541
    },
    {
      "epoch": 0.3448526443278159,
      "grad_norm": 78.70552062988281,
      "learning_rate": 1.5236631424348703e-05,
      "loss": 6.3821,
      "step": 8542
    },
    {
      "epoch": 0.34489301574485265,
      "grad_norm": 266.5612487792969,
      "learning_rate": 1.5235517418614946e-05,
      "loss": 6.6111,
      "step": 8543
    },
    {
      "epoch": 0.34493338716188937,
      "grad_norm": 158.29837036132812,
      "learning_rate": 1.5234403323366069e-05,
      "loss": 5.1329,
      "step": 8544
    },
    {
      "epoch": 0.3449737585789261,
      "grad_norm": 131.1771240234375,
      "learning_rate": 1.5233289138621116e-05,
      "loss": 4.9983,
      "step": 8545
    },
    {
      "epoch": 0.34501412999596287,
      "grad_norm": 114.48736572265625,
      "learning_rate": 1.523217486439914e-05,
      "loss": 3.8787,
      "step": 8546
    },
    {
      "epoch": 0.3450545014129996,
      "grad_norm": 106.31043243408203,
      "learning_rate": 1.5231060500719193e-05,
      "loss": 5.2593,
      "step": 8547
    },
    {
      "epoch": 0.3450948728300363,
      "grad_norm": 113.78414916992188,
      "learning_rate": 1.5229946047600326e-05,
      "loss": 5.044,
      "step": 8548
    },
    {
      "epoch": 0.3451352442470731,
      "grad_norm": 73.73387908935547,
      "learning_rate": 1.5228831505061595e-05,
      "loss": 5.416,
      "step": 8549
    },
    {
      "epoch": 0.3451756156641098,
      "grad_norm": 106.14337158203125,
      "learning_rate": 1.5227716873122055e-05,
      "loss": 5.7011,
      "step": 8550
    },
    {
      "epoch": 0.34521598708114654,
      "grad_norm": 73.6043472290039,
      "learning_rate": 1.5226602151800764e-05,
      "loss": 4.5723,
      "step": 8551
    },
    {
      "epoch": 0.3452563584981833,
      "grad_norm": 76.561279296875,
      "learning_rate": 1.5225487341116783e-05,
      "loss": 5.5513,
      "step": 8552
    },
    {
      "epoch": 0.34529672991522004,
      "grad_norm": 87.41860961914062,
      "learning_rate": 1.5224372441089168e-05,
      "loss": 6.1155,
      "step": 8553
    },
    {
      "epoch": 0.34533710133225676,
      "grad_norm": 109.84825134277344,
      "learning_rate": 1.5223257451736988e-05,
      "loss": 5.5414,
      "step": 8554
    },
    {
      "epoch": 0.3453774727492935,
      "grad_norm": 78.30438995361328,
      "learning_rate": 1.5222142373079298e-05,
      "loss": 5.5093,
      "step": 8555
    },
    {
      "epoch": 0.34541784416633026,
      "grad_norm": 66.76964569091797,
      "learning_rate": 1.5221027205135169e-05,
      "loss": 6.9254,
      "step": 8556
    },
    {
      "epoch": 0.345458215583367,
      "grad_norm": 207.05105590820312,
      "learning_rate": 1.5219911947923667e-05,
      "loss": 5.367,
      "step": 8557
    },
    {
      "epoch": 0.3454985870004037,
      "grad_norm": 204.68507385253906,
      "learning_rate": 1.5218796601463864e-05,
      "loss": 4.8696,
      "step": 8558
    },
    {
      "epoch": 0.3455389584174405,
      "grad_norm": 102.4618911743164,
      "learning_rate": 1.521768116577482e-05,
      "loss": 5.1216,
      "step": 8559
    },
    {
      "epoch": 0.3455793298344772,
      "grad_norm": 94.33433532714844,
      "learning_rate": 1.5216565640875618e-05,
      "loss": 4.8412,
      "step": 8560
    },
    {
      "epoch": 0.3456197012515139,
      "grad_norm": 115.89593505859375,
      "learning_rate": 1.5215450026785318e-05,
      "loss": 4.2594,
      "step": 8561
    },
    {
      "epoch": 0.34566007266855064,
      "grad_norm": 166.90892028808594,
      "learning_rate": 1.5214334323523008e-05,
      "loss": 6.7214,
      "step": 8562
    },
    {
      "epoch": 0.3457004440855874,
      "grad_norm": 86.92699432373047,
      "learning_rate": 1.5213218531107753e-05,
      "loss": 4.7677,
      "step": 8563
    },
    {
      "epoch": 0.34574081550262414,
      "grad_norm": 71.7904052734375,
      "learning_rate": 1.5212102649558637e-05,
      "loss": 4.6533,
      "step": 8564
    },
    {
      "epoch": 0.34578118691966087,
      "grad_norm": 231.2840118408203,
      "learning_rate": 1.521098667889474e-05,
      "loss": 3.9378,
      "step": 8565
    },
    {
      "epoch": 0.34582155833669764,
      "grad_norm": 84.05272674560547,
      "learning_rate": 1.5209870619135134e-05,
      "loss": 6.7253,
      "step": 8566
    },
    {
      "epoch": 0.34586192975373437,
      "grad_norm": 109.4698715209961,
      "learning_rate": 1.5208754470298908e-05,
      "loss": 6.7315,
      "step": 8567
    },
    {
      "epoch": 0.3459023011707711,
      "grad_norm": 96.11783599853516,
      "learning_rate": 1.5207638232405147e-05,
      "loss": 4.1524,
      "step": 8568
    },
    {
      "epoch": 0.3459426725878078,
      "grad_norm": 101.135498046875,
      "learning_rate": 1.5206521905472928e-05,
      "loss": 5.8855,
      "step": 8569
    },
    {
      "epoch": 0.3459830440048446,
      "grad_norm": 85.82677459716797,
      "learning_rate": 1.5205405489521345e-05,
      "loss": 4.4557,
      "step": 8570
    },
    {
      "epoch": 0.3460234154218813,
      "grad_norm": 79.27720642089844,
      "learning_rate": 1.5204288984569482e-05,
      "loss": 4.2758,
      "step": 8571
    },
    {
      "epoch": 0.34606378683891803,
      "grad_norm": 128.88043212890625,
      "learning_rate": 1.5203172390636432e-05,
      "loss": 4.2516,
      "step": 8572
    },
    {
      "epoch": 0.3461041582559548,
      "grad_norm": 80.30848693847656,
      "learning_rate": 1.5202055707741283e-05,
      "loss": 5.4132,
      "step": 8573
    },
    {
      "epoch": 0.34614452967299153,
      "grad_norm": 123.0703125,
      "learning_rate": 1.5200938935903132e-05,
      "loss": 4.1103,
      "step": 8574
    },
    {
      "epoch": 0.34618490109002825,
      "grad_norm": 81.03993225097656,
      "learning_rate": 1.5199822075141067e-05,
      "loss": 5.2847,
      "step": 8575
    },
    {
      "epoch": 0.346225272507065,
      "grad_norm": 90.1724624633789,
      "learning_rate": 1.519870512547419e-05,
      "loss": 5.5016,
      "step": 8576
    },
    {
      "epoch": 0.34626564392410175,
      "grad_norm": 127.75923156738281,
      "learning_rate": 1.5197588086921591e-05,
      "loss": 5.9192,
      "step": 8577
    },
    {
      "epoch": 0.3463060153411385,
      "grad_norm": 85.50880432128906,
      "learning_rate": 1.5196470959502376e-05,
      "loss": 6.9832,
      "step": 8578
    },
    {
      "epoch": 0.3463463867581752,
      "grad_norm": 65.35576629638672,
      "learning_rate": 1.5195353743235641e-05,
      "loss": 3.9587,
      "step": 8579
    },
    {
      "epoch": 0.346386758175212,
      "grad_norm": 90.32763671875,
      "learning_rate": 1.519423643814049e-05,
      "loss": 5.852,
      "step": 8580
    },
    {
      "epoch": 0.3464271295922487,
      "grad_norm": 87.66633605957031,
      "learning_rate": 1.5193119044236024e-05,
      "loss": 3.1572,
      "step": 8581
    },
    {
      "epoch": 0.3464675010092854,
      "grad_norm": 101.7297592163086,
      "learning_rate": 1.5192001561541352e-05,
      "loss": 5.5121,
      "step": 8582
    },
    {
      "epoch": 0.34650787242632214,
      "grad_norm": 113.24867248535156,
      "learning_rate": 1.5190883990075576e-05,
      "loss": 6.8458,
      "step": 8583
    },
    {
      "epoch": 0.3465482438433589,
      "grad_norm": 227.7669219970703,
      "learning_rate": 1.5189766329857803e-05,
      "loss": 5.681,
      "step": 8584
    },
    {
      "epoch": 0.34658861526039564,
      "grad_norm": 108.31494140625,
      "learning_rate": 1.5188648580907145e-05,
      "loss": 5.491,
      "step": 8585
    },
    {
      "epoch": 0.34662898667743236,
      "grad_norm": 101.10933685302734,
      "learning_rate": 1.5187530743242716e-05,
      "loss": 4.0853,
      "step": 8586
    },
    {
      "epoch": 0.34666935809446914,
      "grad_norm": 77.01107025146484,
      "learning_rate": 1.5186412816883623e-05,
      "loss": 4.4906,
      "step": 8587
    },
    {
      "epoch": 0.34670972951150586,
      "grad_norm": 122.07169342041016,
      "learning_rate": 1.5185294801848985e-05,
      "loss": 5.5215,
      "step": 8588
    },
    {
      "epoch": 0.3467501009285426,
      "grad_norm": 100.2854995727539,
      "learning_rate": 1.5184176698157908e-05,
      "loss": 4.4102,
      "step": 8589
    },
    {
      "epoch": 0.3467904723455793,
      "grad_norm": 111.51107025146484,
      "learning_rate": 1.5183058505829521e-05,
      "loss": 4.1725,
      "step": 8590
    },
    {
      "epoch": 0.3468308437626161,
      "grad_norm": 119.45008087158203,
      "learning_rate": 1.5181940224882933e-05,
      "loss": 4.6473,
      "step": 8591
    },
    {
      "epoch": 0.3468712151796528,
      "grad_norm": 122.22504425048828,
      "learning_rate": 1.5180821855337273e-05,
      "loss": 4.9369,
      "step": 8592
    },
    {
      "epoch": 0.3469115865966895,
      "grad_norm": 97.35787200927734,
      "learning_rate": 1.5179703397211653e-05,
      "loss": 4.194,
      "step": 8593
    },
    {
      "epoch": 0.3469519580137263,
      "grad_norm": 78.46851348876953,
      "learning_rate": 1.5178584850525202e-05,
      "loss": 5.3869,
      "step": 8594
    },
    {
      "epoch": 0.346992329430763,
      "grad_norm": 112.10061645507812,
      "learning_rate": 1.517746621529704e-05,
      "loss": 6.4741,
      "step": 8595
    },
    {
      "epoch": 0.34703270084779975,
      "grad_norm": 112.74246215820312,
      "learning_rate": 1.5176347491546302e-05,
      "loss": 6.07,
      "step": 8596
    },
    {
      "epoch": 0.34707307226483647,
      "grad_norm": 88.71807861328125,
      "learning_rate": 1.5175228679292104e-05,
      "loss": 5.1402,
      "step": 8597
    },
    {
      "epoch": 0.34711344368187325,
      "grad_norm": 112.99195098876953,
      "learning_rate": 1.5174109778553585e-05,
      "loss": 4.9866,
      "step": 8598
    },
    {
      "epoch": 0.34715381509890997,
      "grad_norm": 96.68598937988281,
      "learning_rate": 1.5172990789349866e-05,
      "loss": 4.3277,
      "step": 8599
    },
    {
      "epoch": 0.3471941865159467,
      "grad_norm": 82.01419067382812,
      "learning_rate": 1.5171871711700088e-05,
      "loss": 4.5552,
      "step": 8600
    },
    {
      "epoch": 0.34723455793298347,
      "grad_norm": 94.8270492553711,
      "learning_rate": 1.5170752545623382e-05,
      "loss": 4.7823,
      "step": 8601
    },
    {
      "epoch": 0.3472749293500202,
      "grad_norm": 103.77213287353516,
      "learning_rate": 1.516963329113888e-05,
      "loss": 3.4325,
      "step": 8602
    },
    {
      "epoch": 0.3473153007670569,
      "grad_norm": 143.3491668701172,
      "learning_rate": 1.516851394826572e-05,
      "loss": 4.4723,
      "step": 8603
    },
    {
      "epoch": 0.3473556721840937,
      "grad_norm": 83.06920623779297,
      "learning_rate": 1.516739451702304e-05,
      "loss": 7.0311,
      "step": 8604
    },
    {
      "epoch": 0.3473960436011304,
      "grad_norm": 74.50531768798828,
      "learning_rate": 1.5166274997429984e-05,
      "loss": 4.385,
      "step": 8605
    },
    {
      "epoch": 0.34743641501816713,
      "grad_norm": 772.2489624023438,
      "learning_rate": 1.5165155389505689e-05,
      "loss": 4.3369,
      "step": 8606
    },
    {
      "epoch": 0.34747678643520385,
      "grad_norm": 101.76385498046875,
      "learning_rate": 1.5164035693269294e-05,
      "loss": 6.0637,
      "step": 8607
    },
    {
      "epoch": 0.34751715785224063,
      "grad_norm": 106.10472869873047,
      "learning_rate": 1.5162915908739953e-05,
      "loss": 2.9386,
      "step": 8608
    },
    {
      "epoch": 0.34755752926927735,
      "grad_norm": 71.4064712524414,
      "learning_rate": 1.5161796035936803e-05,
      "loss": 6.0076,
      "step": 8609
    },
    {
      "epoch": 0.3475979006863141,
      "grad_norm": 103.82524108886719,
      "learning_rate": 1.5160676074878996e-05,
      "loss": 6.7075,
      "step": 8610
    },
    {
      "epoch": 0.34763827210335085,
      "grad_norm": 99.44961547851562,
      "learning_rate": 1.5159556025585678e-05,
      "loss": 5.3621,
      "step": 8611
    },
    {
      "epoch": 0.3476786435203876,
      "grad_norm": 194.90354919433594,
      "learning_rate": 1.5158435888076001e-05,
      "loss": 7.0117,
      "step": 8612
    },
    {
      "epoch": 0.3477190149374243,
      "grad_norm": 90.57880401611328,
      "learning_rate": 1.5157315662369117e-05,
      "loss": 5.4266,
      "step": 8613
    },
    {
      "epoch": 0.347759386354461,
      "grad_norm": 82.23604583740234,
      "learning_rate": 1.5156195348484178e-05,
      "loss": 4.8324,
      "step": 8614
    },
    {
      "epoch": 0.3477997577714978,
      "grad_norm": 78.23004913330078,
      "learning_rate": 1.5155074946440336e-05,
      "loss": 5.2677,
      "step": 8615
    },
    {
      "epoch": 0.3478401291885345,
      "grad_norm": 112.9850845336914,
      "learning_rate": 1.5153954456256754e-05,
      "loss": 6.6215,
      "step": 8616
    },
    {
      "epoch": 0.34788050060557124,
      "grad_norm": 117.14341735839844,
      "learning_rate": 1.5152833877952584e-05,
      "loss": 6.7341,
      "step": 8617
    },
    {
      "epoch": 0.347920872022608,
      "grad_norm": 72.58382415771484,
      "learning_rate": 1.515171321154699e-05,
      "loss": 6.9172,
      "step": 8618
    },
    {
      "epoch": 0.34796124343964474,
      "grad_norm": 98.69014739990234,
      "learning_rate": 1.5150592457059127e-05,
      "loss": 6.4915,
      "step": 8619
    },
    {
      "epoch": 0.34800161485668146,
      "grad_norm": 101.31795501708984,
      "learning_rate": 1.5149471614508161e-05,
      "loss": 3.1275,
      "step": 8620
    },
    {
      "epoch": 0.3480419862737182,
      "grad_norm": 137.44656372070312,
      "learning_rate": 1.5148350683913257e-05,
      "loss": 5.1978,
      "step": 8621
    },
    {
      "epoch": 0.34808235769075496,
      "grad_norm": 143.10552978515625,
      "learning_rate": 1.5147229665293579e-05,
      "loss": 5.8475,
      "step": 8622
    },
    {
      "epoch": 0.3481227291077917,
      "grad_norm": 78.97909545898438,
      "learning_rate": 1.5146108558668293e-05,
      "loss": 5.7262,
      "step": 8623
    },
    {
      "epoch": 0.3481631005248284,
      "grad_norm": 82.24567413330078,
      "learning_rate": 1.5144987364056568e-05,
      "loss": 6.7188,
      "step": 8624
    },
    {
      "epoch": 0.3482034719418652,
      "grad_norm": 89.62381744384766,
      "learning_rate": 1.5143866081477572e-05,
      "loss": 5.1135,
      "step": 8625
    },
    {
      "epoch": 0.3482438433589019,
      "grad_norm": 66.61176300048828,
      "learning_rate": 1.5142744710950479e-05,
      "loss": 3.868,
      "step": 8626
    },
    {
      "epoch": 0.3482842147759386,
      "grad_norm": 72.33758544921875,
      "learning_rate": 1.5141623252494459e-05,
      "loss": 4.5226,
      "step": 8627
    },
    {
      "epoch": 0.34832458619297535,
      "grad_norm": 85.93387603759766,
      "learning_rate": 1.5140501706128689e-05,
      "loss": 6.2135,
      "step": 8628
    },
    {
      "epoch": 0.3483649576100121,
      "grad_norm": 116.02677154541016,
      "learning_rate": 1.5139380071872343e-05,
      "loss": 5.4923,
      "step": 8629
    },
    {
      "epoch": 0.34840532902704885,
      "grad_norm": 80.85746765136719,
      "learning_rate": 1.51382583497446e-05,
      "loss": 7.1342,
      "step": 8630
    },
    {
      "epoch": 0.34844570044408557,
      "grad_norm": 105.58393096923828,
      "learning_rate": 1.5137136539764639e-05,
      "loss": 5.1558,
      "step": 8631
    },
    {
      "epoch": 0.34848607186112235,
      "grad_norm": 111.11095428466797,
      "learning_rate": 1.5136014641951638e-05,
      "loss": 4.5006,
      "step": 8632
    },
    {
      "epoch": 0.34852644327815907,
      "grad_norm": 62.241241455078125,
      "learning_rate": 1.5134892656324778e-05,
      "loss": 5.1527,
      "step": 8633
    },
    {
      "epoch": 0.3485668146951958,
      "grad_norm": 84.72814178466797,
      "learning_rate": 1.5133770582903246e-05,
      "loss": 5.6744,
      "step": 8634
    },
    {
      "epoch": 0.3486071861122325,
      "grad_norm": 199.10325622558594,
      "learning_rate": 1.5132648421706226e-05,
      "loss": 5.8828,
      "step": 8635
    },
    {
      "epoch": 0.3486475575292693,
      "grad_norm": 97.48018646240234,
      "learning_rate": 1.5131526172752903e-05,
      "loss": 5.0844,
      "step": 8636
    },
    {
      "epoch": 0.348687928946306,
      "grad_norm": 98.27706909179688,
      "learning_rate": 1.5130403836062466e-05,
      "loss": 5.7988,
      "step": 8637
    },
    {
      "epoch": 0.34872830036334274,
      "grad_norm": 135.76649475097656,
      "learning_rate": 1.5129281411654105e-05,
      "loss": 4.9738,
      "step": 8638
    },
    {
      "epoch": 0.3487686717803795,
      "grad_norm": 78.84520721435547,
      "learning_rate": 1.5128158899547005e-05,
      "loss": 4.7835,
      "step": 8639
    },
    {
      "epoch": 0.34880904319741624,
      "grad_norm": 87.74372100830078,
      "learning_rate": 1.5127036299760369e-05,
      "loss": 6.2271,
      "step": 8640
    },
    {
      "epoch": 0.34884941461445296,
      "grad_norm": 88.79778289794922,
      "learning_rate": 1.512591361231338e-05,
      "loss": 4.1285,
      "step": 8641
    },
    {
      "epoch": 0.3488897860314897,
      "grad_norm": 87.85297393798828,
      "learning_rate": 1.512479083722524e-05,
      "loss": 6.0577,
      "step": 8642
    },
    {
      "epoch": 0.34893015744852646,
      "grad_norm": 86.12171936035156,
      "learning_rate": 1.5123667974515143e-05,
      "loss": 5.3876,
      "step": 8643
    },
    {
      "epoch": 0.3489705288655632,
      "grad_norm": 94.16752624511719,
      "learning_rate": 1.5122545024202289e-05,
      "loss": 4.9801,
      "step": 8644
    },
    {
      "epoch": 0.3490109002825999,
      "grad_norm": 59.88206100463867,
      "learning_rate": 1.5121421986305875e-05,
      "loss": 2.8896,
      "step": 8645
    },
    {
      "epoch": 0.3490512716996367,
      "grad_norm": 108.12478637695312,
      "learning_rate": 1.5120298860845108e-05,
      "loss": 4.4693,
      "step": 8646
    },
    {
      "epoch": 0.3490916431166734,
      "grad_norm": 74.7798843383789,
      "learning_rate": 1.5119175647839186e-05,
      "loss": 4.1359,
      "step": 8647
    },
    {
      "epoch": 0.3491320145337101,
      "grad_norm": 91.17086791992188,
      "learning_rate": 1.5118052347307316e-05,
      "loss": 5.0921,
      "step": 8648
    },
    {
      "epoch": 0.34917238595074684,
      "grad_norm": 94.96548461914062,
      "learning_rate": 1.51169289592687e-05,
      "loss": 6.2959,
      "step": 8649
    },
    {
      "epoch": 0.3492127573677836,
      "grad_norm": 70.3574447631836,
      "learning_rate": 1.511580548374255e-05,
      "loss": 6.0253,
      "step": 8650
    },
    {
      "epoch": 0.34925312878482034,
      "grad_norm": 79.15204620361328,
      "learning_rate": 1.5114681920748069e-05,
      "loss": 5.8301,
      "step": 8651
    },
    {
      "epoch": 0.34929350020185707,
      "grad_norm": 119.79074096679688,
      "learning_rate": 1.5113558270304474e-05,
      "loss": 4.9667,
      "step": 8652
    },
    {
      "epoch": 0.34933387161889384,
      "grad_norm": 72.37432861328125,
      "learning_rate": 1.5112434532430971e-05,
      "loss": 5.6136,
      "step": 8653
    },
    {
      "epoch": 0.34937424303593057,
      "grad_norm": 86.02120971679688,
      "learning_rate": 1.5111310707146775e-05,
      "loss": 5.6874,
      "step": 8654
    },
    {
      "epoch": 0.3494146144529673,
      "grad_norm": 68.52619934082031,
      "learning_rate": 1.5110186794471105e-05,
      "loss": 5.257,
      "step": 8655
    },
    {
      "epoch": 0.34945498587000406,
      "grad_norm": 89.34487915039062,
      "learning_rate": 1.5109062794423172e-05,
      "loss": 2.9858,
      "step": 8656
    },
    {
      "epoch": 0.3494953572870408,
      "grad_norm": 66.47618103027344,
      "learning_rate": 1.5107938707022198e-05,
      "loss": 7.0164,
      "step": 8657
    },
    {
      "epoch": 0.3495357287040775,
      "grad_norm": 118.54080200195312,
      "learning_rate": 1.5106814532287399e-05,
      "loss": 6.1135,
      "step": 8658
    },
    {
      "epoch": 0.34957610012111423,
      "grad_norm": 81.44886779785156,
      "learning_rate": 1.5105690270237996e-05,
      "loss": 6.3514,
      "step": 8659
    },
    {
      "epoch": 0.349616471538151,
      "grad_norm": 104.77627563476562,
      "learning_rate": 1.5104565920893213e-05,
      "loss": 5.1487,
      "step": 8660
    },
    {
      "epoch": 0.34965684295518773,
      "grad_norm": 127.378662109375,
      "learning_rate": 1.510344148427227e-05,
      "loss": 5.2378,
      "step": 8661
    },
    {
      "epoch": 0.34969721437222445,
      "grad_norm": 104.7506332397461,
      "learning_rate": 1.5102316960394399e-05,
      "loss": 4.9775,
      "step": 8662
    },
    {
      "epoch": 0.34973758578926123,
      "grad_norm": 164.84580993652344,
      "learning_rate": 1.510119234927882e-05,
      "loss": 4.2964,
      "step": 8663
    },
    {
      "epoch": 0.34977795720629795,
      "grad_norm": 104.34491729736328,
      "learning_rate": 1.5100067650944765e-05,
      "loss": 5.4793,
      "step": 8664
    },
    {
      "epoch": 0.3498183286233347,
      "grad_norm": 121.95703125,
      "learning_rate": 1.5098942865411462e-05,
      "loss": 7.2674,
      "step": 8665
    },
    {
      "epoch": 0.3498587000403714,
      "grad_norm": 155.64280700683594,
      "learning_rate": 1.5097817992698143e-05,
      "loss": 6.5875,
      "step": 8666
    },
    {
      "epoch": 0.3498990714574082,
      "grad_norm": 128.8512725830078,
      "learning_rate": 1.509669303282404e-05,
      "loss": 5.7865,
      "step": 8667
    },
    {
      "epoch": 0.3499394428744449,
      "grad_norm": 89.88139343261719,
      "learning_rate": 1.5095567985808389e-05,
      "loss": 5.499,
      "step": 8668
    },
    {
      "epoch": 0.3499798142914816,
      "grad_norm": 69.86556243896484,
      "learning_rate": 1.5094442851670423e-05,
      "loss": 5.7316,
      "step": 8669
    },
    {
      "epoch": 0.3500201857085184,
      "grad_norm": 90.61079406738281,
      "learning_rate": 1.5093317630429382e-05,
      "loss": 6.5589,
      "step": 8670
    },
    {
      "epoch": 0.3500605571255551,
      "grad_norm": 124.09698486328125,
      "learning_rate": 1.5092192322104504e-05,
      "loss": 6.9083,
      "step": 8671
    },
    {
      "epoch": 0.35010092854259184,
      "grad_norm": 93.41219329833984,
      "learning_rate": 1.5091066926715026e-05,
      "loss": 4.6166,
      "step": 8672
    },
    {
      "epoch": 0.35014129995962856,
      "grad_norm": 72.640380859375,
      "learning_rate": 1.5089941444280194e-05,
      "loss": 5.6248,
      "step": 8673
    },
    {
      "epoch": 0.35018167137666534,
      "grad_norm": 124.62055206298828,
      "learning_rate": 1.508881587481925e-05,
      "loss": 5.5837,
      "step": 8674
    },
    {
      "epoch": 0.35022204279370206,
      "grad_norm": 122.5328598022461,
      "learning_rate": 1.5087690218351437e-05,
      "loss": 3.608,
      "step": 8675
    },
    {
      "epoch": 0.3502624142107388,
      "grad_norm": 103.60785675048828,
      "learning_rate": 1.5086564474896e-05,
      "loss": 4.7797,
      "step": 8676
    },
    {
      "epoch": 0.35030278562777556,
      "grad_norm": 124.64019012451172,
      "learning_rate": 1.508543864447219e-05,
      "loss": 5.5356,
      "step": 8677
    },
    {
      "epoch": 0.3503431570448123,
      "grad_norm": 72.1156234741211,
      "learning_rate": 1.5084312727099255e-05,
      "loss": 5.6143,
      "step": 8678
    },
    {
      "epoch": 0.350383528461849,
      "grad_norm": 70.81604766845703,
      "learning_rate": 1.5083186722796443e-05,
      "loss": 4.6027,
      "step": 8679
    },
    {
      "epoch": 0.3504238998788857,
      "grad_norm": 96.3320083618164,
      "learning_rate": 1.508206063158301e-05,
      "loss": 5.6446,
      "step": 8680
    },
    {
      "epoch": 0.3504642712959225,
      "grad_norm": 63.938507080078125,
      "learning_rate": 1.5080934453478207e-05,
      "loss": 4.2375,
      "step": 8681
    },
    {
      "epoch": 0.3505046427129592,
      "grad_norm": 128.0328826904297,
      "learning_rate": 1.507980818850129e-05,
      "loss": 5.6989,
      "step": 8682
    },
    {
      "epoch": 0.35054501412999595,
      "grad_norm": 106.43314361572266,
      "learning_rate": 1.5078681836671513e-05,
      "loss": 5.5135,
      "step": 8683
    },
    {
      "epoch": 0.3505853855470327,
      "grad_norm": 107.8160400390625,
      "learning_rate": 1.5077555398008139e-05,
      "loss": 5.1581,
      "step": 8684
    },
    {
      "epoch": 0.35062575696406945,
      "grad_norm": 138.09869384765625,
      "learning_rate": 1.5076428872530425e-05,
      "loss": 4.4571,
      "step": 8685
    },
    {
      "epoch": 0.35066612838110617,
      "grad_norm": 84.50794219970703,
      "learning_rate": 1.5075302260257632e-05,
      "loss": 4.4207,
      "step": 8686
    },
    {
      "epoch": 0.3507064997981429,
      "grad_norm": 135.4296875,
      "learning_rate": 1.5074175561209018e-05,
      "loss": 3.8836,
      "step": 8687
    },
    {
      "epoch": 0.35074687121517967,
      "grad_norm": 120.73599243164062,
      "learning_rate": 1.5073048775403856e-05,
      "loss": 7.4361,
      "step": 8688
    },
    {
      "epoch": 0.3507872426322164,
      "grad_norm": 85.71952819824219,
      "learning_rate": 1.5071921902861402e-05,
      "loss": 5.4212,
      "step": 8689
    },
    {
      "epoch": 0.3508276140492531,
      "grad_norm": 105.35398864746094,
      "learning_rate": 1.507079494360093e-05,
      "loss": 5.1216,
      "step": 8690
    },
    {
      "epoch": 0.3508679854662899,
      "grad_norm": 83.00192260742188,
      "learning_rate": 1.5069667897641704e-05,
      "loss": 5.1472,
      "step": 8691
    },
    {
      "epoch": 0.3509083568833266,
      "grad_norm": 133.2711181640625,
      "learning_rate": 1.5068540765002996e-05,
      "loss": 6.1575,
      "step": 8692
    },
    {
      "epoch": 0.35094872830036333,
      "grad_norm": 87.3780517578125,
      "learning_rate": 1.5067413545704078e-05,
      "loss": 4.5657,
      "step": 8693
    },
    {
      "epoch": 0.35098909971740005,
      "grad_norm": 84.94515991210938,
      "learning_rate": 1.506628623976422e-05,
      "loss": 5.6105,
      "step": 8694
    },
    {
      "epoch": 0.35102947113443683,
      "grad_norm": 92.14506530761719,
      "learning_rate": 1.5065158847202699e-05,
      "loss": 5.1181,
      "step": 8695
    },
    {
      "epoch": 0.35106984255147355,
      "grad_norm": 72.9109878540039,
      "learning_rate": 1.506403136803879e-05,
      "loss": 6.2405,
      "step": 8696
    },
    {
      "epoch": 0.3511102139685103,
      "grad_norm": 143.37091064453125,
      "learning_rate": 1.5062903802291767e-05,
      "loss": 4.983,
      "step": 8697
    },
    {
      "epoch": 0.35115058538554705,
      "grad_norm": 84.29902648925781,
      "learning_rate": 1.5061776149980915e-05,
      "loss": 5.7279,
      "step": 8698
    },
    {
      "epoch": 0.3511909568025838,
      "grad_norm": 120.19451141357422,
      "learning_rate": 1.5060648411125509e-05,
      "loss": 5.9317,
      "step": 8699
    },
    {
      "epoch": 0.3512313282196205,
      "grad_norm": 106.05504608154297,
      "learning_rate": 1.5059520585744835e-05,
      "loss": 4.7156,
      "step": 8700
    },
    {
      "epoch": 0.3512716996366572,
      "grad_norm": 111.5919418334961,
      "learning_rate": 1.505839267385817e-05,
      "loss": 5.0321,
      "step": 8701
    },
    {
      "epoch": 0.351312071053694,
      "grad_norm": 88.85112762451172,
      "learning_rate": 1.5057264675484807e-05,
      "loss": 4.7663,
      "step": 8702
    },
    {
      "epoch": 0.3513524424707307,
      "grad_norm": 103.31111907958984,
      "learning_rate": 1.5056136590644024e-05,
      "loss": 6.1421,
      "step": 8703
    },
    {
      "epoch": 0.35139281388776744,
      "grad_norm": 100.23829650878906,
      "learning_rate": 1.5055008419355113e-05,
      "loss": 6.8077,
      "step": 8704
    },
    {
      "epoch": 0.3514331853048042,
      "grad_norm": 78.27212524414062,
      "learning_rate": 1.5053880161637361e-05,
      "loss": 5.0444,
      "step": 8705
    },
    {
      "epoch": 0.35147355672184094,
      "grad_norm": 121.91010284423828,
      "learning_rate": 1.5052751817510065e-05,
      "loss": 7.1177,
      "step": 8706
    },
    {
      "epoch": 0.35151392813887766,
      "grad_norm": 124.87451934814453,
      "learning_rate": 1.5051623386992509e-05,
      "loss": 5.2405,
      "step": 8707
    },
    {
      "epoch": 0.35155429955591444,
      "grad_norm": 163.1982879638672,
      "learning_rate": 1.5050494870103988e-05,
      "loss": 5.7266,
      "step": 8708
    },
    {
      "epoch": 0.35159467097295116,
      "grad_norm": 87.21420288085938,
      "learning_rate": 1.50493662668638e-05,
      "loss": 5.8503,
      "step": 8709
    },
    {
      "epoch": 0.3516350423899879,
      "grad_norm": 103.32099151611328,
      "learning_rate": 1.504823757729124e-05,
      "loss": 4.9526,
      "step": 8710
    },
    {
      "epoch": 0.3516754138070246,
      "grad_norm": 88.51436614990234,
      "learning_rate": 1.5047108801405607e-05,
      "loss": 4.133,
      "step": 8711
    },
    {
      "epoch": 0.3517157852240614,
      "grad_norm": 88.21904754638672,
      "learning_rate": 1.5045979939226196e-05,
      "loss": 4.8069,
      "step": 8712
    },
    {
      "epoch": 0.3517561566410981,
      "grad_norm": 81.22269439697266,
      "learning_rate": 1.5044850990772313e-05,
      "loss": 4.5645,
      "step": 8713
    },
    {
      "epoch": 0.3517965280581348,
      "grad_norm": 77.35594177246094,
      "learning_rate": 1.5043721956063259e-05,
      "loss": 5.1803,
      "step": 8714
    },
    {
      "epoch": 0.3518368994751716,
      "grad_norm": 133.3516082763672,
      "learning_rate": 1.5042592835118337e-05,
      "loss": 4.9136,
      "step": 8715
    },
    {
      "epoch": 0.3518772708922083,
      "grad_norm": 92.5036392211914,
      "learning_rate": 1.5041463627956856e-05,
      "loss": 4.4354,
      "step": 8716
    },
    {
      "epoch": 0.35191764230924505,
      "grad_norm": 86.31660461425781,
      "learning_rate": 1.5040334334598115e-05,
      "loss": 6.8909,
      "step": 8717
    },
    {
      "epoch": 0.35195801372628177,
      "grad_norm": 92.26172637939453,
      "learning_rate": 1.5039204955061428e-05,
      "loss": 5.3135,
      "step": 8718
    },
    {
      "epoch": 0.35199838514331855,
      "grad_norm": 149.75331115722656,
      "learning_rate": 1.5038075489366107e-05,
      "loss": 4.9888,
      "step": 8719
    },
    {
      "epoch": 0.35203875656035527,
      "grad_norm": 92.26962280273438,
      "learning_rate": 1.5036945937531458e-05,
      "loss": 5.7435,
      "step": 8720
    },
    {
      "epoch": 0.352079127977392,
      "grad_norm": 117.03031921386719,
      "learning_rate": 1.5035816299576791e-05,
      "loss": 4.18,
      "step": 8721
    },
    {
      "epoch": 0.35211949939442877,
      "grad_norm": 77.64535522460938,
      "learning_rate": 1.5034686575521429e-05,
      "loss": 3.8812,
      "step": 8722
    },
    {
      "epoch": 0.3521598708114655,
      "grad_norm": 139.77635192871094,
      "learning_rate": 1.5033556765384681e-05,
      "loss": 4.4794,
      "step": 8723
    },
    {
      "epoch": 0.3522002422285022,
      "grad_norm": 152.95175170898438,
      "learning_rate": 1.5032426869185872e-05,
      "loss": 5.4716,
      "step": 8724
    },
    {
      "epoch": 0.35224061364553894,
      "grad_norm": 145.15164184570312,
      "learning_rate": 1.5031296886944308e-05,
      "loss": 6.1911,
      "step": 8725
    },
    {
      "epoch": 0.3522809850625757,
      "grad_norm": 99.5193099975586,
      "learning_rate": 1.5030166818679322e-05,
      "loss": 6.4071,
      "step": 8726
    },
    {
      "epoch": 0.35232135647961244,
      "grad_norm": 98.30803680419922,
      "learning_rate": 1.5029036664410228e-05,
      "loss": 5.09,
      "step": 8727
    },
    {
      "epoch": 0.35236172789664916,
      "grad_norm": 94.36650085449219,
      "learning_rate": 1.5027906424156351e-05,
      "loss": 4.7657,
      "step": 8728
    },
    {
      "epoch": 0.35240209931368593,
      "grad_norm": 101.72418212890625,
      "learning_rate": 1.5026776097937013e-05,
      "loss": 6.23,
      "step": 8729
    },
    {
      "epoch": 0.35244247073072266,
      "grad_norm": 110.68695068359375,
      "learning_rate": 1.5025645685771544e-05,
      "loss": 5.3296,
      "step": 8730
    },
    {
      "epoch": 0.3524828421477594,
      "grad_norm": 463.18890380859375,
      "learning_rate": 1.5024515187679268e-05,
      "loss": 5.44,
      "step": 8731
    },
    {
      "epoch": 0.3525232135647961,
      "grad_norm": 108.12457275390625,
      "learning_rate": 1.5023384603679515e-05,
      "loss": 2.9882,
      "step": 8732
    },
    {
      "epoch": 0.3525635849818329,
      "grad_norm": 85.55570220947266,
      "learning_rate": 1.5022253933791615e-05,
      "loss": 5.4272,
      "step": 8733
    },
    {
      "epoch": 0.3526039563988696,
      "grad_norm": 90.3802719116211,
      "learning_rate": 1.5021123178034906e-05,
      "loss": 6.07,
      "step": 8734
    },
    {
      "epoch": 0.3526443278159063,
      "grad_norm": 94.93507385253906,
      "learning_rate": 1.501999233642871e-05,
      "loss": 4.8666,
      "step": 8735
    },
    {
      "epoch": 0.3526846992329431,
      "grad_norm": 91.70709228515625,
      "learning_rate": 1.5018861408992372e-05,
      "loss": 5.4879,
      "step": 8736
    },
    {
      "epoch": 0.3527250706499798,
      "grad_norm": 74.29344940185547,
      "learning_rate": 1.501773039574522e-05,
      "loss": 6.9932,
      "step": 8737
    },
    {
      "epoch": 0.35276544206701654,
      "grad_norm": 79.59151458740234,
      "learning_rate": 1.5016599296706597e-05,
      "loss": 6.2456,
      "step": 8738
    },
    {
      "epoch": 0.35280581348405327,
      "grad_norm": 69.11196899414062,
      "learning_rate": 1.501546811189584e-05,
      "loss": 5.951,
      "step": 8739
    },
    {
      "epoch": 0.35284618490109004,
      "grad_norm": 122.9038314819336,
      "learning_rate": 1.5014336841332293e-05,
      "loss": 4.7937,
      "step": 8740
    },
    {
      "epoch": 0.35288655631812677,
      "grad_norm": 83.44744873046875,
      "learning_rate": 1.5013205485035292e-05,
      "loss": 4.0778,
      "step": 8741
    },
    {
      "epoch": 0.3529269277351635,
      "grad_norm": 77.43719482421875,
      "learning_rate": 1.5012074043024185e-05,
      "loss": 4.6117,
      "step": 8742
    },
    {
      "epoch": 0.35296729915220026,
      "grad_norm": 90.80736541748047,
      "learning_rate": 1.5010942515318317e-05,
      "loss": 6.1978,
      "step": 8743
    },
    {
      "epoch": 0.353007670569237,
      "grad_norm": 95.78645324707031,
      "learning_rate": 1.5009810901937032e-05,
      "loss": 5.1213,
      "step": 8744
    },
    {
      "epoch": 0.3530480419862737,
      "grad_norm": 85.908447265625,
      "learning_rate": 1.5008679202899678e-05,
      "loss": 7.1006,
      "step": 8745
    },
    {
      "epoch": 0.35308841340331043,
      "grad_norm": 102.05917358398438,
      "learning_rate": 1.500754741822561e-05,
      "loss": 5.898,
      "step": 8746
    },
    {
      "epoch": 0.3531287848203472,
      "grad_norm": 135.2647247314453,
      "learning_rate": 1.500641554793417e-05,
      "loss": 4.7702,
      "step": 8747
    },
    {
      "epoch": 0.35316915623738393,
      "grad_norm": 74.0792007446289,
      "learning_rate": 1.5005283592044719e-05,
      "loss": 5.2769,
      "step": 8748
    },
    {
      "epoch": 0.35320952765442065,
      "grad_norm": 116.26017761230469,
      "learning_rate": 1.5004151550576605e-05,
      "loss": 4.4606,
      "step": 8749
    },
    {
      "epoch": 0.35324989907145743,
      "grad_norm": 80.29051971435547,
      "learning_rate": 1.5003019423549185e-05,
      "loss": 6.288,
      "step": 8750
    },
    {
      "epoch": 0.35329027048849415,
      "grad_norm": 111.0896987915039,
      "learning_rate": 1.5001887210981817e-05,
      "loss": 4.9735,
      "step": 8751
    },
    {
      "epoch": 0.3533306419055309,
      "grad_norm": 81.57398986816406,
      "learning_rate": 1.5000754912893856e-05,
      "loss": 4.7697,
      "step": 8752
    },
    {
      "epoch": 0.3533710133225676,
      "grad_norm": 99.54700469970703,
      "learning_rate": 1.4999622529304661e-05,
      "loss": 4.4335,
      "step": 8753
    },
    {
      "epoch": 0.3534113847396044,
      "grad_norm": 113.30178833007812,
      "learning_rate": 1.4998490060233602e-05,
      "loss": 3.8429,
      "step": 8754
    },
    {
      "epoch": 0.3534517561566411,
      "grad_norm": 109.99787139892578,
      "learning_rate": 1.499735750570003e-05,
      "loss": 5.6444,
      "step": 8755
    },
    {
      "epoch": 0.3534921275736778,
      "grad_norm": 75.8516616821289,
      "learning_rate": 1.4996224865723318e-05,
      "loss": 4.8236,
      "step": 8756
    },
    {
      "epoch": 0.3535324989907146,
      "grad_norm": 104.1716079711914,
      "learning_rate": 1.4995092140322825e-05,
      "loss": 3.4774,
      "step": 8757
    },
    {
      "epoch": 0.3535728704077513,
      "grad_norm": 116.47509765625,
      "learning_rate": 1.4993959329517923e-05,
      "loss": 5.6479,
      "step": 8758
    },
    {
      "epoch": 0.35361324182478804,
      "grad_norm": 120.6810531616211,
      "learning_rate": 1.4992826433327976e-05,
      "loss": 4.394,
      "step": 8759
    },
    {
      "epoch": 0.3536536132418248,
      "grad_norm": 74.83134460449219,
      "learning_rate": 1.499169345177236e-05,
      "loss": 5.7506,
      "step": 8760
    },
    {
      "epoch": 0.35369398465886154,
      "grad_norm": 99.17071533203125,
      "learning_rate": 1.4990560384870439e-05,
      "loss": 3.4212,
      "step": 8761
    },
    {
      "epoch": 0.35373435607589826,
      "grad_norm": 83.07367706298828,
      "learning_rate": 1.498942723264159e-05,
      "loss": 5.7916,
      "step": 8762
    },
    {
      "epoch": 0.353774727492935,
      "grad_norm": 113.84196472167969,
      "learning_rate": 1.4988293995105183e-05,
      "loss": 6.3699,
      "step": 8763
    },
    {
      "epoch": 0.35381509890997176,
      "grad_norm": 91.66001892089844,
      "learning_rate": 1.4987160672280602e-05,
      "loss": 5.4748,
      "step": 8764
    },
    {
      "epoch": 0.3538554703270085,
      "grad_norm": 162.93966674804688,
      "learning_rate": 1.4986027264187217e-05,
      "loss": 6.0535,
      "step": 8765
    },
    {
      "epoch": 0.3538958417440452,
      "grad_norm": 74.06134796142578,
      "learning_rate": 1.4984893770844413e-05,
      "loss": 6.6248,
      "step": 8766
    },
    {
      "epoch": 0.353936213161082,
      "grad_norm": 86.92379760742188,
      "learning_rate": 1.4983760192271562e-05,
      "loss": 4.6793,
      "step": 8767
    },
    {
      "epoch": 0.3539765845781187,
      "grad_norm": 166.97567749023438,
      "learning_rate": 1.4982626528488051e-05,
      "loss": 4.7117,
      "step": 8768
    },
    {
      "epoch": 0.3540169559951554,
      "grad_norm": 137.59701538085938,
      "learning_rate": 1.4981492779513261e-05,
      "loss": 6.5536,
      "step": 8769
    },
    {
      "epoch": 0.35405732741219215,
      "grad_norm": 145.6073760986328,
      "learning_rate": 1.4980358945366576e-05,
      "loss": 5.3774,
      "step": 8770
    },
    {
      "epoch": 0.3540976988292289,
      "grad_norm": 98.43611145019531,
      "learning_rate": 1.4979225026067385e-05,
      "loss": 6.3821,
      "step": 8771
    },
    {
      "epoch": 0.35413807024626565,
      "grad_norm": 671.3471069335938,
      "learning_rate": 1.4978091021635073e-05,
      "loss": 5.7976,
      "step": 8772
    },
    {
      "epoch": 0.35417844166330237,
      "grad_norm": 154.67535400390625,
      "learning_rate": 1.4976956932089028e-05,
      "loss": 5.2962,
      "step": 8773
    },
    {
      "epoch": 0.35421881308033915,
      "grad_norm": 136.01409912109375,
      "learning_rate": 1.4975822757448644e-05,
      "loss": 3.8472,
      "step": 8774
    },
    {
      "epoch": 0.35425918449737587,
      "grad_norm": 120.97064208984375,
      "learning_rate": 1.4974688497733307e-05,
      "loss": 6.9582,
      "step": 8775
    },
    {
      "epoch": 0.3542995559144126,
      "grad_norm": 131.41253662109375,
      "learning_rate": 1.4973554152962418e-05,
      "loss": 5.857,
      "step": 8776
    },
    {
      "epoch": 0.3543399273314493,
      "grad_norm": 74.85475158691406,
      "learning_rate": 1.4972419723155362e-05,
      "loss": 6.0532,
      "step": 8777
    },
    {
      "epoch": 0.3543802987484861,
      "grad_norm": 95.56173706054688,
      "learning_rate": 1.4971285208331545e-05,
      "loss": 6.6838,
      "step": 8778
    },
    {
      "epoch": 0.3544206701655228,
      "grad_norm": 106.63695526123047,
      "learning_rate": 1.4970150608510356e-05,
      "loss": 5.133,
      "step": 8779
    },
    {
      "epoch": 0.35446104158255953,
      "grad_norm": 114.21007537841797,
      "learning_rate": 1.4969015923711197e-05,
      "loss": 5.3113,
      "step": 8780
    },
    {
      "epoch": 0.3545014129995963,
      "grad_norm": 75.16924285888672,
      "learning_rate": 1.4967881153953472e-05,
      "loss": 2.8505,
      "step": 8781
    },
    {
      "epoch": 0.35454178441663303,
      "grad_norm": 100.80567169189453,
      "learning_rate": 1.496674629925658e-05,
      "loss": 5.098,
      "step": 8782
    },
    {
      "epoch": 0.35458215583366975,
      "grad_norm": 87.32557678222656,
      "learning_rate": 1.4965611359639923e-05,
      "loss": 7.1232,
      "step": 8783
    },
    {
      "epoch": 0.3546225272507065,
      "grad_norm": 91.85084533691406,
      "learning_rate": 1.4964476335122909e-05,
      "loss": 6.4016,
      "step": 8784
    },
    {
      "epoch": 0.35466289866774325,
      "grad_norm": 145.65887451171875,
      "learning_rate": 1.496334122572494e-05,
      "loss": 5.1126,
      "step": 8785
    },
    {
      "epoch": 0.35470327008478,
      "grad_norm": 227.0939178466797,
      "learning_rate": 1.4962206031465428e-05,
      "loss": 5.2927,
      "step": 8786
    },
    {
      "epoch": 0.3547436415018167,
      "grad_norm": 106.87454223632812,
      "learning_rate": 1.496107075236378e-05,
      "loss": 4.9532,
      "step": 8787
    },
    {
      "epoch": 0.3547840129188535,
      "grad_norm": 72.49174499511719,
      "learning_rate": 1.4959935388439409e-05,
      "loss": 4.9201,
      "step": 8788
    },
    {
      "epoch": 0.3548243843358902,
      "grad_norm": 98.0425033569336,
      "learning_rate": 1.4958799939711722e-05,
      "loss": 5.731,
      "step": 8789
    },
    {
      "epoch": 0.3548647557529269,
      "grad_norm": 129.52828979492188,
      "learning_rate": 1.4957664406200136e-05,
      "loss": 4.7341,
      "step": 8790
    },
    {
      "epoch": 0.35490512716996364,
      "grad_norm": 80.97358703613281,
      "learning_rate": 1.4956528787924067e-05,
      "loss": 5.1789,
      "step": 8791
    },
    {
      "epoch": 0.3549454985870004,
      "grad_norm": 86.48793029785156,
      "learning_rate": 1.495539308490293e-05,
      "loss": 4.4847,
      "step": 8792
    },
    {
      "epoch": 0.35498587000403714,
      "grad_norm": 73.98668670654297,
      "learning_rate": 1.4954257297156143e-05,
      "loss": 5.7776,
      "step": 8793
    },
    {
      "epoch": 0.35502624142107386,
      "grad_norm": 98.53644561767578,
      "learning_rate": 1.4953121424703126e-05,
      "loss": 6.4842,
      "step": 8794
    },
    {
      "epoch": 0.35506661283811064,
      "grad_norm": 113.6964340209961,
      "learning_rate": 1.4951985467563294e-05,
      "loss": 4.925,
      "step": 8795
    },
    {
      "epoch": 0.35510698425514736,
      "grad_norm": 115.05333709716797,
      "learning_rate": 1.4950849425756083e-05,
      "loss": 4.5514,
      "step": 8796
    },
    {
      "epoch": 0.3551473556721841,
      "grad_norm": 111.04338836669922,
      "learning_rate": 1.49497132993009e-05,
      "loss": 5.6557,
      "step": 8797
    },
    {
      "epoch": 0.3551877270892208,
      "grad_norm": 101.06678771972656,
      "learning_rate": 1.4948577088217182e-05,
      "loss": 5.0357,
      "step": 8798
    },
    {
      "epoch": 0.3552280985062576,
      "grad_norm": 95.6523208618164,
      "learning_rate": 1.494744079252435e-05,
      "loss": 5.9347,
      "step": 8799
    },
    {
      "epoch": 0.3552684699232943,
      "grad_norm": 123.28790283203125,
      "learning_rate": 1.4946304412241834e-05,
      "loss": 4.8975,
      "step": 8800
    },
    {
      "epoch": 0.355308841340331,
      "grad_norm": 88.1064682006836,
      "learning_rate": 1.4945167947389064e-05,
      "loss": 5.6584,
      "step": 8801
    },
    {
      "epoch": 0.3553492127573678,
      "grad_norm": 152.06198120117188,
      "learning_rate": 1.4944031397985468e-05,
      "loss": 5.6222,
      "step": 8802
    },
    {
      "epoch": 0.3553895841744045,
      "grad_norm": 61.507049560546875,
      "learning_rate": 1.4942894764050482e-05,
      "loss": 4.2325,
      "step": 8803
    },
    {
      "epoch": 0.35542995559144125,
      "grad_norm": 142.3750762939453,
      "learning_rate": 1.4941758045603539e-05,
      "loss": 6.9354,
      "step": 8804
    },
    {
      "epoch": 0.35547032700847797,
      "grad_norm": 88.03289031982422,
      "learning_rate": 1.4940621242664068e-05,
      "loss": 4.6173,
      "step": 8805
    },
    {
      "epoch": 0.35551069842551475,
      "grad_norm": 262.731689453125,
      "learning_rate": 1.4939484355251516e-05,
      "loss": 5.6526,
      "step": 8806
    },
    {
      "epoch": 0.35555106984255147,
      "grad_norm": 694.6334228515625,
      "learning_rate": 1.4938347383385314e-05,
      "loss": 5.1726,
      "step": 8807
    },
    {
      "epoch": 0.3555914412595882,
      "grad_norm": 108.84500122070312,
      "learning_rate": 1.4937210327084906e-05,
      "loss": 4.6212,
      "step": 8808
    },
    {
      "epoch": 0.35563181267662497,
      "grad_norm": 121.19515228271484,
      "learning_rate": 1.4936073186369729e-05,
      "loss": 5.4989,
      "step": 8809
    },
    {
      "epoch": 0.3556721840936617,
      "grad_norm": 116.06332397460938,
      "learning_rate": 1.4934935961259225e-05,
      "loss": 4.2726,
      "step": 8810
    },
    {
      "epoch": 0.3557125555106984,
      "grad_norm": 87.75636291503906,
      "learning_rate": 1.4933798651772839e-05,
      "loss": 4.5214,
      "step": 8811
    },
    {
      "epoch": 0.35575292692773514,
      "grad_norm": 91.58477783203125,
      "learning_rate": 1.4932661257930022e-05,
      "loss": 6.6276,
      "step": 8812
    },
    {
      "epoch": 0.3557932983447719,
      "grad_norm": 91.20993041992188,
      "learning_rate": 1.4931523779750214e-05,
      "loss": 6.183,
      "step": 8813
    },
    {
      "epoch": 0.35583366976180864,
      "grad_norm": 87.61544036865234,
      "learning_rate": 1.4930386217252865e-05,
      "loss": 5.168,
      "step": 8814
    },
    {
      "epoch": 0.35587404117884536,
      "grad_norm": 101.59625244140625,
      "learning_rate": 1.4929248570457426e-05,
      "loss": 5.0139,
      "step": 8815
    },
    {
      "epoch": 0.35591441259588213,
      "grad_norm": 93.10003662109375,
      "learning_rate": 1.4928110839383348e-05,
      "loss": 6.169,
      "step": 8816
    },
    {
      "epoch": 0.35595478401291886,
      "grad_norm": 116.92972564697266,
      "learning_rate": 1.492697302405008e-05,
      "loss": 4.736,
      "step": 8817
    },
    {
      "epoch": 0.3559951554299556,
      "grad_norm": 132.1350555419922,
      "learning_rate": 1.4925835124477081e-05,
      "loss": 4.9461,
      "step": 8818
    },
    {
      "epoch": 0.35603552684699236,
      "grad_norm": 155.26243591308594,
      "learning_rate": 1.4924697140683802e-05,
      "loss": 5.4074,
      "step": 8819
    },
    {
      "epoch": 0.3560758982640291,
      "grad_norm": 85.9544906616211,
      "learning_rate": 1.4923559072689701e-05,
      "loss": 6.1145,
      "step": 8820
    },
    {
      "epoch": 0.3561162696810658,
      "grad_norm": 95.20445251464844,
      "learning_rate": 1.492242092051424e-05,
      "loss": 5.1213,
      "step": 8821
    },
    {
      "epoch": 0.3561566410981025,
      "grad_norm": 104.45774841308594,
      "learning_rate": 1.4921282684176875e-05,
      "loss": 5.3284,
      "step": 8822
    },
    {
      "epoch": 0.3561970125151393,
      "grad_norm": 73.8716049194336,
      "learning_rate": 1.4920144363697067e-05,
      "loss": 4.6399,
      "step": 8823
    },
    {
      "epoch": 0.356237383932176,
      "grad_norm": 91.18668365478516,
      "learning_rate": 1.4919005959094282e-05,
      "loss": 6.0411,
      "step": 8824
    },
    {
      "epoch": 0.35627775534921274,
      "grad_norm": 114.73775482177734,
      "learning_rate": 1.4917867470387978e-05,
      "loss": 5.4158,
      "step": 8825
    },
    {
      "epoch": 0.3563181267662495,
      "grad_norm": 95.7513656616211,
      "learning_rate": 1.4916728897597628e-05,
      "loss": 5.5822,
      "step": 8826
    },
    {
      "epoch": 0.35635849818328624,
      "grad_norm": 95.43356323242188,
      "learning_rate": 1.4915590240742694e-05,
      "loss": 5.8273,
      "step": 8827
    },
    {
      "epoch": 0.35639886960032297,
      "grad_norm": 98.1636962890625,
      "learning_rate": 1.4914451499842648e-05,
      "loss": 6.2107,
      "step": 8828
    },
    {
      "epoch": 0.3564392410173597,
      "grad_norm": 102.43901824951172,
      "learning_rate": 1.4913312674916953e-05,
      "loss": 5.5072,
      "step": 8829
    },
    {
      "epoch": 0.35647961243439646,
      "grad_norm": 59.4473876953125,
      "learning_rate": 1.4912173765985085e-05,
      "loss": 5.945,
      "step": 8830
    },
    {
      "epoch": 0.3565199838514332,
      "grad_norm": 87.51011657714844,
      "learning_rate": 1.491103477306652e-05,
      "loss": 5.4617,
      "step": 8831
    },
    {
      "epoch": 0.3565603552684699,
      "grad_norm": 98.60214233398438,
      "learning_rate": 1.4909895696180725e-05,
      "loss": 5.2302,
      "step": 8832
    },
    {
      "epoch": 0.3566007266855067,
      "grad_norm": 133.5031280517578,
      "learning_rate": 1.490875653534718e-05,
      "loss": 5.8516,
      "step": 8833
    },
    {
      "epoch": 0.3566410981025434,
      "grad_norm": 161.13568115234375,
      "learning_rate": 1.4907617290585363e-05,
      "loss": 5.6878,
      "step": 8834
    },
    {
      "epoch": 0.35668146951958013,
      "grad_norm": 75.21280670166016,
      "learning_rate": 1.4906477961914748e-05,
      "loss": 5.0666,
      "step": 8835
    },
    {
      "epoch": 0.35672184093661685,
      "grad_norm": 101.87770080566406,
      "learning_rate": 1.4905338549354819e-05,
      "loss": 6.2343,
      "step": 8836
    },
    {
      "epoch": 0.35676221235365363,
      "grad_norm": 83.6406478881836,
      "learning_rate": 1.4904199052925054e-05,
      "loss": 5.3247,
      "step": 8837
    },
    {
      "epoch": 0.35680258377069035,
      "grad_norm": 85.18995666503906,
      "learning_rate": 1.4903059472644941e-05,
      "loss": 6.7269,
      "step": 8838
    },
    {
      "epoch": 0.3568429551877271,
      "grad_norm": 71.98890686035156,
      "learning_rate": 1.4901919808533955e-05,
      "loss": 4.4488,
      "step": 8839
    },
    {
      "epoch": 0.35688332660476385,
      "grad_norm": 156.56748962402344,
      "learning_rate": 1.490078006061159e-05,
      "loss": 6.301,
      "step": 8840
    },
    {
      "epoch": 0.3569236980218006,
      "grad_norm": 85.56804656982422,
      "learning_rate": 1.489964022889733e-05,
      "loss": 4.6836,
      "step": 8841
    },
    {
      "epoch": 0.3569640694388373,
      "grad_norm": 67.81390380859375,
      "learning_rate": 1.4898500313410664e-05,
      "loss": 5.2656,
      "step": 8842
    },
    {
      "epoch": 0.357004440855874,
      "grad_norm": 185.44955444335938,
      "learning_rate": 1.489736031417108e-05,
      "loss": 5.3085,
      "step": 8843
    },
    {
      "epoch": 0.3570448122729108,
      "grad_norm": 78.9861068725586,
      "learning_rate": 1.4896220231198073e-05,
      "loss": 5.8975,
      "step": 8844
    },
    {
      "epoch": 0.3570851836899475,
      "grad_norm": 86.2502670288086,
      "learning_rate": 1.4895080064511131e-05,
      "loss": 5.6248,
      "step": 8845
    },
    {
      "epoch": 0.35712555510698424,
      "grad_norm": 116.32249450683594,
      "learning_rate": 1.4893939814129752e-05,
      "loss": 4.9414,
      "step": 8846
    },
    {
      "epoch": 0.357165926524021,
      "grad_norm": 80.59905242919922,
      "learning_rate": 1.4892799480073432e-05,
      "loss": 5.6592,
      "step": 8847
    },
    {
      "epoch": 0.35720629794105774,
      "grad_norm": 87.40462493896484,
      "learning_rate": 1.4891659062361665e-05,
      "loss": 5.3043,
      "step": 8848
    },
    {
      "epoch": 0.35724666935809446,
      "grad_norm": 73.80174255371094,
      "learning_rate": 1.489051856101395e-05,
      "loss": 4.4676,
      "step": 8849
    },
    {
      "epoch": 0.3572870407751312,
      "grad_norm": 92.0369644165039,
      "learning_rate": 1.4889377976049786e-05,
      "loss": 5.2623,
      "step": 8850
    },
    {
      "epoch": 0.35732741219216796,
      "grad_norm": 113.75395202636719,
      "learning_rate": 1.4888237307488678e-05,
      "loss": 6.4282,
      "step": 8851
    },
    {
      "epoch": 0.3573677836092047,
      "grad_norm": 73.99334716796875,
      "learning_rate": 1.488709655535013e-05,
      "loss": 7.0687,
      "step": 8852
    },
    {
      "epoch": 0.3574081550262414,
      "grad_norm": 93.78912353515625,
      "learning_rate": 1.488595571965364e-05,
      "loss": 5.6181,
      "step": 8853
    },
    {
      "epoch": 0.3574485264432782,
      "grad_norm": 72.8905258178711,
      "learning_rate": 1.4884814800418717e-05,
      "loss": 4.8713,
      "step": 8854
    },
    {
      "epoch": 0.3574888978603149,
      "grad_norm": 146.46995544433594,
      "learning_rate": 1.4883673797664868e-05,
      "loss": 6.1018,
      "step": 8855
    },
    {
      "epoch": 0.3575292692773516,
      "grad_norm": 99.1534194946289,
      "learning_rate": 1.4882532711411601e-05,
      "loss": 5.8606,
      "step": 8856
    },
    {
      "epoch": 0.35756964069438835,
      "grad_norm": 100.2274398803711,
      "learning_rate": 1.4881391541678427e-05,
      "loss": 6.5067,
      "step": 8857
    },
    {
      "epoch": 0.3576100121114251,
      "grad_norm": 105.86824035644531,
      "learning_rate": 1.4880250288484857e-05,
      "loss": 4.9247,
      "step": 8858
    },
    {
      "epoch": 0.35765038352846185,
      "grad_norm": 79.03643035888672,
      "learning_rate": 1.4879108951850401e-05,
      "loss": 5.0181,
      "step": 8859
    },
    {
      "epoch": 0.35769075494549857,
      "grad_norm": 111.0548095703125,
      "learning_rate": 1.4877967531794574e-05,
      "loss": 3.7209,
      "step": 8860
    },
    {
      "epoch": 0.35773112636253535,
      "grad_norm": 87.84162139892578,
      "learning_rate": 1.4876826028336897e-05,
      "loss": 5.1385,
      "step": 8861
    },
    {
      "epoch": 0.35777149777957207,
      "grad_norm": 81.22015380859375,
      "learning_rate": 1.4875684441496883e-05,
      "loss": 3.5953,
      "step": 8862
    },
    {
      "epoch": 0.3578118691966088,
      "grad_norm": 92.9424819946289,
      "learning_rate": 1.4874542771294049e-05,
      "loss": 4.5634,
      "step": 8863
    },
    {
      "epoch": 0.3578522406136455,
      "grad_norm": 74.8837661743164,
      "learning_rate": 1.4873401017747917e-05,
      "loss": 6.0431,
      "step": 8864
    },
    {
      "epoch": 0.3578926120306823,
      "grad_norm": 70.436767578125,
      "learning_rate": 1.4872259180878009e-05,
      "loss": 4.9392,
      "step": 8865
    },
    {
      "epoch": 0.357932983447719,
      "grad_norm": 121.53055572509766,
      "learning_rate": 1.4871117260703845e-05,
      "loss": 6.6597,
      "step": 8866
    },
    {
      "epoch": 0.35797335486475573,
      "grad_norm": 111.17406463623047,
      "learning_rate": 1.4869975257244951e-05,
      "loss": 4.3931,
      "step": 8867
    },
    {
      "epoch": 0.3580137262817925,
      "grad_norm": 165.4057159423828,
      "learning_rate": 1.4868833170520854e-05,
      "loss": 4.1541,
      "step": 8868
    },
    {
      "epoch": 0.35805409769882923,
      "grad_norm": 85.13111114501953,
      "learning_rate": 1.4867691000551077e-05,
      "loss": 6.304,
      "step": 8869
    },
    {
      "epoch": 0.35809446911586595,
      "grad_norm": 82.85501861572266,
      "learning_rate": 1.4866548747355149e-05,
      "loss": 4.8208,
      "step": 8870
    },
    {
      "epoch": 0.35813484053290273,
      "grad_norm": 94.3434829711914,
      "learning_rate": 1.4865406410952603e-05,
      "loss": 6.0716,
      "step": 8871
    },
    {
      "epoch": 0.35817521194993945,
      "grad_norm": 105.90048217773438,
      "learning_rate": 1.4864263991362972e-05,
      "loss": 5.5024,
      "step": 8872
    },
    {
      "epoch": 0.3582155833669762,
      "grad_norm": 105.62590789794922,
      "learning_rate": 1.4863121488605783e-05,
      "loss": 4.9677,
      "step": 8873
    },
    {
      "epoch": 0.3582559547840129,
      "grad_norm": 74.16838073730469,
      "learning_rate": 1.4861978902700573e-05,
      "loss": 4.5555,
      "step": 8874
    },
    {
      "epoch": 0.3582963262010497,
      "grad_norm": 142.34861755371094,
      "learning_rate": 1.4860836233666878e-05,
      "loss": 4.7663,
      "step": 8875
    },
    {
      "epoch": 0.3583366976180864,
      "grad_norm": 126.59479522705078,
      "learning_rate": 1.4859693481524233e-05,
      "loss": 5.4442,
      "step": 8876
    },
    {
      "epoch": 0.3583770690351231,
      "grad_norm": 151.92578125,
      "learning_rate": 1.485855064629218e-05,
      "loss": 6.0588,
      "step": 8877
    },
    {
      "epoch": 0.3584174404521599,
      "grad_norm": 83.7823715209961,
      "learning_rate": 1.4857407727990254e-05,
      "loss": 7.0936,
      "step": 8878
    },
    {
      "epoch": 0.3584578118691966,
      "grad_norm": 100.5744857788086,
      "learning_rate": 1.4856264726637998e-05,
      "loss": 4.046,
      "step": 8879
    },
    {
      "epoch": 0.35849818328623334,
      "grad_norm": 82.72465515136719,
      "learning_rate": 1.485512164225496e-05,
      "loss": 4.4287,
      "step": 8880
    },
    {
      "epoch": 0.35853855470327006,
      "grad_norm": 160.97607421875,
      "learning_rate": 1.4853978474860675e-05,
      "loss": 4.5844,
      "step": 8881
    },
    {
      "epoch": 0.35857892612030684,
      "grad_norm": 104.75254821777344,
      "learning_rate": 1.4852835224474695e-05,
      "loss": 4.7677,
      "step": 8882
    },
    {
      "epoch": 0.35861929753734356,
      "grad_norm": 71.18038940429688,
      "learning_rate": 1.4851691891116564e-05,
      "loss": 4.1557,
      "step": 8883
    },
    {
      "epoch": 0.3586596689543803,
      "grad_norm": 98.72925567626953,
      "learning_rate": 1.4850548474805832e-05,
      "loss": 6.7131,
      "step": 8884
    },
    {
      "epoch": 0.35870004037141706,
      "grad_norm": 101.97978973388672,
      "learning_rate": 1.4849404975562047e-05,
      "loss": 5.0442,
      "step": 8885
    },
    {
      "epoch": 0.3587404117884538,
      "grad_norm": 108.69641876220703,
      "learning_rate": 1.4848261393404762e-05,
      "loss": 3.6617,
      "step": 8886
    },
    {
      "epoch": 0.3587807832054905,
      "grad_norm": 104.00711822509766,
      "learning_rate": 1.4847117728353528e-05,
      "loss": 5.9959,
      "step": 8887
    },
    {
      "epoch": 0.3588211546225272,
      "grad_norm": 131.0713653564453,
      "learning_rate": 1.48459739804279e-05,
      "loss": 5.9956,
      "step": 8888
    },
    {
      "epoch": 0.358861526039564,
      "grad_norm": 77.83228302001953,
      "learning_rate": 1.4844830149647431e-05,
      "loss": 5.9037,
      "step": 8889
    },
    {
      "epoch": 0.3589018974566007,
      "grad_norm": 104.58064270019531,
      "learning_rate": 1.4843686236031682e-05,
      "loss": 4.0363,
      "step": 8890
    },
    {
      "epoch": 0.35894226887363745,
      "grad_norm": 101.00995635986328,
      "learning_rate": 1.484254223960021e-05,
      "loss": 5.1918,
      "step": 8891
    },
    {
      "epoch": 0.3589826402906742,
      "grad_norm": 87.82205200195312,
      "learning_rate": 1.4841398160372573e-05,
      "loss": 4.3814,
      "step": 8892
    },
    {
      "epoch": 0.35902301170771095,
      "grad_norm": 152.42074584960938,
      "learning_rate": 1.4840253998368332e-05,
      "loss": 4.642,
      "step": 8893
    },
    {
      "epoch": 0.35906338312474767,
      "grad_norm": 95.48905944824219,
      "learning_rate": 1.4839109753607054e-05,
      "loss": 5.4012,
      "step": 8894
    },
    {
      "epoch": 0.3591037545417844,
      "grad_norm": 74.58067321777344,
      "learning_rate": 1.4837965426108296e-05,
      "loss": 4.4067,
      "step": 8895
    },
    {
      "epoch": 0.35914412595882117,
      "grad_norm": 171.8329620361328,
      "learning_rate": 1.483682101589163e-05,
      "loss": 6.1045,
      "step": 8896
    },
    {
      "epoch": 0.3591844973758579,
      "grad_norm": 114.63589477539062,
      "learning_rate": 1.4835676522976614e-05,
      "loss": 5.4643,
      "step": 8897
    },
    {
      "epoch": 0.3592248687928946,
      "grad_norm": 197.34347534179688,
      "learning_rate": 1.4834531947382824e-05,
      "loss": 3.907,
      "step": 8898
    },
    {
      "epoch": 0.3592652402099314,
      "grad_norm": 76.16448974609375,
      "learning_rate": 1.4833387289129827e-05,
      "loss": 3.9987,
      "step": 8899
    },
    {
      "epoch": 0.3593056116269681,
      "grad_norm": 104.1001968383789,
      "learning_rate": 1.4832242548237194e-05,
      "loss": 4.1189,
      "step": 8900
    },
    {
      "epoch": 0.35934598304400484,
      "grad_norm": 77.03291320800781,
      "learning_rate": 1.48310977247245e-05,
      "loss": 5.7294,
      "step": 8901
    },
    {
      "epoch": 0.35938635446104156,
      "grad_norm": 101.80243682861328,
      "learning_rate": 1.4829952818611313e-05,
      "loss": 5.7094,
      "step": 8902
    },
    {
      "epoch": 0.35942672587807833,
      "grad_norm": 117.5005874633789,
      "learning_rate": 1.4828807829917214e-05,
      "loss": 4.7716,
      "step": 8903
    },
    {
      "epoch": 0.35946709729511506,
      "grad_norm": 123.5639877319336,
      "learning_rate": 1.4827662758661776e-05,
      "loss": 4.9114,
      "step": 8904
    },
    {
      "epoch": 0.3595074687121518,
      "grad_norm": 114.02906036376953,
      "learning_rate": 1.4826517604864579e-05,
      "loss": 6.0787,
      "step": 8905
    },
    {
      "epoch": 0.35954784012918856,
      "grad_norm": 142.47947692871094,
      "learning_rate": 1.4825372368545201e-05,
      "loss": 4.742,
      "step": 8906
    },
    {
      "epoch": 0.3595882115462253,
      "grad_norm": 80.92357635498047,
      "learning_rate": 1.4824227049723223e-05,
      "loss": 4.1974,
      "step": 8907
    },
    {
      "epoch": 0.359628582963262,
      "grad_norm": 170.86549377441406,
      "learning_rate": 1.4823081648418228e-05,
      "loss": 6.122,
      "step": 8908
    },
    {
      "epoch": 0.3596689543802987,
      "grad_norm": 132.63470458984375,
      "learning_rate": 1.48219361646498e-05,
      "loss": 3.7083,
      "step": 8909
    },
    {
      "epoch": 0.3597093257973355,
      "grad_norm": 95.48944854736328,
      "learning_rate": 1.4820790598437525e-05,
      "loss": 5.6875,
      "step": 8910
    },
    {
      "epoch": 0.3597496972143722,
      "grad_norm": 127.88106536865234,
      "learning_rate": 1.4819644949800985e-05,
      "loss": 7.096,
      "step": 8911
    },
    {
      "epoch": 0.35979006863140894,
      "grad_norm": 88.12823486328125,
      "learning_rate": 1.4818499218759775e-05,
      "loss": 5.5092,
      "step": 8912
    },
    {
      "epoch": 0.3598304400484457,
      "grad_norm": 124.04251098632812,
      "learning_rate": 1.4817353405333476e-05,
      "loss": 4.8148,
      "step": 8913
    },
    {
      "epoch": 0.35987081146548244,
      "grad_norm": 100.93739318847656,
      "learning_rate": 1.4816207509541685e-05,
      "loss": 5.6795,
      "step": 8914
    },
    {
      "epoch": 0.35991118288251917,
      "grad_norm": 88.50590515136719,
      "learning_rate": 1.4815061531403993e-05,
      "loss": 4.3845,
      "step": 8915
    },
    {
      "epoch": 0.3599515542995559,
      "grad_norm": 115.5939712524414,
      "learning_rate": 1.4813915470939992e-05,
      "loss": 6.0567,
      "step": 8916
    },
    {
      "epoch": 0.35999192571659266,
      "grad_norm": 73.14820861816406,
      "learning_rate": 1.4812769328169278e-05,
      "loss": 3.924,
      "step": 8917
    },
    {
      "epoch": 0.3600322971336294,
      "grad_norm": 116.58731079101562,
      "learning_rate": 1.4811623103111447e-05,
      "loss": 4.3956,
      "step": 8918
    },
    {
      "epoch": 0.3600726685506661,
      "grad_norm": 108.53688049316406,
      "learning_rate": 1.4810476795786098e-05,
      "loss": 5.5714,
      "step": 8919
    },
    {
      "epoch": 0.3601130399677029,
      "grad_norm": 93.8448715209961,
      "learning_rate": 1.4809330406212827e-05,
      "loss": 3.2829,
      "step": 8920
    },
    {
      "epoch": 0.3601534113847396,
      "grad_norm": 99.25374603271484,
      "learning_rate": 1.4808183934411238e-05,
      "loss": 6.461,
      "step": 8921
    },
    {
      "epoch": 0.36019378280177633,
      "grad_norm": 130.45925903320312,
      "learning_rate": 1.480703738040093e-05,
      "loss": 3.7563,
      "step": 8922
    },
    {
      "epoch": 0.3602341542188131,
      "grad_norm": 118.61727905273438,
      "learning_rate": 1.4805890744201512e-05,
      "loss": 4.8418,
      "step": 8923
    },
    {
      "epoch": 0.36027452563584983,
      "grad_norm": 87.10490417480469,
      "learning_rate": 1.480474402583258e-05,
      "loss": 5.1485,
      "step": 8924
    },
    {
      "epoch": 0.36031489705288655,
      "grad_norm": 105.86947631835938,
      "learning_rate": 1.4803597225313749e-05,
      "loss": 4.4134,
      "step": 8925
    },
    {
      "epoch": 0.3603552684699233,
      "grad_norm": 130.74493408203125,
      "learning_rate": 1.480245034266462e-05,
      "loss": 4.0519,
      "step": 8926
    },
    {
      "epoch": 0.36039563988696005,
      "grad_norm": 208.9917755126953,
      "learning_rate": 1.4801303377904808e-05,
      "loss": 5.0499,
      "step": 8927
    },
    {
      "epoch": 0.3604360113039968,
      "grad_norm": 82.19530487060547,
      "learning_rate": 1.4800156331053916e-05,
      "loss": 5.971,
      "step": 8928
    },
    {
      "epoch": 0.3604763827210335,
      "grad_norm": 72.91118621826172,
      "learning_rate": 1.479900920213156e-05,
      "loss": 5.1041,
      "step": 8929
    },
    {
      "epoch": 0.3605167541380703,
      "grad_norm": 88.13479614257812,
      "learning_rate": 1.4797861991157354e-05,
      "loss": 4.8299,
      "step": 8930
    },
    {
      "epoch": 0.360557125555107,
      "grad_norm": 100.55601501464844,
      "learning_rate": 1.4796714698150917e-05,
      "loss": 4.6026,
      "step": 8931
    },
    {
      "epoch": 0.3605974969721437,
      "grad_norm": 90.87042999267578,
      "learning_rate": 1.4795567323131852e-05,
      "loss": 4.9508,
      "step": 8932
    },
    {
      "epoch": 0.36063786838918044,
      "grad_norm": 251.41238403320312,
      "learning_rate": 1.4794419866119788e-05,
      "loss": 5.6343,
      "step": 8933
    },
    {
      "epoch": 0.3606782398062172,
      "grad_norm": 100.63956451416016,
      "learning_rate": 1.479327232713434e-05,
      "loss": 5.6798,
      "step": 8934
    },
    {
      "epoch": 0.36071861122325394,
      "grad_norm": 134.56089782714844,
      "learning_rate": 1.4792124706195128e-05,
      "loss": 5.3231,
      "step": 8935
    },
    {
      "epoch": 0.36075898264029066,
      "grad_norm": 217.29815673828125,
      "learning_rate": 1.4790977003321773e-05,
      "loss": 4.7178,
      "step": 8936
    },
    {
      "epoch": 0.36079935405732744,
      "grad_norm": 104.30968475341797,
      "learning_rate": 1.4789829218533902e-05,
      "loss": 5.3066,
      "step": 8937
    },
    {
      "epoch": 0.36083972547436416,
      "grad_norm": 74.58873748779297,
      "learning_rate": 1.4788681351851133e-05,
      "loss": 4.7413,
      "step": 8938
    },
    {
      "epoch": 0.3608800968914009,
      "grad_norm": 136.5140838623047,
      "learning_rate": 1.4787533403293095e-05,
      "loss": 6.0595,
      "step": 8939
    },
    {
      "epoch": 0.3609204683084376,
      "grad_norm": 140.578369140625,
      "learning_rate": 1.4786385372879418e-05,
      "loss": 5.6185,
      "step": 8940
    },
    {
      "epoch": 0.3609608397254744,
      "grad_norm": 98.34929656982422,
      "learning_rate": 1.4785237260629727e-05,
      "loss": 3.7896,
      "step": 8941
    },
    {
      "epoch": 0.3610012111425111,
      "grad_norm": 102.49340057373047,
      "learning_rate": 1.4784089066563652e-05,
      "loss": 4.3962,
      "step": 8942
    },
    {
      "epoch": 0.3610415825595478,
      "grad_norm": 406.104736328125,
      "learning_rate": 1.4782940790700828e-05,
      "loss": 3.9068,
      "step": 8943
    },
    {
      "epoch": 0.3610819539765846,
      "grad_norm": 94.74649810791016,
      "learning_rate": 1.4781792433060884e-05,
      "loss": 3.6071,
      "step": 8944
    },
    {
      "epoch": 0.3611223253936213,
      "grad_norm": 125.95561981201172,
      "learning_rate": 1.4780643993663457e-05,
      "loss": 5.77,
      "step": 8945
    },
    {
      "epoch": 0.36116269681065805,
      "grad_norm": 91.74636840820312,
      "learning_rate": 1.477949547252818e-05,
      "loss": 4.7778,
      "step": 8946
    },
    {
      "epoch": 0.36120306822769477,
      "grad_norm": 97.13681030273438,
      "learning_rate": 1.4778346869674693e-05,
      "loss": 4.9998,
      "step": 8947
    },
    {
      "epoch": 0.36124343964473155,
      "grad_norm": 89.74176788330078,
      "learning_rate": 1.477719818512263e-05,
      "loss": 5.0308,
      "step": 8948
    },
    {
      "epoch": 0.36128381106176827,
      "grad_norm": 105.4883804321289,
      "learning_rate": 1.4776049418891636e-05,
      "loss": 4.9821,
      "step": 8949
    },
    {
      "epoch": 0.361324182478805,
      "grad_norm": 71.29572296142578,
      "learning_rate": 1.4774900571001347e-05,
      "loss": 5.5815,
      "step": 8950
    },
    {
      "epoch": 0.36136455389584177,
      "grad_norm": 100.81729888916016,
      "learning_rate": 1.4773751641471412e-05,
      "loss": 4.8839,
      "step": 8951
    },
    {
      "epoch": 0.3614049253128785,
      "grad_norm": 383.8350830078125,
      "learning_rate": 1.4772602630321469e-05,
      "loss": 6.0512,
      "step": 8952
    },
    {
      "epoch": 0.3614452967299152,
      "grad_norm": 98.09713745117188,
      "learning_rate": 1.4771453537571168e-05,
      "loss": 6.0319,
      "step": 8953
    },
    {
      "epoch": 0.36148566814695193,
      "grad_norm": 98.46524047851562,
      "learning_rate": 1.4770304363240151e-05,
      "loss": 5.8112,
      "step": 8954
    },
    {
      "epoch": 0.3615260395639887,
      "grad_norm": 112.15991973876953,
      "learning_rate": 1.4769155107348073e-05,
      "loss": 3.3479,
      "step": 8955
    },
    {
      "epoch": 0.36156641098102543,
      "grad_norm": 155.6320037841797,
      "learning_rate": 1.4768005769914574e-05,
      "loss": 5.3262,
      "step": 8956
    },
    {
      "epoch": 0.36160678239806215,
      "grad_norm": 100.6982421875,
      "learning_rate": 1.4766856350959313e-05,
      "loss": 5.449,
      "step": 8957
    },
    {
      "epoch": 0.36164715381509893,
      "grad_norm": 90.63648986816406,
      "learning_rate": 1.476570685050194e-05,
      "loss": 5.6784,
      "step": 8958
    },
    {
      "epoch": 0.36168752523213565,
      "grad_norm": 117.80818176269531,
      "learning_rate": 1.4764557268562108e-05,
      "loss": 6.4535,
      "step": 8959
    },
    {
      "epoch": 0.3617278966491724,
      "grad_norm": 200.91749572753906,
      "learning_rate": 1.4763407605159473e-05,
      "loss": 6.6008,
      "step": 8960
    },
    {
      "epoch": 0.3617682680662091,
      "grad_norm": 136.39280700683594,
      "learning_rate": 1.4762257860313692e-05,
      "loss": 5.7113,
      "step": 8961
    },
    {
      "epoch": 0.3618086394832459,
      "grad_norm": 87.82492065429688,
      "learning_rate": 1.476110803404442e-05,
      "loss": 4.4913,
      "step": 8962
    },
    {
      "epoch": 0.3618490109002826,
      "grad_norm": 83.72960662841797,
      "learning_rate": 1.4759958126371322e-05,
      "loss": 6.2457,
      "step": 8963
    },
    {
      "epoch": 0.3618893823173193,
      "grad_norm": 83.39492797851562,
      "learning_rate": 1.4758808137314053e-05,
      "loss": 4.8916,
      "step": 8964
    },
    {
      "epoch": 0.3619297537343561,
      "grad_norm": 70.8938980102539,
      "learning_rate": 1.475765806689228e-05,
      "loss": 4.0964,
      "step": 8965
    },
    {
      "epoch": 0.3619701251513928,
      "grad_norm": 121.03455352783203,
      "learning_rate": 1.4756507915125663e-05,
      "loss": 5.533,
      "step": 8966
    },
    {
      "epoch": 0.36201049656842954,
      "grad_norm": 102.64643859863281,
      "learning_rate": 1.4755357682033867e-05,
      "loss": 5.652,
      "step": 8967
    },
    {
      "epoch": 0.36205086798546626,
      "grad_norm": 113.41828155517578,
      "learning_rate": 1.4754207367636557e-05,
      "loss": 5.8533,
      "step": 8968
    },
    {
      "epoch": 0.36209123940250304,
      "grad_norm": 88.078369140625,
      "learning_rate": 1.4753056971953403e-05,
      "loss": 4.9643,
      "step": 8969
    },
    {
      "epoch": 0.36213161081953976,
      "grad_norm": 87.36219787597656,
      "learning_rate": 1.4751906495004077e-05,
      "loss": 5.9124,
      "step": 8970
    },
    {
      "epoch": 0.3621719822365765,
      "grad_norm": 88.1889877319336,
      "learning_rate": 1.4750755936808246e-05,
      "loss": 3.9888,
      "step": 8971
    },
    {
      "epoch": 0.36221235365361326,
      "grad_norm": 174.4679718017578,
      "learning_rate": 1.474960529738558e-05,
      "loss": 6.1284,
      "step": 8972
    },
    {
      "epoch": 0.36225272507065,
      "grad_norm": 109.48236083984375,
      "learning_rate": 1.4748454576755759e-05,
      "loss": 5.8104,
      "step": 8973
    },
    {
      "epoch": 0.3622930964876867,
      "grad_norm": 129.80215454101562,
      "learning_rate": 1.4747303774938448e-05,
      "loss": 4.2455,
      "step": 8974
    },
    {
      "epoch": 0.3623334679047235,
      "grad_norm": 89.73558044433594,
      "learning_rate": 1.4746152891953334e-05,
      "loss": 5.7322,
      "step": 8975
    },
    {
      "epoch": 0.3623738393217602,
      "grad_norm": 222.62257385253906,
      "learning_rate": 1.4745001927820085e-05,
      "loss": 5.3553,
      "step": 8976
    },
    {
      "epoch": 0.3624142107387969,
      "grad_norm": 90.55824279785156,
      "learning_rate": 1.4743850882558381e-05,
      "loss": 4.7813,
      "step": 8977
    },
    {
      "epoch": 0.36245458215583365,
      "grad_norm": 123.80171203613281,
      "learning_rate": 1.4742699756187907e-05,
      "loss": 5.1306,
      "step": 8978
    },
    {
      "epoch": 0.3624949535728704,
      "grad_norm": 88.43264770507812,
      "learning_rate": 1.4741548548728344e-05,
      "loss": 5.574,
      "step": 8979
    },
    {
      "epoch": 0.36253532498990715,
      "grad_norm": 149.3486785888672,
      "learning_rate": 1.474039726019937e-05,
      "loss": 5.1483,
      "step": 8980
    },
    {
      "epoch": 0.36257569640694387,
      "grad_norm": 69.01344299316406,
      "learning_rate": 1.4739245890620674e-05,
      "loss": 3.5943,
      "step": 8981
    },
    {
      "epoch": 0.36261606782398065,
      "grad_norm": 81.91552734375,
      "learning_rate": 1.4738094440011941e-05,
      "loss": 6.5331,
      "step": 8982
    },
    {
      "epoch": 0.36265643924101737,
      "grad_norm": 108.1187515258789,
      "learning_rate": 1.4736942908392857e-05,
      "loss": 5.2171,
      "step": 8983
    },
    {
      "epoch": 0.3626968106580541,
      "grad_norm": 104.42694091796875,
      "learning_rate": 1.473579129578311e-05,
      "loss": 5.6172,
      "step": 8984
    },
    {
      "epoch": 0.3627371820750908,
      "grad_norm": 118.4493637084961,
      "learning_rate": 1.4734639602202392e-05,
      "loss": 5.2374,
      "step": 8985
    },
    {
      "epoch": 0.3627775534921276,
      "grad_norm": 109.19994354248047,
      "learning_rate": 1.4733487827670392e-05,
      "loss": 6.1042,
      "step": 8986
    },
    {
      "epoch": 0.3628179249091643,
      "grad_norm": 67.59001159667969,
      "learning_rate": 1.4732335972206804e-05,
      "loss": 4.5739,
      "step": 8987
    },
    {
      "epoch": 0.36285829632620104,
      "grad_norm": 115.67091369628906,
      "learning_rate": 1.473118403583132e-05,
      "loss": 5.8538,
      "step": 8988
    },
    {
      "epoch": 0.3628986677432378,
      "grad_norm": 102.19842529296875,
      "learning_rate": 1.473003201856364e-05,
      "loss": 5.0216,
      "step": 8989
    },
    {
      "epoch": 0.36293903916027453,
      "grad_norm": 98.3726806640625,
      "learning_rate": 1.4728879920423456e-05,
      "loss": 4.6109,
      "step": 8990
    },
    {
      "epoch": 0.36297941057731126,
      "grad_norm": 178.47506713867188,
      "learning_rate": 1.4727727741430468e-05,
      "loss": 5.9484,
      "step": 8991
    },
    {
      "epoch": 0.363019781994348,
      "grad_norm": 91.07086181640625,
      "learning_rate": 1.4726575481604378e-05,
      "loss": 5.2148,
      "step": 8992
    },
    {
      "epoch": 0.36306015341138476,
      "grad_norm": 98.02032470703125,
      "learning_rate": 1.4725423140964882e-05,
      "loss": 5.0402,
      "step": 8993
    },
    {
      "epoch": 0.3631005248284215,
      "grad_norm": 268.3623046875,
      "learning_rate": 1.4724270719531685e-05,
      "loss": 7.3014,
      "step": 8994
    },
    {
      "epoch": 0.3631408962454582,
      "grad_norm": 109.3341293334961,
      "learning_rate": 1.4723118217324492e-05,
      "loss": 5.2865,
      "step": 8995
    },
    {
      "epoch": 0.363181267662495,
      "grad_norm": 137.3427734375,
      "learning_rate": 1.4721965634363006e-05,
      "loss": 6.2008,
      "step": 8996
    },
    {
      "epoch": 0.3632216390795317,
      "grad_norm": 100.88484954833984,
      "learning_rate": 1.4720812970666933e-05,
      "loss": 6.4704,
      "step": 8997
    },
    {
      "epoch": 0.3632620104965684,
      "grad_norm": 98.14071655273438,
      "learning_rate": 1.4719660226255983e-05,
      "loss": 5.6072,
      "step": 8998
    },
    {
      "epoch": 0.36330238191360514,
      "grad_norm": 117.62156677246094,
      "learning_rate": 1.4718507401149866e-05,
      "loss": 6.4805,
      "step": 8999
    },
    {
      "epoch": 0.3633427533306419,
      "grad_norm": 126.17823791503906,
      "learning_rate": 1.471735449536829e-05,
      "loss": 5.5507,
      "step": 9000
    },
    {
      "epoch": 0.36338312474767864,
      "grad_norm": 328.95611572265625,
      "learning_rate": 1.471620150893097e-05,
      "loss": 2.5457,
      "step": 9001
    },
    {
      "epoch": 0.36342349616471536,
      "grad_norm": 80.88835144042969,
      "learning_rate": 1.4715048441857614e-05,
      "loss": 5.0602,
      "step": 9002
    },
    {
      "epoch": 0.36346386758175214,
      "grad_norm": 80.59326171875,
      "learning_rate": 1.4713895294167942e-05,
      "loss": 5.8233,
      "step": 9003
    },
    {
      "epoch": 0.36350423899878886,
      "grad_norm": 315.1146240234375,
      "learning_rate": 1.4712742065881664e-05,
      "loss": 4.042,
      "step": 9004
    },
    {
      "epoch": 0.3635446104158256,
      "grad_norm": 100.86262512207031,
      "learning_rate": 1.4711588757018507e-05,
      "loss": 4.7031,
      "step": 9005
    },
    {
      "epoch": 0.3635849818328623,
      "grad_norm": 84.7246322631836,
      "learning_rate": 1.4710435367598184e-05,
      "loss": 4.079,
      "step": 9006
    },
    {
      "epoch": 0.3636253532498991,
      "grad_norm": 76.48170471191406,
      "learning_rate": 1.4709281897640413e-05,
      "loss": 4.8633,
      "step": 9007
    },
    {
      "epoch": 0.3636657246669358,
      "grad_norm": 63.66037368774414,
      "learning_rate": 1.4708128347164918e-05,
      "loss": 3.9424,
      "step": 9008
    },
    {
      "epoch": 0.36370609608397253,
      "grad_norm": 179.9181671142578,
      "learning_rate": 1.4706974716191426e-05,
      "loss": 4.9212,
      "step": 9009
    },
    {
      "epoch": 0.3637464675010093,
      "grad_norm": 86.51880645751953,
      "learning_rate": 1.4705821004739656e-05,
      "loss": 6.602,
      "step": 9010
    },
    {
      "epoch": 0.36378683891804603,
      "grad_norm": 92.89742279052734,
      "learning_rate": 1.4704667212829339e-05,
      "loss": 6.1294,
      "step": 9011
    },
    {
      "epoch": 0.36382721033508275,
      "grad_norm": 102.08514404296875,
      "learning_rate": 1.4703513340480194e-05,
      "loss": 4.8222,
      "step": 9012
    },
    {
      "epoch": 0.3638675817521195,
      "grad_norm": 83.52067565917969,
      "learning_rate": 1.4702359387711957e-05,
      "loss": 4.9436,
      "step": 9013
    },
    {
      "epoch": 0.36390795316915625,
      "grad_norm": 129.10183715820312,
      "learning_rate": 1.4701205354544352e-05,
      "loss": 5.6421,
      "step": 9014
    },
    {
      "epoch": 0.363948324586193,
      "grad_norm": 97.33811950683594,
      "learning_rate": 1.4700051240997119e-05,
      "loss": 5.417,
      "step": 9015
    },
    {
      "epoch": 0.3639886960032297,
      "grad_norm": 128.78614807128906,
      "learning_rate": 1.4698897047089981e-05,
      "loss": 7.2229,
      "step": 9016
    },
    {
      "epoch": 0.36402906742026647,
      "grad_norm": 120.71571350097656,
      "learning_rate": 1.4697742772842678e-05,
      "loss": 5.3054,
      "step": 9017
    },
    {
      "epoch": 0.3640694388373032,
      "grad_norm": 79.0307388305664,
      "learning_rate": 1.4696588418274942e-05,
      "loss": 6.1794,
      "step": 9018
    },
    {
      "epoch": 0.3641098102543399,
      "grad_norm": 147.204345703125,
      "learning_rate": 1.4695433983406512e-05,
      "loss": 5.7567,
      "step": 9019
    },
    {
      "epoch": 0.36415018167137664,
      "grad_norm": 86.9385757446289,
      "learning_rate": 1.4694279468257127e-05,
      "loss": 5.4497,
      "step": 9020
    },
    {
      "epoch": 0.3641905530884134,
      "grad_norm": 124.550048828125,
      "learning_rate": 1.4693124872846524e-05,
      "loss": 4.2525,
      "step": 9021
    },
    {
      "epoch": 0.36423092450545014,
      "grad_norm": 96.38630676269531,
      "learning_rate": 1.4691970197194444e-05,
      "loss": 5.5187,
      "step": 9022
    },
    {
      "epoch": 0.36427129592248686,
      "grad_norm": 122.5361099243164,
      "learning_rate": 1.4690815441320635e-05,
      "loss": 4.8097,
      "step": 9023
    },
    {
      "epoch": 0.36431166733952364,
      "grad_norm": 72.45814514160156,
      "learning_rate": 1.468966060524483e-05,
      "loss": 5.9218,
      "step": 9024
    },
    {
      "epoch": 0.36435203875656036,
      "grad_norm": 107.05131530761719,
      "learning_rate": 1.4688505688986783e-05,
      "loss": 6.4781,
      "step": 9025
    },
    {
      "epoch": 0.3643924101735971,
      "grad_norm": 99.11902618408203,
      "learning_rate": 1.4687350692566236e-05,
      "loss": 5.4343,
      "step": 9026
    },
    {
      "epoch": 0.36443278159063386,
      "grad_norm": 89.38426208496094,
      "learning_rate": 1.4686195616002937e-05,
      "loss": 3.4754,
      "step": 9027
    },
    {
      "epoch": 0.3644731530076706,
      "grad_norm": 126.6203842163086,
      "learning_rate": 1.4685040459316635e-05,
      "loss": 5.2519,
      "step": 9028
    },
    {
      "epoch": 0.3645135244247073,
      "grad_norm": 65.26992797851562,
      "learning_rate": 1.4683885222527084e-05,
      "loss": 5.0866,
      "step": 9029
    },
    {
      "epoch": 0.364553895841744,
      "grad_norm": 132.3690185546875,
      "learning_rate": 1.4682729905654031e-05,
      "loss": 4.4471,
      "step": 9030
    },
    {
      "epoch": 0.3645942672587808,
      "grad_norm": 87.98917388916016,
      "learning_rate": 1.4681574508717235e-05,
      "loss": 3.7909,
      "step": 9031
    },
    {
      "epoch": 0.3646346386758175,
      "grad_norm": 71.94014739990234,
      "learning_rate": 1.4680419031736445e-05,
      "loss": 5.5158,
      "step": 9032
    },
    {
      "epoch": 0.36467501009285425,
      "grad_norm": 93.13440704345703,
      "learning_rate": 1.4679263474731422e-05,
      "loss": 4.1264,
      "step": 9033
    },
    {
      "epoch": 0.364715381509891,
      "grad_norm": 87.00813293457031,
      "learning_rate": 1.4678107837721917e-05,
      "loss": 5.8147,
      "step": 9034
    },
    {
      "epoch": 0.36475575292692775,
      "grad_norm": 92.51112365722656,
      "learning_rate": 1.4676952120727695e-05,
      "loss": 4.4628,
      "step": 9035
    },
    {
      "epoch": 0.36479612434396447,
      "grad_norm": 111.7314453125,
      "learning_rate": 1.467579632376851e-05,
      "loss": 6.0598,
      "step": 9036
    },
    {
      "epoch": 0.3648364957610012,
      "grad_norm": 64.63682556152344,
      "learning_rate": 1.4674640446864127e-05,
      "loss": 5.2914,
      "step": 9037
    },
    {
      "epoch": 0.36487686717803797,
      "grad_norm": 106.46327209472656,
      "learning_rate": 1.4673484490034309e-05,
      "loss": 5.5788,
      "step": 9038
    },
    {
      "epoch": 0.3649172385950747,
      "grad_norm": 88.47378540039062,
      "learning_rate": 1.4672328453298821e-05,
      "loss": 4.5377,
      "step": 9039
    },
    {
      "epoch": 0.3649576100121114,
      "grad_norm": 99.94683837890625,
      "learning_rate": 1.4671172336677425e-05,
      "loss": 5.373,
      "step": 9040
    },
    {
      "epoch": 0.3649979814291482,
      "grad_norm": 81.63123321533203,
      "learning_rate": 1.4670016140189893e-05,
      "loss": 5.259,
      "step": 9041
    },
    {
      "epoch": 0.3650383528461849,
      "grad_norm": 94.28056335449219,
      "learning_rate": 1.4668859863855986e-05,
      "loss": 5.2453,
      "step": 9042
    },
    {
      "epoch": 0.36507872426322163,
      "grad_norm": 212.68115234375,
      "learning_rate": 1.466770350769548e-05,
      "loss": 5.5647,
      "step": 9043
    },
    {
      "epoch": 0.36511909568025835,
      "grad_norm": 72.87661743164062,
      "learning_rate": 1.4666547071728145e-05,
      "loss": 5.2028,
      "step": 9044
    },
    {
      "epoch": 0.36515946709729513,
      "grad_norm": 94.33318328857422,
      "learning_rate": 1.4665390555973752e-05,
      "loss": 4.6721,
      "step": 9045
    },
    {
      "epoch": 0.36519983851433185,
      "grad_norm": 129.5873565673828,
      "learning_rate": 1.4664233960452075e-05,
      "loss": 4.9488,
      "step": 9046
    },
    {
      "epoch": 0.3652402099313686,
      "grad_norm": 92.46648406982422,
      "learning_rate": 1.4663077285182887e-05,
      "loss": 4.3474,
      "step": 9047
    },
    {
      "epoch": 0.36528058134840535,
      "grad_norm": 99.1619644165039,
      "learning_rate": 1.4661920530185968e-05,
      "loss": 5.4786,
      "step": 9048
    },
    {
      "epoch": 0.3653209527654421,
      "grad_norm": 101.89102935791016,
      "learning_rate": 1.4660763695481095e-05,
      "loss": 5.6092,
      "step": 9049
    },
    {
      "epoch": 0.3653613241824788,
      "grad_norm": 111.09423828125,
      "learning_rate": 1.4659606781088045e-05,
      "loss": 5.9516,
      "step": 9050
    },
    {
      "epoch": 0.3654016955995155,
      "grad_norm": 166.3105926513672,
      "learning_rate": 1.4658449787026603e-05,
      "loss": 6.2096,
      "step": 9051
    },
    {
      "epoch": 0.3654420670165523,
      "grad_norm": 126.53179168701172,
      "learning_rate": 1.4657292713316544e-05,
      "loss": 4.6833,
      "step": 9052
    },
    {
      "epoch": 0.365482438433589,
      "grad_norm": 110.3216552734375,
      "learning_rate": 1.4656135559977658e-05,
      "loss": 2.9983,
      "step": 9053
    },
    {
      "epoch": 0.36552280985062574,
      "grad_norm": 122.03366088867188,
      "learning_rate": 1.4654978327029725e-05,
      "loss": 4.6727,
      "step": 9054
    },
    {
      "epoch": 0.3655631812676625,
      "grad_norm": 91.18334197998047,
      "learning_rate": 1.4653821014492535e-05,
      "loss": 3.7747,
      "step": 9055
    },
    {
      "epoch": 0.36560355268469924,
      "grad_norm": 73.5594711303711,
      "learning_rate": 1.4652663622385869e-05,
      "loss": 4.5587,
      "step": 9056
    },
    {
      "epoch": 0.36564392410173596,
      "grad_norm": 91.19532012939453,
      "learning_rate": 1.4651506150729521e-05,
      "loss": 4.9677,
      "step": 9057
    },
    {
      "epoch": 0.3656842955187727,
      "grad_norm": 119.68102264404297,
      "learning_rate": 1.4650348599543282e-05,
      "loss": 3.7816,
      "step": 9058
    },
    {
      "epoch": 0.36572466693580946,
      "grad_norm": 114.44501495361328,
      "learning_rate": 1.4649190968846942e-05,
      "loss": 3.9752,
      "step": 9059
    },
    {
      "epoch": 0.3657650383528462,
      "grad_norm": 71.28126525878906,
      "learning_rate": 1.464803325866029e-05,
      "loss": 3.7787,
      "step": 9060
    },
    {
      "epoch": 0.3658054097698829,
      "grad_norm": 122.7109375,
      "learning_rate": 1.4646875469003126e-05,
      "loss": 5.2852,
      "step": 9061
    },
    {
      "epoch": 0.3658457811869197,
      "grad_norm": 102.49755859375,
      "learning_rate": 1.4645717599895241e-05,
      "loss": 4.3577,
      "step": 9062
    },
    {
      "epoch": 0.3658861526039564,
      "grad_norm": 77.97149658203125,
      "learning_rate": 1.4644559651356435e-05,
      "loss": 5.2956,
      "step": 9063
    },
    {
      "epoch": 0.3659265240209931,
      "grad_norm": 99.4234848022461,
      "learning_rate": 1.4643401623406504e-05,
      "loss": 5.3978,
      "step": 9064
    },
    {
      "epoch": 0.36596689543802985,
      "grad_norm": 107.54032897949219,
      "learning_rate": 1.464224351606525e-05,
      "loss": 5.0026,
      "step": 9065
    },
    {
      "epoch": 0.3660072668550666,
      "grad_norm": 104.46283721923828,
      "learning_rate": 1.4641085329352466e-05,
      "loss": 6.4755,
      "step": 9066
    },
    {
      "epoch": 0.36604763827210335,
      "grad_norm": 73.15910339355469,
      "learning_rate": 1.4639927063287966e-05,
      "loss": 5.1816,
      "step": 9067
    },
    {
      "epoch": 0.36608800968914007,
      "grad_norm": 104.14559936523438,
      "learning_rate": 1.4638768717891546e-05,
      "loss": 4.0387,
      "step": 9068
    },
    {
      "epoch": 0.36612838110617685,
      "grad_norm": 99.21161651611328,
      "learning_rate": 1.4637610293183015e-05,
      "loss": 5.8764,
      "step": 9069
    },
    {
      "epoch": 0.36616875252321357,
      "grad_norm": 63.362003326416016,
      "learning_rate": 1.4636451789182178e-05,
      "loss": 4.6343,
      "step": 9070
    },
    {
      "epoch": 0.3662091239402503,
      "grad_norm": 84.56118774414062,
      "learning_rate": 1.4635293205908842e-05,
      "loss": 5.4884,
      "step": 9071
    },
    {
      "epoch": 0.366249495357287,
      "grad_norm": 174.1509246826172,
      "learning_rate": 1.4634134543382814e-05,
      "loss": 4.4557,
      "step": 9072
    },
    {
      "epoch": 0.3662898667743238,
      "grad_norm": 161.18417358398438,
      "learning_rate": 1.463297580162391e-05,
      "loss": 4.9601,
      "step": 9073
    },
    {
      "epoch": 0.3663302381913605,
      "grad_norm": 85.74512481689453,
      "learning_rate": 1.4631816980651939e-05,
      "loss": 4.57,
      "step": 9074
    },
    {
      "epoch": 0.36637060960839724,
      "grad_norm": 82.70939636230469,
      "learning_rate": 1.4630658080486711e-05,
      "loss": 5.0531,
      "step": 9075
    },
    {
      "epoch": 0.366410981025434,
      "grad_norm": 103.69824981689453,
      "learning_rate": 1.4629499101148043e-05,
      "loss": 5.1572,
      "step": 9076
    },
    {
      "epoch": 0.36645135244247073,
      "grad_norm": 99.34520721435547,
      "learning_rate": 1.462834004265575e-05,
      "loss": 5.9682,
      "step": 9077
    },
    {
      "epoch": 0.36649172385950746,
      "grad_norm": 227.6074981689453,
      "learning_rate": 1.4627180905029654e-05,
      "loss": 5.1772,
      "step": 9078
    },
    {
      "epoch": 0.36653209527654423,
      "grad_norm": 110.72427368164062,
      "learning_rate": 1.4626021688289567e-05,
      "loss": 4.4871,
      "step": 9079
    },
    {
      "epoch": 0.36657246669358096,
      "grad_norm": 141.67494201660156,
      "learning_rate": 1.4624862392455312e-05,
      "loss": 5.1965,
      "step": 9080
    },
    {
      "epoch": 0.3666128381106177,
      "grad_norm": 82.58264923095703,
      "learning_rate": 1.4623703017546713e-05,
      "loss": 5.031,
      "step": 9081
    },
    {
      "epoch": 0.3666532095276544,
      "grad_norm": 169.9984893798828,
      "learning_rate": 1.4622543563583587e-05,
      "loss": 6.0795,
      "step": 9082
    },
    {
      "epoch": 0.3666935809446912,
      "grad_norm": 148.84100341796875,
      "learning_rate": 1.4621384030585762e-05,
      "loss": 5.8806,
      "step": 9083
    },
    {
      "epoch": 0.3667339523617279,
      "grad_norm": 144.59613037109375,
      "learning_rate": 1.4620224418573059e-05,
      "loss": 5.2183,
      "step": 9084
    },
    {
      "epoch": 0.3667743237787646,
      "grad_norm": 77.37709045410156,
      "learning_rate": 1.4619064727565308e-05,
      "loss": 4.852,
      "step": 9085
    },
    {
      "epoch": 0.3668146951958014,
      "grad_norm": 104.02168273925781,
      "learning_rate": 1.4617904957582336e-05,
      "loss": 4.4848,
      "step": 9086
    },
    {
      "epoch": 0.3668550666128381,
      "grad_norm": 111.95872497558594,
      "learning_rate": 1.4616745108643974e-05,
      "loss": 5.1969,
      "step": 9087
    },
    {
      "epoch": 0.36689543802987484,
      "grad_norm": 118.86318969726562,
      "learning_rate": 1.4615585180770052e-05,
      "loss": 6.246,
      "step": 9088
    },
    {
      "epoch": 0.36693580944691156,
      "grad_norm": 78.96199035644531,
      "learning_rate": 1.4614425173980402e-05,
      "loss": 6.3909,
      "step": 9089
    },
    {
      "epoch": 0.36697618086394834,
      "grad_norm": 117.84820556640625,
      "learning_rate": 1.4613265088294854e-05,
      "loss": 6.3701,
      "step": 9090
    },
    {
      "epoch": 0.36701655228098506,
      "grad_norm": 159.16981506347656,
      "learning_rate": 1.461210492373325e-05,
      "loss": 2.9538,
      "step": 9091
    },
    {
      "epoch": 0.3670569236980218,
      "grad_norm": 115.59449005126953,
      "learning_rate": 1.461094468031542e-05,
      "loss": 4.4847,
      "step": 9092
    },
    {
      "epoch": 0.36709729511505856,
      "grad_norm": 105.5914306640625,
      "learning_rate": 1.4609784358061203e-05,
      "loss": 6.2897,
      "step": 9093
    },
    {
      "epoch": 0.3671376665320953,
      "grad_norm": 95.20335388183594,
      "learning_rate": 1.4608623956990437e-05,
      "loss": 4.5641,
      "step": 9094
    },
    {
      "epoch": 0.367178037949132,
      "grad_norm": 105.12592315673828,
      "learning_rate": 1.4607463477122965e-05,
      "loss": 5.1703,
      "step": 9095
    },
    {
      "epoch": 0.36721840936616873,
      "grad_norm": 117.83466339111328,
      "learning_rate": 1.4606302918478626e-05,
      "loss": 4.556,
      "step": 9096
    },
    {
      "epoch": 0.3672587807832055,
      "grad_norm": 116.29485321044922,
      "learning_rate": 1.4605142281077265e-05,
      "loss": 4.7764,
      "step": 9097
    },
    {
      "epoch": 0.36729915220024223,
      "grad_norm": 125.23332214355469,
      "learning_rate": 1.4603981564938723e-05,
      "loss": 4.5592,
      "step": 9098
    },
    {
      "epoch": 0.36733952361727895,
      "grad_norm": 68.51387786865234,
      "learning_rate": 1.4602820770082848e-05,
      "loss": 5.0079,
      "step": 9099
    },
    {
      "epoch": 0.36737989503431573,
      "grad_norm": 85.21415710449219,
      "learning_rate": 1.4601659896529486e-05,
      "loss": 5.0877,
      "step": 9100
    },
    {
      "epoch": 0.36742026645135245,
      "grad_norm": 150.8819122314453,
      "learning_rate": 1.4600498944298484e-05,
      "loss": 5.3651,
      "step": 9101
    },
    {
      "epoch": 0.3674606378683892,
      "grad_norm": 92.39421844482422,
      "learning_rate": 1.4599337913409695e-05,
      "loss": 5.7914,
      "step": 9102
    },
    {
      "epoch": 0.3675010092854259,
      "grad_norm": 86.08396911621094,
      "learning_rate": 1.459817680388297e-05,
      "loss": 5.9636,
      "step": 9103
    },
    {
      "epoch": 0.36754138070246267,
      "grad_norm": 118.96344757080078,
      "learning_rate": 1.4597015615738155e-05,
      "loss": 3.9096,
      "step": 9104
    },
    {
      "epoch": 0.3675817521194994,
      "grad_norm": 118.74713897705078,
      "learning_rate": 1.4595854348995109e-05,
      "loss": 6.1627,
      "step": 9105
    },
    {
      "epoch": 0.3676221235365361,
      "grad_norm": 86.7867660522461,
      "learning_rate": 1.4594693003673684e-05,
      "loss": 5.8073,
      "step": 9106
    },
    {
      "epoch": 0.3676624949535729,
      "grad_norm": 81.47614288330078,
      "learning_rate": 1.459353157979374e-05,
      "loss": 5.1994,
      "step": 9107
    },
    {
      "epoch": 0.3677028663706096,
      "grad_norm": 101.66119384765625,
      "learning_rate": 1.4592370077375132e-05,
      "loss": 3.5952,
      "step": 9108
    },
    {
      "epoch": 0.36774323778764634,
      "grad_norm": 199.33526611328125,
      "learning_rate": 1.4591208496437723e-05,
      "loss": 5.9649,
      "step": 9109
    },
    {
      "epoch": 0.36778360920468306,
      "grad_norm": 90.71601104736328,
      "learning_rate": 1.4590046837001367e-05,
      "loss": 6.5042,
      "step": 9110
    },
    {
      "epoch": 0.36782398062171984,
      "grad_norm": 70.74453735351562,
      "learning_rate": 1.4588885099085927e-05,
      "loss": 5.5635,
      "step": 9111
    },
    {
      "epoch": 0.36786435203875656,
      "grad_norm": 133.18704223632812,
      "learning_rate": 1.458772328271127e-05,
      "loss": 4.3318,
      "step": 9112
    },
    {
      "epoch": 0.3679047234557933,
      "grad_norm": 81.18244171142578,
      "learning_rate": 1.458656138789726e-05,
      "loss": 4.369,
      "step": 9113
    },
    {
      "epoch": 0.36794509487283006,
      "grad_norm": 61.66715621948242,
      "learning_rate": 1.4585399414663756e-05,
      "loss": 4.7084,
      "step": 9114
    },
    {
      "epoch": 0.3679854662898668,
      "grad_norm": 74.97689819335938,
      "learning_rate": 1.4584237363030631e-05,
      "loss": 4.8365,
      "step": 9115
    },
    {
      "epoch": 0.3680258377069035,
      "grad_norm": 82.46159362792969,
      "learning_rate": 1.4583075233017752e-05,
      "loss": 5.308,
      "step": 9116
    },
    {
      "epoch": 0.3680662091239402,
      "grad_norm": 87.768310546875,
      "learning_rate": 1.4581913024644992e-05,
      "loss": 5.2207,
      "step": 9117
    },
    {
      "epoch": 0.368106580540977,
      "grad_norm": 164.2787628173828,
      "learning_rate": 1.4580750737932215e-05,
      "loss": 6.3465,
      "step": 9118
    },
    {
      "epoch": 0.3681469519580137,
      "grad_norm": 85.4842529296875,
      "learning_rate": 1.45795883728993e-05,
      "loss": 4.6544,
      "step": 9119
    },
    {
      "epoch": 0.36818732337505045,
      "grad_norm": 102.91126251220703,
      "learning_rate": 1.4578425929566116e-05,
      "loss": 5.5229,
      "step": 9120
    },
    {
      "epoch": 0.3682276947920872,
      "grad_norm": 100.22598266601562,
      "learning_rate": 1.457726340795254e-05,
      "loss": 5.9388,
      "step": 9121
    },
    {
      "epoch": 0.36826806620912395,
      "grad_norm": 109.01081085205078,
      "learning_rate": 1.4576100808078448e-05,
      "loss": 3.2103,
      "step": 9122
    },
    {
      "epoch": 0.36830843762616067,
      "grad_norm": 61.327239990234375,
      "learning_rate": 1.457493812996372e-05,
      "loss": 7.4704,
      "step": 9123
    },
    {
      "epoch": 0.3683488090431974,
      "grad_norm": 67.06983184814453,
      "learning_rate": 1.4573775373628235e-05,
      "loss": 3.9094,
      "step": 9124
    },
    {
      "epoch": 0.36838918046023417,
      "grad_norm": 132.6919708251953,
      "learning_rate": 1.4572612539091866e-05,
      "loss": 5.8582,
      "step": 9125
    },
    {
      "epoch": 0.3684295518772709,
      "grad_norm": 71.32010650634766,
      "learning_rate": 1.4571449626374503e-05,
      "loss": 5.0099,
      "step": 9126
    },
    {
      "epoch": 0.3684699232943076,
      "grad_norm": 196.62725830078125,
      "learning_rate": 1.4570286635496027e-05,
      "loss": 5.914,
      "step": 9127
    },
    {
      "epoch": 0.3685102947113444,
      "grad_norm": 149.19973754882812,
      "learning_rate": 1.4569123566476321e-05,
      "loss": 6.7683,
      "step": 9128
    },
    {
      "epoch": 0.3685506661283811,
      "grad_norm": 88.63997650146484,
      "learning_rate": 1.4567960419335274e-05,
      "loss": 4.8455,
      "step": 9129
    },
    {
      "epoch": 0.36859103754541783,
      "grad_norm": 82.58148193359375,
      "learning_rate": 1.4566797194092768e-05,
      "loss": 6.2275,
      "step": 9130
    },
    {
      "epoch": 0.3686314089624546,
      "grad_norm": 76.43982696533203,
      "learning_rate": 1.4565633890768697e-05,
      "loss": 6.4334,
      "step": 9131
    },
    {
      "epoch": 0.36867178037949133,
      "grad_norm": 93.46839141845703,
      "learning_rate": 1.4564470509382947e-05,
      "loss": 6.7768,
      "step": 9132
    },
    {
      "epoch": 0.36871215179652805,
      "grad_norm": 82.80628967285156,
      "learning_rate": 1.456330704995541e-05,
      "loss": 5.0296,
      "step": 9133
    },
    {
      "epoch": 0.3687525232135648,
      "grad_norm": 109.06607818603516,
      "learning_rate": 1.4562143512505977e-05,
      "loss": 4.9552,
      "step": 9134
    },
    {
      "epoch": 0.36879289463060155,
      "grad_norm": 125.49176788330078,
      "learning_rate": 1.4560979897054543e-05,
      "loss": 5.0714,
      "step": 9135
    },
    {
      "epoch": 0.3688332660476383,
      "grad_norm": 92.74117279052734,
      "learning_rate": 1.4559816203621004e-05,
      "loss": 5.6963,
      "step": 9136
    },
    {
      "epoch": 0.368873637464675,
      "grad_norm": 77.05329895019531,
      "learning_rate": 1.4558652432225257e-05,
      "loss": 6.0393,
      "step": 9137
    },
    {
      "epoch": 0.3689140088817118,
      "grad_norm": 121.01712036132812,
      "learning_rate": 1.4557488582887198e-05,
      "loss": 5.41,
      "step": 9138
    },
    {
      "epoch": 0.3689543802987485,
      "grad_norm": 99.40563201904297,
      "learning_rate": 1.455632465562673e-05,
      "loss": 5.9031,
      "step": 9139
    },
    {
      "epoch": 0.3689947517157852,
      "grad_norm": 88.4354248046875,
      "learning_rate": 1.4555160650463744e-05,
      "loss": 6.1915,
      "step": 9140
    },
    {
      "epoch": 0.36903512313282194,
      "grad_norm": 70.0784683227539,
      "learning_rate": 1.4553996567418155e-05,
      "loss": 3.5276,
      "step": 9141
    },
    {
      "epoch": 0.3690754945498587,
      "grad_norm": 82.90131378173828,
      "learning_rate": 1.4552832406509854e-05,
      "loss": 5.2638,
      "step": 9142
    },
    {
      "epoch": 0.36911586596689544,
      "grad_norm": 70.04621124267578,
      "learning_rate": 1.4551668167758752e-05,
      "loss": 5.0681,
      "step": 9143
    },
    {
      "epoch": 0.36915623738393216,
      "grad_norm": 93.2026596069336,
      "learning_rate": 1.4550503851184753e-05,
      "loss": 4.8098,
      "step": 9144
    },
    {
      "epoch": 0.36919660880096894,
      "grad_norm": 174.0178680419922,
      "learning_rate": 1.4549339456807763e-05,
      "loss": 6.8376,
      "step": 9145
    },
    {
      "epoch": 0.36923698021800566,
      "grad_norm": 110.79251098632812,
      "learning_rate": 1.4548174984647691e-05,
      "loss": 5.8586,
      "step": 9146
    },
    {
      "epoch": 0.3692773516350424,
      "grad_norm": 151.4393310546875,
      "learning_rate": 1.4547010434724451e-05,
      "loss": 5.2236,
      "step": 9147
    },
    {
      "epoch": 0.3693177230520791,
      "grad_norm": 86.1923599243164,
      "learning_rate": 1.4545845807057948e-05,
      "loss": 5.8115,
      "step": 9148
    },
    {
      "epoch": 0.3693580944691159,
      "grad_norm": 78.03727722167969,
      "learning_rate": 1.4544681101668099e-05,
      "loss": 4.65,
      "step": 9149
    },
    {
      "epoch": 0.3693984658861526,
      "grad_norm": 71.49221801757812,
      "learning_rate": 1.4543516318574815e-05,
      "loss": 4.6888,
      "step": 9150
    },
    {
      "epoch": 0.3694388373031893,
      "grad_norm": 85.96395111083984,
      "learning_rate": 1.4542351457798012e-05,
      "loss": 6.5142,
      "step": 9151
    },
    {
      "epoch": 0.3694792087202261,
      "grad_norm": 73.5143051147461,
      "learning_rate": 1.4541186519357607e-05,
      "loss": 6.4029,
      "step": 9152
    },
    {
      "epoch": 0.3695195801372628,
      "grad_norm": 87.04914093017578,
      "learning_rate": 1.4540021503273517e-05,
      "loss": 3.6969,
      "step": 9153
    },
    {
      "epoch": 0.36955995155429955,
      "grad_norm": 115.88320922851562,
      "learning_rate": 1.4538856409565658e-05,
      "loss": 5.2885,
      "step": 9154
    },
    {
      "epoch": 0.36960032297133627,
      "grad_norm": 78.45367431640625,
      "learning_rate": 1.4537691238253954e-05,
      "loss": 5.3241,
      "step": 9155
    },
    {
      "epoch": 0.36964069438837305,
      "grad_norm": 95.06685638427734,
      "learning_rate": 1.4536525989358325e-05,
      "loss": 6.0007,
      "step": 9156
    },
    {
      "epoch": 0.36968106580540977,
      "grad_norm": 134.7685546875,
      "learning_rate": 1.4535360662898702e-05,
      "loss": 6.4326,
      "step": 9157
    },
    {
      "epoch": 0.3697214372224465,
      "grad_norm": 94.73878479003906,
      "learning_rate": 1.4534195258894997e-05,
      "loss": 4.877,
      "step": 9158
    },
    {
      "epoch": 0.36976180863948327,
      "grad_norm": 141.98638916015625,
      "learning_rate": 1.4533029777367143e-05,
      "loss": 5.6334,
      "step": 9159
    },
    {
      "epoch": 0.36980218005652,
      "grad_norm": 92.76643371582031,
      "learning_rate": 1.4531864218335062e-05,
      "loss": 3.9062,
      "step": 9160
    },
    {
      "epoch": 0.3698425514735567,
      "grad_norm": 122.71207427978516,
      "learning_rate": 1.4530698581818689e-05,
      "loss": 5.8166,
      "step": 9161
    },
    {
      "epoch": 0.36988292289059344,
      "grad_norm": 77.29754638671875,
      "learning_rate": 1.452953286783795e-05,
      "loss": 4.8078,
      "step": 9162
    },
    {
      "epoch": 0.3699232943076302,
      "grad_norm": 79.37577819824219,
      "learning_rate": 1.4528367076412775e-05,
      "loss": 5.1147,
      "step": 9163
    },
    {
      "epoch": 0.36996366572466693,
      "grad_norm": 76.23871612548828,
      "learning_rate": 1.4527201207563098e-05,
      "loss": 4.3588,
      "step": 9164
    },
    {
      "epoch": 0.37000403714170366,
      "grad_norm": 95.64395904541016,
      "learning_rate": 1.4526035261308851e-05,
      "loss": 4.0462,
      "step": 9165
    },
    {
      "epoch": 0.37004440855874043,
      "grad_norm": 165.86480712890625,
      "learning_rate": 1.452486923766997e-05,
      "loss": 6.0533,
      "step": 9166
    },
    {
      "epoch": 0.37008477997577716,
      "grad_norm": 107.60386657714844,
      "learning_rate": 1.4523703136666392e-05,
      "loss": 4.4925,
      "step": 9167
    },
    {
      "epoch": 0.3701251513928139,
      "grad_norm": 108.59225463867188,
      "learning_rate": 1.4522536958318054e-05,
      "loss": 4.4011,
      "step": 9168
    },
    {
      "epoch": 0.3701655228098506,
      "grad_norm": 231.34307861328125,
      "learning_rate": 1.4521370702644896e-05,
      "loss": 4.2518,
      "step": 9169
    },
    {
      "epoch": 0.3702058942268874,
      "grad_norm": 82.6387939453125,
      "learning_rate": 1.4520204369666857e-05,
      "loss": 6.7523,
      "step": 9170
    },
    {
      "epoch": 0.3702462656439241,
      "grad_norm": 92.0118637084961,
      "learning_rate": 1.451903795940388e-05,
      "loss": 5.903,
      "step": 9171
    },
    {
      "epoch": 0.3702866370609608,
      "grad_norm": 83.41516876220703,
      "learning_rate": 1.4517871471875901e-05,
      "loss": 5.2503,
      "step": 9172
    },
    {
      "epoch": 0.3703270084779976,
      "grad_norm": 113.1795425415039,
      "learning_rate": 1.4516704907102876e-05,
      "loss": 4.7842,
      "step": 9173
    },
    {
      "epoch": 0.3703673798950343,
      "grad_norm": 81.84233856201172,
      "learning_rate": 1.451553826510474e-05,
      "loss": 6.4157,
      "step": 9174
    },
    {
      "epoch": 0.37040775131207104,
      "grad_norm": 85.20685577392578,
      "learning_rate": 1.4514371545901447e-05,
      "loss": 5.3738,
      "step": 9175
    },
    {
      "epoch": 0.37044812272910776,
      "grad_norm": 118.04676055908203,
      "learning_rate": 1.4513204749512939e-05,
      "loss": 5.1887,
      "step": 9176
    },
    {
      "epoch": 0.37048849414614454,
      "grad_norm": 74.3030014038086,
      "learning_rate": 1.4512037875959174e-05,
      "loss": 5.0032,
      "step": 9177
    },
    {
      "epoch": 0.37052886556318126,
      "grad_norm": 70.95153045654297,
      "learning_rate": 1.4510870925260094e-05,
      "loss": 4.6055,
      "step": 9178
    },
    {
      "epoch": 0.370569236980218,
      "grad_norm": 114.38899230957031,
      "learning_rate": 1.4509703897435656e-05,
      "loss": 4.5221,
      "step": 9179
    },
    {
      "epoch": 0.37060960839725476,
      "grad_norm": 77.06072998046875,
      "learning_rate": 1.4508536792505811e-05,
      "loss": 4.7379,
      "step": 9180
    },
    {
      "epoch": 0.3706499798142915,
      "grad_norm": 113.09809112548828,
      "learning_rate": 1.4507369610490518e-05,
      "loss": 4.7613,
      "step": 9181
    },
    {
      "epoch": 0.3706903512313282,
      "grad_norm": 102.9954833984375,
      "learning_rate": 1.4506202351409728e-05,
      "loss": 5.7438,
      "step": 9182
    },
    {
      "epoch": 0.370730722648365,
      "grad_norm": 80.6579360961914,
      "learning_rate": 1.45050350152834e-05,
      "loss": 6.0406,
      "step": 9183
    },
    {
      "epoch": 0.3707710940654017,
      "grad_norm": 125.0224609375,
      "learning_rate": 1.4503867602131494e-05,
      "loss": 6.3762,
      "step": 9184
    },
    {
      "epoch": 0.37081146548243843,
      "grad_norm": 103.00634002685547,
      "learning_rate": 1.4502700111973969e-05,
      "loss": 4.6655,
      "step": 9185
    },
    {
      "epoch": 0.37085183689947515,
      "grad_norm": 80.6435775756836,
      "learning_rate": 1.4501532544830785e-05,
      "loss": 5.8314,
      "step": 9186
    },
    {
      "epoch": 0.37089220831651193,
      "grad_norm": 83.35359954833984,
      "learning_rate": 1.4500364900721911e-05,
      "loss": 4.9743,
      "step": 9187
    },
    {
      "epoch": 0.37093257973354865,
      "grad_norm": 86.77404022216797,
      "learning_rate": 1.4499197179667301e-05,
      "loss": 5.1039,
      "step": 9188
    },
    {
      "epoch": 0.3709729511505854,
      "grad_norm": 211.37228393554688,
      "learning_rate": 1.449802938168693e-05,
      "loss": 4.4449,
      "step": 9189
    },
    {
      "epoch": 0.37101332256762215,
      "grad_norm": 83.48666381835938,
      "learning_rate": 1.449686150680076e-05,
      "loss": 4.9345,
      "step": 9190
    },
    {
      "epoch": 0.37105369398465887,
      "grad_norm": 132.28488159179688,
      "learning_rate": 1.4495693555028755e-05,
      "loss": 5.0752,
      "step": 9191
    },
    {
      "epoch": 0.3710940654016956,
      "grad_norm": 95.19615936279297,
      "learning_rate": 1.4494525526390892e-05,
      "loss": 6.1339,
      "step": 9192
    },
    {
      "epoch": 0.3711344368187323,
      "grad_norm": 134.0594940185547,
      "learning_rate": 1.4493357420907138e-05,
      "loss": 4.6038,
      "step": 9193
    },
    {
      "epoch": 0.3711748082357691,
      "grad_norm": 96.52983093261719,
      "learning_rate": 1.4492189238597463e-05,
      "loss": 4.7176,
      "step": 9194
    },
    {
      "epoch": 0.3712151796528058,
      "grad_norm": 147.96409606933594,
      "learning_rate": 1.4491020979481843e-05,
      "loss": 5.7562,
      "step": 9195
    },
    {
      "epoch": 0.37125555106984254,
      "grad_norm": 78.25296783447266,
      "learning_rate": 1.4489852643580248e-05,
      "loss": 3.7652,
      "step": 9196
    },
    {
      "epoch": 0.3712959224868793,
      "grad_norm": 86.01044464111328,
      "learning_rate": 1.4488684230912664e-05,
      "loss": 5.33,
      "step": 9197
    },
    {
      "epoch": 0.37133629390391604,
      "grad_norm": 108.56460571289062,
      "learning_rate": 1.4487515741499057e-05,
      "loss": 6.6998,
      "step": 9198
    },
    {
      "epoch": 0.37137666532095276,
      "grad_norm": 78.11882019042969,
      "learning_rate": 1.4486347175359412e-05,
      "loss": 5.2527,
      "step": 9199
    },
    {
      "epoch": 0.3714170367379895,
      "grad_norm": 122.14450073242188,
      "learning_rate": 1.4485178532513706e-05,
      "loss": 4.8597,
      "step": 9200
    },
    {
      "epoch": 0.37145740815502626,
      "grad_norm": 253.2539520263672,
      "learning_rate": 1.4484009812981921e-05,
      "loss": 4.6846,
      "step": 9201
    },
    {
      "epoch": 0.371497779572063,
      "grad_norm": 126.24796295166016,
      "learning_rate": 1.4482841016784038e-05,
      "loss": 3.5563,
      "step": 9202
    },
    {
      "epoch": 0.3715381509890997,
      "grad_norm": 79.03180694580078,
      "learning_rate": 1.4481672143940041e-05,
      "loss": 4.7067,
      "step": 9203
    },
    {
      "epoch": 0.3715785224061365,
      "grad_norm": 124.80098724365234,
      "learning_rate": 1.4480503194469918e-05,
      "loss": 2.9029,
      "step": 9204
    },
    {
      "epoch": 0.3716188938231732,
      "grad_norm": 110.24434661865234,
      "learning_rate": 1.4479334168393653e-05,
      "loss": 6.0419,
      "step": 9205
    },
    {
      "epoch": 0.3716592652402099,
      "grad_norm": 85.52017974853516,
      "learning_rate": 1.4478165065731231e-05,
      "loss": 4.3664,
      "step": 9206
    },
    {
      "epoch": 0.37169963665724665,
      "grad_norm": 152.03317260742188,
      "learning_rate": 1.4476995886502646e-05,
      "loss": 3.546,
      "step": 9207
    },
    {
      "epoch": 0.3717400080742834,
      "grad_norm": 1051.6424560546875,
      "learning_rate": 1.4475826630727887e-05,
      "loss": 5.5215,
      "step": 9208
    },
    {
      "epoch": 0.37178037949132015,
      "grad_norm": 111.91055297851562,
      "learning_rate": 1.4474657298426944e-05,
      "loss": 5.519,
      "step": 9209
    },
    {
      "epoch": 0.37182075090835687,
      "grad_norm": 97.43765258789062,
      "learning_rate": 1.4473487889619808e-05,
      "loss": 4.843,
      "step": 9210
    },
    {
      "epoch": 0.37186112232539364,
      "grad_norm": 221.6244659423828,
      "learning_rate": 1.4472318404326478e-05,
      "loss": 6.6181,
      "step": 9211
    },
    {
      "epoch": 0.37190149374243037,
      "grad_norm": 87.27433776855469,
      "learning_rate": 1.4471148842566945e-05,
      "loss": 4.1877,
      "step": 9212
    },
    {
      "epoch": 0.3719418651594671,
      "grad_norm": 74.4671401977539,
      "learning_rate": 1.4469979204361208e-05,
      "loss": 5.8222,
      "step": 9213
    },
    {
      "epoch": 0.3719822365765038,
      "grad_norm": 236.10743713378906,
      "learning_rate": 1.4468809489729263e-05,
      "loss": 4.9821,
      "step": 9214
    },
    {
      "epoch": 0.3720226079935406,
      "grad_norm": 84.3155746459961,
      "learning_rate": 1.4467639698691115e-05,
      "loss": 5.1934,
      "step": 9215
    },
    {
      "epoch": 0.3720629794105773,
      "grad_norm": 111.87686157226562,
      "learning_rate": 1.4466469831266758e-05,
      "loss": 3.8363,
      "step": 9216
    },
    {
      "epoch": 0.37210335082761403,
      "grad_norm": 201.0253143310547,
      "learning_rate": 1.4465299887476198e-05,
      "loss": 7.6136,
      "step": 9217
    },
    {
      "epoch": 0.3721437222446508,
      "grad_norm": 105.76039123535156,
      "learning_rate": 1.4464129867339438e-05,
      "loss": 6.9288,
      "step": 9218
    },
    {
      "epoch": 0.37218409366168753,
      "grad_norm": 171.9276885986328,
      "learning_rate": 1.446295977087648e-05,
      "loss": 4.9352,
      "step": 9219
    },
    {
      "epoch": 0.37222446507872425,
      "grad_norm": 99.98127746582031,
      "learning_rate": 1.4461789598107334e-05,
      "loss": 5.7224,
      "step": 9220
    },
    {
      "epoch": 0.372264836495761,
      "grad_norm": 89.11460876464844,
      "learning_rate": 1.4460619349052003e-05,
      "loss": 4.5748,
      "step": 9221
    },
    {
      "epoch": 0.37230520791279775,
      "grad_norm": 235.8642120361328,
      "learning_rate": 1.4459449023730499e-05,
      "loss": 5.5108,
      "step": 9222
    },
    {
      "epoch": 0.3723455793298345,
      "grad_norm": 80.70413970947266,
      "learning_rate": 1.4458278622162827e-05,
      "loss": 5.2834,
      "step": 9223
    },
    {
      "epoch": 0.3723859507468712,
      "grad_norm": 111.86280822753906,
      "learning_rate": 1.4457108144369003e-05,
      "loss": 5.4302,
      "step": 9224
    },
    {
      "epoch": 0.372426322163908,
      "grad_norm": 183.75863647460938,
      "learning_rate": 1.4455937590369038e-05,
      "loss": 5.5715,
      "step": 9225
    },
    {
      "epoch": 0.3724666935809447,
      "grad_norm": 88.69591522216797,
      "learning_rate": 1.4454766960182947e-05,
      "loss": 3.0353,
      "step": 9226
    },
    {
      "epoch": 0.3725070649979814,
      "grad_norm": 70.0589828491211,
      "learning_rate": 1.4453596253830742e-05,
      "loss": 4.8433,
      "step": 9227
    },
    {
      "epoch": 0.37254743641501814,
      "grad_norm": 154.3365020751953,
      "learning_rate": 1.4452425471332441e-05,
      "loss": 3.8215,
      "step": 9228
    },
    {
      "epoch": 0.3725878078320549,
      "grad_norm": 73.10316467285156,
      "learning_rate": 1.4451254612708063e-05,
      "loss": 5.6101,
      "step": 9229
    },
    {
      "epoch": 0.37262817924909164,
      "grad_norm": 118.93525695800781,
      "learning_rate": 1.4450083677977625e-05,
      "loss": 4.6902,
      "step": 9230
    },
    {
      "epoch": 0.37266855066612836,
      "grad_norm": 131.3292236328125,
      "learning_rate": 1.4448912667161148e-05,
      "loss": 5.6762,
      "step": 9231
    },
    {
      "epoch": 0.37270892208316514,
      "grad_norm": 127.81889343261719,
      "learning_rate": 1.4447741580278651e-05,
      "loss": 5.9442,
      "step": 9232
    },
    {
      "epoch": 0.37274929350020186,
      "grad_norm": 93.32315063476562,
      "learning_rate": 1.4446570417350161e-05,
      "loss": 4.0438,
      "step": 9233
    },
    {
      "epoch": 0.3727896649172386,
      "grad_norm": 74.42321014404297,
      "learning_rate": 1.44453991783957e-05,
      "loss": 4.5366,
      "step": 9234
    },
    {
      "epoch": 0.3728300363342753,
      "grad_norm": 100.7272720336914,
      "learning_rate": 1.4444227863435293e-05,
      "loss": 4.6502,
      "step": 9235
    },
    {
      "epoch": 0.3728704077513121,
      "grad_norm": 114.8003921508789,
      "learning_rate": 1.444305647248897e-05,
      "loss": 4.7559,
      "step": 9236
    },
    {
      "epoch": 0.3729107791683488,
      "grad_norm": 97.24234771728516,
      "learning_rate": 1.4441885005576755e-05,
      "loss": 6.0553,
      "step": 9237
    },
    {
      "epoch": 0.3729511505853855,
      "grad_norm": 116.264892578125,
      "learning_rate": 1.4440713462718679e-05,
      "loss": 7.1916,
      "step": 9238
    },
    {
      "epoch": 0.3729915220024223,
      "grad_norm": 168.14837646484375,
      "learning_rate": 1.4439541843934775e-05,
      "loss": 5.1241,
      "step": 9239
    },
    {
      "epoch": 0.373031893419459,
      "grad_norm": 87.07844543457031,
      "learning_rate": 1.443837014924507e-05,
      "loss": 3.451,
      "step": 9240
    },
    {
      "epoch": 0.37307226483649575,
      "grad_norm": 95.95893096923828,
      "learning_rate": 1.4437198378669598e-05,
      "loss": 5.3985,
      "step": 9241
    },
    {
      "epoch": 0.3731126362535325,
      "grad_norm": 121.41316223144531,
      "learning_rate": 1.4436026532228397e-05,
      "loss": 5.9169,
      "step": 9242
    },
    {
      "epoch": 0.37315300767056925,
      "grad_norm": 135.46322631835938,
      "learning_rate": 1.4434854609941502e-05,
      "loss": 3.9317,
      "step": 9243
    },
    {
      "epoch": 0.37319337908760597,
      "grad_norm": 89.8835678100586,
      "learning_rate": 1.4433682611828948e-05,
      "loss": 4.7982,
      "step": 9244
    },
    {
      "epoch": 0.3732337505046427,
      "grad_norm": 84.97223663330078,
      "learning_rate": 1.4432510537910777e-05,
      "loss": 6.8252,
      "step": 9245
    },
    {
      "epoch": 0.37327412192167947,
      "grad_norm": 103.09564971923828,
      "learning_rate": 1.4431338388207023e-05,
      "loss": 6.1624,
      "step": 9246
    },
    {
      "epoch": 0.3733144933387162,
      "grad_norm": 87.63470458984375,
      "learning_rate": 1.4430166162737735e-05,
      "loss": 5.6569,
      "step": 9247
    },
    {
      "epoch": 0.3733548647557529,
      "grad_norm": 102.5894546508789,
      "learning_rate": 1.4428993861522948e-05,
      "loss": 6.1629,
      "step": 9248
    },
    {
      "epoch": 0.3733952361727897,
      "grad_norm": 120.545654296875,
      "learning_rate": 1.442782148458271e-05,
      "loss": 4.4355,
      "step": 9249
    },
    {
      "epoch": 0.3734356075898264,
      "grad_norm": 124.77922821044922,
      "learning_rate": 1.4426649031937061e-05,
      "loss": 4.9583,
      "step": 9250
    },
    {
      "epoch": 0.37347597900686313,
      "grad_norm": 96.11697387695312,
      "learning_rate": 1.4425476503606055e-05,
      "loss": 5.1584,
      "step": 9251
    },
    {
      "epoch": 0.37351635042389986,
      "grad_norm": 109.39665222167969,
      "learning_rate": 1.442430389960973e-05,
      "loss": 5.6465,
      "step": 9252
    },
    {
      "epoch": 0.37355672184093663,
      "grad_norm": 105.4565200805664,
      "learning_rate": 1.4423131219968143e-05,
      "loss": 5.3917,
      "step": 9253
    },
    {
      "epoch": 0.37359709325797336,
      "grad_norm": 77.36973571777344,
      "learning_rate": 1.442195846470134e-05,
      "loss": 5.7912,
      "step": 9254
    },
    {
      "epoch": 0.3736374646750101,
      "grad_norm": 89.11502838134766,
      "learning_rate": 1.4420785633829376e-05,
      "loss": 3.499,
      "step": 9255
    },
    {
      "epoch": 0.37367783609204686,
      "grad_norm": 143.65396118164062,
      "learning_rate": 1.4419612727372296e-05,
      "loss": 5.6837,
      "step": 9256
    },
    {
      "epoch": 0.3737182075090836,
      "grad_norm": 100.44297790527344,
      "learning_rate": 1.4418439745350163e-05,
      "loss": 5.501,
      "step": 9257
    },
    {
      "epoch": 0.3737585789261203,
      "grad_norm": 89.0344009399414,
      "learning_rate": 1.4417266687783025e-05,
      "loss": 5.738,
      "step": 9258
    },
    {
      "epoch": 0.373798950343157,
      "grad_norm": 125.07625579833984,
      "learning_rate": 1.4416093554690943e-05,
      "loss": 5.5838,
      "step": 9259
    },
    {
      "epoch": 0.3738393217601938,
      "grad_norm": 74.13111114501953,
      "learning_rate": 1.4414920346093974e-05,
      "loss": 6.0833,
      "step": 9260
    },
    {
      "epoch": 0.3738796931772305,
      "grad_norm": 108.08912658691406,
      "learning_rate": 1.4413747062012177e-05,
      "loss": 5.0731,
      "step": 9261
    },
    {
      "epoch": 0.37392006459426724,
      "grad_norm": 85.92058563232422,
      "learning_rate": 1.4412573702465614e-05,
      "loss": 6.7846,
      "step": 9262
    },
    {
      "epoch": 0.373960436011304,
      "grad_norm": 77.9670181274414,
      "learning_rate": 1.441140026747434e-05,
      "loss": 6.2583,
      "step": 9263
    },
    {
      "epoch": 0.37400080742834074,
      "grad_norm": 128.70697021484375,
      "learning_rate": 1.4410226757058423e-05,
      "loss": 5.6167,
      "step": 9264
    },
    {
      "epoch": 0.37404117884537746,
      "grad_norm": 89.04658508300781,
      "learning_rate": 1.4409053171237933e-05,
      "loss": 6.7516,
      "step": 9265
    },
    {
      "epoch": 0.3740815502624142,
      "grad_norm": 83.71443939208984,
      "learning_rate": 1.4407879510032924e-05,
      "loss": 3.8885,
      "step": 9266
    },
    {
      "epoch": 0.37412192167945096,
      "grad_norm": 146.90524291992188,
      "learning_rate": 1.4406705773463471e-05,
      "loss": 5.1335,
      "step": 9267
    },
    {
      "epoch": 0.3741622930964877,
      "grad_norm": 77.71144104003906,
      "learning_rate": 1.440553196154964e-05,
      "loss": 4.084,
      "step": 9268
    },
    {
      "epoch": 0.3742026645135244,
      "grad_norm": 99.4179458618164,
      "learning_rate": 1.44043580743115e-05,
      "loss": 4.0965,
      "step": 9269
    },
    {
      "epoch": 0.3742430359305612,
      "grad_norm": 118.93583679199219,
      "learning_rate": 1.440318411176912e-05,
      "loss": 4.123,
      "step": 9270
    },
    {
      "epoch": 0.3742834073475979,
      "grad_norm": 84.55147552490234,
      "learning_rate": 1.4402010073942574e-05,
      "loss": 6.7351,
      "step": 9271
    },
    {
      "epoch": 0.37432377876463463,
      "grad_norm": 81.77420806884766,
      "learning_rate": 1.4400835960851938e-05,
      "loss": 5.3151,
      "step": 9272
    },
    {
      "epoch": 0.37436415018167135,
      "grad_norm": 131.72511291503906,
      "learning_rate": 1.4399661772517279e-05,
      "loss": 5.4779,
      "step": 9273
    },
    {
      "epoch": 0.37440452159870813,
      "grad_norm": 133.69058227539062,
      "learning_rate": 1.439848750895868e-05,
      "loss": 6.4332,
      "step": 9274
    },
    {
      "epoch": 0.37444489301574485,
      "grad_norm": 100.80475616455078,
      "learning_rate": 1.4397313170196218e-05,
      "loss": 5.1818,
      "step": 9275
    },
    {
      "epoch": 0.3744852644327816,
      "grad_norm": 89.42333984375,
      "learning_rate": 1.4396138756249967e-05,
      "loss": 4.5641,
      "step": 9276
    },
    {
      "epoch": 0.37452563584981835,
      "grad_norm": 88.70097351074219,
      "learning_rate": 1.439496426714001e-05,
      "loss": 4.3983,
      "step": 9277
    },
    {
      "epoch": 0.37456600726685507,
      "grad_norm": 120.41051483154297,
      "learning_rate": 1.4393789702886424e-05,
      "loss": 5.4079,
      "step": 9278
    },
    {
      "epoch": 0.3746063786838918,
      "grad_norm": 78.96178436279297,
      "learning_rate": 1.4392615063509298e-05,
      "loss": 5.5333,
      "step": 9279
    },
    {
      "epoch": 0.3746467501009285,
      "grad_norm": 82.73977661132812,
      "learning_rate": 1.4391440349028712e-05,
      "loss": 5.1308,
      "step": 9280
    },
    {
      "epoch": 0.3746871215179653,
      "grad_norm": 152.99021911621094,
      "learning_rate": 1.439026555946475e-05,
      "loss": 6.0916,
      "step": 9281
    },
    {
      "epoch": 0.374727492935002,
      "grad_norm": 146.3261260986328,
      "learning_rate": 1.4389090694837498e-05,
      "loss": 5.612,
      "step": 9282
    },
    {
      "epoch": 0.37476786435203874,
      "grad_norm": 90.41271209716797,
      "learning_rate": 1.4387915755167043e-05,
      "loss": 5.4708,
      "step": 9283
    },
    {
      "epoch": 0.3748082357690755,
      "grad_norm": 1819.8521728515625,
      "learning_rate": 1.4386740740473477e-05,
      "loss": 5.2804,
      "step": 9284
    },
    {
      "epoch": 0.37484860718611224,
      "grad_norm": 100.28334045410156,
      "learning_rate": 1.438556565077689e-05,
      "loss": 5.9134,
      "step": 9285
    },
    {
      "epoch": 0.37488897860314896,
      "grad_norm": 91.27239990234375,
      "learning_rate": 1.4384390486097368e-05,
      "loss": 5.4083,
      "step": 9286
    },
    {
      "epoch": 0.3749293500201857,
      "grad_norm": 100.78849792480469,
      "learning_rate": 1.438321524645501e-05,
      "loss": 4.8127,
      "step": 9287
    },
    {
      "epoch": 0.37496972143722246,
      "grad_norm": 102.73365783691406,
      "learning_rate": 1.4382039931869904e-05,
      "loss": 6.08,
      "step": 9288
    },
    {
      "epoch": 0.3750100928542592,
      "grad_norm": 160.29022216796875,
      "learning_rate": 1.438086454236215e-05,
      "loss": 5.4757,
      "step": 9289
    },
    {
      "epoch": 0.3750504642712959,
      "grad_norm": 136.853515625,
      "learning_rate": 1.4379689077951841e-05,
      "loss": 7.0443,
      "step": 9290
    },
    {
      "epoch": 0.3750908356883327,
      "grad_norm": 139.4066619873047,
      "learning_rate": 1.4378513538659075e-05,
      "loss": 5.2448,
      "step": 9291
    },
    {
      "epoch": 0.3751312071053694,
      "grad_norm": 86.77659606933594,
      "learning_rate": 1.4377337924503953e-05,
      "loss": 3.6055,
      "step": 9292
    },
    {
      "epoch": 0.3751715785224061,
      "grad_norm": 158.20835876464844,
      "learning_rate": 1.4376162235506577e-05,
      "loss": 5.2077,
      "step": 9293
    },
    {
      "epoch": 0.3752119499394429,
      "grad_norm": 89.39104461669922,
      "learning_rate": 1.4374986471687043e-05,
      "loss": 5.0269,
      "step": 9294
    },
    {
      "epoch": 0.3752523213564796,
      "grad_norm": 121.43612670898438,
      "learning_rate": 1.4373810633065459e-05,
      "loss": 4.4797,
      "step": 9295
    },
    {
      "epoch": 0.37529269277351635,
      "grad_norm": 113.41622161865234,
      "learning_rate": 1.4372634719661925e-05,
      "loss": 5.6587,
      "step": 9296
    },
    {
      "epoch": 0.37533306419055307,
      "grad_norm": 93.58663940429688,
      "learning_rate": 1.437145873149655e-05,
      "loss": 3.7877,
      "step": 9297
    },
    {
      "epoch": 0.37537343560758984,
      "grad_norm": 116.6131362915039,
      "learning_rate": 1.4370282668589436e-05,
      "loss": 5.6106,
      "step": 9298
    },
    {
      "epoch": 0.37541380702462657,
      "grad_norm": 113.31968688964844,
      "learning_rate": 1.4369106530960695e-05,
      "loss": 5.8531,
      "step": 9299
    },
    {
      "epoch": 0.3754541784416633,
      "grad_norm": 106.32715606689453,
      "learning_rate": 1.4367930318630434e-05,
      "loss": 4.9021,
      "step": 9300
    },
    {
      "epoch": 0.37549454985870007,
      "grad_norm": 89.15299987792969,
      "learning_rate": 1.4366754031618766e-05,
      "loss": 3.801,
      "step": 9301
    },
    {
      "epoch": 0.3755349212757368,
      "grad_norm": 91.37376403808594,
      "learning_rate": 1.4365577669945799e-05,
      "loss": 3.0988,
      "step": 9302
    },
    {
      "epoch": 0.3755752926927735,
      "grad_norm": 242.55833435058594,
      "learning_rate": 1.436440123363165e-05,
      "loss": 6.1023,
      "step": 9303
    },
    {
      "epoch": 0.37561566410981023,
      "grad_norm": 119.41439056396484,
      "learning_rate": 1.4363224722696431e-05,
      "loss": 5.7265,
      "step": 9304
    },
    {
      "epoch": 0.375656035526847,
      "grad_norm": 124.00894165039062,
      "learning_rate": 1.4362048137160259e-05,
      "loss": 2.6695,
      "step": 9305
    },
    {
      "epoch": 0.37569640694388373,
      "grad_norm": 133.3777618408203,
      "learning_rate": 1.4360871477043248e-05,
      "loss": 4.5313,
      "step": 9306
    },
    {
      "epoch": 0.37573677836092045,
      "grad_norm": 104.20848083496094,
      "learning_rate": 1.4359694742365519e-05,
      "loss": 5.1938,
      "step": 9307
    },
    {
      "epoch": 0.37577714977795723,
      "grad_norm": 92.19112396240234,
      "learning_rate": 1.4358517933147192e-05,
      "loss": 5.0031,
      "step": 9308
    },
    {
      "epoch": 0.37581752119499395,
      "grad_norm": 270.5786437988281,
      "learning_rate": 1.4357341049408387e-05,
      "loss": 4.2373,
      "step": 9309
    },
    {
      "epoch": 0.3758578926120307,
      "grad_norm": 93.36541748046875,
      "learning_rate": 1.4356164091169222e-05,
      "loss": 5.035,
      "step": 9310
    },
    {
      "epoch": 0.3758982640290674,
      "grad_norm": 75.98831176757812,
      "learning_rate": 1.4354987058449824e-05,
      "loss": 4.4126,
      "step": 9311
    },
    {
      "epoch": 0.3759386354461042,
      "grad_norm": 141.84320068359375,
      "learning_rate": 1.4353809951270317e-05,
      "loss": 7.0013,
      "step": 9312
    },
    {
      "epoch": 0.3759790068631409,
      "grad_norm": 94.77869415283203,
      "learning_rate": 1.4352632769650827e-05,
      "loss": 5.5924,
      "step": 9313
    },
    {
      "epoch": 0.3760193782801776,
      "grad_norm": 98.2086410522461,
      "learning_rate": 1.435145551361148e-05,
      "loss": 4.2056,
      "step": 9314
    },
    {
      "epoch": 0.3760597496972144,
      "grad_norm": 100.00944519042969,
      "learning_rate": 1.4350278183172408e-05,
      "loss": 7.4133,
      "step": 9315
    },
    {
      "epoch": 0.3761001211142511,
      "grad_norm": 100.26580047607422,
      "learning_rate": 1.4349100778353735e-05,
      "loss": 4.0761,
      "step": 9316
    },
    {
      "epoch": 0.37614049253128784,
      "grad_norm": 95.24305725097656,
      "learning_rate": 1.4347923299175596e-05,
      "loss": 4.419,
      "step": 9317
    },
    {
      "epoch": 0.37618086394832456,
      "grad_norm": 121.79546356201172,
      "learning_rate": 1.4346745745658122e-05,
      "loss": 5.6995,
      "step": 9318
    },
    {
      "epoch": 0.37622123536536134,
      "grad_norm": 113.53590393066406,
      "learning_rate": 1.4345568117821444e-05,
      "loss": 5.2308,
      "step": 9319
    },
    {
      "epoch": 0.37626160678239806,
      "grad_norm": 97.06631469726562,
      "learning_rate": 1.43443904156857e-05,
      "loss": 5.8666,
      "step": 9320
    },
    {
      "epoch": 0.3763019781994348,
      "grad_norm": 152.1392822265625,
      "learning_rate": 1.4343212639271025e-05,
      "loss": 5.1799,
      "step": 9321
    },
    {
      "epoch": 0.37634234961647156,
      "grad_norm": 86.51665496826172,
      "learning_rate": 1.4342034788597556e-05,
      "loss": 5.8854,
      "step": 9322
    },
    {
      "epoch": 0.3763827210335083,
      "grad_norm": 136.6768798828125,
      "learning_rate": 1.434085686368543e-05,
      "loss": 4.6063,
      "step": 9323
    },
    {
      "epoch": 0.376423092450545,
      "grad_norm": 81.18684387207031,
      "learning_rate": 1.4339678864554788e-05,
      "loss": 5.8606,
      "step": 9324
    },
    {
      "epoch": 0.3764634638675817,
      "grad_norm": 101.74859619140625,
      "learning_rate": 1.4338500791225775e-05,
      "loss": 4.9123,
      "step": 9325
    },
    {
      "epoch": 0.3765038352846185,
      "grad_norm": 88.60202026367188,
      "learning_rate": 1.4337322643718527e-05,
      "loss": 5.188,
      "step": 9326
    },
    {
      "epoch": 0.3765442067016552,
      "grad_norm": 76.62479400634766,
      "learning_rate": 1.4336144422053194e-05,
      "loss": 5.2787,
      "step": 9327
    },
    {
      "epoch": 0.37658457811869195,
      "grad_norm": 67.1329116821289,
      "learning_rate": 1.4334966126249911e-05,
      "loss": 4.7404,
      "step": 9328
    },
    {
      "epoch": 0.3766249495357287,
      "grad_norm": 118.62560272216797,
      "learning_rate": 1.4333787756328833e-05,
      "loss": 4.3225,
      "step": 9329
    },
    {
      "epoch": 0.37666532095276545,
      "grad_norm": 97.42314147949219,
      "learning_rate": 1.4332609312310104e-05,
      "loss": 6.0998,
      "step": 9330
    },
    {
      "epoch": 0.37670569236980217,
      "grad_norm": 108.40314483642578,
      "learning_rate": 1.4331430794213874e-05,
      "loss": 5.8146,
      "step": 9331
    },
    {
      "epoch": 0.3767460637868389,
      "grad_norm": 81.2721939086914,
      "learning_rate": 1.433025220206029e-05,
      "loss": 5.2428,
      "step": 9332
    },
    {
      "epoch": 0.37678643520387567,
      "grad_norm": 216.214111328125,
      "learning_rate": 1.4329073535869508e-05,
      "loss": 5.1894,
      "step": 9333
    },
    {
      "epoch": 0.3768268066209124,
      "grad_norm": 113.2392807006836,
      "learning_rate": 1.4327894795661677e-05,
      "loss": 5.0073,
      "step": 9334
    },
    {
      "epoch": 0.3768671780379491,
      "grad_norm": 85.44409942626953,
      "learning_rate": 1.4326715981456951e-05,
      "loss": 5.4871,
      "step": 9335
    },
    {
      "epoch": 0.3769075494549859,
      "grad_norm": 87.00192260742188,
      "learning_rate": 1.4325537093275483e-05,
      "loss": 4.9134,
      "step": 9336
    },
    {
      "epoch": 0.3769479208720226,
      "grad_norm": 77.70957946777344,
      "learning_rate": 1.4324358131137434e-05,
      "loss": 6.6106,
      "step": 9337
    },
    {
      "epoch": 0.37698829228905933,
      "grad_norm": 86.20156860351562,
      "learning_rate": 1.432317909506296e-05,
      "loss": 5.1315,
      "step": 9338
    },
    {
      "epoch": 0.37702866370609606,
      "grad_norm": 102.28266143798828,
      "learning_rate": 1.4321999985072218e-05,
      "loss": 7.3497,
      "step": 9339
    },
    {
      "epoch": 0.37706903512313283,
      "grad_norm": 72.86463165283203,
      "learning_rate": 1.4320820801185366e-05,
      "loss": 3.4025,
      "step": 9340
    },
    {
      "epoch": 0.37710940654016956,
      "grad_norm": 104.61544799804688,
      "learning_rate": 1.4319641543422571e-05,
      "loss": 4.7373,
      "step": 9341
    },
    {
      "epoch": 0.3771497779572063,
      "grad_norm": 74.77881622314453,
      "learning_rate": 1.4318462211803991e-05,
      "loss": 7.2485,
      "step": 9342
    },
    {
      "epoch": 0.37719014937424306,
      "grad_norm": 86.70611572265625,
      "learning_rate": 1.4317282806349796e-05,
      "loss": 6.8995,
      "step": 9343
    },
    {
      "epoch": 0.3772305207912798,
      "grad_norm": 92.03070831298828,
      "learning_rate": 1.4316103327080141e-05,
      "loss": 6.3991,
      "step": 9344
    },
    {
      "epoch": 0.3772708922083165,
      "grad_norm": 91.15127563476562,
      "learning_rate": 1.4314923774015201e-05,
      "loss": 4.7054,
      "step": 9345
    },
    {
      "epoch": 0.3773112636253533,
      "grad_norm": 970.6072387695312,
      "learning_rate": 1.431374414717514e-05,
      "loss": 6.9784,
      "step": 9346
    },
    {
      "epoch": 0.37735163504239,
      "grad_norm": 134.30503845214844,
      "learning_rate": 1.4312564446580126e-05,
      "loss": 4.7215,
      "step": 9347
    },
    {
      "epoch": 0.3773920064594267,
      "grad_norm": 90.8364486694336,
      "learning_rate": 1.4311384672250332e-05,
      "loss": 3.8874,
      "step": 9348
    },
    {
      "epoch": 0.37743237787646344,
      "grad_norm": 89.21388244628906,
      "learning_rate": 1.431020482420593e-05,
      "loss": 7.6459,
      "step": 9349
    },
    {
      "epoch": 0.3774727492935002,
      "grad_norm": 123.37228393554688,
      "learning_rate": 1.4309024902467085e-05,
      "loss": 3.8726,
      "step": 9350
    },
    {
      "epoch": 0.37751312071053694,
      "grad_norm": 109.51721954345703,
      "learning_rate": 1.430784490705398e-05,
      "loss": 6.4157,
      "step": 9351
    },
    {
      "epoch": 0.37755349212757366,
      "grad_norm": 91.58989715576172,
      "learning_rate": 1.4306664837986786e-05,
      "loss": 4.8103,
      "step": 9352
    },
    {
      "epoch": 0.37759386354461044,
      "grad_norm": 107.79061126708984,
      "learning_rate": 1.430548469528568e-05,
      "loss": 4.3961,
      "step": 9353
    },
    {
      "epoch": 0.37763423496164716,
      "grad_norm": 73.11061096191406,
      "learning_rate": 1.4304304478970839e-05,
      "loss": 5.5896,
      "step": 9354
    },
    {
      "epoch": 0.3776746063786839,
      "grad_norm": 100.77582550048828,
      "learning_rate": 1.4303124189062446e-05,
      "loss": 5.2199,
      "step": 9355
    },
    {
      "epoch": 0.3777149777957206,
      "grad_norm": 98.1591567993164,
      "learning_rate": 1.4301943825580673e-05,
      "loss": 4.0312,
      "step": 9356
    },
    {
      "epoch": 0.3777553492127574,
      "grad_norm": 83.65184020996094,
      "learning_rate": 1.4300763388545711e-05,
      "loss": 4.9567,
      "step": 9357
    },
    {
      "epoch": 0.3777957206297941,
      "grad_norm": 126.52433776855469,
      "learning_rate": 1.4299582877977736e-05,
      "loss": 5.8823,
      "step": 9358
    },
    {
      "epoch": 0.37783609204683083,
      "grad_norm": 150.1932373046875,
      "learning_rate": 1.4298402293896935e-05,
      "loss": 5.3789,
      "step": 9359
    },
    {
      "epoch": 0.3778764634638676,
      "grad_norm": 96.7654037475586,
      "learning_rate": 1.4297221636323491e-05,
      "loss": 5.6586,
      "step": 9360
    },
    {
      "epoch": 0.37791683488090433,
      "grad_norm": 118.69239044189453,
      "learning_rate": 1.429604090527759e-05,
      "loss": 6.4232,
      "step": 9361
    },
    {
      "epoch": 0.37795720629794105,
      "grad_norm": 72.68580627441406,
      "learning_rate": 1.4294860100779424e-05,
      "loss": 3.4069,
      "step": 9362
    },
    {
      "epoch": 0.3779975777149778,
      "grad_norm": 147.0451202392578,
      "learning_rate": 1.4293679222849179e-05,
      "loss": 4.9685,
      "step": 9363
    },
    {
      "epoch": 0.37803794913201455,
      "grad_norm": 89.95807647705078,
      "learning_rate": 1.4292498271507045e-05,
      "loss": 5.9848,
      "step": 9364
    },
    {
      "epoch": 0.37807832054905127,
      "grad_norm": 95.92564392089844,
      "learning_rate": 1.4291317246773218e-05,
      "loss": 5.6101,
      "step": 9365
    },
    {
      "epoch": 0.378118691966088,
      "grad_norm": 99.16349029541016,
      "learning_rate": 1.4290136148667881e-05,
      "loss": 4.8499,
      "step": 9366
    },
    {
      "epoch": 0.37815906338312477,
      "grad_norm": 62.73418426513672,
      "learning_rate": 1.4288954977211238e-05,
      "loss": 3.7321,
      "step": 9367
    },
    {
      "epoch": 0.3781994348001615,
      "grad_norm": 80.4330825805664,
      "learning_rate": 1.428777373242348e-05,
      "loss": 5.8723,
      "step": 9368
    },
    {
      "epoch": 0.3782398062171982,
      "grad_norm": 80.89337158203125,
      "learning_rate": 1.4286592414324805e-05,
      "loss": 4.172,
      "step": 9369
    },
    {
      "epoch": 0.37828017763423494,
      "grad_norm": 106.14750671386719,
      "learning_rate": 1.4285411022935405e-05,
      "loss": 5.0096,
      "step": 9370
    },
    {
      "epoch": 0.3783205490512717,
      "grad_norm": 268.8523864746094,
      "learning_rate": 1.4284229558275486e-05,
      "loss": 5.0304,
      "step": 9371
    },
    {
      "epoch": 0.37836092046830844,
      "grad_norm": 82.81707000732422,
      "learning_rate": 1.4283048020365247e-05,
      "loss": 4.1757,
      "step": 9372
    },
    {
      "epoch": 0.37840129188534516,
      "grad_norm": 86.57675170898438,
      "learning_rate": 1.4281866409224886e-05,
      "loss": 7.6475,
      "step": 9373
    },
    {
      "epoch": 0.37844166330238194,
      "grad_norm": 78.84817504882812,
      "learning_rate": 1.4280684724874611e-05,
      "loss": 2.76,
      "step": 9374
    },
    {
      "epoch": 0.37848203471941866,
      "grad_norm": 77.48424530029297,
      "learning_rate": 1.4279502967334624e-05,
      "loss": 5.7101,
      "step": 9375
    },
    {
      "epoch": 0.3785224061364554,
      "grad_norm": 109.14185333251953,
      "learning_rate": 1.4278321136625128e-05,
      "loss": 4.136,
      "step": 9376
    },
    {
      "epoch": 0.3785627775534921,
      "grad_norm": 111.5101547241211,
      "learning_rate": 1.4277139232766332e-05,
      "loss": 6.7341,
      "step": 9377
    },
    {
      "epoch": 0.3786031489705289,
      "grad_norm": 104.54025268554688,
      "learning_rate": 1.4275957255778443e-05,
      "loss": 7.1394,
      "step": 9378
    },
    {
      "epoch": 0.3786435203875656,
      "grad_norm": 76.35177612304688,
      "learning_rate": 1.427477520568167e-05,
      "loss": 4.9209,
      "step": 9379
    },
    {
      "epoch": 0.3786838918046023,
      "grad_norm": 136.67030334472656,
      "learning_rate": 1.4273593082496219e-05,
      "loss": 6.6346,
      "step": 9380
    },
    {
      "epoch": 0.3787242632216391,
      "grad_norm": 75.6785888671875,
      "learning_rate": 1.4272410886242312e-05,
      "loss": 4.4549,
      "step": 9381
    },
    {
      "epoch": 0.3787646346386758,
      "grad_norm": 85.04060363769531,
      "learning_rate": 1.4271228616940152e-05,
      "loss": 5.3734,
      "step": 9382
    },
    {
      "epoch": 0.37880500605571255,
      "grad_norm": 88.42935180664062,
      "learning_rate": 1.4270046274609958e-05,
      "loss": 6.1886,
      "step": 9383
    },
    {
      "epoch": 0.37884537747274927,
      "grad_norm": 93.72662353515625,
      "learning_rate": 1.4268863859271943e-05,
      "loss": 3.6484,
      "step": 9384
    },
    {
      "epoch": 0.37888574888978604,
      "grad_norm": 94.44232177734375,
      "learning_rate": 1.4267681370946326e-05,
      "loss": 4.5488,
      "step": 9385
    },
    {
      "epoch": 0.37892612030682277,
      "grad_norm": 176.69479370117188,
      "learning_rate": 1.4266498809653322e-05,
      "loss": 4.6967,
      "step": 9386
    },
    {
      "epoch": 0.3789664917238595,
      "grad_norm": 86.08747100830078,
      "learning_rate": 1.4265316175413153e-05,
      "loss": 5.2176,
      "step": 9387
    },
    {
      "epoch": 0.37900686314089627,
      "grad_norm": 64.13945770263672,
      "learning_rate": 1.4264133468246035e-05,
      "loss": 5.2566,
      "step": 9388
    },
    {
      "epoch": 0.379047234557933,
      "grad_norm": 82.55193328857422,
      "learning_rate": 1.4262950688172195e-05,
      "loss": 5.3615,
      "step": 9389
    },
    {
      "epoch": 0.3790876059749697,
      "grad_norm": 107.88215637207031,
      "learning_rate": 1.4261767835211853e-05,
      "loss": 4.4379,
      "step": 9390
    },
    {
      "epoch": 0.37912797739200643,
      "grad_norm": 80.27366638183594,
      "learning_rate": 1.4260584909385231e-05,
      "loss": 5.5693,
      "step": 9391
    },
    {
      "epoch": 0.3791683488090432,
      "grad_norm": 106.71404266357422,
      "learning_rate": 1.4259401910712556e-05,
      "loss": 6.6473,
      "step": 9392
    },
    {
      "epoch": 0.37920872022607993,
      "grad_norm": 76.74735260009766,
      "learning_rate": 1.4258218839214055e-05,
      "loss": 4.8173,
      "step": 9393
    },
    {
      "epoch": 0.37924909164311665,
      "grad_norm": 120.08229064941406,
      "learning_rate": 1.4257035694909955e-05,
      "loss": 4.9886,
      "step": 9394
    },
    {
      "epoch": 0.37928946306015343,
      "grad_norm": 77.23151397705078,
      "learning_rate": 1.4255852477820487e-05,
      "loss": 5.2634,
      "step": 9395
    },
    {
      "epoch": 0.37932983447719015,
      "grad_norm": 145.2874755859375,
      "learning_rate": 1.4254669187965879e-05,
      "loss": 4.8096,
      "step": 9396
    },
    {
      "epoch": 0.3793702058942269,
      "grad_norm": 66.68575286865234,
      "learning_rate": 1.4253485825366365e-05,
      "loss": 2.6926,
      "step": 9397
    },
    {
      "epoch": 0.37941057731126365,
      "grad_norm": 181.44747924804688,
      "learning_rate": 1.4252302390042175e-05,
      "loss": 5.8457,
      "step": 9398
    },
    {
      "epoch": 0.3794509487283004,
      "grad_norm": 86.74496459960938,
      "learning_rate": 1.4251118882013542e-05,
      "loss": 5.3247,
      "step": 9399
    },
    {
      "epoch": 0.3794913201453371,
      "grad_norm": 143.67169189453125,
      "learning_rate": 1.4249935301300703e-05,
      "loss": 3.9393,
      "step": 9400
    },
    {
      "epoch": 0.3795316915623738,
      "grad_norm": 128.2812042236328,
      "learning_rate": 1.4248751647923898e-05,
      "loss": 4.8655,
      "step": 9401
    },
    {
      "epoch": 0.3795720629794106,
      "grad_norm": 99.04549407958984,
      "learning_rate": 1.424756792190336e-05,
      "loss": 4.8542,
      "step": 9402
    },
    {
      "epoch": 0.3796124343964473,
      "grad_norm": 124.774169921875,
      "learning_rate": 1.4246384123259329e-05,
      "loss": 4.1802,
      "step": 9403
    },
    {
      "epoch": 0.37965280581348404,
      "grad_norm": 81.46092224121094,
      "learning_rate": 1.4245200252012047e-05,
      "loss": 5.9751,
      "step": 9404
    },
    {
      "epoch": 0.3796931772305208,
      "grad_norm": 104.19373321533203,
      "learning_rate": 1.4244016308181752e-05,
      "loss": 4.6712,
      "step": 9405
    },
    {
      "epoch": 0.37973354864755754,
      "grad_norm": 169.20956420898438,
      "learning_rate": 1.4242832291788689e-05,
      "loss": 6.4224,
      "step": 9406
    },
    {
      "epoch": 0.37977392006459426,
      "grad_norm": 66.30662536621094,
      "learning_rate": 1.4241648202853102e-05,
      "loss": 5.1585,
      "step": 9407
    },
    {
      "epoch": 0.379814291481631,
      "grad_norm": 107.47454833984375,
      "learning_rate": 1.4240464041395237e-05,
      "loss": 6.47,
      "step": 9408
    },
    {
      "epoch": 0.37985466289866776,
      "grad_norm": 176.6405487060547,
      "learning_rate": 1.4239279807435336e-05,
      "loss": 5.1155,
      "step": 9409
    },
    {
      "epoch": 0.3798950343157045,
      "grad_norm": 103.98466491699219,
      "learning_rate": 1.4238095500993652e-05,
      "loss": 4.8039,
      "step": 9410
    },
    {
      "epoch": 0.3799354057327412,
      "grad_norm": 90.20958709716797,
      "learning_rate": 1.4236911122090429e-05,
      "loss": 4.7688,
      "step": 9411
    },
    {
      "epoch": 0.379975777149778,
      "grad_norm": 75.12642669677734,
      "learning_rate": 1.4235726670745922e-05,
      "loss": 4.15,
      "step": 9412
    },
    {
      "epoch": 0.3800161485668147,
      "grad_norm": 197.45925903320312,
      "learning_rate": 1.423454214698038e-05,
      "loss": 5.4768,
      "step": 9413
    },
    {
      "epoch": 0.3800565199838514,
      "grad_norm": 111.57864379882812,
      "learning_rate": 1.4233357550814056e-05,
      "loss": 5.8038,
      "step": 9414
    },
    {
      "epoch": 0.38009689140088815,
      "grad_norm": 123.29605102539062,
      "learning_rate": 1.4232172882267201e-05,
      "loss": 6.5752,
      "step": 9415
    },
    {
      "epoch": 0.3801372628179249,
      "grad_norm": 116.04940795898438,
      "learning_rate": 1.4230988141360075e-05,
      "loss": 5.0974,
      "step": 9416
    },
    {
      "epoch": 0.38017763423496165,
      "grad_norm": 192.7432861328125,
      "learning_rate": 1.4229803328112932e-05,
      "loss": 6.7416,
      "step": 9417
    },
    {
      "epoch": 0.38021800565199837,
      "grad_norm": 108.96617126464844,
      "learning_rate": 1.422861844254603e-05,
      "loss": 3.7813,
      "step": 9418
    },
    {
      "epoch": 0.38025837706903515,
      "grad_norm": 109.91629791259766,
      "learning_rate": 1.4227433484679624e-05,
      "loss": 5.0365,
      "step": 9419
    },
    {
      "epoch": 0.38029874848607187,
      "grad_norm": 98.21446228027344,
      "learning_rate": 1.4226248454533977e-05,
      "loss": 5.0313,
      "step": 9420
    },
    {
      "epoch": 0.3803391199031086,
      "grad_norm": 117.58184814453125,
      "learning_rate": 1.422506335212935e-05,
      "loss": 4.8063,
      "step": 9421
    },
    {
      "epoch": 0.3803794913201453,
      "grad_norm": 79.3093032836914,
      "learning_rate": 1.4223878177486012e-05,
      "loss": 3.8795,
      "step": 9422
    },
    {
      "epoch": 0.3804198627371821,
      "grad_norm": 174.87722778320312,
      "learning_rate": 1.4222692930624216e-05,
      "loss": 7.063,
      "step": 9423
    },
    {
      "epoch": 0.3804602341542188,
      "grad_norm": 93.76956176757812,
      "learning_rate": 1.4221507611564234e-05,
      "loss": 4.9478,
      "step": 9424
    },
    {
      "epoch": 0.38050060557125553,
      "grad_norm": 183.93075561523438,
      "learning_rate": 1.4220322220326326e-05,
      "loss": 6.27,
      "step": 9425
    },
    {
      "epoch": 0.3805409769882923,
      "grad_norm": 75.67728424072266,
      "learning_rate": 1.4219136756930767e-05,
      "loss": 4.8575,
      "step": 9426
    },
    {
      "epoch": 0.38058134840532903,
      "grad_norm": 122.07404327392578,
      "learning_rate": 1.4217951221397822e-05,
      "loss": 5.5275,
      "step": 9427
    },
    {
      "epoch": 0.38062171982236576,
      "grad_norm": 83.89984893798828,
      "learning_rate": 1.4216765613747758e-05,
      "loss": 4.7112,
      "step": 9428
    },
    {
      "epoch": 0.3806620912394025,
      "grad_norm": 121.16854095458984,
      "learning_rate": 1.421557993400085e-05,
      "loss": 4.2934,
      "step": 9429
    },
    {
      "epoch": 0.38070246265643926,
      "grad_norm": 102.539794921875,
      "learning_rate": 1.4214394182177369e-05,
      "loss": 6.0498,
      "step": 9430
    },
    {
      "epoch": 0.380742834073476,
      "grad_norm": 95.30396270751953,
      "learning_rate": 1.4213208358297586e-05,
      "loss": 6.5586,
      "step": 9431
    },
    {
      "epoch": 0.3807832054905127,
      "grad_norm": 155.02273559570312,
      "learning_rate": 1.4212022462381785e-05,
      "loss": 5.4663,
      "step": 9432
    },
    {
      "epoch": 0.3808235769075495,
      "grad_norm": 92.26396179199219,
      "learning_rate": 1.4210836494450231e-05,
      "loss": 4.104,
      "step": 9433
    },
    {
      "epoch": 0.3808639483245862,
      "grad_norm": 197.03456115722656,
      "learning_rate": 1.4209650454523208e-05,
      "loss": 5.3289,
      "step": 9434
    },
    {
      "epoch": 0.3809043197416229,
      "grad_norm": 85.77422332763672,
      "learning_rate": 1.420846434262099e-05,
      "loss": 6.3682,
      "step": 9435
    },
    {
      "epoch": 0.38094469115865964,
      "grad_norm": 77.17662811279297,
      "learning_rate": 1.4207278158763861e-05,
      "loss": 5.5985,
      "step": 9436
    },
    {
      "epoch": 0.3809850625756964,
      "grad_norm": 90.62281036376953,
      "learning_rate": 1.4206091902972099e-05,
      "loss": 4.2561,
      "step": 9437
    },
    {
      "epoch": 0.38102543399273314,
      "grad_norm": 73.079833984375,
      "learning_rate": 1.4204905575265991e-05,
      "loss": 5.0101,
      "step": 9438
    },
    {
      "epoch": 0.38106580540976986,
      "grad_norm": 105.226806640625,
      "learning_rate": 1.4203719175665813e-05,
      "loss": 6.2196,
      "step": 9439
    },
    {
      "epoch": 0.38110617682680664,
      "grad_norm": 99.30427551269531,
      "learning_rate": 1.4202532704191853e-05,
      "loss": 6.602,
      "step": 9440
    },
    {
      "epoch": 0.38114654824384336,
      "grad_norm": 114.63990783691406,
      "learning_rate": 1.42013461608644e-05,
      "loss": 5.3869,
      "step": 9441
    },
    {
      "epoch": 0.3811869196608801,
      "grad_norm": 106.48197174072266,
      "learning_rate": 1.4200159545703737e-05,
      "loss": 6.2186,
      "step": 9442
    },
    {
      "epoch": 0.3812272910779168,
      "grad_norm": 172.19064331054688,
      "learning_rate": 1.4198972858730153e-05,
      "loss": 4.1877,
      "step": 9443
    },
    {
      "epoch": 0.3812676624949536,
      "grad_norm": 94.66915893554688,
      "learning_rate": 1.4197786099963943e-05,
      "loss": 5.8956,
      "step": 9444
    },
    {
      "epoch": 0.3813080339119903,
      "grad_norm": 127.61802673339844,
      "learning_rate": 1.4196599269425388e-05,
      "loss": 6.1924,
      "step": 9445
    },
    {
      "epoch": 0.38134840532902703,
      "grad_norm": 72.32410430908203,
      "learning_rate": 1.4195412367134788e-05,
      "loss": 6.782,
      "step": 9446
    },
    {
      "epoch": 0.3813887767460638,
      "grad_norm": 74.2740707397461,
      "learning_rate": 1.4194225393112433e-05,
      "loss": 6.399,
      "step": 9447
    },
    {
      "epoch": 0.38142914816310053,
      "grad_norm": 90.78068542480469,
      "learning_rate": 1.4193038347378618e-05,
      "loss": 5.6821,
      "step": 9448
    },
    {
      "epoch": 0.38146951958013725,
      "grad_norm": 226.4221954345703,
      "learning_rate": 1.419185122995364e-05,
      "loss": 6.7367,
      "step": 9449
    },
    {
      "epoch": 0.38150989099717403,
      "grad_norm": 100.9735107421875,
      "learning_rate": 1.419066404085779e-05,
      "loss": 4.9345,
      "step": 9450
    },
    {
      "epoch": 0.38155026241421075,
      "grad_norm": 78.0103759765625,
      "learning_rate": 1.4189476780111376e-05,
      "loss": 4.5036,
      "step": 9451
    },
    {
      "epoch": 0.38159063383124747,
      "grad_norm": 83.3680191040039,
      "learning_rate": 1.4188289447734688e-05,
      "loss": 5.2279,
      "step": 9452
    },
    {
      "epoch": 0.3816310052482842,
      "grad_norm": 111.2537841796875,
      "learning_rate": 1.4187102043748033e-05,
      "loss": 6.0225,
      "step": 9453
    },
    {
      "epoch": 0.38167137666532097,
      "grad_norm": 62.198631286621094,
      "learning_rate": 1.4185914568171713e-05,
      "loss": 6.1571,
      "step": 9454
    },
    {
      "epoch": 0.3817117480823577,
      "grad_norm": 84.27140808105469,
      "learning_rate": 1.4184727021026026e-05,
      "loss": 5.4743,
      "step": 9455
    },
    {
      "epoch": 0.3817521194993944,
      "grad_norm": 109.04243469238281,
      "learning_rate": 1.418353940233128e-05,
      "loss": 5.7289,
      "step": 9456
    },
    {
      "epoch": 0.3817924909164312,
      "grad_norm": 141.415771484375,
      "learning_rate": 1.4182351712107777e-05,
      "loss": 6.3644,
      "step": 9457
    },
    {
      "epoch": 0.3818328623334679,
      "grad_norm": 162.60369873046875,
      "learning_rate": 1.4181163950375832e-05,
      "loss": 4.0656,
      "step": 9458
    },
    {
      "epoch": 0.38187323375050464,
      "grad_norm": 122.99447631835938,
      "learning_rate": 1.417997611715574e-05,
      "loss": 5.9401,
      "step": 9459
    },
    {
      "epoch": 0.38191360516754136,
      "grad_norm": 94.27217102050781,
      "learning_rate": 1.417878821246782e-05,
      "loss": 5.3458,
      "step": 9460
    },
    {
      "epoch": 0.38195397658457814,
      "grad_norm": 94.06454467773438,
      "learning_rate": 1.417760023633238e-05,
      "loss": 5.6733,
      "step": 9461
    },
    {
      "epoch": 0.38199434800161486,
      "grad_norm": 124.7751235961914,
      "learning_rate": 1.4176412188769732e-05,
      "loss": 4.8636,
      "step": 9462
    },
    {
      "epoch": 0.3820347194186516,
      "grad_norm": 144.74253845214844,
      "learning_rate": 1.4175224069800187e-05,
      "loss": 5.4476,
      "step": 9463
    },
    {
      "epoch": 0.38207509083568836,
      "grad_norm": 174.6633758544922,
      "learning_rate": 1.4174035879444064e-05,
      "loss": 5.492,
      "step": 9464
    },
    {
      "epoch": 0.3821154622527251,
      "grad_norm": 94.79393768310547,
      "learning_rate": 1.417284761772167e-05,
      "loss": 5.6407,
      "step": 9465
    },
    {
      "epoch": 0.3821558336697618,
      "grad_norm": 88.9095687866211,
      "learning_rate": 1.417165928465333e-05,
      "loss": 6.5836,
      "step": 9466
    },
    {
      "epoch": 0.3821962050867985,
      "grad_norm": 80.90718841552734,
      "learning_rate": 1.4170470880259353e-05,
      "loss": 4.2185,
      "step": 9467
    },
    {
      "epoch": 0.3822365765038353,
      "grad_norm": 60.695682525634766,
      "learning_rate": 1.4169282404560067e-05,
      "loss": 4.339,
      "step": 9468
    },
    {
      "epoch": 0.382276947920872,
      "grad_norm": 113.45650482177734,
      "learning_rate": 1.4168093857575785e-05,
      "loss": 5.8965,
      "step": 9469
    },
    {
      "epoch": 0.38231731933790875,
      "grad_norm": 62.56510925292969,
      "learning_rate": 1.4166905239326832e-05,
      "loss": 5.8599,
      "step": 9470
    },
    {
      "epoch": 0.3823576907549455,
      "grad_norm": 69.07328796386719,
      "learning_rate": 1.4165716549833527e-05,
      "loss": 6.1692,
      "step": 9471
    },
    {
      "epoch": 0.38239806217198224,
      "grad_norm": 98.51516723632812,
      "learning_rate": 1.4164527789116202e-05,
      "loss": 6.2522,
      "step": 9472
    },
    {
      "epoch": 0.38243843358901897,
      "grad_norm": 83.21365356445312,
      "learning_rate": 1.4163338957195173e-05,
      "loss": 5.7102,
      "step": 9473
    },
    {
      "epoch": 0.3824788050060557,
      "grad_norm": 124.6690444946289,
      "learning_rate": 1.4162150054090773e-05,
      "loss": 7.0754,
      "step": 9474
    },
    {
      "epoch": 0.38251917642309247,
      "grad_norm": 79.01008605957031,
      "learning_rate": 1.4160961079823323e-05,
      "loss": 4.3264,
      "step": 9475
    },
    {
      "epoch": 0.3825595478401292,
      "grad_norm": 72.83869934082031,
      "learning_rate": 1.4159772034413158e-05,
      "loss": 5.2364,
      "step": 9476
    },
    {
      "epoch": 0.3825999192571659,
      "grad_norm": 198.31895446777344,
      "learning_rate": 1.4158582917880603e-05,
      "loss": 4.9598,
      "step": 9477
    },
    {
      "epoch": 0.3826402906742027,
      "grad_norm": 88.1397705078125,
      "learning_rate": 1.4157393730245995e-05,
      "loss": 5.176,
      "step": 9478
    },
    {
      "epoch": 0.3826806620912394,
      "grad_norm": 76.55961608886719,
      "learning_rate": 1.4156204471529658e-05,
      "loss": 6.1553,
      "step": 9479
    },
    {
      "epoch": 0.38272103350827613,
      "grad_norm": 120.901611328125,
      "learning_rate": 1.415501514175193e-05,
      "loss": 5.6389,
      "step": 9480
    },
    {
      "epoch": 0.38276140492531285,
      "grad_norm": 72.55347442626953,
      "learning_rate": 1.4153825740933145e-05,
      "loss": 5.243,
      "step": 9481
    },
    {
      "epoch": 0.38280177634234963,
      "grad_norm": 130.95126342773438,
      "learning_rate": 1.4152636269093642e-05,
      "loss": 3.9558,
      "step": 9482
    },
    {
      "epoch": 0.38284214775938635,
      "grad_norm": 119.69512939453125,
      "learning_rate": 1.4151446726253758e-05,
      "loss": 5.9991,
      "step": 9483
    },
    {
      "epoch": 0.3828825191764231,
      "grad_norm": 79.14009857177734,
      "learning_rate": 1.4150257112433825e-05,
      "loss": 4.3085,
      "step": 9484
    },
    {
      "epoch": 0.38292289059345985,
      "grad_norm": 117.41752624511719,
      "learning_rate": 1.4149067427654188e-05,
      "loss": 3.9478,
      "step": 9485
    },
    {
      "epoch": 0.3829632620104966,
      "grad_norm": 73.41295623779297,
      "learning_rate": 1.414787767193519e-05,
      "loss": 5.8272,
      "step": 9486
    },
    {
      "epoch": 0.3830036334275333,
      "grad_norm": 126.56414794921875,
      "learning_rate": 1.4146687845297167e-05,
      "loss": 5.3313,
      "step": 9487
    },
    {
      "epoch": 0.38304400484457,
      "grad_norm": 67.56107330322266,
      "learning_rate": 1.4145497947760464e-05,
      "loss": 5.0483,
      "step": 9488
    },
    {
      "epoch": 0.3830843762616068,
      "grad_norm": 98.03704071044922,
      "learning_rate": 1.4144307979345428e-05,
      "loss": 5.2075,
      "step": 9489
    },
    {
      "epoch": 0.3831247476786435,
      "grad_norm": 114.8867416381836,
      "learning_rate": 1.4143117940072403e-05,
      "loss": 4.7538,
      "step": 9490
    },
    {
      "epoch": 0.38316511909568024,
      "grad_norm": 102.9859848022461,
      "learning_rate": 1.4141927829961737e-05,
      "loss": 6.5629,
      "step": 9491
    },
    {
      "epoch": 0.383205490512717,
      "grad_norm": 62.25128936767578,
      "learning_rate": 1.4140737649033776e-05,
      "loss": 4.8663,
      "step": 9492
    },
    {
      "epoch": 0.38324586192975374,
      "grad_norm": 99.0740966796875,
      "learning_rate": 1.413954739730887e-05,
      "loss": 5.4011,
      "step": 9493
    },
    {
      "epoch": 0.38328623334679046,
      "grad_norm": 95.06796264648438,
      "learning_rate": 1.4138357074807374e-05,
      "loss": 5.767,
      "step": 9494
    },
    {
      "epoch": 0.3833266047638272,
      "grad_norm": 80.57510375976562,
      "learning_rate": 1.413716668154963e-05,
      "loss": 6.4853,
      "step": 9495
    },
    {
      "epoch": 0.38336697618086396,
      "grad_norm": 76.34026336669922,
      "learning_rate": 1.4135976217556002e-05,
      "loss": 6.2289,
      "step": 9496
    },
    {
      "epoch": 0.3834073475979007,
      "grad_norm": 98.65409851074219,
      "learning_rate": 1.4134785682846835e-05,
      "loss": 4.4823,
      "step": 9497
    },
    {
      "epoch": 0.3834477190149374,
      "grad_norm": 200.8205108642578,
      "learning_rate": 1.4133595077442489e-05,
      "loss": 4.3032,
      "step": 9498
    },
    {
      "epoch": 0.3834880904319742,
      "grad_norm": 100.89970397949219,
      "learning_rate": 1.413240440136332e-05,
      "loss": 4.1368,
      "step": 9499
    },
    {
      "epoch": 0.3835284618490109,
      "grad_norm": 200.8482666015625,
      "learning_rate": 1.4131213654629688e-05,
      "loss": 3.9712,
      "step": 9500
    },
    {
      "epoch": 0.3835688332660476,
      "grad_norm": 236.13592529296875,
      "learning_rate": 1.4130022837261945e-05,
      "loss": 4.5651,
      "step": 9501
    },
    {
      "epoch": 0.3836092046830844,
      "grad_norm": 91.21553802490234,
      "learning_rate": 1.412883194928046e-05,
      "loss": 4.5619,
      "step": 9502
    },
    {
      "epoch": 0.3836495761001211,
      "grad_norm": 105.77423095703125,
      "learning_rate": 1.4127640990705586e-05,
      "loss": 6.654,
      "step": 9503
    },
    {
      "epoch": 0.38368994751715785,
      "grad_norm": 119.18049621582031,
      "learning_rate": 1.4126449961557695e-05,
      "loss": 6.1801,
      "step": 9504
    },
    {
      "epoch": 0.38373031893419457,
      "grad_norm": 117.08037567138672,
      "learning_rate": 1.4125258861857142e-05,
      "loss": 5.1649,
      "step": 9505
    },
    {
      "epoch": 0.38377069035123135,
      "grad_norm": 74.24005889892578,
      "learning_rate": 1.4124067691624296e-05,
      "loss": 4.2681,
      "step": 9506
    },
    {
      "epoch": 0.38381106176826807,
      "grad_norm": 67.84394836425781,
      "learning_rate": 1.4122876450879525e-05,
      "loss": 5.7448,
      "step": 9507
    },
    {
      "epoch": 0.3838514331853048,
      "grad_norm": 138.4846954345703,
      "learning_rate": 1.4121685139643191e-05,
      "loss": 5.975,
      "step": 9508
    },
    {
      "epoch": 0.38389180460234157,
      "grad_norm": 112.14171600341797,
      "learning_rate": 1.4120493757935666e-05,
      "loss": 5.7751,
      "step": 9509
    },
    {
      "epoch": 0.3839321760193783,
      "grad_norm": 104.1883544921875,
      "learning_rate": 1.4119302305777325e-05,
      "loss": 4.1529,
      "step": 9510
    },
    {
      "epoch": 0.383972547436415,
      "grad_norm": 99.33551025390625,
      "learning_rate": 1.411811078318853e-05,
      "loss": 4.9449,
      "step": 9511
    },
    {
      "epoch": 0.38401291885345173,
      "grad_norm": 91.36558532714844,
      "learning_rate": 1.4116919190189658e-05,
      "loss": 5.8664,
      "step": 9512
    },
    {
      "epoch": 0.3840532902704885,
      "grad_norm": 84.69263458251953,
      "learning_rate": 1.411572752680108e-05,
      "loss": 4.3073,
      "step": 9513
    },
    {
      "epoch": 0.38409366168752523,
      "grad_norm": 185.65574645996094,
      "learning_rate": 1.4114535793043173e-05,
      "loss": 6.5417,
      "step": 9514
    },
    {
      "epoch": 0.38413403310456196,
      "grad_norm": 148.08953857421875,
      "learning_rate": 1.4113343988936313e-05,
      "loss": 4.215,
      "step": 9515
    },
    {
      "epoch": 0.38417440452159873,
      "grad_norm": 93.01469421386719,
      "learning_rate": 1.4112152114500876e-05,
      "loss": 4.8155,
      "step": 9516
    },
    {
      "epoch": 0.38421477593863546,
      "grad_norm": 96.21949005126953,
      "learning_rate": 1.4110960169757239e-05,
      "loss": 5.8245,
      "step": 9517
    },
    {
      "epoch": 0.3842551473556722,
      "grad_norm": 158.0640869140625,
      "learning_rate": 1.4109768154725783e-05,
      "loss": 4.3053,
      "step": 9518
    },
    {
      "epoch": 0.3842955187727089,
      "grad_norm": 83.81981658935547,
      "learning_rate": 1.4108576069426886e-05,
      "loss": 5.4721,
      "step": 9519
    },
    {
      "epoch": 0.3843358901897457,
      "grad_norm": 82.86343383789062,
      "learning_rate": 1.4107383913880938e-05,
      "loss": 6.1095,
      "step": 9520
    },
    {
      "epoch": 0.3843762616067824,
      "grad_norm": 122.2884521484375,
      "learning_rate": 1.4106191688108315e-05,
      "loss": 4.5071,
      "step": 9521
    },
    {
      "epoch": 0.3844166330238191,
      "grad_norm": 77.01835632324219,
      "learning_rate": 1.4104999392129402e-05,
      "loss": 4.5944,
      "step": 9522
    },
    {
      "epoch": 0.3844570044408559,
      "grad_norm": 66.2397232055664,
      "learning_rate": 1.4103807025964585e-05,
      "loss": 6.2437,
      "step": 9523
    },
    {
      "epoch": 0.3844973758578926,
      "grad_norm": 127.18016815185547,
      "learning_rate": 1.4102614589634252e-05,
      "loss": 2.1056,
      "step": 9524
    },
    {
      "epoch": 0.38453774727492934,
      "grad_norm": 97.40646362304688,
      "learning_rate": 1.4101422083158788e-05,
      "loss": 6.2415,
      "step": 9525
    },
    {
      "epoch": 0.38457811869196606,
      "grad_norm": 89.99747467041016,
      "learning_rate": 1.4100229506558586e-05,
      "loss": 5.3603,
      "step": 9526
    },
    {
      "epoch": 0.38461849010900284,
      "grad_norm": 103.55206298828125,
      "learning_rate": 1.4099036859854031e-05,
      "loss": 5.9759,
      "step": 9527
    },
    {
      "epoch": 0.38465886152603956,
      "grad_norm": 111.47320556640625,
      "learning_rate": 1.409784414306552e-05,
      "loss": 6.8845,
      "step": 9528
    },
    {
      "epoch": 0.3846992329430763,
      "grad_norm": 109.3421401977539,
      "learning_rate": 1.4096651356213442e-05,
      "loss": 5.5826,
      "step": 9529
    },
    {
      "epoch": 0.38473960436011306,
      "grad_norm": 140.55047607421875,
      "learning_rate": 1.4095458499318195e-05,
      "loss": 5.2499,
      "step": 9530
    },
    {
      "epoch": 0.3847799757771498,
      "grad_norm": 131.51470947265625,
      "learning_rate": 1.4094265572400167e-05,
      "loss": 6.3474,
      "step": 9531
    },
    {
      "epoch": 0.3848203471941865,
      "grad_norm": 93.03820037841797,
      "learning_rate": 1.4093072575479762e-05,
      "loss": 4.8778,
      "step": 9532
    },
    {
      "epoch": 0.38486071861122323,
      "grad_norm": 121.34851837158203,
      "learning_rate": 1.4091879508577373e-05,
      "loss": 3.6941,
      "step": 9533
    },
    {
      "epoch": 0.38490109002826,
      "grad_norm": 115.6607437133789,
      "learning_rate": 1.4090686371713403e-05,
      "loss": 4.7882,
      "step": 9534
    },
    {
      "epoch": 0.38494146144529673,
      "grad_norm": 90.05876922607422,
      "learning_rate": 1.4089493164908242e-05,
      "loss": 5.0439,
      "step": 9535
    },
    {
      "epoch": 0.38498183286233345,
      "grad_norm": 96.52288055419922,
      "learning_rate": 1.4088299888182302e-05,
      "loss": 6.8458,
      "step": 9536
    },
    {
      "epoch": 0.38502220427937023,
      "grad_norm": 108.34784698486328,
      "learning_rate": 1.4087106541555979e-05,
      "loss": 5.2855,
      "step": 9537
    },
    {
      "epoch": 0.38506257569640695,
      "grad_norm": 104.50023651123047,
      "learning_rate": 1.4085913125049677e-05,
      "loss": 5.1263,
      "step": 9538
    },
    {
      "epoch": 0.38510294711344367,
      "grad_norm": 147.56163024902344,
      "learning_rate": 1.4084719638683804e-05,
      "loss": 4.0115,
      "step": 9539
    },
    {
      "epoch": 0.3851433185304804,
      "grad_norm": 86.99697875976562,
      "learning_rate": 1.4083526082478762e-05,
      "loss": 4.8332,
      "step": 9540
    },
    {
      "epoch": 0.38518368994751717,
      "grad_norm": 98.39898681640625,
      "learning_rate": 1.408233245645496e-05,
      "loss": 5.4897,
      "step": 9541
    },
    {
      "epoch": 0.3852240613645539,
      "grad_norm": 116.31082153320312,
      "learning_rate": 1.4081138760632809e-05,
      "loss": 6.281,
      "step": 9542
    },
    {
      "epoch": 0.3852644327815906,
      "grad_norm": 96.6448745727539,
      "learning_rate": 1.4079944995032713e-05,
      "loss": 4.532,
      "step": 9543
    },
    {
      "epoch": 0.3853048041986274,
      "grad_norm": 66.90399932861328,
      "learning_rate": 1.4078751159675083e-05,
      "loss": 3.7214,
      "step": 9544
    },
    {
      "epoch": 0.3853451756156641,
      "grad_norm": 87.49565887451172,
      "learning_rate": 1.4077557254580336e-05,
      "loss": 5.6821,
      "step": 9545
    },
    {
      "epoch": 0.38538554703270084,
      "grad_norm": 135.11077880859375,
      "learning_rate": 1.407636327976888e-05,
      "loss": 7.0931,
      "step": 9546
    },
    {
      "epoch": 0.38542591844973756,
      "grad_norm": 102.37297821044922,
      "learning_rate": 1.4075169235261132e-05,
      "loss": 5.574,
      "step": 9547
    },
    {
      "epoch": 0.38546628986677434,
      "grad_norm": 76.7230453491211,
      "learning_rate": 1.4073975121077506e-05,
      "loss": 6.0976,
      "step": 9548
    },
    {
      "epoch": 0.38550666128381106,
      "grad_norm": 113.57261657714844,
      "learning_rate": 1.4072780937238414e-05,
      "loss": 4.4209,
      "step": 9549
    },
    {
      "epoch": 0.3855470327008478,
      "grad_norm": 70.14423370361328,
      "learning_rate": 1.4071586683764286e-05,
      "loss": 4.2828,
      "step": 9550
    },
    {
      "epoch": 0.38558740411788456,
      "grad_norm": 85.27923583984375,
      "learning_rate": 1.407039236067553e-05,
      "loss": 3.9867,
      "step": 9551
    },
    {
      "epoch": 0.3856277755349213,
      "grad_norm": 152.34950256347656,
      "learning_rate": 1.4069197967992572e-05,
      "loss": 4.1795,
      "step": 9552
    },
    {
      "epoch": 0.385668146951958,
      "grad_norm": 80.29205322265625,
      "learning_rate": 1.4068003505735827e-05,
      "loss": 6.6241,
      "step": 9553
    },
    {
      "epoch": 0.3857085183689948,
      "grad_norm": 107.7098388671875,
      "learning_rate": 1.4066808973925726e-05,
      "loss": 5.7597,
      "step": 9554
    },
    {
      "epoch": 0.3857488897860315,
      "grad_norm": 121.512451171875,
      "learning_rate": 1.4065614372582685e-05,
      "loss": 5.5852,
      "step": 9555
    },
    {
      "epoch": 0.3857892612030682,
      "grad_norm": 94.89856719970703,
      "learning_rate": 1.4064419701727134e-05,
      "loss": 4.8651,
      "step": 9556
    },
    {
      "epoch": 0.38582963262010495,
      "grad_norm": 108.14376068115234,
      "learning_rate": 1.4063224961379495e-05,
      "loss": 5.88,
      "step": 9557
    },
    {
      "epoch": 0.3858700040371417,
      "grad_norm": 93.1170654296875,
      "learning_rate": 1.4062030151560197e-05,
      "loss": 4.1879,
      "step": 9558
    },
    {
      "epoch": 0.38591037545417844,
      "grad_norm": 123.50813293457031,
      "learning_rate": 1.4060835272289673e-05,
      "loss": 6.4282,
      "step": 9559
    },
    {
      "epoch": 0.38595074687121517,
      "grad_norm": 72.25642395019531,
      "learning_rate": 1.4059640323588346e-05,
      "loss": 5.5936,
      "step": 9560
    },
    {
      "epoch": 0.38599111828825194,
      "grad_norm": 270.4314880371094,
      "learning_rate": 1.405844530547665e-05,
      "loss": 5.3282,
      "step": 9561
    },
    {
      "epoch": 0.38603148970528867,
      "grad_norm": 91.22635650634766,
      "learning_rate": 1.4057250217975016e-05,
      "loss": 5.708,
      "step": 9562
    },
    {
      "epoch": 0.3860718611223254,
      "grad_norm": 77.9441909790039,
      "learning_rate": 1.4056055061103877e-05,
      "loss": 5.7523,
      "step": 9563
    },
    {
      "epoch": 0.3861122325393621,
      "grad_norm": 87.7235107421875,
      "learning_rate": 1.405485983488367e-05,
      "loss": 4.5656,
      "step": 9564
    },
    {
      "epoch": 0.3861526039563989,
      "grad_norm": 91.15422821044922,
      "learning_rate": 1.4053664539334828e-05,
      "loss": 5.1176,
      "step": 9565
    },
    {
      "epoch": 0.3861929753734356,
      "grad_norm": 126.21029663085938,
      "learning_rate": 1.4052469174477787e-05,
      "loss": 4.1236,
      "step": 9566
    },
    {
      "epoch": 0.38623334679047233,
      "grad_norm": 67.54086303710938,
      "learning_rate": 1.4051273740332986e-05,
      "loss": 4.1936,
      "step": 9567
    },
    {
      "epoch": 0.3862737182075091,
      "grad_norm": 80.22245025634766,
      "learning_rate": 1.4050078236920866e-05,
      "loss": 4.6909,
      "step": 9568
    },
    {
      "epoch": 0.38631408962454583,
      "grad_norm": 110.6187515258789,
      "learning_rate": 1.4048882664261865e-05,
      "loss": 5.6115,
      "step": 9569
    },
    {
      "epoch": 0.38635446104158255,
      "grad_norm": 84.4228286743164,
      "learning_rate": 1.4047687022376426e-05,
      "loss": 6.6438,
      "step": 9570
    },
    {
      "epoch": 0.3863948324586193,
      "grad_norm": 86.36178588867188,
      "learning_rate": 1.4046491311284988e-05,
      "loss": 3.9585,
      "step": 9571
    },
    {
      "epoch": 0.38643520387565605,
      "grad_norm": 79.0615463256836,
      "learning_rate": 1.4045295531008003e-05,
      "loss": 5.029,
      "step": 9572
    },
    {
      "epoch": 0.3864755752926928,
      "grad_norm": 102.21102905273438,
      "learning_rate": 1.4044099681565909e-05,
      "loss": 4.723,
      "step": 9573
    },
    {
      "epoch": 0.3865159467097295,
      "grad_norm": 128.68116760253906,
      "learning_rate": 1.4042903762979155e-05,
      "loss": 6.4528,
      "step": 9574
    },
    {
      "epoch": 0.3865563181267663,
      "grad_norm": 118.48135375976562,
      "learning_rate": 1.4041707775268183e-05,
      "loss": 3.7163,
      "step": 9575
    },
    {
      "epoch": 0.386596689543803,
      "grad_norm": 101.76148223876953,
      "learning_rate": 1.4040511718453451e-05,
      "loss": 4.4051,
      "step": 9576
    },
    {
      "epoch": 0.3866370609608397,
      "grad_norm": 96.3560791015625,
      "learning_rate": 1.40393155925554e-05,
      "loss": 5.8805,
      "step": 9577
    },
    {
      "epoch": 0.38667743237787644,
      "grad_norm": 118.15138244628906,
      "learning_rate": 1.4038119397594488e-05,
      "loss": 5.9741,
      "step": 9578
    },
    {
      "epoch": 0.3867178037949132,
      "grad_norm": 65.97718811035156,
      "learning_rate": 1.403692313359116e-05,
      "loss": 5.2567,
      "step": 9579
    },
    {
      "epoch": 0.38675817521194994,
      "grad_norm": 88.23685455322266,
      "learning_rate": 1.4035726800565876e-05,
      "loss": 6.3831,
      "step": 9580
    },
    {
      "epoch": 0.38679854662898666,
      "grad_norm": 99.64154052734375,
      "learning_rate": 1.4034530398539088e-05,
      "loss": 6.0453,
      "step": 9581
    },
    {
      "epoch": 0.38683891804602344,
      "grad_norm": 78.15119934082031,
      "learning_rate": 1.4033333927531252e-05,
      "loss": 3.14,
      "step": 9582
    },
    {
      "epoch": 0.38687928946306016,
      "grad_norm": 68.31311798095703,
      "learning_rate": 1.4032137387562824e-05,
      "loss": 4.0352,
      "step": 9583
    },
    {
      "epoch": 0.3869196608800969,
      "grad_norm": 74.79236602783203,
      "learning_rate": 1.4030940778654264e-05,
      "loss": 6.2136,
      "step": 9584
    },
    {
      "epoch": 0.3869600322971336,
      "grad_norm": 110.43376922607422,
      "learning_rate": 1.4029744100826026e-05,
      "loss": 5.3247,
      "step": 9585
    },
    {
      "epoch": 0.3870004037141704,
      "grad_norm": 84.36026763916016,
      "learning_rate": 1.4028547354098576e-05,
      "loss": 5.2266,
      "step": 9586
    },
    {
      "epoch": 0.3870407751312071,
      "grad_norm": 81.1049575805664,
      "learning_rate": 1.4027350538492372e-05,
      "loss": 5.171,
      "step": 9587
    },
    {
      "epoch": 0.3870811465482438,
      "grad_norm": 67.79434967041016,
      "learning_rate": 1.4026153654027879e-05,
      "loss": 6.6027,
      "step": 9588
    },
    {
      "epoch": 0.3871215179652806,
      "grad_norm": 109.4482192993164,
      "learning_rate": 1.4024956700725561e-05,
      "loss": 4.9365,
      "step": 9589
    },
    {
      "epoch": 0.3871618893823173,
      "grad_norm": 124.03337097167969,
      "learning_rate": 1.4023759678605884e-05,
      "loss": 6.0027,
      "step": 9590
    },
    {
      "epoch": 0.38720226079935405,
      "grad_norm": 111.45262145996094,
      "learning_rate": 1.4022562587689307e-05,
      "loss": 5.9968,
      "step": 9591
    },
    {
      "epoch": 0.38724263221639077,
      "grad_norm": 97.6661148071289,
      "learning_rate": 1.4021365427996309e-05,
      "loss": 5.0892,
      "step": 9592
    },
    {
      "epoch": 0.38728300363342755,
      "grad_norm": 100.95582580566406,
      "learning_rate": 1.402016819954735e-05,
      "loss": 5.8194,
      "step": 9593
    },
    {
      "epoch": 0.38732337505046427,
      "grad_norm": 102.77287292480469,
      "learning_rate": 1.4018970902362905e-05,
      "loss": 4.6241,
      "step": 9594
    },
    {
      "epoch": 0.387363746467501,
      "grad_norm": 412.96990966796875,
      "learning_rate": 1.401777353646344e-05,
      "loss": 4.6104,
      "step": 9595
    },
    {
      "epoch": 0.38740411788453777,
      "grad_norm": 86.93370819091797,
      "learning_rate": 1.401657610186943e-05,
      "loss": 6.8695,
      "step": 9596
    },
    {
      "epoch": 0.3874444893015745,
      "grad_norm": 129.4237518310547,
      "learning_rate": 1.401537859860135e-05,
      "loss": 4.6801,
      "step": 9597
    },
    {
      "epoch": 0.3874848607186112,
      "grad_norm": 151.8002166748047,
      "learning_rate": 1.401418102667967e-05,
      "loss": 5.7893,
      "step": 9598
    },
    {
      "epoch": 0.38752523213564793,
      "grad_norm": 87.23470306396484,
      "learning_rate": 1.4012983386124868e-05,
      "loss": 6.0065,
      "step": 9599
    },
    {
      "epoch": 0.3875656035526847,
      "grad_norm": 108.07160186767578,
      "learning_rate": 1.4011785676957423e-05,
      "loss": 4.077,
      "step": 9600
    },
    {
      "epoch": 0.38760597496972143,
      "grad_norm": 62.8158073425293,
      "learning_rate": 1.4010587899197809e-05,
      "loss": 4.723,
      "step": 9601
    },
    {
      "epoch": 0.38764634638675816,
      "grad_norm": 61.324581146240234,
      "learning_rate": 1.4009390052866509e-05,
      "loss": 3.9659,
      "step": 9602
    },
    {
      "epoch": 0.38768671780379493,
      "grad_norm": 72.73640441894531,
      "learning_rate": 1.4008192137984002e-05,
      "loss": 4.258,
      "step": 9603
    },
    {
      "epoch": 0.38772708922083166,
      "grad_norm": 104.63163757324219,
      "learning_rate": 1.400699415457077e-05,
      "loss": 4.4271,
      "step": 9604
    },
    {
      "epoch": 0.3877674606378684,
      "grad_norm": 106.25909423828125,
      "learning_rate": 1.4005796102647293e-05,
      "loss": 3.5225,
      "step": 9605
    },
    {
      "epoch": 0.3878078320549051,
      "grad_norm": 78.82421112060547,
      "learning_rate": 1.4004597982234056e-05,
      "loss": 4.7236,
      "step": 9606
    },
    {
      "epoch": 0.3878482034719419,
      "grad_norm": 91.01831817626953,
      "learning_rate": 1.4003399793351546e-05,
      "loss": 4.6929,
      "step": 9607
    },
    {
      "epoch": 0.3878885748889786,
      "grad_norm": 81.04521942138672,
      "learning_rate": 1.4002201536020251e-05,
      "loss": 6.7955,
      "step": 9608
    },
    {
      "epoch": 0.3879289463060153,
      "grad_norm": 251.06141662597656,
      "learning_rate": 1.400100321026065e-05,
      "loss": 4.493,
      "step": 9609
    },
    {
      "epoch": 0.3879693177230521,
      "grad_norm": 73.32392883300781,
      "learning_rate": 1.3999804816093243e-05,
      "loss": 4.9676,
      "step": 9610
    },
    {
      "epoch": 0.3880096891400888,
      "grad_norm": 130.61007690429688,
      "learning_rate": 1.3998606353538511e-05,
      "loss": 4.867,
      "step": 9611
    },
    {
      "epoch": 0.38805006055712554,
      "grad_norm": 95.83505249023438,
      "learning_rate": 1.3997407822616948e-05,
      "loss": 5.3704,
      "step": 9612
    },
    {
      "epoch": 0.3880904319741623,
      "grad_norm": 71.9994125366211,
      "learning_rate": 1.3996209223349047e-05,
      "loss": 6.5728,
      "step": 9613
    },
    {
      "epoch": 0.38813080339119904,
      "grad_norm": 121.13922882080078,
      "learning_rate": 1.39950105557553e-05,
      "loss": 6.2691,
      "step": 9614
    },
    {
      "epoch": 0.38817117480823576,
      "grad_norm": 106.8058090209961,
      "learning_rate": 1.3993811819856198e-05,
      "loss": 6.1229,
      "step": 9615
    },
    {
      "epoch": 0.3882115462252725,
      "grad_norm": 97.53498840332031,
      "learning_rate": 1.399261301567224e-05,
      "loss": 5.9634,
      "step": 9616
    },
    {
      "epoch": 0.38825191764230926,
      "grad_norm": 97.55683135986328,
      "learning_rate": 1.3991414143223926e-05,
      "loss": 4.596,
      "step": 9617
    },
    {
      "epoch": 0.388292289059346,
      "grad_norm": 139.16439819335938,
      "learning_rate": 1.3990215202531752e-05,
      "loss": 6.4346,
      "step": 9618
    },
    {
      "epoch": 0.3883326604763827,
      "grad_norm": 83.9403076171875,
      "learning_rate": 1.3989016193616212e-05,
      "loss": 5.1876,
      "step": 9619
    },
    {
      "epoch": 0.3883730318934195,
      "grad_norm": 198.6851348876953,
      "learning_rate": 1.3987817116497813e-05,
      "loss": 5.0467,
      "step": 9620
    },
    {
      "epoch": 0.3884134033104562,
      "grad_norm": 113.31837463378906,
      "learning_rate": 1.3986617971197051e-05,
      "loss": 5.3595,
      "step": 9621
    },
    {
      "epoch": 0.38845377472749293,
      "grad_norm": 93.66902923583984,
      "learning_rate": 1.3985418757734435e-05,
      "loss": 5.5367,
      "step": 9622
    },
    {
      "epoch": 0.38849414614452965,
      "grad_norm": 75.28069305419922,
      "learning_rate": 1.3984219476130463e-05,
      "loss": 4.9278,
      "step": 9623
    },
    {
      "epoch": 0.38853451756156643,
      "grad_norm": 84.96159362792969,
      "learning_rate": 1.3983020126405642e-05,
      "loss": 5.0917,
      "step": 9624
    },
    {
      "epoch": 0.38857488897860315,
      "grad_norm": 79.98539733886719,
      "learning_rate": 1.3981820708580476e-05,
      "loss": 3.9883,
      "step": 9625
    },
    {
      "epoch": 0.38861526039563987,
      "grad_norm": 76.15238952636719,
      "learning_rate": 1.3980621222675474e-05,
      "loss": 4.0372,
      "step": 9626
    },
    {
      "epoch": 0.38865563181267665,
      "grad_norm": 89.18585968017578,
      "learning_rate": 1.3979421668711144e-05,
      "loss": 6.6431,
      "step": 9627
    },
    {
      "epoch": 0.38869600322971337,
      "grad_norm": 127.2226791381836,
      "learning_rate": 1.3978222046708002e-05,
      "loss": 5.0163,
      "step": 9628
    },
    {
      "epoch": 0.3887363746467501,
      "grad_norm": 90.37269592285156,
      "learning_rate": 1.3977022356686547e-05,
      "loss": 6.1758,
      "step": 9629
    },
    {
      "epoch": 0.3887767460637868,
      "grad_norm": 100.61894226074219,
      "learning_rate": 1.39758225986673e-05,
      "loss": 6.2519,
      "step": 9630
    },
    {
      "epoch": 0.3888171174808236,
      "grad_norm": 89.80952453613281,
      "learning_rate": 1.3974622772670768e-05,
      "loss": 5.614,
      "step": 9631
    },
    {
      "epoch": 0.3888574888978603,
      "grad_norm": 94.63456726074219,
      "learning_rate": 1.3973422878717471e-05,
      "loss": 4.8812,
      "step": 9632
    },
    {
      "epoch": 0.38889786031489704,
      "grad_norm": 133.9526824951172,
      "learning_rate": 1.3972222916827918e-05,
      "loss": 5.5163,
      "step": 9633
    },
    {
      "epoch": 0.3889382317319338,
      "grad_norm": 88.28728485107422,
      "learning_rate": 1.3971022887022632e-05,
      "loss": 5.0128,
      "step": 9634
    },
    {
      "epoch": 0.38897860314897054,
      "grad_norm": 115.69851684570312,
      "learning_rate": 1.3969822789322126e-05,
      "loss": 3.5796,
      "step": 9635
    },
    {
      "epoch": 0.38901897456600726,
      "grad_norm": 213.23968505859375,
      "learning_rate": 1.396862262374692e-05,
      "loss": 4.8215,
      "step": 9636
    },
    {
      "epoch": 0.389059345983044,
      "grad_norm": 145.6709747314453,
      "learning_rate": 1.3967422390317534e-05,
      "loss": 6.352,
      "step": 9637
    },
    {
      "epoch": 0.38909971740008076,
      "grad_norm": 70.46807098388672,
      "learning_rate": 1.3966222089054492e-05,
      "loss": 5.9758,
      "step": 9638
    },
    {
      "epoch": 0.3891400888171175,
      "grad_norm": 106.40062713623047,
      "learning_rate": 1.3965021719978312e-05,
      "loss": 5.1292,
      "step": 9639
    },
    {
      "epoch": 0.3891804602341542,
      "grad_norm": 120.88745880126953,
      "learning_rate": 1.396382128310952e-05,
      "loss": 4.7632,
      "step": 9640
    },
    {
      "epoch": 0.389220831651191,
      "grad_norm": 95.13226318359375,
      "learning_rate": 1.3962620778468641e-05,
      "loss": 5.5721,
      "step": 9641
    },
    {
      "epoch": 0.3892612030682277,
      "grad_norm": 106.97667694091797,
      "learning_rate": 1.3961420206076199e-05,
      "loss": 5.8337,
      "step": 9642
    },
    {
      "epoch": 0.3893015744852644,
      "grad_norm": 95.90023040771484,
      "learning_rate": 1.3960219565952724e-05,
      "loss": 4.6557,
      "step": 9643
    },
    {
      "epoch": 0.38934194590230115,
      "grad_norm": 94.09966278076172,
      "learning_rate": 1.3959018858118742e-05,
      "loss": 3.9198,
      "step": 9644
    },
    {
      "epoch": 0.3893823173193379,
      "grad_norm": 84.45587158203125,
      "learning_rate": 1.395781808259478e-05,
      "loss": 5.4043,
      "step": 9645
    },
    {
      "epoch": 0.38942268873637464,
      "grad_norm": 92.30826568603516,
      "learning_rate": 1.3956617239401373e-05,
      "loss": 4.4541,
      "step": 9646
    },
    {
      "epoch": 0.38946306015341137,
      "grad_norm": 76.85587310791016,
      "learning_rate": 1.3955416328559048e-05,
      "loss": 4.6019,
      "step": 9647
    },
    {
      "epoch": 0.38950343157044814,
      "grad_norm": 78.40667724609375,
      "learning_rate": 1.3954215350088345e-05,
      "loss": 4.5905,
      "step": 9648
    },
    {
      "epoch": 0.38954380298748487,
      "grad_norm": 123.1654052734375,
      "learning_rate": 1.395301430400979e-05,
      "loss": 4.9514,
      "step": 9649
    },
    {
      "epoch": 0.3895841744045216,
      "grad_norm": 92.78897094726562,
      "learning_rate": 1.3951813190343924e-05,
      "loss": 4.1297,
      "step": 9650
    },
    {
      "epoch": 0.3896245458215583,
      "grad_norm": 122.8521957397461,
      "learning_rate": 1.3950612009111277e-05,
      "loss": 5.1301,
      "step": 9651
    },
    {
      "epoch": 0.3896649172385951,
      "grad_norm": 118.04666900634766,
      "learning_rate": 1.3949410760332394e-05,
      "loss": 5.5868,
      "step": 9652
    },
    {
      "epoch": 0.3897052886556318,
      "grad_norm": 79.61268615722656,
      "learning_rate": 1.3948209444027804e-05,
      "loss": 5.0388,
      "step": 9653
    },
    {
      "epoch": 0.38974566007266853,
      "grad_norm": 84.05279541015625,
      "learning_rate": 1.3947008060218056e-05,
      "loss": 4.9248,
      "step": 9654
    },
    {
      "epoch": 0.3897860314897053,
      "grad_norm": 179.05517578125,
      "learning_rate": 1.394580660892369e-05,
      "loss": 5.1429,
      "step": 9655
    },
    {
      "epoch": 0.38982640290674203,
      "grad_norm": 118.8216781616211,
      "learning_rate": 1.3944605090165238e-05,
      "loss": 5.4369,
      "step": 9656
    },
    {
      "epoch": 0.38986677432377875,
      "grad_norm": 78.21127319335938,
      "learning_rate": 1.3943403503963253e-05,
      "loss": 4.2983,
      "step": 9657
    },
    {
      "epoch": 0.3899071457408155,
      "grad_norm": 130.04757690429688,
      "learning_rate": 1.394220185033828e-05,
      "loss": 6.0301,
      "step": 9658
    },
    {
      "epoch": 0.38994751715785225,
      "grad_norm": 159.52078247070312,
      "learning_rate": 1.3941000129310858e-05,
      "loss": 4.9842,
      "step": 9659
    },
    {
      "epoch": 0.389987888574889,
      "grad_norm": 75.55535125732422,
      "learning_rate": 1.3939798340901538e-05,
      "loss": 3.7967,
      "step": 9660
    },
    {
      "epoch": 0.3900282599919257,
      "grad_norm": 93.84297943115234,
      "learning_rate": 1.3938596485130865e-05,
      "loss": 4.9917,
      "step": 9661
    },
    {
      "epoch": 0.3900686314089625,
      "grad_norm": 76.23312377929688,
      "learning_rate": 1.3937394562019391e-05,
      "loss": 6.5946,
      "step": 9662
    },
    {
      "epoch": 0.3901090028259992,
      "grad_norm": 181.55975341796875,
      "learning_rate": 1.3936192571587664e-05,
      "loss": 3.9471,
      "step": 9663
    },
    {
      "epoch": 0.3901493742430359,
      "grad_norm": 93.35986328125,
      "learning_rate": 1.3934990513856234e-05,
      "loss": 3.8698,
      "step": 9664
    },
    {
      "epoch": 0.3901897456600727,
      "grad_norm": 77.0617904663086,
      "learning_rate": 1.3933788388845657e-05,
      "loss": 4.8141,
      "step": 9665
    },
    {
      "epoch": 0.3902301170771094,
      "grad_norm": 117.12506866455078,
      "learning_rate": 1.3932586196576484e-05,
      "loss": 3.3503,
      "step": 9666
    },
    {
      "epoch": 0.39027048849414614,
      "grad_norm": 119.61095428466797,
      "learning_rate": 1.393138393706927e-05,
      "loss": 5.6043,
      "step": 9667
    },
    {
      "epoch": 0.39031085991118286,
      "grad_norm": 95.07471466064453,
      "learning_rate": 1.3930181610344571e-05,
      "loss": 4.9689,
      "step": 9668
    },
    {
      "epoch": 0.39035123132821964,
      "grad_norm": 124.7652587890625,
      "learning_rate": 1.3928979216422945e-05,
      "loss": 4.2537,
      "step": 9669
    },
    {
      "epoch": 0.39039160274525636,
      "grad_norm": 114.35406494140625,
      "learning_rate": 1.392777675532495e-05,
      "loss": 4.8854,
      "step": 9670
    },
    {
      "epoch": 0.3904319741622931,
      "grad_norm": 118.11788177490234,
      "learning_rate": 1.3926574227071141e-05,
      "loss": 4.6084,
      "step": 9671
    },
    {
      "epoch": 0.39047234557932986,
      "grad_norm": 98.25094604492188,
      "learning_rate": 1.3925371631682086e-05,
      "loss": 6.5854,
      "step": 9672
    },
    {
      "epoch": 0.3905127169963666,
      "grad_norm": 77.79581451416016,
      "learning_rate": 1.392416896917834e-05,
      "loss": 6.6785,
      "step": 9673
    },
    {
      "epoch": 0.3905530884134033,
      "grad_norm": 88.09902954101562,
      "learning_rate": 1.392296623958047e-05,
      "loss": 5.5362,
      "step": 9674
    },
    {
      "epoch": 0.39059345983044,
      "grad_norm": 89.79370880126953,
      "learning_rate": 1.3921763442909035e-05,
      "loss": 3.579,
      "step": 9675
    },
    {
      "epoch": 0.3906338312474768,
      "grad_norm": 70.40743255615234,
      "learning_rate": 1.3920560579184606e-05,
      "loss": 7.0488,
      "step": 9676
    },
    {
      "epoch": 0.3906742026645135,
      "grad_norm": 127.30722045898438,
      "learning_rate": 1.3919357648427743e-05,
      "loss": 4.8261,
      "step": 9677
    },
    {
      "epoch": 0.39071457408155025,
      "grad_norm": 77.06878662109375,
      "learning_rate": 1.3918154650659022e-05,
      "loss": 5.9198,
      "step": 9678
    },
    {
      "epoch": 0.390754945498587,
      "grad_norm": 145.2030029296875,
      "learning_rate": 1.3916951585899002e-05,
      "loss": 5.0853,
      "step": 9679
    },
    {
      "epoch": 0.39079531691562375,
      "grad_norm": 128.595458984375,
      "learning_rate": 1.3915748454168259e-05,
      "loss": 4.9265,
      "step": 9680
    },
    {
      "epoch": 0.39083568833266047,
      "grad_norm": 154.47225952148438,
      "learning_rate": 1.3914545255487358e-05,
      "loss": 4.851,
      "step": 9681
    },
    {
      "epoch": 0.3908760597496972,
      "grad_norm": 105.0886459350586,
      "learning_rate": 1.3913341989876876e-05,
      "loss": 3.7284,
      "step": 9682
    },
    {
      "epoch": 0.39091643116673397,
      "grad_norm": 81.70248413085938,
      "learning_rate": 1.3912138657357384e-05,
      "loss": 5.773,
      "step": 9683
    },
    {
      "epoch": 0.3909568025837707,
      "grad_norm": 101.28958129882812,
      "learning_rate": 1.3910935257949459e-05,
      "loss": 4.6671,
      "step": 9684
    },
    {
      "epoch": 0.3909971740008074,
      "grad_norm": 105.03034210205078,
      "learning_rate": 1.390973179167367e-05,
      "loss": 4.8089,
      "step": 9685
    },
    {
      "epoch": 0.3910375454178442,
      "grad_norm": 125.61607360839844,
      "learning_rate": 1.3908528258550597e-05,
      "loss": 6.5989,
      "step": 9686
    },
    {
      "epoch": 0.3910779168348809,
      "grad_norm": 95.93203735351562,
      "learning_rate": 1.3907324658600821e-05,
      "loss": 2.9987,
      "step": 9687
    },
    {
      "epoch": 0.39111828825191763,
      "grad_norm": 82.91341400146484,
      "learning_rate": 1.3906120991844917e-05,
      "loss": 5.4122,
      "step": 9688
    },
    {
      "epoch": 0.39115865966895436,
      "grad_norm": 164.4285125732422,
      "learning_rate": 1.3904917258303466e-05,
      "loss": 5.4405,
      "step": 9689
    },
    {
      "epoch": 0.39119903108599113,
      "grad_norm": 179.08306884765625,
      "learning_rate": 1.3903713457997046e-05,
      "loss": 6.2096,
      "step": 9690
    },
    {
      "epoch": 0.39123940250302786,
      "grad_norm": 121.1915512084961,
      "learning_rate": 1.3902509590946243e-05,
      "loss": 5.9033,
      "step": 9691
    },
    {
      "epoch": 0.3912797739200646,
      "grad_norm": 106.57726287841797,
      "learning_rate": 1.390130565717164e-05,
      "loss": 3.53,
      "step": 9692
    },
    {
      "epoch": 0.39132014533710135,
      "grad_norm": 85.73502349853516,
      "learning_rate": 1.390010165669382e-05,
      "loss": 5.2513,
      "step": 9693
    },
    {
      "epoch": 0.3913605167541381,
      "grad_norm": 117.95225524902344,
      "learning_rate": 1.389889758953337e-05,
      "loss": 4.1956,
      "step": 9694
    },
    {
      "epoch": 0.3914008881711748,
      "grad_norm": 96.70246887207031,
      "learning_rate": 1.3897693455710873e-05,
      "loss": 4.0941,
      "step": 9695
    },
    {
      "epoch": 0.3914412595882115,
      "grad_norm": 86.177734375,
      "learning_rate": 1.3896489255246919e-05,
      "loss": 5.486,
      "step": 9696
    },
    {
      "epoch": 0.3914816310052483,
      "grad_norm": 76.86314392089844,
      "learning_rate": 1.38952849881621e-05,
      "loss": 5.4689,
      "step": 9697
    },
    {
      "epoch": 0.391522002422285,
      "grad_norm": 97.7924575805664,
      "learning_rate": 1.3894080654477005e-05,
      "loss": 5.3032,
      "step": 9698
    },
    {
      "epoch": 0.39156237383932174,
      "grad_norm": 90.08085632324219,
      "learning_rate": 1.3892876254212222e-05,
      "loss": 4.8173,
      "step": 9699
    },
    {
      "epoch": 0.3916027452563585,
      "grad_norm": 203.15286254882812,
      "learning_rate": 1.3891671787388348e-05,
      "loss": 5.3635,
      "step": 9700
    },
    {
      "epoch": 0.39164311667339524,
      "grad_norm": 98.86768341064453,
      "learning_rate": 1.389046725402597e-05,
      "loss": 4.4456,
      "step": 9701
    },
    {
      "epoch": 0.39168348809043196,
      "grad_norm": 106.26473236083984,
      "learning_rate": 1.3889262654145692e-05,
      "loss": 4.1596,
      "step": 9702
    },
    {
      "epoch": 0.3917238595074687,
      "grad_norm": 115.03801727294922,
      "learning_rate": 1.38880579877681e-05,
      "loss": 5.9334,
      "step": 9703
    },
    {
      "epoch": 0.39176423092450546,
      "grad_norm": 80.07144927978516,
      "learning_rate": 1.3886853254913797e-05,
      "loss": 4.4706,
      "step": 9704
    },
    {
      "epoch": 0.3918046023415422,
      "grad_norm": 181.51625061035156,
      "learning_rate": 1.388564845560338e-05,
      "loss": 6.8872,
      "step": 9705
    },
    {
      "epoch": 0.3918449737585789,
      "grad_norm": 101.3532485961914,
      "learning_rate": 1.388444358985745e-05,
      "loss": 5.7845,
      "step": 9706
    },
    {
      "epoch": 0.3918853451756157,
      "grad_norm": 96.31616973876953,
      "learning_rate": 1.3883238657696603e-05,
      "loss": 5.1847,
      "step": 9707
    },
    {
      "epoch": 0.3919257165926524,
      "grad_norm": 122.51396942138672,
      "learning_rate": 1.3882033659141445e-05,
      "loss": 4.2564,
      "step": 9708
    },
    {
      "epoch": 0.39196608800968913,
      "grad_norm": 70.5292739868164,
      "learning_rate": 1.3880828594212573e-05,
      "loss": 6.0165,
      "step": 9709
    },
    {
      "epoch": 0.39200645942672585,
      "grad_norm": 142.20944213867188,
      "learning_rate": 1.38796234629306e-05,
      "loss": 5.2546,
      "step": 9710
    },
    {
      "epoch": 0.39204683084376263,
      "grad_norm": 110.53654479980469,
      "learning_rate": 1.387841826531612e-05,
      "loss": 7.5319,
      "step": 9711
    },
    {
      "epoch": 0.39208720226079935,
      "grad_norm": 157.39308166503906,
      "learning_rate": 1.387721300138975e-05,
      "loss": 5.7746,
      "step": 9712
    },
    {
      "epoch": 0.39212757367783607,
      "grad_norm": 86.27949523925781,
      "learning_rate": 1.3876007671172087e-05,
      "loss": 5.0792,
      "step": 9713
    },
    {
      "epoch": 0.39216794509487285,
      "grad_norm": 129.5948944091797,
      "learning_rate": 1.3874802274683745e-05,
      "loss": 5.1829,
      "step": 9714
    },
    {
      "epoch": 0.39220831651190957,
      "grad_norm": 205.20045471191406,
      "learning_rate": 1.3873596811945332e-05,
      "loss": 5.1812,
      "step": 9715
    },
    {
      "epoch": 0.3922486879289463,
      "grad_norm": 71.00083923339844,
      "learning_rate": 1.387239128297746e-05,
      "loss": 4.5676,
      "step": 9716
    },
    {
      "epoch": 0.39228905934598307,
      "grad_norm": 82.43401336669922,
      "learning_rate": 1.3871185687800739e-05,
      "loss": 3.8217,
      "step": 9717
    },
    {
      "epoch": 0.3923294307630198,
      "grad_norm": 92.57286834716797,
      "learning_rate": 1.3869980026435786e-05,
      "loss": 5.4435,
      "step": 9718
    },
    {
      "epoch": 0.3923698021800565,
      "grad_norm": 110.24869537353516,
      "learning_rate": 1.3868774298903206e-05,
      "loss": 6.3105,
      "step": 9719
    },
    {
      "epoch": 0.39241017359709324,
      "grad_norm": 88.04354095458984,
      "learning_rate": 1.3867568505223625e-05,
      "loss": 4.9106,
      "step": 9720
    },
    {
      "epoch": 0.39245054501413,
      "grad_norm": 102.45600891113281,
      "learning_rate": 1.386636264541765e-05,
      "loss": 4.9224,
      "step": 9721
    },
    {
      "epoch": 0.39249091643116674,
      "grad_norm": 123.89952087402344,
      "learning_rate": 1.3865156719505905e-05,
      "loss": 4.4259,
      "step": 9722
    },
    {
      "epoch": 0.39253128784820346,
      "grad_norm": 88.80886840820312,
      "learning_rate": 1.3863950727509004e-05,
      "loss": 5.492,
      "step": 9723
    },
    {
      "epoch": 0.39257165926524024,
      "grad_norm": 89.29267120361328,
      "learning_rate": 1.3862744669447568e-05,
      "loss": 4.2612,
      "step": 9724
    },
    {
      "epoch": 0.39261203068227696,
      "grad_norm": 68.30879211425781,
      "learning_rate": 1.3861538545342219e-05,
      "loss": 4.6675,
      "step": 9725
    },
    {
      "epoch": 0.3926524020993137,
      "grad_norm": 108.4274673461914,
      "learning_rate": 1.386033235521358e-05,
      "loss": 5.8996,
      "step": 9726
    },
    {
      "epoch": 0.3926927735163504,
      "grad_norm": 88.58248901367188,
      "learning_rate": 1.3859126099082269e-05,
      "loss": 6.7712,
      "step": 9727
    },
    {
      "epoch": 0.3927331449333872,
      "grad_norm": 86.03562927246094,
      "learning_rate": 1.3857919776968918e-05,
      "loss": 5.4675,
      "step": 9728
    },
    {
      "epoch": 0.3927735163504239,
      "grad_norm": 98.74466705322266,
      "learning_rate": 1.3856713388894144e-05,
      "loss": 5.3187,
      "step": 9729
    },
    {
      "epoch": 0.3928138877674606,
      "grad_norm": 110.5521240234375,
      "learning_rate": 1.385550693487858e-05,
      "loss": 4.1376,
      "step": 9730
    },
    {
      "epoch": 0.3928542591844974,
      "grad_norm": 105.51371002197266,
      "learning_rate": 1.3854300414942847e-05,
      "loss": 3.9061,
      "step": 9731
    },
    {
      "epoch": 0.3928946306015341,
      "grad_norm": 86.07538604736328,
      "learning_rate": 1.3853093829107582e-05,
      "loss": 5.4806,
      "step": 9732
    },
    {
      "epoch": 0.39293500201857084,
      "grad_norm": 99.21244812011719,
      "learning_rate": 1.3851887177393407e-05,
      "loss": 6.0482,
      "step": 9733
    },
    {
      "epoch": 0.39297537343560757,
      "grad_norm": 163.75071716308594,
      "learning_rate": 1.3850680459820956e-05,
      "loss": 4.4561,
      "step": 9734
    },
    {
      "epoch": 0.39301574485264434,
      "grad_norm": 85.66432189941406,
      "learning_rate": 1.3849473676410858e-05,
      "loss": 6.4407,
      "step": 9735
    },
    {
      "epoch": 0.39305611626968107,
      "grad_norm": 98.87140655517578,
      "learning_rate": 1.3848266827183756e-05,
      "loss": 6.3511,
      "step": 9736
    },
    {
      "epoch": 0.3930964876867178,
      "grad_norm": 97.58049774169922,
      "learning_rate": 1.3847059912160272e-05,
      "loss": 6.0265,
      "step": 9737
    },
    {
      "epoch": 0.39313685910375457,
      "grad_norm": 98.5539779663086,
      "learning_rate": 1.3845852931361052e-05,
      "loss": 6.2563,
      "step": 9738
    },
    {
      "epoch": 0.3931772305207913,
      "grad_norm": 80.74366760253906,
      "learning_rate": 1.3844645884806725e-05,
      "loss": 5.4222,
      "step": 9739
    },
    {
      "epoch": 0.393217601937828,
      "grad_norm": 83.25718688964844,
      "learning_rate": 1.3843438772517932e-05,
      "loss": 4.0356,
      "step": 9740
    },
    {
      "epoch": 0.39325797335486473,
      "grad_norm": 84.2221908569336,
      "learning_rate": 1.384223159451531e-05,
      "loss": 5.6154,
      "step": 9741
    },
    {
      "epoch": 0.3932983447719015,
      "grad_norm": 207.5900421142578,
      "learning_rate": 1.38410243508195e-05,
      "loss": 4.9143,
      "step": 9742
    },
    {
      "epoch": 0.39333871618893823,
      "grad_norm": 91.24169158935547,
      "learning_rate": 1.3839817041451142e-05,
      "loss": 3.9822,
      "step": 9743
    },
    {
      "epoch": 0.39337908760597495,
      "grad_norm": 117.92347717285156,
      "learning_rate": 1.3838609666430882e-05,
      "loss": 5.5125,
      "step": 9744
    },
    {
      "epoch": 0.39341945902301173,
      "grad_norm": 88.4334716796875,
      "learning_rate": 1.3837402225779357e-05,
      "loss": 5.3728,
      "step": 9745
    },
    {
      "epoch": 0.39345983044004845,
      "grad_norm": 101.26399993896484,
      "learning_rate": 1.383619471951722e-05,
      "loss": 6.4304,
      "step": 9746
    },
    {
      "epoch": 0.3935002018570852,
      "grad_norm": 108.15672302246094,
      "learning_rate": 1.3834987147665107e-05,
      "loss": 5.438,
      "step": 9747
    },
    {
      "epoch": 0.3935405732741219,
      "grad_norm": 85.50386810302734,
      "learning_rate": 1.3833779510243673e-05,
      "loss": 5.7865,
      "step": 9748
    },
    {
      "epoch": 0.3935809446911587,
      "grad_norm": 95.5966796875,
      "learning_rate": 1.3832571807273559e-05,
      "loss": 5.9559,
      "step": 9749
    },
    {
      "epoch": 0.3936213161081954,
      "grad_norm": 111.4237060546875,
      "learning_rate": 1.3831364038775416e-05,
      "loss": 6.6914,
      "step": 9750
    },
    {
      "epoch": 0.3936616875252321,
      "grad_norm": 77.05138397216797,
      "learning_rate": 1.3830156204769898e-05,
      "loss": 4.9948,
      "step": 9751
    },
    {
      "epoch": 0.3937020589422689,
      "grad_norm": 181.99871826171875,
      "learning_rate": 1.3828948305277652e-05,
      "loss": 4.4747,
      "step": 9752
    },
    {
      "epoch": 0.3937424303593056,
      "grad_norm": 89.59449768066406,
      "learning_rate": 1.382774034031933e-05,
      "loss": 5.7261,
      "step": 9753
    },
    {
      "epoch": 0.39378280177634234,
      "grad_norm": 96.6485366821289,
      "learning_rate": 1.3826532309915584e-05,
      "loss": 3.433,
      "step": 9754
    },
    {
      "epoch": 0.39382317319337906,
      "grad_norm": 110.79937744140625,
      "learning_rate": 1.3825324214087075e-05,
      "loss": 6.078,
      "step": 9755
    },
    {
      "epoch": 0.39386354461041584,
      "grad_norm": 120.89154815673828,
      "learning_rate": 1.3824116052854454e-05,
      "loss": 5.6558,
      "step": 9756
    },
    {
      "epoch": 0.39390391602745256,
      "grad_norm": 100.70356750488281,
      "learning_rate": 1.3822907826238377e-05,
      "loss": 6.5553,
      "step": 9757
    },
    {
      "epoch": 0.3939442874444893,
      "grad_norm": 296.449462890625,
      "learning_rate": 1.3821699534259505e-05,
      "loss": 5.6976,
      "step": 9758
    },
    {
      "epoch": 0.39398465886152606,
      "grad_norm": 71.98796844482422,
      "learning_rate": 1.3820491176938496e-05,
      "loss": 3.1708,
      "step": 9759
    },
    {
      "epoch": 0.3940250302785628,
      "grad_norm": 128.05567932128906,
      "learning_rate": 1.3819282754296008e-05,
      "loss": 4.008,
      "step": 9760
    },
    {
      "epoch": 0.3940654016955995,
      "grad_norm": 85.74088287353516,
      "learning_rate": 1.3818074266352703e-05,
      "loss": 4.4715,
      "step": 9761
    },
    {
      "epoch": 0.3941057731126362,
      "grad_norm": 102.50245666503906,
      "learning_rate": 1.3816865713129246e-05,
      "loss": 6.1252,
      "step": 9762
    },
    {
      "epoch": 0.394146144529673,
      "grad_norm": 96.60061645507812,
      "learning_rate": 1.3815657094646296e-05,
      "loss": 4.9956,
      "step": 9763
    },
    {
      "epoch": 0.3941865159467097,
      "grad_norm": 100.96378326416016,
      "learning_rate": 1.381444841092452e-05,
      "loss": 4.3068,
      "step": 9764
    },
    {
      "epoch": 0.39422688736374645,
      "grad_norm": 86.97381591796875,
      "learning_rate": 1.3813239661984583e-05,
      "loss": 5.2668,
      "step": 9765
    },
    {
      "epoch": 0.3942672587807832,
      "grad_norm": 397.1275329589844,
      "learning_rate": 1.3812030847847156e-05,
      "loss": 4.83,
      "step": 9766
    },
    {
      "epoch": 0.39430763019781995,
      "grad_norm": 81.53783416748047,
      "learning_rate": 1.3810821968532899e-05,
      "loss": 5.8796,
      "step": 9767
    },
    {
      "epoch": 0.39434800161485667,
      "grad_norm": 99.91616821289062,
      "learning_rate": 1.380961302406249e-05,
      "loss": 4.2644,
      "step": 9768
    },
    {
      "epoch": 0.39438837303189345,
      "grad_norm": 89.16180419921875,
      "learning_rate": 1.3808404014456592e-05,
      "loss": 5.0035,
      "step": 9769
    },
    {
      "epoch": 0.39442874444893017,
      "grad_norm": 115.12789916992188,
      "learning_rate": 1.3807194939735878e-05,
      "loss": 5.7705,
      "step": 9770
    },
    {
      "epoch": 0.3944691158659669,
      "grad_norm": 80.16279602050781,
      "learning_rate": 1.380598579992102e-05,
      "loss": 5.7689,
      "step": 9771
    },
    {
      "epoch": 0.3945094872830036,
      "grad_norm": 148.14529418945312,
      "learning_rate": 1.3804776595032693e-05,
      "loss": 5.3672,
      "step": 9772
    },
    {
      "epoch": 0.3945498587000404,
      "grad_norm": 66.30850982666016,
      "learning_rate": 1.3803567325091571e-05,
      "loss": 5.9654,
      "step": 9773
    },
    {
      "epoch": 0.3945902301170771,
      "grad_norm": 154.11729431152344,
      "learning_rate": 1.3802357990118331e-05,
      "loss": 4.8961,
      "step": 9774
    },
    {
      "epoch": 0.39463060153411383,
      "grad_norm": 75.3212661743164,
      "learning_rate": 1.3801148590133647e-05,
      "loss": 6.1717,
      "step": 9775
    },
    {
      "epoch": 0.3946709729511506,
      "grad_norm": 132.48069763183594,
      "learning_rate": 1.37999391251582e-05,
      "loss": 4.0989,
      "step": 9776
    },
    {
      "epoch": 0.39471134436818733,
      "grad_norm": 92.56013488769531,
      "learning_rate": 1.3798729595212666e-05,
      "loss": 5.3499,
      "step": 9777
    },
    {
      "epoch": 0.39475171578522406,
      "grad_norm": 84.7759780883789,
      "learning_rate": 1.3797520000317728e-05,
      "loss": 5.12,
      "step": 9778
    },
    {
      "epoch": 0.3947920872022608,
      "grad_norm": 70.72569274902344,
      "learning_rate": 1.3796310340494063e-05,
      "loss": 6.207,
      "step": 9779
    },
    {
      "epoch": 0.39483245861929755,
      "grad_norm": 96.65389251708984,
      "learning_rate": 1.379510061576236e-05,
      "loss": 5.9227,
      "step": 9780
    },
    {
      "epoch": 0.3948728300363343,
      "grad_norm": 119.5779800415039,
      "learning_rate": 1.3793890826143297e-05,
      "loss": 6.0613,
      "step": 9781
    },
    {
      "epoch": 0.394913201453371,
      "grad_norm": 93.37263488769531,
      "learning_rate": 1.3792680971657562e-05,
      "loss": 4.8406,
      "step": 9782
    },
    {
      "epoch": 0.3949535728704078,
      "grad_norm": 72.90221405029297,
      "learning_rate": 1.3791471052325837e-05,
      "loss": 4.7304,
      "step": 9783
    },
    {
      "epoch": 0.3949939442874445,
      "grad_norm": 112.3136215209961,
      "learning_rate": 1.379026106816881e-05,
      "loss": 6.5016,
      "step": 9784
    },
    {
      "epoch": 0.3950343157044812,
      "grad_norm": 91.43806457519531,
      "learning_rate": 1.3789051019207169e-05,
      "loss": 6.2762,
      "step": 9785
    },
    {
      "epoch": 0.39507468712151794,
      "grad_norm": 111.08851623535156,
      "learning_rate": 1.3787840905461607e-05,
      "loss": 8.0548,
      "step": 9786
    },
    {
      "epoch": 0.3951150585385547,
      "grad_norm": 132.14466857910156,
      "learning_rate": 1.3786630726952808e-05,
      "loss": 5.1395,
      "step": 9787
    },
    {
      "epoch": 0.39515542995559144,
      "grad_norm": 94.37010192871094,
      "learning_rate": 1.3785420483701468e-05,
      "loss": 6.264,
      "step": 9788
    },
    {
      "epoch": 0.39519580137262816,
      "grad_norm": 139.50534057617188,
      "learning_rate": 1.3784210175728276e-05,
      "loss": 5.5772,
      "step": 9789
    },
    {
      "epoch": 0.39523617278966494,
      "grad_norm": 214.39358520507812,
      "learning_rate": 1.3782999803053927e-05,
      "loss": 5.0043,
      "step": 9790
    },
    {
      "epoch": 0.39527654420670166,
      "grad_norm": 93.78736877441406,
      "learning_rate": 1.3781789365699115e-05,
      "loss": 4.3305,
      "step": 9791
    },
    {
      "epoch": 0.3953169156237384,
      "grad_norm": 78.4153060913086,
      "learning_rate": 1.3780578863684537e-05,
      "loss": 5.1303,
      "step": 9792
    },
    {
      "epoch": 0.3953572870407751,
      "grad_norm": 105.27993774414062,
      "learning_rate": 1.3779368297030888e-05,
      "loss": 4.7562,
      "step": 9793
    },
    {
      "epoch": 0.3953976584578119,
      "grad_norm": 84.95843505859375,
      "learning_rate": 1.3778157665758864e-05,
      "loss": 4.3189,
      "step": 9794
    },
    {
      "epoch": 0.3954380298748486,
      "grad_norm": 94.18071746826172,
      "learning_rate": 1.3776946969889168e-05,
      "loss": 5.5635,
      "step": 9795
    },
    {
      "epoch": 0.39547840129188533,
      "grad_norm": 88.70942687988281,
      "learning_rate": 1.3775736209442502e-05,
      "loss": 5.4195,
      "step": 9796
    },
    {
      "epoch": 0.3955187727089221,
      "grad_norm": 71.65169525146484,
      "learning_rate": 1.377452538443956e-05,
      "loss": 6.385,
      "step": 9797
    },
    {
      "epoch": 0.39555914412595883,
      "grad_norm": 134.05902099609375,
      "learning_rate": 1.3773314494901052e-05,
      "loss": 4.8571,
      "step": 9798
    },
    {
      "epoch": 0.39559951554299555,
      "grad_norm": 96.4971694946289,
      "learning_rate": 1.3772103540847674e-05,
      "loss": 4.5542,
      "step": 9799
    },
    {
      "epoch": 0.39563988696003227,
      "grad_norm": 74.544921875,
      "learning_rate": 1.3770892522300136e-05,
      "loss": 5.0168,
      "step": 9800
    },
    {
      "epoch": 0.39568025837706905,
      "grad_norm": 105.07801055908203,
      "learning_rate": 1.3769681439279141e-05,
      "loss": 4.7311,
      "step": 9801
    },
    {
      "epoch": 0.39572062979410577,
      "grad_norm": 71.5572280883789,
      "learning_rate": 1.3768470291805397e-05,
      "loss": 5.6926,
      "step": 9802
    },
    {
      "epoch": 0.3957610012111425,
      "grad_norm": 85.10242462158203,
      "learning_rate": 1.3767259079899609e-05,
      "loss": 6.0965,
      "step": 9803
    },
    {
      "epoch": 0.39580137262817927,
      "grad_norm": 109.1307373046875,
      "learning_rate": 1.3766047803582487e-05,
      "loss": 3.7373,
      "step": 9804
    },
    {
      "epoch": 0.395841744045216,
      "grad_norm": 102.4127197265625,
      "learning_rate": 1.3764836462874741e-05,
      "loss": 4.4903,
      "step": 9805
    },
    {
      "epoch": 0.3958821154622527,
      "grad_norm": 140.29681396484375,
      "learning_rate": 1.376362505779709e-05,
      "loss": 5.2159,
      "step": 9806
    },
    {
      "epoch": 0.39592248687928944,
      "grad_norm": 90.07621765136719,
      "learning_rate": 1.3762413588370233e-05,
      "loss": 4.1795,
      "step": 9807
    },
    {
      "epoch": 0.3959628582963262,
      "grad_norm": 166.8191680908203,
      "learning_rate": 1.3761202054614892e-05,
      "loss": 5.1099,
      "step": 9808
    },
    {
      "epoch": 0.39600322971336294,
      "grad_norm": 108.2179183959961,
      "learning_rate": 1.3759990456551776e-05,
      "loss": 4.2567,
      "step": 9809
    },
    {
      "epoch": 0.39604360113039966,
      "grad_norm": 98.25475311279297,
      "learning_rate": 1.3758778794201608e-05,
      "loss": 5.1845,
      "step": 9810
    },
    {
      "epoch": 0.39608397254743644,
      "grad_norm": 89.83172607421875,
      "learning_rate": 1.3757567067585098e-05,
      "loss": 5.9032,
      "step": 9811
    },
    {
      "epoch": 0.39612434396447316,
      "grad_norm": 69.09123992919922,
      "learning_rate": 1.3756355276722966e-05,
      "loss": 4.8339,
      "step": 9812
    },
    {
      "epoch": 0.3961647153815099,
      "grad_norm": 91.20081329345703,
      "learning_rate": 1.3755143421635927e-05,
      "loss": 5.429,
      "step": 9813
    },
    {
      "epoch": 0.3962050867985466,
      "grad_norm": 116.60623168945312,
      "learning_rate": 1.375393150234471e-05,
      "loss": 5.902,
      "step": 9814
    },
    {
      "epoch": 0.3962454582155834,
      "grad_norm": 163.6284942626953,
      "learning_rate": 1.3752719518870025e-05,
      "loss": 6.8418,
      "step": 9815
    },
    {
      "epoch": 0.3962858296326201,
      "grad_norm": 125.91740417480469,
      "learning_rate": 1.3751507471232604e-05,
      "loss": 4.7637,
      "step": 9816
    },
    {
      "epoch": 0.3963262010496568,
      "grad_norm": 87.87385559082031,
      "learning_rate": 1.3750295359453164e-05,
      "loss": 5.4925,
      "step": 9817
    },
    {
      "epoch": 0.3963665724666936,
      "grad_norm": 87.29440307617188,
      "learning_rate": 1.3749083183552431e-05,
      "loss": 6.1092,
      "step": 9818
    },
    {
      "epoch": 0.3964069438837303,
      "grad_norm": 108.2324447631836,
      "learning_rate": 1.3747870943551129e-05,
      "loss": 4.3571,
      "step": 9819
    },
    {
      "epoch": 0.39644731530076704,
      "grad_norm": 83.6478500366211,
      "learning_rate": 1.3746658639469987e-05,
      "loss": 5.2343,
      "step": 9820
    },
    {
      "epoch": 0.3964876867178038,
      "grad_norm": 97.35012817382812,
      "learning_rate": 1.374544627132973e-05,
      "loss": 5.6619,
      "step": 9821
    },
    {
      "epoch": 0.39652805813484054,
      "grad_norm": 88.3277359008789,
      "learning_rate": 1.3744233839151089e-05,
      "loss": 4.5548,
      "step": 9822
    },
    {
      "epoch": 0.39656842955187727,
      "grad_norm": 161.11346435546875,
      "learning_rate": 1.3743021342954792e-05,
      "loss": 6.465,
      "step": 9823
    },
    {
      "epoch": 0.396608800968914,
      "grad_norm": 114.64875030517578,
      "learning_rate": 1.3741808782761573e-05,
      "loss": 5.1732,
      "step": 9824
    },
    {
      "epoch": 0.39664917238595077,
      "grad_norm": 165.69403076171875,
      "learning_rate": 1.3740596158592158e-05,
      "loss": 6.7296,
      "step": 9825
    },
    {
      "epoch": 0.3966895438029875,
      "grad_norm": 236.1666259765625,
      "learning_rate": 1.3739383470467287e-05,
      "loss": 6.1458,
      "step": 9826
    },
    {
      "epoch": 0.3967299152200242,
      "grad_norm": 97.74983215332031,
      "learning_rate": 1.3738170718407689e-05,
      "loss": 4.2817,
      "step": 9827
    },
    {
      "epoch": 0.396770286637061,
      "grad_norm": 105.85901641845703,
      "learning_rate": 1.3736957902434103e-05,
      "loss": 6.4493,
      "step": 9828
    },
    {
      "epoch": 0.3968106580540977,
      "grad_norm": 111.50251007080078,
      "learning_rate": 1.3735745022567261e-05,
      "loss": 5.1449,
      "step": 9829
    },
    {
      "epoch": 0.39685102947113443,
      "grad_norm": 115.23919677734375,
      "learning_rate": 1.3734532078827904e-05,
      "loss": 6.553,
      "step": 9830
    },
    {
      "epoch": 0.39689140088817115,
      "grad_norm": 85.42231750488281,
      "learning_rate": 1.3733319071236768e-05,
      "loss": 6.4937,
      "step": 9831
    },
    {
      "epoch": 0.39693177230520793,
      "grad_norm": 98.71804809570312,
      "learning_rate": 1.3732105999814593e-05,
      "loss": 5.618,
      "step": 9832
    },
    {
      "epoch": 0.39697214372224465,
      "grad_norm": 89.21080017089844,
      "learning_rate": 1.3730892864582122e-05,
      "loss": 3.4076,
      "step": 9833
    },
    {
      "epoch": 0.3970125151392814,
      "grad_norm": 125.96713256835938,
      "learning_rate": 1.3729679665560097e-05,
      "loss": 5.8897,
      "step": 9834
    },
    {
      "epoch": 0.39705288655631815,
      "grad_norm": 98.28369140625,
      "learning_rate": 1.3728466402769253e-05,
      "loss": 4.9056,
      "step": 9835
    },
    {
      "epoch": 0.3970932579733549,
      "grad_norm": 82.46700286865234,
      "learning_rate": 1.3727253076230348e-05,
      "loss": 5.4918,
      "step": 9836
    },
    {
      "epoch": 0.3971336293903916,
      "grad_norm": 67.04959106445312,
      "learning_rate": 1.3726039685964113e-05,
      "loss": 6.7999,
      "step": 9837
    },
    {
      "epoch": 0.3971740008074283,
      "grad_norm": 92.01446533203125,
      "learning_rate": 1.3724826231991307e-05,
      "loss": 5.8268,
      "step": 9838
    },
    {
      "epoch": 0.3972143722244651,
      "grad_norm": 105.67363739013672,
      "learning_rate": 1.3723612714332664e-05,
      "loss": 3.5962,
      "step": 9839
    },
    {
      "epoch": 0.3972547436415018,
      "grad_norm": 75.32141876220703,
      "learning_rate": 1.3722399133008942e-05,
      "loss": 4.7731,
      "step": 9840
    },
    {
      "epoch": 0.39729511505853854,
      "grad_norm": 86.68830871582031,
      "learning_rate": 1.3721185488040886e-05,
      "loss": 3.9198,
      "step": 9841
    },
    {
      "epoch": 0.3973354864755753,
      "grad_norm": 105.05224609375,
      "learning_rate": 1.3719971779449249e-05,
      "loss": 4.763,
      "step": 9842
    },
    {
      "epoch": 0.39737585789261204,
      "grad_norm": 76.94025421142578,
      "learning_rate": 1.3718758007254778e-05,
      "loss": 4.9621,
      "step": 9843
    },
    {
      "epoch": 0.39741622930964876,
      "grad_norm": 135.40087890625,
      "learning_rate": 1.3717544171478231e-05,
      "loss": 5.0993,
      "step": 9844
    },
    {
      "epoch": 0.3974566007266855,
      "grad_norm": 157.20933532714844,
      "learning_rate": 1.3716330272140361e-05,
      "loss": 5.6087,
      "step": 9845
    },
    {
      "epoch": 0.39749697214372226,
      "grad_norm": 84.56336975097656,
      "learning_rate": 1.3715116309261923e-05,
      "loss": 4.7882,
      "step": 9846
    },
    {
      "epoch": 0.397537343560759,
      "grad_norm": 86.21202087402344,
      "learning_rate": 1.3713902282863668e-05,
      "loss": 5.8792,
      "step": 9847
    },
    {
      "epoch": 0.3975777149777957,
      "grad_norm": 67.79780578613281,
      "learning_rate": 1.3712688192966359e-05,
      "loss": 4.9117,
      "step": 9848
    },
    {
      "epoch": 0.3976180863948325,
      "grad_norm": 93.33841705322266,
      "learning_rate": 1.3711474039590748e-05,
      "loss": 5.0642,
      "step": 9849
    },
    {
      "epoch": 0.3976584578118692,
      "grad_norm": 100.04960632324219,
      "learning_rate": 1.3710259822757604e-05,
      "loss": 4.615,
      "step": 9850
    },
    {
      "epoch": 0.3976988292289059,
      "grad_norm": 95.88956451416016,
      "learning_rate": 1.3709045542487676e-05,
      "loss": 3.6537,
      "step": 9851
    },
    {
      "epoch": 0.39773920064594265,
      "grad_norm": 105.78748321533203,
      "learning_rate": 1.370783119880173e-05,
      "loss": 4.6682,
      "step": 9852
    },
    {
      "epoch": 0.3977795720629794,
      "grad_norm": 185.01968383789062,
      "learning_rate": 1.3706616791720531e-05,
      "loss": 6.1021,
      "step": 9853
    },
    {
      "epoch": 0.39781994348001615,
      "grad_norm": 75.98744201660156,
      "learning_rate": 1.3705402321264843e-05,
      "loss": 3.7588,
      "step": 9854
    },
    {
      "epoch": 0.39786031489705287,
      "grad_norm": 191.88267517089844,
      "learning_rate": 1.3704187787455422e-05,
      "loss": 4.9644,
      "step": 9855
    },
    {
      "epoch": 0.39790068631408965,
      "grad_norm": 132.3114776611328,
      "learning_rate": 1.3702973190313047e-05,
      "loss": 5.063,
      "step": 9856
    },
    {
      "epoch": 0.39794105773112637,
      "grad_norm": 109.93856048583984,
      "learning_rate": 1.3701758529858472e-05,
      "loss": 4.734,
      "step": 9857
    },
    {
      "epoch": 0.3979814291481631,
      "grad_norm": 95.74446868896484,
      "learning_rate": 1.3700543806112474e-05,
      "loss": 5.4161,
      "step": 9858
    },
    {
      "epoch": 0.3980218005651998,
      "grad_norm": 101.5312271118164,
      "learning_rate": 1.3699329019095814e-05,
      "loss": 4.5609,
      "step": 9859
    },
    {
      "epoch": 0.3980621719822366,
      "grad_norm": 88.95726013183594,
      "learning_rate": 1.369811416882927e-05,
      "loss": 6.7014,
      "step": 9860
    },
    {
      "epoch": 0.3981025433992733,
      "grad_norm": 115.60452270507812,
      "learning_rate": 1.369689925533361e-05,
      "loss": 4.7336,
      "step": 9861
    },
    {
      "epoch": 0.39814291481631003,
      "grad_norm": 192.62603759765625,
      "learning_rate": 1.36956842786296e-05,
      "loss": 6.2181,
      "step": 9862
    },
    {
      "epoch": 0.3981832862333468,
      "grad_norm": 147.64285278320312,
      "learning_rate": 1.3694469238738024e-05,
      "loss": 6.6205,
      "step": 9863
    },
    {
      "epoch": 0.39822365765038353,
      "grad_norm": 107.10977935791016,
      "learning_rate": 1.369325413567965e-05,
      "loss": 5.5712,
      "step": 9864
    },
    {
      "epoch": 0.39826402906742026,
      "grad_norm": 130.22918701171875,
      "learning_rate": 1.3692038969475254e-05,
      "loss": 5.1215,
      "step": 9865
    },
    {
      "epoch": 0.398304400484457,
      "grad_norm": 131.54953002929688,
      "learning_rate": 1.3690823740145613e-05,
      "loss": 4.941,
      "step": 9866
    },
    {
      "epoch": 0.39834477190149375,
      "grad_norm": 138.3892364501953,
      "learning_rate": 1.3689608447711504e-05,
      "loss": 4.8919,
      "step": 9867
    },
    {
      "epoch": 0.3983851433185305,
      "grad_norm": 86.6033706665039,
      "learning_rate": 1.368839309219371e-05,
      "loss": 4.7716,
      "step": 9868
    },
    {
      "epoch": 0.3984255147355672,
      "grad_norm": 129.59010314941406,
      "learning_rate": 1.3687177673613006e-05,
      "loss": 5.6967,
      "step": 9869
    },
    {
      "epoch": 0.398465886152604,
      "grad_norm": 129.67446899414062,
      "learning_rate": 1.3685962191990174e-05,
      "loss": 5.8356,
      "step": 9870
    },
    {
      "epoch": 0.3985062575696407,
      "grad_norm": 126.60520935058594,
      "learning_rate": 1.3684746647345993e-05,
      "loss": 5.0529,
      "step": 9871
    },
    {
      "epoch": 0.3985466289866774,
      "grad_norm": 94.46762084960938,
      "learning_rate": 1.368353103970125e-05,
      "loss": 2.8928,
      "step": 9872
    },
    {
      "epoch": 0.3985870004037142,
      "grad_norm": 146.25213623046875,
      "learning_rate": 1.3682315369076726e-05,
      "loss": 4.7932,
      "step": 9873
    },
    {
      "epoch": 0.3986273718207509,
      "grad_norm": 111.81463623046875,
      "learning_rate": 1.3681099635493212e-05,
      "loss": 4.6397,
      "step": 9874
    },
    {
      "epoch": 0.39866774323778764,
      "grad_norm": 87.74475860595703,
      "learning_rate": 1.3679883838971488e-05,
      "loss": 5.753,
      "step": 9875
    },
    {
      "epoch": 0.39870811465482436,
      "grad_norm": 193.7996063232422,
      "learning_rate": 1.3678667979532346e-05,
      "loss": 5.0315,
      "step": 9876
    },
    {
      "epoch": 0.39874848607186114,
      "grad_norm": 108.96322631835938,
      "learning_rate": 1.367745205719657e-05,
      "loss": 5.3319,
      "step": 9877
    },
    {
      "epoch": 0.39878885748889786,
      "grad_norm": 98.17216491699219,
      "learning_rate": 1.3676236071984949e-05,
      "loss": 4.3288,
      "step": 9878
    },
    {
      "epoch": 0.3988292289059346,
      "grad_norm": 157.4913787841797,
      "learning_rate": 1.367502002391828e-05,
      "loss": 5.0107,
      "step": 9879
    },
    {
      "epoch": 0.39886960032297136,
      "grad_norm": 74.54830169677734,
      "learning_rate": 1.3673803913017347e-05,
      "loss": 4.9746,
      "step": 9880
    },
    {
      "epoch": 0.3989099717400081,
      "grad_norm": 94.2559814453125,
      "learning_rate": 1.367258773930295e-05,
      "loss": 5.9686,
      "step": 9881
    },
    {
      "epoch": 0.3989503431570448,
      "grad_norm": 147.71255493164062,
      "learning_rate": 1.3671371502795876e-05,
      "loss": 6.4869,
      "step": 9882
    },
    {
      "epoch": 0.39899071457408153,
      "grad_norm": 102.50748443603516,
      "learning_rate": 1.367015520351692e-05,
      "loss": 4.2586,
      "step": 9883
    },
    {
      "epoch": 0.3990310859911183,
      "grad_norm": 113.55075073242188,
      "learning_rate": 1.3668938841486883e-05,
      "loss": 4.7444,
      "step": 9884
    },
    {
      "epoch": 0.39907145740815503,
      "grad_norm": 87.32434844970703,
      "learning_rate": 1.366772241672656e-05,
      "loss": 5.4545,
      "step": 9885
    },
    {
      "epoch": 0.39911182882519175,
      "grad_norm": 144.5163116455078,
      "learning_rate": 1.3666505929256749e-05,
      "loss": 3.4968,
      "step": 9886
    },
    {
      "epoch": 0.3991522002422285,
      "grad_norm": 83.48458099365234,
      "learning_rate": 1.3665289379098248e-05,
      "loss": 6.18,
      "step": 9887
    },
    {
      "epoch": 0.39919257165926525,
      "grad_norm": 93.25835418701172,
      "learning_rate": 1.3664072766271856e-05,
      "loss": 5.6045,
      "step": 9888
    },
    {
      "epoch": 0.39923294307630197,
      "grad_norm": 84.04983520507812,
      "learning_rate": 1.366285609079838e-05,
      "loss": 6.6224,
      "step": 9889
    },
    {
      "epoch": 0.3992733144933387,
      "grad_norm": 83.4250259399414,
      "learning_rate": 1.3661639352698615e-05,
      "loss": 5.0704,
      "step": 9890
    },
    {
      "epoch": 0.39931368591037547,
      "grad_norm": 92.85383605957031,
      "learning_rate": 1.3660422551993368e-05,
      "loss": 5.3142,
      "step": 9891
    },
    {
      "epoch": 0.3993540573274122,
      "grad_norm": 67.56437683105469,
      "learning_rate": 1.3659205688703443e-05,
      "loss": 5.6651,
      "step": 9892
    },
    {
      "epoch": 0.3993944287444489,
      "grad_norm": 87.7671127319336,
      "learning_rate": 1.3657988762849645e-05,
      "loss": 5.1085,
      "step": 9893
    },
    {
      "epoch": 0.3994348001614857,
      "grad_norm": 68.30902099609375,
      "learning_rate": 1.365677177445278e-05,
      "loss": 4.052,
      "step": 9894
    },
    {
      "epoch": 0.3994751715785224,
      "grad_norm": 107.5217514038086,
      "learning_rate": 1.3655554723533659e-05,
      "loss": 5.4912,
      "step": 9895
    },
    {
      "epoch": 0.39951554299555914,
      "grad_norm": 189.95101928710938,
      "learning_rate": 1.3654337610113089e-05,
      "loss": 3.5303,
      "step": 9896
    },
    {
      "epoch": 0.39955591441259586,
      "grad_norm": 74.80391693115234,
      "learning_rate": 1.365312043421188e-05,
      "loss": 6.2411,
      "step": 9897
    },
    {
      "epoch": 0.39959628582963264,
      "grad_norm": 86.2298355102539,
      "learning_rate": 1.365190319585084e-05,
      "loss": 5.6256,
      "step": 9898
    },
    {
      "epoch": 0.39963665724666936,
      "grad_norm": 126.14086151123047,
      "learning_rate": 1.3650685895050786e-05,
      "loss": 5.8856,
      "step": 9899
    },
    {
      "epoch": 0.3996770286637061,
      "grad_norm": 76.32518768310547,
      "learning_rate": 1.3649468531832524e-05,
      "loss": 4.9944,
      "step": 9900
    },
    {
      "epoch": 0.39971740008074286,
      "grad_norm": 94.18658447265625,
      "learning_rate": 1.3648251106216877e-05,
      "loss": 3.2245,
      "step": 9901
    },
    {
      "epoch": 0.3997577714977796,
      "grad_norm": 95.83458709716797,
      "learning_rate": 1.364703361822465e-05,
      "loss": 4.9072,
      "step": 9902
    },
    {
      "epoch": 0.3997981429148163,
      "grad_norm": 108.65992736816406,
      "learning_rate": 1.3645816067876668e-05,
      "loss": 7.116,
      "step": 9903
    },
    {
      "epoch": 0.399838514331853,
      "grad_norm": 93.99835205078125,
      "learning_rate": 1.3644598455193743e-05,
      "loss": 3.4025,
      "step": 9904
    },
    {
      "epoch": 0.3998788857488898,
      "grad_norm": 104.2765121459961,
      "learning_rate": 1.3643380780196697e-05,
      "loss": 4.8938,
      "step": 9905
    },
    {
      "epoch": 0.3999192571659265,
      "grad_norm": 124.35428619384766,
      "learning_rate": 1.3642163042906347e-05,
      "loss": 5.5873,
      "step": 9906
    },
    {
      "epoch": 0.39995962858296324,
      "grad_norm": 123.19144439697266,
      "learning_rate": 1.3640945243343516e-05,
      "loss": 4.5664,
      "step": 9907
    },
    {
      "epoch": 0.4,
      "grad_norm": 80.45681762695312,
      "learning_rate": 1.363972738152902e-05,
      "loss": 5.7671,
      "step": 9908
    },
    {
      "epoch": 0.40004037141703674,
      "grad_norm": 101.0565185546875,
      "learning_rate": 1.3638509457483689e-05,
      "loss": 5.1734,
      "step": 9909
    },
    {
      "epoch": 0.40008074283407347,
      "grad_norm": 99.42650604248047,
      "learning_rate": 1.3637291471228338e-05,
      "loss": 4.1309,
      "step": 9910
    },
    {
      "epoch": 0.4001211142511102,
      "grad_norm": 178.565185546875,
      "learning_rate": 1.3636073422783798e-05,
      "loss": 5.0648,
      "step": 9911
    },
    {
      "epoch": 0.40016148566814697,
      "grad_norm": 111.3807373046875,
      "learning_rate": 1.3634855312170897e-05,
      "loss": 6.2251,
      "step": 9912
    },
    {
      "epoch": 0.4002018570851837,
      "grad_norm": 110.2804946899414,
      "learning_rate": 1.3633637139410453e-05,
      "loss": 5.1022,
      "step": 9913
    },
    {
      "epoch": 0.4002422285022204,
      "grad_norm": 119.03135681152344,
      "learning_rate": 1.3632418904523302e-05,
      "loss": 7.1611,
      "step": 9914
    },
    {
      "epoch": 0.4002825999192572,
      "grad_norm": 107.5084228515625,
      "learning_rate": 1.363120060753027e-05,
      "loss": 5.5014,
      "step": 9915
    },
    {
      "epoch": 0.4003229713362939,
      "grad_norm": 115.6515121459961,
      "learning_rate": 1.3629982248452184e-05,
      "loss": 5.4329,
      "step": 9916
    },
    {
      "epoch": 0.40036334275333063,
      "grad_norm": 82.89141845703125,
      "learning_rate": 1.3628763827309883e-05,
      "loss": 4.6664,
      "step": 9917
    },
    {
      "epoch": 0.40040371417036735,
      "grad_norm": 91.70899200439453,
      "learning_rate": 1.3627545344124192e-05,
      "loss": 5.5888,
      "step": 9918
    },
    {
      "epoch": 0.40044408558740413,
      "grad_norm": 127.21209716796875,
      "learning_rate": 1.3626326798915948e-05,
      "loss": 6.2026,
      "step": 9919
    },
    {
      "epoch": 0.40048445700444085,
      "grad_norm": 110.88040161132812,
      "learning_rate": 1.362510819170598e-05,
      "loss": 3.497,
      "step": 9920
    },
    {
      "epoch": 0.4005248284214776,
      "grad_norm": 84.99169158935547,
      "learning_rate": 1.362388952251513e-05,
      "loss": 4.0918,
      "step": 9921
    },
    {
      "epoch": 0.40056519983851435,
      "grad_norm": 89.58877563476562,
      "learning_rate": 1.362267079136423e-05,
      "loss": 5.4499,
      "step": 9922
    },
    {
      "epoch": 0.4006055712555511,
      "grad_norm": 116.62408447265625,
      "learning_rate": 1.3621451998274122e-05,
      "loss": 4.5571,
      "step": 9923
    },
    {
      "epoch": 0.4006459426725878,
      "grad_norm": 117.68245697021484,
      "learning_rate": 1.362023314326564e-05,
      "loss": 4.974,
      "step": 9924
    },
    {
      "epoch": 0.4006863140896246,
      "grad_norm": 119.50753021240234,
      "learning_rate": 1.3619014226359626e-05,
      "loss": 4.84,
      "step": 9925
    },
    {
      "epoch": 0.4007266855066613,
      "grad_norm": 99.94263458251953,
      "learning_rate": 1.3617795247576917e-05,
      "loss": 6.9844,
      "step": 9926
    },
    {
      "epoch": 0.400767056923698,
      "grad_norm": 129.6034698486328,
      "learning_rate": 1.361657620693836e-05,
      "loss": 5.1871,
      "step": 9927
    },
    {
      "epoch": 0.40080742834073474,
      "grad_norm": 123.02598571777344,
      "learning_rate": 1.3615357104464794e-05,
      "loss": 4.7073,
      "step": 9928
    },
    {
      "epoch": 0.4008477997577715,
      "grad_norm": 81.19197845458984,
      "learning_rate": 1.3614137940177066e-05,
      "loss": 5.0953,
      "step": 9929
    },
    {
      "epoch": 0.40088817117480824,
      "grad_norm": 123.8517074584961,
      "learning_rate": 1.3612918714096016e-05,
      "loss": 6.3968,
      "step": 9930
    },
    {
      "epoch": 0.40092854259184496,
      "grad_norm": 114.49932098388672,
      "learning_rate": 1.3611699426242493e-05,
      "loss": 5.8934,
      "step": 9931
    },
    {
      "epoch": 0.40096891400888174,
      "grad_norm": 142.5537109375,
      "learning_rate": 1.3610480076637342e-05,
      "loss": 5.0105,
      "step": 9932
    },
    {
      "epoch": 0.40100928542591846,
      "grad_norm": 86.63528442382812,
      "learning_rate": 1.3609260665301418e-05,
      "loss": 4.7375,
      "step": 9933
    },
    {
      "epoch": 0.4010496568429552,
      "grad_norm": 78.27537536621094,
      "learning_rate": 1.360804119225556e-05,
      "loss": 5.9175,
      "step": 9934
    },
    {
      "epoch": 0.4010900282599919,
      "grad_norm": 91.16817474365234,
      "learning_rate": 1.3606821657520625e-05,
      "loss": 5.1223,
      "step": 9935
    },
    {
      "epoch": 0.4011303996770287,
      "grad_norm": 101.05674743652344,
      "learning_rate": 1.3605602061117465e-05,
      "loss": 7.5294,
      "step": 9936
    },
    {
      "epoch": 0.4011707710940654,
      "grad_norm": 255.07200622558594,
      "learning_rate": 1.3604382403066927e-05,
      "loss": 5.4343,
      "step": 9937
    },
    {
      "epoch": 0.4012111425111021,
      "grad_norm": 72.66639709472656,
      "learning_rate": 1.3603162683389866e-05,
      "loss": 4.065,
      "step": 9938
    },
    {
      "epoch": 0.4012515139281389,
      "grad_norm": 144.0323486328125,
      "learning_rate": 1.3601942902107137e-05,
      "loss": 5.2148,
      "step": 9939
    },
    {
      "epoch": 0.4012918853451756,
      "grad_norm": 102.42202758789062,
      "learning_rate": 1.3600723059239595e-05,
      "loss": 5.324,
      "step": 9940
    },
    {
      "epoch": 0.40133225676221235,
      "grad_norm": 67.60549926757812,
      "learning_rate": 1.3599503154808098e-05,
      "loss": 5.2261,
      "step": 9941
    },
    {
      "epoch": 0.40137262817924907,
      "grad_norm": 136.10989379882812,
      "learning_rate": 1.3598283188833504e-05,
      "loss": 4.1123,
      "step": 9942
    },
    {
      "epoch": 0.40141299959628585,
      "grad_norm": 80.3004379272461,
      "learning_rate": 1.3597063161336667e-05,
      "loss": 4.8396,
      "step": 9943
    },
    {
      "epoch": 0.40145337101332257,
      "grad_norm": 102.13011932373047,
      "learning_rate": 1.359584307233845e-05,
      "loss": 4.2229,
      "step": 9944
    },
    {
      "epoch": 0.4014937424303593,
      "grad_norm": 102.93238830566406,
      "learning_rate": 1.3594622921859716e-05,
      "loss": 6.2326,
      "step": 9945
    },
    {
      "epoch": 0.40153411384739607,
      "grad_norm": 114.82176208496094,
      "learning_rate": 1.3593402709921323e-05,
      "loss": 3.8953,
      "step": 9946
    },
    {
      "epoch": 0.4015744852644328,
      "grad_norm": 110.29104614257812,
      "learning_rate": 1.3592182436544138e-05,
      "loss": 6.3352,
      "step": 9947
    },
    {
      "epoch": 0.4016148566814695,
      "grad_norm": 198.96923828125,
      "learning_rate": 1.3590962101749016e-05,
      "loss": 5.4066,
      "step": 9948
    },
    {
      "epoch": 0.40165522809850623,
      "grad_norm": 82.6927490234375,
      "learning_rate": 1.358974170555683e-05,
      "loss": 5.103,
      "step": 9949
    },
    {
      "epoch": 0.401695599515543,
      "grad_norm": 110.23848724365234,
      "learning_rate": 1.3588521247988444e-05,
      "loss": 5.0895,
      "step": 9950
    },
    {
      "epoch": 0.40173597093257973,
      "grad_norm": 110.00828552246094,
      "learning_rate": 1.3587300729064723e-05,
      "loss": 6.4322,
      "step": 9951
    },
    {
      "epoch": 0.40177634234961646,
      "grad_norm": 105.23509216308594,
      "learning_rate": 1.3586080148806535e-05,
      "loss": 4.3372,
      "step": 9952
    },
    {
      "epoch": 0.40181671376665323,
      "grad_norm": 96.4324722290039,
      "learning_rate": 1.3584859507234756e-05,
      "loss": 5.0206,
      "step": 9953
    },
    {
      "epoch": 0.40185708518368995,
      "grad_norm": 95.32572937011719,
      "learning_rate": 1.3583638804370244e-05,
      "loss": 4.8741,
      "step": 9954
    },
    {
      "epoch": 0.4018974566007267,
      "grad_norm": 193.73179626464844,
      "learning_rate": 1.3582418040233885e-05,
      "loss": 4.9613,
      "step": 9955
    },
    {
      "epoch": 0.4019378280177634,
      "grad_norm": 67.6092529296875,
      "learning_rate": 1.3581197214846537e-05,
      "loss": 6.4877,
      "step": 9956
    },
    {
      "epoch": 0.4019781994348002,
      "grad_norm": 98.83833312988281,
      "learning_rate": 1.357997632822908e-05,
      "loss": 5.0598,
      "step": 9957
    },
    {
      "epoch": 0.4020185708518369,
      "grad_norm": 121.04734802246094,
      "learning_rate": 1.3578755380402388e-05,
      "loss": 4.8894,
      "step": 9958
    },
    {
      "epoch": 0.4020589422688736,
      "grad_norm": 99.77359771728516,
      "learning_rate": 1.357753437138734e-05,
      "loss": 4.0399,
      "step": 9959
    },
    {
      "epoch": 0.4020993136859104,
      "grad_norm": 176.112060546875,
      "learning_rate": 1.3576313301204802e-05,
      "loss": 7.1719,
      "step": 9960
    },
    {
      "epoch": 0.4021396851029471,
      "grad_norm": 104.52452850341797,
      "learning_rate": 1.357509216987566e-05,
      "loss": 6.4743,
      "step": 9961
    },
    {
      "epoch": 0.40218005651998384,
      "grad_norm": 124.9189224243164,
      "learning_rate": 1.3573870977420791e-05,
      "loss": 2.9577,
      "step": 9962
    },
    {
      "epoch": 0.40222042793702056,
      "grad_norm": 89.43659973144531,
      "learning_rate": 1.3572649723861075e-05,
      "loss": 4.9109,
      "step": 9963
    },
    {
      "epoch": 0.40226079935405734,
      "grad_norm": 86.4192886352539,
      "learning_rate": 1.3571428409217389e-05,
      "loss": 4.8663,
      "step": 9964
    },
    {
      "epoch": 0.40230117077109406,
      "grad_norm": 182.28353881835938,
      "learning_rate": 1.3570207033510619e-05,
      "loss": 5.4886,
      "step": 9965
    },
    {
      "epoch": 0.4023415421881308,
      "grad_norm": 144.5015106201172,
      "learning_rate": 1.3568985596761646e-05,
      "loss": 4.5542,
      "step": 9966
    },
    {
      "epoch": 0.40238191360516756,
      "grad_norm": 79.13272857666016,
      "learning_rate": 1.3567764098991352e-05,
      "loss": 6.88,
      "step": 9967
    },
    {
      "epoch": 0.4024222850222043,
      "grad_norm": 63.32184600830078,
      "learning_rate": 1.3566542540220623e-05,
      "loss": 5.8472,
      "step": 9968
    },
    {
      "epoch": 0.402462656439241,
      "grad_norm": 279.29876708984375,
      "learning_rate": 1.3565320920470348e-05,
      "loss": 6.5566,
      "step": 9969
    },
    {
      "epoch": 0.40250302785627773,
      "grad_norm": 104.03968811035156,
      "learning_rate": 1.3564099239761409e-05,
      "loss": 3.4877,
      "step": 9970
    },
    {
      "epoch": 0.4025433992733145,
      "grad_norm": 205.55271911621094,
      "learning_rate": 1.3562877498114695e-05,
      "loss": 5.2348,
      "step": 9971
    },
    {
      "epoch": 0.40258377069035123,
      "grad_norm": 127.35460662841797,
      "learning_rate": 1.3561655695551094e-05,
      "loss": 5.1197,
      "step": 9972
    },
    {
      "epoch": 0.40262414210738795,
      "grad_norm": 90.04676818847656,
      "learning_rate": 1.3560433832091501e-05,
      "loss": 5.6958,
      "step": 9973
    },
    {
      "epoch": 0.4026645135244247,
      "grad_norm": 91.9584732055664,
      "learning_rate": 1.3559211907756803e-05,
      "loss": 5.0691,
      "step": 9974
    },
    {
      "epoch": 0.40270488494146145,
      "grad_norm": 93.05632019042969,
      "learning_rate": 1.3557989922567894e-05,
      "loss": 4.4874,
      "step": 9975
    },
    {
      "epoch": 0.40274525635849817,
      "grad_norm": 88.93419647216797,
      "learning_rate": 1.3556767876545662e-05,
      "loss": 3.0682,
      "step": 9976
    },
    {
      "epoch": 0.40278562777553495,
      "grad_norm": 180.31480407714844,
      "learning_rate": 1.355554576971101e-05,
      "loss": 5.9507,
      "step": 9977
    },
    {
      "epoch": 0.40282599919257167,
      "grad_norm": 101.93328094482422,
      "learning_rate": 1.3554323602084822e-05,
      "loss": 5.6481,
      "step": 9978
    },
    {
      "epoch": 0.4028663706096084,
      "grad_norm": 102.15277862548828,
      "learning_rate": 1.3553101373688007e-05,
      "loss": 4.2009,
      "step": 9979
    },
    {
      "epoch": 0.4029067420266451,
      "grad_norm": 119.141845703125,
      "learning_rate": 1.3551879084541451e-05,
      "loss": 4.7215,
      "step": 9980
    },
    {
      "epoch": 0.4029471134436819,
      "grad_norm": 110.01436614990234,
      "learning_rate": 1.3550656734666056e-05,
      "loss": 6.1845,
      "step": 9981
    },
    {
      "epoch": 0.4029874848607186,
      "grad_norm": 83.09420013427734,
      "learning_rate": 1.3549434324082724e-05,
      "loss": 5.4809,
      "step": 9982
    },
    {
      "epoch": 0.40302785627775534,
      "grad_norm": 92.83761596679688,
      "learning_rate": 1.3548211852812353e-05,
      "loss": 5.486,
      "step": 9983
    },
    {
      "epoch": 0.4030682276947921,
      "grad_norm": 100.5505142211914,
      "learning_rate": 1.3546989320875848e-05,
      "loss": 5.7317,
      "step": 9984
    },
    {
      "epoch": 0.40310859911182884,
      "grad_norm": 90.11148071289062,
      "learning_rate": 1.3545766728294107e-05,
      "loss": 2.4392,
      "step": 9985
    },
    {
      "epoch": 0.40314897052886556,
      "grad_norm": 88.02227783203125,
      "learning_rate": 1.3544544075088033e-05,
      "loss": 5.5911,
      "step": 9986
    },
    {
      "epoch": 0.4031893419459023,
      "grad_norm": 137.91412353515625,
      "learning_rate": 1.3543321361278534e-05,
      "loss": 4.4414,
      "step": 9987
    },
    {
      "epoch": 0.40322971336293906,
      "grad_norm": 122.88093566894531,
      "learning_rate": 1.3542098586886513e-05,
      "loss": 5.8636,
      "step": 9988
    },
    {
      "epoch": 0.4032700847799758,
      "grad_norm": 85.15090942382812,
      "learning_rate": 1.354087575193288e-05,
      "loss": 5.4716,
      "step": 9989
    },
    {
      "epoch": 0.4033104561970125,
      "grad_norm": 117.61284637451172,
      "learning_rate": 1.3539652856438537e-05,
      "loss": 5.9066,
      "step": 9990
    },
    {
      "epoch": 0.4033508276140493,
      "grad_norm": 91.57879638671875,
      "learning_rate": 1.3538429900424397e-05,
      "loss": 4.9552,
      "step": 9991
    },
    {
      "epoch": 0.403391199031086,
      "grad_norm": 91.59081268310547,
      "learning_rate": 1.3537206883911367e-05,
      "loss": 5.7223,
      "step": 9992
    },
    {
      "epoch": 0.4034315704481227,
      "grad_norm": 134.8419189453125,
      "learning_rate": 1.3535983806920363e-05,
      "loss": 5.5088,
      "step": 9993
    },
    {
      "epoch": 0.40347194186515944,
      "grad_norm": 76.02286529541016,
      "learning_rate": 1.3534760669472289e-05,
      "loss": 4.5144,
      "step": 9994
    },
    {
      "epoch": 0.4035123132821962,
      "grad_norm": 72.09246826171875,
      "learning_rate": 1.3533537471588068e-05,
      "loss": 5.6467,
      "step": 9995
    },
    {
      "epoch": 0.40355268469923294,
      "grad_norm": 100.4280014038086,
      "learning_rate": 1.35323142132886e-05,
      "loss": 3.5376,
      "step": 9996
    },
    {
      "epoch": 0.40359305611626967,
      "grad_norm": 68.26614379882812,
      "learning_rate": 1.3531090894594812e-05,
      "loss": 3.9576,
      "step": 9997
    },
    {
      "epoch": 0.40363342753330644,
      "grad_norm": 104.07910919189453,
      "learning_rate": 1.3529867515527615e-05,
      "loss": 5.4762,
      "step": 9998
    },
    {
      "epoch": 0.40367379895034317,
      "grad_norm": 78.60118865966797,
      "learning_rate": 1.3528644076107926e-05,
      "loss": 4.4539,
      "step": 9999
    },
    {
      "epoch": 0.4037141703673799,
      "grad_norm": 72.59718322753906,
      "learning_rate": 1.3527420576356662e-05,
      "loss": 6.0743,
      "step": 10000
    },
    {
      "epoch": 0.4037545417844166,
      "grad_norm": 104.75896453857422,
      "learning_rate": 1.3526197016294745e-05,
      "loss": 4.4795,
      "step": 10001
    },
    {
      "epoch": 0.4037949132014534,
      "grad_norm": 90.55802917480469,
      "learning_rate": 1.3524973395943091e-05,
      "loss": 4.6666,
      "step": 10002
    },
    {
      "epoch": 0.4038352846184901,
      "grad_norm": 98.5560073852539,
      "learning_rate": 1.3523749715322625e-05,
      "loss": 4.6252,
      "step": 10003
    },
    {
      "epoch": 0.40387565603552683,
      "grad_norm": 92.5905990600586,
      "learning_rate": 1.3522525974454266e-05,
      "loss": 5.2039,
      "step": 10004
    },
    {
      "epoch": 0.4039160274525636,
      "grad_norm": 104.65901184082031,
      "learning_rate": 1.3521302173358941e-05,
      "loss": 5.0798,
      "step": 10005
    },
    {
      "epoch": 0.40395639886960033,
      "grad_norm": 119.90254211425781,
      "learning_rate": 1.3520078312057572e-05,
      "loss": 5.3623,
      "step": 10006
    },
    {
      "epoch": 0.40399677028663705,
      "grad_norm": 82.96353912353516,
      "learning_rate": 1.351885439057108e-05,
      "loss": 4.8753,
      "step": 10007
    },
    {
      "epoch": 0.4040371417036738,
      "grad_norm": 79.16500854492188,
      "learning_rate": 1.35176304089204e-05,
      "loss": 4.4483,
      "step": 10008
    },
    {
      "epoch": 0.40407751312071055,
      "grad_norm": 96.217041015625,
      "learning_rate": 1.351640636712645e-05,
      "loss": 5.8008,
      "step": 10009
    },
    {
      "epoch": 0.4041178845377473,
      "grad_norm": 81.70865631103516,
      "learning_rate": 1.3515182265210164e-05,
      "loss": 5.9261,
      "step": 10010
    },
    {
      "epoch": 0.404158255954784,
      "grad_norm": 160.00379943847656,
      "learning_rate": 1.351395810319247e-05,
      "loss": 6.8167,
      "step": 10011
    },
    {
      "epoch": 0.4041986273718208,
      "grad_norm": 90.25914001464844,
      "learning_rate": 1.3512733881094296e-05,
      "loss": 5.0782,
      "step": 10012
    },
    {
      "epoch": 0.4042389987888575,
      "grad_norm": 346.70361328125,
      "learning_rate": 1.3511509598936581e-05,
      "loss": 4.7798,
      "step": 10013
    },
    {
      "epoch": 0.4042793702058942,
      "grad_norm": 136.9358673095703,
      "learning_rate": 1.3510285256740246e-05,
      "loss": 3.4565,
      "step": 10014
    },
    {
      "epoch": 0.40431974162293094,
      "grad_norm": 100.13786315917969,
      "learning_rate": 1.3509060854526234e-05,
      "loss": 4.9682,
      "step": 10015
    },
    {
      "epoch": 0.4043601130399677,
      "grad_norm": 144.26275634765625,
      "learning_rate": 1.3507836392315473e-05,
      "loss": 5.2024,
      "step": 10016
    },
    {
      "epoch": 0.40440048445700444,
      "grad_norm": 93.33014678955078,
      "learning_rate": 1.3506611870128904e-05,
      "loss": 4.7484,
      "step": 10017
    },
    {
      "epoch": 0.40444085587404116,
      "grad_norm": 167.52212524414062,
      "learning_rate": 1.3505387287987457e-05,
      "loss": 6.7344,
      "step": 10018
    },
    {
      "epoch": 0.40448122729107794,
      "grad_norm": 260.50885009765625,
      "learning_rate": 1.3504162645912073e-05,
      "loss": 5.5607,
      "step": 10019
    },
    {
      "epoch": 0.40452159870811466,
      "grad_norm": 640.3532104492188,
      "learning_rate": 1.3502937943923693e-05,
      "loss": 5.05,
      "step": 10020
    },
    {
      "epoch": 0.4045619701251514,
      "grad_norm": 109.2449722290039,
      "learning_rate": 1.3501713182043254e-05,
      "loss": 5.5725,
      "step": 10021
    },
    {
      "epoch": 0.4046023415421881,
      "grad_norm": 86.66970825195312,
      "learning_rate": 1.3500488360291698e-05,
      "loss": 3.9461,
      "step": 10022
    },
    {
      "epoch": 0.4046427129592249,
      "grad_norm": 156.80555725097656,
      "learning_rate": 1.3499263478689964e-05,
      "loss": 6.2212,
      "step": 10023
    },
    {
      "epoch": 0.4046830843762616,
      "grad_norm": 123.40353393554688,
      "learning_rate": 1.3498038537258993e-05,
      "loss": 4.4915,
      "step": 10024
    },
    {
      "epoch": 0.4047234557932983,
      "grad_norm": 92.90618133544922,
      "learning_rate": 1.3496813536019735e-05,
      "loss": 4.468,
      "step": 10025
    },
    {
      "epoch": 0.4047638272103351,
      "grad_norm": 77.51480865478516,
      "learning_rate": 1.349558847499313e-05,
      "loss": 4.8887,
      "step": 10026
    },
    {
      "epoch": 0.4048041986273718,
      "grad_norm": 86.79052734375,
      "learning_rate": 1.3494363354200127e-05,
      "loss": 5.6922,
      "step": 10027
    },
    {
      "epoch": 0.40484457004440855,
      "grad_norm": 126.95555114746094,
      "learning_rate": 1.3493138173661666e-05,
      "loss": 4.8105,
      "step": 10028
    },
    {
      "epoch": 0.40488494146144527,
      "grad_norm": 84.32832336425781,
      "learning_rate": 1.3491912933398702e-05,
      "loss": 4.9676,
      "step": 10029
    },
    {
      "epoch": 0.40492531287848205,
      "grad_norm": 143.48516845703125,
      "learning_rate": 1.3490687633432177e-05,
      "loss": 3.231,
      "step": 10030
    },
    {
      "epoch": 0.40496568429551877,
      "grad_norm": 86.31990814208984,
      "learning_rate": 1.3489462273783052e-05,
      "loss": 5.5798,
      "step": 10031
    },
    {
      "epoch": 0.4050060557125555,
      "grad_norm": 119.41007995605469,
      "learning_rate": 1.3488236854472266e-05,
      "loss": 5.6776,
      "step": 10032
    },
    {
      "epoch": 0.40504642712959227,
      "grad_norm": 171.85250854492188,
      "learning_rate": 1.3487011375520778e-05,
      "loss": 3.3227,
      "step": 10033
    },
    {
      "epoch": 0.405086798546629,
      "grad_norm": 97.6163101196289,
      "learning_rate": 1.3485785836949536e-05,
      "loss": 5.0636,
      "step": 10034
    },
    {
      "epoch": 0.4051271699636657,
      "grad_norm": 94.8985366821289,
      "learning_rate": 1.3484560238779498e-05,
      "loss": 4.3437,
      "step": 10035
    },
    {
      "epoch": 0.4051675413807025,
      "grad_norm": 219.05982971191406,
      "learning_rate": 1.3483334581031615e-05,
      "loss": 4.7306,
      "step": 10036
    },
    {
      "epoch": 0.4052079127977392,
      "grad_norm": 82.55017852783203,
      "learning_rate": 1.3482108863726846e-05,
      "loss": 5.5215,
      "step": 10037
    },
    {
      "epoch": 0.40524828421477593,
      "grad_norm": 166.3076934814453,
      "learning_rate": 1.3480883086886145e-05,
      "loss": 4.6971,
      "step": 10038
    },
    {
      "epoch": 0.40528865563181266,
      "grad_norm": 149.81265258789062,
      "learning_rate": 1.3479657250530473e-05,
      "loss": 4.5883,
      "step": 10039
    },
    {
      "epoch": 0.40532902704884943,
      "grad_norm": 85.87447357177734,
      "learning_rate": 1.3478431354680787e-05,
      "loss": 6.5138,
      "step": 10040
    },
    {
      "epoch": 0.40536939846588615,
      "grad_norm": 92.15064239501953,
      "learning_rate": 1.347720539935805e-05,
      "loss": 6.124,
      "step": 10041
    },
    {
      "epoch": 0.4054097698829229,
      "grad_norm": 95.77992248535156,
      "learning_rate": 1.3475979384583218e-05,
      "loss": 4.7637,
      "step": 10042
    },
    {
      "epoch": 0.40545014129995965,
      "grad_norm": 106.34741973876953,
      "learning_rate": 1.3474753310377258e-05,
      "loss": 4.0659,
      "step": 10043
    },
    {
      "epoch": 0.4054905127169964,
      "grad_norm": 78.27279663085938,
      "learning_rate": 1.3473527176761127e-05,
      "loss": 6.1162,
      "step": 10044
    },
    {
      "epoch": 0.4055308841340331,
      "grad_norm": 211.0213165283203,
      "learning_rate": 1.3472300983755794e-05,
      "loss": 4.4663,
      "step": 10045
    },
    {
      "epoch": 0.4055712555510698,
      "grad_norm": 79.83462524414062,
      "learning_rate": 1.3471074731382224e-05,
      "loss": 5.1384,
      "step": 10046
    },
    {
      "epoch": 0.4056116269681066,
      "grad_norm": 113.06800079345703,
      "learning_rate": 1.3469848419661382e-05,
      "loss": 6.0491,
      "step": 10047
    },
    {
      "epoch": 0.4056519983851433,
      "grad_norm": 90.31826782226562,
      "learning_rate": 1.3468622048614233e-05,
      "loss": 3.6255,
      "step": 10048
    },
    {
      "epoch": 0.40569236980218004,
      "grad_norm": 77.02816009521484,
      "learning_rate": 1.3467395618261744e-05,
      "loss": 3.8453,
      "step": 10049
    },
    {
      "epoch": 0.4057327412192168,
      "grad_norm": 80.68212890625,
      "learning_rate": 1.346616912862489e-05,
      "loss": 5.855,
      "step": 10050
    },
    {
      "epoch": 0.40577311263625354,
      "grad_norm": 104.8309326171875,
      "learning_rate": 1.346494257972464e-05,
      "loss": 4.8059,
      "step": 10051
    },
    {
      "epoch": 0.40581348405329026,
      "grad_norm": 106.13055419921875,
      "learning_rate": 1.3463715971581958e-05,
      "loss": 5.1864,
      "step": 10052
    },
    {
      "epoch": 0.405853855470327,
      "grad_norm": 65.47319030761719,
      "learning_rate": 1.3462489304217828e-05,
      "loss": 5.9652,
      "step": 10053
    },
    {
      "epoch": 0.40589422688736376,
      "grad_norm": 80.28852081298828,
      "learning_rate": 1.3461262577653211e-05,
      "loss": 4.73,
      "step": 10054
    },
    {
      "epoch": 0.4059345983044005,
      "grad_norm": 73.84825897216797,
      "learning_rate": 1.3460035791909089e-05,
      "loss": 5.2839,
      "step": 10055
    },
    {
      "epoch": 0.4059749697214372,
      "grad_norm": 99.5623550415039,
      "learning_rate": 1.3458808947006434e-05,
      "loss": 4.866,
      "step": 10056
    },
    {
      "epoch": 0.406015341138474,
      "grad_norm": 153.0874786376953,
      "learning_rate": 1.3457582042966224e-05,
      "loss": 5.9181,
      "step": 10057
    },
    {
      "epoch": 0.4060557125555107,
      "grad_norm": 100.58978271484375,
      "learning_rate": 1.3456355079809434e-05,
      "loss": 4.4015,
      "step": 10058
    },
    {
      "epoch": 0.40609608397254743,
      "grad_norm": 115.27201080322266,
      "learning_rate": 1.3455128057557043e-05,
      "loss": 5.4761,
      "step": 10059
    },
    {
      "epoch": 0.40613645538958415,
      "grad_norm": 100.23486328125,
      "learning_rate": 1.3453900976230027e-05,
      "loss": 2.8709,
      "step": 10060
    },
    {
      "epoch": 0.4061768268066209,
      "grad_norm": 115.50223541259766,
      "learning_rate": 1.3452673835849377e-05,
      "loss": 6.581,
      "step": 10061
    },
    {
      "epoch": 0.40621719822365765,
      "grad_norm": 91.41007232666016,
      "learning_rate": 1.3451446636436063e-05,
      "loss": 5.9622,
      "step": 10062
    },
    {
      "epoch": 0.40625756964069437,
      "grad_norm": 104.51646423339844,
      "learning_rate": 1.3450219378011071e-05,
      "loss": 6.1178,
      "step": 10063
    },
    {
      "epoch": 0.40629794105773115,
      "grad_norm": 91.1811752319336,
      "learning_rate": 1.3448992060595384e-05,
      "loss": 5.4404,
      "step": 10064
    },
    {
      "epoch": 0.40633831247476787,
      "grad_norm": 131.32290649414062,
      "learning_rate": 1.344776468420999e-05,
      "loss": 5.2076,
      "step": 10065
    },
    {
      "epoch": 0.4063786838918046,
      "grad_norm": 119.97634887695312,
      "learning_rate": 1.3446537248875868e-05,
      "loss": 4.3806,
      "step": 10066
    },
    {
      "epoch": 0.4064190553088413,
      "grad_norm": 92.14669799804688,
      "learning_rate": 1.344530975461401e-05,
      "loss": 5.2044,
      "step": 10067
    },
    {
      "epoch": 0.4064594267258781,
      "grad_norm": 115.68241119384766,
      "learning_rate": 1.3444082201445399e-05,
      "loss": 4.455,
      "step": 10068
    },
    {
      "epoch": 0.4064997981429148,
      "grad_norm": 109.7421875,
      "learning_rate": 1.3442854589391023e-05,
      "loss": 5.1324,
      "step": 10069
    },
    {
      "epoch": 0.40654016955995154,
      "grad_norm": 108.95244598388672,
      "learning_rate": 1.3441626918471875e-05,
      "loss": 5.5584,
      "step": 10070
    },
    {
      "epoch": 0.4065805409769883,
      "grad_norm": 89.95259094238281,
      "learning_rate": 1.3440399188708944e-05,
      "loss": 3.8692,
      "step": 10071
    },
    {
      "epoch": 0.40662091239402504,
      "grad_norm": 96.25947570800781,
      "learning_rate": 1.3439171400123218e-05,
      "loss": 5.821,
      "step": 10072
    },
    {
      "epoch": 0.40666128381106176,
      "grad_norm": 94.86469268798828,
      "learning_rate": 1.3437943552735697e-05,
      "loss": 5.3102,
      "step": 10073
    },
    {
      "epoch": 0.4067016552280985,
      "grad_norm": 82.27680969238281,
      "learning_rate": 1.3436715646567366e-05,
      "loss": 3.0917,
      "step": 10074
    },
    {
      "epoch": 0.40674202664513526,
      "grad_norm": 120.13681030273438,
      "learning_rate": 1.3435487681639224e-05,
      "loss": 4.499,
      "step": 10075
    },
    {
      "epoch": 0.406782398062172,
      "grad_norm": 96.13814544677734,
      "learning_rate": 1.3434259657972263e-05,
      "loss": 4.6532,
      "step": 10076
    },
    {
      "epoch": 0.4068227694792087,
      "grad_norm": 112.08829498291016,
      "learning_rate": 1.3433031575587485e-05,
      "loss": 3.294,
      "step": 10077
    },
    {
      "epoch": 0.4068631408962455,
      "grad_norm": 103.73388671875,
      "learning_rate": 1.3431803434505881e-05,
      "loss": 4.8909,
      "step": 10078
    },
    {
      "epoch": 0.4069035123132822,
      "grad_norm": 87.42671966552734,
      "learning_rate": 1.343057523474845e-05,
      "loss": 6.0851,
      "step": 10079
    },
    {
      "epoch": 0.4069438837303189,
      "grad_norm": 142.49673461914062,
      "learning_rate": 1.3429346976336197e-05,
      "loss": 4.5489,
      "step": 10080
    },
    {
      "epoch": 0.40698425514735564,
      "grad_norm": 139.86285400390625,
      "learning_rate": 1.342811865929012e-05,
      "loss": 5.2375,
      "step": 10081
    },
    {
      "epoch": 0.4070246265643924,
      "grad_norm": 161.6896514892578,
      "learning_rate": 1.3426890283631215e-05,
      "loss": 5.309,
      "step": 10082
    },
    {
      "epoch": 0.40706499798142914,
      "grad_norm": 117.32080078125,
      "learning_rate": 1.3425661849380491e-05,
      "loss": 6.7177,
      "step": 10083
    },
    {
      "epoch": 0.40710536939846587,
      "grad_norm": 184.79798889160156,
      "learning_rate": 1.342443335655895e-05,
      "loss": 4.7463,
      "step": 10084
    },
    {
      "epoch": 0.40714574081550264,
      "grad_norm": 113.6337890625,
      "learning_rate": 1.3423204805187595e-05,
      "loss": 4.8087,
      "step": 10085
    },
    {
      "epoch": 0.40718611223253937,
      "grad_norm": 110.78534698486328,
      "learning_rate": 1.3421976195287428e-05,
      "loss": 3.4263,
      "step": 10086
    },
    {
      "epoch": 0.4072264836495761,
      "grad_norm": 137.0172882080078,
      "learning_rate": 1.3420747526879462e-05,
      "loss": 6.3911,
      "step": 10087
    },
    {
      "epoch": 0.40726685506661287,
      "grad_norm": 103.4755630493164,
      "learning_rate": 1.34195187999847e-05,
      "loss": 6.1474,
      "step": 10088
    },
    {
      "epoch": 0.4073072264836496,
      "grad_norm": 102.11587524414062,
      "learning_rate": 1.3418290014624153e-05,
      "loss": 3.7148,
      "step": 10089
    },
    {
      "epoch": 0.4073475979006863,
      "grad_norm": 106.99130249023438,
      "learning_rate": 1.3417061170818828e-05,
      "loss": 7.4922,
      "step": 10090
    },
    {
      "epoch": 0.40738796931772303,
      "grad_norm": 112.6327896118164,
      "learning_rate": 1.3415832268589737e-05,
      "loss": 5.8244,
      "step": 10091
    },
    {
      "epoch": 0.4074283407347598,
      "grad_norm": 563.2591552734375,
      "learning_rate": 1.341460330795789e-05,
      "loss": 5.9183,
      "step": 10092
    },
    {
      "epoch": 0.40746871215179653,
      "grad_norm": 104.87904357910156,
      "learning_rate": 1.34133742889443e-05,
      "loss": 4.4611,
      "step": 10093
    },
    {
      "epoch": 0.40750908356883325,
      "grad_norm": 80.32937622070312,
      "learning_rate": 1.3412145211569982e-05,
      "loss": 3.9849,
      "step": 10094
    },
    {
      "epoch": 0.40754945498587003,
      "grad_norm": 107.7944564819336,
      "learning_rate": 1.3410916075855949e-05,
      "loss": 4.9923,
      "step": 10095
    },
    {
      "epoch": 0.40758982640290675,
      "grad_norm": 118.14616394042969,
      "learning_rate": 1.3409686881823214e-05,
      "loss": 3.724,
      "step": 10096
    },
    {
      "epoch": 0.4076301978199435,
      "grad_norm": 271.8922119140625,
      "learning_rate": 1.3408457629492797e-05,
      "loss": 5.0856,
      "step": 10097
    },
    {
      "epoch": 0.4076705692369802,
      "grad_norm": 140.9392547607422,
      "learning_rate": 1.3407228318885715e-05,
      "loss": 5.7173,
      "step": 10098
    },
    {
      "epoch": 0.407710940654017,
      "grad_norm": 75.95709228515625,
      "learning_rate": 1.340599895002298e-05,
      "loss": 5.1386,
      "step": 10099
    },
    {
      "epoch": 0.4077513120710537,
      "grad_norm": 94.9743423461914,
      "learning_rate": 1.340476952292562e-05,
      "loss": 3.991,
      "step": 10100
    },
    {
      "epoch": 0.4077916834880904,
      "grad_norm": 83.06111145019531,
      "learning_rate": 1.3403540037614654e-05,
      "loss": 5.9607,
      "step": 10101
    },
    {
      "epoch": 0.4078320549051272,
      "grad_norm": 88.21978759765625,
      "learning_rate": 1.3402310494111097e-05,
      "loss": 5.5847,
      "step": 10102
    },
    {
      "epoch": 0.4078724263221639,
      "grad_norm": 116.76950073242188,
      "learning_rate": 1.340108089243598e-05,
      "loss": 5.6638,
      "step": 10103
    },
    {
      "epoch": 0.40791279773920064,
      "grad_norm": 95.46693420410156,
      "learning_rate": 1.3399851232610319e-05,
      "loss": 4.5574,
      "step": 10104
    },
    {
      "epoch": 0.40795316915623736,
      "grad_norm": 97.44046020507812,
      "learning_rate": 1.3398621514655142e-05,
      "loss": 5.6531,
      "step": 10105
    },
    {
      "epoch": 0.40799354057327414,
      "grad_norm": 120.39663696289062,
      "learning_rate": 1.3397391738591471e-05,
      "loss": 3.8157,
      "step": 10106
    },
    {
      "epoch": 0.40803391199031086,
      "grad_norm": 82.26434326171875,
      "learning_rate": 1.339616190444034e-05,
      "loss": 6.3046,
      "step": 10107
    },
    {
      "epoch": 0.4080742834073476,
      "grad_norm": 105.66761016845703,
      "learning_rate": 1.3394932012222767e-05,
      "loss": 5.1794,
      "step": 10108
    },
    {
      "epoch": 0.40811465482438436,
      "grad_norm": 89.67098236083984,
      "learning_rate": 1.3393702061959783e-05,
      "loss": 6.4019,
      "step": 10109
    },
    {
      "epoch": 0.4081550262414211,
      "grad_norm": 151.58062744140625,
      "learning_rate": 1.339247205367242e-05,
      "loss": 4.1107,
      "step": 10110
    },
    {
      "epoch": 0.4081953976584578,
      "grad_norm": 97.28108978271484,
      "learning_rate": 1.3391241987381711e-05,
      "loss": 3.9911,
      "step": 10111
    },
    {
      "epoch": 0.4082357690754945,
      "grad_norm": 106.11190795898438,
      "learning_rate": 1.3390011863108678e-05,
      "loss": 5.0295,
      "step": 10112
    },
    {
      "epoch": 0.4082761404925313,
      "grad_norm": 107.41531372070312,
      "learning_rate": 1.3388781680874363e-05,
      "loss": 5.5639,
      "step": 10113
    },
    {
      "epoch": 0.408316511909568,
      "grad_norm": 75.5036392211914,
      "learning_rate": 1.338755144069979e-05,
      "loss": 4.9446,
      "step": 10114
    },
    {
      "epoch": 0.40835688332660475,
      "grad_norm": 120.62325286865234,
      "learning_rate": 1.3386321142606e-05,
      "loss": 4.5197,
      "step": 10115
    },
    {
      "epoch": 0.4083972547436415,
      "grad_norm": 110.95451354980469,
      "learning_rate": 1.3385090786614025e-05,
      "loss": 5.9163,
      "step": 10116
    },
    {
      "epoch": 0.40843762616067825,
      "grad_norm": 98.3787612915039,
      "learning_rate": 1.3383860372744903e-05,
      "loss": 5.6739,
      "step": 10117
    },
    {
      "epoch": 0.40847799757771497,
      "grad_norm": 135.16444396972656,
      "learning_rate": 1.3382629901019672e-05,
      "loss": 3.917,
      "step": 10118
    },
    {
      "epoch": 0.4085183689947517,
      "grad_norm": 80.58392333984375,
      "learning_rate": 1.3381399371459368e-05,
      "loss": 6.754,
      "step": 10119
    },
    {
      "epoch": 0.40855874041178847,
      "grad_norm": 101.70562744140625,
      "learning_rate": 1.3380168784085028e-05,
      "loss": 4.1506,
      "step": 10120
    },
    {
      "epoch": 0.4085991118288252,
      "grad_norm": 105.31814575195312,
      "learning_rate": 1.3378938138917698e-05,
      "loss": 4.0771,
      "step": 10121
    },
    {
      "epoch": 0.4086394832458619,
      "grad_norm": 67.92202758789062,
      "learning_rate": 1.3377707435978415e-05,
      "loss": 6.1742,
      "step": 10122
    },
    {
      "epoch": 0.4086798546628987,
      "grad_norm": 84.94874572753906,
      "learning_rate": 1.3376476675288226e-05,
      "loss": 5.0584,
      "step": 10123
    },
    {
      "epoch": 0.4087202260799354,
      "grad_norm": 87.56005096435547,
      "learning_rate": 1.3375245856868164e-05,
      "loss": 5.1587,
      "step": 10124
    },
    {
      "epoch": 0.40876059749697213,
      "grad_norm": 100.33807373046875,
      "learning_rate": 1.3374014980739284e-05,
      "loss": 6.2011,
      "step": 10125
    },
    {
      "epoch": 0.40880096891400886,
      "grad_norm": 99.86756896972656,
      "learning_rate": 1.3372784046922626e-05,
      "loss": 6.0928,
      "step": 10126
    },
    {
      "epoch": 0.40884134033104563,
      "grad_norm": 101.8389892578125,
      "learning_rate": 1.3371553055439237e-05,
      "loss": 4.9333,
      "step": 10127
    },
    {
      "epoch": 0.40888171174808235,
      "grad_norm": 82.93124389648438,
      "learning_rate": 1.3370322006310162e-05,
      "loss": 4.8827,
      "step": 10128
    },
    {
      "epoch": 0.4089220831651191,
      "grad_norm": 104.3852310180664,
      "learning_rate": 1.3369090899556452e-05,
      "loss": 5.8521,
      "step": 10129
    },
    {
      "epoch": 0.40896245458215585,
      "grad_norm": 120.7365951538086,
      "learning_rate": 1.3367859735199156e-05,
      "loss": 5.8885,
      "step": 10130
    },
    {
      "epoch": 0.4090028259991926,
      "grad_norm": 135.47276306152344,
      "learning_rate": 1.3366628513259323e-05,
      "loss": 5.259,
      "step": 10131
    },
    {
      "epoch": 0.4090431974162293,
      "grad_norm": 104.60509490966797,
      "learning_rate": 1.3365397233758004e-05,
      "loss": 3.7108,
      "step": 10132
    },
    {
      "epoch": 0.409083568833266,
      "grad_norm": 101.94896697998047,
      "learning_rate": 1.336416589671625e-05,
      "loss": 4.6993,
      "step": 10133
    },
    {
      "epoch": 0.4091239402503028,
      "grad_norm": 208.6069793701172,
      "learning_rate": 1.3362934502155116e-05,
      "loss": 5.9259,
      "step": 10134
    },
    {
      "epoch": 0.4091643116673395,
      "grad_norm": 111.33734130859375,
      "learning_rate": 1.3361703050095656e-05,
      "loss": 5.2957,
      "step": 10135
    },
    {
      "epoch": 0.40920468308437624,
      "grad_norm": 156.58717346191406,
      "learning_rate": 1.3360471540558921e-05,
      "loss": 5.1061,
      "step": 10136
    },
    {
      "epoch": 0.409245054501413,
      "grad_norm": 98.36089324951172,
      "learning_rate": 1.3359239973565972e-05,
      "loss": 6.6462,
      "step": 10137
    },
    {
      "epoch": 0.40928542591844974,
      "grad_norm": 103.38101959228516,
      "learning_rate": 1.3358008349137862e-05,
      "loss": 5.1295,
      "step": 10138
    },
    {
      "epoch": 0.40932579733548646,
      "grad_norm": 88.64895629882812,
      "learning_rate": 1.3356776667295653e-05,
      "loss": 4.6679,
      "step": 10139
    },
    {
      "epoch": 0.40936616875252324,
      "grad_norm": 92.20588684082031,
      "learning_rate": 1.33555449280604e-05,
      "loss": 4.8697,
      "step": 10140
    },
    {
      "epoch": 0.40940654016955996,
      "grad_norm": 117.84110260009766,
      "learning_rate": 1.3354313131453167e-05,
      "loss": 5.3303,
      "step": 10141
    },
    {
      "epoch": 0.4094469115865967,
      "grad_norm": 128.4618682861328,
      "learning_rate": 1.3353081277495008e-05,
      "loss": 4.1627,
      "step": 10142
    },
    {
      "epoch": 0.4094872830036334,
      "grad_norm": 94.3980941772461,
      "learning_rate": 1.3351849366206992e-05,
      "loss": 5.3188,
      "step": 10143
    },
    {
      "epoch": 0.4095276544206702,
      "grad_norm": 189.1033935546875,
      "learning_rate": 1.335061739761018e-05,
      "loss": 3.9299,
      "step": 10144
    },
    {
      "epoch": 0.4095680258377069,
      "grad_norm": 1590.4337158203125,
      "learning_rate": 1.3349385371725634e-05,
      "loss": 4.7427,
      "step": 10145
    },
    {
      "epoch": 0.40960839725474363,
      "grad_norm": 65.62686920166016,
      "learning_rate": 1.334815328857442e-05,
      "loss": 4.0714,
      "step": 10146
    },
    {
      "epoch": 0.4096487686717804,
      "grad_norm": 174.4590301513672,
      "learning_rate": 1.3346921148177601e-05,
      "loss": 4.5366,
      "step": 10147
    },
    {
      "epoch": 0.4096891400888171,
      "grad_norm": 138.1952362060547,
      "learning_rate": 1.3345688950556246e-05,
      "loss": 4.9561,
      "step": 10148
    },
    {
      "epoch": 0.40972951150585385,
      "grad_norm": 125.59071350097656,
      "learning_rate": 1.3344456695731425e-05,
      "loss": 6.4049,
      "step": 10149
    },
    {
      "epoch": 0.40976988292289057,
      "grad_norm": 104.07972717285156,
      "learning_rate": 1.3343224383724203e-05,
      "loss": 5.7799,
      "step": 10150
    },
    {
      "epoch": 0.40981025433992735,
      "grad_norm": 102.18998718261719,
      "learning_rate": 1.3341992014555651e-05,
      "loss": 3.0123,
      "step": 10151
    },
    {
      "epoch": 0.40985062575696407,
      "grad_norm": 148.01080322265625,
      "learning_rate": 1.334075958824684e-05,
      "loss": 4.4943,
      "step": 10152
    },
    {
      "epoch": 0.4098909971740008,
      "grad_norm": 85.30939483642578,
      "learning_rate": 1.3339527104818842e-05,
      "loss": 4.3314,
      "step": 10153
    },
    {
      "epoch": 0.40993136859103757,
      "grad_norm": 82.5202865600586,
      "learning_rate": 1.3338294564292729e-05,
      "loss": 6.5203,
      "step": 10154
    },
    {
      "epoch": 0.4099717400080743,
      "grad_norm": 80.07940673828125,
      "learning_rate": 1.3337061966689573e-05,
      "loss": 7.0884,
      "step": 10155
    },
    {
      "epoch": 0.410012111425111,
      "grad_norm": 116.40235900878906,
      "learning_rate": 1.333582931203045e-05,
      "loss": 6.1906,
      "step": 10156
    },
    {
      "epoch": 0.41005248284214774,
      "grad_norm": 110.14173889160156,
      "learning_rate": 1.3334596600336436e-05,
      "loss": 4.7938,
      "step": 10157
    },
    {
      "epoch": 0.4100928542591845,
      "grad_norm": 123.69001770019531,
      "learning_rate": 1.3333363831628605e-05,
      "loss": 5.2865,
      "step": 10158
    },
    {
      "epoch": 0.41013322567622124,
      "grad_norm": 90.4808349609375,
      "learning_rate": 1.333213100592804e-05,
      "loss": 6.5768,
      "step": 10159
    },
    {
      "epoch": 0.41017359709325796,
      "grad_norm": 110.53754425048828,
      "learning_rate": 1.3330898123255814e-05,
      "loss": 5.564,
      "step": 10160
    },
    {
      "epoch": 0.41021396851029474,
      "grad_norm": 166.59754943847656,
      "learning_rate": 1.3329665183633009e-05,
      "loss": 5.6468,
      "step": 10161
    },
    {
      "epoch": 0.41025433992733146,
      "grad_norm": 123.2160415649414,
      "learning_rate": 1.3328432187080705e-05,
      "loss": 5.0204,
      "step": 10162
    },
    {
      "epoch": 0.4102947113443682,
      "grad_norm": 103.83979034423828,
      "learning_rate": 1.3327199133619982e-05,
      "loss": 4.9831,
      "step": 10163
    },
    {
      "epoch": 0.4103350827614049,
      "grad_norm": 83.0586929321289,
      "learning_rate": 1.3325966023271923e-05,
      "loss": 6.2539,
      "step": 10164
    },
    {
      "epoch": 0.4103754541784417,
      "grad_norm": 97.14961242675781,
      "learning_rate": 1.3324732856057615e-05,
      "loss": 3.6793,
      "step": 10165
    },
    {
      "epoch": 0.4104158255954784,
      "grad_norm": 69.59477996826172,
      "learning_rate": 1.3323499631998136e-05,
      "loss": 6.6674,
      "step": 10166
    },
    {
      "epoch": 0.4104561970125151,
      "grad_norm": 103.09156799316406,
      "learning_rate": 1.3322266351114572e-05,
      "loss": 5.6508,
      "step": 10167
    },
    {
      "epoch": 0.4104965684295519,
      "grad_norm": 86.5912094116211,
      "learning_rate": 1.3321033013428014e-05,
      "loss": 5.252,
      "step": 10168
    },
    {
      "epoch": 0.4105369398465886,
      "grad_norm": 77.24353790283203,
      "learning_rate": 1.3319799618959548e-05,
      "loss": 5.4286,
      "step": 10169
    },
    {
      "epoch": 0.41057731126362534,
      "grad_norm": 110.56656646728516,
      "learning_rate": 1.3318566167730257e-05,
      "loss": 5.9172,
      "step": 10170
    },
    {
      "epoch": 0.41061768268066207,
      "grad_norm": 112.06303405761719,
      "learning_rate": 1.331733265976124e-05,
      "loss": 5.3736,
      "step": 10171
    },
    {
      "epoch": 0.41065805409769884,
      "grad_norm": 113.67191314697266,
      "learning_rate": 1.3316099095073575e-05,
      "loss": 3.543,
      "step": 10172
    },
    {
      "epoch": 0.41069842551473557,
      "grad_norm": 78.44924926757812,
      "learning_rate": 1.3314865473688362e-05,
      "loss": 5.3771,
      "step": 10173
    },
    {
      "epoch": 0.4107387969317723,
      "grad_norm": 100.90061950683594,
      "learning_rate": 1.3313631795626691e-05,
      "loss": 7.1608,
      "step": 10174
    },
    {
      "epoch": 0.41077916834880907,
      "grad_norm": 94.10598754882812,
      "learning_rate": 1.3312398060909652e-05,
      "loss": 4.6622,
      "step": 10175
    },
    {
      "epoch": 0.4108195397658458,
      "grad_norm": 57.4235725402832,
      "learning_rate": 1.3311164269558343e-05,
      "loss": 5.9691,
      "step": 10176
    },
    {
      "epoch": 0.4108599111828825,
      "grad_norm": 103.3335189819336,
      "learning_rate": 1.3309930421593854e-05,
      "loss": 5.7188,
      "step": 10177
    },
    {
      "epoch": 0.41090028259991923,
      "grad_norm": 72.96768951416016,
      "learning_rate": 1.3308696517037287e-05,
      "loss": 6.1966,
      "step": 10178
    },
    {
      "epoch": 0.410940654016956,
      "grad_norm": 122.53529357910156,
      "learning_rate": 1.3307462555909736e-05,
      "loss": 6.3575,
      "step": 10179
    },
    {
      "epoch": 0.41098102543399273,
      "grad_norm": 80.04006958007812,
      "learning_rate": 1.3306228538232297e-05,
      "loss": 3.7529,
      "step": 10180
    },
    {
      "epoch": 0.41102139685102945,
      "grad_norm": 368.7599792480469,
      "learning_rate": 1.3304994464026072e-05,
      "loss": 5.3936,
      "step": 10181
    },
    {
      "epoch": 0.41106176826806623,
      "grad_norm": 86.76416015625,
      "learning_rate": 1.330376033331216e-05,
      "loss": 6.7972,
      "step": 10182
    },
    {
      "epoch": 0.41110213968510295,
      "grad_norm": 66.76097869873047,
      "learning_rate": 1.330252614611166e-05,
      "loss": 5.6977,
      "step": 10183
    },
    {
      "epoch": 0.4111425111021397,
      "grad_norm": 72.02337646484375,
      "learning_rate": 1.3301291902445675e-05,
      "loss": 5.6173,
      "step": 10184
    },
    {
      "epoch": 0.4111828825191764,
      "grad_norm": 104.27555084228516,
      "learning_rate": 1.3300057602335307e-05,
      "loss": 5.4784,
      "step": 10185
    },
    {
      "epoch": 0.4112232539362132,
      "grad_norm": 73.5774917602539,
      "learning_rate": 1.3298823245801662e-05,
      "loss": 5.8256,
      "step": 10186
    },
    {
      "epoch": 0.4112636253532499,
      "grad_norm": 80.94157409667969,
      "learning_rate": 1.329758883286584e-05,
      "loss": 4.8011,
      "step": 10187
    },
    {
      "epoch": 0.4113039967702866,
      "grad_norm": 112.50121307373047,
      "learning_rate": 1.3296354363548951e-05,
      "loss": 4.6994,
      "step": 10188
    },
    {
      "epoch": 0.4113443681873234,
      "grad_norm": 83.04290771484375,
      "learning_rate": 1.32951198378721e-05,
      "loss": 6.7992,
      "step": 10189
    },
    {
      "epoch": 0.4113847396043601,
      "grad_norm": 76.98265838623047,
      "learning_rate": 1.3293885255856394e-05,
      "loss": 6.0414,
      "step": 10190
    },
    {
      "epoch": 0.41142511102139684,
      "grad_norm": 80.80752563476562,
      "learning_rate": 1.3292650617522943e-05,
      "loss": 4.3165,
      "step": 10191
    },
    {
      "epoch": 0.4114654824384336,
      "grad_norm": 134.37030029296875,
      "learning_rate": 1.3291415922892854e-05,
      "loss": 4.6243,
      "step": 10192
    },
    {
      "epoch": 0.41150585385547034,
      "grad_norm": 150.3105926513672,
      "learning_rate": 1.329018117198724e-05,
      "loss": 4.5645,
      "step": 10193
    },
    {
      "epoch": 0.41154622527250706,
      "grad_norm": 112.21469116210938,
      "learning_rate": 1.3288946364827209e-05,
      "loss": 5.1743,
      "step": 10194
    },
    {
      "epoch": 0.4115865966895438,
      "grad_norm": 102.79222106933594,
      "learning_rate": 1.3287711501433877e-05,
      "loss": 4.423,
      "step": 10195
    },
    {
      "epoch": 0.41162696810658056,
      "grad_norm": 124.86481475830078,
      "learning_rate": 1.3286476581828356e-05,
      "loss": 4.346,
      "step": 10196
    },
    {
      "epoch": 0.4116673395236173,
      "grad_norm": 110.96976470947266,
      "learning_rate": 1.3285241606031758e-05,
      "loss": 6.655,
      "step": 10197
    },
    {
      "epoch": 0.411707710940654,
      "grad_norm": 106.25382232666016,
      "learning_rate": 1.3284006574065201e-05,
      "loss": 5.1974,
      "step": 10198
    },
    {
      "epoch": 0.4117480823576908,
      "grad_norm": 69.6541976928711,
      "learning_rate": 1.3282771485949802e-05,
      "loss": 5.286,
      "step": 10199
    },
    {
      "epoch": 0.4117884537747275,
      "grad_norm": 90.88251495361328,
      "learning_rate": 1.3281536341706676e-05,
      "loss": 5.3244,
      "step": 10200
    },
    {
      "epoch": 0.4118288251917642,
      "grad_norm": 72.98125457763672,
      "learning_rate": 1.328030114135694e-05,
      "loss": 5.1653,
      "step": 10201
    },
    {
      "epoch": 0.41186919660880095,
      "grad_norm": 116.73672485351562,
      "learning_rate": 1.3279065884921716e-05,
      "loss": 4.8546,
      "step": 10202
    },
    {
      "epoch": 0.4119095680258377,
      "grad_norm": 93.95903015136719,
      "learning_rate": 1.3277830572422122e-05,
      "loss": 6.0345,
      "step": 10203
    },
    {
      "epoch": 0.41194993944287445,
      "grad_norm": 346.9225769042969,
      "learning_rate": 1.3276595203879278e-05,
      "loss": 6.1799,
      "step": 10204
    },
    {
      "epoch": 0.41199031085991117,
      "grad_norm": 110.83210754394531,
      "learning_rate": 1.327535977931431e-05,
      "loss": 5.0192,
      "step": 10205
    },
    {
      "epoch": 0.41203068227694795,
      "grad_norm": 134.1188201904297,
      "learning_rate": 1.3274124298748334e-05,
      "loss": 4.5898,
      "step": 10206
    },
    {
      "epoch": 0.41207105369398467,
      "grad_norm": 91.6152572631836,
      "learning_rate": 1.3272888762202482e-05,
      "loss": 4.4309,
      "step": 10207
    },
    {
      "epoch": 0.4121114251110214,
      "grad_norm": 79.46131896972656,
      "learning_rate": 1.3271653169697874e-05,
      "loss": 3.9091,
      "step": 10208
    },
    {
      "epoch": 0.4121517965280581,
      "grad_norm": 92.69355773925781,
      "learning_rate": 1.3270417521255637e-05,
      "loss": 5.7948,
      "step": 10209
    },
    {
      "epoch": 0.4121921679450949,
      "grad_norm": 149.5449981689453,
      "learning_rate": 1.3269181816896897e-05,
      "loss": 5.355,
      "step": 10210
    },
    {
      "epoch": 0.4122325393621316,
      "grad_norm": 88.38428497314453,
      "learning_rate": 1.3267946056642782e-05,
      "loss": 4.7199,
      "step": 10211
    },
    {
      "epoch": 0.41227291077916833,
      "grad_norm": 158.56265258789062,
      "learning_rate": 1.326671024051442e-05,
      "loss": 5.4582,
      "step": 10212
    },
    {
      "epoch": 0.4123132821962051,
      "grad_norm": 89.08345794677734,
      "learning_rate": 1.3265474368532942e-05,
      "loss": 4.521,
      "step": 10213
    },
    {
      "epoch": 0.41235365361324183,
      "grad_norm": 83.41961669921875,
      "learning_rate": 1.3264238440719477e-05,
      "loss": 5.6583,
      "step": 10214
    },
    {
      "epoch": 0.41239402503027855,
      "grad_norm": 91.61426544189453,
      "learning_rate": 1.326300245709516e-05,
      "loss": 4.9128,
      "step": 10215
    },
    {
      "epoch": 0.4124343964473153,
      "grad_norm": 74.51669311523438,
      "learning_rate": 1.3261766417681117e-05,
      "loss": 6.4882,
      "step": 10216
    },
    {
      "epoch": 0.41247476786435205,
      "grad_norm": 113.1467056274414,
      "learning_rate": 1.3260530322498486e-05,
      "loss": 3.5694,
      "step": 10217
    },
    {
      "epoch": 0.4125151392813888,
      "grad_norm": 81.74510192871094,
      "learning_rate": 1.32592941715684e-05,
      "loss": 5.3912,
      "step": 10218
    },
    {
      "epoch": 0.4125555106984255,
      "grad_norm": 99.74324035644531,
      "learning_rate": 1.3258057964911997e-05,
      "loss": 5.1403,
      "step": 10219
    },
    {
      "epoch": 0.4125958821154623,
      "grad_norm": 77.99755096435547,
      "learning_rate": 1.3256821702550408e-05,
      "loss": 6.2998,
      "step": 10220
    },
    {
      "epoch": 0.412636253532499,
      "grad_norm": 70.31307220458984,
      "learning_rate": 1.3255585384504777e-05,
      "loss": 3.3276,
      "step": 10221
    },
    {
      "epoch": 0.4126766249495357,
      "grad_norm": 87.88262176513672,
      "learning_rate": 1.3254349010796233e-05,
      "loss": 3.7483,
      "step": 10222
    },
    {
      "epoch": 0.41271699636657244,
      "grad_norm": 78.67647552490234,
      "learning_rate": 1.3253112581445924e-05,
      "loss": 3.7059,
      "step": 10223
    },
    {
      "epoch": 0.4127573677836092,
      "grad_norm": 74.22998809814453,
      "learning_rate": 1.3251876096474985e-05,
      "loss": 5.5121,
      "step": 10224
    },
    {
      "epoch": 0.41279773920064594,
      "grad_norm": 61.932064056396484,
      "learning_rate": 1.3250639555904559e-05,
      "loss": 4.255,
      "step": 10225
    },
    {
      "epoch": 0.41283811061768266,
      "grad_norm": 109.53895568847656,
      "learning_rate": 1.3249402959755786e-05,
      "loss": 4.2431,
      "step": 10226
    },
    {
      "epoch": 0.41287848203471944,
      "grad_norm": 116.07250213623047,
      "learning_rate": 1.3248166308049813e-05,
      "loss": 3.8072,
      "step": 10227
    },
    {
      "epoch": 0.41291885345175616,
      "grad_norm": 92.01045227050781,
      "learning_rate": 1.3246929600807779e-05,
      "loss": 4.7211,
      "step": 10228
    },
    {
      "epoch": 0.4129592248687929,
      "grad_norm": 90.76921844482422,
      "learning_rate": 1.3245692838050833e-05,
      "loss": 4.6872,
      "step": 10229
    },
    {
      "epoch": 0.4129995962858296,
      "grad_norm": 165.0478515625,
      "learning_rate": 1.3244456019800116e-05,
      "loss": 5.3824,
      "step": 10230
    },
    {
      "epoch": 0.4130399677028664,
      "grad_norm": 92.5075912475586,
      "learning_rate": 1.324321914607678e-05,
      "loss": 5.5769,
      "step": 10231
    },
    {
      "epoch": 0.4130803391199031,
      "grad_norm": 78.55841064453125,
      "learning_rate": 1.3241982216901967e-05,
      "loss": 6.4987,
      "step": 10232
    },
    {
      "epoch": 0.41312071053693983,
      "grad_norm": 66.01636505126953,
      "learning_rate": 1.3240745232296832e-05,
      "loss": 5.7446,
      "step": 10233
    },
    {
      "epoch": 0.4131610819539766,
      "grad_norm": 105.69862365722656,
      "learning_rate": 1.323950819228252e-05,
      "loss": 5.1183,
      "step": 10234
    },
    {
      "epoch": 0.4132014533710133,
      "grad_norm": 88.36262512207031,
      "learning_rate": 1.3238271096880182e-05,
      "loss": 4.7805,
      "step": 10235
    },
    {
      "epoch": 0.41324182478805005,
      "grad_norm": 92.85897827148438,
      "learning_rate": 1.323703394611097e-05,
      "loss": 5.9341,
      "step": 10236
    },
    {
      "epoch": 0.41328219620508677,
      "grad_norm": 93.78348541259766,
      "learning_rate": 1.3235796739996038e-05,
      "loss": 3.9447,
      "step": 10237
    },
    {
      "epoch": 0.41332256762212355,
      "grad_norm": 113.99681091308594,
      "learning_rate": 1.3234559478556537e-05,
      "loss": 6.139,
      "step": 10238
    },
    {
      "epoch": 0.41336293903916027,
      "grad_norm": 78.10201263427734,
      "learning_rate": 1.3233322161813623e-05,
      "loss": 4.6872,
      "step": 10239
    },
    {
      "epoch": 0.413403310456197,
      "grad_norm": 94.2849349975586,
      "learning_rate": 1.323208478978845e-05,
      "loss": 6.5478,
      "step": 10240
    },
    {
      "epoch": 0.41344368187323377,
      "grad_norm": 126.65849304199219,
      "learning_rate": 1.3230847362502172e-05,
      "loss": 4.2011,
      "step": 10241
    },
    {
      "epoch": 0.4134840532902705,
      "grad_norm": 239.1122283935547,
      "learning_rate": 1.3229609879975952e-05,
      "loss": 5.4808,
      "step": 10242
    },
    {
      "epoch": 0.4135244247073072,
      "grad_norm": 135.2839813232422,
      "learning_rate": 1.3228372342230944e-05,
      "loss": 5.3096,
      "step": 10243
    },
    {
      "epoch": 0.413564796124344,
      "grad_norm": 100.40277099609375,
      "learning_rate": 1.3227134749288307e-05,
      "loss": 4.4625,
      "step": 10244
    },
    {
      "epoch": 0.4136051675413807,
      "grad_norm": 86.2430419921875,
      "learning_rate": 1.32258971011692e-05,
      "loss": 5.0564,
      "step": 10245
    },
    {
      "epoch": 0.41364553895841744,
      "grad_norm": 143.55494689941406,
      "learning_rate": 1.3224659397894788e-05,
      "loss": 4.9291,
      "step": 10246
    },
    {
      "epoch": 0.41368591037545416,
      "grad_norm": 111.17201232910156,
      "learning_rate": 1.3223421639486231e-05,
      "loss": 4.073,
      "step": 10247
    },
    {
      "epoch": 0.41372628179249094,
      "grad_norm": 141.06097412109375,
      "learning_rate": 1.3222183825964689e-05,
      "loss": 4.4505,
      "step": 10248
    },
    {
      "epoch": 0.41376665320952766,
      "grad_norm": 74.82798767089844,
      "learning_rate": 1.322094595735133e-05,
      "loss": 4.8589,
      "step": 10249
    },
    {
      "epoch": 0.4138070246265644,
      "grad_norm": 99.38968658447266,
      "learning_rate": 1.3219708033667315e-05,
      "loss": 5.2764,
      "step": 10250
    },
    {
      "epoch": 0.41384739604360116,
      "grad_norm": 104.54359436035156,
      "learning_rate": 1.3218470054933814e-05,
      "loss": 6.6054,
      "step": 10251
    },
    {
      "epoch": 0.4138877674606379,
      "grad_norm": 112.95906829833984,
      "learning_rate": 1.3217232021171987e-05,
      "loss": 5.2496,
      "step": 10252
    },
    {
      "epoch": 0.4139281388776746,
      "grad_norm": 116.2467269897461,
      "learning_rate": 1.3215993932403009e-05,
      "loss": 5.2974,
      "step": 10253
    },
    {
      "epoch": 0.4139685102947113,
      "grad_norm": 95.05587005615234,
      "learning_rate": 1.3214755788648041e-05,
      "loss": 4.5476,
      "step": 10254
    },
    {
      "epoch": 0.4140088817117481,
      "grad_norm": 137.7049102783203,
      "learning_rate": 1.3213517589928256e-05,
      "loss": 4.1454,
      "step": 10255
    },
    {
      "epoch": 0.4140492531287848,
      "grad_norm": 82.08560180664062,
      "learning_rate": 1.3212279336264824e-05,
      "loss": 5.0639,
      "step": 10256
    },
    {
      "epoch": 0.41408962454582154,
      "grad_norm": 141.9514617919922,
      "learning_rate": 1.321104102767892e-05,
      "loss": 7.746,
      "step": 10257
    },
    {
      "epoch": 0.4141299959628583,
      "grad_norm": 423.1868591308594,
      "learning_rate": 1.3209802664191711e-05,
      "loss": 6.3199,
      "step": 10258
    },
    {
      "epoch": 0.41417036737989504,
      "grad_norm": 102.86817169189453,
      "learning_rate": 1.3208564245824373e-05,
      "loss": 5.3954,
      "step": 10259
    },
    {
      "epoch": 0.41421073879693177,
      "grad_norm": 96.8375244140625,
      "learning_rate": 1.3207325772598077e-05,
      "loss": 5.7486,
      "step": 10260
    },
    {
      "epoch": 0.4142511102139685,
      "grad_norm": 144.01834106445312,
      "learning_rate": 1.3206087244534002e-05,
      "loss": 4.5138,
      "step": 10261
    },
    {
      "epoch": 0.41429148163100527,
      "grad_norm": 175.28663635253906,
      "learning_rate": 1.320484866165332e-05,
      "loss": 4.0469,
      "step": 10262
    },
    {
      "epoch": 0.414331853048042,
      "grad_norm": 82.94536590576172,
      "learning_rate": 1.3203610023977211e-05,
      "loss": 3.8814,
      "step": 10263
    },
    {
      "epoch": 0.4143722244650787,
      "grad_norm": 129.82289123535156,
      "learning_rate": 1.320237133152685e-05,
      "loss": 3.9316,
      "step": 10264
    },
    {
      "epoch": 0.4144125958821155,
      "grad_norm": 90.41825866699219,
      "learning_rate": 1.3201132584323418e-05,
      "loss": 3.4962,
      "step": 10265
    },
    {
      "epoch": 0.4144529672991522,
      "grad_norm": 94.19328308105469,
      "learning_rate": 1.3199893782388094e-05,
      "loss": 6.8385,
      "step": 10266
    },
    {
      "epoch": 0.41449333871618893,
      "grad_norm": 110.88924407958984,
      "learning_rate": 1.3198654925742061e-05,
      "loss": 5.3018,
      "step": 10267
    },
    {
      "epoch": 0.41453371013322565,
      "grad_norm": 94.31148529052734,
      "learning_rate": 1.3197416014406494e-05,
      "loss": 4.3783,
      "step": 10268
    },
    {
      "epoch": 0.41457408155026243,
      "grad_norm": 78.80812072753906,
      "learning_rate": 1.3196177048402583e-05,
      "loss": 4.6426,
      "step": 10269
    },
    {
      "epoch": 0.41461445296729915,
      "grad_norm": 84.43794250488281,
      "learning_rate": 1.3194938027751505e-05,
      "loss": 5.5711,
      "step": 10270
    },
    {
      "epoch": 0.4146548243843359,
      "grad_norm": 83.3099594116211,
      "learning_rate": 1.3193698952474451e-05,
      "loss": 4.0955,
      "step": 10271
    },
    {
      "epoch": 0.41469519580137265,
      "grad_norm": 104.5589599609375,
      "learning_rate": 1.31924598225926e-05,
      "loss": 5.2866,
      "step": 10272
    },
    {
      "epoch": 0.4147355672184094,
      "grad_norm": 137.8142547607422,
      "learning_rate": 1.3191220638127144e-05,
      "loss": 4.4015,
      "step": 10273
    },
    {
      "epoch": 0.4147759386354461,
      "grad_norm": 81.32381439208984,
      "learning_rate": 1.3189981399099264e-05,
      "loss": 5.5704,
      "step": 10274
    },
    {
      "epoch": 0.4148163100524828,
      "grad_norm": 125.45112609863281,
      "learning_rate": 1.318874210553015e-05,
      "loss": 6.2528,
      "step": 10275
    },
    {
      "epoch": 0.4148566814695196,
      "grad_norm": 151.84317016601562,
      "learning_rate": 1.3187502757440996e-05,
      "loss": 4.3324,
      "step": 10276
    },
    {
      "epoch": 0.4148970528865563,
      "grad_norm": 117.21306610107422,
      "learning_rate": 1.3186263354852986e-05,
      "loss": 5.5721,
      "step": 10277
    },
    {
      "epoch": 0.41493742430359304,
      "grad_norm": 72.19486999511719,
      "learning_rate": 1.3185023897787312e-05,
      "loss": 5.3692,
      "step": 10278
    },
    {
      "epoch": 0.4149777957206298,
      "grad_norm": 104.79869842529297,
      "learning_rate": 1.3183784386265171e-05,
      "loss": 4.757,
      "step": 10279
    },
    {
      "epoch": 0.41501816713766654,
      "grad_norm": 103.7439193725586,
      "learning_rate": 1.318254482030775e-05,
      "loss": 6.1467,
      "step": 10280
    },
    {
      "epoch": 0.41505853855470326,
      "grad_norm": 172.57444763183594,
      "learning_rate": 1.3181305199936243e-05,
      "loss": 4.9015,
      "step": 10281
    },
    {
      "epoch": 0.41509890997174,
      "grad_norm": 105.8577880859375,
      "learning_rate": 1.3180065525171846e-05,
      "loss": 4.8049,
      "step": 10282
    },
    {
      "epoch": 0.41513928138877676,
      "grad_norm": 118.05464172363281,
      "learning_rate": 1.3178825796035757e-05,
      "loss": 5.0373,
      "step": 10283
    },
    {
      "epoch": 0.4151796528058135,
      "grad_norm": 140.5039825439453,
      "learning_rate": 1.3177586012549167e-05,
      "loss": 6.4515,
      "step": 10284
    },
    {
      "epoch": 0.4152200242228502,
      "grad_norm": 63.96825408935547,
      "learning_rate": 1.3176346174733277e-05,
      "loss": 5.9308,
      "step": 10285
    },
    {
      "epoch": 0.415260395639887,
      "grad_norm": 108.78028106689453,
      "learning_rate": 1.3175106282609285e-05,
      "loss": 5.3579,
      "step": 10286
    },
    {
      "epoch": 0.4153007670569237,
      "grad_norm": 75.74388885498047,
      "learning_rate": 1.3173866336198391e-05,
      "loss": 6.0856,
      "step": 10287
    },
    {
      "epoch": 0.4153411384739604,
      "grad_norm": 91.40930938720703,
      "learning_rate": 1.3172626335521793e-05,
      "loss": 5.895,
      "step": 10288
    },
    {
      "epoch": 0.41538150989099715,
      "grad_norm": 79.82305145263672,
      "learning_rate": 1.3171386280600695e-05,
      "loss": 4.1161,
      "step": 10289
    },
    {
      "epoch": 0.4154218813080339,
      "grad_norm": 82.30936431884766,
      "learning_rate": 1.3170146171456294e-05,
      "loss": 5.7084,
      "step": 10290
    },
    {
      "epoch": 0.41546225272507065,
      "grad_norm": 68.35717010498047,
      "learning_rate": 1.31689060081098e-05,
      "loss": 5.2418,
      "step": 10291
    },
    {
      "epoch": 0.41550262414210737,
      "grad_norm": 78.01203155517578,
      "learning_rate": 1.3167665790582411e-05,
      "loss": 3.3862,
      "step": 10292
    },
    {
      "epoch": 0.41554299555914415,
      "grad_norm": 117.70027923583984,
      "learning_rate": 1.3166425518895336e-05,
      "loss": 5.7773,
      "step": 10293
    },
    {
      "epoch": 0.41558336697618087,
      "grad_norm": 134.2660675048828,
      "learning_rate": 1.3165185193069776e-05,
      "loss": 4.0762,
      "step": 10294
    },
    {
      "epoch": 0.4156237383932176,
      "grad_norm": 91.3561019897461,
      "learning_rate": 1.3163944813126942e-05,
      "loss": 4.887,
      "step": 10295
    },
    {
      "epoch": 0.41566410981025437,
      "grad_norm": 170.96287536621094,
      "learning_rate": 1.3162704379088038e-05,
      "loss": 4.6866,
      "step": 10296
    },
    {
      "epoch": 0.4157044812272911,
      "grad_norm": 100.17198944091797,
      "learning_rate": 1.3161463890974277e-05,
      "loss": 5.98,
      "step": 10297
    },
    {
      "epoch": 0.4157448526443278,
      "grad_norm": 78.65093994140625,
      "learning_rate": 1.3160223348806864e-05,
      "loss": 4.6235,
      "step": 10298
    },
    {
      "epoch": 0.41578522406136453,
      "grad_norm": 110.65062713623047,
      "learning_rate": 1.3158982752607015e-05,
      "loss": 5.3084,
      "step": 10299
    },
    {
      "epoch": 0.4158255954784013,
      "grad_norm": 93.1705551147461,
      "learning_rate": 1.3157742102395934e-05,
      "loss": 4.5018,
      "step": 10300
    },
    {
      "epoch": 0.41586596689543803,
      "grad_norm": 152.4884796142578,
      "learning_rate": 1.3156501398194838e-05,
      "loss": 6.2684,
      "step": 10301
    },
    {
      "epoch": 0.41590633831247475,
      "grad_norm": 103.79059600830078,
      "learning_rate": 1.315526064002494e-05,
      "loss": 4.801,
      "step": 10302
    },
    {
      "epoch": 0.41594670972951153,
      "grad_norm": 117.9211196899414,
      "learning_rate": 1.3154019827907453e-05,
      "loss": 5.638,
      "step": 10303
    },
    {
      "epoch": 0.41598708114654825,
      "grad_norm": 93.39141082763672,
      "learning_rate": 1.3152778961863593e-05,
      "loss": 5.3556,
      "step": 10304
    },
    {
      "epoch": 0.416027452563585,
      "grad_norm": 78.63468933105469,
      "learning_rate": 1.3151538041914572e-05,
      "loss": 4.785,
      "step": 10305
    },
    {
      "epoch": 0.4160678239806217,
      "grad_norm": 91.41194152832031,
      "learning_rate": 1.315029706808161e-05,
      "loss": 5.5511,
      "step": 10306
    },
    {
      "epoch": 0.4161081953976585,
      "grad_norm": 88.16053009033203,
      "learning_rate": 1.3149056040385928e-05,
      "loss": 4.5581,
      "step": 10307
    },
    {
      "epoch": 0.4161485668146952,
      "grad_norm": 109.68169403076172,
      "learning_rate": 1.3147814958848736e-05,
      "loss": 5.2308,
      "step": 10308
    },
    {
      "epoch": 0.4161889382317319,
      "grad_norm": 89.44666290283203,
      "learning_rate": 1.3146573823491267e-05,
      "loss": 4.5842,
      "step": 10309
    },
    {
      "epoch": 0.4162293096487687,
      "grad_norm": 71.95919036865234,
      "learning_rate": 1.3145332634334726e-05,
      "loss": 6.6057,
      "step": 10310
    },
    {
      "epoch": 0.4162696810658054,
      "grad_norm": 89.42508697509766,
      "learning_rate": 1.3144091391400346e-05,
      "loss": 5.1718,
      "step": 10311
    },
    {
      "epoch": 0.41631005248284214,
      "grad_norm": 93.72483825683594,
      "learning_rate": 1.3142850094709343e-05,
      "loss": 4.1105,
      "step": 10312
    },
    {
      "epoch": 0.41635042389987886,
      "grad_norm": 108.94319915771484,
      "learning_rate": 1.3141608744282943e-05,
      "loss": 6.1526,
      "step": 10313
    },
    {
      "epoch": 0.41639079531691564,
      "grad_norm": 105.88538360595703,
      "learning_rate": 1.314036734014237e-05,
      "loss": 4.8109,
      "step": 10314
    },
    {
      "epoch": 0.41643116673395236,
      "grad_norm": 77.01673889160156,
      "learning_rate": 1.3139125882308847e-05,
      "loss": 4.9708,
      "step": 10315
    },
    {
      "epoch": 0.4164715381509891,
      "grad_norm": 90.09730529785156,
      "learning_rate": 1.3137884370803602e-05,
      "loss": 5.7454,
      "step": 10316
    },
    {
      "epoch": 0.41651190956802586,
      "grad_norm": 227.15087890625,
      "learning_rate": 1.3136642805647865e-05,
      "loss": 4.4463,
      "step": 10317
    },
    {
      "epoch": 0.4165522809850626,
      "grad_norm": 113.37831115722656,
      "learning_rate": 1.313540118686286e-05,
      "loss": 4.8096,
      "step": 10318
    },
    {
      "epoch": 0.4165926524020993,
      "grad_norm": 78.29096984863281,
      "learning_rate": 1.3134159514469816e-05,
      "loss": 4.904,
      "step": 10319
    },
    {
      "epoch": 0.41663302381913603,
      "grad_norm": 115.12047576904297,
      "learning_rate": 1.3132917788489961e-05,
      "loss": 5.5076,
      "step": 10320
    },
    {
      "epoch": 0.4166733952361728,
      "grad_norm": 83.456787109375,
      "learning_rate": 1.313167600894453e-05,
      "loss": 6.781,
      "step": 10321
    },
    {
      "epoch": 0.4167137666532095,
      "grad_norm": 108.36117553710938,
      "learning_rate": 1.313043417585475e-05,
      "loss": 5.9834,
      "step": 10322
    },
    {
      "epoch": 0.41675413807024625,
      "grad_norm": 125.3142318725586,
      "learning_rate": 1.3129192289241854e-05,
      "loss": 3.6805,
      "step": 10323
    },
    {
      "epoch": 0.416794509487283,
      "grad_norm": 153.2333221435547,
      "learning_rate": 1.312795034912708e-05,
      "loss": 5.7846,
      "step": 10324
    },
    {
      "epoch": 0.41683488090431975,
      "grad_norm": 79.89454650878906,
      "learning_rate": 1.312670835553166e-05,
      "loss": 5.1077,
      "step": 10325
    },
    {
      "epoch": 0.41687525232135647,
      "grad_norm": 87.62801361083984,
      "learning_rate": 1.3125466308476828e-05,
      "loss": 4.5114,
      "step": 10326
    },
    {
      "epoch": 0.4169156237383932,
      "grad_norm": 122.38023376464844,
      "learning_rate": 1.312422420798382e-05,
      "loss": 5.231,
      "step": 10327
    },
    {
      "epoch": 0.41695599515542997,
      "grad_norm": 129.38555908203125,
      "learning_rate": 1.3122982054073874e-05,
      "loss": 6.6599,
      "step": 10328
    },
    {
      "epoch": 0.4169963665724667,
      "grad_norm": 103.54667663574219,
      "learning_rate": 1.312173984676823e-05,
      "loss": 3.809,
      "step": 10329
    },
    {
      "epoch": 0.4170367379895034,
      "grad_norm": 88.93049621582031,
      "learning_rate": 1.312049758608812e-05,
      "loss": 4.5387,
      "step": 10330
    },
    {
      "epoch": 0.4170771094065402,
      "grad_norm": 82.41105651855469,
      "learning_rate": 1.3119255272054792e-05,
      "loss": 5.9888,
      "step": 10331
    },
    {
      "epoch": 0.4171174808235769,
      "grad_norm": 151.87030029296875,
      "learning_rate": 1.3118012904689479e-05,
      "loss": 7.127,
      "step": 10332
    },
    {
      "epoch": 0.41715785224061364,
      "grad_norm": 100.5879898071289,
      "learning_rate": 1.3116770484013428e-05,
      "loss": 4.7093,
      "step": 10333
    },
    {
      "epoch": 0.41719822365765036,
      "grad_norm": 78.0118179321289,
      "learning_rate": 1.3115528010047877e-05,
      "loss": 6.0086,
      "step": 10334
    },
    {
      "epoch": 0.41723859507468714,
      "grad_norm": 166.5762939453125,
      "learning_rate": 1.3114285482814077e-05,
      "loss": 4.4363,
      "step": 10335
    },
    {
      "epoch": 0.41727896649172386,
      "grad_norm": 60.772403717041016,
      "learning_rate": 1.3113042902333266e-05,
      "loss": 5.4171,
      "step": 10336
    },
    {
      "epoch": 0.4173193379087606,
      "grad_norm": 300.6165771484375,
      "learning_rate": 1.3111800268626693e-05,
      "loss": 3.7881,
      "step": 10337
    },
    {
      "epoch": 0.41735970932579736,
      "grad_norm": 87.16796112060547,
      "learning_rate": 1.31105575817156e-05,
      "loss": 5.7705,
      "step": 10338
    },
    {
      "epoch": 0.4174000807428341,
      "grad_norm": 160.4389190673828,
      "learning_rate": 1.3109314841621237e-05,
      "loss": 6.4169,
      "step": 10339
    },
    {
      "epoch": 0.4174404521598708,
      "grad_norm": 87.7250747680664,
      "learning_rate": 1.3108072048364849e-05,
      "loss": 5.3482,
      "step": 10340
    },
    {
      "epoch": 0.4174808235769075,
      "grad_norm": 152.90478515625,
      "learning_rate": 1.3106829201967689e-05,
      "loss": 4.4069,
      "step": 10341
    },
    {
      "epoch": 0.4175211949939443,
      "grad_norm": 87.28775024414062,
      "learning_rate": 1.3105586302451002e-05,
      "loss": 4.5308,
      "step": 10342
    },
    {
      "epoch": 0.417561566410981,
      "grad_norm": 65.74461364746094,
      "learning_rate": 1.3104343349836042e-05,
      "loss": 5.3533,
      "step": 10343
    },
    {
      "epoch": 0.41760193782801774,
      "grad_norm": 107.93687438964844,
      "learning_rate": 1.310310034414406e-05,
      "loss": 6.1563,
      "step": 10344
    },
    {
      "epoch": 0.4176423092450545,
      "grad_norm": 84.84435272216797,
      "learning_rate": 1.3101857285396311e-05,
      "loss": 6.3486,
      "step": 10345
    },
    {
      "epoch": 0.41768268066209124,
      "grad_norm": 120.47689819335938,
      "learning_rate": 1.310061417361404e-05,
      "loss": 6.364,
      "step": 10346
    },
    {
      "epoch": 0.41772305207912797,
      "grad_norm": 76.39685821533203,
      "learning_rate": 1.3099371008818514e-05,
      "loss": 5.5475,
      "step": 10347
    },
    {
      "epoch": 0.41776342349616474,
      "grad_norm": 116.43647766113281,
      "learning_rate": 1.3098127791030977e-05,
      "loss": 4.686,
      "step": 10348
    },
    {
      "epoch": 0.41780379491320147,
      "grad_norm": 95.71595764160156,
      "learning_rate": 1.3096884520272693e-05,
      "loss": 6.7256,
      "step": 10349
    },
    {
      "epoch": 0.4178441663302382,
      "grad_norm": 108.0133056640625,
      "learning_rate": 1.309564119656491e-05,
      "loss": 5.1271,
      "step": 10350
    },
    {
      "epoch": 0.4178845377472749,
      "grad_norm": 78.27691650390625,
      "learning_rate": 1.3094397819928897e-05,
      "loss": 3.6745,
      "step": 10351
    },
    {
      "epoch": 0.4179249091643117,
      "grad_norm": 210.3724822998047,
      "learning_rate": 1.3093154390385902e-05,
      "loss": 5.5227,
      "step": 10352
    },
    {
      "epoch": 0.4179652805813484,
      "grad_norm": 133.36764526367188,
      "learning_rate": 1.3091910907957193e-05,
      "loss": 5.736,
      "step": 10353
    },
    {
      "epoch": 0.41800565199838513,
      "grad_norm": 113.9085464477539,
      "learning_rate": 1.3090667372664027e-05,
      "loss": 4.2307,
      "step": 10354
    },
    {
      "epoch": 0.4180460234154219,
      "grad_norm": 100.37902069091797,
      "learning_rate": 1.3089423784527665e-05,
      "loss": 3.4691,
      "step": 10355
    },
    {
      "epoch": 0.41808639483245863,
      "grad_norm": 160.03211975097656,
      "learning_rate": 1.3088180143569374e-05,
      "loss": 6.9785,
      "step": 10356
    },
    {
      "epoch": 0.41812676624949535,
      "grad_norm": 193.4099884033203,
      "learning_rate": 1.308693644981041e-05,
      "loss": 4.326,
      "step": 10357
    },
    {
      "epoch": 0.4181671376665321,
      "grad_norm": 103.7607650756836,
      "learning_rate": 1.3085692703272044e-05,
      "loss": 5.6786,
      "step": 10358
    },
    {
      "epoch": 0.41820750908356885,
      "grad_norm": 112.20682525634766,
      "learning_rate": 1.3084448903975538e-05,
      "loss": 6.8945,
      "step": 10359
    },
    {
      "epoch": 0.4182478805006056,
      "grad_norm": 111.11225891113281,
      "learning_rate": 1.3083205051942158e-05,
      "loss": 5.5595,
      "step": 10360
    },
    {
      "epoch": 0.4182882519176423,
      "grad_norm": 77.09156036376953,
      "learning_rate": 1.308196114719317e-05,
      "loss": 3.8549,
      "step": 10361
    },
    {
      "epoch": 0.4183286233346791,
      "grad_norm": 254.37796020507812,
      "learning_rate": 1.3080717189749847e-05,
      "loss": 5.8995,
      "step": 10362
    },
    {
      "epoch": 0.4183689947517158,
      "grad_norm": 108.78751373291016,
      "learning_rate": 1.307947317963345e-05,
      "loss": 5.0486,
      "step": 10363
    },
    {
      "epoch": 0.4184093661687525,
      "grad_norm": 120.85860443115234,
      "learning_rate": 1.3078229116865253e-05,
      "loss": 3.32,
      "step": 10364
    },
    {
      "epoch": 0.41844973758578924,
      "grad_norm": 97.62959289550781,
      "learning_rate": 1.3076985001466528e-05,
      "loss": 5.7958,
      "step": 10365
    },
    {
      "epoch": 0.418490109002826,
      "grad_norm": 87.70549774169922,
      "learning_rate": 1.3075740833458545e-05,
      "loss": 4.9151,
      "step": 10366
    },
    {
      "epoch": 0.41853048041986274,
      "grad_norm": 73.74418640136719,
      "learning_rate": 1.3074496612862575e-05,
      "loss": 2.7466,
      "step": 10367
    },
    {
      "epoch": 0.41857085183689946,
      "grad_norm": 142.71913146972656,
      "learning_rate": 1.3073252339699893e-05,
      "loss": 6.5601,
      "step": 10368
    },
    {
      "epoch": 0.41861122325393624,
      "grad_norm": 97.62605285644531,
      "learning_rate": 1.3072008013991773e-05,
      "loss": 5.33,
      "step": 10369
    },
    {
      "epoch": 0.41865159467097296,
      "grad_norm": 107.672607421875,
      "learning_rate": 1.3070763635759488e-05,
      "loss": 4.2828,
      "step": 10370
    },
    {
      "epoch": 0.4186919660880097,
      "grad_norm": 147.6422119140625,
      "learning_rate": 1.306951920502432e-05,
      "loss": 4.4403,
      "step": 10371
    },
    {
      "epoch": 0.4187323375050464,
      "grad_norm": 104.66436004638672,
      "learning_rate": 1.306827472180754e-05,
      "loss": 5.5977,
      "step": 10372
    },
    {
      "epoch": 0.4187727089220832,
      "grad_norm": 74.93580627441406,
      "learning_rate": 1.3067030186130422e-05,
      "loss": 6.298,
      "step": 10373
    },
    {
      "epoch": 0.4188130803391199,
      "grad_norm": 71.17529296875,
      "learning_rate": 1.3065785598014253e-05,
      "loss": 5.7438,
      "step": 10374
    },
    {
      "epoch": 0.4188534517561566,
      "grad_norm": 142.14544677734375,
      "learning_rate": 1.306454095748031e-05,
      "loss": 5.8551,
      "step": 10375
    },
    {
      "epoch": 0.4188938231731934,
      "grad_norm": 87.98334503173828,
      "learning_rate": 1.3063296264549872e-05,
      "loss": 5.4618,
      "step": 10376
    },
    {
      "epoch": 0.4189341945902301,
      "grad_norm": 80.1598892211914,
      "learning_rate": 1.3062051519244221e-05,
      "loss": 5.9255,
      "step": 10377
    },
    {
      "epoch": 0.41897456600726685,
      "grad_norm": 85.70064544677734,
      "learning_rate": 1.306080672158464e-05,
      "loss": 5.6518,
      "step": 10378
    },
    {
      "epoch": 0.41901493742430357,
      "grad_norm": 98.03816986083984,
      "learning_rate": 1.3059561871592413e-05,
      "loss": 3.7966,
      "step": 10379
    },
    {
      "epoch": 0.41905530884134035,
      "grad_norm": 82.77728271484375,
      "learning_rate": 1.3058316969288823e-05,
      "loss": 5.9386,
      "step": 10380
    },
    {
      "epoch": 0.41909568025837707,
      "grad_norm": 109.64318084716797,
      "learning_rate": 1.3057072014695148e-05,
      "loss": 6.1844,
      "step": 10381
    },
    {
      "epoch": 0.4191360516754138,
      "grad_norm": 92.4883041381836,
      "learning_rate": 1.3055827007832688e-05,
      "loss": 5.7209,
      "step": 10382
    },
    {
      "epoch": 0.41917642309245057,
      "grad_norm": 65.97374725341797,
      "learning_rate": 1.3054581948722717e-05,
      "loss": 5.7364,
      "step": 10383
    },
    {
      "epoch": 0.4192167945094873,
      "grad_norm": 168.42269897460938,
      "learning_rate": 1.305333683738653e-05,
      "loss": 5.9411,
      "step": 10384
    },
    {
      "epoch": 0.419257165926524,
      "grad_norm": 151.35621643066406,
      "learning_rate": 1.3052091673845411e-05,
      "loss": 6.1937,
      "step": 10385
    },
    {
      "epoch": 0.41929753734356073,
      "grad_norm": 276.066162109375,
      "learning_rate": 1.3050846458120656e-05,
      "loss": 4.8666,
      "step": 10386
    },
    {
      "epoch": 0.4193379087605975,
      "grad_norm": 117.22528839111328,
      "learning_rate": 1.3049601190233548e-05,
      "loss": 5.0597,
      "step": 10387
    },
    {
      "epoch": 0.41937828017763423,
      "grad_norm": 115.63420867919922,
      "learning_rate": 1.3048355870205382e-05,
      "loss": 5.0985,
      "step": 10388
    },
    {
      "epoch": 0.41941865159467095,
      "grad_norm": 123.06935119628906,
      "learning_rate": 1.3047110498057446e-05,
      "loss": 4.6492,
      "step": 10389
    },
    {
      "epoch": 0.41945902301170773,
      "grad_norm": 62.547000885009766,
      "learning_rate": 1.304586507381104e-05,
      "loss": 4.3271,
      "step": 10390
    },
    {
      "epoch": 0.41949939442874445,
      "grad_norm": 133.93568420410156,
      "learning_rate": 1.3044619597487452e-05,
      "loss": 5.1637,
      "step": 10391
    },
    {
      "epoch": 0.4195397658457812,
      "grad_norm": 71.44550323486328,
      "learning_rate": 1.3043374069107978e-05,
      "loss": 5.6644,
      "step": 10392
    },
    {
      "epoch": 0.4195801372628179,
      "grad_norm": 116.26429748535156,
      "learning_rate": 1.3042128488693914e-05,
      "loss": 5.6471,
      "step": 10393
    },
    {
      "epoch": 0.4196205086798547,
      "grad_norm": 109.61864471435547,
      "learning_rate": 1.3040882856266557e-05,
      "loss": 3.3093,
      "step": 10394
    },
    {
      "epoch": 0.4196608800968914,
      "grad_norm": 86.15846252441406,
      "learning_rate": 1.3039637171847204e-05,
      "loss": 5.8621,
      "step": 10395
    },
    {
      "epoch": 0.4197012515139281,
      "grad_norm": 151.99227905273438,
      "learning_rate": 1.3038391435457155e-05,
      "loss": 4.3516,
      "step": 10396
    },
    {
      "epoch": 0.4197416229309649,
      "grad_norm": 199.56179809570312,
      "learning_rate": 1.3037145647117706e-05,
      "loss": 5.7623,
      "step": 10397
    },
    {
      "epoch": 0.4197819943480016,
      "grad_norm": 96.32431030273438,
      "learning_rate": 1.303589980685016e-05,
      "loss": 5.2902,
      "step": 10398
    },
    {
      "epoch": 0.41982236576503834,
      "grad_norm": 191.88710021972656,
      "learning_rate": 1.3034653914675816e-05,
      "loss": 5.7821,
      "step": 10399
    },
    {
      "epoch": 0.41986273718207506,
      "grad_norm": 64.80262756347656,
      "learning_rate": 1.3033407970615976e-05,
      "loss": 4.3816,
      "step": 10400
    },
    {
      "epoch": 0.41990310859911184,
      "grad_norm": 78.6489486694336,
      "learning_rate": 1.3032161974691944e-05,
      "loss": 5.1557,
      "step": 10401
    },
    {
      "epoch": 0.41994348001614856,
      "grad_norm": 115.24243927001953,
      "learning_rate": 1.3030915926925026e-05,
      "loss": 5.5484,
      "step": 10402
    },
    {
      "epoch": 0.4199838514331853,
      "grad_norm": 116.45526123046875,
      "learning_rate": 1.3029669827336518e-05,
      "loss": 5.7073,
      "step": 10403
    },
    {
      "epoch": 0.42002422285022206,
      "grad_norm": 80.56488037109375,
      "learning_rate": 1.302842367594773e-05,
      "loss": 5.4123,
      "step": 10404
    },
    {
      "epoch": 0.4200645942672588,
      "grad_norm": 104.33767700195312,
      "learning_rate": 1.3027177472779972e-05,
      "loss": 7.5782,
      "step": 10405
    },
    {
      "epoch": 0.4201049656842955,
      "grad_norm": 105.890869140625,
      "learning_rate": 1.302593121785455e-05,
      "loss": 5.4601,
      "step": 10406
    },
    {
      "epoch": 0.4201453371013323,
      "grad_norm": 113.68897247314453,
      "learning_rate": 1.3024684911192766e-05,
      "loss": 4.7106,
      "step": 10407
    },
    {
      "epoch": 0.420185708518369,
      "grad_norm": 117.05001831054688,
      "learning_rate": 1.3023438552815936e-05,
      "loss": 4.6961,
      "step": 10408
    },
    {
      "epoch": 0.4202260799354057,
      "grad_norm": 95.0750961303711,
      "learning_rate": 1.3022192142745367e-05,
      "loss": 5.654,
      "step": 10409
    },
    {
      "epoch": 0.42026645135244245,
      "grad_norm": 70.94902801513672,
      "learning_rate": 1.3020945681002371e-05,
      "loss": 5.7886,
      "step": 10410
    },
    {
      "epoch": 0.4203068227694792,
      "grad_norm": 96.4576416015625,
      "learning_rate": 1.3019699167608253e-05,
      "loss": 4.734,
      "step": 10411
    },
    {
      "epoch": 0.42034719418651595,
      "grad_norm": 151.9458770751953,
      "learning_rate": 1.3018452602584337e-05,
      "loss": 4.7801,
      "step": 10412
    },
    {
      "epoch": 0.42038756560355267,
      "grad_norm": 97.55258178710938,
      "learning_rate": 1.3017205985951926e-05,
      "loss": 4.5633,
      "step": 10413
    },
    {
      "epoch": 0.42042793702058945,
      "grad_norm": 119.09819793701172,
      "learning_rate": 1.3015959317732336e-05,
      "loss": 4.5418,
      "step": 10414
    },
    {
      "epoch": 0.42046830843762617,
      "grad_norm": 71.83363342285156,
      "learning_rate": 1.3014712597946889e-05,
      "loss": 4.0247,
      "step": 10415
    },
    {
      "epoch": 0.4205086798546629,
      "grad_norm": 81.99392700195312,
      "learning_rate": 1.3013465826616896e-05,
      "loss": 4.7701,
      "step": 10416
    },
    {
      "epoch": 0.4205490512716996,
      "grad_norm": 71.24797821044922,
      "learning_rate": 1.3012219003763673e-05,
      "loss": 6.5631,
      "step": 10417
    },
    {
      "epoch": 0.4205894226887364,
      "grad_norm": 80.79032135009766,
      "learning_rate": 1.301097212940854e-05,
      "loss": 4.2989,
      "step": 10418
    },
    {
      "epoch": 0.4206297941057731,
      "grad_norm": 138.03445434570312,
      "learning_rate": 1.3009725203572814e-05,
      "loss": 4.4692,
      "step": 10419
    },
    {
      "epoch": 0.42067016552280984,
      "grad_norm": 114.97474670410156,
      "learning_rate": 1.3008478226277817e-05,
      "loss": 5.7703,
      "step": 10420
    },
    {
      "epoch": 0.4207105369398466,
      "grad_norm": 94.70630645751953,
      "learning_rate": 1.3007231197544864e-05,
      "loss": 5.4397,
      "step": 10421
    },
    {
      "epoch": 0.42075090835688334,
      "grad_norm": 384.4236755371094,
      "learning_rate": 1.3005984117395284e-05,
      "loss": 4.5237,
      "step": 10422
    },
    {
      "epoch": 0.42079127977392006,
      "grad_norm": 113.48348999023438,
      "learning_rate": 1.3004736985850393e-05,
      "loss": 4.6034,
      "step": 10423
    },
    {
      "epoch": 0.4208316511909568,
      "grad_norm": 146.12794494628906,
      "learning_rate": 1.3003489802931514e-05,
      "loss": 5.0427,
      "step": 10424
    },
    {
      "epoch": 0.42087202260799356,
      "grad_norm": 156.29168701171875,
      "learning_rate": 1.3002242568659976e-05,
      "loss": 3.5773,
      "step": 10425
    },
    {
      "epoch": 0.4209123940250303,
      "grad_norm": 115.91149139404297,
      "learning_rate": 1.30009952830571e-05,
      "loss": 6.0893,
      "step": 10426
    },
    {
      "epoch": 0.420952765442067,
      "grad_norm": 91.53362274169922,
      "learning_rate": 1.2999747946144212e-05,
      "loss": 5.0901,
      "step": 10427
    },
    {
      "epoch": 0.4209931368591038,
      "grad_norm": 93.8650894165039,
      "learning_rate": 1.2998500557942641e-05,
      "loss": 5.1528,
      "step": 10428
    },
    {
      "epoch": 0.4210335082761405,
      "grad_norm": 83.28546905517578,
      "learning_rate": 1.2997253118473709e-05,
      "loss": 4.4322,
      "step": 10429
    },
    {
      "epoch": 0.4210738796931772,
      "grad_norm": 75.28313446044922,
      "learning_rate": 1.2996005627758753e-05,
      "loss": 5.3706,
      "step": 10430
    },
    {
      "epoch": 0.42111425111021394,
      "grad_norm": 101.77680206298828,
      "learning_rate": 1.2994758085819093e-05,
      "loss": 3.8312,
      "step": 10431
    },
    {
      "epoch": 0.4211546225272507,
      "grad_norm": 79.14945220947266,
      "learning_rate": 1.2993510492676062e-05,
      "loss": 5.8708,
      "step": 10432
    },
    {
      "epoch": 0.42119499394428744,
      "grad_norm": 104.89738464355469,
      "learning_rate": 1.2992262848350992e-05,
      "loss": 4.3048,
      "step": 10433
    },
    {
      "epoch": 0.42123536536132417,
      "grad_norm": 63.279296875,
      "learning_rate": 1.2991015152865218e-05,
      "loss": 3.3613,
      "step": 10434
    },
    {
      "epoch": 0.42127573677836094,
      "grad_norm": 90.90534210205078,
      "learning_rate": 1.2989767406240068e-05,
      "loss": 5.8137,
      "step": 10435
    },
    {
      "epoch": 0.42131610819539766,
      "grad_norm": 85.7756576538086,
      "learning_rate": 1.2988519608496877e-05,
      "loss": 4.5815,
      "step": 10436
    },
    {
      "epoch": 0.4213564796124344,
      "grad_norm": 103.64615631103516,
      "learning_rate": 1.2987271759656978e-05,
      "loss": 4.707,
      "step": 10437
    },
    {
      "epoch": 0.4213968510294711,
      "grad_norm": 68.87224578857422,
      "learning_rate": 1.2986023859741711e-05,
      "loss": 4.9382,
      "step": 10438
    },
    {
      "epoch": 0.4214372224465079,
      "grad_norm": 121.33513641357422,
      "learning_rate": 1.2984775908772406e-05,
      "loss": 6.3966,
      "step": 10439
    },
    {
      "epoch": 0.4214775938635446,
      "grad_norm": 108.41507720947266,
      "learning_rate": 1.2983527906770404e-05,
      "loss": 6.2499,
      "step": 10440
    },
    {
      "epoch": 0.42151796528058133,
      "grad_norm": 118.62973022460938,
      "learning_rate": 1.298227985375704e-05,
      "loss": 6.4837,
      "step": 10441
    },
    {
      "epoch": 0.4215583366976181,
      "grad_norm": 97.84812927246094,
      "learning_rate": 1.2981031749753656e-05,
      "loss": 4.8994,
      "step": 10442
    },
    {
      "epoch": 0.42159870811465483,
      "grad_norm": 108.0849609375,
      "learning_rate": 1.297978359478159e-05,
      "loss": 3.7439,
      "step": 10443
    },
    {
      "epoch": 0.42163907953169155,
      "grad_norm": 107.45210266113281,
      "learning_rate": 1.2978535388862184e-05,
      "loss": 3.4745,
      "step": 10444
    },
    {
      "epoch": 0.4216794509487283,
      "grad_norm": 108.3401870727539,
      "learning_rate": 1.2977287132016776e-05,
      "loss": 4.9445,
      "step": 10445
    },
    {
      "epoch": 0.42171982236576505,
      "grad_norm": 81.33787536621094,
      "learning_rate": 1.2976038824266714e-05,
      "loss": 4.0933,
      "step": 10446
    },
    {
      "epoch": 0.4217601937828018,
      "grad_norm": 113.91693878173828,
      "learning_rate": 1.2974790465633336e-05,
      "loss": 5.2383,
      "step": 10447
    },
    {
      "epoch": 0.4218005651998385,
      "grad_norm": 81.22576904296875,
      "learning_rate": 1.2973542056137988e-05,
      "loss": 5.6332,
      "step": 10448
    },
    {
      "epoch": 0.4218409366168753,
      "grad_norm": 86.6356201171875,
      "learning_rate": 1.2972293595802013e-05,
      "loss": 5.3163,
      "step": 10449
    },
    {
      "epoch": 0.421881308033912,
      "grad_norm": 102.20127868652344,
      "learning_rate": 1.297104508464676e-05,
      "loss": 5.2793,
      "step": 10450
    },
    {
      "epoch": 0.4219216794509487,
      "grad_norm": 65.74813079833984,
      "learning_rate": 1.2969796522693572e-05,
      "loss": 3.7875,
      "step": 10451
    },
    {
      "epoch": 0.42196205086798544,
      "grad_norm": 93.29916381835938,
      "learning_rate": 1.2968547909963799e-05,
      "loss": 7.1978,
      "step": 10452
    },
    {
      "epoch": 0.4220024222850222,
      "grad_norm": 117.27104187011719,
      "learning_rate": 1.2967299246478789e-05,
      "loss": 5.231,
      "step": 10453
    },
    {
      "epoch": 0.42204279370205894,
      "grad_norm": 144.15982055664062,
      "learning_rate": 1.2966050532259893e-05,
      "loss": 5.3606,
      "step": 10454
    },
    {
      "epoch": 0.42208316511909566,
      "grad_norm": 85.31808471679688,
      "learning_rate": 1.2964801767328456e-05,
      "loss": 5.6097,
      "step": 10455
    },
    {
      "epoch": 0.42212353653613244,
      "grad_norm": 99.91537475585938,
      "learning_rate": 1.2963552951705834e-05,
      "loss": 5.6725,
      "step": 10456
    },
    {
      "epoch": 0.42216390795316916,
      "grad_norm": 98.73654174804688,
      "learning_rate": 1.2962304085413378e-05,
      "loss": 6.2792,
      "step": 10457
    },
    {
      "epoch": 0.4222042793702059,
      "grad_norm": 81.6457290649414,
      "learning_rate": 1.2961055168472438e-05,
      "loss": 4.7692,
      "step": 10458
    },
    {
      "epoch": 0.42224465078724266,
      "grad_norm": 70.17383575439453,
      "learning_rate": 1.2959806200904367e-05,
      "loss": 5.4301,
      "step": 10459
    },
    {
      "epoch": 0.4222850222042794,
      "grad_norm": 90.64944458007812,
      "learning_rate": 1.2958557182730527e-05,
      "loss": 5.8821,
      "step": 10460
    },
    {
      "epoch": 0.4223253936213161,
      "grad_norm": 87.30228424072266,
      "learning_rate": 1.2957308113972263e-05,
      "loss": 5.0008,
      "step": 10461
    },
    {
      "epoch": 0.4223657650383528,
      "grad_norm": 92.12320709228516,
      "learning_rate": 1.2956058994650937e-05,
      "loss": 5.2432,
      "step": 10462
    },
    {
      "epoch": 0.4224061364553896,
      "grad_norm": 137.69390869140625,
      "learning_rate": 1.2954809824787903e-05,
      "loss": 4.277,
      "step": 10463
    },
    {
      "epoch": 0.4224465078724263,
      "grad_norm": 103.87477111816406,
      "learning_rate": 1.2953560604404527e-05,
      "loss": 4.1451,
      "step": 10464
    },
    {
      "epoch": 0.42248687928946305,
      "grad_norm": 543.7496337890625,
      "learning_rate": 1.2952311333522159e-05,
      "loss": 5.5605,
      "step": 10465
    },
    {
      "epoch": 0.4225272507064998,
      "grad_norm": 166.43194580078125,
      "learning_rate": 1.2951062012162161e-05,
      "loss": 5.4007,
      "step": 10466
    },
    {
      "epoch": 0.42256762212353655,
      "grad_norm": 141.2709197998047,
      "learning_rate": 1.2949812640345894e-05,
      "loss": 4.9084,
      "step": 10467
    },
    {
      "epoch": 0.42260799354057327,
      "grad_norm": 124.8163833618164,
      "learning_rate": 1.2948563218094718e-05,
      "loss": 4.8514,
      "step": 10468
    },
    {
      "epoch": 0.42264836495761,
      "grad_norm": 112.13785552978516,
      "learning_rate": 1.294731374543e-05,
      "loss": 6.4302,
      "step": 10469
    },
    {
      "epoch": 0.42268873637464677,
      "grad_norm": 88.80976104736328,
      "learning_rate": 1.2946064222373096e-05,
      "loss": 5.9362,
      "step": 10470
    },
    {
      "epoch": 0.4227291077916835,
      "grad_norm": 77.24308776855469,
      "learning_rate": 1.2944814648945373e-05,
      "loss": 5.5538,
      "step": 10471
    },
    {
      "epoch": 0.4227694792087202,
      "grad_norm": 115.6822280883789,
      "learning_rate": 1.2943565025168197e-05,
      "loss": 4.2873,
      "step": 10472
    },
    {
      "epoch": 0.422809850625757,
      "grad_norm": 279.9666442871094,
      "learning_rate": 1.2942315351062931e-05,
      "loss": 6.5608,
      "step": 10473
    },
    {
      "epoch": 0.4228502220427937,
      "grad_norm": 113.47037506103516,
      "learning_rate": 1.2941065626650947e-05,
      "loss": 5.1987,
      "step": 10474
    },
    {
      "epoch": 0.42289059345983043,
      "grad_norm": 112.38642883300781,
      "learning_rate": 1.2939815851953608e-05,
      "loss": 5.5211,
      "step": 10475
    },
    {
      "epoch": 0.42293096487686715,
      "grad_norm": 71.53034210205078,
      "learning_rate": 1.2938566026992282e-05,
      "loss": 5.899,
      "step": 10476
    },
    {
      "epoch": 0.42297133629390393,
      "grad_norm": 142.13475036621094,
      "learning_rate": 1.2937316151788337e-05,
      "loss": 6.0555,
      "step": 10477
    },
    {
      "epoch": 0.42301170771094065,
      "grad_norm": 114.3092269897461,
      "learning_rate": 1.2936066226363148e-05,
      "loss": 5.9892,
      "step": 10478
    },
    {
      "epoch": 0.4230520791279774,
      "grad_norm": 77.67780303955078,
      "learning_rate": 1.2934816250738082e-05,
      "loss": 4.331,
      "step": 10479
    },
    {
      "epoch": 0.42309245054501415,
      "grad_norm": 101.14191436767578,
      "learning_rate": 1.2933566224934513e-05,
      "loss": 6.6185,
      "step": 10480
    },
    {
      "epoch": 0.4231328219620509,
      "grad_norm": 83.0066909790039,
      "learning_rate": 1.2932316148973807e-05,
      "loss": 5.328,
      "step": 10481
    },
    {
      "epoch": 0.4231731933790876,
      "grad_norm": 93.73355865478516,
      "learning_rate": 1.2931066022877346e-05,
      "loss": 6.6899,
      "step": 10482
    },
    {
      "epoch": 0.4232135647961243,
      "grad_norm": 91.11919403076172,
      "learning_rate": 1.2929815846666498e-05,
      "loss": 6.3796,
      "step": 10483
    },
    {
      "epoch": 0.4232539362131611,
      "grad_norm": 92.28973388671875,
      "learning_rate": 1.2928565620362642e-05,
      "loss": 5.1251,
      "step": 10484
    },
    {
      "epoch": 0.4232943076301978,
      "grad_norm": 66.800048828125,
      "learning_rate": 1.2927315343987154e-05,
      "loss": 4.0843,
      "step": 10485
    },
    {
      "epoch": 0.42333467904723454,
      "grad_norm": 97.13514709472656,
      "learning_rate": 1.292606501756141e-05,
      "loss": 5.8554,
      "step": 10486
    },
    {
      "epoch": 0.4233750504642713,
      "grad_norm": 115.61336517333984,
      "learning_rate": 1.2924814641106785e-05,
      "loss": 5.3171,
      "step": 10487
    },
    {
      "epoch": 0.42341542188130804,
      "grad_norm": 78.46619415283203,
      "learning_rate": 1.292356421464466e-05,
      "loss": 4.7495,
      "step": 10488
    },
    {
      "epoch": 0.42345579329834476,
      "grad_norm": 102.22953033447266,
      "learning_rate": 1.2922313738196412e-05,
      "loss": 3.6848,
      "step": 10489
    },
    {
      "epoch": 0.4234961647153815,
      "grad_norm": 153.7582244873047,
      "learning_rate": 1.2921063211783427e-05,
      "loss": 3.9464,
      "step": 10490
    },
    {
      "epoch": 0.42353653613241826,
      "grad_norm": 132.11830139160156,
      "learning_rate": 1.291981263542708e-05,
      "loss": 4.6521,
      "step": 10491
    },
    {
      "epoch": 0.423576907549455,
      "grad_norm": 296.0662536621094,
      "learning_rate": 1.2918562009148753e-05,
      "loss": 3.084,
      "step": 10492
    },
    {
      "epoch": 0.4236172789664917,
      "grad_norm": 95.59799194335938,
      "learning_rate": 1.2917311332969833e-05,
      "loss": 5.4523,
      "step": 10493
    },
    {
      "epoch": 0.4236576503835285,
      "grad_norm": 117.54253387451172,
      "learning_rate": 1.2916060606911704e-05,
      "loss": 3.9325,
      "step": 10494
    },
    {
      "epoch": 0.4236980218005652,
      "grad_norm": 102.30755615234375,
      "learning_rate": 1.2914809830995746e-05,
      "loss": 4.9918,
      "step": 10495
    },
    {
      "epoch": 0.4237383932176019,
      "grad_norm": 101.22602844238281,
      "learning_rate": 1.2913559005243348e-05,
      "loss": 5.3443,
      "step": 10496
    },
    {
      "epoch": 0.42377876463463865,
      "grad_norm": 88.96715545654297,
      "learning_rate": 1.2912308129675893e-05,
      "loss": 4.468,
      "step": 10497
    },
    {
      "epoch": 0.4238191360516754,
      "grad_norm": 118.72821807861328,
      "learning_rate": 1.291105720431477e-05,
      "loss": 4.7043,
      "step": 10498
    },
    {
      "epoch": 0.42385950746871215,
      "grad_norm": 90.93807983398438,
      "learning_rate": 1.2909806229181365e-05,
      "loss": 4.7675,
      "step": 10499
    },
    {
      "epoch": 0.42389987888574887,
      "grad_norm": 99.94620513916016,
      "learning_rate": 1.2908555204297073e-05,
      "loss": 6.0765,
      "step": 10500
    },
    {
      "epoch": 0.42394025030278565,
      "grad_norm": 94.93340301513672,
      "learning_rate": 1.2907304129683274e-05,
      "loss": 5.8037,
      "step": 10501
    },
    {
      "epoch": 0.42398062171982237,
      "grad_norm": 103.22908782958984,
      "learning_rate": 1.2906053005361366e-05,
      "loss": 5.4396,
      "step": 10502
    },
    {
      "epoch": 0.4240209931368591,
      "grad_norm": 68.59207916259766,
      "learning_rate": 1.2904801831352736e-05,
      "loss": 4.9598,
      "step": 10503
    },
    {
      "epoch": 0.4240613645538958,
      "grad_norm": 67.27873229980469,
      "learning_rate": 1.2903550607678778e-05,
      "loss": 4.461,
      "step": 10504
    },
    {
      "epoch": 0.4241017359709326,
      "grad_norm": 131.81842041015625,
      "learning_rate": 1.2902299334360886e-05,
      "loss": 4.8333,
      "step": 10505
    },
    {
      "epoch": 0.4241421073879693,
      "grad_norm": 118.64942932128906,
      "learning_rate": 1.2901048011420454e-05,
      "loss": 6.2273,
      "step": 10506
    },
    {
      "epoch": 0.42418247880500604,
      "grad_norm": 101.46829986572266,
      "learning_rate": 1.2899796638878875e-05,
      "loss": 3.0031,
      "step": 10507
    },
    {
      "epoch": 0.4242228502220428,
      "grad_norm": 139.2996368408203,
      "learning_rate": 1.2898545216757544e-05,
      "loss": 2.8782,
      "step": 10508
    },
    {
      "epoch": 0.42426322163907954,
      "grad_norm": 144.82313537597656,
      "learning_rate": 1.2897293745077857e-05,
      "loss": 2.9588,
      "step": 10509
    },
    {
      "epoch": 0.42430359305611626,
      "grad_norm": 97.66851043701172,
      "learning_rate": 1.2896042223861216e-05,
      "loss": 3.9528,
      "step": 10510
    },
    {
      "epoch": 0.42434396447315303,
      "grad_norm": 129.19677734375,
      "learning_rate": 1.2894790653129013e-05,
      "loss": 5.2088,
      "step": 10511
    },
    {
      "epoch": 0.42438433589018976,
      "grad_norm": 87.41993713378906,
      "learning_rate": 1.2893539032902649e-05,
      "loss": 5.0777,
      "step": 10512
    },
    {
      "epoch": 0.4244247073072265,
      "grad_norm": 117.98924255371094,
      "learning_rate": 1.2892287363203526e-05,
      "loss": 4.4003,
      "step": 10513
    },
    {
      "epoch": 0.4244650787242632,
      "grad_norm": 123.72396850585938,
      "learning_rate": 1.2891035644053043e-05,
      "loss": 5.623,
      "step": 10514
    },
    {
      "epoch": 0.4245054501413,
      "grad_norm": 95.85831451416016,
      "learning_rate": 1.2889783875472602e-05,
      "loss": 5.0354,
      "step": 10515
    },
    {
      "epoch": 0.4245458215583367,
      "grad_norm": 89.70660400390625,
      "learning_rate": 1.2888532057483604e-05,
      "loss": 5.1862,
      "step": 10516
    },
    {
      "epoch": 0.4245861929753734,
      "grad_norm": 103.71366882324219,
      "learning_rate": 1.2887280190107454e-05,
      "loss": 5.0754,
      "step": 10517
    },
    {
      "epoch": 0.4246265643924102,
      "grad_norm": 108.44758605957031,
      "learning_rate": 1.2886028273365554e-05,
      "loss": 6.6259,
      "step": 10518
    },
    {
      "epoch": 0.4246669358094469,
      "grad_norm": 90.92899322509766,
      "learning_rate": 1.288477630727931e-05,
      "loss": 5.4768,
      "step": 10519
    },
    {
      "epoch": 0.42470730722648364,
      "grad_norm": 140.95465087890625,
      "learning_rate": 1.2883524291870129e-05,
      "loss": 5.5116,
      "step": 10520
    },
    {
      "epoch": 0.42474767864352037,
      "grad_norm": 102.9776840209961,
      "learning_rate": 1.2882272227159414e-05,
      "loss": 5.195,
      "step": 10521
    },
    {
      "epoch": 0.42478805006055714,
      "grad_norm": 156.94334411621094,
      "learning_rate": 1.2881020113168576e-05,
      "loss": 4.094,
      "step": 10522
    },
    {
      "epoch": 0.42482842147759386,
      "grad_norm": 128.6014404296875,
      "learning_rate": 1.287976794991902e-05,
      "loss": 6.419,
      "step": 10523
    },
    {
      "epoch": 0.4248687928946306,
      "grad_norm": 85.6791763305664,
      "learning_rate": 1.287851573743216e-05,
      "loss": 5.0596,
      "step": 10524
    },
    {
      "epoch": 0.42490916431166736,
      "grad_norm": 102.57130432128906,
      "learning_rate": 1.28772634757294e-05,
      "loss": 5.2045,
      "step": 10525
    },
    {
      "epoch": 0.4249495357287041,
      "grad_norm": 116.74871826171875,
      "learning_rate": 1.2876011164832154e-05,
      "loss": 6.3256,
      "step": 10526
    },
    {
      "epoch": 0.4249899071457408,
      "grad_norm": 110.11992645263672,
      "learning_rate": 1.2874758804761833e-05,
      "loss": 4.6323,
      "step": 10527
    },
    {
      "epoch": 0.42503027856277753,
      "grad_norm": 94.4134521484375,
      "learning_rate": 1.287350639553985e-05,
      "loss": 6.2994,
      "step": 10528
    },
    {
      "epoch": 0.4250706499798143,
      "grad_norm": 120.08726501464844,
      "learning_rate": 1.2872253937187617e-05,
      "loss": 5.1422,
      "step": 10529
    },
    {
      "epoch": 0.42511102139685103,
      "grad_norm": 93.34520721435547,
      "learning_rate": 1.2871001429726549e-05,
      "loss": 4.9577,
      "step": 10530
    },
    {
      "epoch": 0.42515139281388775,
      "grad_norm": 102.2908706665039,
      "learning_rate": 1.2869748873178064e-05,
      "loss": 3.8278,
      "step": 10531
    },
    {
      "epoch": 0.42519176423092453,
      "grad_norm": 94.15699768066406,
      "learning_rate": 1.2868496267563569e-05,
      "loss": 3.3462,
      "step": 10532
    },
    {
      "epoch": 0.42523213564796125,
      "grad_norm": 148.23255920410156,
      "learning_rate": 1.2867243612904488e-05,
      "loss": 4.4203,
      "step": 10533
    },
    {
      "epoch": 0.425272507064998,
      "grad_norm": 67.67922973632812,
      "learning_rate": 1.286599090922224e-05,
      "loss": 4.8391,
      "step": 10534
    },
    {
      "epoch": 0.4253128784820347,
      "grad_norm": 94.31512451171875,
      "learning_rate": 1.2864738156538238e-05,
      "loss": 5.6781,
      "step": 10535
    },
    {
      "epoch": 0.4253532498990715,
      "grad_norm": 151.8422393798828,
      "learning_rate": 1.2863485354873903e-05,
      "loss": 6.4591,
      "step": 10536
    },
    {
      "epoch": 0.4253936213161082,
      "grad_norm": 143.7410888671875,
      "learning_rate": 1.2862232504250653e-05,
      "loss": 4.7507,
      "step": 10537
    },
    {
      "epoch": 0.4254339927331449,
      "grad_norm": 79.33544158935547,
      "learning_rate": 1.2860979604689916e-05,
      "loss": 5.4491,
      "step": 10538
    },
    {
      "epoch": 0.4254743641501817,
      "grad_norm": 87.21061706542969,
      "learning_rate": 1.2859726656213103e-05,
      "loss": 5.4118,
      "step": 10539
    },
    {
      "epoch": 0.4255147355672184,
      "grad_norm": 97.42262268066406,
      "learning_rate": 1.2858473658841644e-05,
      "loss": 5.9451,
      "step": 10540
    },
    {
      "epoch": 0.42555510698425514,
      "grad_norm": 87.49330139160156,
      "learning_rate": 1.2857220612596961e-05,
      "loss": 3.6574,
      "step": 10541
    },
    {
      "epoch": 0.42559547840129186,
      "grad_norm": 125.9598388671875,
      "learning_rate": 1.2855967517500481e-05,
      "loss": 6.1679,
      "step": 10542
    },
    {
      "epoch": 0.42563584981832864,
      "grad_norm": 133.01515197753906,
      "learning_rate": 1.285471437357362e-05,
      "loss": 5.2018,
      "step": 10543
    },
    {
      "epoch": 0.42567622123536536,
      "grad_norm": 96.228515625,
      "learning_rate": 1.2853461180837814e-05,
      "loss": 4.1542,
      "step": 10544
    },
    {
      "epoch": 0.4257165926524021,
      "grad_norm": 76.77877044677734,
      "learning_rate": 1.2852207939314481e-05,
      "loss": 5.9046,
      "step": 10545
    },
    {
      "epoch": 0.42575696406943886,
      "grad_norm": 110.65179443359375,
      "learning_rate": 1.2850954649025056e-05,
      "loss": 4.3626,
      "step": 10546
    },
    {
      "epoch": 0.4257973354864756,
      "grad_norm": 89.93566131591797,
      "learning_rate": 1.2849701309990963e-05,
      "loss": 5.5239,
      "step": 10547
    },
    {
      "epoch": 0.4258377069035123,
      "grad_norm": 94.1953353881836,
      "learning_rate": 1.284844792223363e-05,
      "loss": 3.4331,
      "step": 10548
    },
    {
      "epoch": 0.425878078320549,
      "grad_norm": 109.41735076904297,
      "learning_rate": 1.2847194485774493e-05,
      "loss": 5.7994,
      "step": 10549
    },
    {
      "epoch": 0.4259184497375858,
      "grad_norm": 152.82550048828125,
      "learning_rate": 1.2845941000634975e-05,
      "loss": 4.5227,
      "step": 10550
    },
    {
      "epoch": 0.4259588211546225,
      "grad_norm": 156.0836181640625,
      "learning_rate": 1.2844687466836514e-05,
      "loss": 5.5034,
      "step": 10551
    },
    {
      "epoch": 0.42599919257165925,
      "grad_norm": 87.34576416015625,
      "learning_rate": 1.2843433884400541e-05,
      "loss": 6.2138,
      "step": 10552
    },
    {
      "epoch": 0.426039563988696,
      "grad_norm": 81.86238861083984,
      "learning_rate": 1.2842180253348487e-05,
      "loss": 5.361,
      "step": 10553
    },
    {
      "epoch": 0.42607993540573275,
      "grad_norm": 101.9640121459961,
      "learning_rate": 1.2840926573701787e-05,
      "loss": 4.4204,
      "step": 10554
    },
    {
      "epoch": 0.42612030682276947,
      "grad_norm": 129.0758056640625,
      "learning_rate": 1.2839672845481876e-05,
      "loss": 4.7228,
      "step": 10555
    },
    {
      "epoch": 0.4261606782398062,
      "grad_norm": 98.7244644165039,
      "learning_rate": 1.2838419068710196e-05,
      "loss": 5.4186,
      "step": 10556
    },
    {
      "epoch": 0.42620104965684297,
      "grad_norm": 190.2469024658203,
      "learning_rate": 1.2837165243408172e-05,
      "loss": 5.0558,
      "step": 10557
    },
    {
      "epoch": 0.4262414210738797,
      "grad_norm": 291.3012390136719,
      "learning_rate": 1.283591136959725e-05,
      "loss": 4.4649,
      "step": 10558
    },
    {
      "epoch": 0.4262817924909164,
      "grad_norm": 114.47293090820312,
      "learning_rate": 1.2834657447298863e-05,
      "loss": 5.5865,
      "step": 10559
    },
    {
      "epoch": 0.4263221639079532,
      "grad_norm": 82.6974868774414,
      "learning_rate": 1.2833403476534456e-05,
      "loss": 4.3678,
      "step": 10560
    },
    {
      "epoch": 0.4263625353249899,
      "grad_norm": 110.33636474609375,
      "learning_rate": 1.2832149457325464e-05,
      "loss": 3.1985,
      "step": 10561
    },
    {
      "epoch": 0.42640290674202663,
      "grad_norm": 211.94204711914062,
      "learning_rate": 1.2830895389693334e-05,
      "loss": 5.1294,
      "step": 10562
    },
    {
      "epoch": 0.4264432781590634,
      "grad_norm": 89.926513671875,
      "learning_rate": 1.2829641273659498e-05,
      "loss": 6.0692,
      "step": 10563
    },
    {
      "epoch": 0.42648364957610013,
      "grad_norm": 105.01311492919922,
      "learning_rate": 1.282838710924541e-05,
      "loss": 6.5315,
      "step": 10564
    },
    {
      "epoch": 0.42652402099313685,
      "grad_norm": 105.77883911132812,
      "learning_rate": 1.2827132896472502e-05,
      "loss": 5.6281,
      "step": 10565
    },
    {
      "epoch": 0.4265643924101736,
      "grad_norm": 137.84783935546875,
      "learning_rate": 1.2825878635362226e-05,
      "loss": 5.9761,
      "step": 10566
    },
    {
      "epoch": 0.42660476382721035,
      "grad_norm": 109.4109115600586,
      "learning_rate": 1.2824624325936023e-05,
      "loss": 4.3966,
      "step": 10567
    },
    {
      "epoch": 0.4266451352442471,
      "grad_norm": 126.39497375488281,
      "learning_rate": 1.2823369968215342e-05,
      "loss": 4.294,
      "step": 10568
    },
    {
      "epoch": 0.4266855066612838,
      "grad_norm": 103.1612777709961,
      "learning_rate": 1.2822115562221624e-05,
      "loss": 4.7246,
      "step": 10569
    },
    {
      "epoch": 0.4267258780783206,
      "grad_norm": 118.47811889648438,
      "learning_rate": 1.282086110797632e-05,
      "loss": 4.4959,
      "step": 10570
    },
    {
      "epoch": 0.4267662494953573,
      "grad_norm": 99.30658721923828,
      "learning_rate": 1.2819606605500882e-05,
      "loss": 5.0225,
      "step": 10571
    },
    {
      "epoch": 0.426806620912394,
      "grad_norm": 121.7347640991211,
      "learning_rate": 1.2818352054816753e-05,
      "loss": 5.0832,
      "step": 10572
    },
    {
      "epoch": 0.42684699232943074,
      "grad_norm": 199.4119110107422,
      "learning_rate": 1.2817097455945387e-05,
      "loss": 6.2219,
      "step": 10573
    },
    {
      "epoch": 0.4268873637464675,
      "grad_norm": 134.78163146972656,
      "learning_rate": 1.2815842808908232e-05,
      "loss": 4.2269,
      "step": 10574
    },
    {
      "epoch": 0.42692773516350424,
      "grad_norm": 126.03961944580078,
      "learning_rate": 1.281458811372674e-05,
      "loss": 4.8451,
      "step": 10575
    },
    {
      "epoch": 0.42696810658054096,
      "grad_norm": 91.97830200195312,
      "learning_rate": 1.2813333370422365e-05,
      "loss": 4.9058,
      "step": 10576
    },
    {
      "epoch": 0.42700847799757774,
      "grad_norm": 135.9945068359375,
      "learning_rate": 1.2812078579016558e-05,
      "loss": 4.5105,
      "step": 10577
    },
    {
      "epoch": 0.42704884941461446,
      "grad_norm": 112.97103118896484,
      "learning_rate": 1.2810823739530774e-05,
      "loss": 3.7593,
      "step": 10578
    },
    {
      "epoch": 0.4270892208316512,
      "grad_norm": 103.99992370605469,
      "learning_rate": 1.2809568851986466e-05,
      "loss": 6.1812,
      "step": 10579
    },
    {
      "epoch": 0.4271295922486879,
      "grad_norm": 121.57849884033203,
      "learning_rate": 1.2808313916405093e-05,
      "loss": 6.2114,
      "step": 10580
    },
    {
      "epoch": 0.4271699636657247,
      "grad_norm": 184.70155334472656,
      "learning_rate": 1.280705893280811e-05,
      "loss": 4.19,
      "step": 10581
    },
    {
      "epoch": 0.4272103350827614,
      "grad_norm": 99.18299865722656,
      "learning_rate": 1.2805803901216972e-05,
      "loss": 5.2796,
      "step": 10582
    },
    {
      "epoch": 0.4272507064997981,
      "grad_norm": 108.32032012939453,
      "learning_rate": 1.2804548821653144e-05,
      "loss": 5.0585,
      "step": 10583
    },
    {
      "epoch": 0.4272910779168349,
      "grad_norm": 63.42863845825195,
      "learning_rate": 1.2803293694138077e-05,
      "loss": 5.1714,
      "step": 10584
    },
    {
      "epoch": 0.4273314493338716,
      "grad_norm": 149.74630737304688,
      "learning_rate": 1.2802038518693234e-05,
      "loss": 4.9795,
      "step": 10585
    },
    {
      "epoch": 0.42737182075090835,
      "grad_norm": 192.06468200683594,
      "learning_rate": 1.2800783295340078e-05,
      "loss": 5.6821,
      "step": 10586
    },
    {
      "epoch": 0.42741219216794507,
      "grad_norm": 108.4285659790039,
      "learning_rate": 1.2799528024100068e-05,
      "loss": 4.8523,
      "step": 10587
    },
    {
      "epoch": 0.42745256358498185,
      "grad_norm": 113.80693817138672,
      "learning_rate": 1.2798272704994662e-05,
      "loss": 4.9441,
      "step": 10588
    },
    {
      "epoch": 0.42749293500201857,
      "grad_norm": 117.38764190673828,
      "learning_rate": 1.279701733804533e-05,
      "loss": 5.7184,
      "step": 10589
    },
    {
      "epoch": 0.4275333064190553,
      "grad_norm": 217.08721923828125,
      "learning_rate": 1.2795761923273533e-05,
      "loss": 4.2517,
      "step": 10590
    },
    {
      "epoch": 0.42757367783609207,
      "grad_norm": 68.11651611328125,
      "learning_rate": 1.2794506460700733e-05,
      "loss": 4.5658,
      "step": 10591
    },
    {
      "epoch": 0.4276140492531288,
      "grad_norm": 99.9979476928711,
      "learning_rate": 1.2793250950348402e-05,
      "loss": 6.9129,
      "step": 10592
    },
    {
      "epoch": 0.4276544206701655,
      "grad_norm": 83.23553466796875,
      "learning_rate": 1.2791995392238002e-05,
      "loss": 4.6986,
      "step": 10593
    },
    {
      "epoch": 0.42769479208720224,
      "grad_norm": 98.91389465332031,
      "learning_rate": 1.2790739786391e-05,
      "loss": 6.3043,
      "step": 10594
    },
    {
      "epoch": 0.427735163504239,
      "grad_norm": 96.85334777832031,
      "learning_rate": 1.2789484132828862e-05,
      "loss": 5.1542,
      "step": 10595
    },
    {
      "epoch": 0.42777553492127574,
      "grad_norm": 92.82782745361328,
      "learning_rate": 1.2788228431573062e-05,
      "loss": 4.5793,
      "step": 10596
    },
    {
      "epoch": 0.42781590633831246,
      "grad_norm": 85.76514434814453,
      "learning_rate": 1.2786972682645065e-05,
      "loss": 4.7619,
      "step": 10597
    },
    {
      "epoch": 0.42785627775534923,
      "grad_norm": 81.72380065917969,
      "learning_rate": 1.2785716886066345e-05,
      "loss": 6.5674,
      "step": 10598
    },
    {
      "epoch": 0.42789664917238596,
      "grad_norm": 110.6695785522461,
      "learning_rate": 1.2784461041858372e-05,
      "loss": 4.9924,
      "step": 10599
    },
    {
      "epoch": 0.4279370205894227,
      "grad_norm": 78.54231262207031,
      "learning_rate": 1.2783205150042615e-05,
      "loss": 5.3798,
      "step": 10600
    },
    {
      "epoch": 0.4279773920064594,
      "grad_norm": 114.96614837646484,
      "learning_rate": 1.2781949210640549e-05,
      "loss": 4.5808,
      "step": 10601
    },
    {
      "epoch": 0.4280177634234962,
      "grad_norm": 75.76441955566406,
      "learning_rate": 1.278069322367365e-05,
      "loss": 7.1324,
      "step": 10602
    },
    {
      "epoch": 0.4280581348405329,
      "grad_norm": 72.85884094238281,
      "learning_rate": 1.2779437189163389e-05,
      "loss": 5.5997,
      "step": 10603
    },
    {
      "epoch": 0.4280985062575696,
      "grad_norm": 102.20101928710938,
      "learning_rate": 1.2778181107131242e-05,
      "loss": 6.1854,
      "step": 10604
    },
    {
      "epoch": 0.4281388776746064,
      "grad_norm": 101.60234069824219,
      "learning_rate": 1.2776924977598685e-05,
      "loss": 4.9369,
      "step": 10605
    },
    {
      "epoch": 0.4281792490916431,
      "grad_norm": 80.27693176269531,
      "learning_rate": 1.27756688005872e-05,
      "loss": 5.7555,
      "step": 10606
    },
    {
      "epoch": 0.42821962050867984,
      "grad_norm": 103.49581909179688,
      "learning_rate": 1.2774412576118255e-05,
      "loss": 2.7355,
      "step": 10607
    },
    {
      "epoch": 0.42825999192571657,
      "grad_norm": 90.68856811523438,
      "learning_rate": 1.2773156304213336e-05,
      "loss": 4.0963,
      "step": 10608
    },
    {
      "epoch": 0.42830036334275334,
      "grad_norm": 88.54850769042969,
      "learning_rate": 1.277189998489392e-05,
      "loss": 4.8661,
      "step": 10609
    },
    {
      "epoch": 0.42834073475979006,
      "grad_norm": 87.18345642089844,
      "learning_rate": 1.2770643618181486e-05,
      "loss": 5.9184,
      "step": 10610
    },
    {
      "epoch": 0.4283811061768268,
      "grad_norm": 134.98374938964844,
      "learning_rate": 1.2769387204097516e-05,
      "loss": 4.4954,
      "step": 10611
    },
    {
      "epoch": 0.42842147759386356,
      "grad_norm": 92.40721893310547,
      "learning_rate": 1.2768130742663493e-05,
      "loss": 6.0296,
      "step": 10612
    },
    {
      "epoch": 0.4284618490109003,
      "grad_norm": 188.9901123046875,
      "learning_rate": 1.2766874233900896e-05,
      "loss": 6.3387,
      "step": 10613
    },
    {
      "epoch": 0.428502220427937,
      "grad_norm": 95.05084991455078,
      "learning_rate": 1.2765617677831216e-05,
      "loss": 4.5497,
      "step": 10614
    },
    {
      "epoch": 0.4285425918449738,
      "grad_norm": 91.5427474975586,
      "learning_rate": 1.2764361074475927e-05,
      "loss": 5.5226,
      "step": 10615
    },
    {
      "epoch": 0.4285829632620105,
      "grad_norm": 305.6809387207031,
      "learning_rate": 1.2763104423856523e-05,
      "loss": 5.4794,
      "step": 10616
    },
    {
      "epoch": 0.42862333467904723,
      "grad_norm": 80.83489227294922,
      "learning_rate": 1.276184772599448e-05,
      "loss": 6.5705,
      "step": 10617
    },
    {
      "epoch": 0.42866370609608395,
      "grad_norm": 85.09269714355469,
      "learning_rate": 1.2760590980911296e-05,
      "loss": 4.6342,
      "step": 10618
    },
    {
      "epoch": 0.42870407751312073,
      "grad_norm": 80.41839599609375,
      "learning_rate": 1.275933418862845e-05,
      "loss": 4.9111,
      "step": 10619
    },
    {
      "epoch": 0.42874444893015745,
      "grad_norm": 88.21859741210938,
      "learning_rate": 1.2758077349167431e-05,
      "loss": 6.0149,
      "step": 10620
    },
    {
      "epoch": 0.4287848203471942,
      "grad_norm": 101.50604248046875,
      "learning_rate": 1.275682046254973e-05,
      "loss": 6.0722,
      "step": 10621
    },
    {
      "epoch": 0.42882519176423095,
      "grad_norm": 130.74937438964844,
      "learning_rate": 1.2755563528796842e-05,
      "loss": 5.2734,
      "step": 10622
    },
    {
      "epoch": 0.4288655631812677,
      "grad_norm": 91.71125793457031,
      "learning_rate": 1.2754306547930246e-05,
      "loss": 5.725,
      "step": 10623
    },
    {
      "epoch": 0.4289059345983044,
      "grad_norm": 77.25484466552734,
      "learning_rate": 1.2753049519971446e-05,
      "loss": 3.6927,
      "step": 10624
    },
    {
      "epoch": 0.4289463060153411,
      "grad_norm": 178.22625732421875,
      "learning_rate": 1.2751792444941922e-05,
      "loss": 5.0044,
      "step": 10625
    },
    {
      "epoch": 0.4289866774323779,
      "grad_norm": 187.7103271484375,
      "learning_rate": 1.2750535322863178e-05,
      "loss": 5.5415,
      "step": 10626
    },
    {
      "epoch": 0.4290270488494146,
      "grad_norm": 73.9297103881836,
      "learning_rate": 1.2749278153756698e-05,
      "loss": 7.4336,
      "step": 10627
    },
    {
      "epoch": 0.42906742026645134,
      "grad_norm": 109.09795379638672,
      "learning_rate": 1.2748020937643987e-05,
      "loss": 5.5242,
      "step": 10628
    },
    {
      "epoch": 0.4291077916834881,
      "grad_norm": 219.75440979003906,
      "learning_rate": 1.2746763674546533e-05,
      "loss": 2.98,
      "step": 10629
    },
    {
      "epoch": 0.42914816310052484,
      "grad_norm": 89.88192749023438,
      "learning_rate": 1.2745506364485831e-05,
      "loss": 5.8507,
      "step": 10630
    },
    {
      "epoch": 0.42918853451756156,
      "grad_norm": 79.06483459472656,
      "learning_rate": 1.2744249007483382e-05,
      "loss": 5.1383,
      "step": 10631
    },
    {
      "epoch": 0.4292289059345983,
      "grad_norm": 75.42599487304688,
      "learning_rate": 1.2742991603560687e-05,
      "loss": 5.177,
      "step": 10632
    },
    {
      "epoch": 0.42926927735163506,
      "grad_norm": 109.03411102294922,
      "learning_rate": 1.2741734152739241e-05,
      "loss": 4.754,
      "step": 10633
    },
    {
      "epoch": 0.4293096487686718,
      "grad_norm": 105.50042724609375,
      "learning_rate": 1.2740476655040541e-05,
      "loss": 6.1168,
      "step": 10634
    },
    {
      "epoch": 0.4293500201857085,
      "grad_norm": 90.32139587402344,
      "learning_rate": 1.273921911048609e-05,
      "loss": 5.4145,
      "step": 10635
    },
    {
      "epoch": 0.4293903916027453,
      "grad_norm": 108.07382202148438,
      "learning_rate": 1.273796151909739e-05,
      "loss": 4.7863,
      "step": 10636
    },
    {
      "epoch": 0.429430763019782,
      "grad_norm": 91.25501251220703,
      "learning_rate": 1.273670388089594e-05,
      "loss": 3.864,
      "step": 10637
    },
    {
      "epoch": 0.4294711344368187,
      "grad_norm": 88.62580108642578,
      "learning_rate": 1.2735446195903243e-05,
      "loss": 5.3083,
      "step": 10638
    },
    {
      "epoch": 0.42951150585385545,
      "grad_norm": 99.86181640625,
      "learning_rate": 1.2734188464140807e-05,
      "loss": 5.6464,
      "step": 10639
    },
    {
      "epoch": 0.4295518772708922,
      "grad_norm": 129.19863891601562,
      "learning_rate": 1.2732930685630131e-05,
      "loss": 6.0408,
      "step": 10640
    },
    {
      "epoch": 0.42959224868792895,
      "grad_norm": 97.09325408935547,
      "learning_rate": 1.2731672860392722e-05,
      "loss": 4.9545,
      "step": 10641
    },
    {
      "epoch": 0.42963262010496567,
      "grad_norm": 80.72576904296875,
      "learning_rate": 1.2730414988450091e-05,
      "loss": 5.6136,
      "step": 10642
    },
    {
      "epoch": 0.42967299152200245,
      "grad_norm": 68.85657501220703,
      "learning_rate": 1.2729157069823732e-05,
      "loss": 2.3187,
      "step": 10643
    },
    {
      "epoch": 0.42971336293903917,
      "grad_norm": 84.86083221435547,
      "learning_rate": 1.2727899104535166e-05,
      "loss": 4.9102,
      "step": 10644
    },
    {
      "epoch": 0.4297537343560759,
      "grad_norm": 74.69751739501953,
      "learning_rate": 1.2726641092605892e-05,
      "loss": 4.2773,
      "step": 10645
    },
    {
      "epoch": 0.4297941057731126,
      "grad_norm": 115.15034484863281,
      "learning_rate": 1.2725383034057427e-05,
      "loss": 3.8412,
      "step": 10646
    },
    {
      "epoch": 0.4298344771901494,
      "grad_norm": 73.37838745117188,
      "learning_rate": 1.272412492891127e-05,
      "loss": 6.3146,
      "step": 10647
    },
    {
      "epoch": 0.4298748486071861,
      "grad_norm": 61.659549713134766,
      "learning_rate": 1.272286677718894e-05,
      "loss": 5.2293,
      "step": 10648
    },
    {
      "epoch": 0.42991522002422283,
      "grad_norm": 97.89999389648438,
      "learning_rate": 1.2721608578911946e-05,
      "loss": 5.3823,
      "step": 10649
    },
    {
      "epoch": 0.4299555914412596,
      "grad_norm": 79.42520904541016,
      "learning_rate": 1.2720350334101805e-05,
      "loss": 5.0043,
      "step": 10650
    },
    {
      "epoch": 0.42999596285829633,
      "grad_norm": 80.9129867553711,
      "learning_rate": 1.2719092042780022e-05,
      "loss": 5.4802,
      "step": 10651
    },
    {
      "epoch": 0.43003633427533305,
      "grad_norm": 92.96246337890625,
      "learning_rate": 1.2717833704968116e-05,
      "loss": 5.4614,
      "step": 10652
    },
    {
      "epoch": 0.4300767056923698,
      "grad_norm": 155.47659301757812,
      "learning_rate": 1.2716575320687602e-05,
      "loss": 6.6323,
      "step": 10653
    },
    {
      "epoch": 0.43011707710940655,
      "grad_norm": 208.8515625,
      "learning_rate": 1.271531688995999e-05,
      "loss": 5.0809,
      "step": 10654
    },
    {
      "epoch": 0.4301574485264433,
      "grad_norm": 151.6039276123047,
      "learning_rate": 1.27140584128068e-05,
      "loss": 6.1745,
      "step": 10655
    },
    {
      "epoch": 0.43019781994348,
      "grad_norm": 104.94573211669922,
      "learning_rate": 1.2712799889249552e-05,
      "loss": 5.127,
      "step": 10656
    },
    {
      "epoch": 0.4302381913605168,
      "grad_norm": 125.09154510498047,
      "learning_rate": 1.271154131930976e-05,
      "loss": 6.2308,
      "step": 10657
    },
    {
      "epoch": 0.4302785627775535,
      "grad_norm": 71.8083724975586,
      "learning_rate": 1.271028270300894e-05,
      "loss": 5.9225,
      "step": 10658
    },
    {
      "epoch": 0.4303189341945902,
      "grad_norm": 114.98157501220703,
      "learning_rate": 1.2709024040368617e-05,
      "loss": 6.5789,
      "step": 10659
    },
    {
      "epoch": 0.43035930561162694,
      "grad_norm": 71.46260833740234,
      "learning_rate": 1.2707765331410309e-05,
      "loss": 5.5071,
      "step": 10660
    },
    {
      "epoch": 0.4303996770286637,
      "grad_norm": 87.37716674804688,
      "learning_rate": 1.2706506576155537e-05,
      "loss": 6.3815,
      "step": 10661
    },
    {
      "epoch": 0.43044004844570044,
      "grad_norm": 78.3779525756836,
      "learning_rate": 1.2705247774625823e-05,
      "loss": 6.2662,
      "step": 10662
    },
    {
      "epoch": 0.43048041986273716,
      "grad_norm": 73.9124755859375,
      "learning_rate": 1.270398892684269e-05,
      "loss": 5.0133,
      "step": 10663
    },
    {
      "epoch": 0.43052079127977394,
      "grad_norm": 107.06293487548828,
      "learning_rate": 1.270273003282766e-05,
      "loss": 6.2913,
      "step": 10664
    },
    {
      "epoch": 0.43056116269681066,
      "grad_norm": 86.69734191894531,
      "learning_rate": 1.2701471092602256e-05,
      "loss": 5.4399,
      "step": 10665
    },
    {
      "epoch": 0.4306015341138474,
      "grad_norm": 116.50210571289062,
      "learning_rate": 1.2700212106188008e-05,
      "loss": 4.1902,
      "step": 10666
    },
    {
      "epoch": 0.43064190553088416,
      "grad_norm": 124.68472290039062,
      "learning_rate": 1.2698953073606434e-05,
      "loss": 5.7506,
      "step": 10667
    },
    {
      "epoch": 0.4306822769479209,
      "grad_norm": 111.83118438720703,
      "learning_rate": 1.2697693994879068e-05,
      "loss": 5.1991,
      "step": 10668
    },
    {
      "epoch": 0.4307226483649576,
      "grad_norm": 85.08704376220703,
      "learning_rate": 1.2696434870027433e-05,
      "loss": 4.1202,
      "step": 10669
    },
    {
      "epoch": 0.4307630197819943,
      "grad_norm": 74.82897186279297,
      "learning_rate": 1.269517569907306e-05,
      "loss": 5.798,
      "step": 10670
    },
    {
      "epoch": 0.4308033911990311,
      "grad_norm": 105.88129425048828,
      "learning_rate": 1.2693916482037474e-05,
      "loss": 5.3505,
      "step": 10671
    },
    {
      "epoch": 0.4308437626160678,
      "grad_norm": 72.72120666503906,
      "learning_rate": 1.2692657218942208e-05,
      "loss": 7.2608,
      "step": 10672
    },
    {
      "epoch": 0.43088413403310455,
      "grad_norm": 78.47320556640625,
      "learning_rate": 1.2691397909808792e-05,
      "loss": 4.1396,
      "step": 10673
    },
    {
      "epoch": 0.4309245054501413,
      "grad_norm": 85.88098907470703,
      "learning_rate": 1.2690138554658758e-05,
      "loss": 5.1355,
      "step": 10674
    },
    {
      "epoch": 0.43096487686717805,
      "grad_norm": 84.50798797607422,
      "learning_rate": 1.2688879153513633e-05,
      "loss": 3.9751,
      "step": 10675
    },
    {
      "epoch": 0.43100524828421477,
      "grad_norm": 86.12371063232422,
      "learning_rate": 1.2687619706394955e-05,
      "loss": 5.3414,
      "step": 10676
    },
    {
      "epoch": 0.4310456197012515,
      "grad_norm": 146.73489379882812,
      "learning_rate": 1.2686360213324258e-05,
      "loss": 2.9754,
      "step": 10677
    },
    {
      "epoch": 0.43108599111828827,
      "grad_norm": 86.18433380126953,
      "learning_rate": 1.268510067432307e-05,
      "loss": 6.1279,
      "step": 10678
    },
    {
      "epoch": 0.431126362535325,
      "grad_norm": 85.7718734741211,
      "learning_rate": 1.2683841089412934e-05,
      "loss": 5.3107,
      "step": 10679
    },
    {
      "epoch": 0.4311667339523617,
      "grad_norm": 117.58595275878906,
      "learning_rate": 1.2682581458615384e-05,
      "loss": 4.6865,
      "step": 10680
    },
    {
      "epoch": 0.4312071053693985,
      "grad_norm": 161.47398376464844,
      "learning_rate": 1.2681321781951953e-05,
      "loss": 6.093,
      "step": 10681
    },
    {
      "epoch": 0.4312474767864352,
      "grad_norm": 94.00131225585938,
      "learning_rate": 1.2680062059444181e-05,
      "loss": 3.9169,
      "step": 10682
    },
    {
      "epoch": 0.43128784820347194,
      "grad_norm": 76.67631530761719,
      "learning_rate": 1.2678802291113606e-05,
      "loss": 5.1074,
      "step": 10683
    },
    {
      "epoch": 0.43132821962050866,
      "grad_norm": 119.33629608154297,
      "learning_rate": 1.2677542476981768e-05,
      "loss": 4.6261,
      "step": 10684
    },
    {
      "epoch": 0.43136859103754543,
      "grad_norm": 64.79886627197266,
      "learning_rate": 1.2676282617070206e-05,
      "loss": 5.8165,
      "step": 10685
    },
    {
      "epoch": 0.43140896245458216,
      "grad_norm": 86.00623321533203,
      "learning_rate": 1.267502271140046e-05,
      "loss": 5.3789,
      "step": 10686
    },
    {
      "epoch": 0.4314493338716189,
      "grad_norm": 94.72293090820312,
      "learning_rate": 1.2673762759994073e-05,
      "loss": 5.1086,
      "step": 10687
    },
    {
      "epoch": 0.43148970528865566,
      "grad_norm": 147.84097290039062,
      "learning_rate": 1.2672502762872585e-05,
      "loss": 4.6585,
      "step": 10688
    },
    {
      "epoch": 0.4315300767056924,
      "grad_norm": 138.72735595703125,
      "learning_rate": 1.2671242720057541e-05,
      "loss": 4.615,
      "step": 10689
    },
    {
      "epoch": 0.4315704481227291,
      "grad_norm": 111.46710968017578,
      "learning_rate": 1.2669982631570488e-05,
      "loss": 5.4151,
      "step": 10690
    },
    {
      "epoch": 0.4316108195397658,
      "grad_norm": 76.40265655517578,
      "learning_rate": 1.2668722497432961e-05,
      "loss": 3.7223,
      "step": 10691
    },
    {
      "epoch": 0.4316511909568026,
      "grad_norm": 175.1456756591797,
      "learning_rate": 1.2667462317666517e-05,
      "loss": 4.4687,
      "step": 10692
    },
    {
      "epoch": 0.4316915623738393,
      "grad_norm": 73.28802490234375,
      "learning_rate": 1.2666202092292693e-05,
      "loss": 5.3742,
      "step": 10693
    },
    {
      "epoch": 0.43173193379087604,
      "grad_norm": 92.66090393066406,
      "learning_rate": 1.266494182133304e-05,
      "loss": 3.5005,
      "step": 10694
    },
    {
      "epoch": 0.4317723052079128,
      "grad_norm": 130.61790466308594,
      "learning_rate": 1.2663681504809104e-05,
      "loss": 5.9867,
      "step": 10695
    },
    {
      "epoch": 0.43181267662494954,
      "grad_norm": 118.87228393554688,
      "learning_rate": 1.2662421142742436e-05,
      "loss": 5.4253,
      "step": 10696
    },
    {
      "epoch": 0.43185304804198626,
      "grad_norm": 154.85336303710938,
      "learning_rate": 1.2661160735154583e-05,
      "loss": 6.4882,
      "step": 10697
    },
    {
      "epoch": 0.431893419459023,
      "grad_norm": 90.89994049072266,
      "learning_rate": 1.2659900282067095e-05,
      "loss": 5.2705,
      "step": 10698
    },
    {
      "epoch": 0.43193379087605976,
      "grad_norm": 104.0384292602539,
      "learning_rate": 1.2658639783501523e-05,
      "loss": 3.1101,
      "step": 10699
    },
    {
      "epoch": 0.4319741622930965,
      "grad_norm": 161.3267059326172,
      "learning_rate": 1.2657379239479419e-05,
      "loss": 2.9778,
      "step": 10700
    },
    {
      "epoch": 0.4320145337101332,
      "grad_norm": 103.61318969726562,
      "learning_rate": 1.2656118650022338e-05,
      "loss": 4.7517,
      "step": 10701
    },
    {
      "epoch": 0.43205490512717,
      "grad_norm": 112.94422912597656,
      "learning_rate": 1.265485801515183e-05,
      "loss": 5.5799,
      "step": 10702
    },
    {
      "epoch": 0.4320952765442067,
      "grad_norm": 93.53435516357422,
      "learning_rate": 1.2653597334889447e-05,
      "loss": 5.8583,
      "step": 10703
    },
    {
      "epoch": 0.43213564796124343,
      "grad_norm": 161.20913696289062,
      "learning_rate": 1.265233660925675e-05,
      "loss": 5.3329,
      "step": 10704
    },
    {
      "epoch": 0.43217601937828015,
      "grad_norm": 280.9020080566406,
      "learning_rate": 1.2651075838275287e-05,
      "loss": 5.3938,
      "step": 10705
    },
    {
      "epoch": 0.43221639079531693,
      "grad_norm": 95.2249755859375,
      "learning_rate": 1.264981502196662e-05,
      "loss": 4.9686,
      "step": 10706
    },
    {
      "epoch": 0.43225676221235365,
      "grad_norm": 82.23297882080078,
      "learning_rate": 1.2648554160352302e-05,
      "loss": 6.4795,
      "step": 10707
    },
    {
      "epoch": 0.4322971336293904,
      "grad_norm": 121.00843811035156,
      "learning_rate": 1.2647293253453894e-05,
      "loss": 5.5294,
      "step": 10708
    },
    {
      "epoch": 0.43233750504642715,
      "grad_norm": 90.94705963134766,
      "learning_rate": 1.2646032301292951e-05,
      "loss": 4.3905,
      "step": 10709
    },
    {
      "epoch": 0.4323778764634639,
      "grad_norm": 79.34767150878906,
      "learning_rate": 1.2644771303891039e-05,
      "loss": 5.3832,
      "step": 10710
    },
    {
      "epoch": 0.4324182478805006,
      "grad_norm": 90.22460174560547,
      "learning_rate": 1.2643510261269709e-05,
      "loss": 5.2551,
      "step": 10711
    },
    {
      "epoch": 0.4324586192975373,
      "grad_norm": 191.30548095703125,
      "learning_rate": 1.264224917345053e-05,
      "loss": 5.5859,
      "step": 10712
    },
    {
      "epoch": 0.4324989907145741,
      "grad_norm": 104.035888671875,
      "learning_rate": 1.264098804045506e-05,
      "loss": 4.2886,
      "step": 10713
    },
    {
      "epoch": 0.4325393621316108,
      "grad_norm": 123.01309204101562,
      "learning_rate": 1.2639726862304861e-05,
      "loss": 5.4258,
      "step": 10714
    },
    {
      "epoch": 0.43257973354864754,
      "grad_norm": 114.77949523925781,
      "learning_rate": 1.2638465639021495e-05,
      "loss": 6.3351,
      "step": 10715
    },
    {
      "epoch": 0.4326201049656843,
      "grad_norm": 99.27025604248047,
      "learning_rate": 1.263720437062653e-05,
      "loss": 5.2239,
      "step": 10716
    },
    {
      "epoch": 0.43266047638272104,
      "grad_norm": 174.27410888671875,
      "learning_rate": 1.2635943057141528e-05,
      "loss": 7.0924,
      "step": 10717
    },
    {
      "epoch": 0.43270084779975776,
      "grad_norm": 114.57001495361328,
      "learning_rate": 1.2634681698588054e-05,
      "loss": 7.4814,
      "step": 10718
    },
    {
      "epoch": 0.43274121921679454,
      "grad_norm": 166.83465576171875,
      "learning_rate": 1.2633420294987677e-05,
      "loss": 5.1684,
      "step": 10719
    },
    {
      "epoch": 0.43278159063383126,
      "grad_norm": 116.02073669433594,
      "learning_rate": 1.263215884636196e-05,
      "loss": 5.1024,
      "step": 10720
    },
    {
      "epoch": 0.432821962050868,
      "grad_norm": 120.59697723388672,
      "learning_rate": 1.2630897352732476e-05,
      "loss": 4.6901,
      "step": 10721
    },
    {
      "epoch": 0.4328623334679047,
      "grad_norm": 84.81642150878906,
      "learning_rate": 1.262963581412079e-05,
      "loss": 3.4782,
      "step": 10722
    },
    {
      "epoch": 0.4329027048849415,
      "grad_norm": 85.75321960449219,
      "learning_rate": 1.2628374230548475e-05,
      "loss": 5.0304,
      "step": 10723
    },
    {
      "epoch": 0.4329430763019782,
      "grad_norm": 125.9030990600586,
      "learning_rate": 1.2627112602037094e-05,
      "loss": 3.7838,
      "step": 10724
    },
    {
      "epoch": 0.4329834477190149,
      "grad_norm": 69.52729797363281,
      "learning_rate": 1.2625850928608223e-05,
      "loss": 5.0909,
      "step": 10725
    },
    {
      "epoch": 0.4330238191360517,
      "grad_norm": 73.41195678710938,
      "learning_rate": 1.2624589210283435e-05,
      "loss": 6.1038,
      "step": 10726
    },
    {
      "epoch": 0.4330641905530884,
      "grad_norm": 194.4266815185547,
      "learning_rate": 1.2623327447084298e-05,
      "loss": 5.1676,
      "step": 10727
    },
    {
      "epoch": 0.43310456197012515,
      "grad_norm": 114.40780639648438,
      "learning_rate": 1.2622065639032387e-05,
      "loss": 4.1454,
      "step": 10728
    },
    {
      "epoch": 0.43314493338716187,
      "grad_norm": 78.9025650024414,
      "learning_rate": 1.262080378614928e-05,
      "loss": 4.2529,
      "step": 10729
    },
    {
      "epoch": 0.43318530480419865,
      "grad_norm": 114.31356048583984,
      "learning_rate": 1.2619541888456549e-05,
      "loss": 5.6014,
      "step": 10730
    },
    {
      "epoch": 0.43322567622123537,
      "grad_norm": 84.4608154296875,
      "learning_rate": 1.2618279945975764e-05,
      "loss": 3.285,
      "step": 10731
    },
    {
      "epoch": 0.4332660476382721,
      "grad_norm": 147.48899841308594,
      "learning_rate": 1.261701795872851e-05,
      "loss": 3.7204,
      "step": 10732
    },
    {
      "epoch": 0.43330641905530887,
      "grad_norm": 146.453125,
      "learning_rate": 1.2615755926736358e-05,
      "loss": 4.5676,
      "step": 10733
    },
    {
      "epoch": 0.4333467904723456,
      "grad_norm": 77.40028381347656,
      "learning_rate": 1.2614493850020888e-05,
      "loss": 8.0888,
      "step": 10734
    },
    {
      "epoch": 0.4333871618893823,
      "grad_norm": 93.04265594482422,
      "learning_rate": 1.2613231728603678e-05,
      "loss": 5.2412,
      "step": 10735
    },
    {
      "epoch": 0.43342753330641903,
      "grad_norm": 72.1028060913086,
      "learning_rate": 1.2611969562506307e-05,
      "loss": 4.7463,
      "step": 10736
    },
    {
      "epoch": 0.4334679047234558,
      "grad_norm": 85.05854797363281,
      "learning_rate": 1.261070735175036e-05,
      "loss": 5.0662,
      "step": 10737
    },
    {
      "epoch": 0.43350827614049253,
      "grad_norm": 98.50144958496094,
      "learning_rate": 1.260944509635741e-05,
      "loss": 3.7874,
      "step": 10738
    },
    {
      "epoch": 0.43354864755752925,
      "grad_norm": 118.5203628540039,
      "learning_rate": 1.2608182796349042e-05,
      "loss": 6.8245,
      "step": 10739
    },
    {
      "epoch": 0.43358901897456603,
      "grad_norm": 79.7812271118164,
      "learning_rate": 1.2606920451746843e-05,
      "loss": 2.9747,
      "step": 10740
    },
    {
      "epoch": 0.43362939039160275,
      "grad_norm": 101.29683685302734,
      "learning_rate": 1.2605658062572388e-05,
      "loss": 3.2133,
      "step": 10741
    },
    {
      "epoch": 0.4336697618086395,
      "grad_norm": 94.35948944091797,
      "learning_rate": 1.2604395628847266e-05,
      "loss": 4.7031,
      "step": 10742
    },
    {
      "epoch": 0.4337101332256762,
      "grad_norm": 73.80851745605469,
      "learning_rate": 1.260313315059306e-05,
      "loss": 4.5697,
      "step": 10743
    },
    {
      "epoch": 0.433750504642713,
      "grad_norm": 118.27792358398438,
      "learning_rate": 1.260187062783136e-05,
      "loss": 5.7318,
      "step": 10744
    },
    {
      "epoch": 0.4337908760597497,
      "grad_norm": 107.42525482177734,
      "learning_rate": 1.2600608060583741e-05,
      "loss": 4.0415,
      "step": 10745
    },
    {
      "epoch": 0.4338312474767864,
      "grad_norm": 106.40110778808594,
      "learning_rate": 1.2599345448871798e-05,
      "loss": 5.5309,
      "step": 10746
    },
    {
      "epoch": 0.4338716188938232,
      "grad_norm": 94.43321990966797,
      "learning_rate": 1.2598082792717122e-05,
      "loss": 3.3362,
      "step": 10747
    },
    {
      "epoch": 0.4339119903108599,
      "grad_norm": 91.85264587402344,
      "learning_rate": 1.2596820092141295e-05,
      "loss": 3.8386,
      "step": 10748
    },
    {
      "epoch": 0.43395236172789664,
      "grad_norm": 110.45333099365234,
      "learning_rate": 1.2595557347165907e-05,
      "loss": 5.0719,
      "step": 10749
    },
    {
      "epoch": 0.43399273314493336,
      "grad_norm": 158.73497009277344,
      "learning_rate": 1.2594294557812552e-05,
      "loss": 5.3351,
      "step": 10750
    },
    {
      "epoch": 0.43403310456197014,
      "grad_norm": 113.33519744873047,
      "learning_rate": 1.2593031724102818e-05,
      "loss": 5.2139,
      "step": 10751
    },
    {
      "epoch": 0.43407347597900686,
      "grad_norm": 92.89490509033203,
      "learning_rate": 1.2591768846058296e-05,
      "loss": 5.698,
      "step": 10752
    },
    {
      "epoch": 0.4341138473960436,
      "grad_norm": 100.14112854003906,
      "learning_rate": 1.2590505923700577e-05,
      "loss": 6.8297,
      "step": 10753
    },
    {
      "epoch": 0.43415421881308036,
      "grad_norm": 104.54219055175781,
      "learning_rate": 1.258924295705126e-05,
      "loss": 5.1287,
      "step": 10754
    },
    {
      "epoch": 0.4341945902301171,
      "grad_norm": 103.42386627197266,
      "learning_rate": 1.2587979946131932e-05,
      "loss": 6.3913,
      "step": 10755
    },
    {
      "epoch": 0.4342349616471538,
      "grad_norm": 115.10393524169922,
      "learning_rate": 1.258671689096419e-05,
      "loss": 4.565,
      "step": 10756
    },
    {
      "epoch": 0.4342753330641905,
      "grad_norm": 111.28156280517578,
      "learning_rate": 1.2585453791569631e-05,
      "loss": 5.4532,
      "step": 10757
    },
    {
      "epoch": 0.4343157044812273,
      "grad_norm": 101.38463592529297,
      "learning_rate": 1.2584190647969849e-05,
      "loss": 5.5359,
      "step": 10758
    },
    {
      "epoch": 0.434356075898264,
      "grad_norm": 93.68133544921875,
      "learning_rate": 1.258292746018644e-05,
      "loss": 4.0376,
      "step": 10759
    },
    {
      "epoch": 0.43439644731530075,
      "grad_norm": 108.69033813476562,
      "learning_rate": 1.2581664228241007e-05,
      "loss": 4.3559,
      "step": 10760
    },
    {
      "epoch": 0.4344368187323375,
      "grad_norm": 158.77232360839844,
      "learning_rate": 1.258040095215514e-05,
      "loss": 5.267,
      "step": 10761
    },
    {
      "epoch": 0.43447719014937425,
      "grad_norm": 116.16909790039062,
      "learning_rate": 1.2579137631950447e-05,
      "loss": 5.3307,
      "step": 10762
    },
    {
      "epoch": 0.43451756156641097,
      "grad_norm": 116.3614273071289,
      "learning_rate": 1.2577874267648517e-05,
      "loss": 6.3126,
      "step": 10763
    },
    {
      "epoch": 0.4345579329834477,
      "grad_norm": 181.65892028808594,
      "learning_rate": 1.2576610859270964e-05,
      "loss": 4.9484,
      "step": 10764
    },
    {
      "epoch": 0.43459830440048447,
      "grad_norm": 81.61298370361328,
      "learning_rate": 1.2575347406839376e-05,
      "loss": 5.5676,
      "step": 10765
    },
    {
      "epoch": 0.4346386758175212,
      "grad_norm": 77.78543853759766,
      "learning_rate": 1.2574083910375363e-05,
      "loss": 5.6485,
      "step": 10766
    },
    {
      "epoch": 0.4346790472345579,
      "grad_norm": 69.82691192626953,
      "learning_rate": 1.2572820369900525e-05,
      "loss": 5.6035,
      "step": 10767
    },
    {
      "epoch": 0.4347194186515947,
      "grad_norm": 91.30048370361328,
      "learning_rate": 1.2571556785436472e-05,
      "loss": 5.2667,
      "step": 10768
    },
    {
      "epoch": 0.4347597900686314,
      "grad_norm": 109.5865478515625,
      "learning_rate": 1.2570293157004799e-05,
      "loss": 5.5918,
      "step": 10769
    },
    {
      "epoch": 0.43480016148566814,
      "grad_norm": 101.33014678955078,
      "learning_rate": 1.2569029484627113e-05,
      "loss": 5.6374,
      "step": 10770
    },
    {
      "epoch": 0.4348405329027049,
      "grad_norm": 64.78593444824219,
      "learning_rate": 1.2567765768325024e-05,
      "loss": 5.0641,
      "step": 10771
    },
    {
      "epoch": 0.43488090431974163,
      "grad_norm": 76.8919677734375,
      "learning_rate": 1.2566502008120138e-05,
      "loss": 4.9079,
      "step": 10772
    },
    {
      "epoch": 0.43492127573677836,
      "grad_norm": 102.71637725830078,
      "learning_rate": 1.256523820403406e-05,
      "loss": 4.1779,
      "step": 10773
    },
    {
      "epoch": 0.4349616471538151,
      "grad_norm": 105.79178619384766,
      "learning_rate": 1.25639743560884e-05,
      "loss": 3.1725,
      "step": 10774
    },
    {
      "epoch": 0.43500201857085186,
      "grad_norm": 78.19035339355469,
      "learning_rate": 1.2562710464304764e-05,
      "loss": 4.2987,
      "step": 10775
    },
    {
      "epoch": 0.4350423899878886,
      "grad_norm": 111.31887817382812,
      "learning_rate": 1.2561446528704761e-05,
      "loss": 6.3097,
      "step": 10776
    },
    {
      "epoch": 0.4350827614049253,
      "grad_norm": 89.29319763183594,
      "learning_rate": 1.2560182549310007e-05,
      "loss": 5.0263,
      "step": 10777
    },
    {
      "epoch": 0.4351231328219621,
      "grad_norm": 100.36875915527344,
      "learning_rate": 1.255891852614211e-05,
      "loss": 5.7064,
      "step": 10778
    },
    {
      "epoch": 0.4351635042389988,
      "grad_norm": 156.54917907714844,
      "learning_rate": 1.2557654459222681e-05,
      "loss": 5.6281,
      "step": 10779
    },
    {
      "epoch": 0.4352038756560355,
      "grad_norm": 103.43891906738281,
      "learning_rate": 1.2556390348573334e-05,
      "loss": 5.3924,
      "step": 10780
    },
    {
      "epoch": 0.43524424707307224,
      "grad_norm": 180.16677856445312,
      "learning_rate": 1.255512619421568e-05,
      "loss": 4.5633,
      "step": 10781
    },
    {
      "epoch": 0.435284618490109,
      "grad_norm": 125.17741394042969,
      "learning_rate": 1.2553861996171338e-05,
      "loss": 5.346,
      "step": 10782
    },
    {
      "epoch": 0.43532498990714574,
      "grad_norm": 127.41383361816406,
      "learning_rate": 1.2552597754461917e-05,
      "loss": 4.9851,
      "step": 10783
    },
    {
      "epoch": 0.43536536132418246,
      "grad_norm": 92.9611587524414,
      "learning_rate": 1.2551333469109037e-05,
      "loss": 5.7516,
      "step": 10784
    },
    {
      "epoch": 0.43540573274121924,
      "grad_norm": 112.86514282226562,
      "learning_rate": 1.255006914013431e-05,
      "loss": 4.6801,
      "step": 10785
    },
    {
      "epoch": 0.43544610415825596,
      "grad_norm": 105.2659912109375,
      "learning_rate": 1.2548804767559354e-05,
      "loss": 6.2366,
      "step": 10786
    },
    {
      "epoch": 0.4354864755752927,
      "grad_norm": 81.42233276367188,
      "learning_rate": 1.2547540351405792e-05,
      "loss": 5.1096,
      "step": 10787
    },
    {
      "epoch": 0.4355268469923294,
      "grad_norm": 83.662109375,
      "learning_rate": 1.2546275891695238e-05,
      "loss": 3.2938,
      "step": 10788
    },
    {
      "epoch": 0.4355672184093662,
      "grad_norm": 101.56068420410156,
      "learning_rate": 1.254501138844931e-05,
      "loss": 5.3389,
      "step": 10789
    },
    {
      "epoch": 0.4356075898264029,
      "grad_norm": 100.28440856933594,
      "learning_rate": 1.2543746841689637e-05,
      "loss": 5.3242,
      "step": 10790
    },
    {
      "epoch": 0.43564796124343963,
      "grad_norm": 90.29588317871094,
      "learning_rate": 1.2542482251437827e-05,
      "loss": 5.591,
      "step": 10791
    },
    {
      "epoch": 0.4356883326604764,
      "grad_norm": 134.0095977783203,
      "learning_rate": 1.2541217617715508e-05,
      "loss": 7.024,
      "step": 10792
    },
    {
      "epoch": 0.43572870407751313,
      "grad_norm": 84.92144775390625,
      "learning_rate": 1.2539952940544304e-05,
      "loss": 5.1317,
      "step": 10793
    },
    {
      "epoch": 0.43576907549454985,
      "grad_norm": 92.84159851074219,
      "learning_rate": 1.2538688219945836e-05,
      "loss": 5.0896,
      "step": 10794
    },
    {
      "epoch": 0.4358094469115866,
      "grad_norm": 122.05998229980469,
      "learning_rate": 1.2537423455941726e-05,
      "loss": 6.0912,
      "step": 10795
    },
    {
      "epoch": 0.43584981832862335,
      "grad_norm": 71.5400161743164,
      "learning_rate": 1.25361586485536e-05,
      "loss": 3.9428,
      "step": 10796
    },
    {
      "epoch": 0.4358901897456601,
      "grad_norm": 117.75887298583984,
      "learning_rate": 1.2534893797803082e-05,
      "loss": 4.4856,
      "step": 10797
    },
    {
      "epoch": 0.4359305611626968,
      "grad_norm": 117.70970916748047,
      "learning_rate": 1.2533628903711806e-05,
      "loss": 7.108,
      "step": 10798
    },
    {
      "epoch": 0.43597093257973357,
      "grad_norm": 105.39178466796875,
      "learning_rate": 1.2532363966301386e-05,
      "loss": 3.1247,
      "step": 10799
    },
    {
      "epoch": 0.4360113039967703,
      "grad_norm": 114.02446746826172,
      "learning_rate": 1.253109898559346e-05,
      "loss": 6.9467,
      "step": 10800
    },
    {
      "epoch": 0.436051675413807,
      "grad_norm": 83.15216064453125,
      "learning_rate": 1.2529833961609647e-05,
      "loss": 4.9857,
      "step": 10801
    },
    {
      "epoch": 0.43609204683084374,
      "grad_norm": 84.59674072265625,
      "learning_rate": 1.2528568894371584e-05,
      "loss": 5.6289,
      "step": 10802
    },
    {
      "epoch": 0.4361324182478805,
      "grad_norm": 82.19554138183594,
      "learning_rate": 1.2527303783900898e-05,
      "loss": 3.0445,
      "step": 10803
    },
    {
      "epoch": 0.43617278966491724,
      "grad_norm": 118.71350860595703,
      "learning_rate": 1.2526038630219217e-05,
      "loss": 4.0836,
      "step": 10804
    },
    {
      "epoch": 0.43621316108195396,
      "grad_norm": 119.6129379272461,
      "learning_rate": 1.2524773433348173e-05,
      "loss": 4.1159,
      "step": 10805
    },
    {
      "epoch": 0.43625353249899074,
      "grad_norm": 125.63691711425781,
      "learning_rate": 1.2523508193309403e-05,
      "loss": 5.4838,
      "step": 10806
    },
    {
      "epoch": 0.43629390391602746,
      "grad_norm": 106.70641326904297,
      "learning_rate": 1.2522242910124533e-05,
      "loss": 6.24,
      "step": 10807
    },
    {
      "epoch": 0.4363342753330642,
      "grad_norm": 67.84757232666016,
      "learning_rate": 1.2520977583815202e-05,
      "loss": 5.6315,
      "step": 10808
    },
    {
      "epoch": 0.4363746467501009,
      "grad_norm": 122.01983642578125,
      "learning_rate": 1.2519712214403037e-05,
      "loss": 5.4455,
      "step": 10809
    },
    {
      "epoch": 0.4364150181671377,
      "grad_norm": 115.80464935302734,
      "learning_rate": 1.251844680190968e-05,
      "loss": 5.0793,
      "step": 10810
    },
    {
      "epoch": 0.4364553895841744,
      "grad_norm": 107.57453155517578,
      "learning_rate": 1.2517181346356764e-05,
      "loss": 6.2449,
      "step": 10811
    },
    {
      "epoch": 0.4364957610012111,
      "grad_norm": 102.88275909423828,
      "learning_rate": 1.2515915847765925e-05,
      "loss": 7.2903,
      "step": 10812
    },
    {
      "epoch": 0.4365361324182479,
      "grad_norm": 94.3470687866211,
      "learning_rate": 1.2514650306158797e-05,
      "loss": 5.5254,
      "step": 10813
    },
    {
      "epoch": 0.4365765038352846,
      "grad_norm": 133.80393981933594,
      "learning_rate": 1.2513384721557024e-05,
      "loss": 6.7238,
      "step": 10814
    },
    {
      "epoch": 0.43661687525232135,
      "grad_norm": 82.1087417602539,
      "learning_rate": 1.251211909398224e-05,
      "loss": 5.2795,
      "step": 10815
    },
    {
      "epoch": 0.43665724666935807,
      "grad_norm": 161.96958923339844,
      "learning_rate": 1.2510853423456084e-05,
      "loss": 4.9097,
      "step": 10816
    },
    {
      "epoch": 0.43669761808639485,
      "grad_norm": 78.62599182128906,
      "learning_rate": 1.25095877100002e-05,
      "loss": 3.874,
      "step": 10817
    },
    {
      "epoch": 0.43673798950343157,
      "grad_norm": 94.91002655029297,
      "learning_rate": 1.2508321953636227e-05,
      "loss": 5.6394,
      "step": 10818
    },
    {
      "epoch": 0.4367783609204683,
      "grad_norm": 88.46383666992188,
      "learning_rate": 1.2507056154385804e-05,
      "loss": 4.8954,
      "step": 10819
    },
    {
      "epoch": 0.43681873233750507,
      "grad_norm": 105.80547332763672,
      "learning_rate": 1.2505790312270579e-05,
      "loss": 5.3529,
      "step": 10820
    },
    {
      "epoch": 0.4368591037545418,
      "grad_norm": 146.56301879882812,
      "learning_rate": 1.2504524427312185e-05,
      "loss": 6.6045,
      "step": 10821
    },
    {
      "epoch": 0.4368994751715785,
      "grad_norm": 109.57958221435547,
      "learning_rate": 1.2503258499532278e-05,
      "loss": 4.2846,
      "step": 10822
    },
    {
      "epoch": 0.43693984658861523,
      "grad_norm": 104.4039535522461,
      "learning_rate": 1.2501992528952492e-05,
      "loss": 3.4846,
      "step": 10823
    },
    {
      "epoch": 0.436980218005652,
      "grad_norm": 88.25065612792969,
      "learning_rate": 1.250072651559448e-05,
      "loss": 4.3841,
      "step": 10824
    },
    {
      "epoch": 0.43702058942268873,
      "grad_norm": 80.32769012451172,
      "learning_rate": 1.2499460459479882e-05,
      "loss": 5.5427,
      "step": 10825
    },
    {
      "epoch": 0.43706096083972545,
      "grad_norm": 74.30186462402344,
      "learning_rate": 1.2498194360630344e-05,
      "loss": 5.0725,
      "step": 10826
    },
    {
      "epoch": 0.43710133225676223,
      "grad_norm": 154.18507385253906,
      "learning_rate": 1.2496928219067518e-05,
      "loss": 7.8031,
      "step": 10827
    },
    {
      "epoch": 0.43714170367379895,
      "grad_norm": 110.82579803466797,
      "learning_rate": 1.2495662034813055e-05,
      "loss": 6.1161,
      "step": 10828
    },
    {
      "epoch": 0.4371820750908357,
      "grad_norm": 126.45730590820312,
      "learning_rate": 1.2494395807888593e-05,
      "loss": 3.9306,
      "step": 10829
    },
    {
      "epoch": 0.43722244650787245,
      "grad_norm": 85.13944244384766,
      "learning_rate": 1.249312953831579e-05,
      "loss": 5.3764,
      "step": 10830
    },
    {
      "epoch": 0.4372628179249092,
      "grad_norm": 75.59607696533203,
      "learning_rate": 1.2491863226116293e-05,
      "loss": 4.2538,
      "step": 10831
    },
    {
      "epoch": 0.4373031893419459,
      "grad_norm": 75.05252075195312,
      "learning_rate": 1.2490596871311755e-05,
      "loss": 5.4922,
      "step": 10832
    },
    {
      "epoch": 0.4373435607589826,
      "grad_norm": 69.90910339355469,
      "learning_rate": 1.2489330473923824e-05,
      "loss": 3.3374,
      "step": 10833
    },
    {
      "epoch": 0.4373839321760194,
      "grad_norm": 96.03634643554688,
      "learning_rate": 1.248806403397416e-05,
      "loss": 4.8366,
      "step": 10834
    },
    {
      "epoch": 0.4374243035930561,
      "grad_norm": 302.17486572265625,
      "learning_rate": 1.2486797551484404e-05,
      "loss": 6.0816,
      "step": 10835
    },
    {
      "epoch": 0.43746467501009284,
      "grad_norm": 136.07130432128906,
      "learning_rate": 1.2485531026476221e-05,
      "loss": 2.6492,
      "step": 10836
    },
    {
      "epoch": 0.4375050464271296,
      "grad_norm": 131.36070251464844,
      "learning_rate": 1.248426445897126e-05,
      "loss": 4.6371,
      "step": 10837
    },
    {
      "epoch": 0.43754541784416634,
      "grad_norm": 85.32100677490234,
      "learning_rate": 1.2482997848991178e-05,
      "loss": 6.2336,
      "step": 10838
    },
    {
      "epoch": 0.43758578926120306,
      "grad_norm": 95.02273559570312,
      "learning_rate": 1.2481731196557633e-05,
      "loss": 5.193,
      "step": 10839
    },
    {
      "epoch": 0.4376261606782398,
      "grad_norm": 190.07847595214844,
      "learning_rate": 1.2480464501692278e-05,
      "loss": 6.2547,
      "step": 10840
    },
    {
      "epoch": 0.43766653209527656,
      "grad_norm": 66.79094696044922,
      "learning_rate": 1.2479197764416775e-05,
      "loss": 2.1289,
      "step": 10841
    },
    {
      "epoch": 0.4377069035123133,
      "grad_norm": 227.07398986816406,
      "learning_rate": 1.2477930984752778e-05,
      "loss": 4.3812,
      "step": 10842
    },
    {
      "epoch": 0.43774727492935,
      "grad_norm": 80.19287109375,
      "learning_rate": 1.2476664162721948e-05,
      "loss": 3.8932,
      "step": 10843
    },
    {
      "epoch": 0.4377876463463868,
      "grad_norm": 169.577392578125,
      "learning_rate": 1.2475397298345942e-05,
      "loss": 4.8726,
      "step": 10844
    },
    {
      "epoch": 0.4378280177634235,
      "grad_norm": 235.1566619873047,
      "learning_rate": 1.2474130391646422e-05,
      "loss": 5.0173,
      "step": 10845
    },
    {
      "epoch": 0.4378683891804602,
      "grad_norm": 132.87936401367188,
      "learning_rate": 1.2472863442645052e-05,
      "loss": 5.324,
      "step": 10846
    },
    {
      "epoch": 0.43790876059749695,
      "grad_norm": 68.29861450195312,
      "learning_rate": 1.2471596451363492e-05,
      "loss": 4.0393,
      "step": 10847
    },
    {
      "epoch": 0.4379491320145337,
      "grad_norm": 90.35537719726562,
      "learning_rate": 1.2470329417823403e-05,
      "loss": 5.6239,
      "step": 10848
    },
    {
      "epoch": 0.43798950343157045,
      "grad_norm": 93.5866928100586,
      "learning_rate": 1.2469062342046451e-05,
      "loss": 5.0508,
      "step": 10849
    },
    {
      "epoch": 0.43802987484860717,
      "grad_norm": 142.91311645507812,
      "learning_rate": 1.24677952240543e-05,
      "loss": 4.8062,
      "step": 10850
    },
    {
      "epoch": 0.43807024626564395,
      "grad_norm": 165.865234375,
      "learning_rate": 1.2466528063868612e-05,
      "loss": 5.2735,
      "step": 10851
    },
    {
      "epoch": 0.43811061768268067,
      "grad_norm": 100.97811889648438,
      "learning_rate": 1.2465260861511052e-05,
      "loss": 5.0065,
      "step": 10852
    },
    {
      "epoch": 0.4381509890997174,
      "grad_norm": 118.28631591796875,
      "learning_rate": 1.2463993617003292e-05,
      "loss": 5.6087,
      "step": 10853
    },
    {
      "epoch": 0.4381913605167541,
      "grad_norm": 82.0665512084961,
      "learning_rate": 1.246272633036699e-05,
      "loss": 5.3035,
      "step": 10854
    },
    {
      "epoch": 0.4382317319337909,
      "grad_norm": 132.1261444091797,
      "learning_rate": 1.2461459001623822e-05,
      "loss": 5.6636,
      "step": 10855
    },
    {
      "epoch": 0.4382721033508276,
      "grad_norm": 116.03626251220703,
      "learning_rate": 1.2460191630795451e-05,
      "loss": 4.5001,
      "step": 10856
    },
    {
      "epoch": 0.43831247476786434,
      "grad_norm": 90.73950958251953,
      "learning_rate": 1.245892421790355e-05,
      "loss": 5.35,
      "step": 10857
    },
    {
      "epoch": 0.4383528461849011,
      "grad_norm": 106.33438873291016,
      "learning_rate": 1.2457656762969785e-05,
      "loss": 5.6676,
      "step": 10858
    },
    {
      "epoch": 0.43839321760193783,
      "grad_norm": 122.28158569335938,
      "learning_rate": 1.2456389266015832e-05,
      "loss": 5.6568,
      "step": 10859
    },
    {
      "epoch": 0.43843358901897456,
      "grad_norm": 101.94876861572266,
      "learning_rate": 1.2455121727063352e-05,
      "loss": 6.2818,
      "step": 10860
    },
    {
      "epoch": 0.4384739604360113,
      "grad_norm": 94.98487854003906,
      "learning_rate": 1.2453854146134031e-05,
      "loss": 6.8877,
      "step": 10861
    },
    {
      "epoch": 0.43851433185304806,
      "grad_norm": 79.50425720214844,
      "learning_rate": 1.245258652324953e-05,
      "loss": 5.4377,
      "step": 10862
    },
    {
      "epoch": 0.4385547032700848,
      "grad_norm": 73.18757629394531,
      "learning_rate": 1.2451318858431529e-05,
      "loss": 6.3847,
      "step": 10863
    },
    {
      "epoch": 0.4385950746871215,
      "grad_norm": 102.80915069580078,
      "learning_rate": 1.2450051151701696e-05,
      "loss": 3.8584,
      "step": 10864
    },
    {
      "epoch": 0.4386354461041583,
      "grad_norm": 151.93374633789062,
      "learning_rate": 1.2448783403081713e-05,
      "loss": 7.0552,
      "step": 10865
    },
    {
      "epoch": 0.438675817521195,
      "grad_norm": 121.23501586914062,
      "learning_rate": 1.2447515612593248e-05,
      "loss": 4.0782,
      "step": 10866
    },
    {
      "epoch": 0.4387161889382317,
      "grad_norm": 85.75382995605469,
      "learning_rate": 1.2446247780257987e-05,
      "loss": 5.4178,
      "step": 10867
    },
    {
      "epoch": 0.43875656035526844,
      "grad_norm": 73.0689926147461,
      "learning_rate": 1.2444979906097598e-05,
      "loss": 4.4701,
      "step": 10868
    },
    {
      "epoch": 0.4387969317723052,
      "grad_norm": 86.23473358154297,
      "learning_rate": 1.2443711990133764e-05,
      "loss": 4.5745,
      "step": 10869
    },
    {
      "epoch": 0.43883730318934194,
      "grad_norm": 96.80155944824219,
      "learning_rate": 1.244244403238816e-05,
      "loss": 6.1686,
      "step": 10870
    },
    {
      "epoch": 0.43887767460637866,
      "grad_norm": 109.54499816894531,
      "learning_rate": 1.2441176032882468e-05,
      "loss": 4.9742,
      "step": 10871
    },
    {
      "epoch": 0.43891804602341544,
      "grad_norm": 90.48492431640625,
      "learning_rate": 1.2439907991638363e-05,
      "loss": 5.7537,
      "step": 10872
    },
    {
      "epoch": 0.43895841744045216,
      "grad_norm": 84.64985656738281,
      "learning_rate": 1.2438639908677532e-05,
      "loss": 5.4287,
      "step": 10873
    },
    {
      "epoch": 0.4389987888574889,
      "grad_norm": 117.62967681884766,
      "learning_rate": 1.2437371784021653e-05,
      "loss": 4.8242,
      "step": 10874
    },
    {
      "epoch": 0.4390391602745256,
      "grad_norm": 99.3263168334961,
      "learning_rate": 1.2436103617692405e-05,
      "loss": 4.9084,
      "step": 10875
    },
    {
      "epoch": 0.4390795316915624,
      "grad_norm": 109.70679473876953,
      "learning_rate": 1.2434835409711477e-05,
      "loss": 4.5602,
      "step": 10876
    },
    {
      "epoch": 0.4391199031085991,
      "grad_norm": 86.71737670898438,
      "learning_rate": 1.243356716010055e-05,
      "loss": 5.666,
      "step": 10877
    },
    {
      "epoch": 0.43916027452563583,
      "grad_norm": 111.1494140625,
      "learning_rate": 1.2432298868881306e-05,
      "loss": 4.4363,
      "step": 10878
    },
    {
      "epoch": 0.4392006459426726,
      "grad_norm": 163.16976928710938,
      "learning_rate": 1.243103053607543e-05,
      "loss": 4.6336,
      "step": 10879
    },
    {
      "epoch": 0.43924101735970933,
      "grad_norm": 113.74940490722656,
      "learning_rate": 1.242976216170461e-05,
      "loss": 5.2163,
      "step": 10880
    },
    {
      "epoch": 0.43928138877674605,
      "grad_norm": 130.2457733154297,
      "learning_rate": 1.2428493745790533e-05,
      "loss": 4.9935,
      "step": 10881
    },
    {
      "epoch": 0.43932176019378283,
      "grad_norm": 92.65779113769531,
      "learning_rate": 1.2427225288354881e-05,
      "loss": 6.1495,
      "step": 10882
    },
    {
      "epoch": 0.43936213161081955,
      "grad_norm": 86.8188247680664,
      "learning_rate": 1.2425956789419344e-05,
      "loss": 4.6022,
      "step": 10883
    },
    {
      "epoch": 0.4394025030278563,
      "grad_norm": 97.61709594726562,
      "learning_rate": 1.2424688249005611e-05,
      "loss": 5.0752,
      "step": 10884
    },
    {
      "epoch": 0.439442874444893,
      "grad_norm": 107.37088012695312,
      "learning_rate": 1.2423419667135373e-05,
      "loss": 6.4315,
      "step": 10885
    },
    {
      "epoch": 0.43948324586192977,
      "grad_norm": 165.86524963378906,
      "learning_rate": 1.2422151043830315e-05,
      "loss": 5.851,
      "step": 10886
    },
    {
      "epoch": 0.4395236172789665,
      "grad_norm": 101.64979553222656,
      "learning_rate": 1.2420882379112134e-05,
      "loss": 7.5216,
      "step": 10887
    },
    {
      "epoch": 0.4395639886960032,
      "grad_norm": 97.80375671386719,
      "learning_rate": 1.2419613673002514e-05,
      "loss": 5.7631,
      "step": 10888
    },
    {
      "epoch": 0.43960436011304,
      "grad_norm": 75.51022338867188,
      "learning_rate": 1.2418344925523153e-05,
      "loss": 6.4919,
      "step": 10889
    },
    {
      "epoch": 0.4396447315300767,
      "grad_norm": 140.8380584716797,
      "learning_rate": 1.241707613669574e-05,
      "loss": 5.9627,
      "step": 10890
    },
    {
      "epoch": 0.43968510294711344,
      "grad_norm": 135.2897186279297,
      "learning_rate": 1.2415807306541972e-05,
      "loss": 3.9629,
      "step": 10891
    },
    {
      "epoch": 0.43972547436415016,
      "grad_norm": 80.45240020751953,
      "learning_rate": 1.2414538435083537e-05,
      "loss": 6.0861,
      "step": 10892
    },
    {
      "epoch": 0.43976584578118694,
      "grad_norm": 84.05445098876953,
      "learning_rate": 1.2413269522342136e-05,
      "loss": 5.9295,
      "step": 10893
    },
    {
      "epoch": 0.43980621719822366,
      "grad_norm": 86.94470977783203,
      "learning_rate": 1.2412000568339457e-05,
      "loss": 5.378,
      "step": 10894
    },
    {
      "epoch": 0.4398465886152604,
      "grad_norm": 115.47087097167969,
      "learning_rate": 1.2410731573097201e-05,
      "loss": 5.4954,
      "step": 10895
    },
    {
      "epoch": 0.43988696003229716,
      "grad_norm": 86.40447235107422,
      "learning_rate": 1.2409462536637066e-05,
      "loss": 6.207,
      "step": 10896
    },
    {
      "epoch": 0.4399273314493339,
      "grad_norm": 105.03781127929688,
      "learning_rate": 1.240819345898075e-05,
      "loss": 4.3108,
      "step": 10897
    },
    {
      "epoch": 0.4399677028663706,
      "grad_norm": 84.15904235839844,
      "learning_rate": 1.2406924340149948e-05,
      "loss": 7.3406,
      "step": 10898
    },
    {
      "epoch": 0.4400080742834073,
      "grad_norm": 125.98786926269531,
      "learning_rate": 1.2405655180166363e-05,
      "loss": 4.5324,
      "step": 10899
    },
    {
      "epoch": 0.4400484457004441,
      "grad_norm": 129.4600830078125,
      "learning_rate": 1.2404385979051687e-05,
      "loss": 2.7591,
      "step": 10900
    },
    {
      "epoch": 0.4400888171174808,
      "grad_norm": 124.07709503173828,
      "learning_rate": 1.2403116736827629e-05,
      "loss": 4.8933,
      "step": 10901
    },
    {
      "epoch": 0.44012918853451755,
      "grad_norm": 80.23975372314453,
      "learning_rate": 1.2401847453515888e-05,
      "loss": 5.9152,
      "step": 10902
    },
    {
      "epoch": 0.4401695599515543,
      "grad_norm": 85.71908569335938,
      "learning_rate": 1.2400578129138161e-05,
      "loss": 4.2064,
      "step": 10903
    },
    {
      "epoch": 0.44020993136859105,
      "grad_norm": 85.02687072753906,
      "learning_rate": 1.2399308763716154e-05,
      "loss": 5.5481,
      "step": 10904
    },
    {
      "epoch": 0.44025030278562777,
      "grad_norm": 121.19576263427734,
      "learning_rate": 1.2398039357271569e-05,
      "loss": 4.3539,
      "step": 10905
    },
    {
      "epoch": 0.4402906742026645,
      "grad_norm": 106.2219009399414,
      "learning_rate": 1.2396769909826112e-05,
      "loss": 6.1519,
      "step": 10906
    },
    {
      "epoch": 0.44033104561970127,
      "grad_norm": 363.0920715332031,
      "learning_rate": 1.2395500421401487e-05,
      "loss": 6.6548,
      "step": 10907
    },
    {
      "epoch": 0.440371417036738,
      "grad_norm": 130.03187561035156,
      "learning_rate": 1.2394230892019398e-05,
      "loss": 4.186,
      "step": 10908
    },
    {
      "epoch": 0.4404117884537747,
      "grad_norm": 79.46842193603516,
      "learning_rate": 1.2392961321701553e-05,
      "loss": 4.0621,
      "step": 10909
    },
    {
      "epoch": 0.4404521598708115,
      "grad_norm": 69.26293182373047,
      "learning_rate": 1.2391691710469656e-05,
      "loss": 4.7626,
      "step": 10910
    },
    {
      "epoch": 0.4404925312878482,
      "grad_norm": 159.1634063720703,
      "learning_rate": 1.2390422058345418e-05,
      "loss": 3.9328,
      "step": 10911
    },
    {
      "epoch": 0.44053290270488493,
      "grad_norm": 90.47792053222656,
      "learning_rate": 1.238915236535054e-05,
      "loss": 5.9756,
      "step": 10912
    },
    {
      "epoch": 0.44057327412192165,
      "grad_norm": 71.5369873046875,
      "learning_rate": 1.2387882631506743e-05,
      "loss": 5.534,
      "step": 10913
    },
    {
      "epoch": 0.44061364553895843,
      "grad_norm": 133.6231689453125,
      "learning_rate": 1.2386612856835723e-05,
      "loss": 3.7723,
      "step": 10914
    },
    {
      "epoch": 0.44065401695599515,
      "grad_norm": 100.48680114746094,
      "learning_rate": 1.2385343041359197e-05,
      "loss": 7.3447,
      "step": 10915
    },
    {
      "epoch": 0.4406943883730319,
      "grad_norm": 86.12537384033203,
      "learning_rate": 1.2384073185098874e-05,
      "loss": 4.9822,
      "step": 10916
    },
    {
      "epoch": 0.44073475979006865,
      "grad_norm": 98.60801696777344,
      "learning_rate": 1.238280328807647e-05,
      "loss": 4.7994,
      "step": 10917
    },
    {
      "epoch": 0.4407751312071054,
      "grad_norm": 100.92731475830078,
      "learning_rate": 1.2381533350313693e-05,
      "loss": 4.6548,
      "step": 10918
    },
    {
      "epoch": 0.4408155026241421,
      "grad_norm": 90.24503326416016,
      "learning_rate": 1.2380263371832257e-05,
      "loss": 5.8936,
      "step": 10919
    },
    {
      "epoch": 0.4408558740411788,
      "grad_norm": 119.2989501953125,
      "learning_rate": 1.2378993352653876e-05,
      "loss": 6.0066,
      "step": 10920
    },
    {
      "epoch": 0.4408962454582156,
      "grad_norm": 98.45545196533203,
      "learning_rate": 1.2377723292800265e-05,
      "loss": 5.6915,
      "step": 10921
    },
    {
      "epoch": 0.4409366168752523,
      "grad_norm": 95.23027038574219,
      "learning_rate": 1.2376453192293136e-05,
      "loss": 4.923,
      "step": 10922
    },
    {
      "epoch": 0.44097698829228904,
      "grad_norm": 95.19053649902344,
      "learning_rate": 1.2375183051154207e-05,
      "loss": 5.5941,
      "step": 10923
    },
    {
      "epoch": 0.4410173597093258,
      "grad_norm": 130.83029174804688,
      "learning_rate": 1.2373912869405196e-05,
      "loss": 5.9056,
      "step": 10924
    },
    {
      "epoch": 0.44105773112636254,
      "grad_norm": 152.98843383789062,
      "learning_rate": 1.2372642647067817e-05,
      "loss": 5.5242,
      "step": 10925
    },
    {
      "epoch": 0.44109810254339926,
      "grad_norm": 91.3885726928711,
      "learning_rate": 1.2371372384163787e-05,
      "loss": 5.8705,
      "step": 10926
    },
    {
      "epoch": 0.441138473960436,
      "grad_norm": 83.53356170654297,
      "learning_rate": 1.237010208071483e-05,
      "loss": 3.4166,
      "step": 10927
    },
    {
      "epoch": 0.44117884537747276,
      "grad_norm": 130.89736938476562,
      "learning_rate": 1.2368831736742662e-05,
      "loss": 3.5331,
      "step": 10928
    },
    {
      "epoch": 0.4412192167945095,
      "grad_norm": 85.18113708496094,
      "learning_rate": 1.2367561352269002e-05,
      "loss": 6.0701,
      "step": 10929
    },
    {
      "epoch": 0.4412595882115462,
      "grad_norm": 92.86490631103516,
      "learning_rate": 1.2366290927315572e-05,
      "loss": 4.3025,
      "step": 10930
    },
    {
      "epoch": 0.441299959628583,
      "grad_norm": 109.28509521484375,
      "learning_rate": 1.2365020461904094e-05,
      "loss": 5.0287,
      "step": 10931
    },
    {
      "epoch": 0.4413403310456197,
      "grad_norm": 293.0802917480469,
      "learning_rate": 1.2363749956056286e-05,
      "loss": 5.7977,
      "step": 10932
    },
    {
      "epoch": 0.4413807024626564,
      "grad_norm": 158.1237030029297,
      "learning_rate": 1.2362479409793876e-05,
      "loss": 5.3943,
      "step": 10933
    },
    {
      "epoch": 0.4414210738796932,
      "grad_norm": 105.9314956665039,
      "learning_rate": 1.2361208823138581e-05,
      "loss": 4.7424,
      "step": 10934
    },
    {
      "epoch": 0.4414614452967299,
      "grad_norm": 85.3238754272461,
      "learning_rate": 1.2359938196112133e-05,
      "loss": 4.9629,
      "step": 10935
    },
    {
      "epoch": 0.44150181671376665,
      "grad_norm": 71.0063705444336,
      "learning_rate": 1.235866752873625e-05,
      "loss": 6.0233,
      "step": 10936
    },
    {
      "epoch": 0.44154218813080337,
      "grad_norm": 112.73417663574219,
      "learning_rate": 1.2357396821032663e-05,
      "loss": 4.8583,
      "step": 10937
    },
    {
      "epoch": 0.44158255954784015,
      "grad_norm": 90.76774597167969,
      "learning_rate": 1.2356126073023091e-05,
      "loss": 5.6106,
      "step": 10938
    },
    {
      "epoch": 0.44162293096487687,
      "grad_norm": 126.82186889648438,
      "learning_rate": 1.2354855284729271e-05,
      "loss": 4.502,
      "step": 10939
    },
    {
      "epoch": 0.4416633023819136,
      "grad_norm": 97.66627502441406,
      "learning_rate": 1.235358445617292e-05,
      "loss": 4.6649,
      "step": 10940
    },
    {
      "epoch": 0.44170367379895037,
      "grad_norm": 89.17770385742188,
      "learning_rate": 1.2352313587375776e-05,
      "loss": 4.9557,
      "step": 10941
    },
    {
      "epoch": 0.4417440452159871,
      "grad_norm": 67.388427734375,
      "learning_rate": 1.2351042678359555e-05,
      "loss": 5.7873,
      "step": 10942
    },
    {
      "epoch": 0.4417844166330238,
      "grad_norm": 120.82470703125,
      "learning_rate": 1.2349771729146e-05,
      "loss": 5.8808,
      "step": 10943
    },
    {
      "epoch": 0.44182478805006054,
      "grad_norm": 74.77674865722656,
      "learning_rate": 1.2348500739756835e-05,
      "loss": 4.8954,
      "step": 10944
    },
    {
      "epoch": 0.4418651594670973,
      "grad_norm": 115.0797348022461,
      "learning_rate": 1.2347229710213789e-05,
      "loss": 4.1861,
      "step": 10945
    },
    {
      "epoch": 0.44190553088413403,
      "grad_norm": 93.34678649902344,
      "learning_rate": 1.2345958640538595e-05,
      "loss": 4.3183,
      "step": 10946
    },
    {
      "epoch": 0.44194590230117076,
      "grad_norm": 115.51158142089844,
      "learning_rate": 1.234468753075299e-05,
      "loss": 7.9178,
      "step": 10947
    },
    {
      "epoch": 0.44198627371820753,
      "grad_norm": 94.54987335205078,
      "learning_rate": 1.2343416380878703e-05,
      "loss": 5.5225,
      "step": 10948
    },
    {
      "epoch": 0.44202664513524426,
      "grad_norm": 73.56192016601562,
      "learning_rate": 1.2342145190937469e-05,
      "loss": 5.883,
      "step": 10949
    },
    {
      "epoch": 0.442067016552281,
      "grad_norm": 136.97874450683594,
      "learning_rate": 1.2340873960951018e-05,
      "loss": 5.7961,
      "step": 10950
    },
    {
      "epoch": 0.4421073879693177,
      "grad_norm": 92.91134643554688,
      "learning_rate": 1.2339602690941092e-05,
      "loss": 5.6131,
      "step": 10951
    },
    {
      "epoch": 0.4421477593863545,
      "grad_norm": 106.15806579589844,
      "learning_rate": 1.2338331380929422e-05,
      "loss": 6.0106,
      "step": 10952
    },
    {
      "epoch": 0.4421881308033912,
      "grad_norm": 79.7496109008789,
      "learning_rate": 1.2337060030937744e-05,
      "loss": 4.536,
      "step": 10953
    },
    {
      "epoch": 0.4422285022204279,
      "grad_norm": 88.46705627441406,
      "learning_rate": 1.2335788640987799e-05,
      "loss": 4.983,
      "step": 10954
    },
    {
      "epoch": 0.4422688736374647,
      "grad_norm": 75.11612701416016,
      "learning_rate": 1.2334517211101322e-05,
      "loss": 6.7476,
      "step": 10955
    },
    {
      "epoch": 0.4423092450545014,
      "grad_norm": 76.54798889160156,
      "learning_rate": 1.2333245741300053e-05,
      "loss": 5.2966,
      "step": 10956
    },
    {
      "epoch": 0.44234961647153814,
      "grad_norm": 100.03357696533203,
      "learning_rate": 1.233197423160573e-05,
      "loss": 2.9961,
      "step": 10957
    },
    {
      "epoch": 0.44238998788857486,
      "grad_norm": 82.10073852539062,
      "learning_rate": 1.2330702682040095e-05,
      "loss": 5.637,
      "step": 10958
    },
    {
      "epoch": 0.44243035930561164,
      "grad_norm": 179.6915283203125,
      "learning_rate": 1.2329431092624883e-05,
      "loss": 6.7953,
      "step": 10959
    },
    {
      "epoch": 0.44247073072264836,
      "grad_norm": 84.28987884521484,
      "learning_rate": 1.2328159463381842e-05,
      "loss": 6.4782,
      "step": 10960
    },
    {
      "epoch": 0.4425111021396851,
      "grad_norm": 102.47725677490234,
      "learning_rate": 1.2326887794332709e-05,
      "loss": 6.3105,
      "step": 10961
    },
    {
      "epoch": 0.44255147355672186,
      "grad_norm": 75.31526184082031,
      "learning_rate": 1.232561608549923e-05,
      "loss": 5.0486,
      "step": 10962
    },
    {
      "epoch": 0.4425918449737586,
      "grad_norm": 76.99453735351562,
      "learning_rate": 1.2324344336903145e-05,
      "loss": 4.7041,
      "step": 10963
    },
    {
      "epoch": 0.4426322163907953,
      "grad_norm": 99.54042053222656,
      "learning_rate": 1.23230725485662e-05,
      "loss": 6.0105,
      "step": 10964
    },
    {
      "epoch": 0.44267258780783203,
      "grad_norm": 107.26725769042969,
      "learning_rate": 1.232180072051014e-05,
      "loss": 6.1803,
      "step": 10965
    },
    {
      "epoch": 0.4427129592248688,
      "grad_norm": 76.52234649658203,
      "learning_rate": 1.2320528852756708e-05,
      "loss": 5.8253,
      "step": 10966
    },
    {
      "epoch": 0.44275333064190553,
      "grad_norm": 91.7452392578125,
      "learning_rate": 1.2319256945327653e-05,
      "loss": 5.4268,
      "step": 10967
    },
    {
      "epoch": 0.44279370205894225,
      "grad_norm": 92.46363067626953,
      "learning_rate": 1.231798499824472e-05,
      "loss": 4.7418,
      "step": 10968
    },
    {
      "epoch": 0.44283407347597903,
      "grad_norm": 68.11386108398438,
      "learning_rate": 1.2316713011529656e-05,
      "loss": 5.8888,
      "step": 10969
    },
    {
      "epoch": 0.44287444489301575,
      "grad_norm": 95.47237396240234,
      "learning_rate": 1.2315440985204208e-05,
      "loss": 4.1296,
      "step": 10970
    },
    {
      "epoch": 0.4429148163100525,
      "grad_norm": 146.05931091308594,
      "learning_rate": 1.2314168919290128e-05,
      "loss": 5.0962,
      "step": 10971
    },
    {
      "epoch": 0.4429551877270892,
      "grad_norm": 100.58576202392578,
      "learning_rate": 1.2312896813809162e-05,
      "loss": 4.978,
      "step": 10972
    },
    {
      "epoch": 0.44299555914412597,
      "grad_norm": 112.52092742919922,
      "learning_rate": 1.2311624668783063e-05,
      "loss": 4.2682,
      "step": 10973
    },
    {
      "epoch": 0.4430359305611627,
      "grad_norm": 93.15788269042969,
      "learning_rate": 1.2310352484233576e-05,
      "loss": 5.4471,
      "step": 10974
    },
    {
      "epoch": 0.4430763019781994,
      "grad_norm": 109.01019287109375,
      "learning_rate": 1.2309080260182462e-05,
      "loss": 4.382,
      "step": 10975
    },
    {
      "epoch": 0.4431166733952362,
      "grad_norm": 99.36241149902344,
      "learning_rate": 1.2307807996651465e-05,
      "loss": 5.7095,
      "step": 10976
    },
    {
      "epoch": 0.4431570448122729,
      "grad_norm": 101.49544525146484,
      "learning_rate": 1.2306535693662342e-05,
      "loss": 5.1459,
      "step": 10977
    },
    {
      "epoch": 0.44319741622930964,
      "grad_norm": 86.58634185791016,
      "learning_rate": 1.2305263351236843e-05,
      "loss": 6.1857,
      "step": 10978
    },
    {
      "epoch": 0.44323778764634636,
      "grad_norm": 136.0492401123047,
      "learning_rate": 1.2303990969396727e-05,
      "loss": 4.0142,
      "step": 10979
    },
    {
      "epoch": 0.44327815906338314,
      "grad_norm": 93.88790893554688,
      "learning_rate": 1.230271854816374e-05,
      "loss": 4.3317,
      "step": 10980
    },
    {
      "epoch": 0.44331853048041986,
      "grad_norm": 108.66222381591797,
      "learning_rate": 1.230144608755965e-05,
      "loss": 6.2606,
      "step": 10981
    },
    {
      "epoch": 0.4433589018974566,
      "grad_norm": 70.46728515625,
      "learning_rate": 1.23001735876062e-05,
      "loss": 6.4078,
      "step": 10982
    },
    {
      "epoch": 0.44339927331449336,
      "grad_norm": 82.85064697265625,
      "learning_rate": 1.2298901048325155e-05,
      "loss": 5.3066,
      "step": 10983
    },
    {
      "epoch": 0.4434396447315301,
      "grad_norm": 96.42451477050781,
      "learning_rate": 1.2297628469738273e-05,
      "loss": 4.6747,
      "step": 10984
    },
    {
      "epoch": 0.4434800161485668,
      "grad_norm": 107.80743408203125,
      "learning_rate": 1.2296355851867308e-05,
      "loss": 5.5921,
      "step": 10985
    },
    {
      "epoch": 0.4435203875656036,
      "grad_norm": 103.29229736328125,
      "learning_rate": 1.2295083194734017e-05,
      "loss": 5.2252,
      "step": 10986
    },
    {
      "epoch": 0.4435607589826403,
      "grad_norm": 137.46615600585938,
      "learning_rate": 1.2293810498360167e-05,
      "loss": 7.5373,
      "step": 10987
    },
    {
      "epoch": 0.443601130399677,
      "grad_norm": 101.71404266357422,
      "learning_rate": 1.2292537762767512e-05,
      "loss": 5.9006,
      "step": 10988
    },
    {
      "epoch": 0.44364150181671375,
      "grad_norm": 85.08946228027344,
      "learning_rate": 1.2291264987977815e-05,
      "loss": 4.1485,
      "step": 10989
    },
    {
      "epoch": 0.4436818732337505,
      "grad_norm": 88.66033935546875,
      "learning_rate": 1.2289992174012837e-05,
      "loss": 4.766,
      "step": 10990
    },
    {
      "epoch": 0.44372224465078725,
      "grad_norm": 94.27982330322266,
      "learning_rate": 1.2288719320894343e-05,
      "loss": 5.3313,
      "step": 10991
    },
    {
      "epoch": 0.44376261606782397,
      "grad_norm": 71.8203353881836,
      "learning_rate": 1.228744642864409e-05,
      "loss": 3.6367,
      "step": 10992
    },
    {
      "epoch": 0.44380298748486074,
      "grad_norm": 99.87458038330078,
      "learning_rate": 1.2286173497283845e-05,
      "loss": 5.6268,
      "step": 10993
    },
    {
      "epoch": 0.44384335890189747,
      "grad_norm": 80.11166381835938,
      "learning_rate": 1.228490052683537e-05,
      "loss": 5.1172,
      "step": 10994
    },
    {
      "epoch": 0.4438837303189342,
      "grad_norm": 98.38223266601562,
      "learning_rate": 1.2283627517320435e-05,
      "loss": 5.763,
      "step": 10995
    },
    {
      "epoch": 0.4439241017359709,
      "grad_norm": 87.49798583984375,
      "learning_rate": 1.22823544687608e-05,
      "loss": 5.0125,
      "step": 10996
    },
    {
      "epoch": 0.4439644731530077,
      "grad_norm": 126.63861846923828,
      "learning_rate": 1.2281081381178236e-05,
      "loss": 5.8864,
      "step": 10997
    },
    {
      "epoch": 0.4440048445700444,
      "grad_norm": 124.60421752929688,
      "learning_rate": 1.2279808254594505e-05,
      "loss": 5.024,
      "step": 10998
    },
    {
      "epoch": 0.44404521598708113,
      "grad_norm": 154.95130920410156,
      "learning_rate": 1.2278535089031377e-05,
      "loss": 5.9355,
      "step": 10999
    },
    {
      "epoch": 0.4440855874041179,
      "grad_norm": 98.9628677368164,
      "learning_rate": 1.227726188451062e-05,
      "loss": 4.1703,
      "step": 11000
    },
    {
      "epoch": 0.44412595882115463,
      "grad_norm": 98.57310485839844,
      "learning_rate": 1.2275988641054004e-05,
      "loss": 6.5636,
      "step": 11001
    },
    {
      "epoch": 0.44416633023819135,
      "grad_norm": 133.867919921875,
      "learning_rate": 1.2274715358683293e-05,
      "loss": 3.9502,
      "step": 11002
    },
    {
      "epoch": 0.4442067016552281,
      "grad_norm": 180.4846649169922,
      "learning_rate": 1.2273442037420264e-05,
      "loss": 5.7537,
      "step": 11003
    },
    {
      "epoch": 0.44424707307226485,
      "grad_norm": 161.49742126464844,
      "learning_rate": 1.2272168677286682e-05,
      "loss": 6.2551,
      "step": 11004
    },
    {
      "epoch": 0.4442874444893016,
      "grad_norm": 99.89896392822266,
      "learning_rate": 1.2270895278304326e-05,
      "loss": 4.4772,
      "step": 11005
    },
    {
      "epoch": 0.4443278159063383,
      "grad_norm": 115.71417236328125,
      "learning_rate": 1.226962184049496e-05,
      "loss": 5.8496,
      "step": 11006
    },
    {
      "epoch": 0.4443681873233751,
      "grad_norm": 143.95526123046875,
      "learning_rate": 1.2268348363880364e-05,
      "loss": 7.8142,
      "step": 11007
    },
    {
      "epoch": 0.4444085587404118,
      "grad_norm": 86.5326156616211,
      "learning_rate": 1.2267074848482306e-05,
      "loss": 6.0971,
      "step": 11008
    },
    {
      "epoch": 0.4444489301574485,
      "grad_norm": 83.37399291992188,
      "learning_rate": 1.2265801294322563e-05,
      "loss": 4.7293,
      "step": 11009
    },
    {
      "epoch": 0.44448930157448524,
      "grad_norm": 111.82554626464844,
      "learning_rate": 1.2264527701422908e-05,
      "loss": 4.9539,
      "step": 11010
    },
    {
      "epoch": 0.444529672991522,
      "grad_norm": 108.48834991455078,
      "learning_rate": 1.2263254069805118e-05,
      "loss": 7.0201,
      "step": 11011
    },
    {
      "epoch": 0.44457004440855874,
      "grad_norm": 84.54539489746094,
      "learning_rate": 1.2261980399490968e-05,
      "loss": 4.0629,
      "step": 11012
    },
    {
      "epoch": 0.44461041582559546,
      "grad_norm": 91.50992584228516,
      "learning_rate": 1.2260706690502233e-05,
      "loss": 3.7746,
      "step": 11013
    },
    {
      "epoch": 0.44465078724263224,
      "grad_norm": 107.22315216064453,
      "learning_rate": 1.2259432942860694e-05,
      "loss": 4.7908,
      "step": 11014
    },
    {
      "epoch": 0.44469115865966896,
      "grad_norm": 136.278564453125,
      "learning_rate": 1.2258159156588132e-05,
      "loss": 5.0284,
      "step": 11015
    },
    {
      "epoch": 0.4447315300767057,
      "grad_norm": 108.33246612548828,
      "learning_rate": 1.2256885331706317e-05,
      "loss": 4.1746,
      "step": 11016
    },
    {
      "epoch": 0.4447719014937424,
      "grad_norm": 92.66666412353516,
      "learning_rate": 1.2255611468237033e-05,
      "loss": 4.5872,
      "step": 11017
    },
    {
      "epoch": 0.4448122729107792,
      "grad_norm": 77.36709594726562,
      "learning_rate": 1.2254337566202064e-05,
      "loss": 4.6022,
      "step": 11018
    },
    {
      "epoch": 0.4448526443278159,
      "grad_norm": 119.21136474609375,
      "learning_rate": 1.2253063625623185e-05,
      "loss": 4.5899,
      "step": 11019
    },
    {
      "epoch": 0.4448930157448526,
      "grad_norm": 136.97238159179688,
      "learning_rate": 1.2251789646522176e-05,
      "loss": 6.239,
      "step": 11020
    },
    {
      "epoch": 0.4449333871618894,
      "grad_norm": 133.51715087890625,
      "learning_rate": 1.2250515628920827e-05,
      "loss": 4.3734,
      "step": 11021
    },
    {
      "epoch": 0.4449737585789261,
      "grad_norm": 52.990882873535156,
      "learning_rate": 1.2249241572840913e-05,
      "loss": 4.4428,
      "step": 11022
    },
    {
      "epoch": 0.44501412999596285,
      "grad_norm": 88.5163345336914,
      "learning_rate": 1.2247967478304221e-05,
      "loss": 5.41,
      "step": 11023
    },
    {
      "epoch": 0.44505450141299957,
      "grad_norm": 109.45835876464844,
      "learning_rate": 1.2246693345332534e-05,
      "loss": 5.1697,
      "step": 11024
    },
    {
      "epoch": 0.44509487283003635,
      "grad_norm": 75.62464141845703,
      "learning_rate": 1.2245419173947641e-05,
      "loss": 5.5299,
      "step": 11025
    },
    {
      "epoch": 0.44513524424707307,
      "grad_norm": 100.57221984863281,
      "learning_rate": 1.2244144964171318e-05,
      "loss": 3.4201,
      "step": 11026
    },
    {
      "epoch": 0.4451756156641098,
      "grad_norm": 82.53701782226562,
      "learning_rate": 1.2242870716025362e-05,
      "loss": 5.2771,
      "step": 11027
    },
    {
      "epoch": 0.44521598708114657,
      "grad_norm": 63.857295989990234,
      "learning_rate": 1.224159642953155e-05,
      "loss": 4.0537,
      "step": 11028
    },
    {
      "epoch": 0.4452563584981833,
      "grad_norm": 81.65908813476562,
      "learning_rate": 1.2240322104711677e-05,
      "loss": 4.0397,
      "step": 11029
    },
    {
      "epoch": 0.44529672991522,
      "grad_norm": 118.4681625366211,
      "learning_rate": 1.2239047741587524e-05,
      "loss": 6.2449,
      "step": 11030
    },
    {
      "epoch": 0.44533710133225674,
      "grad_norm": 99.49063873291016,
      "learning_rate": 1.2237773340180888e-05,
      "loss": 4.8816,
      "step": 11031
    },
    {
      "epoch": 0.4453774727492935,
      "grad_norm": 93.88455200195312,
      "learning_rate": 1.2236498900513549e-05,
      "loss": 6.3927,
      "step": 11032
    },
    {
      "epoch": 0.44541784416633023,
      "grad_norm": 127.31471252441406,
      "learning_rate": 1.22352244226073e-05,
      "loss": 3.8933,
      "step": 11033
    },
    {
      "epoch": 0.44545821558336696,
      "grad_norm": 102.6448974609375,
      "learning_rate": 1.2233949906483937e-05,
      "loss": 5.1606,
      "step": 11034
    },
    {
      "epoch": 0.44549858700040373,
      "grad_norm": 95.07569885253906,
      "learning_rate": 1.2232675352165246e-05,
      "loss": 5.0836,
      "step": 11035
    },
    {
      "epoch": 0.44553895841744046,
      "grad_norm": 270.8314208984375,
      "learning_rate": 1.223140075967302e-05,
      "loss": 6.9093,
      "step": 11036
    },
    {
      "epoch": 0.4455793298344772,
      "grad_norm": 134.0818328857422,
      "learning_rate": 1.2230126129029053e-05,
      "loss": 5.1473,
      "step": 11037
    },
    {
      "epoch": 0.44561970125151396,
      "grad_norm": 81.7453384399414,
      "learning_rate": 1.2228851460255135e-05,
      "loss": 4.1405,
      "step": 11038
    },
    {
      "epoch": 0.4456600726685507,
      "grad_norm": 70.18266296386719,
      "learning_rate": 1.2227576753373062e-05,
      "loss": 5.3988,
      "step": 11039
    },
    {
      "epoch": 0.4457004440855874,
      "grad_norm": 160.03363037109375,
      "learning_rate": 1.2226302008404629e-05,
      "loss": 5.3394,
      "step": 11040
    },
    {
      "epoch": 0.4457408155026241,
      "grad_norm": 121.1376724243164,
      "learning_rate": 1.2225027225371633e-05,
      "loss": 6.0821,
      "step": 11041
    },
    {
      "epoch": 0.4457811869196609,
      "grad_norm": 105.89107513427734,
      "learning_rate": 1.2223752404295863e-05,
      "loss": 5.8237,
      "step": 11042
    },
    {
      "epoch": 0.4458215583366976,
      "grad_norm": 91.0435562133789,
      "learning_rate": 1.222247754519912e-05,
      "loss": 4.9694,
      "step": 11043
    },
    {
      "epoch": 0.44586192975373434,
      "grad_norm": 105.07025909423828,
      "learning_rate": 1.22212026481032e-05,
      "loss": 5.1793,
      "step": 11044
    },
    {
      "epoch": 0.4459023011707711,
      "grad_norm": 87.79671478271484,
      "learning_rate": 1.2219927713029906e-05,
      "loss": 6.9835,
      "step": 11045
    },
    {
      "epoch": 0.44594267258780784,
      "grad_norm": 89.07661437988281,
      "learning_rate": 1.221865274000103e-05,
      "loss": 4.7599,
      "step": 11046
    },
    {
      "epoch": 0.44598304400484456,
      "grad_norm": 101.8155517578125,
      "learning_rate": 1.2217377729038374e-05,
      "loss": 4.7951,
      "step": 11047
    },
    {
      "epoch": 0.4460234154218813,
      "grad_norm": 87.04258728027344,
      "learning_rate": 1.2216102680163735e-05,
      "loss": 3.7902,
      "step": 11048
    },
    {
      "epoch": 0.44606378683891806,
      "grad_norm": 5288.7138671875,
      "learning_rate": 1.2214827593398917e-05,
      "loss": 4.8069,
      "step": 11049
    },
    {
      "epoch": 0.4461041582559548,
      "grad_norm": 100.82246398925781,
      "learning_rate": 1.2213552468765717e-05,
      "loss": 6.207,
      "step": 11050
    },
    {
      "epoch": 0.4461445296729915,
      "grad_norm": 188.3072967529297,
      "learning_rate": 1.2212277306285938e-05,
      "loss": 5.5518,
      "step": 11051
    },
    {
      "epoch": 0.4461849010900283,
      "grad_norm": 81.60325622558594,
      "learning_rate": 1.2211002105981385e-05,
      "loss": 6.3707,
      "step": 11052
    },
    {
      "epoch": 0.446225272507065,
      "grad_norm": 336.464111328125,
      "learning_rate": 1.220972686787386e-05,
      "loss": 6.5718,
      "step": 11053
    },
    {
      "epoch": 0.44626564392410173,
      "grad_norm": 84.71871185302734,
      "learning_rate": 1.2208451591985161e-05,
      "loss": 6.9649,
      "step": 11054
    },
    {
      "epoch": 0.44630601534113845,
      "grad_norm": 73.4386978149414,
      "learning_rate": 1.2207176278337103e-05,
      "loss": 6.0553,
      "step": 11055
    },
    {
      "epoch": 0.44634638675817523,
      "grad_norm": 178.9231414794922,
      "learning_rate": 1.2205900926951481e-05,
      "loss": 3.3575,
      "step": 11056
    },
    {
      "epoch": 0.44638675817521195,
      "grad_norm": 201.97979736328125,
      "learning_rate": 1.2204625537850107e-05,
      "loss": 5.5984,
      "step": 11057
    },
    {
      "epoch": 0.4464271295922487,
      "grad_norm": 93.05296325683594,
      "learning_rate": 1.2203350111054782e-05,
      "loss": 6.0048,
      "step": 11058
    },
    {
      "epoch": 0.44646750100928545,
      "grad_norm": 93.84735870361328,
      "learning_rate": 1.2202074646587318e-05,
      "loss": 4.9223,
      "step": 11059
    },
    {
      "epoch": 0.44650787242632217,
      "grad_norm": 113.50018310546875,
      "learning_rate": 1.2200799144469517e-05,
      "loss": 4.5217,
      "step": 11060
    },
    {
      "epoch": 0.4465482438433589,
      "grad_norm": 86.10489654541016,
      "learning_rate": 1.219952360472319e-05,
      "loss": 4.9904,
      "step": 11061
    },
    {
      "epoch": 0.4465886152603956,
      "grad_norm": 100.00135040283203,
      "learning_rate": 1.2198248027370146e-05,
      "loss": 3.5508,
      "step": 11062
    },
    {
      "epoch": 0.4466289866774324,
      "grad_norm": 104.71772003173828,
      "learning_rate": 1.2196972412432196e-05,
      "loss": 4.8498,
      "step": 11063
    },
    {
      "epoch": 0.4466693580944691,
      "grad_norm": 80.44993591308594,
      "learning_rate": 1.2195696759931143e-05,
      "loss": 5.6367,
      "step": 11064
    },
    {
      "epoch": 0.44670972951150584,
      "grad_norm": 80.45621490478516,
      "learning_rate": 1.219442106988881e-05,
      "loss": 5.6145,
      "step": 11065
    },
    {
      "epoch": 0.4467501009285426,
      "grad_norm": 110.2573013305664,
      "learning_rate": 1.2193145342326995e-05,
      "loss": 4.4598,
      "step": 11066
    },
    {
      "epoch": 0.44679047234557934,
      "grad_norm": 82.3207015991211,
      "learning_rate": 1.2191869577267519e-05,
      "loss": 5.7533,
      "step": 11067
    },
    {
      "epoch": 0.44683084376261606,
      "grad_norm": 321.45745849609375,
      "learning_rate": 1.219059377473219e-05,
      "loss": 4.8425,
      "step": 11068
    },
    {
      "epoch": 0.4468712151796528,
      "grad_norm": 119.92153930664062,
      "learning_rate": 1.2189317934742826e-05,
      "loss": 5.9028,
      "step": 11069
    },
    {
      "epoch": 0.44691158659668956,
      "grad_norm": 73.22772216796875,
      "learning_rate": 1.2188042057321234e-05,
      "loss": 5.4759,
      "step": 11070
    },
    {
      "epoch": 0.4469519580137263,
      "grad_norm": 83.47422790527344,
      "learning_rate": 1.2186766142489233e-05,
      "loss": 5.1214,
      "step": 11071
    },
    {
      "epoch": 0.446992329430763,
      "grad_norm": 120.33990478515625,
      "learning_rate": 1.2185490190268636e-05,
      "loss": 4.9075,
      "step": 11072
    },
    {
      "epoch": 0.4470327008477998,
      "grad_norm": 109.89641571044922,
      "learning_rate": 1.2184214200681266e-05,
      "loss": 5.5761,
      "step": 11073
    },
    {
      "epoch": 0.4470730722648365,
      "grad_norm": 80.46892547607422,
      "learning_rate": 1.2182938173748928e-05,
      "loss": 5.3332,
      "step": 11074
    },
    {
      "epoch": 0.4471134436818732,
      "grad_norm": 79.39349365234375,
      "learning_rate": 1.2181662109493449e-05,
      "loss": 6.6836,
      "step": 11075
    },
    {
      "epoch": 0.44715381509890995,
      "grad_norm": 113.51432800292969,
      "learning_rate": 1.2180386007936637e-05,
      "loss": 3.7845,
      "step": 11076
    },
    {
      "epoch": 0.4471941865159467,
      "grad_norm": 86.13336944580078,
      "learning_rate": 1.2179109869100324e-05,
      "loss": 5.8134,
      "step": 11077
    },
    {
      "epoch": 0.44723455793298345,
      "grad_norm": 119.73130798339844,
      "learning_rate": 1.2177833693006314e-05,
      "loss": 5.582,
      "step": 11078
    },
    {
      "epoch": 0.44727492935002017,
      "grad_norm": 56.653717041015625,
      "learning_rate": 1.2176557479676435e-05,
      "loss": 5.13,
      "step": 11079
    },
    {
      "epoch": 0.44731530076705694,
      "grad_norm": 94.27262115478516,
      "learning_rate": 1.2175281229132505e-05,
      "loss": 3.7392,
      "step": 11080
    },
    {
      "epoch": 0.44735567218409367,
      "grad_norm": 86.80482482910156,
      "learning_rate": 1.2174004941396346e-05,
      "loss": 5.8008,
      "step": 11081
    },
    {
      "epoch": 0.4473960436011304,
      "grad_norm": 64.1164779663086,
      "learning_rate": 1.2172728616489779e-05,
      "loss": 5.3606,
      "step": 11082
    },
    {
      "epoch": 0.4474364150181671,
      "grad_norm": 88.25923919677734,
      "learning_rate": 1.217145225443463e-05,
      "loss": 6.2155,
      "step": 11083
    },
    {
      "epoch": 0.4474767864352039,
      "grad_norm": 79.09874725341797,
      "learning_rate": 1.2170175855252715e-05,
      "loss": 5.7958,
      "step": 11084
    },
    {
      "epoch": 0.4475171578522406,
      "grad_norm": 105.6319351196289,
      "learning_rate": 1.216889941896586e-05,
      "loss": 5.4995,
      "step": 11085
    },
    {
      "epoch": 0.44755752926927733,
      "grad_norm": 88.38871002197266,
      "learning_rate": 1.2167622945595892e-05,
      "loss": 5.2855,
      "step": 11086
    },
    {
      "epoch": 0.4475979006863141,
      "grad_norm": 117.91046905517578,
      "learning_rate": 1.2166346435164633e-05,
      "loss": 5.6847,
      "step": 11087
    },
    {
      "epoch": 0.44763827210335083,
      "grad_norm": 70.54335021972656,
      "learning_rate": 1.2165069887693908e-05,
      "loss": 6.6823,
      "step": 11088
    },
    {
      "epoch": 0.44767864352038755,
      "grad_norm": 84.5487289428711,
      "learning_rate": 1.2163793303205545e-05,
      "loss": 5.2056,
      "step": 11089
    },
    {
      "epoch": 0.44771901493742433,
      "grad_norm": 115.82476806640625,
      "learning_rate": 1.2162516681721365e-05,
      "loss": 4.9925,
      "step": 11090
    },
    {
      "epoch": 0.44775938635446105,
      "grad_norm": 100.26533508300781,
      "learning_rate": 1.2161240023263202e-05,
      "loss": 3.5509,
      "step": 11091
    },
    {
      "epoch": 0.4477997577714978,
      "grad_norm": 78.36908721923828,
      "learning_rate": 1.2159963327852882e-05,
      "loss": 4.4981,
      "step": 11092
    },
    {
      "epoch": 0.4478401291885345,
      "grad_norm": 78.31446075439453,
      "learning_rate": 1.2158686595512232e-05,
      "loss": 6.3632,
      "step": 11093
    },
    {
      "epoch": 0.4478805006055713,
      "grad_norm": 64.38711547851562,
      "learning_rate": 1.2157409826263082e-05,
      "loss": 3.9854,
      "step": 11094
    },
    {
      "epoch": 0.447920872022608,
      "grad_norm": 114.0913314819336,
      "learning_rate": 1.2156133020127265e-05,
      "loss": 4.6812,
      "step": 11095
    },
    {
      "epoch": 0.4479612434396447,
      "grad_norm": 130.79544067382812,
      "learning_rate": 1.2154856177126606e-05,
      "loss": 5.1978,
      "step": 11096
    },
    {
      "epoch": 0.4480016148566815,
      "grad_norm": 71.97161865234375,
      "learning_rate": 1.2153579297282941e-05,
      "loss": 6.6572,
      "step": 11097
    },
    {
      "epoch": 0.4480419862737182,
      "grad_norm": 107.75257110595703,
      "learning_rate": 1.2152302380618095e-05,
      "loss": 3.9275,
      "step": 11098
    },
    {
      "epoch": 0.44808235769075494,
      "grad_norm": 71.02590942382812,
      "learning_rate": 1.2151025427153907e-05,
      "loss": 5.2856,
      "step": 11099
    },
    {
      "epoch": 0.44812272910779166,
      "grad_norm": 107.63337707519531,
      "learning_rate": 1.2149748436912206e-05,
      "loss": 5.2455,
      "step": 11100
    },
    {
      "epoch": 0.44816310052482844,
      "grad_norm": 75.8456039428711,
      "learning_rate": 1.2148471409914823e-05,
      "loss": 4.3158,
      "step": 11101
    },
    {
      "epoch": 0.44820347194186516,
      "grad_norm": 95.03414154052734,
      "learning_rate": 1.21471943461836e-05,
      "loss": 6.0923,
      "step": 11102
    },
    {
      "epoch": 0.4482438433589019,
      "grad_norm": 161.60911560058594,
      "learning_rate": 1.2145917245740368e-05,
      "loss": 5.1447,
      "step": 11103
    },
    {
      "epoch": 0.44828421477593866,
      "grad_norm": 100.53604125976562,
      "learning_rate": 1.2144640108606963e-05,
      "loss": 4.9364,
      "step": 11104
    },
    {
      "epoch": 0.4483245861929754,
      "grad_norm": 75.90503692626953,
      "learning_rate": 1.214336293480522e-05,
      "loss": 3.8814,
      "step": 11105
    },
    {
      "epoch": 0.4483649576100121,
      "grad_norm": 149.5391845703125,
      "learning_rate": 1.2142085724356973e-05,
      "loss": 4.3868,
      "step": 11106
    },
    {
      "epoch": 0.4484053290270488,
      "grad_norm": 122.37489318847656,
      "learning_rate": 1.2140808477284064e-05,
      "loss": 4.7255,
      "step": 11107
    },
    {
      "epoch": 0.4484457004440856,
      "grad_norm": 232.4123992919922,
      "learning_rate": 1.213953119360833e-05,
      "loss": 6.4363,
      "step": 11108
    },
    {
      "epoch": 0.4484860718611223,
      "grad_norm": 453.0975646972656,
      "learning_rate": 1.213825387335161e-05,
      "loss": 5.0937,
      "step": 11109
    },
    {
      "epoch": 0.44852644327815905,
      "grad_norm": 87.30015563964844,
      "learning_rate": 1.2136976516535736e-05,
      "loss": 4.4512,
      "step": 11110
    },
    {
      "epoch": 0.4485668146951958,
      "grad_norm": 87.61174774169922,
      "learning_rate": 1.2135699123182558e-05,
      "loss": 5.3097,
      "step": 11111
    },
    {
      "epoch": 0.44860718611223255,
      "grad_norm": 152.33534240722656,
      "learning_rate": 1.2134421693313912e-05,
      "loss": 5.1878,
      "step": 11112
    },
    {
      "epoch": 0.44864755752926927,
      "grad_norm": 95.5900650024414,
      "learning_rate": 1.2133144226951639e-05,
      "loss": 5.1611,
      "step": 11113
    },
    {
      "epoch": 0.448687928946306,
      "grad_norm": 133.60850524902344,
      "learning_rate": 1.213186672411758e-05,
      "loss": 6.0682,
      "step": 11114
    },
    {
      "epoch": 0.44872830036334277,
      "grad_norm": 101.78844451904297,
      "learning_rate": 1.213058918483358e-05,
      "loss": 4.8925,
      "step": 11115
    },
    {
      "epoch": 0.4487686717803795,
      "grad_norm": 130.14291381835938,
      "learning_rate": 1.212931160912148e-05,
      "loss": 4.8741,
      "step": 11116
    },
    {
      "epoch": 0.4488090431974162,
      "grad_norm": 75.6653060913086,
      "learning_rate": 1.2128033997003126e-05,
      "loss": 6.3765,
      "step": 11117
    },
    {
      "epoch": 0.448849414614453,
      "grad_norm": 82.41069030761719,
      "learning_rate": 1.2126756348500354e-05,
      "loss": 4.357,
      "step": 11118
    },
    {
      "epoch": 0.4488897860314897,
      "grad_norm": 105.53828430175781,
      "learning_rate": 1.212547866363502e-05,
      "loss": 3.9258,
      "step": 11119
    },
    {
      "epoch": 0.44893015744852643,
      "grad_norm": 93.76024627685547,
      "learning_rate": 1.212420094242896e-05,
      "loss": 3.5045,
      "step": 11120
    },
    {
      "epoch": 0.44897052886556316,
      "grad_norm": 141.13778686523438,
      "learning_rate": 1.2122923184904028e-05,
      "loss": 5.9017,
      "step": 11121
    },
    {
      "epoch": 0.44901090028259993,
      "grad_norm": 138.13735961914062,
      "learning_rate": 1.2121645391082065e-05,
      "loss": 5.4083,
      "step": 11122
    },
    {
      "epoch": 0.44905127169963666,
      "grad_norm": 120.43302154541016,
      "learning_rate": 1.2120367560984923e-05,
      "loss": 6.5779,
      "step": 11123
    },
    {
      "epoch": 0.4490916431166734,
      "grad_norm": 117.32276153564453,
      "learning_rate": 1.2119089694634443e-05,
      "loss": 5.413,
      "step": 11124
    },
    {
      "epoch": 0.44913201453371016,
      "grad_norm": 61.891273498535156,
      "learning_rate": 1.211781179205248e-05,
      "loss": 2.4803,
      "step": 11125
    },
    {
      "epoch": 0.4491723859507469,
      "grad_norm": 132.56524658203125,
      "learning_rate": 1.211653385326088e-05,
      "loss": 4.5166,
      "step": 11126
    },
    {
      "epoch": 0.4492127573677836,
      "grad_norm": 110.49624633789062,
      "learning_rate": 1.2115255878281496e-05,
      "loss": 5.2527,
      "step": 11127
    },
    {
      "epoch": 0.4492531287848203,
      "grad_norm": 79.08259582519531,
      "learning_rate": 1.2113977867136171e-05,
      "loss": 4.3072,
      "step": 11128
    },
    {
      "epoch": 0.4492935002018571,
      "grad_norm": 181.60772705078125,
      "learning_rate": 1.2112699819846765e-05,
      "loss": 5.1191,
      "step": 11129
    },
    {
      "epoch": 0.4493338716188938,
      "grad_norm": 103.30465698242188,
      "learning_rate": 1.2111421736435125e-05,
      "loss": 4.0322,
      "step": 11130
    },
    {
      "epoch": 0.44937424303593054,
      "grad_norm": 165.3274688720703,
      "learning_rate": 1.2110143616923102e-05,
      "loss": 5.671,
      "step": 11131
    },
    {
      "epoch": 0.4494146144529673,
      "grad_norm": 70.65423583984375,
      "learning_rate": 1.210886546133255e-05,
      "loss": 4.5054,
      "step": 11132
    },
    {
      "epoch": 0.44945498587000404,
      "grad_norm": 127.54769134521484,
      "learning_rate": 1.2107587269685328e-05,
      "loss": 5.7562,
      "step": 11133
    },
    {
      "epoch": 0.44949535728704076,
      "grad_norm": 108.71188354492188,
      "learning_rate": 1.210630904200328e-05,
      "loss": 3.8305,
      "step": 11134
    },
    {
      "epoch": 0.4495357287040775,
      "grad_norm": 102.36903381347656,
      "learning_rate": 1.2105030778308269e-05,
      "loss": 6.2737,
      "step": 11135
    },
    {
      "epoch": 0.44957610012111426,
      "grad_norm": 139.9503173828125,
      "learning_rate": 1.2103752478622146e-05,
      "loss": 4.8656,
      "step": 11136
    },
    {
      "epoch": 0.449616471538151,
      "grad_norm": 79.86703491210938,
      "learning_rate": 1.2102474142966771e-05,
      "loss": 6.4927,
      "step": 11137
    },
    {
      "epoch": 0.4496568429551877,
      "grad_norm": 86.60759735107422,
      "learning_rate": 1.2101195771363994e-05,
      "loss": 6.8327,
      "step": 11138
    },
    {
      "epoch": 0.4496972143722245,
      "grad_norm": 100.7059555053711,
      "learning_rate": 1.209991736383568e-05,
      "loss": 6.4351,
      "step": 11139
    },
    {
      "epoch": 0.4497375857892612,
      "grad_norm": 79.0891342163086,
      "learning_rate": 1.2098638920403678e-05,
      "loss": 5.7163,
      "step": 11140
    },
    {
      "epoch": 0.44977795720629793,
      "grad_norm": 106.65360260009766,
      "learning_rate": 1.2097360441089854e-05,
      "loss": 5.7625,
      "step": 11141
    },
    {
      "epoch": 0.4498183286233347,
      "grad_norm": 74.16148376464844,
      "learning_rate": 1.209608192591606e-05,
      "loss": 5.912,
      "step": 11142
    },
    {
      "epoch": 0.44985870004037143,
      "grad_norm": 131.4083709716797,
      "learning_rate": 1.2094803374904167e-05,
      "loss": 5.2694,
      "step": 11143
    },
    {
      "epoch": 0.44989907145740815,
      "grad_norm": 83.58517456054688,
      "learning_rate": 1.2093524788076023e-05,
      "loss": 2.8678,
      "step": 11144
    },
    {
      "epoch": 0.4499394428744449,
      "grad_norm": 118.39080810546875,
      "learning_rate": 1.2092246165453497e-05,
      "loss": 6.2989,
      "step": 11145
    },
    {
      "epoch": 0.44997981429148165,
      "grad_norm": 95.07862854003906,
      "learning_rate": 1.2090967507058443e-05,
      "loss": 6.4668,
      "step": 11146
    },
    {
      "epoch": 0.45002018570851837,
      "grad_norm": 112.77436065673828,
      "learning_rate": 1.2089688812912731e-05,
      "loss": 4.898,
      "step": 11147
    },
    {
      "epoch": 0.4500605571255551,
      "grad_norm": 85.47467041015625,
      "learning_rate": 1.208841008303822e-05,
      "loss": 4.4619,
      "step": 11148
    },
    {
      "epoch": 0.45010092854259187,
      "grad_norm": 92.49681854248047,
      "learning_rate": 1.2087131317456773e-05,
      "loss": 4.6728,
      "step": 11149
    },
    {
      "epoch": 0.4501412999596286,
      "grad_norm": 71.7533187866211,
      "learning_rate": 1.2085852516190253e-05,
      "loss": 4.7538,
      "step": 11150
    },
    {
      "epoch": 0.4501816713766653,
      "grad_norm": 93.91734313964844,
      "learning_rate": 1.2084573679260526e-05,
      "loss": 5.8596,
      "step": 11151
    },
    {
      "epoch": 0.45022204279370204,
      "grad_norm": 171.0267791748047,
      "learning_rate": 1.2083294806689456e-05,
      "loss": 5.9522,
      "step": 11152
    },
    {
      "epoch": 0.4502624142107388,
      "grad_norm": 126.06265258789062,
      "learning_rate": 1.2082015898498911e-05,
      "loss": 4.4708,
      "step": 11153
    },
    {
      "epoch": 0.45030278562777554,
      "grad_norm": 91.66527557373047,
      "learning_rate": 1.2080736954710757e-05,
      "loss": 3.7049,
      "step": 11154
    },
    {
      "epoch": 0.45034315704481226,
      "grad_norm": 72.76667785644531,
      "learning_rate": 1.207945797534686e-05,
      "loss": 5.2546,
      "step": 11155
    },
    {
      "epoch": 0.45038352846184904,
      "grad_norm": 115.84510040283203,
      "learning_rate": 1.2078178960429085e-05,
      "loss": 5.8418,
      "step": 11156
    },
    {
      "epoch": 0.45042389987888576,
      "grad_norm": 112.86590576171875,
      "learning_rate": 1.2076899909979305e-05,
      "loss": 4.0142,
      "step": 11157
    },
    {
      "epoch": 0.4504642712959225,
      "grad_norm": 86.12017822265625,
      "learning_rate": 1.2075620824019385e-05,
      "loss": 6.7212,
      "step": 11158
    },
    {
      "epoch": 0.4505046427129592,
      "grad_norm": 112.0505599975586,
      "learning_rate": 1.2074341702571195e-05,
      "loss": 6.3863,
      "step": 11159
    },
    {
      "epoch": 0.450545014129996,
      "grad_norm": 94.12602233886719,
      "learning_rate": 1.2073062545656606e-05,
      "loss": 5.1853,
      "step": 11160
    },
    {
      "epoch": 0.4505853855470327,
      "grad_norm": 85.05081176757812,
      "learning_rate": 1.2071783353297492e-05,
      "loss": 5.1048,
      "step": 11161
    },
    {
      "epoch": 0.4506257569640694,
      "grad_norm": 171.67137145996094,
      "learning_rate": 1.2070504125515716e-05,
      "loss": 6.1254,
      "step": 11162
    },
    {
      "epoch": 0.4506661283811062,
      "grad_norm": 94.38302612304688,
      "learning_rate": 1.206922486233316e-05,
      "loss": 4.6772,
      "step": 11163
    },
    {
      "epoch": 0.4507064997981429,
      "grad_norm": 119.57659912109375,
      "learning_rate": 1.2067945563771685e-05,
      "loss": 4.4933,
      "step": 11164
    },
    {
      "epoch": 0.45074687121517965,
      "grad_norm": 104.14472198486328,
      "learning_rate": 1.2066666229853174e-05,
      "loss": 4.0487,
      "step": 11165
    },
    {
      "epoch": 0.45078724263221637,
      "grad_norm": 80.60417175292969,
      "learning_rate": 1.2065386860599494e-05,
      "loss": 4.5092,
      "step": 11166
    },
    {
      "epoch": 0.45082761404925314,
      "grad_norm": 86.40535736083984,
      "learning_rate": 1.2064107456032523e-05,
      "loss": 5.1574,
      "step": 11167
    },
    {
      "epoch": 0.45086798546628987,
      "grad_norm": 76.17354583740234,
      "learning_rate": 1.2062828016174132e-05,
      "loss": 5.823,
      "step": 11168
    },
    {
      "epoch": 0.4509083568833266,
      "grad_norm": 149.655517578125,
      "learning_rate": 1.2061548541046203e-05,
      "loss": 5.4289,
      "step": 11169
    },
    {
      "epoch": 0.45094872830036337,
      "grad_norm": 78.58422088623047,
      "learning_rate": 1.2060269030670605e-05,
      "loss": 5.3228,
      "step": 11170
    },
    {
      "epoch": 0.4509890997174001,
      "grad_norm": 120.65811157226562,
      "learning_rate": 1.2058989485069219e-05,
      "loss": 6.4066,
      "step": 11171
    },
    {
      "epoch": 0.4510294711344368,
      "grad_norm": 79.59131622314453,
      "learning_rate": 1.2057709904263919e-05,
      "loss": 3.737,
      "step": 11172
    },
    {
      "epoch": 0.45106984255147353,
      "grad_norm": 98.87419128417969,
      "learning_rate": 1.2056430288276589e-05,
      "loss": 4.2895,
      "step": 11173
    },
    {
      "epoch": 0.4511102139685103,
      "grad_norm": 78.037841796875,
      "learning_rate": 1.2055150637129099e-05,
      "loss": 5.7437,
      "step": 11174
    },
    {
      "epoch": 0.45115058538554703,
      "grad_norm": 75.02729034423828,
      "learning_rate": 1.2053870950843336e-05,
      "loss": 4.5174,
      "step": 11175
    },
    {
      "epoch": 0.45119095680258375,
      "grad_norm": 101.93506622314453,
      "learning_rate": 1.2052591229441174e-05,
      "loss": 5.0798,
      "step": 11176
    },
    {
      "epoch": 0.45123132821962053,
      "grad_norm": 99.40262603759766,
      "learning_rate": 1.2051311472944493e-05,
      "loss": 4.4117,
      "step": 11177
    },
    {
      "epoch": 0.45127169963665725,
      "grad_norm": 90.77872467041016,
      "learning_rate": 1.2050031681375178e-05,
      "loss": 5.4857,
      "step": 11178
    },
    {
      "epoch": 0.451312071053694,
      "grad_norm": 150.1197509765625,
      "learning_rate": 1.2048751854755108e-05,
      "loss": 4.1304,
      "step": 11179
    },
    {
      "epoch": 0.4513524424707307,
      "grad_norm": 135.42945861816406,
      "learning_rate": 1.2047471993106163e-05,
      "loss": 5.0328,
      "step": 11180
    },
    {
      "epoch": 0.4513928138877675,
      "grad_norm": 103.79365539550781,
      "learning_rate": 1.2046192096450234e-05,
      "loss": 6.1603,
      "step": 11181
    },
    {
      "epoch": 0.4514331853048042,
      "grad_norm": 660.5298461914062,
      "learning_rate": 1.2044912164809193e-05,
      "loss": 4.8008,
      "step": 11182
    },
    {
      "epoch": 0.4514735567218409,
      "grad_norm": 97.99553680419922,
      "learning_rate": 1.2043632198204932e-05,
      "loss": 5.9351,
      "step": 11183
    },
    {
      "epoch": 0.4515139281388777,
      "grad_norm": 75.84133911132812,
      "learning_rate": 1.2042352196659333e-05,
      "loss": 7.1072,
      "step": 11184
    },
    {
      "epoch": 0.4515542995559144,
      "grad_norm": 94.7176742553711,
      "learning_rate": 1.204107216019428e-05,
      "loss": 4.8888,
      "step": 11185
    },
    {
      "epoch": 0.45159467097295114,
      "grad_norm": 132.02293395996094,
      "learning_rate": 1.2039792088831658e-05,
      "loss": 4.4028,
      "step": 11186
    },
    {
      "epoch": 0.45163504238998786,
      "grad_norm": 94.90840911865234,
      "learning_rate": 1.2038511982593357e-05,
      "loss": 4.4404,
      "step": 11187
    },
    {
      "epoch": 0.45167541380702464,
      "grad_norm": 151.32798767089844,
      "learning_rate": 1.2037231841501258e-05,
      "loss": 5.6405,
      "step": 11188
    },
    {
      "epoch": 0.45171578522406136,
      "grad_norm": 125.98816680908203,
      "learning_rate": 1.2035951665577252e-05,
      "loss": 4.4655,
      "step": 11189
    },
    {
      "epoch": 0.4517561566410981,
      "grad_norm": 103.35586547851562,
      "learning_rate": 1.2034671454843227e-05,
      "loss": 4.2942,
      "step": 11190
    },
    {
      "epoch": 0.45179652805813486,
      "grad_norm": 76.5519027709961,
      "learning_rate": 1.2033391209321075e-05,
      "loss": 6.3802,
      "step": 11191
    },
    {
      "epoch": 0.4518368994751716,
      "grad_norm": 74.30990600585938,
      "learning_rate": 1.2032110929032679e-05,
      "loss": 4.8412,
      "step": 11192
    },
    {
      "epoch": 0.4518772708922083,
      "grad_norm": 168.846435546875,
      "learning_rate": 1.2030830613999933e-05,
      "loss": 4.1266,
      "step": 11193
    },
    {
      "epoch": 0.451917642309245,
      "grad_norm": 84.08611297607422,
      "learning_rate": 1.2029550264244726e-05,
      "loss": 6.312,
      "step": 11194
    },
    {
      "epoch": 0.4519580137262818,
      "grad_norm": 138.72877502441406,
      "learning_rate": 1.2028269879788948e-05,
      "loss": 5.9863,
      "step": 11195
    },
    {
      "epoch": 0.4519983851433185,
      "grad_norm": 82.98540496826172,
      "learning_rate": 1.2026989460654488e-05,
      "loss": 5.2807,
      "step": 11196
    },
    {
      "epoch": 0.45203875656035525,
      "grad_norm": 81.47334289550781,
      "learning_rate": 1.2025709006863246e-05,
      "loss": 5.4432,
      "step": 11197
    },
    {
      "epoch": 0.452079127977392,
      "grad_norm": 131.32156372070312,
      "learning_rate": 1.2024428518437108e-05,
      "loss": 5.1643,
      "step": 11198
    },
    {
      "epoch": 0.45211949939442875,
      "grad_norm": 107.41846466064453,
      "learning_rate": 1.202314799539797e-05,
      "loss": 5.4961,
      "step": 11199
    },
    {
      "epoch": 0.45215987081146547,
      "grad_norm": 80.5974349975586,
      "learning_rate": 1.2021867437767724e-05,
      "loss": 5.5145,
      "step": 11200
    },
    {
      "epoch": 0.45220024222850225,
      "grad_norm": 126.5748062133789,
      "learning_rate": 1.2020586845568271e-05,
      "loss": 5.5001,
      "step": 11201
    },
    {
      "epoch": 0.45224061364553897,
      "grad_norm": 96.9032211303711,
      "learning_rate": 1.2019306218821499e-05,
      "loss": 4.595,
      "step": 11202
    },
    {
      "epoch": 0.4522809850625757,
      "grad_norm": 263.12890625,
      "learning_rate": 1.2018025557549306e-05,
      "loss": 4.8612,
      "step": 11203
    },
    {
      "epoch": 0.4523213564796124,
      "grad_norm": 87.63664245605469,
      "learning_rate": 1.2016744861773589e-05,
      "loss": 5.883,
      "step": 11204
    },
    {
      "epoch": 0.4523617278966492,
      "grad_norm": 114.52021789550781,
      "learning_rate": 1.2015464131516243e-05,
      "loss": 6.1054,
      "step": 11205
    },
    {
      "epoch": 0.4524020993136859,
      "grad_norm": 86.92294311523438,
      "learning_rate": 1.2014183366799169e-05,
      "loss": 4.288,
      "step": 11206
    },
    {
      "epoch": 0.45244247073072263,
      "grad_norm": 76.22509765625,
      "learning_rate": 1.201290256764426e-05,
      "loss": 6.2732,
      "step": 11207
    },
    {
      "epoch": 0.4524828421477594,
      "grad_norm": 116.71906280517578,
      "learning_rate": 1.201162173407342e-05,
      "loss": 4.3896,
      "step": 11208
    },
    {
      "epoch": 0.45252321356479613,
      "grad_norm": 81.28243255615234,
      "learning_rate": 1.2010340866108543e-05,
      "loss": 5.3989,
      "step": 11209
    },
    {
      "epoch": 0.45256358498183286,
      "grad_norm": 88.78678894042969,
      "learning_rate": 1.2009059963771532e-05,
      "loss": 6.1319,
      "step": 11210
    },
    {
      "epoch": 0.4526039563988696,
      "grad_norm": 93.4740219116211,
      "learning_rate": 1.200777902708429e-05,
      "loss": 4.4886,
      "step": 11211
    },
    {
      "epoch": 0.45264432781590636,
      "grad_norm": 89.50419616699219,
      "learning_rate": 1.2006498056068711e-05,
      "loss": 4.5841,
      "step": 11212
    },
    {
      "epoch": 0.4526846992329431,
      "grad_norm": 74.5162124633789,
      "learning_rate": 1.2005217050746704e-05,
      "loss": 4.1149,
      "step": 11213
    },
    {
      "epoch": 0.4527250706499798,
      "grad_norm": 117.26213836669922,
      "learning_rate": 1.200393601114017e-05,
      "loss": 6.6478,
      "step": 11214
    },
    {
      "epoch": 0.4527654420670166,
      "grad_norm": 106.82968139648438,
      "learning_rate": 1.2002654937271006e-05,
      "loss": 5.9964,
      "step": 11215
    },
    {
      "epoch": 0.4528058134840533,
      "grad_norm": 106.55311584472656,
      "learning_rate": 1.200137382916112e-05,
      "loss": 5.9524,
      "step": 11216
    },
    {
      "epoch": 0.45284618490109,
      "grad_norm": 70.3532485961914,
      "learning_rate": 1.2000092686832417e-05,
      "loss": 5.9074,
      "step": 11217
    },
    {
      "epoch": 0.45288655631812674,
      "grad_norm": 85.09806060791016,
      "learning_rate": 1.1998811510306796e-05,
      "loss": 7.5397,
      "step": 11218
    },
    {
      "epoch": 0.4529269277351635,
      "grad_norm": 107.78484344482422,
      "learning_rate": 1.1997530299606165e-05,
      "loss": 5.153,
      "step": 11219
    },
    {
      "epoch": 0.45296729915220024,
      "grad_norm": 70.0580062866211,
      "learning_rate": 1.1996249054752433e-05,
      "loss": 4.8323,
      "step": 11220
    },
    {
      "epoch": 0.45300767056923696,
      "grad_norm": 145.51351928710938,
      "learning_rate": 1.1994967775767506e-05,
      "loss": 5.0817,
      "step": 11221
    },
    {
      "epoch": 0.45304804198627374,
      "grad_norm": 110.34487915039062,
      "learning_rate": 1.1993686462673284e-05,
      "loss": 6.2131,
      "step": 11222
    },
    {
      "epoch": 0.45308841340331046,
      "grad_norm": 107.74984741210938,
      "learning_rate": 1.1992405115491683e-05,
      "loss": 6.605,
      "step": 11223
    },
    {
      "epoch": 0.4531287848203472,
      "grad_norm": 159.8262481689453,
      "learning_rate": 1.1991123734244606e-05,
      "loss": 4.4591,
      "step": 11224
    },
    {
      "epoch": 0.4531691562373839,
      "grad_norm": 97.40176391601562,
      "learning_rate": 1.1989842318953964e-05,
      "loss": 5.5266,
      "step": 11225
    },
    {
      "epoch": 0.4532095276544207,
      "grad_norm": 123.8430404663086,
      "learning_rate": 1.198856086964166e-05,
      "loss": 6.6292,
      "step": 11226
    },
    {
      "epoch": 0.4532498990714574,
      "grad_norm": 127.68833923339844,
      "learning_rate": 1.1987279386329615e-05,
      "loss": 5.741,
      "step": 11227
    },
    {
      "epoch": 0.45329027048849413,
      "grad_norm": 97.44669342041016,
      "learning_rate": 1.1985997869039729e-05,
      "loss": 4.7367,
      "step": 11228
    },
    {
      "epoch": 0.4533306419055309,
      "grad_norm": 95.06990814208984,
      "learning_rate": 1.1984716317793917e-05,
      "loss": 5.1813,
      "step": 11229
    },
    {
      "epoch": 0.45337101332256763,
      "grad_norm": 76.457275390625,
      "learning_rate": 1.1983434732614093e-05,
      "loss": 4.7903,
      "step": 11230
    },
    {
      "epoch": 0.45341138473960435,
      "grad_norm": 80.10881042480469,
      "learning_rate": 1.1982153113522166e-05,
      "loss": 5.3852,
      "step": 11231
    },
    {
      "epoch": 0.4534517561566411,
      "grad_norm": 108.63733673095703,
      "learning_rate": 1.1980871460540046e-05,
      "loss": 5.2039,
      "step": 11232
    },
    {
      "epoch": 0.45349212757367785,
      "grad_norm": 87.97930145263672,
      "learning_rate": 1.1979589773689655e-05,
      "loss": 4.9563,
      "step": 11233
    },
    {
      "epoch": 0.45353249899071457,
      "grad_norm": 133.5136260986328,
      "learning_rate": 1.1978308052992897e-05,
      "loss": 5.1746,
      "step": 11234
    },
    {
      "epoch": 0.4535728704077513,
      "grad_norm": 90.44637298583984,
      "learning_rate": 1.1977026298471697e-05,
      "loss": 5.7947,
      "step": 11235
    },
    {
      "epoch": 0.45361324182478807,
      "grad_norm": 82.0143814086914,
      "learning_rate": 1.1975744510147957e-05,
      "loss": 5.3384,
      "step": 11236
    },
    {
      "epoch": 0.4536536132418248,
      "grad_norm": 101.06826782226562,
      "learning_rate": 1.1974462688043604e-05,
      "loss": 5.8238,
      "step": 11237
    },
    {
      "epoch": 0.4536939846588615,
      "grad_norm": 116.80293273925781,
      "learning_rate": 1.197318083218055e-05,
      "loss": 3.7294,
      "step": 11238
    },
    {
      "epoch": 0.45373435607589824,
      "grad_norm": 102.31144714355469,
      "learning_rate": 1.1971898942580709e-05,
      "loss": 6.9234,
      "step": 11239
    },
    {
      "epoch": 0.453774727492935,
      "grad_norm": 118.30909729003906,
      "learning_rate": 1.1970617019266e-05,
      "loss": 5.7837,
      "step": 11240
    },
    {
      "epoch": 0.45381509890997174,
      "grad_norm": 82.43429565429688,
      "learning_rate": 1.1969335062258344e-05,
      "loss": 5.904,
      "step": 11241
    },
    {
      "epoch": 0.45385547032700846,
      "grad_norm": 105.57051849365234,
      "learning_rate": 1.1968053071579657e-05,
      "loss": 6.3937,
      "step": 11242
    },
    {
      "epoch": 0.45389584174404524,
      "grad_norm": 103.22637939453125,
      "learning_rate": 1.1966771047251858e-05,
      "loss": 3.954,
      "step": 11243
    },
    {
      "epoch": 0.45393621316108196,
      "grad_norm": 62.6829833984375,
      "learning_rate": 1.1965488989296865e-05,
      "loss": 6.0451,
      "step": 11244
    },
    {
      "epoch": 0.4539765845781187,
      "grad_norm": 107.6717300415039,
      "learning_rate": 1.1964206897736603e-05,
      "loss": 5.346,
      "step": 11245
    },
    {
      "epoch": 0.4540169559951554,
      "grad_norm": 89.78997802734375,
      "learning_rate": 1.1962924772592987e-05,
      "loss": 6.414,
      "step": 11246
    },
    {
      "epoch": 0.4540573274121922,
      "grad_norm": 78.06779479980469,
      "learning_rate": 1.1961642613887944e-05,
      "loss": 5.2926,
      "step": 11247
    },
    {
      "epoch": 0.4540976988292289,
      "grad_norm": 77.8786392211914,
      "learning_rate": 1.1960360421643391e-05,
      "loss": 4.5519,
      "step": 11248
    },
    {
      "epoch": 0.4541380702462656,
      "grad_norm": 87.08396911621094,
      "learning_rate": 1.1959078195881253e-05,
      "loss": 4.4685,
      "step": 11249
    },
    {
      "epoch": 0.4541784416633024,
      "grad_norm": 67.0778579711914,
      "learning_rate": 1.1957795936623453e-05,
      "loss": 5.5073,
      "step": 11250
    },
    {
      "epoch": 0.4542188130803391,
      "grad_norm": 165.58761596679688,
      "learning_rate": 1.1956513643891916e-05,
      "loss": 4.5288,
      "step": 11251
    },
    {
      "epoch": 0.45425918449737585,
      "grad_norm": 97.188232421875,
      "learning_rate": 1.195523131770856e-05,
      "loss": 5.5274,
      "step": 11252
    },
    {
      "epoch": 0.4542995559144126,
      "grad_norm": 126.58981323242188,
      "learning_rate": 1.1953948958095318e-05,
      "loss": 6.0201,
      "step": 11253
    },
    {
      "epoch": 0.45433992733144934,
      "grad_norm": 98.57341003417969,
      "learning_rate": 1.1952666565074109e-05,
      "loss": 5.6532,
      "step": 11254
    },
    {
      "epoch": 0.45438029874848607,
      "grad_norm": 85.49641418457031,
      "learning_rate": 1.1951384138666866e-05,
      "loss": 5.6755,
      "step": 11255
    },
    {
      "epoch": 0.4544206701655228,
      "grad_norm": 88.24311828613281,
      "learning_rate": 1.1950101678895507e-05,
      "loss": 5.5981,
      "step": 11256
    },
    {
      "epoch": 0.45446104158255957,
      "grad_norm": 77.21753692626953,
      "learning_rate": 1.1948819185781964e-05,
      "loss": 5.7555,
      "step": 11257
    },
    {
      "epoch": 0.4545014129995963,
      "grad_norm": 169.33065795898438,
      "learning_rate": 1.1947536659348162e-05,
      "loss": 5.1663,
      "step": 11258
    },
    {
      "epoch": 0.454541784416633,
      "grad_norm": 106.73129272460938,
      "learning_rate": 1.1946254099616035e-05,
      "loss": 3.5434,
      "step": 11259
    },
    {
      "epoch": 0.4545821558336698,
      "grad_norm": 124.64366912841797,
      "learning_rate": 1.1944971506607503e-05,
      "loss": 4.5161,
      "step": 11260
    },
    {
      "epoch": 0.4546225272507065,
      "grad_norm": 143.05633544921875,
      "learning_rate": 1.1943688880344504e-05,
      "loss": 5.8132,
      "step": 11261
    },
    {
      "epoch": 0.45466289866774323,
      "grad_norm": 88.3592300415039,
      "learning_rate": 1.1942406220848963e-05,
      "loss": 4.7016,
      "step": 11262
    },
    {
      "epoch": 0.45470327008477995,
      "grad_norm": 4377.17822265625,
      "learning_rate": 1.1941123528142813e-05,
      "loss": 5.7686,
      "step": 11263
    },
    {
      "epoch": 0.45474364150181673,
      "grad_norm": 101.18435668945312,
      "learning_rate": 1.1939840802247979e-05,
      "loss": 5.7236,
      "step": 11264
    },
    {
      "epoch": 0.45478401291885345,
      "grad_norm": 114.89850616455078,
      "learning_rate": 1.1938558043186401e-05,
      "loss": 5.5831,
      "step": 11265
    },
    {
      "epoch": 0.4548243843358902,
      "grad_norm": 87.7291259765625,
      "learning_rate": 1.1937275250980008e-05,
      "loss": 5.8738,
      "step": 11266
    },
    {
      "epoch": 0.45486475575292695,
      "grad_norm": 153.70603942871094,
      "learning_rate": 1.1935992425650726e-05,
      "loss": 6.2763,
      "step": 11267
    },
    {
      "epoch": 0.4549051271699637,
      "grad_norm": 81.03101348876953,
      "learning_rate": 1.19347095672205e-05,
      "loss": 6.2584,
      "step": 11268
    },
    {
      "epoch": 0.4549454985870004,
      "grad_norm": 85.55657958984375,
      "learning_rate": 1.1933426675711256e-05,
      "loss": 5.6488,
      "step": 11269
    },
    {
      "epoch": 0.4549858700040371,
      "grad_norm": 94.68363952636719,
      "learning_rate": 1.1932143751144931e-05,
      "loss": 4.5111,
      "step": 11270
    },
    {
      "epoch": 0.4550262414210739,
      "grad_norm": 122.27684783935547,
      "learning_rate": 1.1930860793543462e-05,
      "loss": 3.6213,
      "step": 11271
    },
    {
      "epoch": 0.4550666128381106,
      "grad_norm": 75.54753875732422,
      "learning_rate": 1.192957780292878e-05,
      "loss": 5.3876,
      "step": 11272
    },
    {
      "epoch": 0.45510698425514734,
      "grad_norm": 80.46981048583984,
      "learning_rate": 1.1928294779322825e-05,
      "loss": 4.1941,
      "step": 11273
    },
    {
      "epoch": 0.4551473556721841,
      "grad_norm": 82.641845703125,
      "learning_rate": 1.192701172274753e-05,
      "loss": 5.5479,
      "step": 11274
    },
    {
      "epoch": 0.45518772708922084,
      "grad_norm": 136.8614044189453,
      "learning_rate": 1.1925728633224838e-05,
      "loss": 4.8333,
      "step": 11275
    },
    {
      "epoch": 0.45522809850625756,
      "grad_norm": 95.557373046875,
      "learning_rate": 1.1924445510776682e-05,
      "loss": 5.2513,
      "step": 11276
    },
    {
      "epoch": 0.4552684699232943,
      "grad_norm": 109.60804748535156,
      "learning_rate": 1.1923162355424998e-05,
      "loss": 5.3005,
      "step": 11277
    },
    {
      "epoch": 0.45530884134033106,
      "grad_norm": 85.08305358886719,
      "learning_rate": 1.192187916719173e-05,
      "loss": 5.9201,
      "step": 11278
    },
    {
      "epoch": 0.4553492127573678,
      "grad_norm": 67.27719116210938,
      "learning_rate": 1.192059594609882e-05,
      "loss": 4.7212,
      "step": 11279
    },
    {
      "epoch": 0.4553895841744045,
      "grad_norm": 131.97897338867188,
      "learning_rate": 1.1919312692168198e-05,
      "loss": 7.0762,
      "step": 11280
    },
    {
      "epoch": 0.4554299555914413,
      "grad_norm": 85.5535659790039,
      "learning_rate": 1.1918029405421816e-05,
      "loss": 5.4506,
      "step": 11281
    },
    {
      "epoch": 0.455470327008478,
      "grad_norm": 188.605224609375,
      "learning_rate": 1.1916746085881607e-05,
      "loss": 3.8276,
      "step": 11282
    },
    {
      "epoch": 0.4555106984255147,
      "grad_norm": 310.3922424316406,
      "learning_rate": 1.191546273356952e-05,
      "loss": 3.0155,
      "step": 11283
    },
    {
      "epoch": 0.45555106984255145,
      "grad_norm": 109.8934555053711,
      "learning_rate": 1.191417934850749e-05,
      "loss": 4.7359,
      "step": 11284
    },
    {
      "epoch": 0.4555914412595882,
      "grad_norm": 84.30181121826172,
      "learning_rate": 1.1912895930717465e-05,
      "loss": 5.5241,
      "step": 11285
    },
    {
      "epoch": 0.45563181267662495,
      "grad_norm": 94.31478881835938,
      "learning_rate": 1.1911612480221385e-05,
      "loss": 4.3423,
      "step": 11286
    },
    {
      "epoch": 0.45567218409366167,
      "grad_norm": 142.44430541992188,
      "learning_rate": 1.1910328997041195e-05,
      "loss": 5.4186,
      "step": 11287
    },
    {
      "epoch": 0.45571255551069845,
      "grad_norm": 94.10595703125,
      "learning_rate": 1.190904548119884e-05,
      "loss": 4.2799,
      "step": 11288
    },
    {
      "epoch": 0.45575292692773517,
      "grad_norm": 236.93035888671875,
      "learning_rate": 1.1907761932716266e-05,
      "loss": 4.7234,
      "step": 11289
    },
    {
      "epoch": 0.4557932983447719,
      "grad_norm": 95.41509246826172,
      "learning_rate": 1.1906478351615418e-05,
      "loss": 5.5757,
      "step": 11290
    },
    {
      "epoch": 0.4558336697618086,
      "grad_norm": 106.40838623046875,
      "learning_rate": 1.1905194737918242e-05,
      "loss": 4.9557,
      "step": 11291
    },
    {
      "epoch": 0.4558740411788454,
      "grad_norm": 98.47359466552734,
      "learning_rate": 1.1903911091646684e-05,
      "loss": 5.3135,
      "step": 11292
    },
    {
      "epoch": 0.4559144125958821,
      "grad_norm": 68.40935516357422,
      "learning_rate": 1.1902627412822695e-05,
      "loss": 5.087,
      "step": 11293
    },
    {
      "epoch": 0.45595478401291883,
      "grad_norm": 95.48441314697266,
      "learning_rate": 1.1901343701468218e-05,
      "loss": 5.2337,
      "step": 11294
    },
    {
      "epoch": 0.4559951554299556,
      "grad_norm": 90.04844665527344,
      "learning_rate": 1.1900059957605205e-05,
      "loss": 5.983,
      "step": 11295
    },
    {
      "epoch": 0.45603552684699233,
      "grad_norm": 97.75518798828125,
      "learning_rate": 1.1898776181255603e-05,
      "loss": 5.751,
      "step": 11296
    },
    {
      "epoch": 0.45607589826402906,
      "grad_norm": 105.3473129272461,
      "learning_rate": 1.1897492372441362e-05,
      "loss": 6.3917,
      "step": 11297
    },
    {
      "epoch": 0.4561162696810658,
      "grad_norm": 186.57760620117188,
      "learning_rate": 1.189620853118443e-05,
      "loss": 5.569,
      "step": 11298
    },
    {
      "epoch": 0.45615664109810256,
      "grad_norm": 82.1430892944336,
      "learning_rate": 1.1894924657506763e-05,
      "loss": 4.6243,
      "step": 11299
    },
    {
      "epoch": 0.4561970125151393,
      "grad_norm": 108.63945770263672,
      "learning_rate": 1.1893640751430309e-05,
      "loss": 5.8583,
      "step": 11300
    },
    {
      "epoch": 0.456237383932176,
      "grad_norm": 166.6332550048828,
      "learning_rate": 1.1892356812977021e-05,
      "loss": 7.373,
      "step": 11301
    },
    {
      "epoch": 0.4562777553492128,
      "grad_norm": 86.07841491699219,
      "learning_rate": 1.1891072842168849e-05,
      "loss": 4.7324,
      "step": 11302
    },
    {
      "epoch": 0.4563181267662495,
      "grad_norm": 153.49496459960938,
      "learning_rate": 1.1889788839027749e-05,
      "loss": 6.7374,
      "step": 11303
    },
    {
      "epoch": 0.4563584981832862,
      "grad_norm": 91.06914520263672,
      "learning_rate": 1.1888504803575672e-05,
      "loss": 5.6253,
      "step": 11304
    },
    {
      "epoch": 0.456398869600323,
      "grad_norm": 105.1946792602539,
      "learning_rate": 1.1887220735834572e-05,
      "loss": 5.7398,
      "step": 11305
    },
    {
      "epoch": 0.4564392410173597,
      "grad_norm": 72.02757263183594,
      "learning_rate": 1.1885936635826404e-05,
      "loss": 6.599,
      "step": 11306
    },
    {
      "epoch": 0.45647961243439644,
      "grad_norm": 86.5334701538086,
      "learning_rate": 1.1884652503573124e-05,
      "loss": 5.892,
      "step": 11307
    },
    {
      "epoch": 0.45651998385143316,
      "grad_norm": 168.2843780517578,
      "learning_rate": 1.1883368339096685e-05,
      "loss": 3.4017,
      "step": 11308
    },
    {
      "epoch": 0.45656035526846994,
      "grad_norm": 86.46774291992188,
      "learning_rate": 1.1882084142419053e-05,
      "loss": 6.2476,
      "step": 11309
    },
    {
      "epoch": 0.45660072668550666,
      "grad_norm": 69.7755126953125,
      "learning_rate": 1.1880799913562168e-05,
      "loss": 4.9913,
      "step": 11310
    },
    {
      "epoch": 0.4566410981025434,
      "grad_norm": 103.0035629272461,
      "learning_rate": 1.1879515652548004e-05,
      "loss": 5.7883,
      "step": 11311
    },
    {
      "epoch": 0.45668146951958016,
      "grad_norm": 112.30321502685547,
      "learning_rate": 1.1878231359398506e-05,
      "loss": 5.8186,
      "step": 11312
    },
    {
      "epoch": 0.4567218409366169,
      "grad_norm": 83.7154769897461,
      "learning_rate": 1.1876947034135641e-05,
      "loss": 5.1213,
      "step": 11313
    },
    {
      "epoch": 0.4567622123536536,
      "grad_norm": 89.23181915283203,
      "learning_rate": 1.1875662676781363e-05,
      "loss": 5.848,
      "step": 11314
    },
    {
      "epoch": 0.45680258377069033,
      "grad_norm": 144.19003295898438,
      "learning_rate": 1.1874378287357636e-05,
      "loss": 5.3896,
      "step": 11315
    },
    {
      "epoch": 0.4568429551877271,
      "grad_norm": 81.44641876220703,
      "learning_rate": 1.187309386588641e-05,
      "loss": 5.5332,
      "step": 11316
    },
    {
      "epoch": 0.45688332660476383,
      "grad_norm": 76.3310317993164,
      "learning_rate": 1.1871809412389659e-05,
      "loss": 4.3539,
      "step": 11317
    },
    {
      "epoch": 0.45692369802180055,
      "grad_norm": 97.81434631347656,
      "learning_rate": 1.1870524926889335e-05,
      "loss": 5.7985,
      "step": 11318
    },
    {
      "epoch": 0.45696406943883733,
      "grad_norm": 102.33895874023438,
      "learning_rate": 1.1869240409407406e-05,
      "loss": 5.5019,
      "step": 11319
    },
    {
      "epoch": 0.45700444085587405,
      "grad_norm": 125.39031219482422,
      "learning_rate": 1.186795585996583e-05,
      "loss": 3.8164,
      "step": 11320
    },
    {
      "epoch": 0.45704481227291077,
      "grad_norm": 124.44747924804688,
      "learning_rate": 1.186667127858657e-05,
      "loss": 5.8533,
      "step": 11321
    },
    {
      "epoch": 0.4570851836899475,
      "grad_norm": 91.7364730834961,
      "learning_rate": 1.1865386665291591e-05,
      "loss": 4.3102,
      "step": 11322
    },
    {
      "epoch": 0.45712555510698427,
      "grad_norm": 107.0179672241211,
      "learning_rate": 1.1864102020102854e-05,
      "loss": 5.417,
      "step": 11323
    },
    {
      "epoch": 0.457165926524021,
      "grad_norm": 111.12788391113281,
      "learning_rate": 1.1862817343042327e-05,
      "loss": 5.7417,
      "step": 11324
    },
    {
      "epoch": 0.4572062979410577,
      "grad_norm": 103.17404174804688,
      "learning_rate": 1.1861532634131974e-05,
      "loss": 6.1471,
      "step": 11325
    },
    {
      "epoch": 0.4572466693580945,
      "grad_norm": 302.62255859375,
      "learning_rate": 1.186024789339376e-05,
      "loss": 5.2087,
      "step": 11326
    },
    {
      "epoch": 0.4572870407751312,
      "grad_norm": 100.8694076538086,
      "learning_rate": 1.185896312084965e-05,
      "loss": 4.39,
      "step": 11327
    },
    {
      "epoch": 0.45732741219216794,
      "grad_norm": 88.34700012207031,
      "learning_rate": 1.1857678316521611e-05,
      "loss": 6.1313,
      "step": 11328
    },
    {
      "epoch": 0.45736778360920466,
      "grad_norm": 80.84452056884766,
      "learning_rate": 1.1856393480431609e-05,
      "loss": 5.6843,
      "step": 11329
    },
    {
      "epoch": 0.45740815502624144,
      "grad_norm": 85.23072052001953,
      "learning_rate": 1.1855108612601618e-05,
      "loss": 5.0794,
      "step": 11330
    },
    {
      "epoch": 0.45744852644327816,
      "grad_norm": 94.58679962158203,
      "learning_rate": 1.18538237130536e-05,
      "loss": 4.3204,
      "step": 11331
    },
    {
      "epoch": 0.4574888978603149,
      "grad_norm": 108.42425537109375,
      "learning_rate": 1.1852538781809526e-05,
      "loss": 4.7833,
      "step": 11332
    },
    {
      "epoch": 0.45752926927735166,
      "grad_norm": 113.4568099975586,
      "learning_rate": 1.1851253818891364e-05,
      "loss": 4.3807,
      "step": 11333
    },
    {
      "epoch": 0.4575696406943884,
      "grad_norm": 88.16899871826172,
      "learning_rate": 1.1849968824321086e-05,
      "loss": 4.8462,
      "step": 11334
    },
    {
      "epoch": 0.4576100121114251,
      "grad_norm": 121.21104431152344,
      "learning_rate": 1.184868379812066e-05,
      "loss": 5.1978,
      "step": 11335
    },
    {
      "epoch": 0.4576503835284618,
      "grad_norm": 77.22032165527344,
      "learning_rate": 1.1847398740312057e-05,
      "loss": 4.4583,
      "step": 11336
    },
    {
      "epoch": 0.4576907549454986,
      "grad_norm": 66.5023422241211,
      "learning_rate": 1.184611365091725e-05,
      "loss": 5.1875,
      "step": 11337
    },
    {
      "epoch": 0.4577311263625353,
      "grad_norm": 132.19651794433594,
      "learning_rate": 1.184482852995821e-05,
      "loss": 5.6338,
      "step": 11338
    },
    {
      "epoch": 0.45777149777957205,
      "grad_norm": 117.72511291503906,
      "learning_rate": 1.184354337745691e-05,
      "loss": 4.5624,
      "step": 11339
    },
    {
      "epoch": 0.4578118691966088,
      "grad_norm": 102.08629608154297,
      "learning_rate": 1.1842258193435327e-05,
      "loss": 6.9059,
      "step": 11340
    },
    {
      "epoch": 0.45785224061364554,
      "grad_norm": 127.91092681884766,
      "learning_rate": 1.1840972977915425e-05,
      "loss": 4.5067,
      "step": 11341
    },
    {
      "epoch": 0.45789261203068227,
      "grad_norm": 121.80606079101562,
      "learning_rate": 1.1839687730919191e-05,
      "loss": 5.6266,
      "step": 11342
    },
    {
      "epoch": 0.457932983447719,
      "grad_norm": 93.53726196289062,
      "learning_rate": 1.1838402452468587e-05,
      "loss": 4.1052,
      "step": 11343
    },
    {
      "epoch": 0.45797335486475577,
      "grad_norm": 104.90900421142578,
      "learning_rate": 1.18371171425856e-05,
      "loss": 6.5421,
      "step": 11344
    },
    {
      "epoch": 0.4580137262817925,
      "grad_norm": 79.77487182617188,
      "learning_rate": 1.1835831801292193e-05,
      "loss": 5.0585,
      "step": 11345
    },
    {
      "epoch": 0.4580540976988292,
      "grad_norm": 285.1929626464844,
      "learning_rate": 1.1834546428610354e-05,
      "loss": 5.8468,
      "step": 11346
    },
    {
      "epoch": 0.458094469115866,
      "grad_norm": 95.84935760498047,
      "learning_rate": 1.1833261024562052e-05,
      "loss": 4.3031,
      "step": 11347
    },
    {
      "epoch": 0.4581348405329027,
      "grad_norm": 115.68962097167969,
      "learning_rate": 1.1831975589169268e-05,
      "loss": 3.935,
      "step": 11348
    },
    {
      "epoch": 0.45817521194993943,
      "grad_norm": 94.23290252685547,
      "learning_rate": 1.1830690122453979e-05,
      "loss": 6.1472,
      "step": 11349
    },
    {
      "epoch": 0.45821558336697615,
      "grad_norm": 67.58715057373047,
      "learning_rate": 1.1829404624438166e-05,
      "loss": 6.0315,
      "step": 11350
    },
    {
      "epoch": 0.45825595478401293,
      "grad_norm": 98.8604736328125,
      "learning_rate": 1.1828119095143804e-05,
      "loss": 4.7005,
      "step": 11351
    },
    {
      "epoch": 0.45829632620104965,
      "grad_norm": 102.15674591064453,
      "learning_rate": 1.1826833534592876e-05,
      "loss": 5.7358,
      "step": 11352
    },
    {
      "epoch": 0.4583366976180864,
      "grad_norm": 72.48548889160156,
      "learning_rate": 1.182554794280736e-05,
      "loss": 5.9937,
      "step": 11353
    },
    {
      "epoch": 0.45837706903512315,
      "grad_norm": 103.61705780029297,
      "learning_rate": 1.1824262319809238e-05,
      "loss": 6.2474,
      "step": 11354
    },
    {
      "epoch": 0.4584174404521599,
      "grad_norm": 78.27631378173828,
      "learning_rate": 1.1822976665620489e-05,
      "loss": 5.6331,
      "step": 11355
    },
    {
      "epoch": 0.4584578118691966,
      "grad_norm": 88.62200927734375,
      "learning_rate": 1.1821690980263097e-05,
      "loss": 6.7732,
      "step": 11356
    },
    {
      "epoch": 0.4584981832862334,
      "grad_norm": 116.00852966308594,
      "learning_rate": 1.1820405263759041e-05,
      "loss": 5.2001,
      "step": 11357
    },
    {
      "epoch": 0.4585385547032701,
      "grad_norm": 81.51701354980469,
      "learning_rate": 1.1819119516130306e-05,
      "loss": 3.5648,
      "step": 11358
    },
    {
      "epoch": 0.4585789261203068,
      "grad_norm": 87.03723907470703,
      "learning_rate": 1.1817833737398877e-05,
      "loss": 5.181,
      "step": 11359
    },
    {
      "epoch": 0.45861929753734354,
      "grad_norm": 106.77233123779297,
      "learning_rate": 1.1816547927586737e-05,
      "loss": 6.2602,
      "step": 11360
    },
    {
      "epoch": 0.4586596689543803,
      "grad_norm": 105.36616516113281,
      "learning_rate": 1.1815262086715867e-05,
      "loss": 5.6652,
      "step": 11361
    },
    {
      "epoch": 0.45870004037141704,
      "grad_norm": 105.86030578613281,
      "learning_rate": 1.1813976214808255e-05,
      "loss": 5.7018,
      "step": 11362
    },
    {
      "epoch": 0.45874041178845376,
      "grad_norm": 146.84918212890625,
      "learning_rate": 1.1812690311885885e-05,
      "loss": 3.805,
      "step": 11363
    },
    {
      "epoch": 0.45878078320549054,
      "grad_norm": 155.16220092773438,
      "learning_rate": 1.1811404377970746e-05,
      "loss": 5.881,
      "step": 11364
    },
    {
      "epoch": 0.45882115462252726,
      "grad_norm": 66.3902359008789,
      "learning_rate": 1.1810118413084819e-05,
      "loss": 6.1739,
      "step": 11365
    },
    {
      "epoch": 0.458861526039564,
      "grad_norm": 86.0765609741211,
      "learning_rate": 1.1808832417250095e-05,
      "loss": 4.9791,
      "step": 11366
    },
    {
      "epoch": 0.4589018974566007,
      "grad_norm": 91.84187316894531,
      "learning_rate": 1.1807546390488561e-05,
      "loss": 5.0525,
      "step": 11367
    },
    {
      "epoch": 0.4589422688736375,
      "grad_norm": 122.40250396728516,
      "learning_rate": 1.1806260332822206e-05,
      "loss": 5.6373,
      "step": 11368
    },
    {
      "epoch": 0.4589826402906742,
      "grad_norm": 83.37958526611328,
      "learning_rate": 1.1804974244273015e-05,
      "loss": 4.7695,
      "step": 11369
    },
    {
      "epoch": 0.4590230117077109,
      "grad_norm": 83.75428771972656,
      "learning_rate": 1.1803688124862982e-05,
      "loss": 4.8781,
      "step": 11370
    },
    {
      "epoch": 0.4590633831247477,
      "grad_norm": 95.3494644165039,
      "learning_rate": 1.180240197461409e-05,
      "loss": 5.3514,
      "step": 11371
    },
    {
      "epoch": 0.4591037545417844,
      "grad_norm": 125.89533996582031,
      "learning_rate": 1.1801115793548338e-05,
      "loss": 5.0185,
      "step": 11372
    },
    {
      "epoch": 0.45914412595882115,
      "grad_norm": 143.81268310546875,
      "learning_rate": 1.1799829581687709e-05,
      "loss": 4.5665,
      "step": 11373
    },
    {
      "epoch": 0.45918449737585787,
      "grad_norm": 89.79817962646484,
      "learning_rate": 1.1798543339054199e-05,
      "loss": 6.4284,
      "step": 11374
    },
    {
      "epoch": 0.45922486879289465,
      "grad_norm": 91.91142272949219,
      "learning_rate": 1.1797257065669796e-05,
      "loss": 5.8922,
      "step": 11375
    },
    {
      "epoch": 0.45926524020993137,
      "grad_norm": 117.6109619140625,
      "learning_rate": 1.1795970761556493e-05,
      "loss": 4.2112,
      "step": 11376
    },
    {
      "epoch": 0.4593056116269681,
      "grad_norm": 98.45620727539062,
      "learning_rate": 1.1794684426736285e-05,
      "loss": 3.6427,
      "step": 11377
    },
    {
      "epoch": 0.45934598304400487,
      "grad_norm": 100.0252914428711,
      "learning_rate": 1.1793398061231166e-05,
      "loss": 5.5266,
      "step": 11378
    },
    {
      "epoch": 0.4593863544610416,
      "grad_norm": 98.23056030273438,
      "learning_rate": 1.1792111665063125e-05,
      "loss": 6.6502,
      "step": 11379
    },
    {
      "epoch": 0.4594267258780783,
      "grad_norm": 104.15425109863281,
      "learning_rate": 1.1790825238254162e-05,
      "loss": 4.5631,
      "step": 11380
    },
    {
      "epoch": 0.45946709729511503,
      "grad_norm": 71.89849090576172,
      "learning_rate": 1.1789538780826265e-05,
      "loss": 5.666,
      "step": 11381
    },
    {
      "epoch": 0.4595074687121518,
      "grad_norm": 129.1355743408203,
      "learning_rate": 1.1788252292801438e-05,
      "loss": 3.5133,
      "step": 11382
    },
    {
      "epoch": 0.45954784012918853,
      "grad_norm": 113.67145538330078,
      "learning_rate": 1.1786965774201673e-05,
      "loss": 5.6846,
      "step": 11383
    },
    {
      "epoch": 0.45958821154622526,
      "grad_norm": 151.4600830078125,
      "learning_rate": 1.1785679225048965e-05,
      "loss": 4.958,
      "step": 11384
    },
    {
      "epoch": 0.45962858296326203,
      "grad_norm": 88.45153045654297,
      "learning_rate": 1.178439264536531e-05,
      "loss": 4.0328,
      "step": 11385
    },
    {
      "epoch": 0.45966895438029876,
      "grad_norm": 122.5910415649414,
      "learning_rate": 1.1783106035172708e-05,
      "loss": 5.9985,
      "step": 11386
    },
    {
      "epoch": 0.4597093257973355,
      "grad_norm": 64.33767700195312,
      "learning_rate": 1.1781819394493159e-05,
      "loss": 4.9794,
      "step": 11387
    },
    {
      "epoch": 0.4597496972143722,
      "grad_norm": 108.413818359375,
      "learning_rate": 1.1780532723348659e-05,
      "loss": 3.934,
      "step": 11388
    },
    {
      "epoch": 0.459790068631409,
      "grad_norm": 89.87410736083984,
      "learning_rate": 1.1779246021761206e-05,
      "loss": 5.5025,
      "step": 11389
    },
    {
      "epoch": 0.4598304400484457,
      "grad_norm": 77.33467864990234,
      "learning_rate": 1.1777959289752803e-05,
      "loss": 6.1213,
      "step": 11390
    },
    {
      "epoch": 0.4598708114654824,
      "grad_norm": 114.92139434814453,
      "learning_rate": 1.1776672527345446e-05,
      "loss": 5.2,
      "step": 11391
    },
    {
      "epoch": 0.4599111828825192,
      "grad_norm": 114.52304077148438,
      "learning_rate": 1.1775385734561137e-05,
      "loss": 3.7078,
      "step": 11392
    },
    {
      "epoch": 0.4599515542995559,
      "grad_norm": 104.505615234375,
      "learning_rate": 1.1774098911421879e-05,
      "loss": 3.5424,
      "step": 11393
    },
    {
      "epoch": 0.45999192571659264,
      "grad_norm": 153.5265655517578,
      "learning_rate": 1.1772812057949674e-05,
      "loss": 4.3979,
      "step": 11394
    },
    {
      "epoch": 0.46003229713362936,
      "grad_norm": 116.93144226074219,
      "learning_rate": 1.1771525174166519e-05,
      "loss": 5.5472,
      "step": 11395
    },
    {
      "epoch": 0.46007266855066614,
      "grad_norm": 109.30178833007812,
      "learning_rate": 1.1770238260094423e-05,
      "loss": 4.5677,
      "step": 11396
    },
    {
      "epoch": 0.46011303996770286,
      "grad_norm": 118.9994888305664,
      "learning_rate": 1.1768951315755383e-05,
      "loss": 4.7796,
      "step": 11397
    },
    {
      "epoch": 0.4601534113847396,
      "grad_norm": 142.54518127441406,
      "learning_rate": 1.1767664341171409e-05,
      "loss": 4.5974,
      "step": 11398
    },
    {
      "epoch": 0.46019378280177636,
      "grad_norm": 84.32989501953125,
      "learning_rate": 1.1766377336364503e-05,
      "loss": 5.9901,
      "step": 11399
    },
    {
      "epoch": 0.4602341542188131,
      "grad_norm": 88.84762573242188,
      "learning_rate": 1.176509030135667e-05,
      "loss": 6.4333,
      "step": 11400
    },
    {
      "epoch": 0.4602745256358498,
      "grad_norm": 73.26731872558594,
      "learning_rate": 1.1763803236169913e-05,
      "loss": 4.6481,
      "step": 11401
    },
    {
      "epoch": 0.46031489705288653,
      "grad_norm": 100.68907928466797,
      "learning_rate": 1.1762516140826241e-05,
      "loss": 5.608,
      "step": 11402
    },
    {
      "epoch": 0.4603552684699233,
      "grad_norm": 69.71560668945312,
      "learning_rate": 1.1761229015347657e-05,
      "loss": 4.6732,
      "step": 11403
    },
    {
      "epoch": 0.46039563988696003,
      "grad_norm": 110.96944427490234,
      "learning_rate": 1.1759941859756173e-05,
      "loss": 6.103,
      "step": 11404
    },
    {
      "epoch": 0.46043601130399675,
      "grad_norm": 84.54923248291016,
      "learning_rate": 1.1758654674073789e-05,
      "loss": 5.5067,
      "step": 11405
    },
    {
      "epoch": 0.46047638272103353,
      "grad_norm": 135.23114013671875,
      "learning_rate": 1.1757367458322518e-05,
      "loss": 4.4599,
      "step": 11406
    },
    {
      "epoch": 0.46051675413807025,
      "grad_norm": 114.05848693847656,
      "learning_rate": 1.1756080212524364e-05,
      "loss": 5.3403,
      "step": 11407
    },
    {
      "epoch": 0.46055712555510697,
      "grad_norm": 113.0404281616211,
      "learning_rate": 1.1754792936701345e-05,
      "loss": 4.8002,
      "step": 11408
    },
    {
      "epoch": 0.46059749697214375,
      "grad_norm": 118.4676513671875,
      "learning_rate": 1.1753505630875459e-05,
      "loss": 5.7173,
      "step": 11409
    },
    {
      "epoch": 0.46063786838918047,
      "grad_norm": 132.12496948242188,
      "learning_rate": 1.1752218295068726e-05,
      "loss": 5.3673,
      "step": 11410
    },
    {
      "epoch": 0.4606782398062172,
      "grad_norm": 132.5874481201172,
      "learning_rate": 1.175093092930315e-05,
      "loss": 4.0818,
      "step": 11411
    },
    {
      "epoch": 0.4607186112232539,
      "grad_norm": 171.3387908935547,
      "learning_rate": 1.1749643533600746e-05,
      "loss": 4.0346,
      "step": 11412
    },
    {
      "epoch": 0.4607589826402907,
      "grad_norm": 140.70570373535156,
      "learning_rate": 1.1748356107983521e-05,
      "loss": 4.9527,
      "step": 11413
    },
    {
      "epoch": 0.4607993540573274,
      "grad_norm": 159.36285400390625,
      "learning_rate": 1.174706865247349e-05,
      "loss": 6.6381,
      "step": 11414
    },
    {
      "epoch": 0.46083972547436414,
      "grad_norm": 94.73423767089844,
      "learning_rate": 1.1745781167092663e-05,
      "loss": 6.6837,
      "step": 11415
    },
    {
      "epoch": 0.4608800968914009,
      "grad_norm": 1035.392333984375,
      "learning_rate": 1.1744493651863056e-05,
      "loss": 4.97,
      "step": 11416
    },
    {
      "epoch": 0.46092046830843764,
      "grad_norm": 108.52295684814453,
      "learning_rate": 1.174320610680668e-05,
      "loss": 5.1907,
      "step": 11417
    },
    {
      "epoch": 0.46096083972547436,
      "grad_norm": 100.22118377685547,
      "learning_rate": 1.1741918531945554e-05,
      "loss": 4.7548,
      "step": 11418
    },
    {
      "epoch": 0.4610012111425111,
      "grad_norm": 154.2816162109375,
      "learning_rate": 1.1740630927301683e-05,
      "loss": 4.3549,
      "step": 11419
    },
    {
      "epoch": 0.46104158255954786,
      "grad_norm": 136.62669372558594,
      "learning_rate": 1.1739343292897094e-05,
      "loss": 4.8757,
      "step": 11420
    },
    {
      "epoch": 0.4610819539765846,
      "grad_norm": 126.56205749511719,
      "learning_rate": 1.1738055628753792e-05,
      "loss": 4.9142,
      "step": 11421
    },
    {
      "epoch": 0.4611223253936213,
      "grad_norm": 81.38701629638672,
      "learning_rate": 1.17367679348938e-05,
      "loss": 5.7897,
      "step": 11422
    },
    {
      "epoch": 0.4611626968106581,
      "grad_norm": 125.73680114746094,
      "learning_rate": 1.1735480211339132e-05,
      "loss": 5.7231,
      "step": 11423
    },
    {
      "epoch": 0.4612030682276948,
      "grad_norm": 222.7443389892578,
      "learning_rate": 1.1734192458111803e-05,
      "loss": 4.609,
      "step": 11424
    },
    {
      "epoch": 0.4612434396447315,
      "grad_norm": 127.16219329833984,
      "learning_rate": 1.1732904675233834e-05,
      "loss": 4.3974,
      "step": 11425
    },
    {
      "epoch": 0.46128381106176825,
      "grad_norm": 134.40501403808594,
      "learning_rate": 1.173161686272724e-05,
      "loss": 4.7657,
      "step": 11426
    },
    {
      "epoch": 0.461324182478805,
      "grad_norm": 174.11566162109375,
      "learning_rate": 1.1730329020614042e-05,
      "loss": 5.1205,
      "step": 11427
    },
    {
      "epoch": 0.46136455389584174,
      "grad_norm": 91.30294799804688,
      "learning_rate": 1.172904114891626e-05,
      "loss": 5.7592,
      "step": 11428
    },
    {
      "epoch": 0.46140492531287847,
      "grad_norm": 77.0135726928711,
      "learning_rate": 1.172775324765591e-05,
      "loss": 4.254,
      "step": 11429
    },
    {
      "epoch": 0.46144529672991524,
      "grad_norm": 134.26463317871094,
      "learning_rate": 1.1726465316855016e-05,
      "loss": 5.4101,
      "step": 11430
    },
    {
      "epoch": 0.46148566814695197,
      "grad_norm": 99.15702056884766,
      "learning_rate": 1.1725177356535593e-05,
      "loss": 5.1838,
      "step": 11431
    },
    {
      "epoch": 0.4615260395639887,
      "grad_norm": 89.43666076660156,
      "learning_rate": 1.172388936671967e-05,
      "loss": 5.0755,
      "step": 11432
    },
    {
      "epoch": 0.4615664109810254,
      "grad_norm": 566.6268310546875,
      "learning_rate": 1.1722601347429264e-05,
      "loss": 4.0304,
      "step": 11433
    },
    {
      "epoch": 0.4616067823980622,
      "grad_norm": 90.84701538085938,
      "learning_rate": 1.1721313298686397e-05,
      "loss": 5.7047,
      "step": 11434
    },
    {
      "epoch": 0.4616471538150989,
      "grad_norm": 79.31932067871094,
      "learning_rate": 1.172002522051309e-05,
      "loss": 5.2983,
      "step": 11435
    },
    {
      "epoch": 0.46168752523213563,
      "grad_norm": 113.97957611083984,
      "learning_rate": 1.1718737112931369e-05,
      "loss": 3.6151,
      "step": 11436
    },
    {
      "epoch": 0.4617278966491724,
      "grad_norm": 84.23637390136719,
      "learning_rate": 1.1717448975963254e-05,
      "loss": 5.0943,
      "step": 11437
    },
    {
      "epoch": 0.46176826806620913,
      "grad_norm": 91.34780883789062,
      "learning_rate": 1.1716160809630778e-05,
      "loss": 4.038,
      "step": 11438
    },
    {
      "epoch": 0.46180863948324585,
      "grad_norm": 115.0656967163086,
      "learning_rate": 1.1714872613955957e-05,
      "loss": 5.0224,
      "step": 11439
    },
    {
      "epoch": 0.4618490109002826,
      "grad_norm": 156.5421905517578,
      "learning_rate": 1.171358438896082e-05,
      "loss": 5.0228,
      "step": 11440
    },
    {
      "epoch": 0.46188938231731935,
      "grad_norm": 161.91941833496094,
      "learning_rate": 1.1712296134667389e-05,
      "loss": 5.1752,
      "step": 11441
    },
    {
      "epoch": 0.4619297537343561,
      "grad_norm": 94.39185333251953,
      "learning_rate": 1.1711007851097695e-05,
      "loss": 3.1574,
      "step": 11442
    },
    {
      "epoch": 0.4619701251513928,
      "grad_norm": 67.96414184570312,
      "learning_rate": 1.170971953827376e-05,
      "loss": 5.7616,
      "step": 11443
    },
    {
      "epoch": 0.4620104965684296,
      "grad_norm": 95.85411834716797,
      "learning_rate": 1.1708431196217618e-05,
      "loss": 6.8801,
      "step": 11444
    },
    {
      "epoch": 0.4620508679854663,
      "grad_norm": 93.34026336669922,
      "learning_rate": 1.1707142824951288e-05,
      "loss": 6.5664,
      "step": 11445
    },
    {
      "epoch": 0.462091239402503,
      "grad_norm": 102.66374206542969,
      "learning_rate": 1.17058544244968e-05,
      "loss": 5.9539,
      "step": 11446
    },
    {
      "epoch": 0.46213161081953974,
      "grad_norm": 238.9991455078125,
      "learning_rate": 1.170456599487619e-05,
      "loss": 4.8055,
      "step": 11447
    },
    {
      "epoch": 0.4621719822365765,
      "grad_norm": 115.05136108398438,
      "learning_rate": 1.170327753611148e-05,
      "loss": 5.9903,
      "step": 11448
    },
    {
      "epoch": 0.46221235365361324,
      "grad_norm": 158.6896514892578,
      "learning_rate": 1.17019890482247e-05,
      "loss": 5.2292,
      "step": 11449
    },
    {
      "epoch": 0.46225272507064996,
      "grad_norm": 88.68702697753906,
      "learning_rate": 1.1700700531237883e-05,
      "loss": 4.8241,
      "step": 11450
    },
    {
      "epoch": 0.46229309648768674,
      "grad_norm": 68.58771514892578,
      "learning_rate": 1.1699411985173057e-05,
      "loss": 4.468,
      "step": 11451
    },
    {
      "epoch": 0.46233346790472346,
      "grad_norm": 142.93112182617188,
      "learning_rate": 1.1698123410052258e-05,
      "loss": 6.0286,
      "step": 11452
    },
    {
      "epoch": 0.4623738393217602,
      "grad_norm": 104.72015380859375,
      "learning_rate": 1.169683480589751e-05,
      "loss": 4.9277,
      "step": 11453
    },
    {
      "epoch": 0.4624142107387969,
      "grad_norm": 136.39999389648438,
      "learning_rate": 1.1695546172730854e-05,
      "loss": 5.9491,
      "step": 11454
    },
    {
      "epoch": 0.4624545821558337,
      "grad_norm": 79.85980224609375,
      "learning_rate": 1.1694257510574313e-05,
      "loss": 4.937,
      "step": 11455
    },
    {
      "epoch": 0.4624949535728704,
      "grad_norm": 150.5137481689453,
      "learning_rate": 1.1692968819449926e-05,
      "loss": 4.3018,
      "step": 11456
    },
    {
      "epoch": 0.4625353249899071,
      "grad_norm": 88.3204116821289,
      "learning_rate": 1.1691680099379725e-05,
      "loss": 4.6136,
      "step": 11457
    },
    {
      "epoch": 0.4625756964069439,
      "grad_norm": 96.17295837402344,
      "learning_rate": 1.1690391350385747e-05,
      "loss": 3.1381,
      "step": 11458
    },
    {
      "epoch": 0.4626160678239806,
      "grad_norm": 89.98005676269531,
      "learning_rate": 1.1689102572490023e-05,
      "loss": 3.47,
      "step": 11459
    },
    {
      "epoch": 0.46265643924101735,
      "grad_norm": 92.49930572509766,
      "learning_rate": 1.168781376571459e-05,
      "loss": 5.8191,
      "step": 11460
    },
    {
      "epoch": 0.4626968106580541,
      "grad_norm": 85.61336517333984,
      "learning_rate": 1.1686524930081481e-05,
      "loss": 4.5028,
      "step": 11461
    },
    {
      "epoch": 0.46273718207509085,
      "grad_norm": 99.06327056884766,
      "learning_rate": 1.1685236065612737e-05,
      "loss": 3.7004,
      "step": 11462
    },
    {
      "epoch": 0.46277755349212757,
      "grad_norm": 78.97454833984375,
      "learning_rate": 1.168394717233039e-05,
      "loss": 5.0709,
      "step": 11463
    },
    {
      "epoch": 0.4628179249091643,
      "grad_norm": 130.03269958496094,
      "learning_rate": 1.1682658250256478e-05,
      "loss": 5.1763,
      "step": 11464
    },
    {
      "epoch": 0.46285829632620107,
      "grad_norm": 81.10748291015625,
      "learning_rate": 1.1681369299413041e-05,
      "loss": 4.2509,
      "step": 11465
    },
    {
      "epoch": 0.4628986677432378,
      "grad_norm": 105.22630310058594,
      "learning_rate": 1.1680080319822113e-05,
      "loss": 5.1893,
      "step": 11466
    },
    {
      "epoch": 0.4629390391602745,
      "grad_norm": 94.72172546386719,
      "learning_rate": 1.1678791311505737e-05,
      "loss": 5.3364,
      "step": 11467
    },
    {
      "epoch": 0.4629794105773113,
      "grad_norm": 120.79654693603516,
      "learning_rate": 1.167750227448595e-05,
      "loss": 5.1759,
      "step": 11468
    },
    {
      "epoch": 0.463019781994348,
      "grad_norm": 120.43478393554688,
      "learning_rate": 1.167621320878479e-05,
      "loss": 5.6666,
      "step": 11469
    },
    {
      "epoch": 0.46306015341138473,
      "grad_norm": 162.420166015625,
      "learning_rate": 1.16749241144243e-05,
      "loss": 6.0623,
      "step": 11470
    },
    {
      "epoch": 0.46310052482842146,
      "grad_norm": 106.22510528564453,
      "learning_rate": 1.1673634991426514e-05,
      "loss": 5.2249,
      "step": 11471
    },
    {
      "epoch": 0.46314089624545823,
      "grad_norm": 95.64143371582031,
      "learning_rate": 1.1672345839813484e-05,
      "loss": 6.6077,
      "step": 11472
    },
    {
      "epoch": 0.46318126766249496,
      "grad_norm": 103.9670639038086,
      "learning_rate": 1.1671056659607243e-05,
      "loss": 5.9806,
      "step": 11473
    },
    {
      "epoch": 0.4632216390795317,
      "grad_norm": 78.23538208007812,
      "learning_rate": 1.1669767450829835e-05,
      "loss": 4.7246,
      "step": 11474
    },
    {
      "epoch": 0.46326201049656845,
      "grad_norm": 128.17164611816406,
      "learning_rate": 1.1668478213503301e-05,
      "loss": 4.9548,
      "step": 11475
    },
    {
      "epoch": 0.4633023819136052,
      "grad_norm": 93.04759979248047,
      "learning_rate": 1.1667188947649688e-05,
      "loss": 6.2132,
      "step": 11476
    },
    {
      "epoch": 0.4633427533306419,
      "grad_norm": 85.98114776611328,
      "learning_rate": 1.1665899653291035e-05,
      "loss": 4.8779,
      "step": 11477
    },
    {
      "epoch": 0.4633831247476786,
      "grad_norm": 89.578125,
      "learning_rate": 1.166461033044939e-05,
      "loss": 5.8503,
      "step": 11478
    },
    {
      "epoch": 0.4634234961647154,
      "grad_norm": 83.79509735107422,
      "learning_rate": 1.1663320979146797e-05,
      "loss": 4.0745,
      "step": 11479
    },
    {
      "epoch": 0.4634638675817521,
      "grad_norm": 93.86844635009766,
      "learning_rate": 1.1662031599405297e-05,
      "loss": 5.4051,
      "step": 11480
    },
    {
      "epoch": 0.46350423899878884,
      "grad_norm": 89.09312438964844,
      "learning_rate": 1.166074219124694e-05,
      "loss": 4.9057,
      "step": 11481
    },
    {
      "epoch": 0.4635446104158256,
      "grad_norm": 114.13558959960938,
      "learning_rate": 1.1659452754693767e-05,
      "loss": 5.0665,
      "step": 11482
    },
    {
      "epoch": 0.46358498183286234,
      "grad_norm": 86.83413696289062,
      "learning_rate": 1.1658163289767828e-05,
      "loss": 5.6607,
      "step": 11483
    },
    {
      "epoch": 0.46362535324989906,
      "grad_norm": 87.852294921875,
      "learning_rate": 1.1656873796491169e-05,
      "loss": 5.7473,
      "step": 11484
    },
    {
      "epoch": 0.4636657246669358,
      "grad_norm": 104.32007598876953,
      "learning_rate": 1.1655584274885836e-05,
      "loss": 5.6121,
      "step": 11485
    },
    {
      "epoch": 0.46370609608397256,
      "grad_norm": 76.21558380126953,
      "learning_rate": 1.165429472497388e-05,
      "loss": 3.4092,
      "step": 11486
    },
    {
      "epoch": 0.4637464675010093,
      "grad_norm": 84.6275405883789,
      "learning_rate": 1.1653005146777346e-05,
      "loss": 4.5114,
      "step": 11487
    },
    {
      "epoch": 0.463786838918046,
      "grad_norm": 76.37190246582031,
      "learning_rate": 1.1651715540318285e-05,
      "loss": 6.5516,
      "step": 11488
    },
    {
      "epoch": 0.4638272103350828,
      "grad_norm": 74.15675354003906,
      "learning_rate": 1.1650425905618743e-05,
      "loss": 4.7539,
      "step": 11489
    },
    {
      "epoch": 0.4638675817521195,
      "grad_norm": 113.4619369506836,
      "learning_rate": 1.1649136242700776e-05,
      "loss": 3.7864,
      "step": 11490
    },
    {
      "epoch": 0.46390795316915623,
      "grad_norm": 156.19142150878906,
      "learning_rate": 1.1647846551586425e-05,
      "loss": 5.8999,
      "step": 11491
    },
    {
      "epoch": 0.46394832458619295,
      "grad_norm": 86.99636840820312,
      "learning_rate": 1.1646556832297751e-05,
      "loss": 4.8499,
      "step": 11492
    },
    {
      "epoch": 0.46398869600322973,
      "grad_norm": 86.63218688964844,
      "learning_rate": 1.1645267084856797e-05,
      "loss": 4.6757,
      "step": 11493
    },
    {
      "epoch": 0.46402906742026645,
      "grad_norm": 114.21134185791016,
      "learning_rate": 1.1643977309285618e-05,
      "loss": 4.458,
      "step": 11494
    },
    {
      "epoch": 0.46406943883730317,
      "grad_norm": 116.69654083251953,
      "learning_rate": 1.1642687505606267e-05,
      "loss": 4.0274,
      "step": 11495
    },
    {
      "epoch": 0.46410981025433995,
      "grad_norm": 94.93712615966797,
      "learning_rate": 1.1641397673840795e-05,
      "loss": 4.3587,
      "step": 11496
    },
    {
      "epoch": 0.46415018167137667,
      "grad_norm": 115.77201843261719,
      "learning_rate": 1.1640107814011257e-05,
      "loss": 5.043,
      "step": 11497
    },
    {
      "epoch": 0.4641905530884134,
      "grad_norm": 170.277099609375,
      "learning_rate": 1.1638817926139705e-05,
      "loss": 3.2016,
      "step": 11498
    },
    {
      "epoch": 0.4642309245054501,
      "grad_norm": 92.79261016845703,
      "learning_rate": 1.163752801024819e-05,
      "loss": 4.4464,
      "step": 11499
    },
    {
      "epoch": 0.4642712959224869,
      "grad_norm": 110.66097259521484,
      "learning_rate": 1.1636238066358776e-05,
      "loss": 5.0962,
      "step": 11500
    },
    {
      "epoch": 0.4643116673395236,
      "grad_norm": 111.34561920166016,
      "learning_rate": 1.1634948094493507e-05,
      "loss": 4.7563,
      "step": 11501
    },
    {
      "epoch": 0.46435203875656034,
      "grad_norm": 75.05680847167969,
      "learning_rate": 1.1633658094674447e-05,
      "loss": 4.9752,
      "step": 11502
    },
    {
      "epoch": 0.4643924101735971,
      "grad_norm": 127.9221420288086,
      "learning_rate": 1.1632368066923644e-05,
      "loss": 5.9402,
      "step": 11503
    },
    {
      "epoch": 0.46443278159063384,
      "grad_norm": 123.1265640258789,
      "learning_rate": 1.1631078011263162e-05,
      "loss": 4.8732,
      "step": 11504
    },
    {
      "epoch": 0.46447315300767056,
      "grad_norm": 136.43914794921875,
      "learning_rate": 1.1629787927715054e-05,
      "loss": 5.5786,
      "step": 11505
    },
    {
      "epoch": 0.4645135244247073,
      "grad_norm": 138.1741485595703,
      "learning_rate": 1.1628497816301381e-05,
      "loss": 3.349,
      "step": 11506
    },
    {
      "epoch": 0.46455389584174406,
      "grad_norm": 84.65928649902344,
      "learning_rate": 1.1627207677044194e-05,
      "loss": 3.7353,
      "step": 11507
    },
    {
      "epoch": 0.4645942672587808,
      "grad_norm": 87.62286376953125,
      "learning_rate": 1.1625917509965559e-05,
      "loss": 5.2445,
      "step": 11508
    },
    {
      "epoch": 0.4646346386758175,
      "grad_norm": 119.0675048828125,
      "learning_rate": 1.162462731508753e-05,
      "loss": 4.2682,
      "step": 11509
    },
    {
      "epoch": 0.4646750100928543,
      "grad_norm": 101.35099029541016,
      "learning_rate": 1.1623337092432168e-05,
      "loss": 4.5401,
      "step": 11510
    },
    {
      "epoch": 0.464715381509891,
      "grad_norm": 99.00857543945312,
      "learning_rate": 1.162204684202153e-05,
      "loss": 4.8501,
      "step": 11511
    },
    {
      "epoch": 0.4647557529269277,
      "grad_norm": 629.9584350585938,
      "learning_rate": 1.1620756563877682e-05,
      "loss": 6.2448,
      "step": 11512
    },
    {
      "epoch": 0.4647961243439645,
      "grad_norm": 102.05045318603516,
      "learning_rate": 1.161946625802268e-05,
      "loss": 6.1527,
      "step": 11513
    },
    {
      "epoch": 0.4648364957610012,
      "grad_norm": 111.03556060791016,
      "learning_rate": 1.1618175924478586e-05,
      "loss": 5.263,
      "step": 11514
    },
    {
      "epoch": 0.46487686717803794,
      "grad_norm": 118.84980010986328,
      "learning_rate": 1.1616885563267462e-05,
      "loss": 4.1446,
      "step": 11515
    },
    {
      "epoch": 0.46491723859507467,
      "grad_norm": 76.50511169433594,
      "learning_rate": 1.1615595174411374e-05,
      "loss": 4.3502,
      "step": 11516
    },
    {
      "epoch": 0.46495761001211144,
      "grad_norm": 122.78628540039062,
      "learning_rate": 1.1614304757932377e-05,
      "loss": 5.5854,
      "step": 11517
    },
    {
      "epoch": 0.46499798142914817,
      "grad_norm": 94.8455581665039,
      "learning_rate": 1.1613014313852541e-05,
      "loss": 5.9486,
      "step": 11518
    },
    {
      "epoch": 0.4650383528461849,
      "grad_norm": 71.16871643066406,
      "learning_rate": 1.1611723842193925e-05,
      "loss": 5.3397,
      "step": 11519
    },
    {
      "epoch": 0.46507872426322167,
      "grad_norm": 74.1109848022461,
      "learning_rate": 1.1610433342978597e-05,
      "loss": 4.7568,
      "step": 11520
    },
    {
      "epoch": 0.4651190956802584,
      "grad_norm": 101.39307403564453,
      "learning_rate": 1.160914281622862e-05,
      "loss": 6.1835,
      "step": 11521
    },
    {
      "epoch": 0.4651594670972951,
      "grad_norm": 623.8297119140625,
      "learning_rate": 1.1607852261966057e-05,
      "loss": 3.9079,
      "step": 11522
    },
    {
      "epoch": 0.46519983851433183,
      "grad_norm": 105.16464233398438,
      "learning_rate": 1.1606561680212976e-05,
      "loss": 5.2748,
      "step": 11523
    },
    {
      "epoch": 0.4652402099313686,
      "grad_norm": 148.25765991210938,
      "learning_rate": 1.160527107099144e-05,
      "loss": 3.8838,
      "step": 11524
    },
    {
      "epoch": 0.46528058134840533,
      "grad_norm": 122.23928833007812,
      "learning_rate": 1.1603980434323516e-05,
      "loss": 5.4308,
      "step": 11525
    },
    {
      "epoch": 0.46532095276544205,
      "grad_norm": 107.28919219970703,
      "learning_rate": 1.1602689770231275e-05,
      "loss": 5.8174,
      "step": 11526
    },
    {
      "epoch": 0.46536132418247883,
      "grad_norm": 1176.5322265625,
      "learning_rate": 1.1601399078736782e-05,
      "loss": 5.7039,
      "step": 11527
    },
    {
      "epoch": 0.46540169559951555,
      "grad_norm": 81.15699005126953,
      "learning_rate": 1.1600108359862103e-05,
      "loss": 5.2411,
      "step": 11528
    },
    {
      "epoch": 0.4654420670165523,
      "grad_norm": 125.25972747802734,
      "learning_rate": 1.1598817613629305e-05,
      "loss": 4.9009,
      "step": 11529
    },
    {
      "epoch": 0.465482438433589,
      "grad_norm": 101.84017944335938,
      "learning_rate": 1.1597526840060465e-05,
      "loss": 6.0498,
      "step": 11530
    },
    {
      "epoch": 0.4655228098506258,
      "grad_norm": 80.12164306640625,
      "learning_rate": 1.1596236039177641e-05,
      "loss": 3.6967,
      "step": 11531
    },
    {
      "epoch": 0.4655631812676625,
      "grad_norm": 130.94970703125,
      "learning_rate": 1.1594945211002913e-05,
      "loss": 4.1953,
      "step": 11532
    },
    {
      "epoch": 0.4656035526846992,
      "grad_norm": 119.60303497314453,
      "learning_rate": 1.1593654355558344e-05,
      "loss": 6.1585,
      "step": 11533
    },
    {
      "epoch": 0.465643924101736,
      "grad_norm": 96.33171081542969,
      "learning_rate": 1.1592363472866005e-05,
      "loss": 6.0306,
      "step": 11534
    },
    {
      "epoch": 0.4656842955187727,
      "grad_norm": 97.01878356933594,
      "learning_rate": 1.1591072562947971e-05,
      "loss": 5.3124,
      "step": 11535
    },
    {
      "epoch": 0.46572466693580944,
      "grad_norm": 99.8893051147461,
      "learning_rate": 1.1589781625826312e-05,
      "loss": 6.9681,
      "step": 11536
    },
    {
      "epoch": 0.46576503835284616,
      "grad_norm": 105.37284851074219,
      "learning_rate": 1.15884906615231e-05,
      "loss": 4.9995,
      "step": 11537
    },
    {
      "epoch": 0.46580540976988294,
      "grad_norm": 110.86234283447266,
      "learning_rate": 1.1587199670060407e-05,
      "loss": 5.1046,
      "step": 11538
    },
    {
      "epoch": 0.46584578118691966,
      "grad_norm": 82.51067352294922,
      "learning_rate": 1.1585908651460304e-05,
      "loss": 5.524,
      "step": 11539
    },
    {
      "epoch": 0.4658861526039564,
      "grad_norm": 101.92671966552734,
      "learning_rate": 1.1584617605744871e-05,
      "loss": 5.597,
      "step": 11540
    },
    {
      "epoch": 0.46592652402099316,
      "grad_norm": 92.67342376708984,
      "learning_rate": 1.1583326532936172e-05,
      "loss": 5.5899,
      "step": 11541
    },
    {
      "epoch": 0.4659668954380299,
      "grad_norm": 192.4610595703125,
      "learning_rate": 1.1582035433056288e-05,
      "loss": 4.9447,
      "step": 11542
    },
    {
      "epoch": 0.4660072668550666,
      "grad_norm": 116.21908569335938,
      "learning_rate": 1.1580744306127293e-05,
      "loss": 4.8909,
      "step": 11543
    },
    {
      "epoch": 0.4660476382721033,
      "grad_norm": 114.27201843261719,
      "learning_rate": 1.1579453152171261e-05,
      "loss": 5.4682,
      "step": 11544
    },
    {
      "epoch": 0.4660880096891401,
      "grad_norm": 91.12483978271484,
      "learning_rate": 1.1578161971210268e-05,
      "loss": 5.3438,
      "step": 11545
    },
    {
      "epoch": 0.4661283811061768,
      "grad_norm": 91.02318572998047,
      "learning_rate": 1.1576870763266392e-05,
      "loss": 6.1967,
      "step": 11546
    },
    {
      "epoch": 0.46616875252321355,
      "grad_norm": 130.29502868652344,
      "learning_rate": 1.157557952836171e-05,
      "loss": 5.9577,
      "step": 11547
    },
    {
      "epoch": 0.4662091239402503,
      "grad_norm": 82.96202087402344,
      "learning_rate": 1.1574288266518294e-05,
      "loss": 5.7391,
      "step": 11548
    },
    {
      "epoch": 0.46624949535728705,
      "grad_norm": 82.40866088867188,
      "learning_rate": 1.1572996977758227e-05,
      "loss": 5.2692,
      "step": 11549
    },
    {
      "epoch": 0.46628986677432377,
      "grad_norm": 106.34432220458984,
      "learning_rate": 1.1571705662103585e-05,
      "loss": 4.9953,
      "step": 11550
    },
    {
      "epoch": 0.4663302381913605,
      "grad_norm": 96.58560943603516,
      "learning_rate": 1.1570414319576444e-05,
      "loss": 5.6976,
      "step": 11551
    },
    {
      "epoch": 0.46637060960839727,
      "grad_norm": 116.37220001220703,
      "learning_rate": 1.1569122950198888e-05,
      "loss": 4.7495,
      "step": 11552
    },
    {
      "epoch": 0.466410981025434,
      "grad_norm": 102.13619995117188,
      "learning_rate": 1.1567831553992991e-05,
      "loss": 4.7977,
      "step": 11553
    },
    {
      "epoch": 0.4664513524424707,
      "grad_norm": 75.84406280517578,
      "learning_rate": 1.1566540130980836e-05,
      "loss": 4.8358,
      "step": 11554
    },
    {
      "epoch": 0.4664917238595075,
      "grad_norm": 102.09136962890625,
      "learning_rate": 1.1565248681184503e-05,
      "loss": 5.0906,
      "step": 11555
    },
    {
      "epoch": 0.4665320952765442,
      "grad_norm": 118.73977661132812,
      "learning_rate": 1.1563957204626076e-05,
      "loss": 4.7906,
      "step": 11556
    },
    {
      "epoch": 0.46657246669358093,
      "grad_norm": 101.0272216796875,
      "learning_rate": 1.1562665701327628e-05,
      "loss": 4.0099,
      "step": 11557
    },
    {
      "epoch": 0.46661283811061766,
      "grad_norm": 68.21755981445312,
      "learning_rate": 1.1561374171311251e-05,
      "loss": 5.7243,
      "step": 11558
    },
    {
      "epoch": 0.46665320952765443,
      "grad_norm": 94.50582122802734,
      "learning_rate": 1.1560082614599018e-05,
      "loss": 6.418,
      "step": 11559
    },
    {
      "epoch": 0.46669358094469116,
      "grad_norm": 76.44677734375,
      "learning_rate": 1.1558791031213017e-05,
      "loss": 3.8965,
      "step": 11560
    },
    {
      "epoch": 0.4667339523617279,
      "grad_norm": 129.5094451904297,
      "learning_rate": 1.1557499421175327e-05,
      "loss": 4.4612,
      "step": 11561
    },
    {
      "epoch": 0.46677432377876465,
      "grad_norm": 104.54100036621094,
      "learning_rate": 1.1556207784508038e-05,
      "loss": 3.8804,
      "step": 11562
    },
    {
      "epoch": 0.4668146951958014,
      "grad_norm": 105.29960632324219,
      "learning_rate": 1.1554916121233224e-05,
      "loss": 5.5805,
      "step": 11563
    },
    {
      "epoch": 0.4668550666128381,
      "grad_norm": 62.34330749511719,
      "learning_rate": 1.1553624431372978e-05,
      "loss": 3.9919,
      "step": 11564
    },
    {
      "epoch": 0.4668954380298749,
      "grad_norm": 86.7802963256836,
      "learning_rate": 1.1552332714949384e-05,
      "loss": 4.7883,
      "step": 11565
    },
    {
      "epoch": 0.4669358094469116,
      "grad_norm": 103.35282897949219,
      "learning_rate": 1.1551040971984525e-05,
      "loss": 4.8237,
      "step": 11566
    },
    {
      "epoch": 0.4669761808639483,
      "grad_norm": 118.09978485107422,
      "learning_rate": 1.1549749202500485e-05,
      "loss": 5.8904,
      "step": 11567
    },
    {
      "epoch": 0.46701655228098504,
      "grad_norm": 123.22203063964844,
      "learning_rate": 1.1548457406519355e-05,
      "loss": 5.0766,
      "step": 11568
    },
    {
      "epoch": 0.4670569236980218,
      "grad_norm": 113.79659271240234,
      "learning_rate": 1.1547165584063218e-05,
      "loss": 3.7584,
      "step": 11569
    },
    {
      "epoch": 0.46709729511505854,
      "grad_norm": 153.57940673828125,
      "learning_rate": 1.1545873735154164e-05,
      "loss": 5.1938,
      "step": 11570
    },
    {
      "epoch": 0.46713766653209526,
      "grad_norm": 76.04541015625,
      "learning_rate": 1.1544581859814277e-05,
      "loss": 5.4655,
      "step": 11571
    },
    {
      "epoch": 0.46717803794913204,
      "grad_norm": 140.92446899414062,
      "learning_rate": 1.154328995806565e-05,
      "loss": 5.2659,
      "step": 11572
    },
    {
      "epoch": 0.46721840936616876,
      "grad_norm": 190.91754150390625,
      "learning_rate": 1.1541998029930363e-05,
      "loss": 5.7239,
      "step": 11573
    },
    {
      "epoch": 0.4672587807832055,
      "grad_norm": 96.83007049560547,
      "learning_rate": 1.1540706075430516e-05,
      "loss": 3.957,
      "step": 11574
    },
    {
      "epoch": 0.4672991522002422,
      "grad_norm": 119.46363830566406,
      "learning_rate": 1.1539414094588191e-05,
      "loss": 6.9406,
      "step": 11575
    },
    {
      "epoch": 0.467339523617279,
      "grad_norm": 86.12700653076172,
      "learning_rate": 1.1538122087425481e-05,
      "loss": 5.3337,
      "step": 11576
    },
    {
      "epoch": 0.4673798950343157,
      "grad_norm": 71.84662628173828,
      "learning_rate": 1.1536830053964472e-05,
      "loss": 4.4376,
      "step": 11577
    },
    {
      "epoch": 0.46742026645135243,
      "grad_norm": 173.41159057617188,
      "learning_rate": 1.1535537994227261e-05,
      "loss": 5.9293,
      "step": 11578
    },
    {
      "epoch": 0.4674606378683892,
      "grad_norm": 192.03504943847656,
      "learning_rate": 1.1534245908235935e-05,
      "loss": 4.6475,
      "step": 11579
    },
    {
      "epoch": 0.46750100928542593,
      "grad_norm": 88.38278198242188,
      "learning_rate": 1.1532953796012588e-05,
      "loss": 5.5462,
      "step": 11580
    },
    {
      "epoch": 0.46754138070246265,
      "grad_norm": 87.24532318115234,
      "learning_rate": 1.1531661657579307e-05,
      "loss": 5.4194,
      "step": 11581
    },
    {
      "epoch": 0.46758175211949937,
      "grad_norm": 79.16551971435547,
      "learning_rate": 1.1530369492958189e-05,
      "loss": 4.6052,
      "step": 11582
    },
    {
      "epoch": 0.46762212353653615,
      "grad_norm": 87.57105255126953,
      "learning_rate": 1.1529077302171328e-05,
      "loss": 5.4972,
      "step": 11583
    },
    {
      "epoch": 0.46766249495357287,
      "grad_norm": 126.31663513183594,
      "learning_rate": 1.1527785085240817e-05,
      "loss": 4.1455,
      "step": 11584
    },
    {
      "epoch": 0.4677028663706096,
      "grad_norm": 66.42524719238281,
      "learning_rate": 1.1526492842188746e-05,
      "loss": 5.4001,
      "step": 11585
    },
    {
      "epoch": 0.46774323778764637,
      "grad_norm": 145.28475952148438,
      "learning_rate": 1.1525200573037214e-05,
      "loss": 4.6865,
      "step": 11586
    },
    {
      "epoch": 0.4677836092046831,
      "grad_norm": 81.74526977539062,
      "learning_rate": 1.1523908277808313e-05,
      "loss": 4.8247,
      "step": 11587
    },
    {
      "epoch": 0.4678239806217198,
      "grad_norm": 75.08511352539062,
      "learning_rate": 1.152261595652414e-05,
      "loss": 4.9544,
      "step": 11588
    },
    {
      "epoch": 0.46786435203875654,
      "grad_norm": 109.15904235839844,
      "learning_rate": 1.1521323609206788e-05,
      "loss": 4.7697,
      "step": 11589
    },
    {
      "epoch": 0.4679047234557933,
      "grad_norm": 92.30743408203125,
      "learning_rate": 1.1520031235878357e-05,
      "loss": 6.2368,
      "step": 11590
    },
    {
      "epoch": 0.46794509487283004,
      "grad_norm": 96.83466339111328,
      "learning_rate": 1.151873883656094e-05,
      "loss": 3.6502,
      "step": 11591
    },
    {
      "epoch": 0.46798546628986676,
      "grad_norm": 102.1910171508789,
      "learning_rate": 1.1517446411276634e-05,
      "loss": 5.5236,
      "step": 11592
    },
    {
      "epoch": 0.46802583770690354,
      "grad_norm": 80.60047149658203,
      "learning_rate": 1.1516153960047539e-05,
      "loss": 5.3715,
      "step": 11593
    },
    {
      "epoch": 0.46806620912394026,
      "grad_norm": 121.95807647705078,
      "learning_rate": 1.1514861482895752e-05,
      "loss": 5.1242,
      "step": 11594
    },
    {
      "epoch": 0.468106580540977,
      "grad_norm": 72.71370697021484,
      "learning_rate": 1.151356897984337e-05,
      "loss": 4.676,
      "step": 11595
    },
    {
      "epoch": 0.4681469519580137,
      "grad_norm": 103.6689682006836,
      "learning_rate": 1.1512276450912497e-05,
      "loss": 4.8032,
      "step": 11596
    },
    {
      "epoch": 0.4681873233750505,
      "grad_norm": 125.73866271972656,
      "learning_rate": 1.1510983896125222e-05,
      "loss": 4.2811,
      "step": 11597
    },
    {
      "epoch": 0.4682276947920872,
      "grad_norm": 110.34103393554688,
      "learning_rate": 1.1509691315503657e-05,
      "loss": 5.2863,
      "step": 11598
    },
    {
      "epoch": 0.4682680662091239,
      "grad_norm": 95.30658721923828,
      "learning_rate": 1.1508398709069892e-05,
      "loss": 5.0613,
      "step": 11599
    },
    {
      "epoch": 0.4683084376261607,
      "grad_norm": 114.0606689453125,
      "learning_rate": 1.1507106076846033e-05,
      "loss": 4.4689,
      "step": 11600
    },
    {
      "epoch": 0.4683488090431974,
      "grad_norm": 111.19422912597656,
      "learning_rate": 1.1505813418854179e-05,
      "loss": 7.4944,
      "step": 11601
    },
    {
      "epoch": 0.46838918046023414,
      "grad_norm": 146.78216552734375,
      "learning_rate": 1.1504520735116432e-05,
      "loss": 5.9972,
      "step": 11602
    },
    {
      "epoch": 0.46842955187727087,
      "grad_norm": 93.24690246582031,
      "learning_rate": 1.1503228025654892e-05,
      "loss": 5.4727,
      "step": 11603
    },
    {
      "epoch": 0.46846992329430764,
      "grad_norm": 107.69126892089844,
      "learning_rate": 1.1501935290491664e-05,
      "loss": 5.1071,
      "step": 11604
    },
    {
      "epoch": 0.46851029471134437,
      "grad_norm": 119.60668182373047,
      "learning_rate": 1.150064252964885e-05,
      "loss": 5.1709,
      "step": 11605
    },
    {
      "epoch": 0.4685506661283811,
      "grad_norm": 99.11895751953125,
      "learning_rate": 1.1499349743148553e-05,
      "loss": 4.7118,
      "step": 11606
    },
    {
      "epoch": 0.46859103754541787,
      "grad_norm": 88.94620513916016,
      "learning_rate": 1.1498056931012878e-05,
      "loss": 5.4791,
      "step": 11607
    },
    {
      "epoch": 0.4686314089624546,
      "grad_norm": 85.37638092041016,
      "learning_rate": 1.1496764093263928e-05,
      "loss": 3.1012,
      "step": 11608
    },
    {
      "epoch": 0.4686717803794913,
      "grad_norm": 74.63953399658203,
      "learning_rate": 1.1495471229923805e-05,
      "loss": 4.7624,
      "step": 11609
    },
    {
      "epoch": 0.46871215179652803,
      "grad_norm": 101.90998840332031,
      "learning_rate": 1.1494178341014616e-05,
      "loss": 4.4341,
      "step": 11610
    },
    {
      "epoch": 0.4687525232135648,
      "grad_norm": 105.89138793945312,
      "learning_rate": 1.1492885426558467e-05,
      "loss": 5.1702,
      "step": 11611
    },
    {
      "epoch": 0.46879289463060153,
      "grad_norm": 132.1383514404297,
      "learning_rate": 1.1491592486577463e-05,
      "loss": 4.8471,
      "step": 11612
    },
    {
      "epoch": 0.46883326604763825,
      "grad_norm": 152.12210083007812,
      "learning_rate": 1.149029952109371e-05,
      "loss": 4.1909,
      "step": 11613
    },
    {
      "epoch": 0.46887363746467503,
      "grad_norm": 75.12249755859375,
      "learning_rate": 1.1489006530129318e-05,
      "loss": 3.9814,
      "step": 11614
    },
    {
      "epoch": 0.46891400888171175,
      "grad_norm": 136.9741973876953,
      "learning_rate": 1.148771351370639e-05,
      "loss": 5.2428,
      "step": 11615
    },
    {
      "epoch": 0.4689543802987485,
      "grad_norm": 116.93729400634766,
      "learning_rate": 1.1486420471847036e-05,
      "loss": 4.8163,
      "step": 11616
    },
    {
      "epoch": 0.4689947517157852,
      "grad_norm": 135.17271423339844,
      "learning_rate": 1.1485127404573361e-05,
      "loss": 5.3747,
      "step": 11617
    },
    {
      "epoch": 0.469035123132822,
      "grad_norm": 88.6303939819336,
      "learning_rate": 1.1483834311907479e-05,
      "loss": 4.7205,
      "step": 11618
    },
    {
      "epoch": 0.4690754945498587,
      "grad_norm": 83.64415740966797,
      "learning_rate": 1.148254119387149e-05,
      "loss": 5.1112,
      "step": 11619
    },
    {
      "epoch": 0.4691158659668954,
      "grad_norm": 95.05783081054688,
      "learning_rate": 1.1481248050487513e-05,
      "loss": 4.8362,
      "step": 11620
    },
    {
      "epoch": 0.4691562373839322,
      "grad_norm": 142.5929718017578,
      "learning_rate": 1.147995488177765e-05,
      "loss": 5.1821,
      "step": 11621
    },
    {
      "epoch": 0.4691966088009689,
      "grad_norm": 309.507568359375,
      "learning_rate": 1.1478661687764015e-05,
      "loss": 5.4622,
      "step": 11622
    },
    {
      "epoch": 0.46923698021800564,
      "grad_norm": 86.20367431640625,
      "learning_rate": 1.147736846846872e-05,
      "loss": 3.7595,
      "step": 11623
    },
    {
      "epoch": 0.4692773516350424,
      "grad_norm": 106.98785400390625,
      "learning_rate": 1.1476075223913875e-05,
      "loss": 6.2954,
      "step": 11624
    },
    {
      "epoch": 0.46931772305207914,
      "grad_norm": 87.8243179321289,
      "learning_rate": 1.1474781954121585e-05,
      "loss": 4.6383,
      "step": 11625
    },
    {
      "epoch": 0.46935809446911586,
      "grad_norm": 107.11729431152344,
      "learning_rate": 1.1473488659113973e-05,
      "loss": 5.1917,
      "step": 11626
    },
    {
      "epoch": 0.4693984658861526,
      "grad_norm": 83.97779846191406,
      "learning_rate": 1.1472195338913144e-05,
      "loss": 3.6326,
      "step": 11627
    },
    {
      "epoch": 0.46943883730318936,
      "grad_norm": 90.65746307373047,
      "learning_rate": 1.1470901993541212e-05,
      "loss": 4.1187,
      "step": 11628
    },
    {
      "epoch": 0.4694792087202261,
      "grad_norm": 234.4463348388672,
      "learning_rate": 1.1469608623020291e-05,
      "loss": 4.9913,
      "step": 11629
    },
    {
      "epoch": 0.4695195801372628,
      "grad_norm": 140.7861785888672,
      "learning_rate": 1.1468315227372497e-05,
      "loss": 5.2807,
      "step": 11630
    },
    {
      "epoch": 0.4695599515542996,
      "grad_norm": 70.62196350097656,
      "learning_rate": 1.1467021806619938e-05,
      "loss": 4.7375,
      "step": 11631
    },
    {
      "epoch": 0.4696003229713363,
      "grad_norm": 104.24024963378906,
      "learning_rate": 1.1465728360784732e-05,
      "loss": 3.7543,
      "step": 11632
    },
    {
      "epoch": 0.469640694388373,
      "grad_norm": 82.57798767089844,
      "learning_rate": 1.1464434889888994e-05,
      "loss": 4.3116,
      "step": 11633
    },
    {
      "epoch": 0.46968106580540975,
      "grad_norm": 76.12390899658203,
      "learning_rate": 1.146314139395484e-05,
      "loss": 5.8384,
      "step": 11634
    },
    {
      "epoch": 0.4697214372224465,
      "grad_norm": 102.4637451171875,
      "learning_rate": 1.1461847873004383e-05,
      "loss": 5.3018,
      "step": 11635
    },
    {
      "epoch": 0.46976180863948325,
      "grad_norm": 106.5948257446289,
      "learning_rate": 1.1460554327059744e-05,
      "loss": 5.3241,
      "step": 11636
    },
    {
      "epoch": 0.46980218005651997,
      "grad_norm": 122.33358764648438,
      "learning_rate": 1.1459260756143033e-05,
      "loss": 4.6013,
      "step": 11637
    },
    {
      "epoch": 0.46984255147355675,
      "grad_norm": 130.84808349609375,
      "learning_rate": 1.1457967160276375e-05,
      "loss": 3.6865,
      "step": 11638
    },
    {
      "epoch": 0.46988292289059347,
      "grad_norm": 94.4047622680664,
      "learning_rate": 1.145667353948188e-05,
      "loss": 5.4613,
      "step": 11639
    },
    {
      "epoch": 0.4699232943076302,
      "grad_norm": 109.20816802978516,
      "learning_rate": 1.1455379893781671e-05,
      "loss": 7.0009,
      "step": 11640
    },
    {
      "epoch": 0.4699636657246669,
      "grad_norm": 100.1939468383789,
      "learning_rate": 1.1454086223197862e-05,
      "loss": 3.0742,
      "step": 11641
    },
    {
      "epoch": 0.4700040371417037,
      "grad_norm": 84.0648193359375,
      "learning_rate": 1.1452792527752576e-05,
      "loss": 6.2932,
      "step": 11642
    },
    {
      "epoch": 0.4700444085587404,
      "grad_norm": 159.2592010498047,
      "learning_rate": 1.145149880746793e-05,
      "loss": 3.3758,
      "step": 11643
    },
    {
      "epoch": 0.47008477997577713,
      "grad_norm": 91.60907745361328,
      "learning_rate": 1.1450205062366043e-05,
      "loss": 5.7653,
      "step": 11644
    },
    {
      "epoch": 0.4701251513928139,
      "grad_norm": 144.43052673339844,
      "learning_rate": 1.1448911292469037e-05,
      "loss": 5.8792,
      "step": 11645
    },
    {
      "epoch": 0.47016552280985063,
      "grad_norm": 105.02981567382812,
      "learning_rate": 1.144761749779903e-05,
      "loss": 4.3669,
      "step": 11646
    },
    {
      "epoch": 0.47020589422688736,
      "grad_norm": 312.0472106933594,
      "learning_rate": 1.1446323678378146e-05,
      "loss": 4.1898,
      "step": 11647
    },
    {
      "epoch": 0.4702462656439241,
      "grad_norm": 68.4642562866211,
      "learning_rate": 1.1445029834228505e-05,
      "loss": 5.3994,
      "step": 11648
    },
    {
      "epoch": 0.47028663706096085,
      "grad_norm": 125.07926940917969,
      "learning_rate": 1.1443735965372225e-05,
      "loss": 4.5562,
      "step": 11649
    },
    {
      "epoch": 0.4703270084779976,
      "grad_norm": 85.21600341796875,
      "learning_rate": 1.1442442071831434e-05,
      "loss": 4.8329,
      "step": 11650
    },
    {
      "epoch": 0.4703673798950343,
      "grad_norm": 104.38280487060547,
      "learning_rate": 1.144114815362825e-05,
      "loss": 3.1478,
      "step": 11651
    },
    {
      "epoch": 0.4704077513120711,
      "grad_norm": 208.01742553710938,
      "learning_rate": 1.14398542107848e-05,
      "loss": 3.2153,
      "step": 11652
    },
    {
      "epoch": 0.4704481227291078,
      "grad_norm": 82.6134262084961,
      "learning_rate": 1.1438560243323202e-05,
      "loss": 3.282,
      "step": 11653
    },
    {
      "epoch": 0.4704884941461445,
      "grad_norm": 252.42788696289062,
      "learning_rate": 1.1437266251265587e-05,
      "loss": 3.7096,
      "step": 11654
    },
    {
      "epoch": 0.47052886556318124,
      "grad_norm": 98.8041763305664,
      "learning_rate": 1.1435972234634071e-05,
      "loss": 4.0994,
      "step": 11655
    },
    {
      "epoch": 0.470569236980218,
      "grad_norm": 78.09784698486328,
      "learning_rate": 1.1434678193450788e-05,
      "loss": 5.2229,
      "step": 11656
    },
    {
      "epoch": 0.47060960839725474,
      "grad_norm": 85.65550231933594,
      "learning_rate": 1.1433384127737857e-05,
      "loss": 4.6113,
      "step": 11657
    },
    {
      "epoch": 0.47064997981429146,
      "grad_norm": 136.7467803955078,
      "learning_rate": 1.1432090037517405e-05,
      "loss": 6.478,
      "step": 11658
    },
    {
      "epoch": 0.47069035123132824,
      "grad_norm": 144.38836669921875,
      "learning_rate": 1.1430795922811554e-05,
      "loss": 4.212,
      "step": 11659
    },
    {
      "epoch": 0.47073072264836496,
      "grad_norm": 80.82676696777344,
      "learning_rate": 1.1429501783642439e-05,
      "loss": 6.0945,
      "step": 11660
    },
    {
      "epoch": 0.4707710940654017,
      "grad_norm": 100.89991760253906,
      "learning_rate": 1.1428207620032177e-05,
      "loss": 3.3959,
      "step": 11661
    },
    {
      "epoch": 0.4708114654824384,
      "grad_norm": 87.71073150634766,
      "learning_rate": 1.14269134320029e-05,
      "loss": 6.2589,
      "step": 11662
    },
    {
      "epoch": 0.4708518368994752,
      "grad_norm": 238.3685760498047,
      "learning_rate": 1.1425619219576735e-05,
      "loss": 5.9462,
      "step": 11663
    },
    {
      "epoch": 0.4708922083165119,
      "grad_norm": 89.69088745117188,
      "learning_rate": 1.1424324982775814e-05,
      "loss": 5.6302,
      "step": 11664
    },
    {
      "epoch": 0.47093257973354863,
      "grad_norm": 137.1733856201172,
      "learning_rate": 1.142303072162226e-05,
      "loss": 6.8026,
      "step": 11665
    },
    {
      "epoch": 0.4709729511505854,
      "grad_norm": 125.36875915527344,
      "learning_rate": 1.1421736436138204e-05,
      "loss": 5.3301,
      "step": 11666
    },
    {
      "epoch": 0.47101332256762213,
      "grad_norm": 89.1881103515625,
      "learning_rate": 1.1420442126345774e-05,
      "loss": 4.6735,
      "step": 11667
    },
    {
      "epoch": 0.47105369398465885,
      "grad_norm": 133.93345642089844,
      "learning_rate": 1.1419147792267103e-05,
      "loss": 6.2147,
      "step": 11668
    },
    {
      "epoch": 0.47109406540169557,
      "grad_norm": 106.48384857177734,
      "learning_rate": 1.1417853433924316e-05,
      "loss": 6.0651,
      "step": 11669
    },
    {
      "epoch": 0.47113443681873235,
      "grad_norm": 133.3766326904297,
      "learning_rate": 1.1416559051339546e-05,
      "loss": 4.682,
      "step": 11670
    },
    {
      "epoch": 0.47117480823576907,
      "grad_norm": 139.463623046875,
      "learning_rate": 1.1415264644534924e-05,
      "loss": 5.7043,
      "step": 11671
    },
    {
      "epoch": 0.4712151796528058,
      "grad_norm": 815.4144897460938,
      "learning_rate": 1.1413970213532584e-05,
      "loss": 4.3512,
      "step": 11672
    },
    {
      "epoch": 0.47125555106984257,
      "grad_norm": 113.03472137451172,
      "learning_rate": 1.1412675758354652e-05,
      "loss": 6.705,
      "step": 11673
    },
    {
      "epoch": 0.4712959224868793,
      "grad_norm": 103.41790771484375,
      "learning_rate": 1.1411381279023268e-05,
      "loss": 5.1729,
      "step": 11674
    },
    {
      "epoch": 0.471336293903916,
      "grad_norm": 75.08663177490234,
      "learning_rate": 1.1410086775560556e-05,
      "loss": 6.3935,
      "step": 11675
    },
    {
      "epoch": 0.4713766653209528,
      "grad_norm": 98.3322525024414,
      "learning_rate": 1.1408792247988653e-05,
      "loss": 5.0933,
      "step": 11676
    },
    {
      "epoch": 0.4714170367379895,
      "grad_norm": 111.550537109375,
      "learning_rate": 1.1407497696329695e-05,
      "loss": 6.2245,
      "step": 11677
    },
    {
      "epoch": 0.47145740815502624,
      "grad_norm": 78.53450775146484,
      "learning_rate": 1.1406203120605812e-05,
      "loss": 5.122,
      "step": 11678
    },
    {
      "epoch": 0.47149777957206296,
      "grad_norm": 73.3622055053711,
      "learning_rate": 1.1404908520839138e-05,
      "loss": 4.0552,
      "step": 11679
    },
    {
      "epoch": 0.47153815098909974,
      "grad_norm": 85.67964935302734,
      "learning_rate": 1.1403613897051809e-05,
      "loss": 5.8861,
      "step": 11680
    },
    {
      "epoch": 0.47157852240613646,
      "grad_norm": 105.0510025024414,
      "learning_rate": 1.1402319249265961e-05,
      "loss": 3.9688,
      "step": 11681
    },
    {
      "epoch": 0.4716188938231732,
      "grad_norm": 100.65995025634766,
      "learning_rate": 1.140102457750373e-05,
      "loss": 5.5948,
      "step": 11682
    },
    {
      "epoch": 0.47165926524020996,
      "grad_norm": 71.16455078125,
      "learning_rate": 1.1399729881787247e-05,
      "loss": 5.5579,
      "step": 11683
    },
    {
      "epoch": 0.4716996366572467,
      "grad_norm": 78.97383880615234,
      "learning_rate": 1.1398435162138655e-05,
      "loss": 5.7864,
      "step": 11684
    },
    {
      "epoch": 0.4717400080742834,
      "grad_norm": 146.27464294433594,
      "learning_rate": 1.1397140418580083e-05,
      "loss": 4.2061,
      "step": 11685
    },
    {
      "epoch": 0.4717803794913201,
      "grad_norm": 96.63449096679688,
      "learning_rate": 1.139584565113368e-05,
      "loss": 4.0651,
      "step": 11686
    },
    {
      "epoch": 0.4718207509083569,
      "grad_norm": 438.4908142089844,
      "learning_rate": 1.1394550859821568e-05,
      "loss": 7.2613,
      "step": 11687
    },
    {
      "epoch": 0.4718611223253936,
      "grad_norm": 112.7941665649414,
      "learning_rate": 1.1393256044665897e-05,
      "loss": 5.753,
      "step": 11688
    },
    {
      "epoch": 0.47190149374243034,
      "grad_norm": 75.4794921875,
      "learning_rate": 1.1391961205688799e-05,
      "loss": 4.9899,
      "step": 11689
    },
    {
      "epoch": 0.4719418651594671,
      "grad_norm": 100.5219497680664,
      "learning_rate": 1.1390666342912415e-05,
      "loss": 4.6576,
      "step": 11690
    },
    {
      "epoch": 0.47198223657650384,
      "grad_norm": 96.46929168701172,
      "learning_rate": 1.1389371456358884e-05,
      "loss": 4.8355,
      "step": 11691
    },
    {
      "epoch": 0.47202260799354057,
      "grad_norm": 74.57300567626953,
      "learning_rate": 1.1388076546050348e-05,
      "loss": 3.9156,
      "step": 11692
    },
    {
      "epoch": 0.4720629794105773,
      "grad_norm": 1005.3692626953125,
      "learning_rate": 1.1386781612008942e-05,
      "loss": 5.0101,
      "step": 11693
    },
    {
      "epoch": 0.47210335082761407,
      "grad_norm": 85.6950454711914,
      "learning_rate": 1.138548665425681e-05,
      "loss": 6.5164,
      "step": 11694
    },
    {
      "epoch": 0.4721437222446508,
      "grad_norm": 92.72057342529297,
      "learning_rate": 1.1384191672816092e-05,
      "loss": 4.1518,
      "step": 11695
    },
    {
      "epoch": 0.4721840936616875,
      "grad_norm": 84.3935317993164,
      "learning_rate": 1.138289666770893e-05,
      "loss": 4.6471,
      "step": 11696
    },
    {
      "epoch": 0.4722244650787243,
      "grad_norm": 88.31609344482422,
      "learning_rate": 1.138160163895746e-05,
      "loss": 5.4339,
      "step": 11697
    },
    {
      "epoch": 0.472264836495761,
      "grad_norm": 82.93585968017578,
      "learning_rate": 1.1380306586583831e-05,
      "loss": 3.1882,
      "step": 11698
    },
    {
      "epoch": 0.47230520791279773,
      "grad_norm": 94.58856201171875,
      "learning_rate": 1.137901151061018e-05,
      "loss": 5.5434,
      "step": 11699
    },
    {
      "epoch": 0.47234557932983445,
      "grad_norm": 124.82219696044922,
      "learning_rate": 1.1377716411058655e-05,
      "loss": 5.7674,
      "step": 11700
    },
    {
      "epoch": 0.47238595074687123,
      "grad_norm": 77.43951416015625,
      "learning_rate": 1.1376421287951396e-05,
      "loss": 4.9404,
      "step": 11701
    },
    {
      "epoch": 0.47242632216390795,
      "grad_norm": 108.4242935180664,
      "learning_rate": 1.1375126141310546e-05,
      "loss": 6.2505,
      "step": 11702
    },
    {
      "epoch": 0.4724666935809447,
      "grad_norm": 98.85345458984375,
      "learning_rate": 1.1373830971158251e-05,
      "loss": 5.165,
      "step": 11703
    },
    {
      "epoch": 0.47250706499798145,
      "grad_norm": 103.16912078857422,
      "learning_rate": 1.1372535777516656e-05,
      "loss": 5.6068,
      "step": 11704
    },
    {
      "epoch": 0.4725474364150182,
      "grad_norm": 109.21389770507812,
      "learning_rate": 1.1371240560407904e-05,
      "loss": 4.8584,
      "step": 11705
    },
    {
      "epoch": 0.4725878078320549,
      "grad_norm": 88.89938354492188,
      "learning_rate": 1.136994531985414e-05,
      "loss": 5.231,
      "step": 11706
    },
    {
      "epoch": 0.4726281792490916,
      "grad_norm": 123.99442291259766,
      "learning_rate": 1.1368650055877508e-05,
      "loss": 4.1814,
      "step": 11707
    },
    {
      "epoch": 0.4726685506661284,
      "grad_norm": 166.2327880859375,
      "learning_rate": 1.1367354768500159e-05,
      "loss": 5.8668,
      "step": 11708
    },
    {
      "epoch": 0.4727089220831651,
      "grad_norm": 96.31385803222656,
      "learning_rate": 1.1366059457744234e-05,
      "loss": 6.7709,
      "step": 11709
    },
    {
      "epoch": 0.47274929350020184,
      "grad_norm": 79.23566436767578,
      "learning_rate": 1.1364764123631884e-05,
      "loss": 6.2674,
      "step": 11710
    },
    {
      "epoch": 0.4727896649172386,
      "grad_norm": 119.26738739013672,
      "learning_rate": 1.1363468766185253e-05,
      "loss": 3.4265,
      "step": 11711
    },
    {
      "epoch": 0.47283003633427534,
      "grad_norm": 80.57117462158203,
      "learning_rate": 1.1362173385426492e-05,
      "loss": 5.4083,
      "step": 11712
    },
    {
      "epoch": 0.47287040775131206,
      "grad_norm": 82.53870391845703,
      "learning_rate": 1.1360877981377745e-05,
      "loss": 5.0014,
      "step": 11713
    },
    {
      "epoch": 0.4729107791683488,
      "grad_norm": 72.34688568115234,
      "learning_rate": 1.1359582554061164e-05,
      "loss": 5.6445,
      "step": 11714
    },
    {
      "epoch": 0.47295115058538556,
      "grad_norm": 82.5341796875,
      "learning_rate": 1.1358287103498895e-05,
      "loss": 4.9956,
      "step": 11715
    },
    {
      "epoch": 0.4729915220024223,
      "grad_norm": 88.46041870117188,
      "learning_rate": 1.135699162971309e-05,
      "loss": 5.2478,
      "step": 11716
    },
    {
      "epoch": 0.473031893419459,
      "grad_norm": 110.85264587402344,
      "learning_rate": 1.1355696132725895e-05,
      "loss": 5.5778,
      "step": 11717
    },
    {
      "epoch": 0.4730722648364958,
      "grad_norm": 81.36681365966797,
      "learning_rate": 1.1354400612559465e-05,
      "loss": 4.7062,
      "step": 11718
    },
    {
      "epoch": 0.4731126362535325,
      "grad_norm": 81.30318450927734,
      "learning_rate": 1.1353105069235943e-05,
      "loss": 4.5037,
      "step": 11719
    },
    {
      "epoch": 0.4731530076705692,
      "grad_norm": 98.07538604736328,
      "learning_rate": 1.1351809502777485e-05,
      "loss": 5.6026,
      "step": 11720
    },
    {
      "epoch": 0.47319337908760595,
      "grad_norm": 115.21695709228516,
      "learning_rate": 1.135051391320624e-05,
      "loss": 4.6073,
      "step": 11721
    },
    {
      "epoch": 0.4732337505046427,
      "grad_norm": 215.36146545410156,
      "learning_rate": 1.1349218300544365e-05,
      "loss": 3.6486,
      "step": 11722
    },
    {
      "epoch": 0.47327412192167945,
      "grad_norm": 84.55307006835938,
      "learning_rate": 1.1347922664814006e-05,
      "loss": 5.5778,
      "step": 11723
    },
    {
      "epoch": 0.47331449333871617,
      "grad_norm": 105.85707092285156,
      "learning_rate": 1.1346627006037317e-05,
      "loss": 6.444,
      "step": 11724
    },
    {
      "epoch": 0.47335486475575295,
      "grad_norm": 140.80548095703125,
      "learning_rate": 1.1345331324236451e-05,
      "loss": 5.6006,
      "step": 11725
    },
    {
      "epoch": 0.47339523617278967,
      "grad_norm": 70.24481201171875,
      "learning_rate": 1.134403561943356e-05,
      "loss": 5.1977,
      "step": 11726
    },
    {
      "epoch": 0.4734356075898264,
      "grad_norm": 91.44830322265625,
      "learning_rate": 1.1342739891650798e-05,
      "loss": 6.6076,
      "step": 11727
    },
    {
      "epoch": 0.47347597900686317,
      "grad_norm": 127.66104125976562,
      "learning_rate": 1.1341444140910322e-05,
      "loss": 3.995,
      "step": 11728
    },
    {
      "epoch": 0.4735163504238999,
      "grad_norm": 87.97604370117188,
      "learning_rate": 1.134014836723428e-05,
      "loss": 6.8291,
      "step": 11729
    },
    {
      "epoch": 0.4735567218409366,
      "grad_norm": 98.82374572753906,
      "learning_rate": 1.1338852570644834e-05,
      "loss": 4.988,
      "step": 11730
    },
    {
      "epoch": 0.47359709325797333,
      "grad_norm": 305.75274658203125,
      "learning_rate": 1.1337556751164131e-05,
      "loss": 6.0321,
      "step": 11731
    },
    {
      "epoch": 0.4736374646750101,
      "grad_norm": 94.45333862304688,
      "learning_rate": 1.1336260908814335e-05,
      "loss": 5.9332,
      "step": 11732
    },
    {
      "epoch": 0.47367783609204683,
      "grad_norm": 71.12248992919922,
      "learning_rate": 1.1334965043617596e-05,
      "loss": 5.607,
      "step": 11733
    },
    {
      "epoch": 0.47371820750908356,
      "grad_norm": 135.9860076904297,
      "learning_rate": 1.1333669155596074e-05,
      "loss": 3.9147,
      "step": 11734
    },
    {
      "epoch": 0.47375857892612033,
      "grad_norm": 276.1130676269531,
      "learning_rate": 1.1332373244771922e-05,
      "loss": 4.8961,
      "step": 11735
    },
    {
      "epoch": 0.47379895034315705,
      "grad_norm": 113.60781860351562,
      "learning_rate": 1.1331077311167302e-05,
      "loss": 5.6698,
      "step": 11736
    },
    {
      "epoch": 0.4738393217601938,
      "grad_norm": 71.03173065185547,
      "learning_rate": 1.1329781354804364e-05,
      "loss": 5.4337,
      "step": 11737
    },
    {
      "epoch": 0.4738796931772305,
      "grad_norm": 78.355224609375,
      "learning_rate": 1.1328485375705274e-05,
      "loss": 5.4378,
      "step": 11738
    },
    {
      "epoch": 0.4739200645942673,
      "grad_norm": 158.4001922607422,
      "learning_rate": 1.1327189373892182e-05,
      "loss": 5.5658,
      "step": 11739
    },
    {
      "epoch": 0.473960436011304,
      "grad_norm": 94.5924301147461,
      "learning_rate": 1.1325893349387252e-05,
      "loss": 5.762,
      "step": 11740
    },
    {
      "epoch": 0.4740008074283407,
      "grad_norm": 136.65301513671875,
      "learning_rate": 1.1324597302212643e-05,
      "loss": 4.4207,
      "step": 11741
    },
    {
      "epoch": 0.4740411788453775,
      "grad_norm": 64.75102233886719,
      "learning_rate": 1.1323301232390513e-05,
      "loss": 6.0356,
      "step": 11742
    },
    {
      "epoch": 0.4740815502624142,
      "grad_norm": 122.31041717529297,
      "learning_rate": 1.1322005139943022e-05,
      "loss": 5.0947,
      "step": 11743
    },
    {
      "epoch": 0.47412192167945094,
      "grad_norm": 92.93266296386719,
      "learning_rate": 1.1320709024892331e-05,
      "loss": 4.5223,
      "step": 11744
    },
    {
      "epoch": 0.47416229309648766,
      "grad_norm": 76.565673828125,
      "learning_rate": 1.1319412887260598e-05,
      "loss": 4.3789,
      "step": 11745
    },
    {
      "epoch": 0.47420266451352444,
      "grad_norm": 101.259521484375,
      "learning_rate": 1.131811672706999e-05,
      "loss": 5.349,
      "step": 11746
    },
    {
      "epoch": 0.47424303593056116,
      "grad_norm": 135.55010986328125,
      "learning_rate": 1.1316820544342658e-05,
      "loss": 6.0478,
      "step": 11747
    },
    {
      "epoch": 0.4742834073475979,
      "grad_norm": 142.47433471679688,
      "learning_rate": 1.1315524339100776e-05,
      "loss": 3.2393,
      "step": 11748
    },
    {
      "epoch": 0.47432377876463466,
      "grad_norm": 80.99559783935547,
      "learning_rate": 1.1314228111366493e-05,
      "loss": 5.4432,
      "step": 11749
    },
    {
      "epoch": 0.4743641501816714,
      "grad_norm": 104.12701416015625,
      "learning_rate": 1.1312931861161981e-05,
      "loss": 5.1988,
      "step": 11750
    },
    {
      "epoch": 0.4744045215987081,
      "grad_norm": 81.11797332763672,
      "learning_rate": 1.1311635588509402e-05,
      "loss": 4.6396,
      "step": 11751
    },
    {
      "epoch": 0.47444489301574483,
      "grad_norm": 83.03632354736328,
      "learning_rate": 1.1310339293430916e-05,
      "loss": 5.7196,
      "step": 11752
    },
    {
      "epoch": 0.4744852644327816,
      "grad_norm": 97.87516021728516,
      "learning_rate": 1.1309042975948688e-05,
      "loss": 5.815,
      "step": 11753
    },
    {
      "epoch": 0.47452563584981833,
      "grad_norm": 83.60009002685547,
      "learning_rate": 1.1307746636084885e-05,
      "loss": 4.3312,
      "step": 11754
    },
    {
      "epoch": 0.47456600726685505,
      "grad_norm": 79.67668151855469,
      "learning_rate": 1.1306450273861662e-05,
      "loss": 4.1708,
      "step": 11755
    },
    {
      "epoch": 0.4746063786838918,
      "grad_norm": 174.87246704101562,
      "learning_rate": 1.1305153889301195e-05,
      "loss": 5.9923,
      "step": 11756
    },
    {
      "epoch": 0.47464675010092855,
      "grad_norm": 167.89944458007812,
      "learning_rate": 1.1303857482425643e-05,
      "loss": 5.5374,
      "step": 11757
    },
    {
      "epoch": 0.47468712151796527,
      "grad_norm": 84.98717498779297,
      "learning_rate": 1.1302561053257172e-05,
      "loss": 5.891,
      "step": 11758
    },
    {
      "epoch": 0.474727492935002,
      "grad_norm": 66.27489471435547,
      "learning_rate": 1.1301264601817949e-05,
      "loss": 7.693,
      "step": 11759
    },
    {
      "epoch": 0.47476786435203877,
      "grad_norm": 309.5196533203125,
      "learning_rate": 1.1299968128130138e-05,
      "loss": 5.0845,
      "step": 11760
    },
    {
      "epoch": 0.4748082357690755,
      "grad_norm": 89.36902618408203,
      "learning_rate": 1.129867163221591e-05,
      "loss": 5.5814,
      "step": 11761
    },
    {
      "epoch": 0.4748486071861122,
      "grad_norm": 107.64030456542969,
      "learning_rate": 1.1297375114097431e-05,
      "loss": 5.8011,
      "step": 11762
    },
    {
      "epoch": 0.474888978603149,
      "grad_norm": 97.306640625,
      "learning_rate": 1.1296078573796864e-05,
      "loss": 6.3037,
      "step": 11763
    },
    {
      "epoch": 0.4749293500201857,
      "grad_norm": 102.35012817382812,
      "learning_rate": 1.1294782011336382e-05,
      "loss": 6.4552,
      "step": 11764
    },
    {
      "epoch": 0.47496972143722244,
      "grad_norm": 127.99943542480469,
      "learning_rate": 1.1293485426738152e-05,
      "loss": 3.8127,
      "step": 11765
    },
    {
      "epoch": 0.47501009285425916,
      "grad_norm": 94.57566833496094,
      "learning_rate": 1.1292188820024338e-05,
      "loss": 4.9003,
      "step": 11766
    },
    {
      "epoch": 0.47505046427129594,
      "grad_norm": 119.94623565673828,
      "learning_rate": 1.1290892191217116e-05,
      "loss": 4.2081,
      "step": 11767
    },
    {
      "epoch": 0.47509083568833266,
      "grad_norm": 66.42219543457031,
      "learning_rate": 1.1289595540338652e-05,
      "loss": 4.7517,
      "step": 11768
    },
    {
      "epoch": 0.4751312071053694,
      "grad_norm": 116.10834503173828,
      "learning_rate": 1.1288298867411114e-05,
      "loss": 6.1522,
      "step": 11769
    },
    {
      "epoch": 0.47517157852240616,
      "grad_norm": 96.28800201416016,
      "learning_rate": 1.1287002172456673e-05,
      "loss": 4.4052,
      "step": 11770
    },
    {
      "epoch": 0.4752119499394429,
      "grad_norm": 78.89571380615234,
      "learning_rate": 1.1285705455497502e-05,
      "loss": 4.9284,
      "step": 11771
    },
    {
      "epoch": 0.4752523213564796,
      "grad_norm": 81.28307342529297,
      "learning_rate": 1.128440871655577e-05,
      "loss": 4.8528,
      "step": 11772
    },
    {
      "epoch": 0.4752926927735163,
      "grad_norm": 91.06044006347656,
      "learning_rate": 1.1283111955653648e-05,
      "loss": 5.0095,
      "step": 11773
    },
    {
      "epoch": 0.4753330641905531,
      "grad_norm": 76.04694366455078,
      "learning_rate": 1.1281815172813309e-05,
      "loss": 5.4987,
      "step": 11774
    },
    {
      "epoch": 0.4753734356075898,
      "grad_norm": 150.11361694335938,
      "learning_rate": 1.1280518368056921e-05,
      "loss": 5.2285,
      "step": 11775
    },
    {
      "epoch": 0.47541380702462654,
      "grad_norm": 91.9353256225586,
      "learning_rate": 1.1279221541406663e-05,
      "loss": 6.6785,
      "step": 11776
    },
    {
      "epoch": 0.4754541784416633,
      "grad_norm": 88.5576171875,
      "learning_rate": 1.1277924692884699e-05,
      "loss": 4.5662,
      "step": 11777
    },
    {
      "epoch": 0.47549454985870004,
      "grad_norm": 115.35356140136719,
      "learning_rate": 1.1276627822513209e-05,
      "loss": 6.305,
      "step": 11778
    },
    {
      "epoch": 0.47553492127573677,
      "grad_norm": 107.22635650634766,
      "learning_rate": 1.1275330930314364e-05,
      "loss": 5.4757,
      "step": 11779
    },
    {
      "epoch": 0.47557529269277354,
      "grad_norm": 79.3896484375,
      "learning_rate": 1.127403401631034e-05,
      "loss": 5.1932,
      "step": 11780
    },
    {
      "epoch": 0.47561566410981027,
      "grad_norm": 90.9341049194336,
      "learning_rate": 1.1272737080523308e-05,
      "loss": 5.1821,
      "step": 11781
    },
    {
      "epoch": 0.475656035526847,
      "grad_norm": 97.0086441040039,
      "learning_rate": 1.1271440122975446e-05,
      "loss": 5.9554,
      "step": 11782
    },
    {
      "epoch": 0.4756964069438837,
      "grad_norm": 100.57707214355469,
      "learning_rate": 1.1270143143688923e-05,
      "loss": 3.425,
      "step": 11783
    },
    {
      "epoch": 0.4757367783609205,
      "grad_norm": 127.2597885131836,
      "learning_rate": 1.1268846142685923e-05,
      "loss": 3.878,
      "step": 11784
    },
    {
      "epoch": 0.4757771497779572,
      "grad_norm": 90.0723876953125,
      "learning_rate": 1.1267549119988615e-05,
      "loss": 5.8326,
      "step": 11785
    },
    {
      "epoch": 0.47581752119499393,
      "grad_norm": 146.01873779296875,
      "learning_rate": 1.1266252075619178e-05,
      "loss": 4.0916,
      "step": 11786
    },
    {
      "epoch": 0.4758578926120307,
      "grad_norm": 131.9854278564453,
      "learning_rate": 1.1264955009599783e-05,
      "loss": 4.5447,
      "step": 11787
    },
    {
      "epoch": 0.47589826402906743,
      "grad_norm": 88.92423248291016,
      "learning_rate": 1.1263657921952613e-05,
      "loss": 4.489,
      "step": 11788
    },
    {
      "epoch": 0.47593863544610415,
      "grad_norm": 87.7230453491211,
      "learning_rate": 1.1262360812699844e-05,
      "loss": 5.4423,
      "step": 11789
    },
    {
      "epoch": 0.4759790068631409,
      "grad_norm": 97.18712615966797,
      "learning_rate": 1.1261063681863655e-05,
      "loss": 5.7449,
      "step": 11790
    },
    {
      "epoch": 0.47601937828017765,
      "grad_norm": 88.10584259033203,
      "learning_rate": 1.1259766529466218e-05,
      "loss": 5.7629,
      "step": 11791
    },
    {
      "epoch": 0.4760597496972144,
      "grad_norm": 104.67778015136719,
      "learning_rate": 1.1258469355529719e-05,
      "loss": 3.2374,
      "step": 11792
    },
    {
      "epoch": 0.4761001211142511,
      "grad_norm": 146.53135681152344,
      "learning_rate": 1.1257172160076331e-05,
      "loss": 4.8,
      "step": 11793
    },
    {
      "epoch": 0.4761404925312879,
      "grad_norm": 86.46269989013672,
      "learning_rate": 1.1255874943128236e-05,
      "loss": 5.8472,
      "step": 11794
    },
    {
      "epoch": 0.4761808639483246,
      "grad_norm": 84.35860443115234,
      "learning_rate": 1.125457770470761e-05,
      "loss": 5.4043,
      "step": 11795
    },
    {
      "epoch": 0.4762212353653613,
      "grad_norm": 145.66921997070312,
      "learning_rate": 1.1253280444836637e-05,
      "loss": 5.288,
      "step": 11796
    },
    {
      "epoch": 0.47626160678239804,
      "grad_norm": 123.93172454833984,
      "learning_rate": 1.1251983163537495e-05,
      "loss": 1.9566,
      "step": 11797
    },
    {
      "epoch": 0.4763019781994348,
      "grad_norm": 85.48013305664062,
      "learning_rate": 1.1250685860832362e-05,
      "loss": 4.5981,
      "step": 11798
    },
    {
      "epoch": 0.47634234961647154,
      "grad_norm": 73.7070083618164,
      "learning_rate": 1.1249388536743421e-05,
      "loss": 4.9648,
      "step": 11799
    },
    {
      "epoch": 0.47638272103350826,
      "grad_norm": 120.42778015136719,
      "learning_rate": 1.1248091191292856e-05,
      "loss": 3.4847,
      "step": 11800
    },
    {
      "epoch": 0.47642309245054504,
      "grad_norm": 168.30661010742188,
      "learning_rate": 1.1246793824502846e-05,
      "loss": 5.861,
      "step": 11801
    },
    {
      "epoch": 0.47646346386758176,
      "grad_norm": 98.92201232910156,
      "learning_rate": 1.124549643639557e-05,
      "loss": 4.2385,
      "step": 11802
    },
    {
      "epoch": 0.4765038352846185,
      "grad_norm": 136.42547607421875,
      "learning_rate": 1.124419902699322e-05,
      "loss": 4.795,
      "step": 11803
    },
    {
      "epoch": 0.4765442067016552,
      "grad_norm": 99.08040618896484,
      "learning_rate": 1.1242901596317966e-05,
      "loss": 5.887,
      "step": 11804
    },
    {
      "epoch": 0.476584578118692,
      "grad_norm": 97.90946960449219,
      "learning_rate": 1.1241604144391999e-05,
      "loss": 6.2154,
      "step": 11805
    },
    {
      "epoch": 0.4766249495357287,
      "grad_norm": 108.27442169189453,
      "learning_rate": 1.1240306671237501e-05,
      "loss": 4.6182,
      "step": 11806
    },
    {
      "epoch": 0.4766653209527654,
      "grad_norm": 112.12167358398438,
      "learning_rate": 1.1239009176876656e-05,
      "loss": 5.2496,
      "step": 11807
    },
    {
      "epoch": 0.4767056923698022,
      "grad_norm": 135.2583465576172,
      "learning_rate": 1.1237711661331645e-05,
      "loss": 5.9005,
      "step": 11808
    },
    {
      "epoch": 0.4767460637868389,
      "grad_norm": 535.64990234375,
      "learning_rate": 1.1236414124624655e-05,
      "loss": 4.9982,
      "step": 11809
    },
    {
      "epoch": 0.47678643520387565,
      "grad_norm": 72.91928100585938,
      "learning_rate": 1.123511656677787e-05,
      "loss": 4.4275,
      "step": 11810
    },
    {
      "epoch": 0.47682680662091237,
      "grad_norm": 77.06095123291016,
      "learning_rate": 1.123381898781348e-05,
      "loss": 5.6654,
      "step": 11811
    },
    {
      "epoch": 0.47686717803794915,
      "grad_norm": 94.0400390625,
      "learning_rate": 1.1232521387753664e-05,
      "loss": 5.5757,
      "step": 11812
    },
    {
      "epoch": 0.47690754945498587,
      "grad_norm": 120.26837158203125,
      "learning_rate": 1.1231223766620611e-05,
      "loss": 4.6437,
      "step": 11813
    },
    {
      "epoch": 0.4769479208720226,
      "grad_norm": 109.37944030761719,
      "learning_rate": 1.1229926124436506e-05,
      "loss": 4.4391,
      "step": 11814
    },
    {
      "epoch": 0.47698829228905937,
      "grad_norm": 115.28561401367188,
      "learning_rate": 1.1228628461223539e-05,
      "loss": 4.8845,
      "step": 11815
    },
    {
      "epoch": 0.4770286637060961,
      "grad_norm": 84.52560424804688,
      "learning_rate": 1.122733077700389e-05,
      "loss": 3.7917,
      "step": 11816
    },
    {
      "epoch": 0.4770690351231328,
      "grad_norm": 92.08262634277344,
      "learning_rate": 1.1226033071799753e-05,
      "loss": 5.3993,
      "step": 11817
    },
    {
      "epoch": 0.47710940654016953,
      "grad_norm": 130.1393280029297,
      "learning_rate": 1.1224735345633312e-05,
      "loss": 4.5188,
      "step": 11818
    },
    {
      "epoch": 0.4771497779572063,
      "grad_norm": 123.28121948242188,
      "learning_rate": 1.1223437598526758e-05,
      "loss": 6.0208,
      "step": 11819
    },
    {
      "epoch": 0.47719014937424303,
      "grad_norm": 74.15065002441406,
      "learning_rate": 1.1222139830502277e-05,
      "loss": 6.8801,
      "step": 11820
    },
    {
      "epoch": 0.47723052079127976,
      "grad_norm": 87.49700927734375,
      "learning_rate": 1.1220842041582061e-05,
      "loss": 6.532,
      "step": 11821
    },
    {
      "epoch": 0.47727089220831653,
      "grad_norm": 150.4273223876953,
      "learning_rate": 1.1219544231788295e-05,
      "loss": 2.5069,
      "step": 11822
    },
    {
      "epoch": 0.47731126362535325,
      "grad_norm": 475.95263671875,
      "learning_rate": 1.1218246401143172e-05,
      "loss": 4.662,
      "step": 11823
    },
    {
      "epoch": 0.47735163504239,
      "grad_norm": 81.97747802734375,
      "learning_rate": 1.1216948549668877e-05,
      "loss": 5.2875,
      "step": 11824
    },
    {
      "epoch": 0.4773920064594267,
      "grad_norm": 76.18521881103516,
      "learning_rate": 1.1215650677387608e-05,
      "loss": 4.6834,
      "step": 11825
    },
    {
      "epoch": 0.4774323778764635,
      "grad_norm": 94.10504913330078,
      "learning_rate": 1.121435278432155e-05,
      "loss": 6.3889,
      "step": 11826
    },
    {
      "epoch": 0.4774727492935002,
      "grad_norm": 83.17511749267578,
      "learning_rate": 1.1213054870492894e-05,
      "loss": 5.7956,
      "step": 11827
    },
    {
      "epoch": 0.4775131207105369,
      "grad_norm": 79.5331039428711,
      "learning_rate": 1.1211756935923832e-05,
      "loss": 5.2713,
      "step": 11828
    },
    {
      "epoch": 0.4775534921275737,
      "grad_norm": 83.25688934326172,
      "learning_rate": 1.1210458980636556e-05,
      "loss": 6.2517,
      "step": 11829
    },
    {
      "epoch": 0.4775938635446104,
      "grad_norm": 180.76669311523438,
      "learning_rate": 1.1209161004653258e-05,
      "loss": 5.1308,
      "step": 11830
    },
    {
      "epoch": 0.47763423496164714,
      "grad_norm": 103.1587142944336,
      "learning_rate": 1.1207863007996132e-05,
      "loss": 4.2863,
      "step": 11831
    },
    {
      "epoch": 0.4776746063786839,
      "grad_norm": 69.062744140625,
      "learning_rate": 1.1206564990687367e-05,
      "loss": 5.2459,
      "step": 11832
    },
    {
      "epoch": 0.47771497779572064,
      "grad_norm": 142.6927490234375,
      "learning_rate": 1.120526695274916e-05,
      "loss": 4.7622,
      "step": 11833
    },
    {
      "epoch": 0.47775534921275736,
      "grad_norm": 71.60679626464844,
      "learning_rate": 1.1203968894203703e-05,
      "loss": 4.2412,
      "step": 11834
    },
    {
      "epoch": 0.4777957206297941,
      "grad_norm": 175.98826599121094,
      "learning_rate": 1.120267081507319e-05,
      "loss": 5.8394,
      "step": 11835
    },
    {
      "epoch": 0.47783609204683086,
      "grad_norm": 90.1572494506836,
      "learning_rate": 1.120137271537981e-05,
      "loss": 6.2784,
      "step": 11836
    },
    {
      "epoch": 0.4778764634638676,
      "grad_norm": 158.78382873535156,
      "learning_rate": 1.1200074595145768e-05,
      "loss": 5.6982,
      "step": 11837
    },
    {
      "epoch": 0.4779168348809043,
      "grad_norm": 81.15333557128906,
      "learning_rate": 1.1198776454393248e-05,
      "loss": 4.3487,
      "step": 11838
    },
    {
      "epoch": 0.4779572062979411,
      "grad_norm": 148.1480712890625,
      "learning_rate": 1.1197478293144451e-05,
      "loss": 6.0085,
      "step": 11839
    },
    {
      "epoch": 0.4779975777149778,
      "grad_norm": 101.46141815185547,
      "learning_rate": 1.119618011142157e-05,
      "loss": 4.6355,
      "step": 11840
    },
    {
      "epoch": 0.47803794913201453,
      "grad_norm": 93.2108383178711,
      "learning_rate": 1.1194881909246806e-05,
      "loss": 5.2384,
      "step": 11841
    },
    {
      "epoch": 0.47807832054905125,
      "grad_norm": 72.95335388183594,
      "learning_rate": 1.119358368664235e-05,
      "loss": 5.3063,
      "step": 11842
    },
    {
      "epoch": 0.478118691966088,
      "grad_norm": 98.17469024658203,
      "learning_rate": 1.1192285443630401e-05,
      "loss": 5.2006,
      "step": 11843
    },
    {
      "epoch": 0.47815906338312475,
      "grad_norm": 123.30409240722656,
      "learning_rate": 1.1190987180233156e-05,
      "loss": 4.8231,
      "step": 11844
    },
    {
      "epoch": 0.47819943480016147,
      "grad_norm": 97.43445587158203,
      "learning_rate": 1.118968889647281e-05,
      "loss": 6.1066,
      "step": 11845
    },
    {
      "epoch": 0.47823980621719825,
      "grad_norm": 116.68573760986328,
      "learning_rate": 1.118839059237156e-05,
      "loss": 6.2493,
      "step": 11846
    },
    {
      "epoch": 0.47828017763423497,
      "grad_norm": 143.9910888671875,
      "learning_rate": 1.1187092267951608e-05,
      "loss": 6.9458,
      "step": 11847
    },
    {
      "epoch": 0.4783205490512717,
      "grad_norm": 138.69175720214844,
      "learning_rate": 1.1185793923235151e-05,
      "loss": 5.4338,
      "step": 11848
    },
    {
      "epoch": 0.4783609204683084,
      "grad_norm": 79.49834442138672,
      "learning_rate": 1.1184495558244383e-05,
      "loss": 5.9073,
      "step": 11849
    },
    {
      "epoch": 0.4784012918853452,
      "grad_norm": 80.9334945678711,
      "learning_rate": 1.1183197173001509e-05,
      "loss": 6.9174,
      "step": 11850
    },
    {
      "epoch": 0.4784416633023819,
      "grad_norm": 265.13446044921875,
      "learning_rate": 1.1181898767528729e-05,
      "loss": 4.8267,
      "step": 11851
    },
    {
      "epoch": 0.47848203471941864,
      "grad_norm": 100.56343841552734,
      "learning_rate": 1.1180600341848238e-05,
      "loss": 3.6752,
      "step": 11852
    },
    {
      "epoch": 0.4785224061364554,
      "grad_norm": 78.45415496826172,
      "learning_rate": 1.117930189598224e-05,
      "loss": 5.146,
      "step": 11853
    },
    {
      "epoch": 0.47856277755349214,
      "grad_norm": 184.8787384033203,
      "learning_rate": 1.117800342995293e-05,
      "loss": 6.4061,
      "step": 11854
    },
    {
      "epoch": 0.47860314897052886,
      "grad_norm": 102.65093994140625,
      "learning_rate": 1.1176704943782518e-05,
      "loss": 5.1911,
      "step": 11855
    },
    {
      "epoch": 0.4786435203875656,
      "grad_norm": 88.38677215576172,
      "learning_rate": 1.1175406437493193e-05,
      "loss": 4.9757,
      "step": 11856
    },
    {
      "epoch": 0.47868389180460236,
      "grad_norm": 84.50687408447266,
      "learning_rate": 1.1174107911107166e-05,
      "loss": 5.675,
      "step": 11857
    },
    {
      "epoch": 0.4787242632216391,
      "grad_norm": 205.1859893798828,
      "learning_rate": 1.1172809364646632e-05,
      "loss": 4.6815,
      "step": 11858
    },
    {
      "epoch": 0.4787646346386758,
      "grad_norm": 93.82051086425781,
      "learning_rate": 1.1171510798133798e-05,
      "loss": 2.6821,
      "step": 11859
    },
    {
      "epoch": 0.4788050060557126,
      "grad_norm": 88.86936950683594,
      "learning_rate": 1.1170212211590865e-05,
      "loss": 5.8395,
      "step": 11860
    },
    {
      "epoch": 0.4788453774727493,
      "grad_norm": 128.11537170410156,
      "learning_rate": 1.116891360504004e-05,
      "loss": 5.7559,
      "step": 11861
    },
    {
      "epoch": 0.478885748889786,
      "grad_norm": 93.26876831054688,
      "learning_rate": 1.1167614978503516e-05,
      "loss": 4.2019,
      "step": 11862
    },
    {
      "epoch": 0.47892612030682274,
      "grad_norm": 87.71843719482422,
      "learning_rate": 1.1166316332003509e-05,
      "loss": 3.7974,
      "step": 11863
    },
    {
      "epoch": 0.4789664917238595,
      "grad_norm": 86.77326202392578,
      "learning_rate": 1.116501766556221e-05,
      "loss": 5.8343,
      "step": 11864
    },
    {
      "epoch": 0.47900686314089624,
      "grad_norm": 100.81918334960938,
      "learning_rate": 1.1163718979201833e-05,
      "loss": 5.6149,
      "step": 11865
    },
    {
      "epoch": 0.47904723455793297,
      "grad_norm": 273.47332763671875,
      "learning_rate": 1.1162420272944576e-05,
      "loss": 4.7126,
      "step": 11866
    },
    {
      "epoch": 0.47908760597496974,
      "grad_norm": 77.58617401123047,
      "learning_rate": 1.1161121546812649e-05,
      "loss": 4.9535,
      "step": 11867
    },
    {
      "epoch": 0.47912797739200647,
      "grad_norm": 154.52430725097656,
      "learning_rate": 1.1159822800828255e-05,
      "loss": 4.2325,
      "step": 11868
    },
    {
      "epoch": 0.4791683488090432,
      "grad_norm": 89.7510757446289,
      "learning_rate": 1.1158524035013598e-05,
      "loss": 4.7526,
      "step": 11869
    },
    {
      "epoch": 0.4792087202260799,
      "grad_norm": 95.18685150146484,
      "learning_rate": 1.1157225249390886e-05,
      "loss": 5.5531,
      "step": 11870
    },
    {
      "epoch": 0.4792490916431167,
      "grad_norm": 90.83612823486328,
      "learning_rate": 1.1155926443982325e-05,
      "loss": 5.8081,
      "step": 11871
    },
    {
      "epoch": 0.4792894630601534,
      "grad_norm": 72.38919830322266,
      "learning_rate": 1.1154627618810122e-05,
      "loss": 6.3894,
      "step": 11872
    },
    {
      "epoch": 0.47932983447719013,
      "grad_norm": 91.04747772216797,
      "learning_rate": 1.1153328773896484e-05,
      "loss": 5.5967,
      "step": 11873
    },
    {
      "epoch": 0.4793702058942269,
      "grad_norm": 89.84312438964844,
      "learning_rate": 1.1152029909263613e-05,
      "loss": 5.7031,
      "step": 11874
    },
    {
      "epoch": 0.47941057731126363,
      "grad_norm": 79.46553039550781,
      "learning_rate": 1.1150731024933722e-05,
      "loss": 4.7993,
      "step": 11875
    },
    {
      "epoch": 0.47945094872830035,
      "grad_norm": 106.27503204345703,
      "learning_rate": 1.114943212092902e-05,
      "loss": 4.5882,
      "step": 11876
    },
    {
      "epoch": 0.4794913201453371,
      "grad_norm": 106.38137817382812,
      "learning_rate": 1.1148133197271711e-05,
      "loss": 4.7654,
      "step": 11877
    },
    {
      "epoch": 0.47953169156237385,
      "grad_norm": 87.53977966308594,
      "learning_rate": 1.1146834253984008e-05,
      "loss": 4.8267,
      "step": 11878
    },
    {
      "epoch": 0.4795720629794106,
      "grad_norm": 82.71558380126953,
      "learning_rate": 1.1145535291088111e-05,
      "loss": 4.9027,
      "step": 11879
    },
    {
      "epoch": 0.4796124343964473,
      "grad_norm": 106.46724700927734,
      "learning_rate": 1.114423630860624e-05,
      "loss": 5.7874,
      "step": 11880
    },
    {
      "epoch": 0.4796528058134841,
      "grad_norm": 108.12258911132812,
      "learning_rate": 1.1142937306560601e-05,
      "loss": 5.3868,
      "step": 11881
    },
    {
      "epoch": 0.4796931772305208,
      "grad_norm": 84.59073638916016,
      "learning_rate": 1.1141638284973402e-05,
      "loss": 6.5713,
      "step": 11882
    },
    {
      "epoch": 0.4797335486475575,
      "grad_norm": 101.62577056884766,
      "learning_rate": 1.1140339243866853e-05,
      "loss": 6.5774,
      "step": 11883
    },
    {
      "epoch": 0.4797739200645943,
      "grad_norm": 96.40937042236328,
      "learning_rate": 1.1139040183263168e-05,
      "loss": 5.016,
      "step": 11884
    },
    {
      "epoch": 0.479814291481631,
      "grad_norm": 74.13780975341797,
      "learning_rate": 1.1137741103184556e-05,
      "loss": 5.0427,
      "step": 11885
    },
    {
      "epoch": 0.47985466289866774,
      "grad_norm": 109.32135009765625,
      "learning_rate": 1.1136442003653225e-05,
      "loss": 5.2007,
      "step": 11886
    },
    {
      "epoch": 0.47989503431570446,
      "grad_norm": 192.36502075195312,
      "learning_rate": 1.113514288469139e-05,
      "loss": 7.1853,
      "step": 11887
    },
    {
      "epoch": 0.47993540573274124,
      "grad_norm": 141.23974609375,
      "learning_rate": 1.1133843746321263e-05,
      "loss": 7.6003,
      "step": 11888
    },
    {
      "epoch": 0.47997577714977796,
      "grad_norm": 75.0861587524414,
      "learning_rate": 1.1132544588565057e-05,
      "loss": 6.2851,
      "step": 11889
    },
    {
      "epoch": 0.4800161485668147,
      "grad_norm": 154.30528259277344,
      "learning_rate": 1.1131245411444983e-05,
      "loss": 4.4842,
      "step": 11890
    },
    {
      "epoch": 0.48005651998385146,
      "grad_norm": 85.3221664428711,
      "learning_rate": 1.1129946214983255e-05,
      "loss": 4.8678,
      "step": 11891
    },
    {
      "epoch": 0.4800968914008882,
      "grad_norm": 112.37744903564453,
      "learning_rate": 1.1128646999202079e-05,
      "loss": 5.4347,
      "step": 11892
    },
    {
      "epoch": 0.4801372628179249,
      "grad_norm": 71.0502700805664,
      "learning_rate": 1.1127347764123681e-05,
      "loss": 5.3166,
      "step": 11893
    },
    {
      "epoch": 0.4801776342349616,
      "grad_norm": 76.5761947631836,
      "learning_rate": 1.1126048509770268e-05,
      "loss": 5.6624,
      "step": 11894
    },
    {
      "epoch": 0.4802180056519984,
      "grad_norm": 106.999755859375,
      "learning_rate": 1.1124749236164056e-05,
      "loss": 6.0137,
      "step": 11895
    },
    {
      "epoch": 0.4802583770690351,
      "grad_norm": 82.21212768554688,
      "learning_rate": 1.1123449943327256e-05,
      "loss": 5.859,
      "step": 11896
    },
    {
      "epoch": 0.48029874848607185,
      "grad_norm": 98.1056137084961,
      "learning_rate": 1.1122150631282085e-05,
      "loss": 5.9138,
      "step": 11897
    },
    {
      "epoch": 0.4803391199031086,
      "grad_norm": 123.33497619628906,
      "learning_rate": 1.1120851300050759e-05,
      "loss": 4.1193,
      "step": 11898
    },
    {
      "epoch": 0.48037949132014535,
      "grad_norm": 85.45233917236328,
      "learning_rate": 1.1119551949655495e-05,
      "loss": 5.3006,
      "step": 11899
    },
    {
      "epoch": 0.48041986273718207,
      "grad_norm": 93.52900695800781,
      "learning_rate": 1.1118252580118505e-05,
      "loss": 2.7714,
      "step": 11900
    },
    {
      "epoch": 0.4804602341542188,
      "grad_norm": 120.57173156738281,
      "learning_rate": 1.111695319146201e-05,
      "loss": 5.7174,
      "step": 11901
    },
    {
      "epoch": 0.48050060557125557,
      "grad_norm": 85.07945251464844,
      "learning_rate": 1.1115653783708223e-05,
      "loss": 5.1407,
      "step": 11902
    },
    {
      "epoch": 0.4805409769882923,
      "grad_norm": 86.99456787109375,
      "learning_rate": 1.111435435687936e-05,
      "loss": 6.1409,
      "step": 11903
    },
    {
      "epoch": 0.480581348405329,
      "grad_norm": 66.3205337524414,
      "learning_rate": 1.1113054910997641e-05,
      "loss": 4.2991,
      "step": 11904
    },
    {
      "epoch": 0.4806217198223658,
      "grad_norm": 70.7447280883789,
      "learning_rate": 1.1111755446085281e-05,
      "loss": 7.1591,
      "step": 11905
    },
    {
      "epoch": 0.4806620912394025,
      "grad_norm": 99.44231414794922,
      "learning_rate": 1.11104559621645e-05,
      "loss": 4.2111,
      "step": 11906
    },
    {
      "epoch": 0.48070246265643923,
      "grad_norm": 91.23712921142578,
      "learning_rate": 1.1109156459257513e-05,
      "loss": 5.3481,
      "step": 11907
    },
    {
      "epoch": 0.48074283407347596,
      "grad_norm": 102.09261322021484,
      "learning_rate": 1.110785693738654e-05,
      "loss": 5.8568,
      "step": 11908
    },
    {
      "epoch": 0.48078320549051273,
      "grad_norm": 78.66798400878906,
      "learning_rate": 1.1106557396573804e-05,
      "loss": 5.3874,
      "step": 11909
    },
    {
      "epoch": 0.48082357690754945,
      "grad_norm": 96.99626922607422,
      "learning_rate": 1.1105257836841519e-05,
      "loss": 6.4989,
      "step": 11910
    },
    {
      "epoch": 0.4808639483245862,
      "grad_norm": 85.08737182617188,
      "learning_rate": 1.1103958258211906e-05,
      "loss": 6.2476,
      "step": 11911
    },
    {
      "epoch": 0.48090431974162295,
      "grad_norm": 82.26693725585938,
      "learning_rate": 1.1102658660707186e-05,
      "loss": 4.5431,
      "step": 11912
    },
    {
      "epoch": 0.4809446911586597,
      "grad_norm": 76.25943756103516,
      "learning_rate": 1.1101359044349577e-05,
      "loss": 6.8455,
      "step": 11913
    },
    {
      "epoch": 0.4809850625756964,
      "grad_norm": 90.47928619384766,
      "learning_rate": 1.1100059409161298e-05,
      "loss": 5.4135,
      "step": 11914
    },
    {
      "epoch": 0.4810254339927331,
      "grad_norm": 120.35031127929688,
      "learning_rate": 1.1098759755164576e-05,
      "loss": 3.9006,
      "step": 11915
    },
    {
      "epoch": 0.4810658054097699,
      "grad_norm": 79.82662200927734,
      "learning_rate": 1.1097460082381625e-05,
      "loss": 5.3121,
      "step": 11916
    },
    {
      "epoch": 0.4811061768268066,
      "grad_norm": 84.0395278930664,
      "learning_rate": 1.1096160390834669e-05,
      "loss": 3.6816,
      "step": 11917
    },
    {
      "epoch": 0.48114654824384334,
      "grad_norm": 120.07788848876953,
      "learning_rate": 1.1094860680545931e-05,
      "loss": 4.0533,
      "step": 11918
    },
    {
      "epoch": 0.4811869196608801,
      "grad_norm": 166.78118896484375,
      "learning_rate": 1.1093560951537632e-05,
      "loss": 5.7174,
      "step": 11919
    },
    {
      "epoch": 0.48122729107791684,
      "grad_norm": 135.13909912109375,
      "learning_rate": 1.1092261203831993e-05,
      "loss": 5.3036,
      "step": 11920
    },
    {
      "epoch": 0.48126766249495356,
      "grad_norm": 96.16292572021484,
      "learning_rate": 1.1090961437451242e-05,
      "loss": 5.2535,
      "step": 11921
    },
    {
      "epoch": 0.4813080339119903,
      "grad_norm": 74.53629302978516,
      "learning_rate": 1.1089661652417593e-05,
      "loss": 5.6919,
      "step": 11922
    },
    {
      "epoch": 0.48134840532902706,
      "grad_norm": 103.61875915527344,
      "learning_rate": 1.108836184875328e-05,
      "loss": 5.4508,
      "step": 11923
    },
    {
      "epoch": 0.4813887767460638,
      "grad_norm": 174.6517791748047,
      "learning_rate": 1.1087062026480517e-05,
      "loss": 3.1542,
      "step": 11924
    },
    {
      "epoch": 0.4814291481631005,
      "grad_norm": 82.77085876464844,
      "learning_rate": 1.1085762185621533e-05,
      "loss": 4.5242,
      "step": 11925
    },
    {
      "epoch": 0.4814695195801373,
      "grad_norm": 301.8828125,
      "learning_rate": 1.108446232619855e-05,
      "loss": 6.0577,
      "step": 11926
    },
    {
      "epoch": 0.481509890997174,
      "grad_norm": 139.79989624023438,
      "learning_rate": 1.1083162448233793e-05,
      "loss": 4.1695,
      "step": 11927
    },
    {
      "epoch": 0.48155026241421073,
      "grad_norm": 179.6743621826172,
      "learning_rate": 1.1081862551749487e-05,
      "loss": 4.039,
      "step": 11928
    },
    {
      "epoch": 0.48159063383124745,
      "grad_norm": 100.62433624267578,
      "learning_rate": 1.1080562636767861e-05,
      "loss": 5.3574,
      "step": 11929
    },
    {
      "epoch": 0.4816310052482842,
      "grad_norm": 122.17202758789062,
      "learning_rate": 1.1079262703311136e-05,
      "loss": 6.7324,
      "step": 11930
    },
    {
      "epoch": 0.48167137666532095,
      "grad_norm": 99.4888687133789,
      "learning_rate": 1.1077962751401539e-05,
      "loss": 6.1461,
      "step": 11931
    },
    {
      "epoch": 0.48171174808235767,
      "grad_norm": 99.78629302978516,
      "learning_rate": 1.1076662781061296e-05,
      "loss": 6.2708,
      "step": 11932
    },
    {
      "epoch": 0.48175211949939445,
      "grad_norm": 91.30188751220703,
      "learning_rate": 1.1075362792312634e-05,
      "loss": 5.9257,
      "step": 11933
    },
    {
      "epoch": 0.48179249091643117,
      "grad_norm": 107.62635803222656,
      "learning_rate": 1.1074062785177777e-05,
      "loss": 5.0803,
      "step": 11934
    },
    {
      "epoch": 0.4818328623334679,
      "grad_norm": 77.00515747070312,
      "learning_rate": 1.1072762759678959e-05,
      "loss": 5.7106,
      "step": 11935
    },
    {
      "epoch": 0.48187323375050467,
      "grad_norm": 100.07064056396484,
      "learning_rate": 1.10714627158384e-05,
      "loss": 5.011,
      "step": 11936
    },
    {
      "epoch": 0.4819136051675414,
      "grad_norm": 238.5045623779297,
      "learning_rate": 1.107016265367833e-05,
      "loss": 4.0655,
      "step": 11937
    },
    {
      "epoch": 0.4819539765845781,
      "grad_norm": 99.72366333007812,
      "learning_rate": 1.1068862573220978e-05,
      "loss": 5.3079,
      "step": 11938
    },
    {
      "epoch": 0.48199434800161484,
      "grad_norm": 81.54249572753906,
      "learning_rate": 1.1067562474488572e-05,
      "loss": 4.9761,
      "step": 11939
    },
    {
      "epoch": 0.4820347194186516,
      "grad_norm": 107.87529754638672,
      "learning_rate": 1.106626235750334e-05,
      "loss": 4.1937,
      "step": 11940
    },
    {
      "epoch": 0.48207509083568834,
      "grad_norm": 220.9154052734375,
      "learning_rate": 1.1064962222287515e-05,
      "loss": 5.3786,
      "step": 11941
    },
    {
      "epoch": 0.48211546225272506,
      "grad_norm": 90.27748107910156,
      "learning_rate": 1.106366206886332e-05,
      "loss": 3.2703,
      "step": 11942
    },
    {
      "epoch": 0.48215583366976184,
      "grad_norm": 74.90509033203125,
      "learning_rate": 1.106236189725299e-05,
      "loss": 3.9997,
      "step": 11943
    },
    {
      "epoch": 0.48219620508679856,
      "grad_norm": 219.30780029296875,
      "learning_rate": 1.1061061707478748e-05,
      "loss": 3.9736,
      "step": 11944
    },
    {
      "epoch": 0.4822365765038353,
      "grad_norm": 89.55583190917969,
      "learning_rate": 1.1059761499562831e-05,
      "loss": 3.6972,
      "step": 11945
    },
    {
      "epoch": 0.482276947920872,
      "grad_norm": 75.50180053710938,
      "learning_rate": 1.1058461273527468e-05,
      "loss": 5.1799,
      "step": 11946
    },
    {
      "epoch": 0.4823173193379088,
      "grad_norm": 84.5674819946289,
      "learning_rate": 1.1057161029394886e-05,
      "loss": 5.5416,
      "step": 11947
    },
    {
      "epoch": 0.4823576907549455,
      "grad_norm": 190.4193115234375,
      "learning_rate": 1.1055860767187321e-05,
      "loss": 5.2878,
      "step": 11948
    },
    {
      "epoch": 0.4823980621719822,
      "grad_norm": 110.20558166503906,
      "learning_rate": 1.1054560486927001e-05,
      "loss": 4.2543,
      "step": 11949
    },
    {
      "epoch": 0.482438433589019,
      "grad_norm": 138.2822265625,
      "learning_rate": 1.105326018863616e-05,
      "loss": 3.333,
      "step": 11950
    },
    {
      "epoch": 0.4824788050060557,
      "grad_norm": 91.3213119506836,
      "learning_rate": 1.1051959872337033e-05,
      "loss": 4.9208,
      "step": 11951
    },
    {
      "epoch": 0.48251917642309244,
      "grad_norm": 102.84459686279297,
      "learning_rate": 1.1050659538051844e-05,
      "loss": 5.6327,
      "step": 11952
    },
    {
      "epoch": 0.48255954784012917,
      "grad_norm": 155.85121154785156,
      "learning_rate": 1.1049359185802832e-05,
      "loss": 4.8041,
      "step": 11953
    },
    {
      "epoch": 0.48259991925716594,
      "grad_norm": 91.2020492553711,
      "learning_rate": 1.1048058815612228e-05,
      "loss": 4.2345,
      "step": 11954
    },
    {
      "epoch": 0.48264029067420267,
      "grad_norm": 104.03917694091797,
      "learning_rate": 1.104675842750227e-05,
      "loss": 4.2291,
      "step": 11955
    },
    {
      "epoch": 0.4826806620912394,
      "grad_norm": 83.95647430419922,
      "learning_rate": 1.1045458021495182e-05,
      "loss": 4.0524,
      "step": 11956
    },
    {
      "epoch": 0.48272103350827616,
      "grad_norm": 86.19407653808594,
      "learning_rate": 1.1044157597613204e-05,
      "loss": 5.2554,
      "step": 11957
    },
    {
      "epoch": 0.4827614049253129,
      "grad_norm": 117.650146484375,
      "learning_rate": 1.1042857155878568e-05,
      "loss": 5.6007,
      "step": 11958
    },
    {
      "epoch": 0.4828017763423496,
      "grad_norm": 160.33851623535156,
      "learning_rate": 1.1041556696313511e-05,
      "loss": 5.5035,
      "step": 11959
    },
    {
      "epoch": 0.48284214775938633,
      "grad_norm": 130.57713317871094,
      "learning_rate": 1.1040256218940268e-05,
      "loss": 4.1822,
      "step": 11960
    },
    {
      "epoch": 0.4828825191764231,
      "grad_norm": 114.80471801757812,
      "learning_rate": 1.1038955723781074e-05,
      "loss": 4.1367,
      "step": 11961
    },
    {
      "epoch": 0.48292289059345983,
      "grad_norm": 123.28182220458984,
      "learning_rate": 1.103765521085816e-05,
      "loss": 3.9609,
      "step": 11962
    },
    {
      "epoch": 0.48296326201049655,
      "grad_norm": 85.1010971069336,
      "learning_rate": 1.1036354680193768e-05,
      "loss": 5.8383,
      "step": 11963
    },
    {
      "epoch": 0.48300363342753333,
      "grad_norm": 2162.93701171875,
      "learning_rate": 1.1035054131810127e-05,
      "loss": 6.1893,
      "step": 11964
    },
    {
      "epoch": 0.48304400484457005,
      "grad_norm": 134.7753448486328,
      "learning_rate": 1.1033753565729481e-05,
      "loss": 4.1897,
      "step": 11965
    },
    {
      "epoch": 0.4830843762616068,
      "grad_norm": 85.73986053466797,
      "learning_rate": 1.1032452981974063e-05,
      "loss": 5.7595,
      "step": 11966
    },
    {
      "epoch": 0.4831247476786435,
      "grad_norm": 68.18855285644531,
      "learning_rate": 1.1031152380566109e-05,
      "loss": 3.7146,
      "step": 11967
    },
    {
      "epoch": 0.4831651190956803,
      "grad_norm": 99.51892852783203,
      "learning_rate": 1.1029851761527855e-05,
      "loss": 6.0409,
      "step": 11968
    },
    {
      "epoch": 0.483205490512717,
      "grad_norm": 104.57759857177734,
      "learning_rate": 1.1028551124881544e-05,
      "loss": 5.3002,
      "step": 11969
    },
    {
      "epoch": 0.4832458619297537,
      "grad_norm": 107.7506332397461,
      "learning_rate": 1.1027250470649409e-05,
      "loss": 5.4891,
      "step": 11970
    },
    {
      "epoch": 0.4832862333467905,
      "grad_norm": 92.06665802001953,
      "learning_rate": 1.102594979885369e-05,
      "loss": 5.1857,
      "step": 11971
    },
    {
      "epoch": 0.4833266047638272,
      "grad_norm": 82.13780975341797,
      "learning_rate": 1.1024649109516625e-05,
      "loss": 5.8922,
      "step": 11972
    },
    {
      "epoch": 0.48336697618086394,
      "grad_norm": 160.34844970703125,
      "learning_rate": 1.1023348402660454e-05,
      "loss": 4.916,
      "step": 11973
    },
    {
      "epoch": 0.48340734759790066,
      "grad_norm": 80.46874237060547,
      "learning_rate": 1.1022047678307412e-05,
      "loss": 6.2325,
      "step": 11974
    },
    {
      "epoch": 0.48344771901493744,
      "grad_norm": 146.3715362548828,
      "learning_rate": 1.1020746936479747e-05,
      "loss": 5.8051,
      "step": 11975
    },
    {
      "epoch": 0.48348809043197416,
      "grad_norm": 225.19189453125,
      "learning_rate": 1.1019446177199688e-05,
      "loss": 4.1835,
      "step": 11976
    },
    {
      "epoch": 0.4835284618490109,
      "grad_norm": 78.77611541748047,
      "learning_rate": 1.1018145400489482e-05,
      "loss": 4.6658,
      "step": 11977
    },
    {
      "epoch": 0.48356883326604766,
      "grad_norm": 71.75777435302734,
      "learning_rate": 1.1016844606371364e-05,
      "loss": 3.3787,
      "step": 11978
    },
    {
      "epoch": 0.4836092046830844,
      "grad_norm": 90.13684844970703,
      "learning_rate": 1.1015543794867582e-05,
      "loss": 5.5916,
      "step": 11979
    },
    {
      "epoch": 0.4836495761001211,
      "grad_norm": 118.73210906982422,
      "learning_rate": 1.101424296600037e-05,
      "loss": 5.0702,
      "step": 11980
    },
    {
      "epoch": 0.4836899475171578,
      "grad_norm": 95.66596984863281,
      "learning_rate": 1.1012942119791976e-05,
      "loss": 4.1475,
      "step": 11981
    },
    {
      "epoch": 0.4837303189341946,
      "grad_norm": 117.62069702148438,
      "learning_rate": 1.1011641256264631e-05,
      "loss": 4.9251,
      "step": 11982
    },
    {
      "epoch": 0.4837706903512313,
      "grad_norm": 114.77445983886719,
      "learning_rate": 1.1010340375440589e-05,
      "loss": 4.2936,
      "step": 11983
    },
    {
      "epoch": 0.48381106176826805,
      "grad_norm": 161.80149841308594,
      "learning_rate": 1.1009039477342081e-05,
      "loss": 4.6423,
      "step": 11984
    },
    {
      "epoch": 0.4838514331853048,
      "grad_norm": 191.27017211914062,
      "learning_rate": 1.1007738561991355e-05,
      "loss": 3.7697,
      "step": 11985
    },
    {
      "epoch": 0.48389180460234155,
      "grad_norm": 84.74327850341797,
      "learning_rate": 1.1006437629410654e-05,
      "loss": 5.4847,
      "step": 11986
    },
    {
      "epoch": 0.48393217601937827,
      "grad_norm": 65.20771026611328,
      "learning_rate": 1.1005136679622219e-05,
      "loss": 5.7424,
      "step": 11987
    },
    {
      "epoch": 0.483972547436415,
      "grad_norm": 113.16029357910156,
      "learning_rate": 1.1003835712648294e-05,
      "loss": 4.6172,
      "step": 11988
    },
    {
      "epoch": 0.48401291885345177,
      "grad_norm": 109.39724731445312,
      "learning_rate": 1.1002534728511125e-05,
      "loss": 6.4546,
      "step": 11989
    },
    {
      "epoch": 0.4840532902704885,
      "grad_norm": 87.87232971191406,
      "learning_rate": 1.100123372723295e-05,
      "loss": 6.5909,
      "step": 11990
    },
    {
      "epoch": 0.4840936616875252,
      "grad_norm": 115.04781341552734,
      "learning_rate": 1.0999932708836019e-05,
      "loss": 4.0201,
      "step": 11991
    },
    {
      "epoch": 0.484134033104562,
      "grad_norm": 112.2264633178711,
      "learning_rate": 1.099863167334257e-05,
      "loss": 6.1841,
      "step": 11992
    },
    {
      "epoch": 0.4841744045215987,
      "grad_norm": 91.46548461914062,
      "learning_rate": 1.0997330620774856e-05,
      "loss": 5.2461,
      "step": 11993
    },
    {
      "epoch": 0.48421477593863543,
      "grad_norm": 109.07528686523438,
      "learning_rate": 1.0996029551155113e-05,
      "loss": 4.2312,
      "step": 11994
    },
    {
      "epoch": 0.4842551473556722,
      "grad_norm": 94.30126190185547,
      "learning_rate": 1.099472846450559e-05,
      "loss": 6.0231,
      "step": 11995
    },
    {
      "epoch": 0.48429551877270893,
      "grad_norm": 76.78817749023438,
      "learning_rate": 1.0993427360848537e-05,
      "loss": 2.4348,
      "step": 11996
    },
    {
      "epoch": 0.48433589018974565,
      "grad_norm": 125.5816650390625,
      "learning_rate": 1.0992126240206196e-05,
      "loss": 5.1708,
      "step": 11997
    },
    {
      "epoch": 0.4843762616067824,
      "grad_norm": 101.78448486328125,
      "learning_rate": 1.0990825102600811e-05,
      "loss": 5.219,
      "step": 11998
    },
    {
      "epoch": 0.48441663302381915,
      "grad_norm": 110.84902954101562,
      "learning_rate": 1.098952394805463e-05,
      "loss": 6.2477,
      "step": 11999
    },
    {
      "epoch": 0.4844570044408559,
      "grad_norm": 90.98673248291016,
      "learning_rate": 1.09882227765899e-05,
      "loss": 4.3266,
      "step": 12000
    },
    {
      "epoch": 0.4844973758578926,
      "grad_norm": 132.44796752929688,
      "learning_rate": 1.0986921588228871e-05,
      "loss": 6.7594,
      "step": 12001
    },
    {
      "epoch": 0.4845377472749294,
      "grad_norm": 105.69956970214844,
      "learning_rate": 1.0985620382993787e-05,
      "loss": 4.051,
      "step": 12002
    },
    {
      "epoch": 0.4845781186919661,
      "grad_norm": 96.70112609863281,
      "learning_rate": 1.0984319160906896e-05,
      "loss": 5.7938,
      "step": 12003
    },
    {
      "epoch": 0.4846184901090028,
      "grad_norm": 110.15443420410156,
      "learning_rate": 1.0983017921990441e-05,
      "loss": 6.5527,
      "step": 12004
    },
    {
      "epoch": 0.48465886152603954,
      "grad_norm": 76.940673828125,
      "learning_rate": 1.098171666626668e-05,
      "loss": 4.1097,
      "step": 12005
    },
    {
      "epoch": 0.4846992329430763,
      "grad_norm": 121.0786361694336,
      "learning_rate": 1.0980415393757855e-05,
      "loss": 6.4183,
      "step": 12006
    },
    {
      "epoch": 0.48473960436011304,
      "grad_norm": 114.18782806396484,
      "learning_rate": 1.0979114104486216e-05,
      "loss": 2.6439,
      "step": 12007
    },
    {
      "epoch": 0.48477997577714976,
      "grad_norm": 96.40703582763672,
      "learning_rate": 1.0977812798474012e-05,
      "loss": 5.4297,
      "step": 12008
    },
    {
      "epoch": 0.48482034719418654,
      "grad_norm": 66.7181625366211,
      "learning_rate": 1.0976511475743494e-05,
      "loss": 6.1222,
      "step": 12009
    },
    {
      "epoch": 0.48486071861122326,
      "grad_norm": 89.7480697631836,
      "learning_rate": 1.0975210136316909e-05,
      "loss": 5.0974,
      "step": 12010
    },
    {
      "epoch": 0.48490109002826,
      "grad_norm": 105.36217498779297,
      "learning_rate": 1.0973908780216512e-05,
      "loss": 5.3653,
      "step": 12011
    },
    {
      "epoch": 0.4849414614452967,
      "grad_norm": 70.61023712158203,
      "learning_rate": 1.0972607407464542e-05,
      "loss": 6.8088,
      "step": 12012
    },
    {
      "epoch": 0.4849818328623335,
      "grad_norm": 95.07723236083984,
      "learning_rate": 1.0971306018083262e-05,
      "loss": 3.6238,
      "step": 12013
    },
    {
      "epoch": 0.4850222042793702,
      "grad_norm": 69.46868133544922,
      "learning_rate": 1.0970004612094913e-05,
      "loss": 5.7505,
      "step": 12014
    },
    {
      "epoch": 0.48506257569640693,
      "grad_norm": 99.20114135742188,
      "learning_rate": 1.0968703189521752e-05,
      "loss": 4.8154,
      "step": 12015
    },
    {
      "epoch": 0.4851029471134437,
      "grad_norm": 86.80056762695312,
      "learning_rate": 1.096740175038603e-05,
      "loss": 5.4606,
      "step": 12016
    },
    {
      "epoch": 0.4851433185304804,
      "grad_norm": 84.6800308227539,
      "learning_rate": 1.0966100294709996e-05,
      "loss": 6.5322,
      "step": 12017
    },
    {
      "epoch": 0.48518368994751715,
      "grad_norm": 91.51101684570312,
      "learning_rate": 1.0964798822515904e-05,
      "loss": 4.8908,
      "step": 12018
    },
    {
      "epoch": 0.48522406136455387,
      "grad_norm": 100.18543243408203,
      "learning_rate": 1.0963497333826002e-05,
      "loss": 5.2138,
      "step": 12019
    },
    {
      "epoch": 0.48526443278159065,
      "grad_norm": 91.732421875,
      "learning_rate": 1.0962195828662547e-05,
      "loss": 4.4338,
      "step": 12020
    },
    {
      "epoch": 0.48530480419862737,
      "grad_norm": 97.6726303100586,
      "learning_rate": 1.0960894307047794e-05,
      "loss": 4.896,
      "step": 12021
    },
    {
      "epoch": 0.4853451756156641,
      "grad_norm": 60.15216064453125,
      "learning_rate": 1.0959592769003988e-05,
      "loss": 3.8537,
      "step": 12022
    },
    {
      "epoch": 0.48538554703270087,
      "grad_norm": 131.53932189941406,
      "learning_rate": 1.095829121455339e-05,
      "loss": 4.3961,
      "step": 12023
    },
    {
      "epoch": 0.4854259184497376,
      "grad_norm": 78.00852966308594,
      "learning_rate": 1.0956989643718245e-05,
      "loss": 5.0119,
      "step": 12024
    },
    {
      "epoch": 0.4854662898667743,
      "grad_norm": 90.32046508789062,
      "learning_rate": 1.0955688056520812e-05,
      "loss": 5.096,
      "step": 12025
    },
    {
      "epoch": 0.48550666128381104,
      "grad_norm": 91.77119445800781,
      "learning_rate": 1.0954386452983347e-05,
      "loss": 6.4504,
      "step": 12026
    },
    {
      "epoch": 0.4855470327008478,
      "grad_norm": 140.15798950195312,
      "learning_rate": 1.0953084833128105e-05,
      "loss": 5.1073,
      "step": 12027
    },
    {
      "epoch": 0.48558740411788454,
      "grad_norm": 93.03872680664062,
      "learning_rate": 1.0951783196977335e-05,
      "loss": 4.8016,
      "step": 12028
    },
    {
      "epoch": 0.48562777553492126,
      "grad_norm": 1567.1376953125,
      "learning_rate": 1.0950481544553295e-05,
      "loss": 4.2619,
      "step": 12029
    },
    {
      "epoch": 0.48566814695195804,
      "grad_norm": 105.00836944580078,
      "learning_rate": 1.094917987587824e-05,
      "loss": 5.2008,
      "step": 12030
    },
    {
      "epoch": 0.48570851836899476,
      "grad_norm": 88.47782135009766,
      "learning_rate": 1.0947878190974427e-05,
      "loss": 5.6169,
      "step": 12031
    },
    {
      "epoch": 0.4857488897860315,
      "grad_norm": 110.60387420654297,
      "learning_rate": 1.094657648986411e-05,
      "loss": 5.8441,
      "step": 12032
    },
    {
      "epoch": 0.4857892612030682,
      "grad_norm": 112.46363067626953,
      "learning_rate": 1.0945274772569545e-05,
      "loss": 7.5146,
      "step": 12033
    },
    {
      "epoch": 0.485829632620105,
      "grad_norm": 119.5266342163086,
      "learning_rate": 1.094397303911299e-05,
      "loss": 4.8713,
      "step": 12034
    },
    {
      "epoch": 0.4858700040371417,
      "grad_norm": 99.6246109008789,
      "learning_rate": 1.0942671289516698e-05,
      "loss": 4.0008,
      "step": 12035
    },
    {
      "epoch": 0.4859103754541784,
      "grad_norm": 88.36917114257812,
      "learning_rate": 1.0941369523802927e-05,
      "loss": 4.7538,
      "step": 12036
    },
    {
      "epoch": 0.4859507468712152,
      "grad_norm": 82.0610580444336,
      "learning_rate": 1.094006774199394e-05,
      "loss": 6.1946,
      "step": 12037
    },
    {
      "epoch": 0.4859911182882519,
      "grad_norm": 126.30963134765625,
      "learning_rate": 1.0938765944111987e-05,
      "loss": 5.0717,
      "step": 12038
    },
    {
      "epoch": 0.48603148970528864,
      "grad_norm": 120.97071075439453,
      "learning_rate": 1.093746413017933e-05,
      "loss": 4.63,
      "step": 12039
    },
    {
      "epoch": 0.48607186112232537,
      "grad_norm": 96.16015625,
      "learning_rate": 1.0936162300218222e-05,
      "loss": 4.4443,
      "step": 12040
    },
    {
      "epoch": 0.48611223253936214,
      "grad_norm": 93.88958740234375,
      "learning_rate": 1.093486045425093e-05,
      "loss": 5.6301,
      "step": 12041
    },
    {
      "epoch": 0.48615260395639887,
      "grad_norm": 100.77294921875,
      "learning_rate": 1.0933558592299704e-05,
      "loss": 5.0097,
      "step": 12042
    },
    {
      "epoch": 0.4861929753734356,
      "grad_norm": 103.7817611694336,
      "learning_rate": 1.0932256714386808e-05,
      "loss": 5.0555,
      "step": 12043
    },
    {
      "epoch": 0.48623334679047236,
      "grad_norm": 112.11871337890625,
      "learning_rate": 1.0930954820534496e-05,
      "loss": 5.754,
      "step": 12044
    },
    {
      "epoch": 0.4862737182075091,
      "grad_norm": 74.6978759765625,
      "learning_rate": 1.0929652910765034e-05,
      "loss": 4.8535,
      "step": 12045
    },
    {
      "epoch": 0.4863140896245458,
      "grad_norm": 111.59879302978516,
      "learning_rate": 1.0928350985100675e-05,
      "loss": 3.4339,
      "step": 12046
    },
    {
      "epoch": 0.4863544610415826,
      "grad_norm": 77.93302917480469,
      "learning_rate": 1.0927049043563685e-05,
      "loss": 4.7103,
      "step": 12047
    },
    {
      "epoch": 0.4863948324586193,
      "grad_norm": 137.65565490722656,
      "learning_rate": 1.0925747086176319e-05,
      "loss": 4.4451,
      "step": 12048
    },
    {
      "epoch": 0.48643520387565603,
      "grad_norm": 78.22884368896484,
      "learning_rate": 1.0924445112960842e-05,
      "loss": 5.4503,
      "step": 12049
    },
    {
      "epoch": 0.48647557529269275,
      "grad_norm": 87.5912094116211,
      "learning_rate": 1.0923143123939507e-05,
      "loss": 4.1294,
      "step": 12050
    },
    {
      "epoch": 0.48651594670972953,
      "grad_norm": 241.95623779296875,
      "learning_rate": 1.0921841119134586e-05,
      "loss": 5.5557,
      "step": 12051
    },
    {
      "epoch": 0.48655631812676625,
      "grad_norm": 90.1886978149414,
      "learning_rate": 1.0920539098568333e-05,
      "loss": 4.6046,
      "step": 12052
    },
    {
      "epoch": 0.486596689543803,
      "grad_norm": 117.4673080444336,
      "learning_rate": 1.091923706226301e-05,
      "loss": 5.3122,
      "step": 12053
    },
    {
      "epoch": 0.48663706096083975,
      "grad_norm": 80.55355072021484,
      "learning_rate": 1.091793501024088e-05,
      "loss": 5.3547,
      "step": 12054
    },
    {
      "epoch": 0.4866774323778765,
      "grad_norm": 90.61226654052734,
      "learning_rate": 1.0916632942524205e-05,
      "loss": 5.6812,
      "step": 12055
    },
    {
      "epoch": 0.4867178037949132,
      "grad_norm": 124.93272399902344,
      "learning_rate": 1.0915330859135246e-05,
      "loss": 5.6782,
      "step": 12056
    },
    {
      "epoch": 0.4867581752119499,
      "grad_norm": 71.75875091552734,
      "learning_rate": 1.091402876009627e-05,
      "loss": 4.8738,
      "step": 12057
    },
    {
      "epoch": 0.4867985466289867,
      "grad_norm": 73.03997039794922,
      "learning_rate": 1.0912726645429535e-05,
      "loss": 5.7904,
      "step": 12058
    },
    {
      "epoch": 0.4868389180460234,
      "grad_norm": 218.00302124023438,
      "learning_rate": 1.0911424515157305e-05,
      "loss": 4.6238,
      "step": 12059
    },
    {
      "epoch": 0.48687928946306014,
      "grad_norm": 92.23004913330078,
      "learning_rate": 1.0910122369301843e-05,
      "loss": 6.3919,
      "step": 12060
    },
    {
      "epoch": 0.4869196608800969,
      "grad_norm": 123.59809875488281,
      "learning_rate": 1.0908820207885419e-05,
      "loss": 4.2291,
      "step": 12061
    },
    {
      "epoch": 0.48696003229713364,
      "grad_norm": 102.33086395263672,
      "learning_rate": 1.0907518030930287e-05,
      "loss": 4.7315,
      "step": 12062
    },
    {
      "epoch": 0.48700040371417036,
      "grad_norm": 79.91019439697266,
      "learning_rate": 1.0906215838458717e-05,
      "loss": 6.4449,
      "step": 12063
    },
    {
      "epoch": 0.4870407751312071,
      "grad_norm": 102.41310119628906,
      "learning_rate": 1.090491363049297e-05,
      "loss": 5.0782,
      "step": 12064
    },
    {
      "epoch": 0.48708114654824386,
      "grad_norm": 125.30353546142578,
      "learning_rate": 1.0903611407055315e-05,
      "loss": 5.0233,
      "step": 12065
    },
    {
      "epoch": 0.4871215179652806,
      "grad_norm": 130.27191162109375,
      "learning_rate": 1.0902309168168014e-05,
      "loss": 4.4408,
      "step": 12066
    },
    {
      "epoch": 0.4871618893823173,
      "grad_norm": 125.34172058105469,
      "learning_rate": 1.0901006913853336e-05,
      "loss": 4.0484,
      "step": 12067
    },
    {
      "epoch": 0.4872022607993541,
      "grad_norm": 114.29234313964844,
      "learning_rate": 1.0899704644133543e-05,
      "loss": 5.1503,
      "step": 12068
    },
    {
      "epoch": 0.4872426322163908,
      "grad_norm": 75.48532104492188,
      "learning_rate": 1.0898402359030903e-05,
      "loss": 5.5271,
      "step": 12069
    },
    {
      "epoch": 0.4872830036334275,
      "grad_norm": 78.0648422241211,
      "learning_rate": 1.0897100058567676e-05,
      "loss": 5.7005,
      "step": 12070
    },
    {
      "epoch": 0.48732337505046425,
      "grad_norm": 108.55791473388672,
      "learning_rate": 1.0895797742766136e-05,
      "loss": 5.5088,
      "step": 12071
    },
    {
      "epoch": 0.487363746467501,
      "grad_norm": 77.74081420898438,
      "learning_rate": 1.0894495411648544e-05,
      "loss": 2.5711,
      "step": 12072
    },
    {
      "epoch": 0.48740411788453775,
      "grad_norm": 79.11074829101562,
      "learning_rate": 1.0893193065237173e-05,
      "loss": 5.387,
      "step": 12073
    },
    {
      "epoch": 0.48744448930157447,
      "grad_norm": 143.94029235839844,
      "learning_rate": 1.0891890703554284e-05,
      "loss": 5.8638,
      "step": 12074
    },
    {
      "epoch": 0.48748486071861125,
      "grad_norm": 100.29788970947266,
      "learning_rate": 1.0890588326622146e-05,
      "loss": 5.5861,
      "step": 12075
    },
    {
      "epoch": 0.48752523213564797,
      "grad_norm": 159.85043334960938,
      "learning_rate": 1.0889285934463027e-05,
      "loss": 4.7635,
      "step": 12076
    },
    {
      "epoch": 0.4875656035526847,
      "grad_norm": 77.3243637084961,
      "learning_rate": 1.08879835270992e-05,
      "loss": 5.1999,
      "step": 12077
    },
    {
      "epoch": 0.4876059749697214,
      "grad_norm": 78.09463500976562,
      "learning_rate": 1.0886681104552922e-05,
      "loss": 4.5627,
      "step": 12078
    },
    {
      "epoch": 0.4876463463867582,
      "grad_norm": 92.13517761230469,
      "learning_rate": 1.0885378666846473e-05,
      "loss": 5.2177,
      "step": 12079
    },
    {
      "epoch": 0.4876867178037949,
      "grad_norm": 106.1957778930664,
      "learning_rate": 1.0884076214002113e-05,
      "loss": 4.6093,
      "step": 12080
    },
    {
      "epoch": 0.48772708922083163,
      "grad_norm": 158.80258178710938,
      "learning_rate": 1.0882773746042115e-05,
      "loss": 6.5039,
      "step": 12081
    },
    {
      "epoch": 0.4877674606378684,
      "grad_norm": 205.66073608398438,
      "learning_rate": 1.0881471262988748e-05,
      "loss": 2.9679,
      "step": 12082
    },
    {
      "epoch": 0.48780783205490513,
      "grad_norm": 66.78850555419922,
      "learning_rate": 1.0880168764864279e-05,
      "loss": 3.8232,
      "step": 12083
    },
    {
      "epoch": 0.48784820347194185,
      "grad_norm": 82.0896224975586,
      "learning_rate": 1.087886625169098e-05,
      "loss": 4.0161,
      "step": 12084
    },
    {
      "epoch": 0.4878885748889786,
      "grad_norm": 103.21294403076172,
      "learning_rate": 1.087756372349112e-05,
      "loss": 4.9123,
      "step": 12085
    },
    {
      "epoch": 0.48792894630601535,
      "grad_norm": 67.38056182861328,
      "learning_rate": 1.0876261180286971e-05,
      "loss": 4.7054,
      "step": 12086
    },
    {
      "epoch": 0.4879693177230521,
      "grad_norm": 133.9624481201172,
      "learning_rate": 1.0874958622100802e-05,
      "loss": 5.5953,
      "step": 12087
    },
    {
      "epoch": 0.4880096891400888,
      "grad_norm": 116.4932861328125,
      "learning_rate": 1.0873656048954884e-05,
      "loss": 4.5607,
      "step": 12088
    },
    {
      "epoch": 0.4880500605571256,
      "grad_norm": 92.04733276367188,
      "learning_rate": 1.0872353460871488e-05,
      "loss": 4.7549,
      "step": 12089
    },
    {
      "epoch": 0.4880904319741623,
      "grad_norm": 91.09649658203125,
      "learning_rate": 1.0871050857872882e-05,
      "loss": 5.0862,
      "step": 12090
    },
    {
      "epoch": 0.488130803391199,
      "grad_norm": 110.56644439697266,
      "learning_rate": 1.0869748239981344e-05,
      "loss": 5.3405,
      "step": 12091
    },
    {
      "epoch": 0.48817117480823574,
      "grad_norm": 85.19285583496094,
      "learning_rate": 1.086844560721914e-05,
      "loss": 6.3056,
      "step": 12092
    },
    {
      "epoch": 0.4882115462252725,
      "grad_norm": 168.21278381347656,
      "learning_rate": 1.0867142959608543e-05,
      "loss": 4.488,
      "step": 12093
    },
    {
      "epoch": 0.48825191764230924,
      "grad_norm": 86.74893188476562,
      "learning_rate": 1.0865840297171826e-05,
      "loss": 6.5718,
      "step": 12094
    },
    {
      "epoch": 0.48829228905934596,
      "grad_norm": 69.06551361083984,
      "learning_rate": 1.0864537619931266e-05,
      "loss": 5.9246,
      "step": 12095
    },
    {
      "epoch": 0.48833266047638274,
      "grad_norm": 93.61431884765625,
      "learning_rate": 1.0863234927909128e-05,
      "loss": 4.2554,
      "step": 12096
    },
    {
      "epoch": 0.48837303189341946,
      "grad_norm": 75.54994201660156,
      "learning_rate": 1.0861932221127691e-05,
      "loss": 4.5378,
      "step": 12097
    },
    {
      "epoch": 0.4884134033104562,
      "grad_norm": 82.3620834350586,
      "learning_rate": 1.0860629499609223e-05,
      "loss": 4.9654,
      "step": 12098
    },
    {
      "epoch": 0.48845377472749296,
      "grad_norm": 120.2935791015625,
      "learning_rate": 1.0859326763376003e-05,
      "loss": 6.2484,
      "step": 12099
    },
    {
      "epoch": 0.4884941461445297,
      "grad_norm": 111.10394287109375,
      "learning_rate": 1.08580240124503e-05,
      "loss": 5.3401,
      "step": 12100
    },
    {
      "epoch": 0.4885345175615664,
      "grad_norm": 67.56270599365234,
      "learning_rate": 1.0856721246854391e-05,
      "loss": 4.7078,
      "step": 12101
    },
    {
      "epoch": 0.48857488897860313,
      "grad_norm": 130.3793487548828,
      "learning_rate": 1.085541846661055e-05,
      "loss": 5.5624,
      "step": 12102
    },
    {
      "epoch": 0.4886152603956399,
      "grad_norm": 71.44491577148438,
      "learning_rate": 1.0854115671741049e-05,
      "loss": 5.2768,
      "step": 12103
    },
    {
      "epoch": 0.4886556318126766,
      "grad_norm": 152.5662841796875,
      "learning_rate": 1.0852812862268163e-05,
      "loss": 6.3244,
      "step": 12104
    },
    {
      "epoch": 0.48869600322971335,
      "grad_norm": 107.64002227783203,
      "learning_rate": 1.0851510038214173e-05,
      "loss": 4.0185,
      "step": 12105
    },
    {
      "epoch": 0.4887363746467501,
      "grad_norm": 75.3589096069336,
      "learning_rate": 1.0850207199601346e-05,
      "loss": 5.7307,
      "step": 12106
    },
    {
      "epoch": 0.48877674606378685,
      "grad_norm": 125.5439453125,
      "learning_rate": 1.0848904346451966e-05,
      "loss": 5.4992,
      "step": 12107
    },
    {
      "epoch": 0.48881711748082357,
      "grad_norm": 109.7693862915039,
      "learning_rate": 1.08476014787883e-05,
      "loss": 4.2784,
      "step": 12108
    },
    {
      "epoch": 0.4888574888978603,
      "grad_norm": 129.97393798828125,
      "learning_rate": 1.084629859663263e-05,
      "loss": 5.2837,
      "step": 12109
    },
    {
      "epoch": 0.48889786031489707,
      "grad_norm": 77.8967514038086,
      "learning_rate": 1.0844995700007228e-05,
      "loss": 5.3093,
      "step": 12110
    },
    {
      "epoch": 0.4889382317319338,
      "grad_norm": 102.50684356689453,
      "learning_rate": 1.0843692788934375e-05,
      "loss": 6.5514,
      "step": 12111
    },
    {
      "epoch": 0.4889786031489705,
      "grad_norm": 108.63953399658203,
      "learning_rate": 1.0842389863436343e-05,
      "loss": 6.3585,
      "step": 12112
    },
    {
      "epoch": 0.4890189745660073,
      "grad_norm": 59.98642349243164,
      "learning_rate": 1.0841086923535412e-05,
      "loss": 4.009,
      "step": 12113
    },
    {
      "epoch": 0.489059345983044,
      "grad_norm": 133.35189819335938,
      "learning_rate": 1.0839783969253858e-05,
      "loss": 5.0834,
      "step": 12114
    },
    {
      "epoch": 0.48909971740008074,
      "grad_norm": 109.29170227050781,
      "learning_rate": 1.0838481000613963e-05,
      "loss": 5.4581,
      "step": 12115
    },
    {
      "epoch": 0.48914008881711746,
      "grad_norm": 367.7802429199219,
      "learning_rate": 1.0837178017637997e-05,
      "loss": 5.6157,
      "step": 12116
    },
    {
      "epoch": 0.48918046023415424,
      "grad_norm": 104.6900863647461,
      "learning_rate": 1.0835875020348246e-05,
      "loss": 4.9949,
      "step": 12117
    },
    {
      "epoch": 0.48922083165119096,
      "grad_norm": 92.78795623779297,
      "learning_rate": 1.0834572008766979e-05,
      "loss": 6.4668,
      "step": 12118
    },
    {
      "epoch": 0.4892612030682277,
      "grad_norm": 78.95030975341797,
      "learning_rate": 1.0833268982916482e-05,
      "loss": 5.5431,
      "step": 12119
    },
    {
      "epoch": 0.48930157448526446,
      "grad_norm": 81.680419921875,
      "learning_rate": 1.0831965942819031e-05,
      "loss": 4.9806,
      "step": 12120
    },
    {
      "epoch": 0.4893419459023012,
      "grad_norm": 100.88544464111328,
      "learning_rate": 1.0830662888496908e-05,
      "loss": 5.3606,
      "step": 12121
    },
    {
      "epoch": 0.4893823173193379,
      "grad_norm": 89.87408447265625,
      "learning_rate": 1.0829359819972385e-05,
      "loss": 4.4034,
      "step": 12122
    },
    {
      "epoch": 0.4894226887363746,
      "grad_norm": 76.07608795166016,
      "learning_rate": 1.0828056737267747e-05,
      "loss": 3.6053,
      "step": 12123
    },
    {
      "epoch": 0.4894630601534114,
      "grad_norm": 87.96504974365234,
      "learning_rate": 1.0826753640405271e-05,
      "loss": 5.7611,
      "step": 12124
    },
    {
      "epoch": 0.4895034315704481,
      "grad_norm": 89.43882751464844,
      "learning_rate": 1.0825450529407242e-05,
      "loss": 6.2654,
      "step": 12125
    },
    {
      "epoch": 0.48954380298748484,
      "grad_norm": 84.91669464111328,
      "learning_rate": 1.0824147404295933e-05,
      "loss": 5.3178,
      "step": 12126
    },
    {
      "epoch": 0.4895841744045216,
      "grad_norm": 81.8783950805664,
      "learning_rate": 1.0822844265093633e-05,
      "loss": 5.2735,
      "step": 12127
    },
    {
      "epoch": 0.48962454582155834,
      "grad_norm": 144.141845703125,
      "learning_rate": 1.0821541111822612e-05,
      "loss": 4.6434,
      "step": 12128
    },
    {
      "epoch": 0.48966491723859507,
      "grad_norm": 103.43862915039062,
      "learning_rate": 1.082023794450516e-05,
      "loss": 5.5391,
      "step": 12129
    },
    {
      "epoch": 0.4897052886556318,
      "grad_norm": 96.94815063476562,
      "learning_rate": 1.0818934763163553e-05,
      "loss": 4.7414,
      "step": 12130
    },
    {
      "epoch": 0.48974566007266856,
      "grad_norm": 96.69317626953125,
      "learning_rate": 1.0817631567820076e-05,
      "loss": 7.1951,
      "step": 12131
    },
    {
      "epoch": 0.4897860314897053,
      "grad_norm": 122.1808853149414,
      "learning_rate": 1.0816328358497003e-05,
      "loss": 5.8327,
      "step": 12132
    },
    {
      "epoch": 0.489826402906742,
      "grad_norm": 80.75115203857422,
      "learning_rate": 1.0815025135216625e-05,
      "loss": 5.2219,
      "step": 12133
    },
    {
      "epoch": 0.4898667743237788,
      "grad_norm": 140.54637145996094,
      "learning_rate": 1.0813721898001219e-05,
      "loss": 5.3798,
      "step": 12134
    },
    {
      "epoch": 0.4899071457408155,
      "grad_norm": 83.71112060546875,
      "learning_rate": 1.0812418646873073e-05,
      "loss": 6.0147,
      "step": 12135
    },
    {
      "epoch": 0.48994751715785223,
      "grad_norm": 225.79371643066406,
      "learning_rate": 1.081111538185446e-05,
      "loss": 5.5743,
      "step": 12136
    },
    {
      "epoch": 0.48998788857488895,
      "grad_norm": 114.8694839477539,
      "learning_rate": 1.0809812102967672e-05,
      "loss": 3.4128,
      "step": 12137
    },
    {
      "epoch": 0.49002825999192573,
      "grad_norm": 131.22601318359375,
      "learning_rate": 1.0808508810234985e-05,
      "loss": 3.5049,
      "step": 12138
    },
    {
      "epoch": 0.49006863140896245,
      "grad_norm": 81.16665649414062,
      "learning_rate": 1.0807205503678688e-05,
      "loss": 5.4081,
      "step": 12139
    },
    {
      "epoch": 0.4901090028259992,
      "grad_norm": 74.27438354492188,
      "learning_rate": 1.080590218332106e-05,
      "loss": 6.2347,
      "step": 12140
    },
    {
      "epoch": 0.49014937424303595,
      "grad_norm": 89.14694213867188,
      "learning_rate": 1.0804598849184388e-05,
      "loss": 4.0126,
      "step": 12141
    },
    {
      "epoch": 0.4901897456600727,
      "grad_norm": 180.36770629882812,
      "learning_rate": 1.0803295501290954e-05,
      "loss": 4.6445,
      "step": 12142
    },
    {
      "epoch": 0.4902301170771094,
      "grad_norm": 82.85572814941406,
      "learning_rate": 1.0801992139663042e-05,
      "loss": 6.8454,
      "step": 12143
    },
    {
      "epoch": 0.4902704884941461,
      "grad_norm": 82.25857543945312,
      "learning_rate": 1.0800688764322938e-05,
      "loss": 5.2562,
      "step": 12144
    },
    {
      "epoch": 0.4903108599111829,
      "grad_norm": 86.39472198486328,
      "learning_rate": 1.0799385375292925e-05,
      "loss": 5.8536,
      "step": 12145
    },
    {
      "epoch": 0.4903512313282196,
      "grad_norm": 84.73442840576172,
      "learning_rate": 1.0798081972595289e-05,
      "loss": 4.8056,
      "step": 12146
    },
    {
      "epoch": 0.49039160274525634,
      "grad_norm": 86.59113311767578,
      "learning_rate": 1.0796778556252316e-05,
      "loss": 4.7849,
      "step": 12147
    },
    {
      "epoch": 0.4904319741622931,
      "grad_norm": 103.03153991699219,
      "learning_rate": 1.079547512628629e-05,
      "loss": 6.0843,
      "step": 12148
    },
    {
      "epoch": 0.49047234557932984,
      "grad_norm": 89.46479034423828,
      "learning_rate": 1.0794171682719497e-05,
      "loss": 5.8983,
      "step": 12149
    },
    {
      "epoch": 0.49051271699636656,
      "grad_norm": 88.00788879394531,
      "learning_rate": 1.079286822557422e-05,
      "loss": 5.3041,
      "step": 12150
    },
    {
      "epoch": 0.49055308841340334,
      "grad_norm": 87.70626831054688,
      "learning_rate": 1.0791564754872752e-05,
      "loss": 5.8136,
      "step": 12151
    },
    {
      "epoch": 0.49059345983044006,
      "grad_norm": 176.84207153320312,
      "learning_rate": 1.0790261270637373e-05,
      "loss": 5.9405,
      "step": 12152
    },
    {
      "epoch": 0.4906338312474768,
      "grad_norm": 69.5641860961914,
      "learning_rate": 1.0788957772890371e-05,
      "loss": 4.1211,
      "step": 12153
    },
    {
      "epoch": 0.4906742026645135,
      "grad_norm": 89.5257339477539,
      "learning_rate": 1.0787654261654033e-05,
      "loss": 4.491,
      "step": 12154
    },
    {
      "epoch": 0.4907145740815503,
      "grad_norm": 184.5668182373047,
      "learning_rate": 1.078635073695065e-05,
      "loss": 5.3473,
      "step": 12155
    },
    {
      "epoch": 0.490754945498587,
      "grad_norm": 81.2779769897461,
      "learning_rate": 1.0785047198802503e-05,
      "loss": 5.5924,
      "step": 12156
    },
    {
      "epoch": 0.4907953169156237,
      "grad_norm": 91.0010757446289,
      "learning_rate": 1.0783743647231884e-05,
      "loss": 6.0503,
      "step": 12157
    },
    {
      "epoch": 0.4908356883326605,
      "grad_norm": 174.60870361328125,
      "learning_rate": 1.0782440082261075e-05,
      "loss": 5.6654,
      "step": 12158
    },
    {
      "epoch": 0.4908760597496972,
      "grad_norm": 113.24832153320312,
      "learning_rate": 1.0781136503912371e-05,
      "loss": 4.2323,
      "step": 12159
    },
    {
      "epoch": 0.49091643116673395,
      "grad_norm": 95.01519012451172,
      "learning_rate": 1.0779832912208055e-05,
      "loss": 4.3786,
      "step": 12160
    },
    {
      "epoch": 0.49095680258377067,
      "grad_norm": 75.33203125,
      "learning_rate": 1.077852930717042e-05,
      "loss": 3.9355,
      "step": 12161
    },
    {
      "epoch": 0.49099717400080745,
      "grad_norm": 113.03846740722656,
      "learning_rate": 1.0777225688821749e-05,
      "loss": 4.01,
      "step": 12162
    },
    {
      "epoch": 0.49103754541784417,
      "grad_norm": 82.56861114501953,
      "learning_rate": 1.0775922057184334e-05,
      "loss": 5.3869,
      "step": 12163
    },
    {
      "epoch": 0.4910779168348809,
      "grad_norm": 123.11412048339844,
      "learning_rate": 1.0774618412280464e-05,
      "loss": 3.5889,
      "step": 12164
    },
    {
      "epoch": 0.49111828825191767,
      "grad_norm": 84.46864318847656,
      "learning_rate": 1.0773314754132429e-05,
      "loss": 4.4264,
      "step": 12165
    },
    {
      "epoch": 0.4911586596689544,
      "grad_norm": 83.41239166259766,
      "learning_rate": 1.0772011082762517e-05,
      "loss": 4.8337,
      "step": 12166
    },
    {
      "epoch": 0.4911990310859911,
      "grad_norm": 70.59327697753906,
      "learning_rate": 1.0770707398193018e-05,
      "loss": 6.5175,
      "step": 12167
    },
    {
      "epoch": 0.49123940250302783,
      "grad_norm": 173.10580444335938,
      "learning_rate": 1.0769403700446222e-05,
      "loss": 4.8659,
      "step": 12168
    },
    {
      "epoch": 0.4912797739200646,
      "grad_norm": 134.83265686035156,
      "learning_rate": 1.0768099989544421e-05,
      "loss": 5.9473,
      "step": 12169
    },
    {
      "epoch": 0.49132014533710133,
      "grad_norm": 172.7861328125,
      "learning_rate": 1.0766796265509903e-05,
      "loss": 4.7464,
      "step": 12170
    },
    {
      "epoch": 0.49136051675413805,
      "grad_norm": 99.56076049804688,
      "learning_rate": 1.076549252836496e-05,
      "loss": 4.526,
      "step": 12171
    },
    {
      "epoch": 0.49140088817117483,
      "grad_norm": 103.65513610839844,
      "learning_rate": 1.076418877813188e-05,
      "loss": 5.1749,
      "step": 12172
    },
    {
      "epoch": 0.49144125958821155,
      "grad_norm": 102.39908599853516,
      "learning_rate": 1.0762885014832959e-05,
      "loss": 5.373,
      "step": 12173
    },
    {
      "epoch": 0.4914816310052483,
      "grad_norm": 114.53435516357422,
      "learning_rate": 1.0761581238490482e-05,
      "loss": 3.2508,
      "step": 12174
    },
    {
      "epoch": 0.491522002422285,
      "grad_norm": 81.77143096923828,
      "learning_rate": 1.0760277449126748e-05,
      "loss": 4.8,
      "step": 12175
    },
    {
      "epoch": 0.4915623738393218,
      "grad_norm": 135.4193878173828,
      "learning_rate": 1.0758973646764044e-05,
      "loss": 4.3798,
      "step": 12176
    },
    {
      "epoch": 0.4916027452563585,
      "grad_norm": 255.23037719726562,
      "learning_rate": 1.0757669831424662e-05,
      "loss": 5.353,
      "step": 12177
    },
    {
      "epoch": 0.4916431166733952,
      "grad_norm": 83.4896011352539,
      "learning_rate": 1.0756366003130896e-05,
      "loss": 4.7934,
      "step": 12178
    },
    {
      "epoch": 0.491683488090432,
      "grad_norm": 177.9975128173828,
      "learning_rate": 1.075506216190504e-05,
      "loss": 5.311,
      "step": 12179
    },
    {
      "epoch": 0.4917238595074687,
      "grad_norm": 90.90950775146484,
      "learning_rate": 1.075375830776938e-05,
      "loss": 5.0932,
      "step": 12180
    },
    {
      "epoch": 0.49176423092450544,
      "grad_norm": 113.64595031738281,
      "learning_rate": 1.0752454440746216e-05,
      "loss": 5.4948,
      "step": 12181
    },
    {
      "epoch": 0.49180460234154216,
      "grad_norm": 98.82552337646484,
      "learning_rate": 1.0751150560857834e-05,
      "loss": 5.7427,
      "step": 12182
    },
    {
      "epoch": 0.49184497375857894,
      "grad_norm": 114.03471374511719,
      "learning_rate": 1.0749846668126535e-05,
      "loss": 5.0169,
      "step": 12183
    },
    {
      "epoch": 0.49188534517561566,
      "grad_norm": 111.92913055419922,
      "learning_rate": 1.0748542762574608e-05,
      "loss": 5.1403,
      "step": 12184
    },
    {
      "epoch": 0.4919257165926524,
      "grad_norm": 96.78511810302734,
      "learning_rate": 1.074723884422435e-05,
      "loss": 4.4581,
      "step": 12185
    },
    {
      "epoch": 0.49196608800968916,
      "grad_norm": 232.95755004882812,
      "learning_rate": 1.074593491309805e-05,
      "loss": 6.1764,
      "step": 12186
    },
    {
      "epoch": 0.4920064594267259,
      "grad_norm": 106.49105072021484,
      "learning_rate": 1.0744630969218006e-05,
      "loss": 3.493,
      "step": 12187
    },
    {
      "epoch": 0.4920468308437626,
      "grad_norm": 62.82715606689453,
      "learning_rate": 1.074332701260651e-05,
      "loss": 4.5461,
      "step": 12188
    },
    {
      "epoch": 0.49208720226079933,
      "grad_norm": 94.68704986572266,
      "learning_rate": 1.0742023043285861e-05,
      "loss": 4.2524,
      "step": 12189
    },
    {
      "epoch": 0.4921275736778361,
      "grad_norm": 155.367919921875,
      "learning_rate": 1.0740719061278349e-05,
      "loss": 4.1701,
      "step": 12190
    },
    {
      "epoch": 0.4921679450948728,
      "grad_norm": 100.33899688720703,
      "learning_rate": 1.0739415066606268e-05,
      "loss": 5.8609,
      "step": 12191
    },
    {
      "epoch": 0.49220831651190955,
      "grad_norm": 86.052734375,
      "learning_rate": 1.0738111059291918e-05,
      "loss": 6.1119,
      "step": 12192
    },
    {
      "epoch": 0.4922486879289463,
      "grad_norm": 79.62735748291016,
      "learning_rate": 1.0736807039357594e-05,
      "loss": 3.7183,
      "step": 12193
    },
    {
      "epoch": 0.49228905934598305,
      "grad_norm": 117.7166519165039,
      "learning_rate": 1.0735503006825587e-05,
      "loss": 5.1719,
      "step": 12194
    },
    {
      "epoch": 0.49232943076301977,
      "grad_norm": 85.62725067138672,
      "learning_rate": 1.07341989617182e-05,
      "loss": 4.3847,
      "step": 12195
    },
    {
      "epoch": 0.4923698021800565,
      "grad_norm": 147.57562255859375,
      "learning_rate": 1.0732894904057723e-05,
      "loss": 5.2181,
      "step": 12196
    },
    {
      "epoch": 0.49241017359709327,
      "grad_norm": 102.84234619140625,
      "learning_rate": 1.0731590833866457e-05,
      "loss": 5.9638,
      "step": 12197
    },
    {
      "epoch": 0.49245054501413,
      "grad_norm": 161.28451538085938,
      "learning_rate": 1.0730286751166691e-05,
      "loss": 4.2812,
      "step": 12198
    },
    {
      "epoch": 0.4924909164311667,
      "grad_norm": 93.32481384277344,
      "learning_rate": 1.0728982655980733e-05,
      "loss": 4.7902,
      "step": 12199
    },
    {
      "epoch": 0.4925312878482035,
      "grad_norm": 178.54922485351562,
      "learning_rate": 1.0727678548330868e-05,
      "loss": 5.3778,
      "step": 12200
    },
    {
      "epoch": 0.4925716592652402,
      "grad_norm": 93.72476959228516,
      "learning_rate": 1.0726374428239404e-05,
      "loss": 6.6825,
      "step": 12201
    },
    {
      "epoch": 0.49261203068227694,
      "grad_norm": 108.9031753540039,
      "learning_rate": 1.072507029572863e-05,
      "loss": 4.7694,
      "step": 12202
    },
    {
      "epoch": 0.4926524020993137,
      "grad_norm": 86.66506958007812,
      "learning_rate": 1.072376615082085e-05,
      "loss": 3.2893,
      "step": 12203
    },
    {
      "epoch": 0.49269277351635044,
      "grad_norm": 88.89822387695312,
      "learning_rate": 1.0722461993538355e-05,
      "loss": 5.8823,
      "step": 12204
    },
    {
      "epoch": 0.49273314493338716,
      "grad_norm": 109.73869323730469,
      "learning_rate": 1.0721157823903451e-05,
      "loss": 6.259,
      "step": 12205
    },
    {
      "epoch": 0.4927735163504239,
      "grad_norm": 160.16490173339844,
      "learning_rate": 1.071985364193843e-05,
      "loss": 5.6125,
      "step": 12206
    },
    {
      "epoch": 0.49281388776746066,
      "grad_norm": 137.6885986328125,
      "learning_rate": 1.0718549447665596e-05,
      "loss": 6.3658,
      "step": 12207
    },
    {
      "epoch": 0.4928542591844974,
      "grad_norm": 165.72190856933594,
      "learning_rate": 1.0717245241107238e-05,
      "loss": 5.0678,
      "step": 12208
    },
    {
      "epoch": 0.4928946306015341,
      "grad_norm": 85.89949798583984,
      "learning_rate": 1.0715941022285668e-05,
      "loss": 3.8348,
      "step": 12209
    },
    {
      "epoch": 0.4929350020185709,
      "grad_norm": 80.40352630615234,
      "learning_rate": 1.0714636791223174e-05,
      "loss": 5.0498,
      "step": 12210
    },
    {
      "epoch": 0.4929753734356076,
      "grad_norm": 190.9270782470703,
      "learning_rate": 1.0713332547942062e-05,
      "loss": 4.6971,
      "step": 12211
    },
    {
      "epoch": 0.4930157448526443,
      "grad_norm": 87.34874725341797,
      "learning_rate": 1.0712028292464627e-05,
      "loss": 5.6459,
      "step": 12212
    },
    {
      "epoch": 0.49305611626968104,
      "grad_norm": 87.8160629272461,
      "learning_rate": 1.0710724024813172e-05,
      "loss": 3.3742,
      "step": 12213
    },
    {
      "epoch": 0.4930964876867178,
      "grad_norm": 121.58211517333984,
      "learning_rate": 1.0709419745009997e-05,
      "loss": 5.1074,
      "step": 12214
    },
    {
      "epoch": 0.49313685910375454,
      "grad_norm": 123.18338012695312,
      "learning_rate": 1.0708115453077401e-05,
      "loss": 5.0858,
      "step": 12215
    },
    {
      "epoch": 0.49317723052079127,
      "grad_norm": 79.03206634521484,
      "learning_rate": 1.0706811149037682e-05,
      "loss": 5.6015,
      "step": 12216
    },
    {
      "epoch": 0.49321760193782804,
      "grad_norm": 94.7481460571289,
      "learning_rate": 1.0705506832913146e-05,
      "loss": 4.2617,
      "step": 12217
    },
    {
      "epoch": 0.49325797335486476,
      "grad_norm": 85.64386749267578,
      "learning_rate": 1.0704202504726087e-05,
      "loss": 4.6656,
      "step": 12218
    },
    {
      "epoch": 0.4932983447719015,
      "grad_norm": 59.833065032958984,
      "learning_rate": 1.0702898164498814e-05,
      "loss": 3.0779,
      "step": 12219
    },
    {
      "epoch": 0.4933387161889382,
      "grad_norm": 153.95005798339844,
      "learning_rate": 1.0701593812253619e-05,
      "loss": 5.3818,
      "step": 12220
    },
    {
      "epoch": 0.493379087605975,
      "grad_norm": 87.74018859863281,
      "learning_rate": 1.070028944801281e-05,
      "loss": 6.2223,
      "step": 12221
    },
    {
      "epoch": 0.4934194590230117,
      "grad_norm": 78.68948364257812,
      "learning_rate": 1.0698985071798686e-05,
      "loss": 5.0561,
      "step": 12222
    },
    {
      "epoch": 0.49345983044004843,
      "grad_norm": 103.51437377929688,
      "learning_rate": 1.069768068363355e-05,
      "loss": 5.5269,
      "step": 12223
    },
    {
      "epoch": 0.4935002018570852,
      "grad_norm": 106.68152618408203,
      "learning_rate": 1.0696376283539704e-05,
      "loss": 5.0481,
      "step": 12224
    },
    {
      "epoch": 0.49354057327412193,
      "grad_norm": 141.43663024902344,
      "learning_rate": 1.0695071871539453e-05,
      "loss": 5.7495,
      "step": 12225
    },
    {
      "epoch": 0.49358094469115865,
      "grad_norm": 149.20242309570312,
      "learning_rate": 1.069376744765509e-05,
      "loss": 4.7743,
      "step": 12226
    },
    {
      "epoch": 0.4936213161081954,
      "grad_norm": 81.32028198242188,
      "learning_rate": 1.069246301190893e-05,
      "loss": 6.0414,
      "step": 12227
    },
    {
      "epoch": 0.49366168752523215,
      "grad_norm": 63.5081901550293,
      "learning_rate": 1.0691158564323265e-05,
      "loss": 5.0804,
      "step": 12228
    },
    {
      "epoch": 0.4937020589422689,
      "grad_norm": 100.1009521484375,
      "learning_rate": 1.0689854104920406e-05,
      "loss": 5.7732,
      "step": 12229
    },
    {
      "epoch": 0.4937424303593056,
      "grad_norm": 80.47943115234375,
      "learning_rate": 1.0688549633722646e-05,
      "loss": 6.1474,
      "step": 12230
    },
    {
      "epoch": 0.4937828017763424,
      "grad_norm": 85.23479461669922,
      "learning_rate": 1.0687245150752301e-05,
      "loss": 5.4822,
      "step": 12231
    },
    {
      "epoch": 0.4938231731933791,
      "grad_norm": 105.07331085205078,
      "learning_rate": 1.0685940656031666e-05,
      "loss": 5.146,
      "step": 12232
    },
    {
      "epoch": 0.4938635446104158,
      "grad_norm": 98.65460205078125,
      "learning_rate": 1.068463614958305e-05,
      "loss": 5.899,
      "step": 12233
    },
    {
      "epoch": 0.49390391602745254,
      "grad_norm": 87.78785705566406,
      "learning_rate": 1.0683331631428752e-05,
      "loss": 5.0569,
      "step": 12234
    },
    {
      "epoch": 0.4939442874444893,
      "grad_norm": 97.80513000488281,
      "learning_rate": 1.0682027101591081e-05,
      "loss": 4.5922,
      "step": 12235
    },
    {
      "epoch": 0.49398465886152604,
      "grad_norm": 88.59515380859375,
      "learning_rate": 1.0680722560092338e-05,
      "loss": 5.6815,
      "step": 12236
    },
    {
      "epoch": 0.49402503027856276,
      "grad_norm": 101.58595275878906,
      "learning_rate": 1.067941800695483e-05,
      "loss": 6.0058,
      "step": 12237
    },
    {
      "epoch": 0.49406540169559954,
      "grad_norm": 93.91985321044922,
      "learning_rate": 1.0678113442200861e-05,
      "loss": 6.5201,
      "step": 12238
    },
    {
      "epoch": 0.49410577311263626,
      "grad_norm": 115.7325210571289,
      "learning_rate": 1.0676808865852735e-05,
      "loss": 5.366,
      "step": 12239
    },
    {
      "epoch": 0.494146144529673,
      "grad_norm": 145.0578155517578,
      "learning_rate": 1.0675504277932758e-05,
      "loss": 4.4668,
      "step": 12240
    },
    {
      "epoch": 0.4941865159467097,
      "grad_norm": 83.9842300415039,
      "learning_rate": 1.0674199678463232e-05,
      "loss": 5.2598,
      "step": 12241
    },
    {
      "epoch": 0.4942268873637465,
      "grad_norm": 81.5254135131836,
      "learning_rate": 1.067289506746647e-05,
      "loss": 3.9795,
      "step": 12242
    },
    {
      "epoch": 0.4942672587807832,
      "grad_norm": 88.48371887207031,
      "learning_rate": 1.0671590444964771e-05,
      "loss": 5.3364,
      "step": 12243
    },
    {
      "epoch": 0.4943076301978199,
      "grad_norm": 72.9781723022461,
      "learning_rate": 1.0670285810980446e-05,
      "loss": 4.7324,
      "step": 12244
    },
    {
      "epoch": 0.4943480016148567,
      "grad_norm": 147.7472381591797,
      "learning_rate": 1.0668981165535797e-05,
      "loss": 5.4286,
      "step": 12245
    },
    {
      "epoch": 0.4943883730318934,
      "grad_norm": 87.37765502929688,
      "learning_rate": 1.0667676508653133e-05,
      "loss": 3.6386,
      "step": 12246
    },
    {
      "epoch": 0.49442874444893015,
      "grad_norm": 125.32028198242188,
      "learning_rate": 1.066637184035476e-05,
      "loss": 4.4474,
      "step": 12247
    },
    {
      "epoch": 0.49446911586596687,
      "grad_norm": 156.77621459960938,
      "learning_rate": 1.0665067160662985e-05,
      "loss": 5.3603,
      "step": 12248
    },
    {
      "epoch": 0.49450948728300365,
      "grad_norm": 88.77783966064453,
      "learning_rate": 1.0663762469600115e-05,
      "loss": 6.2343,
      "step": 12249
    },
    {
      "epoch": 0.49454985870004037,
      "grad_norm": 84.34416198730469,
      "learning_rate": 1.0662457767188454e-05,
      "loss": 4.4964,
      "step": 12250
    },
    {
      "epoch": 0.4945902301170771,
      "grad_norm": 259.42987060546875,
      "learning_rate": 1.0661153053450312e-05,
      "loss": 5.0151,
      "step": 12251
    },
    {
      "epoch": 0.49463060153411387,
      "grad_norm": 82.34923553466797,
      "learning_rate": 1.0659848328408e-05,
      "loss": 5.0872,
      "step": 12252
    },
    {
      "epoch": 0.4946709729511506,
      "grad_norm": 79.17070770263672,
      "learning_rate": 1.065854359208382e-05,
      "loss": 5.1322,
      "step": 12253
    },
    {
      "epoch": 0.4947113443681873,
      "grad_norm": 99.52203369140625,
      "learning_rate": 1.0657238844500084e-05,
      "loss": 4.9493,
      "step": 12254
    },
    {
      "epoch": 0.4947517157852241,
      "grad_norm": 85.47605895996094,
      "learning_rate": 1.06559340856791e-05,
      "loss": 4.1651,
      "step": 12255
    },
    {
      "epoch": 0.4947920872022608,
      "grad_norm": 85.13738250732422,
      "learning_rate": 1.0654629315643173e-05,
      "loss": 4.5296,
      "step": 12256
    },
    {
      "epoch": 0.49483245861929753,
      "grad_norm": 157.7255096435547,
      "learning_rate": 1.0653324534414614e-05,
      "loss": 4.99,
      "step": 12257
    },
    {
      "epoch": 0.49487283003633425,
      "grad_norm": 83.84123992919922,
      "learning_rate": 1.0652019742015728e-05,
      "loss": 6.0126,
      "step": 12258
    },
    {
      "epoch": 0.49491320145337103,
      "grad_norm": 154.43032836914062,
      "learning_rate": 1.0650714938468832e-05,
      "loss": 6.2141,
      "step": 12259
    },
    {
      "epoch": 0.49495357287040775,
      "grad_norm": 100.365234375,
      "learning_rate": 1.0649410123796224e-05,
      "loss": 4.0148,
      "step": 12260
    },
    {
      "epoch": 0.4949939442874445,
      "grad_norm": 113.49577331542969,
      "learning_rate": 1.0648105298020224e-05,
      "loss": 4.9781,
      "step": 12261
    },
    {
      "epoch": 0.49503431570448125,
      "grad_norm": 118.23287963867188,
      "learning_rate": 1.0646800461163134e-05,
      "loss": 4.6086,
      "step": 12262
    },
    {
      "epoch": 0.495074687121518,
      "grad_norm": 90.50457763671875,
      "learning_rate": 1.064549561324727e-05,
      "loss": 5.6174,
      "step": 12263
    },
    {
      "epoch": 0.4951150585385547,
      "grad_norm": 84.97427368164062,
      "learning_rate": 1.0644190754294933e-05,
      "loss": 5.2302,
      "step": 12264
    },
    {
      "epoch": 0.4951554299555914,
      "grad_norm": 99.7044677734375,
      "learning_rate": 1.0642885884328445e-05,
      "loss": 4.7465,
      "step": 12265
    },
    {
      "epoch": 0.4951958013726282,
      "grad_norm": 94.86004638671875,
      "learning_rate": 1.0641581003370105e-05,
      "loss": 5.9787,
      "step": 12266
    },
    {
      "epoch": 0.4952361727896649,
      "grad_norm": 140.5902862548828,
      "learning_rate": 1.064027611144223e-05,
      "loss": 5.422,
      "step": 12267
    },
    {
      "epoch": 0.49527654420670164,
      "grad_norm": 91.44464874267578,
      "learning_rate": 1.0638971208567128e-05,
      "loss": 4.7805,
      "step": 12268
    },
    {
      "epoch": 0.4953169156237384,
      "grad_norm": 87.7524185180664,
      "learning_rate": 1.0637666294767113e-05,
      "loss": 5.2178,
      "step": 12269
    },
    {
      "epoch": 0.49535728704077514,
      "grad_norm": 94.43921661376953,
      "learning_rate": 1.063636137006449e-05,
      "loss": 6.2175,
      "step": 12270
    },
    {
      "epoch": 0.49539765845781186,
      "grad_norm": 126.31068420410156,
      "learning_rate": 1.0635056434481571e-05,
      "loss": 5.4971,
      "step": 12271
    },
    {
      "epoch": 0.4954380298748486,
      "grad_norm": 124.94398498535156,
      "learning_rate": 1.0633751488040676e-05,
      "loss": 5.197,
      "step": 12272
    },
    {
      "epoch": 0.49547840129188536,
      "grad_norm": 115.40872192382812,
      "learning_rate": 1.0632446530764108e-05,
      "loss": 5.2938,
      "step": 12273
    },
    {
      "epoch": 0.4955187727089221,
      "grad_norm": 69.51714324951172,
      "learning_rate": 1.0631141562674181e-05,
      "loss": 3.2602,
      "step": 12274
    },
    {
      "epoch": 0.4955591441259588,
      "grad_norm": 208.24232482910156,
      "learning_rate": 1.0629836583793208e-05,
      "loss": 3.8606,
      "step": 12275
    },
    {
      "epoch": 0.4955995155429956,
      "grad_norm": 85.9074935913086,
      "learning_rate": 1.0628531594143498e-05,
      "loss": 4.6078,
      "step": 12276
    },
    {
      "epoch": 0.4956398869600323,
      "grad_norm": 108.10701751708984,
      "learning_rate": 1.0627226593747367e-05,
      "loss": 4.9542,
      "step": 12277
    },
    {
      "epoch": 0.495680258377069,
      "grad_norm": 74.95468139648438,
      "learning_rate": 1.0625921582627127e-05,
      "loss": 4.8132,
      "step": 12278
    },
    {
      "epoch": 0.49572062979410575,
      "grad_norm": 97.03614807128906,
      "learning_rate": 1.0624616560805089e-05,
      "loss": 6.5086,
      "step": 12279
    },
    {
      "epoch": 0.4957610012111425,
      "grad_norm": 78.01211547851562,
      "learning_rate": 1.0623311528303565e-05,
      "loss": 5.831,
      "step": 12280
    },
    {
      "epoch": 0.49580137262817925,
      "grad_norm": 94.80607604980469,
      "learning_rate": 1.0622006485144869e-05,
      "loss": 4.2405,
      "step": 12281
    },
    {
      "epoch": 0.49584174404521597,
      "grad_norm": 82.41816711425781,
      "learning_rate": 1.0620701431351314e-05,
      "loss": 6.2118,
      "step": 12282
    },
    {
      "epoch": 0.49588211546225275,
      "grad_norm": 130.94961547851562,
      "learning_rate": 1.0619396366945216e-05,
      "loss": 4.9704,
      "step": 12283
    },
    {
      "epoch": 0.49592248687928947,
      "grad_norm": 185.67129516601562,
      "learning_rate": 1.0618091291948889e-05,
      "loss": 3.8541,
      "step": 12284
    },
    {
      "epoch": 0.4959628582963262,
      "grad_norm": 78.56844329833984,
      "learning_rate": 1.061678620638464e-05,
      "loss": 5.3554,
      "step": 12285
    },
    {
      "epoch": 0.4960032297133629,
      "grad_norm": 91.28250122070312,
      "learning_rate": 1.061548111027479e-05,
      "loss": 4.3722,
      "step": 12286
    },
    {
      "epoch": 0.4960436011303997,
      "grad_norm": 69.5947036743164,
      "learning_rate": 1.0614176003641647e-05,
      "loss": 4.5692,
      "step": 12287
    },
    {
      "epoch": 0.4960839725474364,
      "grad_norm": 97.06123352050781,
      "learning_rate": 1.0612870886507533e-05,
      "loss": 4.5914,
      "step": 12288
    },
    {
      "epoch": 0.49612434396447314,
      "grad_norm": 120.90510559082031,
      "learning_rate": 1.0611565758894755e-05,
      "loss": 3.5532,
      "step": 12289
    },
    {
      "epoch": 0.4961647153815099,
      "grad_norm": 107.92993927001953,
      "learning_rate": 1.0610260620825632e-05,
      "loss": 5.8208,
      "step": 12290
    },
    {
      "epoch": 0.49620508679854664,
      "grad_norm": 86.28282928466797,
      "learning_rate": 1.0608955472322475e-05,
      "loss": 5.3535,
      "step": 12291
    },
    {
      "epoch": 0.49624545821558336,
      "grad_norm": 161.3451385498047,
      "learning_rate": 1.0607650313407603e-05,
      "loss": 4.6795,
      "step": 12292
    },
    {
      "epoch": 0.4962858296326201,
      "grad_norm": 134.41757202148438,
      "learning_rate": 1.0606345144103327e-05,
      "loss": 5.0034,
      "step": 12293
    },
    {
      "epoch": 0.49632620104965686,
      "grad_norm": 135.78530883789062,
      "learning_rate": 1.060503996443197e-05,
      "loss": 3.1322,
      "step": 12294
    },
    {
      "epoch": 0.4963665724666936,
      "grad_norm": 264.69061279296875,
      "learning_rate": 1.060373477441584e-05,
      "loss": 5.7695,
      "step": 12295
    },
    {
      "epoch": 0.4964069438837303,
      "grad_norm": 149.26687622070312,
      "learning_rate": 1.0602429574077256e-05,
      "loss": 5.575,
      "step": 12296
    },
    {
      "epoch": 0.4964473153007671,
      "grad_norm": 91.68489837646484,
      "learning_rate": 1.060112436343853e-05,
      "loss": 5.0767,
      "step": 12297
    },
    {
      "epoch": 0.4964876867178038,
      "grad_norm": 96.99364471435547,
      "learning_rate": 1.0599819142521986e-05,
      "loss": 4.1425,
      "step": 12298
    },
    {
      "epoch": 0.4965280581348405,
      "grad_norm": 82.87806701660156,
      "learning_rate": 1.0598513911349932e-05,
      "loss": 5.3654,
      "step": 12299
    },
    {
      "epoch": 0.49656842955187724,
      "grad_norm": 90.84515380859375,
      "learning_rate": 1.0597208669944688e-05,
      "loss": 4.3368,
      "step": 12300
    },
    {
      "epoch": 0.496608800968914,
      "grad_norm": 120.48639678955078,
      "learning_rate": 1.059590341832857e-05,
      "loss": 4.7834,
      "step": 12301
    },
    {
      "epoch": 0.49664917238595074,
      "grad_norm": 110.74579620361328,
      "learning_rate": 1.0594598156523896e-05,
      "loss": 3.285,
      "step": 12302
    },
    {
      "epoch": 0.49668954380298747,
      "grad_norm": 112.4175033569336,
      "learning_rate": 1.0593292884552982e-05,
      "loss": 5.2231,
      "step": 12303
    },
    {
      "epoch": 0.49672991522002424,
      "grad_norm": 145.0962677001953,
      "learning_rate": 1.0591987602438146e-05,
      "loss": 5.2371,
      "step": 12304
    },
    {
      "epoch": 0.49677028663706096,
      "grad_norm": 97.54679870605469,
      "learning_rate": 1.0590682310201703e-05,
      "loss": 4.7837,
      "step": 12305
    },
    {
      "epoch": 0.4968106580540977,
      "grad_norm": 78.80978393554688,
      "learning_rate": 1.0589377007865973e-05,
      "loss": 4.3558,
      "step": 12306
    },
    {
      "epoch": 0.49685102947113446,
      "grad_norm": 91.1738052368164,
      "learning_rate": 1.0588071695453271e-05,
      "loss": 6.1626,
      "step": 12307
    },
    {
      "epoch": 0.4968914008881712,
      "grad_norm": 100.7137680053711,
      "learning_rate": 1.0586766372985919e-05,
      "loss": 5.5334,
      "step": 12308
    },
    {
      "epoch": 0.4969317723052079,
      "grad_norm": 76.5125732421875,
      "learning_rate": 1.0585461040486229e-05,
      "loss": 5.6381,
      "step": 12309
    },
    {
      "epoch": 0.49697214372224463,
      "grad_norm": 148.193115234375,
      "learning_rate": 1.0584155697976522e-05,
      "loss": 5.8902,
      "step": 12310
    },
    {
      "epoch": 0.4970125151392814,
      "grad_norm": 422.57415771484375,
      "learning_rate": 1.0582850345479116e-05,
      "loss": 4.4701,
      "step": 12311
    },
    {
      "epoch": 0.49705288655631813,
      "grad_norm": 90.15104675292969,
      "learning_rate": 1.0581544983016334e-05,
      "loss": 5.2125,
      "step": 12312
    },
    {
      "epoch": 0.49709325797335485,
      "grad_norm": 80.25477600097656,
      "learning_rate": 1.0580239610610489e-05,
      "loss": 6.1521,
      "step": 12313
    },
    {
      "epoch": 0.49713362939039163,
      "grad_norm": 102.305908203125,
      "learning_rate": 1.0578934228283903e-05,
      "loss": 2.5768,
      "step": 12314
    },
    {
      "epoch": 0.49717400080742835,
      "grad_norm": 143.14369201660156,
      "learning_rate": 1.0577628836058894e-05,
      "loss": 5.9281,
      "step": 12315
    },
    {
      "epoch": 0.4972143722244651,
      "grad_norm": 143.7633514404297,
      "learning_rate": 1.0576323433957778e-05,
      "loss": 4.6651,
      "step": 12316
    },
    {
      "epoch": 0.4972547436415018,
      "grad_norm": 150.44932556152344,
      "learning_rate": 1.0575018022002879e-05,
      "loss": 3.085,
      "step": 12317
    },
    {
      "epoch": 0.4972951150585386,
      "grad_norm": 174.0931396484375,
      "learning_rate": 1.0573712600216514e-05,
      "loss": 4.8154,
      "step": 12318
    },
    {
      "epoch": 0.4973354864755753,
      "grad_norm": 122.54463958740234,
      "learning_rate": 1.0572407168621002e-05,
      "loss": 5.1796,
      "step": 12319
    },
    {
      "epoch": 0.497375857892612,
      "grad_norm": 106.94583129882812,
      "learning_rate": 1.0571101727238664e-05,
      "loss": 3.6286,
      "step": 12320
    },
    {
      "epoch": 0.4974162293096488,
      "grad_norm": 338.9217224121094,
      "learning_rate": 1.0569796276091821e-05,
      "loss": 5.5484,
      "step": 12321
    },
    {
      "epoch": 0.4974566007266855,
      "grad_norm": 119.3533706665039,
      "learning_rate": 1.0568490815202794e-05,
      "loss": 3.9316,
      "step": 12322
    },
    {
      "epoch": 0.49749697214372224,
      "grad_norm": 98.73458862304688,
      "learning_rate": 1.0567185344593898e-05,
      "loss": 5.0731,
      "step": 12323
    },
    {
      "epoch": 0.49753734356075896,
      "grad_norm": 133.88914489746094,
      "learning_rate": 1.056587986428746e-05,
      "loss": 4.4943,
      "step": 12324
    },
    {
      "epoch": 0.49757771497779574,
      "grad_norm": 78.7278823852539,
      "learning_rate": 1.0564574374305798e-05,
      "loss": 5.111,
      "step": 12325
    },
    {
      "epoch": 0.49761808639483246,
      "grad_norm": 83.28289031982422,
      "learning_rate": 1.0563268874671233e-05,
      "loss": 5.9421,
      "step": 12326
    },
    {
      "epoch": 0.4976584578118692,
      "grad_norm": 61.622188568115234,
      "learning_rate": 1.0561963365406084e-05,
      "loss": 5.8305,
      "step": 12327
    },
    {
      "epoch": 0.49769882922890596,
      "grad_norm": 76.65454864501953,
      "learning_rate": 1.0560657846532674e-05,
      "loss": 5.1613,
      "step": 12328
    },
    {
      "epoch": 0.4977392006459427,
      "grad_norm": 116.6474609375,
      "learning_rate": 1.0559352318073321e-05,
      "loss": 4.3579,
      "step": 12329
    },
    {
      "epoch": 0.4977795720629794,
      "grad_norm": 242.7620849609375,
      "learning_rate": 1.0558046780050353e-05,
      "loss": 4.9408,
      "step": 12330
    },
    {
      "epoch": 0.4978199434800161,
      "grad_norm": 134.54078674316406,
      "learning_rate": 1.0556741232486086e-05,
      "loss": 5.2834,
      "step": 12331
    },
    {
      "epoch": 0.4978603148970529,
      "grad_norm": 212.73443603515625,
      "learning_rate": 1.0555435675402847e-05,
      "loss": 4.9366,
      "step": 12332
    },
    {
      "epoch": 0.4979006863140896,
      "grad_norm": 73.12612915039062,
      "learning_rate": 1.0554130108822952e-05,
      "loss": 4.9548,
      "step": 12333
    },
    {
      "epoch": 0.49794105773112635,
      "grad_norm": 99.08918762207031,
      "learning_rate": 1.055282453276873e-05,
      "loss": 5.2362,
      "step": 12334
    },
    {
      "epoch": 0.4979814291481631,
      "grad_norm": 83.73616027832031,
      "learning_rate": 1.0551518947262496e-05,
      "loss": 4.9571,
      "step": 12335
    },
    {
      "epoch": 0.49802180056519985,
      "grad_norm": 89.66292572021484,
      "learning_rate": 1.055021335232658e-05,
      "loss": 6.3819,
      "step": 12336
    },
    {
      "epoch": 0.49806217198223657,
      "grad_norm": 133.66488647460938,
      "learning_rate": 1.0548907747983295e-05,
      "loss": 2.7063,
      "step": 12337
    },
    {
      "epoch": 0.4981025433992733,
      "grad_norm": 139.31199645996094,
      "learning_rate": 1.0547602134254972e-05,
      "loss": 5.9313,
      "step": 12338
    },
    {
      "epoch": 0.49814291481631007,
      "grad_norm": 117.45655059814453,
      "learning_rate": 1.0546296511163931e-05,
      "loss": 6.5289,
      "step": 12339
    },
    {
      "epoch": 0.4981832862333468,
      "grad_norm": 139.66957092285156,
      "learning_rate": 1.0544990878732495e-05,
      "loss": 5.3537,
      "step": 12340
    },
    {
      "epoch": 0.4982236576503835,
      "grad_norm": 115.14785766601562,
      "learning_rate": 1.0543685236982987e-05,
      "loss": 3.9599,
      "step": 12341
    },
    {
      "epoch": 0.4982640290674203,
      "grad_norm": 399.0462951660156,
      "learning_rate": 1.0542379585937735e-05,
      "loss": 3.7868,
      "step": 12342
    },
    {
      "epoch": 0.498304400484457,
      "grad_norm": 115.67367553710938,
      "learning_rate": 1.0541073925619054e-05,
      "loss": 4.7002,
      "step": 12343
    },
    {
      "epoch": 0.49834477190149373,
      "grad_norm": 73.38674926757812,
      "learning_rate": 1.0539768256049274e-05,
      "loss": 6.4908,
      "step": 12344
    },
    {
      "epoch": 0.49838514331853045,
      "grad_norm": 69.22811889648438,
      "learning_rate": 1.0538462577250716e-05,
      "loss": 4.3561,
      "step": 12345
    },
    {
      "epoch": 0.49842551473556723,
      "grad_norm": 97.3161392211914,
      "learning_rate": 1.053715688924571e-05,
      "loss": 5.0787,
      "step": 12346
    },
    {
      "epoch": 0.49846588615260395,
      "grad_norm": 87.29629516601562,
      "learning_rate": 1.053585119205657e-05,
      "loss": 5.1119,
      "step": 12347
    },
    {
      "epoch": 0.4985062575696407,
      "grad_norm": 73.24781036376953,
      "learning_rate": 1.053454548570563e-05,
      "loss": 4.7021,
      "step": 12348
    },
    {
      "epoch": 0.49854662898667745,
      "grad_norm": 76.88216400146484,
      "learning_rate": 1.0533239770215207e-05,
      "loss": 4.271,
      "step": 12349
    },
    {
      "epoch": 0.4985870004037142,
      "grad_norm": 115.57772827148438,
      "learning_rate": 1.053193404560763e-05,
      "loss": 7.2738,
      "step": 12350
    },
    {
      "epoch": 0.4986273718207509,
      "grad_norm": 64.59748840332031,
      "learning_rate": 1.0530628311905223e-05,
      "loss": 4.5716,
      "step": 12351
    },
    {
      "epoch": 0.4986677432377876,
      "grad_norm": 92.04753112792969,
      "learning_rate": 1.0529322569130314e-05,
      "loss": 5.141,
      "step": 12352
    },
    {
      "epoch": 0.4987081146548244,
      "grad_norm": 72.1214828491211,
      "learning_rate": 1.052801681730522e-05,
      "loss": 6.0773,
      "step": 12353
    },
    {
      "epoch": 0.4987484860718611,
      "grad_norm": 85.8819580078125,
      "learning_rate": 1.0526711056452276e-05,
      "loss": 5.1195,
      "step": 12354
    },
    {
      "epoch": 0.49878885748889784,
      "grad_norm": 81.2221450805664,
      "learning_rate": 1.0525405286593802e-05,
      "loss": 5.8717,
      "step": 12355
    },
    {
      "epoch": 0.4988292289059346,
      "grad_norm": 90.62541198730469,
      "learning_rate": 1.0524099507752124e-05,
      "loss": 5.728,
      "step": 12356
    },
    {
      "epoch": 0.49886960032297134,
      "grad_norm": 96.32506561279297,
      "learning_rate": 1.0522793719949566e-05,
      "loss": 4.9411,
      "step": 12357
    },
    {
      "epoch": 0.49890997174000806,
      "grad_norm": 85.87767791748047,
      "learning_rate": 1.052148792320846e-05,
      "loss": 4.9862,
      "step": 12358
    },
    {
      "epoch": 0.49895034315704484,
      "grad_norm": 97.71939849853516,
      "learning_rate": 1.0520182117551126e-05,
      "loss": 6.2652,
      "step": 12359
    },
    {
      "epoch": 0.49899071457408156,
      "grad_norm": 87.93793487548828,
      "learning_rate": 1.0518876302999892e-05,
      "loss": 3.7347,
      "step": 12360
    },
    {
      "epoch": 0.4990310859911183,
      "grad_norm": 264.630615234375,
      "learning_rate": 1.0517570479577084e-05,
      "loss": 4.8606,
      "step": 12361
    },
    {
      "epoch": 0.499071457408155,
      "grad_norm": 101.1677017211914,
      "learning_rate": 1.0516264647305032e-05,
      "loss": 4.8385,
      "step": 12362
    },
    {
      "epoch": 0.4991118288251918,
      "grad_norm": 75.8511962890625,
      "learning_rate": 1.051495880620606e-05,
      "loss": 5.1039,
      "step": 12363
    },
    {
      "epoch": 0.4991522002422285,
      "grad_norm": 126.7692642211914,
      "learning_rate": 1.0513652956302496e-05,
      "loss": 4.8343,
      "step": 12364
    },
    {
      "epoch": 0.4991925716592652,
      "grad_norm": 154.47848510742188,
      "learning_rate": 1.0512347097616662e-05,
      "loss": 5.6692,
      "step": 12365
    },
    {
      "epoch": 0.499232943076302,
      "grad_norm": 85.4392318725586,
      "learning_rate": 1.0511041230170894e-05,
      "loss": 4.9133,
      "step": 12366
    },
    {
      "epoch": 0.4992733144933387,
      "grad_norm": 111.72361755371094,
      "learning_rate": 1.050973535398751e-05,
      "loss": 6.2546,
      "step": 12367
    },
    {
      "epoch": 0.49931368591037545,
      "grad_norm": 135.70260620117188,
      "learning_rate": 1.0508429469088844e-05,
      "loss": 4.3576,
      "step": 12368
    },
    {
      "epoch": 0.49935405732741217,
      "grad_norm": 78.3464584350586,
      "learning_rate": 1.050712357549722e-05,
      "loss": 5.1919,
      "step": 12369
    },
    {
      "epoch": 0.49939442874444895,
      "grad_norm": 87.93663024902344,
      "learning_rate": 1.0505817673234966e-05,
      "loss": 5.3922,
      "step": 12370
    },
    {
      "epoch": 0.49943480016148567,
      "grad_norm": 90.24364471435547,
      "learning_rate": 1.0504511762324412e-05,
      "loss": 5.2284,
      "step": 12371
    },
    {
      "epoch": 0.4994751715785224,
      "grad_norm": 293.6184387207031,
      "learning_rate": 1.0503205842787887e-05,
      "loss": 4.8853,
      "step": 12372
    },
    {
      "epoch": 0.49951554299555917,
      "grad_norm": 235.38694763183594,
      "learning_rate": 1.0501899914647716e-05,
      "loss": 5.3919,
      "step": 12373
    },
    {
      "epoch": 0.4995559144125959,
      "grad_norm": 74.969482421875,
      "learning_rate": 1.0500593977926228e-05,
      "loss": 4.1388,
      "step": 12374
    },
    {
      "epoch": 0.4995962858296326,
      "grad_norm": 97.7226791381836,
      "learning_rate": 1.0499288032645751e-05,
      "loss": 4.9369,
      "step": 12375
    },
    {
      "epoch": 0.49963665724666934,
      "grad_norm": 121.87741088867188,
      "learning_rate": 1.0497982078828617e-05,
      "loss": 3.8961,
      "step": 12376
    },
    {
      "epoch": 0.4996770286637061,
      "grad_norm": 100.60923767089844,
      "learning_rate": 1.049667611649715e-05,
      "loss": 6.3422,
      "step": 12377
    },
    {
      "epoch": 0.49971740008074284,
      "grad_norm": 72.0873031616211,
      "learning_rate": 1.0495370145673684e-05,
      "loss": 5.4495,
      "step": 12378
    },
    {
      "epoch": 0.49975777149777956,
      "grad_norm": 71.50676727294922,
      "learning_rate": 1.0494064166380544e-05,
      "loss": 3.7306,
      "step": 12379
    },
    {
      "epoch": 0.49979814291481633,
      "grad_norm": 104.94609832763672,
      "learning_rate": 1.0492758178640059e-05,
      "loss": 6.0301,
      "step": 12380
    },
    {
      "epoch": 0.49983851433185306,
      "grad_norm": 130.06739807128906,
      "learning_rate": 1.049145218247456e-05,
      "loss": 5.1382,
      "step": 12381
    },
    {
      "epoch": 0.4998788857488898,
      "grad_norm": 93.09393310546875,
      "learning_rate": 1.0490146177906378e-05,
      "loss": 4.0018,
      "step": 12382
    },
    {
      "epoch": 0.4999192571659265,
      "grad_norm": 74.61780548095703,
      "learning_rate": 1.0488840164957838e-05,
      "loss": 3.8698,
      "step": 12383
    },
    {
      "epoch": 0.4999596285829633,
      "grad_norm": 63.14833450317383,
      "learning_rate": 1.0487534143651278e-05,
      "loss": 5.8036,
      "step": 12384
    },
    {
      "epoch": 0.5,
      "grad_norm": 122.38134765625,
      "learning_rate": 1.0486228114009018e-05,
      "loss": 4.2467,
      "step": 12385
    },
    {
      "epoch": 0.5000403714170367,
      "grad_norm": 168.0176239013672,
      "learning_rate": 1.0484922076053396e-05,
      "loss": 5.8772,
      "step": 12386
    },
    {
      "epoch": 0.5000807428340734,
      "grad_norm": 89.46076202392578,
      "learning_rate": 1.0483616029806737e-05,
      "loss": 6.1602,
      "step": 12387
    },
    {
      "epoch": 0.5001211142511102,
      "grad_norm": 93.3130111694336,
      "learning_rate": 1.0482309975291373e-05,
      "loss": 6.6422,
      "step": 12388
    },
    {
      "epoch": 0.500161485668147,
      "grad_norm": 119.94886016845703,
      "learning_rate": 1.0481003912529634e-05,
      "loss": 5.8598,
      "step": 12389
    },
    {
      "epoch": 0.5002018570851837,
      "grad_norm": 70.91907501220703,
      "learning_rate": 1.0479697841543853e-05,
      "loss": 4.7347,
      "step": 12390
    },
    {
      "epoch": 0.5002422285022204,
      "grad_norm": 96.91693878173828,
      "learning_rate": 1.0478391762356358e-05,
      "loss": 4.2595,
      "step": 12391
    },
    {
      "epoch": 0.5002825999192572,
      "grad_norm": 83.57185363769531,
      "learning_rate": 1.047708567498948e-05,
      "loss": 4.3379,
      "step": 12392
    },
    {
      "epoch": 0.5003229713362939,
      "grad_norm": 85.05377197265625,
      "learning_rate": 1.0475779579465552e-05,
      "loss": 7.3819,
      "step": 12393
    },
    {
      "epoch": 0.5003633427533306,
      "grad_norm": 86.2922592163086,
      "learning_rate": 1.0474473475806906e-05,
      "loss": 5.5702,
      "step": 12394
    },
    {
      "epoch": 0.5004037141703673,
      "grad_norm": 119.15574645996094,
      "learning_rate": 1.047316736403587e-05,
      "loss": 5.5984,
      "step": 12395
    },
    {
      "epoch": 0.5004440855874042,
      "grad_norm": 71.97433471679688,
      "learning_rate": 1.0471861244174778e-05,
      "loss": 2.8415,
      "step": 12396
    },
    {
      "epoch": 0.5004844570044409,
      "grad_norm": 79.99470520019531,
      "learning_rate": 1.047055511624596e-05,
      "loss": 3.995,
      "step": 12397
    },
    {
      "epoch": 0.5005248284214776,
      "grad_norm": 99.15021514892578,
      "learning_rate": 1.0469248980271748e-05,
      "loss": 5.243,
      "step": 12398
    },
    {
      "epoch": 0.5005651998385143,
      "grad_norm": 109.60147094726562,
      "learning_rate": 1.0467942836274472e-05,
      "loss": 4.9713,
      "step": 12399
    },
    {
      "epoch": 0.500605571255551,
      "grad_norm": 119.96503448486328,
      "learning_rate": 1.046663668427647e-05,
      "loss": 5.5164,
      "step": 12400
    },
    {
      "epoch": 0.5006459426725878,
      "grad_norm": 104.23847961425781,
      "learning_rate": 1.0465330524300065e-05,
      "loss": 6.2643,
      "step": 12401
    },
    {
      "epoch": 0.5006863140896245,
      "grad_norm": 96.26876068115234,
      "learning_rate": 1.0464024356367602e-05,
      "loss": 5.7837,
      "step": 12402
    },
    {
      "epoch": 0.5007266855066613,
      "grad_norm": 86.9693374633789,
      "learning_rate": 1.0462718180501401e-05,
      "loss": 5.7276,
      "step": 12403
    },
    {
      "epoch": 0.500767056923698,
      "grad_norm": 103.88660430908203,
      "learning_rate": 1.0461411996723803e-05,
      "loss": 4.5397,
      "step": 12404
    },
    {
      "epoch": 0.5008074283407348,
      "grad_norm": 75.68145751953125,
      "learning_rate": 1.0460105805057134e-05,
      "loss": 5.2695,
      "step": 12405
    },
    {
      "epoch": 0.5008477997577715,
      "grad_norm": 68.33959197998047,
      "learning_rate": 1.0458799605523733e-05,
      "loss": 5.2235,
      "step": 12406
    },
    {
      "epoch": 0.5008881711748082,
      "grad_norm": 94.47962188720703,
      "learning_rate": 1.0457493398145928e-05,
      "loss": 6.0847,
      "step": 12407
    },
    {
      "epoch": 0.5009285425918449,
      "grad_norm": 71.52005767822266,
      "learning_rate": 1.0456187182946053e-05,
      "loss": 5.8828,
      "step": 12408
    },
    {
      "epoch": 0.5009689140088817,
      "grad_norm": 138.4460906982422,
      "learning_rate": 1.0454880959946445e-05,
      "loss": 4.9125,
      "step": 12409
    },
    {
      "epoch": 0.5010092854259185,
      "grad_norm": 73.44055938720703,
      "learning_rate": 1.0453574729169436e-05,
      "loss": 3.015,
      "step": 12410
    },
    {
      "epoch": 0.5010496568429552,
      "grad_norm": 120.42842102050781,
      "learning_rate": 1.0452268490637356e-05,
      "loss": 4.6704,
      "step": 12411
    },
    {
      "epoch": 0.5010900282599919,
      "grad_norm": 86.0228500366211,
      "learning_rate": 1.0450962244372542e-05,
      "loss": 5.2963,
      "step": 12412
    },
    {
      "epoch": 0.5011303996770287,
      "grad_norm": 90.21878051757812,
      "learning_rate": 1.0449655990397325e-05,
      "loss": 4.7547,
      "step": 12413
    },
    {
      "epoch": 0.5011707710940654,
      "grad_norm": 117.19104766845703,
      "learning_rate": 1.0448349728734044e-05,
      "loss": 5.4653,
      "step": 12414
    },
    {
      "epoch": 0.5012111425111021,
      "grad_norm": 86.48612976074219,
      "learning_rate": 1.0447043459405027e-05,
      "loss": 4.8318,
      "step": 12415
    },
    {
      "epoch": 0.5012515139281388,
      "grad_norm": 99.96150207519531,
      "learning_rate": 1.0445737182432611e-05,
      "loss": 4.4341,
      "step": 12416
    },
    {
      "epoch": 0.5012918853451757,
      "grad_norm": 75.51654815673828,
      "learning_rate": 1.044443089783913e-05,
      "loss": 6.6906,
      "step": 12417
    },
    {
      "epoch": 0.5013322567622124,
      "grad_norm": 107.66002655029297,
      "learning_rate": 1.044312460564692e-05,
      "loss": 3.8977,
      "step": 12418
    },
    {
      "epoch": 0.5013726281792491,
      "grad_norm": 91.91012573242188,
      "learning_rate": 1.044181830587831e-05,
      "loss": 4.8502,
      "step": 12419
    },
    {
      "epoch": 0.5014129995962858,
      "grad_norm": 82.56494140625,
      "learning_rate": 1.0440511998555642e-05,
      "loss": 5.2428,
      "step": 12420
    },
    {
      "epoch": 0.5014533710133225,
      "grad_norm": 88.09981536865234,
      "learning_rate": 1.0439205683701248e-05,
      "loss": 5.904,
      "step": 12421
    },
    {
      "epoch": 0.5014937424303593,
      "grad_norm": 107.56220245361328,
      "learning_rate": 1.0437899361337463e-05,
      "loss": 5.1215,
      "step": 12422
    },
    {
      "epoch": 0.501534113847396,
      "grad_norm": 92.47300720214844,
      "learning_rate": 1.0436593031486617e-05,
      "loss": 5.909,
      "step": 12423
    },
    {
      "epoch": 0.5015744852644328,
      "grad_norm": 75.51766967773438,
      "learning_rate": 1.0435286694171054e-05,
      "loss": 3.6256,
      "step": 12424
    },
    {
      "epoch": 0.5016148566814695,
      "grad_norm": 80.31995391845703,
      "learning_rate": 1.0433980349413102e-05,
      "loss": 6.0743,
      "step": 12425
    },
    {
      "epoch": 0.5016552280985063,
      "grad_norm": 111.09959411621094,
      "learning_rate": 1.0432673997235103e-05,
      "loss": 5.6316,
      "step": 12426
    },
    {
      "epoch": 0.501695599515543,
      "grad_norm": 94.15933227539062,
      "learning_rate": 1.0431367637659387e-05,
      "loss": 5.7646,
      "step": 12427
    },
    {
      "epoch": 0.5017359709325797,
      "grad_norm": 95.68907165527344,
      "learning_rate": 1.043006127070829e-05,
      "loss": 5.3436,
      "step": 12428
    },
    {
      "epoch": 0.5017763423496164,
      "grad_norm": 238.489501953125,
      "learning_rate": 1.042875489640415e-05,
      "loss": 5.4595,
      "step": 12429
    },
    {
      "epoch": 0.5018167137666532,
      "grad_norm": 102.56136322021484,
      "learning_rate": 1.0427448514769306e-05,
      "loss": 4.5138,
      "step": 12430
    },
    {
      "epoch": 0.50185708518369,
      "grad_norm": 90.8545913696289,
      "learning_rate": 1.0426142125826087e-05,
      "loss": 5.9093,
      "step": 12431
    },
    {
      "epoch": 0.5018974566007267,
      "grad_norm": 126.65718078613281,
      "learning_rate": 1.0424835729596836e-05,
      "loss": 4.8125,
      "step": 12432
    },
    {
      "epoch": 0.5019378280177634,
      "grad_norm": 81.64298248291016,
      "learning_rate": 1.0423529326103884e-05,
      "loss": 3.988,
      "step": 12433
    },
    {
      "epoch": 0.5019781994348002,
      "grad_norm": 104.82472229003906,
      "learning_rate": 1.0422222915369568e-05,
      "loss": 4.7167,
      "step": 12434
    },
    {
      "epoch": 0.5020185708518369,
      "grad_norm": 77.87156677246094,
      "learning_rate": 1.042091649741623e-05,
      "loss": 4.2509,
      "step": 12435
    },
    {
      "epoch": 0.5020589422688736,
      "grad_norm": 73.42431640625,
      "learning_rate": 1.04196100722662e-05,
      "loss": 5.5114,
      "step": 12436
    },
    {
      "epoch": 0.5020993136859103,
      "grad_norm": 88.12071990966797,
      "learning_rate": 1.0418303639941818e-05,
      "loss": 3.9214,
      "step": 12437
    },
    {
      "epoch": 0.5021396851029472,
      "grad_norm": 73.9209976196289,
      "learning_rate": 1.041699720046542e-05,
      "loss": 5.2952,
      "step": 12438
    },
    {
      "epoch": 0.5021800565199839,
      "grad_norm": 325.9375305175781,
      "learning_rate": 1.0415690753859342e-05,
      "loss": 6.6786,
      "step": 12439
    },
    {
      "epoch": 0.5022204279370206,
      "grad_norm": 75.02355194091797,
      "learning_rate": 1.0414384300145928e-05,
      "loss": 6.3938,
      "step": 12440
    },
    {
      "epoch": 0.5022607993540573,
      "grad_norm": 138.67141723632812,
      "learning_rate": 1.0413077839347505e-05,
      "loss": 6.3417,
      "step": 12441
    },
    {
      "epoch": 0.502301170771094,
      "grad_norm": 118.36157989501953,
      "learning_rate": 1.0411771371486421e-05,
      "loss": 4.6972,
      "step": 12442
    },
    {
      "epoch": 0.5023415421881308,
      "grad_norm": 106.60838317871094,
      "learning_rate": 1.0410464896585003e-05,
      "loss": 4.8659,
      "step": 12443
    },
    {
      "epoch": 0.5023819136051676,
      "grad_norm": 57.93214416503906,
      "learning_rate": 1.0409158414665596e-05,
      "loss": 4.7349,
      "step": 12444
    },
    {
      "epoch": 0.5024222850222043,
      "grad_norm": 105.20451354980469,
      "learning_rate": 1.0407851925750536e-05,
      "loss": 3.6152,
      "step": 12445
    },
    {
      "epoch": 0.502462656439241,
      "grad_norm": 104.14430236816406,
      "learning_rate": 1.040654542986216e-05,
      "loss": 5.1491,
      "step": 12446
    },
    {
      "epoch": 0.5025030278562778,
      "grad_norm": 79.92375183105469,
      "learning_rate": 1.0405238927022807e-05,
      "loss": 4.9837,
      "step": 12447
    },
    {
      "epoch": 0.5025433992733145,
      "grad_norm": 118.09173583984375,
      "learning_rate": 1.0403932417254811e-05,
      "loss": 5.9635,
      "step": 12448
    },
    {
      "epoch": 0.5025837706903512,
      "grad_norm": 100.69623565673828,
      "learning_rate": 1.0402625900580516e-05,
      "loss": 5.6329,
      "step": 12449
    },
    {
      "epoch": 0.5026241421073879,
      "grad_norm": 81.7964096069336,
      "learning_rate": 1.0401319377022261e-05,
      "loss": 6.3385,
      "step": 12450
    },
    {
      "epoch": 0.5026645135244248,
      "grad_norm": 72.44932556152344,
      "learning_rate": 1.0400012846602378e-05,
      "loss": 4.2264,
      "step": 12451
    },
    {
      "epoch": 0.5027048849414615,
      "grad_norm": 83.34280395507812,
      "learning_rate": 1.0398706309343214e-05,
      "loss": 5.1905,
      "step": 12452
    },
    {
      "epoch": 0.5027452563584982,
      "grad_norm": 119.19258117675781,
      "learning_rate": 1.03973997652671e-05,
      "loss": 5.1236,
      "step": 12453
    },
    {
      "epoch": 0.5027856277755349,
      "grad_norm": 98.06712341308594,
      "learning_rate": 1.039609321439638e-05,
      "loss": 5.5958,
      "step": 12454
    },
    {
      "epoch": 0.5028259991925716,
      "grad_norm": 73.71276092529297,
      "learning_rate": 1.0394786656753388e-05,
      "loss": 4.9301,
      "step": 12455
    },
    {
      "epoch": 0.5028663706096084,
      "grad_norm": 103.39414978027344,
      "learning_rate": 1.0393480092360469e-05,
      "loss": 4.276,
      "step": 12456
    },
    {
      "epoch": 0.5029067420266451,
      "grad_norm": 164.56390380859375,
      "learning_rate": 1.0392173521239957e-05,
      "loss": 5.6362,
      "step": 12457
    },
    {
      "epoch": 0.5029471134436819,
      "grad_norm": 76.26854705810547,
      "learning_rate": 1.0390866943414193e-05,
      "loss": 6.182,
      "step": 12458
    },
    {
      "epoch": 0.5029874848607186,
      "grad_norm": 102.85967254638672,
      "learning_rate": 1.0389560358905519e-05,
      "loss": 4.8894,
      "step": 12459
    },
    {
      "epoch": 0.5030278562777554,
      "grad_norm": 124.72345733642578,
      "learning_rate": 1.0388253767736272e-05,
      "loss": 4.7147,
      "step": 12460
    },
    {
      "epoch": 0.5030682276947921,
      "grad_norm": 141.05545043945312,
      "learning_rate": 1.0386947169928791e-05,
      "loss": 5.2085,
      "step": 12461
    },
    {
      "epoch": 0.5031085991118288,
      "grad_norm": 74.03477478027344,
      "learning_rate": 1.0385640565505422e-05,
      "loss": 5.685,
      "step": 12462
    },
    {
      "epoch": 0.5031489705288655,
      "grad_norm": 86.04290771484375,
      "learning_rate": 1.0384333954488492e-05,
      "loss": 6.76,
      "step": 12463
    },
    {
      "epoch": 0.5031893419459023,
      "grad_norm": 72.77583312988281,
      "learning_rate": 1.0383027336900356e-05,
      "loss": 6.3497,
      "step": 12464
    },
    {
      "epoch": 0.5032297133629391,
      "grad_norm": 83.581298828125,
      "learning_rate": 1.0381720712763343e-05,
      "loss": 7.1652,
      "step": 12465
    },
    {
      "epoch": 0.5032700847799758,
      "grad_norm": 117.06063842773438,
      "learning_rate": 1.03804140820998e-05,
      "loss": 3.7654,
      "step": 12466
    },
    {
      "epoch": 0.5033104561970125,
      "grad_norm": 133.58824157714844,
      "learning_rate": 1.0379107444932061e-05,
      "loss": 5.5097,
      "step": 12467
    },
    {
      "epoch": 0.5033508276140493,
      "grad_norm": 110.50960540771484,
      "learning_rate": 1.0377800801282471e-05,
      "loss": 6.4702,
      "step": 12468
    },
    {
      "epoch": 0.503391199031086,
      "grad_norm": 98.94878387451172,
      "learning_rate": 1.0376494151173368e-05,
      "loss": 5.3982,
      "step": 12469
    },
    {
      "epoch": 0.5034315704481227,
      "grad_norm": 140.98904418945312,
      "learning_rate": 1.03751874946271e-05,
      "loss": 5.1142,
      "step": 12470
    },
    {
      "epoch": 0.5034719418651594,
      "grad_norm": 79.04119873046875,
      "learning_rate": 1.0373880831665997e-05,
      "loss": 3.8175,
      "step": 12471
    },
    {
      "epoch": 0.5035123132821963,
      "grad_norm": 72.5232162475586,
      "learning_rate": 1.0372574162312406e-05,
      "loss": 4.8518,
      "step": 12472
    },
    {
      "epoch": 0.503552684699233,
      "grad_norm": 149.03538513183594,
      "learning_rate": 1.0371267486588666e-05,
      "loss": 5.4089,
      "step": 12473
    },
    {
      "epoch": 0.5035930561162697,
      "grad_norm": 124.33946990966797,
      "learning_rate": 1.036996080451712e-05,
      "loss": 3.7388,
      "step": 12474
    },
    {
      "epoch": 0.5036334275333064,
      "grad_norm": 85.56742095947266,
      "learning_rate": 1.036865411612011e-05,
      "loss": 5.9064,
      "step": 12475
    },
    {
      "epoch": 0.5036737989503431,
      "grad_norm": 93.5898208618164,
      "learning_rate": 1.0367347421419974e-05,
      "loss": 5.1857,
      "step": 12476
    },
    {
      "epoch": 0.5037141703673799,
      "grad_norm": 116.76666259765625,
      "learning_rate": 1.0366040720439052e-05,
      "loss": 4.9907,
      "step": 12477
    },
    {
      "epoch": 0.5037545417844166,
      "grad_norm": 101.50263977050781,
      "learning_rate": 1.0364734013199691e-05,
      "loss": 4.1377,
      "step": 12478
    },
    {
      "epoch": 0.5037949132014534,
      "grad_norm": 176.85601806640625,
      "learning_rate": 1.036342729972423e-05,
      "loss": 5.1639,
      "step": 12479
    },
    {
      "epoch": 0.5038352846184901,
      "grad_norm": 111.6061782836914,
      "learning_rate": 1.0362120580035011e-05,
      "loss": 4.8655,
      "step": 12480
    },
    {
      "epoch": 0.5038756560355269,
      "grad_norm": 100.39913177490234,
      "learning_rate": 1.0360813854154375e-05,
      "loss": 5.1804,
      "step": 12481
    },
    {
      "epoch": 0.5039160274525636,
      "grad_norm": 99.96817779541016,
      "learning_rate": 1.0359507122104667e-05,
      "loss": 4.7641,
      "step": 12482
    },
    {
      "epoch": 0.5039563988696003,
      "grad_norm": 88.82640838623047,
      "learning_rate": 1.0358200383908224e-05,
      "loss": 5.4097,
      "step": 12483
    },
    {
      "epoch": 0.503996770286637,
      "grad_norm": 82.36219787597656,
      "learning_rate": 1.0356893639587394e-05,
      "loss": 5.3609,
      "step": 12484
    },
    {
      "epoch": 0.5040371417036738,
      "grad_norm": 70.75988006591797,
      "learning_rate": 1.0355586889164511e-05,
      "loss": 5.3405,
      "step": 12485
    },
    {
      "epoch": 0.5040775131207106,
      "grad_norm": 86.36199951171875,
      "learning_rate": 1.0354280132661926e-05,
      "loss": 5.5242,
      "step": 12486
    },
    {
      "epoch": 0.5041178845377473,
      "grad_norm": 77.20857238769531,
      "learning_rate": 1.0352973370101978e-05,
      "loss": 4.3671,
      "step": 12487
    },
    {
      "epoch": 0.504158255954784,
      "grad_norm": 95.75900268554688,
      "learning_rate": 1.0351666601507008e-05,
      "loss": 6.1121,
      "step": 12488
    },
    {
      "epoch": 0.5041986273718208,
      "grad_norm": 73.86434936523438,
      "learning_rate": 1.0350359826899362e-05,
      "loss": 4.8905,
      "step": 12489
    },
    {
      "epoch": 0.5042389987888575,
      "grad_norm": 97.83943939208984,
      "learning_rate": 1.0349053046301381e-05,
      "loss": 4.4367,
      "step": 12490
    },
    {
      "epoch": 0.5042793702058942,
      "grad_norm": 80.41475677490234,
      "learning_rate": 1.0347746259735408e-05,
      "loss": 4.8391,
      "step": 12491
    },
    {
      "epoch": 0.5043197416229309,
      "grad_norm": 93.72930908203125,
      "learning_rate": 1.0346439467223785e-05,
      "loss": 3.2113,
      "step": 12492
    },
    {
      "epoch": 0.5043601130399678,
      "grad_norm": 89.95237731933594,
      "learning_rate": 1.0345132668788856e-05,
      "loss": 5.7826,
      "step": 12493
    },
    {
      "epoch": 0.5044004844570045,
      "grad_norm": 192.57070922851562,
      "learning_rate": 1.0343825864452968e-05,
      "loss": 4.0832,
      "step": 12494
    },
    {
      "epoch": 0.5044408558740412,
      "grad_norm": 94.4679183959961,
      "learning_rate": 1.0342519054238455e-05,
      "loss": 4.8352,
      "step": 12495
    },
    {
      "epoch": 0.5044812272910779,
      "grad_norm": 78.28472137451172,
      "learning_rate": 1.034121223816767e-05,
      "loss": 2.8482,
      "step": 12496
    },
    {
      "epoch": 0.5045215987081146,
      "grad_norm": 98.28096771240234,
      "learning_rate": 1.033990541626295e-05,
      "loss": 5.0267,
      "step": 12497
    },
    {
      "epoch": 0.5045619701251514,
      "grad_norm": 82.23117065429688,
      "learning_rate": 1.0338598588546643e-05,
      "loss": 6.9156,
      "step": 12498
    },
    {
      "epoch": 0.5046023415421881,
      "grad_norm": 131.261474609375,
      "learning_rate": 1.0337291755041089e-05,
      "loss": 5.6051,
      "step": 12499
    },
    {
      "epoch": 0.5046427129592249,
      "grad_norm": 95.7345199584961,
      "learning_rate": 1.0335984915768636e-05,
      "loss": 4.0894,
      "step": 12500
    },
    {
      "epoch": 0.5046830843762616,
      "grad_norm": 102.72830963134766,
      "learning_rate": 1.0334678070751623e-05,
      "loss": 4.5708,
      "step": 12501
    },
    {
      "epoch": 0.5047234557932984,
      "grad_norm": 92.71964263916016,
      "learning_rate": 1.0333371220012397e-05,
      "loss": 5.2375,
      "step": 12502
    },
    {
      "epoch": 0.5047638272103351,
      "grad_norm": 116.6889877319336,
      "learning_rate": 1.0332064363573304e-05,
      "loss": 6.6447,
      "step": 12503
    },
    {
      "epoch": 0.5048041986273718,
      "grad_norm": 119.14472961425781,
      "learning_rate": 1.0330757501456685e-05,
      "loss": 5.4328,
      "step": 12504
    },
    {
      "epoch": 0.5048445700444085,
      "grad_norm": 233.4661865234375,
      "learning_rate": 1.0329450633684884e-05,
      "loss": 4.2842,
      "step": 12505
    },
    {
      "epoch": 0.5048849414614452,
      "grad_norm": 89.40663146972656,
      "learning_rate": 1.0328143760280244e-05,
      "loss": 5.1697,
      "step": 12506
    },
    {
      "epoch": 0.5049253128784821,
      "grad_norm": 82.51456451416016,
      "learning_rate": 1.0326836881265114e-05,
      "loss": 3.3364,
      "step": 12507
    },
    {
      "epoch": 0.5049656842955188,
      "grad_norm": 81.10655212402344,
      "learning_rate": 1.032552999666184e-05,
      "loss": 5.0579,
      "step": 12508
    },
    {
      "epoch": 0.5050060557125555,
      "grad_norm": 106.01898956298828,
      "learning_rate": 1.032422310649276e-05,
      "loss": 4.6823,
      "step": 12509
    },
    {
      "epoch": 0.5050464271295922,
      "grad_norm": 79.4302749633789,
      "learning_rate": 1.0322916210780223e-05,
      "loss": 5.2932,
      "step": 12510
    },
    {
      "epoch": 0.505086798546629,
      "grad_norm": 77.1211166381836,
      "learning_rate": 1.0321609309546572e-05,
      "loss": 5.6717,
      "step": 12511
    },
    {
      "epoch": 0.5051271699636657,
      "grad_norm": 94.36199188232422,
      "learning_rate": 1.0320302402814154e-05,
      "loss": 4.1269,
      "step": 12512
    },
    {
      "epoch": 0.5051675413807024,
      "grad_norm": 90.18197631835938,
      "learning_rate": 1.031899549060531e-05,
      "loss": 4.5988,
      "step": 12513
    },
    {
      "epoch": 0.5052079127977392,
      "grad_norm": 91.38060760498047,
      "learning_rate": 1.0317688572942393e-05,
      "loss": 5.6235,
      "step": 12514
    },
    {
      "epoch": 0.505248284214776,
      "grad_norm": 90.48583984375,
      "learning_rate": 1.031638164984774e-05,
      "loss": 6.0108,
      "step": 12515
    },
    {
      "epoch": 0.5052886556318127,
      "grad_norm": 87.93815612792969,
      "learning_rate": 1.0315074721343698e-05,
      "loss": 5.641,
      "step": 12516
    },
    {
      "epoch": 0.5053290270488494,
      "grad_norm": 86.93851470947266,
      "learning_rate": 1.0313767787452615e-05,
      "loss": 4.4006,
      "step": 12517
    },
    {
      "epoch": 0.5053693984658861,
      "grad_norm": 117.9226303100586,
      "learning_rate": 1.0312460848196837e-05,
      "loss": 5.0435,
      "step": 12518
    },
    {
      "epoch": 0.5054097698829229,
      "grad_norm": 133.9847412109375,
      "learning_rate": 1.0311153903598707e-05,
      "loss": 4.9612,
      "step": 12519
    },
    {
      "epoch": 0.5054501412999596,
      "grad_norm": 115.45951080322266,
      "learning_rate": 1.0309846953680574e-05,
      "loss": 5.746,
      "step": 12520
    },
    {
      "epoch": 0.5054905127169964,
      "grad_norm": 91.54701232910156,
      "learning_rate": 1.0308539998464779e-05,
      "loss": 4.8857,
      "step": 12521
    },
    {
      "epoch": 0.5055308841340331,
      "grad_norm": 175.14247131347656,
      "learning_rate": 1.0307233037973673e-05,
      "loss": 5.9432,
      "step": 12522
    },
    {
      "epoch": 0.5055712555510699,
      "grad_norm": 147.0518341064453,
      "learning_rate": 1.0305926072229598e-05,
      "loss": 7.0506,
      "step": 12523
    },
    {
      "epoch": 0.5056116269681066,
      "grad_norm": 132.92715454101562,
      "learning_rate": 1.0304619101254903e-05,
      "loss": 5.6582,
      "step": 12524
    },
    {
      "epoch": 0.5056519983851433,
      "grad_norm": 141.49200439453125,
      "learning_rate": 1.0303312125071929e-05,
      "loss": 6.4953,
      "step": 12525
    },
    {
      "epoch": 0.50569236980218,
      "grad_norm": 132.96768188476562,
      "learning_rate": 1.0302005143703028e-05,
      "loss": 3.8257,
      "step": 12526
    },
    {
      "epoch": 0.5057327412192167,
      "grad_norm": 102.60173797607422,
      "learning_rate": 1.0300698157170544e-05,
      "loss": 5.3743,
      "step": 12527
    },
    {
      "epoch": 0.5057731126362536,
      "grad_norm": 132.29786682128906,
      "learning_rate": 1.0299391165496826e-05,
      "loss": 4.7171,
      "step": 12528
    },
    {
      "epoch": 0.5058134840532903,
      "grad_norm": 84.0846939086914,
      "learning_rate": 1.0298084168704215e-05,
      "loss": 5.0076,
      "step": 12529
    },
    {
      "epoch": 0.505853855470327,
      "grad_norm": 138.449951171875,
      "learning_rate": 1.0296777166815063e-05,
      "loss": 4.1958,
      "step": 12530
    },
    {
      "epoch": 0.5058942268873637,
      "grad_norm": 90.70191955566406,
      "learning_rate": 1.0295470159851715e-05,
      "loss": 5.2729,
      "step": 12531
    },
    {
      "epoch": 0.5059345983044005,
      "grad_norm": 88.88806915283203,
      "learning_rate": 1.0294163147836515e-05,
      "loss": 6.0681,
      "step": 12532
    },
    {
      "epoch": 0.5059749697214372,
      "grad_norm": 86.30302429199219,
      "learning_rate": 1.0292856130791813e-05,
      "loss": 7.4008,
      "step": 12533
    },
    {
      "epoch": 0.5060153411384739,
      "grad_norm": 83.15396118164062,
      "learning_rate": 1.0291549108739957e-05,
      "loss": 5.5143,
      "step": 12534
    },
    {
      "epoch": 0.5060557125555107,
      "grad_norm": 102.2159194946289,
      "learning_rate": 1.029024208170329e-05,
      "loss": 4.5828,
      "step": 12535
    },
    {
      "epoch": 0.5060960839725475,
      "grad_norm": 89.18527221679688,
      "learning_rate": 1.0288935049704159e-05,
      "loss": 5.8119,
      "step": 12536
    },
    {
      "epoch": 0.5061364553895842,
      "grad_norm": 133.40155029296875,
      "learning_rate": 1.0287628012764916e-05,
      "loss": 4.9115,
      "step": 12537
    },
    {
      "epoch": 0.5061768268066209,
      "grad_norm": 158.2978973388672,
      "learning_rate": 1.0286320970907906e-05,
      "loss": 5.8539,
      "step": 12538
    },
    {
      "epoch": 0.5062171982236576,
      "grad_norm": 85.51264190673828,
      "learning_rate": 1.0285013924155473e-05,
      "loss": 4.8294,
      "step": 12539
    },
    {
      "epoch": 0.5062575696406943,
      "grad_norm": 76.53141021728516,
      "learning_rate": 1.0283706872529973e-05,
      "loss": 4.8471,
      "step": 12540
    },
    {
      "epoch": 0.5062979410577311,
      "grad_norm": 121.91447448730469,
      "learning_rate": 1.0282399816053743e-05,
      "loss": 5.4701,
      "step": 12541
    },
    {
      "epoch": 0.5063383124747679,
      "grad_norm": 189.04568481445312,
      "learning_rate": 1.028109275474914e-05,
      "loss": 4.3208,
      "step": 12542
    },
    {
      "epoch": 0.5063786838918046,
      "grad_norm": 94.05233764648438,
      "learning_rate": 1.0279785688638504e-05,
      "loss": 5.071,
      "step": 12543
    },
    {
      "epoch": 0.5064190553088413,
      "grad_norm": 88.45707702636719,
      "learning_rate": 1.0278478617744188e-05,
      "loss": 5.1729,
      "step": 12544
    },
    {
      "epoch": 0.5064594267258781,
      "grad_norm": 62.77225112915039,
      "learning_rate": 1.0277171542088536e-05,
      "loss": 6.1661,
      "step": 12545
    },
    {
      "epoch": 0.5064997981429148,
      "grad_norm": 106.94255828857422,
      "learning_rate": 1.0275864461693896e-05,
      "loss": 6.0614,
      "step": 12546
    },
    {
      "epoch": 0.5065401695599515,
      "grad_norm": 92.60466766357422,
      "learning_rate": 1.0274557376582619e-05,
      "loss": 5.9909,
      "step": 12547
    },
    {
      "epoch": 0.5065805409769883,
      "grad_norm": 100.54035949707031,
      "learning_rate": 1.0273250286777057e-05,
      "loss": 5.9727,
      "step": 12548
    },
    {
      "epoch": 0.5066209123940251,
      "grad_norm": 141.9760284423828,
      "learning_rate": 1.0271943192299547e-05,
      "loss": 4.4359,
      "step": 12549
    },
    {
      "epoch": 0.5066612838110618,
      "grad_norm": 84.91121673583984,
      "learning_rate": 1.0270636093172448e-05,
      "loss": 4.2235,
      "step": 12550
    },
    {
      "epoch": 0.5067016552280985,
      "grad_norm": 69.1911849975586,
      "learning_rate": 1.0269328989418101e-05,
      "loss": 4.8169,
      "step": 12551
    },
    {
      "epoch": 0.5067420266451352,
      "grad_norm": 71.93565368652344,
      "learning_rate": 1.0268021881058858e-05,
      "loss": 4.2814,
      "step": 12552
    },
    {
      "epoch": 0.506782398062172,
      "grad_norm": 111.64977264404297,
      "learning_rate": 1.0266714768117067e-05,
      "loss": 4.3654,
      "step": 12553
    },
    {
      "epoch": 0.5068227694792087,
      "grad_norm": 86.52220153808594,
      "learning_rate": 1.0265407650615078e-05,
      "loss": 4.6209,
      "step": 12554
    },
    {
      "epoch": 0.5068631408962455,
      "grad_norm": 88.56787872314453,
      "learning_rate": 1.0264100528575234e-05,
      "loss": 6.3659,
      "step": 12555
    },
    {
      "epoch": 0.5069035123132822,
      "grad_norm": 95.99879455566406,
      "learning_rate": 1.0262793402019889e-05,
      "loss": 4.9401,
      "step": 12556
    },
    {
      "epoch": 0.506943883730319,
      "grad_norm": 92.86264038085938,
      "learning_rate": 1.026148627097139e-05,
      "loss": 4.9141,
      "step": 12557
    },
    {
      "epoch": 0.5069842551473557,
      "grad_norm": 100.94066619873047,
      "learning_rate": 1.026017913545209e-05,
      "loss": 4.6523,
      "step": 12558
    },
    {
      "epoch": 0.5070246265643924,
      "grad_norm": 98.2398452758789,
      "learning_rate": 1.025887199548433e-05,
      "loss": 6.0063,
      "step": 12559
    },
    {
      "epoch": 0.5070649979814291,
      "grad_norm": 130.81524658203125,
      "learning_rate": 1.0257564851090467e-05,
      "loss": 5.4625,
      "step": 12560
    },
    {
      "epoch": 0.5071053693984658,
      "grad_norm": 156.4945068359375,
      "learning_rate": 1.0256257702292844e-05,
      "loss": 4.9573,
      "step": 12561
    },
    {
      "epoch": 0.5071457408155027,
      "grad_norm": 70.15953826904297,
      "learning_rate": 1.0254950549113815e-05,
      "loss": 5.1308,
      "step": 12562
    },
    {
      "epoch": 0.5071861122325394,
      "grad_norm": 87.44303894042969,
      "learning_rate": 1.0253643391575726e-05,
      "loss": 2.8488,
      "step": 12563
    },
    {
      "epoch": 0.5072264836495761,
      "grad_norm": 130.2756805419922,
      "learning_rate": 1.0252336229700926e-05,
      "loss": 5.9808,
      "step": 12564
    },
    {
      "epoch": 0.5072668550666128,
      "grad_norm": 93.2558822631836,
      "learning_rate": 1.0251029063511765e-05,
      "loss": 6.5914,
      "step": 12565
    },
    {
      "epoch": 0.5073072264836496,
      "grad_norm": 86.41358947753906,
      "learning_rate": 1.0249721893030596e-05,
      "loss": 5.6899,
      "step": 12566
    },
    {
      "epoch": 0.5073475979006863,
      "grad_norm": 109.7022705078125,
      "learning_rate": 1.0248414718279762e-05,
      "loss": 3.7607,
      "step": 12567
    },
    {
      "epoch": 0.507387969317723,
      "grad_norm": 92.59844970703125,
      "learning_rate": 1.024710753928162e-05,
      "loss": 7.2285,
      "step": 12568
    },
    {
      "epoch": 0.5074283407347598,
      "grad_norm": 119.88106536865234,
      "learning_rate": 1.0245800356058515e-05,
      "loss": 4.9553,
      "step": 12569
    },
    {
      "epoch": 0.5074687121517966,
      "grad_norm": 68.96285247802734,
      "learning_rate": 1.0244493168632799e-05,
      "loss": 4.6577,
      "step": 12570
    },
    {
      "epoch": 0.5075090835688333,
      "grad_norm": 97.5627670288086,
      "learning_rate": 1.0243185977026819e-05,
      "loss": 5.3051,
      "step": 12571
    },
    {
      "epoch": 0.50754945498587,
      "grad_norm": 70.0755386352539,
      "learning_rate": 1.0241878781262928e-05,
      "loss": 5.0277,
      "step": 12572
    },
    {
      "epoch": 0.5075898264029067,
      "grad_norm": 99.09210205078125,
      "learning_rate": 1.0240571581363475e-05,
      "loss": 3.9565,
      "step": 12573
    },
    {
      "epoch": 0.5076301978199435,
      "grad_norm": 102.54240417480469,
      "learning_rate": 1.0239264377350808e-05,
      "loss": 4.9684,
      "step": 12574
    },
    {
      "epoch": 0.5076705692369802,
      "grad_norm": 83.20569610595703,
      "learning_rate": 1.023795716924728e-05,
      "loss": 5.3451,
      "step": 12575
    },
    {
      "epoch": 0.507710940654017,
      "grad_norm": 101.1772232055664,
      "learning_rate": 1.0236649957075238e-05,
      "loss": 4.2903,
      "step": 12576
    },
    {
      "epoch": 0.5077513120710537,
      "grad_norm": 96.26437377929688,
      "learning_rate": 1.0235342740857035e-05,
      "loss": 5.5797,
      "step": 12577
    },
    {
      "epoch": 0.5077916834880905,
      "grad_norm": 173.3466033935547,
      "learning_rate": 1.0234035520615023e-05,
      "loss": 4.1868,
      "step": 12578
    },
    {
      "epoch": 0.5078320549051272,
      "grad_norm": 72.92229461669922,
      "learning_rate": 1.0232728296371547e-05,
      "loss": 5.3043,
      "step": 12579
    },
    {
      "epoch": 0.5078724263221639,
      "grad_norm": 87.09257507324219,
      "learning_rate": 1.0231421068148963e-05,
      "loss": 5.1489,
      "step": 12580
    },
    {
      "epoch": 0.5079127977392006,
      "grad_norm": 117.40303802490234,
      "learning_rate": 1.0230113835969619e-05,
      "loss": 6.3747,
      "step": 12581
    },
    {
      "epoch": 0.5079531691562373,
      "grad_norm": 86.67627716064453,
      "learning_rate": 1.0228806599855868e-05,
      "loss": 4.9051,
      "step": 12582
    },
    {
      "epoch": 0.5079935405732742,
      "grad_norm": 103.69251251220703,
      "learning_rate": 1.0227499359830054e-05,
      "loss": 4.2549,
      "step": 12583
    },
    {
      "epoch": 0.5080339119903109,
      "grad_norm": 68.4201889038086,
      "learning_rate": 1.0226192115914535e-05,
      "loss": 4.6167,
      "step": 12584
    },
    {
      "epoch": 0.5080742834073476,
      "grad_norm": 74.09211730957031,
      "learning_rate": 1.0224884868131658e-05,
      "loss": 5.1234,
      "step": 12585
    },
    {
      "epoch": 0.5081146548243843,
      "grad_norm": 383.2873840332031,
      "learning_rate": 1.0223577616503773e-05,
      "loss": 5.2301,
      "step": 12586
    },
    {
      "epoch": 0.5081550262414211,
      "grad_norm": 83.85932159423828,
      "learning_rate": 1.0222270361053233e-05,
      "loss": 5.1343,
      "step": 12587
    },
    {
      "epoch": 0.5081953976584578,
      "grad_norm": 81.94287109375,
      "learning_rate": 1.0220963101802393e-05,
      "loss": 5.1582,
      "step": 12588
    },
    {
      "epoch": 0.5082357690754945,
      "grad_norm": 91.20987701416016,
      "learning_rate": 1.0219655838773597e-05,
      "loss": 5.3885,
      "step": 12589
    },
    {
      "epoch": 0.5082761404925313,
      "grad_norm": 107.4759750366211,
      "learning_rate": 1.02183485719892e-05,
      "loss": 4.9926,
      "step": 12590
    },
    {
      "epoch": 0.5083165119095681,
      "grad_norm": 157.79327392578125,
      "learning_rate": 1.0217041301471551e-05,
      "loss": 5.2106,
      "step": 12591
    },
    {
      "epoch": 0.5083568833266048,
      "grad_norm": 169.09068298339844,
      "learning_rate": 1.0215734027243005e-05,
      "loss": 4.3146,
      "step": 12592
    },
    {
      "epoch": 0.5083972547436415,
      "grad_norm": 137.72445678710938,
      "learning_rate": 1.0214426749325907e-05,
      "loss": 5.9683,
      "step": 12593
    },
    {
      "epoch": 0.5084376261606782,
      "grad_norm": 106.02009582519531,
      "learning_rate": 1.0213119467742618e-05,
      "loss": 6.6819,
      "step": 12594
    },
    {
      "epoch": 0.508477997577715,
      "grad_norm": 115.46665954589844,
      "learning_rate": 1.0211812182515479e-05,
      "loss": 5.8261,
      "step": 12595
    },
    {
      "epoch": 0.5085183689947517,
      "grad_norm": 68.11557006835938,
      "learning_rate": 1.0210504893666847e-05,
      "loss": 5.5812,
      "step": 12596
    },
    {
      "epoch": 0.5085587404117885,
      "grad_norm": 99.47942352294922,
      "learning_rate": 1.0209197601219072e-05,
      "loss": 5.833,
      "step": 12597
    },
    {
      "epoch": 0.5085991118288252,
      "grad_norm": 128.22256469726562,
      "learning_rate": 1.020789030519451e-05,
      "loss": 6.2649,
      "step": 12598
    },
    {
      "epoch": 0.508639483245862,
      "grad_norm": 132.2040557861328,
      "learning_rate": 1.0206583005615507e-05,
      "loss": 5.0512,
      "step": 12599
    },
    {
      "epoch": 0.5086798546628987,
      "grad_norm": 115.14161682128906,
      "learning_rate": 1.0205275702504419e-05,
      "loss": 5.8319,
      "step": 12600
    },
    {
      "epoch": 0.5087202260799354,
      "grad_norm": 113.38208770751953,
      "learning_rate": 1.0203968395883592e-05,
      "loss": 3.7879,
      "step": 12601
    },
    {
      "epoch": 0.5087605974969721,
      "grad_norm": 84.83937072753906,
      "learning_rate": 1.0202661085775386e-05,
      "loss": 5.5536,
      "step": 12602
    },
    {
      "epoch": 0.5088009689140088,
      "grad_norm": 75.45215606689453,
      "learning_rate": 1.0201353772202145e-05,
      "loss": 5.0237,
      "step": 12603
    },
    {
      "epoch": 0.5088413403310457,
      "grad_norm": 76.07939147949219,
      "learning_rate": 1.0200046455186226e-05,
      "loss": 4.8441,
      "step": 12604
    },
    {
      "epoch": 0.5088817117480824,
      "grad_norm": 110.95978546142578,
      "learning_rate": 1.019873913474998e-05,
      "loss": 6.0652,
      "step": 12605
    },
    {
      "epoch": 0.5089220831651191,
      "grad_norm": 113.80734252929688,
      "learning_rate": 1.0197431810915759e-05,
      "loss": 4.6304,
      "step": 12606
    },
    {
      "epoch": 0.5089624545821558,
      "grad_norm": 79.2470932006836,
      "learning_rate": 1.0196124483705916e-05,
      "loss": 5.8328,
      "step": 12607
    },
    {
      "epoch": 0.5090028259991926,
      "grad_norm": 106.62620544433594,
      "learning_rate": 1.0194817153142802e-05,
      "loss": 5.9067,
      "step": 12608
    },
    {
      "epoch": 0.5090431974162293,
      "grad_norm": 80.88709259033203,
      "learning_rate": 1.0193509819248769e-05,
      "loss": 6.6849,
      "step": 12609
    },
    {
      "epoch": 0.509083568833266,
      "grad_norm": 84.14592742919922,
      "learning_rate": 1.019220248204617e-05,
      "loss": 4.7553,
      "step": 12610
    },
    {
      "epoch": 0.5091239402503028,
      "grad_norm": 76.09764099121094,
      "learning_rate": 1.0190895141557355e-05,
      "loss": 5.3755,
      "step": 12611
    },
    {
      "epoch": 0.5091643116673396,
      "grad_norm": 87.69213104248047,
      "learning_rate": 1.0189587797804682e-05,
      "loss": 4.3139,
      "step": 12612
    },
    {
      "epoch": 0.5092046830843763,
      "grad_norm": 74.14318084716797,
      "learning_rate": 1.0188280450810502e-05,
      "loss": 4.6195,
      "step": 12613
    },
    {
      "epoch": 0.509245054501413,
      "grad_norm": 393.4039001464844,
      "learning_rate": 1.0186973100597162e-05,
      "loss": 5.3443,
      "step": 12614
    },
    {
      "epoch": 0.5092854259184497,
      "grad_norm": 101.50016021728516,
      "learning_rate": 1.018566574718702e-05,
      "loss": 4.0074,
      "step": 12615
    },
    {
      "epoch": 0.5093257973354864,
      "grad_norm": 127.7047348022461,
      "learning_rate": 1.0184358390602427e-05,
      "loss": 5.5472,
      "step": 12616
    },
    {
      "epoch": 0.5093661687525232,
      "grad_norm": 59.1123046875,
      "learning_rate": 1.0183051030865737e-05,
      "loss": 3.6731,
      "step": 12617
    },
    {
      "epoch": 0.50940654016956,
      "grad_norm": 104.334228515625,
      "learning_rate": 1.0181743667999302e-05,
      "loss": 4.6481,
      "step": 12618
    },
    {
      "epoch": 0.5094469115865967,
      "grad_norm": 87.68170166015625,
      "learning_rate": 1.0180436302025471e-05,
      "loss": 5.0931,
      "step": 12619
    },
    {
      "epoch": 0.5094872830036334,
      "grad_norm": 84.58240509033203,
      "learning_rate": 1.0179128932966607e-05,
      "loss": 5.0858,
      "step": 12620
    },
    {
      "epoch": 0.5095276544206702,
      "grad_norm": 70.5619125366211,
      "learning_rate": 1.0177821560845051e-05,
      "loss": 5.1275,
      "step": 12621
    },
    {
      "epoch": 0.5095680258377069,
      "grad_norm": 85.87469482421875,
      "learning_rate": 1.0176514185683166e-05,
      "loss": 5.8009,
      "step": 12622
    },
    {
      "epoch": 0.5096083972547436,
      "grad_norm": 73.62838745117188,
      "learning_rate": 1.0175206807503297e-05,
      "loss": 4.6256,
      "step": 12623
    },
    {
      "epoch": 0.5096487686717803,
      "grad_norm": 121.39632415771484,
      "learning_rate": 1.01738994263278e-05,
      "loss": 5.5143,
      "step": 12624
    },
    {
      "epoch": 0.5096891400888172,
      "grad_norm": 121.86805725097656,
      "learning_rate": 1.017259204217903e-05,
      "loss": 3.6688,
      "step": 12625
    },
    {
      "epoch": 0.5097295115058539,
      "grad_norm": 77.46708679199219,
      "learning_rate": 1.0171284655079344e-05,
      "loss": 5.2241,
      "step": 12626
    },
    {
      "epoch": 0.5097698829228906,
      "grad_norm": 104.19348907470703,
      "learning_rate": 1.0169977265051083e-05,
      "loss": 4.5177,
      "step": 12627
    },
    {
      "epoch": 0.5098102543399273,
      "grad_norm": 134.89556884765625,
      "learning_rate": 1.0168669872116612e-05,
      "loss": 4.4655,
      "step": 12628
    },
    {
      "epoch": 0.509850625756964,
      "grad_norm": 96.3539047241211,
      "learning_rate": 1.016736247629828e-05,
      "loss": 4.2641,
      "step": 12629
    },
    {
      "epoch": 0.5098909971740008,
      "grad_norm": 150.7725067138672,
      "learning_rate": 1.016605507761844e-05,
      "loss": 4.3648,
      "step": 12630
    },
    {
      "epoch": 0.5099313685910375,
      "grad_norm": 98.82990264892578,
      "learning_rate": 1.0164747676099445e-05,
      "loss": 3.8706,
      "step": 12631
    },
    {
      "epoch": 0.5099717400080743,
      "grad_norm": 84.14069366455078,
      "learning_rate": 1.0163440271763651e-05,
      "loss": 3.9853,
      "step": 12632
    },
    {
      "epoch": 0.510012111425111,
      "grad_norm": 117.7898941040039,
      "learning_rate": 1.0162132864633407e-05,
      "loss": 3.0902,
      "step": 12633
    },
    {
      "epoch": 0.5100524828421478,
      "grad_norm": 79.85958862304688,
      "learning_rate": 1.0160825454731072e-05,
      "loss": 4.2793,
      "step": 12634
    },
    {
      "epoch": 0.5100928542591845,
      "grad_norm": 125.49623107910156,
      "learning_rate": 1.0159518042078995e-05,
      "loss": 5.0894,
      "step": 12635
    },
    {
      "epoch": 0.5101332256762212,
      "grad_norm": 103.2586441040039,
      "learning_rate": 1.0158210626699536e-05,
      "loss": 4.7944,
      "step": 12636
    },
    {
      "epoch": 0.5101735970932579,
      "grad_norm": 137.09674072265625,
      "learning_rate": 1.015690320861504e-05,
      "loss": 3.9543,
      "step": 12637
    },
    {
      "epoch": 0.5102139685102947,
      "grad_norm": 141.27099609375,
      "learning_rate": 1.0155595787847871e-05,
      "loss": 4.5997,
      "step": 12638
    },
    {
      "epoch": 0.5102543399273315,
      "grad_norm": 75.95746612548828,
      "learning_rate": 1.0154288364420373e-05,
      "loss": 4.7278,
      "step": 12639
    },
    {
      "epoch": 0.5102947113443682,
      "grad_norm": 90.96875762939453,
      "learning_rate": 1.0152980938354908e-05,
      "loss": 4.6977,
      "step": 12640
    },
    {
      "epoch": 0.5103350827614049,
      "grad_norm": 117.15678405761719,
      "learning_rate": 1.0151673509673823e-05,
      "loss": 4.7908,
      "step": 12641
    },
    {
      "epoch": 0.5103754541784417,
      "grad_norm": 82.10401916503906,
      "learning_rate": 1.0150366078399478e-05,
      "loss": 6.5741,
      "step": 12642
    },
    {
      "epoch": 0.5104158255954784,
      "grad_norm": 107.5259017944336,
      "learning_rate": 1.0149058644554222e-05,
      "loss": 5.1759,
      "step": 12643
    },
    {
      "epoch": 0.5104561970125151,
      "grad_norm": 111.42878723144531,
      "learning_rate": 1.014775120816041e-05,
      "loss": 5.8096,
      "step": 12644
    },
    {
      "epoch": 0.5104965684295518,
      "grad_norm": 80.2330322265625,
      "learning_rate": 1.0146443769240397e-05,
      "loss": 5.4268,
      "step": 12645
    },
    {
      "epoch": 0.5105369398465887,
      "grad_norm": 86.46367645263672,
      "learning_rate": 1.014513632781654e-05,
      "loss": 5.5918,
      "step": 12646
    },
    {
      "epoch": 0.5105773112636254,
      "grad_norm": 105.83639526367188,
      "learning_rate": 1.014382888391119e-05,
      "loss": 5.2419,
      "step": 12647
    },
    {
      "epoch": 0.5106176826806621,
      "grad_norm": 78.45027923583984,
      "learning_rate": 1.0142521437546704e-05,
      "loss": 5.7264,
      "step": 12648
    },
    {
      "epoch": 0.5106580540976988,
      "grad_norm": 116.6031494140625,
      "learning_rate": 1.0141213988745431e-05,
      "loss": 4.4433,
      "step": 12649
    },
    {
      "epoch": 0.5106984255147355,
      "grad_norm": 83.58644104003906,
      "learning_rate": 1.0139906537529731e-05,
      "loss": 5.4968,
      "step": 12650
    },
    {
      "epoch": 0.5107387969317723,
      "grad_norm": 98.4834976196289,
      "learning_rate": 1.0138599083921955e-05,
      "loss": 6.2069,
      "step": 12651
    },
    {
      "epoch": 0.5107791683488091,
      "grad_norm": 86.93869018554688,
      "learning_rate": 1.013729162794446e-05,
      "loss": 4.189,
      "step": 12652
    },
    {
      "epoch": 0.5108195397658458,
      "grad_norm": 89.55501556396484,
      "learning_rate": 1.0135984169619595e-05,
      "loss": 5.9842,
      "step": 12653
    },
    {
      "epoch": 0.5108599111828825,
      "grad_norm": 95.7597427368164,
      "learning_rate": 1.013467670896972e-05,
      "loss": 5.0741,
      "step": 12654
    },
    {
      "epoch": 0.5109002825999193,
      "grad_norm": 107.59554290771484,
      "learning_rate": 1.0133369246017186e-05,
      "loss": 3.2037,
      "step": 12655
    },
    {
      "epoch": 0.510940654016956,
      "grad_norm": 119.49868774414062,
      "learning_rate": 1.0132061780784353e-05,
      "loss": 5.8765,
      "step": 12656
    },
    {
      "epoch": 0.5109810254339927,
      "grad_norm": 67.87871551513672,
      "learning_rate": 1.013075431329357e-05,
      "loss": 5.7618,
      "step": 12657
    },
    {
      "epoch": 0.5110213968510294,
      "grad_norm": 112.81472778320312,
      "learning_rate": 1.0129446843567194e-05,
      "loss": 6.0362,
      "step": 12658
    },
    {
      "epoch": 0.5110617682680663,
      "grad_norm": 87.58027648925781,
      "learning_rate": 1.012813937162758e-05,
      "loss": 5.7946,
      "step": 12659
    },
    {
      "epoch": 0.511102139685103,
      "grad_norm": 102.37594604492188,
      "learning_rate": 1.012683189749708e-05,
      "loss": 6.0479,
      "step": 12660
    },
    {
      "epoch": 0.5111425111021397,
      "grad_norm": 75.51893615722656,
      "learning_rate": 1.0125524421198052e-05,
      "loss": 4.9167,
      "step": 12661
    },
    {
      "epoch": 0.5111828825191764,
      "grad_norm": 113.6278305053711,
      "learning_rate": 1.012421694275285e-05,
      "loss": 4.9361,
      "step": 12662
    },
    {
      "epoch": 0.5112232539362132,
      "grad_norm": 133.95907592773438,
      "learning_rate": 1.0122909462183825e-05,
      "loss": 5.1082,
      "step": 12663
    },
    {
      "epoch": 0.5112636253532499,
      "grad_norm": 87.06539154052734,
      "learning_rate": 1.0121601979513338e-05,
      "loss": 3.1703,
      "step": 12664
    },
    {
      "epoch": 0.5113039967702866,
      "grad_norm": 78.34358215332031,
      "learning_rate": 1.0120294494763741e-05,
      "loss": 4.8683,
      "step": 12665
    },
    {
      "epoch": 0.5113443681873234,
      "grad_norm": 104.88660430908203,
      "learning_rate": 1.0118987007957389e-05,
      "loss": 3.7737,
      "step": 12666
    },
    {
      "epoch": 0.5113847396043602,
      "grad_norm": 99.61471557617188,
      "learning_rate": 1.0117679519116636e-05,
      "loss": 3.6565,
      "step": 12667
    },
    {
      "epoch": 0.5114251110213969,
      "grad_norm": 156.6334686279297,
      "learning_rate": 1.0116372028263841e-05,
      "loss": 5.8615,
      "step": 12668
    },
    {
      "epoch": 0.5114654824384336,
      "grad_norm": 67.10635375976562,
      "learning_rate": 1.0115064535421353e-05,
      "loss": 5.0886,
      "step": 12669
    },
    {
      "epoch": 0.5115058538554703,
      "grad_norm": 117.31635284423828,
      "learning_rate": 1.0113757040611532e-05,
      "loss": 4.8293,
      "step": 12670
    },
    {
      "epoch": 0.511546225272507,
      "grad_norm": 76.3913803100586,
      "learning_rate": 1.011244954385673e-05,
      "loss": 5.6093,
      "step": 12671
    },
    {
      "epoch": 0.5115865966895438,
      "grad_norm": 162.70669555664062,
      "learning_rate": 1.0111142045179305e-05,
      "loss": 3.7458,
      "step": 12672
    },
    {
      "epoch": 0.5116269681065806,
      "grad_norm": 109.01566314697266,
      "learning_rate": 1.010983454460161e-05,
      "loss": 6.3175,
      "step": 12673
    },
    {
      "epoch": 0.5116673395236173,
      "grad_norm": 116.6216049194336,
      "learning_rate": 1.0108527042145997e-05,
      "loss": 4.8014,
      "step": 12674
    },
    {
      "epoch": 0.511707710940654,
      "grad_norm": 95.65108489990234,
      "learning_rate": 1.0107219537834828e-05,
      "loss": 7.0913,
      "step": 12675
    },
    {
      "epoch": 0.5117480823576908,
      "grad_norm": 66.85746002197266,
      "learning_rate": 1.0105912031690457e-05,
      "loss": 3.4148,
      "step": 12676
    },
    {
      "epoch": 0.5117884537747275,
      "grad_norm": 77.89445495605469,
      "learning_rate": 1.0104604523735236e-05,
      "loss": 4.8951,
      "step": 12677
    },
    {
      "epoch": 0.5118288251917642,
      "grad_norm": 101.37907409667969,
      "learning_rate": 1.0103297013991523e-05,
      "loss": 6.5775,
      "step": 12678
    },
    {
      "epoch": 0.5118691966088009,
      "grad_norm": 135.7537078857422,
      "learning_rate": 1.0101989502481668e-05,
      "loss": 4.8602,
      "step": 12679
    },
    {
      "epoch": 0.5119095680258378,
      "grad_norm": 125.88581085205078,
      "learning_rate": 1.0100681989228037e-05,
      "loss": 4.0509,
      "step": 12680
    },
    {
      "epoch": 0.5119499394428745,
      "grad_norm": 66.89203643798828,
      "learning_rate": 1.0099374474252974e-05,
      "loss": 5.6941,
      "step": 12681
    },
    {
      "epoch": 0.5119903108599112,
      "grad_norm": 88.47601318359375,
      "learning_rate": 1.0098066957578842e-05,
      "loss": 5.0609,
      "step": 12682
    },
    {
      "epoch": 0.5120306822769479,
      "grad_norm": 93.91641235351562,
      "learning_rate": 1.0096759439227994e-05,
      "loss": 4.4303,
      "step": 12683
    },
    {
      "epoch": 0.5120710536939846,
      "grad_norm": 115.92745971679688,
      "learning_rate": 1.0095451919222785e-05,
      "loss": 5.4124,
      "step": 12684
    },
    {
      "epoch": 0.5121114251110214,
      "grad_norm": 82.45675659179688,
      "learning_rate": 1.009414439758557e-05,
      "loss": 5.7121,
      "step": 12685
    },
    {
      "epoch": 0.5121517965280581,
      "grad_norm": 68.69336700439453,
      "learning_rate": 1.0092836874338707e-05,
      "loss": 4.581,
      "step": 12686
    },
    {
      "epoch": 0.5121921679450949,
      "grad_norm": 91.177734375,
      "learning_rate": 1.009152934950455e-05,
      "loss": 3.0482,
      "step": 12687
    },
    {
      "epoch": 0.5122325393621316,
      "grad_norm": 103.2139663696289,
      "learning_rate": 1.0090221823105455e-05,
      "loss": 6.0796,
      "step": 12688
    },
    {
      "epoch": 0.5122729107791684,
      "grad_norm": 95.78581237792969,
      "learning_rate": 1.0088914295163776e-05,
      "loss": 4.4501,
      "step": 12689
    },
    {
      "epoch": 0.5123132821962051,
      "grad_norm": 107.34439849853516,
      "learning_rate": 1.0087606765701873e-05,
      "loss": 4.2138,
      "step": 12690
    },
    {
      "epoch": 0.5123536536132418,
      "grad_norm": 94.14425659179688,
      "learning_rate": 1.0086299234742099e-05,
      "loss": 4.338,
      "step": 12691
    },
    {
      "epoch": 0.5123940250302785,
      "grad_norm": 109.77254486083984,
      "learning_rate": 1.0084991702306807e-05,
      "loss": 4.3282,
      "step": 12692
    },
    {
      "epoch": 0.5124343964473153,
      "grad_norm": 82.76123046875,
      "learning_rate": 1.0083684168418357e-05,
      "loss": 5.7589,
      "step": 12693
    },
    {
      "epoch": 0.5124747678643521,
      "grad_norm": 127.39814758300781,
      "learning_rate": 1.0082376633099102e-05,
      "loss": 6.1903,
      "step": 12694
    },
    {
      "epoch": 0.5125151392813888,
      "grad_norm": 84.37899780273438,
      "learning_rate": 1.0081069096371398e-05,
      "loss": 5.6422,
      "step": 12695
    },
    {
      "epoch": 0.5125555106984255,
      "grad_norm": 126.06707763671875,
      "learning_rate": 1.0079761558257604e-05,
      "loss": 3.8149,
      "step": 12696
    },
    {
      "epoch": 0.5125958821154623,
      "grad_norm": 94.24447631835938,
      "learning_rate": 1.0078454018780073e-05,
      "loss": 4.5752,
      "step": 12697
    },
    {
      "epoch": 0.512636253532499,
      "grad_norm": 70.50334930419922,
      "learning_rate": 1.0077146477961161e-05,
      "loss": 4.0107,
      "step": 12698
    },
    {
      "epoch": 0.5126766249495357,
      "grad_norm": 250.90347290039062,
      "learning_rate": 1.0075838935823225e-05,
      "loss": 5.3689,
      "step": 12699
    },
    {
      "epoch": 0.5127169963665724,
      "grad_norm": 84.2378921508789,
      "learning_rate": 1.0074531392388622e-05,
      "loss": 6.5441,
      "step": 12700
    },
    {
      "epoch": 0.5127573677836093,
      "grad_norm": 84.87985229492188,
      "learning_rate": 1.0073223847679703e-05,
      "loss": 6.2073,
      "step": 12701
    },
    {
      "epoch": 0.512797739200646,
      "grad_norm": 96.331787109375,
      "learning_rate": 1.0071916301718828e-05,
      "loss": 5.5852,
      "step": 12702
    },
    {
      "epoch": 0.5128381106176827,
      "grad_norm": 78.69449615478516,
      "learning_rate": 1.0070608754528352e-05,
      "loss": 3.0045,
      "step": 12703
    },
    {
      "epoch": 0.5128784820347194,
      "grad_norm": 93.54203796386719,
      "learning_rate": 1.0069301206130633e-05,
      "loss": 6.1026,
      "step": 12704
    },
    {
      "epoch": 0.5129188534517561,
      "grad_norm": 112.4684066772461,
      "learning_rate": 1.0067993656548023e-05,
      "loss": 3.4823,
      "step": 12705
    },
    {
      "epoch": 0.5129592248687929,
      "grad_norm": 69.9578857421875,
      "learning_rate": 1.006668610580288e-05,
      "loss": 5.4312,
      "step": 12706
    },
    {
      "epoch": 0.5129995962858296,
      "grad_norm": 111.23653411865234,
      "learning_rate": 1.0065378553917562e-05,
      "loss": 4.2352,
      "step": 12707
    },
    {
      "epoch": 0.5130399677028664,
      "grad_norm": 88.45161437988281,
      "learning_rate": 1.0064071000914426e-05,
      "loss": 5.4891,
      "step": 12708
    },
    {
      "epoch": 0.5130803391199031,
      "grad_norm": 73.45863342285156,
      "learning_rate": 1.0062763446815819e-05,
      "loss": 5.3908,
      "step": 12709
    },
    {
      "epoch": 0.5131207105369399,
      "grad_norm": 76.81320190429688,
      "learning_rate": 1.006145589164411e-05,
      "loss": 6.1228,
      "step": 12710
    },
    {
      "epoch": 0.5131610819539766,
      "grad_norm": 82.16097259521484,
      "learning_rate": 1.0060148335421644e-05,
      "loss": 4.4083,
      "step": 12711
    },
    {
      "epoch": 0.5132014533710133,
      "grad_norm": 83.14142608642578,
      "learning_rate": 1.0058840778170783e-05,
      "loss": 6.7362,
      "step": 12712
    },
    {
      "epoch": 0.51324182478805,
      "grad_norm": 79.01449584960938,
      "learning_rate": 1.0057533219913882e-05,
      "loss": 5.329,
      "step": 12713
    },
    {
      "epoch": 0.5132821962050867,
      "grad_norm": 82.17507934570312,
      "learning_rate": 1.00562256606733e-05,
      "loss": 4.6826,
      "step": 12714
    },
    {
      "epoch": 0.5133225676221236,
      "grad_norm": 125.48909759521484,
      "learning_rate": 1.005491810047139e-05,
      "loss": 2.6007,
      "step": 12715
    },
    {
      "epoch": 0.5133629390391603,
      "grad_norm": 171.10255432128906,
      "learning_rate": 1.0053610539330508e-05,
      "loss": 5.4783,
      "step": 12716
    },
    {
      "epoch": 0.513403310456197,
      "grad_norm": 108.741455078125,
      "learning_rate": 1.0052302977273007e-05,
      "loss": 4.5905,
      "step": 12717
    },
    {
      "epoch": 0.5134436818732337,
      "grad_norm": 89.89995574951172,
      "learning_rate": 1.0050995414321254e-05,
      "loss": 5.2405,
      "step": 12718
    },
    {
      "epoch": 0.5134840532902705,
      "grad_norm": 86.68450164794922,
      "learning_rate": 1.0049687850497594e-05,
      "loss": 4.7804,
      "step": 12719
    },
    {
      "epoch": 0.5135244247073072,
      "grad_norm": 287.5587158203125,
      "learning_rate": 1.004838028582439e-05,
      "loss": 5.3865,
      "step": 12720
    },
    {
      "epoch": 0.5135647961243439,
      "grad_norm": 89.39928436279297,
      "learning_rate": 1.0047072720323994e-05,
      "loss": 5.104,
      "step": 12721
    },
    {
      "epoch": 0.5136051675413807,
      "grad_norm": 72.79562377929688,
      "learning_rate": 1.0045765154018764e-05,
      "loss": 3.4933,
      "step": 12722
    },
    {
      "epoch": 0.5136455389584175,
      "grad_norm": 71.52538299560547,
      "learning_rate": 1.0044457586931058e-05,
      "loss": 3.735,
      "step": 12723
    },
    {
      "epoch": 0.5136859103754542,
      "grad_norm": 79.3344497680664,
      "learning_rate": 1.0043150019083233e-05,
      "loss": 5.3054,
      "step": 12724
    },
    {
      "epoch": 0.5137262817924909,
      "grad_norm": 100.11754608154297,
      "learning_rate": 1.0041842450497638e-05,
      "loss": 5.3487,
      "step": 12725
    },
    {
      "epoch": 0.5137666532095276,
      "grad_norm": 120.17268371582031,
      "learning_rate": 1.0040534881196642e-05,
      "loss": 7.0058,
      "step": 12726
    },
    {
      "epoch": 0.5138070246265644,
      "grad_norm": 96.70887756347656,
      "learning_rate": 1.0039227311202589e-05,
      "loss": 5.7657,
      "step": 12727
    },
    {
      "epoch": 0.5138473960436011,
      "grad_norm": 77.43994903564453,
      "learning_rate": 1.0037919740537842e-05,
      "loss": 6.6841,
      "step": 12728
    },
    {
      "epoch": 0.5138877674606379,
      "grad_norm": 116.20873260498047,
      "learning_rate": 1.0036612169224758e-05,
      "loss": 5.6021,
      "step": 12729
    },
    {
      "epoch": 0.5139281388776746,
      "grad_norm": 105.06221771240234,
      "learning_rate": 1.0035304597285689e-05,
      "loss": 4.9888,
      "step": 12730
    },
    {
      "epoch": 0.5139685102947114,
      "grad_norm": 95.99848937988281,
      "learning_rate": 1.0033997024742994e-05,
      "loss": 6.0452,
      "step": 12731
    },
    {
      "epoch": 0.5140088817117481,
      "grad_norm": 96.4288101196289,
      "learning_rate": 1.0032689451619029e-05,
      "loss": 6.3093,
      "step": 12732
    },
    {
      "epoch": 0.5140492531287848,
      "grad_norm": 140.34027099609375,
      "learning_rate": 1.003138187793615e-05,
      "loss": 4.605,
      "step": 12733
    },
    {
      "epoch": 0.5140896245458215,
      "grad_norm": 120.79237365722656,
      "learning_rate": 1.0030074303716717e-05,
      "loss": 5.6816,
      "step": 12734
    },
    {
      "epoch": 0.5141299959628582,
      "grad_norm": 110.29689025878906,
      "learning_rate": 1.0028766728983082e-05,
      "loss": 5.3406,
      "step": 12735
    },
    {
      "epoch": 0.5141703673798951,
      "grad_norm": 106.82878112792969,
      "learning_rate": 1.0027459153757605e-05,
      "loss": 4.0422,
      "step": 12736
    },
    {
      "epoch": 0.5142107387969318,
      "grad_norm": 132.17210388183594,
      "learning_rate": 1.0026151578062637e-05,
      "loss": 5.0676,
      "step": 12737
    },
    {
      "epoch": 0.5142511102139685,
      "grad_norm": 75.18045806884766,
      "learning_rate": 1.0024844001920542e-05,
      "loss": 4.8622,
      "step": 12738
    },
    {
      "epoch": 0.5142914816310052,
      "grad_norm": 192.52854919433594,
      "learning_rate": 1.0023536425353668e-05,
      "loss": 6.0726,
      "step": 12739
    },
    {
      "epoch": 0.514331853048042,
      "grad_norm": 88.24881744384766,
      "learning_rate": 1.0022228848384382e-05,
      "loss": 4.2451,
      "step": 12740
    },
    {
      "epoch": 0.5143722244650787,
      "grad_norm": 78.46553802490234,
      "learning_rate": 1.0020921271035029e-05,
      "loss": 5.3728,
      "step": 12741
    },
    {
      "epoch": 0.5144125958821154,
      "grad_norm": 73.43034362792969,
      "learning_rate": 1.0019613693327974e-05,
      "loss": 6.331,
      "step": 12742
    },
    {
      "epoch": 0.5144529672991522,
      "grad_norm": 136.8926544189453,
      "learning_rate": 1.0018306115285568e-05,
      "loss": 5.0574,
      "step": 12743
    },
    {
      "epoch": 0.514493338716189,
      "grad_norm": 100.21123504638672,
      "learning_rate": 1.0016998536930174e-05,
      "loss": 5.8578,
      "step": 12744
    },
    {
      "epoch": 0.5145337101332257,
      "grad_norm": 193.20098876953125,
      "learning_rate": 1.001569095828414e-05,
      "loss": 5.9301,
      "step": 12745
    },
    {
      "epoch": 0.5145740815502624,
      "grad_norm": 185.29312133789062,
      "learning_rate": 1.0014383379369833e-05,
      "loss": 5.4341,
      "step": 12746
    },
    {
      "epoch": 0.5146144529672991,
      "grad_norm": 71.41925811767578,
      "learning_rate": 1.00130758002096e-05,
      "loss": 4.8444,
      "step": 12747
    },
    {
      "epoch": 0.5146548243843359,
      "grad_norm": 96.54964447021484,
      "learning_rate": 1.0011768220825805e-05,
      "loss": 5.2236,
      "step": 12748
    },
    {
      "epoch": 0.5146951958013726,
      "grad_norm": 68.35567474365234,
      "learning_rate": 1.0010460641240796e-05,
      "loss": 4.4949,
      "step": 12749
    },
    {
      "epoch": 0.5147355672184094,
      "grad_norm": 66.60358428955078,
      "learning_rate": 1.0009153061476937e-05,
      "loss": 5.9192,
      "step": 12750
    },
    {
      "epoch": 0.5147759386354461,
      "grad_norm": 93.1425552368164,
      "learning_rate": 1.0007845481556581e-05,
      "loss": 4.6372,
      "step": 12751
    },
    {
      "epoch": 0.5148163100524829,
      "grad_norm": 249.6703643798828,
      "learning_rate": 1.0006537901502086e-05,
      "loss": 3.7428,
      "step": 12752
    },
    {
      "epoch": 0.5148566814695196,
      "grad_norm": 118.05361938476562,
      "learning_rate": 1.0005230321335808e-05,
      "loss": 6.339,
      "step": 12753
    },
    {
      "epoch": 0.5148970528865563,
      "grad_norm": 80.02474212646484,
      "learning_rate": 1.0003922741080106e-05,
      "loss": 4.0495,
      "step": 12754
    },
    {
      "epoch": 0.514937424303593,
      "grad_norm": 88.74015045166016,
      "learning_rate": 1.0002615160757332e-05,
      "loss": 5.6015,
      "step": 12755
    },
    {
      "epoch": 0.5149777957206298,
      "grad_norm": 119.32479858398438,
      "learning_rate": 1.0001307580389845e-05,
      "loss": 5.8161,
      "step": 12756
    },
    {
      "epoch": 0.5150181671376666,
      "grad_norm": 114.36346435546875,
      "learning_rate": 1e-05,
      "loss": 4.5961,
      "step": 12757
    },
    {
      "epoch": 0.5150585385547033,
      "grad_norm": 125.90904235839844,
      "learning_rate": 9.998692419610157e-06,
      "loss": 5.5906,
      "step": 12758
    },
    {
      "epoch": 0.51509890997174,
      "grad_norm": 122.13849639892578,
      "learning_rate": 9.99738483924267e-06,
      "loss": 6.9869,
      "step": 12759
    },
    {
      "epoch": 0.5151392813887767,
      "grad_norm": 193.19764709472656,
      "learning_rate": 9.996077258919899e-06,
      "loss": 7.0903,
      "step": 12760
    },
    {
      "epoch": 0.5151796528058135,
      "grad_norm": 92.65805053710938,
      "learning_rate": 9.994769678664195e-06,
      "loss": 5.9744,
      "step": 12761
    },
    {
      "epoch": 0.5152200242228502,
      "grad_norm": 70.1250228881836,
      "learning_rate": 9.993462098497917e-06,
      "loss": 4.11,
      "step": 12762
    },
    {
      "epoch": 0.515260395639887,
      "grad_norm": 78.51193237304688,
      "learning_rate": 9.99215451844342e-06,
      "loss": 4.9869,
      "step": 12763
    },
    {
      "epoch": 0.5153007670569237,
      "grad_norm": 135.12557983398438,
      "learning_rate": 9.990846938523066e-06,
      "loss": 5.6305,
      "step": 12764
    },
    {
      "epoch": 0.5153411384739605,
      "grad_norm": 62.449371337890625,
      "learning_rate": 9.989539358759207e-06,
      "loss": 4.6158,
      "step": 12765
    },
    {
      "epoch": 0.5153815098909972,
      "grad_norm": 291.96502685546875,
      "learning_rate": 9.988231779174199e-06,
      "loss": 4.0822,
      "step": 12766
    },
    {
      "epoch": 0.5154218813080339,
      "grad_norm": 68.17632293701172,
      "learning_rate": 9.9869241997904e-06,
      "loss": 5.08,
      "step": 12767
    },
    {
      "epoch": 0.5154622527250706,
      "grad_norm": 126.22299194335938,
      "learning_rate": 9.985616620630172e-06,
      "loss": 2.9922,
      "step": 12768
    },
    {
      "epoch": 0.5155026241421073,
      "grad_norm": 89.1727523803711,
      "learning_rate": 9.984309041715861e-06,
      "loss": 5.1395,
      "step": 12769
    },
    {
      "epoch": 0.5155429955591442,
      "grad_norm": 79.36923217773438,
      "learning_rate": 9.98300146306983e-06,
      "loss": 6.3285,
      "step": 12770
    },
    {
      "epoch": 0.5155833669761809,
      "grad_norm": 121.5395278930664,
      "learning_rate": 9.981693884714433e-06,
      "loss": 4.9606,
      "step": 12771
    },
    {
      "epoch": 0.5156237383932176,
      "grad_norm": 104.26461791992188,
      "learning_rate": 9.98038630667203e-06,
      "loss": 4.0363,
      "step": 12772
    },
    {
      "epoch": 0.5156641098102543,
      "grad_norm": 155.9984588623047,
      "learning_rate": 9.979078728964974e-06,
      "loss": 4.3382,
      "step": 12773
    },
    {
      "epoch": 0.5157044812272911,
      "grad_norm": 64.20514678955078,
      "learning_rate": 9.977771151615621e-06,
      "loss": 5.9864,
      "step": 12774
    },
    {
      "epoch": 0.5157448526443278,
      "grad_norm": 88.2839126586914,
      "learning_rate": 9.97646357464633e-06,
      "loss": 5.0403,
      "step": 12775
    },
    {
      "epoch": 0.5157852240613645,
      "grad_norm": 190.65158081054688,
      "learning_rate": 9.975155998079463e-06,
      "loss": 5.2439,
      "step": 12776
    },
    {
      "epoch": 0.5158255954784013,
      "grad_norm": 71.80062103271484,
      "learning_rate": 9.973848421937366e-06,
      "loss": 4.9836,
      "step": 12777
    },
    {
      "epoch": 0.5158659668954381,
      "grad_norm": 101.86094665527344,
      "learning_rate": 9.972540846242399e-06,
      "loss": 5.1579,
      "step": 12778
    },
    {
      "epoch": 0.5159063383124748,
      "grad_norm": 221.1688690185547,
      "learning_rate": 9.971233271016918e-06,
      "loss": 6.0701,
      "step": 12779
    },
    {
      "epoch": 0.5159467097295115,
      "grad_norm": 125.98677062988281,
      "learning_rate": 9.969925696283288e-06,
      "loss": 5.2046,
      "step": 12780
    },
    {
      "epoch": 0.5159870811465482,
      "grad_norm": 77.4419937133789,
      "learning_rate": 9.968618122063852e-06,
      "loss": 5.8723,
      "step": 12781
    },
    {
      "epoch": 0.516027452563585,
      "grad_norm": 114.46165466308594,
      "learning_rate": 9.967310548380974e-06,
      "loss": 5.5946,
      "step": 12782
    },
    {
      "epoch": 0.5160678239806217,
      "grad_norm": 158.94189453125,
      "learning_rate": 9.966002975257008e-06,
      "loss": 3.0992,
      "step": 12783
    },
    {
      "epoch": 0.5161081953976585,
      "grad_norm": 139.4281768798828,
      "learning_rate": 9.964695402714314e-06,
      "loss": 4.6761,
      "step": 12784
    },
    {
      "epoch": 0.5161485668146952,
      "grad_norm": 108.35649871826172,
      "learning_rate": 9.963387830775246e-06,
      "loss": 6.2069,
      "step": 12785
    },
    {
      "epoch": 0.516188938231732,
      "grad_norm": 74.62540435791016,
      "learning_rate": 9.96208025946216e-06,
      "loss": 5.0862,
      "step": 12786
    },
    {
      "epoch": 0.5162293096487687,
      "grad_norm": 160.91122436523438,
      "learning_rate": 9.960772688797413e-06,
      "loss": 5.5092,
      "step": 12787
    },
    {
      "epoch": 0.5162696810658054,
      "grad_norm": 103.00935363769531,
      "learning_rate": 9.959465118803365e-06,
      "loss": 2.4539,
      "step": 12788
    },
    {
      "epoch": 0.5163100524828421,
      "grad_norm": 79.74209594726562,
      "learning_rate": 9.958157549502363e-06,
      "loss": 6.0851,
      "step": 12789
    },
    {
      "epoch": 0.5163504238998788,
      "grad_norm": 110.08724975585938,
      "learning_rate": 9.95684998091677e-06,
      "loss": 4.784,
      "step": 12790
    },
    {
      "epoch": 0.5163907953169157,
      "grad_norm": 115.66817474365234,
      "learning_rate": 9.955542413068944e-06,
      "loss": 5.4702,
      "step": 12791
    },
    {
      "epoch": 0.5164311667339524,
      "grad_norm": 74.6824951171875,
      "learning_rate": 9.954234845981239e-06,
      "loss": 5.6048,
      "step": 12792
    },
    {
      "epoch": 0.5164715381509891,
      "grad_norm": 75.62712860107422,
      "learning_rate": 9.95292727967601e-06,
      "loss": 6.5475,
      "step": 12793
    },
    {
      "epoch": 0.5165119095680258,
      "grad_norm": 108.10828399658203,
      "learning_rate": 9.951619714175613e-06,
      "loss": 5.8794,
      "step": 12794
    },
    {
      "epoch": 0.5165522809850626,
      "grad_norm": 69.76911163330078,
      "learning_rate": 9.950312149502408e-06,
      "loss": 5.0398,
      "step": 12795
    },
    {
      "epoch": 0.5165926524020993,
      "grad_norm": 74.78824615478516,
      "learning_rate": 9.949004585678753e-06,
      "loss": 4.8117,
      "step": 12796
    },
    {
      "epoch": 0.516633023819136,
      "grad_norm": 99.57527923583984,
      "learning_rate": 9.947697022726995e-06,
      "loss": 7.0339,
      "step": 12797
    },
    {
      "epoch": 0.5166733952361728,
      "grad_norm": 173.1612091064453,
      "learning_rate": 9.946389460669497e-06,
      "loss": 4.8662,
      "step": 12798
    },
    {
      "epoch": 0.5167137666532096,
      "grad_norm": 88.25115203857422,
      "learning_rate": 9.945081899528612e-06,
      "loss": 5.4241,
      "step": 12799
    },
    {
      "epoch": 0.5167541380702463,
      "grad_norm": 63.665252685546875,
      "learning_rate": 9.943774339326705e-06,
      "loss": 5.1272,
      "step": 12800
    },
    {
      "epoch": 0.516794509487283,
      "grad_norm": 109.06967163085938,
      "learning_rate": 9.94246678008612e-06,
      "loss": 4.8093,
      "step": 12801
    },
    {
      "epoch": 0.5168348809043197,
      "grad_norm": 106.66828155517578,
      "learning_rate": 9.941159221829218e-06,
      "loss": 6.7448,
      "step": 12802
    },
    {
      "epoch": 0.5168752523213564,
      "grad_norm": 111.19959259033203,
      "learning_rate": 9.939851664578358e-06,
      "loss": 3.8175,
      "step": 12803
    },
    {
      "epoch": 0.5169156237383932,
      "grad_norm": 115.04878997802734,
      "learning_rate": 9.938544108355895e-06,
      "loss": 3.6195,
      "step": 12804
    },
    {
      "epoch": 0.51695599515543,
      "grad_norm": 101.14964294433594,
      "learning_rate": 9.937236553184182e-06,
      "loss": 5.8594,
      "step": 12805
    },
    {
      "epoch": 0.5169963665724667,
      "grad_norm": 111.7103042602539,
      "learning_rate": 9.935928999085577e-06,
      "loss": 5.5349,
      "step": 12806
    },
    {
      "epoch": 0.5170367379895034,
      "grad_norm": 79.93992614746094,
      "learning_rate": 9.934621446082438e-06,
      "loss": 5.0137,
      "step": 12807
    },
    {
      "epoch": 0.5170771094065402,
      "grad_norm": 92.61449432373047,
      "learning_rate": 9.933313894197123e-06,
      "loss": 4.8115,
      "step": 12808
    },
    {
      "epoch": 0.5171174808235769,
      "grad_norm": 106.97830200195312,
      "learning_rate": 9.93200634345198e-06,
      "loss": 5.0807,
      "step": 12809
    },
    {
      "epoch": 0.5171578522406136,
      "grad_norm": 70.19828033447266,
      "learning_rate": 9.93069879386937e-06,
      "loss": 5.6708,
      "step": 12810
    },
    {
      "epoch": 0.5171982236576503,
      "grad_norm": 72.42378234863281,
      "learning_rate": 9.92939124547165e-06,
      "loss": 5.6136,
      "step": 12811
    },
    {
      "epoch": 0.5172385950746872,
      "grad_norm": 98.82887268066406,
      "learning_rate": 9.928083698281176e-06,
      "loss": 4.799,
      "step": 12812
    },
    {
      "epoch": 0.5172789664917239,
      "grad_norm": 89.64826202392578,
      "learning_rate": 9.9267761523203e-06,
      "loss": 4.8564,
      "step": 12813
    },
    {
      "epoch": 0.5173193379087606,
      "grad_norm": 158.25123596191406,
      "learning_rate": 9.925468607611381e-06,
      "loss": 5.258,
      "step": 12814
    },
    {
      "epoch": 0.5173597093257973,
      "grad_norm": 66.88819885253906,
      "learning_rate": 9.924161064176777e-06,
      "loss": 5.0035,
      "step": 12815
    },
    {
      "epoch": 0.5174000807428341,
      "grad_norm": 77.27924346923828,
      "learning_rate": 9.922853522038844e-06,
      "loss": 5.8104,
      "step": 12816
    },
    {
      "epoch": 0.5174404521598708,
      "grad_norm": 67.88117218017578,
      "learning_rate": 9.92154598121993e-06,
      "loss": 4.6156,
      "step": 12817
    },
    {
      "epoch": 0.5174808235769075,
      "grad_norm": 95.41297149658203,
      "learning_rate": 9.920238441742398e-06,
      "loss": 4.8837,
      "step": 12818
    },
    {
      "epoch": 0.5175211949939443,
      "grad_norm": 123.39391326904297,
      "learning_rate": 9.918930903628603e-06,
      "loss": 5.7469,
      "step": 12819
    },
    {
      "epoch": 0.517561566410981,
      "grad_norm": 82.71196746826172,
      "learning_rate": 9.9176233669009e-06,
      "loss": 4.5484,
      "step": 12820
    },
    {
      "epoch": 0.5176019378280178,
      "grad_norm": 83.32742309570312,
      "learning_rate": 9.916315831581646e-06,
      "loss": 5.8906,
      "step": 12821
    },
    {
      "epoch": 0.5176423092450545,
      "grad_norm": 168.75587463378906,
      "learning_rate": 9.915008297693195e-06,
      "loss": 5.859,
      "step": 12822
    },
    {
      "epoch": 0.5176826806620912,
      "grad_norm": 94.34134674072266,
      "learning_rate": 9.913700765257903e-06,
      "loss": 5.4101,
      "step": 12823
    },
    {
      "epoch": 0.5177230520791279,
      "grad_norm": 108.3462905883789,
      "learning_rate": 9.912393234298132e-06,
      "loss": 4.101,
      "step": 12824
    },
    {
      "epoch": 0.5177634234961647,
      "grad_norm": 105.431640625,
      "learning_rate": 9.911085704836225e-06,
      "loss": 4.4446,
      "step": 12825
    },
    {
      "epoch": 0.5178037949132015,
      "grad_norm": 331.0045166015625,
      "learning_rate": 9.909778176894546e-06,
      "loss": 5.1882,
      "step": 12826
    },
    {
      "epoch": 0.5178441663302382,
      "grad_norm": 118.92120361328125,
      "learning_rate": 9.90847065049545e-06,
      "loss": 5.3924,
      "step": 12827
    },
    {
      "epoch": 0.5178845377472749,
      "grad_norm": 79.58851623535156,
      "learning_rate": 9.907163125661298e-06,
      "loss": 4.9463,
      "step": 12828
    },
    {
      "epoch": 0.5179249091643117,
      "grad_norm": 144.2375946044922,
      "learning_rate": 9.905855602414433e-06,
      "loss": 5.3378,
      "step": 12829
    },
    {
      "epoch": 0.5179652805813484,
      "grad_norm": 119.22740936279297,
      "learning_rate": 9.904548080777219e-06,
      "loss": 5.0395,
      "step": 12830
    },
    {
      "epoch": 0.5180056519983851,
      "grad_norm": 182.2905731201172,
      "learning_rate": 9.90324056077201e-06,
      "loss": 4.0756,
      "step": 12831
    },
    {
      "epoch": 0.5180460234154218,
      "grad_norm": 81.65303039550781,
      "learning_rate": 9.901933042421161e-06,
      "loss": 5.5524,
      "step": 12832
    },
    {
      "epoch": 0.5180863948324587,
      "grad_norm": 114.00125122070312,
      "learning_rate": 9.900625525747028e-06,
      "loss": 5.4294,
      "step": 12833
    },
    {
      "epoch": 0.5181267662494954,
      "grad_norm": 111.13581848144531,
      "learning_rate": 9.899318010771966e-06,
      "loss": 4.9318,
      "step": 12834
    },
    {
      "epoch": 0.5181671376665321,
      "grad_norm": 87.239501953125,
      "learning_rate": 9.898010497518332e-06,
      "loss": 3.8545,
      "step": 12835
    },
    {
      "epoch": 0.5182075090835688,
      "grad_norm": 80.68714904785156,
      "learning_rate": 9.896702986008482e-06,
      "loss": 6.0596,
      "step": 12836
    },
    {
      "epoch": 0.5182478805006056,
      "grad_norm": 100.60855102539062,
      "learning_rate": 9.89539547626477e-06,
      "loss": 2.9685,
      "step": 12837
    },
    {
      "epoch": 0.5182882519176423,
      "grad_norm": 104.01292419433594,
      "learning_rate": 9.894087968309546e-06,
      "loss": 3.6654,
      "step": 12838
    },
    {
      "epoch": 0.518328623334679,
      "grad_norm": 100.79309844970703,
      "learning_rate": 9.892780462165172e-06,
      "loss": 4.6734,
      "step": 12839
    },
    {
      "epoch": 0.5183689947517158,
      "grad_norm": 83.06945037841797,
      "learning_rate": 9.891472957854005e-06,
      "loss": 5.8161,
      "step": 12840
    },
    {
      "epoch": 0.5184093661687526,
      "grad_norm": 159.0811309814453,
      "learning_rate": 9.890165455398396e-06,
      "loss": 3.758,
      "step": 12841
    },
    {
      "epoch": 0.5184497375857893,
      "grad_norm": 115.12218475341797,
      "learning_rate": 9.888857954820697e-06,
      "loss": 4.2116,
      "step": 12842
    },
    {
      "epoch": 0.518490109002826,
      "grad_norm": 120.03300476074219,
      "learning_rate": 9.887550456143271e-06,
      "loss": 5.9508,
      "step": 12843
    },
    {
      "epoch": 0.5185304804198627,
      "grad_norm": 78.3558349609375,
      "learning_rate": 9.886242959388473e-06,
      "loss": 4.7806,
      "step": 12844
    },
    {
      "epoch": 0.5185708518368994,
      "grad_norm": 93.30827331542969,
      "learning_rate": 9.88493546457865e-06,
      "loss": 6.0468,
      "step": 12845
    },
    {
      "epoch": 0.5186112232539362,
      "grad_norm": 105.41378784179688,
      "learning_rate": 9.883627971736162e-06,
      "loss": 6.2007,
      "step": 12846
    },
    {
      "epoch": 0.518651594670973,
      "grad_norm": 107.74822998046875,
      "learning_rate": 9.882320480883364e-06,
      "loss": 4.5222,
      "step": 12847
    },
    {
      "epoch": 0.5186919660880097,
      "grad_norm": 106.2939453125,
      "learning_rate": 9.881012992042616e-06,
      "loss": 5.2501,
      "step": 12848
    },
    {
      "epoch": 0.5187323375050464,
      "grad_norm": 78.71504211425781,
      "learning_rate": 9.879705505236264e-06,
      "loss": 6.2697,
      "step": 12849
    },
    {
      "epoch": 0.5187727089220832,
      "grad_norm": 95.1829833984375,
      "learning_rate": 9.878398020486665e-06,
      "loss": 5.3079,
      "step": 12850
    },
    {
      "epoch": 0.5188130803391199,
      "grad_norm": 85.20496368408203,
      "learning_rate": 9.877090537816177e-06,
      "loss": 6.9325,
      "step": 12851
    },
    {
      "epoch": 0.5188534517561566,
      "grad_norm": 77.18399810791016,
      "learning_rate": 9.875783057247155e-06,
      "loss": 4.8295,
      "step": 12852
    },
    {
      "epoch": 0.5188938231731933,
      "grad_norm": 65.89772033691406,
      "learning_rate": 9.874475578801952e-06,
      "loss": 4.8092,
      "step": 12853
    },
    {
      "epoch": 0.5189341945902302,
      "grad_norm": 103.0360336303711,
      "learning_rate": 9.873168102502923e-06,
      "loss": 3.8061,
      "step": 12854
    },
    {
      "epoch": 0.5189745660072669,
      "grad_norm": 116.88219451904297,
      "learning_rate": 9.871860628372422e-06,
      "loss": 4.2028,
      "step": 12855
    },
    {
      "epoch": 0.5190149374243036,
      "grad_norm": 147.17111206054688,
      "learning_rate": 9.870553156432811e-06,
      "loss": 5.1275,
      "step": 12856
    },
    {
      "epoch": 0.5190553088413403,
      "grad_norm": 84.92903137207031,
      "learning_rate": 9.869245686706434e-06,
      "loss": 3.7733,
      "step": 12857
    },
    {
      "epoch": 0.519095680258377,
      "grad_norm": 115.32599639892578,
      "learning_rate": 9.86793821921565e-06,
      "loss": 4.7764,
      "step": 12858
    },
    {
      "epoch": 0.5191360516754138,
      "grad_norm": 81.9309310913086,
      "learning_rate": 9.866630753982814e-06,
      "loss": 5.6595,
      "step": 12859
    },
    {
      "epoch": 0.5191764230924505,
      "grad_norm": 123.47638702392578,
      "learning_rate": 9.865323291030285e-06,
      "loss": 5.9303,
      "step": 12860
    },
    {
      "epoch": 0.5192167945094873,
      "grad_norm": 107.7166519165039,
      "learning_rate": 9.864015830380408e-06,
      "loss": 6.8527,
      "step": 12861
    },
    {
      "epoch": 0.519257165926524,
      "grad_norm": 85.40977478027344,
      "learning_rate": 9.862708372055544e-06,
      "loss": 6.2545,
      "step": 12862
    },
    {
      "epoch": 0.5192975373435608,
      "grad_norm": 89.75958251953125,
      "learning_rate": 9.861400916078047e-06,
      "loss": 4.9419,
      "step": 12863
    },
    {
      "epoch": 0.5193379087605975,
      "grad_norm": 179.30996704101562,
      "learning_rate": 9.860093462470274e-06,
      "loss": 4.4002,
      "step": 12864
    },
    {
      "epoch": 0.5193782801776342,
      "grad_norm": 85.0166015625,
      "learning_rate": 9.858786011254572e-06,
      "loss": 6.6861,
      "step": 12865
    },
    {
      "epoch": 0.5194186515946709,
      "grad_norm": 95.4521484375,
      "learning_rate": 9.857478562453299e-06,
      "loss": 6.0334,
      "step": 12866
    },
    {
      "epoch": 0.5194590230117078,
      "grad_norm": 101.57801818847656,
      "learning_rate": 9.856171116088811e-06,
      "loss": 4.6368,
      "step": 12867
    },
    {
      "epoch": 0.5194993944287445,
      "grad_norm": 92.12083435058594,
      "learning_rate": 9.854863672183463e-06,
      "loss": 3.0043,
      "step": 12868
    },
    {
      "epoch": 0.5195397658457812,
      "grad_norm": 68.22545623779297,
      "learning_rate": 9.853556230759606e-06,
      "loss": 5.261,
      "step": 12869
    },
    {
      "epoch": 0.5195801372628179,
      "grad_norm": 100.4686279296875,
      "learning_rate": 9.852248791839594e-06,
      "loss": 6.0416,
      "step": 12870
    },
    {
      "epoch": 0.5196205086798547,
      "grad_norm": 101.98849487304688,
      "learning_rate": 9.850941355445782e-06,
      "loss": 6.1009,
      "step": 12871
    },
    {
      "epoch": 0.5196608800968914,
      "grad_norm": 110.21360778808594,
      "learning_rate": 9.849633921600527e-06,
      "loss": 4.8423,
      "step": 12872
    },
    {
      "epoch": 0.5197012515139281,
      "grad_norm": 101.36345672607422,
      "learning_rate": 9.84832649032618e-06,
      "loss": 4.3873,
      "step": 12873
    },
    {
      "epoch": 0.5197416229309649,
      "grad_norm": 83.65137481689453,
      "learning_rate": 9.847019061645096e-06,
      "loss": 6.1392,
      "step": 12874
    },
    {
      "epoch": 0.5197819943480017,
      "grad_norm": 77.56166076660156,
      "learning_rate": 9.845711635579627e-06,
      "loss": 6.3344,
      "step": 12875
    },
    {
      "epoch": 0.5198223657650384,
      "grad_norm": 206.42352294921875,
      "learning_rate": 9.844404212152134e-06,
      "loss": 4.7442,
      "step": 12876
    },
    {
      "epoch": 0.5198627371820751,
      "grad_norm": 141.4284210205078,
      "learning_rate": 9.843096791384961e-06,
      "loss": 4.3872,
      "step": 12877
    },
    {
      "epoch": 0.5199031085991118,
      "grad_norm": 111.20421600341797,
      "learning_rate": 9.841789373300467e-06,
      "loss": 5.7681,
      "step": 12878
    },
    {
      "epoch": 0.5199434800161485,
      "grad_norm": 83.14710235595703,
      "learning_rate": 9.840481957921005e-06,
      "loss": 4.8665,
      "step": 12879
    },
    {
      "epoch": 0.5199838514331853,
      "grad_norm": 80.2088851928711,
      "learning_rate": 9.839174545268931e-06,
      "loss": 4.7266,
      "step": 12880
    },
    {
      "epoch": 0.5200242228502221,
      "grad_norm": 128.02728271484375,
      "learning_rate": 9.837867135366594e-06,
      "loss": 5.1303,
      "step": 12881
    },
    {
      "epoch": 0.5200645942672588,
      "grad_norm": 83.96272277832031,
      "learning_rate": 9.836559728236352e-06,
      "loss": 4.753,
      "step": 12882
    },
    {
      "epoch": 0.5201049656842955,
      "grad_norm": 94.99073791503906,
      "learning_rate": 9.835252323900557e-06,
      "loss": 4.7622,
      "step": 12883
    },
    {
      "epoch": 0.5201453371013323,
      "grad_norm": 83.73275756835938,
      "learning_rate": 9.833944922381566e-06,
      "loss": 6.5677,
      "step": 12884
    },
    {
      "epoch": 0.520185708518369,
      "grad_norm": 80.26625061035156,
      "learning_rate": 9.832637523701724e-06,
      "loss": 5.1083,
      "step": 12885
    },
    {
      "epoch": 0.5202260799354057,
      "grad_norm": 114.2660903930664,
      "learning_rate": 9.83133012788339e-06,
      "loss": 5.0486,
      "step": 12886
    },
    {
      "epoch": 0.5202664513524424,
      "grad_norm": 69.28216552734375,
      "learning_rate": 9.830022734948917e-06,
      "loss": 3.5357,
      "step": 12887
    },
    {
      "epoch": 0.5203068227694793,
      "grad_norm": 77.46051025390625,
      "learning_rate": 9.828715344920663e-06,
      "loss": 6.046,
      "step": 12888
    },
    {
      "epoch": 0.520347194186516,
      "grad_norm": 112.31704711914062,
      "learning_rate": 9.827407957820973e-06,
      "loss": 4.2333,
      "step": 12889
    },
    {
      "epoch": 0.5203875656035527,
      "grad_norm": 172.84478759765625,
      "learning_rate": 9.826100573672202e-06,
      "loss": 5.4927,
      "step": 12890
    },
    {
      "epoch": 0.5204279370205894,
      "grad_norm": 87.4063949584961,
      "learning_rate": 9.824793192496706e-06,
      "loss": 5.1112,
      "step": 12891
    },
    {
      "epoch": 0.5204683084376261,
      "grad_norm": 84.4478530883789,
      "learning_rate": 9.823485814316839e-06,
      "loss": 4.5486,
      "step": 12892
    },
    {
      "epoch": 0.5205086798546629,
      "grad_norm": 204.2192840576172,
      "learning_rate": 9.822178439154952e-06,
      "loss": 4.968,
      "step": 12893
    },
    {
      "epoch": 0.5205490512716996,
      "grad_norm": 87.60415649414062,
      "learning_rate": 9.820871067033396e-06,
      "loss": 5.212,
      "step": 12894
    },
    {
      "epoch": 0.5205894226887364,
      "grad_norm": 97.37789154052734,
      "learning_rate": 9.819563697974527e-06,
      "loss": 5.8956,
      "step": 12895
    },
    {
      "epoch": 0.5206297941057731,
      "grad_norm": 77.65106201171875,
      "learning_rate": 9.818256332000705e-06,
      "loss": 5.4336,
      "step": 12896
    },
    {
      "epoch": 0.5206701655228099,
      "grad_norm": 78.6217269897461,
      "learning_rate": 9.816948969134267e-06,
      "loss": 5.518,
      "step": 12897
    },
    {
      "epoch": 0.5207105369398466,
      "grad_norm": 71.69676971435547,
      "learning_rate": 9.815641609397576e-06,
      "loss": 5.9029,
      "step": 12898
    },
    {
      "epoch": 0.5207509083568833,
      "grad_norm": 82.45728302001953,
      "learning_rate": 9.814334252812982e-06,
      "loss": 4.2109,
      "step": 12899
    },
    {
      "epoch": 0.52079127977392,
      "grad_norm": 110.88651275634766,
      "learning_rate": 9.813026899402841e-06,
      "loss": 5.7598,
      "step": 12900
    },
    {
      "epoch": 0.5208316511909568,
      "grad_norm": 76.6964111328125,
      "learning_rate": 9.811719549189502e-06,
      "loss": 4.4858,
      "step": 12901
    },
    {
      "epoch": 0.5208720226079936,
      "grad_norm": 95.47428894042969,
      "learning_rate": 9.810412202195319e-06,
      "loss": 5.9338,
      "step": 12902
    },
    {
      "epoch": 0.5209123940250303,
      "grad_norm": 129.4057159423828,
      "learning_rate": 9.809104858442644e-06,
      "loss": 5.5675,
      "step": 12903
    },
    {
      "epoch": 0.520952765442067,
      "grad_norm": 91.80608367919922,
      "learning_rate": 9.807797517953835e-06,
      "loss": 4.6321,
      "step": 12904
    },
    {
      "epoch": 0.5209931368591038,
      "grad_norm": 95.85797119140625,
      "learning_rate": 9.806490180751235e-06,
      "loss": 5.6329,
      "step": 12905
    },
    {
      "epoch": 0.5210335082761405,
      "grad_norm": 121.9384994506836,
      "learning_rate": 9.805182846857201e-06,
      "loss": 4.4446,
      "step": 12906
    },
    {
      "epoch": 0.5210738796931772,
      "grad_norm": 175.7529296875,
      "learning_rate": 9.803875516294086e-06,
      "loss": 5.3271,
      "step": 12907
    },
    {
      "epoch": 0.5211142511102139,
      "grad_norm": 86.88606262207031,
      "learning_rate": 9.802568189084243e-06,
      "loss": 4.2549,
      "step": 12908
    },
    {
      "epoch": 0.5211546225272508,
      "grad_norm": 72.83551788330078,
      "learning_rate": 9.801260865250023e-06,
      "loss": 5.0036,
      "step": 12909
    },
    {
      "epoch": 0.5211949939442875,
      "grad_norm": 92.59716033935547,
      "learning_rate": 9.799953544813776e-06,
      "loss": 5.0489,
      "step": 12910
    },
    {
      "epoch": 0.5212353653613242,
      "grad_norm": 107.7115707397461,
      "learning_rate": 9.798646227797857e-06,
      "loss": 4.9688,
      "step": 12911
    },
    {
      "epoch": 0.5212757367783609,
      "grad_norm": 97.64439392089844,
      "learning_rate": 9.79733891422462e-06,
      "loss": 3.9254,
      "step": 12912
    },
    {
      "epoch": 0.5213161081953976,
      "grad_norm": 146.0726318359375,
      "learning_rate": 9.796031604116411e-06,
      "loss": 3.3452,
      "step": 12913
    },
    {
      "epoch": 0.5213564796124344,
      "grad_norm": 117.06204986572266,
      "learning_rate": 9.794724297495585e-06,
      "loss": 5.1146,
      "step": 12914
    },
    {
      "epoch": 0.5213968510294711,
      "grad_norm": 89.56405639648438,
      "learning_rate": 9.793416994384493e-06,
      "loss": 5.6188,
      "step": 12915
    },
    {
      "epoch": 0.5214372224465079,
      "grad_norm": 89.94197082519531,
      "learning_rate": 9.792109694805496e-06,
      "loss": 5.7091,
      "step": 12916
    },
    {
      "epoch": 0.5214775938635446,
      "grad_norm": 106.65960693359375,
      "learning_rate": 9.790802398780931e-06,
      "loss": 4.7658,
      "step": 12917
    },
    {
      "epoch": 0.5215179652805814,
      "grad_norm": 87.05933380126953,
      "learning_rate": 9.789495106333155e-06,
      "loss": 5.2886,
      "step": 12918
    },
    {
      "epoch": 0.5215583366976181,
      "grad_norm": 150.35104370117188,
      "learning_rate": 9.788187817484523e-06,
      "loss": 4.8042,
      "step": 12919
    },
    {
      "epoch": 0.5215987081146548,
      "grad_norm": 68.88916778564453,
      "learning_rate": 9.786880532257387e-06,
      "loss": 6.358,
      "step": 12920
    },
    {
      "epoch": 0.5216390795316915,
      "grad_norm": 87.40070343017578,
      "learning_rate": 9.785573250674094e-06,
      "loss": 4.3471,
      "step": 12921
    },
    {
      "epoch": 0.5216794509487283,
      "grad_norm": 99.03792572021484,
      "learning_rate": 9.784265972756997e-06,
      "loss": 5.8279,
      "step": 12922
    },
    {
      "epoch": 0.5217198223657651,
      "grad_norm": 132.80856323242188,
      "learning_rate": 9.78295869852845e-06,
      "loss": 4.6194,
      "step": 12923
    },
    {
      "epoch": 0.5217601937828018,
      "grad_norm": 75.25625610351562,
      "learning_rate": 9.781651428010806e-06,
      "loss": 5.3599,
      "step": 12924
    },
    {
      "epoch": 0.5218005651998385,
      "grad_norm": 111.27363586425781,
      "learning_rate": 9.780344161226407e-06,
      "loss": 5.8298,
      "step": 12925
    },
    {
      "epoch": 0.5218409366168753,
      "grad_norm": 100.10823059082031,
      "learning_rate": 9.77903689819761e-06,
      "loss": 5.106,
      "step": 12926
    },
    {
      "epoch": 0.521881308033912,
      "grad_norm": 99.81937408447266,
      "learning_rate": 9.777729638946767e-06,
      "loss": 5.5827,
      "step": 12927
    },
    {
      "epoch": 0.5219216794509487,
      "grad_norm": 102.18961334228516,
      "learning_rate": 9.77642238349623e-06,
      "loss": 6.1903,
      "step": 12928
    },
    {
      "epoch": 0.5219620508679854,
      "grad_norm": 77.56867980957031,
      "learning_rate": 9.775115131868346e-06,
      "loss": 5.7259,
      "step": 12929
    },
    {
      "epoch": 0.5220024222850222,
      "grad_norm": 81.87654876708984,
      "learning_rate": 9.773807884085467e-06,
      "loss": 5.2101,
      "step": 12930
    },
    {
      "epoch": 0.522042793702059,
      "grad_norm": 109.82852172851562,
      "learning_rate": 9.772500640169948e-06,
      "loss": 5.2358,
      "step": 12931
    },
    {
      "epoch": 0.5220831651190957,
      "grad_norm": 138.1845703125,
      "learning_rate": 9.771193400144139e-06,
      "loss": 5.4872,
      "step": 12932
    },
    {
      "epoch": 0.5221235365361324,
      "grad_norm": 117.84406280517578,
      "learning_rate": 9.769886164030383e-06,
      "loss": 4.0298,
      "step": 12933
    },
    {
      "epoch": 0.5221639079531691,
      "grad_norm": 77.14016723632812,
      "learning_rate": 9.768578931851038e-06,
      "loss": 6.229,
      "step": 12934
    },
    {
      "epoch": 0.5222042793702059,
      "grad_norm": 91.89705657958984,
      "learning_rate": 9.767271703628453e-06,
      "loss": 3.0014,
      "step": 12935
    },
    {
      "epoch": 0.5222446507872426,
      "grad_norm": 401.3735046386719,
      "learning_rate": 9.765964479384982e-06,
      "loss": 4.3716,
      "step": 12936
    },
    {
      "epoch": 0.5222850222042794,
      "grad_norm": 120.61125183105469,
      "learning_rate": 9.764657259142968e-06,
      "loss": 5.4591,
      "step": 12937
    },
    {
      "epoch": 0.5223253936213161,
      "grad_norm": 120.35581970214844,
      "learning_rate": 9.763350042924765e-06,
      "loss": 4.9582,
      "step": 12938
    },
    {
      "epoch": 0.5223657650383529,
      "grad_norm": 85.4492416381836,
      "learning_rate": 9.762042830752724e-06,
      "loss": 4.8765,
      "step": 12939
    },
    {
      "epoch": 0.5224061364553896,
      "grad_norm": 88.41842651367188,
      "learning_rate": 9.760735622649197e-06,
      "loss": 5.8752,
      "step": 12940
    },
    {
      "epoch": 0.5224465078724263,
      "grad_norm": 97.00037384033203,
      "learning_rate": 9.759428418636529e-06,
      "loss": 5.3919,
      "step": 12941
    },
    {
      "epoch": 0.522486879289463,
      "grad_norm": 129.23822021484375,
      "learning_rate": 9.758121218737075e-06,
      "loss": 5.0489,
      "step": 12942
    },
    {
      "epoch": 0.5225272507064997,
      "grad_norm": 100.82704162597656,
      "learning_rate": 9.756814022973183e-06,
      "loss": 5.2334,
      "step": 12943
    },
    {
      "epoch": 0.5225676221235366,
      "grad_norm": 131.42506408691406,
      "learning_rate": 9.755506831367206e-06,
      "loss": 5.6929,
      "step": 12944
    },
    {
      "epoch": 0.5226079935405733,
      "grad_norm": 107.33963012695312,
      "learning_rate": 9.754199643941488e-06,
      "loss": 3.5892,
      "step": 12945
    },
    {
      "epoch": 0.52264836495761,
      "grad_norm": 113.70782470703125,
      "learning_rate": 9.752892460718383e-06,
      "loss": 6.1776,
      "step": 12946
    },
    {
      "epoch": 0.5226887363746467,
      "grad_norm": 120.74057006835938,
      "learning_rate": 9.751585281720238e-06,
      "loss": 3.7145,
      "step": 12947
    },
    {
      "epoch": 0.5227291077916835,
      "grad_norm": 98.51691436767578,
      "learning_rate": 9.75027810696941e-06,
      "loss": 5.6692,
      "step": 12948
    },
    {
      "epoch": 0.5227694792087202,
      "grad_norm": 125.0078125,
      "learning_rate": 9.748970936488236e-06,
      "loss": 5.4545,
      "step": 12949
    },
    {
      "epoch": 0.5228098506257569,
      "grad_norm": 70.63623046875,
      "learning_rate": 9.747663770299076e-06,
      "loss": 4.86,
      "step": 12950
    },
    {
      "epoch": 0.5228502220427937,
      "grad_norm": 270.3041076660156,
      "learning_rate": 9.746356608424276e-06,
      "loss": 3.8634,
      "step": 12951
    },
    {
      "epoch": 0.5228905934598305,
      "grad_norm": 80.96704864501953,
      "learning_rate": 9.74504945088619e-06,
      "loss": 5.0931,
      "step": 12952
    },
    {
      "epoch": 0.5229309648768672,
      "grad_norm": 113.44126892089844,
      "learning_rate": 9.743742297707159e-06,
      "loss": 4.2031,
      "step": 12953
    },
    {
      "epoch": 0.5229713362939039,
      "grad_norm": 132.12501525878906,
      "learning_rate": 9.742435148909536e-06,
      "loss": 5.381,
      "step": 12954
    },
    {
      "epoch": 0.5230117077109406,
      "grad_norm": 108.75970458984375,
      "learning_rate": 9.74112800451567e-06,
      "loss": 4.7677,
      "step": 12955
    },
    {
      "epoch": 0.5230520791279774,
      "grad_norm": 69.94961547851562,
      "learning_rate": 9.739820864547916e-06,
      "loss": 5.4508,
      "step": 12956
    },
    {
      "epoch": 0.5230924505450141,
      "grad_norm": 87.51524353027344,
      "learning_rate": 9.738513729028612e-06,
      "loss": 6.969,
      "step": 12957
    },
    {
      "epoch": 0.5231328219620509,
      "grad_norm": 114.27394104003906,
      "learning_rate": 9.737206597980113e-06,
      "loss": 4.4381,
      "step": 12958
    },
    {
      "epoch": 0.5231731933790876,
      "grad_norm": 133.11932373046875,
      "learning_rate": 9.735899471424767e-06,
      "loss": 4.2895,
      "step": 12959
    },
    {
      "epoch": 0.5232135647961244,
      "grad_norm": 81.56620788574219,
      "learning_rate": 9.734592349384927e-06,
      "loss": 5.9854,
      "step": 12960
    },
    {
      "epoch": 0.5232539362131611,
      "grad_norm": 99.2729263305664,
      "learning_rate": 9.733285231882936e-06,
      "loss": 4.0366,
      "step": 12961
    },
    {
      "epoch": 0.5232943076301978,
      "grad_norm": 159.15631103515625,
      "learning_rate": 9.731978118941143e-06,
      "loss": 4.342,
      "step": 12962
    },
    {
      "epoch": 0.5233346790472345,
      "grad_norm": 134.2248077392578,
      "learning_rate": 9.7306710105819e-06,
      "loss": 4.7455,
      "step": 12963
    },
    {
      "epoch": 0.5233750504642712,
      "grad_norm": 109.48128509521484,
      "learning_rate": 9.729363906827557e-06,
      "loss": 6.0928,
      "step": 12964
    },
    {
      "epoch": 0.5234154218813081,
      "grad_norm": 82.13580322265625,
      "learning_rate": 9.728056807700456e-06,
      "loss": 4.7554,
      "step": 12965
    },
    {
      "epoch": 0.5234557932983448,
      "grad_norm": 92.16239929199219,
      "learning_rate": 9.726749713222946e-06,
      "loss": 5.2762,
      "step": 12966
    },
    {
      "epoch": 0.5234961647153815,
      "grad_norm": 76.72637939453125,
      "learning_rate": 9.72544262341738e-06,
      "loss": 5.3175,
      "step": 12967
    },
    {
      "epoch": 0.5235365361324182,
      "grad_norm": 81.75467681884766,
      "learning_rate": 9.724135538306107e-06,
      "loss": 5.3251,
      "step": 12968
    },
    {
      "epoch": 0.523576907549455,
      "grad_norm": 79.39073181152344,
      "learning_rate": 9.722828457911469e-06,
      "loss": 5.7581,
      "step": 12969
    },
    {
      "epoch": 0.5236172789664917,
      "grad_norm": 81.33010864257812,
      "learning_rate": 9.721521382255816e-06,
      "loss": 5.0442,
      "step": 12970
    },
    {
      "epoch": 0.5236576503835285,
      "grad_norm": 77.68692779541016,
      "learning_rate": 9.720214311361497e-06,
      "loss": 6.1565,
      "step": 12971
    },
    {
      "epoch": 0.5236980218005652,
      "grad_norm": 109.3318862915039,
      "learning_rate": 9.718907245250867e-06,
      "loss": 6.0151,
      "step": 12972
    },
    {
      "epoch": 0.523738393217602,
      "grad_norm": 80.23200988769531,
      "learning_rate": 9.717600183946259e-06,
      "loss": 5.5758,
      "step": 12973
    },
    {
      "epoch": 0.5237787646346387,
      "grad_norm": 102.87420654296875,
      "learning_rate": 9.71629312747003e-06,
      "loss": 4.8688,
      "step": 12974
    },
    {
      "epoch": 0.5238191360516754,
      "grad_norm": 129.40667724609375,
      "learning_rate": 9.714986075844525e-06,
      "loss": 5.0815,
      "step": 12975
    },
    {
      "epoch": 0.5238595074687121,
      "grad_norm": 91.48582458496094,
      "learning_rate": 9.7136790290921e-06,
      "loss": 5.6297,
      "step": 12976
    },
    {
      "epoch": 0.5238998788857488,
      "grad_norm": 120.9923095703125,
      "learning_rate": 9.712371987235087e-06,
      "loss": 5.8188,
      "step": 12977
    },
    {
      "epoch": 0.5239402503027857,
      "grad_norm": 69.94945526123047,
      "learning_rate": 9.711064950295843e-06,
      "loss": 3.7875,
      "step": 12978
    },
    {
      "epoch": 0.5239806217198224,
      "grad_norm": 76.48910522460938,
      "learning_rate": 9.709757918296715e-06,
      "loss": 6.2735,
      "step": 12979
    },
    {
      "epoch": 0.5240209931368591,
      "grad_norm": 71.8172378540039,
      "learning_rate": 9.708450891260048e-06,
      "loss": 4.393,
      "step": 12980
    },
    {
      "epoch": 0.5240613645538958,
      "grad_norm": 81.71551513671875,
      "learning_rate": 9.70714386920819e-06,
      "loss": 5.6848,
      "step": 12981
    },
    {
      "epoch": 0.5241017359709326,
      "grad_norm": 73.9614028930664,
      "learning_rate": 9.705836852163486e-06,
      "loss": 4.474,
      "step": 12982
    },
    {
      "epoch": 0.5241421073879693,
      "grad_norm": 139.58120727539062,
      "learning_rate": 9.704529840148286e-06,
      "loss": 4.9527,
      "step": 12983
    },
    {
      "epoch": 0.524182478805006,
      "grad_norm": 106.17616271972656,
      "learning_rate": 9.70322283318494e-06,
      "loss": 5.0419,
      "step": 12984
    },
    {
      "epoch": 0.5242228502220428,
      "grad_norm": 290.56146240234375,
      "learning_rate": 9.701915831295787e-06,
      "loss": 7.7596,
      "step": 12985
    },
    {
      "epoch": 0.5242632216390796,
      "grad_norm": 81.68404388427734,
      "learning_rate": 9.700608834503177e-06,
      "loss": 6.4939,
      "step": 12986
    },
    {
      "epoch": 0.5243035930561163,
      "grad_norm": 109.1982421875,
      "learning_rate": 9.699301842829456e-06,
      "loss": 6.1396,
      "step": 12987
    },
    {
      "epoch": 0.524343964473153,
      "grad_norm": 73.51302337646484,
      "learning_rate": 9.697994856296973e-06,
      "loss": 6.039,
      "step": 12988
    },
    {
      "epoch": 0.5243843358901897,
      "grad_norm": 104.92185974121094,
      "learning_rate": 9.696687874928073e-06,
      "loss": 4.0878,
      "step": 12989
    },
    {
      "epoch": 0.5244247073072265,
      "grad_norm": 83.583984375,
      "learning_rate": 9.6953808987451e-06,
      "loss": 6.7405,
      "step": 12990
    },
    {
      "epoch": 0.5244650787242632,
      "grad_norm": 123.92001342773438,
      "learning_rate": 9.694073927770404e-06,
      "loss": 4.5315,
      "step": 12991
    },
    {
      "epoch": 0.5245054501413,
      "grad_norm": 52.62592697143555,
      "learning_rate": 9.692766962026332e-06,
      "loss": 3.822,
      "step": 12992
    },
    {
      "epoch": 0.5245458215583367,
      "grad_norm": 117.37310028076172,
      "learning_rate": 9.691460001535223e-06,
      "loss": 5.6375,
      "step": 12993
    },
    {
      "epoch": 0.5245861929753735,
      "grad_norm": 126.126220703125,
      "learning_rate": 9.690153046319428e-06,
      "loss": 2.6867,
      "step": 12994
    },
    {
      "epoch": 0.5246265643924102,
      "grad_norm": 71.1406021118164,
      "learning_rate": 9.688846096401293e-06,
      "loss": 6.0054,
      "step": 12995
    },
    {
      "epoch": 0.5246669358094469,
      "grad_norm": 98.34056854248047,
      "learning_rate": 9.687539151803166e-06,
      "loss": 5.2154,
      "step": 12996
    },
    {
      "epoch": 0.5247073072264836,
      "grad_norm": 73.80342102050781,
      "learning_rate": 9.686232212547388e-06,
      "loss": 6.0706,
      "step": 12997
    },
    {
      "epoch": 0.5247476786435203,
      "grad_norm": 129.93121337890625,
      "learning_rate": 9.684925278656305e-06,
      "loss": 5.326,
      "step": 12998
    },
    {
      "epoch": 0.5247880500605572,
      "grad_norm": 105.77167510986328,
      "learning_rate": 9.683618350152264e-06,
      "loss": 4.5301,
      "step": 12999
    },
    {
      "epoch": 0.5248284214775939,
      "grad_norm": 104.35220336914062,
      "learning_rate": 9.682311427057612e-06,
      "loss": 4.6064,
      "step": 13000
    },
    {
      "epoch": 0.5248687928946306,
      "grad_norm": 157.5475311279297,
      "learning_rate": 9.681004509394691e-06,
      "loss": 4.1661,
      "step": 13001
    },
    {
      "epoch": 0.5249091643116673,
      "grad_norm": 106.40040588378906,
      "learning_rate": 9.679697597185848e-06,
      "loss": 5.5993,
      "step": 13002
    },
    {
      "epoch": 0.5249495357287041,
      "grad_norm": 78.08507537841797,
      "learning_rate": 9.67839069045343e-06,
      "loss": 4.6996,
      "step": 13003
    },
    {
      "epoch": 0.5249899071457408,
      "grad_norm": 72.43316650390625,
      "learning_rate": 9.677083789219782e-06,
      "loss": 5.2255,
      "step": 13004
    },
    {
      "epoch": 0.5250302785627775,
      "grad_norm": 66.11639404296875,
      "learning_rate": 9.675776893507244e-06,
      "loss": 5.7693,
      "step": 13005
    },
    {
      "epoch": 0.5250706499798143,
      "grad_norm": 99.2022933959961,
      "learning_rate": 9.674470003338164e-06,
      "loss": 6.6982,
      "step": 13006
    },
    {
      "epoch": 0.5251110213968511,
      "grad_norm": 79.66119384765625,
      "learning_rate": 9.673163118734886e-06,
      "loss": 5.7633,
      "step": 13007
    },
    {
      "epoch": 0.5251513928138878,
      "grad_norm": 109.11022186279297,
      "learning_rate": 9.671856239719758e-06,
      "loss": 6.8119,
      "step": 13008
    },
    {
      "epoch": 0.5251917642309245,
      "grad_norm": 81.70024108886719,
      "learning_rate": 9.67054936631512e-06,
      "loss": 5.1552,
      "step": 13009
    },
    {
      "epoch": 0.5252321356479612,
      "grad_norm": 77.89676666259766,
      "learning_rate": 9.669242498543319e-06,
      "loss": 6.0032,
      "step": 13010
    },
    {
      "epoch": 0.525272507064998,
      "grad_norm": 61.588924407958984,
      "learning_rate": 9.667935636426697e-06,
      "loss": 5.5028,
      "step": 13011
    },
    {
      "epoch": 0.5253128784820347,
      "grad_norm": 87.1718978881836,
      "learning_rate": 9.666628779987608e-06,
      "loss": 4.8767,
      "step": 13012
    },
    {
      "epoch": 0.5253532498990715,
      "grad_norm": 76.24858856201172,
      "learning_rate": 9.66532192924838e-06,
      "loss": 4.2731,
      "step": 13013
    },
    {
      "epoch": 0.5253936213161082,
      "grad_norm": 78.14424133300781,
      "learning_rate": 9.664015084231367e-06,
      "loss": 4.1465,
      "step": 13014
    },
    {
      "epoch": 0.525433992733145,
      "grad_norm": 105.15448760986328,
      "learning_rate": 9.662708244958913e-06,
      "loss": 5.502,
      "step": 13015
    },
    {
      "epoch": 0.5254743641501817,
      "grad_norm": 115.55838775634766,
      "learning_rate": 9.66140141145336e-06,
      "loss": 4.0453,
      "step": 13016
    },
    {
      "epoch": 0.5255147355672184,
      "grad_norm": 79.27276611328125,
      "learning_rate": 9.660094583737051e-06,
      "loss": 3.2888,
      "step": 13017
    },
    {
      "epoch": 0.5255551069842551,
      "grad_norm": 88.66140747070312,
      "learning_rate": 9.658787761832332e-06,
      "loss": 5.9637,
      "step": 13018
    },
    {
      "epoch": 0.5255954784012918,
      "grad_norm": 83.91536712646484,
      "learning_rate": 9.657480945761546e-06,
      "loss": 4.2412,
      "step": 13019
    },
    {
      "epoch": 0.5256358498183287,
      "grad_norm": 97.51986694335938,
      "learning_rate": 9.656174135547038e-06,
      "loss": 4.0486,
      "step": 13020
    },
    {
      "epoch": 0.5256762212353654,
      "grad_norm": 126.95755004882812,
      "learning_rate": 9.654867331211146e-06,
      "loss": 5.2117,
      "step": 13021
    },
    {
      "epoch": 0.5257165926524021,
      "grad_norm": 58.91423034667969,
      "learning_rate": 9.653560532776217e-06,
      "loss": 5.4522,
      "step": 13022
    },
    {
      "epoch": 0.5257569640694388,
      "grad_norm": 167.94879150390625,
      "learning_rate": 9.652253740264594e-06,
      "loss": 5.8829,
      "step": 13023
    },
    {
      "epoch": 0.5257973354864756,
      "grad_norm": 85.93579864501953,
      "learning_rate": 9.650946953698624e-06,
      "loss": 4.0194,
      "step": 13024
    },
    {
      "epoch": 0.5258377069035123,
      "grad_norm": 90.89681243896484,
      "learning_rate": 9.649640173100642e-06,
      "loss": 5.4557,
      "step": 13025
    },
    {
      "epoch": 0.525878078320549,
      "grad_norm": 131.77963256835938,
      "learning_rate": 9.648333398492995e-06,
      "loss": 4.9594,
      "step": 13026
    },
    {
      "epoch": 0.5259184497375858,
      "grad_norm": 118.70753479003906,
      "learning_rate": 9.647026629898024e-06,
      "loss": 5.4376,
      "step": 13027
    },
    {
      "epoch": 0.5259588211546226,
      "grad_norm": 93.29893493652344,
      "learning_rate": 9.645719867338079e-06,
      "loss": 4.8025,
      "step": 13028
    },
    {
      "epoch": 0.5259991925716593,
      "grad_norm": 88.11968994140625,
      "learning_rate": 9.64441311083549e-06,
      "loss": 4.5458,
      "step": 13029
    },
    {
      "epoch": 0.526039563988696,
      "grad_norm": 96.82975006103516,
      "learning_rate": 9.643106360412611e-06,
      "loss": 3.354,
      "step": 13030
    },
    {
      "epoch": 0.5260799354057327,
      "grad_norm": 97.98095703125,
      "learning_rate": 9.641799616091778e-06,
      "loss": 5.3765,
      "step": 13031
    },
    {
      "epoch": 0.5261203068227694,
      "grad_norm": 96.37726593017578,
      "learning_rate": 9.640492877895338e-06,
      "loss": 6.0682,
      "step": 13032
    },
    {
      "epoch": 0.5261606782398062,
      "grad_norm": 115.01148986816406,
      "learning_rate": 9.639186145845629e-06,
      "loss": 5.7156,
      "step": 13033
    },
    {
      "epoch": 0.526201049656843,
      "grad_norm": 77.62154388427734,
      "learning_rate": 9.637879419964992e-06,
      "loss": 4.9054,
      "step": 13034
    },
    {
      "epoch": 0.5262414210738797,
      "grad_norm": 257.57122802734375,
      "learning_rate": 9.636572700275772e-06,
      "loss": 5.0996,
      "step": 13035
    },
    {
      "epoch": 0.5262817924909164,
      "grad_norm": 277.1275329589844,
      "learning_rate": 9.635265986800312e-06,
      "loss": 5.7662,
      "step": 13036
    },
    {
      "epoch": 0.5263221639079532,
      "grad_norm": 169.28756713867188,
      "learning_rate": 9.63395927956095e-06,
      "loss": 4.8701,
      "step": 13037
    },
    {
      "epoch": 0.5263625353249899,
      "grad_norm": 87.74818420410156,
      "learning_rate": 9.63265257858003e-06,
      "loss": 5.5675,
      "step": 13038
    },
    {
      "epoch": 0.5264029067420266,
      "grad_norm": 74.29386138916016,
      "learning_rate": 9.631345883879892e-06,
      "loss": 3.9297,
      "step": 13039
    },
    {
      "epoch": 0.5264432781590633,
      "grad_norm": 92.27644348144531,
      "learning_rate": 9.630039195482883e-06,
      "loss": 3.8369,
      "step": 13040
    },
    {
      "epoch": 0.5264836495761002,
      "grad_norm": 657.1279296875,
      "learning_rate": 9.628732513411336e-06,
      "loss": 4.0581,
      "step": 13041
    },
    {
      "epoch": 0.5265240209931369,
      "grad_norm": 69.46623992919922,
      "learning_rate": 9.627425837687596e-06,
      "loss": 4.7196,
      "step": 13042
    },
    {
      "epoch": 0.5265643924101736,
      "grad_norm": 110.52075958251953,
      "learning_rate": 9.626119168334005e-06,
      "loss": 3.1185,
      "step": 13043
    },
    {
      "epoch": 0.5266047638272103,
      "grad_norm": 140.84738159179688,
      "learning_rate": 9.624812505372907e-06,
      "loss": 5.8139,
      "step": 13044
    },
    {
      "epoch": 0.526645135244247,
      "grad_norm": 102.8874740600586,
      "learning_rate": 9.623505848826635e-06,
      "loss": 5.8283,
      "step": 13045
    },
    {
      "epoch": 0.5266855066612838,
      "grad_norm": 105.41230010986328,
      "learning_rate": 9.622199198717532e-06,
      "loss": 4.5139,
      "step": 13046
    },
    {
      "epoch": 0.5267258780783205,
      "grad_norm": 102.01422882080078,
      "learning_rate": 9.620892555067942e-06,
      "loss": 4.2162,
      "step": 13047
    },
    {
      "epoch": 0.5267662494953573,
      "grad_norm": 418.9230651855469,
      "learning_rate": 9.619585917900205e-06,
      "loss": 5.1839,
      "step": 13048
    },
    {
      "epoch": 0.526806620912394,
      "grad_norm": 125.88333892822266,
      "learning_rate": 9.61827928723666e-06,
      "loss": 6.1963,
      "step": 13049
    },
    {
      "epoch": 0.5268469923294308,
      "grad_norm": 166.1443328857422,
      "learning_rate": 9.616972663099648e-06,
      "loss": 6.0789,
      "step": 13050
    },
    {
      "epoch": 0.5268873637464675,
      "grad_norm": 110.37711334228516,
      "learning_rate": 9.615666045511508e-06,
      "loss": 4.2757,
      "step": 13051
    },
    {
      "epoch": 0.5269277351635042,
      "grad_norm": 82.24394989013672,
      "learning_rate": 9.614359434494585e-06,
      "loss": 5.1604,
      "step": 13052
    },
    {
      "epoch": 0.5269681065805409,
      "grad_norm": 106.59339141845703,
      "learning_rate": 9.61305283007121e-06,
      "loss": 3.6602,
      "step": 13053
    },
    {
      "epoch": 0.5270084779975777,
      "grad_norm": 153.19448852539062,
      "learning_rate": 9.61174623226373e-06,
      "loss": 5.5904,
      "step": 13054
    },
    {
      "epoch": 0.5270488494146145,
      "grad_norm": 83.53247833251953,
      "learning_rate": 9.610439641094483e-06,
      "loss": 5.0576,
      "step": 13055
    },
    {
      "epoch": 0.5270892208316512,
      "grad_norm": 132.05441284179688,
      "learning_rate": 9.609133056585809e-06,
      "loss": 4.6454,
      "step": 13056
    },
    {
      "epoch": 0.5271295922486879,
      "grad_norm": 115.4595947265625,
      "learning_rate": 9.607826478760045e-06,
      "loss": 5.2589,
      "step": 13057
    },
    {
      "epoch": 0.5271699636657247,
      "grad_norm": 93.31535339355469,
      "learning_rate": 9.606519907639532e-06,
      "loss": 5.4081,
      "step": 13058
    },
    {
      "epoch": 0.5272103350827614,
      "grad_norm": 103.85916900634766,
      "learning_rate": 9.605213343246612e-06,
      "loss": 4.3252,
      "step": 13059
    },
    {
      "epoch": 0.5272507064997981,
      "grad_norm": 70.73499298095703,
      "learning_rate": 9.603906785603625e-06,
      "loss": 5.1711,
      "step": 13060
    },
    {
      "epoch": 0.5272910779168348,
      "grad_norm": 111.22834777832031,
      "learning_rate": 9.602600234732903e-06,
      "loss": 6.0366,
      "step": 13061
    },
    {
      "epoch": 0.5273314493338717,
      "grad_norm": 95.24835205078125,
      "learning_rate": 9.601293690656789e-06,
      "loss": 6.7826,
      "step": 13062
    },
    {
      "epoch": 0.5273718207509084,
      "grad_norm": 77.66283416748047,
      "learning_rate": 9.59998715339762e-06,
      "loss": 4.217,
      "step": 13063
    },
    {
      "epoch": 0.5274121921679451,
      "grad_norm": 76.86018371582031,
      "learning_rate": 9.598680622977742e-06,
      "loss": 4.1184,
      "step": 13064
    },
    {
      "epoch": 0.5274525635849818,
      "grad_norm": 175.1450653076172,
      "learning_rate": 9.597374099419485e-06,
      "loss": 6.0216,
      "step": 13065
    },
    {
      "epoch": 0.5274929350020185,
      "grad_norm": 87.392578125,
      "learning_rate": 9.59606758274519e-06,
      "loss": 3.2778,
      "step": 13066
    },
    {
      "epoch": 0.5275333064190553,
      "grad_norm": 75.14799499511719,
      "learning_rate": 9.594761072977196e-06,
      "loss": 5.459,
      "step": 13067
    },
    {
      "epoch": 0.527573677836092,
      "grad_norm": 95.1840591430664,
      "learning_rate": 9.593454570137844e-06,
      "loss": 4.3973,
      "step": 13068
    },
    {
      "epoch": 0.5276140492531288,
      "grad_norm": 95.24482727050781,
      "learning_rate": 9.592148074249467e-06,
      "loss": 5.6103,
      "step": 13069
    },
    {
      "epoch": 0.5276544206701655,
      "grad_norm": 74.90263366699219,
      "learning_rate": 9.590841585334405e-06,
      "loss": 4.5385,
      "step": 13070
    },
    {
      "epoch": 0.5276947920872023,
      "grad_norm": 86.5296630859375,
      "learning_rate": 9.589535103414997e-06,
      "loss": 5.2764,
      "step": 13071
    },
    {
      "epoch": 0.527735163504239,
      "grad_norm": 76.16368103027344,
      "learning_rate": 9.588228628513586e-06,
      "loss": 3.6088,
      "step": 13072
    },
    {
      "epoch": 0.5277755349212757,
      "grad_norm": 95.73694610595703,
      "learning_rate": 9.586922160652497e-06,
      "loss": 5.7251,
      "step": 13073
    },
    {
      "epoch": 0.5278159063383124,
      "grad_norm": 122.9429931640625,
      "learning_rate": 9.585615699854076e-06,
      "loss": 5.1297,
      "step": 13074
    },
    {
      "epoch": 0.5278562777553493,
      "grad_norm": 101.53396606445312,
      "learning_rate": 9.584309246140658e-06,
      "loss": 6.7046,
      "step": 13075
    },
    {
      "epoch": 0.527896649172386,
      "grad_norm": 115.4732666015625,
      "learning_rate": 9.583002799534584e-06,
      "loss": 4.7148,
      "step": 13076
    },
    {
      "epoch": 0.5279370205894227,
      "grad_norm": 91.20856475830078,
      "learning_rate": 9.581696360058186e-06,
      "loss": 5.5582,
      "step": 13077
    },
    {
      "epoch": 0.5279773920064594,
      "grad_norm": 74.97940063476562,
      "learning_rate": 9.580389927733803e-06,
      "loss": 3.1765,
      "step": 13078
    },
    {
      "epoch": 0.5280177634234962,
      "grad_norm": 86.48502349853516,
      "learning_rate": 9.579083502583772e-06,
      "loss": 4.8869,
      "step": 13079
    },
    {
      "epoch": 0.5280581348405329,
      "grad_norm": 105.43688201904297,
      "learning_rate": 9.577777084630435e-06,
      "loss": 6.1357,
      "step": 13080
    },
    {
      "epoch": 0.5280985062575696,
      "grad_norm": 92.6167984008789,
      "learning_rate": 9.57647067389612e-06,
      "loss": 4.5673,
      "step": 13081
    },
    {
      "epoch": 0.5281388776746064,
      "grad_norm": 90.94170379638672,
      "learning_rate": 9.575164270403167e-06,
      "loss": 4.6646,
      "step": 13082
    },
    {
      "epoch": 0.5281792490916432,
      "grad_norm": 76.93077850341797,
      "learning_rate": 9.573857874173915e-06,
      "loss": 5.5492,
      "step": 13083
    },
    {
      "epoch": 0.5282196205086799,
      "grad_norm": 99.79369354248047,
      "learning_rate": 9.5725514852307e-06,
      "loss": 4.7434,
      "step": 13084
    },
    {
      "epoch": 0.5282599919257166,
      "grad_norm": 97.76136016845703,
      "learning_rate": 9.571245103595852e-06,
      "loss": 5.486,
      "step": 13085
    },
    {
      "epoch": 0.5283003633427533,
      "grad_norm": 106.07743072509766,
      "learning_rate": 9.569938729291712e-06,
      "loss": 4.3354,
      "step": 13086
    },
    {
      "epoch": 0.52834073475979,
      "grad_norm": 94.64574432373047,
      "learning_rate": 9.568632362340616e-06,
      "loss": 4.6238,
      "step": 13087
    },
    {
      "epoch": 0.5283811061768268,
      "grad_norm": 169.92837524414062,
      "learning_rate": 9.567326002764902e-06,
      "loss": 4.6201,
      "step": 13088
    },
    {
      "epoch": 0.5284214775938636,
      "grad_norm": 265.29205322265625,
      "learning_rate": 9.5660196505869e-06,
      "loss": 5.8825,
      "step": 13089
    },
    {
      "epoch": 0.5284618490109003,
      "grad_norm": 92.21521759033203,
      "learning_rate": 9.56471330582895e-06,
      "loss": 5.1956,
      "step": 13090
    },
    {
      "epoch": 0.528502220427937,
      "grad_norm": 120.30681610107422,
      "learning_rate": 9.563406968513383e-06,
      "loss": 4.7935,
      "step": 13091
    },
    {
      "epoch": 0.5285425918449738,
      "grad_norm": 79.63423156738281,
      "learning_rate": 9.562100638662544e-06,
      "loss": 4.0916,
      "step": 13092
    },
    {
      "epoch": 0.5285829632620105,
      "grad_norm": 126.64388275146484,
      "learning_rate": 9.560794316298755e-06,
      "loss": 4.2452,
      "step": 13093
    },
    {
      "epoch": 0.5286233346790472,
      "grad_norm": 135.5755615234375,
      "learning_rate": 9.55948800144436e-06,
      "loss": 4.0312,
      "step": 13094
    },
    {
      "epoch": 0.5286637060960839,
      "grad_norm": 127.59461212158203,
      "learning_rate": 9.558181694121692e-06,
      "loss": 5.4593,
      "step": 13095
    },
    {
      "epoch": 0.5287040775131208,
      "grad_norm": 114.36958312988281,
      "learning_rate": 9.556875394353085e-06,
      "loss": 5.8447,
      "step": 13096
    },
    {
      "epoch": 0.5287444489301575,
      "grad_norm": 78.89656829833984,
      "learning_rate": 9.555569102160871e-06,
      "loss": 5.0537,
      "step": 13097
    },
    {
      "epoch": 0.5287848203471942,
      "grad_norm": 113.05970001220703,
      "learning_rate": 9.55426281756739e-06,
      "loss": 2.4521,
      "step": 13098
    },
    {
      "epoch": 0.5288251917642309,
      "grad_norm": 120.99454498291016,
      "learning_rate": 9.552956540594975e-06,
      "loss": 5.4483,
      "step": 13099
    },
    {
      "epoch": 0.5288655631812677,
      "grad_norm": 116.80854797363281,
      "learning_rate": 9.55165027126596e-06,
      "loss": 5.5311,
      "step": 13100
    },
    {
      "epoch": 0.5289059345983044,
      "grad_norm": 77.2682113647461,
      "learning_rate": 9.550344009602676e-06,
      "loss": 5.179,
      "step": 13101
    },
    {
      "epoch": 0.5289463060153411,
      "grad_norm": 93.02064514160156,
      "learning_rate": 9.549037755627461e-06,
      "loss": 5.182,
      "step": 13102
    },
    {
      "epoch": 0.5289866774323779,
      "grad_norm": 73.94316864013672,
      "learning_rate": 9.547731509362646e-06,
      "loss": 5.3098,
      "step": 13103
    },
    {
      "epoch": 0.5290270488494146,
      "grad_norm": 85.95338439941406,
      "learning_rate": 9.54642527083057e-06,
      "loss": 4.8264,
      "step": 13104
    },
    {
      "epoch": 0.5290674202664514,
      "grad_norm": 89.27181243896484,
      "learning_rate": 9.545119040053557e-06,
      "loss": 4.3274,
      "step": 13105
    },
    {
      "epoch": 0.5291077916834881,
      "grad_norm": 87.39707946777344,
      "learning_rate": 9.543812817053949e-06,
      "loss": 4.2496,
      "step": 13106
    },
    {
      "epoch": 0.5291481631005248,
      "grad_norm": 90.55233764648438,
      "learning_rate": 9.542506601854075e-06,
      "loss": 4.9083,
      "step": 13107
    },
    {
      "epoch": 0.5291885345175615,
      "grad_norm": 115.81477355957031,
      "learning_rate": 9.54120039447627e-06,
      "loss": 5.3433,
      "step": 13108
    },
    {
      "epoch": 0.5292289059345983,
      "grad_norm": 170.0333709716797,
      "learning_rate": 9.539894194942868e-06,
      "loss": 6.0437,
      "step": 13109
    },
    {
      "epoch": 0.5292692773516351,
      "grad_norm": 85.02544403076172,
      "learning_rate": 9.5385880032762e-06,
      "loss": 3.9833,
      "step": 13110
    },
    {
      "epoch": 0.5293096487686718,
      "grad_norm": 86.17255401611328,
      "learning_rate": 9.537281819498599e-06,
      "loss": 5.2314,
      "step": 13111
    },
    {
      "epoch": 0.5293500201857085,
      "grad_norm": 95.77899932861328,
      "learning_rate": 9.535975643632403e-06,
      "loss": 7.6847,
      "step": 13112
    },
    {
      "epoch": 0.5293903916027453,
      "grad_norm": 142.7754669189453,
      "learning_rate": 9.534669475699936e-06,
      "loss": 4.8515,
      "step": 13113
    },
    {
      "epoch": 0.529430763019782,
      "grad_norm": 103.55778503417969,
      "learning_rate": 9.533363315723533e-06,
      "loss": 3.6676,
      "step": 13114
    },
    {
      "epoch": 0.5294711344368187,
      "grad_norm": 82.38861846923828,
      "learning_rate": 9.53205716372553e-06,
      "loss": 4.9383,
      "step": 13115
    },
    {
      "epoch": 0.5295115058538554,
      "grad_norm": 97.25376892089844,
      "learning_rate": 9.530751019728258e-06,
      "loss": 6.051,
      "step": 13116
    },
    {
      "epoch": 0.5295518772708923,
      "grad_norm": 101.64241027832031,
      "learning_rate": 9.529444883754044e-06,
      "loss": 4.726,
      "step": 13117
    },
    {
      "epoch": 0.529592248687929,
      "grad_norm": 125.15950775146484,
      "learning_rate": 9.528138755825224e-06,
      "loss": 4.7122,
      "step": 13118
    },
    {
      "epoch": 0.5296326201049657,
      "grad_norm": 106.54061126708984,
      "learning_rate": 9.526832635964132e-06,
      "loss": 5.7616,
      "step": 13119
    },
    {
      "epoch": 0.5296729915220024,
      "grad_norm": 165.63743591308594,
      "learning_rate": 9.525526524193099e-06,
      "loss": 5.0491,
      "step": 13120
    },
    {
      "epoch": 0.5297133629390391,
      "grad_norm": 154.33761596679688,
      "learning_rate": 9.524220420534451e-06,
      "loss": 4.8411,
      "step": 13121
    },
    {
      "epoch": 0.5297537343560759,
      "grad_norm": 87.58538055419922,
      "learning_rate": 9.522914325010521e-06,
      "loss": 6.1481,
      "step": 13122
    },
    {
      "epoch": 0.5297941057731126,
      "grad_norm": 145.31307983398438,
      "learning_rate": 9.521608237643644e-06,
      "loss": 5.5194,
      "step": 13123
    },
    {
      "epoch": 0.5298344771901494,
      "grad_norm": 113.79942321777344,
      "learning_rate": 9.520302158456152e-06,
      "loss": 6.1281,
      "step": 13124
    },
    {
      "epoch": 0.5298748486071861,
      "grad_norm": 79.32284545898438,
      "learning_rate": 9.518996087470369e-06,
      "loss": 5.3723,
      "step": 13125
    },
    {
      "epoch": 0.5299152200242229,
      "grad_norm": 89.62739562988281,
      "learning_rate": 9.517690024708628e-06,
      "loss": 7.0357,
      "step": 13126
    },
    {
      "epoch": 0.5299555914412596,
      "grad_norm": 139.86517333984375,
      "learning_rate": 9.516383970193265e-06,
      "loss": 4.7294,
      "step": 13127
    },
    {
      "epoch": 0.5299959628582963,
      "grad_norm": 103.54766082763672,
      "learning_rate": 9.51507792394661e-06,
      "loss": 4.0789,
      "step": 13128
    },
    {
      "epoch": 0.530036334275333,
      "grad_norm": 154.6918182373047,
      "learning_rate": 9.513771885990985e-06,
      "loss": 3.3061,
      "step": 13129
    },
    {
      "epoch": 0.5300767056923698,
      "grad_norm": 102.92189025878906,
      "learning_rate": 9.512465856348724e-06,
      "loss": 4.151,
      "step": 13130
    },
    {
      "epoch": 0.5301170771094066,
      "grad_norm": 75.92359924316406,
      "learning_rate": 9.51115983504216e-06,
      "loss": 4.8376,
      "step": 13131
    },
    {
      "epoch": 0.5301574485264433,
      "grad_norm": 90.87604522705078,
      "learning_rate": 9.509853822093626e-06,
      "loss": 5.5909,
      "step": 13132
    },
    {
      "epoch": 0.53019781994348,
      "grad_norm": 95.02342224121094,
      "learning_rate": 9.508547817525442e-06,
      "loss": 4.3542,
      "step": 13133
    },
    {
      "epoch": 0.5302381913605168,
      "grad_norm": 88.28479766845703,
      "learning_rate": 9.507241821359944e-06,
      "loss": 5.6648,
      "step": 13134
    },
    {
      "epoch": 0.5302785627775535,
      "grad_norm": 184.44674682617188,
      "learning_rate": 9.50593583361946e-06,
      "loss": 4.4779,
      "step": 13135
    },
    {
      "epoch": 0.5303189341945902,
      "grad_norm": 90.1199951171875,
      "learning_rate": 9.50462985432632e-06,
      "loss": 5.9045,
      "step": 13136
    },
    {
      "epoch": 0.5303593056116269,
      "grad_norm": 84.80854797363281,
      "learning_rate": 9.503323883502853e-06,
      "loss": 5.2719,
      "step": 13137
    },
    {
      "epoch": 0.5303996770286638,
      "grad_norm": 94.09040832519531,
      "learning_rate": 9.502017921171384e-06,
      "loss": 4.9161,
      "step": 13138
    },
    {
      "epoch": 0.5304400484457005,
      "grad_norm": 67.08263397216797,
      "learning_rate": 9.500711967354249e-06,
      "loss": 3.8187,
      "step": 13139
    },
    {
      "epoch": 0.5304804198627372,
      "grad_norm": 105.88143920898438,
      "learning_rate": 9.499406022073776e-06,
      "loss": 3.7126,
      "step": 13140
    },
    {
      "epoch": 0.5305207912797739,
      "grad_norm": 111.41484832763672,
      "learning_rate": 9.498100085352288e-06,
      "loss": 5.6561,
      "step": 13141
    },
    {
      "epoch": 0.5305611626968106,
      "grad_norm": 92.37822723388672,
      "learning_rate": 9.496794157212116e-06,
      "loss": 5.6009,
      "step": 13142
    },
    {
      "epoch": 0.5306015341138474,
      "grad_norm": 99.83678436279297,
      "learning_rate": 9.495488237675588e-06,
      "loss": 6.4678,
      "step": 13143
    },
    {
      "epoch": 0.5306419055308841,
      "grad_norm": 212.10708618164062,
      "learning_rate": 9.494182326765037e-06,
      "loss": 6.7786,
      "step": 13144
    },
    {
      "epoch": 0.5306822769479209,
      "grad_norm": 74.38914489746094,
      "learning_rate": 9.492876424502784e-06,
      "loss": 5.5513,
      "step": 13145
    },
    {
      "epoch": 0.5307226483649576,
      "grad_norm": 118.33104705810547,
      "learning_rate": 9.49157053091116e-06,
      "loss": 4.5701,
      "step": 13146
    },
    {
      "epoch": 0.5307630197819944,
      "grad_norm": 88.75879669189453,
      "learning_rate": 9.49026464601249e-06,
      "loss": 4.9833,
      "step": 13147
    },
    {
      "epoch": 0.5308033911990311,
      "grad_norm": 60.276527404785156,
      "learning_rate": 9.488958769829113e-06,
      "loss": 4.3395,
      "step": 13148
    },
    {
      "epoch": 0.5308437626160678,
      "grad_norm": 77.07808685302734,
      "learning_rate": 9.487652902383341e-06,
      "loss": 5.2213,
      "step": 13149
    },
    {
      "epoch": 0.5308841340331045,
      "grad_norm": 140.00982666015625,
      "learning_rate": 9.486347043697508e-06,
      "loss": 4.6183,
      "step": 13150
    },
    {
      "epoch": 0.5309245054501412,
      "grad_norm": 105.0823745727539,
      "learning_rate": 9.48504119379394e-06,
      "loss": 4.9798,
      "step": 13151
    },
    {
      "epoch": 0.5309648768671781,
      "grad_norm": 105.04080200195312,
      "learning_rate": 9.483735352694971e-06,
      "loss": 5.0458,
      "step": 13152
    },
    {
      "epoch": 0.5310052482842148,
      "grad_norm": 70.3246078491211,
      "learning_rate": 9.482429520422918e-06,
      "loss": 4.3761,
      "step": 13153
    },
    {
      "epoch": 0.5310456197012515,
      "grad_norm": 84.31159210205078,
      "learning_rate": 9.481123697000111e-06,
      "loss": 5.5728,
      "step": 13154
    },
    {
      "epoch": 0.5310859911182882,
      "grad_norm": 139.37796020507812,
      "learning_rate": 9.479817882448877e-06,
      "loss": 6.0821,
      "step": 13155
    },
    {
      "epoch": 0.531126362535325,
      "grad_norm": 99.42225646972656,
      "learning_rate": 9.478512076791545e-06,
      "loss": 5.1957,
      "step": 13156
    },
    {
      "epoch": 0.5311667339523617,
      "grad_norm": 90.97024536132812,
      "learning_rate": 9.477206280050435e-06,
      "loss": 5.003,
      "step": 13157
    },
    {
      "epoch": 0.5312071053693984,
      "grad_norm": 139.72085571289062,
      "learning_rate": 9.47590049224788e-06,
      "loss": 4.313,
      "step": 13158
    },
    {
      "epoch": 0.5312474767864352,
      "grad_norm": 105.22380065917969,
      "learning_rate": 9.4745947134062e-06,
      "loss": 6.0835,
      "step": 13159
    },
    {
      "epoch": 0.531287848203472,
      "grad_norm": 108.67174530029297,
      "learning_rate": 9.473288943547729e-06,
      "loss": 4.353,
      "step": 13160
    },
    {
      "epoch": 0.5313282196205087,
      "grad_norm": 132.3981475830078,
      "learning_rate": 9.471983182694782e-06,
      "loss": 4.1437,
      "step": 13161
    },
    {
      "epoch": 0.5313685910375454,
      "grad_norm": 110.0034408569336,
      "learning_rate": 9.47067743086969e-06,
      "loss": 5.7253,
      "step": 13162
    },
    {
      "epoch": 0.5314089624545821,
      "grad_norm": 369.7892150878906,
      "learning_rate": 9.469371688094777e-06,
      "loss": 4.2857,
      "step": 13163
    },
    {
      "epoch": 0.5314493338716189,
      "grad_norm": 78.46788787841797,
      "learning_rate": 9.468065954392372e-06,
      "loss": 5.7918,
      "step": 13164
    },
    {
      "epoch": 0.5314897052886556,
      "grad_norm": 70.70758056640625,
      "learning_rate": 9.466760229784795e-06,
      "loss": 5.001,
      "step": 13165
    },
    {
      "epoch": 0.5315300767056924,
      "grad_norm": 82.28717041015625,
      "learning_rate": 9.465454514294374e-06,
      "loss": 5.6935,
      "step": 13166
    },
    {
      "epoch": 0.5315704481227291,
      "grad_norm": 70.12214660644531,
      "learning_rate": 9.464148807943432e-06,
      "loss": 4.3864,
      "step": 13167
    },
    {
      "epoch": 0.5316108195397659,
      "grad_norm": 81.5667724609375,
      "learning_rate": 9.462843110754296e-06,
      "loss": 5.2387,
      "step": 13168
    },
    {
      "epoch": 0.5316511909568026,
      "grad_norm": 85.00050354003906,
      "learning_rate": 9.461537422749286e-06,
      "loss": 3.4243,
      "step": 13169
    },
    {
      "epoch": 0.5316915623738393,
      "grad_norm": 74.93621826171875,
      "learning_rate": 9.460231743950728e-06,
      "loss": 5.6339,
      "step": 13170
    },
    {
      "epoch": 0.531731933790876,
      "grad_norm": 101.9364013671875,
      "learning_rate": 9.458926074380948e-06,
      "loss": 4.7326,
      "step": 13171
    },
    {
      "epoch": 0.5317723052079127,
      "grad_norm": 106.99938201904297,
      "learning_rate": 9.457620414062272e-06,
      "loss": 5.7301,
      "step": 13172
    },
    {
      "epoch": 0.5318126766249496,
      "grad_norm": 86.02059173583984,
      "learning_rate": 9.456314763017016e-06,
      "loss": 5.5423,
      "step": 13173
    },
    {
      "epoch": 0.5318530480419863,
      "grad_norm": 157.15158081054688,
      "learning_rate": 9.455009121267508e-06,
      "loss": 5.8593,
      "step": 13174
    },
    {
      "epoch": 0.531893419459023,
      "grad_norm": 147.22622680664062,
      "learning_rate": 9.453703488836072e-06,
      "loss": 5.8368,
      "step": 13175
    },
    {
      "epoch": 0.5319337908760597,
      "grad_norm": 83.52975463867188,
      "learning_rate": 9.452397865745033e-06,
      "loss": 5.0475,
      "step": 13176
    },
    {
      "epoch": 0.5319741622930965,
      "grad_norm": 89.84613800048828,
      "learning_rate": 9.451092252016708e-06,
      "loss": 5.0155,
      "step": 13177
    },
    {
      "epoch": 0.5320145337101332,
      "grad_norm": 110.74368286132812,
      "learning_rate": 9.449786647673425e-06,
      "loss": 5.0379,
      "step": 13178
    },
    {
      "epoch": 0.53205490512717,
      "grad_norm": 135.83941650390625,
      "learning_rate": 9.448481052737505e-06,
      "loss": 6.3747,
      "step": 13179
    },
    {
      "epoch": 0.5320952765442067,
      "grad_norm": 91.29956817626953,
      "learning_rate": 9.447175467231275e-06,
      "loss": 4.5253,
      "step": 13180
    },
    {
      "epoch": 0.5321356479612435,
      "grad_norm": 145.59947204589844,
      "learning_rate": 9.44586989117705e-06,
      "loss": 5.0821,
      "step": 13181
    },
    {
      "epoch": 0.5321760193782802,
      "grad_norm": 90.49530029296875,
      "learning_rate": 9.444564324597155e-06,
      "loss": 5.6132,
      "step": 13182
    },
    {
      "epoch": 0.5322163907953169,
      "grad_norm": 104.56219482421875,
      "learning_rate": 9.443258767513914e-06,
      "loss": 7.0866,
      "step": 13183
    },
    {
      "epoch": 0.5322567622123536,
      "grad_norm": 134.67042541503906,
      "learning_rate": 9.44195321994965e-06,
      "loss": 5.0552,
      "step": 13184
    },
    {
      "epoch": 0.5322971336293904,
      "grad_norm": 93.03292083740234,
      "learning_rate": 9.44064768192668e-06,
      "loss": 3.9199,
      "step": 13185
    },
    {
      "epoch": 0.5323375050464272,
      "grad_norm": 86.80445098876953,
      "learning_rate": 9.43934215346733e-06,
      "loss": 3.1827,
      "step": 13186
    },
    {
      "epoch": 0.5323778764634639,
      "grad_norm": 88.03795623779297,
      "learning_rate": 9.438036634593918e-06,
      "loss": 6.4412,
      "step": 13187
    },
    {
      "epoch": 0.5324182478805006,
      "grad_norm": 145.9460906982422,
      "learning_rate": 9.436731125328772e-06,
      "loss": 5.1468,
      "step": 13188
    },
    {
      "epoch": 0.5324586192975373,
      "grad_norm": 80.02301788330078,
      "learning_rate": 9.435425625694205e-06,
      "loss": 4.9874,
      "step": 13189
    },
    {
      "epoch": 0.5324989907145741,
      "grad_norm": 80.8631362915039,
      "learning_rate": 9.434120135712542e-06,
      "loss": 4.1101,
      "step": 13190
    },
    {
      "epoch": 0.5325393621316108,
      "grad_norm": 111.17793273925781,
      "learning_rate": 9.432814655406102e-06,
      "loss": 6.6595,
      "step": 13191
    },
    {
      "epoch": 0.5325797335486475,
      "grad_norm": 83.39573669433594,
      "learning_rate": 9.431509184797211e-06,
      "loss": 5.0486,
      "step": 13192
    },
    {
      "epoch": 0.5326201049656843,
      "grad_norm": 89.10283660888672,
      "learning_rate": 9.430203723908182e-06,
      "loss": 6.4262,
      "step": 13193
    },
    {
      "epoch": 0.5326604763827211,
      "grad_norm": 109.56734466552734,
      "learning_rate": 9.428898272761338e-06,
      "loss": 5.7642,
      "step": 13194
    },
    {
      "epoch": 0.5327008477997578,
      "grad_norm": 82.9981689453125,
      "learning_rate": 9.427592831379002e-06,
      "loss": 5.1183,
      "step": 13195
    },
    {
      "epoch": 0.5327412192167945,
      "grad_norm": 115.66175842285156,
      "learning_rate": 9.42628739978349e-06,
      "loss": 4.4129,
      "step": 13196
    },
    {
      "epoch": 0.5327815906338312,
      "grad_norm": 78.29141998291016,
      "learning_rate": 9.424981977997126e-06,
      "loss": 4.6417,
      "step": 13197
    },
    {
      "epoch": 0.532821962050868,
      "grad_norm": 102.00598907470703,
      "learning_rate": 9.423676566042225e-06,
      "loss": 6.9493,
      "step": 13198
    },
    {
      "epoch": 0.5328623334679047,
      "grad_norm": 83.22528839111328,
      "learning_rate": 9.422371163941108e-06,
      "loss": 4.5895,
      "step": 13199
    },
    {
      "epoch": 0.5329027048849415,
      "grad_norm": 84.22599792480469,
      "learning_rate": 9.421065771716102e-06,
      "loss": 3.8789,
      "step": 13200
    },
    {
      "epoch": 0.5329430763019782,
      "grad_norm": 78.117919921875,
      "learning_rate": 9.419760389389514e-06,
      "loss": 4.613,
      "step": 13201
    },
    {
      "epoch": 0.532983447719015,
      "grad_norm": 119.49153137207031,
      "learning_rate": 9.418455016983668e-06,
      "loss": 4.6616,
      "step": 13202
    },
    {
      "epoch": 0.5330238191360517,
      "grad_norm": 76.63362884521484,
      "learning_rate": 9.417149654520884e-06,
      "loss": 4.7218,
      "step": 13203
    },
    {
      "epoch": 0.5330641905530884,
      "grad_norm": 106.84495544433594,
      "learning_rate": 9.41584430202348e-06,
      "loss": 4.5779,
      "step": 13204
    },
    {
      "epoch": 0.5331045619701251,
      "grad_norm": 89.95745086669922,
      "learning_rate": 9.414538959513774e-06,
      "loss": 6.0872,
      "step": 13205
    },
    {
      "epoch": 0.5331449333871618,
      "grad_norm": 74.26494598388672,
      "learning_rate": 9.413233627014085e-06,
      "loss": 3.6086,
      "step": 13206
    },
    {
      "epoch": 0.5331853048041987,
      "grad_norm": 92.60574340820312,
      "learning_rate": 9.41192830454673e-06,
      "loss": 3.8101,
      "step": 13207
    },
    {
      "epoch": 0.5332256762212354,
      "grad_norm": 79.6065444946289,
      "learning_rate": 9.410622992134033e-06,
      "loss": 4.5671,
      "step": 13208
    },
    {
      "epoch": 0.5332660476382721,
      "grad_norm": 87.70884704589844,
      "learning_rate": 9.4093176897983e-06,
      "loss": 4.8009,
      "step": 13209
    },
    {
      "epoch": 0.5333064190553088,
      "grad_norm": 104.24310302734375,
      "learning_rate": 9.408012397561856e-06,
      "loss": 4.6859,
      "step": 13210
    },
    {
      "epoch": 0.5333467904723456,
      "grad_norm": 90.17636108398438,
      "learning_rate": 9.40670711544702e-06,
      "loss": 5.9524,
      "step": 13211
    },
    {
      "epoch": 0.5333871618893823,
      "grad_norm": 69.25712585449219,
      "learning_rate": 9.405401843476108e-06,
      "loss": 5.2588,
      "step": 13212
    },
    {
      "epoch": 0.533427533306419,
      "grad_norm": 85.29681396484375,
      "learning_rate": 9.404096581671434e-06,
      "loss": 3.8561,
      "step": 13213
    },
    {
      "epoch": 0.5334679047234558,
      "grad_norm": 168.02601623535156,
      "learning_rate": 9.402791330055315e-06,
      "loss": 4.0379,
      "step": 13214
    },
    {
      "epoch": 0.5335082761404926,
      "grad_norm": 88.55169677734375,
      "learning_rate": 9.401486088650072e-06,
      "loss": 6.8171,
      "step": 13215
    },
    {
      "epoch": 0.5335486475575293,
      "grad_norm": 104.76931762695312,
      "learning_rate": 9.40018085747802e-06,
      "loss": 3.9166,
      "step": 13216
    },
    {
      "epoch": 0.533589018974566,
      "grad_norm": 112.95965576171875,
      "learning_rate": 9.398875636561473e-06,
      "loss": 4.7059,
      "step": 13217
    },
    {
      "epoch": 0.5336293903916027,
      "grad_norm": 75.80835723876953,
      "learning_rate": 9.397570425922748e-06,
      "loss": 4.7062,
      "step": 13218
    },
    {
      "epoch": 0.5336697618086395,
      "grad_norm": 89.05754089355469,
      "learning_rate": 9.396265225584162e-06,
      "loss": 6.0298,
      "step": 13219
    },
    {
      "epoch": 0.5337101332256762,
      "grad_norm": 244.5932159423828,
      "learning_rate": 9.394960035568035e-06,
      "loss": 6.1634,
      "step": 13220
    },
    {
      "epoch": 0.533750504642713,
      "grad_norm": 72.6595458984375,
      "learning_rate": 9.393654855896675e-06,
      "loss": 5.5471,
      "step": 13221
    },
    {
      "epoch": 0.5337908760597497,
      "grad_norm": 110.28121948242188,
      "learning_rate": 9.3923496865924e-06,
      "loss": 5.0927,
      "step": 13222
    },
    {
      "epoch": 0.5338312474767865,
      "grad_norm": 76.60152435302734,
      "learning_rate": 9.391044527677527e-06,
      "loss": 5.6353,
      "step": 13223
    },
    {
      "epoch": 0.5338716188938232,
      "grad_norm": 120.79246520996094,
      "learning_rate": 9.389739379174373e-06,
      "loss": 4.2977,
      "step": 13224
    },
    {
      "epoch": 0.5339119903108599,
      "grad_norm": 128.19384765625,
      "learning_rate": 9.38843424110525e-06,
      "loss": 4.3622,
      "step": 13225
    },
    {
      "epoch": 0.5339523617278966,
      "grad_norm": 108.00481414794922,
      "learning_rate": 9.38712911349247e-06,
      "loss": 7.8263,
      "step": 13226
    },
    {
      "epoch": 0.5339927331449333,
      "grad_norm": 103.60064697265625,
      "learning_rate": 9.385823996358353e-06,
      "loss": 3.3602,
      "step": 13227
    },
    {
      "epoch": 0.5340331045619702,
      "grad_norm": 94.62256622314453,
      "learning_rate": 9.384518889725216e-06,
      "loss": 6.5689,
      "step": 13228
    },
    {
      "epoch": 0.5340734759790069,
      "grad_norm": 110.3812255859375,
      "learning_rate": 9.383213793615364e-06,
      "loss": 5.0677,
      "step": 13229
    },
    {
      "epoch": 0.5341138473960436,
      "grad_norm": 71.09800720214844,
      "learning_rate": 9.381908708051115e-06,
      "loss": 5.0421,
      "step": 13230
    },
    {
      "epoch": 0.5341542188130803,
      "grad_norm": 61.117340087890625,
      "learning_rate": 9.380603633054784e-06,
      "loss": 3.3219,
      "step": 13231
    },
    {
      "epoch": 0.5341945902301171,
      "grad_norm": 118.45166778564453,
      "learning_rate": 9.379298568648688e-06,
      "loss": 5.6295,
      "step": 13232
    },
    {
      "epoch": 0.5342349616471538,
      "grad_norm": 78.5632095336914,
      "learning_rate": 9.377993514855133e-06,
      "loss": 5.6365,
      "step": 13233
    },
    {
      "epoch": 0.5342753330641905,
      "grad_norm": 94.75425720214844,
      "learning_rate": 9.376688471696438e-06,
      "loss": 4.6595,
      "step": 13234
    },
    {
      "epoch": 0.5343157044812273,
      "grad_norm": 78.4682846069336,
      "learning_rate": 9.375383439194913e-06,
      "loss": 5.1909,
      "step": 13235
    },
    {
      "epoch": 0.5343560758982641,
      "grad_norm": 67.70013427734375,
      "learning_rate": 9.374078417372878e-06,
      "loss": 5.4134,
      "step": 13236
    },
    {
      "epoch": 0.5343964473153008,
      "grad_norm": 100.67323303222656,
      "learning_rate": 9.372773406252637e-06,
      "loss": 4.7229,
      "step": 13237
    },
    {
      "epoch": 0.5344368187323375,
      "grad_norm": 77.658447265625,
      "learning_rate": 9.371468405856505e-06,
      "loss": 5.1447,
      "step": 13238
    },
    {
      "epoch": 0.5344771901493742,
      "grad_norm": 96.6758041381836,
      "learning_rate": 9.370163416206794e-06,
      "loss": 5.8323,
      "step": 13239
    },
    {
      "epoch": 0.534517561566411,
      "grad_norm": 96.5758056640625,
      "learning_rate": 9.36885843732582e-06,
      "loss": 5.8922,
      "step": 13240
    },
    {
      "epoch": 0.5345579329834477,
      "grad_norm": 75.83228302001953,
      "learning_rate": 9.367553469235897e-06,
      "loss": 5.0706,
      "step": 13241
    },
    {
      "epoch": 0.5345983044004845,
      "grad_norm": 87.72261810302734,
      "learning_rate": 9.366248511959329e-06,
      "loss": 4.2153,
      "step": 13242
    },
    {
      "epoch": 0.5346386758175212,
      "grad_norm": 137.15086364746094,
      "learning_rate": 9.36494356551843e-06,
      "loss": 5.4482,
      "step": 13243
    },
    {
      "epoch": 0.534679047234558,
      "grad_norm": 99.54090881347656,
      "learning_rate": 9.363638629935513e-06,
      "loss": 5.5738,
      "step": 13244
    },
    {
      "epoch": 0.5347194186515947,
      "grad_norm": 116.42229461669922,
      "learning_rate": 9.362333705232892e-06,
      "loss": 3.8406,
      "step": 13245
    },
    {
      "epoch": 0.5347597900686314,
      "grad_norm": 92.72208404541016,
      "learning_rate": 9.361028791432875e-06,
      "loss": 5.0741,
      "step": 13246
    },
    {
      "epoch": 0.5348001614856681,
      "grad_norm": 98.73151397705078,
      "learning_rate": 9.359723888557771e-06,
      "loss": 3.9415,
      "step": 13247
    },
    {
      "epoch": 0.5348405329027048,
      "grad_norm": 96.18641662597656,
      "learning_rate": 9.358418996629895e-06,
      "loss": 4.3283,
      "step": 13248
    },
    {
      "epoch": 0.5348809043197417,
      "grad_norm": 109.47308349609375,
      "learning_rate": 9.35711411567156e-06,
      "loss": 2.7371,
      "step": 13249
    },
    {
      "epoch": 0.5349212757367784,
      "grad_norm": 103.0194091796875,
      "learning_rate": 9.355809245705068e-06,
      "loss": 5.7362,
      "step": 13250
    },
    {
      "epoch": 0.5349616471538151,
      "grad_norm": 111.73187255859375,
      "learning_rate": 9.354504386752732e-06,
      "loss": 6.0974,
      "step": 13251
    },
    {
      "epoch": 0.5350020185708518,
      "grad_norm": 108.60404205322266,
      "learning_rate": 9.353199538836866e-06,
      "loss": 6.5918,
      "step": 13252
    },
    {
      "epoch": 0.5350423899878886,
      "grad_norm": 77.73123931884766,
      "learning_rate": 9.35189470197978e-06,
      "loss": 5.7031,
      "step": 13253
    },
    {
      "epoch": 0.5350827614049253,
      "grad_norm": 126.16240692138672,
      "learning_rate": 9.350589876203778e-06,
      "loss": 4.6037,
      "step": 13254
    },
    {
      "epoch": 0.535123132821962,
      "grad_norm": 207.9646759033203,
      "learning_rate": 9.349285061531171e-06,
      "loss": 5.9982,
      "step": 13255
    },
    {
      "epoch": 0.5351635042389988,
      "grad_norm": 108.7223129272461,
      "learning_rate": 9.347980257984273e-06,
      "loss": 3.1445,
      "step": 13256
    },
    {
      "epoch": 0.5352038756560356,
      "grad_norm": 86.36082458496094,
      "learning_rate": 9.346675465585391e-06,
      "loss": 5.2915,
      "step": 13257
    },
    {
      "epoch": 0.5352442470730723,
      "grad_norm": 87.90287780761719,
      "learning_rate": 9.34537068435683e-06,
      "loss": 4.174,
      "step": 13258
    },
    {
      "epoch": 0.535284618490109,
      "grad_norm": 99.28327178955078,
      "learning_rate": 9.344065914320904e-06,
      "loss": 5.0474,
      "step": 13259
    },
    {
      "epoch": 0.5353249899071457,
      "grad_norm": 108.0738754272461,
      "learning_rate": 9.342761155499916e-06,
      "loss": 7.3846,
      "step": 13260
    },
    {
      "epoch": 0.5353653613241824,
      "grad_norm": 88.12480926513672,
      "learning_rate": 9.341456407916183e-06,
      "loss": 5.3557,
      "step": 13261
    },
    {
      "epoch": 0.5354057327412192,
      "grad_norm": 72.1934814453125,
      "learning_rate": 9.340151671592004e-06,
      "loss": 5.0353,
      "step": 13262
    },
    {
      "epoch": 0.535446104158256,
      "grad_norm": 75.25057220458984,
      "learning_rate": 9.33884694654969e-06,
      "loss": 4.801,
      "step": 13263
    },
    {
      "epoch": 0.5354864755752927,
      "grad_norm": 135.71368408203125,
      "learning_rate": 9.337542232811549e-06,
      "loss": 5.3403,
      "step": 13264
    },
    {
      "epoch": 0.5355268469923294,
      "grad_norm": 135.9523162841797,
      "learning_rate": 9.33623753039989e-06,
      "loss": 3.6429,
      "step": 13265
    },
    {
      "epoch": 0.5355672184093662,
      "grad_norm": 82.56964874267578,
      "learning_rate": 9.334932839337018e-06,
      "loss": 5.0463,
      "step": 13266
    },
    {
      "epoch": 0.5356075898264029,
      "grad_norm": 98.6256103515625,
      "learning_rate": 9.333628159645241e-06,
      "loss": 3.7653,
      "step": 13267
    },
    {
      "epoch": 0.5356479612434396,
      "grad_norm": 89.85433959960938,
      "learning_rate": 9.332323491346867e-06,
      "loss": 3.7114,
      "step": 13268
    },
    {
      "epoch": 0.5356883326604763,
      "grad_norm": 93.43681335449219,
      "learning_rate": 9.331018834464207e-06,
      "loss": 4.052,
      "step": 13269
    },
    {
      "epoch": 0.5357287040775132,
      "grad_norm": 96.19735717773438,
      "learning_rate": 9.32971418901956e-06,
      "loss": 4.1153,
      "step": 13270
    },
    {
      "epoch": 0.5357690754945499,
      "grad_norm": 123.52063751220703,
      "learning_rate": 9.328409555035232e-06,
      "loss": 5.2105,
      "step": 13271
    },
    {
      "epoch": 0.5358094469115866,
      "grad_norm": 75.55126953125,
      "learning_rate": 9.327104932533532e-06,
      "loss": 5.7306,
      "step": 13272
    },
    {
      "epoch": 0.5358498183286233,
      "grad_norm": 103.39547729492188,
      "learning_rate": 9.325800321536771e-06,
      "loss": 5.0958,
      "step": 13273
    },
    {
      "epoch": 0.53589018974566,
      "grad_norm": 100.80152130126953,
      "learning_rate": 9.324495722067246e-06,
      "loss": 3.4766,
      "step": 13274
    },
    {
      "epoch": 0.5359305611626968,
      "grad_norm": 71.49024963378906,
      "learning_rate": 9.323191134147269e-06,
      "loss": 5.1171,
      "step": 13275
    },
    {
      "epoch": 0.5359709325797335,
      "grad_norm": 117.8778305053711,
      "learning_rate": 9.32188655779914e-06,
      "loss": 5.563,
      "step": 13276
    },
    {
      "epoch": 0.5360113039967703,
      "grad_norm": 163.98553466796875,
      "learning_rate": 9.320581993045174e-06,
      "loss": 6.3591,
      "step": 13277
    },
    {
      "epoch": 0.536051675413807,
      "grad_norm": 75.35773468017578,
      "learning_rate": 9.319277439907663e-06,
      "loss": 4.4838,
      "step": 13278
    },
    {
      "epoch": 0.5360920468308438,
      "grad_norm": 124.28276824951172,
      "learning_rate": 9.31797289840892e-06,
      "loss": 3.6083,
      "step": 13279
    },
    {
      "epoch": 0.5361324182478805,
      "grad_norm": 76.97560119628906,
      "learning_rate": 9.316668368571248e-06,
      "loss": 3.8709,
      "step": 13280
    },
    {
      "epoch": 0.5361727896649172,
      "grad_norm": 83.94290161132812,
      "learning_rate": 9.315363850416955e-06,
      "loss": 6.7893,
      "step": 13281
    },
    {
      "epoch": 0.5362131610819539,
      "grad_norm": 86.98932647705078,
      "learning_rate": 9.314059343968337e-06,
      "loss": 5.5803,
      "step": 13282
    },
    {
      "epoch": 0.5362535324989907,
      "grad_norm": 152.92539978027344,
      "learning_rate": 9.312754849247702e-06,
      "loss": 4.4106,
      "step": 13283
    },
    {
      "epoch": 0.5362939039160275,
      "grad_norm": 84.7159194946289,
      "learning_rate": 9.311450366277355e-06,
      "loss": 5.5362,
      "step": 13284
    },
    {
      "epoch": 0.5363342753330642,
      "grad_norm": 108.08811950683594,
      "learning_rate": 9.310145895079601e-06,
      "loss": 5.0912,
      "step": 13285
    },
    {
      "epoch": 0.5363746467501009,
      "grad_norm": 127.90005493164062,
      "learning_rate": 9.308841435676738e-06,
      "loss": 5.6115,
      "step": 13286
    },
    {
      "epoch": 0.5364150181671377,
      "grad_norm": 85.8809814453125,
      "learning_rate": 9.307536988091074e-06,
      "loss": 4.3832,
      "step": 13287
    },
    {
      "epoch": 0.5364553895841744,
      "grad_norm": 146.19960021972656,
      "learning_rate": 9.30623255234491e-06,
      "loss": 6.5051,
      "step": 13288
    },
    {
      "epoch": 0.5364957610012111,
      "grad_norm": 114.29135131835938,
      "learning_rate": 9.304928128460554e-06,
      "loss": 6.236,
      "step": 13289
    },
    {
      "epoch": 0.5365361324182479,
      "grad_norm": 108.39393615722656,
      "learning_rate": 9.303623716460297e-06,
      "loss": 5.3207,
      "step": 13290
    },
    {
      "epoch": 0.5365765038352847,
      "grad_norm": 95.63158416748047,
      "learning_rate": 9.302319316366451e-06,
      "loss": 4.9782,
      "step": 13291
    },
    {
      "epoch": 0.5366168752523214,
      "grad_norm": 130.3588104248047,
      "learning_rate": 9.301014928201316e-06,
      "loss": 5.7214,
      "step": 13292
    },
    {
      "epoch": 0.5366572466693581,
      "grad_norm": 107.38912200927734,
      "learning_rate": 9.299710551987193e-06,
      "loss": 6.1299,
      "step": 13293
    },
    {
      "epoch": 0.5366976180863948,
      "grad_norm": 71.60869598388672,
      "learning_rate": 9.298406187746385e-06,
      "loss": 5.311,
      "step": 13294
    },
    {
      "epoch": 0.5367379895034315,
      "grad_norm": 100.2457504272461,
      "learning_rate": 9.29710183550119e-06,
      "loss": 4.5168,
      "step": 13295
    },
    {
      "epoch": 0.5367783609204683,
      "grad_norm": 215.12249755859375,
      "learning_rate": 9.295797495273913e-06,
      "loss": 5.5866,
      "step": 13296
    },
    {
      "epoch": 0.5368187323375051,
      "grad_norm": 168.52561950683594,
      "learning_rate": 9.29449316708686e-06,
      "loss": 4.3518,
      "step": 13297
    },
    {
      "epoch": 0.5368591037545418,
      "grad_norm": 85.02974700927734,
      "learning_rate": 9.29318885096232e-06,
      "loss": 5.2831,
      "step": 13298
    },
    {
      "epoch": 0.5368994751715785,
      "grad_norm": 73.64900207519531,
      "learning_rate": 9.291884546922602e-06,
      "loss": 4.0415,
      "step": 13299
    },
    {
      "epoch": 0.5369398465886153,
      "grad_norm": 120.00691223144531,
      "learning_rate": 9.290580254990005e-06,
      "loss": 5.3861,
      "step": 13300
    },
    {
      "epoch": 0.536980218005652,
      "grad_norm": 100.37364959716797,
      "learning_rate": 9.289275975186833e-06,
      "loss": 5.2335,
      "step": 13301
    },
    {
      "epoch": 0.5370205894226887,
      "grad_norm": 85.56075286865234,
      "learning_rate": 9.287971707535376e-06,
      "loss": 3.686,
      "step": 13302
    },
    {
      "epoch": 0.5370609608397254,
      "grad_norm": 74.86972045898438,
      "learning_rate": 9.286667452057942e-06,
      "loss": 4.2762,
      "step": 13303
    },
    {
      "epoch": 0.5371013322567623,
      "grad_norm": 82.3934097290039,
      "learning_rate": 9.285363208776828e-06,
      "loss": 6.5672,
      "step": 13304
    },
    {
      "epoch": 0.537141703673799,
      "grad_norm": 110.11758422851562,
      "learning_rate": 9.284058977714337e-06,
      "loss": 6.3557,
      "step": 13305
    },
    {
      "epoch": 0.5371820750908357,
      "grad_norm": 119.84136199951172,
      "learning_rate": 9.282754758892763e-06,
      "loss": 5.2685,
      "step": 13306
    },
    {
      "epoch": 0.5372224465078724,
      "grad_norm": 82.63426208496094,
      "learning_rate": 9.281450552334407e-06,
      "loss": 5.1663,
      "step": 13307
    },
    {
      "epoch": 0.5372628179249092,
      "grad_norm": 101.46006774902344,
      "learning_rate": 9.280146358061571e-06,
      "loss": 5.5868,
      "step": 13308
    },
    {
      "epoch": 0.5373031893419459,
      "grad_norm": 141.4647216796875,
      "learning_rate": 9.278842176096554e-06,
      "loss": 4.2373,
      "step": 13309
    },
    {
      "epoch": 0.5373435607589826,
      "grad_norm": 72.71875762939453,
      "learning_rate": 9.277538006461648e-06,
      "loss": 4.6822,
      "step": 13310
    },
    {
      "epoch": 0.5373839321760194,
      "grad_norm": 102.59266662597656,
      "learning_rate": 9.276233849179154e-06,
      "loss": 5.3064,
      "step": 13311
    },
    {
      "epoch": 0.5374243035930562,
      "grad_norm": 112.24707794189453,
      "learning_rate": 9.27492970427137e-06,
      "loss": 5.1381,
      "step": 13312
    },
    {
      "epoch": 0.5374646750100929,
      "grad_norm": 186.63250732421875,
      "learning_rate": 9.273625571760601e-06,
      "loss": 5.1712,
      "step": 13313
    },
    {
      "epoch": 0.5375050464271296,
      "grad_norm": 136.45150756835938,
      "learning_rate": 9.272321451669133e-06,
      "loss": 4.0384,
      "step": 13314
    },
    {
      "epoch": 0.5375454178441663,
      "grad_norm": 77.52899932861328,
      "learning_rate": 9.27101734401927e-06,
      "loss": 4.4976,
      "step": 13315
    },
    {
      "epoch": 0.537585789261203,
      "grad_norm": 86.9115982055664,
      "learning_rate": 9.269713248833309e-06,
      "loss": 4.5096,
      "step": 13316
    },
    {
      "epoch": 0.5376261606782398,
      "grad_norm": 189.00155639648438,
      "learning_rate": 9.268409166133548e-06,
      "loss": 5.0722,
      "step": 13317
    },
    {
      "epoch": 0.5376665320952766,
      "grad_norm": 123.84224700927734,
      "learning_rate": 9.267105095942279e-06,
      "loss": 4.4439,
      "step": 13318
    },
    {
      "epoch": 0.5377069035123133,
      "grad_norm": 139.21925354003906,
      "learning_rate": 9.265801038281803e-06,
      "loss": 5.9734,
      "step": 13319
    },
    {
      "epoch": 0.53774727492935,
      "grad_norm": 80.75093841552734,
      "learning_rate": 9.264496993174413e-06,
      "loss": 4.6667,
      "step": 13320
    },
    {
      "epoch": 0.5377876463463868,
      "grad_norm": 116.66474914550781,
      "learning_rate": 9.263192960642411e-06,
      "loss": 5.0155,
      "step": 13321
    },
    {
      "epoch": 0.5378280177634235,
      "grad_norm": 136.52586364746094,
      "learning_rate": 9.261888940708085e-06,
      "loss": 6.1297,
      "step": 13322
    },
    {
      "epoch": 0.5378683891804602,
      "grad_norm": 126.72503662109375,
      "learning_rate": 9.260584933393735e-06,
      "loss": 6.8221,
      "step": 13323
    },
    {
      "epoch": 0.5379087605974969,
      "grad_norm": 108.48876190185547,
      "learning_rate": 9.259280938721655e-06,
      "loss": 3.8707,
      "step": 13324
    },
    {
      "epoch": 0.5379491320145338,
      "grad_norm": 103.61138153076172,
      "learning_rate": 9.257976956714144e-06,
      "loss": 4.5286,
      "step": 13325
    },
    {
      "epoch": 0.5379895034315705,
      "grad_norm": 134.9611358642578,
      "learning_rate": 9.256672987393492e-06,
      "loss": 3.352,
      "step": 13326
    },
    {
      "epoch": 0.5380298748486072,
      "grad_norm": 107.12370300292969,
      "learning_rate": 9.255369030781997e-06,
      "loss": 5.9449,
      "step": 13327
    },
    {
      "epoch": 0.5380702462656439,
      "grad_norm": 314.96337890625,
      "learning_rate": 9.254065086901951e-06,
      "loss": 5.1476,
      "step": 13328
    },
    {
      "epoch": 0.5381106176826806,
      "grad_norm": 205.43922424316406,
      "learning_rate": 9.252761155775656e-06,
      "loss": 4.2782,
      "step": 13329
    },
    {
      "epoch": 0.5381509890997174,
      "grad_norm": 110.35899353027344,
      "learning_rate": 9.251457237425395e-06,
      "loss": 5.1145,
      "step": 13330
    },
    {
      "epoch": 0.5381913605167541,
      "grad_norm": 79.09522247314453,
      "learning_rate": 9.250153331873466e-06,
      "loss": 6.0656,
      "step": 13331
    },
    {
      "epoch": 0.5382317319337909,
      "grad_norm": 136.7174530029297,
      "learning_rate": 9.248849439142167e-06,
      "loss": 4.9297,
      "step": 13332
    },
    {
      "epoch": 0.5382721033508276,
      "grad_norm": 66.6397933959961,
      "learning_rate": 9.24754555925379e-06,
      "loss": 6.7196,
      "step": 13333
    },
    {
      "epoch": 0.5383124747678644,
      "grad_norm": 100.56253814697266,
      "learning_rate": 9.246241692230624e-06,
      "loss": 5.8871,
      "step": 13334
    },
    {
      "epoch": 0.5383528461849011,
      "grad_norm": 108.40010070800781,
      "learning_rate": 9.244937838094964e-06,
      "loss": 4.5554,
      "step": 13335
    },
    {
      "epoch": 0.5383932176019378,
      "grad_norm": 147.58999633789062,
      "learning_rate": 9.243633996869104e-06,
      "loss": 4.7082,
      "step": 13336
    },
    {
      "epoch": 0.5384335890189745,
      "grad_norm": 86.46437072753906,
      "learning_rate": 9.242330168575342e-06,
      "loss": 4.3684,
      "step": 13337
    },
    {
      "epoch": 0.5384739604360113,
      "grad_norm": 95.79300689697266,
      "learning_rate": 9.24102635323596e-06,
      "loss": 4.5267,
      "step": 13338
    },
    {
      "epoch": 0.5385143318530481,
      "grad_norm": 94.75657653808594,
      "learning_rate": 9.239722550873255e-06,
      "loss": 4.8552,
      "step": 13339
    },
    {
      "epoch": 0.5385547032700848,
      "grad_norm": 119.46043395996094,
      "learning_rate": 9.238418761509518e-06,
      "loss": 5.8239,
      "step": 13340
    },
    {
      "epoch": 0.5385950746871215,
      "grad_norm": 99.80037689208984,
      "learning_rate": 9.237114985167046e-06,
      "loss": 4.4712,
      "step": 13341
    },
    {
      "epoch": 0.5386354461041583,
      "grad_norm": 95.60755920410156,
      "learning_rate": 9.235811221868123e-06,
      "loss": 6.0474,
      "step": 13342
    },
    {
      "epoch": 0.538675817521195,
      "grad_norm": 108.77449798583984,
      "learning_rate": 9.234507471635043e-06,
      "loss": 5.5566,
      "step": 13343
    },
    {
      "epoch": 0.5387161889382317,
      "grad_norm": 89.65797424316406,
      "learning_rate": 9.233203734490099e-06,
      "loss": 5.5281,
      "step": 13344
    },
    {
      "epoch": 0.5387565603552684,
      "grad_norm": 78.65510559082031,
      "learning_rate": 9.231900010455584e-06,
      "loss": 5.9368,
      "step": 13345
    },
    {
      "epoch": 0.5387969317723053,
      "grad_norm": 82.88690185546875,
      "learning_rate": 9.23059629955378e-06,
      "loss": 4.3455,
      "step": 13346
    },
    {
      "epoch": 0.538837303189342,
      "grad_norm": 62.841041564941406,
      "learning_rate": 9.229292601806983e-06,
      "loss": 4.4654,
      "step": 13347
    },
    {
      "epoch": 0.5388776746063787,
      "grad_norm": 102.19341278076172,
      "learning_rate": 9.227988917237485e-06,
      "loss": 4.0655,
      "step": 13348
    },
    {
      "epoch": 0.5389180460234154,
      "grad_norm": 129.93954467773438,
      "learning_rate": 9.226685245867576e-06,
      "loss": 5.3836,
      "step": 13349
    },
    {
      "epoch": 0.5389584174404521,
      "grad_norm": 79.07237243652344,
      "learning_rate": 9.22538158771954e-06,
      "loss": 5.2076,
      "step": 13350
    },
    {
      "epoch": 0.5389987888574889,
      "grad_norm": 104.06109619140625,
      "learning_rate": 9.22407794281567e-06,
      "loss": 5.1496,
      "step": 13351
    },
    {
      "epoch": 0.5390391602745256,
      "grad_norm": 74.00099182128906,
      "learning_rate": 9.222774311178253e-06,
      "loss": 4.9051,
      "step": 13352
    },
    {
      "epoch": 0.5390795316915624,
      "grad_norm": 103.62953186035156,
      "learning_rate": 9.221470692829585e-06,
      "loss": 5.1273,
      "step": 13353
    },
    {
      "epoch": 0.5391199031085991,
      "grad_norm": 81.48710632324219,
      "learning_rate": 9.220167087791946e-06,
      "loss": 6.1644,
      "step": 13354
    },
    {
      "epoch": 0.5391602745256359,
      "grad_norm": 109.49107360839844,
      "learning_rate": 9.21886349608763e-06,
      "loss": 7.174,
      "step": 13355
    },
    {
      "epoch": 0.5392006459426726,
      "grad_norm": 85.5233154296875,
      "learning_rate": 9.217559917738925e-06,
      "loss": 4.3574,
      "step": 13356
    },
    {
      "epoch": 0.5392410173597093,
      "grad_norm": 89.69979095458984,
      "learning_rate": 9.216256352768123e-06,
      "loss": 5.0696,
      "step": 13357
    },
    {
      "epoch": 0.539281388776746,
      "grad_norm": 86.95922088623047,
      "learning_rate": 9.2149528011975e-06,
      "loss": 6.4057,
      "step": 13358
    },
    {
      "epoch": 0.5393217601937828,
      "grad_norm": 101.70918273925781,
      "learning_rate": 9.213649263049353e-06,
      "loss": 4.4436,
      "step": 13359
    },
    {
      "epoch": 0.5393621316108196,
      "grad_norm": 86.04937744140625,
      "learning_rate": 9.212345738345967e-06,
      "loss": 4.0876,
      "step": 13360
    },
    {
      "epoch": 0.5394025030278563,
      "grad_norm": 83.47078704833984,
      "learning_rate": 9.211042227109632e-06,
      "loss": 4.6266,
      "step": 13361
    },
    {
      "epoch": 0.539442874444893,
      "grad_norm": 92.21232604980469,
      "learning_rate": 9.209738729362629e-06,
      "loss": 2.0066,
      "step": 13362
    },
    {
      "epoch": 0.5394832458619297,
      "grad_norm": 79.46765899658203,
      "learning_rate": 9.20843524512725e-06,
      "loss": 3.6417,
      "step": 13363
    },
    {
      "epoch": 0.5395236172789665,
      "grad_norm": 71.36479949951172,
      "learning_rate": 9.20713177442578e-06,
      "loss": 4.2242,
      "step": 13364
    },
    {
      "epoch": 0.5395639886960032,
      "grad_norm": 205.5026092529297,
      "learning_rate": 9.205828317280508e-06,
      "loss": 4.758,
      "step": 13365
    },
    {
      "epoch": 0.5396043601130399,
      "grad_norm": 106.27043914794922,
      "learning_rate": 9.204524873713715e-06,
      "loss": 7.0445,
      "step": 13366
    },
    {
      "epoch": 0.5396447315300767,
      "grad_norm": 115.34723663330078,
      "learning_rate": 9.203221443747686e-06,
      "loss": 3.4471,
      "step": 13367
    },
    {
      "epoch": 0.5396851029471135,
      "grad_norm": 99.20551300048828,
      "learning_rate": 9.201918027404711e-06,
      "loss": 3.8275,
      "step": 13368
    },
    {
      "epoch": 0.5397254743641502,
      "grad_norm": 96.51385498046875,
      "learning_rate": 9.200614624707081e-06,
      "loss": 2.3856,
      "step": 13369
    },
    {
      "epoch": 0.5397658457811869,
      "grad_norm": 62.57548141479492,
      "learning_rate": 9.199311235677067e-06,
      "loss": 6.4182,
      "step": 13370
    },
    {
      "epoch": 0.5398062171982236,
      "grad_norm": 100.5118179321289,
      "learning_rate": 9.198007860336962e-06,
      "loss": 4.4169,
      "step": 13371
    },
    {
      "epoch": 0.5398465886152604,
      "grad_norm": 108.26954650878906,
      "learning_rate": 9.196704498709049e-06,
      "loss": 5.0964,
      "step": 13372
    },
    {
      "epoch": 0.5398869600322971,
      "grad_norm": 86.23918151855469,
      "learning_rate": 9.195401150815617e-06,
      "loss": 4.5475,
      "step": 13373
    },
    {
      "epoch": 0.5399273314493339,
      "grad_norm": 87.60366821289062,
      "learning_rate": 9.194097816678941e-06,
      "loss": 4.7422,
      "step": 13374
    },
    {
      "epoch": 0.5399677028663706,
      "grad_norm": 83.70853424072266,
      "learning_rate": 9.192794496321315e-06,
      "loss": 4.1308,
      "step": 13375
    },
    {
      "epoch": 0.5400080742834074,
      "grad_norm": 68.29448699951172,
      "learning_rate": 9.191491189765015e-06,
      "loss": 4.3056,
      "step": 13376
    },
    {
      "epoch": 0.5400484457004441,
      "grad_norm": 164.50318908691406,
      "learning_rate": 9.190187897032333e-06,
      "loss": 5.614,
      "step": 13377
    },
    {
      "epoch": 0.5400888171174808,
      "grad_norm": 120.85812377929688,
      "learning_rate": 9.188884618145543e-06,
      "loss": 5.8593,
      "step": 13378
    },
    {
      "epoch": 0.5401291885345175,
      "grad_norm": 84.8401107788086,
      "learning_rate": 9.18758135312693e-06,
      "loss": 5.0857,
      "step": 13379
    },
    {
      "epoch": 0.5401695599515542,
      "grad_norm": 83.6993637084961,
      "learning_rate": 9.186278101998781e-06,
      "loss": 5.5146,
      "step": 13380
    },
    {
      "epoch": 0.5402099313685911,
      "grad_norm": 128.14630126953125,
      "learning_rate": 9.184974864783378e-06,
      "loss": 4.24,
      "step": 13381
    },
    {
      "epoch": 0.5402503027856278,
      "grad_norm": 204.92425537109375,
      "learning_rate": 9.183671641502998e-06,
      "loss": 6.4248,
      "step": 13382
    },
    {
      "epoch": 0.5402906742026645,
      "grad_norm": 117.16400909423828,
      "learning_rate": 9.18236843217993e-06,
      "loss": 5.0351,
      "step": 13383
    },
    {
      "epoch": 0.5403310456197012,
      "grad_norm": 95.8516845703125,
      "learning_rate": 9.181065236836448e-06,
      "loss": 6.084,
      "step": 13384
    },
    {
      "epoch": 0.540371417036738,
      "grad_norm": 134.2105712890625,
      "learning_rate": 9.179762055494845e-06,
      "loss": 6.3587,
      "step": 13385
    },
    {
      "epoch": 0.5404117884537747,
      "grad_norm": 81.40644073486328,
      "learning_rate": 9.17845888817739e-06,
      "loss": 4.1767,
      "step": 13386
    },
    {
      "epoch": 0.5404521598708114,
      "grad_norm": 72.15196228027344,
      "learning_rate": 9.17715573490637e-06,
      "loss": 3.6094,
      "step": 13387
    },
    {
      "epoch": 0.5404925312878482,
      "grad_norm": 96.23995208740234,
      "learning_rate": 9.175852595704067e-06,
      "loss": 4.8482,
      "step": 13388
    },
    {
      "epoch": 0.540532902704885,
      "grad_norm": 110.29560089111328,
      "learning_rate": 9.174549470592763e-06,
      "loss": 6.4442,
      "step": 13389
    },
    {
      "epoch": 0.5405732741219217,
      "grad_norm": 87.85095977783203,
      "learning_rate": 9.173246359594732e-06,
      "loss": 4.6833,
      "step": 13390
    },
    {
      "epoch": 0.5406136455389584,
      "grad_norm": 95.6358413696289,
      "learning_rate": 9.171943262732256e-06,
      "loss": 4.7847,
      "step": 13391
    },
    {
      "epoch": 0.5406540169559951,
      "grad_norm": 106.68750762939453,
      "learning_rate": 9.170640180027618e-06,
      "loss": 5.9169,
      "step": 13392
    },
    {
      "epoch": 0.5406943883730319,
      "grad_norm": 110.97794342041016,
      "learning_rate": 9.169337111503099e-06,
      "loss": 5.9119,
      "step": 13393
    },
    {
      "epoch": 0.5407347597900687,
      "grad_norm": 132.37098693847656,
      "learning_rate": 9.16803405718097e-06,
      "loss": 5.4533,
      "step": 13394
    },
    {
      "epoch": 0.5407751312071054,
      "grad_norm": 85.32599639892578,
      "learning_rate": 9.16673101708352e-06,
      "loss": 6.0903,
      "step": 13395
    },
    {
      "epoch": 0.5408155026241421,
      "grad_norm": 97.66200256347656,
      "learning_rate": 9.165427991233021e-06,
      "loss": 5.1954,
      "step": 13396
    },
    {
      "epoch": 0.5408558740411789,
      "grad_norm": 195.43907165527344,
      "learning_rate": 9.16412497965176e-06,
      "loss": 4.9728,
      "step": 13397
    },
    {
      "epoch": 0.5408962454582156,
      "grad_norm": 117.55165100097656,
      "learning_rate": 9.162821982362006e-06,
      "loss": 4.2396,
      "step": 13398
    },
    {
      "epoch": 0.5409366168752523,
      "grad_norm": 163.99026489257812,
      "learning_rate": 9.16151899938604e-06,
      "loss": 4.8688,
      "step": 13399
    },
    {
      "epoch": 0.540976988292289,
      "grad_norm": 94.42163848876953,
      "learning_rate": 9.160216030746144e-06,
      "loss": 6.5379,
      "step": 13400
    },
    {
      "epoch": 0.5410173597093259,
      "grad_norm": 140.16647338867188,
      "learning_rate": 9.15891307646459e-06,
      "loss": 3.5205,
      "step": 13401
    },
    {
      "epoch": 0.5410577311263626,
      "grad_norm": 106.49019622802734,
      "learning_rate": 9.15761013656366e-06,
      "loss": 4.2674,
      "step": 13402
    },
    {
      "epoch": 0.5410981025433993,
      "grad_norm": 144.99171447753906,
      "learning_rate": 9.156307211065628e-06,
      "loss": 3.9549,
      "step": 13403
    },
    {
      "epoch": 0.541138473960436,
      "grad_norm": 136.80410766601562,
      "learning_rate": 9.155004299992773e-06,
      "loss": 4.8637,
      "step": 13404
    },
    {
      "epoch": 0.5411788453774727,
      "grad_norm": 111.08712768554688,
      "learning_rate": 9.153701403367377e-06,
      "loss": 5.3508,
      "step": 13405
    },
    {
      "epoch": 0.5412192167945095,
      "grad_norm": 89.27826690673828,
      "learning_rate": 9.152398521211704e-06,
      "loss": 5.2046,
      "step": 13406
    },
    {
      "epoch": 0.5412595882115462,
      "grad_norm": 120.90953826904297,
      "learning_rate": 9.151095653548038e-06,
      "loss": 5.1088,
      "step": 13407
    },
    {
      "epoch": 0.541299959628583,
      "grad_norm": 265.37030029296875,
      "learning_rate": 9.149792800398654e-06,
      "loss": 4.1138,
      "step": 13408
    },
    {
      "epoch": 0.5413403310456197,
      "grad_norm": 81.34393310546875,
      "learning_rate": 9.148489961785832e-06,
      "loss": 4.6539,
      "step": 13409
    },
    {
      "epoch": 0.5413807024626565,
      "grad_norm": 88.56562805175781,
      "learning_rate": 9.147187137731839e-06,
      "loss": 5.4381,
      "step": 13410
    },
    {
      "epoch": 0.5414210738796932,
      "grad_norm": 77.46790313720703,
      "learning_rate": 9.145884328258955e-06,
      "loss": 5.6021,
      "step": 13411
    },
    {
      "epoch": 0.5414614452967299,
      "grad_norm": 77.92671203613281,
      "learning_rate": 9.144581533389454e-06,
      "loss": 3.496,
      "step": 13412
    },
    {
      "epoch": 0.5415018167137666,
      "grad_norm": 104.06611633300781,
      "learning_rate": 9.143278753145612e-06,
      "loss": 6.0794,
      "step": 13413
    },
    {
      "epoch": 0.5415421881308033,
      "grad_norm": 103.45101165771484,
      "learning_rate": 9.141975987549702e-06,
      "loss": 5.4672,
      "step": 13414
    },
    {
      "epoch": 0.5415825595478402,
      "grad_norm": 107.25777435302734,
      "learning_rate": 9.140673236623998e-06,
      "loss": 4.8932,
      "step": 13415
    },
    {
      "epoch": 0.5416229309648769,
      "grad_norm": 71.89405822753906,
      "learning_rate": 9.139370500390777e-06,
      "loss": 3.8917,
      "step": 13416
    },
    {
      "epoch": 0.5416633023819136,
      "grad_norm": 95.5439224243164,
      "learning_rate": 9.138067778872314e-06,
      "loss": 4.4048,
      "step": 13417
    },
    {
      "epoch": 0.5417036737989503,
      "grad_norm": 133.67404174804688,
      "learning_rate": 9.136765072090875e-06,
      "loss": 5.5533,
      "step": 13418
    },
    {
      "epoch": 0.5417440452159871,
      "grad_norm": 174.76853942871094,
      "learning_rate": 9.135462380068737e-06,
      "loss": 4.702,
      "step": 13419
    },
    {
      "epoch": 0.5417844166330238,
      "grad_norm": 147.81263732910156,
      "learning_rate": 9.134159702828174e-06,
      "loss": 5.3076,
      "step": 13420
    },
    {
      "epoch": 0.5418247880500605,
      "grad_norm": 128.25009155273438,
      "learning_rate": 9.13285704039146e-06,
      "loss": 5.6196,
      "step": 13421
    },
    {
      "epoch": 0.5418651594670973,
      "grad_norm": 89.32848358154297,
      "learning_rate": 9.131554392780863e-06,
      "loss": 4.9868,
      "step": 13422
    },
    {
      "epoch": 0.5419055308841341,
      "grad_norm": 96.98342895507812,
      "learning_rate": 9.13025176001866e-06,
      "loss": 5.0476,
      "step": 13423
    },
    {
      "epoch": 0.5419459023011708,
      "grad_norm": 103.23819732666016,
      "learning_rate": 9.128949142127118e-06,
      "loss": 5.783,
      "step": 13424
    },
    {
      "epoch": 0.5419862737182075,
      "grad_norm": 85.09712982177734,
      "learning_rate": 9.127646539128519e-06,
      "loss": 5.7536,
      "step": 13425
    },
    {
      "epoch": 0.5420266451352442,
      "grad_norm": 125.8720703125,
      "learning_rate": 9.12634395104512e-06,
      "loss": 4.7444,
      "step": 13426
    },
    {
      "epoch": 0.542067016552281,
      "grad_norm": 90.47700500488281,
      "learning_rate": 9.125041377899201e-06,
      "loss": 6.7637,
      "step": 13427
    },
    {
      "epoch": 0.5421073879693177,
      "grad_norm": 142.6474151611328,
      "learning_rate": 9.12373881971303e-06,
      "loss": 4.0467,
      "step": 13428
    },
    {
      "epoch": 0.5421477593863545,
      "grad_norm": 68.39449310302734,
      "learning_rate": 9.122436276508882e-06,
      "loss": 5.1207,
      "step": 13429
    },
    {
      "epoch": 0.5421881308033912,
      "grad_norm": 89.88326263427734,
      "learning_rate": 9.121133748309023e-06,
      "loss": 5.1634,
      "step": 13430
    },
    {
      "epoch": 0.542228502220428,
      "grad_norm": 80.77652740478516,
      "learning_rate": 9.119831235135726e-06,
      "loss": 6.6219,
      "step": 13431
    },
    {
      "epoch": 0.5422688736374647,
      "grad_norm": 67.27645111083984,
      "learning_rate": 9.118528737011256e-06,
      "loss": 4.181,
      "step": 13432
    },
    {
      "epoch": 0.5423092450545014,
      "grad_norm": 170.32350158691406,
      "learning_rate": 9.11722625395789e-06,
      "loss": 5.4637,
      "step": 13433
    },
    {
      "epoch": 0.5423496164715381,
      "grad_norm": 102.42219543457031,
      "learning_rate": 9.11592378599789e-06,
      "loss": 3.91,
      "step": 13434
    },
    {
      "epoch": 0.5423899878885748,
      "grad_norm": 86.12384796142578,
      "learning_rate": 9.11462133315353e-06,
      "loss": 3.0657,
      "step": 13435
    },
    {
      "epoch": 0.5424303593056117,
      "grad_norm": 82.15023803710938,
      "learning_rate": 9.11331889544708e-06,
      "loss": 4.3869,
      "step": 13436
    },
    {
      "epoch": 0.5424707307226484,
      "grad_norm": 178.84066772460938,
      "learning_rate": 9.112016472900806e-06,
      "loss": 2.6474,
      "step": 13437
    },
    {
      "epoch": 0.5425111021396851,
      "grad_norm": 119.32250213623047,
      "learning_rate": 9.110714065536976e-06,
      "loss": 4.5139,
      "step": 13438
    },
    {
      "epoch": 0.5425514735567218,
      "grad_norm": 77.8028335571289,
      "learning_rate": 9.109411673377857e-06,
      "loss": 6.2908,
      "step": 13439
    },
    {
      "epoch": 0.5425918449737586,
      "grad_norm": 251.7618408203125,
      "learning_rate": 9.108109296445719e-06,
      "loss": 6.125,
      "step": 13440
    },
    {
      "epoch": 0.5426322163907953,
      "grad_norm": 146.43649291992188,
      "learning_rate": 9.10680693476283e-06,
      "loss": 5.9898,
      "step": 13441
    },
    {
      "epoch": 0.542672587807832,
      "grad_norm": 138.02464294433594,
      "learning_rate": 9.105504588351457e-06,
      "loss": 5.7276,
      "step": 13442
    },
    {
      "epoch": 0.5427129592248688,
      "grad_norm": 85.41825103759766,
      "learning_rate": 9.104202257233866e-06,
      "loss": 6.77,
      "step": 13443
    },
    {
      "epoch": 0.5427533306419056,
      "grad_norm": 108.15375518798828,
      "learning_rate": 9.102899941432324e-06,
      "loss": 4.6783,
      "step": 13444
    },
    {
      "epoch": 0.5427937020589423,
      "grad_norm": 131.94320678710938,
      "learning_rate": 9.101597640969104e-06,
      "loss": 6.394,
      "step": 13445
    },
    {
      "epoch": 0.542834073475979,
      "grad_norm": 92.08158874511719,
      "learning_rate": 9.10029535586646e-06,
      "loss": 5.4484,
      "step": 13446
    },
    {
      "epoch": 0.5428744448930157,
      "grad_norm": 161.12588500976562,
      "learning_rate": 9.098993086146666e-06,
      "loss": 5.2412,
      "step": 13447
    },
    {
      "epoch": 0.5429148163100525,
      "grad_norm": 96.39019012451172,
      "learning_rate": 9.097690831831986e-06,
      "loss": 4.4992,
      "step": 13448
    },
    {
      "epoch": 0.5429551877270892,
      "grad_norm": 84.12596130371094,
      "learning_rate": 9.096388592944686e-06,
      "loss": 4.7968,
      "step": 13449
    },
    {
      "epoch": 0.542995559144126,
      "grad_norm": 95.98359680175781,
      "learning_rate": 9.095086369507031e-06,
      "loss": 5.3463,
      "step": 13450
    },
    {
      "epoch": 0.5430359305611627,
      "grad_norm": 110.1780776977539,
      "learning_rate": 9.093784161541286e-06,
      "loss": 5.6685,
      "step": 13451
    },
    {
      "epoch": 0.5430763019781994,
      "grad_norm": 116.89337921142578,
      "learning_rate": 9.092481969069715e-06,
      "loss": 4.4262,
      "step": 13452
    },
    {
      "epoch": 0.5431166733952362,
      "grad_norm": 105.8340835571289,
      "learning_rate": 9.091179792114586e-06,
      "loss": 7.6182,
      "step": 13453
    },
    {
      "epoch": 0.5431570448122729,
      "grad_norm": 634.22607421875,
      "learning_rate": 9.089877630698159e-06,
      "loss": 4.9061,
      "step": 13454
    },
    {
      "epoch": 0.5431974162293096,
      "grad_norm": 93.27888488769531,
      "learning_rate": 9.088575484842697e-06,
      "loss": 4.0952,
      "step": 13455
    },
    {
      "epoch": 0.5432377876463463,
      "grad_norm": 123.4871597290039,
      "learning_rate": 9.087273354570466e-06,
      "loss": 4.4288,
      "step": 13456
    },
    {
      "epoch": 0.5432781590633832,
      "grad_norm": 87.86742401123047,
      "learning_rate": 9.085971239903735e-06,
      "loss": 5.9799,
      "step": 13457
    },
    {
      "epoch": 0.5433185304804199,
      "grad_norm": 70.76800537109375,
      "learning_rate": 9.084669140864757e-06,
      "loss": 5.1585,
      "step": 13458
    },
    {
      "epoch": 0.5433589018974566,
      "grad_norm": 425.5811767578125,
      "learning_rate": 9.083367057475798e-06,
      "loss": 5.2847,
      "step": 13459
    },
    {
      "epoch": 0.5433992733144933,
      "grad_norm": 85.09548950195312,
      "learning_rate": 9.082064989759123e-06,
      "loss": 7.3528,
      "step": 13460
    },
    {
      "epoch": 0.5434396447315301,
      "grad_norm": 113.49100494384766,
      "learning_rate": 9.080762937736993e-06,
      "loss": 6.7381,
      "step": 13461
    },
    {
      "epoch": 0.5434800161485668,
      "grad_norm": 81.27800750732422,
      "learning_rate": 9.07946090143167e-06,
      "loss": 5.1494,
      "step": 13462
    },
    {
      "epoch": 0.5435203875656035,
      "grad_norm": 85.4567642211914,
      "learning_rate": 9.078158880865416e-06,
      "loss": 4.1973,
      "step": 13463
    },
    {
      "epoch": 0.5435607589826403,
      "grad_norm": 114.38581085205078,
      "learning_rate": 9.076856876060491e-06,
      "loss": 5.5248,
      "step": 13464
    },
    {
      "epoch": 0.5436011303996771,
      "grad_norm": 121.13723754882812,
      "learning_rate": 9.075554887039165e-06,
      "loss": 5.0005,
      "step": 13465
    },
    {
      "epoch": 0.5436415018167138,
      "grad_norm": 87.52826690673828,
      "learning_rate": 9.074252913823685e-06,
      "loss": 6.3869,
      "step": 13466
    },
    {
      "epoch": 0.5436818732337505,
      "grad_norm": 93.28546905517578,
      "learning_rate": 9.072950956436318e-06,
      "loss": 6.155,
      "step": 13467
    },
    {
      "epoch": 0.5437222446507872,
      "grad_norm": 93.20880126953125,
      "learning_rate": 9.071649014899327e-06,
      "loss": 5.258,
      "step": 13468
    },
    {
      "epoch": 0.543762616067824,
      "grad_norm": 92.14383697509766,
      "learning_rate": 9.07034708923497e-06,
      "loss": 4.2957,
      "step": 13469
    },
    {
      "epoch": 0.5438029874848607,
      "grad_norm": 89.14502716064453,
      "learning_rate": 9.069045179465506e-06,
      "loss": 5.3506,
      "step": 13470
    },
    {
      "epoch": 0.5438433589018975,
      "grad_norm": 226.8288116455078,
      "learning_rate": 9.067743285613195e-06,
      "loss": 3.5468,
      "step": 13471
    },
    {
      "epoch": 0.5438837303189342,
      "grad_norm": 178.05288696289062,
      "learning_rate": 9.066441407700296e-06,
      "loss": 7.4837,
      "step": 13472
    },
    {
      "epoch": 0.5439241017359709,
      "grad_norm": 79.9779052734375,
      "learning_rate": 9.065139545749075e-06,
      "loss": 4.4713,
      "step": 13473
    },
    {
      "epoch": 0.5439644731530077,
      "grad_norm": 161.8339385986328,
      "learning_rate": 9.06383769978178e-06,
      "loss": 3.9541,
      "step": 13474
    },
    {
      "epoch": 0.5440048445700444,
      "grad_norm": 101.57024383544922,
      "learning_rate": 9.062535869820674e-06,
      "loss": 4.8842,
      "step": 13475
    },
    {
      "epoch": 0.5440452159870811,
      "grad_norm": 124.9006118774414,
      "learning_rate": 9.061234055888013e-06,
      "loss": 6.3326,
      "step": 13476
    },
    {
      "epoch": 0.5440855874041178,
      "grad_norm": 139.3983154296875,
      "learning_rate": 9.059932258006065e-06,
      "loss": 6.3323,
      "step": 13477
    },
    {
      "epoch": 0.5441259588211547,
      "grad_norm": 87.6942367553711,
      "learning_rate": 9.058630476197075e-06,
      "loss": 5.1868,
      "step": 13478
    },
    {
      "epoch": 0.5441663302381914,
      "grad_norm": 118.65765380859375,
      "learning_rate": 9.057328710483307e-06,
      "loss": 5.698,
      "step": 13479
    },
    {
      "epoch": 0.5442067016552281,
      "grad_norm": 82.64659118652344,
      "learning_rate": 9.056026960887015e-06,
      "loss": 5.0837,
      "step": 13480
    },
    {
      "epoch": 0.5442470730722648,
      "grad_norm": 153.23440551757812,
      "learning_rate": 9.054725227430459e-06,
      "loss": 4.3889,
      "step": 13481
    },
    {
      "epoch": 0.5442874444893016,
      "grad_norm": 81.82972717285156,
      "learning_rate": 9.053423510135892e-06,
      "loss": 5.6722,
      "step": 13482
    },
    {
      "epoch": 0.5443278159063383,
      "grad_norm": 105.75829315185547,
      "learning_rate": 9.052121809025574e-06,
      "loss": 5.9366,
      "step": 13483
    },
    {
      "epoch": 0.544368187323375,
      "grad_norm": 44.80937194824219,
      "learning_rate": 9.05082012412176e-06,
      "loss": 2.2084,
      "step": 13484
    },
    {
      "epoch": 0.5444085587404118,
      "grad_norm": 141.9021759033203,
      "learning_rate": 9.049518455446708e-06,
      "loss": 4.2967,
      "step": 13485
    },
    {
      "epoch": 0.5444489301574486,
      "grad_norm": 71.17105865478516,
      "learning_rate": 9.048216803022668e-06,
      "loss": 4.061,
      "step": 13486
    },
    {
      "epoch": 0.5444893015744853,
      "grad_norm": 115.5749740600586,
      "learning_rate": 9.046915166871898e-06,
      "loss": 5.8133,
      "step": 13487
    },
    {
      "epoch": 0.544529672991522,
      "grad_norm": 87.2171630859375,
      "learning_rate": 9.045613547016653e-06,
      "loss": 5.3804,
      "step": 13488
    },
    {
      "epoch": 0.5445700444085587,
      "grad_norm": 118.3319091796875,
      "learning_rate": 9.04431194347919e-06,
      "loss": 4.8419,
      "step": 13489
    },
    {
      "epoch": 0.5446104158255954,
      "grad_norm": 92.65650177001953,
      "learning_rate": 9.043010356281758e-06,
      "loss": 4.6143,
      "step": 13490
    },
    {
      "epoch": 0.5446507872426322,
      "grad_norm": 82.5089340209961,
      "learning_rate": 9.041708785446615e-06,
      "loss": 4.8843,
      "step": 13491
    },
    {
      "epoch": 0.544691158659669,
      "grad_norm": 109.9066162109375,
      "learning_rate": 9.040407230996013e-06,
      "loss": 4.1651,
      "step": 13492
    },
    {
      "epoch": 0.5447315300767057,
      "grad_norm": 89.72257232666016,
      "learning_rate": 9.039105692952211e-06,
      "loss": 5.1409,
      "step": 13493
    },
    {
      "epoch": 0.5447719014937424,
      "grad_norm": 78.96546936035156,
      "learning_rate": 9.037804171337455e-06,
      "loss": 5.6297,
      "step": 13494
    },
    {
      "epoch": 0.5448122729107792,
      "grad_norm": 79.30349731445312,
      "learning_rate": 9.036502666173999e-06,
      "loss": 5.1574,
      "step": 13495
    },
    {
      "epoch": 0.5448526443278159,
      "grad_norm": 100.99940490722656,
      "learning_rate": 9.035201177484098e-06,
      "loss": 4.9578,
      "step": 13496
    },
    {
      "epoch": 0.5448930157448526,
      "grad_norm": 111.7148666381836,
      "learning_rate": 9.033899705290009e-06,
      "loss": 5.0751,
      "step": 13497
    },
    {
      "epoch": 0.5449333871618894,
      "grad_norm": 118.45548248291016,
      "learning_rate": 9.032598249613974e-06,
      "loss": 5.5637,
      "step": 13498
    },
    {
      "epoch": 0.5449737585789262,
      "grad_norm": 111.6148452758789,
      "learning_rate": 9.031296810478251e-06,
      "loss": 4.4889,
      "step": 13499
    },
    {
      "epoch": 0.5450141299959629,
      "grad_norm": 75.79707336425781,
      "learning_rate": 9.029995387905089e-06,
      "loss": 6.7935,
      "step": 13500
    },
    {
      "epoch": 0.5450545014129996,
      "grad_norm": 428.4623718261719,
      "learning_rate": 9.028693981916743e-06,
      "loss": 5.0877,
      "step": 13501
    },
    {
      "epoch": 0.5450948728300363,
      "grad_norm": 83.88690948486328,
      "learning_rate": 9.02739259253546e-06,
      "loss": 3.8609,
      "step": 13502
    },
    {
      "epoch": 0.545135244247073,
      "grad_norm": 91.07427215576172,
      "learning_rate": 9.026091219783493e-06,
      "loss": 5.5477,
      "step": 13503
    },
    {
      "epoch": 0.5451756156641098,
      "grad_norm": 100.86920166015625,
      "learning_rate": 9.024789863683091e-06,
      "loss": 5.3259,
      "step": 13504
    },
    {
      "epoch": 0.5452159870811466,
      "grad_norm": 101.71846771240234,
      "learning_rate": 9.023488524256509e-06,
      "loss": 5.9666,
      "step": 13505
    },
    {
      "epoch": 0.5452563584981833,
      "grad_norm": 100.55702209472656,
      "learning_rate": 9.02218720152599e-06,
      "loss": 4.4272,
      "step": 13506
    },
    {
      "epoch": 0.54529672991522,
      "grad_norm": 84.2918930053711,
      "learning_rate": 9.020885895513785e-06,
      "loss": 4.6311,
      "step": 13507
    },
    {
      "epoch": 0.5453371013322568,
      "grad_norm": 71.41683959960938,
      "learning_rate": 9.019584606242147e-06,
      "loss": 5.1716,
      "step": 13508
    },
    {
      "epoch": 0.5453774727492935,
      "grad_norm": 166.9352569580078,
      "learning_rate": 9.018283333733322e-06,
      "loss": 4.2463,
      "step": 13509
    },
    {
      "epoch": 0.5454178441663302,
      "grad_norm": 105.99272918701172,
      "learning_rate": 9.01698207800956e-06,
      "loss": 5.5215,
      "step": 13510
    },
    {
      "epoch": 0.5454582155833669,
      "grad_norm": 113.153564453125,
      "learning_rate": 9.015680839093109e-06,
      "loss": 4.4582,
      "step": 13511
    },
    {
      "epoch": 0.5454985870004038,
      "grad_norm": 236.36593627929688,
      "learning_rate": 9.014379617006214e-06,
      "loss": 5.5894,
      "step": 13512
    },
    {
      "epoch": 0.5455389584174405,
      "grad_norm": 111.32707977294922,
      "learning_rate": 9.013078411771134e-06,
      "loss": 4.9287,
      "step": 13513
    },
    {
      "epoch": 0.5455793298344772,
      "grad_norm": 67.09994506835938,
      "learning_rate": 9.011777223410103e-06,
      "loss": 5.9516,
      "step": 13514
    },
    {
      "epoch": 0.5456197012515139,
      "grad_norm": 135.227294921875,
      "learning_rate": 9.010476051945372e-06,
      "loss": 5.8139,
      "step": 13515
    },
    {
      "epoch": 0.5456600726685507,
      "grad_norm": 120.30363464355469,
      "learning_rate": 9.00917489739919e-06,
      "loss": 5.8938,
      "step": 13516
    },
    {
      "epoch": 0.5457004440855874,
      "grad_norm": 163.444091796875,
      "learning_rate": 9.00787375979381e-06,
      "loss": 6.4227,
      "step": 13517
    },
    {
      "epoch": 0.5457408155026241,
      "grad_norm": 92.78904724121094,
      "learning_rate": 9.006572639151466e-06,
      "loss": 6.1943,
      "step": 13518
    },
    {
      "epoch": 0.5457811869196609,
      "grad_norm": 74.22026062011719,
      "learning_rate": 9.005271535494411e-06,
      "loss": 4.7798,
      "step": 13519
    },
    {
      "epoch": 0.5458215583366977,
      "grad_norm": 86.02880859375,
      "learning_rate": 9.00397044884489e-06,
      "loss": 3.8282,
      "step": 13520
    },
    {
      "epoch": 0.5458619297537344,
      "grad_norm": 68.08789825439453,
      "learning_rate": 9.00266937922515e-06,
      "loss": 5.5066,
      "step": 13521
    },
    {
      "epoch": 0.5459023011707711,
      "grad_norm": 69.03258514404297,
      "learning_rate": 9.001368326657432e-06,
      "loss": 3.8568,
      "step": 13522
    },
    {
      "epoch": 0.5459426725878078,
      "grad_norm": 94.16804504394531,
      "learning_rate": 9.000067291163984e-06,
      "loss": 6.0862,
      "step": 13523
    },
    {
      "epoch": 0.5459830440048445,
      "grad_norm": 94.779296875,
      "learning_rate": 8.998766272767052e-06,
      "loss": 4.0457,
      "step": 13524
    },
    {
      "epoch": 0.5460234154218813,
      "grad_norm": 185.70059204101562,
      "learning_rate": 8.997465271488882e-06,
      "loss": 5.1178,
      "step": 13525
    },
    {
      "epoch": 0.5460637868389181,
      "grad_norm": 84.91722106933594,
      "learning_rate": 8.99616428735171e-06,
      "loss": 4.8769,
      "step": 13526
    },
    {
      "epoch": 0.5461041582559548,
      "grad_norm": 86.32234954833984,
      "learning_rate": 8.994863320377784e-06,
      "loss": 4.029,
      "step": 13527
    },
    {
      "epoch": 0.5461445296729915,
      "grad_norm": 119.60604858398438,
      "learning_rate": 8.993562370589348e-06,
      "loss": 5.1172,
      "step": 13528
    },
    {
      "epoch": 0.5461849010900283,
      "grad_norm": 146.6290283203125,
      "learning_rate": 8.992261438008649e-06,
      "loss": 5.1539,
      "step": 13529
    },
    {
      "epoch": 0.546225272507065,
      "grad_norm": 123.45610046386719,
      "learning_rate": 8.990960522657922e-06,
      "loss": 3.9665,
      "step": 13530
    },
    {
      "epoch": 0.5462656439241017,
      "grad_norm": 100.20852661132812,
      "learning_rate": 8.989659624559414e-06,
      "loss": 5.1303,
      "step": 13531
    },
    {
      "epoch": 0.5463060153411384,
      "grad_norm": 88.47834014892578,
      "learning_rate": 8.988358743735369e-06,
      "loss": 5.5893,
      "step": 13532
    },
    {
      "epoch": 0.5463463867581753,
      "grad_norm": 113.28425598144531,
      "learning_rate": 8.98705788020803e-06,
      "loss": 3.8472,
      "step": 13533
    },
    {
      "epoch": 0.546386758175212,
      "grad_norm": 72.97123718261719,
      "learning_rate": 8.985757033999632e-06,
      "loss": 5.3274,
      "step": 13534
    },
    {
      "epoch": 0.5464271295922487,
      "grad_norm": 98.2707290649414,
      "learning_rate": 8.98445620513242e-06,
      "loss": 4.5821,
      "step": 13535
    },
    {
      "epoch": 0.5464675010092854,
      "grad_norm": 107.42525482177734,
      "learning_rate": 8.983155393628636e-06,
      "loss": 5.2871,
      "step": 13536
    },
    {
      "epoch": 0.5465078724263221,
      "grad_norm": 110.94160461425781,
      "learning_rate": 8.981854599510522e-06,
      "loss": 4.8613,
      "step": 13537
    },
    {
      "epoch": 0.5465482438433589,
      "grad_norm": 94.80768585205078,
      "learning_rate": 8.980553822800314e-06,
      "loss": 5.7409,
      "step": 13538
    },
    {
      "epoch": 0.5465886152603956,
      "grad_norm": 89.48336791992188,
      "learning_rate": 8.979253063520256e-06,
      "loss": 5.5622,
      "step": 13539
    },
    {
      "epoch": 0.5466289866774324,
      "grad_norm": 77.75270080566406,
      "learning_rate": 8.977952321692586e-06,
      "loss": 5.1711,
      "step": 13540
    },
    {
      "epoch": 0.5466693580944691,
      "grad_norm": 107.16268157958984,
      "learning_rate": 8.976651597339551e-06,
      "loss": 5.2041,
      "step": 13541
    },
    {
      "epoch": 0.5467097295115059,
      "grad_norm": 83.21221160888672,
      "learning_rate": 8.975350890483378e-06,
      "loss": 5.7805,
      "step": 13542
    },
    {
      "epoch": 0.5467501009285426,
      "grad_norm": 86.4626693725586,
      "learning_rate": 8.974050201146313e-06,
      "loss": 4.3617,
      "step": 13543
    },
    {
      "epoch": 0.5467904723455793,
      "grad_norm": 175.64247131347656,
      "learning_rate": 8.972749529350593e-06,
      "loss": 5.2042,
      "step": 13544
    },
    {
      "epoch": 0.546830843762616,
      "grad_norm": 120.08222198486328,
      "learning_rate": 8.971448875118462e-06,
      "loss": 3.8109,
      "step": 13545
    },
    {
      "epoch": 0.5468712151796528,
      "grad_norm": 100.2588119506836,
      "learning_rate": 8.970148238472149e-06,
      "loss": 6.4475,
      "step": 13546
    },
    {
      "epoch": 0.5469115865966896,
      "grad_norm": 117.32630920410156,
      "learning_rate": 8.968847619433896e-06,
      "loss": 4.6585,
      "step": 13547
    },
    {
      "epoch": 0.5469519580137263,
      "grad_norm": 76.77510070800781,
      "learning_rate": 8.96754701802594e-06,
      "loss": 4.6525,
      "step": 13548
    },
    {
      "epoch": 0.546992329430763,
      "grad_norm": 71.29077911376953,
      "learning_rate": 8.966246434270522e-06,
      "loss": 4.6705,
      "step": 13549
    },
    {
      "epoch": 0.5470327008477998,
      "grad_norm": 91.98865509033203,
      "learning_rate": 8.964945868189875e-06,
      "loss": 5.0525,
      "step": 13550
    },
    {
      "epoch": 0.5470730722648365,
      "grad_norm": 71.12567901611328,
      "learning_rate": 8.963645319806235e-06,
      "loss": 4.5687,
      "step": 13551
    },
    {
      "epoch": 0.5471134436818732,
      "grad_norm": 80.61266326904297,
      "learning_rate": 8.962344789141841e-06,
      "loss": 4.8247,
      "step": 13552
    },
    {
      "epoch": 0.5471538150989099,
      "grad_norm": 70.41328430175781,
      "learning_rate": 8.961044276218931e-06,
      "loss": 6.9012,
      "step": 13553
    },
    {
      "epoch": 0.5471941865159468,
      "grad_norm": 136.21139526367188,
      "learning_rate": 8.959743781059736e-06,
      "loss": 5.9634,
      "step": 13554
    },
    {
      "epoch": 0.5472345579329835,
      "grad_norm": 94.54135131835938,
      "learning_rate": 8.95844330368649e-06,
      "loss": 7.0432,
      "step": 13555
    },
    {
      "epoch": 0.5472749293500202,
      "grad_norm": 85.56639099121094,
      "learning_rate": 8.957142844121433e-06,
      "loss": 2.9426,
      "step": 13556
    },
    {
      "epoch": 0.5473153007670569,
      "grad_norm": 121.64834594726562,
      "learning_rate": 8.955842402386801e-06,
      "loss": 5.7056,
      "step": 13557
    },
    {
      "epoch": 0.5473556721840936,
      "grad_norm": 83.09637451171875,
      "learning_rate": 8.954541978504823e-06,
      "loss": 5.5004,
      "step": 13558
    },
    {
      "epoch": 0.5473960436011304,
      "grad_norm": 111.15070343017578,
      "learning_rate": 8.953241572497734e-06,
      "loss": 6.4114,
      "step": 13559
    },
    {
      "epoch": 0.5474364150181671,
      "grad_norm": 118.62242889404297,
      "learning_rate": 8.951941184387772e-06,
      "loss": 4.6595,
      "step": 13560
    },
    {
      "epoch": 0.5474767864352039,
      "grad_norm": 123.6997299194336,
      "learning_rate": 8.950640814197171e-06,
      "loss": 5.6243,
      "step": 13561
    },
    {
      "epoch": 0.5475171578522406,
      "grad_norm": 86.1698989868164,
      "learning_rate": 8.949340461948158e-06,
      "loss": 5.2195,
      "step": 13562
    },
    {
      "epoch": 0.5475575292692774,
      "grad_norm": 91.17757415771484,
      "learning_rate": 8.94804012766297e-06,
      "loss": 5.4747,
      "step": 13563
    },
    {
      "epoch": 0.5475979006863141,
      "grad_norm": 81.58241271972656,
      "learning_rate": 8.94673981136384e-06,
      "loss": 4.9701,
      "step": 13564
    },
    {
      "epoch": 0.5476382721033508,
      "grad_norm": 72.47957611083984,
      "learning_rate": 8.945439513073002e-06,
      "loss": 6.0184,
      "step": 13565
    },
    {
      "epoch": 0.5476786435203875,
      "grad_norm": 103.70928955078125,
      "learning_rate": 8.944139232812684e-06,
      "loss": 5.018,
      "step": 13566
    },
    {
      "epoch": 0.5477190149374243,
      "grad_norm": 106.29139709472656,
      "learning_rate": 8.942838970605116e-06,
      "loss": 3.9024,
      "step": 13567
    },
    {
      "epoch": 0.5477593863544611,
      "grad_norm": 85.82310485839844,
      "learning_rate": 8.941538726472537e-06,
      "loss": 6.3569,
      "step": 13568
    },
    {
      "epoch": 0.5477997577714978,
      "grad_norm": 97.01042938232422,
      "learning_rate": 8.940238500437172e-06,
      "loss": 5.6895,
      "step": 13569
    },
    {
      "epoch": 0.5478401291885345,
      "grad_norm": 116.26961517333984,
      "learning_rate": 8.938938292521255e-06,
      "loss": 4.5344,
      "step": 13570
    },
    {
      "epoch": 0.5478805006055713,
      "grad_norm": 120.20681762695312,
      "learning_rate": 8.937638102747015e-06,
      "loss": 5.6434,
      "step": 13571
    },
    {
      "epoch": 0.547920872022608,
      "grad_norm": 98.56039428710938,
      "learning_rate": 8.936337931136682e-06,
      "loss": 6.4465,
      "step": 13572
    },
    {
      "epoch": 0.5479612434396447,
      "grad_norm": 112.14021301269531,
      "learning_rate": 8.93503777771249e-06,
      "loss": 5.2682,
      "step": 13573
    },
    {
      "epoch": 0.5480016148566814,
      "grad_norm": 221.1099090576172,
      "learning_rate": 8.933737642496663e-06,
      "loss": 4.2337,
      "step": 13574
    },
    {
      "epoch": 0.5480419862737183,
      "grad_norm": 210.12139892578125,
      "learning_rate": 8.93243752551143e-06,
      "loss": 4.1585,
      "step": 13575
    },
    {
      "epoch": 0.548082357690755,
      "grad_norm": 122.89022827148438,
      "learning_rate": 8.931137426779024e-06,
      "loss": 4.5644,
      "step": 13576
    },
    {
      "epoch": 0.5481227291077917,
      "grad_norm": 88.36494445800781,
      "learning_rate": 8.929837346321674e-06,
      "loss": 4.8944,
      "step": 13577
    },
    {
      "epoch": 0.5481631005248284,
      "grad_norm": 99.3958740234375,
      "learning_rate": 8.928537284161604e-06,
      "loss": 5.6186,
      "step": 13578
    },
    {
      "epoch": 0.5482034719418651,
      "grad_norm": 77.67816925048828,
      "learning_rate": 8.927237240321044e-06,
      "loss": 6.0131,
      "step": 13579
    },
    {
      "epoch": 0.5482438433589019,
      "grad_norm": 73.93993377685547,
      "learning_rate": 8.925937214822223e-06,
      "loss": 5.0602,
      "step": 13580
    },
    {
      "epoch": 0.5482842147759386,
      "grad_norm": 93.92018127441406,
      "learning_rate": 8.924637207687372e-06,
      "loss": 4.8544,
      "step": 13581
    },
    {
      "epoch": 0.5483245861929754,
      "grad_norm": 81.7276840209961,
      "learning_rate": 8.923337218938707e-06,
      "loss": 4.5937,
      "step": 13582
    },
    {
      "epoch": 0.5483649576100121,
      "grad_norm": 113.87460327148438,
      "learning_rate": 8.922037248598463e-06,
      "loss": 5.8027,
      "step": 13583
    },
    {
      "epoch": 0.5484053290270489,
      "grad_norm": 78.59388732910156,
      "learning_rate": 8.920737296688865e-06,
      "loss": 4.5695,
      "step": 13584
    },
    {
      "epoch": 0.5484457004440856,
      "grad_norm": 164.67465209960938,
      "learning_rate": 8.919437363232144e-06,
      "loss": 4.2446,
      "step": 13585
    },
    {
      "epoch": 0.5484860718611223,
      "grad_norm": 78.46976470947266,
      "learning_rate": 8.918137448250515e-06,
      "loss": 4.784,
      "step": 13586
    },
    {
      "epoch": 0.548526443278159,
      "grad_norm": 134.66189575195312,
      "learning_rate": 8.91683755176621e-06,
      "loss": 5.2478,
      "step": 13587
    },
    {
      "epoch": 0.5485668146951957,
      "grad_norm": 132.6804656982422,
      "learning_rate": 8.915537673801453e-06,
      "loss": 4.1785,
      "step": 13588
    },
    {
      "epoch": 0.5486071861122326,
      "grad_norm": 88.04251098632812,
      "learning_rate": 8.914237814378472e-06,
      "loss": 4.1796,
      "step": 13589
    },
    {
      "epoch": 0.5486475575292693,
      "grad_norm": 90.0814437866211,
      "learning_rate": 8.912937973519486e-06,
      "loss": 5.7178,
      "step": 13590
    },
    {
      "epoch": 0.548687928946306,
      "grad_norm": 102.97843170166016,
      "learning_rate": 8.911638151246723e-06,
      "loss": 6.5224,
      "step": 13591
    },
    {
      "epoch": 0.5487283003633427,
      "grad_norm": 84.80990600585938,
      "learning_rate": 8.910338347582405e-06,
      "loss": 5.7912,
      "step": 13592
    },
    {
      "epoch": 0.5487686717803795,
      "grad_norm": 87.7090072631836,
      "learning_rate": 8.909038562548763e-06,
      "loss": 5.9324,
      "step": 13593
    },
    {
      "epoch": 0.5488090431974162,
      "grad_norm": 159.892578125,
      "learning_rate": 8.907738796168009e-06,
      "loss": 6.4225,
      "step": 13594
    },
    {
      "epoch": 0.5488494146144529,
      "grad_norm": 196.1078643798828,
      "learning_rate": 8.90643904846237e-06,
      "loss": 4.206,
      "step": 13595
    },
    {
      "epoch": 0.5488897860314897,
      "grad_norm": 78.43099212646484,
      "learning_rate": 8.90513931945407e-06,
      "loss": 4.6112,
      "step": 13596
    },
    {
      "epoch": 0.5489301574485265,
      "grad_norm": 101.08907318115234,
      "learning_rate": 8.903839609165333e-06,
      "loss": 6.7308,
      "step": 13597
    },
    {
      "epoch": 0.5489705288655632,
      "grad_norm": 129.2641143798828,
      "learning_rate": 8.902539917618378e-06,
      "loss": 4.5553,
      "step": 13598
    },
    {
      "epoch": 0.5490109002825999,
      "grad_norm": 106.53475952148438,
      "learning_rate": 8.901240244835426e-06,
      "loss": 4.0259,
      "step": 13599
    },
    {
      "epoch": 0.5490512716996366,
      "grad_norm": 118.22940826416016,
      "learning_rate": 8.899940590838702e-06,
      "loss": 3.8865,
      "step": 13600
    },
    {
      "epoch": 0.5490916431166734,
      "grad_norm": 98.1136703491211,
      "learning_rate": 8.898640955650428e-06,
      "loss": 5.1634,
      "step": 13601
    },
    {
      "epoch": 0.5491320145337101,
      "grad_norm": 109.63337707519531,
      "learning_rate": 8.897341339292817e-06,
      "loss": 4.5266,
      "step": 13602
    },
    {
      "epoch": 0.5491723859507469,
      "grad_norm": 72.79046630859375,
      "learning_rate": 8.896041741788095e-06,
      "loss": 7.6017,
      "step": 13603
    },
    {
      "epoch": 0.5492127573677836,
      "grad_norm": 107.57217407226562,
      "learning_rate": 8.894742163158481e-06,
      "loss": 4.754,
      "step": 13604
    },
    {
      "epoch": 0.5492531287848204,
      "grad_norm": 98.00397491455078,
      "learning_rate": 8.8934426034262e-06,
      "loss": 5.7673,
      "step": 13605
    },
    {
      "epoch": 0.5492935002018571,
      "grad_norm": 96.77291107177734,
      "learning_rate": 8.892143062613462e-06,
      "loss": 6.0808,
      "step": 13606
    },
    {
      "epoch": 0.5493338716188938,
      "grad_norm": 95.5501480102539,
      "learning_rate": 8.89084354074249e-06,
      "loss": 4.8781,
      "step": 13607
    },
    {
      "epoch": 0.5493742430359305,
      "grad_norm": 86.0048599243164,
      "learning_rate": 8.889544037835504e-06,
      "loss": 4.5334,
      "step": 13608
    },
    {
      "epoch": 0.5494146144529674,
      "grad_norm": 86.4150619506836,
      "learning_rate": 8.888244553914724e-06,
      "loss": 5.751,
      "step": 13609
    },
    {
      "epoch": 0.5494549858700041,
      "grad_norm": 121.98535919189453,
      "learning_rate": 8.886945089002364e-06,
      "loss": 4.6339,
      "step": 13610
    },
    {
      "epoch": 0.5494953572870408,
      "grad_norm": 71.8815689086914,
      "learning_rate": 8.885645643120643e-06,
      "loss": 4.7748,
      "step": 13611
    },
    {
      "epoch": 0.5495357287040775,
      "grad_norm": 107.85076904296875,
      "learning_rate": 8.884346216291779e-06,
      "loss": 5.3991,
      "step": 13612
    },
    {
      "epoch": 0.5495761001211142,
      "grad_norm": 104.972412109375,
      "learning_rate": 8.883046808537995e-06,
      "loss": 4.6325,
      "step": 13613
    },
    {
      "epoch": 0.549616471538151,
      "grad_norm": 128.0458221435547,
      "learning_rate": 8.881747419881496e-06,
      "loss": 4.9922,
      "step": 13614
    },
    {
      "epoch": 0.5496568429551877,
      "grad_norm": 97.31961822509766,
      "learning_rate": 8.880448050344508e-06,
      "loss": 5.373,
      "step": 13615
    },
    {
      "epoch": 0.5496972143722245,
      "grad_norm": 126.79822540283203,
      "learning_rate": 8.879148699949241e-06,
      "loss": 5.3188,
      "step": 13616
    },
    {
      "epoch": 0.5497375857892612,
      "grad_norm": 81.84272766113281,
      "learning_rate": 8.877849368717918e-06,
      "loss": 5.5898,
      "step": 13617
    },
    {
      "epoch": 0.549777957206298,
      "grad_norm": 75.01712036132812,
      "learning_rate": 8.876550056672747e-06,
      "loss": 6.1786,
      "step": 13618
    },
    {
      "epoch": 0.5498183286233347,
      "grad_norm": 101.7120361328125,
      "learning_rate": 8.875250763835948e-06,
      "loss": 7.6775,
      "step": 13619
    },
    {
      "epoch": 0.5498587000403714,
      "grad_norm": 74.39208984375,
      "learning_rate": 8.873951490229733e-06,
      "loss": 5.2452,
      "step": 13620
    },
    {
      "epoch": 0.5498990714574081,
      "grad_norm": 119.9183349609375,
      "learning_rate": 8.872652235876322e-06,
      "loss": 5.504,
      "step": 13621
    },
    {
      "epoch": 0.5499394428744449,
      "grad_norm": 106.97594451904297,
      "learning_rate": 8.871353000797923e-06,
      "loss": 3.3338,
      "step": 13622
    },
    {
      "epoch": 0.5499798142914817,
      "grad_norm": 108.96665954589844,
      "learning_rate": 8.87005378501675e-06,
      "loss": 6.44,
      "step": 13623
    },
    {
      "epoch": 0.5500201857085184,
      "grad_norm": 92.82980346679688,
      "learning_rate": 8.868754588555019e-06,
      "loss": 3.5589,
      "step": 13624
    },
    {
      "epoch": 0.5500605571255551,
      "grad_norm": 81.91725158691406,
      "learning_rate": 8.86745541143495e-06,
      "loss": 4.9194,
      "step": 13625
    },
    {
      "epoch": 0.5501009285425918,
      "grad_norm": 111.36152648925781,
      "learning_rate": 8.86615625367874e-06,
      "loss": 4.1338,
      "step": 13626
    },
    {
      "epoch": 0.5501412999596286,
      "grad_norm": 90.38835144042969,
      "learning_rate": 8.864857115308614e-06,
      "loss": 6.2406,
      "step": 13627
    },
    {
      "epoch": 0.5501816713766653,
      "grad_norm": 99.05628204345703,
      "learning_rate": 8.863557996346778e-06,
      "loss": 5.2381,
      "step": 13628
    },
    {
      "epoch": 0.550222042793702,
      "grad_norm": 88.68401336669922,
      "learning_rate": 8.86225889681545e-06,
      "loss": 6.8192,
      "step": 13629
    },
    {
      "epoch": 0.5502624142107388,
      "grad_norm": 140.0213165283203,
      "learning_rate": 8.860959816736837e-06,
      "loss": 2.8785,
      "step": 13630
    },
    {
      "epoch": 0.5503027856277756,
      "grad_norm": 94.63719940185547,
      "learning_rate": 8.859660756133149e-06,
      "loss": 5.2813,
      "step": 13631
    },
    {
      "epoch": 0.5503431570448123,
      "grad_norm": 214.25601196289062,
      "learning_rate": 8.8583617150266e-06,
      "loss": 4.4306,
      "step": 13632
    },
    {
      "epoch": 0.550383528461849,
      "grad_norm": 101.20842742919922,
      "learning_rate": 8.857062693439404e-06,
      "loss": 3.9312,
      "step": 13633
    },
    {
      "epoch": 0.5504238998788857,
      "grad_norm": 111.8724365234375,
      "learning_rate": 8.855763691393763e-06,
      "loss": 3.625,
      "step": 13634
    },
    {
      "epoch": 0.5504642712959225,
      "grad_norm": 210.97096252441406,
      "learning_rate": 8.85446470891189e-06,
      "loss": 4.5684,
      "step": 13635
    },
    {
      "epoch": 0.5505046427129592,
      "grad_norm": 93.9572982788086,
      "learning_rate": 8.853165746015997e-06,
      "loss": 4.1927,
      "step": 13636
    },
    {
      "epoch": 0.550545014129996,
      "grad_norm": 63.49052429199219,
      "learning_rate": 8.851866802728292e-06,
      "loss": 3.2961,
      "step": 13637
    },
    {
      "epoch": 0.5505853855470327,
      "grad_norm": 135.4992218017578,
      "learning_rate": 8.850567879070984e-06,
      "loss": 5.7493,
      "step": 13638
    },
    {
      "epoch": 0.5506257569640695,
      "grad_norm": 108.6224594116211,
      "learning_rate": 8.84926897506628e-06,
      "loss": 4.9767,
      "step": 13639
    },
    {
      "epoch": 0.5506661283811062,
      "grad_norm": 97.5739517211914,
      "learning_rate": 8.847970090736387e-06,
      "loss": 4.6654,
      "step": 13640
    },
    {
      "epoch": 0.5507064997981429,
      "grad_norm": 106.74333190917969,
      "learning_rate": 8.846671226103523e-06,
      "loss": 4.4741,
      "step": 13641
    },
    {
      "epoch": 0.5507468712151796,
      "grad_norm": 72.40489196777344,
      "learning_rate": 8.845372381189882e-06,
      "loss": 4.3542,
      "step": 13642
    },
    {
      "epoch": 0.5507872426322163,
      "grad_norm": 86.61369323730469,
      "learning_rate": 8.844073556017677e-06,
      "loss": 4.1033,
      "step": 13643
    },
    {
      "epoch": 0.5508276140492532,
      "grad_norm": 98.25677490234375,
      "learning_rate": 8.842774750609114e-06,
      "loss": 4.3764,
      "step": 13644
    },
    {
      "epoch": 0.5508679854662899,
      "grad_norm": 105.76103210449219,
      "learning_rate": 8.841475964986405e-06,
      "loss": 5.3063,
      "step": 13645
    },
    {
      "epoch": 0.5509083568833266,
      "grad_norm": 280.2673645019531,
      "learning_rate": 8.840177199171748e-06,
      "loss": 5.2268,
      "step": 13646
    },
    {
      "epoch": 0.5509487283003633,
      "grad_norm": 102.53498077392578,
      "learning_rate": 8.838878453187353e-06,
      "loss": 6.177,
      "step": 13647
    },
    {
      "epoch": 0.5509890997174001,
      "grad_norm": 86.77947998046875,
      "learning_rate": 8.837579727055424e-06,
      "loss": 5.7284,
      "step": 13648
    },
    {
      "epoch": 0.5510294711344368,
      "grad_norm": 89.02044677734375,
      "learning_rate": 8.836281020798172e-06,
      "loss": 4.256,
      "step": 13649
    },
    {
      "epoch": 0.5510698425514735,
      "grad_norm": 84.57057189941406,
      "learning_rate": 8.834982334437793e-06,
      "loss": 3.8271,
      "step": 13650
    },
    {
      "epoch": 0.5511102139685103,
      "grad_norm": 140.581298828125,
      "learning_rate": 8.833683667996496e-06,
      "loss": 4.6247,
      "step": 13651
    },
    {
      "epoch": 0.5511505853855471,
      "grad_norm": 1237.6922607421875,
      "learning_rate": 8.832385021496484e-06,
      "loss": 5.4098,
      "step": 13652
    },
    {
      "epoch": 0.5511909568025838,
      "grad_norm": 93.49510192871094,
      "learning_rate": 8.831086394959965e-06,
      "loss": 4.8166,
      "step": 13653
    },
    {
      "epoch": 0.5512313282196205,
      "grad_norm": 149.52365112304688,
      "learning_rate": 8.829787788409137e-06,
      "loss": 5.1092,
      "step": 13654
    },
    {
      "epoch": 0.5512716996366572,
      "grad_norm": 134.27603149414062,
      "learning_rate": 8.828489201866203e-06,
      "loss": 3.7993,
      "step": 13655
    },
    {
      "epoch": 0.551312071053694,
      "grad_norm": 99.23345947265625,
      "learning_rate": 8.82719063535337e-06,
      "loss": 5.6682,
      "step": 13656
    },
    {
      "epoch": 0.5513524424707307,
      "grad_norm": 80.18118286132812,
      "learning_rate": 8.82589208889284e-06,
      "loss": 3.9125,
      "step": 13657
    },
    {
      "epoch": 0.5513928138877675,
      "grad_norm": 130.09083557128906,
      "learning_rate": 8.82459356250681e-06,
      "loss": 4.3772,
      "step": 13658
    },
    {
      "epoch": 0.5514331853048042,
      "grad_norm": 85.19488525390625,
      "learning_rate": 8.823295056217487e-06,
      "loss": 4.8081,
      "step": 13659
    },
    {
      "epoch": 0.551473556721841,
      "grad_norm": 71.11713409423828,
      "learning_rate": 8.82199657004707e-06,
      "loss": 5.2534,
      "step": 13660
    },
    {
      "epoch": 0.5515139281388777,
      "grad_norm": 99.40970611572266,
      "learning_rate": 8.820698104017765e-06,
      "loss": 4.7165,
      "step": 13661
    },
    {
      "epoch": 0.5515542995559144,
      "grad_norm": 86.54863739013672,
      "learning_rate": 8.819399658151765e-06,
      "loss": 4.4129,
      "step": 13662
    },
    {
      "epoch": 0.5515946709729511,
      "grad_norm": 103.42918395996094,
      "learning_rate": 8.818101232471273e-06,
      "loss": 5.5841,
      "step": 13663
    },
    {
      "epoch": 0.5516350423899878,
      "grad_norm": 142.6943817138672,
      "learning_rate": 8.816802826998489e-06,
      "loss": 4.6584,
      "step": 13664
    },
    {
      "epoch": 0.5516754138070247,
      "grad_norm": 123.66316986083984,
      "learning_rate": 8.815504441755619e-06,
      "loss": 5.8136,
      "step": 13665
    },
    {
      "epoch": 0.5517157852240614,
      "grad_norm": 112.89738464355469,
      "learning_rate": 8.814206076764854e-06,
      "loss": 5.9112,
      "step": 13666
    },
    {
      "epoch": 0.5517561566410981,
      "grad_norm": 96.82877349853516,
      "learning_rate": 8.812907732048393e-06,
      "loss": 4.5187,
      "step": 13667
    },
    {
      "epoch": 0.5517965280581348,
      "grad_norm": 90.26631164550781,
      "learning_rate": 8.811609407628441e-06,
      "loss": 5.61,
      "step": 13668
    },
    {
      "epoch": 0.5518368994751716,
      "grad_norm": 111.97045135498047,
      "learning_rate": 8.810311103527195e-06,
      "loss": 5.8382,
      "step": 13669
    },
    {
      "epoch": 0.5518772708922083,
      "grad_norm": 100.80522155761719,
      "learning_rate": 8.80901281976685e-06,
      "loss": 4.4271,
      "step": 13670
    },
    {
      "epoch": 0.551917642309245,
      "grad_norm": 85.1618881225586,
      "learning_rate": 8.8077145563696e-06,
      "loss": 5.2846,
      "step": 13671
    },
    {
      "epoch": 0.5519580137262818,
      "grad_norm": 123.49790954589844,
      "learning_rate": 8.806416313357651e-06,
      "loss": 4.6871,
      "step": 13672
    },
    {
      "epoch": 0.5519983851433186,
      "grad_norm": 101.9811019897461,
      "learning_rate": 8.805118090753198e-06,
      "loss": 4.5228,
      "step": 13673
    },
    {
      "epoch": 0.5520387565603553,
      "grad_norm": 84.01543426513672,
      "learning_rate": 8.803819888578432e-06,
      "loss": 6.3264,
      "step": 13674
    },
    {
      "epoch": 0.552079127977392,
      "grad_norm": 78.47848510742188,
      "learning_rate": 8.802521706855552e-06,
      "loss": 5.8037,
      "step": 13675
    },
    {
      "epoch": 0.5521194993944287,
      "grad_norm": 128.99436950683594,
      "learning_rate": 8.801223545606755e-06,
      "loss": 5.523,
      "step": 13676
    },
    {
      "epoch": 0.5521598708114654,
      "grad_norm": 97.99662017822266,
      "learning_rate": 8.799925404854237e-06,
      "loss": 5.3322,
      "step": 13677
    },
    {
      "epoch": 0.5522002422285022,
      "grad_norm": 137.66217041015625,
      "learning_rate": 8.798627284620192e-06,
      "loss": 3.5312,
      "step": 13678
    },
    {
      "epoch": 0.552240613645539,
      "grad_norm": 91.07576751708984,
      "learning_rate": 8.797329184926814e-06,
      "loss": 2.9761,
      "step": 13679
    },
    {
      "epoch": 0.5522809850625757,
      "grad_norm": 97.2206802368164,
      "learning_rate": 8.796031105796299e-06,
      "loss": 3.8503,
      "step": 13680
    },
    {
      "epoch": 0.5523213564796124,
      "grad_norm": 141.5996856689453,
      "learning_rate": 8.794733047250845e-06,
      "loss": 6.8576,
      "step": 13681
    },
    {
      "epoch": 0.5523617278966492,
      "grad_norm": 89.3245620727539,
      "learning_rate": 8.793435009312635e-06,
      "loss": 5.2584,
      "step": 13682
    },
    {
      "epoch": 0.5524020993136859,
      "grad_norm": 160.2249755859375,
      "learning_rate": 8.792136992003871e-06,
      "loss": 4.3524,
      "step": 13683
    },
    {
      "epoch": 0.5524424707307226,
      "grad_norm": 69.78543090820312,
      "learning_rate": 8.790838995346744e-06,
      "loss": 5.6714,
      "step": 13684
    },
    {
      "epoch": 0.5524828421477593,
      "grad_norm": 82.86223602294922,
      "learning_rate": 8.789541019363448e-06,
      "loss": 4.9779,
      "step": 13685
    },
    {
      "epoch": 0.5525232135647962,
      "grad_norm": 125.4063720703125,
      "learning_rate": 8.788243064076171e-06,
      "loss": 4.8054,
      "step": 13686
    },
    {
      "epoch": 0.5525635849818329,
      "grad_norm": 81.37279510498047,
      "learning_rate": 8.786945129507108e-06,
      "loss": 5.1162,
      "step": 13687
    },
    {
      "epoch": 0.5526039563988696,
      "grad_norm": 157.83555603027344,
      "learning_rate": 8.785647215678452e-06,
      "loss": 5.1293,
      "step": 13688
    },
    {
      "epoch": 0.5526443278159063,
      "grad_norm": 251.1357421875,
      "learning_rate": 8.784349322612397e-06,
      "loss": 5.1308,
      "step": 13689
    },
    {
      "epoch": 0.552684699232943,
      "grad_norm": 101.31397247314453,
      "learning_rate": 8.783051450331124e-06,
      "loss": 6.3848,
      "step": 13690
    },
    {
      "epoch": 0.5527250706499798,
      "grad_norm": 88.56031036376953,
      "learning_rate": 8.78175359885683e-06,
      "loss": 5.7939,
      "step": 13691
    },
    {
      "epoch": 0.5527654420670165,
      "grad_norm": 117.455078125,
      "learning_rate": 8.780455768211707e-06,
      "loss": 4.9133,
      "step": 13692
    },
    {
      "epoch": 0.5528058134840533,
      "grad_norm": 110.30485534667969,
      "learning_rate": 8.779157958417944e-06,
      "loss": 4.8049,
      "step": 13693
    },
    {
      "epoch": 0.55284618490109,
      "grad_norm": 65.22249603271484,
      "learning_rate": 8.777860169497725e-06,
      "loss": 4.1323,
      "step": 13694
    },
    {
      "epoch": 0.5528865563181268,
      "grad_norm": 119.62222290039062,
      "learning_rate": 8.776562401473243e-06,
      "loss": 6.5471,
      "step": 13695
    },
    {
      "epoch": 0.5529269277351635,
      "grad_norm": 194.84323120117188,
      "learning_rate": 8.77526465436669e-06,
      "loss": 5.1796,
      "step": 13696
    },
    {
      "epoch": 0.5529672991522002,
      "grad_norm": 76.3346176147461,
      "learning_rate": 8.77396692820025e-06,
      "loss": 4.0665,
      "step": 13697
    },
    {
      "epoch": 0.5530076705692369,
      "grad_norm": 86.77684783935547,
      "learning_rate": 8.772669222996113e-06,
      "loss": 4.8266,
      "step": 13698
    },
    {
      "epoch": 0.5530480419862737,
      "grad_norm": 92.86957550048828,
      "learning_rate": 8.771371538776466e-06,
      "loss": 4.1453,
      "step": 13699
    },
    {
      "epoch": 0.5530884134033105,
      "grad_norm": 95.73811340332031,
      "learning_rate": 8.770073875563494e-06,
      "loss": 3.5048,
      "step": 13700
    },
    {
      "epoch": 0.5531287848203472,
      "grad_norm": 92.6242446899414,
      "learning_rate": 8.768776233379394e-06,
      "loss": 5.4334,
      "step": 13701
    },
    {
      "epoch": 0.5531691562373839,
      "grad_norm": 155.36325073242188,
      "learning_rate": 8.767478612246338e-06,
      "loss": 4.0542,
      "step": 13702
    },
    {
      "epoch": 0.5532095276544207,
      "grad_norm": 103.40432739257812,
      "learning_rate": 8.766181012186523e-06,
      "loss": 4.7935,
      "step": 13703
    },
    {
      "epoch": 0.5532498990714574,
      "grad_norm": 110.54560852050781,
      "learning_rate": 8.76488343322213e-06,
      "loss": 4.8828,
      "step": 13704
    },
    {
      "epoch": 0.5532902704884941,
      "grad_norm": 96.3344497680664,
      "learning_rate": 8.763585875375347e-06,
      "loss": 5.0087,
      "step": 13705
    },
    {
      "epoch": 0.5533306419055308,
      "grad_norm": 101.4189453125,
      "learning_rate": 8.762288338668356e-06,
      "loss": 5.8637,
      "step": 13706
    },
    {
      "epoch": 0.5533710133225677,
      "grad_norm": 115.61272430419922,
      "learning_rate": 8.760990823123346e-06,
      "loss": 4.7315,
      "step": 13707
    },
    {
      "epoch": 0.5534113847396044,
      "grad_norm": 161.06866455078125,
      "learning_rate": 8.7596933287625e-06,
      "loss": 4.6219,
      "step": 13708
    },
    {
      "epoch": 0.5534517561566411,
      "grad_norm": 124.40052795410156,
      "learning_rate": 8.758395855608004e-06,
      "loss": 6.5778,
      "step": 13709
    },
    {
      "epoch": 0.5534921275736778,
      "grad_norm": 70.83123016357422,
      "learning_rate": 8.757098403682038e-06,
      "loss": 5.9512,
      "step": 13710
    },
    {
      "epoch": 0.5535324989907145,
      "grad_norm": 154.91705322265625,
      "learning_rate": 8.755800973006784e-06,
      "loss": 4.9573,
      "step": 13711
    },
    {
      "epoch": 0.5535728704077513,
      "grad_norm": 142.28549194335938,
      "learning_rate": 8.75450356360443e-06,
      "loss": 5.67,
      "step": 13712
    },
    {
      "epoch": 0.5536132418247881,
      "grad_norm": 547.1212768554688,
      "learning_rate": 8.753206175497159e-06,
      "loss": 5.4999,
      "step": 13713
    },
    {
      "epoch": 0.5536536132418248,
      "grad_norm": 99.54264831542969,
      "learning_rate": 8.751908808707148e-06,
      "loss": 4.3309,
      "step": 13714
    },
    {
      "epoch": 0.5536939846588615,
      "grad_norm": 84.66633605957031,
      "learning_rate": 8.75061146325658e-06,
      "loss": 6.3992,
      "step": 13715
    },
    {
      "epoch": 0.5537343560758983,
      "grad_norm": 103.48200225830078,
      "learning_rate": 8.749314139167641e-06,
      "loss": 5.6631,
      "step": 13716
    },
    {
      "epoch": 0.553774727492935,
      "grad_norm": 97.50961303710938,
      "learning_rate": 8.74801683646251e-06,
      "loss": 6.1665,
      "step": 13717
    },
    {
      "epoch": 0.5538150989099717,
      "grad_norm": 86.2636947631836,
      "learning_rate": 8.746719555163367e-06,
      "loss": 5.6566,
      "step": 13718
    },
    {
      "epoch": 0.5538554703270084,
      "grad_norm": 92.2238998413086,
      "learning_rate": 8.745422295292392e-06,
      "loss": 7.8116,
      "step": 13719
    },
    {
      "epoch": 0.5538958417440453,
      "grad_norm": 87.41958618164062,
      "learning_rate": 8.744125056871766e-06,
      "loss": 5.6984,
      "step": 13720
    },
    {
      "epoch": 0.553936213161082,
      "grad_norm": 159.8052520751953,
      "learning_rate": 8.742827839923669e-06,
      "loss": 5.3531,
      "step": 13721
    },
    {
      "epoch": 0.5539765845781187,
      "grad_norm": 268.046142578125,
      "learning_rate": 8.741530644470284e-06,
      "loss": 5.0566,
      "step": 13722
    },
    {
      "epoch": 0.5540169559951554,
      "grad_norm": 91.48175811767578,
      "learning_rate": 8.740233470533783e-06,
      "loss": 4.3131,
      "step": 13723
    },
    {
      "epoch": 0.5540573274121922,
      "grad_norm": 76.05854797363281,
      "learning_rate": 8.738936318136348e-06,
      "loss": 4.0534,
      "step": 13724
    },
    {
      "epoch": 0.5540976988292289,
      "grad_norm": 155.78829956054688,
      "learning_rate": 8.737639187300156e-06,
      "loss": 4.6887,
      "step": 13725
    },
    {
      "epoch": 0.5541380702462656,
      "grad_norm": 69.12923431396484,
      "learning_rate": 8.736342078047389e-06,
      "loss": 6.1138,
      "step": 13726
    },
    {
      "epoch": 0.5541784416633024,
      "grad_norm": 84.09712219238281,
      "learning_rate": 8.735044990400218e-06,
      "loss": 5.2966,
      "step": 13727
    },
    {
      "epoch": 0.5542188130803392,
      "grad_norm": 105.3494873046875,
      "learning_rate": 8.733747924380826e-06,
      "loss": 5.4382,
      "step": 13728
    },
    {
      "epoch": 0.5542591844973759,
      "grad_norm": 71.1461410522461,
      "learning_rate": 8.732450880011387e-06,
      "loss": 4.7101,
      "step": 13729
    },
    {
      "epoch": 0.5542995559144126,
      "grad_norm": 175.15359497070312,
      "learning_rate": 8.731153857314083e-06,
      "loss": 6.2838,
      "step": 13730
    },
    {
      "epoch": 0.5543399273314493,
      "grad_norm": 89.47637939453125,
      "learning_rate": 8.729856856311079e-06,
      "loss": 5.6224,
      "step": 13731
    },
    {
      "epoch": 0.554380298748486,
      "grad_norm": 104.72763061523438,
      "learning_rate": 8.728559877024557e-06,
      "loss": 6.2881,
      "step": 13732
    },
    {
      "epoch": 0.5544206701655228,
      "grad_norm": 68.50404357910156,
      "learning_rate": 8.727262919476692e-06,
      "loss": 4.6234,
      "step": 13733
    },
    {
      "epoch": 0.5544610415825596,
      "grad_norm": 136.08445739746094,
      "learning_rate": 8.725965983689665e-06,
      "loss": 5.1986,
      "step": 13734
    },
    {
      "epoch": 0.5545014129995963,
      "grad_norm": 90.5068130493164,
      "learning_rate": 8.724669069685638e-06,
      "loss": 5.7079,
      "step": 13735
    },
    {
      "epoch": 0.554541784416633,
      "grad_norm": 86.6834945678711,
      "learning_rate": 8.723372177486793e-06,
      "loss": 4.6433,
      "step": 13736
    },
    {
      "epoch": 0.5545821558336698,
      "grad_norm": 172.66827392578125,
      "learning_rate": 8.722075307115303e-06,
      "loss": 5.7116,
      "step": 13737
    },
    {
      "epoch": 0.5546225272507065,
      "grad_norm": 92.43399047851562,
      "learning_rate": 8.720778458593342e-06,
      "loss": 5.4375,
      "step": 13738
    },
    {
      "epoch": 0.5546628986677432,
      "grad_norm": 94.27052307128906,
      "learning_rate": 8.719481631943082e-06,
      "loss": 3.7454,
      "step": 13739
    },
    {
      "epoch": 0.5547032700847799,
      "grad_norm": 101.9205551147461,
      "learning_rate": 8.718184827186694e-06,
      "loss": 5.1911,
      "step": 13740
    },
    {
      "epoch": 0.5547436415018168,
      "grad_norm": 96.92838287353516,
      "learning_rate": 8.716888044346354e-06,
      "loss": 3.7883,
      "step": 13741
    },
    {
      "epoch": 0.5547840129188535,
      "grad_norm": 77.12823486328125,
      "learning_rate": 8.715591283444235e-06,
      "loss": 4.6724,
      "step": 13742
    },
    {
      "epoch": 0.5548243843358902,
      "grad_norm": 248.6671600341797,
      "learning_rate": 8.714294544502502e-06,
      "loss": 5.1125,
      "step": 13743
    },
    {
      "epoch": 0.5548647557529269,
      "grad_norm": 69.9871826171875,
      "learning_rate": 8.712997827543329e-06,
      "loss": 4.4972,
      "step": 13744
    },
    {
      "epoch": 0.5549051271699637,
      "grad_norm": 119.71751403808594,
      "learning_rate": 8.711701132588889e-06,
      "loss": 6.4227,
      "step": 13745
    },
    {
      "epoch": 0.5549454985870004,
      "grad_norm": 92.78300476074219,
      "learning_rate": 8.710404459661353e-06,
      "loss": 4.7295,
      "step": 13746
    },
    {
      "epoch": 0.5549858700040371,
      "grad_norm": 90.49504852294922,
      "learning_rate": 8.709107808782886e-06,
      "loss": 5.6724,
      "step": 13747
    },
    {
      "epoch": 0.5550262414210739,
      "grad_norm": 83.42562866210938,
      "learning_rate": 8.707811179975663e-06,
      "loss": 5.6852,
      "step": 13748
    },
    {
      "epoch": 0.5550666128381107,
      "grad_norm": 131.45274353027344,
      "learning_rate": 8.70651457326185e-06,
      "loss": 5.8125,
      "step": 13749
    },
    {
      "epoch": 0.5551069842551474,
      "grad_norm": 96.90966033935547,
      "learning_rate": 8.705217988663623e-06,
      "loss": 4.9316,
      "step": 13750
    },
    {
      "epoch": 0.5551473556721841,
      "grad_norm": 116.29512786865234,
      "learning_rate": 8.703921426203139e-06,
      "loss": 6.8101,
      "step": 13751
    },
    {
      "epoch": 0.5551877270892208,
      "grad_norm": 131.20896911621094,
      "learning_rate": 8.702624885902574e-06,
      "loss": 4.9609,
      "step": 13752
    },
    {
      "epoch": 0.5552280985062575,
      "grad_norm": 78.18372344970703,
      "learning_rate": 8.701328367784092e-06,
      "loss": 4.9104,
      "step": 13753
    },
    {
      "epoch": 0.5552684699232943,
      "grad_norm": 78.93633270263672,
      "learning_rate": 8.700031871869864e-06,
      "loss": 6.0055,
      "step": 13754
    },
    {
      "epoch": 0.5553088413403311,
      "grad_norm": 92.93480682373047,
      "learning_rate": 8.698735398182055e-06,
      "loss": 4.1482,
      "step": 13755
    },
    {
      "epoch": 0.5553492127573678,
      "grad_norm": 65.17282104492188,
      "learning_rate": 8.69743894674283e-06,
      "loss": 4.6909,
      "step": 13756
    },
    {
      "epoch": 0.5553895841744045,
      "grad_norm": 94.09474182128906,
      "learning_rate": 8.696142517574359e-06,
      "loss": 3.7349,
      "step": 13757
    },
    {
      "epoch": 0.5554299555914413,
      "grad_norm": 88.61360168457031,
      "learning_rate": 8.69484611069881e-06,
      "loss": 6.3029,
      "step": 13758
    },
    {
      "epoch": 0.555470327008478,
      "grad_norm": 93.57957458496094,
      "learning_rate": 8.69354972613834e-06,
      "loss": 6.1222,
      "step": 13759
    },
    {
      "epoch": 0.5555106984255147,
      "grad_norm": 105.09883117675781,
      "learning_rate": 8.69225336391512e-06,
      "loss": 4.7701,
      "step": 13760
    },
    {
      "epoch": 0.5555510698425514,
      "grad_norm": 216.7399444580078,
      "learning_rate": 8.690957024051314e-06,
      "loss": 5.1724,
      "step": 13761
    },
    {
      "epoch": 0.5555914412595883,
      "grad_norm": 144.3200225830078,
      "learning_rate": 8.689660706569089e-06,
      "loss": 5.5554,
      "step": 13762
    },
    {
      "epoch": 0.555631812676625,
      "grad_norm": 129.0269775390625,
      "learning_rate": 8.688364411490603e-06,
      "loss": 5.4589,
      "step": 13763
    },
    {
      "epoch": 0.5556721840936617,
      "grad_norm": 108.37202453613281,
      "learning_rate": 8.68706813883802e-06,
      "loss": 4.7158,
      "step": 13764
    },
    {
      "epoch": 0.5557125555106984,
      "grad_norm": 71.01598358154297,
      "learning_rate": 8.685771888633508e-06,
      "loss": 4.9994,
      "step": 13765
    },
    {
      "epoch": 0.5557529269277351,
      "grad_norm": 123.59508514404297,
      "learning_rate": 8.68447566089923e-06,
      "loss": 5.1337,
      "step": 13766
    },
    {
      "epoch": 0.5557932983447719,
      "grad_norm": 113.45577239990234,
      "learning_rate": 8.683179455657343e-06,
      "loss": 2.9049,
      "step": 13767
    },
    {
      "epoch": 0.5558336697618086,
      "grad_norm": 89.6227798461914,
      "learning_rate": 8.681883272930014e-06,
      "loss": 5.3327,
      "step": 13768
    },
    {
      "epoch": 0.5558740411788454,
      "grad_norm": 110.7804183959961,
      "learning_rate": 8.680587112739402e-06,
      "loss": 5.6714,
      "step": 13769
    },
    {
      "epoch": 0.5559144125958821,
      "grad_norm": 107.28462219238281,
      "learning_rate": 8.679290975107672e-06,
      "loss": 5.4838,
      "step": 13770
    },
    {
      "epoch": 0.5559547840129189,
      "grad_norm": 130.3448944091797,
      "learning_rate": 8.677994860056981e-06,
      "loss": 5.1197,
      "step": 13771
    },
    {
      "epoch": 0.5559951554299556,
      "grad_norm": 96.002685546875,
      "learning_rate": 8.676698767609489e-06,
      "loss": 3.8639,
      "step": 13772
    },
    {
      "epoch": 0.5560355268469923,
      "grad_norm": 100.11735534667969,
      "learning_rate": 8.675402697787359e-06,
      "loss": 5.2247,
      "step": 13773
    },
    {
      "epoch": 0.556075898264029,
      "grad_norm": 95.81315612792969,
      "learning_rate": 8.67410665061275e-06,
      "loss": 5.5608,
      "step": 13774
    },
    {
      "epoch": 0.5561162696810658,
      "grad_norm": 135.3187255859375,
      "learning_rate": 8.67281062610782e-06,
      "loss": 5.3345,
      "step": 13775
    },
    {
      "epoch": 0.5561566410981026,
      "grad_norm": 69.30626678466797,
      "learning_rate": 8.67151462429473e-06,
      "loss": 6.4191,
      "step": 13776
    },
    {
      "epoch": 0.5561970125151393,
      "grad_norm": 72.87730407714844,
      "learning_rate": 8.670218645195636e-06,
      "loss": 5.3251,
      "step": 13777
    },
    {
      "epoch": 0.556237383932176,
      "grad_norm": 92.02873992919922,
      "learning_rate": 8.668922688832703e-06,
      "loss": 4.3553,
      "step": 13778
    },
    {
      "epoch": 0.5562777553492128,
      "grad_norm": 76.20995330810547,
      "learning_rate": 8.66762675522808e-06,
      "loss": 3.8099,
      "step": 13779
    },
    {
      "epoch": 0.5563181267662495,
      "grad_norm": 131.44654846191406,
      "learning_rate": 8.666330844403929e-06,
      "loss": 4.9815,
      "step": 13780
    },
    {
      "epoch": 0.5563584981832862,
      "grad_norm": 83.11277770996094,
      "learning_rate": 8.665034956382404e-06,
      "loss": 3.9293,
      "step": 13781
    },
    {
      "epoch": 0.5563988696003229,
      "grad_norm": 71.3419189453125,
      "learning_rate": 8.663739091185668e-06,
      "loss": 5.8436,
      "step": 13782
    },
    {
      "epoch": 0.5564392410173598,
      "grad_norm": 149.61305236816406,
      "learning_rate": 8.66244324883587e-06,
      "loss": 5.1194,
      "step": 13783
    },
    {
      "epoch": 0.5564796124343965,
      "grad_norm": 102.62308502197266,
      "learning_rate": 8.66114742935517e-06,
      "loss": 5.5556,
      "step": 13784
    },
    {
      "epoch": 0.5565199838514332,
      "grad_norm": 133.9726104736328,
      "learning_rate": 8.659851632765722e-06,
      "loss": 4.3015,
      "step": 13785
    },
    {
      "epoch": 0.5565603552684699,
      "grad_norm": 99.4331283569336,
      "learning_rate": 8.658555859089683e-06,
      "loss": 6.1319,
      "step": 13786
    },
    {
      "epoch": 0.5566007266855066,
      "grad_norm": 90.4173812866211,
      "learning_rate": 8.657260108349204e-06,
      "loss": 3.9466,
      "step": 13787
    },
    {
      "epoch": 0.5566410981025434,
      "grad_norm": 74.87902069091797,
      "learning_rate": 8.655964380566441e-06,
      "loss": 5.8187,
      "step": 13788
    },
    {
      "epoch": 0.5566814695195801,
      "grad_norm": 218.906982421875,
      "learning_rate": 8.65466867576355e-06,
      "loss": 4.803,
      "step": 13789
    },
    {
      "epoch": 0.5567218409366169,
      "grad_norm": 82.75083923339844,
      "learning_rate": 8.653372993962688e-06,
      "loss": 4.9269,
      "step": 13790
    },
    {
      "epoch": 0.5567622123536536,
      "grad_norm": 121.48147583007812,
      "learning_rate": 8.652077335185998e-06,
      "loss": 4.4001,
      "step": 13791
    },
    {
      "epoch": 0.5568025837706904,
      "grad_norm": 75.85684967041016,
      "learning_rate": 8.650781699455637e-06,
      "loss": 4.5301,
      "step": 13792
    },
    {
      "epoch": 0.5568429551877271,
      "grad_norm": 97.29989624023438,
      "learning_rate": 8.64948608679376e-06,
      "loss": 3.5881,
      "step": 13793
    },
    {
      "epoch": 0.5568833266047638,
      "grad_norm": 176.7425079345703,
      "learning_rate": 8.648190497222518e-06,
      "loss": 4.7792,
      "step": 13794
    },
    {
      "epoch": 0.5569236980218005,
      "grad_norm": 114.54711151123047,
      "learning_rate": 8.64689493076406e-06,
      "loss": 4.8541,
      "step": 13795
    },
    {
      "epoch": 0.5569640694388373,
      "grad_norm": 69.87495422363281,
      "learning_rate": 8.64559938744054e-06,
      "loss": 3.7781,
      "step": 13796
    },
    {
      "epoch": 0.5570044408558741,
      "grad_norm": 169.5013885498047,
      "learning_rate": 8.644303867274107e-06,
      "loss": 3.4901,
      "step": 13797
    },
    {
      "epoch": 0.5570448122729108,
      "grad_norm": 101.07544708251953,
      "learning_rate": 8.643008370286914e-06,
      "loss": 5.0653,
      "step": 13798
    },
    {
      "epoch": 0.5570851836899475,
      "grad_norm": 98.48944854736328,
      "learning_rate": 8.641712896501108e-06,
      "loss": 5.7311,
      "step": 13799
    },
    {
      "epoch": 0.5571255551069842,
      "grad_norm": 74.71195220947266,
      "learning_rate": 8.64041744593884e-06,
      "loss": 5.3718,
      "step": 13800
    },
    {
      "epoch": 0.557165926524021,
      "grad_norm": 96.13833618164062,
      "learning_rate": 8.639122018622257e-06,
      "loss": 5.53,
      "step": 13801
    },
    {
      "epoch": 0.5572062979410577,
      "grad_norm": 71.92060089111328,
      "learning_rate": 8.637826614573513e-06,
      "loss": 4.5219,
      "step": 13802
    },
    {
      "epoch": 0.5572466693580944,
      "grad_norm": 103.51309204101562,
      "learning_rate": 8.63653123381475e-06,
      "loss": 5.8495,
      "step": 13803
    },
    {
      "epoch": 0.5572870407751312,
      "grad_norm": 141.41314697265625,
      "learning_rate": 8.63523587636812e-06,
      "loss": 4.5059,
      "step": 13804
    },
    {
      "epoch": 0.557327412192168,
      "grad_norm": 138.9296417236328,
      "learning_rate": 8.633940542255768e-06,
      "loss": 5.894,
      "step": 13805
    },
    {
      "epoch": 0.5573677836092047,
      "grad_norm": 152.18870544433594,
      "learning_rate": 8.632645231499845e-06,
      "loss": 3.8865,
      "step": 13806
    },
    {
      "epoch": 0.5574081550262414,
      "grad_norm": 70.30514526367188,
      "learning_rate": 8.631349944122495e-06,
      "loss": 3.9314,
      "step": 13807
    },
    {
      "epoch": 0.5574485264432781,
      "grad_norm": 81.8716049194336,
      "learning_rate": 8.630054680145862e-06,
      "loss": 6.6554,
      "step": 13808
    },
    {
      "epoch": 0.5574888978603149,
      "grad_norm": 145.90357971191406,
      "learning_rate": 8.628759439592097e-06,
      "loss": 5.5707,
      "step": 13809
    },
    {
      "epoch": 0.5575292692773516,
      "grad_norm": 105.5804214477539,
      "learning_rate": 8.627464222483347e-06,
      "loss": 5.5275,
      "step": 13810
    },
    {
      "epoch": 0.5575696406943884,
      "grad_norm": 99.5003662109375,
      "learning_rate": 8.62616902884175e-06,
      "loss": 4.5049,
      "step": 13811
    },
    {
      "epoch": 0.5576100121114251,
      "grad_norm": 74.47433471679688,
      "learning_rate": 8.624873858689456e-06,
      "loss": 6.0853,
      "step": 13812
    },
    {
      "epoch": 0.5576503835284619,
      "grad_norm": 106.27128601074219,
      "learning_rate": 8.623578712048606e-06,
      "loss": 5.3601,
      "step": 13813
    },
    {
      "epoch": 0.5576907549454986,
      "grad_norm": 107.69236755371094,
      "learning_rate": 8.622283588941348e-06,
      "loss": 5.5556,
      "step": 13814
    },
    {
      "epoch": 0.5577311263625353,
      "grad_norm": 131.35751342773438,
      "learning_rate": 8.620988489389821e-06,
      "loss": 4.5327,
      "step": 13815
    },
    {
      "epoch": 0.557771497779572,
      "grad_norm": 89.7822265625,
      "learning_rate": 8.61969341341617e-06,
      "loss": 5.3991,
      "step": 13816
    },
    {
      "epoch": 0.5578118691966089,
      "grad_norm": 88.39839935302734,
      "learning_rate": 8.618398361042542e-06,
      "loss": 4.1851,
      "step": 13817
    },
    {
      "epoch": 0.5578522406136456,
      "grad_norm": 81.54525756835938,
      "learning_rate": 8.617103332291077e-06,
      "loss": 5.5605,
      "step": 13818
    },
    {
      "epoch": 0.5578926120306823,
      "grad_norm": 114.58744049072266,
      "learning_rate": 8.615808327183913e-06,
      "loss": 5.3569,
      "step": 13819
    },
    {
      "epoch": 0.557932983447719,
      "grad_norm": 133.69093322753906,
      "learning_rate": 8.614513345743192e-06,
      "loss": 4.0787,
      "step": 13820
    },
    {
      "epoch": 0.5579733548647557,
      "grad_norm": 136.3101348876953,
      "learning_rate": 8.613218387991058e-06,
      "loss": 5.6408,
      "step": 13821
    },
    {
      "epoch": 0.5580137262817925,
      "grad_norm": 83.14727783203125,
      "learning_rate": 8.611923453949657e-06,
      "loss": 3.7066,
      "step": 13822
    },
    {
      "epoch": 0.5580540976988292,
      "grad_norm": 73.9988021850586,
      "learning_rate": 8.61062854364112e-06,
      "loss": 5.0758,
      "step": 13823
    },
    {
      "epoch": 0.558094469115866,
      "grad_norm": 76.88267517089844,
      "learning_rate": 8.609333657087587e-06,
      "loss": 5.2056,
      "step": 13824
    },
    {
      "epoch": 0.5581348405329027,
      "grad_norm": 71.78709411621094,
      "learning_rate": 8.608038794311203e-06,
      "loss": 5.9223,
      "step": 13825
    },
    {
      "epoch": 0.5581752119499395,
      "grad_norm": 90.70915985107422,
      "learning_rate": 8.606743955334107e-06,
      "loss": 3.7965,
      "step": 13826
    },
    {
      "epoch": 0.5582155833669762,
      "grad_norm": 132.81654357910156,
      "learning_rate": 8.605449140178435e-06,
      "loss": 5.1527,
      "step": 13827
    },
    {
      "epoch": 0.5582559547840129,
      "grad_norm": 111.2055892944336,
      "learning_rate": 8.604154348866326e-06,
      "loss": 5.7408,
      "step": 13828
    },
    {
      "epoch": 0.5582963262010496,
      "grad_norm": 70.20774841308594,
      "learning_rate": 8.602859581419915e-06,
      "loss": 4.7708,
      "step": 13829
    },
    {
      "epoch": 0.5583366976180864,
      "grad_norm": 85.21271514892578,
      "learning_rate": 8.60156483786135e-06,
      "loss": 5.9063,
      "step": 13830
    },
    {
      "epoch": 0.5583770690351232,
      "grad_norm": 92.17013549804688,
      "learning_rate": 8.600270118212756e-06,
      "loss": 6.3566,
      "step": 13831
    },
    {
      "epoch": 0.5584174404521599,
      "grad_norm": 115.91944122314453,
      "learning_rate": 8.598975422496273e-06,
      "loss": 5.2048,
      "step": 13832
    },
    {
      "epoch": 0.5584578118691966,
      "grad_norm": 194.8197784423828,
      "learning_rate": 8.59768075073404e-06,
      "loss": 4.7934,
      "step": 13833
    },
    {
      "epoch": 0.5584981832862334,
      "grad_norm": 79.95431518554688,
      "learning_rate": 8.596386102948193e-06,
      "loss": 3.4821,
      "step": 13834
    },
    {
      "epoch": 0.5585385547032701,
      "grad_norm": 80.58930206298828,
      "learning_rate": 8.595091479160864e-06,
      "loss": 5.4368,
      "step": 13835
    },
    {
      "epoch": 0.5585789261203068,
      "grad_norm": 92.16148376464844,
      "learning_rate": 8.593796879394191e-06,
      "loss": 6.2675,
      "step": 13836
    },
    {
      "epoch": 0.5586192975373435,
      "grad_norm": 78.92163848876953,
      "learning_rate": 8.592502303670305e-06,
      "loss": 4.8894,
      "step": 13837
    },
    {
      "epoch": 0.5586596689543804,
      "grad_norm": 112.18099212646484,
      "learning_rate": 8.59120775201135e-06,
      "loss": 5.097,
      "step": 13838
    },
    {
      "epoch": 0.5587000403714171,
      "grad_norm": 407.814208984375,
      "learning_rate": 8.589913224439448e-06,
      "loss": 5.1397,
      "step": 13839
    },
    {
      "epoch": 0.5587404117884538,
      "grad_norm": 90.5469970703125,
      "learning_rate": 8.588618720976735e-06,
      "loss": 6.2158,
      "step": 13840
    },
    {
      "epoch": 0.5587807832054905,
      "grad_norm": 180.37408447265625,
      "learning_rate": 8.587324241645348e-06,
      "loss": 5.3225,
      "step": 13841
    },
    {
      "epoch": 0.5588211546225272,
      "grad_norm": 111.31279754638672,
      "learning_rate": 8.58602978646742e-06,
      "loss": 4.9148,
      "step": 13842
    },
    {
      "epoch": 0.558861526039564,
      "grad_norm": 87.66847229003906,
      "learning_rate": 8.58473535546508e-06,
      "loss": 5.669,
      "step": 13843
    },
    {
      "epoch": 0.5589018974566007,
      "grad_norm": 221.4401092529297,
      "learning_rate": 8.583440948660458e-06,
      "loss": 4.7667,
      "step": 13844
    },
    {
      "epoch": 0.5589422688736375,
      "grad_norm": 95.9344711303711,
      "learning_rate": 8.58214656607569e-06,
      "loss": 5.1177,
      "step": 13845
    },
    {
      "epoch": 0.5589826402906742,
      "grad_norm": 103.0114517211914,
      "learning_rate": 8.580852207732904e-06,
      "loss": 5.9945,
      "step": 13846
    },
    {
      "epoch": 0.559023011707711,
      "grad_norm": 86.06121063232422,
      "learning_rate": 8.57955787365423e-06,
      "loss": 6.1299,
      "step": 13847
    },
    {
      "epoch": 0.5590633831247477,
      "grad_norm": 117.23521423339844,
      "learning_rate": 8.578263563861799e-06,
      "loss": 5.0514,
      "step": 13848
    },
    {
      "epoch": 0.5591037545417844,
      "grad_norm": 83.35757446289062,
      "learning_rate": 8.576969278377742e-06,
      "loss": 6.2474,
      "step": 13849
    },
    {
      "epoch": 0.5591441259588211,
      "grad_norm": 64.83439636230469,
      "learning_rate": 8.575675017224191e-06,
      "loss": 3.1373,
      "step": 13850
    },
    {
      "epoch": 0.5591844973758578,
      "grad_norm": 99.588134765625,
      "learning_rate": 8.574380780423268e-06,
      "loss": 4.9962,
      "step": 13851
    },
    {
      "epoch": 0.5592248687928947,
      "grad_norm": 78.94426727294922,
      "learning_rate": 8.573086567997104e-06,
      "loss": 5.2418,
      "step": 13852
    },
    {
      "epoch": 0.5592652402099314,
      "grad_norm": 101.53722381591797,
      "learning_rate": 8.571792379967828e-06,
      "loss": 6.4392,
      "step": 13853
    },
    {
      "epoch": 0.5593056116269681,
      "grad_norm": 84.12858581542969,
      "learning_rate": 8.570498216357568e-06,
      "loss": 5.2924,
      "step": 13854
    },
    {
      "epoch": 0.5593459830440048,
      "grad_norm": 90.85231018066406,
      "learning_rate": 8.569204077188449e-06,
      "loss": 4.6167,
      "step": 13855
    },
    {
      "epoch": 0.5593863544610416,
      "grad_norm": 69.24852752685547,
      "learning_rate": 8.567909962482599e-06,
      "loss": 5.0589,
      "step": 13856
    },
    {
      "epoch": 0.5594267258780783,
      "grad_norm": 81.52294921875,
      "learning_rate": 8.566615872262145e-06,
      "loss": 5.2149,
      "step": 13857
    },
    {
      "epoch": 0.559467097295115,
      "grad_norm": 92.02674865722656,
      "learning_rate": 8.565321806549216e-06,
      "loss": 4.618,
      "step": 13858
    },
    {
      "epoch": 0.5595074687121518,
      "grad_norm": 74.66266632080078,
      "learning_rate": 8.56402776536593e-06,
      "loss": 5.7008,
      "step": 13859
    },
    {
      "epoch": 0.5595478401291886,
      "grad_norm": 110.62931060791016,
      "learning_rate": 8.562733748734415e-06,
      "loss": 5.1288,
      "step": 13860
    },
    {
      "epoch": 0.5595882115462253,
      "grad_norm": 76.7750244140625,
      "learning_rate": 8.561439756676798e-06,
      "loss": 4.2658,
      "step": 13861
    },
    {
      "epoch": 0.559628582963262,
      "grad_norm": 109.45601654052734,
      "learning_rate": 8.560145789215204e-06,
      "loss": 4.5534,
      "step": 13862
    },
    {
      "epoch": 0.5596689543802987,
      "grad_norm": 81.14635467529297,
      "learning_rate": 8.558851846371751e-06,
      "loss": 5.5988,
      "step": 13863
    },
    {
      "epoch": 0.5597093257973355,
      "grad_norm": 87.67066955566406,
      "learning_rate": 8.557557928168568e-06,
      "loss": 5.0853,
      "step": 13864
    },
    {
      "epoch": 0.5597496972143722,
      "grad_norm": 123.76922607421875,
      "learning_rate": 8.556264034627775e-06,
      "loss": 6.8453,
      "step": 13865
    },
    {
      "epoch": 0.559790068631409,
      "grad_norm": 72.75666809082031,
      "learning_rate": 8.554970165771502e-06,
      "loss": 4.3306,
      "step": 13866
    },
    {
      "epoch": 0.5598304400484457,
      "grad_norm": 211.05262756347656,
      "learning_rate": 8.553676321621858e-06,
      "loss": 2.275,
      "step": 13867
    },
    {
      "epoch": 0.5598708114654825,
      "grad_norm": 82.95748138427734,
      "learning_rate": 8.552382502200973e-06,
      "loss": 5.3398,
      "step": 13868
    },
    {
      "epoch": 0.5599111828825192,
      "grad_norm": 106.99583435058594,
      "learning_rate": 8.551088707530965e-06,
      "loss": 5.424,
      "step": 13869
    },
    {
      "epoch": 0.5599515542995559,
      "grad_norm": 68.6309585571289,
      "learning_rate": 8.549794937633962e-06,
      "loss": 4.6694,
      "step": 13870
    },
    {
      "epoch": 0.5599919257165926,
      "grad_norm": 100.41119384765625,
      "learning_rate": 8.548501192532075e-06,
      "loss": 4.7476,
      "step": 13871
    },
    {
      "epoch": 0.5600322971336293,
      "grad_norm": 100.69692993164062,
      "learning_rate": 8.547207472247427e-06,
      "loss": 4.8621,
      "step": 13872
    },
    {
      "epoch": 0.5600726685506662,
      "grad_norm": 110.65544128417969,
      "learning_rate": 8.54591377680214e-06,
      "loss": 3.9463,
      "step": 13873
    },
    {
      "epoch": 0.5601130399677029,
      "grad_norm": 130.319580078125,
      "learning_rate": 8.544620106218334e-06,
      "loss": 6.1869,
      "step": 13874
    },
    {
      "epoch": 0.5601534113847396,
      "grad_norm": 82.79374694824219,
      "learning_rate": 8.543326460518123e-06,
      "loss": 5.5955,
      "step": 13875
    },
    {
      "epoch": 0.5601937828017763,
      "grad_norm": 97.58365631103516,
      "learning_rate": 8.542032839723629e-06,
      "loss": 5.8109,
      "step": 13876
    },
    {
      "epoch": 0.5602341542188131,
      "grad_norm": 96.70162200927734,
      "learning_rate": 8.540739243856966e-06,
      "loss": 4.5573,
      "step": 13877
    },
    {
      "epoch": 0.5602745256358498,
      "grad_norm": 86.87678527832031,
      "learning_rate": 8.539445672940261e-06,
      "loss": 5.5828,
      "step": 13878
    },
    {
      "epoch": 0.5603148970528865,
      "grad_norm": 955.2608642578125,
      "learning_rate": 8.53815212699562e-06,
      "loss": 5.4212,
      "step": 13879
    },
    {
      "epoch": 0.5603552684699233,
      "grad_norm": 91.28193664550781,
      "learning_rate": 8.536858606045163e-06,
      "loss": 5.3362,
      "step": 13880
    },
    {
      "epoch": 0.5603956398869601,
      "grad_norm": 74.52288818359375,
      "learning_rate": 8.535565110111008e-06,
      "loss": 4.1436,
      "step": 13881
    },
    {
      "epoch": 0.5604360113039968,
      "grad_norm": 190.2291717529297,
      "learning_rate": 8.534271639215272e-06,
      "loss": 4.4256,
      "step": 13882
    },
    {
      "epoch": 0.5604763827210335,
      "grad_norm": 74.37435913085938,
      "learning_rate": 8.532978193380066e-06,
      "loss": 4.8595,
      "step": 13883
    },
    {
      "epoch": 0.5605167541380702,
      "grad_norm": 96.76522827148438,
      "learning_rate": 8.531684772627506e-06,
      "loss": 4.3522,
      "step": 13884
    },
    {
      "epoch": 0.560557125555107,
      "grad_norm": 66.94566345214844,
      "learning_rate": 8.53039137697971e-06,
      "loss": 3.9136,
      "step": 13885
    },
    {
      "epoch": 0.5605974969721437,
      "grad_norm": 95.34540557861328,
      "learning_rate": 8.529098006458791e-06,
      "loss": 4.6948,
      "step": 13886
    },
    {
      "epoch": 0.5606378683891805,
      "grad_norm": 177.0673828125,
      "learning_rate": 8.52780466108686e-06,
      "loss": 5.1204,
      "step": 13887
    },
    {
      "epoch": 0.5606782398062172,
      "grad_norm": 90.62190246582031,
      "learning_rate": 8.52651134088603e-06,
      "loss": 4.0931,
      "step": 13888
    },
    {
      "epoch": 0.560718611223254,
      "grad_norm": 98.80091094970703,
      "learning_rate": 8.525218045878413e-06,
      "loss": 4.8137,
      "step": 13889
    },
    {
      "epoch": 0.5607589826402907,
      "grad_norm": 73.64117431640625,
      "learning_rate": 8.523924776086132e-06,
      "loss": 5.3632,
      "step": 13890
    },
    {
      "epoch": 0.5607993540573274,
      "grad_norm": 72.62032318115234,
      "learning_rate": 8.522631531531284e-06,
      "loss": 5.6177,
      "step": 13891
    },
    {
      "epoch": 0.5608397254743641,
      "grad_norm": 90.06629943847656,
      "learning_rate": 8.521338312235987e-06,
      "loss": 5.805,
      "step": 13892
    },
    {
      "epoch": 0.5608800968914008,
      "grad_norm": 111.65652465820312,
      "learning_rate": 8.520045118222351e-06,
      "loss": 6.1359,
      "step": 13893
    },
    {
      "epoch": 0.5609204683084377,
      "grad_norm": 72.94615936279297,
      "learning_rate": 8.518751949512492e-06,
      "loss": 5.3711,
      "step": 13894
    },
    {
      "epoch": 0.5609608397254744,
      "grad_norm": 125.81183624267578,
      "learning_rate": 8.517458806128513e-06,
      "loss": 3.8688,
      "step": 13895
    },
    {
      "epoch": 0.5610012111425111,
      "grad_norm": 71.6593246459961,
      "learning_rate": 8.516165688092526e-06,
      "loss": 4.5391,
      "step": 13896
    },
    {
      "epoch": 0.5610415825595478,
      "grad_norm": 274.9361267089844,
      "learning_rate": 8.51487259542664e-06,
      "loss": 6.0313,
      "step": 13897
    },
    {
      "epoch": 0.5610819539765846,
      "grad_norm": 127.53782653808594,
      "learning_rate": 8.51357952815297e-06,
      "loss": 5.3872,
      "step": 13898
    },
    {
      "epoch": 0.5611223253936213,
      "grad_norm": 75.81255340576172,
      "learning_rate": 8.512286486293612e-06,
      "loss": 5.1211,
      "step": 13899
    },
    {
      "epoch": 0.561162696810658,
      "grad_norm": 102.2791748046875,
      "learning_rate": 8.510993469870685e-06,
      "loss": 6.463,
      "step": 13900
    },
    {
      "epoch": 0.5612030682276948,
      "grad_norm": 97.78120422363281,
      "learning_rate": 8.50970047890629e-06,
      "loss": 5.2727,
      "step": 13901
    },
    {
      "epoch": 0.5612434396447316,
      "grad_norm": 121.739013671875,
      "learning_rate": 8.50840751342254e-06,
      "loss": 4.4437,
      "step": 13902
    },
    {
      "epoch": 0.5612838110617683,
      "grad_norm": 72.73479461669922,
      "learning_rate": 8.507114573441534e-06,
      "loss": 5.7896,
      "step": 13903
    },
    {
      "epoch": 0.561324182478805,
      "grad_norm": 106.6372299194336,
      "learning_rate": 8.505821658985385e-06,
      "loss": 4.283,
      "step": 13904
    },
    {
      "epoch": 0.5613645538958417,
      "grad_norm": 89.92141723632812,
      "learning_rate": 8.504528770076197e-06,
      "loss": 4.4655,
      "step": 13905
    },
    {
      "epoch": 0.5614049253128784,
      "grad_norm": 98.87016296386719,
      "learning_rate": 8.503235906736077e-06,
      "loss": 5.2034,
      "step": 13906
    },
    {
      "epoch": 0.5614452967299152,
      "grad_norm": 129.8688201904297,
      "learning_rate": 8.501943068987127e-06,
      "loss": 6.8708,
      "step": 13907
    },
    {
      "epoch": 0.561485668146952,
      "grad_norm": 130.3511962890625,
      "learning_rate": 8.50065025685145e-06,
      "loss": 4.3215,
      "step": 13908
    },
    {
      "epoch": 0.5615260395639887,
      "grad_norm": 110.0829086303711,
      "learning_rate": 8.49935747035115e-06,
      "loss": 5.5907,
      "step": 13909
    },
    {
      "epoch": 0.5615664109810254,
      "grad_norm": 109.3205337524414,
      "learning_rate": 8.498064709508342e-06,
      "loss": 5.1016,
      "step": 13910
    },
    {
      "epoch": 0.5616067823980622,
      "grad_norm": 111.60856628417969,
      "learning_rate": 8.496771974345113e-06,
      "loss": 5.091,
      "step": 13911
    },
    {
      "epoch": 0.5616471538150989,
      "grad_norm": 98.91755676269531,
      "learning_rate": 8.495479264883573e-06,
      "loss": 4.1113,
      "step": 13912
    },
    {
      "epoch": 0.5616875252321356,
      "grad_norm": 68.81179809570312,
      "learning_rate": 8.494186581145825e-06,
      "loss": 6.2598,
      "step": 13913
    },
    {
      "epoch": 0.5617278966491723,
      "grad_norm": 91.39766693115234,
      "learning_rate": 8.492893923153972e-06,
      "loss": 4.6862,
      "step": 13914
    },
    {
      "epoch": 0.5617682680662092,
      "grad_norm": 110.35779571533203,
      "learning_rate": 8.491601290930111e-06,
      "loss": 5.5462,
      "step": 13915
    },
    {
      "epoch": 0.5618086394832459,
      "grad_norm": 86.22066497802734,
      "learning_rate": 8.490308684496345e-06,
      "loss": 4.4638,
      "step": 13916
    },
    {
      "epoch": 0.5618490109002826,
      "grad_norm": 64.02701568603516,
      "learning_rate": 8.489016103874776e-06,
      "loss": 4.1782,
      "step": 13917
    },
    {
      "epoch": 0.5618893823173193,
      "grad_norm": 123.28485107421875,
      "learning_rate": 8.487723549087509e-06,
      "loss": 6.4335,
      "step": 13918
    },
    {
      "epoch": 0.561929753734356,
      "grad_norm": 103.55935668945312,
      "learning_rate": 8.486431020156632e-06,
      "loss": 5.2228,
      "step": 13919
    },
    {
      "epoch": 0.5619701251513928,
      "grad_norm": 79.7396469116211,
      "learning_rate": 8.48513851710425e-06,
      "loss": 5.2656,
      "step": 13920
    },
    {
      "epoch": 0.5620104965684296,
      "grad_norm": 115.80003356933594,
      "learning_rate": 8.483846039952463e-06,
      "loss": 4.0769,
      "step": 13921
    },
    {
      "epoch": 0.5620508679854663,
      "grad_norm": 88.64791107177734,
      "learning_rate": 8.48255358872337e-06,
      "loss": 3.9009,
      "step": 13922
    },
    {
      "epoch": 0.562091239402503,
      "grad_norm": 81.58036804199219,
      "learning_rate": 8.481261163439064e-06,
      "loss": 6.2317,
      "step": 13923
    },
    {
      "epoch": 0.5621316108195398,
      "grad_norm": 116.47217559814453,
      "learning_rate": 8.479968764121647e-06,
      "loss": 4.869,
      "step": 13924
    },
    {
      "epoch": 0.5621719822365765,
      "grad_norm": 94.82894897460938,
      "learning_rate": 8.478676390793212e-06,
      "loss": 5.1014,
      "step": 13925
    },
    {
      "epoch": 0.5622123536536132,
      "grad_norm": 81.98815155029297,
      "learning_rate": 8.477384043475864e-06,
      "loss": 5.3088,
      "step": 13926
    },
    {
      "epoch": 0.5622527250706499,
      "grad_norm": 79.50923919677734,
      "learning_rate": 8.47609172219169e-06,
      "loss": 4.8315,
      "step": 13927
    },
    {
      "epoch": 0.5622930964876868,
      "grad_norm": 99.2765121459961,
      "learning_rate": 8.47479942696279e-06,
      "loss": 6.215,
      "step": 13928
    },
    {
      "epoch": 0.5623334679047235,
      "grad_norm": 81.70152282714844,
      "learning_rate": 8.473507157811254e-06,
      "loss": 5.3768,
      "step": 13929
    },
    {
      "epoch": 0.5623738393217602,
      "grad_norm": 332.59130859375,
      "learning_rate": 8.47221491475919e-06,
      "loss": 6.1381,
      "step": 13930
    },
    {
      "epoch": 0.5624142107387969,
      "grad_norm": 166.4923095703125,
      "learning_rate": 8.470922697828677e-06,
      "loss": 5.3483,
      "step": 13931
    },
    {
      "epoch": 0.5624545821558337,
      "grad_norm": 93.66645050048828,
      "learning_rate": 8.469630507041815e-06,
      "loss": 4.0258,
      "step": 13932
    },
    {
      "epoch": 0.5624949535728704,
      "grad_norm": 165.55401611328125,
      "learning_rate": 8.468338342420696e-06,
      "loss": 5.9311,
      "step": 13933
    },
    {
      "epoch": 0.5625353249899071,
      "grad_norm": 94.32537078857422,
      "learning_rate": 8.467046203987419e-06,
      "loss": 3.5332,
      "step": 13934
    },
    {
      "epoch": 0.5625756964069439,
      "grad_norm": 83.71742248535156,
      "learning_rate": 8.465754091764069e-06,
      "loss": 5.9462,
      "step": 13935
    },
    {
      "epoch": 0.5626160678239807,
      "grad_norm": 85.81157684326172,
      "learning_rate": 8.464462005772742e-06,
      "loss": 3.8585,
      "step": 13936
    },
    {
      "epoch": 0.5626564392410174,
      "grad_norm": 102.1852798461914,
      "learning_rate": 8.46316994603553e-06,
      "loss": 4.1589,
      "step": 13937
    },
    {
      "epoch": 0.5626968106580541,
      "grad_norm": 92.15320587158203,
      "learning_rate": 8.461877912574525e-06,
      "loss": 3.4182,
      "step": 13938
    },
    {
      "epoch": 0.5627371820750908,
      "grad_norm": 142.79022216796875,
      "learning_rate": 8.460585905411814e-06,
      "loss": 5.5966,
      "step": 13939
    },
    {
      "epoch": 0.5627775534921275,
      "grad_norm": 92.32018280029297,
      "learning_rate": 8.459293924569487e-06,
      "loss": 4.632,
      "step": 13940
    },
    {
      "epoch": 0.5628179249091643,
      "grad_norm": 88.6670150756836,
      "learning_rate": 8.458001970069637e-06,
      "loss": 4.6957,
      "step": 13941
    },
    {
      "epoch": 0.5628582963262011,
      "grad_norm": 112.29560089111328,
      "learning_rate": 8.456710041934355e-06,
      "loss": 4.1686,
      "step": 13942
    },
    {
      "epoch": 0.5628986677432378,
      "grad_norm": 87.86079406738281,
      "learning_rate": 8.455418140185724e-06,
      "loss": 5.5247,
      "step": 13943
    },
    {
      "epoch": 0.5629390391602745,
      "grad_norm": 81.05280303955078,
      "learning_rate": 8.454126264845837e-06,
      "loss": 5.4902,
      "step": 13944
    },
    {
      "epoch": 0.5629794105773113,
      "grad_norm": 180.4116668701172,
      "learning_rate": 8.452834415936782e-06,
      "loss": 4.1532,
      "step": 13945
    },
    {
      "epoch": 0.563019781994348,
      "grad_norm": 100.654052734375,
      "learning_rate": 8.45154259348065e-06,
      "loss": 5.8376,
      "step": 13946
    },
    {
      "epoch": 0.5630601534113847,
      "grad_norm": 106.02278137207031,
      "learning_rate": 8.450250797499517e-06,
      "loss": 4.8707,
      "step": 13947
    },
    {
      "epoch": 0.5631005248284214,
      "grad_norm": 95.0611343383789,
      "learning_rate": 8.448959028015479e-06,
      "loss": 3.6273,
      "step": 13948
    },
    {
      "epoch": 0.5631408962454583,
      "grad_norm": 111.75629425048828,
      "learning_rate": 8.447667285050617e-06,
      "loss": 4.6115,
      "step": 13949
    },
    {
      "epoch": 0.563181267662495,
      "grad_norm": 82.23469543457031,
      "learning_rate": 8.446375568627025e-06,
      "loss": 4.3029,
      "step": 13950
    },
    {
      "epoch": 0.5632216390795317,
      "grad_norm": 93.1567153930664,
      "learning_rate": 8.445083878766777e-06,
      "loss": 3.9924,
      "step": 13951
    },
    {
      "epoch": 0.5632620104965684,
      "grad_norm": 101.01070404052734,
      "learning_rate": 8.443792215491967e-06,
      "loss": 6.7497,
      "step": 13952
    },
    {
      "epoch": 0.5633023819136052,
      "grad_norm": 88.00316619873047,
      "learning_rate": 8.442500578824677e-06,
      "loss": 3.9441,
      "step": 13953
    },
    {
      "epoch": 0.5633427533306419,
      "grad_norm": 74.54371643066406,
      "learning_rate": 8.44120896878699e-06,
      "loss": 4.6921,
      "step": 13954
    },
    {
      "epoch": 0.5633831247476786,
      "grad_norm": 63.99746322631836,
      "learning_rate": 8.439917385400986e-06,
      "loss": 5.0161,
      "step": 13955
    },
    {
      "epoch": 0.5634234961647154,
      "grad_norm": 92.66605377197266,
      "learning_rate": 8.438625828688754e-06,
      "loss": 5.0413,
      "step": 13956
    },
    {
      "epoch": 0.5634638675817522,
      "grad_norm": 99.05591583251953,
      "learning_rate": 8.437334298672372e-06,
      "loss": 4.6902,
      "step": 13957
    },
    {
      "epoch": 0.5635042389987889,
      "grad_norm": 90.5005874633789,
      "learning_rate": 8.43604279537393e-06,
      "loss": 4.6789,
      "step": 13958
    },
    {
      "epoch": 0.5635446104158256,
      "grad_norm": 81.29025268554688,
      "learning_rate": 8.434751318815499e-06,
      "loss": 5.8316,
      "step": 13959
    },
    {
      "epoch": 0.5635849818328623,
      "grad_norm": 96.11933898925781,
      "learning_rate": 8.433459869019167e-06,
      "loss": 5.0087,
      "step": 13960
    },
    {
      "epoch": 0.563625353249899,
      "grad_norm": 142.49404907226562,
      "learning_rate": 8.432168446007012e-06,
      "loss": 5.6255,
      "step": 13961
    },
    {
      "epoch": 0.5636657246669358,
      "grad_norm": 92.62805938720703,
      "learning_rate": 8.430877049801117e-06,
      "loss": 3.6017,
      "step": 13962
    },
    {
      "epoch": 0.5637060960839726,
      "grad_norm": 55.251956939697266,
      "learning_rate": 8.429585680423559e-06,
      "loss": 5.4134,
      "step": 13963
    },
    {
      "epoch": 0.5637464675010093,
      "grad_norm": 135.85655212402344,
      "learning_rate": 8.428294337896419e-06,
      "loss": 5.4571,
      "step": 13964
    },
    {
      "epoch": 0.563786838918046,
      "grad_norm": 91.63912963867188,
      "learning_rate": 8.427003022241775e-06,
      "loss": 5.8651,
      "step": 13965
    },
    {
      "epoch": 0.5638272103350828,
      "grad_norm": 136.66143798828125,
      "learning_rate": 8.425711733481709e-06,
      "loss": 5.8942,
      "step": 13966
    },
    {
      "epoch": 0.5638675817521195,
      "grad_norm": 196.3928680419922,
      "learning_rate": 8.424420471638296e-06,
      "loss": 5.4302,
      "step": 13967
    },
    {
      "epoch": 0.5639079531691562,
      "grad_norm": 114.4255599975586,
      "learning_rate": 8.42312923673361e-06,
      "loss": 6.5125,
      "step": 13968
    },
    {
      "epoch": 0.5639483245861929,
      "grad_norm": 101.3481674194336,
      "learning_rate": 8.421838028789734e-06,
      "loss": 4.9633,
      "step": 13969
    },
    {
      "epoch": 0.5639886960032298,
      "grad_norm": 80.38864135742188,
      "learning_rate": 8.420546847828742e-06,
      "loss": 3.3024,
      "step": 13970
    },
    {
      "epoch": 0.5640290674202665,
      "grad_norm": 126.10040283203125,
      "learning_rate": 8.41925569387271e-06,
      "loss": 3.5448,
      "step": 13971
    },
    {
      "epoch": 0.5640694388373032,
      "grad_norm": 415.10845947265625,
      "learning_rate": 8.417964566943713e-06,
      "loss": 4.2752,
      "step": 13972
    },
    {
      "epoch": 0.5641098102543399,
      "grad_norm": 78.75373840332031,
      "learning_rate": 8.41667346706383e-06,
      "loss": 4.5209,
      "step": 13973
    },
    {
      "epoch": 0.5641501816713766,
      "grad_norm": 95.78952026367188,
      "learning_rate": 8.415382394255136e-06,
      "loss": 6.0697,
      "step": 13974
    },
    {
      "epoch": 0.5641905530884134,
      "grad_norm": 92.67620849609375,
      "learning_rate": 8.414091348539699e-06,
      "loss": 3.8983,
      "step": 13975
    },
    {
      "epoch": 0.5642309245054501,
      "grad_norm": 137.55218505859375,
      "learning_rate": 8.412800329939596e-06,
      "loss": 6.2248,
      "step": 13976
    },
    {
      "epoch": 0.5642712959224869,
      "grad_norm": 95.32948303222656,
      "learning_rate": 8.411509338476902e-06,
      "loss": 4.9664,
      "step": 13977
    },
    {
      "epoch": 0.5643116673395236,
      "grad_norm": 85.4988021850586,
      "learning_rate": 8.410218374173693e-06,
      "loss": 5.1169,
      "step": 13978
    },
    {
      "epoch": 0.5643520387565604,
      "grad_norm": 125.07636260986328,
      "learning_rate": 8.408927437052032e-06,
      "loss": 4.8481,
      "step": 13979
    },
    {
      "epoch": 0.5643924101735971,
      "grad_norm": 87.46563720703125,
      "learning_rate": 8.407636527133998e-06,
      "loss": 2.7108,
      "step": 13980
    },
    {
      "epoch": 0.5644327815906338,
      "grad_norm": 85.35566711425781,
      "learning_rate": 8.406345644441661e-06,
      "loss": 3.7372,
      "step": 13981
    },
    {
      "epoch": 0.5644731530076705,
      "grad_norm": 76.64375305175781,
      "learning_rate": 8.405054788997092e-06,
      "loss": 5.0557,
      "step": 13982
    },
    {
      "epoch": 0.5645135244247073,
      "grad_norm": 83.83501434326172,
      "learning_rate": 8.40376396082236e-06,
      "loss": 3.3258,
      "step": 13983
    },
    {
      "epoch": 0.5645538958417441,
      "grad_norm": 102.9427490234375,
      "learning_rate": 8.402473159939538e-06,
      "loss": 5.5251,
      "step": 13984
    },
    {
      "epoch": 0.5645942672587808,
      "grad_norm": 115.41874694824219,
      "learning_rate": 8.401182386370695e-06,
      "loss": 5.3118,
      "step": 13985
    },
    {
      "epoch": 0.5646346386758175,
      "grad_norm": 159.46237182617188,
      "learning_rate": 8.399891640137902e-06,
      "loss": 6.1505,
      "step": 13986
    },
    {
      "epoch": 0.5646750100928543,
      "grad_norm": 105.216552734375,
      "learning_rate": 8.398600921263223e-06,
      "loss": 4.9635,
      "step": 13987
    },
    {
      "epoch": 0.564715381509891,
      "grad_norm": 73.4241943359375,
      "learning_rate": 8.397310229768726e-06,
      "loss": 4.4977,
      "step": 13988
    },
    {
      "epoch": 0.5647557529269277,
      "grad_norm": 119.21419525146484,
      "learning_rate": 8.396019565676486e-06,
      "loss": 4.065,
      "step": 13989
    },
    {
      "epoch": 0.5647961243439644,
      "grad_norm": 117.83353424072266,
      "learning_rate": 8.394728929008565e-06,
      "loss": 4.9274,
      "step": 13990
    },
    {
      "epoch": 0.5648364957610013,
      "grad_norm": 149.31130981445312,
      "learning_rate": 8.393438319787029e-06,
      "loss": 3.9805,
      "step": 13991
    },
    {
      "epoch": 0.564876867178038,
      "grad_norm": 150.98158264160156,
      "learning_rate": 8.392147738033946e-06,
      "loss": 3.7499,
      "step": 13992
    },
    {
      "epoch": 0.5649172385950747,
      "grad_norm": 107.36272430419922,
      "learning_rate": 8.39085718377138e-06,
      "loss": 5.9369,
      "step": 13993
    },
    {
      "epoch": 0.5649576100121114,
      "grad_norm": 116.07357025146484,
      "learning_rate": 8.389566657021406e-06,
      "loss": 5.4494,
      "step": 13994
    },
    {
      "epoch": 0.5649979814291481,
      "grad_norm": 89.09111022949219,
      "learning_rate": 8.388276157806077e-06,
      "loss": 4.0465,
      "step": 13995
    },
    {
      "epoch": 0.5650383528461849,
      "grad_norm": 85.15125274658203,
      "learning_rate": 8.386985686147462e-06,
      "loss": 4.2173,
      "step": 13996
    },
    {
      "epoch": 0.5650787242632216,
      "grad_norm": 113.38165283203125,
      "learning_rate": 8.385695242067625e-06,
      "loss": 8.0646,
      "step": 13997
    },
    {
      "epoch": 0.5651190956802584,
      "grad_norm": 68.78941345214844,
      "learning_rate": 8.384404825588633e-06,
      "loss": 4.8765,
      "step": 13998
    },
    {
      "epoch": 0.5651594670972951,
      "grad_norm": 89.66084289550781,
      "learning_rate": 8.383114436732542e-06,
      "loss": 5.5357,
      "step": 13999
    },
    {
      "epoch": 0.5651998385143319,
      "grad_norm": 76.24240112304688,
      "learning_rate": 8.381824075521418e-06,
      "loss": 5.5779,
      "step": 14000
    },
    {
      "epoch": 0.5652402099313686,
      "grad_norm": 169.29904174804688,
      "learning_rate": 8.380533741977324e-06,
      "loss": 4.0486,
      "step": 14001
    },
    {
      "epoch": 0.5652805813484053,
      "grad_norm": 228.3102264404297,
      "learning_rate": 8.379243436122323e-06,
      "loss": 4.956,
      "step": 14002
    },
    {
      "epoch": 0.565320952765442,
      "grad_norm": 102.65901184082031,
      "learning_rate": 8.377953157978472e-06,
      "loss": 5.3543,
      "step": 14003
    },
    {
      "epoch": 0.5653613241824788,
      "grad_norm": 205.74581909179688,
      "learning_rate": 8.376662907567836e-06,
      "loss": 4.8563,
      "step": 14004
    },
    {
      "epoch": 0.5654016955995156,
      "grad_norm": 64.54214477539062,
      "learning_rate": 8.375372684912472e-06,
      "loss": 5.8682,
      "step": 14005
    },
    {
      "epoch": 0.5654420670165523,
      "grad_norm": 77.31472778320312,
      "learning_rate": 8.374082490034446e-06,
      "loss": 4.0783,
      "step": 14006
    },
    {
      "epoch": 0.565482438433589,
      "grad_norm": 111.29634094238281,
      "learning_rate": 8.372792322955808e-06,
      "loss": 5.871,
      "step": 14007
    },
    {
      "epoch": 0.5655228098506258,
      "grad_norm": 176.71426391601562,
      "learning_rate": 8.371502183698624e-06,
      "loss": 5.3841,
      "step": 14008
    },
    {
      "epoch": 0.5655631812676625,
      "grad_norm": 72.9297866821289,
      "learning_rate": 8.370212072284947e-06,
      "loss": 4.2395,
      "step": 14009
    },
    {
      "epoch": 0.5656035526846992,
      "grad_norm": 106.09937286376953,
      "learning_rate": 8.36892198873684e-06,
      "loss": 5.6211,
      "step": 14010
    },
    {
      "epoch": 0.5656439241017359,
      "grad_norm": 87.77306365966797,
      "learning_rate": 8.367631933076357e-06,
      "loss": 4.9699,
      "step": 14011
    },
    {
      "epoch": 0.5656842955187728,
      "grad_norm": 84.61943054199219,
      "learning_rate": 8.366341905325555e-06,
      "loss": 4.2117,
      "step": 14012
    },
    {
      "epoch": 0.5657246669358095,
      "grad_norm": 74.10826873779297,
      "learning_rate": 8.365051905506493e-06,
      "loss": 2.8675,
      "step": 14013
    },
    {
      "epoch": 0.5657650383528462,
      "grad_norm": 98.0325927734375,
      "learning_rate": 8.363761933641229e-06,
      "loss": 6.7946,
      "step": 14014
    },
    {
      "epoch": 0.5658054097698829,
      "grad_norm": 100.72403717041016,
      "learning_rate": 8.362471989751811e-06,
      "loss": 4.6937,
      "step": 14015
    },
    {
      "epoch": 0.5658457811869196,
      "grad_norm": 118.30979919433594,
      "learning_rate": 8.361182073860299e-06,
      "loss": 6.1955,
      "step": 14016
    },
    {
      "epoch": 0.5658861526039564,
      "grad_norm": 78.72659301757812,
      "learning_rate": 8.359892185988744e-06,
      "loss": 5.9236,
      "step": 14017
    },
    {
      "epoch": 0.5659265240209931,
      "grad_norm": 72.0840072631836,
      "learning_rate": 8.358602326159208e-06,
      "loss": 3.9617,
      "step": 14018
    },
    {
      "epoch": 0.5659668954380299,
      "grad_norm": 82.40973663330078,
      "learning_rate": 8.357312494393736e-06,
      "loss": 5.6216,
      "step": 14019
    },
    {
      "epoch": 0.5660072668550666,
      "grad_norm": 69.55595397949219,
      "learning_rate": 8.356022690714385e-06,
      "loss": 2.594,
      "step": 14020
    },
    {
      "epoch": 0.5660476382721034,
      "grad_norm": 76.93862915039062,
      "learning_rate": 8.354732915143206e-06,
      "loss": 6.0301,
      "step": 14021
    },
    {
      "epoch": 0.5660880096891401,
      "grad_norm": 63.22636032104492,
      "learning_rate": 8.353443167702254e-06,
      "loss": 3.8482,
      "step": 14022
    },
    {
      "epoch": 0.5661283811061768,
      "grad_norm": 267.9273376464844,
      "learning_rate": 8.352153448413576e-06,
      "loss": 4.7628,
      "step": 14023
    },
    {
      "epoch": 0.5661687525232135,
      "grad_norm": 165.3460693359375,
      "learning_rate": 8.350863757299228e-06,
      "loss": 5.2002,
      "step": 14024
    },
    {
      "epoch": 0.5662091239402502,
      "grad_norm": 89.67111206054688,
      "learning_rate": 8.349574094381257e-06,
      "loss": 4.0916,
      "step": 14025
    },
    {
      "epoch": 0.5662494953572871,
      "grad_norm": 107.32363891601562,
      "learning_rate": 8.34828445968172e-06,
      "loss": 5.6592,
      "step": 14026
    },
    {
      "epoch": 0.5662898667743238,
      "grad_norm": 75.90782928466797,
      "learning_rate": 8.34699485322266e-06,
      "loss": 3.6444,
      "step": 14027
    },
    {
      "epoch": 0.5663302381913605,
      "grad_norm": 73.57650756835938,
      "learning_rate": 8.345705275026124e-06,
      "loss": 4.0273,
      "step": 14028
    },
    {
      "epoch": 0.5663706096083972,
      "grad_norm": 80.85986328125,
      "learning_rate": 8.344415725114166e-06,
      "loss": 5.2209,
      "step": 14029
    },
    {
      "epoch": 0.566410981025434,
      "grad_norm": 229.91290283203125,
      "learning_rate": 8.343126203508835e-06,
      "loss": 4.1497,
      "step": 14030
    },
    {
      "epoch": 0.5664513524424707,
      "grad_norm": 86.02265930175781,
      "learning_rate": 8.341836710232175e-06,
      "loss": 6.0435,
      "step": 14031
    },
    {
      "epoch": 0.5664917238595075,
      "grad_norm": 105.37782287597656,
      "learning_rate": 8.340547245306235e-06,
      "loss": 6.479,
      "step": 14032
    },
    {
      "epoch": 0.5665320952765442,
      "grad_norm": 72.60492706298828,
      "learning_rate": 8.339257808753062e-06,
      "loss": 4.9972,
      "step": 14033
    },
    {
      "epoch": 0.566572466693581,
      "grad_norm": 103.00469970703125,
      "learning_rate": 8.337968400594706e-06,
      "loss": 3.7853,
      "step": 14034
    },
    {
      "epoch": 0.5666128381106177,
      "grad_norm": 78.74235534667969,
      "learning_rate": 8.336679020853206e-06,
      "loss": 5.4948,
      "step": 14035
    },
    {
      "epoch": 0.5666532095276544,
      "grad_norm": 77.73761749267578,
      "learning_rate": 8.335389669550612e-06,
      "loss": 5.3536,
      "step": 14036
    },
    {
      "epoch": 0.5666935809446911,
      "grad_norm": 111.23460388183594,
      "learning_rate": 8.334100346708965e-06,
      "loss": 4.6426,
      "step": 14037
    },
    {
      "epoch": 0.5667339523617279,
      "grad_norm": 76.667724609375,
      "learning_rate": 8.332811052350317e-06,
      "loss": 5.1831,
      "step": 14038
    },
    {
      "epoch": 0.5667743237787647,
      "grad_norm": 234.61099243164062,
      "learning_rate": 8.331521786496704e-06,
      "loss": 5.4956,
      "step": 14039
    },
    {
      "epoch": 0.5668146951958014,
      "grad_norm": 105.764404296875,
      "learning_rate": 8.33023254917017e-06,
      "loss": 4.955,
      "step": 14040
    },
    {
      "epoch": 0.5668550666128381,
      "grad_norm": 79.50122833251953,
      "learning_rate": 8.328943340392762e-06,
      "loss": 3.4651,
      "step": 14041
    },
    {
      "epoch": 0.5668954380298749,
      "grad_norm": 90.14591217041016,
      "learning_rate": 8.327654160186521e-06,
      "loss": 5.1589,
      "step": 14042
    },
    {
      "epoch": 0.5669358094469116,
      "grad_norm": 108.24127197265625,
      "learning_rate": 8.326365008573488e-06,
      "loss": 4.3765,
      "step": 14043
    },
    {
      "epoch": 0.5669761808639483,
      "grad_norm": 107.6142578125,
      "learning_rate": 8.325075885575704e-06,
      "loss": 4.9733,
      "step": 14044
    },
    {
      "epoch": 0.567016552280985,
      "grad_norm": 109.70357513427734,
      "learning_rate": 8.323786791215212e-06,
      "loss": 4.7173,
      "step": 14045
    },
    {
      "epoch": 0.5670569236980219,
      "grad_norm": 80.33646392822266,
      "learning_rate": 8.322497725514055e-06,
      "loss": 6.7848,
      "step": 14046
    },
    {
      "epoch": 0.5670972951150586,
      "grad_norm": 71.72933959960938,
      "learning_rate": 8.321208688494266e-06,
      "loss": 5.8855,
      "step": 14047
    },
    {
      "epoch": 0.5671376665320953,
      "grad_norm": 101.98567199707031,
      "learning_rate": 8.319919680177889e-06,
      "loss": 5.4351,
      "step": 14048
    },
    {
      "epoch": 0.567178037949132,
      "grad_norm": 120.12004852294922,
      "learning_rate": 8.31863070058696e-06,
      "loss": 5.0096,
      "step": 14049
    },
    {
      "epoch": 0.5672184093661687,
      "grad_norm": 99.64173889160156,
      "learning_rate": 8.317341749743524e-06,
      "loss": 6.0338,
      "step": 14050
    },
    {
      "epoch": 0.5672587807832055,
      "grad_norm": 85.74974060058594,
      "learning_rate": 8.316052827669612e-06,
      "loss": 6.0962,
      "step": 14051
    },
    {
      "epoch": 0.5672991522002422,
      "grad_norm": 106.94023895263672,
      "learning_rate": 8.314763934387265e-06,
      "loss": 3.6863,
      "step": 14052
    },
    {
      "epoch": 0.567339523617279,
      "grad_norm": 76.46601104736328,
      "learning_rate": 8.313475069918517e-06,
      "loss": 4.1483,
      "step": 14053
    },
    {
      "epoch": 0.5673798950343157,
      "grad_norm": 74.62926483154297,
      "learning_rate": 8.312186234285415e-06,
      "loss": 5.254,
      "step": 14054
    },
    {
      "epoch": 0.5674202664513525,
      "grad_norm": 79.2612533569336,
      "learning_rate": 8.31089742750998e-06,
      "loss": 4.5254,
      "step": 14055
    },
    {
      "epoch": 0.5674606378683892,
      "grad_norm": 91.513916015625,
      "learning_rate": 8.309608649614256e-06,
      "loss": 5.673,
      "step": 14056
    },
    {
      "epoch": 0.5675010092854259,
      "grad_norm": 143.68228149414062,
      "learning_rate": 8.308319900620275e-06,
      "loss": 4.9797,
      "step": 14057
    },
    {
      "epoch": 0.5675413807024626,
      "grad_norm": 127.956298828125,
      "learning_rate": 8.307031180550077e-06,
      "loss": 5.4137,
      "step": 14058
    },
    {
      "epoch": 0.5675817521194993,
      "grad_norm": 75.30791473388672,
      "learning_rate": 8.30574248942569e-06,
      "loss": 6.3482,
      "step": 14059
    },
    {
      "epoch": 0.5676221235365362,
      "grad_norm": 117.89230346679688,
      "learning_rate": 8.304453827269149e-06,
      "loss": 5.7078,
      "step": 14060
    },
    {
      "epoch": 0.5676624949535729,
      "grad_norm": 102.68270874023438,
      "learning_rate": 8.303165194102489e-06,
      "loss": 5.0691,
      "step": 14061
    },
    {
      "epoch": 0.5677028663706096,
      "grad_norm": 79.94886016845703,
      "learning_rate": 8.301876589947747e-06,
      "loss": 4.912,
      "step": 14062
    },
    {
      "epoch": 0.5677432377876463,
      "grad_norm": 106.69300842285156,
      "learning_rate": 8.300588014826944e-06,
      "loss": 4.867,
      "step": 14063
    },
    {
      "epoch": 0.5677836092046831,
      "grad_norm": 111.56544494628906,
      "learning_rate": 8.299299468762118e-06,
      "loss": 5.5312,
      "step": 14064
    },
    {
      "epoch": 0.5678239806217198,
      "grad_norm": 130.0897979736328,
      "learning_rate": 8.298010951775302e-06,
      "loss": 5.0206,
      "step": 14065
    },
    {
      "epoch": 0.5678643520387565,
      "grad_norm": 94.69355010986328,
      "learning_rate": 8.296722463888525e-06,
      "loss": 5.3416,
      "step": 14066
    },
    {
      "epoch": 0.5679047234557933,
      "grad_norm": 81.82575225830078,
      "learning_rate": 8.295434005123814e-06,
      "loss": 3.5611,
      "step": 14067
    },
    {
      "epoch": 0.5679450948728301,
      "grad_norm": 103.39022827148438,
      "learning_rate": 8.294145575503203e-06,
      "loss": 6.9263,
      "step": 14068
    },
    {
      "epoch": 0.5679854662898668,
      "grad_norm": 89.7146987915039,
      "learning_rate": 8.292857175048715e-06,
      "loss": 4.651,
      "step": 14069
    },
    {
      "epoch": 0.5680258377069035,
      "grad_norm": 100.47369384765625,
      "learning_rate": 8.291568803782387e-06,
      "loss": 4.0467,
      "step": 14070
    },
    {
      "epoch": 0.5680662091239402,
      "grad_norm": 101.85289001464844,
      "learning_rate": 8.290280461726241e-06,
      "loss": 3.9359,
      "step": 14071
    },
    {
      "epoch": 0.568106580540977,
      "grad_norm": 67.00210571289062,
      "learning_rate": 8.288992148902307e-06,
      "loss": 5.1153,
      "step": 14072
    },
    {
      "epoch": 0.5681469519580137,
      "grad_norm": 78.21621704101562,
      "learning_rate": 8.287703865332611e-06,
      "loss": 4.3216,
      "step": 14073
    },
    {
      "epoch": 0.5681873233750505,
      "grad_norm": 75.48265838623047,
      "learning_rate": 8.286415611039185e-06,
      "loss": 3.0513,
      "step": 14074
    },
    {
      "epoch": 0.5682276947920872,
      "grad_norm": 99.80979919433594,
      "learning_rate": 8.285127386044046e-06,
      "loss": 4.0551,
      "step": 14075
    },
    {
      "epoch": 0.568268066209124,
      "grad_norm": 87.65228271484375,
      "learning_rate": 8.283839190369225e-06,
      "loss": 4.3731,
      "step": 14076
    },
    {
      "epoch": 0.5683084376261607,
      "grad_norm": 113.15260314941406,
      "learning_rate": 8.282551024036744e-06,
      "loss": 3.7004,
      "step": 14077
    },
    {
      "epoch": 0.5683488090431974,
      "grad_norm": 91.84239196777344,
      "learning_rate": 8.281262887068635e-06,
      "loss": 5.8922,
      "step": 14078
    },
    {
      "epoch": 0.5683891804602341,
      "grad_norm": 79.09585571289062,
      "learning_rate": 8.279974779486914e-06,
      "loss": 6.9067,
      "step": 14079
    },
    {
      "epoch": 0.5684295518772708,
      "grad_norm": 91.36324310302734,
      "learning_rate": 8.278686701313608e-06,
      "loss": 5.5775,
      "step": 14080
    },
    {
      "epoch": 0.5684699232943077,
      "grad_norm": 93.30179595947266,
      "learning_rate": 8.277398652570738e-06,
      "loss": 5.958,
      "step": 14081
    },
    {
      "epoch": 0.5685102947113444,
      "grad_norm": 81.8832778930664,
      "learning_rate": 8.276110633280335e-06,
      "loss": 2.882,
      "step": 14082
    },
    {
      "epoch": 0.5685506661283811,
      "grad_norm": 83.15425872802734,
      "learning_rate": 8.274822643464409e-06,
      "loss": 3.3457,
      "step": 14083
    },
    {
      "epoch": 0.5685910375454178,
      "grad_norm": 139.8023681640625,
      "learning_rate": 8.273534683144989e-06,
      "loss": 5.1794,
      "step": 14084
    },
    {
      "epoch": 0.5686314089624546,
      "grad_norm": 143.53807067871094,
      "learning_rate": 8.27224675234409e-06,
      "loss": 5.6278,
      "step": 14085
    },
    {
      "epoch": 0.5686717803794913,
      "grad_norm": 113.38233184814453,
      "learning_rate": 8.270958851083746e-06,
      "loss": 4.6861,
      "step": 14086
    },
    {
      "epoch": 0.568712151796528,
      "grad_norm": 95.31273651123047,
      "learning_rate": 8.269670979385961e-06,
      "loss": 4.786,
      "step": 14087
    },
    {
      "epoch": 0.5687525232135648,
      "grad_norm": 124.30706787109375,
      "learning_rate": 8.268383137272764e-06,
      "loss": 4.3476,
      "step": 14088
    },
    {
      "epoch": 0.5687928946306016,
      "grad_norm": 79.87879943847656,
      "learning_rate": 8.26709532476617e-06,
      "loss": 5.6949,
      "step": 14089
    },
    {
      "epoch": 0.5688332660476383,
      "grad_norm": 100.46916198730469,
      "learning_rate": 8.2658075418882e-06,
      "loss": 5.5122,
      "step": 14090
    },
    {
      "epoch": 0.568873637464675,
      "grad_norm": 93.1527328491211,
      "learning_rate": 8.264519788660872e-06,
      "loss": 4.4139,
      "step": 14091
    },
    {
      "epoch": 0.5689140088817117,
      "grad_norm": 110.63861846923828,
      "learning_rate": 8.263232065106202e-06,
      "loss": 4.8444,
      "step": 14092
    },
    {
      "epoch": 0.5689543802987485,
      "grad_norm": 79.17230224609375,
      "learning_rate": 8.261944371246208e-06,
      "loss": 5.0847,
      "step": 14093
    },
    {
      "epoch": 0.5689947517157852,
      "grad_norm": 137.54661560058594,
      "learning_rate": 8.260656707102911e-06,
      "loss": 5.016,
      "step": 14094
    },
    {
      "epoch": 0.569035123132822,
      "grad_norm": 90.14929962158203,
      "learning_rate": 8.259369072698318e-06,
      "loss": 4.8569,
      "step": 14095
    },
    {
      "epoch": 0.5690754945498587,
      "grad_norm": 90.1776123046875,
      "learning_rate": 8.25808146805445e-06,
      "loss": 5.7309,
      "step": 14096
    },
    {
      "epoch": 0.5691158659668955,
      "grad_norm": 84.52204132080078,
      "learning_rate": 8.25679389319332e-06,
      "loss": 5.2672,
      "step": 14097
    },
    {
      "epoch": 0.5691562373839322,
      "grad_norm": 113.42927551269531,
      "learning_rate": 8.255506348136947e-06,
      "loss": 5.896,
      "step": 14098
    },
    {
      "epoch": 0.5691966088009689,
      "grad_norm": 90.87125396728516,
      "learning_rate": 8.254218832907339e-06,
      "loss": 4.163,
      "step": 14099
    },
    {
      "epoch": 0.5692369802180056,
      "grad_norm": 279.4277038574219,
      "learning_rate": 8.252931347526512e-06,
      "loss": 2.7868,
      "step": 14100
    },
    {
      "epoch": 0.5692773516350423,
      "grad_norm": 109.21839141845703,
      "learning_rate": 8.25164389201648e-06,
      "loss": 4.8061,
      "step": 14101
    },
    {
      "epoch": 0.5693177230520792,
      "grad_norm": 132.6114959716797,
      "learning_rate": 8.25035646639926e-06,
      "loss": 4.2882,
      "step": 14102
    },
    {
      "epoch": 0.5693580944691159,
      "grad_norm": 72.50787353515625,
      "learning_rate": 8.249069070696852e-06,
      "loss": 4.9322,
      "step": 14103
    },
    {
      "epoch": 0.5693984658861526,
      "grad_norm": 119.81021881103516,
      "learning_rate": 8.247781704931275e-06,
      "loss": 4.9077,
      "step": 14104
    },
    {
      "epoch": 0.5694388373031893,
      "grad_norm": 88.2247085571289,
      "learning_rate": 8.246494369124541e-06,
      "loss": 5.716,
      "step": 14105
    },
    {
      "epoch": 0.5694792087202261,
      "grad_norm": 112.65699768066406,
      "learning_rate": 8.245207063298661e-06,
      "loss": 5.1546,
      "step": 14106
    },
    {
      "epoch": 0.5695195801372628,
      "grad_norm": 97.21312713623047,
      "learning_rate": 8.243919787475638e-06,
      "loss": 5.271,
      "step": 14107
    },
    {
      "epoch": 0.5695599515542995,
      "grad_norm": 144.49847412109375,
      "learning_rate": 8.242632541677487e-06,
      "loss": 5.0738,
      "step": 14108
    },
    {
      "epoch": 0.5696003229713363,
      "grad_norm": 101.7205581665039,
      "learning_rate": 8.241345325926216e-06,
      "loss": 5.0332,
      "step": 14109
    },
    {
      "epoch": 0.5696406943883731,
      "grad_norm": 106.21136474609375,
      "learning_rate": 8.240058140243834e-06,
      "loss": 4.2405,
      "step": 14110
    },
    {
      "epoch": 0.5696810658054098,
      "grad_norm": 93.05300903320312,
      "learning_rate": 8.238770984652346e-06,
      "loss": 5.5356,
      "step": 14111
    },
    {
      "epoch": 0.5697214372224465,
      "grad_norm": 70.4133071899414,
      "learning_rate": 8.237483859173762e-06,
      "loss": 6.2528,
      "step": 14112
    },
    {
      "epoch": 0.5697618086394832,
      "grad_norm": 71.2757339477539,
      "learning_rate": 8.236196763830088e-06,
      "loss": 4.8129,
      "step": 14113
    },
    {
      "epoch": 0.56980218005652,
      "grad_norm": 147.67025756835938,
      "learning_rate": 8.234909698643333e-06,
      "loss": 4.6828,
      "step": 14114
    },
    {
      "epoch": 0.5698425514735567,
      "grad_norm": 116.2861557006836,
      "learning_rate": 8.2336226636355e-06,
      "loss": 5.7916,
      "step": 14115
    },
    {
      "epoch": 0.5698829228905935,
      "grad_norm": 78.5772705078125,
      "learning_rate": 8.232335658828593e-06,
      "loss": 2.4504,
      "step": 14116
    },
    {
      "epoch": 0.5699232943076302,
      "grad_norm": 86.29695129394531,
      "learning_rate": 8.231048684244619e-06,
      "loss": 6.2941,
      "step": 14117
    },
    {
      "epoch": 0.569963665724667,
      "grad_norm": 82.78607940673828,
      "learning_rate": 8.229761739905582e-06,
      "loss": 6.0415,
      "step": 14118
    },
    {
      "epoch": 0.5700040371417037,
      "grad_norm": 99.21006774902344,
      "learning_rate": 8.228474825833483e-06,
      "loss": 4.9842,
      "step": 14119
    },
    {
      "epoch": 0.5700444085587404,
      "grad_norm": 124.7492904663086,
      "learning_rate": 8.22718794205033e-06,
      "loss": 6.1154,
      "step": 14120
    },
    {
      "epoch": 0.5700847799757771,
      "grad_norm": 152.99050903320312,
      "learning_rate": 8.225901088578123e-06,
      "loss": 5.0912,
      "step": 14121
    },
    {
      "epoch": 0.5701251513928138,
      "grad_norm": 119.98925018310547,
      "learning_rate": 8.224614265438868e-06,
      "loss": 4.1888,
      "step": 14122
    },
    {
      "epoch": 0.5701655228098507,
      "grad_norm": 97.29492950439453,
      "learning_rate": 8.22332747265456e-06,
      "loss": 6.0224,
      "step": 14123
    },
    {
      "epoch": 0.5702058942268874,
      "grad_norm": 83.89924621582031,
      "learning_rate": 8.2220407102472e-06,
      "loss": 5.9192,
      "step": 14124
    },
    {
      "epoch": 0.5702462656439241,
      "grad_norm": 82.50912475585938,
      "learning_rate": 8.220753978238795e-06,
      "loss": 5.1321,
      "step": 14125
    },
    {
      "epoch": 0.5702866370609608,
      "grad_norm": 75.28612518310547,
      "learning_rate": 8.219467276651346e-06,
      "loss": 5.4857,
      "step": 14126
    },
    {
      "epoch": 0.5703270084779976,
      "grad_norm": 119.57453155517578,
      "learning_rate": 8.218180605506844e-06,
      "loss": 4.8419,
      "step": 14127
    },
    {
      "epoch": 0.5703673798950343,
      "grad_norm": 101.46251678466797,
      "learning_rate": 8.216893964827293e-06,
      "loss": 5.6827,
      "step": 14128
    },
    {
      "epoch": 0.570407751312071,
      "grad_norm": 87.98756408691406,
      "learning_rate": 8.215607354634691e-06,
      "loss": 5.1187,
      "step": 14129
    },
    {
      "epoch": 0.5704481227291078,
      "grad_norm": 127.56759643554688,
      "learning_rate": 8.21432077495104e-06,
      "loss": 3.7697,
      "step": 14130
    },
    {
      "epoch": 0.5704884941461446,
      "grad_norm": 149.71241760253906,
      "learning_rate": 8.21303422579833e-06,
      "loss": 4.0768,
      "step": 14131
    },
    {
      "epoch": 0.5705288655631813,
      "grad_norm": 69.45487213134766,
      "learning_rate": 8.211747707198563e-06,
      "loss": 5.5348,
      "step": 14132
    },
    {
      "epoch": 0.570569236980218,
      "grad_norm": 86.02277374267578,
      "learning_rate": 8.210461219173735e-06,
      "loss": 6.4661,
      "step": 14133
    },
    {
      "epoch": 0.5706096083972547,
      "grad_norm": 106.6602554321289,
      "learning_rate": 8.209174761745843e-06,
      "loss": 4.4155,
      "step": 14134
    },
    {
      "epoch": 0.5706499798142914,
      "grad_norm": 72.20940399169922,
      "learning_rate": 8.207888334936879e-06,
      "loss": 4.5754,
      "step": 14135
    },
    {
      "epoch": 0.5706903512313283,
      "grad_norm": 89.0307846069336,
      "learning_rate": 8.20660193876884e-06,
      "loss": 5.7541,
      "step": 14136
    },
    {
      "epoch": 0.570730722648365,
      "grad_norm": 144.9119110107422,
      "learning_rate": 8.205315573263717e-06,
      "loss": 4.5526,
      "step": 14137
    },
    {
      "epoch": 0.5707710940654017,
      "grad_norm": 90.2215805053711,
      "learning_rate": 8.20402923844351e-06,
      "loss": 3.2107,
      "step": 14138
    },
    {
      "epoch": 0.5708114654824384,
      "grad_norm": 87.00431060791016,
      "learning_rate": 8.202742934330208e-06,
      "loss": 5.0264,
      "step": 14139
    },
    {
      "epoch": 0.5708518368994752,
      "grad_norm": 94.71620178222656,
      "learning_rate": 8.201456660945805e-06,
      "loss": 5.1443,
      "step": 14140
    },
    {
      "epoch": 0.5708922083165119,
      "grad_norm": 89.11365509033203,
      "learning_rate": 8.200170418312292e-06,
      "loss": 5.4406,
      "step": 14141
    },
    {
      "epoch": 0.5709325797335486,
      "grad_norm": 91.73219299316406,
      "learning_rate": 8.198884206451667e-06,
      "loss": 6.0359,
      "step": 14142
    },
    {
      "epoch": 0.5709729511505854,
      "grad_norm": 193.54690551757812,
      "learning_rate": 8.197598025385912e-06,
      "loss": 5.8826,
      "step": 14143
    },
    {
      "epoch": 0.5710133225676222,
      "grad_norm": 97.6375732421875,
      "learning_rate": 8.196311875137021e-06,
      "loss": 4.7999,
      "step": 14144
    },
    {
      "epoch": 0.5710536939846589,
      "grad_norm": 173.43321228027344,
      "learning_rate": 8.195025755726985e-06,
      "loss": 5.4153,
      "step": 14145
    },
    {
      "epoch": 0.5710940654016956,
      "grad_norm": 107.64518737792969,
      "learning_rate": 8.193739667177799e-06,
      "loss": 3.3498,
      "step": 14146
    },
    {
      "epoch": 0.5711344368187323,
      "grad_norm": 87.61840057373047,
      "learning_rate": 8.19245360951144e-06,
      "loss": 3.6405,
      "step": 14147
    },
    {
      "epoch": 0.571174808235769,
      "grad_norm": 139.47332763671875,
      "learning_rate": 8.191167582749908e-06,
      "loss": 3.8417,
      "step": 14148
    },
    {
      "epoch": 0.5712151796528058,
      "grad_norm": 94.7091064453125,
      "learning_rate": 8.189881586915183e-06,
      "loss": 5.924,
      "step": 14149
    },
    {
      "epoch": 0.5712555510698426,
      "grad_norm": 1576.1319580078125,
      "learning_rate": 8.188595622029258e-06,
      "loss": 4.0716,
      "step": 14150
    },
    {
      "epoch": 0.5712959224868793,
      "grad_norm": 74.57904052734375,
      "learning_rate": 8.187309688114118e-06,
      "loss": 5.1859,
      "step": 14151
    },
    {
      "epoch": 0.571336293903916,
      "grad_norm": 115.0982666015625,
      "learning_rate": 8.186023785191748e-06,
      "loss": 6.038,
      "step": 14152
    },
    {
      "epoch": 0.5713766653209528,
      "grad_norm": 77.2236557006836,
      "learning_rate": 8.184737913284135e-06,
      "loss": 3.585,
      "step": 14153
    },
    {
      "epoch": 0.5714170367379895,
      "grad_norm": 99.07325744628906,
      "learning_rate": 8.18345207241327e-06,
      "loss": 4.9632,
      "step": 14154
    },
    {
      "epoch": 0.5714574081550262,
      "grad_norm": 89.99821472167969,
      "learning_rate": 8.182166262601127e-06,
      "loss": 6.9446,
      "step": 14155
    },
    {
      "epoch": 0.5714977795720629,
      "grad_norm": 131.6683807373047,
      "learning_rate": 8.180880483869696e-06,
      "loss": 6.5487,
      "step": 14156
    },
    {
      "epoch": 0.5715381509890998,
      "grad_norm": 92.4857177734375,
      "learning_rate": 8.179594736240962e-06,
      "loss": 6.6729,
      "step": 14157
    },
    {
      "epoch": 0.5715785224061365,
      "grad_norm": 90.99359130859375,
      "learning_rate": 8.178309019736908e-06,
      "loss": 5.7276,
      "step": 14158
    },
    {
      "epoch": 0.5716188938231732,
      "grad_norm": 104.02703857421875,
      "learning_rate": 8.177023334379515e-06,
      "loss": 5.449,
      "step": 14159
    },
    {
      "epoch": 0.5716592652402099,
      "grad_norm": 97.77632141113281,
      "learning_rate": 8.175737680190765e-06,
      "loss": 6.7731,
      "step": 14160
    },
    {
      "epoch": 0.5716996366572467,
      "grad_norm": 104.59073638916016,
      "learning_rate": 8.174452057192642e-06,
      "loss": 5.6075,
      "step": 14161
    },
    {
      "epoch": 0.5717400080742834,
      "grad_norm": 95.82573699951172,
      "learning_rate": 8.173166465407129e-06,
      "loss": 6.7606,
      "step": 14162
    },
    {
      "epoch": 0.5717803794913201,
      "grad_norm": 89.35386657714844,
      "learning_rate": 8.1718809048562e-06,
      "loss": 4.3455,
      "step": 14163
    },
    {
      "epoch": 0.5718207509083569,
      "grad_norm": 117.98998260498047,
      "learning_rate": 8.170595375561836e-06,
      "loss": 3.3105,
      "step": 14164
    },
    {
      "epoch": 0.5718611223253937,
      "grad_norm": 85.4251708984375,
      "learning_rate": 8.169309877546021e-06,
      "loss": 5.4775,
      "step": 14165
    },
    {
      "epoch": 0.5719014937424304,
      "grad_norm": 95.78079986572266,
      "learning_rate": 8.168024410830735e-06,
      "loss": 5.7855,
      "step": 14166
    },
    {
      "epoch": 0.5719418651594671,
      "grad_norm": 80.08403015136719,
      "learning_rate": 8.166738975437951e-06,
      "loss": 5.0342,
      "step": 14167
    },
    {
      "epoch": 0.5719822365765038,
      "grad_norm": 113.84835052490234,
      "learning_rate": 8.165453571389649e-06,
      "loss": 6.7474,
      "step": 14168
    },
    {
      "epoch": 0.5720226079935405,
      "grad_norm": 77.5891342163086,
      "learning_rate": 8.164168198707807e-06,
      "loss": 5.2239,
      "step": 14169
    },
    {
      "epoch": 0.5720629794105773,
      "grad_norm": 85.0730209350586,
      "learning_rate": 8.162882857414407e-06,
      "loss": 5.1196,
      "step": 14170
    },
    {
      "epoch": 0.5721033508276141,
      "grad_norm": 91.35454559326172,
      "learning_rate": 8.161597547531415e-06,
      "loss": 6.5838,
      "step": 14171
    },
    {
      "epoch": 0.5721437222446508,
      "grad_norm": 107.97587585449219,
      "learning_rate": 8.160312269080812e-06,
      "loss": 6.2049,
      "step": 14172
    },
    {
      "epoch": 0.5721840936616875,
      "grad_norm": 133.2084503173828,
      "learning_rate": 8.159027022084573e-06,
      "loss": 4.3521,
      "step": 14173
    },
    {
      "epoch": 0.5722244650787243,
      "grad_norm": 106.01570129394531,
      "learning_rate": 8.157741806564678e-06,
      "loss": 4.7447,
      "step": 14174
    },
    {
      "epoch": 0.572264836495761,
      "grad_norm": 78.91256713867188,
      "learning_rate": 8.156456622543092e-06,
      "loss": 4.6114,
      "step": 14175
    },
    {
      "epoch": 0.5723052079127977,
      "grad_norm": 127.99156951904297,
      "learning_rate": 8.155171470041791e-06,
      "loss": 3.9838,
      "step": 14176
    },
    {
      "epoch": 0.5723455793298344,
      "grad_norm": 120.03265380859375,
      "learning_rate": 8.153886349082752e-06,
      "loss": 5.1442,
      "step": 14177
    },
    {
      "epoch": 0.5723859507468713,
      "grad_norm": 110.11656188964844,
      "learning_rate": 8.152601259687947e-06,
      "loss": 4.5213,
      "step": 14178
    },
    {
      "epoch": 0.572426322163908,
      "grad_norm": 119.7095718383789,
      "learning_rate": 8.151316201879344e-06,
      "loss": 5.9207,
      "step": 14179
    },
    {
      "epoch": 0.5724666935809447,
      "grad_norm": 94.20882415771484,
      "learning_rate": 8.150031175678917e-06,
      "loss": 3.7259,
      "step": 14180
    },
    {
      "epoch": 0.5725070649979814,
      "grad_norm": 84.42086029052734,
      "learning_rate": 8.148746181108637e-06,
      "loss": 2.91,
      "step": 14181
    },
    {
      "epoch": 0.5725474364150182,
      "grad_norm": 83.8138198852539,
      "learning_rate": 8.147461218190479e-06,
      "loss": 5.5515,
      "step": 14182
    },
    {
      "epoch": 0.5725878078320549,
      "grad_norm": 89.42378234863281,
      "learning_rate": 8.146176286946403e-06,
      "loss": 3.7461,
      "step": 14183
    },
    {
      "epoch": 0.5726281792490916,
      "grad_norm": 104.02001190185547,
      "learning_rate": 8.144891387398383e-06,
      "loss": 6.8926,
      "step": 14184
    },
    {
      "epoch": 0.5726685506661284,
      "grad_norm": 96.57896423339844,
      "learning_rate": 8.14360651956839e-06,
      "loss": 5.7331,
      "step": 14185
    },
    {
      "epoch": 0.5727089220831652,
      "grad_norm": 87.15370178222656,
      "learning_rate": 8.142321683478392e-06,
      "loss": 3.6926,
      "step": 14186
    },
    {
      "epoch": 0.5727492935002019,
      "grad_norm": 91.59690856933594,
      "learning_rate": 8.141036879150354e-06,
      "loss": 6.0718,
      "step": 14187
    },
    {
      "epoch": 0.5727896649172386,
      "grad_norm": 87.43384552001953,
      "learning_rate": 8.139752106606242e-06,
      "loss": 4.935,
      "step": 14188
    },
    {
      "epoch": 0.5728300363342753,
      "grad_norm": 87.43750762939453,
      "learning_rate": 8.138467365868028e-06,
      "loss": 5.731,
      "step": 14189
    },
    {
      "epoch": 0.572870407751312,
      "grad_norm": 105.60749053955078,
      "learning_rate": 8.137182656957676e-06,
      "loss": 2.7344,
      "step": 14190
    },
    {
      "epoch": 0.5729107791683488,
      "grad_norm": 89.1119384765625,
      "learning_rate": 8.13589797989715e-06,
      "loss": 4.386,
      "step": 14191
    },
    {
      "epoch": 0.5729511505853856,
      "grad_norm": 130.88775634765625,
      "learning_rate": 8.134613334708412e-06,
      "loss": 4.1353,
      "step": 14192
    },
    {
      "epoch": 0.5729915220024223,
      "grad_norm": 119.51248168945312,
      "learning_rate": 8.133328721413432e-06,
      "loss": 4.9615,
      "step": 14193
    },
    {
      "epoch": 0.573031893419459,
      "grad_norm": 63.50917434692383,
      "learning_rate": 8.132044140034175e-06,
      "loss": 2.8079,
      "step": 14194
    },
    {
      "epoch": 0.5730722648364958,
      "grad_norm": 75.54106140136719,
      "learning_rate": 8.130759590592599e-06,
      "loss": 3.1068,
      "step": 14195
    },
    {
      "epoch": 0.5731126362535325,
      "grad_norm": 162.38180541992188,
      "learning_rate": 8.129475073110667e-06,
      "loss": 6.5712,
      "step": 14196
    },
    {
      "epoch": 0.5731530076705692,
      "grad_norm": 115.6552963256836,
      "learning_rate": 8.128190587610343e-06,
      "loss": 5.3476,
      "step": 14197
    },
    {
      "epoch": 0.5731933790876059,
      "grad_norm": 106.47581481933594,
      "learning_rate": 8.126906134113593e-06,
      "loss": 5.2964,
      "step": 14198
    },
    {
      "epoch": 0.5732337505046428,
      "grad_norm": 108.65393829345703,
      "learning_rate": 8.125621712642371e-06,
      "loss": 5.587,
      "step": 14199
    },
    {
      "epoch": 0.5732741219216795,
      "grad_norm": 106.1260986328125,
      "learning_rate": 8.12433732321864e-06,
      "loss": 5.835,
      "step": 14200
    },
    {
      "epoch": 0.5733144933387162,
      "grad_norm": 441.94793701171875,
      "learning_rate": 8.123052965864362e-06,
      "loss": 3.1866,
      "step": 14201
    },
    {
      "epoch": 0.5733548647557529,
      "grad_norm": 174.4940948486328,
      "learning_rate": 8.121768640601494e-06,
      "loss": 4.8108,
      "step": 14202
    },
    {
      "epoch": 0.5733952361727896,
      "grad_norm": 100.80767822265625,
      "learning_rate": 8.120484347452003e-06,
      "loss": 6.1162,
      "step": 14203
    },
    {
      "epoch": 0.5734356075898264,
      "grad_norm": 87.15760803222656,
      "learning_rate": 8.119200086437834e-06,
      "loss": 4.9009,
      "step": 14204
    },
    {
      "epoch": 0.5734759790068631,
      "grad_norm": 125.25083923339844,
      "learning_rate": 8.117915857580952e-06,
      "loss": 4.1022,
      "step": 14205
    },
    {
      "epoch": 0.5735163504238999,
      "grad_norm": 357.4539794921875,
      "learning_rate": 8.116631660903313e-06,
      "loss": 4.9842,
      "step": 14206
    },
    {
      "epoch": 0.5735567218409366,
      "grad_norm": 76.74211883544922,
      "learning_rate": 8.115347496426878e-06,
      "loss": 5.5,
      "step": 14207
    },
    {
      "epoch": 0.5735970932579734,
      "grad_norm": 74.35222625732422,
      "learning_rate": 8.114063364173597e-06,
      "loss": 5.1652,
      "step": 14208
    },
    {
      "epoch": 0.5736374646750101,
      "grad_norm": 92.9540786743164,
      "learning_rate": 8.11277926416543e-06,
      "loss": 6.1673,
      "step": 14209
    },
    {
      "epoch": 0.5736778360920468,
      "grad_norm": 119.62926483154297,
      "learning_rate": 8.11149519642433e-06,
      "loss": 4.739,
      "step": 14210
    },
    {
      "epoch": 0.5737182075090835,
      "grad_norm": 97.91493225097656,
      "learning_rate": 8.110211160972256e-06,
      "loss": 4.9704,
      "step": 14211
    },
    {
      "epoch": 0.5737585789261203,
      "grad_norm": 164.1263427734375,
      "learning_rate": 8.108927157831154e-06,
      "loss": 5.6008,
      "step": 14212
    },
    {
      "epoch": 0.5737989503431571,
      "grad_norm": 135.61265563964844,
      "learning_rate": 8.107643187022982e-06,
      "loss": 5.1552,
      "step": 14213
    },
    {
      "epoch": 0.5738393217601938,
      "grad_norm": 86.7645263671875,
      "learning_rate": 8.106359248569691e-06,
      "loss": 5.4474,
      "step": 14214
    },
    {
      "epoch": 0.5738796931772305,
      "grad_norm": 87.90699768066406,
      "learning_rate": 8.10507534249324e-06,
      "loss": 4.6834,
      "step": 14215
    },
    {
      "epoch": 0.5739200645942673,
      "grad_norm": 83.49117279052734,
      "learning_rate": 8.103791468815573e-06,
      "loss": 3.8406,
      "step": 14216
    },
    {
      "epoch": 0.573960436011304,
      "grad_norm": 71.0898666381836,
      "learning_rate": 8.102507627558642e-06,
      "loss": 4.4936,
      "step": 14217
    },
    {
      "epoch": 0.5740008074283407,
      "grad_norm": 73.77233123779297,
      "learning_rate": 8.1012238187444e-06,
      "loss": 5.685,
      "step": 14218
    },
    {
      "epoch": 0.5740411788453774,
      "grad_norm": 71.72514343261719,
      "learning_rate": 8.0999400423948e-06,
      "loss": 4.5098,
      "step": 14219
    },
    {
      "epoch": 0.5740815502624143,
      "grad_norm": 172.19691467285156,
      "learning_rate": 8.098656298531784e-06,
      "loss": 4.0161,
      "step": 14220
    },
    {
      "epoch": 0.574121921679451,
      "grad_norm": 108.09680938720703,
      "learning_rate": 8.097372587177306e-06,
      "loss": 4.0846,
      "step": 14221
    },
    {
      "epoch": 0.5741622930964877,
      "grad_norm": 106.0416030883789,
      "learning_rate": 8.096088908353316e-06,
      "loss": 4.895,
      "step": 14222
    },
    {
      "epoch": 0.5742026645135244,
      "grad_norm": 138.5083770751953,
      "learning_rate": 8.094805262081761e-06,
      "loss": 6.8178,
      "step": 14223
    },
    {
      "epoch": 0.5742430359305611,
      "grad_norm": 324.1100769042969,
      "learning_rate": 8.093521648384585e-06,
      "loss": 3.1754,
      "step": 14224
    },
    {
      "epoch": 0.5742834073475979,
      "grad_norm": 89.5511703491211,
      "learning_rate": 8.092238067283736e-06,
      "loss": 6.4391,
      "step": 14225
    },
    {
      "epoch": 0.5743237787646346,
      "grad_norm": 98.35005950927734,
      "learning_rate": 8.090954518801162e-06,
      "loss": 5.7103,
      "step": 14226
    },
    {
      "epoch": 0.5743641501816714,
      "grad_norm": 98.11402893066406,
      "learning_rate": 8.089671002958809e-06,
      "loss": 4.4286,
      "step": 14227
    },
    {
      "epoch": 0.5744045215987081,
      "grad_norm": 100.63263702392578,
      "learning_rate": 8.088387519778617e-06,
      "loss": 4.5982,
      "step": 14228
    },
    {
      "epoch": 0.5744448930157449,
      "grad_norm": 91.40799713134766,
      "learning_rate": 8.087104069282538e-06,
      "loss": 6.2518,
      "step": 14229
    },
    {
      "epoch": 0.5744852644327816,
      "grad_norm": 90.86361694335938,
      "learning_rate": 8.085820651492511e-06,
      "loss": 5.5955,
      "step": 14230
    },
    {
      "epoch": 0.5745256358498183,
      "grad_norm": 96.3197021484375,
      "learning_rate": 8.084537266430485e-06,
      "loss": 4.092,
      "step": 14231
    },
    {
      "epoch": 0.574566007266855,
      "grad_norm": 94.39252471923828,
      "learning_rate": 8.083253914118395e-06,
      "loss": 4.4369,
      "step": 14232
    },
    {
      "epoch": 0.5746063786838917,
      "grad_norm": 76.65536499023438,
      "learning_rate": 8.081970594578185e-06,
      "loss": 4.5619,
      "step": 14233
    },
    {
      "epoch": 0.5746467501009286,
      "grad_norm": 73.41915130615234,
      "learning_rate": 8.080687307831802e-06,
      "loss": 6.151,
      "step": 14234
    },
    {
      "epoch": 0.5746871215179653,
      "grad_norm": 79.35912322998047,
      "learning_rate": 8.079404053901186e-06,
      "loss": 5.7612,
      "step": 14235
    },
    {
      "epoch": 0.574727492935002,
      "grad_norm": 69.6437759399414,
      "learning_rate": 8.078120832808274e-06,
      "loss": 4.7319,
      "step": 14236
    },
    {
      "epoch": 0.5747678643520387,
      "grad_norm": 234.94601440429688,
      "learning_rate": 8.076837644575005e-06,
      "loss": 5.1901,
      "step": 14237
    },
    {
      "epoch": 0.5748082357690755,
      "grad_norm": 71.5143814086914,
      "learning_rate": 8.075554489223323e-06,
      "loss": 5.3344,
      "step": 14238
    },
    {
      "epoch": 0.5748486071861122,
      "grad_norm": 76.40152740478516,
      "learning_rate": 8.074271366775167e-06,
      "loss": 4.5878,
      "step": 14239
    },
    {
      "epoch": 0.574888978603149,
      "grad_norm": 107.2484359741211,
      "learning_rate": 8.072988277252472e-06,
      "loss": 5.172,
      "step": 14240
    },
    {
      "epoch": 0.5749293500201857,
      "grad_norm": 120.6003189086914,
      "learning_rate": 8.071705220677177e-06,
      "loss": 5.9308,
      "step": 14241
    },
    {
      "epoch": 0.5749697214372225,
      "grad_norm": 91.10556030273438,
      "learning_rate": 8.070422197071221e-06,
      "loss": 4.2987,
      "step": 14242
    },
    {
      "epoch": 0.5750100928542592,
      "grad_norm": 78.678955078125,
      "learning_rate": 8.069139206456543e-06,
      "loss": 6.0145,
      "step": 14243
    },
    {
      "epoch": 0.5750504642712959,
      "grad_norm": 99.88424682617188,
      "learning_rate": 8.06785624885507e-06,
      "loss": 5.5981,
      "step": 14244
    },
    {
      "epoch": 0.5750908356883326,
      "grad_norm": 175.3199920654297,
      "learning_rate": 8.066573324288745e-06,
      "loss": 3.4618,
      "step": 14245
    },
    {
      "epoch": 0.5751312071053694,
      "grad_norm": 124.63243865966797,
      "learning_rate": 8.0652904327795e-06,
      "loss": 4.3758,
      "step": 14246
    },
    {
      "epoch": 0.5751715785224062,
      "grad_norm": 105.78022766113281,
      "learning_rate": 8.064007574349276e-06,
      "loss": 5.8281,
      "step": 14247
    },
    {
      "epoch": 0.5752119499394429,
      "grad_norm": 112.08300018310547,
      "learning_rate": 8.062724749019997e-06,
      "loss": 5.1495,
      "step": 14248
    },
    {
      "epoch": 0.5752523213564796,
      "grad_norm": 102.24267578125,
      "learning_rate": 8.0614419568136e-06,
      "loss": 5.1937,
      "step": 14249
    },
    {
      "epoch": 0.5752926927735164,
      "grad_norm": 140.29092407226562,
      "learning_rate": 8.060159197752021e-06,
      "loss": 5.4366,
      "step": 14250
    },
    {
      "epoch": 0.5753330641905531,
      "grad_norm": 99.10369873046875,
      "learning_rate": 8.058876471857194e-06,
      "loss": 4.9706,
      "step": 14251
    },
    {
      "epoch": 0.5753734356075898,
      "grad_norm": 130.93035888671875,
      "learning_rate": 8.05759377915104e-06,
      "loss": 5.9547,
      "step": 14252
    },
    {
      "epoch": 0.5754138070246265,
      "grad_norm": 103.1306381225586,
      "learning_rate": 8.056311119655497e-06,
      "loss": 5.7637,
      "step": 14253
    },
    {
      "epoch": 0.5754541784416634,
      "grad_norm": 72.35363006591797,
      "learning_rate": 8.055028493392495e-06,
      "loss": 3.541,
      "step": 14254
    },
    {
      "epoch": 0.5754945498587001,
      "grad_norm": 136.14697265625,
      "learning_rate": 8.05374590038397e-06,
      "loss": 4.2339,
      "step": 14255
    },
    {
      "epoch": 0.5755349212757368,
      "grad_norm": 94.26769256591797,
      "learning_rate": 8.05246334065184e-06,
      "loss": 5.651,
      "step": 14256
    },
    {
      "epoch": 0.5755752926927735,
      "grad_norm": 72.81877136230469,
      "learning_rate": 8.051180814218041e-06,
      "loss": 6.9628,
      "step": 14257
    },
    {
      "epoch": 0.5756156641098102,
      "grad_norm": 78.12760162353516,
      "learning_rate": 8.049898321104495e-06,
      "loss": 5.5347,
      "step": 14258
    },
    {
      "epoch": 0.575656035526847,
      "grad_norm": 107.6164321899414,
      "learning_rate": 8.04861586133314e-06,
      "loss": 6.2907,
      "step": 14259
    },
    {
      "epoch": 0.5756964069438837,
      "grad_norm": 139.49734497070312,
      "learning_rate": 8.047333434925893e-06,
      "loss": 3.9216,
      "step": 14260
    },
    {
      "epoch": 0.5757367783609205,
      "grad_norm": 88.16001892089844,
      "learning_rate": 8.046051041904685e-06,
      "loss": 6.4927,
      "step": 14261
    },
    {
      "epoch": 0.5757771497779572,
      "grad_norm": 122.60759735107422,
      "learning_rate": 8.04476868229144e-06,
      "loss": 4.4709,
      "step": 14262
    },
    {
      "epoch": 0.575817521194994,
      "grad_norm": 98.48444366455078,
      "learning_rate": 8.04348635610809e-06,
      "loss": 5.3873,
      "step": 14263
    },
    {
      "epoch": 0.5758578926120307,
      "grad_norm": 96.4197769165039,
      "learning_rate": 8.04220406337655e-06,
      "loss": 4.7058,
      "step": 14264
    },
    {
      "epoch": 0.5758982640290674,
      "grad_norm": 78.75411987304688,
      "learning_rate": 8.040921804118748e-06,
      "loss": 6.4775,
      "step": 14265
    },
    {
      "epoch": 0.5759386354461041,
      "grad_norm": 110.8072280883789,
      "learning_rate": 8.03963957835661e-06,
      "loss": 5.1577,
      "step": 14266
    },
    {
      "epoch": 0.5759790068631409,
      "grad_norm": 102.1366958618164,
      "learning_rate": 8.03835738611206e-06,
      "loss": 4.0698,
      "step": 14267
    },
    {
      "epoch": 0.5760193782801777,
      "grad_norm": 88.39505767822266,
      "learning_rate": 8.037075227407014e-06,
      "loss": 6.9276,
      "step": 14268
    },
    {
      "epoch": 0.5760597496972144,
      "grad_norm": 94.43378448486328,
      "learning_rate": 8.035793102263398e-06,
      "loss": 4.0075,
      "step": 14269
    },
    {
      "epoch": 0.5761001211142511,
      "grad_norm": 82.36434173583984,
      "learning_rate": 8.034511010703135e-06,
      "loss": 4.0893,
      "step": 14270
    },
    {
      "epoch": 0.5761404925312879,
      "grad_norm": 88.60397338867188,
      "learning_rate": 8.033228952748147e-06,
      "loss": 3.8458,
      "step": 14271
    },
    {
      "epoch": 0.5761808639483246,
      "grad_norm": 117.09211730957031,
      "learning_rate": 8.031946928420346e-06,
      "loss": 4.9671,
      "step": 14272
    },
    {
      "epoch": 0.5762212353653613,
      "grad_norm": 101.92245483398438,
      "learning_rate": 8.030664937741659e-06,
      "loss": 7.2396,
      "step": 14273
    },
    {
      "epoch": 0.576261606782398,
      "grad_norm": 89.44193267822266,
      "learning_rate": 8.029382980734e-06,
      "loss": 7.61,
      "step": 14274
    },
    {
      "epoch": 0.5763019781994349,
      "grad_norm": 106.59483337402344,
      "learning_rate": 8.028101057419296e-06,
      "loss": 4.6001,
      "step": 14275
    },
    {
      "epoch": 0.5763423496164716,
      "grad_norm": 132.5879364013672,
      "learning_rate": 8.026819167819454e-06,
      "loss": 3.9024,
      "step": 14276
    },
    {
      "epoch": 0.5763827210335083,
      "grad_norm": 107.63616943359375,
      "learning_rate": 8.025537311956398e-06,
      "loss": 4.4611,
      "step": 14277
    },
    {
      "epoch": 0.576423092450545,
      "grad_norm": 113.69602966308594,
      "learning_rate": 8.024255489852042e-06,
      "loss": 4.7696,
      "step": 14278
    },
    {
      "epoch": 0.5764634638675817,
      "grad_norm": 105.42196655273438,
      "learning_rate": 8.02297370152831e-06,
      "loss": 4.5851,
      "step": 14279
    },
    {
      "epoch": 0.5765038352846185,
      "grad_norm": 95.04429626464844,
      "learning_rate": 8.021691947007104e-06,
      "loss": 6.0531,
      "step": 14280
    },
    {
      "epoch": 0.5765442067016552,
      "grad_norm": 108.93212127685547,
      "learning_rate": 8.020410226310349e-06,
      "loss": 5.3342,
      "step": 14281
    },
    {
      "epoch": 0.576584578118692,
      "grad_norm": 83.84159851074219,
      "learning_rate": 8.019128539459954e-06,
      "loss": 5.7903,
      "step": 14282
    },
    {
      "epoch": 0.5766249495357287,
      "grad_norm": 85.19134521484375,
      "learning_rate": 8.01784688647784e-06,
      "loss": 5.6565,
      "step": 14283
    },
    {
      "epoch": 0.5766653209527655,
      "grad_norm": 100.28845977783203,
      "learning_rate": 8.016565267385912e-06,
      "loss": 3.1901,
      "step": 14284
    },
    {
      "epoch": 0.5767056923698022,
      "grad_norm": 98.63751220703125,
      "learning_rate": 8.015283682206085e-06,
      "loss": 4.3387,
      "step": 14285
    },
    {
      "epoch": 0.5767460637868389,
      "grad_norm": 119.85771942138672,
      "learning_rate": 8.014002130960274e-06,
      "loss": 4.5193,
      "step": 14286
    },
    {
      "epoch": 0.5767864352038756,
      "grad_norm": 87.4246597290039,
      "learning_rate": 8.01272061367039e-06,
      "loss": 4.3275,
      "step": 14287
    },
    {
      "epoch": 0.5768268066209123,
      "grad_norm": 105.44268035888672,
      "learning_rate": 8.011439130358341e-06,
      "loss": 4.4756,
      "step": 14288
    },
    {
      "epoch": 0.5768671780379492,
      "grad_norm": 218.14157104492188,
      "learning_rate": 8.01015768104604e-06,
      "loss": 5.4438,
      "step": 14289
    },
    {
      "epoch": 0.5769075494549859,
      "grad_norm": 84.54051208496094,
      "learning_rate": 8.008876265755396e-06,
      "loss": 5.2556,
      "step": 14290
    },
    {
      "epoch": 0.5769479208720226,
      "grad_norm": 95.76842498779297,
      "learning_rate": 8.00759488450832e-06,
      "loss": 3.7429,
      "step": 14291
    },
    {
      "epoch": 0.5769882922890593,
      "grad_norm": 105.94477081298828,
      "learning_rate": 8.006313537326718e-06,
      "loss": 4.0783,
      "step": 14292
    },
    {
      "epoch": 0.5770286637060961,
      "grad_norm": 97.43492126464844,
      "learning_rate": 8.005032224232497e-06,
      "loss": 4.7716,
      "step": 14293
    },
    {
      "epoch": 0.5770690351231328,
      "grad_norm": 94.79325103759766,
      "learning_rate": 8.003750945247567e-06,
      "loss": 6.104,
      "step": 14294
    },
    {
      "epoch": 0.5771094065401695,
      "grad_norm": 101.6806640625,
      "learning_rate": 8.002469700393836e-06,
      "loss": 4.517,
      "step": 14295
    },
    {
      "epoch": 0.5771497779572063,
      "grad_norm": 119.16390991210938,
      "learning_rate": 8.001188489693207e-06,
      "loss": 3.2501,
      "step": 14296
    },
    {
      "epoch": 0.5771901493742431,
      "grad_norm": 79.89462280273438,
      "learning_rate": 7.999907313167586e-06,
      "loss": 4.252,
      "step": 14297
    },
    {
      "epoch": 0.5772305207912798,
      "grad_norm": 382.8205261230469,
      "learning_rate": 7.998626170838881e-06,
      "loss": 4.9919,
      "step": 14298
    },
    {
      "epoch": 0.5772708922083165,
      "grad_norm": 106.5171127319336,
      "learning_rate": 7.997345062728999e-06,
      "loss": 5.9417,
      "step": 14299
    },
    {
      "epoch": 0.5773112636253532,
      "grad_norm": 86.7855453491211,
      "learning_rate": 7.996063988859834e-06,
      "loss": 5.3127,
      "step": 14300
    },
    {
      "epoch": 0.57735163504239,
      "grad_norm": 65.04493713378906,
      "learning_rate": 7.994782949253298e-06,
      "loss": 6.6159,
      "step": 14301
    },
    {
      "epoch": 0.5773920064594267,
      "grad_norm": 84.8875503540039,
      "learning_rate": 7.993501943931287e-06,
      "loss": 4.0601,
      "step": 14302
    },
    {
      "epoch": 0.5774323778764635,
      "grad_norm": 91.37765502929688,
      "learning_rate": 7.992220972915715e-06,
      "loss": 5.1232,
      "step": 14303
    },
    {
      "epoch": 0.5774727492935002,
      "grad_norm": 81.27813720703125,
      "learning_rate": 7.990940036228471e-06,
      "loss": 2.917,
      "step": 14304
    },
    {
      "epoch": 0.577513120710537,
      "grad_norm": 92.41002655029297,
      "learning_rate": 7.98965913389146e-06,
      "loss": 4.7279,
      "step": 14305
    },
    {
      "epoch": 0.5775534921275737,
      "grad_norm": 67.86443328857422,
      "learning_rate": 7.988378265926584e-06,
      "loss": 4.5583,
      "step": 14306
    },
    {
      "epoch": 0.5775938635446104,
      "grad_norm": 68.61920166015625,
      "learning_rate": 7.987097432355744e-06,
      "loss": 3.2168,
      "step": 14307
    },
    {
      "epoch": 0.5776342349616471,
      "grad_norm": 68.95020294189453,
      "learning_rate": 7.985816633200836e-06,
      "loss": 4.811,
      "step": 14308
    },
    {
      "epoch": 0.5776746063786838,
      "grad_norm": 126.91065216064453,
      "learning_rate": 7.984535868483759e-06,
      "loss": 4.6153,
      "step": 14309
    },
    {
      "epoch": 0.5777149777957207,
      "grad_norm": 84.74212646484375,
      "learning_rate": 7.983255138226413e-06,
      "loss": 5.6899,
      "step": 14310
    },
    {
      "epoch": 0.5777553492127574,
      "grad_norm": 129.78582763671875,
      "learning_rate": 7.981974442450699e-06,
      "loss": 5.7044,
      "step": 14311
    },
    {
      "epoch": 0.5777957206297941,
      "grad_norm": 98.43025970458984,
      "learning_rate": 7.980693781178504e-06,
      "loss": 3.7424,
      "step": 14312
    },
    {
      "epoch": 0.5778360920468308,
      "grad_norm": 79.57593536376953,
      "learning_rate": 7.97941315443173e-06,
      "loss": 5.5886,
      "step": 14313
    },
    {
      "epoch": 0.5778764634638676,
      "grad_norm": 259.478759765625,
      "learning_rate": 7.978132562232274e-06,
      "loss": 5.3789,
      "step": 14314
    },
    {
      "epoch": 0.5779168348809043,
      "grad_norm": 69.52986907958984,
      "learning_rate": 7.976852004602032e-06,
      "loss": 3.5077,
      "step": 14315
    },
    {
      "epoch": 0.577957206297941,
      "grad_norm": 113.03497314453125,
      "learning_rate": 7.975571481562894e-06,
      "loss": 5.8834,
      "step": 14316
    },
    {
      "epoch": 0.5779975777149778,
      "grad_norm": 79.99971771240234,
      "learning_rate": 7.974290993136756e-06,
      "loss": 4.9581,
      "step": 14317
    },
    {
      "epoch": 0.5780379491320146,
      "grad_norm": 117.17984771728516,
      "learning_rate": 7.973010539345512e-06,
      "loss": 6.6506,
      "step": 14318
    },
    {
      "epoch": 0.5780783205490513,
      "grad_norm": 72.1672134399414,
      "learning_rate": 7.971730120211059e-06,
      "loss": 3.6765,
      "step": 14319
    },
    {
      "epoch": 0.578118691966088,
      "grad_norm": 111.9085464477539,
      "learning_rate": 7.97044973575528e-06,
      "loss": 5.6271,
      "step": 14320
    },
    {
      "epoch": 0.5781590633831247,
      "grad_norm": 200.152099609375,
      "learning_rate": 7.96916938600007e-06,
      "loss": 4.1251,
      "step": 14321
    },
    {
      "epoch": 0.5781994348001614,
      "grad_norm": 109.08758544921875,
      "learning_rate": 7.967889070967323e-06,
      "loss": 5.3007,
      "step": 14322
    },
    {
      "epoch": 0.5782398062171982,
      "grad_norm": 86.42539978027344,
      "learning_rate": 7.96660879067893e-06,
      "loss": 6.4947,
      "step": 14323
    },
    {
      "epoch": 0.578280177634235,
      "grad_norm": 79.65348815917969,
      "learning_rate": 7.965328545156774e-06,
      "loss": 6.3535,
      "step": 14324
    },
    {
      "epoch": 0.5783205490512717,
      "grad_norm": 511.07904052734375,
      "learning_rate": 7.96404833442275e-06,
      "loss": 4.2202,
      "step": 14325
    },
    {
      "epoch": 0.5783609204683084,
      "grad_norm": 65.46442413330078,
      "learning_rate": 7.962768158498745e-06,
      "loss": 4.7385,
      "step": 14326
    },
    {
      "epoch": 0.5784012918853452,
      "grad_norm": 102.03077697753906,
      "learning_rate": 7.96148801740665e-06,
      "loss": 5.5506,
      "step": 14327
    },
    {
      "epoch": 0.5784416633023819,
      "grad_norm": 84.22270965576172,
      "learning_rate": 7.960207911168345e-06,
      "loss": 5.8535,
      "step": 14328
    },
    {
      "epoch": 0.5784820347194186,
      "grad_norm": 86.74846649169922,
      "learning_rate": 7.958927839805723e-06,
      "loss": 4.781,
      "step": 14329
    },
    {
      "epoch": 0.5785224061364553,
      "grad_norm": 148.78948974609375,
      "learning_rate": 7.957647803340669e-06,
      "loss": 5.1323,
      "step": 14330
    },
    {
      "epoch": 0.5785627775534922,
      "grad_norm": 97.44962310791016,
      "learning_rate": 7.956367801795071e-06,
      "loss": 4.6865,
      "step": 14331
    },
    {
      "epoch": 0.5786031489705289,
      "grad_norm": 72.16866302490234,
      "learning_rate": 7.955087835190809e-06,
      "loss": 5.6452,
      "step": 14332
    },
    {
      "epoch": 0.5786435203875656,
      "grad_norm": 78.62215423583984,
      "learning_rate": 7.95380790354977e-06,
      "loss": 3.7954,
      "step": 14333
    },
    {
      "epoch": 0.5786838918046023,
      "grad_norm": 105.97381591796875,
      "learning_rate": 7.952528006893836e-06,
      "loss": 4.5247,
      "step": 14334
    },
    {
      "epoch": 0.5787242632216391,
      "grad_norm": 107.41271209716797,
      "learning_rate": 7.951248145244896e-06,
      "loss": 5.6025,
      "step": 14335
    },
    {
      "epoch": 0.5787646346386758,
      "grad_norm": 92.77323150634766,
      "learning_rate": 7.949968318624824e-06,
      "loss": 6.6308,
      "step": 14336
    },
    {
      "epoch": 0.5788050060557125,
      "grad_norm": 88.19944763183594,
      "learning_rate": 7.948688527055509e-06,
      "loss": 6.6984,
      "step": 14337
    },
    {
      "epoch": 0.5788453774727493,
      "grad_norm": 167.70278930664062,
      "learning_rate": 7.947408770558828e-06,
      "loss": 3.5635,
      "step": 14338
    },
    {
      "epoch": 0.5788857488897861,
      "grad_norm": 164.7913818359375,
      "learning_rate": 7.94612904915667e-06,
      "loss": 4.4032,
      "step": 14339
    },
    {
      "epoch": 0.5789261203068228,
      "grad_norm": 127.64081573486328,
      "learning_rate": 7.944849362870903e-06,
      "loss": 6.208,
      "step": 14340
    },
    {
      "epoch": 0.5789664917238595,
      "grad_norm": 76.3467788696289,
      "learning_rate": 7.943569711723413e-06,
      "loss": 4.9453,
      "step": 14341
    },
    {
      "epoch": 0.5790068631408962,
      "grad_norm": 104.49913787841797,
      "learning_rate": 7.942290095736081e-06,
      "loss": 5.0542,
      "step": 14342
    },
    {
      "epoch": 0.5790472345579329,
      "grad_norm": 80.72598266601562,
      "learning_rate": 7.941010514930786e-06,
      "loss": 4.0247,
      "step": 14343
    },
    {
      "epoch": 0.5790876059749698,
      "grad_norm": 81.97234344482422,
      "learning_rate": 7.939730969329397e-06,
      "loss": 5.2774,
      "step": 14344
    },
    {
      "epoch": 0.5791279773920065,
      "grad_norm": 93.15489959716797,
      "learning_rate": 7.9384514589538e-06,
      "loss": 5.854,
      "step": 14345
    },
    {
      "epoch": 0.5791683488090432,
      "grad_norm": 176.90390014648438,
      "learning_rate": 7.93717198382587e-06,
      "loss": 6.3153,
      "step": 14346
    },
    {
      "epoch": 0.5792087202260799,
      "grad_norm": 98.06405639648438,
      "learning_rate": 7.93589254396748e-06,
      "loss": 6.0374,
      "step": 14347
    },
    {
      "epoch": 0.5792490916431167,
      "grad_norm": 91.91270446777344,
      "learning_rate": 7.934613139400509e-06,
      "loss": 4.4011,
      "step": 14348
    },
    {
      "epoch": 0.5792894630601534,
      "grad_norm": 86.44459533691406,
      "learning_rate": 7.93333377014683e-06,
      "loss": 4.4544,
      "step": 14349
    },
    {
      "epoch": 0.5793298344771901,
      "grad_norm": 115.6298828125,
      "learning_rate": 7.932054436228316e-06,
      "loss": 7.1622,
      "step": 14350
    },
    {
      "epoch": 0.5793702058942269,
      "grad_norm": 89.92456817626953,
      "learning_rate": 7.930775137666846e-06,
      "loss": 5.9197,
      "step": 14351
    },
    {
      "epoch": 0.5794105773112637,
      "grad_norm": 137.70968627929688,
      "learning_rate": 7.929495874484286e-06,
      "loss": 5.9114,
      "step": 14352
    },
    {
      "epoch": 0.5794509487283004,
      "grad_norm": 153.0756378173828,
      "learning_rate": 7.928216646702512e-06,
      "loss": 4.7305,
      "step": 14353
    },
    {
      "epoch": 0.5794913201453371,
      "grad_norm": 104.67647552490234,
      "learning_rate": 7.926937454343394e-06,
      "loss": 4.5958,
      "step": 14354
    },
    {
      "epoch": 0.5795316915623738,
      "grad_norm": 110.39131927490234,
      "learning_rate": 7.925658297428807e-06,
      "loss": 6.268,
      "step": 14355
    },
    {
      "epoch": 0.5795720629794106,
      "grad_norm": 80.19385528564453,
      "learning_rate": 7.92437917598062e-06,
      "loss": 5.3632,
      "step": 14356
    },
    {
      "epoch": 0.5796124343964473,
      "grad_norm": 113.1189193725586,
      "learning_rate": 7.923100090020699e-06,
      "loss": 5.9703,
      "step": 14357
    },
    {
      "epoch": 0.5796528058134841,
      "grad_norm": 93.54129028320312,
      "learning_rate": 7.921821039570916e-06,
      "loss": 5.1384,
      "step": 14358
    },
    {
      "epoch": 0.5796931772305208,
      "grad_norm": 113.07707214355469,
      "learning_rate": 7.920542024653146e-06,
      "loss": 4.4987,
      "step": 14359
    },
    {
      "epoch": 0.5797335486475576,
      "grad_norm": 76.23651885986328,
      "learning_rate": 7.919263045289247e-06,
      "loss": 5.3932,
      "step": 14360
    },
    {
      "epoch": 0.5797739200645943,
      "grad_norm": 95.51433563232422,
      "learning_rate": 7.91798410150109e-06,
      "loss": 4.1704,
      "step": 14361
    },
    {
      "epoch": 0.579814291481631,
      "grad_norm": 77.11404418945312,
      "learning_rate": 7.916705193310546e-06,
      "loss": 4.1903,
      "step": 14362
    },
    {
      "epoch": 0.5798546628986677,
      "grad_norm": 122.7613525390625,
      "learning_rate": 7.915426320739478e-06,
      "loss": 6.247,
      "step": 14363
    },
    {
      "epoch": 0.5798950343157044,
      "grad_norm": 128.71087646484375,
      "learning_rate": 7.91414748380975e-06,
      "loss": 4.0859,
      "step": 14364
    },
    {
      "epoch": 0.5799354057327413,
      "grad_norm": 89.26162719726562,
      "learning_rate": 7.912868682543229e-06,
      "loss": 5.9183,
      "step": 14365
    },
    {
      "epoch": 0.579975777149778,
      "grad_norm": 329.7396545410156,
      "learning_rate": 7.911589916961784e-06,
      "loss": 5.8564,
      "step": 14366
    },
    {
      "epoch": 0.5800161485668147,
      "grad_norm": 141.970703125,
      "learning_rate": 7.910311187087274e-06,
      "loss": 6.2432,
      "step": 14367
    },
    {
      "epoch": 0.5800565199838514,
      "grad_norm": 108.2684326171875,
      "learning_rate": 7.909032492941558e-06,
      "loss": 4.2361,
      "step": 14368
    },
    {
      "epoch": 0.5800968914008882,
      "grad_norm": 74.24406433105469,
      "learning_rate": 7.907753834546506e-06,
      "loss": 5.3982,
      "step": 14369
    },
    {
      "epoch": 0.5801372628179249,
      "grad_norm": 95.76866912841797,
      "learning_rate": 7.906475211923977e-06,
      "loss": 3.9335,
      "step": 14370
    },
    {
      "epoch": 0.5801776342349616,
      "grad_norm": 99.96603393554688,
      "learning_rate": 7.905196625095838e-06,
      "loss": 4.2279,
      "step": 14371
    },
    {
      "epoch": 0.5802180056519984,
      "grad_norm": 76.90855407714844,
      "learning_rate": 7.903918074083941e-06,
      "loss": 4.538,
      "step": 14372
    },
    {
      "epoch": 0.5802583770690352,
      "grad_norm": 84.29559326171875,
      "learning_rate": 7.90263955891015e-06,
      "loss": 4.9999,
      "step": 14373
    },
    {
      "epoch": 0.5802987484860719,
      "grad_norm": 88.46665954589844,
      "learning_rate": 7.901361079596323e-06,
      "loss": 5.3607,
      "step": 14374
    },
    {
      "epoch": 0.5803391199031086,
      "grad_norm": 121.91255950927734,
      "learning_rate": 7.900082636164325e-06,
      "loss": 4.8799,
      "step": 14375
    },
    {
      "epoch": 0.5803794913201453,
      "grad_norm": 79.46170806884766,
      "learning_rate": 7.898804228636008e-06,
      "loss": 5.5068,
      "step": 14376
    },
    {
      "epoch": 0.580419862737182,
      "grad_norm": 69.54142761230469,
      "learning_rate": 7.897525857033232e-06,
      "loss": 5.1206,
      "step": 14377
    },
    {
      "epoch": 0.5804602341542188,
      "grad_norm": 72.8326644897461,
      "learning_rate": 7.896247521377855e-06,
      "loss": 5.3646,
      "step": 14378
    },
    {
      "epoch": 0.5805006055712556,
      "grad_norm": 79.62091827392578,
      "learning_rate": 7.894969221691734e-06,
      "loss": 3.8422,
      "step": 14379
    },
    {
      "epoch": 0.5805409769882923,
      "grad_norm": 131.84768676757812,
      "learning_rate": 7.893690957996723e-06,
      "loss": 5.1476,
      "step": 14380
    },
    {
      "epoch": 0.580581348405329,
      "grad_norm": 84.72789001464844,
      "learning_rate": 7.892412730314676e-06,
      "loss": 4.771,
      "step": 14381
    },
    {
      "epoch": 0.5806217198223658,
      "grad_norm": 99.82666015625,
      "learning_rate": 7.89113453866745e-06,
      "loss": 3.6623,
      "step": 14382
    },
    {
      "epoch": 0.5806620912394025,
      "grad_norm": 105.56062316894531,
      "learning_rate": 7.889856383076902e-06,
      "loss": 5.1583,
      "step": 14383
    },
    {
      "epoch": 0.5807024626564392,
      "grad_norm": 70.95851135253906,
      "learning_rate": 7.888578263564878e-06,
      "loss": 5.4505,
      "step": 14384
    },
    {
      "epoch": 0.5807428340734759,
      "grad_norm": 80.47718048095703,
      "learning_rate": 7.887300180153237e-06,
      "loss": 4.7187,
      "step": 14385
    },
    {
      "epoch": 0.5807832054905128,
      "grad_norm": 88.00260925292969,
      "learning_rate": 7.886022132863829e-06,
      "loss": 5.4325,
      "step": 14386
    },
    {
      "epoch": 0.5808235769075495,
      "grad_norm": 164.22525024414062,
      "learning_rate": 7.88474412171851e-06,
      "loss": 5.7698,
      "step": 14387
    },
    {
      "epoch": 0.5808639483245862,
      "grad_norm": 98.68978881835938,
      "learning_rate": 7.883466146739122e-06,
      "loss": 6.1889,
      "step": 14388
    },
    {
      "epoch": 0.5809043197416229,
      "grad_norm": 89.05982971191406,
      "learning_rate": 7.882188207947523e-06,
      "loss": 4.1745,
      "step": 14389
    },
    {
      "epoch": 0.5809446911586597,
      "grad_norm": 149.63539123535156,
      "learning_rate": 7.880910305365559e-06,
      "loss": 6.1597,
      "step": 14390
    },
    {
      "epoch": 0.5809850625756964,
      "grad_norm": 70.09827423095703,
      "learning_rate": 7.879632439015083e-06,
      "loss": 6.6994,
      "step": 14391
    },
    {
      "epoch": 0.5810254339927331,
      "grad_norm": 65.19692993164062,
      "learning_rate": 7.878354608917938e-06,
      "loss": 4.1477,
      "step": 14392
    },
    {
      "epoch": 0.5810658054097699,
      "grad_norm": 100.14749145507812,
      "learning_rate": 7.877076815095975e-06,
      "loss": 5.9997,
      "step": 14393
    },
    {
      "epoch": 0.5811061768268067,
      "grad_norm": 102.43330383300781,
      "learning_rate": 7.875799057571041e-06,
      "loss": 4.8219,
      "step": 14394
    },
    {
      "epoch": 0.5811465482438434,
      "grad_norm": 89.89734649658203,
      "learning_rate": 7.874521336364984e-06,
      "loss": 2.6863,
      "step": 14395
    },
    {
      "epoch": 0.5811869196608801,
      "grad_norm": 131.36038208007812,
      "learning_rate": 7.873243651499647e-06,
      "loss": 4.7064,
      "step": 14396
    },
    {
      "epoch": 0.5812272910779168,
      "grad_norm": 129.50682067871094,
      "learning_rate": 7.871966002996878e-06,
      "loss": 4.7457,
      "step": 14397
    },
    {
      "epoch": 0.5812676624949535,
      "grad_norm": 133.40525817871094,
      "learning_rate": 7.87068839087852e-06,
      "loss": 5.6993,
      "step": 14398
    },
    {
      "epoch": 0.5813080339119903,
      "grad_norm": 69.47955322265625,
      "learning_rate": 7.869410815166425e-06,
      "loss": 4.9189,
      "step": 14399
    },
    {
      "epoch": 0.5813484053290271,
      "grad_norm": 91.00839233398438,
      "learning_rate": 7.868133275882423e-06,
      "loss": 3.8452,
      "step": 14400
    },
    {
      "epoch": 0.5813887767460638,
      "grad_norm": 113.04051971435547,
      "learning_rate": 7.866855773048363e-06,
      "loss": 4.4332,
      "step": 14401
    },
    {
      "epoch": 0.5814291481631005,
      "grad_norm": 110.93218994140625,
      "learning_rate": 7.86557830668609e-06,
      "loss": 4.509,
      "step": 14402
    },
    {
      "epoch": 0.5814695195801373,
      "grad_norm": 131.0320587158203,
      "learning_rate": 7.864300876817445e-06,
      "loss": 5.5354,
      "step": 14403
    },
    {
      "epoch": 0.581509890997174,
      "grad_norm": 124.67058563232422,
      "learning_rate": 7.863023483464265e-06,
      "loss": 6.0263,
      "step": 14404
    },
    {
      "epoch": 0.5815502624142107,
      "grad_norm": 92.19744873046875,
      "learning_rate": 7.861746126648396e-06,
      "loss": 5.6493,
      "step": 14405
    },
    {
      "epoch": 0.5815906338312474,
      "grad_norm": 84.51506042480469,
      "learning_rate": 7.86046880639167e-06,
      "loss": 4.4904,
      "step": 14406
    },
    {
      "epoch": 0.5816310052482843,
      "grad_norm": 98.0565414428711,
      "learning_rate": 7.85919152271594e-06,
      "loss": 4.5761,
      "step": 14407
    },
    {
      "epoch": 0.581671376665321,
      "grad_norm": 252.1560516357422,
      "learning_rate": 7.85791427564303e-06,
      "loss": 5.3804,
      "step": 14408
    },
    {
      "epoch": 0.5817117480823577,
      "grad_norm": 125.16136932373047,
      "learning_rate": 7.856637065194784e-06,
      "loss": 6.0229,
      "step": 14409
    },
    {
      "epoch": 0.5817521194993944,
      "grad_norm": 79.05769348144531,
      "learning_rate": 7.855359891393039e-06,
      "loss": 6.2427,
      "step": 14410
    },
    {
      "epoch": 0.5817924909164311,
      "grad_norm": 89.36296844482422,
      "learning_rate": 7.854082754259637e-06,
      "loss": 5.0834,
      "step": 14411
    },
    {
      "epoch": 0.5818328623334679,
      "grad_norm": 115.9468765258789,
      "learning_rate": 7.852805653816401e-06,
      "loss": 5.9149,
      "step": 14412
    },
    {
      "epoch": 0.5818732337505046,
      "grad_norm": 84.90460205078125,
      "learning_rate": 7.851528590085178e-06,
      "loss": 5.4522,
      "step": 14413
    },
    {
      "epoch": 0.5819136051675414,
      "grad_norm": 101.5234603881836,
      "learning_rate": 7.850251563087799e-06,
      "loss": 4.4849,
      "step": 14414
    },
    {
      "epoch": 0.5819539765845781,
      "grad_norm": 90.19257354736328,
      "learning_rate": 7.848974572846098e-06,
      "loss": 4.7002,
      "step": 14415
    },
    {
      "epoch": 0.5819943480016149,
      "grad_norm": 142.0600128173828,
      "learning_rate": 7.847697619381909e-06,
      "loss": 4.391,
      "step": 14416
    },
    {
      "epoch": 0.5820347194186516,
      "grad_norm": 95.37030029296875,
      "learning_rate": 7.846420702717064e-06,
      "loss": 4.0298,
      "step": 14417
    },
    {
      "epoch": 0.5820750908356883,
      "grad_norm": 74.26290130615234,
      "learning_rate": 7.845143822873394e-06,
      "loss": 6.0692,
      "step": 14418
    },
    {
      "epoch": 0.582115462252725,
      "grad_norm": 79.10891723632812,
      "learning_rate": 7.84386697987274e-06,
      "loss": 6.0137,
      "step": 14419
    },
    {
      "epoch": 0.5821558336697618,
      "grad_norm": 354.77069091796875,
      "learning_rate": 7.84259017373692e-06,
      "loss": 4.9933,
      "step": 14420
    },
    {
      "epoch": 0.5821962050867986,
      "grad_norm": 71.25848388671875,
      "learning_rate": 7.84131340448777e-06,
      "loss": 4.6311,
      "step": 14421
    },
    {
      "epoch": 0.5822365765038353,
      "grad_norm": 133.99530029296875,
      "learning_rate": 7.84003667214712e-06,
      "loss": 2.8966,
      "step": 14422
    },
    {
      "epoch": 0.582276947920872,
      "grad_norm": 67.93777465820312,
      "learning_rate": 7.8387599767368e-06,
      "loss": 3.7694,
      "step": 14423
    },
    {
      "epoch": 0.5823173193379088,
      "grad_norm": 107.01852416992188,
      "learning_rate": 7.837483318278638e-06,
      "loss": 4.8715,
      "step": 14424
    },
    {
      "epoch": 0.5823576907549455,
      "grad_norm": 98.59809875488281,
      "learning_rate": 7.836206696794459e-06,
      "loss": 5.0902,
      "step": 14425
    },
    {
      "epoch": 0.5823980621719822,
      "grad_norm": 108.03570556640625,
      "learning_rate": 7.834930112306093e-06,
      "loss": 5.2039,
      "step": 14426
    },
    {
      "epoch": 0.5824384335890189,
      "grad_norm": 101.77066040039062,
      "learning_rate": 7.833653564835373e-06,
      "loss": 5.9859,
      "step": 14427
    },
    {
      "epoch": 0.5824788050060558,
      "grad_norm": 146.20175170898438,
      "learning_rate": 7.832377054404112e-06,
      "loss": 5.1132,
      "step": 14428
    },
    {
      "epoch": 0.5825191764230925,
      "grad_norm": 114.0842056274414,
      "learning_rate": 7.831100581034141e-06,
      "loss": 4.5829,
      "step": 14429
    },
    {
      "epoch": 0.5825595478401292,
      "grad_norm": 80.01801300048828,
      "learning_rate": 7.829824144747287e-06,
      "loss": 4.353,
      "step": 14430
    },
    {
      "epoch": 0.5825999192571659,
      "grad_norm": 212.4576416015625,
      "learning_rate": 7.828547745565376e-06,
      "loss": 5.074,
      "step": 14431
    },
    {
      "epoch": 0.5826402906742026,
      "grad_norm": 118.69161987304688,
      "learning_rate": 7.827271383510223e-06,
      "loss": 6.6127,
      "step": 14432
    },
    {
      "epoch": 0.5826806620912394,
      "grad_norm": 87.9241943359375,
      "learning_rate": 7.825995058603656e-06,
      "loss": 5.1555,
      "step": 14433
    },
    {
      "epoch": 0.5827210335082761,
      "grad_norm": 83.71077728271484,
      "learning_rate": 7.824718770867496e-06,
      "loss": 6.0485,
      "step": 14434
    },
    {
      "epoch": 0.5827614049253129,
      "grad_norm": 97.82044982910156,
      "learning_rate": 7.82344252032357e-06,
      "loss": 5.3557,
      "step": 14435
    },
    {
      "epoch": 0.5828017763423496,
      "grad_norm": 123.94273376464844,
      "learning_rate": 7.82216630699369e-06,
      "loss": 6.3297,
      "step": 14436
    },
    {
      "epoch": 0.5828421477593864,
      "grad_norm": 115.75383758544922,
      "learning_rate": 7.820890130899681e-06,
      "loss": 4.4035,
      "step": 14437
    },
    {
      "epoch": 0.5828825191764231,
      "grad_norm": 96.96106719970703,
      "learning_rate": 7.819613992063361e-06,
      "loss": 4.8238,
      "step": 14438
    },
    {
      "epoch": 0.5829228905934598,
      "grad_norm": 109.86612701416016,
      "learning_rate": 7.818337890506556e-06,
      "loss": 4.1863,
      "step": 14439
    },
    {
      "epoch": 0.5829632620104965,
      "grad_norm": 99.94560241699219,
      "learning_rate": 7.817061826251075e-06,
      "loss": 4.029,
      "step": 14440
    },
    {
      "epoch": 0.5830036334275333,
      "grad_norm": 112.33824920654297,
      "learning_rate": 7.815785799318737e-06,
      "loss": 3.6205,
      "step": 14441
    },
    {
      "epoch": 0.5830440048445701,
      "grad_norm": 90.39183807373047,
      "learning_rate": 7.814509809731364e-06,
      "loss": 6.7406,
      "step": 14442
    },
    {
      "epoch": 0.5830843762616068,
      "grad_norm": 77.50218963623047,
      "learning_rate": 7.813233857510769e-06,
      "loss": 4.4629,
      "step": 14443
    },
    {
      "epoch": 0.5831247476786435,
      "grad_norm": 89.95510864257812,
      "learning_rate": 7.811957942678769e-06,
      "loss": 4.817,
      "step": 14444
    },
    {
      "epoch": 0.5831651190956803,
      "grad_norm": 122.2560806274414,
      "learning_rate": 7.810682065257177e-06,
      "loss": 5.4997,
      "step": 14445
    },
    {
      "epoch": 0.583205490512717,
      "grad_norm": 204.09661865234375,
      "learning_rate": 7.80940622526781e-06,
      "loss": 4.8014,
      "step": 14446
    },
    {
      "epoch": 0.5832458619297537,
      "grad_norm": 71.41439819335938,
      "learning_rate": 7.808130422732485e-06,
      "loss": 6.1436,
      "step": 14447
    },
    {
      "epoch": 0.5832862333467904,
      "grad_norm": 88.05543518066406,
      "learning_rate": 7.806854657673008e-06,
      "loss": 5.3377,
      "step": 14448
    },
    {
      "epoch": 0.5833266047638273,
      "grad_norm": 160.34561157226562,
      "learning_rate": 7.805578930111194e-06,
      "loss": 4.2157,
      "step": 14449
    },
    {
      "epoch": 0.583366976180864,
      "grad_norm": 94.47062683105469,
      "learning_rate": 7.804303240068855e-06,
      "loss": 5.9906,
      "step": 14450
    },
    {
      "epoch": 0.5834073475979007,
      "grad_norm": 85.63946533203125,
      "learning_rate": 7.803027587567811e-06,
      "loss": 4.8759,
      "step": 14451
    },
    {
      "epoch": 0.5834477190149374,
      "grad_norm": 89.26014709472656,
      "learning_rate": 7.801751972629857e-06,
      "loss": 4.067,
      "step": 14452
    },
    {
      "epoch": 0.5834880904319741,
      "grad_norm": 242.80709838867188,
      "learning_rate": 7.800476395276814e-06,
      "loss": 3.8334,
      "step": 14453
    },
    {
      "epoch": 0.5835284618490109,
      "grad_norm": 70.7723388671875,
      "learning_rate": 7.799200855530487e-06,
      "loss": 5.2413,
      "step": 14454
    },
    {
      "epoch": 0.5835688332660477,
      "grad_norm": 127.4021987915039,
      "learning_rate": 7.797925353412688e-06,
      "loss": 5.7235,
      "step": 14455
    },
    {
      "epoch": 0.5836092046830844,
      "grad_norm": 115.5003662109375,
      "learning_rate": 7.79664988894522e-06,
      "loss": 5.7935,
      "step": 14456
    },
    {
      "epoch": 0.5836495761001211,
      "grad_norm": 95.70030212402344,
      "learning_rate": 7.795374462149896e-06,
      "loss": 5.5327,
      "step": 14457
    },
    {
      "epoch": 0.5836899475171579,
      "grad_norm": 105.10945892333984,
      "learning_rate": 7.794099073048519e-06,
      "loss": 4.8126,
      "step": 14458
    },
    {
      "epoch": 0.5837303189341946,
      "grad_norm": 69.7792739868164,
      "learning_rate": 7.792823721662902e-06,
      "loss": 4.3054,
      "step": 14459
    },
    {
      "epoch": 0.5837706903512313,
      "grad_norm": 146.63343811035156,
      "learning_rate": 7.79154840801484e-06,
      "loss": 5.5807,
      "step": 14460
    },
    {
      "epoch": 0.583811061768268,
      "grad_norm": 98.82347106933594,
      "learning_rate": 7.790273132126144e-06,
      "loss": 5.286,
      "step": 14461
    },
    {
      "epoch": 0.5838514331853049,
      "grad_norm": 89.2551498413086,
      "learning_rate": 7.788997894018617e-06,
      "loss": 6.0653,
      "step": 14462
    },
    {
      "epoch": 0.5838918046023416,
      "grad_norm": 134.03909301757812,
      "learning_rate": 7.787722693714065e-06,
      "loss": 4.3821,
      "step": 14463
    },
    {
      "epoch": 0.5839321760193783,
      "grad_norm": 88.64759826660156,
      "learning_rate": 7.786447531234286e-06,
      "loss": 5.2939,
      "step": 14464
    },
    {
      "epoch": 0.583972547436415,
      "grad_norm": 107.0956802368164,
      "learning_rate": 7.785172406601086e-06,
      "loss": 4.3764,
      "step": 14465
    },
    {
      "epoch": 0.5840129188534517,
      "grad_norm": 81.98994445800781,
      "learning_rate": 7.783897319836267e-06,
      "loss": 5.0417,
      "step": 14466
    },
    {
      "epoch": 0.5840532902704885,
      "grad_norm": 101.81749725341797,
      "learning_rate": 7.782622270961631e-06,
      "loss": 4.4505,
      "step": 14467
    },
    {
      "epoch": 0.5840936616875252,
      "grad_norm": 113.18348693847656,
      "learning_rate": 7.781347259998972e-06,
      "loss": 6.1517,
      "step": 14468
    },
    {
      "epoch": 0.584134033104562,
      "grad_norm": 78.47669982910156,
      "learning_rate": 7.780072286970096e-06,
      "loss": 4.0831,
      "step": 14469
    },
    {
      "epoch": 0.5841744045215987,
      "grad_norm": 99.56507110595703,
      "learning_rate": 7.778797351896798e-06,
      "loss": 5.5183,
      "step": 14470
    },
    {
      "epoch": 0.5842147759386355,
      "grad_norm": 104.21072387695312,
      "learning_rate": 7.777522454800882e-06,
      "loss": 5.4771,
      "step": 14471
    },
    {
      "epoch": 0.5842551473556722,
      "grad_norm": 77.78263854980469,
      "learning_rate": 7.776247595704139e-06,
      "loss": 5.2134,
      "step": 14472
    },
    {
      "epoch": 0.5842955187727089,
      "grad_norm": 98.03120422363281,
      "learning_rate": 7.77497277462837e-06,
      "loss": 6.1268,
      "step": 14473
    },
    {
      "epoch": 0.5843358901897456,
      "grad_norm": 73.39796447753906,
      "learning_rate": 7.77369799159537e-06,
      "loss": 5.5418,
      "step": 14474
    },
    {
      "epoch": 0.5843762616067824,
      "grad_norm": 102.07316589355469,
      "learning_rate": 7.772423246626941e-06,
      "loss": 6.6557,
      "step": 14475
    },
    {
      "epoch": 0.5844166330238192,
      "grad_norm": 118.7978286743164,
      "learning_rate": 7.771148539744868e-06,
      "loss": 4.8229,
      "step": 14476
    },
    {
      "epoch": 0.5844570044408559,
      "grad_norm": 96.02989959716797,
      "learning_rate": 7.76987387097095e-06,
      "loss": 4.4122,
      "step": 14477
    },
    {
      "epoch": 0.5844973758578926,
      "grad_norm": 183.12908935546875,
      "learning_rate": 7.76859924032698e-06,
      "loss": 4.915,
      "step": 14478
    },
    {
      "epoch": 0.5845377472749294,
      "grad_norm": 101.5511474609375,
      "learning_rate": 7.767324647834759e-06,
      "loss": 4.2212,
      "step": 14479
    },
    {
      "epoch": 0.5845781186919661,
      "grad_norm": 122.54898071289062,
      "learning_rate": 7.766050093516066e-06,
      "loss": 4.5105,
      "step": 14480
    },
    {
      "epoch": 0.5846184901090028,
      "grad_norm": 88.36531829833984,
      "learning_rate": 7.764775577392701e-06,
      "loss": 4.8471,
      "step": 14481
    },
    {
      "epoch": 0.5846588615260395,
      "grad_norm": 72.7674331665039,
      "learning_rate": 7.763501099486455e-06,
      "loss": 4.0105,
      "step": 14482
    },
    {
      "epoch": 0.5846992329430764,
      "grad_norm": 116.73246002197266,
      "learning_rate": 7.762226659819118e-06,
      "loss": 6.6577,
      "step": 14483
    },
    {
      "epoch": 0.5847396043601131,
      "grad_norm": 122.90391540527344,
      "learning_rate": 7.760952258412477e-06,
      "loss": 5.1511,
      "step": 14484
    },
    {
      "epoch": 0.5847799757771498,
      "grad_norm": 85.77362823486328,
      "learning_rate": 7.759677895288326e-06,
      "loss": 4.478,
      "step": 14485
    },
    {
      "epoch": 0.5848203471941865,
      "grad_norm": 108.69645690917969,
      "learning_rate": 7.75840357046845e-06,
      "loss": 5.0896,
      "step": 14486
    },
    {
      "epoch": 0.5848607186112232,
      "grad_norm": 105.67031860351562,
      "learning_rate": 7.757129283974643e-06,
      "loss": 5.294,
      "step": 14487
    },
    {
      "epoch": 0.58490109002826,
      "grad_norm": 84.71109008789062,
      "learning_rate": 7.755855035828684e-06,
      "loss": 6.3548,
      "step": 14488
    },
    {
      "epoch": 0.5849414614452967,
      "grad_norm": 81.76960754394531,
      "learning_rate": 7.754580826052362e-06,
      "loss": 4.6361,
      "step": 14489
    },
    {
      "epoch": 0.5849818328623335,
      "grad_norm": 79.85132598876953,
      "learning_rate": 7.753306654667466e-06,
      "loss": 5.1307,
      "step": 14490
    },
    {
      "epoch": 0.5850222042793702,
      "grad_norm": 168.24996948242188,
      "learning_rate": 7.752032521695782e-06,
      "loss": 4.1468,
      "step": 14491
    },
    {
      "epoch": 0.585062575696407,
      "grad_norm": 149.2554473876953,
      "learning_rate": 7.750758427159089e-06,
      "loss": 5.141,
      "step": 14492
    },
    {
      "epoch": 0.5851029471134437,
      "grad_norm": 86.54906463623047,
      "learning_rate": 7.749484371079175e-06,
      "loss": 3.5446,
      "step": 14493
    },
    {
      "epoch": 0.5851433185304804,
      "grad_norm": 101.81163024902344,
      "learning_rate": 7.748210353477824e-06,
      "loss": 6.0264,
      "step": 14494
    },
    {
      "epoch": 0.5851836899475171,
      "grad_norm": 81.48127746582031,
      "learning_rate": 7.746936374376822e-06,
      "loss": 3.0296,
      "step": 14495
    },
    {
      "epoch": 0.5852240613645538,
      "grad_norm": 78.58982849121094,
      "learning_rate": 7.74566243379794e-06,
      "loss": 6.0739,
      "step": 14496
    },
    {
      "epoch": 0.5852644327815907,
      "grad_norm": 101.47064971923828,
      "learning_rate": 7.744388531762968e-06,
      "loss": 5.0366,
      "step": 14497
    },
    {
      "epoch": 0.5853048041986274,
      "grad_norm": 76.0394515991211,
      "learning_rate": 7.743114668293685e-06,
      "loss": 5.4226,
      "step": 14498
    },
    {
      "epoch": 0.5853451756156641,
      "grad_norm": 126.45831298828125,
      "learning_rate": 7.741840843411873e-06,
      "loss": 3.6124,
      "step": 14499
    },
    {
      "epoch": 0.5853855470327008,
      "grad_norm": 149.01080322265625,
      "learning_rate": 7.740567057139308e-06,
      "loss": 5.4776,
      "step": 14500
    },
    {
      "epoch": 0.5854259184497376,
      "grad_norm": 100.03128051757812,
      "learning_rate": 7.73929330949777e-06,
      "loss": 5.0081,
      "step": 14501
    },
    {
      "epoch": 0.5854662898667743,
      "grad_norm": 87.4947509765625,
      "learning_rate": 7.738019600509036e-06,
      "loss": 6.0154,
      "step": 14502
    },
    {
      "epoch": 0.585506661283811,
      "grad_norm": 65.41189575195312,
      "learning_rate": 7.736745930194887e-06,
      "loss": 5.3303,
      "step": 14503
    },
    {
      "epoch": 0.5855470327008478,
      "grad_norm": 75.49872589111328,
      "learning_rate": 7.735472298577095e-06,
      "loss": 5.1232,
      "step": 14504
    },
    {
      "epoch": 0.5855874041178846,
      "grad_norm": 83.0119400024414,
      "learning_rate": 7.734198705677439e-06,
      "loss": 4.5055,
      "step": 14505
    },
    {
      "epoch": 0.5856277755349213,
      "grad_norm": 113.05559539794922,
      "learning_rate": 7.732925151517694e-06,
      "loss": 3.4188,
      "step": 14506
    },
    {
      "epoch": 0.585668146951958,
      "grad_norm": 94.90985870361328,
      "learning_rate": 7.73165163611964e-06,
      "loss": 5.2565,
      "step": 14507
    },
    {
      "epoch": 0.5857085183689947,
      "grad_norm": 107.02871704101562,
      "learning_rate": 7.730378159505041e-06,
      "loss": 3.9209,
      "step": 14508
    },
    {
      "epoch": 0.5857488897860315,
      "grad_norm": 115.25625610351562,
      "learning_rate": 7.729104721695677e-06,
      "loss": 5.7296,
      "step": 14509
    },
    {
      "epoch": 0.5857892612030682,
      "grad_norm": 114.9549789428711,
      "learning_rate": 7.727831322713318e-06,
      "loss": 5.5158,
      "step": 14510
    },
    {
      "epoch": 0.585829632620105,
      "grad_norm": 114.9505844116211,
      "learning_rate": 7.72655796257974e-06,
      "loss": 4.8659,
      "step": 14511
    },
    {
      "epoch": 0.5858700040371417,
      "grad_norm": 209.70484924316406,
      "learning_rate": 7.72528464131671e-06,
      "loss": 4.8455,
      "step": 14512
    },
    {
      "epoch": 0.5859103754541785,
      "grad_norm": 80.46671295166016,
      "learning_rate": 7.724011358946e-06,
      "loss": 3.8495,
      "step": 14513
    },
    {
      "epoch": 0.5859507468712152,
      "grad_norm": 93.47158813476562,
      "learning_rate": 7.722738115489382e-06,
      "loss": 5.8277,
      "step": 14514
    },
    {
      "epoch": 0.5859911182882519,
      "grad_norm": 80.6344223022461,
      "learning_rate": 7.721464910968628e-06,
      "loss": 4.5577,
      "step": 14515
    },
    {
      "epoch": 0.5860314897052886,
      "grad_norm": 98.6525650024414,
      "learning_rate": 7.720191745405499e-06,
      "loss": 4.7164,
      "step": 14516
    },
    {
      "epoch": 0.5860718611223253,
      "grad_norm": 113.7487564086914,
      "learning_rate": 7.718918618821767e-06,
      "loss": 4.0485,
      "step": 14517
    },
    {
      "epoch": 0.5861122325393622,
      "grad_norm": 101.2760009765625,
      "learning_rate": 7.7176455312392e-06,
      "loss": 4.4187,
      "step": 14518
    },
    {
      "epoch": 0.5861526039563989,
      "grad_norm": 72.92000579833984,
      "learning_rate": 7.71637248267957e-06,
      "loss": 4.7026,
      "step": 14519
    },
    {
      "epoch": 0.5861929753734356,
      "grad_norm": 57.85110092163086,
      "learning_rate": 7.715099473164633e-06,
      "loss": 1.6413,
      "step": 14520
    },
    {
      "epoch": 0.5862333467904723,
      "grad_norm": 100.73087310791016,
      "learning_rate": 7.713826502716158e-06,
      "loss": 4.7799,
      "step": 14521
    },
    {
      "epoch": 0.5862737182075091,
      "grad_norm": 163.53851318359375,
      "learning_rate": 7.712553571355913e-06,
      "loss": 4.2189,
      "step": 14522
    },
    {
      "epoch": 0.5863140896245458,
      "grad_norm": 96.94393157958984,
      "learning_rate": 7.711280679105662e-06,
      "loss": 5.2815,
      "step": 14523
    },
    {
      "epoch": 0.5863544610415825,
      "grad_norm": 100.2402572631836,
      "learning_rate": 7.710007825987165e-06,
      "loss": 5.6176,
      "step": 14524
    },
    {
      "epoch": 0.5863948324586193,
      "grad_norm": 112.79244995117188,
      "learning_rate": 7.708735012022186e-06,
      "loss": 4.2728,
      "step": 14525
    },
    {
      "epoch": 0.5864352038756561,
      "grad_norm": 92.74750518798828,
      "learning_rate": 7.70746223723249e-06,
      "loss": 5.9929,
      "step": 14526
    },
    {
      "epoch": 0.5864755752926928,
      "grad_norm": 89.15430450439453,
      "learning_rate": 7.706189501639838e-06,
      "loss": 4.469,
      "step": 14527
    },
    {
      "epoch": 0.5865159467097295,
      "grad_norm": 80.51898193359375,
      "learning_rate": 7.704916805265985e-06,
      "loss": 4.917,
      "step": 14528
    },
    {
      "epoch": 0.5865563181267662,
      "grad_norm": 104.70865631103516,
      "learning_rate": 7.703644148132697e-06,
      "loss": 4.5902,
      "step": 14529
    },
    {
      "epoch": 0.586596689543803,
      "grad_norm": 114.97325897216797,
      "learning_rate": 7.702371530261729e-06,
      "loss": 6.5232,
      "step": 14530
    },
    {
      "epoch": 0.5866370609608397,
      "grad_norm": 83.65159606933594,
      "learning_rate": 7.701098951674846e-06,
      "loss": 5.7696,
      "step": 14531
    },
    {
      "epoch": 0.5866774323778765,
      "grad_norm": 95.43115234375,
      "learning_rate": 7.699826412393801e-06,
      "loss": 4.3373,
      "step": 14532
    },
    {
      "epoch": 0.5867178037949132,
      "grad_norm": 129.81298828125,
      "learning_rate": 7.698553912440353e-06,
      "loss": 5.6007,
      "step": 14533
    },
    {
      "epoch": 0.58675817521195,
      "grad_norm": 71.69422912597656,
      "learning_rate": 7.697281451836258e-06,
      "loss": 6.3611,
      "step": 14534
    },
    {
      "epoch": 0.5867985466289867,
      "grad_norm": 79.52906799316406,
      "learning_rate": 7.696009030603278e-06,
      "loss": 4.7838,
      "step": 14535
    },
    {
      "epoch": 0.5868389180460234,
      "grad_norm": 89.59284210205078,
      "learning_rate": 7.694736648763159e-06,
      "loss": 5.1356,
      "step": 14536
    },
    {
      "epoch": 0.5868792894630601,
      "grad_norm": 94.38672637939453,
      "learning_rate": 7.69346430633766e-06,
      "loss": 5.8056,
      "step": 14537
    },
    {
      "epoch": 0.5869196608800968,
      "grad_norm": 104.32624053955078,
      "learning_rate": 7.692192003348535e-06,
      "loss": 5.5131,
      "step": 14538
    },
    {
      "epoch": 0.5869600322971337,
      "grad_norm": 74.36624908447266,
      "learning_rate": 7.690919739817543e-06,
      "loss": 4.7377,
      "step": 14539
    },
    {
      "epoch": 0.5870004037141704,
      "grad_norm": 80.72950744628906,
      "learning_rate": 7.689647515766426e-06,
      "loss": 5.3846,
      "step": 14540
    },
    {
      "epoch": 0.5870407751312071,
      "grad_norm": 110.78743743896484,
      "learning_rate": 7.688375331216942e-06,
      "loss": 5.571,
      "step": 14541
    },
    {
      "epoch": 0.5870811465482438,
      "grad_norm": 110.09901428222656,
      "learning_rate": 7.68710318619084e-06,
      "loss": 4.602,
      "step": 14542
    },
    {
      "epoch": 0.5871215179652806,
      "grad_norm": 74.74961853027344,
      "learning_rate": 7.685831080709877e-06,
      "loss": 4.5784,
      "step": 14543
    },
    {
      "epoch": 0.5871618893823173,
      "grad_norm": 95.58615112304688,
      "learning_rate": 7.684559014795794e-06,
      "loss": 4.0016,
      "step": 14544
    },
    {
      "epoch": 0.587202260799354,
      "grad_norm": 89.51258850097656,
      "learning_rate": 7.683286988470348e-06,
      "loss": 3.781,
      "step": 14545
    },
    {
      "epoch": 0.5872426322163908,
      "grad_norm": 134.41941833496094,
      "learning_rate": 7.682015001755282e-06,
      "loss": 5.4571,
      "step": 14546
    },
    {
      "epoch": 0.5872830036334276,
      "grad_norm": 90.20084381103516,
      "learning_rate": 7.680743054672353e-06,
      "loss": 5.3887,
      "step": 14547
    },
    {
      "epoch": 0.5873233750504643,
      "grad_norm": 116.54820251464844,
      "learning_rate": 7.679471147243295e-06,
      "loss": 6.0088,
      "step": 14548
    },
    {
      "epoch": 0.587363746467501,
      "grad_norm": 103.9380874633789,
      "learning_rate": 7.678199279489864e-06,
      "loss": 5.5028,
      "step": 14549
    },
    {
      "epoch": 0.5874041178845377,
      "grad_norm": 85.5473861694336,
      "learning_rate": 7.676927451433802e-06,
      "loss": 5.0435,
      "step": 14550
    },
    {
      "epoch": 0.5874444893015744,
      "grad_norm": 147.3457794189453,
      "learning_rate": 7.67565566309686e-06,
      "loss": 5.2204,
      "step": 14551
    },
    {
      "epoch": 0.5874848607186112,
      "grad_norm": 132.51683044433594,
      "learning_rate": 7.674383914500774e-06,
      "loss": 5.4055,
      "step": 14552
    },
    {
      "epoch": 0.587525232135648,
      "grad_norm": 137.2542724609375,
      "learning_rate": 7.673112205667293e-06,
      "loss": 4.6949,
      "step": 14553
    },
    {
      "epoch": 0.5875656035526847,
      "grad_norm": 92.91556549072266,
      "learning_rate": 7.67184053661816e-06,
      "loss": 4.7064,
      "step": 14554
    },
    {
      "epoch": 0.5876059749697214,
      "grad_norm": 150.06544494628906,
      "learning_rate": 7.67056890737512e-06,
      "loss": 4.0092,
      "step": 14555
    },
    {
      "epoch": 0.5876463463867582,
      "grad_norm": 118.60045623779297,
      "learning_rate": 7.66929731795991e-06,
      "loss": 4.7246,
      "step": 14556
    },
    {
      "epoch": 0.5876867178037949,
      "grad_norm": 134.18496704101562,
      "learning_rate": 7.668025768394271e-06,
      "loss": 3.2341,
      "step": 14557
    },
    {
      "epoch": 0.5877270892208316,
      "grad_norm": 89.15679931640625,
      "learning_rate": 7.666754258699947e-06,
      "loss": 5.9364,
      "step": 14558
    },
    {
      "epoch": 0.5877674606378684,
      "grad_norm": 136.66485595703125,
      "learning_rate": 7.665482788898681e-06,
      "loss": 5.009,
      "step": 14559
    },
    {
      "epoch": 0.5878078320549052,
      "grad_norm": 140.08645629882812,
      "learning_rate": 7.664211359012205e-06,
      "loss": 5.3771,
      "step": 14560
    },
    {
      "epoch": 0.5878482034719419,
      "grad_norm": 70.65520477294922,
      "learning_rate": 7.662939969062259e-06,
      "loss": 4.503,
      "step": 14561
    },
    {
      "epoch": 0.5878885748889786,
      "grad_norm": 150.12628173828125,
      "learning_rate": 7.661668619070581e-06,
      "loss": 4.8279,
      "step": 14562
    },
    {
      "epoch": 0.5879289463060153,
      "grad_norm": 94.63032531738281,
      "learning_rate": 7.660397309058913e-06,
      "loss": 6.6173,
      "step": 14563
    },
    {
      "epoch": 0.587969317723052,
      "grad_norm": 68.37301635742188,
      "learning_rate": 7.659126039048985e-06,
      "loss": 5.9497,
      "step": 14564
    },
    {
      "epoch": 0.5880096891400888,
      "grad_norm": 89.59693145751953,
      "learning_rate": 7.657854809062535e-06,
      "loss": 5.41,
      "step": 14565
    },
    {
      "epoch": 0.5880500605571256,
      "grad_norm": 96.66801452636719,
      "learning_rate": 7.656583619121298e-06,
      "loss": 6.4165,
      "step": 14566
    },
    {
      "epoch": 0.5880904319741623,
      "grad_norm": 119.12200927734375,
      "learning_rate": 7.655312469247014e-06,
      "loss": 6.5041,
      "step": 14567
    },
    {
      "epoch": 0.588130803391199,
      "grad_norm": 771.540771484375,
      "learning_rate": 7.654041359461406e-06,
      "loss": 5.1645,
      "step": 14568
    },
    {
      "epoch": 0.5881711748082358,
      "grad_norm": 152.04507446289062,
      "learning_rate": 7.652770289786214e-06,
      "loss": 5.1015,
      "step": 14569
    },
    {
      "epoch": 0.5882115462252725,
      "grad_norm": 92.57876586914062,
      "learning_rate": 7.651499260243169e-06,
      "loss": 4.6859,
      "step": 14570
    },
    {
      "epoch": 0.5882519176423092,
      "grad_norm": 95.47840118408203,
      "learning_rate": 7.650228270854004e-06,
      "loss": 4.9048,
      "step": 14571
    },
    {
      "epoch": 0.5882922890593459,
      "grad_norm": 173.2707977294922,
      "learning_rate": 7.648957321640446e-06,
      "loss": 4.651,
      "step": 14572
    },
    {
      "epoch": 0.5883326604763828,
      "grad_norm": 77.84302520751953,
      "learning_rate": 7.64768641262423e-06,
      "loss": 5.2949,
      "step": 14573
    },
    {
      "epoch": 0.5883730318934195,
      "grad_norm": 90.1027603149414,
      "learning_rate": 7.64641554382708e-06,
      "loss": 6.0733,
      "step": 14574
    },
    {
      "epoch": 0.5884134033104562,
      "grad_norm": 139.24160766601562,
      "learning_rate": 7.645144715270734e-06,
      "loss": 5.3722,
      "step": 14575
    },
    {
      "epoch": 0.5884537747274929,
      "grad_norm": 88.84114837646484,
      "learning_rate": 7.64387392697691e-06,
      "loss": 5.6742,
      "step": 14576
    },
    {
      "epoch": 0.5884941461445297,
      "grad_norm": 100.69548034667969,
      "learning_rate": 7.64260317896734e-06,
      "loss": 5.5519,
      "step": 14577
    },
    {
      "epoch": 0.5885345175615664,
      "grad_norm": 107.34739685058594,
      "learning_rate": 7.641332471263751e-06,
      "loss": 5.8311,
      "step": 14578
    },
    {
      "epoch": 0.5885748889786031,
      "grad_norm": 95.75117492675781,
      "learning_rate": 7.64006180388787e-06,
      "loss": 5.7531,
      "step": 14579
    },
    {
      "epoch": 0.5886152603956399,
      "grad_norm": 92.8011245727539,
      "learning_rate": 7.63879117686142e-06,
      "loss": 6.2189,
      "step": 14580
    },
    {
      "epoch": 0.5886556318126767,
      "grad_norm": 100.94664001464844,
      "learning_rate": 7.637520590206126e-06,
      "loss": 4.2345,
      "step": 14581
    },
    {
      "epoch": 0.5886960032297134,
      "grad_norm": 78.46546173095703,
      "learning_rate": 7.636250043943715e-06,
      "loss": 5.8523,
      "step": 14582
    },
    {
      "epoch": 0.5887363746467501,
      "grad_norm": 86.81085968017578,
      "learning_rate": 7.634979538095913e-06,
      "loss": 4.1848,
      "step": 14583
    },
    {
      "epoch": 0.5887767460637868,
      "grad_norm": 60.6374397277832,
      "learning_rate": 7.633709072684432e-06,
      "loss": 1.9425,
      "step": 14584
    },
    {
      "epoch": 0.5888171174808235,
      "grad_norm": 133.0304412841797,
      "learning_rate": 7.632438647731e-06,
      "loss": 5.0093,
      "step": 14585
    },
    {
      "epoch": 0.5888574888978603,
      "grad_norm": 85.91999053955078,
      "learning_rate": 7.63116826325734e-06,
      "loss": 5.0751,
      "step": 14586
    },
    {
      "epoch": 0.5888978603148971,
      "grad_norm": 144.1476287841797,
      "learning_rate": 7.629897919285174e-06,
      "loss": 5.7517,
      "step": 14587
    },
    {
      "epoch": 0.5889382317319338,
      "grad_norm": 110.18673706054688,
      "learning_rate": 7.628627615836216e-06,
      "loss": 6.2154,
      "step": 14588
    },
    {
      "epoch": 0.5889786031489705,
      "grad_norm": 106.53900146484375,
      "learning_rate": 7.627357352932187e-06,
      "loss": 5.7865,
      "step": 14589
    },
    {
      "epoch": 0.5890189745660073,
      "grad_norm": 93.97234344482422,
      "learning_rate": 7.626087130594808e-06,
      "loss": 2.3781,
      "step": 14590
    },
    {
      "epoch": 0.589059345983044,
      "grad_norm": 97.88848114013672,
      "learning_rate": 7.624816948845797e-06,
      "loss": 4.7385,
      "step": 14591
    },
    {
      "epoch": 0.5890997174000807,
      "grad_norm": 95.5184555053711,
      "learning_rate": 7.623546807706868e-06,
      "loss": 4.3402,
      "step": 14592
    },
    {
      "epoch": 0.5891400888171174,
      "grad_norm": 84.07974243164062,
      "learning_rate": 7.622276707199738e-06,
      "loss": 4.9533,
      "step": 14593
    },
    {
      "epoch": 0.5891804602341543,
      "grad_norm": 216.05593872070312,
      "learning_rate": 7.621006647346125e-06,
      "loss": 5.2428,
      "step": 14594
    },
    {
      "epoch": 0.589220831651191,
      "grad_norm": 91.93692016601562,
      "learning_rate": 7.619736628167747e-06,
      "loss": 5.3451,
      "step": 14595
    },
    {
      "epoch": 0.5892612030682277,
      "grad_norm": 130.0390167236328,
      "learning_rate": 7.618466649686311e-06,
      "loss": 4.0084,
      "step": 14596
    },
    {
      "epoch": 0.5893015744852644,
      "grad_norm": 124.3607177734375,
      "learning_rate": 7.617196711923532e-06,
      "loss": 5.9123,
      "step": 14597
    },
    {
      "epoch": 0.5893419459023012,
      "grad_norm": 119.68183135986328,
      "learning_rate": 7.6159268149011265e-06,
      "loss": 5.5013,
      "step": 14598
    },
    {
      "epoch": 0.5893823173193379,
      "grad_norm": 109.00935363769531,
      "learning_rate": 7.614656958640807e-06,
      "loss": 5.739,
      "step": 14599
    },
    {
      "epoch": 0.5894226887363746,
      "grad_norm": 100.8873062133789,
      "learning_rate": 7.613387143164281e-06,
      "loss": 4.4293,
      "step": 14600
    },
    {
      "epoch": 0.5894630601534114,
      "grad_norm": 110.9452896118164,
      "learning_rate": 7.612117368493262e-06,
      "loss": 4.771,
      "step": 14601
    },
    {
      "epoch": 0.5895034315704482,
      "grad_norm": 100.07099151611328,
      "learning_rate": 7.610847634649459e-06,
      "loss": 5.7082,
      "step": 14602
    },
    {
      "epoch": 0.5895438029874849,
      "grad_norm": 97.39353942871094,
      "learning_rate": 7.609577941654588e-06,
      "loss": 3.3039,
      "step": 14603
    },
    {
      "epoch": 0.5895841744045216,
      "grad_norm": 69.12229919433594,
      "learning_rate": 7.608308289530347e-06,
      "loss": 2.9105,
      "step": 14604
    },
    {
      "epoch": 0.5896245458215583,
      "grad_norm": 106.51395416259766,
      "learning_rate": 7.60703867829845e-06,
      "loss": 4.6816,
      "step": 14605
    },
    {
      "epoch": 0.589664917238595,
      "grad_norm": 73.81824493408203,
      "learning_rate": 7.605769107980603e-06,
      "loss": 4.3192,
      "step": 14606
    },
    {
      "epoch": 0.5897052886556318,
      "grad_norm": 90.7006607055664,
      "learning_rate": 7.6044995785985174e-06,
      "loss": 6.2731,
      "step": 14607
    },
    {
      "epoch": 0.5897456600726686,
      "grad_norm": 67.95521545410156,
      "learning_rate": 7.603230090173891e-06,
      "loss": 5.2364,
      "step": 14608
    },
    {
      "epoch": 0.5897860314897053,
      "grad_norm": 98.57759094238281,
      "learning_rate": 7.6019606427284335e-06,
      "loss": 3.5713,
      "step": 14609
    },
    {
      "epoch": 0.589826402906742,
      "grad_norm": 85.74556732177734,
      "learning_rate": 7.600691236283849e-06,
      "loss": 5.6129,
      "step": 14610
    },
    {
      "epoch": 0.5898667743237788,
      "grad_norm": 91.47005462646484,
      "learning_rate": 7.599421870861844e-06,
      "loss": 4.2452,
      "step": 14611
    },
    {
      "epoch": 0.5899071457408155,
      "grad_norm": 103.00125885009766,
      "learning_rate": 7.5981525464841166e-06,
      "loss": 5.1705,
      "step": 14612
    },
    {
      "epoch": 0.5899475171578522,
      "grad_norm": 105.5557632446289,
      "learning_rate": 7.596883263172372e-06,
      "loss": 3.9282,
      "step": 14613
    },
    {
      "epoch": 0.5899878885748889,
      "grad_norm": 108.34666442871094,
      "learning_rate": 7.595614020948312e-06,
      "loss": 3.4442,
      "step": 14614
    },
    {
      "epoch": 0.5900282599919258,
      "grad_norm": 78.79293823242188,
      "learning_rate": 7.5943448198336425e-06,
      "loss": 5.9718,
      "step": 14615
    },
    {
      "epoch": 0.5900686314089625,
      "grad_norm": 94.92495727539062,
      "learning_rate": 7.593075659850054e-06,
      "loss": 5.1034,
      "step": 14616
    },
    {
      "epoch": 0.5901090028259992,
      "grad_norm": 229.78858947753906,
      "learning_rate": 7.591806541019252e-06,
      "loss": 4.1352,
      "step": 14617
    },
    {
      "epoch": 0.5901493742430359,
      "grad_norm": 125.36241912841797,
      "learning_rate": 7.590537463362934e-06,
      "loss": 5.1232,
      "step": 14618
    },
    {
      "epoch": 0.5901897456600727,
      "grad_norm": 70.56281280517578,
      "learning_rate": 7.589268426902801e-06,
      "loss": 3.8479,
      "step": 14619
    },
    {
      "epoch": 0.5902301170771094,
      "grad_norm": 78.28890228271484,
      "learning_rate": 7.587999431660546e-06,
      "loss": 5.4697,
      "step": 14620
    },
    {
      "epoch": 0.5902704884941461,
      "grad_norm": 96.98986053466797,
      "learning_rate": 7.5867304776578685e-06,
      "loss": 5.6431,
      "step": 14621
    },
    {
      "epoch": 0.5903108599111829,
      "grad_norm": 80.3880844116211,
      "learning_rate": 7.585461564916465e-06,
      "loss": 5.1729,
      "step": 14622
    },
    {
      "epoch": 0.5903512313282196,
      "grad_norm": 181.6608428955078,
      "learning_rate": 7.584192693458034e-06,
      "loss": 6.156,
      "step": 14623
    },
    {
      "epoch": 0.5903916027452564,
      "grad_norm": 127.8050537109375,
      "learning_rate": 7.582923863304262e-06,
      "loss": 2.474,
      "step": 14624
    },
    {
      "epoch": 0.5904319741622931,
      "grad_norm": 69.21495056152344,
      "learning_rate": 7.581655074476848e-06,
      "loss": 4.7431,
      "step": 14625
    },
    {
      "epoch": 0.5904723455793298,
      "grad_norm": 75.68902587890625,
      "learning_rate": 7.5803863269974855e-06,
      "loss": 4.9693,
      "step": 14626
    },
    {
      "epoch": 0.5905127169963665,
      "grad_norm": 967.5408325195312,
      "learning_rate": 7.57911762088787e-06,
      "loss": 4.8118,
      "step": 14627
    },
    {
      "epoch": 0.5905530884134033,
      "grad_norm": 164.49514770507812,
      "learning_rate": 7.577848956169687e-06,
      "loss": 3.2652,
      "step": 14628
    },
    {
      "epoch": 0.5905934598304401,
      "grad_norm": 143.22128295898438,
      "learning_rate": 7.576580332864631e-06,
      "loss": 5.5069,
      "step": 14629
    },
    {
      "epoch": 0.5906338312474768,
      "grad_norm": 92.16849517822266,
      "learning_rate": 7.575311750994389e-06,
      "loss": 4.972,
      "step": 14630
    },
    {
      "epoch": 0.5906742026645135,
      "grad_norm": 120.11798858642578,
      "learning_rate": 7.574043210580659e-06,
      "loss": 4.1103,
      "step": 14631
    },
    {
      "epoch": 0.5907145740815503,
      "grad_norm": 79.68256378173828,
      "learning_rate": 7.572774711645123e-06,
      "loss": 4.5095,
      "step": 14632
    },
    {
      "epoch": 0.590754945498587,
      "grad_norm": 58.994693756103516,
      "learning_rate": 7.571506254209471e-06,
      "loss": 5.4661,
      "step": 14633
    },
    {
      "epoch": 0.5907953169156237,
      "grad_norm": 83.00118255615234,
      "learning_rate": 7.570237838295391e-06,
      "loss": 2.7817,
      "step": 14634
    },
    {
      "epoch": 0.5908356883326604,
      "grad_norm": 118.05978393554688,
      "learning_rate": 7.5689694639245735e-06,
      "loss": 4.6288,
      "step": 14635
    },
    {
      "epoch": 0.5908760597496973,
      "grad_norm": 195.91055297851562,
      "learning_rate": 7.5677011311186985e-06,
      "loss": 3.8063,
      "step": 14636
    },
    {
      "epoch": 0.590916431166734,
      "grad_norm": 160.24539184570312,
      "learning_rate": 7.566432839899454e-06,
      "loss": 4.3819,
      "step": 14637
    },
    {
      "epoch": 0.5909568025837707,
      "grad_norm": 85.19376373291016,
      "learning_rate": 7.565164590288524e-06,
      "loss": 6.1519,
      "step": 14638
    },
    {
      "epoch": 0.5909971740008074,
      "grad_norm": 89.27011108398438,
      "learning_rate": 7.563896382307596e-06,
      "loss": 5.1908,
      "step": 14639
    },
    {
      "epoch": 0.5910375454178441,
      "grad_norm": 132.00064086914062,
      "learning_rate": 7.562628215978351e-06,
      "loss": 5.0881,
      "step": 14640
    },
    {
      "epoch": 0.5910779168348809,
      "grad_norm": 77.98478698730469,
      "learning_rate": 7.561360091322469e-06,
      "loss": 3.1984,
      "step": 14641
    },
    {
      "epoch": 0.5911182882519176,
      "grad_norm": 83.58161163330078,
      "learning_rate": 7.560092008361637e-06,
      "loss": 5.3123,
      "step": 14642
    },
    {
      "epoch": 0.5911586596689544,
      "grad_norm": 62.65145492553711,
      "learning_rate": 7.5588239671175374e-06,
      "loss": 5.2401,
      "step": 14643
    },
    {
      "epoch": 0.5911990310859911,
      "grad_norm": 248.1136016845703,
      "learning_rate": 7.5575559676118436e-06,
      "loss": 4.9215,
      "step": 14644
    },
    {
      "epoch": 0.5912394025030279,
      "grad_norm": 108.76201629638672,
      "learning_rate": 7.556288009866239e-06,
      "loss": 4.5645,
      "step": 14645
    },
    {
      "epoch": 0.5912797739200646,
      "grad_norm": 101.66602325439453,
      "learning_rate": 7.555020093902403e-06,
      "loss": 5.5521,
      "step": 14646
    },
    {
      "epoch": 0.5913201453371013,
      "grad_norm": 161.2411651611328,
      "learning_rate": 7.553752219742019e-06,
      "loss": 4.334,
      "step": 14647
    },
    {
      "epoch": 0.591360516754138,
      "grad_norm": 126.04740905761719,
      "learning_rate": 7.552484387406753e-06,
      "loss": 7.0357,
      "step": 14648
    },
    {
      "epoch": 0.5914008881711748,
      "grad_norm": 107.87250518798828,
      "learning_rate": 7.5512165969182916e-06,
      "loss": 5.512,
      "step": 14649
    },
    {
      "epoch": 0.5914412595882116,
      "grad_norm": 62.810726165771484,
      "learning_rate": 7.549948848298306e-06,
      "loss": 4.9926,
      "step": 14650
    },
    {
      "epoch": 0.5914816310052483,
      "grad_norm": 96.53567504882812,
      "learning_rate": 7.548681141568477e-06,
      "loss": 4.7178,
      "step": 14651
    },
    {
      "epoch": 0.591522002422285,
      "grad_norm": 94.41651916503906,
      "learning_rate": 7.547413476750474e-06,
      "loss": 5.4424,
      "step": 14652
    },
    {
      "epoch": 0.5915623738393218,
      "grad_norm": 164.22677612304688,
      "learning_rate": 7.546145853865972e-06,
      "loss": 6.3609,
      "step": 14653
    },
    {
      "epoch": 0.5916027452563585,
      "grad_norm": 110.49669647216797,
      "learning_rate": 7.544878272936647e-06,
      "loss": 5.6292,
      "step": 14654
    },
    {
      "epoch": 0.5916431166733952,
      "grad_norm": 89.63702392578125,
      "learning_rate": 7.543610733984174e-06,
      "loss": 3.9448,
      "step": 14655
    },
    {
      "epoch": 0.5916834880904319,
      "grad_norm": 97.15544128417969,
      "learning_rate": 7.5423432370302186e-06,
      "loss": 6.2099,
      "step": 14656
    },
    {
      "epoch": 0.5917238595074688,
      "grad_norm": 86.15923309326172,
      "learning_rate": 7.5410757820964535e-06,
      "loss": 5.0391,
      "step": 14657
    },
    {
      "epoch": 0.5917642309245055,
      "grad_norm": 101.60879516601562,
      "learning_rate": 7.539808369204551e-06,
      "loss": 5.1179,
      "step": 14658
    },
    {
      "epoch": 0.5918046023415422,
      "grad_norm": 71.73849487304688,
      "learning_rate": 7.5385409983761815e-06,
      "loss": 4.6781,
      "step": 14659
    },
    {
      "epoch": 0.5918449737585789,
      "grad_norm": 107.00177001953125,
      "learning_rate": 7.537273669633011e-06,
      "loss": 4.7421,
      "step": 14660
    },
    {
      "epoch": 0.5918853451756156,
      "grad_norm": 65.57695007324219,
      "learning_rate": 7.536006382996711e-06,
      "loss": 3.818,
      "step": 14661
    },
    {
      "epoch": 0.5919257165926524,
      "grad_norm": 76.32087707519531,
      "learning_rate": 7.534739138488948e-06,
      "loss": 4.2387,
      "step": 14662
    },
    {
      "epoch": 0.5919660880096892,
      "grad_norm": 89.16803741455078,
      "learning_rate": 7.533471936131393e-06,
      "loss": 4.6165,
      "step": 14663
    },
    {
      "epoch": 0.5920064594267259,
      "grad_norm": 96.14175415039062,
      "learning_rate": 7.532204775945704e-06,
      "loss": 5.6123,
      "step": 14664
    },
    {
      "epoch": 0.5920468308437626,
      "grad_norm": 72.07460021972656,
      "learning_rate": 7.53093765795355e-06,
      "loss": 3.3661,
      "step": 14665
    },
    {
      "epoch": 0.5920872022607994,
      "grad_norm": 291.9224548339844,
      "learning_rate": 7.529670582176597e-06,
      "loss": 5.5414,
      "step": 14666
    },
    {
      "epoch": 0.5921275736778361,
      "grad_norm": 96.65837860107422,
      "learning_rate": 7.528403548636509e-06,
      "loss": 4.9707,
      "step": 14667
    },
    {
      "epoch": 0.5921679450948728,
      "grad_norm": 111.53782653808594,
      "learning_rate": 7.527136557354951e-06,
      "loss": 5.3925,
      "step": 14668
    },
    {
      "epoch": 0.5922083165119095,
      "grad_norm": 126.33419799804688,
      "learning_rate": 7.52586960835358e-06,
      "loss": 5.2916,
      "step": 14669
    },
    {
      "epoch": 0.5922486879289464,
      "grad_norm": 126.80782318115234,
      "learning_rate": 7.524602701654061e-06,
      "loss": 4.9843,
      "step": 14670
    },
    {
      "epoch": 0.5922890593459831,
      "grad_norm": 89.68549346923828,
      "learning_rate": 7.523335837278059e-06,
      "loss": 4.1469,
      "step": 14671
    },
    {
      "epoch": 0.5923294307630198,
      "grad_norm": 82.99425506591797,
      "learning_rate": 7.522069015247226e-06,
      "loss": 3.8071,
      "step": 14672
    },
    {
      "epoch": 0.5923698021800565,
      "grad_norm": 91.80238342285156,
      "learning_rate": 7.520802235583228e-06,
      "loss": 5.8809,
      "step": 14673
    },
    {
      "epoch": 0.5924101735970932,
      "grad_norm": 99.32718658447266,
      "learning_rate": 7.5195354983077215e-06,
      "loss": 5.6944,
      "step": 14674
    },
    {
      "epoch": 0.59245054501413,
      "grad_norm": 122.4290542602539,
      "learning_rate": 7.5182688034423715e-06,
      "loss": 4.3142,
      "step": 14675
    },
    {
      "epoch": 0.5924909164311667,
      "grad_norm": 366.2208557128906,
      "learning_rate": 7.517002151008825e-06,
      "loss": 2.9683,
      "step": 14676
    },
    {
      "epoch": 0.5925312878482035,
      "grad_norm": 119.217041015625,
      "learning_rate": 7.515735541028742e-06,
      "loss": 5.9623,
      "step": 14677
    },
    {
      "epoch": 0.5925716592652402,
      "grad_norm": 110.9153823852539,
      "learning_rate": 7.514468973523782e-06,
      "loss": 5.2626,
      "step": 14678
    },
    {
      "epoch": 0.592612030682277,
      "grad_norm": 116.4285659790039,
      "learning_rate": 7.5132024485156e-06,
      "loss": 3.7886,
      "step": 14679
    },
    {
      "epoch": 0.5926524020993137,
      "grad_norm": 105.46553802490234,
      "learning_rate": 7.511935966025846e-06,
      "loss": 4.6372,
      "step": 14680
    },
    {
      "epoch": 0.5926927735163504,
      "grad_norm": 281.83453369140625,
      "learning_rate": 7.510669526076178e-06,
      "loss": 5.899,
      "step": 14681
    },
    {
      "epoch": 0.5927331449333871,
      "grad_norm": 96.21363830566406,
      "learning_rate": 7.509403128688248e-06,
      "loss": 5.7084,
      "step": 14682
    },
    {
      "epoch": 0.5927735163504239,
      "grad_norm": 153.07945251464844,
      "learning_rate": 7.508136773883707e-06,
      "loss": 5.8848,
      "step": 14683
    },
    {
      "epoch": 0.5928138877674607,
      "grad_norm": 157.92318725585938,
      "learning_rate": 7.506870461684215e-06,
      "loss": 5.4706,
      "step": 14684
    },
    {
      "epoch": 0.5928542591844974,
      "grad_norm": 77.6312026977539,
      "learning_rate": 7.505604192111411e-06,
      "loss": 3.6291,
      "step": 14685
    },
    {
      "epoch": 0.5928946306015341,
      "grad_norm": 91.80897521972656,
      "learning_rate": 7.504337965186949e-06,
      "loss": 4.1701,
      "step": 14686
    },
    {
      "epoch": 0.5929350020185709,
      "grad_norm": 121.99574279785156,
      "learning_rate": 7.503071780932482e-06,
      "loss": 4.5613,
      "step": 14687
    },
    {
      "epoch": 0.5929753734356076,
      "grad_norm": 104.26856231689453,
      "learning_rate": 7.5018056393696584e-06,
      "loss": 5.5812,
      "step": 14688
    },
    {
      "epoch": 0.5930157448526443,
      "grad_norm": 84.03296661376953,
      "learning_rate": 7.500539540520122e-06,
      "loss": 4.8348,
      "step": 14689
    },
    {
      "epoch": 0.593056116269681,
      "grad_norm": 120.90074157714844,
      "learning_rate": 7.499273484405522e-06,
      "loss": 3.7172,
      "step": 14690
    },
    {
      "epoch": 0.5930964876867179,
      "grad_norm": 139.46273803710938,
      "learning_rate": 7.498007471047508e-06,
      "loss": 4.7209,
      "step": 14691
    },
    {
      "epoch": 0.5931368591037546,
      "grad_norm": 359.8480224609375,
      "learning_rate": 7.496741500467727e-06,
      "loss": 4.7728,
      "step": 14692
    },
    {
      "epoch": 0.5931772305207913,
      "grad_norm": 101.3918685913086,
      "learning_rate": 7.495475572687816e-06,
      "loss": 4.813,
      "step": 14693
    },
    {
      "epoch": 0.593217601937828,
      "grad_norm": 112.30126190185547,
      "learning_rate": 7.494209687729424e-06,
      "loss": 4.0878,
      "step": 14694
    },
    {
      "epoch": 0.5932579733548647,
      "grad_norm": 97.80374145507812,
      "learning_rate": 7.492943845614196e-06,
      "loss": 4.8028,
      "step": 14695
    },
    {
      "epoch": 0.5932983447719015,
      "grad_norm": 101.54855346679688,
      "learning_rate": 7.491678046363777e-06,
      "loss": 5.2176,
      "step": 14696
    },
    {
      "epoch": 0.5933387161889382,
      "grad_norm": 126.11683654785156,
      "learning_rate": 7.490412289999803e-06,
      "loss": 6.5942,
      "step": 14697
    },
    {
      "epoch": 0.593379087605975,
      "grad_norm": 69.96089935302734,
      "learning_rate": 7.4891465765439175e-06,
      "loss": 4.8849,
      "step": 14698
    },
    {
      "epoch": 0.5934194590230117,
      "grad_norm": 93.85377502441406,
      "learning_rate": 7.487880906017764e-06,
      "loss": 5.7217,
      "step": 14699
    },
    {
      "epoch": 0.5934598304400485,
      "grad_norm": 109.65545654296875,
      "learning_rate": 7.4866152784429815e-06,
      "loss": 4.5696,
      "step": 14700
    },
    {
      "epoch": 0.5935002018570852,
      "grad_norm": 151.76138305664062,
      "learning_rate": 7.485349693841206e-06,
      "loss": 4.2276,
      "step": 14701
    },
    {
      "epoch": 0.5935405732741219,
      "grad_norm": 93.57101440429688,
      "learning_rate": 7.484084152234079e-06,
      "loss": 5.1915,
      "step": 14702
    },
    {
      "epoch": 0.5935809446911586,
      "grad_norm": 86.28194427490234,
      "learning_rate": 7.482818653643238e-06,
      "loss": 5.1569,
      "step": 14703
    },
    {
      "epoch": 0.5936213161081954,
      "grad_norm": 95.46849060058594,
      "learning_rate": 7.4815531980903246e-06,
      "loss": 4.8732,
      "step": 14704
    },
    {
      "epoch": 0.5936616875252322,
      "grad_norm": 90.21355438232422,
      "learning_rate": 7.480287785596966e-06,
      "loss": 5.2892,
      "step": 14705
    },
    {
      "epoch": 0.5937020589422689,
      "grad_norm": 100.57847595214844,
      "learning_rate": 7.479022416184802e-06,
      "loss": 6.2547,
      "step": 14706
    },
    {
      "epoch": 0.5937424303593056,
      "grad_norm": 88.12271881103516,
      "learning_rate": 7.477757089875468e-06,
      "loss": 5.7975,
      "step": 14707
    },
    {
      "epoch": 0.5937828017763424,
      "grad_norm": 164.39212036132812,
      "learning_rate": 7.4764918066906e-06,
      "loss": 4.2644,
      "step": 14708
    },
    {
      "epoch": 0.5938231731933791,
      "grad_norm": 81.07402801513672,
      "learning_rate": 7.4752265666518275e-06,
      "loss": 6.267,
      "step": 14709
    },
    {
      "epoch": 0.5938635446104158,
      "grad_norm": 89.21849060058594,
      "learning_rate": 7.473961369780785e-06,
      "loss": 5.5396,
      "step": 14710
    },
    {
      "epoch": 0.5939039160274525,
      "grad_norm": 98.16455078125,
      "learning_rate": 7.472696216099105e-06,
      "loss": 3.7609,
      "step": 14711
    },
    {
      "epoch": 0.5939442874444893,
      "grad_norm": 80.45795440673828,
      "learning_rate": 7.4714311056284195e-06,
      "loss": 4.9649,
      "step": 14712
    },
    {
      "epoch": 0.5939846588615261,
      "grad_norm": 109.38480377197266,
      "learning_rate": 7.470166038390356e-06,
      "loss": 5.1095,
      "step": 14713
    },
    {
      "epoch": 0.5940250302785628,
      "grad_norm": 66.32968139648438,
      "learning_rate": 7.468901014406545e-06,
      "loss": 5.5472,
      "step": 14714
    },
    {
      "epoch": 0.5940654016955995,
      "grad_norm": 101.0186767578125,
      "learning_rate": 7.467636033698615e-06,
      "loss": 6.7941,
      "step": 14715
    },
    {
      "epoch": 0.5941057731126362,
      "grad_norm": 79.3128433227539,
      "learning_rate": 7.466371096288201e-06,
      "loss": 5.915,
      "step": 14716
    },
    {
      "epoch": 0.594146144529673,
      "grad_norm": 80.7952651977539,
      "learning_rate": 7.465106202196919e-06,
      "loss": 3.4208,
      "step": 14717
    },
    {
      "epoch": 0.5941865159467097,
      "grad_norm": 98.50370025634766,
      "learning_rate": 7.4638413514464035e-06,
      "loss": 5.7645,
      "step": 14718
    },
    {
      "epoch": 0.5942268873637465,
      "grad_norm": 78.1889419555664,
      "learning_rate": 7.4625765440582775e-06,
      "loss": 5.2093,
      "step": 14719
    },
    {
      "epoch": 0.5942672587807832,
      "grad_norm": 64.5621566772461,
      "learning_rate": 7.461311780054169e-06,
      "loss": 4.1336,
      "step": 14720
    },
    {
      "epoch": 0.59430763019782,
      "grad_norm": 121.36940002441406,
      "learning_rate": 7.4600470594557e-06,
      "loss": 3.5739,
      "step": 14721
    },
    {
      "epoch": 0.5943480016148567,
      "grad_norm": 110.96485900878906,
      "learning_rate": 7.458782382284493e-06,
      "loss": 4.5205,
      "step": 14722
    },
    {
      "epoch": 0.5943883730318934,
      "grad_norm": 131.95042419433594,
      "learning_rate": 7.457517748562174e-06,
      "loss": 4.5597,
      "step": 14723
    },
    {
      "epoch": 0.5944287444489301,
      "grad_norm": 99.44304656982422,
      "learning_rate": 7.45625315831037e-06,
      "loss": 4.8939,
      "step": 14724
    },
    {
      "epoch": 0.5944691158659668,
      "grad_norm": 151.62728881835938,
      "learning_rate": 7.454988611550691e-06,
      "loss": 6.3031,
      "step": 14725
    },
    {
      "epoch": 0.5945094872830037,
      "grad_norm": 95.92244720458984,
      "learning_rate": 7.453724108304764e-06,
      "loss": 5.5669,
      "step": 14726
    },
    {
      "epoch": 0.5945498587000404,
      "grad_norm": 77.04932403564453,
      "learning_rate": 7.452459648594209e-06,
      "loss": 5.2878,
      "step": 14727
    },
    {
      "epoch": 0.5945902301170771,
      "grad_norm": 105.71910858154297,
      "learning_rate": 7.4511952324406464e-06,
      "loss": 5.071,
      "step": 14728
    },
    {
      "epoch": 0.5946306015341138,
      "grad_norm": 70.58418273925781,
      "learning_rate": 7.449930859865693e-06,
      "loss": 5.0231,
      "step": 14729
    },
    {
      "epoch": 0.5946709729511506,
      "grad_norm": 135.34893798828125,
      "learning_rate": 7.448666530890967e-06,
      "loss": 3.4579,
      "step": 14730
    },
    {
      "epoch": 0.5947113443681873,
      "grad_norm": 91.27085876464844,
      "learning_rate": 7.447402245538085e-06,
      "loss": 7.032,
      "step": 14731
    },
    {
      "epoch": 0.594751715785224,
      "grad_norm": 166.0756072998047,
      "learning_rate": 7.446138003828667e-06,
      "loss": 5.3342,
      "step": 14732
    },
    {
      "epoch": 0.5947920872022608,
      "grad_norm": 115.70755004882812,
      "learning_rate": 7.444873805784321e-06,
      "loss": 4.6228,
      "step": 14733
    },
    {
      "epoch": 0.5948324586192976,
      "grad_norm": 76.27767944335938,
      "learning_rate": 7.443609651426669e-06,
      "loss": 4.9648,
      "step": 14734
    },
    {
      "epoch": 0.5948728300363343,
      "grad_norm": 71.83222198486328,
      "learning_rate": 7.442345540777319e-06,
      "loss": 4.678,
      "step": 14735
    },
    {
      "epoch": 0.594913201453371,
      "grad_norm": 83.20214080810547,
      "learning_rate": 7.441081473857894e-06,
      "loss": 6.4304,
      "step": 14736
    },
    {
      "epoch": 0.5949535728704077,
      "grad_norm": 157.52243041992188,
      "learning_rate": 7.4398174506899955e-06,
      "loss": 3.8208,
      "step": 14737
    },
    {
      "epoch": 0.5949939442874445,
      "grad_norm": 154.4055938720703,
      "learning_rate": 7.43855347129524e-06,
      "loss": 3.6064,
      "step": 14738
    },
    {
      "epoch": 0.5950343157044812,
      "grad_norm": 189.06430053710938,
      "learning_rate": 7.43728953569524e-06,
      "loss": 4.814,
      "step": 14739
    },
    {
      "epoch": 0.595074687121518,
      "grad_norm": 110.6469497680664,
      "learning_rate": 7.436025643911605e-06,
      "loss": 4.7075,
      "step": 14740
    },
    {
      "epoch": 0.5951150585385547,
      "grad_norm": 94.17615509033203,
      "learning_rate": 7.434761795965944e-06,
      "loss": 6.4966,
      "step": 14741
    },
    {
      "epoch": 0.5951554299555915,
      "grad_norm": 75.78939819335938,
      "learning_rate": 7.433497991879863e-06,
      "loss": 6.4576,
      "step": 14742
    },
    {
      "epoch": 0.5951958013726282,
      "grad_norm": 93.13579559326172,
      "learning_rate": 7.432234231674975e-06,
      "loss": 6.0771,
      "step": 14743
    },
    {
      "epoch": 0.5952361727896649,
      "grad_norm": 97.63681030273438,
      "learning_rate": 7.4309705153728905e-06,
      "loss": 5.5157,
      "step": 14744
    },
    {
      "epoch": 0.5952765442067016,
      "grad_norm": 145.97679138183594,
      "learning_rate": 7.4297068429952055e-06,
      "loss": 5.6365,
      "step": 14745
    },
    {
      "epoch": 0.5953169156237383,
      "grad_norm": 140.95188903808594,
      "learning_rate": 7.428443214563532e-06,
      "loss": 6.5523,
      "step": 14746
    },
    {
      "epoch": 0.5953572870407752,
      "grad_norm": 99.99958801269531,
      "learning_rate": 7.4271796300994745e-06,
      "loss": 4.8023,
      "step": 14747
    },
    {
      "epoch": 0.5953976584578119,
      "grad_norm": 95.19613647460938,
      "learning_rate": 7.42591608962464e-06,
      "loss": 3.5028,
      "step": 14748
    },
    {
      "epoch": 0.5954380298748486,
      "grad_norm": 92.8193588256836,
      "learning_rate": 7.424652593160626e-06,
      "loss": 6.9498,
      "step": 14749
    },
    {
      "epoch": 0.5954784012918853,
      "grad_norm": 139.84463500976562,
      "learning_rate": 7.42338914072904e-06,
      "loss": 5.0202,
      "step": 14750
    },
    {
      "epoch": 0.5955187727089221,
      "grad_norm": 85.39323425292969,
      "learning_rate": 7.422125732351482e-06,
      "loss": 5.3361,
      "step": 14751
    },
    {
      "epoch": 0.5955591441259588,
      "grad_norm": 98.13599395751953,
      "learning_rate": 7.420862368049559e-06,
      "loss": 3.2864,
      "step": 14752
    },
    {
      "epoch": 0.5955995155429955,
      "grad_norm": 89.51173400878906,
      "learning_rate": 7.419599047844863e-06,
      "loss": 3.7774,
      "step": 14753
    },
    {
      "epoch": 0.5956398869600323,
      "grad_norm": 143.29135131835938,
      "learning_rate": 7.418335771758997e-06,
      "loss": 4.5049,
      "step": 14754
    },
    {
      "epoch": 0.5956802583770691,
      "grad_norm": 99.7560043334961,
      "learning_rate": 7.41707253981356e-06,
      "loss": 4.5728,
      "step": 14755
    },
    {
      "epoch": 0.5957206297941058,
      "grad_norm": 85.65583038330078,
      "learning_rate": 7.415809352030156e-06,
      "loss": 4.924,
      "step": 14756
    },
    {
      "epoch": 0.5957610012111425,
      "grad_norm": 454.129150390625,
      "learning_rate": 7.414546208430373e-06,
      "loss": 5.1298,
      "step": 14757
    },
    {
      "epoch": 0.5958013726281792,
      "grad_norm": 119.74810791015625,
      "learning_rate": 7.413283109035813e-06,
      "loss": 5.2995,
      "step": 14758
    },
    {
      "epoch": 0.595841744045216,
      "grad_norm": 71.38742065429688,
      "learning_rate": 7.412020053868071e-06,
      "loss": 5.4996,
      "step": 14759
    },
    {
      "epoch": 0.5958821154622527,
      "grad_norm": 94.10301208496094,
      "learning_rate": 7.410757042948744e-06,
      "loss": 5.6775,
      "step": 14760
    },
    {
      "epoch": 0.5959224868792895,
      "grad_norm": 124.42657470703125,
      "learning_rate": 7.409494076299424e-06,
      "loss": 5.4653,
      "step": 14761
    },
    {
      "epoch": 0.5959628582963262,
      "grad_norm": 90.51052856445312,
      "learning_rate": 7.408231153941707e-06,
      "loss": 3.0536,
      "step": 14762
    },
    {
      "epoch": 0.596003229713363,
      "grad_norm": 110.2538070678711,
      "learning_rate": 7.406968275897184e-06,
      "loss": 5.9923,
      "step": 14763
    },
    {
      "epoch": 0.5960436011303997,
      "grad_norm": 103.65495300292969,
      "learning_rate": 7.405705442187452e-06,
      "loss": 5.5553,
      "step": 14764
    },
    {
      "epoch": 0.5960839725474364,
      "grad_norm": 115.42537689208984,
      "learning_rate": 7.404442652834095e-06,
      "loss": 5.6067,
      "step": 14765
    },
    {
      "epoch": 0.5961243439644731,
      "grad_norm": 89.27912902832031,
      "learning_rate": 7.403179907858708e-06,
      "loss": 4.1552,
      "step": 14766
    },
    {
      "epoch": 0.59616471538151,
      "grad_norm": 85.74832916259766,
      "learning_rate": 7.401917207282881e-06,
      "loss": 4.3049,
      "step": 14767
    },
    {
      "epoch": 0.5962050867985467,
      "grad_norm": 126.4527816772461,
      "learning_rate": 7.400654551128204e-06,
      "loss": 8.5069,
      "step": 14768
    },
    {
      "epoch": 0.5962454582155834,
      "grad_norm": 97.0390625,
      "learning_rate": 7.399391939416263e-06,
      "loss": 5.18,
      "step": 14769
    },
    {
      "epoch": 0.5962858296326201,
      "grad_norm": 88.05663299560547,
      "learning_rate": 7.398129372168646e-06,
      "loss": 3.6478,
      "step": 14770
    },
    {
      "epoch": 0.5963262010496568,
      "grad_norm": 128.14898681640625,
      "learning_rate": 7.396866849406941e-06,
      "loss": 6.5165,
      "step": 14771
    },
    {
      "epoch": 0.5963665724666936,
      "grad_norm": 157.38478088378906,
      "learning_rate": 7.3956043711527385e-06,
      "loss": 6.8367,
      "step": 14772
    },
    {
      "epoch": 0.5964069438837303,
      "grad_norm": 102.04774475097656,
      "learning_rate": 7.394341937427616e-06,
      "loss": 5.4748,
      "step": 14773
    },
    {
      "epoch": 0.5964473153007671,
      "grad_norm": 124.71226501464844,
      "learning_rate": 7.3930795482531616e-06,
      "loss": 3.8933,
      "step": 14774
    },
    {
      "epoch": 0.5964876867178038,
      "grad_norm": 150.12857055664062,
      "learning_rate": 7.3918172036509575e-06,
      "loss": 4.4996,
      "step": 14775
    },
    {
      "epoch": 0.5965280581348406,
      "grad_norm": 230.0227813720703,
      "learning_rate": 7.390554903642593e-06,
      "loss": 5.3907,
      "step": 14776
    },
    {
      "epoch": 0.5965684295518773,
      "grad_norm": 128.58717346191406,
      "learning_rate": 7.389292648249643e-06,
      "loss": 5.0005,
      "step": 14777
    },
    {
      "epoch": 0.596608800968914,
      "grad_norm": 88.67289733886719,
      "learning_rate": 7.388030437493694e-06,
      "loss": 5.5944,
      "step": 14778
    },
    {
      "epoch": 0.5966491723859507,
      "grad_norm": 88.68045806884766,
      "learning_rate": 7.386768271396324e-06,
      "loss": 5.0736,
      "step": 14779
    },
    {
      "epoch": 0.5966895438029874,
      "grad_norm": 105.58873748779297,
      "learning_rate": 7.385506149979116e-06,
      "loss": 5.1188,
      "step": 14780
    },
    {
      "epoch": 0.5967299152200243,
      "grad_norm": 88.43230438232422,
      "learning_rate": 7.384244073263645e-06,
      "loss": 5.3242,
      "step": 14781
    },
    {
      "epoch": 0.596770286637061,
      "grad_norm": 233.92584228515625,
      "learning_rate": 7.382982041271493e-06,
      "loss": 4.9903,
      "step": 14782
    },
    {
      "epoch": 0.5968106580540977,
      "grad_norm": 125.76992797851562,
      "learning_rate": 7.381720054024237e-06,
      "loss": 5.9081,
      "step": 14783
    },
    {
      "epoch": 0.5968510294711344,
      "grad_norm": 100.5492172241211,
      "learning_rate": 7.380458111543458e-06,
      "loss": 6.2168,
      "step": 14784
    },
    {
      "epoch": 0.5968914008881712,
      "grad_norm": 169.85720825195312,
      "learning_rate": 7.379196213850724e-06,
      "loss": 4.9786,
      "step": 14785
    },
    {
      "epoch": 0.5969317723052079,
      "grad_norm": 97.3908462524414,
      "learning_rate": 7.377934360967613e-06,
      "loss": 5.2879,
      "step": 14786
    },
    {
      "epoch": 0.5969721437222446,
      "grad_norm": 84.88790893554688,
      "learning_rate": 7.376672552915703e-06,
      "loss": 5.1324,
      "step": 14787
    },
    {
      "epoch": 0.5970125151392814,
      "grad_norm": 109.37419128417969,
      "learning_rate": 7.3754107897165685e-06,
      "loss": 4.1468,
      "step": 14788
    },
    {
      "epoch": 0.5970528865563182,
      "grad_norm": 156.2028045654297,
      "learning_rate": 7.374149071391778e-06,
      "loss": 3.1929,
      "step": 14789
    },
    {
      "epoch": 0.5970932579733549,
      "grad_norm": 100.87886047363281,
      "learning_rate": 7.372887397962907e-06,
      "loss": 4.8513,
      "step": 14790
    },
    {
      "epoch": 0.5971336293903916,
      "grad_norm": 88.2015151977539,
      "learning_rate": 7.371625769451527e-06,
      "loss": 4.9259,
      "step": 14791
    },
    {
      "epoch": 0.5971740008074283,
      "grad_norm": 90.39897918701172,
      "learning_rate": 7.370364185879213e-06,
      "loss": 4.8687,
      "step": 14792
    },
    {
      "epoch": 0.597214372224465,
      "grad_norm": 168.0242919921875,
      "learning_rate": 7.3691026472675255e-06,
      "loss": 4.8779,
      "step": 14793
    },
    {
      "epoch": 0.5972547436415018,
      "grad_norm": 67.62248229980469,
      "learning_rate": 7.36784115363804e-06,
      "loss": 4.6085,
      "step": 14794
    },
    {
      "epoch": 0.5972951150585386,
      "grad_norm": 70.93236541748047,
      "learning_rate": 7.366579705012325e-06,
      "loss": 4.4757,
      "step": 14795
    },
    {
      "epoch": 0.5973354864755753,
      "grad_norm": 73.94470977783203,
      "learning_rate": 7.3653183014119475e-06,
      "loss": 4.4896,
      "step": 14796
    },
    {
      "epoch": 0.597375857892612,
      "grad_norm": 89.35645294189453,
      "learning_rate": 7.364056942858475e-06,
      "loss": 5.4259,
      "step": 14797
    },
    {
      "epoch": 0.5974162293096488,
      "grad_norm": 72.21392059326172,
      "learning_rate": 7.362795629373471e-06,
      "loss": 5.3259,
      "step": 14798
    },
    {
      "epoch": 0.5974566007266855,
      "grad_norm": 99.48673248291016,
      "learning_rate": 7.361534360978505e-06,
      "loss": 4.1649,
      "step": 14799
    },
    {
      "epoch": 0.5974969721437222,
      "grad_norm": 71.48187255859375,
      "learning_rate": 7.360273137695144e-06,
      "loss": 4.6644,
      "step": 14800
    },
    {
      "epoch": 0.5975373435607589,
      "grad_norm": 108.0242919921875,
      "learning_rate": 7.359011959544944e-06,
      "loss": 4.2981,
      "step": 14801
    },
    {
      "epoch": 0.5975777149777958,
      "grad_norm": 77.94445037841797,
      "learning_rate": 7.357750826549472e-06,
      "loss": 4.6444,
      "step": 14802
    },
    {
      "epoch": 0.5976180863948325,
      "grad_norm": 65.26771545410156,
      "learning_rate": 7.356489738730291e-06,
      "loss": 4.2998,
      "step": 14803
    },
    {
      "epoch": 0.5976584578118692,
      "grad_norm": 74.05535888671875,
      "learning_rate": 7.355228696108967e-06,
      "loss": 6.6066,
      "step": 14804
    },
    {
      "epoch": 0.5976988292289059,
      "grad_norm": 90.34732818603516,
      "learning_rate": 7.353967698707052e-06,
      "loss": 5.5634,
      "step": 14805
    },
    {
      "epoch": 0.5977392006459427,
      "grad_norm": 79.05510711669922,
      "learning_rate": 7.35270674654611e-06,
      "loss": 5.0044,
      "step": 14806
    },
    {
      "epoch": 0.5977795720629794,
      "grad_norm": 60.3832893371582,
      "learning_rate": 7.3514458396477e-06,
      "loss": 3.5213,
      "step": 14807
    },
    {
      "epoch": 0.5978199434800161,
      "grad_norm": 96.56244659423828,
      "learning_rate": 7.350184978033386e-06,
      "loss": 3.998,
      "step": 14808
    },
    {
      "epoch": 0.5978603148970529,
      "grad_norm": 91.30521392822266,
      "learning_rate": 7.348924161724716e-06,
      "loss": 4.6009,
      "step": 14809
    },
    {
      "epoch": 0.5979006863140897,
      "grad_norm": 94.33282470703125,
      "learning_rate": 7.347663390743254e-06,
      "loss": 5.0793,
      "step": 14810
    },
    {
      "epoch": 0.5979410577311264,
      "grad_norm": 87.59749603271484,
      "learning_rate": 7.346402665110553e-06,
      "loss": 5.0761,
      "step": 14811
    },
    {
      "epoch": 0.5979814291481631,
      "grad_norm": 128.07640075683594,
      "learning_rate": 7.3451419848481765e-06,
      "loss": 6.0716,
      "step": 14812
    },
    {
      "epoch": 0.5980218005651998,
      "grad_norm": 108.11631774902344,
      "learning_rate": 7.343881349977666e-06,
      "loss": 4.9498,
      "step": 14813
    },
    {
      "epoch": 0.5980621719822365,
      "grad_norm": 131.65626525878906,
      "learning_rate": 7.342620760520583e-06,
      "loss": 4.874,
      "step": 14814
    },
    {
      "epoch": 0.5981025433992733,
      "grad_norm": 98.30108642578125,
      "learning_rate": 7.341360216498479e-06,
      "loss": 4.6787,
      "step": 14815
    },
    {
      "epoch": 0.5981429148163101,
      "grad_norm": 79.28250122070312,
      "learning_rate": 7.340099717932909e-06,
      "loss": 4.0219,
      "step": 14816
    },
    {
      "epoch": 0.5981832862333468,
      "grad_norm": 97.27115631103516,
      "learning_rate": 7.338839264845421e-06,
      "loss": 2.9223,
      "step": 14817
    },
    {
      "epoch": 0.5982236576503835,
      "grad_norm": 99.96615600585938,
      "learning_rate": 7.3375788572575655e-06,
      "loss": 4.6164,
      "step": 14818
    },
    {
      "epoch": 0.5982640290674203,
      "grad_norm": 117.7203369140625,
      "learning_rate": 7.336318495190897e-06,
      "loss": 6.0225,
      "step": 14819
    },
    {
      "epoch": 0.598304400484457,
      "grad_norm": 91.71234893798828,
      "learning_rate": 7.335058178666965e-06,
      "loss": 5.3638,
      "step": 14820
    },
    {
      "epoch": 0.5983447719014937,
      "grad_norm": 91.11293029785156,
      "learning_rate": 7.333797907707311e-06,
      "loss": 4.7678,
      "step": 14821
    },
    {
      "epoch": 0.5983851433185304,
      "grad_norm": 78.60757446289062,
      "learning_rate": 7.332537682333487e-06,
      "loss": 5.6426,
      "step": 14822
    },
    {
      "epoch": 0.5984255147355673,
      "grad_norm": 125.54552459716797,
      "learning_rate": 7.331277502567038e-06,
      "loss": 5.3368,
      "step": 14823
    },
    {
      "epoch": 0.598465886152604,
      "grad_norm": 129.81947326660156,
      "learning_rate": 7.330017368429518e-06,
      "loss": 4.7765,
      "step": 14824
    },
    {
      "epoch": 0.5985062575696407,
      "grad_norm": 74.61341857910156,
      "learning_rate": 7.328757279942462e-06,
      "loss": 5.1894,
      "step": 14825
    },
    {
      "epoch": 0.5985466289866774,
      "grad_norm": 104.4049072265625,
      "learning_rate": 7.327497237127416e-06,
      "loss": 5.0544,
      "step": 14826
    },
    {
      "epoch": 0.5985870004037142,
      "grad_norm": 139.1348114013672,
      "learning_rate": 7.32623724000593e-06,
      "loss": 5.3863,
      "step": 14827
    },
    {
      "epoch": 0.5986273718207509,
      "grad_norm": 92.51769256591797,
      "learning_rate": 7.324977288599544e-06,
      "loss": 5.0686,
      "step": 14828
    },
    {
      "epoch": 0.5986677432377876,
      "grad_norm": 79.60529327392578,
      "learning_rate": 7.323717382929798e-06,
      "loss": 3.4155,
      "step": 14829
    },
    {
      "epoch": 0.5987081146548244,
      "grad_norm": 137.60641479492188,
      "learning_rate": 7.322457523018235e-06,
      "loss": 6.3381,
      "step": 14830
    },
    {
      "epoch": 0.5987484860718612,
      "grad_norm": 85.49626159667969,
      "learning_rate": 7.321197708886396e-06,
      "loss": 5.397,
      "step": 14831
    },
    {
      "epoch": 0.5987888574888979,
      "grad_norm": 75.97003936767578,
      "learning_rate": 7.319937940555824e-06,
      "loss": 5.4919,
      "step": 14832
    },
    {
      "epoch": 0.5988292289059346,
      "grad_norm": 97.58062744140625,
      "learning_rate": 7.318678218048052e-06,
      "loss": 5.5912,
      "step": 14833
    },
    {
      "epoch": 0.5988696003229713,
      "grad_norm": 88.62043762207031,
      "learning_rate": 7.31741854138462e-06,
      "loss": 5.5742,
      "step": 14834
    },
    {
      "epoch": 0.598909971740008,
      "grad_norm": 499.3341064453125,
      "learning_rate": 7.316158910587068e-06,
      "loss": 4.9073,
      "step": 14835
    },
    {
      "epoch": 0.5989503431570448,
      "grad_norm": 88.50043487548828,
      "learning_rate": 7.314899325676932e-06,
      "loss": 6.5825,
      "step": 14836
    },
    {
      "epoch": 0.5989907145740816,
      "grad_norm": 97.92810821533203,
      "learning_rate": 7.313639786675746e-06,
      "loss": 4.7392,
      "step": 14837
    },
    {
      "epoch": 0.5990310859911183,
      "grad_norm": 122.70655822753906,
      "learning_rate": 7.312380293605046e-06,
      "loss": 5.2438,
      "step": 14838
    },
    {
      "epoch": 0.599071457408155,
      "grad_norm": 161.6297149658203,
      "learning_rate": 7.311120846486368e-06,
      "loss": 3.8867,
      "step": 14839
    },
    {
      "epoch": 0.5991118288251918,
      "grad_norm": 89.89632415771484,
      "learning_rate": 7.3098614453412474e-06,
      "loss": 6.0354,
      "step": 14840
    },
    {
      "epoch": 0.5991522002422285,
      "grad_norm": 123.12545013427734,
      "learning_rate": 7.3086020901912104e-06,
      "loss": 4.5521,
      "step": 14841
    },
    {
      "epoch": 0.5991925716592652,
      "grad_norm": 204.2048797607422,
      "learning_rate": 7.3073427810577935e-06,
      "loss": 4.815,
      "step": 14842
    },
    {
      "epoch": 0.5992329430763019,
      "grad_norm": 108.0106430053711,
      "learning_rate": 7.306083517962526e-06,
      "loss": 5.5214,
      "step": 14843
    },
    {
      "epoch": 0.5992733144933388,
      "grad_norm": 103.08214569091797,
      "learning_rate": 7.304824300926945e-06,
      "loss": 5.0101,
      "step": 14844
    },
    {
      "epoch": 0.5993136859103755,
      "grad_norm": 130.71800231933594,
      "learning_rate": 7.303565129972571e-06,
      "loss": 3.5967,
      "step": 14845
    },
    {
      "epoch": 0.5993540573274122,
      "grad_norm": 132.13511657714844,
      "learning_rate": 7.302306005120936e-06,
      "loss": 4.9708,
      "step": 14846
    },
    {
      "epoch": 0.5993944287444489,
      "grad_norm": 317.5114440917969,
      "learning_rate": 7.301046926393568e-06,
      "loss": 6.4925,
      "step": 14847
    },
    {
      "epoch": 0.5994348001614856,
      "grad_norm": 169.7836456298828,
      "learning_rate": 7.299787893811998e-06,
      "loss": 5.8274,
      "step": 14848
    },
    {
      "epoch": 0.5994751715785224,
      "grad_norm": 117.97166442871094,
      "learning_rate": 7.298528907397746e-06,
      "loss": 6.5632,
      "step": 14849
    },
    {
      "epoch": 0.5995155429955591,
      "grad_norm": 117.61019134521484,
      "learning_rate": 7.297269967172344e-06,
      "loss": 5.1353,
      "step": 14850
    },
    {
      "epoch": 0.5995559144125959,
      "grad_norm": 90.04829406738281,
      "learning_rate": 7.296011073157312e-06,
      "loss": 5.0427,
      "step": 14851
    },
    {
      "epoch": 0.5995962858296326,
      "grad_norm": 85.31240844726562,
      "learning_rate": 7.294752225374182e-06,
      "loss": 2.9361,
      "step": 14852
    },
    {
      "epoch": 0.5996366572466694,
      "grad_norm": 74.21347045898438,
      "learning_rate": 7.2934934238444665e-06,
      "loss": 2.0391,
      "step": 14853
    },
    {
      "epoch": 0.5996770286637061,
      "grad_norm": 92.35797119140625,
      "learning_rate": 7.292234668589693e-06,
      "loss": 5.759,
      "step": 14854
    },
    {
      "epoch": 0.5997174000807428,
      "grad_norm": 86.26177978515625,
      "learning_rate": 7.290975959631384e-06,
      "loss": 4.4375,
      "step": 14855
    },
    {
      "epoch": 0.5997577714977795,
      "grad_norm": 103.59772491455078,
      "learning_rate": 7.289717296991063e-06,
      "loss": 6.2866,
      "step": 14856
    },
    {
      "epoch": 0.5997981429148163,
      "grad_norm": 96.88607025146484,
      "learning_rate": 7.288458680690244e-06,
      "loss": 5.9197,
      "step": 14857
    },
    {
      "epoch": 0.5998385143318531,
      "grad_norm": 79.42759704589844,
      "learning_rate": 7.28720011075045e-06,
      "loss": 5.2103,
      "step": 14858
    },
    {
      "epoch": 0.5998788857488898,
      "grad_norm": 76.70242309570312,
      "learning_rate": 7.285941587193199e-06,
      "loss": 5.3295,
      "step": 14859
    },
    {
      "epoch": 0.5999192571659265,
      "grad_norm": 107.96056365966797,
      "learning_rate": 7.284683110040015e-06,
      "loss": 4.8464,
      "step": 14860
    },
    {
      "epoch": 0.5999596285829633,
      "grad_norm": 87.91973114013672,
      "learning_rate": 7.283424679312403e-06,
      "loss": 5.4278,
      "step": 14861
    },
    {
      "epoch": 0.6,
      "grad_norm": 95.22875213623047,
      "learning_rate": 7.282166295031886e-06,
      "loss": 5.1771,
      "step": 14862
    },
    {
      "epoch": 0.6000403714170367,
      "grad_norm": 105.88780212402344,
      "learning_rate": 7.2809079572199795e-06,
      "loss": 4.7638,
      "step": 14863
    },
    {
      "epoch": 0.6000807428340734,
      "grad_norm": 121.10681915283203,
      "learning_rate": 7.279649665898201e-06,
      "loss": 5.7618,
      "step": 14864
    },
    {
      "epoch": 0.6001211142511103,
      "grad_norm": 76.74837493896484,
      "learning_rate": 7.278391421088056e-06,
      "loss": 4.2423,
      "step": 14865
    },
    {
      "epoch": 0.600161485668147,
      "grad_norm": 128.972900390625,
      "learning_rate": 7.277133222811063e-06,
      "loss": 4.6893,
      "step": 14866
    },
    {
      "epoch": 0.6002018570851837,
      "grad_norm": 130.17433166503906,
      "learning_rate": 7.275875071088733e-06,
      "loss": 4.788,
      "step": 14867
    },
    {
      "epoch": 0.6002422285022204,
      "grad_norm": 93.05926513671875,
      "learning_rate": 7.27461696594258e-06,
      "loss": 6.1646,
      "step": 14868
    },
    {
      "epoch": 0.6002825999192571,
      "grad_norm": 92.2562026977539,
      "learning_rate": 7.27335890739411e-06,
      "loss": 6.0666,
      "step": 14869
    },
    {
      "epoch": 0.6003229713362939,
      "grad_norm": 74.66007232666016,
      "learning_rate": 7.272100895464838e-06,
      "loss": 4.4079,
      "step": 14870
    },
    {
      "epoch": 0.6003633427533306,
      "grad_norm": 73.39131164550781,
      "learning_rate": 7.270842930176267e-06,
      "loss": 5.1479,
      "step": 14871
    },
    {
      "epoch": 0.6004037141703674,
      "grad_norm": 98.67135620117188,
      "learning_rate": 7.269585011549916e-06,
      "loss": 5.4425,
      "step": 14872
    },
    {
      "epoch": 0.6004440855874041,
      "grad_norm": 79.67919158935547,
      "learning_rate": 7.268327139607279e-06,
      "loss": 6.4744,
      "step": 14873
    },
    {
      "epoch": 0.6004844570044409,
      "grad_norm": 270.9787902832031,
      "learning_rate": 7.267069314369871e-06,
      "loss": 3.64,
      "step": 14874
    },
    {
      "epoch": 0.6005248284214776,
      "grad_norm": 85.56632995605469,
      "learning_rate": 7.2658115358591955e-06,
      "loss": 4.3855,
      "step": 14875
    },
    {
      "epoch": 0.6005651998385143,
      "grad_norm": 98.39498901367188,
      "learning_rate": 7.264553804096759e-06,
      "loss": 4.9739,
      "step": 14876
    },
    {
      "epoch": 0.600605571255551,
      "grad_norm": 69.73499298095703,
      "learning_rate": 7.263296119104063e-06,
      "loss": 3.46,
      "step": 14877
    },
    {
      "epoch": 0.6006459426725879,
      "grad_norm": 68.73957061767578,
      "learning_rate": 7.2620384809026135e-06,
      "loss": 3.0988,
      "step": 14878
    },
    {
      "epoch": 0.6006863140896246,
      "grad_norm": 83.84542083740234,
      "learning_rate": 7.260780889513911e-06,
      "loss": 5.3854,
      "step": 14879
    },
    {
      "epoch": 0.6007266855066613,
      "grad_norm": 94.18974304199219,
      "learning_rate": 7.2595233449594635e-06,
      "loss": 6.25,
      "step": 14880
    },
    {
      "epoch": 0.600767056923698,
      "grad_norm": 76.77477264404297,
      "learning_rate": 7.258265847260763e-06,
      "loss": 4.6155,
      "step": 14881
    },
    {
      "epoch": 0.6008074283407348,
      "grad_norm": 79.02808380126953,
      "learning_rate": 7.257008396439316e-06,
      "loss": 4.5924,
      "step": 14882
    },
    {
      "epoch": 0.6008477997577715,
      "grad_norm": 69.12642669677734,
      "learning_rate": 7.2557509925166166e-06,
      "loss": 3.4005,
      "step": 14883
    },
    {
      "epoch": 0.6008881711748082,
      "grad_norm": 83.42597961425781,
      "learning_rate": 7.254493635514172e-06,
      "loss": 5.0595,
      "step": 14884
    },
    {
      "epoch": 0.600928542591845,
      "grad_norm": 94.00425720214844,
      "learning_rate": 7.2532363254534724e-06,
      "loss": 4.9049,
      "step": 14885
    },
    {
      "epoch": 0.6009689140088817,
      "grad_norm": 96.22807312011719,
      "learning_rate": 7.251979062356016e-06,
      "loss": 5.7608,
      "step": 14886
    },
    {
      "epoch": 0.6010092854259185,
      "grad_norm": 106.03217315673828,
      "learning_rate": 7.2507218462433015e-06,
      "loss": 4.0287,
      "step": 14887
    },
    {
      "epoch": 0.6010496568429552,
      "grad_norm": 95.89196014404297,
      "learning_rate": 7.249464677136828e-06,
      "loss": 4.8849,
      "step": 14888
    },
    {
      "epoch": 0.6010900282599919,
      "grad_norm": 98.14350891113281,
      "learning_rate": 7.24820755505808e-06,
      "loss": 5.5181,
      "step": 14889
    },
    {
      "epoch": 0.6011303996770286,
      "grad_norm": 90.078857421875,
      "learning_rate": 7.246950480028559e-06,
      "loss": 5.2367,
      "step": 14890
    },
    {
      "epoch": 0.6011707710940654,
      "grad_norm": 106.11904907226562,
      "learning_rate": 7.245693452069754e-06,
      "loss": 3.1439,
      "step": 14891
    },
    {
      "epoch": 0.6012111425111022,
      "grad_norm": 115.57068634033203,
      "learning_rate": 7.2444364712031646e-06,
      "loss": 4.953,
      "step": 14892
    },
    {
      "epoch": 0.6012515139281389,
      "grad_norm": 113.63713836669922,
      "learning_rate": 7.243179537450272e-06,
      "loss": 4.8052,
      "step": 14893
    },
    {
      "epoch": 0.6012918853451756,
      "grad_norm": 125.5439682006836,
      "learning_rate": 7.241922650832571e-06,
      "loss": 6.1194,
      "step": 14894
    },
    {
      "epoch": 0.6013322567622124,
      "grad_norm": 98.85755920410156,
      "learning_rate": 7.240665811371554e-06,
      "loss": 5.5587,
      "step": 14895
    },
    {
      "epoch": 0.6013726281792491,
      "grad_norm": 123.0022201538086,
      "learning_rate": 7.23940901908871e-06,
      "loss": 6.8794,
      "step": 14896
    },
    {
      "epoch": 0.6014129995962858,
      "grad_norm": 117.83700561523438,
      "learning_rate": 7.238152274005521e-06,
      "loss": 4.9534,
      "step": 14897
    },
    {
      "epoch": 0.6014533710133225,
      "grad_norm": 138.9602813720703,
      "learning_rate": 7.236895576143481e-06,
      "loss": 4.0597,
      "step": 14898
    },
    {
      "epoch": 0.6014937424303594,
      "grad_norm": 152.02597045898438,
      "learning_rate": 7.235638925524073e-06,
      "loss": 4.9351,
      "step": 14899
    },
    {
      "epoch": 0.6015341138473961,
      "grad_norm": 148.85704040527344,
      "learning_rate": 7.234382322168789e-06,
      "loss": 4.5705,
      "step": 14900
    },
    {
      "epoch": 0.6015744852644328,
      "grad_norm": 194.5964813232422,
      "learning_rate": 7.233125766099105e-06,
      "loss": 4.1923,
      "step": 14901
    },
    {
      "epoch": 0.6016148566814695,
      "grad_norm": 186.85340881347656,
      "learning_rate": 7.231869257336509e-06,
      "loss": 5.3485,
      "step": 14902
    },
    {
      "epoch": 0.6016552280985062,
      "grad_norm": 142.20985412597656,
      "learning_rate": 7.2306127959024854e-06,
      "loss": 3.7643,
      "step": 14903
    },
    {
      "epoch": 0.601695599515543,
      "grad_norm": 87.18385314941406,
      "learning_rate": 7.229356381818517e-06,
      "loss": 5.1732,
      "step": 14904
    },
    {
      "epoch": 0.6017359709325797,
      "grad_norm": 77.55030822753906,
      "learning_rate": 7.228100015106083e-06,
      "loss": 4.0863,
      "step": 14905
    },
    {
      "epoch": 0.6017763423496165,
      "grad_norm": 116.90074920654297,
      "learning_rate": 7.226843695786665e-06,
      "loss": 5.0367,
      "step": 14906
    },
    {
      "epoch": 0.6018167137666532,
      "grad_norm": 108.35350799560547,
      "learning_rate": 7.225587423881746e-06,
      "loss": 5.652,
      "step": 14907
    },
    {
      "epoch": 0.60185708518369,
      "grad_norm": 95.7121810913086,
      "learning_rate": 7.224331199412806e-06,
      "loss": 4.6822,
      "step": 14908
    },
    {
      "epoch": 0.6018974566007267,
      "grad_norm": 97.1902847290039,
      "learning_rate": 7.2230750224013155e-06,
      "loss": 5.3305,
      "step": 14909
    },
    {
      "epoch": 0.6019378280177634,
      "grad_norm": 82.2431411743164,
      "learning_rate": 7.22181889286876e-06,
      "loss": 5.2651,
      "step": 14910
    },
    {
      "epoch": 0.6019781994348001,
      "grad_norm": 102.07349395751953,
      "learning_rate": 7.220562810836613e-06,
      "loss": 4.5548,
      "step": 14911
    },
    {
      "epoch": 0.6020185708518369,
      "grad_norm": 119.47705078125,
      "learning_rate": 7.2193067763263555e-06,
      "loss": 5.2177,
      "step": 14912
    },
    {
      "epoch": 0.6020589422688737,
      "grad_norm": 112.31953430175781,
      "learning_rate": 7.218050789359454e-06,
      "loss": 5.4676,
      "step": 14913
    },
    {
      "epoch": 0.6020993136859104,
      "grad_norm": 84.99629974365234,
      "learning_rate": 7.216794849957389e-06,
      "loss": 6.1379,
      "step": 14914
    },
    {
      "epoch": 0.6021396851029471,
      "grad_norm": 161.3673095703125,
      "learning_rate": 7.215538958141632e-06,
      "loss": 3.7893,
      "step": 14915
    },
    {
      "epoch": 0.6021800565199839,
      "grad_norm": 102.37725830078125,
      "learning_rate": 7.214283113933658e-06,
      "loss": 5.092,
      "step": 14916
    },
    {
      "epoch": 0.6022204279370206,
      "grad_norm": 92.66571807861328,
      "learning_rate": 7.213027317354937e-06,
      "loss": 5.6549,
      "step": 14917
    },
    {
      "epoch": 0.6022607993540573,
      "grad_norm": 77.70829010009766,
      "learning_rate": 7.21177156842694e-06,
      "loss": 5.169,
      "step": 14918
    },
    {
      "epoch": 0.602301170771094,
      "grad_norm": 67.82280731201172,
      "learning_rate": 7.210515867171139e-06,
      "loss": 5.2166,
      "step": 14919
    },
    {
      "epoch": 0.6023415421881309,
      "grad_norm": 81.99955749511719,
      "learning_rate": 7.209260213609006e-06,
      "loss": 4.8578,
      "step": 14920
    },
    {
      "epoch": 0.6023819136051676,
      "grad_norm": 71.50110626220703,
      "learning_rate": 7.2080046077620024e-06,
      "loss": 3.9924,
      "step": 14921
    },
    {
      "epoch": 0.6024222850222043,
      "grad_norm": 131.5434112548828,
      "learning_rate": 7.2067490496516e-06,
      "loss": 4.9666,
      "step": 14922
    },
    {
      "epoch": 0.602462656439241,
      "grad_norm": 89.99011993408203,
      "learning_rate": 7.205493539299266e-06,
      "loss": 4.3329,
      "step": 14923
    },
    {
      "epoch": 0.6025030278562777,
      "grad_norm": 97.60908508300781,
      "learning_rate": 7.2042380767264706e-06,
      "loss": 5.377,
      "step": 14924
    },
    {
      "epoch": 0.6025433992733145,
      "grad_norm": 143.6219024658203,
      "learning_rate": 7.202982661954671e-06,
      "loss": 5.2084,
      "step": 14925
    },
    {
      "epoch": 0.6025837706903512,
      "grad_norm": 156.86387634277344,
      "learning_rate": 7.2017272950053385e-06,
      "loss": 5.1965,
      "step": 14926
    },
    {
      "epoch": 0.602624142107388,
      "grad_norm": 117.09696197509766,
      "learning_rate": 7.200471975899934e-06,
      "loss": 6.7091,
      "step": 14927
    },
    {
      "epoch": 0.6026645135244247,
      "grad_norm": 110.54261016845703,
      "learning_rate": 7.199216704659926e-06,
      "loss": 5.0499,
      "step": 14928
    },
    {
      "epoch": 0.6027048849414615,
      "grad_norm": 71.85601043701172,
      "learning_rate": 7.197961481306768e-06,
      "loss": 4.3668,
      "step": 14929
    },
    {
      "epoch": 0.6027452563584982,
      "grad_norm": 169.5071258544922,
      "learning_rate": 7.196706305861925e-06,
      "loss": 6.0515,
      "step": 14930
    },
    {
      "epoch": 0.6027856277755349,
      "grad_norm": 185.81710815429688,
      "learning_rate": 7.195451178346857e-06,
      "loss": 3.9831,
      "step": 14931
    },
    {
      "epoch": 0.6028259991925716,
      "grad_norm": 80.29385375976562,
      "learning_rate": 7.194196098783031e-06,
      "loss": 4.1346,
      "step": 14932
    },
    {
      "epoch": 0.6028663706096083,
      "grad_norm": 146.7299041748047,
      "learning_rate": 7.192941067191894e-06,
      "loss": 3.3946,
      "step": 14933
    },
    {
      "epoch": 0.6029067420266452,
      "grad_norm": 86.07636260986328,
      "learning_rate": 7.19168608359491e-06,
      "loss": 5.6111,
      "step": 14934
    },
    {
      "epoch": 0.6029471134436819,
      "grad_norm": 120.4623794555664,
      "learning_rate": 7.190431148013536e-06,
      "loss": 4.9829,
      "step": 14935
    },
    {
      "epoch": 0.6029874848607186,
      "grad_norm": 67.96102905273438,
      "learning_rate": 7.18917626046923e-06,
      "loss": 4.6803,
      "step": 14936
    },
    {
      "epoch": 0.6030278562777553,
      "grad_norm": 134.66387939453125,
      "learning_rate": 7.1879214209834455e-06,
      "loss": 4.8962,
      "step": 14937
    },
    {
      "epoch": 0.6030682276947921,
      "grad_norm": 182.0010986328125,
      "learning_rate": 7.186666629577638e-06,
      "loss": 5.1195,
      "step": 14938
    },
    {
      "epoch": 0.6031085991118288,
      "grad_norm": 95.18441009521484,
      "learning_rate": 7.185411886273261e-06,
      "loss": 5.372,
      "step": 14939
    },
    {
      "epoch": 0.6031489705288655,
      "grad_norm": 120.4944839477539,
      "learning_rate": 7.184157191091772e-06,
      "loss": 4.917,
      "step": 14940
    },
    {
      "epoch": 0.6031893419459023,
      "grad_norm": 88.74324035644531,
      "learning_rate": 7.182902544054616e-06,
      "loss": 4.3126,
      "step": 14941
    },
    {
      "epoch": 0.6032297133629391,
      "grad_norm": 90.06103515625,
      "learning_rate": 7.181647945183248e-06,
      "loss": 7.0373,
      "step": 14942
    },
    {
      "epoch": 0.6032700847799758,
      "grad_norm": 80.98582458496094,
      "learning_rate": 7.180393394499119e-06,
      "loss": 5.865,
      "step": 14943
    },
    {
      "epoch": 0.6033104561970125,
      "grad_norm": 82.84232330322266,
      "learning_rate": 7.17913889202368e-06,
      "loss": 4.4173,
      "step": 14944
    },
    {
      "epoch": 0.6033508276140492,
      "grad_norm": 82.26911163330078,
      "learning_rate": 7.177884437778377e-06,
      "loss": 4.884,
      "step": 14945
    },
    {
      "epoch": 0.603391199031086,
      "grad_norm": 153.21788024902344,
      "learning_rate": 7.17663003178466e-06,
      "loss": 4.9982,
      "step": 14946
    },
    {
      "epoch": 0.6034315704481227,
      "grad_norm": 102.87208557128906,
      "learning_rate": 7.175375674063978e-06,
      "loss": 6.3289,
      "step": 14947
    },
    {
      "epoch": 0.6034719418651595,
      "grad_norm": 137.9168243408203,
      "learning_rate": 7.174121364637778e-06,
      "loss": 4.1369,
      "step": 14948
    },
    {
      "epoch": 0.6035123132821962,
      "grad_norm": 157.8976593017578,
      "learning_rate": 7.172867103527501e-06,
      "loss": 5.3734,
      "step": 14949
    },
    {
      "epoch": 0.603552684699233,
      "grad_norm": 221.5470428466797,
      "learning_rate": 7.171612890754594e-06,
      "loss": 4.5806,
      "step": 14950
    },
    {
      "epoch": 0.6035930561162697,
      "grad_norm": 81.55695343017578,
      "learning_rate": 7.170358726340501e-06,
      "loss": 4.9879,
      "step": 14951
    },
    {
      "epoch": 0.6036334275333064,
      "grad_norm": 111.16075134277344,
      "learning_rate": 7.169104610306672e-06,
      "loss": 5.8637,
      "step": 14952
    },
    {
      "epoch": 0.6036737989503431,
      "grad_norm": 71.1340560913086,
      "learning_rate": 7.167850542674538e-06,
      "loss": 3.5511,
      "step": 14953
    },
    {
      "epoch": 0.6037141703673798,
      "grad_norm": 91.14974975585938,
      "learning_rate": 7.166596523465547e-06,
      "loss": 5.0137,
      "step": 14954
    },
    {
      "epoch": 0.6037545417844167,
      "grad_norm": 146.660888671875,
      "learning_rate": 7.1653425527011375e-06,
      "loss": 5.5699,
      "step": 14955
    },
    {
      "epoch": 0.6037949132014534,
      "grad_norm": 89.78594207763672,
      "learning_rate": 7.164088630402754e-06,
      "loss": 5.1651,
      "step": 14956
    },
    {
      "epoch": 0.6038352846184901,
      "grad_norm": 123.10750579833984,
      "learning_rate": 7.162834756591832e-06,
      "loss": 5.2637,
      "step": 14957
    },
    {
      "epoch": 0.6038756560355268,
      "grad_norm": 121.54536437988281,
      "learning_rate": 7.161580931289809e-06,
      "loss": 4.5327,
      "step": 14958
    },
    {
      "epoch": 0.6039160274525636,
      "grad_norm": 89.19988250732422,
      "learning_rate": 7.160327154518124e-06,
      "loss": 5.4201,
      "step": 14959
    },
    {
      "epoch": 0.6039563988696003,
      "grad_norm": 100.1751480102539,
      "learning_rate": 7.159073426298217e-06,
      "loss": 6.1051,
      "step": 14960
    },
    {
      "epoch": 0.603996770286637,
      "grad_norm": 176.88685607910156,
      "learning_rate": 7.157819746651518e-06,
      "loss": 3.1111,
      "step": 14961
    },
    {
      "epoch": 0.6040371417036738,
      "grad_norm": 120.80036926269531,
      "learning_rate": 7.156566115599464e-06,
      "loss": 5.5431,
      "step": 14962
    },
    {
      "epoch": 0.6040775131207106,
      "grad_norm": 100.58596801757812,
      "learning_rate": 7.155312533163488e-06,
      "loss": 4.2067,
      "step": 14963
    },
    {
      "epoch": 0.6041178845377473,
      "grad_norm": 116.09379577636719,
      "learning_rate": 7.154058999365028e-06,
      "loss": 4.6202,
      "step": 14964
    },
    {
      "epoch": 0.604158255954784,
      "grad_norm": 92.4168472290039,
      "learning_rate": 7.1528055142255105e-06,
      "loss": 4.9813,
      "step": 14965
    },
    {
      "epoch": 0.6041986273718207,
      "grad_norm": 109.16825103759766,
      "learning_rate": 7.151552077766371e-06,
      "loss": 3.891,
      "step": 14966
    },
    {
      "epoch": 0.6042389987888575,
      "grad_norm": 114.79485321044922,
      "learning_rate": 7.150298690009039e-06,
      "loss": 3.9048,
      "step": 14967
    },
    {
      "epoch": 0.6042793702058942,
      "grad_norm": 76.9535903930664,
      "learning_rate": 7.149045350974948e-06,
      "loss": 6.518,
      "step": 14968
    },
    {
      "epoch": 0.604319741622931,
      "grad_norm": 104.61268615722656,
      "learning_rate": 7.1477920606855215e-06,
      "loss": 4.2811,
      "step": 14969
    },
    {
      "epoch": 0.6043601130399677,
      "grad_norm": 105.0789794921875,
      "learning_rate": 7.1465388191621894e-06,
      "loss": 6.4329,
      "step": 14970
    },
    {
      "epoch": 0.6044004844570044,
      "grad_norm": 112.11394500732422,
      "learning_rate": 7.1452856264263805e-06,
      "loss": 5.4994,
      "step": 14971
    },
    {
      "epoch": 0.6044408558740412,
      "grad_norm": 102.61295318603516,
      "learning_rate": 7.144032482499525e-06,
      "loss": 4.1328,
      "step": 14972
    },
    {
      "epoch": 0.6044812272910779,
      "grad_norm": 100.60836791992188,
      "learning_rate": 7.1427793874030415e-06,
      "loss": 4.0732,
      "step": 14973
    },
    {
      "epoch": 0.6045215987081146,
      "grad_norm": 97.19447326660156,
      "learning_rate": 7.141526341158358e-06,
      "loss": 5.2435,
      "step": 14974
    },
    {
      "epoch": 0.6045619701251513,
      "grad_norm": 76.68636322021484,
      "learning_rate": 7.140273343786899e-06,
      "loss": 4.8596,
      "step": 14975
    },
    {
      "epoch": 0.6046023415421882,
      "grad_norm": 84.1271743774414,
      "learning_rate": 7.1390203953100904e-06,
      "loss": 6.3982,
      "step": 14976
    },
    {
      "epoch": 0.6046427129592249,
      "grad_norm": 80.46593475341797,
      "learning_rate": 7.137767495749348e-06,
      "loss": 5.883,
      "step": 14977
    },
    {
      "epoch": 0.6046830843762616,
      "grad_norm": 146.3227996826172,
      "learning_rate": 7.1365146451261e-06,
      "loss": 4.8054,
      "step": 14978
    },
    {
      "epoch": 0.6047234557932983,
      "grad_norm": 199.2099609375,
      "learning_rate": 7.135261843461765e-06,
      "loss": 5.8994,
      "step": 14979
    },
    {
      "epoch": 0.6047638272103351,
      "grad_norm": 78.32441711425781,
      "learning_rate": 7.134009090777765e-06,
      "loss": 3.771,
      "step": 14980
    },
    {
      "epoch": 0.6048041986273718,
      "grad_norm": 152.9949951171875,
      "learning_rate": 7.132756387095514e-06,
      "loss": 5.0789,
      "step": 14981
    },
    {
      "epoch": 0.6048445700444086,
      "grad_norm": 77.74361419677734,
      "learning_rate": 7.1315037324364335e-06,
      "loss": 4.6662,
      "step": 14982
    },
    {
      "epoch": 0.6048849414614453,
      "grad_norm": 89.85750579833984,
      "learning_rate": 7.130251126821941e-06,
      "loss": 4.8677,
      "step": 14983
    },
    {
      "epoch": 0.6049253128784821,
      "grad_norm": 139.08551025390625,
      "learning_rate": 7.128998570273453e-06,
      "loss": 3.4461,
      "step": 14984
    },
    {
      "epoch": 0.6049656842955188,
      "grad_norm": 96.20110321044922,
      "learning_rate": 7.127746062812385e-06,
      "loss": 4.6068,
      "step": 14985
    },
    {
      "epoch": 0.6050060557125555,
      "grad_norm": 110.63602447509766,
      "learning_rate": 7.126493604460152e-06,
      "loss": 3.3462,
      "step": 14986
    },
    {
      "epoch": 0.6050464271295922,
      "grad_norm": 121.41498565673828,
      "learning_rate": 7.125241195238168e-06,
      "loss": 4.9364,
      "step": 14987
    },
    {
      "epoch": 0.605086798546629,
      "grad_norm": 94.06937408447266,
      "learning_rate": 7.12398883516785e-06,
      "loss": 5.948,
      "step": 14988
    },
    {
      "epoch": 0.6051271699636658,
      "grad_norm": 107.08733367919922,
      "learning_rate": 7.122736524270604e-06,
      "loss": 6.279,
      "step": 14989
    },
    {
      "epoch": 0.6051675413807025,
      "grad_norm": 194.56076049804688,
      "learning_rate": 7.1214842625678435e-06,
      "loss": 6.3327,
      "step": 14990
    },
    {
      "epoch": 0.6052079127977392,
      "grad_norm": 109.37356567382812,
      "learning_rate": 7.120232050080981e-06,
      "loss": 6.1448,
      "step": 14991
    },
    {
      "epoch": 0.605248284214776,
      "grad_norm": 131.60545349121094,
      "learning_rate": 7.118979886831427e-06,
      "loss": 4.1259,
      "step": 14992
    },
    {
      "epoch": 0.6052886556318127,
      "grad_norm": 135.0365447998047,
      "learning_rate": 7.117727772840588e-06,
      "loss": 2.918,
      "step": 14993
    },
    {
      "epoch": 0.6053290270488494,
      "grad_norm": 99.15843963623047,
      "learning_rate": 7.116475708129873e-06,
      "loss": 4.9166,
      "step": 14994
    },
    {
      "epoch": 0.6053693984658861,
      "grad_norm": 142.0458221435547,
      "learning_rate": 7.1152236927206896e-06,
      "loss": 5.6445,
      "step": 14995
    },
    {
      "epoch": 0.6054097698829229,
      "grad_norm": 95.31299591064453,
      "learning_rate": 7.11397172663445e-06,
      "loss": 4.581,
      "step": 14996
    },
    {
      "epoch": 0.6054501412999597,
      "grad_norm": 96.49421691894531,
      "learning_rate": 7.11271980989255e-06,
      "loss": 5.2179,
      "step": 14997
    },
    {
      "epoch": 0.6054905127169964,
      "grad_norm": 73.33808898925781,
      "learning_rate": 7.111467942516398e-06,
      "loss": 5.6854,
      "step": 14998
    },
    {
      "epoch": 0.6055308841340331,
      "grad_norm": 143.6119384765625,
      "learning_rate": 7.110216124527399e-06,
      "loss": 3.3702,
      "step": 14999
    },
    {
      "epoch": 0.6055712555510698,
      "grad_norm": 189.03330993652344,
      "learning_rate": 7.108964355946961e-06,
      "loss": 6.1737,
      "step": 15000
    },
    {
      "epoch": 0.6056116269681066,
      "grad_norm": 108.37264251708984,
      "learning_rate": 7.107712636796478e-06,
      "loss": 5.6602,
      "step": 15001
    },
    {
      "epoch": 0.6056519983851433,
      "grad_norm": 81.8187255859375,
      "learning_rate": 7.106460967097353e-06,
      "loss": 6.4594,
      "step": 15002
    },
    {
      "epoch": 0.6056923698021801,
      "grad_norm": 70.10076904296875,
      "learning_rate": 7.1052093468709896e-06,
      "loss": 5.7596,
      "step": 15003
    },
    {
      "epoch": 0.6057327412192168,
      "grad_norm": 120.16050720214844,
      "learning_rate": 7.103957776138789e-06,
      "loss": 7.1551,
      "step": 15004
    },
    {
      "epoch": 0.6057731126362536,
      "grad_norm": 115.00434112548828,
      "learning_rate": 7.102706254922145e-06,
      "loss": 4.9729,
      "step": 15005
    },
    {
      "epoch": 0.6058134840532903,
      "grad_norm": 78.17243957519531,
      "learning_rate": 7.101454783242459e-06,
      "loss": 5.5909,
      "step": 15006
    },
    {
      "epoch": 0.605853855470327,
      "grad_norm": 159.84437561035156,
      "learning_rate": 7.1002033611211274e-06,
      "loss": 4.8076,
      "step": 15007
    },
    {
      "epoch": 0.6058942268873637,
      "grad_norm": 84.899658203125,
      "learning_rate": 7.09895198857955e-06,
      "loss": 4.8401,
      "step": 15008
    },
    {
      "epoch": 0.6059345983044004,
      "grad_norm": 81.84186553955078,
      "learning_rate": 7.097700665639116e-06,
      "loss": 3.9917,
      "step": 15009
    },
    {
      "epoch": 0.6059749697214373,
      "grad_norm": 69.46520233154297,
      "learning_rate": 7.096449392321223e-06,
      "loss": 4.7115,
      "step": 15010
    },
    {
      "epoch": 0.606015341138474,
      "grad_norm": 170.3727569580078,
      "learning_rate": 7.095198168647266e-06,
      "loss": 2.6814,
      "step": 15011
    },
    {
      "epoch": 0.6060557125555107,
      "grad_norm": 486.4050598144531,
      "learning_rate": 7.093946994638638e-06,
      "loss": 4.4747,
      "step": 15012
    },
    {
      "epoch": 0.6060960839725474,
      "grad_norm": 106.24353790283203,
      "learning_rate": 7.092695870316728e-06,
      "loss": 5.4937,
      "step": 15013
    },
    {
      "epoch": 0.6061364553895842,
      "grad_norm": 116.05690002441406,
      "learning_rate": 7.091444795702931e-06,
      "loss": 5.4232,
      "step": 15014
    },
    {
      "epoch": 0.6061768268066209,
      "grad_norm": 98.32486724853516,
      "learning_rate": 7.090193770818634e-06,
      "loss": 4.6271,
      "step": 15015
    },
    {
      "epoch": 0.6062171982236576,
      "grad_norm": 104.54122161865234,
      "learning_rate": 7.088942795685234e-06,
      "loss": 5.5352,
      "step": 15016
    },
    {
      "epoch": 0.6062575696406944,
      "grad_norm": 116.64990234375,
      "learning_rate": 7.0876918703241115e-06,
      "loss": 6.299,
      "step": 15017
    },
    {
      "epoch": 0.6062979410577312,
      "grad_norm": 73.08465576171875,
      "learning_rate": 7.086440994756656e-06,
      "loss": 5.3473,
      "step": 15018
    },
    {
      "epoch": 0.6063383124747679,
      "grad_norm": 100.69449615478516,
      "learning_rate": 7.085190169004254e-06,
      "loss": 5.6228,
      "step": 15019
    },
    {
      "epoch": 0.6063786838918046,
      "grad_norm": 88.8982925415039,
      "learning_rate": 7.083939393088301e-06,
      "loss": 4.2677,
      "step": 15020
    },
    {
      "epoch": 0.6064190553088413,
      "grad_norm": 90.67371368408203,
      "learning_rate": 7.082688667030168e-06,
      "loss": 4.1207,
      "step": 15021
    },
    {
      "epoch": 0.606459426725878,
      "grad_norm": 95.91753387451172,
      "learning_rate": 7.081437990851249e-06,
      "loss": 6.6098,
      "step": 15022
    },
    {
      "epoch": 0.6064997981429148,
      "grad_norm": 112.06429290771484,
      "learning_rate": 7.080187364572924e-06,
      "loss": 1.9878,
      "step": 15023
    },
    {
      "epoch": 0.6065401695599516,
      "grad_norm": 195.41102600097656,
      "learning_rate": 7.078936788216577e-06,
      "loss": 4.7548,
      "step": 15024
    },
    {
      "epoch": 0.6065805409769883,
      "grad_norm": 196.0199432373047,
      "learning_rate": 7.07768626180359e-06,
      "loss": 5.7771,
      "step": 15025
    },
    {
      "epoch": 0.606620912394025,
      "grad_norm": 100.00682830810547,
      "learning_rate": 7.076435785355342e-06,
      "loss": 5.3483,
      "step": 15026
    },
    {
      "epoch": 0.6066612838110618,
      "grad_norm": 111.33900451660156,
      "learning_rate": 7.0751853588932175e-06,
      "loss": 5.6188,
      "step": 15027
    },
    {
      "epoch": 0.6067016552280985,
      "grad_norm": 71.66495513916016,
      "learning_rate": 7.073934982438595e-06,
      "loss": 5.1017,
      "step": 15028
    },
    {
      "epoch": 0.6067420266451352,
      "grad_norm": 74.81322479248047,
      "learning_rate": 7.072684656012849e-06,
      "loss": 4.2327,
      "step": 15029
    },
    {
      "epoch": 0.6067823980621719,
      "grad_norm": 105.4968490600586,
      "learning_rate": 7.0714343796373585e-06,
      "loss": 5.475,
      "step": 15030
    },
    {
      "epoch": 0.6068227694792088,
      "grad_norm": 396.6544494628906,
      "learning_rate": 7.070184153333503e-06,
      "loss": 4.2297,
      "step": 15031
    },
    {
      "epoch": 0.6068631408962455,
      "grad_norm": 105.63656616210938,
      "learning_rate": 7.068933977122658e-06,
      "loss": 4.4209,
      "step": 15032
    },
    {
      "epoch": 0.6069035123132822,
      "grad_norm": 73.05101013183594,
      "learning_rate": 7.0676838510261946e-06,
      "loss": 5.6309,
      "step": 15033
    },
    {
      "epoch": 0.6069438837303189,
      "grad_norm": 122.45840454101562,
      "learning_rate": 7.066433775065491e-06,
      "loss": 5.8194,
      "step": 15034
    },
    {
      "epoch": 0.6069842551473557,
      "grad_norm": 103.60682678222656,
      "learning_rate": 7.0651837492619195e-06,
      "loss": 5.3475,
      "step": 15035
    },
    {
      "epoch": 0.6070246265643924,
      "grad_norm": 92.52304077148438,
      "learning_rate": 7.063933773636856e-06,
      "loss": 4.3837,
      "step": 15036
    },
    {
      "epoch": 0.6070649979814291,
      "grad_norm": 113.2628402709961,
      "learning_rate": 7.062683848211664e-06,
      "loss": 4.3568,
      "step": 15037
    },
    {
      "epoch": 0.6071053693984659,
      "grad_norm": 84.08570861816406,
      "learning_rate": 7.061433973007721e-06,
      "loss": 3.1469,
      "step": 15038
    },
    {
      "epoch": 0.6071457408155027,
      "grad_norm": 91.70633697509766,
      "learning_rate": 7.060184148046394e-06,
      "loss": 3.9924,
      "step": 15039
    },
    {
      "epoch": 0.6071861122325394,
      "grad_norm": 113.00141143798828,
      "learning_rate": 7.0589343733490576e-06,
      "loss": 7.042,
      "step": 15040
    },
    {
      "epoch": 0.6072264836495761,
      "grad_norm": 118.94947814941406,
      "learning_rate": 7.057684648937071e-06,
      "loss": 6.2586,
      "step": 15041
    },
    {
      "epoch": 0.6072668550666128,
      "grad_norm": 90.29856872558594,
      "learning_rate": 7.056434974831806e-06,
      "loss": 3.513,
      "step": 15042
    },
    {
      "epoch": 0.6073072264836495,
      "grad_norm": 74.32392120361328,
      "learning_rate": 7.05518535105463e-06,
      "loss": 4.9583,
      "step": 15043
    },
    {
      "epoch": 0.6073475979006863,
      "grad_norm": 68.44154357910156,
      "learning_rate": 7.053935777626909e-06,
      "loss": 5.4492,
      "step": 15044
    },
    {
      "epoch": 0.6073879693177231,
      "grad_norm": 86.697021484375,
      "learning_rate": 7.052686254570005e-06,
      "loss": 4.8266,
      "step": 15045
    },
    {
      "epoch": 0.6074283407347598,
      "grad_norm": 91.7562255859375,
      "learning_rate": 7.051436781905283e-06,
      "loss": 4.7581,
      "step": 15046
    },
    {
      "epoch": 0.6074687121517965,
      "grad_norm": 98.76973724365234,
      "learning_rate": 7.050187359654107e-06,
      "loss": 5.7255,
      "step": 15047
    },
    {
      "epoch": 0.6075090835688333,
      "grad_norm": 739.8019409179688,
      "learning_rate": 7.048937987837843e-06,
      "loss": 4.7287,
      "step": 15048
    },
    {
      "epoch": 0.60754945498587,
      "grad_norm": 82.19316101074219,
      "learning_rate": 7.047688666477844e-06,
      "loss": 5.6566,
      "step": 15049
    },
    {
      "epoch": 0.6075898264029067,
      "grad_norm": 93.4500732421875,
      "learning_rate": 7.046439395595475e-06,
      "loss": 6.0844,
      "step": 15050
    },
    {
      "epoch": 0.6076301978199434,
      "grad_norm": 79.85862731933594,
      "learning_rate": 7.045190175212096e-06,
      "loss": 3.454,
      "step": 15051
    },
    {
      "epoch": 0.6076705692369803,
      "grad_norm": 131.24290466308594,
      "learning_rate": 7.0439410053490665e-06,
      "loss": 4.4497,
      "step": 15052
    },
    {
      "epoch": 0.607710940654017,
      "grad_norm": 82.7418212890625,
      "learning_rate": 7.042691886027739e-06,
      "loss": 4.7243,
      "step": 15053
    },
    {
      "epoch": 0.6077513120710537,
      "grad_norm": 123.53475952148438,
      "learning_rate": 7.0414428172694774e-06,
      "loss": 5.9281,
      "step": 15054
    },
    {
      "epoch": 0.6077916834880904,
      "grad_norm": 91.18305969238281,
      "learning_rate": 7.0401937990956326e-06,
      "loss": 4.3047,
      "step": 15055
    },
    {
      "epoch": 0.6078320549051272,
      "grad_norm": 133.07579040527344,
      "learning_rate": 7.038944831527568e-06,
      "loss": 5.3134,
      "step": 15056
    },
    {
      "epoch": 0.6078724263221639,
      "grad_norm": 108.92890930175781,
      "learning_rate": 7.037695914586626e-06,
      "loss": 5.2298,
      "step": 15057
    },
    {
      "epoch": 0.6079127977392006,
      "grad_norm": 78.38208770751953,
      "learning_rate": 7.036447048294168e-06,
      "loss": 3.4015,
      "step": 15058
    },
    {
      "epoch": 0.6079531691562374,
      "grad_norm": 97.7892837524414,
      "learning_rate": 7.035198232671545e-06,
      "loss": 7.3721,
      "step": 15059
    },
    {
      "epoch": 0.6079935405732741,
      "grad_norm": 100.7223129272461,
      "learning_rate": 7.033949467740113e-06,
      "loss": 5.3317,
      "step": 15060
    },
    {
      "epoch": 0.6080339119903109,
      "grad_norm": 123.06074523925781,
      "learning_rate": 7.032700753521214e-06,
      "loss": 5.9661,
      "step": 15061
    },
    {
      "epoch": 0.6080742834073476,
      "grad_norm": 102.65020751953125,
      "learning_rate": 7.031452090036204e-06,
      "loss": 6.9418,
      "step": 15062
    },
    {
      "epoch": 0.6081146548243843,
      "grad_norm": 76.90247344970703,
      "learning_rate": 7.03020347730643e-06,
      "loss": 5.3207,
      "step": 15063
    },
    {
      "epoch": 0.608155026241421,
      "grad_norm": 100.63516998291016,
      "learning_rate": 7.028954915353246e-06,
      "loss": 4.9852,
      "step": 15064
    },
    {
      "epoch": 0.6081953976584578,
      "grad_norm": 81.38015747070312,
      "learning_rate": 7.02770640419799e-06,
      "loss": 5.1465,
      "step": 15065
    },
    {
      "epoch": 0.6082357690754946,
      "grad_norm": 157.447509765625,
      "learning_rate": 7.026457943862015e-06,
      "loss": 6.4813,
      "step": 15066
    },
    {
      "epoch": 0.6082761404925313,
      "grad_norm": 88.07406616210938,
      "learning_rate": 7.025209534366666e-06,
      "loss": 3.9055,
      "step": 15067
    },
    {
      "epoch": 0.608316511909568,
      "grad_norm": 82.61373138427734,
      "learning_rate": 7.02396117573329e-06,
      "loss": 6.5142,
      "step": 15068
    },
    {
      "epoch": 0.6083568833266048,
      "grad_norm": 121.04622650146484,
      "learning_rate": 7.0227128679832256e-06,
      "loss": 4.1699,
      "step": 15069
    },
    {
      "epoch": 0.6083972547436415,
      "grad_norm": 95.14502716064453,
      "learning_rate": 7.021464611137818e-06,
      "loss": 4.9771,
      "step": 15070
    },
    {
      "epoch": 0.6084376261606782,
      "grad_norm": 113.2566909790039,
      "learning_rate": 7.020216405218411e-06,
      "loss": 5.828,
      "step": 15071
    },
    {
      "epoch": 0.6084779975777149,
      "grad_norm": 85.72364807128906,
      "learning_rate": 7.018968250246347e-06,
      "loss": 4.431,
      "step": 15072
    },
    {
      "epoch": 0.6085183689947518,
      "grad_norm": 110.73825073242188,
      "learning_rate": 7.017720146242961e-06,
      "loss": 4.3566,
      "step": 15073
    },
    {
      "epoch": 0.6085587404117885,
      "grad_norm": 183.1453094482422,
      "learning_rate": 7.016472093229598e-06,
      "loss": 4.5799,
      "step": 15074
    },
    {
      "epoch": 0.6085991118288252,
      "grad_norm": 144.80690002441406,
      "learning_rate": 7.015224091227595e-06,
      "loss": 5.2596,
      "step": 15075
    },
    {
      "epoch": 0.6086394832458619,
      "grad_norm": 86.83731842041016,
      "learning_rate": 7.013976140258294e-06,
      "loss": 4.7936,
      "step": 15076
    },
    {
      "epoch": 0.6086798546628986,
      "grad_norm": 76.53998565673828,
      "learning_rate": 7.012728240343024e-06,
      "loss": 4.5678,
      "step": 15077
    },
    {
      "epoch": 0.6087202260799354,
      "grad_norm": 93.15299987792969,
      "learning_rate": 7.011480391503125e-06,
      "loss": 4.0091,
      "step": 15078
    },
    {
      "epoch": 0.6087605974969721,
      "grad_norm": 86.58353424072266,
      "learning_rate": 7.010232593759934e-06,
      "loss": 5.7821,
      "step": 15079
    },
    {
      "epoch": 0.6088009689140089,
      "grad_norm": 134.26271057128906,
      "learning_rate": 7.008984847134787e-06,
      "loss": 4.5947,
      "step": 15080
    },
    {
      "epoch": 0.6088413403310456,
      "grad_norm": 75.46617126464844,
      "learning_rate": 7.00773715164901e-06,
      "loss": 5.7441,
      "step": 15081
    },
    {
      "epoch": 0.6088817117480824,
      "grad_norm": 79.44519805908203,
      "learning_rate": 7.006489507323942e-06,
      "loss": 6.3376,
      "step": 15082
    },
    {
      "epoch": 0.6089220831651191,
      "grad_norm": 139.2724609375,
      "learning_rate": 7.0052419141809115e-06,
      "loss": 4.5153,
      "step": 15083
    },
    {
      "epoch": 0.6089624545821558,
      "grad_norm": 82.75340270996094,
      "learning_rate": 7.003994372241253e-06,
      "loss": 5.9018,
      "step": 15084
    },
    {
      "epoch": 0.6090028259991925,
      "grad_norm": 844.4374389648438,
      "learning_rate": 7.002746881526292e-06,
      "loss": 4.8892,
      "step": 15085
    },
    {
      "epoch": 0.6090431974162294,
      "grad_norm": 94.32119750976562,
      "learning_rate": 7.001499442057362e-06,
      "loss": 5.3969,
      "step": 15086
    },
    {
      "epoch": 0.6090835688332661,
      "grad_norm": 108.87174224853516,
      "learning_rate": 7.000252053855788e-06,
      "loss": 4.3909,
      "step": 15087
    },
    {
      "epoch": 0.6091239402503028,
      "grad_norm": 179.68698120117188,
      "learning_rate": 6.999004716942904e-06,
      "loss": 4.021,
      "step": 15088
    },
    {
      "epoch": 0.6091643116673395,
      "grad_norm": 103.8793716430664,
      "learning_rate": 6.997757431340027e-06,
      "loss": 5.8429,
      "step": 15089
    },
    {
      "epoch": 0.6092046830843763,
      "grad_norm": 88.47335052490234,
      "learning_rate": 6.996510197068488e-06,
      "loss": 4.3741,
      "step": 15090
    },
    {
      "epoch": 0.609245054501413,
      "grad_norm": 97.12983703613281,
      "learning_rate": 6.99526301414961e-06,
      "loss": 4.8158,
      "step": 15091
    },
    {
      "epoch": 0.6092854259184497,
      "grad_norm": 81.7937240600586,
      "learning_rate": 6.9940158826047215e-06,
      "loss": 5.9871,
      "step": 15092
    },
    {
      "epoch": 0.6093257973354865,
      "grad_norm": 65.62971496582031,
      "learning_rate": 6.992768802455139e-06,
      "loss": 5.5596,
      "step": 15093
    },
    {
      "epoch": 0.6093661687525233,
      "grad_norm": 153.6053466796875,
      "learning_rate": 6.991521773722186e-06,
      "loss": 5.3202,
      "step": 15094
    },
    {
      "epoch": 0.60940654016956,
      "grad_norm": 81.53184509277344,
      "learning_rate": 6.990274796427188e-06,
      "loss": 4.1742,
      "step": 15095
    },
    {
      "epoch": 0.6094469115865967,
      "grad_norm": 157.23294067382812,
      "learning_rate": 6.989027870591465e-06,
      "loss": 5.3905,
      "step": 15096
    },
    {
      "epoch": 0.6094872830036334,
      "grad_norm": 108.30322265625,
      "learning_rate": 6.987780996236331e-06,
      "loss": 5.45,
      "step": 15097
    },
    {
      "epoch": 0.6095276544206701,
      "grad_norm": 77.45555114746094,
      "learning_rate": 6.986534173383107e-06,
      "loss": 4.8746,
      "step": 15098
    },
    {
      "epoch": 0.6095680258377069,
      "grad_norm": 110.1894302368164,
      "learning_rate": 6.985287402053113e-06,
      "loss": 4.5261,
      "step": 15099
    },
    {
      "epoch": 0.6096083972547437,
      "grad_norm": 136.19674682617188,
      "learning_rate": 6.984040682267665e-06,
      "loss": 5.6805,
      "step": 15100
    },
    {
      "epoch": 0.6096487686717804,
      "grad_norm": 92.21732330322266,
      "learning_rate": 6.9827940140480776e-06,
      "loss": 4.9299,
      "step": 15101
    },
    {
      "epoch": 0.6096891400888171,
      "grad_norm": 99.1090316772461,
      "learning_rate": 6.981547397415667e-06,
      "loss": 5.7496,
      "step": 15102
    },
    {
      "epoch": 0.6097295115058539,
      "grad_norm": 83.1258544921875,
      "learning_rate": 6.980300832391746e-06,
      "loss": 5.8244,
      "step": 15103
    },
    {
      "epoch": 0.6097698829228906,
      "grad_norm": 67.05236053466797,
      "learning_rate": 6.979054318997636e-06,
      "loss": 5.3187,
      "step": 15104
    },
    {
      "epoch": 0.6098102543399273,
      "grad_norm": 93.47991180419922,
      "learning_rate": 6.977807857254637e-06,
      "loss": 5.4586,
      "step": 15105
    },
    {
      "epoch": 0.609850625756964,
      "grad_norm": 92.6666488647461,
      "learning_rate": 6.976561447184066e-06,
      "loss": 4.4477,
      "step": 15106
    },
    {
      "epoch": 0.6098909971740009,
      "grad_norm": 99.28629302978516,
      "learning_rate": 6.975315088807234e-06,
      "loss": 6.0028,
      "step": 15107
    },
    {
      "epoch": 0.6099313685910376,
      "grad_norm": 161.82249450683594,
      "learning_rate": 6.974068782145455e-06,
      "loss": 3.9892,
      "step": 15108
    },
    {
      "epoch": 0.6099717400080743,
      "grad_norm": 71.29645538330078,
      "learning_rate": 6.972822527220031e-06,
      "loss": 5.707,
      "step": 15109
    },
    {
      "epoch": 0.610012111425111,
      "grad_norm": 75.9632797241211,
      "learning_rate": 6.971576324052272e-06,
      "loss": 5.6912,
      "step": 15110
    },
    {
      "epoch": 0.6100524828421477,
      "grad_norm": 88.53861999511719,
      "learning_rate": 6.970330172663486e-06,
      "loss": 5.3203,
      "step": 15111
    },
    {
      "epoch": 0.6100928542591845,
      "grad_norm": 98.1282730102539,
      "learning_rate": 6.969084073074981e-06,
      "loss": 6.1947,
      "step": 15112
    },
    {
      "epoch": 0.6101332256762212,
      "grad_norm": 122.84011840820312,
      "learning_rate": 6.967838025308058e-06,
      "loss": 5.5479,
      "step": 15113
    },
    {
      "epoch": 0.610173597093258,
      "grad_norm": 88.92857360839844,
      "learning_rate": 6.966592029384025e-06,
      "loss": 4.7228,
      "step": 15114
    },
    {
      "epoch": 0.6102139685102947,
      "grad_norm": 102.79180145263672,
      "learning_rate": 6.965346085324185e-06,
      "loss": 5.6471,
      "step": 15115
    },
    {
      "epoch": 0.6102543399273315,
      "grad_norm": 102.54583740234375,
      "learning_rate": 6.964100193149843e-06,
      "loss": 5.5038,
      "step": 15116
    },
    {
      "epoch": 0.6102947113443682,
      "grad_norm": 172.59866333007812,
      "learning_rate": 6.962854352882295e-06,
      "loss": 6.1496,
      "step": 15117
    },
    {
      "epoch": 0.6103350827614049,
      "grad_norm": 78.25037384033203,
      "learning_rate": 6.961608564542848e-06,
      "loss": 3.8741,
      "step": 15118
    },
    {
      "epoch": 0.6103754541784416,
      "grad_norm": 88.60417175292969,
      "learning_rate": 6.960362828152796e-06,
      "loss": 6.7863,
      "step": 15119
    },
    {
      "epoch": 0.6104158255954784,
      "grad_norm": 127.10777282714844,
      "learning_rate": 6.959117143733445e-06,
      "loss": 3.5132,
      "step": 15120
    },
    {
      "epoch": 0.6104561970125152,
      "grad_norm": 84.58621978759766,
      "learning_rate": 6.957871511306088e-06,
      "loss": 6.0046,
      "step": 15121
    },
    {
      "epoch": 0.6104965684295519,
      "grad_norm": 84.90023803710938,
      "learning_rate": 6.956625930892023e-06,
      "loss": 5.991,
      "step": 15122
    },
    {
      "epoch": 0.6105369398465886,
      "grad_norm": 96.15371704101562,
      "learning_rate": 6.955380402512549e-06,
      "loss": 5.1861,
      "step": 15123
    },
    {
      "epoch": 0.6105773112636254,
      "grad_norm": 78.1334228515625,
      "learning_rate": 6.9541349261889645e-06,
      "loss": 4.0833,
      "step": 15124
    },
    {
      "epoch": 0.6106176826806621,
      "grad_norm": 82.61891174316406,
      "learning_rate": 6.952889501942557e-06,
      "loss": 5.5158,
      "step": 15125
    },
    {
      "epoch": 0.6106580540976988,
      "grad_norm": 134.23106384277344,
      "learning_rate": 6.951644129794621e-06,
      "loss": 4.79,
      "step": 15126
    },
    {
      "epoch": 0.6106984255147355,
      "grad_norm": 92.37723541259766,
      "learning_rate": 6.9503988097664545e-06,
      "loss": 5.2561,
      "step": 15127
    },
    {
      "epoch": 0.6107387969317724,
      "grad_norm": 112.17794036865234,
      "learning_rate": 6.94915354187935e-06,
      "loss": 5.1316,
      "step": 15128
    },
    {
      "epoch": 0.6107791683488091,
      "grad_norm": 72.70626831054688,
      "learning_rate": 6.947908326154591e-06,
      "loss": 4.7266,
      "step": 15129
    },
    {
      "epoch": 0.6108195397658458,
      "grad_norm": 95.03360748291016,
      "learning_rate": 6.946663162613473e-06,
      "loss": 5.9162,
      "step": 15130
    },
    {
      "epoch": 0.6108599111828825,
      "grad_norm": 83.19910430908203,
      "learning_rate": 6.945418051277284e-06,
      "loss": 6.2608,
      "step": 15131
    },
    {
      "epoch": 0.6109002825999192,
      "grad_norm": 83.55135345458984,
      "learning_rate": 6.944172992167318e-06,
      "loss": 4.9697,
      "step": 15132
    },
    {
      "epoch": 0.610940654016956,
      "grad_norm": 67.71369171142578,
      "learning_rate": 6.9429279853048525e-06,
      "loss": 4.0572,
      "step": 15133
    },
    {
      "epoch": 0.6109810254339927,
      "grad_norm": 76.66702270507812,
      "learning_rate": 6.941683030711181e-06,
      "loss": 5.6106,
      "step": 15134
    },
    {
      "epoch": 0.6110213968510295,
      "grad_norm": 85.84552764892578,
      "learning_rate": 6.94043812840759e-06,
      "loss": 5.2536,
      "step": 15135
    },
    {
      "epoch": 0.6110617682680662,
      "grad_norm": 104.89287567138672,
      "learning_rate": 6.939193278415364e-06,
      "loss": 4.5622,
      "step": 15136
    },
    {
      "epoch": 0.611102139685103,
      "grad_norm": 80.54127502441406,
      "learning_rate": 6.937948480755783e-06,
      "loss": 5.7171,
      "step": 15137
    },
    {
      "epoch": 0.6111425111021397,
      "grad_norm": 76.27001190185547,
      "learning_rate": 6.936703735450131e-06,
      "loss": 5.3824,
      "step": 15138
    },
    {
      "epoch": 0.6111828825191764,
      "grad_norm": 101.58572387695312,
      "learning_rate": 6.935459042519692e-06,
      "loss": 5.3497,
      "step": 15139
    },
    {
      "epoch": 0.6112232539362131,
      "grad_norm": 145.7339630126953,
      "learning_rate": 6.93421440198575e-06,
      "loss": 5.3217,
      "step": 15140
    },
    {
      "epoch": 0.6112636253532499,
      "grad_norm": 105.39871215820312,
      "learning_rate": 6.932969813869581e-06,
      "loss": 5.564,
      "step": 15141
    },
    {
      "epoch": 0.6113039967702867,
      "grad_norm": 93.3230209350586,
      "learning_rate": 6.931725278192465e-06,
      "loss": 4.888,
      "step": 15142
    },
    {
      "epoch": 0.6113443681873234,
      "grad_norm": 129.59999084472656,
      "learning_rate": 6.930480794975682e-06,
      "loss": 4.5375,
      "step": 15143
    },
    {
      "epoch": 0.6113847396043601,
      "grad_norm": 98.91693878173828,
      "learning_rate": 6.929236364240515e-06,
      "loss": 4.6027,
      "step": 15144
    },
    {
      "epoch": 0.6114251110213968,
      "grad_norm": 68.29186248779297,
      "learning_rate": 6.927991986008229e-06,
      "loss": 4.3859,
      "step": 15145
    },
    {
      "epoch": 0.6114654824384336,
      "grad_norm": 99.36985778808594,
      "learning_rate": 6.926747660300109e-06,
      "loss": 6.3933,
      "step": 15146
    },
    {
      "epoch": 0.6115058538554703,
      "grad_norm": 202.35415649414062,
      "learning_rate": 6.925503387137426e-06,
      "loss": 4.8104,
      "step": 15147
    },
    {
      "epoch": 0.611546225272507,
      "grad_norm": 81.34494018554688,
      "learning_rate": 6.924259166541459e-06,
      "loss": 4.7717,
      "step": 15148
    },
    {
      "epoch": 0.6115865966895438,
      "grad_norm": 123.78457641601562,
      "learning_rate": 6.923014998533475e-06,
      "loss": 4.9107,
      "step": 15149
    },
    {
      "epoch": 0.6116269681065806,
      "grad_norm": 102.37825012207031,
      "learning_rate": 6.9217708831347484e-06,
      "loss": 3.5554,
      "step": 15150
    },
    {
      "epoch": 0.6116673395236173,
      "grad_norm": 130.73829650878906,
      "learning_rate": 6.920526820366552e-06,
      "loss": 5.6139,
      "step": 15151
    },
    {
      "epoch": 0.611707710940654,
      "grad_norm": 81.71302795410156,
      "learning_rate": 6.9192828102501586e-06,
      "loss": 6.1515,
      "step": 15152
    },
    {
      "epoch": 0.6117480823576907,
      "grad_norm": 72.27137756347656,
      "learning_rate": 6.918038852806832e-06,
      "loss": 6.1037,
      "step": 15153
    },
    {
      "epoch": 0.6117884537747275,
      "grad_norm": 82.39888763427734,
      "learning_rate": 6.916794948057844e-06,
      "loss": 3.5845,
      "step": 15154
    },
    {
      "epoch": 0.6118288251917642,
      "grad_norm": 119.07827758789062,
      "learning_rate": 6.915551096024464e-06,
      "loss": 4.7029,
      "step": 15155
    },
    {
      "epoch": 0.611869196608801,
      "grad_norm": 86.65225219726562,
      "learning_rate": 6.91430729672796e-06,
      "loss": 4.8743,
      "step": 15156
    },
    {
      "epoch": 0.6119095680258377,
      "grad_norm": 122.87557983398438,
      "learning_rate": 6.913063550189592e-06,
      "loss": 4.1063,
      "step": 15157
    },
    {
      "epoch": 0.6119499394428745,
      "grad_norm": 80.89852142333984,
      "learning_rate": 6.91181985643063e-06,
      "loss": 4.2182,
      "step": 15158
    },
    {
      "epoch": 0.6119903108599112,
      "grad_norm": 79.88987731933594,
      "learning_rate": 6.910576215472335e-06,
      "loss": 6.409,
      "step": 15159
    },
    {
      "epoch": 0.6120306822769479,
      "grad_norm": 71.9741439819336,
      "learning_rate": 6.909332627335974e-06,
      "loss": 4.839,
      "step": 15160
    },
    {
      "epoch": 0.6120710536939846,
      "grad_norm": 93.67591094970703,
      "learning_rate": 6.908089092042809e-06,
      "loss": 3.9449,
      "step": 15161
    },
    {
      "epoch": 0.6121114251110213,
      "grad_norm": 75.66788482666016,
      "learning_rate": 6.906845609614099e-06,
      "loss": 3.8105,
      "step": 15162
    },
    {
      "epoch": 0.6121517965280582,
      "grad_norm": 103.9827880859375,
      "learning_rate": 6.905602180071106e-06,
      "loss": 6.8336,
      "step": 15163
    },
    {
      "epoch": 0.6121921679450949,
      "grad_norm": 67.09298706054688,
      "learning_rate": 6.9043588034350896e-06,
      "loss": 3.1348,
      "step": 15164
    },
    {
      "epoch": 0.6122325393621316,
      "grad_norm": 80.63284301757812,
      "learning_rate": 6.903115479727313e-06,
      "loss": 5.0721,
      "step": 15165
    },
    {
      "epoch": 0.6122729107791683,
      "grad_norm": 82.49278259277344,
      "learning_rate": 6.901872208969026e-06,
      "loss": 5.9184,
      "step": 15166
    },
    {
      "epoch": 0.6123132821962051,
      "grad_norm": 79.43946838378906,
      "learning_rate": 6.900628991181488e-06,
      "loss": 4.7268,
      "step": 15167
    },
    {
      "epoch": 0.6123536536132418,
      "grad_norm": 57.06392288208008,
      "learning_rate": 6.8993858263859595e-06,
      "loss": 4.9646,
      "step": 15168
    },
    {
      "epoch": 0.6123940250302785,
      "grad_norm": 102.5808334350586,
      "learning_rate": 6.898142714603695e-06,
      "loss": 4.5576,
      "step": 15169
    },
    {
      "epoch": 0.6124343964473153,
      "grad_norm": 93.43791961669922,
      "learning_rate": 6.896899655855943e-06,
      "loss": 6.9038,
      "step": 15170
    },
    {
      "epoch": 0.6124747678643521,
      "grad_norm": 78.49934387207031,
      "learning_rate": 6.895656650163961e-06,
      "loss": 4.2937,
      "step": 15171
    },
    {
      "epoch": 0.6125151392813888,
      "grad_norm": 87.853515625,
      "learning_rate": 6.894413697549001e-06,
      "loss": 4.3694,
      "step": 15172
    },
    {
      "epoch": 0.6125555106984255,
      "grad_norm": 86.14189147949219,
      "learning_rate": 6.893170798032316e-06,
      "loss": 6.5838,
      "step": 15173
    },
    {
      "epoch": 0.6125958821154622,
      "grad_norm": 110.67008972167969,
      "learning_rate": 6.891927951635155e-06,
      "loss": 5.562,
      "step": 15174
    },
    {
      "epoch": 0.612636253532499,
      "grad_norm": 133.58421325683594,
      "learning_rate": 6.890685158378767e-06,
      "loss": 6.1736,
      "step": 15175
    },
    {
      "epoch": 0.6126766249495357,
      "grad_norm": 94.73664855957031,
      "learning_rate": 6.889442418284402e-06,
      "loss": 5.4153,
      "step": 15176
    },
    {
      "epoch": 0.6127169963665725,
      "grad_norm": 81.26744842529297,
      "learning_rate": 6.8881997313733115e-06,
      "loss": 5.0756,
      "step": 15177
    },
    {
      "epoch": 0.6127573677836092,
      "grad_norm": 82.79173278808594,
      "learning_rate": 6.886957097666735e-06,
      "loss": 4.4645,
      "step": 15178
    },
    {
      "epoch": 0.612797739200646,
      "grad_norm": 119.53426361083984,
      "learning_rate": 6.885714517185924e-06,
      "loss": 6.1033,
      "step": 15179
    },
    {
      "epoch": 0.6128381106176827,
      "grad_norm": 84.23468017578125,
      "learning_rate": 6.884471989952122e-06,
      "loss": 5.3482,
      "step": 15180
    },
    {
      "epoch": 0.6128784820347194,
      "grad_norm": 58.61577606201172,
      "learning_rate": 6.883229515986576e-06,
      "loss": 3.6206,
      "step": 15181
    },
    {
      "epoch": 0.6129188534517561,
      "grad_norm": 94.0147933959961,
      "learning_rate": 6.881987095310524e-06,
      "loss": 5.1862,
      "step": 15182
    },
    {
      "epoch": 0.6129592248687928,
      "grad_norm": 86.74114990234375,
      "learning_rate": 6.880744727945212e-06,
      "loss": 3.9512,
      "step": 15183
    },
    {
      "epoch": 0.6129995962858297,
      "grad_norm": 78.88015747070312,
      "learning_rate": 6.87950241391188e-06,
      "loss": 4.8971,
      "step": 15184
    },
    {
      "epoch": 0.6130399677028664,
      "grad_norm": 154.11380004882812,
      "learning_rate": 6.878260153231776e-06,
      "loss": 3.7046,
      "step": 15185
    },
    {
      "epoch": 0.6130803391199031,
      "grad_norm": 103.55506134033203,
      "learning_rate": 6.877017945926129e-06,
      "loss": 5.277,
      "step": 15186
    },
    {
      "epoch": 0.6131207105369398,
      "grad_norm": 127.71022033691406,
      "learning_rate": 6.8757757920161815e-06,
      "loss": 5.1575,
      "step": 15187
    },
    {
      "epoch": 0.6131610819539766,
      "grad_norm": 68.75235748291016,
      "learning_rate": 6.874533691523173e-06,
      "loss": 5.8909,
      "step": 15188
    },
    {
      "epoch": 0.6132014533710133,
      "grad_norm": 79.62515258789062,
      "learning_rate": 6.873291644468343e-06,
      "loss": 4.1595,
      "step": 15189
    },
    {
      "epoch": 0.61324182478805,
      "grad_norm": 161.0369110107422,
      "learning_rate": 6.872049650872923e-06,
      "loss": 6.6076,
      "step": 15190
    },
    {
      "epoch": 0.6132821962050868,
      "grad_norm": 72.38026428222656,
      "learning_rate": 6.870807710758148e-06,
      "loss": 4.4504,
      "step": 15191
    },
    {
      "epoch": 0.6133225676221236,
      "grad_norm": 92.7933578491211,
      "learning_rate": 6.869565824145253e-06,
      "loss": 4.0564,
      "step": 15192
    },
    {
      "epoch": 0.6133629390391603,
      "grad_norm": 98.10575866699219,
      "learning_rate": 6.868323991055477e-06,
      "loss": 4.3405,
      "step": 15193
    },
    {
      "epoch": 0.613403310456197,
      "grad_norm": 65.36554718017578,
      "learning_rate": 6.867082211510043e-06,
      "loss": 5.0778,
      "step": 15194
    },
    {
      "epoch": 0.6134436818732337,
      "grad_norm": 79.92652893066406,
      "learning_rate": 6.865840485530188e-06,
      "loss": 4.4588,
      "step": 15195
    },
    {
      "epoch": 0.6134840532902704,
      "grad_norm": 97.40131378173828,
      "learning_rate": 6.864598813137142e-06,
      "loss": 5.1838,
      "step": 15196
    },
    {
      "epoch": 0.6135244247073073,
      "grad_norm": 88.97731018066406,
      "learning_rate": 6.863357194352139e-06,
      "loss": 5.0693,
      "step": 15197
    },
    {
      "epoch": 0.613564796124344,
      "grad_norm": 110.11348724365234,
      "learning_rate": 6.8621156291964e-06,
      "loss": 6.2827,
      "step": 15198
    },
    {
      "epoch": 0.6136051675413807,
      "grad_norm": 108.91427612304688,
      "learning_rate": 6.860874117691155e-06,
      "loss": 5.4324,
      "step": 15199
    },
    {
      "epoch": 0.6136455389584174,
      "grad_norm": 93.18843078613281,
      "learning_rate": 6.859632659857634e-06,
      "loss": 5.0084,
      "step": 15200
    },
    {
      "epoch": 0.6136859103754542,
      "grad_norm": 151.41197204589844,
      "learning_rate": 6.85839125571706e-06,
      "loss": 4.7476,
      "step": 15201
    },
    {
      "epoch": 0.6137262817924909,
      "grad_norm": 104.83277130126953,
      "learning_rate": 6.857149905290661e-06,
      "loss": 3.5198,
      "step": 15202
    },
    {
      "epoch": 0.6137666532095276,
      "grad_norm": 104.25192260742188,
      "learning_rate": 6.8559086085996576e-06,
      "loss": 5.5396,
      "step": 15203
    },
    {
      "epoch": 0.6138070246265644,
      "grad_norm": 61.35212326049805,
      "learning_rate": 6.854667365665275e-06,
      "loss": 3.1794,
      "step": 15204
    },
    {
      "epoch": 0.6138473960436012,
      "grad_norm": 113.508056640625,
      "learning_rate": 6.853426176508739e-06,
      "loss": 5.4057,
      "step": 15205
    },
    {
      "epoch": 0.6138877674606379,
      "grad_norm": 87.00123596191406,
      "learning_rate": 6.8521850411512645e-06,
      "loss": 5.5162,
      "step": 15206
    },
    {
      "epoch": 0.6139281388776746,
      "grad_norm": 94.0113754272461,
      "learning_rate": 6.850943959614076e-06,
      "loss": 5.5558,
      "step": 15207
    },
    {
      "epoch": 0.6139685102947113,
      "grad_norm": 126.33253479003906,
      "learning_rate": 6.849702931918391e-06,
      "loss": 4.8885,
      "step": 15208
    },
    {
      "epoch": 0.6140088817117481,
      "grad_norm": 72.10474395751953,
      "learning_rate": 6.848461958085431e-06,
      "loss": 4.7403,
      "step": 15209
    },
    {
      "epoch": 0.6140492531287848,
      "grad_norm": 88.69518280029297,
      "learning_rate": 6.847221038136411e-06,
      "loss": 4.1524,
      "step": 15210
    },
    {
      "epoch": 0.6140896245458216,
      "grad_norm": 308.5381164550781,
      "learning_rate": 6.845980172092549e-06,
      "loss": 4.8907,
      "step": 15211
    },
    {
      "epoch": 0.6141299959628583,
      "grad_norm": 109.91422271728516,
      "learning_rate": 6.844739359975061e-06,
      "loss": 6.1099,
      "step": 15212
    },
    {
      "epoch": 0.614170367379895,
      "grad_norm": 108.9521484375,
      "learning_rate": 6.843498601805165e-06,
      "loss": 5.2748,
      "step": 15213
    },
    {
      "epoch": 0.6142107387969318,
      "grad_norm": 99.45661163330078,
      "learning_rate": 6.842257897604068e-06,
      "loss": 4.8639,
      "step": 15214
    },
    {
      "epoch": 0.6142511102139685,
      "grad_norm": 100.8272476196289,
      "learning_rate": 6.841017247392988e-06,
      "loss": 4.82,
      "step": 15215
    },
    {
      "epoch": 0.6142914816310052,
      "grad_norm": 99.45758056640625,
      "learning_rate": 6.839776651193135e-06,
      "loss": 4.31,
      "step": 15216
    },
    {
      "epoch": 0.6143318530480419,
      "grad_norm": 98.27986907958984,
      "learning_rate": 6.838536109025727e-06,
      "loss": 5.642,
      "step": 15217
    },
    {
      "epoch": 0.6143722244650788,
      "grad_norm": 88.04923248291016,
      "learning_rate": 6.837295620911966e-06,
      "loss": 7.0283,
      "step": 15218
    },
    {
      "epoch": 0.6144125958821155,
      "grad_norm": 108.32057189941406,
      "learning_rate": 6.836055186873062e-06,
      "loss": 5.4828,
      "step": 15219
    },
    {
      "epoch": 0.6144529672991522,
      "grad_norm": 155.0875701904297,
      "learning_rate": 6.834814806930227e-06,
      "loss": 5.8021,
      "step": 15220
    },
    {
      "epoch": 0.6144933387161889,
      "grad_norm": 182.01858520507812,
      "learning_rate": 6.83357448110467e-06,
      "loss": 5.0266,
      "step": 15221
    },
    {
      "epoch": 0.6145337101332257,
      "grad_norm": 101.1726303100586,
      "learning_rate": 6.832334209417593e-06,
      "loss": 6.1985,
      "step": 15222
    },
    {
      "epoch": 0.6145740815502624,
      "grad_norm": 94.61016082763672,
      "learning_rate": 6.831093991890203e-06,
      "loss": 5.0551,
      "step": 15223
    },
    {
      "epoch": 0.6146144529672991,
      "grad_norm": 120.07305908203125,
      "learning_rate": 6.8298538285437065e-06,
      "loss": 5.6024,
      "step": 15224
    },
    {
      "epoch": 0.6146548243843359,
      "grad_norm": 109.21212768554688,
      "learning_rate": 6.8286137193993104e-06,
      "loss": 6.9406,
      "step": 15225
    },
    {
      "epoch": 0.6146951958013727,
      "grad_norm": 77.33344268798828,
      "learning_rate": 6.82737366447821e-06,
      "loss": 5.6994,
      "step": 15226
    },
    {
      "epoch": 0.6147355672184094,
      "grad_norm": 91.5445556640625,
      "learning_rate": 6.826133663801612e-06,
      "loss": 5.6322,
      "step": 15227
    },
    {
      "epoch": 0.6147759386354461,
      "grad_norm": 195.1837615966797,
      "learning_rate": 6.824893717390716e-06,
      "loss": 5.3172,
      "step": 15228
    },
    {
      "epoch": 0.6148163100524828,
      "grad_norm": 123.21321868896484,
      "learning_rate": 6.823653825266725e-06,
      "loss": 4.599,
      "step": 15229
    },
    {
      "epoch": 0.6148566814695196,
      "grad_norm": 162.07217407226562,
      "learning_rate": 6.822413987450835e-06,
      "loss": 5.039,
      "step": 15230
    },
    {
      "epoch": 0.6148970528865563,
      "grad_norm": 76.22655487060547,
      "learning_rate": 6.821174203964246e-06,
      "loss": 5.728,
      "step": 15231
    },
    {
      "epoch": 0.6149374243035931,
      "grad_norm": 86.54718017578125,
      "learning_rate": 6.8199344748281535e-06,
      "loss": 4.7329,
      "step": 15232
    },
    {
      "epoch": 0.6149777957206298,
      "grad_norm": 1641.407470703125,
      "learning_rate": 6.818694800063761e-06,
      "loss": 4.5734,
      "step": 15233
    },
    {
      "epoch": 0.6150181671376665,
      "grad_norm": 78.4503173828125,
      "learning_rate": 6.817455179692254e-06,
      "loss": 5.1023,
      "step": 15234
    },
    {
      "epoch": 0.6150585385547033,
      "grad_norm": 71.19731140136719,
      "learning_rate": 6.816215613734831e-06,
      "loss": 5.0446,
      "step": 15235
    },
    {
      "epoch": 0.61509890997174,
      "grad_norm": 116.13725280761719,
      "learning_rate": 6.814976102212687e-06,
      "loss": 5.4532,
      "step": 15236
    },
    {
      "epoch": 0.6151392813887767,
      "grad_norm": 112.68749237060547,
      "learning_rate": 6.813736645147017e-06,
      "loss": 5.6909,
      "step": 15237
    },
    {
      "epoch": 0.6151796528058134,
      "grad_norm": 145.34927368164062,
      "learning_rate": 6.812497242559008e-06,
      "loss": 5.7608,
      "step": 15238
    },
    {
      "epoch": 0.6152200242228503,
      "grad_norm": 80.2515640258789,
      "learning_rate": 6.811257894469851e-06,
      "loss": 6.2909,
      "step": 15239
    },
    {
      "epoch": 0.615260395639887,
      "grad_norm": 93.13349151611328,
      "learning_rate": 6.810018600900739e-06,
      "loss": 5.3004,
      "step": 15240
    },
    {
      "epoch": 0.6153007670569237,
      "grad_norm": 63.88608932495117,
      "learning_rate": 6.8087793618728614e-06,
      "loss": 5.9995,
      "step": 15241
    },
    {
      "epoch": 0.6153411384739604,
      "grad_norm": 90.4736099243164,
      "learning_rate": 6.807540177407403e-06,
      "loss": 5.3204,
      "step": 15242
    },
    {
      "epoch": 0.6153815098909972,
      "grad_norm": 94.68321228027344,
      "learning_rate": 6.806301047525552e-06,
      "loss": 5.1568,
      "step": 15243
    },
    {
      "epoch": 0.6154218813080339,
      "grad_norm": 91.60997009277344,
      "learning_rate": 6.805061972248495e-06,
      "loss": 5.7542,
      "step": 15244
    },
    {
      "epoch": 0.6154622527250706,
      "grad_norm": 89.1355972290039,
      "learning_rate": 6.803822951597423e-06,
      "loss": 6.3749,
      "step": 15245
    },
    {
      "epoch": 0.6155026241421074,
      "grad_norm": 108.88475036621094,
      "learning_rate": 6.80258398559351e-06,
      "loss": 6.4005,
      "step": 15246
    },
    {
      "epoch": 0.6155429955591442,
      "grad_norm": 104.40006256103516,
      "learning_rate": 6.801345074257944e-06,
      "loss": 5.0581,
      "step": 15247
    },
    {
      "epoch": 0.6155833669761809,
      "grad_norm": 97.11724853515625,
      "learning_rate": 6.800106217611908e-06,
      "loss": 5.0118,
      "step": 15248
    },
    {
      "epoch": 0.6156237383932176,
      "grad_norm": 93.82893371582031,
      "learning_rate": 6.798867415676584e-06,
      "loss": 6.936,
      "step": 15249
    },
    {
      "epoch": 0.6156641098102543,
      "grad_norm": 92.50856018066406,
      "learning_rate": 6.797628668473153e-06,
      "loss": 5.7028,
      "step": 15250
    },
    {
      "epoch": 0.615704481227291,
      "grad_norm": 90.6905288696289,
      "learning_rate": 6.796389976022791e-06,
      "loss": 4.5779,
      "step": 15251
    },
    {
      "epoch": 0.6157448526443278,
      "grad_norm": 104.4693603515625,
      "learning_rate": 6.795151338346681e-06,
      "loss": 4.512,
      "step": 15252
    },
    {
      "epoch": 0.6157852240613646,
      "grad_norm": 85.85360717773438,
      "learning_rate": 6.793912755466004e-06,
      "loss": 5.9195,
      "step": 15253
    },
    {
      "epoch": 0.6158255954784013,
      "grad_norm": 148.24169921875,
      "learning_rate": 6.792674227401926e-06,
      "loss": 4.4302,
      "step": 15254
    },
    {
      "epoch": 0.615865966895438,
      "grad_norm": 131.10157775878906,
      "learning_rate": 6.79143575417563e-06,
      "loss": 5.6048,
      "step": 15255
    },
    {
      "epoch": 0.6159063383124748,
      "grad_norm": 103.961181640625,
      "learning_rate": 6.7901973358082905e-06,
      "loss": 4.9264,
      "step": 15256
    },
    {
      "epoch": 0.6159467097295115,
      "grad_norm": 166.39297485351562,
      "learning_rate": 6.788958972321084e-06,
      "loss": 7.871,
      "step": 15257
    },
    {
      "epoch": 0.6159870811465482,
      "grad_norm": 112.48033905029297,
      "learning_rate": 6.7877206637351775e-06,
      "loss": 5.6368,
      "step": 15258
    },
    {
      "epoch": 0.6160274525635849,
      "grad_norm": 91.21461486816406,
      "learning_rate": 6.786482410071747e-06,
      "loss": 4.5464,
      "step": 15259
    },
    {
      "epoch": 0.6160678239806218,
      "grad_norm": 76.45587921142578,
      "learning_rate": 6.785244211351962e-06,
      "loss": 4.4083,
      "step": 15260
    },
    {
      "epoch": 0.6161081953976585,
      "grad_norm": 85.60636901855469,
      "learning_rate": 6.784006067596998e-06,
      "loss": 4.0089,
      "step": 15261
    },
    {
      "epoch": 0.6161485668146952,
      "grad_norm": 111.75762939453125,
      "learning_rate": 6.7827679788280156e-06,
      "loss": 5.3865,
      "step": 15262
    },
    {
      "epoch": 0.6161889382317319,
      "grad_norm": 89.06884002685547,
      "learning_rate": 6.78152994506619e-06,
      "loss": 4.2993,
      "step": 15263
    },
    {
      "epoch": 0.6162293096487687,
      "grad_norm": 99.86695861816406,
      "learning_rate": 6.780291966332686e-06,
      "loss": 5.4487,
      "step": 15264
    },
    {
      "epoch": 0.6162696810658054,
      "grad_norm": 85.6677017211914,
      "learning_rate": 6.779054042648674e-06,
      "loss": 4.9742,
      "step": 15265
    },
    {
      "epoch": 0.6163100524828421,
      "grad_norm": 98.50967407226562,
      "learning_rate": 6.777816174035315e-06,
      "loss": 7.0109,
      "step": 15266
    },
    {
      "epoch": 0.6163504238998789,
      "grad_norm": 132.92137145996094,
      "learning_rate": 6.776578360513773e-06,
      "loss": 4.6722,
      "step": 15267
    },
    {
      "epoch": 0.6163907953169157,
      "grad_norm": 84.17823791503906,
      "learning_rate": 6.775340602105214e-06,
      "loss": 5.4026,
      "step": 15268
    },
    {
      "epoch": 0.6164311667339524,
      "grad_norm": 81.8692855834961,
      "learning_rate": 6.774102898830802e-06,
      "loss": 4.8039,
      "step": 15269
    },
    {
      "epoch": 0.6164715381509891,
      "grad_norm": 101.010009765625,
      "learning_rate": 6.772865250711696e-06,
      "loss": 3.9705,
      "step": 15270
    },
    {
      "epoch": 0.6165119095680258,
      "grad_norm": 101.63807678222656,
      "learning_rate": 6.771627657769059e-06,
      "loss": 5.1512,
      "step": 15271
    },
    {
      "epoch": 0.6165522809850625,
      "grad_norm": 87.0859146118164,
      "learning_rate": 6.77039012002405e-06,
      "loss": 6.0271,
      "step": 15272
    },
    {
      "epoch": 0.6165926524020993,
      "grad_norm": 87.78706359863281,
      "learning_rate": 6.769152637497831e-06,
      "loss": 4.5931,
      "step": 15273
    },
    {
      "epoch": 0.6166330238191361,
      "grad_norm": 103.81919860839844,
      "learning_rate": 6.767915210211555e-06,
      "loss": 5.8381,
      "step": 15274
    },
    {
      "epoch": 0.6166733952361728,
      "grad_norm": 75.8786849975586,
      "learning_rate": 6.7666778381863805e-06,
      "loss": 4.799,
      "step": 15275
    },
    {
      "epoch": 0.6167137666532095,
      "grad_norm": 76.60758209228516,
      "learning_rate": 6.7654405214434645e-06,
      "loss": 5.341,
      "step": 15276
    },
    {
      "epoch": 0.6167541380702463,
      "grad_norm": 126.0286636352539,
      "learning_rate": 6.7642032600039674e-06,
      "loss": 5.5266,
      "step": 15277
    },
    {
      "epoch": 0.616794509487283,
      "grad_norm": 100.1476058959961,
      "learning_rate": 6.762966053889033e-06,
      "loss": 5.0641,
      "step": 15278
    },
    {
      "epoch": 0.6168348809043197,
      "grad_norm": 97.38457489013672,
      "learning_rate": 6.761728903119822e-06,
      "loss": 5.0082,
      "step": 15279
    },
    {
      "epoch": 0.6168752523213564,
      "grad_norm": 66.26468658447266,
      "learning_rate": 6.760491807717483e-06,
      "loss": 4.7604,
      "step": 15280
    },
    {
      "epoch": 0.6169156237383933,
      "grad_norm": 99.829833984375,
      "learning_rate": 6.759254767703172e-06,
      "loss": 4.7943,
      "step": 15281
    },
    {
      "epoch": 0.61695599515543,
      "grad_norm": 104.51441955566406,
      "learning_rate": 6.758017783098034e-06,
      "loss": 4.0621,
      "step": 15282
    },
    {
      "epoch": 0.6169963665724667,
      "grad_norm": 144.5611572265625,
      "learning_rate": 6.756780853923223e-06,
      "loss": 6.4172,
      "step": 15283
    },
    {
      "epoch": 0.6170367379895034,
      "grad_norm": 125.00214385986328,
      "learning_rate": 6.755543980199884e-06,
      "loss": 3.6691,
      "step": 15284
    },
    {
      "epoch": 0.6170771094065401,
      "grad_norm": 76.79956817626953,
      "learning_rate": 6.7543071619491715e-06,
      "loss": 5.6904,
      "step": 15285
    },
    {
      "epoch": 0.6171174808235769,
      "grad_norm": 81.88777923583984,
      "learning_rate": 6.7530703991922245e-06,
      "loss": 5.9498,
      "step": 15286
    },
    {
      "epoch": 0.6171578522406136,
      "grad_norm": 69.25776672363281,
      "learning_rate": 6.75183369195019e-06,
      "loss": 6.2216,
      "step": 15287
    },
    {
      "epoch": 0.6171982236576504,
      "grad_norm": 88.86699676513672,
      "learning_rate": 6.750597040244215e-06,
      "loss": 4.5452,
      "step": 15288
    },
    {
      "epoch": 0.6172385950746871,
      "grad_norm": 105.62861633300781,
      "learning_rate": 6.7493604440954456e-06,
      "loss": 5.4616,
      "step": 15289
    },
    {
      "epoch": 0.6172789664917239,
      "grad_norm": 77.83900451660156,
      "learning_rate": 6.748123903525018e-06,
      "loss": 2.0611,
      "step": 15290
    },
    {
      "epoch": 0.6173193379087606,
      "grad_norm": 71.51216888427734,
      "learning_rate": 6.746887418554078e-06,
      "loss": 6.512,
      "step": 15291
    },
    {
      "epoch": 0.6173597093257973,
      "grad_norm": 123.60004425048828,
      "learning_rate": 6.745650989203767e-06,
      "loss": 5.5152,
      "step": 15292
    },
    {
      "epoch": 0.617400080742834,
      "grad_norm": 294.43328857421875,
      "learning_rate": 6.744414615495229e-06,
      "loss": 6.1956,
      "step": 15293
    },
    {
      "epoch": 0.6174404521598708,
      "grad_norm": 113.36720275878906,
      "learning_rate": 6.743178297449595e-06,
      "loss": 5.603,
      "step": 15294
    },
    {
      "epoch": 0.6174808235769076,
      "grad_norm": 73.13615417480469,
      "learning_rate": 6.7419420350880075e-06,
      "loss": 4.1515,
      "step": 15295
    },
    {
      "epoch": 0.6175211949939443,
      "grad_norm": 85.96794891357422,
      "learning_rate": 6.7407058284316015e-06,
      "loss": 5.9934,
      "step": 15296
    },
    {
      "epoch": 0.617561566410981,
      "grad_norm": 125.28406524658203,
      "learning_rate": 6.739469677501518e-06,
      "loss": 6.8335,
      "step": 15297
    },
    {
      "epoch": 0.6176019378280178,
      "grad_norm": 100.11094665527344,
      "learning_rate": 6.738233582318886e-06,
      "loss": 5.5435,
      "step": 15298
    },
    {
      "epoch": 0.6176423092450545,
      "grad_norm": 67.12054443359375,
      "learning_rate": 6.736997542904844e-06,
      "loss": 6.8974,
      "step": 15299
    },
    {
      "epoch": 0.6176826806620912,
      "grad_norm": 90.13162994384766,
      "learning_rate": 6.7357615592805246e-06,
      "loss": 5.6743,
      "step": 15300
    },
    {
      "epoch": 0.617723052079128,
      "grad_norm": 73.59878540039062,
      "learning_rate": 6.7345256314670616e-06,
      "loss": 4.7904,
      "step": 15301
    },
    {
      "epoch": 0.6177634234961648,
      "grad_norm": 133.57232666015625,
      "learning_rate": 6.733289759485584e-06,
      "loss": 3.1825,
      "step": 15302
    },
    {
      "epoch": 0.6178037949132015,
      "grad_norm": 89.8676986694336,
      "learning_rate": 6.732053943357221e-06,
      "loss": 5.1051,
      "step": 15303
    },
    {
      "epoch": 0.6178441663302382,
      "grad_norm": 100.48129272460938,
      "learning_rate": 6.730818183103105e-06,
      "loss": 2.4585,
      "step": 15304
    },
    {
      "epoch": 0.6178845377472749,
      "grad_norm": 101.36132049560547,
      "learning_rate": 6.7295824787443676e-06,
      "loss": 4.3577,
      "step": 15305
    },
    {
      "epoch": 0.6179249091643116,
      "grad_norm": 87.0278549194336,
      "learning_rate": 6.72834683030213e-06,
      "loss": 3.3009,
      "step": 15306
    },
    {
      "epoch": 0.6179652805813484,
      "grad_norm": 72.3229751586914,
      "learning_rate": 6.727111237797521e-06,
      "loss": 5.1077,
      "step": 15307
    },
    {
      "epoch": 0.6180056519983852,
      "grad_norm": 77.90935516357422,
      "learning_rate": 6.725875701251667e-06,
      "loss": 5.7766,
      "step": 15308
    },
    {
      "epoch": 0.6180460234154219,
      "grad_norm": 63.44882583618164,
      "learning_rate": 6.724640220685695e-06,
      "loss": 6.3959,
      "step": 15309
    },
    {
      "epoch": 0.6180863948324586,
      "grad_norm": 166.58323669433594,
      "learning_rate": 6.723404796120725e-06,
      "loss": 3.2681,
      "step": 15310
    },
    {
      "epoch": 0.6181267662494954,
      "grad_norm": 110.42008209228516,
      "learning_rate": 6.722169427577881e-06,
      "loss": 4.4638,
      "step": 15311
    },
    {
      "epoch": 0.6181671376665321,
      "grad_norm": 63.292938232421875,
      "learning_rate": 6.720934115078286e-06,
      "loss": 3.3223,
      "step": 15312
    },
    {
      "epoch": 0.6182075090835688,
      "grad_norm": 131.99607849121094,
      "learning_rate": 6.719698858643064e-06,
      "loss": 5.1,
      "step": 15313
    },
    {
      "epoch": 0.6182478805006055,
      "grad_norm": 127.16340637207031,
      "learning_rate": 6.718463658293329e-06,
      "loss": 5.3483,
      "step": 15314
    },
    {
      "epoch": 0.6182882519176424,
      "grad_norm": 94.08757019042969,
      "learning_rate": 6.7172285140502e-06,
      "loss": 6.8884,
      "step": 15315
    },
    {
      "epoch": 0.6183286233346791,
      "grad_norm": 87.1301040649414,
      "learning_rate": 6.715993425934799e-06,
      "loss": 4.7184,
      "step": 15316
    },
    {
      "epoch": 0.6183689947517158,
      "grad_norm": 75.84696960449219,
      "learning_rate": 6.714758393968244e-06,
      "loss": 5.0154,
      "step": 15317
    },
    {
      "epoch": 0.6184093661687525,
      "grad_norm": 70.5610122680664,
      "learning_rate": 6.713523418171647e-06,
      "loss": 5.2728,
      "step": 15318
    },
    {
      "epoch": 0.6184497375857892,
      "grad_norm": 87.20799255371094,
      "learning_rate": 6.712288498566124e-06,
      "loss": 4.7397,
      "step": 15319
    },
    {
      "epoch": 0.618490109002826,
      "grad_norm": 90.5838851928711,
      "learning_rate": 6.711053635172791e-06,
      "loss": 5.1873,
      "step": 15320
    },
    {
      "epoch": 0.6185304804198627,
      "grad_norm": 108.61682891845703,
      "learning_rate": 6.709818828012764e-06,
      "loss": 5.6063,
      "step": 15321
    },
    {
      "epoch": 0.6185708518368995,
      "grad_norm": 81.88551330566406,
      "learning_rate": 6.70858407710715e-06,
      "loss": 5.3676,
      "step": 15322
    },
    {
      "epoch": 0.6186112232539362,
      "grad_norm": 92.86756896972656,
      "learning_rate": 6.70734938247706e-06,
      "loss": 5.4227,
      "step": 15323
    },
    {
      "epoch": 0.618651594670973,
      "grad_norm": 103.08333587646484,
      "learning_rate": 6.706114744143607e-06,
      "loss": 4.6936,
      "step": 15324
    },
    {
      "epoch": 0.6186919660880097,
      "grad_norm": 87.23209381103516,
      "learning_rate": 6.704880162127904e-06,
      "loss": 5.0715,
      "step": 15325
    },
    {
      "epoch": 0.6187323375050464,
      "grad_norm": 162.26368713378906,
      "learning_rate": 6.703645636451053e-06,
      "loss": 3.7552,
      "step": 15326
    },
    {
      "epoch": 0.6187727089220831,
      "grad_norm": 96.1218032836914,
      "learning_rate": 6.7024111671341615e-06,
      "loss": 4.8762,
      "step": 15327
    },
    {
      "epoch": 0.6188130803391199,
      "grad_norm": 82.39615631103516,
      "learning_rate": 6.701176754198342e-06,
      "loss": 3.9287,
      "step": 15328
    },
    {
      "epoch": 0.6188534517561567,
      "grad_norm": 84.5676498413086,
      "learning_rate": 6.699942397664696e-06,
      "loss": 3.2092,
      "step": 15329
    },
    {
      "epoch": 0.6188938231731934,
      "grad_norm": 111.64920806884766,
      "learning_rate": 6.698708097554327e-06,
      "loss": 5.418,
      "step": 15330
    },
    {
      "epoch": 0.6189341945902301,
      "grad_norm": 85.40535736083984,
      "learning_rate": 6.697473853888341e-06,
      "loss": 4.6734,
      "step": 15331
    },
    {
      "epoch": 0.6189745660072669,
      "grad_norm": 83.21040344238281,
      "learning_rate": 6.696239666687841e-06,
      "loss": 4.1187,
      "step": 15332
    },
    {
      "epoch": 0.6190149374243036,
      "grad_norm": 122.04055786132812,
      "learning_rate": 6.695005535973931e-06,
      "loss": 5.5096,
      "step": 15333
    },
    {
      "epoch": 0.6190553088413403,
      "grad_norm": 101.30186462402344,
      "learning_rate": 6.693771461767705e-06,
      "loss": 4.7802,
      "step": 15334
    },
    {
      "epoch": 0.619095680258377,
      "grad_norm": 80.20789337158203,
      "learning_rate": 6.6925374440902655e-06,
      "loss": 3.5826,
      "step": 15335
    },
    {
      "epoch": 0.6191360516754139,
      "grad_norm": 115.28067779541016,
      "learning_rate": 6.691303482962714e-06,
      "loss": 5.8233,
      "step": 15336
    },
    {
      "epoch": 0.6191764230924506,
      "grad_norm": 82.89185333251953,
      "learning_rate": 6.6900695784061465e-06,
      "loss": 4.7696,
      "step": 15337
    },
    {
      "epoch": 0.6192167945094873,
      "grad_norm": 109.31224060058594,
      "learning_rate": 6.68883573044166e-06,
      "loss": 5.3382,
      "step": 15338
    },
    {
      "epoch": 0.619257165926524,
      "grad_norm": 143.4357147216797,
      "learning_rate": 6.687601939090349e-06,
      "loss": 6.1895,
      "step": 15339
    },
    {
      "epoch": 0.6192975373435607,
      "grad_norm": 326.617431640625,
      "learning_rate": 6.686368204373311e-06,
      "loss": 3.7128,
      "step": 15340
    },
    {
      "epoch": 0.6193379087605975,
      "grad_norm": 115.99409484863281,
      "learning_rate": 6.685134526311641e-06,
      "loss": 5.6848,
      "step": 15341
    },
    {
      "epoch": 0.6193782801776342,
      "grad_norm": 132.0762176513672,
      "learning_rate": 6.683900904926427e-06,
      "loss": 4.61,
      "step": 15342
    },
    {
      "epoch": 0.619418651594671,
      "grad_norm": 147.29678344726562,
      "learning_rate": 6.682667340238764e-06,
      "loss": 4.8423,
      "step": 15343
    },
    {
      "epoch": 0.6194590230117077,
      "grad_norm": 79.57469940185547,
      "learning_rate": 6.681433832269743e-06,
      "loss": 4.4884,
      "step": 15344
    },
    {
      "epoch": 0.6194993944287445,
      "grad_norm": 122.80319213867188,
      "learning_rate": 6.680200381040457e-06,
      "loss": 4.5434,
      "step": 15345
    },
    {
      "epoch": 0.6195397658457812,
      "grad_norm": 83.68798828125,
      "learning_rate": 6.678966986571989e-06,
      "loss": 5.2236,
      "step": 15346
    },
    {
      "epoch": 0.6195801372628179,
      "grad_norm": 268.5706787109375,
      "learning_rate": 6.67773364888543e-06,
      "loss": 6.1576,
      "step": 15347
    },
    {
      "epoch": 0.6196205086798546,
      "grad_norm": 110.00431823730469,
      "learning_rate": 6.676500368001868e-06,
      "loss": 4.1096,
      "step": 15348
    },
    {
      "epoch": 0.6196608800968914,
      "grad_norm": 89.74700164794922,
      "learning_rate": 6.67526714394239e-06,
      "loss": 4.4594,
      "step": 15349
    },
    {
      "epoch": 0.6197012515139282,
      "grad_norm": 81.09233856201172,
      "learning_rate": 6.674033976728079e-06,
      "loss": 3.2817,
      "step": 15350
    },
    {
      "epoch": 0.6197416229309649,
      "grad_norm": 90.73677825927734,
      "learning_rate": 6.672800866380021e-06,
      "loss": 6.7265,
      "step": 15351
    },
    {
      "epoch": 0.6197819943480016,
      "grad_norm": 71.51158905029297,
      "learning_rate": 6.671567812919297e-06,
      "loss": 4.1583,
      "step": 15352
    },
    {
      "epoch": 0.6198223657650384,
      "grad_norm": 101.43916320800781,
      "learning_rate": 6.670334816366995e-06,
      "loss": 4.3513,
      "step": 15353
    },
    {
      "epoch": 0.6198627371820751,
      "grad_norm": 73.544921875,
      "learning_rate": 6.669101876744189e-06,
      "loss": 2.2225,
      "step": 15354
    },
    {
      "epoch": 0.6199031085991118,
      "grad_norm": 109.39041137695312,
      "learning_rate": 6.667868994071962e-06,
      "loss": 5.2158,
      "step": 15355
    },
    {
      "epoch": 0.6199434800161485,
      "grad_norm": 100.38644409179688,
      "learning_rate": 6.666636168371396e-06,
      "loss": 4.8703,
      "step": 15356
    },
    {
      "epoch": 0.6199838514331854,
      "grad_norm": 92.29209899902344,
      "learning_rate": 6.665403399663568e-06,
      "loss": 5.3231,
      "step": 15357
    },
    {
      "epoch": 0.6200242228502221,
      "grad_norm": 91.43850708007812,
      "learning_rate": 6.6641706879695535e-06,
      "loss": 3.6642,
      "step": 15358
    },
    {
      "epoch": 0.6200645942672588,
      "grad_norm": 88.84017944335938,
      "learning_rate": 6.662938033310429e-06,
      "loss": 3.5159,
      "step": 15359
    },
    {
      "epoch": 0.6201049656842955,
      "grad_norm": 374.8038635253906,
      "learning_rate": 6.661705435707273e-06,
      "loss": 4.0848,
      "step": 15360
    },
    {
      "epoch": 0.6201453371013322,
      "grad_norm": 101.23371124267578,
      "learning_rate": 6.660472895181162e-06,
      "loss": 5.1473,
      "step": 15361
    },
    {
      "epoch": 0.620185708518369,
      "grad_norm": 101.53878784179688,
      "learning_rate": 6.659240411753163e-06,
      "loss": 4.4115,
      "step": 15362
    },
    {
      "epoch": 0.6202260799354057,
      "grad_norm": 133.25216674804688,
      "learning_rate": 6.658007985444351e-06,
      "loss": 5.758,
      "step": 15363
    },
    {
      "epoch": 0.6202664513524425,
      "grad_norm": 82.9168701171875,
      "learning_rate": 6.656775616275799e-06,
      "loss": 4.5778,
      "step": 15364
    },
    {
      "epoch": 0.6203068227694792,
      "grad_norm": 92.88570404052734,
      "learning_rate": 6.65554330426858e-06,
      "loss": 5.6721,
      "step": 15365
    },
    {
      "epoch": 0.620347194186516,
      "grad_norm": 97.1352310180664,
      "learning_rate": 6.654311049443758e-06,
      "loss": 5.8755,
      "step": 15366
    },
    {
      "epoch": 0.6203875656035527,
      "grad_norm": 192.83543395996094,
      "learning_rate": 6.653078851822403e-06,
      "loss": 4.9639,
      "step": 15367
    },
    {
      "epoch": 0.6204279370205894,
      "grad_norm": 83.95835876464844,
      "learning_rate": 6.651846711425585e-06,
      "loss": 5.8836,
      "step": 15368
    },
    {
      "epoch": 0.6204683084376261,
      "grad_norm": 90.88090515136719,
      "learning_rate": 6.65061462827437e-06,
      "loss": 4.3281,
      "step": 15369
    },
    {
      "epoch": 0.6205086798546628,
      "grad_norm": 66.31250762939453,
      "learning_rate": 6.649382602389824e-06,
      "loss": 4.2378,
      "step": 15370
    },
    {
      "epoch": 0.6205490512716997,
      "grad_norm": 75.14350128173828,
      "learning_rate": 6.648150633793009e-06,
      "loss": 4.4766,
      "step": 15371
    },
    {
      "epoch": 0.6205894226887364,
      "grad_norm": 110.10342407226562,
      "learning_rate": 6.6469187225049915e-06,
      "loss": 4.1065,
      "step": 15372
    },
    {
      "epoch": 0.6206297941057731,
      "grad_norm": 725.0841064453125,
      "learning_rate": 6.645686868546839e-06,
      "loss": 4.0319,
      "step": 15373
    },
    {
      "epoch": 0.6206701655228098,
      "grad_norm": 90.06910705566406,
      "learning_rate": 6.644455071939604e-06,
      "loss": 5.0495,
      "step": 15374
    },
    {
      "epoch": 0.6207105369398466,
      "grad_norm": 139.0943145751953,
      "learning_rate": 6.643223332704349e-06,
      "loss": 6.3365,
      "step": 15375
    },
    {
      "epoch": 0.6207509083568833,
      "grad_norm": 64.68318176269531,
      "learning_rate": 6.64199165086214e-06,
      "loss": 6.6425,
      "step": 15376
    },
    {
      "epoch": 0.62079127977392,
      "grad_norm": 79.34615325927734,
      "learning_rate": 6.640760026434032e-06,
      "loss": 3.96,
      "step": 15377
    },
    {
      "epoch": 0.6208316511909568,
      "grad_norm": 71.16232299804688,
      "learning_rate": 6.639528459441083e-06,
      "loss": 5.167,
      "step": 15378
    },
    {
      "epoch": 0.6208720226079936,
      "grad_norm": 104.87725067138672,
      "learning_rate": 6.638296949904348e-06,
      "loss": 6.1138,
      "step": 15379
    },
    {
      "epoch": 0.6209123940250303,
      "grad_norm": 110.55170440673828,
      "learning_rate": 6.637065497844886e-06,
      "loss": 5.5446,
      "step": 15380
    },
    {
      "epoch": 0.620952765442067,
      "grad_norm": 74.28702545166016,
      "learning_rate": 6.635834103283753e-06,
      "loss": 4.2564,
      "step": 15381
    },
    {
      "epoch": 0.6209931368591037,
      "grad_norm": 94.9197769165039,
      "learning_rate": 6.634602766242e-06,
      "loss": 5.0043,
      "step": 15382
    },
    {
      "epoch": 0.6210335082761405,
      "grad_norm": 136.0020751953125,
      "learning_rate": 6.6333714867406795e-06,
      "loss": 5.6035,
      "step": 15383
    },
    {
      "epoch": 0.6210738796931772,
      "grad_norm": 91.36505126953125,
      "learning_rate": 6.632140264800846e-06,
      "loss": 3.6084,
      "step": 15384
    },
    {
      "epoch": 0.621114251110214,
      "grad_norm": 100.4331283569336,
      "learning_rate": 6.630909100443552e-06,
      "loss": 4.5781,
      "step": 15385
    },
    {
      "epoch": 0.6211546225272507,
      "grad_norm": 208.2967529296875,
      "learning_rate": 6.629677993689842e-06,
      "loss": 4.6034,
      "step": 15386
    },
    {
      "epoch": 0.6211949939442875,
      "grad_norm": 72.51233673095703,
      "learning_rate": 6.628446944560767e-06,
      "loss": 6.2506,
      "step": 15387
    },
    {
      "epoch": 0.6212353653613242,
      "grad_norm": 113.48328399658203,
      "learning_rate": 6.627215953077377e-06,
      "loss": 4.6267,
      "step": 15388
    },
    {
      "epoch": 0.6212757367783609,
      "grad_norm": 89.68829345703125,
      "learning_rate": 6.625985019260719e-06,
      "loss": 4.3973,
      "step": 15389
    },
    {
      "epoch": 0.6213161081953976,
      "grad_norm": 79.07293701171875,
      "learning_rate": 6.624754143131837e-06,
      "loss": 5.6418,
      "step": 15390
    },
    {
      "epoch": 0.6213564796124343,
      "grad_norm": 137.1891326904297,
      "learning_rate": 6.623523324711778e-06,
      "loss": 5.5636,
      "step": 15391
    },
    {
      "epoch": 0.6213968510294712,
      "grad_norm": 91.12741088867188,
      "learning_rate": 6.622292564021585e-06,
      "loss": 2.9322,
      "step": 15392
    },
    {
      "epoch": 0.6214372224465079,
      "grad_norm": 110.17015075683594,
      "learning_rate": 6.621061861082307e-06,
      "loss": 5.1023,
      "step": 15393
    },
    {
      "epoch": 0.6214775938635446,
      "grad_norm": 79.95769500732422,
      "learning_rate": 6.619831215914974e-06,
      "loss": 5.5833,
      "step": 15394
    },
    {
      "epoch": 0.6215179652805813,
      "grad_norm": 78.17754364013672,
      "learning_rate": 6.618600628540637e-06,
      "loss": 5.1695,
      "step": 15395
    },
    {
      "epoch": 0.6215583366976181,
      "grad_norm": 93.52200317382812,
      "learning_rate": 6.6173700989803315e-06,
      "loss": 2.7122,
      "step": 15396
    },
    {
      "epoch": 0.6215987081146548,
      "grad_norm": 73.87753295898438,
      "learning_rate": 6.616139627255099e-06,
      "loss": 4.8664,
      "step": 15397
    },
    {
      "epoch": 0.6216390795316915,
      "grad_norm": 79.07318115234375,
      "learning_rate": 6.614909213385976e-06,
      "loss": 4.9035,
      "step": 15398
    },
    {
      "epoch": 0.6216794509487283,
      "grad_norm": 65.2998275756836,
      "learning_rate": 6.613678857394001e-06,
      "loss": 3.8405,
      "step": 15399
    },
    {
      "epoch": 0.6217198223657651,
      "grad_norm": 111.21546936035156,
      "learning_rate": 6.6124485593002106e-06,
      "loss": 5.7045,
      "step": 15400
    },
    {
      "epoch": 0.6217601937828018,
      "grad_norm": 73.010009765625,
      "learning_rate": 6.611218319125643e-06,
      "loss": 5.8228,
      "step": 15401
    },
    {
      "epoch": 0.6218005651998385,
      "grad_norm": 158.7651824951172,
      "learning_rate": 6.609988136891325e-06,
      "loss": 4.9819,
      "step": 15402
    },
    {
      "epoch": 0.6218409366168752,
      "grad_norm": 67.48844909667969,
      "learning_rate": 6.6087580126182924e-06,
      "loss": 4.7785,
      "step": 15403
    },
    {
      "epoch": 0.621881308033912,
      "grad_norm": 188.37066650390625,
      "learning_rate": 6.607527946327579e-06,
      "loss": 6.4473,
      "step": 15404
    },
    {
      "epoch": 0.6219216794509488,
      "grad_norm": 73.50249481201172,
      "learning_rate": 6.6062979380402185e-06,
      "loss": 5.2936,
      "step": 15405
    },
    {
      "epoch": 0.6219620508679855,
      "grad_norm": 79.85748291015625,
      "learning_rate": 6.605067987777236e-06,
      "loss": 6.2812,
      "step": 15406
    },
    {
      "epoch": 0.6220024222850222,
      "grad_norm": 79.14411926269531,
      "learning_rate": 6.603838095559663e-06,
      "loss": 5.9097,
      "step": 15407
    },
    {
      "epoch": 0.622042793702059,
      "grad_norm": 72.78382873535156,
      "learning_rate": 6.602608261408528e-06,
      "loss": 4.042,
      "step": 15408
    },
    {
      "epoch": 0.6220831651190957,
      "grad_norm": 76.64612579345703,
      "learning_rate": 6.6013784853448625e-06,
      "loss": 3.7306,
      "step": 15409
    },
    {
      "epoch": 0.6221235365361324,
      "grad_norm": 77.15248107910156,
      "learning_rate": 6.600148767389684e-06,
      "loss": 4.8748,
      "step": 15410
    },
    {
      "epoch": 0.6221639079531691,
      "grad_norm": 77.04931640625,
      "learning_rate": 6.5989191075640236e-06,
      "loss": 3.7407,
      "step": 15411
    },
    {
      "epoch": 0.622204279370206,
      "grad_norm": 84.62828826904297,
      "learning_rate": 6.5976895058889025e-06,
      "loss": 5.9488,
      "step": 15412
    },
    {
      "epoch": 0.6222446507872427,
      "grad_norm": 91.82209014892578,
      "learning_rate": 6.5964599623853505e-06,
      "loss": 5.1229,
      "step": 15413
    },
    {
      "epoch": 0.6222850222042794,
      "grad_norm": 93.13436126708984,
      "learning_rate": 6.5952304770743825e-06,
      "loss": 4.7131,
      "step": 15414
    },
    {
      "epoch": 0.6223253936213161,
      "grad_norm": 110.1066665649414,
      "learning_rate": 6.594001049977021e-06,
      "loss": 3.8559,
      "step": 15415
    },
    {
      "epoch": 0.6223657650383528,
      "grad_norm": 76.74835205078125,
      "learning_rate": 6.592771681114289e-06,
      "loss": 6.3875,
      "step": 15416
    },
    {
      "epoch": 0.6224061364553896,
      "grad_norm": 76.35624694824219,
      "learning_rate": 6.591542370507207e-06,
      "loss": 4.7371,
      "step": 15417
    },
    {
      "epoch": 0.6224465078724263,
      "grad_norm": 178.069580078125,
      "learning_rate": 6.590313118176789e-06,
      "loss": 4.755,
      "step": 15418
    },
    {
      "epoch": 0.6224868792894631,
      "grad_norm": 106.32182312011719,
      "learning_rate": 6.589083924144054e-06,
      "loss": 5.403,
      "step": 15419
    },
    {
      "epoch": 0.6225272507064998,
      "grad_norm": 217.470947265625,
      "learning_rate": 6.587854788430019e-06,
      "loss": 5.7582,
      "step": 15420
    },
    {
      "epoch": 0.6225676221235366,
      "grad_norm": 126.02434539794922,
      "learning_rate": 6.586625711055703e-06,
      "loss": 4.0397,
      "step": 15421
    },
    {
      "epoch": 0.6226079935405733,
      "grad_norm": 63.73207092285156,
      "learning_rate": 6.585396692042113e-06,
      "loss": 7.3623,
      "step": 15422
    },
    {
      "epoch": 0.62264836495761,
      "grad_norm": 238.4598388671875,
      "learning_rate": 6.584167731410266e-06,
      "loss": 5.75,
      "step": 15423
    },
    {
      "epoch": 0.6226887363746467,
      "grad_norm": 90.73844909667969,
      "learning_rate": 6.582938829181174e-06,
      "loss": 5.7775,
      "step": 15424
    },
    {
      "epoch": 0.6227291077916834,
      "grad_norm": 101.71595764160156,
      "learning_rate": 6.581709985375851e-06,
      "loss": 4.8152,
      "step": 15425
    },
    {
      "epoch": 0.6227694792087203,
      "grad_norm": 100.93080139160156,
      "learning_rate": 6.580481200015302e-06,
      "loss": 4.4071,
      "step": 15426
    },
    {
      "epoch": 0.622809850625757,
      "grad_norm": 70.09378814697266,
      "learning_rate": 6.57925247312054e-06,
      "loss": 2.2049,
      "step": 15427
    },
    {
      "epoch": 0.6228502220427937,
      "grad_norm": 294.8962097167969,
      "learning_rate": 6.578023804712572e-06,
      "loss": 3.513,
      "step": 15428
    },
    {
      "epoch": 0.6228905934598304,
      "grad_norm": 105.5246353149414,
      "learning_rate": 6.576795194812411e-06,
      "loss": 5.0233,
      "step": 15429
    },
    {
      "epoch": 0.6229309648768672,
      "grad_norm": 86.30955505371094,
      "learning_rate": 6.575566643441054e-06,
      "loss": 4.7382,
      "step": 15430
    },
    {
      "epoch": 0.6229713362939039,
      "grad_norm": 119.27633666992188,
      "learning_rate": 6.57433815061951e-06,
      "loss": 5.9629,
      "step": 15431
    },
    {
      "epoch": 0.6230117077109406,
      "grad_norm": 96.9399185180664,
      "learning_rate": 6.573109716368785e-06,
      "loss": 5.2315,
      "step": 15432
    },
    {
      "epoch": 0.6230520791279774,
      "grad_norm": 99.01590728759766,
      "learning_rate": 6.571881340709886e-06,
      "loss": 4.9987,
      "step": 15433
    },
    {
      "epoch": 0.6230924505450142,
      "grad_norm": 87.10150909423828,
      "learning_rate": 6.570653023663807e-06,
      "loss": 2.7278,
      "step": 15434
    },
    {
      "epoch": 0.6231328219620509,
      "grad_norm": 113.03492736816406,
      "learning_rate": 6.569424765251551e-06,
      "loss": 7.2414,
      "step": 15435
    },
    {
      "epoch": 0.6231731933790876,
      "grad_norm": 108.93977355957031,
      "learning_rate": 6.568196565494123e-06,
      "loss": 4.0164,
      "step": 15436
    },
    {
      "epoch": 0.6232135647961243,
      "grad_norm": 79.66472625732422,
      "learning_rate": 6.566968424412521e-06,
      "loss": 4.1755,
      "step": 15437
    },
    {
      "epoch": 0.623253936213161,
      "grad_norm": 152.5521240234375,
      "learning_rate": 6.56574034202774e-06,
      "loss": 3.8261,
      "step": 15438
    },
    {
      "epoch": 0.6232943076301978,
      "grad_norm": 113.92403411865234,
      "learning_rate": 6.564512318360778e-06,
      "loss": 3.5305,
      "step": 15439
    },
    {
      "epoch": 0.6233346790472346,
      "grad_norm": 101.389892578125,
      "learning_rate": 6.5632843534326354e-06,
      "loss": 3.8242,
      "step": 15440
    },
    {
      "epoch": 0.6233750504642713,
      "grad_norm": 111.70850372314453,
      "learning_rate": 6.5620564472643085e-06,
      "loss": 3.6687,
      "step": 15441
    },
    {
      "epoch": 0.623415421881308,
      "grad_norm": 99.62991333007812,
      "learning_rate": 6.560828599876784e-06,
      "loss": 5.05,
      "step": 15442
    },
    {
      "epoch": 0.6234557932983448,
      "grad_norm": 94.90670776367188,
      "learning_rate": 6.5596008112910595e-06,
      "loss": 5.3223,
      "step": 15443
    },
    {
      "epoch": 0.6234961647153815,
      "grad_norm": 82.27989959716797,
      "learning_rate": 6.5583730815281266e-06,
      "loss": 6.1858,
      "step": 15444
    },
    {
      "epoch": 0.6235365361324182,
      "grad_norm": 121.99264526367188,
      "learning_rate": 6.557145410608979e-06,
      "loss": 6.3728,
      "step": 15445
    },
    {
      "epoch": 0.6235769075494549,
      "grad_norm": 85.16875457763672,
      "learning_rate": 6.555917798554606e-06,
      "loss": 4.0057,
      "step": 15446
    },
    {
      "epoch": 0.6236172789664918,
      "grad_norm": 127.306396484375,
      "learning_rate": 6.554690245385993e-06,
      "loss": 3.5888,
      "step": 15447
    },
    {
      "epoch": 0.6236576503835285,
      "grad_norm": 207.33828735351562,
      "learning_rate": 6.553462751124131e-06,
      "loss": 4.1428,
      "step": 15448
    },
    {
      "epoch": 0.6236980218005652,
      "grad_norm": 90.90850067138672,
      "learning_rate": 6.552235315790014e-06,
      "loss": 5.4403,
      "step": 15449
    },
    {
      "epoch": 0.6237383932176019,
      "grad_norm": 153.63870239257812,
      "learning_rate": 6.551007939404616e-06,
      "loss": 4.5069,
      "step": 15450
    },
    {
      "epoch": 0.6237787646346387,
      "grad_norm": 118.19204711914062,
      "learning_rate": 6.549780621988931e-06,
      "loss": 5.8185,
      "step": 15451
    },
    {
      "epoch": 0.6238191360516754,
      "grad_norm": 99.38166809082031,
      "learning_rate": 6.548553363563939e-06,
      "loss": 4.0963,
      "step": 15452
    },
    {
      "epoch": 0.6238595074687121,
      "grad_norm": 92.21057891845703,
      "learning_rate": 6.547326164150629e-06,
      "loss": 6.0667,
      "step": 15453
    },
    {
      "epoch": 0.6238998788857489,
      "grad_norm": 119.16060638427734,
      "learning_rate": 6.546099023769974e-06,
      "loss": 4.6365,
      "step": 15454
    },
    {
      "epoch": 0.6239402503027857,
      "grad_norm": 91.99610137939453,
      "learning_rate": 6.544871942442961e-06,
      "loss": 4.9608,
      "step": 15455
    },
    {
      "epoch": 0.6239806217198224,
      "grad_norm": 87.16822814941406,
      "learning_rate": 6.5436449201905695e-06,
      "loss": 5.1271,
      "step": 15456
    },
    {
      "epoch": 0.6240209931368591,
      "grad_norm": 101.91020202636719,
      "learning_rate": 6.542417957033781e-06,
      "loss": 5.3149,
      "step": 15457
    },
    {
      "epoch": 0.6240613645538958,
      "grad_norm": 93.97846221923828,
      "learning_rate": 6.541191052993568e-06,
      "loss": 5.2207,
      "step": 15458
    },
    {
      "epoch": 0.6241017359709325,
      "grad_norm": 122.02873229980469,
      "learning_rate": 6.539964208090912e-06,
      "loss": 4.476,
      "step": 15459
    },
    {
      "epoch": 0.6241421073879693,
      "grad_norm": 143.0734100341797,
      "learning_rate": 6.538737422346789e-06,
      "loss": 3.8628,
      "step": 15460
    },
    {
      "epoch": 0.6241824788050061,
      "grad_norm": 125.57079315185547,
      "learning_rate": 6.5375106957821765e-06,
      "loss": 4.3814,
      "step": 15461
    },
    {
      "epoch": 0.6242228502220428,
      "grad_norm": 117.49222564697266,
      "learning_rate": 6.536284028418043e-06,
      "loss": 4.5678,
      "step": 15462
    },
    {
      "epoch": 0.6242632216390795,
      "grad_norm": 94.4518814086914,
      "learning_rate": 6.535057420275363e-06,
      "loss": 6.0528,
      "step": 15463
    },
    {
      "epoch": 0.6243035930561163,
      "grad_norm": 83.64407348632812,
      "learning_rate": 6.53383087137511e-06,
      "loss": 5.2943,
      "step": 15464
    },
    {
      "epoch": 0.624343964473153,
      "grad_norm": 111.0434341430664,
      "learning_rate": 6.532604381738257e-06,
      "loss": 4.297,
      "step": 15465
    },
    {
      "epoch": 0.6243843358901897,
      "grad_norm": 83.84178161621094,
      "learning_rate": 6.531377951385771e-06,
      "loss": 4.3963,
      "step": 15466
    },
    {
      "epoch": 0.6244247073072264,
      "grad_norm": 89.19962310791016,
      "learning_rate": 6.530151580338621e-06,
      "loss": 5.5254,
      "step": 15467
    },
    {
      "epoch": 0.6244650787242633,
      "grad_norm": 137.39816284179688,
      "learning_rate": 6.528925268617776e-06,
      "loss": 3.7914,
      "step": 15468
    },
    {
      "epoch": 0.6245054501413,
      "grad_norm": 141.97872924804688,
      "learning_rate": 6.527699016244209e-06,
      "loss": 3.4832,
      "step": 15469
    },
    {
      "epoch": 0.6245458215583367,
      "grad_norm": 87.96973419189453,
      "learning_rate": 6.5264728232388755e-06,
      "loss": 4.56,
      "step": 15470
    },
    {
      "epoch": 0.6245861929753734,
      "grad_norm": 151.2353515625,
      "learning_rate": 6.5252466896227455e-06,
      "loss": 4.2639,
      "step": 15471
    },
    {
      "epoch": 0.6246265643924102,
      "grad_norm": 99.0444107055664,
      "learning_rate": 6.524020615416783e-06,
      "loss": 6.1753,
      "step": 15472
    },
    {
      "epoch": 0.6246669358094469,
      "grad_norm": 120.6518325805664,
      "learning_rate": 6.522794600641955e-06,
      "loss": 6.0584,
      "step": 15473
    },
    {
      "epoch": 0.6247073072264836,
      "grad_norm": 107.77307891845703,
      "learning_rate": 6.521568645319215e-06,
      "loss": 5.342,
      "step": 15474
    },
    {
      "epoch": 0.6247476786435204,
      "grad_norm": 146.7619171142578,
      "learning_rate": 6.520342749469529e-06,
      "loss": 5.6781,
      "step": 15475
    },
    {
      "epoch": 0.6247880500605572,
      "grad_norm": 97.6932373046875,
      "learning_rate": 6.519116913113856e-06,
      "loss": 4.0948,
      "step": 15476
    },
    {
      "epoch": 0.6248284214775939,
      "grad_norm": 107.31643676757812,
      "learning_rate": 6.517891136273158e-06,
      "loss": 5.5863,
      "step": 15477
    },
    {
      "epoch": 0.6248687928946306,
      "grad_norm": 123.91244506835938,
      "learning_rate": 6.5166654189683886e-06,
      "loss": 4.5856,
      "step": 15478
    },
    {
      "epoch": 0.6249091643116673,
      "grad_norm": 99.3900146484375,
      "learning_rate": 6.515439761220506e-06,
      "loss": 5.6169,
      "step": 15479
    },
    {
      "epoch": 0.624949535728704,
      "grad_norm": 100.43476104736328,
      "learning_rate": 6.5142141630504654e-06,
      "loss": 6.5068,
      "step": 15480
    },
    {
      "epoch": 0.6249899071457408,
      "grad_norm": 85.10704040527344,
      "learning_rate": 6.512988624479227e-06,
      "loss": 5.2633,
      "step": 15481
    },
    {
      "epoch": 0.6250302785627776,
      "grad_norm": 102.7247314453125,
      "learning_rate": 6.511763145527737e-06,
      "loss": 3.999,
      "step": 15482
    },
    {
      "epoch": 0.6250706499798143,
      "grad_norm": 85.8324966430664,
      "learning_rate": 6.510537726216951e-06,
      "loss": 5.402,
      "step": 15483
    },
    {
      "epoch": 0.625111021396851,
      "grad_norm": 85.99805450439453,
      "learning_rate": 6.509312366567822e-06,
      "loss": 5.5358,
      "step": 15484
    },
    {
      "epoch": 0.6251513928138878,
      "grad_norm": 100.70098876953125,
      "learning_rate": 6.508087066601302e-06,
      "loss": 4.8429,
      "step": 15485
    },
    {
      "epoch": 0.6251917642309245,
      "grad_norm": 75.96065521240234,
      "learning_rate": 6.506861826338338e-06,
      "loss": 6.2293,
      "step": 15486
    },
    {
      "epoch": 0.6252321356479612,
      "grad_norm": 92.03007507324219,
      "learning_rate": 6.505636645799878e-06,
      "loss": 5.8697,
      "step": 15487
    },
    {
      "epoch": 0.6252725070649979,
      "grad_norm": 105.51961517333984,
      "learning_rate": 6.504411525006872e-06,
      "loss": 4.4779,
      "step": 15488
    },
    {
      "epoch": 0.6253128784820348,
      "grad_norm": 79.6859359741211,
      "learning_rate": 6.50318646398027e-06,
      "loss": 3.7006,
      "step": 15489
    },
    {
      "epoch": 0.6253532498990715,
      "grad_norm": 154.54306030273438,
      "learning_rate": 6.50196146274101e-06,
      "loss": 4.148,
      "step": 15490
    },
    {
      "epoch": 0.6253936213161082,
      "grad_norm": 83.45893096923828,
      "learning_rate": 6.50073652131004e-06,
      "loss": 5.0537,
      "step": 15491
    },
    {
      "epoch": 0.6254339927331449,
      "grad_norm": 130.5682830810547,
      "learning_rate": 6.499511639708305e-06,
      "loss": 5.5089,
      "step": 15492
    },
    {
      "epoch": 0.6254743641501816,
      "grad_norm": 155.1271209716797,
      "learning_rate": 6.49828681795675e-06,
      "loss": 4.8247,
      "step": 15493
    },
    {
      "epoch": 0.6255147355672184,
      "grad_norm": 81.00763702392578,
      "learning_rate": 6.49706205607631e-06,
      "loss": 5.4678,
      "step": 15494
    },
    {
      "epoch": 0.6255551069842551,
      "grad_norm": 133.16551208496094,
      "learning_rate": 6.495837354087928e-06,
      "loss": 2.0974,
      "step": 15495
    },
    {
      "epoch": 0.6255954784012919,
      "grad_norm": 98.32731628417969,
      "learning_rate": 6.494612712012545e-06,
      "loss": 2.9337,
      "step": 15496
    },
    {
      "epoch": 0.6256358498183286,
      "grad_norm": 119.12008666992188,
      "learning_rate": 6.493388129871101e-06,
      "loss": 5.3316,
      "step": 15497
    },
    {
      "epoch": 0.6256762212353654,
      "grad_norm": 102.93743896484375,
      "learning_rate": 6.49216360768453e-06,
      "loss": 3.5753,
      "step": 15498
    },
    {
      "epoch": 0.6257165926524021,
      "grad_norm": 73.46363830566406,
      "learning_rate": 6.490939145473769e-06,
      "loss": 4.2595,
      "step": 15499
    },
    {
      "epoch": 0.6257569640694388,
      "grad_norm": 68.68716430664062,
      "learning_rate": 6.489714743259755e-06,
      "loss": 3.1727,
      "step": 15500
    },
    {
      "epoch": 0.6257973354864755,
      "grad_norm": 161.6457061767578,
      "learning_rate": 6.488490401063425e-06,
      "loss": 5.0762,
      "step": 15501
    },
    {
      "epoch": 0.6258377069035123,
      "grad_norm": 118.27096557617188,
      "learning_rate": 6.487266118905706e-06,
      "loss": 4.9312,
      "step": 15502
    },
    {
      "epoch": 0.6258780783205491,
      "grad_norm": 116.36491394042969,
      "learning_rate": 6.486041896807533e-06,
      "loss": 5.9272,
      "step": 15503
    },
    {
      "epoch": 0.6259184497375858,
      "grad_norm": 76.83333587646484,
      "learning_rate": 6.484817734789839e-06,
      "loss": 5.4758,
      "step": 15504
    },
    {
      "epoch": 0.6259588211546225,
      "grad_norm": 98.76023864746094,
      "learning_rate": 6.483593632873554e-06,
      "loss": 6.3283,
      "step": 15505
    },
    {
      "epoch": 0.6259991925716593,
      "grad_norm": 76.58493041992188,
      "learning_rate": 6.4823695910796045e-06,
      "loss": 6.2975,
      "step": 15506
    },
    {
      "epoch": 0.626039563988696,
      "grad_norm": 124.69244384765625,
      "learning_rate": 6.481145609428921e-06,
      "loss": 5.0539,
      "step": 15507
    },
    {
      "epoch": 0.6260799354057327,
      "grad_norm": 72.2063980102539,
      "learning_rate": 6.47992168794243e-06,
      "loss": 2.8877,
      "step": 15508
    },
    {
      "epoch": 0.6261203068227695,
      "grad_norm": 65.66043090820312,
      "learning_rate": 6.478697826641062e-06,
      "loss": 4.1176,
      "step": 15509
    },
    {
      "epoch": 0.6261606782398063,
      "grad_norm": 60.67626953125,
      "learning_rate": 6.477474025545736e-06,
      "loss": 5.0726,
      "step": 15510
    },
    {
      "epoch": 0.626201049656843,
      "grad_norm": 69.14634704589844,
      "learning_rate": 6.476250284677377e-06,
      "loss": 4.3207,
      "step": 15511
    },
    {
      "epoch": 0.6262414210738797,
      "grad_norm": 88.17572021484375,
      "learning_rate": 6.475026604056911e-06,
      "loss": 4.2311,
      "step": 15512
    },
    {
      "epoch": 0.6262817924909164,
      "grad_norm": 91.33909606933594,
      "learning_rate": 6.473802983705259e-06,
      "loss": 4.0431,
      "step": 15513
    },
    {
      "epoch": 0.6263221639079531,
      "grad_norm": 90.22227478027344,
      "learning_rate": 6.47257942364334e-06,
      "loss": 4.7872,
      "step": 15514
    },
    {
      "epoch": 0.6263625353249899,
      "grad_norm": 86.18204498291016,
      "learning_rate": 6.471355923892076e-06,
      "loss": 6.299,
      "step": 15515
    },
    {
      "epoch": 0.6264029067420267,
      "grad_norm": 95.21533203125,
      "learning_rate": 6.470132484472387e-06,
      "loss": 6.0346,
      "step": 15516
    },
    {
      "epoch": 0.6264432781590634,
      "grad_norm": 201.64276123046875,
      "learning_rate": 6.468909105405192e-06,
      "loss": 4.4098,
      "step": 15517
    },
    {
      "epoch": 0.6264836495761001,
      "grad_norm": 289.40789794921875,
      "learning_rate": 6.467685786711402e-06,
      "loss": 4.912,
      "step": 15518
    },
    {
      "epoch": 0.6265240209931369,
      "grad_norm": 106.5746841430664,
      "learning_rate": 6.466462528411936e-06,
      "loss": 5.4907,
      "step": 15519
    },
    {
      "epoch": 0.6265643924101736,
      "grad_norm": 114.62418365478516,
      "learning_rate": 6.4652393305277105e-06,
      "loss": 5.6506,
      "step": 15520
    },
    {
      "epoch": 0.6266047638272103,
      "grad_norm": 77.13677215576172,
      "learning_rate": 6.464016193079642e-06,
      "loss": 6.2132,
      "step": 15521
    },
    {
      "epoch": 0.626645135244247,
      "grad_norm": 95.04682159423828,
      "learning_rate": 6.462793116088635e-06,
      "loss": 7.0196,
      "step": 15522
    },
    {
      "epoch": 0.6266855066612839,
      "grad_norm": 110.08967590332031,
      "learning_rate": 6.461570099575606e-06,
      "loss": 4.5088,
      "step": 15523
    },
    {
      "epoch": 0.6267258780783206,
      "grad_norm": 94.35706329345703,
      "learning_rate": 6.460347143561465e-06,
      "loss": 5.2451,
      "step": 15524
    },
    {
      "epoch": 0.6267662494953573,
      "grad_norm": 86.72767639160156,
      "learning_rate": 6.459124248067126e-06,
      "loss": 4.1461,
      "step": 15525
    },
    {
      "epoch": 0.626806620912394,
      "grad_norm": 76.32442474365234,
      "learning_rate": 6.45790141311349e-06,
      "loss": 5.1541,
      "step": 15526
    },
    {
      "epoch": 0.6268469923294308,
      "grad_norm": 86.93584442138672,
      "learning_rate": 6.456678638721467e-06,
      "loss": 5.9151,
      "step": 15527
    },
    {
      "epoch": 0.6268873637464675,
      "grad_norm": 77.43367004394531,
      "learning_rate": 6.455455924911968e-06,
      "loss": 4.5708,
      "step": 15528
    },
    {
      "epoch": 0.6269277351635042,
      "grad_norm": 85.18489837646484,
      "learning_rate": 6.454233271705899e-06,
      "loss": 3.8592,
      "step": 15529
    },
    {
      "epoch": 0.626968106580541,
      "grad_norm": 107.12654876708984,
      "learning_rate": 6.453010679124157e-06,
      "loss": 4.8381,
      "step": 15530
    },
    {
      "epoch": 0.6270084779975778,
      "grad_norm": 135.00770568847656,
      "learning_rate": 6.4517881471876475e-06,
      "loss": 5.0794,
      "step": 15531
    },
    {
      "epoch": 0.6270488494146145,
      "grad_norm": 209.2467498779297,
      "learning_rate": 6.450565675917277e-06,
      "loss": 4.4121,
      "step": 15532
    },
    {
      "epoch": 0.6270892208316512,
      "grad_norm": 73.1801528930664,
      "learning_rate": 6.449343265333947e-06,
      "loss": 5.9234,
      "step": 15533
    },
    {
      "epoch": 0.6271295922486879,
      "grad_norm": 113.16432189941406,
      "learning_rate": 6.448120915458553e-06,
      "loss": 5.4861,
      "step": 15534
    },
    {
      "epoch": 0.6271699636657246,
      "grad_norm": 87.28841400146484,
      "learning_rate": 6.446898626311997e-06,
      "loss": 5.9613,
      "step": 15535
    },
    {
      "epoch": 0.6272103350827614,
      "grad_norm": 62.659122467041016,
      "learning_rate": 6.445676397915178e-06,
      "loss": 4.7627,
      "step": 15536
    },
    {
      "epoch": 0.6272507064997982,
      "grad_norm": 94.60504913330078,
      "learning_rate": 6.444454230288997e-06,
      "loss": 5.0559,
      "step": 15537
    },
    {
      "epoch": 0.6272910779168349,
      "grad_norm": 115.3642578125,
      "learning_rate": 6.443232123454342e-06,
      "loss": 6.1109,
      "step": 15538
    },
    {
      "epoch": 0.6273314493338716,
      "grad_norm": 96.36842346191406,
      "learning_rate": 6.44201007743211e-06,
      "loss": 5.3736,
      "step": 15539
    },
    {
      "epoch": 0.6273718207509084,
      "grad_norm": 63.766334533691406,
      "learning_rate": 6.440788092243199e-06,
      "loss": 3.2982,
      "step": 15540
    },
    {
      "epoch": 0.6274121921679451,
      "grad_norm": 132.61399841308594,
      "learning_rate": 6.439566167908505e-06,
      "loss": 5.5224,
      "step": 15541
    },
    {
      "epoch": 0.6274525635849818,
      "grad_norm": 97.35105895996094,
      "learning_rate": 6.438344304448909e-06,
      "loss": 5.1978,
      "step": 15542
    },
    {
      "epoch": 0.6274929350020185,
      "grad_norm": 121.4608154296875,
      "learning_rate": 6.437122501885309e-06,
      "loss": 4.3694,
      "step": 15543
    },
    {
      "epoch": 0.6275333064190554,
      "grad_norm": 74.65548706054688,
      "learning_rate": 6.435900760238595e-06,
      "loss": 5.5667,
      "step": 15544
    },
    {
      "epoch": 0.6275736778360921,
      "grad_norm": 86.57757568359375,
      "learning_rate": 6.434679079529658e-06,
      "loss": 4.9287,
      "step": 15545
    },
    {
      "epoch": 0.6276140492531288,
      "grad_norm": 96.17584228515625,
      "learning_rate": 6.43345745977938e-06,
      "loss": 5.7817,
      "step": 15546
    },
    {
      "epoch": 0.6276544206701655,
      "grad_norm": 101.06555938720703,
      "learning_rate": 6.43223590100865e-06,
      "loss": 4.4418,
      "step": 15547
    },
    {
      "epoch": 0.6276947920872022,
      "grad_norm": 86.21603393554688,
      "learning_rate": 6.431014403238356e-06,
      "loss": 4.2646,
      "step": 15548
    },
    {
      "epoch": 0.627735163504239,
      "grad_norm": 103.92151641845703,
      "learning_rate": 6.429792966489385e-06,
      "loss": 4.6351,
      "step": 15549
    },
    {
      "epoch": 0.6277755349212757,
      "grad_norm": 132.75851440429688,
      "learning_rate": 6.428571590782614e-06,
      "loss": 5.1045,
      "step": 15550
    },
    {
      "epoch": 0.6278159063383125,
      "grad_norm": 67.63011932373047,
      "learning_rate": 6.427350276138929e-06,
      "loss": 4.1449,
      "step": 15551
    },
    {
      "epoch": 0.6278562777553492,
      "grad_norm": 70.03147888183594,
      "learning_rate": 6.42612902257921e-06,
      "loss": 4.4611,
      "step": 15552
    },
    {
      "epoch": 0.627896649172386,
      "grad_norm": 94.82052612304688,
      "learning_rate": 6.424907830124343e-06,
      "loss": 4.8268,
      "step": 15553
    },
    {
      "epoch": 0.6279370205894227,
      "grad_norm": 158.82749938964844,
      "learning_rate": 6.4236866987951994e-06,
      "loss": 5.4296,
      "step": 15554
    },
    {
      "epoch": 0.6279773920064594,
      "grad_norm": 115.77527618408203,
      "learning_rate": 6.422465628612664e-06,
      "loss": 5.2242,
      "step": 15555
    },
    {
      "epoch": 0.6280177634234961,
      "grad_norm": 73.6752700805664,
      "learning_rate": 6.421244619597612e-06,
      "loss": 5.1821,
      "step": 15556
    },
    {
      "epoch": 0.6280581348405329,
      "grad_norm": 92.61103820800781,
      "learning_rate": 6.420023671770923e-06,
      "loss": 4.9704,
      "step": 15557
    },
    {
      "epoch": 0.6280985062575697,
      "grad_norm": 91.51678466796875,
      "learning_rate": 6.418802785153466e-06,
      "loss": 6.2653,
      "step": 15558
    },
    {
      "epoch": 0.6281388776746064,
      "grad_norm": 74.73603057861328,
      "learning_rate": 6.417581959766119e-06,
      "loss": 5.6929,
      "step": 15559
    },
    {
      "epoch": 0.6281792490916431,
      "grad_norm": 83.44775390625,
      "learning_rate": 6.416361195629755e-06,
      "loss": 5.8387,
      "step": 15560
    },
    {
      "epoch": 0.6282196205086799,
      "grad_norm": 84.10950469970703,
      "learning_rate": 6.415140492765249e-06,
      "loss": 5.4733,
      "step": 15561
    },
    {
      "epoch": 0.6282599919257166,
      "grad_norm": 95.81703186035156,
      "learning_rate": 6.413919851193466e-06,
      "loss": 5.3291,
      "step": 15562
    },
    {
      "epoch": 0.6283003633427533,
      "grad_norm": 80.90137481689453,
      "learning_rate": 6.412699270935281e-06,
      "loss": 4.6759,
      "step": 15563
    },
    {
      "epoch": 0.62834073475979,
      "grad_norm": 120.6703872680664,
      "learning_rate": 6.41147875201156e-06,
      "loss": 4.4928,
      "step": 15564
    },
    {
      "epoch": 0.6283811061768269,
      "grad_norm": 73.88253784179688,
      "learning_rate": 6.4102582944431735e-06,
      "loss": 5.2017,
      "step": 15565
    },
    {
      "epoch": 0.6284214775938636,
      "grad_norm": 112.97540283203125,
      "learning_rate": 6.409037898250987e-06,
      "loss": 4.7855,
      "step": 15566
    },
    {
      "epoch": 0.6284618490109003,
      "grad_norm": 73.65728759765625,
      "learning_rate": 6.407817563455867e-06,
      "loss": 4.104,
      "step": 15567
    },
    {
      "epoch": 0.628502220427937,
      "grad_norm": 102.64999389648438,
      "learning_rate": 6.406597290078677e-06,
      "loss": 4.2553,
      "step": 15568
    },
    {
      "epoch": 0.6285425918449737,
      "grad_norm": 94.6607666015625,
      "learning_rate": 6.405377078140287e-06,
      "loss": 5.9482,
      "step": 15569
    },
    {
      "epoch": 0.6285829632620105,
      "grad_norm": 247.24681091308594,
      "learning_rate": 6.404156927661551e-06,
      "loss": 3.9695,
      "step": 15570
    },
    {
      "epoch": 0.6286233346790472,
      "grad_norm": 123.0394058227539,
      "learning_rate": 6.402936838663334e-06,
      "loss": 6.2747,
      "step": 15571
    },
    {
      "epoch": 0.628663706096084,
      "grad_norm": 115.1880111694336,
      "learning_rate": 6.401716811166499e-06,
      "loss": 6.0372,
      "step": 15572
    },
    {
      "epoch": 0.6287040775131207,
      "grad_norm": 195.91859436035156,
      "learning_rate": 6.400496845191903e-06,
      "loss": 3.3731,
      "step": 15573
    },
    {
      "epoch": 0.6287444489301575,
      "grad_norm": 78.64543151855469,
      "learning_rate": 6.399276940760406e-06,
      "loss": 5.1204,
      "step": 15574
    },
    {
      "epoch": 0.6287848203471942,
      "grad_norm": 101.18346405029297,
      "learning_rate": 6.3980570978928645e-06,
      "loss": 5.5836,
      "step": 15575
    },
    {
      "epoch": 0.6288251917642309,
      "grad_norm": 143.0545196533203,
      "learning_rate": 6.396837316610136e-06,
      "loss": 5.8516,
      "step": 15576
    },
    {
      "epoch": 0.6288655631812676,
      "grad_norm": 132.90432739257812,
      "learning_rate": 6.395617596933079e-06,
      "loss": 5.2435,
      "step": 15577
    },
    {
      "epoch": 0.6289059345983044,
      "grad_norm": 77.05824279785156,
      "learning_rate": 6.39439793888254e-06,
      "loss": 4.5276,
      "step": 15578
    },
    {
      "epoch": 0.6289463060153412,
      "grad_norm": 88.52106475830078,
      "learning_rate": 6.393178342479376e-06,
      "loss": 4.284,
      "step": 15579
    },
    {
      "epoch": 0.6289866774323779,
      "grad_norm": 99.63385772705078,
      "learning_rate": 6.3919588077444405e-06,
      "loss": 5.1391,
      "step": 15580
    },
    {
      "epoch": 0.6290270488494146,
      "grad_norm": 89.88388061523438,
      "learning_rate": 6.390739334698588e-06,
      "loss": 3.7915,
      "step": 15581
    },
    {
      "epoch": 0.6290674202664513,
      "grad_norm": 107.92877960205078,
      "learning_rate": 6.38951992336266e-06,
      "loss": 6.517,
      "step": 15582
    },
    {
      "epoch": 0.6291077916834881,
      "grad_norm": 106.99325561523438,
      "learning_rate": 6.388300573757511e-06,
      "loss": 3.9944,
      "step": 15583
    },
    {
      "epoch": 0.6291481631005248,
      "grad_norm": 104.45789337158203,
      "learning_rate": 6.3870812859039874e-06,
      "loss": 4.162,
      "step": 15584
    },
    {
      "epoch": 0.6291885345175615,
      "grad_norm": 83.10025024414062,
      "learning_rate": 6.385862059822941e-06,
      "loss": 5.2577,
      "step": 15585
    },
    {
      "epoch": 0.6292289059345983,
      "grad_norm": 73.12694549560547,
      "learning_rate": 6.384642895535209e-06,
      "loss": 3.7748,
      "step": 15586
    },
    {
      "epoch": 0.6292692773516351,
      "grad_norm": 133.6226043701172,
      "learning_rate": 6.383423793061643e-06,
      "loss": 5.6444,
      "step": 15587
    },
    {
      "epoch": 0.6293096487686718,
      "grad_norm": 130.43675231933594,
      "learning_rate": 6.382204752423084e-06,
      "loss": 3.5757,
      "step": 15588
    },
    {
      "epoch": 0.6293500201857085,
      "grad_norm": 162.61245727539062,
      "learning_rate": 6.38098577364038e-06,
      "loss": 5.3954,
      "step": 15589
    },
    {
      "epoch": 0.6293903916027452,
      "grad_norm": 97.4022216796875,
      "learning_rate": 6.379766856734364e-06,
      "loss": 5.0232,
      "step": 15590
    },
    {
      "epoch": 0.629430763019782,
      "grad_norm": 120.01815032958984,
      "learning_rate": 6.378548001725881e-06,
      "loss": 3.694,
      "step": 15591
    },
    {
      "epoch": 0.6294711344368187,
      "grad_norm": 92.05241394042969,
      "learning_rate": 6.377329208635769e-06,
      "loss": 4.9696,
      "step": 15592
    },
    {
      "epoch": 0.6295115058538555,
      "grad_norm": 84.35311126708984,
      "learning_rate": 6.376110477484873e-06,
      "loss": 5.2937,
      "step": 15593
    },
    {
      "epoch": 0.6295518772708922,
      "grad_norm": 81.97806549072266,
      "learning_rate": 6.374891808294022e-06,
      "loss": 4.6265,
      "step": 15594
    },
    {
      "epoch": 0.629592248687929,
      "grad_norm": 170.5729217529297,
      "learning_rate": 6.373673201084054e-06,
      "loss": 4.7503,
      "step": 15595
    },
    {
      "epoch": 0.6296326201049657,
      "grad_norm": 130.99359130859375,
      "learning_rate": 6.372454655875808e-06,
      "loss": 4.337,
      "step": 15596
    },
    {
      "epoch": 0.6296729915220024,
      "grad_norm": 109.15947723388672,
      "learning_rate": 6.3712361726901205e-06,
      "loss": 4.9861,
      "step": 15597
    },
    {
      "epoch": 0.6297133629390391,
      "grad_norm": 98.45796203613281,
      "learning_rate": 6.3700177515478166e-06,
      "loss": 5.6815,
      "step": 15598
    },
    {
      "epoch": 0.6297537343560758,
      "grad_norm": 79.76239776611328,
      "learning_rate": 6.368799392469733e-06,
      "loss": 4.7133,
      "step": 15599
    },
    {
      "epoch": 0.6297941057731127,
      "grad_norm": 113.47289276123047,
      "learning_rate": 6.3675810954767e-06,
      "loss": 6.4321,
      "step": 15600
    },
    {
      "epoch": 0.6298344771901494,
      "grad_norm": 96.39791870117188,
      "learning_rate": 6.36636286058955e-06,
      "loss": 4.9649,
      "step": 15601
    },
    {
      "epoch": 0.6298748486071861,
      "grad_norm": 119.69029235839844,
      "learning_rate": 6.365144687829108e-06,
      "loss": 3.6837,
      "step": 15602
    },
    {
      "epoch": 0.6299152200242228,
      "grad_norm": 155.12237548828125,
      "learning_rate": 6.363926577216204e-06,
      "loss": 6.079,
      "step": 15603
    },
    {
      "epoch": 0.6299555914412596,
      "grad_norm": 116.34342956542969,
      "learning_rate": 6.362708528771665e-06,
      "loss": 5.2936,
      "step": 15604
    },
    {
      "epoch": 0.6299959628582963,
      "grad_norm": 97.58760833740234,
      "learning_rate": 6.361490542516317e-06,
      "loss": 6.0661,
      "step": 15605
    },
    {
      "epoch": 0.630036334275333,
      "grad_norm": 75.24996948242188,
      "learning_rate": 6.360272618470984e-06,
      "loss": 5.5299,
      "step": 15606
    },
    {
      "epoch": 0.6300767056923698,
      "grad_norm": 120.21451568603516,
      "learning_rate": 6.359054756656488e-06,
      "loss": 4.8688,
      "step": 15607
    },
    {
      "epoch": 0.6301170771094066,
      "grad_norm": 83.70540618896484,
      "learning_rate": 6.3578369570936535e-06,
      "loss": 4.2034,
      "step": 15608
    },
    {
      "epoch": 0.6301574485264433,
      "grad_norm": 95.31781768798828,
      "learning_rate": 6.356619219803307e-06,
      "loss": 4.1894,
      "step": 15609
    },
    {
      "epoch": 0.63019781994348,
      "grad_norm": 102.98094177246094,
      "learning_rate": 6.35540154480626e-06,
      "loss": 5.7,
      "step": 15610
    },
    {
      "epoch": 0.6302381913605167,
      "grad_norm": 119.9601058959961,
      "learning_rate": 6.354183932123335e-06,
      "loss": 5.1065,
      "step": 15611
    },
    {
      "epoch": 0.6302785627775535,
      "grad_norm": 81.57649230957031,
      "learning_rate": 6.352966381775352e-06,
      "loss": 5.2474,
      "step": 15612
    },
    {
      "epoch": 0.6303189341945902,
      "grad_norm": 300.3497009277344,
      "learning_rate": 6.3517488937831285e-06,
      "loss": 5.0136,
      "step": 15613
    },
    {
      "epoch": 0.630359305611627,
      "grad_norm": 96.92818450927734,
      "learning_rate": 6.350531468167478e-06,
      "loss": 4.1829,
      "step": 15614
    },
    {
      "epoch": 0.6303996770286637,
      "grad_norm": 185.14683532714844,
      "learning_rate": 6.349314104949218e-06,
      "loss": 5.3095,
      "step": 15615
    },
    {
      "epoch": 0.6304400484457005,
      "grad_norm": 224.52423095703125,
      "learning_rate": 6.348096804149161e-06,
      "loss": 3.8144,
      "step": 15616
    },
    {
      "epoch": 0.6304804198627372,
      "grad_norm": 129.4156951904297,
      "learning_rate": 6.346879565788124e-06,
      "loss": 4.9939,
      "step": 15617
    },
    {
      "epoch": 0.6305207912797739,
      "grad_norm": 142.01475524902344,
      "learning_rate": 6.345662389886914e-06,
      "loss": 4.7186,
      "step": 15618
    },
    {
      "epoch": 0.6305611626968106,
      "grad_norm": 103.06940460205078,
      "learning_rate": 6.3444452764663426e-06,
      "loss": 5.0754,
      "step": 15619
    },
    {
      "epoch": 0.6306015341138475,
      "grad_norm": 113.67555236816406,
      "learning_rate": 6.34322822554722e-06,
      "loss": 5.2826,
      "step": 15620
    },
    {
      "epoch": 0.6306419055308842,
      "grad_norm": 115.61483001708984,
      "learning_rate": 6.342011237150358e-06,
      "loss": 6.5005,
      "step": 15621
    },
    {
      "epoch": 0.6306822769479209,
      "grad_norm": 95.79280853271484,
      "learning_rate": 6.340794311296561e-06,
      "loss": 5.3267,
      "step": 15622
    },
    {
      "epoch": 0.6307226483649576,
      "grad_norm": 102.79826354980469,
      "learning_rate": 6.339577448006635e-06,
      "loss": 4.2796,
      "step": 15623
    },
    {
      "epoch": 0.6307630197819943,
      "grad_norm": 89.74763488769531,
      "learning_rate": 6.338360647301387e-06,
      "loss": 5.1757,
      "step": 15624
    },
    {
      "epoch": 0.6308033911990311,
      "grad_norm": 80.91704559326172,
      "learning_rate": 6.337143909201627e-06,
      "loss": 4.8702,
      "step": 15625
    },
    {
      "epoch": 0.6308437626160678,
      "grad_norm": 168.82664489746094,
      "learning_rate": 6.335927233728145e-06,
      "loss": 5.4068,
      "step": 15626
    },
    {
      "epoch": 0.6308841340331046,
      "grad_norm": 161.5408935546875,
      "learning_rate": 6.334710620901754e-06,
      "loss": 5.1457,
      "step": 15627
    },
    {
      "epoch": 0.6309245054501413,
      "grad_norm": 97.41340637207031,
      "learning_rate": 6.333494070743252e-06,
      "loss": 4.7513,
      "step": 15628
    },
    {
      "epoch": 0.6309648768671781,
      "grad_norm": 79.04658508300781,
      "learning_rate": 6.332277583273444e-06,
      "loss": 4.1049,
      "step": 15629
    },
    {
      "epoch": 0.6310052482842148,
      "grad_norm": 101.9944076538086,
      "learning_rate": 6.33106115851312e-06,
      "loss": 5.8773,
      "step": 15630
    },
    {
      "epoch": 0.6310456197012515,
      "grad_norm": 72.03726959228516,
      "learning_rate": 6.329844796483083e-06,
      "loss": 5.1817,
      "step": 15631
    },
    {
      "epoch": 0.6310859911182882,
      "grad_norm": 81.69588470458984,
      "learning_rate": 6.3286284972041275e-06,
      "loss": 3.7935,
      "step": 15632
    },
    {
      "epoch": 0.631126362535325,
      "grad_norm": 129.65098571777344,
      "learning_rate": 6.327412260697055e-06,
      "loss": 5.0471,
      "step": 15633
    },
    {
      "epoch": 0.6311667339523618,
      "grad_norm": 128.09811401367188,
      "learning_rate": 6.326196086982656e-06,
      "loss": 6.5653,
      "step": 15634
    },
    {
      "epoch": 0.6312071053693985,
      "grad_norm": 76.77549743652344,
      "learning_rate": 6.324979976081723e-06,
      "loss": 5.1548,
      "step": 15635
    },
    {
      "epoch": 0.6312474767864352,
      "grad_norm": 101.65813446044922,
      "learning_rate": 6.323763928015052e-06,
      "loss": 5.2925,
      "step": 15636
    },
    {
      "epoch": 0.631287848203472,
      "grad_norm": 81.34661865234375,
      "learning_rate": 6.322547942803437e-06,
      "loss": 5.219,
      "step": 15637
    },
    {
      "epoch": 0.6313282196205087,
      "grad_norm": 106.14039611816406,
      "learning_rate": 6.321332020467658e-06,
      "loss": 6.3059,
      "step": 15638
    },
    {
      "epoch": 0.6313685910375454,
      "grad_norm": 71.96897888183594,
      "learning_rate": 6.320116161028515e-06,
      "loss": 5.9748,
      "step": 15639
    },
    {
      "epoch": 0.6314089624545821,
      "grad_norm": 148.08822631835938,
      "learning_rate": 6.318900364506791e-06,
      "loss": 5.723,
      "step": 15640
    },
    {
      "epoch": 0.631449333871619,
      "grad_norm": 92.8070068359375,
      "learning_rate": 6.317684630923273e-06,
      "loss": 5.3535,
      "step": 15641
    },
    {
      "epoch": 0.6314897052886557,
      "grad_norm": 96.7383804321289,
      "learning_rate": 6.316468960298754e-06,
      "loss": 5.1163,
      "step": 15642
    },
    {
      "epoch": 0.6315300767056924,
      "grad_norm": 102.60424041748047,
      "learning_rate": 6.315253352654009e-06,
      "loss": 5.6062,
      "step": 15643
    },
    {
      "epoch": 0.6315704481227291,
      "grad_norm": 225.595703125,
      "learning_rate": 6.31403780800983e-06,
      "loss": 5.0962,
      "step": 15644
    },
    {
      "epoch": 0.6316108195397658,
      "grad_norm": 148.52764892578125,
      "learning_rate": 6.312822326386996e-06,
      "loss": 5.9619,
      "step": 15645
    },
    {
      "epoch": 0.6316511909568026,
      "grad_norm": 87.86540222167969,
      "learning_rate": 6.311606907806294e-06,
      "loss": 5.6126,
      "step": 15646
    },
    {
      "epoch": 0.6316915623738393,
      "grad_norm": 120.15270233154297,
      "learning_rate": 6.310391552288497e-06,
      "loss": 4.1431,
      "step": 15647
    },
    {
      "epoch": 0.6317319337908761,
      "grad_norm": 108.81008911132812,
      "learning_rate": 6.309176259854388e-06,
      "loss": 5.7476,
      "step": 15648
    },
    {
      "epoch": 0.6317723052079128,
      "grad_norm": 102.9189682006836,
      "learning_rate": 6.307961030524748e-06,
      "loss": 5.5636,
      "step": 15649
    },
    {
      "epoch": 0.6318126766249496,
      "grad_norm": 88.94054412841797,
      "learning_rate": 6.306745864320355e-06,
      "loss": 5.6701,
      "step": 15650
    },
    {
      "epoch": 0.6318530480419863,
      "grad_norm": 93.81879425048828,
      "learning_rate": 6.30553076126198e-06,
      "loss": 4.2929,
      "step": 15651
    },
    {
      "epoch": 0.631893419459023,
      "grad_norm": 140.46170043945312,
      "learning_rate": 6.304315721370402e-06,
      "loss": 3.5146,
      "step": 15652
    },
    {
      "epoch": 0.6319337908760597,
      "grad_norm": 135.46124267578125,
      "learning_rate": 6.303100744666395e-06,
      "loss": 5.4868,
      "step": 15653
    },
    {
      "epoch": 0.6319741622930964,
      "grad_norm": 73.71045684814453,
      "learning_rate": 6.3018858311707344e-06,
      "loss": 4.4437,
      "step": 15654
    },
    {
      "epoch": 0.6320145337101333,
      "grad_norm": 108.53911590576172,
      "learning_rate": 6.300670980904188e-06,
      "loss": 5.2828,
      "step": 15655
    },
    {
      "epoch": 0.63205490512717,
      "grad_norm": 77.44209289550781,
      "learning_rate": 6.29945619388753e-06,
      "loss": 4.3986,
      "step": 15656
    },
    {
      "epoch": 0.6320952765442067,
      "grad_norm": 77.64119720458984,
      "learning_rate": 6.298241470141528e-06,
      "loss": 5.4891,
      "step": 15657
    },
    {
      "epoch": 0.6321356479612434,
      "grad_norm": 73.667236328125,
      "learning_rate": 6.297026809686959e-06,
      "loss": 5.0562,
      "step": 15658
    },
    {
      "epoch": 0.6321760193782802,
      "grad_norm": 129.44235229492188,
      "learning_rate": 6.295812212544578e-06,
      "loss": 4.865,
      "step": 15659
    },
    {
      "epoch": 0.6322163907953169,
      "grad_norm": 100.73023223876953,
      "learning_rate": 6.29459767873516e-06,
      "loss": 4.289,
      "step": 15660
    },
    {
      "epoch": 0.6322567622123536,
      "grad_norm": 74.84738159179688,
      "learning_rate": 6.293383208279469e-06,
      "loss": 4.768,
      "step": 15661
    },
    {
      "epoch": 0.6322971336293904,
      "grad_norm": 115.99102020263672,
      "learning_rate": 6.29216880119827e-06,
      "loss": 5.8377,
      "step": 15662
    },
    {
      "epoch": 0.6323375050464272,
      "grad_norm": 81.05705261230469,
      "learning_rate": 6.290954457512326e-06,
      "loss": 4.4736,
      "step": 15663
    },
    {
      "epoch": 0.6323778764634639,
      "grad_norm": 99.73564910888672,
      "learning_rate": 6.289740177242399e-06,
      "loss": 4.6924,
      "step": 15664
    },
    {
      "epoch": 0.6324182478805006,
      "grad_norm": 135.39013671875,
      "learning_rate": 6.288525960409251e-06,
      "loss": 6.0408,
      "step": 15665
    },
    {
      "epoch": 0.6324586192975373,
      "grad_norm": 94.55406188964844,
      "learning_rate": 6.287311807033646e-06,
      "loss": 6.5249,
      "step": 15666
    },
    {
      "epoch": 0.632498990714574,
      "grad_norm": 131.54913330078125,
      "learning_rate": 6.286097717136335e-06,
      "loss": 6.6071,
      "step": 15667
    },
    {
      "epoch": 0.6325393621316108,
      "grad_norm": 98.44320678710938,
      "learning_rate": 6.2848836907380805e-06,
      "loss": 4.733,
      "step": 15668
    },
    {
      "epoch": 0.6325797335486476,
      "grad_norm": 91.60980224609375,
      "learning_rate": 6.283669727859639e-06,
      "loss": 4.9401,
      "step": 15669
    },
    {
      "epoch": 0.6326201049656843,
      "grad_norm": 84.75231170654297,
      "learning_rate": 6.282455828521772e-06,
      "loss": 4.3043,
      "step": 15670
    },
    {
      "epoch": 0.632660476382721,
      "grad_norm": 72.95435333251953,
      "learning_rate": 6.281241992745224e-06,
      "loss": 4.1599,
      "step": 15671
    },
    {
      "epoch": 0.6327008477997578,
      "grad_norm": 72.98220825195312,
      "learning_rate": 6.2800282205507555e-06,
      "loss": 5.1752,
      "step": 15672
    },
    {
      "epoch": 0.6327412192167945,
      "grad_norm": 95.43392181396484,
      "learning_rate": 6.278814511959117e-06,
      "loss": 4.7855,
      "step": 15673
    },
    {
      "epoch": 0.6327815906338312,
      "grad_norm": 97.4222183227539,
      "learning_rate": 6.277600866991063e-06,
      "loss": 5.0124,
      "step": 15674
    },
    {
      "epoch": 0.6328219620508679,
      "grad_norm": 72.10237121582031,
      "learning_rate": 6.276387285667338e-06,
      "loss": 5.3011,
      "step": 15675
    },
    {
      "epoch": 0.6328623334679048,
      "grad_norm": 79.71903228759766,
      "learning_rate": 6.2751737680086975e-06,
      "loss": 5.9786,
      "step": 15676
    },
    {
      "epoch": 0.6329027048849415,
      "grad_norm": 80.5411148071289,
      "learning_rate": 6.2739603140358865e-06,
      "loss": 4.6513,
      "step": 15677
    },
    {
      "epoch": 0.6329430763019782,
      "grad_norm": 83.71794128417969,
      "learning_rate": 6.272746923769657e-06,
      "loss": 5.0589,
      "step": 15678
    },
    {
      "epoch": 0.6329834477190149,
      "grad_norm": 88.99451446533203,
      "learning_rate": 6.271533597230748e-06,
      "loss": 6.1074,
      "step": 15679
    },
    {
      "epoch": 0.6330238191360517,
      "grad_norm": 137.11106872558594,
      "learning_rate": 6.270320334439908e-06,
      "loss": 4.3379,
      "step": 15680
    },
    {
      "epoch": 0.6330641905530884,
      "grad_norm": 73.13856506347656,
      "learning_rate": 6.26910713541788e-06,
      "loss": 5.2469,
      "step": 15681
    },
    {
      "epoch": 0.6331045619701251,
      "grad_norm": 89.94049835205078,
      "learning_rate": 6.26789400018541e-06,
      "loss": 4.5815,
      "step": 15682
    },
    {
      "epoch": 0.6331449333871619,
      "grad_norm": 116.04022216796875,
      "learning_rate": 6.266680928763236e-06,
      "loss": 2.1318,
      "step": 15683
    },
    {
      "epoch": 0.6331853048041987,
      "grad_norm": 83.96791076660156,
      "learning_rate": 6.265467921172099e-06,
      "loss": 4.8636,
      "step": 15684
    },
    {
      "epoch": 0.6332256762212354,
      "grad_norm": 97.37279510498047,
      "learning_rate": 6.26425497743274e-06,
      "loss": 4.5501,
      "step": 15685
    },
    {
      "epoch": 0.6332660476382721,
      "grad_norm": 93.7419662475586,
      "learning_rate": 6.263042097565902e-06,
      "loss": 5.4315,
      "step": 15686
    },
    {
      "epoch": 0.6333064190553088,
      "grad_norm": 108.36610412597656,
      "learning_rate": 6.261829281592313e-06,
      "loss": 6.6375,
      "step": 15687
    },
    {
      "epoch": 0.6333467904723455,
      "grad_norm": 167.7283935546875,
      "learning_rate": 6.260616529532715e-06,
      "loss": 4.9221,
      "step": 15688
    },
    {
      "epoch": 0.6333871618893823,
      "grad_norm": 111.35334014892578,
      "learning_rate": 6.259403841407842e-06,
      "loss": 5.348,
      "step": 15689
    },
    {
      "epoch": 0.6334275333064191,
      "grad_norm": 115.4997787475586,
      "learning_rate": 6.258191217238432e-06,
      "loss": 5.1691,
      "step": 15690
    },
    {
      "epoch": 0.6334679047234558,
      "grad_norm": 191.86708068847656,
      "learning_rate": 6.256978657045211e-06,
      "loss": 3.4342,
      "step": 15691
    },
    {
      "epoch": 0.6335082761404925,
      "grad_norm": 128.0228271484375,
      "learning_rate": 6.255766160848913e-06,
      "loss": 4.3583,
      "step": 15692
    },
    {
      "epoch": 0.6335486475575293,
      "grad_norm": 105.21160125732422,
      "learning_rate": 6.254553728670272e-06,
      "loss": 6.0271,
      "step": 15693
    },
    {
      "epoch": 0.633589018974566,
      "grad_norm": 102.41786193847656,
      "learning_rate": 6.2533413605300165e-06,
      "loss": 4.495,
      "step": 15694
    },
    {
      "epoch": 0.6336293903916027,
      "grad_norm": 87.3792495727539,
      "learning_rate": 6.252129056448874e-06,
      "loss": 5.1334,
      "step": 15695
    },
    {
      "epoch": 0.6336697618086394,
      "grad_norm": 98.8723373413086,
      "learning_rate": 6.250916816447572e-06,
      "loss": 5.5551,
      "step": 15696
    },
    {
      "epoch": 0.6337101332256763,
      "grad_norm": 91.72966003417969,
      "learning_rate": 6.249704640546839e-06,
      "loss": 4.9158,
      "step": 15697
    },
    {
      "epoch": 0.633750504642713,
      "grad_norm": 110.07892608642578,
      "learning_rate": 6.2484925287674e-06,
      "loss": 5.4285,
      "step": 15698
    },
    {
      "epoch": 0.6337908760597497,
      "grad_norm": 112.85054779052734,
      "learning_rate": 6.247280481129977e-06,
      "loss": 5.3766,
      "step": 15699
    },
    {
      "epoch": 0.6338312474767864,
      "grad_norm": 78.44147491455078,
      "learning_rate": 6.246068497655293e-06,
      "loss": 3.5023,
      "step": 15700
    },
    {
      "epoch": 0.6338716188938232,
      "grad_norm": 94.79393768310547,
      "learning_rate": 6.244856578364073e-06,
      "loss": 4.7017,
      "step": 15701
    },
    {
      "epoch": 0.6339119903108599,
      "grad_norm": 109.71342468261719,
      "learning_rate": 6.243644723277038e-06,
      "loss": 3.5561,
      "step": 15702
    },
    {
      "epoch": 0.6339523617278966,
      "grad_norm": 80.84796142578125,
      "learning_rate": 6.242432932414905e-06,
      "loss": 4.4654,
      "step": 15703
    },
    {
      "epoch": 0.6339927331449334,
      "grad_norm": 99.26174926757812,
      "learning_rate": 6.241221205798393e-06,
      "loss": 4.0813,
      "step": 15704
    },
    {
      "epoch": 0.6340331045619702,
      "grad_norm": 93.6109619140625,
      "learning_rate": 6.240009543448222e-06,
      "loss": 5.6351,
      "step": 15705
    },
    {
      "epoch": 0.6340734759790069,
      "grad_norm": 81.17045593261719,
      "learning_rate": 6.238797945385112e-06,
      "loss": 4.2175,
      "step": 15706
    },
    {
      "epoch": 0.6341138473960436,
      "grad_norm": 104.41255950927734,
      "learning_rate": 6.23758641162977e-06,
      "loss": 6.4629,
      "step": 15707
    },
    {
      "epoch": 0.6341542188130803,
      "grad_norm": 87.6516342163086,
      "learning_rate": 6.2363749422029144e-06,
      "loss": 4.2855,
      "step": 15708
    },
    {
      "epoch": 0.634194590230117,
      "grad_norm": 122.43666076660156,
      "learning_rate": 6.235163537125257e-06,
      "loss": 3.802,
      "step": 15709
    },
    {
      "epoch": 0.6342349616471538,
      "grad_norm": 83.85144805908203,
      "learning_rate": 6.233952196417515e-06,
      "loss": 4.2643,
      "step": 15710
    },
    {
      "epoch": 0.6342753330641906,
      "grad_norm": 85.3882064819336,
      "learning_rate": 6.232740920100394e-06,
      "loss": 6.3939,
      "step": 15711
    },
    {
      "epoch": 0.6343157044812273,
      "grad_norm": 98.23442077636719,
      "learning_rate": 6.231529708194606e-06,
      "loss": 4.7057,
      "step": 15712
    },
    {
      "epoch": 0.634356075898264,
      "grad_norm": 88.4146728515625,
      "learning_rate": 6.230318560720863e-06,
      "loss": 5.8396,
      "step": 15713
    },
    {
      "epoch": 0.6343964473153008,
      "grad_norm": 88.58523559570312,
      "learning_rate": 6.229107477699869e-06,
      "loss": 5.3505,
      "step": 15714
    },
    {
      "epoch": 0.6344368187323375,
      "grad_norm": 134.56845092773438,
      "learning_rate": 6.227896459152328e-06,
      "loss": 5.7543,
      "step": 15715
    },
    {
      "epoch": 0.6344771901493742,
      "grad_norm": 100.28543853759766,
      "learning_rate": 6.226685505098951e-06,
      "loss": 5.7675,
      "step": 15716
    },
    {
      "epoch": 0.6345175615664109,
      "grad_norm": 100.27076721191406,
      "learning_rate": 6.225474615560439e-06,
      "loss": 4.5752,
      "step": 15717
    },
    {
      "epoch": 0.6345579329834478,
      "grad_norm": 97.23091888427734,
      "learning_rate": 6.224263790557502e-06,
      "loss": 5.2443,
      "step": 15718
    },
    {
      "epoch": 0.6345983044004845,
      "grad_norm": 148.6534423828125,
      "learning_rate": 6.223053030110832e-06,
      "loss": 7.6997,
      "step": 15719
    },
    {
      "epoch": 0.6346386758175212,
      "grad_norm": 82.00448608398438,
      "learning_rate": 6.2218423342411375e-06,
      "loss": 4.4347,
      "step": 15720
    },
    {
      "epoch": 0.6346790472345579,
      "grad_norm": 98.93158721923828,
      "learning_rate": 6.220631702969115e-06,
      "loss": 5.0136,
      "step": 15721
    },
    {
      "epoch": 0.6347194186515946,
      "grad_norm": 116.99283599853516,
      "learning_rate": 6.219421136315467e-06,
      "loss": 5.8254,
      "step": 15722
    },
    {
      "epoch": 0.6347597900686314,
      "grad_norm": 88.00917053222656,
      "learning_rate": 6.2182106343008876e-06,
      "loss": 4.4095,
      "step": 15723
    },
    {
      "epoch": 0.6348001614856682,
      "grad_norm": 113.92463684082031,
      "learning_rate": 6.217000196946075e-06,
      "loss": 5.6139,
      "step": 15724
    },
    {
      "epoch": 0.6348405329027049,
      "grad_norm": 91.34590148925781,
      "learning_rate": 6.215789824271725e-06,
      "loss": 5.6516,
      "step": 15725
    },
    {
      "epoch": 0.6348809043197416,
      "grad_norm": 103.3228759765625,
      "learning_rate": 6.214579516298536e-06,
      "loss": 6.1195,
      "step": 15726
    },
    {
      "epoch": 0.6349212757367784,
      "grad_norm": 76.08854675292969,
      "learning_rate": 6.213369273047195e-06,
      "loss": 4.7989,
      "step": 15727
    },
    {
      "epoch": 0.6349616471538151,
      "grad_norm": 85.24433135986328,
      "learning_rate": 6.212159094538397e-06,
      "loss": 4.9557,
      "step": 15728
    },
    {
      "epoch": 0.6350020185708518,
      "grad_norm": 74.08898162841797,
      "learning_rate": 6.210948980792832e-06,
      "loss": 4.9087,
      "step": 15729
    },
    {
      "epoch": 0.6350423899878885,
      "grad_norm": 109.69390869140625,
      "learning_rate": 6.209738931831193e-06,
      "loss": 3.8046,
      "step": 15730
    },
    {
      "epoch": 0.6350827614049254,
      "grad_norm": 91.09781646728516,
      "learning_rate": 6.208528947674166e-06,
      "loss": 4.3677,
      "step": 15731
    },
    {
      "epoch": 0.6351231328219621,
      "grad_norm": 110.41105651855469,
      "learning_rate": 6.20731902834244e-06,
      "loss": 5.2471,
      "step": 15732
    },
    {
      "epoch": 0.6351635042389988,
      "grad_norm": 82.9543685913086,
      "learning_rate": 6.2061091738567026e-06,
      "loss": 4.4412,
      "step": 15733
    },
    {
      "epoch": 0.6352038756560355,
      "grad_norm": 106.6967544555664,
      "learning_rate": 6.2048993842376435e-06,
      "loss": 5.3562,
      "step": 15734
    },
    {
      "epoch": 0.6352442470730723,
      "grad_norm": 78.5594482421875,
      "learning_rate": 6.203689659505938e-06,
      "loss": 5.2256,
      "step": 15735
    },
    {
      "epoch": 0.635284618490109,
      "grad_norm": 90.60147094726562,
      "learning_rate": 6.2024799996822735e-06,
      "loss": 4.3578,
      "step": 15736
    },
    {
      "epoch": 0.6353249899071457,
      "grad_norm": 84.9573745727539,
      "learning_rate": 6.201270404787334e-06,
      "loss": 4.0485,
      "step": 15737
    },
    {
      "epoch": 0.6353653613241825,
      "grad_norm": 395.9751281738281,
      "learning_rate": 6.2000608748418045e-06,
      "loss": 4.9619,
      "step": 15738
    },
    {
      "epoch": 0.6354057327412193,
      "grad_norm": 68.14525604248047,
      "learning_rate": 6.198851409866356e-06,
      "loss": 4.3382,
      "step": 15739
    },
    {
      "epoch": 0.635446104158256,
      "grad_norm": 123.6962661743164,
      "learning_rate": 6.197642009881672e-06,
      "loss": 4.9452,
      "step": 15740
    },
    {
      "epoch": 0.6354864755752927,
      "grad_norm": 135.14906311035156,
      "learning_rate": 6.19643267490843e-06,
      "loss": 4.3571,
      "step": 15741
    },
    {
      "epoch": 0.6355268469923294,
      "grad_norm": 78.78971099853516,
      "learning_rate": 6.19522340496731e-06,
      "loss": 5.9999,
      "step": 15742
    },
    {
      "epoch": 0.6355672184093661,
      "grad_norm": 101.11678314208984,
      "learning_rate": 6.1940142000789835e-06,
      "loss": 5.7862,
      "step": 15743
    },
    {
      "epoch": 0.6356075898264029,
      "grad_norm": 84.34478759765625,
      "learning_rate": 6.192805060264126e-06,
      "loss": 5.361,
      "step": 15744
    },
    {
      "epoch": 0.6356479612434397,
      "grad_norm": 130.63546752929688,
      "learning_rate": 6.19159598554341e-06,
      "loss": 4.6904,
      "step": 15745
    },
    {
      "epoch": 0.6356883326604764,
      "grad_norm": 93.39932250976562,
      "learning_rate": 6.1903869759375155e-06,
      "loss": 3.875,
      "step": 15746
    },
    {
      "epoch": 0.6357287040775131,
      "grad_norm": 107.8160400390625,
      "learning_rate": 6.189178031467103e-06,
      "loss": 4.2719,
      "step": 15747
    },
    {
      "epoch": 0.6357690754945499,
      "grad_norm": 139.95065307617188,
      "learning_rate": 6.187969152152848e-06,
      "loss": 5.1027,
      "step": 15748
    },
    {
      "epoch": 0.6358094469115866,
      "grad_norm": 83.7325210571289,
      "learning_rate": 6.186760338015417e-06,
      "loss": 6.3137,
      "step": 15749
    },
    {
      "epoch": 0.6358498183286233,
      "grad_norm": 120.87854766845703,
      "learning_rate": 6.185551589075483e-06,
      "loss": 5.8657,
      "step": 15750
    },
    {
      "epoch": 0.63589018974566,
      "grad_norm": 101.19244384765625,
      "learning_rate": 6.184342905353706e-06,
      "loss": 4.0525,
      "step": 15751
    },
    {
      "epoch": 0.6359305611626969,
      "grad_norm": 79.10448455810547,
      "learning_rate": 6.183134286870757e-06,
      "loss": 5.0965,
      "step": 15752
    },
    {
      "epoch": 0.6359709325797336,
      "grad_norm": 74.23410034179688,
      "learning_rate": 6.1819257336472985e-06,
      "loss": 5.1418,
      "step": 15753
    },
    {
      "epoch": 0.6360113039967703,
      "grad_norm": 118.44235229492188,
      "learning_rate": 6.180717245703997e-06,
      "loss": 4.5454,
      "step": 15754
    },
    {
      "epoch": 0.636051675413807,
      "grad_norm": 163.8735809326172,
      "learning_rate": 6.179508823061508e-06,
      "loss": 6.1522,
      "step": 15755
    },
    {
      "epoch": 0.6360920468308437,
      "grad_norm": 84.0123519897461,
      "learning_rate": 6.178300465740497e-06,
      "loss": 4.4055,
      "step": 15756
    },
    {
      "epoch": 0.6361324182478805,
      "grad_norm": 116.56967163085938,
      "learning_rate": 6.177092173761623e-06,
      "loss": 4.4003,
      "step": 15757
    },
    {
      "epoch": 0.6361727896649172,
      "grad_norm": 86.50112915039062,
      "learning_rate": 6.1758839471455515e-06,
      "loss": 4.7523,
      "step": 15758
    },
    {
      "epoch": 0.636213161081954,
      "grad_norm": 76.40522766113281,
      "learning_rate": 6.174675785912929e-06,
      "loss": 5.9306,
      "step": 15759
    },
    {
      "epoch": 0.6362535324989907,
      "grad_norm": 72.75349426269531,
      "learning_rate": 6.1734676900844184e-06,
      "loss": 6.2793,
      "step": 15760
    },
    {
      "epoch": 0.6362939039160275,
      "grad_norm": 154.63497924804688,
      "learning_rate": 6.172259659680675e-06,
      "loss": 5.7844,
      "step": 15761
    },
    {
      "epoch": 0.6363342753330642,
      "grad_norm": 144.11985778808594,
      "learning_rate": 6.171051694722354e-06,
      "loss": 3.7779,
      "step": 15762
    },
    {
      "epoch": 0.6363746467501009,
      "grad_norm": 61.431640625,
      "learning_rate": 6.169843795230106e-06,
      "loss": 4.422,
      "step": 15763
    },
    {
      "epoch": 0.6364150181671376,
      "grad_norm": 93.718017578125,
      "learning_rate": 6.168635961224585e-06,
      "loss": 5.2111,
      "step": 15764
    },
    {
      "epoch": 0.6364553895841744,
      "grad_norm": 103.15897369384766,
      "learning_rate": 6.167428192726444e-06,
      "loss": 3.4041,
      "step": 15765
    },
    {
      "epoch": 0.6364957610012112,
      "grad_norm": 90.67617797851562,
      "learning_rate": 6.166220489756333e-06,
      "loss": 5.9964,
      "step": 15766
    },
    {
      "epoch": 0.6365361324182479,
      "grad_norm": 68.7520980834961,
      "learning_rate": 6.165012852334895e-06,
      "loss": 3.6509,
      "step": 15767
    },
    {
      "epoch": 0.6365765038352846,
      "grad_norm": 106.06431579589844,
      "learning_rate": 6.163805280482784e-06,
      "loss": 6.5218,
      "step": 15768
    },
    {
      "epoch": 0.6366168752523214,
      "grad_norm": 93.07833099365234,
      "learning_rate": 6.162597774220643e-06,
      "loss": 4.3577,
      "step": 15769
    },
    {
      "epoch": 0.6366572466693581,
      "grad_norm": 123.6806640625,
      "learning_rate": 6.161390333569122e-06,
      "loss": 6.2089,
      "step": 15770
    },
    {
      "epoch": 0.6366976180863948,
      "grad_norm": 115.01174926757812,
      "learning_rate": 6.16018295854886e-06,
      "loss": 5.2319,
      "step": 15771
    },
    {
      "epoch": 0.6367379895034315,
      "grad_norm": 71.45692443847656,
      "learning_rate": 6.158975649180502e-06,
      "loss": 5.8257,
      "step": 15772
    },
    {
      "epoch": 0.6367783609204684,
      "grad_norm": 98.16273498535156,
      "learning_rate": 6.157768405484692e-06,
      "loss": 6.0126,
      "step": 15773
    },
    {
      "epoch": 0.6368187323375051,
      "grad_norm": 90.05367279052734,
      "learning_rate": 6.156561227482073e-06,
      "loss": 6.3243,
      "step": 15774
    },
    {
      "epoch": 0.6368591037545418,
      "grad_norm": 81.33000946044922,
      "learning_rate": 6.1553541151932794e-06,
      "loss": 4.322,
      "step": 15775
    },
    {
      "epoch": 0.6368994751715785,
      "grad_norm": 76.77405548095703,
      "learning_rate": 6.154147068638952e-06,
      "loss": 4.847,
      "step": 15776
    },
    {
      "epoch": 0.6369398465886152,
      "grad_norm": 127.823486328125,
      "learning_rate": 6.1529400878397275e-06,
      "loss": 5.3261,
      "step": 15777
    },
    {
      "epoch": 0.636980218005652,
      "grad_norm": 109.87516784667969,
      "learning_rate": 6.15173317281625e-06,
      "loss": 5.4835,
      "step": 15778
    },
    {
      "epoch": 0.6370205894226887,
      "grad_norm": 91.8301010131836,
      "learning_rate": 6.150526323589143e-06,
      "loss": 5.9189,
      "step": 15779
    },
    {
      "epoch": 0.6370609608397255,
      "grad_norm": 102.85137939453125,
      "learning_rate": 6.149319540179048e-06,
      "loss": 5.6018,
      "step": 15780
    },
    {
      "epoch": 0.6371013322567622,
      "grad_norm": 93.87940216064453,
      "learning_rate": 6.148112822606598e-06,
      "loss": 3.561,
      "step": 15781
    },
    {
      "epoch": 0.637141703673799,
      "grad_norm": 94.94954681396484,
      "learning_rate": 6.1469061708924234e-06,
      "loss": 5.1961,
      "step": 15782
    },
    {
      "epoch": 0.6371820750908357,
      "grad_norm": 81.36114501953125,
      "learning_rate": 6.145699585057154e-06,
      "loss": 5.4185,
      "step": 15783
    },
    {
      "epoch": 0.6372224465078724,
      "grad_norm": 446.357666015625,
      "learning_rate": 6.144493065121423e-06,
      "loss": 5.6105,
      "step": 15784
    },
    {
      "epoch": 0.6372628179249091,
      "grad_norm": 99.10736083984375,
      "learning_rate": 6.143286611105856e-06,
      "loss": 5.1308,
      "step": 15785
    },
    {
      "epoch": 0.6373031893419459,
      "grad_norm": 83.32333374023438,
      "learning_rate": 6.142080223031087e-06,
      "loss": 5.0839,
      "step": 15786
    },
    {
      "epoch": 0.6373435607589827,
      "grad_norm": 112.29238891601562,
      "learning_rate": 6.140873900917732e-06,
      "loss": 6.4918,
      "step": 15787
    },
    {
      "epoch": 0.6373839321760194,
      "grad_norm": 127.7974624633789,
      "learning_rate": 6.1396676447864224e-06,
      "loss": 5.0134,
      "step": 15788
    },
    {
      "epoch": 0.6374243035930561,
      "grad_norm": 103.12632751464844,
      "learning_rate": 6.138461454657783e-06,
      "loss": 5.7616,
      "step": 15789
    },
    {
      "epoch": 0.6374646750100929,
      "grad_norm": 82.97885131835938,
      "learning_rate": 6.137255330552434e-06,
      "loss": 6.3934,
      "step": 15790
    },
    {
      "epoch": 0.6375050464271296,
      "grad_norm": 108.86686706542969,
      "learning_rate": 6.136049272490998e-06,
      "loss": 4.8033,
      "step": 15791
    },
    {
      "epoch": 0.6375454178441663,
      "grad_norm": 80.90213775634766,
      "learning_rate": 6.1348432804940984e-06,
      "loss": 4.932,
      "step": 15792
    },
    {
      "epoch": 0.637585789261203,
      "grad_norm": 127.14435577392578,
      "learning_rate": 6.133637354582351e-06,
      "loss": 4.6179,
      "step": 15793
    },
    {
      "epoch": 0.6376261606782399,
      "grad_norm": 119.48255157470703,
      "learning_rate": 6.132431494776381e-06,
      "loss": 4.8566,
      "step": 15794
    },
    {
      "epoch": 0.6376665320952766,
      "grad_norm": 100.66305541992188,
      "learning_rate": 6.131225701096797e-06,
      "loss": 5.4518,
      "step": 15795
    },
    {
      "epoch": 0.6377069035123133,
      "grad_norm": 91.5376968383789,
      "learning_rate": 6.130019973564219e-06,
      "loss": 5.4343,
      "step": 15796
    },
    {
      "epoch": 0.63774727492935,
      "grad_norm": 121.40276336669922,
      "learning_rate": 6.128814312199262e-06,
      "loss": 6.7992,
      "step": 15797
    },
    {
      "epoch": 0.6377876463463867,
      "grad_norm": 149.08677673339844,
      "learning_rate": 6.127608717022544e-06,
      "loss": 4.2236,
      "step": 15798
    },
    {
      "epoch": 0.6378280177634235,
      "grad_norm": 87.08724975585938,
      "learning_rate": 6.1264031880546705e-06,
      "loss": 5.3293,
      "step": 15799
    },
    {
      "epoch": 0.6378683891804602,
      "grad_norm": 100.08477020263672,
      "learning_rate": 6.125197725316259e-06,
      "loss": 7.0921,
      "step": 15800
    },
    {
      "epoch": 0.637908760597497,
      "grad_norm": 77.49007415771484,
      "learning_rate": 6.123992328827916e-06,
      "loss": 4.5022,
      "step": 15801
    },
    {
      "epoch": 0.6379491320145337,
      "grad_norm": 76.56304931640625,
      "learning_rate": 6.122786998610255e-06,
      "loss": 4.8336,
      "step": 15802
    },
    {
      "epoch": 0.6379895034315705,
      "grad_norm": 82.56864929199219,
      "learning_rate": 6.121581734683882e-06,
      "loss": 3.5948,
      "step": 15803
    },
    {
      "epoch": 0.6380298748486072,
      "grad_norm": 74.6773910522461,
      "learning_rate": 6.120376537069403e-06,
      "loss": 6.3868,
      "step": 15804
    },
    {
      "epoch": 0.6380702462656439,
      "grad_norm": 109.56387329101562,
      "learning_rate": 6.119171405787425e-06,
      "loss": 4.5408,
      "step": 15805
    },
    {
      "epoch": 0.6381106176826806,
      "grad_norm": 86.93719482421875,
      "learning_rate": 6.11796634085856e-06,
      "loss": 5.0999,
      "step": 15806
    },
    {
      "epoch": 0.6381509890997173,
      "grad_norm": 83.10493469238281,
      "learning_rate": 6.1167613423034e-06,
      "loss": 3.7303,
      "step": 15807
    },
    {
      "epoch": 0.6381913605167542,
      "grad_norm": 101.93653869628906,
      "learning_rate": 6.1155564101425535e-06,
      "loss": 6.3468,
      "step": 15808
    },
    {
      "epoch": 0.6382317319337909,
      "grad_norm": 86.37621307373047,
      "learning_rate": 6.1143515443966205e-06,
      "loss": 3.6791,
      "step": 15809
    },
    {
      "epoch": 0.6382721033508276,
      "grad_norm": 78.1769790649414,
      "learning_rate": 6.113146745086205e-06,
      "loss": 5.3731,
      "step": 15810
    },
    {
      "epoch": 0.6383124747678643,
      "grad_norm": 116.91449737548828,
      "learning_rate": 6.111942012231902e-06,
      "loss": 5.1984,
      "step": 15811
    },
    {
      "epoch": 0.6383528461849011,
      "grad_norm": 80.72530364990234,
      "learning_rate": 6.1107373458543115e-06,
      "loss": 4.2208,
      "step": 15812
    },
    {
      "epoch": 0.6383932176019378,
      "grad_norm": 118.55453491210938,
      "learning_rate": 6.10953274597403e-06,
      "loss": 4.9693,
      "step": 15813
    },
    {
      "epoch": 0.6384335890189745,
      "grad_norm": 82.7566909790039,
      "learning_rate": 6.1083282126116585e-06,
      "loss": 5.3052,
      "step": 15814
    },
    {
      "epoch": 0.6384739604360113,
      "grad_norm": 83.91492462158203,
      "learning_rate": 6.107123745787782e-06,
      "loss": 4.5724,
      "step": 15815
    },
    {
      "epoch": 0.6385143318530481,
      "grad_norm": 84.70465087890625,
      "learning_rate": 6.105919345522998e-06,
      "loss": 5.31,
      "step": 15816
    },
    {
      "epoch": 0.6385547032700848,
      "grad_norm": 130.82557678222656,
      "learning_rate": 6.104715011837901e-06,
      "loss": 4.153,
      "step": 15817
    },
    {
      "epoch": 0.6385950746871215,
      "grad_norm": 86.7561264038086,
      "learning_rate": 6.103510744753083e-06,
      "loss": 3.7921,
      "step": 15818
    },
    {
      "epoch": 0.6386354461041582,
      "grad_norm": 108.56673431396484,
      "learning_rate": 6.1023065442891305e-06,
      "loss": 6.4288,
      "step": 15819
    },
    {
      "epoch": 0.638675817521195,
      "grad_norm": 81.88420104980469,
      "learning_rate": 6.101102410466634e-06,
      "loss": 6.0688,
      "step": 15820
    },
    {
      "epoch": 0.6387161889382317,
      "grad_norm": 87.38983154296875,
      "learning_rate": 6.099898343306181e-06,
      "loss": 3.4485,
      "step": 15821
    },
    {
      "epoch": 0.6387565603552685,
      "grad_norm": 77.94245147705078,
      "learning_rate": 6.098694342828364e-06,
      "loss": 6.1136,
      "step": 15822
    },
    {
      "epoch": 0.6387969317723052,
      "grad_norm": 87.78758239746094,
      "learning_rate": 6.097490409053759e-06,
      "loss": 5.5909,
      "step": 15823
    },
    {
      "epoch": 0.638837303189342,
      "grad_norm": 79.86152648925781,
      "learning_rate": 6.096286542002956e-06,
      "loss": 4.7524,
      "step": 15824
    },
    {
      "epoch": 0.6388776746063787,
      "grad_norm": 107.5616683959961,
      "learning_rate": 6.095082741696537e-06,
      "loss": 4.9078,
      "step": 15825
    },
    {
      "epoch": 0.6389180460234154,
      "grad_norm": 82.40894317626953,
      "learning_rate": 6.093879008155088e-06,
      "loss": 6.5772,
      "step": 15826
    },
    {
      "epoch": 0.6389584174404521,
      "grad_norm": 120.74549865722656,
      "learning_rate": 6.092675341399182e-06,
      "loss": 4.7476,
      "step": 15827
    },
    {
      "epoch": 0.638998788857489,
      "grad_norm": 95.67568969726562,
      "learning_rate": 6.091471741449404e-06,
      "loss": 4.7714,
      "step": 15828
    },
    {
      "epoch": 0.6390391602745257,
      "grad_norm": 101.05445861816406,
      "learning_rate": 6.090268208326332e-06,
      "loss": 4.8204,
      "step": 15829
    },
    {
      "epoch": 0.6390795316915624,
      "grad_norm": 112.04314422607422,
      "learning_rate": 6.089064742050546e-06,
      "loss": 4.9989,
      "step": 15830
    },
    {
      "epoch": 0.6391199031085991,
      "grad_norm": 137.79904174804688,
      "learning_rate": 6.087861342642618e-06,
      "loss": 4.489,
      "step": 15831
    },
    {
      "epoch": 0.6391602745256358,
      "grad_norm": 103.01248931884766,
      "learning_rate": 6.086658010123125e-06,
      "loss": 5.0571,
      "step": 15832
    },
    {
      "epoch": 0.6392006459426726,
      "grad_norm": 110.91861724853516,
      "learning_rate": 6.0854547445126436e-06,
      "loss": 2.804,
      "step": 15833
    },
    {
      "epoch": 0.6392410173597093,
      "grad_norm": 85.85712432861328,
      "learning_rate": 6.084251545831748e-06,
      "loss": 5.4648,
      "step": 15834
    },
    {
      "epoch": 0.6392813887767461,
      "grad_norm": 95.91402435302734,
      "learning_rate": 6.0830484141010024e-06,
      "loss": 4.9822,
      "step": 15835
    },
    {
      "epoch": 0.6393217601937828,
      "grad_norm": 113.60425567626953,
      "learning_rate": 6.0818453493409824e-06,
      "loss": 3.3052,
      "step": 15836
    },
    {
      "epoch": 0.6393621316108196,
      "grad_norm": 121.29145050048828,
      "learning_rate": 6.080642351572257e-06,
      "loss": 3.4996,
      "step": 15837
    },
    {
      "epoch": 0.6394025030278563,
      "grad_norm": 91.10934448242188,
      "learning_rate": 6.079439420815398e-06,
      "loss": 3.7113,
      "step": 15838
    },
    {
      "epoch": 0.639442874444893,
      "grad_norm": 83.51142883300781,
      "learning_rate": 6.078236557090966e-06,
      "loss": 4.4182,
      "step": 15839
    },
    {
      "epoch": 0.6394832458619297,
      "grad_norm": 75.8481216430664,
      "learning_rate": 6.077033760419533e-06,
      "loss": 6.0787,
      "step": 15840
    },
    {
      "epoch": 0.6395236172789664,
      "grad_norm": 97.37663269042969,
      "learning_rate": 6.075831030821661e-06,
      "loss": 5.0632,
      "step": 15841
    },
    {
      "epoch": 0.6395639886960033,
      "grad_norm": 119.95405578613281,
      "learning_rate": 6.074628368317918e-06,
      "loss": 5.3234,
      "step": 15842
    },
    {
      "epoch": 0.63960436011304,
      "grad_norm": 56.89725875854492,
      "learning_rate": 6.0734257729288615e-06,
      "loss": 5.6775,
      "step": 15843
    },
    {
      "epoch": 0.6396447315300767,
      "grad_norm": 105.02408599853516,
      "learning_rate": 6.072223244675054e-06,
      "loss": 3.89,
      "step": 15844
    },
    {
      "epoch": 0.6396851029471134,
      "grad_norm": 105.90262603759766,
      "learning_rate": 6.071020783577057e-06,
      "loss": 5.3819,
      "step": 15845
    },
    {
      "epoch": 0.6397254743641502,
      "grad_norm": 126.89423370361328,
      "learning_rate": 6.069818389655433e-06,
      "loss": 5.3263,
      "step": 15846
    },
    {
      "epoch": 0.6397658457811869,
      "grad_norm": 71.16464233398438,
      "learning_rate": 6.068616062930735e-06,
      "loss": 5.348,
      "step": 15847
    },
    {
      "epoch": 0.6398062171982236,
      "grad_norm": 146.24278259277344,
      "learning_rate": 6.06741380342352e-06,
      "loss": 4.8949,
      "step": 15848
    },
    {
      "epoch": 0.6398465886152604,
      "grad_norm": 171.57444763183594,
      "learning_rate": 6.066211611154346e-06,
      "loss": 5.1283,
      "step": 15849
    },
    {
      "epoch": 0.6398869600322972,
      "grad_norm": 81.83682250976562,
      "learning_rate": 6.065009486143769e-06,
      "loss": 5.6497,
      "step": 15850
    },
    {
      "epoch": 0.6399273314493339,
      "grad_norm": 81.31877136230469,
      "learning_rate": 6.0638074284123404e-06,
      "loss": 4.3718,
      "step": 15851
    },
    {
      "epoch": 0.6399677028663706,
      "grad_norm": 122.67634582519531,
      "learning_rate": 6.062605437980611e-06,
      "loss": 4.6122,
      "step": 15852
    },
    {
      "epoch": 0.6400080742834073,
      "grad_norm": 110.73149871826172,
      "learning_rate": 6.061403514869135e-06,
      "loss": 4.7344,
      "step": 15853
    },
    {
      "epoch": 0.6400484457004441,
      "grad_norm": 740.1578369140625,
      "learning_rate": 6.060201659098467e-06,
      "loss": 5.2009,
      "step": 15854
    },
    {
      "epoch": 0.6400888171174808,
      "grad_norm": 121.89791107177734,
      "learning_rate": 6.0589998706891465e-06,
      "loss": 4.2306,
      "step": 15855
    },
    {
      "epoch": 0.6401291885345176,
      "grad_norm": 103.79999542236328,
      "learning_rate": 6.057798149661723e-06,
      "loss": 3.2918,
      "step": 15856
    },
    {
      "epoch": 0.6401695599515543,
      "grad_norm": 71.97604370117188,
      "learning_rate": 6.0565964960367466e-06,
      "loss": 3.58,
      "step": 15857
    },
    {
      "epoch": 0.6402099313685911,
      "grad_norm": 98.87033081054688,
      "learning_rate": 6.055394909834764e-06,
      "loss": 5.6889,
      "step": 15858
    },
    {
      "epoch": 0.6402503027856278,
      "grad_norm": 135.81332397460938,
      "learning_rate": 6.054193391076315e-06,
      "loss": 4.6623,
      "step": 15859
    },
    {
      "epoch": 0.6402906742026645,
      "grad_norm": 148.9156494140625,
      "learning_rate": 6.052991939781944e-06,
      "loss": 4.3592,
      "step": 15860
    },
    {
      "epoch": 0.6403310456197012,
      "grad_norm": 109.90254974365234,
      "learning_rate": 6.0517905559721956e-06,
      "loss": 5.2582,
      "step": 15861
    },
    {
      "epoch": 0.6403714170367379,
      "grad_norm": 83.9889907836914,
      "learning_rate": 6.0505892396676125e-06,
      "loss": 5.2986,
      "step": 15862
    },
    {
      "epoch": 0.6404117884537748,
      "grad_norm": 110.2366943359375,
      "learning_rate": 6.049387990888726e-06,
      "loss": 5.7916,
      "step": 15863
    },
    {
      "epoch": 0.6404521598708115,
      "grad_norm": 361.141357421875,
      "learning_rate": 6.04818680965608e-06,
      "loss": 5.2331,
      "step": 15864
    },
    {
      "epoch": 0.6404925312878482,
      "grad_norm": 121.1598129272461,
      "learning_rate": 6.0469856959902105e-06,
      "loss": 5.0738,
      "step": 15865
    },
    {
      "epoch": 0.6405329027048849,
      "grad_norm": 77.57550048828125,
      "learning_rate": 6.04578464991166e-06,
      "loss": 4.0334,
      "step": 15866
    },
    {
      "epoch": 0.6405732741219217,
      "grad_norm": 130.90440368652344,
      "learning_rate": 6.044583671440954e-06,
      "loss": 5.4108,
      "step": 15867
    },
    {
      "epoch": 0.6406136455389584,
      "grad_norm": 276.0523376464844,
      "learning_rate": 6.043382760598631e-06,
      "loss": 3.4473,
      "step": 15868
    },
    {
      "epoch": 0.6406540169559951,
      "grad_norm": 141.2278289794922,
      "learning_rate": 6.042181917405222e-06,
      "loss": 5.4926,
      "step": 15869
    },
    {
      "epoch": 0.6406943883730319,
      "grad_norm": 103.49552154541016,
      "learning_rate": 6.040981141881263e-06,
      "loss": 6.3873,
      "step": 15870
    },
    {
      "epoch": 0.6407347597900687,
      "grad_norm": 103.8110580444336,
      "learning_rate": 6.039780434047279e-06,
      "loss": 5.9873,
      "step": 15871
    },
    {
      "epoch": 0.6407751312071054,
      "grad_norm": 116.44357299804688,
      "learning_rate": 6.038579793923802e-06,
      "loss": 7.0522,
      "step": 15872
    },
    {
      "epoch": 0.6408155026241421,
      "grad_norm": 91.96878814697266,
      "learning_rate": 6.037379221531361e-06,
      "loss": 4.2479,
      "step": 15873
    },
    {
      "epoch": 0.6408558740411788,
      "grad_norm": 215.4889373779297,
      "learning_rate": 6.036178716890484e-06,
      "loss": 4.239,
      "step": 15874
    },
    {
      "epoch": 0.6408962454582156,
      "grad_norm": 114.42630004882812,
      "learning_rate": 6.034978280021692e-06,
      "loss": 4.988,
      "step": 15875
    },
    {
      "epoch": 0.6409366168752523,
      "grad_norm": 92.59300994873047,
      "learning_rate": 6.033777910945512e-06,
      "loss": 6.5088,
      "step": 15876
    },
    {
      "epoch": 0.6409769882922891,
      "grad_norm": 63.149967193603516,
      "learning_rate": 6.032577609682468e-06,
      "loss": 5.3638,
      "step": 15877
    },
    {
      "epoch": 0.6410173597093258,
      "grad_norm": 100.72883605957031,
      "learning_rate": 6.031377376253085e-06,
      "loss": 5.7081,
      "step": 15878
    },
    {
      "epoch": 0.6410577311263626,
      "grad_norm": 106.5587158203125,
      "learning_rate": 6.030177210677879e-06,
      "loss": 4.9949,
      "step": 15879
    },
    {
      "epoch": 0.6410981025433993,
      "grad_norm": 218.82492065429688,
      "learning_rate": 6.028977112977372e-06,
      "loss": 4.6379,
      "step": 15880
    },
    {
      "epoch": 0.641138473960436,
      "grad_norm": 82.40045166015625,
      "learning_rate": 6.027777083172084e-06,
      "loss": 4.8048,
      "step": 15881
    },
    {
      "epoch": 0.6411788453774727,
      "grad_norm": 89.53121185302734,
      "learning_rate": 6.026577121282535e-06,
      "loss": 5.7896,
      "step": 15882
    },
    {
      "epoch": 0.6412192167945094,
      "grad_norm": 98.00033569335938,
      "learning_rate": 6.025377227329235e-06,
      "loss": 5.0352,
      "step": 15883
    },
    {
      "epoch": 0.6412595882115463,
      "grad_norm": 57.20762252807617,
      "learning_rate": 6.024177401332705e-06,
      "loss": 5.5499,
      "step": 15884
    },
    {
      "epoch": 0.641299959628583,
      "grad_norm": 69.88262939453125,
      "learning_rate": 6.022977643313455e-06,
      "loss": 4.7761,
      "step": 15885
    },
    {
      "epoch": 0.6413403310456197,
      "grad_norm": 68.76067352294922,
      "learning_rate": 6.021777953292005e-06,
      "loss": 5.5778,
      "step": 15886
    },
    {
      "epoch": 0.6413807024626564,
      "grad_norm": 90.44497680664062,
      "learning_rate": 6.0205783312888576e-06,
      "loss": 5.0973,
      "step": 15887
    },
    {
      "epoch": 0.6414210738796932,
      "grad_norm": 96.67256927490234,
      "learning_rate": 6.0193787773245296e-06,
      "loss": 5.403,
      "step": 15888
    },
    {
      "epoch": 0.6414614452967299,
      "grad_norm": 123.13215637207031,
      "learning_rate": 6.018179291419527e-06,
      "loss": 5.595,
      "step": 15889
    },
    {
      "epoch": 0.6415018167137666,
      "grad_norm": 71.26459503173828,
      "learning_rate": 6.016979873594364e-06,
      "loss": 5.8742,
      "step": 15890
    },
    {
      "epoch": 0.6415421881308034,
      "grad_norm": 91.20833587646484,
      "learning_rate": 6.015780523869541e-06,
      "loss": 4.7252,
      "step": 15891
    },
    {
      "epoch": 0.6415825595478402,
      "grad_norm": 89.67597961425781,
      "learning_rate": 6.014581242265567e-06,
      "loss": 5.3342,
      "step": 15892
    },
    {
      "epoch": 0.6416229309648769,
      "grad_norm": 158.0019073486328,
      "learning_rate": 6.013382028802949e-06,
      "loss": 5.3388,
      "step": 15893
    },
    {
      "epoch": 0.6416633023819136,
      "grad_norm": 92.75804901123047,
      "learning_rate": 6.012182883502191e-06,
      "loss": 4.2854,
      "step": 15894
    },
    {
      "epoch": 0.6417036737989503,
      "grad_norm": 87.95328521728516,
      "learning_rate": 6.01098380638379e-06,
      "loss": 5.46,
      "step": 15895
    },
    {
      "epoch": 0.641744045215987,
      "grad_norm": 112.5186996459961,
      "learning_rate": 6.009784797468252e-06,
      "loss": 5.4098,
      "step": 15896
    },
    {
      "epoch": 0.6417844166330238,
      "grad_norm": 96.37947845458984,
      "learning_rate": 6.008585856776074e-06,
      "loss": 5.4377,
      "step": 15897
    },
    {
      "epoch": 0.6418247880500606,
      "grad_norm": 85.52256774902344,
      "learning_rate": 6.007386984327761e-06,
      "loss": 4.496,
      "step": 15898
    },
    {
      "epoch": 0.6418651594670973,
      "grad_norm": 83.06526947021484,
      "learning_rate": 6.006188180143805e-06,
      "loss": 5.198,
      "step": 15899
    },
    {
      "epoch": 0.641905530884134,
      "grad_norm": 107.95287322998047,
      "learning_rate": 6.004989444244704e-06,
      "loss": 4.2772,
      "step": 15900
    },
    {
      "epoch": 0.6419459023011708,
      "grad_norm": 93.55790710449219,
      "learning_rate": 6.003790776650955e-06,
      "loss": 6.2805,
      "step": 15901
    },
    {
      "epoch": 0.6419862737182075,
      "grad_norm": 73.80546569824219,
      "learning_rate": 6.002592177383056e-06,
      "loss": 5.1102,
      "step": 15902
    },
    {
      "epoch": 0.6420266451352442,
      "grad_norm": 75.9261245727539,
      "learning_rate": 6.001393646461493e-06,
      "loss": 6.2644,
      "step": 15903
    },
    {
      "epoch": 0.6420670165522809,
      "grad_norm": 77.71613311767578,
      "learning_rate": 6.0001951839067595e-06,
      "loss": 5.2221,
      "step": 15904
    },
    {
      "epoch": 0.6421073879693178,
      "grad_norm": 81.07170104980469,
      "learning_rate": 5.9989967897393486e-06,
      "loss": 5.0033,
      "step": 15905
    },
    {
      "epoch": 0.6421477593863545,
      "grad_norm": 76.5001220703125,
      "learning_rate": 5.997798463979755e-06,
      "loss": 4.1572,
      "step": 15906
    },
    {
      "epoch": 0.6421881308033912,
      "grad_norm": 85.35631561279297,
      "learning_rate": 5.996600206648457e-06,
      "loss": 5.3637,
      "step": 15907
    },
    {
      "epoch": 0.6422285022204279,
      "grad_norm": 81.55127716064453,
      "learning_rate": 5.995402017765946e-06,
      "loss": 6.1417,
      "step": 15908
    },
    {
      "epoch": 0.6422688736374647,
      "grad_norm": 124.32612609863281,
      "learning_rate": 5.99420389735271e-06,
      "loss": 6.0893,
      "step": 15909
    },
    {
      "epoch": 0.6423092450545014,
      "grad_norm": 74.38945007324219,
      "learning_rate": 5.993005845429234e-06,
      "loss": 6.3607,
      "step": 15910
    },
    {
      "epoch": 0.6423496164715381,
      "grad_norm": 126.83070373535156,
      "learning_rate": 5.991807862016e-06,
      "loss": 4.2311,
      "step": 15911
    },
    {
      "epoch": 0.6423899878885749,
      "grad_norm": 91.66413879394531,
      "learning_rate": 5.990609947133492e-06,
      "loss": 5.5915,
      "step": 15912
    },
    {
      "epoch": 0.6424303593056117,
      "grad_norm": 95.9952163696289,
      "learning_rate": 5.989412100802191e-06,
      "loss": 4.9961,
      "step": 15913
    },
    {
      "epoch": 0.6424707307226484,
      "grad_norm": 106.61151885986328,
      "learning_rate": 5.988214323042581e-06,
      "loss": 5.4319,
      "step": 15914
    },
    {
      "epoch": 0.6425111021396851,
      "grad_norm": 69.69518280029297,
      "learning_rate": 5.9870166138751354e-06,
      "loss": 5.6285,
      "step": 15915
    },
    {
      "epoch": 0.6425514735567218,
      "grad_norm": 94.1554946899414,
      "learning_rate": 5.985818973320334e-06,
      "loss": 5.2648,
      "step": 15916
    },
    {
      "epoch": 0.6425918449737585,
      "grad_norm": 87.41268157958984,
      "learning_rate": 5.984621401398654e-06,
      "loss": 5.8211,
      "step": 15917
    },
    {
      "epoch": 0.6426322163907953,
      "grad_norm": 110.55016326904297,
      "learning_rate": 5.983423898130574e-06,
      "loss": 5.3128,
      "step": 15918
    },
    {
      "epoch": 0.6426725878078321,
      "grad_norm": 91.13557434082031,
      "learning_rate": 5.982226463536562e-06,
      "loss": 6.0615,
      "step": 15919
    },
    {
      "epoch": 0.6427129592248688,
      "grad_norm": 72.8949966430664,
      "learning_rate": 5.981029097637099e-06,
      "loss": 4.0915,
      "step": 15920
    },
    {
      "epoch": 0.6427533306419055,
      "grad_norm": 99.45980834960938,
      "learning_rate": 5.979831800452651e-06,
      "loss": 4.3282,
      "step": 15921
    },
    {
      "epoch": 0.6427937020589423,
      "grad_norm": 150.29029846191406,
      "learning_rate": 5.978634572003695e-06,
      "loss": 5.4532,
      "step": 15922
    },
    {
      "epoch": 0.642834073475979,
      "grad_norm": 97.2147445678711,
      "learning_rate": 5.977437412310694e-06,
      "loss": 5.603,
      "step": 15923
    },
    {
      "epoch": 0.6428744448930157,
      "grad_norm": 98.53893280029297,
      "learning_rate": 5.97624032139412e-06,
      "loss": 5.921,
      "step": 15924
    },
    {
      "epoch": 0.6429148163100524,
      "grad_norm": 92.01011657714844,
      "learning_rate": 5.975043299274441e-06,
      "loss": 3.6762,
      "step": 15925
    },
    {
      "epoch": 0.6429551877270893,
      "grad_norm": 107.0214614868164,
      "learning_rate": 5.973846345972123e-06,
      "loss": 5.9632,
      "step": 15926
    },
    {
      "epoch": 0.642995559144126,
      "grad_norm": 79.88274383544922,
      "learning_rate": 5.97264946150763e-06,
      "loss": 6.0449,
      "step": 15927
    },
    {
      "epoch": 0.6430359305611627,
      "grad_norm": 112.92074584960938,
      "learning_rate": 5.971452645901426e-06,
      "loss": 5.5453,
      "step": 15928
    },
    {
      "epoch": 0.6430763019781994,
      "grad_norm": 100.80120086669922,
      "learning_rate": 5.970255899173976e-06,
      "loss": 6.1675,
      "step": 15929
    },
    {
      "epoch": 0.6431166733952361,
      "grad_norm": 102.1064682006836,
      "learning_rate": 5.969059221345742e-06,
      "loss": 4.409,
      "step": 15930
    },
    {
      "epoch": 0.6431570448122729,
      "grad_norm": 102.3572006225586,
      "learning_rate": 5.967862612437179e-06,
      "loss": 4.7564,
      "step": 15931
    },
    {
      "epoch": 0.6431974162293097,
      "grad_norm": 128.27734375,
      "learning_rate": 5.9666660724687495e-06,
      "loss": 4.3327,
      "step": 15932
    },
    {
      "epoch": 0.6432377876463464,
      "grad_norm": 116.86225891113281,
      "learning_rate": 5.965469601460912e-06,
      "loss": 2.8307,
      "step": 15933
    },
    {
      "epoch": 0.6432781590633831,
      "grad_norm": 80.95905303955078,
      "learning_rate": 5.964273199434128e-06,
      "loss": 3.8361,
      "step": 15934
    },
    {
      "epoch": 0.6433185304804199,
      "grad_norm": 106.25672149658203,
      "learning_rate": 5.963076866408844e-06,
      "loss": 6.9592,
      "step": 15935
    },
    {
      "epoch": 0.6433589018974566,
      "grad_norm": 104.63652038574219,
      "learning_rate": 5.961880602405516e-06,
      "loss": 5.264,
      "step": 15936
    },
    {
      "epoch": 0.6433992733144933,
      "grad_norm": 100.45660400390625,
      "learning_rate": 5.960684407444603e-06,
      "loss": 5.1234,
      "step": 15937
    },
    {
      "epoch": 0.64343964473153,
      "grad_norm": 159.87803649902344,
      "learning_rate": 5.959488281546555e-06,
      "loss": 5.3004,
      "step": 15938
    },
    {
      "epoch": 0.6434800161485669,
      "grad_norm": 90.35325622558594,
      "learning_rate": 5.958292224731819e-06,
      "loss": 5.2387,
      "step": 15939
    },
    {
      "epoch": 0.6435203875656036,
      "grad_norm": 69.72223663330078,
      "learning_rate": 5.95709623702085e-06,
      "loss": 6.0363,
      "step": 15940
    },
    {
      "epoch": 0.6435607589826403,
      "grad_norm": 80.06324768066406,
      "learning_rate": 5.955900318434092e-06,
      "loss": 4.5176,
      "step": 15941
    },
    {
      "epoch": 0.643601130399677,
      "grad_norm": 99.48526000976562,
      "learning_rate": 5.9547044689920006e-06,
      "loss": 4.3463,
      "step": 15942
    },
    {
      "epoch": 0.6436415018167138,
      "grad_norm": 95.97340393066406,
      "learning_rate": 5.953508688715013e-06,
      "loss": 4.2124,
      "step": 15943
    },
    {
      "epoch": 0.6436818732337505,
      "grad_norm": 91.21449279785156,
      "learning_rate": 5.952312977623575e-06,
      "loss": 4.976,
      "step": 15944
    },
    {
      "epoch": 0.6437222446507872,
      "grad_norm": 105.90042877197266,
      "learning_rate": 5.951117335738137e-06,
      "loss": 5.2856,
      "step": 15945
    },
    {
      "epoch": 0.643762616067824,
      "grad_norm": 94.83587646484375,
      "learning_rate": 5.9499217630791364e-06,
      "loss": 5.8481,
      "step": 15946
    },
    {
      "epoch": 0.6438029874848608,
      "grad_norm": 78.36524963378906,
      "learning_rate": 5.948726259667015e-06,
      "loss": 7.553,
      "step": 15947
    },
    {
      "epoch": 0.6438433589018975,
      "grad_norm": 97.19338989257812,
      "learning_rate": 5.947530825522214e-06,
      "loss": 4.4923,
      "step": 15948
    },
    {
      "epoch": 0.6438837303189342,
      "grad_norm": 95.2007064819336,
      "learning_rate": 5.946335460665174e-06,
      "loss": 4.6324,
      "step": 15949
    },
    {
      "epoch": 0.6439241017359709,
      "grad_norm": 84.03246307373047,
      "learning_rate": 5.945140165116334e-06,
      "loss": 6.9413,
      "step": 15950
    },
    {
      "epoch": 0.6439644731530076,
      "grad_norm": 83.54584503173828,
      "learning_rate": 5.9439449388961246e-06,
      "loss": 4.2038,
      "step": 15951
    },
    {
      "epoch": 0.6440048445700444,
      "grad_norm": 71.1390151977539,
      "learning_rate": 5.942749782024986e-06,
      "loss": 5.1144,
      "step": 15952
    },
    {
      "epoch": 0.6440452159870812,
      "grad_norm": 112.747314453125,
      "learning_rate": 5.941554694523351e-06,
      "loss": 4.1735,
      "step": 15953
    },
    {
      "epoch": 0.6440855874041179,
      "grad_norm": 97.82938385009766,
      "learning_rate": 5.940359676411658e-06,
      "loss": 5.7311,
      "step": 15954
    },
    {
      "epoch": 0.6441259588211546,
      "grad_norm": 72.29622650146484,
      "learning_rate": 5.9391647277103314e-06,
      "loss": 4.6764,
      "step": 15955
    },
    {
      "epoch": 0.6441663302381914,
      "grad_norm": 71.66752624511719,
      "learning_rate": 5.937969848439804e-06,
      "loss": 3.8072,
      "step": 15956
    },
    {
      "epoch": 0.6442067016552281,
      "grad_norm": 86.77714538574219,
      "learning_rate": 5.936775038620507e-06,
      "loss": 4.046,
      "step": 15957
    },
    {
      "epoch": 0.6442470730722648,
      "grad_norm": 106.00064849853516,
      "learning_rate": 5.935580298272871e-06,
      "loss": 4.5334,
      "step": 15958
    },
    {
      "epoch": 0.6442874444893015,
      "grad_norm": 113.87271881103516,
      "learning_rate": 5.934385627417319e-06,
      "loss": 5.7752,
      "step": 15959
    },
    {
      "epoch": 0.6443278159063384,
      "grad_norm": 93.1441650390625,
      "learning_rate": 5.933191026074277e-06,
      "loss": 6.0606,
      "step": 15960
    },
    {
      "epoch": 0.6443681873233751,
      "grad_norm": 76.78791809082031,
      "learning_rate": 5.931996494264173e-06,
      "loss": 5.0715,
      "step": 15961
    },
    {
      "epoch": 0.6444085587404118,
      "grad_norm": 83.03729248046875,
      "learning_rate": 5.930802032007433e-06,
      "loss": 5.335,
      "step": 15962
    },
    {
      "epoch": 0.6444489301574485,
      "grad_norm": 90.57855224609375,
      "learning_rate": 5.929607639324473e-06,
      "loss": 6.5138,
      "step": 15963
    },
    {
      "epoch": 0.6444893015744853,
      "grad_norm": 87.73871612548828,
      "learning_rate": 5.928413316235717e-06,
      "loss": 3.7723,
      "step": 15964
    },
    {
      "epoch": 0.644529672991522,
      "grad_norm": 99.70323944091797,
      "learning_rate": 5.927219062761585e-06,
      "loss": 4.3455,
      "step": 15965
    },
    {
      "epoch": 0.6445700444085587,
      "grad_norm": 119.11695098876953,
      "learning_rate": 5.926024878922498e-06,
      "loss": 6.0233,
      "step": 15966
    },
    {
      "epoch": 0.6446104158255955,
      "grad_norm": 80.60739135742188,
      "learning_rate": 5.924830764738872e-06,
      "loss": 5.2665,
      "step": 15967
    },
    {
      "epoch": 0.6446507872426323,
      "grad_norm": 89.53125762939453,
      "learning_rate": 5.923636720231122e-06,
      "loss": 4.1031,
      "step": 15968
    },
    {
      "epoch": 0.644691158659669,
      "grad_norm": 122.00308990478516,
      "learning_rate": 5.922442745419666e-06,
      "loss": 4.2546,
      "step": 15969
    },
    {
      "epoch": 0.6447315300767057,
      "grad_norm": 107.45442962646484,
      "learning_rate": 5.9212488403249204e-06,
      "loss": 3.2463,
      "step": 15970
    },
    {
      "epoch": 0.6447719014937424,
      "grad_norm": 64.71852111816406,
      "learning_rate": 5.920055004967291e-06,
      "loss": 4.5618,
      "step": 15971
    },
    {
      "epoch": 0.6448122729107791,
      "grad_norm": 106.61335754394531,
      "learning_rate": 5.9188612393671954e-06,
      "loss": 4.5436,
      "step": 15972
    },
    {
      "epoch": 0.6448526443278159,
      "grad_norm": 131.18658447265625,
      "learning_rate": 5.91766754354504e-06,
      "loss": 5.0096,
      "step": 15973
    },
    {
      "epoch": 0.6448930157448527,
      "grad_norm": 77.21076965332031,
      "learning_rate": 5.916473917521241e-06,
      "loss": 4.7453,
      "step": 15974
    },
    {
      "epoch": 0.6449333871618894,
      "grad_norm": 95.59138488769531,
      "learning_rate": 5.9152803613162e-06,
      "loss": 5.2,
      "step": 15975
    },
    {
      "epoch": 0.6449737585789261,
      "grad_norm": 85.13409423828125,
      "learning_rate": 5.914086874950326e-06,
      "loss": 5.4334,
      "step": 15976
    },
    {
      "epoch": 0.6450141299959629,
      "grad_norm": 99.53750610351562,
      "learning_rate": 5.912893458444024e-06,
      "loss": 2.7609,
      "step": 15977
    },
    {
      "epoch": 0.6450545014129996,
      "grad_norm": 86.94369506835938,
      "learning_rate": 5.911700111817703e-06,
      "loss": 5.383,
      "step": 15978
    },
    {
      "epoch": 0.6450948728300363,
      "grad_norm": 221.32119750976562,
      "learning_rate": 5.910506835091761e-06,
      "loss": 3.3829,
      "step": 15979
    },
    {
      "epoch": 0.645135244247073,
      "grad_norm": 166.7691650390625,
      "learning_rate": 5.9093136282866014e-06,
      "loss": 4.2275,
      "step": 15980
    },
    {
      "epoch": 0.6451756156641099,
      "grad_norm": 123.01100158691406,
      "learning_rate": 5.908120491422627e-06,
      "loss": 5.3508,
      "step": 15981
    },
    {
      "epoch": 0.6452159870811466,
      "grad_norm": 211.10939025878906,
      "learning_rate": 5.906927424520241e-06,
      "loss": 5.3957,
      "step": 15982
    },
    {
      "epoch": 0.6452563584981833,
      "grad_norm": 84.5771255493164,
      "learning_rate": 5.905734427599834e-06,
      "loss": 4.6443,
      "step": 15983
    },
    {
      "epoch": 0.64529672991522,
      "grad_norm": 91.72285461425781,
      "learning_rate": 5.904541500681808e-06,
      "loss": 4.9813,
      "step": 15984
    },
    {
      "epoch": 0.6453371013322567,
      "grad_norm": 148.20921325683594,
      "learning_rate": 5.9033486437865584e-06,
      "loss": 4.7591,
      "step": 15985
    },
    {
      "epoch": 0.6453774727492935,
      "grad_norm": 84.22760009765625,
      "learning_rate": 5.9021558569344814e-06,
      "loss": 5.9857,
      "step": 15986
    },
    {
      "epoch": 0.6454178441663302,
      "grad_norm": 85.17417907714844,
      "learning_rate": 5.90096314014597e-06,
      "loss": 4.5696,
      "step": 15987
    },
    {
      "epoch": 0.645458215583367,
      "grad_norm": 678.9989013671875,
      "learning_rate": 5.899770493441417e-06,
      "loss": 5.3333,
      "step": 15988
    },
    {
      "epoch": 0.6454985870004037,
      "grad_norm": 85.21841430664062,
      "learning_rate": 5.898577916841213e-06,
      "loss": 6.4878,
      "step": 15989
    },
    {
      "epoch": 0.6455389584174405,
      "grad_norm": 128.92015075683594,
      "learning_rate": 5.897385410365752e-06,
      "loss": 4.9308,
      "step": 15990
    },
    {
      "epoch": 0.6455793298344772,
      "grad_norm": 119.7442855834961,
      "learning_rate": 5.8961929740354175e-06,
      "loss": 7.0462,
      "step": 15991
    },
    {
      "epoch": 0.6456197012515139,
      "grad_norm": 80.12376403808594,
      "learning_rate": 5.8950006078706004e-06,
      "loss": 4.8289,
      "step": 15992
    },
    {
      "epoch": 0.6456600726685506,
      "grad_norm": 72.17241668701172,
      "learning_rate": 5.893808311891686e-06,
      "loss": 4.2907,
      "step": 15993
    },
    {
      "epoch": 0.6457004440855874,
      "grad_norm": 89.36685180664062,
      "learning_rate": 5.892616086119065e-06,
      "loss": 4.8398,
      "step": 15994
    },
    {
      "epoch": 0.6457408155026242,
      "grad_norm": 191.14230346679688,
      "learning_rate": 5.891423930573115e-06,
      "loss": 5.6862,
      "step": 15995
    },
    {
      "epoch": 0.6457811869196609,
      "grad_norm": 120.90306091308594,
      "learning_rate": 5.8902318452742195e-06,
      "loss": 3.0203,
      "step": 15996
    },
    {
      "epoch": 0.6458215583366976,
      "grad_norm": 295.4842529296875,
      "learning_rate": 5.8890398302427644e-06,
      "loss": 4.2195,
      "step": 15997
    },
    {
      "epoch": 0.6458619297537344,
      "grad_norm": 85.26703643798828,
      "learning_rate": 5.887847885499129e-06,
      "loss": 5.7945,
      "step": 15998
    },
    {
      "epoch": 0.6459023011707711,
      "grad_norm": 144.1226043701172,
      "learning_rate": 5.886656011063691e-06,
      "loss": 4.9614,
      "step": 15999
    },
    {
      "epoch": 0.6459426725878078,
      "grad_norm": 108.80854034423828,
      "learning_rate": 5.885464206956828e-06,
      "loss": 5.6156,
      "step": 16000
    },
    {
      "epoch": 0.6459830440048445,
      "grad_norm": 141.8448028564453,
      "learning_rate": 5.884272473198921e-06,
      "loss": 4.8732,
      "step": 16001
    },
    {
      "epoch": 0.6460234154218814,
      "grad_norm": 72.0770263671875,
      "learning_rate": 5.883080809810348e-06,
      "loss": 4.5663,
      "step": 16002
    },
    {
      "epoch": 0.6460637868389181,
      "grad_norm": 85.46367645263672,
      "learning_rate": 5.881889216811474e-06,
      "loss": 5.2107,
      "step": 16003
    },
    {
      "epoch": 0.6461041582559548,
      "grad_norm": 104.64769744873047,
      "learning_rate": 5.880697694222679e-06,
      "loss": 4.8397,
      "step": 16004
    },
    {
      "epoch": 0.6461445296729915,
      "grad_norm": 69.78890991210938,
      "learning_rate": 5.879506242064332e-06,
      "loss": 5.6651,
      "step": 16005
    },
    {
      "epoch": 0.6461849010900282,
      "grad_norm": 108.44638061523438,
      "learning_rate": 5.878314860356811e-06,
      "loss": 6.1006,
      "step": 16006
    },
    {
      "epoch": 0.646225272507065,
      "grad_norm": 107.19962310791016,
      "learning_rate": 5.877123549120478e-06,
      "loss": 5.0715,
      "step": 16007
    },
    {
      "epoch": 0.6462656439241017,
      "grad_norm": 89.56803894042969,
      "learning_rate": 5.875932308375705e-06,
      "loss": 4.5062,
      "step": 16008
    },
    {
      "epoch": 0.6463060153411385,
      "grad_norm": 72.918701171875,
      "learning_rate": 5.874741138142859e-06,
      "loss": 5.2355,
      "step": 16009
    },
    {
      "epoch": 0.6463463867581752,
      "grad_norm": 73.33659362792969,
      "learning_rate": 5.87355003844231e-06,
      "loss": 5.6386,
      "step": 16010
    },
    {
      "epoch": 0.646386758175212,
      "grad_norm": 122.12515258789062,
      "learning_rate": 5.872359009294415e-06,
      "loss": 5.2997,
      "step": 16011
    },
    {
      "epoch": 0.6464271295922487,
      "grad_norm": 92.49072265625,
      "learning_rate": 5.871168050719542e-06,
      "loss": 5.0169,
      "step": 16012
    },
    {
      "epoch": 0.6464675010092854,
      "grad_norm": 95.56439208984375,
      "learning_rate": 5.869977162738055e-06,
      "loss": 4.9889,
      "step": 16013
    },
    {
      "epoch": 0.6465078724263221,
      "grad_norm": 190.820556640625,
      "learning_rate": 5.868786345370316e-06,
      "loss": 6.3772,
      "step": 16014
    },
    {
      "epoch": 0.6465482438433588,
      "grad_norm": 174.60162353515625,
      "learning_rate": 5.867595598636684e-06,
      "loss": 6.0062,
      "step": 16015
    },
    {
      "epoch": 0.6465886152603957,
      "grad_norm": 95.10061645507812,
      "learning_rate": 5.866404922557514e-06,
      "loss": 4.2363,
      "step": 16016
    },
    {
      "epoch": 0.6466289866774324,
      "grad_norm": 116.10327911376953,
      "learning_rate": 5.865214317153168e-06,
      "loss": 4.6517,
      "step": 16017
    },
    {
      "epoch": 0.6466693580944691,
      "grad_norm": 98.483642578125,
      "learning_rate": 5.864023782444004e-06,
      "loss": 4.5046,
      "step": 16018
    },
    {
      "epoch": 0.6467097295115058,
      "grad_norm": 76.55609130859375,
      "learning_rate": 5.862833318450372e-06,
      "loss": 4.5616,
      "step": 16019
    },
    {
      "epoch": 0.6467501009285426,
      "grad_norm": 84.57935333251953,
      "learning_rate": 5.86164292519263e-06,
      "loss": 4.8584,
      "step": 16020
    },
    {
      "epoch": 0.6467904723455793,
      "grad_norm": 119.51860809326172,
      "learning_rate": 5.8604526026911305e-06,
      "loss": 5.7021,
      "step": 16021
    },
    {
      "epoch": 0.646830843762616,
      "grad_norm": 78.4375,
      "learning_rate": 5.859262350966228e-06,
      "loss": 4.7858,
      "step": 16022
    },
    {
      "epoch": 0.6468712151796528,
      "grad_norm": 109.33578491210938,
      "learning_rate": 5.858072170038267e-06,
      "loss": 7.1023,
      "step": 16023
    },
    {
      "epoch": 0.6469115865966896,
      "grad_norm": 78.48851013183594,
      "learning_rate": 5.8568820599276e-06,
      "loss": 3.8755,
      "step": 16024
    },
    {
      "epoch": 0.6469519580137263,
      "grad_norm": 80.68064880371094,
      "learning_rate": 5.8556920206545735e-06,
      "loss": 3.4427,
      "step": 16025
    },
    {
      "epoch": 0.646992329430763,
      "grad_norm": 103.72265625,
      "learning_rate": 5.854502052239539e-06,
      "loss": 5.0096,
      "step": 16026
    },
    {
      "epoch": 0.6470327008477997,
      "grad_norm": 148.10069274902344,
      "learning_rate": 5.853312154702837e-06,
      "loss": 4.4705,
      "step": 16027
    },
    {
      "epoch": 0.6470730722648365,
      "grad_norm": 85.854736328125,
      "learning_rate": 5.852122328064813e-06,
      "loss": 3.9805,
      "step": 16028
    },
    {
      "epoch": 0.6471134436818732,
      "grad_norm": 96.58466339111328,
      "learning_rate": 5.850932572345811e-06,
      "loss": 4.2804,
      "step": 16029
    },
    {
      "epoch": 0.64715381509891,
      "grad_norm": 84.44953918457031,
      "learning_rate": 5.849742887566179e-06,
      "loss": 5.826,
      "step": 16030
    },
    {
      "epoch": 0.6471941865159467,
      "grad_norm": 85.32713317871094,
      "learning_rate": 5.848553273746247e-06,
      "loss": 4.6486,
      "step": 16031
    },
    {
      "epoch": 0.6472345579329835,
      "grad_norm": 98.344482421875,
      "learning_rate": 5.847363730906359e-06,
      "loss": 5.8713,
      "step": 16032
    },
    {
      "epoch": 0.6472749293500202,
      "grad_norm": 113.7667007446289,
      "learning_rate": 5.8461742590668546e-06,
      "loss": 4.8645,
      "step": 16033
    },
    {
      "epoch": 0.6473153007670569,
      "grad_norm": 73.21881866455078,
      "learning_rate": 5.844984858248073e-06,
      "loss": 5.4562,
      "step": 16034
    },
    {
      "epoch": 0.6473556721840936,
      "grad_norm": 85.22470092773438,
      "learning_rate": 5.843795528470346e-06,
      "loss": 5.4206,
      "step": 16035
    },
    {
      "epoch": 0.6473960436011303,
      "grad_norm": 248.08737182617188,
      "learning_rate": 5.84260626975401e-06,
      "loss": 5.3001,
      "step": 16036
    },
    {
      "epoch": 0.6474364150181672,
      "grad_norm": 134.9375,
      "learning_rate": 5.841417082119397e-06,
      "loss": 4.7051,
      "step": 16037
    },
    {
      "epoch": 0.6474767864352039,
      "grad_norm": 63.92168045043945,
      "learning_rate": 5.8402279655868466e-06,
      "loss": 4.1994,
      "step": 16038
    },
    {
      "epoch": 0.6475171578522406,
      "grad_norm": 130.6625213623047,
      "learning_rate": 5.8390389201766784e-06,
      "loss": 5.3942,
      "step": 16039
    },
    {
      "epoch": 0.6475575292692773,
      "grad_norm": 116.82142639160156,
      "learning_rate": 5.837849945909229e-06,
      "loss": 5.6978,
      "step": 16040
    },
    {
      "epoch": 0.6475979006863141,
      "grad_norm": 159.74855041503906,
      "learning_rate": 5.836661042804827e-06,
      "loss": 4.8163,
      "step": 16041
    },
    {
      "epoch": 0.6476382721033508,
      "grad_norm": 129.66433715820312,
      "learning_rate": 5.8354722108838036e-06,
      "loss": 4.1314,
      "step": 16042
    },
    {
      "epoch": 0.6476786435203876,
      "grad_norm": 85.2699203491211,
      "learning_rate": 5.834283450166475e-06,
      "loss": 4.4932,
      "step": 16043
    },
    {
      "epoch": 0.6477190149374243,
      "grad_norm": 106.27961730957031,
      "learning_rate": 5.833094760673172e-06,
      "loss": 4.755,
      "step": 16044
    },
    {
      "epoch": 0.6477593863544611,
      "grad_norm": 86.96888732910156,
      "learning_rate": 5.831906142424218e-06,
      "loss": 5.779,
      "step": 16045
    },
    {
      "epoch": 0.6477997577714978,
      "grad_norm": 174.42974853515625,
      "learning_rate": 5.830717595439941e-06,
      "loss": 5.1354,
      "step": 16046
    },
    {
      "epoch": 0.6478401291885345,
      "grad_norm": 134.78700256347656,
      "learning_rate": 5.829529119740651e-06,
      "loss": 5.3797,
      "step": 16047
    },
    {
      "epoch": 0.6478805006055712,
      "grad_norm": 79.21454620361328,
      "learning_rate": 5.828340715346676e-06,
      "loss": 5.8238,
      "step": 16048
    },
    {
      "epoch": 0.647920872022608,
      "grad_norm": 201.69784545898438,
      "learning_rate": 5.8271523822783335e-06,
      "loss": 4.804,
      "step": 16049
    },
    {
      "epoch": 0.6479612434396448,
      "grad_norm": 87.54828643798828,
      "learning_rate": 5.82596412055594e-06,
      "loss": 6.7187,
      "step": 16050
    },
    {
      "epoch": 0.6480016148566815,
      "grad_norm": 119.11707305908203,
      "learning_rate": 5.824775930199813e-06,
      "loss": 5.9959,
      "step": 16051
    },
    {
      "epoch": 0.6480419862737182,
      "grad_norm": 69.00691986083984,
      "learning_rate": 5.823587811230268e-06,
      "loss": 5.1183,
      "step": 16052
    },
    {
      "epoch": 0.648082357690755,
      "grad_norm": 94.87435150146484,
      "learning_rate": 5.822399763667619e-06,
      "loss": 4.4499,
      "step": 16053
    },
    {
      "epoch": 0.6481227291077917,
      "grad_norm": 85.86979675292969,
      "learning_rate": 5.821211787532181e-06,
      "loss": 4.6553,
      "step": 16054
    },
    {
      "epoch": 0.6481631005248284,
      "grad_norm": 94.67179870605469,
      "learning_rate": 5.8200238828442625e-06,
      "loss": 4.663,
      "step": 16055
    },
    {
      "epoch": 0.6482034719418651,
      "grad_norm": 127.505126953125,
      "learning_rate": 5.8188360496241725e-06,
      "loss": 5.4153,
      "step": 16056
    },
    {
      "epoch": 0.648243843358902,
      "grad_norm": 82.93897247314453,
      "learning_rate": 5.8176482878922225e-06,
      "loss": 4.387,
      "step": 16057
    },
    {
      "epoch": 0.6482842147759387,
      "grad_norm": 106.08487701416016,
      "learning_rate": 5.816460597668725e-06,
      "loss": 5.9648,
      "step": 16058
    },
    {
      "epoch": 0.6483245861929754,
      "grad_norm": 78.37225341796875,
      "learning_rate": 5.815272978973977e-06,
      "loss": 3.9862,
      "step": 16059
    },
    {
      "epoch": 0.6483649576100121,
      "grad_norm": 78.0284652709961,
      "learning_rate": 5.814085431828289e-06,
      "loss": 4.4258,
      "step": 16060
    },
    {
      "epoch": 0.6484053290270488,
      "grad_norm": 151.28013610839844,
      "learning_rate": 5.812897956251966e-06,
      "loss": 2.7454,
      "step": 16061
    },
    {
      "epoch": 0.6484457004440856,
      "grad_norm": 74.78446960449219,
      "learning_rate": 5.811710552265315e-06,
      "loss": 5.9254,
      "step": 16062
    },
    {
      "epoch": 0.6484860718611223,
      "grad_norm": 118.1983413696289,
      "learning_rate": 5.810523219888628e-06,
      "loss": 4.6502,
      "step": 16063
    },
    {
      "epoch": 0.6485264432781591,
      "grad_norm": 94.65782165527344,
      "learning_rate": 5.80933595914221e-06,
      "loss": 4.7083,
      "step": 16064
    },
    {
      "epoch": 0.6485668146951958,
      "grad_norm": 88.77841186523438,
      "learning_rate": 5.808148770046363e-06,
      "loss": 5.1018,
      "step": 16065
    },
    {
      "epoch": 0.6486071861122326,
      "grad_norm": 83.46530151367188,
      "learning_rate": 5.806961652621388e-06,
      "loss": 4.2839,
      "step": 16066
    },
    {
      "epoch": 0.6486475575292693,
      "grad_norm": 70.05848693847656,
      "learning_rate": 5.805774606887571e-06,
      "loss": 5.2746,
      "step": 16067
    },
    {
      "epoch": 0.648687928946306,
      "grad_norm": 134.1524658203125,
      "learning_rate": 5.804587632865215e-06,
      "loss": 3.9459,
      "step": 16068
    },
    {
      "epoch": 0.6487283003633427,
      "grad_norm": 91.62350463867188,
      "learning_rate": 5.803400730574613e-06,
      "loss": 4.9426,
      "step": 16069
    },
    {
      "epoch": 0.6487686717803794,
      "grad_norm": 83.84314727783203,
      "learning_rate": 5.80221390003606e-06,
      "loss": 5.275,
      "step": 16070
    },
    {
      "epoch": 0.6488090431974163,
      "grad_norm": 96.71356201171875,
      "learning_rate": 5.801027141269846e-06,
      "loss": 4.6745,
      "step": 16071
    },
    {
      "epoch": 0.648849414614453,
      "grad_norm": 82.19953918457031,
      "learning_rate": 5.799840454296263e-06,
      "loss": 5.7687,
      "step": 16072
    },
    {
      "epoch": 0.6488897860314897,
      "grad_norm": 89.33208465576172,
      "learning_rate": 5.7986538391356e-06,
      "loss": 5.9856,
      "step": 16073
    },
    {
      "epoch": 0.6489301574485264,
      "grad_norm": 84.08877563476562,
      "learning_rate": 5.797467295808149e-06,
      "loss": 5.1887,
      "step": 16074
    },
    {
      "epoch": 0.6489705288655632,
      "grad_norm": 182.27975463867188,
      "learning_rate": 5.79628082433419e-06,
      "loss": 4.6615,
      "step": 16075
    },
    {
      "epoch": 0.6490109002825999,
      "grad_norm": 92.85230255126953,
      "learning_rate": 5.7950944247340115e-06,
      "loss": 5.0881,
      "step": 16076
    },
    {
      "epoch": 0.6490512716996366,
      "grad_norm": 112.26139831542969,
      "learning_rate": 5.7939080970279e-06,
      "loss": 3.4672,
      "step": 16077
    },
    {
      "epoch": 0.6490916431166734,
      "grad_norm": 147.11622619628906,
      "learning_rate": 5.792721841236143e-06,
      "loss": 5.3466,
      "step": 16078
    },
    {
      "epoch": 0.6491320145337102,
      "grad_norm": 117.67313385009766,
      "learning_rate": 5.791535657379013e-06,
      "loss": 5.4417,
      "step": 16079
    },
    {
      "epoch": 0.6491723859507469,
      "grad_norm": 105.02924346923828,
      "learning_rate": 5.790349545476796e-06,
      "loss": 4.8577,
      "step": 16080
    },
    {
      "epoch": 0.6492127573677836,
      "grad_norm": 118.00958251953125,
      "learning_rate": 5.789163505549771e-06,
      "loss": 5.9001,
      "step": 16081
    },
    {
      "epoch": 0.6492531287848203,
      "grad_norm": 82.4742431640625,
      "learning_rate": 5.787977537618221e-06,
      "loss": 4.9877,
      "step": 16082
    },
    {
      "epoch": 0.649293500201857,
      "grad_norm": 102.06480407714844,
      "learning_rate": 5.786791641702416e-06,
      "loss": 5.8469,
      "step": 16083
    },
    {
      "epoch": 0.6493338716188938,
      "grad_norm": 87.85889434814453,
      "learning_rate": 5.7856058178226345e-06,
      "loss": 5.3493,
      "step": 16084
    },
    {
      "epoch": 0.6493742430359306,
      "grad_norm": 74.73558807373047,
      "learning_rate": 5.784420065999154e-06,
      "loss": 3.8813,
      "step": 16085
    },
    {
      "epoch": 0.6494146144529673,
      "grad_norm": 82.05480194091797,
      "learning_rate": 5.783234386252248e-06,
      "loss": 4.2499,
      "step": 16086
    },
    {
      "epoch": 0.649454985870004,
      "grad_norm": 68.4787368774414,
      "learning_rate": 5.782048778602185e-06,
      "loss": 5.9454,
      "step": 16087
    },
    {
      "epoch": 0.6494953572870408,
      "grad_norm": 93.31725311279297,
      "learning_rate": 5.780863243069237e-06,
      "loss": 4.8769,
      "step": 16088
    },
    {
      "epoch": 0.6495357287040775,
      "grad_norm": 99.20236206054688,
      "learning_rate": 5.779677779673676e-06,
      "loss": 5.5145,
      "step": 16089
    },
    {
      "epoch": 0.6495761001211142,
      "grad_norm": 86.23356628417969,
      "learning_rate": 5.778492388435769e-06,
      "loss": 3.6764,
      "step": 16090
    },
    {
      "epoch": 0.6496164715381509,
      "grad_norm": 123.40129852294922,
      "learning_rate": 5.7773070693757835e-06,
      "loss": 6.047,
      "step": 16091
    },
    {
      "epoch": 0.6496568429551878,
      "grad_norm": 87.97357177734375,
      "learning_rate": 5.776121822513989e-06,
      "loss": 4.9954,
      "step": 16092
    },
    {
      "epoch": 0.6496972143722245,
      "grad_norm": 116.68492889404297,
      "learning_rate": 5.774936647870646e-06,
      "loss": 4.9235,
      "step": 16093
    },
    {
      "epoch": 0.6497375857892612,
      "grad_norm": 66.73546600341797,
      "learning_rate": 5.773751545466024e-06,
      "loss": 3.7458,
      "step": 16094
    },
    {
      "epoch": 0.6497779572062979,
      "grad_norm": 100.17718505859375,
      "learning_rate": 5.772566515320379e-06,
      "loss": 4.3096,
      "step": 16095
    },
    {
      "epoch": 0.6498183286233347,
      "grad_norm": 106.20581817626953,
      "learning_rate": 5.7713815574539735e-06,
      "loss": 3.9078,
      "step": 16096
    },
    {
      "epoch": 0.6498587000403714,
      "grad_norm": 82.07901000976562,
      "learning_rate": 5.770196671887069e-06,
      "loss": 3.9118,
      "step": 16097
    },
    {
      "epoch": 0.6498990714574081,
      "grad_norm": 133.39378356933594,
      "learning_rate": 5.769011858639928e-06,
      "loss": 3.9798,
      "step": 16098
    },
    {
      "epoch": 0.6499394428744449,
      "grad_norm": 101.6785888671875,
      "learning_rate": 5.7678271177328005e-06,
      "loss": 4.6742,
      "step": 16099
    },
    {
      "epoch": 0.6499798142914817,
      "grad_norm": 100.3077163696289,
      "learning_rate": 5.766642449185946e-06,
      "loss": 5.1671,
      "step": 16100
    },
    {
      "epoch": 0.6500201857085184,
      "grad_norm": 83.88363647460938,
      "learning_rate": 5.765457853019621e-06,
      "loss": 5.1884,
      "step": 16101
    },
    {
      "epoch": 0.6500605571255551,
      "grad_norm": 118.03150939941406,
      "learning_rate": 5.764273329254082e-06,
      "loss": 6.2577,
      "step": 16102
    },
    {
      "epoch": 0.6501009285425918,
      "grad_norm": 68.2518081665039,
      "learning_rate": 5.763088877909574e-06,
      "loss": 4.6028,
      "step": 16103
    },
    {
      "epoch": 0.6501412999596285,
      "grad_norm": 92.2651596069336,
      "learning_rate": 5.761904499006352e-06,
      "loss": 5.4049,
      "step": 16104
    },
    {
      "epoch": 0.6501816713766653,
      "grad_norm": 79.64795684814453,
      "learning_rate": 5.760720192564666e-06,
      "loss": 4.7969,
      "step": 16105
    },
    {
      "epoch": 0.6502220427937021,
      "grad_norm": 88.21990203857422,
      "learning_rate": 5.75953595860477e-06,
      "loss": 5.1333,
      "step": 16106
    },
    {
      "epoch": 0.6502624142107388,
      "grad_norm": 90.64071655273438,
      "learning_rate": 5.758351797146903e-06,
      "loss": 3.111,
      "step": 16107
    },
    {
      "epoch": 0.6503027856277755,
      "grad_norm": 109.70606994628906,
      "learning_rate": 5.757167708211315e-06,
      "loss": 5.6398,
      "step": 16108
    },
    {
      "epoch": 0.6503431570448123,
      "grad_norm": 106.29185485839844,
      "learning_rate": 5.755983691818252e-06,
      "loss": 5.5761,
      "step": 16109
    },
    {
      "epoch": 0.650383528461849,
      "grad_norm": 218.95506286621094,
      "learning_rate": 5.754799747987956e-06,
      "loss": 4.6619,
      "step": 16110
    },
    {
      "epoch": 0.6504238998788857,
      "grad_norm": 305.71826171875,
      "learning_rate": 5.7536158767406726e-06,
      "loss": 5.6968,
      "step": 16111
    },
    {
      "epoch": 0.6504642712959224,
      "grad_norm": 80.40032196044922,
      "learning_rate": 5.75243207809664e-06,
      "loss": 4.2176,
      "step": 16112
    },
    {
      "epoch": 0.6505046427129593,
      "grad_norm": 75.81674194335938,
      "learning_rate": 5.751248352076102e-06,
      "loss": 3.9193,
      "step": 16113
    },
    {
      "epoch": 0.650545014129996,
      "grad_norm": 90.51351928710938,
      "learning_rate": 5.750064698699298e-06,
      "loss": 4.6262,
      "step": 16114
    },
    {
      "epoch": 0.6505853855470327,
      "grad_norm": 95.59381866455078,
      "learning_rate": 5.74888111798646e-06,
      "loss": 6.4719,
      "step": 16115
    },
    {
      "epoch": 0.6506257569640694,
      "grad_norm": 90.2071762084961,
      "learning_rate": 5.7476976099578276e-06,
      "loss": 5.5453,
      "step": 16116
    },
    {
      "epoch": 0.6506661283811062,
      "grad_norm": 132.8812255859375,
      "learning_rate": 5.746514174633637e-06,
      "loss": 4.4294,
      "step": 16117
    },
    {
      "epoch": 0.6507064997981429,
      "grad_norm": 85.85498809814453,
      "learning_rate": 5.745330812034125e-06,
      "loss": 5.1063,
      "step": 16118
    },
    {
      "epoch": 0.6507468712151796,
      "grad_norm": 81.09674835205078,
      "learning_rate": 5.744147522179515e-06,
      "loss": 5.1056,
      "step": 16119
    },
    {
      "epoch": 0.6507872426322164,
      "grad_norm": 142.35073852539062,
      "learning_rate": 5.742964305090047e-06,
      "loss": 6.0576,
      "step": 16120
    },
    {
      "epoch": 0.6508276140492532,
      "grad_norm": 69.7660140991211,
      "learning_rate": 5.741781160785947e-06,
      "loss": 6.0653,
      "step": 16121
    },
    {
      "epoch": 0.6508679854662899,
      "grad_norm": 99.71804809570312,
      "learning_rate": 5.740598089287446e-06,
      "loss": 4.028,
      "step": 16122
    },
    {
      "epoch": 0.6509083568833266,
      "grad_norm": 89.2920150756836,
      "learning_rate": 5.739415090614775e-06,
      "loss": 4.243,
      "step": 16123
    },
    {
      "epoch": 0.6509487283003633,
      "grad_norm": 105.86309814453125,
      "learning_rate": 5.738232164788154e-06,
      "loss": 3.9961,
      "step": 16124
    },
    {
      "epoch": 0.6509890997174,
      "grad_norm": 76.34966278076172,
      "learning_rate": 5.737049311827809e-06,
      "loss": 4.381,
      "step": 16125
    },
    {
      "epoch": 0.6510294711344368,
      "grad_norm": 142.2386016845703,
      "learning_rate": 5.735866531753966e-06,
      "loss": 4.7239,
      "step": 16126
    },
    {
      "epoch": 0.6510698425514736,
      "grad_norm": 98.78717041015625,
      "learning_rate": 5.734683824586853e-06,
      "loss": 4.5291,
      "step": 16127
    },
    {
      "epoch": 0.6511102139685103,
      "grad_norm": 107.0935287475586,
      "learning_rate": 5.733501190346682e-06,
      "loss": 5.559,
      "step": 16128
    },
    {
      "epoch": 0.651150585385547,
      "grad_norm": 67.39969635009766,
      "learning_rate": 5.732318629053678e-06,
      "loss": 4.5237,
      "step": 16129
    },
    {
      "epoch": 0.6511909568025838,
      "grad_norm": 112.4928207397461,
      "learning_rate": 5.73113614072806e-06,
      "loss": 1.2571,
      "step": 16130
    },
    {
      "epoch": 0.6512313282196205,
      "grad_norm": 80.94929504394531,
      "learning_rate": 5.729953725390045e-06,
      "loss": 3.2998,
      "step": 16131
    },
    {
      "epoch": 0.6512716996366572,
      "grad_norm": 118.75724029541016,
      "learning_rate": 5.72877138305985e-06,
      "loss": 4.3454,
      "step": 16132
    },
    {
      "epoch": 0.6513120710536939,
      "grad_norm": 67.22399139404297,
      "learning_rate": 5.72758911375769e-06,
      "loss": 5.3862,
      "step": 16133
    },
    {
      "epoch": 0.6513524424707308,
      "grad_norm": 120.72520446777344,
      "learning_rate": 5.726406917503779e-06,
      "loss": 3.9291,
      "step": 16134
    },
    {
      "epoch": 0.6513928138877675,
      "grad_norm": 151.12124633789062,
      "learning_rate": 5.725224794318336e-06,
      "loss": 5.3633,
      "step": 16135
    },
    {
      "epoch": 0.6514331853048042,
      "grad_norm": 117.5463638305664,
      "learning_rate": 5.724042744221561e-06,
      "loss": 4.5522,
      "step": 16136
    },
    {
      "epoch": 0.6514735567218409,
      "grad_norm": 120.29212188720703,
      "learning_rate": 5.72286076723367e-06,
      "loss": 4.3555,
      "step": 16137
    },
    {
      "epoch": 0.6515139281388777,
      "grad_norm": 100.14582061767578,
      "learning_rate": 5.721678863374873e-06,
      "loss": 4.313,
      "step": 16138
    },
    {
      "epoch": 0.6515542995559144,
      "grad_norm": 75.75341033935547,
      "learning_rate": 5.72049703266538e-06,
      "loss": 5.7339,
      "step": 16139
    },
    {
      "epoch": 0.6515946709729511,
      "grad_norm": 116.268310546875,
      "learning_rate": 5.7193152751253914e-06,
      "loss": 5.4977,
      "step": 16140
    },
    {
      "epoch": 0.6516350423899879,
      "grad_norm": 81.89106750488281,
      "learning_rate": 5.718133590775114e-06,
      "loss": 5.7276,
      "step": 16141
    },
    {
      "epoch": 0.6516754138070247,
      "grad_norm": 122.6802749633789,
      "learning_rate": 5.716951979634755e-06,
      "loss": 5.4649,
      "step": 16142
    },
    {
      "epoch": 0.6517157852240614,
      "grad_norm": 114.03240203857422,
      "learning_rate": 5.7157704417245175e-06,
      "loss": 5.7424,
      "step": 16143
    },
    {
      "epoch": 0.6517561566410981,
      "grad_norm": 104.3568115234375,
      "learning_rate": 5.714588977064599e-06,
      "loss": 4.7925,
      "step": 16144
    },
    {
      "epoch": 0.6517965280581348,
      "grad_norm": 92.39812469482422,
      "learning_rate": 5.713407585675201e-06,
      "loss": 5.9394,
      "step": 16145
    },
    {
      "epoch": 0.6518368994751715,
      "grad_norm": 91.7237319946289,
      "learning_rate": 5.7122262675765235e-06,
      "loss": 5.1661,
      "step": 16146
    },
    {
      "epoch": 0.6518772708922084,
      "grad_norm": 149.01638793945312,
      "learning_rate": 5.711045022788768e-06,
      "loss": 6.3441,
      "step": 16147
    },
    {
      "epoch": 0.6519176423092451,
      "grad_norm": 95.55508422851562,
      "learning_rate": 5.709863851332123e-06,
      "loss": 5.1595,
      "step": 16148
    },
    {
      "epoch": 0.6519580137262818,
      "grad_norm": 182.5460968017578,
      "learning_rate": 5.7086827532267885e-06,
      "loss": 4.4957,
      "step": 16149
    },
    {
      "epoch": 0.6519983851433185,
      "grad_norm": 160.9015655517578,
      "learning_rate": 5.707501728492958e-06,
      "loss": 5.788,
      "step": 16150
    },
    {
      "epoch": 0.6520387565603553,
      "grad_norm": 90.02217102050781,
      "learning_rate": 5.706320777150825e-06,
      "loss": 4.1913,
      "step": 16151
    },
    {
      "epoch": 0.652079127977392,
      "grad_norm": 83.87806701660156,
      "learning_rate": 5.705139899220577e-06,
      "loss": 5.0163,
      "step": 16152
    },
    {
      "epoch": 0.6521194993944287,
      "grad_norm": 84.79415130615234,
      "learning_rate": 5.703959094722411e-06,
      "loss": 5.3685,
      "step": 16153
    },
    {
      "epoch": 0.6521598708114655,
      "grad_norm": 152.3244171142578,
      "learning_rate": 5.702778363676511e-06,
      "loss": 2.9446,
      "step": 16154
    },
    {
      "epoch": 0.6522002422285023,
      "grad_norm": 80.69916534423828,
      "learning_rate": 5.70159770610307e-06,
      "loss": 4.5527,
      "step": 16155
    },
    {
      "epoch": 0.652240613645539,
      "grad_norm": 73.36785888671875,
      "learning_rate": 5.700417122022267e-06,
      "loss": 3.8353,
      "step": 16156
    },
    {
      "epoch": 0.6522809850625757,
      "grad_norm": 87.03109741210938,
      "learning_rate": 5.69923661145429e-06,
      "loss": 4.685,
      "step": 16157
    },
    {
      "epoch": 0.6523213564796124,
      "grad_norm": 101.5355453491211,
      "learning_rate": 5.698056174419326e-06,
      "loss": 5.1224,
      "step": 16158
    },
    {
      "epoch": 0.6523617278966491,
      "grad_norm": 86.318115234375,
      "learning_rate": 5.6968758109375585e-06,
      "loss": 5.4723,
      "step": 16159
    },
    {
      "epoch": 0.6524020993136859,
      "grad_norm": 211.10995483398438,
      "learning_rate": 5.695695521029163e-06,
      "loss": 5.499,
      "step": 16160
    },
    {
      "epoch": 0.6524424707307227,
      "grad_norm": 126.0676040649414,
      "learning_rate": 5.694515304714322e-06,
      "loss": 4.4394,
      "step": 16161
    },
    {
      "epoch": 0.6524828421477594,
      "grad_norm": 93.41410827636719,
      "learning_rate": 5.693335162013216e-06,
      "loss": 4.4121,
      "step": 16162
    },
    {
      "epoch": 0.6525232135647961,
      "grad_norm": 97.00178527832031,
      "learning_rate": 5.692155092946025e-06,
      "loss": 5.7258,
      "step": 16163
    },
    {
      "epoch": 0.6525635849818329,
      "grad_norm": 79.09780883789062,
      "learning_rate": 5.690975097532918e-06,
      "loss": 4.4161,
      "step": 16164
    },
    {
      "epoch": 0.6526039563988696,
      "grad_norm": 97.80340576171875,
      "learning_rate": 5.6897951757940766e-06,
      "loss": 5.4318,
      "step": 16165
    },
    {
      "epoch": 0.6526443278159063,
      "grad_norm": 124.31221771240234,
      "learning_rate": 5.688615327749671e-06,
      "loss": 5.1621,
      "step": 16166
    },
    {
      "epoch": 0.652684699232943,
      "grad_norm": 73.28093719482422,
      "learning_rate": 5.68743555341988e-06,
      "loss": 4.0167,
      "step": 16167
    },
    {
      "epoch": 0.6527250706499799,
      "grad_norm": 83.61029815673828,
      "learning_rate": 5.686255852824866e-06,
      "loss": 5.4811,
      "step": 16168
    },
    {
      "epoch": 0.6527654420670166,
      "grad_norm": 85.6436996459961,
      "learning_rate": 5.685076225984803e-06,
      "loss": 5.8277,
      "step": 16169
    },
    {
      "epoch": 0.6528058134840533,
      "grad_norm": 91.93872833251953,
      "learning_rate": 5.683896672919862e-06,
      "loss": 5.2054,
      "step": 16170
    },
    {
      "epoch": 0.65284618490109,
      "grad_norm": 102.43895721435547,
      "learning_rate": 5.682717193650209e-06,
      "loss": 4.5637,
      "step": 16171
    },
    {
      "epoch": 0.6528865563181268,
      "grad_norm": 86.7010726928711,
      "learning_rate": 5.68153778819601e-06,
      "loss": 5.4627,
      "step": 16172
    },
    {
      "epoch": 0.6529269277351635,
      "grad_norm": 85.755126953125,
      "learning_rate": 5.680358456577429e-06,
      "loss": 4.8621,
      "step": 16173
    },
    {
      "epoch": 0.6529672991522002,
      "grad_norm": 178.48208618164062,
      "learning_rate": 5.679179198814633e-06,
      "loss": 3.0541,
      "step": 16174
    },
    {
      "epoch": 0.653007670569237,
      "grad_norm": 80.2386703491211,
      "learning_rate": 5.678000014927787e-06,
      "loss": 5.5942,
      "step": 16175
    },
    {
      "epoch": 0.6530480419862738,
      "grad_norm": 100.38355255126953,
      "learning_rate": 5.676820904937044e-06,
      "loss": 5.0325,
      "step": 16176
    },
    {
      "epoch": 0.6530884134033105,
      "grad_norm": 73.25164031982422,
      "learning_rate": 5.675641868862567e-06,
      "loss": 4.3693,
      "step": 16177
    },
    {
      "epoch": 0.6531287848203472,
      "grad_norm": 104.37210845947266,
      "learning_rate": 5.674462906724516e-06,
      "loss": 4.1444,
      "step": 16178
    },
    {
      "epoch": 0.6531691562373839,
      "grad_norm": 86.02989959716797,
      "learning_rate": 5.673284018543054e-06,
      "loss": 4.9499,
      "step": 16179
    },
    {
      "epoch": 0.6532095276544206,
      "grad_norm": 118.67191314697266,
      "learning_rate": 5.672105204338327e-06,
      "loss": 4.7529,
      "step": 16180
    },
    {
      "epoch": 0.6532498990714574,
      "grad_norm": 107.85919189453125,
      "learning_rate": 5.670926464130495e-06,
      "loss": 4.0851,
      "step": 16181
    },
    {
      "epoch": 0.6532902704884942,
      "grad_norm": 123.56932067871094,
      "learning_rate": 5.669747797939711e-06,
      "loss": 4.6267,
      "step": 16182
    },
    {
      "epoch": 0.6533306419055309,
      "grad_norm": 88.410888671875,
      "learning_rate": 5.668569205786131e-06,
      "loss": 5.6467,
      "step": 16183
    },
    {
      "epoch": 0.6533710133225676,
      "grad_norm": 92.85417175292969,
      "learning_rate": 5.6673906876899e-06,
      "loss": 4.8827,
      "step": 16184
    },
    {
      "epoch": 0.6534113847396044,
      "grad_norm": 80.54947662353516,
      "learning_rate": 5.66621224367117e-06,
      "loss": 5.9972,
      "step": 16185
    },
    {
      "epoch": 0.6534517561566411,
      "grad_norm": 79.29020690917969,
      "learning_rate": 5.665033873750091e-06,
      "loss": 5.1481,
      "step": 16186
    },
    {
      "epoch": 0.6534921275736778,
      "grad_norm": 86.58200073242188,
      "learning_rate": 5.663855577946809e-06,
      "loss": 4.1294,
      "step": 16187
    },
    {
      "epoch": 0.6535324989907145,
      "grad_norm": 98.94043731689453,
      "learning_rate": 5.662677356281472e-06,
      "loss": 5.9315,
      "step": 16188
    },
    {
      "epoch": 0.6535728704077514,
      "grad_norm": 129.41043090820312,
      "learning_rate": 5.661499208774224e-06,
      "loss": 6.8024,
      "step": 16189
    },
    {
      "epoch": 0.6536132418247881,
      "grad_norm": 70.72063446044922,
      "learning_rate": 5.660321135445208e-06,
      "loss": 5.7815,
      "step": 16190
    },
    {
      "epoch": 0.6536536132418248,
      "grad_norm": 108.62535095214844,
      "learning_rate": 5.659143136314571e-06,
      "loss": 5.544,
      "step": 16191
    },
    {
      "epoch": 0.6536939846588615,
      "grad_norm": 90.53184509277344,
      "learning_rate": 5.657965211402446e-06,
      "loss": 5.2693,
      "step": 16192
    },
    {
      "epoch": 0.6537343560758982,
      "grad_norm": 112.12197875976562,
      "learning_rate": 5.656787360728976e-06,
      "loss": 4.847,
      "step": 16193
    },
    {
      "epoch": 0.653774727492935,
      "grad_norm": 69.67908477783203,
      "learning_rate": 5.6556095843143e-06,
      "loss": 4.9119,
      "step": 16194
    },
    {
      "epoch": 0.6538150989099717,
      "grad_norm": 146.28631591796875,
      "learning_rate": 5.6544318821785595e-06,
      "loss": 5.645,
      "step": 16195
    },
    {
      "epoch": 0.6538554703270085,
      "grad_norm": 116.84615325927734,
      "learning_rate": 5.653254254341881e-06,
      "loss": 4.6515,
      "step": 16196
    },
    {
      "epoch": 0.6538958417440452,
      "grad_norm": 73.34254455566406,
      "learning_rate": 5.6520767008244056e-06,
      "loss": 3.9157,
      "step": 16197
    },
    {
      "epoch": 0.653936213161082,
      "grad_norm": 90.65877532958984,
      "learning_rate": 5.650899221646266e-06,
      "loss": 5.2692,
      "step": 16198
    },
    {
      "epoch": 0.6539765845781187,
      "grad_norm": 144.08551025390625,
      "learning_rate": 5.649721816827596e-06,
      "loss": 4.9205,
      "step": 16199
    },
    {
      "epoch": 0.6540169559951554,
      "grad_norm": 91.1072998046875,
      "learning_rate": 5.648544486388522e-06,
      "loss": 5.2743,
      "step": 16200
    },
    {
      "epoch": 0.6540573274121921,
      "grad_norm": 93.02052307128906,
      "learning_rate": 5.647367230349174e-06,
      "loss": 5.4513,
      "step": 16201
    },
    {
      "epoch": 0.6540976988292289,
      "grad_norm": 107.19036102294922,
      "learning_rate": 5.646190048729685e-06,
      "loss": 3.6298,
      "step": 16202
    },
    {
      "epoch": 0.6541380702462657,
      "grad_norm": 94.15128326416016,
      "learning_rate": 5.6450129415501805e-06,
      "loss": 5.4691,
      "step": 16203
    },
    {
      "epoch": 0.6541784416633024,
      "grad_norm": 74.35154724121094,
      "learning_rate": 5.643835908830784e-06,
      "loss": 4.6887,
      "step": 16204
    },
    {
      "epoch": 0.6542188130803391,
      "grad_norm": 86.81391143798828,
      "learning_rate": 5.642658950591618e-06,
      "loss": 3.532,
      "step": 16205
    },
    {
      "epoch": 0.6542591844973759,
      "grad_norm": 260.9339904785156,
      "learning_rate": 5.641482066852811e-06,
      "loss": 6.448,
      "step": 16206
    },
    {
      "epoch": 0.6542995559144126,
      "grad_norm": 104.46971130371094,
      "learning_rate": 5.640305257634482e-06,
      "loss": 6.3589,
      "step": 16207
    },
    {
      "epoch": 0.6543399273314493,
      "grad_norm": 93.89424133300781,
      "learning_rate": 5.6391285229567515e-06,
      "loss": 4.2886,
      "step": 16208
    },
    {
      "epoch": 0.654380298748486,
      "grad_norm": 100.13371276855469,
      "learning_rate": 5.637951862839741e-06,
      "loss": 3.2077,
      "step": 16209
    },
    {
      "epoch": 0.6544206701655229,
      "grad_norm": 130.44485473632812,
      "learning_rate": 5.636775277303568e-06,
      "loss": 4.6635,
      "step": 16210
    },
    {
      "epoch": 0.6544610415825596,
      "grad_norm": 84.67031860351562,
      "learning_rate": 5.635598766368353e-06,
      "loss": 5.0122,
      "step": 16211
    },
    {
      "epoch": 0.6545014129995963,
      "grad_norm": 90.88968658447266,
      "learning_rate": 5.634422330054203e-06,
      "loss": 5.0036,
      "step": 16212
    },
    {
      "epoch": 0.654541784416633,
      "grad_norm": 71.38694763183594,
      "learning_rate": 5.6332459683812356e-06,
      "loss": 7.0652,
      "step": 16213
    },
    {
      "epoch": 0.6545821558336697,
      "grad_norm": 77.1858901977539,
      "learning_rate": 5.632069681369567e-06,
      "loss": 3.3596,
      "step": 16214
    },
    {
      "epoch": 0.6546225272507065,
      "grad_norm": 483.7280578613281,
      "learning_rate": 5.630893469039309e-06,
      "loss": 5.4762,
      "step": 16215
    },
    {
      "epoch": 0.6546628986677432,
      "grad_norm": 94.61253356933594,
      "learning_rate": 5.629717331410567e-06,
      "loss": 5.2864,
      "step": 16216
    },
    {
      "epoch": 0.65470327008478,
      "grad_norm": 89.48384094238281,
      "learning_rate": 5.628541268503453e-06,
      "loss": 6.8165,
      "step": 16217
    },
    {
      "epoch": 0.6547436415018167,
      "grad_norm": 77.70584106445312,
      "learning_rate": 5.627365280338076e-06,
      "loss": 4.1961,
      "step": 16218
    },
    {
      "epoch": 0.6547840129188535,
      "grad_norm": 74.28028869628906,
      "learning_rate": 5.626189366934546e-06,
      "loss": 5.2075,
      "step": 16219
    },
    {
      "epoch": 0.6548243843358902,
      "grad_norm": 113.4105453491211,
      "learning_rate": 5.6250135283129595e-06,
      "loss": 3.3971,
      "step": 16220
    },
    {
      "epoch": 0.6548647557529269,
      "grad_norm": 118.67986297607422,
      "learning_rate": 5.623837764493425e-06,
      "loss": 5.6836,
      "step": 16221
    },
    {
      "epoch": 0.6549051271699636,
      "grad_norm": 99.98274993896484,
      "learning_rate": 5.622662075496047e-06,
      "loss": 4.326,
      "step": 16222
    },
    {
      "epoch": 0.6549454985870004,
      "grad_norm": 93.25511169433594,
      "learning_rate": 5.6214864613409284e-06,
      "loss": 5.8871,
      "step": 16223
    },
    {
      "epoch": 0.6549858700040372,
      "grad_norm": 91.66133880615234,
      "learning_rate": 5.620310922048165e-06,
      "loss": 5.3311,
      "step": 16224
    },
    {
      "epoch": 0.6550262414210739,
      "grad_norm": 69.44862365722656,
      "learning_rate": 5.619135457637854e-06,
      "loss": 5.0682,
      "step": 16225
    },
    {
      "epoch": 0.6550666128381106,
      "grad_norm": 278.34307861328125,
      "learning_rate": 5.6179600681301e-06,
      "loss": 5.3893,
      "step": 16226
    },
    {
      "epoch": 0.6551069842551474,
      "grad_norm": 144.0454864501953,
      "learning_rate": 5.616784753544992e-06,
      "loss": 6.0723,
      "step": 16227
    },
    {
      "epoch": 0.6551473556721841,
      "grad_norm": 127.39881896972656,
      "learning_rate": 5.615609513902633e-06,
      "loss": 5.0537,
      "step": 16228
    },
    {
      "epoch": 0.6551877270892208,
      "grad_norm": 86.50609588623047,
      "learning_rate": 5.614434349223113e-06,
      "loss": 4.5499,
      "step": 16229
    },
    {
      "epoch": 0.6552280985062575,
      "grad_norm": 96.95072937011719,
      "learning_rate": 5.613259259526522e-06,
      "loss": 5.7682,
      "step": 16230
    },
    {
      "epoch": 0.6552684699232943,
      "grad_norm": 106.31878662109375,
      "learning_rate": 5.612084244832958e-06,
      "loss": 4.784,
      "step": 16231
    },
    {
      "epoch": 0.6553088413403311,
      "grad_norm": 117.73869323730469,
      "learning_rate": 5.610909305162505e-06,
      "loss": 4.0264,
      "step": 16232
    },
    {
      "epoch": 0.6553492127573678,
      "grad_norm": 89.41802215576172,
      "learning_rate": 5.609734440535253e-06,
      "loss": 5.2723,
      "step": 16233
    },
    {
      "epoch": 0.6553895841744045,
      "grad_norm": 60.27815246582031,
      "learning_rate": 5.608559650971289e-06,
      "loss": 3.2051,
      "step": 16234
    },
    {
      "epoch": 0.6554299555914412,
      "grad_norm": 132.72132873535156,
      "learning_rate": 5.607384936490705e-06,
      "loss": 5.4543,
      "step": 16235
    },
    {
      "epoch": 0.655470327008478,
      "grad_norm": 80.29422760009766,
      "learning_rate": 5.606210297113577e-06,
      "loss": 3.9985,
      "step": 16236
    },
    {
      "epoch": 0.6555106984255147,
      "grad_norm": 103.06859588623047,
      "learning_rate": 5.6050357328599935e-06,
      "loss": 4.4433,
      "step": 16237
    },
    {
      "epoch": 0.6555510698425515,
      "grad_norm": 120.57620239257812,
      "learning_rate": 5.603861243750035e-06,
      "loss": 4.4253,
      "step": 16238
    },
    {
      "epoch": 0.6555914412595882,
      "grad_norm": 79.2862319946289,
      "learning_rate": 5.602686829803787e-06,
      "loss": 5.4896,
      "step": 16239
    },
    {
      "epoch": 0.655631812676625,
      "grad_norm": 92.3687744140625,
      "learning_rate": 5.6015124910413235e-06,
      "loss": 6.2276,
      "step": 16240
    },
    {
      "epoch": 0.6556721840936617,
      "grad_norm": 90.92499542236328,
      "learning_rate": 5.6003382274827235e-06,
      "loss": 5.5356,
      "step": 16241
    },
    {
      "epoch": 0.6557125555106984,
      "grad_norm": 109.78438568115234,
      "learning_rate": 5.599164039148066e-06,
      "loss": 4.5042,
      "step": 16242
    },
    {
      "epoch": 0.6557529269277351,
      "grad_norm": 146.07888793945312,
      "learning_rate": 5.597989926057432e-06,
      "loss": 5.2566,
      "step": 16243
    },
    {
      "epoch": 0.6557932983447718,
      "grad_norm": 123.95216369628906,
      "learning_rate": 5.5968158882308855e-06,
      "loss": 4.9599,
      "step": 16244
    },
    {
      "epoch": 0.6558336697618087,
      "grad_norm": 88.97149658203125,
      "learning_rate": 5.595641925688506e-06,
      "loss": 5.4719,
      "step": 16245
    },
    {
      "epoch": 0.6558740411788454,
      "grad_norm": 94.99031829833984,
      "learning_rate": 5.594468038450365e-06,
      "loss": 5.4811,
      "step": 16246
    },
    {
      "epoch": 0.6559144125958821,
      "grad_norm": 132.3415069580078,
      "learning_rate": 5.593294226536532e-06,
      "loss": 2.953,
      "step": 16247
    },
    {
      "epoch": 0.6559547840129188,
      "grad_norm": 64.36168670654297,
      "learning_rate": 5.5921204899670765e-06,
      "loss": 5.3207,
      "step": 16248
    },
    {
      "epoch": 0.6559951554299556,
      "grad_norm": 238.73504638671875,
      "learning_rate": 5.59094682876207e-06,
      "loss": 3.856,
      "step": 16249
    },
    {
      "epoch": 0.6560355268469923,
      "grad_norm": 119.50540161132812,
      "learning_rate": 5.589773242941574e-06,
      "loss": 5.9968,
      "step": 16250
    },
    {
      "epoch": 0.6560758982640291,
      "grad_norm": 136.65673828125,
      "learning_rate": 5.588599732525663e-06,
      "loss": 3.3765,
      "step": 16251
    },
    {
      "epoch": 0.6561162696810658,
      "grad_norm": 88.35623168945312,
      "learning_rate": 5.587426297534391e-06,
      "loss": 4.7343,
      "step": 16252
    },
    {
      "epoch": 0.6561566410981026,
      "grad_norm": 77.4673080444336,
      "learning_rate": 5.586252937987824e-06,
      "loss": 4.9343,
      "step": 16253
    },
    {
      "epoch": 0.6561970125151393,
      "grad_norm": 137.6798858642578,
      "learning_rate": 5.585079653906026e-06,
      "loss": 4.4117,
      "step": 16254
    },
    {
      "epoch": 0.656237383932176,
      "grad_norm": 82.51077270507812,
      "learning_rate": 5.58390644530906e-06,
      "loss": 5.6679,
      "step": 16255
    },
    {
      "epoch": 0.6562777553492127,
      "grad_norm": 104.77752685546875,
      "learning_rate": 5.582733312216977e-06,
      "loss": 4.8854,
      "step": 16256
    },
    {
      "epoch": 0.6563181267662495,
      "grad_norm": 145.52294921875,
      "learning_rate": 5.58156025464984e-06,
      "loss": 4.5913,
      "step": 16257
    },
    {
      "epoch": 0.6563584981832863,
      "grad_norm": 98.48844146728516,
      "learning_rate": 5.580387272627705e-06,
      "loss": 4.2426,
      "step": 16258
    },
    {
      "epoch": 0.656398869600323,
      "grad_norm": 67.38381958007812,
      "learning_rate": 5.579214366170631e-06,
      "loss": 3.954,
      "step": 16259
    },
    {
      "epoch": 0.6564392410173597,
      "grad_norm": 106.54562377929688,
      "learning_rate": 5.578041535298664e-06,
      "loss": 5.359,
      "step": 16260
    },
    {
      "epoch": 0.6564796124343965,
      "grad_norm": 87.27342224121094,
      "learning_rate": 5.57686878003186e-06,
      "loss": 4.4266,
      "step": 16261
    },
    {
      "epoch": 0.6565199838514332,
      "grad_norm": 86.6816635131836,
      "learning_rate": 5.575696100390271e-06,
      "loss": 5.9249,
      "step": 16262
    },
    {
      "epoch": 0.6565603552684699,
      "grad_norm": 92.08695983886719,
      "learning_rate": 5.574523496393952e-06,
      "loss": 4.7862,
      "step": 16263
    },
    {
      "epoch": 0.6566007266855066,
      "grad_norm": 71.85548400878906,
      "learning_rate": 5.573350968062943e-06,
      "loss": 4.014,
      "step": 16264
    },
    {
      "epoch": 0.6566410981025435,
      "grad_norm": 89.78916931152344,
      "learning_rate": 5.572178515417295e-06,
      "loss": 5.48,
      "step": 16265
    },
    {
      "epoch": 0.6566814695195802,
      "grad_norm": 83.27403259277344,
      "learning_rate": 5.5710061384770554e-06,
      "loss": 5.2272,
      "step": 16266
    },
    {
      "epoch": 0.6567218409366169,
      "grad_norm": 84.92509460449219,
      "learning_rate": 5.569833837262269e-06,
      "loss": 5.5826,
      "step": 16267
    },
    {
      "epoch": 0.6567622123536536,
      "grad_norm": 84.55003356933594,
      "learning_rate": 5.568661611792976e-06,
      "loss": 3.9775,
      "step": 16268
    },
    {
      "epoch": 0.6568025837706903,
      "grad_norm": 92.26461029052734,
      "learning_rate": 5.567489462089224e-06,
      "loss": 6.4551,
      "step": 16269
    },
    {
      "epoch": 0.6568429551877271,
      "grad_norm": 70.7203140258789,
      "learning_rate": 5.566317388171049e-06,
      "loss": 5.7717,
      "step": 16270
    },
    {
      "epoch": 0.6568833266047638,
      "grad_norm": 98.925048828125,
      "learning_rate": 5.5651453900585e-06,
      "loss": 4.8111,
      "step": 16271
    },
    {
      "epoch": 0.6569236980218006,
      "grad_norm": 77.35419464111328,
      "learning_rate": 5.563973467771604e-06,
      "loss": 4.7023,
      "step": 16272
    },
    {
      "epoch": 0.6569640694388373,
      "grad_norm": 81.9743423461914,
      "learning_rate": 5.562801621330402e-06,
      "loss": 4.8049,
      "step": 16273
    },
    {
      "epoch": 0.6570044408558741,
      "grad_norm": 178.94033813476562,
      "learning_rate": 5.561629850754932e-06,
      "loss": 4.4018,
      "step": 16274
    },
    {
      "epoch": 0.6570448122729108,
      "grad_norm": 110.84487915039062,
      "learning_rate": 5.560458156065231e-06,
      "loss": 4.8512,
      "step": 16275
    },
    {
      "epoch": 0.6570851836899475,
      "grad_norm": 77.53158569335938,
      "learning_rate": 5.559286537281323e-06,
      "loss": 5.6253,
      "step": 16276
    },
    {
      "epoch": 0.6571255551069842,
      "grad_norm": 93.59347534179688,
      "learning_rate": 5.558114994423247e-06,
      "loss": 4.0377,
      "step": 16277
    },
    {
      "epoch": 0.657165926524021,
      "grad_norm": 88.66988372802734,
      "learning_rate": 5.556943527511032e-06,
      "loss": 4.9233,
      "step": 16278
    },
    {
      "epoch": 0.6572062979410578,
      "grad_norm": 112.8025894165039,
      "learning_rate": 5.55577213656471e-06,
      "loss": 6.445,
      "step": 16279
    },
    {
      "epoch": 0.6572466693580945,
      "grad_norm": 144.68663024902344,
      "learning_rate": 5.5546008216043034e-06,
      "loss": 3.7999,
      "step": 16280
    },
    {
      "epoch": 0.6572870407751312,
      "grad_norm": 84.61499786376953,
      "learning_rate": 5.553429582649842e-06,
      "loss": 5.0415,
      "step": 16281
    },
    {
      "epoch": 0.657327412192168,
      "grad_norm": 104.5011978149414,
      "learning_rate": 5.5522584197213504e-06,
      "loss": 5.5753,
      "step": 16282
    },
    {
      "epoch": 0.6573677836092047,
      "grad_norm": 105.82567596435547,
      "learning_rate": 5.5510873328388585e-06,
      "loss": 4.5237,
      "step": 16283
    },
    {
      "epoch": 0.6574081550262414,
      "grad_norm": 78.34632110595703,
      "learning_rate": 5.549916322022381e-06,
      "loss": 6.6062,
      "step": 16284
    },
    {
      "epoch": 0.6574485264432781,
      "grad_norm": 102.30855560302734,
      "learning_rate": 5.54874538729194e-06,
      "loss": 5.4407,
      "step": 16285
    },
    {
      "epoch": 0.657488897860315,
      "grad_norm": 89.53559875488281,
      "learning_rate": 5.547574528667561e-06,
      "loss": 4.8103,
      "step": 16286
    },
    {
      "epoch": 0.6575292692773517,
      "grad_norm": 98.90814971923828,
      "learning_rate": 5.54640374616926e-06,
      "loss": 5.7197,
      "step": 16287
    },
    {
      "epoch": 0.6575696406943884,
      "grad_norm": 79.3031234741211,
      "learning_rate": 5.545233039817055e-06,
      "loss": 4.756,
      "step": 16288
    },
    {
      "epoch": 0.6576100121114251,
      "grad_norm": 119.94554138183594,
      "learning_rate": 5.544062409630961e-06,
      "loss": 4.9522,
      "step": 16289
    },
    {
      "epoch": 0.6576503835284618,
      "grad_norm": 76.00064086914062,
      "learning_rate": 5.542891855630996e-06,
      "loss": 4.4071,
      "step": 16290
    },
    {
      "epoch": 0.6576907549454986,
      "grad_norm": 106.7916030883789,
      "learning_rate": 5.541721377837176e-06,
      "loss": 4.0493,
      "step": 16291
    },
    {
      "epoch": 0.6577311263625353,
      "grad_norm": 101.63836669921875,
      "learning_rate": 5.5405509762695055e-06,
      "loss": 5.005,
      "step": 16292
    },
    {
      "epoch": 0.6577714977795721,
      "grad_norm": 91.3772201538086,
      "learning_rate": 5.539380650947998e-06,
      "loss": 5.3319,
      "step": 16293
    },
    {
      "epoch": 0.6578118691966088,
      "grad_norm": 85.07942962646484,
      "learning_rate": 5.538210401892667e-06,
      "loss": 5.9815,
      "step": 16294
    },
    {
      "epoch": 0.6578522406136456,
      "grad_norm": 101.1849365234375,
      "learning_rate": 5.537040229123523e-06,
      "loss": 5.4414,
      "step": 16295
    },
    {
      "epoch": 0.6578926120306823,
      "grad_norm": 102.9888687133789,
      "learning_rate": 5.535870132660566e-06,
      "loss": 4.5483,
      "step": 16296
    },
    {
      "epoch": 0.657932983447719,
      "grad_norm": 82.46147918701172,
      "learning_rate": 5.534700112523803e-06,
      "loss": 4.5712,
      "step": 16297
    },
    {
      "epoch": 0.6579733548647557,
      "grad_norm": 79.06491088867188,
      "learning_rate": 5.533530168733242e-06,
      "loss": 5.0432,
      "step": 16298
    },
    {
      "epoch": 0.6580137262817924,
      "grad_norm": 91.047119140625,
      "learning_rate": 5.53236030130889e-06,
      "loss": 4.5771,
      "step": 16299
    },
    {
      "epoch": 0.6580540976988293,
      "grad_norm": 67.04924011230469,
      "learning_rate": 5.531190510270739e-06,
      "loss": 5.4426,
      "step": 16300
    },
    {
      "epoch": 0.658094469115866,
      "grad_norm": 93.02203369140625,
      "learning_rate": 5.5300207956387955e-06,
      "loss": 4.8305,
      "step": 16301
    },
    {
      "epoch": 0.6581348405329027,
      "grad_norm": 61.384315490722656,
      "learning_rate": 5.528851157433057e-06,
      "loss": 4.8076,
      "step": 16302
    },
    {
      "epoch": 0.6581752119499394,
      "grad_norm": 114.84459686279297,
      "learning_rate": 5.527681595673528e-06,
      "loss": 5.1405,
      "step": 16303
    },
    {
      "epoch": 0.6582155833669762,
      "grad_norm": 104.95144653320312,
      "learning_rate": 5.5265121103801954e-06,
      "loss": 6.0807,
      "step": 16304
    },
    {
      "epoch": 0.6582559547840129,
      "grad_norm": 118.39945983886719,
      "learning_rate": 5.525342701573061e-06,
      "loss": 5.4292,
      "step": 16305
    },
    {
      "epoch": 0.6582963262010496,
      "grad_norm": 122.55701446533203,
      "learning_rate": 5.524173369272117e-06,
      "loss": 4.3255,
      "step": 16306
    },
    {
      "epoch": 0.6583366976180864,
      "grad_norm": 100.31637573242188,
      "learning_rate": 5.523004113497354e-06,
      "loss": 6.0014,
      "step": 16307
    },
    {
      "epoch": 0.6583770690351232,
      "grad_norm": 87.44943237304688,
      "learning_rate": 5.521834934268768e-06,
      "loss": 5.3336,
      "step": 16308
    },
    {
      "epoch": 0.6584174404521599,
      "grad_norm": 248.93411254882812,
      "learning_rate": 5.5206658316063474e-06,
      "loss": 3.8756,
      "step": 16309
    },
    {
      "epoch": 0.6584578118691966,
      "grad_norm": 89.67155456542969,
      "learning_rate": 5.519496805530081e-06,
      "loss": 4.9832,
      "step": 16310
    },
    {
      "epoch": 0.6584981832862333,
      "grad_norm": 85.6822509765625,
      "learning_rate": 5.518327856059961e-06,
      "loss": 4.3655,
      "step": 16311
    },
    {
      "epoch": 0.65853855470327,
      "grad_norm": 111.21859741210938,
      "learning_rate": 5.517158983215964e-06,
      "loss": 4.1403,
      "step": 16312
    },
    {
      "epoch": 0.6585789261203068,
      "grad_norm": 81.4065933227539,
      "learning_rate": 5.515990187018081e-06,
      "loss": 4.2659,
      "step": 16313
    },
    {
      "epoch": 0.6586192975373436,
      "grad_norm": 81.33578491210938,
      "learning_rate": 5.514821467486295e-06,
      "loss": 5.7714,
      "step": 16314
    },
    {
      "epoch": 0.6586596689543803,
      "grad_norm": 118.31026458740234,
      "learning_rate": 5.513652824640592e-06,
      "loss": 5.1631,
      "step": 16315
    },
    {
      "epoch": 0.658700040371417,
      "grad_norm": 154.6815185546875,
      "learning_rate": 5.512484258500945e-06,
      "loss": 4.9509,
      "step": 16316
    },
    {
      "epoch": 0.6587404117884538,
      "grad_norm": 90.64147186279297,
      "learning_rate": 5.511315769087338e-06,
      "loss": 5.7643,
      "step": 16317
    },
    {
      "epoch": 0.6587807832054905,
      "grad_norm": 112.16465759277344,
      "learning_rate": 5.510147356419751e-06,
      "loss": 4.6969,
      "step": 16318
    },
    {
      "epoch": 0.6588211546225272,
      "grad_norm": 86.94160461425781,
      "learning_rate": 5.508979020518161e-06,
      "loss": 3.3839,
      "step": 16319
    },
    {
      "epoch": 0.6588615260395639,
      "grad_norm": 87.2848892211914,
      "learning_rate": 5.507810761402541e-06,
      "loss": 5.0885,
      "step": 16320
    },
    {
      "epoch": 0.6589018974566008,
      "grad_norm": 108.72237396240234,
      "learning_rate": 5.5066425790928665e-06,
      "loss": 4.6678,
      "step": 16321
    },
    {
      "epoch": 0.6589422688736375,
      "grad_norm": 241.5196533203125,
      "learning_rate": 5.50547447360911e-06,
      "loss": 4.2519,
      "step": 16322
    },
    {
      "epoch": 0.6589826402906742,
      "grad_norm": 225.80654907226562,
      "learning_rate": 5.50430644497125e-06,
      "loss": 6.4364,
      "step": 16323
    },
    {
      "epoch": 0.6590230117077109,
      "grad_norm": 398.400390625,
      "learning_rate": 5.503138493199247e-06,
      "loss": 3.854,
      "step": 16324
    },
    {
      "epoch": 0.6590633831247477,
      "grad_norm": 77.49492645263672,
      "learning_rate": 5.5019706183130735e-06,
      "loss": 5.8157,
      "step": 16325
    },
    {
      "epoch": 0.6591037545417844,
      "grad_norm": 117.05693054199219,
      "learning_rate": 5.5008028203327005e-06,
      "loss": 3.5949,
      "step": 16326
    },
    {
      "epoch": 0.6591441259588211,
      "grad_norm": 71.60979461669922,
      "learning_rate": 5.499635099278092e-06,
      "loss": 5.2087,
      "step": 16327
    },
    {
      "epoch": 0.6591844973758579,
      "grad_norm": 116.45230865478516,
      "learning_rate": 5.4984674551692145e-06,
      "loss": 5.3073,
      "step": 16328
    },
    {
      "epoch": 0.6592248687928947,
      "grad_norm": 108.06346893310547,
      "learning_rate": 5.4972998880260305e-06,
      "loss": 5.4234,
      "step": 16329
    },
    {
      "epoch": 0.6592652402099314,
      "grad_norm": 72.84989166259766,
      "learning_rate": 5.4961323978685055e-06,
      "loss": 2.3331,
      "step": 16330
    },
    {
      "epoch": 0.6593056116269681,
      "grad_norm": 93.29823303222656,
      "learning_rate": 5.494964984716603e-06,
      "loss": 4.7837,
      "step": 16331
    },
    {
      "epoch": 0.6593459830440048,
      "grad_norm": 130.1192626953125,
      "learning_rate": 5.493797648590276e-06,
      "loss": 5.4576,
      "step": 16332
    },
    {
      "epoch": 0.6593863544610415,
      "grad_norm": 102.57746887207031,
      "learning_rate": 5.492630389509485e-06,
      "loss": 4.2711,
      "step": 16333
    },
    {
      "epoch": 0.6594267258780783,
      "grad_norm": 110.53817749023438,
      "learning_rate": 5.4914632074941886e-06,
      "loss": 6.4463,
      "step": 16334
    },
    {
      "epoch": 0.6594670972951151,
      "grad_norm": 90.22206115722656,
      "learning_rate": 5.490296102564348e-06,
      "loss": 5.9637,
      "step": 16335
    },
    {
      "epoch": 0.6595074687121518,
      "grad_norm": 68.3949966430664,
      "learning_rate": 5.4891290747399094e-06,
      "loss": 4.428,
      "step": 16336
    },
    {
      "epoch": 0.6595478401291885,
      "grad_norm": 101.49002838134766,
      "learning_rate": 5.487962124040829e-06,
      "loss": 5.599,
      "step": 16337
    },
    {
      "epoch": 0.6595882115462253,
      "grad_norm": 95.41395568847656,
      "learning_rate": 5.48679525048706e-06,
      "loss": 3.9976,
      "step": 16338
    },
    {
      "epoch": 0.659628582963262,
      "grad_norm": 103.0191421508789,
      "learning_rate": 5.485628454098558e-06,
      "loss": 6.2065,
      "step": 16339
    },
    {
      "epoch": 0.6596689543802987,
      "grad_norm": 73.5465316772461,
      "learning_rate": 5.484461734895263e-06,
      "loss": 6.4457,
      "step": 16340
    },
    {
      "epoch": 0.6597093257973354,
      "grad_norm": 162.7572784423828,
      "learning_rate": 5.483295092897128e-06,
      "loss": 2.9499,
      "step": 16341
    },
    {
      "epoch": 0.6597496972143723,
      "grad_norm": 92.24748229980469,
      "learning_rate": 5.4821285281241e-06,
      "loss": 4.815,
      "step": 16342
    },
    {
      "epoch": 0.659790068631409,
      "grad_norm": 82.11738586425781,
      "learning_rate": 5.480962040596128e-06,
      "loss": 5.1452,
      "step": 16343
    },
    {
      "epoch": 0.6598304400484457,
      "grad_norm": 78.45171356201172,
      "learning_rate": 5.479795630333149e-06,
      "loss": 6.0539,
      "step": 16344
    },
    {
      "epoch": 0.6598708114654824,
      "grad_norm": 205.8335723876953,
      "learning_rate": 5.478629297355108e-06,
      "loss": 5.5493,
      "step": 16345
    },
    {
      "epoch": 0.6599111828825192,
      "grad_norm": 179.9967498779297,
      "learning_rate": 5.4774630416819485e-06,
      "loss": 4.6944,
      "step": 16346
    },
    {
      "epoch": 0.6599515542995559,
      "grad_norm": 79.86595153808594,
      "learning_rate": 5.4762968633336096e-06,
      "loss": 4.0832,
      "step": 16347
    },
    {
      "epoch": 0.6599919257165926,
      "grad_norm": 83.17390441894531,
      "learning_rate": 5.475130762330032e-06,
      "loss": 4.5709,
      "step": 16348
    },
    {
      "epoch": 0.6600322971336294,
      "grad_norm": 86.7496109008789,
      "learning_rate": 5.473964738691151e-06,
      "loss": 4.3838,
      "step": 16349
    },
    {
      "epoch": 0.6600726685506662,
      "grad_norm": 392.01129150390625,
      "learning_rate": 5.472798792436904e-06,
      "loss": 5.4579,
      "step": 16350
    },
    {
      "epoch": 0.6601130399677029,
      "grad_norm": 87.64741516113281,
      "learning_rate": 5.471632923587229e-06,
      "loss": 3.8799,
      "step": 16351
    },
    {
      "epoch": 0.6601534113847396,
      "grad_norm": 137.2361602783203,
      "learning_rate": 5.470467132162054e-06,
      "loss": 5.7817,
      "step": 16352
    },
    {
      "epoch": 0.6601937828017763,
      "grad_norm": 87.37609100341797,
      "learning_rate": 5.4693014181813135e-06,
      "loss": 4.7419,
      "step": 16353
    },
    {
      "epoch": 0.660234154218813,
      "grad_norm": 102.21085357666016,
      "learning_rate": 5.468135781664938e-06,
      "loss": 5.5869,
      "step": 16354
    },
    {
      "epoch": 0.6602745256358499,
      "grad_norm": 91.66319274902344,
      "learning_rate": 5.466970222632863e-06,
      "loss": 4.4095,
      "step": 16355
    },
    {
      "epoch": 0.6603148970528866,
      "grad_norm": 81.600341796875,
      "learning_rate": 5.465804741105007e-06,
      "loss": 6.1258,
      "step": 16356
    },
    {
      "epoch": 0.6603552684699233,
      "grad_norm": 75.10853576660156,
      "learning_rate": 5.464639337101303e-06,
      "loss": 3.8395,
      "step": 16357
    },
    {
      "epoch": 0.66039563988696,
      "grad_norm": 88.07373046875,
      "learning_rate": 5.463474010641674e-06,
      "loss": 5.5855,
      "step": 16358
    },
    {
      "epoch": 0.6604360113039968,
      "grad_norm": 93.73104858398438,
      "learning_rate": 5.4623087617460515e-06,
      "loss": 4.1668,
      "step": 16359
    },
    {
      "epoch": 0.6604763827210335,
      "grad_norm": 87.9524154663086,
      "learning_rate": 5.461143590434347e-06,
      "loss": 5.584,
      "step": 16360
    },
    {
      "epoch": 0.6605167541380702,
      "grad_norm": 81.0165786743164,
      "learning_rate": 5.459978496726489e-06,
      "loss": 5.4301,
      "step": 16361
    },
    {
      "epoch": 0.660557125555107,
      "grad_norm": 75.08515930175781,
      "learning_rate": 5.458813480642397e-06,
      "loss": 5.266,
      "step": 16362
    },
    {
      "epoch": 0.6605974969721438,
      "grad_norm": 145.3917999267578,
      "learning_rate": 5.457648542201995e-06,
      "loss": 4.2951,
      "step": 16363
    },
    {
      "epoch": 0.6606378683891805,
      "grad_norm": 111.0876235961914,
      "learning_rate": 5.45648368142519e-06,
      "loss": 4.5884,
      "step": 16364
    },
    {
      "epoch": 0.6606782398062172,
      "grad_norm": 98.84843444824219,
      "learning_rate": 5.455318898331905e-06,
      "loss": 6.3701,
      "step": 16365
    },
    {
      "epoch": 0.6607186112232539,
      "grad_norm": 73.83981323242188,
      "learning_rate": 5.454154192942054e-06,
      "loss": 5.8082,
      "step": 16366
    },
    {
      "epoch": 0.6607589826402906,
      "grad_norm": 74.77002716064453,
      "learning_rate": 5.452989565275552e-06,
      "loss": 3.1132,
      "step": 16367
    },
    {
      "epoch": 0.6607993540573274,
      "grad_norm": 109.21798706054688,
      "learning_rate": 5.45182501535231e-06,
      "loss": 4.1882,
      "step": 16368
    },
    {
      "epoch": 0.6608397254743642,
      "grad_norm": 239.40078735351562,
      "learning_rate": 5.450660543192239e-06,
      "loss": 3.0687,
      "step": 16369
    },
    {
      "epoch": 0.6608800968914009,
      "grad_norm": 134.91041564941406,
      "learning_rate": 5.44949614881525e-06,
      "loss": 4.6459,
      "step": 16370
    },
    {
      "epoch": 0.6609204683084376,
      "grad_norm": 90.57087707519531,
      "learning_rate": 5.448331832241254e-06,
      "loss": 5.5991,
      "step": 16371
    },
    {
      "epoch": 0.6609608397254744,
      "grad_norm": 94.726318359375,
      "learning_rate": 5.447167593490151e-06,
      "loss": 4.6225,
      "step": 16372
    },
    {
      "epoch": 0.6610012111425111,
      "grad_norm": 75.82965087890625,
      "learning_rate": 5.4460034325818504e-06,
      "loss": 4.0656,
      "step": 16373
    },
    {
      "epoch": 0.6610415825595478,
      "grad_norm": 117.5810317993164,
      "learning_rate": 5.444839349536255e-06,
      "loss": 3.9804,
      "step": 16374
    },
    {
      "epoch": 0.6610819539765845,
      "grad_norm": 116.3459701538086,
      "learning_rate": 5.443675344373276e-06,
      "loss": 5.2021,
      "step": 16375
    },
    {
      "epoch": 0.6611223253936214,
      "grad_norm": 123.19605255126953,
      "learning_rate": 5.442511417112804e-06,
      "loss": 6.6299,
      "step": 16376
    },
    {
      "epoch": 0.6611626968106581,
      "grad_norm": 100.68495178222656,
      "learning_rate": 5.441347567774744e-06,
      "loss": 6.2071,
      "step": 16377
    },
    {
      "epoch": 0.6612030682276948,
      "grad_norm": 147.93670654296875,
      "learning_rate": 5.440183796378997e-06,
      "loss": 5.7587,
      "step": 16378
    },
    {
      "epoch": 0.6612434396447315,
      "grad_norm": 132.99964904785156,
      "learning_rate": 5.439020102945462e-06,
      "loss": 5.6136,
      "step": 16379
    },
    {
      "epoch": 0.6612838110617683,
      "grad_norm": 171.72019958496094,
      "learning_rate": 5.437856487494029e-06,
      "loss": 3.7769,
      "step": 16380
    },
    {
      "epoch": 0.661324182478805,
      "grad_norm": 96.03404235839844,
      "learning_rate": 5.4366929500445955e-06,
      "loss": 5.2935,
      "step": 16381
    },
    {
      "epoch": 0.6613645538958417,
      "grad_norm": 99.76327514648438,
      "learning_rate": 5.4355294906170575e-06,
      "loss": 6.3867,
      "step": 16382
    },
    {
      "epoch": 0.6614049253128785,
      "grad_norm": 92.61405181884766,
      "learning_rate": 5.434366109231306e-06,
      "loss": 4.2565,
      "step": 16383
    },
    {
      "epoch": 0.6614452967299153,
      "grad_norm": 94.81967163085938,
      "learning_rate": 5.433202805907231e-06,
      "loss": 3.8057,
      "step": 16384
    },
    {
      "epoch": 0.661485668146952,
      "grad_norm": 111.1231918334961,
      "learning_rate": 5.4320395806647276e-06,
      "loss": 5.3491,
      "step": 16385
    },
    {
      "epoch": 0.6615260395639887,
      "grad_norm": 119.0278549194336,
      "learning_rate": 5.430876433523682e-06,
      "loss": 5.2633,
      "step": 16386
    },
    {
      "epoch": 0.6615664109810254,
      "grad_norm": 76.35501098632812,
      "learning_rate": 5.429713364503975e-06,
      "loss": 4.796,
      "step": 16387
    },
    {
      "epoch": 0.6616067823980621,
      "grad_norm": 120.09774017333984,
      "learning_rate": 5.4285503736254985e-06,
      "loss": 3.9189,
      "step": 16388
    },
    {
      "epoch": 0.6616471538150989,
      "grad_norm": 87.53458404541016,
      "learning_rate": 5.427387460908136e-06,
      "loss": 2.9886,
      "step": 16389
    },
    {
      "epoch": 0.6616875252321357,
      "grad_norm": 78.37472534179688,
      "learning_rate": 5.426224626371768e-06,
      "loss": 4.755,
      "step": 16390
    },
    {
      "epoch": 0.6617278966491724,
      "grad_norm": 78.30398559570312,
      "learning_rate": 5.425061870036283e-06,
      "loss": 6.521,
      "step": 16391
    },
    {
      "epoch": 0.6617682680662091,
      "grad_norm": 83.43441772460938,
      "learning_rate": 5.423899191921553e-06,
      "loss": 7.0647,
      "step": 16392
    },
    {
      "epoch": 0.6618086394832459,
      "grad_norm": 87.3736572265625,
      "learning_rate": 5.422736592047461e-06,
      "loss": 3.8401,
      "step": 16393
    },
    {
      "epoch": 0.6618490109002826,
      "grad_norm": 138.65151977539062,
      "learning_rate": 5.421574070433886e-06,
      "loss": 5.8438,
      "step": 16394
    },
    {
      "epoch": 0.6618893823173193,
      "grad_norm": 90.76464080810547,
      "learning_rate": 5.420411627100706e-06,
      "loss": 4.415,
      "step": 16395
    },
    {
      "epoch": 0.661929753734356,
      "grad_norm": 127.16507720947266,
      "learning_rate": 5.419249262067788e-06,
      "loss": 4.8621,
      "step": 16396
    },
    {
      "epoch": 0.6619701251513929,
      "grad_norm": 142.6042938232422,
      "learning_rate": 5.418086975355012e-06,
      "loss": 5.2554,
      "step": 16397
    },
    {
      "epoch": 0.6620104965684296,
      "grad_norm": 86.82295227050781,
      "learning_rate": 5.416924766982249e-06,
      "loss": 3.0947,
      "step": 16398
    },
    {
      "epoch": 0.6620508679854663,
      "grad_norm": 108.8096923828125,
      "learning_rate": 5.415762636969373e-06,
      "loss": 4.8381,
      "step": 16399
    },
    {
      "epoch": 0.662091239402503,
      "grad_norm": 145.56793212890625,
      "learning_rate": 5.41460058533625e-06,
      "loss": 5.3724,
      "step": 16400
    },
    {
      "epoch": 0.6621316108195398,
      "grad_norm": 79.0333480834961,
      "learning_rate": 5.413438612102746e-06,
      "loss": 4.4245,
      "step": 16401
    },
    {
      "epoch": 0.6621719822365765,
      "grad_norm": 141.75064086914062,
      "learning_rate": 5.412276717288732e-06,
      "loss": 3.5411,
      "step": 16402
    },
    {
      "epoch": 0.6622123536536132,
      "grad_norm": 351.5192565917969,
      "learning_rate": 5.411114900914074e-06,
      "loss": 5.0025,
      "step": 16403
    },
    {
      "epoch": 0.66225272507065,
      "grad_norm": 67.52816772460938,
      "learning_rate": 5.409953162998636e-06,
      "loss": 4.7112,
      "step": 16404
    },
    {
      "epoch": 0.6622930964876867,
      "grad_norm": 92.97975158691406,
      "learning_rate": 5.408791503562278e-06,
      "loss": 5.2588,
      "step": 16405
    },
    {
      "epoch": 0.6623334679047235,
      "grad_norm": 87.50171661376953,
      "learning_rate": 5.407629922624866e-06,
      "loss": 4.1763,
      "step": 16406
    },
    {
      "epoch": 0.6623738393217602,
      "grad_norm": 97.31946563720703,
      "learning_rate": 5.406468420206261e-06,
      "loss": 5.4415,
      "step": 16407
    },
    {
      "epoch": 0.6624142107387969,
      "grad_norm": 75.70097351074219,
      "learning_rate": 5.405306996326317e-06,
      "loss": 5.7322,
      "step": 16408
    },
    {
      "epoch": 0.6624545821558336,
      "grad_norm": 79.63190460205078,
      "learning_rate": 5.404145651004893e-06,
      "loss": 4.8821,
      "step": 16409
    },
    {
      "epoch": 0.6624949535728704,
      "grad_norm": 105.75358581542969,
      "learning_rate": 5.4029843842618466e-06,
      "loss": 5.3277,
      "step": 16410
    },
    {
      "epoch": 0.6625353249899072,
      "grad_norm": 131.02017211914062,
      "learning_rate": 5.401823196117037e-06,
      "loss": 5.6404,
      "step": 16411
    },
    {
      "epoch": 0.6625756964069439,
      "grad_norm": 85.475830078125,
      "learning_rate": 5.400662086590307e-06,
      "loss": 4.7946,
      "step": 16412
    },
    {
      "epoch": 0.6626160678239806,
      "grad_norm": 125.28673553466797,
      "learning_rate": 5.399501055701517e-06,
      "loss": 2.9318,
      "step": 16413
    },
    {
      "epoch": 0.6626564392410174,
      "grad_norm": 100.9976806640625,
      "learning_rate": 5.398340103470517e-06,
      "loss": 4.2711,
      "step": 16414
    },
    {
      "epoch": 0.6626968106580541,
      "grad_norm": 86.02886199951172,
      "learning_rate": 5.397179229917158e-06,
      "loss": 3.43,
      "step": 16415
    },
    {
      "epoch": 0.6627371820750908,
      "grad_norm": 86.64727020263672,
      "learning_rate": 5.396018435061281e-06,
      "loss": 4.5167,
      "step": 16416
    },
    {
      "epoch": 0.6627775534921275,
      "grad_norm": 161.75946044921875,
      "learning_rate": 5.394857718922739e-06,
      "loss": 5.2179,
      "step": 16417
    },
    {
      "epoch": 0.6628179249091644,
      "grad_norm": 141.10598754882812,
      "learning_rate": 5.393697081521376e-06,
      "loss": 5.4081,
      "step": 16418
    },
    {
      "epoch": 0.6628582963262011,
      "grad_norm": 77.13404083251953,
      "learning_rate": 5.39253652287704e-06,
      "loss": 6.506,
      "step": 16419
    },
    {
      "epoch": 0.6628986677432378,
      "grad_norm": 78.6281509399414,
      "learning_rate": 5.391376043009567e-06,
      "loss": 4.6831,
      "step": 16420
    },
    {
      "epoch": 0.6629390391602745,
      "grad_norm": 134.4427490234375,
      "learning_rate": 5.3902156419388005e-06,
      "loss": 4.0913,
      "step": 16421
    },
    {
      "epoch": 0.6629794105773112,
      "grad_norm": 254.16612243652344,
      "learning_rate": 5.389055319684583e-06,
      "loss": 4.6844,
      "step": 16422
    },
    {
      "epoch": 0.663019781994348,
      "grad_norm": 93.40235900878906,
      "learning_rate": 5.387895076266751e-06,
      "loss": 4.6021,
      "step": 16423
    },
    {
      "epoch": 0.6630601534113847,
      "grad_norm": 96.10953521728516,
      "learning_rate": 5.3867349117051445e-06,
      "loss": 4.6706,
      "step": 16424
    },
    {
      "epoch": 0.6631005248284215,
      "grad_norm": 77.76949310302734,
      "learning_rate": 5.385574826019598e-06,
      "loss": 5.6661,
      "step": 16425
    },
    {
      "epoch": 0.6631408962454582,
      "grad_norm": 110.17676544189453,
      "learning_rate": 5.384414819229946e-06,
      "loss": 5.4614,
      "step": 16426
    },
    {
      "epoch": 0.663181267662495,
      "grad_norm": 81.208984375,
      "learning_rate": 5.383254891356027e-06,
      "loss": 4.1607,
      "step": 16427
    },
    {
      "epoch": 0.6632216390795317,
      "grad_norm": 90.94585418701172,
      "learning_rate": 5.382095042417665e-06,
      "loss": 4.2915,
      "step": 16428
    },
    {
      "epoch": 0.6632620104965684,
      "grad_norm": 79.14633178710938,
      "learning_rate": 5.380935272434693e-06,
      "loss": 5.9822,
      "step": 16429
    },
    {
      "epoch": 0.6633023819136051,
      "grad_norm": 100.17823791503906,
      "learning_rate": 5.379775581426943e-06,
      "loss": 5.3362,
      "step": 16430
    },
    {
      "epoch": 0.6633427533306419,
      "grad_norm": 72.55207824707031,
      "learning_rate": 5.3786159694142434e-06,
      "loss": 4.54,
      "step": 16431
    },
    {
      "epoch": 0.6633831247476787,
      "grad_norm": 85.61563873291016,
      "learning_rate": 5.377456436416417e-06,
      "loss": 4.8365,
      "step": 16432
    },
    {
      "epoch": 0.6634234961647154,
      "grad_norm": 105.71306610107422,
      "learning_rate": 5.3762969824532905e-06,
      "loss": 3.9808,
      "step": 16433
    },
    {
      "epoch": 0.6634638675817521,
      "grad_norm": 103.08934783935547,
      "learning_rate": 5.375137607544687e-06,
      "loss": 3.3891,
      "step": 16434
    },
    {
      "epoch": 0.6635042389987889,
      "grad_norm": 207.79173278808594,
      "learning_rate": 5.373978311710437e-06,
      "loss": 5.1182,
      "step": 16435
    },
    {
      "epoch": 0.6635446104158256,
      "grad_norm": 134.9118194580078,
      "learning_rate": 5.3728190949703496e-06,
      "loss": 6.5887,
      "step": 16436
    },
    {
      "epoch": 0.6635849818328623,
      "grad_norm": 111.37039947509766,
      "learning_rate": 5.37165995734425e-06,
      "loss": 4.2405,
      "step": 16437
    },
    {
      "epoch": 0.663625353249899,
      "grad_norm": 90.8591079711914,
      "learning_rate": 5.37050089885196e-06,
      "loss": 6.0725,
      "step": 16438
    },
    {
      "epoch": 0.6636657246669359,
      "grad_norm": 74.8650131225586,
      "learning_rate": 5.369341919513296e-06,
      "loss": 3.3778,
      "step": 16439
    },
    {
      "epoch": 0.6637060960839726,
      "grad_norm": 76.48722076416016,
      "learning_rate": 5.368183019348068e-06,
      "loss": 5.0484,
      "step": 16440
    },
    {
      "epoch": 0.6637464675010093,
      "grad_norm": 120.49018096923828,
      "learning_rate": 5.367024198376094e-06,
      "loss": 5.6013,
      "step": 16441
    },
    {
      "epoch": 0.663786838918046,
      "grad_norm": 87.1622314453125,
      "learning_rate": 5.365865456617188e-06,
      "loss": 5.5402,
      "step": 16442
    },
    {
      "epoch": 0.6638272103350827,
      "grad_norm": 69.0114974975586,
      "learning_rate": 5.3647067940911614e-06,
      "loss": 4.2681,
      "step": 16443
    },
    {
      "epoch": 0.6638675817521195,
      "grad_norm": 108.33523559570312,
      "learning_rate": 5.363548210817824e-06,
      "loss": 3.5902,
      "step": 16444
    },
    {
      "epoch": 0.6639079531691562,
      "grad_norm": 77.64393615722656,
      "learning_rate": 5.362389706816984e-06,
      "loss": 4.8553,
      "step": 16445
    },
    {
      "epoch": 0.663948324586193,
      "grad_norm": 83.45675659179688,
      "learning_rate": 5.361231282108453e-06,
      "loss": 3.8321,
      "step": 16446
    },
    {
      "epoch": 0.6639886960032297,
      "grad_norm": 149.5196990966797,
      "learning_rate": 5.360072936712036e-06,
      "loss": 5.7425,
      "step": 16447
    },
    {
      "epoch": 0.6640290674202665,
      "grad_norm": 76.31415557861328,
      "learning_rate": 5.358914670647535e-06,
      "loss": 5.213,
      "step": 16448
    },
    {
      "epoch": 0.6640694388373032,
      "grad_norm": 109.30809020996094,
      "learning_rate": 5.3577564839347555e-06,
      "loss": 3.5455,
      "step": 16449
    },
    {
      "epoch": 0.6641098102543399,
      "grad_norm": 166.78564453125,
      "learning_rate": 5.356598376593498e-06,
      "loss": 4.5991,
      "step": 16450
    },
    {
      "epoch": 0.6641501816713766,
      "grad_norm": 100.47919464111328,
      "learning_rate": 5.35544034864357e-06,
      "loss": 4.4116,
      "step": 16451
    },
    {
      "epoch": 0.6641905530884133,
      "grad_norm": 90.4995346069336,
      "learning_rate": 5.354282400104762e-06,
      "loss": 4.0801,
      "step": 16452
    },
    {
      "epoch": 0.6642309245054502,
      "grad_norm": 105.00138854980469,
      "learning_rate": 5.353124530996876e-06,
      "loss": 4.7854,
      "step": 16453
    },
    {
      "epoch": 0.6642712959224869,
      "grad_norm": 100.87288665771484,
      "learning_rate": 5.351966741339711e-06,
      "loss": 7.2739,
      "step": 16454
    },
    {
      "epoch": 0.6643116673395236,
      "grad_norm": 105.26033782958984,
      "learning_rate": 5.350809031153064e-06,
      "loss": 5.8297,
      "step": 16455
    },
    {
      "epoch": 0.6643520387565603,
      "grad_norm": 102.16160583496094,
      "learning_rate": 5.3496514004567215e-06,
      "loss": 5.7521,
      "step": 16456
    },
    {
      "epoch": 0.6643924101735971,
      "grad_norm": 76.18336486816406,
      "learning_rate": 5.348493849270479e-06,
      "loss": 5.6552,
      "step": 16457
    },
    {
      "epoch": 0.6644327815906338,
      "grad_norm": 127.5245361328125,
      "learning_rate": 5.3473363776141325e-06,
      "loss": 4.2331,
      "step": 16458
    },
    {
      "epoch": 0.6644731530076705,
      "grad_norm": 77.2266616821289,
      "learning_rate": 5.346178985507473e-06,
      "loss": 3.7641,
      "step": 16459
    },
    {
      "epoch": 0.6645135244247073,
      "grad_norm": 103.78749084472656,
      "learning_rate": 5.34502167297028e-06,
      "loss": 5.6398,
      "step": 16460
    },
    {
      "epoch": 0.6645538958417441,
      "grad_norm": 78.56348419189453,
      "learning_rate": 5.343864440022346e-06,
      "loss": 5.5542,
      "step": 16461
    },
    {
      "epoch": 0.6645942672587808,
      "grad_norm": 80.78118896484375,
      "learning_rate": 5.342707286683459e-06,
      "loss": 6.4762,
      "step": 16462
    },
    {
      "epoch": 0.6646346386758175,
      "grad_norm": 120.12943267822266,
      "learning_rate": 5.341550212973401e-06,
      "loss": 4.1584,
      "step": 16463
    },
    {
      "epoch": 0.6646750100928542,
      "grad_norm": 81.2234115600586,
      "learning_rate": 5.340393218911956e-06,
      "loss": 5.698,
      "step": 16464
    },
    {
      "epoch": 0.664715381509891,
      "grad_norm": 94.32040405273438,
      "learning_rate": 5.339236304518905e-06,
      "loss": 2.9161,
      "step": 16465
    },
    {
      "epoch": 0.6647557529269278,
      "grad_norm": 234.01771545410156,
      "learning_rate": 5.3380794698140306e-06,
      "loss": 3.2776,
      "step": 16466
    },
    {
      "epoch": 0.6647961243439645,
      "grad_norm": 64.90586853027344,
      "learning_rate": 5.336922714817115e-06,
      "loss": 3.9104,
      "step": 16467
    },
    {
      "epoch": 0.6648364957610012,
      "grad_norm": 62.409889221191406,
      "learning_rate": 5.335766039547927e-06,
      "loss": 4.3857,
      "step": 16468
    },
    {
      "epoch": 0.664876867178038,
      "grad_norm": 69.04808807373047,
      "learning_rate": 5.334609444026248e-06,
      "loss": 6.4536,
      "step": 16469
    },
    {
      "epoch": 0.6649172385950747,
      "grad_norm": 122.9095230102539,
      "learning_rate": 5.333452928271855e-06,
      "loss": 3.5051,
      "step": 16470
    },
    {
      "epoch": 0.6649576100121114,
      "grad_norm": 79.67312622070312,
      "learning_rate": 5.332296492304523e-06,
      "loss": 4.3735,
      "step": 16471
    },
    {
      "epoch": 0.6649979814291481,
      "grad_norm": 66.18096160888672,
      "learning_rate": 5.3311401361440155e-06,
      "loss": 5.8857,
      "step": 16472
    },
    {
      "epoch": 0.665038352846185,
      "grad_norm": 116.92874908447266,
      "learning_rate": 5.3299838598101106e-06,
      "loss": 4.0502,
      "step": 16473
    },
    {
      "epoch": 0.6650787242632217,
      "grad_norm": 135.80950927734375,
      "learning_rate": 5.328827663322575e-06,
      "loss": 5.6608,
      "step": 16474
    },
    {
      "epoch": 0.6651190956802584,
      "grad_norm": 89.54518127441406,
      "learning_rate": 5.327671546701184e-06,
      "loss": 5.7138,
      "step": 16475
    },
    {
      "epoch": 0.6651594670972951,
      "grad_norm": 193.2248077392578,
      "learning_rate": 5.3265155099656955e-06,
      "loss": 4.8878,
      "step": 16476
    },
    {
      "epoch": 0.6651998385143318,
      "grad_norm": 79.8753662109375,
      "learning_rate": 5.325359553135876e-06,
      "loss": 3.7584,
      "step": 16477
    },
    {
      "epoch": 0.6652402099313686,
      "grad_norm": 79.13995361328125,
      "learning_rate": 5.324203676231493e-06,
      "loss": 3.981,
      "step": 16478
    },
    {
      "epoch": 0.6652805813484053,
      "grad_norm": 79.910888671875,
      "learning_rate": 5.323047879272313e-06,
      "loss": 4.6086,
      "step": 16479
    },
    {
      "epoch": 0.6653209527654421,
      "grad_norm": 102.65412902832031,
      "learning_rate": 5.3218921622780885e-06,
      "loss": 4.2778,
      "step": 16480
    },
    {
      "epoch": 0.6653613241824788,
      "grad_norm": 92.28491973876953,
      "learning_rate": 5.320736525268584e-06,
      "loss": 5.0646,
      "step": 16481
    },
    {
      "epoch": 0.6654016955995156,
      "grad_norm": 92.24959564208984,
      "learning_rate": 5.319580968263558e-06,
      "loss": 5.7796,
      "step": 16482
    },
    {
      "epoch": 0.6654420670165523,
      "grad_norm": 127.0453109741211,
      "learning_rate": 5.318425491282766e-06,
      "loss": 4.9652,
      "step": 16483
    },
    {
      "epoch": 0.665482438433589,
      "grad_norm": 83.03398132324219,
      "learning_rate": 5.317270094345967e-06,
      "loss": 5.9262,
      "step": 16484
    },
    {
      "epoch": 0.6655228098506257,
      "grad_norm": 91.32828521728516,
      "learning_rate": 5.316114777472916e-06,
      "loss": 3.1027,
      "step": 16485
    },
    {
      "epoch": 0.6655631812676625,
      "grad_norm": 67.38568115234375,
      "learning_rate": 5.314959540683364e-06,
      "loss": 5.1661,
      "step": 16486
    },
    {
      "epoch": 0.6656035526846993,
      "grad_norm": 148.87100219726562,
      "learning_rate": 5.313804383997066e-06,
      "loss": 5.4856,
      "step": 16487
    },
    {
      "epoch": 0.665643924101736,
      "grad_norm": 114.753662109375,
      "learning_rate": 5.312649307433768e-06,
      "loss": 4.6299,
      "step": 16488
    },
    {
      "epoch": 0.6656842955187727,
      "grad_norm": 99.56295776367188,
      "learning_rate": 5.31149431101322e-06,
      "loss": 5.4191,
      "step": 16489
    },
    {
      "epoch": 0.6657246669358095,
      "grad_norm": 108.95689392089844,
      "learning_rate": 5.310339394755171e-06,
      "loss": 5.1723,
      "step": 16490
    },
    {
      "epoch": 0.6657650383528462,
      "grad_norm": 87.58670043945312,
      "learning_rate": 5.309184558679371e-06,
      "loss": 5.8071,
      "step": 16491
    },
    {
      "epoch": 0.6658054097698829,
      "grad_norm": 75.10384368896484,
      "learning_rate": 5.308029802805557e-06,
      "loss": 4.5285,
      "step": 16492
    },
    {
      "epoch": 0.6658457811869196,
      "grad_norm": 105.83804321289062,
      "learning_rate": 5.3068751271534765e-06,
      "loss": 6.3838,
      "step": 16493
    },
    {
      "epoch": 0.6658861526039564,
      "grad_norm": 87.8528823852539,
      "learning_rate": 5.305720531742874e-06,
      "loss": 5.1853,
      "step": 16494
    },
    {
      "epoch": 0.6659265240209932,
      "grad_norm": 66.12396240234375,
      "learning_rate": 5.304566016593491e-06,
      "loss": 5.3561,
      "step": 16495
    },
    {
      "epoch": 0.6659668954380299,
      "grad_norm": 82.54061126708984,
      "learning_rate": 5.3034115817250605e-06,
      "loss": 4.4999,
      "step": 16496
    },
    {
      "epoch": 0.6660072668550666,
      "grad_norm": 101.73332977294922,
      "learning_rate": 5.302257227157326e-06,
      "loss": 4.6564,
      "step": 16497
    },
    {
      "epoch": 0.6660476382721033,
      "grad_norm": 59.307376861572266,
      "learning_rate": 5.301102952910021e-06,
      "loss": 4.9557,
      "step": 16498
    },
    {
      "epoch": 0.6660880096891401,
      "grad_norm": 84.2588882446289,
      "learning_rate": 5.299948759002888e-06,
      "loss": 4.4215,
      "step": 16499
    },
    {
      "epoch": 0.6661283811061768,
      "grad_norm": 75.86892700195312,
      "learning_rate": 5.298794645455652e-06,
      "loss": 6.0454,
      "step": 16500
    },
    {
      "epoch": 0.6661687525232136,
      "grad_norm": 81.98973083496094,
      "learning_rate": 5.297640612288048e-06,
      "loss": 4.5622,
      "step": 16501
    },
    {
      "epoch": 0.6662091239402503,
      "grad_norm": 91.42001342773438,
      "learning_rate": 5.29648665951981e-06,
      "loss": 5.128,
      "step": 16502
    },
    {
      "epoch": 0.6662494953572871,
      "grad_norm": 92.89936828613281,
      "learning_rate": 5.2953327871706665e-06,
      "loss": 5.2732,
      "step": 16503
    },
    {
      "epoch": 0.6662898667743238,
      "grad_norm": 83.76992797851562,
      "learning_rate": 5.294178995260345e-06,
      "loss": 5.1068,
      "step": 16504
    },
    {
      "epoch": 0.6663302381913605,
      "grad_norm": 82.86909484863281,
      "learning_rate": 5.293025283808575e-06,
      "loss": 5.8918,
      "step": 16505
    },
    {
      "epoch": 0.6663706096083972,
      "grad_norm": 116.89071655273438,
      "learning_rate": 5.2918716528350815e-06,
      "loss": 5.3356,
      "step": 16506
    },
    {
      "epoch": 0.666410981025434,
      "grad_norm": 102.09809112548828,
      "learning_rate": 5.290718102359591e-06,
      "loss": 6.2371,
      "step": 16507
    },
    {
      "epoch": 0.6664513524424708,
      "grad_norm": 114.84791564941406,
      "learning_rate": 5.28956463240182e-06,
      "loss": 5.0268,
      "step": 16508
    },
    {
      "epoch": 0.6664917238595075,
      "grad_norm": 98.60004425048828,
      "learning_rate": 5.288411242981495e-06,
      "loss": 5.4688,
      "step": 16509
    },
    {
      "epoch": 0.6665320952765442,
      "grad_norm": 88.0105972290039,
      "learning_rate": 5.287257934118335e-06,
      "loss": 5.5627,
      "step": 16510
    },
    {
      "epoch": 0.666572466693581,
      "grad_norm": 79.58831787109375,
      "learning_rate": 5.286104705832065e-06,
      "loss": 4.9779,
      "step": 16511
    },
    {
      "epoch": 0.6666128381106177,
      "grad_norm": 109.09911346435547,
      "learning_rate": 5.2849515581423905e-06,
      "loss": 4.6619,
      "step": 16512
    },
    {
      "epoch": 0.6666532095276544,
      "grad_norm": 124.84293365478516,
      "learning_rate": 5.283798491069035e-06,
      "loss": 5.5257,
      "step": 16513
    },
    {
      "epoch": 0.6666935809446911,
      "grad_norm": 119.74034881591797,
      "learning_rate": 5.282645504631711e-06,
      "loss": 4.9204,
      "step": 16514
    },
    {
      "epoch": 0.6667339523617279,
      "grad_norm": 80.72509002685547,
      "learning_rate": 5.281492598850138e-06,
      "loss": 5.1412,
      "step": 16515
    },
    {
      "epoch": 0.6667743237787647,
      "grad_norm": 77.54069519042969,
      "learning_rate": 5.280339773744019e-06,
      "loss": 4.0269,
      "step": 16516
    },
    {
      "epoch": 0.6668146951958014,
      "grad_norm": 108.68915557861328,
      "learning_rate": 5.279187029333068e-06,
      "loss": 6.9122,
      "step": 16517
    },
    {
      "epoch": 0.6668550666128381,
      "grad_norm": 99.114013671875,
      "learning_rate": 5.278034365636997e-06,
      "loss": 5.9591,
      "step": 16518
    },
    {
      "epoch": 0.6668954380298748,
      "grad_norm": 81.8416519165039,
      "learning_rate": 5.2768817826755135e-06,
      "loss": 5.0742,
      "step": 16519
    },
    {
      "epoch": 0.6669358094469116,
      "grad_norm": 62.579322814941406,
      "learning_rate": 5.275729280468319e-06,
      "loss": 3.7603,
      "step": 16520
    },
    {
      "epoch": 0.6669761808639483,
      "grad_norm": 66.0103530883789,
      "learning_rate": 5.274576859035122e-06,
      "loss": 3.4023,
      "step": 16521
    },
    {
      "epoch": 0.6670165522809851,
      "grad_norm": 83.75697326660156,
      "learning_rate": 5.273424518395627e-06,
      "loss": 4.1029,
      "step": 16522
    },
    {
      "epoch": 0.6670569236980218,
      "grad_norm": 103.94435119628906,
      "learning_rate": 5.272272258569533e-06,
      "loss": 5.0437,
      "step": 16523
    },
    {
      "epoch": 0.6670972951150586,
      "grad_norm": 89.0786361694336,
      "learning_rate": 5.2711200795765464e-06,
      "loss": 5.1673,
      "step": 16524
    },
    {
      "epoch": 0.6671376665320953,
      "grad_norm": 93.94269561767578,
      "learning_rate": 5.269967981436361e-06,
      "loss": 6.5219,
      "step": 16525
    },
    {
      "epoch": 0.667178037949132,
      "grad_norm": 139.58094787597656,
      "learning_rate": 5.268815964168678e-06,
      "loss": 5.5345,
      "step": 16526
    },
    {
      "epoch": 0.6672184093661687,
      "grad_norm": 89.24468231201172,
      "learning_rate": 5.2676640277932e-06,
      "loss": 4.5677,
      "step": 16527
    },
    {
      "epoch": 0.6672587807832054,
      "grad_norm": 81.13768005371094,
      "learning_rate": 5.26651217232961e-06,
      "loss": 4.0152,
      "step": 16528
    },
    {
      "epoch": 0.6672991522002423,
      "grad_norm": 124.68412017822266,
      "learning_rate": 5.26536039779761e-06,
      "loss": 3.6372,
      "step": 16529
    },
    {
      "epoch": 0.667339523617279,
      "grad_norm": 81.02356719970703,
      "learning_rate": 5.26420870421689e-06,
      "loss": 5.927,
      "step": 16530
    },
    {
      "epoch": 0.6673798950343157,
      "grad_norm": 96.46947479248047,
      "learning_rate": 5.263057091607146e-06,
      "loss": 5.764,
      "step": 16531
    },
    {
      "epoch": 0.6674202664513524,
      "grad_norm": 104.98332977294922,
      "learning_rate": 5.2619055599880605e-06,
      "loss": 3.4817,
      "step": 16532
    },
    {
      "epoch": 0.6674606378683892,
      "grad_norm": 87.63444519042969,
      "learning_rate": 5.260754109379327e-06,
      "loss": 3.9199,
      "step": 16533
    },
    {
      "epoch": 0.6675010092854259,
      "grad_norm": 82.35443115234375,
      "learning_rate": 5.25960273980063e-06,
      "loss": 4.9711,
      "step": 16534
    },
    {
      "epoch": 0.6675413807024626,
      "grad_norm": 135.26315307617188,
      "learning_rate": 5.25845145127166e-06,
      "loss": 5.1018,
      "step": 16535
    },
    {
      "epoch": 0.6675817521194994,
      "grad_norm": 122.98065185546875,
      "learning_rate": 5.257300243812096e-06,
      "loss": 5.2749,
      "step": 16536
    },
    {
      "epoch": 0.6676221235365362,
      "grad_norm": 91.91094970703125,
      "learning_rate": 5.25614911744162e-06,
      "loss": 4.6578,
      "step": 16537
    },
    {
      "epoch": 0.6676624949535729,
      "grad_norm": 76.73806762695312,
      "learning_rate": 5.2549980721799186e-06,
      "loss": 5.7762,
      "step": 16538
    },
    {
      "epoch": 0.6677028663706096,
      "grad_norm": 80.6091079711914,
      "learning_rate": 5.253847108046674e-06,
      "loss": 4.9393,
      "step": 16539
    },
    {
      "epoch": 0.6677432377876463,
      "grad_norm": 74.81322479248047,
      "learning_rate": 5.2526962250615555e-06,
      "loss": 5.842,
      "step": 16540
    },
    {
      "epoch": 0.667783609204683,
      "grad_norm": 112.37995910644531,
      "learning_rate": 5.251545423244247e-06,
      "loss": 4.9693,
      "step": 16541
    },
    {
      "epoch": 0.6678239806217198,
      "grad_norm": 166.69485473632812,
      "learning_rate": 5.250394702614421e-06,
      "loss": 4.4466,
      "step": 16542
    },
    {
      "epoch": 0.6678643520387566,
      "grad_norm": 93.11425018310547,
      "learning_rate": 5.249244063191756e-06,
      "loss": 6.332,
      "step": 16543
    },
    {
      "epoch": 0.6679047234557933,
      "grad_norm": 114.7891616821289,
      "learning_rate": 5.2480935049959235e-06,
      "loss": 4.7232,
      "step": 16544
    },
    {
      "epoch": 0.66794509487283,
      "grad_norm": 1088.95654296875,
      "learning_rate": 5.2469430280465965e-06,
      "loss": 4.6804,
      "step": 16545
    },
    {
      "epoch": 0.6679854662898668,
      "grad_norm": 99.52642822265625,
      "learning_rate": 5.245792632363443e-06,
      "loss": 5.5411,
      "step": 16546
    },
    {
      "epoch": 0.6680258377069035,
      "grad_norm": 97.57511901855469,
      "learning_rate": 5.244642317966139e-06,
      "loss": 6.446,
      "step": 16547
    },
    {
      "epoch": 0.6680662091239402,
      "grad_norm": 122.94220733642578,
      "learning_rate": 5.243492084874342e-06,
      "loss": 6.3293,
      "step": 16548
    },
    {
      "epoch": 0.6681065805409769,
      "grad_norm": 108.94145202636719,
      "learning_rate": 5.242341933107724e-06,
      "loss": 4.8423,
      "step": 16549
    },
    {
      "epoch": 0.6681469519580138,
      "grad_norm": 87.1651611328125,
      "learning_rate": 5.241191862685948e-06,
      "loss": 6.1654,
      "step": 16550
    },
    {
      "epoch": 0.6681873233750505,
      "grad_norm": 217.07188415527344,
      "learning_rate": 5.240041873628683e-06,
      "loss": 5.8813,
      "step": 16551
    },
    {
      "epoch": 0.6682276947920872,
      "grad_norm": 137.24449157714844,
      "learning_rate": 5.238891965955581e-06,
      "loss": 4.9786,
      "step": 16552
    },
    {
      "epoch": 0.6682680662091239,
      "grad_norm": 99.50501251220703,
      "learning_rate": 5.2377421396863105e-06,
      "loss": 6.2887,
      "step": 16553
    },
    {
      "epoch": 0.6683084376261607,
      "grad_norm": 103.46690368652344,
      "learning_rate": 5.236592394840529e-06,
      "loss": 4.4943,
      "step": 16554
    },
    {
      "epoch": 0.6683488090431974,
      "grad_norm": 90.61489868164062,
      "learning_rate": 5.2354427314378965e-06,
      "loss": 4.6564,
      "step": 16555
    },
    {
      "epoch": 0.6683891804602341,
      "grad_norm": 93.0645751953125,
      "learning_rate": 5.234293149498064e-06,
      "loss": 4.5395,
      "step": 16556
    },
    {
      "epoch": 0.6684295518772709,
      "grad_norm": 81.550537109375,
      "learning_rate": 5.233143649040691e-06,
      "loss": 4.805,
      "step": 16557
    },
    {
      "epoch": 0.6684699232943077,
      "grad_norm": 94.46495056152344,
      "learning_rate": 5.231994230085429e-06,
      "loss": 5.2577,
      "step": 16558
    },
    {
      "epoch": 0.6685102947113444,
      "grad_norm": 67.4230728149414,
      "learning_rate": 5.230844892651935e-06,
      "loss": 6.011,
      "step": 16559
    },
    {
      "epoch": 0.6685506661283811,
      "grad_norm": 106.16758728027344,
      "learning_rate": 5.229695636759853e-06,
      "loss": 5.8292,
      "step": 16560
    },
    {
      "epoch": 0.6685910375454178,
      "grad_norm": 114.37557220458984,
      "learning_rate": 5.228546462428838e-06,
      "loss": 3.4827,
      "step": 16561
    },
    {
      "epoch": 0.6686314089624545,
      "grad_norm": 108.77495574951172,
      "learning_rate": 5.227397369678534e-06,
      "loss": 5.0502,
      "step": 16562
    },
    {
      "epoch": 0.6686717803794913,
      "grad_norm": 89.03629302978516,
      "learning_rate": 5.226248358528592e-06,
      "loss": 6.0481,
      "step": 16563
    },
    {
      "epoch": 0.6687121517965281,
      "grad_norm": 67.88323974609375,
      "learning_rate": 5.225099428998654e-06,
      "loss": 4.4098,
      "step": 16564
    },
    {
      "epoch": 0.6687525232135648,
      "grad_norm": 106.71315002441406,
      "learning_rate": 5.223950581108366e-06,
      "loss": 5.5318,
      "step": 16565
    },
    {
      "epoch": 0.6687928946306015,
      "grad_norm": 108.57746124267578,
      "learning_rate": 5.22280181487737e-06,
      "loss": 4.1625,
      "step": 16566
    },
    {
      "epoch": 0.6688332660476383,
      "grad_norm": 135.32998657226562,
      "learning_rate": 5.221653130325312e-06,
      "loss": 5.9326,
      "step": 16567
    },
    {
      "epoch": 0.668873637464675,
      "grad_norm": 137.51417541503906,
      "learning_rate": 5.220504527471822e-06,
      "loss": 2.8168,
      "step": 16568
    },
    {
      "epoch": 0.6689140088817117,
      "grad_norm": 94.68984985351562,
      "learning_rate": 5.219356006336544e-06,
      "loss": 5.1079,
      "step": 16569
    },
    {
      "epoch": 0.6689543802987485,
      "grad_norm": 65.06312561035156,
      "learning_rate": 5.2182075669391164e-06,
      "loss": 4.7094,
      "step": 16570
    },
    {
      "epoch": 0.6689947517157853,
      "grad_norm": 84.93265533447266,
      "learning_rate": 5.2170592092991756e-06,
      "loss": 5.3033,
      "step": 16571
    },
    {
      "epoch": 0.669035123132822,
      "grad_norm": 110.00460815429688,
      "learning_rate": 5.21591093343635e-06,
      "loss": 5.3483,
      "step": 16572
    },
    {
      "epoch": 0.6690754945498587,
      "grad_norm": 146.15234375,
      "learning_rate": 5.214762739370276e-06,
      "loss": 3.9742,
      "step": 16573
    },
    {
      "epoch": 0.6691158659668954,
      "grad_norm": 206.8583221435547,
      "learning_rate": 5.213614627120584e-06,
      "loss": 5.4046,
      "step": 16574
    },
    {
      "epoch": 0.6691562373839322,
      "grad_norm": 100.0559310913086,
      "learning_rate": 5.212466596706909e-06,
      "loss": 5.3478,
      "step": 16575
    },
    {
      "epoch": 0.6691966088009689,
      "grad_norm": 78.25248718261719,
      "learning_rate": 5.2113186481488706e-06,
      "loss": 3.3665,
      "step": 16576
    },
    {
      "epoch": 0.6692369802180057,
      "grad_norm": 131.66554260253906,
      "learning_rate": 5.210170781466103e-06,
      "loss": 5.3567,
      "step": 16577
    },
    {
      "epoch": 0.6692773516350424,
      "grad_norm": 136.48040771484375,
      "learning_rate": 5.209022996678229e-06,
      "loss": 4.2068,
      "step": 16578
    },
    {
      "epoch": 0.6693177230520791,
      "grad_norm": 75.76214599609375,
      "learning_rate": 5.2078752938048784e-06,
      "loss": 4.2702,
      "step": 16579
    },
    {
      "epoch": 0.6693580944691159,
      "grad_norm": 92.62419128417969,
      "learning_rate": 5.206727672865666e-06,
      "loss": 5.7737,
      "step": 16580
    },
    {
      "epoch": 0.6693984658861526,
      "grad_norm": 102.9451675415039,
      "learning_rate": 5.205580133880216e-06,
      "loss": 6.545,
      "step": 16581
    },
    {
      "epoch": 0.6694388373031893,
      "grad_norm": 111.42947387695312,
      "learning_rate": 5.20443267686815e-06,
      "loss": 6.3253,
      "step": 16582
    },
    {
      "epoch": 0.669479208720226,
      "grad_norm": 79.08899688720703,
      "learning_rate": 5.2032853018490884e-06,
      "loss": 5.048,
      "step": 16583
    },
    {
      "epoch": 0.6695195801372629,
      "grad_norm": 76.9413833618164,
      "learning_rate": 5.202138008842646e-06,
      "loss": 4.0227,
      "step": 16584
    },
    {
      "epoch": 0.6695599515542996,
      "grad_norm": 83.9953384399414,
      "learning_rate": 5.200990797868439e-06,
      "loss": 5.9095,
      "step": 16585
    },
    {
      "epoch": 0.6696003229713363,
      "grad_norm": 95.48844146728516,
      "learning_rate": 5.199843668946085e-06,
      "loss": 4.4365,
      "step": 16586
    },
    {
      "epoch": 0.669640694388373,
      "grad_norm": 63.07307052612305,
      "learning_rate": 5.198696622095198e-06,
      "loss": 5.0085,
      "step": 16587
    },
    {
      "epoch": 0.6696810658054098,
      "grad_norm": 99.79113006591797,
      "learning_rate": 5.1975496573353825e-06,
      "loss": 4.7039,
      "step": 16588
    },
    {
      "epoch": 0.6697214372224465,
      "grad_norm": 118.44088745117188,
      "learning_rate": 5.196402774686253e-06,
      "loss": 4.4884,
      "step": 16589
    },
    {
      "epoch": 0.6697618086394832,
      "grad_norm": 86.03807067871094,
      "learning_rate": 5.19525597416742e-06,
      "loss": 5.6185,
      "step": 16590
    },
    {
      "epoch": 0.66980218005652,
      "grad_norm": 90.35914611816406,
      "learning_rate": 5.194109255798494e-06,
      "loss": 5.9801,
      "step": 16591
    },
    {
      "epoch": 0.6698425514735568,
      "grad_norm": 118.37285614013672,
      "learning_rate": 5.192962619599071e-06,
      "loss": 6.6852,
      "step": 16592
    },
    {
      "epoch": 0.6698829228905935,
      "grad_norm": 97.77340698242188,
      "learning_rate": 5.191816065588764e-06,
      "loss": 5.3884,
      "step": 16593
    },
    {
      "epoch": 0.6699232943076302,
      "grad_norm": 79.45673370361328,
      "learning_rate": 5.190669593787174e-06,
      "loss": 4.487,
      "step": 16594
    },
    {
      "epoch": 0.6699636657246669,
      "grad_norm": 92.66743469238281,
      "learning_rate": 5.189523204213908e-06,
      "loss": 6.1363,
      "step": 16595
    },
    {
      "epoch": 0.6700040371417036,
      "grad_norm": 97.42760467529297,
      "learning_rate": 5.188376896888556e-06,
      "loss": 5.4468,
      "step": 16596
    },
    {
      "epoch": 0.6700444085587404,
      "grad_norm": 122.79510498046875,
      "learning_rate": 5.187230671830725e-06,
      "loss": 4.8505,
      "step": 16597
    },
    {
      "epoch": 0.6700847799757772,
      "grad_norm": 102.26968383789062,
      "learning_rate": 5.18608452906001e-06,
      "loss": 5.3407,
      "step": 16598
    },
    {
      "epoch": 0.6701251513928139,
      "grad_norm": 86.98381042480469,
      "learning_rate": 5.184938468596009e-06,
      "loss": 4.2053,
      "step": 16599
    },
    {
      "epoch": 0.6701655228098506,
      "grad_norm": 138.83682250976562,
      "learning_rate": 5.183792490458316e-06,
      "loss": 6.307,
      "step": 16600
    },
    {
      "epoch": 0.6702058942268874,
      "grad_norm": 76.30164337158203,
      "learning_rate": 5.182646594666525e-06,
      "loss": 3.4484,
      "step": 16601
    },
    {
      "epoch": 0.6702462656439241,
      "grad_norm": 65.57455444335938,
      "learning_rate": 5.1815007812402275e-06,
      "loss": 5.2857,
      "step": 16602
    },
    {
      "epoch": 0.6702866370609608,
      "grad_norm": 72.91973114013672,
      "learning_rate": 5.180355050199013e-06,
      "loss": 6.4225,
      "step": 16603
    },
    {
      "epoch": 0.6703270084779975,
      "grad_norm": 117.14923095703125,
      "learning_rate": 5.179209401562478e-06,
      "loss": 4.6614,
      "step": 16604
    },
    {
      "epoch": 0.6703673798950344,
      "grad_norm": 67.53864288330078,
      "learning_rate": 5.178063835350201e-06,
      "loss": 6.0284,
      "step": 16605
    },
    {
      "epoch": 0.6704077513120711,
      "grad_norm": 96.33057403564453,
      "learning_rate": 5.176918351581772e-06,
      "loss": 5.041,
      "step": 16606
    },
    {
      "epoch": 0.6704481227291078,
      "grad_norm": 93.58274841308594,
      "learning_rate": 5.175772950276777e-06,
      "loss": 4.199,
      "step": 16607
    },
    {
      "epoch": 0.6704884941461445,
      "grad_norm": 97.44945526123047,
      "learning_rate": 5.174627631454803e-06,
      "loss": 5.8339,
      "step": 16608
    },
    {
      "epoch": 0.6705288655631813,
      "grad_norm": 140.73452758789062,
      "learning_rate": 5.173482395135425e-06,
      "loss": 5.718,
      "step": 16609
    },
    {
      "epoch": 0.670569236980218,
      "grad_norm": 92.65853881835938,
      "learning_rate": 5.172337241338226e-06,
      "loss": 4.7515,
      "step": 16610
    },
    {
      "epoch": 0.6706096083972547,
      "grad_norm": 100.3568115234375,
      "learning_rate": 5.171192170082787e-06,
      "loss": 4.1541,
      "step": 16611
    },
    {
      "epoch": 0.6706499798142915,
      "grad_norm": 102.60905456542969,
      "learning_rate": 5.170047181388691e-06,
      "loss": 2.7519,
      "step": 16612
    },
    {
      "epoch": 0.6706903512313283,
      "grad_norm": 190.12258911132812,
      "learning_rate": 5.168902275275505e-06,
      "loss": 5.9209,
      "step": 16613
    },
    {
      "epoch": 0.670730722648365,
      "grad_norm": 418.2333984375,
      "learning_rate": 5.167757451762807e-06,
      "loss": 6.4842,
      "step": 16614
    },
    {
      "epoch": 0.6707710940654017,
      "grad_norm": 122.5167007446289,
      "learning_rate": 5.166612710870175e-06,
      "loss": 5.8296,
      "step": 16615
    },
    {
      "epoch": 0.6708114654824384,
      "grad_norm": 92.42333221435547,
      "learning_rate": 5.165468052617181e-06,
      "loss": 5.4013,
      "step": 16616
    },
    {
      "epoch": 0.6708518368994751,
      "grad_norm": 66.3647689819336,
      "learning_rate": 5.164323477023391e-06,
      "loss": 4.9948,
      "step": 16617
    },
    {
      "epoch": 0.6708922083165119,
      "grad_norm": 97.8836669921875,
      "learning_rate": 5.163178984108377e-06,
      "loss": 5.9602,
      "step": 16618
    },
    {
      "epoch": 0.6709325797335487,
      "grad_norm": 85.67021179199219,
      "learning_rate": 5.162034573891708e-06,
      "loss": 5.9709,
      "step": 16619
    },
    {
      "epoch": 0.6709729511505854,
      "grad_norm": 80.77433013916016,
      "learning_rate": 5.160890246392949e-06,
      "loss": 5.0533,
      "step": 16620
    },
    {
      "epoch": 0.6710133225676221,
      "grad_norm": 84.00727844238281,
      "learning_rate": 5.159746001631667e-06,
      "loss": 6.156,
      "step": 16621
    },
    {
      "epoch": 0.6710536939846589,
      "grad_norm": 96.65888977050781,
      "learning_rate": 5.1586018396274266e-06,
      "loss": 5.3561,
      "step": 16622
    },
    {
      "epoch": 0.6710940654016956,
      "grad_norm": 84.39686584472656,
      "learning_rate": 5.157457760399789e-06,
      "loss": 5.1128,
      "step": 16623
    },
    {
      "epoch": 0.6711344368187323,
      "grad_norm": 81.36992645263672,
      "learning_rate": 5.156313763968319e-06,
      "loss": 3.5221,
      "step": 16624
    },
    {
      "epoch": 0.671174808235769,
      "grad_norm": 83.23538970947266,
      "learning_rate": 5.15516985035257e-06,
      "loss": 5.2452,
      "step": 16625
    },
    {
      "epoch": 0.6712151796528059,
      "grad_norm": 88.73213195800781,
      "learning_rate": 5.154026019572102e-06,
      "loss": 6.0207,
      "step": 16626
    },
    {
      "epoch": 0.6712555510698426,
      "grad_norm": 82.0159912109375,
      "learning_rate": 5.152882271646473e-06,
      "loss": 3.8214,
      "step": 16627
    },
    {
      "epoch": 0.6712959224868793,
      "grad_norm": 89.92203521728516,
      "learning_rate": 5.151738606595242e-06,
      "loss": 5.5347,
      "step": 16628
    },
    {
      "epoch": 0.671336293903916,
      "grad_norm": 98.73749542236328,
      "learning_rate": 5.150595024437956e-06,
      "loss": 4.5741,
      "step": 16629
    },
    {
      "epoch": 0.6713766653209527,
      "grad_norm": 91.2245864868164,
      "learning_rate": 5.14945152519417e-06,
      "loss": 5.4937,
      "step": 16630
    },
    {
      "epoch": 0.6714170367379895,
      "grad_norm": 128.94070434570312,
      "learning_rate": 5.148308108883437e-06,
      "loss": 4.1907,
      "step": 16631
    },
    {
      "epoch": 0.6714574081550262,
      "grad_norm": 74.74588775634766,
      "learning_rate": 5.14716477552531e-06,
      "loss": 5.2176,
      "step": 16632
    },
    {
      "epoch": 0.671497779572063,
      "grad_norm": 96.92118072509766,
      "learning_rate": 5.146021525139329e-06,
      "loss": 5.6468,
      "step": 16633
    },
    {
      "epoch": 0.6715381509890997,
      "grad_norm": 94.76103973388672,
      "learning_rate": 5.144878357745045e-06,
      "loss": 4.91,
      "step": 16634
    },
    {
      "epoch": 0.6715785224061365,
      "grad_norm": 105.4586181640625,
      "learning_rate": 5.143735273362002e-06,
      "loss": 2.7658,
      "step": 16635
    },
    {
      "epoch": 0.6716188938231732,
      "grad_norm": 123.67118835449219,
      "learning_rate": 5.142592272009751e-06,
      "loss": 4.5269,
      "step": 16636
    },
    {
      "epoch": 0.6716592652402099,
      "grad_norm": 77.52833557128906,
      "learning_rate": 5.141449353707826e-06,
      "loss": 5.6528,
      "step": 16637
    },
    {
      "epoch": 0.6716996366572466,
      "grad_norm": 102.33296966552734,
      "learning_rate": 5.140306518475771e-06,
      "loss": 4.8676,
      "step": 16638
    },
    {
      "epoch": 0.6717400080742834,
      "grad_norm": 95.88081359863281,
      "learning_rate": 5.139163766333125e-06,
      "loss": 5.8498,
      "step": 16639
    },
    {
      "epoch": 0.6717803794913202,
      "grad_norm": 170.7677459716797,
      "learning_rate": 5.138021097299428e-06,
      "loss": 6.2274,
      "step": 16640
    },
    {
      "epoch": 0.6718207509083569,
      "grad_norm": 96.08441162109375,
      "learning_rate": 5.136878511394219e-06,
      "loss": 4.4201,
      "step": 16641
    },
    {
      "epoch": 0.6718611223253936,
      "grad_norm": 68.64300537109375,
      "learning_rate": 5.135736008637029e-06,
      "loss": 4.9387,
      "step": 16642
    },
    {
      "epoch": 0.6719014937424304,
      "grad_norm": 128.302001953125,
      "learning_rate": 5.134593589047396e-06,
      "loss": 3.0091,
      "step": 16643
    },
    {
      "epoch": 0.6719418651594671,
      "grad_norm": 100.61744689941406,
      "learning_rate": 5.1334512526448545e-06,
      "loss": 2.8558,
      "step": 16644
    },
    {
      "epoch": 0.6719822365765038,
      "grad_norm": 97.28057098388672,
      "learning_rate": 5.1323089994489276e-06,
      "loss": 3.1511,
      "step": 16645
    },
    {
      "epoch": 0.6720226079935405,
      "grad_norm": 104.79019927978516,
      "learning_rate": 5.13116682947915e-06,
      "loss": 5.7919,
      "step": 16646
    },
    {
      "epoch": 0.6720629794105774,
      "grad_norm": 91.30011749267578,
      "learning_rate": 5.130024742755051e-06,
      "loss": 4.8987,
      "step": 16647
    },
    {
      "epoch": 0.6721033508276141,
      "grad_norm": 106.45574188232422,
      "learning_rate": 5.1288827392961595e-06,
      "loss": 6.3343,
      "step": 16648
    },
    {
      "epoch": 0.6721437222446508,
      "grad_norm": 84.2997055053711,
      "learning_rate": 5.127740819121996e-06,
      "loss": 4.7596,
      "step": 16649
    },
    {
      "epoch": 0.6721840936616875,
      "grad_norm": 91.25301361083984,
      "learning_rate": 5.126598982252085e-06,
      "loss": 5.1459,
      "step": 16650
    },
    {
      "epoch": 0.6722244650787242,
      "grad_norm": 109.1593246459961,
      "learning_rate": 5.1254572287059526e-06,
      "loss": 4.6743,
      "step": 16651
    },
    {
      "epoch": 0.672264836495761,
      "grad_norm": 133.4529266357422,
      "learning_rate": 5.124315558503122e-06,
      "loss": 3.8591,
      "step": 16652
    },
    {
      "epoch": 0.6723052079127977,
      "grad_norm": 88.6049575805664,
      "learning_rate": 5.123173971663106e-06,
      "loss": 4.4708,
      "step": 16653
    },
    {
      "epoch": 0.6723455793298345,
      "grad_norm": 114.6514663696289,
      "learning_rate": 5.1220324682054265e-06,
      "loss": 5.0792,
      "step": 16654
    },
    {
      "epoch": 0.6723859507468712,
      "grad_norm": 132.8684844970703,
      "learning_rate": 5.120891048149601e-06,
      "loss": 3.7926,
      "step": 16655
    },
    {
      "epoch": 0.672426322163908,
      "grad_norm": 86.9875717163086,
      "learning_rate": 5.119749711515149e-06,
      "loss": 4.0258,
      "step": 16656
    },
    {
      "epoch": 0.6724666935809447,
      "grad_norm": 76.2268295288086,
      "learning_rate": 5.118608458321579e-06,
      "loss": 5.6647,
      "step": 16657
    },
    {
      "epoch": 0.6725070649979814,
      "grad_norm": 112.70470428466797,
      "learning_rate": 5.117467288588402e-06,
      "loss": 4.1861,
      "step": 16658
    },
    {
      "epoch": 0.6725474364150181,
      "grad_norm": 146.6857147216797,
      "learning_rate": 5.116326202335136e-06,
      "loss": 5.2041,
      "step": 16659
    },
    {
      "epoch": 0.6725878078320549,
      "grad_norm": 514.8234252929688,
      "learning_rate": 5.115185199581285e-06,
      "loss": 5.5657,
      "step": 16660
    },
    {
      "epoch": 0.6726281792490917,
      "grad_norm": 101.91090393066406,
      "learning_rate": 5.114044280346362e-06,
      "loss": 5.6236,
      "step": 16661
    },
    {
      "epoch": 0.6726685506661284,
      "grad_norm": 81.82518005371094,
      "learning_rate": 5.112903444649872e-06,
      "loss": 4.3604,
      "step": 16662
    },
    {
      "epoch": 0.6727089220831651,
      "grad_norm": 85.87344360351562,
      "learning_rate": 5.11176269251132e-06,
      "loss": 3.7104,
      "step": 16663
    },
    {
      "epoch": 0.6727492935002019,
      "grad_norm": 126.82139587402344,
      "learning_rate": 5.110622023950216e-06,
      "loss": 4.6455,
      "step": 16664
    },
    {
      "epoch": 0.6727896649172386,
      "grad_norm": 88.35629272460938,
      "learning_rate": 5.109481438986053e-06,
      "loss": 5.2708,
      "step": 16665
    },
    {
      "epoch": 0.6728300363342753,
      "grad_norm": 86.76167297363281,
      "learning_rate": 5.108340937638339e-06,
      "loss": 4.6634,
      "step": 16666
    },
    {
      "epoch": 0.672870407751312,
      "grad_norm": 89.72985076904297,
      "learning_rate": 5.107200519926569e-06,
      "loss": 4.5759,
      "step": 16667
    },
    {
      "epoch": 0.6729107791683488,
      "grad_norm": 82.34402465820312,
      "learning_rate": 5.106060185870252e-06,
      "loss": 4.6599,
      "step": 16668
    },
    {
      "epoch": 0.6729511505853856,
      "grad_norm": 102.41370391845703,
      "learning_rate": 5.104919935488873e-06,
      "loss": 6.5592,
      "step": 16669
    },
    {
      "epoch": 0.6729915220024223,
      "grad_norm": 82.80134582519531,
      "learning_rate": 5.10377976880193e-06,
      "loss": 6.8277,
      "step": 16670
    },
    {
      "epoch": 0.673031893419459,
      "grad_norm": 116.60069274902344,
      "learning_rate": 5.102639685828921e-06,
      "loss": 3.7084,
      "step": 16671
    },
    {
      "epoch": 0.6730722648364957,
      "grad_norm": 111.66375732421875,
      "learning_rate": 5.101499686589341e-06,
      "loss": 4.7391,
      "step": 16672
    },
    {
      "epoch": 0.6731126362535325,
      "grad_norm": 81.60449981689453,
      "learning_rate": 5.100359771102674e-06,
      "loss": 4.3214,
      "step": 16673
    },
    {
      "epoch": 0.6731530076705693,
      "grad_norm": 98.12540435791016,
      "learning_rate": 5.099219939388413e-06,
      "loss": 5.356,
      "step": 16674
    },
    {
      "epoch": 0.673193379087606,
      "grad_norm": 94.24836730957031,
      "learning_rate": 5.098080191466046e-06,
      "loss": 3.8503,
      "step": 16675
    },
    {
      "epoch": 0.6732337505046427,
      "grad_norm": 73.78854370117188,
      "learning_rate": 5.0969405273550665e-06,
      "loss": 5.0374,
      "step": 16676
    },
    {
      "epoch": 0.6732741219216795,
      "grad_norm": 117.3122787475586,
      "learning_rate": 5.095800947074949e-06,
      "loss": 5.6152,
      "step": 16677
    },
    {
      "epoch": 0.6733144933387162,
      "grad_norm": 139.48468017578125,
      "learning_rate": 5.094661450645184e-06,
      "loss": 5.743,
      "step": 16678
    },
    {
      "epoch": 0.6733548647557529,
      "grad_norm": 76.48577117919922,
      "learning_rate": 5.093522038085255e-06,
      "loss": 4.5076,
      "step": 16679
    },
    {
      "epoch": 0.6733952361727896,
      "grad_norm": 101.27998352050781,
      "learning_rate": 5.092382709414639e-06,
      "loss": 5.832,
      "step": 16680
    },
    {
      "epoch": 0.6734356075898265,
      "grad_norm": 67.49271392822266,
      "learning_rate": 5.09124346465282e-06,
      "loss": 5.649,
      "step": 16681
    },
    {
      "epoch": 0.6734759790068632,
      "grad_norm": 126.82351684570312,
      "learning_rate": 5.090104303819275e-06,
      "loss": 4.4339,
      "step": 16682
    },
    {
      "epoch": 0.6735163504238999,
      "grad_norm": 783.1954345703125,
      "learning_rate": 5.08896522693348e-06,
      "loss": 5.505,
      "step": 16683
    },
    {
      "epoch": 0.6735567218409366,
      "grad_norm": 93.82462310791016,
      "learning_rate": 5.087826234014916e-06,
      "loss": 5.4222,
      "step": 16684
    },
    {
      "epoch": 0.6735970932579733,
      "grad_norm": 207.7335205078125,
      "learning_rate": 5.0866873250830495e-06,
      "loss": 5.6781,
      "step": 16685
    },
    {
      "epoch": 0.6736374646750101,
      "grad_norm": 77.89397430419922,
      "learning_rate": 5.085548500157356e-06,
      "loss": 4.9738,
      "step": 16686
    },
    {
      "epoch": 0.6736778360920468,
      "grad_norm": 206.4005584716797,
      "learning_rate": 5.084409759257307e-06,
      "loss": 5.0147,
      "step": 16687
    },
    {
      "epoch": 0.6737182075090836,
      "grad_norm": 82.25287628173828,
      "learning_rate": 5.083271102402376e-06,
      "loss": 5.869,
      "step": 16688
    },
    {
      "epoch": 0.6737585789261203,
      "grad_norm": 69.0936050415039,
      "learning_rate": 5.082132529612024e-06,
      "loss": 2.3828,
      "step": 16689
    },
    {
      "epoch": 0.6737989503431571,
      "grad_norm": 96.2963638305664,
      "learning_rate": 5.080994040905721e-06,
      "loss": 6.0774,
      "step": 16690
    },
    {
      "epoch": 0.6738393217601938,
      "grad_norm": 73.70341491699219,
      "learning_rate": 5.0798556363029336e-06,
      "loss": 5.3613,
      "step": 16691
    },
    {
      "epoch": 0.6738796931772305,
      "grad_norm": 77.60755157470703,
      "learning_rate": 5.078717315823129e-06,
      "loss": 6.2585,
      "step": 16692
    },
    {
      "epoch": 0.6739200645942672,
      "grad_norm": 99.67152404785156,
      "learning_rate": 5.0775790794857636e-06,
      "loss": 4.855,
      "step": 16693
    },
    {
      "epoch": 0.673960436011304,
      "grad_norm": 81.64730834960938,
      "learning_rate": 5.076440927310301e-06,
      "loss": 5.6148,
      "step": 16694
    },
    {
      "epoch": 0.6740008074283408,
      "grad_norm": 64.00462341308594,
      "learning_rate": 5.0753028593162e-06,
      "loss": 3.6142,
      "step": 16695
    },
    {
      "epoch": 0.6740411788453775,
      "grad_norm": 84.48309326171875,
      "learning_rate": 5.074164875522925e-06,
      "loss": 5.0759,
      "step": 16696
    },
    {
      "epoch": 0.6740815502624142,
      "grad_norm": 121.72993469238281,
      "learning_rate": 5.073026975949924e-06,
      "loss": 4.87,
      "step": 16697
    },
    {
      "epoch": 0.674121921679451,
      "grad_norm": 96.15689849853516,
      "learning_rate": 5.071889160616656e-06,
      "loss": 6.975,
      "step": 16698
    },
    {
      "epoch": 0.6741622930964877,
      "grad_norm": 86.62171173095703,
      "learning_rate": 5.070751429542576e-06,
      "loss": 4.3122,
      "step": 16699
    },
    {
      "epoch": 0.6742026645135244,
      "grad_norm": 149.0670166015625,
      "learning_rate": 5.0696137827471355e-06,
      "loss": 5.9533,
      "step": 16700
    },
    {
      "epoch": 0.6742430359305611,
      "grad_norm": 826.8849487304688,
      "learning_rate": 5.068476220249787e-06,
      "loss": 4.1511,
      "step": 16701
    },
    {
      "epoch": 0.674283407347598,
      "grad_norm": 76.3194351196289,
      "learning_rate": 5.067338742069978e-06,
      "loss": 5.1835,
      "step": 16702
    },
    {
      "epoch": 0.6743237787646347,
      "grad_norm": 65.1965103149414,
      "learning_rate": 5.066201348227159e-06,
      "loss": 2.6418,
      "step": 16703
    },
    {
      "epoch": 0.6743641501816714,
      "grad_norm": 129.3074951171875,
      "learning_rate": 5.065064038740779e-06,
      "loss": 5.2396,
      "step": 16704
    },
    {
      "epoch": 0.6744045215987081,
      "grad_norm": 94.64239501953125,
      "learning_rate": 5.063926813630275e-06,
      "loss": 5.4152,
      "step": 16705
    },
    {
      "epoch": 0.6744448930157448,
      "grad_norm": 129.00244140625,
      "learning_rate": 5.062789672915097e-06,
      "loss": 6.3194,
      "step": 16706
    },
    {
      "epoch": 0.6744852644327816,
      "grad_norm": 121.30118560791016,
      "learning_rate": 5.061652616614687e-06,
      "loss": 4.5264,
      "step": 16707
    },
    {
      "epoch": 0.6745256358498183,
      "grad_norm": 76.5483627319336,
      "learning_rate": 5.060515644748488e-06,
      "loss": 5.7287,
      "step": 16708
    },
    {
      "epoch": 0.6745660072668551,
      "grad_norm": 122.86473846435547,
      "learning_rate": 5.059378757335933e-06,
      "loss": 5.3676,
      "step": 16709
    },
    {
      "epoch": 0.6746063786838918,
      "grad_norm": 98.09765625,
      "learning_rate": 5.058241954396465e-06,
      "loss": 4.2397,
      "step": 16710
    },
    {
      "epoch": 0.6746467501009286,
      "grad_norm": 84.41004180908203,
      "learning_rate": 5.0571052359495196e-06,
      "loss": 5.5689,
      "step": 16711
    },
    {
      "epoch": 0.6746871215179653,
      "grad_norm": 80.8344955444336,
      "learning_rate": 5.055968602014536e-06,
      "loss": 3.7133,
      "step": 16712
    },
    {
      "epoch": 0.674727492935002,
      "grad_norm": 64.12679290771484,
      "learning_rate": 5.05483205261094e-06,
      "loss": 4.0161,
      "step": 16713
    },
    {
      "epoch": 0.6747678643520387,
      "grad_norm": 89.92912292480469,
      "learning_rate": 5.05369558775817e-06,
      "loss": 4.4841,
      "step": 16714
    },
    {
      "epoch": 0.6748082357690754,
      "grad_norm": 2645.916015625,
      "learning_rate": 5.052559207475653e-06,
      "loss": 3.6739,
      "step": 16715
    },
    {
      "epoch": 0.6748486071861123,
      "grad_norm": 173.54421997070312,
      "learning_rate": 5.051422911782824e-06,
      "loss": 4.0146,
      "step": 16716
    },
    {
      "epoch": 0.674888978603149,
      "grad_norm": 92.43649291992188,
      "learning_rate": 5.050286700699104e-06,
      "loss": 5.3659,
      "step": 16717
    },
    {
      "epoch": 0.6749293500201857,
      "grad_norm": 152.09178161621094,
      "learning_rate": 5.049150574243924e-06,
      "loss": 3.485,
      "step": 16718
    },
    {
      "epoch": 0.6749697214372224,
      "grad_norm": 93.9373550415039,
      "learning_rate": 5.048014532436707e-06,
      "loss": 4.7961,
      "step": 16719
    },
    {
      "epoch": 0.6750100928542592,
      "grad_norm": 68.2120590209961,
      "learning_rate": 5.046878575296879e-06,
      "loss": 3.6224,
      "step": 16720
    },
    {
      "epoch": 0.6750504642712959,
      "grad_norm": 91.60281372070312,
      "learning_rate": 5.045742702843859e-06,
      "loss": 4.7876,
      "step": 16721
    },
    {
      "epoch": 0.6750908356883326,
      "grad_norm": 87.37191009521484,
      "learning_rate": 5.044606915097071e-06,
      "loss": 6.3429,
      "step": 16722
    },
    {
      "epoch": 0.6751312071053694,
      "grad_norm": 169.07989501953125,
      "learning_rate": 5.0434712120759325e-06,
      "loss": 6.2463,
      "step": 16723
    },
    {
      "epoch": 0.6751715785224062,
      "grad_norm": 118.31253051757812,
      "learning_rate": 5.042335593799867e-06,
      "loss": 4.9135,
      "step": 16724
    },
    {
      "epoch": 0.6752119499394429,
      "grad_norm": 130.4200439453125,
      "learning_rate": 5.041200060288281e-06,
      "loss": 5.2862,
      "step": 16725
    },
    {
      "epoch": 0.6752523213564796,
      "grad_norm": 90.7537612915039,
      "learning_rate": 5.040064611560594e-06,
      "loss": 4.7275,
      "step": 16726
    },
    {
      "epoch": 0.6752926927735163,
      "grad_norm": 91.10688781738281,
      "learning_rate": 5.038929247636221e-06,
      "loss": 6.7766,
      "step": 16727
    },
    {
      "epoch": 0.6753330641905531,
      "grad_norm": 92.07176971435547,
      "learning_rate": 5.0377939685345765e-06,
      "loss": 5.746,
      "step": 16728
    },
    {
      "epoch": 0.6753734356075898,
      "grad_norm": 91.0113296508789,
      "learning_rate": 5.036658774275063e-06,
      "loss": 7.6726,
      "step": 16729
    },
    {
      "epoch": 0.6754138070246266,
      "grad_norm": 149.0604705810547,
      "learning_rate": 5.035523664877096e-06,
      "loss": 4.3048,
      "step": 16730
    },
    {
      "epoch": 0.6754541784416633,
      "grad_norm": 79.93486785888672,
      "learning_rate": 5.0343886403600795e-06,
      "loss": 6.2353,
      "step": 16731
    },
    {
      "epoch": 0.6754945498587,
      "grad_norm": 95.30554962158203,
      "learning_rate": 5.0332537007434255e-06,
      "loss": 5.7044,
      "step": 16732
    },
    {
      "epoch": 0.6755349212757368,
      "grad_norm": 146.35809326171875,
      "learning_rate": 5.032118846046532e-06,
      "loss": 4.9985,
      "step": 16733
    },
    {
      "epoch": 0.6755752926927735,
      "grad_norm": 113.6451644897461,
      "learning_rate": 5.030984076288805e-06,
      "loss": 5.0388,
      "step": 16734
    },
    {
      "epoch": 0.6756156641098102,
      "grad_norm": 74.29456329345703,
      "learning_rate": 5.0298493914896475e-06,
      "loss": 4.0377,
      "step": 16735
    },
    {
      "epoch": 0.6756560355268469,
      "grad_norm": 85.08837127685547,
      "learning_rate": 5.028714791668463e-06,
      "loss": 3.6684,
      "step": 16736
    },
    {
      "epoch": 0.6756964069438838,
      "grad_norm": 85.81285095214844,
      "learning_rate": 5.027580276844642e-06,
      "loss": 3.9581,
      "step": 16737
    },
    {
      "epoch": 0.6757367783609205,
      "grad_norm": 71.3017578125,
      "learning_rate": 5.026445847037588e-06,
      "loss": 5.7636,
      "step": 16738
    },
    {
      "epoch": 0.6757771497779572,
      "grad_norm": 233.8521728515625,
      "learning_rate": 5.025311502266695e-06,
      "loss": 5.9558,
      "step": 16739
    },
    {
      "epoch": 0.6758175211949939,
      "grad_norm": 121.23543548583984,
      "learning_rate": 5.024177242551358e-06,
      "loss": 3.8222,
      "step": 16740
    },
    {
      "epoch": 0.6758578926120307,
      "grad_norm": 90.00035858154297,
      "learning_rate": 5.023043067910973e-06,
      "loss": 3.9603,
      "step": 16741
    },
    {
      "epoch": 0.6758982640290674,
      "grad_norm": 67.76567840576172,
      "learning_rate": 5.0219089783649275e-06,
      "loss": 5.7065,
      "step": 16742
    },
    {
      "epoch": 0.6759386354461041,
      "grad_norm": 122.5343246459961,
      "learning_rate": 5.020774973932614e-06,
      "loss": 3.7896,
      "step": 16743
    },
    {
      "epoch": 0.6759790068631409,
      "grad_norm": 123.8775863647461,
      "learning_rate": 5.019641054633426e-06,
      "loss": 7.0682,
      "step": 16744
    },
    {
      "epoch": 0.6760193782801777,
      "grad_norm": 140.26895141601562,
      "learning_rate": 5.018507220486741e-06,
      "loss": 5.1016,
      "step": 16745
    },
    {
      "epoch": 0.6760597496972144,
      "grad_norm": 95.68089294433594,
      "learning_rate": 5.017373471511951e-06,
      "loss": 5.3129,
      "step": 16746
    },
    {
      "epoch": 0.6761001211142511,
      "grad_norm": 124.80997467041016,
      "learning_rate": 5.0162398077284395e-06,
      "loss": 4.401,
      "step": 16747
    },
    {
      "epoch": 0.6761404925312878,
      "grad_norm": 89.90096282958984,
      "learning_rate": 5.015106229155593e-06,
      "loss": 4.3936,
      "step": 16748
    },
    {
      "epoch": 0.6761808639483246,
      "grad_norm": 98.45164489746094,
      "learning_rate": 5.0139727358127845e-06,
      "loss": 5.285,
      "step": 16749
    },
    {
      "epoch": 0.6762212353653613,
      "grad_norm": 93.34999084472656,
      "learning_rate": 5.0128393277194e-06,
      "loss": 5.4873,
      "step": 16750
    },
    {
      "epoch": 0.6762616067823981,
      "grad_norm": 88.5638198852539,
      "learning_rate": 5.011706004894817e-06,
      "loss": 5.228,
      "step": 16751
    },
    {
      "epoch": 0.6763019781994348,
      "grad_norm": 81.80194091796875,
      "learning_rate": 5.010572767358417e-06,
      "loss": 6.2255,
      "step": 16752
    },
    {
      "epoch": 0.6763423496164715,
      "grad_norm": 93.16120147705078,
      "learning_rate": 5.0094396151295665e-06,
      "loss": 4.1764,
      "step": 16753
    },
    {
      "epoch": 0.6763827210335083,
      "grad_norm": 88.78392791748047,
      "learning_rate": 5.0083065482276465e-06,
      "loss": 6.1352,
      "step": 16754
    },
    {
      "epoch": 0.676423092450545,
      "grad_norm": 81.8484878540039,
      "learning_rate": 5.007173566672026e-06,
      "loss": 4.9255,
      "step": 16755
    },
    {
      "epoch": 0.6764634638675817,
      "grad_norm": 122.4563980102539,
      "learning_rate": 5.006040670482083e-06,
      "loss": 6.3677,
      "step": 16756
    },
    {
      "epoch": 0.6765038352846184,
      "grad_norm": 88.58245086669922,
      "learning_rate": 5.004907859677179e-06,
      "loss": 4.7,
      "step": 16757
    },
    {
      "epoch": 0.6765442067016553,
      "grad_norm": 97.80509948730469,
      "learning_rate": 5.003775134276687e-06,
      "loss": 5.0529,
      "step": 16758
    },
    {
      "epoch": 0.676584578118692,
      "grad_norm": 82.65841674804688,
      "learning_rate": 5.002642494299973e-06,
      "loss": 3.7357,
      "step": 16759
    },
    {
      "epoch": 0.6766249495357287,
      "grad_norm": 74.06480407714844,
      "learning_rate": 5.001509939766402e-06,
      "loss": 4.0127,
      "step": 16760
    },
    {
      "epoch": 0.6766653209527654,
      "grad_norm": 86.24430084228516,
      "learning_rate": 5.000377470695339e-06,
      "loss": 3.3672,
      "step": 16761
    },
    {
      "epoch": 0.6767056923698022,
      "grad_norm": 98.91124725341797,
      "learning_rate": 4.999245087106146e-06,
      "loss": 4.8825,
      "step": 16762
    },
    {
      "epoch": 0.6767460637868389,
      "grad_norm": 88.58089447021484,
      "learning_rate": 4.998112789018186e-06,
      "loss": 4.3082,
      "step": 16763
    },
    {
      "epoch": 0.6767864352038756,
      "grad_norm": 126.13162994384766,
      "learning_rate": 4.996980576450819e-06,
      "loss": 4.0781,
      "step": 16764
    },
    {
      "epoch": 0.6768268066209124,
      "grad_norm": 76.45073699951172,
      "learning_rate": 4.995848449423398e-06,
      "loss": 4.4936,
      "step": 16765
    },
    {
      "epoch": 0.6768671780379492,
      "grad_norm": 105.6711654663086,
      "learning_rate": 4.9947164079552835e-06,
      "loss": 5.7777,
      "step": 16766
    },
    {
      "epoch": 0.6769075494549859,
      "grad_norm": 99.53886413574219,
      "learning_rate": 4.9935844520658296e-06,
      "loss": 5.4385,
      "step": 16767
    },
    {
      "epoch": 0.6769479208720226,
      "grad_norm": 89.19683074951172,
      "learning_rate": 4.992452581774394e-06,
      "loss": 4.9926,
      "step": 16768
    },
    {
      "epoch": 0.6769882922890593,
      "grad_norm": 113.06758880615234,
      "learning_rate": 4.991320797100323e-06,
      "loss": 4.6827,
      "step": 16769
    },
    {
      "epoch": 0.677028663706096,
      "grad_norm": 147.8348388671875,
      "learning_rate": 4.990189098062971e-06,
      "loss": 4.9108,
      "step": 16770
    },
    {
      "epoch": 0.6770690351231328,
      "grad_norm": 96.53105926513672,
      "learning_rate": 4.989057484681685e-06,
      "loss": 4.2181,
      "step": 16771
    },
    {
      "epoch": 0.6771094065401696,
      "grad_norm": 69.30353546142578,
      "learning_rate": 4.987925956975819e-06,
      "loss": 6.2257,
      "step": 16772
    },
    {
      "epoch": 0.6771497779572063,
      "grad_norm": 86.62349700927734,
      "learning_rate": 4.986794514964711e-06,
      "loss": 4.6832,
      "step": 16773
    },
    {
      "epoch": 0.677190149374243,
      "grad_norm": 75.71986389160156,
      "learning_rate": 4.985663158667711e-06,
      "loss": 5.0394,
      "step": 16774
    },
    {
      "epoch": 0.6772305207912798,
      "grad_norm": 104.98165893554688,
      "learning_rate": 4.984531888104161e-06,
      "loss": 2.9392,
      "step": 16775
    },
    {
      "epoch": 0.6772708922083165,
      "grad_norm": 72.94073486328125,
      "learning_rate": 4.983400703293408e-06,
      "loss": 4.0483,
      "step": 16776
    },
    {
      "epoch": 0.6773112636253532,
      "grad_norm": 80.9172592163086,
      "learning_rate": 4.982269604254785e-06,
      "loss": 5.1851,
      "step": 16777
    },
    {
      "epoch": 0.67735163504239,
      "grad_norm": 70.83802795410156,
      "learning_rate": 4.981138591007634e-06,
      "loss": 4.2646,
      "step": 16778
    },
    {
      "epoch": 0.6773920064594268,
      "grad_norm": 207.07923889160156,
      "learning_rate": 4.9800076635712925e-06,
      "loss": 6.1123,
      "step": 16779
    },
    {
      "epoch": 0.6774323778764635,
      "grad_norm": 99.8297119140625,
      "learning_rate": 4.978876821965097e-06,
      "loss": 6.0864,
      "step": 16780
    },
    {
      "epoch": 0.6774727492935002,
      "grad_norm": 89.0007553100586,
      "learning_rate": 4.977746066208384e-06,
      "loss": 4.5879,
      "step": 16781
    },
    {
      "epoch": 0.6775131207105369,
      "grad_norm": 79.9525375366211,
      "learning_rate": 4.976615396320485e-06,
      "loss": 3.6461,
      "step": 16782
    },
    {
      "epoch": 0.6775534921275737,
      "grad_norm": 87.60391235351562,
      "learning_rate": 4.975484812320734e-06,
      "loss": 5.3038,
      "step": 16783
    },
    {
      "epoch": 0.6775938635446104,
      "grad_norm": 79.66681671142578,
      "learning_rate": 4.974354314228461e-06,
      "loss": 5.545,
      "step": 16784
    },
    {
      "epoch": 0.6776342349616472,
      "grad_norm": 84.5287857055664,
      "learning_rate": 4.97322390206299e-06,
      "loss": 6.2684,
      "step": 16785
    },
    {
      "epoch": 0.6776746063786839,
      "grad_norm": 100.4414291381836,
      "learning_rate": 4.972093575843653e-06,
      "loss": 4.7803,
      "step": 16786
    },
    {
      "epoch": 0.6777149777957207,
      "grad_norm": 83.2159194946289,
      "learning_rate": 4.970963335589775e-06,
      "loss": 5.7417,
      "step": 16787
    },
    {
      "epoch": 0.6777553492127574,
      "grad_norm": 131.17491149902344,
      "learning_rate": 4.969833181320683e-06,
      "loss": 5.6815,
      "step": 16788
    },
    {
      "epoch": 0.6777957206297941,
      "grad_norm": 75.05694580078125,
      "learning_rate": 4.968703113055694e-06,
      "loss": 4.0767,
      "step": 16789
    },
    {
      "epoch": 0.6778360920468308,
      "grad_norm": 112.54602813720703,
      "learning_rate": 4.967573130814132e-06,
      "loss": 4.6587,
      "step": 16790
    },
    {
      "epoch": 0.6778764634638675,
      "grad_norm": 80.4886245727539,
      "learning_rate": 4.966443234615319e-06,
      "loss": 6.133,
      "step": 16791
    },
    {
      "epoch": 0.6779168348809044,
      "grad_norm": 75.9893798828125,
      "learning_rate": 4.965313424478576e-06,
      "loss": 5.3499,
      "step": 16792
    },
    {
      "epoch": 0.6779572062979411,
      "grad_norm": 112.84243774414062,
      "learning_rate": 4.964183700423213e-06,
      "loss": 6.4442,
      "step": 16793
    },
    {
      "epoch": 0.6779975777149778,
      "grad_norm": 113.67247772216797,
      "learning_rate": 4.9630540624685485e-06,
      "loss": 5.8466,
      "step": 16794
    },
    {
      "epoch": 0.6780379491320145,
      "grad_norm": 86.8468246459961,
      "learning_rate": 4.9619245106338975e-06,
      "loss": 5.6349,
      "step": 16795
    },
    {
      "epoch": 0.6780783205490513,
      "grad_norm": 193.91055297851562,
      "learning_rate": 4.960795044938574e-06,
      "loss": 4.2592,
      "step": 16796
    },
    {
      "epoch": 0.678118691966088,
      "grad_norm": 102.19733428955078,
      "learning_rate": 4.959665665401886e-06,
      "loss": 3.3348,
      "step": 16797
    },
    {
      "epoch": 0.6781590633831247,
      "grad_norm": 106.64688873291016,
      "learning_rate": 4.95853637204315e-06,
      "loss": 3.5508,
      "step": 16798
    },
    {
      "epoch": 0.6781994348001615,
      "grad_norm": 80.9615707397461,
      "learning_rate": 4.957407164881666e-06,
      "loss": 5.1679,
      "step": 16799
    },
    {
      "epoch": 0.6782398062171983,
      "grad_norm": 99.58003234863281,
      "learning_rate": 4.9562780439367434e-06,
      "loss": 4.764,
      "step": 16800
    },
    {
      "epoch": 0.678280177634235,
      "grad_norm": 103.94139862060547,
      "learning_rate": 4.955149009227688e-06,
      "loss": 5.6213,
      "step": 16801
    },
    {
      "epoch": 0.6783205490512717,
      "grad_norm": 142.72207641601562,
      "learning_rate": 4.954020060773804e-06,
      "loss": 4.9981,
      "step": 16802
    },
    {
      "epoch": 0.6783609204683084,
      "grad_norm": 116.3259048461914,
      "learning_rate": 4.952891198594395e-06,
      "loss": 4.0605,
      "step": 16803
    },
    {
      "epoch": 0.6784012918853451,
      "grad_norm": 123.37813568115234,
      "learning_rate": 4.951762422708763e-06,
      "loss": 3.7193,
      "step": 16804
    },
    {
      "epoch": 0.6784416633023819,
      "grad_norm": 122.92720031738281,
      "learning_rate": 4.950633733136202e-06,
      "loss": 6.5902,
      "step": 16805
    },
    {
      "epoch": 0.6784820347194187,
      "grad_norm": 77.11591339111328,
      "learning_rate": 4.949505129896013e-06,
      "loss": 5.6244,
      "step": 16806
    },
    {
      "epoch": 0.6785224061364554,
      "grad_norm": 114.44964599609375,
      "learning_rate": 4.9483766130074925e-06,
      "loss": 5.4773,
      "step": 16807
    },
    {
      "epoch": 0.6785627775534921,
      "grad_norm": 81.22996520996094,
      "learning_rate": 4.94724818248994e-06,
      "loss": 3.3199,
      "step": 16808
    },
    {
      "epoch": 0.6786031489705289,
      "grad_norm": 104.05867004394531,
      "learning_rate": 4.9461198383626395e-06,
      "loss": 4.3888,
      "step": 16809
    },
    {
      "epoch": 0.6786435203875656,
      "grad_norm": 110.38933563232422,
      "learning_rate": 4.944991580644889e-06,
      "loss": 5.1877,
      "step": 16810
    },
    {
      "epoch": 0.6786838918046023,
      "grad_norm": 78.86143493652344,
      "learning_rate": 4.943863409355978e-06,
      "loss": 5.8637,
      "step": 16811
    },
    {
      "epoch": 0.678724263221639,
      "grad_norm": 86.29202270507812,
      "learning_rate": 4.942735324515198e-06,
      "loss": 4.9889,
      "step": 16812
    },
    {
      "epoch": 0.6787646346386759,
      "grad_norm": 78.48983764648438,
      "learning_rate": 4.941607326141833e-06,
      "loss": 5.5108,
      "step": 16813
    },
    {
      "epoch": 0.6788050060557126,
      "grad_norm": 95.51824951171875,
      "learning_rate": 4.940479414255169e-06,
      "loss": 5.0827,
      "step": 16814
    },
    {
      "epoch": 0.6788453774727493,
      "grad_norm": 87.99897003173828,
      "learning_rate": 4.9393515888744936e-06,
      "loss": 4.4536,
      "step": 16815
    },
    {
      "epoch": 0.678885748889786,
      "grad_norm": 77.81808471679688,
      "learning_rate": 4.938223850019087e-06,
      "loss": 3.9465,
      "step": 16816
    },
    {
      "epoch": 0.6789261203068228,
      "grad_norm": 81.67127227783203,
      "learning_rate": 4.937096197708233e-06,
      "loss": 5.6013,
      "step": 16817
    },
    {
      "epoch": 0.6789664917238595,
      "grad_norm": 120.48275756835938,
      "learning_rate": 4.935968631961212e-06,
      "loss": 5.0316,
      "step": 16818
    },
    {
      "epoch": 0.6790068631408962,
      "grad_norm": 114.33748626708984,
      "learning_rate": 4.934841152797302e-06,
      "loss": 5.5037,
      "step": 16819
    },
    {
      "epoch": 0.679047234557933,
      "grad_norm": 182.68768310546875,
      "learning_rate": 4.933713760235783e-06,
      "loss": 5.0233,
      "step": 16820
    },
    {
      "epoch": 0.6790876059749698,
      "grad_norm": 152.50868225097656,
      "learning_rate": 4.932586454295925e-06,
      "loss": 6.0655,
      "step": 16821
    },
    {
      "epoch": 0.6791279773920065,
      "grad_norm": 89.88096618652344,
      "learning_rate": 4.931459234997005e-06,
      "loss": 5.1883,
      "step": 16822
    },
    {
      "epoch": 0.6791683488090432,
      "grad_norm": 165.39382934570312,
      "learning_rate": 4.9303321023582964e-06,
      "loss": 5.0777,
      "step": 16823
    },
    {
      "epoch": 0.6792087202260799,
      "grad_norm": 120.42854309082031,
      "learning_rate": 4.9292050563990735e-06,
      "loss": 4.6566,
      "step": 16824
    },
    {
      "epoch": 0.6792490916431166,
      "grad_norm": 76.84736633300781,
      "learning_rate": 4.9280780971386e-06,
      "loss": 3.7975,
      "step": 16825
    },
    {
      "epoch": 0.6792894630601534,
      "grad_norm": 113.93426513671875,
      "learning_rate": 4.926951224596147e-06,
      "loss": 5.5849,
      "step": 16826
    },
    {
      "epoch": 0.6793298344771902,
      "grad_norm": 94.3968276977539,
      "learning_rate": 4.925824438790982e-06,
      "loss": 3.9781,
      "step": 16827
    },
    {
      "epoch": 0.6793702058942269,
      "grad_norm": 71.39727020263672,
      "learning_rate": 4.9246977397423736e-06,
      "loss": 3.3663,
      "step": 16828
    },
    {
      "epoch": 0.6794105773112636,
      "grad_norm": 152.83633422851562,
      "learning_rate": 4.923571127469579e-06,
      "loss": 4.2185,
      "step": 16829
    },
    {
      "epoch": 0.6794509487283004,
      "grad_norm": 92.75272369384766,
      "learning_rate": 4.922444601991863e-06,
      "loss": 5.7927,
      "step": 16830
    },
    {
      "epoch": 0.6794913201453371,
      "grad_norm": 91.67752075195312,
      "learning_rate": 4.921318163328487e-06,
      "loss": 4.2582,
      "step": 16831
    },
    {
      "epoch": 0.6795316915623738,
      "grad_norm": 564.1155395507812,
      "learning_rate": 4.920191811498714e-06,
      "loss": 5.7869,
      "step": 16832
    },
    {
      "epoch": 0.6795720629794105,
      "grad_norm": 89.8603515625,
      "learning_rate": 4.919065546521797e-06,
      "loss": 6.2605,
      "step": 16833
    },
    {
      "epoch": 0.6796124343964474,
      "grad_norm": 79.4034194946289,
      "learning_rate": 4.917939368416993e-06,
      "loss": 5.8665,
      "step": 16834
    },
    {
      "epoch": 0.6796528058134841,
      "grad_norm": 204.1481170654297,
      "learning_rate": 4.91681327720356e-06,
      "loss": 4.0596,
      "step": 16835
    },
    {
      "epoch": 0.6796931772305208,
      "grad_norm": 116.03759002685547,
      "learning_rate": 4.915687272900747e-06,
      "loss": 6.1275,
      "step": 16836
    },
    {
      "epoch": 0.6797335486475575,
      "grad_norm": 94.12071990966797,
      "learning_rate": 4.914561355527812e-06,
      "loss": 5.6663,
      "step": 16837
    },
    {
      "epoch": 0.6797739200645943,
      "grad_norm": 133.12620544433594,
      "learning_rate": 4.913435525104e-06,
      "loss": 4.8332,
      "step": 16838
    },
    {
      "epoch": 0.679814291481631,
      "grad_norm": 85.0476303100586,
      "learning_rate": 4.912309781648564e-06,
      "loss": 3.6769,
      "step": 16839
    },
    {
      "epoch": 0.6798546628986677,
      "grad_norm": 112.66523742675781,
      "learning_rate": 4.911184125180754e-06,
      "loss": 4.4359,
      "step": 16840
    },
    {
      "epoch": 0.6798950343157045,
      "grad_norm": 100.05783081054688,
      "learning_rate": 4.910058555719806e-06,
      "loss": 4.5732,
      "step": 16841
    },
    {
      "epoch": 0.6799354057327412,
      "grad_norm": 111.11900329589844,
      "learning_rate": 4.908933073284974e-06,
      "loss": 5.8671,
      "step": 16842
    },
    {
      "epoch": 0.679975777149778,
      "grad_norm": 75.9126205444336,
      "learning_rate": 4.907807677895496e-06,
      "loss": 4.884,
      "step": 16843
    },
    {
      "epoch": 0.6800161485668147,
      "grad_norm": 84.95527648925781,
      "learning_rate": 4.90668236957062e-06,
      "loss": 4.9267,
      "step": 16844
    },
    {
      "epoch": 0.6800565199838514,
      "grad_norm": 85.14541625976562,
      "learning_rate": 4.905557148329578e-06,
      "loss": 6.6057,
      "step": 16845
    },
    {
      "epoch": 0.6800968914008881,
      "grad_norm": 77.35601806640625,
      "learning_rate": 4.904432014191613e-06,
      "loss": 4.697,
      "step": 16846
    },
    {
      "epoch": 0.6801372628179249,
      "grad_norm": 67.92073059082031,
      "learning_rate": 4.903306967175961e-06,
      "loss": 4.2553,
      "step": 16847
    },
    {
      "epoch": 0.6801776342349617,
      "grad_norm": 81.23719024658203,
      "learning_rate": 4.902182007301862e-06,
      "loss": 3.621,
      "step": 16848
    },
    {
      "epoch": 0.6802180056519984,
      "grad_norm": 93.20755767822266,
      "learning_rate": 4.901057134588542e-06,
      "loss": 3.6668,
      "step": 16849
    },
    {
      "epoch": 0.6802583770690351,
      "grad_norm": 86.82103729248047,
      "learning_rate": 4.899932349055238e-06,
      "loss": 4.3429,
      "step": 16850
    },
    {
      "epoch": 0.6802987484860719,
      "grad_norm": 91.65776062011719,
      "learning_rate": 4.898807650721183e-06,
      "loss": 6.11,
      "step": 16851
    },
    {
      "epoch": 0.6803391199031086,
      "grad_norm": 260.35784912109375,
      "learning_rate": 4.897683039605607e-06,
      "loss": 6.4268,
      "step": 16852
    },
    {
      "epoch": 0.6803794913201453,
      "grad_norm": 466.3518371582031,
      "learning_rate": 4.896558515727734e-06,
      "loss": 5.0163,
      "step": 16853
    },
    {
      "epoch": 0.680419862737182,
      "grad_norm": 73.57485961914062,
      "learning_rate": 4.8954340791067925e-06,
      "loss": 4.9274,
      "step": 16854
    },
    {
      "epoch": 0.6804602341542189,
      "grad_norm": 124.6329574584961,
      "learning_rate": 4.894309729762008e-06,
      "loss": 5.2572,
      "step": 16855
    },
    {
      "epoch": 0.6805006055712556,
      "grad_norm": 95.93815612792969,
      "learning_rate": 4.8931854677126045e-06,
      "loss": 5.7488,
      "step": 16856
    },
    {
      "epoch": 0.6805409769882923,
      "grad_norm": 105.49072265625,
      "learning_rate": 4.892061292977805e-06,
      "loss": 5.588,
      "step": 16857
    },
    {
      "epoch": 0.680581348405329,
      "grad_norm": 72.41352081298828,
      "learning_rate": 4.890937205576828e-06,
      "loss": 6.2626,
      "step": 16858
    },
    {
      "epoch": 0.6806217198223657,
      "grad_norm": 77.7691421508789,
      "learning_rate": 4.889813205528895e-06,
      "loss": 3.7858,
      "step": 16859
    },
    {
      "epoch": 0.6806620912394025,
      "grad_norm": 93.09346008300781,
      "learning_rate": 4.888689292853226e-06,
      "loss": 5.9873,
      "step": 16860
    },
    {
      "epoch": 0.6807024626564392,
      "grad_norm": 152.5189666748047,
      "learning_rate": 4.887565467569032e-06,
      "loss": 5.0617,
      "step": 16861
    },
    {
      "epoch": 0.680742834073476,
      "grad_norm": 75.70718383789062,
      "learning_rate": 4.886441729695529e-06,
      "loss": 5.1367,
      "step": 16862
    },
    {
      "epoch": 0.6807832054905127,
      "grad_norm": 109.83490753173828,
      "learning_rate": 4.885318079251932e-06,
      "loss": 6.0848,
      "step": 16863
    },
    {
      "epoch": 0.6808235769075495,
      "grad_norm": 73.27339172363281,
      "learning_rate": 4.8841945162574554e-06,
      "loss": 5.519,
      "step": 16864
    },
    {
      "epoch": 0.6808639483245862,
      "grad_norm": 102.47218322753906,
      "learning_rate": 4.883071040731303e-06,
      "loss": 5.2973,
      "step": 16865
    },
    {
      "epoch": 0.6809043197416229,
      "grad_norm": 85.10556030273438,
      "learning_rate": 4.881947652692688e-06,
      "loss": 4.9147,
      "step": 16866
    },
    {
      "epoch": 0.6809446911586596,
      "grad_norm": 90.68754577636719,
      "learning_rate": 4.8808243521608145e-06,
      "loss": 5.3257,
      "step": 16867
    },
    {
      "epoch": 0.6809850625756964,
      "grad_norm": 104.91261291503906,
      "learning_rate": 4.879701139154895e-06,
      "loss": 2.9377,
      "step": 16868
    },
    {
      "epoch": 0.6810254339927332,
      "grad_norm": 87.51148223876953,
      "learning_rate": 4.8785780136941265e-06,
      "loss": 3.474,
      "step": 16869
    },
    {
      "epoch": 0.6810658054097699,
      "grad_norm": 69.5558853149414,
      "learning_rate": 4.877454975797714e-06,
      "loss": 3.8196,
      "step": 16870
    },
    {
      "epoch": 0.6811061768268066,
      "grad_norm": 118.21382141113281,
      "learning_rate": 4.876332025484859e-06,
      "loss": 3.807,
      "step": 16871
    },
    {
      "epoch": 0.6811465482438434,
      "grad_norm": 96.67748260498047,
      "learning_rate": 4.875209162774765e-06,
      "loss": 4.1091,
      "step": 16872
    },
    {
      "epoch": 0.6811869196608801,
      "grad_norm": 238.6080780029297,
      "learning_rate": 4.874086387686624e-06,
      "loss": 5.1419,
      "step": 16873
    },
    {
      "epoch": 0.6812272910779168,
      "grad_norm": 139.4178466796875,
      "learning_rate": 4.8729637002396364e-06,
      "loss": 4.1741,
      "step": 16874
    },
    {
      "epoch": 0.6812676624949535,
      "grad_norm": 140.43228149414062,
      "learning_rate": 4.871841100452997e-06,
      "loss": 6.4544,
      "step": 16875
    },
    {
      "epoch": 0.6813080339119904,
      "grad_norm": 831.9501342773438,
      "learning_rate": 4.8707185883458985e-06,
      "loss": 5.9423,
      "step": 16876
    },
    {
      "epoch": 0.6813484053290271,
      "grad_norm": 106.82048034667969,
      "learning_rate": 4.8695961639375355e-06,
      "loss": 4.0035,
      "step": 16877
    },
    {
      "epoch": 0.6813887767460638,
      "grad_norm": 87.57030487060547,
      "learning_rate": 4.868473827247097e-06,
      "loss": 5.6603,
      "step": 16878
    },
    {
      "epoch": 0.6814291481631005,
      "grad_norm": 95.17327117919922,
      "learning_rate": 4.867351578293774e-06,
      "loss": 4.5468,
      "step": 16879
    },
    {
      "epoch": 0.6814695195801372,
      "grad_norm": 77.92425537109375,
      "learning_rate": 4.866229417096756e-06,
      "loss": 4.2684,
      "step": 16880
    },
    {
      "epoch": 0.681509890997174,
      "grad_norm": 132.30931091308594,
      "learning_rate": 4.865107343675224e-06,
      "loss": 5.1811,
      "step": 16881
    },
    {
      "epoch": 0.6815502624142107,
      "grad_norm": 99.5606918334961,
      "learning_rate": 4.8639853580483656e-06,
      "loss": 4.8985,
      "step": 16882
    },
    {
      "epoch": 0.6815906338312475,
      "grad_norm": 88.66024017333984,
      "learning_rate": 4.862863460235363e-06,
      "loss": 5.1479,
      "step": 16883
    },
    {
      "epoch": 0.6816310052482842,
      "grad_norm": 100.85706329345703,
      "learning_rate": 4.861741650255404e-06,
      "loss": 3.5118,
      "step": 16884
    },
    {
      "epoch": 0.681671376665321,
      "grad_norm": 73.9541015625,
      "learning_rate": 4.86061992812766e-06,
      "loss": 4.552,
      "step": 16885
    },
    {
      "epoch": 0.6817117480823577,
      "grad_norm": 94.07562255859375,
      "learning_rate": 4.859498293871314e-06,
      "loss": 5.3744,
      "step": 16886
    },
    {
      "epoch": 0.6817521194993944,
      "grad_norm": 92.18101501464844,
      "learning_rate": 4.8583767475055435e-06,
      "loss": 4.6233,
      "step": 16887
    },
    {
      "epoch": 0.6817924909164311,
      "grad_norm": 786.74169921875,
      "learning_rate": 4.857255289049527e-06,
      "loss": 5.5027,
      "step": 16888
    },
    {
      "epoch": 0.681832862333468,
      "grad_norm": 104.87982940673828,
      "learning_rate": 4.856133918522432e-06,
      "loss": 6.6534,
      "step": 16889
    },
    {
      "epoch": 0.6818732337505047,
      "grad_norm": 95.88318634033203,
      "learning_rate": 4.855012635943436e-06,
      "loss": 4.7956,
      "step": 16890
    },
    {
      "epoch": 0.6819136051675414,
      "grad_norm": 126.51411437988281,
      "learning_rate": 4.853891441331709e-06,
      "loss": 3.9266,
      "step": 16891
    },
    {
      "epoch": 0.6819539765845781,
      "grad_norm": 74.83723449707031,
      "learning_rate": 4.852770334706426e-06,
      "loss": 4.8434,
      "step": 16892
    },
    {
      "epoch": 0.6819943480016148,
      "grad_norm": 154.3409423828125,
      "learning_rate": 4.851649316086746e-06,
      "loss": 4.2425,
      "step": 16893
    },
    {
      "epoch": 0.6820347194186516,
      "grad_norm": 71.84025573730469,
      "learning_rate": 4.85052838549184e-06,
      "loss": 5.1895,
      "step": 16894
    },
    {
      "epoch": 0.6820750908356883,
      "grad_norm": 110.72489166259766,
      "learning_rate": 4.849407542940875e-06,
      "loss": 5.4828,
      "step": 16895
    },
    {
      "epoch": 0.6821154622527251,
      "grad_norm": 119.10173034667969,
      "learning_rate": 4.8482867884530124e-06,
      "loss": 5.1586,
      "step": 16896
    },
    {
      "epoch": 0.6821558336697618,
      "grad_norm": 96.86906433105469,
      "learning_rate": 4.8471661220474166e-06,
      "loss": 6.3581,
      "step": 16897
    },
    {
      "epoch": 0.6821962050867986,
      "grad_norm": 84.26685333251953,
      "learning_rate": 4.846045543743247e-06,
      "loss": 5.6146,
      "step": 16898
    },
    {
      "epoch": 0.6822365765038353,
      "grad_norm": 166.00286865234375,
      "learning_rate": 4.844925053559664e-06,
      "loss": 2.7673,
      "step": 16899
    },
    {
      "epoch": 0.682276947920872,
      "grad_norm": 108.2186279296875,
      "learning_rate": 4.843804651515826e-06,
      "loss": 5.2047,
      "step": 16900
    },
    {
      "epoch": 0.6823173193379087,
      "grad_norm": 67.0573501586914,
      "learning_rate": 4.842684337630887e-06,
      "loss": 5.1786,
      "step": 16901
    },
    {
      "epoch": 0.6823576907549455,
      "grad_norm": 89.97595977783203,
      "learning_rate": 4.841564111924001e-06,
      "loss": 5.1627,
      "step": 16902
    },
    {
      "epoch": 0.6823980621719823,
      "grad_norm": 86.58121490478516,
      "learning_rate": 4.840443974414323e-06,
      "loss": 4.2359,
      "step": 16903
    },
    {
      "epoch": 0.682438433589019,
      "grad_norm": 78.28315734863281,
      "learning_rate": 4.839323925121009e-06,
      "loss": 6.0697,
      "step": 16904
    },
    {
      "epoch": 0.6824788050060557,
      "grad_norm": 131.2845001220703,
      "learning_rate": 4.8382039640632e-06,
      "loss": 4.8071,
      "step": 16905
    },
    {
      "epoch": 0.6825191764230925,
      "grad_norm": 95.47614288330078,
      "learning_rate": 4.83708409126005e-06,
      "loss": 5.9127,
      "step": 16906
    },
    {
      "epoch": 0.6825595478401292,
      "grad_norm": 138.6815185546875,
      "learning_rate": 4.835964306730706e-06,
      "loss": 5.5573,
      "step": 16907
    },
    {
      "epoch": 0.6825999192571659,
      "grad_norm": 71.82324981689453,
      "learning_rate": 4.834844610494317e-06,
      "loss": 3.6103,
      "step": 16908
    },
    {
      "epoch": 0.6826402906742026,
      "grad_norm": 166.4097137451172,
      "learning_rate": 4.83372500257002e-06,
      "loss": 4.083,
      "step": 16909
    },
    {
      "epoch": 0.6826806620912395,
      "grad_norm": 104.78590393066406,
      "learning_rate": 4.832605482976962e-06,
      "loss": 4.3721,
      "step": 16910
    },
    {
      "epoch": 0.6827210335082762,
      "grad_norm": 81.06268310546875,
      "learning_rate": 4.831486051734282e-06,
      "loss": 4.8592,
      "step": 16911
    },
    {
      "epoch": 0.6827614049253129,
      "grad_norm": 180.0702362060547,
      "learning_rate": 4.830366708861126e-06,
      "loss": 3.358,
      "step": 16912
    },
    {
      "epoch": 0.6828017763423496,
      "grad_norm": 80.54481506347656,
      "learning_rate": 4.829247454376623e-06,
      "loss": 4.9375,
      "step": 16913
    },
    {
      "epoch": 0.6828421477593863,
      "grad_norm": 76.0305404663086,
      "learning_rate": 4.8281282882999146e-06,
      "loss": 4.5053,
      "step": 16914
    },
    {
      "epoch": 0.6828825191764231,
      "grad_norm": 149.39910888671875,
      "learning_rate": 4.827009210650136e-06,
      "loss": 3.8813,
      "step": 16915
    },
    {
      "epoch": 0.6829228905934598,
      "grad_norm": 122.4734115600586,
      "learning_rate": 4.825890221446418e-06,
      "loss": 5.207,
      "step": 16916
    },
    {
      "epoch": 0.6829632620104966,
      "grad_norm": 77.41138458251953,
      "learning_rate": 4.824771320707896e-06,
      "loss": 4.7946,
      "step": 16917
    },
    {
      "epoch": 0.6830036334275333,
      "grad_norm": 77.9317855834961,
      "learning_rate": 4.8236525084537e-06,
      "loss": 6.6892,
      "step": 16918
    },
    {
      "epoch": 0.6830440048445701,
      "grad_norm": 101.2374496459961,
      "learning_rate": 4.8225337847029574e-06,
      "loss": 4.0854,
      "step": 16919
    },
    {
      "epoch": 0.6830843762616068,
      "grad_norm": 108.67253112792969,
      "learning_rate": 4.821415149474801e-06,
      "loss": 6.6675,
      "step": 16920
    },
    {
      "epoch": 0.6831247476786435,
      "grad_norm": 94.34730529785156,
      "learning_rate": 4.820296602788349e-06,
      "loss": 5.3629,
      "step": 16921
    },
    {
      "epoch": 0.6831651190956802,
      "grad_norm": 82.08920288085938,
      "learning_rate": 4.8191781446627295e-06,
      "loss": 5.8582,
      "step": 16922
    },
    {
      "epoch": 0.683205490512717,
      "grad_norm": 85.43265533447266,
      "learning_rate": 4.818059775117066e-06,
      "loss": 4.796,
      "step": 16923
    },
    {
      "epoch": 0.6832458619297538,
      "grad_norm": 136.6990509033203,
      "learning_rate": 4.816941494170483e-06,
      "loss": 5.481,
      "step": 16924
    },
    {
      "epoch": 0.6832862333467905,
      "grad_norm": 84.74027252197266,
      "learning_rate": 4.815823301842093e-06,
      "loss": 4.4454,
      "step": 16925
    },
    {
      "epoch": 0.6833266047638272,
      "grad_norm": 80.35762023925781,
      "learning_rate": 4.814705198151019e-06,
      "loss": 5.2545,
      "step": 16926
    },
    {
      "epoch": 0.683366976180864,
      "grad_norm": 97.01428985595703,
      "learning_rate": 4.813587183116377e-06,
      "loss": 5.3628,
      "step": 16927
    },
    {
      "epoch": 0.6834073475979007,
      "grad_norm": 162.28546142578125,
      "learning_rate": 4.812469256757288e-06,
      "loss": 6.8191,
      "step": 16928
    },
    {
      "epoch": 0.6834477190149374,
      "grad_norm": 81.48583221435547,
      "learning_rate": 4.811351419092857e-06,
      "loss": 5.4002,
      "step": 16929
    },
    {
      "epoch": 0.6834880904319741,
      "grad_norm": 80.2685546875,
      "learning_rate": 4.8102336701421995e-06,
      "loss": 5.0591,
      "step": 16930
    },
    {
      "epoch": 0.683528461849011,
      "grad_norm": 60.22751998901367,
      "learning_rate": 4.809116009924427e-06,
      "loss": 4.7659,
      "step": 16931
    },
    {
      "epoch": 0.6835688332660477,
      "grad_norm": 123.32714080810547,
      "learning_rate": 4.807998438458654e-06,
      "loss": 4.0825,
      "step": 16932
    },
    {
      "epoch": 0.6836092046830844,
      "grad_norm": 101.49790954589844,
      "learning_rate": 4.80688095576398e-06,
      "loss": 4.4022,
      "step": 16933
    },
    {
      "epoch": 0.6836495761001211,
      "grad_norm": 97.51996612548828,
      "learning_rate": 4.805763561859513e-06,
      "loss": 5.2475,
      "step": 16934
    },
    {
      "epoch": 0.6836899475171578,
      "grad_norm": 198.56048583984375,
      "learning_rate": 4.8046462567643615e-06,
      "loss": 5.7318,
      "step": 16935
    },
    {
      "epoch": 0.6837303189341946,
      "grad_norm": 76.81675720214844,
      "learning_rate": 4.803529040497626e-06,
      "loss": 4.6862,
      "step": 16936
    },
    {
      "epoch": 0.6837706903512313,
      "grad_norm": 97.12998962402344,
      "learning_rate": 4.802411913078409e-06,
      "loss": 4.1005,
      "step": 16937
    },
    {
      "epoch": 0.6838110617682681,
      "grad_norm": 121.36680603027344,
      "learning_rate": 4.8012948745258126e-06,
      "loss": 5.069,
      "step": 16938
    },
    {
      "epoch": 0.6838514331853048,
      "grad_norm": 157.36660766601562,
      "learning_rate": 4.800177924858933e-06,
      "loss": 6.2032,
      "step": 16939
    },
    {
      "epoch": 0.6838918046023416,
      "grad_norm": 93.05772399902344,
      "learning_rate": 4.799061064096872e-06,
      "loss": 5.331,
      "step": 16940
    },
    {
      "epoch": 0.6839321760193783,
      "grad_norm": 110.49874877929688,
      "learning_rate": 4.797944292258718e-06,
      "loss": 5.0066,
      "step": 16941
    },
    {
      "epoch": 0.683972547436415,
      "grad_norm": 97.30767059326172,
      "learning_rate": 4.79682760936357e-06,
      "loss": 3.8555,
      "step": 16942
    },
    {
      "epoch": 0.6840129188534517,
      "grad_norm": 73.1950912475586,
      "learning_rate": 4.795711015430518e-06,
      "loss": 5.1886,
      "step": 16943
    },
    {
      "epoch": 0.6840532902704884,
      "grad_norm": 125.3678207397461,
      "learning_rate": 4.79459451047866e-06,
      "loss": 4.6722,
      "step": 16944
    },
    {
      "epoch": 0.6840936616875253,
      "grad_norm": 95.70816802978516,
      "learning_rate": 4.793478094527075e-06,
      "loss": 5.8247,
      "step": 16945
    },
    {
      "epoch": 0.684134033104562,
      "grad_norm": 90.6054916381836,
      "learning_rate": 4.792361767594858e-06,
      "loss": 2.756,
      "step": 16946
    },
    {
      "epoch": 0.6841744045215987,
      "grad_norm": 135.3959503173828,
      "learning_rate": 4.791245529701093e-06,
      "loss": 3.9855,
      "step": 16947
    },
    {
      "epoch": 0.6842147759386354,
      "grad_norm": 73.09820556640625,
      "learning_rate": 4.790129380864871e-06,
      "loss": 5.6178,
      "step": 16948
    },
    {
      "epoch": 0.6842551473556722,
      "grad_norm": 74.66812133789062,
      "learning_rate": 4.789013321105266e-06,
      "loss": 4.8375,
      "step": 16949
    },
    {
      "epoch": 0.6842955187727089,
      "grad_norm": 132.09886169433594,
      "learning_rate": 4.7878973504413654e-06,
      "loss": 4.7882,
      "step": 16950
    },
    {
      "epoch": 0.6843358901897456,
      "grad_norm": 120.56031036376953,
      "learning_rate": 4.786781468892247e-06,
      "loss": 5.7327,
      "step": 16951
    },
    {
      "epoch": 0.6843762616067824,
      "grad_norm": 88.71340942382812,
      "learning_rate": 4.785665676476998e-06,
      "loss": 4.6729,
      "step": 16952
    },
    {
      "epoch": 0.6844166330238192,
      "grad_norm": 97.6800308227539,
      "learning_rate": 4.784549973214685e-06,
      "loss": 3.2073,
      "step": 16953
    },
    {
      "epoch": 0.6844570044408559,
      "grad_norm": 86.99856567382812,
      "learning_rate": 4.783434359124388e-06,
      "loss": 5.4509,
      "step": 16954
    },
    {
      "epoch": 0.6844973758578926,
      "grad_norm": 78.90994262695312,
      "learning_rate": 4.782318834225182e-06,
      "loss": 4.2126,
      "step": 16955
    },
    {
      "epoch": 0.6845377472749293,
      "grad_norm": 138.28347778320312,
      "learning_rate": 4.78120339853614e-06,
      "loss": 4.9738,
      "step": 16956
    },
    {
      "epoch": 0.684578118691966,
      "grad_norm": 84.24940490722656,
      "learning_rate": 4.780088052076333e-06,
      "loss": 4.5751,
      "step": 16957
    },
    {
      "epoch": 0.6846184901090028,
      "grad_norm": 105.1917495727539,
      "learning_rate": 4.778972794864831e-06,
      "loss": 5.7506,
      "step": 16958
    },
    {
      "epoch": 0.6846588615260396,
      "grad_norm": 94.34325408935547,
      "learning_rate": 4.777857626920702e-06,
      "loss": 4.793,
      "step": 16959
    },
    {
      "epoch": 0.6846992329430763,
      "grad_norm": 80.12842559814453,
      "learning_rate": 4.776742548263018e-06,
      "loss": 5.3934,
      "step": 16960
    },
    {
      "epoch": 0.684739604360113,
      "grad_norm": 83.70255279541016,
      "learning_rate": 4.775627558910835e-06,
      "loss": 5.2863,
      "step": 16961
    },
    {
      "epoch": 0.6847799757771498,
      "grad_norm": 94.493896484375,
      "learning_rate": 4.7745126588832205e-06,
      "loss": 3.5249,
      "step": 16962
    },
    {
      "epoch": 0.6848203471941865,
      "grad_norm": 95.9891128540039,
      "learning_rate": 4.773397848199236e-06,
      "loss": 4.9539,
      "step": 16963
    },
    {
      "epoch": 0.6848607186112232,
      "grad_norm": 104.0252914428711,
      "learning_rate": 4.772283126877949e-06,
      "loss": 3.2361,
      "step": 16964
    },
    {
      "epoch": 0.6849010900282599,
      "grad_norm": 222.56777954101562,
      "learning_rate": 4.771168494938407e-06,
      "loss": 4.277,
      "step": 16965
    },
    {
      "epoch": 0.6849414614452968,
      "grad_norm": 104.31761169433594,
      "learning_rate": 4.770053952399676e-06,
      "loss": 5.6267,
      "step": 16966
    },
    {
      "epoch": 0.6849818328623335,
      "grad_norm": 82.98165893554688,
      "learning_rate": 4.768939499280808e-06,
      "loss": 5.2084,
      "step": 16967
    },
    {
      "epoch": 0.6850222042793702,
      "grad_norm": 94.34164428710938,
      "learning_rate": 4.7678251356008646e-06,
      "loss": 4.3239,
      "step": 16968
    },
    {
      "epoch": 0.6850625756964069,
      "grad_norm": 2074.218994140625,
      "learning_rate": 4.766710861378887e-06,
      "loss": 4.9315,
      "step": 16969
    },
    {
      "epoch": 0.6851029471134437,
      "grad_norm": 112.12747955322266,
      "learning_rate": 4.7655966766339355e-06,
      "loss": 4.2896,
      "step": 16970
    },
    {
      "epoch": 0.6851433185304804,
      "grad_norm": 109.93431091308594,
      "learning_rate": 4.764482581385056e-06,
      "loss": 5.3423,
      "step": 16971
    },
    {
      "epoch": 0.6851836899475171,
      "grad_norm": 111.7449951171875,
      "learning_rate": 4.763368575651302e-06,
      "loss": 4.7622,
      "step": 16972
    },
    {
      "epoch": 0.6852240613645539,
      "grad_norm": 90.84312438964844,
      "learning_rate": 4.762254659451715e-06,
      "loss": 3.9963,
      "step": 16973
    },
    {
      "epoch": 0.6852644327815907,
      "grad_norm": 88.42977142333984,
      "learning_rate": 4.761140832805342e-06,
      "loss": 5.0216,
      "step": 16974
    },
    {
      "epoch": 0.6853048041986274,
      "grad_norm": 151.0232696533203,
      "learning_rate": 4.7600270957312265e-06,
      "loss": 4.2425,
      "step": 16975
    },
    {
      "epoch": 0.6853451756156641,
      "grad_norm": 81.92987060546875,
      "learning_rate": 4.758913448248411e-06,
      "loss": 5.5067,
      "step": 16976
    },
    {
      "epoch": 0.6853855470327008,
      "grad_norm": 71.38133239746094,
      "learning_rate": 4.757799890375937e-06,
      "loss": 5.2896,
      "step": 16977
    },
    {
      "epoch": 0.6854259184497375,
      "grad_norm": 88.24390411376953,
      "learning_rate": 4.756686422132844e-06,
      "loss": 4.6921,
      "step": 16978
    },
    {
      "epoch": 0.6854662898667743,
      "grad_norm": 71.16626739501953,
      "learning_rate": 4.755573043538169e-06,
      "loss": 5.4981,
      "step": 16979
    },
    {
      "epoch": 0.6855066612838111,
      "grad_norm": 192.16763305664062,
      "learning_rate": 4.754459754610952e-06,
      "loss": 4.0184,
      "step": 16980
    },
    {
      "epoch": 0.6855470327008478,
      "grad_norm": 76.0222396850586,
      "learning_rate": 4.75334655537022e-06,
      "loss": 6.0257,
      "step": 16981
    },
    {
      "epoch": 0.6855874041178845,
      "grad_norm": 89.39772033691406,
      "learning_rate": 4.75223344583501e-06,
      "loss": 5.3832,
      "step": 16982
    },
    {
      "epoch": 0.6856277755349213,
      "grad_norm": 73.33684539794922,
      "learning_rate": 4.751120426024353e-06,
      "loss": 4.6694,
      "step": 16983
    },
    {
      "epoch": 0.685668146951958,
      "grad_norm": 89.70756530761719,
      "learning_rate": 4.750007495957286e-06,
      "loss": 4.7874,
      "step": 16984
    },
    {
      "epoch": 0.6857085183689947,
      "grad_norm": 104.4280014038086,
      "learning_rate": 4.748894655652826e-06,
      "loss": 6.0312,
      "step": 16985
    },
    {
      "epoch": 0.6857488897860314,
      "grad_norm": 99.55610656738281,
      "learning_rate": 4.747781905130005e-06,
      "loss": 5.7724,
      "step": 16986
    },
    {
      "epoch": 0.6857892612030683,
      "grad_norm": 78.1626968383789,
      "learning_rate": 4.746669244407848e-06,
      "loss": 5.4685,
      "step": 16987
    },
    {
      "epoch": 0.685829632620105,
      "grad_norm": 164.65269470214844,
      "learning_rate": 4.745556673505384e-06,
      "loss": 5.2338,
      "step": 16988
    },
    {
      "epoch": 0.6858700040371417,
      "grad_norm": 86.71009063720703,
      "learning_rate": 4.744444192441627e-06,
      "loss": 5.8065,
      "step": 16989
    },
    {
      "epoch": 0.6859103754541784,
      "grad_norm": 99.29137420654297,
      "learning_rate": 4.743331801235601e-06,
      "loss": 5.3006,
      "step": 16990
    },
    {
      "epoch": 0.6859507468712152,
      "grad_norm": 76.34793853759766,
      "learning_rate": 4.742219499906328e-06,
      "loss": 6.0525,
      "step": 16991
    },
    {
      "epoch": 0.6859911182882519,
      "grad_norm": 120.36714935302734,
      "learning_rate": 4.741107288472826e-06,
      "loss": 4.4327,
      "step": 16992
    },
    {
      "epoch": 0.6860314897052887,
      "grad_norm": 88.08200073242188,
      "learning_rate": 4.739995166954105e-06,
      "loss": 4.2483,
      "step": 16993
    },
    {
      "epoch": 0.6860718611223254,
      "grad_norm": 79.47383880615234,
      "learning_rate": 4.738883135369184e-06,
      "loss": 5.8405,
      "step": 16994
    },
    {
      "epoch": 0.6861122325393622,
      "grad_norm": 142.84939575195312,
      "learning_rate": 4.737771193737076e-06,
      "loss": 4.4833,
      "step": 16995
    },
    {
      "epoch": 0.6861526039563989,
      "grad_norm": 97.1917495727539,
      "learning_rate": 4.736659342076793e-06,
      "loss": 4.7549,
      "step": 16996
    },
    {
      "epoch": 0.6861929753734356,
      "grad_norm": 150.0635223388672,
      "learning_rate": 4.735547580407342e-06,
      "loss": 6.8036,
      "step": 16997
    },
    {
      "epoch": 0.6862333467904723,
      "grad_norm": 92.23660278320312,
      "learning_rate": 4.734435908747736e-06,
      "loss": 4.8397,
      "step": 16998
    },
    {
      "epoch": 0.686273718207509,
      "grad_norm": 104.6307144165039,
      "learning_rate": 4.73332432711698e-06,
      "loss": 4.054,
      "step": 16999
    },
    {
      "epoch": 0.6863140896245459,
      "grad_norm": 106.79364013671875,
      "learning_rate": 4.732212835534083e-06,
      "loss": 2.8685,
      "step": 17000
    },
    {
      "epoch": 0.6863544610415826,
      "grad_norm": 114.37445831298828,
      "learning_rate": 4.731101434018042e-06,
      "loss": 7.2884,
      "step": 17001
    },
    {
      "epoch": 0.6863948324586193,
      "grad_norm": 94.11799621582031,
      "learning_rate": 4.729990122587862e-06,
      "loss": 3.7925,
      "step": 17002
    },
    {
      "epoch": 0.686435203875656,
      "grad_norm": 198.26588439941406,
      "learning_rate": 4.728878901262545e-06,
      "loss": 4.6033,
      "step": 17003
    },
    {
      "epoch": 0.6864755752926928,
      "grad_norm": 104.82333374023438,
      "learning_rate": 4.727767770061093e-06,
      "loss": 4.253,
      "step": 17004
    },
    {
      "epoch": 0.6865159467097295,
      "grad_norm": 92.91316223144531,
      "learning_rate": 4.726656729002497e-06,
      "loss": 5.6159,
      "step": 17005
    },
    {
      "epoch": 0.6865563181267662,
      "grad_norm": 115.01342010498047,
      "learning_rate": 4.725545778105757e-06,
      "loss": 5.4217,
      "step": 17006
    },
    {
      "epoch": 0.686596689543803,
      "grad_norm": 95.7535171508789,
      "learning_rate": 4.724434917389868e-06,
      "loss": 4.8459,
      "step": 17007
    },
    {
      "epoch": 0.6866370609608398,
      "grad_norm": 485.66082763671875,
      "learning_rate": 4.723324146873826e-06,
      "loss": 5.3418,
      "step": 17008
    },
    {
      "epoch": 0.6866774323778765,
      "grad_norm": 132.54922485351562,
      "learning_rate": 4.722213466576614e-06,
      "loss": 5.2342,
      "step": 17009
    },
    {
      "epoch": 0.6867178037949132,
      "grad_norm": 78.15478515625,
      "learning_rate": 4.721102876517228e-06,
      "loss": 3.5418,
      "step": 17010
    },
    {
      "epoch": 0.6867581752119499,
      "grad_norm": 76.86993408203125,
      "learning_rate": 4.719992376714657e-06,
      "loss": 5.7533,
      "step": 17011
    },
    {
      "epoch": 0.6867985466289867,
      "grad_norm": 90.39675903320312,
      "learning_rate": 4.718881967187885e-06,
      "loss": 5.3016,
      "step": 17012
    },
    {
      "epoch": 0.6868389180460234,
      "grad_norm": 120.50993347167969,
      "learning_rate": 4.7177716479559e-06,
      "loss": 4.3774,
      "step": 17013
    },
    {
      "epoch": 0.6868792894630602,
      "grad_norm": 110.49047088623047,
      "learning_rate": 4.7166614190376845e-06,
      "loss": 6.0975,
      "step": 17014
    },
    {
      "epoch": 0.6869196608800969,
      "grad_norm": 117.05172729492188,
      "learning_rate": 4.715551280452222e-06,
      "loss": 6.4373,
      "step": 17015
    },
    {
      "epoch": 0.6869600322971336,
      "grad_norm": 154.67581176757812,
      "learning_rate": 4.714441232218495e-06,
      "loss": 5.0796,
      "step": 17016
    },
    {
      "epoch": 0.6870004037141704,
      "grad_norm": 86.38187408447266,
      "learning_rate": 4.7133312743554785e-06,
      "loss": 4.9316,
      "step": 17017
    },
    {
      "epoch": 0.6870407751312071,
      "grad_norm": 246.36312866210938,
      "learning_rate": 4.71222140688215e-06,
      "loss": 5.0741,
      "step": 17018
    },
    {
      "epoch": 0.6870811465482438,
      "grad_norm": 93.28076171875,
      "learning_rate": 4.711111629817489e-06,
      "loss": 2.4249,
      "step": 17019
    },
    {
      "epoch": 0.6871215179652805,
      "grad_norm": 76.65847778320312,
      "learning_rate": 4.710001943180471e-06,
      "loss": 5.1053,
      "step": 17020
    },
    {
      "epoch": 0.6871618893823174,
      "grad_norm": 115.41678619384766,
      "learning_rate": 4.708892346990064e-06,
      "loss": 6.0435,
      "step": 17021
    },
    {
      "epoch": 0.6872022607993541,
      "grad_norm": 116.03292083740234,
      "learning_rate": 4.7077828412652416e-06,
      "loss": 2.9547,
      "step": 17022
    },
    {
      "epoch": 0.6872426322163908,
      "grad_norm": 126.91175842285156,
      "learning_rate": 4.706673426024975e-06,
      "loss": 5.5591,
      "step": 17023
    },
    {
      "epoch": 0.6872830036334275,
      "grad_norm": 101.26835632324219,
      "learning_rate": 4.7055641012882356e-06,
      "loss": 6.3632,
      "step": 17024
    },
    {
      "epoch": 0.6873233750504643,
      "grad_norm": 199.23130798339844,
      "learning_rate": 4.7044548670739836e-06,
      "loss": 6.1665,
      "step": 17025
    },
    {
      "epoch": 0.687363746467501,
      "grad_norm": 86.05679321289062,
      "learning_rate": 4.703345723401186e-06,
      "loss": 3.0054,
      "step": 17026
    },
    {
      "epoch": 0.6874041178845377,
      "grad_norm": 91.64612579345703,
      "learning_rate": 4.702236670288809e-06,
      "loss": 6.2798,
      "step": 17027
    },
    {
      "epoch": 0.6874444893015745,
      "grad_norm": 195.0056915283203,
      "learning_rate": 4.701127707755817e-06,
      "loss": 4.7492,
      "step": 17028
    },
    {
      "epoch": 0.6874848607186113,
      "grad_norm": 92.375732421875,
      "learning_rate": 4.700018835821163e-06,
      "loss": 5.6303,
      "step": 17029
    },
    {
      "epoch": 0.687525232135648,
      "grad_norm": 130.12925720214844,
      "learning_rate": 4.698910054503811e-06,
      "loss": 6.082,
      "step": 17030
    },
    {
      "epoch": 0.6875656035526847,
      "grad_norm": 70.02368927001953,
      "learning_rate": 4.697801363822719e-06,
      "loss": 5.8632,
      "step": 17031
    },
    {
      "epoch": 0.6876059749697214,
      "grad_norm": 130.8658447265625,
      "learning_rate": 4.6966927637968416e-06,
      "loss": 5.2668,
      "step": 17032
    },
    {
      "epoch": 0.6876463463867581,
      "grad_norm": 92.12146759033203,
      "learning_rate": 4.695584254445134e-06,
      "loss": 4.5837,
      "step": 17033
    },
    {
      "epoch": 0.6876867178037949,
      "grad_norm": 102.11316680908203,
      "learning_rate": 4.694475835786549e-06,
      "loss": 6.1712,
      "step": 17034
    },
    {
      "epoch": 0.6877270892208317,
      "grad_norm": 70.6112289428711,
      "learning_rate": 4.693367507840038e-06,
      "loss": 4.5216,
      "step": 17035
    },
    {
      "epoch": 0.6877674606378684,
      "grad_norm": 72.59883117675781,
      "learning_rate": 4.692259270624554e-06,
      "loss": 5.455,
      "step": 17036
    },
    {
      "epoch": 0.6878078320549051,
      "grad_norm": 96.95551300048828,
      "learning_rate": 4.691151124159038e-06,
      "loss": 6.9038,
      "step": 17037
    },
    {
      "epoch": 0.6878482034719419,
      "grad_norm": 85.78760528564453,
      "learning_rate": 4.690043068462442e-06,
      "loss": 4.6047,
      "step": 17038
    },
    {
      "epoch": 0.6878885748889786,
      "grad_norm": 87.19424438476562,
      "learning_rate": 4.688935103553709e-06,
      "loss": 4.801,
      "step": 17039
    },
    {
      "epoch": 0.6879289463060153,
      "grad_norm": 75.41035461425781,
      "learning_rate": 4.687827229451787e-06,
      "loss": 5.6365,
      "step": 17040
    },
    {
      "epoch": 0.687969317723052,
      "grad_norm": 101.61270141601562,
      "learning_rate": 4.686719446175612e-06,
      "loss": 4.8319,
      "step": 17041
    },
    {
      "epoch": 0.6880096891400889,
      "grad_norm": 99.34553527832031,
      "learning_rate": 4.685611753744125e-06,
      "loss": 4.3645,
      "step": 17042
    },
    {
      "epoch": 0.6880500605571256,
      "grad_norm": 135.02456665039062,
      "learning_rate": 4.684504152176269e-06,
      "loss": 5.7518,
      "step": 17043
    },
    {
      "epoch": 0.6880904319741623,
      "grad_norm": 86.2132797241211,
      "learning_rate": 4.683396641490982e-06,
      "loss": 5.4426,
      "step": 17044
    },
    {
      "epoch": 0.688130803391199,
      "grad_norm": 89.91523742675781,
      "learning_rate": 4.6822892217071935e-06,
      "loss": 3.2479,
      "step": 17045
    },
    {
      "epoch": 0.6881711748082358,
      "grad_norm": 95.4117431640625,
      "learning_rate": 4.681181892843842e-06,
      "loss": 4.6773,
      "step": 17046
    },
    {
      "epoch": 0.6882115462252725,
      "grad_norm": 87.41275787353516,
      "learning_rate": 4.680074654919861e-06,
      "loss": 5.7671,
      "step": 17047
    },
    {
      "epoch": 0.6882519176423092,
      "grad_norm": 98.90067291259766,
      "learning_rate": 4.678967507954182e-06,
      "loss": 4.5238,
      "step": 17048
    },
    {
      "epoch": 0.688292289059346,
      "grad_norm": 92.7584228515625,
      "learning_rate": 4.677860451965731e-06,
      "loss": 5.6219,
      "step": 17049
    },
    {
      "epoch": 0.6883326604763828,
      "grad_norm": 79.10855865478516,
      "learning_rate": 4.676753486973437e-06,
      "loss": 3.7108,
      "step": 17050
    },
    {
      "epoch": 0.6883730318934195,
      "grad_norm": 93.342041015625,
      "learning_rate": 4.675646612996228e-06,
      "loss": 4.9618,
      "step": 17051
    },
    {
      "epoch": 0.6884134033104562,
      "grad_norm": 87.2967529296875,
      "learning_rate": 4.67453983005303e-06,
      "loss": 5.4906,
      "step": 17052
    },
    {
      "epoch": 0.6884537747274929,
      "grad_norm": 123.3769760131836,
      "learning_rate": 4.673433138162763e-06,
      "loss": 4.9655,
      "step": 17053
    },
    {
      "epoch": 0.6884941461445296,
      "grad_norm": 98.21309661865234,
      "learning_rate": 4.672326537344353e-06,
      "loss": 5.5939,
      "step": 17054
    },
    {
      "epoch": 0.6885345175615664,
      "grad_norm": 113.2170639038086,
      "learning_rate": 4.671220027616716e-06,
      "loss": 5.0255,
      "step": 17055
    },
    {
      "epoch": 0.6885748889786032,
      "grad_norm": 101.03866577148438,
      "learning_rate": 4.670113608998779e-06,
      "loss": 4.5015,
      "step": 17056
    },
    {
      "epoch": 0.6886152603956399,
      "grad_norm": 61.443511962890625,
      "learning_rate": 4.669007281509448e-06,
      "loss": 4.98,
      "step": 17057
    },
    {
      "epoch": 0.6886556318126766,
      "grad_norm": 101.72517395019531,
      "learning_rate": 4.667901045167643e-06,
      "loss": 3.9131,
      "step": 17058
    },
    {
      "epoch": 0.6886960032297134,
      "grad_norm": 126.70796966552734,
      "learning_rate": 4.66679489999228e-06,
      "loss": 4.904,
      "step": 17059
    },
    {
      "epoch": 0.6887363746467501,
      "grad_norm": 72.65261840820312,
      "learning_rate": 4.665688846002273e-06,
      "loss": 5.0152,
      "step": 17060
    },
    {
      "epoch": 0.6887767460637868,
      "grad_norm": 195.80389404296875,
      "learning_rate": 4.664582883216527e-06,
      "loss": 4.3108,
      "step": 17061
    },
    {
      "epoch": 0.6888171174808235,
      "grad_norm": 93.9770278930664,
      "learning_rate": 4.663477011653955e-06,
      "loss": 6.3388,
      "step": 17062
    },
    {
      "epoch": 0.6888574888978604,
      "grad_norm": 137.5541229248047,
      "learning_rate": 4.662371231333465e-06,
      "loss": 4.9171,
      "step": 17063
    },
    {
      "epoch": 0.6888978603148971,
      "grad_norm": 97.53215789794922,
      "learning_rate": 4.661265542273965e-06,
      "loss": 6.28,
      "step": 17064
    },
    {
      "epoch": 0.6889382317319338,
      "grad_norm": 90.95106506347656,
      "learning_rate": 4.660159944494355e-06,
      "loss": 3.981,
      "step": 17065
    },
    {
      "epoch": 0.6889786031489705,
      "grad_norm": 89.33422088623047,
      "learning_rate": 4.659054438013539e-06,
      "loss": 4.7459,
      "step": 17066
    },
    {
      "epoch": 0.6890189745660072,
      "grad_norm": 125.33790588378906,
      "learning_rate": 4.657949022850421e-06,
      "loss": 7.0498,
      "step": 17067
    },
    {
      "epoch": 0.689059345983044,
      "grad_norm": 76.16047668457031,
      "learning_rate": 4.656843699023904e-06,
      "loss": 6.1191,
      "step": 17068
    },
    {
      "epoch": 0.6890997174000807,
      "grad_norm": 85.57247161865234,
      "learning_rate": 4.6557384665528785e-06,
      "loss": 4.6527,
      "step": 17069
    },
    {
      "epoch": 0.6891400888171175,
      "grad_norm": 138.7149658203125,
      "learning_rate": 4.654633325456246e-06,
      "loss": 3.8999,
      "step": 17070
    },
    {
      "epoch": 0.6891804602341542,
      "grad_norm": 131.45196533203125,
      "learning_rate": 4.6535282757529e-06,
      "loss": 4.0894,
      "step": 17071
    },
    {
      "epoch": 0.689220831651191,
      "grad_norm": 98.69526672363281,
      "learning_rate": 4.6524233174617375e-06,
      "loss": 4.9447,
      "step": 17072
    },
    {
      "epoch": 0.6892612030682277,
      "grad_norm": 100.84600830078125,
      "learning_rate": 4.651318450601647e-06,
      "loss": 5.7141,
      "step": 17073
    },
    {
      "epoch": 0.6893015744852644,
      "grad_norm": 116.01129150390625,
      "learning_rate": 4.6502136751915215e-06,
      "loss": 4.6921,
      "step": 17074
    },
    {
      "epoch": 0.6893419459023011,
      "grad_norm": 71.35448455810547,
      "learning_rate": 4.649108991250249e-06,
      "loss": 6.2994,
      "step": 17075
    },
    {
      "epoch": 0.6893823173193379,
      "grad_norm": 198.12867736816406,
      "learning_rate": 4.648004398796723e-06,
      "loss": 3.7148,
      "step": 17076
    },
    {
      "epoch": 0.6894226887363747,
      "grad_norm": 89.36797332763672,
      "learning_rate": 4.646899897849819e-06,
      "loss": 5.3205,
      "step": 17077
    },
    {
      "epoch": 0.6894630601534114,
      "grad_norm": 104.30048370361328,
      "learning_rate": 4.645795488428426e-06,
      "loss": 4.3414,
      "step": 17078
    },
    {
      "epoch": 0.6895034315704481,
      "grad_norm": 121.9161376953125,
      "learning_rate": 4.6446911705514275e-06,
      "loss": 5.6745,
      "step": 17079
    },
    {
      "epoch": 0.6895438029874849,
      "grad_norm": 84.74394989013672,
      "learning_rate": 4.643586944237709e-06,
      "loss": 5.8072,
      "step": 17080
    },
    {
      "epoch": 0.6895841744045216,
      "grad_norm": 94.10033416748047,
      "learning_rate": 4.642482809506142e-06,
      "loss": 4.9877,
      "step": 17081
    },
    {
      "epoch": 0.6896245458215583,
      "grad_norm": 80.77791595458984,
      "learning_rate": 4.641378766375606e-06,
      "loss": 5.3902,
      "step": 17082
    },
    {
      "epoch": 0.689664917238595,
      "grad_norm": 161.79225158691406,
      "learning_rate": 4.640274814864982e-06,
      "loss": 3.7771,
      "step": 17083
    },
    {
      "epoch": 0.6897052886556319,
      "grad_norm": 131.00820922851562,
      "learning_rate": 4.639170954993142e-06,
      "loss": 5.6555,
      "step": 17084
    },
    {
      "epoch": 0.6897456600726686,
      "grad_norm": 71.13349914550781,
      "learning_rate": 4.6380671867789636e-06,
      "loss": 5.1793,
      "step": 17085
    },
    {
      "epoch": 0.6897860314897053,
      "grad_norm": 115.0200424194336,
      "learning_rate": 4.636963510241313e-06,
      "loss": 6.3538,
      "step": 17086
    },
    {
      "epoch": 0.689826402906742,
      "grad_norm": 103.71098327636719,
      "learning_rate": 4.635859925399061e-06,
      "loss": 6.3157,
      "step": 17087
    },
    {
      "epoch": 0.6898667743237787,
      "grad_norm": 117.82437896728516,
      "learning_rate": 4.634756432271079e-06,
      "loss": 5.5704,
      "step": 17088
    },
    {
      "epoch": 0.6899071457408155,
      "grad_norm": 127.70804595947266,
      "learning_rate": 4.633653030876237e-06,
      "loss": 4.6329,
      "step": 17089
    },
    {
      "epoch": 0.6899475171578522,
      "grad_norm": 102.88946533203125,
      "learning_rate": 4.6325497212333935e-06,
      "loss": 5.4519,
      "step": 17090
    },
    {
      "epoch": 0.689987888574889,
      "grad_norm": 78.16486358642578,
      "learning_rate": 4.6314465033614155e-06,
      "loss": 4.713,
      "step": 17091
    },
    {
      "epoch": 0.6900282599919257,
      "grad_norm": 93.90751647949219,
      "learning_rate": 4.630343377279165e-06,
      "loss": 5.01,
      "step": 17092
    },
    {
      "epoch": 0.6900686314089625,
      "grad_norm": 96.42132568359375,
      "learning_rate": 4.629240343005506e-06,
      "loss": 3.3959,
      "step": 17093
    },
    {
      "epoch": 0.6901090028259992,
      "grad_norm": 80.82756805419922,
      "learning_rate": 4.628137400559292e-06,
      "loss": 6.3321,
      "step": 17094
    },
    {
      "epoch": 0.6901493742430359,
      "grad_norm": 80.63006591796875,
      "learning_rate": 4.627034549959387e-06,
      "loss": 4.5446,
      "step": 17095
    },
    {
      "epoch": 0.6901897456600726,
      "grad_norm": 88.37491607666016,
      "learning_rate": 4.625931791224644e-06,
      "loss": 3.8683,
      "step": 17096
    },
    {
      "epoch": 0.6902301170771095,
      "grad_norm": 137.01162719726562,
      "learning_rate": 4.6248291243739205e-06,
      "loss": 4.9818,
      "step": 17097
    },
    {
      "epoch": 0.6902704884941462,
      "grad_norm": 131.9752655029297,
      "learning_rate": 4.623726549426065e-06,
      "loss": 2.677,
      "step": 17098
    },
    {
      "epoch": 0.6903108599111829,
      "grad_norm": 103.31920623779297,
      "learning_rate": 4.62262406639993e-06,
      "loss": 5.8383,
      "step": 17099
    },
    {
      "epoch": 0.6903512313282196,
      "grad_norm": 82.65609741210938,
      "learning_rate": 4.621521675314367e-06,
      "loss": 6.4044,
      "step": 17100
    },
    {
      "epoch": 0.6903916027452563,
      "grad_norm": 87.50019836425781,
      "learning_rate": 4.620419376188227e-06,
      "loss": 5.558,
      "step": 17101
    },
    {
      "epoch": 0.6904319741622931,
      "grad_norm": 124.3536148071289,
      "learning_rate": 4.619317169040349e-06,
      "loss": 5.9651,
      "step": 17102
    },
    {
      "epoch": 0.6904723455793298,
      "grad_norm": 99.88121795654297,
      "learning_rate": 4.618215053889584e-06,
      "loss": 3.5901,
      "step": 17103
    },
    {
      "epoch": 0.6905127169963666,
      "grad_norm": 97.2399673461914,
      "learning_rate": 4.617113030754773e-06,
      "loss": 5.3433,
      "step": 17104
    },
    {
      "epoch": 0.6905530884134033,
      "grad_norm": 109.12648010253906,
      "learning_rate": 4.616011099654763e-06,
      "loss": 4.6026,
      "step": 17105
    },
    {
      "epoch": 0.6905934598304401,
      "grad_norm": 179.03396606445312,
      "learning_rate": 4.614909260608388e-06,
      "loss": 5.582,
      "step": 17106
    },
    {
      "epoch": 0.6906338312474768,
      "grad_norm": 80.928955078125,
      "learning_rate": 4.61380751363449e-06,
      "loss": 5.7994,
      "step": 17107
    },
    {
      "epoch": 0.6906742026645135,
      "grad_norm": 76.85791015625,
      "learning_rate": 4.612705858751903e-06,
      "loss": 3.5732,
      "step": 17108
    },
    {
      "epoch": 0.6907145740815502,
      "grad_norm": 87.95665740966797,
      "learning_rate": 4.6116042959794725e-06,
      "loss": 4.2738,
      "step": 17109
    },
    {
      "epoch": 0.690754945498587,
      "grad_norm": 94.52227783203125,
      "learning_rate": 4.610502825336021e-06,
      "loss": 6.859,
      "step": 17110
    },
    {
      "epoch": 0.6907953169156238,
      "grad_norm": 84.1344223022461,
      "learning_rate": 4.609401446840384e-06,
      "loss": 5.4861,
      "step": 17111
    },
    {
      "epoch": 0.6908356883326605,
      "grad_norm": 75.54227447509766,
      "learning_rate": 4.608300160511396e-06,
      "loss": 5.6418,
      "step": 17112
    },
    {
      "epoch": 0.6908760597496972,
      "grad_norm": 100.59651947021484,
      "learning_rate": 4.6071989663678854e-06,
      "loss": 5.4179,
      "step": 17113
    },
    {
      "epoch": 0.690916431166734,
      "grad_norm": 77.17646789550781,
      "learning_rate": 4.606097864428678e-06,
      "loss": 3.6508,
      "step": 17114
    },
    {
      "epoch": 0.6909568025837707,
      "grad_norm": 124.45343017578125,
      "learning_rate": 4.604996854712603e-06,
      "loss": 4.577,
      "step": 17115
    },
    {
      "epoch": 0.6909971740008074,
      "grad_norm": 130.6267547607422,
      "learning_rate": 4.603895937238481e-06,
      "loss": 3.9382,
      "step": 17116
    },
    {
      "epoch": 0.6910375454178441,
      "grad_norm": 99.80443572998047,
      "learning_rate": 4.602795112025144e-06,
      "loss": 5.2561,
      "step": 17117
    },
    {
      "epoch": 0.691077916834881,
      "grad_norm": 105.19087982177734,
      "learning_rate": 4.601694379091401e-06,
      "loss": 3.9073,
      "step": 17118
    },
    {
      "epoch": 0.6911182882519177,
      "grad_norm": 107.5442123413086,
      "learning_rate": 4.60059373845608e-06,
      "loss": 5.9101,
      "step": 17119
    },
    {
      "epoch": 0.6911586596689544,
      "grad_norm": 97.32266235351562,
      "learning_rate": 4.599493190137998e-06,
      "loss": 4.663,
      "step": 17120
    },
    {
      "epoch": 0.6911990310859911,
      "grad_norm": 79.4442367553711,
      "learning_rate": 4.598392734155974e-06,
      "loss": 5.3361,
      "step": 17121
    },
    {
      "epoch": 0.6912394025030278,
      "grad_norm": 97.82955169677734,
      "learning_rate": 4.597292370528816e-06,
      "loss": 4.233,
      "step": 17122
    },
    {
      "epoch": 0.6912797739200646,
      "grad_norm": 108.12471771240234,
      "learning_rate": 4.596192099275344e-06,
      "loss": 6.0607,
      "step": 17123
    },
    {
      "epoch": 0.6913201453371013,
      "grad_norm": 88.0953140258789,
      "learning_rate": 4.5950919204143674e-06,
      "loss": 4.5379,
      "step": 17124
    },
    {
      "epoch": 0.6913605167541381,
      "grad_norm": 99.10258483886719,
      "learning_rate": 4.593991833964702e-06,
      "loss": 5.8301,
      "step": 17125
    },
    {
      "epoch": 0.6914008881711748,
      "grad_norm": 68.546142578125,
      "learning_rate": 4.5928918399451476e-06,
      "loss": 3.1802,
      "step": 17126
    },
    {
      "epoch": 0.6914412595882116,
      "grad_norm": 68.02074432373047,
      "learning_rate": 4.5917919383745165e-06,
      "loss": 4.2015,
      "step": 17127
    },
    {
      "epoch": 0.6914816310052483,
      "grad_norm": 83.9935302734375,
      "learning_rate": 4.590692129271614e-06,
      "loss": 3.7359,
      "step": 17128
    },
    {
      "epoch": 0.691522002422285,
      "grad_norm": 106.62675476074219,
      "learning_rate": 4.5895924126552495e-06,
      "loss": 5.394,
      "step": 17129
    },
    {
      "epoch": 0.6915623738393217,
      "grad_norm": 74.84194946289062,
      "learning_rate": 4.588492788544215e-06,
      "loss": 5.4107,
      "step": 17130
    },
    {
      "epoch": 0.6916027452563585,
      "grad_norm": 119.2150650024414,
      "learning_rate": 4.587393256957319e-06,
      "loss": 3.1393,
      "step": 17131
    },
    {
      "epoch": 0.6916431166733953,
      "grad_norm": 72.12953186035156,
      "learning_rate": 4.586293817913358e-06,
      "loss": 4.9708,
      "step": 17132
    },
    {
      "epoch": 0.691683488090432,
      "grad_norm": 148.06346130371094,
      "learning_rate": 4.58519447143113e-06,
      "loss": 6.1246,
      "step": 17133
    },
    {
      "epoch": 0.6917238595074687,
      "grad_norm": 118.51099395751953,
      "learning_rate": 4.584095217529433e-06,
      "loss": 4.801,
      "step": 17134
    },
    {
      "epoch": 0.6917642309245055,
      "grad_norm": 163.0244598388672,
      "learning_rate": 4.58299605622706e-06,
      "loss": 4.3526,
      "step": 17135
    },
    {
      "epoch": 0.6918046023415422,
      "grad_norm": 176.34280395507812,
      "learning_rate": 4.581896987542805e-06,
      "loss": 3.8263,
      "step": 17136
    },
    {
      "epoch": 0.6918449737585789,
      "grad_norm": 74.03275299072266,
      "learning_rate": 4.580798011495464e-06,
      "loss": 4.3632,
      "step": 17137
    },
    {
      "epoch": 0.6918853451756156,
      "grad_norm": 100.900390625,
      "learning_rate": 4.579699128103817e-06,
      "loss": 3.8803,
      "step": 17138
    },
    {
      "epoch": 0.6919257165926525,
      "grad_norm": 86.93675994873047,
      "learning_rate": 4.578600337386659e-06,
      "loss": 5.254,
      "step": 17139
    },
    {
      "epoch": 0.6919660880096892,
      "grad_norm": 92.31816864013672,
      "learning_rate": 4.577501639362774e-06,
      "loss": 5.1086,
      "step": 17140
    },
    {
      "epoch": 0.6920064594267259,
      "grad_norm": 94.97288513183594,
      "learning_rate": 4.576403034050954e-06,
      "loss": 5.1021,
      "step": 17141
    },
    {
      "epoch": 0.6920468308437626,
      "grad_norm": 97.30461883544922,
      "learning_rate": 4.575304521469972e-06,
      "loss": 5.3537,
      "step": 17142
    },
    {
      "epoch": 0.6920872022607993,
      "grad_norm": 119.98014831542969,
      "learning_rate": 4.5742061016386164e-06,
      "loss": 5.5412,
      "step": 17143
    },
    {
      "epoch": 0.6921275736778361,
      "grad_norm": 118.98849487304688,
      "learning_rate": 4.573107774575665e-06,
      "loss": 3.3945,
      "step": 17144
    },
    {
      "epoch": 0.6921679450948728,
      "grad_norm": 125.66064453125,
      "learning_rate": 4.572009540299902e-06,
      "loss": 3.7419,
      "step": 17145
    },
    {
      "epoch": 0.6922083165119096,
      "grad_norm": 117.7176742553711,
      "learning_rate": 4.570911398830097e-06,
      "loss": 4.4015,
      "step": 17146
    },
    {
      "epoch": 0.6922486879289463,
      "grad_norm": 81.38261413574219,
      "learning_rate": 4.5698133501850295e-06,
      "loss": 5.8635,
      "step": 17147
    },
    {
      "epoch": 0.6922890593459831,
      "grad_norm": 102.02591705322266,
      "learning_rate": 4.568715394383473e-06,
      "loss": 5.5493,
      "step": 17148
    },
    {
      "epoch": 0.6923294307630198,
      "grad_norm": 382.7682800292969,
      "learning_rate": 4.567617531444204e-06,
      "loss": 4.2482,
      "step": 17149
    },
    {
      "epoch": 0.6923698021800565,
      "grad_norm": 152.20553588867188,
      "learning_rate": 4.5665197613859855e-06,
      "loss": 5.8588,
      "step": 17150
    },
    {
      "epoch": 0.6924101735970932,
      "grad_norm": 80.55037689208984,
      "learning_rate": 4.565422084227592e-06,
      "loss": 4.1681,
      "step": 17151
    },
    {
      "epoch": 0.69245054501413,
      "grad_norm": 94.43540954589844,
      "learning_rate": 4.56432449998779e-06,
      "loss": 5.6569,
      "step": 17152
    },
    {
      "epoch": 0.6924909164311668,
      "grad_norm": 93.12147521972656,
      "learning_rate": 4.563227008685346e-06,
      "loss": 4.1883,
      "step": 17153
    },
    {
      "epoch": 0.6925312878482035,
      "grad_norm": 108.9204330444336,
      "learning_rate": 4.562129610339024e-06,
      "loss": 4.9851,
      "step": 17154
    },
    {
      "epoch": 0.6925716592652402,
      "grad_norm": 64.44027709960938,
      "learning_rate": 4.561032304967586e-06,
      "loss": 3.0664,
      "step": 17155
    },
    {
      "epoch": 0.692612030682277,
      "grad_norm": 80.28748321533203,
      "learning_rate": 4.559935092589797e-06,
      "loss": 5.326,
      "step": 17156
    },
    {
      "epoch": 0.6926524020993137,
      "grad_norm": 93.39856719970703,
      "learning_rate": 4.558837973224417e-06,
      "loss": 3.8743,
      "step": 17157
    },
    {
      "epoch": 0.6926927735163504,
      "grad_norm": 203.29217529296875,
      "learning_rate": 4.557740946890198e-06,
      "loss": 4.6742,
      "step": 17158
    },
    {
      "epoch": 0.6927331449333871,
      "grad_norm": 89.76689910888672,
      "learning_rate": 4.5566440136059e-06,
      "loss": 5.3559,
      "step": 17159
    },
    {
      "epoch": 0.692773516350424,
      "grad_norm": 152.2233428955078,
      "learning_rate": 4.555547173390279e-06,
      "loss": 5.1144,
      "step": 17160
    },
    {
      "epoch": 0.6928138877674607,
      "grad_norm": 90.57421875,
      "learning_rate": 4.55445042626209e-06,
      "loss": 4.8051,
      "step": 17161
    },
    {
      "epoch": 0.6928542591844974,
      "grad_norm": 93.61825561523438,
      "learning_rate": 4.55335377224008e-06,
      "loss": 5.1487,
      "step": 17162
    },
    {
      "epoch": 0.6928946306015341,
      "grad_norm": 88.93045043945312,
      "learning_rate": 4.552257211343e-06,
      "loss": 5.3529,
      "step": 17163
    },
    {
      "epoch": 0.6929350020185708,
      "grad_norm": 98.22745513916016,
      "learning_rate": 4.551160743589602e-06,
      "loss": 3.9563,
      "step": 17164
    },
    {
      "epoch": 0.6929753734356076,
      "grad_norm": 2813.566650390625,
      "learning_rate": 4.550064368998636e-06,
      "loss": 3.2632,
      "step": 17165
    },
    {
      "epoch": 0.6930157448526443,
      "grad_norm": 93.87882995605469,
      "learning_rate": 4.548968087588836e-06,
      "loss": 5.4597,
      "step": 17166
    },
    {
      "epoch": 0.6930561162696811,
      "grad_norm": 98.95352172851562,
      "learning_rate": 4.547871899378955e-06,
      "loss": 5.4318,
      "step": 17167
    },
    {
      "epoch": 0.6930964876867178,
      "grad_norm": 92.78903198242188,
      "learning_rate": 4.546775804387732e-06,
      "loss": 4.1333,
      "step": 17168
    },
    {
      "epoch": 0.6931368591037546,
      "grad_norm": 126.68997955322266,
      "learning_rate": 4.545679802633912e-06,
      "loss": 5.8267,
      "step": 17169
    },
    {
      "epoch": 0.6931772305207913,
      "grad_norm": 123.48992156982422,
      "learning_rate": 4.544583894136227e-06,
      "loss": 4.2825,
      "step": 17170
    },
    {
      "epoch": 0.693217601937828,
      "grad_norm": 104.8251953125,
      "learning_rate": 4.543488078913419e-06,
      "loss": 4.5212,
      "step": 17171
    },
    {
      "epoch": 0.6932579733548647,
      "grad_norm": 85.28126525878906,
      "learning_rate": 4.542392356984222e-06,
      "loss": 5.2901,
      "step": 17172
    },
    {
      "epoch": 0.6932983447719014,
      "grad_norm": 174.9890899658203,
      "learning_rate": 4.541296728367373e-06,
      "loss": 4.3984,
      "step": 17173
    },
    {
      "epoch": 0.6933387161889383,
      "grad_norm": 73.23120880126953,
      "learning_rate": 4.5402011930816e-06,
      "loss": 4.276,
      "step": 17174
    },
    {
      "epoch": 0.693379087605975,
      "grad_norm": 75.75433349609375,
      "learning_rate": 4.539105751145639e-06,
      "loss": 4.5336,
      "step": 17175
    },
    {
      "epoch": 0.6934194590230117,
      "grad_norm": 103.39324951171875,
      "learning_rate": 4.5380104025782176e-06,
      "loss": 5.3522,
      "step": 17176
    },
    {
      "epoch": 0.6934598304400484,
      "grad_norm": 86.8774642944336,
      "learning_rate": 4.536915147398066e-06,
      "loss": 5.8291,
      "step": 17177
    },
    {
      "epoch": 0.6935002018570852,
      "grad_norm": 79.5683822631836,
      "learning_rate": 4.535819985623906e-06,
      "loss": 3.3351,
      "step": 17178
    },
    {
      "epoch": 0.6935405732741219,
      "grad_norm": 142.1754150390625,
      "learning_rate": 4.5347249172744645e-06,
      "loss": 5.6667,
      "step": 17179
    },
    {
      "epoch": 0.6935809446911586,
      "grad_norm": 130.44485473632812,
      "learning_rate": 4.533629942368464e-06,
      "loss": 5.5804,
      "step": 17180
    },
    {
      "epoch": 0.6936213161081954,
      "grad_norm": 155.5223846435547,
      "learning_rate": 4.5325350609246296e-06,
      "loss": 3.909,
      "step": 17181
    },
    {
      "epoch": 0.6936616875252322,
      "grad_norm": 125.8198471069336,
      "learning_rate": 4.531440272961675e-06,
      "loss": 3.7209,
      "step": 17182
    },
    {
      "epoch": 0.6937020589422689,
      "grad_norm": 145.0132293701172,
      "learning_rate": 4.530345578498323e-06,
      "loss": 4.9489,
      "step": 17183
    },
    {
      "epoch": 0.6937424303593056,
      "grad_norm": 103.28964233398438,
      "learning_rate": 4.529250977553288e-06,
      "loss": 4.092,
      "step": 17184
    },
    {
      "epoch": 0.6937828017763423,
      "grad_norm": 114.26598358154297,
      "learning_rate": 4.52815647014529e-06,
      "loss": 4.9454,
      "step": 17185
    },
    {
      "epoch": 0.693823173193379,
      "grad_norm": 95.07886505126953,
      "learning_rate": 4.527062056293034e-06,
      "loss": 4.2514,
      "step": 17186
    },
    {
      "epoch": 0.6938635446104158,
      "grad_norm": 116.2730484008789,
      "learning_rate": 4.5259677360152375e-06,
      "loss": 5.7649,
      "step": 17187
    },
    {
      "epoch": 0.6939039160274526,
      "grad_norm": 95.62627410888672,
      "learning_rate": 4.52487350933061e-06,
      "loss": 4.5648,
      "step": 17188
    },
    {
      "epoch": 0.6939442874444893,
      "grad_norm": 95.97819519042969,
      "learning_rate": 4.523779376257863e-06,
      "loss": 3.1658,
      "step": 17189
    },
    {
      "epoch": 0.693984658861526,
      "grad_norm": 108.7374267578125,
      "learning_rate": 4.522685336815699e-06,
      "loss": 4.3132,
      "step": 17190
    },
    {
      "epoch": 0.6940250302785628,
      "grad_norm": 77.47393035888672,
      "learning_rate": 4.521591391022824e-06,
      "loss": 6.1099,
      "step": 17191
    },
    {
      "epoch": 0.6940654016955995,
      "grad_norm": 85.49739074707031,
      "learning_rate": 4.5204975388979425e-06,
      "loss": 3.9073,
      "step": 17192
    },
    {
      "epoch": 0.6941057731126362,
      "grad_norm": 113.27161407470703,
      "learning_rate": 4.519403780459758e-06,
      "loss": 4.8189,
      "step": 17193
    },
    {
      "epoch": 0.6941461445296729,
      "grad_norm": 84.26387023925781,
      "learning_rate": 4.518310115726972e-06,
      "loss": 5.6981,
      "step": 17194
    },
    {
      "epoch": 0.6941865159467098,
      "grad_norm": 86.12577819824219,
      "learning_rate": 4.51721654471828e-06,
      "loss": 5.4726,
      "step": 17195
    },
    {
      "epoch": 0.6942268873637465,
      "grad_norm": 178.40029907226562,
      "learning_rate": 4.516123067452384e-06,
      "loss": 5.8413,
      "step": 17196
    },
    {
      "epoch": 0.6942672587807832,
      "grad_norm": 137.4790802001953,
      "learning_rate": 4.51502968394798e-06,
      "loss": 4.5731,
      "step": 17197
    },
    {
      "epoch": 0.6943076301978199,
      "grad_norm": 81.25787353515625,
      "learning_rate": 4.513936394223758e-06,
      "loss": 4.5235,
      "step": 17198
    },
    {
      "epoch": 0.6943480016148567,
      "grad_norm": 118.94619750976562,
      "learning_rate": 4.512843198298411e-06,
      "loss": 5.2783,
      "step": 17199
    },
    {
      "epoch": 0.6943883730318934,
      "grad_norm": 129.18980407714844,
      "learning_rate": 4.511750096190633e-06,
      "loss": 5.9828,
      "step": 17200
    },
    {
      "epoch": 0.6944287444489301,
      "grad_norm": 73.60225677490234,
      "learning_rate": 4.510657087919116e-06,
      "loss": 4.5483,
      "step": 17201
    },
    {
      "epoch": 0.6944691158659669,
      "grad_norm": 116.57772064208984,
      "learning_rate": 4.509564173502541e-06,
      "loss": 5.1743,
      "step": 17202
    },
    {
      "epoch": 0.6945094872830037,
      "grad_norm": 91.56251525878906,
      "learning_rate": 4.508471352959597e-06,
      "loss": 4.8603,
      "step": 17203
    },
    {
      "epoch": 0.6945498587000404,
      "grad_norm": 109.00216674804688,
      "learning_rate": 4.507378626308968e-06,
      "loss": 4.3589,
      "step": 17204
    },
    {
      "epoch": 0.6945902301170771,
      "grad_norm": 101.62020111083984,
      "learning_rate": 4.506285993569344e-06,
      "loss": 5.5858,
      "step": 17205
    },
    {
      "epoch": 0.6946306015341138,
      "grad_norm": 138.25253295898438,
      "learning_rate": 4.505193454759395e-06,
      "loss": 4.8141,
      "step": 17206
    },
    {
      "epoch": 0.6946709729511505,
      "grad_norm": 90.2122573852539,
      "learning_rate": 4.504101009897809e-06,
      "loss": 4.4484,
      "step": 17207
    },
    {
      "epoch": 0.6947113443681874,
      "grad_norm": 83.0337905883789,
      "learning_rate": 4.50300865900326e-06,
      "loss": 5.0311,
      "step": 17208
    },
    {
      "epoch": 0.6947517157852241,
      "grad_norm": 113.36331939697266,
      "learning_rate": 4.501916402094427e-06,
      "loss": 4.6015,
      "step": 17209
    },
    {
      "epoch": 0.6947920872022608,
      "grad_norm": 139.0012664794922,
      "learning_rate": 4.500824239189986e-06,
      "loss": 5.2,
      "step": 17210
    },
    {
      "epoch": 0.6948324586192975,
      "grad_norm": 95.84803771972656,
      "learning_rate": 4.499732170308611e-06,
      "loss": 6.4889,
      "step": 17211
    },
    {
      "epoch": 0.6948728300363343,
      "grad_norm": 176.69888305664062,
      "learning_rate": 4.498640195468968e-06,
      "loss": 5.2072,
      "step": 17212
    },
    {
      "epoch": 0.694913201453371,
      "grad_norm": 106.75960540771484,
      "learning_rate": 4.4975483146897324e-06,
      "loss": 5.4391,
      "step": 17213
    },
    {
      "epoch": 0.6949535728704077,
      "grad_norm": 135.53111267089844,
      "learning_rate": 4.496456527989571e-06,
      "loss": 6.0813,
      "step": 17214
    },
    {
      "epoch": 0.6949939442874445,
      "grad_norm": 102.76921081542969,
      "learning_rate": 4.4953648353871514e-06,
      "loss": 6.0898,
      "step": 17215
    },
    {
      "epoch": 0.6950343157044813,
      "grad_norm": 115.37709045410156,
      "learning_rate": 4.494273236901137e-06,
      "loss": 4.2216,
      "step": 17216
    },
    {
      "epoch": 0.695074687121518,
      "grad_norm": 118.57982635498047,
      "learning_rate": 4.4931817325502e-06,
      "loss": 4.538,
      "step": 17217
    },
    {
      "epoch": 0.6951150585385547,
      "grad_norm": 126.69984436035156,
      "learning_rate": 4.492090322352991e-06,
      "loss": 5.639,
      "step": 17218
    },
    {
      "epoch": 0.6951554299555914,
      "grad_norm": 180.75131225585938,
      "learning_rate": 4.490999006328175e-06,
      "loss": 6.3679,
      "step": 17219
    },
    {
      "epoch": 0.6951958013726282,
      "grad_norm": 113.38743591308594,
      "learning_rate": 4.4899077844944126e-06,
      "loss": 4.6416,
      "step": 17220
    },
    {
      "epoch": 0.6952361727896649,
      "grad_norm": 75.44873809814453,
      "learning_rate": 4.488816656870364e-06,
      "loss": 6.0681,
      "step": 17221
    },
    {
      "epoch": 0.6952765442067017,
      "grad_norm": 84.16446685791016,
      "learning_rate": 4.487725623474676e-06,
      "loss": 4.987,
      "step": 17222
    },
    {
      "epoch": 0.6953169156237384,
      "grad_norm": 78.88418579101562,
      "learning_rate": 4.486634684326009e-06,
      "loss": 5.4385,
      "step": 17223
    },
    {
      "epoch": 0.6953572870407752,
      "grad_norm": 78.72006225585938,
      "learning_rate": 4.485543839443013e-06,
      "loss": 4.7424,
      "step": 17224
    },
    {
      "epoch": 0.6953976584578119,
      "grad_norm": 93.34327697753906,
      "learning_rate": 4.4844530888443446e-06,
      "loss": 5.3977,
      "step": 17225
    },
    {
      "epoch": 0.6954380298748486,
      "grad_norm": 76.19733428955078,
      "learning_rate": 4.483362432548645e-06,
      "loss": 5.1508,
      "step": 17226
    },
    {
      "epoch": 0.6954784012918853,
      "grad_norm": 102.95402526855469,
      "learning_rate": 4.482271870574566e-06,
      "loss": 6.7392,
      "step": 17227
    },
    {
      "epoch": 0.695518772708922,
      "grad_norm": 93.67293548583984,
      "learning_rate": 4.481181402940752e-06,
      "loss": 5.5414,
      "step": 17228
    },
    {
      "epoch": 0.6955591441259589,
      "grad_norm": 140.53138732910156,
      "learning_rate": 4.480091029665849e-06,
      "loss": 5.0218,
      "step": 17229
    },
    {
      "epoch": 0.6955995155429956,
      "grad_norm": 91.6513671875,
      "learning_rate": 4.4790007507685005e-06,
      "loss": 5.9984,
      "step": 17230
    },
    {
      "epoch": 0.6956398869600323,
      "grad_norm": 92.98265075683594,
      "learning_rate": 4.477910566267346e-06,
      "loss": 4.6596,
      "step": 17231
    },
    {
      "epoch": 0.695680258377069,
      "grad_norm": 98.38272094726562,
      "learning_rate": 4.4768204761810256e-06,
      "loss": 4.4429,
      "step": 17232
    },
    {
      "epoch": 0.6957206297941058,
      "grad_norm": 82.6249771118164,
      "learning_rate": 4.475730480528181e-06,
      "loss": 5.3395,
      "step": 17233
    },
    {
      "epoch": 0.6957610012111425,
      "grad_norm": 84.36469268798828,
      "learning_rate": 4.474640579327442e-06,
      "loss": 5.565,
      "step": 17234
    },
    {
      "epoch": 0.6958013726281792,
      "grad_norm": 85.49687194824219,
      "learning_rate": 4.473550772597446e-06,
      "loss": 6.2258,
      "step": 17235
    },
    {
      "epoch": 0.695841744045216,
      "grad_norm": 109.73884582519531,
      "learning_rate": 4.472461060356828e-06,
      "loss": 4.8482,
      "step": 17236
    },
    {
      "epoch": 0.6958821154622528,
      "grad_norm": 122.35250854492188,
      "learning_rate": 4.47137144262422e-06,
      "loss": 5.0182,
      "step": 17237
    },
    {
      "epoch": 0.6959224868792895,
      "grad_norm": 146.9282989501953,
      "learning_rate": 4.470281919418247e-06,
      "loss": 5.5574,
      "step": 17238
    },
    {
      "epoch": 0.6959628582963262,
      "grad_norm": 100.15381622314453,
      "learning_rate": 4.4691924907575405e-06,
      "loss": 4.9517,
      "step": 17239
    },
    {
      "epoch": 0.6960032297133629,
      "grad_norm": 89.10322570800781,
      "learning_rate": 4.468103156660727e-06,
      "loss": 4.2894,
      "step": 17240
    },
    {
      "epoch": 0.6960436011303996,
      "grad_norm": 79.81146240234375,
      "learning_rate": 4.467013917146434e-06,
      "loss": 4.4694,
      "step": 17241
    },
    {
      "epoch": 0.6960839725474364,
      "grad_norm": 74.93058776855469,
      "learning_rate": 4.465924772233281e-06,
      "loss": 5.997,
      "step": 17242
    },
    {
      "epoch": 0.6961243439644732,
      "grad_norm": 81.04644012451172,
      "learning_rate": 4.464835721939891e-06,
      "loss": 3.9421,
      "step": 17243
    },
    {
      "epoch": 0.6961647153815099,
      "grad_norm": 452.2286376953125,
      "learning_rate": 4.463746766284883e-06,
      "loss": 5.6258,
      "step": 17244
    },
    {
      "epoch": 0.6962050867985466,
      "grad_norm": 105.47612762451172,
      "learning_rate": 4.462657905286882e-06,
      "loss": 5.192,
      "step": 17245
    },
    {
      "epoch": 0.6962454582155834,
      "grad_norm": 118.16793823242188,
      "learning_rate": 4.461569138964498e-06,
      "loss": 4.3186,
      "step": 17246
    },
    {
      "epoch": 0.6962858296326201,
      "grad_norm": 65.27139282226562,
      "learning_rate": 4.460480467336347e-06,
      "loss": 5.2659,
      "step": 17247
    },
    {
      "epoch": 0.6963262010496568,
      "grad_norm": 103.11061096191406,
      "learning_rate": 4.459391890421043e-06,
      "loss": 4.1369,
      "step": 17248
    },
    {
      "epoch": 0.6963665724666935,
      "grad_norm": 109.3372802734375,
      "learning_rate": 4.458303408237201e-06,
      "loss": 3.0969,
      "step": 17249
    },
    {
      "epoch": 0.6964069438837304,
      "grad_norm": 123.82856750488281,
      "learning_rate": 4.45721502080343e-06,
      "loss": 6.2805,
      "step": 17250
    },
    {
      "epoch": 0.6964473153007671,
      "grad_norm": 78.3170166015625,
      "learning_rate": 4.456126728138339e-06,
      "loss": 5.523,
      "step": 17251
    },
    {
      "epoch": 0.6964876867178038,
      "grad_norm": 97.39141845703125,
      "learning_rate": 4.455038530260535e-06,
      "loss": 5.8615,
      "step": 17252
    },
    {
      "epoch": 0.6965280581348405,
      "grad_norm": 87.52153778076172,
      "learning_rate": 4.4539504271886265e-06,
      "loss": 4.3893,
      "step": 17253
    },
    {
      "epoch": 0.6965684295518773,
      "grad_norm": 72.36328887939453,
      "learning_rate": 4.452862418941212e-06,
      "loss": 4.8854,
      "step": 17254
    },
    {
      "epoch": 0.696608800968914,
      "grad_norm": 87.23793029785156,
      "learning_rate": 4.451774505536896e-06,
      "loss": 4.9023,
      "step": 17255
    },
    {
      "epoch": 0.6966491723859507,
      "grad_norm": 115.17391204833984,
      "learning_rate": 4.45068668699428e-06,
      "loss": 4.312,
      "step": 17256
    },
    {
      "epoch": 0.6966895438029875,
      "grad_norm": 72.44102478027344,
      "learning_rate": 4.449598963331967e-06,
      "loss": 4.5967,
      "step": 17257
    },
    {
      "epoch": 0.6967299152200243,
      "grad_norm": 91.56978607177734,
      "learning_rate": 4.448511334568547e-06,
      "loss": 6.2558,
      "step": 17258
    },
    {
      "epoch": 0.696770286637061,
      "grad_norm": 124.1073226928711,
      "learning_rate": 4.447423800722619e-06,
      "loss": 4.3189,
      "step": 17259
    },
    {
      "epoch": 0.6968106580540977,
      "grad_norm": 90.74759674072266,
      "learning_rate": 4.4463363618127774e-06,
      "loss": 6.2665,
      "step": 17260
    },
    {
      "epoch": 0.6968510294711344,
      "grad_norm": 114.81510162353516,
      "learning_rate": 4.4452490178576205e-06,
      "loss": 4.5124,
      "step": 17261
    },
    {
      "epoch": 0.6968914008881711,
      "grad_norm": 71.22483825683594,
      "learning_rate": 4.4441617688757295e-06,
      "loss": 4.725,
      "step": 17262
    },
    {
      "epoch": 0.6969317723052079,
      "grad_norm": 1391.0740966796875,
      "learning_rate": 4.443074614885699e-06,
      "loss": 3.649,
      "step": 17263
    },
    {
      "epoch": 0.6969721437222447,
      "grad_norm": 232.603515625,
      "learning_rate": 4.441987555906115e-06,
      "loss": 2.6637,
      "step": 17264
    },
    {
      "epoch": 0.6970125151392814,
      "grad_norm": 99.37599182128906,
      "learning_rate": 4.44090059195557e-06,
      "loss": 5.7955,
      "step": 17265
    },
    {
      "epoch": 0.6970528865563181,
      "grad_norm": 90.28948211669922,
      "learning_rate": 4.43981372305264e-06,
      "loss": 4.8909,
      "step": 17266
    },
    {
      "epoch": 0.6970932579733549,
      "grad_norm": 97.41936492919922,
      "learning_rate": 4.438726949215909e-06,
      "loss": 7.2226,
      "step": 17267
    },
    {
      "epoch": 0.6971336293903916,
      "grad_norm": 79.83853912353516,
      "learning_rate": 4.437640270463963e-06,
      "loss": 5.5163,
      "step": 17268
    },
    {
      "epoch": 0.6971740008074283,
      "grad_norm": 76.4693832397461,
      "learning_rate": 4.436553686815378e-06,
      "loss": 4.3267,
      "step": 17269
    },
    {
      "epoch": 0.697214372224465,
      "grad_norm": 97.90230560302734,
      "learning_rate": 4.435467198288734e-06,
      "loss": 4.0145,
      "step": 17270
    },
    {
      "epoch": 0.6972547436415019,
      "grad_norm": 104.2063217163086,
      "learning_rate": 4.434380804902607e-06,
      "loss": 5.0642,
      "step": 17271
    },
    {
      "epoch": 0.6972951150585386,
      "grad_norm": 138.7152099609375,
      "learning_rate": 4.43329450667557e-06,
      "loss": 3.6989,
      "step": 17272
    },
    {
      "epoch": 0.6973354864755753,
      "grad_norm": 107.15251922607422,
      "learning_rate": 4.4322083036262045e-06,
      "loss": 4.063,
      "step": 17273
    },
    {
      "epoch": 0.697375857892612,
      "grad_norm": 91.34921264648438,
      "learning_rate": 4.431122195773069e-06,
      "loss": 6.1484,
      "step": 17274
    },
    {
      "epoch": 0.6974162293096487,
      "grad_norm": 101.62439727783203,
      "learning_rate": 4.4300361831347405e-06,
      "loss": 4.2843,
      "step": 17275
    },
    {
      "epoch": 0.6974566007266855,
      "grad_norm": 109.3066177368164,
      "learning_rate": 4.428950265729786e-06,
      "loss": 6.0033,
      "step": 17276
    },
    {
      "epoch": 0.6974969721437222,
      "grad_norm": 80.01844787597656,
      "learning_rate": 4.427864443576777e-06,
      "loss": 5.3109,
      "step": 17277
    },
    {
      "epoch": 0.697537343560759,
      "grad_norm": 84.94218444824219,
      "learning_rate": 4.42677871669427e-06,
      "loss": 5.5775,
      "step": 17278
    },
    {
      "epoch": 0.6975777149777957,
      "grad_norm": 109.82984161376953,
      "learning_rate": 4.4256930851008325e-06,
      "loss": 4.7589,
      "step": 17279
    },
    {
      "epoch": 0.6976180863948325,
      "grad_norm": 82.8894271850586,
      "learning_rate": 4.424607548815028e-06,
      "loss": 6.8566,
      "step": 17280
    },
    {
      "epoch": 0.6976584578118692,
      "grad_norm": 81.46297454833984,
      "learning_rate": 4.423522107855417e-06,
      "loss": 4.8574,
      "step": 17281
    },
    {
      "epoch": 0.6976988292289059,
      "grad_norm": 82.36959075927734,
      "learning_rate": 4.422436762240552e-06,
      "loss": 3.8239,
      "step": 17282
    },
    {
      "epoch": 0.6977392006459426,
      "grad_norm": 117.88728332519531,
      "learning_rate": 4.421351511988994e-06,
      "loss": 3.5548,
      "step": 17283
    },
    {
      "epoch": 0.6977795720629794,
      "grad_norm": 120.12472534179688,
      "learning_rate": 4.420266357119298e-06,
      "loss": 5.427,
      "step": 17284
    },
    {
      "epoch": 0.6978199434800162,
      "grad_norm": 74.41817474365234,
      "learning_rate": 4.419181297650023e-06,
      "loss": 4.6757,
      "step": 17285
    },
    {
      "epoch": 0.6978603148970529,
      "grad_norm": 96.4836196899414,
      "learning_rate": 4.41809633359971e-06,
      "loss": 4.8231,
      "step": 17286
    },
    {
      "epoch": 0.6979006863140896,
      "grad_norm": 134.53988647460938,
      "learning_rate": 4.417011464986916e-06,
      "loss": 6.0199,
      "step": 17287
    },
    {
      "epoch": 0.6979410577311264,
      "grad_norm": 87.72429656982422,
      "learning_rate": 4.415926691830188e-06,
      "loss": 4.5249,
      "step": 17288
    },
    {
      "epoch": 0.6979814291481631,
      "grad_norm": 125.12052154541016,
      "learning_rate": 4.414842014148074e-06,
      "loss": 5.8876,
      "step": 17289
    },
    {
      "epoch": 0.6980218005651998,
      "grad_norm": 99.25638580322266,
      "learning_rate": 4.413757431959119e-06,
      "loss": 4.504,
      "step": 17290
    },
    {
      "epoch": 0.6980621719822365,
      "grad_norm": 104.997802734375,
      "learning_rate": 4.4126729452818686e-06,
      "loss": 5.5698,
      "step": 17291
    },
    {
      "epoch": 0.6981025433992734,
      "grad_norm": 94.2094497680664,
      "learning_rate": 4.411588554134862e-06,
      "loss": 4.3975,
      "step": 17292
    },
    {
      "epoch": 0.6981429148163101,
      "grad_norm": 98.091064453125,
      "learning_rate": 4.410504258536646e-06,
      "loss": 3.7929,
      "step": 17293
    },
    {
      "epoch": 0.6981832862333468,
      "grad_norm": 110.20220947265625,
      "learning_rate": 4.40942005850575e-06,
      "loss": 5.4083,
      "step": 17294
    },
    {
      "epoch": 0.6982236576503835,
      "grad_norm": 120.9521484375,
      "learning_rate": 4.4083359540607166e-06,
      "loss": 4.2043,
      "step": 17295
    },
    {
      "epoch": 0.6982640290674202,
      "grad_norm": 95.00968170166016,
      "learning_rate": 4.407251945220081e-06,
      "loss": 6.0431,
      "step": 17296
    },
    {
      "epoch": 0.698304400484457,
      "grad_norm": 110.50133514404297,
      "learning_rate": 4.4061680320023805e-06,
      "loss": 3.306,
      "step": 17297
    },
    {
      "epoch": 0.6983447719014937,
      "grad_norm": 157.99639892578125,
      "learning_rate": 4.405084214426141e-06,
      "loss": 5.8325,
      "step": 17298
    },
    {
      "epoch": 0.6983851433185305,
      "grad_norm": 120.79037475585938,
      "learning_rate": 4.404000492509895e-06,
      "loss": 5.177,
      "step": 17299
    },
    {
      "epoch": 0.6984255147355672,
      "grad_norm": 110.02008819580078,
      "learning_rate": 4.402916866272174e-06,
      "loss": 4.9332,
      "step": 17300
    },
    {
      "epoch": 0.698465886152604,
      "grad_norm": 109.12020874023438,
      "learning_rate": 4.401833335731509e-06,
      "loss": 4.663,
      "step": 17301
    },
    {
      "epoch": 0.6985062575696407,
      "grad_norm": 111.38286590576172,
      "learning_rate": 4.400749900906416e-06,
      "loss": 4.4916,
      "step": 17302
    },
    {
      "epoch": 0.6985466289866774,
      "grad_norm": 151.26266479492188,
      "learning_rate": 4.399666561815426e-06,
      "loss": 5.6916,
      "step": 17303
    },
    {
      "epoch": 0.6985870004037141,
      "grad_norm": 126.06092834472656,
      "learning_rate": 4.398583318477059e-06,
      "loss": 6.223,
      "step": 17304
    },
    {
      "epoch": 0.6986273718207509,
      "grad_norm": 150.01451110839844,
      "learning_rate": 4.397500170909841e-06,
      "loss": 5.8639,
      "step": 17305
    },
    {
      "epoch": 0.6986677432377877,
      "grad_norm": 71.11611938476562,
      "learning_rate": 4.396417119132284e-06,
      "loss": 6.0928,
      "step": 17306
    },
    {
      "epoch": 0.6987081146548244,
      "grad_norm": 88.76323699951172,
      "learning_rate": 4.395334163162908e-06,
      "loss": 4.0032,
      "step": 17307
    },
    {
      "epoch": 0.6987484860718611,
      "grad_norm": 84.86945343017578,
      "learning_rate": 4.39425130302023e-06,
      "loss": 3.9712,
      "step": 17308
    },
    {
      "epoch": 0.6987888574888979,
      "grad_norm": 79.29949188232422,
      "learning_rate": 4.393168538722765e-06,
      "loss": 5.4911,
      "step": 17309
    },
    {
      "epoch": 0.6988292289059346,
      "grad_norm": 105.49249267578125,
      "learning_rate": 4.392085870289026e-06,
      "loss": 4.595,
      "step": 17310
    },
    {
      "epoch": 0.6988696003229713,
      "grad_norm": 113.08409881591797,
      "learning_rate": 4.391003297737521e-06,
      "loss": 4.6015,
      "step": 17311
    },
    {
      "epoch": 0.6989099717400081,
      "grad_norm": 108.34979248046875,
      "learning_rate": 4.389920821086764e-06,
      "loss": 4.7236,
      "step": 17312
    },
    {
      "epoch": 0.6989503431570449,
      "grad_norm": 73.15120697021484,
      "learning_rate": 4.388838440355262e-06,
      "loss": 5.3878,
      "step": 17313
    },
    {
      "epoch": 0.6989907145740816,
      "grad_norm": 79.79866027832031,
      "learning_rate": 4.387756155561518e-06,
      "loss": 5.3736,
      "step": 17314
    },
    {
      "epoch": 0.6990310859911183,
      "grad_norm": 90.17932891845703,
      "learning_rate": 4.386673966724037e-06,
      "loss": 5.4574,
      "step": 17315
    },
    {
      "epoch": 0.699071457408155,
      "grad_norm": 97.59152221679688,
      "learning_rate": 4.385591873861322e-06,
      "loss": 5.1214,
      "step": 17316
    },
    {
      "epoch": 0.6991118288251917,
      "grad_norm": 110.96312713623047,
      "learning_rate": 4.3845098769918815e-06,
      "loss": 5.1619,
      "step": 17317
    },
    {
      "epoch": 0.6991522002422285,
      "grad_norm": 92.21922302246094,
      "learning_rate": 4.383427976134204e-06,
      "loss": 4.7113,
      "step": 17318
    },
    {
      "epoch": 0.6991925716592653,
      "grad_norm": 83.86016082763672,
      "learning_rate": 4.382346171306793e-06,
      "loss": 4.5985,
      "step": 17319
    },
    {
      "epoch": 0.699232943076302,
      "grad_norm": 108.69779968261719,
      "learning_rate": 4.381264462528143e-06,
      "loss": 3.5188,
      "step": 17320
    },
    {
      "epoch": 0.6992733144933387,
      "grad_norm": 184.54617309570312,
      "learning_rate": 4.380182849816756e-06,
      "loss": 5.5829,
      "step": 17321
    },
    {
      "epoch": 0.6993136859103755,
      "grad_norm": 144.4804229736328,
      "learning_rate": 4.379101333191113e-06,
      "loss": 5.674,
      "step": 17322
    },
    {
      "epoch": 0.6993540573274122,
      "grad_norm": 87.65226745605469,
      "learning_rate": 4.378019912669712e-06,
      "loss": 4.9208,
      "step": 17323
    },
    {
      "epoch": 0.6993944287444489,
      "grad_norm": 86.65724182128906,
      "learning_rate": 4.376938588271044e-06,
      "loss": 5.211,
      "step": 17324
    },
    {
      "epoch": 0.6994348001614856,
      "grad_norm": 104.49736785888672,
      "learning_rate": 4.375857360013597e-06,
      "loss": 7.264,
      "step": 17325
    },
    {
      "epoch": 0.6994751715785225,
      "grad_norm": 81.48190307617188,
      "learning_rate": 4.374776227915852e-06,
      "loss": 6.1962,
      "step": 17326
    },
    {
      "epoch": 0.6995155429955592,
      "grad_norm": 151.84205627441406,
      "learning_rate": 4.373695191996299e-06,
      "loss": 5.891,
      "step": 17327
    },
    {
      "epoch": 0.6995559144125959,
      "grad_norm": 100.6440200805664,
      "learning_rate": 4.37261425227342e-06,
      "loss": 6.2271,
      "step": 17328
    },
    {
      "epoch": 0.6995962858296326,
      "grad_norm": 67.41336822509766,
      "learning_rate": 4.371533408765695e-06,
      "loss": 5.9666,
      "step": 17329
    },
    {
      "epoch": 0.6996366572466693,
      "grad_norm": 93.9764404296875,
      "learning_rate": 4.370452661491606e-06,
      "loss": 4.8639,
      "step": 17330
    },
    {
      "epoch": 0.6996770286637061,
      "grad_norm": 78.2952651977539,
      "learning_rate": 4.3693720104696314e-06,
      "loss": 5.0745,
      "step": 17331
    },
    {
      "epoch": 0.6997174000807428,
      "grad_norm": 127.05030059814453,
      "learning_rate": 4.368291455718245e-06,
      "loss": 7.2449,
      "step": 17332
    },
    {
      "epoch": 0.6997577714977796,
      "grad_norm": 107.59941864013672,
      "learning_rate": 4.36721099725593e-06,
      "loss": 4.017,
      "step": 17333
    },
    {
      "epoch": 0.6997981429148163,
      "grad_norm": 96.2642593383789,
      "learning_rate": 4.366130635101148e-06,
      "loss": 5.2316,
      "step": 17334
    },
    {
      "epoch": 0.6998385143318531,
      "grad_norm": 84.57457733154297,
      "learning_rate": 4.365050369272377e-06,
      "loss": 3.8481,
      "step": 17335
    },
    {
      "epoch": 0.6998788857488898,
      "grad_norm": 105.74762725830078,
      "learning_rate": 4.363970199788085e-06,
      "loss": 6.3498,
      "step": 17336
    },
    {
      "epoch": 0.6999192571659265,
      "grad_norm": 92.51458740234375,
      "learning_rate": 4.3628901266667455e-06,
      "loss": 4.415,
      "step": 17337
    },
    {
      "epoch": 0.6999596285829632,
      "grad_norm": 95.78955078125,
      "learning_rate": 4.361810149926818e-06,
      "loss": 4.6074,
      "step": 17338
    },
    {
      "epoch": 0.7,
      "grad_norm": 482.1076965332031,
      "learning_rate": 4.36073026958677e-06,
      "loss": 5.0539,
      "step": 17339
    },
    {
      "epoch": 0.7000403714170368,
      "grad_norm": 121.8583755493164,
      "learning_rate": 4.359650485665066e-06,
      "loss": 4.0872,
      "step": 17340
    },
    {
      "epoch": 0.7000807428340735,
      "grad_norm": 81.69019317626953,
      "learning_rate": 4.358570798180173e-06,
      "loss": 6.7174,
      "step": 17341
    },
    {
      "epoch": 0.7001211142511102,
      "grad_norm": 120.38992309570312,
      "learning_rate": 4.357491207150541e-06,
      "loss": 5.3661,
      "step": 17342
    },
    {
      "epoch": 0.700161485668147,
      "grad_norm": 159.30313110351562,
      "learning_rate": 4.356411712594631e-06,
      "loss": 5.7069,
      "step": 17343
    },
    {
      "epoch": 0.7002018570851837,
      "grad_norm": 122.84163665771484,
      "learning_rate": 4.355332314530905e-06,
      "loss": 4.9787,
      "step": 17344
    },
    {
      "epoch": 0.7002422285022204,
      "grad_norm": 164.9550323486328,
      "learning_rate": 4.354253012977817e-06,
      "loss": 4.8527,
      "step": 17345
    },
    {
      "epoch": 0.7002825999192571,
      "grad_norm": 84.99024200439453,
      "learning_rate": 4.3531738079538155e-06,
      "loss": 3.266,
      "step": 17346
    },
    {
      "epoch": 0.700322971336294,
      "grad_norm": 167.92727661132812,
      "learning_rate": 4.352094699477356e-06,
      "loss": 5.8986,
      "step": 17347
    },
    {
      "epoch": 0.7003633427533307,
      "grad_norm": 120.80859375,
      "learning_rate": 4.351015687566887e-06,
      "loss": 4.3948,
      "step": 17348
    },
    {
      "epoch": 0.7004037141703674,
      "grad_norm": 106.52874755859375,
      "learning_rate": 4.34993677224086e-06,
      "loss": 4.8263,
      "step": 17349
    },
    {
      "epoch": 0.7004440855874041,
      "grad_norm": 110.69287872314453,
      "learning_rate": 4.34885795351772e-06,
      "loss": 3.9761,
      "step": 17350
    },
    {
      "epoch": 0.7004844570044408,
      "grad_norm": 105.62761688232422,
      "learning_rate": 4.3477792314159125e-06,
      "loss": 4.6708,
      "step": 17351
    },
    {
      "epoch": 0.7005248284214776,
      "grad_norm": 110.25444793701172,
      "learning_rate": 4.346700605953881e-06,
      "loss": 4.7836,
      "step": 17352
    },
    {
      "epoch": 0.7005651998385143,
      "grad_norm": 119.03524017333984,
      "learning_rate": 4.345622077150072e-06,
      "loss": 5.0078,
      "step": 17353
    },
    {
      "epoch": 0.7006055712555511,
      "grad_norm": 129.3905487060547,
      "learning_rate": 4.344543645022918e-06,
      "loss": 4.3032,
      "step": 17354
    },
    {
      "epoch": 0.7006459426725878,
      "grad_norm": 67.74393463134766,
      "learning_rate": 4.34346530959086e-06,
      "loss": 4.6038,
      "step": 17355
    },
    {
      "epoch": 0.7006863140896246,
      "grad_norm": 146.3258056640625,
      "learning_rate": 4.342387070872337e-06,
      "loss": 4.9495,
      "step": 17356
    },
    {
      "epoch": 0.7007266855066613,
      "grad_norm": 94.64432525634766,
      "learning_rate": 4.341308928885787e-06,
      "loss": 4.947,
      "step": 17357
    },
    {
      "epoch": 0.700767056923698,
      "grad_norm": 84.0778579711914,
      "learning_rate": 4.340230883649638e-06,
      "loss": 5.4671,
      "step": 17358
    },
    {
      "epoch": 0.7008074283407347,
      "grad_norm": 88.32299041748047,
      "learning_rate": 4.339152935182322e-06,
      "loss": 4.7422,
      "step": 17359
    },
    {
      "epoch": 0.7008477997577715,
      "grad_norm": 80.34912872314453,
      "learning_rate": 4.338075083502273e-06,
      "loss": 5.878,
      "step": 17360
    },
    {
      "epoch": 0.7008881711748083,
      "grad_norm": 72.02247619628906,
      "learning_rate": 4.336997328627923e-06,
      "loss": 5.1385,
      "step": 17361
    },
    {
      "epoch": 0.700928542591845,
      "grad_norm": 164.96047973632812,
      "learning_rate": 4.33591967057769e-06,
      "loss": 5.4526,
      "step": 17362
    },
    {
      "epoch": 0.7009689140088817,
      "grad_norm": 88.71004486083984,
      "learning_rate": 4.334842109370004e-06,
      "loss": 4.2101,
      "step": 17363
    },
    {
      "epoch": 0.7010092854259184,
      "grad_norm": 87.0636978149414,
      "learning_rate": 4.3337646450232895e-06,
      "loss": 5.5785,
      "step": 17364
    },
    {
      "epoch": 0.7010496568429552,
      "grad_norm": 93.36476135253906,
      "learning_rate": 4.332687277555972e-06,
      "loss": 4.2252,
      "step": 17365
    },
    {
      "epoch": 0.7010900282599919,
      "grad_norm": 74.94815826416016,
      "learning_rate": 4.331610006986462e-06,
      "loss": 4.7377,
      "step": 17366
    },
    {
      "epoch": 0.7011303996770286,
      "grad_norm": 72.94271850585938,
      "learning_rate": 4.330532833333187e-06,
      "loss": 4.1685,
      "step": 17367
    },
    {
      "epoch": 0.7011707710940654,
      "grad_norm": 139.27197265625,
      "learning_rate": 4.329455756614559e-06,
      "loss": 6.1545,
      "step": 17368
    },
    {
      "epoch": 0.7012111425111022,
      "grad_norm": 150.43898010253906,
      "learning_rate": 4.328378776848997e-06,
      "loss": 4.5038,
      "step": 17369
    },
    {
      "epoch": 0.7012515139281389,
      "grad_norm": 145.66094970703125,
      "learning_rate": 4.3273018940549135e-06,
      "loss": 5.2288,
      "step": 17370
    },
    {
      "epoch": 0.7012918853451756,
      "grad_norm": 87.8629379272461,
      "learning_rate": 4.326225108250721e-06,
      "loss": 5.7582,
      "step": 17371
    },
    {
      "epoch": 0.7013322567622123,
      "grad_norm": 178.4451904296875,
      "learning_rate": 4.325148419454831e-06,
      "loss": 6.3529,
      "step": 17372
    },
    {
      "epoch": 0.7013726281792491,
      "grad_norm": 91.6837387084961,
      "learning_rate": 4.324071827685654e-06,
      "loss": 5.2402,
      "step": 17373
    },
    {
      "epoch": 0.7014129995962858,
      "grad_norm": 75.84447479248047,
      "learning_rate": 4.32299533296159e-06,
      "loss": 3.9368,
      "step": 17374
    },
    {
      "epoch": 0.7014533710133226,
      "grad_norm": 86.5787353515625,
      "learning_rate": 4.321918935301051e-06,
      "loss": 5.5347,
      "step": 17375
    },
    {
      "epoch": 0.7014937424303593,
      "grad_norm": 107.02930450439453,
      "learning_rate": 4.3208426347224374e-06,
      "loss": 4.431,
      "step": 17376
    },
    {
      "epoch": 0.7015341138473961,
      "grad_norm": 102.23687744140625,
      "learning_rate": 4.319766431244157e-06,
      "loss": 6.2287,
      "step": 17377
    },
    {
      "epoch": 0.7015744852644328,
      "grad_norm": 78.05802154541016,
      "learning_rate": 4.318690324884602e-06,
      "loss": 4.6653,
      "step": 17378
    },
    {
      "epoch": 0.7016148566814695,
      "grad_norm": 85.8141860961914,
      "learning_rate": 4.317614315662178e-06,
      "loss": 5.5404,
      "step": 17379
    },
    {
      "epoch": 0.7016552280985062,
      "grad_norm": 102.32984924316406,
      "learning_rate": 4.3165384035952776e-06,
      "loss": 5.5859,
      "step": 17380
    },
    {
      "epoch": 0.701695599515543,
      "grad_norm": 101.01371765136719,
      "learning_rate": 4.315462588702303e-06,
      "loss": 6.0089,
      "step": 17381
    },
    {
      "epoch": 0.7017359709325798,
      "grad_norm": 101.46328735351562,
      "learning_rate": 4.31438687100164e-06,
      "loss": 6.5332,
      "step": 17382
    },
    {
      "epoch": 0.7017763423496165,
      "grad_norm": 95.67948150634766,
      "learning_rate": 4.313311250511685e-06,
      "loss": 4.1074,
      "step": 17383
    },
    {
      "epoch": 0.7018167137666532,
      "grad_norm": 87.739501953125,
      "learning_rate": 4.312235727250827e-06,
      "loss": 4.4372,
      "step": 17384
    },
    {
      "epoch": 0.7018570851836899,
      "grad_norm": 84.71841430664062,
      "learning_rate": 4.311160301237462e-06,
      "loss": 5.0773,
      "step": 17385
    },
    {
      "epoch": 0.7018974566007267,
      "grad_norm": 96.36897277832031,
      "learning_rate": 4.310084972489966e-06,
      "loss": 5.1797,
      "step": 17386
    },
    {
      "epoch": 0.7019378280177634,
      "grad_norm": 113.65068817138672,
      "learning_rate": 4.309009741026731e-06,
      "loss": 6.3456,
      "step": 17387
    },
    {
      "epoch": 0.7019781994348001,
      "grad_norm": 120.18802642822266,
      "learning_rate": 4.307934606866139e-06,
      "loss": 3.4896,
      "step": 17388
    },
    {
      "epoch": 0.7020185708518369,
      "grad_norm": 169.3610382080078,
      "learning_rate": 4.306859570026575e-06,
      "loss": 3.7808,
      "step": 17389
    },
    {
      "epoch": 0.7020589422688737,
      "grad_norm": 69.82437133789062,
      "learning_rate": 4.3057846305264164e-06,
      "loss": 4.5378,
      "step": 17390
    },
    {
      "epoch": 0.7020993136859104,
      "grad_norm": 131.7162628173828,
      "learning_rate": 4.304709788384044e-06,
      "loss": 4.8555,
      "step": 17391
    },
    {
      "epoch": 0.7021396851029471,
      "grad_norm": 69.8065185546875,
      "learning_rate": 4.303635043617834e-06,
      "loss": 5.8424,
      "step": 17392
    },
    {
      "epoch": 0.7021800565199838,
      "grad_norm": 110.44876861572266,
      "learning_rate": 4.302560396246168e-06,
      "loss": 6.0542,
      "step": 17393
    },
    {
      "epoch": 0.7022204279370206,
      "grad_norm": 104.86949920654297,
      "learning_rate": 4.301485846287411e-06,
      "loss": 4.6376,
      "step": 17394
    },
    {
      "epoch": 0.7022607993540573,
      "grad_norm": 61.249752044677734,
      "learning_rate": 4.300411393759938e-06,
      "loss": 6.0626,
      "step": 17395
    },
    {
      "epoch": 0.7023011707710941,
      "grad_norm": 87.48478698730469,
      "learning_rate": 4.299337038682121e-06,
      "loss": 5.5059,
      "step": 17396
    },
    {
      "epoch": 0.7023415421881308,
      "grad_norm": 135.34689331054688,
      "learning_rate": 4.298262781072333e-06,
      "loss": 5.1338,
      "step": 17397
    },
    {
      "epoch": 0.7023819136051676,
      "grad_norm": 86.1022720336914,
      "learning_rate": 4.297188620948932e-06,
      "loss": 6.1968,
      "step": 17398
    },
    {
      "epoch": 0.7024222850222043,
      "grad_norm": 163.37103271484375,
      "learning_rate": 4.29611455833029e-06,
      "loss": 6.3009,
      "step": 17399
    },
    {
      "epoch": 0.702462656439241,
      "grad_norm": 77.55812072753906,
      "learning_rate": 4.2950405932347685e-06,
      "loss": 5.927,
      "step": 17400
    },
    {
      "epoch": 0.7025030278562777,
      "grad_norm": 94.47950744628906,
      "learning_rate": 4.293966725680735e-06,
      "loss": 4.7381,
      "step": 17401
    },
    {
      "epoch": 0.7025433992733144,
      "grad_norm": 87.24727630615234,
      "learning_rate": 4.292892955686541e-06,
      "loss": 4.0477,
      "step": 17402
    },
    {
      "epoch": 0.7025837706903513,
      "grad_norm": 101.1517333984375,
      "learning_rate": 4.2918192832705515e-06,
      "loss": 5.9272,
      "step": 17403
    },
    {
      "epoch": 0.702624142107388,
      "grad_norm": 121.6559829711914,
      "learning_rate": 4.290745708451124e-06,
      "loss": 3.7138,
      "step": 17404
    },
    {
      "epoch": 0.7026645135244247,
      "grad_norm": 152.37684631347656,
      "learning_rate": 4.289672231246614e-06,
      "loss": 3.7233,
      "step": 17405
    },
    {
      "epoch": 0.7027048849414614,
      "grad_norm": 322.1108093261719,
      "learning_rate": 4.288598851675372e-06,
      "loss": 5.2159,
      "step": 17406
    },
    {
      "epoch": 0.7027452563584982,
      "grad_norm": 107.32520294189453,
      "learning_rate": 4.287525569755752e-06,
      "loss": 3.6486,
      "step": 17407
    },
    {
      "epoch": 0.7027856277755349,
      "grad_norm": 122.30745697021484,
      "learning_rate": 4.2864523855061055e-06,
      "loss": 4.894,
      "step": 17408
    },
    {
      "epoch": 0.7028259991925716,
      "grad_norm": 102.22049713134766,
      "learning_rate": 4.285379298944782e-06,
      "loss": 4.8151,
      "step": 17409
    },
    {
      "epoch": 0.7028663706096084,
      "grad_norm": 117.54335021972656,
      "learning_rate": 4.284306310090127e-06,
      "loss": 5.093,
      "step": 17410
    },
    {
      "epoch": 0.7029067420266452,
      "grad_norm": 79.77732849121094,
      "learning_rate": 4.283233418960486e-06,
      "loss": 4.3694,
      "step": 17411
    },
    {
      "epoch": 0.7029471134436819,
      "grad_norm": 85.28683471679688,
      "learning_rate": 4.282160625574205e-06,
      "loss": 5.3762,
      "step": 17412
    },
    {
      "epoch": 0.7029874848607186,
      "grad_norm": 75.68037414550781,
      "learning_rate": 4.28108792994963e-06,
      "loss": 3.326,
      "step": 17413
    },
    {
      "epoch": 0.7030278562777553,
      "grad_norm": 63.51252365112305,
      "learning_rate": 4.280015332105091e-06,
      "loss": 3.7575,
      "step": 17414
    },
    {
      "epoch": 0.703068227694792,
      "grad_norm": 80.01803588867188,
      "learning_rate": 4.278942832058934e-06,
      "loss": 4.0121,
      "step": 17415
    },
    {
      "epoch": 0.7031085991118289,
      "grad_norm": 76.49838256835938,
      "learning_rate": 4.277870429829495e-06,
      "loss": 6.5606,
      "step": 17416
    },
    {
      "epoch": 0.7031489705288656,
      "grad_norm": 87.90613555908203,
      "learning_rate": 4.276798125435112e-06,
      "loss": 6.0278,
      "step": 17417
    },
    {
      "epoch": 0.7031893419459023,
      "grad_norm": 89.14688110351562,
      "learning_rate": 4.275725918894113e-06,
      "loss": 4.3467,
      "step": 17418
    },
    {
      "epoch": 0.703229713362939,
      "grad_norm": 98.2719955444336,
      "learning_rate": 4.274653810224835e-06,
      "loss": 4.6889,
      "step": 17419
    },
    {
      "epoch": 0.7032700847799758,
      "grad_norm": 86.83567810058594,
      "learning_rate": 4.273581799445606e-06,
      "loss": 5.2809,
      "step": 17420
    },
    {
      "epoch": 0.7033104561970125,
      "grad_norm": 121.45221710205078,
      "learning_rate": 4.272509886574759e-06,
      "loss": 3.4321,
      "step": 17421
    },
    {
      "epoch": 0.7033508276140492,
      "grad_norm": 116.32593536376953,
      "learning_rate": 4.271438071630617e-06,
      "loss": 5.3516,
      "step": 17422
    },
    {
      "epoch": 0.703391199031086,
      "grad_norm": 68.3155517578125,
      "learning_rate": 4.270366354631506e-06,
      "loss": 3.2573,
      "step": 17423
    },
    {
      "epoch": 0.7034315704481228,
      "grad_norm": 82.12701416015625,
      "learning_rate": 4.26929473559575e-06,
      "loss": 5.3758,
      "step": 17424
    },
    {
      "epoch": 0.7034719418651595,
      "grad_norm": 80.28794860839844,
      "learning_rate": 4.268223214541671e-06,
      "loss": 5.1415,
      "step": 17425
    },
    {
      "epoch": 0.7035123132821962,
      "grad_norm": 81.83232879638672,
      "learning_rate": 4.267151791487592e-06,
      "loss": 5.9681,
      "step": 17426
    },
    {
      "epoch": 0.7035526846992329,
      "grad_norm": 92.69026947021484,
      "learning_rate": 4.26608046645183e-06,
      "loss": 4.822,
      "step": 17427
    },
    {
      "epoch": 0.7035930561162697,
      "grad_norm": 85.46112823486328,
      "learning_rate": 4.265009239452701e-06,
      "loss": 3.745,
      "step": 17428
    },
    {
      "epoch": 0.7036334275333064,
      "grad_norm": 96.68354797363281,
      "learning_rate": 4.263938110508527e-06,
      "loss": 5.9507,
      "step": 17429
    },
    {
      "epoch": 0.7036737989503432,
      "grad_norm": 97.37110900878906,
      "learning_rate": 4.2628670796376125e-06,
      "loss": 4.9962,
      "step": 17430
    },
    {
      "epoch": 0.7037141703673799,
      "grad_norm": 379.4971008300781,
      "learning_rate": 4.261796146858274e-06,
      "loss": 4.8715,
      "step": 17431
    },
    {
      "epoch": 0.7037545417844167,
      "grad_norm": 79.1907958984375,
      "learning_rate": 4.260725312188821e-06,
      "loss": 5.6544,
      "step": 17432
    },
    {
      "epoch": 0.7037949132014534,
      "grad_norm": 82.3531723022461,
      "learning_rate": 4.259654575647567e-06,
      "loss": 5.0758,
      "step": 17433
    },
    {
      "epoch": 0.7038352846184901,
      "grad_norm": 220.47177124023438,
      "learning_rate": 4.258583937252811e-06,
      "loss": 5.8998,
      "step": 17434
    },
    {
      "epoch": 0.7038756560355268,
      "grad_norm": 65.62348175048828,
      "learning_rate": 4.257513397022864e-06,
      "loss": 5.3969,
      "step": 17435
    },
    {
      "epoch": 0.7039160274525635,
      "grad_norm": 88.90896606445312,
      "learning_rate": 4.256442954976027e-06,
      "loss": 4.891,
      "step": 17436
    },
    {
      "epoch": 0.7039563988696004,
      "grad_norm": 81.76351928710938,
      "learning_rate": 4.255372611130607e-06,
      "loss": 4.9713,
      "step": 17437
    },
    {
      "epoch": 0.7039967702866371,
      "grad_norm": 72.10315704345703,
      "learning_rate": 4.254302365504898e-06,
      "loss": 4.6392,
      "step": 17438
    },
    {
      "epoch": 0.7040371417036738,
      "grad_norm": 102.3854751586914,
      "learning_rate": 4.253232218117199e-06,
      "loss": 4.2484,
      "step": 17439
    },
    {
      "epoch": 0.7040775131207105,
      "grad_norm": 90.67841339111328,
      "learning_rate": 4.252162168985811e-06,
      "loss": 4.7046,
      "step": 17440
    },
    {
      "epoch": 0.7041178845377473,
      "grad_norm": 66.88239288330078,
      "learning_rate": 4.251092218129033e-06,
      "loss": 6.1846,
      "step": 17441
    },
    {
      "epoch": 0.704158255954784,
      "grad_norm": 107.27864074707031,
      "learning_rate": 4.250022365565148e-06,
      "loss": 4.3213,
      "step": 17442
    },
    {
      "epoch": 0.7041986273718207,
      "grad_norm": 114.44102478027344,
      "learning_rate": 4.248952611312452e-06,
      "loss": 3.6474,
      "step": 17443
    },
    {
      "epoch": 0.7042389987888575,
      "grad_norm": 112.1856918334961,
      "learning_rate": 4.2478829553892384e-06,
      "loss": 3.9474,
      "step": 17444
    },
    {
      "epoch": 0.7042793702058943,
      "grad_norm": 73.4468002319336,
      "learning_rate": 4.2468133978137945e-06,
      "loss": 5.173,
      "step": 17445
    },
    {
      "epoch": 0.704319741622931,
      "grad_norm": 199.99806213378906,
      "learning_rate": 4.245743938604406e-06,
      "loss": 4.2651,
      "step": 17446
    },
    {
      "epoch": 0.7043601130399677,
      "grad_norm": 78.670654296875,
      "learning_rate": 4.24467457777936e-06,
      "loss": 5.4964,
      "step": 17447
    },
    {
      "epoch": 0.7044004844570044,
      "grad_norm": 80.72975158691406,
      "learning_rate": 4.243605315356938e-06,
      "loss": 3.9793,
      "step": 17448
    },
    {
      "epoch": 0.7044408558740411,
      "grad_norm": 98.54410552978516,
      "learning_rate": 4.242536151355428e-06,
      "loss": 5.0235,
      "step": 17449
    },
    {
      "epoch": 0.7044812272910779,
      "grad_norm": 90.47726440429688,
      "learning_rate": 4.2414670857931e-06,
      "loss": 5.4672,
      "step": 17450
    },
    {
      "epoch": 0.7045215987081147,
      "grad_norm": 84.4603271484375,
      "learning_rate": 4.240398118688238e-06,
      "loss": 5.6263,
      "step": 17451
    },
    {
      "epoch": 0.7045619701251514,
      "grad_norm": 108.2198715209961,
      "learning_rate": 4.239329250059119e-06,
      "loss": 3.3242,
      "step": 17452
    },
    {
      "epoch": 0.7046023415421881,
      "grad_norm": 116.96259307861328,
      "learning_rate": 4.23826047992402e-06,
      "loss": 5.5771,
      "step": 17453
    },
    {
      "epoch": 0.7046427129592249,
      "grad_norm": 70.88396453857422,
      "learning_rate": 4.23719180830121e-06,
      "loss": 4.7362,
      "step": 17454
    },
    {
      "epoch": 0.7046830843762616,
      "grad_norm": 72.68640899658203,
      "learning_rate": 4.236123235208962e-06,
      "loss": 5.932,
      "step": 17455
    },
    {
      "epoch": 0.7047234557932983,
      "grad_norm": 89.43583679199219,
      "learning_rate": 4.235054760665548e-06,
      "loss": 3.6451,
      "step": 17456
    },
    {
      "epoch": 0.704763827210335,
      "grad_norm": 74.79299926757812,
      "learning_rate": 4.2339863846892385e-06,
      "loss": 4.7386,
      "step": 17457
    },
    {
      "epoch": 0.7048041986273719,
      "grad_norm": 128.75042724609375,
      "learning_rate": 4.232918107298294e-06,
      "loss": 5.1173,
      "step": 17458
    },
    {
      "epoch": 0.7048445700444086,
      "grad_norm": 72.18842315673828,
      "learning_rate": 4.2318499285109825e-06,
      "loss": 5.5654,
      "step": 17459
    },
    {
      "epoch": 0.7048849414614453,
      "grad_norm": 185.90818786621094,
      "learning_rate": 4.230781848345568e-06,
      "loss": 5.6875,
      "step": 17460
    },
    {
      "epoch": 0.704925312878482,
      "grad_norm": 75.88716125488281,
      "learning_rate": 4.229713866820314e-06,
      "loss": 4.8029,
      "step": 17461
    },
    {
      "epoch": 0.7049656842955188,
      "grad_norm": 1909.755126953125,
      "learning_rate": 4.228645983953476e-06,
      "loss": 4.2113,
      "step": 17462
    },
    {
      "epoch": 0.7050060557125555,
      "grad_norm": 104.46845245361328,
      "learning_rate": 4.227578199763315e-06,
      "loss": 5.2912,
      "step": 17463
    },
    {
      "epoch": 0.7050464271295922,
      "grad_norm": 116.19153594970703,
      "learning_rate": 4.226510514268086e-06,
      "loss": 4.6153,
      "step": 17464
    },
    {
      "epoch": 0.705086798546629,
      "grad_norm": 112.54148864746094,
      "learning_rate": 4.225442927486047e-06,
      "loss": 5.3845,
      "step": 17465
    },
    {
      "epoch": 0.7051271699636658,
      "grad_norm": 142.89488220214844,
      "learning_rate": 4.224375439435448e-06,
      "loss": 5.8266,
      "step": 17466
    },
    {
      "epoch": 0.7051675413807025,
      "grad_norm": 103.69859313964844,
      "learning_rate": 4.223308050134541e-06,
      "loss": 5.6602,
      "step": 17467
    },
    {
      "epoch": 0.7052079127977392,
      "grad_norm": 132.59341430664062,
      "learning_rate": 4.222240759601578e-06,
      "loss": 5.9902,
      "step": 17468
    },
    {
      "epoch": 0.7052482842147759,
      "grad_norm": 71.17195892333984,
      "learning_rate": 4.221173567854809e-06,
      "loss": 6.0135,
      "step": 17469
    },
    {
      "epoch": 0.7052886556318126,
      "grad_norm": 146.0151824951172,
      "learning_rate": 4.220106474912474e-06,
      "loss": 4.1055,
      "step": 17470
    },
    {
      "epoch": 0.7053290270488494,
      "grad_norm": 87.98837280273438,
      "learning_rate": 4.219039480792822e-06,
      "loss": 4.4576,
      "step": 17471
    },
    {
      "epoch": 0.7053693984658862,
      "grad_norm": 83.3315658569336,
      "learning_rate": 4.217972585514095e-06,
      "loss": 4.7973,
      "step": 17472
    },
    {
      "epoch": 0.7054097698829229,
      "grad_norm": 80.12334442138672,
      "learning_rate": 4.216905789094537e-06,
      "loss": 4.7664,
      "step": 17473
    },
    {
      "epoch": 0.7054501412999596,
      "grad_norm": 94.27526092529297,
      "learning_rate": 4.215839091552384e-06,
      "loss": 4.3019,
      "step": 17474
    },
    {
      "epoch": 0.7054905127169964,
      "grad_norm": 117.60560607910156,
      "learning_rate": 4.214772492905874e-06,
      "loss": 4.7864,
      "step": 17475
    },
    {
      "epoch": 0.7055308841340331,
      "grad_norm": 96.02710723876953,
      "learning_rate": 4.213705993173246e-06,
      "loss": 4.8532,
      "step": 17476
    },
    {
      "epoch": 0.7055712555510698,
      "grad_norm": 128.96383666992188,
      "learning_rate": 4.212639592372737e-06,
      "loss": 5.6936,
      "step": 17477
    },
    {
      "epoch": 0.7056116269681065,
      "grad_norm": 87.41405487060547,
      "learning_rate": 4.211573290522573e-06,
      "loss": 5.0684,
      "step": 17478
    },
    {
      "epoch": 0.7056519983851434,
      "grad_norm": 80.17459869384766,
      "learning_rate": 4.2105070876409885e-06,
      "loss": 4.5539,
      "step": 17479
    },
    {
      "epoch": 0.7056923698021801,
      "grad_norm": 96.97879791259766,
      "learning_rate": 4.209440983746212e-06,
      "loss": 6.1138,
      "step": 17480
    },
    {
      "epoch": 0.7057327412192168,
      "grad_norm": 125.90375518798828,
      "learning_rate": 4.208374978856479e-06,
      "loss": 5.6762,
      "step": 17481
    },
    {
      "epoch": 0.7057731126362535,
      "grad_norm": 96.83435821533203,
      "learning_rate": 4.207309072990004e-06,
      "loss": 2.1961,
      "step": 17482
    },
    {
      "epoch": 0.7058134840532903,
      "grad_norm": 74.71318817138672,
      "learning_rate": 4.206243266165018e-06,
      "loss": 5.0358,
      "step": 17483
    },
    {
      "epoch": 0.705853855470327,
      "grad_norm": 81.80601501464844,
      "learning_rate": 4.205177558399741e-06,
      "loss": 4.9075,
      "step": 17484
    },
    {
      "epoch": 0.7058942268873637,
      "grad_norm": 81.85638427734375,
      "learning_rate": 4.204111949712397e-06,
      "loss": 5.9569,
      "step": 17485
    },
    {
      "epoch": 0.7059345983044005,
      "grad_norm": 77.87214660644531,
      "learning_rate": 4.203046440121205e-06,
      "loss": 3.9884,
      "step": 17486
    },
    {
      "epoch": 0.7059749697214373,
      "grad_norm": 129.4368133544922,
      "learning_rate": 4.201981029644381e-06,
      "loss": 4.1269,
      "step": 17487
    },
    {
      "epoch": 0.706015341138474,
      "grad_norm": 102.57234954833984,
      "learning_rate": 4.200915718300142e-06,
      "loss": 5.0744,
      "step": 17488
    },
    {
      "epoch": 0.7060557125555107,
      "grad_norm": 106.99811553955078,
      "learning_rate": 4.1998505061067076e-06,
      "loss": 5.6557,
      "step": 17489
    },
    {
      "epoch": 0.7060960839725474,
      "grad_norm": 86.83851623535156,
      "learning_rate": 4.19878539308228e-06,
      "loss": 5.2999,
      "step": 17490
    },
    {
      "epoch": 0.7061364553895841,
      "grad_norm": 116.98324584960938,
      "learning_rate": 4.197720379245076e-06,
      "loss": 3.4035,
      "step": 17491
    },
    {
      "epoch": 0.7061768268066209,
      "grad_norm": 65.93125915527344,
      "learning_rate": 4.196655464613304e-06,
      "loss": 4.7604,
      "step": 17492
    },
    {
      "epoch": 0.7062171982236577,
      "grad_norm": 89.632080078125,
      "learning_rate": 4.195590649205175e-06,
      "loss": 3.3646,
      "step": 17493
    },
    {
      "epoch": 0.7062575696406944,
      "grad_norm": 85.40963745117188,
      "learning_rate": 4.194525933038889e-06,
      "loss": 4.1835,
      "step": 17494
    },
    {
      "epoch": 0.7062979410577311,
      "grad_norm": 93.09020233154297,
      "learning_rate": 4.193461316132652e-06,
      "loss": 4.3864,
      "step": 17495
    },
    {
      "epoch": 0.7063383124747679,
      "grad_norm": 79.91688537597656,
      "learning_rate": 4.192396798504667e-06,
      "loss": 5.0984,
      "step": 17496
    },
    {
      "epoch": 0.7063786838918046,
      "grad_norm": 106.81500244140625,
      "learning_rate": 4.191332380173139e-06,
      "loss": 5.1146,
      "step": 17497
    },
    {
      "epoch": 0.7064190553088413,
      "grad_norm": 75.9444808959961,
      "learning_rate": 4.190268061156259e-06,
      "loss": 4.7669,
      "step": 17498
    },
    {
      "epoch": 0.706459426725878,
      "grad_norm": 127.94029235839844,
      "learning_rate": 4.189203841472229e-06,
      "loss": 5.3254,
      "step": 17499
    },
    {
      "epoch": 0.7064997981429149,
      "grad_norm": 98.27606964111328,
      "learning_rate": 4.188139721139245e-06,
      "loss": 5.5411,
      "step": 17500
    },
    {
      "epoch": 0.7065401695599516,
      "grad_norm": 78.71244812011719,
      "learning_rate": 4.187075700175502e-06,
      "loss": 5.4582,
      "step": 17501
    },
    {
      "epoch": 0.7065805409769883,
      "grad_norm": 117.34400177001953,
      "learning_rate": 4.186011778599188e-06,
      "loss": 4.9744,
      "step": 17502
    },
    {
      "epoch": 0.706620912394025,
      "grad_norm": 81.65499114990234,
      "learning_rate": 4.184947956428496e-06,
      "loss": 4.5962,
      "step": 17503
    },
    {
      "epoch": 0.7066612838110617,
      "grad_norm": 122.167236328125,
      "learning_rate": 4.183884233681614e-06,
      "loss": 5.4354,
      "step": 17504
    },
    {
      "epoch": 0.7067016552280985,
      "grad_norm": 131.0602569580078,
      "learning_rate": 4.18282061037673e-06,
      "loss": 6.4026,
      "step": 17505
    },
    {
      "epoch": 0.7067420266451352,
      "grad_norm": 142.65940856933594,
      "learning_rate": 4.18175708653203e-06,
      "loss": 5.4271,
      "step": 17506
    },
    {
      "epoch": 0.706782398062172,
      "grad_norm": 135.2217254638672,
      "learning_rate": 4.180693662165697e-06,
      "loss": 4.3919,
      "step": 17507
    },
    {
      "epoch": 0.7068227694792087,
      "grad_norm": 88.064453125,
      "learning_rate": 4.1796303372959135e-06,
      "loss": 4.3642,
      "step": 17508
    },
    {
      "epoch": 0.7068631408962455,
      "grad_norm": 100.38252258300781,
      "learning_rate": 4.178567111940863e-06,
      "loss": 4.6371,
      "step": 17509
    },
    {
      "epoch": 0.7069035123132822,
      "grad_norm": 95.37480926513672,
      "learning_rate": 4.1775039861187185e-06,
      "loss": 3.409,
      "step": 17510
    },
    {
      "epoch": 0.7069438837303189,
      "grad_norm": 107.15955352783203,
      "learning_rate": 4.176440959847658e-06,
      "loss": 5.8724,
      "step": 17511
    },
    {
      "epoch": 0.7069842551473556,
      "grad_norm": 107.84090423583984,
      "learning_rate": 4.175378033145858e-06,
      "loss": 4.2889,
      "step": 17512
    },
    {
      "epoch": 0.7070246265643924,
      "grad_norm": 89.7055892944336,
      "learning_rate": 4.174315206031496e-06,
      "loss": 6.1332,
      "step": 17513
    },
    {
      "epoch": 0.7070649979814292,
      "grad_norm": 90.84703063964844,
      "learning_rate": 4.173252478522737e-06,
      "loss": 5.3308,
      "step": 17514
    },
    {
      "epoch": 0.7071053693984659,
      "grad_norm": 96.19097900390625,
      "learning_rate": 4.172189850637753e-06,
      "loss": 5.5451,
      "step": 17515
    },
    {
      "epoch": 0.7071457408155026,
      "grad_norm": 104.92955017089844,
      "learning_rate": 4.171127322394714e-06,
      "loss": 3.204,
      "step": 17516
    },
    {
      "epoch": 0.7071861122325394,
      "grad_norm": 77.59272766113281,
      "learning_rate": 4.170064893811792e-06,
      "loss": 3.4252,
      "step": 17517
    },
    {
      "epoch": 0.7072264836495761,
      "grad_norm": 84.51937866210938,
      "learning_rate": 4.169002564907141e-06,
      "loss": 5.9148,
      "step": 17518
    },
    {
      "epoch": 0.7072668550666128,
      "grad_norm": 119.82936096191406,
      "learning_rate": 4.167940335698929e-06,
      "loss": 6.3672,
      "step": 17519
    },
    {
      "epoch": 0.7073072264836496,
      "grad_norm": 115.05181121826172,
      "learning_rate": 4.166878206205319e-06,
      "loss": 6.4565,
      "step": 17520
    },
    {
      "epoch": 0.7073475979006864,
      "grad_norm": 80.13578796386719,
      "learning_rate": 4.165816176444475e-06,
      "loss": 5.0689,
      "step": 17521
    },
    {
      "epoch": 0.7073879693177231,
      "grad_norm": 123.82413482666016,
      "learning_rate": 4.164754246434547e-06,
      "loss": 4.5048,
      "step": 17522
    },
    {
      "epoch": 0.7074283407347598,
      "grad_norm": 81.0962142944336,
      "learning_rate": 4.1636924161936945e-06,
      "loss": 4.5495,
      "step": 17523
    },
    {
      "epoch": 0.7074687121517965,
      "grad_norm": 113.12208557128906,
      "learning_rate": 4.162630685740072e-06,
      "loss": 3.8848,
      "step": 17524
    },
    {
      "epoch": 0.7075090835688332,
      "grad_norm": 97.212890625,
      "learning_rate": 4.1615690550918355e-06,
      "loss": 4.687,
      "step": 17525
    },
    {
      "epoch": 0.70754945498587,
      "grad_norm": 141.06985473632812,
      "learning_rate": 4.160507524267132e-06,
      "loss": 4.4284,
      "step": 17526
    },
    {
      "epoch": 0.7075898264029068,
      "grad_norm": 97.2698974609375,
      "learning_rate": 4.159446093284115e-06,
      "loss": 6.8471,
      "step": 17527
    },
    {
      "epoch": 0.7076301978199435,
      "grad_norm": 148.08131408691406,
      "learning_rate": 4.158384762160932e-06,
      "loss": 5.9233,
      "step": 17528
    },
    {
      "epoch": 0.7076705692369802,
      "grad_norm": 115.52387237548828,
      "learning_rate": 4.157323530915731e-06,
      "loss": 5.0836,
      "step": 17529
    },
    {
      "epoch": 0.707710940654017,
      "grad_norm": 156.4593048095703,
      "learning_rate": 4.156262399566651e-06,
      "loss": 5.9448,
      "step": 17530
    },
    {
      "epoch": 0.7077513120710537,
      "grad_norm": 74.50350189208984,
      "learning_rate": 4.155201368131837e-06,
      "loss": 6.803,
      "step": 17531
    },
    {
      "epoch": 0.7077916834880904,
      "grad_norm": 80.03437042236328,
      "learning_rate": 4.154140436629431e-06,
      "loss": 5.7784,
      "step": 17532
    },
    {
      "epoch": 0.7078320549051271,
      "grad_norm": 94.16007995605469,
      "learning_rate": 4.153079605077577e-06,
      "loss": 4.4381,
      "step": 17533
    },
    {
      "epoch": 0.707872426322164,
      "grad_norm": 91.4002685546875,
      "learning_rate": 4.152018873494403e-06,
      "loss": 4.5753,
      "step": 17534
    },
    {
      "epoch": 0.7079127977392007,
      "grad_norm": 101.22566223144531,
      "learning_rate": 4.1509582418980525e-06,
      "loss": 5.1082,
      "step": 17535
    },
    {
      "epoch": 0.7079531691562374,
      "grad_norm": 138.13206481933594,
      "learning_rate": 4.149897710306656e-06,
      "loss": 4.5437,
      "step": 17536
    },
    {
      "epoch": 0.7079935405732741,
      "grad_norm": 284.2968444824219,
      "learning_rate": 4.148837278738353e-06,
      "loss": 5.4749,
      "step": 17537
    },
    {
      "epoch": 0.7080339119903108,
      "grad_norm": 93.25231170654297,
      "learning_rate": 4.1477769472112625e-06,
      "loss": 5.5989,
      "step": 17538
    },
    {
      "epoch": 0.7080742834073476,
      "grad_norm": 100.62871551513672,
      "learning_rate": 4.146716715743523e-06,
      "loss": 6.5472,
      "step": 17539
    },
    {
      "epoch": 0.7081146548243843,
      "grad_norm": 110.58538818359375,
      "learning_rate": 4.1456565843532585e-06,
      "loss": 4.5412,
      "step": 17540
    },
    {
      "epoch": 0.7081550262414211,
      "grad_norm": 149.87258911132812,
      "learning_rate": 4.144596553058598e-06,
      "loss": 3.7173,
      "step": 17541
    },
    {
      "epoch": 0.7081953976584578,
      "grad_norm": 144.853515625,
      "learning_rate": 4.143536621877661e-06,
      "loss": 5.0949,
      "step": 17542
    },
    {
      "epoch": 0.7082357690754946,
      "grad_norm": 98.54685974121094,
      "learning_rate": 4.1424767908285715e-06,
      "loss": 5.5291,
      "step": 17543
    },
    {
      "epoch": 0.7082761404925313,
      "grad_norm": 115.67800903320312,
      "learning_rate": 4.14141705992945e-06,
      "loss": 5.2667,
      "step": 17544
    },
    {
      "epoch": 0.708316511909568,
      "grad_norm": 120.53949737548828,
      "learning_rate": 4.1403574291984165e-06,
      "loss": 5.0303,
      "step": 17545
    },
    {
      "epoch": 0.7083568833266047,
      "grad_norm": 117.24689483642578,
      "learning_rate": 4.139297898653587e-06,
      "loss": 4.5481,
      "step": 17546
    },
    {
      "epoch": 0.7083972547436415,
      "grad_norm": 169.16659545898438,
      "learning_rate": 4.1382384683130784e-06,
      "loss": 4.0328,
      "step": 17547
    },
    {
      "epoch": 0.7084376261606783,
      "grad_norm": 73.21780395507812,
      "learning_rate": 4.137179138195003e-06,
      "loss": 4.8013,
      "step": 17548
    },
    {
      "epoch": 0.708477997577715,
      "grad_norm": 91.03184509277344,
      "learning_rate": 4.136119908317479e-06,
      "loss": 6.231,
      "step": 17549
    },
    {
      "epoch": 0.7085183689947517,
      "grad_norm": 115.31707000732422,
      "learning_rate": 4.135060778698606e-06,
      "loss": 4.565,
      "step": 17550
    },
    {
      "epoch": 0.7085587404117885,
      "grad_norm": 73.20504760742188,
      "learning_rate": 4.134001749356498e-06,
      "loss": 4.6081,
      "step": 17551
    },
    {
      "epoch": 0.7085991118288252,
      "grad_norm": 135.38369750976562,
      "learning_rate": 4.132942820309263e-06,
      "loss": 5.523,
      "step": 17552
    },
    {
      "epoch": 0.7086394832458619,
      "grad_norm": 99.91130065917969,
      "learning_rate": 4.131883991575009e-06,
      "loss": 4.2715,
      "step": 17553
    },
    {
      "epoch": 0.7086798546628986,
      "grad_norm": 84.94727325439453,
      "learning_rate": 4.130825263171833e-06,
      "loss": 3.3025,
      "step": 17554
    },
    {
      "epoch": 0.7087202260799355,
      "grad_norm": 135.15521240234375,
      "learning_rate": 4.1297666351178376e-06,
      "loss": 5.7744,
      "step": 17555
    },
    {
      "epoch": 0.7087605974969722,
      "grad_norm": 96.07112884521484,
      "learning_rate": 4.128708107431125e-06,
      "loss": 4.375,
      "step": 17556
    },
    {
      "epoch": 0.7088009689140089,
      "grad_norm": 82.16055297851562,
      "learning_rate": 4.127649680129798e-06,
      "loss": 3.685,
      "step": 17557
    },
    {
      "epoch": 0.7088413403310456,
      "grad_norm": 82.45211791992188,
      "learning_rate": 4.126591353231945e-06,
      "loss": 2.554,
      "step": 17558
    },
    {
      "epoch": 0.7088817117480823,
      "grad_norm": 132.78106689453125,
      "learning_rate": 4.125533126755663e-06,
      "loss": 5.7054,
      "step": 17559
    },
    {
      "epoch": 0.7089220831651191,
      "grad_norm": 132.6204071044922,
      "learning_rate": 4.124475000719047e-06,
      "loss": 5.6545,
      "step": 17560
    },
    {
      "epoch": 0.7089624545821558,
      "grad_norm": 87.21307373046875,
      "learning_rate": 4.123416975140192e-06,
      "loss": 3.9036,
      "step": 17561
    },
    {
      "epoch": 0.7090028259991926,
      "grad_norm": 94.49993896484375,
      "learning_rate": 4.122359050037181e-06,
      "loss": 3.4687,
      "step": 17562
    },
    {
      "epoch": 0.7090431974162293,
      "grad_norm": 88.75548553466797,
      "learning_rate": 4.1213012254281035e-06,
      "loss": 5.439,
      "step": 17563
    },
    {
      "epoch": 0.7090835688332661,
      "grad_norm": 87.84976196289062,
      "learning_rate": 4.120243501331049e-06,
      "loss": 5.1697,
      "step": 17564
    },
    {
      "epoch": 0.7091239402503028,
      "grad_norm": 129.95314025878906,
      "learning_rate": 4.119185877764099e-06,
      "loss": 5.6324,
      "step": 17565
    },
    {
      "epoch": 0.7091643116673395,
      "grad_norm": 74.14359283447266,
      "learning_rate": 4.118128354745339e-06,
      "loss": 6.0849,
      "step": 17566
    },
    {
      "epoch": 0.7092046830843762,
      "grad_norm": 70.04253387451172,
      "learning_rate": 4.117070932292848e-06,
      "loss": 5.5982,
      "step": 17567
    },
    {
      "epoch": 0.709245054501413,
      "grad_norm": 173.92872619628906,
      "learning_rate": 4.1160136104247075e-06,
      "loss": 3.9494,
      "step": 17568
    },
    {
      "epoch": 0.7092854259184498,
      "grad_norm": 116.45716094970703,
      "learning_rate": 4.114956389158993e-06,
      "loss": 5.3139,
      "step": 17569
    },
    {
      "epoch": 0.7093257973354865,
      "grad_norm": 138.10923767089844,
      "learning_rate": 4.113899268513786e-06,
      "loss": 5.686,
      "step": 17570
    },
    {
      "epoch": 0.7093661687525232,
      "grad_norm": 138.29119873046875,
      "learning_rate": 4.112842248507153e-06,
      "loss": 4.2822,
      "step": 17571
    },
    {
      "epoch": 0.70940654016956,
      "grad_norm": 12095.8251953125,
      "learning_rate": 4.111785329157169e-06,
      "loss": 4.4854,
      "step": 17572
    },
    {
      "epoch": 0.7094469115865967,
      "grad_norm": 103.19252014160156,
      "learning_rate": 4.110728510481907e-06,
      "loss": 3.8358,
      "step": 17573
    },
    {
      "epoch": 0.7094872830036334,
      "grad_norm": 142.07388305664062,
      "learning_rate": 4.109671792499438e-06,
      "loss": 3.4731,
      "step": 17574
    },
    {
      "epoch": 0.7095276544206701,
      "grad_norm": 82.57913208007812,
      "learning_rate": 4.1086151752278235e-06,
      "loss": 6.8291,
      "step": 17575
    },
    {
      "epoch": 0.709568025837707,
      "grad_norm": 108.90830993652344,
      "learning_rate": 4.107558658685131e-06,
      "loss": 5.33,
      "step": 17576
    },
    {
      "epoch": 0.7096083972547437,
      "grad_norm": 74.37932586669922,
      "learning_rate": 4.106502242889428e-06,
      "loss": 4.9489,
      "step": 17577
    },
    {
      "epoch": 0.7096487686717804,
      "grad_norm": 99.35686492919922,
      "learning_rate": 4.105445927858775e-06,
      "loss": 4.5881,
      "step": 17578
    },
    {
      "epoch": 0.7096891400888171,
      "grad_norm": 133.55775451660156,
      "learning_rate": 4.1043897136112294e-06,
      "loss": 4.519,
      "step": 17579
    },
    {
      "epoch": 0.7097295115058538,
      "grad_norm": 122.25548553466797,
      "learning_rate": 4.103333600164853e-06,
      "loss": 5.2459,
      "step": 17580
    },
    {
      "epoch": 0.7097698829228906,
      "grad_norm": 100.9704360961914,
      "learning_rate": 4.102277587537702e-06,
      "loss": 4.6638,
      "step": 17581
    },
    {
      "epoch": 0.7098102543399273,
      "grad_norm": 107.29340362548828,
      "learning_rate": 4.101221675747835e-06,
      "loss": 5.4598,
      "step": 17582
    },
    {
      "epoch": 0.7098506257569641,
      "grad_norm": 103.0799560546875,
      "learning_rate": 4.1001658648133e-06,
      "loss": 4.8946,
      "step": 17583
    },
    {
      "epoch": 0.7098909971740008,
      "grad_norm": 71.99356079101562,
      "learning_rate": 4.099110154752151e-06,
      "loss": 4.3227,
      "step": 17584
    },
    {
      "epoch": 0.7099313685910376,
      "grad_norm": 182.5372772216797,
      "learning_rate": 4.09805454558244e-06,
      "loss": 6.3838,
      "step": 17585
    },
    {
      "epoch": 0.7099717400080743,
      "grad_norm": 128.89308166503906,
      "learning_rate": 4.096999037322213e-06,
      "loss": 3.8146,
      "step": 17586
    },
    {
      "epoch": 0.710012111425111,
      "grad_norm": 106.49101257324219,
      "learning_rate": 4.095943629989518e-06,
      "loss": 5.1706,
      "step": 17587
    },
    {
      "epoch": 0.7100524828421477,
      "grad_norm": 98.44068145751953,
      "learning_rate": 4.094888323602401e-06,
      "loss": 3.5421,
      "step": 17588
    },
    {
      "epoch": 0.7100928542591844,
      "grad_norm": 80.65339660644531,
      "learning_rate": 4.093833118178902e-06,
      "loss": 5.6726,
      "step": 17589
    },
    {
      "epoch": 0.7101332256762213,
      "grad_norm": 123.28880310058594,
      "learning_rate": 4.092778013737071e-06,
      "loss": 3.5168,
      "step": 17590
    },
    {
      "epoch": 0.710173597093258,
      "grad_norm": 104.37599182128906,
      "learning_rate": 4.091723010294936e-06,
      "loss": 4.2614,
      "step": 17591
    },
    {
      "epoch": 0.7102139685102947,
      "grad_norm": 219.46353149414062,
      "learning_rate": 4.0906681078705424e-06,
      "loss": 5.7801,
      "step": 17592
    },
    {
      "epoch": 0.7102543399273314,
      "grad_norm": 103.10688018798828,
      "learning_rate": 4.089613306481923e-06,
      "loss": 5.5875,
      "step": 17593
    },
    {
      "epoch": 0.7102947113443682,
      "grad_norm": 93.71278381347656,
      "learning_rate": 4.0885586061471196e-06,
      "loss": 5.9018,
      "step": 17594
    },
    {
      "epoch": 0.7103350827614049,
      "grad_norm": 72.39855194091797,
      "learning_rate": 4.087504006884156e-06,
      "loss": 3.6523,
      "step": 17595
    },
    {
      "epoch": 0.7103754541784416,
      "grad_norm": 121.38134002685547,
      "learning_rate": 4.086449508711066e-06,
      "loss": 5.521,
      "step": 17596
    },
    {
      "epoch": 0.7104158255954784,
      "grad_norm": 123.9483871459961,
      "learning_rate": 4.085395111645881e-06,
      "loss": 5.2989,
      "step": 17597
    },
    {
      "epoch": 0.7104561970125152,
      "grad_norm": 74.36858367919922,
      "learning_rate": 4.0843408157066325e-06,
      "loss": 4.3265,
      "step": 17598
    },
    {
      "epoch": 0.7104965684295519,
      "grad_norm": 78.47185516357422,
      "learning_rate": 4.083286620911338e-06,
      "loss": 5.8805,
      "step": 17599
    },
    {
      "epoch": 0.7105369398465886,
      "grad_norm": 100.0918960571289,
      "learning_rate": 4.082232527278025e-06,
      "loss": 3.0596,
      "step": 17600
    },
    {
      "epoch": 0.7105773112636253,
      "grad_norm": 114.51181030273438,
      "learning_rate": 4.0811785348247164e-06,
      "loss": 5.2581,
      "step": 17601
    },
    {
      "epoch": 0.710617682680662,
      "grad_norm": 115.06060791015625,
      "learning_rate": 4.080124643569439e-06,
      "loss": 4.7248,
      "step": 17602
    },
    {
      "epoch": 0.7106580540976988,
      "grad_norm": 70.6260986328125,
      "learning_rate": 4.079070853530201e-06,
      "loss": 4.0108,
      "step": 17603
    },
    {
      "epoch": 0.7106984255147356,
      "grad_norm": 83.93367004394531,
      "learning_rate": 4.078017164725025e-06,
      "loss": 5.9037,
      "step": 17604
    },
    {
      "epoch": 0.7107387969317723,
      "grad_norm": 110.96695709228516,
      "learning_rate": 4.076963577171928e-06,
      "loss": 3.2993,
      "step": 17605
    },
    {
      "epoch": 0.710779168348809,
      "grad_norm": 130.2057647705078,
      "learning_rate": 4.0759100908889215e-06,
      "loss": 3.9629,
      "step": 17606
    },
    {
      "epoch": 0.7108195397658458,
      "grad_norm": 75.57747650146484,
      "learning_rate": 4.074856705894018e-06,
      "loss": 4.4964,
      "step": 17607
    },
    {
      "epoch": 0.7108599111828825,
      "grad_norm": 95.34211730957031,
      "learning_rate": 4.07380342220523e-06,
      "loss": 4.5581,
      "step": 17608
    },
    {
      "epoch": 0.7109002825999192,
      "grad_norm": 63.10707092285156,
      "learning_rate": 4.072750239840564e-06,
      "loss": 4.3602,
      "step": 17609
    },
    {
      "epoch": 0.7109406540169559,
      "grad_norm": 133.83181762695312,
      "learning_rate": 4.071697158818032e-06,
      "loss": 5.4044,
      "step": 17610
    },
    {
      "epoch": 0.7109810254339928,
      "grad_norm": 87.79003143310547,
      "learning_rate": 4.07064417915563e-06,
      "loss": 5.5576,
      "step": 17611
    },
    {
      "epoch": 0.7110213968510295,
      "grad_norm": 157.00015258789062,
      "learning_rate": 4.069591300871368e-06,
      "loss": 6.2337,
      "step": 17612
    },
    {
      "epoch": 0.7110617682680662,
      "grad_norm": 97.46466064453125,
      "learning_rate": 4.068538523983246e-06,
      "loss": 5.3595,
      "step": 17613
    },
    {
      "epoch": 0.7111021396851029,
      "grad_norm": 107.0324935913086,
      "learning_rate": 4.067485848509268e-06,
      "loss": 5.6413,
      "step": 17614
    },
    {
      "epoch": 0.7111425111021397,
      "grad_norm": 81.27306365966797,
      "learning_rate": 4.066433274467425e-06,
      "loss": 5.2834,
      "step": 17615
    },
    {
      "epoch": 0.7111828825191764,
      "grad_norm": 100.81019592285156,
      "learning_rate": 4.065380801875717e-06,
      "loss": 5.6386,
      "step": 17616
    },
    {
      "epoch": 0.7112232539362131,
      "grad_norm": 121.64298248291016,
      "learning_rate": 4.064328430752139e-06,
      "loss": 5.8981,
      "step": 17617
    },
    {
      "epoch": 0.7112636253532499,
      "grad_norm": 112.16523742675781,
      "learning_rate": 4.063276161114687e-06,
      "loss": 3.1265,
      "step": 17618
    },
    {
      "epoch": 0.7113039967702867,
      "grad_norm": 95.49142456054688,
      "learning_rate": 4.062223992981347e-06,
      "loss": 4.1356,
      "step": 17619
    },
    {
      "epoch": 0.7113443681873234,
      "grad_norm": 98.86456298828125,
      "learning_rate": 4.0611719263701096e-06,
      "loss": 5.2276,
      "step": 17620
    },
    {
      "epoch": 0.7113847396043601,
      "grad_norm": 97.20356750488281,
      "learning_rate": 4.060119961298965e-06,
      "loss": 4.1424,
      "step": 17621
    },
    {
      "epoch": 0.7114251110213968,
      "grad_norm": 115.22003936767578,
      "learning_rate": 4.059068097785898e-06,
      "loss": 5.263,
      "step": 17622
    },
    {
      "epoch": 0.7114654824384335,
      "grad_norm": 115.24889373779297,
      "learning_rate": 4.058016335848898e-06,
      "loss": 4.1674,
      "step": 17623
    },
    {
      "epoch": 0.7115058538554703,
      "grad_norm": 99.23304748535156,
      "learning_rate": 4.056964675505938e-06,
      "loss": 4.5424,
      "step": 17624
    },
    {
      "epoch": 0.7115462252725071,
      "grad_norm": 125.79566192626953,
      "learning_rate": 4.055913116775006e-06,
      "loss": 5.1807,
      "step": 17625
    },
    {
      "epoch": 0.7115865966895438,
      "grad_norm": 129.15316772460938,
      "learning_rate": 4.054861659674077e-06,
      "loss": 6.726,
      "step": 17626
    },
    {
      "epoch": 0.7116269681065805,
      "grad_norm": 103.002197265625,
      "learning_rate": 4.053810304221132e-06,
      "loss": 4.818,
      "step": 17627
    },
    {
      "epoch": 0.7116673395236173,
      "grad_norm": 71.6264877319336,
      "learning_rate": 4.052759050434145e-06,
      "loss": 4.2174,
      "step": 17628
    },
    {
      "epoch": 0.711707710940654,
      "grad_norm": 100.41856384277344,
      "learning_rate": 4.051707898331091e-06,
      "loss": 4.6771,
      "step": 17629
    },
    {
      "epoch": 0.7117480823576907,
      "grad_norm": 165.56382751464844,
      "learning_rate": 4.050656847929945e-06,
      "loss": 5.7322,
      "step": 17630
    },
    {
      "epoch": 0.7117884537747275,
      "grad_norm": 97.22734069824219,
      "learning_rate": 4.049605899248671e-06,
      "loss": 5.6996,
      "step": 17631
    },
    {
      "epoch": 0.7118288251917643,
      "grad_norm": 86.46915435791016,
      "learning_rate": 4.04855505230524e-06,
      "loss": 5.7783,
      "step": 17632
    },
    {
      "epoch": 0.711869196608801,
      "grad_norm": 100.5543441772461,
      "learning_rate": 4.04750430711762e-06,
      "loss": 6.0165,
      "step": 17633
    },
    {
      "epoch": 0.7119095680258377,
      "grad_norm": 84.17948150634766,
      "learning_rate": 4.046453663703781e-06,
      "loss": 5.9443,
      "step": 17634
    },
    {
      "epoch": 0.7119499394428744,
      "grad_norm": 114.75892639160156,
      "learning_rate": 4.045403122081677e-06,
      "loss": 4.9618,
      "step": 17635
    },
    {
      "epoch": 0.7119903108599112,
      "grad_norm": 154.6898651123047,
      "learning_rate": 4.044352682269276e-06,
      "loss": 5.4153,
      "step": 17636
    },
    {
      "epoch": 0.7120306822769479,
      "grad_norm": 81.45439147949219,
      "learning_rate": 4.043302344284535e-06,
      "loss": 3.9258,
      "step": 17637
    },
    {
      "epoch": 0.7120710536939847,
      "grad_norm": 105.97073364257812,
      "learning_rate": 4.042252108145419e-06,
      "loss": 4.9699,
      "step": 17638
    },
    {
      "epoch": 0.7121114251110214,
      "grad_norm": 93.28255462646484,
      "learning_rate": 4.041201973869876e-06,
      "loss": 4.021,
      "step": 17639
    },
    {
      "epoch": 0.7121517965280582,
      "grad_norm": 124.29417419433594,
      "learning_rate": 4.040151941475864e-06,
      "loss": 3.7702,
      "step": 17640
    },
    {
      "epoch": 0.7121921679450949,
      "grad_norm": 73.16331481933594,
      "learning_rate": 4.039102010981336e-06,
      "loss": 4.4762,
      "step": 17641
    },
    {
      "epoch": 0.7122325393621316,
      "grad_norm": 101.87493133544922,
      "learning_rate": 4.0380521824042455e-06,
      "loss": 5.4454,
      "step": 17642
    },
    {
      "epoch": 0.7122729107791683,
      "grad_norm": 332.4655456542969,
      "learning_rate": 4.037002455762539e-06,
      "loss": 5.3125,
      "step": 17643
    },
    {
      "epoch": 0.712313282196205,
      "grad_norm": 89.48041534423828,
      "learning_rate": 4.035952831074166e-06,
      "loss": 5.6313,
      "step": 17644
    },
    {
      "epoch": 0.7123536536132419,
      "grad_norm": 155.9021453857422,
      "learning_rate": 4.034903308357073e-06,
      "loss": 4.0934,
      "step": 17645
    },
    {
      "epoch": 0.7123940250302786,
      "grad_norm": 61.97191619873047,
      "learning_rate": 4.033853887629208e-06,
      "loss": 5.0587,
      "step": 17646
    },
    {
      "epoch": 0.7124343964473153,
      "grad_norm": 91.28559112548828,
      "learning_rate": 4.032804568908505e-06,
      "loss": 5.5622,
      "step": 17647
    },
    {
      "epoch": 0.712474767864352,
      "grad_norm": 91.52850341796875,
      "learning_rate": 4.03175535221291e-06,
      "loss": 4.2515,
      "step": 17648
    },
    {
      "epoch": 0.7125151392813888,
      "grad_norm": 71.40499877929688,
      "learning_rate": 4.030706237560361e-06,
      "loss": 5.386,
      "step": 17649
    },
    {
      "epoch": 0.7125555106984255,
      "grad_norm": 116.14238739013672,
      "learning_rate": 4.0296572249688e-06,
      "loss": 4.462,
      "step": 17650
    },
    {
      "epoch": 0.7125958821154622,
      "grad_norm": 82.24763488769531,
      "learning_rate": 4.028608314456156e-06,
      "loss": 4.9704,
      "step": 17651
    },
    {
      "epoch": 0.712636253532499,
      "grad_norm": 129.90362548828125,
      "learning_rate": 4.027559506040365e-06,
      "loss": 4.8498,
      "step": 17652
    },
    {
      "epoch": 0.7126766249495358,
      "grad_norm": 90.01744079589844,
      "learning_rate": 4.026510799739359e-06,
      "loss": 3.9173,
      "step": 17653
    },
    {
      "epoch": 0.7127169963665725,
      "grad_norm": 78.59066009521484,
      "learning_rate": 4.025462195571074e-06,
      "loss": 4.8578,
      "step": 17654
    },
    {
      "epoch": 0.7127573677836092,
      "grad_norm": 90.84062957763672,
      "learning_rate": 4.0244136935534305e-06,
      "loss": 4.9365,
      "step": 17655
    },
    {
      "epoch": 0.7127977392006459,
      "grad_norm": 111.04154205322266,
      "learning_rate": 4.0233652937043575e-06,
      "loss": 5.9506,
      "step": 17656
    },
    {
      "epoch": 0.7128381106176827,
      "grad_norm": 87.24419403076172,
      "learning_rate": 4.022316996041783e-06,
      "loss": 4.2758,
      "step": 17657
    },
    {
      "epoch": 0.7128784820347194,
      "grad_norm": 83.35517120361328,
      "learning_rate": 4.0212688005836305e-06,
      "loss": 4.9176,
      "step": 17658
    },
    {
      "epoch": 0.7129188534517562,
      "grad_norm": 85.35934448242188,
      "learning_rate": 4.020220707347819e-06,
      "loss": 5.4206,
      "step": 17659
    },
    {
      "epoch": 0.7129592248687929,
      "grad_norm": 83.54087829589844,
      "learning_rate": 4.0191727163522676e-06,
      "loss": 4.0414,
      "step": 17660
    },
    {
      "epoch": 0.7129995962858297,
      "grad_norm": 164.3092803955078,
      "learning_rate": 4.018124827614898e-06,
      "loss": 4.8873,
      "step": 17661
    },
    {
      "epoch": 0.7130399677028664,
      "grad_norm": 70.32485961914062,
      "learning_rate": 4.017077041153624e-06,
      "loss": 3.843,
      "step": 17662
    },
    {
      "epoch": 0.7130803391199031,
      "grad_norm": 94.51859283447266,
      "learning_rate": 4.016029356986363e-06,
      "loss": 4.532,
      "step": 17663
    },
    {
      "epoch": 0.7131207105369398,
      "grad_norm": 87.70446014404297,
      "learning_rate": 4.014981775131025e-06,
      "loss": 4.5841,
      "step": 17664
    },
    {
      "epoch": 0.7131610819539765,
      "grad_norm": 90.18306732177734,
      "learning_rate": 4.013934295605523e-06,
      "loss": 4.5634,
      "step": 17665
    },
    {
      "epoch": 0.7132014533710134,
      "grad_norm": 117.01422882080078,
      "learning_rate": 4.01288691842777e-06,
      "loss": 6.6207,
      "step": 17666
    },
    {
      "epoch": 0.7132418247880501,
      "grad_norm": 85.81262969970703,
      "learning_rate": 4.011839643615666e-06,
      "loss": 4.7625,
      "step": 17667
    },
    {
      "epoch": 0.7132821962050868,
      "grad_norm": 75.11327362060547,
      "learning_rate": 4.010792471187121e-06,
      "loss": 5.455,
      "step": 17668
    },
    {
      "epoch": 0.7133225676221235,
      "grad_norm": 91.42825317382812,
      "learning_rate": 4.009745401160038e-06,
      "loss": 3.9935,
      "step": 17669
    },
    {
      "epoch": 0.7133629390391603,
      "grad_norm": 202.37913513183594,
      "learning_rate": 4.008698433552324e-06,
      "loss": 4.6231,
      "step": 17670
    },
    {
      "epoch": 0.713403310456197,
      "grad_norm": 74.78121948242188,
      "learning_rate": 4.007651568381873e-06,
      "loss": 5.0318,
      "step": 17671
    },
    {
      "epoch": 0.7134436818732337,
      "grad_norm": 89.13544464111328,
      "learning_rate": 4.006604805666587e-06,
      "loss": 3.9064,
      "step": 17672
    },
    {
      "epoch": 0.7134840532902705,
      "grad_norm": 95.24861907958984,
      "learning_rate": 4.005558145424364e-06,
      "loss": 4.1423,
      "step": 17673
    },
    {
      "epoch": 0.7135244247073073,
      "grad_norm": 132.09100341796875,
      "learning_rate": 4.004511587673101e-06,
      "loss": 3.1557,
      "step": 17674
    },
    {
      "epoch": 0.713564796124344,
      "grad_norm": 77.58879852294922,
      "learning_rate": 4.0034651324306875e-06,
      "loss": 4.9907,
      "step": 17675
    },
    {
      "epoch": 0.7136051675413807,
      "grad_norm": 99.336181640625,
      "learning_rate": 4.002418779715016e-06,
      "loss": 4.576,
      "step": 17676
    },
    {
      "epoch": 0.7136455389584174,
      "grad_norm": 122.79950714111328,
      "learning_rate": 4.001372529543977e-06,
      "loss": 4.058,
      "step": 17677
    },
    {
      "epoch": 0.7136859103754541,
      "grad_norm": 227.36598205566406,
      "learning_rate": 4.000326381935465e-06,
      "loss": 3.6292,
      "step": 17678
    },
    {
      "epoch": 0.7137262817924909,
      "grad_norm": 333.4105224609375,
      "learning_rate": 3.9992803369073584e-06,
      "loss": 4.5684,
      "step": 17679
    },
    {
      "epoch": 0.7137666532095277,
      "grad_norm": 77.33755493164062,
      "learning_rate": 3.998234394477545e-06,
      "loss": 3.4321,
      "step": 17680
    },
    {
      "epoch": 0.7138070246265644,
      "grad_norm": 87.36589813232422,
      "learning_rate": 3.997188554663908e-06,
      "loss": 5.5219,
      "step": 17681
    },
    {
      "epoch": 0.7138473960436011,
      "grad_norm": 99.27922058105469,
      "learning_rate": 3.996142817484329e-06,
      "loss": 4.0867,
      "step": 17682
    },
    {
      "epoch": 0.7138877674606379,
      "grad_norm": 83.43024444580078,
      "learning_rate": 3.9950971829566875e-06,
      "loss": 3.516,
      "step": 17683
    },
    {
      "epoch": 0.7139281388776746,
      "grad_norm": 119.14126586914062,
      "learning_rate": 3.994051651098863e-06,
      "loss": 4.8042,
      "step": 17684
    },
    {
      "epoch": 0.7139685102947113,
      "grad_norm": 86.6948471069336,
      "learning_rate": 3.9930062219287294e-06,
      "loss": 3.6555,
      "step": 17685
    },
    {
      "epoch": 0.714008881711748,
      "grad_norm": 87.94772338867188,
      "learning_rate": 3.991960895464167e-06,
      "loss": 6.4483,
      "step": 17686
    },
    {
      "epoch": 0.7140492531287849,
      "grad_norm": 111.46044921875,
      "learning_rate": 3.990915671723038e-06,
      "loss": 5.7951,
      "step": 17687
    },
    {
      "epoch": 0.7140896245458216,
      "grad_norm": 105.7868881225586,
      "learning_rate": 3.989870550723219e-06,
      "loss": 5.6642,
      "step": 17688
    },
    {
      "epoch": 0.7141299959628583,
      "grad_norm": 105.67193603515625,
      "learning_rate": 3.98882553248258e-06,
      "loss": 3.6458,
      "step": 17689
    },
    {
      "epoch": 0.714170367379895,
      "grad_norm": 104.71162414550781,
      "learning_rate": 3.98778061701899e-06,
      "loss": 4.4187,
      "step": 17690
    },
    {
      "epoch": 0.7142107387969318,
      "grad_norm": 70.6385269165039,
      "learning_rate": 3.986735804350309e-06,
      "loss": 4.715,
      "step": 17691
    },
    {
      "epoch": 0.7142511102139685,
      "grad_norm": 195.2193603515625,
      "learning_rate": 3.985691094494404e-06,
      "loss": 3.7343,
      "step": 17692
    },
    {
      "epoch": 0.7142914816310052,
      "grad_norm": 120.46024322509766,
      "learning_rate": 3.984646487469136e-06,
      "loss": 3.6995,
      "step": 17693
    },
    {
      "epoch": 0.714331853048042,
      "grad_norm": 123.69707489013672,
      "learning_rate": 3.98360198329237e-06,
      "loss": 5.9952,
      "step": 17694
    },
    {
      "epoch": 0.7143722244650788,
      "grad_norm": 170.39523315429688,
      "learning_rate": 3.982557581981958e-06,
      "loss": 4.2332,
      "step": 17695
    },
    {
      "epoch": 0.7144125958821155,
      "grad_norm": 81.71182250976562,
      "learning_rate": 3.981513283555758e-06,
      "loss": 5.3055,
      "step": 17696
    },
    {
      "epoch": 0.7144529672991522,
      "grad_norm": 80.29853820800781,
      "learning_rate": 3.980469088031627e-06,
      "loss": 4.559,
      "step": 17697
    },
    {
      "epoch": 0.7144933387161889,
      "grad_norm": 111.36511993408203,
      "learning_rate": 3.9794249954274235e-06,
      "loss": 5.0528,
      "step": 17698
    },
    {
      "epoch": 0.7145337101332256,
      "grad_norm": 82.11096954345703,
      "learning_rate": 3.978381005760988e-06,
      "loss": 4.5837,
      "step": 17699
    },
    {
      "epoch": 0.7145740815502624,
      "grad_norm": 113.66324615478516,
      "learning_rate": 3.977337119050176e-06,
      "loss": 6.6173,
      "step": 17700
    },
    {
      "epoch": 0.7146144529672992,
      "grad_norm": 124.62158203125,
      "learning_rate": 3.976293335312836e-06,
      "loss": 3.7925,
      "step": 17701
    },
    {
      "epoch": 0.7146548243843359,
      "grad_norm": 118.66146850585938,
      "learning_rate": 3.975249654566813e-06,
      "loss": 5.1551,
      "step": 17702
    },
    {
      "epoch": 0.7146951958013726,
      "grad_norm": 73.44977569580078,
      "learning_rate": 3.974206076829953e-06,
      "loss": 6.5911,
      "step": 17703
    },
    {
      "epoch": 0.7147355672184094,
      "grad_norm": 74.81501007080078,
      "learning_rate": 3.973162602120097e-06,
      "loss": 6.3,
      "step": 17704
    },
    {
      "epoch": 0.7147759386354461,
      "grad_norm": 92.34711456298828,
      "learning_rate": 3.972119230455087e-06,
      "loss": 5.9145,
      "step": 17705
    },
    {
      "epoch": 0.7148163100524828,
      "grad_norm": 89.08169555664062,
      "learning_rate": 3.971075961852766e-06,
      "loss": 5.624,
      "step": 17706
    },
    {
      "epoch": 0.7148566814695195,
      "grad_norm": 103.75914001464844,
      "learning_rate": 3.970032796330964e-06,
      "loss": 5.3105,
      "step": 17707
    },
    {
      "epoch": 0.7148970528865564,
      "grad_norm": 96.46377563476562,
      "learning_rate": 3.96898973390752e-06,
      "loss": 4.9678,
      "step": 17708
    },
    {
      "epoch": 0.7149374243035931,
      "grad_norm": 117.93333435058594,
      "learning_rate": 3.967946774600268e-06,
      "loss": 5.681,
      "step": 17709
    },
    {
      "epoch": 0.7149777957206298,
      "grad_norm": 96.57735443115234,
      "learning_rate": 3.966903918427043e-06,
      "loss": 4.0288,
      "step": 17710
    },
    {
      "epoch": 0.7150181671376665,
      "grad_norm": 105.63084411621094,
      "learning_rate": 3.96586116540567e-06,
      "loss": 4.4605,
      "step": 17711
    },
    {
      "epoch": 0.7150585385547032,
      "grad_norm": 113.22299194335938,
      "learning_rate": 3.964818515553981e-06,
      "loss": 5.2943,
      "step": 17712
    },
    {
      "epoch": 0.71509890997174,
      "grad_norm": 96.94397735595703,
      "learning_rate": 3.9637759688898015e-06,
      "loss": 5.0315,
      "step": 17713
    },
    {
      "epoch": 0.7151392813887767,
      "grad_norm": 93.5164566040039,
      "learning_rate": 3.962733525430961e-06,
      "loss": 4.7351,
      "step": 17714
    },
    {
      "epoch": 0.7151796528058135,
      "grad_norm": 122.00526428222656,
      "learning_rate": 3.961691185195275e-06,
      "loss": 4.9988,
      "step": 17715
    },
    {
      "epoch": 0.7152200242228502,
      "grad_norm": 82.40245056152344,
      "learning_rate": 3.9606489482005706e-06,
      "loss": 5.5779,
      "step": 17716
    },
    {
      "epoch": 0.715260395639887,
      "grad_norm": 105.33287811279297,
      "learning_rate": 3.959606814464665e-06,
      "loss": 7.8593,
      "step": 17717
    },
    {
      "epoch": 0.7153007670569237,
      "grad_norm": 90.39852142333984,
      "learning_rate": 3.958564784005382e-06,
      "loss": 6.6207,
      "step": 17718
    },
    {
      "epoch": 0.7153411384739604,
      "grad_norm": 89.2710189819336,
      "learning_rate": 3.95752285684053e-06,
      "loss": 5.4973,
      "step": 17719
    },
    {
      "epoch": 0.7153815098909971,
      "grad_norm": 332.8781433105469,
      "learning_rate": 3.956481032987926e-06,
      "loss": 4.6139,
      "step": 17720
    },
    {
      "epoch": 0.7154218813080339,
      "grad_norm": 123.39241027832031,
      "learning_rate": 3.955439312465384e-06,
      "loss": 4.6739,
      "step": 17721
    },
    {
      "epoch": 0.7154622527250707,
      "grad_norm": 178.9936981201172,
      "learning_rate": 3.954397695290715e-06,
      "loss": 5.4286,
      "step": 17722
    },
    {
      "epoch": 0.7155026241421074,
      "grad_norm": 92.97830200195312,
      "learning_rate": 3.9533561814817274e-06,
      "loss": 5.2451,
      "step": 17723
    },
    {
      "epoch": 0.7155429955591441,
      "grad_norm": 101.87069702148438,
      "learning_rate": 3.95231477105623e-06,
      "loss": 5.408,
      "step": 17724
    },
    {
      "epoch": 0.7155833669761809,
      "grad_norm": 110.549072265625,
      "learning_rate": 3.951273464032027e-06,
      "loss": 4.9261,
      "step": 17725
    },
    {
      "epoch": 0.7156237383932176,
      "grad_norm": 94.31126403808594,
      "learning_rate": 3.950232260426927e-06,
      "loss": 4.1634,
      "step": 17726
    },
    {
      "epoch": 0.7156641098102543,
      "grad_norm": 91.15518188476562,
      "learning_rate": 3.949191160258724e-06,
      "loss": 5.0958,
      "step": 17727
    },
    {
      "epoch": 0.715704481227291,
      "grad_norm": 169.75856018066406,
      "learning_rate": 3.948150163545223e-06,
      "loss": 3.9963,
      "step": 17728
    },
    {
      "epoch": 0.7157448526443279,
      "grad_norm": 88.62460327148438,
      "learning_rate": 3.947109270304221e-06,
      "loss": 5.9822,
      "step": 17729
    },
    {
      "epoch": 0.7157852240613646,
      "grad_norm": 98.57259368896484,
      "learning_rate": 3.946068480553521e-06,
      "loss": 3.5235,
      "step": 17730
    },
    {
      "epoch": 0.7158255954784013,
      "grad_norm": 90.64157104492188,
      "learning_rate": 3.9450277943109095e-06,
      "loss": 3.994,
      "step": 17731
    },
    {
      "epoch": 0.715865966895438,
      "grad_norm": 84.12831115722656,
      "learning_rate": 3.943987211594183e-06,
      "loss": 5.4902,
      "step": 17732
    },
    {
      "epoch": 0.7159063383124747,
      "grad_norm": 85.00389099121094,
      "learning_rate": 3.942946732421133e-06,
      "loss": 5.1849,
      "step": 17733
    },
    {
      "epoch": 0.7159467097295115,
      "grad_norm": 86.48611450195312,
      "learning_rate": 3.941906356809553e-06,
      "loss": 5.7137,
      "step": 17734
    },
    {
      "epoch": 0.7159870811465483,
      "grad_norm": 106.99701690673828,
      "learning_rate": 3.940866084777224e-06,
      "loss": 4.8776,
      "step": 17735
    },
    {
      "epoch": 0.716027452563585,
      "grad_norm": 145.1290283203125,
      "learning_rate": 3.939825916341936e-06,
      "loss": 6.5094,
      "step": 17736
    },
    {
      "epoch": 0.7160678239806217,
      "grad_norm": 83.1431655883789,
      "learning_rate": 3.938785851521474e-06,
      "loss": 4.846,
      "step": 17737
    },
    {
      "epoch": 0.7161081953976585,
      "grad_norm": 88.04331970214844,
      "learning_rate": 3.937745890333623e-06,
      "loss": 3.8912,
      "step": 17738
    },
    {
      "epoch": 0.7161485668146952,
      "grad_norm": 86.5091781616211,
      "learning_rate": 3.936706032796158e-06,
      "loss": 5.9761,
      "step": 17739
    },
    {
      "epoch": 0.7161889382317319,
      "grad_norm": 87.73513793945312,
      "learning_rate": 3.935666278926861e-06,
      "loss": 5.5075,
      "step": 17740
    },
    {
      "epoch": 0.7162293096487686,
      "grad_norm": 135.6137237548828,
      "learning_rate": 3.934626628743508e-06,
      "loss": 3.8847,
      "step": 17741
    },
    {
      "epoch": 0.7162696810658055,
      "grad_norm": 92.44713592529297,
      "learning_rate": 3.9335870822638764e-06,
      "loss": 5.74,
      "step": 17742
    },
    {
      "epoch": 0.7163100524828422,
      "grad_norm": 88.6521987915039,
      "learning_rate": 3.93254763950574e-06,
      "loss": 6.4789,
      "step": 17743
    },
    {
      "epoch": 0.7163504238998789,
      "grad_norm": 112.65415954589844,
      "learning_rate": 3.931508300486872e-06,
      "loss": 5.279,
      "step": 17744
    },
    {
      "epoch": 0.7163907953169156,
      "grad_norm": 101.60844421386719,
      "learning_rate": 3.930469065225039e-06,
      "loss": 5.1424,
      "step": 17745
    },
    {
      "epoch": 0.7164311667339524,
      "grad_norm": 85.23753356933594,
      "learning_rate": 3.929429933738016e-06,
      "loss": 5.335,
      "step": 17746
    },
    {
      "epoch": 0.7164715381509891,
      "grad_norm": 101.99732971191406,
      "learning_rate": 3.928390906043561e-06,
      "loss": 5.4089,
      "step": 17747
    },
    {
      "epoch": 0.7165119095680258,
      "grad_norm": 81.1592788696289,
      "learning_rate": 3.927351982159444e-06,
      "loss": 5.2122,
      "step": 17748
    },
    {
      "epoch": 0.7165522809850626,
      "grad_norm": 77.07974243164062,
      "learning_rate": 3.926313162103427e-06,
      "loss": 5.1204,
      "step": 17749
    },
    {
      "epoch": 0.7165926524020994,
      "grad_norm": 78.53263092041016,
      "learning_rate": 3.925274445893276e-06,
      "loss": 4.8021,
      "step": 17750
    },
    {
      "epoch": 0.7166330238191361,
      "grad_norm": 86.9522476196289,
      "learning_rate": 3.924235833546742e-06,
      "loss": 5.5486,
      "step": 17751
    },
    {
      "epoch": 0.7166733952361728,
      "grad_norm": 109.50180053710938,
      "learning_rate": 3.923197325081588e-06,
      "loss": 3.1714,
      "step": 17752
    },
    {
      "epoch": 0.7167137666532095,
      "grad_norm": 96.34722900390625,
      "learning_rate": 3.92215892051557e-06,
      "loss": 5.2506,
      "step": 17753
    },
    {
      "epoch": 0.7167541380702462,
      "grad_norm": 76.17414093017578,
      "learning_rate": 3.921120619866444e-06,
      "loss": 4.9375,
      "step": 17754
    },
    {
      "epoch": 0.716794509487283,
      "grad_norm": 95.05596160888672,
      "learning_rate": 3.920082423151957e-06,
      "loss": 5.4153,
      "step": 17755
    },
    {
      "epoch": 0.7168348809043198,
      "grad_norm": 125.62779235839844,
      "learning_rate": 3.919044330389864e-06,
      "loss": 4.6951,
      "step": 17756
    },
    {
      "epoch": 0.7168752523213565,
      "grad_norm": 107.43189239501953,
      "learning_rate": 3.918006341597912e-06,
      "loss": 4.1895,
      "step": 17757
    },
    {
      "epoch": 0.7169156237383932,
      "grad_norm": 117.99968719482422,
      "learning_rate": 3.916968456793853e-06,
      "loss": 5.2939,
      "step": 17758
    },
    {
      "epoch": 0.71695599515543,
      "grad_norm": 67.1069107055664,
      "learning_rate": 3.915930675995424e-06,
      "loss": 4.511,
      "step": 17759
    },
    {
      "epoch": 0.7169963665724667,
      "grad_norm": 111.34992218017578,
      "learning_rate": 3.9148929992203735e-06,
      "loss": 3.0607,
      "step": 17760
    },
    {
      "epoch": 0.7170367379895034,
      "grad_norm": 90.46053314208984,
      "learning_rate": 3.913855426486444e-06,
      "loss": 4.8076,
      "step": 17761
    },
    {
      "epoch": 0.7170771094065401,
      "grad_norm": 126.16655731201172,
      "learning_rate": 3.9128179578113736e-06,
      "loss": 4.8402,
      "step": 17762
    },
    {
      "epoch": 0.717117480823577,
      "grad_norm": 149.18116760253906,
      "learning_rate": 3.911780593212901e-06,
      "loss": 4.0537,
      "step": 17763
    },
    {
      "epoch": 0.7171578522406137,
      "grad_norm": 82.5365982055664,
      "learning_rate": 3.910743332708764e-06,
      "loss": 4.5242,
      "step": 17764
    },
    {
      "epoch": 0.7171982236576504,
      "grad_norm": 74.80662536621094,
      "learning_rate": 3.909706176316697e-06,
      "loss": 4.9167,
      "step": 17765
    },
    {
      "epoch": 0.7172385950746871,
      "grad_norm": 78.03086853027344,
      "learning_rate": 3.908669124054435e-06,
      "loss": 4.6591,
      "step": 17766
    },
    {
      "epoch": 0.7172789664917238,
      "grad_norm": 110.69685363769531,
      "learning_rate": 3.907632175939704e-06,
      "loss": 5.9208,
      "step": 17767
    },
    {
      "epoch": 0.7173193379087606,
      "grad_norm": 106.65408325195312,
      "learning_rate": 3.906595331990235e-06,
      "loss": 3.6854,
      "step": 17768
    },
    {
      "epoch": 0.7173597093257973,
      "grad_norm": 65.53394317626953,
      "learning_rate": 3.905558592223758e-06,
      "loss": 5.1935,
      "step": 17769
    },
    {
      "epoch": 0.7174000807428341,
      "grad_norm": 80.6233139038086,
      "learning_rate": 3.904521956658001e-06,
      "loss": 6.1194,
      "step": 17770
    },
    {
      "epoch": 0.7174404521598708,
      "grad_norm": 106.06517791748047,
      "learning_rate": 3.903485425310681e-06,
      "loss": 4.5074,
      "step": 17771
    },
    {
      "epoch": 0.7174808235769076,
      "grad_norm": 110.9975357055664,
      "learning_rate": 3.902448998199523e-06,
      "loss": 5.1077,
      "step": 17772
    },
    {
      "epoch": 0.7175211949939443,
      "grad_norm": 67.60743713378906,
      "learning_rate": 3.901412675342248e-06,
      "loss": 4.3585,
      "step": 17773
    },
    {
      "epoch": 0.717561566410981,
      "grad_norm": 88.05708312988281,
      "learning_rate": 3.900376456756579e-06,
      "loss": 4.5257,
      "step": 17774
    },
    {
      "epoch": 0.7176019378280177,
      "grad_norm": 124.63139343261719,
      "learning_rate": 3.899340342460226e-06,
      "loss": 5.7215,
      "step": 17775
    },
    {
      "epoch": 0.7176423092450545,
      "grad_norm": 90.8363037109375,
      "learning_rate": 3.898304332470907e-06,
      "loss": 5.4624,
      "step": 17776
    },
    {
      "epoch": 0.7176826806620913,
      "grad_norm": 129.80845642089844,
      "learning_rate": 3.897268426806335e-06,
      "loss": 6.3362,
      "step": 17777
    },
    {
      "epoch": 0.717723052079128,
      "grad_norm": 93.91983032226562,
      "learning_rate": 3.896232625484226e-06,
      "loss": 3.8484,
      "step": 17778
    },
    {
      "epoch": 0.7177634234961647,
      "grad_norm": 72.12295532226562,
      "learning_rate": 3.8951969285222825e-06,
      "loss": 3.6226,
      "step": 17779
    },
    {
      "epoch": 0.7178037949132015,
      "grad_norm": 95.18919372558594,
      "learning_rate": 3.894161335938214e-06,
      "loss": 4.8205,
      "step": 17780
    },
    {
      "epoch": 0.7178441663302382,
      "grad_norm": 90.8933334350586,
      "learning_rate": 3.89312584774973e-06,
      "loss": 5.4199,
      "step": 17781
    },
    {
      "epoch": 0.7178845377472749,
      "grad_norm": 121.96383666992188,
      "learning_rate": 3.892090463974534e-06,
      "loss": 6.0051,
      "step": 17782
    },
    {
      "epoch": 0.7179249091643116,
      "grad_norm": 68.4494400024414,
      "learning_rate": 3.891055184630327e-06,
      "loss": 2.7492,
      "step": 17783
    },
    {
      "epoch": 0.7179652805813485,
      "grad_norm": 84.53829956054688,
      "learning_rate": 3.890020009734812e-06,
      "loss": 4.7179,
      "step": 17784
    },
    {
      "epoch": 0.7180056519983852,
      "grad_norm": 161.6858367919922,
      "learning_rate": 3.888984939305686e-06,
      "loss": 4.9532,
      "step": 17785
    },
    {
      "epoch": 0.7180460234154219,
      "grad_norm": 87.85582733154297,
      "learning_rate": 3.887949973360651e-06,
      "loss": 4.9563,
      "step": 17786
    },
    {
      "epoch": 0.7180863948324586,
      "grad_norm": 68.15233612060547,
      "learning_rate": 3.886915111917395e-06,
      "loss": 6.0054,
      "step": 17787
    },
    {
      "epoch": 0.7181267662494953,
      "grad_norm": 102.83809661865234,
      "learning_rate": 3.885880354993615e-06,
      "loss": 4.9633,
      "step": 17788
    },
    {
      "epoch": 0.7181671376665321,
      "grad_norm": 81.43882751464844,
      "learning_rate": 3.8848457026070054e-06,
      "loss": 5.6303,
      "step": 17789
    },
    {
      "epoch": 0.7182075090835688,
      "grad_norm": 72.49557495117188,
      "learning_rate": 3.883811154775257e-06,
      "loss": 5.1327,
      "step": 17790
    },
    {
      "epoch": 0.7182478805006056,
      "grad_norm": 86.10554504394531,
      "learning_rate": 3.882776711516051e-06,
      "loss": 5.5593,
      "step": 17791
    },
    {
      "epoch": 0.7182882519176423,
      "grad_norm": 59.35210037231445,
      "learning_rate": 3.881742372847079e-06,
      "loss": 4.6292,
      "step": 17792
    },
    {
      "epoch": 0.7183286233346791,
      "grad_norm": 138.49330139160156,
      "learning_rate": 3.880708138786026e-06,
      "loss": 4.7007,
      "step": 17793
    },
    {
      "epoch": 0.7183689947517158,
      "grad_norm": 142.83364868164062,
      "learning_rate": 3.879674009350577e-06,
      "loss": 4.6061,
      "step": 17794
    },
    {
      "epoch": 0.7184093661687525,
      "grad_norm": 90.33470916748047,
      "learning_rate": 3.878639984558408e-06,
      "loss": 3.5775,
      "step": 17795
    },
    {
      "epoch": 0.7184497375857892,
      "grad_norm": 100.12967681884766,
      "learning_rate": 3.877606064427199e-06,
      "loss": 2.9131,
      "step": 17796
    },
    {
      "epoch": 0.718490109002826,
      "grad_norm": 107.56356048583984,
      "learning_rate": 3.876572248974631e-06,
      "loss": 4.1189,
      "step": 17797
    },
    {
      "epoch": 0.7185304804198628,
      "grad_norm": 90.8962631225586,
      "learning_rate": 3.875538538218382e-06,
      "loss": 4.4169,
      "step": 17798
    },
    {
      "epoch": 0.7185708518368995,
      "grad_norm": 112.43630981445312,
      "learning_rate": 3.874504932176117e-06,
      "loss": 5.1886,
      "step": 17799
    },
    {
      "epoch": 0.7186112232539362,
      "grad_norm": 124.2145767211914,
      "learning_rate": 3.873471430865515e-06,
      "loss": 5.8678,
      "step": 17800
    },
    {
      "epoch": 0.718651594670973,
      "grad_norm": 78.53247833251953,
      "learning_rate": 3.872438034304246e-06,
      "loss": 4.5374,
      "step": 17801
    },
    {
      "epoch": 0.7186919660880097,
      "grad_norm": 99.65565490722656,
      "learning_rate": 3.871404742509975e-06,
      "loss": 5.593,
      "step": 17802
    },
    {
      "epoch": 0.7187323375050464,
      "grad_norm": 89.12149047851562,
      "learning_rate": 3.870371555500374e-06,
      "loss": 5.7258,
      "step": 17803
    },
    {
      "epoch": 0.7187727089220831,
      "grad_norm": 123.19215393066406,
      "learning_rate": 3.869338473293105e-06,
      "loss": 5.854,
      "step": 17804
    },
    {
      "epoch": 0.71881308033912,
      "grad_norm": 87.99706268310547,
      "learning_rate": 3.868305495905832e-06,
      "loss": 6.3297,
      "step": 17805
    },
    {
      "epoch": 0.7188534517561567,
      "grad_norm": 82.02203369140625,
      "learning_rate": 3.867272623356219e-06,
      "loss": 4.7395,
      "step": 17806
    },
    {
      "epoch": 0.7188938231731934,
      "grad_norm": 108.61325073242188,
      "learning_rate": 3.866239855661922e-06,
      "loss": 3.8084,
      "step": 17807
    },
    {
      "epoch": 0.7189341945902301,
      "grad_norm": 72.87926483154297,
      "learning_rate": 3.865207192840598e-06,
      "loss": 4.986,
      "step": 17808
    },
    {
      "epoch": 0.7189745660072668,
      "grad_norm": 137.42430114746094,
      "learning_rate": 3.864174634909906e-06,
      "loss": 5.2869,
      "step": 17809
    },
    {
      "epoch": 0.7190149374243036,
      "grad_norm": 146.07713317871094,
      "learning_rate": 3.863142181887502e-06,
      "loss": 5.456,
      "step": 17810
    },
    {
      "epoch": 0.7190553088413403,
      "grad_norm": 103.5013427734375,
      "learning_rate": 3.862109833791034e-06,
      "loss": 4.8644,
      "step": 17811
    },
    {
      "epoch": 0.7190956802583771,
      "grad_norm": 88.9433822631836,
      "learning_rate": 3.861077590638155e-06,
      "loss": 5.208,
      "step": 17812
    },
    {
      "epoch": 0.7191360516754138,
      "grad_norm": 88.09757995605469,
      "learning_rate": 3.860045452446514e-06,
      "loss": 2.631,
      "step": 17813
    },
    {
      "epoch": 0.7191764230924506,
      "grad_norm": 67.83744812011719,
      "learning_rate": 3.85901341923376e-06,
      "loss": 5.3419,
      "step": 17814
    },
    {
      "epoch": 0.7192167945094873,
      "grad_norm": 103.69971466064453,
      "learning_rate": 3.857981491017533e-06,
      "loss": 4.2638,
      "step": 17815
    },
    {
      "epoch": 0.719257165926524,
      "grad_norm": 115.85110473632812,
      "learning_rate": 3.85694966781548e-06,
      "loss": 5.6985,
      "step": 17816
    },
    {
      "epoch": 0.7192975373435607,
      "grad_norm": 149.4862823486328,
      "learning_rate": 3.855917949645244e-06,
      "loss": 4.5383,
      "step": 17817
    },
    {
      "epoch": 0.7193379087605974,
      "grad_norm": 82.35457611083984,
      "learning_rate": 3.854886336524465e-06,
      "loss": 5.2044,
      "step": 17818
    },
    {
      "epoch": 0.7193782801776343,
      "grad_norm": 153.36729431152344,
      "learning_rate": 3.853854828470778e-06,
      "loss": 4.4067,
      "step": 17819
    },
    {
      "epoch": 0.719418651594671,
      "grad_norm": 1609.24609375,
      "learning_rate": 3.852823425501821e-06,
      "loss": 4.1933,
      "step": 17820
    },
    {
      "epoch": 0.7194590230117077,
      "grad_norm": 107.15605163574219,
      "learning_rate": 3.851792127635227e-06,
      "loss": 5.0201,
      "step": 17821
    },
    {
      "epoch": 0.7194993944287444,
      "grad_norm": 107.42726135253906,
      "learning_rate": 3.850760934888632e-06,
      "loss": 5.1774,
      "step": 17822
    },
    {
      "epoch": 0.7195397658457812,
      "grad_norm": 115.2490234375,
      "learning_rate": 3.849729847279665e-06,
      "loss": 4.7312,
      "step": 17823
    },
    {
      "epoch": 0.7195801372628179,
      "grad_norm": 84.19264221191406,
      "learning_rate": 3.848698864825956e-06,
      "loss": 6.8747,
      "step": 17824
    },
    {
      "epoch": 0.7196205086798546,
      "grad_norm": 61.65154266357422,
      "learning_rate": 3.847667987545131e-06,
      "loss": 4.9684,
      "step": 17825
    },
    {
      "epoch": 0.7196608800968914,
      "grad_norm": 123.80884552001953,
      "learning_rate": 3.8466372154548225e-06,
      "loss": 3.2046,
      "step": 17826
    },
    {
      "epoch": 0.7197012515139282,
      "grad_norm": 102.39370727539062,
      "learning_rate": 3.845606548572643e-06,
      "loss": 5.104,
      "step": 17827
    },
    {
      "epoch": 0.7197416229309649,
      "grad_norm": 65.88726043701172,
      "learning_rate": 3.8445759869162215e-06,
      "loss": 3.8591,
      "step": 17828
    },
    {
      "epoch": 0.7197819943480016,
      "grad_norm": 108.6934585571289,
      "learning_rate": 3.843545530503175e-06,
      "loss": 5.0199,
      "step": 17829
    },
    {
      "epoch": 0.7198223657650383,
      "grad_norm": 81.16317749023438,
      "learning_rate": 3.842515179351128e-06,
      "loss": 5.2694,
      "step": 17830
    },
    {
      "epoch": 0.719862737182075,
      "grad_norm": 117.75979614257812,
      "learning_rate": 3.841484933477689e-06,
      "loss": 4.4106,
      "step": 17831
    },
    {
      "epoch": 0.7199031085991118,
      "grad_norm": 81.69473266601562,
      "learning_rate": 3.840454792900477e-06,
      "loss": 4.7926,
      "step": 17832
    },
    {
      "epoch": 0.7199434800161486,
      "grad_norm": 93.72245788574219,
      "learning_rate": 3.839424757637103e-06,
      "loss": 5.9337,
      "step": 17833
    },
    {
      "epoch": 0.7199838514331853,
      "grad_norm": 87.32365417480469,
      "learning_rate": 3.838394827705185e-06,
      "loss": 5.6454,
      "step": 17834
    },
    {
      "epoch": 0.720024222850222,
      "grad_norm": 72.08695983886719,
      "learning_rate": 3.837365003122322e-06,
      "loss": 5.3032,
      "step": 17835
    },
    {
      "epoch": 0.7200645942672588,
      "grad_norm": 100.24179077148438,
      "learning_rate": 3.836335283906127e-06,
      "loss": 3.0942,
      "step": 17836
    },
    {
      "epoch": 0.7201049656842955,
      "grad_norm": 105.57195281982422,
      "learning_rate": 3.835305670074206e-06,
      "loss": 5.0813,
      "step": 17837
    },
    {
      "epoch": 0.7201453371013322,
      "grad_norm": 134.5303192138672,
      "learning_rate": 3.834276161644163e-06,
      "loss": 3.5585,
      "step": 17838
    },
    {
      "epoch": 0.720185708518369,
      "grad_norm": 81.42672729492188,
      "learning_rate": 3.833246758633599e-06,
      "loss": 6.3352,
      "step": 17839
    },
    {
      "epoch": 0.7202260799354058,
      "grad_norm": 101.57672882080078,
      "learning_rate": 3.832217461060115e-06,
      "loss": 5.9599,
      "step": 17840
    },
    {
      "epoch": 0.7202664513524425,
      "grad_norm": 100.54145050048828,
      "learning_rate": 3.8311882689413095e-06,
      "loss": 4.9524,
      "step": 17841
    },
    {
      "epoch": 0.7203068227694792,
      "grad_norm": 84.00010681152344,
      "learning_rate": 3.8301591822947824e-06,
      "loss": 2.9757,
      "step": 17842
    },
    {
      "epoch": 0.7203471941865159,
      "grad_norm": 141.87286376953125,
      "learning_rate": 3.829130201138122e-06,
      "loss": 3.6106,
      "step": 17843
    },
    {
      "epoch": 0.7203875656035527,
      "grad_norm": 90.8973159790039,
      "learning_rate": 3.828101325488926e-06,
      "loss": 4.5262,
      "step": 17844
    },
    {
      "epoch": 0.7204279370205894,
      "grad_norm": 85.34207153320312,
      "learning_rate": 3.8270725553647856e-06,
      "loss": 5.8931,
      "step": 17845
    },
    {
      "epoch": 0.7204683084376262,
      "grad_norm": 141.3523406982422,
      "learning_rate": 3.826043890783292e-06,
      "loss": 5.3845,
      "step": 17846
    },
    {
      "epoch": 0.7205086798546629,
      "grad_norm": 99.04891204833984,
      "learning_rate": 3.825015331762028e-06,
      "loss": 4.0431,
      "step": 17847
    },
    {
      "epoch": 0.7205490512716997,
      "grad_norm": 147.33807373046875,
      "learning_rate": 3.823986878318582e-06,
      "loss": 4.2556,
      "step": 17848
    },
    {
      "epoch": 0.7205894226887364,
      "grad_norm": 86.84320831298828,
      "learning_rate": 3.822958530470538e-06,
      "loss": 4.8696,
      "step": 17849
    },
    {
      "epoch": 0.7206297941057731,
      "grad_norm": 194.22621154785156,
      "learning_rate": 3.8219302882354815e-06,
      "loss": 4.3158,
      "step": 17850
    },
    {
      "epoch": 0.7206701655228098,
      "grad_norm": 113.97061157226562,
      "learning_rate": 3.820902151630987e-06,
      "loss": 4.9696,
      "step": 17851
    },
    {
      "epoch": 0.7207105369398465,
      "grad_norm": 87.69742584228516,
      "learning_rate": 3.819874120674638e-06,
      "loss": 3.0021,
      "step": 17852
    },
    {
      "epoch": 0.7207509083568834,
      "grad_norm": 96.93183135986328,
      "learning_rate": 3.818846195384008e-06,
      "loss": 4.8681,
      "step": 17853
    },
    {
      "epoch": 0.7207912797739201,
      "grad_norm": 118.77726745605469,
      "learning_rate": 3.817818375776679e-06,
      "loss": 5.1031,
      "step": 17854
    },
    {
      "epoch": 0.7208316511909568,
      "grad_norm": 77.08646392822266,
      "learning_rate": 3.816790661870216e-06,
      "loss": 4.8426,
      "step": 17855
    },
    {
      "epoch": 0.7208720226079935,
      "grad_norm": 76.32363891601562,
      "learning_rate": 3.815763053682192e-06,
      "loss": 5.0494,
      "step": 17856
    },
    {
      "epoch": 0.7209123940250303,
      "grad_norm": 94.65681457519531,
      "learning_rate": 3.8147355512301795e-06,
      "loss": 4.8691,
      "step": 17857
    },
    {
      "epoch": 0.720952765442067,
      "grad_norm": 1284.81982421875,
      "learning_rate": 3.8137081545317454e-06,
      "loss": 4.925,
      "step": 17858
    },
    {
      "epoch": 0.7209931368591037,
      "grad_norm": 71.1159896850586,
      "learning_rate": 3.8126808636044563e-06,
      "loss": 5.0281,
      "step": 17859
    },
    {
      "epoch": 0.7210335082761405,
      "grad_norm": 69.28164672851562,
      "learning_rate": 3.8116536784658763e-06,
      "loss": 2.5848,
      "step": 17860
    },
    {
      "epoch": 0.7210738796931773,
      "grad_norm": 161.48646545410156,
      "learning_rate": 3.8106265991335665e-06,
      "loss": 5.0075,
      "step": 17861
    },
    {
      "epoch": 0.721114251110214,
      "grad_norm": 86.18171691894531,
      "learning_rate": 3.8095996256250922e-06,
      "loss": 4.7799,
      "step": 17862
    },
    {
      "epoch": 0.7211546225272507,
      "grad_norm": 94.55469512939453,
      "learning_rate": 3.8085727579580057e-06,
      "loss": 5.0711,
      "step": 17863
    },
    {
      "epoch": 0.7211949939442874,
      "grad_norm": 97.95748138427734,
      "learning_rate": 3.807545996149867e-06,
      "loss": 5.0017,
      "step": 17864
    },
    {
      "epoch": 0.7212353653613242,
      "grad_norm": 81.1093521118164,
      "learning_rate": 3.806519340218231e-06,
      "loss": 4.7661,
      "step": 17865
    },
    {
      "epoch": 0.7212757367783609,
      "grad_norm": 121.74951934814453,
      "learning_rate": 3.805492790180656e-06,
      "loss": 5.1665,
      "step": 17866
    },
    {
      "epoch": 0.7213161081953977,
      "grad_norm": 83.22969055175781,
      "learning_rate": 3.8044663460546847e-06,
      "loss": 3.8689,
      "step": 17867
    },
    {
      "epoch": 0.7213564796124344,
      "grad_norm": 80.14474487304688,
      "learning_rate": 3.8034400078578714e-06,
      "loss": 3.7728,
      "step": 17868
    },
    {
      "epoch": 0.7213968510294712,
      "grad_norm": 96.47117614746094,
      "learning_rate": 3.802413775607764e-06,
      "loss": 6.1685,
      "step": 17869
    },
    {
      "epoch": 0.7214372224465079,
      "grad_norm": 101.84151458740234,
      "learning_rate": 3.8013876493219116e-06,
      "loss": 6.0073,
      "step": 17870
    },
    {
      "epoch": 0.7214775938635446,
      "grad_norm": 81.46430969238281,
      "learning_rate": 3.8003616290178526e-06,
      "loss": 5.7492,
      "step": 17871
    },
    {
      "epoch": 0.7215179652805813,
      "grad_norm": 108.0719223022461,
      "learning_rate": 3.7993357147131327e-06,
      "loss": 3.9866,
      "step": 17872
    },
    {
      "epoch": 0.721558336697618,
      "grad_norm": 107.58767700195312,
      "learning_rate": 3.7983099064252925e-06,
      "loss": 4.8333,
      "step": 17873
    },
    {
      "epoch": 0.7215987081146549,
      "grad_norm": 238.87124633789062,
      "learning_rate": 3.797284204171874e-06,
      "loss": 4.4228,
      "step": 17874
    },
    {
      "epoch": 0.7216390795316916,
      "grad_norm": 293.3580017089844,
      "learning_rate": 3.796258607970408e-06,
      "loss": 5.6448,
      "step": 17875
    },
    {
      "epoch": 0.7216794509487283,
      "grad_norm": 148.99826049804688,
      "learning_rate": 3.795233117838433e-06,
      "loss": 5.8093,
      "step": 17876
    },
    {
      "epoch": 0.721719822365765,
      "grad_norm": 134.1521759033203,
      "learning_rate": 3.7942077337934823e-06,
      "loss": 3.8802,
      "step": 17877
    },
    {
      "epoch": 0.7217601937828018,
      "grad_norm": 81.3379898071289,
      "learning_rate": 3.793182455853087e-06,
      "loss": 5.4697,
      "step": 17878
    },
    {
      "epoch": 0.7218005651998385,
      "grad_norm": 129.11337280273438,
      "learning_rate": 3.79215728403478e-06,
      "loss": 5.0755,
      "step": 17879
    },
    {
      "epoch": 0.7218409366168752,
      "grad_norm": 136.8218536376953,
      "learning_rate": 3.7911322183560852e-06,
      "loss": 6.1199,
      "step": 17880
    },
    {
      "epoch": 0.721881308033912,
      "grad_norm": 96.3621597290039,
      "learning_rate": 3.790107258834531e-06,
      "loss": 4.9312,
      "step": 17881
    },
    {
      "epoch": 0.7219216794509488,
      "grad_norm": 146.0897979736328,
      "learning_rate": 3.7890824054876452e-06,
      "loss": 4.6822,
      "step": 17882
    },
    {
      "epoch": 0.7219620508679855,
      "grad_norm": 151.07867431640625,
      "learning_rate": 3.788057658332943e-06,
      "loss": 3.6708,
      "step": 17883
    },
    {
      "epoch": 0.7220024222850222,
      "grad_norm": 89.15691375732422,
      "learning_rate": 3.78703301738795e-06,
      "loss": 4.7447,
      "step": 17884
    },
    {
      "epoch": 0.7220427937020589,
      "grad_norm": 95.90312194824219,
      "learning_rate": 3.786008482670184e-06,
      "loss": 3.4623,
      "step": 17885
    },
    {
      "epoch": 0.7220831651190956,
      "grad_norm": 113.36502075195312,
      "learning_rate": 3.784984054197164e-06,
      "loss": 4.1728,
      "step": 17886
    },
    {
      "epoch": 0.7221235365361324,
      "grad_norm": 180.03079223632812,
      "learning_rate": 3.783959731986402e-06,
      "loss": 5.1221,
      "step": 17887
    },
    {
      "epoch": 0.7221639079531692,
      "grad_norm": 114.16494750976562,
      "learning_rate": 3.7829355160554115e-06,
      "loss": 5.8893,
      "step": 17888
    },
    {
      "epoch": 0.7222042793702059,
      "grad_norm": 93.18740844726562,
      "learning_rate": 3.7819114064217066e-06,
      "loss": 5.411,
      "step": 17889
    },
    {
      "epoch": 0.7222446507872426,
      "grad_norm": 77.48762512207031,
      "learning_rate": 3.7808874031028e-06,
      "loss": 5.0353,
      "step": 17890
    },
    {
      "epoch": 0.7222850222042794,
      "grad_norm": 80.074951171875,
      "learning_rate": 3.7798635061161926e-06,
      "loss": 5.8506,
      "step": 17891
    },
    {
      "epoch": 0.7223253936213161,
      "grad_norm": 68.02997589111328,
      "learning_rate": 3.778839715479393e-06,
      "loss": 4.8384,
      "step": 17892
    },
    {
      "epoch": 0.7223657650383528,
      "grad_norm": 136.4193878173828,
      "learning_rate": 3.7778160312099086e-06,
      "loss": 4.7656,
      "step": 17893
    },
    {
      "epoch": 0.7224061364553895,
      "grad_norm": 86.64573669433594,
      "learning_rate": 3.7767924533252433e-06,
      "loss": 5.8013,
      "step": 17894
    },
    {
      "epoch": 0.7224465078724264,
      "grad_norm": 82.04637908935547,
      "learning_rate": 3.77576898184289e-06,
      "loss": 3.8377,
      "step": 17895
    },
    {
      "epoch": 0.7224868792894631,
      "grad_norm": 176.369873046875,
      "learning_rate": 3.7747456167803544e-06,
      "loss": 4.0477,
      "step": 17896
    },
    {
      "epoch": 0.7225272507064998,
      "grad_norm": 101.14390563964844,
      "learning_rate": 3.7737223581551307e-06,
      "loss": 4.0241,
      "step": 17897
    },
    {
      "epoch": 0.7225676221235365,
      "grad_norm": 103.17893981933594,
      "learning_rate": 3.7726992059847146e-06,
      "loss": 6.5153,
      "step": 17898
    },
    {
      "epoch": 0.7226079935405733,
      "grad_norm": 103.0093002319336,
      "learning_rate": 3.7716761602866014e-06,
      "loss": 5.573,
      "step": 17899
    },
    {
      "epoch": 0.72264836495761,
      "grad_norm": 104.6667251586914,
      "learning_rate": 3.770653221078281e-06,
      "loss": 6.9973,
      "step": 17900
    },
    {
      "epoch": 0.7226887363746467,
      "grad_norm": 91.62788391113281,
      "learning_rate": 3.7696303883772443e-06,
      "loss": 4.7237,
      "step": 17901
    },
    {
      "epoch": 0.7227291077916835,
      "grad_norm": 256.78338623046875,
      "learning_rate": 3.7686076622009826e-06,
      "loss": 5.8755,
      "step": 17902
    },
    {
      "epoch": 0.7227694792087203,
      "grad_norm": 96.87505340576172,
      "learning_rate": 3.7675850425669745e-06,
      "loss": 4.865,
      "step": 17903
    },
    {
      "epoch": 0.722809850625757,
      "grad_norm": 92.11466217041016,
      "learning_rate": 3.766562529492709e-06,
      "loss": 3.6625,
      "step": 17904
    },
    {
      "epoch": 0.7228502220427937,
      "grad_norm": 205.73129272460938,
      "learning_rate": 3.765540122995668e-06,
      "loss": 3.7096,
      "step": 17905
    },
    {
      "epoch": 0.7228905934598304,
      "grad_norm": 85.28720092773438,
      "learning_rate": 3.7645178230933368e-06,
      "loss": 4.522,
      "step": 17906
    },
    {
      "epoch": 0.7229309648768671,
      "grad_norm": 101.22223663330078,
      "learning_rate": 3.7634956298031844e-06,
      "loss": 5.2673,
      "step": 17907
    },
    {
      "epoch": 0.7229713362939039,
      "grad_norm": 97.72893524169922,
      "learning_rate": 3.7624735431426952e-06,
      "loss": 5.7556,
      "step": 17908
    },
    {
      "epoch": 0.7230117077109407,
      "grad_norm": 103.98007202148438,
      "learning_rate": 3.761451563129341e-06,
      "loss": 6.1602,
      "step": 17909
    },
    {
      "epoch": 0.7230520791279774,
      "grad_norm": 173.1437530517578,
      "learning_rate": 3.7604296897806023e-06,
      "loss": 5.2679,
      "step": 17910
    },
    {
      "epoch": 0.7230924505450141,
      "grad_norm": 82.8119888305664,
      "learning_rate": 3.759407923113941e-06,
      "loss": 6.1711,
      "step": 17911
    },
    {
      "epoch": 0.7231328219620509,
      "grad_norm": 73.12384796142578,
      "learning_rate": 3.7583862631468316e-06,
      "loss": 3.7095,
      "step": 17912
    },
    {
      "epoch": 0.7231731933790876,
      "grad_norm": 105.6054458618164,
      "learning_rate": 3.7573647098967424e-06,
      "loss": 6.6645,
      "step": 17913
    },
    {
      "epoch": 0.7232135647961243,
      "grad_norm": 82.33493041992188,
      "learning_rate": 3.7563432633811414e-06,
      "loss": 5.1929,
      "step": 17914
    },
    {
      "epoch": 0.723253936213161,
      "grad_norm": 101.59346008300781,
      "learning_rate": 3.7553219236174877e-06,
      "loss": 3.5958,
      "step": 17915
    },
    {
      "epoch": 0.7232943076301979,
      "grad_norm": 104.25603485107422,
      "learning_rate": 3.7543006906232472e-06,
      "loss": 3.63,
      "step": 17916
    },
    {
      "epoch": 0.7233346790472346,
      "grad_norm": 65.47854614257812,
      "learning_rate": 3.7532795644158793e-06,
      "loss": 4.2504,
      "step": 17917
    },
    {
      "epoch": 0.7233750504642713,
      "grad_norm": 100.97763061523438,
      "learning_rate": 3.7522585450128448e-06,
      "loss": 4.7168,
      "step": 17918
    },
    {
      "epoch": 0.723415421881308,
      "grad_norm": 80.2391357421875,
      "learning_rate": 3.751237632431599e-06,
      "loss": 4.7435,
      "step": 17919
    },
    {
      "epoch": 0.7234557932983448,
      "grad_norm": 111.26048278808594,
      "learning_rate": 3.750216826689598e-06,
      "loss": 4.3668,
      "step": 17920
    },
    {
      "epoch": 0.7234961647153815,
      "grad_norm": 82.87154388427734,
      "learning_rate": 3.7491961278042943e-06,
      "loss": 4.1268,
      "step": 17921
    },
    {
      "epoch": 0.7235365361324182,
      "grad_norm": 98.9007568359375,
      "learning_rate": 3.7481755357931437e-06,
      "loss": 3.6807,
      "step": 17922
    },
    {
      "epoch": 0.723576907549455,
      "grad_norm": 98.64599609375,
      "learning_rate": 3.74715505067359e-06,
      "loss": 4.4761,
      "step": 17923
    },
    {
      "epoch": 0.7236172789664918,
      "grad_norm": 99.11631774902344,
      "learning_rate": 3.7461346724630822e-06,
      "loss": 5.5909,
      "step": 17924
    },
    {
      "epoch": 0.7236576503835285,
      "grad_norm": 117.77110290527344,
      "learning_rate": 3.7451144011790687e-06,
      "loss": 5.528,
      "step": 17925
    },
    {
      "epoch": 0.7236980218005652,
      "grad_norm": 71.34911346435547,
      "learning_rate": 3.7440942368389956e-06,
      "loss": 5.5543,
      "step": 17926
    },
    {
      "epoch": 0.7237383932176019,
      "grad_norm": 102.88658142089844,
      "learning_rate": 3.7430741794602987e-06,
      "loss": 6.2091,
      "step": 17927
    },
    {
      "epoch": 0.7237787646346386,
      "grad_norm": 90.30516052246094,
      "learning_rate": 3.7420542290604235e-06,
      "loss": 5.8298,
      "step": 17928
    },
    {
      "epoch": 0.7238191360516754,
      "grad_norm": 88.6335678100586,
      "learning_rate": 3.7410343856568067e-06,
      "loss": 4.962,
      "step": 17929
    },
    {
      "epoch": 0.7238595074687122,
      "grad_norm": 119.73008728027344,
      "learning_rate": 3.7400146492668898e-06,
      "loss": 4.3874,
      "step": 17930
    },
    {
      "epoch": 0.7238998788857489,
      "grad_norm": 267.8570251464844,
      "learning_rate": 3.7389950199081006e-06,
      "loss": 6.4143,
      "step": 17931
    },
    {
      "epoch": 0.7239402503027856,
      "grad_norm": 151.36138916015625,
      "learning_rate": 3.7379754975978765e-06,
      "loss": 4.7072,
      "step": 17932
    },
    {
      "epoch": 0.7239806217198224,
      "grad_norm": 108.36542510986328,
      "learning_rate": 3.736956082353648e-06,
      "loss": 4.9301,
      "step": 17933
    },
    {
      "epoch": 0.7240209931368591,
      "grad_norm": 94.98316955566406,
      "learning_rate": 3.7359367741928485e-06,
      "loss": 4.9239,
      "step": 17934
    },
    {
      "epoch": 0.7240613645538958,
      "grad_norm": 71.12796020507812,
      "learning_rate": 3.7349175731329e-06,
      "loss": 4.5071,
      "step": 17935
    },
    {
      "epoch": 0.7241017359709325,
      "grad_norm": 91.58940887451172,
      "learning_rate": 3.7338984791912304e-06,
      "loss": 4.9291,
      "step": 17936
    },
    {
      "epoch": 0.7241421073879694,
      "grad_norm": 130.65365600585938,
      "learning_rate": 3.7328794923852653e-06,
      "loss": 5.5909,
      "step": 17937
    },
    {
      "epoch": 0.7241824788050061,
      "grad_norm": 101.73654174804688,
      "learning_rate": 3.7318606127324252e-06,
      "loss": 5.2897,
      "step": 17938
    },
    {
      "epoch": 0.7242228502220428,
      "grad_norm": 113.65044403076172,
      "learning_rate": 3.730841840250131e-06,
      "loss": 3.5692,
      "step": 17939
    },
    {
      "epoch": 0.7242632216390795,
      "grad_norm": 69.16468811035156,
      "learning_rate": 3.7298231749558024e-06,
      "loss": 3.199,
      "step": 17940
    },
    {
      "epoch": 0.7243035930561162,
      "grad_norm": 124.18905639648438,
      "learning_rate": 3.7288046168668556e-06,
      "loss": 6.5592,
      "step": 17941
    },
    {
      "epoch": 0.724343964473153,
      "grad_norm": 105.62406921386719,
      "learning_rate": 3.72778616600071e-06,
      "loss": 4.2978,
      "step": 17942
    },
    {
      "epoch": 0.7243843358901898,
      "grad_norm": 82.68024444580078,
      "learning_rate": 3.7267678223747693e-06,
      "loss": 4.696,
      "step": 17943
    },
    {
      "epoch": 0.7244247073072265,
      "grad_norm": 70.0086441040039,
      "learning_rate": 3.725749586006451e-06,
      "loss": 4.9046,
      "step": 17944
    },
    {
      "epoch": 0.7244650787242632,
      "grad_norm": 131.22557067871094,
      "learning_rate": 3.724731456913162e-06,
      "loss": 4.2017,
      "step": 17945
    },
    {
      "epoch": 0.7245054501413,
      "grad_norm": 84.98475646972656,
      "learning_rate": 3.723713435112315e-06,
      "loss": 3.5536,
      "step": 17946
    },
    {
      "epoch": 0.7245458215583367,
      "grad_norm": 84.73064422607422,
      "learning_rate": 3.72269552062131e-06,
      "loss": 6.0701,
      "step": 17947
    },
    {
      "epoch": 0.7245861929753734,
      "grad_norm": 114.03317260742188,
      "learning_rate": 3.721677713457552e-06,
      "loss": 5.9924,
      "step": 17948
    },
    {
      "epoch": 0.7246265643924101,
      "grad_norm": 1636.1619873046875,
      "learning_rate": 3.7206600136384442e-06,
      "loss": 5.4199,
      "step": 17949
    },
    {
      "epoch": 0.724666935809447,
      "grad_norm": 87.40363311767578,
      "learning_rate": 3.7196424211813895e-06,
      "loss": 4.3427,
      "step": 17950
    },
    {
      "epoch": 0.7247073072264837,
      "grad_norm": 111.65367126464844,
      "learning_rate": 3.718624936103782e-06,
      "loss": 5.1039,
      "step": 17951
    },
    {
      "epoch": 0.7247476786435204,
      "grad_norm": 87.02066802978516,
      "learning_rate": 3.7176075584230187e-06,
      "loss": 5.253,
      "step": 17952
    },
    {
      "epoch": 0.7247880500605571,
      "grad_norm": 97.8545150756836,
      "learning_rate": 3.716590288156496e-06,
      "loss": 3.9487,
      "step": 17953
    },
    {
      "epoch": 0.7248284214775939,
      "grad_norm": 111.96134185791016,
      "learning_rate": 3.715573125321611e-06,
      "loss": 5.5876,
      "step": 17954
    },
    {
      "epoch": 0.7248687928946306,
      "grad_norm": 87.39720153808594,
      "learning_rate": 3.7145560699357454e-06,
      "loss": 5.1559,
      "step": 17955
    },
    {
      "epoch": 0.7249091643116673,
      "grad_norm": 104.0103530883789,
      "learning_rate": 3.7135391220162943e-06,
      "loss": 3.5562,
      "step": 17956
    },
    {
      "epoch": 0.7249495357287041,
      "grad_norm": 65.91722106933594,
      "learning_rate": 3.712522281580644e-06,
      "loss": 3.1617,
      "step": 17957
    },
    {
      "epoch": 0.7249899071457409,
      "grad_norm": 158.99307250976562,
      "learning_rate": 3.7115055486461804e-06,
      "loss": 5.208,
      "step": 17958
    },
    {
      "epoch": 0.7250302785627776,
      "grad_norm": 64.59117889404297,
      "learning_rate": 3.7104889232302875e-06,
      "loss": 5.0485,
      "step": 17959
    },
    {
      "epoch": 0.7250706499798143,
      "grad_norm": 71.47188568115234,
      "learning_rate": 3.709472405350346e-06,
      "loss": 5.3766,
      "step": 17960
    },
    {
      "epoch": 0.725111021396851,
      "grad_norm": 111.14030456542969,
      "learning_rate": 3.708455995023739e-06,
      "loss": 4.6129,
      "step": 17961
    },
    {
      "epoch": 0.7251513928138877,
      "grad_norm": 104.95970153808594,
      "learning_rate": 3.707439692267845e-06,
      "loss": 4.4685,
      "step": 17962
    },
    {
      "epoch": 0.7251917642309245,
      "grad_norm": 79.3929214477539,
      "learning_rate": 3.7064234971000335e-06,
      "loss": 4.2962,
      "step": 17963
    },
    {
      "epoch": 0.7252321356479613,
      "grad_norm": 104.16120147705078,
      "learning_rate": 3.7054074095376845e-06,
      "loss": 3.9364,
      "step": 17964
    },
    {
      "epoch": 0.725272507064998,
      "grad_norm": 83.17855072021484,
      "learning_rate": 3.7043914295981698e-06,
      "loss": 4.8802,
      "step": 17965
    },
    {
      "epoch": 0.7253128784820347,
      "grad_norm": 90.4112548828125,
      "learning_rate": 3.703375557298864e-06,
      "loss": 4.5231,
      "step": 17966
    },
    {
      "epoch": 0.7253532498990715,
      "grad_norm": 122.2471694946289,
      "learning_rate": 3.7023597926571287e-06,
      "loss": 4.9969,
      "step": 17967
    },
    {
      "epoch": 0.7253936213161082,
      "grad_norm": 105.70979309082031,
      "learning_rate": 3.701344135690336e-06,
      "loss": 6.1154,
      "step": 17968
    },
    {
      "epoch": 0.7254339927331449,
      "grad_norm": 73.29557800292969,
      "learning_rate": 3.700328586415849e-06,
      "loss": 4.3368,
      "step": 17969
    },
    {
      "epoch": 0.7254743641501816,
      "grad_norm": 83.22669219970703,
      "learning_rate": 3.6993131448510366e-06,
      "loss": 4.7709,
      "step": 17970
    },
    {
      "epoch": 0.7255147355672185,
      "grad_norm": 115.56688690185547,
      "learning_rate": 3.698297811013252e-06,
      "loss": 6.0472,
      "step": 17971
    },
    {
      "epoch": 0.7255551069842552,
      "grad_norm": 121.65177917480469,
      "learning_rate": 3.6972825849198592e-06,
      "loss": 6.4209,
      "step": 17972
    },
    {
      "epoch": 0.7255954784012919,
      "grad_norm": 51.77739715576172,
      "learning_rate": 3.696267466588217e-06,
      "loss": 1.5751,
      "step": 17973
    },
    {
      "epoch": 0.7256358498183286,
      "grad_norm": 181.01345825195312,
      "learning_rate": 3.695252456035684e-06,
      "loss": 3.8102,
      "step": 17974
    },
    {
      "epoch": 0.7256762212353653,
      "grad_norm": 81.35627746582031,
      "learning_rate": 3.694237553279608e-06,
      "loss": 3.5299,
      "step": 17975
    },
    {
      "epoch": 0.7257165926524021,
      "grad_norm": 76.34280395507812,
      "learning_rate": 3.6932227583373447e-06,
      "loss": 4.4159,
      "step": 17976
    },
    {
      "epoch": 0.7257569640694388,
      "grad_norm": 80.2529525756836,
      "learning_rate": 3.692208071226245e-06,
      "loss": 6.885,
      "step": 17977
    },
    {
      "epoch": 0.7257973354864756,
      "grad_norm": 73.48104095458984,
      "learning_rate": 3.6911934919636572e-06,
      "loss": 5.8828,
      "step": 17978
    },
    {
      "epoch": 0.7258377069035123,
      "grad_norm": 104.58683776855469,
      "learning_rate": 3.6901790205669297e-06,
      "loss": 3.2867,
      "step": 17979
    },
    {
      "epoch": 0.7258780783205491,
      "grad_norm": 152.3020782470703,
      "learning_rate": 3.6891646570534056e-06,
      "loss": 4.8839,
      "step": 17980
    },
    {
      "epoch": 0.7259184497375858,
      "grad_norm": 176.8861083984375,
      "learning_rate": 3.688150401440429e-06,
      "loss": 4.5556,
      "step": 17981
    },
    {
      "epoch": 0.7259588211546225,
      "grad_norm": 91.32635498046875,
      "learning_rate": 3.687136253745345e-06,
      "loss": 4.5059,
      "step": 17982
    },
    {
      "epoch": 0.7259991925716592,
      "grad_norm": 129.37071228027344,
      "learning_rate": 3.6861222139854857e-06,
      "loss": 5.331,
      "step": 17983
    },
    {
      "epoch": 0.726039563988696,
      "grad_norm": 90.2817611694336,
      "learning_rate": 3.6851082821781923e-06,
      "loss": 5.4866,
      "step": 17984
    },
    {
      "epoch": 0.7260799354057328,
      "grad_norm": 99.18350219726562,
      "learning_rate": 3.684094458340801e-06,
      "loss": 3.8297,
      "step": 17985
    },
    {
      "epoch": 0.7261203068227695,
      "grad_norm": 67.0342025756836,
      "learning_rate": 3.68308074249065e-06,
      "loss": 4.1927,
      "step": 17986
    },
    {
      "epoch": 0.7261606782398062,
      "grad_norm": 86.6060562133789,
      "learning_rate": 3.6820671346450633e-06,
      "loss": 5.3084,
      "step": 17987
    },
    {
      "epoch": 0.726201049656843,
      "grad_norm": 127.28715515136719,
      "learning_rate": 3.681053634821374e-06,
      "loss": 5.0996,
      "step": 17988
    },
    {
      "epoch": 0.7262414210738797,
      "grad_norm": 107.2988052368164,
      "learning_rate": 3.680040243036914e-06,
      "loss": 3.6759,
      "step": 17989
    },
    {
      "epoch": 0.7262817924909164,
      "grad_norm": 203.8661651611328,
      "learning_rate": 3.67902695930901e-06,
      "loss": 4.2279,
      "step": 17990
    },
    {
      "epoch": 0.7263221639079531,
      "grad_norm": 127.75428009033203,
      "learning_rate": 3.6780137836549813e-06,
      "loss": 5.0534,
      "step": 17991
    },
    {
      "epoch": 0.72636253532499,
      "grad_norm": 81.79066467285156,
      "learning_rate": 3.6770007160921527e-06,
      "loss": 5.1742,
      "step": 17992
    },
    {
      "epoch": 0.7264029067420267,
      "grad_norm": 92.9368896484375,
      "learning_rate": 3.6759877566378477e-06,
      "loss": 4.4428,
      "step": 17993
    },
    {
      "epoch": 0.7264432781590634,
      "grad_norm": 79.09672546386719,
      "learning_rate": 3.6749749053093884e-06,
      "loss": 4.8264,
      "step": 17994
    },
    {
      "epoch": 0.7264836495761001,
      "grad_norm": 160.8951873779297,
      "learning_rate": 3.673962162124084e-06,
      "loss": 3.932,
      "step": 17995
    },
    {
      "epoch": 0.7265240209931368,
      "grad_norm": 172.14889526367188,
      "learning_rate": 3.6729495270992543e-06,
      "loss": 4.3805,
      "step": 17996
    },
    {
      "epoch": 0.7265643924101736,
      "grad_norm": 93.06327819824219,
      "learning_rate": 3.6719370002522137e-06,
      "loss": 5.7599,
      "step": 17997
    },
    {
      "epoch": 0.7266047638272103,
      "grad_norm": 75.07257843017578,
      "learning_rate": 3.6709245816002723e-06,
      "loss": 4.3462,
      "step": 17998
    },
    {
      "epoch": 0.7266451352442471,
      "grad_norm": 129.26634216308594,
      "learning_rate": 3.6699122711607414e-06,
      "loss": 5.5074,
      "step": 17999
    },
    {
      "epoch": 0.7266855066612838,
      "grad_norm": 81.87654876708984,
      "learning_rate": 3.668900068950929e-06,
      "loss": 4.4098,
      "step": 18000
    },
    {
      "epoch": 0.7267258780783206,
      "grad_norm": 217.66001892089844,
      "learning_rate": 3.6678879749881413e-06,
      "loss": 3.2315,
      "step": 18001
    },
    {
      "epoch": 0.7267662494953573,
      "grad_norm": 65.7498779296875,
      "learning_rate": 3.6668759892896866e-06,
      "loss": 5.9368,
      "step": 18002
    },
    {
      "epoch": 0.726806620912394,
      "grad_norm": 110.35144805908203,
      "learning_rate": 3.665864111872859e-06,
      "loss": 5.8172,
      "step": 18003
    },
    {
      "epoch": 0.7268469923294307,
      "grad_norm": 98.38239288330078,
      "learning_rate": 3.6648523427549642e-06,
      "loss": 4.1251,
      "step": 18004
    },
    {
      "epoch": 0.7268873637464675,
      "grad_norm": 102.27417755126953,
      "learning_rate": 3.6638406819533013e-06,
      "loss": 4.9305,
      "step": 18005
    },
    {
      "epoch": 0.7269277351635043,
      "grad_norm": 71.2011947631836,
      "learning_rate": 3.662829129485169e-06,
      "loss": 5.0657,
      "step": 18006
    },
    {
      "epoch": 0.726968106580541,
      "grad_norm": 92.56856536865234,
      "learning_rate": 3.661817685367858e-06,
      "loss": 5.6239,
      "step": 18007
    },
    {
      "epoch": 0.7270084779975777,
      "grad_norm": 125.72632598876953,
      "learning_rate": 3.660806349618664e-06,
      "loss": 4.1711,
      "step": 18008
    },
    {
      "epoch": 0.7270488494146145,
      "grad_norm": 76.73615264892578,
      "learning_rate": 3.6597951222548777e-06,
      "loss": 4.0222,
      "step": 18009
    },
    {
      "epoch": 0.7270892208316512,
      "grad_norm": 121.9287338256836,
      "learning_rate": 3.6587840032937926e-06,
      "loss": 6.1712,
      "step": 18010
    },
    {
      "epoch": 0.7271295922486879,
      "grad_norm": 113.43319702148438,
      "learning_rate": 3.6577729927526894e-06,
      "loss": 3.8716,
      "step": 18011
    },
    {
      "epoch": 0.7271699636657246,
      "grad_norm": 214.79754638671875,
      "learning_rate": 3.6567620906488588e-06,
      "loss": 4.5004,
      "step": 18012
    },
    {
      "epoch": 0.7272103350827614,
      "grad_norm": 95.76398468017578,
      "learning_rate": 3.6557512969995846e-06,
      "loss": 6.3166,
      "step": 18013
    },
    {
      "epoch": 0.7272507064997982,
      "grad_norm": 87.95272064208984,
      "learning_rate": 3.6547406118221505e-06,
      "loss": 4.1496,
      "step": 18014
    },
    {
      "epoch": 0.7272910779168349,
      "grad_norm": 130.54222106933594,
      "learning_rate": 3.6537300351338313e-06,
      "loss": 6.2885,
      "step": 18015
    },
    {
      "epoch": 0.7273314493338716,
      "grad_norm": 112.42909240722656,
      "learning_rate": 3.652719566951909e-06,
      "loss": 4.7901,
      "step": 18016
    },
    {
      "epoch": 0.7273718207509083,
      "grad_norm": 88.59220123291016,
      "learning_rate": 3.65170920729366e-06,
      "loss": 4.1591,
      "step": 18017
    },
    {
      "epoch": 0.7274121921679451,
      "grad_norm": 112.13987731933594,
      "learning_rate": 3.6506989561763595e-06,
      "loss": 4.9349,
      "step": 18018
    },
    {
      "epoch": 0.7274525635849818,
      "grad_norm": 76.9133529663086,
      "learning_rate": 3.6496888136172803e-06,
      "loss": 3.9494,
      "step": 18019
    },
    {
      "epoch": 0.7274929350020186,
      "grad_norm": 78.81522369384766,
      "learning_rate": 3.6486787796336942e-06,
      "loss": 3.2284,
      "step": 18020
    },
    {
      "epoch": 0.7275333064190553,
      "grad_norm": 80.08096313476562,
      "learning_rate": 3.6476688542428686e-06,
      "loss": 4.7833,
      "step": 18021
    },
    {
      "epoch": 0.7275736778360921,
      "grad_norm": 160.62173461914062,
      "learning_rate": 3.646659037462075e-06,
      "loss": 4.7982,
      "step": 18022
    },
    {
      "epoch": 0.7276140492531288,
      "grad_norm": 79.16381072998047,
      "learning_rate": 3.645649329308573e-06,
      "loss": 4.6506,
      "step": 18023
    },
    {
      "epoch": 0.7276544206701655,
      "grad_norm": 81.25304412841797,
      "learning_rate": 3.644639729799628e-06,
      "loss": 4.3058,
      "step": 18024
    },
    {
      "epoch": 0.7276947920872022,
      "grad_norm": 94.97905731201172,
      "learning_rate": 3.6436302389525037e-06,
      "loss": 7.4364,
      "step": 18025
    },
    {
      "epoch": 0.727735163504239,
      "grad_norm": 97.47245025634766,
      "learning_rate": 3.642620856784462e-06,
      "loss": 5.7425,
      "step": 18026
    },
    {
      "epoch": 0.7277755349212758,
      "grad_norm": 85.84269714355469,
      "learning_rate": 3.6416115833127553e-06,
      "loss": 4.6702,
      "step": 18027
    },
    {
      "epoch": 0.7278159063383125,
      "grad_norm": 93.8109359741211,
      "learning_rate": 3.640602418554642e-06,
      "loss": 5.7917,
      "step": 18028
    },
    {
      "epoch": 0.7278562777553492,
      "grad_norm": 176.88592529296875,
      "learning_rate": 3.6395933625273772e-06,
      "loss": 6.5686,
      "step": 18029
    },
    {
      "epoch": 0.727896649172386,
      "grad_norm": 87.10833740234375,
      "learning_rate": 3.6385844152482163e-06,
      "loss": 4.3281,
      "step": 18030
    },
    {
      "epoch": 0.7279370205894227,
      "grad_norm": 127.56300354003906,
      "learning_rate": 3.6375755767344047e-06,
      "loss": 3.9688,
      "step": 18031
    },
    {
      "epoch": 0.7279773920064594,
      "grad_norm": 90.22537994384766,
      "learning_rate": 3.636566847003192e-06,
      "loss": 4.5143,
      "step": 18032
    },
    {
      "epoch": 0.7280177634234961,
      "grad_norm": 76.26840209960938,
      "learning_rate": 3.6355582260718282e-06,
      "loss": 5.6703,
      "step": 18033
    },
    {
      "epoch": 0.7280581348405329,
      "grad_norm": 173.7085723876953,
      "learning_rate": 3.6345497139575546e-06,
      "loss": 5.4569,
      "step": 18034
    },
    {
      "epoch": 0.7280985062575697,
      "grad_norm": 76.1874771118164,
      "learning_rate": 3.6335413106776185e-06,
      "loss": 3.1968,
      "step": 18035
    },
    {
      "epoch": 0.7281388776746064,
      "grad_norm": 103.5145492553711,
      "learning_rate": 3.6325330162492613e-06,
      "loss": 5.4786,
      "step": 18036
    },
    {
      "epoch": 0.7281792490916431,
      "grad_norm": 140.96043395996094,
      "learning_rate": 3.6315248306897177e-06,
      "loss": 4.7745,
      "step": 18037
    },
    {
      "epoch": 0.7282196205086798,
      "grad_norm": 83.7674789428711,
      "learning_rate": 3.6305167540162277e-06,
      "loss": 4.8336,
      "step": 18038
    },
    {
      "epoch": 0.7282599919257166,
      "grad_norm": 84.16117858886719,
      "learning_rate": 3.6295087862460275e-06,
      "loss": 4.6052,
      "step": 18039
    },
    {
      "epoch": 0.7283003633427533,
      "grad_norm": 69.20670318603516,
      "learning_rate": 3.628500927396351e-06,
      "loss": 4.7191,
      "step": 18040
    },
    {
      "epoch": 0.7283407347597901,
      "grad_norm": 73.168212890625,
      "learning_rate": 3.62749317748443e-06,
      "loss": 4.5071,
      "step": 18041
    },
    {
      "epoch": 0.7283811061768268,
      "grad_norm": 80.82475280761719,
      "learning_rate": 3.6264855365274986e-06,
      "loss": 3.8589,
      "step": 18042
    },
    {
      "epoch": 0.7284214775938636,
      "grad_norm": 106.34104919433594,
      "learning_rate": 3.625478004542777e-06,
      "loss": 4.931,
      "step": 18043
    },
    {
      "epoch": 0.7284618490109003,
      "grad_norm": 103.94633483886719,
      "learning_rate": 3.6244705815474968e-06,
      "loss": 4.9451,
      "step": 18044
    },
    {
      "epoch": 0.728502220427937,
      "grad_norm": 103.48503112792969,
      "learning_rate": 3.623463267558881e-06,
      "loss": 5.1133,
      "step": 18045
    },
    {
      "epoch": 0.7285425918449737,
      "grad_norm": 96.72053527832031,
      "learning_rate": 3.622456062594154e-06,
      "loss": 5.6732,
      "step": 18046
    },
    {
      "epoch": 0.7285829632620104,
      "grad_norm": 92.27340698242188,
      "learning_rate": 3.621448966670539e-06,
      "loss": 4.9869,
      "step": 18047
    },
    {
      "epoch": 0.7286233346790473,
      "grad_norm": 86.84000396728516,
      "learning_rate": 3.620441979805247e-06,
      "loss": 6.0458,
      "step": 18048
    },
    {
      "epoch": 0.728663706096084,
      "grad_norm": 193.50564575195312,
      "learning_rate": 3.6194351020155004e-06,
      "loss": 4.0371,
      "step": 18049
    },
    {
      "epoch": 0.7287040775131207,
      "grad_norm": 112.90701293945312,
      "learning_rate": 3.6184283333185146e-06,
      "loss": 5.7382,
      "step": 18050
    },
    {
      "epoch": 0.7287444489301574,
      "grad_norm": 106.72332763671875,
      "learning_rate": 3.617421673731505e-06,
      "loss": 4.7267,
      "step": 18051
    },
    {
      "epoch": 0.7287848203471942,
      "grad_norm": 144.98410034179688,
      "learning_rate": 3.616415123271677e-06,
      "loss": 5.4904,
      "step": 18052
    },
    {
      "epoch": 0.7288251917642309,
      "grad_norm": 107.48487091064453,
      "learning_rate": 3.615408681956244e-06,
      "loss": 5.9752,
      "step": 18053
    },
    {
      "epoch": 0.7288655631812677,
      "grad_norm": 86.4496078491211,
      "learning_rate": 3.614402349802413e-06,
      "loss": 5.5133,
      "step": 18054
    },
    {
      "epoch": 0.7289059345983044,
      "grad_norm": 133.5157012939453,
      "learning_rate": 3.613396126827391e-06,
      "loss": 4.8963,
      "step": 18055
    },
    {
      "epoch": 0.7289463060153412,
      "grad_norm": 119.78353881835938,
      "learning_rate": 3.6123900130483813e-06,
      "loss": 4.1234,
      "step": 18056
    },
    {
      "epoch": 0.7289866774323779,
      "grad_norm": 126.0653305053711,
      "learning_rate": 3.611384008482586e-06,
      "loss": 5.7149,
      "step": 18057
    },
    {
      "epoch": 0.7290270488494146,
      "grad_norm": 119.47489166259766,
      "learning_rate": 3.610378113147205e-06,
      "loss": 4.4627,
      "step": 18058
    },
    {
      "epoch": 0.7290674202664513,
      "grad_norm": 97.0337142944336,
      "learning_rate": 3.6093723270594416e-06,
      "loss": 6.2936,
      "step": 18059
    },
    {
      "epoch": 0.729107791683488,
      "grad_norm": 82.12908172607422,
      "learning_rate": 3.6083666502364844e-06,
      "loss": 4.6524,
      "step": 18060
    },
    {
      "epoch": 0.7291481631005249,
      "grad_norm": 168.4633331298828,
      "learning_rate": 3.607361082695533e-06,
      "loss": 5.412,
      "step": 18061
    },
    {
      "epoch": 0.7291885345175616,
      "grad_norm": 93.69996643066406,
      "learning_rate": 3.606355624453778e-06,
      "loss": 5.4853,
      "step": 18062
    },
    {
      "epoch": 0.7292289059345983,
      "grad_norm": 114.74527740478516,
      "learning_rate": 3.605350275528415e-06,
      "loss": 5.6111,
      "step": 18063
    },
    {
      "epoch": 0.729269277351635,
      "grad_norm": 108.36177825927734,
      "learning_rate": 3.6043450359366274e-06,
      "loss": 5.9294,
      "step": 18064
    },
    {
      "epoch": 0.7293096487686718,
      "grad_norm": 111.32461547851562,
      "learning_rate": 3.6033399056956043e-06,
      "loss": 4.7039,
      "step": 18065
    },
    {
      "epoch": 0.7293500201857085,
      "grad_norm": 530.3883056640625,
      "learning_rate": 3.6023348848225305e-06,
      "loss": 6.1543,
      "step": 18066
    },
    {
      "epoch": 0.7293903916027452,
      "grad_norm": 104.03025817871094,
      "learning_rate": 3.6013299733345953e-06,
      "loss": 3.745,
      "step": 18067
    },
    {
      "epoch": 0.729430763019782,
      "grad_norm": 140.92593383789062,
      "learning_rate": 3.6003251712489718e-06,
      "loss": 5.8173,
      "step": 18068
    },
    {
      "epoch": 0.7294711344368188,
      "grad_norm": 76.6927490234375,
      "learning_rate": 3.599320478582843e-06,
      "loss": 3.8868,
      "step": 18069
    },
    {
      "epoch": 0.7295115058538555,
      "grad_norm": 83.60400390625,
      "learning_rate": 3.598315895353388e-06,
      "loss": 6.3921,
      "step": 18070
    },
    {
      "epoch": 0.7295518772708922,
      "grad_norm": 104.6292495727539,
      "learning_rate": 3.597311421577785e-06,
      "loss": 3.488,
      "step": 18071
    },
    {
      "epoch": 0.7295922486879289,
      "grad_norm": 94.30764770507812,
      "learning_rate": 3.5963070572732027e-06,
      "loss": 4.4685,
      "step": 18072
    },
    {
      "epoch": 0.7296326201049657,
      "grad_norm": 104.99429321289062,
      "learning_rate": 3.5953028024568158e-06,
      "loss": 5.4687,
      "step": 18073
    },
    {
      "epoch": 0.7296729915220024,
      "grad_norm": 67.77989196777344,
      "learning_rate": 3.594298657145794e-06,
      "loss": 5.2724,
      "step": 18074
    },
    {
      "epoch": 0.7297133629390392,
      "grad_norm": 81.05203247070312,
      "learning_rate": 3.5932946213573084e-06,
      "loss": 4.3299,
      "step": 18075
    },
    {
      "epoch": 0.7297537343560759,
      "grad_norm": 115.96562194824219,
      "learning_rate": 3.5922906951085224e-06,
      "loss": 5.017,
      "step": 18076
    },
    {
      "epoch": 0.7297941057731127,
      "grad_norm": 104.38267517089844,
      "learning_rate": 3.5912868784166033e-06,
      "loss": 5.6628,
      "step": 18077
    },
    {
      "epoch": 0.7298344771901494,
      "grad_norm": 76.68762969970703,
      "learning_rate": 3.590283171298713e-06,
      "loss": 4.3301,
      "step": 18078
    },
    {
      "epoch": 0.7298748486071861,
      "grad_norm": 185.80482482910156,
      "learning_rate": 3.5892795737720163e-06,
      "loss": 4.7763,
      "step": 18079
    },
    {
      "epoch": 0.7299152200242228,
      "grad_norm": 88.88678741455078,
      "learning_rate": 3.5882760858536656e-06,
      "loss": 5.5968,
      "step": 18080
    },
    {
      "epoch": 0.7299555914412595,
      "grad_norm": 101.44071197509766,
      "learning_rate": 3.5872727075608207e-06,
      "loss": 5.3131,
      "step": 18081
    },
    {
      "epoch": 0.7299959628582964,
      "grad_norm": 93.76854705810547,
      "learning_rate": 3.5862694389106377e-06,
      "loss": 4.9613,
      "step": 18082
    },
    {
      "epoch": 0.7300363342753331,
      "grad_norm": 102.1598892211914,
      "learning_rate": 3.585266279920274e-06,
      "loss": 3.0681,
      "step": 18083
    },
    {
      "epoch": 0.7300767056923698,
      "grad_norm": 106.72230529785156,
      "learning_rate": 3.584263230606875e-06,
      "loss": 4.0329,
      "step": 18084
    },
    {
      "epoch": 0.7301170771094065,
      "grad_norm": 109.87185668945312,
      "learning_rate": 3.5832602909875915e-06,
      "loss": 5.2358,
      "step": 18085
    },
    {
      "epoch": 0.7301574485264433,
      "grad_norm": 81.16600036621094,
      "learning_rate": 3.5822574610795725e-06,
      "loss": 5.4235,
      "step": 18086
    },
    {
      "epoch": 0.73019781994348,
      "grad_norm": 90.07756805419922,
      "learning_rate": 3.5812547408999688e-06,
      "loss": 5.6849,
      "step": 18087
    },
    {
      "epoch": 0.7302381913605167,
      "grad_norm": 94.36962127685547,
      "learning_rate": 3.5802521304659164e-06,
      "loss": 5.0301,
      "step": 18088
    },
    {
      "epoch": 0.7302785627775535,
      "grad_norm": 72.66909790039062,
      "learning_rate": 3.5792496297945613e-06,
      "loss": 5.5868,
      "step": 18089
    },
    {
      "epoch": 0.7303189341945903,
      "grad_norm": 80.9309310913086,
      "learning_rate": 3.578247238903044e-06,
      "loss": 5.1395,
      "step": 18090
    },
    {
      "epoch": 0.730359305611627,
      "grad_norm": 110.39935302734375,
      "learning_rate": 3.5772449578085067e-06,
      "loss": 3.9355,
      "step": 18091
    },
    {
      "epoch": 0.7303996770286637,
      "grad_norm": 85.40391540527344,
      "learning_rate": 3.576242786528079e-06,
      "loss": 4.1223,
      "step": 18092
    },
    {
      "epoch": 0.7304400484457004,
      "grad_norm": 82.14274597167969,
      "learning_rate": 3.575240725078899e-06,
      "loss": 5.1238,
      "step": 18093
    },
    {
      "epoch": 0.7304804198627372,
      "grad_norm": 105.60936737060547,
      "learning_rate": 3.5742387734781005e-06,
      "loss": 4.0486,
      "step": 18094
    },
    {
      "epoch": 0.7305207912797739,
      "grad_norm": 135.35824584960938,
      "learning_rate": 3.573236931742813e-06,
      "loss": 4.5001,
      "step": 18095
    },
    {
      "epoch": 0.7305611626968107,
      "grad_norm": 75.45301818847656,
      "learning_rate": 3.572235199890166e-06,
      "loss": 3.5778,
      "step": 18096
    },
    {
      "epoch": 0.7306015341138474,
      "grad_norm": 94.14761352539062,
      "learning_rate": 3.5712335779372876e-06,
      "loss": 6.724,
      "step": 18097
    },
    {
      "epoch": 0.7306419055308842,
      "grad_norm": 87.08666229248047,
      "learning_rate": 3.5702320659013022e-06,
      "loss": 4.7499,
      "step": 18098
    },
    {
      "epoch": 0.7306822769479209,
      "grad_norm": 148.40342712402344,
      "learning_rate": 3.569230663799338e-06,
      "loss": 3.4516,
      "step": 18099
    },
    {
      "epoch": 0.7307226483649576,
      "grad_norm": 82.26408386230469,
      "learning_rate": 3.5682293716485095e-06,
      "loss": 4.7888,
      "step": 18100
    },
    {
      "epoch": 0.7307630197819943,
      "grad_norm": 95.16742706298828,
      "learning_rate": 3.567228189465939e-06,
      "loss": 6.1947,
      "step": 18101
    },
    {
      "epoch": 0.730803391199031,
      "grad_norm": 142.14761352539062,
      "learning_rate": 3.5662271172687448e-06,
      "loss": 5.6162,
      "step": 18102
    },
    {
      "epoch": 0.7308437626160679,
      "grad_norm": 81.56121063232422,
      "learning_rate": 3.5652261550740462e-06,
      "loss": 5.5031,
      "step": 18103
    },
    {
      "epoch": 0.7308841340331046,
      "grad_norm": 103.00833892822266,
      "learning_rate": 3.5642253028989503e-06,
      "loss": 4.2319,
      "step": 18104
    },
    {
      "epoch": 0.7309245054501413,
      "grad_norm": 96.12062072753906,
      "learning_rate": 3.5632245607605744e-06,
      "loss": 5.5406,
      "step": 18105
    },
    {
      "epoch": 0.730964876867178,
      "grad_norm": 83.19917297363281,
      "learning_rate": 3.5622239286760263e-06,
      "loss": 4.0771,
      "step": 18106
    },
    {
      "epoch": 0.7310052482842148,
      "grad_norm": 77.60733032226562,
      "learning_rate": 3.5612234066624195e-06,
      "loss": 4.9288,
      "step": 18107
    },
    {
      "epoch": 0.7310456197012515,
      "grad_norm": 87.40750122070312,
      "learning_rate": 3.560222994736854e-06,
      "loss": 4.712,
      "step": 18108
    },
    {
      "epoch": 0.7310859911182882,
      "grad_norm": 141.2601776123047,
      "learning_rate": 3.5592226929164366e-06,
      "loss": 2.9807,
      "step": 18109
    },
    {
      "epoch": 0.731126362535325,
      "grad_norm": 111.46031188964844,
      "learning_rate": 3.5582225012182714e-06,
      "loss": 6.0959,
      "step": 18110
    },
    {
      "epoch": 0.7311667339523618,
      "grad_norm": 75.37986755371094,
      "learning_rate": 3.557222419659462e-06,
      "loss": 5.0194,
      "step": 18111
    },
    {
      "epoch": 0.7312071053693985,
      "grad_norm": 114.52542114257812,
      "learning_rate": 3.556222448257101e-06,
      "loss": 4.0022,
      "step": 18112
    },
    {
      "epoch": 0.7312474767864352,
      "grad_norm": 92.60711669921875,
      "learning_rate": 3.5552225870282877e-06,
      "loss": 5.3463,
      "step": 18113
    },
    {
      "epoch": 0.7312878482034719,
      "grad_norm": 103.48616027832031,
      "learning_rate": 3.55422283599012e-06,
      "loss": 3.9789,
      "step": 18114
    },
    {
      "epoch": 0.7313282196205086,
      "grad_norm": 68.95068359375,
      "learning_rate": 3.5532231951596887e-06,
      "loss": 5.3228,
      "step": 18115
    },
    {
      "epoch": 0.7313685910375454,
      "grad_norm": 76.89649963378906,
      "learning_rate": 3.5522236645540874e-06,
      "loss": 4.6104,
      "step": 18116
    },
    {
      "epoch": 0.7314089624545822,
      "grad_norm": 102.46253204345703,
      "learning_rate": 3.5512242441904044e-06,
      "loss": 6.7281,
      "step": 18117
    },
    {
      "epoch": 0.7314493338716189,
      "grad_norm": 94.44284057617188,
      "learning_rate": 3.5502249340857277e-06,
      "loss": 5.6071,
      "step": 18118
    },
    {
      "epoch": 0.7314897052886556,
      "grad_norm": 90.58264923095703,
      "learning_rate": 3.5492257342571467e-06,
      "loss": 4.7903,
      "step": 18119
    },
    {
      "epoch": 0.7315300767056924,
      "grad_norm": 84.32487487792969,
      "learning_rate": 3.5482266447217384e-06,
      "loss": 5.0252,
      "step": 18120
    },
    {
      "epoch": 0.7315704481227291,
      "grad_norm": 88.86888122558594,
      "learning_rate": 3.54722766549659e-06,
      "loss": 4.0175,
      "step": 18121
    },
    {
      "epoch": 0.7316108195397658,
      "grad_norm": 98.65170288085938,
      "learning_rate": 3.5462287965987785e-06,
      "loss": 5.4279,
      "step": 18122
    },
    {
      "epoch": 0.7316511909568025,
      "grad_norm": 66.10623931884766,
      "learning_rate": 3.5452300380453873e-06,
      "loss": 5.0229,
      "step": 18123
    },
    {
      "epoch": 0.7316915623738394,
      "grad_norm": 82.2586669921875,
      "learning_rate": 3.5442313898534875e-06,
      "loss": 3.5747,
      "step": 18124
    },
    {
      "epoch": 0.7317319337908761,
      "grad_norm": 79.86699676513672,
      "learning_rate": 3.543232852040155e-06,
      "loss": 4.6962,
      "step": 18125
    },
    {
      "epoch": 0.7317723052079128,
      "grad_norm": 116.25892639160156,
      "learning_rate": 3.5422344246224626e-06,
      "loss": 5.8137,
      "step": 18126
    },
    {
      "epoch": 0.7318126766249495,
      "grad_norm": 83.84918212890625,
      "learning_rate": 3.541236107617485e-06,
      "loss": 5.0969,
      "step": 18127
    },
    {
      "epoch": 0.7318530480419863,
      "grad_norm": 93.3967056274414,
      "learning_rate": 3.540237901042285e-06,
      "loss": 6.0622,
      "step": 18128
    },
    {
      "epoch": 0.731893419459023,
      "grad_norm": 71.5904769897461,
      "learning_rate": 3.539239804913932e-06,
      "loss": 5.218,
      "step": 18129
    },
    {
      "epoch": 0.7319337908760597,
      "grad_norm": 88.69476318359375,
      "learning_rate": 3.5382418192494906e-06,
      "loss": 3.9113,
      "step": 18130
    },
    {
      "epoch": 0.7319741622930965,
      "grad_norm": 118.4410400390625,
      "learning_rate": 3.5372439440660287e-06,
      "loss": 4.4712,
      "step": 18131
    },
    {
      "epoch": 0.7320145337101333,
      "grad_norm": 105.7321548461914,
      "learning_rate": 3.5362461793806003e-06,
      "loss": 6.1621,
      "step": 18132
    },
    {
      "epoch": 0.73205490512717,
      "grad_norm": 118.3463134765625,
      "learning_rate": 3.5352485252102685e-06,
      "loss": 5.1272,
      "step": 18133
    },
    {
      "epoch": 0.7320952765442067,
      "grad_norm": 122.05208587646484,
      "learning_rate": 3.5342509815720904e-06,
      "loss": 5.5136,
      "step": 18134
    },
    {
      "epoch": 0.7321356479612434,
      "grad_norm": 93.15264129638672,
      "learning_rate": 3.533253548483123e-06,
      "loss": 3.7455,
      "step": 18135
    },
    {
      "epoch": 0.7321760193782801,
      "grad_norm": 116.38140869140625,
      "learning_rate": 3.5322562259604176e-06,
      "loss": 5.7886,
      "step": 18136
    },
    {
      "epoch": 0.7322163907953169,
      "grad_norm": 127.04289245605469,
      "learning_rate": 3.5312590140210278e-06,
      "loss": 5.5791,
      "step": 18137
    },
    {
      "epoch": 0.7322567622123537,
      "grad_norm": 91.80952453613281,
      "learning_rate": 3.5302619126820036e-06,
      "loss": 4.4382,
      "step": 18138
    },
    {
      "epoch": 0.7322971336293904,
      "grad_norm": 90.17301940917969,
      "learning_rate": 3.529264921960396e-06,
      "loss": 5.1667,
      "step": 18139
    },
    {
      "epoch": 0.7323375050464271,
      "grad_norm": 124.82711029052734,
      "learning_rate": 3.528268041873244e-06,
      "loss": 4.9783,
      "step": 18140
    },
    {
      "epoch": 0.7323778764634639,
      "grad_norm": 83.41763305664062,
      "learning_rate": 3.5272712724375968e-06,
      "loss": 4.439,
      "step": 18141
    },
    {
      "epoch": 0.7324182478805006,
      "grad_norm": 70.69956970214844,
      "learning_rate": 3.526274613670496e-06,
      "loss": 3.2427,
      "step": 18142
    },
    {
      "epoch": 0.7324586192975373,
      "grad_norm": 112.2885971069336,
      "learning_rate": 3.525278065588985e-06,
      "loss": 4.4534,
      "step": 18143
    },
    {
      "epoch": 0.732498990714574,
      "grad_norm": 92.70172882080078,
      "learning_rate": 3.5242816282100956e-06,
      "loss": 3.5021,
      "step": 18144
    },
    {
      "epoch": 0.7325393621316109,
      "grad_norm": 96.15994262695312,
      "learning_rate": 3.5232853015508693e-06,
      "loss": 4.4818,
      "step": 18145
    },
    {
      "epoch": 0.7325797335486476,
      "grad_norm": 426.2038879394531,
      "learning_rate": 3.52228908562834e-06,
      "loss": 4.593,
      "step": 18146
    },
    {
      "epoch": 0.7326201049656843,
      "grad_norm": 166.62835693359375,
      "learning_rate": 3.521292980459544e-06,
      "loss": 4.651,
      "step": 18147
    },
    {
      "epoch": 0.732660476382721,
      "grad_norm": 71.31587219238281,
      "learning_rate": 3.5202969860615055e-06,
      "loss": 6.6932,
      "step": 18148
    },
    {
      "epoch": 0.7327008477997577,
      "grad_norm": 128.38250732421875,
      "learning_rate": 3.519301102451258e-06,
      "loss": 4.9602,
      "step": 18149
    },
    {
      "epoch": 0.7327412192167945,
      "grad_norm": 106.4964599609375,
      "learning_rate": 3.518305329645828e-06,
      "loss": 5.9969,
      "step": 18150
    },
    {
      "epoch": 0.7327815906338312,
      "grad_norm": 85.05279541015625,
      "learning_rate": 3.5173096676622444e-06,
      "loss": 4.5091,
      "step": 18151
    },
    {
      "epoch": 0.732821962050868,
      "grad_norm": 100.09090423583984,
      "learning_rate": 3.516314116517523e-06,
      "loss": 6.4247,
      "step": 18152
    },
    {
      "epoch": 0.7328623334679047,
      "grad_norm": 87.89404296875,
      "learning_rate": 3.5153186762286904e-06,
      "loss": 5.0193,
      "step": 18153
    },
    {
      "epoch": 0.7329027048849415,
      "grad_norm": 113.31315612792969,
      "learning_rate": 3.514323346812766e-06,
      "loss": 5.4131,
      "step": 18154
    },
    {
      "epoch": 0.7329430763019782,
      "grad_norm": 83.0095443725586,
      "learning_rate": 3.513328128286767e-06,
      "loss": 4.7147,
      "step": 18155
    },
    {
      "epoch": 0.7329834477190149,
      "grad_norm": 76.33621215820312,
      "learning_rate": 3.5123330206677086e-06,
      "loss": 4.6945,
      "step": 18156
    },
    {
      "epoch": 0.7330238191360516,
      "grad_norm": 144.18966674804688,
      "learning_rate": 3.511338023972607e-06,
      "loss": 4.8148,
      "step": 18157
    },
    {
      "epoch": 0.7330641905530885,
      "grad_norm": 155.38246154785156,
      "learning_rate": 3.5103431382184727e-06,
      "loss": 4.5453,
      "step": 18158
    },
    {
      "epoch": 0.7331045619701252,
      "grad_norm": 90.60112762451172,
      "learning_rate": 3.509348363422319e-06,
      "loss": 4.9593,
      "step": 18159
    },
    {
      "epoch": 0.7331449333871619,
      "grad_norm": 81.1425552368164,
      "learning_rate": 3.508353699601149e-06,
      "loss": 4.1313,
      "step": 18160
    },
    {
      "epoch": 0.7331853048041986,
      "grad_norm": 630.88330078125,
      "learning_rate": 3.5073591467719713e-06,
      "loss": 5.3044,
      "step": 18161
    },
    {
      "epoch": 0.7332256762212354,
      "grad_norm": 86.09651947021484,
      "learning_rate": 3.5063647049517913e-06,
      "loss": 4.8637,
      "step": 18162
    },
    {
      "epoch": 0.7332660476382721,
      "grad_norm": 82.24443817138672,
      "learning_rate": 3.505370374157614e-06,
      "loss": 5.4863,
      "step": 18163
    },
    {
      "epoch": 0.7333064190553088,
      "grad_norm": 68.81047058105469,
      "learning_rate": 3.5043761544064335e-06,
      "loss": 4.3,
      "step": 18164
    },
    {
      "epoch": 0.7333467904723456,
      "grad_norm": 94.14801788330078,
      "learning_rate": 3.5033820457152534e-06,
      "loss": 4.9242,
      "step": 18165
    },
    {
      "epoch": 0.7333871618893824,
      "grad_norm": 91.72529602050781,
      "learning_rate": 3.5023880481010696e-06,
      "loss": 5.6737,
      "step": 18166
    },
    {
      "epoch": 0.7334275333064191,
      "grad_norm": 92.69254302978516,
      "learning_rate": 3.50139416158088e-06,
      "loss": 4.3508,
      "step": 18167
    },
    {
      "epoch": 0.7334679047234558,
      "grad_norm": 107.50543212890625,
      "learning_rate": 3.5004003861716706e-06,
      "loss": 5.3083,
      "step": 18168
    },
    {
      "epoch": 0.7335082761404925,
      "grad_norm": 90.42256164550781,
      "learning_rate": 3.4994067218904383e-06,
      "loss": 4.3456,
      "step": 18169
    },
    {
      "epoch": 0.7335486475575292,
      "grad_norm": 99.72090911865234,
      "learning_rate": 3.498413168754171e-06,
      "loss": 3.1286,
      "step": 18170
    },
    {
      "epoch": 0.733589018974566,
      "grad_norm": 83.50874328613281,
      "learning_rate": 3.4974197267798594e-06,
      "loss": 4.1726,
      "step": 18171
    },
    {
      "epoch": 0.7336293903916028,
      "grad_norm": 106.29731750488281,
      "learning_rate": 3.4964263959844825e-06,
      "loss": 5.7859,
      "step": 18172
    },
    {
      "epoch": 0.7336697618086395,
      "grad_norm": 103.21289825439453,
      "learning_rate": 3.495433176385027e-06,
      "loss": 5.6746,
      "step": 18173
    },
    {
      "epoch": 0.7337101332256762,
      "grad_norm": 86.13014221191406,
      "learning_rate": 3.4944400679984745e-06,
      "loss": 4.943,
      "step": 18174
    },
    {
      "epoch": 0.733750504642713,
      "grad_norm": 105.81267547607422,
      "learning_rate": 3.4934470708418056e-06,
      "loss": 5.2885,
      "step": 18175
    },
    {
      "epoch": 0.7337908760597497,
      "grad_norm": 93.56062316894531,
      "learning_rate": 3.4924541849319982e-06,
      "loss": 4.2454,
      "step": 18176
    },
    {
      "epoch": 0.7338312474767864,
      "grad_norm": 97.49480438232422,
      "learning_rate": 3.4914614102860277e-06,
      "loss": 5.5977,
      "step": 18177
    },
    {
      "epoch": 0.7338716188938231,
      "grad_norm": 160.67620849609375,
      "learning_rate": 3.49046874692087e-06,
      "loss": 4.9274,
      "step": 18178
    },
    {
      "epoch": 0.73391199031086,
      "grad_norm": 117.27987670898438,
      "learning_rate": 3.489476194853497e-06,
      "loss": 4.3129,
      "step": 18179
    },
    {
      "epoch": 0.7339523617278967,
      "grad_norm": 111.4631118774414,
      "learning_rate": 3.4884837541008764e-06,
      "loss": 4.3545,
      "step": 18180
    },
    {
      "epoch": 0.7339927331449334,
      "grad_norm": 115.48921966552734,
      "learning_rate": 3.487491424679977e-06,
      "loss": 4.2833,
      "step": 18181
    },
    {
      "epoch": 0.7340331045619701,
      "grad_norm": 121.50566864013672,
      "learning_rate": 3.4864992066077663e-06,
      "loss": 6.0369,
      "step": 18182
    },
    {
      "epoch": 0.7340734759790069,
      "grad_norm": 116.26089477539062,
      "learning_rate": 3.485507099901213e-06,
      "loss": 5.2586,
      "step": 18183
    },
    {
      "epoch": 0.7341138473960436,
      "grad_norm": 108.23125457763672,
      "learning_rate": 3.484515104577273e-06,
      "loss": 4.2452,
      "step": 18184
    },
    {
      "epoch": 0.7341542188130803,
      "grad_norm": 93.98872375488281,
      "learning_rate": 3.4835232206529092e-06,
      "loss": 4.2815,
      "step": 18185
    },
    {
      "epoch": 0.7341945902301171,
      "grad_norm": 104.70662689208984,
      "learning_rate": 3.4825314481450813e-06,
      "loss": 6.1476,
      "step": 18186
    },
    {
      "epoch": 0.7342349616471538,
      "grad_norm": 133.17506408691406,
      "learning_rate": 3.481539787070749e-06,
      "loss": 5.37,
      "step": 18187
    },
    {
      "epoch": 0.7342753330641906,
      "grad_norm": 79.08183288574219,
      "learning_rate": 3.4805482374468623e-06,
      "loss": 3.6404,
      "step": 18188
    },
    {
      "epoch": 0.7343157044812273,
      "grad_norm": 117.91228485107422,
      "learning_rate": 3.4795567992903767e-06,
      "loss": 5.3194,
      "step": 18189
    },
    {
      "epoch": 0.734356075898264,
      "grad_norm": 69.71687316894531,
      "learning_rate": 3.4785654726182427e-06,
      "loss": 4.6395,
      "step": 18190
    },
    {
      "epoch": 0.7343964473153007,
      "grad_norm": 148.08935546875,
      "learning_rate": 3.4775742574474147e-06,
      "loss": 5.5674,
      "step": 18191
    },
    {
      "epoch": 0.7344368187323375,
      "grad_norm": 82.35719299316406,
      "learning_rate": 3.4765831537948325e-06,
      "loss": 3.8433,
      "step": 18192
    },
    {
      "epoch": 0.7344771901493743,
      "grad_norm": 120.21810150146484,
      "learning_rate": 3.4755921616774447e-06,
      "loss": 5.4381,
      "step": 18193
    },
    {
      "epoch": 0.734517561566411,
      "grad_norm": 105.14202880859375,
      "learning_rate": 3.4746012811121955e-06,
      "loss": 3.0924,
      "step": 18194
    },
    {
      "epoch": 0.7345579329834477,
      "grad_norm": 127.0068359375,
      "learning_rate": 3.473610512116026e-06,
      "loss": 4.303,
      "step": 18195
    },
    {
      "epoch": 0.7345983044004845,
      "grad_norm": 99.85395812988281,
      "learning_rate": 3.4726198547058764e-06,
      "loss": 5.2017,
      "step": 18196
    },
    {
      "epoch": 0.7346386758175212,
      "grad_norm": 112.23309326171875,
      "learning_rate": 3.4716293088986862e-06,
      "loss": 5.6861,
      "step": 18197
    },
    {
      "epoch": 0.7346790472345579,
      "grad_norm": 76.13556671142578,
      "learning_rate": 3.4706388747113883e-06,
      "loss": 4.4235,
      "step": 18198
    },
    {
      "epoch": 0.7347194186515946,
      "grad_norm": 74.81914520263672,
      "learning_rate": 3.4696485521609237e-06,
      "loss": 4.1362,
      "step": 18199
    },
    {
      "epoch": 0.7347597900686315,
      "grad_norm": 113.07313537597656,
      "learning_rate": 3.468658341264214e-06,
      "loss": 3.8934,
      "step": 18200
    },
    {
      "epoch": 0.7348001614856682,
      "grad_norm": 85.77615356445312,
      "learning_rate": 3.4676682420381957e-06,
      "loss": 5.6891,
      "step": 18201
    },
    {
      "epoch": 0.7348405329027049,
      "grad_norm": 133.37648010253906,
      "learning_rate": 3.4666782544997968e-06,
      "loss": 3.9622,
      "step": 18202
    },
    {
      "epoch": 0.7348809043197416,
      "grad_norm": 84.50895690917969,
      "learning_rate": 3.4656883786659467e-06,
      "loss": 5.1703,
      "step": 18203
    },
    {
      "epoch": 0.7349212757367783,
      "grad_norm": 133.7012176513672,
      "learning_rate": 3.4646986145535635e-06,
      "loss": 5.1609,
      "step": 18204
    },
    {
      "epoch": 0.7349616471538151,
      "grad_norm": 114.82904052734375,
      "learning_rate": 3.4637089621795715e-06,
      "loss": 1.6927,
      "step": 18205
    },
    {
      "epoch": 0.7350020185708518,
      "grad_norm": 74.40596771240234,
      "learning_rate": 3.4627194215608937e-06,
      "loss": 5.1322,
      "step": 18206
    },
    {
      "epoch": 0.7350423899878886,
      "grad_norm": 86.80769348144531,
      "learning_rate": 3.461729992714452e-06,
      "loss": 6.6272,
      "step": 18207
    },
    {
      "epoch": 0.7350827614049253,
      "grad_norm": 74.03765106201172,
      "learning_rate": 3.460740675657156e-06,
      "loss": 3.4249,
      "step": 18208
    },
    {
      "epoch": 0.7351231328219621,
      "grad_norm": 93.82564544677734,
      "learning_rate": 3.4597514704059243e-06,
      "loss": 4.9622,
      "step": 18209
    },
    {
      "epoch": 0.7351635042389988,
      "grad_norm": 129.70352172851562,
      "learning_rate": 3.458762376977669e-06,
      "loss": 4.8885,
      "step": 18210
    },
    {
      "epoch": 0.7352038756560355,
      "grad_norm": 117.6334228515625,
      "learning_rate": 3.457773395389307e-06,
      "loss": 5.0467,
      "step": 18211
    },
    {
      "epoch": 0.7352442470730722,
      "grad_norm": 78.51325225830078,
      "learning_rate": 3.456784525657737e-06,
      "loss": 5.318,
      "step": 18212
    },
    {
      "epoch": 0.735284618490109,
      "grad_norm": 81.22012329101562,
      "learning_rate": 3.4557957677998744e-06,
      "loss": 5.0946,
      "step": 18213
    },
    {
      "epoch": 0.7353249899071458,
      "grad_norm": 244.0416717529297,
      "learning_rate": 3.4548071218326205e-06,
      "loss": 3.7345,
      "step": 18214
    },
    {
      "epoch": 0.7353653613241825,
      "grad_norm": 74.33301544189453,
      "learning_rate": 3.4538185877728825e-06,
      "loss": 2.9613,
      "step": 18215
    },
    {
      "epoch": 0.7354057327412192,
      "grad_norm": 108.41587829589844,
      "learning_rate": 3.4528301656375594e-06,
      "loss": 5.0961,
      "step": 18216
    },
    {
      "epoch": 0.735446104158256,
      "grad_norm": 208.22743225097656,
      "learning_rate": 3.451841855443552e-06,
      "loss": 5.3017,
      "step": 18217
    },
    {
      "epoch": 0.7354864755752927,
      "grad_norm": 154.8599853515625,
      "learning_rate": 3.4508536572077566e-06,
      "loss": 5.0841,
      "step": 18218
    },
    {
      "epoch": 0.7355268469923294,
      "grad_norm": 110.68135070800781,
      "learning_rate": 3.449865570947075e-06,
      "loss": 4.5647,
      "step": 18219
    },
    {
      "epoch": 0.7355672184093661,
      "grad_norm": 116.19918823242188,
      "learning_rate": 3.4488775966783927e-06,
      "loss": 6.0003,
      "step": 18220
    },
    {
      "epoch": 0.735607589826403,
      "grad_norm": 116.99247741699219,
      "learning_rate": 3.4478897344186057e-06,
      "loss": 4.4327,
      "step": 18221
    },
    {
      "epoch": 0.7356479612434397,
      "grad_norm": 88.42815399169922,
      "learning_rate": 3.446901984184604e-06,
      "loss": 4.3473,
      "step": 18222
    },
    {
      "epoch": 0.7356883326604764,
      "grad_norm": 97.19910430908203,
      "learning_rate": 3.445914345993279e-06,
      "loss": 4.3892,
      "step": 18223
    },
    {
      "epoch": 0.7357287040775131,
      "grad_norm": 117.10578155517578,
      "learning_rate": 3.4449268198615102e-06,
      "loss": 5.2168,
      "step": 18224
    },
    {
      "epoch": 0.7357690754945498,
      "grad_norm": 85.98870086669922,
      "learning_rate": 3.443939405806186e-06,
      "loss": 5.088,
      "step": 18225
    },
    {
      "epoch": 0.7358094469115866,
      "grad_norm": 95.16568756103516,
      "learning_rate": 3.442952103844188e-06,
      "loss": 5.4457,
      "step": 18226
    },
    {
      "epoch": 0.7358498183286233,
      "grad_norm": 143.47509765625,
      "learning_rate": 3.4419649139924007e-06,
      "loss": 4.5481,
      "step": 18227
    },
    {
      "epoch": 0.7358901897456601,
      "grad_norm": 92.69002532958984,
      "learning_rate": 3.440977836267696e-06,
      "loss": 6.6736,
      "step": 18228
    },
    {
      "epoch": 0.7359305611626968,
      "grad_norm": 86.80541229248047,
      "learning_rate": 3.4399908706869545e-06,
      "loss": 6.9557,
      "step": 18229
    },
    {
      "epoch": 0.7359709325797336,
      "grad_norm": 64.31966400146484,
      "learning_rate": 3.4390040172670492e-06,
      "loss": 5.2223,
      "step": 18230
    },
    {
      "epoch": 0.7360113039967703,
      "grad_norm": 122.44896697998047,
      "learning_rate": 3.4380172760248587e-06,
      "loss": 4.8677,
      "step": 18231
    },
    {
      "epoch": 0.736051675413807,
      "grad_norm": 84.86624145507812,
      "learning_rate": 3.437030646977245e-06,
      "loss": 4.1006,
      "step": 18232
    },
    {
      "epoch": 0.7360920468308437,
      "grad_norm": 83.63941192626953,
      "learning_rate": 3.436044130141082e-06,
      "loss": 4.8889,
      "step": 18233
    },
    {
      "epoch": 0.7361324182478804,
      "grad_norm": 106.85336303710938,
      "learning_rate": 3.4350577255332363e-06,
      "loss": 5.2371,
      "step": 18234
    },
    {
      "epoch": 0.7361727896649173,
      "grad_norm": 71.27495574951172,
      "learning_rate": 3.4340714331705737e-06,
      "loss": 4.7351,
      "step": 18235
    },
    {
      "epoch": 0.736213161081954,
      "grad_norm": 129.4925079345703,
      "learning_rate": 3.4330852530699564e-06,
      "loss": 7.3111,
      "step": 18236
    },
    {
      "epoch": 0.7362535324989907,
      "grad_norm": 82.63720703125,
      "learning_rate": 3.4320991852482476e-06,
      "loss": 3.3363,
      "step": 18237
    },
    {
      "epoch": 0.7362939039160274,
      "grad_norm": 109.35303497314453,
      "learning_rate": 3.431113229722305e-06,
      "loss": 4.7105,
      "step": 18238
    },
    {
      "epoch": 0.7363342753330642,
      "grad_norm": 106.83343505859375,
      "learning_rate": 3.43012738650899e-06,
      "loss": 4.056,
      "step": 18239
    },
    {
      "epoch": 0.7363746467501009,
      "grad_norm": 164.2075958251953,
      "learning_rate": 3.429141655625151e-06,
      "loss": 5.6151,
      "step": 18240
    },
    {
      "epoch": 0.7364150181671376,
      "grad_norm": 104.57124328613281,
      "learning_rate": 3.4281560370876465e-06,
      "loss": 5.6461,
      "step": 18241
    },
    {
      "epoch": 0.7364553895841744,
      "grad_norm": 89.41574096679688,
      "learning_rate": 3.4271705309133273e-06,
      "loss": 6.5456,
      "step": 18242
    },
    {
      "epoch": 0.7364957610012112,
      "grad_norm": 66.91667938232422,
      "learning_rate": 3.426185137119046e-06,
      "loss": 5.3324,
      "step": 18243
    },
    {
      "epoch": 0.7365361324182479,
      "grad_norm": 79.3685531616211,
      "learning_rate": 3.425199855721645e-06,
      "loss": 4.039,
      "step": 18244
    },
    {
      "epoch": 0.7365765038352846,
      "grad_norm": 73.75249481201172,
      "learning_rate": 3.424214686737973e-06,
      "loss": 5.6512,
      "step": 18245
    },
    {
      "epoch": 0.7366168752523213,
      "grad_norm": 89.38762664794922,
      "learning_rate": 3.4232296301848745e-06,
      "loss": 5.6761,
      "step": 18246
    },
    {
      "epoch": 0.7366572466693581,
      "grad_norm": 89.98011016845703,
      "learning_rate": 3.4222446860791948e-06,
      "loss": 4.5685,
      "step": 18247
    },
    {
      "epoch": 0.7366976180863948,
      "grad_norm": 104.61258697509766,
      "learning_rate": 3.4212598544377674e-06,
      "loss": 4.0999,
      "step": 18248
    },
    {
      "epoch": 0.7367379895034316,
      "grad_norm": 67.08551788330078,
      "learning_rate": 3.4202751352774344e-06,
      "loss": 2.4372,
      "step": 18249
    },
    {
      "epoch": 0.7367783609204683,
      "grad_norm": 91.44666290283203,
      "learning_rate": 3.4192905286150314e-06,
      "loss": 5.8252,
      "step": 18250
    },
    {
      "epoch": 0.7368187323375051,
      "grad_norm": 103.33592987060547,
      "learning_rate": 3.4183060344673937e-06,
      "loss": 6.1914,
      "step": 18251
    },
    {
      "epoch": 0.7368591037545418,
      "grad_norm": 168.67918395996094,
      "learning_rate": 3.417321652851353e-06,
      "loss": 5.3646,
      "step": 18252
    },
    {
      "epoch": 0.7368994751715785,
      "grad_norm": 124.11573791503906,
      "learning_rate": 3.416337383783741e-06,
      "loss": 4.4907,
      "step": 18253
    },
    {
      "epoch": 0.7369398465886152,
      "grad_norm": 71.63899993896484,
      "learning_rate": 3.415353227281385e-06,
      "loss": 3.8444,
      "step": 18254
    },
    {
      "epoch": 0.7369802180056519,
      "grad_norm": 81.04208374023438,
      "learning_rate": 3.414369183361116e-06,
      "loss": 6.1456,
      "step": 18255
    },
    {
      "epoch": 0.7370205894226888,
      "grad_norm": 86.11397552490234,
      "learning_rate": 3.4133852520397526e-06,
      "loss": 5.7754,
      "step": 18256
    },
    {
      "epoch": 0.7370609608397255,
      "grad_norm": 88.67327880859375,
      "learning_rate": 3.4124014333341203e-06,
      "loss": 5.4427,
      "step": 18257
    },
    {
      "epoch": 0.7371013322567622,
      "grad_norm": 80.86819458007812,
      "learning_rate": 3.4114177272610404e-06,
      "loss": 5.9953,
      "step": 18258
    },
    {
      "epoch": 0.7371417036737989,
      "grad_norm": 144.0452880859375,
      "learning_rate": 3.410434133837335e-06,
      "loss": 5.4955,
      "step": 18259
    },
    {
      "epoch": 0.7371820750908357,
      "grad_norm": 76.87208557128906,
      "learning_rate": 3.4094506530798144e-06,
      "loss": 3.5309,
      "step": 18260
    },
    {
      "epoch": 0.7372224465078724,
      "grad_norm": 90.1191635131836,
      "learning_rate": 3.408467285005298e-06,
      "loss": 5.0439,
      "step": 18261
    },
    {
      "epoch": 0.7372628179249092,
      "grad_norm": 81.58587646484375,
      "learning_rate": 3.407484029630599e-06,
      "loss": 4.5649,
      "step": 18262
    },
    {
      "epoch": 0.7373031893419459,
      "grad_norm": 75.98954772949219,
      "learning_rate": 3.4065008869725323e-06,
      "loss": 4.7359,
      "step": 18263
    },
    {
      "epoch": 0.7373435607589827,
      "grad_norm": 118.26091003417969,
      "learning_rate": 3.4055178570478986e-06,
      "loss": 5.2224,
      "step": 18264
    },
    {
      "epoch": 0.7373839321760194,
      "grad_norm": 261.0732116699219,
      "learning_rate": 3.4045349398735116e-06,
      "loss": 5.3076,
      "step": 18265
    },
    {
      "epoch": 0.7374243035930561,
      "grad_norm": 92.8950424194336,
      "learning_rate": 3.4035521354661748e-06,
      "loss": 4.3895,
      "step": 18266
    },
    {
      "epoch": 0.7374646750100928,
      "grad_norm": 73.60409545898438,
      "learning_rate": 3.402569443842697e-06,
      "loss": 6.3804,
      "step": 18267
    },
    {
      "epoch": 0.7375050464271296,
      "grad_norm": 79.80377197265625,
      "learning_rate": 3.4015868650198713e-06,
      "loss": 4.7023,
      "step": 18268
    },
    {
      "epoch": 0.7375454178441664,
      "grad_norm": 177.1580047607422,
      "learning_rate": 3.4006043990145034e-06,
      "loss": 4.5295,
      "step": 18269
    },
    {
      "epoch": 0.7375857892612031,
      "grad_norm": 133.7570037841797,
      "learning_rate": 3.399622045843388e-06,
      "loss": 4.5995,
      "step": 18270
    },
    {
      "epoch": 0.7376261606782398,
      "grad_norm": 125.02853393554688,
      "learning_rate": 3.3986398055233226e-06,
      "loss": 3.228,
      "step": 18271
    },
    {
      "epoch": 0.7376665320952766,
      "grad_norm": 92.62535095214844,
      "learning_rate": 3.3976576780711013e-06,
      "loss": 6.7886,
      "step": 18272
    },
    {
      "epoch": 0.7377069035123133,
      "grad_norm": 112.79570007324219,
      "learning_rate": 3.396675663503517e-06,
      "loss": 4.0864,
      "step": 18273
    },
    {
      "epoch": 0.73774727492935,
      "grad_norm": 129.81617736816406,
      "learning_rate": 3.3956937618373586e-06,
      "loss": 6.1803,
      "step": 18274
    },
    {
      "epoch": 0.7377876463463867,
      "grad_norm": 84.56253814697266,
      "learning_rate": 3.3947119730894175e-06,
      "loss": 4.2672,
      "step": 18275
    },
    {
      "epoch": 0.7378280177634235,
      "grad_norm": 133.70994567871094,
      "learning_rate": 3.393730297276474e-06,
      "loss": 6.2042,
      "step": 18276
    },
    {
      "epoch": 0.7378683891804603,
      "grad_norm": 103.41273498535156,
      "learning_rate": 3.3927487344153165e-06,
      "loss": 5.0945,
      "step": 18277
    },
    {
      "epoch": 0.737908760597497,
      "grad_norm": 136.10202026367188,
      "learning_rate": 3.391767284522726e-06,
      "loss": 4.8696,
      "step": 18278
    },
    {
      "epoch": 0.7379491320145337,
      "grad_norm": 109.20802307128906,
      "learning_rate": 3.390785947615487e-06,
      "loss": 3.0145,
      "step": 18279
    },
    {
      "epoch": 0.7379895034315704,
      "grad_norm": 114.5710678100586,
      "learning_rate": 3.3898047237103714e-06,
      "loss": 4.0866,
      "step": 18280
    },
    {
      "epoch": 0.7380298748486072,
      "grad_norm": 137.01846313476562,
      "learning_rate": 3.388823612824159e-06,
      "loss": 4.6029,
      "step": 18281
    },
    {
      "epoch": 0.7380702462656439,
      "grad_norm": 75.84923553466797,
      "learning_rate": 3.3878426149736243e-06,
      "loss": 5.4761,
      "step": 18282
    },
    {
      "epoch": 0.7381106176826807,
      "grad_norm": 91.13080596923828,
      "learning_rate": 3.3868617301755446e-06,
      "loss": 3.7505,
      "step": 18283
    },
    {
      "epoch": 0.7381509890997174,
      "grad_norm": 215.3432159423828,
      "learning_rate": 3.385880958446682e-06,
      "loss": 7.3688,
      "step": 18284
    },
    {
      "epoch": 0.7381913605167542,
      "grad_norm": 85.92041015625,
      "learning_rate": 3.3849002998038117e-06,
      "loss": 5.4717,
      "step": 18285
    },
    {
      "epoch": 0.7382317319337909,
      "grad_norm": 80.64630126953125,
      "learning_rate": 3.383919754263697e-06,
      "loss": 3.6084,
      "step": 18286
    },
    {
      "epoch": 0.7382721033508276,
      "grad_norm": 82.87368774414062,
      "learning_rate": 3.3829393218431096e-06,
      "loss": 3.7122,
      "step": 18287
    },
    {
      "epoch": 0.7383124747678643,
      "grad_norm": 91.38121795654297,
      "learning_rate": 3.3819590025588034e-06,
      "loss": 6.6669,
      "step": 18288
    },
    {
      "epoch": 0.738352846184901,
      "grad_norm": 143.37643432617188,
      "learning_rate": 3.380978796427544e-06,
      "loss": 5.4929,
      "step": 18289
    },
    {
      "epoch": 0.7383932176019379,
      "grad_norm": 81.59027862548828,
      "learning_rate": 3.379998703466092e-06,
      "loss": 4.1637,
      "step": 18290
    },
    {
      "epoch": 0.7384335890189746,
      "grad_norm": 143.9515838623047,
      "learning_rate": 3.379018723691202e-06,
      "loss": 6.3972,
      "step": 18291
    },
    {
      "epoch": 0.7384739604360113,
      "grad_norm": 65.41053771972656,
      "learning_rate": 3.378038857119632e-06,
      "loss": 6.4932,
      "step": 18292
    },
    {
      "epoch": 0.738514331853048,
      "grad_norm": 109.29496765136719,
      "learning_rate": 3.3770591037681333e-06,
      "loss": 3.7292,
      "step": 18293
    },
    {
      "epoch": 0.7385547032700848,
      "grad_norm": 331.7079772949219,
      "learning_rate": 3.376079463653458e-06,
      "loss": 4.7246,
      "step": 18294
    },
    {
      "epoch": 0.7385950746871215,
      "grad_norm": 63.28571701049805,
      "learning_rate": 3.37509993679236e-06,
      "loss": 4.8453,
      "step": 18295
    },
    {
      "epoch": 0.7386354461041582,
      "grad_norm": 117.83737182617188,
      "learning_rate": 3.374120523201578e-06,
      "loss": 4.5869,
      "step": 18296
    },
    {
      "epoch": 0.738675817521195,
      "grad_norm": 92.8818588256836,
      "learning_rate": 3.3731412228978644e-06,
      "loss": 6.2501,
      "step": 18297
    },
    {
      "epoch": 0.7387161889382318,
      "grad_norm": 178.14633178710938,
      "learning_rate": 3.37216203589796e-06,
      "loss": 4.1659,
      "step": 18298
    },
    {
      "epoch": 0.7387565603552685,
      "grad_norm": 129.64431762695312,
      "learning_rate": 3.371182962218611e-06,
      "loss": 5.6631,
      "step": 18299
    },
    {
      "epoch": 0.7387969317723052,
      "grad_norm": 118.33184051513672,
      "learning_rate": 3.3702040018765516e-06,
      "loss": 4.0427,
      "step": 18300
    },
    {
      "epoch": 0.7388373031893419,
      "grad_norm": 120.4029769897461,
      "learning_rate": 3.3692251548885213e-06,
      "loss": 6.8277,
      "step": 18301
    },
    {
      "epoch": 0.7388776746063787,
      "grad_norm": 140.29075622558594,
      "learning_rate": 3.3682464212712563e-06,
      "loss": 6.036,
      "step": 18302
    },
    {
      "epoch": 0.7389180460234154,
      "grad_norm": 102.98318481445312,
      "learning_rate": 3.367267801041496e-06,
      "loss": 5.335,
      "step": 18303
    },
    {
      "epoch": 0.7389584174404522,
      "grad_norm": 145.4521942138672,
      "learning_rate": 3.366289294215963e-06,
      "loss": 4.3986,
      "step": 18304
    },
    {
      "epoch": 0.7389987888574889,
      "grad_norm": 121.66343688964844,
      "learning_rate": 3.3653109008113927e-06,
      "loss": 5.6644,
      "step": 18305
    },
    {
      "epoch": 0.7390391602745257,
      "grad_norm": 94.39958953857422,
      "learning_rate": 3.3643326208445137e-06,
      "loss": 4.5168,
      "step": 18306
    },
    {
      "epoch": 0.7390795316915624,
      "grad_norm": 70.31929779052734,
      "learning_rate": 3.363354454332054e-06,
      "loss": 6.7557,
      "step": 18307
    },
    {
      "epoch": 0.7391199031085991,
      "grad_norm": 114.47013854980469,
      "learning_rate": 3.3623764012907322e-06,
      "loss": 5.3096,
      "step": 18308
    },
    {
      "epoch": 0.7391602745256358,
      "grad_norm": 73.15333557128906,
      "learning_rate": 3.361398461737274e-06,
      "loss": 4.9042,
      "step": 18309
    },
    {
      "epoch": 0.7392006459426725,
      "grad_norm": 107.99948120117188,
      "learning_rate": 3.3604206356883994e-06,
      "loss": 3.6126,
      "step": 18310
    },
    {
      "epoch": 0.7392410173597094,
      "grad_norm": 143.50144958496094,
      "learning_rate": 3.359442923160828e-06,
      "loss": 6.2026,
      "step": 18311
    },
    {
      "epoch": 0.7392813887767461,
      "grad_norm": 131.56951904296875,
      "learning_rate": 3.358465324171276e-06,
      "loss": 6.0942,
      "step": 18312
    },
    {
      "epoch": 0.7393217601937828,
      "grad_norm": 84.4505386352539,
      "learning_rate": 3.3574878387364572e-06,
      "loss": 5.879,
      "step": 18313
    },
    {
      "epoch": 0.7393621316108195,
      "grad_norm": 123.69866180419922,
      "learning_rate": 3.3565104668730864e-06,
      "loss": 3.066,
      "step": 18314
    },
    {
      "epoch": 0.7394025030278563,
      "grad_norm": 79.21975708007812,
      "learning_rate": 3.355533208597874e-06,
      "loss": 4.9154,
      "step": 18315
    },
    {
      "epoch": 0.739442874444893,
      "grad_norm": 147.41201782226562,
      "learning_rate": 3.3545560639275265e-06,
      "loss": 4.4972,
      "step": 18316
    },
    {
      "epoch": 0.7394832458619297,
      "grad_norm": 105.62739562988281,
      "learning_rate": 3.353579032878751e-06,
      "loss": 3.4583,
      "step": 18317
    },
    {
      "epoch": 0.7395236172789665,
      "grad_norm": 89.81172180175781,
      "learning_rate": 3.3526021154682543e-06,
      "loss": 6.4398,
      "step": 18318
    },
    {
      "epoch": 0.7395639886960033,
      "grad_norm": 85.08378601074219,
      "learning_rate": 3.3516253117127408e-06,
      "loss": 2.9016,
      "step": 18319
    },
    {
      "epoch": 0.73960436011304,
      "grad_norm": 295.8628845214844,
      "learning_rate": 3.350648621628907e-06,
      "loss": 3.6199,
      "step": 18320
    },
    {
      "epoch": 0.7396447315300767,
      "grad_norm": 93.9594497680664,
      "learning_rate": 3.349672045233454e-06,
      "loss": 4.9357,
      "step": 18321
    },
    {
      "epoch": 0.7396851029471134,
      "grad_norm": 129.71730041503906,
      "learning_rate": 3.34869558254308e-06,
      "loss": 5.9088,
      "step": 18322
    },
    {
      "epoch": 0.7397254743641501,
      "grad_norm": 86.36590576171875,
      "learning_rate": 3.3477192335744822e-06,
      "loss": 4.4963,
      "step": 18323
    },
    {
      "epoch": 0.7397658457811869,
      "grad_norm": 88.30248260498047,
      "learning_rate": 3.3467429983443477e-06,
      "loss": 5.0143,
      "step": 18324
    },
    {
      "epoch": 0.7398062171982237,
      "grad_norm": 76.65409851074219,
      "learning_rate": 3.345766876869373e-06,
      "loss": 4.0548,
      "step": 18325
    },
    {
      "epoch": 0.7398465886152604,
      "grad_norm": 70.29314422607422,
      "learning_rate": 3.3447908691662446e-06,
      "loss": 4.3037,
      "step": 18326
    },
    {
      "epoch": 0.7398869600322971,
      "grad_norm": 139.23220825195312,
      "learning_rate": 3.3438149752516546e-06,
      "loss": 4.6435,
      "step": 18327
    },
    {
      "epoch": 0.7399273314493339,
      "grad_norm": 101.29421997070312,
      "learning_rate": 3.3428391951422825e-06,
      "loss": 4.0535,
      "step": 18328
    },
    {
      "epoch": 0.7399677028663706,
      "grad_norm": 74.45484161376953,
      "learning_rate": 3.341863528854815e-06,
      "loss": 5.8413,
      "step": 18329
    },
    {
      "epoch": 0.7400080742834073,
      "grad_norm": 116.86412811279297,
      "learning_rate": 3.340887976405932e-06,
      "loss": 4.5125,
      "step": 18330
    },
    {
      "epoch": 0.740048445700444,
      "grad_norm": 81.3237075805664,
      "learning_rate": 3.3399125378123152e-06,
      "loss": 5.8729,
      "step": 18331
    },
    {
      "epoch": 0.7400888171174809,
      "grad_norm": 72.18485260009766,
      "learning_rate": 3.3389372130906415e-06,
      "loss": 5.4628,
      "step": 18332
    },
    {
      "epoch": 0.7401291885345176,
      "grad_norm": 105.70015716552734,
      "learning_rate": 3.3379620022575866e-06,
      "loss": 2.6041,
      "step": 18333
    },
    {
      "epoch": 0.7401695599515543,
      "grad_norm": 122.42825317382812,
      "learning_rate": 3.3369869053298255e-06,
      "loss": 3.6357,
      "step": 18334
    },
    {
      "epoch": 0.740209931368591,
      "grad_norm": 147.5387725830078,
      "learning_rate": 3.3360119223240305e-06,
      "loss": 5.0153,
      "step": 18335
    },
    {
      "epoch": 0.7402503027856278,
      "grad_norm": 113.25949096679688,
      "learning_rate": 3.335037053256869e-06,
      "loss": 6.1588,
      "step": 18336
    },
    {
      "epoch": 0.7402906742026645,
      "grad_norm": 74.92230987548828,
      "learning_rate": 3.334062298145009e-06,
      "loss": 5.1586,
      "step": 18337
    },
    {
      "epoch": 0.7403310456197012,
      "grad_norm": 110.46375274658203,
      "learning_rate": 3.3330876570051173e-06,
      "loss": 5.2242,
      "step": 18338
    },
    {
      "epoch": 0.740371417036738,
      "grad_norm": 136.9752197265625,
      "learning_rate": 3.3321131298538612e-06,
      "loss": 4.7234,
      "step": 18339
    },
    {
      "epoch": 0.7404117884537748,
      "grad_norm": 156.9711456298828,
      "learning_rate": 3.331138716707898e-06,
      "loss": 6.0399,
      "step": 18340
    },
    {
      "epoch": 0.7404521598708115,
      "grad_norm": 80.1180648803711,
      "learning_rate": 3.330164417583889e-06,
      "loss": 4.9783,
      "step": 18341
    },
    {
      "epoch": 0.7404925312878482,
      "grad_norm": 83.58401489257812,
      "learning_rate": 3.3291902324984924e-06,
      "loss": 4.0626,
      "step": 18342
    },
    {
      "epoch": 0.7405329027048849,
      "grad_norm": 87.76453399658203,
      "learning_rate": 3.3282161614683705e-06,
      "loss": 5.1425,
      "step": 18343
    },
    {
      "epoch": 0.7405732741219216,
      "grad_norm": 72.61039733886719,
      "learning_rate": 3.3272422045101682e-06,
      "loss": 5.4022,
      "step": 18344
    },
    {
      "epoch": 0.7406136455389584,
      "grad_norm": 101.35118103027344,
      "learning_rate": 3.326268361640542e-06,
      "loss": 5.3759,
      "step": 18345
    },
    {
      "epoch": 0.7406540169559952,
      "grad_norm": 122.82416534423828,
      "learning_rate": 3.3252946328761427e-06,
      "loss": 5.0285,
      "step": 18346
    },
    {
      "epoch": 0.7406943883730319,
      "grad_norm": 87.4691390991211,
      "learning_rate": 3.3243210182336226e-06,
      "loss": 5.95,
      "step": 18347
    },
    {
      "epoch": 0.7407347597900686,
      "grad_norm": 82.73084259033203,
      "learning_rate": 3.3233475177296205e-06,
      "loss": 5.969,
      "step": 18348
    },
    {
      "epoch": 0.7407751312071054,
      "grad_norm": 99.92607879638672,
      "learning_rate": 3.3223741313807845e-06,
      "loss": 3.7497,
      "step": 18349
    },
    {
      "epoch": 0.7408155026241421,
      "grad_norm": 168.57188415527344,
      "learning_rate": 3.321400859203757e-06,
      "loss": 5.235,
      "step": 18350
    },
    {
      "epoch": 0.7408558740411788,
      "grad_norm": 118.61162567138672,
      "learning_rate": 3.3204277012151807e-06,
      "loss": 4.2747,
      "step": 18351
    },
    {
      "epoch": 0.7408962454582155,
      "grad_norm": 85.14500427246094,
      "learning_rate": 3.3194546574316923e-06,
      "loss": 4.1849,
      "step": 18352
    },
    {
      "epoch": 0.7409366168752524,
      "grad_norm": 78.42446899414062,
      "learning_rate": 3.318481727869928e-06,
      "loss": 3.416,
      "step": 18353
    },
    {
      "epoch": 0.7409769882922891,
      "grad_norm": 83.52201080322266,
      "learning_rate": 3.3175089125465244e-06,
      "loss": 5.0054,
      "step": 18354
    },
    {
      "epoch": 0.7410173597093258,
      "grad_norm": 92.84588623046875,
      "learning_rate": 3.3165362114781176e-06,
      "loss": 4.199,
      "step": 18355
    },
    {
      "epoch": 0.7410577311263625,
      "grad_norm": 91.3167953491211,
      "learning_rate": 3.3155636246813294e-06,
      "loss": 6.287,
      "step": 18356
    },
    {
      "epoch": 0.7410981025433993,
      "grad_norm": 110.95919036865234,
      "learning_rate": 3.3145911521727957e-06,
      "loss": 4.7126,
      "step": 18357
    },
    {
      "epoch": 0.741138473960436,
      "grad_norm": 100.33045196533203,
      "learning_rate": 3.313618793969141e-06,
      "loss": 6.5495,
      "step": 18358
    },
    {
      "epoch": 0.7411788453774727,
      "grad_norm": 154.947265625,
      "learning_rate": 3.312646550086994e-06,
      "loss": 6.5576,
      "step": 18359
    },
    {
      "epoch": 0.7412192167945095,
      "grad_norm": 111.34819030761719,
      "learning_rate": 3.311674420542973e-06,
      "loss": 4.3123,
      "step": 18360
    },
    {
      "epoch": 0.7412595882115462,
      "grad_norm": 91.18799591064453,
      "learning_rate": 3.3107024053536994e-06,
      "loss": 4.2107,
      "step": 18361
    },
    {
      "epoch": 0.741299959628583,
      "grad_norm": 97.05606079101562,
      "learning_rate": 3.3097305045357953e-06,
      "loss": 3.8254,
      "step": 18362
    },
    {
      "epoch": 0.7413403310456197,
      "grad_norm": 90.55087280273438,
      "learning_rate": 3.30875871810588e-06,
      "loss": 5.5585,
      "step": 18363
    },
    {
      "epoch": 0.7413807024626564,
      "grad_norm": 107.149658203125,
      "learning_rate": 3.307787046080562e-06,
      "loss": 4.0937,
      "step": 18364
    },
    {
      "epoch": 0.7414210738796931,
      "grad_norm": 90.74315643310547,
      "learning_rate": 3.306815488476457e-06,
      "loss": 6.167,
      "step": 18365
    },
    {
      "epoch": 0.74146144529673,
      "grad_norm": 82.22135925292969,
      "learning_rate": 3.305844045310178e-06,
      "loss": 5.2846,
      "step": 18366
    },
    {
      "epoch": 0.7415018167137667,
      "grad_norm": 117.9947509765625,
      "learning_rate": 3.3048727165983373e-06,
      "loss": 6.1097,
      "step": 18367
    },
    {
      "epoch": 0.7415421881308034,
      "grad_norm": 108.54613494873047,
      "learning_rate": 3.303901502357536e-06,
      "loss": 4.7449,
      "step": 18368
    },
    {
      "epoch": 0.7415825595478401,
      "grad_norm": 104.45184326171875,
      "learning_rate": 3.302930402604382e-06,
      "loss": 5.7495,
      "step": 18369
    },
    {
      "epoch": 0.7416229309648769,
      "grad_norm": 104.95489501953125,
      "learning_rate": 3.3019594173554794e-06,
      "loss": 4.3314,
      "step": 18370
    },
    {
      "epoch": 0.7416633023819136,
      "grad_norm": 219.8870391845703,
      "learning_rate": 3.3009885466274303e-06,
      "loss": 3.3273,
      "step": 18371
    },
    {
      "epoch": 0.7417036737989503,
      "grad_norm": 79.65248107910156,
      "learning_rate": 3.3000177904368324e-06,
      "loss": 6.0327,
      "step": 18372
    },
    {
      "epoch": 0.7417440452159871,
      "grad_norm": 133.51353454589844,
      "learning_rate": 3.299047148800286e-06,
      "loss": 4.4716,
      "step": 18373
    },
    {
      "epoch": 0.7417844166330239,
      "grad_norm": 118.2264175415039,
      "learning_rate": 3.2980766217343852e-06,
      "loss": 3.7058,
      "step": 18374
    },
    {
      "epoch": 0.7418247880500606,
      "grad_norm": 140.8677978515625,
      "learning_rate": 3.297106209255727e-06,
      "loss": 4.0243,
      "step": 18375
    },
    {
      "epoch": 0.7418651594670973,
      "grad_norm": 268.70599365234375,
      "learning_rate": 3.2961359113808968e-06,
      "loss": 5.2724,
      "step": 18376
    },
    {
      "epoch": 0.741905530884134,
      "grad_norm": 87.98077392578125,
      "learning_rate": 3.2951657281264883e-06,
      "loss": 7.073,
      "step": 18377
    },
    {
      "epoch": 0.7419459023011707,
      "grad_norm": 87.17716217041016,
      "learning_rate": 3.2941956595090886e-06,
      "loss": 5.2443,
      "step": 18378
    },
    {
      "epoch": 0.7419862737182075,
      "grad_norm": 93.24468994140625,
      "learning_rate": 3.2932257055452877e-06,
      "loss": 5.7416,
      "step": 18379
    },
    {
      "epoch": 0.7420266451352443,
      "grad_norm": 120.37889099121094,
      "learning_rate": 3.2922558662516624e-06,
      "loss": 3.7867,
      "step": 18380
    },
    {
      "epoch": 0.742067016552281,
      "grad_norm": 84.54313659667969,
      "learning_rate": 3.2912861416447985e-06,
      "loss": 4.3152,
      "step": 18381
    },
    {
      "epoch": 0.7421073879693177,
      "grad_norm": 182.23182678222656,
      "learning_rate": 3.2903165317412745e-06,
      "loss": 4.1452,
      "step": 18382
    },
    {
      "epoch": 0.7421477593863545,
      "grad_norm": 74.82049560546875,
      "learning_rate": 3.289347036557673e-06,
      "loss": 6.1488,
      "step": 18383
    },
    {
      "epoch": 0.7421881308033912,
      "grad_norm": 77.8791275024414,
      "learning_rate": 3.288377656110564e-06,
      "loss": 5.1384,
      "step": 18384
    },
    {
      "epoch": 0.7422285022204279,
      "grad_norm": 108.10321044921875,
      "learning_rate": 3.2874083904165244e-06,
      "loss": 4.7613,
      "step": 18385
    },
    {
      "epoch": 0.7422688736374646,
      "grad_norm": 98.42650604248047,
      "learning_rate": 3.2864392394921263e-06,
      "loss": 5.0862,
      "step": 18386
    },
    {
      "epoch": 0.7423092450545015,
      "grad_norm": 79.9175033569336,
      "learning_rate": 3.285470203353942e-06,
      "loss": 3.1662,
      "step": 18387
    },
    {
      "epoch": 0.7423496164715382,
      "grad_norm": 103.33541870117188,
      "learning_rate": 3.284501282018535e-06,
      "loss": 4.4755,
      "step": 18388
    },
    {
      "epoch": 0.7423899878885749,
      "grad_norm": 80.88642120361328,
      "learning_rate": 3.2835324755024753e-06,
      "loss": 4.2483,
      "step": 18389
    },
    {
      "epoch": 0.7424303593056116,
      "grad_norm": 112.63166809082031,
      "learning_rate": 3.2825637838223256e-06,
      "loss": 4.0281,
      "step": 18390
    },
    {
      "epoch": 0.7424707307226484,
      "grad_norm": 84.5052261352539,
      "learning_rate": 3.281595206994649e-06,
      "loss": 5.0931,
      "step": 18391
    },
    {
      "epoch": 0.7425111021396851,
      "grad_norm": 84.96820068359375,
      "learning_rate": 3.280626745036005e-06,
      "loss": 5.0506,
      "step": 18392
    },
    {
      "epoch": 0.7425514735567218,
      "grad_norm": 1136.20361328125,
      "learning_rate": 3.2796583979629527e-06,
      "loss": 4.5319,
      "step": 18393
    },
    {
      "epoch": 0.7425918449737586,
      "grad_norm": 86.6367416381836,
      "learning_rate": 3.27869016579205e-06,
      "loss": 6.366,
      "step": 18394
    },
    {
      "epoch": 0.7426322163907954,
      "grad_norm": 91.18032836914062,
      "learning_rate": 3.277722048539852e-06,
      "loss": 3.4505,
      "step": 18395
    },
    {
      "epoch": 0.7426725878078321,
      "grad_norm": 111.06874084472656,
      "learning_rate": 3.2767540462229065e-06,
      "loss": 5.3383,
      "step": 18396
    },
    {
      "epoch": 0.7427129592248688,
      "grad_norm": 75.24071502685547,
      "learning_rate": 3.2757861588577666e-06,
      "loss": 4.2971,
      "step": 18397
    },
    {
      "epoch": 0.7427533306419055,
      "grad_norm": 70.25016021728516,
      "learning_rate": 3.2748183864609815e-06,
      "loss": 4.4639,
      "step": 18398
    },
    {
      "epoch": 0.7427937020589422,
      "grad_norm": 111.15967559814453,
      "learning_rate": 3.2738507290491017e-06,
      "loss": 4.9657,
      "step": 18399
    },
    {
      "epoch": 0.742834073475979,
      "grad_norm": 86.07328033447266,
      "learning_rate": 3.2728831866386635e-06,
      "loss": 4.9957,
      "step": 18400
    },
    {
      "epoch": 0.7428744448930158,
      "grad_norm": 88.5811996459961,
      "learning_rate": 3.2719157592462147e-06,
      "loss": 5.5913,
      "step": 18401
    },
    {
      "epoch": 0.7429148163100525,
      "grad_norm": 78.37162017822266,
      "learning_rate": 3.2709484468882945e-06,
      "loss": 5.5471,
      "step": 18402
    },
    {
      "epoch": 0.7429551877270892,
      "grad_norm": 103.35121154785156,
      "learning_rate": 3.2699812495814454e-06,
      "loss": 3.6959,
      "step": 18403
    },
    {
      "epoch": 0.742995559144126,
      "grad_norm": 112.26286315917969,
      "learning_rate": 3.269014167342198e-06,
      "loss": 4.78,
      "step": 18404
    },
    {
      "epoch": 0.7430359305611627,
      "grad_norm": 117.42578887939453,
      "learning_rate": 3.268047200187091e-06,
      "loss": 4.8966,
      "step": 18405
    },
    {
      "epoch": 0.7430763019781994,
      "grad_norm": 76.92755889892578,
      "learning_rate": 3.2670803481326563e-06,
      "loss": 4.0467,
      "step": 18406
    },
    {
      "epoch": 0.7431166733952361,
      "grad_norm": 164.06947326660156,
      "learning_rate": 3.266113611195428e-06,
      "loss": 4.116,
      "step": 18407
    },
    {
      "epoch": 0.743157044812273,
      "grad_norm": 121.58839416503906,
      "learning_rate": 3.2651469893919297e-06,
      "loss": 4.6255,
      "step": 18408
    },
    {
      "epoch": 0.7431974162293097,
      "grad_norm": 120.59607696533203,
      "learning_rate": 3.2641804827386904e-06,
      "loss": 3.2336,
      "step": 18409
    },
    {
      "epoch": 0.7432377876463464,
      "grad_norm": 82.34420013427734,
      "learning_rate": 3.2632140912522357e-06,
      "loss": 4.0952,
      "step": 18410
    },
    {
      "epoch": 0.7432781590633831,
      "grad_norm": 81.4994888305664,
      "learning_rate": 3.262247814949089e-06,
      "loss": 3.5227,
      "step": 18411
    },
    {
      "epoch": 0.7433185304804198,
      "grad_norm": 125.95669555664062,
      "learning_rate": 3.2612816538457703e-06,
      "loss": 7.2751,
      "step": 18412
    },
    {
      "epoch": 0.7433589018974566,
      "grad_norm": 94.08305358886719,
      "learning_rate": 3.2603156079587993e-06,
      "loss": 5.1299,
      "step": 18413
    },
    {
      "epoch": 0.7433992733144933,
      "grad_norm": 85.70441436767578,
      "learning_rate": 3.259349677304693e-06,
      "loss": 4.1488,
      "step": 18414
    },
    {
      "epoch": 0.7434396447315301,
      "grad_norm": 88.25456237792969,
      "learning_rate": 3.2583838618999698e-06,
      "loss": 4.3206,
      "step": 18415
    },
    {
      "epoch": 0.7434800161485668,
      "grad_norm": 112.62284088134766,
      "learning_rate": 3.2574181617611367e-06,
      "loss": 4.3577,
      "step": 18416
    },
    {
      "epoch": 0.7435203875656036,
      "grad_norm": 90.59919738769531,
      "learning_rate": 3.2564525769047083e-06,
      "loss": 4.7592,
      "step": 18417
    },
    {
      "epoch": 0.7435607589826403,
      "grad_norm": 86.36058044433594,
      "learning_rate": 3.255487107347193e-06,
      "loss": 4.687,
      "step": 18418
    },
    {
      "epoch": 0.743601130399677,
      "grad_norm": 95.8678207397461,
      "learning_rate": 3.254521753105102e-06,
      "loss": 4.0956,
      "step": 18419
    },
    {
      "epoch": 0.7436415018167137,
      "grad_norm": 93.8951187133789,
      "learning_rate": 3.2535565141949346e-06,
      "loss": 4.5447,
      "step": 18420
    },
    {
      "epoch": 0.7436818732337505,
      "grad_norm": 91.38758087158203,
      "learning_rate": 3.2525913906331964e-06,
      "loss": 5.878,
      "step": 18421
    },
    {
      "epoch": 0.7437222446507873,
      "grad_norm": 125.90348815917969,
      "learning_rate": 3.251626382436388e-06,
      "loss": 6.1091,
      "step": 18422
    },
    {
      "epoch": 0.743762616067824,
      "grad_norm": 106.23027038574219,
      "learning_rate": 3.250661489621014e-06,
      "loss": 3.4661,
      "step": 18423
    },
    {
      "epoch": 0.7438029874848607,
      "grad_norm": 96.60459899902344,
      "learning_rate": 3.249696712203565e-06,
      "loss": 5.6778,
      "step": 18424
    },
    {
      "epoch": 0.7438433589018975,
      "grad_norm": 113.23308563232422,
      "learning_rate": 3.2487320502005384e-06,
      "loss": 3.9936,
      "step": 18425
    },
    {
      "epoch": 0.7438837303189342,
      "grad_norm": 119.20074462890625,
      "learning_rate": 3.2477675036284286e-06,
      "loss": 3.6635,
      "step": 18426
    },
    {
      "epoch": 0.7439241017359709,
      "grad_norm": 124.46678161621094,
      "learning_rate": 3.24680307250373e-06,
      "loss": 5.52,
      "step": 18427
    },
    {
      "epoch": 0.7439644731530076,
      "grad_norm": 99.51240539550781,
      "learning_rate": 3.245838756842926e-06,
      "loss": 5.3732,
      "step": 18428
    },
    {
      "epoch": 0.7440048445700445,
      "grad_norm": 80.77886199951172,
      "learning_rate": 3.2448745566625073e-06,
      "loss": 3.6032,
      "step": 18429
    },
    {
      "epoch": 0.7440452159870812,
      "grad_norm": 88.96788024902344,
      "learning_rate": 3.2439104719789595e-06,
      "loss": 5.1491,
      "step": 18430
    },
    {
      "epoch": 0.7440855874041179,
      "grad_norm": 97.87133026123047,
      "learning_rate": 3.2429465028087647e-06,
      "loss": 4.7941,
      "step": 18431
    },
    {
      "epoch": 0.7441259588211546,
      "grad_norm": 78.16514587402344,
      "learning_rate": 3.241982649168407e-06,
      "loss": 6.2681,
      "step": 18432
    },
    {
      "epoch": 0.7441663302381913,
      "grad_norm": 107.23798370361328,
      "learning_rate": 3.241018911074365e-06,
      "loss": 4.9892,
      "step": 18433
    },
    {
      "epoch": 0.7442067016552281,
      "grad_norm": 97.95706176757812,
      "learning_rate": 3.240055288543116e-06,
      "loss": 5.0845,
      "step": 18434
    },
    {
      "epoch": 0.7442470730722648,
      "grad_norm": 95.7224349975586,
      "learning_rate": 3.239091781591138e-06,
      "loss": 6.2529,
      "step": 18435
    },
    {
      "epoch": 0.7442874444893016,
      "grad_norm": 68.94152069091797,
      "learning_rate": 3.2381283902349015e-06,
      "loss": 4.6361,
      "step": 18436
    },
    {
      "epoch": 0.7443278159063383,
      "grad_norm": 130.05018615722656,
      "learning_rate": 3.2371651144908776e-06,
      "loss": 4.0712,
      "step": 18437
    },
    {
      "epoch": 0.7443681873233751,
      "grad_norm": 132.97984313964844,
      "learning_rate": 3.23620195437554e-06,
      "loss": 4.3262,
      "step": 18438
    },
    {
      "epoch": 0.7444085587404118,
      "grad_norm": 88.1798095703125,
      "learning_rate": 3.2352389099053562e-06,
      "loss": 4.5515,
      "step": 18439
    },
    {
      "epoch": 0.7444489301574485,
      "grad_norm": 144.64308166503906,
      "learning_rate": 3.234275981096787e-06,
      "loss": 6.0915,
      "step": 18440
    },
    {
      "epoch": 0.7444893015744852,
      "grad_norm": 91.09664154052734,
      "learning_rate": 3.2333131679662986e-06,
      "loss": 5.9856,
      "step": 18441
    },
    {
      "epoch": 0.744529672991522,
      "grad_norm": 66.38771057128906,
      "learning_rate": 3.2323504705303556e-06,
      "loss": 3.3272,
      "step": 18442
    },
    {
      "epoch": 0.7445700444085588,
      "grad_norm": 93.14642333984375,
      "learning_rate": 3.2313878888054184e-06,
      "loss": 3.8858,
      "step": 18443
    },
    {
      "epoch": 0.7446104158255955,
      "grad_norm": 86.09276580810547,
      "learning_rate": 3.230425422807939e-06,
      "loss": 5.6668,
      "step": 18444
    },
    {
      "epoch": 0.7446507872426322,
      "grad_norm": 65.00184631347656,
      "learning_rate": 3.2294630725543763e-06,
      "loss": 5.1435,
      "step": 18445
    },
    {
      "epoch": 0.744691158659669,
      "grad_norm": 87.02645111083984,
      "learning_rate": 3.228500838061185e-06,
      "loss": 5.597,
      "step": 18446
    },
    {
      "epoch": 0.7447315300767057,
      "grad_norm": 97.34603118896484,
      "learning_rate": 3.227538719344817e-06,
      "loss": 4.9579,
      "step": 18447
    },
    {
      "epoch": 0.7447719014937424,
      "grad_norm": 101.33834838867188,
      "learning_rate": 3.2265767164217254e-06,
      "loss": 5.315,
      "step": 18448
    },
    {
      "epoch": 0.7448122729107791,
      "grad_norm": 89.0091552734375,
      "learning_rate": 3.225614829308352e-06,
      "loss": 4.09,
      "step": 18449
    },
    {
      "epoch": 0.744852644327816,
      "grad_norm": 68.80635833740234,
      "learning_rate": 3.2246530580211453e-06,
      "loss": 6.0556,
      "step": 18450
    },
    {
      "epoch": 0.7448930157448527,
      "grad_norm": 150.546630859375,
      "learning_rate": 3.2236914025765496e-06,
      "loss": 5.1717,
      "step": 18451
    },
    {
      "epoch": 0.7449333871618894,
      "grad_norm": 83.39421844482422,
      "learning_rate": 3.2227298629910063e-06,
      "loss": 5.5289,
      "step": 18452
    },
    {
      "epoch": 0.7449737585789261,
      "grad_norm": 372.8992919921875,
      "learning_rate": 3.2217684392809576e-06,
      "loss": 4.7699,
      "step": 18453
    },
    {
      "epoch": 0.7450141299959628,
      "grad_norm": 89.0141830444336,
      "learning_rate": 3.22080713146284e-06,
      "loss": 4.3576,
      "step": 18454
    },
    {
      "epoch": 0.7450545014129996,
      "grad_norm": 200.76390075683594,
      "learning_rate": 3.2198459395530934e-06,
      "loss": 4.6931,
      "step": 18455
    },
    {
      "epoch": 0.7450948728300363,
      "grad_norm": 89.01493835449219,
      "learning_rate": 3.218884863568145e-06,
      "loss": 5.1392,
      "step": 18456
    },
    {
      "epoch": 0.7451352442470731,
      "grad_norm": 94.92123413085938,
      "learning_rate": 3.217923903524429e-06,
      "loss": 4.1155,
      "step": 18457
    },
    {
      "epoch": 0.7451756156641098,
      "grad_norm": 179.03756713867188,
      "learning_rate": 3.2169630594383783e-06,
      "loss": 3.5824,
      "step": 18458
    },
    {
      "epoch": 0.7452159870811466,
      "grad_norm": 113.6865234375,
      "learning_rate": 3.216002331326423e-06,
      "loss": 6.1269,
      "step": 18459
    },
    {
      "epoch": 0.7452563584981833,
      "grad_norm": 145.9973907470703,
      "learning_rate": 3.215041719204983e-06,
      "loss": 5.9266,
      "step": 18460
    },
    {
      "epoch": 0.74529672991522,
      "grad_norm": 72.27515411376953,
      "learning_rate": 3.214081223090485e-06,
      "loss": 4.3461,
      "step": 18461
    },
    {
      "epoch": 0.7453371013322567,
      "grad_norm": 110.46398162841797,
      "learning_rate": 3.213120842999352e-06,
      "loss": 5.394,
      "step": 18462
    },
    {
      "epoch": 0.7453774727492934,
      "grad_norm": 107.69842529296875,
      "learning_rate": 3.212160578948007e-06,
      "loss": 5.8001,
      "step": 18463
    },
    {
      "epoch": 0.7454178441663303,
      "grad_norm": 107.31808471679688,
      "learning_rate": 3.2112004309528633e-06,
      "loss": 6.1741,
      "step": 18464
    },
    {
      "epoch": 0.745458215583367,
      "grad_norm": 117.63056182861328,
      "learning_rate": 3.2102403990303376e-06,
      "loss": 4.5893,
      "step": 18465
    },
    {
      "epoch": 0.7454985870004037,
      "grad_norm": 91.3443374633789,
      "learning_rate": 3.209280483196847e-06,
      "loss": 6.3857,
      "step": 18466
    },
    {
      "epoch": 0.7455389584174404,
      "grad_norm": 93.74369812011719,
      "learning_rate": 3.208320683468802e-06,
      "loss": 4.3975,
      "step": 18467
    },
    {
      "epoch": 0.7455793298344772,
      "grad_norm": 84.40602111816406,
      "learning_rate": 3.207360999862614e-06,
      "loss": 5.5198,
      "step": 18468
    },
    {
      "epoch": 0.7456197012515139,
      "grad_norm": 91.80689239501953,
      "learning_rate": 3.206401432394691e-06,
      "loss": 5.7962,
      "step": 18469
    },
    {
      "epoch": 0.7456600726685506,
      "grad_norm": 114.7034912109375,
      "learning_rate": 3.2054419810814397e-06,
      "loss": 3.7668,
      "step": 18470
    },
    {
      "epoch": 0.7457004440855874,
      "grad_norm": 100.10247802734375,
      "learning_rate": 3.2044826459392664e-06,
      "loss": 5.6876,
      "step": 18471
    },
    {
      "epoch": 0.7457408155026242,
      "grad_norm": 102.74681854248047,
      "learning_rate": 3.203523426984568e-06,
      "loss": 5.2743,
      "step": 18472
    },
    {
      "epoch": 0.7457811869196609,
      "grad_norm": 100.71549224853516,
      "learning_rate": 3.202564324233749e-06,
      "loss": 3.8834,
      "step": 18473
    },
    {
      "epoch": 0.7458215583366976,
      "grad_norm": 100.4878158569336,
      "learning_rate": 3.2016053377032065e-06,
      "loss": 6.7843,
      "step": 18474
    },
    {
      "epoch": 0.7458619297537343,
      "grad_norm": 130.4586944580078,
      "learning_rate": 3.2006464674093397e-06,
      "loss": 5.2844,
      "step": 18475
    },
    {
      "epoch": 0.745902301170771,
      "grad_norm": 73.43623352050781,
      "learning_rate": 3.199687713368539e-06,
      "loss": 3.9769,
      "step": 18476
    },
    {
      "epoch": 0.7459426725878079,
      "grad_norm": 108.78655242919922,
      "learning_rate": 3.1987290755971977e-06,
      "loss": 4.4987,
      "step": 18477
    },
    {
      "epoch": 0.7459830440048446,
      "grad_norm": 130.16595458984375,
      "learning_rate": 3.1977705541117065e-06,
      "loss": 5.2339,
      "step": 18478
    },
    {
      "epoch": 0.7460234154218813,
      "grad_norm": 122.79186248779297,
      "learning_rate": 3.1968121489284587e-06,
      "loss": 4.7371,
      "step": 18479
    },
    {
      "epoch": 0.746063786838918,
      "grad_norm": 73.66914367675781,
      "learning_rate": 3.195853860063832e-06,
      "loss": 5.4981,
      "step": 18480
    },
    {
      "epoch": 0.7461041582559548,
      "grad_norm": 89.49967956542969,
      "learning_rate": 3.1948956875342176e-06,
      "loss": 5.2474,
      "step": 18481
    },
    {
      "epoch": 0.7461445296729915,
      "grad_norm": 77.92986297607422,
      "learning_rate": 3.1939376313559942e-06,
      "loss": 6.3277,
      "step": 18482
    },
    {
      "epoch": 0.7461849010900282,
      "grad_norm": 98.67324829101562,
      "learning_rate": 3.192979691545548e-06,
      "loss": 5.8035,
      "step": 18483
    },
    {
      "epoch": 0.746225272507065,
      "grad_norm": 91.30819702148438,
      "learning_rate": 3.1920218681192493e-06,
      "loss": 5.4586,
      "step": 18484
    },
    {
      "epoch": 0.7462656439241018,
      "grad_norm": 77.54778289794922,
      "learning_rate": 3.19106416109348e-06,
      "loss": 5.0425,
      "step": 18485
    },
    {
      "epoch": 0.7463060153411385,
      "grad_norm": 82.51608276367188,
      "learning_rate": 3.1901065704846135e-06,
      "loss": 5.0599,
      "step": 18486
    },
    {
      "epoch": 0.7463463867581752,
      "grad_norm": 73.21670532226562,
      "learning_rate": 3.189149096309022e-06,
      "loss": 5.2313,
      "step": 18487
    },
    {
      "epoch": 0.7463867581752119,
      "grad_norm": 131.0918426513672,
      "learning_rate": 3.1881917385830764e-06,
      "loss": 5.4338,
      "step": 18488
    },
    {
      "epoch": 0.7464271295922487,
      "grad_norm": 113.83526611328125,
      "learning_rate": 3.187234497323145e-06,
      "loss": 4.1071,
      "step": 18489
    },
    {
      "epoch": 0.7464675010092854,
      "grad_norm": 100.47492980957031,
      "learning_rate": 3.1862773725455955e-06,
      "loss": 5.5972,
      "step": 18490
    },
    {
      "epoch": 0.7465078724263222,
      "grad_norm": 218.390380859375,
      "learning_rate": 3.185320364266795e-06,
      "loss": 4.3111,
      "step": 18491
    },
    {
      "epoch": 0.7465482438433589,
      "grad_norm": 91.61853790283203,
      "learning_rate": 3.184363472503099e-06,
      "loss": 4.7743,
      "step": 18492
    },
    {
      "epoch": 0.7465886152603957,
      "grad_norm": 91.48413848876953,
      "learning_rate": 3.183406697270872e-06,
      "loss": 6.3684,
      "step": 18493
    },
    {
      "epoch": 0.7466289866774324,
      "grad_norm": 88.19535827636719,
      "learning_rate": 3.182450038586473e-06,
      "loss": 4.9099,
      "step": 18494
    },
    {
      "epoch": 0.7466693580944691,
      "grad_norm": 115.2630615234375,
      "learning_rate": 3.1814934964662612e-06,
      "loss": 5.1884,
      "step": 18495
    },
    {
      "epoch": 0.7467097295115058,
      "grad_norm": 91.98076629638672,
      "learning_rate": 3.1805370709265857e-06,
      "loss": 6.3177,
      "step": 18496
    },
    {
      "epoch": 0.7467501009285425,
      "grad_norm": 116.70350646972656,
      "learning_rate": 3.1795807619838017e-06,
      "loss": 4.681,
      "step": 18497
    },
    {
      "epoch": 0.7467904723455794,
      "grad_norm": 90.98284912109375,
      "learning_rate": 3.1786245696542584e-06,
      "loss": 4.6015,
      "step": 18498
    },
    {
      "epoch": 0.7468308437626161,
      "grad_norm": 80.52275848388672,
      "learning_rate": 3.17766849395431e-06,
      "loss": 5.8275,
      "step": 18499
    },
    {
      "epoch": 0.7468712151796528,
      "grad_norm": 113.87226104736328,
      "learning_rate": 3.176712534900296e-06,
      "loss": 5.6598,
      "step": 18500
    },
    {
      "epoch": 0.7469115865966895,
      "grad_norm": 83.81155395507812,
      "learning_rate": 3.175756692508564e-06,
      "loss": 4.8327,
      "step": 18501
    },
    {
      "epoch": 0.7469519580137263,
      "grad_norm": 76.58091735839844,
      "learning_rate": 3.174800966795457e-06,
      "loss": 5.6971,
      "step": 18502
    },
    {
      "epoch": 0.746992329430763,
      "grad_norm": 58.48830795288086,
      "learning_rate": 3.173845357777319e-06,
      "loss": 5.1565,
      "step": 18503
    },
    {
      "epoch": 0.7470327008477997,
      "grad_norm": 96.15792083740234,
      "learning_rate": 3.1728898654704808e-06,
      "loss": 4.105,
      "step": 18504
    },
    {
      "epoch": 0.7470730722648365,
      "grad_norm": 111.22298431396484,
      "learning_rate": 3.171934489891284e-06,
      "loss": 3.8272,
      "step": 18505
    },
    {
      "epoch": 0.7471134436818733,
      "grad_norm": 98.97628784179688,
      "learning_rate": 3.170979231056063e-06,
      "loss": 5.2211,
      "step": 18506
    },
    {
      "epoch": 0.74715381509891,
      "grad_norm": 142.5177459716797,
      "learning_rate": 3.1700240889811495e-06,
      "loss": 6.3213,
      "step": 18507
    },
    {
      "epoch": 0.7471941865159467,
      "grad_norm": 93.07129669189453,
      "learning_rate": 3.169069063682876e-06,
      "loss": 4.4073,
      "step": 18508
    },
    {
      "epoch": 0.7472345579329834,
      "grad_norm": 106.93978881835938,
      "learning_rate": 3.1681141551775694e-06,
      "loss": 5.7607,
      "step": 18509
    },
    {
      "epoch": 0.7472749293500202,
      "grad_norm": 114.29650115966797,
      "learning_rate": 3.1671593634815577e-06,
      "loss": 1.9069,
      "step": 18510
    },
    {
      "epoch": 0.7473153007670569,
      "grad_norm": 144.634033203125,
      "learning_rate": 3.1662046886111677e-06,
      "loss": 4.3902,
      "step": 18511
    },
    {
      "epoch": 0.7473556721840937,
      "grad_norm": 82.22310638427734,
      "learning_rate": 3.1652501305827165e-06,
      "loss": 5.9988,
      "step": 18512
    },
    {
      "epoch": 0.7473960436011304,
      "grad_norm": 82.40677642822266,
      "learning_rate": 3.1642956894125278e-06,
      "loss": 5.8002,
      "step": 18513
    },
    {
      "epoch": 0.7474364150181672,
      "grad_norm": 75.71802520751953,
      "learning_rate": 3.1633413651169196e-06,
      "loss": 4.6912,
      "step": 18514
    },
    {
      "epoch": 0.7474767864352039,
      "grad_norm": 83.58234405517578,
      "learning_rate": 3.1623871577122144e-06,
      "loss": 4.9724,
      "step": 18515
    },
    {
      "epoch": 0.7475171578522406,
      "grad_norm": 116.51041412353516,
      "learning_rate": 3.1614330672147175e-06,
      "loss": 4.9648,
      "step": 18516
    },
    {
      "epoch": 0.7475575292692773,
      "grad_norm": 76.5637435913086,
      "learning_rate": 3.160479093640746e-06,
      "loss": 4.6015,
      "step": 18517
    },
    {
      "epoch": 0.747597900686314,
      "grad_norm": 90.5601577758789,
      "learning_rate": 3.1595252370066098e-06,
      "loss": 5.8034,
      "step": 18518
    },
    {
      "epoch": 0.7476382721033509,
      "grad_norm": 100.87130737304688,
      "learning_rate": 3.158571497328622e-06,
      "loss": 4.8785,
      "step": 18519
    },
    {
      "epoch": 0.7476786435203876,
      "grad_norm": 72.19685363769531,
      "learning_rate": 3.1576178746230833e-06,
      "loss": 4.9889,
      "step": 18520
    },
    {
      "epoch": 0.7477190149374243,
      "grad_norm": 79.83649444580078,
      "learning_rate": 3.1566643689063004e-06,
      "loss": 5.3295,
      "step": 18521
    },
    {
      "epoch": 0.747759386354461,
      "grad_norm": 416.7921142578125,
      "learning_rate": 3.1557109801945763e-06,
      "loss": 4.8328,
      "step": 18522
    },
    {
      "epoch": 0.7477997577714978,
      "grad_norm": 74.19148254394531,
      "learning_rate": 3.1547577085042147e-06,
      "loss": 4.7295,
      "step": 18523
    },
    {
      "epoch": 0.7478401291885345,
      "grad_norm": 89.95901489257812,
      "learning_rate": 3.1538045538515083e-06,
      "loss": 6.6077,
      "step": 18524
    },
    {
      "epoch": 0.7478805006055712,
      "grad_norm": 91.17949676513672,
      "learning_rate": 3.1528515162527574e-06,
      "loss": 4.515,
      "step": 18525
    },
    {
      "epoch": 0.747920872022608,
      "grad_norm": 103.63665008544922,
      "learning_rate": 3.1518985957242563e-06,
      "loss": 5.6148,
      "step": 18526
    },
    {
      "epoch": 0.7479612434396448,
      "grad_norm": 158.73992919921875,
      "learning_rate": 3.150945792282298e-06,
      "loss": 4.7806,
      "step": 18527
    },
    {
      "epoch": 0.7480016148566815,
      "grad_norm": 155.4584503173828,
      "learning_rate": 3.149993105943173e-06,
      "loss": 5.8836,
      "step": 18528
    },
    {
      "epoch": 0.7480419862737182,
      "grad_norm": 68.47810363769531,
      "learning_rate": 3.1490405367231704e-06,
      "loss": 5.8283,
      "step": 18529
    },
    {
      "epoch": 0.7480823576907549,
      "grad_norm": 83.26817321777344,
      "learning_rate": 3.148088084638575e-06,
      "loss": 6.6244,
      "step": 18530
    },
    {
      "epoch": 0.7481227291077917,
      "grad_norm": 130.55023193359375,
      "learning_rate": 3.1471357497056745e-06,
      "loss": 4.6407,
      "step": 18531
    },
    {
      "epoch": 0.7481631005248284,
      "grad_norm": 85.93771362304688,
      "learning_rate": 3.1461835319407518e-06,
      "loss": 5.6901,
      "step": 18532
    },
    {
      "epoch": 0.7482034719418652,
      "grad_norm": 115.1086196899414,
      "learning_rate": 3.145231431360084e-06,
      "loss": 5.9193,
      "step": 18533
    },
    {
      "epoch": 0.7482438433589019,
      "grad_norm": 92.10901641845703,
      "learning_rate": 3.144279447979951e-06,
      "loss": 4.3732,
      "step": 18534
    },
    {
      "epoch": 0.7482842147759386,
      "grad_norm": 127.32320404052734,
      "learning_rate": 3.1433275818166296e-06,
      "loss": 5.7867,
      "step": 18535
    },
    {
      "epoch": 0.7483245861929754,
      "grad_norm": 141.51052856445312,
      "learning_rate": 3.1423758328863994e-06,
      "loss": 4.9264,
      "step": 18536
    },
    {
      "epoch": 0.7483649576100121,
      "grad_norm": 114.815673828125,
      "learning_rate": 3.141424201205525e-06,
      "loss": 4.2198,
      "step": 18537
    },
    {
      "epoch": 0.7484053290270488,
      "grad_norm": 143.37181091308594,
      "learning_rate": 3.1404726867902814e-06,
      "loss": 4.4631,
      "step": 18538
    },
    {
      "epoch": 0.7484457004440855,
      "grad_norm": 106.49366760253906,
      "learning_rate": 3.139521289656936e-06,
      "loss": 3.1526,
      "step": 18539
    },
    {
      "epoch": 0.7484860718611224,
      "grad_norm": 134.52537536621094,
      "learning_rate": 3.13857000982176e-06,
      "loss": 4.0253,
      "step": 18540
    },
    {
      "epoch": 0.7485264432781591,
      "grad_norm": 81.6397705078125,
      "learning_rate": 3.1376188473010106e-06,
      "loss": 3.0604,
      "step": 18541
    },
    {
      "epoch": 0.7485668146951958,
      "grad_norm": 89.92945098876953,
      "learning_rate": 3.1366678021109544e-06,
      "loss": 4.9534,
      "step": 18542
    },
    {
      "epoch": 0.7486071861122325,
      "grad_norm": 91.25080108642578,
      "learning_rate": 3.135716874267851e-06,
      "loss": 4.8154,
      "step": 18543
    },
    {
      "epoch": 0.7486475575292693,
      "grad_norm": 75.94107818603516,
      "learning_rate": 3.134766063787964e-06,
      "loss": 4.2906,
      "step": 18544
    },
    {
      "epoch": 0.748687928946306,
      "grad_norm": 125.1317138671875,
      "learning_rate": 3.133815370687543e-06,
      "loss": 6.094,
      "step": 18545
    },
    {
      "epoch": 0.7487283003633427,
      "grad_norm": 126.79341125488281,
      "learning_rate": 3.1328647949828438e-06,
      "loss": 5.4333,
      "step": 18546
    },
    {
      "epoch": 0.7487686717803795,
      "grad_norm": 97.75593566894531,
      "learning_rate": 3.1319143366901217e-06,
      "loss": 4.4941,
      "step": 18547
    },
    {
      "epoch": 0.7488090431974163,
      "grad_norm": 69.31257629394531,
      "learning_rate": 3.1309639958256264e-06,
      "loss": 4.7229,
      "step": 18548
    },
    {
      "epoch": 0.748849414614453,
      "grad_norm": 92.88790893554688,
      "learning_rate": 3.1300137724056056e-06,
      "loss": 5.6877,
      "step": 18549
    },
    {
      "epoch": 0.7488897860314897,
      "grad_norm": 91.83602905273438,
      "learning_rate": 3.129063666446308e-06,
      "loss": 4.2654,
      "step": 18550
    },
    {
      "epoch": 0.7489301574485264,
      "grad_norm": 129.2333221435547,
      "learning_rate": 3.1281136779639766e-06,
      "loss": 3.2813,
      "step": 18551
    },
    {
      "epoch": 0.7489705288655631,
      "grad_norm": 103.00863647460938,
      "learning_rate": 3.1271638069748567e-06,
      "loss": 6.15,
      "step": 18552
    },
    {
      "epoch": 0.7490109002825999,
      "grad_norm": 129.9466094970703,
      "learning_rate": 3.1262140534951834e-06,
      "loss": 6.9179,
      "step": 18553
    },
    {
      "epoch": 0.7490512716996367,
      "grad_norm": 101.30378723144531,
      "learning_rate": 3.1252644175411985e-06,
      "loss": 4.1689,
      "step": 18554
    },
    {
      "epoch": 0.7490916431166734,
      "grad_norm": 136.7465057373047,
      "learning_rate": 3.1243148991291393e-06,
      "loss": 2.7075,
      "step": 18555
    },
    {
      "epoch": 0.7491320145337101,
      "grad_norm": 89.93222045898438,
      "learning_rate": 3.1233654982752417e-06,
      "loss": 4.5671,
      "step": 18556
    },
    {
      "epoch": 0.7491723859507469,
      "grad_norm": 2522.708740234375,
      "learning_rate": 3.1224162149957327e-06,
      "loss": 3.6868,
      "step": 18557
    },
    {
      "epoch": 0.7492127573677836,
      "grad_norm": 116.58443450927734,
      "learning_rate": 3.1214670493068465e-06,
      "loss": 4.9264,
      "step": 18558
    },
    {
      "epoch": 0.7492531287848203,
      "grad_norm": 142.50714111328125,
      "learning_rate": 3.1205180012248114e-06,
      "loss": 5.4264,
      "step": 18559
    },
    {
      "epoch": 0.749293500201857,
      "grad_norm": 87.40544891357422,
      "learning_rate": 3.1195690707658567e-06,
      "loss": 4.7129,
      "step": 18560
    },
    {
      "epoch": 0.7493338716188939,
      "grad_norm": 114.73985290527344,
      "learning_rate": 3.1186202579462e-06,
      "loss": 4.3589,
      "step": 18561
    },
    {
      "epoch": 0.7493742430359306,
      "grad_norm": 99.43403625488281,
      "learning_rate": 3.1176715627820687e-06,
      "loss": 3.1127,
      "step": 18562
    },
    {
      "epoch": 0.7494146144529673,
      "grad_norm": 159.6890411376953,
      "learning_rate": 3.116722985289682e-06,
      "loss": 4.4537,
      "step": 18563
    },
    {
      "epoch": 0.749454985870004,
      "grad_norm": 112.13094329833984,
      "learning_rate": 3.1157745254852623e-06,
      "loss": 5.1,
      "step": 18564
    },
    {
      "epoch": 0.7494953572870408,
      "grad_norm": 105.93819427490234,
      "learning_rate": 3.114826183385018e-06,
      "loss": 3.9171,
      "step": 18565
    },
    {
      "epoch": 0.7495357287040775,
      "grad_norm": 103.22140502929688,
      "learning_rate": 3.113877959005169e-06,
      "loss": 4.344,
      "step": 18566
    },
    {
      "epoch": 0.7495761001211142,
      "grad_norm": 180.8795166015625,
      "learning_rate": 3.1129298523619268e-06,
      "loss": 5.188,
      "step": 18567
    },
    {
      "epoch": 0.749616471538151,
      "grad_norm": 76.88095092773438,
      "learning_rate": 3.1119818634715004e-06,
      "loss": 5.2314,
      "step": 18568
    },
    {
      "epoch": 0.7496568429551878,
      "grad_norm": 99.25016021728516,
      "learning_rate": 3.1110339923501e-06,
      "loss": 4.6983,
      "step": 18569
    },
    {
      "epoch": 0.7496972143722245,
      "grad_norm": 156.94654846191406,
      "learning_rate": 3.110086239013932e-06,
      "loss": 5.8479,
      "step": 18570
    },
    {
      "epoch": 0.7497375857892612,
      "grad_norm": 77.01211547851562,
      "learning_rate": 3.109138603479199e-06,
      "loss": 5.1126,
      "step": 18571
    },
    {
      "epoch": 0.7497779572062979,
      "grad_norm": 109.26884460449219,
      "learning_rate": 3.108191085762109e-06,
      "loss": 3.7232,
      "step": 18572
    },
    {
      "epoch": 0.7498183286233346,
      "grad_norm": 141.70277404785156,
      "learning_rate": 3.1072436858788548e-06,
      "loss": 4.6921,
      "step": 18573
    },
    {
      "epoch": 0.7498587000403714,
      "grad_norm": 92.6317138671875,
      "learning_rate": 3.106296403845638e-06,
      "loss": 5.7691,
      "step": 18574
    },
    {
      "epoch": 0.7498990714574082,
      "grad_norm": 163.95077514648438,
      "learning_rate": 3.105349239678653e-06,
      "loss": 3.6137,
      "step": 18575
    },
    {
      "epoch": 0.7499394428744449,
      "grad_norm": 80.3536148071289,
      "learning_rate": 3.1044021933941004e-06,
      "loss": 3.0973,
      "step": 18576
    },
    {
      "epoch": 0.7499798142914816,
      "grad_norm": 111.85819244384766,
      "learning_rate": 3.103455265008164e-06,
      "loss": 4.6128,
      "step": 18577
    },
    {
      "epoch": 0.7500201857085184,
      "grad_norm": 86.44755554199219,
      "learning_rate": 3.1025084545370387e-06,
      "loss": 3.4468,
      "step": 18578
    },
    {
      "epoch": 0.7500605571255551,
      "grad_norm": 115.35015106201172,
      "learning_rate": 3.101561761996912e-06,
      "loss": 4.1523,
      "step": 18579
    },
    {
      "epoch": 0.7501009285425918,
      "grad_norm": 75.95147705078125,
      "learning_rate": 3.1006151874039726e-06,
      "loss": 1.8827,
      "step": 18580
    },
    {
      "epoch": 0.7501412999596286,
      "grad_norm": 72.2266845703125,
      "learning_rate": 3.0996687307744e-06,
      "loss": 3.998,
      "step": 18581
    },
    {
      "epoch": 0.7501816713766654,
      "grad_norm": 84.0718994140625,
      "learning_rate": 3.098722392124378e-06,
      "loss": 3.4449,
      "step": 18582
    },
    {
      "epoch": 0.7502220427937021,
      "grad_norm": 120.53119659423828,
      "learning_rate": 3.0977761714700883e-06,
      "loss": 5.5505,
      "step": 18583
    },
    {
      "epoch": 0.7502624142107388,
      "grad_norm": 176.1216583251953,
      "learning_rate": 3.096830068827711e-06,
      "loss": 4.8653,
      "step": 18584
    },
    {
      "epoch": 0.7503027856277755,
      "grad_norm": 87.73690795898438,
      "learning_rate": 3.095884084213415e-06,
      "loss": 5.1605,
      "step": 18585
    },
    {
      "epoch": 0.7503431570448122,
      "grad_norm": 81.49224853515625,
      "learning_rate": 3.0949382176433806e-06,
      "loss": 4.8269,
      "step": 18586
    },
    {
      "epoch": 0.750383528461849,
      "grad_norm": 173.8786163330078,
      "learning_rate": 3.0939924691337775e-06,
      "loss": 4.128,
      "step": 18587
    },
    {
      "epoch": 0.7504238998788858,
      "grad_norm": 76.28604888916016,
      "learning_rate": 3.0930468387007763e-06,
      "loss": 6.8126,
      "step": 18588
    },
    {
      "epoch": 0.7504642712959225,
      "grad_norm": 278.5660095214844,
      "learning_rate": 3.092101326360546e-06,
      "loss": 4.7695,
      "step": 18589
    },
    {
      "epoch": 0.7505046427129592,
      "grad_norm": 65.51589965820312,
      "learning_rate": 3.0911559321292504e-06,
      "loss": 5.2595,
      "step": 18590
    },
    {
      "epoch": 0.750545014129996,
      "grad_norm": 74.9963150024414,
      "learning_rate": 3.0902106560230573e-06,
      "loss": 4.6839,
      "step": 18591
    },
    {
      "epoch": 0.7505853855470327,
      "grad_norm": 84.78135681152344,
      "learning_rate": 3.0892654980581283e-06,
      "loss": 5.0336,
      "step": 18592
    },
    {
      "epoch": 0.7506257569640694,
      "grad_norm": 107.50274658203125,
      "learning_rate": 3.0883204582506176e-06,
      "loss": 5.0172,
      "step": 18593
    },
    {
      "epoch": 0.7506661283811061,
      "grad_norm": 106.2278823852539,
      "learning_rate": 3.087375536616689e-06,
      "loss": 4.5331,
      "step": 18594
    },
    {
      "epoch": 0.750706499798143,
      "grad_norm": 94.43932342529297,
      "learning_rate": 3.086430733172495e-06,
      "loss": 6.0633,
      "step": 18595
    },
    {
      "epoch": 0.7507468712151797,
      "grad_norm": 91.47894287109375,
      "learning_rate": 3.0854860479341954e-06,
      "loss": 5.2234,
      "step": 18596
    },
    {
      "epoch": 0.7507872426322164,
      "grad_norm": 105.41983032226562,
      "learning_rate": 3.0845414809179342e-06,
      "loss": 3.6781,
      "step": 18597
    },
    {
      "epoch": 0.7508276140492531,
      "grad_norm": 161.2706756591797,
      "learning_rate": 3.0835970321398644e-06,
      "loss": 4.1945,
      "step": 18598
    },
    {
      "epoch": 0.7508679854662899,
      "grad_norm": 84.14263153076172,
      "learning_rate": 3.0826527016161356e-06,
      "loss": 4.2441,
      "step": 18599
    },
    {
      "epoch": 0.7509083568833266,
      "grad_norm": 83.6750717163086,
      "learning_rate": 3.0817084893628946e-06,
      "loss": 3.4794,
      "step": 18600
    },
    {
      "epoch": 0.7509487283003633,
      "grad_norm": 123.39905548095703,
      "learning_rate": 3.0807643953962796e-06,
      "loss": 5.4535,
      "step": 18601
    },
    {
      "epoch": 0.7509890997174001,
      "grad_norm": 91.97113037109375,
      "learning_rate": 3.079820419732437e-06,
      "loss": 5.1137,
      "step": 18602
    },
    {
      "epoch": 0.7510294711344369,
      "grad_norm": 92.71083068847656,
      "learning_rate": 3.078876562387505e-06,
      "loss": 3.2249,
      "step": 18603
    },
    {
      "epoch": 0.7510698425514736,
      "grad_norm": 72.9375,
      "learning_rate": 3.077932823377624e-06,
      "loss": 4.3958,
      "step": 18604
    },
    {
      "epoch": 0.7511102139685103,
      "grad_norm": 69.98292541503906,
      "learning_rate": 3.076989202718925e-06,
      "loss": 3.9288,
      "step": 18605
    },
    {
      "epoch": 0.751150585385547,
      "grad_norm": 126.69145965576172,
      "learning_rate": 3.076045700427546e-06,
      "loss": 5.5609,
      "step": 18606
    },
    {
      "epoch": 0.7511909568025837,
      "grad_norm": 115.79766082763672,
      "learning_rate": 3.075102316519615e-06,
      "loss": 5.4718,
      "step": 18607
    },
    {
      "epoch": 0.7512313282196205,
      "grad_norm": 99.19979858398438,
      "learning_rate": 3.0741590510112638e-06,
      "loss": 4.5234,
      "step": 18608
    },
    {
      "epoch": 0.7512716996366573,
      "grad_norm": 122.60383605957031,
      "learning_rate": 3.0732159039186194e-06,
      "loss": 4.6467,
      "step": 18609
    },
    {
      "epoch": 0.751312071053694,
      "grad_norm": 163.47116088867188,
      "learning_rate": 3.072272875257809e-06,
      "loss": 4.6576,
      "step": 18610
    },
    {
      "epoch": 0.7513524424707307,
      "grad_norm": 98.20511627197266,
      "learning_rate": 3.0713299650449545e-06,
      "loss": 3.2462,
      "step": 18611
    },
    {
      "epoch": 0.7513928138877675,
      "grad_norm": 120.9020767211914,
      "learning_rate": 3.070387173296181e-06,
      "loss": 4.8371,
      "step": 18612
    },
    {
      "epoch": 0.7514331853048042,
      "grad_norm": 105.03784942626953,
      "learning_rate": 3.0694445000276017e-06,
      "loss": 5.3095,
      "step": 18613
    },
    {
      "epoch": 0.7514735567218409,
      "grad_norm": 103.65589904785156,
      "learning_rate": 3.0685019452553378e-06,
      "loss": 5.7014,
      "step": 18614
    },
    {
      "epoch": 0.7515139281388776,
      "grad_norm": 181.78875732421875,
      "learning_rate": 3.0675595089955044e-06,
      "loss": 4.2294,
      "step": 18615
    },
    {
      "epoch": 0.7515542995559145,
      "grad_norm": 70.29061126708984,
      "learning_rate": 3.0666171912642186e-06,
      "loss": 4.9409,
      "step": 18616
    },
    {
      "epoch": 0.7515946709729512,
      "grad_norm": 59.70857620239258,
      "learning_rate": 3.065674992077584e-06,
      "loss": 4.0005,
      "step": 18617
    },
    {
      "epoch": 0.7516350423899879,
      "grad_norm": 126.76216888427734,
      "learning_rate": 3.0647329114517143e-06,
      "loss": 6.0509,
      "step": 18618
    },
    {
      "epoch": 0.7516754138070246,
      "grad_norm": 87.81543731689453,
      "learning_rate": 3.0637909494027175e-06,
      "loss": 5.0173,
      "step": 18619
    },
    {
      "epoch": 0.7517157852240614,
      "grad_norm": 111.8680419921875,
      "learning_rate": 3.0628491059467015e-06,
      "loss": 4.7466,
      "step": 18620
    },
    {
      "epoch": 0.7517561566410981,
      "grad_norm": 111.86115264892578,
      "learning_rate": 3.0619073810997625e-06,
      "loss": 4.0462,
      "step": 18621
    },
    {
      "epoch": 0.7517965280581348,
      "grad_norm": 83.2537612915039,
      "learning_rate": 3.0609657748780065e-06,
      "loss": 4.9647,
      "step": 18622
    },
    {
      "epoch": 0.7518368994751716,
      "grad_norm": 92.33472442626953,
      "learning_rate": 3.060024287297532e-06,
      "loss": 4.2504,
      "step": 18623
    },
    {
      "epoch": 0.7518772708922083,
      "grad_norm": 109.76759338378906,
      "learning_rate": 3.0590829183744387e-06,
      "loss": 5.085,
      "step": 18624
    },
    {
      "epoch": 0.7519176423092451,
      "grad_norm": 74.77719116210938,
      "learning_rate": 3.0581416681248153e-06,
      "loss": 4.6005,
      "step": 18625
    },
    {
      "epoch": 0.7519580137262818,
      "grad_norm": 125.94602966308594,
      "learning_rate": 3.0572005365647605e-06,
      "loss": 6.0178,
      "step": 18626
    },
    {
      "epoch": 0.7519983851433185,
      "grad_norm": 95.09905242919922,
      "learning_rate": 3.056259523710363e-06,
      "loss": 4.6571,
      "step": 18627
    },
    {
      "epoch": 0.7520387565603552,
      "grad_norm": 91.48646545410156,
      "learning_rate": 3.0553186295777137e-06,
      "loss": 4.7235,
      "step": 18628
    },
    {
      "epoch": 0.752079127977392,
      "grad_norm": 101.26073455810547,
      "learning_rate": 3.0543778541828984e-06,
      "loss": 5.5525,
      "step": 18629
    },
    {
      "epoch": 0.7521194993944288,
      "grad_norm": 88.8955307006836,
      "learning_rate": 3.053437197542002e-06,
      "loss": 5.7523,
      "step": 18630
    },
    {
      "epoch": 0.7521598708114655,
      "grad_norm": 139.56675720214844,
      "learning_rate": 3.052496659671109e-06,
      "loss": 5.0224,
      "step": 18631
    },
    {
      "epoch": 0.7522002422285022,
      "grad_norm": 104.40364837646484,
      "learning_rate": 3.051556240586302e-06,
      "loss": 3.9328,
      "step": 18632
    },
    {
      "epoch": 0.752240613645539,
      "grad_norm": 101.43001556396484,
      "learning_rate": 3.0506159403036552e-06,
      "loss": 4.9314,
      "step": 18633
    },
    {
      "epoch": 0.7522809850625757,
      "grad_norm": 209.3549041748047,
      "learning_rate": 3.0496757588392467e-06,
      "loss": 6.151,
      "step": 18634
    },
    {
      "epoch": 0.7523213564796124,
      "grad_norm": 72.4827880859375,
      "learning_rate": 3.048735696209154e-06,
      "loss": 3.4621,
      "step": 18635
    },
    {
      "epoch": 0.7523617278966491,
      "grad_norm": 95.3387451171875,
      "learning_rate": 3.047795752429451e-06,
      "loss": 4.9318,
      "step": 18636
    },
    {
      "epoch": 0.752402099313686,
      "grad_norm": 98.42753601074219,
      "learning_rate": 3.0468559275162025e-06,
      "loss": 5.8361,
      "step": 18637
    },
    {
      "epoch": 0.7524424707307227,
      "grad_norm": 142.0778045654297,
      "learning_rate": 3.045916221485481e-06,
      "loss": 4.7758,
      "step": 18638
    },
    {
      "epoch": 0.7524828421477594,
      "grad_norm": 96.615966796875,
      "learning_rate": 3.0449766343533536e-06,
      "loss": 5.1433,
      "step": 18639
    },
    {
      "epoch": 0.7525232135647961,
      "grad_norm": 153.03224182128906,
      "learning_rate": 3.044037166135887e-06,
      "loss": 5.4637,
      "step": 18640
    },
    {
      "epoch": 0.7525635849818328,
      "grad_norm": 91.2318115234375,
      "learning_rate": 3.043097816849139e-06,
      "loss": 4.0309,
      "step": 18641
    },
    {
      "epoch": 0.7526039563988696,
      "grad_norm": 100.06661224365234,
      "learning_rate": 3.042158586509173e-06,
      "loss": 5.3939,
      "step": 18642
    },
    {
      "epoch": 0.7526443278159063,
      "grad_norm": 82.50520324707031,
      "learning_rate": 3.0412194751320467e-06,
      "loss": 4.2564,
      "step": 18643
    },
    {
      "epoch": 0.7526846992329431,
      "grad_norm": 103.99314880371094,
      "learning_rate": 3.0402804827338216e-06,
      "loss": 4.3705,
      "step": 18644
    },
    {
      "epoch": 0.7527250706499798,
      "grad_norm": 140.32289123535156,
      "learning_rate": 3.039341609330544e-06,
      "loss": 5.4233,
      "step": 18645
    },
    {
      "epoch": 0.7527654420670166,
      "grad_norm": 91.15987396240234,
      "learning_rate": 3.0384028549382707e-06,
      "loss": 5.2822,
      "step": 18646
    },
    {
      "epoch": 0.7528058134840533,
      "grad_norm": 74.6319351196289,
      "learning_rate": 3.037464219573053e-06,
      "loss": 3.843,
      "step": 18647
    },
    {
      "epoch": 0.75284618490109,
      "grad_norm": 127.7280502319336,
      "learning_rate": 3.0365257032509376e-06,
      "loss": 5.2473,
      "step": 18648
    },
    {
      "epoch": 0.7528865563181267,
      "grad_norm": 108.08081817626953,
      "learning_rate": 3.0355873059879714e-06,
      "loss": 2.1905,
      "step": 18649
    },
    {
      "epoch": 0.7529269277351635,
      "grad_norm": 167.29495239257812,
      "learning_rate": 3.0346490278002007e-06,
      "loss": 5.6958,
      "step": 18650
    },
    {
      "epoch": 0.7529672991522003,
      "grad_norm": 132.49771118164062,
      "learning_rate": 3.0337108687036653e-06,
      "loss": 3.1458,
      "step": 18651
    },
    {
      "epoch": 0.753007670569237,
      "grad_norm": 101.34667205810547,
      "learning_rate": 3.0327728287144097e-06,
      "loss": 4.6577,
      "step": 18652
    },
    {
      "epoch": 0.7530480419862737,
      "grad_norm": 100.85482025146484,
      "learning_rate": 3.0318349078484653e-06,
      "loss": 3.8944,
      "step": 18653
    },
    {
      "epoch": 0.7530884134033105,
      "grad_norm": 74.78487396240234,
      "learning_rate": 3.030897106121874e-06,
      "loss": 4.5209,
      "step": 18654
    },
    {
      "epoch": 0.7531287848203472,
      "grad_norm": 84.45988464355469,
      "learning_rate": 3.029959423550667e-06,
      "loss": 4.8743,
      "step": 18655
    },
    {
      "epoch": 0.7531691562373839,
      "grad_norm": 120.44917297363281,
      "learning_rate": 3.02902186015088e-06,
      "loss": 6.2623,
      "step": 18656
    },
    {
      "epoch": 0.7532095276544206,
      "grad_norm": 120.72090148925781,
      "learning_rate": 3.0280844159385394e-06,
      "loss": 3.9528,
      "step": 18657
    },
    {
      "epoch": 0.7532498990714575,
      "grad_norm": 85.95531463623047,
      "learning_rate": 3.0271470909296743e-06,
      "loss": 4.8848,
      "step": 18658
    },
    {
      "epoch": 0.7532902704884942,
      "grad_norm": 104.46329498291016,
      "learning_rate": 3.02620988514031e-06,
      "loss": 4.1193,
      "step": 18659
    },
    {
      "epoch": 0.7533306419055309,
      "grad_norm": 93.8287124633789,
      "learning_rate": 3.025272798586476e-06,
      "loss": 5.7461,
      "step": 18660
    },
    {
      "epoch": 0.7533710133225676,
      "grad_norm": 110.29188537597656,
      "learning_rate": 3.0243358312841864e-06,
      "loss": 4.5147,
      "step": 18661
    },
    {
      "epoch": 0.7534113847396043,
      "grad_norm": 125.8672866821289,
      "learning_rate": 3.0233989832494647e-06,
      "loss": 5.0441,
      "step": 18662
    },
    {
      "epoch": 0.7534517561566411,
      "grad_norm": 93.06647491455078,
      "learning_rate": 3.0224622544983285e-06,
      "loss": 5.2068,
      "step": 18663
    },
    {
      "epoch": 0.7534921275736778,
      "grad_norm": 116.45672607421875,
      "learning_rate": 3.0215256450467943e-06,
      "loss": 4.5119,
      "step": 18664
    },
    {
      "epoch": 0.7535324989907146,
      "grad_norm": 104.53024291992188,
      "learning_rate": 3.0205891549108757e-06,
      "loss": 4.458,
      "step": 18665
    },
    {
      "epoch": 0.7535728704077513,
      "grad_norm": 87.92798614501953,
      "learning_rate": 3.0196527841065846e-06,
      "loss": 5.6779,
      "step": 18666
    },
    {
      "epoch": 0.7536132418247881,
      "grad_norm": 144.84982299804688,
      "learning_rate": 3.0187165326499303e-06,
      "loss": 4.7631,
      "step": 18667
    },
    {
      "epoch": 0.7536536132418248,
      "grad_norm": 79.65505981445312,
      "learning_rate": 3.0177804005569244e-06,
      "loss": 5.2928,
      "step": 18668
    },
    {
      "epoch": 0.7536939846588615,
      "grad_norm": 132.6295166015625,
      "learning_rate": 3.016844387843566e-06,
      "loss": 3.6816,
      "step": 18669
    },
    {
      "epoch": 0.7537343560758982,
      "grad_norm": 92.63013458251953,
      "learning_rate": 3.015908494525861e-06,
      "loss": 4.8144,
      "step": 18670
    },
    {
      "epoch": 0.753774727492935,
      "grad_norm": 796.7803955078125,
      "learning_rate": 3.014972720619812e-06,
      "loss": 5.9571,
      "step": 18671
    },
    {
      "epoch": 0.7538150989099718,
      "grad_norm": 119.94341278076172,
      "learning_rate": 3.014037066141422e-06,
      "loss": 4.4967,
      "step": 18672
    },
    {
      "epoch": 0.7538554703270085,
      "grad_norm": 95.10968780517578,
      "learning_rate": 3.0131015311066814e-06,
      "loss": 3.6416,
      "step": 18673
    },
    {
      "epoch": 0.7538958417440452,
      "grad_norm": 70.24320983886719,
      "learning_rate": 3.0121661155315883e-06,
      "loss": 5.052,
      "step": 18674
    },
    {
      "epoch": 0.753936213161082,
      "grad_norm": 124.15052032470703,
      "learning_rate": 3.0112308194321373e-06,
      "loss": 5.7956,
      "step": 18675
    },
    {
      "epoch": 0.7539765845781187,
      "grad_norm": 139.04928588867188,
      "learning_rate": 3.0102956428243226e-06,
      "loss": 4.7141,
      "step": 18676
    },
    {
      "epoch": 0.7540169559951554,
      "grad_norm": 82.47632598876953,
      "learning_rate": 3.009360585724127e-06,
      "loss": 5.2711,
      "step": 18677
    },
    {
      "epoch": 0.7540573274121921,
      "grad_norm": 82.8448486328125,
      "learning_rate": 3.0084256481475414e-06,
      "loss": 3.6898,
      "step": 18678
    },
    {
      "epoch": 0.754097698829229,
      "grad_norm": 88.72457885742188,
      "learning_rate": 3.00749083011055e-06,
      "loss": 3.6493,
      "step": 18679
    },
    {
      "epoch": 0.7541380702462657,
      "grad_norm": 99.91867065429688,
      "learning_rate": 3.0065561316291404e-06,
      "loss": 5.978,
      "step": 18680
    },
    {
      "epoch": 0.7541784416633024,
      "grad_norm": 86.25947570800781,
      "learning_rate": 3.0056215527192867e-06,
      "loss": 5.8583,
      "step": 18681
    },
    {
      "epoch": 0.7542188130803391,
      "grad_norm": 132.75184631347656,
      "learning_rate": 3.004687093396972e-06,
      "loss": 5.7099,
      "step": 18682
    },
    {
      "epoch": 0.7542591844973758,
      "grad_norm": 132.1101531982422,
      "learning_rate": 3.003752753678172e-06,
      "loss": 4.8151,
      "step": 18683
    },
    {
      "epoch": 0.7542995559144126,
      "grad_norm": 135.16177368164062,
      "learning_rate": 3.002818533578863e-06,
      "loss": 2.8685,
      "step": 18684
    },
    {
      "epoch": 0.7543399273314494,
      "grad_norm": 108.9643325805664,
      "learning_rate": 3.0018844331150175e-06,
      "loss": 4.5226,
      "step": 18685
    },
    {
      "epoch": 0.7543802987484861,
      "grad_norm": 138.7592315673828,
      "learning_rate": 3.0009504523026055e-06,
      "loss": 5.6931,
      "step": 18686
    },
    {
      "epoch": 0.7544206701655228,
      "grad_norm": 66.82849884033203,
      "learning_rate": 3.0000165911575975e-06,
      "loss": 4.9615,
      "step": 18687
    },
    {
      "epoch": 0.7544610415825596,
      "grad_norm": 78.8473129272461,
      "learning_rate": 2.9990828496959635e-06,
      "loss": 5.0282,
      "step": 18688
    },
    {
      "epoch": 0.7545014129995963,
      "grad_norm": 166.2843017578125,
      "learning_rate": 2.99814922793366e-06,
      "loss": 2.6943,
      "step": 18689
    },
    {
      "epoch": 0.754541784416633,
      "grad_norm": 110.282470703125,
      "learning_rate": 2.997215725886655e-06,
      "loss": 7.5476,
      "step": 18690
    },
    {
      "epoch": 0.7545821558336697,
      "grad_norm": 97.79183197021484,
      "learning_rate": 2.9962823435709087e-06,
      "loss": 5.5387,
      "step": 18691
    },
    {
      "epoch": 0.7546225272507066,
      "grad_norm": 74.7618408203125,
      "learning_rate": 2.9953490810023823e-06,
      "loss": 5.0321,
      "step": 18692
    },
    {
      "epoch": 0.7546628986677433,
      "grad_norm": 131.09629821777344,
      "learning_rate": 2.9944159381970263e-06,
      "loss": 5.015,
      "step": 18693
    },
    {
      "epoch": 0.75470327008478,
      "grad_norm": 98.81375122070312,
      "learning_rate": 2.9934829151707976e-06,
      "loss": 4.8492,
      "step": 18694
    },
    {
      "epoch": 0.7547436415018167,
      "grad_norm": 89.0088882446289,
      "learning_rate": 2.9925500119396513e-06,
      "loss": 5.2551,
      "step": 18695
    },
    {
      "epoch": 0.7547840129188534,
      "grad_norm": 114.90646362304688,
      "learning_rate": 2.9916172285195388e-06,
      "loss": 6.2015,
      "step": 18696
    },
    {
      "epoch": 0.7548243843358902,
      "grad_norm": 189.34674072265625,
      "learning_rate": 2.990684564926403e-06,
      "loss": 4.8052,
      "step": 18697
    },
    {
      "epoch": 0.7548647557529269,
      "grad_norm": 137.65025329589844,
      "learning_rate": 2.9897520211761922e-06,
      "loss": 6.7054,
      "step": 18698
    },
    {
      "epoch": 0.7549051271699637,
      "grad_norm": 165.20657348632812,
      "learning_rate": 2.988819597284852e-06,
      "loss": 3.7224,
      "step": 18699
    },
    {
      "epoch": 0.7549454985870004,
      "grad_norm": 107.3153076171875,
      "learning_rate": 2.987887293268328e-06,
      "loss": 5.4672,
      "step": 18700
    },
    {
      "epoch": 0.7549858700040372,
      "grad_norm": 98.4930191040039,
      "learning_rate": 2.986955109142554e-06,
      "loss": 5.3953,
      "step": 18701
    },
    {
      "epoch": 0.7550262414210739,
      "grad_norm": 82.74871063232422,
      "learning_rate": 2.9860230449234705e-06,
      "loss": 3.7606,
      "step": 18702
    },
    {
      "epoch": 0.7550666128381106,
      "grad_norm": 154.82310485839844,
      "learning_rate": 2.9850911006270144e-06,
      "loss": 6.4364,
      "step": 18703
    },
    {
      "epoch": 0.7551069842551473,
      "grad_norm": 93.03240203857422,
      "learning_rate": 2.9841592762691195e-06,
      "loss": 5.4367,
      "step": 18704
    },
    {
      "epoch": 0.755147355672184,
      "grad_norm": 100.53645324707031,
      "learning_rate": 2.983227571865718e-06,
      "loss": 4.9288,
      "step": 18705
    },
    {
      "epoch": 0.7551877270892209,
      "grad_norm": 103.15665435791016,
      "learning_rate": 2.9822959874327383e-06,
      "loss": 4.5415,
      "step": 18706
    },
    {
      "epoch": 0.7552280985062576,
      "grad_norm": 131.80108642578125,
      "learning_rate": 2.981364522986111e-06,
      "loss": 6.3215,
      "step": 18707
    },
    {
      "epoch": 0.7552684699232943,
      "grad_norm": 143.20498657226562,
      "learning_rate": 2.980433178541764e-06,
      "loss": 3.9349,
      "step": 18708
    },
    {
      "epoch": 0.755308841340331,
      "grad_norm": 104.60794830322266,
      "learning_rate": 2.9795019541156133e-06,
      "loss": 3.8055,
      "step": 18709
    },
    {
      "epoch": 0.7553492127573678,
      "grad_norm": 86.0743637084961,
      "learning_rate": 2.9785708497235866e-06,
      "loss": 4.2167,
      "step": 18710
    },
    {
      "epoch": 0.7553895841744045,
      "grad_norm": 121.76589965820312,
      "learning_rate": 2.977639865381602e-06,
      "loss": 4.5036,
      "step": 18711
    },
    {
      "epoch": 0.7554299555914412,
      "grad_norm": 86.17860412597656,
      "learning_rate": 2.97670900110558e-06,
      "loss": 5.1075,
      "step": 18712
    },
    {
      "epoch": 0.755470327008478,
      "grad_norm": 87.47584533691406,
      "learning_rate": 2.97577825691143e-06,
      "loss": 4.5651,
      "step": 18713
    },
    {
      "epoch": 0.7555106984255148,
      "grad_norm": 124.81244659423828,
      "learning_rate": 2.9748476328150688e-06,
      "loss": 5.0048,
      "step": 18714
    },
    {
      "epoch": 0.7555510698425515,
      "grad_norm": 126.82270050048828,
      "learning_rate": 2.9739171288324097e-06,
      "loss": 4.5612,
      "step": 18715
    },
    {
      "epoch": 0.7555914412595882,
      "grad_norm": 74.6983413696289,
      "learning_rate": 2.9729867449793627e-06,
      "loss": 5.5338,
      "step": 18716
    },
    {
      "epoch": 0.7556318126766249,
      "grad_norm": 75.49854278564453,
      "learning_rate": 2.97205648127183e-06,
      "loss": 4.67,
      "step": 18717
    },
    {
      "epoch": 0.7556721840936617,
      "grad_norm": 76.70748901367188,
      "learning_rate": 2.9711263377257204e-06,
      "loss": 3.9991,
      "step": 18718
    },
    {
      "epoch": 0.7557125555106984,
      "grad_norm": 111.44364929199219,
      "learning_rate": 2.9701963143569366e-06,
      "loss": 4.5559,
      "step": 18719
    },
    {
      "epoch": 0.7557529269277352,
      "grad_norm": 144.78555297851562,
      "learning_rate": 2.969266411181383e-06,
      "loss": 4.7493,
      "step": 18720
    },
    {
      "epoch": 0.7557932983447719,
      "grad_norm": 78.95931243896484,
      "learning_rate": 2.9683366282149527e-06,
      "loss": 4.9438,
      "step": 18721
    },
    {
      "epoch": 0.7558336697618087,
      "grad_norm": 99.77582550048828,
      "learning_rate": 2.967406965473546e-06,
      "loss": 5.8627,
      "step": 18722
    },
    {
      "epoch": 0.7558740411788454,
      "grad_norm": 152.705078125,
      "learning_rate": 2.9664774229730587e-06,
      "loss": 3.2187,
      "step": 18723
    },
    {
      "epoch": 0.7559144125958821,
      "grad_norm": 138.59146118164062,
      "learning_rate": 2.9655480007293825e-06,
      "loss": 5.5768,
      "step": 18724
    },
    {
      "epoch": 0.7559547840129188,
      "grad_norm": 69.858642578125,
      "learning_rate": 2.964618698758408e-06,
      "loss": 3.8706,
      "step": 18725
    },
    {
      "epoch": 0.7559951554299555,
      "grad_norm": 80.2884521484375,
      "learning_rate": 2.963689517076026e-06,
      "loss": 5.9644,
      "step": 18726
    },
    {
      "epoch": 0.7560355268469924,
      "grad_norm": 95.0284194946289,
      "learning_rate": 2.9627604556981216e-06,
      "loss": 4.0191,
      "step": 18727
    },
    {
      "epoch": 0.7560758982640291,
      "grad_norm": 128.79348754882812,
      "learning_rate": 2.961831514640584e-06,
      "loss": 3.9885,
      "step": 18728
    },
    {
      "epoch": 0.7561162696810658,
      "grad_norm": 152.05523681640625,
      "learning_rate": 2.9609026939192885e-06,
      "loss": 4.2512,
      "step": 18729
    },
    {
      "epoch": 0.7561566410981025,
      "grad_norm": 122.16399383544922,
      "learning_rate": 2.9599739935501204e-06,
      "loss": 5.5414,
      "step": 18730
    },
    {
      "epoch": 0.7561970125151393,
      "grad_norm": 67.34852600097656,
      "learning_rate": 2.9590454135489564e-06,
      "loss": 4.478,
      "step": 18731
    },
    {
      "epoch": 0.756237383932176,
      "grad_norm": 120.23512268066406,
      "learning_rate": 2.958116953931678e-06,
      "loss": 4.5008,
      "step": 18732
    },
    {
      "epoch": 0.7562777553492127,
      "grad_norm": 128.5689239501953,
      "learning_rate": 2.9571886147141515e-06,
      "loss": 4.3156,
      "step": 18733
    },
    {
      "epoch": 0.7563181267662495,
      "grad_norm": 136.3596954345703,
      "learning_rate": 2.9562603959122536e-06,
      "loss": 4.1677,
      "step": 18734
    },
    {
      "epoch": 0.7563584981832863,
      "grad_norm": 136.78799438476562,
      "learning_rate": 2.955332297541855e-06,
      "loss": 4.9574,
      "step": 18735
    },
    {
      "epoch": 0.756398869600323,
      "grad_norm": 95.82199096679688,
      "learning_rate": 2.9544043196188276e-06,
      "loss": 4.661,
      "step": 18736
    },
    {
      "epoch": 0.7564392410173597,
      "grad_norm": 73.40798950195312,
      "learning_rate": 2.9534764621590293e-06,
      "loss": 5.2767,
      "step": 18737
    },
    {
      "epoch": 0.7564796124343964,
      "grad_norm": 87.42599487304688,
      "learning_rate": 2.9525487251783302e-06,
      "loss": 4.5867,
      "step": 18738
    },
    {
      "epoch": 0.7565199838514332,
      "grad_norm": 85.55730438232422,
      "learning_rate": 2.9516211086925894e-06,
      "loss": 5.306,
      "step": 18739
    },
    {
      "epoch": 0.7565603552684699,
      "grad_norm": 87.1041488647461,
      "learning_rate": 2.950693612717672e-06,
      "loss": 5.4638,
      "step": 18740
    },
    {
      "epoch": 0.7566007266855067,
      "grad_norm": 142.5732879638672,
      "learning_rate": 2.94976623726943e-06,
      "loss": 3.7582,
      "step": 18741
    },
    {
      "epoch": 0.7566410981025434,
      "grad_norm": 98.12005615234375,
      "learning_rate": 2.948838982363721e-06,
      "loss": 5.0378,
      "step": 18742
    },
    {
      "epoch": 0.7566814695195802,
      "grad_norm": 158.3925018310547,
      "learning_rate": 2.9479118480164002e-06,
      "loss": 3.4516,
      "step": 18743
    },
    {
      "epoch": 0.7567218409366169,
      "grad_norm": 99.60600280761719,
      "learning_rate": 2.9469848342433193e-06,
      "loss": 4.4723,
      "step": 18744
    },
    {
      "epoch": 0.7567622123536536,
      "grad_norm": 139.42503356933594,
      "learning_rate": 2.9460579410603272e-06,
      "loss": 6.2402,
      "step": 18745
    },
    {
      "epoch": 0.7568025837706903,
      "grad_norm": 169.32919311523438,
      "learning_rate": 2.945131168483273e-06,
      "loss": 5.6734,
      "step": 18746
    },
    {
      "epoch": 0.756842955187727,
      "grad_norm": 127.55071258544922,
      "learning_rate": 2.944204516528001e-06,
      "loss": 3.6132,
      "step": 18747
    },
    {
      "epoch": 0.7568833266047639,
      "grad_norm": 75.84904479980469,
      "learning_rate": 2.9432779852103576e-06,
      "loss": 3.6178,
      "step": 18748
    },
    {
      "epoch": 0.7569236980218006,
      "grad_norm": 111.724365234375,
      "learning_rate": 2.9423515745461794e-06,
      "loss": 5.5356,
      "step": 18749
    },
    {
      "epoch": 0.7569640694388373,
      "grad_norm": 83.51021575927734,
      "learning_rate": 2.9414252845513093e-06,
      "loss": 5.2214,
      "step": 18750
    },
    {
      "epoch": 0.757004440855874,
      "grad_norm": 99.22997283935547,
      "learning_rate": 2.940499115241583e-06,
      "loss": 5.2477,
      "step": 18751
    },
    {
      "epoch": 0.7570448122729108,
      "grad_norm": 89.04503631591797,
      "learning_rate": 2.9395730666328405e-06,
      "loss": 6.2583,
      "step": 18752
    },
    {
      "epoch": 0.7570851836899475,
      "grad_norm": 97.46593475341797,
      "learning_rate": 2.9386471387409076e-06,
      "loss": 4.9234,
      "step": 18753
    },
    {
      "epoch": 0.7571255551069842,
      "grad_norm": 106.76626586914062,
      "learning_rate": 2.937721331581619e-06,
      "loss": 3.5353,
      "step": 18754
    },
    {
      "epoch": 0.757165926524021,
      "grad_norm": 90.37041473388672,
      "learning_rate": 2.936795645170805e-06,
      "loss": 4.0815,
      "step": 18755
    },
    {
      "epoch": 0.7572062979410578,
      "grad_norm": 220.98052978515625,
      "learning_rate": 2.935870079524293e-06,
      "loss": 5.3054,
      "step": 18756
    },
    {
      "epoch": 0.7572466693580945,
      "grad_norm": 112.02436828613281,
      "learning_rate": 2.9349446346579057e-06,
      "loss": 5.7887,
      "step": 18757
    },
    {
      "epoch": 0.7572870407751312,
      "grad_norm": 98.12741088867188,
      "learning_rate": 2.9340193105874658e-06,
      "loss": 4.1023,
      "step": 18758
    },
    {
      "epoch": 0.7573274121921679,
      "grad_norm": 92.93268585205078,
      "learning_rate": 2.9330941073287956e-06,
      "loss": 5.6273,
      "step": 18759
    },
    {
      "epoch": 0.7573677836092046,
      "grad_norm": 91.82853698730469,
      "learning_rate": 2.9321690248977176e-06,
      "loss": 4.801,
      "step": 18760
    },
    {
      "epoch": 0.7574081550262414,
      "grad_norm": 107.12726593017578,
      "learning_rate": 2.9312440633100414e-06,
      "loss": 4.6239,
      "step": 18761
    },
    {
      "epoch": 0.7574485264432782,
      "grad_norm": 87.7756118774414,
      "learning_rate": 2.9303192225815844e-06,
      "loss": 4.4176,
      "step": 18762
    },
    {
      "epoch": 0.7574888978603149,
      "grad_norm": 64.3674545288086,
      "learning_rate": 2.92939450272816e-06,
      "loss": 5.8779,
      "step": 18763
    },
    {
      "epoch": 0.7575292692773516,
      "grad_norm": 177.8526153564453,
      "learning_rate": 2.9284699037655782e-06,
      "loss": 4.9017,
      "step": 18764
    },
    {
      "epoch": 0.7575696406943884,
      "grad_norm": 103.00899505615234,
      "learning_rate": 2.927545425709648e-06,
      "loss": 4.8655,
      "step": 18765
    },
    {
      "epoch": 0.7576100121114251,
      "grad_norm": 109.1463623046875,
      "learning_rate": 2.9266210685761766e-06,
      "loss": 5.1906,
      "step": 18766
    },
    {
      "epoch": 0.7576503835284618,
      "grad_norm": 97.33605194091797,
      "learning_rate": 2.9256968323809664e-06,
      "loss": 5.3723,
      "step": 18767
    },
    {
      "epoch": 0.7576907549454985,
      "grad_norm": 85.4611587524414,
      "learning_rate": 2.9247727171398243e-06,
      "loss": 5.5105,
      "step": 18768
    },
    {
      "epoch": 0.7577311263625354,
      "grad_norm": 96.07029724121094,
      "learning_rate": 2.923848722868543e-06,
      "loss": 4.8151,
      "step": 18769
    },
    {
      "epoch": 0.7577714977795721,
      "grad_norm": 86.70113372802734,
      "learning_rate": 2.9229248495829254e-06,
      "loss": 4.4591,
      "step": 18770
    },
    {
      "epoch": 0.7578118691966088,
      "grad_norm": 110.39323425292969,
      "learning_rate": 2.9220010972987665e-06,
      "loss": 4.5249,
      "step": 18771
    },
    {
      "epoch": 0.7578522406136455,
      "grad_norm": 139.9979248046875,
      "learning_rate": 2.921077466031863e-06,
      "loss": 6.2555,
      "step": 18772
    },
    {
      "epoch": 0.7578926120306823,
      "grad_norm": 108.40655517578125,
      "learning_rate": 2.920153955798002e-06,
      "loss": 5.4057,
      "step": 18773
    },
    {
      "epoch": 0.757932983447719,
      "grad_norm": 103.6406021118164,
      "learning_rate": 2.919230566612976e-06,
      "loss": 3.7392,
      "step": 18774
    },
    {
      "epoch": 0.7579733548647557,
      "grad_norm": 100.33356475830078,
      "learning_rate": 2.918307298492572e-06,
      "loss": 5.6659,
      "step": 18775
    },
    {
      "epoch": 0.7580137262817925,
      "grad_norm": 90.43357849121094,
      "learning_rate": 2.917384151452579e-06,
      "loss": 5.3317,
      "step": 18776
    },
    {
      "epoch": 0.7580540976988293,
      "grad_norm": 107.82775115966797,
      "learning_rate": 2.916461125508775e-06,
      "loss": 4.1221,
      "step": 18777
    },
    {
      "epoch": 0.758094469115866,
      "grad_norm": 109.60066986083984,
      "learning_rate": 2.9155382206769445e-06,
      "loss": 4.6979,
      "step": 18778
    },
    {
      "epoch": 0.7581348405329027,
      "grad_norm": 83.8216323852539,
      "learning_rate": 2.9146154369728674e-06,
      "loss": 4.8043,
      "step": 18779
    },
    {
      "epoch": 0.7581752119499394,
      "grad_norm": 230.30648803710938,
      "learning_rate": 2.9136927744123246e-06,
      "loss": 5.1769,
      "step": 18780
    },
    {
      "epoch": 0.7582155833669761,
      "grad_norm": 97.66735076904297,
      "learning_rate": 2.912770233011083e-06,
      "loss": 4.8628,
      "step": 18781
    },
    {
      "epoch": 0.7582559547840129,
      "grad_norm": 80.61068725585938,
      "learning_rate": 2.9118478127849213e-06,
      "loss": 5.0843,
      "step": 18782
    },
    {
      "epoch": 0.7582963262010497,
      "grad_norm": 121.81881713867188,
      "learning_rate": 2.9109255137496094e-06,
      "loss": 4.1717,
      "step": 18783
    },
    {
      "epoch": 0.7583366976180864,
      "grad_norm": 86.82499694824219,
      "learning_rate": 2.910003335920918e-06,
      "loss": 5.1726,
      "step": 18784
    },
    {
      "epoch": 0.7583770690351231,
      "grad_norm": 99.89166259765625,
      "learning_rate": 2.909081279314613e-06,
      "loss": 5.2375,
      "step": 18785
    },
    {
      "epoch": 0.7584174404521599,
      "grad_norm": 125.50452423095703,
      "learning_rate": 2.9081593439464585e-06,
      "loss": 5.7722,
      "step": 18786
    },
    {
      "epoch": 0.7584578118691966,
      "grad_norm": 100.50590515136719,
      "learning_rate": 2.9072375298322196e-06,
      "loss": 3.2835,
      "step": 18787
    },
    {
      "epoch": 0.7584981832862333,
      "grad_norm": 152.71356201171875,
      "learning_rate": 2.906315836987659e-06,
      "loss": 5.4176,
      "step": 18788
    },
    {
      "epoch": 0.75853855470327,
      "grad_norm": 93.06849670410156,
      "learning_rate": 2.9053942654285292e-06,
      "loss": 5.0263,
      "step": 18789
    },
    {
      "epoch": 0.7585789261203069,
      "grad_norm": 122.08878326416016,
      "learning_rate": 2.9044728151705912e-06,
      "loss": 5.2439,
      "step": 18790
    },
    {
      "epoch": 0.7586192975373436,
      "grad_norm": 75.58720397949219,
      "learning_rate": 2.903551486229599e-06,
      "loss": 4.1767,
      "step": 18791
    },
    {
      "epoch": 0.7586596689543803,
      "grad_norm": 77.33699035644531,
      "learning_rate": 2.9026302786213066e-06,
      "loss": 5.5858,
      "step": 18792
    },
    {
      "epoch": 0.758700040371417,
      "grad_norm": 68.71314239501953,
      "learning_rate": 2.9017091923614616e-06,
      "loss": 5.0353,
      "step": 18793
    },
    {
      "epoch": 0.7587404117884538,
      "grad_norm": 86.78785705566406,
      "learning_rate": 2.9007882274658126e-06,
      "loss": 3.995,
      "step": 18794
    },
    {
      "epoch": 0.7587807832054905,
      "grad_norm": 78.79523468017578,
      "learning_rate": 2.8998673839501066e-06,
      "loss": 4.4373,
      "step": 18795
    },
    {
      "epoch": 0.7588211546225273,
      "grad_norm": 121.43204498291016,
      "learning_rate": 2.898946661830092e-06,
      "loss": 4.7739,
      "step": 18796
    },
    {
      "epoch": 0.758861526039564,
      "grad_norm": 91.85486602783203,
      "learning_rate": 2.8980260611215048e-06,
      "loss": 4.3752,
      "step": 18797
    },
    {
      "epoch": 0.7589018974566007,
      "grad_norm": 138.76611328125,
      "learning_rate": 2.8971055818400873e-06,
      "loss": 4.1715,
      "step": 18798
    },
    {
      "epoch": 0.7589422688736375,
      "grad_norm": 111.21858215332031,
      "learning_rate": 2.8961852240015777e-06,
      "loss": 3.858,
      "step": 18799
    },
    {
      "epoch": 0.7589826402906742,
      "grad_norm": 111.75344848632812,
      "learning_rate": 2.895264987621715e-06,
      "loss": 4.7705,
      "step": 18800
    },
    {
      "epoch": 0.7590230117077109,
      "grad_norm": 138.32064819335938,
      "learning_rate": 2.894344872716227e-06,
      "loss": 5.1689,
      "step": 18801
    },
    {
      "epoch": 0.7590633831247476,
      "grad_norm": 94.79840087890625,
      "learning_rate": 2.8934248793008503e-06,
      "loss": 6.0606,
      "step": 18802
    },
    {
      "epoch": 0.7591037545417845,
      "grad_norm": 114.25177001953125,
      "learning_rate": 2.8925050073913117e-06,
      "loss": 5.3856,
      "step": 18803
    },
    {
      "epoch": 0.7591441259588212,
      "grad_norm": 80.87480926513672,
      "learning_rate": 2.891585257003341e-06,
      "loss": 6.0571,
      "step": 18804
    },
    {
      "epoch": 0.7591844973758579,
      "grad_norm": 98.8765869140625,
      "learning_rate": 2.8906656281526624e-06,
      "loss": 5.1444,
      "step": 18805
    },
    {
      "epoch": 0.7592248687928946,
      "grad_norm": 81.99474334716797,
      "learning_rate": 2.889746120855e-06,
      "loss": 6.5871,
      "step": 18806
    },
    {
      "epoch": 0.7592652402099314,
      "grad_norm": 114.6265640258789,
      "learning_rate": 2.8888267351260756e-06,
      "loss": 4.6874,
      "step": 18807
    },
    {
      "epoch": 0.7593056116269681,
      "grad_norm": 90.47736358642578,
      "learning_rate": 2.887907470981612e-06,
      "loss": 6.1007,
      "step": 18808
    },
    {
      "epoch": 0.7593459830440048,
      "grad_norm": 83.06402587890625,
      "learning_rate": 2.886988328437319e-06,
      "loss": 4.1243,
      "step": 18809
    },
    {
      "epoch": 0.7593863544610416,
      "grad_norm": 69.02702331542969,
      "learning_rate": 2.886069307508915e-06,
      "loss": 4.7455,
      "step": 18810
    },
    {
      "epoch": 0.7594267258780784,
      "grad_norm": 88.59939575195312,
      "learning_rate": 2.8851504082121153e-06,
      "loss": 5.9184,
      "step": 18811
    },
    {
      "epoch": 0.7594670972951151,
      "grad_norm": 101.3111343383789,
      "learning_rate": 2.8842316305626306e-06,
      "loss": 3.3309,
      "step": 18812
    },
    {
      "epoch": 0.7595074687121518,
      "grad_norm": 152.4743194580078,
      "learning_rate": 2.8833129745761656e-06,
      "loss": 6.0101,
      "step": 18813
    },
    {
      "epoch": 0.7595478401291885,
      "grad_norm": 69.11695861816406,
      "learning_rate": 2.8823944402684314e-06,
      "loss": 5.9782,
      "step": 18814
    },
    {
      "epoch": 0.7595882115462252,
      "grad_norm": 111.31700897216797,
      "learning_rate": 2.8814760276551303e-06,
      "loss": 4.9359,
      "step": 18815
    },
    {
      "epoch": 0.759628582963262,
      "grad_norm": 92.33055877685547,
      "learning_rate": 2.8805577367519687e-06,
      "loss": 4.3601,
      "step": 18816
    },
    {
      "epoch": 0.7596689543802988,
      "grad_norm": 86.44982147216797,
      "learning_rate": 2.8796395675746437e-06,
      "loss": 4.3678,
      "step": 18817
    },
    {
      "epoch": 0.7597093257973355,
      "grad_norm": 77.71382141113281,
      "learning_rate": 2.878721520138853e-06,
      "loss": 4.212,
      "step": 18818
    },
    {
      "epoch": 0.7597496972143722,
      "grad_norm": 138.51637268066406,
      "learning_rate": 2.8778035944602954e-06,
      "loss": 4.9164,
      "step": 18819
    },
    {
      "epoch": 0.759790068631409,
      "grad_norm": 89.50785827636719,
      "learning_rate": 2.8768857905546688e-06,
      "loss": 3.9852,
      "step": 18820
    },
    {
      "epoch": 0.7598304400484457,
      "grad_norm": 220.2985382080078,
      "learning_rate": 2.875968108437657e-06,
      "loss": 2.8701,
      "step": 18821
    },
    {
      "epoch": 0.7598708114654824,
      "grad_norm": 104.37639617919922,
      "learning_rate": 2.8750505481249556e-06,
      "loss": 5.5598,
      "step": 18822
    },
    {
      "epoch": 0.7599111828825191,
      "grad_norm": 94.38158416748047,
      "learning_rate": 2.874133109632251e-06,
      "loss": 5.3632,
      "step": 18823
    },
    {
      "epoch": 0.759951554299556,
      "grad_norm": 113.67283630371094,
      "learning_rate": 2.8732157929752314e-06,
      "loss": 5.725,
      "step": 18824
    },
    {
      "epoch": 0.7599919257165927,
      "grad_norm": 134.78590393066406,
      "learning_rate": 2.8722985981695783e-06,
      "loss": 4.5075,
      "step": 18825
    },
    {
      "epoch": 0.7600322971336294,
      "grad_norm": 89.70169067382812,
      "learning_rate": 2.871381525230976e-06,
      "loss": 4.0955,
      "step": 18826
    },
    {
      "epoch": 0.7600726685506661,
      "grad_norm": 79.73286437988281,
      "learning_rate": 2.8704645741751024e-06,
      "loss": 4.6023,
      "step": 18827
    },
    {
      "epoch": 0.7601130399677029,
      "grad_norm": 180.68701171875,
      "learning_rate": 2.869547745017639e-06,
      "loss": 3.7652,
      "step": 18828
    },
    {
      "epoch": 0.7601534113847396,
      "grad_norm": 133.54351806640625,
      "learning_rate": 2.8686310377742556e-06,
      "loss": 5.1855,
      "step": 18829
    },
    {
      "epoch": 0.7601937828017763,
      "grad_norm": 99.32160186767578,
      "learning_rate": 2.8677144524606284e-06,
      "loss": 4.0743,
      "step": 18830
    },
    {
      "epoch": 0.7602341542188131,
      "grad_norm": 83.8301773071289,
      "learning_rate": 2.866797989092429e-06,
      "loss": 5.4211,
      "step": 18831
    },
    {
      "epoch": 0.7602745256358499,
      "grad_norm": 136.83441162109375,
      "learning_rate": 2.8658816476853293e-06,
      "loss": 4.0261,
      "step": 18832
    },
    {
      "epoch": 0.7603148970528866,
      "grad_norm": 133.3417205810547,
      "learning_rate": 2.864965428254991e-06,
      "loss": 3.153,
      "step": 18833
    },
    {
      "epoch": 0.7603552684699233,
      "grad_norm": 98.18804931640625,
      "learning_rate": 2.8640493308170836e-06,
      "loss": 6.3869,
      "step": 18834
    },
    {
      "epoch": 0.76039563988696,
      "grad_norm": 143.67034912109375,
      "learning_rate": 2.8631333553872674e-06,
      "loss": 6.0263,
      "step": 18835
    },
    {
      "epoch": 0.7604360113039967,
      "grad_norm": 95.07557678222656,
      "learning_rate": 2.8622175019812095e-06,
      "loss": 4.0952,
      "step": 18836
    },
    {
      "epoch": 0.7604763827210335,
      "grad_norm": 72.56642150878906,
      "learning_rate": 2.861301770614561e-06,
      "loss": 5.2631,
      "step": 18837
    },
    {
      "epoch": 0.7605167541380703,
      "grad_norm": 94.63641357421875,
      "learning_rate": 2.8603861613029814e-06,
      "loss": 5.4117,
      "step": 18838
    },
    {
      "epoch": 0.760557125555107,
      "grad_norm": 100.32952117919922,
      "learning_rate": 2.859470674062127e-06,
      "loss": 4.8902,
      "step": 18839
    },
    {
      "epoch": 0.7605974969721437,
      "grad_norm": 62.07404327392578,
      "learning_rate": 2.8585553089076523e-06,
      "loss": 3.4677,
      "step": 18840
    },
    {
      "epoch": 0.7606378683891805,
      "grad_norm": 68.56147766113281,
      "learning_rate": 2.857640065855203e-06,
      "loss": 4.5785,
      "step": 18841
    },
    {
      "epoch": 0.7606782398062172,
      "grad_norm": 95.69542694091797,
      "learning_rate": 2.856724944920429e-06,
      "loss": 6.0616,
      "step": 18842
    },
    {
      "epoch": 0.7607186112232539,
      "grad_norm": 121.80833435058594,
      "learning_rate": 2.855809946118978e-06,
      "loss": 4.4314,
      "step": 18843
    },
    {
      "epoch": 0.7607589826402906,
      "grad_norm": 154.8052978515625,
      "learning_rate": 2.8548950694664947e-06,
      "loss": 4.8312,
      "step": 18844
    },
    {
      "epoch": 0.7607993540573275,
      "grad_norm": 119.51502227783203,
      "learning_rate": 2.8539803149786196e-06,
      "loss": 4.2732,
      "step": 18845
    },
    {
      "epoch": 0.7608397254743642,
      "grad_norm": 156.56005859375,
      "learning_rate": 2.8530656826709946e-06,
      "loss": 5.7497,
      "step": 18846
    },
    {
      "epoch": 0.7608800968914009,
      "grad_norm": 66.59431457519531,
      "learning_rate": 2.8521511725592576e-06,
      "loss": 4.6021,
      "step": 18847
    },
    {
      "epoch": 0.7609204683084376,
      "grad_norm": 103.53659057617188,
      "learning_rate": 2.851236784659046e-06,
      "loss": 4.1004,
      "step": 18848
    },
    {
      "epoch": 0.7609608397254743,
      "grad_norm": 93.06250762939453,
      "learning_rate": 2.8503225189859898e-06,
      "loss": 4.0272,
      "step": 18849
    },
    {
      "epoch": 0.7610012111425111,
      "grad_norm": 87.07349395751953,
      "learning_rate": 2.8494083755557224e-06,
      "loss": 4.0086,
      "step": 18850
    },
    {
      "epoch": 0.7610415825595478,
      "grad_norm": 81.75727844238281,
      "learning_rate": 2.848494354383874e-06,
      "loss": 4.4958,
      "step": 18851
    },
    {
      "epoch": 0.7610819539765846,
      "grad_norm": 113.28118896484375,
      "learning_rate": 2.8475804554860754e-06,
      "loss": 6.5454,
      "step": 18852
    },
    {
      "epoch": 0.7611223253936213,
      "grad_norm": 97.5205078125,
      "learning_rate": 2.8466666788779452e-06,
      "loss": 3.9632,
      "step": 18853
    },
    {
      "epoch": 0.7611626968106581,
      "grad_norm": 120.79683685302734,
      "learning_rate": 2.845753024575111e-06,
      "loss": 4.3465,
      "step": 18854
    },
    {
      "epoch": 0.7612030682276948,
      "grad_norm": 79.7225570678711,
      "learning_rate": 2.8448394925931945e-06,
      "loss": 6.002,
      "step": 18855
    },
    {
      "epoch": 0.7612434396447315,
      "grad_norm": 86.0674057006836,
      "learning_rate": 2.8439260829478154e-06,
      "loss": 5.8964,
      "step": 18856
    },
    {
      "epoch": 0.7612838110617682,
      "grad_norm": 86.65770721435547,
      "learning_rate": 2.8430127956545884e-06,
      "loss": 5.4214,
      "step": 18857
    },
    {
      "epoch": 0.761324182478805,
      "grad_norm": 110.52310943603516,
      "learning_rate": 2.8420996307291293e-06,
      "loss": 4.9021,
      "step": 18858
    },
    {
      "epoch": 0.7613645538958418,
      "grad_norm": 339.2856140136719,
      "learning_rate": 2.8411865881870516e-06,
      "loss": 5.1806,
      "step": 18859
    },
    {
      "epoch": 0.7614049253128785,
      "grad_norm": 145.66061401367188,
      "learning_rate": 2.840273668043969e-06,
      "loss": 4.6577,
      "step": 18860
    },
    {
      "epoch": 0.7614452967299152,
      "grad_norm": 86.92364501953125,
      "learning_rate": 2.8393608703154842e-06,
      "loss": 5.2469,
      "step": 18861
    },
    {
      "epoch": 0.761485668146952,
      "grad_norm": 86.6092758178711,
      "learning_rate": 2.8384481950172073e-06,
      "loss": 4.3832,
      "step": 18862
    },
    {
      "epoch": 0.7615260395639887,
      "grad_norm": 137.7977752685547,
      "learning_rate": 2.837535642164744e-06,
      "loss": 5.6609,
      "step": 18863
    },
    {
      "epoch": 0.7615664109810254,
      "grad_norm": 106.58269500732422,
      "learning_rate": 2.8366232117736936e-06,
      "loss": 6.1621,
      "step": 18864
    },
    {
      "epoch": 0.7616067823980621,
      "grad_norm": 81.8154067993164,
      "learning_rate": 2.8357109038596597e-06,
      "loss": 4.7209,
      "step": 18865
    },
    {
      "epoch": 0.761647153815099,
      "grad_norm": 80.6703872680664,
      "learning_rate": 2.8347987184382398e-06,
      "loss": 4.3963,
      "step": 18866
    },
    {
      "epoch": 0.7616875252321357,
      "grad_norm": 92.07306671142578,
      "learning_rate": 2.83388665552503e-06,
      "loss": 4.1579,
      "step": 18867
    },
    {
      "epoch": 0.7617278966491724,
      "grad_norm": 87.39704132080078,
      "learning_rate": 2.832974715135627e-06,
      "loss": 4.6044,
      "step": 18868
    },
    {
      "epoch": 0.7617682680662091,
      "grad_norm": 88.73017120361328,
      "learning_rate": 2.832062897285617e-06,
      "loss": 3.4983,
      "step": 18869
    },
    {
      "epoch": 0.7618086394832458,
      "grad_norm": 316.3397521972656,
      "learning_rate": 2.8311512019905928e-06,
      "loss": 4.9482,
      "step": 18870
    },
    {
      "epoch": 0.7618490109002826,
      "grad_norm": 95.8139877319336,
      "learning_rate": 2.8302396292661438e-06,
      "loss": 6.051,
      "step": 18871
    },
    {
      "epoch": 0.7618893823173193,
      "grad_norm": 118.78770446777344,
      "learning_rate": 2.829328179127857e-06,
      "loss": 3.9901,
      "step": 18872
    },
    {
      "epoch": 0.7619297537343561,
      "grad_norm": 103.7037582397461,
      "learning_rate": 2.82841685159131e-06,
      "loss": 5.8332,
      "step": 18873
    },
    {
      "epoch": 0.7619701251513928,
      "grad_norm": 128.54757690429688,
      "learning_rate": 2.8275056466720884e-06,
      "loss": 5.2873,
      "step": 18874
    },
    {
      "epoch": 0.7620104965684296,
      "grad_norm": 118.36895751953125,
      "learning_rate": 2.8265945643857705e-06,
      "loss": 4.9338,
      "step": 18875
    },
    {
      "epoch": 0.7620508679854663,
      "grad_norm": 72.01324462890625,
      "learning_rate": 2.8256836047479387e-06,
      "loss": 4.6419,
      "step": 18876
    },
    {
      "epoch": 0.762091239402503,
      "grad_norm": 111.45030975341797,
      "learning_rate": 2.824772767774161e-06,
      "loss": 4.4331,
      "step": 18877
    },
    {
      "epoch": 0.7621316108195397,
      "grad_norm": 78.90939331054688,
      "learning_rate": 2.8238620534800132e-06,
      "loss": 4.9786,
      "step": 18878
    },
    {
      "epoch": 0.7621719822365765,
      "grad_norm": 138.21652221679688,
      "learning_rate": 2.8229514618810662e-06,
      "loss": 4.8024,
      "step": 18879
    },
    {
      "epoch": 0.7622123536536133,
      "grad_norm": 70.95770263671875,
      "learning_rate": 2.8220409929928906e-06,
      "loss": 4.1867,
      "step": 18880
    },
    {
      "epoch": 0.76225272507065,
      "grad_norm": 82.9676513671875,
      "learning_rate": 2.821130646831052e-06,
      "loss": 3.9,
      "step": 18881
    },
    {
      "epoch": 0.7622930964876867,
      "grad_norm": 93.36544799804688,
      "learning_rate": 2.820220423411114e-06,
      "loss": 4.7984,
      "step": 18882
    },
    {
      "epoch": 0.7623334679047234,
      "grad_norm": 145.1737823486328,
      "learning_rate": 2.819310322748643e-06,
      "loss": 4.9413,
      "step": 18883
    },
    {
      "epoch": 0.7623738393217602,
      "grad_norm": 114.83168029785156,
      "learning_rate": 2.818400344859199e-06,
      "loss": 5.16,
      "step": 18884
    },
    {
      "epoch": 0.7624142107387969,
      "grad_norm": 91.9611587524414,
      "learning_rate": 2.8174904897583366e-06,
      "loss": 4.2581,
      "step": 18885
    },
    {
      "epoch": 0.7624545821558336,
      "grad_norm": 108.9251480102539,
      "learning_rate": 2.8165807574616146e-06,
      "loss": 5.8837,
      "step": 18886
    },
    {
      "epoch": 0.7624949535728704,
      "grad_norm": 92.51874542236328,
      "learning_rate": 2.8156711479845856e-06,
      "loss": 4.1531,
      "step": 18887
    },
    {
      "epoch": 0.7625353249899072,
      "grad_norm": 82.02157592773438,
      "learning_rate": 2.8147616613428077e-06,
      "loss": 4.7435,
      "step": 18888
    },
    {
      "epoch": 0.7625756964069439,
      "grad_norm": 116.52408599853516,
      "learning_rate": 2.8138522975518233e-06,
      "loss": 4.4682,
      "step": 18889
    },
    {
      "epoch": 0.7626160678239806,
      "grad_norm": 105.85101318359375,
      "learning_rate": 2.8129430566271844e-06,
      "loss": 3.6582,
      "step": 18890
    },
    {
      "epoch": 0.7626564392410173,
      "grad_norm": 83.6472396850586,
      "learning_rate": 2.8120339385844354e-06,
      "loss": 5.167,
      "step": 18891
    },
    {
      "epoch": 0.7626968106580541,
      "grad_norm": 62.53126907348633,
      "learning_rate": 2.811124943439124e-06,
      "loss": 4.0908,
      "step": 18892
    },
    {
      "epoch": 0.7627371820750908,
      "grad_norm": 92.69508361816406,
      "learning_rate": 2.810216071206786e-06,
      "loss": 3.7371,
      "step": 18893
    },
    {
      "epoch": 0.7627775534921276,
      "grad_norm": 103.1263198852539,
      "learning_rate": 2.8093073219029644e-06,
      "loss": 4.2117,
      "step": 18894
    },
    {
      "epoch": 0.7628179249091643,
      "grad_norm": 93.62918853759766,
      "learning_rate": 2.808398695543195e-06,
      "loss": 3.1773,
      "step": 18895
    },
    {
      "epoch": 0.7628582963262011,
      "grad_norm": 83.45883178710938,
      "learning_rate": 2.8074901921430188e-06,
      "loss": 4.9965,
      "step": 18896
    },
    {
      "epoch": 0.7628986677432378,
      "grad_norm": 88.8759536743164,
      "learning_rate": 2.80658181171796e-06,
      "loss": 2.1662,
      "step": 18897
    },
    {
      "epoch": 0.7629390391602745,
      "grad_norm": 96.700927734375,
      "learning_rate": 2.805673554283556e-06,
      "loss": 3.5,
      "step": 18898
    },
    {
      "epoch": 0.7629794105773112,
      "grad_norm": 173.3385009765625,
      "learning_rate": 2.8047654198553333e-06,
      "loss": 4.1653,
      "step": 18899
    },
    {
      "epoch": 0.7630197819943481,
      "grad_norm": 117.14579772949219,
      "learning_rate": 2.803857408448819e-06,
      "loss": 3.9524,
      "step": 18900
    },
    {
      "epoch": 0.7630601534113848,
      "grad_norm": 152.0054168701172,
      "learning_rate": 2.8029495200795397e-06,
      "loss": 5.6018,
      "step": 18901
    },
    {
      "epoch": 0.7631005248284215,
      "grad_norm": 101.45462036132812,
      "learning_rate": 2.8020417547630175e-06,
      "loss": 4.6492,
      "step": 18902
    },
    {
      "epoch": 0.7631408962454582,
      "grad_norm": 103.75619506835938,
      "learning_rate": 2.8011341125147727e-06,
      "loss": 4.3808,
      "step": 18903
    },
    {
      "epoch": 0.7631812676624949,
      "grad_norm": 82.79342651367188,
      "learning_rate": 2.800226593350327e-06,
      "loss": 5.852,
      "step": 18904
    },
    {
      "epoch": 0.7632216390795317,
      "grad_norm": 102.37897491455078,
      "learning_rate": 2.7993191972851906e-06,
      "loss": 6.5757,
      "step": 18905
    },
    {
      "epoch": 0.7632620104965684,
      "grad_norm": 72.44282531738281,
      "learning_rate": 2.798411924334882e-06,
      "loss": 5.5828,
      "step": 18906
    },
    {
      "epoch": 0.7633023819136052,
      "grad_norm": 96.52583312988281,
      "learning_rate": 2.797504774514912e-06,
      "loss": 4.2471,
      "step": 18907
    },
    {
      "epoch": 0.7633427533306419,
      "grad_norm": 80.09529876708984,
      "learning_rate": 2.7965977478407937e-06,
      "loss": 4.4742,
      "step": 18908
    },
    {
      "epoch": 0.7633831247476787,
      "grad_norm": 100.35897827148438,
      "learning_rate": 2.7956908443280305e-06,
      "loss": 4.8981,
      "step": 18909
    },
    {
      "epoch": 0.7634234961647154,
      "grad_norm": 107.13512420654297,
      "learning_rate": 2.7947840639921308e-06,
      "loss": 5.9633,
      "step": 18910
    },
    {
      "epoch": 0.7634638675817521,
      "grad_norm": 92.6975326538086,
      "learning_rate": 2.7938774068485975e-06,
      "loss": 4.0908,
      "step": 18911
    },
    {
      "epoch": 0.7635042389987888,
      "grad_norm": 161.93006896972656,
      "learning_rate": 2.7929708729129368e-06,
      "loss": 4.6614,
      "step": 18912
    },
    {
      "epoch": 0.7635446104158256,
      "grad_norm": 84.92190551757812,
      "learning_rate": 2.792064462200641e-06,
      "loss": 5.4121,
      "step": 18913
    },
    {
      "epoch": 0.7635849818328624,
      "grad_norm": 70.31819152832031,
      "learning_rate": 2.791158174727211e-06,
      "loss": 5.1897,
      "step": 18914
    },
    {
      "epoch": 0.7636253532498991,
      "grad_norm": 123.05732727050781,
      "learning_rate": 2.7902520105081422e-06,
      "loss": 6.9296,
      "step": 18915
    },
    {
      "epoch": 0.7636657246669358,
      "grad_norm": 84.83163452148438,
      "learning_rate": 2.789345969558932e-06,
      "loss": 5.5146,
      "step": 18916
    },
    {
      "epoch": 0.7637060960839726,
      "grad_norm": 84.05602264404297,
      "learning_rate": 2.788440051895064e-06,
      "loss": 3.9084,
      "step": 18917
    },
    {
      "epoch": 0.7637464675010093,
      "grad_norm": 357.2261962890625,
      "learning_rate": 2.7875342575320308e-06,
      "loss": 5.0696,
      "step": 18918
    },
    {
      "epoch": 0.763786838918046,
      "grad_norm": 109.0792465209961,
      "learning_rate": 2.7866285864853205e-06,
      "loss": 4.8793,
      "step": 18919
    },
    {
      "epoch": 0.7638272103350827,
      "grad_norm": 86.12458801269531,
      "learning_rate": 2.785723038770416e-06,
      "loss": 4.9445,
      "step": 18920
    },
    {
      "epoch": 0.7638675817521196,
      "grad_norm": 97.9971694946289,
      "learning_rate": 2.7848176144028003e-06,
      "loss": 5.9067,
      "step": 18921
    },
    {
      "epoch": 0.7639079531691563,
      "grad_norm": 95.44083404541016,
      "learning_rate": 2.7839123133979562e-06,
      "loss": 4.5829,
      "step": 18922
    },
    {
      "epoch": 0.763948324586193,
      "grad_norm": 87.32354736328125,
      "learning_rate": 2.7830071357713595e-06,
      "loss": 4.0845,
      "step": 18923
    },
    {
      "epoch": 0.7639886960032297,
      "grad_norm": 184.50491333007812,
      "learning_rate": 2.7821020815384926e-06,
      "loss": 4.5537,
      "step": 18924
    },
    {
      "epoch": 0.7640290674202664,
      "grad_norm": 82.669677734375,
      "learning_rate": 2.7811971507148206e-06,
      "loss": 4.832,
      "step": 18925
    },
    {
      "epoch": 0.7640694388373032,
      "grad_norm": 122.14314270019531,
      "learning_rate": 2.78029234331582e-06,
      "loss": 5.1168,
      "step": 18926
    },
    {
      "epoch": 0.7641098102543399,
      "grad_norm": 94.97164154052734,
      "learning_rate": 2.7793876593569624e-06,
      "loss": 5.3965,
      "step": 18927
    },
    {
      "epoch": 0.7641501816713767,
      "grad_norm": 84.31423950195312,
      "learning_rate": 2.778483098853717e-06,
      "loss": 3.9388,
      "step": 18928
    },
    {
      "epoch": 0.7641905530884134,
      "grad_norm": 89.6620101928711,
      "learning_rate": 2.777578661821544e-06,
      "loss": 5.186,
      "step": 18929
    },
    {
      "epoch": 0.7642309245054502,
      "grad_norm": 82.39546203613281,
      "learning_rate": 2.7766743482759097e-06,
      "loss": 5.4334,
      "step": 18930
    },
    {
      "epoch": 0.7642712959224869,
      "grad_norm": 290.73565673828125,
      "learning_rate": 2.7757701582322773e-06,
      "loss": 4.5486,
      "step": 18931
    },
    {
      "epoch": 0.7643116673395236,
      "grad_norm": 131.20614624023438,
      "learning_rate": 2.7748660917061077e-06,
      "loss": 5.6443,
      "step": 18932
    },
    {
      "epoch": 0.7643520387565603,
      "grad_norm": 118.46524810791016,
      "learning_rate": 2.773962148712853e-06,
      "loss": 5.4403,
      "step": 18933
    },
    {
      "epoch": 0.764392410173597,
      "grad_norm": 76.16297912597656,
      "learning_rate": 2.7730583292679725e-06,
      "loss": 4.2859,
      "step": 18934
    },
    {
      "epoch": 0.7644327815906339,
      "grad_norm": 125.59685516357422,
      "learning_rate": 2.7721546333869174e-06,
      "loss": 5.5482,
      "step": 18935
    },
    {
      "epoch": 0.7644731530076706,
      "grad_norm": 103.55188751220703,
      "learning_rate": 2.771251061085144e-06,
      "loss": 6.184,
      "step": 18936
    },
    {
      "epoch": 0.7645135244247073,
      "grad_norm": 101.0804443359375,
      "learning_rate": 2.770347612378094e-06,
      "loss": 4.2497,
      "step": 18937
    },
    {
      "epoch": 0.764553895841744,
      "grad_norm": 70.29032135009766,
      "learning_rate": 2.7694442872812153e-06,
      "loss": 3.9904,
      "step": 18938
    },
    {
      "epoch": 0.7645942672587808,
      "grad_norm": 105.77493286132812,
      "learning_rate": 2.768541085809957e-06,
      "loss": 4.4578,
      "step": 18939
    },
    {
      "epoch": 0.7646346386758175,
      "grad_norm": 121.84812927246094,
      "learning_rate": 2.7676380079797582e-06,
      "loss": 5.0005,
      "step": 18940
    },
    {
      "epoch": 0.7646750100928542,
      "grad_norm": 245.24452209472656,
      "learning_rate": 2.7667350538060612e-06,
      "loss": 4.1354,
      "step": 18941
    },
    {
      "epoch": 0.764715381509891,
      "grad_norm": 113.0905532836914,
      "learning_rate": 2.7658322233043046e-06,
      "loss": 5.453,
      "step": 18942
    },
    {
      "epoch": 0.7647557529269278,
      "grad_norm": 110.14747619628906,
      "learning_rate": 2.7649295164899225e-06,
      "loss": 4.8794,
      "step": 18943
    },
    {
      "epoch": 0.7647961243439645,
      "grad_norm": 107.3844223022461,
      "learning_rate": 2.7640269333783543e-06,
      "loss": 5.4733,
      "step": 18944
    },
    {
      "epoch": 0.7648364957610012,
      "grad_norm": 88.23529052734375,
      "learning_rate": 2.763124473985026e-06,
      "loss": 6.4755,
      "step": 18945
    },
    {
      "epoch": 0.7648768671780379,
      "grad_norm": 84.5820083618164,
      "learning_rate": 2.7622221383253687e-06,
      "loss": 3.3852,
      "step": 18946
    },
    {
      "epoch": 0.7649172385950747,
      "grad_norm": 128.0131072998047,
      "learning_rate": 2.7613199264148117e-06,
      "loss": 4.5088,
      "step": 18947
    },
    {
      "epoch": 0.7649576100121114,
      "grad_norm": 102.5512924194336,
      "learning_rate": 2.760417838268784e-06,
      "loss": 5.1105,
      "step": 18948
    },
    {
      "epoch": 0.7649979814291482,
      "grad_norm": 257.9134826660156,
      "learning_rate": 2.759515873902703e-06,
      "loss": 4.942,
      "step": 18949
    },
    {
      "epoch": 0.7650383528461849,
      "grad_norm": 100.01181030273438,
      "learning_rate": 2.758614033331991e-06,
      "loss": 3.9937,
      "step": 18950
    },
    {
      "epoch": 0.7650787242632217,
      "grad_norm": 78.57788848876953,
      "learning_rate": 2.757712316572071e-06,
      "loss": 4.4821,
      "step": 18951
    },
    {
      "epoch": 0.7651190956802584,
      "grad_norm": 92.08595275878906,
      "learning_rate": 2.75681072363836e-06,
      "loss": 3.3554,
      "step": 18952
    },
    {
      "epoch": 0.7651594670972951,
      "grad_norm": 69.34210205078125,
      "learning_rate": 2.7559092545462694e-06,
      "loss": 5.0495,
      "step": 18953
    },
    {
      "epoch": 0.7651998385143318,
      "grad_norm": 151.55958557128906,
      "learning_rate": 2.755007909311215e-06,
      "loss": 4.3319,
      "step": 18954
    },
    {
      "epoch": 0.7652402099313685,
      "grad_norm": 157.57949829101562,
      "learning_rate": 2.7541066879486056e-06,
      "loss": 4.8181,
      "step": 18955
    },
    {
      "epoch": 0.7652805813484054,
      "grad_norm": 105.96701049804688,
      "learning_rate": 2.753205590473855e-06,
      "loss": 4.9372,
      "step": 18956
    },
    {
      "epoch": 0.7653209527654421,
      "grad_norm": 110.51189422607422,
      "learning_rate": 2.752304616902364e-06,
      "loss": 5.3565,
      "step": 18957
    },
    {
      "epoch": 0.7653613241824788,
      "grad_norm": 72.07836151123047,
      "learning_rate": 2.7514037672495387e-06,
      "loss": 4.6544,
      "step": 18958
    },
    {
      "epoch": 0.7654016955995155,
      "grad_norm": 115.23313903808594,
      "learning_rate": 2.750503041530783e-06,
      "loss": 7.0603,
      "step": 18959
    },
    {
      "epoch": 0.7654420670165523,
      "grad_norm": 105.3818359375,
      "learning_rate": 2.749602439761496e-06,
      "loss": 5.0304,
      "step": 18960
    },
    {
      "epoch": 0.765482438433589,
      "grad_norm": 100.71315002441406,
      "learning_rate": 2.7487019619570764e-06,
      "loss": 4.2406,
      "step": 18961
    },
    {
      "epoch": 0.7655228098506257,
      "grad_norm": 103.245849609375,
      "learning_rate": 2.7478016081329195e-06,
      "loss": 6.0474,
      "step": 18962
    },
    {
      "epoch": 0.7655631812676625,
      "grad_norm": 93.17992401123047,
      "learning_rate": 2.746901378304422e-06,
      "loss": 5.1372,
      "step": 18963
    },
    {
      "epoch": 0.7656035526846993,
      "grad_norm": 79.89063262939453,
      "learning_rate": 2.7460012724869745e-06,
      "loss": 5.6986,
      "step": 18964
    },
    {
      "epoch": 0.765643924101736,
      "grad_norm": 80.67448425292969,
      "learning_rate": 2.7451012906959638e-06,
      "loss": 5.6046,
      "step": 18965
    },
    {
      "epoch": 0.7656842955187727,
      "grad_norm": 78.18607330322266,
      "learning_rate": 2.7442014329467804e-06,
      "loss": 4.412,
      "step": 18966
    },
    {
      "epoch": 0.7657246669358094,
      "grad_norm": 66.83345031738281,
      "learning_rate": 2.743301699254808e-06,
      "loss": 5.0675,
      "step": 18967
    },
    {
      "epoch": 0.7657650383528462,
      "grad_norm": 96.0954818725586,
      "learning_rate": 2.742402089635434e-06,
      "loss": 6.671,
      "step": 18968
    },
    {
      "epoch": 0.7658054097698829,
      "grad_norm": 77.00531768798828,
      "learning_rate": 2.7415026041040348e-06,
      "loss": 3.857,
      "step": 18969
    },
    {
      "epoch": 0.7658457811869197,
      "grad_norm": 121.99005126953125,
      "learning_rate": 2.74060324267599e-06,
      "loss": 5.7201,
      "step": 18970
    },
    {
      "epoch": 0.7658861526039564,
      "grad_norm": 97.33634948730469,
      "learning_rate": 2.739704005366678e-06,
      "loss": 6.3419,
      "step": 18971
    },
    {
      "epoch": 0.7659265240209931,
      "grad_norm": 85.63719177246094,
      "learning_rate": 2.7388048921914768e-06,
      "loss": 5.5947,
      "step": 18972
    },
    {
      "epoch": 0.7659668954380299,
      "grad_norm": 69.97173309326172,
      "learning_rate": 2.737905903165753e-06,
      "loss": 4.8117,
      "step": 18973
    },
    {
      "epoch": 0.7660072668550666,
      "grad_norm": 95.92130279541016,
      "learning_rate": 2.7370070383048795e-06,
      "loss": 5.0049,
      "step": 18974
    },
    {
      "epoch": 0.7660476382721033,
      "grad_norm": 58.357872009277344,
      "learning_rate": 2.736108297624226e-06,
      "loss": 3.0218,
      "step": 18975
    },
    {
      "epoch": 0.76608800968914,
      "grad_norm": 127.64167785644531,
      "learning_rate": 2.7352096811391603e-06,
      "loss": 4.2737,
      "step": 18976
    },
    {
      "epoch": 0.7661283811061769,
      "grad_norm": 76.30264282226562,
      "learning_rate": 2.734311188865042e-06,
      "loss": 3.946,
      "step": 18977
    },
    {
      "epoch": 0.7661687525232136,
      "grad_norm": 95.39183044433594,
      "learning_rate": 2.7334128208172363e-06,
      "loss": 4.7685,
      "step": 18978
    },
    {
      "epoch": 0.7662091239402503,
      "grad_norm": 73.23946380615234,
      "learning_rate": 2.7325145770111016e-06,
      "loss": 4.0737,
      "step": 18979
    },
    {
      "epoch": 0.766249495357287,
      "grad_norm": 92.82557678222656,
      "learning_rate": 2.7316164574619973e-06,
      "loss": 4.6665,
      "step": 18980
    },
    {
      "epoch": 0.7662898667743238,
      "grad_norm": 76.7312240600586,
      "learning_rate": 2.730718462185279e-06,
      "loss": 4.5464,
      "step": 18981
    },
    {
      "epoch": 0.7663302381913605,
      "grad_norm": 109.36544036865234,
      "learning_rate": 2.7298205911962994e-06,
      "loss": 5.5939,
      "step": 18982
    },
    {
      "epoch": 0.7663706096083972,
      "grad_norm": 88.53455352783203,
      "learning_rate": 2.728922844510411e-06,
      "loss": 6.1018,
      "step": 18983
    },
    {
      "epoch": 0.766410981025434,
      "grad_norm": 91.7300796508789,
      "learning_rate": 2.728025222142966e-06,
      "loss": 5.6121,
      "step": 18984
    },
    {
      "epoch": 0.7664513524424708,
      "grad_norm": 103.99272155761719,
      "learning_rate": 2.7271277241093053e-06,
      "loss": 6.4419,
      "step": 18985
    },
    {
      "epoch": 0.7664917238595075,
      "grad_norm": 126.14393615722656,
      "learning_rate": 2.726230350424777e-06,
      "loss": 5.9985,
      "step": 18986
    },
    {
      "epoch": 0.7665320952765442,
      "grad_norm": 129.54576110839844,
      "learning_rate": 2.7253331011047234e-06,
      "loss": 6.2197,
      "step": 18987
    },
    {
      "epoch": 0.7665724666935809,
      "grad_norm": 163.56185913085938,
      "learning_rate": 2.72443597616449e-06,
      "loss": 6.5618,
      "step": 18988
    },
    {
      "epoch": 0.7666128381106176,
      "grad_norm": 76.78201293945312,
      "learning_rate": 2.7235389756194097e-06,
      "loss": 4.3706,
      "step": 18989
    },
    {
      "epoch": 0.7666532095276544,
      "grad_norm": 91.16673278808594,
      "learning_rate": 2.72264209948482e-06,
      "loss": 4.5069,
      "step": 18990
    },
    {
      "epoch": 0.7666935809446912,
      "grad_norm": 88.76620483398438,
      "learning_rate": 2.721745347776057e-06,
      "loss": 5.6104,
      "step": 18991
    },
    {
      "epoch": 0.7667339523617279,
      "grad_norm": 132.06582641601562,
      "learning_rate": 2.7208487205084556e-06,
      "loss": 4.222,
      "step": 18992
    },
    {
      "epoch": 0.7667743237787646,
      "grad_norm": 102.79657745361328,
      "learning_rate": 2.7199522176973394e-06,
      "loss": 5.4679,
      "step": 18993
    },
    {
      "epoch": 0.7668146951958014,
      "grad_norm": 103.45552825927734,
      "learning_rate": 2.7190558393580403e-06,
      "loss": 6.6481,
      "step": 18994
    },
    {
      "epoch": 0.7668550666128381,
      "grad_norm": 89.55865478515625,
      "learning_rate": 2.7181595855058853e-06,
      "loss": 4.704,
      "step": 18995
    },
    {
      "epoch": 0.7668954380298748,
      "grad_norm": 93.30480194091797,
      "learning_rate": 2.717263456156198e-06,
      "loss": 4.2759,
      "step": 18996
    },
    {
      "epoch": 0.7669358094469115,
      "grad_norm": 99.38739776611328,
      "learning_rate": 2.7163674513242977e-06,
      "loss": 5.7649,
      "step": 18997
    },
    {
      "epoch": 0.7669761808639484,
      "grad_norm": 203.9826202392578,
      "learning_rate": 2.7154715710255054e-06,
      "loss": 5.2597,
      "step": 18998
    },
    {
      "epoch": 0.7670165522809851,
      "grad_norm": 95.18502044677734,
      "learning_rate": 2.7145758152751378e-06,
      "loss": 4.9444,
      "step": 18999
    },
    {
      "epoch": 0.7670569236980218,
      "grad_norm": 91.10872650146484,
      "learning_rate": 2.713680184088512e-06,
      "loss": 5.9642,
      "step": 19000
    },
    {
      "epoch": 0.7670972951150585,
      "grad_norm": 88.08240509033203,
      "learning_rate": 2.7127846774809397e-06,
      "loss": 4.4314,
      "step": 19001
    },
    {
      "epoch": 0.7671376665320953,
      "grad_norm": 84.41671752929688,
      "learning_rate": 2.7118892954677333e-06,
      "loss": 3.7914,
      "step": 19002
    },
    {
      "epoch": 0.767178037949132,
      "grad_norm": 73.88875579833984,
      "learning_rate": 2.7109940380641997e-06,
      "loss": 5.4124,
      "step": 19003
    },
    {
      "epoch": 0.7672184093661688,
      "grad_norm": 87.70793914794922,
      "learning_rate": 2.7100989052856486e-06,
      "loss": 4.745,
      "step": 19004
    },
    {
      "epoch": 0.7672587807832055,
      "grad_norm": 109.94369506835938,
      "learning_rate": 2.7092038971473845e-06,
      "loss": 5.2245,
      "step": 19005
    },
    {
      "epoch": 0.7672991522002423,
      "grad_norm": 167.97055053710938,
      "learning_rate": 2.708309013664706e-06,
      "loss": 5.2983,
      "step": 19006
    },
    {
      "epoch": 0.767339523617279,
      "grad_norm": 96.47344207763672,
      "learning_rate": 2.707414254852917e-06,
      "loss": 4.525,
      "step": 19007
    },
    {
      "epoch": 0.7673798950343157,
      "grad_norm": 123.60173034667969,
      "learning_rate": 2.7065196207273136e-06,
      "loss": 5.8031,
      "step": 19008
    },
    {
      "epoch": 0.7674202664513524,
      "grad_norm": 116.81089782714844,
      "learning_rate": 2.7056251113031973e-06,
      "loss": 4.9493,
      "step": 19009
    },
    {
      "epoch": 0.7674606378683891,
      "grad_norm": 89.56075286865234,
      "learning_rate": 2.7047307265958546e-06,
      "loss": 4.9245,
      "step": 19010
    },
    {
      "epoch": 0.767501009285426,
      "grad_norm": 89.4737548828125,
      "learning_rate": 2.7038364666205797e-06,
      "loss": 4.7885,
      "step": 19011
    },
    {
      "epoch": 0.7675413807024627,
      "grad_norm": 78.79157257080078,
      "learning_rate": 2.702942331392665e-06,
      "loss": 5.051,
      "step": 19012
    },
    {
      "epoch": 0.7675817521194994,
      "grad_norm": 110.23296356201172,
      "learning_rate": 2.7020483209273982e-06,
      "loss": 5.5775,
      "step": 19013
    },
    {
      "epoch": 0.7676221235365361,
      "grad_norm": 86.84122467041016,
      "learning_rate": 2.701154435240061e-06,
      "loss": 5.4269,
      "step": 19014
    },
    {
      "epoch": 0.7676624949535729,
      "grad_norm": 261.8231506347656,
      "learning_rate": 2.7002606743459393e-06,
      "loss": 4.4003,
      "step": 19015
    },
    {
      "epoch": 0.7677028663706096,
      "grad_norm": 89.5096206665039,
      "learning_rate": 2.699367038260313e-06,
      "loss": 4.3298,
      "step": 19016
    },
    {
      "epoch": 0.7677432377876463,
      "grad_norm": 998.3075561523438,
      "learning_rate": 2.6984735269984663e-06,
      "loss": 4.82,
      "step": 19017
    },
    {
      "epoch": 0.7677836092046831,
      "grad_norm": 85.2576675415039,
      "learning_rate": 2.6975801405756683e-06,
      "loss": 5.6583,
      "step": 19018
    },
    {
      "epoch": 0.7678239806217199,
      "grad_norm": 117.70403289794922,
      "learning_rate": 2.6966868790071976e-06,
      "loss": 6.1514,
      "step": 19019
    },
    {
      "epoch": 0.7678643520387566,
      "grad_norm": 93.42221069335938,
      "learning_rate": 2.695793742308327e-06,
      "loss": 5.0728,
      "step": 19020
    },
    {
      "epoch": 0.7679047234557933,
      "grad_norm": 107.9045181274414,
      "learning_rate": 2.6949007304943264e-06,
      "loss": 5.184,
      "step": 19021
    },
    {
      "epoch": 0.76794509487283,
      "grad_norm": 78.9941635131836,
      "learning_rate": 2.6940078435804638e-06,
      "loss": 5.4167,
      "step": 19022
    },
    {
      "epoch": 0.7679854662898667,
      "grad_norm": 124.33866882324219,
      "learning_rate": 2.6931150815820083e-06,
      "loss": 3.765,
      "step": 19023
    },
    {
      "epoch": 0.7680258377069035,
      "grad_norm": 116.82257080078125,
      "learning_rate": 2.692222444514221e-06,
      "loss": 4.613,
      "step": 19024
    },
    {
      "epoch": 0.7680662091239403,
      "grad_norm": 81.9206771850586,
      "learning_rate": 2.6913299323923682e-06,
      "loss": 5.2984,
      "step": 19025
    },
    {
      "epoch": 0.768106580540977,
      "grad_norm": 120.57233428955078,
      "learning_rate": 2.690437545231703e-06,
      "loss": 4.7765,
      "step": 19026
    },
    {
      "epoch": 0.7681469519580137,
      "grad_norm": 63.91312789916992,
      "learning_rate": 2.689545283047488e-06,
      "loss": 3.5772,
      "step": 19027
    },
    {
      "epoch": 0.7681873233750505,
      "grad_norm": 123.68157196044922,
      "learning_rate": 2.688653145854977e-06,
      "loss": 4.1365,
      "step": 19028
    },
    {
      "epoch": 0.7682276947920872,
      "grad_norm": 122.00442504882812,
      "learning_rate": 2.6877611336694265e-06,
      "loss": 4.7373,
      "step": 19029
    },
    {
      "epoch": 0.7682680662091239,
      "grad_norm": 115.83918762207031,
      "learning_rate": 2.6868692465060832e-06,
      "loss": 4.3109,
      "step": 19030
    },
    {
      "epoch": 0.7683084376261606,
      "grad_norm": 137.49607849121094,
      "learning_rate": 2.6859774843801978e-06,
      "loss": 4.992,
      "step": 19031
    },
    {
      "epoch": 0.7683488090431975,
      "grad_norm": 101.23880004882812,
      "learning_rate": 2.685085847307017e-06,
      "loss": 3.8797,
      "step": 19032
    },
    {
      "epoch": 0.7683891804602342,
      "grad_norm": 98.4432144165039,
      "learning_rate": 2.684194335301792e-06,
      "loss": 4.4261,
      "step": 19033
    },
    {
      "epoch": 0.7684295518772709,
      "grad_norm": 195.17306518554688,
      "learning_rate": 2.6833029483797557e-06,
      "loss": 3.7488,
      "step": 19034
    },
    {
      "epoch": 0.7684699232943076,
      "grad_norm": 89.34574890136719,
      "learning_rate": 2.6824116865561535e-06,
      "loss": 5.0613,
      "step": 19035
    },
    {
      "epoch": 0.7685102947113444,
      "grad_norm": 88.76305389404297,
      "learning_rate": 2.6815205498462236e-06,
      "loss": 5.6056,
      "step": 19036
    },
    {
      "epoch": 0.7685506661283811,
      "grad_norm": 86.31171417236328,
      "learning_rate": 2.6806295382652048e-06,
      "loss": 4.4897,
      "step": 19037
    },
    {
      "epoch": 0.7685910375454178,
      "grad_norm": 249.92352294921875,
      "learning_rate": 2.6797386518283276e-06,
      "loss": 5.2515,
      "step": 19038
    },
    {
      "epoch": 0.7686314089624546,
      "grad_norm": 114.9353256225586,
      "learning_rate": 2.678847890550824e-06,
      "loss": 5.0159,
      "step": 19039
    },
    {
      "epoch": 0.7686717803794914,
      "grad_norm": 113.75701904296875,
      "learning_rate": 2.677957254447926e-06,
      "loss": 4.6084,
      "step": 19040
    },
    {
      "epoch": 0.7687121517965281,
      "grad_norm": 152.6111602783203,
      "learning_rate": 2.6770667435348597e-06,
      "loss": 4.1655,
      "step": 19041
    },
    {
      "epoch": 0.7687525232135648,
      "grad_norm": 101.9818115234375,
      "learning_rate": 2.676176357826853e-06,
      "loss": 3.0883,
      "step": 19042
    },
    {
      "epoch": 0.7687928946306015,
      "grad_norm": 139.86888122558594,
      "learning_rate": 2.675286097339128e-06,
      "loss": 6.6842,
      "step": 19043
    },
    {
      "epoch": 0.7688332660476382,
      "grad_norm": 178.2399444580078,
      "learning_rate": 2.6743959620869066e-06,
      "loss": 3.8832,
      "step": 19044
    },
    {
      "epoch": 0.768873637464675,
      "grad_norm": 107.78803253173828,
      "learning_rate": 2.67350595208541e-06,
      "loss": 5.8256,
      "step": 19045
    },
    {
      "epoch": 0.7689140088817118,
      "grad_norm": 102.14669799804688,
      "learning_rate": 2.6726160673498504e-06,
      "loss": 4.1113,
      "step": 19046
    },
    {
      "epoch": 0.7689543802987485,
      "grad_norm": 120.45973205566406,
      "learning_rate": 2.6717263078954447e-06,
      "loss": 5.2274,
      "step": 19047
    },
    {
      "epoch": 0.7689947517157852,
      "grad_norm": 132.34832763671875,
      "learning_rate": 2.670836673737407e-06,
      "loss": 5.6988,
      "step": 19048
    },
    {
      "epoch": 0.769035123132822,
      "grad_norm": 91.5418472290039,
      "learning_rate": 2.6699471648909504e-06,
      "loss": 4.9349,
      "step": 19049
    },
    {
      "epoch": 0.7690754945498587,
      "grad_norm": 94.98133850097656,
      "learning_rate": 2.669057781371277e-06,
      "loss": 5.5933,
      "step": 19050
    },
    {
      "epoch": 0.7691158659668954,
      "grad_norm": 108.73570251464844,
      "learning_rate": 2.6681685231935974e-06,
      "loss": 4.3795,
      "step": 19051
    },
    {
      "epoch": 0.7691562373839321,
      "grad_norm": 111.60334014892578,
      "learning_rate": 2.667279390373114e-06,
      "loss": 6.6851,
      "step": 19052
    },
    {
      "epoch": 0.769196608800969,
      "grad_norm": 99.99417114257812,
      "learning_rate": 2.6663903829250337e-06,
      "loss": 5.2765,
      "step": 19053
    },
    {
      "epoch": 0.7692369802180057,
      "grad_norm": 110.13368225097656,
      "learning_rate": 2.6655015008645492e-06,
      "loss": 5.2732,
      "step": 19054
    },
    {
      "epoch": 0.7692773516350424,
      "grad_norm": 92.83056640625,
      "learning_rate": 2.6646127442068615e-06,
      "loss": 2.777,
      "step": 19055
    },
    {
      "epoch": 0.7693177230520791,
      "grad_norm": 71.60945129394531,
      "learning_rate": 2.6637241129671666e-06,
      "loss": 3.086,
      "step": 19056
    },
    {
      "epoch": 0.7693580944691158,
      "grad_norm": 141.61595153808594,
      "learning_rate": 2.662835607160662e-06,
      "loss": 2.8149,
      "step": 19057
    },
    {
      "epoch": 0.7693984658861526,
      "grad_norm": 122.42619323730469,
      "learning_rate": 2.6619472268025306e-06,
      "loss": 4.8423,
      "step": 19058
    },
    {
      "epoch": 0.7694388373031893,
      "grad_norm": 99.09090423583984,
      "learning_rate": 2.661058971907967e-06,
      "loss": 6.0228,
      "step": 19059
    },
    {
      "epoch": 0.7694792087202261,
      "grad_norm": 108.36788940429688,
      "learning_rate": 2.660170842492157e-06,
      "loss": 6.0897,
      "step": 19060
    },
    {
      "epoch": 0.7695195801372628,
      "grad_norm": 256.45501708984375,
      "learning_rate": 2.659282838570285e-06,
      "loss": 5.5442,
      "step": 19061
    },
    {
      "epoch": 0.7695599515542996,
      "grad_norm": 74.21375274658203,
      "learning_rate": 2.6583949601575356e-06,
      "loss": 5.0368,
      "step": 19062
    },
    {
      "epoch": 0.7696003229713363,
      "grad_norm": 159.41827392578125,
      "learning_rate": 2.6575072072690888e-06,
      "loss": 4.446,
      "step": 19063
    },
    {
      "epoch": 0.769640694388373,
      "grad_norm": 126.47200775146484,
      "learning_rate": 2.6566195799201222e-06,
      "loss": 5.8458,
      "step": 19064
    },
    {
      "epoch": 0.7696810658054097,
      "grad_norm": 140.1324920654297,
      "learning_rate": 2.6557320781258157e-06,
      "loss": 6.1907,
      "step": 19065
    },
    {
      "epoch": 0.7697214372224465,
      "grad_norm": 77.75140380859375,
      "learning_rate": 2.6548447019013377e-06,
      "loss": 4.6055,
      "step": 19066
    },
    {
      "epoch": 0.7697618086394833,
      "grad_norm": 85.09014892578125,
      "learning_rate": 2.653957451261864e-06,
      "loss": 4.7717,
      "step": 19067
    },
    {
      "epoch": 0.76980218005652,
      "grad_norm": 84.599609375,
      "learning_rate": 2.653070326222562e-06,
      "loss": 5.5367,
      "step": 19068
    },
    {
      "epoch": 0.7698425514735567,
      "grad_norm": 80.99762725830078,
      "learning_rate": 2.652183326798605e-06,
      "loss": 4.4745,
      "step": 19069
    },
    {
      "epoch": 0.7698829228905935,
      "grad_norm": 117.05068969726562,
      "learning_rate": 2.651296453005152e-06,
      "loss": 4.2466,
      "step": 19070
    },
    {
      "epoch": 0.7699232943076302,
      "grad_norm": 89.14024353027344,
      "learning_rate": 2.6504097048573686e-06,
      "loss": 4.6629,
      "step": 19071
    },
    {
      "epoch": 0.7699636657246669,
      "grad_norm": 132.9286651611328,
      "learning_rate": 2.6495230823704175e-06,
      "loss": 5.2119,
      "step": 19072
    },
    {
      "epoch": 0.7700040371417036,
      "grad_norm": 106.68798828125,
      "learning_rate": 2.648636585559459e-06,
      "loss": 3.0913,
      "step": 19073
    },
    {
      "epoch": 0.7700444085587405,
      "grad_norm": 76.2298355102539,
      "learning_rate": 2.647750214439646e-06,
      "loss": 5.4164,
      "step": 19074
    },
    {
      "epoch": 0.7700847799757772,
      "grad_norm": 97.36701202392578,
      "learning_rate": 2.6468639690261357e-06,
      "loss": 4.4883,
      "step": 19075
    },
    {
      "epoch": 0.7701251513928139,
      "grad_norm": 83.01041412353516,
      "learning_rate": 2.6459778493340806e-06,
      "loss": 5.8352,
      "step": 19076
    },
    {
      "epoch": 0.7701655228098506,
      "grad_norm": 99.67121887207031,
      "learning_rate": 2.645091855378631e-06,
      "loss": 5.4718,
      "step": 19077
    },
    {
      "epoch": 0.7702058942268873,
      "grad_norm": 144.61358642578125,
      "learning_rate": 2.644205987174936e-06,
      "loss": 4.6346,
      "step": 19078
    },
    {
      "epoch": 0.7702462656439241,
      "grad_norm": 109.77588653564453,
      "learning_rate": 2.6433202447381413e-06,
      "loss": 4.7424,
      "step": 19079
    },
    {
      "epoch": 0.7702866370609608,
      "grad_norm": 102.90213775634766,
      "learning_rate": 2.6424346280833948e-06,
      "loss": 5.56,
      "step": 19080
    },
    {
      "epoch": 0.7703270084779976,
      "grad_norm": 80.77776336669922,
      "learning_rate": 2.6415491372258316e-06,
      "loss": 5.9891,
      "step": 19081
    },
    {
      "epoch": 0.7703673798950343,
      "grad_norm": 73.40120697021484,
      "learning_rate": 2.640663772180595e-06,
      "loss": 4.1013,
      "step": 19082
    },
    {
      "epoch": 0.7704077513120711,
      "grad_norm": 80.88568878173828,
      "learning_rate": 2.639778532962822e-06,
      "loss": 4.7004,
      "step": 19083
    },
    {
      "epoch": 0.7704481227291078,
      "grad_norm": 83.07941436767578,
      "learning_rate": 2.6388934195876505e-06,
      "loss": 4.7585,
      "step": 19084
    },
    {
      "epoch": 0.7704884941461445,
      "grad_norm": 120.07256317138672,
      "learning_rate": 2.6380084320702136e-06,
      "loss": 3.7998,
      "step": 19085
    },
    {
      "epoch": 0.7705288655631812,
      "grad_norm": 127.05562591552734,
      "learning_rate": 2.6371235704256394e-06,
      "loss": 4.1641,
      "step": 19086
    },
    {
      "epoch": 0.770569236980218,
      "grad_norm": 109.65933227539062,
      "learning_rate": 2.6362388346690572e-06,
      "loss": 5.9558,
      "step": 19087
    },
    {
      "epoch": 0.7706096083972548,
      "grad_norm": 93.71080780029297,
      "learning_rate": 2.635354224815596e-06,
      "loss": 3.0169,
      "step": 19088
    },
    {
      "epoch": 0.7706499798142915,
      "grad_norm": 83.96878814697266,
      "learning_rate": 2.634469740880382e-06,
      "loss": 2.418,
      "step": 19089
    },
    {
      "epoch": 0.7706903512313282,
      "grad_norm": 138.06707763671875,
      "learning_rate": 2.633585382878534e-06,
      "loss": 4.7088,
      "step": 19090
    },
    {
      "epoch": 0.770730722648365,
      "grad_norm": 112.49639892578125,
      "learning_rate": 2.632701150825173e-06,
      "loss": 5.2932,
      "step": 19091
    },
    {
      "epoch": 0.7707710940654017,
      "grad_norm": 131.92059326171875,
      "learning_rate": 2.6318170447354198e-06,
      "loss": 6.0244,
      "step": 19092
    },
    {
      "epoch": 0.7708114654824384,
      "grad_norm": 113.5728530883789,
      "learning_rate": 2.63093306462439e-06,
      "loss": 5.804,
      "step": 19093
    },
    {
      "epoch": 0.7708518368994751,
      "grad_norm": 92.86446380615234,
      "learning_rate": 2.6300492105071964e-06,
      "loss": 4.9325,
      "step": 19094
    },
    {
      "epoch": 0.770892208316512,
      "grad_norm": 161.7028045654297,
      "learning_rate": 2.629165482398949e-06,
      "loss": 5.6528,
      "step": 19095
    },
    {
      "epoch": 0.7709325797335487,
      "grad_norm": 86.85629272460938,
      "learning_rate": 2.6282818803147615e-06,
      "loss": 5.3086,
      "step": 19096
    },
    {
      "epoch": 0.7709729511505854,
      "grad_norm": 92.08740234375,
      "learning_rate": 2.6273984042697387e-06,
      "loss": 4.371,
      "step": 19097
    },
    {
      "epoch": 0.7710133225676221,
      "grad_norm": 78.80965423583984,
      "learning_rate": 2.6265150542789863e-06,
      "loss": 5.1873,
      "step": 19098
    },
    {
      "epoch": 0.7710536939846588,
      "grad_norm": 101.73651123046875,
      "learning_rate": 2.625631830357609e-06,
      "loss": 7.389,
      "step": 19099
    },
    {
      "epoch": 0.7710940654016956,
      "grad_norm": 84.59127807617188,
      "learning_rate": 2.6247487325207065e-06,
      "loss": 3.2883,
      "step": 19100
    },
    {
      "epoch": 0.7711344368187323,
      "grad_norm": 139.3022918701172,
      "learning_rate": 2.6238657607833817e-06,
      "loss": 5.7227,
      "step": 19101
    },
    {
      "epoch": 0.7711748082357691,
      "grad_norm": 147.5357208251953,
      "learning_rate": 2.6229829151607254e-06,
      "loss": 6.1238,
      "step": 19102
    },
    {
      "epoch": 0.7712151796528058,
      "grad_norm": 87.69392395019531,
      "learning_rate": 2.6221001956678336e-06,
      "loss": 3.711,
      "step": 19103
    },
    {
      "epoch": 0.7712555510698426,
      "grad_norm": 88.1605224609375,
      "learning_rate": 2.621217602319801e-06,
      "loss": 5.4232,
      "step": 19104
    },
    {
      "epoch": 0.7712959224868793,
      "grad_norm": 90.32806396484375,
      "learning_rate": 2.6203351351317185e-06,
      "loss": 3.9524,
      "step": 19105
    },
    {
      "epoch": 0.771336293903916,
      "grad_norm": 79.8164291381836,
      "learning_rate": 2.61945279411867e-06,
      "loss": 5.6275,
      "step": 19106
    },
    {
      "epoch": 0.7713766653209527,
      "grad_norm": 215.23887634277344,
      "learning_rate": 2.618570579295744e-06,
      "loss": 4.377,
      "step": 19107
    },
    {
      "epoch": 0.7714170367379896,
      "grad_norm": 73.71119689941406,
      "learning_rate": 2.6176884906780244e-06,
      "loss": 4.9283,
      "step": 19108
    },
    {
      "epoch": 0.7714574081550263,
      "grad_norm": 92.99851989746094,
      "learning_rate": 2.6168065282805945e-06,
      "loss": 5.3816,
      "step": 19109
    },
    {
      "epoch": 0.771497779572063,
      "grad_norm": 101.71398162841797,
      "learning_rate": 2.61592469211853e-06,
      "loss": 5.744,
      "step": 19110
    },
    {
      "epoch": 0.7715381509890997,
      "grad_norm": 76.21215057373047,
      "learning_rate": 2.61504298220691e-06,
      "loss": 4.0369,
      "step": 19111
    },
    {
      "epoch": 0.7715785224061364,
      "grad_norm": 131.16893005371094,
      "learning_rate": 2.6141613985608093e-06,
      "loss": 5.4066,
      "step": 19112
    },
    {
      "epoch": 0.7716188938231732,
      "grad_norm": 77.71867370605469,
      "learning_rate": 2.613279941195306e-06,
      "loss": 4.0525,
      "step": 19113
    },
    {
      "epoch": 0.7716592652402099,
      "grad_norm": 107.70590209960938,
      "learning_rate": 2.612398610125463e-06,
      "loss": 5.691,
      "step": 19114
    },
    {
      "epoch": 0.7716996366572467,
      "grad_norm": 82.48686218261719,
      "learning_rate": 2.6115174053663517e-06,
      "loss": 6.3557,
      "step": 19115
    },
    {
      "epoch": 0.7717400080742834,
      "grad_norm": 99.65131378173828,
      "learning_rate": 2.6106363269330393e-06,
      "loss": 5.3211,
      "step": 19116
    },
    {
      "epoch": 0.7717803794913202,
      "grad_norm": 84.69849395751953,
      "learning_rate": 2.6097553748405914e-06,
      "loss": 6.1815,
      "step": 19117
    },
    {
      "epoch": 0.7718207509083569,
      "grad_norm": 76.82052612304688,
      "learning_rate": 2.6088745491040678e-06,
      "loss": 4.5361,
      "step": 19118
    },
    {
      "epoch": 0.7718611223253936,
      "grad_norm": 92.67589569091797,
      "learning_rate": 2.6079938497385304e-06,
      "loss": 3.6321,
      "step": 19119
    },
    {
      "epoch": 0.7719014937424303,
      "grad_norm": 110.86865997314453,
      "learning_rate": 2.6071132767590367e-06,
      "loss": 4.9015,
      "step": 19120
    },
    {
      "epoch": 0.7719418651594671,
      "grad_norm": 102.806396484375,
      "learning_rate": 2.6062328301806463e-06,
      "loss": 5.3276,
      "step": 19121
    },
    {
      "epoch": 0.7719822365765039,
      "grad_norm": 75.15737915039062,
      "learning_rate": 2.605352510018404e-06,
      "loss": 5.3208,
      "step": 19122
    },
    {
      "epoch": 0.7720226079935406,
      "grad_norm": 132.0384521484375,
      "learning_rate": 2.6044723162873675e-06,
      "loss": 5.2109,
      "step": 19123
    },
    {
      "epoch": 0.7720629794105773,
      "grad_norm": 73.97681427001953,
      "learning_rate": 2.6035922490025844e-06,
      "loss": 5.6133,
      "step": 19124
    },
    {
      "epoch": 0.772103350827614,
      "grad_norm": 137.79766845703125,
      "learning_rate": 2.602712308179104e-06,
      "loss": 5.1083,
      "step": 19125
    },
    {
      "epoch": 0.7721437222446508,
      "grad_norm": 89.89429473876953,
      "learning_rate": 2.601832493831967e-06,
      "loss": 4.5499,
      "step": 19126
    },
    {
      "epoch": 0.7721840936616875,
      "grad_norm": 415.8948974609375,
      "learning_rate": 2.600952805976219e-06,
      "loss": 5.6208,
      "step": 19127
    },
    {
      "epoch": 0.7722244650787242,
      "grad_norm": 102.42790222167969,
      "learning_rate": 2.6000732446268982e-06,
      "loss": 6.2658,
      "step": 19128
    },
    {
      "epoch": 0.772264836495761,
      "grad_norm": 84.18821716308594,
      "learning_rate": 2.5991938097990497e-06,
      "loss": 5.2126,
      "step": 19129
    },
    {
      "epoch": 0.7723052079127978,
      "grad_norm": 124.8435287475586,
      "learning_rate": 2.5983145015077003e-06,
      "loss": 3.7197,
      "step": 19130
    },
    {
      "epoch": 0.7723455793298345,
      "grad_norm": 75.78219604492188,
      "learning_rate": 2.5974353197678902e-06,
      "loss": 5.3653,
      "step": 19131
    },
    {
      "epoch": 0.7723859507468712,
      "grad_norm": 108.45063781738281,
      "learning_rate": 2.5965562645946497e-06,
      "loss": 4.9591,
      "step": 19132
    },
    {
      "epoch": 0.7724263221639079,
      "grad_norm": 104.93436431884766,
      "learning_rate": 2.5956773360030107e-06,
      "loss": 4.7296,
      "step": 19133
    },
    {
      "epoch": 0.7724666935809447,
      "grad_norm": 119.25012969970703,
      "learning_rate": 2.5947985340079974e-06,
      "loss": 4.6829,
      "step": 19134
    },
    {
      "epoch": 0.7725070649979814,
      "grad_norm": 113.41058349609375,
      "learning_rate": 2.5939198586246352e-06,
      "loss": 6.5072,
      "step": 19135
    },
    {
      "epoch": 0.7725474364150182,
      "grad_norm": 87.4283447265625,
      "learning_rate": 2.59304130986795e-06,
      "loss": 4.6499,
      "step": 19136
    },
    {
      "epoch": 0.7725878078320549,
      "grad_norm": 98.16060638427734,
      "learning_rate": 2.5921628877529616e-06,
      "loss": 3.455,
      "step": 19137
    },
    {
      "epoch": 0.7726281792490917,
      "grad_norm": 101.1607437133789,
      "learning_rate": 2.5912845922946896e-06,
      "loss": 4.5035,
      "step": 19138
    },
    {
      "epoch": 0.7726685506661284,
      "grad_norm": 112.49175262451172,
      "learning_rate": 2.5904064235081506e-06,
      "loss": 6.2629,
      "step": 19139
    },
    {
      "epoch": 0.7727089220831651,
      "grad_norm": 66.9305191040039,
      "learning_rate": 2.5895283814083583e-06,
      "loss": 3.3578,
      "step": 19140
    },
    {
      "epoch": 0.7727492935002018,
      "grad_norm": 64.1286849975586,
      "learning_rate": 2.5886504660103295e-06,
      "loss": 4.7133,
      "step": 19141
    },
    {
      "epoch": 0.7727896649172386,
      "grad_norm": 69.22711181640625,
      "learning_rate": 2.587772677329068e-06,
      "loss": 5.524,
      "step": 19142
    },
    {
      "epoch": 0.7728300363342754,
      "grad_norm": 97.27715301513672,
      "learning_rate": 2.5868950153795847e-06,
      "loss": 6.1776,
      "step": 19143
    },
    {
      "epoch": 0.7728704077513121,
      "grad_norm": 92.07927703857422,
      "learning_rate": 2.5860174801768857e-06,
      "loss": 4.67,
      "step": 19144
    },
    {
      "epoch": 0.7729107791683488,
      "grad_norm": 88.0382308959961,
      "learning_rate": 2.5851400717359787e-06,
      "loss": 4.687,
      "step": 19145
    },
    {
      "epoch": 0.7729511505853855,
      "grad_norm": 105.95040893554688,
      "learning_rate": 2.5842627900718575e-06,
      "loss": 5.1136,
      "step": 19146
    },
    {
      "epoch": 0.7729915220024223,
      "grad_norm": 86.93440246582031,
      "learning_rate": 2.5833856351995256e-06,
      "loss": 4.4694,
      "step": 19147
    },
    {
      "epoch": 0.773031893419459,
      "grad_norm": 102.78530883789062,
      "learning_rate": 2.582508607133981e-06,
      "loss": 5.6819,
      "step": 19148
    },
    {
      "epoch": 0.7730722648364957,
      "grad_norm": 84.90056610107422,
      "learning_rate": 2.5816317058902207e-06,
      "loss": 4.6377,
      "step": 19149
    },
    {
      "epoch": 0.7731126362535325,
      "grad_norm": 111.42955780029297,
      "learning_rate": 2.5807549314832327e-06,
      "loss": 5.3454,
      "step": 19150
    },
    {
      "epoch": 0.7731530076705693,
      "grad_norm": 145.3705596923828,
      "learning_rate": 2.57987828392801e-06,
      "loss": 3.5108,
      "step": 19151
    },
    {
      "epoch": 0.773193379087606,
      "grad_norm": 80.64521789550781,
      "learning_rate": 2.5790017632395404e-06,
      "loss": 4.677,
      "step": 19152
    },
    {
      "epoch": 0.7732337505046427,
      "grad_norm": 83.10147094726562,
      "learning_rate": 2.5781253694328156e-06,
      "loss": 3.351,
      "step": 19153
    },
    {
      "epoch": 0.7732741219216794,
      "grad_norm": 122.97138977050781,
      "learning_rate": 2.5772491025228117e-06,
      "loss": 6.4038,
      "step": 19154
    },
    {
      "epoch": 0.7733144933387162,
      "grad_norm": 124.55724334716797,
      "learning_rate": 2.576372962524516e-06,
      "loss": 5.1353,
      "step": 19155
    },
    {
      "epoch": 0.7733548647557529,
      "grad_norm": 113.47697448730469,
      "learning_rate": 2.575496949452906e-06,
      "loss": 3.1512,
      "step": 19156
    },
    {
      "epoch": 0.7733952361727897,
      "grad_norm": 73.41040802001953,
      "learning_rate": 2.574621063322962e-06,
      "loss": 3.8607,
      "step": 19157
    },
    {
      "epoch": 0.7734356075898264,
      "grad_norm": 116.63786315917969,
      "learning_rate": 2.573745304149657e-06,
      "loss": 4.9554,
      "step": 19158
    },
    {
      "epoch": 0.7734759790068632,
      "grad_norm": 121.53776550292969,
      "learning_rate": 2.5728696719479664e-06,
      "loss": 4.2877,
      "step": 19159
    },
    {
      "epoch": 0.7735163504238999,
      "grad_norm": 129.6900177001953,
      "learning_rate": 2.571994166732861e-06,
      "loss": 3.87,
      "step": 19160
    },
    {
      "epoch": 0.7735567218409366,
      "grad_norm": 86.13273620605469,
      "learning_rate": 2.5711187885193123e-06,
      "loss": 4.0994,
      "step": 19161
    },
    {
      "epoch": 0.7735970932579733,
      "grad_norm": 124.516357421875,
      "learning_rate": 2.570243537322281e-06,
      "loss": 5.4905,
      "step": 19162
    },
    {
      "epoch": 0.77363746467501,
      "grad_norm": 99.42558288574219,
      "learning_rate": 2.5693684131567367e-06,
      "loss": 4.7091,
      "step": 19163
    },
    {
      "epoch": 0.7736778360920469,
      "grad_norm": 127.93028259277344,
      "learning_rate": 2.5684934160376395e-06,
      "loss": 6.4419,
      "step": 19164
    },
    {
      "epoch": 0.7737182075090836,
      "grad_norm": 89.56018829345703,
      "learning_rate": 2.5676185459799553e-06,
      "loss": 4.5886,
      "step": 19165
    },
    {
      "epoch": 0.7737585789261203,
      "grad_norm": 151.37850952148438,
      "learning_rate": 2.5667438029986347e-06,
      "loss": 4.3259,
      "step": 19166
    },
    {
      "epoch": 0.773798950343157,
      "grad_norm": 132.02699279785156,
      "learning_rate": 2.565869187108637e-06,
      "loss": 6.0325,
      "step": 19167
    },
    {
      "epoch": 0.7738393217601938,
      "grad_norm": 91.843505859375,
      "learning_rate": 2.5649946983249164e-06,
      "loss": 5.0127,
      "step": 19168
    },
    {
      "epoch": 0.7738796931772305,
      "grad_norm": 86.65823364257812,
      "learning_rate": 2.564120336662428e-06,
      "loss": 5.7506,
      "step": 19169
    },
    {
      "epoch": 0.7739200645942672,
      "grad_norm": 74.72952270507812,
      "learning_rate": 2.563246102136113e-06,
      "loss": 4.9246,
      "step": 19170
    },
    {
      "epoch": 0.773960436011304,
      "grad_norm": 105.81350708007812,
      "learning_rate": 2.562371994760925e-06,
      "loss": 4.772,
      "step": 19171
    },
    {
      "epoch": 0.7740008074283408,
      "grad_norm": 95.54031372070312,
      "learning_rate": 2.5614980145518074e-06,
      "loss": 5.5024,
      "step": 19172
    },
    {
      "epoch": 0.7740411788453775,
      "grad_norm": 96.75582885742188,
      "learning_rate": 2.560624161523707e-06,
      "loss": 2.836,
      "step": 19173
    },
    {
      "epoch": 0.7740815502624142,
      "grad_norm": 73.669677734375,
      "learning_rate": 2.5597504356915583e-06,
      "loss": 6.3923,
      "step": 19174
    },
    {
      "epoch": 0.7741219216794509,
      "grad_norm": 93.73635864257812,
      "learning_rate": 2.5588768370703032e-06,
      "loss": 5.8771,
      "step": 19175
    },
    {
      "epoch": 0.7741622930964877,
      "grad_norm": 83.80142974853516,
      "learning_rate": 2.558003365674877e-06,
      "loss": 3.0213,
      "step": 19176
    },
    {
      "epoch": 0.7742026645135244,
      "grad_norm": 162.431884765625,
      "learning_rate": 2.5571300215202154e-06,
      "loss": 4.3643,
      "step": 19177
    },
    {
      "epoch": 0.7742430359305612,
      "grad_norm": 81.32335662841797,
      "learning_rate": 2.55625680462125e-06,
      "loss": 4.2665,
      "step": 19178
    },
    {
      "epoch": 0.7742834073475979,
      "grad_norm": 116.04594421386719,
      "learning_rate": 2.555383714992911e-06,
      "loss": 6.1558,
      "step": 19179
    },
    {
      "epoch": 0.7743237787646347,
      "grad_norm": 220.92666625976562,
      "learning_rate": 2.5545107526501256e-06,
      "loss": 4.3139,
      "step": 19180
    },
    {
      "epoch": 0.7743641501816714,
      "grad_norm": 226.87454223632812,
      "learning_rate": 2.553637917607823e-06,
      "loss": 4.8417,
      "step": 19181
    },
    {
      "epoch": 0.7744045215987081,
      "grad_norm": 90.766357421875,
      "learning_rate": 2.552765209880921e-06,
      "loss": 3.6602,
      "step": 19182
    },
    {
      "epoch": 0.7744448930157448,
      "grad_norm": 68.137451171875,
      "learning_rate": 2.5518926294843427e-06,
      "loss": 3.9569,
      "step": 19183
    },
    {
      "epoch": 0.7744852644327815,
      "grad_norm": 134.177734375,
      "learning_rate": 2.551020176433009e-06,
      "loss": 4.2395,
      "step": 19184
    },
    {
      "epoch": 0.7745256358498184,
      "grad_norm": 75.605712890625,
      "learning_rate": 2.5501478507418375e-06,
      "loss": 4.7737,
      "step": 19185
    },
    {
      "epoch": 0.7745660072668551,
      "grad_norm": 92.63485717773438,
      "learning_rate": 2.549275652425738e-06,
      "loss": 5.7214,
      "step": 19186
    },
    {
      "epoch": 0.7746063786838918,
      "grad_norm": 82.89752197265625,
      "learning_rate": 2.548403581499627e-06,
      "loss": 2.7499,
      "step": 19187
    },
    {
      "epoch": 0.7746467501009285,
      "grad_norm": 59.84473419189453,
      "learning_rate": 2.5475316379784133e-06,
      "loss": 4.4395,
      "step": 19188
    },
    {
      "epoch": 0.7746871215179653,
      "grad_norm": 85.92578887939453,
      "learning_rate": 2.546659821877009e-06,
      "loss": 5.657,
      "step": 19189
    },
    {
      "epoch": 0.774727492935002,
      "grad_norm": 118.78221893310547,
      "learning_rate": 2.545788133210314e-06,
      "loss": 5.0054,
      "step": 19190
    },
    {
      "epoch": 0.7747678643520387,
      "grad_norm": 93.16352844238281,
      "learning_rate": 2.5449165719932355e-06,
      "loss": 5.6123,
      "step": 19191
    },
    {
      "epoch": 0.7748082357690755,
      "grad_norm": 111.603759765625,
      "learning_rate": 2.544045138240675e-06,
      "loss": 4.8033,
      "step": 19192
    },
    {
      "epoch": 0.7748486071861123,
      "grad_norm": 72.44287872314453,
      "learning_rate": 2.5431738319675337e-06,
      "loss": 5.4645,
      "step": 19193
    },
    {
      "epoch": 0.774888978603149,
      "grad_norm": 89.66188049316406,
      "learning_rate": 2.542302653188704e-06,
      "loss": 5.0838,
      "step": 19194
    },
    {
      "epoch": 0.7749293500201857,
      "grad_norm": 72.03343200683594,
      "learning_rate": 2.5414316019190842e-06,
      "loss": 6.198,
      "step": 19195
    },
    {
      "epoch": 0.7749697214372224,
      "grad_norm": 84.41323852539062,
      "learning_rate": 2.5405606781735682e-06,
      "loss": 6.1382,
      "step": 19196
    },
    {
      "epoch": 0.7750100928542591,
      "grad_norm": 88.9063491821289,
      "learning_rate": 2.5396898819670444e-06,
      "loss": 4.9713,
      "step": 19197
    },
    {
      "epoch": 0.7750504642712959,
      "grad_norm": 68.05691528320312,
      "learning_rate": 2.538819213314403e-06,
      "loss": 5.8855,
      "step": 19198
    },
    {
      "epoch": 0.7750908356883327,
      "grad_norm": 114.96426391601562,
      "learning_rate": 2.537948672230529e-06,
      "loss": 5.4027,
      "step": 19199
    },
    {
      "epoch": 0.7751312071053694,
      "grad_norm": 66.98800659179688,
      "learning_rate": 2.537078258730308e-06,
      "loss": 3.8256,
      "step": 19200
    },
    {
      "epoch": 0.7751715785224061,
      "grad_norm": 111.48900604248047,
      "learning_rate": 2.536207972828625e-06,
      "loss": 6.0624,
      "step": 19201
    },
    {
      "epoch": 0.7752119499394429,
      "grad_norm": 86.541259765625,
      "learning_rate": 2.535337814540353e-06,
      "loss": 4.8373,
      "step": 19202
    },
    {
      "epoch": 0.7752523213564796,
      "grad_norm": 104.56082916259766,
      "learning_rate": 2.534467783880373e-06,
      "loss": 4.4507,
      "step": 19203
    },
    {
      "epoch": 0.7752926927735163,
      "grad_norm": 105.47827911376953,
      "learning_rate": 2.533597880863562e-06,
      "loss": 3.4022,
      "step": 19204
    },
    {
      "epoch": 0.775333064190553,
      "grad_norm": 92.66878509521484,
      "learning_rate": 2.5327281055047936e-06,
      "loss": 4.6665,
      "step": 19205
    },
    {
      "epoch": 0.7753734356075899,
      "grad_norm": 96.35285949707031,
      "learning_rate": 2.531858457818934e-06,
      "loss": 4.4201,
      "step": 19206
    },
    {
      "epoch": 0.7754138070246266,
      "grad_norm": 114.85746002197266,
      "learning_rate": 2.5309889378208562e-06,
      "loss": 5.4912,
      "step": 19207
    },
    {
      "epoch": 0.7754541784416633,
      "grad_norm": 85.5346908569336,
      "learning_rate": 2.5301195455254258e-06,
      "loss": 6.2347,
      "step": 19208
    },
    {
      "epoch": 0.7754945498587,
      "grad_norm": 79.14006042480469,
      "learning_rate": 2.5292502809475107e-06,
      "loss": 4.1987,
      "step": 19209
    },
    {
      "epoch": 0.7755349212757368,
      "grad_norm": 84.6247787475586,
      "learning_rate": 2.528381144101968e-06,
      "loss": 6.4924,
      "step": 19210
    },
    {
      "epoch": 0.7755752926927735,
      "grad_norm": 100.05183410644531,
      "learning_rate": 2.52751213500366e-06,
      "loss": 3.8746,
      "step": 19211
    },
    {
      "epoch": 0.7756156641098102,
      "grad_norm": 113.58340454101562,
      "learning_rate": 2.5266432536674444e-06,
      "loss": 5.4203,
      "step": 19212
    },
    {
      "epoch": 0.775656035526847,
      "grad_norm": 114.05101776123047,
      "learning_rate": 2.525774500108181e-06,
      "loss": 5.2935,
      "step": 19213
    },
    {
      "epoch": 0.7756964069438838,
      "grad_norm": 161.07371520996094,
      "learning_rate": 2.5249058743407174e-06,
      "loss": 4.3443,
      "step": 19214
    },
    {
      "epoch": 0.7757367783609205,
      "grad_norm": 98.1182632446289,
      "learning_rate": 2.5240373763799074e-06,
      "loss": 4.5227,
      "step": 19215
    },
    {
      "epoch": 0.7757771497779572,
      "grad_norm": 105.80677032470703,
      "learning_rate": 2.5231690062406e-06,
      "loss": 3.7233,
      "step": 19216
    },
    {
      "epoch": 0.7758175211949939,
      "grad_norm": 89.92451477050781,
      "learning_rate": 2.522300763937644e-06,
      "loss": 4.7083,
      "step": 19217
    },
    {
      "epoch": 0.7758578926120306,
      "grad_norm": 92.01773834228516,
      "learning_rate": 2.5214326494858834e-06,
      "loss": 4.2787,
      "step": 19218
    },
    {
      "epoch": 0.7758982640290675,
      "grad_norm": 104.14125061035156,
      "learning_rate": 2.52056466290016e-06,
      "loss": 5.3094,
      "step": 19219
    },
    {
      "epoch": 0.7759386354461042,
      "grad_norm": 78.2478256225586,
      "learning_rate": 2.5196968041953153e-06,
      "loss": 5.2707,
      "step": 19220
    },
    {
      "epoch": 0.7759790068631409,
      "grad_norm": 105.71757507324219,
      "learning_rate": 2.5188290733861897e-06,
      "loss": 4.8023,
      "step": 19221
    },
    {
      "epoch": 0.7760193782801776,
      "grad_norm": 69.0725326538086,
      "learning_rate": 2.5179614704876153e-06,
      "loss": 5.1184,
      "step": 19222
    },
    {
      "epoch": 0.7760597496972144,
      "grad_norm": 106.84400939941406,
      "learning_rate": 2.517093995514428e-06,
      "loss": 5.4791,
      "step": 19223
    },
    {
      "epoch": 0.7761001211142511,
      "grad_norm": 109.84101104736328,
      "learning_rate": 2.516226648481459e-06,
      "loss": 4.515,
      "step": 19224
    },
    {
      "epoch": 0.7761404925312878,
      "grad_norm": 115.90067291259766,
      "learning_rate": 2.515359429403541e-06,
      "loss": 5.3148,
      "step": 19225
    },
    {
      "epoch": 0.7761808639483246,
      "grad_norm": 94.7411117553711,
      "learning_rate": 2.5144923382954965e-06,
      "loss": 3.957,
      "step": 19226
    },
    {
      "epoch": 0.7762212353653614,
      "grad_norm": 99.18795013427734,
      "learning_rate": 2.513625375172154e-06,
      "loss": 4.0024,
      "step": 19227
    },
    {
      "epoch": 0.7762616067823981,
      "grad_norm": 106.43871307373047,
      "learning_rate": 2.5127585400483345e-06,
      "loss": 5.5602,
      "step": 19228
    },
    {
      "epoch": 0.7763019781994348,
      "grad_norm": 79.62443542480469,
      "learning_rate": 2.5118918329388632e-06,
      "loss": 3.0413,
      "step": 19229
    },
    {
      "epoch": 0.7763423496164715,
      "grad_norm": 83.24755859375,
      "learning_rate": 2.5110252538585524e-06,
      "loss": 4.7594,
      "step": 19230
    },
    {
      "epoch": 0.7763827210335082,
      "grad_norm": 105.21461486816406,
      "learning_rate": 2.5101588028222225e-06,
      "loss": 4.2808,
      "step": 19231
    },
    {
      "epoch": 0.776423092450545,
      "grad_norm": 91.12574768066406,
      "learning_rate": 2.509292479844686e-06,
      "loss": 5.5118,
      "step": 19232
    },
    {
      "epoch": 0.7764634638675818,
      "grad_norm": 150.54722595214844,
      "learning_rate": 2.5084262849407593e-06,
      "loss": 5.0944,
      "step": 19233
    },
    {
      "epoch": 0.7765038352846185,
      "grad_norm": 93.66814422607422,
      "learning_rate": 2.5075602181252455e-06,
      "loss": 4.2158,
      "step": 19234
    },
    {
      "epoch": 0.7765442067016552,
      "grad_norm": 181.5543975830078,
      "learning_rate": 2.5066942794129567e-06,
      "loss": 7.923,
      "step": 19235
    },
    {
      "epoch": 0.776584578118692,
      "grad_norm": 74.35708618164062,
      "learning_rate": 2.5058284688186986e-06,
      "loss": 3.9172,
      "step": 19236
    },
    {
      "epoch": 0.7766249495357287,
      "grad_norm": 97.89195251464844,
      "learning_rate": 2.504962786357272e-06,
      "loss": 4.8862,
      "step": 19237
    },
    {
      "epoch": 0.7766653209527654,
      "grad_norm": 109.17638397216797,
      "learning_rate": 2.50409723204348e-06,
      "loss": 3.0103,
      "step": 19238
    },
    {
      "epoch": 0.7767056923698021,
      "grad_norm": 121.97586059570312,
      "learning_rate": 2.503231805892121e-06,
      "loss": 3.9824,
      "step": 19239
    },
    {
      "epoch": 0.776746063786839,
      "grad_norm": 90.86832427978516,
      "learning_rate": 2.502366507917993e-06,
      "loss": 4.2042,
      "step": 19240
    },
    {
      "epoch": 0.7767864352038757,
      "grad_norm": 101.60736846923828,
      "learning_rate": 2.501501338135891e-06,
      "loss": 4.0798,
      "step": 19241
    },
    {
      "epoch": 0.7768268066209124,
      "grad_norm": 92.86701202392578,
      "learning_rate": 2.5006362965606034e-06,
      "loss": 3.4788,
      "step": 19242
    },
    {
      "epoch": 0.7768671780379491,
      "grad_norm": 82.57831573486328,
      "learning_rate": 2.4997713832069228e-06,
      "loss": 5.5161,
      "step": 19243
    },
    {
      "epoch": 0.7769075494549859,
      "grad_norm": 117.20643615722656,
      "learning_rate": 2.4989065980896363e-06,
      "loss": 5.3193,
      "step": 19244
    },
    {
      "epoch": 0.7769479208720226,
      "grad_norm": 79.57453918457031,
      "learning_rate": 2.498041941223536e-06,
      "loss": 5.1816,
      "step": 19245
    },
    {
      "epoch": 0.7769882922890593,
      "grad_norm": 88.5714111328125,
      "learning_rate": 2.497177412623395e-06,
      "loss": 4.5843,
      "step": 19246
    },
    {
      "epoch": 0.7770286637060961,
      "grad_norm": 84.38983917236328,
      "learning_rate": 2.4963130123040013e-06,
      "loss": 5.3527,
      "step": 19247
    },
    {
      "epoch": 0.7770690351231329,
      "grad_norm": 68.04292297363281,
      "learning_rate": 2.495448740280132e-06,
      "loss": 4.5404,
      "step": 19248
    },
    {
      "epoch": 0.7771094065401696,
      "grad_norm": 89.42326354980469,
      "learning_rate": 2.4945845965665683e-06,
      "loss": 5.5766,
      "step": 19249
    },
    {
      "epoch": 0.7771497779572063,
      "grad_norm": 94.42096710205078,
      "learning_rate": 2.493720581178077e-06,
      "loss": 6.4522,
      "step": 19250
    },
    {
      "epoch": 0.777190149374243,
      "grad_norm": 127.59058380126953,
      "learning_rate": 2.4928566941294373e-06,
      "loss": 6.2019,
      "step": 19251
    },
    {
      "epoch": 0.7772305207912797,
      "grad_norm": 167.3470458984375,
      "learning_rate": 2.4919929354354176e-06,
      "loss": 4.1645,
      "step": 19252
    },
    {
      "epoch": 0.7772708922083165,
      "grad_norm": 169.270263671875,
      "learning_rate": 2.4911293051107886e-06,
      "loss": 4.6214,
      "step": 19253
    },
    {
      "epoch": 0.7773112636253533,
      "grad_norm": 194.74520874023438,
      "learning_rate": 2.490265803170311e-06,
      "loss": 5.3305,
      "step": 19254
    },
    {
      "epoch": 0.77735163504239,
      "grad_norm": 82.04663848876953,
      "learning_rate": 2.489402429628751e-06,
      "loss": 4.2717,
      "step": 19255
    },
    {
      "epoch": 0.7773920064594267,
      "grad_norm": 68.03521728515625,
      "learning_rate": 2.488539184500872e-06,
      "loss": 3.9348,
      "step": 19256
    },
    {
      "epoch": 0.7774323778764635,
      "grad_norm": 579.3009643554688,
      "learning_rate": 2.487676067801432e-06,
      "loss": 5.5044,
      "step": 19257
    },
    {
      "epoch": 0.7774727492935002,
      "grad_norm": 117.05616760253906,
      "learning_rate": 2.4868130795451895e-06,
      "loss": 4.2631,
      "step": 19258
    },
    {
      "epoch": 0.7775131207105369,
      "grad_norm": 92.37178802490234,
      "learning_rate": 2.4859502197468975e-06,
      "loss": 4.2098,
      "step": 19259
    },
    {
      "epoch": 0.7775534921275736,
      "grad_norm": 91.83537292480469,
      "learning_rate": 2.4850874884213117e-06,
      "loss": 5.4443,
      "step": 19260
    },
    {
      "epoch": 0.7775938635446105,
      "grad_norm": 104.56205749511719,
      "learning_rate": 2.484224885583183e-06,
      "loss": 5.2747,
      "step": 19261
    },
    {
      "epoch": 0.7776342349616472,
      "grad_norm": 84.18220520019531,
      "learning_rate": 2.4833624112472556e-06,
      "loss": 2.5664,
      "step": 19262
    },
    {
      "epoch": 0.7776746063786839,
      "grad_norm": 90.91661071777344,
      "learning_rate": 2.482500065428278e-06,
      "loss": 6.4798,
      "step": 19263
    },
    {
      "epoch": 0.7777149777957206,
      "grad_norm": 127.93830871582031,
      "learning_rate": 2.481637848140995e-06,
      "loss": 4.1504,
      "step": 19264
    },
    {
      "epoch": 0.7777553492127574,
      "grad_norm": 142.4624786376953,
      "learning_rate": 2.4807757594001514e-06,
      "loss": 3.9627,
      "step": 19265
    },
    {
      "epoch": 0.7777957206297941,
      "grad_norm": 74.81239318847656,
      "learning_rate": 2.47991379922048e-06,
      "loss": 6.174,
      "step": 19266
    },
    {
      "epoch": 0.7778360920468308,
      "grad_norm": 89.02159118652344,
      "learning_rate": 2.479051967616722e-06,
      "loss": 5.4506,
      "step": 19267
    },
    {
      "epoch": 0.7778764634638676,
      "grad_norm": 90.38802337646484,
      "learning_rate": 2.478190264603614e-06,
      "loss": 4.203,
      "step": 19268
    },
    {
      "epoch": 0.7779168348809044,
      "grad_norm": 442.57269287109375,
      "learning_rate": 2.477328690195889e-06,
      "loss": 3.1272,
      "step": 19269
    },
    {
      "epoch": 0.7779572062979411,
      "grad_norm": 135.4778289794922,
      "learning_rate": 2.476467244408275e-06,
      "loss": 4.7654,
      "step": 19270
    },
    {
      "epoch": 0.7779975777149778,
      "grad_norm": 90.46234893798828,
      "learning_rate": 2.475605927255502e-06,
      "loss": 5.2882,
      "step": 19271
    },
    {
      "epoch": 0.7780379491320145,
      "grad_norm": 144.76119995117188,
      "learning_rate": 2.474744738752296e-06,
      "loss": 4.7503,
      "step": 19272
    },
    {
      "epoch": 0.7780783205490512,
      "grad_norm": 96.45369720458984,
      "learning_rate": 2.473883678913387e-06,
      "loss": 4.6688,
      "step": 19273
    },
    {
      "epoch": 0.778118691966088,
      "grad_norm": 107.35406494140625,
      "learning_rate": 2.4730227477534874e-06,
      "loss": 5.4983,
      "step": 19274
    },
    {
      "epoch": 0.7781590633831248,
      "grad_norm": 86.16718292236328,
      "learning_rate": 2.4721619452873235e-06,
      "loss": 5.1959,
      "step": 19275
    },
    {
      "epoch": 0.7781994348001615,
      "grad_norm": 83.54412841796875,
      "learning_rate": 2.4713012715296116e-06,
      "loss": 6.1255,
      "step": 19276
    },
    {
      "epoch": 0.7782398062171982,
      "grad_norm": 93.58325958251953,
      "learning_rate": 2.470440726495066e-06,
      "loss": 4.9738,
      "step": 19277
    },
    {
      "epoch": 0.778280177634235,
      "grad_norm": 77.48503112792969,
      "learning_rate": 2.4695803101984026e-06,
      "loss": 5.975,
      "step": 19278
    },
    {
      "epoch": 0.7783205490512717,
      "grad_norm": 136.01316833496094,
      "learning_rate": 2.4687200226543306e-06,
      "loss": 5.2541,
      "step": 19279
    },
    {
      "epoch": 0.7783609204683084,
      "grad_norm": 105.08331298828125,
      "learning_rate": 2.4678598638775586e-06,
      "loss": 5.8789,
      "step": 19280
    },
    {
      "epoch": 0.7784012918853451,
      "grad_norm": 163.10618591308594,
      "learning_rate": 2.4669998338827985e-06,
      "loss": 5.6475,
      "step": 19281
    },
    {
      "epoch": 0.778441663302382,
      "grad_norm": 136.60028076171875,
      "learning_rate": 2.4661399326847457e-06,
      "loss": 5.2812,
      "step": 19282
    },
    {
      "epoch": 0.7784820347194187,
      "grad_norm": 94.32093048095703,
      "learning_rate": 2.465280160298108e-06,
      "loss": 4.5325,
      "step": 19283
    },
    {
      "epoch": 0.7785224061364554,
      "grad_norm": 100.044189453125,
      "learning_rate": 2.4644205167375857e-06,
      "loss": 4.8007,
      "step": 19284
    },
    {
      "epoch": 0.7785627775534921,
      "grad_norm": 59.62934112548828,
      "learning_rate": 2.463561002017878e-06,
      "loss": 2.4587,
      "step": 19285
    },
    {
      "epoch": 0.7786031489705288,
      "grad_norm": 110.610595703125,
      "learning_rate": 2.4627016161536743e-06,
      "loss": 5.9493,
      "step": 19286
    },
    {
      "epoch": 0.7786435203875656,
      "grad_norm": 97.53663635253906,
      "learning_rate": 2.461842359159674e-06,
      "loss": 5.2097,
      "step": 19287
    },
    {
      "epoch": 0.7786838918046023,
      "grad_norm": 77.33258056640625,
      "learning_rate": 2.4609832310505654e-06,
      "loss": 4.7796,
      "step": 19288
    },
    {
      "epoch": 0.7787242632216391,
      "grad_norm": 92.54219818115234,
      "learning_rate": 2.4601242318410413e-06,
      "loss": 6.1086,
      "step": 19289
    },
    {
      "epoch": 0.7787646346386758,
      "grad_norm": 76.42054748535156,
      "learning_rate": 2.459265361545784e-06,
      "loss": 4.0216,
      "step": 19290
    },
    {
      "epoch": 0.7788050060557126,
      "grad_norm": 114.01819610595703,
      "learning_rate": 2.4584066201794786e-06,
      "loss": 4.1365,
      "step": 19291
    },
    {
      "epoch": 0.7788453774727493,
      "grad_norm": 83.32522583007812,
      "learning_rate": 2.4575480077568103e-06,
      "loss": 5.6859,
      "step": 19292
    },
    {
      "epoch": 0.778885748889786,
      "grad_norm": 88.98555755615234,
      "learning_rate": 2.4566895242924583e-06,
      "loss": 5.1565,
      "step": 19293
    },
    {
      "epoch": 0.7789261203068227,
      "grad_norm": 91.30939483642578,
      "learning_rate": 2.4558311698011006e-06,
      "loss": 4.9608,
      "step": 19294
    },
    {
      "epoch": 0.7789664917238595,
      "grad_norm": 75.20945739746094,
      "learning_rate": 2.4549729442974122e-06,
      "loss": 4.5167,
      "step": 19295
    },
    {
      "epoch": 0.7790068631408963,
      "grad_norm": 108.66744232177734,
      "learning_rate": 2.4541148477960688e-06,
      "loss": 4.8863,
      "step": 19296
    },
    {
      "epoch": 0.779047234557933,
      "grad_norm": 93.90121459960938,
      "learning_rate": 2.4532568803117417e-06,
      "loss": 6.3659,
      "step": 19297
    },
    {
      "epoch": 0.7790876059749697,
      "grad_norm": 89.65451049804688,
      "learning_rate": 2.452399041859097e-06,
      "loss": 5.4508,
      "step": 19298
    },
    {
      "epoch": 0.7791279773920065,
      "grad_norm": 115.85309600830078,
      "learning_rate": 2.451541332452805e-06,
      "loss": 3.8885,
      "step": 19299
    },
    {
      "epoch": 0.7791683488090432,
      "grad_norm": 66.1731185913086,
      "learning_rate": 2.450683752107528e-06,
      "loss": 3.4144,
      "step": 19300
    },
    {
      "epoch": 0.7792087202260799,
      "grad_norm": 126.73468017578125,
      "learning_rate": 2.449826300837933e-06,
      "loss": 5.7147,
      "step": 19301
    },
    {
      "epoch": 0.7792490916431166,
      "grad_norm": 107.6939926147461,
      "learning_rate": 2.4489689786586745e-06,
      "loss": 4.5645,
      "step": 19302
    },
    {
      "epoch": 0.7792894630601535,
      "grad_norm": 168.01158142089844,
      "learning_rate": 2.4481117855844127e-06,
      "loss": 5.7371,
      "step": 19303
    },
    {
      "epoch": 0.7793298344771902,
      "grad_norm": 121.01724243164062,
      "learning_rate": 2.447254721629806e-06,
      "loss": 6.1196,
      "step": 19304
    },
    {
      "epoch": 0.7793702058942269,
      "grad_norm": 102.49560546875,
      "learning_rate": 2.446397786809508e-06,
      "loss": 4.9134,
      "step": 19305
    },
    {
      "epoch": 0.7794105773112636,
      "grad_norm": 135.12950134277344,
      "learning_rate": 2.4455409811381668e-06,
      "loss": 5.6351,
      "step": 19306
    },
    {
      "epoch": 0.7794509487283003,
      "grad_norm": 109.34790802001953,
      "learning_rate": 2.4446843046304347e-06,
      "loss": 4.6215,
      "step": 19307
    },
    {
      "epoch": 0.7794913201453371,
      "grad_norm": 82.6393051147461,
      "learning_rate": 2.443827757300957e-06,
      "loss": 5.6102,
      "step": 19308
    },
    {
      "epoch": 0.7795316915623738,
      "grad_norm": 103.29341125488281,
      "learning_rate": 2.442971339164383e-06,
      "loss": 5.1159,
      "step": 19309
    },
    {
      "epoch": 0.7795720629794106,
      "grad_norm": 136.3611602783203,
      "learning_rate": 2.4421150502353486e-06,
      "loss": 5.203,
      "step": 19310
    },
    {
      "epoch": 0.7796124343964473,
      "grad_norm": 139.16993713378906,
      "learning_rate": 2.4412588905284986e-06,
      "loss": 3.8276,
      "step": 19311
    },
    {
      "epoch": 0.7796528058134841,
      "grad_norm": 65.4464340209961,
      "learning_rate": 2.4404028600584704e-06,
      "loss": 3.8519,
      "step": 19312
    },
    {
      "epoch": 0.7796931772305208,
      "grad_norm": 148.24517822265625,
      "learning_rate": 2.4395469588399e-06,
      "loss": 4.1744,
      "step": 19313
    },
    {
      "epoch": 0.7797335486475575,
      "grad_norm": 104.31578826904297,
      "learning_rate": 2.438691186887423e-06,
      "loss": 5.6852,
      "step": 19314
    },
    {
      "epoch": 0.7797739200645942,
      "grad_norm": 189.3644561767578,
      "learning_rate": 2.4378355442156685e-06,
      "loss": 4.5265,
      "step": 19315
    },
    {
      "epoch": 0.779814291481631,
      "grad_norm": 91.64038848876953,
      "learning_rate": 2.4369800308392677e-06,
      "loss": 6.0724,
      "step": 19316
    },
    {
      "epoch": 0.7798546628986678,
      "grad_norm": 72.33930206298828,
      "learning_rate": 2.4361246467728508e-06,
      "loss": 5.3102,
      "step": 19317
    },
    {
      "epoch": 0.7798950343157045,
      "grad_norm": 96.35675048828125,
      "learning_rate": 2.4352693920310354e-06,
      "loss": 4.6909,
      "step": 19318
    },
    {
      "epoch": 0.7799354057327412,
      "grad_norm": 115.38550567626953,
      "learning_rate": 2.4344142666284487e-06,
      "loss": 5.843,
      "step": 19319
    },
    {
      "epoch": 0.779975777149778,
      "grad_norm": 74.44692993164062,
      "learning_rate": 2.4335592705797116e-06,
      "loss": 5.3193,
      "step": 19320
    },
    {
      "epoch": 0.7800161485668147,
      "grad_norm": 84.10440826416016,
      "learning_rate": 2.4327044038994452e-06,
      "loss": 4.3432,
      "step": 19321
    },
    {
      "epoch": 0.7800565199838514,
      "grad_norm": 88.83965301513672,
      "learning_rate": 2.4318496666022596e-06,
      "loss": 6.4341,
      "step": 19322
    },
    {
      "epoch": 0.7800968914008882,
      "grad_norm": 135.6503448486328,
      "learning_rate": 2.430995058702772e-06,
      "loss": 5.6454,
      "step": 19323
    },
    {
      "epoch": 0.780137262817925,
      "grad_norm": 85.58521270751953,
      "learning_rate": 2.4301405802155932e-06,
      "loss": 4.6123,
      "step": 19324
    },
    {
      "epoch": 0.7801776342349617,
      "grad_norm": 107.48946380615234,
      "learning_rate": 2.4292862311553377e-06,
      "loss": 3.2254,
      "step": 19325
    },
    {
      "epoch": 0.7802180056519984,
      "grad_norm": 89.00237274169922,
      "learning_rate": 2.4284320115366044e-06,
      "loss": 6.8138,
      "step": 19326
    },
    {
      "epoch": 0.7802583770690351,
      "grad_norm": 83.01286315917969,
      "learning_rate": 2.4275779213740037e-06,
      "loss": 3.1836,
      "step": 19327
    },
    {
      "epoch": 0.7802987484860718,
      "grad_norm": 87.13201141357422,
      "learning_rate": 2.426723960682138e-06,
      "loss": 4.0794,
      "step": 19328
    },
    {
      "epoch": 0.7803391199031086,
      "grad_norm": 120.36272430419922,
      "learning_rate": 2.4258701294756106e-06,
      "loss": 5.499,
      "step": 19329
    },
    {
      "epoch": 0.7803794913201454,
      "grad_norm": 90.9664077758789,
      "learning_rate": 2.4250164277690124e-06,
      "loss": 3.6139,
      "step": 19330
    },
    {
      "epoch": 0.7804198627371821,
      "grad_norm": 151.53309631347656,
      "learning_rate": 2.424162855576946e-06,
      "loss": 4.7731,
      "step": 19331
    },
    {
      "epoch": 0.7804602341542188,
      "grad_norm": 81.49723052978516,
      "learning_rate": 2.4233094129140045e-06,
      "loss": 4.4198,
      "step": 19332
    },
    {
      "epoch": 0.7805006055712556,
      "grad_norm": 98.5841293334961,
      "learning_rate": 2.4224560997947776e-06,
      "loss": 5.3317,
      "step": 19333
    },
    {
      "epoch": 0.7805409769882923,
      "grad_norm": 91.85001373291016,
      "learning_rate": 2.421602916233857e-06,
      "loss": 4.4748,
      "step": 19334
    },
    {
      "epoch": 0.780581348405329,
      "grad_norm": 103.70654296875,
      "learning_rate": 2.4207498622458303e-06,
      "loss": 5.1231,
      "step": 19335
    },
    {
      "epoch": 0.7806217198223657,
      "grad_norm": 92.21241760253906,
      "learning_rate": 2.419896937845282e-06,
      "loss": 3.9713,
      "step": 19336
    },
    {
      "epoch": 0.7806620912394026,
      "grad_norm": 76.48741912841797,
      "learning_rate": 2.4190441430467983e-06,
      "loss": 4.0557,
      "step": 19337
    },
    {
      "epoch": 0.7807024626564393,
      "grad_norm": 87.23286437988281,
      "learning_rate": 2.418191477864954e-06,
      "loss": 4.6429,
      "step": 19338
    },
    {
      "epoch": 0.780742834073476,
      "grad_norm": 105.78620147705078,
      "learning_rate": 2.4173389423143303e-06,
      "loss": 4.5531,
      "step": 19339
    },
    {
      "epoch": 0.7807832054905127,
      "grad_norm": 96.69558715820312,
      "learning_rate": 2.4164865364095035e-06,
      "loss": 3.7206,
      "step": 19340
    },
    {
      "epoch": 0.7808235769075494,
      "grad_norm": 88.64210510253906,
      "learning_rate": 2.4156342601650516e-06,
      "loss": 4.8443,
      "step": 19341
    },
    {
      "epoch": 0.7808639483245862,
      "grad_norm": 104.10177612304688,
      "learning_rate": 2.4147821135955408e-06,
      "loss": 6.7481,
      "step": 19342
    },
    {
      "epoch": 0.7809043197416229,
      "grad_norm": 82.62474060058594,
      "learning_rate": 2.4139300967155423e-06,
      "loss": 5.6435,
      "step": 19343
    },
    {
      "epoch": 0.7809446911586597,
      "grad_norm": 109.01053619384766,
      "learning_rate": 2.4130782095396242e-06,
      "loss": 4.451,
      "step": 19344
    },
    {
      "epoch": 0.7809850625756964,
      "grad_norm": 72.38427734375,
      "learning_rate": 2.4122264520823555e-06,
      "loss": 5.3748,
      "step": 19345
    },
    {
      "epoch": 0.7810254339927332,
      "grad_norm": 95.84457397460938,
      "learning_rate": 2.4113748243582925e-06,
      "loss": 4.6133,
      "step": 19346
    },
    {
      "epoch": 0.7810658054097699,
      "grad_norm": 80.36327362060547,
      "learning_rate": 2.4105233263819993e-06,
      "loss": 4.2702,
      "step": 19347
    },
    {
      "epoch": 0.7811061768268066,
      "grad_norm": 74.8851547241211,
      "learning_rate": 2.409671958168035e-06,
      "loss": 5.0663,
      "step": 19348
    },
    {
      "epoch": 0.7811465482438433,
      "grad_norm": 87.22693634033203,
      "learning_rate": 2.4088207197309577e-06,
      "loss": 5.4892,
      "step": 19349
    },
    {
      "epoch": 0.78118691966088,
      "grad_norm": 84.82025909423828,
      "learning_rate": 2.4079696110853166e-06,
      "loss": 6.1265,
      "step": 19350
    },
    {
      "epoch": 0.7812272910779169,
      "grad_norm": 81.41191864013672,
      "learning_rate": 2.4071186322456663e-06,
      "loss": 6.6568,
      "step": 19351
    },
    {
      "epoch": 0.7812676624949536,
      "grad_norm": 128.98329162597656,
      "learning_rate": 2.406267783226558e-06,
      "loss": 4.0841,
      "step": 19352
    },
    {
      "epoch": 0.7813080339119903,
      "grad_norm": 129.82186889648438,
      "learning_rate": 2.4054170640425367e-06,
      "loss": 3.7616,
      "step": 19353
    },
    {
      "epoch": 0.781348405329027,
      "grad_norm": 155.83297729492188,
      "learning_rate": 2.40456647470815e-06,
      "loss": 4.0262,
      "step": 19354
    },
    {
      "epoch": 0.7813887767460638,
      "grad_norm": 73.66890716552734,
      "learning_rate": 2.403716015237939e-06,
      "loss": 5.8561,
      "step": 19355
    },
    {
      "epoch": 0.7814291481631005,
      "grad_norm": 119.42759704589844,
      "learning_rate": 2.402865685646446e-06,
      "loss": 5.0914,
      "step": 19356
    },
    {
      "epoch": 0.7814695195801372,
      "grad_norm": 86.05310821533203,
      "learning_rate": 2.4020154859482112e-06,
      "loss": 4.8752,
      "step": 19357
    },
    {
      "epoch": 0.781509890997174,
      "grad_norm": 97.18885803222656,
      "learning_rate": 2.401165416157767e-06,
      "loss": 2.9082,
      "step": 19358
    },
    {
      "epoch": 0.7815502624142108,
      "grad_norm": 123.03450012207031,
      "learning_rate": 2.4003154762896496e-06,
      "loss": 3.8729,
      "step": 19359
    },
    {
      "epoch": 0.7815906338312475,
      "grad_norm": 401.8789367675781,
      "learning_rate": 2.3994656663583913e-06,
      "loss": 6.1528,
      "step": 19360
    },
    {
      "epoch": 0.7816310052482842,
      "grad_norm": 91.70983123779297,
      "learning_rate": 2.398615986378524e-06,
      "loss": 4.8117,
      "step": 19361
    },
    {
      "epoch": 0.7816713766653209,
      "grad_norm": 103.64556121826172,
      "learning_rate": 2.39776643636457e-06,
      "loss": 3.973,
      "step": 19362
    },
    {
      "epoch": 0.7817117480823577,
      "grad_norm": 118.55928039550781,
      "learning_rate": 2.3969170163310575e-06,
      "loss": 4.7849,
      "step": 19363
    },
    {
      "epoch": 0.7817521194993944,
      "grad_norm": 87.73514556884766,
      "learning_rate": 2.396067726292509e-06,
      "loss": 4.1497,
      "step": 19364
    },
    {
      "epoch": 0.7817924909164312,
      "grad_norm": 200.28919982910156,
      "learning_rate": 2.39521856626345e-06,
      "loss": 5.7879,
      "step": 19365
    },
    {
      "epoch": 0.7818328623334679,
      "grad_norm": 88.42097473144531,
      "learning_rate": 2.394369536258392e-06,
      "loss": 5.0047,
      "step": 19366
    },
    {
      "epoch": 0.7818732337505047,
      "grad_norm": 83.08248901367188,
      "learning_rate": 2.393520636291854e-06,
      "loss": 4.5872,
      "step": 19367
    },
    {
      "epoch": 0.7819136051675414,
      "grad_norm": 103.04534912109375,
      "learning_rate": 2.39267186637835e-06,
      "loss": 5.6105,
      "step": 19368
    },
    {
      "epoch": 0.7819539765845781,
      "grad_norm": 94.95037841796875,
      "learning_rate": 2.391823226532396e-06,
      "loss": 4.9437,
      "step": 19369
    },
    {
      "epoch": 0.7819943480016148,
      "grad_norm": 87.44254302978516,
      "learning_rate": 2.390974716768496e-06,
      "loss": 5.2982,
      "step": 19370
    },
    {
      "epoch": 0.7820347194186515,
      "grad_norm": 107.59477233886719,
      "learning_rate": 2.3901263371011606e-06,
      "loss": 5.0001,
      "step": 19371
    },
    {
      "epoch": 0.7820750908356884,
      "grad_norm": 97.09319305419922,
      "learning_rate": 2.389278087544894e-06,
      "loss": 5.0243,
      "step": 19372
    },
    {
      "epoch": 0.7821154622527251,
      "grad_norm": 98.7015380859375,
      "learning_rate": 2.388429968114199e-06,
      "loss": 4.925,
      "step": 19373
    },
    {
      "epoch": 0.7821558336697618,
      "grad_norm": 102.43551635742188,
      "learning_rate": 2.3875819788235776e-06,
      "loss": 3.9269,
      "step": 19374
    },
    {
      "epoch": 0.7821962050867985,
      "grad_norm": 92.05709838867188,
      "learning_rate": 2.3867341196875283e-06,
      "loss": 3.8705,
      "step": 19375
    },
    {
      "epoch": 0.7822365765038353,
      "grad_norm": 94.16485595703125,
      "learning_rate": 2.3858863907205475e-06,
      "loss": 4.6771,
      "step": 19376
    },
    {
      "epoch": 0.782276947920872,
      "grad_norm": 137.83241271972656,
      "learning_rate": 2.385038791937132e-06,
      "loss": 4.4748,
      "step": 19377
    },
    {
      "epoch": 0.7823173193379087,
      "grad_norm": 102.87186431884766,
      "learning_rate": 2.3841913233517676e-06,
      "loss": 4.3413,
      "step": 19378
    },
    {
      "epoch": 0.7823576907549455,
      "grad_norm": 132.1216583251953,
      "learning_rate": 2.3833439849789486e-06,
      "loss": 4.2001,
      "step": 19379
    },
    {
      "epoch": 0.7823980621719823,
      "grad_norm": 98.84493255615234,
      "learning_rate": 2.3824967768331596e-06,
      "loss": 5.7089,
      "step": 19380
    },
    {
      "epoch": 0.782438433589019,
      "grad_norm": 71.86198425292969,
      "learning_rate": 2.3816496989288918e-06,
      "loss": 4.9944,
      "step": 19381
    },
    {
      "epoch": 0.7824788050060557,
      "grad_norm": 125.95067596435547,
      "learning_rate": 2.380802751280621e-06,
      "loss": 5.3617,
      "step": 19382
    },
    {
      "epoch": 0.7825191764230924,
      "grad_norm": 87.6925277709961,
      "learning_rate": 2.3799559339028298e-06,
      "loss": 5.9039,
      "step": 19383
    },
    {
      "epoch": 0.7825595478401292,
      "grad_norm": 84.82413482666016,
      "learning_rate": 2.3791092468099995e-06,
      "loss": 5.0928,
      "step": 19384
    },
    {
      "epoch": 0.7825999192571659,
      "grad_norm": 65.87251281738281,
      "learning_rate": 2.378262690016607e-06,
      "loss": 3.4778,
      "step": 19385
    },
    {
      "epoch": 0.7826402906742027,
      "grad_norm": 92.57394409179688,
      "learning_rate": 2.3774162635371223e-06,
      "loss": 3.3839,
      "step": 19386
    },
    {
      "epoch": 0.7826806620912394,
      "grad_norm": 92.34393310546875,
      "learning_rate": 2.3765699673860186e-06,
      "loss": 5.6292,
      "step": 19387
    },
    {
      "epoch": 0.7827210335082762,
      "grad_norm": 139.75270080566406,
      "learning_rate": 2.3757238015777673e-06,
      "loss": 4.4641,
      "step": 19388
    },
    {
      "epoch": 0.7827614049253129,
      "grad_norm": 111.04620361328125,
      "learning_rate": 2.374877766126837e-06,
      "loss": 6.5489,
      "step": 19389
    },
    {
      "epoch": 0.7828017763423496,
      "grad_norm": 103.90907287597656,
      "learning_rate": 2.3740318610476887e-06,
      "loss": 6.9236,
      "step": 19390
    },
    {
      "epoch": 0.7828421477593863,
      "grad_norm": 160.8944549560547,
      "learning_rate": 2.3731860863547884e-06,
      "loss": 4.1325,
      "step": 19391
    },
    {
      "epoch": 0.782882519176423,
      "grad_norm": 103.18616485595703,
      "learning_rate": 2.372340442062595e-06,
      "loss": 5.9028,
      "step": 19392
    },
    {
      "epoch": 0.7829228905934599,
      "grad_norm": 98.68602752685547,
      "learning_rate": 2.3714949281855683e-06,
      "loss": 5.1433,
      "step": 19393
    },
    {
      "epoch": 0.7829632620104966,
      "grad_norm": 119.26522064208984,
      "learning_rate": 2.370649544738165e-06,
      "loss": 4.7705,
      "step": 19394
    },
    {
      "epoch": 0.7830036334275333,
      "grad_norm": 118.61392211914062,
      "learning_rate": 2.3698042917348385e-06,
      "loss": 5.0355,
      "step": 19395
    },
    {
      "epoch": 0.78304400484457,
      "grad_norm": 86.14042663574219,
      "learning_rate": 2.3689591691900404e-06,
      "loss": 4.0021,
      "step": 19396
    },
    {
      "epoch": 0.7830843762616068,
      "grad_norm": 104.79826354980469,
      "learning_rate": 2.3681141771182237e-06,
      "loss": 6.0932,
      "step": 19397
    },
    {
      "epoch": 0.7831247476786435,
      "grad_norm": 126.75373077392578,
      "learning_rate": 2.36726931553383e-06,
      "loss": 5.401,
      "step": 19398
    },
    {
      "epoch": 0.7831651190956802,
      "grad_norm": 94.58580780029297,
      "learning_rate": 2.3664245844513078e-06,
      "loss": 4.2386,
      "step": 19399
    },
    {
      "epoch": 0.783205490512717,
      "grad_norm": 94.73677825927734,
      "learning_rate": 2.3655799838850992e-06,
      "loss": 3.9284,
      "step": 19400
    },
    {
      "epoch": 0.7832458619297538,
      "grad_norm": 82.51789093017578,
      "learning_rate": 2.3647355138496485e-06,
      "loss": 4.4965,
      "step": 19401
    },
    {
      "epoch": 0.7832862333467905,
      "grad_norm": 134.12188720703125,
      "learning_rate": 2.3638911743593875e-06,
      "loss": 5.2319,
      "step": 19402
    },
    {
      "epoch": 0.7833266047638272,
      "grad_norm": 145.3008575439453,
      "learning_rate": 2.363046965428756e-06,
      "loss": 5.9419,
      "step": 19403
    },
    {
      "epoch": 0.7833669761808639,
      "grad_norm": 153.67088317871094,
      "learning_rate": 2.362202887072187e-06,
      "loss": 4.9404,
      "step": 19404
    },
    {
      "epoch": 0.7834073475979006,
      "grad_norm": 83.83617401123047,
      "learning_rate": 2.3613589393041168e-06,
      "loss": 4.5577,
      "step": 19405
    },
    {
      "epoch": 0.7834477190149374,
      "grad_norm": 94.35250854492188,
      "learning_rate": 2.360515122138969e-06,
      "loss": 4.433,
      "step": 19406
    },
    {
      "epoch": 0.7834880904319742,
      "grad_norm": 103.83745574951172,
      "learning_rate": 2.359671435591172e-06,
      "loss": 5.5151,
      "step": 19407
    },
    {
      "epoch": 0.7835284618490109,
      "grad_norm": 103.2291488647461,
      "learning_rate": 2.358827879675152e-06,
      "loss": 5.4541,
      "step": 19408
    },
    {
      "epoch": 0.7835688332660476,
      "grad_norm": 94.4020767211914,
      "learning_rate": 2.3579844544053356e-06,
      "loss": 3.8855,
      "step": 19409
    },
    {
      "epoch": 0.7836092046830844,
      "grad_norm": 72.83444213867188,
      "learning_rate": 2.3571411597961358e-06,
      "loss": 4.2974,
      "step": 19410
    },
    {
      "epoch": 0.7836495761001211,
      "grad_norm": 90.00557708740234,
      "learning_rate": 2.3562979958619757e-06,
      "loss": 4.3391,
      "step": 19411
    },
    {
      "epoch": 0.7836899475171578,
      "grad_norm": 135.38150024414062,
      "learning_rate": 2.3554549626172694e-06,
      "loss": 5.3886,
      "step": 19412
    },
    {
      "epoch": 0.7837303189341945,
      "grad_norm": 77.5755386352539,
      "learning_rate": 2.3546120600764333e-06,
      "loss": 5.9577,
      "step": 19413
    },
    {
      "epoch": 0.7837706903512314,
      "grad_norm": 165.4624786376953,
      "learning_rate": 2.3537692882538776e-06,
      "loss": 6.1898,
      "step": 19414
    },
    {
      "epoch": 0.7838110617682681,
      "grad_norm": 127.7789077758789,
      "learning_rate": 2.352926647164011e-06,
      "loss": 6.6668,
      "step": 19415
    },
    {
      "epoch": 0.7838514331853048,
      "grad_norm": 110.45613098144531,
      "learning_rate": 2.352084136821242e-06,
      "loss": 5.1043,
      "step": 19416
    },
    {
      "epoch": 0.7838918046023415,
      "grad_norm": 89.4006118774414,
      "learning_rate": 2.351241757239977e-06,
      "loss": 2.7894,
      "step": 19417
    },
    {
      "epoch": 0.7839321760193783,
      "grad_norm": 97.59905242919922,
      "learning_rate": 2.3503995084346155e-06,
      "loss": 5.1375,
      "step": 19418
    },
    {
      "epoch": 0.783972547436415,
      "grad_norm": 104.69080352783203,
      "learning_rate": 2.349557390419558e-06,
      "loss": 4.9966,
      "step": 19419
    },
    {
      "epoch": 0.7840129188534517,
      "grad_norm": 126.3221206665039,
      "learning_rate": 2.3487154032092054e-06,
      "loss": 4.6374,
      "step": 19420
    },
    {
      "epoch": 0.7840532902704885,
      "grad_norm": 188.2510986328125,
      "learning_rate": 2.347873546817954e-06,
      "loss": 4.7133,
      "step": 19421
    },
    {
      "epoch": 0.7840936616875253,
      "grad_norm": 83.1041030883789,
      "learning_rate": 2.347031821260194e-06,
      "loss": 4.4419,
      "step": 19422
    },
    {
      "epoch": 0.784134033104562,
      "grad_norm": 90.15696716308594,
      "learning_rate": 2.3461902265503177e-06,
      "loss": 3.8593,
      "step": 19423
    },
    {
      "epoch": 0.7841744045215987,
      "grad_norm": 152.52716064453125,
      "learning_rate": 2.345348762702717e-06,
      "loss": 5.8685,
      "step": 19424
    },
    {
      "epoch": 0.7842147759386354,
      "grad_norm": 104.09303283691406,
      "learning_rate": 2.344507429731779e-06,
      "loss": 4.9584,
      "step": 19425
    },
    {
      "epoch": 0.7842551473556721,
      "grad_norm": 183.30813598632812,
      "learning_rate": 2.343666227651885e-06,
      "loss": 4.9774,
      "step": 19426
    },
    {
      "epoch": 0.784295518772709,
      "grad_norm": 108.3692398071289,
      "learning_rate": 2.3428251564774196e-06,
      "loss": 5.5146,
      "step": 19427
    },
    {
      "epoch": 0.7843358901897457,
      "grad_norm": 84.6540298461914,
      "learning_rate": 2.3419842162227634e-06,
      "loss": 4.2869,
      "step": 19428
    },
    {
      "epoch": 0.7843762616067824,
      "grad_norm": 103.4767837524414,
      "learning_rate": 2.3411434069022976e-06,
      "loss": 5.6595,
      "step": 19429
    },
    {
      "epoch": 0.7844166330238191,
      "grad_norm": 88.0323486328125,
      "learning_rate": 2.3403027285303913e-06,
      "loss": 3.8848,
      "step": 19430
    },
    {
      "epoch": 0.7844570044408559,
      "grad_norm": 122.89190673828125,
      "learning_rate": 2.3394621811214223e-06,
      "loss": 6.2921,
      "step": 19431
    },
    {
      "epoch": 0.7844973758578926,
      "grad_norm": 74.82921600341797,
      "learning_rate": 2.3386217646897602e-06,
      "loss": 4.8777,
      "step": 19432
    },
    {
      "epoch": 0.7845377472749293,
      "grad_norm": 136.0699920654297,
      "learning_rate": 2.3377814792497766e-06,
      "loss": 6.2598,
      "step": 19433
    },
    {
      "epoch": 0.7845781186919661,
      "grad_norm": 89.81340026855469,
      "learning_rate": 2.3369413248158367e-06,
      "loss": 4.2804,
      "step": 19434
    },
    {
      "epoch": 0.7846184901090029,
      "grad_norm": 110.58062744140625,
      "learning_rate": 2.336101301402305e-06,
      "loss": 5.396,
      "step": 19435
    },
    {
      "epoch": 0.7846588615260396,
      "grad_norm": 92.4283218383789,
      "learning_rate": 2.335261409023545e-06,
      "loss": 5.3683,
      "step": 19436
    },
    {
      "epoch": 0.7846992329430763,
      "grad_norm": 109.35614013671875,
      "learning_rate": 2.3344216476939198e-06,
      "loss": 3.994,
      "step": 19437
    },
    {
      "epoch": 0.784739604360113,
      "grad_norm": 94.22940826416016,
      "learning_rate": 2.3335820174277803e-06,
      "loss": 5.5793,
      "step": 19438
    },
    {
      "epoch": 0.7847799757771498,
      "grad_norm": 73.73445129394531,
      "learning_rate": 2.332742518239487e-06,
      "loss": 5.4783,
      "step": 19439
    },
    {
      "epoch": 0.7848203471941865,
      "grad_norm": 113.73442077636719,
      "learning_rate": 2.3319031501433907e-06,
      "loss": 7.5674,
      "step": 19440
    },
    {
      "epoch": 0.7848607186112233,
      "grad_norm": 144.33773803710938,
      "learning_rate": 2.331063913153848e-06,
      "loss": 6.2718,
      "step": 19441
    },
    {
      "epoch": 0.78490109002826,
      "grad_norm": 144.2947998046875,
      "learning_rate": 2.330224807285202e-06,
      "loss": 5.9211,
      "step": 19442
    },
    {
      "epoch": 0.7849414614452968,
      "grad_norm": 78.2944564819336,
      "learning_rate": 2.3293858325518004e-06,
      "loss": 5.0774,
      "step": 19443
    },
    {
      "epoch": 0.7849818328623335,
      "grad_norm": 100.9841079711914,
      "learning_rate": 2.32854698896799e-06,
      "loss": 6.7701,
      "step": 19444
    },
    {
      "epoch": 0.7850222042793702,
      "grad_norm": 124.83875274658203,
      "learning_rate": 2.327708276548114e-06,
      "loss": 6.4735,
      "step": 19445
    },
    {
      "epoch": 0.7850625756964069,
      "grad_norm": 163.0514678955078,
      "learning_rate": 2.3268696953065073e-06,
      "loss": 3.5272,
      "step": 19446
    },
    {
      "epoch": 0.7851029471134436,
      "grad_norm": 124.08783721923828,
      "learning_rate": 2.326031245257512e-06,
      "loss": 2.6335,
      "step": 19447
    },
    {
      "epoch": 0.7851433185304805,
      "grad_norm": 88.88045501708984,
      "learning_rate": 2.3251929264154605e-06,
      "loss": 5.5424,
      "step": 19448
    },
    {
      "epoch": 0.7851836899475172,
      "grad_norm": 126.78388977050781,
      "learning_rate": 2.324354738794692e-06,
      "loss": 4.8331,
      "step": 19449
    },
    {
      "epoch": 0.7852240613645539,
      "grad_norm": 90.97944641113281,
      "learning_rate": 2.323516682409531e-06,
      "loss": 4.4358,
      "step": 19450
    },
    {
      "epoch": 0.7852644327815906,
      "grad_norm": 71.89251708984375,
      "learning_rate": 2.322678757274308e-06,
      "loss": 5.3033,
      "step": 19451
    },
    {
      "epoch": 0.7853048041986274,
      "grad_norm": 101.67005157470703,
      "learning_rate": 2.321840963403351e-06,
      "loss": 4.409,
      "step": 19452
    },
    {
      "epoch": 0.7853451756156641,
      "grad_norm": 82.69095611572266,
      "learning_rate": 2.3210033008109845e-06,
      "loss": 6.1496,
      "step": 19453
    },
    {
      "epoch": 0.7853855470327008,
      "grad_norm": 93.89242553710938,
      "learning_rate": 2.3201657695115288e-06,
      "loss": 6.8315,
      "step": 19454
    },
    {
      "epoch": 0.7854259184497376,
      "grad_norm": 99.63716125488281,
      "learning_rate": 2.3193283695193047e-06,
      "loss": 6.0194,
      "step": 19455
    },
    {
      "epoch": 0.7854662898667744,
      "grad_norm": 84.0854721069336,
      "learning_rate": 2.31849110084863e-06,
      "loss": 5.4906,
      "step": 19456
    },
    {
      "epoch": 0.7855066612838111,
      "grad_norm": 70.78446197509766,
      "learning_rate": 2.3176539635138227e-06,
      "loss": 5.6828,
      "step": 19457
    },
    {
      "epoch": 0.7855470327008478,
      "grad_norm": 133.67291259765625,
      "learning_rate": 2.3168169575291898e-06,
      "loss": 4.4733,
      "step": 19458
    },
    {
      "epoch": 0.7855874041178845,
      "grad_norm": 89.53968048095703,
      "learning_rate": 2.3159800829090462e-06,
      "loss": 3.2652,
      "step": 19459
    },
    {
      "epoch": 0.7856277755349212,
      "grad_norm": 114.43378448486328,
      "learning_rate": 2.3151433396676994e-06,
      "loss": 4.7865,
      "step": 19460
    },
    {
      "epoch": 0.785668146951958,
      "grad_norm": 87.69220733642578,
      "learning_rate": 2.3143067278194586e-06,
      "loss": 4.5364,
      "step": 19461
    },
    {
      "epoch": 0.7857085183689948,
      "grad_norm": 75.7619857788086,
      "learning_rate": 2.313470247378623e-06,
      "loss": 4.5003,
      "step": 19462
    },
    {
      "epoch": 0.7857488897860315,
      "grad_norm": 115.48193359375,
      "learning_rate": 2.3126338983594965e-06,
      "loss": 4.0835,
      "step": 19463
    },
    {
      "epoch": 0.7857892612030682,
      "grad_norm": 112.32337188720703,
      "learning_rate": 2.3117976807763786e-06,
      "loss": 4.4549,
      "step": 19464
    },
    {
      "epoch": 0.785829632620105,
      "grad_norm": 101.01607513427734,
      "learning_rate": 2.3109615946435716e-06,
      "loss": 4.714,
      "step": 19465
    },
    {
      "epoch": 0.7858700040371417,
      "grad_norm": 122.25311279296875,
      "learning_rate": 2.3101256399753615e-06,
      "loss": 4.7511,
      "step": 19466
    },
    {
      "epoch": 0.7859103754541784,
      "grad_norm": 112.12123107910156,
      "learning_rate": 2.3092898167860465e-06,
      "loss": 4.3882,
      "step": 19467
    },
    {
      "epoch": 0.7859507468712151,
      "grad_norm": 82.75230407714844,
      "learning_rate": 2.3084541250899163e-06,
      "loss": 4.3569,
      "step": 19468
    },
    {
      "epoch": 0.785991118288252,
      "grad_norm": 83.00931549072266,
      "learning_rate": 2.3076185649012627e-06,
      "loss": 4.7753,
      "step": 19469
    },
    {
      "epoch": 0.7860314897052887,
      "grad_norm": 94.97897338867188,
      "learning_rate": 2.3067831362343663e-06,
      "loss": 4.4601,
      "step": 19470
    },
    {
      "epoch": 0.7860718611223254,
      "grad_norm": 84.8256607055664,
      "learning_rate": 2.3059478391035127e-06,
      "loss": 4.392,
      "step": 19471
    },
    {
      "epoch": 0.7861122325393621,
      "grad_norm": 82.68346405029297,
      "learning_rate": 2.3051126735229834e-06,
      "loss": 5.1239,
      "step": 19472
    },
    {
      "epoch": 0.7861526039563989,
      "grad_norm": 123.48751068115234,
      "learning_rate": 2.3042776395070598e-06,
      "loss": 4.6442,
      "step": 19473
    },
    {
      "epoch": 0.7861929753734356,
      "grad_norm": 118.63872528076172,
      "learning_rate": 2.3034427370700176e-06,
      "loss": 4.5218,
      "step": 19474
    },
    {
      "epoch": 0.7862333467904723,
      "grad_norm": 101.08978271484375,
      "learning_rate": 2.302607966226131e-06,
      "loss": 5.1915,
      "step": 19475
    },
    {
      "epoch": 0.7862737182075091,
      "grad_norm": 71.57938385009766,
      "learning_rate": 2.3017733269896734e-06,
      "loss": 5.5542,
      "step": 19476
    },
    {
      "epoch": 0.7863140896245459,
      "grad_norm": 78.54586791992188,
      "learning_rate": 2.3009388193749194e-06,
      "loss": 4.2296,
      "step": 19477
    },
    {
      "epoch": 0.7863544610415826,
      "grad_norm": 147.88499450683594,
      "learning_rate": 2.3001044433961293e-06,
      "loss": 6.6083,
      "step": 19478
    },
    {
      "epoch": 0.7863948324586193,
      "grad_norm": 620.8251342773438,
      "learning_rate": 2.2992701990675726e-06,
      "loss": 3.879,
      "step": 19479
    },
    {
      "epoch": 0.786435203875656,
      "grad_norm": 146.02963256835938,
      "learning_rate": 2.298436086403514e-06,
      "loss": 5.328,
      "step": 19480
    },
    {
      "epoch": 0.7864755752926927,
      "grad_norm": 102.18342590332031,
      "learning_rate": 2.297602105418215e-06,
      "loss": 5.282,
      "step": 19481
    },
    {
      "epoch": 0.7865159467097295,
      "grad_norm": 135.59446716308594,
      "learning_rate": 2.296768256125932e-06,
      "loss": 4.6971,
      "step": 19482
    },
    {
      "epoch": 0.7865563181267663,
      "grad_norm": 83.33187866210938,
      "learning_rate": 2.2959345385409225e-06,
      "loss": 4.5949,
      "step": 19483
    },
    {
      "epoch": 0.786596689543803,
      "grad_norm": 78.34803009033203,
      "learning_rate": 2.2951009526774437e-06,
      "loss": 3.7016,
      "step": 19484
    },
    {
      "epoch": 0.7866370609608397,
      "grad_norm": 74.21683502197266,
      "learning_rate": 2.294267498549745e-06,
      "loss": 3.5907,
      "step": 19485
    },
    {
      "epoch": 0.7866774323778765,
      "grad_norm": 108.03353118896484,
      "learning_rate": 2.2934341761720813e-06,
      "loss": 5.4422,
      "step": 19486
    },
    {
      "epoch": 0.7867178037949132,
      "grad_norm": 77.48036193847656,
      "learning_rate": 2.292600985558694e-06,
      "loss": 5.274,
      "step": 19487
    },
    {
      "epoch": 0.7867581752119499,
      "grad_norm": 93.87657165527344,
      "learning_rate": 2.2917679267238313e-06,
      "loss": 5.9437,
      "step": 19488
    },
    {
      "epoch": 0.7867985466289866,
      "grad_norm": 101.35819244384766,
      "learning_rate": 2.2909349996817374e-06,
      "loss": 6.2142,
      "step": 19489
    },
    {
      "epoch": 0.7868389180460235,
      "grad_norm": 189.6416778564453,
      "learning_rate": 2.2901022044466534e-06,
      "loss": 3.979,
      "step": 19490
    },
    {
      "epoch": 0.7868792894630602,
      "grad_norm": 80.15281677246094,
      "learning_rate": 2.289269541032818e-06,
      "loss": 5.0724,
      "step": 19491
    },
    {
      "epoch": 0.7869196608800969,
      "grad_norm": 98.28810119628906,
      "learning_rate": 2.2884370094544707e-06,
      "loss": 5.1847,
      "step": 19492
    },
    {
      "epoch": 0.7869600322971336,
      "grad_norm": 125.80240631103516,
      "learning_rate": 2.2876046097258386e-06,
      "loss": 4.6673,
      "step": 19493
    },
    {
      "epoch": 0.7870004037141703,
      "grad_norm": 100.06868743896484,
      "learning_rate": 2.2867723418611587e-06,
      "loss": 5.245,
      "step": 19494
    },
    {
      "epoch": 0.7870407751312071,
      "grad_norm": 88.5499038696289,
      "learning_rate": 2.2859402058746607e-06,
      "loss": 5.2697,
      "step": 19495
    },
    {
      "epoch": 0.7870811465482438,
      "grad_norm": 92.61017608642578,
      "learning_rate": 2.2851082017805704e-06,
      "loss": 4.4049,
      "step": 19496
    },
    {
      "epoch": 0.7871215179652806,
      "grad_norm": 87.80203247070312,
      "learning_rate": 2.284276329593115e-06,
      "loss": 4.9132,
      "step": 19497
    },
    {
      "epoch": 0.7871618893823173,
      "grad_norm": 90.66165161132812,
      "learning_rate": 2.283444589326519e-06,
      "loss": 5.259,
      "step": 19498
    },
    {
      "epoch": 0.7872022607993541,
      "grad_norm": 122.6019515991211,
      "learning_rate": 2.282612980994998e-06,
      "loss": 5.964,
      "step": 19499
    },
    {
      "epoch": 0.7872426322163908,
      "grad_norm": 68.44367980957031,
      "learning_rate": 2.2817815046127743e-06,
      "loss": 3.7791,
      "step": 19500
    },
    {
      "epoch": 0.7872830036334275,
      "grad_norm": 106.87040710449219,
      "learning_rate": 2.280950160194064e-06,
      "loss": 4.2875,
      "step": 19501
    },
    {
      "epoch": 0.7873233750504642,
      "grad_norm": 96.68707275390625,
      "learning_rate": 2.280118947753083e-06,
      "loss": 4.3621,
      "step": 19502
    },
    {
      "epoch": 0.787363746467501,
      "grad_norm": 90.64717102050781,
      "learning_rate": 2.2792878673040375e-06,
      "loss": 5.5817,
      "step": 19503
    },
    {
      "epoch": 0.7874041178845378,
      "grad_norm": 96.30323791503906,
      "learning_rate": 2.2784569188611416e-06,
      "loss": 5.8765,
      "step": 19504
    },
    {
      "epoch": 0.7874444893015745,
      "grad_norm": 68.44245147705078,
      "learning_rate": 2.2776261024386005e-06,
      "loss": 4.9364,
      "step": 19505
    },
    {
      "epoch": 0.7874848607186112,
      "grad_norm": 118.72911834716797,
      "learning_rate": 2.2767954180506226e-06,
      "loss": 3.6101,
      "step": 19506
    },
    {
      "epoch": 0.787525232135648,
      "grad_norm": 123.87041473388672,
      "learning_rate": 2.2759648657114054e-06,
      "loss": 6.0167,
      "step": 19507
    },
    {
      "epoch": 0.7875656035526847,
      "grad_norm": 88.70125579833984,
      "learning_rate": 2.275134445435152e-06,
      "loss": 6.0244,
      "step": 19508
    },
    {
      "epoch": 0.7876059749697214,
      "grad_norm": 78.30506134033203,
      "learning_rate": 2.2743041572360614e-06,
      "loss": 6.0072,
      "step": 19509
    },
    {
      "epoch": 0.7876463463867581,
      "grad_norm": 97.47463989257812,
      "learning_rate": 2.2734740011283276e-06,
      "loss": 5.4111,
      "step": 19510
    },
    {
      "epoch": 0.787686717803795,
      "grad_norm": 101.35163879394531,
      "learning_rate": 2.2726439771261467e-06,
      "loss": 4.8137,
      "step": 19511
    },
    {
      "epoch": 0.7877270892208317,
      "grad_norm": 103.61961364746094,
      "learning_rate": 2.271814085243709e-06,
      "loss": 5.8926,
      "step": 19512
    },
    {
      "epoch": 0.7877674606378684,
      "grad_norm": 153.5477294921875,
      "learning_rate": 2.2709843254952036e-06,
      "loss": 5.8676,
      "step": 19513
    },
    {
      "epoch": 0.7878078320549051,
      "grad_norm": 93.19094848632812,
      "learning_rate": 2.2701546978948206e-06,
      "loss": 3.8513,
      "step": 19514
    },
    {
      "epoch": 0.7878482034719418,
      "grad_norm": 89.16609191894531,
      "learning_rate": 2.269325202456739e-06,
      "loss": 4.8339,
      "step": 19515
    },
    {
      "epoch": 0.7878885748889786,
      "grad_norm": 93.1484375,
      "learning_rate": 2.268495839195144e-06,
      "loss": 4.5973,
      "step": 19516
    },
    {
      "epoch": 0.7879289463060153,
      "grad_norm": 167.86178588867188,
      "learning_rate": 2.2676666081242147e-06,
      "loss": 4.6509,
      "step": 19517
    },
    {
      "epoch": 0.7879693177230521,
      "grad_norm": 106.74365997314453,
      "learning_rate": 2.266837509258135e-06,
      "loss": 4.0309,
      "step": 19518
    },
    {
      "epoch": 0.7880096891400888,
      "grad_norm": 145.22032165527344,
      "learning_rate": 2.266008542611071e-06,
      "loss": 3.3907,
      "step": 19519
    },
    {
      "epoch": 0.7880500605571256,
      "grad_norm": 143.62335205078125,
      "learning_rate": 2.2651797081972016e-06,
      "loss": 4.0689,
      "step": 19520
    },
    {
      "epoch": 0.7880904319741623,
      "grad_norm": 95.70297241210938,
      "learning_rate": 2.264351006030697e-06,
      "loss": 2.5718,
      "step": 19521
    },
    {
      "epoch": 0.788130803391199,
      "grad_norm": 266.05181884765625,
      "learning_rate": 2.263522436125729e-06,
      "loss": 6.0365,
      "step": 19522
    },
    {
      "epoch": 0.7881711748082357,
      "grad_norm": 96.07327270507812,
      "learning_rate": 2.262693998496459e-06,
      "loss": 3.848,
      "step": 19523
    },
    {
      "epoch": 0.7882115462252725,
      "grad_norm": 95.67788696289062,
      "learning_rate": 2.2618656931570537e-06,
      "loss": 3.2559,
      "step": 19524
    },
    {
      "epoch": 0.7882519176423093,
      "grad_norm": 84.7958984375,
      "learning_rate": 2.261037520121674e-06,
      "loss": 3.381,
      "step": 19525
    },
    {
      "epoch": 0.788292289059346,
      "grad_norm": 108.06792449951172,
      "learning_rate": 2.260209479404485e-06,
      "loss": 5.3467,
      "step": 19526
    },
    {
      "epoch": 0.7883326604763827,
      "grad_norm": 73.20597076416016,
      "learning_rate": 2.2593815710196364e-06,
      "loss": 5.7669,
      "step": 19527
    },
    {
      "epoch": 0.7883730318934195,
      "grad_norm": 72.49542236328125,
      "learning_rate": 2.2585537949812887e-06,
      "loss": 5.6156,
      "step": 19528
    },
    {
      "epoch": 0.7884134033104562,
      "grad_norm": 94.43667602539062,
      "learning_rate": 2.257726151303592e-06,
      "loss": 6.4546,
      "step": 19529
    },
    {
      "epoch": 0.7884537747274929,
      "grad_norm": 127.83448791503906,
      "learning_rate": 2.2568986400006997e-06,
      "loss": 5.9377,
      "step": 19530
    },
    {
      "epoch": 0.7884941461445297,
      "grad_norm": 74.58174896240234,
      "learning_rate": 2.2560712610867588e-06,
      "loss": 5.3689,
      "step": 19531
    },
    {
      "epoch": 0.7885345175615665,
      "grad_norm": 105.6476058959961,
      "learning_rate": 2.2552440145759157e-06,
      "loss": 5.4203,
      "step": 19532
    },
    {
      "epoch": 0.7885748889786032,
      "grad_norm": 100.85969543457031,
      "learning_rate": 2.254416900482315e-06,
      "loss": 3.8637,
      "step": 19533
    },
    {
      "epoch": 0.7886152603956399,
      "grad_norm": 233.50787353515625,
      "learning_rate": 2.2535899188201007e-06,
      "loss": 3.8531,
      "step": 19534
    },
    {
      "epoch": 0.7886556318126766,
      "grad_norm": 91.11548614501953,
      "learning_rate": 2.2527630696034064e-06,
      "loss": 4.3653,
      "step": 19535
    },
    {
      "epoch": 0.7886960032297133,
      "grad_norm": 135.83164978027344,
      "learning_rate": 2.2519363528463732e-06,
      "loss": 3.2689,
      "step": 19536
    },
    {
      "epoch": 0.7887363746467501,
      "grad_norm": 105.45420837402344,
      "learning_rate": 2.2511097685631344e-06,
      "loss": 5.9233,
      "step": 19537
    },
    {
      "epoch": 0.7887767460637869,
      "grad_norm": 64.4628677368164,
      "learning_rate": 2.2502833167678272e-06,
      "loss": 6.7581,
      "step": 19538
    },
    {
      "epoch": 0.7888171174808236,
      "grad_norm": 77.49232482910156,
      "learning_rate": 2.249456997474576e-06,
      "loss": 5.9706,
      "step": 19539
    },
    {
      "epoch": 0.7888574888978603,
      "grad_norm": 79.72754669189453,
      "learning_rate": 2.2486308106975106e-06,
      "loss": 4.9131,
      "step": 19540
    },
    {
      "epoch": 0.7888978603148971,
      "grad_norm": 73.32881927490234,
      "learning_rate": 2.2478047564507577e-06,
      "loss": 5.4089,
      "step": 19541
    },
    {
      "epoch": 0.7889382317319338,
      "grad_norm": 81.98989868164062,
      "learning_rate": 2.2469788347484433e-06,
      "loss": 3.5752,
      "step": 19542
    },
    {
      "epoch": 0.7889786031489705,
      "grad_norm": 83.80503845214844,
      "learning_rate": 2.246153045604684e-06,
      "loss": 3.299,
      "step": 19543
    },
    {
      "epoch": 0.7890189745660072,
      "grad_norm": 97.76366424560547,
      "learning_rate": 2.2453273890336014e-06,
      "loss": 4.8086,
      "step": 19544
    },
    {
      "epoch": 0.7890593459830441,
      "grad_norm": 96.14582061767578,
      "learning_rate": 2.244501865049311e-06,
      "loss": 3.6634,
      "step": 19545
    },
    {
      "epoch": 0.7890997174000808,
      "grad_norm": 81.94551849365234,
      "learning_rate": 2.243676473665932e-06,
      "loss": 4.7134,
      "step": 19546
    },
    {
      "epoch": 0.7891400888171175,
      "grad_norm": 90.54621124267578,
      "learning_rate": 2.24285121489757e-06,
      "loss": 4.97,
      "step": 19547
    },
    {
      "epoch": 0.7891804602341542,
      "grad_norm": 156.59718322753906,
      "learning_rate": 2.2420260887583377e-06,
      "loss": 3.5692,
      "step": 19548
    },
    {
      "epoch": 0.789220831651191,
      "grad_norm": 94.38896179199219,
      "learning_rate": 2.241201095262343e-06,
      "loss": 4.4796,
      "step": 19549
    },
    {
      "epoch": 0.7892612030682277,
      "grad_norm": 127.31207275390625,
      "learning_rate": 2.240376234423691e-06,
      "loss": 3.7938,
      "step": 19550
    },
    {
      "epoch": 0.7893015744852644,
      "grad_norm": 105.16236877441406,
      "learning_rate": 2.2395515062564864e-06,
      "loss": 3.825,
      "step": 19551
    },
    {
      "epoch": 0.7893419459023012,
      "grad_norm": 113.83074951171875,
      "learning_rate": 2.2387269107748287e-06,
      "loss": 4.1902,
      "step": 19552
    },
    {
      "epoch": 0.789382317319338,
      "grad_norm": 104.73888397216797,
      "learning_rate": 2.237902447992817e-06,
      "loss": 5.1897,
      "step": 19553
    },
    {
      "epoch": 0.7894226887363747,
      "grad_norm": 132.21343994140625,
      "learning_rate": 2.2370781179245503e-06,
      "loss": 4.6596,
      "step": 19554
    },
    {
      "epoch": 0.7894630601534114,
      "grad_norm": 85.47183990478516,
      "learning_rate": 2.236253920584117e-06,
      "loss": 5.0656,
      "step": 19555
    },
    {
      "epoch": 0.7895034315704481,
      "grad_norm": 85.02385711669922,
      "learning_rate": 2.2354298559856124e-06,
      "loss": 5.9301,
      "step": 19556
    },
    {
      "epoch": 0.7895438029874848,
      "grad_norm": 123.8359375,
      "learning_rate": 2.234605924143126e-06,
      "loss": 4.8415,
      "step": 19557
    },
    {
      "epoch": 0.7895841744045216,
      "grad_norm": 92.78052520751953,
      "learning_rate": 2.233782125070747e-06,
      "loss": 6.3612,
      "step": 19558
    },
    {
      "epoch": 0.7896245458215584,
      "grad_norm": 161.2150115966797,
      "learning_rate": 2.232958458782556e-06,
      "loss": 5.64,
      "step": 19559
    },
    {
      "epoch": 0.7896649172385951,
      "grad_norm": 106.49654388427734,
      "learning_rate": 2.2321349252926373e-06,
      "loss": 6.5517,
      "step": 19560
    },
    {
      "epoch": 0.7897052886556318,
      "grad_norm": 74.557373046875,
      "learning_rate": 2.231311524615073e-06,
      "loss": 3.9164,
      "step": 19561
    },
    {
      "epoch": 0.7897456600726686,
      "grad_norm": 86.88783264160156,
      "learning_rate": 2.230488256763943e-06,
      "loss": 4.7956,
      "step": 19562
    },
    {
      "epoch": 0.7897860314897053,
      "grad_norm": 92.73299407958984,
      "learning_rate": 2.2296651217533172e-06,
      "loss": 3.7182,
      "step": 19563
    },
    {
      "epoch": 0.789826402906742,
      "grad_norm": 102.2027587890625,
      "learning_rate": 2.228842119597273e-06,
      "loss": 3.5453,
      "step": 19564
    },
    {
      "epoch": 0.7898667743237787,
      "grad_norm": 101.3669204711914,
      "learning_rate": 2.2280192503098818e-06,
      "loss": 5.1304,
      "step": 19565
    },
    {
      "epoch": 0.7899071457408156,
      "grad_norm": 121.14636993408203,
      "learning_rate": 2.227196513905215e-06,
      "loss": 5.9905,
      "step": 19566
    },
    {
      "epoch": 0.7899475171578523,
      "grad_norm": 117.17819213867188,
      "learning_rate": 2.2263739103973347e-06,
      "loss": 4.3046,
      "step": 19567
    },
    {
      "epoch": 0.789987888574889,
      "grad_norm": 97.41657257080078,
      "learning_rate": 2.2255514398003077e-06,
      "loss": 5.5702,
      "step": 19568
    },
    {
      "epoch": 0.7900282599919257,
      "grad_norm": 158.9696502685547,
      "learning_rate": 2.224729102128197e-06,
      "loss": 5.8232,
      "step": 19569
    },
    {
      "epoch": 0.7900686314089624,
      "grad_norm": 88.53515625,
      "learning_rate": 2.2239068973950608e-06,
      "loss": 5.2689,
      "step": 19570
    },
    {
      "epoch": 0.7901090028259992,
      "grad_norm": 122.19959259033203,
      "learning_rate": 2.2230848256149593e-06,
      "loss": 6.6661,
      "step": 19571
    },
    {
      "epoch": 0.7901493742430359,
      "grad_norm": 92.3643569946289,
      "learning_rate": 2.222262886801946e-06,
      "loss": 4.6013,
      "step": 19572
    },
    {
      "epoch": 0.7901897456600727,
      "grad_norm": 147.3372802734375,
      "learning_rate": 2.221441080970076e-06,
      "loss": 3.3997,
      "step": 19573
    },
    {
      "epoch": 0.7902301170771094,
      "grad_norm": 79.67024230957031,
      "learning_rate": 2.220619408133401e-06,
      "loss": 5.8113,
      "step": 19574
    },
    {
      "epoch": 0.7902704884941462,
      "grad_norm": 105.39678192138672,
      "learning_rate": 2.2197978683059663e-06,
      "loss": 5.9165,
      "step": 19575
    },
    {
      "epoch": 0.7903108599111829,
      "grad_norm": 75.2393569946289,
      "learning_rate": 2.2189764615018195e-06,
      "loss": 5.1005,
      "step": 19576
    },
    {
      "epoch": 0.7903512313282196,
      "grad_norm": 80.70845031738281,
      "learning_rate": 2.2181551877350048e-06,
      "loss": 4.9067,
      "step": 19577
    },
    {
      "epoch": 0.7903916027452563,
      "grad_norm": 95.32295989990234,
      "learning_rate": 2.217334047019567e-06,
      "loss": 6.1455,
      "step": 19578
    },
    {
      "epoch": 0.790431974162293,
      "grad_norm": 151.78973388671875,
      "learning_rate": 2.2165130393695413e-06,
      "loss": 4.3735,
      "step": 19579
    },
    {
      "epoch": 0.7904723455793299,
      "grad_norm": 100.42428588867188,
      "learning_rate": 2.215692164798966e-06,
      "loss": 4.0348,
      "step": 19580
    },
    {
      "epoch": 0.7905127169963666,
      "grad_norm": 83.06370544433594,
      "learning_rate": 2.2148714233218784e-06,
      "loss": 4.7988,
      "step": 19581
    },
    {
      "epoch": 0.7905530884134033,
      "grad_norm": 120.14512634277344,
      "learning_rate": 2.214050814952311e-06,
      "loss": 4.9442,
      "step": 19582
    },
    {
      "epoch": 0.79059345983044,
      "grad_norm": 128.0951385498047,
      "learning_rate": 2.213230339704291e-06,
      "loss": 4.5347,
      "step": 19583
    },
    {
      "epoch": 0.7906338312474768,
      "grad_norm": 70.21984100341797,
      "learning_rate": 2.212409997591849e-06,
      "loss": 4.4022,
      "step": 19584
    },
    {
      "epoch": 0.7906742026645135,
      "grad_norm": 94.65347290039062,
      "learning_rate": 2.2115897886290104e-06,
      "loss": 5.2621,
      "step": 19585
    },
    {
      "epoch": 0.7907145740815502,
      "grad_norm": 90.73651123046875,
      "learning_rate": 2.210769712829802e-06,
      "loss": 3.9339,
      "step": 19586
    },
    {
      "epoch": 0.790754945498587,
      "grad_norm": 146.35704040527344,
      "learning_rate": 2.2099497702082406e-06,
      "loss": 5.4645,
      "step": 19587
    },
    {
      "epoch": 0.7907953169156238,
      "grad_norm": 94.60655212402344,
      "learning_rate": 2.2091299607783458e-06,
      "loss": 7.5334,
      "step": 19588
    },
    {
      "epoch": 0.7908356883326605,
      "grad_norm": 102.94526672363281,
      "learning_rate": 2.208310284554136e-06,
      "loss": 6.2259,
      "step": 19589
    },
    {
      "epoch": 0.7908760597496972,
      "grad_norm": 118.8542251586914,
      "learning_rate": 2.207490741549626e-06,
      "loss": 5.2665,
      "step": 19590
    },
    {
      "epoch": 0.7909164311667339,
      "grad_norm": 84.6368637084961,
      "learning_rate": 2.2066713317788267e-06,
      "loss": 5.3842,
      "step": 19591
    },
    {
      "epoch": 0.7909568025837707,
      "grad_norm": 102.59358978271484,
      "learning_rate": 2.2058520552557495e-06,
      "loss": 4.611,
      "step": 19592
    },
    {
      "epoch": 0.7909971740008074,
      "grad_norm": 80.55184173583984,
      "learning_rate": 2.2050329119944016e-06,
      "loss": 4.4243,
      "step": 19593
    },
    {
      "epoch": 0.7910375454178442,
      "grad_norm": 112.5670394897461,
      "learning_rate": 2.2042139020087903e-06,
      "loss": 5.149,
      "step": 19594
    },
    {
      "epoch": 0.7910779168348809,
      "grad_norm": 92.6130142211914,
      "learning_rate": 2.2033950253129143e-06,
      "loss": 4.6137,
      "step": 19595
    },
    {
      "epoch": 0.7911182882519177,
      "grad_norm": 98.21839904785156,
      "learning_rate": 2.2025762819207775e-06,
      "loss": 3.6346,
      "step": 19596
    },
    {
      "epoch": 0.7911586596689544,
      "grad_norm": 81.0025405883789,
      "learning_rate": 2.2017576718463775e-06,
      "loss": 6.9026,
      "step": 19597
    },
    {
      "epoch": 0.7911990310859911,
      "grad_norm": 85.69610595703125,
      "learning_rate": 2.200939195103714e-06,
      "loss": 5.1826,
      "step": 19598
    },
    {
      "epoch": 0.7912394025030278,
      "grad_norm": 120.40605926513672,
      "learning_rate": 2.200120851706775e-06,
      "loss": 5.0842,
      "step": 19599
    },
    {
      "epoch": 0.7912797739200645,
      "grad_norm": 93.228759765625,
      "learning_rate": 2.1993026416695563e-06,
      "loss": 5.8329,
      "step": 19600
    },
    {
      "epoch": 0.7913201453371014,
      "grad_norm": 102.79473876953125,
      "learning_rate": 2.198484565006046e-06,
      "loss": 5.2249,
      "step": 19601
    },
    {
      "epoch": 0.7913605167541381,
      "grad_norm": 74.94233703613281,
      "learning_rate": 2.1976666217302344e-06,
      "loss": 6.0204,
      "step": 19602
    },
    {
      "epoch": 0.7914008881711748,
      "grad_norm": 119.33772277832031,
      "learning_rate": 2.1968488118561017e-06,
      "loss": 4.7682,
      "step": 19603
    },
    {
      "epoch": 0.7914412595882115,
      "grad_norm": 70.53060913085938,
      "learning_rate": 2.1960311353976318e-06,
      "loss": 5.2834,
      "step": 19604
    },
    {
      "epoch": 0.7914816310052483,
      "grad_norm": 78.53620910644531,
      "learning_rate": 2.1952135923688065e-06,
      "loss": 4.0809,
      "step": 19605
    },
    {
      "epoch": 0.791522002422285,
      "grad_norm": 125.39649200439453,
      "learning_rate": 2.194396182783606e-06,
      "loss": 5.4592,
      "step": 19606
    },
    {
      "epoch": 0.7915623738393217,
      "grad_norm": 251.93759155273438,
      "learning_rate": 2.193578906656e-06,
      "loss": 5.4518,
      "step": 19607
    },
    {
      "epoch": 0.7916027452563585,
      "grad_norm": 102.92957305908203,
      "learning_rate": 2.1927617639999655e-06,
      "loss": 4.6911,
      "step": 19608
    },
    {
      "epoch": 0.7916431166733953,
      "grad_norm": 82.04002380371094,
      "learning_rate": 2.191944754829475e-06,
      "loss": 5.7294,
      "step": 19609
    },
    {
      "epoch": 0.791683488090432,
      "grad_norm": 85.20066833496094,
      "learning_rate": 2.1911278791584944e-06,
      "loss": 4.3429,
      "step": 19610
    },
    {
      "epoch": 0.7917238595074687,
      "grad_norm": 115.19427490234375,
      "learning_rate": 2.1903111370009924e-06,
      "loss": 5.7142,
      "step": 19611
    },
    {
      "epoch": 0.7917642309245054,
      "grad_norm": 117.88629913330078,
      "learning_rate": 2.1894945283709335e-06,
      "loss": 4.7097,
      "step": 19612
    },
    {
      "epoch": 0.7918046023415422,
      "grad_norm": 103.05929565429688,
      "learning_rate": 2.1886780532822795e-06,
      "loss": 4.6896,
      "step": 19613
    },
    {
      "epoch": 0.7918449737585789,
      "grad_norm": 79.86615753173828,
      "learning_rate": 2.187861711748992e-06,
      "loss": 5.844,
      "step": 19614
    },
    {
      "epoch": 0.7918853451756157,
      "grad_norm": 89.8000717163086,
      "learning_rate": 2.1870455037850246e-06,
      "loss": 5.7213,
      "step": 19615
    },
    {
      "epoch": 0.7919257165926524,
      "grad_norm": 170.2668914794922,
      "learning_rate": 2.186229429404334e-06,
      "loss": 6.2293,
      "step": 19616
    },
    {
      "epoch": 0.7919660880096892,
      "grad_norm": 144.1271514892578,
      "learning_rate": 2.185413488620873e-06,
      "loss": 2.9685,
      "step": 19617
    },
    {
      "epoch": 0.7920064594267259,
      "grad_norm": 95.30744934082031,
      "learning_rate": 2.184597681448597e-06,
      "loss": 5.1454,
      "step": 19618
    },
    {
      "epoch": 0.7920468308437626,
      "grad_norm": 92.78120422363281,
      "learning_rate": 2.183782007901447e-06,
      "loss": 3.6476,
      "step": 19619
    },
    {
      "epoch": 0.7920872022607993,
      "grad_norm": 79.57080078125,
      "learning_rate": 2.182966467993373e-06,
      "loss": 4.6148,
      "step": 19620
    },
    {
      "epoch": 0.792127573677836,
      "grad_norm": 75.82179260253906,
      "learning_rate": 2.182151061738318e-06,
      "loss": 4.9314,
      "step": 19621
    },
    {
      "epoch": 0.7921679450948729,
      "grad_norm": 119.12226104736328,
      "learning_rate": 2.1813357891502263e-06,
      "loss": 4.2662,
      "step": 19622
    },
    {
      "epoch": 0.7922083165119096,
      "grad_norm": 107.42998504638672,
      "learning_rate": 2.180520650243033e-06,
      "loss": 3.8399,
      "step": 19623
    },
    {
      "epoch": 0.7922486879289463,
      "grad_norm": 114.79550170898438,
      "learning_rate": 2.1797056450306754e-06,
      "loss": 4.4256,
      "step": 19624
    },
    {
      "epoch": 0.792289059345983,
      "grad_norm": 84.92724609375,
      "learning_rate": 2.1788907735270903e-06,
      "loss": 3.6862,
      "step": 19625
    },
    {
      "epoch": 0.7923294307630198,
      "grad_norm": 74.6241683959961,
      "learning_rate": 2.1780760357462127e-06,
      "loss": 4.6301,
      "step": 19626
    },
    {
      "epoch": 0.7923698021800565,
      "grad_norm": 123.5699691772461,
      "learning_rate": 2.1772614317019657e-06,
      "loss": 6.0185,
      "step": 19627
    },
    {
      "epoch": 0.7924101735970932,
      "grad_norm": 77.76066589355469,
      "learning_rate": 2.1764469614082805e-06,
      "loss": 5.2416,
      "step": 19628
    },
    {
      "epoch": 0.79245054501413,
      "grad_norm": 152.7704620361328,
      "learning_rate": 2.175632624879084e-06,
      "loss": 6.5245,
      "step": 19629
    },
    {
      "epoch": 0.7924909164311668,
      "grad_norm": 113.8268051147461,
      "learning_rate": 2.1748184221282976e-06,
      "loss": 4.7847,
      "step": 19630
    },
    {
      "epoch": 0.7925312878482035,
      "grad_norm": 88.90865325927734,
      "learning_rate": 2.174004353169844e-06,
      "loss": 5.1763,
      "step": 19631
    },
    {
      "epoch": 0.7925716592652402,
      "grad_norm": 81.61813354492188,
      "learning_rate": 2.17319041801764e-06,
      "loss": 5.5974,
      "step": 19632
    },
    {
      "epoch": 0.7926120306822769,
      "grad_norm": 175.05242919921875,
      "learning_rate": 2.172376616685603e-06,
      "loss": 4.9055,
      "step": 19633
    },
    {
      "epoch": 0.7926524020993136,
      "grad_norm": 108.0479736328125,
      "learning_rate": 2.17156294918765e-06,
      "loss": 4.0193,
      "step": 19634
    },
    {
      "epoch": 0.7926927735163504,
      "grad_norm": 100.33062744140625,
      "learning_rate": 2.1707494155376872e-06,
      "loss": 4.2489,
      "step": 19635
    },
    {
      "epoch": 0.7927331449333872,
      "grad_norm": 135.93618774414062,
      "learning_rate": 2.1699360157496276e-06,
      "loss": 2.6931,
      "step": 19636
    },
    {
      "epoch": 0.7927735163504239,
      "grad_norm": 101.3781967163086,
      "learning_rate": 2.169122749837376e-06,
      "loss": 4.6637,
      "step": 19637
    },
    {
      "epoch": 0.7928138877674606,
      "grad_norm": 63.83954620361328,
      "learning_rate": 2.168309617814842e-06,
      "loss": 3.6633,
      "step": 19638
    },
    {
      "epoch": 0.7928542591844974,
      "grad_norm": 107.25337219238281,
      "learning_rate": 2.1674966196959237e-06,
      "loss": 5.5346,
      "step": 19639
    },
    {
      "epoch": 0.7928946306015341,
      "grad_norm": 83.9108657836914,
      "learning_rate": 2.166683755494522e-06,
      "loss": 4.4223,
      "step": 19640
    },
    {
      "epoch": 0.7929350020185708,
      "grad_norm": 91.90751647949219,
      "learning_rate": 2.1658710252245363e-06,
      "loss": 5.4977,
      "step": 19641
    },
    {
      "epoch": 0.7929753734356076,
      "grad_norm": 83.81392669677734,
      "learning_rate": 2.1650584288998643e-06,
      "loss": 5.3984,
      "step": 19642
    },
    {
      "epoch": 0.7930157448526444,
      "grad_norm": 260.87310791015625,
      "learning_rate": 2.1642459665343953e-06,
      "loss": 3.7759,
      "step": 19643
    },
    {
      "epoch": 0.7930561162696811,
      "grad_norm": 85.67504119873047,
      "learning_rate": 2.1634336381420218e-06,
      "loss": 5.0451,
      "step": 19644
    },
    {
      "epoch": 0.7930964876867178,
      "grad_norm": 87.03802490234375,
      "learning_rate": 2.1626214437366343e-06,
      "loss": 5.3923,
      "step": 19645
    },
    {
      "epoch": 0.7931368591037545,
      "grad_norm": 76.94185638427734,
      "learning_rate": 2.161809383332121e-06,
      "loss": 4.3002,
      "step": 19646
    },
    {
      "epoch": 0.7931772305207913,
      "grad_norm": 90.91857147216797,
      "learning_rate": 2.1609974569423595e-06,
      "loss": 6.0749,
      "step": 19647
    },
    {
      "epoch": 0.793217601937828,
      "grad_norm": 118.89105987548828,
      "learning_rate": 2.160185664581238e-06,
      "loss": 4.9977,
      "step": 19648
    },
    {
      "epoch": 0.7932579733548648,
      "grad_norm": 88.7786865234375,
      "learning_rate": 2.1593740062626335e-06,
      "loss": 4.8996,
      "step": 19649
    },
    {
      "epoch": 0.7932983447719015,
      "grad_norm": 110.31170654296875,
      "learning_rate": 2.158562482000425e-06,
      "loss": 4.1198,
      "step": 19650
    },
    {
      "epoch": 0.7933387161889383,
      "grad_norm": 91.02664184570312,
      "learning_rate": 2.1577510918084854e-06,
      "loss": 5.2594,
      "step": 19651
    },
    {
      "epoch": 0.793379087605975,
      "grad_norm": 86.84777069091797,
      "learning_rate": 2.156939835700691e-06,
      "loss": 3.2094,
      "step": 19652
    },
    {
      "epoch": 0.7934194590230117,
      "grad_norm": 81.11231994628906,
      "learning_rate": 2.1561287136909094e-06,
      "loss": 3.6658,
      "step": 19653
    },
    {
      "epoch": 0.7934598304400484,
      "grad_norm": 140.90489196777344,
      "learning_rate": 2.1553177257930123e-06,
      "loss": 5.4227,
      "step": 19654
    },
    {
      "epoch": 0.7935002018570851,
      "grad_norm": 92.09137725830078,
      "learning_rate": 2.154506872020862e-06,
      "loss": 6.1828,
      "step": 19655
    },
    {
      "epoch": 0.793540573274122,
      "grad_norm": 172.31967163085938,
      "learning_rate": 2.1536961523883238e-06,
      "loss": 5.5165,
      "step": 19656
    },
    {
      "epoch": 0.7935809446911587,
      "grad_norm": 113.4796371459961,
      "learning_rate": 2.1528855669092574e-06,
      "loss": 4.5181,
      "step": 19657
    },
    {
      "epoch": 0.7936213161081954,
      "grad_norm": 94.31916046142578,
      "learning_rate": 2.1520751155975273e-06,
      "loss": 5.2894,
      "step": 19658
    },
    {
      "epoch": 0.7936616875252321,
      "grad_norm": 82.43107604980469,
      "learning_rate": 2.151264798466984e-06,
      "loss": 4.6172,
      "step": 19659
    },
    {
      "epoch": 0.7937020589422689,
      "grad_norm": 85.91117858886719,
      "learning_rate": 2.150454615531484e-06,
      "loss": 4.1627,
      "step": 19660
    },
    {
      "epoch": 0.7937424303593056,
      "grad_norm": 98.0356216430664,
      "learning_rate": 2.1496445668048806e-06,
      "loss": 3.7754,
      "step": 19661
    },
    {
      "epoch": 0.7937828017763423,
      "grad_norm": 134.97344970703125,
      "learning_rate": 2.148834652301025e-06,
      "loss": 5.6759,
      "step": 19662
    },
    {
      "epoch": 0.7938231731933791,
      "grad_norm": 80.40328979492188,
      "learning_rate": 2.1480248720337602e-06,
      "loss": 4.2086,
      "step": 19663
    },
    {
      "epoch": 0.7938635446104159,
      "grad_norm": 67.97486877441406,
      "learning_rate": 2.1472152260169355e-06,
      "loss": 3.4982,
      "step": 19664
    },
    {
      "epoch": 0.7939039160274526,
      "grad_norm": 99.54109191894531,
      "learning_rate": 2.1464057142643925e-06,
      "loss": 5.6307,
      "step": 19665
    },
    {
      "epoch": 0.7939442874444893,
      "grad_norm": 121.3062744140625,
      "learning_rate": 2.1455963367899756e-06,
      "loss": 4.9662,
      "step": 19666
    },
    {
      "epoch": 0.793984658861526,
      "grad_norm": 83.16185760498047,
      "learning_rate": 2.1447870936075165e-06,
      "loss": 4.0795,
      "step": 19667
    },
    {
      "epoch": 0.7940250302785627,
      "grad_norm": 173.61056518554688,
      "learning_rate": 2.143977984730855e-06,
      "loss": 5.12,
      "step": 19668
    },
    {
      "epoch": 0.7940654016955995,
      "grad_norm": 112.98497009277344,
      "learning_rate": 2.1431690101738244e-06,
      "loss": 4.4636,
      "step": 19669
    },
    {
      "epoch": 0.7941057731126363,
      "grad_norm": 83.13621520996094,
      "learning_rate": 2.1423601699502573e-06,
      "loss": 5.9046,
      "step": 19670
    },
    {
      "epoch": 0.794146144529673,
      "grad_norm": 317.51190185546875,
      "learning_rate": 2.141551464073982e-06,
      "loss": 4.0533,
      "step": 19671
    },
    {
      "epoch": 0.7941865159467097,
      "grad_norm": 73.01262664794922,
      "learning_rate": 2.140742892558826e-06,
      "loss": 5.1168,
      "step": 19672
    },
    {
      "epoch": 0.7942268873637465,
      "grad_norm": 88.85034942626953,
      "learning_rate": 2.139934455418614e-06,
      "loss": 4.0277,
      "step": 19673
    },
    {
      "epoch": 0.7942672587807832,
      "grad_norm": 71.91212463378906,
      "learning_rate": 2.1391261526671714e-06,
      "loss": 5.6752,
      "step": 19674
    },
    {
      "epoch": 0.7943076301978199,
      "grad_norm": 78.98052215576172,
      "learning_rate": 2.1383179843183124e-06,
      "loss": 4.8128,
      "step": 19675
    },
    {
      "epoch": 0.7943480016148566,
      "grad_norm": 96.33195495605469,
      "learning_rate": 2.1375099503858567e-06,
      "loss": 3.4287,
      "step": 19676
    },
    {
      "epoch": 0.7943883730318935,
      "grad_norm": 148.810546875,
      "learning_rate": 2.136702050883621e-06,
      "loss": 6.201,
      "step": 19677
    },
    {
      "epoch": 0.7944287444489302,
      "grad_norm": 114.15724182128906,
      "learning_rate": 2.1358942858254206e-06,
      "loss": 5.0824,
      "step": 19678
    },
    {
      "epoch": 0.7944691158659669,
      "grad_norm": 90.62751770019531,
      "learning_rate": 2.1350866552250614e-06,
      "loss": 4.074,
      "step": 19679
    },
    {
      "epoch": 0.7945094872830036,
      "grad_norm": 113.86088562011719,
      "learning_rate": 2.134279159096354e-06,
      "loss": 5.2349,
      "step": 19680
    },
    {
      "epoch": 0.7945498587000404,
      "grad_norm": 177.84376525878906,
      "learning_rate": 2.133471797453106e-06,
      "loss": 5.5155,
      "step": 19681
    },
    {
      "epoch": 0.7945902301170771,
      "grad_norm": 74.56450653076172,
      "learning_rate": 2.132664570309123e-06,
      "loss": 5.3913,
      "step": 19682
    },
    {
      "epoch": 0.7946306015341138,
      "grad_norm": 88.91998291015625,
      "learning_rate": 2.131857477678202e-06,
      "loss": 3.855,
      "step": 19683
    },
    {
      "epoch": 0.7946709729511506,
      "grad_norm": 81.40926361083984,
      "learning_rate": 2.131050519574144e-06,
      "loss": 4.3813,
      "step": 19684
    },
    {
      "epoch": 0.7947113443681874,
      "grad_norm": 102.29206085205078,
      "learning_rate": 2.1302436960107464e-06,
      "loss": 4.8937,
      "step": 19685
    },
    {
      "epoch": 0.7947517157852241,
      "grad_norm": 1177.649169921875,
      "learning_rate": 2.129437007001808e-06,
      "loss": 6.1813,
      "step": 19686
    },
    {
      "epoch": 0.7947920872022608,
      "grad_norm": 103.37911224365234,
      "learning_rate": 2.1286304525611146e-06,
      "loss": 5.8232,
      "step": 19687
    },
    {
      "epoch": 0.7948324586192975,
      "grad_norm": 88.18904876708984,
      "learning_rate": 2.1278240327024602e-06,
      "loss": 7.1672,
      "step": 19688
    },
    {
      "epoch": 0.7948728300363342,
      "grad_norm": 111.11271667480469,
      "learning_rate": 2.1270177474396315e-06,
      "loss": 6.1818,
      "step": 19689
    },
    {
      "epoch": 0.794913201453371,
      "grad_norm": 77.99604034423828,
      "learning_rate": 2.126211596786414e-06,
      "loss": 6.5179,
      "step": 19690
    },
    {
      "epoch": 0.7949535728704078,
      "grad_norm": 79.39608764648438,
      "learning_rate": 2.125405580756592e-06,
      "loss": 4.4177,
      "step": 19691
    },
    {
      "epoch": 0.7949939442874445,
      "grad_norm": 150.5105438232422,
      "learning_rate": 2.124599699363946e-06,
      "loss": 6.2825,
      "step": 19692
    },
    {
      "epoch": 0.7950343157044812,
      "grad_norm": 80.64231872558594,
      "learning_rate": 2.123793952622255e-06,
      "loss": 5.5964,
      "step": 19693
    },
    {
      "epoch": 0.795074687121518,
      "grad_norm": 94.08223724365234,
      "learning_rate": 2.1229883405452977e-06,
      "loss": 4.0479,
      "step": 19694
    },
    {
      "epoch": 0.7951150585385547,
      "grad_norm": 98.89307403564453,
      "learning_rate": 2.1221828631468433e-06,
      "loss": 6.0169,
      "step": 19695
    },
    {
      "epoch": 0.7951554299555914,
      "grad_norm": 95.05789184570312,
      "learning_rate": 2.1213775204406663e-06,
      "loss": 4.9606,
      "step": 19696
    },
    {
      "epoch": 0.7951958013726281,
      "grad_norm": 137.38665771484375,
      "learning_rate": 2.1205723124405342e-06,
      "loss": 5.0864,
      "step": 19697
    },
    {
      "epoch": 0.795236172789665,
      "grad_norm": 117.13589477539062,
      "learning_rate": 2.11976723916022e-06,
      "loss": 4.843,
      "step": 19698
    },
    {
      "epoch": 0.7952765442067017,
      "grad_norm": 89.77861022949219,
      "learning_rate": 2.1189623006134805e-06,
      "loss": 5.2262,
      "step": 19699
    },
    {
      "epoch": 0.7953169156237384,
      "grad_norm": 80.53252410888672,
      "learning_rate": 2.118157496814083e-06,
      "loss": 3.9481,
      "step": 19700
    },
    {
      "epoch": 0.7953572870407751,
      "grad_norm": 110.17123413085938,
      "learning_rate": 2.117352827775786e-06,
      "loss": 5.5078,
      "step": 19701
    },
    {
      "epoch": 0.7953976584578119,
      "grad_norm": 130.0944366455078,
      "learning_rate": 2.1165482935123517e-06,
      "loss": 4.6759,
      "step": 19702
    },
    {
      "epoch": 0.7954380298748486,
      "grad_norm": 95.80096435546875,
      "learning_rate": 2.1157438940375296e-06,
      "loss": 5.0371,
      "step": 19703
    },
    {
      "epoch": 0.7954784012918853,
      "grad_norm": 78.81940460205078,
      "learning_rate": 2.114939629365076e-06,
      "loss": 3.5819,
      "step": 19704
    },
    {
      "epoch": 0.7955187727089221,
      "grad_norm": 84.07837677001953,
      "learning_rate": 2.1141354995087415e-06,
      "loss": 4.5806,
      "step": 19705
    },
    {
      "epoch": 0.7955591441259589,
      "grad_norm": 112.1850814819336,
      "learning_rate": 2.1133315044822757e-06,
      "loss": 4.852,
      "step": 19706
    },
    {
      "epoch": 0.7955995155429956,
      "grad_norm": 87.4197998046875,
      "learning_rate": 2.1125276442994236e-06,
      "loss": 4.2184,
      "step": 19707
    },
    {
      "epoch": 0.7956398869600323,
      "grad_norm": 94.7915267944336,
      "learning_rate": 2.11172391897393e-06,
      "loss": 5.6614,
      "step": 19708
    },
    {
      "epoch": 0.795680258377069,
      "grad_norm": 146.02493286132812,
      "learning_rate": 2.110920328519538e-06,
      "loss": 3.408,
      "step": 19709
    },
    {
      "epoch": 0.7957206297941057,
      "grad_norm": 105.10797882080078,
      "learning_rate": 2.1101168729499887e-06,
      "loss": 5.0786,
      "step": 19710
    },
    {
      "epoch": 0.7957610012111425,
      "grad_norm": 104.86905670166016,
      "learning_rate": 2.109313552279013e-06,
      "loss": 3.8804,
      "step": 19711
    },
    {
      "epoch": 0.7958013726281793,
      "grad_norm": 66.356201171875,
      "learning_rate": 2.108510366520351e-06,
      "loss": 5.8479,
      "step": 19712
    },
    {
      "epoch": 0.795841744045216,
      "grad_norm": 119.10662078857422,
      "learning_rate": 2.1077073156877324e-06,
      "loss": 4.4522,
      "step": 19713
    },
    {
      "epoch": 0.7958821154622527,
      "grad_norm": 76.81900024414062,
      "learning_rate": 2.1069043997948922e-06,
      "loss": 4.2253,
      "step": 19714
    },
    {
      "epoch": 0.7959224868792895,
      "grad_norm": 74.09542083740234,
      "learning_rate": 2.1061016188555518e-06,
      "loss": 4.4171,
      "step": 19715
    },
    {
      "epoch": 0.7959628582963262,
      "grad_norm": 96.33818054199219,
      "learning_rate": 2.1052989728834406e-06,
      "loss": 4.9975,
      "step": 19716
    },
    {
      "epoch": 0.7960032297133629,
      "grad_norm": 199.05770874023438,
      "learning_rate": 2.1044964618922816e-06,
      "loss": 5.1065,
      "step": 19717
    },
    {
      "epoch": 0.7960436011303996,
      "grad_norm": 101.76060485839844,
      "learning_rate": 2.1036940858957987e-06,
      "loss": 4.9503,
      "step": 19718
    },
    {
      "epoch": 0.7960839725474365,
      "grad_norm": 85.88347625732422,
      "learning_rate": 2.102891844907704e-06,
      "loss": 4.2241,
      "step": 19719
    },
    {
      "epoch": 0.7961243439644732,
      "grad_norm": 109.90596771240234,
      "learning_rate": 2.102089738941718e-06,
      "loss": 5.7191,
      "step": 19720
    },
    {
      "epoch": 0.7961647153815099,
      "grad_norm": 87.0253677368164,
      "learning_rate": 2.101287768011555e-06,
      "loss": 2.5383,
      "step": 19721
    },
    {
      "epoch": 0.7962050867985466,
      "grad_norm": 85.43608093261719,
      "learning_rate": 2.100485932130929e-06,
      "loss": 5.3466,
      "step": 19722
    },
    {
      "epoch": 0.7962454582155833,
      "grad_norm": 81.68061065673828,
      "learning_rate": 2.0996842313135446e-06,
      "loss": 4.8997,
      "step": 19723
    },
    {
      "epoch": 0.7962858296326201,
      "grad_norm": 112.13092041015625,
      "learning_rate": 2.098882665573111e-06,
      "loss": 4.9057,
      "step": 19724
    },
    {
      "epoch": 0.7963262010496568,
      "grad_norm": 181.1343231201172,
      "learning_rate": 2.0980812349233326e-06,
      "loss": 5.076,
      "step": 19725
    },
    {
      "epoch": 0.7963665724666936,
      "grad_norm": 113.2261962890625,
      "learning_rate": 2.097279939377913e-06,
      "loss": 5.247,
      "step": 19726
    },
    {
      "epoch": 0.7964069438837303,
      "grad_norm": 90.89230346679688,
      "learning_rate": 2.0964787789505527e-06,
      "loss": 4.0751,
      "step": 19727
    },
    {
      "epoch": 0.7964473153007671,
      "grad_norm": 81.0256118774414,
      "learning_rate": 2.095677753654949e-06,
      "loss": 4.7008,
      "step": 19728
    },
    {
      "epoch": 0.7964876867178038,
      "grad_norm": 98.47317504882812,
      "learning_rate": 2.094876863504798e-06,
      "loss": 5.6911,
      "step": 19729
    },
    {
      "epoch": 0.7965280581348405,
      "grad_norm": 109.35302734375,
      "learning_rate": 2.0940761085137964e-06,
      "loss": 4.5548,
      "step": 19730
    },
    {
      "epoch": 0.7965684295518772,
      "grad_norm": 116.97059631347656,
      "learning_rate": 2.0932754886956275e-06,
      "loss": 5.2604,
      "step": 19731
    },
    {
      "epoch": 0.796608800968914,
      "grad_norm": 99.77278900146484,
      "learning_rate": 2.0924750040639843e-06,
      "loss": 4.5499,
      "step": 19732
    },
    {
      "epoch": 0.7966491723859508,
      "grad_norm": 103.8173599243164,
      "learning_rate": 2.091674654632554e-06,
      "loss": 5.2845,
      "step": 19733
    },
    {
      "epoch": 0.7966895438029875,
      "grad_norm": 91.72120666503906,
      "learning_rate": 2.0908744404150218e-06,
      "loss": 4.0331,
      "step": 19734
    },
    {
      "epoch": 0.7967299152200242,
      "grad_norm": 57.500240325927734,
      "learning_rate": 2.090074361425065e-06,
      "loss": 5.0787,
      "step": 19735
    },
    {
      "epoch": 0.796770286637061,
      "grad_norm": 117.76081085205078,
      "learning_rate": 2.089274417676366e-06,
      "loss": 5.3713,
      "step": 19736
    },
    {
      "epoch": 0.7968106580540977,
      "grad_norm": 80.68966674804688,
      "learning_rate": 2.0884746091826014e-06,
      "loss": 5.1179,
      "step": 19737
    },
    {
      "epoch": 0.7968510294711344,
      "grad_norm": 182.04888916015625,
      "learning_rate": 2.087674935957449e-06,
      "loss": 3.8875,
      "step": 19738
    },
    {
      "epoch": 0.7968914008881711,
      "grad_norm": 98.09272766113281,
      "learning_rate": 2.086875398014575e-06,
      "loss": 3.3532,
      "step": 19739
    },
    {
      "epoch": 0.796931772305208,
      "grad_norm": 96.79994201660156,
      "learning_rate": 2.0860759953676547e-06,
      "loss": 4.3879,
      "step": 19740
    },
    {
      "epoch": 0.7969721437222447,
      "grad_norm": 113.43407440185547,
      "learning_rate": 2.085276728030353e-06,
      "loss": 5.4493,
      "step": 19741
    },
    {
      "epoch": 0.7970125151392814,
      "grad_norm": 87.26295471191406,
      "learning_rate": 2.084477596016341e-06,
      "loss": 6.8118,
      "step": 19742
    },
    {
      "epoch": 0.7970528865563181,
      "grad_norm": 92.39054870605469,
      "learning_rate": 2.083678599339275e-06,
      "loss": 5.9387,
      "step": 19743
    },
    {
      "epoch": 0.7970932579733548,
      "grad_norm": 675.9773559570312,
      "learning_rate": 2.0828797380128185e-06,
      "loss": 5.6736,
      "step": 19744
    },
    {
      "epoch": 0.7971336293903916,
      "grad_norm": 80.91800689697266,
      "learning_rate": 2.082081012050632e-06,
      "loss": 5.2348,
      "step": 19745
    },
    {
      "epoch": 0.7971740008074284,
      "grad_norm": 136.89443969726562,
      "learning_rate": 2.0812824214663695e-06,
      "loss": 5.3059,
      "step": 19746
    },
    {
      "epoch": 0.7972143722244651,
      "grad_norm": 103.36329650878906,
      "learning_rate": 2.080483966273685e-06,
      "loss": 5.0259,
      "step": 19747
    },
    {
      "epoch": 0.7972547436415018,
      "grad_norm": 81.96186065673828,
      "learning_rate": 2.0796856464862337e-06,
      "loss": 5.5907,
      "step": 19748
    },
    {
      "epoch": 0.7972951150585386,
      "grad_norm": 93.25100708007812,
      "learning_rate": 2.0788874621176615e-06,
      "loss": 5.855,
      "step": 19749
    },
    {
      "epoch": 0.7973354864755753,
      "grad_norm": 131.3795166015625,
      "learning_rate": 2.0780894131816188e-06,
      "loss": 5.4414,
      "step": 19750
    },
    {
      "epoch": 0.797375857892612,
      "grad_norm": 80.05248260498047,
      "learning_rate": 2.0772914996917457e-06,
      "loss": 3.4155,
      "step": 19751
    },
    {
      "epoch": 0.7974162293096487,
      "grad_norm": 94.06546020507812,
      "learning_rate": 2.0764937216616864e-06,
      "loss": 6.3895,
      "step": 19752
    },
    {
      "epoch": 0.7974566007266856,
      "grad_norm": 79.00340270996094,
      "learning_rate": 2.075696079105083e-06,
      "loss": 5.1059,
      "step": 19753
    },
    {
      "epoch": 0.7974969721437223,
      "grad_norm": 94.4520492553711,
      "learning_rate": 2.0748985720355743e-06,
      "loss": 4.7478,
      "step": 19754
    },
    {
      "epoch": 0.797537343560759,
      "grad_norm": 86.59770202636719,
      "learning_rate": 2.0741012004667904e-06,
      "loss": 4.1098,
      "step": 19755
    },
    {
      "epoch": 0.7975777149777957,
      "grad_norm": 75.27023315429688,
      "learning_rate": 2.0733039644123688e-06,
      "loss": 4.9879,
      "step": 19756
    },
    {
      "epoch": 0.7976180863948324,
      "grad_norm": 80.8658676147461,
      "learning_rate": 2.072506863885938e-06,
      "loss": 5.2707,
      "step": 19757
    },
    {
      "epoch": 0.7976584578118692,
      "grad_norm": 145.41510009765625,
      "learning_rate": 2.071709898901131e-06,
      "loss": 4.9223,
      "step": 19758
    },
    {
      "epoch": 0.7976988292289059,
      "grad_norm": 144.71511840820312,
      "learning_rate": 2.0709130694715685e-06,
      "loss": 5.2371,
      "step": 19759
    },
    {
      "epoch": 0.7977392006459427,
      "grad_norm": 81.53475189208984,
      "learning_rate": 2.0701163756108764e-06,
      "loss": 3.8691,
      "step": 19760
    },
    {
      "epoch": 0.7977795720629794,
      "grad_norm": 90.55292510986328,
      "learning_rate": 2.069319817332677e-06,
      "loss": 5.8138,
      "step": 19761
    },
    {
      "epoch": 0.7978199434800162,
      "grad_norm": 115.7326889038086,
      "learning_rate": 2.0685233946505913e-06,
      "loss": 4.344,
      "step": 19762
    },
    {
      "epoch": 0.7978603148970529,
      "grad_norm": 80.07012939453125,
      "learning_rate": 2.0677271075782324e-06,
      "loss": 5.0771,
      "step": 19763
    },
    {
      "epoch": 0.7979006863140896,
      "grad_norm": 79.2984619140625,
      "learning_rate": 2.0669309561292163e-06,
      "loss": 4.2123,
      "step": 19764
    },
    {
      "epoch": 0.7979410577311263,
      "grad_norm": 146.5369415283203,
      "learning_rate": 2.066134940317155e-06,
      "loss": 6.6731,
      "step": 19765
    },
    {
      "epoch": 0.7979814291481631,
      "grad_norm": 128.9067840576172,
      "learning_rate": 2.06533906015566e-06,
      "loss": 3.2767,
      "step": 19766
    },
    {
      "epoch": 0.7980218005651999,
      "grad_norm": 109.6405258178711,
      "learning_rate": 2.064543315658338e-06,
      "loss": 5.129,
      "step": 19767
    },
    {
      "epoch": 0.7980621719822366,
      "grad_norm": 129.92115783691406,
      "learning_rate": 2.063747706838796e-06,
      "loss": 5.4702,
      "step": 19768
    },
    {
      "epoch": 0.7981025433992733,
      "grad_norm": 76.0819320678711,
      "learning_rate": 2.0629522337106346e-06,
      "loss": 6.4857,
      "step": 19769
    },
    {
      "epoch": 0.7981429148163101,
      "grad_norm": 176.11465454101562,
      "learning_rate": 2.0621568962874583e-06,
      "loss": 4.5708,
      "step": 19770
    },
    {
      "epoch": 0.7981832862333468,
      "grad_norm": 126.23444366455078,
      "learning_rate": 2.0613616945828608e-06,
      "loss": 4.5054,
      "step": 19771
    },
    {
      "epoch": 0.7982236576503835,
      "grad_norm": 117.98355102539062,
      "learning_rate": 2.0605666286104396e-06,
      "loss": 6.8755,
      "step": 19772
    },
    {
      "epoch": 0.7982640290674202,
      "grad_norm": 87.02987670898438,
      "learning_rate": 2.059771698383789e-06,
      "loss": 3.0028,
      "step": 19773
    },
    {
      "epoch": 0.798304400484457,
      "grad_norm": 123.5592041015625,
      "learning_rate": 2.0589769039165043e-06,
      "loss": 5.2676,
      "step": 19774
    },
    {
      "epoch": 0.7983447719014938,
      "grad_norm": 510.3374328613281,
      "learning_rate": 2.0581822452221678e-06,
      "loss": 5.3733,
      "step": 19775
    },
    {
      "epoch": 0.7983851433185305,
      "grad_norm": 69.60773468017578,
      "learning_rate": 2.057387722314369e-06,
      "loss": 4.3873,
      "step": 19776
    },
    {
      "epoch": 0.7984255147355672,
      "grad_norm": 144.43113708496094,
      "learning_rate": 2.0565933352066934e-06,
      "loss": 5.0597,
      "step": 19777
    },
    {
      "epoch": 0.7984658861526039,
      "grad_norm": 110.63792419433594,
      "learning_rate": 2.055799083912725e-06,
      "loss": 6.8239,
      "step": 19778
    },
    {
      "epoch": 0.7985062575696407,
      "grad_norm": 98.487548828125,
      "learning_rate": 2.0550049684460383e-06,
      "loss": 4.6265,
      "step": 19779
    },
    {
      "epoch": 0.7985466289866774,
      "grad_norm": 80.65746307373047,
      "learning_rate": 2.054210988820213e-06,
      "loss": 3.4317,
      "step": 19780
    },
    {
      "epoch": 0.7985870004037142,
      "grad_norm": 123.55683135986328,
      "learning_rate": 2.053417145048827e-06,
      "loss": 3.5818,
      "step": 19781
    },
    {
      "epoch": 0.7986273718207509,
      "grad_norm": 76.77195739746094,
      "learning_rate": 2.0526234371454523e-06,
      "loss": 5.6444,
      "step": 19782
    },
    {
      "epoch": 0.7986677432377877,
      "grad_norm": 92.82097625732422,
      "learning_rate": 2.051829865123657e-06,
      "loss": 4.8508,
      "step": 19783
    },
    {
      "epoch": 0.7987081146548244,
      "grad_norm": 90.7065658569336,
      "learning_rate": 2.05103642899701e-06,
      "loss": 3.984,
      "step": 19784
    },
    {
      "epoch": 0.7987484860718611,
      "grad_norm": 98.69512939453125,
      "learning_rate": 2.050243128779077e-06,
      "loss": 5.452,
      "step": 19785
    },
    {
      "epoch": 0.7987888574888978,
      "grad_norm": 101.16536712646484,
      "learning_rate": 2.049449964483423e-06,
      "loss": 5.85,
      "step": 19786
    },
    {
      "epoch": 0.7988292289059346,
      "grad_norm": 87.65889739990234,
      "learning_rate": 2.0486569361236097e-06,
      "loss": 5.1153,
      "step": 19787
    },
    {
      "epoch": 0.7988696003229714,
      "grad_norm": 87.34794616699219,
      "learning_rate": 2.0478640437131946e-06,
      "loss": 5.8823,
      "step": 19788
    },
    {
      "epoch": 0.7989099717400081,
      "grad_norm": 91.80537414550781,
      "learning_rate": 2.047071287265735e-06,
      "loss": 5.2069,
      "step": 19789
    },
    {
      "epoch": 0.7989503431570448,
      "grad_norm": 140.04029846191406,
      "learning_rate": 2.0462786667947864e-06,
      "loss": 5.2477,
      "step": 19790
    },
    {
      "epoch": 0.7989907145740816,
      "grad_norm": 96.6421127319336,
      "learning_rate": 2.0454861823138983e-06,
      "loss": 4.2537,
      "step": 19791
    },
    {
      "epoch": 0.7990310859911183,
      "grad_norm": 83.81838989257812,
      "learning_rate": 2.044693833836621e-06,
      "loss": 5.5329,
      "step": 19792
    },
    {
      "epoch": 0.799071457408155,
      "grad_norm": 70.44894409179688,
      "learning_rate": 2.0439016213765005e-06,
      "loss": 5.554,
      "step": 19793
    },
    {
      "epoch": 0.7991118288251917,
      "grad_norm": 128.33071899414062,
      "learning_rate": 2.0431095449470873e-06,
      "loss": 3.9258,
      "step": 19794
    },
    {
      "epoch": 0.7991522002422285,
      "grad_norm": 94.4820785522461,
      "learning_rate": 2.0423176045619177e-06,
      "loss": 5.4319,
      "step": 19795
    },
    {
      "epoch": 0.7991925716592653,
      "grad_norm": 137.49403381347656,
      "learning_rate": 2.0415258002345352e-06,
      "loss": 6.4563,
      "step": 19796
    },
    {
      "epoch": 0.799232943076302,
      "grad_norm": 92.17435455322266,
      "learning_rate": 2.0407341319784756e-06,
      "loss": 4.4231,
      "step": 19797
    },
    {
      "epoch": 0.7992733144933387,
      "grad_norm": 168.25511169433594,
      "learning_rate": 2.0399425998072798e-06,
      "loss": 4.5848,
      "step": 19798
    },
    {
      "epoch": 0.7993136859103754,
      "grad_norm": 112.4339828491211,
      "learning_rate": 2.039151203734475e-06,
      "loss": 4.9062,
      "step": 19799
    },
    {
      "epoch": 0.7993540573274122,
      "grad_norm": 80.2802734375,
      "learning_rate": 2.0383599437735934e-06,
      "loss": 4.1028,
      "step": 19800
    },
    {
      "epoch": 0.7993944287444489,
      "grad_norm": 102.55132293701172,
      "learning_rate": 2.037568819938166e-06,
      "loss": 5.0346,
      "step": 19801
    },
    {
      "epoch": 0.7994348001614857,
      "grad_norm": 73.60181427001953,
      "learning_rate": 2.036777832241721e-06,
      "loss": 5.3066,
      "step": 19802
    },
    {
      "epoch": 0.7994751715785224,
      "grad_norm": 94.0516357421875,
      "learning_rate": 2.035986980697777e-06,
      "loss": 5.5341,
      "step": 19803
    },
    {
      "epoch": 0.7995155429955592,
      "grad_norm": 95.22498321533203,
      "learning_rate": 2.0351962653198578e-06,
      "loss": 3.9658,
      "step": 19804
    },
    {
      "epoch": 0.7995559144125959,
      "grad_norm": 105.79341888427734,
      "learning_rate": 2.034405686121483e-06,
      "loss": 4.5674,
      "step": 19805
    },
    {
      "epoch": 0.7995962858296326,
      "grad_norm": 94.2880630493164,
      "learning_rate": 2.0336152431161704e-06,
      "loss": 5.3285,
      "step": 19806
    },
    {
      "epoch": 0.7996366572466693,
      "grad_norm": 116.1760025024414,
      "learning_rate": 2.0328249363174346e-06,
      "loss": 5.4366,
      "step": 19807
    },
    {
      "epoch": 0.799677028663706,
      "grad_norm": 92.8310775756836,
      "learning_rate": 2.032034765738787e-06,
      "loss": 5.1475,
      "step": 19808
    },
    {
      "epoch": 0.7997174000807429,
      "grad_norm": 213.74351501464844,
      "learning_rate": 2.031244731393739e-06,
      "loss": 6.1653,
      "step": 19809
    },
    {
      "epoch": 0.7997577714977796,
      "grad_norm": 168.74880981445312,
      "learning_rate": 2.0304548332958007e-06,
      "loss": 3.4531,
      "step": 19810
    },
    {
      "epoch": 0.7997981429148163,
      "grad_norm": 100.03748321533203,
      "learning_rate": 2.029665071458471e-06,
      "loss": 4.138,
      "step": 19811
    },
    {
      "epoch": 0.799838514331853,
      "grad_norm": 98.07404327392578,
      "learning_rate": 2.0288754458952566e-06,
      "loss": 4.4005,
      "step": 19812
    },
    {
      "epoch": 0.7998788857488898,
      "grad_norm": 83.06731414794922,
      "learning_rate": 2.028085956619659e-06,
      "loss": 4.6146,
      "step": 19813
    },
    {
      "epoch": 0.7999192571659265,
      "grad_norm": 79.87245178222656,
      "learning_rate": 2.027296603645178e-06,
      "loss": 4.6269,
      "step": 19814
    },
    {
      "epoch": 0.7999596285829632,
      "grad_norm": 67.05561828613281,
      "learning_rate": 2.0265073869853046e-06,
      "loss": 5.6447,
      "step": 19815
    },
    {
      "epoch": 0.8,
      "grad_norm": 99.7608871459961,
      "learning_rate": 2.0257183066535347e-06,
      "loss": 5.2707,
      "step": 19816
    },
    {
      "epoch": 0.8000403714170368,
      "grad_norm": 90.54853820800781,
      "learning_rate": 2.0249293626633613e-06,
      "loss": 5.4313,
      "step": 19817
    },
    {
      "epoch": 0.8000807428340735,
      "grad_norm": 114.62517547607422,
      "learning_rate": 2.0241405550282757e-06,
      "loss": 4.6369,
      "step": 19818
    },
    {
      "epoch": 0.8001211142511102,
      "grad_norm": 86.75992584228516,
      "learning_rate": 2.0233518837617584e-06,
      "loss": 4.8751,
      "step": 19819
    },
    {
      "epoch": 0.8001614856681469,
      "grad_norm": 93.85023498535156,
      "learning_rate": 2.022563348877298e-06,
      "loss": 4.3469,
      "step": 19820
    },
    {
      "epoch": 0.8002018570851837,
      "grad_norm": 105.19690704345703,
      "learning_rate": 2.021774950388374e-06,
      "loss": 5.0904,
      "step": 19821
    },
    {
      "epoch": 0.8002422285022204,
      "grad_norm": 104.79380798339844,
      "learning_rate": 2.020986688308472e-06,
      "loss": 3.5936,
      "step": 19822
    },
    {
      "epoch": 0.8002825999192572,
      "grad_norm": 130.68878173828125,
      "learning_rate": 2.0201985626510624e-06,
      "loss": 4.3142,
      "step": 19823
    },
    {
      "epoch": 0.8003229713362939,
      "grad_norm": 129.3004608154297,
      "learning_rate": 2.019410573429622e-06,
      "loss": 5.5955,
      "step": 19824
    },
    {
      "epoch": 0.8003633427533307,
      "grad_norm": 113.29901885986328,
      "learning_rate": 2.018622720657626e-06,
      "loss": 4.8925,
      "step": 19825
    },
    {
      "epoch": 0.8004037141703674,
      "grad_norm": 107.13336181640625,
      "learning_rate": 2.017835004348543e-06,
      "loss": 5.2308,
      "step": 19826
    },
    {
      "epoch": 0.8004440855874041,
      "grad_norm": 80.13639831542969,
      "learning_rate": 2.0170474245158422e-06,
      "loss": 5.2098,
      "step": 19827
    },
    {
      "epoch": 0.8004844570044408,
      "grad_norm": 81.97503662109375,
      "learning_rate": 2.0162599811729888e-06,
      "loss": 4.012,
      "step": 19828
    },
    {
      "epoch": 0.8005248284214775,
      "grad_norm": 71.61138153076172,
      "learning_rate": 2.015472674333446e-06,
      "loss": 4.2371,
      "step": 19829
    },
    {
      "epoch": 0.8005651998385144,
      "grad_norm": 91.02489471435547,
      "learning_rate": 2.0146855040106784e-06,
      "loss": 5.1763,
      "step": 19830
    },
    {
      "epoch": 0.8006055712555511,
      "grad_norm": 103.0244369506836,
      "learning_rate": 2.0138984702181386e-06,
      "loss": 3.4321,
      "step": 19831
    },
    {
      "epoch": 0.8006459426725878,
      "grad_norm": 102.29212951660156,
      "learning_rate": 2.0131115729692864e-06,
      "loss": 4.78,
      "step": 19832
    },
    {
      "epoch": 0.8006863140896245,
      "grad_norm": 80.1527099609375,
      "learning_rate": 2.0123248122775752e-06,
      "loss": 4.3153,
      "step": 19833
    },
    {
      "epoch": 0.8007266855066613,
      "grad_norm": 91.95993041992188,
      "learning_rate": 2.01153818815646e-06,
      "loss": 4.1392,
      "step": 19834
    },
    {
      "epoch": 0.800767056923698,
      "grad_norm": 97.77391052246094,
      "learning_rate": 2.0107517006193856e-06,
      "loss": 4.958,
      "step": 19835
    },
    {
      "epoch": 0.8008074283407347,
      "grad_norm": 101.49929809570312,
      "learning_rate": 2.0099653496798e-06,
      "loss": 6.3488,
      "step": 19836
    },
    {
      "epoch": 0.8008477997577715,
      "grad_norm": 85.89916229248047,
      "learning_rate": 2.009179135351148e-06,
      "loss": 4.7264,
      "step": 19837
    },
    {
      "epoch": 0.8008881711748083,
      "grad_norm": 79.72669219970703,
      "learning_rate": 2.0083930576468768e-06,
      "loss": 5.2723,
      "step": 19838
    },
    {
      "epoch": 0.800928542591845,
      "grad_norm": 81.01325988769531,
      "learning_rate": 2.0076071165804188e-06,
      "loss": 4.1731,
      "step": 19839
    },
    {
      "epoch": 0.8009689140088817,
      "grad_norm": 123.93915557861328,
      "learning_rate": 2.006821312165217e-06,
      "loss": 3.8278,
      "step": 19840
    },
    {
      "epoch": 0.8010092854259184,
      "grad_norm": 126.99859619140625,
      "learning_rate": 2.0060356444147044e-06,
      "loss": 5.0393,
      "step": 19841
    },
    {
      "epoch": 0.8010496568429551,
      "grad_norm": 86.97594451904297,
      "learning_rate": 2.0052501133423176e-06,
      "loss": 4.5134,
      "step": 19842
    },
    {
      "epoch": 0.8010900282599919,
      "grad_norm": 106.38371276855469,
      "learning_rate": 2.004464718961483e-06,
      "loss": 4.2983,
      "step": 19843
    },
    {
      "epoch": 0.8011303996770287,
      "grad_norm": 85.02818298339844,
      "learning_rate": 2.0036794612856304e-06,
      "loss": 4.3693,
      "step": 19844
    },
    {
      "epoch": 0.8011707710940654,
      "grad_norm": 83.96593475341797,
      "learning_rate": 2.002894340328185e-06,
      "loss": 3.7883,
      "step": 19845
    },
    {
      "epoch": 0.8012111425111021,
      "grad_norm": 97.35083770751953,
      "learning_rate": 2.002109356102574e-06,
      "loss": 4.9723,
      "step": 19846
    },
    {
      "epoch": 0.8012515139281389,
      "grad_norm": 178.2904815673828,
      "learning_rate": 2.0013245086222145e-06,
      "loss": 5.7031,
      "step": 19847
    },
    {
      "epoch": 0.8012918853451756,
      "grad_norm": 94.6496810913086,
      "learning_rate": 2.000539797900529e-06,
      "loss": 5.8741,
      "step": 19848
    },
    {
      "epoch": 0.8013322567622123,
      "grad_norm": 96.13359069824219,
      "learning_rate": 1.9997552239509323e-06,
      "loss": 5.8933,
      "step": 19849
    },
    {
      "epoch": 0.8013726281792491,
      "grad_norm": 107.6246566772461,
      "learning_rate": 1.9989707867868423e-06,
      "loss": 5.1698,
      "step": 19850
    },
    {
      "epoch": 0.8014129995962859,
      "grad_norm": 78.93834686279297,
      "learning_rate": 1.998186486421665e-06,
      "loss": 5.9624,
      "step": 19851
    },
    {
      "epoch": 0.8014533710133226,
      "grad_norm": 71.03306579589844,
      "learning_rate": 1.9974023228688143e-06,
      "loss": 4.2671,
      "step": 19852
    },
    {
      "epoch": 0.8014937424303593,
      "grad_norm": 87.93898010253906,
      "learning_rate": 1.996618296141696e-06,
      "loss": 4.5958,
      "step": 19853
    },
    {
      "epoch": 0.801534113847396,
      "grad_norm": 102.4128189086914,
      "learning_rate": 1.995834406253717e-06,
      "loss": 3.6813,
      "step": 19854
    },
    {
      "epoch": 0.8015744852644328,
      "grad_norm": 103.90756225585938,
      "learning_rate": 1.995050653218277e-06,
      "loss": 6.4521,
      "step": 19855
    },
    {
      "epoch": 0.8016148566814695,
      "grad_norm": 112.49143981933594,
      "learning_rate": 1.9942670370487784e-06,
      "loss": 5.8587,
      "step": 19856
    },
    {
      "epoch": 0.8016552280985063,
      "grad_norm": 103.64004516601562,
      "learning_rate": 1.9934835577586176e-06,
      "loss": 4.8188,
      "step": 19857
    },
    {
      "epoch": 0.801695599515543,
      "grad_norm": 103.92110443115234,
      "learning_rate": 1.992700215361194e-06,
      "loss": 4.999,
      "step": 19858
    },
    {
      "epoch": 0.8017359709325798,
      "grad_norm": 142.89210510253906,
      "learning_rate": 1.991917009869897e-06,
      "loss": 4.4599,
      "step": 19859
    },
    {
      "epoch": 0.8017763423496165,
      "grad_norm": 116.78304290771484,
      "learning_rate": 1.991133941298118e-06,
      "loss": 4.7883,
      "step": 19860
    },
    {
      "epoch": 0.8018167137666532,
      "grad_norm": 162.3749237060547,
      "learning_rate": 1.990351009659247e-06,
      "loss": 4.9892,
      "step": 19861
    },
    {
      "epoch": 0.8018570851836899,
      "grad_norm": 125.51326751708984,
      "learning_rate": 1.9895682149666718e-06,
      "loss": 4.7141,
      "step": 19862
    },
    {
      "epoch": 0.8018974566007266,
      "grad_norm": 108.71121978759766,
      "learning_rate": 1.988785557233772e-06,
      "loss": 4.8403,
      "step": 19863
    },
    {
      "epoch": 0.8019378280177635,
      "grad_norm": 90.55335235595703,
      "learning_rate": 1.988003036473932e-06,
      "loss": 5.3872,
      "step": 19864
    },
    {
      "epoch": 0.8019781994348002,
      "grad_norm": 86.95062255859375,
      "learning_rate": 1.987220652700531e-06,
      "loss": 6.2529,
      "step": 19865
    },
    {
      "epoch": 0.8020185708518369,
      "grad_norm": 87.91899108886719,
      "learning_rate": 1.9864384059269446e-06,
      "loss": 5.5034,
      "step": 19866
    },
    {
      "epoch": 0.8020589422688736,
      "grad_norm": 78.45049285888672,
      "learning_rate": 1.9856562961665493e-06,
      "loss": 4.7182,
      "step": 19867
    },
    {
      "epoch": 0.8020993136859104,
      "grad_norm": 88.66496276855469,
      "learning_rate": 1.984874323432716e-06,
      "loss": 5.8123,
      "step": 19868
    },
    {
      "epoch": 0.8021396851029471,
      "grad_norm": 85.97154998779297,
      "learning_rate": 1.9840924877388157e-06,
      "loss": 4.8263,
      "step": 19869
    },
    {
      "epoch": 0.8021800565199838,
      "grad_norm": 94.43909454345703,
      "learning_rate": 1.9833107890982183e-06,
      "loss": 5.2215,
      "step": 19870
    },
    {
      "epoch": 0.8022204279370206,
      "grad_norm": 91.33715057373047,
      "learning_rate": 1.9825292275242824e-06,
      "loss": 5.6612,
      "step": 19871
    },
    {
      "epoch": 0.8022607993540574,
      "grad_norm": 65.57890319824219,
      "learning_rate": 1.981747803030375e-06,
      "loss": 5.8815,
      "step": 19872
    },
    {
      "epoch": 0.8023011707710941,
      "grad_norm": 104.4654541015625,
      "learning_rate": 1.980966515629856e-06,
      "loss": 4.1027,
      "step": 19873
    },
    {
      "epoch": 0.8023415421881308,
      "grad_norm": 178.53036499023438,
      "learning_rate": 1.980185365336087e-06,
      "loss": 5.1617,
      "step": 19874
    },
    {
      "epoch": 0.8023819136051675,
      "grad_norm": 146.82949829101562,
      "learning_rate": 1.9794043521624175e-06,
      "loss": 5.6533,
      "step": 19875
    },
    {
      "epoch": 0.8024222850222043,
      "grad_norm": 137.38925170898438,
      "learning_rate": 1.9786234761222044e-06,
      "loss": 6.0329,
      "step": 19876
    },
    {
      "epoch": 0.802462656439241,
      "grad_norm": 70.8827133178711,
      "learning_rate": 1.977842737228799e-06,
      "loss": 4.1092,
      "step": 19877
    },
    {
      "epoch": 0.8025030278562778,
      "grad_norm": 159.6173553466797,
      "learning_rate": 1.977062135495551e-06,
      "loss": 4.9933,
      "step": 19878
    },
    {
      "epoch": 0.8025433992733145,
      "grad_norm": 81.68444061279297,
      "learning_rate": 1.9762816709358048e-06,
      "loss": 6.1117,
      "step": 19879
    },
    {
      "epoch": 0.8025837706903513,
      "grad_norm": 81.73402404785156,
      "learning_rate": 1.9755013435629032e-06,
      "loss": 6.7751,
      "step": 19880
    },
    {
      "epoch": 0.802624142107388,
      "grad_norm": 86.63756561279297,
      "learning_rate": 1.974721153390191e-06,
      "loss": 5.1022,
      "step": 19881
    },
    {
      "epoch": 0.8026645135244247,
      "grad_norm": 116.5472412109375,
      "learning_rate": 1.9739411004310105e-06,
      "loss": 5.7203,
      "step": 19882
    },
    {
      "epoch": 0.8027048849414614,
      "grad_norm": 88.37193298339844,
      "learning_rate": 1.973161184698691e-06,
      "loss": 4.9703,
      "step": 19883
    },
    {
      "epoch": 0.8027452563584981,
      "grad_norm": 100.72036743164062,
      "learning_rate": 1.9723814062065715e-06,
      "loss": 5.1477,
      "step": 19884
    },
    {
      "epoch": 0.802785627775535,
      "grad_norm": 96.5000991821289,
      "learning_rate": 1.971601764967984e-06,
      "loss": 4.1848,
      "step": 19885
    },
    {
      "epoch": 0.8028259991925717,
      "grad_norm": 130.51014709472656,
      "learning_rate": 1.970822260996259e-06,
      "loss": 5.8407,
      "step": 19886
    },
    {
      "epoch": 0.8028663706096084,
      "grad_norm": 71.28435516357422,
      "learning_rate": 1.9700428943047232e-06,
      "loss": 5.0034,
      "step": 19887
    },
    {
      "epoch": 0.8029067420266451,
      "grad_norm": 106.470947265625,
      "learning_rate": 1.9692636649067033e-06,
      "loss": 5.0029,
      "step": 19888
    },
    {
      "epoch": 0.8029471134436819,
      "grad_norm": 85.77281951904297,
      "learning_rate": 1.9684845728155223e-06,
      "loss": 4.58,
      "step": 19889
    },
    {
      "epoch": 0.8029874848607186,
      "grad_norm": 92.19379425048828,
      "learning_rate": 1.967705618044501e-06,
      "loss": 4.0206,
      "step": 19890
    },
    {
      "epoch": 0.8030278562777553,
      "grad_norm": 85.26347351074219,
      "learning_rate": 1.9669268006069554e-06,
      "loss": 4.9339,
      "step": 19891
    },
    {
      "epoch": 0.8030682276947921,
      "grad_norm": 152.95700073242188,
      "learning_rate": 1.9661481205162024e-06,
      "loss": 4.5218,
      "step": 19892
    },
    {
      "epoch": 0.8031085991118289,
      "grad_norm": 87.93585968017578,
      "learning_rate": 1.9653695777855565e-06,
      "loss": 3.9052,
      "step": 19893
    },
    {
      "epoch": 0.8031489705288656,
      "grad_norm": 99.33814239501953,
      "learning_rate": 1.964591172428332e-06,
      "loss": 5.4906,
      "step": 19894
    },
    {
      "epoch": 0.8031893419459023,
      "grad_norm": 137.48834228515625,
      "learning_rate": 1.96381290445783e-06,
      "loss": 4.7182,
      "step": 19895
    },
    {
      "epoch": 0.803229713362939,
      "grad_norm": 69.82697296142578,
      "learning_rate": 1.9630347738873635e-06,
      "loss": 4.4405,
      "step": 19896
    },
    {
      "epoch": 0.8032700847799757,
      "grad_norm": 116.83033752441406,
      "learning_rate": 1.9622567807302328e-06,
      "loss": 5.1277,
      "step": 19897
    },
    {
      "epoch": 0.8033104561970125,
      "grad_norm": 120.26140594482422,
      "learning_rate": 1.9614789249997458e-06,
      "loss": 5.812,
      "step": 19898
    },
    {
      "epoch": 0.8033508276140493,
      "grad_norm": 100.84056091308594,
      "learning_rate": 1.9607012067091945e-06,
      "loss": 5.5418,
      "step": 19899
    },
    {
      "epoch": 0.803391199031086,
      "grad_norm": 83.52295684814453,
      "learning_rate": 1.959923625871879e-06,
      "loss": 4.9521,
      "step": 19900
    },
    {
      "epoch": 0.8034315704481227,
      "grad_norm": 109.68305969238281,
      "learning_rate": 1.959146182501096e-06,
      "loss": 3.857,
      "step": 19901
    },
    {
      "epoch": 0.8034719418651595,
      "grad_norm": 714.59765625,
      "learning_rate": 1.958368876610135e-06,
      "loss": 4.611,
      "step": 19902
    },
    {
      "epoch": 0.8035123132821962,
      "grad_norm": 100.47418212890625,
      "learning_rate": 1.9575917082122886e-06,
      "loss": 3.4412,
      "step": 19903
    },
    {
      "epoch": 0.8035526846992329,
      "grad_norm": 116.77210235595703,
      "learning_rate": 1.9568146773208428e-06,
      "loss": 4.8549,
      "step": 19904
    },
    {
      "epoch": 0.8035930561162696,
      "grad_norm": 87.74402618408203,
      "learning_rate": 1.9560377839490872e-06,
      "loss": 5.1248,
      "step": 19905
    },
    {
      "epoch": 0.8036334275333065,
      "grad_norm": 99.02212524414062,
      "learning_rate": 1.955261028110299e-06,
      "loss": 5.8124,
      "step": 19906
    },
    {
      "epoch": 0.8036737989503432,
      "grad_norm": 80.35950469970703,
      "learning_rate": 1.9544844098177605e-06,
      "loss": 5.7674,
      "step": 19907
    },
    {
      "epoch": 0.8037141703673799,
      "grad_norm": 87.20018005371094,
      "learning_rate": 1.9537079290847515e-06,
      "loss": 5.6158,
      "step": 19908
    },
    {
      "epoch": 0.8037545417844166,
      "grad_norm": 99.59430694580078,
      "learning_rate": 1.952931585924548e-06,
      "loss": 5.8262,
      "step": 19909
    },
    {
      "epoch": 0.8037949132014534,
      "grad_norm": 113.3664779663086,
      "learning_rate": 1.9521553803504256e-06,
      "loss": 4.7593,
      "step": 19910
    },
    {
      "epoch": 0.8038352846184901,
      "grad_norm": 89.69617462158203,
      "learning_rate": 1.9513793123756498e-06,
      "loss": 4.7757,
      "step": 19911
    },
    {
      "epoch": 0.8038756560355268,
      "grad_norm": 92.29318237304688,
      "learning_rate": 1.950603382013493e-06,
      "loss": 5.972,
      "step": 19912
    },
    {
      "epoch": 0.8039160274525636,
      "grad_norm": 77.53382873535156,
      "learning_rate": 1.9498275892772224e-06,
      "loss": 3.928,
      "step": 19913
    },
    {
      "epoch": 0.8039563988696004,
      "grad_norm": 79.6983413696289,
      "learning_rate": 1.949051934180104e-06,
      "loss": 4.3804,
      "step": 19914
    },
    {
      "epoch": 0.8039967702866371,
      "grad_norm": 148.68292236328125,
      "learning_rate": 1.948276416735394e-06,
      "loss": 4.7072,
      "step": 19915
    },
    {
      "epoch": 0.8040371417036738,
      "grad_norm": 77.7805404663086,
      "learning_rate": 1.947501036956355e-06,
      "loss": 6.0929,
      "step": 19916
    },
    {
      "epoch": 0.8040775131207105,
      "grad_norm": 72.47895812988281,
      "learning_rate": 1.9467257948562445e-06,
      "loss": 4.7613,
      "step": 19917
    },
    {
      "epoch": 0.8041178845377472,
      "grad_norm": 88.33903503417969,
      "learning_rate": 1.9459506904483207e-06,
      "loss": 5.4691,
      "step": 19918
    },
    {
      "epoch": 0.804158255954784,
      "grad_norm": 81.97711944580078,
      "learning_rate": 1.945175723745829e-06,
      "loss": 4.8549,
      "step": 19919
    },
    {
      "epoch": 0.8041986273718208,
      "grad_norm": 75.69701385498047,
      "learning_rate": 1.944400894762023e-06,
      "loss": 4.1529,
      "step": 19920
    },
    {
      "epoch": 0.8042389987888575,
      "grad_norm": 79.0694580078125,
      "learning_rate": 1.9436262035101504e-06,
      "loss": 2.6357,
      "step": 19921
    },
    {
      "epoch": 0.8042793702058942,
      "grad_norm": 87.35128021240234,
      "learning_rate": 1.942851650003458e-06,
      "loss": 4.7462,
      "step": 19922
    },
    {
      "epoch": 0.804319741622931,
      "grad_norm": 72.38924407958984,
      "learning_rate": 1.9420772342551863e-06,
      "loss": 3.2706,
      "step": 19923
    },
    {
      "epoch": 0.8043601130399677,
      "grad_norm": 167.103759765625,
      "learning_rate": 1.9413029562785778e-06,
      "loss": 4.3105,
      "step": 19924
    },
    {
      "epoch": 0.8044004844570044,
      "grad_norm": 85.5535888671875,
      "learning_rate": 1.9405288160868707e-06,
      "loss": 3.3445,
      "step": 19925
    },
    {
      "epoch": 0.8044408558740411,
      "grad_norm": 91.87897491455078,
      "learning_rate": 1.939754813693302e-06,
      "loss": 4.5525,
      "step": 19926
    },
    {
      "epoch": 0.804481227291078,
      "grad_norm": 92.06736755371094,
      "learning_rate": 1.938980949111102e-06,
      "loss": 4.9967,
      "step": 19927
    },
    {
      "epoch": 0.8045215987081147,
      "grad_norm": 79.90362548828125,
      "learning_rate": 1.938207222353504e-06,
      "loss": 4.6719,
      "step": 19928
    },
    {
      "epoch": 0.8045619701251514,
      "grad_norm": 81.09478759765625,
      "learning_rate": 1.9374336334337374e-06,
      "loss": 6.3925,
      "step": 19929
    },
    {
      "epoch": 0.8046023415421881,
      "grad_norm": 146.21810913085938,
      "learning_rate": 1.93666018236503e-06,
      "loss": 4.2445,
      "step": 19930
    },
    {
      "epoch": 0.8046427129592248,
      "grad_norm": 193.43540954589844,
      "learning_rate": 1.9358868691606026e-06,
      "loss": 4.7505,
      "step": 19931
    },
    {
      "epoch": 0.8046830843762616,
      "grad_norm": 117.73844146728516,
      "learning_rate": 1.9351136938336777e-06,
      "loss": 4.4943,
      "step": 19932
    },
    {
      "epoch": 0.8047234557932983,
      "grad_norm": 73.81306457519531,
      "learning_rate": 1.9343406563974765e-06,
      "loss": 5.538,
      "step": 19933
    },
    {
      "epoch": 0.8047638272103351,
      "grad_norm": 82.03588104248047,
      "learning_rate": 1.933567756865218e-06,
      "loss": 4.2238,
      "step": 19934
    },
    {
      "epoch": 0.8048041986273718,
      "grad_norm": 95.39710235595703,
      "learning_rate": 1.932794995250111e-06,
      "loss": 3.3551,
      "step": 19935
    },
    {
      "epoch": 0.8048445700444086,
      "grad_norm": 72.06916046142578,
      "learning_rate": 1.932022371565372e-06,
      "loss": 1.9538,
      "step": 19936
    },
    {
      "epoch": 0.8048849414614453,
      "grad_norm": 118.33118438720703,
      "learning_rate": 1.9312498858242103e-06,
      "loss": 5.1517,
      "step": 19937
    },
    {
      "epoch": 0.804925312878482,
      "grad_norm": 124.67432403564453,
      "learning_rate": 1.9304775380398356e-06,
      "loss": 4.4741,
      "step": 19938
    },
    {
      "epoch": 0.8049656842955187,
      "grad_norm": 88.82445526123047,
      "learning_rate": 1.9297053282254494e-06,
      "loss": 6.3161,
      "step": 19939
    },
    {
      "epoch": 0.8050060557125555,
      "grad_norm": 144.89430236816406,
      "learning_rate": 1.9289332563942564e-06,
      "loss": 4.524,
      "step": 19940
    },
    {
      "epoch": 0.8050464271295923,
      "grad_norm": 143.2859344482422,
      "learning_rate": 1.9281613225594573e-06,
      "loss": 2.9076,
      "step": 19941
    },
    {
      "epoch": 0.805086798546629,
      "grad_norm": 151.73130798339844,
      "learning_rate": 1.9273895267342504e-06,
      "loss": 4.5836,
      "step": 19942
    },
    {
      "epoch": 0.8051271699636657,
      "grad_norm": 98.37702178955078,
      "learning_rate": 1.9266178689318315e-06,
      "loss": 3.1866,
      "step": 19943
    },
    {
      "epoch": 0.8051675413807025,
      "grad_norm": 90.50162506103516,
      "learning_rate": 1.9258463491653956e-06,
      "loss": 4.2257,
      "step": 19944
    },
    {
      "epoch": 0.8052079127977392,
      "grad_norm": 107.47422790527344,
      "learning_rate": 1.9250749674481316e-06,
      "loss": 4.9921,
      "step": 19945
    },
    {
      "epoch": 0.8052482842147759,
      "grad_norm": 84.47586822509766,
      "learning_rate": 1.9243037237932315e-06,
      "loss": 4.9204,
      "step": 19946
    },
    {
      "epoch": 0.8052886556318126,
      "grad_norm": 114.30821990966797,
      "learning_rate": 1.923532618213878e-06,
      "loss": 4.7009,
      "step": 19947
    },
    {
      "epoch": 0.8053290270488495,
      "grad_norm": 90.74706268310547,
      "learning_rate": 1.922761650723256e-06,
      "loss": 5.8722,
      "step": 19948
    },
    {
      "epoch": 0.8053693984658862,
      "grad_norm": 185.9127197265625,
      "learning_rate": 1.9219908213345484e-06,
      "loss": 4.577,
      "step": 19949
    },
    {
      "epoch": 0.8054097698829229,
      "grad_norm": 87.71748352050781,
      "learning_rate": 1.9212201300609367e-06,
      "loss": 4.3408,
      "step": 19950
    },
    {
      "epoch": 0.8054501412999596,
      "grad_norm": 135.3666534423828,
      "learning_rate": 1.920449576915594e-06,
      "loss": 5.9913,
      "step": 19951
    },
    {
      "epoch": 0.8054905127169963,
      "grad_norm": 88.36744689941406,
      "learning_rate": 1.919679161911695e-06,
      "loss": 4.5945,
      "step": 19952
    },
    {
      "epoch": 0.8055308841340331,
      "grad_norm": 147.87399291992188,
      "learning_rate": 1.918908885062414e-06,
      "loss": 6.0765,
      "step": 19953
    },
    {
      "epoch": 0.8055712555510699,
      "grad_norm": 91.65596771240234,
      "learning_rate": 1.918138746380923e-06,
      "loss": 5.4602,
      "step": 19954
    },
    {
      "epoch": 0.8056116269681066,
      "grad_norm": 97.52651977539062,
      "learning_rate": 1.9173687458803845e-06,
      "loss": 5.0347,
      "step": 19955
    },
    {
      "epoch": 0.8056519983851433,
      "grad_norm": 115.75858306884766,
      "learning_rate": 1.916598883573966e-06,
      "loss": 5.0707,
      "step": 19956
    },
    {
      "epoch": 0.8056923698021801,
      "grad_norm": 100.85939025878906,
      "learning_rate": 1.91582915947483e-06,
      "loss": 5.1984,
      "step": 19957
    },
    {
      "epoch": 0.8057327412192168,
      "grad_norm": 99.16017150878906,
      "learning_rate": 1.915059573596141e-06,
      "loss": 4.2802,
      "step": 19958
    },
    {
      "epoch": 0.8057731126362535,
      "grad_norm": 92.57837677001953,
      "learning_rate": 1.9142901259510504e-06,
      "loss": 6.1161,
      "step": 19959
    },
    {
      "epoch": 0.8058134840532902,
      "grad_norm": 121.74063873291016,
      "learning_rate": 1.913520816552716e-06,
      "loss": 4.6232,
      "step": 19960
    },
    {
      "epoch": 0.8058538554703271,
      "grad_norm": 123.61788940429688,
      "learning_rate": 1.9127516454142948e-06,
      "loss": 5.2147,
      "step": 19961
    },
    {
      "epoch": 0.8058942268873638,
      "grad_norm": 115.12115478515625,
      "learning_rate": 1.9119826125489338e-06,
      "loss": 4.4379,
      "step": 19962
    },
    {
      "epoch": 0.8059345983044005,
      "grad_norm": 107.48206329345703,
      "learning_rate": 1.911213717969783e-06,
      "loss": 5.1634,
      "step": 19963
    },
    {
      "epoch": 0.8059749697214372,
      "grad_norm": 104.54961395263672,
      "learning_rate": 1.91044496168999e-06,
      "loss": 5.4861,
      "step": 19964
    },
    {
      "epoch": 0.806015341138474,
      "grad_norm": 121.93537902832031,
      "learning_rate": 1.9096763437226974e-06,
      "loss": 5.1007,
      "step": 19965
    },
    {
      "epoch": 0.8060557125555107,
      "grad_norm": 94.21411895751953,
      "learning_rate": 1.9089078640810464e-06,
      "loss": 5.1771,
      "step": 19966
    },
    {
      "epoch": 0.8060960839725474,
      "grad_norm": 134.3336944580078,
      "learning_rate": 1.9081395227781806e-06,
      "loss": 5.6353,
      "step": 19967
    },
    {
      "epoch": 0.8061364553895842,
      "grad_norm": 304.5849304199219,
      "learning_rate": 1.9073713198272293e-06,
      "loss": 4.8049,
      "step": 19968
    },
    {
      "epoch": 0.806176826806621,
      "grad_norm": 130.27687072753906,
      "learning_rate": 1.9066032552413317e-06,
      "loss": 3.5595,
      "step": 19969
    },
    {
      "epoch": 0.8062171982236577,
      "grad_norm": 160.00216674804688,
      "learning_rate": 1.9058353290336185e-06,
      "loss": 5.6252,
      "step": 19970
    },
    {
      "epoch": 0.8062575696406944,
      "grad_norm": 82.14730834960938,
      "learning_rate": 1.9050675412172225e-06,
      "loss": 5.0958,
      "step": 19971
    },
    {
      "epoch": 0.8062979410577311,
      "grad_norm": 104.07258605957031,
      "learning_rate": 1.9042998918052669e-06,
      "loss": 5.5679,
      "step": 19972
    },
    {
      "epoch": 0.8063383124747678,
      "grad_norm": 81.77925872802734,
      "learning_rate": 1.9035323808108773e-06,
      "loss": 4.9372,
      "step": 19973
    },
    {
      "epoch": 0.8063786838918046,
      "grad_norm": 112.34452056884766,
      "learning_rate": 1.902765008247177e-06,
      "loss": 4.3659,
      "step": 19974
    },
    {
      "epoch": 0.8064190553088414,
      "grad_norm": 88.56558990478516,
      "learning_rate": 1.90199777412729e-06,
      "loss": 5.2915,
      "step": 19975
    },
    {
      "epoch": 0.8064594267258781,
      "grad_norm": 108.63282012939453,
      "learning_rate": 1.9012306784643286e-06,
      "loss": 6.3345,
      "step": 19976
    },
    {
      "epoch": 0.8064997981429148,
      "grad_norm": 317.13134765625,
      "learning_rate": 1.9004637212714095e-06,
      "loss": 5.0385,
      "step": 19977
    },
    {
      "epoch": 0.8065401695599516,
      "grad_norm": 109.46589660644531,
      "learning_rate": 1.8996969025616474e-06,
      "loss": 3.6383,
      "step": 19978
    },
    {
      "epoch": 0.8065805409769883,
      "grad_norm": 105.0233154296875,
      "learning_rate": 1.8989302223481565e-06,
      "loss": 4.693,
      "step": 19979
    },
    {
      "epoch": 0.806620912394025,
      "grad_norm": 124.44672393798828,
      "learning_rate": 1.8981636806440373e-06,
      "loss": 5.8109,
      "step": 19980
    },
    {
      "epoch": 0.8066612838110617,
      "grad_norm": 97.693359375,
      "learning_rate": 1.8973972774624006e-06,
      "loss": 5.8765,
      "step": 19981
    },
    {
      "epoch": 0.8067016552280986,
      "grad_norm": 104.88398742675781,
      "learning_rate": 1.89663101281635e-06,
      "loss": 3.6014,
      "step": 19982
    },
    {
      "epoch": 0.8067420266451353,
      "grad_norm": 93.00357055664062,
      "learning_rate": 1.8958648867189855e-06,
      "loss": 4.2946,
      "step": 19983
    },
    {
      "epoch": 0.806782398062172,
      "grad_norm": 98.96775817871094,
      "learning_rate": 1.8950988991834074e-06,
      "loss": 4.3626,
      "step": 19984
    },
    {
      "epoch": 0.8068227694792087,
      "grad_norm": 117.19115447998047,
      "learning_rate": 1.894333050222712e-06,
      "loss": 2.6794,
      "step": 19985
    },
    {
      "epoch": 0.8068631408962454,
      "grad_norm": 110.40118408203125,
      "learning_rate": 1.893567339849992e-06,
      "loss": 5.6219,
      "step": 19986
    },
    {
      "epoch": 0.8069035123132822,
      "grad_norm": 109.33575439453125,
      "learning_rate": 1.892801768078345e-06,
      "loss": 5.0652,
      "step": 19987
    },
    {
      "epoch": 0.8069438837303189,
      "grad_norm": 62.81851577758789,
      "learning_rate": 1.8920363349208526e-06,
      "loss": 4.2887,
      "step": 19988
    },
    {
      "epoch": 0.8069842551473557,
      "grad_norm": 69.4846420288086,
      "learning_rate": 1.8912710403906054e-06,
      "loss": 3.9449,
      "step": 19989
    },
    {
      "epoch": 0.8070246265643924,
      "grad_norm": 177.10494995117188,
      "learning_rate": 1.8905058845006885e-06,
      "loss": 4.9313,
      "step": 19990
    },
    {
      "epoch": 0.8070649979814292,
      "grad_norm": 88.08106231689453,
      "learning_rate": 1.889740867264186e-06,
      "loss": 4.8183,
      "step": 19991
    },
    {
      "epoch": 0.8071053693984659,
      "grad_norm": 94.57151794433594,
      "learning_rate": 1.888975988694174e-06,
      "loss": 6.3004,
      "step": 19992
    },
    {
      "epoch": 0.8071457408155026,
      "grad_norm": 82.97151947021484,
      "learning_rate": 1.888211248803732e-06,
      "loss": 2.6628,
      "step": 19993
    },
    {
      "epoch": 0.8071861122325393,
      "grad_norm": 82.78780364990234,
      "learning_rate": 1.887446647605935e-06,
      "loss": 6.0349,
      "step": 19994
    },
    {
      "epoch": 0.807226483649576,
      "grad_norm": 107.94291687011719,
      "learning_rate": 1.8866821851138583e-06,
      "loss": 4.8109,
      "step": 19995
    },
    {
      "epoch": 0.8072668550666129,
      "grad_norm": 114.547119140625,
      "learning_rate": 1.8859178613405681e-06,
      "loss": 6.2654,
      "step": 19996
    },
    {
      "epoch": 0.8073072264836496,
      "grad_norm": 140.96441650390625,
      "learning_rate": 1.885153676299134e-06,
      "loss": 5.77,
      "step": 19997
    },
    {
      "epoch": 0.8073475979006863,
      "grad_norm": 90.01333618164062,
      "learning_rate": 1.8843896300026232e-06,
      "loss": 4.4621,
      "step": 19998
    },
    {
      "epoch": 0.807387969317723,
      "grad_norm": 82.87274169921875,
      "learning_rate": 1.8836257224641008e-06,
      "loss": 4.7186,
      "step": 19999
    },
    {
      "epoch": 0.8074283407347598,
      "grad_norm": 132.0242156982422,
      "learning_rate": 1.8828619536966231e-06,
      "loss": 5.6447,
      "step": 20000
    },
    {
      "epoch": 0.8074687121517965,
      "grad_norm": 139.65830993652344,
      "learning_rate": 1.8820983237132518e-06,
      "loss": 3.0809,
      "step": 20001
    },
    {
      "epoch": 0.8075090835688332,
      "grad_norm": 83.57992553710938,
      "learning_rate": 1.8813348325270409e-06,
      "loss": 5.6963,
      "step": 20002
    },
    {
      "epoch": 0.80754945498587,
      "grad_norm": 101.27349853515625,
      "learning_rate": 1.8805714801510467e-06,
      "loss": 5.6804,
      "step": 20003
    },
    {
      "epoch": 0.8075898264029068,
      "grad_norm": 91.1968994140625,
      "learning_rate": 1.8798082665983209e-06,
      "loss": 5.1537,
      "step": 20004
    },
    {
      "epoch": 0.8076301978199435,
      "grad_norm": 186.82223510742188,
      "learning_rate": 1.8790451918819109e-06,
      "loss": 3.6616,
      "step": 20005
    },
    {
      "epoch": 0.8076705692369802,
      "grad_norm": 127.53563690185547,
      "learning_rate": 1.8782822560148638e-06,
      "loss": 4.1134,
      "step": 20006
    },
    {
      "epoch": 0.8077109406540169,
      "grad_norm": 80.4791488647461,
      "learning_rate": 1.8775194590102275e-06,
      "loss": 5.3212,
      "step": 20007
    },
    {
      "epoch": 0.8077513120710537,
      "grad_norm": 88.63394165039062,
      "learning_rate": 1.8767568008810387e-06,
      "loss": 5.2284,
      "step": 20008
    },
    {
      "epoch": 0.8077916834880904,
      "grad_norm": 103.98745727539062,
      "learning_rate": 1.8759942816403387e-06,
      "loss": 5.2773,
      "step": 20009
    },
    {
      "epoch": 0.8078320549051272,
      "grad_norm": 92.64569091796875,
      "learning_rate": 1.8752319013011666e-06,
      "loss": 5.8402,
      "step": 20010
    },
    {
      "epoch": 0.8078724263221639,
      "grad_norm": 85.06692504882812,
      "learning_rate": 1.8744696598765588e-06,
      "loss": 3.8274,
      "step": 20011
    },
    {
      "epoch": 0.8079127977392007,
      "grad_norm": 119.12686157226562,
      "learning_rate": 1.8737075573795427e-06,
      "loss": 5.8442,
      "step": 20012
    },
    {
      "epoch": 0.8079531691562374,
      "grad_norm": 98.70730590820312,
      "learning_rate": 1.87294559382315e-06,
      "loss": 4.9898,
      "step": 20013
    },
    {
      "epoch": 0.8079935405732741,
      "grad_norm": 97.9526138305664,
      "learning_rate": 1.8721837692204115e-06,
      "loss": 4.9458,
      "step": 20014
    },
    {
      "epoch": 0.8080339119903108,
      "grad_norm": 119.34683990478516,
      "learning_rate": 1.871422083584352e-06,
      "loss": 3.6033,
      "step": 20015
    },
    {
      "epoch": 0.8080742834073475,
      "grad_norm": 115.36712646484375,
      "learning_rate": 1.8706605369279906e-06,
      "loss": 4.3872,
      "step": 20016
    },
    {
      "epoch": 0.8081146548243844,
      "grad_norm": 90.05770874023438,
      "learning_rate": 1.869899129264351e-06,
      "loss": 4.1693,
      "step": 20017
    },
    {
      "epoch": 0.8081550262414211,
      "grad_norm": 101.09712219238281,
      "learning_rate": 1.8691378606064514e-06,
      "loss": 4.2453,
      "step": 20018
    },
    {
      "epoch": 0.8081953976584578,
      "grad_norm": 90.53900909423828,
      "learning_rate": 1.8683767309673096e-06,
      "loss": 3.2856,
      "step": 20019
    },
    {
      "epoch": 0.8082357690754945,
      "grad_norm": 116.81132507324219,
      "learning_rate": 1.8676157403599348e-06,
      "loss": 4.8446,
      "step": 20020
    },
    {
      "epoch": 0.8082761404925313,
      "grad_norm": 89.97486877441406,
      "learning_rate": 1.8668548887973404e-06,
      "loss": 4.0587,
      "step": 20021
    },
    {
      "epoch": 0.808316511909568,
      "grad_norm": 92.26175689697266,
      "learning_rate": 1.8660941762925343e-06,
      "loss": 5.3555,
      "step": 20022
    },
    {
      "epoch": 0.8083568833266047,
      "grad_norm": 105.9290771484375,
      "learning_rate": 1.8653336028585245e-06,
      "loss": 3.408,
      "step": 20023
    },
    {
      "epoch": 0.8083972547436415,
      "grad_norm": 313.3800354003906,
      "learning_rate": 1.8645731685083146e-06,
      "loss": 4.4009,
      "step": 20024
    },
    {
      "epoch": 0.8084376261606783,
      "grad_norm": 102.04651641845703,
      "learning_rate": 1.8638128732549055e-06,
      "loss": 4.3584,
      "step": 20025
    },
    {
      "epoch": 0.808477997577715,
      "grad_norm": 67.66150665283203,
      "learning_rate": 1.8630527171112966e-06,
      "loss": 4.5883,
      "step": 20026
    },
    {
      "epoch": 0.8085183689947517,
      "grad_norm": 78.28242492675781,
      "learning_rate": 1.8622927000904878e-06,
      "loss": 5.9642,
      "step": 20027
    },
    {
      "epoch": 0.8085587404117884,
      "grad_norm": 167.41879272460938,
      "learning_rate": 1.8615328222054685e-06,
      "loss": 5.5911,
      "step": 20028
    },
    {
      "epoch": 0.8085991118288252,
      "grad_norm": 92.62703704833984,
      "learning_rate": 1.860773083469234e-06,
      "loss": 4.1433,
      "step": 20029
    },
    {
      "epoch": 0.8086394832458619,
      "grad_norm": 147.63926696777344,
      "learning_rate": 1.8600134838947724e-06,
      "loss": 4.6217,
      "step": 20030
    },
    {
      "epoch": 0.8086798546628987,
      "grad_norm": 77.67898559570312,
      "learning_rate": 1.859254023495074e-06,
      "loss": 4.2752,
      "step": 20031
    },
    {
      "epoch": 0.8087202260799354,
      "grad_norm": 78.423828125,
      "learning_rate": 1.85849470228312e-06,
      "loss": 4.9633,
      "step": 20032
    },
    {
      "epoch": 0.8087605974969722,
      "grad_norm": 106.78073120117188,
      "learning_rate": 1.857735520271895e-06,
      "loss": 4.3369,
      "step": 20033
    },
    {
      "epoch": 0.8088009689140089,
      "grad_norm": 116.80989837646484,
      "learning_rate": 1.8569764774743792e-06,
      "loss": 6.292,
      "step": 20034
    },
    {
      "epoch": 0.8088413403310456,
      "grad_norm": 71.97473907470703,
      "learning_rate": 1.8562175739035527e-06,
      "loss": 5.2096,
      "step": 20035
    },
    {
      "epoch": 0.8088817117480823,
      "grad_norm": 108.8878402709961,
      "learning_rate": 1.8554588095723857e-06,
      "loss": 4.8613,
      "step": 20036
    },
    {
      "epoch": 0.808922083165119,
      "grad_norm": 87.7884292602539,
      "learning_rate": 1.854700184493854e-06,
      "loss": 3.5467,
      "step": 20037
    },
    {
      "epoch": 0.8089624545821559,
      "grad_norm": 111.87871551513672,
      "learning_rate": 1.853941698680929e-06,
      "loss": 4.6481,
      "step": 20038
    },
    {
      "epoch": 0.8090028259991926,
      "grad_norm": 91.50904083251953,
      "learning_rate": 1.8531833521465804e-06,
      "loss": 5.0249,
      "step": 20039
    },
    {
      "epoch": 0.8090431974162293,
      "grad_norm": 118.441650390625,
      "learning_rate": 1.85242514490377e-06,
      "loss": 3.7362,
      "step": 20040
    },
    {
      "epoch": 0.809083568833266,
      "grad_norm": 68.31563568115234,
      "learning_rate": 1.8516670769654642e-06,
      "loss": 4.3114,
      "step": 20041
    },
    {
      "epoch": 0.8091239402503028,
      "grad_norm": 125.56021881103516,
      "learning_rate": 1.8509091483446228e-06,
      "loss": 5.537,
      "step": 20042
    },
    {
      "epoch": 0.8091643116673395,
      "grad_norm": 88.43345642089844,
      "learning_rate": 1.8501513590542064e-06,
      "loss": 5.9235,
      "step": 20043
    },
    {
      "epoch": 0.8092046830843762,
      "grad_norm": 129.46763610839844,
      "learning_rate": 1.8493937091071702e-06,
      "loss": 4.2266,
      "step": 20044
    },
    {
      "epoch": 0.809245054501413,
      "grad_norm": 77.76493835449219,
      "learning_rate": 1.848636198516468e-06,
      "loss": 4.8624,
      "step": 20045
    },
    {
      "epoch": 0.8092854259184498,
      "grad_norm": 85.1604995727539,
      "learning_rate": 1.8478788272950532e-06,
      "loss": 5.9682,
      "step": 20046
    },
    {
      "epoch": 0.8093257973354865,
      "grad_norm": 159.7749786376953,
      "learning_rate": 1.8471215954558752e-06,
      "loss": 5.3004,
      "step": 20047
    },
    {
      "epoch": 0.8093661687525232,
      "grad_norm": 87.38849639892578,
      "learning_rate": 1.8463645030118783e-06,
      "loss": 2.9635,
      "step": 20048
    },
    {
      "epoch": 0.8094065401695599,
      "grad_norm": 98.64319610595703,
      "learning_rate": 1.8456075499760074e-06,
      "loss": 5.3006,
      "step": 20049
    },
    {
      "epoch": 0.8094469115865967,
      "grad_norm": 86.29573822021484,
      "learning_rate": 1.8448507363612057e-06,
      "loss": 5.816,
      "step": 20050
    },
    {
      "epoch": 0.8094872830036334,
      "grad_norm": 104.787109375,
      "learning_rate": 1.844094062180415e-06,
      "loss": 4.9967,
      "step": 20051
    },
    {
      "epoch": 0.8095276544206702,
      "grad_norm": 94.61693572998047,
      "learning_rate": 1.8433375274465681e-06,
      "loss": 2.9783,
      "step": 20052
    },
    {
      "epoch": 0.8095680258377069,
      "grad_norm": 96.87020111083984,
      "learning_rate": 1.8425811321726017e-06,
      "loss": 5.716,
      "step": 20053
    },
    {
      "epoch": 0.8096083972547437,
      "grad_norm": 122.0209732055664,
      "learning_rate": 1.8418248763714497e-06,
      "loss": 5.6145,
      "step": 20054
    },
    {
      "epoch": 0.8096487686717804,
      "grad_norm": 75.8925552368164,
      "learning_rate": 1.841068760056044e-06,
      "loss": 5.0456,
      "step": 20055
    },
    {
      "epoch": 0.8096891400888171,
      "grad_norm": 83.77389526367188,
      "learning_rate": 1.8403127832393074e-06,
      "loss": 4.7154,
      "step": 20056
    },
    {
      "epoch": 0.8097295115058538,
      "grad_norm": 115.79987335205078,
      "learning_rate": 1.8395569459341688e-06,
      "loss": 4.589,
      "step": 20057
    },
    {
      "epoch": 0.8097698829228905,
      "grad_norm": 104.8531265258789,
      "learning_rate": 1.83880124815355e-06,
      "loss": 5.037,
      "step": 20058
    },
    {
      "epoch": 0.8098102543399274,
      "grad_norm": 85.03567504882812,
      "learning_rate": 1.8380456899103737e-06,
      "loss": 4.7979,
      "step": 20059
    },
    {
      "epoch": 0.8098506257569641,
      "grad_norm": 92.64002990722656,
      "learning_rate": 1.8372902712175555e-06,
      "loss": 6.5243,
      "step": 20060
    },
    {
      "epoch": 0.8098909971740008,
      "grad_norm": 75.50655364990234,
      "learning_rate": 1.8365349920880116e-06,
      "loss": 5.7652,
      "step": 20061
    },
    {
      "epoch": 0.8099313685910375,
      "grad_norm": 88.4065933227539,
      "learning_rate": 1.835779852534656e-06,
      "loss": 5.3735,
      "step": 20062
    },
    {
      "epoch": 0.8099717400080743,
      "grad_norm": 90.27302551269531,
      "learning_rate": 1.8350248525703996e-06,
      "loss": 5.0471,
      "step": 20063
    },
    {
      "epoch": 0.810012111425111,
      "grad_norm": 141.56214904785156,
      "learning_rate": 1.834269992208153e-06,
      "loss": 5.2726,
      "step": 20064
    },
    {
      "epoch": 0.8100524828421478,
      "grad_norm": 85.8858642578125,
      "learning_rate": 1.8335152714608206e-06,
      "loss": 4.5907,
      "step": 20065
    },
    {
      "epoch": 0.8100928542591845,
      "grad_norm": 76.37788391113281,
      "learning_rate": 1.8327606903413064e-06,
      "loss": 6.5218,
      "step": 20066
    },
    {
      "epoch": 0.8101332256762213,
      "grad_norm": 92.38036346435547,
      "learning_rate": 1.8320062488625157e-06,
      "loss": 3.4323,
      "step": 20067
    },
    {
      "epoch": 0.810173597093258,
      "grad_norm": 108.30485534667969,
      "learning_rate": 1.8312519470373425e-06,
      "loss": 3.9838,
      "step": 20068
    },
    {
      "epoch": 0.8102139685102947,
      "grad_norm": 66.71000671386719,
      "learning_rate": 1.8304977848786843e-06,
      "loss": 5.7129,
      "step": 20069
    },
    {
      "epoch": 0.8102543399273314,
      "grad_norm": 96.64508819580078,
      "learning_rate": 1.8297437623994374e-06,
      "loss": 3.6985,
      "step": 20070
    },
    {
      "epoch": 0.8102947113443681,
      "grad_norm": 129.2093963623047,
      "learning_rate": 1.828989879612496e-06,
      "loss": 4.3321,
      "step": 20071
    },
    {
      "epoch": 0.810335082761405,
      "grad_norm": 59.540771484375,
      "learning_rate": 1.8282361365307444e-06,
      "loss": 4.1739,
      "step": 20072
    },
    {
      "epoch": 0.8103754541784417,
      "grad_norm": 101.4326400756836,
      "learning_rate": 1.827482533167072e-06,
      "loss": 5.5138,
      "step": 20073
    },
    {
      "epoch": 0.8104158255954784,
      "grad_norm": 156.1681671142578,
      "learning_rate": 1.8267290695343642e-06,
      "loss": 5.3703,
      "step": 20074
    },
    {
      "epoch": 0.8104561970125151,
      "grad_norm": 105.89287567138672,
      "learning_rate": 1.8259757456455062e-06,
      "loss": 5.0671,
      "step": 20075
    },
    {
      "epoch": 0.8104965684295519,
      "grad_norm": 70.56285858154297,
      "learning_rate": 1.8252225615133724e-06,
      "loss": 3.2891,
      "step": 20076
    },
    {
      "epoch": 0.8105369398465886,
      "grad_norm": 135.55784606933594,
      "learning_rate": 1.8244695171508442e-06,
      "loss": 4.5003,
      "step": 20077
    },
    {
      "epoch": 0.8105773112636253,
      "grad_norm": 75.01765441894531,
      "learning_rate": 1.823716612570795e-06,
      "loss": 5.1641,
      "step": 20078
    },
    {
      "epoch": 0.8106176826806621,
      "grad_norm": 92.9873275756836,
      "learning_rate": 1.8229638477861012e-06,
      "loss": 5.4435,
      "step": 20079
    },
    {
      "epoch": 0.8106580540976989,
      "grad_norm": 96.05240631103516,
      "learning_rate": 1.822211222809629e-06,
      "loss": 4.7311,
      "step": 20080
    },
    {
      "epoch": 0.8106984255147356,
      "grad_norm": 75.28317260742188,
      "learning_rate": 1.821458737654248e-06,
      "loss": 5.4556,
      "step": 20081
    },
    {
      "epoch": 0.8107387969317723,
      "grad_norm": 174.951416015625,
      "learning_rate": 1.820706392332824e-06,
      "loss": 4.6847,
      "step": 20082
    },
    {
      "epoch": 0.810779168348809,
      "grad_norm": 100.92861938476562,
      "learning_rate": 1.8199541868582216e-06,
      "loss": 4.9591,
      "step": 20083
    },
    {
      "epoch": 0.8108195397658458,
      "grad_norm": 90.86032104492188,
      "learning_rate": 1.8192021212432997e-06,
      "loss": 2.9218,
      "step": 20084
    },
    {
      "epoch": 0.8108599111828825,
      "grad_norm": 103.50444030761719,
      "learning_rate": 1.8184501955009193e-06,
      "loss": 5.0402,
      "step": 20085
    },
    {
      "epoch": 0.8109002825999193,
      "grad_norm": 121.04483032226562,
      "learning_rate": 1.8176984096439344e-06,
      "loss": 5.4148,
      "step": 20086
    },
    {
      "epoch": 0.810940654016956,
      "grad_norm": 82.39888000488281,
      "learning_rate": 1.8169467636852023e-06,
      "loss": 4.0219,
      "step": 20087
    },
    {
      "epoch": 0.8109810254339928,
      "grad_norm": 192.24038696289062,
      "learning_rate": 1.8161952576375686e-06,
      "loss": 5.7747,
      "step": 20088
    },
    {
      "epoch": 0.8110213968510295,
      "grad_norm": 95.81869506835938,
      "learning_rate": 1.815443891513886e-06,
      "loss": 5.1776,
      "step": 20089
    },
    {
      "epoch": 0.8110617682680662,
      "grad_norm": 97.35590362548828,
      "learning_rate": 1.8146926653270014e-06,
      "loss": 5.3173,
      "step": 20090
    },
    {
      "epoch": 0.8111021396851029,
      "grad_norm": 112.74895477294922,
      "learning_rate": 1.8139415790897597e-06,
      "loss": 5.1194,
      "step": 20091
    },
    {
      "epoch": 0.8111425111021396,
      "grad_norm": 144.3260498046875,
      "learning_rate": 1.8131906328149983e-06,
      "loss": 4.6707,
      "step": 20092
    },
    {
      "epoch": 0.8111828825191765,
      "grad_norm": 107.32653045654297,
      "learning_rate": 1.8124398265155606e-06,
      "loss": 6.2555,
      "step": 20093
    },
    {
      "epoch": 0.8112232539362132,
      "grad_norm": 91.26448059082031,
      "learning_rate": 1.811689160204282e-06,
      "loss": 5.1413,
      "step": 20094
    },
    {
      "epoch": 0.8112636253532499,
      "grad_norm": 93.12823486328125,
      "learning_rate": 1.8109386338940005e-06,
      "loss": 4.2646,
      "step": 20095
    },
    {
      "epoch": 0.8113039967702866,
      "grad_norm": 83.5047378540039,
      "learning_rate": 1.8101882475975418e-06,
      "loss": 4.3756,
      "step": 20096
    },
    {
      "epoch": 0.8113443681873234,
      "grad_norm": 99.43062591552734,
      "learning_rate": 1.8094380013277402e-06,
      "loss": 3.9561,
      "step": 20097
    },
    {
      "epoch": 0.8113847396043601,
      "grad_norm": 106.74401092529297,
      "learning_rate": 1.808687895097423e-06,
      "loss": 5.5633,
      "step": 20098
    },
    {
      "epoch": 0.8114251110213968,
      "grad_norm": 78.46356201171875,
      "learning_rate": 1.8079379289194166e-06,
      "loss": 5.6685,
      "step": 20099
    },
    {
      "epoch": 0.8114654824384336,
      "grad_norm": 99.86650848388672,
      "learning_rate": 1.8071881028065398e-06,
      "loss": 4.802,
      "step": 20100
    },
    {
      "epoch": 0.8115058538554704,
      "grad_norm": 90.69123840332031,
      "learning_rate": 1.8064384167716142e-06,
      "loss": 5.0049,
      "step": 20101
    },
    {
      "epoch": 0.8115462252725071,
      "grad_norm": 122.6351318359375,
      "learning_rate": 1.8056888708274589e-06,
      "loss": 5.6887,
      "step": 20102
    },
    {
      "epoch": 0.8115865966895438,
      "grad_norm": 123.21646881103516,
      "learning_rate": 1.804939464986889e-06,
      "loss": 4.1543,
      "step": 20103
    },
    {
      "epoch": 0.8116269681065805,
      "grad_norm": 77.90873718261719,
      "learning_rate": 1.8041901992627165e-06,
      "loss": 4.4733,
      "step": 20104
    },
    {
      "epoch": 0.8116673395236172,
      "grad_norm": 77.66724395751953,
      "learning_rate": 1.8034410736677533e-06,
      "loss": 6.1005,
      "step": 20105
    },
    {
      "epoch": 0.811707710940654,
      "grad_norm": 104.86717987060547,
      "learning_rate": 1.802692088214808e-06,
      "loss": 3.1018,
      "step": 20106
    },
    {
      "epoch": 0.8117480823576908,
      "grad_norm": 101.74434661865234,
      "learning_rate": 1.8019432429166883e-06,
      "loss": 5.3342,
      "step": 20107
    },
    {
      "epoch": 0.8117884537747275,
      "grad_norm": 181.50169372558594,
      "learning_rate": 1.8011945377861916e-06,
      "loss": 6.1536,
      "step": 20108
    },
    {
      "epoch": 0.8118288251917642,
      "grad_norm": 87.76322937011719,
      "learning_rate": 1.8004459728361245e-06,
      "loss": 4.7421,
      "step": 20109
    },
    {
      "epoch": 0.811869196608801,
      "grad_norm": 68.24285125732422,
      "learning_rate": 1.7996975480792822e-06,
      "loss": 4.136,
      "step": 20110
    },
    {
      "epoch": 0.8119095680258377,
      "grad_norm": 94.5552978515625,
      "learning_rate": 1.7989492635284667e-06,
      "loss": 3.5063,
      "step": 20111
    },
    {
      "epoch": 0.8119499394428744,
      "grad_norm": 174.835205078125,
      "learning_rate": 1.7982011191964643e-06,
      "loss": 2.878,
      "step": 20112
    },
    {
      "epoch": 0.8119903108599111,
      "grad_norm": 86.66033172607422,
      "learning_rate": 1.7974531150960718e-06,
      "loss": 3.5755,
      "step": 20113
    },
    {
      "epoch": 0.812030682276948,
      "grad_norm": 78.32103729248047,
      "learning_rate": 1.7967052512400752e-06,
      "loss": 3.5677,
      "step": 20114
    },
    {
      "epoch": 0.8120710536939847,
      "grad_norm": 118.03507995605469,
      "learning_rate": 1.7959575276412656e-06,
      "loss": 3.8512,
      "step": 20115
    },
    {
      "epoch": 0.8121114251110214,
      "grad_norm": 121.37603759765625,
      "learning_rate": 1.7952099443124216e-06,
      "loss": 5.7974,
      "step": 20116
    },
    {
      "epoch": 0.8121517965280581,
      "grad_norm": 101.49754333496094,
      "learning_rate": 1.7944625012663285e-06,
      "loss": 5.2981,
      "step": 20117
    },
    {
      "epoch": 0.8121921679450949,
      "grad_norm": 116.1924819946289,
      "learning_rate": 1.7937151985157653e-06,
      "loss": 5.5671,
      "step": 20118
    },
    {
      "epoch": 0.8122325393621316,
      "grad_norm": 165.52574157714844,
      "learning_rate": 1.792968036073509e-06,
      "loss": 3.9292,
      "step": 20119
    },
    {
      "epoch": 0.8122729107791683,
      "grad_norm": 93.41809844970703,
      "learning_rate": 1.7922210139523345e-06,
      "loss": 5.1313,
      "step": 20120
    },
    {
      "epoch": 0.8123132821962051,
      "grad_norm": 122.30672454833984,
      "learning_rate": 1.7914741321650142e-06,
      "loss": 3.9993,
      "step": 20121
    },
    {
      "epoch": 0.8123536536132419,
      "grad_norm": 97.34335327148438,
      "learning_rate": 1.7907273907243183e-06,
      "loss": 3.5916,
      "step": 20122
    },
    {
      "epoch": 0.8123940250302786,
      "grad_norm": 88.00617218017578,
      "learning_rate": 1.7899807896430165e-06,
      "loss": 4.4906,
      "step": 20123
    },
    {
      "epoch": 0.8124343964473153,
      "grad_norm": 112.90426635742188,
      "learning_rate": 1.7892343289338688e-06,
      "loss": 3.8551,
      "step": 20124
    },
    {
      "epoch": 0.812474767864352,
      "grad_norm": 94.6688003540039,
      "learning_rate": 1.7884880086096401e-06,
      "loss": 3.8992,
      "step": 20125
    },
    {
      "epoch": 0.8125151392813887,
      "grad_norm": 67.0893325805664,
      "learning_rate": 1.7877418286830905e-06,
      "loss": 4.6356,
      "step": 20126
    },
    {
      "epoch": 0.8125555106984255,
      "grad_norm": 87.3851318359375,
      "learning_rate": 1.786995789166983e-06,
      "loss": 6.4718,
      "step": 20127
    },
    {
      "epoch": 0.8125958821154623,
      "grad_norm": 91.8754653930664,
      "learning_rate": 1.7862498900740655e-06,
      "loss": 4.3352,
      "step": 20128
    },
    {
      "epoch": 0.812636253532499,
      "grad_norm": 138.2151336669922,
      "learning_rate": 1.785504131417094e-06,
      "loss": 6.2068,
      "step": 20129
    },
    {
      "epoch": 0.8126766249495357,
      "grad_norm": 109.58610534667969,
      "learning_rate": 1.7847585132088197e-06,
      "loss": 3.6342,
      "step": 20130
    },
    {
      "epoch": 0.8127169963665725,
      "grad_norm": 103.35404968261719,
      "learning_rate": 1.7840130354619933e-06,
      "loss": 3.8079,
      "step": 20131
    },
    {
      "epoch": 0.8127573677836092,
      "grad_norm": 102.11552429199219,
      "learning_rate": 1.783267698189356e-06,
      "loss": 4.3751,
      "step": 20132
    },
    {
      "epoch": 0.8127977392006459,
      "grad_norm": 113.1011962890625,
      "learning_rate": 1.782522501403654e-06,
      "loss": 2.8278,
      "step": 20133
    },
    {
      "epoch": 0.8128381106176826,
      "grad_norm": 96.06639099121094,
      "learning_rate": 1.7817774451176273e-06,
      "loss": 4.5997,
      "step": 20134
    },
    {
      "epoch": 0.8128784820347195,
      "grad_norm": 90.95309448242188,
      "learning_rate": 1.781032529344018e-06,
      "loss": 6.9918,
      "step": 20135
    },
    {
      "epoch": 0.8129188534517562,
      "grad_norm": 125.78604888916016,
      "learning_rate": 1.7802877540955576e-06,
      "loss": 5.247,
      "step": 20136
    },
    {
      "epoch": 0.8129592248687929,
      "grad_norm": 68.26280975341797,
      "learning_rate": 1.779543119384982e-06,
      "loss": 4.3499,
      "step": 20137
    },
    {
      "epoch": 0.8129995962858296,
      "grad_norm": 123.081298828125,
      "learning_rate": 1.7787986252250234e-06,
      "loss": 4.7264,
      "step": 20138
    },
    {
      "epoch": 0.8130399677028664,
      "grad_norm": 82.22115325927734,
      "learning_rate": 1.77805427162841e-06,
      "loss": 5.5843,
      "step": 20139
    },
    {
      "epoch": 0.8130803391199031,
      "grad_norm": 106.22337341308594,
      "learning_rate": 1.7773100586078685e-06,
      "loss": 5.7873,
      "step": 20140
    },
    {
      "epoch": 0.8131207105369398,
      "grad_norm": 90.79662322998047,
      "learning_rate": 1.7765659861761241e-06,
      "loss": 4.9187,
      "step": 20141
    },
    {
      "epoch": 0.8131610819539766,
      "grad_norm": 104.4548568725586,
      "learning_rate": 1.7758220543458981e-06,
      "loss": 4.0688,
      "step": 20142
    },
    {
      "epoch": 0.8132014533710133,
      "grad_norm": 182.2954559326172,
      "learning_rate": 1.7750782631299124e-06,
      "loss": 3.9499,
      "step": 20143
    },
    {
      "epoch": 0.8132418247880501,
      "grad_norm": 68.15888977050781,
      "learning_rate": 1.77433461254088e-06,
      "loss": 4.2929,
      "step": 20144
    },
    {
      "epoch": 0.8132821962050868,
      "grad_norm": 112.4361343383789,
      "learning_rate": 1.7735911025915176e-06,
      "loss": 5.532,
      "step": 20145
    },
    {
      "epoch": 0.8133225676221235,
      "grad_norm": 88.57998657226562,
      "learning_rate": 1.772847733294536e-06,
      "loss": 4.1626,
      "step": 20146
    },
    {
      "epoch": 0.8133629390391602,
      "grad_norm": 102.92826080322266,
      "learning_rate": 1.7721045046626495e-06,
      "loss": 3.4288,
      "step": 20147
    },
    {
      "epoch": 0.813403310456197,
      "grad_norm": 94.91573333740234,
      "learning_rate": 1.7713614167085601e-06,
      "loss": 7.2123,
      "step": 20148
    },
    {
      "epoch": 0.8134436818732338,
      "grad_norm": 112.35758209228516,
      "learning_rate": 1.7706184694449756e-06,
      "loss": 5.7663,
      "step": 20149
    },
    {
      "epoch": 0.8134840532902705,
      "grad_norm": 84.54270935058594,
      "learning_rate": 1.769875662884598e-06,
      "loss": 6.3045,
      "step": 20150
    },
    {
      "epoch": 0.8135244247073072,
      "grad_norm": 93.1266860961914,
      "learning_rate": 1.7691329970401306e-06,
      "loss": 5.2703,
      "step": 20151
    },
    {
      "epoch": 0.813564796124344,
      "grad_norm": 91.80001068115234,
      "learning_rate": 1.7683904719242651e-06,
      "loss": 5.6515,
      "step": 20152
    },
    {
      "epoch": 0.8136051675413807,
      "grad_norm": 97.19761657714844,
      "learning_rate": 1.7676480875497016e-06,
      "loss": 4.8851,
      "step": 20153
    },
    {
      "epoch": 0.8136455389584174,
      "grad_norm": 77.11381530761719,
      "learning_rate": 1.7669058439291309e-06,
      "loss": 3.1624,
      "step": 20154
    },
    {
      "epoch": 0.8136859103754541,
      "grad_norm": 102.96891784667969,
      "learning_rate": 1.7661637410752474e-06,
      "loss": 3.5622,
      "step": 20155
    },
    {
      "epoch": 0.813726281792491,
      "grad_norm": 104.36261749267578,
      "learning_rate": 1.7654217790007333e-06,
      "loss": 4.3352,
      "step": 20156
    },
    {
      "epoch": 0.8137666532095277,
      "grad_norm": 194.50027465820312,
      "learning_rate": 1.7646799577182783e-06,
      "loss": 3.7074,
      "step": 20157
    },
    {
      "epoch": 0.8138070246265644,
      "grad_norm": 95.20226287841797,
      "learning_rate": 1.7639382772405656e-06,
      "loss": 3.7048,
      "step": 20158
    },
    {
      "epoch": 0.8138473960436011,
      "grad_norm": 153.9757843017578,
      "learning_rate": 1.763196737580276e-06,
      "loss": 5.4745,
      "step": 20159
    },
    {
      "epoch": 0.8138877674606378,
      "grad_norm": 136.86048889160156,
      "learning_rate": 1.7624553387500865e-06,
      "loss": 4.4034,
      "step": 20160
    },
    {
      "epoch": 0.8139281388776746,
      "grad_norm": 94.1732406616211,
      "learning_rate": 1.7617140807626753e-06,
      "loss": 2.6191,
      "step": 20161
    },
    {
      "epoch": 0.8139685102947113,
      "grad_norm": 85.84164428710938,
      "learning_rate": 1.760972963630716e-06,
      "loss": 4.3785,
      "step": 20162
    },
    {
      "epoch": 0.8140088817117481,
      "grad_norm": 82.7778549194336,
      "learning_rate": 1.7602319873668815e-06,
      "loss": 4.2879,
      "step": 20163
    },
    {
      "epoch": 0.8140492531287848,
      "grad_norm": 86.67816925048828,
      "learning_rate": 1.7594911519838364e-06,
      "loss": 4.8469,
      "step": 20164
    },
    {
      "epoch": 0.8140896245458216,
      "grad_norm": 168.60389709472656,
      "learning_rate": 1.7587504574942505e-06,
      "loss": 4.9035,
      "step": 20165
    },
    {
      "epoch": 0.8141299959628583,
      "grad_norm": 104.38928985595703,
      "learning_rate": 1.7580099039107857e-06,
      "loss": 5.3235,
      "step": 20166
    },
    {
      "epoch": 0.814170367379895,
      "grad_norm": 60.895023345947266,
      "learning_rate": 1.7572694912461087e-06,
      "loss": 2.1395,
      "step": 20167
    },
    {
      "epoch": 0.8142107387969317,
      "grad_norm": 81.1627197265625,
      "learning_rate": 1.7565292195128713e-06,
      "loss": 4.9931,
      "step": 20168
    },
    {
      "epoch": 0.8142511102139686,
      "grad_norm": 90.88697814941406,
      "learning_rate": 1.755789088723736e-06,
      "loss": 3.3276,
      "step": 20169
    },
    {
      "epoch": 0.8142914816310053,
      "grad_norm": 142.85414123535156,
      "learning_rate": 1.755049098891355e-06,
      "loss": 4.8431,
      "step": 20170
    },
    {
      "epoch": 0.814331853048042,
      "grad_norm": 117.16133880615234,
      "learning_rate": 1.7543092500283832e-06,
      "loss": 6.3472,
      "step": 20171
    },
    {
      "epoch": 0.8143722244650787,
      "grad_norm": 82.7948226928711,
      "learning_rate": 1.7535695421474664e-06,
      "loss": 4.3018,
      "step": 20172
    },
    {
      "epoch": 0.8144125958821155,
      "grad_norm": 86.0673599243164,
      "learning_rate": 1.7528299752612533e-06,
      "loss": 6.693,
      "step": 20173
    },
    {
      "epoch": 0.8144529672991522,
      "grad_norm": 86.79714965820312,
      "learning_rate": 1.7520905493823882e-06,
      "loss": 4.3958,
      "step": 20174
    },
    {
      "epoch": 0.8144933387161889,
      "grad_norm": 151.2772674560547,
      "learning_rate": 1.7513512645235188e-06,
      "loss": 4.6602,
      "step": 20175
    },
    {
      "epoch": 0.8145337101332257,
      "grad_norm": 131.27040100097656,
      "learning_rate": 1.750612120697277e-06,
      "loss": 4.6041,
      "step": 20176
    },
    {
      "epoch": 0.8145740815502625,
      "grad_norm": 92.64866638183594,
      "learning_rate": 1.7498731179163041e-06,
      "loss": 3.9095,
      "step": 20177
    },
    {
      "epoch": 0.8146144529672992,
      "grad_norm": 101.79623413085938,
      "learning_rate": 1.7491342561932356e-06,
      "loss": 4.2719,
      "step": 20178
    },
    {
      "epoch": 0.8146548243843359,
      "grad_norm": 95.9203872680664,
      "learning_rate": 1.7483955355407046e-06,
      "loss": 5.1526,
      "step": 20179
    },
    {
      "epoch": 0.8146951958013726,
      "grad_norm": 100.98262786865234,
      "learning_rate": 1.7476569559713408e-06,
      "loss": 4.0279,
      "step": 20180
    },
    {
      "epoch": 0.8147355672184093,
      "grad_norm": 79.54981994628906,
      "learning_rate": 1.7469185174977721e-06,
      "loss": 3.7267,
      "step": 20181
    },
    {
      "epoch": 0.8147759386354461,
      "grad_norm": 111.73282623291016,
      "learning_rate": 1.746180220132625e-06,
      "loss": 4.1896,
      "step": 20182
    },
    {
      "epoch": 0.8148163100524829,
      "grad_norm": 85.22251892089844,
      "learning_rate": 1.745442063888525e-06,
      "loss": 5.8738,
      "step": 20183
    },
    {
      "epoch": 0.8148566814695196,
      "grad_norm": 111.62245178222656,
      "learning_rate": 1.7447040487780864e-06,
      "loss": 4.3977,
      "step": 20184
    },
    {
      "epoch": 0.8148970528865563,
      "grad_norm": 92.1823959350586,
      "learning_rate": 1.7439661748139314e-06,
      "loss": 4.4951,
      "step": 20185
    },
    {
      "epoch": 0.8149374243035931,
      "grad_norm": 147.06793212890625,
      "learning_rate": 1.7432284420086755e-06,
      "loss": 7.6819,
      "step": 20186
    },
    {
      "epoch": 0.8149777957206298,
      "grad_norm": 99.39035034179688,
      "learning_rate": 1.742490850374935e-06,
      "loss": 4.9071,
      "step": 20187
    },
    {
      "epoch": 0.8150181671376665,
      "grad_norm": 121.56684112548828,
      "learning_rate": 1.7417533999253156e-06,
      "loss": 5.117,
      "step": 20188
    },
    {
      "epoch": 0.8150585385547032,
      "grad_norm": 78.89522552490234,
      "learning_rate": 1.7410160906724295e-06,
      "loss": 4.0741,
      "step": 20189
    },
    {
      "epoch": 0.8150989099717401,
      "grad_norm": 86.27969360351562,
      "learning_rate": 1.7402789226288807e-06,
      "loss": 4.0211,
      "step": 20190
    },
    {
      "epoch": 0.8151392813887768,
      "grad_norm": 92.65269470214844,
      "learning_rate": 1.7395418958072786e-06,
      "loss": 6.1458,
      "step": 20191
    },
    {
      "epoch": 0.8151796528058135,
      "grad_norm": 113.40271759033203,
      "learning_rate": 1.738805010220217e-06,
      "loss": 3.9435,
      "step": 20192
    },
    {
      "epoch": 0.8152200242228502,
      "grad_norm": 66.34453582763672,
      "learning_rate": 1.7380682658802984e-06,
      "loss": 5.1262,
      "step": 20193
    },
    {
      "epoch": 0.815260395639887,
      "grad_norm": 105.5736083984375,
      "learning_rate": 1.7373316628001191e-06,
      "loss": 4.4284,
      "step": 20194
    },
    {
      "epoch": 0.8153007670569237,
      "grad_norm": 115.26700592041016,
      "learning_rate": 1.736595200992277e-06,
      "loss": 4.8347,
      "step": 20195
    },
    {
      "epoch": 0.8153411384739604,
      "grad_norm": 138.79592895507812,
      "learning_rate": 1.7358588804693578e-06,
      "loss": 4.5005,
      "step": 20196
    },
    {
      "epoch": 0.8153815098909972,
      "grad_norm": 110.18468475341797,
      "learning_rate": 1.7351227012439542e-06,
      "loss": 5.2008,
      "step": 20197
    },
    {
      "epoch": 0.815421881308034,
      "grad_norm": 137.96951293945312,
      "learning_rate": 1.734386663328651e-06,
      "loss": 6.0752,
      "step": 20198
    },
    {
      "epoch": 0.8154622527250707,
      "grad_norm": 86.72530364990234,
      "learning_rate": 1.7336507667360358e-06,
      "loss": 5.0838,
      "step": 20199
    },
    {
      "epoch": 0.8155026241421074,
      "grad_norm": 143.29417419433594,
      "learning_rate": 1.7329150114786885e-06,
      "loss": 4.8353,
      "step": 20200
    },
    {
      "epoch": 0.8155429955591441,
      "grad_norm": 135.38328552246094,
      "learning_rate": 1.7321793975691892e-06,
      "loss": 5.6049,
      "step": 20201
    },
    {
      "epoch": 0.8155833669761808,
      "grad_norm": 104.3456802368164,
      "learning_rate": 1.7314439250201154e-06,
      "loss": 5.5755,
      "step": 20202
    },
    {
      "epoch": 0.8156237383932176,
      "grad_norm": 102.73570251464844,
      "learning_rate": 1.730708593844045e-06,
      "loss": 5.0965,
      "step": 20203
    },
    {
      "epoch": 0.8156641098102544,
      "grad_norm": 93.43711853027344,
      "learning_rate": 1.7299734040535454e-06,
      "loss": 4.9984,
      "step": 20204
    },
    {
      "epoch": 0.8157044812272911,
      "grad_norm": 88.52405548095703,
      "learning_rate": 1.729238355661188e-06,
      "loss": 6.4121,
      "step": 20205
    },
    {
      "epoch": 0.8157448526443278,
      "grad_norm": 109.66407012939453,
      "learning_rate": 1.7285034486795405e-06,
      "loss": 5.7636,
      "step": 20206
    },
    {
      "epoch": 0.8157852240613646,
      "grad_norm": 70.37561798095703,
      "learning_rate": 1.7277686831211727e-06,
      "loss": 3.6865,
      "step": 20207
    },
    {
      "epoch": 0.8158255954784013,
      "grad_norm": 73.66851806640625,
      "learning_rate": 1.72703405899864e-06,
      "loss": 4.4913,
      "step": 20208
    },
    {
      "epoch": 0.815865966895438,
      "grad_norm": 124.65752410888672,
      "learning_rate": 1.726299576324506e-06,
      "loss": 4.1962,
      "step": 20209
    },
    {
      "epoch": 0.8159063383124747,
      "grad_norm": 80.29850006103516,
      "learning_rate": 1.7255652351113283e-06,
      "loss": 4.5611,
      "step": 20210
    },
    {
      "epoch": 0.8159467097295116,
      "grad_norm": 77.90470123291016,
      "learning_rate": 1.7248310353716669e-06,
      "loss": 4.7546,
      "step": 20211
    },
    {
      "epoch": 0.8159870811465483,
      "grad_norm": 105.84625244140625,
      "learning_rate": 1.7240969771180672e-06,
      "loss": 4.2057,
      "step": 20212
    },
    {
      "epoch": 0.816027452563585,
      "grad_norm": 114.32536315917969,
      "learning_rate": 1.7233630603630836e-06,
      "loss": 5.4293,
      "step": 20213
    },
    {
      "epoch": 0.8160678239806217,
      "grad_norm": 85.26521301269531,
      "learning_rate": 1.722629285119265e-06,
      "loss": 5.3146,
      "step": 20214
    },
    {
      "epoch": 0.8161081953976584,
      "grad_norm": 236.2149658203125,
      "learning_rate": 1.7218956513991581e-06,
      "loss": 4.7434,
      "step": 20215
    },
    {
      "epoch": 0.8161485668146952,
      "grad_norm": 87.79656219482422,
      "learning_rate": 1.721162159215304e-06,
      "loss": 5.6522,
      "step": 20216
    },
    {
      "epoch": 0.8161889382317319,
      "grad_norm": 65.62417602539062,
      "learning_rate": 1.7204288085802423e-06,
      "loss": 5.0801,
      "step": 20217
    },
    {
      "epoch": 0.8162293096487687,
      "grad_norm": 80.64846801757812,
      "learning_rate": 1.7196955995065156e-06,
      "loss": 1.808,
      "step": 20218
    },
    {
      "epoch": 0.8162696810658054,
      "grad_norm": 160.09925842285156,
      "learning_rate": 1.7189625320066573e-06,
      "loss": 3.5153,
      "step": 20219
    },
    {
      "epoch": 0.8163100524828422,
      "grad_norm": 1276.633056640625,
      "learning_rate": 1.7182296060932024e-06,
      "loss": 5.3724,
      "step": 20220
    },
    {
      "epoch": 0.8163504238998789,
      "grad_norm": 144.4215545654297,
      "learning_rate": 1.7174968217786813e-06,
      "loss": 5.9629,
      "step": 20221
    },
    {
      "epoch": 0.8163907953169156,
      "grad_norm": 107.0958251953125,
      "learning_rate": 1.716764179075624e-06,
      "loss": 5.6165,
      "step": 20222
    },
    {
      "epoch": 0.8164311667339523,
      "grad_norm": 140.8633575439453,
      "learning_rate": 1.716031677996559e-06,
      "loss": 4.2602,
      "step": 20223
    },
    {
      "epoch": 0.816471538150989,
      "grad_norm": 107.82534790039062,
      "learning_rate": 1.7152993185540057e-06,
      "loss": 3.5209,
      "step": 20224
    },
    {
      "epoch": 0.8165119095680259,
      "grad_norm": 133.06976318359375,
      "learning_rate": 1.7145671007604881e-06,
      "loss": 5.7181,
      "step": 20225
    },
    {
      "epoch": 0.8165522809850626,
      "grad_norm": 98.36953735351562,
      "learning_rate": 1.713835024628524e-06,
      "loss": 4.7993,
      "step": 20226
    },
    {
      "epoch": 0.8165926524020993,
      "grad_norm": 146.6813507080078,
      "learning_rate": 1.7131030901706347e-06,
      "loss": 4.1215,
      "step": 20227
    },
    {
      "epoch": 0.816633023819136,
      "grad_norm": 84.61125183105469,
      "learning_rate": 1.7123712973993289e-06,
      "loss": 4.1268,
      "step": 20228
    },
    {
      "epoch": 0.8166733952361728,
      "grad_norm": 78.89315795898438,
      "learning_rate": 1.7116396463271202e-06,
      "loss": 5.0167,
      "step": 20229
    },
    {
      "epoch": 0.8167137666532095,
      "grad_norm": 108.42107391357422,
      "learning_rate": 1.7109081369665193e-06,
      "loss": 5.392,
      "step": 20230
    },
    {
      "epoch": 0.8167541380702462,
      "grad_norm": 102.08304595947266,
      "learning_rate": 1.7101767693300352e-06,
      "loss": 7.1779,
      "step": 20231
    },
    {
      "epoch": 0.816794509487283,
      "grad_norm": 131.09063720703125,
      "learning_rate": 1.7094455434301683e-06,
      "loss": 4.862,
      "step": 20232
    },
    {
      "epoch": 0.8168348809043198,
      "grad_norm": 95.06757354736328,
      "learning_rate": 1.7087144592794225e-06,
      "loss": 4.3329,
      "step": 20233
    },
    {
      "epoch": 0.8168752523213565,
      "grad_norm": 107.81855010986328,
      "learning_rate": 1.707983516890297e-06,
      "loss": 5.2286,
      "step": 20234
    },
    {
      "epoch": 0.8169156237383932,
      "grad_norm": 75.46249389648438,
      "learning_rate": 1.707252716275294e-06,
      "loss": 2.9224,
      "step": 20235
    },
    {
      "epoch": 0.8169559951554299,
      "grad_norm": 73.006591796875,
      "learning_rate": 1.706522057446901e-06,
      "loss": 5.2944,
      "step": 20236
    },
    {
      "epoch": 0.8169963665724667,
      "grad_norm": 150.97642517089844,
      "learning_rate": 1.705791540417615e-06,
      "loss": 5.2313,
      "step": 20237
    },
    {
      "epoch": 0.8170367379895034,
      "grad_norm": 120.28109741210938,
      "learning_rate": 1.705061165199925e-06,
      "loss": 4.7045,
      "step": 20238
    },
    {
      "epoch": 0.8170771094065402,
      "grad_norm": 70.0794448852539,
      "learning_rate": 1.7043309318063184e-06,
      "loss": 4.2678,
      "step": 20239
    },
    {
      "epoch": 0.8171174808235769,
      "grad_norm": 244.95407104492188,
      "learning_rate": 1.703600840249281e-06,
      "loss": 5.2139,
      "step": 20240
    },
    {
      "epoch": 0.8171578522406137,
      "grad_norm": 90.56480407714844,
      "learning_rate": 1.7028708905412971e-06,
      "loss": 4.1661,
      "step": 20241
    },
    {
      "epoch": 0.8171982236576504,
      "grad_norm": 95.44979095458984,
      "learning_rate": 1.7021410826948448e-06,
      "loss": 4.1267,
      "step": 20242
    },
    {
      "epoch": 0.8172385950746871,
      "grad_norm": 89.94476318359375,
      "learning_rate": 1.701411416722406e-06,
      "loss": 4.243,
      "step": 20243
    },
    {
      "epoch": 0.8172789664917238,
      "grad_norm": 83.07984924316406,
      "learning_rate": 1.7006818926364498e-06,
      "loss": 4.8835,
      "step": 20244
    },
    {
      "epoch": 0.8173193379087605,
      "grad_norm": 97.35112762451172,
      "learning_rate": 1.699952510449454e-06,
      "loss": 5.8034,
      "step": 20245
    },
    {
      "epoch": 0.8173597093257974,
      "grad_norm": 92.96513366699219,
      "learning_rate": 1.6992232701738888e-06,
      "loss": 5.9114,
      "step": 20246
    },
    {
      "epoch": 0.8174000807428341,
      "grad_norm": 129.74081420898438,
      "learning_rate": 1.6984941718222236e-06,
      "loss": 4.7834,
      "step": 20247
    },
    {
      "epoch": 0.8174404521598708,
      "grad_norm": 93.50009155273438,
      "learning_rate": 1.6977652154069201e-06,
      "loss": 5.2728,
      "step": 20248
    },
    {
      "epoch": 0.8174808235769075,
      "grad_norm": 128.66510009765625,
      "learning_rate": 1.6970364009404439e-06,
      "loss": 4.7728,
      "step": 20249
    },
    {
      "epoch": 0.8175211949939443,
      "grad_norm": 145.66419982910156,
      "learning_rate": 1.696307728435257e-06,
      "loss": 3.587,
      "step": 20250
    },
    {
      "epoch": 0.817561566410981,
      "grad_norm": 102.20533752441406,
      "learning_rate": 1.6955791979038205e-06,
      "loss": 4.5101,
      "step": 20251
    },
    {
      "epoch": 0.8176019378280177,
      "grad_norm": 105.68679809570312,
      "learning_rate": 1.6948508093585836e-06,
      "loss": 5.9434,
      "step": 20252
    },
    {
      "epoch": 0.8176423092450545,
      "grad_norm": 83.66674041748047,
      "learning_rate": 1.694122562812005e-06,
      "loss": 5.8392,
      "step": 20253
    },
    {
      "epoch": 0.8176826806620913,
      "grad_norm": 124.5389175415039,
      "learning_rate": 1.6933944582765348e-06,
      "loss": 5.1065,
      "step": 20254
    },
    {
      "epoch": 0.817723052079128,
      "grad_norm": 104.03020477294922,
      "learning_rate": 1.6926664957646243e-06,
      "loss": 5.4312,
      "step": 20255
    },
    {
      "epoch": 0.8177634234961647,
      "grad_norm": 104.57576751708984,
      "learning_rate": 1.6919386752887157e-06,
      "loss": 5.6961,
      "step": 20256
    },
    {
      "epoch": 0.8178037949132014,
      "grad_norm": 164.04592895507812,
      "learning_rate": 1.6912109968612545e-06,
      "loss": 6.1456,
      "step": 20257
    },
    {
      "epoch": 0.8178441663302382,
      "grad_norm": 90.14067077636719,
      "learning_rate": 1.690483460494683e-06,
      "loss": 6.5128,
      "step": 20258
    },
    {
      "epoch": 0.8178845377472749,
      "grad_norm": 59.58245849609375,
      "learning_rate": 1.68975606620144e-06,
      "loss": 4.7986,
      "step": 20259
    },
    {
      "epoch": 0.8179249091643117,
      "grad_norm": 90.54064178466797,
      "learning_rate": 1.6890288139939625e-06,
      "loss": 5.9018,
      "step": 20260
    },
    {
      "epoch": 0.8179652805813484,
      "grad_norm": 145.16671752929688,
      "learning_rate": 1.688301703884686e-06,
      "loss": 5.7221,
      "step": 20261
    },
    {
      "epoch": 0.8180056519983852,
      "grad_norm": 78.5468978881836,
      "learning_rate": 1.6875747358860395e-06,
      "loss": 4.308,
      "step": 20262
    },
    {
      "epoch": 0.8180460234154219,
      "grad_norm": 74.50695037841797,
      "learning_rate": 1.6868479100104573e-06,
      "loss": 5.3132,
      "step": 20263
    },
    {
      "epoch": 0.8180863948324586,
      "grad_norm": 85.05786895751953,
      "learning_rate": 1.6861212262703608e-06,
      "loss": 5.2246,
      "step": 20264
    },
    {
      "epoch": 0.8181267662494953,
      "grad_norm": 122.39893341064453,
      "learning_rate": 1.6853946846781765e-06,
      "loss": 5.0554,
      "step": 20265
    },
    {
      "epoch": 0.818167137666532,
      "grad_norm": 79.82596588134766,
      "learning_rate": 1.6846682852463282e-06,
      "loss": 5.2634,
      "step": 20266
    },
    {
      "epoch": 0.8182075090835689,
      "grad_norm": 176.4664306640625,
      "learning_rate": 1.6839420279872353e-06,
      "loss": 6.4522,
      "step": 20267
    },
    {
      "epoch": 0.8182478805006056,
      "grad_norm": 84.1817398071289,
      "learning_rate": 1.683215912913313e-06,
      "loss": 5.3363,
      "step": 20268
    },
    {
      "epoch": 0.8182882519176423,
      "grad_norm": 103.7376708984375,
      "learning_rate": 1.6824899400369776e-06,
      "loss": 4.9005,
      "step": 20269
    },
    {
      "epoch": 0.818328623334679,
      "grad_norm": 91.86627197265625,
      "learning_rate": 1.6817641093706416e-06,
      "loss": 5.3377,
      "step": 20270
    },
    {
      "epoch": 0.8183689947517158,
      "grad_norm": 90.0680923461914,
      "learning_rate": 1.6810384209267173e-06,
      "loss": 4.2182,
      "step": 20271
    },
    {
      "epoch": 0.8184093661687525,
      "grad_norm": 98.38471984863281,
      "learning_rate": 1.6803128747176067e-06,
      "loss": 5.4734,
      "step": 20272
    },
    {
      "epoch": 0.8184497375857893,
      "grad_norm": 173.30484008789062,
      "learning_rate": 1.679587470755718e-06,
      "loss": 4.4477,
      "step": 20273
    },
    {
      "epoch": 0.818490109002826,
      "grad_norm": 178.8437042236328,
      "learning_rate": 1.6788622090534545e-06,
      "loss": 6.8719,
      "step": 20274
    },
    {
      "epoch": 0.8185304804198628,
      "grad_norm": 107.2406234741211,
      "learning_rate": 1.6781370896232185e-06,
      "loss": 5.1132,
      "step": 20275
    },
    {
      "epoch": 0.8185708518368995,
      "grad_norm": 80.47403717041016,
      "learning_rate": 1.6774121124774023e-06,
      "loss": 5.6408,
      "step": 20276
    },
    {
      "epoch": 0.8186112232539362,
      "grad_norm": 81.5254135131836,
      "learning_rate": 1.676687277628405e-06,
      "loss": 6.1623,
      "step": 20277
    },
    {
      "epoch": 0.8186515946709729,
      "grad_norm": 106.7886962890625,
      "learning_rate": 1.6759625850886185e-06,
      "loss": 3.5164,
      "step": 20278
    },
    {
      "epoch": 0.8186919660880096,
      "grad_norm": 88.72006225585938,
      "learning_rate": 1.6752380348704333e-06,
      "loss": 4.095,
      "step": 20279
    },
    {
      "epoch": 0.8187323375050465,
      "grad_norm": 78.5360107421875,
      "learning_rate": 1.6745136269862383e-06,
      "loss": 5.252,
      "step": 20280
    },
    {
      "epoch": 0.8187727089220832,
      "grad_norm": 86.99854278564453,
      "learning_rate": 1.673789361448419e-06,
      "loss": 5.7272,
      "step": 20281
    },
    {
      "epoch": 0.8188130803391199,
      "grad_norm": 128.55482482910156,
      "learning_rate": 1.673065238269359e-06,
      "loss": 6.1426,
      "step": 20282
    },
    {
      "epoch": 0.8188534517561566,
      "grad_norm": 105.76383209228516,
      "learning_rate": 1.6723412574614406e-06,
      "loss": 6.3957,
      "step": 20283
    },
    {
      "epoch": 0.8188938231731934,
      "grad_norm": 86.30826568603516,
      "learning_rate": 1.6716174190370382e-06,
      "loss": 5.544,
      "step": 20284
    },
    {
      "epoch": 0.8189341945902301,
      "grad_norm": 79.00663757324219,
      "learning_rate": 1.6708937230085298e-06,
      "loss": 4.6329,
      "step": 20285
    },
    {
      "epoch": 0.8189745660072668,
      "grad_norm": 89.37637329101562,
      "learning_rate": 1.6701701693882877e-06,
      "loss": 4.0791,
      "step": 20286
    },
    {
      "epoch": 0.8190149374243036,
      "grad_norm": 122.43272399902344,
      "learning_rate": 1.6694467581886874e-06,
      "loss": 5.3455,
      "step": 20287
    },
    {
      "epoch": 0.8190553088413404,
      "grad_norm": 70.98564910888672,
      "learning_rate": 1.6687234894220928e-06,
      "loss": 5.1912,
      "step": 20288
    },
    {
      "epoch": 0.8190956802583771,
      "grad_norm": 90.53993225097656,
      "learning_rate": 1.66800036310087e-06,
      "loss": 4.6025,
      "step": 20289
    },
    {
      "epoch": 0.8191360516754138,
      "grad_norm": 87.60546112060547,
      "learning_rate": 1.6672773792373853e-06,
      "loss": 5.7376,
      "step": 20290
    },
    {
      "epoch": 0.8191764230924505,
      "grad_norm": 160.4211883544922,
      "learning_rate": 1.6665545378440018e-06,
      "loss": 4.9642,
      "step": 20291
    },
    {
      "epoch": 0.8192167945094873,
      "grad_norm": 94.81300354003906,
      "learning_rate": 1.6658318389330719e-06,
      "loss": 4.3761,
      "step": 20292
    },
    {
      "epoch": 0.819257165926524,
      "grad_norm": 87.445068359375,
      "learning_rate": 1.6651092825169569e-06,
      "loss": 4.9534,
      "step": 20293
    },
    {
      "epoch": 0.8192975373435608,
      "grad_norm": 167.15982055664062,
      "learning_rate": 1.6643868686080101e-06,
      "loss": 6.2563,
      "step": 20294
    },
    {
      "epoch": 0.8193379087605975,
      "grad_norm": 100.91182708740234,
      "learning_rate": 1.663664597218585e-06,
      "loss": 4.4077,
      "step": 20295
    },
    {
      "epoch": 0.8193782801776343,
      "grad_norm": 146.0043182373047,
      "learning_rate": 1.6629424683610252e-06,
      "loss": 3.8647,
      "step": 20296
    },
    {
      "epoch": 0.819418651594671,
      "grad_norm": 159.33441162109375,
      "learning_rate": 1.6622204820476817e-06,
      "loss": 5.4659,
      "step": 20297
    },
    {
      "epoch": 0.8194590230117077,
      "grad_norm": 79.95879364013672,
      "learning_rate": 1.661498638290897e-06,
      "loss": 3.8246,
      "step": 20298
    },
    {
      "epoch": 0.8194993944287444,
      "grad_norm": 110.96102142333984,
      "learning_rate": 1.6607769371030146e-06,
      "loss": 5.7578,
      "step": 20299
    },
    {
      "epoch": 0.8195397658457811,
      "grad_norm": 88.56998443603516,
      "learning_rate": 1.6600553784963735e-06,
      "loss": 4.8785,
      "step": 20300
    },
    {
      "epoch": 0.819580137262818,
      "grad_norm": 90.58668518066406,
      "learning_rate": 1.6593339624833093e-06,
      "loss": 4.3902,
      "step": 20301
    },
    {
      "epoch": 0.8196205086798547,
      "grad_norm": 70.8233871459961,
      "learning_rate": 1.6586126890761579e-06,
      "loss": 5.4884,
      "step": 20302
    },
    {
      "epoch": 0.8196608800968914,
      "grad_norm": 96.16735076904297,
      "learning_rate": 1.6578915582872536e-06,
      "loss": 4.5121,
      "step": 20303
    },
    {
      "epoch": 0.8197012515139281,
      "grad_norm": 92.1661605834961,
      "learning_rate": 1.6571705701289199e-06,
      "loss": 4.7039,
      "step": 20304
    },
    {
      "epoch": 0.8197416229309649,
      "grad_norm": 76.04612731933594,
      "learning_rate": 1.6564497246134891e-06,
      "loss": 4.3626,
      "step": 20305
    },
    {
      "epoch": 0.8197819943480016,
      "grad_norm": 101.07038116455078,
      "learning_rate": 1.655729021753283e-06,
      "loss": 4.8514,
      "step": 20306
    },
    {
      "epoch": 0.8198223657650383,
      "grad_norm": 118.13655853271484,
      "learning_rate": 1.6550084615606288e-06,
      "loss": 4.4166,
      "step": 20307
    },
    {
      "epoch": 0.8198627371820751,
      "grad_norm": 95.11048126220703,
      "learning_rate": 1.6542880440478404e-06,
      "loss": 4.6796,
      "step": 20308
    },
    {
      "epoch": 0.8199031085991119,
      "grad_norm": 87.84678649902344,
      "learning_rate": 1.6535677692272367e-06,
      "loss": 4.392,
      "step": 20309
    },
    {
      "epoch": 0.8199434800161486,
      "grad_norm": 100.07083892822266,
      "learning_rate": 1.6528476371111358e-06,
      "loss": 4.8972,
      "step": 20310
    },
    {
      "epoch": 0.8199838514331853,
      "grad_norm": 93.36553192138672,
      "learning_rate": 1.6521276477118487e-06,
      "loss": 4.9694,
      "step": 20311
    },
    {
      "epoch": 0.820024222850222,
      "grad_norm": 83.66754150390625,
      "learning_rate": 1.6514078010416846e-06,
      "loss": 5.6806,
      "step": 20312
    },
    {
      "epoch": 0.8200645942672588,
      "grad_norm": 77.83170318603516,
      "learning_rate": 1.6506880971129502e-06,
      "loss": 4.5919,
      "step": 20313
    },
    {
      "epoch": 0.8201049656842955,
      "grad_norm": 121.45185852050781,
      "learning_rate": 1.6499685359379524e-06,
      "loss": 3.9216,
      "step": 20314
    },
    {
      "epoch": 0.8201453371013323,
      "grad_norm": 113.53365325927734,
      "learning_rate": 1.649249117528995e-06,
      "loss": 4.008,
      "step": 20315
    },
    {
      "epoch": 0.820185708518369,
      "grad_norm": 79.2745132446289,
      "learning_rate": 1.6485298418983764e-06,
      "loss": 4.1749,
      "step": 20316
    },
    {
      "epoch": 0.8202260799354057,
      "grad_norm": 105.52467346191406,
      "learning_rate": 1.6478107090583984e-06,
      "loss": 5.6949,
      "step": 20317
    },
    {
      "epoch": 0.8202664513524425,
      "grad_norm": 167.00506591796875,
      "learning_rate": 1.647091719021351e-06,
      "loss": 4.2891,
      "step": 20318
    },
    {
      "epoch": 0.8203068227694792,
      "grad_norm": 60.906166076660156,
      "learning_rate": 1.6463728717995298e-06,
      "loss": 2.9842,
      "step": 20319
    },
    {
      "epoch": 0.8203471941865159,
      "grad_norm": 91.95663452148438,
      "learning_rate": 1.645654167405225e-06,
      "loss": 4.5242,
      "step": 20320
    },
    {
      "epoch": 0.8203875656035526,
      "grad_norm": 71.14897155761719,
      "learning_rate": 1.644935605850725e-06,
      "loss": 4.1137,
      "step": 20321
    },
    {
      "epoch": 0.8204279370205895,
      "grad_norm": 113.68083953857422,
      "learning_rate": 1.644217187148316e-06,
      "loss": 4.9742,
      "step": 20322
    },
    {
      "epoch": 0.8204683084376262,
      "grad_norm": 138.0810089111328,
      "learning_rate": 1.6434989113102828e-06,
      "loss": 5.6052,
      "step": 20323
    },
    {
      "epoch": 0.8205086798546629,
      "grad_norm": 73.72801208496094,
      "learning_rate": 1.6427807783489025e-06,
      "loss": 5.5159,
      "step": 20324
    },
    {
      "epoch": 0.8205490512716996,
      "grad_norm": 103.88739013671875,
      "learning_rate": 1.642062788276455e-06,
      "loss": 3.7366,
      "step": 20325
    },
    {
      "epoch": 0.8205894226887364,
      "grad_norm": 85.75897216796875,
      "learning_rate": 1.6413449411052173e-06,
      "loss": 5.5386,
      "step": 20326
    },
    {
      "epoch": 0.8206297941057731,
      "grad_norm": 83.54170989990234,
      "learning_rate": 1.640627236847465e-06,
      "loss": 4.3131,
      "step": 20327
    },
    {
      "epoch": 0.8206701655228098,
      "grad_norm": 82.17906951904297,
      "learning_rate": 1.6399096755154632e-06,
      "loss": 4.2986,
      "step": 20328
    },
    {
      "epoch": 0.8207105369398466,
      "grad_norm": 92.49856567382812,
      "learning_rate": 1.6391922571214835e-06,
      "loss": 4.8684,
      "step": 20329
    },
    {
      "epoch": 0.8207509083568834,
      "grad_norm": 81.73340606689453,
      "learning_rate": 1.6384749816777922e-06,
      "loss": 2.451,
      "step": 20330
    },
    {
      "epoch": 0.8207912797739201,
      "grad_norm": 90.81360626220703,
      "learning_rate": 1.6377578491966562e-06,
      "loss": 4.2665,
      "step": 20331
    },
    {
      "epoch": 0.8208316511909568,
      "grad_norm": 70.76492309570312,
      "learning_rate": 1.637040859690332e-06,
      "loss": 3.632,
      "step": 20332
    },
    {
      "epoch": 0.8208720226079935,
      "grad_norm": 73.8199462890625,
      "learning_rate": 1.636324013171079e-06,
      "loss": 5.0173,
      "step": 20333
    },
    {
      "epoch": 0.8209123940250302,
      "grad_norm": 78.54412078857422,
      "learning_rate": 1.6356073096511549e-06,
      "loss": 4.4306,
      "step": 20334
    },
    {
      "epoch": 0.820952765442067,
      "grad_norm": 92.6790542602539,
      "learning_rate": 1.6348907491428146e-06,
      "loss": 4.7261,
      "step": 20335
    },
    {
      "epoch": 0.8209931368591038,
      "grad_norm": 94.3174057006836,
      "learning_rate": 1.634174331658307e-06,
      "loss": 5.5462,
      "step": 20336
    },
    {
      "epoch": 0.8210335082761405,
      "grad_norm": 199.45555114746094,
      "learning_rate": 1.6334580572098834e-06,
      "loss": 3.5934,
      "step": 20337
    },
    {
      "epoch": 0.8210738796931772,
      "grad_norm": 140.21481323242188,
      "learning_rate": 1.63274192580979e-06,
      "loss": 3.7985,
      "step": 20338
    },
    {
      "epoch": 0.821114251110214,
      "grad_norm": 122.66568756103516,
      "learning_rate": 1.6320259374702729e-06,
      "loss": 4.3781,
      "step": 20339
    },
    {
      "epoch": 0.8211546225272507,
      "grad_norm": 80.99308776855469,
      "learning_rate": 1.6313100922035697e-06,
      "loss": 5.4038,
      "step": 20340
    },
    {
      "epoch": 0.8211949939442874,
      "grad_norm": 124.08416748046875,
      "learning_rate": 1.6305943900219213e-06,
      "loss": 3.0955,
      "step": 20341
    },
    {
      "epoch": 0.8212353653613241,
      "grad_norm": 98.18190002441406,
      "learning_rate": 1.6298788309375646e-06,
      "loss": 2.98,
      "step": 20342
    },
    {
      "epoch": 0.821275736778361,
      "grad_norm": 106.29109191894531,
      "learning_rate": 1.6291634149627366e-06,
      "loss": 4.7303,
      "step": 20343
    },
    {
      "epoch": 0.8213161081953977,
      "grad_norm": 140.1630401611328,
      "learning_rate": 1.6284481421096644e-06,
      "loss": 5.8771,
      "step": 20344
    },
    {
      "epoch": 0.8213564796124344,
      "grad_norm": 141.4964599609375,
      "learning_rate": 1.627733012390581e-06,
      "loss": 5.4167,
      "step": 20345
    },
    {
      "epoch": 0.8213968510294711,
      "grad_norm": 101.96046447753906,
      "learning_rate": 1.6270180258177115e-06,
      "loss": 5.248,
      "step": 20346
    },
    {
      "epoch": 0.8214372224465079,
      "grad_norm": 83.16922760009766,
      "learning_rate": 1.6263031824032837e-06,
      "loss": 6.2061,
      "step": 20347
    },
    {
      "epoch": 0.8214775938635446,
      "grad_norm": 76.14967346191406,
      "learning_rate": 1.6255884821595159e-06,
      "loss": 5.0222,
      "step": 20348
    },
    {
      "epoch": 0.8215179652805813,
      "grad_norm": 90.07779693603516,
      "learning_rate": 1.624873925098629e-06,
      "loss": 5.0063,
      "step": 20349
    },
    {
      "epoch": 0.8215583366976181,
      "grad_norm": 116.65035247802734,
      "learning_rate": 1.62415951123284e-06,
      "loss": 6.4346,
      "step": 20350
    },
    {
      "epoch": 0.8215987081146549,
      "grad_norm": 103.46505737304688,
      "learning_rate": 1.6234452405743673e-06,
      "loss": 5.2075,
      "step": 20351
    },
    {
      "epoch": 0.8216390795316916,
      "grad_norm": 89.80841064453125,
      "learning_rate": 1.6227311131354184e-06,
      "loss": 6.2437,
      "step": 20352
    },
    {
      "epoch": 0.8216794509487283,
      "grad_norm": 119.97257232666016,
      "learning_rate": 1.622017128928205e-06,
      "loss": 4.5645,
      "step": 20353
    },
    {
      "epoch": 0.821719822365765,
      "grad_norm": 67.91820526123047,
      "learning_rate": 1.6213032879649338e-06,
      "loss": 4.981,
      "step": 20354
    },
    {
      "epoch": 0.8217601937828017,
      "grad_norm": 74.62338256835938,
      "learning_rate": 1.6205895902578118e-06,
      "loss": 6.1497,
      "step": 20355
    },
    {
      "epoch": 0.8218005651998385,
      "grad_norm": 83.82921600341797,
      "learning_rate": 1.6198760358190401e-06,
      "loss": 3.9323,
      "step": 20356
    },
    {
      "epoch": 0.8218409366168753,
      "grad_norm": 111.0337905883789,
      "learning_rate": 1.6191626246608194e-06,
      "loss": 4.5339,
      "step": 20357
    },
    {
      "epoch": 0.821881308033912,
      "grad_norm": 114.37266540527344,
      "learning_rate": 1.6184493567953463e-06,
      "loss": 5.9716,
      "step": 20358
    },
    {
      "epoch": 0.8219216794509487,
      "grad_norm": 116.85078430175781,
      "learning_rate": 1.6177362322348201e-06,
      "loss": 5.7308,
      "step": 20359
    },
    {
      "epoch": 0.8219620508679855,
      "grad_norm": 111.9199447631836,
      "learning_rate": 1.6170232509914275e-06,
      "loss": 5.959,
      "step": 20360
    },
    {
      "epoch": 0.8220024222850222,
      "grad_norm": 104.59949493408203,
      "learning_rate": 1.6163104130773621e-06,
      "loss": 4.0481,
      "step": 20361
    },
    {
      "epoch": 0.8220427937020589,
      "grad_norm": 118.11067962646484,
      "learning_rate": 1.6155977185048111e-06,
      "loss": 5.7924,
      "step": 20362
    },
    {
      "epoch": 0.8220831651190956,
      "grad_norm": 86.390380859375,
      "learning_rate": 1.6148851672859622e-06,
      "loss": 5.4381,
      "step": 20363
    },
    {
      "epoch": 0.8221235365361325,
      "grad_norm": 217.80332946777344,
      "learning_rate": 1.6141727594329936e-06,
      "loss": 6.2013,
      "step": 20364
    },
    {
      "epoch": 0.8221639079531692,
      "grad_norm": 120.46581268310547,
      "learning_rate": 1.6134604949580901e-06,
      "loss": 3.8824,
      "step": 20365
    },
    {
      "epoch": 0.8222042793702059,
      "grad_norm": 170.93626403808594,
      "learning_rate": 1.6127483738734261e-06,
      "loss": 5.3994,
      "step": 20366
    },
    {
      "epoch": 0.8222446507872426,
      "grad_norm": 83.23920440673828,
      "learning_rate": 1.6120363961911833e-06,
      "loss": 4.7398,
      "step": 20367
    },
    {
      "epoch": 0.8222850222042793,
      "grad_norm": 75.54782104492188,
      "learning_rate": 1.6113245619235274e-06,
      "loss": 3.9127,
      "step": 20368
    },
    {
      "epoch": 0.8223253936213161,
      "grad_norm": 87.2802963256836,
      "learning_rate": 1.610612871082633e-06,
      "loss": 5.5074,
      "step": 20369
    },
    {
      "epoch": 0.8223657650383528,
      "grad_norm": 154.87356567382812,
      "learning_rate": 1.6099013236806681e-06,
      "loss": 3.5004,
      "step": 20370
    },
    {
      "epoch": 0.8224061364553896,
      "grad_norm": 131.76536560058594,
      "learning_rate": 1.6091899197298e-06,
      "loss": 3.6525,
      "step": 20371
    },
    {
      "epoch": 0.8224465078724263,
      "grad_norm": 142.4345245361328,
      "learning_rate": 1.6084786592421887e-06,
      "loss": 5.468,
      "step": 20372
    },
    {
      "epoch": 0.8224868792894631,
      "grad_norm": 96.48140716552734,
      "learning_rate": 1.6077675422299966e-06,
      "loss": 5.1392,
      "step": 20373
    },
    {
      "epoch": 0.8225272507064998,
      "grad_norm": 110.54866790771484,
      "learning_rate": 1.6070565687053818e-06,
      "loss": 5.7016,
      "step": 20374
    },
    {
      "epoch": 0.8225676221235365,
      "grad_norm": 99.94535827636719,
      "learning_rate": 1.6063457386805004e-06,
      "loss": 4.4812,
      "step": 20375
    },
    {
      "epoch": 0.8226079935405732,
      "grad_norm": 129.6982421875,
      "learning_rate": 1.6056350521675067e-06,
      "loss": 4.9446,
      "step": 20376
    },
    {
      "epoch": 0.82264836495761,
      "grad_norm": 131.56405639648438,
      "learning_rate": 1.6049245091785526e-06,
      "loss": 5.473,
      "step": 20377
    },
    {
      "epoch": 0.8226887363746468,
      "grad_norm": 1033.0592041015625,
      "learning_rate": 1.6042141097257834e-06,
      "loss": 5.8289,
      "step": 20378
    },
    {
      "epoch": 0.8227291077916835,
      "grad_norm": 103.30858612060547,
      "learning_rate": 1.603503853821351e-06,
      "loss": 5.5245,
      "step": 20379
    },
    {
      "epoch": 0.8227694792087202,
      "grad_norm": 75.63916778564453,
      "learning_rate": 1.602793741477393e-06,
      "loss": 3.7145,
      "step": 20380
    },
    {
      "epoch": 0.822809850625757,
      "grad_norm": 106.5302963256836,
      "learning_rate": 1.6020837727060522e-06,
      "loss": 4.5614,
      "step": 20381
    },
    {
      "epoch": 0.8228502220427937,
      "grad_norm": 301.4417419433594,
      "learning_rate": 1.6013739475194701e-06,
      "loss": 4.4585,
      "step": 20382
    },
    {
      "epoch": 0.8228905934598304,
      "grad_norm": 129.25416564941406,
      "learning_rate": 1.6006642659297822e-06,
      "loss": 5.653,
      "step": 20383
    },
    {
      "epoch": 0.8229309648768672,
      "grad_norm": 90.9826889038086,
      "learning_rate": 1.599954727949119e-06,
      "loss": 3.6091,
      "step": 20384
    },
    {
      "epoch": 0.822971336293904,
      "grad_norm": 86.00394439697266,
      "learning_rate": 1.5992453335896153e-06,
      "loss": 6.3937,
      "step": 20385
    },
    {
      "epoch": 0.8230117077109407,
      "grad_norm": 88.96752166748047,
      "learning_rate": 1.598536082863399e-06,
      "loss": 4.7136,
      "step": 20386
    },
    {
      "epoch": 0.8230520791279774,
      "grad_norm": 119.50972747802734,
      "learning_rate": 1.5978269757825993e-06,
      "loss": 4.4146,
      "step": 20387
    },
    {
      "epoch": 0.8230924505450141,
      "grad_norm": 119.68778228759766,
      "learning_rate": 1.5971180123593356e-06,
      "loss": 7.1711,
      "step": 20388
    },
    {
      "epoch": 0.8231328219620508,
      "grad_norm": 73.8104019165039,
      "learning_rate": 1.5964091926057312e-06,
      "loss": 3.7869,
      "step": 20389
    },
    {
      "epoch": 0.8231731933790876,
      "grad_norm": 102.30935668945312,
      "learning_rate": 1.5957005165339056e-06,
      "loss": 7.024,
      "step": 20390
    },
    {
      "epoch": 0.8232135647961244,
      "grad_norm": 115.73014831542969,
      "learning_rate": 1.5949919841559791e-06,
      "loss": 6.2553,
      "step": 20391
    },
    {
      "epoch": 0.8232539362131611,
      "grad_norm": 85.8036880493164,
      "learning_rate": 1.5942835954840597e-06,
      "loss": 4.916,
      "step": 20392
    },
    {
      "epoch": 0.8232943076301978,
      "grad_norm": 104.87808227539062,
      "learning_rate": 1.593575350530262e-06,
      "loss": 2.9201,
      "step": 20393
    },
    {
      "epoch": 0.8233346790472346,
      "grad_norm": 104.93494415283203,
      "learning_rate": 1.5928672493066955e-06,
      "loss": 2.5455,
      "step": 20394
    },
    {
      "epoch": 0.8233750504642713,
      "grad_norm": 106.55059814453125,
      "learning_rate": 1.5921592918254658e-06,
      "loss": 5.3909,
      "step": 20395
    },
    {
      "epoch": 0.823415421881308,
      "grad_norm": 83.8047866821289,
      "learning_rate": 1.59145147809868e-06,
      "loss": 5.4637,
      "step": 20396
    },
    {
      "epoch": 0.8234557932983447,
      "grad_norm": 87.06877136230469,
      "learning_rate": 1.5907438081384386e-06,
      "loss": 2.8275,
      "step": 20397
    },
    {
      "epoch": 0.8234961647153816,
      "grad_norm": 121.35433197021484,
      "learning_rate": 1.5900362819568404e-06,
      "loss": 4.5714,
      "step": 20398
    },
    {
      "epoch": 0.8235365361324183,
      "grad_norm": 113.64525604248047,
      "learning_rate": 1.589328899565985e-06,
      "loss": 4.1067,
      "step": 20399
    },
    {
      "epoch": 0.823576907549455,
      "grad_norm": 63.64020538330078,
      "learning_rate": 1.5886216609779636e-06,
      "loss": 4.0574,
      "step": 20400
    },
    {
      "epoch": 0.8236172789664917,
      "grad_norm": 70.86627197265625,
      "learning_rate": 1.5879145662048701e-06,
      "loss": 5.8087,
      "step": 20401
    },
    {
      "epoch": 0.8236576503835285,
      "grad_norm": 103.83971405029297,
      "learning_rate": 1.5872076152587923e-06,
      "loss": 3.2716,
      "step": 20402
    },
    {
      "epoch": 0.8236980218005652,
      "grad_norm": 144.54322814941406,
      "learning_rate": 1.586500808151823e-06,
      "loss": 5.5696,
      "step": 20403
    },
    {
      "epoch": 0.8237383932176019,
      "grad_norm": 146.44296264648438,
      "learning_rate": 1.58579414489604e-06,
      "loss": 3.8932,
      "step": 20404
    },
    {
      "epoch": 0.8237787646346387,
      "grad_norm": 364.4522399902344,
      "learning_rate": 1.5850876255035287e-06,
      "loss": 4.6799,
      "step": 20405
    },
    {
      "epoch": 0.8238191360516754,
      "grad_norm": 143.67198181152344,
      "learning_rate": 1.5843812499863687e-06,
      "loss": 5.4023,
      "step": 20406
    },
    {
      "epoch": 0.8238595074687122,
      "grad_norm": 94.50958251953125,
      "learning_rate": 1.5836750183566396e-06,
      "loss": 5.1994,
      "step": 20407
    },
    {
      "epoch": 0.8238998788857489,
      "grad_norm": 91.4705810546875,
      "learning_rate": 1.5829689306264119e-06,
      "loss": 4.0283,
      "step": 20408
    },
    {
      "epoch": 0.8239402503027856,
      "grad_norm": 165.75067138671875,
      "learning_rate": 1.58226298680776e-06,
      "loss": 6.0233,
      "step": 20409
    },
    {
      "epoch": 0.8239806217198223,
      "grad_norm": 96.73428344726562,
      "learning_rate": 1.5815571869127544e-06,
      "loss": 3.5115,
      "step": 20410
    },
    {
      "epoch": 0.8240209931368591,
      "grad_norm": 150.685546875,
      "learning_rate": 1.5808515309534655e-06,
      "loss": 5.0566,
      "step": 20411
    },
    {
      "epoch": 0.8240613645538959,
      "grad_norm": 119.76342010498047,
      "learning_rate": 1.5801460189419514e-06,
      "loss": 4.5147,
      "step": 20412
    },
    {
      "epoch": 0.8241017359709326,
      "grad_norm": 77.88240051269531,
      "learning_rate": 1.5794406508902805e-06,
      "loss": 4.6941,
      "step": 20413
    },
    {
      "epoch": 0.8241421073879693,
      "grad_norm": 132.3431396484375,
      "learning_rate": 1.5787354268105093e-06,
      "loss": 5.5949,
      "step": 20414
    },
    {
      "epoch": 0.8241824788050061,
      "grad_norm": 136.30136108398438,
      "learning_rate": 1.5780303467146984e-06,
      "loss": 5.7352,
      "step": 20415
    },
    {
      "epoch": 0.8242228502220428,
      "grad_norm": 67.0621109008789,
      "learning_rate": 1.5773254106149016e-06,
      "loss": 5.2166,
      "step": 20416
    },
    {
      "epoch": 0.8242632216390795,
      "grad_norm": 95.78509521484375,
      "learning_rate": 1.5766206185231713e-06,
      "loss": 4.693,
      "step": 20417
    },
    {
      "epoch": 0.8243035930561162,
      "grad_norm": 80.42741394042969,
      "learning_rate": 1.5759159704515591e-06,
      "loss": 5.2363,
      "step": 20418
    },
    {
      "epoch": 0.8243439644731531,
      "grad_norm": 98.45306396484375,
      "learning_rate": 1.5752114664121144e-06,
      "loss": 3.6624,
      "step": 20419
    },
    {
      "epoch": 0.8243843358901898,
      "grad_norm": 82.82672119140625,
      "learning_rate": 1.5745071064168781e-06,
      "loss": 5.4767,
      "step": 20420
    },
    {
      "epoch": 0.8244247073072265,
      "grad_norm": 112.20731353759766,
      "learning_rate": 1.5738028904778947e-06,
      "loss": 4.7745,
      "step": 20421
    },
    {
      "epoch": 0.8244650787242632,
      "grad_norm": 119.5711669921875,
      "learning_rate": 1.5730988186072061e-06,
      "loss": 5.7487,
      "step": 20422
    },
    {
      "epoch": 0.8245054501413,
      "grad_norm": 87.35833740234375,
      "learning_rate": 1.5723948908168507e-06,
      "loss": 4.8962,
      "step": 20423
    },
    {
      "epoch": 0.8245458215583367,
      "grad_norm": 108.74443817138672,
      "learning_rate": 1.5716911071188611e-06,
      "loss": 3.8113,
      "step": 20424
    },
    {
      "epoch": 0.8245861929753734,
      "grad_norm": 102.67092895507812,
      "learning_rate": 1.5709874675252723e-06,
      "loss": 5.1944,
      "step": 20425
    },
    {
      "epoch": 0.8246265643924102,
      "grad_norm": 86.87537384033203,
      "learning_rate": 1.5702839720481143e-06,
      "loss": 4.5259,
      "step": 20426
    },
    {
      "epoch": 0.8246669358094469,
      "grad_norm": 120.36539459228516,
      "learning_rate": 1.569580620699419e-06,
      "loss": 5.3373,
      "step": 20427
    },
    {
      "epoch": 0.8247073072264837,
      "grad_norm": 116.54007720947266,
      "learning_rate": 1.5688774134912055e-06,
      "loss": 4.8119,
      "step": 20428
    },
    {
      "epoch": 0.8247476786435204,
      "grad_norm": 101.12763977050781,
      "learning_rate": 1.5681743504354997e-06,
      "loss": 4.6881,
      "step": 20429
    },
    {
      "epoch": 0.8247880500605571,
      "grad_norm": 147.5100555419922,
      "learning_rate": 1.567471431544323e-06,
      "loss": 5.4349,
      "step": 20430
    },
    {
      "epoch": 0.8248284214775938,
      "grad_norm": 117.88339233398438,
      "learning_rate": 1.5667686568296959e-06,
      "loss": 5.2571,
      "step": 20431
    },
    {
      "epoch": 0.8248687928946306,
      "grad_norm": 96.78143310546875,
      "learning_rate": 1.56606602630363e-06,
      "loss": 4.6795,
      "step": 20432
    },
    {
      "epoch": 0.8249091643116674,
      "grad_norm": 99.77559661865234,
      "learning_rate": 1.5653635399781397e-06,
      "loss": 6.3825,
      "step": 20433
    },
    {
      "epoch": 0.8249495357287041,
      "grad_norm": 148.51280212402344,
      "learning_rate": 1.5646611978652359e-06,
      "loss": 4.3617,
      "step": 20434
    },
    {
      "epoch": 0.8249899071457408,
      "grad_norm": 75.60789489746094,
      "learning_rate": 1.5639589999769288e-06,
      "loss": 4.2956,
      "step": 20435
    },
    {
      "epoch": 0.8250302785627776,
      "grad_norm": 62.10014724731445,
      "learning_rate": 1.5632569463252234e-06,
      "loss": 4.5261,
      "step": 20436
    },
    {
      "epoch": 0.8250706499798143,
      "grad_norm": 96.50638580322266,
      "learning_rate": 1.562555036922122e-06,
      "loss": 3.5042,
      "step": 20437
    },
    {
      "epoch": 0.825111021396851,
      "grad_norm": 112.30040740966797,
      "learning_rate": 1.5618532717796276e-06,
      "loss": 6.5338,
      "step": 20438
    },
    {
      "epoch": 0.8251513928138877,
      "grad_norm": 91.43868255615234,
      "learning_rate": 1.5611516509097391e-06,
      "loss": 4.9914,
      "step": 20439
    },
    {
      "epoch": 0.8251917642309246,
      "grad_norm": 86.10191345214844,
      "learning_rate": 1.5604501743244494e-06,
      "loss": 4.4259,
      "step": 20440
    },
    {
      "epoch": 0.8252321356479613,
      "grad_norm": 84.19517517089844,
      "learning_rate": 1.5597488420357543e-06,
      "loss": 5.1697,
      "step": 20441
    },
    {
      "epoch": 0.825272507064998,
      "grad_norm": 85.00335693359375,
      "learning_rate": 1.5590476540556442e-06,
      "loss": 4.0299,
      "step": 20442
    },
    {
      "epoch": 0.8253128784820347,
      "grad_norm": 108.4662857055664,
      "learning_rate": 1.558346610396111e-06,
      "loss": 5.0123,
      "step": 20443
    },
    {
      "epoch": 0.8253532498990714,
      "grad_norm": 111.4054183959961,
      "learning_rate": 1.5576457110691344e-06,
      "loss": 5.6714,
      "step": 20444
    },
    {
      "epoch": 0.8253936213161082,
      "grad_norm": 110.77620697021484,
      "learning_rate": 1.5569449560867035e-06,
      "loss": 5.4339,
      "step": 20445
    },
    {
      "epoch": 0.8254339927331449,
      "grad_norm": 105.17280578613281,
      "learning_rate": 1.5562443454607968e-06,
      "loss": 2.2606,
      "step": 20446
    },
    {
      "epoch": 0.8254743641501817,
      "grad_norm": 115.3470458984375,
      "learning_rate": 1.555543879203395e-06,
      "loss": 4.9889,
      "step": 20447
    },
    {
      "epoch": 0.8255147355672184,
      "grad_norm": 102.70378112792969,
      "learning_rate": 1.5548435573264754e-06,
      "loss": 5.1095,
      "step": 20448
    },
    {
      "epoch": 0.8255551069842552,
      "grad_norm": 76.58208465576172,
      "learning_rate": 1.5541433798420081e-06,
      "loss": 4.2949,
      "step": 20449
    },
    {
      "epoch": 0.8255954784012919,
      "grad_norm": 100.63365936279297,
      "learning_rate": 1.5534433467619658e-06,
      "loss": 4.2268,
      "step": 20450
    },
    {
      "epoch": 0.8256358498183286,
      "grad_norm": 99.83901977539062,
      "learning_rate": 1.5527434580983192e-06,
      "loss": 4.7036,
      "step": 20451
    },
    {
      "epoch": 0.8256762212353653,
      "grad_norm": 114.59521484375,
      "learning_rate": 1.552043713863035e-06,
      "loss": 4.8269,
      "step": 20452
    },
    {
      "epoch": 0.825716592652402,
      "grad_norm": 130.79945373535156,
      "learning_rate": 1.5513441140680741e-06,
      "loss": 3.7149,
      "step": 20453
    },
    {
      "epoch": 0.8257569640694389,
      "grad_norm": 167.59031677246094,
      "learning_rate": 1.5506446587253999e-06,
      "loss": 3.4844,
      "step": 20454
    },
    {
      "epoch": 0.8257973354864756,
      "grad_norm": 95.02925872802734,
      "learning_rate": 1.5499453478469718e-06,
      "loss": 4.8027,
      "step": 20455
    },
    {
      "epoch": 0.8258377069035123,
      "grad_norm": 88.27237701416016,
      "learning_rate": 1.5492461814447447e-06,
      "loss": 5.4209,
      "step": 20456
    },
    {
      "epoch": 0.825878078320549,
      "grad_norm": 148.24957275390625,
      "learning_rate": 1.5485471595306756e-06,
      "loss": 4.0523,
      "step": 20457
    },
    {
      "epoch": 0.8259184497375858,
      "grad_norm": 116.81082153320312,
      "learning_rate": 1.547848282116713e-06,
      "loss": 4.2259,
      "step": 20458
    },
    {
      "epoch": 0.8259588211546225,
      "grad_norm": 109.3330078125,
      "learning_rate": 1.547149549214808e-06,
      "loss": 3.2586,
      "step": 20459
    },
    {
      "epoch": 0.8259991925716592,
      "grad_norm": 124.65068054199219,
      "learning_rate": 1.5464509608369093e-06,
      "loss": 5.2328,
      "step": 20460
    },
    {
      "epoch": 0.826039563988696,
      "grad_norm": 160.70542907714844,
      "learning_rate": 1.5457525169949573e-06,
      "loss": 4.0117,
      "step": 20461
    },
    {
      "epoch": 0.8260799354057328,
      "grad_norm": 74.69095611572266,
      "learning_rate": 1.5450542177008942e-06,
      "loss": 3.0624,
      "step": 20462
    },
    {
      "epoch": 0.8261203068227695,
      "grad_norm": 150.65879821777344,
      "learning_rate": 1.5443560629666599e-06,
      "loss": 4.328,
      "step": 20463
    },
    {
      "epoch": 0.8261606782398062,
      "grad_norm": 98.79105377197266,
      "learning_rate": 1.5436580528041945e-06,
      "loss": 5.0163,
      "step": 20464
    },
    {
      "epoch": 0.8262010496568429,
      "grad_norm": 89.48591613769531,
      "learning_rate": 1.5429601872254274e-06,
      "loss": 4.9156,
      "step": 20465
    },
    {
      "epoch": 0.8262414210738797,
      "grad_norm": 96.8685073852539,
      "learning_rate": 1.5422624662422925e-06,
      "loss": 4.9733,
      "step": 20466
    },
    {
      "epoch": 0.8262817924909164,
      "grad_norm": 112.7743148803711,
      "learning_rate": 1.541564889866718e-06,
      "loss": 5.3305,
      "step": 20467
    },
    {
      "epoch": 0.8263221639079532,
      "grad_norm": 75.2378158569336,
      "learning_rate": 1.5408674581106353e-06,
      "loss": 4.2013,
      "step": 20468
    },
    {
      "epoch": 0.8263625353249899,
      "grad_norm": 133.42820739746094,
      "learning_rate": 1.540170170985963e-06,
      "loss": 4.8616,
      "step": 20469
    },
    {
      "epoch": 0.8264029067420267,
      "grad_norm": 101.62496185302734,
      "learning_rate": 1.5394730285046256e-06,
      "loss": 6.0467,
      "step": 20470
    },
    {
      "epoch": 0.8264432781590634,
      "grad_norm": 100.94296264648438,
      "learning_rate": 1.5387760306785438e-06,
      "loss": 6.4942,
      "step": 20471
    },
    {
      "epoch": 0.8264836495761001,
      "grad_norm": 104.91911315917969,
      "learning_rate": 1.5380791775196347e-06,
      "loss": 4.4225,
      "step": 20472
    },
    {
      "epoch": 0.8265240209931368,
      "grad_norm": 85.91936492919922,
      "learning_rate": 1.5373824690398088e-06,
      "loss": 2.965,
      "step": 20473
    },
    {
      "epoch": 0.8265643924101735,
      "grad_norm": 96.48275756835938,
      "learning_rate": 1.5366859052509818e-06,
      "loss": 3.9315,
      "step": 20474
    },
    {
      "epoch": 0.8266047638272104,
      "grad_norm": 243.3362579345703,
      "learning_rate": 1.5359894861650625e-06,
      "loss": 4.0394,
      "step": 20475
    },
    {
      "epoch": 0.8266451352442471,
      "grad_norm": 57.665191650390625,
      "learning_rate": 1.5352932117939578e-06,
      "loss": 5.3699,
      "step": 20476
    },
    {
      "epoch": 0.8266855066612838,
      "grad_norm": 90.73592376708984,
      "learning_rate": 1.5345970821495736e-06,
      "loss": 3.9164,
      "step": 20477
    },
    {
      "epoch": 0.8267258780783205,
      "grad_norm": 71.06803131103516,
      "learning_rate": 1.5339010972438107e-06,
      "loss": 5.265,
      "step": 20478
    },
    {
      "epoch": 0.8267662494953573,
      "grad_norm": 73.31663513183594,
      "learning_rate": 1.533205257088568e-06,
      "loss": 4.2207,
      "step": 20479
    },
    {
      "epoch": 0.826806620912394,
      "grad_norm": 96.49635314941406,
      "learning_rate": 1.5325095616957476e-06,
      "loss": 5.1581,
      "step": 20480
    },
    {
      "epoch": 0.8268469923294307,
      "grad_norm": 101.05738830566406,
      "learning_rate": 1.5318140110772384e-06,
      "loss": 4.1583,
      "step": 20481
    },
    {
      "epoch": 0.8268873637464675,
      "grad_norm": 102.0488510131836,
      "learning_rate": 1.5311186052449345e-06,
      "loss": 4.3368,
      "step": 20482
    },
    {
      "epoch": 0.8269277351635043,
      "grad_norm": 118.95763397216797,
      "learning_rate": 1.5304233442107263e-06,
      "loss": 5.0538,
      "step": 20483
    },
    {
      "epoch": 0.826968106580541,
      "grad_norm": 79.98201751708984,
      "learning_rate": 1.5297282279865045e-06,
      "loss": 5.5156,
      "step": 20484
    },
    {
      "epoch": 0.8270084779975777,
      "grad_norm": 109.07801818847656,
      "learning_rate": 1.529033256584147e-06,
      "loss": 4.8305,
      "step": 20485
    },
    {
      "epoch": 0.8270488494146144,
      "grad_norm": 196.84449768066406,
      "learning_rate": 1.5283384300155391e-06,
      "loss": 5.6686,
      "step": 20486
    },
    {
      "epoch": 0.8270892208316512,
      "grad_norm": 87.03519439697266,
      "learning_rate": 1.527643748292562e-06,
      "loss": 5.308,
      "step": 20487
    },
    {
      "epoch": 0.827129592248688,
      "grad_norm": 116.42005920410156,
      "learning_rate": 1.5269492114270957e-06,
      "loss": 5.2698,
      "step": 20488
    },
    {
      "epoch": 0.8271699636657247,
      "grad_norm": 138.30844116210938,
      "learning_rate": 1.5262548194310079e-06,
      "loss": 4.602,
      "step": 20489
    },
    {
      "epoch": 0.8272103350827614,
      "grad_norm": 106.05509948730469,
      "learning_rate": 1.525560572316176e-06,
      "loss": 4.8349,
      "step": 20490
    },
    {
      "epoch": 0.8272507064997981,
      "grad_norm": 102.67504119873047,
      "learning_rate": 1.5248664700944693e-06,
      "loss": 5.3605,
      "step": 20491
    },
    {
      "epoch": 0.8272910779168349,
      "grad_norm": 118.40853118896484,
      "learning_rate": 1.5241725127777573e-06,
      "loss": 4.6345,
      "step": 20492
    },
    {
      "epoch": 0.8273314493338716,
      "grad_norm": 100.48501586914062,
      "learning_rate": 1.5234787003778994e-06,
      "loss": 5.8158,
      "step": 20493
    },
    {
      "epoch": 0.8273718207509083,
      "grad_norm": 99.25135040283203,
      "learning_rate": 1.522785032906763e-06,
      "loss": 3.9491,
      "step": 20494
    },
    {
      "epoch": 0.8274121921679451,
      "grad_norm": 115.91553497314453,
      "learning_rate": 1.5220915103762058e-06,
      "loss": 5.4501,
      "step": 20495
    },
    {
      "epoch": 0.8274525635849819,
      "grad_norm": 122.45628356933594,
      "learning_rate": 1.5213981327980875e-06,
      "loss": 4.7096,
      "step": 20496
    },
    {
      "epoch": 0.8274929350020186,
      "grad_norm": 102.42324829101562,
      "learning_rate": 1.520704900184261e-06,
      "loss": 4.4967,
      "step": 20497
    },
    {
      "epoch": 0.8275333064190553,
      "grad_norm": 153.10987854003906,
      "learning_rate": 1.5200118125465812e-06,
      "loss": 3.684,
      "step": 20498
    },
    {
      "epoch": 0.827573677836092,
      "grad_norm": 170.39486694335938,
      "learning_rate": 1.5193188698968965e-06,
      "loss": 6.4419,
      "step": 20499
    },
    {
      "epoch": 0.8276140492531288,
      "grad_norm": 82.77574157714844,
      "learning_rate": 1.518626072247058e-06,
      "loss": 4.2526,
      "step": 20500
    },
    {
      "epoch": 0.8276544206701655,
      "grad_norm": 84.76980590820312,
      "learning_rate": 1.5179334196089056e-06,
      "loss": 4.5984,
      "step": 20501
    },
    {
      "epoch": 0.8276947920872023,
      "grad_norm": 94.20552062988281,
      "learning_rate": 1.5172409119942855e-06,
      "loss": 5.5142,
      "step": 20502
    },
    {
      "epoch": 0.827735163504239,
      "grad_norm": 91.21726989746094,
      "learning_rate": 1.5165485494150356e-06,
      "loss": 5.4274,
      "step": 20503
    },
    {
      "epoch": 0.8277755349212758,
      "grad_norm": 82.65022277832031,
      "learning_rate": 1.5158563318829989e-06,
      "loss": 5.6856,
      "step": 20504
    },
    {
      "epoch": 0.8278159063383125,
      "grad_norm": 74.33430480957031,
      "learning_rate": 1.5151642594100046e-06,
      "loss": 4.8143,
      "step": 20505
    },
    {
      "epoch": 0.8278562777553492,
      "grad_norm": 271.8466491699219,
      "learning_rate": 1.5144723320078868e-06,
      "loss": 5.9854,
      "step": 20506
    },
    {
      "epoch": 0.8278966491723859,
      "grad_norm": 88.28822326660156,
      "learning_rate": 1.5137805496884784e-06,
      "loss": 5.0456,
      "step": 20507
    },
    {
      "epoch": 0.8279370205894226,
      "grad_norm": 85.9671401977539,
      "learning_rate": 1.5130889124636072e-06,
      "loss": 5.4165,
      "step": 20508
    },
    {
      "epoch": 0.8279773920064595,
      "grad_norm": 67.29844665527344,
      "learning_rate": 1.5123974203450953e-06,
      "loss": 4.2411,
      "step": 20509
    },
    {
      "epoch": 0.8280177634234962,
      "grad_norm": 101.87200927734375,
      "learning_rate": 1.5117060733447676e-06,
      "loss": 3.2528,
      "step": 20510
    },
    {
      "epoch": 0.8280581348405329,
      "grad_norm": 97.42137908935547,
      "learning_rate": 1.5110148714744444e-06,
      "loss": 3.8425,
      "step": 20511
    },
    {
      "epoch": 0.8280985062575696,
      "grad_norm": 129.04771423339844,
      "learning_rate": 1.5103238147459465e-06,
      "loss": 6.3708,
      "step": 20512
    },
    {
      "epoch": 0.8281388776746064,
      "grad_norm": 194.500244140625,
      "learning_rate": 1.5096329031710832e-06,
      "loss": 4.6722,
      "step": 20513
    },
    {
      "epoch": 0.8281792490916431,
      "grad_norm": 88.03912353515625,
      "learning_rate": 1.508942136761672e-06,
      "loss": 4.9353,
      "step": 20514
    },
    {
      "epoch": 0.8282196205086798,
      "grad_norm": 180.232421875,
      "learning_rate": 1.5082515155295208e-06,
      "loss": 5.306,
      "step": 20515
    },
    {
      "epoch": 0.8282599919257166,
      "grad_norm": 136.26083374023438,
      "learning_rate": 1.5075610394864404e-06,
      "loss": 5.6952,
      "step": 20516
    },
    {
      "epoch": 0.8283003633427534,
      "grad_norm": 92.50850677490234,
      "learning_rate": 1.5068707086442335e-06,
      "loss": 2.6698,
      "step": 20517
    },
    {
      "epoch": 0.8283407347597901,
      "grad_norm": 69.17990112304688,
      "learning_rate": 1.5061805230147054e-06,
      "loss": 3.8439,
      "step": 20518
    },
    {
      "epoch": 0.8283811061768268,
      "grad_norm": 72.62261199951172,
      "learning_rate": 1.5054904826096562e-06,
      "loss": 5.598,
      "step": 20519
    },
    {
      "epoch": 0.8284214775938635,
      "grad_norm": 112.789794921875,
      "learning_rate": 1.5048005874408856e-06,
      "loss": 4.5394,
      "step": 20520
    },
    {
      "epoch": 0.8284618490109003,
      "grad_norm": 176.18064880371094,
      "learning_rate": 1.5041108375201851e-06,
      "loss": 3.913,
      "step": 20521
    },
    {
      "epoch": 0.828502220427937,
      "grad_norm": 113.76455688476562,
      "learning_rate": 1.50342123285935e-06,
      "loss": 5.5343,
      "step": 20522
    },
    {
      "epoch": 0.8285425918449738,
      "grad_norm": 89.15158081054688,
      "learning_rate": 1.5027317734701708e-06,
      "loss": 4.9933,
      "step": 20523
    },
    {
      "epoch": 0.8285829632620105,
      "grad_norm": 74.67382049560547,
      "learning_rate": 1.5020424593644378e-06,
      "loss": 3.2328,
      "step": 20524
    },
    {
      "epoch": 0.8286233346790473,
      "grad_norm": 89.86526489257812,
      "learning_rate": 1.501353290553932e-06,
      "loss": 4.6874,
      "step": 20525
    },
    {
      "epoch": 0.828663706096084,
      "grad_norm": 73.97279357910156,
      "learning_rate": 1.5006642670504401e-06,
      "loss": 6.0661,
      "step": 20526
    },
    {
      "epoch": 0.8287040775131207,
      "grad_norm": 118.28316497802734,
      "learning_rate": 1.4999753888657408e-06,
      "loss": 5.0507,
      "step": 20527
    },
    {
      "epoch": 0.8287444489301574,
      "grad_norm": 93.65156555175781,
      "learning_rate": 1.4992866560116159e-06,
      "loss": 3.1486,
      "step": 20528
    },
    {
      "epoch": 0.8287848203471941,
      "grad_norm": 236.7372283935547,
      "learning_rate": 1.4985980684998368e-06,
      "loss": 5.2611,
      "step": 20529
    },
    {
      "epoch": 0.828825191764231,
      "grad_norm": 119.46990966796875,
      "learning_rate": 1.497909626342179e-06,
      "loss": 4.5308,
      "step": 20530
    },
    {
      "epoch": 0.8288655631812677,
      "grad_norm": 96.67349243164062,
      "learning_rate": 1.4972213295504111e-06,
      "loss": 4.8861,
      "step": 20531
    },
    {
      "epoch": 0.8289059345983044,
      "grad_norm": 70.32978820800781,
      "learning_rate": 1.4965331781363046e-06,
      "loss": 4.9565,
      "step": 20532
    },
    {
      "epoch": 0.8289463060153411,
      "grad_norm": 98.68276977539062,
      "learning_rate": 1.495845172111623e-06,
      "loss": 5.6715,
      "step": 20533
    },
    {
      "epoch": 0.8289866774323779,
      "grad_norm": 68.36381530761719,
      "learning_rate": 1.4951573114881302e-06,
      "loss": 5.2725,
      "step": 20534
    },
    {
      "epoch": 0.8290270488494146,
      "grad_norm": 87.30720520019531,
      "learning_rate": 1.4944695962775878e-06,
      "loss": 7.0687,
      "step": 20535
    },
    {
      "epoch": 0.8290674202664513,
      "grad_norm": 92.4002456665039,
      "learning_rate": 1.4937820264917558e-06,
      "loss": 3.2989,
      "step": 20536
    },
    {
      "epoch": 0.8291077916834881,
      "grad_norm": 117.74617004394531,
      "learning_rate": 1.4930946021423842e-06,
      "loss": 4.772,
      "step": 20537
    },
    {
      "epoch": 0.8291481631005249,
      "grad_norm": 82.75022888183594,
      "learning_rate": 1.4924073232412306e-06,
      "loss": 5.8472,
      "step": 20538
    },
    {
      "epoch": 0.8291885345175616,
      "grad_norm": 142.4265899658203,
      "learning_rate": 1.4917201898000456e-06,
      "loss": 4.3243,
      "step": 20539
    },
    {
      "epoch": 0.8292289059345983,
      "grad_norm": 180.8815155029297,
      "learning_rate": 1.4910332018305796e-06,
      "loss": 3.9275,
      "step": 20540
    },
    {
      "epoch": 0.829269277351635,
      "grad_norm": 142.89898681640625,
      "learning_rate": 1.4903463593445722e-06,
      "loss": 6.0853,
      "step": 20541
    },
    {
      "epoch": 0.8293096487686717,
      "grad_norm": 77.27404022216797,
      "learning_rate": 1.4896596623537719e-06,
      "loss": 3.9118,
      "step": 20542
    },
    {
      "epoch": 0.8293500201857085,
      "grad_norm": 107.78871154785156,
      "learning_rate": 1.488973110869918e-06,
      "loss": 5.7836,
      "step": 20543
    },
    {
      "epoch": 0.8293903916027453,
      "grad_norm": 145.3656768798828,
      "learning_rate": 1.488286704904751e-06,
      "loss": 3.8589,
      "step": 20544
    },
    {
      "epoch": 0.829430763019782,
      "grad_norm": 80.83798217773438,
      "learning_rate": 1.4876004444700032e-06,
      "loss": 5.2642,
      "step": 20545
    },
    {
      "epoch": 0.8294711344368187,
      "grad_norm": 81.751708984375,
      "learning_rate": 1.4869143295774102e-06,
      "loss": 4.1988,
      "step": 20546
    },
    {
      "epoch": 0.8295115058538555,
      "grad_norm": 94.84827423095703,
      "learning_rate": 1.4862283602387017e-06,
      "loss": 4.9416,
      "step": 20547
    },
    {
      "epoch": 0.8295518772708922,
      "grad_norm": 76.51567077636719,
      "learning_rate": 1.4855425364656106e-06,
      "loss": 4.3902,
      "step": 20548
    },
    {
      "epoch": 0.8295922486879289,
      "grad_norm": 79.94607543945312,
      "learning_rate": 1.4848568582698564e-06,
      "loss": 6.2137,
      "step": 20549
    },
    {
      "epoch": 0.8296326201049656,
      "grad_norm": 92.13233184814453,
      "learning_rate": 1.4841713256631662e-06,
      "loss": 5.0297,
      "step": 20550
    },
    {
      "epoch": 0.8296729915220025,
      "grad_norm": 84.90983581542969,
      "learning_rate": 1.4834859386572598e-06,
      "loss": 4.8729,
      "step": 20551
    },
    {
      "epoch": 0.8297133629390392,
      "grad_norm": 86.10075378417969,
      "learning_rate": 1.4828006972638564e-06,
      "loss": 4.1374,
      "step": 20552
    },
    {
      "epoch": 0.8297537343560759,
      "grad_norm": 77.77435302734375,
      "learning_rate": 1.4821156014946724e-06,
      "loss": 4.3597,
      "step": 20553
    },
    {
      "epoch": 0.8297941057731126,
      "grad_norm": 74.4740219116211,
      "learning_rate": 1.4814306513614207e-06,
      "loss": 6.0486,
      "step": 20554
    },
    {
      "epoch": 0.8298344771901494,
      "grad_norm": 233.9626007080078,
      "learning_rate": 1.480745846875813e-06,
      "loss": 5.9761,
      "step": 20555
    },
    {
      "epoch": 0.8298748486071861,
      "grad_norm": 94.21021270751953,
      "learning_rate": 1.4800611880495585e-06,
      "loss": 4.6904,
      "step": 20556
    },
    {
      "epoch": 0.8299152200242228,
      "grad_norm": 96.98526000976562,
      "learning_rate": 1.4793766748943616e-06,
      "loss": 3.0451,
      "step": 20557
    },
    {
      "epoch": 0.8299555914412596,
      "grad_norm": 74.64399719238281,
      "learning_rate": 1.478692307421925e-06,
      "loss": 5.5533,
      "step": 20558
    },
    {
      "epoch": 0.8299959628582964,
      "grad_norm": 100.60912322998047,
      "learning_rate": 1.4780080856439516e-06,
      "loss": 5.064,
      "step": 20559
    },
    {
      "epoch": 0.8300363342753331,
      "grad_norm": 76.84906768798828,
      "learning_rate": 1.477324009572143e-06,
      "loss": 3.5726,
      "step": 20560
    },
    {
      "epoch": 0.8300767056923698,
      "grad_norm": 177.56781005859375,
      "learning_rate": 1.476640079218189e-06,
      "loss": 3.9496,
      "step": 20561
    },
    {
      "epoch": 0.8301170771094065,
      "grad_norm": 98.02188110351562,
      "learning_rate": 1.4759562945937855e-06,
      "loss": 5.2751,
      "step": 20562
    },
    {
      "epoch": 0.8301574485264432,
      "grad_norm": 85.16604614257812,
      "learning_rate": 1.4752726557106256e-06,
      "loss": 5.1101,
      "step": 20563
    },
    {
      "epoch": 0.83019781994348,
      "grad_norm": 108.350341796875,
      "learning_rate": 1.4745891625803977e-06,
      "loss": 4.3276,
      "step": 20564
    },
    {
      "epoch": 0.8302381913605168,
      "grad_norm": 107.84001159667969,
      "learning_rate": 1.4739058152147845e-06,
      "loss": 6.5988,
      "step": 20565
    },
    {
      "epoch": 0.8302785627775535,
      "grad_norm": 134.1155242919922,
      "learning_rate": 1.4732226136254734e-06,
      "loss": 5.2841,
      "step": 20566
    },
    {
      "epoch": 0.8303189341945902,
      "grad_norm": 194.7286834716797,
      "learning_rate": 1.472539557824143e-06,
      "loss": 4.4428,
      "step": 20567
    },
    {
      "epoch": 0.830359305611627,
      "grad_norm": 75.82612609863281,
      "learning_rate": 1.471856647822476e-06,
      "loss": 3.8622,
      "step": 20568
    },
    {
      "epoch": 0.8303996770286637,
      "grad_norm": 189.45333862304688,
      "learning_rate": 1.4711738836321432e-06,
      "loss": 4.5894,
      "step": 20569
    },
    {
      "epoch": 0.8304400484457004,
      "grad_norm": 101.77274322509766,
      "learning_rate": 1.4704912652648195e-06,
      "loss": 4.8719,
      "step": 20570
    },
    {
      "epoch": 0.8304804198627371,
      "grad_norm": 132.514404296875,
      "learning_rate": 1.4698087927321791e-06,
      "loss": 3.753,
      "step": 20571
    },
    {
      "epoch": 0.830520791279774,
      "grad_norm": 95.5210189819336,
      "learning_rate": 1.4691264660458881e-06,
      "loss": 3.5325,
      "step": 20572
    },
    {
      "epoch": 0.8305611626968107,
      "grad_norm": 136.76388549804688,
      "learning_rate": 1.468444285217614e-06,
      "loss": 5.5508,
      "step": 20573
    },
    {
      "epoch": 0.8306015341138474,
      "grad_norm": 145.9730682373047,
      "learning_rate": 1.4677622502590195e-06,
      "loss": 4.8029,
      "step": 20574
    },
    {
      "epoch": 0.8306419055308841,
      "grad_norm": 178.2523956298828,
      "learning_rate": 1.4670803611817664e-06,
      "loss": 4.6378,
      "step": 20575
    },
    {
      "epoch": 0.8306822769479209,
      "grad_norm": 66.5331039428711,
      "learning_rate": 1.4663986179975154e-06,
      "loss": 4.8003,
      "step": 20576
    },
    {
      "epoch": 0.8307226483649576,
      "grad_norm": 109.32594299316406,
      "learning_rate": 1.4657170207179194e-06,
      "loss": 5.6422,
      "step": 20577
    },
    {
      "epoch": 0.8307630197819943,
      "grad_norm": 131.74502563476562,
      "learning_rate": 1.4650355693546325e-06,
      "loss": 4.8301,
      "step": 20578
    },
    {
      "epoch": 0.8308033911990311,
      "grad_norm": 115.686767578125,
      "learning_rate": 1.4643542639193064e-06,
      "loss": 4.9491,
      "step": 20579
    },
    {
      "epoch": 0.8308437626160678,
      "grad_norm": 97.07913208007812,
      "learning_rate": 1.463673104423593e-06,
      "loss": 4.7194,
      "step": 20580
    },
    {
      "epoch": 0.8308841340331046,
      "grad_norm": 88.40151977539062,
      "learning_rate": 1.462992090879134e-06,
      "loss": 5.7524,
      "step": 20581
    },
    {
      "epoch": 0.8309245054501413,
      "grad_norm": 141.32949829101562,
      "learning_rate": 1.4623112232975744e-06,
      "loss": 4.7501,
      "step": 20582
    },
    {
      "epoch": 0.830964876867178,
      "grad_norm": 107.13179016113281,
      "learning_rate": 1.4616305016905552e-06,
      "loss": 4.8848,
      "step": 20583
    },
    {
      "epoch": 0.8310052482842147,
      "grad_norm": 222.91636657714844,
      "learning_rate": 1.460949926069719e-06,
      "loss": 3.7989,
      "step": 20584
    },
    {
      "epoch": 0.8310456197012515,
      "grad_norm": 83.34454345703125,
      "learning_rate": 1.4602694964466967e-06,
      "loss": 5.4176,
      "step": 20585
    },
    {
      "epoch": 0.8310859911182883,
      "grad_norm": 76.3445816040039,
      "learning_rate": 1.4595892128331247e-06,
      "loss": 3.2548,
      "step": 20586
    },
    {
      "epoch": 0.831126362535325,
      "grad_norm": 99.77006530761719,
      "learning_rate": 1.4589090752406333e-06,
      "loss": 4.8438,
      "step": 20587
    },
    {
      "epoch": 0.8311667339523617,
      "grad_norm": 96.29983520507812,
      "learning_rate": 1.4582290836808544e-06,
      "loss": 6.582,
      "step": 20588
    },
    {
      "epoch": 0.8312071053693985,
      "grad_norm": 76.68054962158203,
      "learning_rate": 1.457549238165409e-06,
      "loss": 4.4847,
      "step": 20589
    },
    {
      "epoch": 0.8312474767864352,
      "grad_norm": 88.25238037109375,
      "learning_rate": 1.456869538705924e-06,
      "loss": 5.6416,
      "step": 20590
    },
    {
      "epoch": 0.8312878482034719,
      "grad_norm": 82.84918975830078,
      "learning_rate": 1.4561899853140194e-06,
      "loss": 5.8377,
      "step": 20591
    },
    {
      "epoch": 0.8313282196205087,
      "grad_norm": 89.51940155029297,
      "learning_rate": 1.4555105780013158e-06,
      "loss": 4.2391,
      "step": 20592
    },
    {
      "epoch": 0.8313685910375455,
      "grad_norm": 94.53829193115234,
      "learning_rate": 1.4548313167794282e-06,
      "loss": 5.6775,
      "step": 20593
    },
    {
      "epoch": 0.8314089624545822,
      "grad_norm": 71.88850402832031,
      "learning_rate": 1.4541522016599697e-06,
      "loss": 2.7803,
      "step": 20594
    },
    {
      "epoch": 0.8314493338716189,
      "grad_norm": 78.76835632324219,
      "learning_rate": 1.4534732326545543e-06,
      "loss": 5.1711,
      "step": 20595
    },
    {
      "epoch": 0.8314897052886556,
      "grad_norm": 77.82195281982422,
      "learning_rate": 1.4527944097747904e-06,
      "loss": 4.2057,
      "step": 20596
    },
    {
      "epoch": 0.8315300767056923,
      "grad_norm": 93.11428833007812,
      "learning_rate": 1.4521157330322798e-06,
      "loss": 5.1203,
      "step": 20597
    },
    {
      "epoch": 0.8315704481227291,
      "grad_norm": 86.22803497314453,
      "learning_rate": 1.4514372024386314e-06,
      "loss": 5.6552,
      "step": 20598
    },
    {
      "epoch": 0.8316108195397659,
      "grad_norm": 89.55953216552734,
      "learning_rate": 1.4507588180054433e-06,
      "loss": 4.176,
      "step": 20599
    },
    {
      "epoch": 0.8316511909568026,
      "grad_norm": 106.45148468017578,
      "learning_rate": 1.4500805797443173e-06,
      "loss": 5.1557,
      "step": 20600
    },
    {
      "epoch": 0.8316915623738393,
      "grad_norm": 226.00779724121094,
      "learning_rate": 1.4494024876668467e-06,
      "loss": 4.788,
      "step": 20601
    },
    {
      "epoch": 0.8317319337908761,
      "grad_norm": 73.86083221435547,
      "learning_rate": 1.4487245417846263e-06,
      "loss": 4.4719,
      "step": 20602
    },
    {
      "epoch": 0.8317723052079128,
      "grad_norm": 139.90826416015625,
      "learning_rate": 1.448046742109247e-06,
      "loss": 5.1255,
      "step": 20603
    },
    {
      "epoch": 0.8318126766249495,
      "grad_norm": 102.71208190917969,
      "learning_rate": 1.4473690886523007e-06,
      "loss": 4.693,
      "step": 20604
    },
    {
      "epoch": 0.8318530480419862,
      "grad_norm": 85.78606414794922,
      "learning_rate": 1.446691581425369e-06,
      "loss": 5.814,
      "step": 20605
    },
    {
      "epoch": 0.8318934194590231,
      "grad_norm": 143.61380004882812,
      "learning_rate": 1.4460142204400373e-06,
      "loss": 4.3313,
      "step": 20606
    },
    {
      "epoch": 0.8319337908760598,
      "grad_norm": 92.8137435913086,
      "learning_rate": 1.4453370057078886e-06,
      "loss": 4.2615,
      "step": 20607
    },
    {
      "epoch": 0.8319741622930965,
      "grad_norm": 68.52033996582031,
      "learning_rate": 1.4446599372405023e-06,
      "loss": 4.181,
      "step": 20608
    },
    {
      "epoch": 0.8320145337101332,
      "grad_norm": 87.73193359375,
      "learning_rate": 1.4439830150494504e-06,
      "loss": 3.4305,
      "step": 20609
    },
    {
      "epoch": 0.83205490512717,
      "grad_norm": 75.96551513671875,
      "learning_rate": 1.4433062391463094e-06,
      "loss": 4.856,
      "step": 20610
    },
    {
      "epoch": 0.8320952765442067,
      "grad_norm": 80.89532470703125,
      "learning_rate": 1.4426296095426494e-06,
      "loss": 4.3322,
      "step": 20611
    },
    {
      "epoch": 0.8321356479612434,
      "grad_norm": 116.43003845214844,
      "learning_rate": 1.4419531262500409e-06,
      "loss": 4.1135,
      "step": 20612
    },
    {
      "epoch": 0.8321760193782802,
      "grad_norm": 90.91912078857422,
      "learning_rate": 1.4412767892800495e-06,
      "loss": 4.8027,
      "step": 20613
    },
    {
      "epoch": 0.832216390795317,
      "grad_norm": 117.71955871582031,
      "learning_rate": 1.4406005986442396e-06,
      "loss": 4.9934,
      "step": 20614
    },
    {
      "epoch": 0.8322567622123537,
      "grad_norm": 76.08090209960938,
      "learning_rate": 1.4399245543541706e-06,
      "loss": 3.8351,
      "step": 20615
    },
    {
      "epoch": 0.8322971336293904,
      "grad_norm": 89.90103149414062,
      "learning_rate": 1.439248656421406e-06,
      "loss": 4.4486,
      "step": 20616
    },
    {
      "epoch": 0.8323375050464271,
      "grad_norm": 261.5254821777344,
      "learning_rate": 1.438572904857496e-06,
      "loss": 5.0221,
      "step": 20617
    },
    {
      "epoch": 0.8323778764634638,
      "grad_norm": 96.3944091796875,
      "learning_rate": 1.4378972996739971e-06,
      "loss": 5.1765,
      "step": 20618
    },
    {
      "epoch": 0.8324182478805006,
      "grad_norm": 114.66255950927734,
      "learning_rate": 1.437221840882459e-06,
      "loss": 5.7248,
      "step": 20619
    },
    {
      "epoch": 0.8324586192975374,
      "grad_norm": 98.47114562988281,
      "learning_rate": 1.4365465284944357e-06,
      "loss": 5.3253,
      "step": 20620
    },
    {
      "epoch": 0.8324989907145741,
      "grad_norm": 93.2956314086914,
      "learning_rate": 1.4358713625214671e-06,
      "loss": 4.6751,
      "step": 20621
    },
    {
      "epoch": 0.8325393621316108,
      "grad_norm": 284.8984375,
      "learning_rate": 1.4351963429750991e-06,
      "loss": 5.2686,
      "step": 20622
    },
    {
      "epoch": 0.8325797335486476,
      "grad_norm": 720.6304931640625,
      "learning_rate": 1.434521469866874e-06,
      "loss": 4.6666,
      "step": 20623
    },
    {
      "epoch": 0.8326201049656843,
      "grad_norm": 96.42735290527344,
      "learning_rate": 1.4338467432083313e-06,
      "loss": 5.3711,
      "step": 20624
    },
    {
      "epoch": 0.832660476382721,
      "grad_norm": 101.39768981933594,
      "learning_rate": 1.433172163011004e-06,
      "loss": 4.2488,
      "step": 20625
    },
    {
      "epoch": 0.8327008477997577,
      "grad_norm": 99.19609832763672,
      "learning_rate": 1.4324977292864284e-06,
      "loss": 4.1584,
      "step": 20626
    },
    {
      "epoch": 0.8327412192167946,
      "grad_norm": 91.88157653808594,
      "learning_rate": 1.4318234420461342e-06,
      "loss": 5.8704,
      "step": 20627
    },
    {
      "epoch": 0.8327815906338313,
      "grad_norm": 111.40673828125,
      "learning_rate": 1.4311493013016542e-06,
      "loss": 5.0782,
      "step": 20628
    },
    {
      "epoch": 0.832821962050868,
      "grad_norm": 83.22486877441406,
      "learning_rate": 1.4304753070645083e-06,
      "loss": 3.6737,
      "step": 20629
    },
    {
      "epoch": 0.8328623334679047,
      "grad_norm": 97.31183624267578,
      "learning_rate": 1.429801459346224e-06,
      "loss": 5.4504,
      "step": 20630
    },
    {
      "epoch": 0.8329027048849414,
      "grad_norm": 73.8436279296875,
      "learning_rate": 1.4291277581583207e-06,
      "loss": 5.6438,
      "step": 20631
    },
    {
      "epoch": 0.8329430763019782,
      "grad_norm": 76.86113739013672,
      "learning_rate": 1.4284542035123194e-06,
      "loss": 4.3155,
      "step": 20632
    },
    {
      "epoch": 0.8329834477190149,
      "grad_norm": 244.90179443359375,
      "learning_rate": 1.4277807954197353e-06,
      "loss": 4.5266,
      "step": 20633
    },
    {
      "epoch": 0.8330238191360517,
      "grad_norm": 112.00877380371094,
      "learning_rate": 1.4271075338920815e-06,
      "loss": 6.4118,
      "step": 20634
    },
    {
      "epoch": 0.8330641905530884,
      "grad_norm": 98.87430572509766,
      "learning_rate": 1.4264344189408697e-06,
      "loss": 4.3417,
      "step": 20635
    },
    {
      "epoch": 0.8331045619701252,
      "grad_norm": 80.62248229980469,
      "learning_rate": 1.4257614505776118e-06,
      "loss": 3.9714,
      "step": 20636
    },
    {
      "epoch": 0.8331449333871619,
      "grad_norm": 105.1454086303711,
      "learning_rate": 1.4250886288138076e-06,
      "loss": 4.3556,
      "step": 20637
    },
    {
      "epoch": 0.8331853048041986,
      "grad_norm": 98.49795532226562,
      "learning_rate": 1.4244159536609636e-06,
      "loss": 5.7533,
      "step": 20638
    },
    {
      "epoch": 0.8332256762212353,
      "grad_norm": 106.59420013427734,
      "learning_rate": 1.4237434251305825e-06,
      "loss": 5.2384,
      "step": 20639
    },
    {
      "epoch": 0.8332660476382721,
      "grad_norm": 86.53277587890625,
      "learning_rate": 1.4230710432341632e-06,
      "loss": 4.2455,
      "step": 20640
    },
    {
      "epoch": 0.8333064190553089,
      "grad_norm": 86.36876678466797,
      "learning_rate": 1.4223988079831996e-06,
      "loss": 4.6882,
      "step": 20641
    },
    {
      "epoch": 0.8333467904723456,
      "grad_norm": 177.7490692138672,
      "learning_rate": 1.4217267193891849e-06,
      "loss": 4.4456,
      "step": 20642
    },
    {
      "epoch": 0.8333871618893823,
      "grad_norm": 130.39866638183594,
      "learning_rate": 1.4210547774636118e-06,
      "loss": 4.8582,
      "step": 20643
    },
    {
      "epoch": 0.833427533306419,
      "grad_norm": 105.09256744384766,
      "learning_rate": 1.4203829822179715e-06,
      "loss": 5.0656,
      "step": 20644
    },
    {
      "epoch": 0.8334679047234558,
      "grad_norm": 95.16401672363281,
      "learning_rate": 1.4197113336637446e-06,
      "loss": 5.586,
      "step": 20645
    },
    {
      "epoch": 0.8335082761404925,
      "grad_norm": 89.94481658935547,
      "learning_rate": 1.4190398318124177e-06,
      "loss": 5.7564,
      "step": 20646
    },
    {
      "epoch": 0.8335486475575292,
      "grad_norm": 83.64257049560547,
      "learning_rate": 1.4183684766754713e-06,
      "loss": 3.5403,
      "step": 20647
    },
    {
      "epoch": 0.833589018974566,
      "grad_norm": 124.39457702636719,
      "learning_rate": 1.4176972682643874e-06,
      "loss": 7.3483,
      "step": 20648
    },
    {
      "epoch": 0.8336293903916028,
      "grad_norm": 84.25316619873047,
      "learning_rate": 1.417026206590636e-06,
      "loss": 4.6248,
      "step": 20649
    },
    {
      "epoch": 0.8336697618086395,
      "grad_norm": 89.8311767578125,
      "learning_rate": 1.4163552916656942e-06,
      "loss": 4.5012,
      "step": 20650
    },
    {
      "epoch": 0.8337101332256762,
      "grad_norm": 91.50507354736328,
      "learning_rate": 1.415684523501033e-06,
      "loss": 4.1634,
      "step": 20651
    },
    {
      "epoch": 0.8337505046427129,
      "grad_norm": 180.0217742919922,
      "learning_rate": 1.41501390210812e-06,
      "loss": 5.4548,
      "step": 20652
    },
    {
      "epoch": 0.8337908760597497,
      "grad_norm": 105.4185791015625,
      "learning_rate": 1.4143434274984224e-06,
      "loss": 5.3192,
      "step": 20653
    },
    {
      "epoch": 0.8338312474767864,
      "grad_norm": 114.206298828125,
      "learning_rate": 1.4136730996834025e-06,
      "loss": 4.8294,
      "step": 20654
    },
    {
      "epoch": 0.8338716188938232,
      "grad_norm": 96.46851348876953,
      "learning_rate": 1.413002918674523e-06,
      "loss": 5.1555,
      "step": 20655
    },
    {
      "epoch": 0.8339119903108599,
      "grad_norm": 89.58814239501953,
      "learning_rate": 1.412332884483244e-06,
      "loss": 4.7313,
      "step": 20656
    },
    {
      "epoch": 0.8339523617278967,
      "grad_norm": 99.34536743164062,
      "learning_rate": 1.4116629971210161e-06,
      "loss": 5.866,
      "step": 20657
    },
    {
      "epoch": 0.8339927331449334,
      "grad_norm": 112.7306137084961,
      "learning_rate": 1.4109932565992957e-06,
      "loss": 3.833,
      "step": 20658
    },
    {
      "epoch": 0.8340331045619701,
      "grad_norm": 75.61194610595703,
      "learning_rate": 1.4103236629295337e-06,
      "loss": 4.1801,
      "step": 20659
    },
    {
      "epoch": 0.8340734759790068,
      "grad_norm": 107.09629821777344,
      "learning_rate": 1.4096542161231808e-06,
      "loss": 3.9496,
      "step": 20660
    },
    {
      "epoch": 0.8341138473960436,
      "grad_norm": 94.8289566040039,
      "learning_rate": 1.408984916191679e-06,
      "loss": 4.4486,
      "step": 20661
    },
    {
      "epoch": 0.8341542188130804,
      "grad_norm": 115.868896484375,
      "learning_rate": 1.4083157631464738e-06,
      "loss": 4.7116,
      "step": 20662
    },
    {
      "epoch": 0.8341945902301171,
      "grad_norm": 91.51541900634766,
      "learning_rate": 1.407646756999006e-06,
      "loss": 5.5792,
      "step": 20663
    },
    {
      "epoch": 0.8342349616471538,
      "grad_norm": 90.55810546875,
      "learning_rate": 1.4069778977607163e-06,
      "loss": 4.8034,
      "step": 20664
    },
    {
      "epoch": 0.8342753330641905,
      "grad_norm": 105.51766204833984,
      "learning_rate": 1.4063091854430367e-06,
      "loss": 4.755,
      "step": 20665
    },
    {
      "epoch": 0.8343157044812273,
      "grad_norm": 91.92585754394531,
      "learning_rate": 1.4056406200574014e-06,
      "loss": 4.6506,
      "step": 20666
    },
    {
      "epoch": 0.834356075898264,
      "grad_norm": 103.90672302246094,
      "learning_rate": 1.4049722016152423e-06,
      "loss": 3.4993,
      "step": 20667
    },
    {
      "epoch": 0.8343964473153007,
      "grad_norm": 79.6224594116211,
      "learning_rate": 1.4043039301279904e-06,
      "loss": 5.0586,
      "step": 20668
    },
    {
      "epoch": 0.8344368187323375,
      "grad_norm": 127.19525909423828,
      "learning_rate": 1.4036358056070666e-06,
      "loss": 5.9639,
      "step": 20669
    },
    {
      "epoch": 0.8344771901493743,
      "grad_norm": 145.00613403320312,
      "learning_rate": 1.402967828063897e-06,
      "loss": 5.3429,
      "step": 20670
    },
    {
      "epoch": 0.834517561566411,
      "grad_norm": 100.26947784423828,
      "learning_rate": 1.4022999975099015e-06,
      "loss": 5.3482,
      "step": 20671
    },
    {
      "epoch": 0.8345579329834477,
      "grad_norm": 75.12128448486328,
      "learning_rate": 1.401632313956499e-06,
      "loss": 4.2204,
      "step": 20672
    },
    {
      "epoch": 0.8345983044004844,
      "grad_norm": 83.68258666992188,
      "learning_rate": 1.4009647774151046e-06,
      "loss": 4.5357,
      "step": 20673
    },
    {
      "epoch": 0.8346386758175212,
      "grad_norm": 107.9892807006836,
      "learning_rate": 1.4002973878971338e-06,
      "loss": 5.0569,
      "step": 20674
    },
    {
      "epoch": 0.8346790472345579,
      "grad_norm": 116.36490631103516,
      "learning_rate": 1.3996301454139949e-06,
      "loss": 5.795,
      "step": 20675
    },
    {
      "epoch": 0.8347194186515947,
      "grad_norm": 446.46539306640625,
      "learning_rate": 1.398963049977099e-06,
      "loss": 5.1757,
      "step": 20676
    },
    {
      "epoch": 0.8347597900686314,
      "grad_norm": 164.4442138671875,
      "learning_rate": 1.3982961015978492e-06,
      "loss": 4.5428,
      "step": 20677
    },
    {
      "epoch": 0.8348001614856682,
      "grad_norm": 85.67887115478516,
      "learning_rate": 1.3976293002876485e-06,
      "loss": 2.7311,
      "step": 20678
    },
    {
      "epoch": 0.8348405329027049,
      "grad_norm": 126.92366790771484,
      "learning_rate": 1.3969626460578989e-06,
      "loss": 5.2192,
      "step": 20679
    },
    {
      "epoch": 0.8348809043197416,
      "grad_norm": 192.89801025390625,
      "learning_rate": 1.3962961389200013e-06,
      "loss": 5.9527,
      "step": 20680
    },
    {
      "epoch": 0.8349212757367783,
      "grad_norm": 112.2469253540039,
      "learning_rate": 1.3956297788853468e-06,
      "loss": 5.0107,
      "step": 20681
    },
    {
      "epoch": 0.834961647153815,
      "grad_norm": 84.34880828857422,
      "learning_rate": 1.3949635659653294e-06,
      "loss": 4.4695,
      "step": 20682
    },
    {
      "epoch": 0.8350020185708519,
      "grad_norm": 114.09156799316406,
      "learning_rate": 1.394297500171341e-06,
      "loss": 5.2161,
      "step": 20683
    },
    {
      "epoch": 0.8350423899878886,
      "grad_norm": 76.55435180664062,
      "learning_rate": 1.3936315815147716e-06,
      "loss": 4.7929,
      "step": 20684
    },
    {
      "epoch": 0.8350827614049253,
      "grad_norm": 133.698486328125,
      "learning_rate": 1.3929658100070032e-06,
      "loss": 3.9082,
      "step": 20685
    },
    {
      "epoch": 0.835123132821962,
      "grad_norm": 93.78007507324219,
      "learning_rate": 1.392300185659421e-06,
      "loss": 4.5437,
      "step": 20686
    },
    {
      "epoch": 0.8351635042389988,
      "grad_norm": 90.14443969726562,
      "learning_rate": 1.3916347084834047e-06,
      "loss": 4.7871,
      "step": 20687
    },
    {
      "epoch": 0.8352038756560355,
      "grad_norm": 136.36387634277344,
      "learning_rate": 1.3909693784903344e-06,
      "loss": 5.191,
      "step": 20688
    },
    {
      "epoch": 0.8352442470730722,
      "grad_norm": 101.60565185546875,
      "learning_rate": 1.390304195691583e-06,
      "loss": 6.3578,
      "step": 20689
    },
    {
      "epoch": 0.835284618490109,
      "grad_norm": 116.93328094482422,
      "learning_rate": 1.389639160098525e-06,
      "loss": 5.308,
      "step": 20690
    },
    {
      "epoch": 0.8353249899071458,
      "grad_norm": 96.3694839477539,
      "learning_rate": 1.3889742717225296e-06,
      "loss": 4.2766,
      "step": 20691
    },
    {
      "epoch": 0.8353653613241825,
      "grad_norm": 80.46695709228516,
      "learning_rate": 1.3883095305749661e-06,
      "loss": 4.1417,
      "step": 20692
    },
    {
      "epoch": 0.8354057327412192,
      "grad_norm": 99.98011779785156,
      "learning_rate": 1.3876449366672007e-06,
      "loss": 4.5571,
      "step": 20693
    },
    {
      "epoch": 0.8354461041582559,
      "grad_norm": 78.65151977539062,
      "learning_rate": 1.3869804900105955e-06,
      "loss": 4.8955,
      "step": 20694
    },
    {
      "epoch": 0.8354864755752927,
      "grad_norm": 127.017333984375,
      "learning_rate": 1.3863161906165102e-06,
      "loss": 5.386,
      "step": 20695
    },
    {
      "epoch": 0.8355268469923295,
      "grad_norm": 96.05106353759766,
      "learning_rate": 1.3856520384963069e-06,
      "loss": 5.3915,
      "step": 20696
    },
    {
      "epoch": 0.8355672184093662,
      "grad_norm": 96.64859771728516,
      "learning_rate": 1.384988033661334e-06,
      "loss": 4.8118,
      "step": 20697
    },
    {
      "epoch": 0.8356075898264029,
      "grad_norm": 73.22400665283203,
      "learning_rate": 1.3843241761229499e-06,
      "loss": 5.0224,
      "step": 20698
    },
    {
      "epoch": 0.8356479612434397,
      "grad_norm": 103.83341217041016,
      "learning_rate": 1.3836604658925023e-06,
      "loss": 6.0448,
      "step": 20699
    },
    {
      "epoch": 0.8356883326604764,
      "grad_norm": 101.88729095458984,
      "learning_rate": 1.3829969029813429e-06,
      "loss": 4.7992,
      "step": 20700
    },
    {
      "epoch": 0.8357287040775131,
      "grad_norm": 195.65902709960938,
      "learning_rate": 1.382333487400812e-06,
      "loss": 3.129,
      "step": 20701
    },
    {
      "epoch": 0.8357690754945498,
      "grad_norm": 85.17601776123047,
      "learning_rate": 1.3816702191622544e-06,
      "loss": 4.5991,
      "step": 20702
    },
    {
      "epoch": 0.8358094469115867,
      "grad_norm": 83.54581451416016,
      "learning_rate": 1.3810070982770119e-06,
      "loss": 2.8822,
      "step": 20703
    },
    {
      "epoch": 0.8358498183286234,
      "grad_norm": 100.77900695800781,
      "learning_rate": 1.380344124756422e-06,
      "loss": 4.2228,
      "step": 20704
    },
    {
      "epoch": 0.8358901897456601,
      "grad_norm": 113.34273529052734,
      "learning_rate": 1.379681298611818e-06,
      "loss": 4.9086,
      "step": 20705
    },
    {
      "epoch": 0.8359305611626968,
      "grad_norm": 80.16291809082031,
      "learning_rate": 1.3790186198545342e-06,
      "loss": 4.279,
      "step": 20706
    },
    {
      "epoch": 0.8359709325797335,
      "grad_norm": 119.97500610351562,
      "learning_rate": 1.378356088495899e-06,
      "loss": 6.1891,
      "step": 20707
    },
    {
      "epoch": 0.8360113039967703,
      "grad_norm": 131.4002227783203,
      "learning_rate": 1.3776937045472449e-06,
      "loss": 4.8521,
      "step": 20708
    },
    {
      "epoch": 0.836051675413807,
      "grad_norm": 79.47119903564453,
      "learning_rate": 1.3770314680198927e-06,
      "loss": 6.568,
      "step": 20709
    },
    {
      "epoch": 0.8360920468308438,
      "grad_norm": 92.09786224365234,
      "learning_rate": 1.3763693789251643e-06,
      "loss": 3.6813,
      "step": 20710
    },
    {
      "epoch": 0.8361324182478805,
      "grad_norm": 90.2834701538086,
      "learning_rate": 1.3757074372743828e-06,
      "loss": 5.1957,
      "step": 20711
    },
    {
      "epoch": 0.8361727896649173,
      "grad_norm": 85.34314727783203,
      "learning_rate": 1.375045643078865e-06,
      "loss": 5.2467,
      "step": 20712
    },
    {
      "epoch": 0.836213161081954,
      "grad_norm": 137.0951690673828,
      "learning_rate": 1.3743839963499261e-06,
      "loss": 5.4393,
      "step": 20713
    },
    {
      "epoch": 0.8362535324989907,
      "grad_norm": 139.67269897460938,
      "learning_rate": 1.3737224970988793e-06,
      "loss": 6.5643,
      "step": 20714
    },
    {
      "epoch": 0.8362939039160274,
      "grad_norm": 79.38671875,
      "learning_rate": 1.3730611453370323e-06,
      "loss": 5.4836,
      "step": 20715
    },
    {
      "epoch": 0.8363342753330641,
      "grad_norm": 91.77081298828125,
      "learning_rate": 1.3723999410756984e-06,
      "loss": 3.4779,
      "step": 20716
    },
    {
      "epoch": 0.836374646750101,
      "grad_norm": 89.48699951171875,
      "learning_rate": 1.3717388843261748e-06,
      "loss": 4.9318,
      "step": 20717
    },
    {
      "epoch": 0.8364150181671377,
      "grad_norm": 72.13571166992188,
      "learning_rate": 1.3710779750997694e-06,
      "loss": 5.9815,
      "step": 20718
    },
    {
      "epoch": 0.8364553895841744,
      "grad_norm": 117.87150573730469,
      "learning_rate": 1.3704172134077798e-06,
      "loss": 5.3175,
      "step": 20719
    },
    {
      "epoch": 0.8364957610012111,
      "grad_norm": 165.23825073242188,
      "learning_rate": 1.3697565992615058e-06,
      "loss": 4.808,
      "step": 20720
    },
    {
      "epoch": 0.8365361324182479,
      "grad_norm": 88.07093811035156,
      "learning_rate": 1.3690961326722395e-06,
      "loss": 3.4627,
      "step": 20721
    },
    {
      "epoch": 0.8365765038352846,
      "grad_norm": 99.34557342529297,
      "learning_rate": 1.3684358136512744e-06,
      "loss": 4.503,
      "step": 20722
    },
    {
      "epoch": 0.8366168752523213,
      "grad_norm": 98.16602325439453,
      "learning_rate": 1.367775642209901e-06,
      "loss": 4.6834,
      "step": 20723
    },
    {
      "epoch": 0.8366572466693581,
      "grad_norm": 114.19049072265625,
      "learning_rate": 1.3671156183594091e-06,
      "loss": 4.7812,
      "step": 20724
    },
    {
      "epoch": 0.8366976180863949,
      "grad_norm": 119.3510971069336,
      "learning_rate": 1.366455742111078e-06,
      "loss": 5.8237,
      "step": 20725
    },
    {
      "epoch": 0.8367379895034316,
      "grad_norm": 81.47602844238281,
      "learning_rate": 1.3657960134761928e-06,
      "loss": 5.5871,
      "step": 20726
    },
    {
      "epoch": 0.8367783609204683,
      "grad_norm": 161.38308715820312,
      "learning_rate": 1.3651364324660344e-06,
      "loss": 5.3618,
      "step": 20727
    },
    {
      "epoch": 0.836818732337505,
      "grad_norm": 108.0195541381836,
      "learning_rate": 1.3644769990918782e-06,
      "loss": 4.5375,
      "step": 20728
    },
    {
      "epoch": 0.8368591037545418,
      "grad_norm": 97.99452209472656,
      "learning_rate": 1.363817713365e-06,
      "loss": 4.5534,
      "step": 20729
    },
    {
      "epoch": 0.8368994751715785,
      "grad_norm": 137.28451538085938,
      "learning_rate": 1.3631585752966747e-06,
      "loss": 6.1216,
      "step": 20730
    },
    {
      "epoch": 0.8369398465886153,
      "grad_norm": 77.9480209350586,
      "learning_rate": 1.3624995848981671e-06,
      "loss": 5.6238,
      "step": 20731
    },
    {
      "epoch": 0.836980218005652,
      "grad_norm": 72.64112854003906,
      "learning_rate": 1.3618407421807456e-06,
      "loss": 4.1327,
      "step": 20732
    },
    {
      "epoch": 0.8370205894226888,
      "grad_norm": 107.74735260009766,
      "learning_rate": 1.361182047155677e-06,
      "loss": 5.8271,
      "step": 20733
    },
    {
      "epoch": 0.8370609608397255,
      "grad_norm": 84.17476654052734,
      "learning_rate": 1.3605234998342209e-06,
      "loss": 3.3741,
      "step": 20734
    },
    {
      "epoch": 0.8371013322567622,
      "grad_norm": 73.93509674072266,
      "learning_rate": 1.3598651002276398e-06,
      "loss": 4.2153,
      "step": 20735
    },
    {
      "epoch": 0.8371417036737989,
      "grad_norm": 135.59144592285156,
      "learning_rate": 1.3592068483471909e-06,
      "loss": 4.7523,
      "step": 20736
    },
    {
      "epoch": 0.8371820750908356,
      "grad_norm": 94.15314483642578,
      "learning_rate": 1.3585487442041256e-06,
      "loss": 3.9373,
      "step": 20737
    },
    {
      "epoch": 0.8372224465078725,
      "grad_norm": 86.44725036621094,
      "learning_rate": 1.3578907878096969e-06,
      "loss": 5.0319,
      "step": 20738
    },
    {
      "epoch": 0.8372628179249092,
      "grad_norm": 97.92799377441406,
      "learning_rate": 1.3572329791751548e-06,
      "loss": 4.7855,
      "step": 20739
    },
    {
      "epoch": 0.8373031893419459,
      "grad_norm": 90.61442565917969,
      "learning_rate": 1.356575318311749e-06,
      "loss": 4.4195,
      "step": 20740
    },
    {
      "epoch": 0.8373435607589826,
      "grad_norm": 102.23116302490234,
      "learning_rate": 1.3559178052307188e-06,
      "loss": 4.3847,
      "step": 20741
    },
    {
      "epoch": 0.8373839321760194,
      "grad_norm": 174.4544219970703,
      "learning_rate": 1.355260439943309e-06,
      "loss": 5.8538,
      "step": 20742
    },
    {
      "epoch": 0.8374243035930561,
      "grad_norm": 153.0377960205078,
      "learning_rate": 1.3546032224607586e-06,
      "loss": 5.3958,
      "step": 20743
    },
    {
      "epoch": 0.8374646750100928,
      "grad_norm": 539.0427856445312,
      "learning_rate": 1.3539461527943077e-06,
      "loss": 4.0786,
      "step": 20744
    },
    {
      "epoch": 0.8375050464271296,
      "grad_norm": 94.3427963256836,
      "learning_rate": 1.3532892309551848e-06,
      "loss": 4.417,
      "step": 20745
    },
    {
      "epoch": 0.8375454178441664,
      "grad_norm": 100.56224060058594,
      "learning_rate": 1.3526324569546245e-06,
      "loss": 4.7346,
      "step": 20746
    },
    {
      "epoch": 0.8375857892612031,
      "grad_norm": 120.5916976928711,
      "learning_rate": 1.3519758308038556e-06,
      "loss": 6.4593,
      "step": 20747
    },
    {
      "epoch": 0.8376261606782398,
      "grad_norm": 82.96517181396484,
      "learning_rate": 1.3513193525141056e-06,
      "loss": 4.6363,
      "step": 20748
    },
    {
      "epoch": 0.8376665320952765,
      "grad_norm": 81.85678100585938,
      "learning_rate": 1.3506630220965988e-06,
      "loss": 4.761,
      "step": 20749
    },
    {
      "epoch": 0.8377069035123133,
      "grad_norm": 68.78125,
      "learning_rate": 1.3500068395625577e-06,
      "loss": 5.1541,
      "step": 20750
    },
    {
      "epoch": 0.83774727492935,
      "grad_norm": 93.47547149658203,
      "learning_rate": 1.3493508049231986e-06,
      "loss": 3.5309,
      "step": 20751
    },
    {
      "epoch": 0.8377876463463868,
      "grad_norm": 103.4242172241211,
      "learning_rate": 1.3486949181897435e-06,
      "loss": 6.8351,
      "step": 20752
    },
    {
      "epoch": 0.8378280177634235,
      "grad_norm": 120.06205749511719,
      "learning_rate": 1.3480391793734005e-06,
      "loss": 6.3822,
      "step": 20753
    },
    {
      "epoch": 0.8378683891804602,
      "grad_norm": 67.40303802490234,
      "learning_rate": 1.3473835884853837e-06,
      "loss": 5.7377,
      "step": 20754
    },
    {
      "epoch": 0.837908760597497,
      "grad_norm": 108.95995330810547,
      "learning_rate": 1.3467281455369019e-06,
      "loss": 3.1304,
      "step": 20755
    },
    {
      "epoch": 0.8379491320145337,
      "grad_norm": 70.77369689941406,
      "learning_rate": 1.3460728505391652e-06,
      "loss": 4.7371,
      "step": 20756
    },
    {
      "epoch": 0.8379895034315704,
      "grad_norm": 103.61397552490234,
      "learning_rate": 1.3454177035033712e-06,
      "loss": 4.7605,
      "step": 20757
    },
    {
      "epoch": 0.8380298748486071,
      "grad_norm": 144.0583953857422,
      "learning_rate": 1.3447627044407253e-06,
      "loss": 5.5075,
      "step": 20758
    },
    {
      "epoch": 0.838070246265644,
      "grad_norm": 90.78367614746094,
      "learning_rate": 1.3441078533624253e-06,
      "loss": 6.0885,
      "step": 20759
    },
    {
      "epoch": 0.8381106176826807,
      "grad_norm": 74.6025161743164,
      "learning_rate": 1.34345315027967e-06,
      "loss": 5.604,
      "step": 20760
    },
    {
      "epoch": 0.8381509890997174,
      "grad_norm": 110.0645523071289,
      "learning_rate": 1.3427985952036493e-06,
      "loss": 3.9587,
      "step": 20761
    },
    {
      "epoch": 0.8381913605167541,
      "grad_norm": 95.48501586914062,
      "learning_rate": 1.3421441881455565e-06,
      "loss": 4.4025,
      "step": 20762
    },
    {
      "epoch": 0.8382317319337909,
      "grad_norm": 66.8095932006836,
      "learning_rate": 1.3414899291165805e-06,
      "loss": 3.6532,
      "step": 20763
    },
    {
      "epoch": 0.8382721033508276,
      "grad_norm": 85.66835021972656,
      "learning_rate": 1.3408358181279103e-06,
      "loss": 3.2554,
      "step": 20764
    },
    {
      "epoch": 0.8383124747678643,
      "grad_norm": 95.71192169189453,
      "learning_rate": 1.3401818551907241e-06,
      "loss": 3.4694,
      "step": 20765
    },
    {
      "epoch": 0.8383528461849011,
      "grad_norm": 109.69544982910156,
      "learning_rate": 1.339528040316207e-06,
      "loss": 3.6135,
      "step": 20766
    },
    {
      "epoch": 0.8383932176019379,
      "grad_norm": 92.44312286376953,
      "learning_rate": 1.3388743735155363e-06,
      "loss": 4.3678,
      "step": 20767
    },
    {
      "epoch": 0.8384335890189746,
      "grad_norm": 75.61518859863281,
      "learning_rate": 1.3382208547998888e-06,
      "loss": 5.7915,
      "step": 20768
    },
    {
      "epoch": 0.8384739604360113,
      "grad_norm": 133.0439453125,
      "learning_rate": 1.3375674841804376e-06,
      "loss": 5.2016,
      "step": 20769
    },
    {
      "epoch": 0.838514331853048,
      "grad_norm": 88.21878814697266,
      "learning_rate": 1.336914261668354e-06,
      "loss": 6.2904,
      "step": 20770
    },
    {
      "epoch": 0.8385547032700847,
      "grad_norm": 76.50404357910156,
      "learning_rate": 1.3362611872748067e-06,
      "loss": 5.4293,
      "step": 20771
    },
    {
      "epoch": 0.8385950746871215,
      "grad_norm": 98.9074935913086,
      "learning_rate": 1.3356082610109644e-06,
      "loss": 4.9617,
      "step": 20772
    },
    {
      "epoch": 0.8386354461041583,
      "grad_norm": 96.3636703491211,
      "learning_rate": 1.3349554828879864e-06,
      "loss": 4.3621,
      "step": 20773
    },
    {
      "epoch": 0.838675817521195,
      "grad_norm": 136.914306640625,
      "learning_rate": 1.3343028529170343e-06,
      "loss": 4.0883,
      "step": 20774
    },
    {
      "epoch": 0.8387161889382317,
      "grad_norm": 88.78767395019531,
      "learning_rate": 1.3336503711092686e-06,
      "loss": 6.5049,
      "step": 20775
    },
    {
      "epoch": 0.8387565603552685,
      "grad_norm": 129.2324676513672,
      "learning_rate": 1.3329980374758455e-06,
      "loss": 3.1947,
      "step": 20776
    },
    {
      "epoch": 0.8387969317723052,
      "grad_norm": 103.12049865722656,
      "learning_rate": 1.3323458520279153e-06,
      "loss": 4.9374,
      "step": 20777
    },
    {
      "epoch": 0.8388373031893419,
      "grad_norm": 136.1335906982422,
      "learning_rate": 1.33169381477663e-06,
      "loss": 5.4218,
      "step": 20778
    },
    {
      "epoch": 0.8388776746063786,
      "grad_norm": 154.44557189941406,
      "learning_rate": 1.3310419257331385e-06,
      "loss": 5.7286,
      "step": 20779
    },
    {
      "epoch": 0.8389180460234155,
      "grad_norm": 97.71334838867188,
      "learning_rate": 1.3303901849085898e-06,
      "loss": 4.1122,
      "step": 20780
    },
    {
      "epoch": 0.8389584174404522,
      "grad_norm": 124.57941436767578,
      "learning_rate": 1.3297385923141215e-06,
      "loss": 6.4077,
      "step": 20781
    },
    {
      "epoch": 0.8389987888574889,
      "grad_norm": 133.5218505859375,
      "learning_rate": 1.3290871479608758e-06,
      "loss": 4.7037,
      "step": 20782
    },
    {
      "epoch": 0.8390391602745256,
      "grad_norm": 197.00575256347656,
      "learning_rate": 1.3284358518599928e-06,
      "loss": 5.4564,
      "step": 20783
    },
    {
      "epoch": 0.8390795316915624,
      "grad_norm": 89.76177978515625,
      "learning_rate": 1.3277847040226088e-06,
      "loss": 5.6009,
      "step": 20784
    },
    {
      "epoch": 0.8391199031085991,
      "grad_norm": 120.3857421875,
      "learning_rate": 1.3271337044598542e-06,
      "loss": 4.9941,
      "step": 20785
    },
    {
      "epoch": 0.8391602745256358,
      "grad_norm": 108.82364654541016,
      "learning_rate": 1.3264828531828588e-06,
      "loss": 5.7702,
      "step": 20786
    },
    {
      "epoch": 0.8392006459426726,
      "grad_norm": 87.84596252441406,
      "learning_rate": 1.3258321502027539e-06,
      "loss": 5.5936,
      "step": 20787
    },
    {
      "epoch": 0.8392410173597094,
      "grad_norm": 67.1761474609375,
      "learning_rate": 1.3251815955306635e-06,
      "loss": 3.0692,
      "step": 20788
    },
    {
      "epoch": 0.8392813887767461,
      "grad_norm": 136.7399444580078,
      "learning_rate": 1.3245311891777102e-06,
      "loss": 6.1613,
      "step": 20789
    },
    {
      "epoch": 0.8393217601937828,
      "grad_norm": 100.91128540039062,
      "learning_rate": 1.323880931155015e-06,
      "loss": 5.0897,
      "step": 20790
    },
    {
      "epoch": 0.8393621316108195,
      "grad_norm": 133.75218200683594,
      "learning_rate": 1.3232308214736955e-06,
      "loss": 4.4063,
      "step": 20791
    },
    {
      "epoch": 0.8394025030278562,
      "grad_norm": 103.18569946289062,
      "learning_rate": 1.3225808601448698e-06,
      "loss": 5.1444,
      "step": 20792
    },
    {
      "epoch": 0.839442874444893,
      "grad_norm": 95.623291015625,
      "learning_rate": 1.3219310471796454e-06,
      "loss": 4.47,
      "step": 20793
    },
    {
      "epoch": 0.8394832458619298,
      "grad_norm": 114.7974853515625,
      "learning_rate": 1.3212813825891357e-06,
      "loss": 3.768,
      "step": 20794
    },
    {
      "epoch": 0.8395236172789665,
      "grad_norm": 199.27403259277344,
      "learning_rate": 1.3206318663844486e-06,
      "loss": 5.2292,
      "step": 20795
    },
    {
      "epoch": 0.8395639886960032,
      "grad_norm": 96.45721435546875,
      "learning_rate": 1.3199824985766906e-06,
      "loss": 6.3517,
      "step": 20796
    },
    {
      "epoch": 0.83960436011304,
      "grad_norm": 97.57927703857422,
      "learning_rate": 1.3193332791769597e-06,
      "loss": 5.1166,
      "step": 20797
    },
    {
      "epoch": 0.8396447315300767,
      "grad_norm": 80.69548797607422,
      "learning_rate": 1.3186842081963603e-06,
      "loss": 4.7268,
      "step": 20798
    },
    {
      "epoch": 0.8396851029471134,
      "grad_norm": 79.86611938476562,
      "learning_rate": 1.3180352856459877e-06,
      "loss": 5.0681,
      "step": 20799
    },
    {
      "epoch": 0.8397254743641501,
      "grad_norm": 129.6678924560547,
      "learning_rate": 1.31738651153694e-06,
      "loss": 4.8538,
      "step": 20800
    },
    {
      "epoch": 0.839765845781187,
      "grad_norm": 99.74149322509766,
      "learning_rate": 1.316737885880306e-06,
      "loss": 4.867,
      "step": 20801
    },
    {
      "epoch": 0.8398062171982237,
      "grad_norm": 87.11067199707031,
      "learning_rate": 1.3160894086871768e-06,
      "loss": 4.3391,
      "step": 20802
    },
    {
      "epoch": 0.8398465886152604,
      "grad_norm": 71.97103118896484,
      "learning_rate": 1.3154410799686402e-06,
      "loss": 3.4506,
      "step": 20803
    },
    {
      "epoch": 0.8398869600322971,
      "grad_norm": 93.22403717041016,
      "learning_rate": 1.314792899735784e-06,
      "loss": 3.8526,
      "step": 20804
    },
    {
      "epoch": 0.8399273314493338,
      "grad_norm": 89.36182403564453,
      "learning_rate": 1.3141448679996849e-06,
      "loss": 5.705,
      "step": 20805
    },
    {
      "epoch": 0.8399677028663706,
      "grad_norm": 92.42964172363281,
      "learning_rate": 1.3134969847714262e-06,
      "loss": 4.462,
      "step": 20806
    },
    {
      "epoch": 0.8400080742834074,
      "grad_norm": 90.72949981689453,
      "learning_rate": 1.3128492500620838e-06,
      "loss": 3.0014,
      "step": 20807
    },
    {
      "epoch": 0.8400484457004441,
      "grad_norm": 72.5210189819336,
      "learning_rate": 1.312201663882734e-06,
      "loss": 3.2299,
      "step": 20808
    },
    {
      "epoch": 0.8400888171174808,
      "grad_norm": 115.58159637451172,
      "learning_rate": 1.311554226244448e-06,
      "loss": 4.6697,
      "step": 20809
    },
    {
      "epoch": 0.8401291885345176,
      "grad_norm": 80.40332794189453,
      "learning_rate": 1.3109069371582961e-06,
      "loss": 4.7938,
      "step": 20810
    },
    {
      "epoch": 0.8401695599515543,
      "grad_norm": 110.74178314208984,
      "learning_rate": 1.3102597966353448e-06,
      "loss": 4.0288,
      "step": 20811
    },
    {
      "epoch": 0.840209931368591,
      "grad_norm": 82.33021545410156,
      "learning_rate": 1.3096128046866619e-06,
      "loss": 4.9035,
      "step": 20812
    },
    {
      "epoch": 0.8402503027856277,
      "grad_norm": 101.79910278320312,
      "learning_rate": 1.308965961323303e-06,
      "loss": 5.7864,
      "step": 20813
    },
    {
      "epoch": 0.8402906742026646,
      "grad_norm": 100.4284439086914,
      "learning_rate": 1.3083192665563315e-06,
      "loss": 5.2619,
      "step": 20814
    },
    {
      "epoch": 0.8403310456197013,
      "grad_norm": 109.34962463378906,
      "learning_rate": 1.3076727203968054e-06,
      "loss": 4.3771,
      "step": 20815
    },
    {
      "epoch": 0.840371417036738,
      "grad_norm": 66.58356475830078,
      "learning_rate": 1.307026322855779e-06,
      "loss": 5.4684,
      "step": 20816
    },
    {
      "epoch": 0.8404117884537747,
      "grad_norm": 82.59697723388672,
      "learning_rate": 1.3063800739443e-06,
      "loss": 5.2891,
      "step": 20817
    },
    {
      "epoch": 0.8404521598708115,
      "grad_norm": 104.10381317138672,
      "learning_rate": 1.305733973673421e-06,
      "loss": 5.3276,
      "step": 20818
    },
    {
      "epoch": 0.8404925312878482,
      "grad_norm": 89.3401870727539,
      "learning_rate": 1.3050880220541895e-06,
      "loss": 3.5579,
      "step": 20819
    },
    {
      "epoch": 0.8405329027048849,
      "grad_norm": 77.20881652832031,
      "learning_rate": 1.304442219097649e-06,
      "loss": 5.0608,
      "step": 20820
    },
    {
      "epoch": 0.8405732741219217,
      "grad_norm": 85.09742736816406,
      "learning_rate": 1.303796564814841e-06,
      "loss": 4.0784,
      "step": 20821
    },
    {
      "epoch": 0.8406136455389585,
      "grad_norm": 92.31368255615234,
      "learning_rate": 1.3031510592168028e-06,
      "loss": 3.4932,
      "step": 20822
    },
    {
      "epoch": 0.8406540169559952,
      "grad_norm": 81.62789154052734,
      "learning_rate": 1.3025057023145738e-06,
      "loss": 4.9167,
      "step": 20823
    },
    {
      "epoch": 0.8406943883730319,
      "grad_norm": 104.50179290771484,
      "learning_rate": 1.3018604941191892e-06,
      "loss": 4.7741,
      "step": 20824
    },
    {
      "epoch": 0.8407347597900686,
      "grad_norm": 148.70062255859375,
      "learning_rate": 1.3012154346416772e-06,
      "loss": 5.0678,
      "step": 20825
    },
    {
      "epoch": 0.8407751312071053,
      "grad_norm": 91.42191314697266,
      "learning_rate": 1.3005705238930666e-06,
      "loss": 4.7548,
      "step": 20826
    },
    {
      "epoch": 0.8408155026241421,
      "grad_norm": 187.41885375976562,
      "learning_rate": 1.2999257618843862e-06,
      "loss": 4.8586,
      "step": 20827
    },
    {
      "epoch": 0.8408558740411789,
      "grad_norm": 91.15899658203125,
      "learning_rate": 1.2992811486266587e-06,
      "loss": 4.9257,
      "step": 20828
    },
    {
      "epoch": 0.8408962454582156,
      "grad_norm": 154.76522827148438,
      "learning_rate": 1.2986366841309073e-06,
      "loss": 5.5593,
      "step": 20829
    },
    {
      "epoch": 0.8409366168752523,
      "grad_norm": 97.25311279296875,
      "learning_rate": 1.2979923684081485e-06,
      "loss": 6.2137,
      "step": 20830
    },
    {
      "epoch": 0.8409769882922891,
      "grad_norm": 147.01039123535156,
      "learning_rate": 1.2973482014693995e-06,
      "loss": 4.3167,
      "step": 20831
    },
    {
      "epoch": 0.8410173597093258,
      "grad_norm": 68.69646453857422,
      "learning_rate": 1.2967041833256755e-06,
      "loss": 6.1649,
      "step": 20832
    },
    {
      "epoch": 0.8410577311263625,
      "grad_norm": 83.6778793334961,
      "learning_rate": 1.2960603139879857e-06,
      "loss": 4.6785,
      "step": 20833
    },
    {
      "epoch": 0.8410981025433992,
      "grad_norm": 78.43673706054688,
      "learning_rate": 1.295416593467338e-06,
      "loss": 5.7465,
      "step": 20834
    },
    {
      "epoch": 0.8411384739604361,
      "grad_norm": 84.30950927734375,
      "learning_rate": 1.2947730217747412e-06,
      "loss": 5.3578,
      "step": 20835
    },
    {
      "epoch": 0.8411788453774728,
      "grad_norm": 103.1680908203125,
      "learning_rate": 1.2941295989211988e-06,
      "loss": 5.8796,
      "step": 20836
    },
    {
      "epoch": 0.8412192167945095,
      "grad_norm": 174.7863311767578,
      "learning_rate": 1.2934863249177087e-06,
      "loss": 4.8134,
      "step": 20837
    },
    {
      "epoch": 0.8412595882115462,
      "grad_norm": 70.30364990234375,
      "learning_rate": 1.292843199775271e-06,
      "loss": 5.9504,
      "step": 20838
    },
    {
      "epoch": 0.841299959628583,
      "grad_norm": 100.88379669189453,
      "learning_rate": 1.2922002235048824e-06,
      "loss": 3.7043,
      "step": 20839
    },
    {
      "epoch": 0.8413403310456197,
      "grad_norm": 71.63880920410156,
      "learning_rate": 1.2915573961175376e-06,
      "loss": 4.5955,
      "step": 20840
    },
    {
      "epoch": 0.8413807024626564,
      "grad_norm": 78.31897735595703,
      "learning_rate": 1.2909147176242232e-06,
      "loss": 4.714,
      "step": 20841
    },
    {
      "epoch": 0.8414210738796932,
      "grad_norm": 109.96257019042969,
      "learning_rate": 1.2902721880359304e-06,
      "loss": 5.1527,
      "step": 20842
    },
    {
      "epoch": 0.84146144529673,
      "grad_norm": 78.45755767822266,
      "learning_rate": 1.289629807363645e-06,
      "loss": 3.7007,
      "step": 20843
    },
    {
      "epoch": 0.8415018167137667,
      "grad_norm": 102.8080062866211,
      "learning_rate": 1.2889875756183512e-06,
      "loss": 5.0453,
      "step": 20844
    },
    {
      "epoch": 0.8415421881308034,
      "grad_norm": 88.92378234863281,
      "learning_rate": 1.288345492811026e-06,
      "loss": 3.9031,
      "step": 20845
    },
    {
      "epoch": 0.8415825595478401,
      "grad_norm": 123.95745849609375,
      "learning_rate": 1.2877035589526499e-06,
      "loss": 3.2007,
      "step": 20846
    },
    {
      "epoch": 0.8416229309648768,
      "grad_norm": 126.90313720703125,
      "learning_rate": 1.287061774054198e-06,
      "loss": 5.3141,
      "step": 20847
    },
    {
      "epoch": 0.8416633023819136,
      "grad_norm": 130.54776000976562,
      "learning_rate": 1.286420138126644e-06,
      "loss": 4.3601,
      "step": 20848
    },
    {
      "epoch": 0.8417036737989504,
      "grad_norm": 80.8821029663086,
      "learning_rate": 1.2857786511809578e-06,
      "loss": 4.627,
      "step": 20849
    },
    {
      "epoch": 0.8417440452159871,
      "grad_norm": 97.25579071044922,
      "learning_rate": 1.2851373132281063e-06,
      "loss": 3.5446,
      "step": 20850
    },
    {
      "epoch": 0.8417844166330238,
      "grad_norm": 65.60910034179688,
      "learning_rate": 1.2844961242790576e-06,
      "loss": 5.8125,
      "step": 20851
    },
    {
      "epoch": 0.8418247880500606,
      "grad_norm": 120.68794250488281,
      "learning_rate": 1.283855084344775e-06,
      "loss": 3.8024,
      "step": 20852
    },
    {
      "epoch": 0.8418651594670973,
      "grad_norm": 158.32847595214844,
      "learning_rate": 1.2832141934362141e-06,
      "loss": 5.0082,
      "step": 20853
    },
    {
      "epoch": 0.841905530884134,
      "grad_norm": 86.65982055664062,
      "learning_rate": 1.2825734515643363e-06,
      "loss": 4.7632,
      "step": 20854
    },
    {
      "epoch": 0.8419459023011707,
      "grad_norm": 86.40419006347656,
      "learning_rate": 1.2819328587400948e-06,
      "loss": 4.8735,
      "step": 20855
    },
    {
      "epoch": 0.8419862737182076,
      "grad_norm": 83.39673614501953,
      "learning_rate": 1.2812924149744465e-06,
      "loss": 6.2232,
      "step": 20856
    },
    {
      "epoch": 0.8420266451352443,
      "grad_norm": 142.72776794433594,
      "learning_rate": 1.280652120278335e-06,
      "loss": 7.1882,
      "step": 20857
    },
    {
      "epoch": 0.842067016552281,
      "grad_norm": 71.7986831665039,
      "learning_rate": 1.2800119746627116e-06,
      "loss": 3.6314,
      "step": 20858
    },
    {
      "epoch": 0.8421073879693177,
      "grad_norm": 75.99092102050781,
      "learning_rate": 1.2793719781385206e-06,
      "loss": 4.4345,
      "step": 20859
    },
    {
      "epoch": 0.8421477593863544,
      "grad_norm": 77.52892303466797,
      "learning_rate": 1.2787321307167066e-06,
      "loss": 5.6317,
      "step": 20860
    },
    {
      "epoch": 0.8421881308033912,
      "grad_norm": 113.17842102050781,
      "learning_rate": 1.2780924324082057e-06,
      "loss": 6.1199,
      "step": 20861
    },
    {
      "epoch": 0.8422285022204279,
      "grad_norm": 111.48226928710938,
      "learning_rate": 1.2774528832239575e-06,
      "loss": 5.1442,
      "step": 20862
    },
    {
      "epoch": 0.8422688736374647,
      "grad_norm": 97.04344940185547,
      "learning_rate": 1.2768134831748957e-06,
      "loss": 5.4955,
      "step": 20863
    },
    {
      "epoch": 0.8423092450545014,
      "grad_norm": 104.93413543701172,
      "learning_rate": 1.276174232271956e-06,
      "loss": 5.2571,
      "step": 20864
    },
    {
      "epoch": 0.8423496164715382,
      "grad_norm": 79.36848449707031,
      "learning_rate": 1.2755351305260622e-06,
      "loss": 5.7122,
      "step": 20865
    },
    {
      "epoch": 0.8423899878885749,
      "grad_norm": 106.9091796875,
      "learning_rate": 1.274896177948146e-06,
      "loss": 4.443,
      "step": 20866
    },
    {
      "epoch": 0.8424303593056116,
      "grad_norm": 121.44320678710938,
      "learning_rate": 1.2742573745491293e-06,
      "loss": 4.7174,
      "step": 20867
    },
    {
      "epoch": 0.8424707307226483,
      "grad_norm": 158.33021545410156,
      "learning_rate": 1.2736187203399365e-06,
      "loss": 5.059,
      "step": 20868
    },
    {
      "epoch": 0.842511102139685,
      "grad_norm": 91.04371643066406,
      "learning_rate": 1.2729802153314851e-06,
      "loss": 5.4791,
      "step": 20869
    },
    {
      "epoch": 0.8425514735567219,
      "grad_norm": 89.05036926269531,
      "learning_rate": 1.2723418595346925e-06,
      "loss": 4.2355,
      "step": 20870
    },
    {
      "epoch": 0.8425918449737586,
      "grad_norm": 129.0382843017578,
      "learning_rate": 1.2717036529604754e-06,
      "loss": 5.4081,
      "step": 20871
    },
    {
      "epoch": 0.8426322163907953,
      "grad_norm": 134.5979766845703,
      "learning_rate": 1.2710655956197448e-06,
      "loss": 6.2164,
      "step": 20872
    },
    {
      "epoch": 0.842672587807832,
      "grad_norm": 114.77124786376953,
      "learning_rate": 1.2704276875234068e-06,
      "loss": 6.267,
      "step": 20873
    },
    {
      "epoch": 0.8427129592248688,
      "grad_norm": 90.11260986328125,
      "learning_rate": 1.2697899286823702e-06,
      "loss": 5.9387,
      "step": 20874
    },
    {
      "epoch": 0.8427533306419055,
      "grad_norm": 78.00878143310547,
      "learning_rate": 1.2691523191075395e-06,
      "loss": 4.3756,
      "step": 20875
    },
    {
      "epoch": 0.8427937020589422,
      "grad_norm": 151.33309936523438,
      "learning_rate": 1.2685148588098185e-06,
      "loss": 5.2222,
      "step": 20876
    },
    {
      "epoch": 0.842834073475979,
      "grad_norm": 69.32770538330078,
      "learning_rate": 1.2678775478001015e-06,
      "loss": 4.3997,
      "step": 20877
    },
    {
      "epoch": 0.8428744448930158,
      "grad_norm": 116.137451171875,
      "learning_rate": 1.267240386089288e-06,
      "loss": 5.8504,
      "step": 20878
    },
    {
      "epoch": 0.8429148163100525,
      "grad_norm": 76.5595703125,
      "learning_rate": 1.266603373688271e-06,
      "loss": 4.7522,
      "step": 20879
    },
    {
      "epoch": 0.8429551877270892,
      "grad_norm": 101.6651840209961,
      "learning_rate": 1.2659665106079444e-06,
      "loss": 5.3384,
      "step": 20880
    },
    {
      "epoch": 0.8429955591441259,
      "grad_norm": 95.49639129638672,
      "learning_rate": 1.2653297968591926e-06,
      "loss": 4.9834,
      "step": 20881
    },
    {
      "epoch": 0.8430359305611627,
      "grad_norm": 99.58517456054688,
      "learning_rate": 1.2646932324529049e-06,
      "loss": 4.6391,
      "step": 20882
    },
    {
      "epoch": 0.8430763019781994,
      "grad_norm": 104.4239501953125,
      "learning_rate": 1.2640568173999646e-06,
      "loss": 5.2286,
      "step": 20883
    },
    {
      "epoch": 0.8431166733952362,
      "grad_norm": 84.65894317626953,
      "learning_rate": 1.2634205517112552e-06,
      "loss": 6.4216,
      "step": 20884
    },
    {
      "epoch": 0.8431570448122729,
      "grad_norm": 94.80441284179688,
      "learning_rate": 1.2627844353976504e-06,
      "loss": 4.9109,
      "step": 20885
    },
    {
      "epoch": 0.8431974162293097,
      "grad_norm": 119.32025146484375,
      "learning_rate": 1.2621484684700281e-06,
      "loss": 6.9326,
      "step": 20886
    },
    {
      "epoch": 0.8432377876463464,
      "grad_norm": 90.51126861572266,
      "learning_rate": 1.2615126509392639e-06,
      "loss": 4.8247,
      "step": 20887
    },
    {
      "epoch": 0.8432781590633831,
      "grad_norm": 93.61332702636719,
      "learning_rate": 1.260876982816227e-06,
      "loss": 6.6074,
      "step": 20888
    },
    {
      "epoch": 0.8433185304804198,
      "grad_norm": 73.10279083251953,
      "learning_rate": 1.2602414641117866e-06,
      "loss": 5.0078,
      "step": 20889
    },
    {
      "epoch": 0.8433589018974565,
      "grad_norm": 77.86600494384766,
      "learning_rate": 1.2596060948368083e-06,
      "loss": 4.4858,
      "step": 20890
    },
    {
      "epoch": 0.8433992733144934,
      "grad_norm": 189.0519561767578,
      "learning_rate": 1.2589708750021555e-06,
      "loss": 4.4257,
      "step": 20891
    },
    {
      "epoch": 0.8434396447315301,
      "grad_norm": 87.55287170410156,
      "learning_rate": 1.258335804618691e-06,
      "loss": 5.4398,
      "step": 20892
    },
    {
      "epoch": 0.8434800161485668,
      "grad_norm": 97.69432067871094,
      "learning_rate": 1.2577008836972682e-06,
      "loss": 3.9139,
      "step": 20893
    },
    {
      "epoch": 0.8435203875656035,
      "grad_norm": 98.51444244384766,
      "learning_rate": 1.257066112248747e-06,
      "loss": 3.3895,
      "step": 20894
    },
    {
      "epoch": 0.8435607589826403,
      "grad_norm": 77.45858001708984,
      "learning_rate": 1.256431490283978e-06,
      "loss": 3.9615,
      "step": 20895
    },
    {
      "epoch": 0.843601130399677,
      "grad_norm": 82.67858123779297,
      "learning_rate": 1.2557970178138157e-06,
      "loss": 4.7996,
      "step": 20896
    },
    {
      "epoch": 0.8436415018167137,
      "grad_norm": 115.04928588867188,
      "learning_rate": 1.2551626948491025e-06,
      "loss": 3.6685,
      "step": 20897
    },
    {
      "epoch": 0.8436818732337505,
      "grad_norm": 107.50713348388672,
      "learning_rate": 1.2545285214006864e-06,
      "loss": 5.6341,
      "step": 20898
    },
    {
      "epoch": 0.8437222446507873,
      "grad_norm": 87.90291595458984,
      "learning_rate": 1.253894497479411e-06,
      "loss": 5.0303,
      "step": 20899
    },
    {
      "epoch": 0.843762616067824,
      "grad_norm": 104.93889617919922,
      "learning_rate": 1.2532606230961176e-06,
      "loss": 2.3536,
      "step": 20900
    },
    {
      "epoch": 0.8438029874848607,
      "grad_norm": 100.85298156738281,
      "learning_rate": 1.2526268982616407e-06,
      "loss": 4.8045,
      "step": 20901
    },
    {
      "epoch": 0.8438433589018974,
      "grad_norm": 77.25485229492188,
      "learning_rate": 1.2519933229868175e-06,
      "loss": 5.3795,
      "step": 20902
    },
    {
      "epoch": 0.8438837303189342,
      "grad_norm": 95.78577423095703,
      "learning_rate": 1.2513598972824792e-06,
      "loss": 4.0976,
      "step": 20903
    },
    {
      "epoch": 0.8439241017359709,
      "grad_norm": 103.03639221191406,
      "learning_rate": 1.2507266211594604e-06,
      "loss": 4.9044,
      "step": 20904
    },
    {
      "epoch": 0.8439644731530077,
      "grad_norm": 84.3770980834961,
      "learning_rate": 1.2500934946285836e-06,
      "loss": 4.9528,
      "step": 20905
    },
    {
      "epoch": 0.8440048445700444,
      "grad_norm": 71.36823272705078,
      "learning_rate": 1.2494605177006746e-06,
      "loss": 4.4443,
      "step": 20906
    },
    {
      "epoch": 0.8440452159870812,
      "grad_norm": 111.1731948852539,
      "learning_rate": 1.2488276903865571e-06,
      "loss": 4.7219,
      "step": 20907
    },
    {
      "epoch": 0.8440855874041179,
      "grad_norm": 93.28646087646484,
      "learning_rate": 1.24819501269705e-06,
      "loss": 6.2697,
      "step": 20908
    },
    {
      "epoch": 0.8441259588211546,
      "grad_norm": 98.76571655273438,
      "learning_rate": 1.2475624846429723e-06,
      "loss": 5.9674,
      "step": 20909
    },
    {
      "epoch": 0.8441663302381913,
      "grad_norm": 146.03317260742188,
      "learning_rate": 1.2469301062351368e-06,
      "loss": 4.8703,
      "step": 20910
    },
    {
      "epoch": 0.8442067016552282,
      "grad_norm": 127.3192138671875,
      "learning_rate": 1.246297877484357e-06,
      "loss": 3.9559,
      "step": 20911
    },
    {
      "epoch": 0.8442470730722649,
      "grad_norm": 94.52242279052734,
      "learning_rate": 1.2456657984014443e-06,
      "loss": 4.4678,
      "step": 20912
    },
    {
      "epoch": 0.8442874444893016,
      "grad_norm": 169.3961944580078,
      "learning_rate": 1.245033868997202e-06,
      "loss": 4.918,
      "step": 20913
    },
    {
      "epoch": 0.8443278159063383,
      "grad_norm": 103.51587677001953,
      "learning_rate": 1.244402089282436e-06,
      "loss": 2.7976,
      "step": 20914
    },
    {
      "epoch": 0.844368187323375,
      "grad_norm": 114.11058807373047,
      "learning_rate": 1.2437704592679478e-06,
      "loss": 5.2165,
      "step": 20915
    },
    {
      "epoch": 0.8444085587404118,
      "grad_norm": 99.67086791992188,
      "learning_rate": 1.2431389789645399e-06,
      "loss": 4.1383,
      "step": 20916
    },
    {
      "epoch": 0.8444489301574485,
      "grad_norm": 114.83358764648438,
      "learning_rate": 1.2425076483830045e-06,
      "loss": 6.1283,
      "step": 20917
    },
    {
      "epoch": 0.8444893015744853,
      "grad_norm": 77.8823471069336,
      "learning_rate": 1.2418764675341388e-06,
      "loss": 4.9773,
      "step": 20918
    },
    {
      "epoch": 0.844529672991522,
      "grad_norm": 99.6378173828125,
      "learning_rate": 1.2412454364287329e-06,
      "loss": 4.652,
      "step": 20919
    },
    {
      "epoch": 0.8445700444085588,
      "grad_norm": 95.9516372680664,
      "learning_rate": 1.2406145550775795e-06,
      "loss": 4.4079,
      "step": 20920
    },
    {
      "epoch": 0.8446104158255955,
      "grad_norm": 82.29264831542969,
      "learning_rate": 1.2399838234914607e-06,
      "loss": 4.4903,
      "step": 20921
    },
    {
      "epoch": 0.8446507872426322,
      "grad_norm": 76.47246551513672,
      "learning_rate": 1.2393532416811615e-06,
      "loss": 4.824,
      "step": 20922
    },
    {
      "epoch": 0.8446911586596689,
      "grad_norm": 116.71135711669922,
      "learning_rate": 1.2387228096574654e-06,
      "loss": 5.3797,
      "step": 20923
    },
    {
      "epoch": 0.8447315300767057,
      "grad_norm": 102.88832092285156,
      "learning_rate": 1.2380925274311518e-06,
      "loss": 5.6237,
      "step": 20924
    },
    {
      "epoch": 0.8447719014937425,
      "grad_norm": 92.43461608886719,
      "learning_rate": 1.2374623950129927e-06,
      "loss": 4.9062,
      "step": 20925
    },
    {
      "epoch": 0.8448122729107792,
      "grad_norm": 112.89231872558594,
      "learning_rate": 1.2368324124137654e-06,
      "loss": 5.5004,
      "step": 20926
    },
    {
      "epoch": 0.8448526443278159,
      "grad_norm": 138.91038513183594,
      "learning_rate": 1.23620257964424e-06,
      "loss": 5.9286,
      "step": 20927
    },
    {
      "epoch": 0.8448930157448526,
      "grad_norm": 86.44486236572266,
      "learning_rate": 1.235572896715186e-06,
      "loss": 5.319,
      "step": 20928
    },
    {
      "epoch": 0.8449333871618894,
      "grad_norm": 83.5255355834961,
      "learning_rate": 1.2349433636373687e-06,
      "loss": 4.4718,
      "step": 20929
    },
    {
      "epoch": 0.8449737585789261,
      "grad_norm": 132.9522705078125,
      "learning_rate": 1.234313980421552e-06,
      "loss": 4.7484,
      "step": 20930
    },
    {
      "epoch": 0.8450141299959628,
      "grad_norm": 146.13095092773438,
      "learning_rate": 1.2336847470784975e-06,
      "loss": 4.523,
      "step": 20931
    },
    {
      "epoch": 0.8450545014129996,
      "grad_norm": 72.87489318847656,
      "learning_rate": 1.2330556636189629e-06,
      "loss": 5.6201,
      "step": 20932
    },
    {
      "epoch": 0.8450948728300364,
      "grad_norm": 73.1280288696289,
      "learning_rate": 1.232426730053705e-06,
      "loss": 4.3494,
      "step": 20933
    },
    {
      "epoch": 0.8451352442470731,
      "grad_norm": 395.6648254394531,
      "learning_rate": 1.231797946393476e-06,
      "loss": 6.0249,
      "step": 20934
    },
    {
      "epoch": 0.8451756156641098,
      "grad_norm": 113.24626922607422,
      "learning_rate": 1.2311693126490255e-06,
      "loss": 3.9175,
      "step": 20935
    },
    {
      "epoch": 0.8452159870811465,
      "grad_norm": 96.63506317138672,
      "learning_rate": 1.2305408288311038e-06,
      "loss": 5.4617,
      "step": 20936
    },
    {
      "epoch": 0.8452563584981833,
      "grad_norm": 82.50157928466797,
      "learning_rate": 1.229912494950457e-06,
      "loss": 4.3322,
      "step": 20937
    },
    {
      "epoch": 0.84529672991522,
      "grad_norm": 89.64640808105469,
      "learning_rate": 1.2292843110178254e-06,
      "loss": 3.2188,
      "step": 20938
    },
    {
      "epoch": 0.8453371013322568,
      "grad_norm": 123.38435363769531,
      "learning_rate": 1.2286562770439514e-06,
      "loss": 3.1956,
      "step": 20939
    },
    {
      "epoch": 0.8453774727492935,
      "grad_norm": 115.84037017822266,
      "learning_rate": 1.228028393039572e-06,
      "loss": 4.7196,
      "step": 20940
    },
    {
      "epoch": 0.8454178441663303,
      "grad_norm": 77.3773193359375,
      "learning_rate": 1.2274006590154254e-06,
      "loss": 6.0622,
      "step": 20941
    },
    {
      "epoch": 0.845458215583367,
      "grad_norm": 83.98635864257812,
      "learning_rate": 1.2267730749822393e-06,
      "loss": 3.8903,
      "step": 20942
    },
    {
      "epoch": 0.8454985870004037,
      "grad_norm": 117.02928924560547,
      "learning_rate": 1.2261456409507476e-06,
      "loss": 3.8189,
      "step": 20943
    },
    {
      "epoch": 0.8455389584174404,
      "grad_norm": 106.89020538330078,
      "learning_rate": 1.2255183569316764e-06,
      "loss": 5.2216,
      "step": 20944
    },
    {
      "epoch": 0.8455793298344771,
      "grad_norm": 95.58472442626953,
      "learning_rate": 1.2248912229357512e-06,
      "loss": 4.7344,
      "step": 20945
    },
    {
      "epoch": 0.845619701251514,
      "grad_norm": 64.8414535522461,
      "learning_rate": 1.2242642389736959e-06,
      "loss": 5.7403,
      "step": 20946
    },
    {
      "epoch": 0.8456600726685507,
      "grad_norm": 93.87722778320312,
      "learning_rate": 1.2236374050562282e-06,
      "loss": 5.0318,
      "step": 20947
    },
    {
      "epoch": 0.8457004440855874,
      "grad_norm": 123.1841049194336,
      "learning_rate": 1.2230107211940667e-06,
      "loss": 4.6633,
      "step": 20948
    },
    {
      "epoch": 0.8457408155026241,
      "grad_norm": 108.26016235351562,
      "learning_rate": 1.2223841873979292e-06,
      "loss": 5.3396,
      "step": 20949
    },
    {
      "epoch": 0.8457811869196609,
      "grad_norm": 93.90486907958984,
      "learning_rate": 1.2217578036785215e-06,
      "loss": 4.5056,
      "step": 20950
    },
    {
      "epoch": 0.8458215583366976,
      "grad_norm": 102.87751770019531,
      "learning_rate": 1.2211315700465576e-06,
      "loss": 4.7839,
      "step": 20951
    },
    {
      "epoch": 0.8458619297537343,
      "grad_norm": 76.20039367675781,
      "learning_rate": 1.2205054865127431e-06,
      "loss": 4.6247,
      "step": 20952
    },
    {
      "epoch": 0.8459023011707711,
      "grad_norm": 103.760009765625,
      "learning_rate": 1.219879553087785e-06,
      "loss": 3.2372,
      "step": 20953
    },
    {
      "epoch": 0.8459426725878079,
      "grad_norm": 83.06729125976562,
      "learning_rate": 1.2192537697823826e-06,
      "loss": 4.0927,
      "step": 20954
    },
    {
      "epoch": 0.8459830440048446,
      "grad_norm": 140.45826721191406,
      "learning_rate": 1.218628136607235e-06,
      "loss": 4.1254,
      "step": 20955
    },
    {
      "epoch": 0.8460234154218813,
      "grad_norm": 76.4916763305664,
      "learning_rate": 1.2180026535730405e-06,
      "loss": 3.5109,
      "step": 20956
    },
    {
      "epoch": 0.846063786838918,
      "grad_norm": 93.2079086303711,
      "learning_rate": 1.217377320690496e-06,
      "loss": 5.8549,
      "step": 20957
    },
    {
      "epoch": 0.8461041582559548,
      "grad_norm": 228.15858459472656,
      "learning_rate": 1.2167521379702873e-06,
      "loss": 4.8902,
      "step": 20958
    },
    {
      "epoch": 0.8461445296729915,
      "grad_norm": 65.8132553100586,
      "learning_rate": 1.2161271054231083e-06,
      "loss": 3.5026,
      "step": 20959
    },
    {
      "epoch": 0.8461849010900283,
      "grad_norm": 119.52533721923828,
      "learning_rate": 1.2155022230596425e-06,
      "loss": 4.5698,
      "step": 20960
    },
    {
      "epoch": 0.846225272507065,
      "grad_norm": 153.6275177001953,
      "learning_rate": 1.2148774908905782e-06,
      "loss": 5.3488,
      "step": 20961
    },
    {
      "epoch": 0.8462656439241018,
      "grad_norm": 108.0366439819336,
      "learning_rate": 1.214252908926592e-06,
      "loss": 3.521,
      "step": 20962
    },
    {
      "epoch": 0.8463060153411385,
      "grad_norm": 151.48277282714844,
      "learning_rate": 1.2136284771783646e-06,
      "loss": 4.2567,
      "step": 20963
    },
    {
      "epoch": 0.8463463867581752,
      "grad_norm": 86.15007781982422,
      "learning_rate": 1.2130041956565718e-06,
      "loss": 4.7561,
      "step": 20964
    },
    {
      "epoch": 0.8463867581752119,
      "grad_norm": 114.82584381103516,
      "learning_rate": 1.2123800643718897e-06,
      "loss": 4.0914,
      "step": 20965
    },
    {
      "epoch": 0.8464271295922486,
      "grad_norm": 85.05149841308594,
      "learning_rate": 1.211756083334986e-06,
      "loss": 5.8656,
      "step": 20966
    },
    {
      "epoch": 0.8464675010092855,
      "grad_norm": 90.6574935913086,
      "learning_rate": 1.2111322525565329e-06,
      "loss": 5.9088,
      "step": 20967
    },
    {
      "epoch": 0.8465078724263222,
      "grad_norm": 84.98021697998047,
      "learning_rate": 1.2105085720471942e-06,
      "loss": 4.5895,
      "step": 20968
    },
    {
      "epoch": 0.8465482438433589,
      "grad_norm": 69.10905456542969,
      "learning_rate": 1.2098850418176366e-06,
      "loss": 3.7529,
      "step": 20969
    },
    {
      "epoch": 0.8465886152603956,
      "grad_norm": 153.1175079345703,
      "learning_rate": 1.2092616618785158e-06,
      "loss": 3.5376,
      "step": 20970
    },
    {
      "epoch": 0.8466289866774324,
      "grad_norm": 103.23204803466797,
      "learning_rate": 1.2086384322404932e-06,
      "loss": 5.336,
      "step": 20971
    },
    {
      "epoch": 0.8466693580944691,
      "grad_norm": 102.67990112304688,
      "learning_rate": 1.2080153529142235e-06,
      "loss": 3.9617,
      "step": 20972
    },
    {
      "epoch": 0.8467097295115058,
      "grad_norm": 140.82057189941406,
      "learning_rate": 1.2073924239103629e-06,
      "loss": 4.4862,
      "step": 20973
    },
    {
      "epoch": 0.8467501009285426,
      "grad_norm": 96.19841003417969,
      "learning_rate": 1.2067696452395572e-06,
      "loss": 5.5536,
      "step": 20974
    },
    {
      "epoch": 0.8467904723455794,
      "grad_norm": 88.57594299316406,
      "learning_rate": 1.2061470169124568e-06,
      "loss": 2.3013,
      "step": 20975
    },
    {
      "epoch": 0.8468308437626161,
      "grad_norm": 116.52664184570312,
      "learning_rate": 1.2055245389397085e-06,
      "loss": 5.5408,
      "step": 20976
    },
    {
      "epoch": 0.8468712151796528,
      "grad_norm": 112.31103515625,
      "learning_rate": 1.2049022113319554e-06,
      "loss": 4.8062,
      "step": 20977
    },
    {
      "epoch": 0.8469115865966895,
      "grad_norm": 88.15109252929688,
      "learning_rate": 1.204280034099834e-06,
      "loss": 4.1641,
      "step": 20978
    },
    {
      "epoch": 0.8469519580137262,
      "grad_norm": 95.79328155517578,
      "learning_rate": 1.203658007253984e-06,
      "loss": 4.4158,
      "step": 20979
    },
    {
      "epoch": 0.846992329430763,
      "grad_norm": 123.85271453857422,
      "learning_rate": 1.203036130805042e-06,
      "loss": 3.165,
      "step": 20980
    },
    {
      "epoch": 0.8470327008477998,
      "grad_norm": 98.50346374511719,
      "learning_rate": 1.2024144047636422e-06,
      "loss": 4.3728,
      "step": 20981
    },
    {
      "epoch": 0.8470730722648365,
      "grad_norm": 110.56993103027344,
      "learning_rate": 1.201792829140409e-06,
      "loss": 5.6823,
      "step": 20982
    },
    {
      "epoch": 0.8471134436818732,
      "grad_norm": 79.10232543945312,
      "learning_rate": 1.2011714039459744e-06,
      "loss": 5.0355,
      "step": 20983
    },
    {
      "epoch": 0.84715381509891,
      "grad_norm": 84.60009765625,
      "learning_rate": 1.2005501291909615e-06,
      "loss": 3.4041,
      "step": 20984
    },
    {
      "epoch": 0.8471941865159467,
      "grad_norm": 179.55673217773438,
      "learning_rate": 1.1999290048859934e-06,
      "loss": 4.9673,
      "step": 20985
    },
    {
      "epoch": 0.8472345579329834,
      "grad_norm": 105.20747375488281,
      "learning_rate": 1.1993080310416894e-06,
      "loss": 5.0983,
      "step": 20986
    },
    {
      "epoch": 0.8472749293500201,
      "grad_norm": 85.17156982421875,
      "learning_rate": 1.1986872076686662e-06,
      "loss": 4.7825,
      "step": 20987
    },
    {
      "epoch": 0.847315300767057,
      "grad_norm": 110.12821197509766,
      "learning_rate": 1.198066534777541e-06,
      "loss": 4.855,
      "step": 20988
    },
    {
      "epoch": 0.8473556721840937,
      "grad_norm": 122.51261138916016,
      "learning_rate": 1.1974460123789244e-06,
      "loss": 4.7437,
      "step": 20989
    },
    {
      "epoch": 0.8473960436011304,
      "grad_norm": 132.02662658691406,
      "learning_rate": 1.1968256404834243e-06,
      "loss": 3.9158,
      "step": 20990
    },
    {
      "epoch": 0.8474364150181671,
      "grad_norm": 193.862060546875,
      "learning_rate": 1.1962054191016492e-06,
      "loss": 5.966,
      "step": 20991
    },
    {
      "epoch": 0.8474767864352039,
      "grad_norm": 80.5673599243164,
      "learning_rate": 1.1955853482442026e-06,
      "loss": 3.216,
      "step": 20992
    },
    {
      "epoch": 0.8475171578522406,
      "grad_norm": 81.33606719970703,
      "learning_rate": 1.1949654279216894e-06,
      "loss": 4.7733,
      "step": 20993
    },
    {
      "epoch": 0.8475575292692773,
      "grad_norm": 120.46341705322266,
      "learning_rate": 1.1943456581447033e-06,
      "loss": 5.8797,
      "step": 20994
    },
    {
      "epoch": 0.8475979006863141,
      "grad_norm": 95.19979858398438,
      "learning_rate": 1.1937260389238437e-06,
      "loss": 4.5289,
      "step": 20995
    },
    {
      "epoch": 0.8476382721033509,
      "grad_norm": 94.31954956054688,
      "learning_rate": 1.1931065702697054e-06,
      "loss": 5.0199,
      "step": 20996
    },
    {
      "epoch": 0.8476786435203876,
      "grad_norm": 132.1021270751953,
      "learning_rate": 1.192487252192881e-06,
      "loss": 5.2312,
      "step": 20997
    },
    {
      "epoch": 0.8477190149374243,
      "grad_norm": 112.73086547851562,
      "learning_rate": 1.1918680847039553e-06,
      "loss": 4.7046,
      "step": 20998
    },
    {
      "epoch": 0.847759386354461,
      "grad_norm": 110.80297088623047,
      "learning_rate": 1.1912490678135168e-06,
      "loss": 4.54,
      "step": 20999
    },
    {
      "epoch": 0.8477997577714977,
      "grad_norm": 88.18987274169922,
      "learning_rate": 1.1906302015321502e-06,
      "loss": 5.1499,
      "step": 21000
    },
    {
      "epoch": 0.8478401291885345,
      "grad_norm": 76.33027648925781,
      "learning_rate": 1.190011485870437e-06,
      "loss": 4.3557,
      "step": 21001
    },
    {
      "epoch": 0.8478805006055713,
      "grad_norm": 161.79315185546875,
      "learning_rate": 1.189392920838952e-06,
      "loss": 5.0062,
      "step": 21002
    },
    {
      "epoch": 0.847920872022608,
      "grad_norm": 77.95028686523438,
      "learning_rate": 1.1887745064482748e-06,
      "loss": 4.9983,
      "step": 21003
    },
    {
      "epoch": 0.8479612434396447,
      "grad_norm": 145.50265502929688,
      "learning_rate": 1.1881562427089776e-06,
      "loss": 6.6682,
      "step": 21004
    },
    {
      "epoch": 0.8480016148566815,
      "grad_norm": 78.29804992675781,
      "learning_rate": 1.1875381296316312e-06,
      "loss": 2.5363,
      "step": 21005
    },
    {
      "epoch": 0.8480419862737182,
      "grad_norm": 114.44371795654297,
      "learning_rate": 1.186920167226805e-06,
      "loss": 4.4934,
      "step": 21006
    },
    {
      "epoch": 0.8480823576907549,
      "grad_norm": 120.18775177001953,
      "learning_rate": 1.1863023555050635e-06,
      "loss": 5.2991,
      "step": 21007
    },
    {
      "epoch": 0.8481227291077916,
      "grad_norm": 1506.7747802734375,
      "learning_rate": 1.1856846944769696e-06,
      "loss": 5.3364,
      "step": 21008
    },
    {
      "epoch": 0.8481631005248285,
      "grad_norm": 112.34136962890625,
      "learning_rate": 1.1850671841530881e-06,
      "loss": 6.4885,
      "step": 21009
    },
    {
      "epoch": 0.8482034719418652,
      "grad_norm": 111.52299499511719,
      "learning_rate": 1.1844498245439696e-06,
      "loss": 5.7903,
      "step": 21010
    },
    {
      "epoch": 0.8482438433589019,
      "grad_norm": 204.9984130859375,
      "learning_rate": 1.1838326156601743e-06,
      "loss": 4.4586,
      "step": 21011
    },
    {
      "epoch": 0.8482842147759386,
      "grad_norm": 76.78099822998047,
      "learning_rate": 1.183215557512254e-06,
      "loss": 5.1404,
      "step": 21012
    },
    {
      "epoch": 0.8483245861929753,
      "grad_norm": 90.15126037597656,
      "learning_rate": 1.182598650110761e-06,
      "loss": 5.4395,
      "step": 21013
    },
    {
      "epoch": 0.8483649576100121,
      "grad_norm": 103.61844635009766,
      "learning_rate": 1.1819818934662386e-06,
      "loss": 5.7811,
      "step": 21014
    },
    {
      "epoch": 0.8484053290270489,
      "grad_norm": 155.75540161132812,
      "learning_rate": 1.1813652875892344e-06,
      "loss": 5.7451,
      "step": 21015
    },
    {
      "epoch": 0.8484457004440856,
      "grad_norm": 116.92313385009766,
      "learning_rate": 1.1807488324902905e-06,
      "loss": 3.8688,
      "step": 21016
    },
    {
      "epoch": 0.8484860718611223,
      "grad_norm": 190.73622131347656,
      "learning_rate": 1.1801325281799502e-06,
      "loss": 6.772,
      "step": 21017
    },
    {
      "epoch": 0.8485264432781591,
      "grad_norm": 79.01681518554688,
      "learning_rate": 1.1795163746687454e-06,
      "loss": 4.5315,
      "step": 21018
    },
    {
      "epoch": 0.8485668146951958,
      "grad_norm": 146.51014709472656,
      "learning_rate": 1.1789003719672121e-06,
      "loss": 3.652,
      "step": 21019
    },
    {
      "epoch": 0.8486071861122325,
      "grad_norm": 107.15296173095703,
      "learning_rate": 1.178284520085885e-06,
      "loss": 5.694,
      "step": 21020
    },
    {
      "epoch": 0.8486475575292692,
      "grad_norm": 127.01394653320312,
      "learning_rate": 1.1776688190352936e-06,
      "loss": 3.8989,
      "step": 21021
    },
    {
      "epoch": 0.8486879289463061,
      "grad_norm": 136.83872985839844,
      "learning_rate": 1.1770532688259628e-06,
      "loss": 4.3571,
      "step": 21022
    },
    {
      "epoch": 0.8487283003633428,
      "grad_norm": 120.40540313720703,
      "learning_rate": 1.1764378694684176e-06,
      "loss": 6.3046,
      "step": 21023
    },
    {
      "epoch": 0.8487686717803795,
      "grad_norm": 94.6242904663086,
      "learning_rate": 1.1758226209731794e-06,
      "loss": 5.2727,
      "step": 21024
    },
    {
      "epoch": 0.8488090431974162,
      "grad_norm": 98.80216217041016,
      "learning_rate": 1.1752075233507698e-06,
      "loss": 5.9201,
      "step": 21025
    },
    {
      "epoch": 0.848849414614453,
      "grad_norm": 118.9493179321289,
      "learning_rate": 1.1745925766117027e-06,
      "loss": 3.6801,
      "step": 21026
    },
    {
      "epoch": 0.8488897860314897,
      "grad_norm": 79.4681625366211,
      "learning_rate": 1.1739777807664953e-06,
      "loss": 5.6274,
      "step": 21027
    },
    {
      "epoch": 0.8489301574485264,
      "grad_norm": 79.70944213867188,
      "learning_rate": 1.173363135825657e-06,
      "loss": 4.3537,
      "step": 21028
    },
    {
      "epoch": 0.8489705288655632,
      "grad_norm": 95.34112548828125,
      "learning_rate": 1.1727486417996992e-06,
      "loss": 4.877,
      "step": 21029
    },
    {
      "epoch": 0.8490109002826,
      "grad_norm": 120.75194549560547,
      "learning_rate": 1.1721342986991246e-06,
      "loss": 5.1895,
      "step": 21030
    },
    {
      "epoch": 0.8490512716996367,
      "grad_norm": 87.3873062133789,
      "learning_rate": 1.1715201065344384e-06,
      "loss": 3.9304,
      "step": 21031
    },
    {
      "epoch": 0.8490916431166734,
      "grad_norm": 117.71795654296875,
      "learning_rate": 1.170906065316143e-06,
      "loss": 6.0348,
      "step": 21032
    },
    {
      "epoch": 0.8491320145337101,
      "grad_norm": 97.88257598876953,
      "learning_rate": 1.1702921750547392e-06,
      "loss": 4.4014,
      "step": 21033
    },
    {
      "epoch": 0.8491723859507468,
      "grad_norm": 85.96202850341797,
      "learning_rate": 1.1696784357607171e-06,
      "loss": 4.1814,
      "step": 21034
    },
    {
      "epoch": 0.8492127573677836,
      "grad_norm": 109.6456527709961,
      "learning_rate": 1.1690648474445744e-06,
      "loss": 4.9165,
      "step": 21035
    },
    {
      "epoch": 0.8492531287848204,
      "grad_norm": 80.68937683105469,
      "learning_rate": 1.1684514101168e-06,
      "loss": 5.2437,
      "step": 21036
    },
    {
      "epoch": 0.8492935002018571,
      "grad_norm": 73.14678955078125,
      "learning_rate": 1.1678381237878867e-06,
      "loss": 5.4146,
      "step": 21037
    },
    {
      "epoch": 0.8493338716188938,
      "grad_norm": 101.97759246826172,
      "learning_rate": 1.167224988468314e-06,
      "loss": 4.6611,
      "step": 21038
    },
    {
      "epoch": 0.8493742430359306,
      "grad_norm": 117.62931060791016,
      "learning_rate": 1.166612004168568e-06,
      "loss": 5.6088,
      "step": 21039
    },
    {
      "epoch": 0.8494146144529673,
      "grad_norm": 152.22715759277344,
      "learning_rate": 1.1659991708991292e-06,
      "loss": 5.8572,
      "step": 21040
    },
    {
      "epoch": 0.849454985870004,
      "grad_norm": 104.13101959228516,
      "learning_rate": 1.165386488670478e-06,
      "loss": 5.1498,
      "step": 21041
    },
    {
      "epoch": 0.8494953572870407,
      "grad_norm": 110.09239959716797,
      "learning_rate": 1.1647739574930862e-06,
      "loss": 5.4952,
      "step": 21042
    },
    {
      "epoch": 0.8495357287040776,
      "grad_norm": 88.752197265625,
      "learning_rate": 1.1641615773774274e-06,
      "loss": 4.3802,
      "step": 21043
    },
    {
      "epoch": 0.8495761001211143,
      "grad_norm": 167.317138671875,
      "learning_rate": 1.1635493483339722e-06,
      "loss": 6.4148,
      "step": 21044
    },
    {
      "epoch": 0.849616471538151,
      "grad_norm": 71.73871612548828,
      "learning_rate": 1.162937270373188e-06,
      "loss": 4.0209,
      "step": 21045
    },
    {
      "epoch": 0.8496568429551877,
      "grad_norm": 108.88036346435547,
      "learning_rate": 1.1623253435055415e-06,
      "loss": 4.8874,
      "step": 21046
    },
    {
      "epoch": 0.8496972143722245,
      "grad_norm": 83.77503967285156,
      "learning_rate": 1.1617135677414937e-06,
      "loss": 5.4633,
      "step": 21047
    },
    {
      "epoch": 0.8497375857892612,
      "grad_norm": 100.68179321289062,
      "learning_rate": 1.161101943091505e-06,
      "loss": 5.0781,
      "step": 21048
    },
    {
      "epoch": 0.8497779572062979,
      "grad_norm": 70.64859008789062,
      "learning_rate": 1.1604904695660346e-06,
      "loss": 2.7346,
      "step": 21049
    },
    {
      "epoch": 0.8498183286233347,
      "grad_norm": 96.28355407714844,
      "learning_rate": 1.1598791471755344e-06,
      "loss": 5.5536,
      "step": 21050
    },
    {
      "epoch": 0.8498587000403715,
      "grad_norm": 81.12089538574219,
      "learning_rate": 1.159267975930457e-06,
      "loss": 4.3813,
      "step": 21051
    },
    {
      "epoch": 0.8498990714574082,
      "grad_norm": 76.58889770507812,
      "learning_rate": 1.158656955841252e-06,
      "loss": 4.4571,
      "step": 21052
    },
    {
      "epoch": 0.8499394428744449,
      "grad_norm": 105.77559661865234,
      "learning_rate": 1.1580460869183697e-06,
      "loss": 6.3319,
      "step": 21053
    },
    {
      "epoch": 0.8499798142914816,
      "grad_norm": 121.54707336425781,
      "learning_rate": 1.1574353691722494e-06,
      "loss": 4.8102,
      "step": 21054
    },
    {
      "epoch": 0.8500201857085183,
      "grad_norm": 80.58353424072266,
      "learning_rate": 1.1568248026133366e-06,
      "loss": 5.0803,
      "step": 21055
    },
    {
      "epoch": 0.8500605571255551,
      "grad_norm": 103.7407455444336,
      "learning_rate": 1.1562143872520681e-06,
      "loss": 4.9152,
      "step": 21056
    },
    {
      "epoch": 0.8501009285425919,
      "grad_norm": 97.99169921875,
      "learning_rate": 1.1556041230988846e-06,
      "loss": 6.0205,
      "step": 21057
    },
    {
      "epoch": 0.8501412999596286,
      "grad_norm": 116.66212463378906,
      "learning_rate": 1.1549940101642154e-06,
      "loss": 5.1233,
      "step": 21058
    },
    {
      "epoch": 0.8501816713766653,
      "grad_norm": 72.88399505615234,
      "learning_rate": 1.1543840484584935e-06,
      "loss": 4.745,
      "step": 21059
    },
    {
      "epoch": 0.8502220427937021,
      "grad_norm": 146.57159423828125,
      "learning_rate": 1.153774237992148e-06,
      "loss": 6.5175,
      "step": 21060
    },
    {
      "epoch": 0.8502624142107388,
      "grad_norm": 95.41424560546875,
      "learning_rate": 1.1531645787756085e-06,
      "loss": 3.7012,
      "step": 21061
    },
    {
      "epoch": 0.8503027856277755,
      "grad_norm": 60.46418762207031,
      "learning_rate": 1.1525550708192933e-06,
      "loss": 4.7093,
      "step": 21062
    },
    {
      "epoch": 0.8503431570448122,
      "grad_norm": 76.74882507324219,
      "learning_rate": 1.1519457141336266e-06,
      "loss": 4.6455,
      "step": 21063
    },
    {
      "epoch": 0.8503835284618491,
      "grad_norm": 108.40746307373047,
      "learning_rate": 1.1513365087290262e-06,
      "loss": 6.6163,
      "step": 21064
    },
    {
      "epoch": 0.8504238998788858,
      "grad_norm": 161.85882568359375,
      "learning_rate": 1.1507274546159087e-06,
      "loss": 5.0598,
      "step": 21065
    },
    {
      "epoch": 0.8504642712959225,
      "grad_norm": 89.69302368164062,
      "learning_rate": 1.1501185518046875e-06,
      "loss": 5.267,
      "step": 21066
    },
    {
      "epoch": 0.8505046427129592,
      "grad_norm": 132.01890563964844,
      "learning_rate": 1.1495098003057725e-06,
      "loss": 7.1789,
      "step": 21067
    },
    {
      "epoch": 0.850545014129996,
      "grad_norm": 89.25529479980469,
      "learning_rate": 1.148901200129573e-06,
      "loss": 5.2431,
      "step": 21068
    },
    {
      "epoch": 0.8505853855470327,
      "grad_norm": 88.7705078125,
      "learning_rate": 1.1482927512864961e-06,
      "loss": 4.7074,
      "step": 21069
    },
    {
      "epoch": 0.8506257569640694,
      "grad_norm": 100.86001586914062,
      "learning_rate": 1.1476844537869403e-06,
      "loss": 5.6079,
      "step": 21070
    },
    {
      "epoch": 0.8506661283811062,
      "grad_norm": 81.57772827148438,
      "learning_rate": 1.1470763076413093e-06,
      "loss": 5.8645,
      "step": 21071
    },
    {
      "epoch": 0.850706499798143,
      "grad_norm": 84.29741668701172,
      "learning_rate": 1.1464683128599995e-06,
      "loss": 5.7353,
      "step": 21072
    },
    {
      "epoch": 0.8507468712151797,
      "grad_norm": 94.74272155761719,
      "learning_rate": 1.1458604694534104e-06,
      "loss": 5.9672,
      "step": 21073
    },
    {
      "epoch": 0.8507872426322164,
      "grad_norm": 90.88701629638672,
      "learning_rate": 1.1452527774319277e-06,
      "loss": 4.9451,
      "step": 21074
    },
    {
      "epoch": 0.8508276140492531,
      "grad_norm": 105.1067886352539,
      "learning_rate": 1.1446452368059468e-06,
      "loss": 4.5936,
      "step": 21075
    },
    {
      "epoch": 0.8508679854662898,
      "grad_norm": 170.73512268066406,
      "learning_rate": 1.1440378475858526e-06,
      "loss": 5.0911,
      "step": 21076
    },
    {
      "epoch": 0.8509083568833266,
      "grad_norm": 83.32601165771484,
      "learning_rate": 1.1434306097820337e-06,
      "loss": 4.545,
      "step": 21077
    },
    {
      "epoch": 0.8509487283003634,
      "grad_norm": 86.9651870727539,
      "learning_rate": 1.142823523404868e-06,
      "loss": 6.0076,
      "step": 21078
    },
    {
      "epoch": 0.8509890997174001,
      "grad_norm": 253.69847106933594,
      "learning_rate": 1.1422165884647364e-06,
      "loss": 5.6804,
      "step": 21079
    },
    {
      "epoch": 0.8510294711344368,
      "grad_norm": 121.96200561523438,
      "learning_rate": 1.141609804972017e-06,
      "loss": 5.5491,
      "step": 21080
    },
    {
      "epoch": 0.8510698425514736,
      "grad_norm": 112.22386169433594,
      "learning_rate": 1.141003172937085e-06,
      "loss": 3.5402,
      "step": 21081
    },
    {
      "epoch": 0.8511102139685103,
      "grad_norm": 110.97826385498047,
      "learning_rate": 1.1403966923703114e-06,
      "loss": 4.2981,
      "step": 21082
    },
    {
      "epoch": 0.851150585385547,
      "grad_norm": 96.40363311767578,
      "learning_rate": 1.1397903632820638e-06,
      "loss": 5.0457,
      "step": 21083
    },
    {
      "epoch": 0.8511909568025837,
      "grad_norm": 63.79092025756836,
      "learning_rate": 1.1391841856827123e-06,
      "loss": 5.2628,
      "step": 21084
    },
    {
      "epoch": 0.8512313282196206,
      "grad_norm": 88.20114135742188,
      "learning_rate": 1.1385781595826196e-06,
      "loss": 5.2835,
      "step": 21085
    },
    {
      "epoch": 0.8512716996366573,
      "grad_norm": 81.9908447265625,
      "learning_rate": 1.137972284992147e-06,
      "loss": 6.0661,
      "step": 21086
    },
    {
      "epoch": 0.851312071053694,
      "grad_norm": 82.43226623535156,
      "learning_rate": 1.1373665619216545e-06,
      "loss": 5.2114,
      "step": 21087
    },
    {
      "epoch": 0.8513524424707307,
      "grad_norm": 83.7392578125,
      "learning_rate": 1.136760990381498e-06,
      "loss": 4.4007,
      "step": 21088
    },
    {
      "epoch": 0.8513928138877674,
      "grad_norm": 122.08792114257812,
      "learning_rate": 1.1361555703820326e-06,
      "loss": 3.8646,
      "step": 21089
    },
    {
      "epoch": 0.8514331853048042,
      "grad_norm": 68.06731414794922,
      "learning_rate": 1.1355503019336078e-06,
      "loss": 4.5968,
      "step": 21090
    },
    {
      "epoch": 0.8514735567218409,
      "grad_norm": 357.0022888183594,
      "learning_rate": 1.134945185046572e-06,
      "loss": 2.8032,
      "step": 21091
    },
    {
      "epoch": 0.8515139281388777,
      "grad_norm": 99.56805419921875,
      "learning_rate": 1.1343402197312715e-06,
      "loss": 5.2545,
      "step": 21092
    },
    {
      "epoch": 0.8515542995559144,
      "grad_norm": 85.5050277709961,
      "learning_rate": 1.1337354059980533e-06,
      "loss": 3.9619,
      "step": 21093
    },
    {
      "epoch": 0.8515946709729512,
      "grad_norm": 103.30941009521484,
      "learning_rate": 1.1331307438572537e-06,
      "loss": 3.9463,
      "step": 21094
    },
    {
      "epoch": 0.8516350423899879,
      "grad_norm": 471.8302001953125,
      "learning_rate": 1.1325262333192121e-06,
      "loss": 3.4523,
      "step": 21095
    },
    {
      "epoch": 0.8516754138070246,
      "grad_norm": 76.52373504638672,
      "learning_rate": 1.131921874394265e-06,
      "loss": 3.2028,
      "step": 21096
    },
    {
      "epoch": 0.8517157852240613,
      "grad_norm": 265.2095947265625,
      "learning_rate": 1.1313176670927473e-06,
      "loss": 5.5872,
      "step": 21097
    },
    {
      "epoch": 0.851756156641098,
      "grad_norm": 72.49463653564453,
      "learning_rate": 1.1307136114249862e-06,
      "loss": 5.0576,
      "step": 21098
    },
    {
      "epoch": 0.8517965280581349,
      "grad_norm": 85.83214569091797,
      "learning_rate": 1.1301097074013102e-06,
      "loss": 2.7173,
      "step": 21099
    },
    {
      "epoch": 0.8518368994751716,
      "grad_norm": 405.9644775390625,
      "learning_rate": 1.1295059550320465e-06,
      "loss": 4.45,
      "step": 21100
    },
    {
      "epoch": 0.8518772708922083,
      "grad_norm": 174.73013305664062,
      "learning_rate": 1.128902354327518e-06,
      "loss": 5.5109,
      "step": 21101
    },
    {
      "epoch": 0.851917642309245,
      "grad_norm": 130.76223754882812,
      "learning_rate": 1.1282989052980419e-06,
      "loss": 6.0736,
      "step": 21102
    },
    {
      "epoch": 0.8519580137262818,
      "grad_norm": 124.9169692993164,
      "learning_rate": 1.1276956079539381e-06,
      "loss": 5.5945,
      "step": 21103
    },
    {
      "epoch": 0.8519983851433185,
      "grad_norm": 149.1466522216797,
      "learning_rate": 1.12709246230552e-06,
      "loss": 5.2548,
      "step": 21104
    },
    {
      "epoch": 0.8520387565603552,
      "grad_norm": 123.19134521484375,
      "learning_rate": 1.1264894683631023e-06,
      "loss": 4.5946,
      "step": 21105
    },
    {
      "epoch": 0.852079127977392,
      "grad_norm": 107.2651138305664,
      "learning_rate": 1.1258866261369928e-06,
      "loss": 1.7957,
      "step": 21106
    },
    {
      "epoch": 0.8521194993944288,
      "grad_norm": 110.57307434082031,
      "learning_rate": 1.1252839356375e-06,
      "loss": 3.8198,
      "step": 21107
    },
    {
      "epoch": 0.8521598708114655,
      "grad_norm": 104.0322494506836,
      "learning_rate": 1.1246813968749281e-06,
      "loss": 3.7961,
      "step": 21108
    },
    {
      "epoch": 0.8522002422285022,
      "grad_norm": 74.1217269897461,
      "learning_rate": 1.12407900985958e-06,
      "loss": 4.4035,
      "step": 21109
    },
    {
      "epoch": 0.8522406136455389,
      "grad_norm": 83.76164245605469,
      "learning_rate": 1.1234767746017527e-06,
      "loss": 4.7936,
      "step": 21110
    },
    {
      "epoch": 0.8522809850625757,
      "grad_norm": 75.9721450805664,
      "learning_rate": 1.1228746911117449e-06,
      "loss": 5.1714,
      "step": 21111
    },
    {
      "epoch": 0.8523213564796124,
      "grad_norm": 146.19337463378906,
      "learning_rate": 1.1222727593998495e-06,
      "loss": 7.0197,
      "step": 21112
    },
    {
      "epoch": 0.8523617278966492,
      "grad_norm": 88.8703842163086,
      "learning_rate": 1.1216709794763614e-06,
      "loss": 6.7964,
      "step": 21113
    },
    {
      "epoch": 0.8524020993136859,
      "grad_norm": 114.38135528564453,
      "learning_rate": 1.1210693513515647e-06,
      "loss": 5.7009,
      "step": 21114
    },
    {
      "epoch": 0.8524424707307227,
      "grad_norm": 98.7558822631836,
      "learning_rate": 1.120467875035749e-06,
      "loss": 5.269,
      "step": 21115
    },
    {
      "epoch": 0.8524828421477594,
      "grad_norm": 77.94163513183594,
      "learning_rate": 1.1198665505391958e-06,
      "loss": 4.4736,
      "step": 21116
    },
    {
      "epoch": 0.8525232135647961,
      "grad_norm": 75.3013687133789,
      "learning_rate": 1.1192653778721918e-06,
      "loss": 4.3065,
      "step": 21117
    },
    {
      "epoch": 0.8525635849818328,
      "grad_norm": 82.26536560058594,
      "learning_rate": 1.1186643570450096e-06,
      "loss": 5.1284,
      "step": 21118
    },
    {
      "epoch": 0.8526039563988697,
      "grad_norm": 78.75489807128906,
      "learning_rate": 1.1180634880679264e-06,
      "loss": 3.3785,
      "step": 21119
    },
    {
      "epoch": 0.8526443278159064,
      "grad_norm": 121.44536590576172,
      "learning_rate": 1.1174627709512175e-06,
      "loss": 5.2944,
      "step": 21120
    },
    {
      "epoch": 0.8526846992329431,
      "grad_norm": 101.21125030517578,
      "learning_rate": 1.1168622057051548e-06,
      "loss": 4.6202,
      "step": 21121
    },
    {
      "epoch": 0.8527250706499798,
      "grad_norm": 95.63616943359375,
      "learning_rate": 1.1162617923400032e-06,
      "loss": 4.0597,
      "step": 21122
    },
    {
      "epoch": 0.8527654420670165,
      "grad_norm": 97.75437927246094,
      "learning_rate": 1.1156615308660291e-06,
      "loss": 5.6713,
      "step": 21123
    },
    {
      "epoch": 0.8528058134840533,
      "grad_norm": 81.36334991455078,
      "learning_rate": 1.1150614212934974e-06,
      "loss": 3.819,
      "step": 21124
    },
    {
      "epoch": 0.85284618490109,
      "grad_norm": 109.26480102539062,
      "learning_rate": 1.1144614636326666e-06,
      "loss": 4.2392,
      "step": 21125
    },
    {
      "epoch": 0.8528865563181268,
      "grad_norm": 79.50385284423828,
      "learning_rate": 1.1138616578937965e-06,
      "loss": 5.0901,
      "step": 21126
    },
    {
      "epoch": 0.8529269277351635,
      "grad_norm": 79.25601959228516,
      "learning_rate": 1.11326200408714e-06,
      "loss": 3.3125,
      "step": 21127
    },
    {
      "epoch": 0.8529672991522003,
      "grad_norm": 106.85610961914062,
      "learning_rate": 1.1126625022229531e-06,
      "loss": 4.9167,
      "step": 21128
    },
    {
      "epoch": 0.853007670569237,
      "grad_norm": 103.76776885986328,
      "learning_rate": 1.1120631523114844e-06,
      "loss": 3.8965,
      "step": 21129
    },
    {
      "epoch": 0.8530480419862737,
      "grad_norm": 97.90851593017578,
      "learning_rate": 1.11146395436298e-06,
      "loss": 5.9437,
      "step": 21130
    },
    {
      "epoch": 0.8530884134033104,
      "grad_norm": 91.85980224609375,
      "learning_rate": 1.1108649083876855e-06,
      "loss": 5.6413,
      "step": 21131
    },
    {
      "epoch": 0.8531287848203472,
      "grad_norm": 113.14212036132812,
      "learning_rate": 1.1102660143958432e-06,
      "loss": 4.6763,
      "step": 21132
    },
    {
      "epoch": 0.853169156237384,
      "grad_norm": 131.30812072753906,
      "learning_rate": 1.1096672723976953e-06,
      "loss": 5.0886,
      "step": 21133
    },
    {
      "epoch": 0.8532095276544207,
      "grad_norm": 81.78485870361328,
      "learning_rate": 1.1090686824034736e-06,
      "loss": 5.5218,
      "step": 21134
    },
    {
      "epoch": 0.8532498990714574,
      "grad_norm": 78.11510467529297,
      "learning_rate": 1.1084702444234175e-06,
      "loss": 5.3926,
      "step": 21135
    },
    {
      "epoch": 0.8532902704884942,
      "grad_norm": 133.8459930419922,
      "learning_rate": 1.1078719584677555e-06,
      "loss": 5.0492,
      "step": 21136
    },
    {
      "epoch": 0.8533306419055309,
      "grad_norm": 107.12168884277344,
      "learning_rate": 1.1072738245467207e-06,
      "loss": 5.153,
      "step": 21137
    },
    {
      "epoch": 0.8533710133225676,
      "grad_norm": 5451.3955078125,
      "learning_rate": 1.106675842670536e-06,
      "loss": 5.1757,
      "step": 21138
    },
    {
      "epoch": 0.8534113847396043,
      "grad_norm": 75.75413513183594,
      "learning_rate": 1.1060780128494264e-06,
      "loss": 4.2972,
      "step": 21139
    },
    {
      "epoch": 0.8534517561566412,
      "grad_norm": 76.75553894042969,
      "learning_rate": 1.1054803350936138e-06,
      "loss": 4.5084,
      "step": 21140
    },
    {
      "epoch": 0.8534921275736779,
      "grad_norm": 114.76996612548828,
      "learning_rate": 1.104882809413318e-06,
      "loss": 5.8087,
      "step": 21141
    },
    {
      "epoch": 0.8535324989907146,
      "grad_norm": 119.05671691894531,
      "learning_rate": 1.104285435818756e-06,
      "loss": 3.8376,
      "step": 21142
    },
    {
      "epoch": 0.8535728704077513,
      "grad_norm": 66.9393310546875,
      "learning_rate": 1.1036882143201388e-06,
      "loss": 5.1674,
      "step": 21143
    },
    {
      "epoch": 0.853613241824788,
      "grad_norm": 85.11837768554688,
      "learning_rate": 1.1030911449276782e-06,
      "loss": 5.0109,
      "step": 21144
    },
    {
      "epoch": 0.8536536132418248,
      "grad_norm": 95.11493682861328,
      "learning_rate": 1.1024942276515839e-06,
      "loss": 3.591,
      "step": 21145
    },
    {
      "epoch": 0.8536939846588615,
      "grad_norm": 103.91835021972656,
      "learning_rate": 1.1018974625020596e-06,
      "loss": 3.8942,
      "step": 21146
    },
    {
      "epoch": 0.8537343560758983,
      "grad_norm": 114.26899719238281,
      "learning_rate": 1.1013008494893118e-06,
      "loss": 5.7768,
      "step": 21147
    },
    {
      "epoch": 0.853774727492935,
      "grad_norm": 105.10625457763672,
      "learning_rate": 1.100704388623538e-06,
      "loss": 4.7997,
      "step": 21148
    },
    {
      "epoch": 0.8538150989099718,
      "grad_norm": 84.3094253540039,
      "learning_rate": 1.100108079914941e-06,
      "loss": 5.1667,
      "step": 21149
    },
    {
      "epoch": 0.8538554703270085,
      "grad_norm": 97.90304565429688,
      "learning_rate": 1.0995119233737107e-06,
      "loss": 3.7859,
      "step": 21150
    },
    {
      "epoch": 0.8538958417440452,
      "grad_norm": 108.29054260253906,
      "learning_rate": 1.0989159190100429e-06,
      "loss": 5.0477,
      "step": 21151
    },
    {
      "epoch": 0.8539362131610819,
      "grad_norm": 83.93783569335938,
      "learning_rate": 1.0983200668341266e-06,
      "loss": 4.9466,
      "step": 21152
    },
    {
      "epoch": 0.8539765845781186,
      "grad_norm": 124.70082092285156,
      "learning_rate": 1.097724366856152e-06,
      "loss": 4.7503,
      "step": 21153
    },
    {
      "epoch": 0.8540169559951555,
      "grad_norm": 84.95504760742188,
      "learning_rate": 1.0971288190863005e-06,
      "loss": 5.0057,
      "step": 21154
    },
    {
      "epoch": 0.8540573274121922,
      "grad_norm": 265.12994384765625,
      "learning_rate": 1.096533423534758e-06,
      "loss": 4.1145,
      "step": 21155
    },
    {
      "epoch": 0.8540976988292289,
      "grad_norm": 118.8050537109375,
      "learning_rate": 1.0959381802117008e-06,
      "loss": 2.8933,
      "step": 21156
    },
    {
      "epoch": 0.8541380702462656,
      "grad_norm": 99.70855712890625,
      "learning_rate": 1.095343089127312e-06,
      "loss": 5.9267,
      "step": 21157
    },
    {
      "epoch": 0.8541784416633024,
      "grad_norm": 79.03816986083984,
      "learning_rate": 1.094748150291759e-06,
      "loss": 4.3104,
      "step": 21158
    },
    {
      "epoch": 0.8542188130803391,
      "grad_norm": 81.83454132080078,
      "learning_rate": 1.094153363715218e-06,
      "loss": 4.3776,
      "step": 21159
    },
    {
      "epoch": 0.8542591844973758,
      "grad_norm": 87.24610137939453,
      "learning_rate": 1.0935587294078588e-06,
      "loss": 4.3788,
      "step": 21160
    },
    {
      "epoch": 0.8542995559144126,
      "grad_norm": 134.81494140625,
      "learning_rate": 1.0929642473798464e-06,
      "loss": 5.5549,
      "step": 21161
    },
    {
      "epoch": 0.8543399273314494,
      "grad_norm": 129.39022827148438,
      "learning_rate": 1.092369917641345e-06,
      "loss": 4.8649,
      "step": 21162
    },
    {
      "epoch": 0.8543802987484861,
      "grad_norm": 96.1605224609375,
      "learning_rate": 1.0917757402025185e-06,
      "loss": 3.805,
      "step": 21163
    },
    {
      "epoch": 0.8544206701655228,
      "grad_norm": 101.92582702636719,
      "learning_rate": 1.0911817150735249e-06,
      "loss": 4.5563,
      "step": 21164
    },
    {
      "epoch": 0.8544610415825595,
      "grad_norm": 107.57111358642578,
      "learning_rate": 1.0905878422645211e-06,
      "loss": 4.9235,
      "step": 21165
    },
    {
      "epoch": 0.8545014129995963,
      "grad_norm": 114.30657196044922,
      "learning_rate": 1.0899941217856592e-06,
      "loss": 6.1971,
      "step": 21166
    },
    {
      "epoch": 0.854541784416633,
      "grad_norm": 4780.52685546875,
      "learning_rate": 1.089400553647091e-06,
      "loss": 4.6801,
      "step": 21167
    },
    {
      "epoch": 0.8545821558336698,
      "grad_norm": 92.70185852050781,
      "learning_rate": 1.088807137858966e-06,
      "loss": 4.8417,
      "step": 21168
    },
    {
      "epoch": 0.8546225272507065,
      "grad_norm": 92.29887390136719,
      "learning_rate": 1.0882138744314309e-06,
      "loss": 6.221,
      "step": 21169
    },
    {
      "epoch": 0.8546628986677433,
      "grad_norm": 117.50833129882812,
      "learning_rate": 1.0876207633746272e-06,
      "loss": 4.3767,
      "step": 21170
    },
    {
      "epoch": 0.85470327008478,
      "grad_norm": 93.22380828857422,
      "learning_rate": 1.0870278046986959e-06,
      "loss": 3.8823,
      "step": 21171
    },
    {
      "epoch": 0.8547436415018167,
      "grad_norm": 143.04356384277344,
      "learning_rate": 1.0864349984137756e-06,
      "loss": 3.7485,
      "step": 21172
    },
    {
      "epoch": 0.8547840129188534,
      "grad_norm": 111.62427520751953,
      "learning_rate": 1.0858423445300047e-06,
      "loss": 4.7515,
      "step": 21173
    },
    {
      "epoch": 0.8548243843358901,
      "grad_norm": 101.177490234375,
      "learning_rate": 1.085249843057512e-06,
      "loss": 3.5651,
      "step": 21174
    },
    {
      "epoch": 0.854864755752927,
      "grad_norm": 92.7442398071289,
      "learning_rate": 1.0846574940064291e-06,
      "loss": 4.9492,
      "step": 21175
    },
    {
      "epoch": 0.8549051271699637,
      "grad_norm": 92.52972412109375,
      "learning_rate": 1.0840652973868837e-06,
      "loss": 5.3284,
      "step": 21176
    },
    {
      "epoch": 0.8549454985870004,
      "grad_norm": 72.21139526367188,
      "learning_rate": 1.0834732532090043e-06,
      "loss": 4.6108,
      "step": 21177
    },
    {
      "epoch": 0.8549858700040371,
      "grad_norm": 126.96884155273438,
      "learning_rate": 1.0828813614829082e-06,
      "loss": 5.8952,
      "step": 21178
    },
    {
      "epoch": 0.8550262414210739,
      "grad_norm": 1374.677978515625,
      "learning_rate": 1.0822896222187186e-06,
      "loss": 4.4485,
      "step": 21179
    },
    {
      "epoch": 0.8550666128381106,
      "grad_norm": 91.8199234008789,
      "learning_rate": 1.0816980354265505e-06,
      "loss": 5.7253,
      "step": 21180
    },
    {
      "epoch": 0.8551069842551473,
      "grad_norm": 176.36375427246094,
      "learning_rate": 1.0811066011165217e-06,
      "loss": 5.6402,
      "step": 21181
    },
    {
      "epoch": 0.8551473556721841,
      "grad_norm": 108.55168914794922,
      "learning_rate": 1.0805153192987429e-06,
      "loss": 4.0747,
      "step": 21182
    },
    {
      "epoch": 0.8551877270892209,
      "grad_norm": 101.97166442871094,
      "learning_rate": 1.0799241899833225e-06,
      "loss": 5.3228,
      "step": 21183
    },
    {
      "epoch": 0.8552280985062576,
      "grad_norm": 130.1104736328125,
      "learning_rate": 1.0793332131803692e-06,
      "loss": 4.7141,
      "step": 21184
    },
    {
      "epoch": 0.8552684699232943,
      "grad_norm": 126.14445495605469,
      "learning_rate": 1.0787423888999892e-06,
      "loss": 5.0698,
      "step": 21185
    },
    {
      "epoch": 0.855308841340331,
      "grad_norm": 109.18312072753906,
      "learning_rate": 1.0781517171522794e-06,
      "loss": 5.0109,
      "step": 21186
    },
    {
      "epoch": 0.8553492127573677,
      "grad_norm": 106.55010986328125,
      "learning_rate": 1.077561197947341e-06,
      "loss": 5.1061,
      "step": 21187
    },
    {
      "epoch": 0.8553895841744045,
      "grad_norm": 196.97238159179688,
      "learning_rate": 1.076970831295271e-06,
      "loss": 6.593,
      "step": 21188
    },
    {
      "epoch": 0.8554299555914413,
      "grad_norm": 116.834716796875,
      "learning_rate": 1.0763806172061654e-06,
      "loss": 4.7714,
      "step": 21189
    },
    {
      "epoch": 0.855470327008478,
      "grad_norm": 88.02447509765625,
      "learning_rate": 1.0757905556901104e-06,
      "loss": 3.3378,
      "step": 21190
    },
    {
      "epoch": 0.8555106984255147,
      "grad_norm": 102.47309112548828,
      "learning_rate": 1.0752006467571974e-06,
      "loss": 5.0943,
      "step": 21191
    },
    {
      "epoch": 0.8555510698425515,
      "grad_norm": 87.89483642578125,
      "learning_rate": 1.0746108904175122e-06,
      "loss": 4.8064,
      "step": 21192
    },
    {
      "epoch": 0.8555914412595882,
      "grad_norm": 73.77296447753906,
      "learning_rate": 1.0740212866811406e-06,
      "loss": 2.22,
      "step": 21193
    },
    {
      "epoch": 0.8556318126766249,
      "grad_norm": 104.76457214355469,
      "learning_rate": 1.0734318355581597e-06,
      "loss": 5.8667,
      "step": 21194
    },
    {
      "epoch": 0.8556721840936616,
      "grad_norm": 128.33950805664062,
      "learning_rate": 1.0728425370586494e-06,
      "loss": 4.3096,
      "step": 21195
    },
    {
      "epoch": 0.8557125555106985,
      "grad_norm": 109.15348052978516,
      "learning_rate": 1.0722533911926857e-06,
      "loss": 5.5985,
      "step": 21196
    },
    {
      "epoch": 0.8557529269277352,
      "grad_norm": 115.68878936767578,
      "learning_rate": 1.0716643979703423e-06,
      "loss": 3.2878,
      "step": 21197
    },
    {
      "epoch": 0.8557932983447719,
      "grad_norm": 75.12602996826172,
      "learning_rate": 1.0710755574016873e-06,
      "loss": 4.2035,
      "step": 21198
    },
    {
      "epoch": 0.8558336697618086,
      "grad_norm": 77.99374389648438,
      "learning_rate": 1.0704868694967896e-06,
      "loss": 4.062,
      "step": 21199
    },
    {
      "epoch": 0.8558740411788454,
      "grad_norm": 81.0879135131836,
      "learning_rate": 1.0698983342657142e-06,
      "loss": 4.5607,
      "step": 21200
    },
    {
      "epoch": 0.8559144125958821,
      "grad_norm": 79.98387145996094,
      "learning_rate": 1.0693099517185235e-06,
      "loss": 4.6174,
      "step": 21201
    },
    {
      "epoch": 0.8559547840129188,
      "grad_norm": 4817.82568359375,
      "learning_rate": 1.068721721865279e-06,
      "loss": 2.8199,
      "step": 21202
    },
    {
      "epoch": 0.8559951554299556,
      "grad_norm": 148.10043334960938,
      "learning_rate": 1.0681336447160362e-06,
      "loss": 4.5153,
      "step": 21203
    },
    {
      "epoch": 0.8560355268469924,
      "grad_norm": 95.49690246582031,
      "learning_rate": 1.0675457202808515e-06,
      "loss": 5.3577,
      "step": 21204
    },
    {
      "epoch": 0.8560758982640291,
      "grad_norm": 94.06682586669922,
      "learning_rate": 1.066957948569778e-06,
      "loss": 3.059,
      "step": 21205
    },
    {
      "epoch": 0.8561162696810658,
      "grad_norm": 105.75885009765625,
      "learning_rate": 1.0663703295928607e-06,
      "loss": 3.5726,
      "step": 21206
    },
    {
      "epoch": 0.8561566410981025,
      "grad_norm": 96.61134338378906,
      "learning_rate": 1.0657828633601497e-06,
      "loss": 5.8529,
      "step": 21207
    },
    {
      "epoch": 0.8561970125151392,
      "grad_norm": 82.950927734375,
      "learning_rate": 1.0651955498816891e-06,
      "loss": 4.7495,
      "step": 21208
    },
    {
      "epoch": 0.856237383932176,
      "grad_norm": 88.77803039550781,
      "learning_rate": 1.064608389167522e-06,
      "loss": 6.051,
      "step": 21209
    },
    {
      "epoch": 0.8562777553492128,
      "grad_norm": 98.35498809814453,
      "learning_rate": 1.0640213812276834e-06,
      "loss": 5.1914,
      "step": 21210
    },
    {
      "epoch": 0.8563181267662495,
      "grad_norm": 118.18077850341797,
      "learning_rate": 1.0634345260722135e-06,
      "loss": 4.607,
      "step": 21211
    },
    {
      "epoch": 0.8563584981832862,
      "grad_norm": 93.60338592529297,
      "learning_rate": 1.0628478237111438e-06,
      "loss": 5.9114,
      "step": 21212
    },
    {
      "epoch": 0.856398869600323,
      "grad_norm": 240.07606506347656,
      "learning_rate": 1.062261274154509e-06,
      "loss": 5.5038,
      "step": 21213
    },
    {
      "epoch": 0.8564392410173597,
      "grad_norm": 103.41666412353516,
      "learning_rate": 1.061674877412333e-06,
      "loss": 4.6276,
      "step": 21214
    },
    {
      "epoch": 0.8564796124343964,
      "grad_norm": 90.32757568359375,
      "learning_rate": 1.0610886334946436e-06,
      "loss": 5.5907,
      "step": 21215
    },
    {
      "epoch": 0.8565199838514331,
      "grad_norm": 109.38014221191406,
      "learning_rate": 1.0605025424114657e-06,
      "loss": 2.9655,
      "step": 21216
    },
    {
      "epoch": 0.85656035526847,
      "grad_norm": 63.56283950805664,
      "learning_rate": 1.0599166041728204e-06,
      "loss": 2.9755,
      "step": 21217
    },
    {
      "epoch": 0.8566007266855067,
      "grad_norm": 94.61552429199219,
      "learning_rate": 1.059330818788723e-06,
      "loss": 6.2952,
      "step": 21218
    },
    {
      "epoch": 0.8566410981025434,
      "grad_norm": 105.95364379882812,
      "learning_rate": 1.05874518626919e-06,
      "loss": 4.9001,
      "step": 21219
    },
    {
      "epoch": 0.8566814695195801,
      "grad_norm": 104.85182189941406,
      "learning_rate": 1.0581597066242356e-06,
      "loss": 6.2184,
      "step": 21220
    },
    {
      "epoch": 0.8567218409366169,
      "grad_norm": 107.8985595703125,
      "learning_rate": 1.0575743798638694e-06,
      "loss": 4.5764,
      "step": 21221
    },
    {
      "epoch": 0.8567622123536536,
      "grad_norm": 91.66808319091797,
      "learning_rate": 1.056989205998099e-06,
      "loss": 5.4946,
      "step": 21222
    },
    {
      "epoch": 0.8568025837706903,
      "grad_norm": 108.56015014648438,
      "learning_rate": 1.05640418503693e-06,
      "loss": 5.2379,
      "step": 21223
    },
    {
      "epoch": 0.8568429551877271,
      "grad_norm": 101.71385955810547,
      "learning_rate": 1.0558193169903652e-06,
      "loss": 5.7346,
      "step": 21224
    },
    {
      "epoch": 0.8568833266047639,
      "grad_norm": 95.93002319335938,
      "learning_rate": 1.0552346018684046e-06,
      "loss": 5.1513,
      "step": 21225
    },
    {
      "epoch": 0.8569236980218006,
      "grad_norm": 106.03702545166016,
      "learning_rate": 1.0546500396810443e-06,
      "loss": 4.5333,
      "step": 21226
    },
    {
      "epoch": 0.8569640694388373,
      "grad_norm": 88.05113220214844,
      "learning_rate": 1.054065630438278e-06,
      "loss": 6.4688,
      "step": 21227
    },
    {
      "epoch": 0.857004440855874,
      "grad_norm": 9551.552734375,
      "learning_rate": 1.0534813741501005e-06,
      "loss": 6.2363,
      "step": 21228
    },
    {
      "epoch": 0.8570448122729107,
      "grad_norm": 87.86534881591797,
      "learning_rate": 1.0528972708265006e-06,
      "loss": 4.8704,
      "step": 21229
    },
    {
      "epoch": 0.8570851836899476,
      "grad_norm": 62.52537155151367,
      "learning_rate": 1.0523133204774626e-06,
      "loss": 2.0572,
      "step": 21230
    },
    {
      "epoch": 0.8571255551069843,
      "grad_norm": 108.94261169433594,
      "learning_rate": 1.051729523112973e-06,
      "loss": 5.1014,
      "step": 21231
    },
    {
      "epoch": 0.857165926524021,
      "grad_norm": 103.74674987792969,
      "learning_rate": 1.0511458787430128e-06,
      "loss": 5.1532,
      "step": 21232
    },
    {
      "epoch": 0.8572062979410577,
      "grad_norm": 80.71795654296875,
      "learning_rate": 1.0505623873775628e-06,
      "loss": 5.0726,
      "step": 21233
    },
    {
      "epoch": 0.8572466693580945,
      "grad_norm": 87.18062591552734,
      "learning_rate": 1.049979049026596e-06,
      "loss": 4.796,
      "step": 21234
    },
    {
      "epoch": 0.8572870407751312,
      "grad_norm": 118.27082824707031,
      "learning_rate": 1.0493958637000867e-06,
      "loss": 5.2766,
      "step": 21235
    },
    {
      "epoch": 0.8573274121921679,
      "grad_norm": 113.79377746582031,
      "learning_rate": 1.0488128314080082e-06,
      "loss": 5.3996,
      "step": 21236
    },
    {
      "epoch": 0.8573677836092047,
      "grad_norm": 69.56758117675781,
      "learning_rate": 1.048229952160329e-06,
      "loss": 4.9237,
      "step": 21237
    },
    {
      "epoch": 0.8574081550262415,
      "grad_norm": 102.16536712646484,
      "learning_rate": 1.0476472259670112e-06,
      "loss": 4.9889,
      "step": 21238
    },
    {
      "epoch": 0.8574485264432782,
      "grad_norm": 86.95552825927734,
      "learning_rate": 1.0470646528380214e-06,
      "loss": 4.2223,
      "step": 21239
    },
    {
      "epoch": 0.8574888978603149,
      "grad_norm": 100.7134780883789,
      "learning_rate": 1.0464822327833191e-06,
      "loss": 3.9377,
      "step": 21240
    },
    {
      "epoch": 0.8575292692773516,
      "grad_norm": 96.59870147705078,
      "learning_rate": 1.0458999658128632e-06,
      "loss": 6.3662,
      "step": 21241
    },
    {
      "epoch": 0.8575696406943883,
      "grad_norm": 81.28784942626953,
      "learning_rate": 1.045317851936608e-06,
      "loss": 5.3942,
      "step": 21242
    },
    {
      "epoch": 0.8576100121114251,
      "grad_norm": 100.70301818847656,
      "learning_rate": 1.0447358911645067e-06,
      "loss": 6.0926,
      "step": 21243
    },
    {
      "epoch": 0.8576503835284619,
      "grad_norm": 92.31935119628906,
      "learning_rate": 1.0441540835065101e-06,
      "loss": 3.4207,
      "step": 21244
    },
    {
      "epoch": 0.8576907549454986,
      "grad_norm": 128.68157958984375,
      "learning_rate": 1.0435724289725668e-06,
      "loss": 6.1663,
      "step": 21245
    },
    {
      "epoch": 0.8577311263625353,
      "grad_norm": 179.76246643066406,
      "learning_rate": 1.042990927572618e-06,
      "loss": 5.3053,
      "step": 21246
    },
    {
      "epoch": 0.8577714977795721,
      "grad_norm": 63.048011779785156,
      "learning_rate": 1.042409579316609e-06,
      "loss": 3.3856,
      "step": 21247
    },
    {
      "epoch": 0.8578118691966088,
      "grad_norm": 89.16693878173828,
      "learning_rate": 1.0418283842144782e-06,
      "loss": 3.0952,
      "step": 21248
    },
    {
      "epoch": 0.8578522406136455,
      "grad_norm": 123.59259796142578,
      "learning_rate": 1.041247342276166e-06,
      "loss": 4.6371,
      "step": 21249
    },
    {
      "epoch": 0.8578926120306822,
      "grad_norm": 134.4342041015625,
      "learning_rate": 1.0406664535116018e-06,
      "loss": 4.753,
      "step": 21250
    },
    {
      "epoch": 0.8579329834477191,
      "grad_norm": 96.16983795166016,
      "learning_rate": 1.040085717930719e-06,
      "loss": 6.7186,
      "step": 21251
    },
    {
      "epoch": 0.8579733548647558,
      "grad_norm": 100.38426208496094,
      "learning_rate": 1.039505135543447e-06,
      "loss": 2.3325,
      "step": 21252
    },
    {
      "epoch": 0.8580137262817925,
      "grad_norm": 121.76806640625,
      "learning_rate": 1.0389247063597162e-06,
      "loss": 5.0394,
      "step": 21253
    },
    {
      "epoch": 0.8580540976988292,
      "grad_norm": 175.79086303710938,
      "learning_rate": 1.0383444303894453e-06,
      "loss": 4.9329,
      "step": 21254
    },
    {
      "epoch": 0.858094469115866,
      "grad_norm": 80.81752014160156,
      "learning_rate": 1.037764307642557e-06,
      "loss": 4.5866,
      "step": 21255
    },
    {
      "epoch": 0.8581348405329027,
      "grad_norm": 142.98118591308594,
      "learning_rate": 1.0371843381289715e-06,
      "loss": 2.7001,
      "step": 21256
    },
    {
      "epoch": 0.8581752119499394,
      "grad_norm": 139.51217651367188,
      "learning_rate": 1.0366045218586053e-06,
      "loss": 5.1373,
      "step": 21257
    },
    {
      "epoch": 0.8582155833669762,
      "grad_norm": 125.11044311523438,
      "learning_rate": 1.0360248588413691e-06,
      "loss": 4.7415,
      "step": 21258
    },
    {
      "epoch": 0.858255954784013,
      "grad_norm": 81.05472564697266,
      "learning_rate": 1.0354453490871753e-06,
      "loss": 2.5449,
      "step": 21259
    },
    {
      "epoch": 0.8582963262010497,
      "grad_norm": 100.86151123046875,
      "learning_rate": 1.0348659926059323e-06,
      "loss": 5.1256,
      "step": 21260
    },
    {
      "epoch": 0.8583366976180864,
      "grad_norm": 89.17401885986328,
      "learning_rate": 1.034286789407546e-06,
      "loss": 5.0616,
      "step": 21261
    },
    {
      "epoch": 0.8583770690351231,
      "grad_norm": 64.22962188720703,
      "learning_rate": 1.033707739501919e-06,
      "loss": 5.5128,
      "step": 21262
    },
    {
      "epoch": 0.8584174404521598,
      "grad_norm": 92.4654312133789,
      "learning_rate": 1.0331288428989517e-06,
      "loss": 5.9557,
      "step": 21263
    },
    {
      "epoch": 0.8584578118691966,
      "grad_norm": 99.7177734375,
      "learning_rate": 1.0325500996085426e-06,
      "loss": 4.5165,
      "step": 21264
    },
    {
      "epoch": 0.8584981832862334,
      "grad_norm": 85.98663330078125,
      "learning_rate": 1.031971509640587e-06,
      "loss": 5.556,
      "step": 21265
    },
    {
      "epoch": 0.8585385547032701,
      "grad_norm": 115.25959777832031,
      "learning_rate": 1.0313930730049758e-06,
      "loss": 4.5267,
      "step": 21266
    },
    {
      "epoch": 0.8585789261203068,
      "grad_norm": 82.02840423583984,
      "learning_rate": 1.0308147897115995e-06,
      "loss": 6.341,
      "step": 21267
    },
    {
      "epoch": 0.8586192975373436,
      "grad_norm": 112.1173095703125,
      "learning_rate": 1.0302366597703462e-06,
      "loss": 4.5785,
      "step": 21268
    },
    {
      "epoch": 0.8586596689543803,
      "grad_norm": 79.78590393066406,
      "learning_rate": 1.0296586831911014e-06,
      "loss": 5.5562,
      "step": 21269
    },
    {
      "epoch": 0.858700040371417,
      "grad_norm": 81.61784362792969,
      "learning_rate": 1.0290808599837442e-06,
      "loss": 3.9179,
      "step": 21270
    },
    {
      "epoch": 0.8587404117884537,
      "grad_norm": 92.5789794921875,
      "learning_rate": 1.0285031901581566e-06,
      "loss": 5.7505,
      "step": 21271
    },
    {
      "epoch": 0.8587807832054906,
      "grad_norm": 88.37294006347656,
      "learning_rate": 1.0279256737242127e-06,
      "loss": 3.6456,
      "step": 21272
    },
    {
      "epoch": 0.8588211546225273,
      "grad_norm": 137.17108154296875,
      "learning_rate": 1.0273483106917915e-06,
      "loss": 6.1336,
      "step": 21273
    },
    {
      "epoch": 0.858861526039564,
      "grad_norm": 86.58626556396484,
      "learning_rate": 1.0267711010707594e-06,
      "loss": 5.6803,
      "step": 21274
    },
    {
      "epoch": 0.8589018974566007,
      "grad_norm": 83.32563018798828,
      "learning_rate": 1.0261940448709883e-06,
      "loss": 5.6502,
      "step": 21275
    },
    {
      "epoch": 0.8589422688736374,
      "grad_norm": 123.38133239746094,
      "learning_rate": 1.025617142102342e-06,
      "loss": 3.1385,
      "step": 21276
    },
    {
      "epoch": 0.8589826402906742,
      "grad_norm": 105.54997253417969,
      "learning_rate": 1.0250403927746888e-06,
      "loss": 2.8591,
      "step": 21277
    },
    {
      "epoch": 0.8590230117077109,
      "grad_norm": 97.73517608642578,
      "learning_rate": 1.0244637968978855e-06,
      "loss": 6.2697,
      "step": 21278
    },
    {
      "epoch": 0.8590633831247477,
      "grad_norm": 79.6026382446289,
      "learning_rate": 1.0238873544817906e-06,
      "loss": 5.0642,
      "step": 21279
    },
    {
      "epoch": 0.8591037545417844,
      "grad_norm": 81.57005310058594,
      "learning_rate": 1.023311065536262e-06,
      "loss": 4.9793,
      "step": 21280
    },
    {
      "epoch": 0.8591441259588212,
      "grad_norm": 98.616943359375,
      "learning_rate": 1.022734930071152e-06,
      "loss": 3.7399,
      "step": 21281
    },
    {
      "epoch": 0.8591844973758579,
      "grad_norm": 102.96621704101562,
      "learning_rate": 1.0221589480963111e-06,
      "loss": 5.7712,
      "step": 21282
    },
    {
      "epoch": 0.8592248687928946,
      "grad_norm": 90.09619903564453,
      "learning_rate": 1.0215831196215875e-06,
      "loss": 5.1862,
      "step": 21283
    },
    {
      "epoch": 0.8592652402099313,
      "grad_norm": 124.7348403930664,
      "learning_rate": 1.0210074446568274e-06,
      "loss": 6.3495,
      "step": 21284
    },
    {
      "epoch": 0.8593056116269681,
      "grad_norm": 116.94223022460938,
      "learning_rate": 1.020431923211873e-06,
      "loss": 7.3833,
      "step": 21285
    },
    {
      "epoch": 0.8593459830440049,
      "grad_norm": 76.94965362548828,
      "learning_rate": 1.0198565552965621e-06,
      "loss": 5.1358,
      "step": 21286
    },
    {
      "epoch": 0.8593863544610416,
      "grad_norm": 87.59954071044922,
      "learning_rate": 1.0192813409207348e-06,
      "loss": 4.4135,
      "step": 21287
    },
    {
      "epoch": 0.8594267258780783,
      "grad_norm": 67.05567169189453,
      "learning_rate": 1.0187062800942248e-06,
      "loss": 4.2947,
      "step": 21288
    },
    {
      "epoch": 0.8594670972951151,
      "grad_norm": 128.7158203125,
      "learning_rate": 1.0181313728268671e-06,
      "loss": 5.7791,
      "step": 21289
    },
    {
      "epoch": 0.8595074687121518,
      "grad_norm": 71.18495178222656,
      "learning_rate": 1.0175566191284859e-06,
      "loss": 5.7833,
      "step": 21290
    },
    {
      "epoch": 0.8595478401291885,
      "grad_norm": 110.50572204589844,
      "learning_rate": 1.0169820190089108e-06,
      "loss": 4.9386,
      "step": 21291
    },
    {
      "epoch": 0.8595882115462252,
      "grad_norm": 93.21839141845703,
      "learning_rate": 1.0164075724779666e-06,
      "loss": 6.4909,
      "step": 21292
    },
    {
      "epoch": 0.8596285829632621,
      "grad_norm": 98.58417510986328,
      "learning_rate": 1.0158332795454774e-06,
      "loss": 5.6203,
      "step": 21293
    },
    {
      "epoch": 0.8596689543802988,
      "grad_norm": 79.45056915283203,
      "learning_rate": 1.0152591402212563e-06,
      "loss": 5.5829,
      "step": 21294
    },
    {
      "epoch": 0.8597093257973355,
      "grad_norm": 91.9634017944336,
      "learning_rate": 1.014685154515125e-06,
      "loss": 6.2108,
      "step": 21295
    },
    {
      "epoch": 0.8597496972143722,
      "grad_norm": 72.72213745117188,
      "learning_rate": 1.0141113224368936e-06,
      "loss": 5.1072,
      "step": 21296
    },
    {
      "epoch": 0.8597900686314089,
      "grad_norm": 67.34846496582031,
      "learning_rate": 1.0135376439963773e-06,
      "loss": 5.5886,
      "step": 21297
    },
    {
      "epoch": 0.8598304400484457,
      "grad_norm": 94.53658294677734,
      "learning_rate": 1.0129641192033811e-06,
      "loss": 5.431,
      "step": 21298
    },
    {
      "epoch": 0.8598708114654824,
      "grad_norm": 290.5487365722656,
      "learning_rate": 1.012390748067711e-06,
      "loss": 3.7995,
      "step": 21299
    },
    {
      "epoch": 0.8599111828825192,
      "grad_norm": 93.19038391113281,
      "learning_rate": 1.011817530599173e-06,
      "loss": 5.4547,
      "step": 21300
    },
    {
      "epoch": 0.8599515542995559,
      "grad_norm": 73.0396499633789,
      "learning_rate": 1.011244466807565e-06,
      "loss": 4.5414,
      "step": 21301
    },
    {
      "epoch": 0.8599919257165927,
      "grad_norm": 71.49687194824219,
      "learning_rate": 1.010671556702687e-06,
      "loss": 5.129,
      "step": 21302
    },
    {
      "epoch": 0.8600322971336294,
      "grad_norm": 78.39071655273438,
      "learning_rate": 1.0100988002943335e-06,
      "loss": 6.1926,
      "step": 21303
    },
    {
      "epoch": 0.8600726685506661,
      "grad_norm": 147.59523010253906,
      "learning_rate": 1.0095261975922987e-06,
      "loss": 5.6928,
      "step": 21304
    },
    {
      "epoch": 0.8601130399677028,
      "grad_norm": 68.8983383178711,
      "learning_rate": 1.0089537486063728e-06,
      "loss": 4.7268,
      "step": 21305
    },
    {
      "epoch": 0.8601534113847396,
      "grad_norm": 106.9955062866211,
      "learning_rate": 1.0083814533463398e-06,
      "loss": 4.2495,
      "step": 21306
    },
    {
      "epoch": 0.8601937828017764,
      "grad_norm": 113.34778594970703,
      "learning_rate": 1.0078093118219878e-06,
      "loss": 4.7637,
      "step": 21307
    },
    {
      "epoch": 0.8602341542188131,
      "grad_norm": 94.52644348144531,
      "learning_rate": 1.0072373240430976e-06,
      "loss": 4.0412,
      "step": 21308
    },
    {
      "epoch": 0.8602745256358498,
      "grad_norm": 88.9233627319336,
      "learning_rate": 1.0066654900194517e-06,
      "loss": 5.244,
      "step": 21309
    },
    {
      "epoch": 0.8603148970528866,
      "grad_norm": 106.46173095703125,
      "learning_rate": 1.006093809760823e-06,
      "loss": 4.5076,
      "step": 21310
    },
    {
      "epoch": 0.8603552684699233,
      "grad_norm": 145.60842895507812,
      "learning_rate": 1.0055222832769884e-06,
      "loss": 4.5113,
      "step": 21311
    },
    {
      "epoch": 0.86039563988696,
      "grad_norm": 100.9859390258789,
      "learning_rate": 1.004950910577719e-06,
      "loss": 4.5255,
      "step": 21312
    },
    {
      "epoch": 0.8604360113039967,
      "grad_norm": 147.0537567138672,
      "learning_rate": 1.0043796916727856e-06,
      "loss": 4.4848,
      "step": 21313
    },
    {
      "epoch": 0.8604763827210336,
      "grad_norm": 151.626708984375,
      "learning_rate": 1.0038086265719505e-06,
      "loss": 6.0415,
      "step": 21314
    },
    {
      "epoch": 0.8605167541380703,
      "grad_norm": 107.09378051757812,
      "learning_rate": 1.0032377152849816e-06,
      "loss": 5.3267,
      "step": 21315
    },
    {
      "epoch": 0.860557125555107,
      "grad_norm": 76.53790283203125,
      "learning_rate": 1.0026669578216385e-06,
      "loss": 3.884,
      "step": 21316
    },
    {
      "epoch": 0.8605974969721437,
      "grad_norm": 120.95457458496094,
      "learning_rate": 1.0020963541916817e-06,
      "loss": 5.0848,
      "step": 21317
    },
    {
      "epoch": 0.8606378683891804,
      "grad_norm": 83.15113830566406,
      "learning_rate": 1.0015259044048642e-06,
      "loss": 4.6286,
      "step": 21318
    },
    {
      "epoch": 0.8606782398062172,
      "grad_norm": 90.97748565673828,
      "learning_rate": 1.0009556084709405e-06,
      "loss": 5.7868,
      "step": 21319
    },
    {
      "epoch": 0.8607186112232539,
      "grad_norm": 81.13371276855469,
      "learning_rate": 1.0003854663996615e-06,
      "loss": 4.1627,
      "step": 21320
    },
    {
      "epoch": 0.8607589826402907,
      "grad_norm": 65.98823547363281,
      "learning_rate": 9.998154782007752e-07,
      "loss": 3.6346,
      "step": 21321
    },
    {
      "epoch": 0.8607993540573274,
      "grad_norm": 81.72796630859375,
      "learning_rate": 9.992456438840281e-07,
      "loss": 5.415,
      "step": 21322
    },
    {
      "epoch": 0.8608397254743642,
      "grad_norm": 121.36666107177734,
      "learning_rate": 9.986759634591613e-07,
      "loss": 4.6593,
      "step": 21323
    },
    {
      "epoch": 0.8608800968914009,
      "grad_norm": 110.37574005126953,
      "learning_rate": 9.981064369359173e-07,
      "loss": 3.5564,
      "step": 21324
    },
    {
      "epoch": 0.8609204683084376,
      "grad_norm": 111.03277587890625,
      "learning_rate": 9.975370643240335e-07,
      "loss": 3.8773,
      "step": 21325
    },
    {
      "epoch": 0.8609608397254743,
      "grad_norm": 112.9854965209961,
      "learning_rate": 9.96967845633241e-07,
      "loss": 5.517,
      "step": 21326
    },
    {
      "epoch": 0.861001211142511,
      "grad_norm": 117.11937713623047,
      "learning_rate": 9.963987808732756e-07,
      "loss": 4.5369,
      "step": 21327
    },
    {
      "epoch": 0.8610415825595479,
      "grad_norm": 101.37564849853516,
      "learning_rate": 9.95829870053865e-07,
      "loss": 5.4319,
      "step": 21328
    },
    {
      "epoch": 0.8610819539765846,
      "grad_norm": 82.72781372070312,
      "learning_rate": 9.9526111318474e-07,
      "loss": 5.553,
      "step": 21329
    },
    {
      "epoch": 0.8611223253936213,
      "grad_norm": 119.23153686523438,
      "learning_rate": 9.94692510275621e-07,
      "loss": 4.8018,
      "step": 21330
    },
    {
      "epoch": 0.861162696810658,
      "grad_norm": 80.06905364990234,
      "learning_rate": 9.941240613362313e-07,
      "loss": 4.3463,
      "step": 21331
    },
    {
      "epoch": 0.8612030682276948,
      "grad_norm": 79.42462921142578,
      "learning_rate": 9.935557663762884e-07,
      "loss": 4.7372,
      "step": 21332
    },
    {
      "epoch": 0.8612434396447315,
      "grad_norm": 113.55914306640625,
      "learning_rate": 9.929876254055138e-07,
      "loss": 5.1132,
      "step": 21333
    },
    {
      "epoch": 0.8612838110617683,
      "grad_norm": 106.89547729492188,
      "learning_rate": 9.92419638433615e-07,
      "loss": 4.0728,
      "step": 21334
    },
    {
      "epoch": 0.861324182478805,
      "grad_norm": 93.24302673339844,
      "learning_rate": 9.918518054703063e-07,
      "loss": 3.5099,
      "step": 21335
    },
    {
      "epoch": 0.8613645538958418,
      "grad_norm": 104.23756408691406,
      "learning_rate": 9.91284126525297e-07,
      "loss": 4.4016,
      "step": 21336
    },
    {
      "epoch": 0.8614049253128785,
      "grad_norm": 116.7765121459961,
      "learning_rate": 9.907166016082936e-07,
      "loss": 4.3313,
      "step": 21337
    },
    {
      "epoch": 0.8614452967299152,
      "grad_norm": 120.44263458251953,
      "learning_rate": 9.90149230728996e-07,
      "loss": 4.8676,
      "step": 21338
    },
    {
      "epoch": 0.8614856681469519,
      "grad_norm": 471.7138366699219,
      "learning_rate": 9.89582013897108e-07,
      "loss": 4.0204,
      "step": 21339
    },
    {
      "epoch": 0.8615260395639887,
      "grad_norm": 94.8929214477539,
      "learning_rate": 9.890149511223256e-07,
      "loss": 2.8318,
      "step": 21340
    },
    {
      "epoch": 0.8615664109810255,
      "grad_norm": 67.5451889038086,
      "learning_rate": 9.884480424143462e-07,
      "loss": 3.7257,
      "step": 21341
    },
    {
      "epoch": 0.8616067823980622,
      "grad_norm": 89.45523834228516,
      "learning_rate": 9.878812877828614e-07,
      "loss": 5.6243,
      "step": 21342
    },
    {
      "epoch": 0.8616471538150989,
      "grad_norm": 120.27705383300781,
      "learning_rate": 9.873146872375617e-07,
      "loss": 3.9598,
      "step": 21343
    },
    {
      "epoch": 0.8616875252321357,
      "grad_norm": 102.6404800415039,
      "learning_rate": 9.867482407881346e-07,
      "loss": 4.6339,
      "step": 21344
    },
    {
      "epoch": 0.8617278966491724,
      "grad_norm": 86.05159759521484,
      "learning_rate": 9.86181948444267e-07,
      "loss": 5.2071,
      "step": 21345
    },
    {
      "epoch": 0.8617682680662091,
      "grad_norm": 97.85977935791016,
      "learning_rate": 9.856158102156378e-07,
      "loss": 5.606,
      "step": 21346
    },
    {
      "epoch": 0.8618086394832458,
      "grad_norm": 111.7083511352539,
      "learning_rate": 9.85049826111928e-07,
      "loss": 4.1607,
      "step": 21347
    },
    {
      "epoch": 0.8618490109002827,
      "grad_norm": 113.66283416748047,
      "learning_rate": 9.844839961428142e-07,
      "loss": 4.7927,
      "step": 21348
    },
    {
      "epoch": 0.8618893823173194,
      "grad_norm": 96.22069549560547,
      "learning_rate": 9.839183203179736e-07,
      "loss": 5.5011,
      "step": 21349
    },
    {
      "epoch": 0.8619297537343561,
      "grad_norm": 100.21343231201172,
      "learning_rate": 9.833527986470748e-07,
      "loss": 4.6868,
      "step": 21350
    },
    {
      "epoch": 0.8619701251513928,
      "grad_norm": 107.58956909179688,
      "learning_rate": 9.827874311397867e-07,
      "loss": 5.21,
      "step": 21351
    },
    {
      "epoch": 0.8620104965684295,
      "grad_norm": 129.17446899414062,
      "learning_rate": 9.82222217805776e-07,
      "loss": 5.2655,
      "step": 21352
    },
    {
      "epoch": 0.8620508679854663,
      "grad_norm": 141.47738647460938,
      "learning_rate": 9.816571586547108e-07,
      "loss": 5.909,
      "step": 21353
    },
    {
      "epoch": 0.862091239402503,
      "grad_norm": 88.68965911865234,
      "learning_rate": 9.810922536962464e-07,
      "loss": 5.4701,
      "step": 21354
    },
    {
      "epoch": 0.8621316108195398,
      "grad_norm": 135.10208129882812,
      "learning_rate": 9.805275029400442e-07,
      "loss": 4.0604,
      "step": 21355
    },
    {
      "epoch": 0.8621719822365765,
      "grad_norm": 197.01524353027344,
      "learning_rate": 9.799629063957584e-07,
      "loss": 4.9373,
      "step": 21356
    },
    {
      "epoch": 0.8622123536536133,
      "grad_norm": 89.74749755859375,
      "learning_rate": 9.793984640730448e-07,
      "loss": 5.8177,
      "step": 21357
    },
    {
      "epoch": 0.86225272507065,
      "grad_norm": 115.50382995605469,
      "learning_rate": 9.788341759815522e-07,
      "loss": 4.3213,
      "step": 21358
    },
    {
      "epoch": 0.8622930964876867,
      "grad_norm": 121.05706787109375,
      "learning_rate": 9.782700421309299e-07,
      "loss": 4.3599,
      "step": 21359
    },
    {
      "epoch": 0.8623334679047234,
      "grad_norm": 75.33959197998047,
      "learning_rate": 9.777060625308221e-07,
      "loss": 4.5838,
      "step": 21360
    },
    {
      "epoch": 0.8623738393217601,
      "grad_norm": 92.13782501220703,
      "learning_rate": 9.771422371908746e-07,
      "loss": 6.0859,
      "step": 21361
    },
    {
      "epoch": 0.862414210738797,
      "grad_norm": 92.41876983642578,
      "learning_rate": 9.765785661207228e-07,
      "loss": 6.3733,
      "step": 21362
    },
    {
      "epoch": 0.8624545821558337,
      "grad_norm": 161.04840087890625,
      "learning_rate": 9.760150493300059e-07,
      "loss": 5.4885,
      "step": 21363
    },
    {
      "epoch": 0.8624949535728704,
      "grad_norm": 113.78021240234375,
      "learning_rate": 9.754516868283591e-07,
      "loss": 3.8147,
      "step": 21364
    },
    {
      "epoch": 0.8625353249899071,
      "grad_norm": 83.77200317382812,
      "learning_rate": 9.748884786254164e-07,
      "loss": 4.2423,
      "step": 21365
    },
    {
      "epoch": 0.8625756964069439,
      "grad_norm": 104.56539154052734,
      "learning_rate": 9.743254247308042e-07,
      "loss": 5.9274,
      "step": 21366
    },
    {
      "epoch": 0.8626160678239806,
      "grad_norm": 107.9581298828125,
      "learning_rate": 9.737625251541505e-07,
      "loss": 5.8118,
      "step": 21367
    },
    {
      "epoch": 0.8626564392410173,
      "grad_norm": 76.875244140625,
      "learning_rate": 9.731997799050807e-07,
      "loss": 6.4095,
      "step": 21368
    },
    {
      "epoch": 0.8626968106580541,
      "grad_norm": 72.27626037597656,
      "learning_rate": 9.726371889932162e-07,
      "loss": 4.809,
      "step": 21369
    },
    {
      "epoch": 0.8627371820750909,
      "grad_norm": 104.06232452392578,
      "learning_rate": 9.720747524281738e-07,
      "loss": 5.8021,
      "step": 21370
    },
    {
      "epoch": 0.8627775534921276,
      "grad_norm": 119.61763763427734,
      "learning_rate": 9.715124702195722e-07,
      "loss": 4.5678,
      "step": 21371
    },
    {
      "epoch": 0.8628179249091643,
      "grad_norm": 109.26634979248047,
      "learning_rate": 9.709503423770239e-07,
      "loss": 4.5804,
      "step": 21372
    },
    {
      "epoch": 0.862858296326201,
      "grad_norm": 109.32766723632812,
      "learning_rate": 9.703883689101411e-07,
      "loss": 4.5529,
      "step": 21373
    },
    {
      "epoch": 0.8628986677432378,
      "grad_norm": 85.49931335449219,
      "learning_rate": 9.698265498285309e-07,
      "loss": 5.8917,
      "step": 21374
    },
    {
      "epoch": 0.8629390391602745,
      "grad_norm": 82.80999755859375,
      "learning_rate": 9.692648851417996e-07,
      "loss": 6.202,
      "step": 21375
    },
    {
      "epoch": 0.8629794105773113,
      "grad_norm": 106.56202697753906,
      "learning_rate": 9.68703374859551e-07,
      "loss": 5.9769,
      "step": 21376
    },
    {
      "epoch": 0.863019781994348,
      "grad_norm": 94.30058288574219,
      "learning_rate": 9.681420189913837e-07,
      "loss": 5.3172,
      "step": 21377
    },
    {
      "epoch": 0.8630601534113848,
      "grad_norm": 66.92985534667969,
      "learning_rate": 9.67580817546898e-07,
      "loss": 4.3087,
      "step": 21378
    },
    {
      "epoch": 0.8631005248284215,
      "grad_norm": 118.401123046875,
      "learning_rate": 9.670197705356877e-07,
      "loss": 4.7861,
      "step": 21379
    },
    {
      "epoch": 0.8631408962454582,
      "grad_norm": 124.92144012451172,
      "learning_rate": 9.664588779673457e-07,
      "loss": 5.843,
      "step": 21380
    },
    {
      "epoch": 0.8631812676624949,
      "grad_norm": 161.23667907714844,
      "learning_rate": 9.658981398514645e-07,
      "loss": 4.1906,
      "step": 21381
    },
    {
      "epoch": 0.8632216390795316,
      "grad_norm": 126.85106658935547,
      "learning_rate": 9.653375561976263e-07,
      "loss": 6.2094,
      "step": 21382
    },
    {
      "epoch": 0.8632620104965685,
      "grad_norm": 165.32923889160156,
      "learning_rate": 9.647771270154194e-07,
      "loss": 3.9878,
      "step": 21383
    },
    {
      "epoch": 0.8633023819136052,
      "grad_norm": 139.46351623535156,
      "learning_rate": 9.642168523144247e-07,
      "loss": 5.3518,
      "step": 21384
    },
    {
      "epoch": 0.8633427533306419,
      "grad_norm": 151.62429809570312,
      "learning_rate": 9.636567321042233e-07,
      "loss": 5.2597,
      "step": 21385
    },
    {
      "epoch": 0.8633831247476786,
      "grad_norm": 94.48567962646484,
      "learning_rate": 9.63096766394388e-07,
      "loss": 5.1609,
      "step": 21386
    },
    {
      "epoch": 0.8634234961647154,
      "grad_norm": 101.20035552978516,
      "learning_rate": 9.625369551944963e-07,
      "loss": 4.6119,
      "step": 21387
    },
    {
      "epoch": 0.8634638675817521,
      "grad_norm": 222.76039123535156,
      "learning_rate": 9.619772985141173e-07,
      "loss": 6.1711,
      "step": 21388
    },
    {
      "epoch": 0.8635042389987888,
      "grad_norm": 64.32736206054688,
      "learning_rate": 9.614177963628246e-07,
      "loss": 5.8881,
      "step": 21389
    },
    {
      "epoch": 0.8635446104158256,
      "grad_norm": 92.65770721435547,
      "learning_rate": 9.608584487501781e-07,
      "loss": 4.6598,
      "step": 21390
    },
    {
      "epoch": 0.8635849818328624,
      "grad_norm": 172.67909240722656,
      "learning_rate": 9.602992556857448e-07,
      "loss": 4.1991,
      "step": 21391
    },
    {
      "epoch": 0.8636253532498991,
      "grad_norm": 79.53937530517578,
      "learning_rate": 9.597402171790848e-07,
      "loss": 4.4417,
      "step": 21392
    },
    {
      "epoch": 0.8636657246669358,
      "grad_norm": 58.169063568115234,
      "learning_rate": 9.591813332397582e-07,
      "loss": 5.8655,
      "step": 21393
    },
    {
      "epoch": 0.8637060960839725,
      "grad_norm": 94.79190063476562,
      "learning_rate": 9.586226038773173e-07,
      "loss": 5.2364,
      "step": 21394
    },
    {
      "epoch": 0.8637464675010093,
      "grad_norm": 75.15308380126953,
      "learning_rate": 9.580640291013166e-07,
      "loss": 4.9574,
      "step": 21395
    },
    {
      "epoch": 0.863786838918046,
      "grad_norm": 84.42302703857422,
      "learning_rate": 9.575056089213075e-07,
      "loss": 5.8409,
      "step": 21396
    },
    {
      "epoch": 0.8638272103350828,
      "grad_norm": 95.02254486083984,
      "learning_rate": 9.569473433468356e-07,
      "loss": 4.8672,
      "step": 21397
    },
    {
      "epoch": 0.8638675817521195,
      "grad_norm": 86.59199523925781,
      "learning_rate": 9.563892323874468e-07,
      "loss": 4.152,
      "step": 21398
    },
    {
      "epoch": 0.8639079531691563,
      "grad_norm": 75.32305145263672,
      "learning_rate": 9.558312760526856e-07,
      "loss": 3.3388,
      "step": 21399
    },
    {
      "epoch": 0.863948324586193,
      "grad_norm": 118.64326477050781,
      "learning_rate": 9.552734743520886e-07,
      "loss": 4.6932,
      "step": 21400
    },
    {
      "epoch": 0.8639886960032297,
      "grad_norm": 59.68312072753906,
      "learning_rate": 9.547158272951961e-07,
      "loss": 4.1376,
      "step": 21401
    },
    {
      "epoch": 0.8640290674202664,
      "grad_norm": 122.62237548828125,
      "learning_rate": 9.541583348915383e-07,
      "loss": 4.5877,
      "step": 21402
    },
    {
      "epoch": 0.8640694388373031,
      "grad_norm": 130.2911834716797,
      "learning_rate": 9.53600997150651e-07,
      "loss": 6.5403,
      "step": 21403
    },
    {
      "epoch": 0.86410981025434,
      "grad_norm": 119.97721862792969,
      "learning_rate": 9.530438140820608e-07,
      "loss": 4.0249,
      "step": 21404
    },
    {
      "epoch": 0.8641501816713767,
      "grad_norm": 106.92154693603516,
      "learning_rate": 9.524867856952969e-07,
      "loss": 5.8705,
      "step": 21405
    },
    {
      "epoch": 0.8641905530884134,
      "grad_norm": 92.88895416259766,
      "learning_rate": 9.519299119998793e-07,
      "loss": 6.2076,
      "step": 21406
    },
    {
      "epoch": 0.8642309245054501,
      "grad_norm": 120.85380554199219,
      "learning_rate": 9.513731930053327e-07,
      "loss": 4.3158,
      "step": 21407
    },
    {
      "epoch": 0.8642712959224869,
      "grad_norm": 120.77440643310547,
      "learning_rate": 9.508166287211739e-07,
      "loss": 4.2055,
      "step": 21408
    },
    {
      "epoch": 0.8643116673395236,
      "grad_norm": 134.0771942138672,
      "learning_rate": 9.502602191569188e-07,
      "loss": 5.3749,
      "step": 21409
    },
    {
      "epoch": 0.8643520387565603,
      "grad_norm": 102.06181335449219,
      "learning_rate": 9.49703964322084e-07,
      "loss": 4.8635,
      "step": 21410
    },
    {
      "epoch": 0.8643924101735971,
      "grad_norm": 110.17222595214844,
      "learning_rate": 9.491478642261753e-07,
      "loss": 5.2148,
      "step": 21411
    },
    {
      "epoch": 0.8644327815906339,
      "grad_norm": 85.96905517578125,
      "learning_rate": 9.485919188787029e-07,
      "loss": 3.7871,
      "step": 21412
    },
    {
      "epoch": 0.8644731530076706,
      "grad_norm": 130.09017944335938,
      "learning_rate": 9.480361282891715e-07,
      "loss": 5.4966,
      "step": 21413
    },
    {
      "epoch": 0.8645135244247073,
      "grad_norm": 62.445613861083984,
      "learning_rate": 9.474804924670866e-07,
      "loss": 5.2232,
      "step": 21414
    },
    {
      "epoch": 0.864553895841744,
      "grad_norm": 87.68209075927734,
      "learning_rate": 9.469250114219453e-07,
      "loss": 4.4402,
      "step": 21415
    },
    {
      "epoch": 0.8645942672587807,
      "grad_norm": 94.45240020751953,
      "learning_rate": 9.463696851632442e-07,
      "loss": 6.3022,
      "step": 21416
    },
    {
      "epoch": 0.8646346386758175,
      "grad_norm": 94.8478012084961,
      "learning_rate": 9.458145137004804e-07,
      "loss": 3.2305,
      "step": 21417
    },
    {
      "epoch": 0.8646750100928543,
      "grad_norm": 85.93236541748047,
      "learning_rate": 9.452594970431462e-07,
      "loss": 4.2402,
      "step": 21418
    },
    {
      "epoch": 0.864715381509891,
      "grad_norm": 86.5402603149414,
      "learning_rate": 9.447046352007295e-07,
      "loss": 3.7929,
      "step": 21419
    },
    {
      "epoch": 0.8647557529269277,
      "grad_norm": 141.97605895996094,
      "learning_rate": 9.441499281827182e-07,
      "loss": 3.5735,
      "step": 21420
    },
    {
      "epoch": 0.8647961243439645,
      "grad_norm": 94.91761779785156,
      "learning_rate": 9.435953759985961e-07,
      "loss": 3.6391,
      "step": 21421
    },
    {
      "epoch": 0.8648364957610012,
      "grad_norm": 78.12629699707031,
      "learning_rate": 9.430409786578465e-07,
      "loss": 5.1631,
      "step": 21422
    },
    {
      "epoch": 0.8648768671780379,
      "grad_norm": 77.70430755615234,
      "learning_rate": 9.424867361699442e-07,
      "loss": 5.9937,
      "step": 21423
    },
    {
      "epoch": 0.8649172385950746,
      "grad_norm": 74.75315856933594,
      "learning_rate": 9.419326485443692e-07,
      "loss": 3.8102,
      "step": 21424
    },
    {
      "epoch": 0.8649576100121115,
      "grad_norm": 103.60582733154297,
      "learning_rate": 9.413787157905929e-07,
      "loss": 3.9733,
      "step": 21425
    },
    {
      "epoch": 0.8649979814291482,
      "grad_norm": 87.25312042236328,
      "learning_rate": 9.408249379180889e-07,
      "loss": 5.4467,
      "step": 21426
    },
    {
      "epoch": 0.8650383528461849,
      "grad_norm": 180.13677978515625,
      "learning_rate": 9.40271314936323e-07,
      "loss": 5.4398,
      "step": 21427
    },
    {
      "epoch": 0.8650787242632216,
      "grad_norm": 76.33854675292969,
      "learning_rate": 9.397178468547608e-07,
      "loss": 4.728,
      "step": 21428
    },
    {
      "epoch": 0.8651190956802584,
      "grad_norm": 78.99646759033203,
      "learning_rate": 9.391645336828658e-07,
      "loss": 4.695,
      "step": 21429
    },
    {
      "epoch": 0.8651594670972951,
      "grad_norm": 82.77113342285156,
      "learning_rate": 9.386113754301018e-07,
      "loss": 3.7036,
      "step": 21430
    },
    {
      "epoch": 0.8651998385143318,
      "grad_norm": 88.86363220214844,
      "learning_rate": 9.380583721059211e-07,
      "loss": 4.1985,
      "step": 21431
    },
    {
      "epoch": 0.8652402099313686,
      "grad_norm": 113.3113784790039,
      "learning_rate": 9.375055237197806e-07,
      "loss": 5.1696,
      "step": 21432
    },
    {
      "epoch": 0.8652805813484054,
      "grad_norm": 117.18927764892578,
      "learning_rate": 9.369528302811326e-07,
      "loss": 3.5804,
      "step": 21433
    },
    {
      "epoch": 0.8653209527654421,
      "grad_norm": 94.53557586669922,
      "learning_rate": 9.364002917994308e-07,
      "loss": 4.6963,
      "step": 21434
    },
    {
      "epoch": 0.8653613241824788,
      "grad_norm": 74.44216918945312,
      "learning_rate": 9.358479082841165e-07,
      "loss": 3.9142,
      "step": 21435
    },
    {
      "epoch": 0.8654016955995155,
      "grad_norm": 122.03825378417969,
      "learning_rate": 9.352956797446367e-07,
      "loss": 5.3868,
      "step": 21436
    },
    {
      "epoch": 0.8654420670165522,
      "grad_norm": 137.24951171875,
      "learning_rate": 9.347436061904325e-07,
      "loss": 6.9458,
      "step": 21437
    },
    {
      "epoch": 0.8654824384335891,
      "grad_norm": 82.6765365600586,
      "learning_rate": 9.341916876309443e-07,
      "loss": 4.3071,
      "step": 21438
    },
    {
      "epoch": 0.8655228098506258,
      "grad_norm": 697.2628784179688,
      "learning_rate": 9.336399240756078e-07,
      "loss": 5.4734,
      "step": 21439
    },
    {
      "epoch": 0.8655631812676625,
      "grad_norm": 72.82533264160156,
      "learning_rate": 9.330883155338577e-07,
      "loss": 4.8089,
      "step": 21440
    },
    {
      "epoch": 0.8656035526846992,
      "grad_norm": 126.94469451904297,
      "learning_rate": 9.325368620151243e-07,
      "loss": 5.6079,
      "step": 21441
    },
    {
      "epoch": 0.865643924101736,
      "grad_norm": 72.12958526611328,
      "learning_rate": 9.319855635288388e-07,
      "loss": 4.3686,
      "step": 21442
    },
    {
      "epoch": 0.8656842955187727,
      "grad_norm": 90.7397232055664,
      "learning_rate": 9.314344200844227e-07,
      "loss": 4.325,
      "step": 21443
    },
    {
      "epoch": 0.8657246669358094,
      "grad_norm": 176.07423400878906,
      "learning_rate": 9.308834316913007e-07,
      "loss": 5.9765,
      "step": 21444
    },
    {
      "epoch": 0.8657650383528462,
      "grad_norm": 85.6297836303711,
      "learning_rate": 9.303325983588952e-07,
      "loss": 4.85,
      "step": 21445
    },
    {
      "epoch": 0.865805409769883,
      "grad_norm": 120.8936767578125,
      "learning_rate": 9.29781920096624e-07,
      "loss": 3.9364,
      "step": 21446
    },
    {
      "epoch": 0.8658457811869197,
      "grad_norm": 109.20722961425781,
      "learning_rate": 9.292313969139e-07,
      "loss": 5.7222,
      "step": 21447
    },
    {
      "epoch": 0.8658861526039564,
      "grad_norm": 113.12175750732422,
      "learning_rate": 9.286810288201375e-07,
      "loss": 4.2323,
      "step": 21448
    },
    {
      "epoch": 0.8659265240209931,
      "grad_norm": 81.39623260498047,
      "learning_rate": 9.28130815824746e-07,
      "loss": 5.0819,
      "step": 21449
    },
    {
      "epoch": 0.8659668954380298,
      "grad_norm": 92.48604583740234,
      "learning_rate": 9.275807579371354e-07,
      "loss": 5.2111,
      "step": 21450
    },
    {
      "epoch": 0.8660072668550666,
      "grad_norm": 90.73558044433594,
      "learning_rate": 9.270308551667062e-07,
      "loss": 3.2063,
      "step": 21451
    },
    {
      "epoch": 0.8660476382721034,
      "grad_norm": 100.51266479492188,
      "learning_rate": 9.264811075228619e-07,
      "loss": 4.3259,
      "step": 21452
    },
    {
      "epoch": 0.8660880096891401,
      "grad_norm": 66.56234741210938,
      "learning_rate": 9.259315150150039e-07,
      "loss": 4.4314,
      "step": 21453
    },
    {
      "epoch": 0.8661283811061768,
      "grad_norm": 116.30713653564453,
      "learning_rate": 9.25382077652528e-07,
      "loss": 5.205,
      "step": 21454
    },
    {
      "epoch": 0.8661687525232136,
      "grad_norm": 96.00311279296875,
      "learning_rate": 9.248327954448266e-07,
      "loss": 4.972,
      "step": 21455
    },
    {
      "epoch": 0.8662091239402503,
      "grad_norm": 98.49205017089844,
      "learning_rate": 9.242836684012924e-07,
      "loss": 4.2796,
      "step": 21456
    },
    {
      "epoch": 0.866249495357287,
      "grad_norm": 96.98500061035156,
      "learning_rate": 9.237346965313143e-07,
      "loss": 4.6642,
      "step": 21457
    },
    {
      "epoch": 0.8662898667743237,
      "grad_norm": 98.86029052734375,
      "learning_rate": 9.231858798442783e-07,
      "loss": 3.7087,
      "step": 21458
    },
    {
      "epoch": 0.8663302381913606,
      "grad_norm": 99.9793701171875,
      "learning_rate": 9.226372183495669e-07,
      "loss": 5.0924,
      "step": 21459
    },
    {
      "epoch": 0.8663706096083973,
      "grad_norm": 119.07400512695312,
      "learning_rate": 9.220887120565625e-07,
      "loss": 6.0886,
      "step": 21460
    },
    {
      "epoch": 0.866410981025434,
      "grad_norm": 122.9769287109375,
      "learning_rate": 9.215403609746432e-07,
      "loss": 5.9437,
      "step": 21461
    },
    {
      "epoch": 0.8664513524424707,
      "grad_norm": 91.10205078125,
      "learning_rate": 9.209921651131848e-07,
      "loss": 6.3227,
      "step": 21462
    },
    {
      "epoch": 0.8664917238595075,
      "grad_norm": 92.66339874267578,
      "learning_rate": 9.204441244815588e-07,
      "loss": 4.6221,
      "step": 21463
    },
    {
      "epoch": 0.8665320952765442,
      "grad_norm": 121.07550048828125,
      "learning_rate": 9.198962390891353e-07,
      "loss": 5.7429,
      "step": 21464
    },
    {
      "epoch": 0.8665724666935809,
      "grad_norm": 108.72099304199219,
      "learning_rate": 9.193485089452824e-07,
      "loss": 5.3771,
      "step": 21465
    },
    {
      "epoch": 0.8666128381106177,
      "grad_norm": 78.09969329833984,
      "learning_rate": 9.188009340593674e-07,
      "loss": 5.2832,
      "step": 21466
    },
    {
      "epoch": 0.8666532095276545,
      "grad_norm": 98.5585708618164,
      "learning_rate": 9.182535144407479e-07,
      "loss": 4.1803,
      "step": 21467
    },
    {
      "epoch": 0.8666935809446912,
      "grad_norm": 83.90753936767578,
      "learning_rate": 9.177062500987865e-07,
      "loss": 3.2352,
      "step": 21468
    },
    {
      "epoch": 0.8667339523617279,
      "grad_norm": 98.13201141357422,
      "learning_rate": 9.171591410428393e-07,
      "loss": 4.6284,
      "step": 21469
    },
    {
      "epoch": 0.8667743237787646,
      "grad_norm": 75.80256652832031,
      "learning_rate": 9.16612187282262e-07,
      "loss": 5.4317,
      "step": 21470
    },
    {
      "epoch": 0.8668146951958013,
      "grad_norm": 101.925537109375,
      "learning_rate": 9.160653888264049e-07,
      "loss": 5.5219,
      "step": 21471
    },
    {
      "epoch": 0.8668550666128381,
      "grad_norm": 77.15689849853516,
      "learning_rate": 9.15518745684616e-07,
      "loss": 4.2437,
      "step": 21472
    },
    {
      "epoch": 0.8668954380298749,
      "grad_norm": 107.51020050048828,
      "learning_rate": 9.149722578662424e-07,
      "loss": 4.058,
      "step": 21473
    },
    {
      "epoch": 0.8669358094469116,
      "grad_norm": 80.19196319580078,
      "learning_rate": 9.144259253806309e-07,
      "loss": 3.6008,
      "step": 21474
    },
    {
      "epoch": 0.8669761808639483,
      "grad_norm": 90.95337677001953,
      "learning_rate": 9.138797482371165e-07,
      "loss": 7.084,
      "step": 21475
    },
    {
      "epoch": 0.8670165522809851,
      "grad_norm": 123.8241195678711,
      "learning_rate": 9.133337264450426e-07,
      "loss": 5.4524,
      "step": 21476
    },
    {
      "epoch": 0.8670569236980218,
      "grad_norm": 78.71260833740234,
      "learning_rate": 9.127878600137418e-07,
      "loss": 4.607,
      "step": 21477
    },
    {
      "epoch": 0.8670972951150585,
      "grad_norm": 114.48271179199219,
      "learning_rate": 9.12242148952549e-07,
      "loss": 4.0661,
      "step": 21478
    },
    {
      "epoch": 0.8671376665320952,
      "grad_norm": 121.42676544189453,
      "learning_rate": 9.116965932707933e-07,
      "loss": 4.7435,
      "step": 21479
    },
    {
      "epoch": 0.8671780379491321,
      "grad_norm": 123.9840316772461,
      "learning_rate": 9.111511929778039e-07,
      "loss": 5.0199,
      "step": 21480
    },
    {
      "epoch": 0.8672184093661688,
      "grad_norm": 74.45635986328125,
      "learning_rate": 9.106059480829044e-07,
      "loss": 4.3846,
      "step": 21481
    },
    {
      "epoch": 0.8672587807832055,
      "grad_norm": 77.62781524658203,
      "learning_rate": 9.100608585954207e-07,
      "loss": 5.9601,
      "step": 21482
    },
    {
      "epoch": 0.8672991522002422,
      "grad_norm": 115.7463150024414,
      "learning_rate": 9.095159245246677e-07,
      "loss": 4.8439,
      "step": 21483
    },
    {
      "epoch": 0.867339523617279,
      "grad_norm": 111.0759048461914,
      "learning_rate": 9.089711458799644e-07,
      "loss": 3.8881,
      "step": 21484
    },
    {
      "epoch": 0.8673798950343157,
      "grad_norm": 89.46248626708984,
      "learning_rate": 9.084265226706257e-07,
      "loss": 6.0281,
      "step": 21485
    },
    {
      "epoch": 0.8674202664513524,
      "grad_norm": 70.67271423339844,
      "learning_rate": 9.078820549059641e-07,
      "loss": 5.5802,
      "step": 21486
    },
    {
      "epoch": 0.8674606378683892,
      "grad_norm": 84.71551513671875,
      "learning_rate": 9.073377425952867e-07,
      "loss": 5.3953,
      "step": 21487
    },
    {
      "epoch": 0.867501009285426,
      "grad_norm": 130.1609344482422,
      "learning_rate": 9.067935857479004e-07,
      "loss": 4.9316,
      "step": 21488
    },
    {
      "epoch": 0.8675413807024627,
      "grad_norm": 123.02215576171875,
      "learning_rate": 9.0624958437311e-07,
      "loss": 5.202,
      "step": 21489
    },
    {
      "epoch": 0.8675817521194994,
      "grad_norm": 103.4935531616211,
      "learning_rate": 9.057057384802182e-07,
      "loss": 4.7684,
      "step": 21490
    },
    {
      "epoch": 0.8676221235365361,
      "grad_norm": 100.77642059326172,
      "learning_rate": 9.051620480785195e-07,
      "loss": 4.4613,
      "step": 21491
    },
    {
      "epoch": 0.8676624949535728,
      "grad_norm": 105.30431365966797,
      "learning_rate": 9.046185131773122e-07,
      "loss": 5.2282,
      "step": 21492
    },
    {
      "epoch": 0.8677028663706096,
      "grad_norm": 115.61251068115234,
      "learning_rate": 9.040751337858889e-07,
      "loss": 4.9824,
      "step": 21493
    },
    {
      "epoch": 0.8677432377876464,
      "grad_norm": 96.3099365234375,
      "learning_rate": 9.03531909913542e-07,
      "loss": 3.0184,
      "step": 21494
    },
    {
      "epoch": 0.8677836092046831,
      "grad_norm": 109.9767837524414,
      "learning_rate": 9.029888415695554e-07,
      "loss": 4.2123,
      "step": 21495
    },
    {
      "epoch": 0.8678239806217198,
      "grad_norm": 219.8634490966797,
      "learning_rate": 9.02445928763217e-07,
      "loss": 5.7569,
      "step": 21496
    },
    {
      "epoch": 0.8678643520387566,
      "grad_norm": 67.6231918334961,
      "learning_rate": 9.019031715038095e-07,
      "loss": 5.33,
      "step": 21497
    },
    {
      "epoch": 0.8679047234557933,
      "grad_norm": 84.77314758300781,
      "learning_rate": 9.013605698006112e-07,
      "loss": 4.1548,
      "step": 21498
    },
    {
      "epoch": 0.86794509487283,
      "grad_norm": 104.11506652832031,
      "learning_rate": 9.00818123662901e-07,
      "loss": 5.3241,
      "step": 21499
    },
    {
      "epoch": 0.8679854662898667,
      "grad_norm": 105.52911376953125,
      "learning_rate": 9.002758330999528e-07,
      "loss": 4.4127,
      "step": 21500
    },
    {
      "epoch": 0.8680258377069036,
      "grad_norm": 80.68473815917969,
      "learning_rate": 8.99733698121038e-07,
      "loss": 5.1396,
      "step": 21501
    },
    {
      "epoch": 0.8680662091239403,
      "grad_norm": 76.12870788574219,
      "learning_rate": 8.991917187354282e-07,
      "loss": 4.0427,
      "step": 21502
    },
    {
      "epoch": 0.868106580540977,
      "grad_norm": 128.31234741210938,
      "learning_rate": 8.986498949523858e-07,
      "loss": 5.519,
      "step": 21503
    },
    {
      "epoch": 0.8681469519580137,
      "grad_norm": 88.4912109375,
      "learning_rate": 8.98108226781178e-07,
      "loss": 6.3309,
      "step": 21504
    },
    {
      "epoch": 0.8681873233750504,
      "grad_norm": 107.54794311523438,
      "learning_rate": 8.975667142310651e-07,
      "loss": 4.0393,
      "step": 21505
    },
    {
      "epoch": 0.8682276947920872,
      "grad_norm": 88.8200454711914,
      "learning_rate": 8.970253573113063e-07,
      "loss": 6.7243,
      "step": 21506
    },
    {
      "epoch": 0.8682680662091239,
      "grad_norm": 145.07925415039062,
      "learning_rate": 8.964841560311566e-07,
      "loss": 5.2689,
      "step": 21507
    },
    {
      "epoch": 0.8683084376261607,
      "grad_norm": 97.9660415649414,
      "learning_rate": 8.959431103998683e-07,
      "loss": 3.6513,
      "step": 21508
    },
    {
      "epoch": 0.8683488090431974,
      "grad_norm": 81.04985809326172,
      "learning_rate": 8.954022204266944e-07,
      "loss": 4.3156,
      "step": 21509
    },
    {
      "epoch": 0.8683891804602342,
      "grad_norm": 106.613525390625,
      "learning_rate": 8.948614861208827e-07,
      "loss": 4.5752,
      "step": 21510
    },
    {
      "epoch": 0.8684295518772709,
      "grad_norm": 127.66246795654297,
      "learning_rate": 8.943209074916759e-07,
      "loss": 3.1683,
      "step": 21511
    },
    {
      "epoch": 0.8684699232943076,
      "grad_norm": 138.19387817382812,
      "learning_rate": 8.937804845483189e-07,
      "loss": 4.4797,
      "step": 21512
    },
    {
      "epoch": 0.8685102947113443,
      "grad_norm": 86.38613891601562,
      "learning_rate": 8.932402173000498e-07,
      "loss": 3.7829,
      "step": 21513
    },
    {
      "epoch": 0.868550666128381,
      "grad_norm": 167.92776489257812,
      "learning_rate": 8.927001057561102e-07,
      "loss": 6.4754,
      "step": 21514
    },
    {
      "epoch": 0.8685910375454179,
      "grad_norm": 128.2450408935547,
      "learning_rate": 8.921601499257293e-07,
      "loss": 4.061,
      "step": 21515
    },
    {
      "epoch": 0.8686314089624546,
      "grad_norm": 82.81360626220703,
      "learning_rate": 8.916203498181419e-07,
      "loss": 3.8379,
      "step": 21516
    },
    {
      "epoch": 0.8686717803794913,
      "grad_norm": 104.51152038574219,
      "learning_rate": 8.910807054425774e-07,
      "loss": 5.8745,
      "step": 21517
    },
    {
      "epoch": 0.868712151796528,
      "grad_norm": 88.08118438720703,
      "learning_rate": 8.905412168082605e-07,
      "loss": 3.4538,
      "step": 21518
    },
    {
      "epoch": 0.8687525232135648,
      "grad_norm": 74.4432601928711,
      "learning_rate": 8.900018839244184e-07,
      "loss": 5.081,
      "step": 21519
    },
    {
      "epoch": 0.8687928946306015,
      "grad_norm": 117.09091186523438,
      "learning_rate": 8.894627068002693e-07,
      "loss": 5.4515,
      "step": 21520
    },
    {
      "epoch": 0.8688332660476382,
      "grad_norm": 101.27162170410156,
      "learning_rate": 8.889236854450334e-07,
      "loss": 3.6212,
      "step": 21521
    },
    {
      "epoch": 0.868873637464675,
      "grad_norm": 85.11417388916016,
      "learning_rate": 8.883848198679291e-07,
      "loss": 5.5994,
      "step": 21522
    },
    {
      "epoch": 0.8689140088817118,
      "grad_norm": 215.58815002441406,
      "learning_rate": 8.878461100781654e-07,
      "loss": 4.2713,
      "step": 21523
    },
    {
      "epoch": 0.8689543802987485,
      "grad_norm": 88.9636001586914,
      "learning_rate": 8.873075560849542e-07,
      "loss": 4.8613,
      "step": 21524
    },
    {
      "epoch": 0.8689947517157852,
      "grad_norm": 112.9217529296875,
      "learning_rate": 8.867691578975046e-07,
      "loss": 4.829,
      "step": 21525
    },
    {
      "epoch": 0.8690351231328219,
      "grad_norm": 118.20466613769531,
      "learning_rate": 8.862309155250237e-07,
      "loss": 3.9411,
      "step": 21526
    },
    {
      "epoch": 0.8690754945498587,
      "grad_norm": 189.68359375,
      "learning_rate": 8.856928289767097e-07,
      "loss": 5.9514,
      "step": 21527
    },
    {
      "epoch": 0.8691158659668954,
      "grad_norm": 115.59623718261719,
      "learning_rate": 8.851548982617642e-07,
      "loss": 5.9602,
      "step": 21528
    },
    {
      "epoch": 0.8691562373839322,
      "grad_norm": 86.43834686279297,
      "learning_rate": 8.846171233893864e-07,
      "loss": 6.0274,
      "step": 21529
    },
    {
      "epoch": 0.8691966088009689,
      "grad_norm": 122.113525390625,
      "learning_rate": 8.840795043687711e-07,
      "loss": 3.6991,
      "step": 21530
    },
    {
      "epoch": 0.8692369802180057,
      "grad_norm": 65.1045913696289,
      "learning_rate": 8.835420412091067e-07,
      "loss": 3.583,
      "step": 21531
    },
    {
      "epoch": 0.8692773516350424,
      "grad_norm": 120.0614242553711,
      "learning_rate": 8.830047339195857e-07,
      "loss": 5.0504,
      "step": 21532
    },
    {
      "epoch": 0.8693177230520791,
      "grad_norm": 165.04177856445312,
      "learning_rate": 8.824675825093932e-07,
      "loss": 4.2775,
      "step": 21533
    },
    {
      "epoch": 0.8693580944691158,
      "grad_norm": 115.28942108154297,
      "learning_rate": 8.81930586987717e-07,
      "loss": 5.7032,
      "step": 21534
    },
    {
      "epoch": 0.8693984658861525,
      "grad_norm": 220.5915985107422,
      "learning_rate": 8.813937473637335e-07,
      "loss": 4.8095,
      "step": 21535
    },
    {
      "epoch": 0.8694388373031894,
      "grad_norm": 95.11366271972656,
      "learning_rate": 8.808570636466229e-07,
      "loss": 4.9985,
      "step": 21536
    },
    {
      "epoch": 0.8694792087202261,
      "grad_norm": 80.48696899414062,
      "learning_rate": 8.803205358455624e-07,
      "loss": 3.937,
      "step": 21537
    },
    {
      "epoch": 0.8695195801372628,
      "grad_norm": 97.37828826904297,
      "learning_rate": 8.797841639697235e-07,
      "loss": 5.2528,
      "step": 21538
    },
    {
      "epoch": 0.8695599515542995,
      "grad_norm": 98.76175689697266,
      "learning_rate": 8.79247948028279e-07,
      "loss": 5.4584,
      "step": 21539
    },
    {
      "epoch": 0.8696003229713363,
      "grad_norm": 121.97811889648438,
      "learning_rate": 8.787118880303968e-07,
      "loss": 4.805,
      "step": 21540
    },
    {
      "epoch": 0.869640694388373,
      "grad_norm": 105.912841796875,
      "learning_rate": 8.78175983985241e-07,
      "loss": 4.4355,
      "step": 21541
    },
    {
      "epoch": 0.8696810658054098,
      "grad_norm": 89.23114776611328,
      "learning_rate": 8.776402359019764e-07,
      "loss": 3.9737,
      "step": 21542
    },
    {
      "epoch": 0.8697214372224465,
      "grad_norm": 104.2757568359375,
      "learning_rate": 8.771046437897601e-07,
      "loss": 5.7653,
      "step": 21543
    },
    {
      "epoch": 0.8697618086394833,
      "grad_norm": 97.53553771972656,
      "learning_rate": 8.765692076577503e-07,
      "loss": 4.8258,
      "step": 21544
    },
    {
      "epoch": 0.86980218005652,
      "grad_norm": 79.07793426513672,
      "learning_rate": 8.760339275151031e-07,
      "loss": 5.3684,
      "step": 21545
    },
    {
      "epoch": 0.8698425514735567,
      "grad_norm": 146.13270568847656,
      "learning_rate": 8.75498803370971e-07,
      "loss": 4.6462,
      "step": 21546
    },
    {
      "epoch": 0.8698829228905934,
      "grad_norm": 105.34465026855469,
      "learning_rate": 8.749638352345002e-07,
      "loss": 4.3498,
      "step": 21547
    },
    {
      "epoch": 0.8699232943076302,
      "grad_norm": 123.79653930664062,
      "learning_rate": 8.7442902311484e-07,
      "loss": 5.1493,
      "step": 21548
    },
    {
      "epoch": 0.869963665724667,
      "grad_norm": 74.3729019165039,
      "learning_rate": 8.738943670211331e-07,
      "loss": 3.8348,
      "step": 21549
    },
    {
      "epoch": 0.8700040371417037,
      "grad_norm": 76.36542510986328,
      "learning_rate": 8.733598669625243e-07,
      "loss": 6.5861,
      "step": 21550
    },
    {
      "epoch": 0.8700444085587404,
      "grad_norm": 91.29656219482422,
      "learning_rate": 8.728255229481464e-07,
      "loss": 5.4349,
      "step": 21551
    },
    {
      "epoch": 0.8700847799757772,
      "grad_norm": 99.97618103027344,
      "learning_rate": 8.722913349871398e-07,
      "loss": 4.8861,
      "step": 21552
    },
    {
      "epoch": 0.8701251513928139,
      "grad_norm": 116.21635437011719,
      "learning_rate": 8.717573030886362e-07,
      "loss": 5.5051,
      "step": 21553
    },
    {
      "epoch": 0.8701655228098506,
      "grad_norm": 119.38756561279297,
      "learning_rate": 8.712234272617681e-07,
      "loss": 4.6827,
      "step": 21554
    },
    {
      "epoch": 0.8702058942268873,
      "grad_norm": 76.13774108886719,
      "learning_rate": 8.706897075156607e-07,
      "loss": 4.3463,
      "step": 21555
    },
    {
      "epoch": 0.8702462656439242,
      "grad_norm": 88.59832763671875,
      "learning_rate": 8.701561438594408e-07,
      "loss": 5.1565,
      "step": 21556
    },
    {
      "epoch": 0.8702866370609609,
      "grad_norm": 82.6231460571289,
      "learning_rate": 8.696227363022314e-07,
      "loss": 5.3303,
      "step": 21557
    },
    {
      "epoch": 0.8703270084779976,
      "grad_norm": 74.18130493164062,
      "learning_rate": 8.690894848531517e-07,
      "loss": 4.5498,
      "step": 21558
    },
    {
      "epoch": 0.8703673798950343,
      "grad_norm": 90.67101287841797,
      "learning_rate": 8.6855638952132e-07,
      "loss": 4.8854,
      "step": 21559
    },
    {
      "epoch": 0.870407751312071,
      "grad_norm": 85.2548599243164,
      "learning_rate": 8.680234503158503e-07,
      "loss": 4.7261,
      "step": 21560
    },
    {
      "epoch": 0.8704481227291078,
      "grad_norm": 88.47227478027344,
      "learning_rate": 8.67490667245855e-07,
      "loss": 2.9062,
      "step": 21561
    },
    {
      "epoch": 0.8704884941461445,
      "grad_norm": 78.64070129394531,
      "learning_rate": 8.669580403204459e-07,
      "loss": 2.8085,
      "step": 21562
    },
    {
      "epoch": 0.8705288655631813,
      "grad_norm": 89.73379516601562,
      "learning_rate": 8.664255695487244e-07,
      "loss": 5.1425,
      "step": 21563
    },
    {
      "epoch": 0.870569236980218,
      "grad_norm": 112.36111450195312,
      "learning_rate": 8.65893254939797e-07,
      "loss": 3.8877,
      "step": 21564
    },
    {
      "epoch": 0.8706096083972548,
      "grad_norm": 128.2533721923828,
      "learning_rate": 8.653610965027658e-07,
      "loss": 5.6518,
      "step": 21565
    },
    {
      "epoch": 0.8706499798142915,
      "grad_norm": 137.05882263183594,
      "learning_rate": 8.648290942467308e-07,
      "loss": 4.4084,
      "step": 21566
    },
    {
      "epoch": 0.8706903512313282,
      "grad_norm": 91.13420867919922,
      "learning_rate": 8.642972481807844e-07,
      "loss": 4.943,
      "step": 21567
    },
    {
      "epoch": 0.8707307226483649,
      "grad_norm": 107.63934326171875,
      "learning_rate": 8.637655583140215e-07,
      "loss": 4.8065,
      "step": 21568
    },
    {
      "epoch": 0.8707710940654017,
      "grad_norm": 79.9874267578125,
      "learning_rate": 8.632340246555326e-07,
      "loss": 4.9961,
      "step": 21569
    },
    {
      "epoch": 0.8708114654824385,
      "grad_norm": 260.9300537109375,
      "learning_rate": 8.627026472144085e-07,
      "loss": 5.3884,
      "step": 21570
    },
    {
      "epoch": 0.8708518368994752,
      "grad_norm": 82.10203552246094,
      "learning_rate": 8.621714259997294e-07,
      "loss": 4.5099,
      "step": 21571
    },
    {
      "epoch": 0.8708922083165119,
      "grad_norm": 94.43180847167969,
      "learning_rate": 8.616403610205814e-07,
      "loss": 5.6501,
      "step": 21572
    },
    {
      "epoch": 0.8709325797335487,
      "grad_norm": 80.62677764892578,
      "learning_rate": 8.61109452286043e-07,
      "loss": 3.0629,
      "step": 21573
    },
    {
      "epoch": 0.8709729511505854,
      "grad_norm": 115.8067398071289,
      "learning_rate": 8.605786998051924e-07,
      "loss": 5.797,
      "step": 21574
    },
    {
      "epoch": 0.8710133225676221,
      "grad_norm": 131.471923828125,
      "learning_rate": 8.600481035871033e-07,
      "loss": 6.8854,
      "step": 21575
    },
    {
      "epoch": 0.8710536939846588,
      "grad_norm": 107.18645477294922,
      "learning_rate": 8.595176636408498e-07,
      "loss": 6.0637,
      "step": 21576
    },
    {
      "epoch": 0.8710940654016956,
      "grad_norm": 107.75466918945312,
      "learning_rate": 8.589873799754978e-07,
      "loss": 4.292,
      "step": 21577
    },
    {
      "epoch": 0.8711344368187324,
      "grad_norm": 100.82649993896484,
      "learning_rate": 8.584572526001189e-07,
      "loss": 4.6407,
      "step": 21578
    },
    {
      "epoch": 0.8711748082357691,
      "grad_norm": 102.59514617919922,
      "learning_rate": 8.579272815237715e-07,
      "loss": 5.4175,
      "step": 21579
    },
    {
      "epoch": 0.8712151796528058,
      "grad_norm": 137.4250946044922,
      "learning_rate": 8.573974667555196e-07,
      "loss": 4.3038,
      "step": 21580
    },
    {
      "epoch": 0.8712555510698425,
      "grad_norm": 115.6281509399414,
      "learning_rate": 8.568678083044213e-07,
      "loss": 4.4083,
      "step": 21581
    },
    {
      "epoch": 0.8712959224868793,
      "grad_norm": 61.45205307006836,
      "learning_rate": 8.56338306179535e-07,
      "loss": 2.8987,
      "step": 21582
    },
    {
      "epoch": 0.871336293903916,
      "grad_norm": 138.12136840820312,
      "learning_rate": 8.5580896038991e-07,
      "loss": 3.1296,
      "step": 21583
    },
    {
      "epoch": 0.8713766653209528,
      "grad_norm": 86.1984634399414,
      "learning_rate": 8.552797709445981e-07,
      "loss": 6.8408,
      "step": 21584
    },
    {
      "epoch": 0.8714170367379895,
      "grad_norm": 114.9237289428711,
      "learning_rate": 8.547507378526476e-07,
      "loss": 5.6843,
      "step": 21585
    },
    {
      "epoch": 0.8714574081550263,
      "grad_norm": 115.04891204833984,
      "learning_rate": 8.542218611231057e-07,
      "loss": 5.1323,
      "step": 21586
    },
    {
      "epoch": 0.871497779572063,
      "grad_norm": 96.53380584716797,
      "learning_rate": 8.536931407650118e-07,
      "loss": 4.4455,
      "step": 21587
    },
    {
      "epoch": 0.8715381509890997,
      "grad_norm": 98.58303833007812,
      "learning_rate": 8.531645767874064e-07,
      "loss": 5.6201,
      "step": 21588
    },
    {
      "epoch": 0.8715785224061364,
      "grad_norm": 73.01985168457031,
      "learning_rate": 8.526361691993268e-07,
      "loss": 4.0118,
      "step": 21589
    },
    {
      "epoch": 0.8716188938231731,
      "grad_norm": 93.5145492553711,
      "learning_rate": 8.521079180098102e-07,
      "loss": 4.6441,
      "step": 21590
    },
    {
      "epoch": 0.87165926524021,
      "grad_norm": 120.93038940429688,
      "learning_rate": 8.515798232278849e-07,
      "loss": 5.737,
      "step": 21591
    },
    {
      "epoch": 0.8716996366572467,
      "grad_norm": 82.07840728759766,
      "learning_rate": 8.510518848625816e-07,
      "loss": 3.6818,
      "step": 21592
    },
    {
      "epoch": 0.8717400080742834,
      "grad_norm": 111.73517608642578,
      "learning_rate": 8.505241029229272e-07,
      "loss": 4.0659,
      "step": 21593
    },
    {
      "epoch": 0.8717803794913201,
      "grad_norm": 89.81424713134766,
      "learning_rate": 8.499964774179437e-07,
      "loss": 4.3158,
      "step": 21594
    },
    {
      "epoch": 0.8718207509083569,
      "grad_norm": 65.21111297607422,
      "learning_rate": 8.494690083566548e-07,
      "loss": 5.3234,
      "step": 21595
    },
    {
      "epoch": 0.8718611223253936,
      "grad_norm": 95.872314453125,
      "learning_rate": 8.489416957480779e-07,
      "loss": 5.7789,
      "step": 21596
    },
    {
      "epoch": 0.8719014937424303,
      "grad_norm": 103.30371856689453,
      "learning_rate": 8.484145396012278e-07,
      "loss": 4.4655,
      "step": 21597
    },
    {
      "epoch": 0.8719418651594671,
      "grad_norm": 81.21953582763672,
      "learning_rate": 8.478875399251208e-07,
      "loss": 4.698,
      "step": 21598
    },
    {
      "epoch": 0.8719822365765039,
      "grad_norm": 102.03661346435547,
      "learning_rate": 8.473606967287639e-07,
      "loss": 5.83,
      "step": 21599
    },
    {
      "epoch": 0.8720226079935406,
      "grad_norm": 291.6111145019531,
      "learning_rate": 8.468340100211659e-07,
      "loss": 6.2537,
      "step": 21600
    },
    {
      "epoch": 0.8720629794105773,
      "grad_norm": 597.5225830078125,
      "learning_rate": 8.463074798113313e-07,
      "loss": 3.4283,
      "step": 21601
    },
    {
      "epoch": 0.872103350827614,
      "grad_norm": 93.99083709716797,
      "learning_rate": 8.457811061082666e-07,
      "loss": 3.468,
      "step": 21602
    },
    {
      "epoch": 0.8721437222446508,
      "grad_norm": 120.74433898925781,
      "learning_rate": 8.452548889209655e-07,
      "loss": 4.8256,
      "step": 21603
    },
    {
      "epoch": 0.8721840936616875,
      "grad_norm": 96.21558380126953,
      "learning_rate": 8.447288282584276e-07,
      "loss": 3.8986,
      "step": 21604
    },
    {
      "epoch": 0.8722244650787243,
      "grad_norm": 90.78890228271484,
      "learning_rate": 8.44202924129649e-07,
      "loss": 3.9557,
      "step": 21605
    },
    {
      "epoch": 0.872264836495761,
      "grad_norm": 99.72294616699219,
      "learning_rate": 8.436771765436202e-07,
      "loss": 3.3401,
      "step": 21606
    },
    {
      "epoch": 0.8723052079127978,
      "grad_norm": 124.69793701171875,
      "learning_rate": 8.431515855093298e-07,
      "loss": 4.9021,
      "step": 21607
    },
    {
      "epoch": 0.8723455793298345,
      "grad_norm": 78.18077850341797,
      "learning_rate": 8.426261510357636e-07,
      "loss": 2.7878,
      "step": 21608
    },
    {
      "epoch": 0.8723859507468712,
      "grad_norm": 82.29708862304688,
      "learning_rate": 8.421008731319069e-07,
      "loss": 4.3906,
      "step": 21609
    },
    {
      "epoch": 0.8724263221639079,
      "grad_norm": 92.83751678466797,
      "learning_rate": 8.415757518067413e-07,
      "loss": 5.4679,
      "step": 21610
    },
    {
      "epoch": 0.8724666935809446,
      "grad_norm": 139.9375457763672,
      "learning_rate": 8.41050787069242e-07,
      "loss": 4.1898,
      "step": 21611
    },
    {
      "epoch": 0.8725070649979815,
      "grad_norm": 87.0579605102539,
      "learning_rate": 8.40525978928387e-07,
      "loss": 5.8867,
      "step": 21612
    },
    {
      "epoch": 0.8725474364150182,
      "grad_norm": 99.35382080078125,
      "learning_rate": 8.400013273931484e-07,
      "loss": 5.9785,
      "step": 21613
    },
    {
      "epoch": 0.8725878078320549,
      "grad_norm": 125.46144104003906,
      "learning_rate": 8.394768324724977e-07,
      "loss": 3.7223,
      "step": 21614
    },
    {
      "epoch": 0.8726281792490916,
      "grad_norm": 70.83985137939453,
      "learning_rate": 8.389524941754012e-07,
      "loss": 3.5017,
      "step": 21615
    },
    {
      "epoch": 0.8726685506661284,
      "grad_norm": 73.35834503173828,
      "learning_rate": 8.38428312510825e-07,
      "loss": 4.5635,
      "step": 21616
    },
    {
      "epoch": 0.8727089220831651,
      "grad_norm": 107.62016296386719,
      "learning_rate": 8.379042874877308e-07,
      "loss": 5.0146,
      "step": 21617
    },
    {
      "epoch": 0.8727492935002018,
      "grad_norm": 83.07213592529297,
      "learning_rate": 8.373804191150792e-07,
      "loss": 4.7164,
      "step": 21618
    },
    {
      "epoch": 0.8727896649172386,
      "grad_norm": 80.50506591796875,
      "learning_rate": 8.368567074018252e-07,
      "loss": 5.2051,
      "step": 21619
    },
    {
      "epoch": 0.8728300363342754,
      "grad_norm": 90.9854507446289,
      "learning_rate": 8.36333152356924e-07,
      "loss": 4.1057,
      "step": 21620
    },
    {
      "epoch": 0.8728704077513121,
      "grad_norm": 77.9693832397461,
      "learning_rate": 8.358097539893261e-07,
      "loss": 3.2126,
      "step": 21621
    },
    {
      "epoch": 0.8729107791683488,
      "grad_norm": 91.09614562988281,
      "learning_rate": 8.352865123079845e-07,
      "loss": 4.5056,
      "step": 21622
    },
    {
      "epoch": 0.8729511505853855,
      "grad_norm": 69.4326171875,
      "learning_rate": 8.347634273218397e-07,
      "loss": 5.3237,
      "step": 21623
    },
    {
      "epoch": 0.8729915220024222,
      "grad_norm": 88.02821350097656,
      "learning_rate": 8.34240499039839e-07,
      "loss": 4.1645,
      "step": 21624
    },
    {
      "epoch": 0.873031893419459,
      "grad_norm": 94.50605773925781,
      "learning_rate": 8.337177274709218e-07,
      "loss": 4.7463,
      "step": 21625
    },
    {
      "epoch": 0.8730722648364958,
      "grad_norm": 150.1566925048828,
      "learning_rate": 8.331951126240279e-07,
      "loss": 3.658,
      "step": 21626
    },
    {
      "epoch": 0.8731126362535325,
      "grad_norm": 74.6488037109375,
      "learning_rate": 8.326726545080899e-07,
      "loss": 4.5582,
      "step": 21627
    },
    {
      "epoch": 0.8731530076705692,
      "grad_norm": 119.95555877685547,
      "learning_rate": 8.32150353132043e-07,
      "loss": 3.3896,
      "step": 21628
    },
    {
      "epoch": 0.873193379087606,
      "grad_norm": 110.69661712646484,
      "learning_rate": 8.316282085048155e-07,
      "loss": 3.871,
      "step": 21629
    },
    {
      "epoch": 0.8732337505046427,
      "grad_norm": 108.34918212890625,
      "learning_rate": 8.311062206353382e-07,
      "loss": 5.6562,
      "step": 21630
    },
    {
      "epoch": 0.8732741219216794,
      "grad_norm": 110.37028503417969,
      "learning_rate": 8.305843895325327e-07,
      "loss": 6.8862,
      "step": 21631
    },
    {
      "epoch": 0.8733144933387161,
      "grad_norm": 100.47962188720703,
      "learning_rate": 8.300627152053209e-07,
      "loss": 4.0295,
      "step": 21632
    },
    {
      "epoch": 0.873354864755753,
      "grad_norm": 100.06324768066406,
      "learning_rate": 8.295411976626233e-07,
      "loss": 5.3629,
      "step": 21633
    },
    {
      "epoch": 0.8733952361727897,
      "grad_norm": 101.98625183105469,
      "learning_rate": 8.290198369133573e-07,
      "loss": 6.4104,
      "step": 21634
    },
    {
      "epoch": 0.8734356075898264,
      "grad_norm": 84.2477798461914,
      "learning_rate": 8.284986329664357e-07,
      "loss": 4.0814,
      "step": 21635
    },
    {
      "epoch": 0.8734759790068631,
      "grad_norm": 84.6027603149414,
      "learning_rate": 8.279775858307715e-07,
      "loss": 5.4865,
      "step": 21636
    },
    {
      "epoch": 0.8735163504238999,
      "grad_norm": 70.48814392089844,
      "learning_rate": 8.274566955152719e-07,
      "loss": 4.4576,
      "step": 21637
    },
    {
      "epoch": 0.8735567218409366,
      "grad_norm": 92.00206756591797,
      "learning_rate": 8.269359620288442e-07,
      "loss": 5.7442,
      "step": 21638
    },
    {
      "epoch": 0.8735970932579733,
      "grad_norm": 134.2718505859375,
      "learning_rate": 8.264153853803903e-07,
      "loss": 6.4234,
      "step": 21639
    },
    {
      "epoch": 0.8736374646750101,
      "grad_norm": 95.6432113647461,
      "learning_rate": 8.258949655788107e-07,
      "loss": 4.7864,
      "step": 21640
    },
    {
      "epoch": 0.8736778360920469,
      "grad_norm": 124.95325469970703,
      "learning_rate": 8.25374702633005e-07,
      "loss": 4.5797,
      "step": 21641
    },
    {
      "epoch": 0.8737182075090836,
      "grad_norm": 77.87784576416016,
      "learning_rate": 8.248545965518684e-07,
      "loss": 3.6946,
      "step": 21642
    },
    {
      "epoch": 0.8737585789261203,
      "grad_norm": 101.97937774658203,
      "learning_rate": 8.243346473442915e-07,
      "loss": 5.1488,
      "step": 21643
    },
    {
      "epoch": 0.873798950343157,
      "grad_norm": 101.29225158691406,
      "learning_rate": 8.238148550191649e-07,
      "loss": 5.195,
      "step": 21644
    },
    {
      "epoch": 0.8738393217601937,
      "grad_norm": 127.29335021972656,
      "learning_rate": 8.232952195853772e-07,
      "loss": 4.7377,
      "step": 21645
    },
    {
      "epoch": 0.8738796931772305,
      "grad_norm": 80.8647232055664,
      "learning_rate": 8.227757410518134e-07,
      "loss": 5.7828,
      "step": 21646
    },
    {
      "epoch": 0.8739200645942673,
      "grad_norm": 146.4538116455078,
      "learning_rate": 8.222564194273519e-07,
      "loss": 5.6817,
      "step": 21647
    },
    {
      "epoch": 0.873960436011304,
      "grad_norm": 77.68527221679688,
      "learning_rate": 8.217372547208746e-07,
      "loss": 5.9877,
      "step": 21648
    },
    {
      "epoch": 0.8740008074283407,
      "grad_norm": 125.8057632446289,
      "learning_rate": 8.212182469412567e-07,
      "loss": 4.6567,
      "step": 21649
    },
    {
      "epoch": 0.8740411788453775,
      "grad_norm": 122.48533630371094,
      "learning_rate": 8.206993960973753e-07,
      "loss": 4.15,
      "step": 21650
    },
    {
      "epoch": 0.8740815502624142,
      "grad_norm": 66.14053344726562,
      "learning_rate": 8.201807021980967e-07,
      "loss": 5.6809,
      "step": 21651
    },
    {
      "epoch": 0.8741219216794509,
      "grad_norm": 93.45648956298828,
      "learning_rate": 8.196621652522929e-07,
      "loss": 4.5258,
      "step": 21652
    },
    {
      "epoch": 0.8741622930964877,
      "grad_norm": 85.46748352050781,
      "learning_rate": 8.191437852688278e-07,
      "loss": 4.316,
      "step": 21653
    },
    {
      "epoch": 0.8742026645135245,
      "grad_norm": 87.18309783935547,
      "learning_rate": 8.186255622565642e-07,
      "loss": 6.0321,
      "step": 21654
    },
    {
      "epoch": 0.8742430359305612,
      "grad_norm": 99.02711486816406,
      "learning_rate": 8.18107496224364e-07,
      "loss": 5.6317,
      "step": 21655
    },
    {
      "epoch": 0.8742834073475979,
      "grad_norm": 71.3347396850586,
      "learning_rate": 8.175895871810846e-07,
      "loss": 5.0349,
      "step": 21656
    },
    {
      "epoch": 0.8743237787646346,
      "grad_norm": 138.50827026367188,
      "learning_rate": 8.17071835135581e-07,
      "loss": 3.7589,
      "step": 21657
    },
    {
      "epoch": 0.8743641501816714,
      "grad_norm": 157.33912658691406,
      "learning_rate": 8.165542400967064e-07,
      "loss": 4.9508,
      "step": 21658
    },
    {
      "epoch": 0.8744045215987081,
      "grad_norm": 77.82861328125,
      "learning_rate": 8.160368020733089e-07,
      "loss": 3.4332,
      "step": 21659
    },
    {
      "epoch": 0.8744448930157449,
      "grad_norm": 79.95885467529297,
      "learning_rate": 8.155195210742351e-07,
      "loss": 5.4776,
      "step": 21660
    },
    {
      "epoch": 0.8744852644327816,
      "grad_norm": 135.37351989746094,
      "learning_rate": 8.150023971083298e-07,
      "loss": 6.9469,
      "step": 21661
    },
    {
      "epoch": 0.8745256358498184,
      "grad_norm": 118.07852172851562,
      "learning_rate": 8.144854301844374e-07,
      "loss": 4.5778,
      "step": 21662
    },
    {
      "epoch": 0.8745660072668551,
      "grad_norm": 102.06261444091797,
      "learning_rate": 8.139686203113917e-07,
      "loss": 3.6092,
      "step": 21663
    },
    {
      "epoch": 0.8746063786838918,
      "grad_norm": 102.80125427246094,
      "learning_rate": 8.134519674980324e-07,
      "loss": 4.8358,
      "step": 21664
    },
    {
      "epoch": 0.8746467501009285,
      "grad_norm": 84.96939849853516,
      "learning_rate": 8.129354717531923e-07,
      "loss": 4.2678,
      "step": 21665
    },
    {
      "epoch": 0.8746871215179652,
      "grad_norm": 142.8640594482422,
      "learning_rate": 8.124191330857045e-07,
      "loss": 4.803,
      "step": 21666
    },
    {
      "epoch": 0.8747274929350021,
      "grad_norm": 142.760498046875,
      "learning_rate": 8.11902951504393e-07,
      "loss": 5.1832,
      "step": 21667
    },
    {
      "epoch": 0.8747678643520388,
      "grad_norm": 84.50106811523438,
      "learning_rate": 8.113869270180852e-07,
      "loss": 5.2379,
      "step": 21668
    },
    {
      "epoch": 0.8748082357690755,
      "grad_norm": 168.08834838867188,
      "learning_rate": 8.10871059635604e-07,
      "loss": 6.9429,
      "step": 21669
    },
    {
      "epoch": 0.8748486071861122,
      "grad_norm": 80.8226547241211,
      "learning_rate": 8.103553493657712e-07,
      "loss": 4.2405,
      "step": 21670
    },
    {
      "epoch": 0.874888978603149,
      "grad_norm": 80.21951293945312,
      "learning_rate": 8.09839796217401e-07,
      "loss": 5.2018,
      "step": 21671
    },
    {
      "epoch": 0.8749293500201857,
      "grad_norm": 102.91728210449219,
      "learning_rate": 8.093244001993084e-07,
      "loss": 4.6903,
      "step": 21672
    },
    {
      "epoch": 0.8749697214372224,
      "grad_norm": 356.77325439453125,
      "learning_rate": 8.088091613203087e-07,
      "loss": 4.7234,
      "step": 21673
    },
    {
      "epoch": 0.8750100928542592,
      "grad_norm": 77.22355651855469,
      "learning_rate": 8.08294079589208e-07,
      "loss": 5.0785,
      "step": 21674
    },
    {
      "epoch": 0.875050464271296,
      "grad_norm": 105.41425323486328,
      "learning_rate": 8.07779155014815e-07,
      "loss": 4.4642,
      "step": 21675
    },
    {
      "epoch": 0.8750908356883327,
      "grad_norm": 108.27706909179688,
      "learning_rate": 8.072643876059327e-07,
      "loss": 4.7034,
      "step": 21676
    },
    {
      "epoch": 0.8751312071053694,
      "grad_norm": 119.47718811035156,
      "learning_rate": 8.06749777371364e-07,
      "loss": 3.7961,
      "step": 21677
    },
    {
      "epoch": 0.8751715785224061,
      "grad_norm": 89.87022399902344,
      "learning_rate": 8.062353243199062e-07,
      "loss": 5.8496,
      "step": 21678
    },
    {
      "epoch": 0.8752119499394428,
      "grad_norm": 115.4216079711914,
      "learning_rate": 8.057210284603545e-07,
      "loss": 6.3609,
      "step": 21679
    },
    {
      "epoch": 0.8752523213564796,
      "grad_norm": 107.56219482421875,
      "learning_rate": 8.052068898015031e-07,
      "loss": 6.2264,
      "step": 21680
    },
    {
      "epoch": 0.8752926927735164,
      "grad_norm": 98.02459716796875,
      "learning_rate": 8.046929083521426e-07,
      "loss": 5.4967,
      "step": 21681
    },
    {
      "epoch": 0.8753330641905531,
      "grad_norm": 173.89697265625,
      "learning_rate": 8.041790841210628e-07,
      "loss": 4.5986,
      "step": 21682
    },
    {
      "epoch": 0.8753734356075898,
      "grad_norm": 84.85148620605469,
      "learning_rate": 8.036654171170443e-07,
      "loss": 3.4587,
      "step": 21683
    },
    {
      "epoch": 0.8754138070246266,
      "grad_norm": 94.7029037475586,
      "learning_rate": 8.031519073488725e-07,
      "loss": 2.5002,
      "step": 21684
    },
    {
      "epoch": 0.8754541784416633,
      "grad_norm": 104.35051727294922,
      "learning_rate": 8.02638554825328e-07,
      "loss": 4.6779,
      "step": 21685
    },
    {
      "epoch": 0.8754945498587,
      "grad_norm": 80.74030303955078,
      "learning_rate": 8.02125359555187e-07,
      "loss": 6.0415,
      "step": 21686
    },
    {
      "epoch": 0.8755349212757367,
      "grad_norm": 180.90170288085938,
      "learning_rate": 8.016123215472227e-07,
      "loss": 5.7029,
      "step": 21687
    },
    {
      "epoch": 0.8755752926927736,
      "grad_norm": 116.20418548583984,
      "learning_rate": 8.01099440810208e-07,
      "loss": 4.1888,
      "step": 21688
    },
    {
      "epoch": 0.8756156641098103,
      "grad_norm": 88.6954574584961,
      "learning_rate": 8.005867173529114e-07,
      "loss": 4.8469,
      "step": 21689
    },
    {
      "epoch": 0.875656035526847,
      "grad_norm": 94.33956146240234,
      "learning_rate": 8.000741511841026e-07,
      "loss": 3.9082,
      "step": 21690
    },
    {
      "epoch": 0.8756964069438837,
      "grad_norm": 90.12859344482422,
      "learning_rate": 7.99561742312539e-07,
      "loss": 3.4601,
      "step": 21691
    },
    {
      "epoch": 0.8757367783609205,
      "grad_norm": 82.33692932128906,
      "learning_rate": 7.990494907469859e-07,
      "loss": 4.7761,
      "step": 21692
    },
    {
      "epoch": 0.8757771497779572,
      "grad_norm": 135.3341827392578,
      "learning_rate": 7.985373964962017e-07,
      "loss": 2.5599,
      "step": 21693
    },
    {
      "epoch": 0.8758175211949939,
      "grad_norm": 109.62201690673828,
      "learning_rate": 7.980254595689396e-07,
      "loss": 6.0308,
      "step": 21694
    },
    {
      "epoch": 0.8758578926120307,
      "grad_norm": 143.150634765625,
      "learning_rate": 7.975136799739546e-07,
      "loss": 3.9726,
      "step": 21695
    },
    {
      "epoch": 0.8758982640290675,
      "grad_norm": 137.88839721679688,
      "learning_rate": 7.970020577199966e-07,
      "loss": 5.3381,
      "step": 21696
    },
    {
      "epoch": 0.8759386354461042,
      "grad_norm": 132.67184448242188,
      "learning_rate": 7.964905928158129e-07,
      "loss": 5.4754,
      "step": 21697
    },
    {
      "epoch": 0.8759790068631409,
      "grad_norm": 88.22350311279297,
      "learning_rate": 7.959792852701487e-07,
      "loss": 3.9747,
      "step": 21698
    },
    {
      "epoch": 0.8760193782801776,
      "grad_norm": 119.92843627929688,
      "learning_rate": 7.954681350917459e-07,
      "loss": 4.2597,
      "step": 21699
    },
    {
      "epoch": 0.8760597496972143,
      "grad_norm": 104.86431121826172,
      "learning_rate": 7.949571422893431e-07,
      "loss": 4.312,
      "step": 21700
    },
    {
      "epoch": 0.8761001211142511,
      "grad_norm": 90.16465759277344,
      "learning_rate": 7.944463068716779e-07,
      "loss": 7.7541,
      "step": 21701
    },
    {
      "epoch": 0.8761404925312879,
      "grad_norm": 89.44022369384766,
      "learning_rate": 7.939356288474854e-07,
      "loss": 4.9897,
      "step": 21702
    },
    {
      "epoch": 0.8761808639483246,
      "grad_norm": 214.24253845214844,
      "learning_rate": 7.934251082254951e-07,
      "loss": 5.0463,
      "step": 21703
    },
    {
      "epoch": 0.8762212353653613,
      "grad_norm": 163.152099609375,
      "learning_rate": 7.929147450144359e-07,
      "loss": 4.6905,
      "step": 21704
    },
    {
      "epoch": 0.8762616067823981,
      "grad_norm": 96.40087890625,
      "learning_rate": 7.924045392230351e-07,
      "loss": 4.7236,
      "step": 21705
    },
    {
      "epoch": 0.8763019781994348,
      "grad_norm": 169.29898071289062,
      "learning_rate": 7.918944908600168e-07,
      "loss": 5.8478,
      "step": 21706
    },
    {
      "epoch": 0.8763423496164715,
      "grad_norm": 110.12230682373047,
      "learning_rate": 7.913845999340985e-07,
      "loss": 4.4027,
      "step": 21707
    },
    {
      "epoch": 0.8763827210335082,
      "grad_norm": 137.93971252441406,
      "learning_rate": 7.90874866454e-07,
      "loss": 5.8803,
      "step": 21708
    },
    {
      "epoch": 0.8764230924505451,
      "grad_norm": 84.6080322265625,
      "learning_rate": 7.903652904284354e-07,
      "loss": 5.6257,
      "step": 21709
    },
    {
      "epoch": 0.8764634638675818,
      "grad_norm": 128.4032745361328,
      "learning_rate": 7.898558718661209e-07,
      "loss": 3.4545,
      "step": 21710
    },
    {
      "epoch": 0.8765038352846185,
      "grad_norm": 86.41111755371094,
      "learning_rate": 7.893466107757631e-07,
      "loss": 4.434,
      "step": 21711
    },
    {
      "epoch": 0.8765442067016552,
      "grad_norm": 106.50602722167969,
      "learning_rate": 7.888375071660681e-07,
      "loss": 2.3575,
      "step": 21712
    },
    {
      "epoch": 0.876584578118692,
      "grad_norm": 407.7773742675781,
      "learning_rate": 7.883285610457425e-07,
      "loss": 4.0991,
      "step": 21713
    },
    {
      "epoch": 0.8766249495357287,
      "grad_norm": 134.88636779785156,
      "learning_rate": 7.878197724234881e-07,
      "loss": 5.8352,
      "step": 21714
    },
    {
      "epoch": 0.8766653209527654,
      "grad_norm": 89.05345153808594,
      "learning_rate": 7.873111413080037e-07,
      "loss": 5.7187,
      "step": 21715
    },
    {
      "epoch": 0.8767056923698022,
      "grad_norm": 102.25666809082031,
      "learning_rate": 7.868026677079854e-07,
      "loss": 4.4218,
      "step": 21716
    },
    {
      "epoch": 0.876746063786839,
      "grad_norm": 96.04570770263672,
      "learning_rate": 7.862943516321275e-07,
      "loss": 4.8891,
      "step": 21717
    },
    {
      "epoch": 0.8767864352038757,
      "grad_norm": 107.44122314453125,
      "learning_rate": 7.85786193089122e-07,
      "loss": 4.8399,
      "step": 21718
    },
    {
      "epoch": 0.8768268066209124,
      "grad_norm": 73.0683822631836,
      "learning_rate": 7.852781920876539e-07,
      "loss": 5.7265,
      "step": 21719
    },
    {
      "epoch": 0.8768671780379491,
      "grad_norm": 78.09314727783203,
      "learning_rate": 7.847703486364111e-07,
      "loss": 4.4339,
      "step": 21720
    },
    {
      "epoch": 0.8769075494549858,
      "grad_norm": 113.18419647216797,
      "learning_rate": 7.842626627440763e-07,
      "loss": 6.2139,
      "step": 21721
    },
    {
      "epoch": 0.8769479208720226,
      "grad_norm": 108.58734893798828,
      "learning_rate": 7.837551344193317e-07,
      "loss": 4.4447,
      "step": 21722
    },
    {
      "epoch": 0.8769882922890594,
      "grad_norm": 90.61760711669922,
      "learning_rate": 7.832477636708502e-07,
      "loss": 3.4861,
      "step": 21723
    },
    {
      "epoch": 0.8770286637060961,
      "grad_norm": 87.418701171875,
      "learning_rate": 7.827405505073094e-07,
      "loss": 3.753,
      "step": 21724
    },
    {
      "epoch": 0.8770690351231328,
      "grad_norm": 89.89160919189453,
      "learning_rate": 7.822334949373822e-07,
      "loss": 3.5695,
      "step": 21725
    },
    {
      "epoch": 0.8771094065401696,
      "grad_norm": 98.58895874023438,
      "learning_rate": 7.817265969697385e-07,
      "loss": 5.147,
      "step": 21726
    },
    {
      "epoch": 0.8771497779572063,
      "grad_norm": 102.03216552734375,
      "learning_rate": 7.812198566130413e-07,
      "loss": 2.8986,
      "step": 21727
    },
    {
      "epoch": 0.877190149374243,
      "grad_norm": 80.9366226196289,
      "learning_rate": 7.807132738759581e-07,
      "loss": 6.73,
      "step": 21728
    },
    {
      "epoch": 0.8772305207912797,
      "grad_norm": 91.56088256835938,
      "learning_rate": 7.802068487671488e-07,
      "loss": 4.1195,
      "step": 21729
    },
    {
      "epoch": 0.8772708922083166,
      "grad_norm": 111.54931640625,
      "learning_rate": 7.797005812952752e-07,
      "loss": 3.8779,
      "step": 21730
    },
    {
      "epoch": 0.8773112636253533,
      "grad_norm": 95.58714294433594,
      "learning_rate": 7.791944714689881e-07,
      "loss": 4.4139,
      "step": 21731
    },
    {
      "epoch": 0.87735163504239,
      "grad_norm": 180.14065551757812,
      "learning_rate": 7.78688519296944e-07,
      "loss": 2.8463,
      "step": 21732
    },
    {
      "epoch": 0.8773920064594267,
      "grad_norm": 183.36166381835938,
      "learning_rate": 7.781827247877927e-07,
      "loss": 4.5569,
      "step": 21733
    },
    {
      "epoch": 0.8774323778764634,
      "grad_norm": 121.60181427001953,
      "learning_rate": 7.776770879501816e-07,
      "loss": 4.3542,
      "step": 21734
    },
    {
      "epoch": 0.8774727492935002,
      "grad_norm": 97.83626556396484,
      "learning_rate": 7.771716087927572e-07,
      "loss": 3.4468,
      "step": 21735
    },
    {
      "epoch": 0.8775131207105369,
      "grad_norm": 79.31881713867188,
      "learning_rate": 7.766662873241615e-07,
      "loss": 3.9836,
      "step": 21736
    },
    {
      "epoch": 0.8775534921275737,
      "grad_norm": 94.78187561035156,
      "learning_rate": 7.761611235530341e-07,
      "loss": 3.4129,
      "step": 21737
    },
    {
      "epoch": 0.8775938635446104,
      "grad_norm": 75.39099884033203,
      "learning_rate": 7.756561174880151e-07,
      "loss": 4.3083,
      "step": 21738
    },
    {
      "epoch": 0.8776342349616472,
      "grad_norm": 133.049072265625,
      "learning_rate": 7.751512691377328e-07,
      "loss": 2.7613,
      "step": 21739
    },
    {
      "epoch": 0.8776746063786839,
      "grad_norm": 121.0418472290039,
      "learning_rate": 7.746465785108226e-07,
      "loss": 3.7498,
      "step": 21740
    },
    {
      "epoch": 0.8777149777957206,
      "grad_norm": 113.681884765625,
      "learning_rate": 7.741420456159144e-07,
      "loss": 6.0016,
      "step": 21741
    },
    {
      "epoch": 0.8777553492127573,
      "grad_norm": 128.44139099121094,
      "learning_rate": 7.736376704616333e-07,
      "loss": 4.5073,
      "step": 21742
    },
    {
      "epoch": 0.877795720629794,
      "grad_norm": 104.8276596069336,
      "learning_rate": 7.731334530566026e-07,
      "loss": 6.0912,
      "step": 21743
    },
    {
      "epoch": 0.8778360920468309,
      "grad_norm": 111.64613342285156,
      "learning_rate": 7.72629393409442e-07,
      "loss": 5.0824,
      "step": 21744
    },
    {
      "epoch": 0.8778764634638676,
      "grad_norm": 179.3893280029297,
      "learning_rate": 7.721254915287724e-07,
      "loss": 5.7008,
      "step": 21745
    },
    {
      "epoch": 0.8779168348809043,
      "grad_norm": 74.37760162353516,
      "learning_rate": 7.716217474232101e-07,
      "loss": 5.4226,
      "step": 21746
    },
    {
      "epoch": 0.877957206297941,
      "grad_norm": 102.36148071289062,
      "learning_rate": 7.71118161101364e-07,
      "loss": 3.7975,
      "step": 21747
    },
    {
      "epoch": 0.8779975777149778,
      "grad_norm": 94.76980590820312,
      "learning_rate": 7.706147325718461e-07,
      "loss": 5.2863,
      "step": 21748
    },
    {
      "epoch": 0.8780379491320145,
      "grad_norm": 97.97067260742188,
      "learning_rate": 7.701114618432648e-07,
      "loss": 5.4511,
      "step": 21749
    },
    {
      "epoch": 0.8780783205490512,
      "grad_norm": 116.0995101928711,
      "learning_rate": 7.696083489242256e-07,
      "loss": 5.9744,
      "step": 21750
    },
    {
      "epoch": 0.878118691966088,
      "grad_norm": 122.39115905761719,
      "learning_rate": 7.691053938233273e-07,
      "loss": 4.9306,
      "step": 21751
    },
    {
      "epoch": 0.8781590633831248,
      "grad_norm": 180.291015625,
      "learning_rate": 7.686025965491706e-07,
      "loss": 4.5449,
      "step": 21752
    },
    {
      "epoch": 0.8781994348001615,
      "grad_norm": 81.38825225830078,
      "learning_rate": 7.680999571103531e-07,
      "loss": 3.9661,
      "step": 21753
    },
    {
      "epoch": 0.8782398062171982,
      "grad_norm": 114.65352630615234,
      "learning_rate": 7.675974755154691e-07,
      "loss": 5.2887,
      "step": 21754
    },
    {
      "epoch": 0.8782801776342349,
      "grad_norm": 93.53853607177734,
      "learning_rate": 7.670951517731073e-07,
      "loss": 5.4767,
      "step": 21755
    },
    {
      "epoch": 0.8783205490512717,
      "grad_norm": 131.0955810546875,
      "learning_rate": 7.665929858918597e-07,
      "loss": 6.1788,
      "step": 21756
    },
    {
      "epoch": 0.8783609204683085,
      "grad_norm": 85.3004150390625,
      "learning_rate": 7.660909778803094e-07,
      "loss": 5.4025,
      "step": 21757
    },
    {
      "epoch": 0.8784012918853452,
      "grad_norm": 79.66443634033203,
      "learning_rate": 7.655891277470429e-07,
      "loss": 5.0405,
      "step": 21758
    },
    {
      "epoch": 0.8784416633023819,
      "grad_norm": 83.55916595458984,
      "learning_rate": 7.650874355006365e-07,
      "loss": 6.2811,
      "step": 21759
    },
    {
      "epoch": 0.8784820347194187,
      "grad_norm": 99.3174819946289,
      "learning_rate": 7.645859011496704e-07,
      "loss": 3.9207,
      "step": 21760
    },
    {
      "epoch": 0.8785224061364554,
      "grad_norm": 74.44341278076172,
      "learning_rate": 7.640845247027184e-07,
      "loss": 4.3596,
      "step": 21761
    },
    {
      "epoch": 0.8785627775534921,
      "grad_norm": 110.03824615478516,
      "learning_rate": 7.635833061683562e-07,
      "loss": 3.5188,
      "step": 21762
    },
    {
      "epoch": 0.8786031489705288,
      "grad_norm": 126.78071594238281,
      "learning_rate": 7.630822455551478e-07,
      "loss": 5.5604,
      "step": 21763
    },
    {
      "epoch": 0.8786435203875657,
      "grad_norm": 133.4119873046875,
      "learning_rate": 7.625813428716644e-07,
      "loss": 5.0512,
      "step": 21764
    },
    {
      "epoch": 0.8786838918046024,
      "grad_norm": 90.14653015136719,
      "learning_rate": 7.620805981264689e-07,
      "loss": 6.6635,
      "step": 21765
    },
    {
      "epoch": 0.8787242632216391,
      "grad_norm": 122.80167388916016,
      "learning_rate": 7.615800113281246e-07,
      "loss": 2.3364,
      "step": 21766
    },
    {
      "epoch": 0.8787646346386758,
      "grad_norm": 126.8514404296875,
      "learning_rate": 7.610795824851869e-07,
      "loss": 4.4355,
      "step": 21767
    },
    {
      "epoch": 0.8788050060557125,
      "grad_norm": 173.72689819335938,
      "learning_rate": 7.605793116062144e-07,
      "loss": 4.5914,
      "step": 21768
    },
    {
      "epoch": 0.8788453774727493,
      "grad_norm": 128.64820861816406,
      "learning_rate": 7.600791986997591e-07,
      "loss": 4.4963,
      "step": 21769
    },
    {
      "epoch": 0.878885748889786,
      "grad_norm": 191.64016723632812,
      "learning_rate": 7.595792437743743e-07,
      "loss": 5.3299,
      "step": 21770
    },
    {
      "epoch": 0.8789261203068228,
      "grad_norm": 121.35284423828125,
      "learning_rate": 7.590794468386053e-07,
      "loss": 5.6442,
      "step": 21771
    },
    {
      "epoch": 0.8789664917238595,
      "grad_norm": 127.36064910888672,
      "learning_rate": 7.585798079009987e-07,
      "loss": 5.5895,
      "step": 21772
    },
    {
      "epoch": 0.8790068631408963,
      "grad_norm": 112.61198425292969,
      "learning_rate": 7.580803269700976e-07,
      "loss": 5.2212,
      "step": 21773
    },
    {
      "epoch": 0.879047234557933,
      "grad_norm": 84.94528198242188,
      "learning_rate": 7.575810040544429e-07,
      "loss": 4.8737,
      "step": 21774
    },
    {
      "epoch": 0.8790876059749697,
      "grad_norm": 99.97213745117188,
      "learning_rate": 7.570818391625689e-07,
      "loss": 6.2735,
      "step": 21775
    },
    {
      "epoch": 0.8791279773920064,
      "grad_norm": 91.1861572265625,
      "learning_rate": 7.565828323030122e-07,
      "loss": 6.4974,
      "step": 21776
    },
    {
      "epoch": 0.8791683488090432,
      "grad_norm": 91.51402282714844,
      "learning_rate": 7.560839834843037e-07,
      "loss": 5.8553,
      "step": 21777
    },
    {
      "epoch": 0.87920872022608,
      "grad_norm": 111.07503509521484,
      "learning_rate": 7.555852927149743e-07,
      "loss": 4.9203,
      "step": 21778
    },
    {
      "epoch": 0.8792490916431167,
      "grad_norm": 193.68002319335938,
      "learning_rate": 7.550867600035483e-07,
      "loss": 6.1496,
      "step": 21779
    },
    {
      "epoch": 0.8792894630601534,
      "grad_norm": 60.68867874145508,
      "learning_rate": 7.54588385358549e-07,
      "loss": 2.7268,
      "step": 21780
    },
    {
      "epoch": 0.8793298344771902,
      "grad_norm": 71.01819610595703,
      "learning_rate": 7.540901687884994e-07,
      "loss": 5.6905,
      "step": 21781
    },
    {
      "epoch": 0.8793702058942269,
      "grad_norm": 85.41544342041016,
      "learning_rate": 7.535921103019194e-07,
      "loss": 6.3889,
      "step": 21782
    },
    {
      "epoch": 0.8794105773112636,
      "grad_norm": 91.9921875,
      "learning_rate": 7.530942099073202e-07,
      "loss": 3.4472,
      "step": 21783
    },
    {
      "epoch": 0.8794509487283003,
      "grad_norm": 86.52517700195312,
      "learning_rate": 7.525964676132169e-07,
      "loss": 4.1961,
      "step": 21784
    },
    {
      "epoch": 0.8794913201453372,
      "grad_norm": 71.43353271484375,
      "learning_rate": 7.520988834281195e-07,
      "loss": 4.6848,
      "step": 21785
    },
    {
      "epoch": 0.8795316915623739,
      "grad_norm": 93.07698822021484,
      "learning_rate": 7.516014573605379e-07,
      "loss": 5.0401,
      "step": 21786
    },
    {
      "epoch": 0.8795720629794106,
      "grad_norm": 91.01232147216797,
      "learning_rate": 7.51104189418973e-07,
      "loss": 5.824,
      "step": 21787
    },
    {
      "epoch": 0.8796124343964473,
      "grad_norm": 92.11149597167969,
      "learning_rate": 7.506070796119281e-07,
      "loss": 5.8222,
      "step": 21788
    },
    {
      "epoch": 0.879652805813484,
      "grad_norm": 73.68006134033203,
      "learning_rate": 7.501101279479039e-07,
      "loss": 4.3514,
      "step": 21789
    },
    {
      "epoch": 0.8796931772305208,
      "grad_norm": 91.78612518310547,
      "learning_rate": 7.496133344353974e-07,
      "loss": 4.9735,
      "step": 21790
    },
    {
      "epoch": 0.8797335486475575,
      "grad_norm": 127.61553955078125,
      "learning_rate": 7.491166990829001e-07,
      "loss": 5.2509,
      "step": 21791
    },
    {
      "epoch": 0.8797739200645943,
      "grad_norm": 115.57611846923828,
      "learning_rate": 7.486202218989058e-07,
      "loss": 4.8476,
      "step": 21792
    },
    {
      "epoch": 0.879814291481631,
      "grad_norm": 304.1369323730469,
      "learning_rate": 7.481239028919018e-07,
      "loss": 5.3535,
      "step": 21793
    },
    {
      "epoch": 0.8798546628986678,
      "grad_norm": 98.74215698242188,
      "learning_rate": 7.476277420703759e-07,
      "loss": 4.9211,
      "step": 21794
    },
    {
      "epoch": 0.8798950343157045,
      "grad_norm": 58.5318603515625,
      "learning_rate": 7.471317394428079e-07,
      "loss": 2.189,
      "step": 21795
    },
    {
      "epoch": 0.8799354057327412,
      "grad_norm": 80.30406951904297,
      "learning_rate": 7.466358950176811e-07,
      "loss": 4.3418,
      "step": 21796
    },
    {
      "epoch": 0.8799757771497779,
      "grad_norm": 85.48258209228516,
      "learning_rate": 7.461402088034719e-07,
      "loss": 5.4886,
      "step": 21797
    },
    {
      "epoch": 0.8800161485668146,
      "grad_norm": 95.27513885498047,
      "learning_rate": 7.45644680808657e-07,
      "loss": 6.2354,
      "step": 21798
    },
    {
      "epoch": 0.8800565199838515,
      "grad_norm": 185.21827697753906,
      "learning_rate": 7.451493110417063e-07,
      "loss": 5.2215,
      "step": 21799
    },
    {
      "epoch": 0.8800968914008882,
      "grad_norm": 76.1452865600586,
      "learning_rate": 7.446540995110907e-07,
      "loss": 4.5629,
      "step": 21800
    },
    {
      "epoch": 0.8801372628179249,
      "grad_norm": 91.76260375976562,
      "learning_rate": 7.441590462252779e-07,
      "loss": 3.6555,
      "step": 21801
    },
    {
      "epoch": 0.8801776342349616,
      "grad_norm": 86.12191009521484,
      "learning_rate": 7.436641511927322e-07,
      "loss": 3.5006,
      "step": 21802
    },
    {
      "epoch": 0.8802180056519984,
      "grad_norm": 109.26505279541016,
      "learning_rate": 7.431694144219126e-07,
      "loss": 5.0049,
      "step": 21803
    },
    {
      "epoch": 0.8802583770690351,
      "grad_norm": 79.18289947509766,
      "learning_rate": 7.42674835921281e-07,
      "loss": 4.7962,
      "step": 21804
    },
    {
      "epoch": 0.8802987484860718,
      "grad_norm": 93.26051330566406,
      "learning_rate": 7.421804156992907e-07,
      "loss": 5.2505,
      "step": 21805
    },
    {
      "epoch": 0.8803391199031086,
      "grad_norm": 89.0224838256836,
      "learning_rate": 7.416861537643994e-07,
      "loss": 6.5066,
      "step": 21806
    },
    {
      "epoch": 0.8803794913201454,
      "grad_norm": 194.92784118652344,
      "learning_rate": 7.411920501250536e-07,
      "loss": 5.4847,
      "step": 21807
    },
    {
      "epoch": 0.8804198627371821,
      "grad_norm": 91.45177459716797,
      "learning_rate": 7.406981047897021e-07,
      "loss": 6.1941,
      "step": 21808
    },
    {
      "epoch": 0.8804602341542188,
      "grad_norm": 90.77765655517578,
      "learning_rate": 7.402043177667917e-07,
      "loss": 4.381,
      "step": 21809
    },
    {
      "epoch": 0.8805006055712555,
      "grad_norm": 466.6659240722656,
      "learning_rate": 7.397106890647643e-07,
      "loss": 5.9204,
      "step": 21810
    },
    {
      "epoch": 0.8805409769882923,
      "grad_norm": 96.05229187011719,
      "learning_rate": 7.392172186920598e-07,
      "loss": 3.9012,
      "step": 21811
    },
    {
      "epoch": 0.880581348405329,
      "grad_norm": 73.98544311523438,
      "learning_rate": 7.38723906657115e-07,
      "loss": 3.1644,
      "step": 21812
    },
    {
      "epoch": 0.8806217198223658,
      "grad_norm": 97.23600769042969,
      "learning_rate": 7.382307529683641e-07,
      "loss": 4.6604,
      "step": 21813
    },
    {
      "epoch": 0.8806620912394025,
      "grad_norm": 79.61952209472656,
      "learning_rate": 7.377377576342426e-07,
      "loss": 3.7594,
      "step": 21814
    },
    {
      "epoch": 0.8807024626564393,
      "grad_norm": 89.68916320800781,
      "learning_rate": 7.372449206631738e-07,
      "loss": 3.8097,
      "step": 21815
    },
    {
      "epoch": 0.880742834073476,
      "grad_norm": 101.52051544189453,
      "learning_rate": 7.367522420635876e-07,
      "loss": 5.0361,
      "step": 21816
    },
    {
      "epoch": 0.8807832054905127,
      "grad_norm": 103.41100311279297,
      "learning_rate": 7.362597218439061e-07,
      "loss": 4.3585,
      "step": 21817
    },
    {
      "epoch": 0.8808235769075494,
      "grad_norm": 100.2960433959961,
      "learning_rate": 7.357673600125526e-07,
      "loss": 5.6865,
      "step": 21818
    },
    {
      "epoch": 0.8808639483245861,
      "grad_norm": 73.06879425048828,
      "learning_rate": 7.352751565779426e-07,
      "loss": 4.0669,
      "step": 21819
    },
    {
      "epoch": 0.880904319741623,
      "grad_norm": 92.71707153320312,
      "learning_rate": 7.347831115484927e-07,
      "loss": 2.2724,
      "step": 21820
    },
    {
      "epoch": 0.8809446911586597,
      "grad_norm": 105.41205596923828,
      "learning_rate": 7.342912249326161e-07,
      "loss": 5.2124,
      "step": 21821
    },
    {
      "epoch": 0.8809850625756964,
      "grad_norm": 83.84627532958984,
      "learning_rate": 7.33799496738723e-07,
      "loss": 5.8059,
      "step": 21822
    },
    {
      "epoch": 0.8810254339927331,
      "grad_norm": 85.5617904663086,
      "learning_rate": 7.333079269752197e-07,
      "loss": 4.6022,
      "step": 21823
    },
    {
      "epoch": 0.8810658054097699,
      "grad_norm": 129.50152587890625,
      "learning_rate": 7.328165156505107e-07,
      "loss": 4.8083,
      "step": 21824
    },
    {
      "epoch": 0.8811061768268066,
      "grad_norm": 90.7642593383789,
      "learning_rate": 7.323252627730004e-07,
      "loss": 3.0704,
      "step": 21825
    },
    {
      "epoch": 0.8811465482438433,
      "grad_norm": 77.42548370361328,
      "learning_rate": 7.318341683510866e-07,
      "loss": 3.0301,
      "step": 21826
    },
    {
      "epoch": 0.8811869196608801,
      "grad_norm": 128.2386016845703,
      "learning_rate": 7.313432323931646e-07,
      "loss": 4.3439,
      "step": 21827
    },
    {
      "epoch": 0.8812272910779169,
      "grad_norm": 82.77460479736328,
      "learning_rate": 7.308524549076302e-07,
      "loss": 4.4949,
      "step": 21828
    },
    {
      "epoch": 0.8812676624949536,
      "grad_norm": 123.55992889404297,
      "learning_rate": 7.30361835902873e-07,
      "loss": 4.8939,
      "step": 21829
    },
    {
      "epoch": 0.8813080339119903,
      "grad_norm": 89.51597595214844,
      "learning_rate": 7.298713753872821e-07,
      "loss": 6.4336,
      "step": 21830
    },
    {
      "epoch": 0.881348405329027,
      "grad_norm": 123.41996002197266,
      "learning_rate": 7.293810733692441e-07,
      "loss": 5.849,
      "step": 21831
    },
    {
      "epoch": 0.8813887767460638,
      "grad_norm": 92.5799560546875,
      "learning_rate": 7.288909298571412e-07,
      "loss": 3.1666,
      "step": 21832
    },
    {
      "epoch": 0.8814291481631005,
      "grad_norm": 83.04730224609375,
      "learning_rate": 7.284009448593532e-07,
      "loss": 5.7668,
      "step": 21833
    },
    {
      "epoch": 0.8814695195801373,
      "grad_norm": 126.13864135742188,
      "learning_rate": 7.279111183842602e-07,
      "loss": 5.1612,
      "step": 21834
    },
    {
      "epoch": 0.881509890997174,
      "grad_norm": 105.6893081665039,
      "learning_rate": 7.274214504402333e-07,
      "loss": 5.8075,
      "step": 21835
    },
    {
      "epoch": 0.8815502624142108,
      "grad_norm": 69.08118438720703,
      "learning_rate": 7.269319410356468e-07,
      "loss": 5.8562,
      "step": 21836
    },
    {
      "epoch": 0.8815906338312475,
      "grad_norm": 83.94727325439453,
      "learning_rate": 7.264425901788708e-07,
      "loss": 4.5873,
      "step": 21837
    },
    {
      "epoch": 0.8816310052482842,
      "grad_norm": 89.66702270507812,
      "learning_rate": 7.259533978782718e-07,
      "loss": 3.3292,
      "step": 21838
    },
    {
      "epoch": 0.8816713766653209,
      "grad_norm": 144.31617736816406,
      "learning_rate": 7.254643641422121e-07,
      "loss": 5.1391,
      "step": 21839
    },
    {
      "epoch": 0.8817117480823576,
      "grad_norm": 89.93815612792969,
      "learning_rate": 7.249754889790539e-07,
      "loss": 5.0075,
      "step": 21840
    },
    {
      "epoch": 0.8817521194993945,
      "grad_norm": 96.0818099975586,
      "learning_rate": 7.244867723971561e-07,
      "loss": 6.0271,
      "step": 21841
    },
    {
      "epoch": 0.8817924909164312,
      "grad_norm": 103.23464965820312,
      "learning_rate": 7.239982144048763e-07,
      "loss": 4.9369,
      "step": 21842
    },
    {
      "epoch": 0.8818328623334679,
      "grad_norm": 114.75096893310547,
      "learning_rate": 7.235098150105646e-07,
      "loss": 4.1198,
      "step": 21843
    },
    {
      "epoch": 0.8818732337505046,
      "grad_norm": 90.29881286621094,
      "learning_rate": 7.230215742225721e-07,
      "loss": 4.5574,
      "step": 21844
    },
    {
      "epoch": 0.8819136051675414,
      "grad_norm": 209.90756225585938,
      "learning_rate": 7.225334920492477e-07,
      "loss": 5.948,
      "step": 21845
    },
    {
      "epoch": 0.8819539765845781,
      "grad_norm": 99.87511444091797,
      "learning_rate": 7.22045568498938e-07,
      "loss": 4.7442,
      "step": 21846
    },
    {
      "epoch": 0.8819943480016148,
      "grad_norm": 105.2567367553711,
      "learning_rate": 7.215578035799819e-07,
      "loss": 4.063,
      "step": 21847
    },
    {
      "epoch": 0.8820347194186516,
      "grad_norm": 96.37026977539062,
      "learning_rate": 7.210701973007206e-07,
      "loss": 4.578,
      "step": 21848
    },
    {
      "epoch": 0.8820750908356884,
      "grad_norm": 71.40208435058594,
      "learning_rate": 7.205827496694917e-07,
      "loss": 4.814,
      "step": 21849
    },
    {
      "epoch": 0.8821154622527251,
      "grad_norm": 72.55815124511719,
      "learning_rate": 7.200954606946275e-07,
      "loss": 1.8517,
      "step": 21850
    },
    {
      "epoch": 0.8821558336697618,
      "grad_norm": 88.294677734375,
      "learning_rate": 7.196083303844625e-07,
      "loss": 3.4886,
      "step": 21851
    },
    {
      "epoch": 0.8821962050867985,
      "grad_norm": 98.67677307128906,
      "learning_rate": 7.191213587473222e-07,
      "loss": 7.0845,
      "step": 21852
    },
    {
      "epoch": 0.8822365765038352,
      "grad_norm": 239.36546325683594,
      "learning_rate": 7.186345457915344e-07,
      "loss": 4.7717,
      "step": 21853
    },
    {
      "epoch": 0.882276947920872,
      "grad_norm": 88.35028839111328,
      "learning_rate": 7.181478915254247e-07,
      "loss": 5.3899,
      "step": 21854
    },
    {
      "epoch": 0.8823173193379088,
      "grad_norm": 78.2269515991211,
      "learning_rate": 7.176613959573098e-07,
      "loss": 5.2852,
      "step": 21855
    },
    {
      "epoch": 0.8823576907549455,
      "grad_norm": 93.62662506103516,
      "learning_rate": 7.171750590955084e-07,
      "loss": 5.4371,
      "step": 21856
    },
    {
      "epoch": 0.8823980621719822,
      "grad_norm": 104.64319610595703,
      "learning_rate": 7.166888809483374e-07,
      "loss": 4.7596,
      "step": 21857
    },
    {
      "epoch": 0.882438433589019,
      "grad_norm": 124.64581298828125,
      "learning_rate": 7.16202861524109e-07,
      "loss": 5.6503,
      "step": 21858
    },
    {
      "epoch": 0.8824788050060557,
      "grad_norm": 85.10987091064453,
      "learning_rate": 7.15717000831132e-07,
      "loss": 5.2983,
      "step": 21859
    },
    {
      "epoch": 0.8825191764230924,
      "grad_norm": 117.57777404785156,
      "learning_rate": 7.152312988777133e-07,
      "loss": 4.8843,
      "step": 21860
    },
    {
      "epoch": 0.8825595478401292,
      "grad_norm": 171.3600311279297,
      "learning_rate": 7.147457556721582e-07,
      "loss": 4.9235,
      "step": 21861
    },
    {
      "epoch": 0.882599919257166,
      "grad_norm": 83.93344116210938,
      "learning_rate": 7.142603712227691e-07,
      "loss": 4.8274,
      "step": 21862
    },
    {
      "epoch": 0.8826402906742027,
      "grad_norm": 119.50225067138672,
      "learning_rate": 7.137751455378427e-07,
      "loss": 3.6358,
      "step": 21863
    },
    {
      "epoch": 0.8826806620912394,
      "grad_norm": 87.3230972290039,
      "learning_rate": 7.132900786256758e-07,
      "loss": 2.9321,
      "step": 21864
    },
    {
      "epoch": 0.8827210335082761,
      "grad_norm": 93.49151611328125,
      "learning_rate": 7.128051704945638e-07,
      "loss": 5.8721,
      "step": 21865
    },
    {
      "epoch": 0.8827614049253129,
      "grad_norm": 84.34266662597656,
      "learning_rate": 7.123204211527968e-07,
      "loss": 5.7334,
      "step": 21866
    },
    {
      "epoch": 0.8828017763423496,
      "grad_norm": 93.27501678466797,
      "learning_rate": 7.118358306086604e-07,
      "loss": 6.6712,
      "step": 21867
    },
    {
      "epoch": 0.8828421477593864,
      "grad_norm": 100.23542022705078,
      "learning_rate": 7.113513988704424e-07,
      "loss": 5.0889,
      "step": 21868
    },
    {
      "epoch": 0.8828825191764231,
      "grad_norm": 158.7731170654297,
      "learning_rate": 7.108671259464251e-07,
      "loss": 5.434,
      "step": 21869
    },
    {
      "epoch": 0.8829228905934599,
      "grad_norm": 141.20870971679688,
      "learning_rate": 7.103830118448885e-07,
      "loss": 5.611,
      "step": 21870
    },
    {
      "epoch": 0.8829632620104966,
      "grad_norm": 115.33069610595703,
      "learning_rate": 7.098990565741093e-07,
      "loss": 5.2511,
      "step": 21871
    },
    {
      "epoch": 0.8830036334275333,
      "grad_norm": 103.5365982055664,
      "learning_rate": 7.09415260142362e-07,
      "loss": 5.5936,
      "step": 21872
    },
    {
      "epoch": 0.88304400484457,
      "grad_norm": 88.11951446533203,
      "learning_rate": 7.08931622557919e-07,
      "loss": 4.7476,
      "step": 21873
    },
    {
      "epoch": 0.8830843762616067,
      "grad_norm": 89.9136962890625,
      "learning_rate": 7.084481438290502e-07,
      "loss": 4.9372,
      "step": 21874
    },
    {
      "epoch": 0.8831247476786436,
      "grad_norm": 114.82233428955078,
      "learning_rate": 7.079648239640202e-07,
      "loss": 3.8337,
      "step": 21875
    },
    {
      "epoch": 0.8831651190956803,
      "grad_norm": 103.87654113769531,
      "learning_rate": 7.074816629710923e-07,
      "loss": 4.3725,
      "step": 21876
    },
    {
      "epoch": 0.883205490512717,
      "grad_norm": 121.49421691894531,
      "learning_rate": 7.069986608585289e-07,
      "loss": 4.34,
      "step": 21877
    },
    {
      "epoch": 0.8832458619297537,
      "grad_norm": 81.6307144165039,
      "learning_rate": 7.065158176345887e-07,
      "loss": 4.9035,
      "step": 21878
    },
    {
      "epoch": 0.8832862333467905,
      "grad_norm": 81.7341079711914,
      "learning_rate": 7.060331333075254e-07,
      "loss": 3.2715,
      "step": 21879
    },
    {
      "epoch": 0.8833266047638272,
      "grad_norm": 84.56355285644531,
      "learning_rate": 7.055506078855912e-07,
      "loss": 5.8472,
      "step": 21880
    },
    {
      "epoch": 0.8833669761808639,
      "grad_norm": 139.911376953125,
      "learning_rate": 7.050682413770382e-07,
      "loss": 5.8246,
      "step": 21881
    },
    {
      "epoch": 0.8834073475979007,
      "grad_norm": 89.91520690917969,
      "learning_rate": 7.045860337901156e-07,
      "loss": 4.3652,
      "step": 21882
    },
    {
      "epoch": 0.8834477190149375,
      "grad_norm": 74.0549545288086,
      "learning_rate": 7.041039851330633e-07,
      "loss": 4.6812,
      "step": 21883
    },
    {
      "epoch": 0.8834880904319742,
      "grad_norm": 88.08700561523438,
      "learning_rate": 7.036220954141248e-07,
      "loss": 4.2222,
      "step": 21884
    },
    {
      "epoch": 0.8835284618490109,
      "grad_norm": 172.86671447753906,
      "learning_rate": 7.031403646415402e-07,
      "loss": 4.7963,
      "step": 21885
    },
    {
      "epoch": 0.8835688332660476,
      "grad_norm": 139.9503173828125,
      "learning_rate": 7.026587928235473e-07,
      "loss": 6.8496,
      "step": 21886
    },
    {
      "epoch": 0.8836092046830843,
      "grad_norm": 88.35234069824219,
      "learning_rate": 7.021773799683762e-07,
      "loss": 4.5335,
      "step": 21887
    },
    {
      "epoch": 0.8836495761001211,
      "grad_norm": 130.25794982910156,
      "learning_rate": 7.016961260842603e-07,
      "loss": 4.1278,
      "step": 21888
    },
    {
      "epoch": 0.8836899475171579,
      "grad_norm": 101.37537384033203,
      "learning_rate": 7.012150311794275e-07,
      "loss": 4.3486,
      "step": 21889
    },
    {
      "epoch": 0.8837303189341946,
      "grad_norm": 106.30224609375,
      "learning_rate": 7.007340952621034e-07,
      "loss": 4.7722,
      "step": 21890
    },
    {
      "epoch": 0.8837706903512313,
      "grad_norm": 98.25,
      "learning_rate": 7.002533183405113e-07,
      "loss": 4.5441,
      "step": 21891
    },
    {
      "epoch": 0.8838110617682681,
      "grad_norm": 450.7222595214844,
      "learning_rate": 6.997727004228705e-07,
      "loss": 4.1473,
      "step": 21892
    },
    {
      "epoch": 0.8838514331853048,
      "grad_norm": 91.29002380371094,
      "learning_rate": 6.992922415173997e-07,
      "loss": 4.2358,
      "step": 21893
    },
    {
      "epoch": 0.8838918046023415,
      "grad_norm": 91.0141830444336,
      "learning_rate": 6.988119416323114e-07,
      "loss": 6.3765,
      "step": 21894
    },
    {
      "epoch": 0.8839321760193782,
      "grad_norm": 99.75525665283203,
      "learning_rate": 6.983318007758221e-07,
      "loss": 6.6898,
      "step": 21895
    },
    {
      "epoch": 0.8839725474364151,
      "grad_norm": 72.12174987792969,
      "learning_rate": 6.978518189561356e-07,
      "loss": 4.954,
      "step": 21896
    },
    {
      "epoch": 0.8840129188534518,
      "grad_norm": 118.27668762207031,
      "learning_rate": 6.973719961814618e-07,
      "loss": 6.0944,
      "step": 21897
    },
    {
      "epoch": 0.8840532902704885,
      "grad_norm": 81.83804321289062,
      "learning_rate": 6.96892332460003e-07,
      "loss": 3.6943,
      "step": 21898
    },
    {
      "epoch": 0.8840936616875252,
      "grad_norm": 98.62992095947266,
      "learning_rate": 6.964128277999626e-07,
      "loss": 4.9505,
      "step": 21899
    },
    {
      "epoch": 0.884134033104562,
      "grad_norm": 93.72579193115234,
      "learning_rate": 6.959334822095354e-07,
      "loss": 5.1758,
      "step": 21900
    },
    {
      "epoch": 0.8841744045215987,
      "grad_norm": 121.44007873535156,
      "learning_rate": 6.954542956969201e-07,
      "loss": 5.6082,
      "step": 21901
    },
    {
      "epoch": 0.8842147759386354,
      "grad_norm": 104.46347045898438,
      "learning_rate": 6.949752682703081e-07,
      "loss": 4.1329,
      "step": 21902
    },
    {
      "epoch": 0.8842551473556722,
      "grad_norm": 97.06919860839844,
      "learning_rate": 6.944963999378918e-07,
      "loss": 6.2578,
      "step": 21903
    },
    {
      "epoch": 0.884295518772709,
      "grad_norm": 105.31707763671875,
      "learning_rate": 6.940176907078555e-07,
      "loss": 3.1239,
      "step": 21904
    },
    {
      "epoch": 0.8843358901897457,
      "grad_norm": 112.30674743652344,
      "learning_rate": 6.935391405883851e-07,
      "loss": 5.3425,
      "step": 21905
    },
    {
      "epoch": 0.8843762616067824,
      "grad_norm": 93.31320190429688,
      "learning_rate": 6.93060749587664e-07,
      "loss": 5.0418,
      "step": 21906
    },
    {
      "epoch": 0.8844166330238191,
      "grad_norm": 178.06622314453125,
      "learning_rate": 6.925825177138723e-07,
      "loss": 4.7998,
      "step": 21907
    },
    {
      "epoch": 0.8844570044408558,
      "grad_norm": 158.96507263183594,
      "learning_rate": 6.921044449751835e-07,
      "loss": 4.1176,
      "step": 21908
    },
    {
      "epoch": 0.8844973758578926,
      "grad_norm": 107.73719024658203,
      "learning_rate": 6.916265313797732e-07,
      "loss": 4.9163,
      "step": 21909
    },
    {
      "epoch": 0.8845377472749294,
      "grad_norm": 80.84407806396484,
      "learning_rate": 6.911487769358127e-07,
      "loss": 3.7861,
      "step": 21910
    },
    {
      "epoch": 0.8845781186919661,
      "grad_norm": 105.1221694946289,
      "learning_rate": 6.906711816514711e-07,
      "loss": 6.4965,
      "step": 21911
    },
    {
      "epoch": 0.8846184901090028,
      "grad_norm": 100.57685852050781,
      "learning_rate": 6.901937455349128e-07,
      "loss": 5.6075,
      "step": 21912
    },
    {
      "epoch": 0.8846588615260396,
      "grad_norm": 124.13013458251953,
      "learning_rate": 6.897164685943003e-07,
      "loss": 5.2937,
      "step": 21913
    },
    {
      "epoch": 0.8846992329430763,
      "grad_norm": 101.0923080444336,
      "learning_rate": 6.89239350837797e-07,
      "loss": 4.3755,
      "step": 21914
    },
    {
      "epoch": 0.884739604360113,
      "grad_norm": 78.6761474609375,
      "learning_rate": 6.887623922735586e-07,
      "loss": 3.348,
      "step": 21915
    },
    {
      "epoch": 0.8847799757771497,
      "grad_norm": 196.80587768554688,
      "learning_rate": 6.882855929097398e-07,
      "loss": 3.4095,
      "step": 21916
    },
    {
      "epoch": 0.8848203471941866,
      "grad_norm": 99.67860412597656,
      "learning_rate": 6.878089527544918e-07,
      "loss": 4.5636,
      "step": 21917
    },
    {
      "epoch": 0.8848607186112233,
      "grad_norm": 77.88136291503906,
      "learning_rate": 6.873324718159657e-07,
      "loss": 4.6976,
      "step": 21918
    },
    {
      "epoch": 0.88490109002826,
      "grad_norm": 86.8540267944336,
      "learning_rate": 6.868561501023096e-07,
      "loss": 4.525,
      "step": 21919
    },
    {
      "epoch": 0.8849414614452967,
      "grad_norm": 139.94581604003906,
      "learning_rate": 6.863799876216637e-07,
      "loss": 5.4925,
      "step": 21920
    },
    {
      "epoch": 0.8849818328623335,
      "grad_norm": 110.04594421386719,
      "learning_rate": 6.859039843821713e-07,
      "loss": 4.9884,
      "step": 21921
    },
    {
      "epoch": 0.8850222042793702,
      "grad_norm": 73.31556701660156,
      "learning_rate": 6.854281403919705e-07,
      "loss": 4.2636,
      "step": 21922
    },
    {
      "epoch": 0.8850625756964069,
      "grad_norm": 82.216552734375,
      "learning_rate": 6.849524556591991e-07,
      "loss": 5.1338,
      "step": 21923
    },
    {
      "epoch": 0.8851029471134437,
      "grad_norm": 106.59896850585938,
      "learning_rate": 6.844769301919874e-07,
      "loss": 4.2448,
      "step": 21924
    },
    {
      "epoch": 0.8851433185304804,
      "grad_norm": 88.99685668945312,
      "learning_rate": 6.840015639984665e-07,
      "loss": 5.877,
      "step": 21925
    },
    {
      "epoch": 0.8851836899475172,
      "grad_norm": 182.22862243652344,
      "learning_rate": 6.835263570867645e-07,
      "loss": 3.8077,
      "step": 21926
    },
    {
      "epoch": 0.8852240613645539,
      "grad_norm": 106.76053619384766,
      "learning_rate": 6.830513094650071e-07,
      "loss": 4.6697,
      "step": 21927
    },
    {
      "epoch": 0.8852644327815906,
      "grad_norm": 153.2103271484375,
      "learning_rate": 6.825764211413144e-07,
      "loss": 3.6662,
      "step": 21928
    },
    {
      "epoch": 0.8853048041986273,
      "grad_norm": 84.18281555175781,
      "learning_rate": 6.821016921238077e-07,
      "loss": 6.734,
      "step": 21929
    },
    {
      "epoch": 0.8853451756156641,
      "grad_norm": 87.70294189453125,
      "learning_rate": 6.816271224206028e-07,
      "loss": 4.1188,
      "step": 21930
    },
    {
      "epoch": 0.8853855470327009,
      "grad_norm": 85.6084976196289,
      "learning_rate": 6.811527120398143e-07,
      "loss": 4.4443,
      "step": 21931
    },
    {
      "epoch": 0.8854259184497376,
      "grad_norm": 185.9727783203125,
      "learning_rate": 6.806784609895533e-07,
      "loss": 5.3016,
      "step": 21932
    },
    {
      "epoch": 0.8854662898667743,
      "grad_norm": 84.32737731933594,
      "learning_rate": 6.802043692779292e-07,
      "loss": 2.7339,
      "step": 21933
    },
    {
      "epoch": 0.8855066612838111,
      "grad_norm": 110.05551147460938,
      "learning_rate": 6.797304369130464e-07,
      "loss": 4.1676,
      "step": 21934
    },
    {
      "epoch": 0.8855470327008478,
      "grad_norm": 146.76708984375,
      "learning_rate": 6.792566639030096e-07,
      "loss": 5.4371,
      "step": 21935
    },
    {
      "epoch": 0.8855874041178845,
      "grad_norm": 83.99665069580078,
      "learning_rate": 6.787830502559179e-07,
      "loss": 4.4956,
      "step": 21936
    },
    {
      "epoch": 0.8856277755349212,
      "grad_norm": 108.87350463867188,
      "learning_rate": 6.783095959798691e-07,
      "loss": 4.7585,
      "step": 21937
    },
    {
      "epoch": 0.8856681469519581,
      "grad_norm": 105.19760131835938,
      "learning_rate": 6.77836301082958e-07,
      "loss": 5.1364,
      "step": 21938
    },
    {
      "epoch": 0.8857085183689948,
      "grad_norm": 95.7623062133789,
      "learning_rate": 6.773631655732793e-07,
      "loss": 4.715,
      "step": 21939
    },
    {
      "epoch": 0.8857488897860315,
      "grad_norm": 113.1798095703125,
      "learning_rate": 6.768901894589197e-07,
      "loss": 3.6263,
      "step": 21940
    },
    {
      "epoch": 0.8857892612030682,
      "grad_norm": 79.76509094238281,
      "learning_rate": 6.764173727479661e-07,
      "loss": 4.5547,
      "step": 21941
    },
    {
      "epoch": 0.885829632620105,
      "grad_norm": 83.12648010253906,
      "learning_rate": 6.759447154485033e-07,
      "loss": 4.9361,
      "step": 21942
    },
    {
      "epoch": 0.8858700040371417,
      "grad_norm": 97.50048065185547,
      "learning_rate": 6.754722175686135e-07,
      "loss": 4.927,
      "step": 21943
    },
    {
      "epoch": 0.8859103754541784,
      "grad_norm": 403.2351989746094,
      "learning_rate": 6.749998791163736e-07,
      "loss": 5.0933,
      "step": 21944
    },
    {
      "epoch": 0.8859507468712152,
      "grad_norm": 101.22335815429688,
      "learning_rate": 6.745277000998607e-07,
      "loss": 4.2249,
      "step": 21945
    },
    {
      "epoch": 0.8859911182882519,
      "grad_norm": 136.91400146484375,
      "learning_rate": 6.74055680527147e-07,
      "loss": 5.9275,
      "step": 21946
    },
    {
      "epoch": 0.8860314897052887,
      "grad_norm": 91.43929290771484,
      "learning_rate": 6.735838204063049e-07,
      "loss": 4.9023,
      "step": 21947
    },
    {
      "epoch": 0.8860718611223254,
      "grad_norm": 97.79569244384766,
      "learning_rate": 6.731121197454005e-07,
      "loss": 4.1901,
      "step": 21948
    },
    {
      "epoch": 0.8861122325393621,
      "grad_norm": 145.9103546142578,
      "learning_rate": 6.72640578552498e-07,
      "loss": 4.6746,
      "step": 21949
    },
    {
      "epoch": 0.8861526039563988,
      "grad_norm": 103.24336242675781,
      "learning_rate": 6.721691968356603e-07,
      "loss": 2.7508,
      "step": 21950
    },
    {
      "epoch": 0.8861929753734356,
      "grad_norm": 78.5317611694336,
      "learning_rate": 6.716979746029473e-07,
      "loss": 6.1273,
      "step": 21951
    },
    {
      "epoch": 0.8862333467904724,
      "grad_norm": 100.58294677734375,
      "learning_rate": 6.712269118624159e-07,
      "loss": 4.8255,
      "step": 21952
    },
    {
      "epoch": 0.8862737182075091,
      "grad_norm": 84.92002868652344,
      "learning_rate": 6.70756008622121e-07,
      "loss": 4.8015,
      "step": 21953
    },
    {
      "epoch": 0.8863140896245458,
      "grad_norm": 73.52210235595703,
      "learning_rate": 6.702852648901126e-07,
      "loss": 5.177,
      "step": 21954
    },
    {
      "epoch": 0.8863544610415826,
      "grad_norm": 87.38639068603516,
      "learning_rate": 6.69814680674441e-07,
      "loss": 4.879,
      "step": 21955
    },
    {
      "epoch": 0.8863948324586193,
      "grad_norm": 94.8726577758789,
      "learning_rate": 6.693442559831487e-07,
      "loss": 4.6962,
      "step": 21956
    },
    {
      "epoch": 0.886435203875656,
      "grad_norm": 136.13841247558594,
      "learning_rate": 6.688739908242814e-07,
      "loss": 4.8481,
      "step": 21957
    },
    {
      "epoch": 0.8864755752926927,
      "grad_norm": 92.99578857421875,
      "learning_rate": 6.684038852058794e-07,
      "loss": 4.2644,
      "step": 21958
    },
    {
      "epoch": 0.8865159467097296,
      "grad_norm": 129.103759765625,
      "learning_rate": 6.679339391359818e-07,
      "loss": 3.0786,
      "step": 21959
    },
    {
      "epoch": 0.8865563181267663,
      "grad_norm": 146.2781219482422,
      "learning_rate": 6.6746415262262e-07,
      "loss": 5.5581,
      "step": 21960
    },
    {
      "epoch": 0.886596689543803,
      "grad_norm": 119.17465209960938,
      "learning_rate": 6.669945256738287e-07,
      "loss": 2.8457,
      "step": 21961
    },
    {
      "epoch": 0.8866370609608397,
      "grad_norm": 126.80008697509766,
      "learning_rate": 6.665250582976357e-07,
      "loss": 5.8066,
      "step": 21962
    },
    {
      "epoch": 0.8866774323778764,
      "grad_norm": 170.48167419433594,
      "learning_rate": 6.660557505020715e-07,
      "loss": 5.6291,
      "step": 21963
    },
    {
      "epoch": 0.8867178037949132,
      "grad_norm": 107.26781463623047,
      "learning_rate": 6.655866022951551e-07,
      "loss": 5.5688,
      "step": 21964
    },
    {
      "epoch": 0.88675817521195,
      "grad_norm": 95.55729675292969,
      "learning_rate": 6.651176136849113e-07,
      "loss": 5.6552,
      "step": 21965
    },
    {
      "epoch": 0.8867985466289867,
      "grad_norm": 85.62303161621094,
      "learning_rate": 6.646487846793581e-07,
      "loss": 4.0302,
      "step": 21966
    },
    {
      "epoch": 0.8868389180460234,
      "grad_norm": 186.61712646484375,
      "learning_rate": 6.641801152865124e-07,
      "loss": 5.5225,
      "step": 21967
    },
    {
      "epoch": 0.8868792894630602,
      "grad_norm": 91.03670501708984,
      "learning_rate": 6.637116055143833e-07,
      "loss": 4.8932,
      "step": 21968
    },
    {
      "epoch": 0.8869196608800969,
      "grad_norm": 85.59288787841797,
      "learning_rate": 6.632432553709856e-07,
      "loss": 4.2764,
      "step": 21969
    },
    {
      "epoch": 0.8869600322971336,
      "grad_norm": 99.52684020996094,
      "learning_rate": 6.627750648643239e-07,
      "loss": 6.2689,
      "step": 21970
    },
    {
      "epoch": 0.8870004037141703,
      "grad_norm": 105.39096069335938,
      "learning_rate": 6.623070340024051e-07,
      "loss": 6.2806,
      "step": 21971
    },
    {
      "epoch": 0.8870407751312072,
      "grad_norm": 122.26895141601562,
      "learning_rate": 6.618391627932319e-07,
      "loss": 5.676,
      "step": 21972
    },
    {
      "epoch": 0.8870811465482439,
      "grad_norm": 144.1465606689453,
      "learning_rate": 6.61371451244801e-07,
      "loss": 6.7066,
      "step": 21973
    },
    {
      "epoch": 0.8871215179652806,
      "grad_norm": 101.4389877319336,
      "learning_rate": 6.609038993651129e-07,
      "loss": 4.201,
      "step": 21974
    },
    {
      "epoch": 0.8871618893823173,
      "grad_norm": 171.8413543701172,
      "learning_rate": 6.604365071621599e-07,
      "loss": 3.2312,
      "step": 21975
    },
    {
      "epoch": 0.887202260799354,
      "grad_norm": 88.27732849121094,
      "learning_rate": 6.599692746439323e-07,
      "loss": 4.8543,
      "step": 21976
    },
    {
      "epoch": 0.8872426322163908,
      "grad_norm": 88.33709716796875,
      "learning_rate": 6.595022018184194e-07,
      "loss": 4.9496,
      "step": 21977
    },
    {
      "epoch": 0.8872830036334275,
      "grad_norm": 112.14306640625,
      "learning_rate": 6.590352886936069e-07,
      "loss": 5.0619,
      "step": 21978
    },
    {
      "epoch": 0.8873233750504643,
      "grad_norm": 82.44580078125,
      "learning_rate": 6.585685352774806e-07,
      "loss": 5.204,
      "step": 21979
    },
    {
      "epoch": 0.887363746467501,
      "grad_norm": 199.50779724121094,
      "learning_rate": 6.581019415780165e-07,
      "loss": 5.6071,
      "step": 21980
    },
    {
      "epoch": 0.8874041178845378,
      "grad_norm": 78.64119720458984,
      "learning_rate": 6.576355076031948e-07,
      "loss": 5.9134,
      "step": 21981
    },
    {
      "epoch": 0.8874444893015745,
      "grad_norm": 94.444580078125,
      "learning_rate": 6.571692333609892e-07,
      "loss": 4.8537,
      "step": 21982
    },
    {
      "epoch": 0.8874848607186112,
      "grad_norm": 128.75851440429688,
      "learning_rate": 6.567031188593742e-07,
      "loss": 3.9159,
      "step": 21983
    },
    {
      "epoch": 0.8875252321356479,
      "grad_norm": 88.65243530273438,
      "learning_rate": 6.56237164106317e-07,
      "loss": 5.1399,
      "step": 21984
    },
    {
      "epoch": 0.8875656035526847,
      "grad_norm": 121.67798614501953,
      "learning_rate": 6.557713691097844e-07,
      "loss": 4.7203,
      "step": 21985
    },
    {
      "epoch": 0.8876059749697215,
      "grad_norm": 102.33349609375,
      "learning_rate": 6.553057338777403e-07,
      "loss": 4.8688,
      "step": 21986
    },
    {
      "epoch": 0.8876463463867582,
      "grad_norm": 88.90603637695312,
      "learning_rate": 6.548402584181479e-07,
      "loss": 5.0658,
      "step": 21987
    },
    {
      "epoch": 0.8876867178037949,
      "grad_norm": 113.13143920898438,
      "learning_rate": 6.543749427389645e-07,
      "loss": 5.2705,
      "step": 21988
    },
    {
      "epoch": 0.8877270892208317,
      "grad_norm": 74.4134521484375,
      "learning_rate": 6.539097868481448e-07,
      "loss": 5.3779,
      "step": 21989
    },
    {
      "epoch": 0.8877674606378684,
      "grad_norm": 106.21776580810547,
      "learning_rate": 6.534447907536445e-07,
      "loss": 6.4685,
      "step": 21990
    },
    {
      "epoch": 0.8878078320549051,
      "grad_norm": 86.96946716308594,
      "learning_rate": 6.52979954463413e-07,
      "loss": 4.6485,
      "step": 21991
    },
    {
      "epoch": 0.8878482034719418,
      "grad_norm": 82.35333251953125,
      "learning_rate": 6.525152779853972e-07,
      "loss": 5.159,
      "step": 21992
    },
    {
      "epoch": 0.8878885748889787,
      "grad_norm": 74.98358917236328,
      "learning_rate": 6.520507613275406e-07,
      "loss": 4.4905,
      "step": 21993
    },
    {
      "epoch": 0.8879289463060154,
      "grad_norm": 135.18890380859375,
      "learning_rate": 6.515864044977882e-07,
      "loss": 4.6938,
      "step": 21994
    },
    {
      "epoch": 0.8879693177230521,
      "grad_norm": 141.85621643066406,
      "learning_rate": 6.51122207504079e-07,
      "loss": 4.5234,
      "step": 21995
    },
    {
      "epoch": 0.8880096891400888,
      "grad_norm": 90.80941009521484,
      "learning_rate": 6.506581703543469e-07,
      "loss": 4.8269,
      "step": 21996
    },
    {
      "epoch": 0.8880500605571255,
      "grad_norm": 66.84796905517578,
      "learning_rate": 6.501942930565286e-07,
      "loss": 4.0641,
      "step": 21997
    },
    {
      "epoch": 0.8880904319741623,
      "grad_norm": 95.31982421875,
      "learning_rate": 6.497305756185545e-07,
      "loss": 4.3029,
      "step": 21998
    },
    {
      "epoch": 0.888130803391199,
      "grad_norm": 136.0639190673828,
      "learning_rate": 6.492670180483551e-07,
      "loss": 4.1035,
      "step": 21999
    },
    {
      "epoch": 0.8881711748082358,
      "grad_norm": 90.67755889892578,
      "learning_rate": 6.488036203538517e-07,
      "loss": 5.5893,
      "step": 22000
    },
    {
      "epoch": 0.8882115462252725,
      "grad_norm": 132.72438049316406,
      "learning_rate": 6.483403825429702e-07,
      "loss": 3.9723,
      "step": 22001
    },
    {
      "epoch": 0.8882519176423093,
      "grad_norm": 102.9636459350586,
      "learning_rate": 6.478773046236297e-07,
      "loss": 5.0011,
      "step": 22002
    },
    {
      "epoch": 0.888292289059346,
      "grad_norm": 238.70062255859375,
      "learning_rate": 6.474143866037508e-07,
      "loss": 3.6608,
      "step": 22003
    },
    {
      "epoch": 0.8883326604763827,
      "grad_norm": 81.61260986328125,
      "learning_rate": 6.469516284912447e-07,
      "loss": 4.6327,
      "step": 22004
    },
    {
      "epoch": 0.8883730318934194,
      "grad_norm": 184.86326599121094,
      "learning_rate": 6.464890302940241e-07,
      "loss": 3.6642,
      "step": 22005
    },
    {
      "epoch": 0.8884134033104562,
      "grad_norm": 115.76302337646484,
      "learning_rate": 6.460265920199992e-07,
      "loss": 4.8814,
      "step": 22006
    },
    {
      "epoch": 0.888453774727493,
      "grad_norm": 77.81167602539062,
      "learning_rate": 6.455643136770773e-07,
      "loss": 5.1612,
      "step": 22007
    },
    {
      "epoch": 0.8884941461445297,
      "grad_norm": 84.5443344116211,
      "learning_rate": 6.451021952731606e-07,
      "loss": 6.3958,
      "step": 22008
    },
    {
      "epoch": 0.8885345175615664,
      "grad_norm": 125.16146087646484,
      "learning_rate": 6.446402368161519e-07,
      "loss": 5.5417,
      "step": 22009
    },
    {
      "epoch": 0.8885748889786032,
      "grad_norm": 104.23011016845703,
      "learning_rate": 6.441784383139482e-07,
      "loss": 4.0791,
      "step": 22010
    },
    {
      "epoch": 0.8886152603956399,
      "grad_norm": 109.3659439086914,
      "learning_rate": 6.437167997744475e-07,
      "loss": 4.7515,
      "step": 22011
    },
    {
      "epoch": 0.8886556318126766,
      "grad_norm": 99.13375091552734,
      "learning_rate": 6.432553212055393e-07,
      "loss": 3.7588,
      "step": 22012
    },
    {
      "epoch": 0.8886960032297133,
      "grad_norm": 117.19674682617188,
      "learning_rate": 6.427940026151158e-07,
      "loss": 4.6316,
      "step": 22013
    },
    {
      "epoch": 0.8887363746467501,
      "grad_norm": 81.1776123046875,
      "learning_rate": 6.423328440110644e-07,
      "loss": 4.7512,
      "step": 22014
    },
    {
      "epoch": 0.8887767460637869,
      "grad_norm": 85.04167938232422,
      "learning_rate": 6.418718454012707e-07,
      "loss": 5.1834,
      "step": 22015
    },
    {
      "epoch": 0.8888171174808236,
      "grad_norm": 143.84925842285156,
      "learning_rate": 6.414110067936141e-07,
      "loss": 5.0751,
      "step": 22016
    },
    {
      "epoch": 0.8888574888978603,
      "grad_norm": 110.85453033447266,
      "learning_rate": 6.40950328195975e-07,
      "loss": 5.5449,
      "step": 22017
    },
    {
      "epoch": 0.888897860314897,
      "grad_norm": 86.29145812988281,
      "learning_rate": 6.404898096162315e-07,
      "loss": 7.0767,
      "step": 22018
    },
    {
      "epoch": 0.8889382317319338,
      "grad_norm": 102.82392120361328,
      "learning_rate": 6.400294510622574e-07,
      "loss": 5.1254,
      "step": 22019
    },
    {
      "epoch": 0.8889786031489705,
      "grad_norm": 119.5918960571289,
      "learning_rate": 6.395692525419206e-07,
      "loss": 6.3379,
      "step": 22020
    },
    {
      "epoch": 0.8890189745660073,
      "grad_norm": 106.64188385009766,
      "learning_rate": 6.391092140630917e-07,
      "loss": 4.7094,
      "step": 22021
    },
    {
      "epoch": 0.889059345983044,
      "grad_norm": 107.91250610351562,
      "learning_rate": 6.386493356336354e-07,
      "loss": 3.4863,
      "step": 22022
    },
    {
      "epoch": 0.8890997174000808,
      "grad_norm": 69.80248260498047,
      "learning_rate": 6.381896172614177e-07,
      "loss": 5.6102,
      "step": 22023
    },
    {
      "epoch": 0.8891400888171175,
      "grad_norm": 106.79476928710938,
      "learning_rate": 6.377300589542934e-07,
      "loss": 4.919,
      "step": 22024
    },
    {
      "epoch": 0.8891804602341542,
      "grad_norm": 89.5041275024414,
      "learning_rate": 6.37270660720124e-07,
      "loss": 3.7534,
      "step": 22025
    },
    {
      "epoch": 0.8892208316511909,
      "grad_norm": 94.58013916015625,
      "learning_rate": 6.36811422566762e-07,
      "loss": 5.65,
      "step": 22026
    },
    {
      "epoch": 0.8892612030682276,
      "grad_norm": 95.78425598144531,
      "learning_rate": 6.363523445020603e-07,
      "loss": 3.5454,
      "step": 22027
    },
    {
      "epoch": 0.8893015744852645,
      "grad_norm": 211.04513549804688,
      "learning_rate": 6.358934265338668e-07,
      "loss": 6.2975,
      "step": 22028
    },
    {
      "epoch": 0.8893419459023012,
      "grad_norm": 99.13294219970703,
      "learning_rate": 6.354346686700308e-07,
      "loss": 3.8363,
      "step": 22029
    },
    {
      "epoch": 0.8893823173193379,
      "grad_norm": 85.12696075439453,
      "learning_rate": 6.349760709183928e-07,
      "loss": 6.3062,
      "step": 22030
    },
    {
      "epoch": 0.8894226887363746,
      "grad_norm": 155.992919921875,
      "learning_rate": 6.345176332867964e-07,
      "loss": 5.0735,
      "step": 22031
    },
    {
      "epoch": 0.8894630601534114,
      "grad_norm": 84.78739166259766,
      "learning_rate": 6.340593557830777e-07,
      "loss": 6.7705,
      "step": 22032
    },
    {
      "epoch": 0.8895034315704481,
      "grad_norm": 162.46163940429688,
      "learning_rate": 6.336012384150725e-07,
      "loss": 4.6409,
      "step": 22033
    },
    {
      "epoch": 0.8895438029874848,
      "grad_norm": 170.6492462158203,
      "learning_rate": 6.331432811906146e-07,
      "loss": 5.4164,
      "step": 22034
    },
    {
      "epoch": 0.8895841744045216,
      "grad_norm": 78.24700927734375,
      "learning_rate": 6.326854841175345e-07,
      "loss": 4.2493,
      "step": 22035
    },
    {
      "epoch": 0.8896245458215584,
      "grad_norm": 85.00286102294922,
      "learning_rate": 6.32227847203657e-07,
      "loss": 4.8506,
      "step": 22036
    },
    {
      "epoch": 0.8896649172385951,
      "grad_norm": 108.76850891113281,
      "learning_rate": 6.317703704568079e-07,
      "loss": 4.7219,
      "step": 22037
    },
    {
      "epoch": 0.8897052886556318,
      "grad_norm": 86.0943603515625,
      "learning_rate": 6.3131305388481e-07,
      "loss": 4.2597,
      "step": 22038
    },
    {
      "epoch": 0.8897456600726685,
      "grad_norm": 136.9871368408203,
      "learning_rate": 6.308558974954815e-07,
      "loss": 5.2083,
      "step": 22039
    },
    {
      "epoch": 0.8897860314897053,
      "grad_norm": 145.57962036132812,
      "learning_rate": 6.303989012966383e-07,
      "loss": 5.1306,
      "step": 22040
    },
    {
      "epoch": 0.889826402906742,
      "grad_norm": 78.34488677978516,
      "learning_rate": 6.29942065296093e-07,
      "loss": 5.593,
      "step": 22041
    },
    {
      "epoch": 0.8898667743237788,
      "grad_norm": 111.66033935546875,
      "learning_rate": 6.294853895016583e-07,
      "loss": 4.0447,
      "step": 22042
    },
    {
      "epoch": 0.8899071457408155,
      "grad_norm": 110.33847045898438,
      "learning_rate": 6.290288739211436e-07,
      "loss": 6.1985,
      "step": 22043
    },
    {
      "epoch": 0.8899475171578523,
      "grad_norm": 160.02479553222656,
      "learning_rate": 6.285725185623503e-07,
      "loss": 4.473,
      "step": 22044
    },
    {
      "epoch": 0.889987888574889,
      "grad_norm": 139.03277587890625,
      "learning_rate": 6.281163234330834e-07,
      "loss": 5.0323,
      "step": 22045
    },
    {
      "epoch": 0.8900282599919257,
      "grad_norm": 101.65989685058594,
      "learning_rate": 6.276602885411432e-07,
      "loss": 5.606,
      "step": 22046
    },
    {
      "epoch": 0.8900686314089624,
      "grad_norm": 140.5269775390625,
      "learning_rate": 6.272044138943245e-07,
      "loss": 5.1351,
      "step": 22047
    },
    {
      "epoch": 0.8901090028259991,
      "grad_norm": 88.70951080322266,
      "learning_rate": 6.267486995004246e-07,
      "loss": 5.4766,
      "step": 22048
    },
    {
      "epoch": 0.890149374243036,
      "grad_norm": 119.98600769042969,
      "learning_rate": 6.262931453672328e-07,
      "loss": 4.5854,
      "step": 22049
    },
    {
      "epoch": 0.8901897456600727,
      "grad_norm": 146.29896545410156,
      "learning_rate": 6.258377515025393e-07,
      "loss": 3.6966,
      "step": 22050
    },
    {
      "epoch": 0.8902301170771094,
      "grad_norm": 94.21179962158203,
      "learning_rate": 6.253825179141327e-07,
      "loss": 3.8907,
      "step": 22051
    },
    {
      "epoch": 0.8902704884941461,
      "grad_norm": 94.73442840576172,
      "learning_rate": 6.249274446097908e-07,
      "loss": 3.7886,
      "step": 22052
    },
    {
      "epoch": 0.8903108599111829,
      "grad_norm": 115.58451080322266,
      "learning_rate": 6.244725315972977e-07,
      "loss": 4.3705,
      "step": 22053
    },
    {
      "epoch": 0.8903512313282196,
      "grad_norm": 69.85709381103516,
      "learning_rate": 6.240177788844304e-07,
      "loss": 2.0997,
      "step": 22054
    },
    {
      "epoch": 0.8903916027452563,
      "grad_norm": 90.01547241210938,
      "learning_rate": 6.235631864789671e-07,
      "loss": 6.5101,
      "step": 22055
    },
    {
      "epoch": 0.8904319741622931,
      "grad_norm": 80.02731323242188,
      "learning_rate": 6.23108754388676e-07,
      "loss": 4.6467,
      "step": 22056
    },
    {
      "epoch": 0.8904723455793299,
      "grad_norm": 86.45463562011719,
      "learning_rate": 6.226544826213276e-07,
      "loss": 3.9046,
      "step": 22057
    },
    {
      "epoch": 0.8905127169963666,
      "grad_norm": 88.40174102783203,
      "learning_rate": 6.222003711846913e-07,
      "loss": 5.2472,
      "step": 22058
    },
    {
      "epoch": 0.8905530884134033,
      "grad_norm": 126.06620025634766,
      "learning_rate": 6.217464200865308e-07,
      "loss": 5.6109,
      "step": 22059
    },
    {
      "epoch": 0.89059345983044,
      "grad_norm": 84.88575744628906,
      "learning_rate": 6.212926293346044e-07,
      "loss": 5.0204,
      "step": 22060
    },
    {
      "epoch": 0.8906338312474767,
      "grad_norm": 122.35784912109375,
      "learning_rate": 6.208389989366737e-07,
      "loss": 6.0216,
      "step": 22061
    },
    {
      "epoch": 0.8906742026645135,
      "grad_norm": 128.82647705078125,
      "learning_rate": 6.203855289004945e-07,
      "loss": 4.6922,
      "step": 22062
    },
    {
      "epoch": 0.8907145740815503,
      "grad_norm": 91.7864990234375,
      "learning_rate": 6.199322192338219e-07,
      "loss": 4.2721,
      "step": 22063
    },
    {
      "epoch": 0.890754945498587,
      "grad_norm": 93.72029876708984,
      "learning_rate": 6.19479069944402e-07,
      "loss": 4.7511,
      "step": 22064
    },
    {
      "epoch": 0.8907953169156237,
      "grad_norm": 106.28621673583984,
      "learning_rate": 6.190260810399851e-07,
      "loss": 6.7528,
      "step": 22065
    },
    {
      "epoch": 0.8908356883326605,
      "grad_norm": 955.6690063476562,
      "learning_rate": 6.185732525283161e-07,
      "loss": 5.0712,
      "step": 22066
    },
    {
      "epoch": 0.8908760597496972,
      "grad_norm": 117.06193542480469,
      "learning_rate": 6.181205844171367e-07,
      "loss": 6.2387,
      "step": 22067
    },
    {
      "epoch": 0.8909164311667339,
      "grad_norm": 59.38593673706055,
      "learning_rate": 6.176680767141874e-07,
      "loss": 3.3492,
      "step": 22068
    },
    {
      "epoch": 0.8909568025837706,
      "grad_norm": 104.59880065917969,
      "learning_rate": 6.172157294272052e-07,
      "loss": 4.909,
      "step": 22069
    },
    {
      "epoch": 0.8909971740008075,
      "grad_norm": 88.9273681640625,
      "learning_rate": 6.167635425639229e-07,
      "loss": 4.8429,
      "step": 22070
    },
    {
      "epoch": 0.8910375454178442,
      "grad_norm": 84.61991119384766,
      "learning_rate": 6.163115161320743e-07,
      "loss": 4.4602,
      "step": 22071
    },
    {
      "epoch": 0.8910779168348809,
      "grad_norm": 83.092529296875,
      "learning_rate": 6.158596501393854e-07,
      "loss": 5.9027,
      "step": 22072
    },
    {
      "epoch": 0.8911182882519176,
      "grad_norm": 95.98385620117188,
      "learning_rate": 6.154079445935824e-07,
      "loss": 4.4146,
      "step": 22073
    },
    {
      "epoch": 0.8911586596689544,
      "grad_norm": 88.8419189453125,
      "learning_rate": 6.149563995023889e-07,
      "loss": 5.3847,
      "step": 22074
    },
    {
      "epoch": 0.8911990310859911,
      "grad_norm": 102.9003677368164,
      "learning_rate": 6.145050148735265e-07,
      "loss": 6.6076,
      "step": 22075
    },
    {
      "epoch": 0.8912394025030279,
      "grad_norm": 100.07376861572266,
      "learning_rate": 6.140537907147104e-07,
      "loss": 4.9559,
      "step": 22076
    },
    {
      "epoch": 0.8912797739200646,
      "grad_norm": 69.59947204589844,
      "learning_rate": 6.136027270336576e-07,
      "loss": 4.7148,
      "step": 22077
    },
    {
      "epoch": 0.8913201453371014,
      "grad_norm": 83.61347961425781,
      "learning_rate": 6.131518238380785e-07,
      "loss": 5.9865,
      "step": 22078
    },
    {
      "epoch": 0.8913605167541381,
      "grad_norm": 114.04682922363281,
      "learning_rate": 6.127010811356848e-07,
      "loss": 5.8289,
      "step": 22079
    },
    {
      "epoch": 0.8914008881711748,
      "grad_norm": 151.96800231933594,
      "learning_rate": 6.122504989341804e-07,
      "loss": 4.3906,
      "step": 22080
    },
    {
      "epoch": 0.8914412595882115,
      "grad_norm": 116.59722137451172,
      "learning_rate": 6.118000772412702e-07,
      "loss": 4.0711,
      "step": 22081
    },
    {
      "epoch": 0.8914816310052482,
      "grad_norm": 90.63642120361328,
      "learning_rate": 6.113498160646558e-07,
      "loss": 3.9018,
      "step": 22082
    },
    {
      "epoch": 0.8915220024222851,
      "grad_norm": 136.89585876464844,
      "learning_rate": 6.108997154120366e-07,
      "loss": 4.2613,
      "step": 22083
    },
    {
      "epoch": 0.8915623738393218,
      "grad_norm": 82.07551574707031,
      "learning_rate": 6.104497752911065e-07,
      "loss": 3.2674,
      "step": 22084
    },
    {
      "epoch": 0.8916027452563585,
      "grad_norm": 138.2168731689453,
      "learning_rate": 6.099999957095593e-07,
      "loss": 6.0087,
      "step": 22085
    },
    {
      "epoch": 0.8916431166733952,
      "grad_norm": 84.57284545898438,
      "learning_rate": 6.095503766750843e-07,
      "loss": 4.4141,
      "step": 22086
    },
    {
      "epoch": 0.891683488090432,
      "grad_norm": 107.1231918334961,
      "learning_rate": 6.091009181953689e-07,
      "loss": 6.2359,
      "step": 22087
    },
    {
      "epoch": 0.8917238595074687,
      "grad_norm": 95.97657775878906,
      "learning_rate": 6.08651620278099e-07,
      "loss": 4.7117,
      "step": 22088
    },
    {
      "epoch": 0.8917642309245054,
      "grad_norm": 72.1729507446289,
      "learning_rate": 6.082024829309563e-07,
      "loss": 5.163,
      "step": 22089
    },
    {
      "epoch": 0.8918046023415422,
      "grad_norm": 84.54696655273438,
      "learning_rate": 6.077535061616202e-07,
      "loss": 4.7081,
      "step": 22090
    },
    {
      "epoch": 0.891844973758579,
      "grad_norm": 112.334716796875,
      "learning_rate": 6.073046899777668e-07,
      "loss": 5.2856,
      "step": 22091
    },
    {
      "epoch": 0.8918853451756157,
      "grad_norm": 74.763671875,
      "learning_rate": 6.0685603438707e-07,
      "loss": 6.2507,
      "step": 22092
    },
    {
      "epoch": 0.8919257165926524,
      "grad_norm": 92.10406494140625,
      "learning_rate": 6.064075393971991e-07,
      "loss": 4.3316,
      "step": 22093
    },
    {
      "epoch": 0.8919660880096891,
      "grad_norm": 103.1393814086914,
      "learning_rate": 6.059592050158236e-07,
      "loss": 3.6017,
      "step": 22094
    },
    {
      "epoch": 0.8920064594267259,
      "grad_norm": 74.6188735961914,
      "learning_rate": 6.055110312506107e-07,
      "loss": 3.5627,
      "step": 22095
    },
    {
      "epoch": 0.8920468308437626,
      "grad_norm": 114.07083129882812,
      "learning_rate": 6.05063018109221e-07,
      "loss": 4.1919,
      "step": 22096
    },
    {
      "epoch": 0.8920872022607994,
      "grad_norm": 80.7175064086914,
      "learning_rate": 6.046151655993138e-07,
      "loss": 4.4803,
      "step": 22097
    },
    {
      "epoch": 0.8921275736778361,
      "grad_norm": 110.3361587524414,
      "learning_rate": 6.041674737285475e-07,
      "loss": 5.3392,
      "step": 22098
    },
    {
      "epoch": 0.8921679450948728,
      "grad_norm": 110.23013305664062,
      "learning_rate": 6.037199425045781e-07,
      "loss": 3.6111,
      "step": 22099
    },
    {
      "epoch": 0.8922083165119096,
      "grad_norm": 109.00849151611328,
      "learning_rate": 6.032725719350552e-07,
      "loss": 6.0918,
      "step": 22100
    },
    {
      "epoch": 0.8922486879289463,
      "grad_norm": 102.18498992919922,
      "learning_rate": 6.02825362027627e-07,
      "loss": 4.8321,
      "step": 22101
    },
    {
      "epoch": 0.892289059345983,
      "grad_norm": 84.00743103027344,
      "learning_rate": 6.023783127899418e-07,
      "loss": 4.8532,
      "step": 22102
    },
    {
      "epoch": 0.8923294307630197,
      "grad_norm": 94.16035461425781,
      "learning_rate": 6.019314242296448e-07,
      "loss": 5.316,
      "step": 22103
    },
    {
      "epoch": 0.8923698021800566,
      "grad_norm": 89.09208679199219,
      "learning_rate": 6.014846963543718e-07,
      "loss": 4.2211,
      "step": 22104
    },
    {
      "epoch": 0.8924101735970933,
      "grad_norm": 80.44597625732422,
      "learning_rate": 6.010381291717638e-07,
      "loss": 4.6022,
      "step": 22105
    },
    {
      "epoch": 0.89245054501413,
      "grad_norm": 65.69156646728516,
      "learning_rate": 6.005917226894553e-07,
      "loss": 2.5588,
      "step": 22106
    },
    {
      "epoch": 0.8924909164311667,
      "grad_norm": 77.02264404296875,
      "learning_rate": 6.001454769150794e-07,
      "loss": 5.5454,
      "step": 22107
    },
    {
      "epoch": 0.8925312878482035,
      "grad_norm": 119.95105743408203,
      "learning_rate": 5.996993918562655e-07,
      "loss": 5.8846,
      "step": 22108
    },
    {
      "epoch": 0.8925716592652402,
      "grad_norm": 145.37391662597656,
      "learning_rate": 5.992534675206408e-07,
      "loss": 6.2899,
      "step": 22109
    },
    {
      "epoch": 0.8926120306822769,
      "grad_norm": 92.69239044189453,
      "learning_rate": 5.98807703915829e-07,
      "loss": 5.7522,
      "step": 22110
    },
    {
      "epoch": 0.8926524020993137,
      "grad_norm": 158.17762756347656,
      "learning_rate": 5.983621010494534e-07,
      "loss": 6.7908,
      "step": 22111
    },
    {
      "epoch": 0.8926927735163505,
      "grad_norm": 108.51716613769531,
      "learning_rate": 5.979166589291308e-07,
      "loss": 4.5157,
      "step": 22112
    },
    {
      "epoch": 0.8927331449333872,
      "grad_norm": 87.5414047241211,
      "learning_rate": 5.974713775624775e-07,
      "loss": 5.2843,
      "step": 22113
    },
    {
      "epoch": 0.8927735163504239,
      "grad_norm": 104.27908325195312,
      "learning_rate": 5.970262569571072e-07,
      "loss": 5.8662,
      "step": 22114
    },
    {
      "epoch": 0.8928138877674606,
      "grad_norm": 91.4066162109375,
      "learning_rate": 5.965812971206331e-07,
      "loss": 4.2622,
      "step": 22115
    },
    {
      "epoch": 0.8928542591844973,
      "grad_norm": 95.49861907958984,
      "learning_rate": 5.961364980606577e-07,
      "loss": 3.3944,
      "step": 22116
    },
    {
      "epoch": 0.8928946306015341,
      "grad_norm": 105.07048034667969,
      "learning_rate": 5.956918597847883e-07,
      "loss": 4.9606,
      "step": 22117
    },
    {
      "epoch": 0.8929350020185709,
      "grad_norm": 84.935302734375,
      "learning_rate": 5.952473823006277e-07,
      "loss": 5.1632,
      "step": 22118
    },
    {
      "epoch": 0.8929753734356076,
      "grad_norm": 116.32563018798828,
      "learning_rate": 5.948030656157767e-07,
      "loss": 4.4382,
      "step": 22119
    },
    {
      "epoch": 0.8930157448526443,
      "grad_norm": 110.64112854003906,
      "learning_rate": 5.943589097378288e-07,
      "loss": 3.6749,
      "step": 22120
    },
    {
      "epoch": 0.8930561162696811,
      "grad_norm": 90.16870880126953,
      "learning_rate": 5.939149146743794e-07,
      "loss": 3.7708,
      "step": 22121
    },
    {
      "epoch": 0.8930964876867178,
      "grad_norm": 89.73616790771484,
      "learning_rate": 5.934710804330213e-07,
      "loss": 3.7587,
      "step": 22122
    },
    {
      "epoch": 0.8931368591037545,
      "grad_norm": 200.02842712402344,
      "learning_rate": 5.930274070213427e-07,
      "loss": 6.0048,
      "step": 22123
    },
    {
      "epoch": 0.8931772305207912,
      "grad_norm": 90.4449691772461,
      "learning_rate": 5.925838944469276e-07,
      "loss": 3.3949,
      "step": 22124
    },
    {
      "epoch": 0.8932176019378281,
      "grad_norm": 122.59254455566406,
      "learning_rate": 5.921405427173598e-07,
      "loss": 6.411,
      "step": 22125
    },
    {
      "epoch": 0.8932579733548648,
      "grad_norm": 105.7110595703125,
      "learning_rate": 5.916973518402191e-07,
      "loss": 5.1555,
      "step": 22126
    },
    {
      "epoch": 0.8932983447719015,
      "grad_norm": 97.70176696777344,
      "learning_rate": 5.912543218230848e-07,
      "loss": 4.6083,
      "step": 22127
    },
    {
      "epoch": 0.8933387161889382,
      "grad_norm": 120.3316879272461,
      "learning_rate": 5.908114526735297e-07,
      "loss": 5.1322,
      "step": 22128
    },
    {
      "epoch": 0.893379087605975,
      "grad_norm": 69.94747161865234,
      "learning_rate": 5.90368744399128e-07,
      "loss": 4.2982,
      "step": 22129
    },
    {
      "epoch": 0.8934194590230117,
      "grad_norm": 91.56261444091797,
      "learning_rate": 5.899261970074466e-07,
      "loss": 5.4314,
      "step": 22130
    },
    {
      "epoch": 0.8934598304400484,
      "grad_norm": 113.77702331542969,
      "learning_rate": 5.894838105060551e-07,
      "loss": 5.9467,
      "step": 22131
    },
    {
      "epoch": 0.8935002018570852,
      "grad_norm": 86.09317016601562,
      "learning_rate": 5.890415849025132e-07,
      "loss": 4.9482,
      "step": 22132
    },
    {
      "epoch": 0.893540573274122,
      "grad_norm": 115.09226989746094,
      "learning_rate": 5.885995202043848e-07,
      "loss": 5.646,
      "step": 22133
    },
    {
      "epoch": 0.8935809446911587,
      "grad_norm": 179.82977294921875,
      "learning_rate": 5.88157616419226e-07,
      "loss": 5.2378,
      "step": 22134
    },
    {
      "epoch": 0.8936213161081954,
      "grad_norm": 100.40093231201172,
      "learning_rate": 5.877158735545963e-07,
      "loss": 5.1561,
      "step": 22135
    },
    {
      "epoch": 0.8936616875252321,
      "grad_norm": 96.71932983398438,
      "learning_rate": 5.872742916180452e-07,
      "loss": 4.1174,
      "step": 22136
    },
    {
      "epoch": 0.8937020589422688,
      "grad_norm": 88.01446533203125,
      "learning_rate": 5.868328706171222e-07,
      "loss": 5.935,
      "step": 22137
    },
    {
      "epoch": 0.8937424303593056,
      "grad_norm": 78.20186614990234,
      "learning_rate": 5.863916105593759e-07,
      "loss": 2.8832,
      "step": 22138
    },
    {
      "epoch": 0.8937828017763424,
      "grad_norm": 97.947021484375,
      "learning_rate": 5.859505114523523e-07,
      "loss": 4.6297,
      "step": 22139
    },
    {
      "epoch": 0.8938231731933791,
      "grad_norm": 71.3355941772461,
      "learning_rate": 5.85509573303591e-07,
      "loss": 4.7384,
      "step": 22140
    },
    {
      "epoch": 0.8938635446104158,
      "grad_norm": 118.87326049804688,
      "learning_rate": 5.850687961206303e-07,
      "loss": 4.8472,
      "step": 22141
    },
    {
      "epoch": 0.8939039160274526,
      "grad_norm": 86.61434936523438,
      "learning_rate": 5.846281799110087e-07,
      "loss": 5.1296,
      "step": 22142
    },
    {
      "epoch": 0.8939442874444893,
      "grad_norm": 72.60541534423828,
      "learning_rate": 5.841877246822592e-07,
      "loss": 4.8374,
      "step": 22143
    },
    {
      "epoch": 0.893984658861526,
      "grad_norm": 72.21868896484375,
      "learning_rate": 5.837474304419111e-07,
      "loss": 3.2975,
      "step": 22144
    },
    {
      "epoch": 0.8940250302785627,
      "grad_norm": 82.2406997680664,
      "learning_rate": 5.83307297197494e-07,
      "loss": 5.438,
      "step": 22145
    },
    {
      "epoch": 0.8940654016955996,
      "grad_norm": 55.5771598815918,
      "learning_rate": 5.82867324956532e-07,
      "loss": 3.0633,
      "step": 22146
    },
    {
      "epoch": 0.8941057731126363,
      "grad_norm": 79.67669677734375,
      "learning_rate": 5.824275137265478e-07,
      "loss": 4.2181,
      "step": 22147
    },
    {
      "epoch": 0.894146144529673,
      "grad_norm": 127.01692962646484,
      "learning_rate": 5.819878635150622e-07,
      "loss": 6.6232,
      "step": 22148
    },
    {
      "epoch": 0.8941865159467097,
      "grad_norm": 164.71078491210938,
      "learning_rate": 5.815483743295913e-07,
      "loss": 5.2777,
      "step": 22149
    },
    {
      "epoch": 0.8942268873637464,
      "grad_norm": 103.77708435058594,
      "learning_rate": 5.811090461776492e-07,
      "loss": 4.7248,
      "step": 22150
    },
    {
      "epoch": 0.8942672587807832,
      "grad_norm": 80.8008041381836,
      "learning_rate": 5.806698790667498e-07,
      "loss": 5.6743,
      "step": 22151
    },
    {
      "epoch": 0.8943076301978199,
      "grad_norm": 107.16869354248047,
      "learning_rate": 5.802308730043971e-07,
      "loss": 5.201,
      "step": 22152
    },
    {
      "epoch": 0.8943480016148567,
      "grad_norm": 100.68659973144531,
      "learning_rate": 5.797920279981007e-07,
      "loss": 4.3144,
      "step": 22153
    },
    {
      "epoch": 0.8943883730318934,
      "grad_norm": 93.8189697265625,
      "learning_rate": 5.793533440553634e-07,
      "loss": 4.7501,
      "step": 22154
    },
    {
      "epoch": 0.8944287444489302,
      "grad_norm": 98.03131103515625,
      "learning_rate": 5.789148211836859e-07,
      "loss": 3.0622,
      "step": 22155
    },
    {
      "epoch": 0.8944691158659669,
      "grad_norm": 134.03817749023438,
      "learning_rate": 5.784764593905634e-07,
      "loss": 6.5182,
      "step": 22156
    },
    {
      "epoch": 0.8945094872830036,
      "grad_norm": 86.46245574951172,
      "learning_rate": 5.780382586834932e-07,
      "loss": 4.0147,
      "step": 22157
    },
    {
      "epoch": 0.8945498587000403,
      "grad_norm": 103.28498840332031,
      "learning_rate": 5.776002190699659e-07,
      "loss": 4.411,
      "step": 22158
    },
    {
      "epoch": 0.8945902301170771,
      "grad_norm": 131.64585876464844,
      "learning_rate": 5.771623405574744e-07,
      "loss": 6.567,
      "step": 22159
    },
    {
      "epoch": 0.8946306015341139,
      "grad_norm": 86.7354965209961,
      "learning_rate": 5.767246231535007e-07,
      "loss": 3.8298,
      "step": 22160
    },
    {
      "epoch": 0.8946709729511506,
      "grad_norm": 84.4815673828125,
      "learning_rate": 5.762870668655318e-07,
      "loss": 5.2363,
      "step": 22161
    },
    {
      "epoch": 0.8947113443681873,
      "grad_norm": 80.18733215332031,
      "learning_rate": 5.758496717010476e-07,
      "loss": 4.8389,
      "step": 22162
    },
    {
      "epoch": 0.894751715785224,
      "grad_norm": 70.78968811035156,
      "learning_rate": 5.754124376675297e-07,
      "loss": 4.4766,
      "step": 22163
    },
    {
      "epoch": 0.8947920872022608,
      "grad_norm": 132.519775390625,
      "learning_rate": 5.749753647724477e-07,
      "loss": 5.0701,
      "step": 22164
    },
    {
      "epoch": 0.8948324586192975,
      "grad_norm": 91.75542449951172,
      "learning_rate": 5.745384530232789e-07,
      "loss": 5.5403,
      "step": 22165
    },
    {
      "epoch": 0.8948728300363342,
      "grad_norm": 81.71941375732422,
      "learning_rate": 5.74101702427493e-07,
      "loss": 4.3734,
      "step": 22166
    },
    {
      "epoch": 0.894913201453371,
      "grad_norm": 74.36247253417969,
      "learning_rate": 5.736651129925564e-07,
      "loss": 5.6238,
      "step": 22167
    },
    {
      "epoch": 0.8949535728704078,
      "grad_norm": 107.64109802246094,
      "learning_rate": 5.732286847259349e-07,
      "loss": 5.473,
      "step": 22168
    },
    {
      "epoch": 0.8949939442874445,
      "grad_norm": 108.93636322021484,
      "learning_rate": 5.727924176350885e-07,
      "loss": 5.2363,
      "step": 22169
    },
    {
      "epoch": 0.8950343157044812,
      "grad_norm": 93.08174133300781,
      "learning_rate": 5.723563117274788e-07,
      "loss": 6.1808,
      "step": 22170
    },
    {
      "epoch": 0.8950746871215179,
      "grad_norm": 80.45684814453125,
      "learning_rate": 5.719203670105622e-07,
      "loss": 5.2281,
      "step": 22171
    },
    {
      "epoch": 0.8951150585385547,
      "grad_norm": 145.68833923339844,
      "learning_rate": 5.714845834917893e-07,
      "loss": 5.5247,
      "step": 22172
    },
    {
      "epoch": 0.8951554299555914,
      "grad_norm": 84.55091857910156,
      "learning_rate": 5.710489611786141e-07,
      "loss": 4.0956,
      "step": 22173
    },
    {
      "epoch": 0.8951958013726282,
      "grad_norm": 104.09212493896484,
      "learning_rate": 5.706135000784818e-07,
      "loss": 4.3207,
      "step": 22174
    },
    {
      "epoch": 0.8952361727896649,
      "grad_norm": 77.84410858154297,
      "learning_rate": 5.701782001988421e-07,
      "loss": 5.0494,
      "step": 22175
    },
    {
      "epoch": 0.8952765442067017,
      "grad_norm": 93.03956604003906,
      "learning_rate": 5.697430615471334e-07,
      "loss": 5.6748,
      "step": 22176
    },
    {
      "epoch": 0.8953169156237384,
      "grad_norm": 66.15117645263672,
      "learning_rate": 5.693080841307963e-07,
      "loss": 4.8119,
      "step": 22177
    },
    {
      "epoch": 0.8953572870407751,
      "grad_norm": 154.1392364501953,
      "learning_rate": 5.688732679572695e-07,
      "loss": 4.1385,
      "step": 22178
    },
    {
      "epoch": 0.8953976584578118,
      "grad_norm": 99.91673278808594,
      "learning_rate": 5.684386130339881e-07,
      "loss": 5.1942,
      "step": 22179
    },
    {
      "epoch": 0.8954380298748487,
      "grad_norm": 63.95222854614258,
      "learning_rate": 5.680041193683794e-07,
      "loss": 4.6687,
      "step": 22180
    },
    {
      "epoch": 0.8954784012918854,
      "grad_norm": 101.39496612548828,
      "learning_rate": 5.675697869678764e-07,
      "loss": 4.9492,
      "step": 22181
    },
    {
      "epoch": 0.8955187727089221,
      "grad_norm": 172.38043212890625,
      "learning_rate": 5.671356158399033e-07,
      "loss": 4.9391,
      "step": 22182
    },
    {
      "epoch": 0.8955591441259588,
      "grad_norm": 166.6881866455078,
      "learning_rate": 5.66701605991884e-07,
      "loss": 4.2181,
      "step": 22183
    },
    {
      "epoch": 0.8955995155429956,
      "grad_norm": 114.78488159179688,
      "learning_rate": 5.66267757431238e-07,
      "loss": 4.6295,
      "step": 22184
    },
    {
      "epoch": 0.8956398869600323,
      "grad_norm": 102.63912200927734,
      "learning_rate": 5.658340701653853e-07,
      "loss": 4.3808,
      "step": 22185
    },
    {
      "epoch": 0.895680258377069,
      "grad_norm": 76.43671417236328,
      "learning_rate": 5.654005442017396e-07,
      "loss": 2.6991,
      "step": 22186
    },
    {
      "epoch": 0.8957206297941058,
      "grad_norm": 127.16573333740234,
      "learning_rate": 5.649671795477129e-07,
      "loss": 6.61,
      "step": 22187
    },
    {
      "epoch": 0.8957610012111425,
      "grad_norm": 119.16141510009766,
      "learning_rate": 5.64533976210715e-07,
      "loss": 3.8602,
      "step": 22188
    },
    {
      "epoch": 0.8958013726281793,
      "grad_norm": 88.2501449584961,
      "learning_rate": 5.641009341981529e-07,
      "loss": 4.7733,
      "step": 22189
    },
    {
      "epoch": 0.895841744045216,
      "grad_norm": 100.12330627441406,
      "learning_rate": 5.636680535174299e-07,
      "loss": 4.7688,
      "step": 22190
    },
    {
      "epoch": 0.8958821154622527,
      "grad_norm": 118.1636734008789,
      "learning_rate": 5.632353341759489e-07,
      "loss": 5.1899,
      "step": 22191
    },
    {
      "epoch": 0.8959224868792894,
      "grad_norm": 178.2654266357422,
      "learning_rate": 5.628027761811062e-07,
      "loss": 6.0357,
      "step": 22192
    },
    {
      "epoch": 0.8959628582963262,
      "grad_norm": 91.84559631347656,
      "learning_rate": 5.62370379540299e-07,
      "loss": 5.3834,
      "step": 22193
    },
    {
      "epoch": 0.896003229713363,
      "grad_norm": 97.54778289794922,
      "learning_rate": 5.619381442609196e-07,
      "loss": 5.1837,
      "step": 22194
    },
    {
      "epoch": 0.8960436011303997,
      "grad_norm": 91.98741912841797,
      "learning_rate": 5.615060703503594e-07,
      "loss": 4.7656,
      "step": 22195
    },
    {
      "epoch": 0.8960839725474364,
      "grad_norm": 91.03131866455078,
      "learning_rate": 5.61074157816004e-07,
      "loss": 6.2319,
      "step": 22196
    },
    {
      "epoch": 0.8961243439644732,
      "grad_norm": 87.65833282470703,
      "learning_rate": 5.606424066652383e-07,
      "loss": 3.5114,
      "step": 22197
    },
    {
      "epoch": 0.8961647153815099,
      "grad_norm": 113.56822204589844,
      "learning_rate": 5.602108169054454e-07,
      "loss": 5.9342,
      "step": 22198
    },
    {
      "epoch": 0.8962050867985466,
      "grad_norm": 77.0754623413086,
      "learning_rate": 5.597793885440062e-07,
      "loss": 3.4516,
      "step": 22199
    },
    {
      "epoch": 0.8962454582155833,
      "grad_norm": 105.04750061035156,
      "learning_rate": 5.593481215882934e-07,
      "loss": 5.9242,
      "step": 22200
    },
    {
      "epoch": 0.8962858296326202,
      "grad_norm": 97.32230377197266,
      "learning_rate": 5.589170160456814e-07,
      "loss": 5.4525,
      "step": 22201
    },
    {
      "epoch": 0.8963262010496569,
      "grad_norm": 226.29307556152344,
      "learning_rate": 5.584860719235418e-07,
      "loss": 3.758,
      "step": 22202
    },
    {
      "epoch": 0.8963665724666936,
      "grad_norm": 89.91986083984375,
      "learning_rate": 5.580552892292435e-07,
      "loss": 6.0952,
      "step": 22203
    },
    {
      "epoch": 0.8964069438837303,
      "grad_norm": 103.68740844726562,
      "learning_rate": 5.576246679701514e-07,
      "loss": 3.7889,
      "step": 22204
    },
    {
      "epoch": 0.896447315300767,
      "grad_norm": 93.65164947509766,
      "learning_rate": 5.571942081536274e-07,
      "loss": 6.1514,
      "step": 22205
    },
    {
      "epoch": 0.8964876867178038,
      "grad_norm": 89.08332824707031,
      "learning_rate": 5.567639097870326e-07,
      "loss": 4.9127,
      "step": 22206
    },
    {
      "epoch": 0.8965280581348405,
      "grad_norm": 104.24264526367188,
      "learning_rate": 5.563337728777241e-07,
      "loss": 3.7393,
      "step": 22207
    },
    {
      "epoch": 0.8965684295518773,
      "grad_norm": 102.16027069091797,
      "learning_rate": 5.559037974330539e-07,
      "loss": 5.3456,
      "step": 22208
    },
    {
      "epoch": 0.896608800968914,
      "grad_norm": 112.33320617675781,
      "learning_rate": 5.554739834603762e-07,
      "loss": 4.7421,
      "step": 22209
    },
    {
      "epoch": 0.8966491723859508,
      "grad_norm": 84.81233215332031,
      "learning_rate": 5.550443309670383e-07,
      "loss": 4.7765,
      "step": 22210
    },
    {
      "epoch": 0.8966895438029875,
      "grad_norm": 103.91699981689453,
      "learning_rate": 5.546148399603879e-07,
      "loss": 5.0243,
      "step": 22211
    },
    {
      "epoch": 0.8967299152200242,
      "grad_norm": 86.77416229248047,
      "learning_rate": 5.541855104477656e-07,
      "loss": 6.317,
      "step": 22212
    },
    {
      "epoch": 0.8967702866370609,
      "grad_norm": 183.5253143310547,
      "learning_rate": 5.537563424365144e-07,
      "loss": 6.4313,
      "step": 22213
    },
    {
      "epoch": 0.8968106580540977,
      "grad_norm": 110.17927551269531,
      "learning_rate": 5.533273359339708e-07,
      "loss": 5.8322,
      "step": 22214
    },
    {
      "epoch": 0.8968510294711345,
      "grad_norm": 68.12606811523438,
      "learning_rate": 5.528984909474711e-07,
      "loss": 4.0585,
      "step": 22215
    },
    {
      "epoch": 0.8968914008881712,
      "grad_norm": 79.74324035644531,
      "learning_rate": 5.52469807484346e-07,
      "loss": 5.4121,
      "step": 22216
    },
    {
      "epoch": 0.8969317723052079,
      "grad_norm": 95.8720932006836,
      "learning_rate": 5.520412855519253e-07,
      "loss": 4.9092,
      "step": 22217
    },
    {
      "epoch": 0.8969721437222447,
      "grad_norm": 78.61936950683594,
      "learning_rate": 5.516129251575364e-07,
      "loss": 5.67,
      "step": 22218
    },
    {
      "epoch": 0.8970125151392814,
      "grad_norm": 74.99798583984375,
      "learning_rate": 5.511847263085035e-07,
      "loss": 6.7751,
      "step": 22219
    },
    {
      "epoch": 0.8970528865563181,
      "grad_norm": 125.15117645263672,
      "learning_rate": 5.507566890121463e-07,
      "loss": 4.9666,
      "step": 22220
    },
    {
      "epoch": 0.8970932579733548,
      "grad_norm": 80.82162475585938,
      "learning_rate": 5.503288132757845e-07,
      "loss": 5.8546,
      "step": 22221
    },
    {
      "epoch": 0.8971336293903917,
      "grad_norm": 140.84329223632812,
      "learning_rate": 5.499010991067322e-07,
      "loss": 3.645,
      "step": 22222
    },
    {
      "epoch": 0.8971740008074284,
      "grad_norm": 88.91178894042969,
      "learning_rate": 5.494735465123046e-07,
      "loss": 2.7578,
      "step": 22223
    },
    {
      "epoch": 0.8972143722244651,
      "grad_norm": 92.39681243896484,
      "learning_rate": 5.490461554998105e-07,
      "loss": 5.641,
      "step": 22224
    },
    {
      "epoch": 0.8972547436415018,
      "grad_norm": 99.76556396484375,
      "learning_rate": 5.486189260765573e-07,
      "loss": 6.5023,
      "step": 22225
    },
    {
      "epoch": 0.8972951150585385,
      "grad_norm": 69.6183853149414,
      "learning_rate": 5.481918582498503e-07,
      "loss": 6.1981,
      "step": 22226
    },
    {
      "epoch": 0.8973354864755753,
      "grad_norm": 80.30249786376953,
      "learning_rate": 5.477649520269923e-07,
      "loss": 4.2798,
      "step": 22227
    },
    {
      "epoch": 0.897375857892612,
      "grad_norm": 74.56135559082031,
      "learning_rate": 5.4733820741528e-07,
      "loss": 4.7708,
      "step": 22228
    },
    {
      "epoch": 0.8974162293096488,
      "grad_norm": 70.6631088256836,
      "learning_rate": 5.469116244220107e-07,
      "loss": 3.7474,
      "step": 22229
    },
    {
      "epoch": 0.8974566007266855,
      "grad_norm": 87.0374526977539,
      "learning_rate": 5.464852030544776e-07,
      "loss": 4.7611,
      "step": 22230
    },
    {
      "epoch": 0.8974969721437223,
      "grad_norm": 90.99645233154297,
      "learning_rate": 5.460589433199737e-07,
      "loss": 6.1434,
      "step": 22231
    },
    {
      "epoch": 0.897537343560759,
      "grad_norm": 81.85359191894531,
      "learning_rate": 5.456328452257842e-07,
      "loss": 5.3003,
      "step": 22232
    },
    {
      "epoch": 0.8975777149777957,
      "grad_norm": 117.08990478515625,
      "learning_rate": 5.452069087791945e-07,
      "loss": 5.6699,
      "step": 22233
    },
    {
      "epoch": 0.8976180863948324,
      "grad_norm": 87.34859466552734,
      "learning_rate": 5.447811339874887e-07,
      "loss": 5.6839,
      "step": 22234
    },
    {
      "epoch": 0.8976584578118691,
      "grad_norm": 78.49981689453125,
      "learning_rate": 5.443555208579476e-07,
      "loss": 5.8575,
      "step": 22235
    },
    {
      "epoch": 0.897698829228906,
      "grad_norm": 106.21318054199219,
      "learning_rate": 5.439300693978456e-07,
      "loss": 4.7413,
      "step": 22236
    },
    {
      "epoch": 0.8977392006459427,
      "grad_norm": 90.10127258300781,
      "learning_rate": 5.435047796144576e-07,
      "loss": 5.0719,
      "step": 22237
    },
    {
      "epoch": 0.8977795720629794,
      "grad_norm": 70.25289916992188,
      "learning_rate": 5.430796515150549e-07,
      "loss": 3.7134,
      "step": 22238
    },
    {
      "epoch": 0.8978199434800161,
      "grad_norm": 104.18506622314453,
      "learning_rate": 5.426546851069081e-07,
      "loss": 3.4647,
      "step": 22239
    },
    {
      "epoch": 0.8978603148970529,
      "grad_norm": 89.51729583740234,
      "learning_rate": 5.422298803972803e-07,
      "loss": 6.3195,
      "step": 22240
    },
    {
      "epoch": 0.8979006863140896,
      "grad_norm": 89.659912109375,
      "learning_rate": 5.418052373934368e-07,
      "loss": 5.5541,
      "step": 22241
    },
    {
      "epoch": 0.8979410577311263,
      "grad_norm": 140.81227111816406,
      "learning_rate": 5.413807561026363e-07,
      "loss": 5.8521,
      "step": 22242
    },
    {
      "epoch": 0.8979814291481631,
      "grad_norm": 171.41964721679688,
      "learning_rate": 5.409564365321374e-07,
      "loss": 5.4986,
      "step": 22243
    },
    {
      "epoch": 0.8980218005651999,
      "grad_norm": 154.88804626464844,
      "learning_rate": 5.405322786891953e-07,
      "loss": 3.4203,
      "step": 22244
    },
    {
      "epoch": 0.8980621719822366,
      "grad_norm": 115.21269989013672,
      "learning_rate": 5.401082825810622e-07,
      "loss": 4.9647,
      "step": 22245
    },
    {
      "epoch": 0.8981025433992733,
      "grad_norm": 80.23225402832031,
      "learning_rate": 5.396844482149865e-07,
      "loss": 5.1429,
      "step": 22246
    },
    {
      "epoch": 0.89814291481631,
      "grad_norm": 137.25559997558594,
      "learning_rate": 5.392607755982171e-07,
      "loss": 4.6836,
      "step": 22247
    },
    {
      "epoch": 0.8981832862333468,
      "grad_norm": 123.99828338623047,
      "learning_rate": 5.388372647379947e-07,
      "loss": 4.4833,
      "step": 22248
    },
    {
      "epoch": 0.8982236576503835,
      "grad_norm": 111.4687271118164,
      "learning_rate": 5.384139156415614e-07,
      "loss": 4.0082,
      "step": 22249
    },
    {
      "epoch": 0.8982640290674203,
      "grad_norm": 105.80667114257812,
      "learning_rate": 5.379907283161556e-07,
      "loss": 6.0127,
      "step": 22250
    },
    {
      "epoch": 0.898304400484457,
      "grad_norm": 119.85104370117188,
      "learning_rate": 5.375677027690152e-07,
      "loss": 6.2135,
      "step": 22251
    },
    {
      "epoch": 0.8983447719014938,
      "grad_norm": 115.16409301757812,
      "learning_rate": 5.371448390073686e-07,
      "loss": 4.1785,
      "step": 22252
    },
    {
      "epoch": 0.8983851433185305,
      "grad_norm": 108.15596771240234,
      "learning_rate": 5.367221370384479e-07,
      "loss": 4.1359,
      "step": 22253
    },
    {
      "epoch": 0.8984255147355672,
      "grad_norm": 77.349365234375,
      "learning_rate": 5.362995968694795e-07,
      "loss": 5.4207,
      "step": 22254
    },
    {
      "epoch": 0.8984658861526039,
      "grad_norm": 95.63055419921875,
      "learning_rate": 5.358772185076911e-07,
      "loss": 4.3753,
      "step": 22255
    },
    {
      "epoch": 0.8985062575696406,
      "grad_norm": 119.73139953613281,
      "learning_rate": 5.354550019603011e-07,
      "loss": 4.2643,
      "step": 22256
    },
    {
      "epoch": 0.8985466289866775,
      "grad_norm": 102.61849212646484,
      "learning_rate": 5.350329472345283e-07,
      "loss": 3.4476,
      "step": 22257
    },
    {
      "epoch": 0.8985870004037142,
      "grad_norm": 78.03978729248047,
      "learning_rate": 5.346110543375904e-07,
      "loss": 5.339,
      "step": 22258
    },
    {
      "epoch": 0.8986273718207509,
      "grad_norm": 119.1361312866211,
      "learning_rate": 5.341893232767004e-07,
      "loss": 4.0431,
      "step": 22259
    },
    {
      "epoch": 0.8986677432377876,
      "grad_norm": 83.64266967773438,
      "learning_rate": 5.337677540590691e-07,
      "loss": 5.1181,
      "step": 22260
    },
    {
      "epoch": 0.8987081146548244,
      "grad_norm": 87.9653549194336,
      "learning_rate": 5.33346346691902e-07,
      "loss": 4.923,
      "step": 22261
    },
    {
      "epoch": 0.8987484860718611,
      "grad_norm": 60.81095504760742,
      "learning_rate": 5.329251011824077e-07,
      "loss": 4.4444,
      "step": 22262
    },
    {
      "epoch": 0.8987888574888978,
      "grad_norm": 82.35003662109375,
      "learning_rate": 5.32504017537786e-07,
      "loss": 4.5434,
      "step": 22263
    },
    {
      "epoch": 0.8988292289059346,
      "grad_norm": 90.72843933105469,
      "learning_rate": 5.320830957652379e-07,
      "loss": 5.2243,
      "step": 22264
    },
    {
      "epoch": 0.8988696003229714,
      "grad_norm": 66.90219116210938,
      "learning_rate": 5.316623358719597e-07,
      "loss": 2.812,
      "step": 22265
    },
    {
      "epoch": 0.8989099717400081,
      "grad_norm": 89.84307861328125,
      "learning_rate": 5.312417378651458e-07,
      "loss": 5.3128,
      "step": 22266
    },
    {
      "epoch": 0.8989503431570448,
      "grad_norm": 80.75397491455078,
      "learning_rate": 5.308213017519881e-07,
      "loss": 3.7578,
      "step": 22267
    },
    {
      "epoch": 0.8989907145740815,
      "grad_norm": 110.78396606445312,
      "learning_rate": 5.304010275396721e-07,
      "loss": 5.1588,
      "step": 22268
    },
    {
      "epoch": 0.8990310859911183,
      "grad_norm": 95.44877624511719,
      "learning_rate": 5.299809152353863e-07,
      "loss": 6.1491,
      "step": 22269
    },
    {
      "epoch": 0.899071457408155,
      "grad_norm": 66.64427185058594,
      "learning_rate": 5.295609648463118e-07,
      "loss": 4.1425,
      "step": 22270
    },
    {
      "epoch": 0.8991118288251918,
      "grad_norm": 88.85433959960938,
      "learning_rate": 5.291411763796317e-07,
      "loss": 5.5473,
      "step": 22271
    },
    {
      "epoch": 0.8991522002422285,
      "grad_norm": 80.24382019042969,
      "learning_rate": 5.287215498425202e-07,
      "loss": 4.4824,
      "step": 22272
    },
    {
      "epoch": 0.8991925716592652,
      "grad_norm": 76.54267120361328,
      "learning_rate": 5.283020852421528e-07,
      "loss": 3.4553,
      "step": 22273
    },
    {
      "epoch": 0.899232943076302,
      "grad_norm": 88.29766082763672,
      "learning_rate": 5.278827825857013e-07,
      "loss": 4.7914,
      "step": 22274
    },
    {
      "epoch": 0.8992733144933387,
      "grad_norm": 102.38438415527344,
      "learning_rate": 5.27463641880338e-07,
      "loss": 5.9077,
      "step": 22275
    },
    {
      "epoch": 0.8993136859103754,
      "grad_norm": 122.11541748046875,
      "learning_rate": 5.270446631332238e-07,
      "loss": 3.8021,
      "step": 22276
    },
    {
      "epoch": 0.8993540573274121,
      "grad_norm": 115.64624786376953,
      "learning_rate": 5.266258463515261e-07,
      "loss": 4.6937,
      "step": 22277
    },
    {
      "epoch": 0.899394428744449,
      "grad_norm": 176.333740234375,
      "learning_rate": 5.262071915424027e-07,
      "loss": 4.7207,
      "step": 22278
    },
    {
      "epoch": 0.8994348001614857,
      "grad_norm": 79.46144104003906,
      "learning_rate": 5.257886987130168e-07,
      "loss": 5.1417,
      "step": 22279
    },
    {
      "epoch": 0.8994751715785224,
      "grad_norm": 75.45707702636719,
      "learning_rate": 5.25370367870518e-07,
      "loss": 3.807,
      "step": 22280
    },
    {
      "epoch": 0.8995155429955591,
      "grad_norm": 84.22412109375,
      "learning_rate": 5.249521990220607e-07,
      "loss": 4.0916,
      "step": 22281
    },
    {
      "epoch": 0.8995559144125959,
      "grad_norm": 92.76805114746094,
      "learning_rate": 5.245341921747948e-07,
      "loss": 5.2057,
      "step": 22282
    },
    {
      "epoch": 0.8995962858296326,
      "grad_norm": 103.89718627929688,
      "learning_rate": 5.24116347335869e-07,
      "loss": 5.9485,
      "step": 22283
    },
    {
      "epoch": 0.8996366572466694,
      "grad_norm": 97.65420532226562,
      "learning_rate": 5.236986645124242e-07,
      "loss": 3.4711,
      "step": 22284
    },
    {
      "epoch": 0.8996770286637061,
      "grad_norm": 94.27603912353516,
      "learning_rate": 5.232811437116048e-07,
      "loss": 4.3559,
      "step": 22285
    },
    {
      "epoch": 0.8997174000807429,
      "grad_norm": 103.9849853515625,
      "learning_rate": 5.228637849405471e-07,
      "loss": 2.3753,
      "step": 22286
    },
    {
      "epoch": 0.8997577714977796,
      "grad_norm": 118.54547882080078,
      "learning_rate": 5.22446588206389e-07,
      "loss": 5.821,
      "step": 22287
    },
    {
      "epoch": 0.8997981429148163,
      "grad_norm": 77.27655792236328,
      "learning_rate": 5.220295535162622e-07,
      "loss": 5.5268,
      "step": 22288
    },
    {
      "epoch": 0.899838514331853,
      "grad_norm": 101.75880432128906,
      "learning_rate": 5.21612680877297e-07,
      "loss": 5.7445,
      "step": 22289
    },
    {
      "epoch": 0.8998788857488897,
      "grad_norm": 59.73939895629883,
      "learning_rate": 5.211959702966218e-07,
      "loss": 3.9299,
      "step": 22290
    },
    {
      "epoch": 0.8999192571659266,
      "grad_norm": 122.0277328491211,
      "learning_rate": 5.20779421781361e-07,
      "loss": 5.8205,
      "step": 22291
    },
    {
      "epoch": 0.8999596285829633,
      "grad_norm": 93.65282440185547,
      "learning_rate": 5.203630353386369e-07,
      "loss": 4.348,
      "step": 22292
    },
    {
      "epoch": 0.9,
      "grad_norm": 135.9044952392578,
      "learning_rate": 5.199468109755668e-07,
      "loss": 4.1687,
      "step": 22293
    },
    {
      "epoch": 0.9000403714170367,
      "grad_norm": 100.4062728881836,
      "learning_rate": 5.195307486992695e-07,
      "loss": 5.2809,
      "step": 22294
    },
    {
      "epoch": 0.9000807428340735,
      "grad_norm": 820.2073974609375,
      "learning_rate": 5.191148485168595e-07,
      "loss": 4.8821,
      "step": 22295
    },
    {
      "epoch": 0.9001211142511102,
      "grad_norm": 93.66302490234375,
      "learning_rate": 5.186991104354445e-07,
      "loss": 4.8382,
      "step": 22296
    },
    {
      "epoch": 0.9001614856681469,
      "grad_norm": 71.80845642089844,
      "learning_rate": 5.182835344621339e-07,
      "loss": 5.8927,
      "step": 22297
    },
    {
      "epoch": 0.9002018570851837,
      "grad_norm": 79.69146728515625,
      "learning_rate": 5.178681206040337e-07,
      "loss": 1.9932,
      "step": 22298
    },
    {
      "epoch": 0.9002422285022205,
      "grad_norm": 83.81266021728516,
      "learning_rate": 5.174528688682479e-07,
      "loss": 5.4634,
      "step": 22299
    },
    {
      "epoch": 0.9002825999192572,
      "grad_norm": 258.15899658203125,
      "learning_rate": 5.170377792618731e-07,
      "loss": 5.3075,
      "step": 22300
    },
    {
      "epoch": 0.9003229713362939,
      "grad_norm": 120.40428161621094,
      "learning_rate": 5.16622851792008e-07,
      "loss": 4.3787,
      "step": 22301
    },
    {
      "epoch": 0.9003633427533306,
      "grad_norm": 88.10865783691406,
      "learning_rate": 5.16208086465747e-07,
      "loss": 6.7954,
      "step": 22302
    },
    {
      "epoch": 0.9004037141703674,
      "grad_norm": 114.33622741699219,
      "learning_rate": 5.157934832901812e-07,
      "loss": 4.9973,
      "step": 22303
    },
    {
      "epoch": 0.9004440855874041,
      "grad_norm": 110.20873260498047,
      "learning_rate": 5.153790422723992e-07,
      "loss": 4.5852,
      "step": 22304
    },
    {
      "epoch": 0.9004844570044409,
      "grad_norm": 79.60488891601562,
      "learning_rate": 5.149647634194887e-07,
      "loss": 3.6532,
      "step": 22305
    },
    {
      "epoch": 0.9005248284214776,
      "grad_norm": 123.05815887451172,
      "learning_rate": 5.145506467385308e-07,
      "loss": 4.9077,
      "step": 22306
    },
    {
      "epoch": 0.9005651998385144,
      "grad_norm": 117.9265365600586,
      "learning_rate": 5.141366922366075e-07,
      "loss": 5.9875,
      "step": 22307
    },
    {
      "epoch": 0.9006055712555511,
      "grad_norm": 81.63688659667969,
      "learning_rate": 5.137228999207954e-07,
      "loss": 4.477,
      "step": 22308
    },
    {
      "epoch": 0.9006459426725878,
      "grad_norm": 127.10723876953125,
      "learning_rate": 5.13309269798169e-07,
      "loss": 3.9859,
      "step": 22309
    },
    {
      "epoch": 0.9006863140896245,
      "grad_norm": 138.22569274902344,
      "learning_rate": 5.128958018758013e-07,
      "loss": 4.4011,
      "step": 22310
    },
    {
      "epoch": 0.9007266855066612,
      "grad_norm": 87.04959106445312,
      "learning_rate": 5.124824961607622e-07,
      "loss": 4.9199,
      "step": 22311
    },
    {
      "epoch": 0.9007670569236981,
      "grad_norm": 82.88275146484375,
      "learning_rate": 5.120693526601173e-07,
      "loss": 4.9736,
      "step": 22312
    },
    {
      "epoch": 0.9008074283407348,
      "grad_norm": 88.00534057617188,
      "learning_rate": 5.116563713809297e-07,
      "loss": 3.1917,
      "step": 22313
    },
    {
      "epoch": 0.9008477997577715,
      "grad_norm": 112.59088897705078,
      "learning_rate": 5.112435523302617e-07,
      "loss": 3.5186,
      "step": 22314
    },
    {
      "epoch": 0.9008881711748082,
      "grad_norm": 101.65625762939453,
      "learning_rate": 5.10830895515172e-07,
      "loss": 5.0018,
      "step": 22315
    },
    {
      "epoch": 0.900928542591845,
      "grad_norm": 104.47882080078125,
      "learning_rate": 5.104184009427137e-07,
      "loss": 4.991,
      "step": 22316
    },
    {
      "epoch": 0.9009689140088817,
      "grad_norm": 165.50140380859375,
      "learning_rate": 5.100060686199415e-07,
      "loss": 6.4794,
      "step": 22317
    },
    {
      "epoch": 0.9010092854259184,
      "grad_norm": 85.97270202636719,
      "learning_rate": 5.095938985539051e-07,
      "loss": 3.9204,
      "step": 22318
    },
    {
      "epoch": 0.9010496568429552,
      "grad_norm": 84.55967712402344,
      "learning_rate": 5.091818907516522e-07,
      "loss": 3.5756,
      "step": 22319
    },
    {
      "epoch": 0.901090028259992,
      "grad_norm": 100.27854919433594,
      "learning_rate": 5.087700452202248e-07,
      "loss": 4.6414,
      "step": 22320
    },
    {
      "epoch": 0.9011303996770287,
      "grad_norm": 121.29882049560547,
      "learning_rate": 5.083583619666665e-07,
      "loss": 5.084,
      "step": 22321
    },
    {
      "epoch": 0.9011707710940654,
      "grad_norm": 101.6343994140625,
      "learning_rate": 5.079468409980148e-07,
      "loss": 3.8309,
      "step": 22322
    },
    {
      "epoch": 0.9012111425111021,
      "grad_norm": 77.79082489013672,
      "learning_rate": 5.075354823213074e-07,
      "loss": 5.3474,
      "step": 22323
    },
    {
      "epoch": 0.9012515139281388,
      "grad_norm": 115.69766235351562,
      "learning_rate": 5.071242859435766e-07,
      "loss": 3.3967,
      "step": 22324
    },
    {
      "epoch": 0.9012918853451756,
      "grad_norm": 100.46199035644531,
      "learning_rate": 5.067132518718532e-07,
      "loss": 4.0848,
      "step": 22325
    },
    {
      "epoch": 0.9013322567622124,
      "grad_norm": 171.09640502929688,
      "learning_rate": 5.063023801131639e-07,
      "loss": 4.4199,
      "step": 22326
    },
    {
      "epoch": 0.9013726281792491,
      "grad_norm": 81.20843505859375,
      "learning_rate": 5.058916706745364e-07,
      "loss": 3.8844,
      "step": 22327
    },
    {
      "epoch": 0.9014129995962858,
      "grad_norm": 80.81874084472656,
      "learning_rate": 5.054811235629898e-07,
      "loss": 3.3341,
      "step": 22328
    },
    {
      "epoch": 0.9014533710133226,
      "grad_norm": 76.28361511230469,
      "learning_rate": 5.050707387855447e-07,
      "loss": 5.4014,
      "step": 22329
    },
    {
      "epoch": 0.9014937424303593,
      "grad_norm": 70.93566131591797,
      "learning_rate": 5.04660516349218e-07,
      "loss": 2.6436,
      "step": 22330
    },
    {
      "epoch": 0.901534113847396,
      "grad_norm": 102.9181137084961,
      "learning_rate": 5.042504562610239e-07,
      "loss": 5.884,
      "step": 22331
    },
    {
      "epoch": 0.9015744852644327,
      "grad_norm": 126.90064239501953,
      "learning_rate": 5.038405585279715e-07,
      "loss": 6.0149,
      "step": 22332
    },
    {
      "epoch": 0.9016148566814696,
      "grad_norm": 82.71871948242188,
      "learning_rate": 5.034308231570717e-07,
      "loss": 2.8915,
      "step": 22333
    },
    {
      "epoch": 0.9016552280985063,
      "grad_norm": 112.81470489501953,
      "learning_rate": 5.030212501553277e-07,
      "loss": 5.1438,
      "step": 22334
    },
    {
      "epoch": 0.901695599515543,
      "grad_norm": 94.93179321289062,
      "learning_rate": 5.026118395297442e-07,
      "loss": 4.7055,
      "step": 22335
    },
    {
      "epoch": 0.9017359709325797,
      "grad_norm": 102.68367767333984,
      "learning_rate": 5.022025912873196e-07,
      "loss": 5.6829,
      "step": 22336
    },
    {
      "epoch": 0.9017763423496165,
      "grad_norm": 82.50768280029297,
      "learning_rate": 5.017935054350521e-07,
      "loss": 4.3429,
      "step": 22337
    },
    {
      "epoch": 0.9018167137666532,
      "grad_norm": 115.25578308105469,
      "learning_rate": 5.013845819799346e-07,
      "loss": 4.9865,
      "step": 22338
    },
    {
      "epoch": 0.9018570851836899,
      "grad_norm": 89.93363189697266,
      "learning_rate": 5.009758209289617e-07,
      "loss": 3.967,
      "step": 22339
    },
    {
      "epoch": 0.9018974566007267,
      "grad_norm": 127.10270690917969,
      "learning_rate": 5.005672222891189e-07,
      "loss": 4.9833,
      "step": 22340
    },
    {
      "epoch": 0.9019378280177635,
      "grad_norm": 72.01376342773438,
      "learning_rate": 5.001587860673951e-07,
      "loss": 3.8442,
      "step": 22341
    },
    {
      "epoch": 0.9019781994348002,
      "grad_norm": 124.68915557861328,
      "learning_rate": 4.997505122707714e-07,
      "loss": 3.4303,
      "step": 22342
    },
    {
      "epoch": 0.9020185708518369,
      "grad_norm": 90.79701232910156,
      "learning_rate": 4.993424009062287e-07,
      "loss": 3.4771,
      "step": 22343
    },
    {
      "epoch": 0.9020589422688736,
      "grad_norm": 82.85934448242188,
      "learning_rate": 4.989344519807459e-07,
      "loss": 5.3737,
      "step": 22344
    },
    {
      "epoch": 0.9020993136859103,
      "grad_norm": 84.76104736328125,
      "learning_rate": 4.985266655012977e-07,
      "loss": 4.6588,
      "step": 22345
    },
    {
      "epoch": 0.9021396851029471,
      "grad_norm": 173.2526092529297,
      "learning_rate": 4.981190414748549e-07,
      "loss": 4.3028,
      "step": 22346
    },
    {
      "epoch": 0.9021800565199839,
      "grad_norm": 75.68580627441406,
      "learning_rate": 4.977115799083909e-07,
      "loss": 3.7749,
      "step": 22347
    },
    {
      "epoch": 0.9022204279370206,
      "grad_norm": 103.39042663574219,
      "learning_rate": 4.973042808088669e-07,
      "loss": 5.4609,
      "step": 22348
    },
    {
      "epoch": 0.9022607993540573,
      "grad_norm": 127.18437957763672,
      "learning_rate": 4.968971441832493e-07,
      "loss": 6.4203,
      "step": 22349
    },
    {
      "epoch": 0.9023011707710941,
      "grad_norm": 77.28620910644531,
      "learning_rate": 4.964901700385006e-07,
      "loss": 6.4239,
      "step": 22350
    },
    {
      "epoch": 0.9023415421881308,
      "grad_norm": 140.64718627929688,
      "learning_rate": 4.960833583815772e-07,
      "loss": 4.7825,
      "step": 22351
    },
    {
      "epoch": 0.9023819136051675,
      "grad_norm": 109.3256607055664,
      "learning_rate": 4.956767092194348e-07,
      "loss": 5.2708,
      "step": 22352
    },
    {
      "epoch": 0.9024222850222042,
      "grad_norm": 73.96847534179688,
      "learning_rate": 4.952702225590266e-07,
      "loss": 2.4269,
      "step": 22353
    },
    {
      "epoch": 0.9024626564392411,
      "grad_norm": 131.62611389160156,
      "learning_rate": 4.948638984073018e-07,
      "loss": 5.6489,
      "step": 22354
    },
    {
      "epoch": 0.9025030278562778,
      "grad_norm": 149.56227111816406,
      "learning_rate": 4.9445773677121e-07,
      "loss": 3.3354,
      "step": 22355
    },
    {
      "epoch": 0.9025433992733145,
      "grad_norm": 79.9239730834961,
      "learning_rate": 4.940517376576925e-07,
      "loss": 5.678,
      "step": 22356
    },
    {
      "epoch": 0.9025837706903512,
      "grad_norm": 115.7404556274414,
      "learning_rate": 4.936459010736927e-07,
      "loss": 5.8294,
      "step": 22357
    },
    {
      "epoch": 0.902624142107388,
      "grad_norm": 66.8744125366211,
      "learning_rate": 4.932402270261482e-07,
      "loss": 5.2874,
      "step": 22358
    },
    {
      "epoch": 0.9026645135244247,
      "grad_norm": 88.3043441772461,
      "learning_rate": 4.928347155219981e-07,
      "loss": 4.2619,
      "step": 22359
    },
    {
      "epoch": 0.9027048849414614,
      "grad_norm": 79.4736099243164,
      "learning_rate": 4.924293665681712e-07,
      "loss": 3.4657,
      "step": 22360
    },
    {
      "epoch": 0.9027452563584982,
      "grad_norm": 129.49378967285156,
      "learning_rate": 4.920241801716019e-07,
      "loss": 5.3979,
      "step": 22361
    },
    {
      "epoch": 0.902785627775535,
      "grad_norm": 90.26878356933594,
      "learning_rate": 4.916191563392148e-07,
      "loss": 4.1733,
      "step": 22362
    },
    {
      "epoch": 0.9028259991925717,
      "grad_norm": 70.26277923583984,
      "learning_rate": 4.912142950779375e-07,
      "loss": 6.4979,
      "step": 22363
    },
    {
      "epoch": 0.9028663706096084,
      "grad_norm": 408.99432373046875,
      "learning_rate": 4.908095963946901e-07,
      "loss": 4.5429,
      "step": 22364
    },
    {
      "epoch": 0.9029067420266451,
      "grad_norm": 87.83260345458984,
      "learning_rate": 4.904050602963939e-07,
      "loss": 5.3395,
      "step": 22365
    },
    {
      "epoch": 0.9029471134436818,
      "grad_norm": 126.1635513305664,
      "learning_rate": 4.900006867899642e-07,
      "loss": 4.4105,
      "step": 22366
    },
    {
      "epoch": 0.9029874848607186,
      "grad_norm": 124.2735824584961,
      "learning_rate": 4.89596475882318e-07,
      "loss": 4.7043,
      "step": 22367
    },
    {
      "epoch": 0.9030278562777554,
      "grad_norm": 99.40200805664062,
      "learning_rate": 4.891924275803606e-07,
      "loss": 5.6095,
      "step": 22368
    },
    {
      "epoch": 0.9030682276947921,
      "grad_norm": 112.48171997070312,
      "learning_rate": 4.887885418910044e-07,
      "loss": 4.8019,
      "step": 22369
    },
    {
      "epoch": 0.9031085991118288,
      "grad_norm": 97.5560531616211,
      "learning_rate": 4.883848188211537e-07,
      "loss": 5.0793,
      "step": 22370
    },
    {
      "epoch": 0.9031489705288656,
      "grad_norm": 129.0457305908203,
      "learning_rate": 4.879812583777111e-07,
      "loss": 4.3309,
      "step": 22371
    },
    {
      "epoch": 0.9031893419459023,
      "grad_norm": 141.38607788085938,
      "learning_rate": 4.875778605675785e-07,
      "loss": 4.6045,
      "step": 22372
    },
    {
      "epoch": 0.903229713362939,
      "grad_norm": 156.69227600097656,
      "learning_rate": 4.871746253976495e-07,
      "loss": 4.8124,
      "step": 22373
    },
    {
      "epoch": 0.9032700847799757,
      "grad_norm": 88.93952941894531,
      "learning_rate": 4.867715528748207e-07,
      "loss": 4.0379,
      "step": 22374
    },
    {
      "epoch": 0.9033104561970126,
      "grad_norm": 110.35044860839844,
      "learning_rate": 4.863686430059833e-07,
      "loss": 5.3421,
      "step": 22375
    },
    {
      "epoch": 0.9033508276140493,
      "grad_norm": 2970.505859375,
      "learning_rate": 4.859658957980274e-07,
      "loss": 4.9522,
      "step": 22376
    },
    {
      "epoch": 0.903391199031086,
      "grad_norm": 103.64529418945312,
      "learning_rate": 4.855633112578361e-07,
      "loss": 5.2753,
      "step": 22377
    },
    {
      "epoch": 0.9034315704481227,
      "grad_norm": 77.52518463134766,
      "learning_rate": 4.851608893922954e-07,
      "loss": 4.564,
      "step": 22378
    },
    {
      "epoch": 0.9034719418651594,
      "grad_norm": 94.588623046875,
      "learning_rate": 4.847586302082841e-07,
      "loss": 4.3,
      "step": 22379
    },
    {
      "epoch": 0.9035123132821962,
      "grad_norm": 78.0440673828125,
      "learning_rate": 4.843565337126821e-07,
      "loss": 3.826,
      "step": 22380
    },
    {
      "epoch": 0.9035526846992329,
      "grad_norm": 107.96239471435547,
      "learning_rate": 4.839545999123607e-07,
      "loss": 5.1668,
      "step": 22381
    },
    {
      "epoch": 0.9035930561162697,
      "grad_norm": 76.9148178100586,
      "learning_rate": 4.835528288141944e-07,
      "loss": 4.6608,
      "step": 22382
    },
    {
      "epoch": 0.9036334275333064,
      "grad_norm": 141.50025939941406,
      "learning_rate": 4.831512204250533e-07,
      "loss": 5.6937,
      "step": 22383
    },
    {
      "epoch": 0.9036737989503432,
      "grad_norm": 147.2183837890625,
      "learning_rate": 4.827497747518018e-07,
      "loss": 5.979,
      "step": 22384
    },
    {
      "epoch": 0.9037141703673799,
      "grad_norm": 91.39412689208984,
      "learning_rate": 4.823484918013055e-07,
      "loss": 6.2477,
      "step": 22385
    },
    {
      "epoch": 0.9037545417844166,
      "grad_norm": 106.41748809814453,
      "learning_rate": 4.819473715804235e-07,
      "loss": 4.3343,
      "step": 22386
    },
    {
      "epoch": 0.9037949132014533,
      "grad_norm": 67.89105224609375,
      "learning_rate": 4.815464140960157e-07,
      "loss": 3.8368,
      "step": 22387
    },
    {
      "epoch": 0.90383528461849,
      "grad_norm": 91.287841796875,
      "learning_rate": 4.811456193549391e-07,
      "loss": 4.7516,
      "step": 22388
    },
    {
      "epoch": 0.9038756560355269,
      "grad_norm": 157.5707550048828,
      "learning_rate": 4.807449873640424e-07,
      "loss": 5.166,
      "step": 22389
    },
    {
      "epoch": 0.9039160274525636,
      "grad_norm": 91.85687255859375,
      "learning_rate": 4.803445181301781e-07,
      "loss": 4.5544,
      "step": 22390
    },
    {
      "epoch": 0.9039563988696003,
      "grad_norm": 122.85448455810547,
      "learning_rate": 4.799442116601916e-07,
      "loss": 5.4778,
      "step": 22391
    },
    {
      "epoch": 0.903996770286637,
      "grad_norm": 80.6786880493164,
      "learning_rate": 4.795440679609298e-07,
      "loss": 4.385,
      "step": 22392
    },
    {
      "epoch": 0.9040371417036738,
      "grad_norm": 86.90010070800781,
      "learning_rate": 4.791440870392306e-07,
      "loss": 4.4454,
      "step": 22393
    },
    {
      "epoch": 0.9040775131207105,
      "grad_norm": 77.3084487915039,
      "learning_rate": 4.78744268901935e-07,
      "loss": 4.8444,
      "step": 22394
    },
    {
      "epoch": 0.9041178845377473,
      "grad_norm": 102.67790985107422,
      "learning_rate": 4.783446135558789e-07,
      "loss": 4.1965,
      "step": 22395
    },
    {
      "epoch": 0.904158255954784,
      "grad_norm": 88.91201782226562,
      "learning_rate": 4.779451210078967e-07,
      "loss": 4.1036,
      "step": 22396
    },
    {
      "epoch": 0.9041986273718208,
      "grad_norm": 71.70606231689453,
      "learning_rate": 4.775457912648162e-07,
      "loss": 5.0707,
      "step": 22397
    },
    {
      "epoch": 0.9042389987888575,
      "grad_norm": 141.7477264404297,
      "learning_rate": 4.771466243334666e-07,
      "loss": 4.5759,
      "step": 22398
    },
    {
      "epoch": 0.9042793702058942,
      "grad_norm": 99.70954895019531,
      "learning_rate": 4.767476202206711e-07,
      "loss": 6.0906,
      "step": 22399
    },
    {
      "epoch": 0.9043197416229309,
      "grad_norm": 115.43096923828125,
      "learning_rate": 4.7634877893325324e-07,
      "loss": 5.4078,
      "step": 22400
    },
    {
      "epoch": 0.9043601130399677,
      "grad_norm": 138.55233764648438,
      "learning_rate": 4.75950100478032e-07,
      "loss": 5.955,
      "step": 22401
    },
    {
      "epoch": 0.9044004844570045,
      "grad_norm": 100.17737579345703,
      "learning_rate": 4.755515848618242e-07,
      "loss": 4.9621,
      "step": 22402
    },
    {
      "epoch": 0.9044408558740412,
      "grad_norm": 94.77423095703125,
      "learning_rate": 4.7515323209144206e-07,
      "loss": 3.8628,
      "step": 22403
    },
    {
      "epoch": 0.9044812272910779,
      "grad_norm": 94.09651947021484,
      "learning_rate": 4.7475504217370016e-07,
      "loss": 5.2104,
      "step": 22404
    },
    {
      "epoch": 0.9045215987081147,
      "grad_norm": 109.49967193603516,
      "learning_rate": 4.74357015115402e-07,
      "loss": 6.2197,
      "step": 22405
    },
    {
      "epoch": 0.9045619701251514,
      "grad_norm": 105.41790771484375,
      "learning_rate": 4.739591509233543e-07,
      "loss": 3.4997,
      "step": 22406
    },
    {
      "epoch": 0.9046023415421881,
      "grad_norm": 118.0584487915039,
      "learning_rate": 4.735614496043617e-07,
      "loss": 4.8808,
      "step": 22407
    },
    {
      "epoch": 0.9046427129592248,
      "grad_norm": 109.89080810546875,
      "learning_rate": 4.731639111652231e-07,
      "loss": 4.4427,
      "step": 22408
    },
    {
      "epoch": 0.9046830843762617,
      "grad_norm": 109.50072479248047,
      "learning_rate": 4.727665356127331e-07,
      "loss": 4.4581,
      "step": 22409
    },
    {
      "epoch": 0.9047234557932984,
      "grad_norm": 83.2691650390625,
      "learning_rate": 4.7236932295368743e-07,
      "loss": 2.9862,
      "step": 22410
    },
    {
      "epoch": 0.9047638272103351,
      "grad_norm": 115.7483901977539,
      "learning_rate": 4.7197227319487835e-07,
      "loss": 5.1167,
      "step": 22411
    },
    {
      "epoch": 0.9048041986273718,
      "grad_norm": 85.29025268554688,
      "learning_rate": 4.7157538634309497e-07,
      "loss": 3.2263,
      "step": 22412
    },
    {
      "epoch": 0.9048445700444085,
      "grad_norm": 67.95417785644531,
      "learning_rate": 4.7117866240512064e-07,
      "loss": 5.2984,
      "step": 22413
    },
    {
      "epoch": 0.9048849414614453,
      "grad_norm": 101.1150894165039,
      "learning_rate": 4.7078210138774006e-07,
      "loss": 3.7391,
      "step": 22414
    },
    {
      "epoch": 0.904925312878482,
      "grad_norm": 90.91217803955078,
      "learning_rate": 4.7038570329773324e-07,
      "loss": 4.6126,
      "step": 22415
    },
    {
      "epoch": 0.9049656842955188,
      "grad_norm": 167.17251586914062,
      "learning_rate": 4.699894681418793e-07,
      "loss": 5.2692,
      "step": 22416
    },
    {
      "epoch": 0.9050060557125555,
      "grad_norm": 103.9270248413086,
      "learning_rate": 4.6959339592694944e-07,
      "loss": 5.7984,
      "step": 22417
    },
    {
      "epoch": 0.9050464271295923,
      "grad_norm": 89.94378662109375,
      "learning_rate": 4.691974866597171e-07,
      "loss": 3.8494,
      "step": 22418
    },
    {
      "epoch": 0.905086798546629,
      "grad_norm": 99.00591278076172,
      "learning_rate": 4.6880174034695245e-07,
      "loss": 5.0295,
      "step": 22419
    },
    {
      "epoch": 0.9051271699636657,
      "grad_norm": 75.59033966064453,
      "learning_rate": 4.684061569954201e-07,
      "loss": 5.5718,
      "step": 22420
    },
    {
      "epoch": 0.9051675413807024,
      "grad_norm": 76.63172912597656,
      "learning_rate": 4.680107366118858e-07,
      "loss": 4.6754,
      "step": 22421
    },
    {
      "epoch": 0.9052079127977392,
      "grad_norm": 85.64898681640625,
      "learning_rate": 4.6761547920310845e-07,
      "loss": 5.7387,
      "step": 22422
    },
    {
      "epoch": 0.905248284214776,
      "grad_norm": 106.2939682006836,
      "learning_rate": 4.6722038477584606e-07,
      "loss": 5.8869,
      "step": 22423
    },
    {
      "epoch": 0.9052886556318127,
      "grad_norm": 107.84813690185547,
      "learning_rate": 4.6682545333685656e-07,
      "loss": 4.7728,
      "step": 22424
    },
    {
      "epoch": 0.9053290270488494,
      "grad_norm": 110.32526397705078,
      "learning_rate": 4.66430684892889e-07,
      "loss": 6.2424,
      "step": 22425
    },
    {
      "epoch": 0.9053693984658862,
      "grad_norm": 85.99134826660156,
      "learning_rate": 4.660360794506946e-07,
      "loss": 5.0635,
      "step": 22426
    },
    {
      "epoch": 0.9054097698829229,
      "grad_norm": 131.1644744873047,
      "learning_rate": 4.6564163701701916e-07,
      "loss": 4.5625,
      "step": 22427
    },
    {
      "epoch": 0.9054501412999596,
      "grad_norm": 83.54269409179688,
      "learning_rate": 4.652473575986083e-07,
      "loss": 5.1478,
      "step": 22428
    },
    {
      "epoch": 0.9054905127169963,
      "grad_norm": 156.50767517089844,
      "learning_rate": 4.648532412022022e-07,
      "loss": 3.9279,
      "step": 22429
    },
    {
      "epoch": 0.9055308841340332,
      "grad_norm": 137.0169219970703,
      "learning_rate": 4.6445928783453887e-07,
      "loss": 5.161,
      "step": 22430
    },
    {
      "epoch": 0.9055712555510699,
      "grad_norm": 103.68931579589844,
      "learning_rate": 4.640654975023551e-07,
      "loss": 2.8281,
      "step": 22431
    },
    {
      "epoch": 0.9056116269681066,
      "grad_norm": 90.47085571289062,
      "learning_rate": 4.6367187021238327e-07,
      "loss": 5.8243,
      "step": 22432
    },
    {
      "epoch": 0.9056519983851433,
      "grad_norm": 153.1283721923828,
      "learning_rate": 4.6327840597135355e-07,
      "loss": 3.8427,
      "step": 22433
    },
    {
      "epoch": 0.90569236980218,
      "grad_norm": 99.94062805175781,
      "learning_rate": 4.628851047859928e-07,
      "loss": 4.2745,
      "step": 22434
    },
    {
      "epoch": 0.9057327412192168,
      "grad_norm": 88.54903411865234,
      "learning_rate": 4.6249196666302567e-07,
      "loss": 4.0573,
      "step": 22435
    },
    {
      "epoch": 0.9057731126362535,
      "grad_norm": 93.18513488769531,
      "learning_rate": 4.620989916091767e-07,
      "loss": 5.0926,
      "step": 22436
    },
    {
      "epoch": 0.9058134840532903,
      "grad_norm": 104.38349151611328,
      "learning_rate": 4.617061796311595e-07,
      "loss": 4.7303,
      "step": 22437
    },
    {
      "epoch": 0.905853855470327,
      "grad_norm": 120.97600555419922,
      "learning_rate": 4.6131353073569416e-07,
      "loss": 5.2874,
      "step": 22438
    },
    {
      "epoch": 0.9058942268873638,
      "grad_norm": 93.55475616455078,
      "learning_rate": 4.609210449294932e-07,
      "loss": 4.6183,
      "step": 22439
    },
    {
      "epoch": 0.9059345983044005,
      "grad_norm": 118.47880554199219,
      "learning_rate": 4.605287222192667e-07,
      "loss": 4.7593,
      "step": 22440
    },
    {
      "epoch": 0.9059749697214372,
      "grad_norm": 109.39500427246094,
      "learning_rate": 4.601365626117227e-07,
      "loss": 2.3656,
      "step": 22441
    },
    {
      "epoch": 0.9060153411384739,
      "grad_norm": 136.4063720703125,
      "learning_rate": 4.597445661135658e-07,
      "loss": 4.874,
      "step": 22442
    },
    {
      "epoch": 0.9060557125555107,
      "grad_norm": 107.11824798583984,
      "learning_rate": 4.593527327314995e-07,
      "loss": 4.7786,
      "step": 22443
    },
    {
      "epoch": 0.9060960839725475,
      "grad_norm": 117.13494110107422,
      "learning_rate": 4.5896106247222295e-07,
      "loss": 5.211,
      "step": 22444
    },
    {
      "epoch": 0.9061364553895842,
      "grad_norm": 80.76514434814453,
      "learning_rate": 4.5856955534243185e-07,
      "loss": 4.838,
      "step": 22445
    },
    {
      "epoch": 0.9061768268066209,
      "grad_norm": 93.05986785888672,
      "learning_rate": 4.581782113488198e-07,
      "loss": 4.3179,
      "step": 22446
    },
    {
      "epoch": 0.9062171982236576,
      "grad_norm": 106.09046173095703,
      "learning_rate": 4.5778703049807914e-07,
      "loss": 5.0972,
      "step": 22447
    },
    {
      "epoch": 0.9062575696406944,
      "grad_norm": 238.13226318359375,
      "learning_rate": 4.5739601279689907e-07,
      "loss": 5.1605,
      "step": 22448
    },
    {
      "epoch": 0.9062979410577311,
      "grad_norm": 147.77645874023438,
      "learning_rate": 4.570051582519619e-07,
      "loss": 4.627,
      "step": 22449
    },
    {
      "epoch": 0.9063383124747678,
      "grad_norm": 87.32525634765625,
      "learning_rate": 4.5661446686995123e-07,
      "loss": 4.2398,
      "step": 22450
    },
    {
      "epoch": 0.9063786838918046,
      "grad_norm": 126.15809631347656,
      "learning_rate": 4.5622393865754843e-07,
      "loss": 4.2965,
      "step": 22451
    },
    {
      "epoch": 0.9064190553088414,
      "grad_norm": 94.64641571044922,
      "learning_rate": 4.5583357362143146e-07,
      "loss": 6.5299,
      "step": 22452
    },
    {
      "epoch": 0.9064594267258781,
      "grad_norm": 100.0042495727539,
      "learning_rate": 4.554433717682727e-07,
      "loss": 4.7988,
      "step": 22453
    },
    {
      "epoch": 0.9064997981429148,
      "grad_norm": 75.78203582763672,
      "learning_rate": 4.550533331047424e-07,
      "loss": 4.7286,
      "step": 22454
    },
    {
      "epoch": 0.9065401695599515,
      "grad_norm": 83.40655517578125,
      "learning_rate": 4.5466345763751195e-07,
      "loss": 4.2268,
      "step": 22455
    },
    {
      "epoch": 0.9065805409769883,
      "grad_norm": 78.98027801513672,
      "learning_rate": 4.542737453732482e-07,
      "loss": 2.5271,
      "step": 22456
    },
    {
      "epoch": 0.906620912394025,
      "grad_norm": 79.84823608398438,
      "learning_rate": 4.5388419631861024e-07,
      "loss": 5.2895,
      "step": 22457
    },
    {
      "epoch": 0.9066612838110618,
      "grad_norm": 63.08488845825195,
      "learning_rate": 4.534948104802617e-07,
      "loss": 3.7906,
      "step": 22458
    },
    {
      "epoch": 0.9067016552280985,
      "grad_norm": 83.7266616821289,
      "learning_rate": 4.5310558786485827e-07,
      "loss": 4.8221,
      "step": 22459
    },
    {
      "epoch": 0.9067420266451353,
      "grad_norm": 93.5596923828125,
      "learning_rate": 4.527165284790558e-07,
      "loss": 3.8701,
      "step": 22460
    },
    {
      "epoch": 0.906782398062172,
      "grad_norm": 115.7232894897461,
      "learning_rate": 4.5232763232950674e-07,
      "loss": 4.7316,
      "step": 22461
    },
    {
      "epoch": 0.9068227694792087,
      "grad_norm": 126.92146301269531,
      "learning_rate": 4.5193889942285906e-07,
      "loss": 5.7975,
      "step": 22462
    },
    {
      "epoch": 0.9068631408962454,
      "grad_norm": 101.18562316894531,
      "learning_rate": 4.5155032976576084e-07,
      "loss": 4.1716,
      "step": 22463
    },
    {
      "epoch": 0.9069035123132821,
      "grad_norm": 187.59295654296875,
      "learning_rate": 4.5116192336485564e-07,
      "loss": 6.4524,
      "step": 22464
    },
    {
      "epoch": 0.906943883730319,
      "grad_norm": 90.98876190185547,
      "learning_rate": 4.507736802267815e-07,
      "loss": 5.3669,
      "step": 22465
    },
    {
      "epoch": 0.9069842551473557,
      "grad_norm": 96.1548080444336,
      "learning_rate": 4.503856003581786e-07,
      "loss": 3.6853,
      "step": 22466
    },
    {
      "epoch": 0.9070246265643924,
      "grad_norm": 70.443359375,
      "learning_rate": 4.499976837656828e-07,
      "loss": 5.025,
      "step": 22467
    },
    {
      "epoch": 0.9070649979814291,
      "grad_norm": 120.790283203125,
      "learning_rate": 4.4960993045592647e-07,
      "loss": 6.8431,
      "step": 22468
    },
    {
      "epoch": 0.9071053693984659,
      "grad_norm": 77.20984649658203,
      "learning_rate": 4.492223404355367e-07,
      "loss": 6.363,
      "step": 22469
    },
    {
      "epoch": 0.9071457408155026,
      "grad_norm": 91.85658264160156,
      "learning_rate": 4.488349137111436e-07,
      "loss": 4.4585,
      "step": 22470
    },
    {
      "epoch": 0.9071861122325393,
      "grad_norm": 91.37772369384766,
      "learning_rate": 4.4844765028936866e-07,
      "loss": 4.6652,
      "step": 22471
    },
    {
      "epoch": 0.9072264836495761,
      "grad_norm": 89.17977905273438,
      "learning_rate": 4.480605501768365e-07,
      "loss": 5.6572,
      "step": 22472
    },
    {
      "epoch": 0.9072668550666129,
      "grad_norm": 93.0127182006836,
      "learning_rate": 4.476736133801618e-07,
      "loss": 4.9043,
      "step": 22473
    },
    {
      "epoch": 0.9073072264836496,
      "grad_norm": 91.78466033935547,
      "learning_rate": 4.4728683990596267e-07,
      "loss": 5.6742,
      "step": 22474
    },
    {
      "epoch": 0.9073475979006863,
      "grad_norm": 109.11512756347656,
      "learning_rate": 4.4690022976085046e-07,
      "loss": 5.1546,
      "step": 22475
    },
    {
      "epoch": 0.907387969317723,
      "grad_norm": 104.73614501953125,
      "learning_rate": 4.4651378295143765e-07,
      "loss": 5.4254,
      "step": 22476
    },
    {
      "epoch": 0.9074283407347598,
      "grad_norm": 124.16744995117188,
      "learning_rate": 4.46127499484329e-07,
      "loss": 5.58,
      "step": 22477
    },
    {
      "epoch": 0.9074687121517965,
      "grad_norm": 103.63839721679688,
      "learning_rate": 4.4574137936613025e-07,
      "loss": 5.0578,
      "step": 22478
    },
    {
      "epoch": 0.9075090835688333,
      "grad_norm": 97.16342163085938,
      "learning_rate": 4.45355422603444e-07,
      "loss": 4.3416,
      "step": 22479
    },
    {
      "epoch": 0.90754945498587,
      "grad_norm": 102.39470672607422,
      "learning_rate": 4.449696292028671e-07,
      "loss": 4.5366,
      "step": 22480
    },
    {
      "epoch": 0.9075898264029068,
      "grad_norm": 177.2664031982422,
      "learning_rate": 4.445839991709977e-07,
      "loss": 3.4174,
      "step": 22481
    },
    {
      "epoch": 0.9076301978199435,
      "grad_norm": 83.10539245605469,
      "learning_rate": 4.4419853251442824e-07,
      "loss": 4.4551,
      "step": 22482
    },
    {
      "epoch": 0.9076705692369802,
      "grad_norm": 89.84965515136719,
      "learning_rate": 4.43813229239749e-07,
      "loss": 6.5201,
      "step": 22483
    },
    {
      "epoch": 0.9077109406540169,
      "grad_norm": 107.11605072021484,
      "learning_rate": 4.4342808935355033e-07,
      "loss": 4.3208,
      "step": 22484
    },
    {
      "epoch": 0.9077513120710536,
      "grad_norm": 139.6434326171875,
      "learning_rate": 4.430431128624135e-07,
      "loss": 3.9485,
      "step": 22485
    },
    {
      "epoch": 0.9077916834880905,
      "grad_norm": 145.94549560546875,
      "learning_rate": 4.426582997729223e-07,
      "loss": 3.9966,
      "step": 22486
    },
    {
      "epoch": 0.9078320549051272,
      "grad_norm": 172.1487274169922,
      "learning_rate": 4.422736500916558e-07,
      "loss": 3.593,
      "step": 22487
    },
    {
      "epoch": 0.9078724263221639,
      "grad_norm": 118.97185516357422,
      "learning_rate": 4.4188916382519207e-07,
      "loss": 6.6301,
      "step": 22488
    },
    {
      "epoch": 0.9079127977392006,
      "grad_norm": 73.5216293334961,
      "learning_rate": 4.415048409801037e-07,
      "loss": 5.1335,
      "step": 22489
    },
    {
      "epoch": 0.9079531691562374,
      "grad_norm": 86.60685729980469,
      "learning_rate": 4.4112068156296097e-07,
      "loss": 5.6793,
      "step": 22490
    },
    {
      "epoch": 0.9079935405732741,
      "grad_norm": 127.93831634521484,
      "learning_rate": 4.40736685580333e-07,
      "loss": 4.3774,
      "step": 22491
    },
    {
      "epoch": 0.9080339119903108,
      "grad_norm": 101.65040588378906,
      "learning_rate": 4.403528530387868e-07,
      "loss": 6.1888,
      "step": 22492
    },
    {
      "epoch": 0.9080742834073476,
      "grad_norm": 83.19622039794922,
      "learning_rate": 4.3996918394488277e-07,
      "loss": 3.7441,
      "step": 22493
    },
    {
      "epoch": 0.9081146548243844,
      "grad_norm": 89.45145416259766,
      "learning_rate": 4.3958567830517996e-07,
      "loss": 5.8672,
      "step": 22494
    },
    {
      "epoch": 0.9081550262414211,
      "grad_norm": 90.10511016845703,
      "learning_rate": 4.3920233612623763e-07,
      "loss": 5.2084,
      "step": 22495
    },
    {
      "epoch": 0.9081953976584578,
      "grad_norm": 101.84371185302734,
      "learning_rate": 4.388191574146106e-07,
      "loss": 3.3707,
      "step": 22496
    },
    {
      "epoch": 0.9082357690754945,
      "grad_norm": 98.13864135742188,
      "learning_rate": 4.384361421768479e-07,
      "loss": 3.1568,
      "step": 22497
    },
    {
      "epoch": 0.9082761404925312,
      "grad_norm": 116.6824951171875,
      "learning_rate": 4.3805329041949895e-07,
      "loss": 4.2302,
      "step": 22498
    },
    {
      "epoch": 0.9083165119095681,
      "grad_norm": 62.628761291503906,
      "learning_rate": 4.376706021491106e-07,
      "loss": 3.6865,
      "step": 22499
    },
    {
      "epoch": 0.9083568833266048,
      "grad_norm": 135.7997283935547,
      "learning_rate": 4.3728807737222543e-07,
      "loss": 4.9048,
      "step": 22500
    },
    {
      "epoch": 0.9083972547436415,
      "grad_norm": 374.7464294433594,
      "learning_rate": 4.3690571609538267e-07,
      "loss": 4.9398,
      "step": 22501
    },
    {
      "epoch": 0.9084376261606782,
      "grad_norm": 129.143798828125,
      "learning_rate": 4.365235183251215e-07,
      "loss": 2.3293,
      "step": 22502
    },
    {
      "epoch": 0.908477997577715,
      "grad_norm": 83.83985137939453,
      "learning_rate": 4.361414840679745e-07,
      "loss": 3.4499,
      "step": 22503
    },
    {
      "epoch": 0.9085183689947517,
      "grad_norm": 111.06790161132812,
      "learning_rate": 4.357596133304776e-07,
      "loss": 4.6276,
      "step": 22504
    },
    {
      "epoch": 0.9085587404117884,
      "grad_norm": 116.57075500488281,
      "learning_rate": 4.353779061191554e-07,
      "loss": 4.6322,
      "step": 22505
    },
    {
      "epoch": 0.9085991118288252,
      "grad_norm": 113.7650375366211,
      "learning_rate": 4.3499636244053623e-07,
      "loss": 5.1198,
      "step": 22506
    },
    {
      "epoch": 0.908639483245862,
      "grad_norm": 129.7688446044922,
      "learning_rate": 4.3461498230114363e-07,
      "loss": 5.6787,
      "step": 22507
    },
    {
      "epoch": 0.9086798546628987,
      "grad_norm": 80.25376892089844,
      "learning_rate": 4.342337657074991e-07,
      "loss": 3.36,
      "step": 22508
    },
    {
      "epoch": 0.9087202260799354,
      "grad_norm": 152.2401123046875,
      "learning_rate": 4.338527126661174e-07,
      "loss": 4.24,
      "step": 22509
    },
    {
      "epoch": 0.9087605974969721,
      "grad_norm": 92.08744812011719,
      "learning_rate": 4.3347182318351665e-07,
      "loss": 3.5867,
      "step": 22510
    },
    {
      "epoch": 0.9088009689140089,
      "grad_norm": 130.42196655273438,
      "learning_rate": 4.3309109726620835e-07,
      "loss": 3.7532,
      "step": 22511
    },
    {
      "epoch": 0.9088413403310456,
      "grad_norm": 152.4855499267578,
      "learning_rate": 4.327105349207028e-07,
      "loss": 5.834,
      "step": 22512
    },
    {
      "epoch": 0.9088817117480824,
      "grad_norm": 104.69366455078125,
      "learning_rate": 4.3233013615350484e-07,
      "loss": 4.1273,
      "step": 22513
    },
    {
      "epoch": 0.9089220831651191,
      "grad_norm": 105.61294555664062,
      "learning_rate": 4.319499009711192e-07,
      "loss": 4.7682,
      "step": 22514
    },
    {
      "epoch": 0.9089624545821559,
      "grad_norm": 88.21257019042969,
      "learning_rate": 4.3156982938004744e-07,
      "loss": 5.8509,
      "step": 22515
    },
    {
      "epoch": 0.9090028259991926,
      "grad_norm": 124.72928619384766,
      "learning_rate": 4.311899213867887e-07,
      "loss": 4.4569,
      "step": 22516
    },
    {
      "epoch": 0.9090431974162293,
      "grad_norm": 102.91280364990234,
      "learning_rate": 4.3081017699783677e-07,
      "loss": 4.3722,
      "step": 22517
    },
    {
      "epoch": 0.909083568833266,
      "grad_norm": 97.23497772216797,
      "learning_rate": 4.304305962196842e-07,
      "loss": 3.4603,
      "step": 22518
    },
    {
      "epoch": 0.9091239402503027,
      "grad_norm": 106.81236267089844,
      "learning_rate": 4.300511790588224e-07,
      "loss": 6.1192,
      "step": 22519
    },
    {
      "epoch": 0.9091643116673396,
      "grad_norm": 87.59659576416016,
      "learning_rate": 4.296719255217385e-07,
      "loss": 5.196,
      "step": 22520
    },
    {
      "epoch": 0.9092046830843763,
      "grad_norm": 124.11820983886719,
      "learning_rate": 4.292928356149162e-07,
      "loss": 5.3213,
      "step": 22521
    },
    {
      "epoch": 0.909245054501413,
      "grad_norm": 102.29633331298828,
      "learning_rate": 4.289139093448369e-07,
      "loss": 3.6278,
      "step": 22522
    },
    {
      "epoch": 0.9092854259184497,
      "grad_norm": 98.9267807006836,
      "learning_rate": 4.285351467179799e-07,
      "loss": 4.144,
      "step": 22523
    },
    {
      "epoch": 0.9093257973354865,
      "grad_norm": 81.6502914428711,
      "learning_rate": 4.2815654774082117e-07,
      "loss": 3.6786,
      "step": 22524
    },
    {
      "epoch": 0.9093661687525232,
      "grad_norm": 82.22447204589844,
      "learning_rate": 4.2777811241983327e-07,
      "loss": 4.7671,
      "step": 22525
    },
    {
      "epoch": 0.9094065401695599,
      "grad_norm": 103.70043182373047,
      "learning_rate": 4.273998407614877e-07,
      "loss": 4.246,
      "step": 22526
    },
    {
      "epoch": 0.9094469115865967,
      "grad_norm": 99.3535385131836,
      "learning_rate": 4.270217327722503e-07,
      "loss": 5.9213,
      "step": 22527
    },
    {
      "epoch": 0.9094872830036335,
      "grad_norm": 88.92020416259766,
      "learning_rate": 4.2664378845858723e-07,
      "loss": 5.8627,
      "step": 22528
    },
    {
      "epoch": 0.9095276544206702,
      "grad_norm": 86.60113525390625,
      "learning_rate": 4.2626600782695984e-07,
      "loss": 4.19,
      "step": 22529
    },
    {
      "epoch": 0.9095680258377069,
      "grad_norm": 102.5181884765625,
      "learning_rate": 4.2588839088382736e-07,
      "loss": 4.5339,
      "step": 22530
    },
    {
      "epoch": 0.9096083972547436,
      "grad_norm": 129.9383087158203,
      "learning_rate": 4.255109376356459e-07,
      "loss": 5.4064,
      "step": 22531
    },
    {
      "epoch": 0.9096487686717804,
      "grad_norm": 81.18122100830078,
      "learning_rate": 4.2513364808887127e-07,
      "loss": 5.0662,
      "step": 22532
    },
    {
      "epoch": 0.9096891400888171,
      "grad_norm": 96.34146118164062,
      "learning_rate": 4.2475652224995056e-07,
      "loss": 4.1025,
      "step": 22533
    },
    {
      "epoch": 0.9097295115058539,
      "grad_norm": 92.45654296875,
      "learning_rate": 4.243795601253331e-07,
      "loss": 4.9714,
      "step": 22534
    },
    {
      "epoch": 0.9097698829228906,
      "grad_norm": 78.66818237304688,
      "learning_rate": 4.240027617214648e-07,
      "loss": 3.7524,
      "step": 22535
    },
    {
      "epoch": 0.9098102543399273,
      "grad_norm": 110.74456024169922,
      "learning_rate": 4.2362612704478943e-07,
      "loss": 4.9862,
      "step": 22536
    },
    {
      "epoch": 0.9098506257569641,
      "grad_norm": 101.22157287597656,
      "learning_rate": 4.2324965610174295e-07,
      "loss": 4.2722,
      "step": 22537
    },
    {
      "epoch": 0.9098909971740008,
      "grad_norm": 89.31285858154297,
      "learning_rate": 4.2287334889876463e-07,
      "loss": 4.8105,
      "step": 22538
    },
    {
      "epoch": 0.9099313685910375,
      "grad_norm": 135.58592224121094,
      "learning_rate": 4.224972054422871e-07,
      "loss": 4.3903,
      "step": 22539
    },
    {
      "epoch": 0.9099717400080742,
      "grad_norm": 113.68267059326172,
      "learning_rate": 4.2212122573874303e-07,
      "loss": 4.7438,
      "step": 22540
    },
    {
      "epoch": 0.9100121114251111,
      "grad_norm": 124.98662567138672,
      "learning_rate": 4.2174540979455946e-07,
      "loss": 5.9569,
      "step": 22541
    },
    {
      "epoch": 0.9100524828421478,
      "grad_norm": 74.42266845703125,
      "learning_rate": 4.213697576161635e-07,
      "loss": 3.881,
      "step": 22542
    },
    {
      "epoch": 0.9100928542591845,
      "grad_norm": 104.63728332519531,
      "learning_rate": 4.2099426920997554e-07,
      "loss": 3.1857,
      "step": 22543
    },
    {
      "epoch": 0.9101332256762212,
      "grad_norm": 127.31468963623047,
      "learning_rate": 4.206189445824194e-07,
      "loss": 5.5327,
      "step": 22544
    },
    {
      "epoch": 0.910173597093258,
      "grad_norm": 451.0248107910156,
      "learning_rate": 4.202437837399076e-07,
      "loss": 4.5629,
      "step": 22545
    },
    {
      "epoch": 0.9102139685102947,
      "grad_norm": 81.5298080444336,
      "learning_rate": 4.1986878668885733e-07,
      "loss": 4.5401,
      "step": 22546
    },
    {
      "epoch": 0.9102543399273314,
      "grad_norm": 111.64356994628906,
      "learning_rate": 4.1949395343568013e-07,
      "loss": 6.6865,
      "step": 22547
    },
    {
      "epoch": 0.9102947113443682,
      "grad_norm": 75.12097930908203,
      "learning_rate": 4.1911928398678523e-07,
      "loss": 4.1526,
      "step": 22548
    },
    {
      "epoch": 0.910335082761405,
      "grad_norm": 89.09241485595703,
      "learning_rate": 4.187447783485754e-07,
      "loss": 3.8244,
      "step": 22549
    },
    {
      "epoch": 0.9103754541784417,
      "grad_norm": 119.7037582397461,
      "learning_rate": 4.183704365274566e-07,
      "loss": 4.0812,
      "step": 22550
    },
    {
      "epoch": 0.9104158255954784,
      "grad_norm": 92.9798812866211,
      "learning_rate": 4.1799625852982916e-07,
      "loss": 4.6162,
      "step": 22551
    },
    {
      "epoch": 0.9104561970125151,
      "grad_norm": 179.648681640625,
      "learning_rate": 4.1762224436209143e-07,
      "loss": 5.1069,
      "step": 22552
    },
    {
      "epoch": 0.9104965684295518,
      "grad_norm": 119.7918701171875,
      "learning_rate": 4.172483940306349e-07,
      "loss": 4.4922,
      "step": 22553
    },
    {
      "epoch": 0.9105369398465886,
      "grad_norm": 93.80738067626953,
      "learning_rate": 4.168747075418533e-07,
      "loss": 5.3382,
      "step": 22554
    },
    {
      "epoch": 0.9105773112636254,
      "grad_norm": 88.37979125976562,
      "learning_rate": 4.165011849021372e-07,
      "loss": 5.6166,
      "step": 22555
    },
    {
      "epoch": 0.9106176826806621,
      "grad_norm": 109.9511489868164,
      "learning_rate": 4.1612782611787137e-07,
      "loss": 4.419,
      "step": 22556
    },
    {
      "epoch": 0.9106580540976988,
      "grad_norm": 100.2838134765625,
      "learning_rate": 4.157546311954397e-07,
      "loss": 5.3653,
      "step": 22557
    },
    {
      "epoch": 0.9106984255147356,
      "grad_norm": 92.03547668457031,
      "learning_rate": 4.153816001412225e-07,
      "loss": 5.0203,
      "step": 22558
    },
    {
      "epoch": 0.9107387969317723,
      "grad_norm": 99.47691345214844,
      "learning_rate": 4.1500873296159814e-07,
      "loss": 4.9352,
      "step": 22559
    },
    {
      "epoch": 0.910779168348809,
      "grad_norm": 160.21881103515625,
      "learning_rate": 4.1463602966294257e-07,
      "loss": 6.3551,
      "step": 22560
    },
    {
      "epoch": 0.9108195397658457,
      "grad_norm": 76.3650894165039,
      "learning_rate": 4.142634902516274e-07,
      "loss": 4.7561,
      "step": 22561
    },
    {
      "epoch": 0.9108599111828826,
      "grad_norm": 96.8921127319336,
      "learning_rate": 4.138911147340219e-07,
      "loss": 3.8808,
      "step": 22562
    },
    {
      "epoch": 0.9109002825999193,
      "grad_norm": 75.24861145019531,
      "learning_rate": 4.1351890311649324e-07,
      "loss": 5.4515,
      "step": 22563
    },
    {
      "epoch": 0.910940654016956,
      "grad_norm": 97.03852081298828,
      "learning_rate": 4.131468554054063e-07,
      "loss": 4.9189,
      "step": 22564
    },
    {
      "epoch": 0.9109810254339927,
      "grad_norm": 73.19166564941406,
      "learning_rate": 4.1277497160712054e-07,
      "loss": 5.3167,
      "step": 22565
    },
    {
      "epoch": 0.9110213968510295,
      "grad_norm": 83.43738555908203,
      "learning_rate": 4.124032517279941e-07,
      "loss": 4.3451,
      "step": 22566
    },
    {
      "epoch": 0.9110617682680662,
      "grad_norm": 93.0482406616211,
      "learning_rate": 4.120316957743842e-07,
      "loss": 5.3975,
      "step": 22567
    },
    {
      "epoch": 0.9111021396851029,
      "grad_norm": 123.19404602050781,
      "learning_rate": 4.116603037526434e-07,
      "loss": 5.3626,
      "step": 22568
    },
    {
      "epoch": 0.9111425111021397,
      "grad_norm": 85.02410888671875,
      "learning_rate": 4.1128907566912014e-07,
      "loss": 5.4751,
      "step": 22569
    },
    {
      "epoch": 0.9111828825191765,
      "grad_norm": 104.6572494506836,
      "learning_rate": 4.1091801153016254e-07,
      "loss": 4.7744,
      "step": 22570
    },
    {
      "epoch": 0.9112232539362132,
      "grad_norm": 141.13214111328125,
      "learning_rate": 4.1054711134211447e-07,
      "loss": 5.292,
      "step": 22571
    },
    {
      "epoch": 0.9112636253532499,
      "grad_norm": 141.6839599609375,
      "learning_rate": 4.101763751113197e-07,
      "loss": 6.0956,
      "step": 22572
    },
    {
      "epoch": 0.9113039967702866,
      "grad_norm": 98.67327880859375,
      "learning_rate": 4.098058028441132e-07,
      "loss": 5.6618,
      "step": 22573
    },
    {
      "epoch": 0.9113443681873233,
      "grad_norm": 187.19981384277344,
      "learning_rate": 4.094353945468332e-07,
      "loss": 5.9586,
      "step": 22574
    },
    {
      "epoch": 0.9113847396043601,
      "grad_norm": 66.92394256591797,
      "learning_rate": 4.090651502258114e-07,
      "loss": 3.1547,
      "step": 22575
    },
    {
      "epoch": 0.9114251110213969,
      "grad_norm": 266.60345458984375,
      "learning_rate": 4.0869506988738153e-07,
      "loss": 4.3619,
      "step": 22576
    },
    {
      "epoch": 0.9114654824384336,
      "grad_norm": 89.43921661376953,
      "learning_rate": 4.083251535378663e-07,
      "loss": 4.0404,
      "step": 22577
    },
    {
      "epoch": 0.9115058538554703,
      "grad_norm": 111.62078094482422,
      "learning_rate": 4.07955401183594e-07,
      "loss": 5.2359,
      "step": 22578
    },
    {
      "epoch": 0.9115462252725071,
      "grad_norm": 131.68624877929688,
      "learning_rate": 4.075858128308852e-07,
      "loss": 5.2136,
      "step": 22579
    },
    {
      "epoch": 0.9115865966895438,
      "grad_norm": 99.27706146240234,
      "learning_rate": 4.0721638848605804e-07,
      "loss": 3.8808,
      "step": 22580
    },
    {
      "epoch": 0.9116269681065805,
      "grad_norm": 100.4251708984375,
      "learning_rate": 4.0684712815543095e-07,
      "loss": 6.1782,
      "step": 22581
    },
    {
      "epoch": 0.9116673395236172,
      "grad_norm": 81.13536834716797,
      "learning_rate": 4.064780318453154e-07,
      "loss": 3.7813,
      "step": 22582
    },
    {
      "epoch": 0.9117077109406541,
      "grad_norm": 113.23802185058594,
      "learning_rate": 4.0610909956202313e-07,
      "loss": 3.8633,
      "step": 22583
    },
    {
      "epoch": 0.9117480823576908,
      "grad_norm": 117.6124038696289,
      "learning_rate": 4.057403313118635e-07,
      "loss": 3.6286,
      "step": 22584
    },
    {
      "epoch": 0.9117884537747275,
      "grad_norm": 103.43693542480469,
      "learning_rate": 4.053717271011393e-07,
      "loss": 5.1867,
      "step": 22585
    },
    {
      "epoch": 0.9118288251917642,
      "grad_norm": 113.75435638427734,
      "learning_rate": 4.050032869361531e-07,
      "loss": 4.3826,
      "step": 22586
    },
    {
      "epoch": 0.911869196608801,
      "grad_norm": 111.82300567626953,
      "learning_rate": 4.0463501082320443e-07,
      "loss": 5.1808,
      "step": 22587
    },
    {
      "epoch": 0.9119095680258377,
      "grad_norm": 141.70318603515625,
      "learning_rate": 4.042668987685916e-07,
      "loss": 4.4719,
      "step": 22588
    },
    {
      "epoch": 0.9119499394428744,
      "grad_norm": 90.19123840332031,
      "learning_rate": 4.038989507786062e-07,
      "loss": 5.4219,
      "step": 22589
    },
    {
      "epoch": 0.9119903108599112,
      "grad_norm": 94.49925994873047,
      "learning_rate": 4.035311668595399e-07,
      "loss": 7.0753,
      "step": 22590
    },
    {
      "epoch": 0.912030682276948,
      "grad_norm": 110.95101928710938,
      "learning_rate": 4.031635470176809e-07,
      "loss": 6.9246,
      "step": 22591
    },
    {
      "epoch": 0.9120710536939847,
      "grad_norm": 92.4830093383789,
      "learning_rate": 4.0279609125931764e-07,
      "loss": 5.6796,
      "step": 22592
    },
    {
      "epoch": 0.9121114251110214,
      "grad_norm": 78.18988800048828,
      "learning_rate": 4.024287995907283e-07,
      "loss": 6.3329,
      "step": 22593
    },
    {
      "epoch": 0.9121517965280581,
      "grad_norm": 140.47174072265625,
      "learning_rate": 4.020616720181936e-07,
      "loss": 4.989,
      "step": 22594
    },
    {
      "epoch": 0.9121921679450948,
      "grad_norm": 76.74373626708984,
      "learning_rate": 4.0169470854799275e-07,
      "loss": 4.1001,
      "step": 22595
    },
    {
      "epoch": 0.9122325393621316,
      "grad_norm": 106.86807250976562,
      "learning_rate": 4.0132790918639753e-07,
      "loss": 5.1297,
      "step": 22596
    },
    {
      "epoch": 0.9122729107791684,
      "grad_norm": 104.22102355957031,
      "learning_rate": 4.0096127393968175e-07,
      "loss": 5.0611,
      "step": 22597
    },
    {
      "epoch": 0.9123132821962051,
      "grad_norm": 83.23107147216797,
      "learning_rate": 4.0059480281411266e-07,
      "loss": 5.1237,
      "step": 22598
    },
    {
      "epoch": 0.9123536536132418,
      "grad_norm": 98.37237548828125,
      "learning_rate": 4.0022849581595634e-07,
      "loss": 6.2404,
      "step": 22599
    },
    {
      "epoch": 0.9123940250302786,
      "grad_norm": 114.25121307373047,
      "learning_rate": 3.998623529514756e-07,
      "loss": 4.991,
      "step": 22600
    },
    {
      "epoch": 0.9124343964473153,
      "grad_norm": 227.28407287597656,
      "learning_rate": 3.994963742269298e-07,
      "loss": 5.4952,
      "step": 22601
    },
    {
      "epoch": 0.912474767864352,
      "grad_norm": 76.23579406738281,
      "learning_rate": 3.991305596485784e-07,
      "loss": 5.0406,
      "step": 22602
    },
    {
      "epoch": 0.9125151392813888,
      "grad_norm": 183.0198516845703,
      "learning_rate": 3.987649092226742e-07,
      "loss": 5.2352,
      "step": 22603
    },
    {
      "epoch": 0.9125555106984256,
      "grad_norm": 152.33154296875,
      "learning_rate": 3.9839942295546993e-07,
      "loss": 3.2335,
      "step": 22604
    },
    {
      "epoch": 0.9125958821154623,
      "grad_norm": 79.70323944091797,
      "learning_rate": 3.9803410085321403e-07,
      "loss": 2.8856,
      "step": 22605
    },
    {
      "epoch": 0.912636253532499,
      "grad_norm": 68.10095977783203,
      "learning_rate": 3.976689429221525e-07,
      "loss": 5.1023,
      "step": 22606
    },
    {
      "epoch": 0.9126766249495357,
      "grad_norm": 85.27254486083984,
      "learning_rate": 3.9730394916852824e-07,
      "loss": 5.0024,
      "step": 22607
    },
    {
      "epoch": 0.9127169963665724,
      "grad_norm": 87.83293151855469,
      "learning_rate": 3.96939119598585e-07,
      "loss": 2.2137,
      "step": 22608
    },
    {
      "epoch": 0.9127573677836092,
      "grad_norm": 124.67765808105469,
      "learning_rate": 3.9657445421855676e-07,
      "loss": 4.1633,
      "step": 22609
    },
    {
      "epoch": 0.912797739200646,
      "grad_norm": 107.17337036132812,
      "learning_rate": 3.9620995303467855e-07,
      "loss": 5.9763,
      "step": 22610
    },
    {
      "epoch": 0.9128381106176827,
      "grad_norm": 95.24571228027344,
      "learning_rate": 3.9584561605318427e-07,
      "loss": 4.6445,
      "step": 22611
    },
    {
      "epoch": 0.9128784820347194,
      "grad_norm": 86.7852783203125,
      "learning_rate": 3.9548144328030447e-07,
      "loss": 5.3297,
      "step": 22612
    },
    {
      "epoch": 0.9129188534517562,
      "grad_norm": 87.9393081665039,
      "learning_rate": 3.9511743472226194e-07,
      "loss": 3.0379,
      "step": 22613
    },
    {
      "epoch": 0.9129592248687929,
      "grad_norm": 112.31690216064453,
      "learning_rate": 3.9475359038528285e-07,
      "loss": 5.7595,
      "step": 22614
    },
    {
      "epoch": 0.9129995962858296,
      "grad_norm": 122.46952819824219,
      "learning_rate": 3.943899102755866e-07,
      "loss": 4.8651,
      "step": 22615
    },
    {
      "epoch": 0.9130399677028663,
      "grad_norm": 133.63063049316406,
      "learning_rate": 3.940263943993927e-07,
      "loss": 3.7611,
      "step": 22616
    },
    {
      "epoch": 0.9130803391199032,
      "grad_norm": 76.22112274169922,
      "learning_rate": 3.9366304276291624e-07,
      "loss": 5.6254,
      "step": 22617
    },
    {
      "epoch": 0.9131207105369399,
      "grad_norm": 85.454833984375,
      "learning_rate": 3.932998553723688e-07,
      "loss": 3.8273,
      "step": 22618
    },
    {
      "epoch": 0.9131610819539766,
      "grad_norm": 94.4485092163086,
      "learning_rate": 3.9293683223396105e-07,
      "loss": 5.6785,
      "step": 22619
    },
    {
      "epoch": 0.9132014533710133,
      "grad_norm": 79.18775939941406,
      "learning_rate": 3.9257397335390025e-07,
      "loss": 4.0568,
      "step": 22620
    },
    {
      "epoch": 0.91324182478805,
      "grad_norm": 84.53801727294922,
      "learning_rate": 3.92211278738388e-07,
      "loss": 4.195,
      "step": 22621
    },
    {
      "epoch": 0.9132821962050868,
      "grad_norm": 174.9625244140625,
      "learning_rate": 3.918487483936273e-07,
      "loss": 6.4596,
      "step": 22622
    },
    {
      "epoch": 0.9133225676221235,
      "grad_norm": 84.42135620117188,
      "learning_rate": 3.914863823258164e-07,
      "loss": 6.5981,
      "step": 22623
    },
    {
      "epoch": 0.9133629390391603,
      "grad_norm": 97.0696029663086,
      "learning_rate": 3.911241805411525e-07,
      "loss": 4.7768,
      "step": 22624
    },
    {
      "epoch": 0.913403310456197,
      "grad_norm": 100.06537628173828,
      "learning_rate": 3.9076214304582525e-07,
      "loss": 4.571,
      "step": 22625
    },
    {
      "epoch": 0.9134436818732338,
      "grad_norm": 111.6766586303711,
      "learning_rate": 3.904002698460263e-07,
      "loss": 3.2754,
      "step": 22626
    },
    {
      "epoch": 0.9134840532902705,
      "grad_norm": 108.2803955078125,
      "learning_rate": 3.900385609479429e-07,
      "loss": 4.0204,
      "step": 22627
    },
    {
      "epoch": 0.9135244247073072,
      "grad_norm": 101.48302459716797,
      "learning_rate": 3.896770163577601e-07,
      "loss": 4.9786,
      "step": 22628
    },
    {
      "epoch": 0.9135647961243439,
      "grad_norm": 82.10169219970703,
      "learning_rate": 3.8931563608165746e-07,
      "loss": 3.6139,
      "step": 22629
    },
    {
      "epoch": 0.9136051675413807,
      "grad_norm": 81.22858428955078,
      "learning_rate": 3.889544201258144e-07,
      "loss": 5.8022,
      "step": 22630
    },
    {
      "epoch": 0.9136455389584175,
      "grad_norm": 105.87516021728516,
      "learning_rate": 3.8859336849640827e-07,
      "loss": 4.8557,
      "step": 22631
    },
    {
      "epoch": 0.9136859103754542,
      "grad_norm": 85.90286254882812,
      "learning_rate": 3.8823248119961187e-07,
      "loss": 5.0981,
      "step": 22632
    },
    {
      "epoch": 0.9137262817924909,
      "grad_norm": 142.4116973876953,
      "learning_rate": 3.878717582415936e-07,
      "loss": 4.0621,
      "step": 22633
    },
    {
      "epoch": 0.9137666532095277,
      "grad_norm": 82.55957794189453,
      "learning_rate": 3.87511199628523e-07,
      "loss": 4.5399,
      "step": 22634
    },
    {
      "epoch": 0.9138070246265644,
      "grad_norm": 114.13426971435547,
      "learning_rate": 3.87150805366564e-07,
      "loss": 4.654,
      "step": 22635
    },
    {
      "epoch": 0.9138473960436011,
      "grad_norm": 112.8268051147461,
      "learning_rate": 3.867905754618784e-07,
      "loss": 5.4061,
      "step": 22636
    },
    {
      "epoch": 0.9138877674606378,
      "grad_norm": 122.40874481201172,
      "learning_rate": 3.864305099206245e-07,
      "loss": 4.7653,
      "step": 22637
    },
    {
      "epoch": 0.9139281388776747,
      "grad_norm": 125.93388366699219,
      "learning_rate": 3.860706087489607e-07,
      "loss": 5.3338,
      "step": 22638
    },
    {
      "epoch": 0.9139685102947114,
      "grad_norm": 103.00972747802734,
      "learning_rate": 3.8571087195304003e-07,
      "loss": 3.3904,
      "step": 22639
    },
    {
      "epoch": 0.9140088817117481,
      "grad_norm": 110.1002426147461,
      "learning_rate": 3.8535129953901184e-07,
      "loss": 5.2923,
      "step": 22640
    },
    {
      "epoch": 0.9140492531287848,
      "grad_norm": 71.8766098022461,
      "learning_rate": 3.8499189151302465e-07,
      "loss": 5.033,
      "step": 22641
    },
    {
      "epoch": 0.9140896245458215,
      "grad_norm": 143.31663513183594,
      "learning_rate": 3.846326478812235e-07,
      "loss": 5.0497,
      "step": 22642
    },
    {
      "epoch": 0.9141299959628583,
      "grad_norm": 74.921142578125,
      "learning_rate": 3.842735686497501e-07,
      "loss": 3.7522,
      "step": 22643
    },
    {
      "epoch": 0.914170367379895,
      "grad_norm": 89.09510040283203,
      "learning_rate": 3.8391465382474623e-07,
      "loss": 4.7662,
      "step": 22644
    },
    {
      "epoch": 0.9142107387969318,
      "grad_norm": 103.00955963134766,
      "learning_rate": 3.8355590341234484e-07,
      "loss": 3.81,
      "step": 22645
    },
    {
      "epoch": 0.9142511102139685,
      "grad_norm": 91.85102844238281,
      "learning_rate": 3.831973174186809e-07,
      "loss": 3.9037,
      "step": 22646
    },
    {
      "epoch": 0.9142914816310053,
      "grad_norm": 99.6747817993164,
      "learning_rate": 3.828388958498874e-07,
      "loss": 5.4671,
      "step": 22647
    },
    {
      "epoch": 0.914331853048042,
      "grad_norm": 96.06372833251953,
      "learning_rate": 3.824806387120916e-07,
      "loss": 3.6952,
      "step": 22648
    },
    {
      "epoch": 0.9143722244650787,
      "grad_norm": 82.8200912475586,
      "learning_rate": 3.821225460114164e-07,
      "loss": 5.2862,
      "step": 22649
    },
    {
      "epoch": 0.9144125958821154,
      "grad_norm": 99.60179138183594,
      "learning_rate": 3.8176461775398687e-07,
      "loss": 5.8134,
      "step": 22650
    },
    {
      "epoch": 0.9144529672991522,
      "grad_norm": 77.32080841064453,
      "learning_rate": 3.814068539459226e-07,
      "loss": 5.8131,
      "step": 22651
    },
    {
      "epoch": 0.914493338716189,
      "grad_norm": 78.4613265991211,
      "learning_rate": 3.8104925459334084e-07,
      "loss": 5.1914,
      "step": 22652
    },
    {
      "epoch": 0.9145337101332257,
      "grad_norm": 122.10295867919922,
      "learning_rate": 3.8069181970235457e-07,
      "loss": 5.4091,
      "step": 22653
    },
    {
      "epoch": 0.9145740815502624,
      "grad_norm": 86.80963897705078,
      "learning_rate": 3.8033454927907444e-07,
      "loss": 4.4161,
      "step": 22654
    },
    {
      "epoch": 0.9146144529672992,
      "grad_norm": 65.81777954101562,
      "learning_rate": 3.7997744332960993e-07,
      "loss": 5.7796,
      "step": 22655
    },
    {
      "epoch": 0.9146548243843359,
      "grad_norm": 85.01106262207031,
      "learning_rate": 3.796205018600674e-07,
      "loss": 4.1074,
      "step": 22656
    },
    {
      "epoch": 0.9146951958013726,
      "grad_norm": 225.710693359375,
      "learning_rate": 3.792637248765496e-07,
      "loss": 6.013,
      "step": 22657
    },
    {
      "epoch": 0.9147355672184093,
      "grad_norm": 150.30665588378906,
      "learning_rate": 3.789071123851551e-07,
      "loss": 6.0981,
      "step": 22658
    },
    {
      "epoch": 0.9147759386354462,
      "grad_norm": 186.49545288085938,
      "learning_rate": 3.7855066439198227e-07,
      "loss": 3.931,
      "step": 22659
    },
    {
      "epoch": 0.9148163100524829,
      "grad_norm": 78.8907241821289,
      "learning_rate": 3.781943809031263e-07,
      "loss": 4.502,
      "step": 22660
    },
    {
      "epoch": 0.9148566814695196,
      "grad_norm": 128.0828094482422,
      "learning_rate": 3.7783826192467786e-07,
      "loss": 5.4671,
      "step": 22661
    },
    {
      "epoch": 0.9148970528865563,
      "grad_norm": 88.85189819335938,
      "learning_rate": 3.7748230746272427e-07,
      "loss": 4.6224,
      "step": 22662
    },
    {
      "epoch": 0.914937424303593,
      "grad_norm": 156.9713897705078,
      "learning_rate": 3.7712651752335403e-07,
      "loss": 4.9264,
      "step": 22663
    },
    {
      "epoch": 0.9149777957206298,
      "grad_norm": 123.22698211669922,
      "learning_rate": 3.7677089211265005e-07,
      "loss": 5.5736,
      "step": 22664
    },
    {
      "epoch": 0.9150181671376665,
      "grad_norm": 127.3927001953125,
      "learning_rate": 3.764154312366908e-07,
      "loss": 6.593,
      "step": 22665
    },
    {
      "epoch": 0.9150585385547033,
      "grad_norm": 153.8368377685547,
      "learning_rate": 3.7606013490155577e-07,
      "loss": 4.5323,
      "step": 22666
    },
    {
      "epoch": 0.91509890997174,
      "grad_norm": 108.22089385986328,
      "learning_rate": 3.75705003113318e-07,
      "loss": 3.7726,
      "step": 22667
    },
    {
      "epoch": 0.9151392813887768,
      "grad_norm": 115.42121887207031,
      "learning_rate": 3.753500358780515e-07,
      "loss": 5.3143,
      "step": 22668
    },
    {
      "epoch": 0.9151796528058135,
      "grad_norm": 147.30563354492188,
      "learning_rate": 3.7499523320182363e-07,
      "loss": 3.7654,
      "step": 22669
    },
    {
      "epoch": 0.9152200242228502,
      "grad_norm": 152.74159240722656,
      "learning_rate": 3.746405950907006e-07,
      "loss": 4.9698,
      "step": 22670
    },
    {
      "epoch": 0.9152603956398869,
      "grad_norm": 98.33500671386719,
      "learning_rate": 3.742861215507476e-07,
      "loss": 4.9457,
      "step": 22671
    },
    {
      "epoch": 0.9153007670569236,
      "grad_norm": 78.96183013916016,
      "learning_rate": 3.7393181258802427e-07,
      "loss": 5.6289,
      "step": 22672
    },
    {
      "epoch": 0.9153411384739605,
      "grad_norm": 107.10411071777344,
      "learning_rate": 3.735776682085879e-07,
      "loss": 5.6663,
      "step": 22673
    },
    {
      "epoch": 0.9153815098909972,
      "grad_norm": 102.72467803955078,
      "learning_rate": 3.7322368841849366e-07,
      "loss": 4.6826,
      "step": 22674
    },
    {
      "epoch": 0.9154218813080339,
      "grad_norm": 126.04558563232422,
      "learning_rate": 3.728698732237945e-07,
      "loss": 4.5909,
      "step": 22675
    },
    {
      "epoch": 0.9154622527250706,
      "grad_norm": 81.76039123535156,
      "learning_rate": 3.725162226305401e-07,
      "loss": 4.1388,
      "step": 22676
    },
    {
      "epoch": 0.9155026241421074,
      "grad_norm": 101.75255584716797,
      "learning_rate": 3.7216273664477555e-07,
      "loss": 5.4623,
      "step": 22677
    },
    {
      "epoch": 0.9155429955591441,
      "grad_norm": 77.86776733398438,
      "learning_rate": 3.7180941527254597e-07,
      "loss": 5.8009,
      "step": 22678
    },
    {
      "epoch": 0.9155833669761808,
      "grad_norm": 93.76472473144531,
      "learning_rate": 3.714562585198922e-07,
      "loss": 3.6939,
      "step": 22679
    },
    {
      "epoch": 0.9156237383932176,
      "grad_norm": 94.96702575683594,
      "learning_rate": 3.7110326639285375e-07,
      "loss": 5.187,
      "step": 22680
    },
    {
      "epoch": 0.9156641098102544,
      "grad_norm": 109.14698028564453,
      "learning_rate": 3.7075043889746254e-07,
      "loss": 3.9302,
      "step": 22681
    },
    {
      "epoch": 0.9157044812272911,
      "grad_norm": 92.28284454345703,
      "learning_rate": 3.703977760397526e-07,
      "loss": 4.9502,
      "step": 22682
    },
    {
      "epoch": 0.9157448526443278,
      "grad_norm": 82.69190979003906,
      "learning_rate": 3.700452778257546e-07,
      "loss": 3.5936,
      "step": 22683
    },
    {
      "epoch": 0.9157852240613645,
      "grad_norm": 98.3817367553711,
      "learning_rate": 3.696929442614949e-07,
      "loss": 4.4532,
      "step": 22684
    },
    {
      "epoch": 0.9158255954784013,
      "grad_norm": 92.01377868652344,
      "learning_rate": 3.693407753529976e-07,
      "loss": 3.6504,
      "step": 22685
    },
    {
      "epoch": 0.915865966895438,
      "grad_norm": 113.9681167602539,
      "learning_rate": 3.689887711062834e-07,
      "loss": 4.5738,
      "step": 22686
    },
    {
      "epoch": 0.9159063383124748,
      "grad_norm": 117.98764038085938,
      "learning_rate": 3.686369315273708e-07,
      "loss": 4.6625,
      "step": 22687
    },
    {
      "epoch": 0.9159467097295115,
      "grad_norm": 121.26146697998047,
      "learning_rate": 3.682852566222772e-07,
      "loss": 5.0629,
      "step": 22688
    },
    {
      "epoch": 0.9159870811465483,
      "grad_norm": 113.45915985107422,
      "learning_rate": 3.679337463970123e-07,
      "loss": 4.3065,
      "step": 22689
    },
    {
      "epoch": 0.916027452563585,
      "grad_norm": 96.41285705566406,
      "learning_rate": 3.6758240085758903e-07,
      "loss": 5.4316,
      "step": 22690
    },
    {
      "epoch": 0.9160678239806217,
      "grad_norm": 97.29840850830078,
      "learning_rate": 3.6723122001001253e-07,
      "loss": 6.4095,
      "step": 22691
    },
    {
      "epoch": 0.9161081953976584,
      "grad_norm": 84.35616302490234,
      "learning_rate": 3.6688020386028923e-07,
      "loss": 5.4522,
      "step": 22692
    },
    {
      "epoch": 0.9161485668146951,
      "grad_norm": 103.38031005859375,
      "learning_rate": 3.6652935241441867e-07,
      "loss": 4.9484,
      "step": 22693
    },
    {
      "epoch": 0.916188938231732,
      "grad_norm": 82.4813003540039,
      "learning_rate": 3.6617866567840053e-07,
      "loss": 5.4138,
      "step": 22694
    },
    {
      "epoch": 0.9162293096487687,
      "grad_norm": 133.1181640625,
      "learning_rate": 3.6582814365823007e-07,
      "loss": 5.2579,
      "step": 22695
    },
    {
      "epoch": 0.9162696810658054,
      "grad_norm": 63.58335876464844,
      "learning_rate": 3.6547778635990126e-07,
      "loss": 3.9734,
      "step": 22696
    },
    {
      "epoch": 0.9163100524828421,
      "grad_norm": 110.82901763916016,
      "learning_rate": 3.65127593789405e-07,
      "loss": 3.7807,
      "step": 22697
    },
    {
      "epoch": 0.9163504238998789,
      "grad_norm": 109.54617309570312,
      "learning_rate": 3.647775659527264e-07,
      "loss": 4.2511,
      "step": 22698
    },
    {
      "epoch": 0.9163907953169156,
      "grad_norm": 71.21273803710938,
      "learning_rate": 3.6442770285585185e-07,
      "loss": 5.9988,
      "step": 22699
    },
    {
      "epoch": 0.9164311667339523,
      "grad_norm": 120.31818389892578,
      "learning_rate": 3.6407800450476427e-07,
      "loss": 5.3352,
      "step": 22700
    },
    {
      "epoch": 0.9164715381509891,
      "grad_norm": 143.7984161376953,
      "learning_rate": 3.6372847090544005e-07,
      "loss": 4.6798,
      "step": 22701
    },
    {
      "epoch": 0.9165119095680259,
      "grad_norm": 93.83148193359375,
      "learning_rate": 3.6337910206385775e-07,
      "loss": 4.2147,
      "step": 22702
    },
    {
      "epoch": 0.9165522809850626,
      "grad_norm": 89.37691497802734,
      "learning_rate": 3.630298979859881e-07,
      "loss": 4.5896,
      "step": 22703
    },
    {
      "epoch": 0.9165926524020993,
      "grad_norm": 101.22674560546875,
      "learning_rate": 3.6268085867780525e-07,
      "loss": 5.1231,
      "step": 22704
    },
    {
      "epoch": 0.916633023819136,
      "grad_norm": 163.00486755371094,
      "learning_rate": 3.623319841452733e-07,
      "loss": 5.0682,
      "step": 22705
    },
    {
      "epoch": 0.9166733952361727,
      "grad_norm": 122.2223892211914,
      "learning_rate": 3.619832743943597e-07,
      "loss": 4.6648,
      "step": 22706
    },
    {
      "epoch": 0.9167137666532096,
      "grad_norm": 91.50286865234375,
      "learning_rate": 3.616347294310252e-07,
      "loss": 5.3321,
      "step": 22707
    },
    {
      "epoch": 0.9167541380702463,
      "grad_norm": 189.49464416503906,
      "learning_rate": 3.6128634926123064e-07,
      "loss": 3.5431,
      "step": 22708
    },
    {
      "epoch": 0.916794509487283,
      "grad_norm": 64.93148040771484,
      "learning_rate": 3.609381338909301e-07,
      "loss": 4.414,
      "step": 22709
    },
    {
      "epoch": 0.9168348809043197,
      "grad_norm": 92.88813018798828,
      "learning_rate": 3.605900833260789e-07,
      "loss": 4.91,
      "step": 22710
    },
    {
      "epoch": 0.9168752523213565,
      "grad_norm": 75.33854675292969,
      "learning_rate": 3.602421975726278e-07,
      "loss": 3.7726,
      "step": 22711
    },
    {
      "epoch": 0.9169156237383932,
      "grad_norm": 128.65435791015625,
      "learning_rate": 3.598944766365264e-07,
      "loss": 7.4851,
      "step": 22712
    },
    {
      "epoch": 0.9169559951554299,
      "grad_norm": 96.20549774169922,
      "learning_rate": 3.5954692052371566e-07,
      "loss": 4.9577,
      "step": 22713
    },
    {
      "epoch": 0.9169963665724667,
      "grad_norm": 88.64264678955078,
      "learning_rate": 3.591995292401418e-07,
      "loss": 4.671,
      "step": 22714
    },
    {
      "epoch": 0.9170367379895035,
      "grad_norm": 106.84842681884766,
      "learning_rate": 3.5885230279174234e-07,
      "loss": 4.073,
      "step": 22715
    },
    {
      "epoch": 0.9170771094065402,
      "grad_norm": 94.78504943847656,
      "learning_rate": 3.5850524118445476e-07,
      "loss": 5.2085,
      "step": 22716
    },
    {
      "epoch": 0.9171174808235769,
      "grad_norm": 80.99810028076172,
      "learning_rate": 3.5815834442421316e-07,
      "loss": 4.8352,
      "step": 22717
    },
    {
      "epoch": 0.9171578522406136,
      "grad_norm": 117.10347747802734,
      "learning_rate": 3.578116125169484e-07,
      "loss": 3.4127,
      "step": 22718
    },
    {
      "epoch": 0.9171982236576504,
      "grad_norm": 85.65065002441406,
      "learning_rate": 3.574650454685902e-07,
      "loss": 6.6042,
      "step": 22719
    },
    {
      "epoch": 0.9172385950746871,
      "grad_norm": 117.102294921875,
      "learning_rate": 3.5711864328506265e-07,
      "loss": 5.4325,
      "step": 22720
    },
    {
      "epoch": 0.9172789664917239,
      "grad_norm": 100.80538177490234,
      "learning_rate": 3.567724059722877e-07,
      "loss": 5.8569,
      "step": 22721
    },
    {
      "epoch": 0.9173193379087606,
      "grad_norm": 62.292545318603516,
      "learning_rate": 3.5642633353618616e-07,
      "loss": 5.6402,
      "step": 22722
    },
    {
      "epoch": 0.9173597093257974,
      "grad_norm": 118.27741241455078,
      "learning_rate": 3.560804259826744e-07,
      "loss": 4.8908,
      "step": 22723
    },
    {
      "epoch": 0.9174000807428341,
      "grad_norm": 96.56974792480469,
      "learning_rate": 3.557346833176689e-07,
      "loss": 3.9734,
      "step": 22724
    },
    {
      "epoch": 0.9174404521598708,
      "grad_norm": 70.00497436523438,
      "learning_rate": 3.553891055470782e-07,
      "loss": 5.3618,
      "step": 22725
    },
    {
      "epoch": 0.9174808235769075,
      "grad_norm": 78.0644760131836,
      "learning_rate": 3.550436926768119e-07,
      "loss": 4.8866,
      "step": 22726
    },
    {
      "epoch": 0.9175211949939442,
      "grad_norm": 76.707275390625,
      "learning_rate": 3.5469844471277546e-07,
      "loss": 5.0873,
      "step": 22727
    },
    {
      "epoch": 0.9175615664109811,
      "grad_norm": 108.37959289550781,
      "learning_rate": 3.5435336166087407e-07,
      "loss": 3.9851,
      "step": 22728
    },
    {
      "epoch": 0.9176019378280178,
      "grad_norm": 91.53804016113281,
      "learning_rate": 3.5400844352700415e-07,
      "loss": 4.4866,
      "step": 22729
    },
    {
      "epoch": 0.9176423092450545,
      "grad_norm": 82.3850326538086,
      "learning_rate": 3.536636903170654e-07,
      "loss": 2.9631,
      "step": 22730
    },
    {
      "epoch": 0.9176826806620912,
      "grad_norm": 97.99948120117188,
      "learning_rate": 3.53319102036952e-07,
      "loss": 4.193,
      "step": 22731
    },
    {
      "epoch": 0.917723052079128,
      "grad_norm": 123.70942687988281,
      "learning_rate": 3.5297467869255584e-07,
      "loss": 4.0132,
      "step": 22732
    },
    {
      "epoch": 0.9177634234961647,
      "grad_norm": 97.58342742919922,
      "learning_rate": 3.5263042028976456e-07,
      "loss": 5.218,
      "step": 22733
    },
    {
      "epoch": 0.9178037949132014,
      "grad_norm": 73.70455932617188,
      "learning_rate": 3.522863268344645e-07,
      "loss": 4.7259,
      "step": 22734
    },
    {
      "epoch": 0.9178441663302382,
      "grad_norm": 94.6280517578125,
      "learning_rate": 3.5194239833253876e-07,
      "loss": 4.8314,
      "step": 22735
    },
    {
      "epoch": 0.917884537747275,
      "grad_norm": 101.47674560546875,
      "learning_rate": 3.515986347898692e-07,
      "loss": 4.4956,
      "step": 22736
    },
    {
      "epoch": 0.9179249091643117,
      "grad_norm": 81.99053192138672,
      "learning_rate": 3.512550362123313e-07,
      "loss": 5.3775,
      "step": 22737
    },
    {
      "epoch": 0.9179652805813484,
      "grad_norm": 91.39334869384766,
      "learning_rate": 3.5091160260580124e-07,
      "loss": 5.5928,
      "step": 22738
    },
    {
      "epoch": 0.9180056519983851,
      "grad_norm": 145.10687255859375,
      "learning_rate": 3.505683339761512e-07,
      "loss": 4.8723,
      "step": 22739
    },
    {
      "epoch": 0.9180460234154219,
      "grad_norm": 210.27149963378906,
      "learning_rate": 3.502252303292497e-07,
      "loss": 5.1256,
      "step": 22740
    },
    {
      "epoch": 0.9180863948324586,
      "grad_norm": 99.61337280273438,
      "learning_rate": 3.4988229167096323e-07,
      "loss": 5.2225,
      "step": 22741
    },
    {
      "epoch": 0.9181267662494954,
      "grad_norm": 127.72071075439453,
      "learning_rate": 3.495395180071537e-07,
      "loss": 5.7052,
      "step": 22742
    },
    {
      "epoch": 0.9181671376665321,
      "grad_norm": 92.15733337402344,
      "learning_rate": 3.4919690934368313e-07,
      "loss": 4.2427,
      "step": 22743
    },
    {
      "epoch": 0.9182075090835689,
      "grad_norm": 218.9600067138672,
      "learning_rate": 3.4885446568641014e-07,
      "loss": 4.7091,
      "step": 22744
    },
    {
      "epoch": 0.9182478805006056,
      "grad_norm": 86.67737579345703,
      "learning_rate": 3.4851218704118786e-07,
      "loss": 5.472,
      "step": 22745
    },
    {
      "epoch": 0.9182882519176423,
      "grad_norm": 108.81254577636719,
      "learning_rate": 3.481700734138693e-07,
      "loss": 3.9033,
      "step": 22746
    },
    {
      "epoch": 0.918328623334679,
      "grad_norm": 79.05237579345703,
      "learning_rate": 3.4782812481030436e-07,
      "loss": 5.8032,
      "step": 22747
    },
    {
      "epoch": 0.9183689947517157,
      "grad_norm": 144.05039978027344,
      "learning_rate": 3.474863412363394e-07,
      "loss": 5.0607,
      "step": 22748
    },
    {
      "epoch": 0.9184093661687526,
      "grad_norm": 119.55010986328125,
      "learning_rate": 3.471447226978175e-07,
      "loss": 5.1609,
      "step": 22749
    },
    {
      "epoch": 0.9184497375857893,
      "grad_norm": 114.33485412597656,
      "learning_rate": 3.468032692005796e-07,
      "loss": 3.9584,
      "step": 22750
    },
    {
      "epoch": 0.918490109002826,
      "grad_norm": 519.9002075195312,
      "learning_rate": 3.4646198075046323e-07,
      "loss": 4.5536,
      "step": 22751
    },
    {
      "epoch": 0.9185304804198627,
      "grad_norm": 90.1140365600586,
      "learning_rate": 3.4612085735330593e-07,
      "loss": 5.9437,
      "step": 22752
    },
    {
      "epoch": 0.9185708518368995,
      "grad_norm": 96.67884063720703,
      "learning_rate": 3.4577989901493745e-07,
      "loss": 6.6135,
      "step": 22753
    },
    {
      "epoch": 0.9186112232539362,
      "grad_norm": 82.76031494140625,
      "learning_rate": 3.4543910574118875e-07,
      "loss": 4.2474,
      "step": 22754
    },
    {
      "epoch": 0.9186515946709729,
      "grad_norm": 97.7188720703125,
      "learning_rate": 3.4509847753788516e-07,
      "loss": 4.581,
      "step": 22755
    },
    {
      "epoch": 0.9186919660880097,
      "grad_norm": 96.70735168457031,
      "learning_rate": 3.447580144108531e-07,
      "loss": 4.3732,
      "step": 22756
    },
    {
      "epoch": 0.9187323375050465,
      "grad_norm": 122.18965148925781,
      "learning_rate": 3.444177163659124e-07,
      "loss": 5.7827,
      "step": 22757
    },
    {
      "epoch": 0.9187727089220832,
      "grad_norm": 71.3169174194336,
      "learning_rate": 3.4407758340888055e-07,
      "loss": 4.2471,
      "step": 22758
    },
    {
      "epoch": 0.9188130803391199,
      "grad_norm": 77.1686782836914,
      "learning_rate": 3.4373761554557404e-07,
      "loss": 5.1505,
      "step": 22759
    },
    {
      "epoch": 0.9188534517561566,
      "grad_norm": 141.6078643798828,
      "learning_rate": 3.433978127818061e-07,
      "loss": 4.8925,
      "step": 22760
    },
    {
      "epoch": 0.9188938231731933,
      "grad_norm": 156.67001342773438,
      "learning_rate": 3.430581751233852e-07,
      "loss": 4.9355,
      "step": 22761
    },
    {
      "epoch": 0.9189341945902301,
      "grad_norm": 102.80379486083984,
      "learning_rate": 3.427187025761192e-07,
      "loss": 5.9476,
      "step": 22762
    },
    {
      "epoch": 0.9189745660072669,
      "grad_norm": 124.94197082519531,
      "learning_rate": 3.4237939514581095e-07,
      "loss": 6.2,
      "step": 22763
    },
    {
      "epoch": 0.9190149374243036,
      "grad_norm": 84.37350463867188,
      "learning_rate": 3.420402528382649e-07,
      "loss": 2.772,
      "step": 22764
    },
    {
      "epoch": 0.9190553088413403,
      "grad_norm": 79.0319595336914,
      "learning_rate": 3.4170127565927636e-07,
      "loss": 4.4031,
      "step": 22765
    },
    {
      "epoch": 0.9190956802583771,
      "grad_norm": 86.0551528930664,
      "learning_rate": 3.4136246361464175e-07,
      "loss": 5.4958,
      "step": 22766
    },
    {
      "epoch": 0.9191360516754138,
      "grad_norm": 78.5472183227539,
      "learning_rate": 3.410238167101554e-07,
      "loss": 4.6308,
      "step": 22767
    },
    {
      "epoch": 0.9191764230924505,
      "grad_norm": 114.96055603027344,
      "learning_rate": 3.4068533495160703e-07,
      "loss": 6.2771,
      "step": 22768
    },
    {
      "epoch": 0.9192167945094872,
      "grad_norm": 107.22159576416016,
      "learning_rate": 3.403470183447821e-07,
      "loss": 3.8322,
      "step": 22769
    },
    {
      "epoch": 0.9192571659265241,
      "grad_norm": 90.5762710571289,
      "learning_rate": 3.40008866895466e-07,
      "loss": 4.0696,
      "step": 22770
    },
    {
      "epoch": 0.9192975373435608,
      "grad_norm": 83.5518798828125,
      "learning_rate": 3.3967088060944067e-07,
      "loss": 6.5852,
      "step": 22771
    },
    {
      "epoch": 0.9193379087605975,
      "grad_norm": 87.9593734741211,
      "learning_rate": 3.3933305949248596e-07,
      "loss": 5.4149,
      "step": 22772
    },
    {
      "epoch": 0.9193782801776342,
      "grad_norm": 79.22050476074219,
      "learning_rate": 3.389954035503762e-07,
      "loss": 4.3527,
      "step": 22773
    },
    {
      "epoch": 0.919418651594671,
      "grad_norm": 72.51850128173828,
      "learning_rate": 3.386579127888845e-07,
      "loss": 5.5108,
      "step": 22774
    },
    {
      "epoch": 0.9194590230117077,
      "grad_norm": 79.94557189941406,
      "learning_rate": 3.383205872137818e-07,
      "loss": 6.4399,
      "step": 22775
    },
    {
      "epoch": 0.9194993944287444,
      "grad_norm": 81.85205078125,
      "learning_rate": 3.379834268308346e-07,
      "loss": 4.2611,
      "step": 22776
    },
    {
      "epoch": 0.9195397658457812,
      "grad_norm": 384.9471435546875,
      "learning_rate": 3.3764643164580945e-07,
      "loss": 5.0644,
      "step": 22777
    },
    {
      "epoch": 0.919580137262818,
      "grad_norm": 89.9983901977539,
      "learning_rate": 3.3730960166446614e-07,
      "loss": 6.9601,
      "step": 22778
    },
    {
      "epoch": 0.9196205086798547,
      "grad_norm": 81.60006713867188,
      "learning_rate": 3.369729368925656e-07,
      "loss": 5.6959,
      "step": 22779
    },
    {
      "epoch": 0.9196608800968914,
      "grad_norm": 83.12841033935547,
      "learning_rate": 3.3663643733586325e-07,
      "loss": 5.3338,
      "step": 22780
    },
    {
      "epoch": 0.9197012515139281,
      "grad_norm": 145.7046661376953,
      "learning_rate": 3.363001030001123e-07,
      "loss": 5.0714,
      "step": 22781
    },
    {
      "epoch": 0.9197416229309648,
      "grad_norm": 148.2429962158203,
      "learning_rate": 3.3596393389106254e-07,
      "loss": 4.3235,
      "step": 22782
    },
    {
      "epoch": 0.9197819943480016,
      "grad_norm": 94.3645248413086,
      "learning_rate": 3.356279300144616e-07,
      "loss": 5.2615,
      "step": 22783
    },
    {
      "epoch": 0.9198223657650384,
      "grad_norm": 101.9640121459961,
      "learning_rate": 3.352920913760571e-07,
      "loss": 5.3648,
      "step": 22784
    },
    {
      "epoch": 0.9198627371820751,
      "grad_norm": 113.3882064819336,
      "learning_rate": 3.3495641798158784e-07,
      "loss": 5.6842,
      "step": 22785
    },
    {
      "epoch": 0.9199031085991118,
      "grad_norm": 109.82084655761719,
      "learning_rate": 3.3462090983679473e-07,
      "loss": 4.3836,
      "step": 22786
    },
    {
      "epoch": 0.9199434800161486,
      "grad_norm": 114.20182037353516,
      "learning_rate": 3.342855669474132e-07,
      "loss": 5.0102,
      "step": 22787
    },
    {
      "epoch": 0.9199838514331853,
      "grad_norm": 95.47083282470703,
      "learning_rate": 3.3395038931917867e-07,
      "loss": 4.9651,
      "step": 22788
    },
    {
      "epoch": 0.920024222850222,
      "grad_norm": 98.65284729003906,
      "learning_rate": 3.336153769578199e-07,
      "loss": 7.1271,
      "step": 22789
    },
    {
      "epoch": 0.9200645942672587,
      "grad_norm": 90.63582611083984,
      "learning_rate": 3.3328052986906554e-07,
      "loss": 5.3608,
      "step": 22790
    },
    {
      "epoch": 0.9201049656842956,
      "grad_norm": 78.84636688232422,
      "learning_rate": 3.3294584805864004e-07,
      "loss": 4.422,
      "step": 22791
    },
    {
      "epoch": 0.9201453371013323,
      "grad_norm": 117.4156494140625,
      "learning_rate": 3.3261133153226764e-07,
      "loss": 3.8208,
      "step": 22792
    },
    {
      "epoch": 0.920185708518369,
      "grad_norm": 83.45409393310547,
      "learning_rate": 3.32276980295666e-07,
      "loss": 5.2707,
      "step": 22793
    },
    {
      "epoch": 0.9202260799354057,
      "grad_norm": 100.73680114746094,
      "learning_rate": 3.319427943545517e-07,
      "loss": 5.3307,
      "step": 22794
    },
    {
      "epoch": 0.9202664513524424,
      "grad_norm": 124.50297546386719,
      "learning_rate": 3.3160877371463893e-07,
      "loss": 4.7615,
      "step": 22795
    },
    {
      "epoch": 0.9203068227694792,
      "grad_norm": 115.43106842041016,
      "learning_rate": 3.312749183816388e-07,
      "loss": 5.141,
      "step": 22796
    },
    {
      "epoch": 0.9203471941865159,
      "grad_norm": 130.197998046875,
      "learning_rate": 3.3094122836126006e-07,
      "loss": 7.1246,
      "step": 22797
    },
    {
      "epoch": 0.9203875656035527,
      "grad_norm": 161.4303436279297,
      "learning_rate": 3.3060770365920704e-07,
      "loss": 5.4281,
      "step": 22798
    },
    {
      "epoch": 0.9204279370205894,
      "grad_norm": 126.13578033447266,
      "learning_rate": 3.302743442811829e-07,
      "loss": 7.8418,
      "step": 22799
    },
    {
      "epoch": 0.9204683084376262,
      "grad_norm": 83.90165710449219,
      "learning_rate": 3.299411502328875e-07,
      "loss": 4.522,
      "step": 22800
    },
    {
      "epoch": 0.9205086798546629,
      "grad_norm": 117.27066802978516,
      "learning_rate": 3.296081215200164e-07,
      "loss": 4.0304,
      "step": 22801
    },
    {
      "epoch": 0.9205490512716996,
      "grad_norm": 96.1364517211914,
      "learning_rate": 3.29275258148265e-07,
      "loss": 6.0201,
      "step": 22802
    },
    {
      "epoch": 0.9205894226887363,
      "grad_norm": 117.00517272949219,
      "learning_rate": 3.2894256012332313e-07,
      "loss": 5.099,
      "step": 22803
    },
    {
      "epoch": 0.9206297941057731,
      "grad_norm": 83.09611511230469,
      "learning_rate": 3.286100274508819e-07,
      "loss": 4.8418,
      "step": 22804
    },
    {
      "epoch": 0.9206701655228099,
      "grad_norm": 81.68856048583984,
      "learning_rate": 3.282776601366233e-07,
      "loss": 5.6687,
      "step": 22805
    },
    {
      "epoch": 0.9207105369398466,
      "grad_norm": 106.02503967285156,
      "learning_rate": 3.279454581862318e-07,
      "loss": 3.772,
      "step": 22806
    },
    {
      "epoch": 0.9207509083568833,
      "grad_norm": 101.89085388183594,
      "learning_rate": 3.2761342160538723e-07,
      "loss": 4.4912,
      "step": 22807
    },
    {
      "epoch": 0.9207912797739201,
      "grad_norm": 194.8026885986328,
      "learning_rate": 3.2728155039976727e-07,
      "loss": 5.4716,
      "step": 22808
    },
    {
      "epoch": 0.9208316511909568,
      "grad_norm": 80.03031921386719,
      "learning_rate": 3.2694984457504407e-07,
      "loss": 5.2641,
      "step": 22809
    },
    {
      "epoch": 0.9208720226079935,
      "grad_norm": 78.43696594238281,
      "learning_rate": 3.266183041368909e-07,
      "loss": 3.7144,
      "step": 22810
    },
    {
      "epoch": 0.9209123940250302,
      "grad_norm": 73.99943542480469,
      "learning_rate": 3.262869290909754e-07,
      "loss": 4.2752,
      "step": 22811
    },
    {
      "epoch": 0.9209527654420671,
      "grad_norm": 97.69986724853516,
      "learning_rate": 3.25955719442963e-07,
      "loss": 5.0477,
      "step": 22812
    },
    {
      "epoch": 0.9209931368591038,
      "grad_norm": 70.8699722290039,
      "learning_rate": 3.2562467519851815e-07,
      "loss": 3.8772,
      "step": 22813
    },
    {
      "epoch": 0.9210335082761405,
      "grad_norm": 109.25887298583984,
      "learning_rate": 3.252937963632996e-07,
      "loss": 7.2638,
      "step": 22814
    },
    {
      "epoch": 0.9210738796931772,
      "grad_norm": 74.72107696533203,
      "learning_rate": 3.2496308294296507e-07,
      "loss": 4.4446,
      "step": 22815
    },
    {
      "epoch": 0.9211142511102139,
      "grad_norm": 92.10595703125,
      "learning_rate": 3.2463253494317113e-07,
      "loss": 5.2657,
      "step": 22816
    },
    {
      "epoch": 0.9211546225272507,
      "grad_norm": 120.44562530517578,
      "learning_rate": 3.2430215236956554e-07,
      "loss": 3.8619,
      "step": 22817
    },
    {
      "epoch": 0.9211949939442875,
      "grad_norm": 158.61148071289062,
      "learning_rate": 3.239719352277981e-07,
      "loss": 2.155,
      "step": 22818
    },
    {
      "epoch": 0.9212353653613242,
      "grad_norm": 70.44840240478516,
      "learning_rate": 3.2364188352351555e-07,
      "loss": 4.1923,
      "step": 22819
    },
    {
      "epoch": 0.9212757367783609,
      "grad_norm": 98.38549041748047,
      "learning_rate": 3.2331199726236326e-07,
      "loss": 5.7034,
      "step": 22820
    },
    {
      "epoch": 0.9213161081953977,
      "grad_norm": 89.87480163574219,
      "learning_rate": 3.229822764499768e-07,
      "loss": 3.7495,
      "step": 22821
    },
    {
      "epoch": 0.9213564796124344,
      "grad_norm": 120.06729888916016,
      "learning_rate": 3.2265272109199605e-07,
      "loss": 4.7348,
      "step": 22822
    },
    {
      "epoch": 0.9213968510294711,
      "grad_norm": 115.54451751708984,
      "learning_rate": 3.2232333119405546e-07,
      "loss": 4.5985,
      "step": 22823
    },
    {
      "epoch": 0.9214372224465078,
      "grad_norm": 112.1573715209961,
      "learning_rate": 3.219941067617882e-07,
      "loss": 5.7713,
      "step": 22824
    },
    {
      "epoch": 0.9214775938635447,
      "grad_norm": 76.83240509033203,
      "learning_rate": 3.21665047800821e-07,
      "loss": 4.8226,
      "step": 22825
    },
    {
      "epoch": 0.9215179652805814,
      "grad_norm": 111.41816711425781,
      "learning_rate": 3.2133615431678144e-07,
      "loss": 5.1142,
      "step": 22826
    },
    {
      "epoch": 0.9215583366976181,
      "grad_norm": 73.10643005371094,
      "learning_rate": 3.210074263152918e-07,
      "loss": 5.4113,
      "step": 22827
    },
    {
      "epoch": 0.9215987081146548,
      "grad_norm": 133.48988342285156,
      "learning_rate": 3.206788638019742e-07,
      "loss": 2.9549,
      "step": 22828
    },
    {
      "epoch": 0.9216390795316916,
      "grad_norm": 135.49783325195312,
      "learning_rate": 3.2035046678244417e-07,
      "loss": 2.4825,
      "step": 22829
    },
    {
      "epoch": 0.9216794509487283,
      "grad_norm": 95.50886535644531,
      "learning_rate": 3.2002223526231725e-07,
      "loss": 4.5729,
      "step": 22830
    },
    {
      "epoch": 0.921719822365765,
      "grad_norm": 106.50752258300781,
      "learning_rate": 3.196941692472055e-07,
      "loss": 4.8026,
      "step": 22831
    },
    {
      "epoch": 0.9217601937828018,
      "grad_norm": 103.06763458251953,
      "learning_rate": 3.1936626874271904e-07,
      "loss": 5.5415,
      "step": 22832
    },
    {
      "epoch": 0.9218005651998386,
      "grad_norm": 132.25892639160156,
      "learning_rate": 3.1903853375446326e-07,
      "loss": 5.453,
      "step": 22833
    },
    {
      "epoch": 0.9218409366168753,
      "grad_norm": 103.99195861816406,
      "learning_rate": 3.1871096428804147e-07,
      "loss": 5.0567,
      "step": 22834
    },
    {
      "epoch": 0.921881308033912,
      "grad_norm": 99.63018035888672,
      "learning_rate": 3.1838356034905594e-07,
      "loss": 4.1508,
      "step": 22835
    },
    {
      "epoch": 0.9219216794509487,
      "grad_norm": 99.4348373413086,
      "learning_rate": 3.180563219431032e-07,
      "loss": 4.9282,
      "step": 22836
    },
    {
      "epoch": 0.9219620508679854,
      "grad_norm": 97.47554016113281,
      "learning_rate": 3.1772924907577774e-07,
      "loss": 6.7316,
      "step": 22837
    },
    {
      "epoch": 0.9220024222850222,
      "grad_norm": 117.69046020507812,
      "learning_rate": 3.174023417526717e-07,
      "loss": 4.7069,
      "step": 22838
    },
    {
      "epoch": 0.922042793702059,
      "grad_norm": 121.683349609375,
      "learning_rate": 3.170755999793762e-07,
      "loss": 4.634,
      "step": 22839
    },
    {
      "epoch": 0.9220831651190957,
      "grad_norm": 102.97972106933594,
      "learning_rate": 3.167490237614779e-07,
      "loss": 4.9754,
      "step": 22840
    },
    {
      "epoch": 0.9221235365361324,
      "grad_norm": 74.91168975830078,
      "learning_rate": 3.1642261310455666e-07,
      "loss": 4.21,
      "step": 22841
    },
    {
      "epoch": 0.9221639079531692,
      "grad_norm": 115.711181640625,
      "learning_rate": 3.1609636801419705e-07,
      "loss": 4.8764,
      "step": 22842
    },
    {
      "epoch": 0.9222042793702059,
      "grad_norm": 90.43753814697266,
      "learning_rate": 3.157702884959757e-07,
      "loss": 5.3098,
      "step": 22843
    },
    {
      "epoch": 0.9222446507872426,
      "grad_norm": 145.19973754882812,
      "learning_rate": 3.154443745554703e-07,
      "loss": 4.4665,
      "step": 22844
    },
    {
      "epoch": 0.9222850222042793,
      "grad_norm": 106.30068969726562,
      "learning_rate": 3.151186261982486e-07,
      "loss": 4.3729,
      "step": 22845
    },
    {
      "epoch": 0.9223253936213162,
      "grad_norm": 90.14371490478516,
      "learning_rate": 3.1479304342988293e-07,
      "loss": 5.2316,
      "step": 22846
    },
    {
      "epoch": 0.9223657650383529,
      "grad_norm": 78.03197479248047,
      "learning_rate": 3.1446762625593985e-07,
      "loss": 4.3406,
      "step": 22847
    },
    {
      "epoch": 0.9224061364553896,
      "grad_norm": 70.2151870727539,
      "learning_rate": 3.141423746819827e-07,
      "loss": 5.4765,
      "step": 22848
    },
    {
      "epoch": 0.9224465078724263,
      "grad_norm": 80.21728515625,
      "learning_rate": 3.1381728871357374e-07,
      "loss": 4.6418,
      "step": 22849
    },
    {
      "epoch": 0.922486879289463,
      "grad_norm": 97.13685607910156,
      "learning_rate": 3.134923683562696e-07,
      "loss": 2.2913,
      "step": 22850
    },
    {
      "epoch": 0.9225272507064998,
      "grad_norm": 79.87442779541016,
      "learning_rate": 3.13167613615627e-07,
      "loss": 5.6999,
      "step": 22851
    },
    {
      "epoch": 0.9225676221235365,
      "grad_norm": 110.30182647705078,
      "learning_rate": 3.1284302449719696e-07,
      "loss": 3.2472,
      "step": 22852
    },
    {
      "epoch": 0.9226079935405733,
      "grad_norm": 79.7334976196289,
      "learning_rate": 3.125186010065295e-07,
      "loss": 6.3747,
      "step": 22853
    },
    {
      "epoch": 0.92264836495761,
      "grad_norm": 81.1725845336914,
      "learning_rate": 3.1219434314917364e-07,
      "loss": 6.1937,
      "step": 22854
    },
    {
      "epoch": 0.9226887363746468,
      "grad_norm": 82.05269622802734,
      "learning_rate": 3.1187025093067036e-07,
      "loss": 4.8954,
      "step": 22855
    },
    {
      "epoch": 0.9227291077916835,
      "grad_norm": 90.58586120605469,
      "learning_rate": 3.115463243565631e-07,
      "loss": 4.0885,
      "step": 22856
    },
    {
      "epoch": 0.9227694792087202,
      "grad_norm": 99.61129760742188,
      "learning_rate": 3.1122256343239066e-07,
      "loss": 4.0185,
      "step": 22857
    },
    {
      "epoch": 0.9228098506257569,
      "grad_norm": 94.4243392944336,
      "learning_rate": 3.108989681636865e-07,
      "loss": 4.6955,
      "step": 22858
    },
    {
      "epoch": 0.9228502220427937,
      "grad_norm": 86.07547760009766,
      "learning_rate": 3.105755385559839e-07,
      "loss": 5.9547,
      "step": 22859
    },
    {
      "epoch": 0.9228905934598305,
      "grad_norm": 128.63575744628906,
      "learning_rate": 3.102522746148129e-07,
      "loss": 4.4871,
      "step": 22860
    },
    {
      "epoch": 0.9229309648768672,
      "grad_norm": 82.5073013305664,
      "learning_rate": 3.099291763457024e-07,
      "loss": 4.9245,
      "step": 22861
    },
    {
      "epoch": 0.9229713362939039,
      "grad_norm": 66.5837173461914,
      "learning_rate": 3.096062437541736e-07,
      "loss": 4.0456,
      "step": 22862
    },
    {
      "epoch": 0.9230117077109407,
      "grad_norm": 117.13372802734375,
      "learning_rate": 3.092834768457498e-07,
      "loss": 5.6456,
      "step": 22863
    },
    {
      "epoch": 0.9230520791279774,
      "grad_norm": 83.46444702148438,
      "learning_rate": 3.089608756259488e-07,
      "loss": 3.9506,
      "step": 22864
    },
    {
      "epoch": 0.9230924505450141,
      "grad_norm": 93.11503601074219,
      "learning_rate": 3.0863844010028843e-07,
      "loss": 2.6627,
      "step": 22865
    },
    {
      "epoch": 0.9231328219620508,
      "grad_norm": 80.00973510742188,
      "learning_rate": 3.083161702742776e-07,
      "loss": 4.5176,
      "step": 22866
    },
    {
      "epoch": 0.9231731933790877,
      "grad_norm": 89.80233764648438,
      "learning_rate": 3.079940661534298e-07,
      "loss": 5.5712,
      "step": 22867
    },
    {
      "epoch": 0.9232135647961244,
      "grad_norm": 255.31964111328125,
      "learning_rate": 3.0767212774324927e-07,
      "loss": 4.6856,
      "step": 22868
    },
    {
      "epoch": 0.9232539362131611,
      "grad_norm": 101.10903930664062,
      "learning_rate": 3.0735035504924515e-07,
      "loss": 5.2552,
      "step": 22869
    },
    {
      "epoch": 0.9232943076301978,
      "grad_norm": 105.43309783935547,
      "learning_rate": 3.0702874807691295e-07,
      "loss": 5.0598,
      "step": 22870
    },
    {
      "epoch": 0.9233346790472345,
      "grad_norm": 112.0667495727539,
      "learning_rate": 3.0670730683175607e-07,
      "loss": 3.2139,
      "step": 22871
    },
    {
      "epoch": 0.9233750504642713,
      "grad_norm": 93.44860076904297,
      "learning_rate": 3.0638603131926793e-07,
      "loss": 4.477,
      "step": 22872
    },
    {
      "epoch": 0.923415421881308,
      "grad_norm": 110.30669403076172,
      "learning_rate": 3.060649215449429e-07,
      "loss": 5.0293,
      "step": 22873
    },
    {
      "epoch": 0.9234557932983448,
      "grad_norm": 101.36346435546875,
      "learning_rate": 3.057439775142701e-07,
      "loss": 6.7374,
      "step": 22874
    },
    {
      "epoch": 0.9234961647153815,
      "grad_norm": 78.5107650756836,
      "learning_rate": 3.0542319923273834e-07,
      "loss": 5.8782,
      "step": 22875
    },
    {
      "epoch": 0.9235365361324183,
      "grad_norm": 97.57706451416016,
      "learning_rate": 3.0510258670583103e-07,
      "loss": 4.3016,
      "step": 22876
    },
    {
      "epoch": 0.923576907549455,
      "grad_norm": 102.8721923828125,
      "learning_rate": 3.047821399390316e-07,
      "loss": 3.8803,
      "step": 22877
    },
    {
      "epoch": 0.9236172789664917,
      "grad_norm": 86.51770782470703,
      "learning_rate": 3.0446185893781564e-07,
      "loss": 5.2083,
      "step": 22878
    },
    {
      "epoch": 0.9236576503835284,
      "grad_norm": 121.81934356689453,
      "learning_rate": 3.0414174370766214e-07,
      "loss": 4.8146,
      "step": 22879
    },
    {
      "epoch": 0.9236980218005651,
      "grad_norm": 100.55595397949219,
      "learning_rate": 3.0382179425404337e-07,
      "loss": 4.1328,
      "step": 22880
    },
    {
      "epoch": 0.923738393217602,
      "grad_norm": 86.57759857177734,
      "learning_rate": 3.0350201058243047e-07,
      "loss": 5.0464,
      "step": 22881
    },
    {
      "epoch": 0.9237787646346387,
      "grad_norm": 85.04528045654297,
      "learning_rate": 3.031823926982891e-07,
      "loss": 5.0045,
      "step": 22882
    },
    {
      "epoch": 0.9238191360516754,
      "grad_norm": 90.81562042236328,
      "learning_rate": 3.028629406070849e-07,
      "loss": 5.6913,
      "step": 22883
    },
    {
      "epoch": 0.9238595074687121,
      "grad_norm": 80.07302856445312,
      "learning_rate": 3.025436543142801e-07,
      "loss": 4.6402,
      "step": 22884
    },
    {
      "epoch": 0.9238998788857489,
      "grad_norm": 97.1998291015625,
      "learning_rate": 3.022245338253349e-07,
      "loss": 5.0466,
      "step": 22885
    },
    {
      "epoch": 0.9239402503027856,
      "grad_norm": 135.70730590820312,
      "learning_rate": 3.0190557914570373e-07,
      "loss": 5.5973,
      "step": 22886
    },
    {
      "epoch": 0.9239806217198223,
      "grad_norm": 120.86174011230469,
      "learning_rate": 3.0158679028084e-07,
      "loss": 5.5702,
      "step": 22887
    },
    {
      "epoch": 0.9240209931368591,
      "grad_norm": 83.9153060913086,
      "learning_rate": 3.012681672361939e-07,
      "loss": 5.5965,
      "step": 22888
    },
    {
      "epoch": 0.9240613645538959,
      "grad_norm": 88.23119354248047,
      "learning_rate": 3.009497100172165e-07,
      "loss": 4.2906,
      "step": 22889
    },
    {
      "epoch": 0.9241017359709326,
      "grad_norm": 118.22557067871094,
      "learning_rate": 3.00631418629348e-07,
      "loss": 5.9132,
      "step": 22890
    },
    {
      "epoch": 0.9241421073879693,
      "grad_norm": 77.46731567382812,
      "learning_rate": 3.003132930780328e-07,
      "loss": 4.4385,
      "step": 22891
    },
    {
      "epoch": 0.924182478805006,
      "grad_norm": 71.12183380126953,
      "learning_rate": 2.9999533336871e-07,
      "loss": 4.5782,
      "step": 22892
    },
    {
      "epoch": 0.9242228502220428,
      "grad_norm": 97.66431427001953,
      "learning_rate": 2.9967753950681635e-07,
      "loss": 5.057,
      "step": 22893
    },
    {
      "epoch": 0.9242632216390795,
      "grad_norm": 93.23458862304688,
      "learning_rate": 2.993599114977841e-07,
      "loss": 3.1806,
      "step": 22894
    },
    {
      "epoch": 0.9243035930561163,
      "grad_norm": 135.1039276123047,
      "learning_rate": 2.9904244934704454e-07,
      "loss": 4.0115,
      "step": 22895
    },
    {
      "epoch": 0.924343964473153,
      "grad_norm": 85.46237182617188,
      "learning_rate": 2.987251530600266e-07,
      "loss": 4.8226,
      "step": 22896
    },
    {
      "epoch": 0.9243843358901898,
      "grad_norm": 163.92759704589844,
      "learning_rate": 2.9840802264215485e-07,
      "loss": 4.9116,
      "step": 22897
    },
    {
      "epoch": 0.9244247073072265,
      "grad_norm": 84.19226837158203,
      "learning_rate": 2.980910580988505e-07,
      "loss": 4.3544,
      "step": 22898
    },
    {
      "epoch": 0.9244650787242632,
      "grad_norm": 79.5144271850586,
      "learning_rate": 2.9777425943553374e-07,
      "loss": 4.3634,
      "step": 22899
    },
    {
      "epoch": 0.9245054501412999,
      "grad_norm": 70.96249389648438,
      "learning_rate": 2.974576266576201e-07,
      "loss": 4.5798,
      "step": 22900
    },
    {
      "epoch": 0.9245458215583366,
      "grad_norm": 111.11547088623047,
      "learning_rate": 2.971411597705254e-07,
      "loss": 4.1511,
      "step": 22901
    },
    {
      "epoch": 0.9245861929753735,
      "grad_norm": 69.85163879394531,
      "learning_rate": 2.968248587796574e-07,
      "loss": 2.6668,
      "step": 22902
    },
    {
      "epoch": 0.9246265643924102,
      "grad_norm": 88.8320541381836,
      "learning_rate": 2.9650872369042627e-07,
      "loss": 5.7485,
      "step": 22903
    },
    {
      "epoch": 0.9246669358094469,
      "grad_norm": 79.81106567382812,
      "learning_rate": 2.961927545082366e-07,
      "loss": 5.1658,
      "step": 22904
    },
    {
      "epoch": 0.9247073072264836,
      "grad_norm": 86.25841522216797,
      "learning_rate": 2.958769512384918e-07,
      "loss": 5.3331,
      "step": 22905
    },
    {
      "epoch": 0.9247476786435204,
      "grad_norm": 94.92288208007812,
      "learning_rate": 2.9556131388658984e-07,
      "loss": 4.3287,
      "step": 22906
    },
    {
      "epoch": 0.9247880500605571,
      "grad_norm": 76.2978515625,
      "learning_rate": 2.952458424579274e-07,
      "loss": 6.2865,
      "step": 22907
    },
    {
      "epoch": 0.9248284214775938,
      "grad_norm": 145.5060272216797,
      "learning_rate": 2.949305369578992e-07,
      "loss": 3.7945,
      "step": 22908
    },
    {
      "epoch": 0.9248687928946306,
      "grad_norm": 112.87857818603516,
      "learning_rate": 2.9461539739189637e-07,
      "loss": 5.5811,
      "step": 22909
    },
    {
      "epoch": 0.9249091643116674,
      "grad_norm": 71.2091064453125,
      "learning_rate": 2.943004237653058e-07,
      "loss": 4.4326,
      "step": 22910
    },
    {
      "epoch": 0.9249495357287041,
      "grad_norm": 153.101318359375,
      "learning_rate": 2.9398561608351415e-07,
      "loss": 6.4755,
      "step": 22911
    },
    {
      "epoch": 0.9249899071457408,
      "grad_norm": 98.02384185791016,
      "learning_rate": 2.9367097435190285e-07,
      "loss": 4.2422,
      "step": 22912
    },
    {
      "epoch": 0.9250302785627775,
      "grad_norm": 88.552734375,
      "learning_rate": 2.933564985758508e-07,
      "loss": 5.2565,
      "step": 22913
    },
    {
      "epoch": 0.9250706499798143,
      "grad_norm": 87.08605194091797,
      "learning_rate": 2.930421887607371e-07,
      "loss": 4.3235,
      "step": 22914
    },
    {
      "epoch": 0.925111021396851,
      "grad_norm": 97.12252807617188,
      "learning_rate": 2.927280449119352e-07,
      "loss": 4.586,
      "step": 22915
    },
    {
      "epoch": 0.9251513928138878,
      "grad_norm": 98.65166473388672,
      "learning_rate": 2.924140670348141e-07,
      "loss": 4.1174,
      "step": 22916
    },
    {
      "epoch": 0.9251917642309245,
      "grad_norm": 117.26163482666016,
      "learning_rate": 2.9210025513474627e-07,
      "loss": 3.0325,
      "step": 22917
    },
    {
      "epoch": 0.9252321356479613,
      "grad_norm": 82.79524230957031,
      "learning_rate": 2.917866092170929e-07,
      "loss": 5.0955,
      "step": 22918
    },
    {
      "epoch": 0.925272507064998,
      "grad_norm": 87.52493286132812,
      "learning_rate": 2.914731292872175e-07,
      "loss": 4.1247,
      "step": 22919
    },
    {
      "epoch": 0.9253128784820347,
      "grad_norm": 101.22689056396484,
      "learning_rate": 2.911598153504813e-07,
      "loss": 2.7746,
      "step": 22920
    },
    {
      "epoch": 0.9253532498990714,
      "grad_norm": 110.6612319946289,
      "learning_rate": 2.908466674122412e-07,
      "loss": 5.7039,
      "step": 22921
    },
    {
      "epoch": 0.9253936213161083,
      "grad_norm": 116.55506896972656,
      "learning_rate": 2.9053368547784953e-07,
      "loss": 5.2931,
      "step": 22922
    },
    {
      "epoch": 0.925433992733145,
      "grad_norm": 89.18485260009766,
      "learning_rate": 2.902208695526587e-07,
      "loss": 4.6514,
      "step": 22923
    },
    {
      "epoch": 0.9254743641501817,
      "grad_norm": 95.66285705566406,
      "learning_rate": 2.8990821964201665e-07,
      "loss": 5.3204,
      "step": 22924
    },
    {
      "epoch": 0.9255147355672184,
      "grad_norm": 83.26244354248047,
      "learning_rate": 2.895957357512713e-07,
      "loss": 4.7458,
      "step": 22925
    },
    {
      "epoch": 0.9255551069842551,
      "grad_norm": 75.89494323730469,
      "learning_rate": 2.8928341788576174e-07,
      "loss": 4.9416,
      "step": 22926
    },
    {
      "epoch": 0.9255954784012919,
      "grad_norm": 110.92401123046875,
      "learning_rate": 2.8897126605082924e-07,
      "loss": 5.7315,
      "step": 22927
    },
    {
      "epoch": 0.9256358498183286,
      "grad_norm": 203.64768981933594,
      "learning_rate": 2.886592802518118e-07,
      "loss": 4.523,
      "step": 22928
    },
    {
      "epoch": 0.9256762212353654,
      "grad_norm": 176.33204650878906,
      "learning_rate": 2.8834746049404394e-07,
      "loss": 3.5792,
      "step": 22929
    },
    {
      "epoch": 0.9257165926524021,
      "grad_norm": 117.0156478881836,
      "learning_rate": 2.880358067828548e-07,
      "loss": 3.6423,
      "step": 22930
    },
    {
      "epoch": 0.9257569640694389,
      "grad_norm": 155.43768310546875,
      "learning_rate": 2.8772431912357457e-07,
      "loss": 5.799,
      "step": 22931
    },
    {
      "epoch": 0.9257973354864756,
      "grad_norm": 121.78008270263672,
      "learning_rate": 2.8741299752152896e-07,
      "loss": 5.786,
      "step": 22932
    },
    {
      "epoch": 0.9258377069035123,
      "grad_norm": 85.75669860839844,
      "learning_rate": 2.871018419820404e-07,
      "loss": 4.2959,
      "step": 22933
    },
    {
      "epoch": 0.925878078320549,
      "grad_norm": 76.87791442871094,
      "learning_rate": 2.8679085251042905e-07,
      "loss": 3.4914,
      "step": 22934
    },
    {
      "epoch": 0.9259184497375857,
      "grad_norm": 76.3725357055664,
      "learning_rate": 2.864800291120118e-07,
      "loss": 2.7882,
      "step": 22935
    },
    {
      "epoch": 0.9259588211546226,
      "grad_norm": 137.32151794433594,
      "learning_rate": 2.8616937179210436e-07,
      "loss": 3.8394,
      "step": 22936
    },
    {
      "epoch": 0.9259991925716593,
      "grad_norm": 73.06491088867188,
      "learning_rate": 2.8585888055601694e-07,
      "loss": 3.914,
      "step": 22937
    },
    {
      "epoch": 0.926039563988696,
      "grad_norm": 116.34954833984375,
      "learning_rate": 2.855485554090587e-07,
      "loss": 5.1662,
      "step": 22938
    },
    {
      "epoch": 0.9260799354057327,
      "grad_norm": 122.08248901367188,
      "learning_rate": 2.8523839635653415e-07,
      "loss": 5.8265,
      "step": 22939
    },
    {
      "epoch": 0.9261203068227695,
      "grad_norm": 78.96370697021484,
      "learning_rate": 2.849284034037481e-07,
      "loss": 5.7723,
      "step": 22940
    },
    {
      "epoch": 0.9261606782398062,
      "grad_norm": 121.19998931884766,
      "learning_rate": 2.846185765560017e-07,
      "loss": 4.4411,
      "step": 22941
    },
    {
      "epoch": 0.9262010496568429,
      "grad_norm": 93.5731430053711,
      "learning_rate": 2.8430891581858854e-07,
      "loss": 5.8857,
      "step": 22942
    },
    {
      "epoch": 0.9262414210738797,
      "grad_norm": 92.22334289550781,
      "learning_rate": 2.839994211968056e-07,
      "loss": 3.2033,
      "step": 22943
    },
    {
      "epoch": 0.9262817924909165,
      "grad_norm": 77.76206970214844,
      "learning_rate": 2.8369009269594406e-07,
      "loss": 4.4943,
      "step": 22944
    },
    {
      "epoch": 0.9263221639079532,
      "grad_norm": 120.34394073486328,
      "learning_rate": 2.833809303212931e-07,
      "loss": 5.7863,
      "step": 22945
    },
    {
      "epoch": 0.9263625353249899,
      "grad_norm": 86.0201187133789,
      "learning_rate": 2.830719340781374e-07,
      "loss": 5.4617,
      "step": 22946
    },
    {
      "epoch": 0.9264029067420266,
      "grad_norm": 100.51100158691406,
      "learning_rate": 2.827631039717615e-07,
      "loss": 4.4336,
      "step": 22947
    },
    {
      "epoch": 0.9264432781590634,
      "grad_norm": 104.46954345703125,
      "learning_rate": 2.824544400074447e-07,
      "loss": 4.0839,
      "step": 22948
    },
    {
      "epoch": 0.9264836495761001,
      "grad_norm": 236.6206817626953,
      "learning_rate": 2.821459421904671e-07,
      "loss": 4.1499,
      "step": 22949
    },
    {
      "epoch": 0.9265240209931369,
      "grad_norm": 80.49705505371094,
      "learning_rate": 2.8183761052609894e-07,
      "loss": 5.1429,
      "step": 22950
    },
    {
      "epoch": 0.9265643924101736,
      "grad_norm": 114.81463623046875,
      "learning_rate": 2.8152944501961374e-07,
      "loss": 4.178,
      "step": 22951
    },
    {
      "epoch": 0.9266047638272104,
      "grad_norm": 99.76274871826172,
      "learning_rate": 2.812214456762818e-07,
      "loss": 5.9777,
      "step": 22952
    },
    {
      "epoch": 0.9266451352442471,
      "grad_norm": 83.84960174560547,
      "learning_rate": 2.809136125013678e-07,
      "loss": 5.9043,
      "step": 22953
    },
    {
      "epoch": 0.9266855066612838,
      "grad_norm": 115.88263702392578,
      "learning_rate": 2.806059455001353e-07,
      "loss": 4.848,
      "step": 22954
    },
    {
      "epoch": 0.9267258780783205,
      "grad_norm": 92.01142883300781,
      "learning_rate": 2.8029844467784453e-07,
      "loss": 4.4883,
      "step": 22955
    },
    {
      "epoch": 0.9267662494953572,
      "grad_norm": 95.61767578125,
      "learning_rate": 2.799911100397523e-07,
      "loss": 4.324,
      "step": 22956
    },
    {
      "epoch": 0.9268066209123941,
      "grad_norm": 70.14269256591797,
      "learning_rate": 2.796839415911168e-07,
      "loss": 3.7722,
      "step": 22957
    },
    {
      "epoch": 0.9268469923294308,
      "grad_norm": 76.59635925292969,
      "learning_rate": 2.793769393371859e-07,
      "loss": 5.615,
      "step": 22958
    },
    {
      "epoch": 0.9268873637464675,
      "grad_norm": 83.98115539550781,
      "learning_rate": 2.7907010328320994e-07,
      "loss": 3.6844,
      "step": 22959
    },
    {
      "epoch": 0.9269277351635042,
      "grad_norm": 99.34868621826172,
      "learning_rate": 2.787634334344347e-07,
      "loss": 2.9382,
      "step": 22960
    },
    {
      "epoch": 0.926968106580541,
      "grad_norm": 115.66401672363281,
      "learning_rate": 2.7845692979610594e-07,
      "loss": 6.2085,
      "step": 22961
    },
    {
      "epoch": 0.9270084779975777,
      "grad_norm": 128.76287841796875,
      "learning_rate": 2.7815059237346065e-07,
      "loss": 4.6879,
      "step": 22962
    },
    {
      "epoch": 0.9270488494146144,
      "grad_norm": 102.76624298095703,
      "learning_rate": 2.7784442117173794e-07,
      "loss": 5.4048,
      "step": 22963
    },
    {
      "epoch": 0.9270892208316512,
      "grad_norm": 106.93294525146484,
      "learning_rate": 2.7753841619617363e-07,
      "loss": 4.575,
      "step": 22964
    },
    {
      "epoch": 0.927129592248688,
      "grad_norm": 84.13268280029297,
      "learning_rate": 2.7723257745199905e-07,
      "loss": 4.1766,
      "step": 22965
    },
    {
      "epoch": 0.9271699636657247,
      "grad_norm": 74.02491760253906,
      "learning_rate": 2.769269049444423e-07,
      "loss": 5.4426,
      "step": 22966
    },
    {
      "epoch": 0.9272103350827614,
      "grad_norm": 74.19788360595703,
      "learning_rate": 2.766213986787303e-07,
      "loss": 5.2008,
      "step": 22967
    },
    {
      "epoch": 0.9272507064997981,
      "grad_norm": 166.48126220703125,
      "learning_rate": 2.763160586600866e-07,
      "loss": 5.5304,
      "step": 22968
    },
    {
      "epoch": 0.9272910779168348,
      "grad_norm": 218.7227325439453,
      "learning_rate": 2.760108848937337e-07,
      "loss": 3.3662,
      "step": 22969
    },
    {
      "epoch": 0.9273314493338716,
      "grad_norm": 85.54016876220703,
      "learning_rate": 2.757058773848864e-07,
      "loss": 5.6226,
      "step": 22970
    },
    {
      "epoch": 0.9273718207509084,
      "grad_norm": 95.2634048461914,
      "learning_rate": 2.7540103613875934e-07,
      "loss": 4.7547,
      "step": 22971
    },
    {
      "epoch": 0.9274121921679451,
      "grad_norm": 278.0708923339844,
      "learning_rate": 2.750963611605673e-07,
      "loss": 4.9995,
      "step": 22972
    },
    {
      "epoch": 0.9274525635849818,
      "grad_norm": 118.39059448242188,
      "learning_rate": 2.747918524555182e-07,
      "loss": 5.2456,
      "step": 22973
    },
    {
      "epoch": 0.9274929350020186,
      "grad_norm": 154.7207489013672,
      "learning_rate": 2.74487510028818e-07,
      "loss": 5.3439,
      "step": 22974
    },
    {
      "epoch": 0.9275333064190553,
      "grad_norm": 121.4267349243164,
      "learning_rate": 2.741833338856703e-07,
      "loss": 4.4912,
      "step": 22975
    },
    {
      "epoch": 0.927573677836092,
      "grad_norm": 85.98677825927734,
      "learning_rate": 2.738793240312765e-07,
      "loss": 4.3541,
      "step": 22976
    },
    {
      "epoch": 0.9276140492531287,
      "grad_norm": 99.74962615966797,
      "learning_rate": 2.735754804708357e-07,
      "loss": 5.3197,
      "step": 22977
    },
    {
      "epoch": 0.9276544206701656,
      "grad_norm": 86.49958801269531,
      "learning_rate": 2.7327180320953937e-07,
      "loss": 5.7218,
      "step": 22978
    },
    {
      "epoch": 0.9276947920872023,
      "grad_norm": 99.80229949951172,
      "learning_rate": 2.7296829225258225e-07,
      "loss": 3.9724,
      "step": 22979
    },
    {
      "epoch": 0.927735163504239,
      "grad_norm": 72.33187866210938,
      "learning_rate": 2.726649476051535e-07,
      "loss": 5.8348,
      "step": 22980
    },
    {
      "epoch": 0.9277755349212757,
      "grad_norm": 170.62416076660156,
      "learning_rate": 2.7236176927243893e-07,
      "loss": 6.2324,
      "step": 22981
    },
    {
      "epoch": 0.9278159063383125,
      "grad_norm": 105.30476379394531,
      "learning_rate": 2.720587572596223e-07,
      "loss": 5.7748,
      "step": 22982
    },
    {
      "epoch": 0.9278562777553492,
      "grad_norm": 132.32571411132812,
      "learning_rate": 2.7175591157188487e-07,
      "loss": 5.1776,
      "step": 22983
    },
    {
      "epoch": 0.9278966491723859,
      "grad_norm": 90.1666259765625,
      "learning_rate": 2.7145323221440366e-07,
      "loss": 5.6081,
      "step": 22984
    },
    {
      "epoch": 0.9279370205894227,
      "grad_norm": 116.22374725341797,
      "learning_rate": 2.711507191923557e-07,
      "loss": 5.1667,
      "step": 22985
    },
    {
      "epoch": 0.9279773920064595,
      "grad_norm": 183.26971435546875,
      "learning_rate": 2.708483725109112e-07,
      "loss": 5.2707,
      "step": 22986
    },
    {
      "epoch": 0.9280177634234962,
      "grad_norm": 126.21022033691406,
      "learning_rate": 2.7054619217523947e-07,
      "loss": 5.8285,
      "step": 22987
    },
    {
      "epoch": 0.9280581348405329,
      "grad_norm": 83.28057861328125,
      "learning_rate": 2.7024417819050854e-07,
      "loss": 5.3368,
      "step": 22988
    },
    {
      "epoch": 0.9280985062575696,
      "grad_norm": 100.97227478027344,
      "learning_rate": 2.6994233056188203e-07,
      "loss": 4.4491,
      "step": 22989
    },
    {
      "epoch": 0.9281388776746063,
      "grad_norm": 77.05416870117188,
      "learning_rate": 2.696406492945203e-07,
      "loss": 5.1917,
      "step": 22990
    },
    {
      "epoch": 0.9281792490916431,
      "grad_norm": 111.06532287597656,
      "learning_rate": 2.693391343935803e-07,
      "loss": 6.9854,
      "step": 22991
    },
    {
      "epoch": 0.9282196205086799,
      "grad_norm": 171.1475830078125,
      "learning_rate": 2.69037785864219e-07,
      "loss": 7.2445,
      "step": 22992
    },
    {
      "epoch": 0.9282599919257166,
      "grad_norm": 129.6367645263672,
      "learning_rate": 2.6873660371158795e-07,
      "loss": 5.6019,
      "step": 22993
    },
    {
      "epoch": 0.9283003633427533,
      "grad_norm": 70.8696060180664,
      "learning_rate": 2.6843558794083737e-07,
      "loss": 1.6371,
      "step": 22994
    },
    {
      "epoch": 0.9283407347597901,
      "grad_norm": 77.66356658935547,
      "learning_rate": 2.6813473855711314e-07,
      "loss": 5.2243,
      "step": 22995
    },
    {
      "epoch": 0.9283811061768268,
      "grad_norm": 88.83014678955078,
      "learning_rate": 2.6783405556555895e-07,
      "loss": 4.3702,
      "step": 22996
    },
    {
      "epoch": 0.9284214775938635,
      "grad_norm": 113.74588012695312,
      "learning_rate": 2.675335389713174e-07,
      "loss": 5.4551,
      "step": 22997
    },
    {
      "epoch": 0.9284618490109002,
      "grad_norm": 97.21730041503906,
      "learning_rate": 2.6723318877952433e-07,
      "loss": 5.8288,
      "step": 22998
    },
    {
      "epoch": 0.9285022204279371,
      "grad_norm": 88.11976623535156,
      "learning_rate": 2.669330049953167e-07,
      "loss": 5.3785,
      "step": 22999
    },
    {
      "epoch": 0.9285425918449738,
      "grad_norm": 68.0280532836914,
      "learning_rate": 2.6663298762382604e-07,
      "loss": 5.1462,
      "step": 23000
    },
    {
      "epoch": 0.9285829632620105,
      "grad_norm": 113.8435287475586,
      "learning_rate": 2.6633313667018266e-07,
      "loss": 5.6687,
      "step": 23001
    },
    {
      "epoch": 0.9286233346790472,
      "grad_norm": 113.14227294921875,
      "learning_rate": 2.660334521395125e-07,
      "loss": 4.6536,
      "step": 23002
    },
    {
      "epoch": 0.928663706096084,
      "grad_norm": 90.88905334472656,
      "learning_rate": 2.657339340369391e-07,
      "loss": 5.7087,
      "step": 23003
    },
    {
      "epoch": 0.9287040775131207,
      "grad_norm": 84.60231018066406,
      "learning_rate": 2.6543458236758526e-07,
      "loss": 3.4036,
      "step": 23004
    },
    {
      "epoch": 0.9287444489301574,
      "grad_norm": 79.28224182128906,
      "learning_rate": 2.651353971365689e-07,
      "loss": 4.7472,
      "step": 23005
    },
    {
      "epoch": 0.9287848203471942,
      "grad_norm": 85.07609558105469,
      "learning_rate": 2.6483637834900375e-07,
      "loss": 5.2742,
      "step": 23006
    },
    {
      "epoch": 0.928825191764231,
      "grad_norm": 106.09434509277344,
      "learning_rate": 2.645375260100025e-07,
      "loss": 4.2806,
      "step": 23007
    },
    {
      "epoch": 0.9288655631812677,
      "grad_norm": 95.77806091308594,
      "learning_rate": 2.642388401246765e-07,
      "loss": 6.1194,
      "step": 23008
    },
    {
      "epoch": 0.9289059345983044,
      "grad_norm": 243.19815063476562,
      "learning_rate": 2.639403206981317e-07,
      "loss": 4.4224,
      "step": 23009
    },
    {
      "epoch": 0.9289463060153411,
      "grad_norm": 96.7423095703125,
      "learning_rate": 2.636419677354718e-07,
      "loss": 5.8554,
      "step": 23010
    },
    {
      "epoch": 0.9289866774323778,
      "grad_norm": 79.80320739746094,
      "learning_rate": 2.633437812417994e-07,
      "loss": 3.4638,
      "step": 23011
    },
    {
      "epoch": 0.9290270488494146,
      "grad_norm": 87.20736694335938,
      "learning_rate": 2.6304576122221035e-07,
      "loss": 4.0975,
      "step": 23012
    },
    {
      "epoch": 0.9290674202664514,
      "grad_norm": 106.5561294555664,
      "learning_rate": 2.627479076818007e-07,
      "loss": 4.8655,
      "step": 23013
    },
    {
      "epoch": 0.9291077916834881,
      "grad_norm": 93.96823120117188,
      "learning_rate": 2.6245022062566515e-07,
      "loss": 3.7188,
      "step": 23014
    },
    {
      "epoch": 0.9291481631005248,
      "grad_norm": 92.35401153564453,
      "learning_rate": 2.621527000588908e-07,
      "loss": 4.9714,
      "step": 23015
    },
    {
      "epoch": 0.9291885345175616,
      "grad_norm": 148.19808959960938,
      "learning_rate": 2.618553459865669e-07,
      "loss": 4.8076,
      "step": 23016
    },
    {
      "epoch": 0.9292289059345983,
      "grad_norm": 226.0510711669922,
      "learning_rate": 2.615581584137761e-07,
      "loss": 4.1191,
      "step": 23017
    },
    {
      "epoch": 0.929269277351635,
      "grad_norm": 77.80319213867188,
      "learning_rate": 2.6126113734559976e-07,
      "loss": 4.1198,
      "step": 23018
    },
    {
      "epoch": 0.9293096487686717,
      "grad_norm": 165.6751251220703,
      "learning_rate": 2.6096428278711616e-07,
      "loss": 5.9751,
      "step": 23019
    },
    {
      "epoch": 0.9293500201857086,
      "grad_norm": 94.37190246582031,
      "learning_rate": 2.606675947434012e-07,
      "loss": 6.7412,
      "step": 23020
    },
    {
      "epoch": 0.9293903916027453,
      "grad_norm": 93.69302368164062,
      "learning_rate": 2.6037107321952857e-07,
      "loss": 4.7509,
      "step": 23021
    },
    {
      "epoch": 0.929430763019782,
      "grad_norm": 87.03816223144531,
      "learning_rate": 2.600747182205654e-07,
      "loss": 3.9295,
      "step": 23022
    },
    {
      "epoch": 0.9294711344368187,
      "grad_norm": 110.23792266845703,
      "learning_rate": 2.5977852975158094e-07,
      "loss": 5.1257,
      "step": 23023
    },
    {
      "epoch": 0.9295115058538554,
      "grad_norm": 84.15716552734375,
      "learning_rate": 2.594825078176377e-07,
      "loss": 5.6898,
      "step": 23024
    },
    {
      "epoch": 0.9295518772708922,
      "grad_norm": 83.0863037109375,
      "learning_rate": 2.591866524237985e-07,
      "loss": 4.807,
      "step": 23025
    },
    {
      "epoch": 0.929592248687929,
      "grad_norm": 124.7624740600586,
      "learning_rate": 2.5889096357512135e-07,
      "loss": 3.6387,
      "step": 23026
    },
    {
      "epoch": 0.9296326201049657,
      "grad_norm": 98.25462341308594,
      "learning_rate": 2.5859544127666116e-07,
      "loss": 6.0259,
      "step": 23027
    },
    {
      "epoch": 0.9296729915220024,
      "grad_norm": 69.46448516845703,
      "learning_rate": 2.5830008553347053e-07,
      "loss": 3.6788,
      "step": 23028
    },
    {
      "epoch": 0.9297133629390392,
      "grad_norm": 96.48990631103516,
      "learning_rate": 2.58004896350601e-07,
      "loss": 4.825,
      "step": 23029
    },
    {
      "epoch": 0.9297537343560759,
      "grad_norm": 99.57444763183594,
      "learning_rate": 2.577098737330974e-07,
      "loss": 5.3378,
      "step": 23030
    },
    {
      "epoch": 0.9297941057731126,
      "grad_norm": 94.61903381347656,
      "learning_rate": 2.5741501768600575e-07,
      "loss": 4.4833,
      "step": 23031
    },
    {
      "epoch": 0.9298344771901493,
      "grad_norm": 76.11964416503906,
      "learning_rate": 2.571203282143664e-07,
      "loss": 5.3737,
      "step": 23032
    },
    {
      "epoch": 0.9298748486071862,
      "grad_norm": 73.40550231933594,
      "learning_rate": 2.5682580532321976e-07,
      "loss": 4.6594,
      "step": 23033
    },
    {
      "epoch": 0.9299152200242229,
      "grad_norm": 79.75296783447266,
      "learning_rate": 2.5653144901759854e-07,
      "loss": 5.0715,
      "step": 23034
    },
    {
      "epoch": 0.9299555914412596,
      "grad_norm": 103.69868469238281,
      "learning_rate": 2.562372593025364e-07,
      "loss": 6.0166,
      "step": 23035
    },
    {
      "epoch": 0.9299959628582963,
      "grad_norm": 66.62490844726562,
      "learning_rate": 2.5594323618306493e-07,
      "loss": 4.3965,
      "step": 23036
    },
    {
      "epoch": 0.930036334275333,
      "grad_norm": 75.48802185058594,
      "learning_rate": 2.556493796642101e-07,
      "loss": 4.9427,
      "step": 23037
    },
    {
      "epoch": 0.9300767056923698,
      "grad_norm": 97.61711883544922,
      "learning_rate": 2.553556897509957e-07,
      "loss": 4.6033,
      "step": 23038
    },
    {
      "epoch": 0.9301170771094065,
      "grad_norm": 150.1419219970703,
      "learning_rate": 2.5506216644844315e-07,
      "loss": 6.0643,
      "step": 23039
    },
    {
      "epoch": 0.9301574485264433,
      "grad_norm": 77.94293212890625,
      "learning_rate": 2.5476880976157195e-07,
      "loss": 2.8657,
      "step": 23040
    },
    {
      "epoch": 0.93019781994348,
      "grad_norm": 77.07274627685547,
      "learning_rate": 2.5447561969539793e-07,
      "loss": 5.454,
      "step": 23041
    },
    {
      "epoch": 0.9302381913605168,
      "grad_norm": 67.14599609375,
      "learning_rate": 2.541825962549327e-07,
      "loss": 4.1141,
      "step": 23042
    },
    {
      "epoch": 0.9302785627775535,
      "grad_norm": 101.29727172851562,
      "learning_rate": 2.538897394451867e-07,
      "loss": 4.0336,
      "step": 23043
    },
    {
      "epoch": 0.9303189341945902,
      "grad_norm": 83.1655502319336,
      "learning_rate": 2.535970492711681e-07,
      "loss": 5.0245,
      "step": 23044
    },
    {
      "epoch": 0.9303593056116269,
      "grad_norm": 98.69086456298828,
      "learning_rate": 2.5330452573788076e-07,
      "loss": 4.9757,
      "step": 23045
    },
    {
      "epoch": 0.9303996770286637,
      "grad_norm": 125.93120574951172,
      "learning_rate": 2.530121688503251e-07,
      "loss": 6.9173,
      "step": 23046
    },
    {
      "epoch": 0.9304400484457005,
      "grad_norm": 73.19053649902344,
      "learning_rate": 2.527199786135004e-07,
      "loss": 4.2698,
      "step": 23047
    },
    {
      "epoch": 0.9304804198627372,
      "grad_norm": 132.76434326171875,
      "learning_rate": 2.524279550324027e-07,
      "loss": 4.9815,
      "step": 23048
    },
    {
      "epoch": 0.9305207912797739,
      "grad_norm": 83.62255096435547,
      "learning_rate": 2.5213609811202467e-07,
      "loss": 5.5679,
      "step": 23049
    },
    {
      "epoch": 0.9305611626968107,
      "grad_norm": 125.5215835571289,
      "learning_rate": 2.5184440785735564e-07,
      "loss": 6.0901,
      "step": 23050
    },
    {
      "epoch": 0.9306015341138474,
      "grad_norm": 174.02552795410156,
      "learning_rate": 2.51552884273385e-07,
      "loss": 3.9862,
      "step": 23051
    },
    {
      "epoch": 0.9306419055308841,
      "grad_norm": 89.38224029541016,
      "learning_rate": 2.5126152736509536e-07,
      "loss": 6.1858,
      "step": 23052
    },
    {
      "epoch": 0.9306822769479208,
      "grad_norm": 120.8930892944336,
      "learning_rate": 2.5097033713746946e-07,
      "loss": 5.4164,
      "step": 23053
    },
    {
      "epoch": 0.9307226483649577,
      "grad_norm": 130.0065155029297,
      "learning_rate": 2.5067931359548435e-07,
      "loss": 4.8206,
      "step": 23054
    },
    {
      "epoch": 0.9307630197819944,
      "grad_norm": 90.9667739868164,
      "learning_rate": 2.503884567441162e-07,
      "loss": 6.6011,
      "step": 23055
    },
    {
      "epoch": 0.9308033911990311,
      "grad_norm": 104.31787109375,
      "learning_rate": 2.500977665883386e-07,
      "loss": 4.0646,
      "step": 23056
    },
    {
      "epoch": 0.9308437626160678,
      "grad_norm": 136.80853271484375,
      "learning_rate": 2.4980724313312223e-07,
      "loss": 7.0265,
      "step": 23057
    },
    {
      "epoch": 0.9308841340331045,
      "grad_norm": 177.24562072753906,
      "learning_rate": 2.495168863834341e-07,
      "loss": 6.4861,
      "step": 23058
    },
    {
      "epoch": 0.9309245054501413,
      "grad_norm": 127.47924041748047,
      "learning_rate": 2.492266963442369e-07,
      "loss": 5.1917,
      "step": 23059
    },
    {
      "epoch": 0.930964876867178,
      "grad_norm": 92.83094787597656,
      "learning_rate": 2.489366730204934e-07,
      "loss": 5.5933,
      "step": 23060
    },
    {
      "epoch": 0.9310052482842148,
      "grad_norm": 88.201416015625,
      "learning_rate": 2.4864681641716403e-07,
      "loss": 4.7635,
      "step": 23061
    },
    {
      "epoch": 0.9310456197012515,
      "grad_norm": 105.21389770507812,
      "learning_rate": 2.4835712653920263e-07,
      "loss": 6.1199,
      "step": 23062
    },
    {
      "epoch": 0.9310859911182883,
      "grad_norm": 141.64166259765625,
      "learning_rate": 2.480676033915619e-07,
      "loss": 5.8156,
      "step": 23063
    },
    {
      "epoch": 0.931126362535325,
      "grad_norm": 130.87017822265625,
      "learning_rate": 2.4777824697919227e-07,
      "loss": 3.1597,
      "step": 23064
    },
    {
      "epoch": 0.9311667339523617,
      "grad_norm": 86.5291976928711,
      "learning_rate": 2.474890573070432e-07,
      "loss": 5.2551,
      "step": 23065
    },
    {
      "epoch": 0.9312071053693984,
      "grad_norm": 143.21217346191406,
      "learning_rate": 2.4720003438005623e-07,
      "loss": 5.0625,
      "step": 23066
    },
    {
      "epoch": 0.9312474767864352,
      "grad_norm": 154.97048950195312,
      "learning_rate": 2.469111782031741e-07,
      "loss": 5.0834,
      "step": 23067
    },
    {
      "epoch": 0.931287848203472,
      "grad_norm": 87.46099853515625,
      "learning_rate": 2.466224887813362e-07,
      "loss": 3.9975,
      "step": 23068
    },
    {
      "epoch": 0.9313282196205087,
      "grad_norm": 110.60053253173828,
      "learning_rate": 2.463339661194775e-07,
      "loss": 6.6852,
      "step": 23069
    },
    {
      "epoch": 0.9313685910375454,
      "grad_norm": 149.0240936279297,
      "learning_rate": 2.460456102225317e-07,
      "loss": 3.3869,
      "step": 23070
    },
    {
      "epoch": 0.9314089624545822,
      "grad_norm": 94.63004302978516,
      "learning_rate": 2.4575742109542945e-07,
      "loss": 5.0668,
      "step": 23071
    },
    {
      "epoch": 0.9314493338716189,
      "grad_norm": 61.885074615478516,
      "learning_rate": 2.4546939874309674e-07,
      "loss": 4.9306,
      "step": 23072
    },
    {
      "epoch": 0.9314897052886556,
      "grad_norm": 283.0090637207031,
      "learning_rate": 2.451815431704596e-07,
      "loss": 5.1855,
      "step": 23073
    },
    {
      "epoch": 0.9315300767056923,
      "grad_norm": 94.61748504638672,
      "learning_rate": 2.4489385438243863e-07,
      "loss": 3.944,
      "step": 23074
    },
    {
      "epoch": 0.9315704481227292,
      "grad_norm": 92.66385650634766,
      "learning_rate": 2.4460633238395315e-07,
      "loss": 4.6195,
      "step": 23075
    },
    {
      "epoch": 0.9316108195397659,
      "grad_norm": 128.8609619140625,
      "learning_rate": 2.4431897717991816e-07,
      "loss": 4.9335,
      "step": 23076
    },
    {
      "epoch": 0.9316511909568026,
      "grad_norm": 71.85601806640625,
      "learning_rate": 2.4403178877524857e-07,
      "loss": 3.8245,
      "step": 23077
    },
    {
      "epoch": 0.9316915623738393,
      "grad_norm": 134.5134735107422,
      "learning_rate": 2.437447671748538e-07,
      "loss": 5.2033,
      "step": 23078
    },
    {
      "epoch": 0.931731933790876,
      "grad_norm": 73.04548645019531,
      "learning_rate": 2.4345791238364005e-07,
      "loss": 4.8261,
      "step": 23079
    },
    {
      "epoch": 0.9317723052079128,
      "grad_norm": 101.19158935546875,
      "learning_rate": 2.431712244065121e-07,
      "loss": 5.5031,
      "step": 23080
    },
    {
      "epoch": 0.9318126766249495,
      "grad_norm": 127.46076965332031,
      "learning_rate": 2.4288470324837386e-07,
      "loss": 5.1186,
      "step": 23081
    },
    {
      "epoch": 0.9318530480419863,
      "grad_norm": 200.57931518554688,
      "learning_rate": 2.425983489141226e-07,
      "loss": 5.4737,
      "step": 23082
    },
    {
      "epoch": 0.931893419459023,
      "grad_norm": 109.13872528076172,
      "learning_rate": 2.423121614086532e-07,
      "loss": 4.8314,
      "step": 23083
    },
    {
      "epoch": 0.9319337908760598,
      "grad_norm": 94.12617492675781,
      "learning_rate": 2.420261407368607e-07,
      "loss": 4.8896,
      "step": 23084
    },
    {
      "epoch": 0.9319741622930965,
      "grad_norm": 113.25369262695312,
      "learning_rate": 2.417402869036356e-07,
      "loss": 4.9006,
      "step": 23085
    },
    {
      "epoch": 0.9320145337101332,
      "grad_norm": 87.67366027832031,
      "learning_rate": 2.41454599913864e-07,
      "loss": 3.829,
      "step": 23086
    },
    {
      "epoch": 0.9320549051271699,
      "grad_norm": 113.09141540527344,
      "learning_rate": 2.4116907977242976e-07,
      "loss": 4.4741,
      "step": 23087
    },
    {
      "epoch": 0.9320952765442067,
      "grad_norm": 116.02146911621094,
      "learning_rate": 2.4088372648421564e-07,
      "loss": 4.8779,
      "step": 23088
    },
    {
      "epoch": 0.9321356479612435,
      "grad_norm": 84.17823028564453,
      "learning_rate": 2.4059854005410224e-07,
      "loss": 3.7167,
      "step": 23089
    },
    {
      "epoch": 0.9321760193782802,
      "grad_norm": 147.5225067138672,
      "learning_rate": 2.4031352048696223e-07,
      "loss": 3.9416,
      "step": 23090
    },
    {
      "epoch": 0.9322163907953169,
      "grad_norm": 99.08173370361328,
      "learning_rate": 2.4002866778767176e-07,
      "loss": 4.102,
      "step": 23091
    },
    {
      "epoch": 0.9322567622123537,
      "grad_norm": 147.1240234375,
      "learning_rate": 2.3974398196109916e-07,
      "loss": 5.6679,
      "step": 23092
    },
    {
      "epoch": 0.9322971336293904,
      "grad_norm": 86.9189224243164,
      "learning_rate": 2.3945946301211385e-07,
      "loss": 3.9635,
      "step": 23093
    },
    {
      "epoch": 0.9323375050464271,
      "grad_norm": 89.41912841796875,
      "learning_rate": 2.3917511094557866e-07,
      "loss": 4.9312,
      "step": 23094
    },
    {
      "epoch": 0.9323778764634638,
      "grad_norm": 162.81488037109375,
      "learning_rate": 2.388909257663552e-07,
      "loss": 4.4311,
      "step": 23095
    },
    {
      "epoch": 0.9324182478805007,
      "grad_norm": 87.78697204589844,
      "learning_rate": 2.386069074793029e-07,
      "loss": 3.6589,
      "step": 23096
    },
    {
      "epoch": 0.9324586192975374,
      "grad_norm": 139.82237243652344,
      "learning_rate": 2.3832305608927907e-07,
      "loss": 4.2369,
      "step": 23097
    },
    {
      "epoch": 0.9324989907145741,
      "grad_norm": 100.27062225341797,
      "learning_rate": 2.3803937160113534e-07,
      "loss": 4.5835,
      "step": 23098
    },
    {
      "epoch": 0.9325393621316108,
      "grad_norm": 120.58192443847656,
      "learning_rate": 2.3775585401972224e-07,
      "loss": 4.495,
      "step": 23099
    },
    {
      "epoch": 0.9325797335486475,
      "grad_norm": 100.47618865966797,
      "learning_rate": 2.3747250334988813e-07,
      "loss": 4.6103,
      "step": 23100
    },
    {
      "epoch": 0.9326201049656843,
      "grad_norm": 88.4310302734375,
      "learning_rate": 2.3718931959647807e-07,
      "loss": 4.005,
      "step": 23101
    },
    {
      "epoch": 0.932660476382721,
      "grad_norm": 81.63895416259766,
      "learning_rate": 2.3690630276433146e-07,
      "loss": 6.0528,
      "step": 23102
    },
    {
      "epoch": 0.9327008477997578,
      "grad_norm": 92.35709381103516,
      "learning_rate": 2.366234528582878e-07,
      "loss": 3.6762,
      "step": 23103
    },
    {
      "epoch": 0.9327412192167945,
      "grad_norm": 107.90167236328125,
      "learning_rate": 2.3634076988318545e-07,
      "loss": 5.6285,
      "step": 23104
    },
    {
      "epoch": 0.9327815906338313,
      "grad_norm": 71.79551696777344,
      "learning_rate": 2.3605825384385604e-07,
      "loss": 3.9728,
      "step": 23105
    },
    {
      "epoch": 0.932821962050868,
      "grad_norm": 151.2427978515625,
      "learning_rate": 2.3577590474512912e-07,
      "loss": 5.2408,
      "step": 23106
    },
    {
      "epoch": 0.9328623334679047,
      "grad_norm": 111.21495056152344,
      "learning_rate": 2.3549372259183412e-07,
      "loss": 5.1945,
      "step": 23107
    },
    {
      "epoch": 0.9329027048849414,
      "grad_norm": 87.34171295166016,
      "learning_rate": 2.3521170738879384e-07,
      "loss": 4.1822,
      "step": 23108
    },
    {
      "epoch": 0.9329430763019781,
      "grad_norm": 189.10641479492188,
      "learning_rate": 2.349298591408311e-07,
      "loss": 6.3686,
      "step": 23109
    },
    {
      "epoch": 0.932983447719015,
      "grad_norm": 75.14032745361328,
      "learning_rate": 2.3464817785276427e-07,
      "loss": 3.2697,
      "step": 23110
    },
    {
      "epoch": 0.9330238191360517,
      "grad_norm": 154.3158721923828,
      "learning_rate": 2.3436666352941062e-07,
      "loss": 5.8177,
      "step": 23111
    },
    {
      "epoch": 0.9330641905530884,
      "grad_norm": 120.98570251464844,
      "learning_rate": 2.3408531617558294e-07,
      "loss": 5.0528,
      "step": 23112
    },
    {
      "epoch": 0.9331045619701251,
      "grad_norm": 79.93293762207031,
      "learning_rate": 2.3380413579609073e-07,
      "loss": 3.8833,
      "step": 23113
    },
    {
      "epoch": 0.9331449333871619,
      "grad_norm": 87.46537017822266,
      "learning_rate": 2.3352312239574238e-07,
      "loss": 5.3996,
      "step": 23114
    },
    {
      "epoch": 0.9331853048041986,
      "grad_norm": 81.52508544921875,
      "learning_rate": 2.3324227597934178e-07,
      "loss": 3.5274,
      "step": 23115
    },
    {
      "epoch": 0.9332256762212353,
      "grad_norm": 72.42925262451172,
      "learning_rate": 2.329615965516918e-07,
      "loss": 5.3941,
      "step": 23116
    },
    {
      "epoch": 0.9332660476382721,
      "grad_norm": 86.59378051757812,
      "learning_rate": 2.3268108411759084e-07,
      "loss": 4.1696,
      "step": 23117
    },
    {
      "epoch": 0.9333064190553089,
      "grad_norm": 101.1113510131836,
      "learning_rate": 2.32400738681835e-07,
      "loss": 5.0272,
      "step": 23118
    },
    {
      "epoch": 0.9333467904723456,
      "grad_norm": 77.3752670288086,
      "learning_rate": 2.3212056024921714e-07,
      "loss": 3.9326,
      "step": 23119
    },
    {
      "epoch": 0.9333871618893823,
      "grad_norm": 85.11540985107422,
      "learning_rate": 2.3184054882452788e-07,
      "loss": 6.2634,
      "step": 23120
    },
    {
      "epoch": 0.933427533306419,
      "grad_norm": 256.42852783203125,
      "learning_rate": 2.3156070441255674e-07,
      "loss": 5.0435,
      "step": 23121
    },
    {
      "epoch": 0.9334679047234558,
      "grad_norm": 91.38800811767578,
      "learning_rate": 2.312810270180843e-07,
      "loss": 4.651,
      "step": 23122
    },
    {
      "epoch": 0.9335082761404925,
      "grad_norm": 86.9216537475586,
      "learning_rate": 2.3100151664589564e-07,
      "loss": 4.1671,
      "step": 23123
    },
    {
      "epoch": 0.9335486475575293,
      "grad_norm": 131.5276336669922,
      "learning_rate": 2.3072217330076918e-07,
      "loss": 5.097,
      "step": 23124
    },
    {
      "epoch": 0.933589018974566,
      "grad_norm": 85.25292205810547,
      "learning_rate": 2.3044299698748107e-07,
      "loss": 6.4896,
      "step": 23125
    },
    {
      "epoch": 0.9336293903916028,
      "grad_norm": 111.01222229003906,
      "learning_rate": 2.3016398771080305e-07,
      "loss": 3.2908,
      "step": 23126
    },
    {
      "epoch": 0.9336697618086395,
      "grad_norm": 103.09221649169922,
      "learning_rate": 2.298851454755069e-07,
      "loss": 5.7536,
      "step": 23127
    },
    {
      "epoch": 0.9337101332256762,
      "grad_norm": 80.1744384765625,
      "learning_rate": 2.2960647028635874e-07,
      "loss": 5.8039,
      "step": 23128
    },
    {
      "epoch": 0.9337505046427129,
      "grad_norm": 100.73096466064453,
      "learning_rate": 2.2932796214812591e-07,
      "loss": 4.594,
      "step": 23129
    },
    {
      "epoch": 0.9337908760597498,
      "grad_norm": 118.69239807128906,
      "learning_rate": 2.2904962106556795e-07,
      "loss": 5.9788,
      "step": 23130
    },
    {
      "epoch": 0.9338312474767865,
      "grad_norm": 94.21318054199219,
      "learning_rate": 2.2877144704344544e-07,
      "loss": 4.6594,
      "step": 23131
    },
    {
      "epoch": 0.9338716188938232,
      "grad_norm": 161.00013732910156,
      "learning_rate": 2.2849344008651352e-07,
      "loss": 4.9669,
      "step": 23132
    },
    {
      "epoch": 0.9339119903108599,
      "grad_norm": 175.56338500976562,
      "learning_rate": 2.2821560019952614e-07,
      "loss": 4.3043,
      "step": 23133
    },
    {
      "epoch": 0.9339523617278966,
      "grad_norm": 98.43998718261719,
      "learning_rate": 2.2793792738723285e-07,
      "loss": 4.7152,
      "step": 23134
    },
    {
      "epoch": 0.9339927331449334,
      "grad_norm": 71.18773651123047,
      "learning_rate": 2.2766042165438095e-07,
      "loss": 4.8773,
      "step": 23135
    },
    {
      "epoch": 0.9340331045619701,
      "grad_norm": 102.58000183105469,
      "learning_rate": 2.2738308300571554e-07,
      "loss": 4.571,
      "step": 23136
    },
    {
      "epoch": 0.9340734759790069,
      "grad_norm": 88.82757568359375,
      "learning_rate": 2.271059114459806e-07,
      "loss": 5.2648,
      "step": 23137
    },
    {
      "epoch": 0.9341138473960436,
      "grad_norm": 142.84690856933594,
      "learning_rate": 2.2682890697991233e-07,
      "loss": 3.8253,
      "step": 23138
    },
    {
      "epoch": 0.9341542188130804,
      "grad_norm": 83.7101821899414,
      "learning_rate": 2.2655206961224697e-07,
      "loss": 5.2122,
      "step": 23139
    },
    {
      "epoch": 0.9341945902301171,
      "grad_norm": 145.16476440429688,
      "learning_rate": 2.2627539934771848e-07,
      "loss": 3.4681,
      "step": 23140
    },
    {
      "epoch": 0.9342349616471538,
      "grad_norm": 92.49868774414062,
      "learning_rate": 2.2599889619105864e-07,
      "loss": 5.2538,
      "step": 23141
    },
    {
      "epoch": 0.9342753330641905,
      "grad_norm": 71.7634048461914,
      "learning_rate": 2.2572256014699368e-07,
      "loss": 4.5631,
      "step": 23142
    },
    {
      "epoch": 0.9343157044812272,
      "grad_norm": 133.5796661376953,
      "learning_rate": 2.2544639122024758e-07,
      "loss": 5.4771,
      "step": 23143
    },
    {
      "epoch": 0.9343560758982641,
      "grad_norm": 104.44393157958984,
      "learning_rate": 2.2517038941554325e-07,
      "loss": 4.4356,
      "step": 23144
    },
    {
      "epoch": 0.9343964473153008,
      "grad_norm": 113.84563446044922,
      "learning_rate": 2.2489455473759912e-07,
      "loss": 5.1175,
      "step": 23145
    },
    {
      "epoch": 0.9344368187323375,
      "grad_norm": 2205.692626953125,
      "learning_rate": 2.2461888719113256e-07,
      "loss": 5.2401,
      "step": 23146
    },
    {
      "epoch": 0.9344771901493742,
      "grad_norm": 104.17420196533203,
      "learning_rate": 2.2434338678085422e-07,
      "loss": 4.5525,
      "step": 23147
    },
    {
      "epoch": 0.934517561566411,
      "grad_norm": 83.00444793701172,
      "learning_rate": 2.2406805351147698e-07,
      "loss": 5.2948,
      "step": 23148
    },
    {
      "epoch": 0.9345579329834477,
      "grad_norm": 151.9661102294922,
      "learning_rate": 2.237928873877071e-07,
      "loss": 4.8709,
      "step": 23149
    },
    {
      "epoch": 0.9345983044004844,
      "grad_norm": 79.16020965576172,
      "learning_rate": 2.2351788841425083e-07,
      "loss": 5.3654,
      "step": 23150
    },
    {
      "epoch": 0.9346386758175212,
      "grad_norm": 137.2648468017578,
      "learning_rate": 2.232430565958077e-07,
      "loss": 5.9611,
      "step": 23151
    },
    {
      "epoch": 0.934679047234558,
      "grad_norm": 122.9312973022461,
      "learning_rate": 2.2296839193707843e-07,
      "loss": 7.252,
      "step": 23152
    },
    {
      "epoch": 0.9347194186515947,
      "grad_norm": 115.85506439208984,
      "learning_rate": 2.2269389444276036e-07,
      "loss": 4.6868,
      "step": 23153
    },
    {
      "epoch": 0.9347597900686314,
      "grad_norm": 103.10995483398438,
      "learning_rate": 2.2241956411754306e-07,
      "loss": 3.4615,
      "step": 23154
    },
    {
      "epoch": 0.9348001614856681,
      "grad_norm": 107.74483489990234,
      "learning_rate": 2.2214540096611948e-07,
      "loss": 5.8237,
      "step": 23155
    },
    {
      "epoch": 0.9348405329027049,
      "grad_norm": 114.79338073730469,
      "learning_rate": 2.2187140499317583e-07,
      "loss": 3.8487,
      "step": 23156
    },
    {
      "epoch": 0.9348809043197416,
      "grad_norm": 81.24885559082031,
      "learning_rate": 2.215975762033995e-07,
      "loss": 3.8623,
      "step": 23157
    },
    {
      "epoch": 0.9349212757367784,
      "grad_norm": 74.49149322509766,
      "learning_rate": 2.2132391460146895e-07,
      "loss": 5.077,
      "step": 23158
    },
    {
      "epoch": 0.9349616471538151,
      "grad_norm": 143.61398315429688,
      "learning_rate": 2.210504201920649e-07,
      "loss": 3.7617,
      "step": 23159
    },
    {
      "epoch": 0.9350020185708519,
      "grad_norm": 105.5287094116211,
      "learning_rate": 2.207770929798625e-07,
      "loss": 4.095,
      "step": 23160
    },
    {
      "epoch": 0.9350423899878886,
      "grad_norm": 116.16649627685547,
      "learning_rate": 2.2050393296953688e-07,
      "loss": 4.8582,
      "step": 23161
    },
    {
      "epoch": 0.9350827614049253,
      "grad_norm": 85.98919677734375,
      "learning_rate": 2.2023094016575652e-07,
      "loss": 3.7589,
      "step": 23162
    },
    {
      "epoch": 0.935123132821962,
      "grad_norm": 87.07571411132812,
      "learning_rate": 2.1995811457318993e-07,
      "loss": 5.2747,
      "step": 23163
    },
    {
      "epoch": 0.9351635042389987,
      "grad_norm": 103.68574523925781,
      "learning_rate": 2.1968545619650117e-07,
      "loss": 3.4427,
      "step": 23164
    },
    {
      "epoch": 0.9352038756560356,
      "grad_norm": 91.78366088867188,
      "learning_rate": 2.1941296504035314e-07,
      "loss": 5.8151,
      "step": 23165
    },
    {
      "epoch": 0.9352442470730723,
      "grad_norm": 100.3848648071289,
      "learning_rate": 2.1914064110940437e-07,
      "loss": 4.3804,
      "step": 23166
    },
    {
      "epoch": 0.935284618490109,
      "grad_norm": 178.9583740234375,
      "learning_rate": 2.1886848440831e-07,
      "loss": 6.7092,
      "step": 23167
    },
    {
      "epoch": 0.9353249899071457,
      "grad_norm": 114.75052642822266,
      "learning_rate": 2.1859649494172297e-07,
      "loss": 3.3283,
      "step": 23168
    },
    {
      "epoch": 0.9353653613241825,
      "grad_norm": 115.60090637207031,
      "learning_rate": 2.1832467271429626e-07,
      "loss": 3.3775,
      "step": 23169
    },
    {
      "epoch": 0.9354057327412192,
      "grad_norm": 125.48016357421875,
      "learning_rate": 2.18053017730675e-07,
      "loss": 5.5725,
      "step": 23170
    },
    {
      "epoch": 0.9354461041582559,
      "grad_norm": 199.41604614257812,
      "learning_rate": 2.177815299955044e-07,
      "loss": 5.3145,
      "step": 23171
    },
    {
      "epoch": 0.9354864755752927,
      "grad_norm": 112.61550903320312,
      "learning_rate": 2.1751020951342628e-07,
      "loss": 5.3115,
      "step": 23172
    },
    {
      "epoch": 0.9355268469923295,
      "grad_norm": 73.59479522705078,
      "learning_rate": 2.1723905628908138e-07,
      "loss": 5.4159,
      "step": 23173
    },
    {
      "epoch": 0.9355672184093662,
      "grad_norm": 172.8760223388672,
      "learning_rate": 2.1696807032710266e-07,
      "loss": 5.426,
      "step": 23174
    },
    {
      "epoch": 0.9356075898264029,
      "grad_norm": 87.3919906616211,
      "learning_rate": 2.166972516321264e-07,
      "loss": 4.8765,
      "step": 23175
    },
    {
      "epoch": 0.9356479612434396,
      "grad_norm": 101.46339416503906,
      "learning_rate": 2.1642660020878003e-07,
      "loss": 4.7689,
      "step": 23176
    },
    {
      "epoch": 0.9356883326604764,
      "grad_norm": 78.34974670410156,
      "learning_rate": 2.161561160616943e-07,
      "loss": 4.1692,
      "step": 23177
    },
    {
      "epoch": 0.9357287040775131,
      "grad_norm": 118.3932876586914,
      "learning_rate": 2.158857991954899e-07,
      "loss": 5.1961,
      "step": 23178
    },
    {
      "epoch": 0.9357690754945499,
      "grad_norm": 92.17059326171875,
      "learning_rate": 2.156156496147921e-07,
      "loss": 5.0988,
      "step": 23179
    },
    {
      "epoch": 0.9358094469115866,
      "grad_norm": 90.90113067626953,
      "learning_rate": 2.1534566732421826e-07,
      "loss": 5.2934,
      "step": 23180
    },
    {
      "epoch": 0.9358498183286234,
      "grad_norm": 96.84628295898438,
      "learning_rate": 2.1507585232838467e-07,
      "loss": 5.4829,
      "step": 23181
    },
    {
      "epoch": 0.9358901897456601,
      "grad_norm": 101.58161163330078,
      "learning_rate": 2.148062046319044e-07,
      "loss": 4.4291,
      "step": 23182
    },
    {
      "epoch": 0.9359305611626968,
      "grad_norm": 312.5467529296875,
      "learning_rate": 2.145367242393881e-07,
      "loss": 2.6062,
      "step": 23183
    },
    {
      "epoch": 0.9359709325797335,
      "grad_norm": 158.39158630371094,
      "learning_rate": 2.1426741115544326e-07,
      "loss": 5.7739,
      "step": 23184
    },
    {
      "epoch": 0.9360113039967702,
      "grad_norm": 101.78408813476562,
      "learning_rate": 2.1399826538467505e-07,
      "loss": 4.9096,
      "step": 23185
    },
    {
      "epoch": 0.9360516754138071,
      "grad_norm": 102.62739562988281,
      "learning_rate": 2.1372928693168314e-07,
      "loss": 5.2128,
      "step": 23186
    },
    {
      "epoch": 0.9360920468308438,
      "grad_norm": 103.1064224243164,
      "learning_rate": 2.1346047580106832e-07,
      "loss": 4.1236,
      "step": 23187
    },
    {
      "epoch": 0.9361324182478805,
      "grad_norm": 80.24118041992188,
      "learning_rate": 2.1319183199742578e-07,
      "loss": 4.3111,
      "step": 23188
    },
    {
      "epoch": 0.9361727896649172,
      "grad_norm": 160.68350219726562,
      "learning_rate": 2.129233555253496e-07,
      "loss": 5.5852,
      "step": 23189
    },
    {
      "epoch": 0.936213161081954,
      "grad_norm": 76.26615905761719,
      "learning_rate": 2.1265504638942946e-07,
      "loss": 4.629,
      "step": 23190
    },
    {
      "epoch": 0.9362535324989907,
      "grad_norm": 87.20669555664062,
      "learning_rate": 2.1238690459425283e-07,
      "loss": 5.5996,
      "step": 23191
    },
    {
      "epoch": 0.9362939039160274,
      "grad_norm": 101.06501007080078,
      "learning_rate": 2.1211893014440488e-07,
      "loss": 5.1359,
      "step": 23192
    },
    {
      "epoch": 0.9363342753330642,
      "grad_norm": 82.8780288696289,
      "learning_rate": 2.1185112304446643e-07,
      "loss": 5.3418,
      "step": 23193
    },
    {
      "epoch": 0.936374646750101,
      "grad_norm": 225.99302673339844,
      "learning_rate": 2.115834832990171e-07,
      "loss": 5.2034,
      "step": 23194
    },
    {
      "epoch": 0.9364150181671377,
      "grad_norm": 84.93968963623047,
      "learning_rate": 2.113160109126322e-07,
      "loss": 5.9349,
      "step": 23195
    },
    {
      "epoch": 0.9364553895841744,
      "grad_norm": 130.39686584472656,
      "learning_rate": 2.110487058898858e-07,
      "loss": 6.4993,
      "step": 23196
    },
    {
      "epoch": 0.9364957610012111,
      "grad_norm": 100.91755676269531,
      "learning_rate": 2.1078156823534756e-07,
      "loss": 5.2914,
      "step": 23197
    },
    {
      "epoch": 0.9365361324182478,
      "grad_norm": 66.28380584716797,
      "learning_rate": 2.1051459795358498e-07,
      "loss": 5.0899,
      "step": 23198
    },
    {
      "epoch": 0.9365765038352846,
      "grad_norm": 97.4460220336914,
      "learning_rate": 2.1024779504916215e-07,
      "loss": 6.3579,
      "step": 23199
    },
    {
      "epoch": 0.9366168752523214,
      "grad_norm": 161.7010498046875,
      "learning_rate": 2.099811595266421e-07,
      "loss": 4.9399,
      "step": 23200
    },
    {
      "epoch": 0.9366572466693581,
      "grad_norm": 170.62454223632812,
      "learning_rate": 2.0971469139058342e-07,
      "loss": 5.1994,
      "step": 23201
    },
    {
      "epoch": 0.9366976180863948,
      "grad_norm": 75.47547149658203,
      "learning_rate": 2.0944839064554024e-07,
      "loss": 5.4876,
      "step": 23202
    },
    {
      "epoch": 0.9367379895034316,
      "grad_norm": 76.48249816894531,
      "learning_rate": 2.0918225729606777e-07,
      "loss": 5.8376,
      "step": 23203
    },
    {
      "epoch": 0.9367783609204683,
      "grad_norm": 127.51068878173828,
      "learning_rate": 2.0891629134671466e-07,
      "loss": 5.0032,
      "step": 23204
    },
    {
      "epoch": 0.936818732337505,
      "grad_norm": 87.71070861816406,
      "learning_rate": 2.0865049280203053e-07,
      "loss": 4.4333,
      "step": 23205
    },
    {
      "epoch": 0.9368591037545417,
      "grad_norm": 101.87447357177734,
      "learning_rate": 2.0838486166655735e-07,
      "loss": 7.3354,
      "step": 23206
    },
    {
      "epoch": 0.9368994751715786,
      "grad_norm": 84.74162292480469,
      "learning_rate": 2.0811939794483814e-07,
      "loss": 4.7698,
      "step": 23207
    },
    {
      "epoch": 0.9369398465886153,
      "grad_norm": 123.1986312866211,
      "learning_rate": 2.0785410164141152e-07,
      "loss": 4.3279,
      "step": 23208
    },
    {
      "epoch": 0.936980218005652,
      "grad_norm": 90.93185424804688,
      "learning_rate": 2.075889727608127e-07,
      "loss": 4.02,
      "step": 23209
    },
    {
      "epoch": 0.9370205894226887,
      "grad_norm": 80.3601303100586,
      "learning_rate": 2.0732401130757696e-07,
      "loss": 4.8777,
      "step": 23210
    },
    {
      "epoch": 0.9370609608397255,
      "grad_norm": 111.30606842041016,
      "learning_rate": 2.070592172862318e-07,
      "loss": 5.2636,
      "step": 23211
    },
    {
      "epoch": 0.9371013322567622,
      "grad_norm": 197.4454803466797,
      "learning_rate": 2.0679459070130693e-07,
      "loss": 4.2429,
      "step": 23212
    },
    {
      "epoch": 0.9371417036737989,
      "grad_norm": 94.99808502197266,
      "learning_rate": 2.0653013155732649e-07,
      "loss": 4.8868,
      "step": 23213
    },
    {
      "epoch": 0.9371820750908357,
      "grad_norm": 148.59625244140625,
      "learning_rate": 2.062658398588102e-07,
      "loss": 4.6263,
      "step": 23214
    },
    {
      "epoch": 0.9372224465078725,
      "grad_norm": 85.47725677490234,
      "learning_rate": 2.060017156102778e-07,
      "loss": 5.5202,
      "step": 23215
    },
    {
      "epoch": 0.9372628179249092,
      "grad_norm": 116.20777893066406,
      "learning_rate": 2.057377588162457e-07,
      "loss": 6.2092,
      "step": 23216
    },
    {
      "epoch": 0.9373031893419459,
      "grad_norm": 117.78826904296875,
      "learning_rate": 2.054739694812269e-07,
      "loss": 5.1536,
      "step": 23217
    },
    {
      "epoch": 0.9373435607589826,
      "grad_norm": 110.52606964111328,
      "learning_rate": 2.0521034760973112e-07,
      "loss": 5.8512,
      "step": 23218
    },
    {
      "epoch": 0.9373839321760193,
      "grad_norm": 96.79527282714844,
      "learning_rate": 2.0494689320626594e-07,
      "loss": 3.8002,
      "step": 23219
    },
    {
      "epoch": 0.9374243035930561,
      "grad_norm": 99.20795440673828,
      "learning_rate": 2.0468360627533656e-07,
      "loss": 3.4076,
      "step": 23220
    },
    {
      "epoch": 0.9374646750100929,
      "grad_norm": 121.39485168457031,
      "learning_rate": 2.0442048682144389e-07,
      "loss": 5.2312,
      "step": 23221
    },
    {
      "epoch": 0.9375050464271296,
      "grad_norm": 95.8022689819336,
      "learning_rate": 2.041575348490854e-07,
      "loss": 6.01,
      "step": 23222
    },
    {
      "epoch": 0.9375454178441663,
      "grad_norm": 93.53583526611328,
      "learning_rate": 2.0389475036275862e-07,
      "loss": 4.2388,
      "step": 23223
    },
    {
      "epoch": 0.9375857892612031,
      "grad_norm": 73.92292022705078,
      "learning_rate": 2.0363213336695554e-07,
      "loss": 4.5481,
      "step": 23224
    },
    {
      "epoch": 0.9376261606782398,
      "grad_norm": 105.41410827636719,
      "learning_rate": 2.0336968386616696e-07,
      "loss": 6.8586,
      "step": 23225
    },
    {
      "epoch": 0.9376665320952765,
      "grad_norm": 93.58451080322266,
      "learning_rate": 2.0310740186488043e-07,
      "loss": 4.328,
      "step": 23226
    },
    {
      "epoch": 0.9377069035123132,
      "grad_norm": 107.85625457763672,
      "learning_rate": 2.0284528736758013e-07,
      "loss": 5.4283,
      "step": 23227
    },
    {
      "epoch": 0.9377472749293501,
      "grad_norm": 134.6234893798828,
      "learning_rate": 2.0258334037874694e-07,
      "loss": 5.0656,
      "step": 23228
    },
    {
      "epoch": 0.9377876463463868,
      "grad_norm": 68.66304779052734,
      "learning_rate": 2.023215609028606e-07,
      "loss": 4.941,
      "step": 23229
    },
    {
      "epoch": 0.9378280177634235,
      "grad_norm": 97.7191390991211,
      "learning_rate": 2.020599489443953e-07,
      "loss": 5.1541,
      "step": 23230
    },
    {
      "epoch": 0.9378683891804602,
      "grad_norm": 129.16091918945312,
      "learning_rate": 2.0179850450782634e-07,
      "loss": 4.9492,
      "step": 23231
    },
    {
      "epoch": 0.937908760597497,
      "grad_norm": 90.1644287109375,
      "learning_rate": 2.0153722759762128e-07,
      "loss": 5.7179,
      "step": 23232
    },
    {
      "epoch": 0.9379491320145337,
      "grad_norm": 118.34033203125,
      "learning_rate": 2.0127611821824988e-07,
      "loss": 4.6142,
      "step": 23233
    },
    {
      "epoch": 0.9379895034315704,
      "grad_norm": 85.67288208007812,
      "learning_rate": 2.010151763741741e-07,
      "loss": 4.5823,
      "step": 23234
    },
    {
      "epoch": 0.9380298748486072,
      "grad_norm": 98.04625701904297,
      "learning_rate": 2.0075440206985708e-07,
      "loss": 5.4943,
      "step": 23235
    },
    {
      "epoch": 0.938070246265644,
      "grad_norm": 98.84949493408203,
      "learning_rate": 2.0049379530975632e-07,
      "loss": 6.0232,
      "step": 23236
    },
    {
      "epoch": 0.9381106176826807,
      "grad_norm": 72.3447265625,
      "learning_rate": 2.002333560983294e-07,
      "loss": 6.225,
      "step": 23237
    },
    {
      "epoch": 0.9381509890997174,
      "grad_norm": 140.87693786621094,
      "learning_rate": 1.9997308444002716e-07,
      "loss": 2.2955,
      "step": 23238
    },
    {
      "epoch": 0.9381913605167541,
      "grad_norm": 89.90604400634766,
      "learning_rate": 1.9971298033930053e-07,
      "loss": 6.1576,
      "step": 23239
    },
    {
      "epoch": 0.9382317319337908,
      "grad_norm": 75.32659912109375,
      "learning_rate": 1.9945304380059593e-07,
      "loss": 6.3505,
      "step": 23240
    },
    {
      "epoch": 0.9382721033508277,
      "grad_norm": 68.1341781616211,
      "learning_rate": 1.991932748283598e-07,
      "loss": 2.9002,
      "step": 23241
    },
    {
      "epoch": 0.9383124747678644,
      "grad_norm": 84.69742584228516,
      "learning_rate": 1.9893367342703084e-07,
      "loss": 4.5591,
      "step": 23242
    },
    {
      "epoch": 0.9383528461849011,
      "grad_norm": 136.91908264160156,
      "learning_rate": 1.986742396010499e-07,
      "loss": 5.0728,
      "step": 23243
    },
    {
      "epoch": 0.9383932176019378,
      "grad_norm": 68.48617553710938,
      "learning_rate": 1.9841497335485127e-07,
      "loss": 6.2306,
      "step": 23244
    },
    {
      "epoch": 0.9384335890189746,
      "grad_norm": 92.32537078857422,
      "learning_rate": 1.98155874692868e-07,
      "loss": 4.6392,
      "step": 23245
    },
    {
      "epoch": 0.9384739604360113,
      "grad_norm": 85.72908020019531,
      "learning_rate": 1.9789694361952992e-07,
      "loss": 5.2431,
      "step": 23246
    },
    {
      "epoch": 0.938514331853048,
      "grad_norm": 101.89620208740234,
      "learning_rate": 1.9763818013926572e-07,
      "loss": 5.1401,
      "step": 23247
    },
    {
      "epoch": 0.9385547032700848,
      "grad_norm": 156.69940185546875,
      "learning_rate": 1.973795842564974e-07,
      "loss": 4.0434,
      "step": 23248
    },
    {
      "epoch": 0.9385950746871216,
      "grad_norm": 126.00204467773438,
      "learning_rate": 1.9712115597564806e-07,
      "loss": 4.3024,
      "step": 23249
    },
    {
      "epoch": 0.9386354461041583,
      "grad_norm": 110.29121398925781,
      "learning_rate": 1.9686289530113535e-07,
      "loss": 5.6505,
      "step": 23250
    },
    {
      "epoch": 0.938675817521195,
      "grad_norm": 159.63088989257812,
      "learning_rate": 1.9660480223737567e-07,
      "loss": 4.5482,
      "step": 23251
    },
    {
      "epoch": 0.9387161889382317,
      "grad_norm": 83.32654571533203,
      "learning_rate": 1.9634687678878107e-07,
      "loss": 5.4715,
      "step": 23252
    },
    {
      "epoch": 0.9387565603552684,
      "grad_norm": 115.12811279296875,
      "learning_rate": 1.9608911895976134e-07,
      "loss": 5.3484,
      "step": 23253
    },
    {
      "epoch": 0.9387969317723052,
      "grad_norm": 83.8087158203125,
      "learning_rate": 1.958315287547241e-07,
      "loss": 3.6903,
      "step": 23254
    },
    {
      "epoch": 0.938837303189342,
      "grad_norm": 91.43632507324219,
      "learning_rate": 1.955741061780736e-07,
      "loss": 5.2098,
      "step": 23255
    },
    {
      "epoch": 0.9388776746063787,
      "grad_norm": 87.80996704101562,
      "learning_rate": 1.9531685123421074e-07,
      "loss": 6.1709,
      "step": 23256
    },
    {
      "epoch": 0.9389180460234154,
      "grad_norm": 74.11325073242188,
      "learning_rate": 1.9505976392753423e-07,
      "loss": 6.1319,
      "step": 23257
    },
    {
      "epoch": 0.9389584174404522,
      "grad_norm": 139.91725158691406,
      "learning_rate": 1.9480284426243944e-07,
      "loss": 5.5496,
      "step": 23258
    },
    {
      "epoch": 0.9389987888574889,
      "grad_norm": 68.39385986328125,
      "learning_rate": 1.9454609224331955e-07,
      "loss": 4.7452,
      "step": 23259
    },
    {
      "epoch": 0.9390391602745256,
      "grad_norm": 138.2127685546875,
      "learning_rate": 1.9428950787456325e-07,
      "loss": 7.7862,
      "step": 23260
    },
    {
      "epoch": 0.9390795316915623,
      "grad_norm": 110.85188293457031,
      "learning_rate": 1.9403309116055924e-07,
      "loss": 5.5303,
      "step": 23261
    },
    {
      "epoch": 0.9391199031085992,
      "grad_norm": 101.99816131591797,
      "learning_rate": 1.937768421056907e-07,
      "loss": 4.9113,
      "step": 23262
    },
    {
      "epoch": 0.9391602745256359,
      "grad_norm": 92.43083953857422,
      "learning_rate": 1.9352076071433857e-07,
      "loss": 2.6602,
      "step": 23263
    },
    {
      "epoch": 0.9392006459426726,
      "grad_norm": 69.09087371826172,
      "learning_rate": 1.9326484699088156e-07,
      "loss": 2.7644,
      "step": 23264
    },
    {
      "epoch": 0.9392410173597093,
      "grad_norm": 112.08940887451172,
      "learning_rate": 1.9300910093969504e-07,
      "loss": 4.4434,
      "step": 23265
    },
    {
      "epoch": 0.939281388776746,
      "grad_norm": 86.74578857421875,
      "learning_rate": 1.9275352256515223e-07,
      "loss": 6.938,
      "step": 23266
    },
    {
      "epoch": 0.9393217601937828,
      "grad_norm": 103.41246795654297,
      "learning_rate": 1.9249811187162294e-07,
      "loss": 2.2541,
      "step": 23267
    },
    {
      "epoch": 0.9393621316108195,
      "grad_norm": 154.43862915039062,
      "learning_rate": 1.9224286886347365e-07,
      "loss": 4.809,
      "step": 23268
    },
    {
      "epoch": 0.9394025030278563,
      "grad_norm": 82.76249694824219,
      "learning_rate": 1.919877935450687e-07,
      "loss": 3.6335,
      "step": 23269
    },
    {
      "epoch": 0.939442874444893,
      "grad_norm": 158.27487182617188,
      "learning_rate": 1.9173288592076899e-07,
      "loss": 5.7622,
      "step": 23270
    },
    {
      "epoch": 0.9394832458619298,
      "grad_norm": 245.56019592285156,
      "learning_rate": 1.914781459949333e-07,
      "loss": 4.7347,
      "step": 23271
    },
    {
      "epoch": 0.9395236172789665,
      "grad_norm": 110.74337005615234,
      "learning_rate": 1.9122357377191593e-07,
      "loss": 4.8108,
      "step": 23272
    },
    {
      "epoch": 0.9395639886960032,
      "grad_norm": 89.59851837158203,
      "learning_rate": 1.909691692560711e-07,
      "loss": 5.3765,
      "step": 23273
    },
    {
      "epoch": 0.9396043601130399,
      "grad_norm": 127.70529174804688,
      "learning_rate": 1.9071493245174765e-07,
      "loss": 4.6855,
      "step": 23274
    },
    {
      "epoch": 0.9396447315300767,
      "grad_norm": 73.50939178466797,
      "learning_rate": 1.9046086336329207e-07,
      "loss": 5.7838,
      "step": 23275
    },
    {
      "epoch": 0.9396851029471135,
      "grad_norm": 87.8728256225586,
      "learning_rate": 1.9020696199504861e-07,
      "loss": 4.9219,
      "step": 23276
    },
    {
      "epoch": 0.9397254743641502,
      "grad_norm": 93.79116821289062,
      "learning_rate": 1.8995322835135942e-07,
      "loss": 5.3146,
      "step": 23277
    },
    {
      "epoch": 0.9397658457811869,
      "grad_norm": 115.95452880859375,
      "learning_rate": 1.8969966243656213e-07,
      "loss": 6.0478,
      "step": 23278
    },
    {
      "epoch": 0.9398062171982237,
      "grad_norm": 155.41761779785156,
      "learning_rate": 1.8944626425499103e-07,
      "loss": 5.2934,
      "step": 23279
    },
    {
      "epoch": 0.9398465886152604,
      "grad_norm": 170.03016662597656,
      "learning_rate": 1.8919303381097931e-07,
      "loss": 4.577,
      "step": 23280
    },
    {
      "epoch": 0.9398869600322971,
      "grad_norm": 84.31615447998047,
      "learning_rate": 1.8893997110885797e-07,
      "loss": 5.0684,
      "step": 23281
    },
    {
      "epoch": 0.9399273314493338,
      "grad_norm": 103.0772705078125,
      "learning_rate": 1.8868707615295246e-07,
      "loss": 5.6969,
      "step": 23282
    },
    {
      "epoch": 0.9399677028663707,
      "grad_norm": 104.4600601196289,
      "learning_rate": 1.8843434894758595e-07,
      "loss": 5.8088,
      "step": 23283
    },
    {
      "epoch": 0.9400080742834074,
      "grad_norm": 89.87240600585938,
      "learning_rate": 1.8818178949708056e-07,
      "loss": 6.0133,
      "step": 23284
    },
    {
      "epoch": 0.9400484457004441,
      "grad_norm": 128.19178771972656,
      "learning_rate": 1.8792939780575502e-07,
      "loss": 4.8841,
      "step": 23285
    },
    {
      "epoch": 0.9400888171174808,
      "grad_norm": 196.0076141357422,
      "learning_rate": 1.876771738779226e-07,
      "loss": 5.0573,
      "step": 23286
    },
    {
      "epoch": 0.9401291885345175,
      "grad_norm": 92.15154266357422,
      "learning_rate": 1.8742511771789872e-07,
      "loss": 4.9482,
      "step": 23287
    },
    {
      "epoch": 0.9401695599515543,
      "grad_norm": 82.59415435791016,
      "learning_rate": 1.8717322932998994e-07,
      "loss": 3.8485,
      "step": 23288
    },
    {
      "epoch": 0.940209931368591,
      "grad_norm": 110.14556884765625,
      "learning_rate": 1.8692150871850613e-07,
      "loss": 4.9242,
      "step": 23289
    },
    {
      "epoch": 0.9402503027856278,
      "grad_norm": 95.38063049316406,
      "learning_rate": 1.8666995588774827e-07,
      "loss": 2.7465,
      "step": 23290
    },
    {
      "epoch": 0.9402906742026645,
      "grad_norm": 143.9506072998047,
      "learning_rate": 1.8641857084201743e-07,
      "loss": 6.0763,
      "step": 23291
    },
    {
      "epoch": 0.9403310456197013,
      "grad_norm": 86.33056640625,
      "learning_rate": 1.8616735358561345e-07,
      "loss": 4.468,
      "step": 23292
    },
    {
      "epoch": 0.940371417036738,
      "grad_norm": 62.18556594848633,
      "learning_rate": 1.859163041228318e-07,
      "loss": 4.9724,
      "step": 23293
    },
    {
      "epoch": 0.9404117884537747,
      "grad_norm": 112.0438461303711,
      "learning_rate": 1.856654224579635e-07,
      "loss": 5.216,
      "step": 23294
    },
    {
      "epoch": 0.9404521598708114,
      "grad_norm": 112.73745727539062,
      "learning_rate": 1.8541470859529732e-07,
      "loss": 4.2651,
      "step": 23295
    },
    {
      "epoch": 0.9404925312878482,
      "grad_norm": 86.55382537841797,
      "learning_rate": 1.8516416253912094e-07,
      "loss": 4.4277,
      "step": 23296
    },
    {
      "epoch": 0.940532902704885,
      "grad_norm": 102.16466522216797,
      "learning_rate": 1.8491378429371987e-07,
      "loss": 3.5657,
      "step": 23297
    },
    {
      "epoch": 0.9405732741219217,
      "grad_norm": 90.49309539794922,
      "learning_rate": 1.8466357386337174e-07,
      "loss": 4.906,
      "step": 23298
    },
    {
      "epoch": 0.9406136455389584,
      "grad_norm": 102.12326049804688,
      "learning_rate": 1.844135312523554e-07,
      "loss": 4.6042,
      "step": 23299
    },
    {
      "epoch": 0.9406540169559952,
      "grad_norm": 102.900634765625,
      "learning_rate": 1.8416365646494737e-07,
      "loss": 5.5186,
      "step": 23300
    },
    {
      "epoch": 0.9406943883730319,
      "grad_norm": 74.47734832763672,
      "learning_rate": 1.8391394950541986e-07,
      "loss": 3.6133,
      "step": 23301
    },
    {
      "epoch": 0.9407347597900686,
      "grad_norm": 77.0361557006836,
      "learning_rate": 1.836644103780416e-07,
      "loss": 6.1573,
      "step": 23302
    },
    {
      "epoch": 0.9407751312071053,
      "grad_norm": 114.20268249511719,
      "learning_rate": 1.8341503908707815e-07,
      "loss": 4.6138,
      "step": 23303
    },
    {
      "epoch": 0.9408155026241422,
      "grad_norm": 108.26156616210938,
      "learning_rate": 1.831658356367938e-07,
      "loss": 4.1078,
      "step": 23304
    },
    {
      "epoch": 0.9408558740411789,
      "grad_norm": 124.67711639404297,
      "learning_rate": 1.8291680003145074e-07,
      "loss": 5.9029,
      "step": 23305
    },
    {
      "epoch": 0.9408962454582156,
      "grad_norm": 107.36396026611328,
      "learning_rate": 1.8266793227530556e-07,
      "loss": 4.1852,
      "step": 23306
    },
    {
      "epoch": 0.9409366168752523,
      "grad_norm": 113.54024505615234,
      "learning_rate": 1.8241923237261372e-07,
      "loss": 4.5718,
      "step": 23307
    },
    {
      "epoch": 0.940976988292289,
      "grad_norm": 101.6369400024414,
      "learning_rate": 1.821707003276263e-07,
      "loss": 5.0469,
      "step": 23308
    },
    {
      "epoch": 0.9410173597093258,
      "grad_norm": 87.03919982910156,
      "learning_rate": 1.8192233614459541e-07,
      "loss": 4.6304,
      "step": 23309
    },
    {
      "epoch": 0.9410577311263625,
      "grad_norm": 72.80662536621094,
      "learning_rate": 1.816741398277655e-07,
      "loss": 3.7618,
      "step": 23310
    },
    {
      "epoch": 0.9410981025433993,
      "grad_norm": 89.16940307617188,
      "learning_rate": 1.8142611138137865e-07,
      "loss": 3.7454,
      "step": 23311
    },
    {
      "epoch": 0.941138473960436,
      "grad_norm": 122.65840911865234,
      "learning_rate": 1.811782508096782e-07,
      "loss": 5.4585,
      "step": 23312
    },
    {
      "epoch": 0.9411788453774728,
      "grad_norm": 72.10539245605469,
      "learning_rate": 1.8093055811690185e-07,
      "loss": 4.4668,
      "step": 23313
    },
    {
      "epoch": 0.9412192167945095,
      "grad_norm": 124.11565399169922,
      "learning_rate": 1.8068303330728398e-07,
      "loss": 4.1585,
      "step": 23314
    },
    {
      "epoch": 0.9412595882115462,
      "grad_norm": 125.09503936767578,
      "learning_rate": 1.8043567638505454e-07,
      "loss": 5.189,
      "step": 23315
    },
    {
      "epoch": 0.9412999596285829,
      "grad_norm": 84.50464630126953,
      "learning_rate": 1.8018848735444572e-07,
      "loss": 4.1426,
      "step": 23316
    },
    {
      "epoch": 0.9413403310456196,
      "grad_norm": 139.3563232421875,
      "learning_rate": 1.7994146621968412e-07,
      "loss": 4.0763,
      "step": 23317
    },
    {
      "epoch": 0.9413807024626565,
      "grad_norm": 95.0840072631836,
      "learning_rate": 1.796946129849908e-07,
      "loss": 6.0101,
      "step": 23318
    },
    {
      "epoch": 0.9414210738796932,
      "grad_norm": 100.51185607910156,
      "learning_rate": 1.7944792765458686e-07,
      "loss": 5.3507,
      "step": 23319
    },
    {
      "epoch": 0.9414614452967299,
      "grad_norm": 95.34344482421875,
      "learning_rate": 1.7920141023269112e-07,
      "loss": 4.7069,
      "step": 23320
    },
    {
      "epoch": 0.9415018167137666,
      "grad_norm": 79.28133392333984,
      "learning_rate": 1.789550607235191e-07,
      "loss": 4.8709,
      "step": 23321
    },
    {
      "epoch": 0.9415421881308034,
      "grad_norm": 76.73028564453125,
      "learning_rate": 1.7870887913128076e-07,
      "loss": 2.6217,
      "step": 23322
    },
    {
      "epoch": 0.9415825595478401,
      "grad_norm": 122.11972045898438,
      "learning_rate": 1.784628654601861e-07,
      "loss": 4.7374,
      "step": 23323
    },
    {
      "epoch": 0.9416229309648768,
      "grad_norm": 88.90525817871094,
      "learning_rate": 1.7821701971444172e-07,
      "loss": 6.0193,
      "step": 23324
    },
    {
      "epoch": 0.9416633023819136,
      "grad_norm": 171.28396606445312,
      "learning_rate": 1.7797134189825093e-07,
      "loss": 3.2115,
      "step": 23325
    },
    {
      "epoch": 0.9417036737989504,
      "grad_norm": 95.19829559326172,
      "learning_rate": 1.7772583201581372e-07,
      "loss": 3.736,
      "step": 23326
    },
    {
      "epoch": 0.9417440452159871,
      "grad_norm": 82.90232849121094,
      "learning_rate": 1.7748049007132894e-07,
      "loss": 4.6351,
      "step": 23327
    },
    {
      "epoch": 0.9417844166330238,
      "grad_norm": 70.42276763916016,
      "learning_rate": 1.7723531606898993e-07,
      "loss": 4.6609,
      "step": 23328
    },
    {
      "epoch": 0.9418247880500605,
      "grad_norm": 109.52259826660156,
      "learning_rate": 1.7699031001298884e-07,
      "loss": 5.5121,
      "step": 23329
    },
    {
      "epoch": 0.9418651594670973,
      "grad_norm": 116.74906921386719,
      "learning_rate": 1.767454719075168e-07,
      "loss": 5.4205,
      "step": 23330
    },
    {
      "epoch": 0.941905530884134,
      "grad_norm": 109.01258850097656,
      "learning_rate": 1.765008017567571e-07,
      "loss": 4.1923,
      "step": 23331
    },
    {
      "epoch": 0.9419459023011708,
      "grad_norm": 102.82200622558594,
      "learning_rate": 1.762562995648931e-07,
      "loss": 5.0565,
      "step": 23332
    },
    {
      "epoch": 0.9419862737182075,
      "grad_norm": 99.80052185058594,
      "learning_rate": 1.7601196533610808e-07,
      "loss": 5.2032,
      "step": 23333
    },
    {
      "epoch": 0.9420266451352443,
      "grad_norm": 108.76042175292969,
      "learning_rate": 1.7576779907457763e-07,
      "loss": 6.0082,
      "step": 23334
    },
    {
      "epoch": 0.942067016552281,
      "grad_norm": 107.40282440185547,
      "learning_rate": 1.7552380078447506e-07,
      "loss": 5.96,
      "step": 23335
    },
    {
      "epoch": 0.9421073879693177,
      "grad_norm": 74.6143798828125,
      "learning_rate": 1.752799704699748e-07,
      "loss": 3.4796,
      "step": 23336
    },
    {
      "epoch": 0.9421477593863544,
      "grad_norm": 66.7068099975586,
      "learning_rate": 1.7503630813524462e-07,
      "loss": 4.2559,
      "step": 23337
    },
    {
      "epoch": 0.9421881308033911,
      "grad_norm": 77.75861358642578,
      "learning_rate": 1.7479281378445122e-07,
      "loss": 4.6728,
      "step": 23338
    },
    {
      "epoch": 0.942228502220428,
      "grad_norm": 73.75926208496094,
      "learning_rate": 1.745494874217557e-07,
      "loss": 4.5414,
      "step": 23339
    },
    {
      "epoch": 0.9422688736374647,
      "grad_norm": 76.2998275756836,
      "learning_rate": 1.7430632905132027e-07,
      "loss": 5.6173,
      "step": 23340
    },
    {
      "epoch": 0.9423092450545014,
      "grad_norm": 241.8179473876953,
      "learning_rate": 1.7406333867730274e-07,
      "loss": 4.4308,
      "step": 23341
    },
    {
      "epoch": 0.9423496164715381,
      "grad_norm": 159.8427734375,
      "learning_rate": 1.7382051630385755e-07,
      "loss": 5.8131,
      "step": 23342
    },
    {
      "epoch": 0.9423899878885749,
      "grad_norm": 89.14810943603516,
      "learning_rate": 1.735778619351336e-07,
      "loss": 4.4952,
      "step": 23343
    },
    {
      "epoch": 0.9424303593056116,
      "grad_norm": 73.11318969726562,
      "learning_rate": 1.7333537557528312e-07,
      "loss": 4.9775,
      "step": 23344
    },
    {
      "epoch": 0.9424707307226484,
      "grad_norm": 126.11132049560547,
      "learning_rate": 1.730930572284506e-07,
      "loss": 3.7118,
      "step": 23345
    },
    {
      "epoch": 0.9425111021396851,
      "grad_norm": 83.30760192871094,
      "learning_rate": 1.7285090689877937e-07,
      "loss": 4.9225,
      "step": 23346
    },
    {
      "epoch": 0.9425514735567219,
      "grad_norm": 110.67520141601562,
      "learning_rate": 1.7260892459040946e-07,
      "loss": 4.6307,
      "step": 23347
    },
    {
      "epoch": 0.9425918449737586,
      "grad_norm": 90.40443420410156,
      "learning_rate": 1.7236711030747865e-07,
      "loss": 4.7031,
      "step": 23348
    },
    {
      "epoch": 0.9426322163907953,
      "grad_norm": 123.15250396728516,
      "learning_rate": 1.7212546405412033e-07,
      "loss": 5.6863,
      "step": 23349
    },
    {
      "epoch": 0.942672587807832,
      "grad_norm": 91.89221954345703,
      "learning_rate": 1.7188398583446785e-07,
      "loss": 4.6942,
      "step": 23350
    },
    {
      "epoch": 0.9427129592248688,
      "grad_norm": 87.5309829711914,
      "learning_rate": 1.7164267565264903e-07,
      "loss": 4.4545,
      "step": 23351
    },
    {
      "epoch": 0.9427533306419056,
      "grad_norm": 101.76374816894531,
      "learning_rate": 1.714015335127883e-07,
      "loss": 6.3288,
      "step": 23352
    },
    {
      "epoch": 0.9427937020589423,
      "grad_norm": 113.65143585205078,
      "learning_rate": 1.7116055941901023e-07,
      "loss": 4.2916,
      "step": 23353
    },
    {
      "epoch": 0.942834073475979,
      "grad_norm": 72.75715637207031,
      "learning_rate": 1.7091975337543587e-07,
      "loss": 3.949,
      "step": 23354
    },
    {
      "epoch": 0.9428744448930158,
      "grad_norm": 109.62086486816406,
      "learning_rate": 1.7067911538618086e-07,
      "loss": 5.4548,
      "step": 23355
    },
    {
      "epoch": 0.9429148163100525,
      "grad_norm": 130.8909912109375,
      "learning_rate": 1.704386454553586e-07,
      "loss": 4.6925,
      "step": 23356
    },
    {
      "epoch": 0.9429551877270892,
      "grad_norm": 88.60652923583984,
      "learning_rate": 1.7019834358708243e-07,
      "loss": 4.3418,
      "step": 23357
    },
    {
      "epoch": 0.9429955591441259,
      "grad_norm": 164.54705810546875,
      "learning_rate": 1.6995820978546128e-07,
      "loss": 5.4402,
      "step": 23358
    },
    {
      "epoch": 0.9430359305611627,
      "grad_norm": 79.77801513671875,
      "learning_rate": 1.6971824405459858e-07,
      "loss": 5.0903,
      "step": 23359
    },
    {
      "epoch": 0.9430763019781995,
      "grad_norm": 77.1105728149414,
      "learning_rate": 1.6947844639859878e-07,
      "loss": 4.3923,
      "step": 23360
    },
    {
      "epoch": 0.9431166733952362,
      "grad_norm": 111.66971588134766,
      "learning_rate": 1.6923881682156197e-07,
      "loss": 5.2789,
      "step": 23361
    },
    {
      "epoch": 0.9431570448122729,
      "grad_norm": 174.2747802734375,
      "learning_rate": 1.6899935532758592e-07,
      "loss": 4.6939,
      "step": 23362
    },
    {
      "epoch": 0.9431974162293096,
      "grad_norm": 84.56130981445312,
      "learning_rate": 1.68760061920763e-07,
      "loss": 4.5409,
      "step": 23363
    },
    {
      "epoch": 0.9432377876463464,
      "grad_norm": 95.07917785644531,
      "learning_rate": 1.6852093660518433e-07,
      "loss": 5.7605,
      "step": 23364
    },
    {
      "epoch": 0.9432781590633831,
      "grad_norm": 111.6653823852539,
      "learning_rate": 1.6828197938494106e-07,
      "loss": 5.5108,
      "step": 23365
    },
    {
      "epoch": 0.9433185304804199,
      "grad_norm": 65.8758773803711,
      "learning_rate": 1.6804319026411664e-07,
      "loss": 3.1022,
      "step": 23366
    },
    {
      "epoch": 0.9433589018974566,
      "grad_norm": 89.43379974365234,
      "learning_rate": 1.6780456924679333e-07,
      "loss": 5.4386,
      "step": 23367
    },
    {
      "epoch": 0.9433992733144934,
      "grad_norm": 128.39649963378906,
      "learning_rate": 1.6756611633705345e-07,
      "loss": 4.8246,
      "step": 23368
    },
    {
      "epoch": 0.9434396447315301,
      "grad_norm": 94.80696105957031,
      "learning_rate": 1.6732783153897148e-07,
      "loss": 4.9101,
      "step": 23369
    },
    {
      "epoch": 0.9434800161485668,
      "grad_norm": 168.03582763671875,
      "learning_rate": 1.6708971485662418e-07,
      "loss": 5.6662,
      "step": 23370
    },
    {
      "epoch": 0.9435203875656035,
      "grad_norm": 166.53868103027344,
      "learning_rate": 1.668517662940805e-07,
      "loss": 5.4528,
      "step": 23371
    },
    {
      "epoch": 0.9435607589826402,
      "grad_norm": 136.9049835205078,
      "learning_rate": 1.6661398585540945e-07,
      "loss": 6.9249,
      "step": 23372
    },
    {
      "epoch": 0.9436011303996771,
      "grad_norm": 109.058837890625,
      "learning_rate": 1.663763735446766e-07,
      "loss": 3.1511,
      "step": 23373
    },
    {
      "epoch": 0.9436415018167138,
      "grad_norm": 94.4574966430664,
      "learning_rate": 1.6613892936594545e-07,
      "loss": 4.6322,
      "step": 23374
    },
    {
      "epoch": 0.9436818732337505,
      "grad_norm": 81.69310760498047,
      "learning_rate": 1.659016533232738e-07,
      "loss": 5.0585,
      "step": 23375
    },
    {
      "epoch": 0.9437222446507872,
      "grad_norm": 95.07978820800781,
      "learning_rate": 1.6566454542071953e-07,
      "loss": 5.28,
      "step": 23376
    },
    {
      "epoch": 0.943762616067824,
      "grad_norm": 98.7281494140625,
      "learning_rate": 1.654276056623372e-07,
      "loss": 5.4827,
      "step": 23377
    },
    {
      "epoch": 0.9438029874848607,
      "grad_norm": 87.62051391601562,
      "learning_rate": 1.6519083405217794e-07,
      "loss": 5.1435,
      "step": 23378
    },
    {
      "epoch": 0.9438433589018974,
      "grad_norm": 114.6735610961914,
      "learning_rate": 1.6495423059428972e-07,
      "loss": 4.8233,
      "step": 23379
    },
    {
      "epoch": 0.9438837303189342,
      "grad_norm": 84.68445587158203,
      "learning_rate": 1.6471779529271593e-07,
      "loss": 4.8442,
      "step": 23380
    },
    {
      "epoch": 0.943924101735971,
      "grad_norm": 87.97686004638672,
      "learning_rate": 1.644815281515022e-07,
      "loss": 4.9789,
      "step": 23381
    },
    {
      "epoch": 0.9439644731530077,
      "grad_norm": 75.93741607666016,
      "learning_rate": 1.6424542917468643e-07,
      "loss": 4.1991,
      "step": 23382
    },
    {
      "epoch": 0.9440048445700444,
      "grad_norm": 100.74280548095703,
      "learning_rate": 1.640094983663054e-07,
      "loss": 4.7623,
      "step": 23383
    },
    {
      "epoch": 0.9440452159870811,
      "grad_norm": 78.48239135742188,
      "learning_rate": 1.637737357303937e-07,
      "loss": 4.7714,
      "step": 23384
    },
    {
      "epoch": 0.9440855874041179,
      "grad_norm": 72.89205932617188,
      "learning_rate": 1.6353814127098132e-07,
      "loss": 2.6392,
      "step": 23385
    },
    {
      "epoch": 0.9441259588211546,
      "grad_norm": 79.19983673095703,
      "learning_rate": 1.6330271499209738e-07,
      "loss": 3.5922,
      "step": 23386
    },
    {
      "epoch": 0.9441663302381914,
      "grad_norm": 102.90919494628906,
      "learning_rate": 1.6306745689776638e-07,
      "loss": 4.5195,
      "step": 23387
    },
    {
      "epoch": 0.9442067016552281,
      "grad_norm": 77.8619155883789,
      "learning_rate": 1.6283236699201067e-07,
      "loss": 6.0712,
      "step": 23388
    },
    {
      "epoch": 0.9442470730722649,
      "grad_norm": 106.5718002319336,
      "learning_rate": 1.6259744527885035e-07,
      "loss": 5.2164,
      "step": 23389
    },
    {
      "epoch": 0.9442874444893016,
      "grad_norm": 98.66714477539062,
      "learning_rate": 1.6236269176230223e-07,
      "loss": 5.1885,
      "step": 23390
    },
    {
      "epoch": 0.9443278159063383,
      "grad_norm": 99.38749694824219,
      "learning_rate": 1.6212810644637977e-07,
      "loss": 3.3266,
      "step": 23391
    },
    {
      "epoch": 0.944368187323375,
      "grad_norm": 80.12418365478516,
      "learning_rate": 1.6189368933509197e-07,
      "loss": 4.5442,
      "step": 23392
    },
    {
      "epoch": 0.9444085587404117,
      "grad_norm": 93.23056030273438,
      "learning_rate": 1.6165944043245006e-07,
      "loss": 4.9071,
      "step": 23393
    },
    {
      "epoch": 0.9444489301574486,
      "grad_norm": 123.34768676757812,
      "learning_rate": 1.6142535974245755e-07,
      "loss": 3.6106,
      "step": 23394
    },
    {
      "epoch": 0.9444893015744853,
      "grad_norm": 94.12653350830078,
      "learning_rate": 1.611914472691156e-07,
      "loss": 4.8658,
      "step": 23395
    },
    {
      "epoch": 0.944529672991522,
      "grad_norm": 100.3525161743164,
      "learning_rate": 1.6095770301642554e-07,
      "loss": 5.7852,
      "step": 23396
    },
    {
      "epoch": 0.9445700444085587,
      "grad_norm": 115.00255584716797,
      "learning_rate": 1.6072412698838193e-07,
      "loss": 5.2397,
      "step": 23397
    },
    {
      "epoch": 0.9446104158255955,
      "grad_norm": 65.71975708007812,
      "learning_rate": 1.604907191889804e-07,
      "loss": 2.743,
      "step": 23398
    },
    {
      "epoch": 0.9446507872426322,
      "grad_norm": 118.46603393554688,
      "learning_rate": 1.6025747962221004e-07,
      "loss": 4.5885,
      "step": 23399
    },
    {
      "epoch": 0.9446911586596689,
      "grad_norm": 98.0482406616211,
      "learning_rate": 1.600244082920599e-07,
      "loss": 4.3598,
      "step": 23400
    },
    {
      "epoch": 0.9447315300767057,
      "grad_norm": 85.9462890625,
      "learning_rate": 1.597915052025134e-07,
      "loss": 4.5184,
      "step": 23401
    },
    {
      "epoch": 0.9447719014937425,
      "grad_norm": 85.462890625,
      "learning_rate": 1.5955877035755518e-07,
      "loss": 4.0627,
      "step": 23402
    },
    {
      "epoch": 0.9448122729107792,
      "grad_norm": 73.86613464355469,
      "learning_rate": 1.5932620376116205e-07,
      "loss": 4.7345,
      "step": 23403
    },
    {
      "epoch": 0.9448526443278159,
      "grad_norm": 121.13035583496094,
      "learning_rate": 1.5909380541731078e-07,
      "loss": 2.799,
      "step": 23404
    },
    {
      "epoch": 0.9448930157448526,
      "grad_norm": 163.9878387451172,
      "learning_rate": 1.5886157532997603e-07,
      "loss": 4.2066,
      "step": 23405
    },
    {
      "epoch": 0.9449333871618893,
      "grad_norm": 103.30545043945312,
      "learning_rate": 1.5862951350312684e-07,
      "loss": 3.6273,
      "step": 23406
    },
    {
      "epoch": 0.9449737585789261,
      "grad_norm": 93.63838195800781,
      "learning_rate": 1.5839761994073222e-07,
      "loss": 3.576,
      "step": 23407
    },
    {
      "epoch": 0.9450141299959629,
      "grad_norm": 82.20018768310547,
      "learning_rate": 1.581658946467557e-07,
      "loss": 5.1676,
      "step": 23408
    },
    {
      "epoch": 0.9450545014129996,
      "grad_norm": 102.23738861083984,
      "learning_rate": 1.5793433762516076e-07,
      "loss": 3.9824,
      "step": 23409
    },
    {
      "epoch": 0.9450948728300363,
      "grad_norm": 96.38211059570312,
      "learning_rate": 1.5770294887990644e-07,
      "loss": 4.4358,
      "step": 23410
    },
    {
      "epoch": 0.9451352442470731,
      "grad_norm": 236.89561462402344,
      "learning_rate": 1.574717284149474e-07,
      "loss": 2.8082,
      "step": 23411
    },
    {
      "epoch": 0.9451756156641098,
      "grad_norm": 81.86417388916016,
      "learning_rate": 1.5724067623423821e-07,
      "loss": 4.8429,
      "step": 23412
    },
    {
      "epoch": 0.9452159870811465,
      "grad_norm": 133.34242248535156,
      "learning_rate": 1.5700979234172797e-07,
      "loss": 5.3878,
      "step": 23413
    },
    {
      "epoch": 0.9452563584981832,
      "grad_norm": 78.19644165039062,
      "learning_rate": 1.5677907674136682e-07,
      "loss": 4.5616,
      "step": 23414
    },
    {
      "epoch": 0.9452967299152201,
      "grad_norm": 98.97969055175781,
      "learning_rate": 1.5654852943709718e-07,
      "loss": 4.744,
      "step": 23415
    },
    {
      "epoch": 0.9453371013322568,
      "grad_norm": 90.61376190185547,
      "learning_rate": 1.5631815043286147e-07,
      "loss": 5.5408,
      "step": 23416
    },
    {
      "epoch": 0.9453774727492935,
      "grad_norm": 160.63671875,
      "learning_rate": 1.5608793973259872e-07,
      "loss": 4.8678,
      "step": 23417
    },
    {
      "epoch": 0.9454178441663302,
      "grad_norm": 78.87458038330078,
      "learning_rate": 1.5585789734024582e-07,
      "loss": 4.7584,
      "step": 23418
    },
    {
      "epoch": 0.945458215583367,
      "grad_norm": 96.5418701171875,
      "learning_rate": 1.5562802325973402e-07,
      "loss": 5.496,
      "step": 23419
    },
    {
      "epoch": 0.9454985870004037,
      "grad_norm": 85.4947280883789,
      "learning_rate": 1.5539831749499468e-07,
      "loss": 3.7475,
      "step": 23420
    },
    {
      "epoch": 0.9455389584174404,
      "grad_norm": 138.25390625,
      "learning_rate": 1.5516878004995573e-07,
      "loss": 4.0266,
      "step": 23421
    },
    {
      "epoch": 0.9455793298344772,
      "grad_norm": 88.92536926269531,
      "learning_rate": 1.549394109285418e-07,
      "loss": 5.2661,
      "step": 23422
    },
    {
      "epoch": 0.945619701251514,
      "grad_norm": 90.41559600830078,
      "learning_rate": 1.5471021013467314e-07,
      "loss": 4.2287,
      "step": 23423
    },
    {
      "epoch": 0.9456600726685507,
      "grad_norm": 172.6840057373047,
      "learning_rate": 1.54481177672271e-07,
      "loss": 5.5964,
      "step": 23424
    },
    {
      "epoch": 0.9457004440855874,
      "grad_norm": 112.3782730102539,
      "learning_rate": 1.5425231354524783e-07,
      "loss": 5.5084,
      "step": 23425
    },
    {
      "epoch": 0.9457408155026241,
      "grad_norm": 81.51949310302734,
      "learning_rate": 1.540236177575194e-07,
      "loss": 5.4941,
      "step": 23426
    },
    {
      "epoch": 0.9457811869196608,
      "grad_norm": 83.12091064453125,
      "learning_rate": 1.537950903129959e-07,
      "loss": 4.8389,
      "step": 23427
    },
    {
      "epoch": 0.9458215583366976,
      "grad_norm": 88.28627014160156,
      "learning_rate": 1.5356673121558309e-07,
      "loss": 5.5646,
      "step": 23428
    },
    {
      "epoch": 0.9458619297537344,
      "grad_norm": 105.5340347290039,
      "learning_rate": 1.5333854046918562e-07,
      "loss": 4.3578,
      "step": 23429
    },
    {
      "epoch": 0.9459023011707711,
      "grad_norm": 71.97372436523438,
      "learning_rate": 1.5311051807770704e-07,
      "loss": 4.8318,
      "step": 23430
    },
    {
      "epoch": 0.9459426725878078,
      "grad_norm": 113.43230438232422,
      "learning_rate": 1.5288266404504315e-07,
      "loss": 4.0807,
      "step": 23431
    },
    {
      "epoch": 0.9459830440048446,
      "grad_norm": 103.78352355957031,
      "learning_rate": 1.526549783750908e-07,
      "loss": 5.1754,
      "step": 23432
    },
    {
      "epoch": 0.9460234154218813,
      "grad_norm": 97.61885833740234,
      "learning_rate": 1.5242746107174356e-07,
      "loss": 3.7702,
      "step": 23433
    },
    {
      "epoch": 0.946063786838918,
      "grad_norm": 107.96746826171875,
      "learning_rate": 1.5220011213889163e-07,
      "loss": 4.0865,
      "step": 23434
    },
    {
      "epoch": 0.9461041582559547,
      "grad_norm": 74.63192749023438,
      "learning_rate": 1.5197293158042083e-07,
      "loss": 5.9382,
      "step": 23435
    },
    {
      "epoch": 0.9461445296729916,
      "grad_norm": 88.15798950195312,
      "learning_rate": 1.5174591940021689e-07,
      "loss": 3.862,
      "step": 23436
    },
    {
      "epoch": 0.9461849010900283,
      "grad_norm": 87.20934295654297,
      "learning_rate": 1.5151907560215895e-07,
      "loss": 2.6165,
      "step": 23437
    },
    {
      "epoch": 0.946225272507065,
      "grad_norm": 92.09217071533203,
      "learning_rate": 1.5129240019012837e-07,
      "loss": 4.7361,
      "step": 23438
    },
    {
      "epoch": 0.9462656439241017,
      "grad_norm": 442.3162536621094,
      "learning_rate": 1.5106589316799868e-07,
      "loss": 3.9926,
      "step": 23439
    },
    {
      "epoch": 0.9463060153411385,
      "grad_norm": 96.27411651611328,
      "learning_rate": 1.508395545396435e-07,
      "loss": 4.5333,
      "step": 23440
    },
    {
      "epoch": 0.9463463867581752,
      "grad_norm": 99.81668853759766,
      "learning_rate": 1.506133843089319e-07,
      "loss": 5.6231,
      "step": 23441
    },
    {
      "epoch": 0.9463867581752119,
      "grad_norm": 105.8818130493164,
      "learning_rate": 1.5038738247973195e-07,
      "loss": 4.6721,
      "step": 23442
    },
    {
      "epoch": 0.9464271295922487,
      "grad_norm": 107.64179992675781,
      "learning_rate": 1.501615490559072e-07,
      "loss": 4.9141,
      "step": 23443
    },
    {
      "epoch": 0.9464675010092855,
      "grad_norm": 105.44293975830078,
      "learning_rate": 1.4993588404131898e-07,
      "loss": 6.8598,
      "step": 23444
    },
    {
      "epoch": 0.9465078724263222,
      "grad_norm": 100.15155792236328,
      "learning_rate": 1.4971038743982536e-07,
      "loss": 5.1743,
      "step": 23445
    },
    {
      "epoch": 0.9465482438433589,
      "grad_norm": 86.37325286865234,
      "learning_rate": 1.4948505925528323e-07,
      "loss": 5.5426,
      "step": 23446
    },
    {
      "epoch": 0.9465886152603956,
      "grad_norm": 114.45409393310547,
      "learning_rate": 1.4925989949154286e-07,
      "loss": 5.9812,
      "step": 23447
    },
    {
      "epoch": 0.9466289866774323,
      "grad_norm": 88.54264068603516,
      "learning_rate": 1.4903490815245558e-07,
      "loss": 3.9644,
      "step": 23448
    },
    {
      "epoch": 0.9466693580944692,
      "grad_norm": 101.32720947265625,
      "learning_rate": 1.4881008524186725e-07,
      "loss": 4.115,
      "step": 23449
    },
    {
      "epoch": 0.9467097295115059,
      "grad_norm": 69.22126007080078,
      "learning_rate": 1.4858543076362252e-07,
      "loss": 5.0087,
      "step": 23450
    },
    {
      "epoch": 0.9467501009285426,
      "grad_norm": 90.60977935791016,
      "learning_rate": 1.4836094472156171e-07,
      "loss": 4.6269,
      "step": 23451
    },
    {
      "epoch": 0.9467904723455793,
      "grad_norm": 87.90603637695312,
      "learning_rate": 1.4813662711952391e-07,
      "loss": 5.0685,
      "step": 23452
    },
    {
      "epoch": 0.9468308437626161,
      "grad_norm": 142.07960510253906,
      "learning_rate": 1.479124779613439e-07,
      "loss": 5.4978,
      "step": 23453
    },
    {
      "epoch": 0.9468712151796528,
      "grad_norm": 224.0133819580078,
      "learning_rate": 1.476884972508552e-07,
      "loss": 4.6927,
      "step": 23454
    },
    {
      "epoch": 0.9469115865966895,
      "grad_norm": 77.9225845336914,
      "learning_rate": 1.4746468499188593e-07,
      "loss": 3.9258,
      "step": 23455
    },
    {
      "epoch": 0.9469519580137263,
      "grad_norm": 104.0975570678711,
      "learning_rate": 1.472410411882619e-07,
      "loss": 5.617,
      "step": 23456
    },
    {
      "epoch": 0.9469923294307631,
      "grad_norm": 68.32843017578125,
      "learning_rate": 1.4701756584381e-07,
      "loss": 4.5284,
      "step": 23457
    },
    {
      "epoch": 0.9470327008477998,
      "grad_norm": 668.9830932617188,
      "learning_rate": 1.4679425896234832e-07,
      "loss": 5.2611,
      "step": 23458
    },
    {
      "epoch": 0.9470730722648365,
      "grad_norm": 106.1003646850586,
      "learning_rate": 1.4657112054769606e-07,
      "loss": 5.4328,
      "step": 23459
    },
    {
      "epoch": 0.9471134436818732,
      "grad_norm": 69.20957946777344,
      "learning_rate": 1.4634815060366902e-07,
      "loss": 5.0915,
      "step": 23460
    },
    {
      "epoch": 0.94715381509891,
      "grad_norm": 178.68637084960938,
      "learning_rate": 1.461253491340775e-07,
      "loss": 5.5824,
      "step": 23461
    },
    {
      "epoch": 0.9471941865159467,
      "grad_norm": 75.31133270263672,
      "learning_rate": 1.4590271614273178e-07,
      "loss": 3.1251,
      "step": 23462
    },
    {
      "epoch": 0.9472345579329835,
      "grad_norm": 186.87478637695312,
      "learning_rate": 1.456802516334399e-07,
      "loss": 5.302,
      "step": 23463
    },
    {
      "epoch": 0.9472749293500202,
      "grad_norm": 64.51629638671875,
      "learning_rate": 1.454579556100033e-07,
      "loss": 5.9421,
      "step": 23464
    },
    {
      "epoch": 0.947315300767057,
      "grad_norm": 84.23885345458984,
      "learning_rate": 1.452358280762245e-07,
      "loss": 5.1662,
      "step": 23465
    },
    {
      "epoch": 0.9473556721840937,
      "grad_norm": 103.58866119384766,
      "learning_rate": 1.450138690359004e-07,
      "loss": 4.0412,
      "step": 23466
    },
    {
      "epoch": 0.9473960436011304,
      "grad_norm": 72.62330627441406,
      "learning_rate": 1.4479207849282583e-07,
      "loss": 5.1009,
      "step": 23467
    },
    {
      "epoch": 0.9474364150181671,
      "grad_norm": 114.46047973632812,
      "learning_rate": 1.4457045645079214e-07,
      "loss": 5.2955,
      "step": 23468
    },
    {
      "epoch": 0.9474767864352038,
      "grad_norm": 102.81128692626953,
      "learning_rate": 1.4434900291359078e-07,
      "loss": 5.6098,
      "step": 23469
    },
    {
      "epoch": 0.9475171578522407,
      "grad_norm": 102.22993469238281,
      "learning_rate": 1.4412771788500756e-07,
      "loss": 6.5379,
      "step": 23470
    },
    {
      "epoch": 0.9475575292692774,
      "grad_norm": 131.67262268066406,
      "learning_rate": 1.4390660136882395e-07,
      "loss": 4.9103,
      "step": 23471
    },
    {
      "epoch": 0.9475979006863141,
      "grad_norm": 94.23959350585938,
      "learning_rate": 1.4368565336882135e-07,
      "loss": 6.0488,
      "step": 23472
    },
    {
      "epoch": 0.9476382721033508,
      "grad_norm": 74.98171997070312,
      "learning_rate": 1.4346487388877895e-07,
      "loss": 5.3058,
      "step": 23473
    },
    {
      "epoch": 0.9476786435203876,
      "grad_norm": 96.8865966796875,
      "learning_rate": 1.432442629324715e-07,
      "loss": 4.0448,
      "step": 23474
    },
    {
      "epoch": 0.9477190149374243,
      "grad_norm": 77.41345977783203,
      "learning_rate": 1.4302382050366824e-07,
      "loss": 4.4455,
      "step": 23475
    },
    {
      "epoch": 0.947759386354461,
      "grad_norm": 84.4078369140625,
      "learning_rate": 1.428035466061406e-07,
      "loss": 4.7554,
      "step": 23476
    },
    {
      "epoch": 0.9477997577714978,
      "grad_norm": 93.84336853027344,
      "learning_rate": 1.425834412436533e-07,
      "loss": 4.3992,
      "step": 23477
    },
    {
      "epoch": 0.9478401291885346,
      "grad_norm": 81.26663208007812,
      "learning_rate": 1.4236350441997227e-07,
      "loss": 4.5954,
      "step": 23478
    },
    {
      "epoch": 0.9478805006055713,
      "grad_norm": 78.41304779052734,
      "learning_rate": 1.4214373613885446e-07,
      "loss": 5.6657,
      "step": 23479
    },
    {
      "epoch": 0.947920872022608,
      "grad_norm": 94.88307189941406,
      "learning_rate": 1.4192413640406022e-07,
      "loss": 6.4889,
      "step": 23480
    },
    {
      "epoch": 0.9479612434396447,
      "grad_norm": 89.47784423828125,
      "learning_rate": 1.417047052193421e-07,
      "loss": 6.1312,
      "step": 23481
    },
    {
      "epoch": 0.9480016148566814,
      "grad_norm": 79.62063598632812,
      "learning_rate": 1.4148544258845265e-07,
      "loss": 5.9586,
      "step": 23482
    },
    {
      "epoch": 0.9480419862737182,
      "grad_norm": 72.62651824951172,
      "learning_rate": 1.412663485151411e-07,
      "loss": 4.5763,
      "step": 23483
    },
    {
      "epoch": 0.948082357690755,
      "grad_norm": 72.5088119506836,
      "learning_rate": 1.410474230031522e-07,
      "loss": 4.0869,
      "step": 23484
    },
    {
      "epoch": 0.9481227291077917,
      "grad_norm": 94.45965576171875,
      "learning_rate": 1.408286660562319e-07,
      "loss": 5.5522,
      "step": 23485
    },
    {
      "epoch": 0.9481631005248284,
      "grad_norm": 155.08090209960938,
      "learning_rate": 1.406100776781183e-07,
      "loss": 5.0079,
      "step": 23486
    },
    {
      "epoch": 0.9482034719418652,
      "grad_norm": 117.1147689819336,
      "learning_rate": 1.403916578725484e-07,
      "loss": 4.3872,
      "step": 23487
    },
    {
      "epoch": 0.9482438433589019,
      "grad_norm": 87.63102722167969,
      "learning_rate": 1.4017340664325695e-07,
      "loss": 5.4332,
      "step": 23488
    },
    {
      "epoch": 0.9482842147759386,
      "grad_norm": 148.13963317871094,
      "learning_rate": 1.399553239939766e-07,
      "loss": 4.6675,
      "step": 23489
    },
    {
      "epoch": 0.9483245861929753,
      "grad_norm": 129.1683807373047,
      "learning_rate": 1.3973740992843544e-07,
      "loss": 5.1453,
      "step": 23490
    },
    {
      "epoch": 0.9483649576100122,
      "grad_norm": 112.82264709472656,
      "learning_rate": 1.3951966445035825e-07,
      "loss": 3.4613,
      "step": 23491
    },
    {
      "epoch": 0.9484053290270489,
      "grad_norm": 88.46441650390625,
      "learning_rate": 1.3930208756346986e-07,
      "loss": 5.6601,
      "step": 23492
    },
    {
      "epoch": 0.9484457004440856,
      "grad_norm": 109.57769775390625,
      "learning_rate": 1.3908467927148838e-07,
      "loss": 5.3133,
      "step": 23493
    },
    {
      "epoch": 0.9484860718611223,
      "grad_norm": 105.5500259399414,
      "learning_rate": 1.388674395781342e-07,
      "loss": 5.8667,
      "step": 23494
    },
    {
      "epoch": 0.948526443278159,
      "grad_norm": 204.34942626953125,
      "learning_rate": 1.3865036848711766e-07,
      "loss": 5.5834,
      "step": 23495
    },
    {
      "epoch": 0.9485668146951958,
      "grad_norm": 116.87007141113281,
      "learning_rate": 1.3843346600215136e-07,
      "loss": 5.3726,
      "step": 23496
    },
    {
      "epoch": 0.9486071861122325,
      "grad_norm": 104.86283111572266,
      "learning_rate": 1.3821673212694563e-07,
      "loss": 5.2866,
      "step": 23497
    },
    {
      "epoch": 0.9486475575292693,
      "grad_norm": 124.57530212402344,
      "learning_rate": 1.3800016686520422e-07,
      "loss": 5.7975,
      "step": 23498
    },
    {
      "epoch": 0.948687928946306,
      "grad_norm": 101.03741455078125,
      "learning_rate": 1.3778377022063083e-07,
      "loss": 4.2074,
      "step": 23499
    },
    {
      "epoch": 0.9487283003633428,
      "grad_norm": 99.90209197998047,
      "learning_rate": 1.375675421969247e-07,
      "loss": 5.2132,
      "step": 23500
    },
    {
      "epoch": 0.9487686717803795,
      "grad_norm": 94.3629379272461,
      "learning_rate": 1.37351482797784e-07,
      "loss": 4.9365,
      "step": 23501
    },
    {
      "epoch": 0.9488090431974162,
      "grad_norm": 90.55705261230469,
      "learning_rate": 1.371355920269013e-07,
      "loss": 4.9281,
      "step": 23502
    },
    {
      "epoch": 0.9488494146144529,
      "grad_norm": 102.58924102783203,
      "learning_rate": 1.3691986988796813e-07,
      "loss": 4.9624,
      "step": 23503
    },
    {
      "epoch": 0.9488897860314897,
      "grad_norm": 101.14608001708984,
      "learning_rate": 1.3670431638467375e-07,
      "loss": 5.7046,
      "step": 23504
    },
    {
      "epoch": 0.9489301574485265,
      "grad_norm": 91.25798034667969,
      "learning_rate": 1.3648893152070298e-07,
      "loss": 4.6231,
      "step": 23505
    },
    {
      "epoch": 0.9489705288655632,
      "grad_norm": 85.873291015625,
      "learning_rate": 1.3627371529973843e-07,
      "loss": 3.9821,
      "step": 23506
    },
    {
      "epoch": 0.9490109002825999,
      "grad_norm": 109.21429443359375,
      "learning_rate": 1.360586677254605e-07,
      "loss": 4.8236,
      "step": 23507
    },
    {
      "epoch": 0.9490512716996367,
      "grad_norm": 98.84723663330078,
      "learning_rate": 1.3584378880154514e-07,
      "loss": 5.4968,
      "step": 23508
    },
    {
      "epoch": 0.9490916431166734,
      "grad_norm": 159.9993438720703,
      "learning_rate": 1.3562907853166606e-07,
      "loss": 4.1222,
      "step": 23509
    },
    {
      "epoch": 0.9491320145337101,
      "grad_norm": 101.11576080322266,
      "learning_rate": 1.354145369194959e-07,
      "loss": 6.2479,
      "step": 23510
    },
    {
      "epoch": 0.9491723859507468,
      "grad_norm": 98.96575164794922,
      "learning_rate": 1.352001639687006e-07,
      "loss": 5.4193,
      "step": 23511
    },
    {
      "epoch": 0.9492127573677837,
      "grad_norm": 85.12117767333984,
      "learning_rate": 1.3498595968294615e-07,
      "loss": 5.0917,
      "step": 23512
    },
    {
      "epoch": 0.9492531287848204,
      "grad_norm": 157.14794921875,
      "learning_rate": 1.3477192406589622e-07,
      "loss": 6.2682,
      "step": 23513
    },
    {
      "epoch": 0.9492935002018571,
      "grad_norm": 143.14981079101562,
      "learning_rate": 1.3455805712121018e-07,
      "loss": 4.6867,
      "step": 23514
    },
    {
      "epoch": 0.9493338716188938,
      "grad_norm": 79.78633880615234,
      "learning_rate": 1.3434435885254282e-07,
      "loss": 5.7361,
      "step": 23515
    },
    {
      "epoch": 0.9493742430359305,
      "grad_norm": 74.52513122558594,
      "learning_rate": 1.3413082926354794e-07,
      "loss": 3.9682,
      "step": 23516
    },
    {
      "epoch": 0.9494146144529673,
      "grad_norm": 152.88687133789062,
      "learning_rate": 1.3391746835787923e-07,
      "loss": 4.0847,
      "step": 23517
    },
    {
      "epoch": 0.949454985870004,
      "grad_norm": 101.67003631591797,
      "learning_rate": 1.337042761391827e-07,
      "loss": 4.0941,
      "step": 23518
    },
    {
      "epoch": 0.9494953572870408,
      "grad_norm": 89.27474212646484,
      "learning_rate": 1.3349125261110209e-07,
      "loss": 6.3422,
      "step": 23519
    },
    {
      "epoch": 0.9495357287040775,
      "grad_norm": 91.12220001220703,
      "learning_rate": 1.3327839777728224e-07,
      "loss": 6.0915,
      "step": 23520
    },
    {
      "epoch": 0.9495761001211143,
      "grad_norm": 85.3311767578125,
      "learning_rate": 1.330657116413614e-07,
      "loss": 4.1596,
      "step": 23521
    },
    {
      "epoch": 0.949616471538151,
      "grad_norm": 103.69060516357422,
      "learning_rate": 1.328531942069755e-07,
      "loss": 5.0975,
      "step": 23522
    },
    {
      "epoch": 0.9496568429551877,
      "grad_norm": 74.51079559326172,
      "learning_rate": 1.3264084547775834e-07,
      "loss": 5.3259,
      "step": 23523
    },
    {
      "epoch": 0.9496972143722244,
      "grad_norm": 98.16220092773438,
      "learning_rate": 1.3242866545734147e-07,
      "loss": 5.2392,
      "step": 23524
    },
    {
      "epoch": 0.9497375857892612,
      "grad_norm": 99.11540222167969,
      "learning_rate": 1.3221665414935082e-07,
      "loss": 4.0694,
      "step": 23525
    },
    {
      "epoch": 0.949777957206298,
      "grad_norm": 85.66642761230469,
      "learning_rate": 1.3200481155741462e-07,
      "loss": 4.1385,
      "step": 23526
    },
    {
      "epoch": 0.9498183286233347,
      "grad_norm": 145.174072265625,
      "learning_rate": 1.3179313768515112e-07,
      "loss": 5.6343,
      "step": 23527
    },
    {
      "epoch": 0.9498587000403714,
      "grad_norm": 82.49110412597656,
      "learning_rate": 1.3158163253618073e-07,
      "loss": 5.7285,
      "step": 23528
    },
    {
      "epoch": 0.9498990714574082,
      "grad_norm": 107.52765655517578,
      "learning_rate": 1.3137029611412054e-07,
      "loss": 5.983,
      "step": 23529
    },
    {
      "epoch": 0.9499394428744449,
      "grad_norm": 109.25914001464844,
      "learning_rate": 1.3115912842258437e-07,
      "loss": 2.9851,
      "step": 23530
    },
    {
      "epoch": 0.9499798142914816,
      "grad_norm": 73.87615966796875,
      "learning_rate": 1.3094812946518042e-07,
      "loss": 4.2018,
      "step": 23531
    },
    {
      "epoch": 0.9500201857085183,
      "grad_norm": 90.84846496582031,
      "learning_rate": 1.3073729924551805e-07,
      "loss": 5.5838,
      "step": 23532
    },
    {
      "epoch": 0.9500605571255551,
      "grad_norm": 90.71680450439453,
      "learning_rate": 1.30526637767201e-07,
      "loss": 4.9731,
      "step": 23533
    },
    {
      "epoch": 0.9501009285425919,
      "grad_norm": 395.20941162109375,
      "learning_rate": 1.3031614503383305e-07,
      "loss": 4.6751,
      "step": 23534
    },
    {
      "epoch": 0.9501412999596286,
      "grad_norm": 82.32951354980469,
      "learning_rate": 1.3010582104901025e-07,
      "loss": 5.7443,
      "step": 23535
    },
    {
      "epoch": 0.9501816713766653,
      "grad_norm": 123.74683380126953,
      "learning_rate": 1.2989566581632974e-07,
      "loss": 3.6252,
      "step": 23536
    },
    {
      "epoch": 0.950222042793702,
      "grad_norm": 105.53448486328125,
      "learning_rate": 1.2968567933938637e-07,
      "loss": 5.4167,
      "step": 23537
    },
    {
      "epoch": 0.9502624142107388,
      "grad_norm": 135.6752471923828,
      "learning_rate": 1.2947586162176839e-07,
      "loss": 5.0371,
      "step": 23538
    },
    {
      "epoch": 0.9503027856277755,
      "grad_norm": 80.58573150634766,
      "learning_rate": 1.2926621266706408e-07,
      "loss": 5.5169,
      "step": 23539
    },
    {
      "epoch": 0.9503431570448123,
      "grad_norm": 120.01957702636719,
      "learning_rate": 1.2905673247885718e-07,
      "loss": 5.8883,
      "step": 23540
    },
    {
      "epoch": 0.950383528461849,
      "grad_norm": 273.4504089355469,
      "learning_rate": 1.2884742106072933e-07,
      "loss": 6.032,
      "step": 23541
    },
    {
      "epoch": 0.9504238998788858,
      "grad_norm": 113.26219940185547,
      "learning_rate": 1.2863827841625986e-07,
      "loss": 3.1786,
      "step": 23542
    },
    {
      "epoch": 0.9504642712959225,
      "grad_norm": 116.76525115966797,
      "learning_rate": 1.284293045490259e-07,
      "loss": 4.9463,
      "step": 23543
    },
    {
      "epoch": 0.9505046427129592,
      "grad_norm": 195.77159118652344,
      "learning_rate": 1.2822049946259797e-07,
      "loss": 4.4726,
      "step": 23544
    },
    {
      "epoch": 0.9505450141299959,
      "grad_norm": 112.77780151367188,
      "learning_rate": 1.2801186316054758e-07,
      "loss": 5.869,
      "step": 23545
    },
    {
      "epoch": 0.9505853855470326,
      "grad_norm": 82.7923583984375,
      "learning_rate": 1.2780339564644194e-07,
      "loss": 5.1605,
      "step": 23546
    },
    {
      "epoch": 0.9506257569640695,
      "grad_norm": 92.26396179199219,
      "learning_rate": 1.2759509692384486e-07,
      "loss": 4.6185,
      "step": 23547
    },
    {
      "epoch": 0.9506661283811062,
      "grad_norm": 241.01290893554688,
      "learning_rate": 1.273869669963168e-07,
      "loss": 4.5356,
      "step": 23548
    },
    {
      "epoch": 0.9507064997981429,
      "grad_norm": 65.07087707519531,
      "learning_rate": 1.2717900586741827e-07,
      "loss": 4.504,
      "step": 23549
    },
    {
      "epoch": 0.9507468712151796,
      "grad_norm": 104.252685546875,
      "learning_rate": 1.269712135407053e-07,
      "loss": 5.1184,
      "step": 23550
    },
    {
      "epoch": 0.9507872426322164,
      "grad_norm": 105.01123046875,
      "learning_rate": 1.2676359001972727e-07,
      "loss": 5.9989,
      "step": 23551
    },
    {
      "epoch": 0.9508276140492531,
      "grad_norm": 182.56932067871094,
      "learning_rate": 1.265561353080369e-07,
      "loss": 4.9428,
      "step": 23552
    },
    {
      "epoch": 0.9508679854662899,
      "grad_norm": 89.77279663085938,
      "learning_rate": 1.2634884940918136e-07,
      "loss": 5.2464,
      "step": 23553
    },
    {
      "epoch": 0.9509083568833266,
      "grad_norm": 75.11480712890625,
      "learning_rate": 1.2614173232670336e-07,
      "loss": 4.742,
      "step": 23554
    },
    {
      "epoch": 0.9509487283003634,
      "grad_norm": 79.62335968017578,
      "learning_rate": 1.259347840641445e-07,
      "loss": 6.2938,
      "step": 23555
    },
    {
      "epoch": 0.9509890997174001,
      "grad_norm": 166.28305053710938,
      "learning_rate": 1.257280046250431e-07,
      "loss": 4.4153,
      "step": 23556
    },
    {
      "epoch": 0.9510294711344368,
      "grad_norm": 80.779541015625,
      "learning_rate": 1.2552139401293406e-07,
      "loss": 5.4438,
      "step": 23557
    },
    {
      "epoch": 0.9510698425514735,
      "grad_norm": 81.93794250488281,
      "learning_rate": 1.2531495223135238e-07,
      "loss": 5.9479,
      "step": 23558
    },
    {
      "epoch": 0.9511102139685103,
      "grad_norm": 119.5263442993164,
      "learning_rate": 1.2510867928382408e-07,
      "loss": 5.8455,
      "step": 23559
    },
    {
      "epoch": 0.9511505853855471,
      "grad_norm": 117.56775665283203,
      "learning_rate": 1.2490257517387862e-07,
      "loss": 4.798,
      "step": 23560
    },
    {
      "epoch": 0.9511909568025838,
      "grad_norm": 122.96662139892578,
      "learning_rate": 1.2469663990503867e-07,
      "loss": 4.3513,
      "step": 23561
    },
    {
      "epoch": 0.9512313282196205,
      "grad_norm": 103.52525329589844,
      "learning_rate": 1.24490873480827e-07,
      "loss": 3.8776,
      "step": 23562
    },
    {
      "epoch": 0.9512716996366573,
      "grad_norm": 85.13980865478516,
      "learning_rate": 1.2428527590475858e-07,
      "loss": 5.0378,
      "step": 23563
    },
    {
      "epoch": 0.951312071053694,
      "grad_norm": 82.79486083984375,
      "learning_rate": 1.2407984718035172e-07,
      "loss": 4.8415,
      "step": 23564
    },
    {
      "epoch": 0.9513524424707307,
      "grad_norm": 93.5823745727539,
      "learning_rate": 1.238745873111169e-07,
      "loss": 4.8705,
      "step": 23565
    },
    {
      "epoch": 0.9513928138877674,
      "grad_norm": 112.95487976074219,
      "learning_rate": 1.236694963005658e-07,
      "loss": 4.0927,
      "step": 23566
    },
    {
      "epoch": 0.9514331853048043,
      "grad_norm": 102.76586151123047,
      "learning_rate": 1.2346457415220226e-07,
      "loss": 5.2824,
      "step": 23567
    },
    {
      "epoch": 0.951473556721841,
      "grad_norm": 101.59136962890625,
      "learning_rate": 1.2325982086953124e-07,
      "loss": 5.4,
      "step": 23568
    },
    {
      "epoch": 0.9515139281388777,
      "grad_norm": 89.28386688232422,
      "learning_rate": 1.2305523645605222e-07,
      "loss": 4.076,
      "step": 23569
    },
    {
      "epoch": 0.9515542995559144,
      "grad_norm": 1193.406982421875,
      "learning_rate": 1.228508209152668e-07,
      "loss": 5.6556,
      "step": 23570
    },
    {
      "epoch": 0.9515946709729511,
      "grad_norm": 61.42901611328125,
      "learning_rate": 1.226465742506655e-07,
      "loss": 4.4363,
      "step": 23571
    },
    {
      "epoch": 0.9516350423899879,
      "grad_norm": 93.63581085205078,
      "learning_rate": 1.2244249646574335e-07,
      "loss": 5.1723,
      "step": 23572
    },
    {
      "epoch": 0.9516754138070246,
      "grad_norm": 88.85091400146484,
      "learning_rate": 1.2223858756398865e-07,
      "loss": 4.6664,
      "step": 23573
    },
    {
      "epoch": 0.9517157852240614,
      "grad_norm": 76.926025390625,
      "learning_rate": 1.2203484754888862e-07,
      "loss": 4.1814,
      "step": 23574
    },
    {
      "epoch": 0.9517561566410981,
      "grad_norm": 101.72649383544922,
      "learning_rate": 1.218312764239249e-07,
      "loss": 6.0687,
      "step": 23575
    },
    {
      "epoch": 0.9517965280581349,
      "grad_norm": 91.70291137695312,
      "learning_rate": 1.2162787419257916e-07,
      "loss": 4.3724,
      "step": 23576
    },
    {
      "epoch": 0.9518368994751716,
      "grad_norm": 80.65593719482422,
      "learning_rate": 1.2142464085832973e-07,
      "loss": 6.2584,
      "step": 23577
    },
    {
      "epoch": 0.9518772708922083,
      "grad_norm": 114.4541015625,
      "learning_rate": 1.212215764246505e-07,
      "loss": 4.5624,
      "step": 23578
    },
    {
      "epoch": 0.951917642309245,
      "grad_norm": 88.03678894042969,
      "learning_rate": 1.2101868089501422e-07,
      "loss": 5.5421,
      "step": 23579
    },
    {
      "epoch": 0.9519580137262817,
      "grad_norm": 110.62537384033203,
      "learning_rate": 1.2081595427288816e-07,
      "loss": 5.2611,
      "step": 23580
    },
    {
      "epoch": 0.9519983851433186,
      "grad_norm": 122.26831817626953,
      "learning_rate": 1.2061339656174066e-07,
      "loss": 5.3425,
      "step": 23581
    },
    {
      "epoch": 0.9520387565603553,
      "grad_norm": 139.900634765625,
      "learning_rate": 1.2041100776503333e-07,
      "loss": 6.2299,
      "step": 23582
    },
    {
      "epoch": 0.952079127977392,
      "grad_norm": 95.675048828125,
      "learning_rate": 1.2020878788622792e-07,
      "loss": 3.9656,
      "step": 23583
    },
    {
      "epoch": 0.9521194993944287,
      "grad_norm": 81.1706771850586,
      "learning_rate": 1.200067369287805e-07,
      "loss": 4.0268,
      "step": 23584
    },
    {
      "epoch": 0.9521598708114655,
      "grad_norm": 165.2782440185547,
      "learning_rate": 1.1980485489614723e-07,
      "loss": 4.9053,
      "step": 23585
    },
    {
      "epoch": 0.9522002422285022,
      "grad_norm": 90.870361328125,
      "learning_rate": 1.196031417917798e-07,
      "loss": 4.5629,
      "step": 23586
    },
    {
      "epoch": 0.9522406136455389,
      "grad_norm": 89.72393035888672,
      "learning_rate": 1.1940159761912539e-07,
      "loss": 4.9804,
      "step": 23587
    },
    {
      "epoch": 0.9522809850625757,
      "grad_norm": 148.67623901367188,
      "learning_rate": 1.192002223816302e-07,
      "loss": 5.1349,
      "step": 23588
    },
    {
      "epoch": 0.9523213564796125,
      "grad_norm": 87.89830780029297,
      "learning_rate": 1.1899901608273812e-07,
      "loss": 6.1082,
      "step": 23589
    },
    {
      "epoch": 0.9523617278966492,
      "grad_norm": 91.98677062988281,
      "learning_rate": 1.1879797872588972e-07,
      "loss": 3.402,
      "step": 23590
    },
    {
      "epoch": 0.9524020993136859,
      "grad_norm": 102.46646118164062,
      "learning_rate": 1.1859711031452115e-07,
      "loss": 6.0777,
      "step": 23591
    },
    {
      "epoch": 0.9524424707307226,
      "grad_norm": 95.8339614868164,
      "learning_rate": 1.1839641085206744e-07,
      "loss": 3.4796,
      "step": 23592
    },
    {
      "epoch": 0.9524828421477594,
      "grad_norm": 67.59619903564453,
      "learning_rate": 1.1819588034196027e-07,
      "loss": 6.4461,
      "step": 23593
    },
    {
      "epoch": 0.9525232135647961,
      "grad_norm": 140.72691345214844,
      "learning_rate": 1.1799551878762804e-07,
      "loss": 5.141,
      "step": 23594
    },
    {
      "epoch": 0.9525635849818329,
      "grad_norm": 86.6102294921875,
      "learning_rate": 1.1779532619249579e-07,
      "loss": 4.7898,
      "step": 23595
    },
    {
      "epoch": 0.9526039563988696,
      "grad_norm": 126.06558227539062,
      "learning_rate": 1.175953025599863e-07,
      "loss": 4.8782,
      "step": 23596
    },
    {
      "epoch": 0.9526443278159064,
      "grad_norm": 92.60002136230469,
      "learning_rate": 1.173954478935202e-07,
      "loss": 5.3986,
      "step": 23597
    },
    {
      "epoch": 0.9526846992329431,
      "grad_norm": 107.58094787597656,
      "learning_rate": 1.1719576219651585e-07,
      "loss": 4.9681,
      "step": 23598
    },
    {
      "epoch": 0.9527250706499798,
      "grad_norm": 120.02651977539062,
      "learning_rate": 1.1699624547238387e-07,
      "loss": 5.9716,
      "step": 23599
    },
    {
      "epoch": 0.9527654420670165,
      "grad_norm": 106.12165832519531,
      "learning_rate": 1.1679689772453928e-07,
      "loss": 3.8133,
      "step": 23600
    },
    {
      "epoch": 0.9528058134840532,
      "grad_norm": 98.06094360351562,
      "learning_rate": 1.1659771895638716e-07,
      "loss": 5.8003,
      "step": 23601
    },
    {
      "epoch": 0.9528461849010901,
      "grad_norm": 95.37679290771484,
      "learning_rate": 1.1639870917133589e-07,
      "loss": 6.1086,
      "step": 23602
    },
    {
      "epoch": 0.9528865563181268,
      "grad_norm": 75.2603988647461,
      "learning_rate": 1.1619986837278607e-07,
      "loss": 4.103,
      "step": 23603
    },
    {
      "epoch": 0.9529269277351635,
      "grad_norm": 92.39254760742188,
      "learning_rate": 1.1600119656413944e-07,
      "loss": 4.2372,
      "step": 23604
    },
    {
      "epoch": 0.9529672991522002,
      "grad_norm": 94.5145263671875,
      "learning_rate": 1.1580269374878994e-07,
      "loss": 4.7147,
      "step": 23605
    },
    {
      "epoch": 0.953007670569237,
      "grad_norm": 109.97904205322266,
      "learning_rate": 1.1560435993013486e-07,
      "loss": 4.0406,
      "step": 23606
    },
    {
      "epoch": 0.9530480419862737,
      "grad_norm": 77.57693481445312,
      "learning_rate": 1.1540619511156259e-07,
      "loss": 5.5304,
      "step": 23607
    },
    {
      "epoch": 0.9530884134033104,
      "grad_norm": 96.0622329711914,
      "learning_rate": 1.1520819929646155e-07,
      "loss": 2.8858,
      "step": 23608
    },
    {
      "epoch": 0.9531287848203472,
      "grad_norm": 92.77301025390625,
      "learning_rate": 1.1501037248821789e-07,
      "loss": 6.0502,
      "step": 23609
    },
    {
      "epoch": 0.953169156237384,
      "grad_norm": 80.51029968261719,
      "learning_rate": 1.148127146902156e-07,
      "loss": 5.0415,
      "step": 23610
    },
    {
      "epoch": 0.9532095276544207,
      "grad_norm": 102.29541015625,
      "learning_rate": 1.1461522590582975e-07,
      "loss": 5.3225,
      "step": 23611
    },
    {
      "epoch": 0.9532498990714574,
      "grad_norm": 89.22274780273438,
      "learning_rate": 1.1441790613844095e-07,
      "loss": 5.3462,
      "step": 23612
    },
    {
      "epoch": 0.9532902704884941,
      "grad_norm": 87.91080474853516,
      "learning_rate": 1.1422075539142096e-07,
      "loss": 4.7913,
      "step": 23613
    },
    {
      "epoch": 0.9533306419055309,
      "grad_norm": 95.15121459960938,
      "learning_rate": 1.1402377366814155e-07,
      "loss": 4.5317,
      "step": 23614
    },
    {
      "epoch": 0.9533710133225676,
      "grad_norm": 90.4070053100586,
      "learning_rate": 1.1382696097197e-07,
      "loss": 3.2156,
      "step": 23615
    },
    {
      "epoch": 0.9534113847396044,
      "grad_norm": 151.59568786621094,
      "learning_rate": 1.1363031730627138e-07,
      "loss": 6.4614,
      "step": 23616
    },
    {
      "epoch": 0.9534517561566411,
      "grad_norm": 118.57486724853516,
      "learning_rate": 1.134338426744086e-07,
      "loss": 5.05,
      "step": 23617
    },
    {
      "epoch": 0.9534921275736779,
      "grad_norm": 99.02428436279297,
      "learning_rate": 1.1323753707974005e-07,
      "loss": 5.6904,
      "step": 23618
    },
    {
      "epoch": 0.9535324989907146,
      "grad_norm": 95.86539459228516,
      "learning_rate": 1.1304140052562196e-07,
      "loss": 5.0604,
      "step": 23619
    },
    {
      "epoch": 0.9535728704077513,
      "grad_norm": 126.18829345703125,
      "learning_rate": 1.128454330154083e-07,
      "loss": 6.3831,
      "step": 23620
    },
    {
      "epoch": 0.953613241824788,
      "grad_norm": 75.18143463134766,
      "learning_rate": 1.126496345524497e-07,
      "loss": 2.5239,
      "step": 23621
    },
    {
      "epoch": 0.9536536132418247,
      "grad_norm": 87.83613586425781,
      "learning_rate": 1.1245400514009352e-07,
      "loss": 4.1101,
      "step": 23622
    },
    {
      "epoch": 0.9536939846588616,
      "grad_norm": 74.35033416748047,
      "learning_rate": 1.1225854478168486e-07,
      "loss": 4.9596,
      "step": 23623
    },
    {
      "epoch": 0.9537343560758983,
      "grad_norm": 73.26258087158203,
      "learning_rate": 1.1206325348056546e-07,
      "loss": 5.0972,
      "step": 23624
    },
    {
      "epoch": 0.953774727492935,
      "grad_norm": 97.3137435913086,
      "learning_rate": 1.1186813124007379e-07,
      "loss": 4.0857,
      "step": 23625
    },
    {
      "epoch": 0.9538150989099717,
      "grad_norm": 76.83415222167969,
      "learning_rate": 1.1167317806354827e-07,
      "loss": 5.6029,
      "step": 23626
    },
    {
      "epoch": 0.9538554703270085,
      "grad_norm": 87.08146667480469,
      "learning_rate": 1.1147839395431958e-07,
      "loss": 4.8297,
      "step": 23627
    },
    {
      "epoch": 0.9538958417440452,
      "grad_norm": 100.21143341064453,
      "learning_rate": 1.112837789157184e-07,
      "loss": 3.7032,
      "step": 23628
    },
    {
      "epoch": 0.9539362131610819,
      "grad_norm": 87.87541198730469,
      "learning_rate": 1.1108933295107315e-07,
      "loss": 3.3246,
      "step": 23629
    },
    {
      "epoch": 0.9539765845781187,
      "grad_norm": 87.38030242919922,
      "learning_rate": 1.1089505606370898e-07,
      "loss": 2.9651,
      "step": 23630
    },
    {
      "epoch": 0.9540169559951555,
      "grad_norm": 110.04883575439453,
      "learning_rate": 1.1070094825694655e-07,
      "loss": 4.7848,
      "step": 23631
    },
    {
      "epoch": 0.9540573274121922,
      "grad_norm": 86.4165267944336,
      "learning_rate": 1.1050700953410321e-07,
      "loss": 4.3186,
      "step": 23632
    },
    {
      "epoch": 0.9540976988292289,
      "grad_norm": 84.65303802490234,
      "learning_rate": 1.1031323989849741e-07,
      "loss": 5.292,
      "step": 23633
    },
    {
      "epoch": 0.9541380702462656,
      "grad_norm": 131.8012237548828,
      "learning_rate": 1.1011963935344206e-07,
      "loss": 6.0911,
      "step": 23634
    },
    {
      "epoch": 0.9541784416633023,
      "grad_norm": 121.16036987304688,
      "learning_rate": 1.0992620790224451e-07,
      "loss": 6.041,
      "step": 23635
    },
    {
      "epoch": 0.9542188130803391,
      "grad_norm": 170.3990478515625,
      "learning_rate": 1.0973294554821545e-07,
      "loss": 3.9915,
      "step": 23636
    },
    {
      "epoch": 0.9542591844973759,
      "grad_norm": 83.53238677978516,
      "learning_rate": 1.095398522946567e-07,
      "loss": 5.5488,
      "step": 23637
    },
    {
      "epoch": 0.9542995559144126,
      "grad_norm": 93.96733093261719,
      "learning_rate": 1.0934692814487003e-07,
      "loss": 4.1134,
      "step": 23638
    },
    {
      "epoch": 0.9543399273314493,
      "grad_norm": 110.66548156738281,
      "learning_rate": 1.0915417310215504e-07,
      "loss": 3.5131,
      "step": 23639
    },
    {
      "epoch": 0.9543802987484861,
      "grad_norm": 79.43424987792969,
      "learning_rate": 1.0896158716980687e-07,
      "loss": 5.6374,
      "step": 23640
    },
    {
      "epoch": 0.9544206701655228,
      "grad_norm": 132.02503967285156,
      "learning_rate": 1.0876917035111844e-07,
      "loss": 4.5174,
      "step": 23641
    },
    {
      "epoch": 0.9544610415825595,
      "grad_norm": 102.77488708496094,
      "learning_rate": 1.0857692264938047e-07,
      "loss": 4.8916,
      "step": 23642
    },
    {
      "epoch": 0.9545014129995962,
      "grad_norm": 139.38650512695312,
      "learning_rate": 1.0838484406787809e-07,
      "loss": 3.7783,
      "step": 23643
    },
    {
      "epoch": 0.9545417844166331,
      "grad_norm": 90.467529296875,
      "learning_rate": 1.0819293460989643e-07,
      "loss": 4.9234,
      "step": 23644
    },
    {
      "epoch": 0.9545821558336698,
      "grad_norm": 115.08272552490234,
      "learning_rate": 1.0800119427871625e-07,
      "loss": 4.1926,
      "step": 23645
    },
    {
      "epoch": 0.9546225272507065,
      "grad_norm": 75.05757141113281,
      "learning_rate": 1.07809623077616e-07,
      "loss": 5.0485,
      "step": 23646
    },
    {
      "epoch": 0.9546628986677432,
      "grad_norm": 62.08082962036133,
      "learning_rate": 1.0761822100987196e-07,
      "loss": 2.4142,
      "step": 23647
    },
    {
      "epoch": 0.95470327008478,
      "grad_norm": 94.18429565429688,
      "learning_rate": 1.0742698807875595e-07,
      "loss": 6.0496,
      "step": 23648
    },
    {
      "epoch": 0.9547436415018167,
      "grad_norm": 93.35931396484375,
      "learning_rate": 1.0723592428753649e-07,
      "loss": 4.5548,
      "step": 23649
    },
    {
      "epoch": 0.9547840129188534,
      "grad_norm": 108.53254699707031,
      "learning_rate": 1.0704502963948316e-07,
      "loss": 3.3441,
      "step": 23650
    },
    {
      "epoch": 0.9548243843358902,
      "grad_norm": 114.54383087158203,
      "learning_rate": 1.0685430413785669e-07,
      "loss": 5.1194,
      "step": 23651
    },
    {
      "epoch": 0.954864755752927,
      "grad_norm": 98.9874496459961,
      "learning_rate": 1.0666374778592004e-07,
      "loss": 5.1639,
      "step": 23652
    },
    {
      "epoch": 0.9549051271699637,
      "grad_norm": 84.10762023925781,
      "learning_rate": 1.0647336058693058e-07,
      "loss": 5.4856,
      "step": 23653
    },
    {
      "epoch": 0.9549454985870004,
      "grad_norm": 88.3775863647461,
      "learning_rate": 1.0628314254414462e-07,
      "loss": 5.0035,
      "step": 23654
    },
    {
      "epoch": 0.9549858700040371,
      "grad_norm": 90.14675903320312,
      "learning_rate": 1.0609309366081178e-07,
      "loss": 5.4384,
      "step": 23655
    },
    {
      "epoch": 0.9550262414210738,
      "grad_norm": 84.57687377929688,
      "learning_rate": 1.0590321394018388e-07,
      "loss": 3.5481,
      "step": 23656
    },
    {
      "epoch": 0.9550666128381106,
      "grad_norm": 81.76416778564453,
      "learning_rate": 1.0571350338550723e-07,
      "loss": 4.1807,
      "step": 23657
    },
    {
      "epoch": 0.9551069842551474,
      "grad_norm": 1073.740966796875,
      "learning_rate": 1.0552396200002368e-07,
      "loss": 5.5195,
      "step": 23658
    },
    {
      "epoch": 0.9551473556721841,
      "grad_norm": 101.58869934082031,
      "learning_rate": 1.0533458978697508e-07,
      "loss": 3.9781,
      "step": 23659
    },
    {
      "epoch": 0.9551877270892208,
      "grad_norm": 96.4172134399414,
      "learning_rate": 1.0514538674959996e-07,
      "loss": 5.0497,
      "step": 23660
    },
    {
      "epoch": 0.9552280985062576,
      "grad_norm": 172.7449493408203,
      "learning_rate": 1.0495635289113238e-07,
      "loss": 4.1687,
      "step": 23661
    },
    {
      "epoch": 0.9552684699232943,
      "grad_norm": 137.0776824951172,
      "learning_rate": 1.0476748821480532e-07,
      "loss": 4.1494,
      "step": 23662
    },
    {
      "epoch": 0.955308841340331,
      "grad_norm": 164.07383728027344,
      "learning_rate": 1.0457879272384619e-07,
      "loss": 4.8134,
      "step": 23663
    },
    {
      "epoch": 0.9553492127573678,
      "grad_norm": 190.60333251953125,
      "learning_rate": 1.0439026642148243e-07,
      "loss": 3.1679,
      "step": 23664
    },
    {
      "epoch": 0.9553895841744046,
      "grad_norm": 104.63532257080078,
      "learning_rate": 1.042019093109381e-07,
      "loss": 4.6624,
      "step": 23665
    },
    {
      "epoch": 0.9554299555914413,
      "grad_norm": 140.4925994873047,
      "learning_rate": 1.0401372139543176e-07,
      "loss": 5.1232,
      "step": 23666
    },
    {
      "epoch": 0.955470327008478,
      "grad_norm": 102.6717529296875,
      "learning_rate": 1.0382570267818304e-07,
      "loss": 4.6476,
      "step": 23667
    },
    {
      "epoch": 0.9555106984255147,
      "grad_norm": 91.11074829101562,
      "learning_rate": 1.0363785316240382e-07,
      "loss": 5.4677,
      "step": 23668
    },
    {
      "epoch": 0.9555510698425514,
      "grad_norm": 130.2726593017578,
      "learning_rate": 1.0345017285130932e-07,
      "loss": 5.6797,
      "step": 23669
    },
    {
      "epoch": 0.9555914412595882,
      "grad_norm": 106.03641510009766,
      "learning_rate": 1.0326266174810584e-07,
      "loss": 4.5591,
      "step": 23670
    },
    {
      "epoch": 0.955631812676625,
      "grad_norm": 87.79188537597656,
      "learning_rate": 1.0307531985600084e-07,
      "loss": 4.5691,
      "step": 23671
    },
    {
      "epoch": 0.9556721840936617,
      "grad_norm": 89.41844177246094,
      "learning_rate": 1.0288814717819618e-07,
      "loss": 4.6658,
      "step": 23672
    },
    {
      "epoch": 0.9557125555106984,
      "grad_norm": 103.27342224121094,
      "learning_rate": 1.0270114371789375e-07,
      "loss": 4.7538,
      "step": 23673
    },
    {
      "epoch": 0.9557529269277352,
      "grad_norm": 114.78275299072266,
      "learning_rate": 1.0251430947828988e-07,
      "loss": 4.7221,
      "step": 23674
    },
    {
      "epoch": 0.9557932983447719,
      "grad_norm": 57.52989196777344,
      "learning_rate": 1.0232764446257759e-07,
      "loss": 3.8918,
      "step": 23675
    },
    {
      "epoch": 0.9558336697618086,
      "grad_norm": 107.9859848022461,
      "learning_rate": 1.0214114867395097e-07,
      "loss": 6.547,
      "step": 23676
    },
    {
      "epoch": 0.9558740411788453,
      "grad_norm": 107.48580169677734,
      "learning_rate": 1.0195482211559749e-07,
      "loss": 5.6147,
      "step": 23677
    },
    {
      "epoch": 0.9559144125958822,
      "grad_norm": 74.15185546875,
      "learning_rate": 1.0176866479070236e-07,
      "loss": 6.5113,
      "step": 23678
    },
    {
      "epoch": 0.9559547840129189,
      "grad_norm": 89.51912689208984,
      "learning_rate": 1.015826767024497e-07,
      "loss": 5.7689,
      "step": 23679
    },
    {
      "epoch": 0.9559951554299556,
      "grad_norm": 89.24038696289062,
      "learning_rate": 1.013968578540181e-07,
      "loss": 4.4178,
      "step": 23680
    },
    {
      "epoch": 0.9560355268469923,
      "grad_norm": 159.08631896972656,
      "learning_rate": 1.012112082485861e-07,
      "loss": 5.0913,
      "step": 23681
    },
    {
      "epoch": 0.9560758982640291,
      "grad_norm": 92.58529663085938,
      "learning_rate": 1.0102572788932674e-07,
      "loss": 6.1696,
      "step": 23682
    },
    {
      "epoch": 0.9561162696810658,
      "grad_norm": 102.77973175048828,
      "learning_rate": 1.0084041677941192e-07,
      "loss": 6.0079,
      "step": 23683
    },
    {
      "epoch": 0.9561566410981025,
      "grad_norm": 289.76190185546875,
      "learning_rate": 1.006552749220091e-07,
      "loss": 5.2547,
      "step": 23684
    },
    {
      "epoch": 0.9561970125151393,
      "grad_norm": 81.25361633300781,
      "learning_rate": 1.0047030232028465e-07,
      "loss": 4.5871,
      "step": 23685
    },
    {
      "epoch": 0.956237383932176,
      "grad_norm": 65.01806640625,
      "learning_rate": 1.002854989774027e-07,
      "loss": 5.1587,
      "step": 23686
    },
    {
      "epoch": 0.9562777553492128,
      "grad_norm": 102.1993179321289,
      "learning_rate": 1.001008648965196e-07,
      "loss": 5.6169,
      "step": 23687
    },
    {
      "epoch": 0.9563181267662495,
      "grad_norm": 124.30810546875,
      "learning_rate": 9.991640008079395e-08,
      "loss": 5.2752,
      "step": 23688
    },
    {
      "epoch": 0.9563584981832862,
      "grad_norm": 751.0946044921875,
      "learning_rate": 9.973210453337878e-08,
      "loss": 3.7266,
      "step": 23689
    },
    {
      "epoch": 0.9563988696003229,
      "grad_norm": 97.90181732177734,
      "learning_rate": 9.954797825742713e-08,
      "loss": 4.7679,
      "step": 23690
    },
    {
      "epoch": 0.9564392410173597,
      "grad_norm": 95.630859375,
      "learning_rate": 9.936402125608535e-08,
      "loss": 3.9913,
      "step": 23691
    },
    {
      "epoch": 0.9564796124343965,
      "grad_norm": 99.04286193847656,
      "learning_rate": 9.918023353249873e-08,
      "loss": 3.5581,
      "step": 23692
    },
    {
      "epoch": 0.9565199838514332,
      "grad_norm": 117.60368347167969,
      "learning_rate": 9.899661508981028e-08,
      "loss": 3.9357,
      "step": 23693
    },
    {
      "epoch": 0.9565603552684699,
      "grad_norm": 123.20440673828125,
      "learning_rate": 9.881316593115975e-08,
      "loss": 4.7214,
      "step": 23694
    },
    {
      "epoch": 0.9566007266855067,
      "grad_norm": 109.01927185058594,
      "learning_rate": 9.862988605968238e-08,
      "loss": 4.9402,
      "step": 23695
    },
    {
      "epoch": 0.9566410981025434,
      "grad_norm": 91.8118896484375,
      "learning_rate": 9.844677547851345e-08,
      "loss": 5.2894,
      "step": 23696
    },
    {
      "epoch": 0.9566814695195801,
      "grad_norm": 78.57535552978516,
      "learning_rate": 9.826383419078156e-08,
      "loss": 4.196,
      "step": 23697
    },
    {
      "epoch": 0.9567218409366168,
      "grad_norm": 105.10386657714844,
      "learning_rate": 9.808106219961643e-08,
      "loss": 4.2444,
      "step": 23698
    },
    {
      "epoch": 0.9567622123536537,
      "grad_norm": 165.26846313476562,
      "learning_rate": 9.789845950814337e-08,
      "loss": 5.1909,
      "step": 23699
    },
    {
      "epoch": 0.9568025837706904,
      "grad_norm": 75.67578887939453,
      "learning_rate": 9.771602611948317e-08,
      "loss": 4.3924,
      "step": 23700
    },
    {
      "epoch": 0.9568429551877271,
      "grad_norm": 88.01982116699219,
      "learning_rate": 9.753376203675447e-08,
      "loss": 6.7242,
      "step": 23701
    },
    {
      "epoch": 0.9568833266047638,
      "grad_norm": 148.5210723876953,
      "learning_rate": 9.735166726307588e-08,
      "loss": 4.4738,
      "step": 23702
    },
    {
      "epoch": 0.9569236980218006,
      "grad_norm": 146.51785278320312,
      "learning_rate": 9.716974180155825e-08,
      "loss": 2.8268,
      "step": 23703
    },
    {
      "epoch": 0.9569640694388373,
      "grad_norm": 153.0769500732422,
      "learning_rate": 9.698798565531464e-08,
      "loss": 4.9184,
      "step": 23704
    },
    {
      "epoch": 0.957004440855874,
      "grad_norm": 82.8370590209961,
      "learning_rate": 9.680639882745035e-08,
      "loss": 4.2181,
      "step": 23705
    },
    {
      "epoch": 0.9570448122729108,
      "grad_norm": 73.54093933105469,
      "learning_rate": 9.662498132107067e-08,
      "loss": 5.1493,
      "step": 23706
    },
    {
      "epoch": 0.9570851836899475,
      "grad_norm": 133.2559051513672,
      "learning_rate": 9.644373313927868e-08,
      "loss": 3.5523,
      "step": 23707
    },
    {
      "epoch": 0.9571255551069843,
      "grad_norm": 96.97223663330078,
      "learning_rate": 9.62626542851719e-08,
      "loss": 4.5424,
      "step": 23708
    },
    {
      "epoch": 0.957165926524021,
      "grad_norm": 198.29222106933594,
      "learning_rate": 9.608174476184674e-08,
      "loss": 5.0097,
      "step": 23709
    },
    {
      "epoch": 0.9572062979410577,
      "grad_norm": 107.75772094726562,
      "learning_rate": 9.590100457239737e-08,
      "loss": 4.3825,
      "step": 23710
    },
    {
      "epoch": 0.9572466693580944,
      "grad_norm": 92.75166320800781,
      "learning_rate": 9.572043371991135e-08,
      "loss": 4.3371,
      "step": 23711
    },
    {
      "epoch": 0.9572870407751312,
      "grad_norm": 151.4779510498047,
      "learning_rate": 9.554003220747842e-08,
      "loss": 4.6477,
      "step": 23712
    },
    {
      "epoch": 0.957327412192168,
      "grad_norm": 75.7150650024414,
      "learning_rate": 9.53598000381828e-08,
      "loss": 4.1905,
      "step": 23713
    },
    {
      "epoch": 0.9573677836092047,
      "grad_norm": 105.33071899414062,
      "learning_rate": 9.517973721510642e-08,
      "loss": 5.19,
      "step": 23714
    },
    {
      "epoch": 0.9574081550262414,
      "grad_norm": 70.32556915283203,
      "learning_rate": 9.499984374132575e-08,
      "loss": 3.4763,
      "step": 23715
    },
    {
      "epoch": 0.9574485264432782,
      "grad_norm": 94.19166564941406,
      "learning_rate": 9.482011961991833e-08,
      "loss": 6.0323,
      "step": 23716
    },
    {
      "epoch": 0.9574888978603149,
      "grad_norm": 90.44891357421875,
      "learning_rate": 9.464056485395612e-08,
      "loss": 4.0372,
      "step": 23717
    },
    {
      "epoch": 0.9575292692773516,
      "grad_norm": 87.17178344726562,
      "learning_rate": 9.446117944651e-08,
      "loss": 5.636,
      "step": 23718
    },
    {
      "epoch": 0.9575696406943883,
      "grad_norm": 111.77412414550781,
      "learning_rate": 9.428196340064643e-08,
      "loss": 4.1152,
      "step": 23719
    },
    {
      "epoch": 0.9576100121114252,
      "grad_norm": 76.57078552246094,
      "learning_rate": 9.410291671943072e-08,
      "loss": 4.379,
      "step": 23720
    },
    {
      "epoch": 0.9576503835284619,
      "grad_norm": 93.19457244873047,
      "learning_rate": 9.392403940592154e-08,
      "loss": 5.3025,
      "step": 23721
    },
    {
      "epoch": 0.9576907549454986,
      "grad_norm": 112.03606414794922,
      "learning_rate": 9.374533146318088e-08,
      "loss": 6.2817,
      "step": 23722
    },
    {
      "epoch": 0.9577311263625353,
      "grad_norm": 89.48372650146484,
      "learning_rate": 9.356679289426074e-08,
      "loss": 5.4094,
      "step": 23723
    },
    {
      "epoch": 0.957771497779572,
      "grad_norm": 103.03594970703125,
      "learning_rate": 9.338842370221534e-08,
      "loss": 4.1622,
      "step": 23724
    },
    {
      "epoch": 0.9578118691966088,
      "grad_norm": 101.00064849853516,
      "learning_rate": 9.32102238900956e-08,
      "loss": 4.5122,
      "step": 23725
    },
    {
      "epoch": 0.9578522406136455,
      "grad_norm": 102.5660171508789,
      "learning_rate": 9.303219346094572e-08,
      "loss": 6.1446,
      "step": 23726
    },
    {
      "epoch": 0.9578926120306823,
      "grad_norm": 78.59412384033203,
      "learning_rate": 9.285433241781216e-08,
      "loss": 3.7298,
      "step": 23727
    },
    {
      "epoch": 0.957932983447719,
      "grad_norm": 110.82037353515625,
      "learning_rate": 9.267664076373362e-08,
      "loss": 4.8199,
      "step": 23728
    },
    {
      "epoch": 0.9579733548647558,
      "grad_norm": 114.29714965820312,
      "learning_rate": 9.249911850174876e-08,
      "loss": 3.9717,
      "step": 23729
    },
    {
      "epoch": 0.9580137262817925,
      "grad_norm": 84.99884796142578,
      "learning_rate": 9.232176563489514e-08,
      "loss": 4.303,
      "step": 23730
    },
    {
      "epoch": 0.9580540976988292,
      "grad_norm": 87.85133361816406,
      "learning_rate": 9.214458216620148e-08,
      "loss": 5.2314,
      "step": 23731
    },
    {
      "epoch": 0.9580944691158659,
      "grad_norm": 104.56558990478516,
      "learning_rate": 9.196756809869977e-08,
      "loss": 5.8068,
      "step": 23732
    },
    {
      "epoch": 0.9581348405329027,
      "grad_norm": 204.95730590820312,
      "learning_rate": 9.17907234354154e-08,
      "loss": 5.4973,
      "step": 23733
    },
    {
      "epoch": 0.9581752119499395,
      "grad_norm": 82.30531311035156,
      "learning_rate": 9.161404817937258e-08,
      "loss": 4.2623,
      "step": 23734
    },
    {
      "epoch": 0.9582155833669762,
      "grad_norm": 122.05884552001953,
      "learning_rate": 9.143754233359226e-08,
      "loss": 5.6745,
      "step": 23735
    },
    {
      "epoch": 0.9582559547840129,
      "grad_norm": 107.29194641113281,
      "learning_rate": 9.126120590109088e-08,
      "loss": 3.6744,
      "step": 23736
    },
    {
      "epoch": 0.9582963262010497,
      "grad_norm": 72.75707244873047,
      "learning_rate": 9.108503888488496e-08,
      "loss": 5.462,
      "step": 23737
    },
    {
      "epoch": 0.9583366976180864,
      "grad_norm": 91.4765625,
      "learning_rate": 9.090904128798539e-08,
      "loss": 3.1956,
      "step": 23738
    },
    {
      "epoch": 0.9583770690351231,
      "grad_norm": 434.4512634277344,
      "learning_rate": 9.0733213113402e-08,
      "loss": 4.1401,
      "step": 23739
    },
    {
      "epoch": 0.9584174404521598,
      "grad_norm": 78.6449966430664,
      "learning_rate": 9.055755436414016e-08,
      "loss": 5.3665,
      "step": 23740
    },
    {
      "epoch": 0.9584578118691967,
      "grad_norm": 110.2723617553711,
      "learning_rate": 9.038206504320524e-08,
      "loss": 4.3026,
      "step": 23741
    },
    {
      "epoch": 0.9584981832862334,
      "grad_norm": 150.11053466796875,
      "learning_rate": 9.020674515359595e-08,
      "loss": 4.8866,
      "step": 23742
    },
    {
      "epoch": 0.9585385547032701,
      "grad_norm": 86.31147003173828,
      "learning_rate": 9.003159469830991e-08,
      "loss": 4.0702,
      "step": 23743
    },
    {
      "epoch": 0.9585789261203068,
      "grad_norm": 121.67501068115234,
      "learning_rate": 8.985661368034248e-08,
      "loss": 5.1898,
      "step": 23744
    },
    {
      "epoch": 0.9586192975373435,
      "grad_norm": 118.91590881347656,
      "learning_rate": 8.968180210268462e-08,
      "loss": 4.6722,
      "step": 23745
    },
    {
      "epoch": 0.9586596689543803,
      "grad_norm": 132.46690368652344,
      "learning_rate": 8.950715996832615e-08,
      "loss": 5.9972,
      "step": 23746
    },
    {
      "epoch": 0.958700040371417,
      "grad_norm": 124.03034210205078,
      "learning_rate": 8.933268728025246e-08,
      "loss": 3.8261,
      "step": 23747
    },
    {
      "epoch": 0.9587404117884538,
      "grad_norm": 147.8499298095703,
      "learning_rate": 8.915838404144673e-08,
      "loss": 4.4941,
      "step": 23748
    },
    {
      "epoch": 0.9587807832054905,
      "grad_norm": 79.68620300292969,
      "learning_rate": 8.898425025488989e-08,
      "loss": 5.6091,
      "step": 23749
    },
    {
      "epoch": 0.9588211546225273,
      "grad_norm": 133.30380249023438,
      "learning_rate": 8.881028592355844e-08,
      "loss": 4.8736,
      "step": 23750
    },
    {
      "epoch": 0.958861526039564,
      "grad_norm": 113.90988159179688,
      "learning_rate": 8.86364910504256e-08,
      "loss": 5.9717,
      "step": 23751
    },
    {
      "epoch": 0.9589018974566007,
      "grad_norm": 102.62415313720703,
      "learning_rate": 8.846286563846451e-08,
      "loss": 6.22,
      "step": 23752
    },
    {
      "epoch": 0.9589422688736374,
      "grad_norm": 156.27488708496094,
      "learning_rate": 8.828940969064392e-08,
      "loss": 3.6487,
      "step": 23753
    },
    {
      "epoch": 0.9589826402906741,
      "grad_norm": 92.97322845458984,
      "learning_rate": 8.811612320992813e-08,
      "loss": 4.9503,
      "step": 23754
    },
    {
      "epoch": 0.959023011707711,
      "grad_norm": 76.44532012939453,
      "learning_rate": 8.794300619928142e-08,
      "loss": 4.7533,
      "step": 23755
    },
    {
      "epoch": 0.9590633831247477,
      "grad_norm": 121.9549789428711,
      "learning_rate": 8.777005866166255e-08,
      "loss": 5.522,
      "step": 23756
    },
    {
      "epoch": 0.9591037545417844,
      "grad_norm": 112.236328125,
      "learning_rate": 8.759728060002915e-08,
      "loss": 4.8525,
      "step": 23757
    },
    {
      "epoch": 0.9591441259588211,
      "grad_norm": 90.26423645019531,
      "learning_rate": 8.74246720173344e-08,
      "loss": 5.5598,
      "step": 23758
    },
    {
      "epoch": 0.9591844973758579,
      "grad_norm": 100.7726821899414,
      "learning_rate": 8.72522329165304e-08,
      "loss": 5.751,
      "step": 23759
    },
    {
      "epoch": 0.9592248687928946,
      "grad_norm": 112.22048950195312,
      "learning_rate": 8.707996330056589e-08,
      "loss": 2.3982,
      "step": 23760
    },
    {
      "epoch": 0.9592652402099313,
      "grad_norm": 107.00475311279297,
      "learning_rate": 8.69078631723852e-08,
      "loss": 5.3851,
      "step": 23761
    },
    {
      "epoch": 0.9593056116269681,
      "grad_norm": 106.33858489990234,
      "learning_rate": 8.673593253493261e-08,
      "loss": 4.3114,
      "step": 23762
    },
    {
      "epoch": 0.9593459830440049,
      "grad_norm": 111.85786437988281,
      "learning_rate": 8.65641713911447e-08,
      "loss": 6.4658,
      "step": 23763
    },
    {
      "epoch": 0.9593863544610416,
      "grad_norm": 98.26245880126953,
      "learning_rate": 8.639257974396131e-08,
      "loss": 6.4334,
      "step": 23764
    },
    {
      "epoch": 0.9594267258780783,
      "grad_norm": 74.60102081298828,
      "learning_rate": 8.622115759631345e-08,
      "loss": 4.7407,
      "step": 23765
    },
    {
      "epoch": 0.959467097295115,
      "grad_norm": 93.33533477783203,
      "learning_rate": 8.604990495113542e-08,
      "loss": 3.0994,
      "step": 23766
    },
    {
      "epoch": 0.9595074687121518,
      "grad_norm": 134.6202850341797,
      "learning_rate": 8.587882181135155e-08,
      "loss": 4.5589,
      "step": 23767
    },
    {
      "epoch": 0.9595478401291886,
      "grad_norm": 100.98176574707031,
      "learning_rate": 8.57079081798895e-08,
      "loss": 4.6812,
      "step": 23768
    },
    {
      "epoch": 0.9595882115462253,
      "grad_norm": 99.25225067138672,
      "learning_rate": 8.553716405967027e-08,
      "loss": 5.768,
      "step": 23769
    },
    {
      "epoch": 0.959628582963262,
      "grad_norm": 113.166015625,
      "learning_rate": 8.536658945361376e-08,
      "loss": 5.205,
      "step": 23770
    },
    {
      "epoch": 0.9596689543802988,
      "grad_norm": 90.74765014648438,
      "learning_rate": 8.51961843646365e-08,
      "loss": 4.5227,
      "step": 23771
    },
    {
      "epoch": 0.9597093257973355,
      "grad_norm": 120.07179260253906,
      "learning_rate": 8.502594879565174e-08,
      "loss": 4.6787,
      "step": 23772
    },
    {
      "epoch": 0.9597496972143722,
      "grad_norm": 93.91319274902344,
      "learning_rate": 8.485588274956935e-08,
      "loss": 5.4998,
      "step": 23773
    },
    {
      "epoch": 0.9597900686314089,
      "grad_norm": 88.7701416015625,
      "learning_rate": 8.468598622929924e-08,
      "loss": 5.0258,
      "step": 23774
    },
    {
      "epoch": 0.9598304400484458,
      "grad_norm": 109.06509399414062,
      "learning_rate": 8.451625923774354e-08,
      "loss": 5.11,
      "step": 23775
    },
    {
      "epoch": 0.9598708114654825,
      "grad_norm": 99.99127960205078,
      "learning_rate": 8.434670177780546e-08,
      "loss": 3.2208,
      "step": 23776
    },
    {
      "epoch": 0.9599111828825192,
      "grad_norm": 119.8577880859375,
      "learning_rate": 8.417731385238492e-08,
      "loss": 5.434,
      "step": 23777
    },
    {
      "epoch": 0.9599515542995559,
      "grad_norm": 95.37686157226562,
      "learning_rate": 8.400809546437627e-08,
      "loss": 5.8667,
      "step": 23778
    },
    {
      "epoch": 0.9599919257165926,
      "grad_norm": 72.47169494628906,
      "learning_rate": 8.383904661667386e-08,
      "loss": 5.4378,
      "step": 23779
    },
    {
      "epoch": 0.9600322971336294,
      "grad_norm": 86.14002227783203,
      "learning_rate": 8.36701673121687e-08,
      "loss": 5.1698,
      "step": 23780
    },
    {
      "epoch": 0.9600726685506661,
      "grad_norm": 107.214111328125,
      "learning_rate": 8.350145755374628e-08,
      "loss": 3.7306,
      "step": 23781
    },
    {
      "epoch": 0.9601130399677029,
      "grad_norm": 91.43147277832031,
      "learning_rate": 8.333291734429316e-08,
      "loss": 4.0145,
      "step": 23782
    },
    {
      "epoch": 0.9601534113847396,
      "grad_norm": 128.16342163085938,
      "learning_rate": 8.316454668669039e-08,
      "loss": 3.289,
      "step": 23783
    },
    {
      "epoch": 0.9601937828017764,
      "grad_norm": 84.23561096191406,
      "learning_rate": 8.299634558381564e-08,
      "loss": 4.5564,
      "step": 23784
    },
    {
      "epoch": 0.9602341542188131,
      "grad_norm": 88.02139282226562,
      "learning_rate": 8.282831403854552e-08,
      "loss": 3.5634,
      "step": 23785
    },
    {
      "epoch": 0.9602745256358498,
      "grad_norm": 115.39505004882812,
      "learning_rate": 8.266045205375328e-08,
      "loss": 4.3522,
      "step": 23786
    },
    {
      "epoch": 0.9603148970528865,
      "grad_norm": 61.141082763671875,
      "learning_rate": 8.249275963230773e-08,
      "loss": 4.5762,
      "step": 23787
    },
    {
      "epoch": 0.9603552684699233,
      "grad_norm": 84.0378189086914,
      "learning_rate": 8.232523677707771e-08,
      "loss": 6.2817,
      "step": 23788
    },
    {
      "epoch": 0.9603956398869601,
      "grad_norm": 88.16509246826172,
      "learning_rate": 8.215788349092647e-08,
      "loss": 4.8234,
      "step": 23789
    },
    {
      "epoch": 0.9604360113039968,
      "grad_norm": 141.23104858398438,
      "learning_rate": 8.199069977671725e-08,
      "loss": 3.7435,
      "step": 23790
    },
    {
      "epoch": 0.9604763827210335,
      "grad_norm": 82.44243621826172,
      "learning_rate": 8.182368563730447e-08,
      "loss": 3.6993,
      "step": 23791
    },
    {
      "epoch": 0.9605167541380703,
      "grad_norm": 78.70658874511719,
      "learning_rate": 8.165684107554694e-08,
      "loss": 5.2795,
      "step": 23792
    },
    {
      "epoch": 0.960557125555107,
      "grad_norm": 120.2384033203125,
      "learning_rate": 8.149016609429683e-08,
      "loss": 3.6418,
      "step": 23793
    },
    {
      "epoch": 0.9605974969721437,
      "grad_norm": 208.2991943359375,
      "learning_rate": 8.132366069640297e-08,
      "loss": 5.9077,
      "step": 23794
    },
    {
      "epoch": 0.9606378683891804,
      "grad_norm": 195.51951599121094,
      "learning_rate": 8.115732488471306e-08,
      "loss": 5.749,
      "step": 23795
    },
    {
      "epoch": 0.9606782398062172,
      "grad_norm": 85.99337768554688,
      "learning_rate": 8.099115866207042e-08,
      "loss": 5.2843,
      "step": 23796
    },
    {
      "epoch": 0.960718611223254,
      "grad_norm": 95.02203369140625,
      "learning_rate": 8.082516203131718e-08,
      "loss": 3.3513,
      "step": 23797
    },
    {
      "epoch": 0.9607589826402907,
      "grad_norm": 123.72469329833984,
      "learning_rate": 8.065933499529e-08,
      "loss": 5.0334,
      "step": 23798
    },
    {
      "epoch": 0.9607993540573274,
      "grad_norm": 96.8279037475586,
      "learning_rate": 8.049367755682436e-08,
      "loss": 3.6436,
      "step": 23799
    },
    {
      "epoch": 0.9608397254743641,
      "grad_norm": 129.42015075683594,
      "learning_rate": 8.032818971875355e-08,
      "loss": 4.8844,
      "step": 23800
    },
    {
      "epoch": 0.9608800968914009,
      "grad_norm": 113.05139923095703,
      "learning_rate": 8.016287148390645e-08,
      "loss": 3.8915,
      "step": 23801
    },
    {
      "epoch": 0.9609204683084376,
      "grad_norm": 74.49011993408203,
      "learning_rate": 7.999772285511077e-08,
      "loss": 4.6019,
      "step": 23802
    },
    {
      "epoch": 0.9609608397254744,
      "grad_norm": 106.22749328613281,
      "learning_rate": 7.983274383518869e-08,
      "loss": 4.9188,
      "step": 23803
    },
    {
      "epoch": 0.9610012111425111,
      "grad_norm": 116.83207702636719,
      "learning_rate": 7.96679344269613e-08,
      "loss": 5.9887,
      "step": 23804
    },
    {
      "epoch": 0.9610415825595479,
      "grad_norm": 106.70825958251953,
      "learning_rate": 7.950329463324636e-08,
      "loss": 5.0791,
      "step": 23805
    },
    {
      "epoch": 0.9610819539765846,
      "grad_norm": 81.293212890625,
      "learning_rate": 7.933882445685936e-08,
      "loss": 4.0049,
      "step": 23806
    },
    {
      "epoch": 0.9611223253936213,
      "grad_norm": 94.99821472167969,
      "learning_rate": 7.917452390061253e-08,
      "loss": 5.2386,
      "step": 23807
    },
    {
      "epoch": 0.961162696810658,
      "grad_norm": 73.53334045410156,
      "learning_rate": 7.901039296731361e-08,
      "loss": 5.3543,
      "step": 23808
    },
    {
      "epoch": 0.9612030682276947,
      "grad_norm": 70.73745727539062,
      "learning_rate": 7.884643165976924e-08,
      "loss": 3.21,
      "step": 23809
    },
    {
      "epoch": 0.9612434396447316,
      "grad_norm": 77.0933837890625,
      "learning_rate": 7.868263998078496e-08,
      "loss": 5.1504,
      "step": 23810
    },
    {
      "epoch": 0.9612838110617683,
      "grad_norm": 129.41954040527344,
      "learning_rate": 7.851901793315853e-08,
      "loss": 4.9062,
      "step": 23811
    },
    {
      "epoch": 0.961324182478805,
      "grad_norm": 104.70265197753906,
      "learning_rate": 7.835556551968882e-08,
      "loss": 2.5442,
      "step": 23812
    },
    {
      "epoch": 0.9613645538958417,
      "grad_norm": 102.74755096435547,
      "learning_rate": 7.819228274317025e-08,
      "loss": 4.1823,
      "step": 23813
    },
    {
      "epoch": 0.9614049253128785,
      "grad_norm": 125.30953979492188,
      "learning_rate": 7.802916960639506e-08,
      "loss": 5.2073,
      "step": 23814
    },
    {
      "epoch": 0.9614452967299152,
      "grad_norm": 99.395751953125,
      "learning_rate": 7.7866226112151e-08,
      "loss": 4.7585,
      "step": 23815
    },
    {
      "epoch": 0.9614856681469519,
      "grad_norm": 114.91094207763672,
      "learning_rate": 7.770345226322473e-08,
      "loss": 4.7109,
      "step": 23816
    },
    {
      "epoch": 0.9615260395639887,
      "grad_norm": 116.93559265136719,
      "learning_rate": 7.754084806239959e-08,
      "loss": 4.9208,
      "step": 23817
    },
    {
      "epoch": 0.9615664109810255,
      "grad_norm": 87.18009948730469,
      "learning_rate": 7.737841351245445e-08,
      "loss": 5.2821,
      "step": 23818
    },
    {
      "epoch": 0.9616067823980622,
      "grad_norm": 101.79122924804688,
      "learning_rate": 7.721614861616822e-08,
      "loss": 3.64,
      "step": 23819
    },
    {
      "epoch": 0.9616471538150989,
      "grad_norm": 75.53543090820312,
      "learning_rate": 7.705405337631422e-08,
      "loss": 4.7412,
      "step": 23820
    },
    {
      "epoch": 0.9616875252321356,
      "grad_norm": 107.08808135986328,
      "learning_rate": 7.689212779566357e-08,
      "loss": 4.7043,
      "step": 23821
    },
    {
      "epoch": 0.9617278966491724,
      "grad_norm": 137.2444610595703,
      "learning_rate": 7.673037187698629e-08,
      "loss": 6.608,
      "step": 23822
    },
    {
      "epoch": 0.9617682680662091,
      "grad_norm": 104.29357147216797,
      "learning_rate": 7.656878562304793e-08,
      "loss": 6.3794,
      "step": 23823
    },
    {
      "epoch": 0.9618086394832459,
      "grad_norm": 78.15036010742188,
      "learning_rate": 7.64073690366085e-08,
      "loss": 4.3931,
      "step": 23824
    },
    {
      "epoch": 0.9618490109002826,
      "grad_norm": 71.94137573242188,
      "learning_rate": 7.624612212043137e-08,
      "loss": 5.7172,
      "step": 23825
    },
    {
      "epoch": 0.9618893823173194,
      "grad_norm": 149.60946655273438,
      "learning_rate": 7.608504487727097e-08,
      "loss": 3.5476,
      "step": 23826
    },
    {
      "epoch": 0.9619297537343561,
      "grad_norm": 70.90367126464844,
      "learning_rate": 7.592413730988291e-08,
      "loss": 5.0675,
      "step": 23827
    },
    {
      "epoch": 0.9619701251513928,
      "grad_norm": 108.49105072021484,
      "learning_rate": 7.576339942101718e-08,
      "loss": 5.1598,
      "step": 23828
    },
    {
      "epoch": 0.9620104965684295,
      "grad_norm": 307.1209411621094,
      "learning_rate": 7.560283121342271e-08,
      "loss": 4.567,
      "step": 23829
    },
    {
      "epoch": 0.9620508679854662,
      "grad_norm": 135.65786743164062,
      "learning_rate": 7.544243268984509e-08,
      "loss": 5.3681,
      "step": 23830
    },
    {
      "epoch": 0.9620912394025031,
      "grad_norm": 131.11712646484375,
      "learning_rate": 7.528220385302653e-08,
      "loss": 6.2235,
      "step": 23831
    },
    {
      "epoch": 0.9621316108195398,
      "grad_norm": 96.14771270751953,
      "learning_rate": 7.5122144705706e-08,
      "loss": 3.938,
      "step": 23832
    },
    {
      "epoch": 0.9621719822365765,
      "grad_norm": 91.38296508789062,
      "learning_rate": 7.496225525062018e-08,
      "loss": 6.1907,
      "step": 23833
    },
    {
      "epoch": 0.9622123536536132,
      "grad_norm": 81.92670440673828,
      "learning_rate": 7.480253549050354e-08,
      "loss": 5.3101,
      "step": 23834
    },
    {
      "epoch": 0.96225272507065,
      "grad_norm": 85.78988647460938,
      "learning_rate": 7.464298542808612e-08,
      "loss": 4.669,
      "step": 23835
    },
    {
      "epoch": 0.9622930964876867,
      "grad_norm": 73.0750503540039,
      "learning_rate": 7.448360506609687e-08,
      "loss": 3.6162,
      "step": 23836
    },
    {
      "epoch": 0.9623334679047234,
      "grad_norm": 94.69686126708984,
      "learning_rate": 7.432439440726024e-08,
      "loss": 3.875,
      "step": 23837
    },
    {
      "epoch": 0.9623738393217602,
      "grad_norm": 71.25617218017578,
      "learning_rate": 7.416535345429743e-08,
      "loss": 5.8946,
      "step": 23838
    },
    {
      "epoch": 0.962414210738797,
      "grad_norm": 91.67200469970703,
      "learning_rate": 7.400648220992957e-08,
      "loss": 4.8718,
      "step": 23839
    },
    {
      "epoch": 0.9624545821558337,
      "grad_norm": 93.21744537353516,
      "learning_rate": 7.384778067687115e-08,
      "loss": 6.263,
      "step": 23840
    },
    {
      "epoch": 0.9624949535728704,
      "grad_norm": 88.13302612304688,
      "learning_rate": 7.368924885783669e-08,
      "loss": 5.2474,
      "step": 23841
    },
    {
      "epoch": 0.9625353249899071,
      "grad_norm": 84.97530364990234,
      "learning_rate": 7.353088675553732e-08,
      "loss": 4.3582,
      "step": 23842
    },
    {
      "epoch": 0.9625756964069438,
      "grad_norm": 92.25150299072266,
      "learning_rate": 7.33726943726798e-08,
      "loss": 3.5237,
      "step": 23843
    },
    {
      "epoch": 0.9626160678239806,
      "grad_norm": 148.64642333984375,
      "learning_rate": 7.32146717119675e-08,
      "loss": 3.5729,
      "step": 23844
    },
    {
      "epoch": 0.9626564392410174,
      "grad_norm": 98.32735443115234,
      "learning_rate": 7.305681877610493e-08,
      "loss": 5.3864,
      "step": 23845
    },
    {
      "epoch": 0.9626968106580541,
      "grad_norm": 95.94263458251953,
      "learning_rate": 7.289913556778883e-08,
      "loss": 5.3752,
      "step": 23846
    },
    {
      "epoch": 0.9627371820750908,
      "grad_norm": 104.61463928222656,
      "learning_rate": 7.274162208971702e-08,
      "loss": 5.0913,
      "step": 23847
    },
    {
      "epoch": 0.9627775534921276,
      "grad_norm": 196.3179473876953,
      "learning_rate": 7.25842783445807e-08,
      "loss": 3.9383,
      "step": 23848
    },
    {
      "epoch": 0.9628179249091643,
      "grad_norm": 90.31390380859375,
      "learning_rate": 7.242710433507105e-08,
      "loss": 5.0168,
      "step": 23849
    },
    {
      "epoch": 0.962858296326201,
      "grad_norm": 133.57888793945312,
      "learning_rate": 7.227010006387591e-08,
      "loss": 4.3575,
      "step": 23850
    },
    {
      "epoch": 0.9628986677432377,
      "grad_norm": 78.05300903320312,
      "learning_rate": 7.21132655336787e-08,
      "loss": 4.1839,
      "step": 23851
    },
    {
      "epoch": 0.9629390391602746,
      "grad_norm": 82.08668518066406,
      "learning_rate": 7.19566007471617e-08,
      "loss": 4.3242,
      "step": 23852
    },
    {
      "epoch": 0.9629794105773113,
      "grad_norm": 89.24471282958984,
      "learning_rate": 7.18001057070028e-08,
      "loss": 4.9464,
      "step": 23853
    },
    {
      "epoch": 0.963019781994348,
      "grad_norm": 179.16212463378906,
      "learning_rate": 7.164378041587871e-08,
      "loss": 4.7357,
      "step": 23854
    },
    {
      "epoch": 0.9630601534113847,
      "grad_norm": 74.51239013671875,
      "learning_rate": 7.148762487646066e-08,
      "loss": 6.1319,
      "step": 23855
    },
    {
      "epoch": 0.9631005248284215,
      "grad_norm": 122.86751556396484,
      "learning_rate": 7.133163909141982e-08,
      "loss": 3.6226,
      "step": 23856
    },
    {
      "epoch": 0.9631408962454582,
      "grad_norm": 90.53645324707031,
      "learning_rate": 7.117582306342296e-08,
      "loss": 4.5297,
      "step": 23857
    },
    {
      "epoch": 0.9631812676624949,
      "grad_norm": 156.308349609375,
      "learning_rate": 7.102017679513462e-08,
      "loss": 6.0772,
      "step": 23858
    },
    {
      "epoch": 0.9632216390795317,
      "grad_norm": 80.7350082397461,
      "learning_rate": 7.086470028921488e-08,
      "loss": 4.3942,
      "step": 23859
    },
    {
      "epoch": 0.9632620104965685,
      "grad_norm": 122.34687042236328,
      "learning_rate": 7.070939354832274e-08,
      "loss": 5.0356,
      "step": 23860
    },
    {
      "epoch": 0.9633023819136052,
      "grad_norm": 93.28264617919922,
      "learning_rate": 7.055425657511384e-08,
      "loss": 3.5746,
      "step": 23861
    },
    {
      "epoch": 0.9633427533306419,
      "grad_norm": 109.92386627197266,
      "learning_rate": 7.039928937223939e-08,
      "loss": 3.8794,
      "step": 23862
    },
    {
      "epoch": 0.9633831247476786,
      "grad_norm": 99.93302154541016,
      "learning_rate": 7.024449194235061e-08,
      "loss": 4.242,
      "step": 23863
    },
    {
      "epoch": 0.9634234961647153,
      "grad_norm": 79.03084564208984,
      "learning_rate": 7.008986428809317e-08,
      "loss": 4.8017,
      "step": 23864
    },
    {
      "epoch": 0.9634638675817521,
      "grad_norm": 94.21102142333984,
      "learning_rate": 6.99354064121105e-08,
      "loss": 5.1518,
      "step": 23865
    },
    {
      "epoch": 0.9635042389987889,
      "grad_norm": 150.4044952392578,
      "learning_rate": 6.978111831704382e-08,
      "loss": 6.7848,
      "step": 23866
    },
    {
      "epoch": 0.9635446104158256,
      "grad_norm": 172.8143768310547,
      "learning_rate": 6.962700000553214e-08,
      "loss": 4.8059,
      "step": 23867
    },
    {
      "epoch": 0.9635849818328623,
      "grad_norm": 209.0765380859375,
      "learning_rate": 6.947305148020888e-08,
      "loss": 3.2525,
      "step": 23868
    },
    {
      "epoch": 0.9636253532498991,
      "grad_norm": 81.77412414550781,
      "learning_rate": 6.931927274370753e-08,
      "loss": 4.6174,
      "step": 23869
    },
    {
      "epoch": 0.9636657246669358,
      "grad_norm": 103.36084747314453,
      "learning_rate": 6.916566379865708e-08,
      "loss": 2.5716,
      "step": 23870
    },
    {
      "epoch": 0.9637060960839725,
      "grad_norm": 105.98815155029297,
      "learning_rate": 6.90122246476832e-08,
      "loss": 4.9002,
      "step": 23871
    },
    {
      "epoch": 0.9637464675010093,
      "grad_norm": 75.64022827148438,
      "learning_rate": 6.885895529340936e-08,
      "loss": 2.6216,
      "step": 23872
    },
    {
      "epoch": 0.9637868389180461,
      "grad_norm": 126.81615447998047,
      "learning_rate": 6.870585573845678e-08,
      "loss": 4.7207,
      "step": 23873
    },
    {
      "epoch": 0.9638272103350828,
      "grad_norm": 96.18606567382812,
      "learning_rate": 6.855292598544338e-08,
      "loss": 4.2892,
      "step": 23874
    },
    {
      "epoch": 0.9638675817521195,
      "grad_norm": 103.25565338134766,
      "learning_rate": 6.840016603698263e-08,
      "loss": 3.8643,
      "step": 23875
    },
    {
      "epoch": 0.9639079531691562,
      "grad_norm": 109.06312561035156,
      "learning_rate": 6.824757589568687e-08,
      "loss": 5.9995,
      "step": 23876
    },
    {
      "epoch": 0.963948324586193,
      "grad_norm": 109.55420684814453,
      "learning_rate": 6.809515556416624e-08,
      "loss": 4.8803,
      "step": 23877
    },
    {
      "epoch": 0.9639886960032297,
      "grad_norm": 105.08541107177734,
      "learning_rate": 6.794290504502532e-08,
      "loss": 5.0263,
      "step": 23878
    },
    {
      "epoch": 0.9640290674202665,
      "grad_norm": 72.34828186035156,
      "learning_rate": 6.779082434086759e-08,
      "loss": 6.0305,
      "step": 23879
    },
    {
      "epoch": 0.9640694388373032,
      "grad_norm": 95.25981140136719,
      "learning_rate": 6.763891345429318e-08,
      "loss": 5.0342,
      "step": 23880
    },
    {
      "epoch": 0.96410981025434,
      "grad_norm": 114.91239166259766,
      "learning_rate": 6.748717238790004e-08,
      "loss": 5.1989,
      "step": 23881
    },
    {
      "epoch": 0.9641501816713767,
      "grad_norm": 85.40841674804688,
      "learning_rate": 6.733560114428273e-08,
      "loss": 3.828,
      "step": 23882
    },
    {
      "epoch": 0.9641905530884134,
      "grad_norm": 120.97427368164062,
      "learning_rate": 6.71841997260314e-08,
      "loss": 5.9264,
      "step": 23883
    },
    {
      "epoch": 0.9642309245054501,
      "grad_norm": 102.91988372802734,
      "learning_rate": 6.703296813573512e-08,
      "loss": 4.5715,
      "step": 23884
    },
    {
      "epoch": 0.9642712959224868,
      "grad_norm": 108.00580596923828,
      "learning_rate": 6.688190637598069e-08,
      "loss": 4.2459,
      "step": 23885
    },
    {
      "epoch": 0.9643116673395237,
      "grad_norm": 83.1582260131836,
      "learning_rate": 6.673101444934938e-08,
      "loss": 5.537,
      "step": 23886
    },
    {
      "epoch": 0.9643520387565604,
      "grad_norm": 156.38507080078125,
      "learning_rate": 6.658029235842356e-08,
      "loss": 6.3876,
      "step": 23887
    },
    {
      "epoch": 0.9643924101735971,
      "grad_norm": 102.6960220336914,
      "learning_rate": 6.642974010577674e-08,
      "loss": 3.8308,
      "step": 23888
    },
    {
      "epoch": 0.9644327815906338,
      "grad_norm": 92.41825103759766,
      "learning_rate": 6.627935769398575e-08,
      "loss": 4.5833,
      "step": 23889
    },
    {
      "epoch": 0.9644731530076706,
      "grad_norm": 64.2735366821289,
      "learning_rate": 6.612914512562074e-08,
      "loss": 5.4147,
      "step": 23890
    },
    {
      "epoch": 0.9645135244247073,
      "grad_norm": 79.95897674560547,
      "learning_rate": 6.597910240324967e-08,
      "loss": 3.9579,
      "step": 23891
    },
    {
      "epoch": 0.964553895841744,
      "grad_norm": 73.46068572998047,
      "learning_rate": 6.582922952943827e-08,
      "loss": 5.2947,
      "step": 23892
    },
    {
      "epoch": 0.9645942672587808,
      "grad_norm": 128.58026123046875,
      "learning_rate": 6.567952650675002e-08,
      "loss": 6.2174,
      "step": 23893
    },
    {
      "epoch": 0.9646346386758176,
      "grad_norm": 69.08468627929688,
      "learning_rate": 6.55299933377418e-08,
      "loss": 6.391,
      "step": 23894
    },
    {
      "epoch": 0.9646750100928543,
      "grad_norm": 81.6243896484375,
      "learning_rate": 6.538063002497375e-08,
      "loss": 4.0947,
      "step": 23895
    },
    {
      "epoch": 0.964715381509891,
      "grad_norm": 87.49816131591797,
      "learning_rate": 6.523143657099717e-08,
      "loss": 4.2938,
      "step": 23896
    },
    {
      "epoch": 0.9647557529269277,
      "grad_norm": 108.71394348144531,
      "learning_rate": 6.508241297836338e-08,
      "loss": 3.4043,
      "step": 23897
    },
    {
      "epoch": 0.9647961243439644,
      "grad_norm": 104.3987045288086,
      "learning_rate": 6.493355924962141e-08,
      "loss": 3.6757,
      "step": 23898
    },
    {
      "epoch": 0.9648364957610012,
      "grad_norm": 129.1227569580078,
      "learning_rate": 6.478487538731482e-08,
      "loss": 6.1419,
      "step": 23899
    },
    {
      "epoch": 0.964876867178038,
      "grad_norm": 168.66586303710938,
      "learning_rate": 6.46363613939871e-08,
      "loss": 5.9587,
      "step": 23900
    },
    {
      "epoch": 0.9649172385950747,
      "grad_norm": 124.45206451416016,
      "learning_rate": 6.448801727217623e-08,
      "loss": 4.186,
      "step": 23901
    },
    {
      "epoch": 0.9649576100121114,
      "grad_norm": 120.15858459472656,
      "learning_rate": 6.43398430244202e-08,
      "loss": 4.9678,
      "step": 23902
    },
    {
      "epoch": 0.9649979814291482,
      "grad_norm": 106.6346435546875,
      "learning_rate": 6.419183865325029e-08,
      "loss": 4.5133,
      "step": 23903
    },
    {
      "epoch": 0.9650383528461849,
      "grad_norm": 97.71534729003906,
      "learning_rate": 6.404400416119894e-08,
      "loss": 4.8201,
      "step": 23904
    },
    {
      "epoch": 0.9650787242632216,
      "grad_norm": 89.94776153564453,
      "learning_rate": 6.3896339550793e-08,
      "loss": 5.2809,
      "step": 23905
    },
    {
      "epoch": 0.9651190956802583,
      "grad_norm": 72.74781036376953,
      "learning_rate": 6.374884482455602e-08,
      "loss": 3.1486,
      "step": 23906
    },
    {
      "epoch": 0.9651594670972952,
      "grad_norm": 90.4874267578125,
      "learning_rate": 6.360151998501263e-08,
      "loss": 5.0415,
      "step": 23907
    },
    {
      "epoch": 0.9651998385143319,
      "grad_norm": 102.13629913330078,
      "learning_rate": 6.345436503467861e-08,
      "loss": 3.9265,
      "step": 23908
    },
    {
      "epoch": 0.9652402099313686,
      "grad_norm": 223.2011260986328,
      "learning_rate": 6.330737997607194e-08,
      "loss": 3.101,
      "step": 23909
    },
    {
      "epoch": 0.9652805813484053,
      "grad_norm": 74.78175354003906,
      "learning_rate": 6.316056481170618e-08,
      "loss": 5.2457,
      "step": 23910
    },
    {
      "epoch": 0.965320952765442,
      "grad_norm": 101.6577377319336,
      "learning_rate": 6.30139195440893e-08,
      "loss": 5.9549,
      "step": 23911
    },
    {
      "epoch": 0.9653613241824788,
      "grad_norm": 78.5511245727539,
      "learning_rate": 6.286744417573043e-08,
      "loss": 4.0676,
      "step": 23912
    },
    {
      "epoch": 0.9654016955995155,
      "grad_norm": 130.8444061279297,
      "learning_rate": 6.27211387091331e-08,
      "loss": 3.9669,
      "step": 23913
    },
    {
      "epoch": 0.9654420670165523,
      "grad_norm": 85.7083969116211,
      "learning_rate": 6.257500314679976e-08,
      "loss": 5.6583,
      "step": 23914
    },
    {
      "epoch": 0.965482438433589,
      "grad_norm": 105.7531967163086,
      "learning_rate": 6.24290374912273e-08,
      "loss": 5.0037,
      "step": 23915
    },
    {
      "epoch": 0.9655228098506258,
      "grad_norm": 139.17526245117188,
      "learning_rate": 6.228324174491262e-08,
      "loss": 4.3434,
      "step": 23916
    },
    {
      "epoch": 0.9655631812676625,
      "grad_norm": 127.17064666748047,
      "learning_rate": 6.213761591034929e-08,
      "loss": 4.4642,
      "step": 23917
    },
    {
      "epoch": 0.9656035526846992,
      "grad_norm": 101.8849105834961,
      "learning_rate": 6.199215999002528e-08,
      "loss": 5.1878,
      "step": 23918
    },
    {
      "epoch": 0.9656439241017359,
      "grad_norm": 77.51582336425781,
      "learning_rate": 6.184687398642863e-08,
      "loss": 5.2541,
      "step": 23919
    },
    {
      "epoch": 0.9656842955187727,
      "grad_norm": 105.44208526611328,
      "learning_rate": 6.170175790204292e-08,
      "loss": 4.7347,
      "step": 23920
    },
    {
      "epoch": 0.9657246669358095,
      "grad_norm": 88.06505584716797,
      "learning_rate": 6.155681173934947e-08,
      "loss": 4.5037,
      "step": 23921
    },
    {
      "epoch": 0.9657650383528462,
      "grad_norm": 103.54046630859375,
      "learning_rate": 6.141203550082741e-08,
      "loss": 5.0524,
      "step": 23922
    },
    {
      "epoch": 0.9658054097698829,
      "grad_norm": 86.87590026855469,
      "learning_rate": 6.126742918895146e-08,
      "loss": 4.0982,
      "step": 23923
    },
    {
      "epoch": 0.9658457811869197,
      "grad_norm": 98.34092712402344,
      "learning_rate": 6.112299280619294e-08,
      "loss": 5.9802,
      "step": 23924
    },
    {
      "epoch": 0.9658861526039564,
      "grad_norm": 77.54405212402344,
      "learning_rate": 6.097872635502211e-08,
      "loss": 4.0653,
      "step": 23925
    },
    {
      "epoch": 0.9659265240209931,
      "grad_norm": 94.54569244384766,
      "learning_rate": 6.08346298379059e-08,
      "loss": 5.7633,
      "step": 23926
    },
    {
      "epoch": 0.9659668954380298,
      "grad_norm": 118.69453430175781,
      "learning_rate": 6.069070325730897e-08,
      "loss": 5.6614,
      "step": 23927
    },
    {
      "epoch": 0.9660072668550667,
      "grad_norm": 82.21311950683594,
      "learning_rate": 6.05469466156905e-08,
      "loss": 5.5477,
      "step": 23928
    },
    {
      "epoch": 0.9660476382721034,
      "grad_norm": 89.50126647949219,
      "learning_rate": 6.04033599155085e-08,
      "loss": 3.8109,
      "step": 23929
    },
    {
      "epoch": 0.9660880096891401,
      "grad_norm": 134.27589416503906,
      "learning_rate": 6.02599431592188e-08,
      "loss": 4.444,
      "step": 23930
    },
    {
      "epoch": 0.9661283811061768,
      "grad_norm": 126.60257720947266,
      "learning_rate": 6.011669634927276e-08,
      "loss": 4.6678,
      "step": 23931
    },
    {
      "epoch": 0.9661687525232135,
      "grad_norm": 106.09724426269531,
      "learning_rate": 5.997361948812064e-08,
      "loss": 3.4963,
      "step": 23932
    },
    {
      "epoch": 0.9662091239402503,
      "grad_norm": 100.08695220947266,
      "learning_rate": 5.983071257820717e-08,
      "loss": 3.53,
      "step": 23933
    },
    {
      "epoch": 0.966249495357287,
      "grad_norm": 120.487060546875,
      "learning_rate": 5.968797562197592e-08,
      "loss": 5.2669,
      "step": 23934
    },
    {
      "epoch": 0.9662898667743238,
      "grad_norm": 93.18586730957031,
      "learning_rate": 5.9545408621869414e-08,
      "loss": 6.0353,
      "step": 23935
    },
    {
      "epoch": 0.9663302381913605,
      "grad_norm": 64.56645202636719,
      "learning_rate": 5.940301158032347e-08,
      "loss": 5.068,
      "step": 23936
    },
    {
      "epoch": 0.9663706096083973,
      "grad_norm": 94.33765411376953,
      "learning_rate": 5.926078449977279e-08,
      "loss": 4.3873,
      "step": 23937
    },
    {
      "epoch": 0.966410981025434,
      "grad_norm": 79.67808532714844,
      "learning_rate": 5.9118727382648785e-08,
      "loss": 4.8734,
      "step": 23938
    },
    {
      "epoch": 0.9664513524424707,
      "grad_norm": 91.81649017333984,
      "learning_rate": 5.897684023138284e-08,
      "loss": 3.7481,
      "step": 23939
    },
    {
      "epoch": 0.9664917238595074,
      "grad_norm": 80.421630859375,
      "learning_rate": 5.8835123048397444e-08,
      "loss": 4.0675,
      "step": 23940
    },
    {
      "epoch": 0.9665320952765442,
      "grad_norm": 115.48143005371094,
      "learning_rate": 5.869357583611734e-08,
      "loss": 4.2722,
      "step": 23941
    },
    {
      "epoch": 0.966572466693581,
      "grad_norm": 121.3809814453125,
      "learning_rate": 5.855219859696282e-08,
      "loss": 4.5486,
      "step": 23942
    },
    {
      "epoch": 0.9666128381106177,
      "grad_norm": 108.28184509277344,
      "learning_rate": 5.841099133335082e-08,
      "loss": 5.471,
      "step": 23943
    },
    {
      "epoch": 0.9666532095276544,
      "grad_norm": 90.95744323730469,
      "learning_rate": 5.826995404769498e-08,
      "loss": 5.6908,
      "step": 23944
    },
    {
      "epoch": 0.9666935809446912,
      "grad_norm": 142.105712890625,
      "learning_rate": 5.8129086742407806e-08,
      "loss": 5.0497,
      "step": 23945
    },
    {
      "epoch": 0.9667339523617279,
      "grad_norm": 82.62969207763672,
      "learning_rate": 5.7988389419896266e-08,
      "loss": 3.2256,
      "step": 23946
    },
    {
      "epoch": 0.9667743237787646,
      "grad_norm": 80.4624252319336,
      "learning_rate": 5.7847862082568426e-08,
      "loss": 6.2784,
      "step": 23947
    },
    {
      "epoch": 0.9668146951958013,
      "grad_norm": 100.3480224609375,
      "learning_rate": 5.770750473282571e-08,
      "loss": 4.7554,
      "step": 23948
    },
    {
      "epoch": 0.9668550666128382,
      "grad_norm": 375.5757751464844,
      "learning_rate": 5.75673173730662e-08,
      "loss": 5.378,
      "step": 23949
    },
    {
      "epoch": 0.9668954380298749,
      "grad_norm": 79.01595306396484,
      "learning_rate": 5.742730000568908e-08,
      "loss": 2.613,
      "step": 23950
    },
    {
      "epoch": 0.9669358094469116,
      "grad_norm": 103.75507354736328,
      "learning_rate": 5.728745263308799e-08,
      "loss": 5.0631,
      "step": 23951
    },
    {
      "epoch": 0.9669761808639483,
      "grad_norm": 132.14627075195312,
      "learning_rate": 5.714777525765325e-08,
      "loss": 5.7776,
      "step": 23952
    },
    {
      "epoch": 0.967016552280985,
      "grad_norm": 122.64647674560547,
      "learning_rate": 5.700826788177294e-08,
      "loss": 4.2629,
      "step": 23953
    },
    {
      "epoch": 0.9670569236980218,
      "grad_norm": 101.39669799804688,
      "learning_rate": 5.6868930507832934e-08,
      "loss": 4.8839,
      "step": 23954
    },
    {
      "epoch": 0.9670972951150585,
      "grad_norm": 106.88768768310547,
      "learning_rate": 5.6729763138215765e-08,
      "loss": 6.0735,
      "step": 23955
    },
    {
      "epoch": 0.9671376665320953,
      "grad_norm": 202.68479919433594,
      "learning_rate": 5.659076577529954e-08,
      "loss": 3.8611,
      "step": 23956
    },
    {
      "epoch": 0.967178037949132,
      "grad_norm": 84.72271728515625,
      "learning_rate": 5.645193842146235e-08,
      "loss": 5.0007,
      "step": 23957
    },
    {
      "epoch": 0.9672184093661688,
      "grad_norm": 107.68905639648438,
      "learning_rate": 5.631328107907674e-08,
      "loss": 6.144,
      "step": 23958
    },
    {
      "epoch": 0.9672587807832055,
      "grad_norm": 72.60667419433594,
      "learning_rate": 5.6174793750515265e-08,
      "loss": 5.4072,
      "step": 23959
    },
    {
      "epoch": 0.9672991522002422,
      "grad_norm": 107.92581176757812,
      "learning_rate": 5.6036476438142694e-08,
      "loss": 5.5191,
      "step": 23960
    },
    {
      "epoch": 0.9673395236172789,
      "grad_norm": 80.4281997680664,
      "learning_rate": 5.589832914432603e-08,
      "loss": 4.8538,
      "step": 23961
    },
    {
      "epoch": 0.9673798950343157,
      "grad_norm": 87.92267608642578,
      "learning_rate": 5.5760351871426696e-08,
      "loss": 4.8549,
      "step": 23962
    },
    {
      "epoch": 0.9674202664513525,
      "grad_norm": 114.6522445678711,
      "learning_rate": 5.5622544621805055e-08,
      "loss": 4.3858,
      "step": 23963
    },
    {
      "epoch": 0.9674606378683892,
      "grad_norm": 152.24917602539062,
      "learning_rate": 5.548490739781476e-08,
      "loss": 5.0879,
      "step": 23964
    },
    {
      "epoch": 0.9675010092854259,
      "grad_norm": 90.58613586425781,
      "learning_rate": 5.5347440201810596e-08,
      "loss": 5.2479,
      "step": 23965
    },
    {
      "epoch": 0.9675413807024627,
      "grad_norm": 84.12938690185547,
      "learning_rate": 5.5210143036142914e-08,
      "loss": 5.5047,
      "step": 23966
    },
    {
      "epoch": 0.9675817521194994,
      "grad_norm": 73.88316345214844,
      "learning_rate": 5.507301590315872e-08,
      "loss": 5.8043,
      "step": 23967
    },
    {
      "epoch": 0.9676221235365361,
      "grad_norm": 74.38880157470703,
      "learning_rate": 5.493605880520392e-08,
      "loss": 4.6114,
      "step": 23968
    },
    {
      "epoch": 0.9676624949535728,
      "grad_norm": 104.17770385742188,
      "learning_rate": 5.479927174461774e-08,
      "loss": 6.5572,
      "step": 23969
    },
    {
      "epoch": 0.9677028663706096,
      "grad_norm": 104.95466613769531,
      "learning_rate": 5.466265472374055e-08,
      "loss": 4.6517,
      "step": 23970
    },
    {
      "epoch": 0.9677432377876464,
      "grad_norm": 94.38333129882812,
      "learning_rate": 5.452620774490935e-08,
      "loss": 5.7717,
      "step": 23971
    },
    {
      "epoch": 0.9677836092046831,
      "grad_norm": 89.12199401855469,
      "learning_rate": 5.4389930810454515e-08,
      "loss": 3.4844,
      "step": 23972
    },
    {
      "epoch": 0.9678239806217198,
      "grad_norm": 124.18482971191406,
      "learning_rate": 5.42538239227075e-08,
      "loss": 6.3638,
      "step": 23973
    },
    {
      "epoch": 0.9678643520387565,
      "grad_norm": 88.0678482055664,
      "learning_rate": 5.411788708399424e-08,
      "loss": 3.7697,
      "step": 23974
    },
    {
      "epoch": 0.9679047234557933,
      "grad_norm": 154.3600311279297,
      "learning_rate": 5.3982120296640625e-08,
      "loss": 4.6511,
      "step": 23975
    },
    {
      "epoch": 0.96794509487283,
      "grad_norm": 68.47300720214844,
      "learning_rate": 5.3846523562967044e-08,
      "loss": 4.2265,
      "step": 23976
    },
    {
      "epoch": 0.9679854662898668,
      "grad_norm": 94.5809555053711,
      "learning_rate": 5.371109688529163e-08,
      "loss": 5.0499,
      "step": 23977
    },
    {
      "epoch": 0.9680258377069035,
      "grad_norm": 69.50584411621094,
      "learning_rate": 5.3575840265930326e-08,
      "loss": 5.6904,
      "step": 23978
    },
    {
      "epoch": 0.9680662091239403,
      "grad_norm": 76.48766326904297,
      "learning_rate": 5.344075370719681e-08,
      "loss": 4.1063,
      "step": 23979
    },
    {
      "epoch": 0.968106580540977,
      "grad_norm": 142.24380493164062,
      "learning_rate": 5.330583721139815e-08,
      "loss": 5.8669,
      "step": 23980
    },
    {
      "epoch": 0.9681469519580137,
      "grad_norm": 124.38912200927734,
      "learning_rate": 5.317109078084248e-08,
      "loss": 3.1731,
      "step": 23981
    },
    {
      "epoch": 0.9681873233750504,
      "grad_norm": 162.42391967773438,
      "learning_rate": 5.303651441783353e-08,
      "loss": 6.3156,
      "step": 23982
    },
    {
      "epoch": 0.9682276947920873,
      "grad_norm": 108.6296615600586,
      "learning_rate": 5.290210812467389e-08,
      "loss": 5.2809,
      "step": 23983
    },
    {
      "epoch": 0.968268066209124,
      "grad_norm": 114.5119857788086,
      "learning_rate": 5.27678719036584e-08,
      "loss": 6.6461,
      "step": 23984
    },
    {
      "epoch": 0.9683084376261607,
      "grad_norm": 110.07296752929688,
      "learning_rate": 5.26338057570841e-08,
      "loss": 4.7204,
      "step": 23985
    },
    {
      "epoch": 0.9683488090431974,
      "grad_norm": 127.40071868896484,
      "learning_rate": 5.249990968724361e-08,
      "loss": 5.6283,
      "step": 23986
    },
    {
      "epoch": 0.9683891804602341,
      "grad_norm": 107.38494110107422,
      "learning_rate": 5.23661836964251e-08,
      "loss": 3.9027,
      "step": 23987
    },
    {
      "epoch": 0.9684295518772709,
      "grad_norm": 85.83878326416016,
      "learning_rate": 5.223262778691562e-08,
      "loss": 4.9368,
      "step": 23988
    },
    {
      "epoch": 0.9684699232943076,
      "grad_norm": 104.63510131835938,
      "learning_rate": 5.209924196099892e-08,
      "loss": 7.0083,
      "step": 23989
    },
    {
      "epoch": 0.9685102947113444,
      "grad_norm": 106.71320343017578,
      "learning_rate": 5.196602622095426e-08,
      "loss": 5.7651,
      "step": 23990
    },
    {
      "epoch": 0.9685506661283811,
      "grad_norm": 100.55472564697266,
      "learning_rate": 5.183298056906094e-08,
      "loss": 3.2365,
      "step": 23991
    },
    {
      "epoch": 0.9685910375454179,
      "grad_norm": 139.51609802246094,
      "learning_rate": 5.170010500759271e-08,
      "loss": 4.8781,
      "step": 23992
    },
    {
      "epoch": 0.9686314089624546,
      "grad_norm": 83.66014099121094,
      "learning_rate": 5.156739953882106e-08,
      "loss": 5.5036,
      "step": 23993
    },
    {
      "epoch": 0.9686717803794913,
      "grad_norm": 93.25995635986328,
      "learning_rate": 5.143486416501642e-08,
      "loss": 5.5161,
      "step": 23994
    },
    {
      "epoch": 0.968712151796528,
      "grad_norm": 87.10062408447266,
      "learning_rate": 5.130249888844363e-08,
      "loss": 3.8511,
      "step": 23995
    },
    {
      "epoch": 0.9687525232135648,
      "grad_norm": 72.31916046142578,
      "learning_rate": 5.117030371136644e-08,
      "loss": 4.1448,
      "step": 23996
    },
    {
      "epoch": 0.9687928946306016,
      "grad_norm": 88.21025085449219,
      "learning_rate": 5.103827863604527e-08,
      "loss": 5.964,
      "step": 23997
    },
    {
      "epoch": 0.9688332660476383,
      "grad_norm": 102.12450408935547,
      "learning_rate": 5.0906423664737195e-08,
      "loss": 3.5948,
      "step": 23998
    },
    {
      "epoch": 0.968873637464675,
      "grad_norm": 81.8555908203125,
      "learning_rate": 5.077473879969597e-08,
      "loss": 5.2666,
      "step": 23999
    },
    {
      "epoch": 0.9689140088817118,
      "grad_norm": 108.32312774658203,
      "learning_rate": 5.064322404317312e-08,
      "loss": 5.0768,
      "step": 24000
    },
    {
      "epoch": 0.9689543802987485,
      "grad_norm": 70.80984497070312,
      "learning_rate": 5.051187939741797e-08,
      "loss": 5.0407,
      "step": 24001
    },
    {
      "epoch": 0.9689947517157852,
      "grad_norm": 83.81584167480469,
      "learning_rate": 5.03807048646765e-08,
      "loss": 4.1663,
      "step": 24002
    },
    {
      "epoch": 0.9690351231328219,
      "grad_norm": 99.85438537597656,
      "learning_rate": 5.024970044719135e-08,
      "loss": 3.6131,
      "step": 24003
    },
    {
      "epoch": 0.9690754945498588,
      "grad_norm": 88.56026458740234,
      "learning_rate": 5.011886614720074e-08,
      "loss": 4.2088,
      "step": 24004
    },
    {
      "epoch": 0.9691158659668955,
      "grad_norm": 90.70729064941406,
      "learning_rate": 4.9988201966943986e-08,
      "loss": 5.8015,
      "step": 24005
    },
    {
      "epoch": 0.9691562373839322,
      "grad_norm": 118.97652435302734,
      "learning_rate": 4.9857707908652633e-08,
      "loss": 4.7818,
      "step": 24006
    },
    {
      "epoch": 0.9691966088009689,
      "grad_norm": 93.31233978271484,
      "learning_rate": 4.972738397456045e-08,
      "loss": 4.6969,
      "step": 24007
    },
    {
      "epoch": 0.9692369802180056,
      "grad_norm": 101.01014709472656,
      "learning_rate": 4.959723016689455e-08,
      "loss": 5.7762,
      "step": 24008
    },
    {
      "epoch": 0.9692773516350424,
      "grad_norm": 118.2716293334961,
      "learning_rate": 4.9467246487878703e-08,
      "loss": 4.4167,
      "step": 24009
    },
    {
      "epoch": 0.9693177230520791,
      "grad_norm": 148.33157348632812,
      "learning_rate": 4.9337432939737804e-08,
      "loss": 4.4184,
      "step": 24010
    },
    {
      "epoch": 0.9693580944691159,
      "grad_norm": 174.30520629882812,
      "learning_rate": 4.9207789524691186e-08,
      "loss": 4.8872,
      "step": 24011
    },
    {
      "epoch": 0.9693984658861526,
      "grad_norm": 100.55644226074219,
      "learning_rate": 4.907831624495374e-08,
      "loss": 6.2192,
      "step": 24012
    },
    {
      "epoch": 0.9694388373031894,
      "grad_norm": 103.49522399902344,
      "learning_rate": 4.894901310273925e-08,
      "loss": 4.368,
      "step": 24013
    },
    {
      "epoch": 0.9694792087202261,
      "grad_norm": 90.3116455078125,
      "learning_rate": 4.8819880100260396e-08,
      "loss": 5.1732,
      "step": 24014
    },
    {
      "epoch": 0.9695195801372628,
      "grad_norm": 88.1348648071289,
      "learning_rate": 4.869091723972319e-08,
      "loss": 4.1896,
      "step": 24015
    },
    {
      "epoch": 0.9695599515542995,
      "grad_norm": 95.4085922241211,
      "learning_rate": 4.8562124523333643e-08,
      "loss": 6.2624,
      "step": 24016
    },
    {
      "epoch": 0.9696003229713362,
      "grad_norm": 90.49087524414062,
      "learning_rate": 4.843350195329333e-08,
      "loss": 4.7121,
      "step": 24017
    },
    {
      "epoch": 0.9696406943883731,
      "grad_norm": 104.73362731933594,
      "learning_rate": 4.8305049531801596e-08,
      "loss": 4.65,
      "step": 24018
    },
    {
      "epoch": 0.9696810658054098,
      "grad_norm": 90.10771179199219,
      "learning_rate": 4.8176767261054466e-08,
      "loss": 4.7042,
      "step": 24019
    },
    {
      "epoch": 0.9697214372224465,
      "grad_norm": 113.11283874511719,
      "learning_rate": 4.804865514324575e-08,
      "loss": 5.6494,
      "step": 24020
    },
    {
      "epoch": 0.9697618086394832,
      "grad_norm": 88.64685821533203,
      "learning_rate": 4.7920713180565905e-08,
      "loss": 4.485,
      "step": 24021
    },
    {
      "epoch": 0.96980218005652,
      "grad_norm": 108.80067443847656,
      "learning_rate": 4.779294137520096e-08,
      "loss": 4.7818,
      "step": 24022
    },
    {
      "epoch": 0.9698425514735567,
      "grad_norm": 137.45399475097656,
      "learning_rate": 4.7665339729338066e-08,
      "loss": 3.4743,
      "step": 24023
    },
    {
      "epoch": 0.9698829228905934,
      "grad_norm": 132.56668090820312,
      "learning_rate": 4.753790824515658e-08,
      "loss": 3.933,
      "step": 24024
    },
    {
      "epoch": 0.9699232943076302,
      "grad_norm": 117.01780700683594,
      "learning_rate": 4.7410646924836986e-08,
      "loss": 6.0879,
      "step": 24025
    },
    {
      "epoch": 0.969963665724667,
      "grad_norm": 87.04484558105469,
      "learning_rate": 4.72835557705531e-08,
      "loss": 4.5939,
      "step": 24026
    },
    {
      "epoch": 0.9700040371417037,
      "grad_norm": 94.4834213256836,
      "learning_rate": 4.715663478447985e-08,
      "loss": 3.9653,
      "step": 24027
    },
    {
      "epoch": 0.9700444085587404,
      "grad_norm": 76.602783203125,
      "learning_rate": 4.7029883968786605e-08,
      "loss": 4.7501,
      "step": 24028
    },
    {
      "epoch": 0.9700847799757771,
      "grad_norm": 112.92937469482422,
      "learning_rate": 4.6903303325640527e-08,
      "loss": 4.3902,
      "step": 24029
    },
    {
      "epoch": 0.9701251513928139,
      "grad_norm": 90.81034088134766,
      "learning_rate": 4.677689285720544e-08,
      "loss": 6.1414,
      "step": 24030
    },
    {
      "epoch": 0.9701655228098506,
      "grad_norm": 96.54830932617188,
      "learning_rate": 4.665065256564405e-08,
      "loss": 4.6063,
      "step": 24031
    },
    {
      "epoch": 0.9702058942268874,
      "grad_norm": 83.57511901855469,
      "learning_rate": 4.652458245311242e-08,
      "loss": 5.6961,
      "step": 24032
    },
    {
      "epoch": 0.9702462656439241,
      "grad_norm": 125.74958038330078,
      "learning_rate": 4.6398682521768823e-08,
      "loss": 3.8925,
      "step": 24033
    },
    {
      "epoch": 0.9702866370609609,
      "grad_norm": 82.03631591796875,
      "learning_rate": 4.627295277376265e-08,
      "loss": 6.5341,
      "step": 24034
    },
    {
      "epoch": 0.9703270084779976,
      "grad_norm": 84.74950408935547,
      "learning_rate": 4.6147393211246615e-08,
      "loss": 4.4702,
      "step": 24035
    },
    {
      "epoch": 0.9703673798950343,
      "grad_norm": 89.65979766845703,
      "learning_rate": 4.6022003836365683e-08,
      "loss": 3.2066,
      "step": 24036
    },
    {
      "epoch": 0.970407751312071,
      "grad_norm": 84.55767059326172,
      "learning_rate": 4.5896784651264795e-08,
      "loss": 5.3102,
      "step": 24037
    },
    {
      "epoch": 0.9704481227291077,
      "grad_norm": 106.40353393554688,
      "learning_rate": 4.577173565808335e-08,
      "loss": 4.4414,
      "step": 24038
    },
    {
      "epoch": 0.9704884941461446,
      "grad_norm": 78.53417205810547,
      "learning_rate": 4.564685685896186e-08,
      "loss": 4.3327,
      "step": 24039
    },
    {
      "epoch": 0.9705288655631813,
      "grad_norm": 141.87158203125,
      "learning_rate": 4.552214825603307e-08,
      "loss": 5.9894,
      "step": 24040
    },
    {
      "epoch": 0.970569236980218,
      "grad_norm": 111.1119155883789,
      "learning_rate": 4.539760985143083e-08,
      "loss": 6.2165,
      "step": 24041
    },
    {
      "epoch": 0.9706096083972547,
      "grad_norm": 181.456298828125,
      "learning_rate": 4.527324164728342e-08,
      "loss": 4.8159,
      "step": 24042
    },
    {
      "epoch": 0.9706499798142915,
      "grad_norm": 86.7371826171875,
      "learning_rate": 4.514904364571804e-08,
      "loss": 4.4869,
      "step": 24043
    },
    {
      "epoch": 0.9706903512313282,
      "grad_norm": 95.70841979980469,
      "learning_rate": 4.502501584885743e-08,
      "loss": 4.4508,
      "step": 24044
    },
    {
      "epoch": 0.9707307226483649,
      "grad_norm": 82.45770263671875,
      "learning_rate": 4.490115825882213e-08,
      "loss": 3.7949,
      "step": 24045
    },
    {
      "epoch": 0.9707710940654017,
      "grad_norm": 85.89962005615234,
      "learning_rate": 4.4777470877730436e-08,
      "loss": 4.2395,
      "step": 24046
    },
    {
      "epoch": 0.9708114654824385,
      "grad_norm": 125.30494689941406,
      "learning_rate": 4.465395370769732e-08,
      "loss": 4.2045,
      "step": 24047
    },
    {
      "epoch": 0.9708518368994752,
      "grad_norm": 101.700439453125,
      "learning_rate": 4.4530606750833315e-08,
      "loss": 4.2275,
      "step": 24048
    },
    {
      "epoch": 0.9708922083165119,
      "grad_norm": 78.9072494506836,
      "learning_rate": 4.4407430009248964e-08,
      "loss": 3.3421,
      "step": 24049
    },
    {
      "epoch": 0.9709325797335486,
      "grad_norm": 91.87614440917969,
      "learning_rate": 4.428442348504924e-08,
      "loss": 3.3122,
      "step": 24050
    },
    {
      "epoch": 0.9709729511505854,
      "grad_norm": 87.09494018554688,
      "learning_rate": 4.4161587180338025e-08,
      "loss": 3.7864,
      "step": 24051
    },
    {
      "epoch": 0.9710133225676221,
      "grad_norm": 70.60575103759766,
      "learning_rate": 4.403892109721475e-08,
      "loss": 4.3729,
      "step": 24052
    },
    {
      "epoch": 0.9710536939846589,
      "grad_norm": 105.82415008544922,
      "learning_rate": 4.3916425237776615e-08,
      "loss": 5.4889,
      "step": 24053
    },
    {
      "epoch": 0.9710940654016956,
      "grad_norm": 75.14356231689453,
      "learning_rate": 4.3794099604118625e-08,
      "loss": 4.7645,
      "step": 24054
    },
    {
      "epoch": 0.9711344368187323,
      "grad_norm": 63.586029052734375,
      "learning_rate": 4.3671944198333536e-08,
      "loss": 3.0517,
      "step": 24055
    },
    {
      "epoch": 0.9711748082357691,
      "grad_norm": 106.62284088134766,
      "learning_rate": 4.354995902250747e-08,
      "loss": 4.487,
      "step": 24056
    },
    {
      "epoch": 0.9712151796528058,
      "grad_norm": 111.01109313964844,
      "learning_rate": 4.342814407872653e-08,
      "loss": 4.3298,
      "step": 24057
    },
    {
      "epoch": 0.9712555510698425,
      "grad_norm": 99.22830963134766,
      "learning_rate": 4.33064993690746e-08,
      "loss": 3.9219,
      "step": 24058
    },
    {
      "epoch": 0.9712959224868792,
      "grad_norm": 107.955078125,
      "learning_rate": 4.3185024895631145e-08,
      "loss": 4.7617,
      "step": 24059
    },
    {
      "epoch": 0.9713362939039161,
      "grad_norm": 83.13482666015625,
      "learning_rate": 4.3063720660473376e-08,
      "loss": 3.8331,
      "step": 24060
    },
    {
      "epoch": 0.9713766653209528,
      "grad_norm": 93.7205810546875,
      "learning_rate": 4.294258666567408e-08,
      "loss": 4.9368,
      "step": 24061
    },
    {
      "epoch": 0.9714170367379895,
      "grad_norm": 98.73521423339844,
      "learning_rate": 4.282162291330494e-08,
      "loss": 3.8451,
      "step": 24062
    },
    {
      "epoch": 0.9714574081550262,
      "grad_norm": 83.0536117553711,
      "learning_rate": 4.2700829405435405e-08,
      "loss": 4.1693,
      "step": 24063
    },
    {
      "epoch": 0.971497779572063,
      "grad_norm": 73.34379577636719,
      "learning_rate": 4.258020614412939e-08,
      "loss": 4.6379,
      "step": 24064
    },
    {
      "epoch": 0.9715381509890997,
      "grad_norm": 82.65321350097656,
      "learning_rate": 4.245975313144857e-08,
      "loss": 4.0719,
      "step": 24065
    },
    {
      "epoch": 0.9715785224061364,
      "grad_norm": 128.10292053222656,
      "learning_rate": 4.2339470369454626e-08,
      "loss": 6.7154,
      "step": 24066
    },
    {
      "epoch": 0.9716188938231732,
      "grad_norm": 143.4794158935547,
      "learning_rate": 4.221935786020259e-08,
      "loss": 5.7431,
      "step": 24067
    },
    {
      "epoch": 0.97165926524021,
      "grad_norm": 234.7710723876953,
      "learning_rate": 4.2099415605746376e-08,
      "loss": 4.1679,
      "step": 24068
    },
    {
      "epoch": 0.9716996366572467,
      "grad_norm": 98.6550521850586,
      "learning_rate": 4.1979643608136554e-08,
      "loss": 6.7165,
      "step": 24069
    },
    {
      "epoch": 0.9717400080742834,
      "grad_norm": 111.96217346191406,
      "learning_rate": 4.18600418694215e-08,
      "loss": 5.8964,
      "step": 24070
    },
    {
      "epoch": 0.9717803794913201,
      "grad_norm": 107.23155975341797,
      "learning_rate": 4.174061039164623e-08,
      "loss": 5.8659,
      "step": 24071
    },
    {
      "epoch": 0.9718207509083568,
      "grad_norm": 94.16155242919922,
      "learning_rate": 4.162134917685134e-08,
      "loss": 4.6117,
      "step": 24072
    },
    {
      "epoch": 0.9718611223253936,
      "grad_norm": 123.98189544677734,
      "learning_rate": 4.150225822707743e-08,
      "loss": 3.399,
      "step": 24073
    },
    {
      "epoch": 0.9719014937424304,
      "grad_norm": 99.64453125,
      "learning_rate": 4.1383337544359526e-08,
      "loss": 5.4102,
      "step": 24074
    },
    {
      "epoch": 0.9719418651594671,
      "grad_norm": 103.93753051757812,
      "learning_rate": 4.126458713073267e-08,
      "loss": 3.8838,
      "step": 24075
    },
    {
      "epoch": 0.9719822365765038,
      "grad_norm": 96.12177276611328,
      "learning_rate": 4.114600698822524e-08,
      "loss": 3.8199,
      "step": 24076
    },
    {
      "epoch": 0.9720226079935406,
      "grad_norm": 116.56879425048828,
      "learning_rate": 4.102759711886561e-08,
      "loss": 6.2267,
      "step": 24077
    },
    {
      "epoch": 0.9720629794105773,
      "grad_norm": 97.13323974609375,
      "learning_rate": 4.090935752467773e-08,
      "loss": 5.4473,
      "step": 24078
    },
    {
      "epoch": 0.972103350827614,
      "grad_norm": 74.80702209472656,
      "learning_rate": 4.079128820768552e-08,
      "loss": 4.4409,
      "step": 24079
    },
    {
      "epoch": 0.9721437222446507,
      "grad_norm": 111.0179443359375,
      "learning_rate": 4.067338916990404e-08,
      "loss": 5.6645,
      "step": 24080
    },
    {
      "epoch": 0.9721840936616876,
      "grad_norm": 79.71233367919922,
      "learning_rate": 4.055566041335168e-08,
      "loss": 3.7897,
      "step": 24081
    },
    {
      "epoch": 0.9722244650787243,
      "grad_norm": 128.7581329345703,
      "learning_rate": 4.0438101940040166e-08,
      "loss": 5.1547,
      "step": 24082
    },
    {
      "epoch": 0.972264836495761,
      "grad_norm": 161.59832763671875,
      "learning_rate": 4.03207137519801e-08,
      "loss": 5.6681,
      "step": 24083
    },
    {
      "epoch": 0.9723052079127977,
      "grad_norm": 120.30055236816406,
      "learning_rate": 4.0203495851177666e-08,
      "loss": 5.7158,
      "step": 24084
    },
    {
      "epoch": 0.9723455793298345,
      "grad_norm": 250.83615112304688,
      "learning_rate": 4.008644823963903e-08,
      "loss": 6.1818,
      "step": 24085
    },
    {
      "epoch": 0.9723859507468712,
      "grad_norm": 94.37122344970703,
      "learning_rate": 3.99695709193626e-08,
      "loss": 6.6574,
      "step": 24086
    },
    {
      "epoch": 0.972426322163908,
      "grad_norm": 122.85588073730469,
      "learning_rate": 3.985286389234899e-08,
      "loss": 5.2926,
      "step": 24087
    },
    {
      "epoch": 0.9724666935809447,
      "grad_norm": 83.07066345214844,
      "learning_rate": 3.973632716059217e-08,
      "loss": 5.3738,
      "step": 24088
    },
    {
      "epoch": 0.9725070649979815,
      "grad_norm": 166.28948974609375,
      "learning_rate": 3.961996072608609e-08,
      "loss": 5.2981,
      "step": 24089
    },
    {
      "epoch": 0.9725474364150182,
      "grad_norm": 97.661376953125,
      "learning_rate": 3.950376459081917e-08,
      "loss": 5.7121,
      "step": 24090
    },
    {
      "epoch": 0.9725878078320549,
      "grad_norm": 82.16880798339844,
      "learning_rate": 3.9387738756777594e-08,
      "loss": 6.6028,
      "step": 24091
    },
    {
      "epoch": 0.9726281792490916,
      "grad_norm": 73.79237365722656,
      "learning_rate": 3.927188322594755e-08,
      "loss": 4.3975,
      "step": 24092
    },
    {
      "epoch": 0.9726685506661283,
      "grad_norm": 63.94937515258789,
      "learning_rate": 3.9156198000307453e-08,
      "loss": 5.0407,
      "step": 24093
    },
    {
      "epoch": 0.9727089220831652,
      "grad_norm": 117.34278106689453,
      "learning_rate": 3.904068308183573e-08,
      "loss": 6.8042,
      "step": 24094
    },
    {
      "epoch": 0.9727492935002019,
      "grad_norm": 100.9996109008789,
      "learning_rate": 3.892533847250857e-08,
      "loss": 4.284,
      "step": 24095
    },
    {
      "epoch": 0.9727896649172386,
      "grad_norm": 129.22727966308594,
      "learning_rate": 3.881016417429773e-08,
      "loss": 2.7724,
      "step": 24096
    },
    {
      "epoch": 0.9728300363342753,
      "grad_norm": 86.96748352050781,
      "learning_rate": 3.869516018917052e-08,
      "loss": 5.7701,
      "step": 24097
    },
    {
      "epoch": 0.9728704077513121,
      "grad_norm": 91.74299621582031,
      "learning_rate": 3.858032651909538e-08,
      "loss": 3.5921,
      "step": 24098
    },
    {
      "epoch": 0.9729107791683488,
      "grad_norm": 89.91734313964844,
      "learning_rate": 3.846566316603628e-08,
      "loss": 5.1102,
      "step": 24099
    },
    {
      "epoch": 0.9729511505853855,
      "grad_norm": 99.06015014648438,
      "learning_rate": 3.8351170131951664e-08,
      "loss": 6.3787,
      "step": 24100
    },
    {
      "epoch": 0.9729915220024223,
      "grad_norm": 116.12855529785156,
      "learning_rate": 3.823684741879885e-08,
      "loss": 6.3307,
      "step": 24101
    },
    {
      "epoch": 0.9730318934194591,
      "grad_norm": 67.19268798828125,
      "learning_rate": 3.812269502853516e-08,
      "loss": 5.001,
      "step": 24102
    },
    {
      "epoch": 0.9730722648364958,
      "grad_norm": 91.4465560913086,
      "learning_rate": 3.800871296311015e-08,
      "loss": 4.9036,
      "step": 24103
    },
    {
      "epoch": 0.9731126362535325,
      "grad_norm": 91.63272857666016,
      "learning_rate": 3.789490122447337e-08,
      "loss": 4.9625,
      "step": 24104
    },
    {
      "epoch": 0.9731530076705692,
      "grad_norm": 68.67870330810547,
      "learning_rate": 3.7781259814569926e-08,
      "loss": 5.177,
      "step": 24105
    },
    {
      "epoch": 0.973193379087606,
      "grad_norm": 122.35575866699219,
      "learning_rate": 3.766778873534383e-08,
      "loss": 3.9924,
      "step": 24106
    },
    {
      "epoch": 0.9732337505046427,
      "grad_norm": 91.37310028076172,
      "learning_rate": 3.7554487988734624e-08,
      "loss": 6.0243,
      "step": 24107
    },
    {
      "epoch": 0.9732741219216795,
      "grad_norm": 153.5599365234375,
      "learning_rate": 3.744135757668077e-08,
      "loss": 4.579,
      "step": 24108
    },
    {
      "epoch": 0.9733144933387162,
      "grad_norm": 109.75706481933594,
      "learning_rate": 3.7328397501114055e-08,
      "loss": 3.9771,
      "step": 24109
    },
    {
      "epoch": 0.973354864755753,
      "grad_norm": 101.4853744506836,
      "learning_rate": 3.7215607763967375e-08,
      "loss": 7.3582,
      "step": 24110
    },
    {
      "epoch": 0.9733952361727897,
      "grad_norm": 80.49604797363281,
      "learning_rate": 3.71029883671703e-08,
      "loss": 5.3339,
      "step": 24111
    },
    {
      "epoch": 0.9734356075898264,
      "grad_norm": 199.89039611816406,
      "learning_rate": 3.6990539312645736e-08,
      "loss": 5.7357,
      "step": 24112
    },
    {
      "epoch": 0.9734759790068631,
      "grad_norm": 101.48125457763672,
      "learning_rate": 3.6878260602317695e-08,
      "loss": 5.6721,
      "step": 24113
    },
    {
      "epoch": 0.9735163504238998,
      "grad_norm": 70.97824096679688,
      "learning_rate": 3.6766152238106866e-08,
      "loss": 3.7818,
      "step": 24114
    },
    {
      "epoch": 0.9735567218409367,
      "grad_norm": 142.77932739257812,
      "learning_rate": 3.6654214221928383e-08,
      "loss": 3.983,
      "step": 24115
    },
    {
      "epoch": 0.9735970932579734,
      "grad_norm": 102.61634063720703,
      "learning_rate": 3.654244655569739e-08,
      "loss": 4.1372,
      "step": 24116
    },
    {
      "epoch": 0.9736374646750101,
      "grad_norm": 84.08287048339844,
      "learning_rate": 3.643084924132345e-08,
      "loss": 4.6383,
      "step": 24117
    },
    {
      "epoch": 0.9736778360920468,
      "grad_norm": 139.06532287597656,
      "learning_rate": 3.6319422280715055e-08,
      "loss": 4.3914,
      "step": 24118
    },
    {
      "epoch": 0.9737182075090836,
      "grad_norm": 138.65087890625,
      "learning_rate": 3.620816567577845e-08,
      "loss": 4.6969,
      "step": 24119
    },
    {
      "epoch": 0.9737585789261203,
      "grad_norm": 160.8650665283203,
      "learning_rate": 3.609707942841545e-08,
      "loss": 5.407,
      "step": 24120
    },
    {
      "epoch": 0.973798950343157,
      "grad_norm": 90.7686538696289,
      "learning_rate": 3.598616354052342e-08,
      "loss": 4.6638,
      "step": 24121
    },
    {
      "epoch": 0.9738393217601938,
      "grad_norm": 78.42813110351562,
      "learning_rate": 3.587541801400196e-08,
      "loss": 2.1089,
      "step": 24122
    },
    {
      "epoch": 0.9738796931772306,
      "grad_norm": 126.15559387207031,
      "learning_rate": 3.576484285074178e-08,
      "loss": 5.456,
      "step": 24123
    },
    {
      "epoch": 0.9739200645942673,
      "grad_norm": 133.00408935546875,
      "learning_rate": 3.5654438052635795e-08,
      "loss": 4.6427,
      "step": 24124
    },
    {
      "epoch": 0.973960436011304,
      "grad_norm": 123.52645874023438,
      "learning_rate": 3.554420362156919e-08,
      "loss": 5.4581,
      "step": 24125
    },
    {
      "epoch": 0.9740008074283407,
      "grad_norm": 81.36222839355469,
      "learning_rate": 3.5434139559429315e-08,
      "loss": 6.0865,
      "step": 24126
    },
    {
      "epoch": 0.9740411788453774,
      "grad_norm": 109.41683197021484,
      "learning_rate": 3.53242458680958e-08,
      "loss": 3.2541,
      "step": 24127
    },
    {
      "epoch": 0.9740815502624142,
      "grad_norm": 82.89448547363281,
      "learning_rate": 3.521452254944935e-08,
      "loss": 5.3908,
      "step": 24128
    },
    {
      "epoch": 0.974121921679451,
      "grad_norm": 86.91744995117188,
      "learning_rate": 3.510496960536403e-08,
      "loss": 5.662,
      "step": 24129
    },
    {
      "epoch": 0.9741622930964877,
      "grad_norm": 151.94540405273438,
      "learning_rate": 3.4995587037715e-08,
      "loss": 5.5873,
      "step": 24130
    },
    {
      "epoch": 0.9742026645135244,
      "grad_norm": 103.04182434082031,
      "learning_rate": 3.488637484837076e-08,
      "loss": 4.1256,
      "step": 24131
    },
    {
      "epoch": 0.9742430359305612,
      "grad_norm": 111.53389739990234,
      "learning_rate": 3.4777333039199836e-08,
      "loss": 4.9018,
      "step": 24132
    },
    {
      "epoch": 0.9742834073475979,
      "grad_norm": 90.73957061767578,
      "learning_rate": 3.466846161206516e-08,
      "loss": 5.2725,
      "step": 24133
    },
    {
      "epoch": 0.9743237787646346,
      "grad_norm": 73.42420959472656,
      "learning_rate": 3.45597605688297e-08,
      "loss": 4.8405,
      "step": 24134
    },
    {
      "epoch": 0.9743641501816713,
      "grad_norm": 101.86048889160156,
      "learning_rate": 3.445122991135197e-08,
      "loss": 4.3542,
      "step": 24135
    },
    {
      "epoch": 0.9744045215987082,
      "grad_norm": 85.69441986083984,
      "learning_rate": 3.434286964148603e-08,
      "loss": 5.8211,
      "step": 24136
    },
    {
      "epoch": 0.9744448930157449,
      "grad_norm": 74.34864807128906,
      "learning_rate": 3.4234679761085965e-08,
      "loss": 3.6646,
      "step": 24137
    },
    {
      "epoch": 0.9744852644327816,
      "grad_norm": 110.2194595336914,
      "learning_rate": 3.412666027200029e-08,
      "loss": 5.0781,
      "step": 24138
    },
    {
      "epoch": 0.9745256358498183,
      "grad_norm": 82.64154052734375,
      "learning_rate": 3.401881117607864e-08,
      "loss": 4.5243,
      "step": 24139
    },
    {
      "epoch": 0.974566007266855,
      "grad_norm": 136.66587829589844,
      "learning_rate": 3.3911132475161754e-08,
      "loss": 4.4393,
      "step": 24140
    },
    {
      "epoch": 0.9746063786838918,
      "grad_norm": 207.97021484375,
      "learning_rate": 3.380362417109151e-08,
      "loss": 4.9627,
      "step": 24141
    },
    {
      "epoch": 0.9746467501009285,
      "grad_norm": 95.60173797607422,
      "learning_rate": 3.3696286265707537e-08,
      "loss": 5.4353,
      "step": 24142
    },
    {
      "epoch": 0.9746871215179653,
      "grad_norm": 90.61688232421875,
      "learning_rate": 3.358911876084392e-08,
      "loss": 4.2894,
      "step": 24143
    },
    {
      "epoch": 0.974727492935002,
      "grad_norm": 99.85952758789062,
      "learning_rate": 3.348212165833253e-08,
      "loss": 5.0376,
      "step": 24144
    },
    {
      "epoch": 0.9747678643520388,
      "grad_norm": 220.28660583496094,
      "learning_rate": 3.3375294960004134e-08,
      "loss": 3.2534,
      "step": 24145
    },
    {
      "epoch": 0.9748082357690755,
      "grad_norm": 79.58822631835938,
      "learning_rate": 3.326863866768393e-08,
      "loss": 4.6662,
      "step": 24146
    },
    {
      "epoch": 0.9748486071861122,
      "grad_norm": 73.17507934570312,
      "learning_rate": 3.316215278319601e-08,
      "loss": 4.1732,
      "step": 24147
    },
    {
      "epoch": 0.9748889786031489,
      "grad_norm": 83.65029907226562,
      "learning_rate": 3.305583730836115e-08,
      "loss": 5.6881,
      "step": 24148
    },
    {
      "epoch": 0.9749293500201857,
      "grad_norm": 66.54222106933594,
      "learning_rate": 3.294969224499789e-08,
      "loss": 4.451,
      "step": 24149
    },
    {
      "epoch": 0.9749697214372225,
      "grad_norm": 258.3258972167969,
      "learning_rate": 3.284371759491922e-08,
      "loss": 3.9919,
      "step": 24150
    },
    {
      "epoch": 0.9750100928542592,
      "grad_norm": 338.6152648925781,
      "learning_rate": 3.273791335993815e-08,
      "loss": 4.1991,
      "step": 24151
    },
    {
      "epoch": 0.9750504642712959,
      "grad_norm": 88.56181335449219,
      "learning_rate": 3.263227954186321e-08,
      "loss": 5.3372,
      "step": 24152
    },
    {
      "epoch": 0.9750908356883327,
      "grad_norm": 63.3147087097168,
      "learning_rate": 3.252681614250075e-08,
      "loss": 4.3589,
      "step": 24153
    },
    {
      "epoch": 0.9751312071053694,
      "grad_norm": 125.94698333740234,
      "learning_rate": 3.242152316365488e-08,
      "loss": 5.8519,
      "step": 24154
    },
    {
      "epoch": 0.9751715785224061,
      "grad_norm": 99.57487487792969,
      "learning_rate": 3.231640060712416e-08,
      "loss": 4.6718,
      "step": 24155
    },
    {
      "epoch": 0.9752119499394428,
      "grad_norm": 93.68817138671875,
      "learning_rate": 3.221144847470714e-08,
      "loss": 5.4585,
      "step": 24156
    },
    {
      "epoch": 0.9752523213564797,
      "grad_norm": 180.951904296875,
      "learning_rate": 3.210666676819796e-08,
      "loss": 6.7272,
      "step": 24157
    },
    {
      "epoch": 0.9752926927735164,
      "grad_norm": 112.98075866699219,
      "learning_rate": 3.20020554893874e-08,
      "loss": 2.5501,
      "step": 24158
    },
    {
      "epoch": 0.9753330641905531,
      "grad_norm": 90.85069274902344,
      "learning_rate": 3.1897614640066245e-08,
      "loss": 5.4482,
      "step": 24159
    },
    {
      "epoch": 0.9753734356075898,
      "grad_norm": 96.05306243896484,
      "learning_rate": 3.179334422201752e-08,
      "loss": 5.4621,
      "step": 24160
    },
    {
      "epoch": 0.9754138070246265,
      "grad_norm": 104.74532318115234,
      "learning_rate": 3.1689244237025354e-08,
      "loss": 4.6768,
      "step": 24161
    },
    {
      "epoch": 0.9754541784416633,
      "grad_norm": 96.4226303100586,
      "learning_rate": 3.158531468686943e-08,
      "loss": 3.9013,
      "step": 24162
    },
    {
      "epoch": 0.9754945498587,
      "grad_norm": 80.80107879638672,
      "learning_rate": 3.148155557332722e-08,
      "loss": 4.8735,
      "step": 24163
    },
    {
      "epoch": 0.9755349212757368,
      "grad_norm": 84.05427551269531,
      "learning_rate": 3.137796689817063e-08,
      "loss": 4.6973,
      "step": 24164
    },
    {
      "epoch": 0.9755752926927735,
      "grad_norm": 136.84625244140625,
      "learning_rate": 3.127454866317381e-08,
      "loss": 4.7441,
      "step": 24165
    },
    {
      "epoch": 0.9756156641098103,
      "grad_norm": 122.16694641113281,
      "learning_rate": 3.117130087010312e-08,
      "loss": 5.3549,
      "step": 24166
    },
    {
      "epoch": 0.975656035526847,
      "grad_norm": 83.23509216308594,
      "learning_rate": 3.106822352072492e-08,
      "loss": 4.5155,
      "step": 24167
    },
    {
      "epoch": 0.9756964069438837,
      "grad_norm": 76.38314819335938,
      "learning_rate": 3.096531661680002e-08,
      "loss": 2.849,
      "step": 24168
    },
    {
      "epoch": 0.9757367783609204,
      "grad_norm": 90.52787780761719,
      "learning_rate": 3.086258016008925e-08,
      "loss": 5.2441,
      "step": 24169
    },
    {
      "epoch": 0.9757771497779572,
      "grad_norm": 111.63871002197266,
      "learning_rate": 3.076001415234897e-08,
      "loss": 4.4244,
      "step": 24170
    },
    {
      "epoch": 0.975817521194994,
      "grad_norm": 89.27055358886719,
      "learning_rate": 3.065761859533223e-08,
      "loss": 3.5227,
      "step": 24171
    },
    {
      "epoch": 0.9758578926120307,
      "grad_norm": 77.46955108642578,
      "learning_rate": 3.0555393490790955e-08,
      "loss": 5.548,
      "step": 24172
    },
    {
      "epoch": 0.9758982640290674,
      "grad_norm": 133.82635498046875,
      "learning_rate": 3.045333884047152e-08,
      "loss": 4.6569,
      "step": 24173
    },
    {
      "epoch": 0.9759386354461042,
      "grad_norm": 91.40788269042969,
      "learning_rate": 3.035145464611922e-08,
      "loss": 5.7294,
      "step": 24174
    },
    {
      "epoch": 0.9759790068631409,
      "grad_norm": 79.70791625976562,
      "learning_rate": 3.0249740909477076e-08,
      "loss": 4.476,
      "step": 24175
    },
    {
      "epoch": 0.9760193782801776,
      "grad_norm": 176.5110626220703,
      "learning_rate": 3.0148197632282606e-08,
      "loss": 5.0852,
      "step": 24176
    },
    {
      "epoch": 0.9760597496972143,
      "grad_norm": 116.7760009765625,
      "learning_rate": 3.004682481627219e-08,
      "loss": 5.7716,
      "step": 24177
    },
    {
      "epoch": 0.9761001211142512,
      "grad_norm": 103.86564636230469,
      "learning_rate": 2.994562246318e-08,
      "loss": 5.097,
      "step": 24178
    },
    {
      "epoch": 0.9761404925312879,
      "grad_norm": 160.025634765625,
      "learning_rate": 2.984459057473688e-08,
      "loss": 4.5147,
      "step": 24179
    },
    {
      "epoch": 0.9761808639483246,
      "grad_norm": 122.74090576171875,
      "learning_rate": 2.9743729152668098e-08,
      "loss": 4.3472,
      "step": 24180
    },
    {
      "epoch": 0.9762212353653613,
      "grad_norm": 104.02387237548828,
      "learning_rate": 2.964303819869896e-08,
      "loss": 3.8453,
      "step": 24181
    },
    {
      "epoch": 0.976261606782398,
      "grad_norm": 80.55073547363281,
      "learning_rate": 2.954251771455252e-08,
      "loss": 3.4723,
      "step": 24182
    },
    {
      "epoch": 0.9763019781994348,
      "grad_norm": 103.1366958618164,
      "learning_rate": 2.9442167701946302e-08,
      "loss": 5.0412,
      "step": 24183
    },
    {
      "epoch": 0.9763423496164715,
      "grad_norm": 89.7015380859375,
      "learning_rate": 2.9341988162595593e-08,
      "loss": 5.7676,
      "step": 24184
    },
    {
      "epoch": 0.9763827210335083,
      "grad_norm": 142.3585205078125,
      "learning_rate": 2.9241979098213468e-08,
      "loss": 3.9153,
      "step": 24185
    },
    {
      "epoch": 0.976423092450545,
      "grad_norm": 94.66627502441406,
      "learning_rate": 2.9142140510510786e-08,
      "loss": 5.0944,
      "step": 24186
    },
    {
      "epoch": 0.9764634638675818,
      "grad_norm": 98.68775939941406,
      "learning_rate": 2.9042472401193954e-08,
      "loss": 4.5959,
      "step": 24187
    },
    {
      "epoch": 0.9765038352846185,
      "grad_norm": 89.26831817626953,
      "learning_rate": 2.894297477196606e-08,
      "loss": 3.2989,
      "step": 24188
    },
    {
      "epoch": 0.9765442067016552,
      "grad_norm": 130.61399841308594,
      "learning_rate": 2.884364762453018e-08,
      "loss": 4.7497,
      "step": 24189
    },
    {
      "epoch": 0.9765845781186919,
      "grad_norm": 104.56414031982422,
      "learning_rate": 2.874449096058274e-08,
      "loss": 4.4406,
      "step": 24190
    },
    {
      "epoch": 0.9766249495357288,
      "grad_norm": 85.24852752685547,
      "learning_rate": 2.8645504781820153e-08,
      "loss": 5.341,
      "step": 24191
    },
    {
      "epoch": 0.9766653209527655,
      "grad_norm": 113.83270263671875,
      "learning_rate": 2.854668908993552e-08,
      "loss": 4.6411,
      "step": 24192
    },
    {
      "epoch": 0.9767056923698022,
      "grad_norm": 150.54638671875,
      "learning_rate": 2.8448043886616373e-08,
      "loss": 4.7454,
      "step": 24193
    },
    {
      "epoch": 0.9767460637868389,
      "grad_norm": 69.10486602783203,
      "learning_rate": 2.834956917355136e-08,
      "loss": 4.3622,
      "step": 24194
    },
    {
      "epoch": 0.9767864352038756,
      "grad_norm": 112.7175521850586,
      "learning_rate": 2.8251264952422475e-08,
      "loss": 5.3356,
      "step": 24195
    },
    {
      "epoch": 0.9768268066209124,
      "grad_norm": 102.03553009033203,
      "learning_rate": 2.8153131224911702e-08,
      "loss": 3.6963,
      "step": 24196
    },
    {
      "epoch": 0.9768671780379491,
      "grad_norm": 73.24832153320312,
      "learning_rate": 2.8055167992696587e-08,
      "loss": 5.1752,
      "step": 24197
    },
    {
      "epoch": 0.9769075494549859,
      "grad_norm": 167.49644470214844,
      "learning_rate": 2.7957375257451347e-08,
      "loss": 3.2513,
      "step": 24198
    },
    {
      "epoch": 0.9769479208720226,
      "grad_norm": 97.8916244506836,
      "learning_rate": 2.785975302084909e-08,
      "loss": 4.6567,
      "step": 24199
    },
    {
      "epoch": 0.9769882922890594,
      "grad_norm": 104.80640411376953,
      "learning_rate": 2.776230128455848e-08,
      "loss": 4.6204,
      "step": 24200
    },
    {
      "epoch": 0.9770286637060961,
      "grad_norm": 106.83808898925781,
      "learning_rate": 2.766502005024485e-08,
      "loss": 4.9004,
      "step": 24201
    },
    {
      "epoch": 0.9770690351231328,
      "grad_norm": 87.9694595336914,
      "learning_rate": 2.7567909319572428e-08,
      "loss": 3.4916,
      "step": 24202
    },
    {
      "epoch": 0.9771094065401695,
      "grad_norm": 94.17845916748047,
      "learning_rate": 2.7470969094203214e-08,
      "loss": 4.4822,
      "step": 24203
    },
    {
      "epoch": 0.9771497779572063,
      "grad_norm": 142.45877075195312,
      "learning_rate": 2.7374199375791443e-08,
      "loss": 3.2398,
      "step": 24204
    },
    {
      "epoch": 0.9771901493742431,
      "grad_norm": 98.39291381835938,
      "learning_rate": 2.7277600165992457e-08,
      "loss": 6.3588,
      "step": 24205
    },
    {
      "epoch": 0.9772305207912798,
      "grad_norm": 105.4410629272461,
      "learning_rate": 2.718117146645938e-08,
      "loss": 5.3086,
      "step": 24206
    },
    {
      "epoch": 0.9772708922083165,
      "grad_norm": 75.21265411376953,
      "learning_rate": 2.7084913278840887e-08,
      "loss": 5.2649,
      "step": 24207
    },
    {
      "epoch": 0.9773112636253533,
      "grad_norm": 112.712646484375,
      "learning_rate": 2.698882560478011e-08,
      "loss": 5.4779,
      "step": 24208
    },
    {
      "epoch": 0.97735163504239,
      "grad_norm": 100.67095947265625,
      "learning_rate": 2.6892908445923514e-08,
      "loss": 4.0131,
      "step": 24209
    },
    {
      "epoch": 0.9773920064594267,
      "grad_norm": 87.74022674560547,
      "learning_rate": 2.679716180390757e-08,
      "loss": 4.6632,
      "step": 24210
    },
    {
      "epoch": 0.9774323778764634,
      "grad_norm": 87.10667419433594,
      "learning_rate": 2.6701585680372068e-08,
      "loss": 5.0526,
      "step": 24211
    },
    {
      "epoch": 0.9774727492935003,
      "grad_norm": 89.71870422363281,
      "learning_rate": 2.660618007695015e-08,
      "loss": 4.425,
      "step": 24212
    },
    {
      "epoch": 0.977513120710537,
      "grad_norm": 120.1150131225586,
      "learning_rate": 2.6510944995272736e-08,
      "loss": 5.0766,
      "step": 24213
    },
    {
      "epoch": 0.9775534921275737,
      "grad_norm": 120.7522964477539,
      "learning_rate": 2.6415880436968523e-08,
      "loss": 5.0083,
      "step": 24214
    },
    {
      "epoch": 0.9775938635446104,
      "grad_norm": 86.53372955322266,
      "learning_rate": 2.6320986403662875e-08,
      "loss": 5.6042,
      "step": 24215
    },
    {
      "epoch": 0.9776342349616471,
      "grad_norm": 85.0262680053711,
      "learning_rate": 2.6226262896978937e-08,
      "loss": 4.5647,
      "step": 24216
    },
    {
      "epoch": 0.9776746063786839,
      "grad_norm": 105.94287109375,
      "learning_rate": 2.613170991853431e-08,
      "loss": 5.8371,
      "step": 24217
    },
    {
      "epoch": 0.9777149777957206,
      "grad_norm": 109.7455062866211,
      "learning_rate": 2.6037327469946583e-08,
      "loss": 5.1745,
      "step": 24218
    },
    {
      "epoch": 0.9777553492127574,
      "grad_norm": 95.37692260742188,
      "learning_rate": 2.5943115552831134e-08,
      "loss": 3.906,
      "step": 24219
    },
    {
      "epoch": 0.9777957206297941,
      "grad_norm": 77.05304718017578,
      "learning_rate": 2.5849074168795564e-08,
      "loss": 4.1023,
      "step": 24220
    },
    {
      "epoch": 0.9778360920468309,
      "grad_norm": 107.4511489868164,
      "learning_rate": 2.5755203319450806e-08,
      "loss": 5.222,
      "step": 24221
    },
    {
      "epoch": 0.9778764634638676,
      "grad_norm": 92.94122314453125,
      "learning_rate": 2.5661503006400024e-08,
      "loss": 5.8899,
      "step": 24222
    },
    {
      "epoch": 0.9779168348809043,
      "grad_norm": 95.20179748535156,
      "learning_rate": 2.5567973231245268e-08,
      "loss": 3.1055,
      "step": 24223
    },
    {
      "epoch": 0.977957206297941,
      "grad_norm": 90.2045669555664,
      "learning_rate": 2.5474613995587482e-08,
      "loss": 3.6423,
      "step": 24224
    },
    {
      "epoch": 0.9779975777149778,
      "grad_norm": 104.23688507080078,
      "learning_rate": 2.5381425301019837e-08,
      "loss": 5.9017,
      "step": 24225
    },
    {
      "epoch": 0.9780379491320146,
      "grad_norm": 194.7521514892578,
      "learning_rate": 2.5288407149138827e-08,
      "loss": 5.2763,
      "step": 24226
    },
    {
      "epoch": 0.9780783205490513,
      "grad_norm": 89.94068908691406,
      "learning_rate": 2.5195559541533188e-08,
      "loss": 3.5864,
      "step": 24227
    },
    {
      "epoch": 0.978118691966088,
      "grad_norm": 100.16913604736328,
      "learning_rate": 2.5102882479790535e-08,
      "loss": 4.6612,
      "step": 24228
    },
    {
      "epoch": 0.9781590633831247,
      "grad_norm": 91.4625015258789,
      "learning_rate": 2.501037596549516e-08,
      "loss": 5.0112,
      "step": 24229
    },
    {
      "epoch": 0.9781994348001615,
      "grad_norm": 90.48236083984375,
      "learning_rate": 2.4918040000229127e-08,
      "loss": 3.9931,
      "step": 24230
    },
    {
      "epoch": 0.9782398062171982,
      "grad_norm": 91.87263488769531,
      "learning_rate": 2.4825874585571174e-08,
      "loss": 3.3542,
      "step": 24231
    },
    {
      "epoch": 0.9782801776342349,
      "grad_norm": 116.25688934326172,
      "learning_rate": 2.473387972309782e-08,
      "loss": 6.226,
      "step": 24232
    },
    {
      "epoch": 0.9783205490512717,
      "grad_norm": 97.53034973144531,
      "learning_rate": 2.464205541438003e-08,
      "loss": 6.2027,
      "step": 24233
    },
    {
      "epoch": 0.9783609204683085,
      "grad_norm": 183.87094116210938,
      "learning_rate": 2.4550401660989877e-08,
      "loss": 3.9162,
      "step": 24234
    },
    {
      "epoch": 0.9784012918853452,
      "grad_norm": 162.63697814941406,
      "learning_rate": 2.4458918464493887e-08,
      "loss": 5.0221,
      "step": 24235
    },
    {
      "epoch": 0.9784416633023819,
      "grad_norm": 120.1479263305664,
      "learning_rate": 2.4367605826454143e-08,
      "loss": 3.1525,
      "step": 24236
    },
    {
      "epoch": 0.9784820347194186,
      "grad_norm": 77.49336242675781,
      "learning_rate": 2.4276463748434954e-08,
      "loss": 3.5614,
      "step": 24237
    },
    {
      "epoch": 0.9785224061364554,
      "grad_norm": 164.81723022460938,
      "learning_rate": 2.4185492231992847e-08,
      "loss": 6.2065,
      "step": 24238
    },
    {
      "epoch": 0.9785627775534921,
      "grad_norm": 84.08087921142578,
      "learning_rate": 2.4094691278684357e-08,
      "loss": 3.9862,
      "step": 24239
    },
    {
      "epoch": 0.9786031489705289,
      "grad_norm": 95.40733337402344,
      "learning_rate": 2.4004060890060464e-08,
      "loss": 3.9957,
      "step": 24240
    },
    {
      "epoch": 0.9786435203875656,
      "grad_norm": 108.7059097290039,
      "learning_rate": 2.3913601067671044e-08,
      "loss": 4.7921,
      "step": 24241
    },
    {
      "epoch": 0.9786838918046024,
      "grad_norm": 580.0824584960938,
      "learning_rate": 2.3823311813063744e-08,
      "loss": 4.3874,
      "step": 24242
    },
    {
      "epoch": 0.9787242632216391,
      "grad_norm": 214.73724365234375,
      "learning_rate": 2.3733193127781773e-08,
      "loss": 5.0376,
      "step": 24243
    },
    {
      "epoch": 0.9787646346386758,
      "grad_norm": 8447.2333984375,
      "learning_rate": 2.3643245013366124e-08,
      "loss": 4.1139,
      "step": 24244
    },
    {
      "epoch": 0.9788050060557125,
      "grad_norm": 83.57733154296875,
      "learning_rate": 2.3553467471354452e-08,
      "loss": 3.4461,
      "step": 24245
    },
    {
      "epoch": 0.9788453774727492,
      "grad_norm": 69.49048614501953,
      "learning_rate": 2.346386050328109e-08,
      "loss": 5.624,
      "step": 24246
    },
    {
      "epoch": 0.9788857488897861,
      "grad_norm": 93.02165222167969,
      "learning_rate": 2.3374424110679252e-08,
      "loss": 4.6326,
      "step": 24247
    },
    {
      "epoch": 0.9789261203068228,
      "grad_norm": 72.49433898925781,
      "learning_rate": 2.3285158295077715e-08,
      "loss": 4.6435,
      "step": 24248
    },
    {
      "epoch": 0.9789664917238595,
      "grad_norm": 587.8882446289062,
      "learning_rate": 2.3196063058003038e-08,
      "loss": 4.2358,
      "step": 24249
    },
    {
      "epoch": 0.9790068631408962,
      "grad_norm": 90.74046325683594,
      "learning_rate": 2.3107138400977335e-08,
      "loss": 7.4422,
      "step": 24250
    },
    {
      "epoch": 0.979047234557933,
      "grad_norm": 76.84210205078125,
      "learning_rate": 2.3018384325522724e-08,
      "loss": 4.3384,
      "step": 24251
    },
    {
      "epoch": 0.9790876059749697,
      "grad_norm": 58.54420471191406,
      "learning_rate": 2.2929800833155768e-08,
      "loss": 4.4027,
      "step": 24252
    },
    {
      "epoch": 0.9791279773920064,
      "grad_norm": 109.70036315917969,
      "learning_rate": 2.284138792539081e-08,
      "loss": 6.5423,
      "step": 24253
    },
    {
      "epoch": 0.9791683488090432,
      "grad_norm": 99.12134552001953,
      "learning_rate": 2.275314560373887e-08,
      "loss": 5.996,
      "step": 24254
    },
    {
      "epoch": 0.97920872022608,
      "grad_norm": 153.67422485351562,
      "learning_rate": 2.2665073869710952e-08,
      "loss": 3.8033,
      "step": 24255
    },
    {
      "epoch": 0.9792490916431167,
      "grad_norm": 104.775634765625,
      "learning_rate": 2.2577172724811413e-08,
      "loss": 4.8771,
      "step": 24256
    },
    {
      "epoch": 0.9792894630601534,
      "grad_norm": 94.84732055664062,
      "learning_rate": 2.2489442170542387e-08,
      "loss": 4.2651,
      "step": 24257
    },
    {
      "epoch": 0.9793298344771901,
      "grad_norm": 120.30623626708984,
      "learning_rate": 2.2401882208406e-08,
      "loss": 3.9856,
      "step": 24258
    },
    {
      "epoch": 0.9793702058942269,
      "grad_norm": 89.19287109375,
      "learning_rate": 2.2314492839896617e-08,
      "loss": 5.4497,
      "step": 24259
    },
    {
      "epoch": 0.9794105773112636,
      "grad_norm": 100.36161804199219,
      "learning_rate": 2.2227274066510816e-08,
      "loss": 5.2028,
      "step": 24260
    },
    {
      "epoch": 0.9794509487283004,
      "grad_norm": 95.21324920654297,
      "learning_rate": 2.214022588973852e-08,
      "loss": 4.3484,
      "step": 24261
    },
    {
      "epoch": 0.9794913201453371,
      "grad_norm": 118.13062286376953,
      "learning_rate": 2.2053348311068533e-08,
      "loss": 4.3717,
      "step": 24262
    },
    {
      "epoch": 0.9795316915623739,
      "grad_norm": 135.4764862060547,
      "learning_rate": 2.1966641331986337e-08,
      "loss": 3.4021,
      "step": 24263
    },
    {
      "epoch": 0.9795720629794106,
      "grad_norm": 93.41265106201172,
      "learning_rate": 2.188010495397408e-08,
      "loss": 5.4448,
      "step": 24264
    },
    {
      "epoch": 0.9796124343964473,
      "grad_norm": 89.13238525390625,
      "learning_rate": 2.1793739178510575e-08,
      "loss": 4.8914,
      "step": 24265
    },
    {
      "epoch": 0.979652805813484,
      "grad_norm": 95.00562286376953,
      "learning_rate": 2.1707544007074643e-08,
      "loss": 4.4143,
      "step": 24266
    },
    {
      "epoch": 0.9796931772305207,
      "grad_norm": 113.87862396240234,
      "learning_rate": 2.162151944113733e-08,
      "loss": 6.5607,
      "step": 24267
    },
    {
      "epoch": 0.9797335486475576,
      "grad_norm": 78.03431701660156,
      "learning_rate": 2.1535665482171898e-08,
      "loss": 4.1221,
      "step": 24268
    },
    {
      "epoch": 0.9797739200645943,
      "grad_norm": 98.59757995605469,
      "learning_rate": 2.1449982131644952e-08,
      "loss": 5.3544,
      "step": 24269
    },
    {
      "epoch": 0.979814291481631,
      "grad_norm": 166.30897521972656,
      "learning_rate": 2.136446939102088e-08,
      "loss": 5.005,
      "step": 24270
    },
    {
      "epoch": 0.9798546628986677,
      "grad_norm": 165.46066284179688,
      "learning_rate": 2.1279127261764067e-08,
      "loss": 5.0318,
      "step": 24271
    },
    {
      "epoch": 0.9798950343157045,
      "grad_norm": 87.97732543945312,
      "learning_rate": 2.1193955745331117e-08,
      "loss": 5.4762,
      "step": 24272
    },
    {
      "epoch": 0.9799354057327412,
      "grad_norm": 97.2804183959961,
      "learning_rate": 2.1108954843179764e-08,
      "loss": 5.0916,
      "step": 24273
    },
    {
      "epoch": 0.9799757771497779,
      "grad_norm": 86.91374969482422,
      "learning_rate": 2.1024124556763283e-08,
      "loss": 5.5756,
      "step": 24274
    },
    {
      "epoch": 0.9800161485668147,
      "grad_norm": 104.77050018310547,
      "learning_rate": 2.0939464887531623e-08,
      "loss": 5.6069,
      "step": 24275
    },
    {
      "epoch": 0.9800565199838515,
      "grad_norm": 187.60910034179688,
      "learning_rate": 2.0854975836931413e-08,
      "loss": 5.1454,
      "step": 24276
    },
    {
      "epoch": 0.9800968914008882,
      "grad_norm": 114.43087768554688,
      "learning_rate": 2.0770657406408156e-08,
      "loss": 4.6935,
      "step": 24277
    },
    {
      "epoch": 0.9801372628179249,
      "grad_norm": 80.37287139892578,
      "learning_rate": 2.0686509597404037e-08,
      "loss": 4.1355,
      "step": 24278
    },
    {
      "epoch": 0.9801776342349616,
      "grad_norm": 136.36944580078125,
      "learning_rate": 2.06025324113579e-08,
      "loss": 5.2494,
      "step": 24279
    },
    {
      "epoch": 0.9802180056519983,
      "grad_norm": 84.45442199707031,
      "learning_rate": 2.0518725849704157e-08,
      "loss": 4.9283,
      "step": 24280
    },
    {
      "epoch": 0.9802583770690351,
      "grad_norm": 89.84544372558594,
      "learning_rate": 2.0435089913876105e-08,
      "loss": 3.5387,
      "step": 24281
    },
    {
      "epoch": 0.9802987484860719,
      "grad_norm": 148.22901916503906,
      "learning_rate": 2.035162460530482e-08,
      "loss": 3.9671,
      "step": 24282
    },
    {
      "epoch": 0.9803391199031086,
      "grad_norm": 231.81417846679688,
      "learning_rate": 2.026832992541583e-08,
      "loss": 4.9112,
      "step": 24283
    },
    {
      "epoch": 0.9803794913201453,
      "grad_norm": 76.89204406738281,
      "learning_rate": 2.018520587563466e-08,
      "loss": 5.236,
      "step": 24284
    },
    {
      "epoch": 0.9804198627371821,
      "grad_norm": 89.76484680175781,
      "learning_rate": 2.010225245738129e-08,
      "loss": 5.22,
      "step": 24285
    },
    {
      "epoch": 0.9804602341542188,
      "grad_norm": 78.71076202392578,
      "learning_rate": 2.0019469672075687e-08,
      "loss": 4.7341,
      "step": 24286
    },
    {
      "epoch": 0.9805006055712555,
      "grad_norm": 112.00630187988281,
      "learning_rate": 1.993685752113117e-08,
      "loss": 4.7787,
      "step": 24287
    },
    {
      "epoch": 0.9805409769882922,
      "grad_norm": 122.8899917602539,
      "learning_rate": 1.9854416005962163e-08,
      "loss": 5.1107,
      "step": 24288
    },
    {
      "epoch": 0.9805813484053291,
      "grad_norm": 76.67436981201172,
      "learning_rate": 1.9772145127976428e-08,
      "loss": 3.0898,
      "step": 24289
    },
    {
      "epoch": 0.9806217198223658,
      "grad_norm": 319.7543029785156,
      "learning_rate": 1.9690044888581727e-08,
      "loss": 3.0708,
      "step": 24290
    },
    {
      "epoch": 0.9806620912394025,
      "grad_norm": 91.81755065917969,
      "learning_rate": 1.9608115289181385e-08,
      "loss": 4.9753,
      "step": 24291
    },
    {
      "epoch": 0.9807024626564392,
      "grad_norm": 158.2270965576172,
      "learning_rate": 1.952635633117761e-08,
      "loss": 5.3199,
      "step": 24292
    },
    {
      "epoch": 0.980742834073476,
      "grad_norm": 117.46561431884766,
      "learning_rate": 1.9444768015965954e-08,
      "loss": 4.3439,
      "step": 24293
    },
    {
      "epoch": 0.9807832054905127,
      "grad_norm": 75.66886901855469,
      "learning_rate": 1.936335034494308e-08,
      "loss": 5.4222,
      "step": 24294
    },
    {
      "epoch": 0.9808235769075495,
      "grad_norm": 191.04043579101562,
      "learning_rate": 1.928210331950009e-08,
      "loss": 4.0498,
      "step": 24295
    },
    {
      "epoch": 0.9808639483245862,
      "grad_norm": 94.13301086425781,
      "learning_rate": 1.920102694102588e-08,
      "loss": 3.4402,
      "step": 24296
    },
    {
      "epoch": 0.980904319741623,
      "grad_norm": 175.72940063476562,
      "learning_rate": 1.9120121210908227e-08,
      "loss": 4.7642,
      "step": 24297
    },
    {
      "epoch": 0.9809446911586597,
      "grad_norm": 86.0757827758789,
      "learning_rate": 1.903938613052936e-08,
      "loss": 4.54,
      "step": 24298
    },
    {
      "epoch": 0.9809850625756964,
      "grad_norm": 110.2520980834961,
      "learning_rate": 1.895882170126928e-08,
      "loss": 3.7528,
      "step": 24299
    },
    {
      "epoch": 0.9810254339927331,
      "grad_norm": 87.26069641113281,
      "learning_rate": 1.887842792450689e-08,
      "loss": 3.9413,
      "step": 24300
    },
    {
      "epoch": 0.9810658054097698,
      "grad_norm": 124.96944427490234,
      "learning_rate": 1.8798204801614428e-08,
      "loss": 3.2941,
      "step": 24301
    },
    {
      "epoch": 0.9811061768268067,
      "grad_norm": 70.00060272216797,
      "learning_rate": 1.8718152333965234e-08,
      "loss": 4.8257,
      "step": 24302
    },
    {
      "epoch": 0.9811465482438434,
      "grad_norm": 106.91712188720703,
      "learning_rate": 1.8638270522928215e-08,
      "loss": 2.7435,
      "step": 24303
    },
    {
      "epoch": 0.9811869196608801,
      "grad_norm": 267.80072021484375,
      "learning_rate": 1.8558559369868946e-08,
      "loss": 4.3032,
      "step": 24304
    },
    {
      "epoch": 0.9812272910779168,
      "grad_norm": 108.74339294433594,
      "learning_rate": 1.8479018876149667e-08,
      "loss": 5.011,
      "step": 24305
    },
    {
      "epoch": 0.9812676624949536,
      "grad_norm": 79.40176391601562,
      "learning_rate": 1.8399649043130408e-08,
      "loss": 5.5055,
      "step": 24306
    },
    {
      "epoch": 0.9813080339119903,
      "grad_norm": 102.5616455078125,
      "learning_rate": 1.8320449872167855e-08,
      "loss": 4.4645,
      "step": 24307
    },
    {
      "epoch": 0.981348405329027,
      "grad_norm": 202.4443359375,
      "learning_rate": 1.8241421364617594e-08,
      "loss": 5.5308,
      "step": 24308
    },
    {
      "epoch": 0.9813887767460638,
      "grad_norm": 161.63571166992188,
      "learning_rate": 1.8162563521829657e-08,
      "loss": 5.5228,
      "step": 24309
    },
    {
      "epoch": 0.9814291481631006,
      "grad_norm": 67.19136810302734,
      "learning_rate": 1.8083876345151853e-08,
      "loss": 4.3565,
      "step": 24310
    },
    {
      "epoch": 0.9814695195801373,
      "grad_norm": 119.20537567138672,
      "learning_rate": 1.800535983593088e-08,
      "loss": 5.5759,
      "step": 24311
    },
    {
      "epoch": 0.981509890997174,
      "grad_norm": 105.08942413330078,
      "learning_rate": 1.7927013995509002e-08,
      "loss": 4.7836,
      "step": 24312
    },
    {
      "epoch": 0.9815502624142107,
      "grad_norm": 76.2053451538086,
      "learning_rate": 1.7848838825225146e-08,
      "loss": 3.2687,
      "step": 24313
    },
    {
      "epoch": 0.9815906338312474,
      "grad_norm": 137.72166442871094,
      "learning_rate": 1.777083432641602e-08,
      "loss": 4.129,
      "step": 24314
    },
    {
      "epoch": 0.9816310052482842,
      "grad_norm": 114.57099914550781,
      "learning_rate": 1.7693000500416112e-08,
      "loss": 3.5712,
      "step": 24315
    },
    {
      "epoch": 0.981671376665321,
      "grad_norm": 113.7745590209961,
      "learning_rate": 1.761533734855547e-08,
      "loss": 5.0753,
      "step": 24316
    },
    {
      "epoch": 0.9817117480823577,
      "grad_norm": 103.9146499633789,
      "learning_rate": 1.7537844872160813e-08,
      "loss": 5.3145,
      "step": 24317
    },
    {
      "epoch": 0.9817521194993944,
      "grad_norm": 78.10242462158203,
      "learning_rate": 1.7460523072558854e-08,
      "loss": 4.4888,
      "step": 24318
    },
    {
      "epoch": 0.9817924909164312,
      "grad_norm": 98.8741226196289,
      "learning_rate": 1.738337195107187e-08,
      "loss": 5.2422,
      "step": 24319
    },
    {
      "epoch": 0.9818328623334679,
      "grad_norm": 97.53467559814453,
      "learning_rate": 1.7306391509017695e-08,
      "loss": 4.2124,
      "step": 24320
    },
    {
      "epoch": 0.9818732337505046,
      "grad_norm": 67.66891479492188,
      "learning_rate": 1.7229581747713055e-08,
      "loss": 4.1992,
      "step": 24321
    },
    {
      "epoch": 0.9819136051675413,
      "grad_norm": 124.0648422241211,
      "learning_rate": 1.715294266847023e-08,
      "loss": 5.5413,
      "step": 24322
    },
    {
      "epoch": 0.9819539765845782,
      "grad_norm": 81.3372802734375,
      "learning_rate": 1.7076474272601506e-08,
      "loss": 6.2886,
      "step": 24323
    },
    {
      "epoch": 0.9819943480016149,
      "grad_norm": 92.97021484375,
      "learning_rate": 1.7000176561412508e-08,
      "loss": 4.4143,
      "step": 24324
    },
    {
      "epoch": 0.9820347194186516,
      "grad_norm": 71.40120697021484,
      "learning_rate": 1.6924049536208853e-08,
      "loss": 4.9344,
      "step": 24325
    },
    {
      "epoch": 0.9820750908356883,
      "grad_norm": 77.1940689086914,
      "learning_rate": 1.6848093198291726e-08,
      "loss": 3.734,
      "step": 24326
    },
    {
      "epoch": 0.9821154622527251,
      "grad_norm": 77.2271499633789,
      "learning_rate": 1.6772307548960086e-08,
      "loss": 3.5031,
      "step": 24327
    },
    {
      "epoch": 0.9821558336697618,
      "grad_norm": 87.7691879272461,
      "learning_rate": 1.6696692589508457e-08,
      "loss": 6.3411,
      "step": 24328
    },
    {
      "epoch": 0.9821962050867985,
      "grad_norm": 77.94480895996094,
      "learning_rate": 1.6621248321232464e-08,
      "loss": 5.7405,
      "step": 24329
    },
    {
      "epoch": 0.9822365765038353,
      "grad_norm": 111.51642608642578,
      "learning_rate": 1.6545974745418858e-08,
      "loss": 5.9733,
      "step": 24330
    },
    {
      "epoch": 0.9822769479208721,
      "grad_norm": 101.35230255126953,
      "learning_rate": 1.6470871863356607e-08,
      "loss": 5.4548,
      "step": 24331
    },
    {
      "epoch": 0.9823173193379088,
      "grad_norm": 78.53108978271484,
      "learning_rate": 1.639593967632913e-08,
      "loss": 5.8597,
      "step": 24332
    },
    {
      "epoch": 0.9823576907549455,
      "grad_norm": 74.86492156982422,
      "learning_rate": 1.6321178185617627e-08,
      "loss": 5.0738,
      "step": 24333
    },
    {
      "epoch": 0.9823980621719822,
      "grad_norm": 121.860107421875,
      "learning_rate": 1.6246587392499957e-08,
      "loss": 4.7844,
      "step": 24334
    },
    {
      "epoch": 0.982438433589019,
      "grad_norm": 109.03343200683594,
      "learning_rate": 1.617216729825288e-08,
      "loss": 5.163,
      "step": 24335
    },
    {
      "epoch": 0.9824788050060557,
      "grad_norm": 124.29428100585938,
      "learning_rate": 1.6097917904147608e-08,
      "loss": 6.1366,
      "step": 24336
    },
    {
      "epoch": 0.9825191764230925,
      "grad_norm": 85.91680145263672,
      "learning_rate": 1.602383921145312e-08,
      "loss": 5.242,
      "step": 24337
    },
    {
      "epoch": 0.9825595478401292,
      "grad_norm": 139.84938049316406,
      "learning_rate": 1.5949931221437287e-08,
      "loss": 4.5614,
      "step": 24338
    },
    {
      "epoch": 0.9825999192571659,
      "grad_norm": 77.36549377441406,
      "learning_rate": 1.587619393536355e-08,
      "loss": 5.8041,
      "step": 24339
    },
    {
      "epoch": 0.9826402906742027,
      "grad_norm": 98.69252014160156,
      "learning_rate": 1.5802627354492006e-08,
      "loss": 4.0887,
      "step": 24340
    },
    {
      "epoch": 0.9826806620912394,
      "grad_norm": 87.09274291992188,
      "learning_rate": 1.5729231480080542e-08,
      "loss": 3.583,
      "step": 24341
    },
    {
      "epoch": 0.9827210335082761,
      "grad_norm": 98.1358642578125,
      "learning_rate": 1.5656006313383708e-08,
      "loss": 6.0434,
      "step": 24342
    },
    {
      "epoch": 0.9827614049253128,
      "grad_norm": 81.09566497802734,
      "learning_rate": 1.5582951855654947e-08,
      "loss": 6.0084,
      "step": 24343
    },
    {
      "epoch": 0.9828017763423497,
      "grad_norm": 99.76234436035156,
      "learning_rate": 1.5510068108142152e-08,
      "loss": 5.1683,
      "step": 24344
    },
    {
      "epoch": 0.9828421477593864,
      "grad_norm": 91.88113403320312,
      "learning_rate": 1.5437355072092098e-08,
      "loss": 5.3581,
      "step": 24345
    },
    {
      "epoch": 0.9828825191764231,
      "grad_norm": 99.43408966064453,
      "learning_rate": 1.5364812748747127e-08,
      "loss": 5.4199,
      "step": 24346
    },
    {
      "epoch": 0.9829228905934598,
      "grad_norm": 92.54912567138672,
      "learning_rate": 1.5292441139348468e-08,
      "loss": 4.6752,
      "step": 24347
    },
    {
      "epoch": 0.9829632620104966,
      "grad_norm": 79.68824768066406,
      "learning_rate": 1.522024024513291e-08,
      "loss": 6.0752,
      "step": 24348
    },
    {
      "epoch": 0.9830036334275333,
      "grad_norm": 92.8027114868164,
      "learning_rate": 1.514821006733502e-08,
      "loss": 5.7859,
      "step": 24349
    },
    {
      "epoch": 0.98304400484457,
      "grad_norm": 82.47821807861328,
      "learning_rate": 1.507635060718604e-08,
      "loss": 4.1082,
      "step": 24350
    },
    {
      "epoch": 0.9830843762616068,
      "grad_norm": 80.5840072631836,
      "learning_rate": 1.500466186591498e-08,
      "loss": 4.5134,
      "step": 24351
    },
    {
      "epoch": 0.9831247476786436,
      "grad_norm": 102.21283721923828,
      "learning_rate": 1.4933143844748643e-08,
      "loss": 4.4403,
      "step": 24352
    },
    {
      "epoch": 0.9831651190956803,
      "grad_norm": 98.16291809082031,
      "learning_rate": 1.486179654490827e-08,
      "loss": 5.0856,
      "step": 24353
    },
    {
      "epoch": 0.983205490512717,
      "grad_norm": 96.7361068725586,
      "learning_rate": 1.4790619967613994e-08,
      "loss": 4.878,
      "step": 24354
    },
    {
      "epoch": 0.9832458619297537,
      "grad_norm": 85.98735809326172,
      "learning_rate": 1.4719614114082626e-08,
      "loss": 6.7086,
      "step": 24355
    },
    {
      "epoch": 0.9832862333467904,
      "grad_norm": 72.27359771728516,
      "learning_rate": 1.4648778985528745e-08,
      "loss": 4.7179,
      "step": 24356
    },
    {
      "epoch": 0.9833266047638272,
      "grad_norm": 76.01217651367188,
      "learning_rate": 1.4578114583163606e-08,
      "loss": 3.3772,
      "step": 24357
    },
    {
      "epoch": 0.983366976180864,
      "grad_norm": 93.27103424072266,
      "learning_rate": 1.450762090819513e-08,
      "loss": 6.4163,
      "step": 24358
    },
    {
      "epoch": 0.9834073475979007,
      "grad_norm": 101.69227600097656,
      "learning_rate": 1.4437297961827912e-08,
      "loss": 4.7906,
      "step": 24359
    },
    {
      "epoch": 0.9834477190149374,
      "grad_norm": 87.15127563476562,
      "learning_rate": 1.4367145745265431e-08,
      "loss": 3.521,
      "step": 24360
    },
    {
      "epoch": 0.9834880904319742,
      "grad_norm": 83.16119384765625,
      "learning_rate": 1.429716425970673e-08,
      "loss": 4.9164,
      "step": 24361
    },
    {
      "epoch": 0.9835284618490109,
      "grad_norm": 90.74700164794922,
      "learning_rate": 1.4227353506348629e-08,
      "loss": 4.7784,
      "step": 24362
    },
    {
      "epoch": 0.9835688332660476,
      "grad_norm": 116.62755584716797,
      "learning_rate": 1.4157713486383507e-08,
      "loss": 5.1384,
      "step": 24363
    },
    {
      "epoch": 0.9836092046830843,
      "grad_norm": 70.76917266845703,
      "learning_rate": 1.4088244201002632e-08,
      "loss": 6.1691,
      "step": 24364
    },
    {
      "epoch": 0.9836495761001212,
      "grad_norm": 107.25180053710938,
      "learning_rate": 1.4018945651395055e-08,
      "loss": 4.8342,
      "step": 24365
    },
    {
      "epoch": 0.9836899475171579,
      "grad_norm": 105.8665771484375,
      "learning_rate": 1.3949817838743162e-08,
      "loss": 5.6223,
      "step": 24366
    },
    {
      "epoch": 0.9837303189341946,
      "grad_norm": 95.8415298461914,
      "learning_rate": 1.3880860764231563e-08,
      "loss": 6.5941,
      "step": 24367
    },
    {
      "epoch": 0.9837706903512313,
      "grad_norm": 106.59191131591797,
      "learning_rate": 1.3812074429037092e-08,
      "loss": 4.3208,
      "step": 24368
    },
    {
      "epoch": 0.983811061768268,
      "grad_norm": 115.9537353515625,
      "learning_rate": 1.3743458834337697e-08,
      "loss": 5.6573,
      "step": 24369
    },
    {
      "epoch": 0.9838514331853048,
      "grad_norm": 106.95660400390625,
      "learning_rate": 1.3675013981304664e-08,
      "loss": 4.8097,
      "step": 24370
    },
    {
      "epoch": 0.9838918046023415,
      "grad_norm": 75.4280014038086,
      "learning_rate": 1.3606739871109275e-08,
      "loss": 3.7307,
      "step": 24371
    },
    {
      "epoch": 0.9839321760193783,
      "grad_norm": 97.77295684814453,
      "learning_rate": 1.353863650491949e-08,
      "loss": 3.6297,
      "step": 24372
    },
    {
      "epoch": 0.983972547436415,
      "grad_norm": 80.23566436767578,
      "learning_rate": 1.3470703883897707e-08,
      "loss": 3.2613,
      "step": 24373
    },
    {
      "epoch": 0.9840129188534518,
      "grad_norm": 98.19511413574219,
      "learning_rate": 1.3402942009207442e-08,
      "loss": 4.3524,
      "step": 24374
    },
    {
      "epoch": 0.9840532902704885,
      "grad_norm": 120.97477722167969,
      "learning_rate": 1.3335350882006659e-08,
      "loss": 3.8533,
      "step": 24375
    },
    {
      "epoch": 0.9840936616875252,
      "grad_norm": 108.18887329101562,
      "learning_rate": 1.3267930503451098e-08,
      "loss": 5.7691,
      "step": 24376
    },
    {
      "epoch": 0.9841340331045619,
      "grad_norm": 106.3763198852539,
      "learning_rate": 1.3200680874692063e-08,
      "loss": 4.7152,
      "step": 24377
    },
    {
      "epoch": 0.9841744045215987,
      "grad_norm": 106.87039947509766,
      "learning_rate": 1.3133601996881962e-08,
      "loss": 5.0919,
      "step": 24378
    },
    {
      "epoch": 0.9842147759386355,
      "grad_norm": 94.6419906616211,
      "learning_rate": 1.3066693871165437e-08,
      "loss": 4.4991,
      "step": 24379
    },
    {
      "epoch": 0.9842551473556722,
      "grad_norm": 92.78506469726562,
      "learning_rate": 1.2999956498687127e-08,
      "loss": 4.687,
      "step": 24380
    },
    {
      "epoch": 0.9842955187727089,
      "grad_norm": 101.44475555419922,
      "learning_rate": 1.2933389880588343e-08,
      "loss": 3.7479,
      "step": 24381
    },
    {
      "epoch": 0.9843358901897457,
      "grad_norm": 95.98804473876953,
      "learning_rate": 1.2866994018007062e-08,
      "loss": 4.0392,
      "step": 24382
    },
    {
      "epoch": 0.9843762616067824,
      "grad_norm": 95.29615783691406,
      "learning_rate": 1.2800768912079043e-08,
      "loss": 6.9506,
      "step": 24383
    },
    {
      "epoch": 0.9844166330238191,
      "grad_norm": 83.41742706298828,
      "learning_rate": 1.2734714563935601e-08,
      "loss": 5.9613,
      "step": 24384
    },
    {
      "epoch": 0.9844570044408558,
      "grad_norm": 500.50799560546875,
      "learning_rate": 1.2668830974706948e-08,
      "loss": 5.8691,
      "step": 24385
    },
    {
      "epoch": 0.9844973758578927,
      "grad_norm": 83.9079360961914,
      "learning_rate": 1.2603118145518844e-08,
      "loss": 6.6357,
      "step": 24386
    },
    {
      "epoch": 0.9845377472749294,
      "grad_norm": 118.8978500366211,
      "learning_rate": 1.253757607749484e-08,
      "loss": 4.8991,
      "step": 24387
    },
    {
      "epoch": 0.9845781186919661,
      "grad_norm": 156.0125732421875,
      "learning_rate": 1.2472204771756257e-08,
      "loss": 5.0809,
      "step": 24388
    },
    {
      "epoch": 0.9846184901090028,
      "grad_norm": 95.24671173095703,
      "learning_rate": 1.2407004229421093e-08,
      "loss": 5.8164,
      "step": 24389
    },
    {
      "epoch": 0.9846588615260395,
      "grad_norm": 130.06459045410156,
      "learning_rate": 1.23419744516029e-08,
      "loss": 4.7876,
      "step": 24390
    },
    {
      "epoch": 0.9846992329430763,
      "grad_norm": 77.47630310058594,
      "learning_rate": 1.2277115439414122e-08,
      "loss": 3.7458,
      "step": 24391
    },
    {
      "epoch": 0.984739604360113,
      "grad_norm": 79.13795471191406,
      "learning_rate": 1.2212427193963872e-08,
      "loss": 4.4853,
      "step": 24392
    },
    {
      "epoch": 0.9847799757771498,
      "grad_norm": 101.87388610839844,
      "learning_rate": 1.214790971635682e-08,
      "loss": 5.0361,
      "step": 24393
    },
    {
      "epoch": 0.9848203471941865,
      "grad_norm": 78.41143798828125,
      "learning_rate": 1.208356300769875e-08,
      "loss": 4.1366,
      "step": 24394
    },
    {
      "epoch": 0.9848607186112233,
      "grad_norm": 86.65097045898438,
      "learning_rate": 1.2019387069087674e-08,
      "loss": 5.4967,
      "step": 24395
    },
    {
      "epoch": 0.98490109002826,
      "grad_norm": 101.78794860839844,
      "learning_rate": 1.19553819016216e-08,
      "loss": 4.0496,
      "step": 24396
    },
    {
      "epoch": 0.9849414614452967,
      "grad_norm": 98.62894439697266,
      "learning_rate": 1.1891547506394097e-08,
      "loss": 4.931,
      "step": 24397
    },
    {
      "epoch": 0.9849818328623334,
      "grad_norm": 82.11763763427734,
      "learning_rate": 1.1827883884497626e-08,
      "loss": 4.8447,
      "step": 24398
    },
    {
      "epoch": 0.9850222042793702,
      "grad_norm": 86.9968032836914,
      "learning_rate": 1.1764391037020206e-08,
      "loss": 3.752,
      "step": 24399
    },
    {
      "epoch": 0.985062575696407,
      "grad_norm": 103.11318969726562,
      "learning_rate": 1.1701068965047634e-08,
      "loss": 3.9877,
      "step": 24400
    },
    {
      "epoch": 0.9851029471134437,
      "grad_norm": 91.84455871582031,
      "learning_rate": 1.163791766966238e-08,
      "loss": 5.279,
      "step": 24401
    },
    {
      "epoch": 0.9851433185304804,
      "grad_norm": 143.42343139648438,
      "learning_rate": 1.1574937151944687e-08,
      "loss": 5.1717,
      "step": 24402
    },
    {
      "epoch": 0.9851836899475171,
      "grad_norm": 132.7056884765625,
      "learning_rate": 1.1512127412970364e-08,
      "loss": 5.2477,
      "step": 24403
    },
    {
      "epoch": 0.9852240613645539,
      "grad_norm": 97.60881042480469,
      "learning_rate": 1.1449488453814106e-08,
      "loss": 3.9305,
      "step": 24404
    },
    {
      "epoch": 0.9852644327815906,
      "grad_norm": 91.48435974121094,
      "learning_rate": 1.1387020275546168e-08,
      "loss": 4.8857,
      "step": 24405
    },
    {
      "epoch": 0.9853048041986274,
      "grad_norm": 106.78365325927734,
      "learning_rate": 1.1324722879235695e-08,
      "loss": 4.5757,
      "step": 24406
    },
    {
      "epoch": 0.9853451756156641,
      "grad_norm": 134.2041473388672,
      "learning_rate": 1.126259626594628e-08,
      "loss": 4.6326,
      "step": 24407
    },
    {
      "epoch": 0.9853855470327009,
      "grad_norm": 94.95989227294922,
      "learning_rate": 1.1200640436742627e-08,
      "loss": 5.1689,
      "step": 24408
    },
    {
      "epoch": 0.9854259184497376,
      "grad_norm": 114.21383666992188,
      "learning_rate": 1.1138855392681668e-08,
      "loss": 6.2563,
      "step": 24409
    },
    {
      "epoch": 0.9854662898667743,
      "grad_norm": 70.20820617675781,
      "learning_rate": 1.1077241134820338e-08,
      "loss": 6.3655,
      "step": 24410
    },
    {
      "epoch": 0.985506661283811,
      "grad_norm": 101.25658416748047,
      "learning_rate": 1.1015797664212235e-08,
      "loss": 5.2944,
      "step": 24411
    },
    {
      "epoch": 0.9855470327008478,
      "grad_norm": 94.49381256103516,
      "learning_rate": 1.0954524981908743e-08,
      "loss": 5.4636,
      "step": 24412
    },
    {
      "epoch": 0.9855874041178846,
      "grad_norm": 171.5458984375,
      "learning_rate": 1.0893423088956801e-08,
      "loss": 3.8892,
      "step": 24413
    },
    {
      "epoch": 0.9856277755349213,
      "grad_norm": 126.72518157958984,
      "learning_rate": 1.0832491986400018e-08,
      "loss": 4.2761,
      "step": 24414
    },
    {
      "epoch": 0.985668146951958,
      "grad_norm": 83.29296112060547,
      "learning_rate": 1.0771731675282005e-08,
      "loss": 4.5114,
      "step": 24415
    },
    {
      "epoch": 0.9857085183689948,
      "grad_norm": 110.39627838134766,
      "learning_rate": 1.071114215664193e-08,
      "loss": 5.079,
      "step": 24416
    },
    {
      "epoch": 0.9857488897860315,
      "grad_norm": 75.45741271972656,
      "learning_rate": 1.065072343151341e-08,
      "loss": 5.6377,
      "step": 24417
    },
    {
      "epoch": 0.9857892612030682,
      "grad_norm": 76.34683990478516,
      "learning_rate": 1.0590475500930064e-08,
      "loss": 7.1881,
      "step": 24418
    },
    {
      "epoch": 0.9858296326201049,
      "grad_norm": 84.00394439697266,
      "learning_rate": 1.0530398365923289e-08,
      "loss": 5.0168,
      "step": 24419
    },
    {
      "epoch": 0.9858700040371418,
      "grad_norm": 99.22877502441406,
      "learning_rate": 1.0470492027520041e-08,
      "loss": 3.3923,
      "step": 24420
    },
    {
      "epoch": 0.9859103754541785,
      "grad_norm": 84.48151397705078,
      "learning_rate": 1.0410756486743944e-08,
      "loss": 4.873,
      "step": 24421
    },
    {
      "epoch": 0.9859507468712152,
      "grad_norm": 88.05806732177734,
      "learning_rate": 1.0351191744616406e-08,
      "loss": 3.3214,
      "step": 24422
    },
    {
      "epoch": 0.9859911182882519,
      "grad_norm": 80.45618438720703,
      "learning_rate": 1.0291797802156611e-08,
      "loss": 4.7729,
      "step": 24423
    },
    {
      "epoch": 0.9860314897052886,
      "grad_norm": 110.88355255126953,
      "learning_rate": 1.0232574660379302e-08,
      "loss": 3.6813,
      "step": 24424
    },
    {
      "epoch": 0.9860718611223254,
      "grad_norm": 95.63288879394531,
      "learning_rate": 1.0173522320295893e-08,
      "loss": 5.2497,
      "step": 24425
    },
    {
      "epoch": 0.9861122325393621,
      "grad_norm": 96.42849731445312,
      "learning_rate": 1.0114640782918905e-08,
      "loss": 5.3818,
      "step": 24426
    },
    {
      "epoch": 0.9861526039563989,
      "grad_norm": 148.83197021484375,
      "learning_rate": 1.0055930049251983e-08,
      "loss": 5.4158,
      "step": 24427
    },
    {
      "epoch": 0.9861929753734356,
      "grad_norm": 91.34784698486328,
      "learning_rate": 9.997390120302098e-09,
      "loss": 1.844,
      "step": 24428
    },
    {
      "epoch": 0.9862333467904724,
      "grad_norm": 108.97679901123047,
      "learning_rate": 9.93902099706734e-09,
      "loss": 5.2482,
      "step": 24429
    },
    {
      "epoch": 0.9862737182075091,
      "grad_norm": 99.17559051513672,
      "learning_rate": 9.880822680546908e-09,
      "loss": 4.2516,
      "step": 24430
    },
    {
      "epoch": 0.9863140896245458,
      "grad_norm": 147.56161499023438,
      "learning_rate": 9.822795171735567e-09,
      "loss": 6.3483,
      "step": 24431
    },
    {
      "epoch": 0.9863544610415825,
      "grad_norm": 81.04779815673828,
      "learning_rate": 9.764938471626961e-09,
      "loss": 4.8967,
      "step": 24432
    },
    {
      "epoch": 0.9863948324586193,
      "grad_norm": 140.88668823242188,
      "learning_rate": 9.707252581208083e-09,
      "loss": 5.5245,
      "step": 24433
    },
    {
      "epoch": 0.9864352038756561,
      "grad_norm": 69.88136291503906,
      "learning_rate": 9.649737501465917e-09,
      "loss": 4.5066,
      "step": 24434
    },
    {
      "epoch": 0.9864755752926928,
      "grad_norm": 101.678955078125,
      "learning_rate": 9.592393233385233e-09,
      "loss": 4.6303,
      "step": 24435
    },
    {
      "epoch": 0.9865159467097295,
      "grad_norm": 89.18993377685547,
      "learning_rate": 9.535219777944138e-09,
      "loss": 5.4703,
      "step": 24436
    },
    {
      "epoch": 0.9865563181267663,
      "grad_norm": 101.30998229980469,
      "learning_rate": 9.478217136121848e-09,
      "loss": 3.9668,
      "step": 24437
    },
    {
      "epoch": 0.986596689543803,
      "grad_norm": 127.7142333984375,
      "learning_rate": 9.421385308893139e-09,
      "loss": 4.7601,
      "step": 24438
    },
    {
      "epoch": 0.9866370609608397,
      "grad_norm": 107.42593383789062,
      "learning_rate": 9.364724297228344e-09,
      "loss": 3.0162,
      "step": 24439
    },
    {
      "epoch": 0.9866774323778764,
      "grad_norm": 98.70769500732422,
      "learning_rate": 9.308234102096691e-09,
      "loss": 4.0913,
      "step": 24440
    },
    {
      "epoch": 0.9867178037949133,
      "grad_norm": 90.96492004394531,
      "learning_rate": 9.251914724465183e-09,
      "loss": 4.7228,
      "step": 24441
    },
    {
      "epoch": 0.98675817521195,
      "grad_norm": 113.55204772949219,
      "learning_rate": 9.195766165295272e-09,
      "loss": 5.107,
      "step": 24442
    },
    {
      "epoch": 0.9867985466289867,
      "grad_norm": 100.36882781982422,
      "learning_rate": 9.139788425547303e-09,
      "loss": 4.5554,
      "step": 24443
    },
    {
      "epoch": 0.9868389180460234,
      "grad_norm": 140.84002685546875,
      "learning_rate": 9.083981506179396e-09,
      "loss": 5.6662,
      "step": 24444
    },
    {
      "epoch": 0.9868792894630601,
      "grad_norm": 109.77669525146484,
      "learning_rate": 9.028345408144123e-09,
      "loss": 5.8893,
      "step": 24445
    },
    {
      "epoch": 0.9869196608800969,
      "grad_norm": 89.93600463867188,
      "learning_rate": 8.972880132394057e-09,
      "loss": 3.4001,
      "step": 24446
    },
    {
      "epoch": 0.9869600322971336,
      "grad_norm": 88.52454376220703,
      "learning_rate": 8.917585679877327e-09,
      "loss": 3.8529,
      "step": 24447
    },
    {
      "epoch": 0.9870004037141704,
      "grad_norm": 111.82058715820312,
      "learning_rate": 8.862462051537624e-09,
      "loss": 5.073,
      "step": 24448
    },
    {
      "epoch": 0.9870407751312071,
      "grad_norm": 98.8494873046875,
      "learning_rate": 8.807509248319745e-09,
      "loss": 4.0381,
      "step": 24449
    },
    {
      "epoch": 0.9870811465482439,
      "grad_norm": 101.57471466064453,
      "learning_rate": 8.752727271162942e-09,
      "loss": 5.6857,
      "step": 24450
    },
    {
      "epoch": 0.9871215179652806,
      "grad_norm": 98.51171112060547,
      "learning_rate": 8.69811612100202e-09,
      "loss": 5.7404,
      "step": 24451
    },
    {
      "epoch": 0.9871618893823173,
      "grad_norm": 70.93238067626953,
      "learning_rate": 8.6436757987729e-09,
      "loss": 5.6723,
      "step": 24452
    },
    {
      "epoch": 0.987202260799354,
      "grad_norm": 94.60203552246094,
      "learning_rate": 8.589406305405945e-09,
      "loss": 3.475,
      "step": 24453
    },
    {
      "epoch": 0.9872426322163907,
      "grad_norm": 78.01213073730469,
      "learning_rate": 8.535307641827085e-09,
      "loss": 4.1014,
      "step": 24454
    },
    {
      "epoch": 0.9872830036334276,
      "grad_norm": 105.97747802734375,
      "learning_rate": 8.481379808963353e-09,
      "loss": 3.8076,
      "step": 24455
    },
    {
      "epoch": 0.9873233750504643,
      "grad_norm": 72.4966812133789,
      "learning_rate": 8.427622807736235e-09,
      "loss": 5.2919,
      "step": 24456
    },
    {
      "epoch": 0.987363746467501,
      "grad_norm": 111.35460662841797,
      "learning_rate": 8.374036639064997e-09,
      "loss": 5.293,
      "step": 24457
    },
    {
      "epoch": 0.9874041178845377,
      "grad_norm": 115.66706848144531,
      "learning_rate": 8.320621303865573e-09,
      "loss": 4.8612,
      "step": 24458
    },
    {
      "epoch": 0.9874444893015745,
      "grad_norm": 103.86703491210938,
      "learning_rate": 8.267376803050564e-09,
      "loss": 4.7175,
      "step": 24459
    },
    {
      "epoch": 0.9874848607186112,
      "grad_norm": 124.1633529663086,
      "learning_rate": 8.214303137531465e-09,
      "loss": 5.3115,
      "step": 24460
    },
    {
      "epoch": 0.9875252321356479,
      "grad_norm": 68.43225860595703,
      "learning_rate": 8.161400308215328e-09,
      "loss": 6.2806,
      "step": 24461
    },
    {
      "epoch": 0.9875656035526847,
      "grad_norm": 91.75182342529297,
      "learning_rate": 8.108668316006984e-09,
      "loss": 4.4542,
      "step": 24462
    },
    {
      "epoch": 0.9876059749697215,
      "grad_norm": 101.2827377319336,
      "learning_rate": 8.056107161806825e-09,
      "loss": 5.0283,
      "step": 24463
    },
    {
      "epoch": 0.9876463463867582,
      "grad_norm": 111.71894073486328,
      "learning_rate": 8.003716846515242e-09,
      "loss": 5.5158,
      "step": 24464
    },
    {
      "epoch": 0.9876867178037949,
      "grad_norm": 115.71814727783203,
      "learning_rate": 7.951497371025962e-09,
      "loss": 4.8721,
      "step": 24465
    },
    {
      "epoch": 0.9877270892208316,
      "grad_norm": 70.47990417480469,
      "learning_rate": 7.899448736233829e-09,
      "loss": 5.0403,
      "step": 24466
    },
    {
      "epoch": 0.9877674606378684,
      "grad_norm": 75.05716705322266,
      "learning_rate": 7.847570943028126e-09,
      "loss": 3.3696,
      "step": 24467
    },
    {
      "epoch": 0.9878078320549051,
      "grad_norm": 94.67184448242188,
      "learning_rate": 7.795863992294816e-09,
      "loss": 2.7803,
      "step": 24468
    },
    {
      "epoch": 0.9878482034719419,
      "grad_norm": 88.61277770996094,
      "learning_rate": 7.744327884919856e-09,
      "loss": 3.3298,
      "step": 24469
    },
    {
      "epoch": 0.9878885748889786,
      "grad_norm": 110.56016540527344,
      "learning_rate": 7.69296262178254e-09,
      "loss": 5.9468,
      "step": 24470
    },
    {
      "epoch": 0.9879289463060154,
      "grad_norm": 92.33744049072266,
      "learning_rate": 7.641768203762168e-09,
      "loss": 5.4622,
      "step": 24471
    },
    {
      "epoch": 0.9879693177230521,
      "grad_norm": 69.88642883300781,
      "learning_rate": 7.590744631733593e-09,
      "loss": 2.7948,
      "step": 24472
    },
    {
      "epoch": 0.9880096891400888,
      "grad_norm": 75.67164611816406,
      "learning_rate": 7.539891906568342e-09,
      "loss": 3.3444,
      "step": 24473
    },
    {
      "epoch": 0.9880500605571255,
      "grad_norm": 72.79216003417969,
      "learning_rate": 7.48921002913794e-09,
      "loss": 4.3005,
      "step": 24474
    },
    {
      "epoch": 0.9880904319741622,
      "grad_norm": 81.9466781616211,
      "learning_rate": 7.438699000307248e-09,
      "loss": 4.1179,
      "step": 24475
    },
    {
      "epoch": 0.9881308033911991,
      "grad_norm": 90.05806732177734,
      "learning_rate": 7.388358820941133e-09,
      "loss": 5.6566,
      "step": 24476
    },
    {
      "epoch": 0.9881711748082358,
      "grad_norm": 114.5804443359375,
      "learning_rate": 7.338189491900016e-09,
      "loss": 4.8256,
      "step": 24477
    },
    {
      "epoch": 0.9882115462252725,
      "grad_norm": 111.4085464477539,
      "learning_rate": 7.288191014040991e-09,
      "loss": 4.5513,
      "step": 24478
    },
    {
      "epoch": 0.9882519176423092,
      "grad_norm": 89.99536895751953,
      "learning_rate": 7.238363388218927e-09,
      "loss": 4.7657,
      "step": 24479
    },
    {
      "epoch": 0.988292289059346,
      "grad_norm": 122.66210174560547,
      "learning_rate": 7.188706615285368e-09,
      "loss": 5.5258,
      "step": 24480
    },
    {
      "epoch": 0.9883326604763827,
      "grad_norm": 82.00325012207031,
      "learning_rate": 7.139220696090743e-09,
      "loss": 4.4812,
      "step": 24481
    },
    {
      "epoch": 0.9883730318934194,
      "grad_norm": 129.640380859375,
      "learning_rate": 7.089905631481042e-09,
      "loss": 4.7528,
      "step": 24482
    },
    {
      "epoch": 0.9884134033104562,
      "grad_norm": 93.81925201416016,
      "learning_rate": 7.040761422297815e-09,
      "loss": 3.2708,
      "step": 24483
    },
    {
      "epoch": 0.988453774727493,
      "grad_norm": 69.04833221435547,
      "learning_rate": 6.9917880693837204e-09,
      "loss": 4.4786,
      "step": 24484
    },
    {
      "epoch": 0.9884941461445297,
      "grad_norm": 179.05348205566406,
      "learning_rate": 6.942985573573647e-09,
      "loss": 6.6672,
      "step": 24485
    },
    {
      "epoch": 0.9885345175615664,
      "grad_norm": 126.9984359741211,
      "learning_rate": 6.89435393570248e-09,
      "loss": 4.1114,
      "step": 24486
    },
    {
      "epoch": 0.9885748889786031,
      "grad_norm": 99.56047821044922,
      "learning_rate": 6.845893156602889e-09,
      "loss": 5.4312,
      "step": 24487
    },
    {
      "epoch": 0.9886152603956398,
      "grad_norm": 81.79984283447266,
      "learning_rate": 6.797603237103101e-09,
      "loss": 5.0048,
      "step": 24488
    },
    {
      "epoch": 0.9886556318126766,
      "grad_norm": 186.35519409179688,
      "learning_rate": 6.7494841780280094e-09,
      "loss": 4.6788,
      "step": 24489
    },
    {
      "epoch": 0.9886960032297134,
      "grad_norm": 104.40576171875,
      "learning_rate": 6.701535980201401e-09,
      "loss": 4.7618,
      "step": 24490
    },
    {
      "epoch": 0.9887363746467501,
      "grad_norm": 80.36439514160156,
      "learning_rate": 6.6537586444426205e-09,
      "loss": 5.5358,
      "step": 24491
    },
    {
      "epoch": 0.9887767460637868,
      "grad_norm": 75.1994400024414,
      "learning_rate": 6.606152171567681e-09,
      "loss": 5.1615,
      "step": 24492
    },
    {
      "epoch": 0.9888171174808236,
      "grad_norm": 84.8632583618164,
      "learning_rate": 6.558716562391487e-09,
      "loss": 2.7124,
      "step": 24493
    },
    {
      "epoch": 0.9888574888978603,
      "grad_norm": 77.38408660888672,
      "learning_rate": 6.511451817725612e-09,
      "loss": 4.9478,
      "step": 24494
    },
    {
      "epoch": 0.988897860314897,
      "grad_norm": 80.51343536376953,
      "learning_rate": 6.464357938377186e-09,
      "loss": 3.2633,
      "step": 24495
    },
    {
      "epoch": 0.9889382317319337,
      "grad_norm": 92.11589050292969,
      "learning_rate": 6.417434925151122e-09,
      "loss": 4.3183,
      "step": 24496
    },
    {
      "epoch": 0.9889786031489706,
      "grad_norm": 102.77232360839844,
      "learning_rate": 6.370682778850113e-09,
      "loss": 5.1615,
      "step": 24497
    },
    {
      "epoch": 0.9890189745660073,
      "grad_norm": 106.21446990966797,
      "learning_rate": 6.3241015002735165e-09,
      "loss": 5.0668,
      "step": 24498
    },
    {
      "epoch": 0.989059345983044,
      "grad_norm": 101.56300354003906,
      "learning_rate": 6.277691090218474e-09,
      "loss": 5.4198,
      "step": 24499
    },
    {
      "epoch": 0.9890997174000807,
      "grad_norm": 108.31633758544922,
      "learning_rate": 6.231451549477685e-09,
      "loss": 5.2339,
      "step": 24500
    },
    {
      "epoch": 0.9891400888171175,
      "grad_norm": 116.65299224853516,
      "learning_rate": 6.185382878842738e-09,
      "loss": 5.1192,
      "step": 24501
    },
    {
      "epoch": 0.9891804602341542,
      "grad_norm": 121.61404418945312,
      "learning_rate": 6.139485079099672e-09,
      "loss": 6.4404,
      "step": 24502
    },
    {
      "epoch": 0.9892208316511909,
      "grad_norm": 88.52312469482422,
      "learning_rate": 6.093758151033413e-09,
      "loss": 3.9008,
      "step": 24503
    },
    {
      "epoch": 0.9892612030682277,
      "grad_norm": 84.35124206542969,
      "learning_rate": 6.0482020954277795e-09,
      "loss": 6.647,
      "step": 24504
    },
    {
      "epoch": 0.9893015744852645,
      "grad_norm": 98.97713470458984,
      "learning_rate": 6.0028169130599275e-09,
      "loss": 7.216,
      "step": 24505
    },
    {
      "epoch": 0.9893419459023012,
      "grad_norm": 104.12388610839844,
      "learning_rate": 5.9576026047059035e-09,
      "loss": 6.1516,
      "step": 24506
    },
    {
      "epoch": 0.9893823173193379,
      "grad_norm": 105.67024993896484,
      "learning_rate": 5.912559171138421e-09,
      "loss": 5.187,
      "step": 24507
    },
    {
      "epoch": 0.9894226887363746,
      "grad_norm": 80.22716522216797,
      "learning_rate": 5.867686613130197e-09,
      "loss": 5.7534,
      "step": 24508
    },
    {
      "epoch": 0.9894630601534113,
      "grad_norm": 107.45426177978516,
      "learning_rate": 5.822984931445064e-09,
      "loss": 5.8831,
      "step": 24509
    },
    {
      "epoch": 0.9895034315704482,
      "grad_norm": 120.62779998779297,
      "learning_rate": 5.7784541268490755e-09,
      "loss": 5.5323,
      "step": 24510
    },
    {
      "epoch": 0.9895438029874849,
      "grad_norm": 104.32878112792969,
      "learning_rate": 5.734094200103846e-09,
      "loss": 3.8825,
      "step": 24511
    },
    {
      "epoch": 0.9895841744045216,
      "grad_norm": 258.3460388183594,
      "learning_rate": 5.689905151966546e-09,
      "loss": 4.431,
      "step": 24512
    },
    {
      "epoch": 0.9896245458215583,
      "grad_norm": 116.77789306640625,
      "learning_rate": 5.645886983194349e-09,
      "loss": 4.4907,
      "step": 24513
    },
    {
      "epoch": 0.9896649172385951,
      "grad_norm": 145.84388732910156,
      "learning_rate": 5.602039694538875e-09,
      "loss": 4.615,
      "step": 24514
    },
    {
      "epoch": 0.9897052886556318,
      "grad_norm": 116.61652374267578,
      "learning_rate": 5.558363286749524e-09,
      "loss": 5.7005,
      "step": 24515
    },
    {
      "epoch": 0.9897456600726685,
      "grad_norm": 107.59900665283203,
      "learning_rate": 5.514857760573478e-09,
      "loss": 4.4037,
      "step": 24516
    },
    {
      "epoch": 0.9897860314897053,
      "grad_norm": 114.69893646240234,
      "learning_rate": 5.471523116754585e-09,
      "loss": 3.4564,
      "step": 24517
    },
    {
      "epoch": 0.9898264029067421,
      "grad_norm": 103.145263671875,
      "learning_rate": 5.428359356033364e-09,
      "loss": 4.9096,
      "step": 24518
    },
    {
      "epoch": 0.9898667743237788,
      "grad_norm": 132.2434539794922,
      "learning_rate": 5.385366479149224e-09,
      "loss": 5.3514,
      "step": 24519
    },
    {
      "epoch": 0.9899071457408155,
      "grad_norm": 88.49430847167969,
      "learning_rate": 5.342544486834911e-09,
      "loss": 5.7559,
      "step": 24520
    },
    {
      "epoch": 0.9899475171578522,
      "grad_norm": 91.99633026123047,
      "learning_rate": 5.2998933798242834e-09,
      "loss": 5.0244,
      "step": 24521
    },
    {
      "epoch": 0.989987888574889,
      "grad_norm": 88.78832244873047,
      "learning_rate": 5.257413158846758e-09,
      "loss": 4.7999,
      "step": 24522
    },
    {
      "epoch": 0.9900282599919257,
      "grad_norm": 99.15006256103516,
      "learning_rate": 5.21510382462842e-09,
      "loss": 5.4359,
      "step": 24523
    },
    {
      "epoch": 0.9900686314089625,
      "grad_norm": 78.4573745727539,
      "learning_rate": 5.172965377890915e-09,
      "loss": 5.0457,
      "step": 24524
    },
    {
      "epoch": 0.9901090028259992,
      "grad_norm": 131.87429809570312,
      "learning_rate": 5.130997819356997e-09,
      "loss": 3.8709,
      "step": 24525
    },
    {
      "epoch": 0.990149374243036,
      "grad_norm": 113.25084686279297,
      "learning_rate": 5.0892011497427614e-09,
      "loss": 4.0213,
      "step": 24526
    },
    {
      "epoch": 0.9901897456600727,
      "grad_norm": 97.43614196777344,
      "learning_rate": 5.047575369764301e-09,
      "loss": 4.5844,
      "step": 24527
    },
    {
      "epoch": 0.9902301170771094,
      "grad_norm": 94.72722625732422,
      "learning_rate": 5.00612048013216e-09,
      "loss": 6.1304,
      "step": 24528
    },
    {
      "epoch": 0.9902704884941461,
      "grad_norm": 86.32913970947266,
      "learning_rate": 4.964836481554658e-09,
      "loss": 5.236,
      "step": 24529
    },
    {
      "epoch": 0.9903108599111828,
      "grad_norm": 88.54869842529297,
      "learning_rate": 4.923723374739009e-09,
      "loss": 4.9338,
      "step": 24530
    },
    {
      "epoch": 0.9903512313282197,
      "grad_norm": 76.498291015625,
      "learning_rate": 4.882781160386873e-09,
      "loss": 4.8979,
      "step": 24531
    },
    {
      "epoch": 0.9903916027452564,
      "grad_norm": 156.4144287109375,
      "learning_rate": 4.8420098391999125e-09,
      "loss": 3.7905,
      "step": 24532
    },
    {
      "epoch": 0.9904319741622931,
      "grad_norm": 95.36177825927734,
      "learning_rate": 4.801409411873126e-09,
      "loss": 4.671,
      "step": 24533
    },
    {
      "epoch": 0.9904723455793298,
      "grad_norm": 113.16976928710938,
      "learning_rate": 4.760979879102623e-09,
      "loss": 4.6387,
      "step": 24534
    },
    {
      "epoch": 0.9905127169963666,
      "grad_norm": 74.52059173583984,
      "learning_rate": 4.7207212415789624e-09,
      "loss": 5.5798,
      "step": 24535
    },
    {
      "epoch": 0.9905530884134033,
      "grad_norm": 102.40383911132812,
      "learning_rate": 4.680633499989373e-09,
      "loss": 5.2051,
      "step": 24536
    },
    {
      "epoch": 0.99059345983044,
      "grad_norm": 101.1458969116211,
      "learning_rate": 4.640716655021082e-09,
      "loss": 5.3277,
      "step": 24537
    },
    {
      "epoch": 0.9906338312474768,
      "grad_norm": 91.78395080566406,
      "learning_rate": 4.600970707354657e-09,
      "loss": 5.9463,
      "step": 24538
    },
    {
      "epoch": 0.9906742026645136,
      "grad_norm": 109.56053161621094,
      "learning_rate": 4.561395657671774e-09,
      "loss": 6.2028,
      "step": 24539
    },
    {
      "epoch": 0.9907145740815503,
      "grad_norm": 67.34354400634766,
      "learning_rate": 4.52199150664745e-09,
      "loss": 5.6166,
      "step": 24540
    },
    {
      "epoch": 0.990754945498587,
      "grad_norm": 77.07832336425781,
      "learning_rate": 4.482758254955588e-09,
      "loss": 5.0618,
      "step": 24541
    },
    {
      "epoch": 0.9907953169156237,
      "grad_norm": 93.1126708984375,
      "learning_rate": 4.443695903267875e-09,
      "loss": 5.6898,
      "step": 24542
    },
    {
      "epoch": 0.9908356883326604,
      "grad_norm": 73.24250793457031,
      "learning_rate": 4.404804452251554e-09,
      "loss": 5.9713,
      "step": 24543
    },
    {
      "epoch": 0.9908760597496972,
      "grad_norm": 92.80513763427734,
      "learning_rate": 4.366083902571649e-09,
      "loss": 5.5232,
      "step": 24544
    },
    {
      "epoch": 0.990916431166734,
      "grad_norm": 97.99968719482422,
      "learning_rate": 4.327534254890964e-09,
      "loss": 4.6685,
      "step": 24545
    },
    {
      "epoch": 0.9909568025837707,
      "grad_norm": 104.23102569580078,
      "learning_rate": 4.289155509866749e-09,
      "loss": 5.655,
      "step": 24546
    },
    {
      "epoch": 0.9909971740008074,
      "grad_norm": 88.59882354736328,
      "learning_rate": 4.250947668157368e-09,
      "loss": 5.1354,
      "step": 24547
    },
    {
      "epoch": 0.9910375454178442,
      "grad_norm": 136.44256591796875,
      "learning_rate": 4.212910730414521e-09,
      "loss": 3.8922,
      "step": 24548
    },
    {
      "epoch": 0.9910779168348809,
      "grad_norm": 78.6749038696289,
      "learning_rate": 4.175044697288799e-09,
      "loss": 5.1927,
      "step": 24549
    },
    {
      "epoch": 0.9911182882519176,
      "grad_norm": 158.16038513183594,
      "learning_rate": 4.137349569427462e-09,
      "loss": 5.7234,
      "step": 24550
    },
    {
      "epoch": 0.9911586596689543,
      "grad_norm": 96.04301452636719,
      "learning_rate": 4.09982534747666e-09,
      "loss": 4.0815,
      "step": 24551
    },
    {
      "epoch": 0.9911990310859912,
      "grad_norm": 156.96200561523438,
      "learning_rate": 4.062472032075881e-09,
      "loss": 7.0466,
      "step": 24552
    },
    {
      "epoch": 0.9912394025030279,
      "grad_norm": 86.87237548828125,
      "learning_rate": 4.025289623864614e-09,
      "loss": 4.5633,
      "step": 24553
    },
    {
      "epoch": 0.9912797739200646,
      "grad_norm": 88.63813018798828,
      "learning_rate": 3.988278123479017e-09,
      "loss": 3.765,
      "step": 24554
    },
    {
      "epoch": 0.9913201453371013,
      "grad_norm": 89.45057678222656,
      "learning_rate": 3.951437531550806e-09,
      "loss": 3.6437,
      "step": 24555
    },
    {
      "epoch": 0.991360516754138,
      "grad_norm": 127.50738525390625,
      "learning_rate": 3.914767848711698e-09,
      "loss": 4.8387,
      "step": 24556
    },
    {
      "epoch": 0.9914008881711748,
      "grad_norm": 93.61769104003906,
      "learning_rate": 3.878269075586749e-09,
      "loss": 5.7036,
      "step": 24557
    },
    {
      "epoch": 0.9914412595882115,
      "grad_norm": 77.52468872070312,
      "learning_rate": 3.841941212801015e-09,
      "loss": 4.8824,
      "step": 24558
    },
    {
      "epoch": 0.9914816310052483,
      "grad_norm": 125.8381576538086,
      "learning_rate": 3.8057842609751095e-09,
      "loss": 4.2338,
      "step": 24559
    },
    {
      "epoch": 0.991522002422285,
      "grad_norm": 98.62630462646484,
      "learning_rate": 3.7697982207274275e-09,
      "loss": 3.2493,
      "step": 24560
    },
    {
      "epoch": 0.9915623738393218,
      "grad_norm": 114.787841796875,
      "learning_rate": 3.733983092674143e-09,
      "loss": 4.5585,
      "step": 24561
    },
    {
      "epoch": 0.9916027452563585,
      "grad_norm": 108.78958129882812,
      "learning_rate": 3.698338877425878e-09,
      "loss": 4.4927,
      "step": 24562
    },
    {
      "epoch": 0.9916431166733952,
      "grad_norm": 105.60529327392578,
      "learning_rate": 3.662865575593255e-09,
      "loss": 5.4484,
      "step": 24563
    },
    {
      "epoch": 0.9916834880904319,
      "grad_norm": 145.94224548339844,
      "learning_rate": 3.6275631877824568e-09,
      "loss": 4.9455,
      "step": 24564
    },
    {
      "epoch": 0.9917238595074687,
      "grad_norm": 93.13809967041016,
      "learning_rate": 3.592431714597444e-09,
      "loss": 4.5407,
      "step": 24565
    },
    {
      "epoch": 0.9917642309245055,
      "grad_norm": 101.83806610107422,
      "learning_rate": 3.5574711566388474e-09,
      "loss": 5.0385,
      "step": 24566
    },
    {
      "epoch": 0.9918046023415422,
      "grad_norm": 79.85114288330078,
      "learning_rate": 3.522681514502857e-09,
      "loss": 6.0541,
      "step": 24567
    },
    {
      "epoch": 0.9918449737585789,
      "grad_norm": 73.53584289550781,
      "learning_rate": 3.4880627887867724e-09,
      "loss": 4.2491,
      "step": 24568
    },
    {
      "epoch": 0.9918853451756157,
      "grad_norm": 89.91436767578125,
      "learning_rate": 3.4536149800812323e-09,
      "loss": 4.9324,
      "step": 24569
    },
    {
      "epoch": 0.9919257165926524,
      "grad_norm": 296.5624084472656,
      "learning_rate": 3.4193380889746553e-09,
      "loss": 3.9574,
      "step": 24570
    },
    {
      "epoch": 0.9919660880096891,
      "grad_norm": 157.18589782714844,
      "learning_rate": 3.3852321160532386e-09,
      "loss": 4.0406,
      "step": 24571
    },
    {
      "epoch": 0.9920064594267258,
      "grad_norm": 74.47377014160156,
      "learning_rate": 3.3512970619020703e-09,
      "loss": 4.8884,
      "step": 24572
    },
    {
      "epoch": 0.9920468308437627,
      "grad_norm": 82.48374938964844,
      "learning_rate": 3.3175329270984657e-09,
      "loss": 5.566,
      "step": 24573
    },
    {
      "epoch": 0.9920872022607994,
      "grad_norm": 104.77002716064453,
      "learning_rate": 3.2839397122219617e-09,
      "loss": 3.6581,
      "step": 24574
    },
    {
      "epoch": 0.9921275736778361,
      "grad_norm": 70.0002670288086,
      "learning_rate": 3.2505174178454337e-09,
      "loss": 3.9323,
      "step": 24575
    },
    {
      "epoch": 0.9921679450948728,
      "grad_norm": 106.62885284423828,
      "learning_rate": 3.2172660445417556e-09,
      "loss": 4.3621,
      "step": 24576
    },
    {
      "epoch": 0.9922083165119095,
      "grad_norm": 139.14028930664062,
      "learning_rate": 3.1841855928782526e-09,
      "loss": 4.5722,
      "step": 24577
    },
    {
      "epoch": 0.9922486879289463,
      "grad_norm": 76.13961029052734,
      "learning_rate": 3.151276063421138e-09,
      "loss": 4.843,
      "step": 24578
    },
    {
      "epoch": 0.992289059345983,
      "grad_norm": 85.20195770263672,
      "learning_rate": 3.1185374567321847e-09,
      "loss": 4.3832,
      "step": 24579
    },
    {
      "epoch": 0.9923294307630198,
      "grad_norm": 85.42625427246094,
      "learning_rate": 3.0859697733731655e-09,
      "loss": 5.6273,
      "step": 24580
    },
    {
      "epoch": 0.9923698021800565,
      "grad_norm": 99.24276733398438,
      "learning_rate": 3.0535730138980814e-09,
      "loss": 5.1635,
      "step": 24581
    },
    {
      "epoch": 0.9924101735970933,
      "grad_norm": 98.59866333007812,
      "learning_rate": 3.0213471788631544e-09,
      "loss": 3.8103,
      "step": 24582
    },
    {
      "epoch": 0.99245054501413,
      "grad_norm": 107.15174865722656,
      "learning_rate": 2.989292268819055e-09,
      "loss": 3.5688,
      "step": 24583
    },
    {
      "epoch": 0.9924909164311667,
      "grad_norm": 114.13658905029297,
      "learning_rate": 2.9574082843120133e-09,
      "loss": 5.6931,
      "step": 24584
    },
    {
      "epoch": 0.9925312878482034,
      "grad_norm": 89.62726593017578,
      "learning_rate": 2.925695225889369e-09,
      "loss": 5.6323,
      "step": 24585
    },
    {
      "epoch": 0.9925716592652402,
      "grad_norm": 93.30294036865234,
      "learning_rate": 2.8941530940929107e-09,
      "loss": 5.4591,
      "step": 24586
    },
    {
      "epoch": 0.992612030682277,
      "grad_norm": 114.41621398925781,
      "learning_rate": 2.862781889461097e-09,
      "loss": 5.0315,
      "step": 24587
    },
    {
      "epoch": 0.9926524020993137,
      "grad_norm": 104.4436264038086,
      "learning_rate": 2.8315816125301654e-09,
      "loss": 4.1974,
      "step": 24588
    },
    {
      "epoch": 0.9926927735163504,
      "grad_norm": 91.75146484375,
      "learning_rate": 2.8005522638352435e-09,
      "loss": 5.3648,
      "step": 24589
    },
    {
      "epoch": 0.9927331449333872,
      "grad_norm": 118.75975036621094,
      "learning_rate": 2.769693843904797e-09,
      "loss": 3.2847,
      "step": 24590
    },
    {
      "epoch": 0.9927735163504239,
      "grad_norm": 156.32777404785156,
      "learning_rate": 2.739006353267293e-09,
      "loss": 5.7375,
      "step": 24591
    },
    {
      "epoch": 0.9928138877674606,
      "grad_norm": 98.04434204101562,
      "learning_rate": 2.7084897924478658e-09,
      "loss": 4.8766,
      "step": 24592
    },
    {
      "epoch": 0.9928542591844973,
      "grad_norm": 78.4273681640625,
      "learning_rate": 2.6781441619683213e-09,
      "loss": 4.867,
      "step": 24593
    },
    {
      "epoch": 0.9928946306015342,
      "grad_norm": 88.05889129638672,
      "learning_rate": 2.6479694623460228e-09,
      "loss": 4.4609,
      "step": 24594
    },
    {
      "epoch": 0.9929350020185709,
      "grad_norm": 85.7508544921875,
      "learning_rate": 2.6179656940983344e-09,
      "loss": 4.6498,
      "step": 24595
    },
    {
      "epoch": 0.9929753734356076,
      "grad_norm": 81.90533447265625,
      "learning_rate": 2.5881328577381794e-09,
      "loss": 3.5816,
      "step": 24596
    },
    {
      "epoch": 0.9930157448526443,
      "grad_norm": 106.90711975097656,
      "learning_rate": 2.55847095377626e-09,
      "loss": 4.8825,
      "step": 24597
    },
    {
      "epoch": 0.993056116269681,
      "grad_norm": 107.3477783203125,
      "learning_rate": 2.5289799827177275e-09,
      "loss": 5.6265,
      "step": 24598
    },
    {
      "epoch": 0.9930964876867178,
      "grad_norm": 515.24365234375,
      "learning_rate": 2.4996599450677338e-09,
      "loss": 4.9383,
      "step": 24599
    },
    {
      "epoch": 0.9931368591037545,
      "grad_norm": 77.70519256591797,
      "learning_rate": 2.4705108413280997e-09,
      "loss": 3.6334,
      "step": 24600
    },
    {
      "epoch": 0.9931772305207913,
      "grad_norm": 92.39806365966797,
      "learning_rate": 2.4415326719962052e-09,
      "loss": 5.3574,
      "step": 24601
    },
    {
      "epoch": 0.993217601937828,
      "grad_norm": 87.67471313476562,
      "learning_rate": 2.4127254375683195e-09,
      "loss": 4.2963,
      "step": 24602
    },
    {
      "epoch": 0.9932579733548648,
      "grad_norm": 117.5634994506836,
      "learning_rate": 2.3840891385373822e-09,
      "loss": 5.6715,
      "step": 24603
    },
    {
      "epoch": 0.9932983447719015,
      "grad_norm": 69.24528503417969,
      "learning_rate": 2.355623775391891e-09,
      "loss": 4.4453,
      "step": 24604
    },
    {
      "epoch": 0.9933387161889382,
      "grad_norm": 106.8982925415039,
      "learning_rate": 2.327329348619234e-09,
      "loss": 4.9905,
      "step": 24605
    },
    {
      "epoch": 0.9933790876059749,
      "grad_norm": 119.2882308959961,
      "learning_rate": 2.299205858702358e-09,
      "loss": 4.9964,
      "step": 24606
    },
    {
      "epoch": 0.9934194590230117,
      "grad_norm": 69.6381607055664,
      "learning_rate": 2.2712533061242106e-09,
      "loss": 5.1504,
      "step": 24607
    },
    {
      "epoch": 0.9934598304400485,
      "grad_norm": 102.80023956298828,
      "learning_rate": 2.243471691359966e-09,
      "loss": 6.2466,
      "step": 24608
    },
    {
      "epoch": 0.9935002018570852,
      "grad_norm": 74.19002532958984,
      "learning_rate": 2.215861014885912e-09,
      "loss": 4.2931,
      "step": 24609
    },
    {
      "epoch": 0.9935405732741219,
      "grad_norm": 1003.4058227539062,
      "learning_rate": 2.1884212771750013e-09,
      "loss": 4.8148,
      "step": 24610
    },
    {
      "epoch": 0.9935809446911587,
      "grad_norm": 90.09949493408203,
      "learning_rate": 2.161152478694639e-09,
      "loss": 3.3604,
      "step": 24611
    },
    {
      "epoch": 0.9936213161081954,
      "grad_norm": 73.46119689941406,
      "learning_rate": 2.1340546199133396e-09,
      "loss": 4.6496,
      "step": 24612
    },
    {
      "epoch": 0.9936616875252321,
      "grad_norm": 123.96826934814453,
      "learning_rate": 2.1071277012918444e-09,
      "loss": 6.4306,
      "step": 24613
    },
    {
      "epoch": 0.9937020589422689,
      "grad_norm": 74.81474304199219,
      "learning_rate": 2.0803717232920073e-09,
      "loss": 5.3678,
      "step": 24614
    },
    {
      "epoch": 0.9937424303593057,
      "grad_norm": 100.39288330078125,
      "learning_rate": 2.0537866863701294e-09,
      "loss": 4.3826,
      "step": 24615
    },
    {
      "epoch": 0.9937828017763424,
      "grad_norm": 104.28193664550781,
      "learning_rate": 2.0273725909825127e-09,
      "loss": 4.7526,
      "step": 24616
    },
    {
      "epoch": 0.9938231731933791,
      "grad_norm": 88.1957778930664,
      "learning_rate": 2.0011294375799074e-09,
      "loss": 4.947,
      "step": 24617
    },
    {
      "epoch": 0.9938635446104158,
      "grad_norm": 73.68981170654297,
      "learning_rate": 1.9750572266108436e-09,
      "loss": 5.5186,
      "step": 24618
    },
    {
      "epoch": 0.9939039160274525,
      "grad_norm": 97.46051788330078,
      "learning_rate": 1.9491559585216315e-09,
      "loss": 5.4038,
      "step": 24619
    },
    {
      "epoch": 0.9939442874444893,
      "grad_norm": 208.78817749023438,
      "learning_rate": 1.923425633754139e-09,
      "loss": 5.0334,
      "step": 24620
    },
    {
      "epoch": 0.9939846588615261,
      "grad_norm": 83.10792541503906,
      "learning_rate": 1.8978662527491255e-09,
      "loss": 4.8383,
      "step": 24621
    },
    {
      "epoch": 0.9940250302785628,
      "grad_norm": 109.48448181152344,
      "learning_rate": 1.872477815942908e-09,
      "loss": 4.6252,
      "step": 24622
    },
    {
      "epoch": 0.9940654016955995,
      "grad_norm": 134.6053466796875,
      "learning_rate": 1.8472603237706942e-09,
      "loss": 6.3457,
      "step": 24623
    },
    {
      "epoch": 0.9941057731126363,
      "grad_norm": 175.40170288085938,
      "learning_rate": 1.8222137766621407e-09,
      "loss": 5.4622,
      "step": 24624
    },
    {
      "epoch": 0.994146144529673,
      "grad_norm": 108.29621887207031,
      "learning_rate": 1.7973381750469032e-09,
      "loss": 5.4741,
      "step": 24625
    },
    {
      "epoch": 0.9941865159467097,
      "grad_norm": 102.4649658203125,
      "learning_rate": 1.7726335193501976e-09,
      "loss": 4.6092,
      "step": 24626
    },
    {
      "epoch": 0.9942268873637464,
      "grad_norm": 79.50275421142578,
      "learning_rate": 1.7480998099927981e-09,
      "loss": 5.4898,
      "step": 24627
    },
    {
      "epoch": 0.9942672587807833,
      "grad_norm": 111.19223022460938,
      "learning_rate": 1.7237370473965898e-09,
      "loss": 5.4578,
      "step": 24628
    },
    {
      "epoch": 0.99430763019782,
      "grad_norm": 84.4725112915039,
      "learning_rate": 1.6995452319756856e-09,
      "loss": 5.3462,
      "step": 24629
    },
    {
      "epoch": 0.9943480016148567,
      "grad_norm": 103.42748260498047,
      "learning_rate": 1.675524364145309e-09,
      "loss": 4.1126,
      "step": 24630
    },
    {
      "epoch": 0.9943883730318934,
      "grad_norm": 110.14374542236328,
      "learning_rate": 1.6516744443151323e-09,
      "loss": 4.8261,
      "step": 24631
    },
    {
      "epoch": 0.9944287444489301,
      "grad_norm": 84.60274505615234,
      "learning_rate": 1.6279954728948278e-09,
      "loss": 5.164,
      "step": 24632
    },
    {
      "epoch": 0.9944691158659669,
      "grad_norm": 115.75111389160156,
      "learning_rate": 1.6044874502862962e-09,
      "loss": 5.0357,
      "step": 24633
    },
    {
      "epoch": 0.9945094872830036,
      "grad_norm": 162.7484893798828,
      "learning_rate": 1.5811503768947688e-09,
      "loss": 5.3157,
      "step": 24634
    },
    {
      "epoch": 0.9945498587000404,
      "grad_norm": 172.31044006347656,
      "learning_rate": 1.5579842531165956e-09,
      "loss": 3.8579,
      "step": 24635
    },
    {
      "epoch": 0.9945902301170771,
      "grad_norm": 70.33779907226562,
      "learning_rate": 1.534989079348126e-09,
      "loss": 5.6003,
      "step": 24636
    },
    {
      "epoch": 0.9946306015341139,
      "grad_norm": 75.70853424072266,
      "learning_rate": 1.5121648559845992e-09,
      "loss": 4.8293,
      "step": 24637
    },
    {
      "epoch": 0.9946709729511506,
      "grad_norm": 113.50301361083984,
      "learning_rate": 1.4895115834145935e-09,
      "loss": 2.6459,
      "step": 24638
    },
    {
      "epoch": 0.9947113443681873,
      "grad_norm": 82.54877471923828,
      "learning_rate": 1.4670292620255766e-09,
      "loss": 5.9694,
      "step": 24639
    },
    {
      "epoch": 0.994751715785224,
      "grad_norm": 72.99303436279297,
      "learning_rate": 1.4447178922016858e-09,
      "loss": 5.3517,
      "step": 24640
    },
    {
      "epoch": 0.9947920872022608,
      "grad_norm": 132.4003143310547,
      "learning_rate": 1.4225774743248377e-09,
      "loss": 4.8671,
      "step": 24641
    },
    {
      "epoch": 0.9948324586192976,
      "grad_norm": 83.70667266845703,
      "learning_rate": 1.4006080087747286e-09,
      "loss": 5.5051,
      "step": 24642
    },
    {
      "epoch": 0.9948728300363343,
      "grad_norm": 109.51715850830078,
      "learning_rate": 1.3788094959243936e-09,
      "loss": 5.6666,
      "step": 24643
    },
    {
      "epoch": 0.994913201453371,
      "grad_norm": 146.05117797851562,
      "learning_rate": 1.357181936149088e-09,
      "loss": 4.7257,
      "step": 24644
    },
    {
      "epoch": 0.9949535728704078,
      "grad_norm": 119.6917495727539,
      "learning_rate": 1.3357253298162954e-09,
      "loss": 5.3666,
      "step": 24645
    },
    {
      "epoch": 0.9949939442874445,
      "grad_norm": 79.3295669555664,
      "learning_rate": 1.31443967729461e-09,
      "loss": 4.6804,
      "step": 24646
    },
    {
      "epoch": 0.9950343157044812,
      "grad_norm": 90.91357421875,
      "learning_rate": 1.293324978948185e-09,
      "loss": 5.5405,
      "step": 24647
    },
    {
      "epoch": 0.9950746871215179,
      "grad_norm": 77.12947845458984,
      "learning_rate": 1.2723812351367327e-09,
      "loss": 5.2421,
      "step": 24648
    },
    {
      "epoch": 0.9951150585385548,
      "grad_norm": 72.1380615234375,
      "learning_rate": 1.2516084462177447e-09,
      "loss": 5.3611,
      "step": 24649
    },
    {
      "epoch": 0.9951554299555915,
      "grad_norm": 83.49496459960938,
      "learning_rate": 1.2310066125487131e-09,
      "loss": 6.4828,
      "step": 24650
    },
    {
      "epoch": 0.9951958013726282,
      "grad_norm": 110.21385192871094,
      "learning_rate": 1.210575734480468e-09,
      "loss": 5.6625,
      "step": 24651
    },
    {
      "epoch": 0.9952361727896649,
      "grad_norm": 93.71632385253906,
      "learning_rate": 1.19031581236273e-09,
      "loss": 4.4968,
      "step": 24652
    },
    {
      "epoch": 0.9952765442067016,
      "grad_norm": 165.1602783203125,
      "learning_rate": 1.1702268465407785e-09,
      "loss": 3.8997,
      "step": 24653
    },
    {
      "epoch": 0.9953169156237384,
      "grad_norm": 87.17581939697266,
      "learning_rate": 1.1503088373598925e-09,
      "loss": 5.6526,
      "step": 24654
    },
    {
      "epoch": 0.9953572870407751,
      "grad_norm": 174.7425537109375,
      "learning_rate": 1.1305617851598006e-09,
      "loss": 4.0664,
      "step": 24655
    },
    {
      "epoch": 0.9953976584578119,
      "grad_norm": 121.1791000366211,
      "learning_rate": 1.1109856902769e-09,
      "loss": 3.7289,
      "step": 24656
    },
    {
      "epoch": 0.9954380298748486,
      "grad_norm": 130.5452880859375,
      "learning_rate": 1.0915805530486989e-09,
      "loss": 4.5265,
      "step": 24657
    },
    {
      "epoch": 0.9954784012918854,
      "grad_norm": 89.89440155029297,
      "learning_rate": 1.072346373803823e-09,
      "loss": 5.8513,
      "step": 24658
    },
    {
      "epoch": 0.9955187727089221,
      "grad_norm": 236.92323303222656,
      "learning_rate": 1.053283152873119e-09,
      "loss": 4.8338,
      "step": 24659
    },
    {
      "epoch": 0.9955591441259588,
      "grad_norm": 63.737030029296875,
      "learning_rate": 1.0343908905829926e-09,
      "loss": 4.1089,
      "step": 24660
    },
    {
      "epoch": 0.9955995155429955,
      "grad_norm": 83.58997344970703,
      "learning_rate": 1.0156695872542977e-09,
      "loss": 5.7139,
      "step": 24661
    },
    {
      "epoch": 0.9956398869600322,
      "grad_norm": 91.6823959350586,
      "learning_rate": 9.971192432089993e-10,
      "loss": 6.2234,
      "step": 24662
    },
    {
      "epoch": 0.9956802583770691,
      "grad_norm": 114.10812377929688,
      "learning_rate": 9.78739858763511e-10,
      "loss": 4.3874,
      "step": 24663
    },
    {
      "epoch": 0.9957206297941058,
      "grad_norm": 109.0263442993164,
      "learning_rate": 9.60531434232026e-10,
      "loss": 4.1982,
      "step": 24664
    },
    {
      "epoch": 0.9957610012111425,
      "grad_norm": 93.81777954101562,
      "learning_rate": 9.424939699265168e-10,
      "loss": 6.6495,
      "step": 24665
    },
    {
      "epoch": 0.9958013726281792,
      "grad_norm": 100.5233154296875,
      "learning_rate": 9.246274661545151e-10,
      "loss": 4.1426,
      "step": 24666
    },
    {
      "epoch": 0.995841744045216,
      "grad_norm": 105.7243881225586,
      "learning_rate": 9.069319232213325e-10,
      "loss": 4.4533,
      "step": 24667
    },
    {
      "epoch": 0.9958821154622527,
      "grad_norm": 181.30792236328125,
      "learning_rate": 8.894073414300597e-10,
      "loss": 4.6962,
      "step": 24668
    },
    {
      "epoch": 0.9959224868792894,
      "grad_norm": 70.00132751464844,
      "learning_rate": 8.720537210815671e-10,
      "loss": 5.77,
      "step": 24669
    },
    {
      "epoch": 0.9959628582963262,
      "grad_norm": 72.94628143310547,
      "learning_rate": 8.54871062470064e-10,
      "loss": 5.7076,
      "step": 24670
    },
    {
      "epoch": 0.996003229713363,
      "grad_norm": 92.53389739990234,
      "learning_rate": 8.378593658908696e-10,
      "loss": 6.1252,
      "step": 24671
    },
    {
      "epoch": 0.9960436011303997,
      "grad_norm": 88.80926513671875,
      "learning_rate": 8.210186316348623e-10,
      "loss": 6.5425,
      "step": 24672
    },
    {
      "epoch": 0.9960839725474364,
      "grad_norm": 84.24150085449219,
      "learning_rate": 8.043488599895899e-10,
      "loss": 5.8684,
      "step": 24673
    },
    {
      "epoch": 0.9961243439644731,
      "grad_norm": 100.16633605957031,
      "learning_rate": 7.878500512403797e-10,
      "loss": 6.2861,
      "step": 24674
    },
    {
      "epoch": 0.9961647153815099,
      "grad_norm": 93.61556243896484,
      "learning_rate": 7.715222056692284e-10,
      "loss": 3.8396,
      "step": 24675
    },
    {
      "epoch": 0.9962050867985466,
      "grad_norm": 118.59092712402344,
      "learning_rate": 7.553653235548019e-10,
      "loss": 5.4695,
      "step": 24676
    },
    {
      "epoch": 0.9962454582155834,
      "grad_norm": 68.7528076171875,
      "learning_rate": 7.393794051735459e-10,
      "loss": 6.0207,
      "step": 24677
    },
    {
      "epoch": 0.9962858296326201,
      "grad_norm": 137.77044677734375,
      "learning_rate": 7.235644507985751e-10,
      "loss": 4.6841,
      "step": 24678
    },
    {
      "epoch": 0.9963262010496569,
      "grad_norm": 97.1368408203125,
      "learning_rate": 7.079204607018941e-10,
      "loss": 4.2195,
      "step": 24679
    },
    {
      "epoch": 0.9963665724666936,
      "grad_norm": 106.99542999267578,
      "learning_rate": 6.924474351488464e-10,
      "loss": 4.5666,
      "step": 24680
    },
    {
      "epoch": 0.9964069438837303,
      "grad_norm": 95.03190612792969,
      "learning_rate": 6.771453744047752e-10,
      "loss": 5.1683,
      "step": 24681
    },
    {
      "epoch": 0.996447315300767,
      "grad_norm": 80.67123413085938,
      "learning_rate": 6.62014278731693e-10,
      "loss": 4.9802,
      "step": 24682
    },
    {
      "epoch": 0.9964876867178037,
      "grad_norm": 242.62176513671875,
      "learning_rate": 6.470541483882819e-10,
      "loss": 5.6054,
      "step": 24683
    },
    {
      "epoch": 0.9965280581348406,
      "grad_norm": 152.04750061035156,
      "learning_rate": 6.322649836298933e-10,
      "loss": 3.3443,
      "step": 24684
    },
    {
      "epoch": 0.9965684295518773,
      "grad_norm": 776.4803466796875,
      "learning_rate": 6.176467847096579e-10,
      "loss": 3.4164,
      "step": 24685
    },
    {
      "epoch": 0.996608800968914,
      "grad_norm": 135.2609405517578,
      "learning_rate": 6.031995518773758e-10,
      "loss": 4.496,
      "step": 24686
    },
    {
      "epoch": 0.9966491723859507,
      "grad_norm": 124.77436828613281,
      "learning_rate": 5.889232853795168e-10,
      "loss": 4.8014,
      "step": 24687
    },
    {
      "epoch": 0.9966895438029875,
      "grad_norm": 99.91903686523438,
      "learning_rate": 5.748179854614399e-10,
      "loss": 4.7075,
      "step": 24688
    },
    {
      "epoch": 0.9967299152200242,
      "grad_norm": 98.79084777832031,
      "learning_rate": 5.608836523640637e-10,
      "loss": 4.1946,
      "step": 24689
    },
    {
      "epoch": 0.9967702866370609,
      "grad_norm": 276.32745361328125,
      "learning_rate": 5.471202863238656e-10,
      "loss": 5.361,
      "step": 24690
    },
    {
      "epoch": 0.9968106580540977,
      "grad_norm": 80.69317626953125,
      "learning_rate": 5.335278875784333e-10,
      "loss": 4.2994,
      "step": 24691
    },
    {
      "epoch": 0.9968510294711345,
      "grad_norm": 85.05365753173828,
      "learning_rate": 5.201064563586933e-10,
      "loss": 3.8835,
      "step": 24692
    },
    {
      "epoch": 0.9968914008881712,
      "grad_norm": 78.21004486083984,
      "learning_rate": 5.068559928955719e-10,
      "loss": 4.6977,
      "step": 24693
    },
    {
      "epoch": 0.9969317723052079,
      "grad_norm": 91.00283813476562,
      "learning_rate": 4.937764974144444e-10,
      "loss": 4.9881,
      "step": 24694
    },
    {
      "epoch": 0.9969721437222446,
      "grad_norm": 97.04171752929688,
      "learning_rate": 4.808679701384656e-10,
      "loss": 4.687,
      "step": 24695
    },
    {
      "epoch": 0.9970125151392814,
      "grad_norm": 75.58805847167969,
      "learning_rate": 4.6813041128968e-10,
      "loss": 5.8816,
      "step": 24696
    },
    {
      "epoch": 0.9970528865563181,
      "grad_norm": 115.7783203125,
      "learning_rate": 4.5556382108569165e-10,
      "loss": 5.22,
      "step": 24697
    },
    {
      "epoch": 0.9970932579733549,
      "grad_norm": 96.03129577636719,
      "learning_rate": 4.431681997396631e-10,
      "loss": 3.8458,
      "step": 24698
    },
    {
      "epoch": 0.9971336293903916,
      "grad_norm": 114.88199615478516,
      "learning_rate": 4.309435474658674e-10,
      "loss": 5.4524,
      "step": 24699
    },
    {
      "epoch": 0.9971740008074284,
      "grad_norm": 104.3581771850586,
      "learning_rate": 4.188898644719164e-10,
      "loss": 5.5823,
      "step": 24700
    },
    {
      "epoch": 0.9972143722244651,
      "grad_norm": 71.12039184570312,
      "learning_rate": 4.0700715096431144e-10,
      "loss": 4.0564,
      "step": 24701
    },
    {
      "epoch": 0.9972547436415018,
      "grad_norm": 90.66929626464844,
      "learning_rate": 3.9529540714622337e-10,
      "loss": 4.3761,
      "step": 24702
    },
    {
      "epoch": 0.9972951150585385,
      "grad_norm": 206.65536499023438,
      "learning_rate": 3.837546332174924e-10,
      "loss": 3.0538,
      "step": 24703
    },
    {
      "epoch": 0.9973354864755752,
      "grad_norm": 430.7817077636719,
      "learning_rate": 3.7238482937684837e-10,
      "loss": 4.1499,
      "step": 24704
    },
    {
      "epoch": 0.9973758578926121,
      "grad_norm": 91.28179931640625,
      "learning_rate": 3.611859958163599e-10,
      "loss": 5.25,
      "step": 24705
    },
    {
      "epoch": 0.9974162293096488,
      "grad_norm": 72.80616760253906,
      "learning_rate": 3.5015813272920587e-10,
      "loss": 5.2178,
      "step": 24706
    },
    {
      "epoch": 0.9974566007266855,
      "grad_norm": 100.59264373779297,
      "learning_rate": 3.39301240304124e-10,
      "loss": 4.8972,
      "step": 24707
    },
    {
      "epoch": 0.9974969721437222,
      "grad_norm": 85.77400970458984,
      "learning_rate": 3.2861531872541154e-10,
      "loss": 3.9247,
      "step": 24708
    },
    {
      "epoch": 0.997537343560759,
      "grad_norm": 108.04559326171875,
      "learning_rate": 3.181003681762551e-10,
      "loss": 4.3988,
      "step": 24709
    },
    {
      "epoch": 0.9975777149777957,
      "grad_norm": 78.99446105957031,
      "learning_rate": 3.0775638883651094e-10,
      "loss": 5.0596,
      "step": 24710
    },
    {
      "epoch": 0.9976180863948324,
      "grad_norm": 90.99554443359375,
      "learning_rate": 2.9758338088381465e-10,
      "loss": 5.8159,
      "step": 24711
    },
    {
      "epoch": 0.9976584578118692,
      "grad_norm": 102.01513671875,
      "learning_rate": 2.875813444913611e-10,
      "loss": 4.1829,
      "step": 24712
    },
    {
      "epoch": 0.997698829228906,
      "grad_norm": 90.36701202392578,
      "learning_rate": 2.777502798301246e-10,
      "loss": 5.1459,
      "step": 24713
    },
    {
      "epoch": 0.9977392006459427,
      "grad_norm": 91.91376495361328,
      "learning_rate": 2.6809018706885905e-10,
      "loss": 6.1588,
      "step": 24714
    },
    {
      "epoch": 0.9977795720629794,
      "grad_norm": 131.86703491210938,
      "learning_rate": 2.586010663707672e-10,
      "loss": 2.1957,
      "step": 24715
    },
    {
      "epoch": 0.9978199434800161,
      "grad_norm": 106.28445434570312,
      "learning_rate": 2.4928291790127236e-10,
      "loss": 3.6988,
      "step": 24716
    },
    {
      "epoch": 0.9978603148970528,
      "grad_norm": 86.35812377929688,
      "learning_rate": 2.401357418169159e-10,
      "loss": 5.8615,
      "step": 24717
    },
    {
      "epoch": 0.9979006863140897,
      "grad_norm": 90.0082015991211,
      "learning_rate": 2.3115953827534954e-10,
      "loss": 4.8769,
      "step": 24718
    },
    {
      "epoch": 0.9979410577311264,
      "grad_norm": 109.24131774902344,
      "learning_rate": 2.2235430742978405e-10,
      "loss": 5.775,
      "step": 24719
    },
    {
      "epoch": 0.9979814291481631,
      "grad_norm": 79.43798828125,
      "learning_rate": 2.137200494300995e-10,
      "loss": 5.4534,
      "step": 24720
    },
    {
      "epoch": 0.9980218005651998,
      "grad_norm": 225.06976318359375,
      "learning_rate": 2.0525676442617603e-10,
      "loss": 6.0119,
      "step": 24721
    },
    {
      "epoch": 0.9980621719822366,
      "grad_norm": 100.74758911132812,
      "learning_rate": 1.969644525601222e-10,
      "loss": 3.9297,
      "step": 24722
    },
    {
      "epoch": 0.9981025433992733,
      "grad_norm": 84.57122039794922,
      "learning_rate": 1.8884311397515676e-10,
      "loss": 6.3949,
      "step": 24723
    },
    {
      "epoch": 0.99814291481631,
      "grad_norm": 109.83338165283203,
      "learning_rate": 1.8089274880894737e-10,
      "loss": 5.4248,
      "step": 24724
    },
    {
      "epoch": 0.9981832862333468,
      "grad_norm": 75.43324279785156,
      "learning_rate": 1.7311335719916167e-10,
      "loss": 5.4054,
      "step": 24725
    },
    {
      "epoch": 0.9982236576503836,
      "grad_norm": 86.4859619140625,
      "learning_rate": 1.65504939276806e-10,
      "loss": 4.5198,
      "step": 24726
    },
    {
      "epoch": 0.9982640290674203,
      "grad_norm": 79.32684326171875,
      "learning_rate": 1.5806749517399688e-10,
      "loss": 5.5907,
      "step": 24727
    },
    {
      "epoch": 0.998304400484457,
      "grad_norm": 89.19843292236328,
      "learning_rate": 1.5080102501618953e-10,
      "loss": 3.8689,
      "step": 24728
    },
    {
      "epoch": 0.9983447719014937,
      "grad_norm": 100.95204162597656,
      "learning_rate": 1.4370552892883916e-10,
      "loss": 5.8101,
      "step": 24729
    },
    {
      "epoch": 0.9983851433185305,
      "grad_norm": 111.3499526977539,
      "learning_rate": 1.3678100703296006e-10,
      "loss": 5.7564,
      "step": 24730
    },
    {
      "epoch": 0.9984255147355672,
      "grad_norm": 88.82865905761719,
      "learning_rate": 1.3002745944734608e-10,
      "loss": 3.909,
      "step": 24731
    },
    {
      "epoch": 0.998465886152604,
      "grad_norm": 79.53868103027344,
      "learning_rate": 1.2344488628523998e-10,
      "loss": 5.0829,
      "step": 24732
    },
    {
      "epoch": 0.9985062575696407,
      "grad_norm": 101.272705078125,
      "learning_rate": 1.1703328766210499e-10,
      "loss": 5.5759,
      "step": 24733
    },
    {
      "epoch": 0.9985466289866775,
      "grad_norm": 120.03935241699219,
      "learning_rate": 1.1079266368563268e-10,
      "loss": 3.6341,
      "step": 24734
    },
    {
      "epoch": 0.9985870004037142,
      "grad_norm": 98.15605926513672,
      "learning_rate": 1.0472301446351474e-10,
      "loss": 4.0653,
      "step": 24735
    },
    {
      "epoch": 0.9986273718207509,
      "grad_norm": 120.90754699707031,
      "learning_rate": 9.882434009900189e-11,
      "loss": 4.561,
      "step": 24736
    },
    {
      "epoch": 0.9986677432377876,
      "grad_norm": 71.16024017333984,
      "learning_rate": 9.309664069312441e-11,
      "loss": 4.871,
      "step": 24737
    },
    {
      "epoch": 0.9987081146548243,
      "grad_norm": 84.27359771728516,
      "learning_rate": 8.753991634469216e-11,
      "loss": 5.7922,
      "step": 24738
    },
    {
      "epoch": 0.9987484860718612,
      "grad_norm": 117.44512176513672,
      "learning_rate": 8.215416714696389e-11,
      "loss": 5.5257,
      "step": 24739
    },
    {
      "epoch": 0.9987888574888979,
      "grad_norm": 100.15077209472656,
      "learning_rate": 7.693939319319832e-11,
      "loss": 4.1085,
      "step": 24740
    },
    {
      "epoch": 0.9988292289059346,
      "grad_norm": 89.006591796875,
      "learning_rate": 7.189559457221329e-11,
      "loss": 4.3729,
      "step": 24741
    },
    {
      "epoch": 0.9988696003229713,
      "grad_norm": 145.2025146484375,
      "learning_rate": 6.702277136949598e-11,
      "loss": 4.5177,
      "step": 24742
    },
    {
      "epoch": 0.9989099717400081,
      "grad_norm": 77.67106628417969,
      "learning_rate": 6.232092367053355e-11,
      "loss": 3.9838,
      "step": 24743
    },
    {
      "epoch": 0.9989503431570448,
      "grad_norm": 151.1380157470703,
      "learning_rate": 5.779005155304163e-11,
      "loss": 4.9918,
      "step": 24744
    },
    {
      "epoch": 0.9989907145740815,
      "grad_norm": 87.45000457763672,
      "learning_rate": 5.343015509695626e-11,
      "loss": 6.1322,
      "step": 24745
    },
    {
      "epoch": 0.9990310859911183,
      "grad_norm": 101.51659393310547,
      "learning_rate": 4.9241234374441946e-11,
      "loss": 4.9971,
      "step": 24746
    },
    {
      "epoch": 0.9990714574081551,
      "grad_norm": 107.593994140625,
      "learning_rate": 4.5223289459883636e-11,
      "loss": 6.4338,
      "step": 24747
    },
    {
      "epoch": 0.9991118288251918,
      "grad_norm": 85.12853240966797,
      "learning_rate": 4.1376320418784476e-11,
      "loss": 3.8588,
      "step": 24748
    },
    {
      "epoch": 0.9991522002422285,
      "grad_norm": 112.38548278808594,
      "learning_rate": 3.77003273199783e-11,
      "loss": 5.6646,
      "step": 24749
    },
    {
      "epoch": 0.9991925716592652,
      "grad_norm": 104.73077392578125,
      "learning_rate": 3.419531022452738e-11,
      "loss": 5.8562,
      "step": 24750
    },
    {
      "epoch": 0.999232943076302,
      "grad_norm": 91.66246795654297,
      "learning_rate": 3.086126919238375e-11,
      "loss": 5.1687,
      "step": 24751
    },
    {
      "epoch": 0.9992733144933387,
      "grad_norm": 80.69548034667969,
      "learning_rate": 2.7698204281279005e-11,
      "loss": 5.0903,
      "step": 24752
    },
    {
      "epoch": 0.9993136859103755,
      "grad_norm": 120.99761199951172,
      "learning_rate": 2.4706115544503863e-11,
      "loss": 5.4386,
      "step": 24753
    },
    {
      "epoch": 0.9993540573274122,
      "grad_norm": 97.7916030883789,
      "learning_rate": 2.1885003034238795e-11,
      "loss": 4.9335,
      "step": 24754
    },
    {
      "epoch": 0.999394428744449,
      "grad_norm": 133.52456665039062,
      "learning_rate": 1.923486679711317e-11,
      "loss": 5.5256,
      "step": 24755
    },
    {
      "epoch": 0.9994348001614857,
      "grad_norm": 196.42337036132812,
      "learning_rate": 1.6755706879756363e-11,
      "loss": 4.488,
      "step": 24756
    },
    {
      "epoch": 0.9994751715785224,
      "grad_norm": 100.37749481201172,
      "learning_rate": 1.4447523324356838e-11,
      "loss": 4.3115,
      "step": 24757
    },
    {
      "epoch": 0.9995155429955591,
      "grad_norm": 97.87213897705078,
      "learning_rate": 1.231031617088263e-11,
      "loss": 6.1394,
      "step": 24758
    },
    {
      "epoch": 0.9995559144125958,
      "grad_norm": 95.47859191894531,
      "learning_rate": 1.0344085454860875e-11,
      "loss": 4.1774,
      "step": 24759
    },
    {
      "epoch": 0.9995962858296327,
      "grad_norm": 93.83086395263672,
      "learning_rate": 8.548831210708486e-12,
      "loss": 4.9235,
      "step": 24760
    },
    {
      "epoch": 0.9996366572466694,
      "grad_norm": 136.488037109375,
      "learning_rate": 6.924553467291262e-12,
      "loss": 5.4066,
      "step": 24761
    },
    {
      "epoch": 0.9996770286637061,
      "grad_norm": 88.60922241210938,
      "learning_rate": 5.471252255695447e-12,
      "loss": 6.1521,
      "step": 24762
    },
    {
      "epoch": 0.9997174000807428,
      "grad_norm": 158.7760772705078,
      "learning_rate": 4.18892759701528e-12,
      "loss": 5.9307,
      "step": 24763
    },
    {
      "epoch": 0.9997577714977796,
      "grad_norm": 95.1839828491211,
      "learning_rate": 3.077579516785889e-12,
      "loss": 4.8384,
      "step": 24764
    },
    {
      "epoch": 0.9997981429148163,
      "grad_norm": 80.76246643066406,
      "learning_rate": 2.1372080305503973e-12,
      "loss": 4.27,
      "step": 24765
    },
    {
      "epoch": 0.999838514331853,
      "grad_norm": 201.98439025878906,
      "learning_rate": 1.3678131571825959e-12,
      "loss": 4.0777,
      "step": 24766
    },
    {
      "epoch": 0.9998788857488898,
      "grad_norm": 90.56375122070312,
      "learning_rate": 7.693949088949382e-13,
      "loss": 5.3761,
      "step": 24767
    },
    {
      "epoch": 0.9999192571659266,
      "grad_norm": 73.69155883789062,
      "learning_rate": 3.4195329567943135e-13,
      "loss": 5.1349,
      "step": 24768
    },
    {
      "epoch": 0.9999596285829633,
      "grad_norm": 75.61074829101562,
      "learning_rate": 8.54883241974136e-14,
      "loss": 1.165,
      "step": 24769
    },
    {
      "epoch": 1.0,
      "grad_norm": 168.19725036621094,
      "learning_rate": 0.0,
      "loss": 6.2623,
      "step": 24770
    },
    {
      "epoch": 1.0,
      "step": 24770,
      "total_flos": 1.4213815936026166e+17,
      "train_loss": 5.2030843579003925,
      "train_runtime": 8590.2737,
      "train_samples_per_second": 23.068,
      "train_steps_per_second": 2.883
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 24770,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 1.4213815936026166e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}